{
  "best_metric": 11.5,
  "best_model_checkpoint": "miner_id_24/checkpoint-1000",
  "epoch": 2.4009603841536613,
  "eval_steps": 1000,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006002400960384153,
      "grad_norm": 1.8054486645269208e-05,
      "learning_rate": 6.666666666666667e-06,
      "loss": 23.0,
      "step": 1
    },
    {
      "epoch": 0.0006002400960384153,
      "eval_loss": 11.5,
      "eval_runtime": 5.3215,
      "eval_samples_per_second": 263.649,
      "eval_steps_per_second": 33.074,
      "step": 1
    },
    {
      "epoch": 0.0012004801920768306,
      "grad_norm": 2.727241735556163e-05,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 23.0,
      "step": 2
    },
    {
      "epoch": 0.001800720288115246,
      "grad_norm": 2.7739322831621394e-05,
      "learning_rate": 2e-05,
      "loss": 23.0,
      "step": 3
    },
    {
      "epoch": 0.0024009603841536613,
      "grad_norm": 2.7980300728813745e-05,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 23.0,
      "step": 4
    },
    {
      "epoch": 0.003001200480192077,
      "grad_norm": 3.064045085920952e-05,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 23.0,
      "step": 5
    },
    {
      "epoch": 0.003601440576230492,
      "grad_norm": 2.1872214347240515e-05,
      "learning_rate": 4e-05,
      "loss": 23.0,
      "step": 6
    },
    {
      "epoch": 0.004201680672268907,
      "grad_norm": 3.888764695147984e-05,
      "learning_rate": 4.666666666666667e-05,
      "loss": 23.0,
      "step": 7
    },
    {
      "epoch": 0.004801920768307323,
      "grad_norm": 2.100978235830553e-05,
      "learning_rate": 5.333333333333333e-05,
      "loss": 23.0,
      "step": 8
    },
    {
      "epoch": 0.005402160864345739,
      "grad_norm": 2.6072715627378784e-05,
      "learning_rate": 6e-05,
      "loss": 23.0,
      "step": 9
    },
    {
      "epoch": 0.006002400960384154,
      "grad_norm": 1.7852040400612168e-05,
      "learning_rate": 6.666666666666667e-05,
      "loss": 23.0,
      "step": 10
    },
    {
      "epoch": 0.006602641056422569,
      "grad_norm": 2.71389744739281e-05,
      "learning_rate": 7.333333333333333e-05,
      "loss": 23.0,
      "step": 11
    },
    {
      "epoch": 0.007202881152460984,
      "grad_norm": 2.6971414627041668e-05,
      "learning_rate": 8e-05,
      "loss": 23.0,
      "step": 12
    },
    {
      "epoch": 0.0078031212484993995,
      "grad_norm": 2.012933873629663e-05,
      "learning_rate": 8.666666666666667e-05,
      "loss": 23.0,
      "step": 13
    },
    {
      "epoch": 0.008403361344537815,
      "grad_norm": 2.6451836674823426e-05,
      "learning_rate": 9.333333333333334e-05,
      "loss": 23.0,
      "step": 14
    },
    {
      "epoch": 0.00900360144057623,
      "grad_norm": 3.4380853321636096e-05,
      "learning_rate": 0.0001,
      "loss": 23.0,
      "step": 15
    },
    {
      "epoch": 0.009603841536614645,
      "grad_norm": 2.7640449843602255e-05,
      "learning_rate": 0.00010666666666666667,
      "loss": 23.0,
      "step": 16
    },
    {
      "epoch": 0.01020408163265306,
      "grad_norm": 1.8410619304631837e-05,
      "learning_rate": 0.00011333333333333334,
      "loss": 23.0,
      "step": 17
    },
    {
      "epoch": 0.010804321728691477,
      "grad_norm": 2.8392196327331476e-05,
      "learning_rate": 0.00012,
      "loss": 23.0,
      "step": 18
    },
    {
      "epoch": 0.011404561824729893,
      "grad_norm": 2.826511081366334e-05,
      "learning_rate": 0.00012666666666666666,
      "loss": 23.0,
      "step": 19
    },
    {
      "epoch": 0.012004801920768308,
      "grad_norm": 3.356917659402825e-05,
      "learning_rate": 0.00013333333333333334,
      "loss": 23.0,
      "step": 20
    },
    {
      "epoch": 0.012605042016806723,
      "grad_norm": 2.246521216875408e-05,
      "learning_rate": 0.00014,
      "loss": 23.0,
      "step": 21
    },
    {
      "epoch": 0.013205282112845138,
      "grad_norm": 2.638386831677053e-05,
      "learning_rate": 0.00014666666666666666,
      "loss": 23.0,
      "step": 22
    },
    {
      "epoch": 0.013805522208883553,
      "grad_norm": 3.957346052629873e-05,
      "learning_rate": 0.00015333333333333334,
      "loss": 23.0,
      "step": 23
    },
    {
      "epoch": 0.014405762304921969,
      "grad_norm": 2.2368900317815132e-05,
      "learning_rate": 0.00016,
      "loss": 23.0,
      "step": 24
    },
    {
      "epoch": 0.015006002400960384,
      "grad_norm": 2.7279023925075307e-05,
      "learning_rate": 0.0001666666666666667,
      "loss": 23.0,
      "step": 25
    },
    {
      "epoch": 0.015606242496998799,
      "grad_norm": 2.1262459995341487e-05,
      "learning_rate": 0.00017333333333333334,
      "loss": 23.0,
      "step": 26
    },
    {
      "epoch": 0.016206482593037214,
      "grad_norm": 2.396126728854142e-05,
      "learning_rate": 0.00018,
      "loss": 23.0,
      "step": 27
    },
    {
      "epoch": 0.01680672268907563,
      "grad_norm": 1.7820493667386472e-05,
      "learning_rate": 0.0001866666666666667,
      "loss": 23.0,
      "step": 28
    },
    {
      "epoch": 0.017406962785114045,
      "grad_norm": 2.8694874345092103e-05,
      "learning_rate": 0.00019333333333333333,
      "loss": 23.0,
      "step": 29
    },
    {
      "epoch": 0.01800720288115246,
      "grad_norm": 2.7091318770544603e-05,
      "learning_rate": 0.0002,
      "loss": 23.0,
      "step": 30
    },
    {
      "epoch": 0.018607442977190875,
      "grad_norm": 3.307936276542023e-05,
      "learning_rate": 0.0001999999982156286,
      "loss": 23.0,
      "step": 31
    },
    {
      "epoch": 0.01920768307322929,
      "grad_norm": 3.513343108352274e-05,
      "learning_rate": 0.00019999999286251453,
      "loss": 23.0,
      "step": 32
    },
    {
      "epoch": 0.019807923169267706,
      "grad_norm": 1.9473052816465497e-05,
      "learning_rate": 0.00019999998394065792,
      "loss": 23.0,
      "step": 33
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 2.3083173800841905e-05,
      "learning_rate": 0.00019999997145005914,
      "loss": 23.0,
      "step": 34
    },
    {
      "epoch": 0.02100840336134454,
      "grad_norm": 3.38441641360987e-05,
      "learning_rate": 0.0001999999553907186,
      "loss": 23.0,
      "step": 35
    },
    {
      "epoch": 0.021608643457382955,
      "grad_norm": 3.0117924325168133e-05,
      "learning_rate": 0.00019999993576263686,
      "loss": 23.0,
      "step": 36
    },
    {
      "epoch": 0.02220888355342137,
      "grad_norm": 2.34516519412864e-05,
      "learning_rate": 0.00019999991256581466,
      "loss": 23.0,
      "step": 37
    },
    {
      "epoch": 0.022809123649459785,
      "grad_norm": 3.247776839998551e-05,
      "learning_rate": 0.0001999998858002528,
      "loss": 23.0,
      "step": 38
    },
    {
      "epoch": 0.0234093637454982,
      "grad_norm": 2.7047510229749605e-05,
      "learning_rate": 0.00019999985546595225,
      "loss": 23.0,
      "step": 39
    },
    {
      "epoch": 0.024009603841536616,
      "grad_norm": 2.9209651984274387e-05,
      "learning_rate": 0.00019999982156291407,
      "loss": 23.0,
      "step": 40
    },
    {
      "epoch": 0.02460984393757503,
      "grad_norm": 2.5179328076774254e-05,
      "learning_rate": 0.00019999978409113953,
      "loss": 23.0,
      "step": 41
    },
    {
      "epoch": 0.025210084033613446,
      "grad_norm": 2.729222433117684e-05,
      "learning_rate": 0.0001999997430506299,
      "loss": 23.0,
      "step": 42
    },
    {
      "epoch": 0.02581032412965186,
      "grad_norm": 3.7065288779558614e-05,
      "learning_rate": 0.0001999996984413867,
      "loss": 23.0,
      "step": 43
    },
    {
      "epoch": 0.026410564225690276,
      "grad_norm": 3.8142901757964864e-05,
      "learning_rate": 0.00019999965026341144,
      "loss": 23.0,
      "step": 44
    },
    {
      "epoch": 0.02701080432172869,
      "grad_norm": 3.7107929529156536e-05,
      "learning_rate": 0.00019999959851670594,
      "loss": 23.0,
      "step": 45
    },
    {
      "epoch": 0.027611044417767107,
      "grad_norm": 4.2656189179979265e-05,
      "learning_rate": 0.00019999954320127196,
      "loss": 23.0,
      "step": 46
    },
    {
      "epoch": 0.028211284513805522,
      "grad_norm": 3.613159424276091e-05,
      "learning_rate": 0.00019999948431711153,
      "loss": 23.0,
      "step": 47
    },
    {
      "epoch": 0.028811524609843937,
      "grad_norm": 4.232913488522172e-05,
      "learning_rate": 0.00019999942186422673,
      "loss": 23.0,
      "step": 48
    },
    {
      "epoch": 0.029411764705882353,
      "grad_norm": 3.183892840752378e-05,
      "learning_rate": 0.00019999935584261982,
      "loss": 23.0,
      "step": 49
    },
    {
      "epoch": 0.030012004801920768,
      "grad_norm": 3.41582708642818e-05,
      "learning_rate": 0.0001999992862522931,
      "loss": 23.0,
      "step": 50
    },
    {
      "epoch": 0.030612244897959183,
      "grad_norm": 4.701867146650329e-05,
      "learning_rate": 0.0001999992130932491,
      "loss": 23.0,
      "step": 51
    },
    {
      "epoch": 0.031212484993997598,
      "grad_norm": 8.062861888902262e-05,
      "learning_rate": 0.00019999913636549037,
      "loss": 23.0,
      "step": 52
    },
    {
      "epoch": 0.03181272509003601,
      "grad_norm": 4.4446765969041735e-05,
      "learning_rate": 0.00019999905606901973,
      "loss": 23.0,
      "step": 53
    },
    {
      "epoch": 0.03241296518607443,
      "grad_norm": 7.205520523712039e-05,
      "learning_rate": 0.00019999897220384,
      "loss": 23.0,
      "step": 54
    },
    {
      "epoch": 0.033013205282112844,
      "grad_norm": 2.287816460011527e-05,
      "learning_rate": 0.0001999988847699542,
      "loss": 23.0,
      "step": 55
    },
    {
      "epoch": 0.03361344537815126,
      "grad_norm": 5.946783130639233e-05,
      "learning_rate": 0.0001999987937673654,
      "loss": 23.0,
      "step": 56
    },
    {
      "epoch": 0.034213685474189674,
      "grad_norm": 2.981833131343592e-05,
      "learning_rate": 0.0001999986991960769,
      "loss": 23.0,
      "step": 57
    },
    {
      "epoch": 0.03481392557022809,
      "grad_norm": 3.543734783306718e-05,
      "learning_rate": 0.00019999860105609206,
      "loss": 23.0,
      "step": 58
    },
    {
      "epoch": 0.035414165666266505,
      "grad_norm": 4.3358595576137304e-05,
      "learning_rate": 0.00019999849934741433,
      "loss": 23.0,
      "step": 59
    },
    {
      "epoch": 0.03601440576230492,
      "grad_norm": 4.9726917495718226e-05,
      "learning_rate": 0.00019999839407004743,
      "loss": 23.0,
      "step": 60
    },
    {
      "epoch": 0.036614645858343335,
      "grad_norm": 4.8018726374721155e-05,
      "learning_rate": 0.00019999828522399504,
      "loss": 23.0,
      "step": 61
    },
    {
      "epoch": 0.03721488595438175,
      "grad_norm": 4.2448387830518186e-05,
      "learning_rate": 0.0001999981728092611,
      "loss": 23.0,
      "step": 62
    },
    {
      "epoch": 0.037815126050420166,
      "grad_norm": 6.509755621664226e-05,
      "learning_rate": 0.0001999980568258496,
      "loss": 23.0,
      "step": 63
    },
    {
      "epoch": 0.03841536614645858,
      "grad_norm": 0.00011501594417495653,
      "learning_rate": 0.00019999793727376467,
      "loss": 23.0,
      "step": 64
    },
    {
      "epoch": 0.039015606242496996,
      "grad_norm": 6.020090222591534e-05,
      "learning_rate": 0.00019999781415301057,
      "loss": 23.0,
      "step": 65
    },
    {
      "epoch": 0.03961584633853541,
      "grad_norm": 7.283796003321186e-05,
      "learning_rate": 0.00019999768746359174,
      "loss": 23.0,
      "step": 66
    },
    {
      "epoch": 0.040216086434573826,
      "grad_norm": 9.18004079721868e-05,
      "learning_rate": 0.00019999755720551262,
      "loss": 23.0,
      "step": 67
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 6.185177335282788e-05,
      "learning_rate": 0.00019999742337877793,
      "loss": 23.0,
      "step": 68
    },
    {
      "epoch": 0.04141656662665066,
      "grad_norm": 5.33986822119914e-05,
      "learning_rate": 0.00019999728598339247,
      "loss": 23.0,
      "step": 69
    },
    {
      "epoch": 0.04201680672268908,
      "grad_norm": 9.752002370078117e-05,
      "learning_rate": 0.00019999714501936104,
      "loss": 23.0,
      "step": 70
    },
    {
      "epoch": 0.042617046818727494,
      "grad_norm": 0.0001452133001293987,
      "learning_rate": 0.00019999700048668875,
      "loss": 23.0,
      "step": 71
    },
    {
      "epoch": 0.04321728691476591,
      "grad_norm": 6.744873826391995e-05,
      "learning_rate": 0.00019999685238538073,
      "loss": 23.0,
      "step": 72
    },
    {
      "epoch": 0.043817527010804325,
      "grad_norm": 9.654606401454657e-05,
      "learning_rate": 0.00019999670071544227,
      "loss": 23.0,
      "step": 73
    },
    {
      "epoch": 0.04441776710684274,
      "grad_norm": 7.846657535992563e-05,
      "learning_rate": 0.0001999965454768788,
      "loss": 23.0,
      "step": 74
    },
    {
      "epoch": 0.045018007202881155,
      "grad_norm": 8.308366523124278e-05,
      "learning_rate": 0.00019999638666969583,
      "loss": 23.0,
      "step": 75
    },
    {
      "epoch": 0.04561824729891957,
      "grad_norm": 9.814684744924307e-05,
      "learning_rate": 0.00019999622429389905,
      "loss": 23.0,
      "step": 76
    },
    {
      "epoch": 0.046218487394957986,
      "grad_norm": 9.901592420646921e-05,
      "learning_rate": 0.0001999960583494942,
      "loss": 23.0,
      "step": 77
    },
    {
      "epoch": 0.0468187274909964,
      "grad_norm": 5.98341939621605e-05,
      "learning_rate": 0.00019999588883648733,
      "loss": 23.0,
      "step": 78
    },
    {
      "epoch": 0.047418967587034816,
      "grad_norm": 7.498155173379928e-05,
      "learning_rate": 0.00019999571575488433,
      "loss": 23.0,
      "step": 79
    },
    {
      "epoch": 0.04801920768307323,
      "grad_norm": 0.00012995477300137281,
      "learning_rate": 0.00019999553910469148,
      "loss": 23.0,
      "step": 80
    },
    {
      "epoch": 0.048619447779111646,
      "grad_norm": 7.154367631301284e-05,
      "learning_rate": 0.00019999535888591508,
      "loss": 23.0,
      "step": 81
    },
    {
      "epoch": 0.04921968787515006,
      "grad_norm": 6.733864574925974e-05,
      "learning_rate": 0.00019999517509856153,
      "loss": 23.0,
      "step": 82
    },
    {
      "epoch": 0.04981992797118848,
      "grad_norm": 6.913344986969605e-05,
      "learning_rate": 0.00019999498774263738,
      "loss": 23.0,
      "step": 83
    },
    {
      "epoch": 0.05042016806722689,
      "grad_norm": 7.432701386278495e-05,
      "learning_rate": 0.00019999479681814935,
      "loss": 23.0,
      "step": 84
    },
    {
      "epoch": 0.05102040816326531,
      "grad_norm": 0.00011996757530141622,
      "learning_rate": 0.00019999460232510426,
      "loss": 23.0,
      "step": 85
    },
    {
      "epoch": 0.05162064825930372,
      "grad_norm": 5.151230652700178e-05,
      "learning_rate": 0.000199994404263509,
      "loss": 23.0,
      "step": 86
    },
    {
      "epoch": 0.05222088835534214,
      "grad_norm": 4.7111861931625754e-05,
      "learning_rate": 0.00019999420263337066,
      "loss": 23.0,
      "step": 87
    },
    {
      "epoch": 0.05282112845138055,
      "grad_norm": 0.00010227777238469571,
      "learning_rate": 0.00019999399743469648,
      "loss": 23.0,
      "step": 88
    },
    {
      "epoch": 0.05342136854741897,
      "grad_norm": 6.315651990007609e-05,
      "learning_rate": 0.00019999378866749373,
      "loss": 23.0,
      "step": 89
    },
    {
      "epoch": 0.05402160864345738,
      "grad_norm": 6.648873386438936e-05,
      "learning_rate": 0.00019999357633176986,
      "loss": 23.0,
      "step": 90
    },
    {
      "epoch": 0.0546218487394958,
      "grad_norm": 0.00017616510740481317,
      "learning_rate": 0.0001999933604275325,
      "loss": 23.0,
      "step": 91
    },
    {
      "epoch": 0.055222088835534214,
      "grad_norm": 0.00013492346624843776,
      "learning_rate": 0.0001999931409547893,
      "loss": 23.0,
      "step": 92
    },
    {
      "epoch": 0.05582232893157263,
      "grad_norm": 0.00010630957694957033,
      "learning_rate": 0.00019999291791354811,
      "loss": 23.0,
      "step": 93
    },
    {
      "epoch": 0.056422569027611044,
      "grad_norm": 0.00013615177886094898,
      "learning_rate": 0.0001999926913038169,
      "loss": 23.0,
      "step": 94
    },
    {
      "epoch": 0.05702280912364946,
      "grad_norm": 0.00013460556510835886,
      "learning_rate": 0.00019999246112560376,
      "loss": 23.0,
      "step": 95
    },
    {
      "epoch": 0.057623049219687875,
      "grad_norm": 0.00021354878845158964,
      "learning_rate": 0.00019999222737891685,
      "loss": 23.0,
      "step": 96
    },
    {
      "epoch": 0.05822328931572629,
      "grad_norm": 0.00010901458881562576,
      "learning_rate": 0.0001999919900637646,
      "loss": 23.0,
      "step": 97
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 0.00015199890185613185,
      "learning_rate": 0.00019999174918015542,
      "loss": 23.0,
      "step": 98
    },
    {
      "epoch": 0.05942376950780312,
      "grad_norm": 0.00010835780994966626,
      "learning_rate": 0.0001999915047280979,
      "loss": 23.0,
      "step": 99
    },
    {
      "epoch": 0.060024009603841535,
      "grad_norm": 0.0001574648340465501,
      "learning_rate": 0.00019999125670760082,
      "loss": 23.0,
      "step": 100
    },
    {
      "epoch": 0.06062424969987995,
      "grad_norm": 9.17216602829285e-05,
      "learning_rate": 0.00019999100511867296,
      "loss": 23.0,
      "step": 101
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 0.00016906972450669855,
      "learning_rate": 0.00019999074996132336,
      "loss": 23.0,
      "step": 102
    },
    {
      "epoch": 0.06182472989195678,
      "grad_norm": 0.00029384472873061895,
      "learning_rate": 0.00019999049123556111,
      "loss": 23.0,
      "step": 103
    },
    {
      "epoch": 0.062424969987995196,
      "grad_norm": 0.0001368739758618176,
      "learning_rate": 0.0001999902289413954,
      "loss": 23.0,
      "step": 104
    },
    {
      "epoch": 0.06302521008403361,
      "grad_norm": 6.585185474250466e-05,
      "learning_rate": 0.00019998996307883565,
      "loss": 23.0,
      "step": 105
    },
    {
      "epoch": 0.06362545018007203,
      "grad_norm": 0.0002683665370568633,
      "learning_rate": 0.00019998969364789132,
      "loss": 23.0,
      "step": 106
    },
    {
      "epoch": 0.06422569027611044,
      "grad_norm": 0.0001274021778954193,
      "learning_rate": 0.00019998942064857202,
      "loss": 23.0,
      "step": 107
    },
    {
      "epoch": 0.06482593037214886,
      "grad_norm": 0.00012011367653030902,
      "learning_rate": 0.00019998914408088752,
      "loss": 23.0,
      "step": 108
    },
    {
      "epoch": 0.06542617046818727,
      "grad_norm": 0.00017174589447677135,
      "learning_rate": 0.00019998886394484764,
      "loss": 23.0,
      "step": 109
    },
    {
      "epoch": 0.06602641056422569,
      "grad_norm": 0.00015978926967363805,
      "learning_rate": 0.00019998858024046243,
      "loss": 23.0,
      "step": 110
    },
    {
      "epoch": 0.0666266506602641,
      "grad_norm": 0.00011685083882184699,
      "learning_rate": 0.000199988292967742,
      "loss": 23.0,
      "step": 111
    },
    {
      "epoch": 0.06722689075630252,
      "grad_norm": 0.00012352668272797018,
      "learning_rate": 0.00019998800212669656,
      "loss": 23.0,
      "step": 112
    },
    {
      "epoch": 0.06782713085234093,
      "grad_norm": 0.00013859743194188923,
      "learning_rate": 0.00019998770771733658,
      "loss": 23.0,
      "step": 113
    },
    {
      "epoch": 0.06842737094837935,
      "grad_norm": 0.00013077691255602986,
      "learning_rate": 0.00019998740973967244,
      "loss": 23.0,
      "step": 114
    },
    {
      "epoch": 0.06902761104441776,
      "grad_norm": 0.00024535699049010873,
      "learning_rate": 0.0001999871081937149,
      "loss": 23.0,
      "step": 115
    },
    {
      "epoch": 0.06962785114045618,
      "grad_norm": 0.00018474037642590702,
      "learning_rate": 0.00019998680307947464,
      "loss": 23.0,
      "step": 116
    },
    {
      "epoch": 0.0702280912364946,
      "grad_norm": 0.0001755363482516259,
      "learning_rate": 0.00019998649439696258,
      "loss": 23.0,
      "step": 117
    },
    {
      "epoch": 0.07082833133253301,
      "grad_norm": 0.00021930545335635543,
      "learning_rate": 0.00019998618214618974,
      "loss": 23.0,
      "step": 118
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.00014808699779678136,
      "learning_rate": 0.00019998586632716725,
      "loss": 23.0,
      "step": 119
    },
    {
      "epoch": 0.07202881152460984,
      "grad_norm": 0.00011154917592648417,
      "learning_rate": 0.0001999855469399064,
      "loss": 23.0,
      "step": 120
    },
    {
      "epoch": 0.07262905162064826,
      "grad_norm": 0.00016951010911725461,
      "learning_rate": 0.00019998522398441851,
      "loss": 23.0,
      "step": 121
    },
    {
      "epoch": 0.07322929171668667,
      "grad_norm": 0.0003669381549116224,
      "learning_rate": 0.00019998489746071526,
      "loss": 23.0,
      "step": 122
    },
    {
      "epoch": 0.07382953181272509,
      "grad_norm": 0.00013156913337297738,
      "learning_rate": 0.00019998456736880816,
      "loss": 23.0,
      "step": 123
    },
    {
      "epoch": 0.0744297719087635,
      "grad_norm": 0.0001488238776801154,
      "learning_rate": 0.00019998423370870902,
      "loss": 23.0,
      "step": 124
    },
    {
      "epoch": 0.07503001200480192,
      "grad_norm": 0.00010793282854137942,
      "learning_rate": 0.00019998389648042978,
      "loss": 23.0,
      "step": 125
    },
    {
      "epoch": 0.07563025210084033,
      "grad_norm": 0.00041736027924343944,
      "learning_rate": 0.0001999835556839825,
      "loss": 23.0,
      "step": 126
    },
    {
      "epoch": 0.07623049219687875,
      "grad_norm": 0.00013802212197333574,
      "learning_rate": 0.0001999832113193793,
      "loss": 23.0,
      "step": 127
    },
    {
      "epoch": 0.07683073229291716,
      "grad_norm": 0.00014316567103378475,
      "learning_rate": 0.00019998286338663242,
      "loss": 23.0,
      "step": 128
    },
    {
      "epoch": 0.07743097238895558,
      "grad_norm": 0.0005471606855280697,
      "learning_rate": 0.0001999825118857544,
      "loss": 23.0,
      "step": 129
    },
    {
      "epoch": 0.07803121248499399,
      "grad_norm": 0.00032816806924529374,
      "learning_rate": 0.00019998215681675768,
      "loss": 23.0,
      "step": 130
    },
    {
      "epoch": 0.07863145258103241,
      "grad_norm": 0.0002986426989082247,
      "learning_rate": 0.000199981798179655,
      "loss": 23.0,
      "step": 131
    },
    {
      "epoch": 0.07923169267707082,
      "grad_norm": 0.0002443274424877018,
      "learning_rate": 0.0001999814359744591,
      "loss": 23.0,
      "step": 132
    },
    {
      "epoch": 0.07983193277310924,
      "grad_norm": 0.00016547096311114728,
      "learning_rate": 0.00019998107020118293,
      "loss": 23.0,
      "step": 133
    },
    {
      "epoch": 0.08043217286914765,
      "grad_norm": 0.0002686209627427161,
      "learning_rate": 0.00019998070085983956,
      "loss": 23.0,
      "step": 134
    },
    {
      "epoch": 0.08103241296518607,
      "grad_norm": 0.00021909335919190198,
      "learning_rate": 0.00019998032795044216,
      "loss": 23.0,
      "step": 135
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.00032937468495219946,
      "learning_rate": 0.00019997995147300404,
      "loss": 23.0,
      "step": 136
    },
    {
      "epoch": 0.0822328931572629,
      "grad_norm": 0.0002638402802404016,
      "learning_rate": 0.0001999795714275386,
      "loss": 23.0,
      "step": 137
    },
    {
      "epoch": 0.08283313325330131,
      "grad_norm": 0.0002468010934535414,
      "learning_rate": 0.00019997918781405947,
      "loss": 23.0,
      "step": 138
    },
    {
      "epoch": 0.08343337334933974,
      "grad_norm": 0.0004808467347174883,
      "learning_rate": 0.00019997880063258027,
      "loss": 23.0,
      "step": 139
    },
    {
      "epoch": 0.08403361344537816,
      "grad_norm": 0.0004602950648404658,
      "learning_rate": 0.00019997840988311488,
      "loss": 23.0,
      "step": 140
    },
    {
      "epoch": 0.08463385354141657,
      "grad_norm": 0.0004139028023928404,
      "learning_rate": 0.00019997801556567718,
      "loss": 23.0,
      "step": 141
    },
    {
      "epoch": 0.08523409363745499,
      "grad_norm": 0.0002881626132875681,
      "learning_rate": 0.00019997761768028133,
      "loss": 23.0,
      "step": 142
    },
    {
      "epoch": 0.0858343337334934,
      "grad_norm": 0.00026587379397824407,
      "learning_rate": 0.00019997721622694146,
      "loss": 23.0,
      "step": 143
    },
    {
      "epoch": 0.08643457382953182,
      "grad_norm": 0.00015115221322048455,
      "learning_rate": 0.0001999768112056719,
      "loss": 23.0,
      "step": 144
    },
    {
      "epoch": 0.08703481392557023,
      "grad_norm": 0.0003756193327717483,
      "learning_rate": 0.00019997640261648712,
      "loss": 23.0,
      "step": 145
    },
    {
      "epoch": 0.08763505402160865,
      "grad_norm": 0.0002508624747861177,
      "learning_rate": 0.0001999759904594017,
      "loss": 23.0,
      "step": 146
    },
    {
      "epoch": 0.08823529411764706,
      "grad_norm": 0.0003372096107341349,
      "learning_rate": 0.00019997557473443037,
      "loss": 23.0,
      "step": 147
    },
    {
      "epoch": 0.08883553421368548,
      "grad_norm": 0.00023483167751692235,
      "learning_rate": 0.0001999751554415879,
      "loss": 23.0,
      "step": 148
    },
    {
      "epoch": 0.0894357743097239,
      "grad_norm": 0.00021300539083313197,
      "learning_rate": 0.00019997473258088935,
      "loss": 23.0,
      "step": 149
    },
    {
      "epoch": 0.09003601440576231,
      "grad_norm": 0.0002520008129067719,
      "learning_rate": 0.00019997430615234976,
      "loss": 23.0,
      "step": 150
    },
    {
      "epoch": 0.09063625450180073,
      "grad_norm": 0.0002747656835708767,
      "learning_rate": 0.00019997387615598432,
      "loss": 23.0,
      "step": 151
    },
    {
      "epoch": 0.09123649459783914,
      "grad_norm": 0.00019271083874627948,
      "learning_rate": 0.0001999734425918084,
      "loss": 23.0,
      "step": 152
    },
    {
      "epoch": 0.09183673469387756,
      "grad_norm": 0.00039355031913146377,
      "learning_rate": 0.0001999730054598375,
      "loss": 23.0,
      "step": 153
    },
    {
      "epoch": 0.09243697478991597,
      "grad_norm": 0.000462890398921445,
      "learning_rate": 0.0001999725647600872,
      "loss": 23.0,
      "step": 154
    },
    {
      "epoch": 0.09303721488595439,
      "grad_norm": 0.0004298924177419394,
      "learning_rate": 0.00019997212049257317,
      "loss": 23.0,
      "step": 155
    },
    {
      "epoch": 0.0936374549819928,
      "grad_norm": 0.00041610770858824253,
      "learning_rate": 0.00019997167265731135,
      "loss": 23.0,
      "step": 156
    },
    {
      "epoch": 0.09423769507803122,
      "grad_norm": 0.00030013787909410894,
      "learning_rate": 0.0001999712212543177,
      "loss": 23.0,
      "step": 157
    },
    {
      "epoch": 0.09483793517406963,
      "grad_norm": 0.00035850016865879297,
      "learning_rate": 0.00019997076628360834,
      "loss": 23.0,
      "step": 158
    },
    {
      "epoch": 0.09543817527010805,
      "grad_norm": 0.00044739991426467896,
      "learning_rate": 0.00019997030774519944,
      "loss": 23.0,
      "step": 159
    },
    {
      "epoch": 0.09603841536614646,
      "grad_norm": 0.0003083112242165953,
      "learning_rate": 0.00019996984563910747,
      "loss": 23.0,
      "step": 160
    },
    {
      "epoch": 0.09663865546218488,
      "grad_norm": 0.0003727200091816485,
      "learning_rate": 0.00019996937996534883,
      "loss": 23.0,
      "step": 161
    },
    {
      "epoch": 0.09723889555822329,
      "grad_norm": 0.00040038779843598604,
      "learning_rate": 0.00019996891072394016,
      "loss": 23.0,
      "step": 162
    },
    {
      "epoch": 0.09783913565426171,
      "grad_norm": 0.000661612895783037,
      "learning_rate": 0.00019996843791489824,
      "loss": 23.0,
      "step": 163
    },
    {
      "epoch": 0.09843937575030012,
      "grad_norm": 0.00018554573762230575,
      "learning_rate": 0.00019996796153823988,
      "loss": 23.0,
      "step": 164
    },
    {
      "epoch": 0.09903961584633854,
      "grad_norm": 0.0003709488082677126,
      "learning_rate": 0.00019996748159398215,
      "loss": 23.0,
      "step": 165
    },
    {
      "epoch": 0.09963985594237695,
      "grad_norm": 0.00026602845173329115,
      "learning_rate": 0.00019996699808214215,
      "loss": 23.0,
      "step": 166
    },
    {
      "epoch": 0.10024009603841537,
      "grad_norm": 0.0003050805244129151,
      "learning_rate": 0.00019996651100273713,
      "loss": 23.0,
      "step": 167
    },
    {
      "epoch": 0.10084033613445378,
      "grad_norm": 0.0006026671035215259,
      "learning_rate": 0.00019996602035578448,
      "loss": 23.0,
      "step": 168
    },
    {
      "epoch": 0.1014405762304922,
      "grad_norm": 0.00023183878511190414,
      "learning_rate": 0.0001999655261413017,
      "loss": 23.0,
      "step": 169
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 0.00046876040869392455,
      "learning_rate": 0.00019996502835930644,
      "loss": 23.0,
      "step": 170
    },
    {
      "epoch": 0.10264105642256903,
      "grad_norm": 0.00031560304341837764,
      "learning_rate": 0.00019996452700981644,
      "loss": 23.0,
      "step": 171
    },
    {
      "epoch": 0.10324129651860744,
      "grad_norm": 0.0005790601135231555,
      "learning_rate": 0.00019996402209284964,
      "loss": 23.0,
      "step": 172
    },
    {
      "epoch": 0.10384153661464586,
      "grad_norm": 0.00031595080508850515,
      "learning_rate": 0.00019996351360842402,
      "loss": 23.0,
      "step": 173
    },
    {
      "epoch": 0.10444177671068428,
      "grad_norm": 0.0004917896003462374,
      "learning_rate": 0.00019996300155655775,
      "loss": 23.0,
      "step": 174
    },
    {
      "epoch": 0.10504201680672269,
      "grad_norm": 0.00028941556229256094,
      "learning_rate": 0.00019996248593726908,
      "loss": 23.0,
      "step": 175
    },
    {
      "epoch": 0.1056422569027611,
      "grad_norm": 0.0005954919033683836,
      "learning_rate": 0.0001999619667505764,
      "loss": 23.0,
      "step": 176
    },
    {
      "epoch": 0.10624249699879952,
      "grad_norm": 0.0004171891196165234,
      "learning_rate": 0.00019996144399649825,
      "loss": 23.0,
      "step": 177
    },
    {
      "epoch": 0.10684273709483794,
      "grad_norm": 0.00026635677204467356,
      "learning_rate": 0.00019996091767505333,
      "loss": 23.0,
      "step": 178
    },
    {
      "epoch": 0.10744297719087635,
      "grad_norm": 0.00038086227141320705,
      "learning_rate": 0.00019996038778626038,
      "loss": 23.0,
      "step": 179
    },
    {
      "epoch": 0.10804321728691477,
      "grad_norm": 0.0007933559827506542,
      "learning_rate": 0.00019995985433013833,
      "loss": 23.0,
      "step": 180
    },
    {
      "epoch": 0.10864345738295318,
      "grad_norm": 0.00028796150581911206,
      "learning_rate": 0.00019995931730670618,
      "loss": 23.0,
      "step": 181
    },
    {
      "epoch": 0.1092436974789916,
      "grad_norm": 0.0004604592395480722,
      "learning_rate": 0.00019995877671598314,
      "loss": 23.0,
      "step": 182
    },
    {
      "epoch": 0.10984393757503001,
      "grad_norm": 0.0002964911109302193,
      "learning_rate": 0.00019995823255798846,
      "loss": 23.0,
      "step": 183
    },
    {
      "epoch": 0.11044417767106843,
      "grad_norm": 0.00046794061199761927,
      "learning_rate": 0.00019995768483274163,
      "loss": 23.0,
      "step": 184
    },
    {
      "epoch": 0.11104441776710684,
      "grad_norm": 0.00039684621151536703,
      "learning_rate": 0.0001999571335402621,
      "loss": 23.0,
      "step": 185
    },
    {
      "epoch": 0.11164465786314526,
      "grad_norm": 0.0006988070090301335,
      "learning_rate": 0.00019995657868056963,
      "loss": 23.0,
      "step": 186
    },
    {
      "epoch": 0.11224489795918367,
      "grad_norm": 0.000538469001185149,
      "learning_rate": 0.00019995602025368395,
      "loss": 23.0,
      "step": 187
    },
    {
      "epoch": 0.11284513805522209,
      "grad_norm": 0.0002729691914282739,
      "learning_rate": 0.00019995545825962508,
      "loss": 23.0,
      "step": 188
    },
    {
      "epoch": 0.1134453781512605,
      "grad_norm": 0.00039077899418771267,
      "learning_rate": 0.00019995489269841296,
      "loss": 23.0,
      "step": 189
    },
    {
      "epoch": 0.11404561824729892,
      "grad_norm": 0.0004048171394970268,
      "learning_rate": 0.00019995432357006786,
      "loss": 23.0,
      "step": 190
    },
    {
      "epoch": 0.11464585834333733,
      "grad_norm": 0.0005215174751356244,
      "learning_rate": 0.00019995375087461005,
      "loss": 23.0,
      "step": 191
    },
    {
      "epoch": 0.11524609843937575,
      "grad_norm": 0.0004714977403637022,
      "learning_rate": 0.00019995317461206003,
      "loss": 23.0,
      "step": 192
    },
    {
      "epoch": 0.11584633853541416,
      "grad_norm": 0.0005683483323082328,
      "learning_rate": 0.0001999525947824383,
      "loss": 23.0,
      "step": 193
    },
    {
      "epoch": 0.11644657863145258,
      "grad_norm": 0.00026816027821041644,
      "learning_rate": 0.00019995201138576553,
      "loss": 23.0,
      "step": 194
    },
    {
      "epoch": 0.117046818727491,
      "grad_norm": 0.00048822627286426723,
      "learning_rate": 0.0001999514244220626,
      "loss": 23.0,
      "step": 195
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.0006117990124039352,
      "learning_rate": 0.00019995083389135046,
      "loss": 23.0,
      "step": 196
    },
    {
      "epoch": 0.11824729891956783,
      "grad_norm": 0.0009073524852283299,
      "learning_rate": 0.00019995023979365014,
      "loss": 23.0,
      "step": 197
    },
    {
      "epoch": 0.11884753901560624,
      "grad_norm": 0.0006450892542488873,
      "learning_rate": 0.00019994964212898287,
      "loss": 23.0,
      "step": 198
    },
    {
      "epoch": 0.11944777911164466,
      "grad_norm": 0.0005826326669193804,
      "learning_rate": 0.00019994904089737,
      "loss": 23.0,
      "step": 199
    },
    {
      "epoch": 0.12004801920768307,
      "grad_norm": 0.0007119604852050543,
      "learning_rate": 0.00019994843609883294,
      "loss": 23.0,
      "step": 200
    },
    {
      "epoch": 0.12064825930372149,
      "grad_norm": 0.0005299604963511229,
      "learning_rate": 0.0001999478277333933,
      "loss": 23.0,
      "step": 201
    },
    {
      "epoch": 0.1212484993997599,
      "grad_norm": 0.0006639006896875799,
      "learning_rate": 0.00019994721580107277,
      "loss": 23.0,
      "step": 202
    },
    {
      "epoch": 0.12184873949579832,
      "grad_norm": 0.0006667767884209752,
      "learning_rate": 0.0001999466003018932,
      "loss": 23.0,
      "step": 203
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 0.00042130585643462837,
      "learning_rate": 0.00019994598123587655,
      "loss": 23.0,
      "step": 204
    },
    {
      "epoch": 0.12304921968787515,
      "grad_norm": 0.0011373097077012062,
      "learning_rate": 0.00019994535860304495,
      "loss": 23.0,
      "step": 205
    },
    {
      "epoch": 0.12364945978391356,
      "grad_norm": 0.0016360898735001683,
      "learning_rate": 0.0001999447324034206,
      "loss": 23.0,
      "step": 206
    },
    {
      "epoch": 0.12424969987995198,
      "grad_norm": 0.0004542885872069746,
      "learning_rate": 0.0001999441026370258,
      "loss": 23.0,
      "step": 207
    },
    {
      "epoch": 0.12484993997599039,
      "grad_norm": 0.0004969018045812845,
      "learning_rate": 0.00019994346930388312,
      "loss": 23.0,
      "step": 208
    },
    {
      "epoch": 0.12545018007202882,
      "grad_norm": 0.000546945258975029,
      "learning_rate": 0.00019994283240401504,
      "loss": 23.0,
      "step": 209
    },
    {
      "epoch": 0.12605042016806722,
      "grad_norm": 0.0007595718489028513,
      "learning_rate": 0.0001999421919374444,
      "loss": 23.0,
      "step": 210
    },
    {
      "epoch": 0.12665066026410565,
      "grad_norm": 0.000410681328503415,
      "learning_rate": 0.000199941547904194,
      "loss": 23.0,
      "step": 211
    },
    {
      "epoch": 0.12725090036014405,
      "grad_norm": 0.0005949096521362662,
      "learning_rate": 0.00019994090030428684,
      "loss": 23.0,
      "step": 212
    },
    {
      "epoch": 0.12785114045618248,
      "grad_norm": 0.0006973805138841271,
      "learning_rate": 0.000199940249137746,
      "loss": 23.0,
      "step": 213
    },
    {
      "epoch": 0.12845138055222088,
      "grad_norm": 0.0003782823623623699,
      "learning_rate": 0.00019993959440459474,
      "loss": 23.0,
      "step": 214
    },
    {
      "epoch": 0.1290516206482593,
      "grad_norm": 0.0008317967294715345,
      "learning_rate": 0.00019993893610485648,
      "loss": 23.0,
      "step": 215
    },
    {
      "epoch": 0.12965186074429771,
      "grad_norm": 0.000514440587721765,
      "learning_rate": 0.00019993827423855464,
      "loss": 23.0,
      "step": 216
    },
    {
      "epoch": 0.13025210084033614,
      "grad_norm": 0.0004617723752744496,
      "learning_rate": 0.00019993760880571287,
      "loss": 23.0,
      "step": 217
    },
    {
      "epoch": 0.13085234093637454,
      "grad_norm": 0.0009777372470125556,
      "learning_rate": 0.0001999369398063549,
      "loss": 23.0,
      "step": 218
    },
    {
      "epoch": 0.13145258103241297,
      "grad_norm": 0.0005835364572703838,
      "learning_rate": 0.00019993626724050463,
      "loss": 23.0,
      "step": 219
    },
    {
      "epoch": 0.13205282112845138,
      "grad_norm": 0.0005018878728151321,
      "learning_rate": 0.00019993559110818602,
      "loss": 23.0,
      "step": 220
    },
    {
      "epoch": 0.1326530612244898,
      "grad_norm": 0.0006578268366865814,
      "learning_rate": 0.00019993491140942327,
      "loss": 23.0,
      "step": 221
    },
    {
      "epoch": 0.1332533013205282,
      "grad_norm": 0.0007726168259978294,
      "learning_rate": 0.0001999342281442406,
      "loss": 23.0,
      "step": 222
    },
    {
      "epoch": 0.13385354141656663,
      "grad_norm": 0.0008200584561564028,
      "learning_rate": 0.00019993354131266232,
      "loss": 23.0,
      "step": 223
    },
    {
      "epoch": 0.13445378151260504,
      "grad_norm": 0.0005691193509846926,
      "learning_rate": 0.00019993285091471306,
      "loss": 23.0,
      "step": 224
    },
    {
      "epoch": 0.13505402160864347,
      "grad_norm": 0.0006812798674218357,
      "learning_rate": 0.00019993215695041742,
      "loss": 23.0,
      "step": 225
    },
    {
      "epoch": 0.13565426170468187,
      "grad_norm": 0.0006009634817019105,
      "learning_rate": 0.00019993145941980012,
      "loss": 23.0,
      "step": 226
    },
    {
      "epoch": 0.1362545018007203,
      "grad_norm": 0.0005906272563152015,
      "learning_rate": 0.0001999307583228861,
      "loss": 23.0,
      "step": 227
    },
    {
      "epoch": 0.1368547418967587,
      "grad_norm": 0.0009002196020446718,
      "learning_rate": 0.0001999300536597004,
      "loss": 23.0,
      "step": 228
    },
    {
      "epoch": 0.13745498199279713,
      "grad_norm": 0.0007381000905297697,
      "learning_rate": 0.0001999293454302681,
      "loss": 23.0,
      "step": 229
    },
    {
      "epoch": 0.13805522208883553,
      "grad_norm": 0.0004923134110867977,
      "learning_rate": 0.00019992863363461455,
      "loss": 23.0,
      "step": 230
    },
    {
      "epoch": 0.13865546218487396,
      "grad_norm": 0.0006840071291662753,
      "learning_rate": 0.00019992791827276504,
      "loss": 23.0,
      "step": 231
    },
    {
      "epoch": 0.13925570228091236,
      "grad_norm": 0.0008459037635475397,
      "learning_rate": 0.00019992719934474522,
      "loss": 23.0,
      "step": 232
    },
    {
      "epoch": 0.1398559423769508,
      "grad_norm": 0.0010143484687432647,
      "learning_rate": 0.0001999264768505807,
      "loss": 23.0,
      "step": 233
    },
    {
      "epoch": 0.1404561824729892,
      "grad_norm": 0.00038011017022654414,
      "learning_rate": 0.00019992575079029728,
      "loss": 23.0,
      "step": 234
    },
    {
      "epoch": 0.14105642256902762,
      "grad_norm": 0.0005543521838262677,
      "learning_rate": 0.0001999250211639208,
      "loss": 23.0,
      "step": 235
    },
    {
      "epoch": 0.14165666266506602,
      "grad_norm": 0.00035856247995980084,
      "learning_rate": 0.00019992428797147737,
      "loss": 23.0,
      "step": 236
    },
    {
      "epoch": 0.14225690276110445,
      "grad_norm": 0.0007775566191412508,
      "learning_rate": 0.00019992355121299318,
      "loss": 23.0,
      "step": 237
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.0007267727633006871,
      "learning_rate": 0.00019992281088849445,
      "loss": 23.0,
      "step": 238
    },
    {
      "epoch": 0.14345738295318128,
      "grad_norm": 0.0004261432040948421,
      "learning_rate": 0.00019992206699800762,
      "loss": 23.0,
      "step": 239
    },
    {
      "epoch": 0.14405762304921968,
      "grad_norm": 0.0008677646401338279,
      "learning_rate": 0.0001999213195415593,
      "loss": 23.0,
      "step": 240
    },
    {
      "epoch": 0.1446578631452581,
      "grad_norm": 0.0005699426401406527,
      "learning_rate": 0.00019992056851917607,
      "loss": 23.0,
      "step": 241
    },
    {
      "epoch": 0.1452581032412965,
      "grad_norm": 0.0007082479423843324,
      "learning_rate": 0.00019991981393088475,
      "loss": 23.0,
      "step": 242
    },
    {
      "epoch": 0.14585834333733494,
      "grad_norm": 0.000806904979981482,
      "learning_rate": 0.00019991905577671233,
      "loss": 23.0,
      "step": 243
    },
    {
      "epoch": 0.14645858343337334,
      "grad_norm": 0.0006296361098065972,
      "learning_rate": 0.00019991829405668584,
      "loss": 23.0,
      "step": 244
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 0.0009300713427364826,
      "learning_rate": 0.00019991752877083246,
      "loss": 23.0,
      "step": 245
    },
    {
      "epoch": 0.14765906362545017,
      "grad_norm": 0.0010097339982166886,
      "learning_rate": 0.00019991675991917946,
      "loss": 23.0,
      "step": 246
    },
    {
      "epoch": 0.1482593037214886,
      "grad_norm": 0.0007776516722515225,
      "learning_rate": 0.00019991598750175436,
      "loss": 23.0,
      "step": 247
    },
    {
      "epoch": 0.148859543817527,
      "grad_norm": 0.00047628002357669175,
      "learning_rate": 0.00019991521151858463,
      "loss": 23.0,
      "step": 248
    },
    {
      "epoch": 0.14945978391356543,
      "grad_norm": 0.00046940191532485187,
      "learning_rate": 0.00019991443196969806,
      "loss": 23.0,
      "step": 249
    },
    {
      "epoch": 0.15006002400960383,
      "grad_norm": 0.0007664930308237672,
      "learning_rate": 0.0001999136488551224,
      "loss": 23.0,
      "step": 250
    },
    {
      "epoch": 0.15066026410564226,
      "grad_norm": 0.000685586070176214,
      "learning_rate": 0.00019991286217488564,
      "loss": 23.0,
      "step": 251
    },
    {
      "epoch": 0.15126050420168066,
      "grad_norm": 0.0010549027938395739,
      "learning_rate": 0.0001999120719290158,
      "loss": 23.0,
      "step": 252
    },
    {
      "epoch": 0.1518607442977191,
      "grad_norm": 0.0005077957175672054,
      "learning_rate": 0.00019991127811754114,
      "loss": 23.0,
      "step": 253
    },
    {
      "epoch": 0.1524609843937575,
      "grad_norm": 0.0005034700152464211,
      "learning_rate": 0.00019991048074048998,
      "loss": 23.0,
      "step": 254
    },
    {
      "epoch": 0.15306122448979592,
      "grad_norm": 0.0007512738811783493,
      "learning_rate": 0.00019990967979789074,
      "loss": 23.0,
      "step": 255
    },
    {
      "epoch": 0.15366146458583432,
      "grad_norm": 0.00046786750317551196,
      "learning_rate": 0.00019990887528977203,
      "loss": 23.0,
      "step": 256
    },
    {
      "epoch": 0.15426170468187275,
      "grad_norm": 0.001064613345079124,
      "learning_rate": 0.00019990806721616257,
      "loss": 23.0,
      "step": 257
    },
    {
      "epoch": 0.15486194477791115,
      "grad_norm": 0.00039759057108312845,
      "learning_rate": 0.00019990725557709118,
      "loss": 23.0,
      "step": 258
    },
    {
      "epoch": 0.15546218487394958,
      "grad_norm": 0.0007065651589073241,
      "learning_rate": 0.0001999064403725868,
      "loss": 23.0,
      "step": 259
    },
    {
      "epoch": 0.15606242496998798,
      "grad_norm": 0.00046774890506640077,
      "learning_rate": 0.0001999056216026786,
      "loss": 23.0,
      "step": 260
    },
    {
      "epoch": 0.1566626650660264,
      "grad_norm": 0.0005046619917266071,
      "learning_rate": 0.00019990479926739575,
      "loss": 23.0,
      "step": 261
    },
    {
      "epoch": 0.15726290516206481,
      "grad_norm": 0.0005047253216616809,
      "learning_rate": 0.00019990397336676759,
      "loss": 23.0,
      "step": 262
    },
    {
      "epoch": 0.15786314525810324,
      "grad_norm": 0.0007587158470414579,
      "learning_rate": 0.00019990314390082358,
      "loss": 23.0,
      "step": 263
    },
    {
      "epoch": 0.15846338535414164,
      "grad_norm": 0.0008176558185368776,
      "learning_rate": 0.00019990231086959337,
      "loss": 23.0,
      "step": 264
    },
    {
      "epoch": 0.15906362545018007,
      "grad_norm": 0.0005377682973630726,
      "learning_rate": 0.00019990147427310665,
      "loss": 23.0,
      "step": 265
    },
    {
      "epoch": 0.15966386554621848,
      "grad_norm": 0.0006355341174639761,
      "learning_rate": 0.0001999006341113933,
      "loss": 23.0,
      "step": 266
    },
    {
      "epoch": 0.1602641056422569,
      "grad_norm": 0.0007314942195080221,
      "learning_rate": 0.0001998997903844833,
      "loss": 23.0,
      "step": 267
    },
    {
      "epoch": 0.1608643457382953,
      "grad_norm": 0.0007350289961323142,
      "learning_rate": 0.00019989894309240674,
      "loss": 23.0,
      "step": 268
    },
    {
      "epoch": 0.16146458583433373,
      "grad_norm": 0.0008466751896776259,
      "learning_rate": 0.00019989809223519388,
      "loss": 23.0,
      "step": 269
    },
    {
      "epoch": 0.16206482593037214,
      "grad_norm": 0.0009016711846925318,
      "learning_rate": 0.0001998972378128751,
      "loss": 23.0,
      "step": 270
    },
    {
      "epoch": 0.16266506602641057,
      "grad_norm": 0.0005349046550691128,
      "learning_rate": 0.00019989637982548083,
      "loss": 23.0,
      "step": 271
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.0005445227725431323,
      "learning_rate": 0.0001998955182730417,
      "loss": 23.0,
      "step": 272
    },
    {
      "epoch": 0.1638655462184874,
      "grad_norm": 0.00042219756869599223,
      "learning_rate": 0.00019989465315558856,
      "loss": 23.0,
      "step": 273
    },
    {
      "epoch": 0.1644657863145258,
      "grad_norm": 0.0010625041322782636,
      "learning_rate": 0.00019989378447315216,
      "loss": 23.0,
      "step": 274
    },
    {
      "epoch": 0.16506602641056423,
      "grad_norm": 0.000675600953400135,
      "learning_rate": 0.00019989291222576356,
      "loss": 23.0,
      "step": 275
    },
    {
      "epoch": 0.16566626650660263,
      "grad_norm": 0.0005503667634911835,
      "learning_rate": 0.00019989203641345392,
      "loss": 23.0,
      "step": 276
    },
    {
      "epoch": 0.16626650660264106,
      "grad_norm": 0.0006727102445438504,
      "learning_rate": 0.00019989115703625442,
      "loss": 23.0,
      "step": 277
    },
    {
      "epoch": 0.16686674669867949,
      "grad_norm": 0.0004575340135488659,
      "learning_rate": 0.00019989027409419646,
      "loss": 23.0,
      "step": 278
    },
    {
      "epoch": 0.1674669867947179,
      "grad_norm": 0.0006102299084886909,
      "learning_rate": 0.00019988938758731158,
      "loss": 23.0,
      "step": 279
    },
    {
      "epoch": 0.16806722689075632,
      "grad_norm": 0.0005805498803965747,
      "learning_rate": 0.0001998884975156314,
      "loss": 23.0,
      "step": 280
    },
    {
      "epoch": 0.16866746698679472,
      "grad_norm": 0.0008877067011781037,
      "learning_rate": 0.00019988760387918767,
      "loss": 23.0,
      "step": 281
    },
    {
      "epoch": 0.16926770708283315,
      "grad_norm": 0.0008169253706000745,
      "learning_rate": 0.00019988670667801234,
      "loss": 23.0,
      "step": 282
    },
    {
      "epoch": 0.16986794717887155,
      "grad_norm": 0.0007667009485885501,
      "learning_rate": 0.00019988580591213736,
      "loss": 23.0,
      "step": 283
    },
    {
      "epoch": 0.17046818727490998,
      "grad_norm": 0.0006645764224231243,
      "learning_rate": 0.0001998849015815949,
      "loss": 23.0,
      "step": 284
    },
    {
      "epoch": 0.17106842737094838,
      "grad_norm": 0.0006818636902607977,
      "learning_rate": 0.00019988399368641723,
      "loss": 23.0,
      "step": 285
    },
    {
      "epoch": 0.1716686674669868,
      "grad_norm": 0.0002914422075264156,
      "learning_rate": 0.00019988308222663678,
      "loss": 23.0,
      "step": 286
    },
    {
      "epoch": 0.1722689075630252,
      "grad_norm": 0.0006489845691248775,
      "learning_rate": 0.00019988216720228606,
      "loss": 23.0,
      "step": 287
    },
    {
      "epoch": 0.17286914765906364,
      "grad_norm": 0.0005175631958991289,
      "learning_rate": 0.00019988124861339772,
      "loss": 23.0,
      "step": 288
    },
    {
      "epoch": 0.17346938775510204,
      "grad_norm": 0.000415219459682703,
      "learning_rate": 0.00019988032646000453,
      "loss": 23.0,
      "step": 289
    },
    {
      "epoch": 0.17406962785114047,
      "grad_norm": 0.0004089119902346283,
      "learning_rate": 0.0001998794007421394,
      "loss": 23.0,
      "step": 290
    },
    {
      "epoch": 0.17466986794717887,
      "grad_norm": 0.0007048040861263871,
      "learning_rate": 0.0001998784714598354,
      "loss": 23.0,
      "step": 291
    },
    {
      "epoch": 0.1752701080432173,
      "grad_norm": 0.00052332115592435,
      "learning_rate": 0.00019987753861312567,
      "loss": 23.0,
      "step": 292
    },
    {
      "epoch": 0.1758703481392557,
      "grad_norm": 0.000617658079136163,
      "learning_rate": 0.0001998766022020435,
      "loss": 23.0,
      "step": 293
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 0.0005852694157510996,
      "learning_rate": 0.0001998756622266223,
      "loss": 23.0,
      "step": 294
    },
    {
      "epoch": 0.17707082833133253,
      "grad_norm": 0.000565545866265893,
      "learning_rate": 0.00019987471868689562,
      "loss": 23.0,
      "step": 295
    },
    {
      "epoch": 0.17767106842737096,
      "grad_norm": 0.0007343909819610417,
      "learning_rate": 0.00019987377158289715,
      "loss": 23.0,
      "step": 296
    },
    {
      "epoch": 0.17827130852340936,
      "grad_norm": 0.0007629229803569615,
      "learning_rate": 0.0001998728209146607,
      "loss": 23.0,
      "step": 297
    },
    {
      "epoch": 0.1788715486194478,
      "grad_norm": 0.0007439067121595144,
      "learning_rate": 0.00019987186668222014,
      "loss": 23.0,
      "step": 298
    },
    {
      "epoch": 0.1794717887154862,
      "grad_norm": 0.0007367929792962968,
      "learning_rate": 0.00019987090888560956,
      "loss": 23.0,
      "step": 299
    },
    {
      "epoch": 0.18007202881152462,
      "grad_norm": 0.0017375691095367074,
      "learning_rate": 0.00019986994752486318,
      "loss": 23.0,
      "step": 300
    },
    {
      "epoch": 0.18067226890756302,
      "grad_norm": 0.0009452842641621828,
      "learning_rate": 0.00019986898260001526,
      "loss": 23.0,
      "step": 301
    },
    {
      "epoch": 0.18127250900360145,
      "grad_norm": 0.0005082595744170249,
      "learning_rate": 0.00019986801411110022,
      "loss": 23.0,
      "step": 302
    },
    {
      "epoch": 0.18187274909963985,
      "grad_norm": 0.0008125613094307482,
      "learning_rate": 0.00019986704205815264,
      "loss": 23.0,
      "step": 303
    },
    {
      "epoch": 0.18247298919567828,
      "grad_norm": 0.000651197973638773,
      "learning_rate": 0.00019986606644120723,
      "loss": 23.0,
      "step": 304
    },
    {
      "epoch": 0.18307322929171668,
      "grad_norm": 0.0007686960743740201,
      "learning_rate": 0.0001998650872602988,
      "loss": 23.0,
      "step": 305
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 0.00037045447970740497,
      "learning_rate": 0.0001998641045154623,
      "loss": 23.0,
      "step": 306
    },
    {
      "epoch": 0.1842737094837935,
      "grad_norm": 0.0011275415308773518,
      "learning_rate": 0.00019986311820673278,
      "loss": 23.0,
      "step": 307
    },
    {
      "epoch": 0.18487394957983194,
      "grad_norm": 0.000977221643552184,
      "learning_rate": 0.00019986212833414543,
      "loss": 23.0,
      "step": 308
    },
    {
      "epoch": 0.18547418967587034,
      "grad_norm": 0.0009870711946859956,
      "learning_rate": 0.00019986113489773563,
      "loss": 23.0,
      "step": 309
    },
    {
      "epoch": 0.18607442977190877,
      "grad_norm": 0.0010398519225418568,
      "learning_rate": 0.0001998601378975388,
      "loss": 23.0,
      "step": 310
    },
    {
      "epoch": 0.18667466986794717,
      "grad_norm": 0.0007000160403549671,
      "learning_rate": 0.00019985913733359048,
      "loss": 23.0,
      "step": 311
    },
    {
      "epoch": 0.1872749099639856,
      "grad_norm": 0.0006655632168985903,
      "learning_rate": 0.00019985813320592643,
      "loss": 23.0,
      "step": 312
    },
    {
      "epoch": 0.187875150060024,
      "grad_norm": 0.0005049473256804049,
      "learning_rate": 0.0001998571255145825,
      "loss": 23.0,
      "step": 313
    },
    {
      "epoch": 0.18847539015606243,
      "grad_norm": 0.0008667523507028818,
      "learning_rate": 0.00019985611425959462,
      "loss": 23.0,
      "step": 314
    },
    {
      "epoch": 0.18907563025210083,
      "grad_norm": 0.0008610632503405213,
      "learning_rate": 0.00019985509944099884,
      "loss": 23.0,
      "step": 315
    },
    {
      "epoch": 0.18967587034813926,
      "grad_norm": 0.0006444462342187762,
      "learning_rate": 0.00019985408105883146,
      "loss": 23.0,
      "step": 316
    },
    {
      "epoch": 0.19027611044417767,
      "grad_norm": 0.0011097770184278488,
      "learning_rate": 0.00019985305911312878,
      "loss": 23.0,
      "step": 317
    },
    {
      "epoch": 0.1908763505402161,
      "grad_norm": 0.000943410734180361,
      "learning_rate": 0.00019985203360392727,
      "loss": 23.0,
      "step": 318
    },
    {
      "epoch": 0.1914765906362545,
      "grad_norm": 0.000921803992241621,
      "learning_rate": 0.00019985100453126352,
      "loss": 23.0,
      "step": 319
    },
    {
      "epoch": 0.19207683073229292,
      "grad_norm": 0.000856006343383342,
      "learning_rate": 0.00019984997189517425,
      "loss": 23.0,
      "step": 320
    },
    {
      "epoch": 0.19267707082833133,
      "grad_norm": 0.0008737395983189344,
      "learning_rate": 0.00019984893569569633,
      "loss": 23.0,
      "step": 321
    },
    {
      "epoch": 0.19327731092436976,
      "grad_norm": 0.0007819237653166056,
      "learning_rate": 0.00019984789593286678,
      "loss": 23.0,
      "step": 322
    },
    {
      "epoch": 0.19387755102040816,
      "grad_norm": 0.0008402220555581152,
      "learning_rate": 0.00019984685260672263,
      "loss": 23.0,
      "step": 323
    },
    {
      "epoch": 0.19447779111644659,
      "grad_norm": 0.0003997522871941328,
      "learning_rate": 0.00019984580571730115,
      "loss": 23.0,
      "step": 324
    },
    {
      "epoch": 0.195078031212485,
      "grad_norm": 0.0004815506108570844,
      "learning_rate": 0.0001998447552646397,
      "loss": 23.0,
      "step": 325
    },
    {
      "epoch": 0.19567827130852342,
      "grad_norm": 0.00059786211932078,
      "learning_rate": 0.00019984370124877577,
      "loss": 23.0,
      "step": 326
    },
    {
      "epoch": 0.19627851140456182,
      "grad_norm": 0.0006378625403158367,
      "learning_rate": 0.00019984264366974697,
      "loss": 23.0,
      "step": 327
    },
    {
      "epoch": 0.19687875150060025,
      "grad_norm": 0.00038927397690713406,
      "learning_rate": 0.00019984158252759105,
      "loss": 23.0,
      "step": 328
    },
    {
      "epoch": 0.19747899159663865,
      "grad_norm": 0.0007686428143642843,
      "learning_rate": 0.00019984051782234589,
      "loss": 23.0,
      "step": 329
    },
    {
      "epoch": 0.19807923169267708,
      "grad_norm": 0.0006654906319454312,
      "learning_rate": 0.00019983944955404942,
      "loss": 23.0,
      "step": 330
    },
    {
      "epoch": 0.19867947178871548,
      "grad_norm": 0.00034374112146906555,
      "learning_rate": 0.00019983837772273984,
      "loss": 23.0,
      "step": 331
    },
    {
      "epoch": 0.1992797118847539,
      "grad_norm": 0.00040528926183469594,
      "learning_rate": 0.00019983730232845537,
      "loss": 23.0,
      "step": 332
    },
    {
      "epoch": 0.1998799519807923,
      "grad_norm": 0.0007493866141885519,
      "learning_rate": 0.0001998362233712344,
      "loss": 23.0,
      "step": 333
    },
    {
      "epoch": 0.20048019207683074,
      "grad_norm": 0.0008692885749042034,
      "learning_rate": 0.0001998351408511154,
      "loss": 23.0,
      "step": 334
    },
    {
      "epoch": 0.20108043217286914,
      "grad_norm": 0.0005017250659875572,
      "learning_rate": 0.00019983405476813707,
      "loss": 23.0,
      "step": 335
    },
    {
      "epoch": 0.20168067226890757,
      "grad_norm": 0.0005804816028103232,
      "learning_rate": 0.0001998329651223381,
      "loss": 23.0,
      "step": 336
    },
    {
      "epoch": 0.20228091236494597,
      "grad_norm": 0.0006715598865412176,
      "learning_rate": 0.00019983187191375743,
      "loss": 23.0,
      "step": 337
    },
    {
      "epoch": 0.2028811524609844,
      "grad_norm": 0.00039470859337598085,
      "learning_rate": 0.00019983077514243403,
      "loss": 23.0,
      "step": 338
    },
    {
      "epoch": 0.2034813925570228,
      "grad_norm": 0.0004575586353894323,
      "learning_rate": 0.00019982967480840707,
      "loss": 23.0,
      "step": 339
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 0.0005345721147023141,
      "learning_rate": 0.0001998285709117158,
      "loss": 23.0,
      "step": 340
    },
    {
      "epoch": 0.20468187274909963,
      "grad_norm": 0.0005476151127368212,
      "learning_rate": 0.0001998274634523996,
      "loss": 23.0,
      "step": 341
    },
    {
      "epoch": 0.20528211284513806,
      "grad_norm": 0.0011936496011912823,
      "learning_rate": 0.00019982635243049807,
      "loss": 23.0,
      "step": 342
    },
    {
      "epoch": 0.20588235294117646,
      "grad_norm": 0.0005647265934385359,
      "learning_rate": 0.0001998252378460508,
      "loss": 23.0,
      "step": 343
    },
    {
      "epoch": 0.2064825930372149,
      "grad_norm": 0.000607253925409168,
      "learning_rate": 0.00019982411969909753,
      "loss": 23.0,
      "step": 344
    },
    {
      "epoch": 0.2070828331332533,
      "grad_norm": 0.00046976387966424227,
      "learning_rate": 0.00019982299798967823,
      "loss": 23.0,
      "step": 345
    },
    {
      "epoch": 0.20768307322929172,
      "grad_norm": 0.00045033791684545577,
      "learning_rate": 0.00019982187271783293,
      "loss": 23.0,
      "step": 346
    },
    {
      "epoch": 0.20828331332533012,
      "grad_norm": 0.000813774939160794,
      "learning_rate": 0.0001998207438836017,
      "loss": 23.0,
      "step": 347
    },
    {
      "epoch": 0.20888355342136855,
      "grad_norm": 0.000533080892637372,
      "learning_rate": 0.00019981961148702495,
      "loss": 23.0,
      "step": 348
    },
    {
      "epoch": 0.20948379351740695,
      "grad_norm": 0.000772426079493016,
      "learning_rate": 0.000199818475528143,
      "loss": 23.0,
      "step": 349
    },
    {
      "epoch": 0.21008403361344538,
      "grad_norm": 0.0005271838745102286,
      "learning_rate": 0.00019981733600699645,
      "loss": 23.0,
      "step": 350
    },
    {
      "epoch": 0.21068427370948378,
      "grad_norm": 0.0005881227552890778,
      "learning_rate": 0.00019981619292362593,
      "loss": 23.0,
      "step": 351
    },
    {
      "epoch": 0.2112845138055222,
      "grad_norm": 0.0006613187142647803,
      "learning_rate": 0.00019981504627807227,
      "loss": 23.0,
      "step": 352
    },
    {
      "epoch": 0.2118847539015606,
      "grad_norm": 0.000673049536999315,
      "learning_rate": 0.0001998138960703763,
      "loss": 23.0,
      "step": 353
    },
    {
      "epoch": 0.21248499399759904,
      "grad_norm": 0.0004623003478627652,
      "learning_rate": 0.00019981274230057916,
      "loss": 23.0,
      "step": 354
    },
    {
      "epoch": 0.21308523409363744,
      "grad_norm": 0.0009141325135715306,
      "learning_rate": 0.00019981158496872202,
      "loss": 23.0,
      "step": 355
    },
    {
      "epoch": 0.21368547418967587,
      "grad_norm": 0.0004129754670429975,
      "learning_rate": 0.00019981042407484615,
      "loss": 23.0,
      "step": 356
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.0007131838356144726,
      "learning_rate": 0.00019980925961899295,
      "loss": 23.0,
      "step": 357
    },
    {
      "epoch": 0.2148859543817527,
      "grad_norm": 0.00048736046301200986,
      "learning_rate": 0.00019980809160120405,
      "loss": 23.0,
      "step": 358
    },
    {
      "epoch": 0.2154861944777911,
      "grad_norm": 0.0008709396352060139,
      "learning_rate": 0.0001998069200215211,
      "loss": 23.0,
      "step": 359
    },
    {
      "epoch": 0.21608643457382953,
      "grad_norm": 0.0008411543094553053,
      "learning_rate": 0.0001998057448799859,
      "loss": 23.0,
      "step": 360
    },
    {
      "epoch": 0.21668667466986793,
      "grad_norm": 0.00045339285861700773,
      "learning_rate": 0.0001998045661766404,
      "loss": 23.0,
      "step": 361
    },
    {
      "epoch": 0.21728691476590636,
      "grad_norm": 0.0008114994852803648,
      "learning_rate": 0.00019980338391152662,
      "loss": 23.0,
      "step": 362
    },
    {
      "epoch": 0.21788715486194477,
      "grad_norm": 0.0009748362936079502,
      "learning_rate": 0.00019980219808468682,
      "loss": 23.0,
      "step": 363
    },
    {
      "epoch": 0.2184873949579832,
      "grad_norm": 0.0005413390463218093,
      "learning_rate": 0.0001998010086961633,
      "loss": 23.0,
      "step": 364
    },
    {
      "epoch": 0.2190876350540216,
      "grad_norm": 0.0009059528820216656,
      "learning_rate": 0.00019979981574599852,
      "loss": 23.0,
      "step": 365
    },
    {
      "epoch": 0.21968787515006002,
      "grad_norm": 0.0004444699443411082,
      "learning_rate": 0.000199798619234235,
      "loss": 23.0,
      "step": 366
    },
    {
      "epoch": 0.22028811524609843,
      "grad_norm": 0.0008990266942419112,
      "learning_rate": 0.0001997974191609155,
      "loss": 23.0,
      "step": 367
    },
    {
      "epoch": 0.22088835534213686,
      "grad_norm": 0.0004023246292490512,
      "learning_rate": 0.00019979621552608276,
      "loss": 23.0,
      "step": 368
    },
    {
      "epoch": 0.22148859543817526,
      "grad_norm": 0.0007271401700563729,
      "learning_rate": 0.00019979500832977985,
      "loss": 23.0,
      "step": 369
    },
    {
      "epoch": 0.22208883553421369,
      "grad_norm": 0.0003374427615199238,
      "learning_rate": 0.00019979379757204978,
      "loss": 23.0,
      "step": 370
    },
    {
      "epoch": 0.22268907563025211,
      "grad_norm": 0.000754833163227886,
      "learning_rate": 0.0001997925832529358,
      "loss": 23.0,
      "step": 371
    },
    {
      "epoch": 0.22328931572629052,
      "grad_norm": 0.000961002369876951,
      "learning_rate": 0.00019979136537248118,
      "loss": 23.0,
      "step": 372
    },
    {
      "epoch": 0.22388955582232895,
      "grad_norm": 0.0005781974759884179,
      "learning_rate": 0.00019979014393072943,
      "loss": 23.0,
      "step": 373
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 0.00044546660501509905,
      "learning_rate": 0.00019978891892772415,
      "loss": 23.0,
      "step": 374
    },
    {
      "epoch": 0.22509003601440578,
      "grad_norm": 0.0005686086369678378,
      "learning_rate": 0.00019978769036350905,
      "loss": 23.0,
      "step": 375
    },
    {
      "epoch": 0.22569027611044418,
      "grad_norm": 0.000627652567345649,
      "learning_rate": 0.00019978645823812797,
      "loss": 23.0,
      "step": 376
    },
    {
      "epoch": 0.2262905162064826,
      "grad_norm": 0.0006272499449551105,
      "learning_rate": 0.00019978522255162485,
      "loss": 23.0,
      "step": 377
    },
    {
      "epoch": 0.226890756302521,
      "grad_norm": 0.0007022761856205761,
      "learning_rate": 0.0001997839833040438,
      "loss": 23.0,
      "step": 378
    },
    {
      "epoch": 0.22749099639855944,
      "grad_norm": 0.0006133838323876262,
      "learning_rate": 0.0001997827404954291,
      "loss": 23.0,
      "step": 379
    },
    {
      "epoch": 0.22809123649459784,
      "grad_norm": 0.0004374498676043004,
      "learning_rate": 0.00019978149412582506,
      "loss": 23.0,
      "step": 380
    },
    {
      "epoch": 0.22869147659063627,
      "grad_norm": 0.00046753635979257524,
      "learning_rate": 0.00019978024419527613,
      "loss": 23.0,
      "step": 381
    },
    {
      "epoch": 0.22929171668667467,
      "grad_norm": 0.0006269077421166003,
      "learning_rate": 0.00019977899070382698,
      "loss": 23.0,
      "step": 382
    },
    {
      "epoch": 0.2298919567827131,
      "grad_norm": 0.0008202579920180142,
      "learning_rate": 0.00019977773365152231,
      "loss": 23.0,
      "step": 383
    },
    {
      "epoch": 0.2304921968787515,
      "grad_norm": 0.0009668364655226469,
      "learning_rate": 0.00019977647303840698,
      "loss": 23.0,
      "step": 384
    },
    {
      "epoch": 0.23109243697478993,
      "grad_norm": 0.0004292426456231624,
      "learning_rate": 0.00019977520886452598,
      "loss": 23.0,
      "step": 385
    },
    {
      "epoch": 0.23169267707082833,
      "grad_norm": 0.00030157389119267464,
      "learning_rate": 0.0001997739411299244,
      "loss": 23.0,
      "step": 386
    },
    {
      "epoch": 0.23229291716686676,
      "grad_norm": 0.00035860264324583113,
      "learning_rate": 0.00019977266983464757,
      "loss": 23.0,
      "step": 387
    },
    {
      "epoch": 0.23289315726290516,
      "grad_norm": 0.00041682066512294114,
      "learning_rate": 0.00019977139497874078,
      "loss": 23.0,
      "step": 388
    },
    {
      "epoch": 0.2334933973589436,
      "grad_norm": 0.0006654221215285361,
      "learning_rate": 0.00019977011656224954,
      "loss": 23.0,
      "step": 389
    },
    {
      "epoch": 0.234093637454982,
      "grad_norm": 0.0003358942340128124,
      "learning_rate": 0.00019976883458521947,
      "loss": 23.0,
      "step": 390
    },
    {
      "epoch": 0.23469387755102042,
      "grad_norm": 0.0004877279279753566,
      "learning_rate": 0.00019976754904769633,
      "loss": 23.0,
      "step": 391
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.0004985998966731131,
      "learning_rate": 0.000199766259949726,
      "loss": 23.0,
      "step": 392
    },
    {
      "epoch": 0.23589435774309725,
      "grad_norm": 0.0006408619810827076,
      "learning_rate": 0.0001997649672913545,
      "loss": 23.0,
      "step": 393
    },
    {
      "epoch": 0.23649459783913565,
      "grad_norm": 0.00035928620491176844,
      "learning_rate": 0.00019976367107262793,
      "loss": 23.0,
      "step": 394
    },
    {
      "epoch": 0.23709483793517408,
      "grad_norm": 0.0007407786906696856,
      "learning_rate": 0.00019976237129359255,
      "loss": 23.0,
      "step": 395
    },
    {
      "epoch": 0.23769507803121248,
      "grad_norm": 0.0005418708897195756,
      "learning_rate": 0.00019976106795429477,
      "loss": 23.0,
      "step": 396
    },
    {
      "epoch": 0.2382953181272509,
      "grad_norm": 0.0006996035808697343,
      "learning_rate": 0.00019975976105478108,
      "loss": 23.0,
      "step": 397
    },
    {
      "epoch": 0.2388955582232893,
      "grad_norm": 0.0005328754195943475,
      "learning_rate": 0.00019975845059509815,
      "loss": 23.0,
      "step": 398
    },
    {
      "epoch": 0.23949579831932774,
      "grad_norm": 0.0009513835539110005,
      "learning_rate": 0.00019975713657529275,
      "loss": 23.0,
      "step": 399
    },
    {
      "epoch": 0.24009603841536614,
      "grad_norm": 0.0006411597714759409,
      "learning_rate": 0.00019975581899541172,
      "loss": 23.0,
      "step": 400
    },
    {
      "epoch": 0.24069627851140457,
      "grad_norm": 0.001117164152674377,
      "learning_rate": 0.00019975449785550212,
      "loss": 23.0,
      "step": 401
    },
    {
      "epoch": 0.24129651860744297,
      "grad_norm": 0.0010399962775409222,
      "learning_rate": 0.00019975317315561108,
      "loss": 23.0,
      "step": 402
    },
    {
      "epoch": 0.2418967587034814,
      "grad_norm": 0.0008498011156916618,
      "learning_rate": 0.00019975184489578587,
      "loss": 23.0,
      "step": 403
    },
    {
      "epoch": 0.2424969987995198,
      "grad_norm": 0.0005976161919534206,
      "learning_rate": 0.0001997505130760739,
      "loss": 23.0,
      "step": 404
    },
    {
      "epoch": 0.24309723889555823,
      "grad_norm": 0.00085114233661443,
      "learning_rate": 0.00019974917769652274,
      "loss": 23.0,
      "step": 405
    },
    {
      "epoch": 0.24369747899159663,
      "grad_norm": 0.0006555085419677198,
      "learning_rate": 0.00019974783875718,
      "loss": 23.0,
      "step": 406
    },
    {
      "epoch": 0.24429771908763506,
      "grad_norm": 0.00048095532110892236,
      "learning_rate": 0.0001997464962580935,
      "loss": 23.0,
      "step": 407
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.0006680062506347895,
      "learning_rate": 0.00019974515019931114,
      "loss": 23.0,
      "step": 408
    },
    {
      "epoch": 0.2454981992797119,
      "grad_norm": 0.0006157276802696288,
      "learning_rate": 0.0001997438005808809,
      "loss": 23.0,
      "step": 409
    },
    {
      "epoch": 0.2460984393757503,
      "grad_norm": 0.0004931309376843274,
      "learning_rate": 0.000199742447402851,
      "loss": 23.0,
      "step": 410
    },
    {
      "epoch": 0.24669867947178872,
      "grad_norm": 0.0005629705265164375,
      "learning_rate": 0.00019974109066526975,
      "loss": 23.0,
      "step": 411
    },
    {
      "epoch": 0.24729891956782712,
      "grad_norm": 0.0009498510626144707,
      "learning_rate": 0.00019973973036818552,
      "loss": 23.0,
      "step": 412
    },
    {
      "epoch": 0.24789915966386555,
      "grad_norm": 0.000444902601884678,
      "learning_rate": 0.00019973836651164683,
      "loss": 23.0,
      "step": 413
    },
    {
      "epoch": 0.24849939975990396,
      "grad_norm": 0.0004672042850870639,
      "learning_rate": 0.00019973699909570244,
      "loss": 23.0,
      "step": 414
    },
    {
      "epoch": 0.24909963985594238,
      "grad_norm": 0.0008003438706509769,
      "learning_rate": 0.0001997356281204011,
      "loss": 23.0,
      "step": 415
    },
    {
      "epoch": 0.24969987995198079,
      "grad_norm": 0.0003626558172982186,
      "learning_rate": 0.00019973425358579176,
      "loss": 23.0,
      "step": 416
    },
    {
      "epoch": 0.2503001200480192,
      "grad_norm": 0.0004940864164382219,
      "learning_rate": 0.00019973287549192344,
      "loss": 23.0,
      "step": 417
    },
    {
      "epoch": 0.25090036014405764,
      "grad_norm": 0.0007219372782856226,
      "learning_rate": 0.00019973149383884536,
      "loss": 23.0,
      "step": 418
    },
    {
      "epoch": 0.251500600240096,
      "grad_norm": 0.0007677687099203467,
      "learning_rate": 0.00019973010862660677,
      "loss": 23.0,
      "step": 419
    },
    {
      "epoch": 0.25210084033613445,
      "grad_norm": 0.0005680742906406522,
      "learning_rate": 0.00019972871985525717,
      "loss": 23.0,
      "step": 420
    },
    {
      "epoch": 0.2527010804321729,
      "grad_norm": 0.0012042596936225891,
      "learning_rate": 0.00019972732752484605,
      "loss": 23.0,
      "step": 421
    },
    {
      "epoch": 0.2533013205282113,
      "grad_norm": 0.0010296528926119208,
      "learning_rate": 0.0001997259316354232,
      "loss": 23.0,
      "step": 422
    },
    {
      "epoch": 0.2539015606242497,
      "grad_norm": 0.0006402980070561171,
      "learning_rate": 0.00019972453218703834,
      "loss": 23.0,
      "step": 423
    },
    {
      "epoch": 0.2545018007202881,
      "grad_norm": 0.0006200941861607134,
      "learning_rate": 0.00019972312917974143,
      "loss": 23.0,
      "step": 424
    },
    {
      "epoch": 0.25510204081632654,
      "grad_norm": 0.0005011442117393017,
      "learning_rate": 0.00019972172261358259,
      "loss": 23.0,
      "step": 425
    },
    {
      "epoch": 0.25570228091236497,
      "grad_norm": 0.0007742932066321373,
      "learning_rate": 0.00019972031248861196,
      "loss": 23.0,
      "step": 426
    },
    {
      "epoch": 0.25630252100840334,
      "grad_norm": 0.0013804651098325849,
      "learning_rate": 0.0001997188988048799,
      "loss": 23.0,
      "step": 427
    },
    {
      "epoch": 0.25690276110444177,
      "grad_norm": 0.000599585531745106,
      "learning_rate": 0.00019971748156243687,
      "loss": 23.0,
      "step": 428
    },
    {
      "epoch": 0.2575030012004802,
      "grad_norm": 0.00047438021283596754,
      "learning_rate": 0.0001997160607613334,
      "loss": 23.0,
      "step": 429
    },
    {
      "epoch": 0.2581032412965186,
      "grad_norm": 0.0008628160576336086,
      "learning_rate": 0.00019971463640162024,
      "loss": 23.0,
      "step": 430
    },
    {
      "epoch": 0.258703481392557,
      "grad_norm": 0.0007876844028942287,
      "learning_rate": 0.00019971320848334817,
      "loss": 23.0,
      "step": 431
    },
    {
      "epoch": 0.25930372148859543,
      "grad_norm": 0.000673728296533227,
      "learning_rate": 0.00019971177700656823,
      "loss": 23.0,
      "step": 432
    },
    {
      "epoch": 0.25990396158463386,
      "grad_norm": 0.0009601108031347394,
      "learning_rate": 0.0001997103419713314,
      "loss": 23.0,
      "step": 433
    },
    {
      "epoch": 0.2605042016806723,
      "grad_norm": 0.0006990964757278562,
      "learning_rate": 0.000199708903377689,
      "loss": 23.0,
      "step": 434
    },
    {
      "epoch": 0.26110444177671066,
      "grad_norm": 0.0005273604765534401,
      "learning_rate": 0.0001997074612256923,
      "loss": 23.0,
      "step": 435
    },
    {
      "epoch": 0.2617046818727491,
      "grad_norm": 0.0015036009717732668,
      "learning_rate": 0.00019970601551539278,
      "loss": 23.0,
      "step": 436
    },
    {
      "epoch": 0.2623049219687875,
      "grad_norm": 0.000941345700994134,
      "learning_rate": 0.00019970456624684207,
      "loss": 23.0,
      "step": 437
    },
    {
      "epoch": 0.26290516206482595,
      "grad_norm": 0.0005987265612930059,
      "learning_rate": 0.00019970311342009184,
      "loss": 23.0,
      "step": 438
    },
    {
      "epoch": 0.2635054021608643,
      "grad_norm": 0.00040275233914144337,
      "learning_rate": 0.00019970165703519396,
      "loss": 23.0,
      "step": 439
    },
    {
      "epoch": 0.26410564225690275,
      "grad_norm": 0.0006889562937431037,
      "learning_rate": 0.0001997001970922004,
      "loss": 23.0,
      "step": 440
    },
    {
      "epoch": 0.2647058823529412,
      "grad_norm": 0.0005851078312844038,
      "learning_rate": 0.00019969873359116326,
      "loss": 23.0,
      "step": 441
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 0.0005285036750137806,
      "learning_rate": 0.00019969726653213478,
      "loss": 23.0,
      "step": 442
    },
    {
      "epoch": 0.265906362545018,
      "grad_norm": 0.000706472375895828,
      "learning_rate": 0.00019969579591516733,
      "loss": 23.0,
      "step": 443
    },
    {
      "epoch": 0.2665066026410564,
      "grad_norm": 0.0005748828989453614,
      "learning_rate": 0.00019969432174031336,
      "loss": 23.0,
      "step": 444
    },
    {
      "epoch": 0.26710684273709484,
      "grad_norm": 0.0009791315533220768,
      "learning_rate": 0.0001996928440076255,
      "loss": 23.0,
      "step": 445
    },
    {
      "epoch": 0.26770708283313327,
      "grad_norm": 0.00048265812802128494,
      "learning_rate": 0.00019969136271715644,
      "loss": 23.0,
      "step": 446
    },
    {
      "epoch": 0.26830732292917164,
      "grad_norm": 0.0005793205345980823,
      "learning_rate": 0.00019968987786895913,
      "loss": 23.0,
      "step": 447
    },
    {
      "epoch": 0.2689075630252101,
      "grad_norm": 0.0006732153124175966,
      "learning_rate": 0.00019968838946308649,
      "loss": 23.0,
      "step": 448
    },
    {
      "epoch": 0.2695078031212485,
      "grad_norm": 0.0007057045004330575,
      "learning_rate": 0.00019968689749959165,
      "loss": 23.0,
      "step": 449
    },
    {
      "epoch": 0.27010804321728693,
      "grad_norm": 0.0005959949921816587,
      "learning_rate": 0.00019968540197852787,
      "loss": 23.0,
      "step": 450
    },
    {
      "epoch": 0.2707082833133253,
      "grad_norm": 0.0011223838664591312,
      "learning_rate": 0.00019968390289994852,
      "loss": 23.0,
      "step": 451
    },
    {
      "epoch": 0.27130852340936373,
      "grad_norm": 0.00040181170334108174,
      "learning_rate": 0.0001996824002639071,
      "loss": 23.0,
      "step": 452
    },
    {
      "epoch": 0.27190876350540216,
      "grad_norm": 0.0006953475531190634,
      "learning_rate": 0.0001996808940704572,
      "loss": 23.0,
      "step": 453
    },
    {
      "epoch": 0.2725090036014406,
      "grad_norm": 0.0008081067353487015,
      "learning_rate": 0.00019967938431965262,
      "loss": 23.0,
      "step": 454
    },
    {
      "epoch": 0.27310924369747897,
      "grad_norm": 0.0007403423660434783,
      "learning_rate": 0.00019967787101154722,
      "loss": 23.0,
      "step": 455
    },
    {
      "epoch": 0.2737094837935174,
      "grad_norm": 0.0007606777944602072,
      "learning_rate": 0.000199676354146195,
      "loss": 23.0,
      "step": 456
    },
    {
      "epoch": 0.2743097238895558,
      "grad_norm": 0.000691116729285568,
      "learning_rate": 0.00019967483372365012,
      "loss": 23.0,
      "step": 457
    },
    {
      "epoch": 0.27490996398559425,
      "grad_norm": 0.0007377572474069893,
      "learning_rate": 0.0001996733097439668,
      "loss": 23.0,
      "step": 458
    },
    {
      "epoch": 0.2755102040816326,
      "grad_norm": 0.000592804339248687,
      "learning_rate": 0.00019967178220719944,
      "loss": 23.0,
      "step": 459
    },
    {
      "epoch": 0.27611044417767105,
      "grad_norm": 0.0005102523718960583,
      "learning_rate": 0.00019967025111340255,
      "loss": 23.0,
      "step": 460
    },
    {
      "epoch": 0.2767106842737095,
      "grad_norm": 0.0010906271636486053,
      "learning_rate": 0.0001996687164626308,
      "loss": 23.0,
      "step": 461
    },
    {
      "epoch": 0.2773109243697479,
      "grad_norm": 0.0008513031643815339,
      "learning_rate": 0.00019966717825493894,
      "loss": 23.0,
      "step": 462
    },
    {
      "epoch": 0.27791116446578634,
      "grad_norm": 0.0007379608578048646,
      "learning_rate": 0.00019966563649038185,
      "loss": 23.0,
      "step": 463
    },
    {
      "epoch": 0.2785114045618247,
      "grad_norm": 0.0006293412297964096,
      "learning_rate": 0.0001996640911690146,
      "loss": 23.0,
      "step": 464
    },
    {
      "epoch": 0.27911164465786314,
      "grad_norm": 0.0005843341932632029,
      "learning_rate": 0.00019966254229089226,
      "loss": 23.0,
      "step": 465
    },
    {
      "epoch": 0.2797118847539016,
      "grad_norm": 0.00046797405229881406,
      "learning_rate": 0.00019966098985607015,
      "loss": 23.0,
      "step": 466
    },
    {
      "epoch": 0.28031212484994,
      "grad_norm": 0.0007364925113506615,
      "learning_rate": 0.0001996594338646037,
      "loss": 23.0,
      "step": 467
    },
    {
      "epoch": 0.2809123649459784,
      "grad_norm": 0.0011836322955787182,
      "learning_rate": 0.0001996578743165484,
      "loss": 23.0,
      "step": 468
    },
    {
      "epoch": 0.2815126050420168,
      "grad_norm": 0.0005399160436354578,
      "learning_rate": 0.00019965631121195992,
      "loss": 23.0,
      "step": 469
    },
    {
      "epoch": 0.28211284513805523,
      "grad_norm": 0.0007591730100102723,
      "learning_rate": 0.000199654744550894,
      "loss": 23.0,
      "step": 470
    },
    {
      "epoch": 0.28271308523409366,
      "grad_norm": 0.0005403325776569545,
      "learning_rate": 0.00019965317433340662,
      "loss": 23.0,
      "step": 471
    },
    {
      "epoch": 0.28331332533013204,
      "grad_norm": 0.0005854291375726461,
      "learning_rate": 0.00019965160055955382,
      "loss": 23.0,
      "step": 472
    },
    {
      "epoch": 0.28391356542617047,
      "grad_norm": 0.00045302778016775846,
      "learning_rate": 0.0001996500232293917,
      "loss": 23.0,
      "step": 473
    },
    {
      "epoch": 0.2845138055222089,
      "grad_norm": 0.0004636446828953922,
      "learning_rate": 0.00019964844234297656,
      "loss": 23.0,
      "step": 474
    },
    {
      "epoch": 0.2851140456182473,
      "grad_norm": 0.0004668672336265445,
      "learning_rate": 0.00019964685790036484,
      "loss": 23.0,
      "step": 475
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.0009069807711057365,
      "learning_rate": 0.0001996452699016131,
      "loss": 23.0,
      "step": 476
    },
    {
      "epoch": 0.2863145258103241,
      "grad_norm": 0.0005298946634866297,
      "learning_rate": 0.000199643678346778,
      "loss": 23.0,
      "step": 477
    },
    {
      "epoch": 0.28691476590636256,
      "grad_norm": 0.0009359425166621804,
      "learning_rate": 0.00019964208323591633,
      "loss": 23.0,
      "step": 478
    },
    {
      "epoch": 0.287515006002401,
      "grad_norm": 0.0008660585153847933,
      "learning_rate": 0.000199640484569085,
      "loss": 23.0,
      "step": 479
    },
    {
      "epoch": 0.28811524609843936,
      "grad_norm": 0.00040566539973951876,
      "learning_rate": 0.00019963888234634112,
      "loss": 23.0,
      "step": 480
    },
    {
      "epoch": 0.2887154861944778,
      "grad_norm": 0.0003369592595845461,
      "learning_rate": 0.0001996372765677418,
      "loss": 23.0,
      "step": 481
    },
    {
      "epoch": 0.2893157262905162,
      "grad_norm": 0.0015372679335996509,
      "learning_rate": 0.0001996356672333444,
      "loss": 23.0,
      "step": 482
    },
    {
      "epoch": 0.28991596638655465,
      "grad_norm": 0.000425403646659106,
      "learning_rate": 0.0001996340543432063,
      "loss": 23.0,
      "step": 483
    },
    {
      "epoch": 0.290516206482593,
      "grad_norm": 0.001078411820344627,
      "learning_rate": 0.00019963243789738512,
      "loss": 23.0,
      "step": 484
    },
    {
      "epoch": 0.29111644657863145,
      "grad_norm": 0.000449808721896261,
      "learning_rate": 0.00019963081789593851,
      "loss": 23.0,
      "step": 485
    },
    {
      "epoch": 0.2917166866746699,
      "grad_norm": 0.0013598357327282429,
      "learning_rate": 0.0001996291943389243,
      "loss": 23.0,
      "step": 486
    },
    {
      "epoch": 0.2923169267707083,
      "grad_norm": 0.0015207430114969611,
      "learning_rate": 0.0001996275672264004,
      "loss": 23.0,
      "step": 487
    },
    {
      "epoch": 0.2929171668667467,
      "grad_norm": 0.0006574721192009747,
      "learning_rate": 0.0001996259365584249,
      "loss": 23.0,
      "step": 488
    },
    {
      "epoch": 0.2935174069627851,
      "grad_norm": 0.000553544145077467,
      "learning_rate": 0.000199624302335056,
      "loss": 23.0,
      "step": 489
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 0.0004415796138346195,
      "learning_rate": 0.000199622664556352,
      "loss": 23.0,
      "step": 490
    },
    {
      "epoch": 0.29471788715486197,
      "grad_norm": 0.0005506374291144311,
      "learning_rate": 0.00019962102322237138,
      "loss": 23.0,
      "step": 491
    },
    {
      "epoch": 0.29531812725090034,
      "grad_norm": 0.0005820810911245644,
      "learning_rate": 0.0001996193783331727,
      "loss": 23.0,
      "step": 492
    },
    {
      "epoch": 0.29591836734693877,
      "grad_norm": 0.0007612020126543939,
      "learning_rate": 0.00019961772988881464,
      "loss": 23.0,
      "step": 493
    },
    {
      "epoch": 0.2965186074429772,
      "grad_norm": 0.0010789800435304642,
      "learning_rate": 0.00019961607788935605,
      "loss": 23.0,
      "step": 494
    },
    {
      "epoch": 0.29711884753901563,
      "grad_norm": 0.00042243878124281764,
      "learning_rate": 0.0001996144223348559,
      "loss": 23.0,
      "step": 495
    },
    {
      "epoch": 0.297719087635054,
      "grad_norm": 0.0007266091415658593,
      "learning_rate": 0.00019961276322537325,
      "loss": 23.0,
      "step": 496
    },
    {
      "epoch": 0.29831932773109243,
      "grad_norm": 0.0008625111076980829,
      "learning_rate": 0.0001996111005609673,
      "loss": 23.0,
      "step": 497
    },
    {
      "epoch": 0.29891956782713086,
      "grad_norm": 0.000400591641664505,
      "learning_rate": 0.00019960943434169743,
      "loss": 23.0,
      "step": 498
    },
    {
      "epoch": 0.2995198079231693,
      "grad_norm": 0.0005319428746588528,
      "learning_rate": 0.00019960776456762305,
      "loss": 23.0,
      "step": 499
    },
    {
      "epoch": 0.30012004801920766,
      "grad_norm": 0.0007574324845336378,
      "learning_rate": 0.00019960609123880377,
      "loss": 23.0,
      "step": 500
    },
    {
      "epoch": 0.3007202881152461,
      "grad_norm": 0.0008873497718013823,
      "learning_rate": 0.00019960441435529932,
      "loss": 23.0,
      "step": 501
    },
    {
      "epoch": 0.3013205282112845,
      "grad_norm": 0.0003605948877520859,
      "learning_rate": 0.00019960273391716955,
      "loss": 23.0,
      "step": 502
    },
    {
      "epoch": 0.30192076830732295,
      "grad_norm": 0.0008765537058934569,
      "learning_rate": 0.0001996010499244744,
      "loss": 23.0,
      "step": 503
    },
    {
      "epoch": 0.3025210084033613,
      "grad_norm": 0.0005614471738226712,
      "learning_rate": 0.00019959936237727398,
      "loss": 23.0,
      "step": 504
    },
    {
      "epoch": 0.30312124849939975,
      "grad_norm": 0.0006175639573484659,
      "learning_rate": 0.00019959767127562854,
      "loss": 23.0,
      "step": 505
    },
    {
      "epoch": 0.3037214885954382,
      "grad_norm": 0.0006772951455786824,
      "learning_rate": 0.00019959597661959837,
      "loss": 23.0,
      "step": 506
    },
    {
      "epoch": 0.3043217286914766,
      "grad_norm": 0.0007493697921745479,
      "learning_rate": 0.000199594278409244,
      "loss": 23.0,
      "step": 507
    },
    {
      "epoch": 0.304921968787515,
      "grad_norm": 0.0013073779409751296,
      "learning_rate": 0.00019959257664462605,
      "loss": 23.0,
      "step": 508
    },
    {
      "epoch": 0.3055222088835534,
      "grad_norm": 0.001169772120192647,
      "learning_rate": 0.00019959087132580519,
      "loss": 23.0,
      "step": 509
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 0.0010430820984765887,
      "learning_rate": 0.00019958916245284232,
      "loss": 23.0,
      "step": 510
    },
    {
      "epoch": 0.3067226890756303,
      "grad_norm": 0.0007382632466033101,
      "learning_rate": 0.0001995874500257984,
      "loss": 23.0,
      "step": 511
    },
    {
      "epoch": 0.30732292917166865,
      "grad_norm": 0.00046590864076279104,
      "learning_rate": 0.0001995857340447346,
      "loss": 23.0,
      "step": 512
    },
    {
      "epoch": 0.3079231692677071,
      "grad_norm": 0.0008095836383290589,
      "learning_rate": 0.00019958401450971208,
      "loss": 23.0,
      "step": 513
    },
    {
      "epoch": 0.3085234093637455,
      "grad_norm": 0.0004994541523046792,
      "learning_rate": 0.00019958229142079224,
      "loss": 23.0,
      "step": 514
    },
    {
      "epoch": 0.30912364945978393,
      "grad_norm": 0.00032915148767642677,
      "learning_rate": 0.0001995805647780366,
      "loss": 23.0,
      "step": 515
    },
    {
      "epoch": 0.3097238895558223,
      "grad_norm": 0.00047977123176679015,
      "learning_rate": 0.00019957883458150675,
      "loss": 23.0,
      "step": 516
    },
    {
      "epoch": 0.31032412965186074,
      "grad_norm": 0.0004798262962140143,
      "learning_rate": 0.00019957710083126444,
      "loss": 23.0,
      "step": 517
    },
    {
      "epoch": 0.31092436974789917,
      "grad_norm": 0.0012143745552748442,
      "learning_rate": 0.00019957536352737152,
      "loss": 23.0,
      "step": 518
    },
    {
      "epoch": 0.3115246098439376,
      "grad_norm": 0.0008633871912024915,
      "learning_rate": 0.00019957362266989005,
      "loss": 23.0,
      "step": 519
    },
    {
      "epoch": 0.31212484993997597,
      "grad_norm": 0.0005252179689705372,
      "learning_rate": 0.0001995718782588821,
      "loss": 23.0,
      "step": 520
    },
    {
      "epoch": 0.3127250900360144,
      "grad_norm": 0.0013168033910915256,
      "learning_rate": 0.00019957013029440993,
      "loss": 23.0,
      "step": 521
    },
    {
      "epoch": 0.3133253301320528,
      "grad_norm": 0.00026732790865935385,
      "learning_rate": 0.00019956837877653597,
      "loss": 23.0,
      "step": 522
    },
    {
      "epoch": 0.31392557022809126,
      "grad_norm": 0.0005732500576414168,
      "learning_rate": 0.0001995666237053227,
      "loss": 23.0,
      "step": 523
    },
    {
      "epoch": 0.31452581032412963,
      "grad_norm": 0.0003575915179681033,
      "learning_rate": 0.0001995648650808327,
      "loss": 23.0,
      "step": 524
    },
    {
      "epoch": 0.31512605042016806,
      "grad_norm": 0.0006335443467833102,
      "learning_rate": 0.0001995631029031288,
      "loss": 23.0,
      "step": 525
    },
    {
      "epoch": 0.3157262905162065,
      "grad_norm": 0.0018182954518124461,
      "learning_rate": 0.00019956133717227386,
      "loss": 23.0,
      "step": 526
    },
    {
      "epoch": 0.3163265306122449,
      "grad_norm": 0.0011957690585404634,
      "learning_rate": 0.0001995595678883309,
      "loss": 23.0,
      "step": 527
    },
    {
      "epoch": 0.3169267707082833,
      "grad_norm": 0.00040781046845950186,
      "learning_rate": 0.00019955779505136304,
      "loss": 23.0,
      "step": 528
    },
    {
      "epoch": 0.3175270108043217,
      "grad_norm": 0.0010567825520411134,
      "learning_rate": 0.00019955601866143358,
      "loss": 23.0,
      "step": 529
    },
    {
      "epoch": 0.31812725090036015,
      "grad_norm": 0.0007379227317869663,
      "learning_rate": 0.00019955423871860593,
      "loss": 23.0,
      "step": 530
    },
    {
      "epoch": 0.3187274909963986,
      "grad_norm": 0.0005250342655926943,
      "learning_rate": 0.00019955245522294357,
      "loss": 23.0,
      "step": 531
    },
    {
      "epoch": 0.31932773109243695,
      "grad_norm": 0.0020976525265723467,
      "learning_rate": 0.00019955066817451016,
      "loss": 23.0,
      "step": 532
    },
    {
      "epoch": 0.3199279711884754,
      "grad_norm": 0.0012484453618526459,
      "learning_rate": 0.00019954887757336946,
      "loss": 23.0,
      "step": 533
    },
    {
      "epoch": 0.3205282112845138,
      "grad_norm": 0.00038985523860901594,
      "learning_rate": 0.0001995470834195854,
      "loss": 23.0,
      "step": 534
    },
    {
      "epoch": 0.32112845138055224,
      "grad_norm": 0.0005392651073634624,
      "learning_rate": 0.00019954528571322202,
      "loss": 23.0,
      "step": 535
    },
    {
      "epoch": 0.3217286914765906,
      "grad_norm": 0.00044334621634334326,
      "learning_rate": 0.00019954348445434345,
      "loss": 23.0,
      "step": 536
    },
    {
      "epoch": 0.32232893157262904,
      "grad_norm": 0.001021339907310903,
      "learning_rate": 0.00019954167964301393,
      "loss": 23.0,
      "step": 537
    },
    {
      "epoch": 0.32292917166866747,
      "grad_norm": 0.0005136961117386818,
      "learning_rate": 0.00019953987127929796,
      "loss": 23.0,
      "step": 538
    },
    {
      "epoch": 0.3235294117647059,
      "grad_norm": 0.0010431839618831873,
      "learning_rate": 0.00019953805936326002,
      "loss": 23.0,
      "step": 539
    },
    {
      "epoch": 0.3241296518607443,
      "grad_norm": 0.0008680829196237028,
      "learning_rate": 0.0001995362438949648,
      "loss": 23.0,
      "step": 540
    },
    {
      "epoch": 0.3247298919567827,
      "grad_norm": 0.00047551849274896085,
      "learning_rate": 0.00019953442487447704,
      "loss": 23.0,
      "step": 541
    },
    {
      "epoch": 0.32533013205282113,
      "grad_norm": 0.000665904488414526,
      "learning_rate": 0.00019953260230186172,
      "loss": 23.0,
      "step": 542
    },
    {
      "epoch": 0.32593037214885956,
      "grad_norm": 0.0007536861230619252,
      "learning_rate": 0.00019953077617718384,
      "loss": 23.0,
      "step": 543
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.0005331055726855993,
      "learning_rate": 0.00019952894650050856,
      "loss": 23.0,
      "step": 544
    },
    {
      "epoch": 0.32713085234093636,
      "grad_norm": 0.0006826044409535825,
      "learning_rate": 0.00019952711327190123,
      "loss": 23.0,
      "step": 545
    },
    {
      "epoch": 0.3277310924369748,
      "grad_norm": 0.001005376921966672,
      "learning_rate": 0.00019952527649142723,
      "loss": 23.0,
      "step": 546
    },
    {
      "epoch": 0.3283313325330132,
      "grad_norm": 0.0010133546311408281,
      "learning_rate": 0.00019952343615915212,
      "loss": 23.0,
      "step": 547
    },
    {
      "epoch": 0.3289315726290516,
      "grad_norm": 0.0015964146004989743,
      "learning_rate": 0.00019952159227514159,
      "loss": 23.0,
      "step": 548
    },
    {
      "epoch": 0.32953181272509,
      "grad_norm": 0.0006953754345886409,
      "learning_rate": 0.0001995197448394614,
      "loss": 23.0,
      "step": 549
    },
    {
      "epoch": 0.33013205282112845,
      "grad_norm": 0.0009064920013770461,
      "learning_rate": 0.00019951789385217757,
      "loss": 23.0,
      "step": 550
    },
    {
      "epoch": 0.3307322929171669,
      "grad_norm": 0.0009463357273489237,
      "learning_rate": 0.00019951603931335602,
      "loss": 23.0,
      "step": 551
    },
    {
      "epoch": 0.33133253301320525,
      "grad_norm": 0.0007558033103123307,
      "learning_rate": 0.00019951418122306306,
      "loss": 23.0,
      "step": 552
    },
    {
      "epoch": 0.3319327731092437,
      "grad_norm": 0.0005100120906718075,
      "learning_rate": 0.00019951231958136498,
      "loss": 23.0,
      "step": 553
    },
    {
      "epoch": 0.3325330132052821,
      "grad_norm": 0.0008068382157944143,
      "learning_rate": 0.00019951045438832817,
      "loss": 23.0,
      "step": 554
    },
    {
      "epoch": 0.33313325330132054,
      "grad_norm": 0.0005732056451961398,
      "learning_rate": 0.00019950858564401916,
      "loss": 23.0,
      "step": 555
    },
    {
      "epoch": 0.33373349339735897,
      "grad_norm": 0.0005752798751927912,
      "learning_rate": 0.00019950671334850476,
      "loss": 23.0,
      "step": 556
    },
    {
      "epoch": 0.33433373349339734,
      "grad_norm": 0.0005713048158213496,
      "learning_rate": 0.0001995048375018517,
      "loss": 23.0,
      "step": 557
    },
    {
      "epoch": 0.3349339735894358,
      "grad_norm": 0.000498209788929671,
      "learning_rate": 0.00019950295810412696,
      "loss": 23.0,
      "step": 558
    },
    {
      "epoch": 0.3355342136854742,
      "grad_norm": 0.0008248278754763305,
      "learning_rate": 0.00019950107515539757,
      "loss": 23.0,
      "step": 559
    },
    {
      "epoch": 0.33613445378151263,
      "grad_norm": 0.00075948127778247,
      "learning_rate": 0.00019949918865573078,
      "loss": 23.0,
      "step": 560
    },
    {
      "epoch": 0.336734693877551,
      "grad_norm": 0.0006507814396172762,
      "learning_rate": 0.00019949729860519385,
      "loss": 23.0,
      "step": 561
    },
    {
      "epoch": 0.33733493397358943,
      "grad_norm": 0.0006093204719945788,
      "learning_rate": 0.00019949540500385432,
      "loss": 23.0,
      "step": 562
    },
    {
      "epoch": 0.33793517406962786,
      "grad_norm": 0.0009938915027305484,
      "learning_rate": 0.00019949350785177972,
      "loss": 23.0,
      "step": 563
    },
    {
      "epoch": 0.3385354141656663,
      "grad_norm": 0.0005058671813458204,
      "learning_rate": 0.00019949160714903772,
      "loss": 23.0,
      "step": 564
    },
    {
      "epoch": 0.33913565426170467,
      "grad_norm": 0.0005110659985803068,
      "learning_rate": 0.00019948970289569617,
      "loss": 23.0,
      "step": 565
    },
    {
      "epoch": 0.3397358943577431,
      "grad_norm": 0.000917077821213752,
      "learning_rate": 0.00019948779509182306,
      "loss": 23.0,
      "step": 566
    },
    {
      "epoch": 0.3403361344537815,
      "grad_norm": 0.0008047555456869304,
      "learning_rate": 0.0001994858837374865,
      "loss": 23.0,
      "step": 567
    },
    {
      "epoch": 0.34093637454981995,
      "grad_norm": 0.0007102153031155467,
      "learning_rate": 0.0001994839688327546,
      "loss": 23.0,
      "step": 568
    },
    {
      "epoch": 0.3415366146458583,
      "grad_norm": 0.0006414945237338543,
      "learning_rate": 0.0001994820503776958,
      "loss": 23.0,
      "step": 569
    },
    {
      "epoch": 0.34213685474189676,
      "grad_norm": 0.0003412929072510451,
      "learning_rate": 0.0001994801283723785,
      "loss": 23.0,
      "step": 570
    },
    {
      "epoch": 0.3427370948379352,
      "grad_norm": 0.0008194602560251951,
      "learning_rate": 0.0001994782028168713,
      "loss": 23.0,
      "step": 571
    },
    {
      "epoch": 0.3433373349339736,
      "grad_norm": 0.0007939741481095552,
      "learning_rate": 0.0001994762737112429,
      "loss": 23.0,
      "step": 572
    },
    {
      "epoch": 0.343937575030012,
      "grad_norm": 0.0010254220105707645,
      "learning_rate": 0.00019947434105556223,
      "loss": 23.0,
      "step": 573
    },
    {
      "epoch": 0.3445378151260504,
      "grad_norm": 0.0005377535708248615,
      "learning_rate": 0.0001994724048498982,
      "loss": 23.0,
      "step": 574
    },
    {
      "epoch": 0.34513805522208885,
      "grad_norm": 0.0005842957179993391,
      "learning_rate": 0.0001994704650943199,
      "loss": 23.0,
      "step": 575
    },
    {
      "epoch": 0.3457382953181273,
      "grad_norm": 0.0005292725982144475,
      "learning_rate": 0.00019946852178889658,
      "loss": 23.0,
      "step": 576
    },
    {
      "epoch": 0.34633853541416565,
      "grad_norm": 0.00041592164780013263,
      "learning_rate": 0.00019946657493369755,
      "loss": 23.0,
      "step": 577
    },
    {
      "epoch": 0.3469387755102041,
      "grad_norm": 0.0002317595499334857,
      "learning_rate": 0.00019946462452879237,
      "loss": 23.0,
      "step": 578
    },
    {
      "epoch": 0.3475390156062425,
      "grad_norm": 0.0005468165036290884,
      "learning_rate": 0.00019946267057425056,
      "loss": 23.0,
      "step": 579
    },
    {
      "epoch": 0.34813925570228094,
      "grad_norm": 0.0007788361399434507,
      "learning_rate": 0.00019946071307014188,
      "loss": 23.0,
      "step": 580
    },
    {
      "epoch": 0.3487394957983193,
      "grad_norm": 0.0006015602848492563,
      "learning_rate": 0.00019945875201653621,
      "loss": 23.0,
      "step": 581
    },
    {
      "epoch": 0.34933973589435774,
      "grad_norm": 0.0006226922268979251,
      "learning_rate": 0.00019945678741350352,
      "loss": 23.0,
      "step": 582
    },
    {
      "epoch": 0.34993997599039617,
      "grad_norm": 0.0010158922523260117,
      "learning_rate": 0.00019945481926111392,
      "loss": 23.0,
      "step": 583
    },
    {
      "epoch": 0.3505402160864346,
      "grad_norm": 0.0007838707533665001,
      "learning_rate": 0.00019945284755943765,
      "loss": 23.0,
      "step": 584
    },
    {
      "epoch": 0.35114045618247297,
      "grad_norm": 0.000780130154453218,
      "learning_rate": 0.00019945087230854505,
      "loss": 23.0,
      "step": 585
    },
    {
      "epoch": 0.3517406962785114,
      "grad_norm": 0.0010645820293575525,
      "learning_rate": 0.00019944889350850668,
      "loss": 23.0,
      "step": 586
    },
    {
      "epoch": 0.35234093637454983,
      "grad_norm": 0.0005978618864901364,
      "learning_rate": 0.00019944691115939307,
      "loss": 23.0,
      "step": 587
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.0006172691937536001,
      "learning_rate": 0.00019944492526127502,
      "loss": 23.0,
      "step": 588
    },
    {
      "epoch": 0.35354141656662663,
      "grad_norm": 0.0006004899623803794,
      "learning_rate": 0.00019944293581422343,
      "loss": 23.0,
      "step": 589
    },
    {
      "epoch": 0.35414165666266506,
      "grad_norm": 0.0004828960227314383,
      "learning_rate": 0.0001994409428183092,
      "loss": 23.0,
      "step": 590
    },
    {
      "epoch": 0.3547418967587035,
      "grad_norm": 0.00032999063841998577,
      "learning_rate": 0.00019943894627360355,
      "loss": 23.0,
      "step": 591
    },
    {
      "epoch": 0.3553421368547419,
      "grad_norm": 0.0004524400574155152,
      "learning_rate": 0.00019943694618017768,
      "loss": 23.0,
      "step": 592
    },
    {
      "epoch": 0.3559423769507803,
      "grad_norm": 0.0006170731503516436,
      "learning_rate": 0.000199434942538103,
      "loss": 23.0,
      "step": 593
    },
    {
      "epoch": 0.3565426170468187,
      "grad_norm": 0.0010504665551707149,
      "learning_rate": 0.00019943293534745098,
      "loss": 23.0,
      "step": 594
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.0006527103250846267,
      "learning_rate": 0.00019943092460829326,
      "loss": 23.0,
      "step": 595
    },
    {
      "epoch": 0.3577430972388956,
      "grad_norm": 0.0005603809840977192,
      "learning_rate": 0.00019942891032070163,
      "loss": 23.0,
      "step": 596
    },
    {
      "epoch": 0.35834333733493395,
      "grad_norm": 0.0008743616635911167,
      "learning_rate": 0.00019942689248474794,
      "loss": 23.0,
      "step": 597
    },
    {
      "epoch": 0.3589435774309724,
      "grad_norm": 0.0012338734231889248,
      "learning_rate": 0.00019942487110050422,
      "loss": 23.0,
      "step": 598
    },
    {
      "epoch": 0.3595438175270108,
      "grad_norm": 0.0007378676091320813,
      "learning_rate": 0.0001994228461680426,
      "loss": 23.0,
      "step": 599
    },
    {
      "epoch": 0.36014405762304924,
      "grad_norm": 0.0008557065157219768,
      "learning_rate": 0.00019942081768743535,
      "loss": 23.0,
      "step": 600
    },
    {
      "epoch": 0.3607442977190876,
      "grad_norm": 0.00044995357166044414,
      "learning_rate": 0.00019941878565875486,
      "loss": 23.0,
      "step": 601
    },
    {
      "epoch": 0.36134453781512604,
      "grad_norm": 0.0010641762055456638,
      "learning_rate": 0.00019941675008207366,
      "loss": 23.0,
      "step": 602
    },
    {
      "epoch": 0.3619447779111645,
      "grad_norm": 0.0004161446413490921,
      "learning_rate": 0.00019941471095746437,
      "loss": 23.0,
      "step": 603
    },
    {
      "epoch": 0.3625450180072029,
      "grad_norm": 0.0010861216578632593,
      "learning_rate": 0.00019941266828499973,
      "loss": 23.0,
      "step": 604
    },
    {
      "epoch": 0.3631452581032413,
      "grad_norm": 0.0004336018755566329,
      "learning_rate": 0.00019941062206475273,
      "loss": 23.0,
      "step": 605
    },
    {
      "epoch": 0.3637454981992797,
      "grad_norm": 0.0007256188546307385,
      "learning_rate": 0.0001994085722967963,
      "loss": 23.0,
      "step": 606
    },
    {
      "epoch": 0.36434573829531813,
      "grad_norm": 0.0005281938938423991,
      "learning_rate": 0.00019940651898120367,
      "loss": 23.0,
      "step": 607
    },
    {
      "epoch": 0.36494597839135656,
      "grad_norm": 0.00047858108882792294,
      "learning_rate": 0.00019940446211804807,
      "loss": 23.0,
      "step": 608
    },
    {
      "epoch": 0.36554621848739494,
      "grad_norm": 0.0003085435600951314,
      "learning_rate": 0.0001994024017074029,
      "loss": 23.0,
      "step": 609
    },
    {
      "epoch": 0.36614645858343337,
      "grad_norm": 0.00042341326479800045,
      "learning_rate": 0.00019940033774934173,
      "loss": 23.0,
      "step": 610
    },
    {
      "epoch": 0.3667466986794718,
      "grad_norm": 0.000549526244867593,
      "learning_rate": 0.00019939827024393817,
      "loss": 23.0,
      "step": 611
    },
    {
      "epoch": 0.3673469387755102,
      "grad_norm": 0.0006585200899280608,
      "learning_rate": 0.00019939619919126606,
      "loss": 23.0,
      "step": 612
    },
    {
      "epoch": 0.3679471788715486,
      "grad_norm": 0.000494246429298073,
      "learning_rate": 0.00019939412459139925,
      "loss": 23.0,
      "step": 613
    },
    {
      "epoch": 0.368547418967587,
      "grad_norm": 0.0006791771738789976,
      "learning_rate": 0.00019939204644441182,
      "loss": 23.0,
      "step": 614
    },
    {
      "epoch": 0.36914765906362546,
      "grad_norm": 0.0006217627669684589,
      "learning_rate": 0.00019938996475037793,
      "loss": 23.0,
      "step": 615
    },
    {
      "epoch": 0.3697478991596639,
      "grad_norm": 0.0007271217764355242,
      "learning_rate": 0.00019938787950937186,
      "loss": 23.0,
      "step": 616
    },
    {
      "epoch": 0.37034813925570226,
      "grad_norm": 0.000804447045084089,
      "learning_rate": 0.000199385790721468,
      "loss": 23.0,
      "step": 617
    },
    {
      "epoch": 0.3709483793517407,
      "grad_norm": 0.0016518638003617525,
      "learning_rate": 0.00019938369838674093,
      "loss": 23.0,
      "step": 618
    },
    {
      "epoch": 0.3715486194477791,
      "grad_norm": 0.0004141997778788209,
      "learning_rate": 0.00019938160250526532,
      "loss": 23.0,
      "step": 619
    },
    {
      "epoch": 0.37214885954381755,
      "grad_norm": 0.0006064656772650778,
      "learning_rate": 0.00019937950307711595,
      "loss": 23.0,
      "step": 620
    },
    {
      "epoch": 0.3727490996398559,
      "grad_norm": 0.0011940618278458714,
      "learning_rate": 0.00019937740010236773,
      "loss": 23.0,
      "step": 621
    },
    {
      "epoch": 0.37334933973589435,
      "grad_norm": 0.0008350508869625628,
      "learning_rate": 0.00019937529358109575,
      "loss": 23.0,
      "step": 622
    },
    {
      "epoch": 0.3739495798319328,
      "grad_norm": 0.0005108820623718202,
      "learning_rate": 0.00019937318351337515,
      "loss": 23.0,
      "step": 623
    },
    {
      "epoch": 0.3745498199279712,
      "grad_norm": 0.0004091697046533227,
      "learning_rate": 0.00019937106989928127,
      "loss": 23.0,
      "step": 624
    },
    {
      "epoch": 0.3751500600240096,
      "grad_norm": 0.0006039454601705074,
      "learning_rate": 0.00019936895273888952,
      "loss": 23.0,
      "step": 625
    },
    {
      "epoch": 0.375750300120048,
      "grad_norm": 0.0002572101657278836,
      "learning_rate": 0.00019936683203227544,
      "loss": 23.0,
      "step": 626
    },
    {
      "epoch": 0.37635054021608644,
      "grad_norm": 0.0002840797242242843,
      "learning_rate": 0.00019936470777951474,
      "loss": 23.0,
      "step": 627
    },
    {
      "epoch": 0.37695078031212487,
      "grad_norm": 0.0005371993756853044,
      "learning_rate": 0.0001993625799806832,
      "loss": 23.0,
      "step": 628
    },
    {
      "epoch": 0.37755102040816324,
      "grad_norm": 0.0005530430353246629,
      "learning_rate": 0.00019936044863585678,
      "loss": 23.0,
      "step": 629
    },
    {
      "epoch": 0.37815126050420167,
      "grad_norm": 0.0003712098114192486,
      "learning_rate": 0.00019935831374511153,
      "loss": 23.0,
      "step": 630
    },
    {
      "epoch": 0.3787515006002401,
      "grad_norm": 0.0006810548948124051,
      "learning_rate": 0.00019935617530852367,
      "loss": 23.0,
      "step": 631
    },
    {
      "epoch": 0.3793517406962785,
      "grad_norm": 0.0009463174501433969,
      "learning_rate": 0.00019935403332616944,
      "loss": 23.0,
      "step": 632
    },
    {
      "epoch": 0.3799519807923169,
      "grad_norm": 0.0003082946641370654,
      "learning_rate": 0.00019935188779812534,
      "loss": 23.0,
      "step": 633
    },
    {
      "epoch": 0.38055222088835533,
      "grad_norm": 0.0007540938095189631,
      "learning_rate": 0.00019934973872446797,
      "loss": 23.0,
      "step": 634
    },
    {
      "epoch": 0.38115246098439376,
      "grad_norm": 0.0006780321127735078,
      "learning_rate": 0.00019934758610527396,
      "loss": 23.0,
      "step": 635
    },
    {
      "epoch": 0.3817527010804322,
      "grad_norm": 0.0007656219531781971,
      "learning_rate": 0.00019934542994062013,
      "loss": 23.0,
      "step": 636
    },
    {
      "epoch": 0.38235294117647056,
      "grad_norm": 0.0005332346190698445,
      "learning_rate": 0.00019934327023058348,
      "loss": 23.0,
      "step": 637
    },
    {
      "epoch": 0.382953181272509,
      "grad_norm": 0.001408933661878109,
      "learning_rate": 0.00019934110697524104,
      "loss": 23.0,
      "step": 638
    },
    {
      "epoch": 0.3835534213685474,
      "grad_norm": 0.0006234024185687304,
      "learning_rate": 0.00019933894017467004,
      "loss": 23.0,
      "step": 639
    },
    {
      "epoch": 0.38415366146458585,
      "grad_norm": 0.0009115723660215735,
      "learning_rate": 0.0001993367698289478,
      "loss": 23.0,
      "step": 640
    },
    {
      "epoch": 0.3847539015606242,
      "grad_norm": 0.0003554839931894094,
      "learning_rate": 0.00019933459593815176,
      "loss": 23.0,
      "step": 641
    },
    {
      "epoch": 0.38535414165666265,
      "grad_norm": 0.0003548678068909794,
      "learning_rate": 0.0001993324185023595,
      "loss": 23.0,
      "step": 642
    },
    {
      "epoch": 0.3859543817527011,
      "grad_norm": 0.0009928151266649365,
      "learning_rate": 0.00019933023752164876,
      "loss": 23.0,
      "step": 643
    },
    {
      "epoch": 0.3865546218487395,
      "grad_norm": 0.0005392222083173692,
      "learning_rate": 0.00019932805299609734,
      "loss": 23.0,
      "step": 644
    },
    {
      "epoch": 0.3871548619447779,
      "grad_norm": 0.00045687207602895796,
      "learning_rate": 0.0001993258649257832,
      "loss": 23.0,
      "step": 645
    },
    {
      "epoch": 0.3877551020408163,
      "grad_norm": 0.0005536451353691518,
      "learning_rate": 0.00019932367331078444,
      "loss": 23.0,
      "step": 646
    },
    {
      "epoch": 0.38835534213685474,
      "grad_norm": 0.0006612985744141042,
      "learning_rate": 0.00019932147815117927,
      "loss": 23.0,
      "step": 647
    },
    {
      "epoch": 0.38895558223289317,
      "grad_norm": 0.0006857411353848875,
      "learning_rate": 0.00019931927944704604,
      "loss": 23.0,
      "step": 648
    },
    {
      "epoch": 0.3895558223289316,
      "grad_norm": 0.0006747473962605,
      "learning_rate": 0.0001993170771984632,
      "loss": 23.0,
      "step": 649
    },
    {
      "epoch": 0.39015606242497,
      "grad_norm": 0.0006530406535603106,
      "learning_rate": 0.00019931487140550935,
      "loss": 23.0,
      "step": 650
    },
    {
      "epoch": 0.3907563025210084,
      "grad_norm": 0.000865319452714175,
      "learning_rate": 0.00019931266206826321,
      "loss": 23.0,
      "step": 651
    },
    {
      "epoch": 0.39135654261704683,
      "grad_norm": 0.0008172982488758862,
      "learning_rate": 0.00019931044918680363,
      "loss": 23.0,
      "step": 652
    },
    {
      "epoch": 0.39195678271308526,
      "grad_norm": 0.0006401525461114943,
      "learning_rate": 0.00019930823276120957,
      "loss": 23.0,
      "step": 653
    },
    {
      "epoch": 0.39255702280912363,
      "grad_norm": 0.001018511364236474,
      "learning_rate": 0.00019930601279156013,
      "loss": 23.0,
      "step": 654
    },
    {
      "epoch": 0.39315726290516206,
      "grad_norm": 0.001182066509500146,
      "learning_rate": 0.00019930378927793453,
      "loss": 23.0,
      "step": 655
    },
    {
      "epoch": 0.3937575030012005,
      "grad_norm": 0.0002946962777059525,
      "learning_rate": 0.00019930156222041216,
      "loss": 23.0,
      "step": 656
    },
    {
      "epoch": 0.3943577430972389,
      "grad_norm": 0.0008485467988066375,
      "learning_rate": 0.00019929933161907246,
      "loss": 23.0,
      "step": 657
    },
    {
      "epoch": 0.3949579831932773,
      "grad_norm": 0.000641645397990942,
      "learning_rate": 0.00019929709747399502,
      "loss": 23.0,
      "step": 658
    },
    {
      "epoch": 0.3955582232893157,
      "grad_norm": 0.0013788726646453142,
      "learning_rate": 0.00019929485978525962,
      "loss": 23.0,
      "step": 659
    },
    {
      "epoch": 0.39615846338535415,
      "grad_norm": 0.0006469928775914013,
      "learning_rate": 0.0001992926185529461,
      "loss": 23.0,
      "step": 660
    },
    {
      "epoch": 0.3967587034813926,
      "grad_norm": 0.0006551746046170592,
      "learning_rate": 0.00019929037377713438,
      "loss": 23.0,
      "step": 661
    },
    {
      "epoch": 0.39735894357743096,
      "grad_norm": 0.0003588496765587479,
      "learning_rate": 0.00019928812545790464,
      "loss": 23.0,
      "step": 662
    },
    {
      "epoch": 0.3979591836734694,
      "grad_norm": 0.00034284719731658697,
      "learning_rate": 0.00019928587359533716,
      "loss": 23.0,
      "step": 663
    },
    {
      "epoch": 0.3985594237695078,
      "grad_norm": 0.0008505964069627225,
      "learning_rate": 0.0001992836181895122,
      "loss": 23.0,
      "step": 664
    },
    {
      "epoch": 0.39915966386554624,
      "grad_norm": 0.0005401921225711703,
      "learning_rate": 0.00019928135924051027,
      "loss": 23.0,
      "step": 665
    },
    {
      "epoch": 0.3997599039615846,
      "grad_norm": 0.0008277631714008749,
      "learning_rate": 0.00019927909674841208,
      "loss": 23.0,
      "step": 666
    },
    {
      "epoch": 0.40036014405762305,
      "grad_norm": 0.0004785002674907446,
      "learning_rate": 0.00019927683071329825,
      "loss": 23.0,
      "step": 667
    },
    {
      "epoch": 0.4009603841536615,
      "grad_norm": 0.0007532955496571958,
      "learning_rate": 0.00019927456113524972,
      "loss": 23.0,
      "step": 668
    },
    {
      "epoch": 0.4015606242496999,
      "grad_norm": 0.0003629917628131807,
      "learning_rate": 0.00019927228801434746,
      "loss": 23.0,
      "step": 669
    },
    {
      "epoch": 0.4021608643457383,
      "grad_norm": 0.0006129452376626432,
      "learning_rate": 0.0001992700113506726,
      "loss": 23.0,
      "step": 670
    },
    {
      "epoch": 0.4027611044417767,
      "grad_norm": 0.00048521102871745825,
      "learning_rate": 0.00019926773114430638,
      "loss": 23.0,
      "step": 671
    },
    {
      "epoch": 0.40336134453781514,
      "grad_norm": 0.0007425924413837492,
      "learning_rate": 0.0001992654473953302,
      "loss": 23.0,
      "step": 672
    },
    {
      "epoch": 0.40396158463385357,
      "grad_norm": 0.0003996037703473121,
      "learning_rate": 0.00019926316010382555,
      "loss": 23.0,
      "step": 673
    },
    {
      "epoch": 0.40456182472989194,
      "grad_norm": 0.0006881665321998298,
      "learning_rate": 0.00019926086926987403,
      "loss": 23.0,
      "step": 674
    },
    {
      "epoch": 0.40516206482593037,
      "grad_norm": 0.0005098696565255523,
      "learning_rate": 0.00019925857489355744,
      "loss": 23.0,
      "step": 675
    },
    {
      "epoch": 0.4057623049219688,
      "grad_norm": 0.000523019814863801,
      "learning_rate": 0.00019925627697495762,
      "loss": 23.0,
      "step": 676
    },
    {
      "epoch": 0.4063625450180072,
      "grad_norm": 0.0005873084883205593,
      "learning_rate": 0.0001992539755141566,
      "loss": 23.0,
      "step": 677
    },
    {
      "epoch": 0.4069627851140456,
      "grad_norm": 0.0009000878199003637,
      "learning_rate": 0.0001992516705112365,
      "loss": 23.0,
      "step": 678
    },
    {
      "epoch": 0.40756302521008403,
      "grad_norm": 0.00041991635225713253,
      "learning_rate": 0.0001992493619662796,
      "loss": 23.0,
      "step": 679
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 0.0004964503459632397,
      "learning_rate": 0.00019924704987936824,
      "loss": 23.0,
      "step": 680
    },
    {
      "epoch": 0.4087635054021609,
      "grad_norm": 0.0012351791374385357,
      "learning_rate": 0.00019924473425058498,
      "loss": 23.0,
      "step": 681
    },
    {
      "epoch": 0.40936374549819926,
      "grad_norm": 0.0009568947134539485,
      "learning_rate": 0.00019924241508001244,
      "loss": 23.0,
      "step": 682
    },
    {
      "epoch": 0.4099639855942377,
      "grad_norm": 0.0006957779405638576,
      "learning_rate": 0.00019924009236773339,
      "loss": 23.0,
      "step": 683
    },
    {
      "epoch": 0.4105642256902761,
      "grad_norm": 0.0007713912636972964,
      "learning_rate": 0.00019923776611383072,
      "loss": 23.0,
      "step": 684
    },
    {
      "epoch": 0.41116446578631455,
      "grad_norm": 0.0005927034071646631,
      "learning_rate": 0.00019923543631838743,
      "loss": 23.0,
      "step": 685
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 0.0012774481438100338,
      "learning_rate": 0.00019923310298148672,
      "loss": 23.0,
      "step": 686
    },
    {
      "epoch": 0.41236494597839135,
      "grad_norm": 0.0006231871666386724,
      "learning_rate": 0.0001992307661032118,
      "loss": 23.0,
      "step": 687
    },
    {
      "epoch": 0.4129651860744298,
      "grad_norm": 0.0008017731597647071,
      "learning_rate": 0.00019922842568364607,
      "loss": 23.0,
      "step": 688
    },
    {
      "epoch": 0.4135654261704682,
      "grad_norm": 0.0005019367672502995,
      "learning_rate": 0.0001992260817228731,
      "loss": 23.0,
      "step": 689
    },
    {
      "epoch": 0.4141656662665066,
      "grad_norm": 0.000567154842428863,
      "learning_rate": 0.0001992237342209765,
      "loss": 23.0,
      "step": 690
    },
    {
      "epoch": 0.414765906362545,
      "grad_norm": 0.00036567161441780627,
      "learning_rate": 0.00019922138317804007,
      "loss": 23.0,
      "step": 691
    },
    {
      "epoch": 0.41536614645858344,
      "grad_norm": 0.0007109650177881122,
      "learning_rate": 0.0001992190285941477,
      "loss": 23.0,
      "step": 692
    },
    {
      "epoch": 0.41596638655462187,
      "grad_norm": 0.000749673810787499,
      "learning_rate": 0.0001992166704693834,
      "loss": 23.0,
      "step": 693
    },
    {
      "epoch": 0.41656662665066024,
      "grad_norm": 0.0009893577080219984,
      "learning_rate": 0.00019921430880383138,
      "loss": 23.0,
      "step": 694
    },
    {
      "epoch": 0.4171668667466987,
      "grad_norm": 0.00036272848956286907,
      "learning_rate": 0.0001992119435975759,
      "loss": 23.0,
      "step": 695
    },
    {
      "epoch": 0.4177671068427371,
      "grad_norm": 0.0006051092641428113,
      "learning_rate": 0.00019920957485070132,
      "loss": 23.0,
      "step": 696
    },
    {
      "epoch": 0.41836734693877553,
      "grad_norm": 0.0004827500961255282,
      "learning_rate": 0.00019920720256329222,
      "loss": 23.0,
      "step": 697
    },
    {
      "epoch": 0.4189675870348139,
      "grad_norm": 0.001281979726627469,
      "learning_rate": 0.00019920482673543326,
      "loss": 23.0,
      "step": 698
    },
    {
      "epoch": 0.41956782713085233,
      "grad_norm": 0.0011693161213770509,
      "learning_rate": 0.00019920244736720922,
      "loss": 23.0,
      "step": 699
    },
    {
      "epoch": 0.42016806722689076,
      "grad_norm": 0.0006273601320572197,
      "learning_rate": 0.000199200064458705,
      "loss": 23.0,
      "step": 700
    },
    {
      "epoch": 0.4207683073229292,
      "grad_norm": 0.003010783577337861,
      "learning_rate": 0.00019919767801000564,
      "loss": 23.0,
      "step": 701
    },
    {
      "epoch": 0.42136854741896757,
      "grad_norm": 0.0008471604087390006,
      "learning_rate": 0.00019919528802119632,
      "loss": 23.0,
      "step": 702
    },
    {
      "epoch": 0.421968787515006,
      "grad_norm": 0.0008838615613058209,
      "learning_rate": 0.00019919289449236237,
      "loss": 23.0,
      "step": 703
    },
    {
      "epoch": 0.4225690276110444,
      "grad_norm": 0.0009554813150316477,
      "learning_rate": 0.00019919049742358916,
      "loss": 23.0,
      "step": 704
    },
    {
      "epoch": 0.42316926770708285,
      "grad_norm": 0.0005560766439884901,
      "learning_rate": 0.00019918809681496223,
      "loss": 23.0,
      "step": 705
    },
    {
      "epoch": 0.4237695078031212,
      "grad_norm": 0.0004707193002104759,
      "learning_rate": 0.00019918569266656725,
      "loss": 23.0,
      "step": 706
    },
    {
      "epoch": 0.42436974789915966,
      "grad_norm": 0.000530709105078131,
      "learning_rate": 0.00019918328497849007,
      "loss": 23.0,
      "step": 707
    },
    {
      "epoch": 0.4249699879951981,
      "grad_norm": 0.0006235828041099012,
      "learning_rate": 0.00019918087375081657,
      "loss": 23.0,
      "step": 708
    },
    {
      "epoch": 0.4255702280912365,
      "grad_norm": 0.000593838223721832,
      "learning_rate": 0.0001991784589836328,
      "loss": 23.0,
      "step": 709
    },
    {
      "epoch": 0.4261704681872749,
      "grad_norm": 0.0005683311028406024,
      "learning_rate": 0.00019917604067702495,
      "loss": 23.0,
      "step": 710
    },
    {
      "epoch": 0.4267707082833133,
      "grad_norm": 0.0009590789559297264,
      "learning_rate": 0.00019917361883107931,
      "loss": 23.0,
      "step": 711
    },
    {
      "epoch": 0.42737094837935174,
      "grad_norm": 0.0005070093320682645,
      "learning_rate": 0.00019917119344588235,
      "loss": 23.0,
      "step": 712
    },
    {
      "epoch": 0.4279711884753902,
      "grad_norm": 0.00040911248652264476,
      "learning_rate": 0.00019916876452152056,
      "loss": 23.0,
      "step": 713
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.00044645063462667167,
      "learning_rate": 0.00019916633205808068,
      "loss": 23.0,
      "step": 714
    },
    {
      "epoch": 0.429171668667467,
      "grad_norm": 0.0003959390742238611,
      "learning_rate": 0.0001991638960556495,
      "loss": 23.0,
      "step": 715
    },
    {
      "epoch": 0.4297719087635054,
      "grad_norm": 0.0006004471215419471,
      "learning_rate": 0.00019916145651431393,
      "loss": 23.0,
      "step": 716
    },
    {
      "epoch": 0.43037214885954383,
      "grad_norm": 0.00035068404395133257,
      "learning_rate": 0.00019915901343416103,
      "loss": 23.0,
      "step": 717
    },
    {
      "epoch": 0.4309723889555822,
      "grad_norm": 0.0003599017218220979,
      "learning_rate": 0.00019915656681527807,
      "loss": 23.0,
      "step": 718
    },
    {
      "epoch": 0.43157262905162064,
      "grad_norm": 0.0003325287252664566,
      "learning_rate": 0.00019915411665775225,
      "loss": 23.0,
      "step": 719
    },
    {
      "epoch": 0.43217286914765907,
      "grad_norm": 0.0007046267855912447,
      "learning_rate": 0.00019915166296167108,
      "loss": 23.0,
      "step": 720
    },
    {
      "epoch": 0.4327731092436975,
      "grad_norm": 0.0003630674909800291,
      "learning_rate": 0.0001991492057271221,
      "loss": 23.0,
      "step": 721
    },
    {
      "epoch": 0.43337334933973587,
      "grad_norm": 0.0003503916668705642,
      "learning_rate": 0.000199146744954193,
      "loss": 23.0,
      "step": 722
    },
    {
      "epoch": 0.4339735894357743,
      "grad_norm": 0.0007532882154919207,
      "learning_rate": 0.00019914428064297165,
      "loss": 23.0,
      "step": 723
    },
    {
      "epoch": 0.4345738295318127,
      "grad_norm": 0.0012118295999243855,
      "learning_rate": 0.0001991418127935459,
      "loss": 23.0,
      "step": 724
    },
    {
      "epoch": 0.43517406962785116,
      "grad_norm": 0.0005586919724009931,
      "learning_rate": 0.00019913934140600392,
      "loss": 23.0,
      "step": 725
    },
    {
      "epoch": 0.43577430972388953,
      "grad_norm": 0.0007068076520226896,
      "learning_rate": 0.00019913686648043383,
      "loss": 23.0,
      "step": 726
    },
    {
      "epoch": 0.43637454981992796,
      "grad_norm": 0.0005361078074201941,
      "learning_rate": 0.00019913438801692402,
      "loss": 23.0,
      "step": 727
    },
    {
      "epoch": 0.4369747899159664,
      "grad_norm": 0.0009473745012655854,
      "learning_rate": 0.00019913190601556288,
      "loss": 23.0,
      "step": 728
    },
    {
      "epoch": 0.4375750300120048,
      "grad_norm": 0.0005555340903811157,
      "learning_rate": 0.00019912942047643902,
      "loss": 23.0,
      "step": 729
    },
    {
      "epoch": 0.4381752701080432,
      "grad_norm": 0.0014016212662681937,
      "learning_rate": 0.0001991269313996411,
      "loss": 23.0,
      "step": 730
    },
    {
      "epoch": 0.4387755102040816,
      "grad_norm": 0.00027204706566408277,
      "learning_rate": 0.00019912443878525803,
      "loss": 23.0,
      "step": 731
    },
    {
      "epoch": 0.43937575030012005,
      "grad_norm": 0.0005901534459553659,
      "learning_rate": 0.0001991219426333787,
      "loss": 23.0,
      "step": 732
    },
    {
      "epoch": 0.4399759903961585,
      "grad_norm": 0.00031070367549546063,
      "learning_rate": 0.00019911944294409223,
      "loss": 23.0,
      "step": 733
    },
    {
      "epoch": 0.44057623049219685,
      "grad_norm": 0.0007562999962829053,
      "learning_rate": 0.00019911693971748777,
      "loss": 23.0,
      "step": 734
    },
    {
      "epoch": 0.4411764705882353,
      "grad_norm": 0.000757384579628706,
      "learning_rate": 0.00019911443295365472,
      "loss": 23.0,
      "step": 735
    },
    {
      "epoch": 0.4417767106842737,
      "grad_norm": 0.0009893305832520127,
      "learning_rate": 0.00019911192265268247,
      "loss": 23.0,
      "step": 736
    },
    {
      "epoch": 0.44237695078031214,
      "grad_norm": 0.0008873580954968929,
      "learning_rate": 0.0001991094088146607,
      "loss": 23.0,
      "step": 737
    },
    {
      "epoch": 0.4429771908763505,
      "grad_norm": 0.0006723484839312732,
      "learning_rate": 0.00019910689143967905,
      "loss": 23.0,
      "step": 738
    },
    {
      "epoch": 0.44357743097238894,
      "grad_norm": 0.0008835842600092292,
      "learning_rate": 0.00019910437052782736,
      "loss": 23.0,
      "step": 739
    },
    {
      "epoch": 0.44417767106842737,
      "grad_norm": 0.00031283125281333923,
      "learning_rate": 0.00019910184607919564,
      "loss": 23.0,
      "step": 740
    },
    {
      "epoch": 0.4447779111644658,
      "grad_norm": 0.0006608405965380371,
      "learning_rate": 0.00019909931809387394,
      "loss": 23.0,
      "step": 741
    },
    {
      "epoch": 0.44537815126050423,
      "grad_norm": 0.0005028719315305352,
      "learning_rate": 0.0001990967865719525,
      "loss": 23.0,
      "step": 742
    },
    {
      "epoch": 0.4459783913565426,
      "grad_norm": 0.000783328665420413,
      "learning_rate": 0.00019909425151352163,
      "loss": 23.0,
      "step": 743
    },
    {
      "epoch": 0.44657863145258103,
      "grad_norm": 0.0007136694621294737,
      "learning_rate": 0.00019909171291867183,
      "loss": 23.0,
      "step": 744
    },
    {
      "epoch": 0.44717887154861946,
      "grad_norm": 0.0006191918510012329,
      "learning_rate": 0.00019908917078749372,
      "loss": 23.0,
      "step": 745
    },
    {
      "epoch": 0.4477791116446579,
      "grad_norm": 0.0007277901167981327,
      "learning_rate": 0.00019908662512007797,
      "loss": 23.0,
      "step": 746
    },
    {
      "epoch": 0.44837935174069626,
      "grad_norm": 0.0006948673981241882,
      "learning_rate": 0.00019908407591651546,
      "loss": 23.0,
      "step": 747
    },
    {
      "epoch": 0.4489795918367347,
      "grad_norm": 0.0010942791122943163,
      "learning_rate": 0.00019908152317689714,
      "loss": 23.0,
      "step": 748
    },
    {
      "epoch": 0.4495798319327731,
      "grad_norm": 0.0005513243959285319,
      "learning_rate": 0.00019907896690131414,
      "loss": 23.0,
      "step": 749
    },
    {
      "epoch": 0.45018007202881155,
      "grad_norm": 0.0003519784368108958,
      "learning_rate": 0.00019907640708985766,
      "loss": 23.0,
      "step": 750
    },
    {
      "epoch": 0.4507803121248499,
      "grad_norm": 0.0008878299267962575,
      "learning_rate": 0.0001990738437426191,
      "loss": 23.0,
      "step": 751
    },
    {
      "epoch": 0.45138055222088835,
      "grad_norm": 0.0005296217277646065,
      "learning_rate": 0.00019907127685968984,
      "loss": 23.0,
      "step": 752
    },
    {
      "epoch": 0.4519807923169268,
      "grad_norm": 0.0009205266833305359,
      "learning_rate": 0.0001990687064411616,
      "loss": 23.0,
      "step": 753
    },
    {
      "epoch": 0.4525810324129652,
      "grad_norm": 0.0008753446745686233,
      "learning_rate": 0.00019906613248712606,
      "loss": 23.0,
      "step": 754
    },
    {
      "epoch": 0.4531812725090036,
      "grad_norm": 0.0008117556571960449,
      "learning_rate": 0.00019906355499767507,
      "loss": 23.0,
      "step": 755
    },
    {
      "epoch": 0.453781512605042,
      "grad_norm": 0.0009617914911359549,
      "learning_rate": 0.00019906097397290063,
      "loss": 23.0,
      "step": 756
    },
    {
      "epoch": 0.45438175270108044,
      "grad_norm": 0.0005935687222518027,
      "learning_rate": 0.00019905838941289484,
      "loss": 23.0,
      "step": 757
    },
    {
      "epoch": 0.4549819927971189,
      "grad_norm": 0.0007953408057801425,
      "learning_rate": 0.00019905580131774993,
      "loss": 23.0,
      "step": 758
    },
    {
      "epoch": 0.45558223289315725,
      "grad_norm": 0.0006210616556927562,
      "learning_rate": 0.00019905320968755828,
      "loss": 23.0,
      "step": 759
    },
    {
      "epoch": 0.4561824729891957,
      "grad_norm": 0.0006589915719814599,
      "learning_rate": 0.00019905061452241238,
      "loss": 23.0,
      "step": 760
    },
    {
      "epoch": 0.4567827130852341,
      "grad_norm": 0.0009153984137810767,
      "learning_rate": 0.00019904801582240484,
      "loss": 23.0,
      "step": 761
    },
    {
      "epoch": 0.45738295318127253,
      "grad_norm": 0.0017457930371165276,
      "learning_rate": 0.00019904541358762838,
      "loss": 23.0,
      "step": 762
    },
    {
      "epoch": 0.4579831932773109,
      "grad_norm": 0.0007837511948309839,
      "learning_rate": 0.0001990428078181759,
      "loss": 23.0,
      "step": 763
    },
    {
      "epoch": 0.45858343337334934,
      "grad_norm": 0.0005806999397464097,
      "learning_rate": 0.00019904019851414034,
      "loss": 23.0,
      "step": 764
    },
    {
      "epoch": 0.45918367346938777,
      "grad_norm": 0.0006374964723363519,
      "learning_rate": 0.0001990375856756149,
      "loss": 23.0,
      "step": 765
    },
    {
      "epoch": 0.4597839135654262,
      "grad_norm": 0.0009351090411655605,
      "learning_rate": 0.00019903496930269278,
      "loss": 23.0,
      "step": 766
    },
    {
      "epoch": 0.46038415366146457,
      "grad_norm": 0.0006809848127886653,
      "learning_rate": 0.00019903234939546734,
      "loss": 23.0,
      "step": 767
    },
    {
      "epoch": 0.460984393757503,
      "grad_norm": 0.000654295610729605,
      "learning_rate": 0.00019902972595403209,
      "loss": 23.0,
      "step": 768
    },
    {
      "epoch": 0.4615846338535414,
      "grad_norm": 0.0006075274432078004,
      "learning_rate": 0.0001990270989784807,
      "loss": 23.0,
      "step": 769
    },
    {
      "epoch": 0.46218487394957986,
      "grad_norm": 0.00044143499690108,
      "learning_rate": 0.0001990244684689068,
      "loss": 23.0,
      "step": 770
    },
    {
      "epoch": 0.46278511404561823,
      "grad_norm": 0.0013147240970283747,
      "learning_rate": 0.0001990218344254044,
      "loss": 23.0,
      "step": 771
    },
    {
      "epoch": 0.46338535414165666,
      "grad_norm": 0.00033725271350704134,
      "learning_rate": 0.0001990191968480674,
      "loss": 23.0,
      "step": 772
    },
    {
      "epoch": 0.4639855942376951,
      "grad_norm": 0.0009398331167176366,
      "learning_rate": 0.00019901655573698998,
      "loss": 23.0,
      "step": 773
    },
    {
      "epoch": 0.4645858343337335,
      "grad_norm": 0.0007297866977751255,
      "learning_rate": 0.0001990139110922664,
      "loss": 23.0,
      "step": 774
    },
    {
      "epoch": 0.4651860744297719,
      "grad_norm": 0.0007459200569428504,
      "learning_rate": 0.00019901126291399103,
      "loss": 23.0,
      "step": 775
    },
    {
      "epoch": 0.4657863145258103,
      "grad_norm": 0.0008847762364894152,
      "learning_rate": 0.0001990086112022584,
      "loss": 23.0,
      "step": 776
    },
    {
      "epoch": 0.46638655462184875,
      "grad_norm": 0.00059562484966591,
      "learning_rate": 0.0001990059559571631,
      "loss": 23.0,
      "step": 777
    },
    {
      "epoch": 0.4669867947178872,
      "grad_norm": 0.0007079718634486198,
      "learning_rate": 0.00019900329717879988,
      "loss": 23.0,
      "step": 778
    },
    {
      "epoch": 0.46758703481392555,
      "grad_norm": 0.0006528205703943968,
      "learning_rate": 0.00019900063486726364,
      "loss": 23.0,
      "step": 779
    },
    {
      "epoch": 0.468187274909964,
      "grad_norm": 0.0007662418647669256,
      "learning_rate": 0.00019899796902264944,
      "loss": 23.0,
      "step": 780
    },
    {
      "epoch": 0.4687875150060024,
      "grad_norm": 0.0009094589622691274,
      "learning_rate": 0.00019899529964505236,
      "loss": 23.0,
      "step": 781
    },
    {
      "epoch": 0.46938775510204084,
      "grad_norm": 0.0007676285458728671,
      "learning_rate": 0.0001989926267345677,
      "loss": 23.0,
      "step": 782
    },
    {
      "epoch": 0.4699879951980792,
      "grad_norm": 0.00048528422485105693,
      "learning_rate": 0.0001989899502912908,
      "loss": 23.0,
      "step": 783
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.0008444074192084372,
      "learning_rate": 0.00019898727031531722,
      "loss": 23.0,
      "step": 784
    },
    {
      "epoch": 0.47118847539015607,
      "grad_norm": 0.0017573406221345067,
      "learning_rate": 0.0001989845868067426,
      "loss": 23.0,
      "step": 785
    },
    {
      "epoch": 0.4717887154861945,
      "grad_norm": 0.0013762153685092926,
      "learning_rate": 0.0001989818997656627,
      "loss": 23.0,
      "step": 786
    },
    {
      "epoch": 0.4723889555822329,
      "grad_norm": 0.0007878051837906241,
      "learning_rate": 0.00019897920919217338,
      "loss": 23.0,
      "step": 787
    },
    {
      "epoch": 0.4729891956782713,
      "grad_norm": 0.0008741064229980111,
      "learning_rate": 0.0001989765150863707,
      "loss": 23.0,
      "step": 788
    },
    {
      "epoch": 0.47358943577430973,
      "grad_norm": 0.0005487430025823414,
      "learning_rate": 0.0001989738174483508,
      "loss": 23.0,
      "step": 789
    },
    {
      "epoch": 0.47418967587034816,
      "grad_norm": 0.0016626424621790648,
      "learning_rate": 0.00019897111627820998,
      "loss": 23.0,
      "step": 790
    },
    {
      "epoch": 0.47478991596638653,
      "grad_norm": 0.0006386625464074314,
      "learning_rate": 0.0001989684115760446,
      "loss": 23.0,
      "step": 791
    },
    {
      "epoch": 0.47539015606242496,
      "grad_norm": 0.0006513031548820436,
      "learning_rate": 0.00019896570334195112,
      "loss": 23.0,
      "step": 792
    },
    {
      "epoch": 0.4759903961584634,
      "grad_norm": 0.0005831916932947934,
      "learning_rate": 0.00019896299157602632,
      "loss": 23.0,
      "step": 793
    },
    {
      "epoch": 0.4765906362545018,
      "grad_norm": 0.0008199085714295506,
      "learning_rate": 0.0001989602762783669,
      "loss": 23.0,
      "step": 794
    },
    {
      "epoch": 0.4771908763505402,
      "grad_norm": 0.0009538012091070414,
      "learning_rate": 0.00019895755744906975,
      "loss": 23.0,
      "step": 795
    },
    {
      "epoch": 0.4777911164465786,
      "grad_norm": 0.0003274022601544857,
      "learning_rate": 0.00019895483508823193,
      "loss": 23.0,
      "step": 796
    },
    {
      "epoch": 0.47839135654261705,
      "grad_norm": 0.00060317013412714,
      "learning_rate": 0.0001989521091959506,
      "loss": 23.0,
      "step": 797
    },
    {
      "epoch": 0.4789915966386555,
      "grad_norm": 0.0006186945829540491,
      "learning_rate": 0.00019894937977232301,
      "loss": 23.0,
      "step": 798
    },
    {
      "epoch": 0.47959183673469385,
      "grad_norm": 0.0016297240508720279,
      "learning_rate": 0.0001989466468174466,
      "loss": 23.0,
      "step": 799
    },
    {
      "epoch": 0.4801920768307323,
      "grad_norm": 0.0004947784473188221,
      "learning_rate": 0.0001989439103314189,
      "loss": 23.0,
      "step": 800
    },
    {
      "epoch": 0.4807923169267707,
      "grad_norm": 0.0003158549952786416,
      "learning_rate": 0.00019894117031433756,
      "loss": 23.0,
      "step": 801
    },
    {
      "epoch": 0.48139255702280914,
      "grad_norm": 0.0005081993876956403,
      "learning_rate": 0.00019893842676630032,
      "loss": 23.0,
      "step": 802
    },
    {
      "epoch": 0.4819927971188475,
      "grad_norm": 0.0011078780516982079,
      "learning_rate": 0.00019893567968740517,
      "loss": 23.0,
      "step": 803
    },
    {
      "epoch": 0.48259303721488594,
      "grad_norm": 0.0006481404998339713,
      "learning_rate": 0.00019893292907775006,
      "loss": 23.0,
      "step": 804
    },
    {
      "epoch": 0.4831932773109244,
      "grad_norm": 0.001042915158905089,
      "learning_rate": 0.0001989301749374332,
      "loss": 23.0,
      "step": 805
    },
    {
      "epoch": 0.4837935174069628,
      "grad_norm": 0.0006100856116972864,
      "learning_rate": 0.00019892741726655294,
      "loss": 23.0,
      "step": 806
    },
    {
      "epoch": 0.4843937575030012,
      "grad_norm": 0.0006573795690201223,
      "learning_rate": 0.0001989246560652076,
      "loss": 23.0,
      "step": 807
    },
    {
      "epoch": 0.4849939975990396,
      "grad_norm": 0.0009064198238775134,
      "learning_rate": 0.00019892189133349572,
      "loss": 23.0,
      "step": 808
    },
    {
      "epoch": 0.48559423769507803,
      "grad_norm": 0.0012316376669332385,
      "learning_rate": 0.000198919123071516,
      "loss": 23.0,
      "step": 809
    },
    {
      "epoch": 0.48619447779111646,
      "grad_norm": 0.0006688730791211128,
      "learning_rate": 0.00019891635127936724,
      "loss": 23.0,
      "step": 810
    },
    {
      "epoch": 0.48679471788715484,
      "grad_norm": 0.0014101287815719843,
      "learning_rate": 0.00019891357595714835,
      "loss": 23.0,
      "step": 811
    },
    {
      "epoch": 0.48739495798319327,
      "grad_norm": 0.0005622003809548914,
      "learning_rate": 0.00019891079710495838,
      "loss": 23.0,
      "step": 812
    },
    {
      "epoch": 0.4879951980792317,
      "grad_norm": 0.000873049721121788,
      "learning_rate": 0.00019890801472289646,
      "loss": 23.0,
      "step": 813
    },
    {
      "epoch": 0.4885954381752701,
      "grad_norm": 0.0007714332896284759,
      "learning_rate": 0.00019890522881106193,
      "loss": 23.0,
      "step": 814
    },
    {
      "epoch": 0.4891956782713085,
      "grad_norm": 0.0008941619307734072,
      "learning_rate": 0.0001989024393695542,
      "loss": 23.0,
      "step": 815
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 0.0009259075159206986,
      "learning_rate": 0.00019889964639847283,
      "loss": 23.0,
      "step": 816
    },
    {
      "epoch": 0.49039615846338536,
      "grad_norm": 0.0004040625353809446,
      "learning_rate": 0.00019889684989791745,
      "loss": 23.0,
      "step": 817
    },
    {
      "epoch": 0.4909963985594238,
      "grad_norm": 0.0011171556543558836,
      "learning_rate": 0.0001988940498679879,
      "loss": 23.0,
      "step": 818
    },
    {
      "epoch": 0.49159663865546216,
      "grad_norm": 0.00037770162452943623,
      "learning_rate": 0.0001988912463087841,
      "loss": 23.0,
      "step": 819
    },
    {
      "epoch": 0.4921968787515006,
      "grad_norm": 0.00041134696220979095,
      "learning_rate": 0.00019888843922040612,
      "loss": 23.0,
      "step": 820
    },
    {
      "epoch": 0.492797118847539,
      "grad_norm": 0.0006298330845311284,
      "learning_rate": 0.00019888562860295408,
      "loss": 23.0,
      "step": 821
    },
    {
      "epoch": 0.49339735894357745,
      "grad_norm": 0.0006055657286196947,
      "learning_rate": 0.00019888281445652837,
      "loss": 23.0,
      "step": 822
    },
    {
      "epoch": 0.4939975990396158,
      "grad_norm": 0.001203834661282599,
      "learning_rate": 0.00019887999678122934,
      "loss": 23.0,
      "step": 823
    },
    {
      "epoch": 0.49459783913565425,
      "grad_norm": 0.0006844853051006794,
      "learning_rate": 0.00019887717557715755,
      "loss": 23.0,
      "step": 824
    },
    {
      "epoch": 0.4951980792316927,
      "grad_norm": 0.0004448948602657765,
      "learning_rate": 0.00019887435084441372,
      "loss": 23.0,
      "step": 825
    },
    {
      "epoch": 0.4957983193277311,
      "grad_norm": 0.0004217177047394216,
      "learning_rate": 0.00019887152258309863,
      "loss": 23.0,
      "step": 826
    },
    {
      "epoch": 0.4963985594237695,
      "grad_norm": 0.0006263357936404645,
      "learning_rate": 0.00019886869079331328,
      "loss": 23.0,
      "step": 827
    },
    {
      "epoch": 0.4969987995198079,
      "grad_norm": 0.0007113023893907666,
      "learning_rate": 0.00019886585547515862,
      "loss": 23.0,
      "step": 828
    },
    {
      "epoch": 0.49759903961584634,
      "grad_norm": 0.0014177537523210049,
      "learning_rate": 0.00019886301662873593,
      "loss": 23.0,
      "step": 829
    },
    {
      "epoch": 0.49819927971188477,
      "grad_norm": 0.0008082055719569325,
      "learning_rate": 0.00019886017425414643,
      "loss": 23.0,
      "step": 830
    },
    {
      "epoch": 0.49879951980792314,
      "grad_norm": 0.0005536688840948045,
      "learning_rate": 0.00019885732835149168,
      "loss": 23.0,
      "step": 831
    },
    {
      "epoch": 0.49939975990396157,
      "grad_norm": 0.0006400597630999982,
      "learning_rate": 0.00019885447892087314,
      "loss": 23.0,
      "step": 832
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.0005208631628192961,
      "learning_rate": 0.00019885162596239254,
      "loss": 23.0,
      "step": 833
    },
    {
      "epoch": 0.5006002400960384,
      "grad_norm": 0.0007002478232607245,
      "learning_rate": 0.00019884876947615166,
      "loss": 23.0,
      "step": 834
    },
    {
      "epoch": 0.5012004801920769,
      "grad_norm": 0.0004122959799133241,
      "learning_rate": 0.00019884590946225248,
      "loss": 23.0,
      "step": 835
    },
    {
      "epoch": 0.5018007202881153,
      "grad_norm": 0.0006469865329563618,
      "learning_rate": 0.00019884304592079707,
      "loss": 23.0,
      "step": 836
    },
    {
      "epoch": 0.5024009603841537,
      "grad_norm": 0.000518444343470037,
      "learning_rate": 0.00019884017885188758,
      "loss": 23.0,
      "step": 837
    },
    {
      "epoch": 0.503001200480192,
      "grad_norm": 0.0012191816931590438,
      "learning_rate": 0.00019883730825562637,
      "loss": 23.0,
      "step": 838
    },
    {
      "epoch": 0.5036014405762305,
      "grad_norm": 0.0007649417966604233,
      "learning_rate": 0.00019883443413211586,
      "loss": 23.0,
      "step": 839
    },
    {
      "epoch": 0.5042016806722689,
      "grad_norm": 0.0008528571343049407,
      "learning_rate": 0.00019883155648145863,
      "loss": 23.0,
      "step": 840
    },
    {
      "epoch": 0.5048019207683073,
      "grad_norm": 0.0006200968637131155,
      "learning_rate": 0.00019882867530375737,
      "loss": 23.0,
      "step": 841
    },
    {
      "epoch": 0.5054021608643458,
      "grad_norm": 0.0011856872588396072,
      "learning_rate": 0.00019882579059911492,
      "loss": 23.0,
      "step": 842
    },
    {
      "epoch": 0.5060024009603842,
      "grad_norm": 0.0005892259650863707,
      "learning_rate": 0.0001988229023676342,
      "loss": 23.0,
      "step": 843
    },
    {
      "epoch": 0.5066026410564226,
      "grad_norm": 0.001340447342954576,
      "learning_rate": 0.00019882001060941832,
      "loss": 23.0,
      "step": 844
    },
    {
      "epoch": 0.507202881152461,
      "grad_norm": 0.0006858596461825073,
      "learning_rate": 0.00019881711532457043,
      "loss": 23.0,
      "step": 845
    },
    {
      "epoch": 0.5078031212484994,
      "grad_norm": 0.0013606062857434154,
      "learning_rate": 0.0001988142165131939,
      "loss": 23.0,
      "step": 846
    },
    {
      "epoch": 0.5084033613445378,
      "grad_norm": 0.0010817131260409951,
      "learning_rate": 0.00019881131417539215,
      "loss": 23.0,
      "step": 847
    },
    {
      "epoch": 0.5090036014405762,
      "grad_norm": 0.0007884291699156165,
      "learning_rate": 0.00019880840831126877,
      "loss": 23.0,
      "step": 848
    },
    {
      "epoch": 0.5096038415366146,
      "grad_norm": 0.0008213819819502532,
      "learning_rate": 0.00019880549892092747,
      "loss": 23.0,
      "step": 849
    },
    {
      "epoch": 0.5102040816326531,
      "grad_norm": 0.0007805051864124835,
      "learning_rate": 0.0001988025860044721,
      "loss": 23.0,
      "step": 850
    },
    {
      "epoch": 0.5108043217286915,
      "grad_norm": 0.001384717645123601,
      "learning_rate": 0.00019879966956200655,
      "loss": 23.0,
      "step": 851
    },
    {
      "epoch": 0.5114045618247299,
      "grad_norm": 0.0005536239477805793,
      "learning_rate": 0.0001987967495936349,
      "loss": 23.0,
      "step": 852
    },
    {
      "epoch": 0.5120048019207684,
      "grad_norm": 0.0010214439826086164,
      "learning_rate": 0.00019879382609946147,
      "loss": 23.0,
      "step": 853
    },
    {
      "epoch": 0.5126050420168067,
      "grad_norm": 0.0006562453927472234,
      "learning_rate": 0.00019879089907959047,
      "loss": 23.0,
      "step": 854
    },
    {
      "epoch": 0.5132052821128451,
      "grad_norm": 0.0006304988055489957,
      "learning_rate": 0.00019878796853412643,
      "loss": 23.0,
      "step": 855
    },
    {
      "epoch": 0.5138055222088835,
      "grad_norm": 0.0008602405432611704,
      "learning_rate": 0.00019878503446317387,
      "loss": 23.0,
      "step": 856
    },
    {
      "epoch": 0.514405762304922,
      "grad_norm": 0.00042852506157942116,
      "learning_rate": 0.00019878209686683754,
      "loss": 23.0,
      "step": 857
    },
    {
      "epoch": 0.5150060024009604,
      "grad_norm": 0.0005161712178960443,
      "learning_rate": 0.00019877915574522233,
      "loss": 23.0,
      "step": 858
    },
    {
      "epoch": 0.5156062424969988,
      "grad_norm": 0.000647779437713325,
      "learning_rate": 0.00019877621109843307,
      "loss": 23.0,
      "step": 859
    },
    {
      "epoch": 0.5162064825930373,
      "grad_norm": 0.0008087221649475396,
      "learning_rate": 0.00019877326292657498,
      "loss": 23.0,
      "step": 860
    },
    {
      "epoch": 0.5168067226890757,
      "grad_norm": 0.0024992094840854406,
      "learning_rate": 0.00019877031122975315,
      "loss": 23.0,
      "step": 861
    },
    {
      "epoch": 0.517406962785114,
      "grad_norm": 0.0006555477739311755,
      "learning_rate": 0.00019876735600807299,
      "loss": 23.0,
      "step": 862
    },
    {
      "epoch": 0.5180072028811524,
      "grad_norm": 0.00048210759996436536,
      "learning_rate": 0.00019876439726164,
      "loss": 23.0,
      "step": 863
    },
    {
      "epoch": 0.5186074429771909,
      "grad_norm": 0.0006051068194210529,
      "learning_rate": 0.00019876143499055968,
      "loss": 23.0,
      "step": 864
    },
    {
      "epoch": 0.5192076830732293,
      "grad_norm": 0.0005166485789231956,
      "learning_rate": 0.00019875846919493778,
      "loss": 23.0,
      "step": 865
    },
    {
      "epoch": 0.5198079231692677,
      "grad_norm": 0.00048096582759171724,
      "learning_rate": 0.0001987554998748802,
      "loss": 23.0,
      "step": 866
    },
    {
      "epoch": 0.5204081632653061,
      "grad_norm": 0.0007085447432473302,
      "learning_rate": 0.00019875252703049285,
      "loss": 23.0,
      "step": 867
    },
    {
      "epoch": 0.5210084033613446,
      "grad_norm": 0.0005757782491855323,
      "learning_rate": 0.0001987495506618818,
      "loss": 23.0,
      "step": 868
    },
    {
      "epoch": 0.521608643457383,
      "grad_norm": 0.0017492113402113318,
      "learning_rate": 0.00019874657076915332,
      "loss": 23.0,
      "step": 869
    },
    {
      "epoch": 0.5222088835534213,
      "grad_norm": 0.00044582822010852396,
      "learning_rate": 0.00019874358735241375,
      "loss": 23.0,
      "step": 870
    },
    {
      "epoch": 0.5228091236494598,
      "grad_norm": 0.0008416534401476383,
      "learning_rate": 0.00019874060041176952,
      "loss": 23.0,
      "step": 871
    },
    {
      "epoch": 0.5234093637454982,
      "grad_norm": 0.0010793217225000262,
      "learning_rate": 0.00019873760994732726,
      "loss": 23.0,
      "step": 872
    },
    {
      "epoch": 0.5240096038415366,
      "grad_norm": 0.000765673175919801,
      "learning_rate": 0.00019873461595919367,
      "loss": 23.0,
      "step": 873
    },
    {
      "epoch": 0.524609843937575,
      "grad_norm": 0.004173364955931902,
      "learning_rate": 0.00019873161844747564,
      "loss": 23.0,
      "step": 874
    },
    {
      "epoch": 0.5252100840336135,
      "grad_norm": 0.00026861258083954453,
      "learning_rate": 0.00019872861741228011,
      "loss": 23.0,
      "step": 875
    },
    {
      "epoch": 0.5258103241296519,
      "grad_norm": 0.0014661707682535052,
      "learning_rate": 0.0001987256128537142,
      "loss": 23.0,
      "step": 876
    },
    {
      "epoch": 0.5264105642256903,
      "grad_norm": 0.0008170696091838181,
      "learning_rate": 0.0001987226047718851,
      "loss": 23.0,
      "step": 877
    },
    {
      "epoch": 0.5270108043217286,
      "grad_norm": 0.00043921300675719976,
      "learning_rate": 0.00019871959316690015,
      "loss": 23.0,
      "step": 878
    },
    {
      "epoch": 0.5276110444177671,
      "grad_norm": 0.0004785111523233354,
      "learning_rate": 0.0001987165780388669,
      "loss": 23.0,
      "step": 879
    },
    {
      "epoch": 0.5282112845138055,
      "grad_norm": 0.0008355296449735761,
      "learning_rate": 0.00019871355938789288,
      "loss": 23.0,
      "step": 880
    },
    {
      "epoch": 0.5288115246098439,
      "grad_norm": 0.0006070987437851727,
      "learning_rate": 0.00019871053721408588,
      "loss": 23.0,
      "step": 881
    },
    {
      "epoch": 0.5294117647058824,
      "grad_norm": 0.000812305836006999,
      "learning_rate": 0.00019870751151755372,
      "loss": 23.0,
      "step": 882
    },
    {
      "epoch": 0.5300120048019208,
      "grad_norm": 0.0011360760545358062,
      "learning_rate": 0.00019870448229840435,
      "loss": 23.0,
      "step": 883
    },
    {
      "epoch": 0.5306122448979592,
      "grad_norm": 0.000924605003092438,
      "learning_rate": 0.0001987014495567459,
      "loss": 23.0,
      "step": 884
    },
    {
      "epoch": 0.5312124849939976,
      "grad_norm": 0.0007006872328929603,
      "learning_rate": 0.00019869841329268664,
      "loss": 23.0,
      "step": 885
    },
    {
      "epoch": 0.531812725090036,
      "grad_norm": 0.0006443238235078752,
      "learning_rate": 0.00019869537350633486,
      "loss": 23.0,
      "step": 886
    },
    {
      "epoch": 0.5324129651860744,
      "grad_norm": 0.0008577117114327848,
      "learning_rate": 0.0001986923301977991,
      "loss": 23.0,
      "step": 887
    },
    {
      "epoch": 0.5330132052821128,
      "grad_norm": 0.0007688856567256153,
      "learning_rate": 0.00019868928336718792,
      "loss": 23.0,
      "step": 888
    },
    {
      "epoch": 0.5336134453781513,
      "grad_norm": 0.0004415331350173801,
      "learning_rate": 0.0001986862330146101,
      "loss": 23.0,
      "step": 889
    },
    {
      "epoch": 0.5342136854741897,
      "grad_norm": 0.0004884483059868217,
      "learning_rate": 0.00019868317914017446,
      "loss": 23.0,
      "step": 890
    },
    {
      "epoch": 0.5348139255702281,
      "grad_norm": 0.0005003849510103464,
      "learning_rate": 0.00019868012174399,
      "loss": 23.0,
      "step": 891
    },
    {
      "epoch": 0.5354141656662665,
      "grad_norm": 0.0009174185106530786,
      "learning_rate": 0.00019867706082616585,
      "loss": 23.0,
      "step": 892
    },
    {
      "epoch": 0.536014405762305,
      "grad_norm": 0.0009697056957520545,
      "learning_rate": 0.0001986739963868112,
      "loss": 23.0,
      "step": 893
    },
    {
      "epoch": 0.5366146458583433,
      "grad_norm": 0.000463727192254737,
      "learning_rate": 0.00019867092842603548,
      "loss": 23.0,
      "step": 894
    },
    {
      "epoch": 0.5372148859543817,
      "grad_norm": 0.0004660410340875387,
      "learning_rate": 0.0001986678569439481,
      "loss": 23.0,
      "step": 895
    },
    {
      "epoch": 0.5378151260504201,
      "grad_norm": 0.0007628395687788725,
      "learning_rate": 0.00019866478194065871,
      "loss": 23.0,
      "step": 896
    },
    {
      "epoch": 0.5384153661464586,
      "grad_norm": 0.0005763001972809434,
      "learning_rate": 0.00019866170341627704,
      "loss": 23.0,
      "step": 897
    },
    {
      "epoch": 0.539015606242497,
      "grad_norm": 0.0007523773238062859,
      "learning_rate": 0.000198658621370913,
      "loss": 23.0,
      "step": 898
    },
    {
      "epoch": 0.5396158463385354,
      "grad_norm": 0.00048348799464292824,
      "learning_rate": 0.0001986555358046765,
      "loss": 23.0,
      "step": 899
    },
    {
      "epoch": 0.5402160864345739,
      "grad_norm": 0.0004331173258833587,
      "learning_rate": 0.00019865244671767772,
      "loss": 23.0,
      "step": 900
    },
    {
      "epoch": 0.5408163265306123,
      "grad_norm": 0.0005526962340809405,
      "learning_rate": 0.0001986493541100269,
      "loss": 23.0,
      "step": 901
    },
    {
      "epoch": 0.5414165666266506,
      "grad_norm": 0.0016207111766561866,
      "learning_rate": 0.00019864625798183435,
      "loss": 23.0,
      "step": 902
    },
    {
      "epoch": 0.542016806722689,
      "grad_norm": 0.0004141308309044689,
      "learning_rate": 0.00019864315833321064,
      "loss": 23.0,
      "step": 903
    },
    {
      "epoch": 0.5426170468187275,
      "grad_norm": 0.0006656160694546998,
      "learning_rate": 0.00019864005516426634,
      "loss": 23.0,
      "step": 904
    },
    {
      "epoch": 0.5432172869147659,
      "grad_norm": 0.0007548914290964603,
      "learning_rate": 0.00019863694847511217,
      "loss": 23.0,
      "step": 905
    },
    {
      "epoch": 0.5438175270108043,
      "grad_norm": 0.0011304601794108748,
      "learning_rate": 0.00019863383826585905,
      "loss": 23.0,
      "step": 906
    },
    {
      "epoch": 0.5444177671068428,
      "grad_norm": 0.0009428642224520445,
      "learning_rate": 0.000198630724536618,
      "loss": 23.0,
      "step": 907
    },
    {
      "epoch": 0.5450180072028812,
      "grad_norm": 0.0005785591201856732,
      "learning_rate": 0.00019862760728750004,
      "loss": 23.0,
      "step": 908
    },
    {
      "epoch": 0.5456182472989196,
      "grad_norm": 0.00038587188464589417,
      "learning_rate": 0.00019862448651861652,
      "loss": 23.0,
      "step": 909
    },
    {
      "epoch": 0.5462184873949579,
      "grad_norm": 0.00045680327457375824,
      "learning_rate": 0.00019862136223007876,
      "loss": 23.0,
      "step": 910
    },
    {
      "epoch": 0.5468187274909964,
      "grad_norm": 0.0005812095478177071,
      "learning_rate": 0.00019861823442199824,
      "loss": 23.0,
      "step": 911
    },
    {
      "epoch": 0.5474189675870348,
      "grad_norm": 0.0011624370235949755,
      "learning_rate": 0.00019861510309448664,
      "loss": 23.0,
      "step": 912
    },
    {
      "epoch": 0.5480192076830732,
      "grad_norm": 0.0005838541546836495,
      "learning_rate": 0.00019861196824765568,
      "loss": 23.0,
      "step": 913
    },
    {
      "epoch": 0.5486194477791116,
      "grad_norm": 0.0005749358097091317,
      "learning_rate": 0.0001986088298816172,
      "loss": 23.0,
      "step": 914
    },
    {
      "epoch": 0.5492196878751501,
      "grad_norm": 0.0007014405564405024,
      "learning_rate": 0.00019860568799648327,
      "loss": 23.0,
      "step": 915
    },
    {
      "epoch": 0.5498199279711885,
      "grad_norm": 0.0005809336435049772,
      "learning_rate": 0.00019860254259236596,
      "loss": 23.0,
      "step": 916
    },
    {
      "epoch": 0.5504201680672269,
      "grad_norm": 0.0006069474620744586,
      "learning_rate": 0.00019859939366937755,
      "loss": 23.0,
      "step": 917
    },
    {
      "epoch": 0.5510204081632653,
      "grad_norm": 0.0013985512778162956,
      "learning_rate": 0.0001985962412276304,
      "loss": 23.0,
      "step": 918
    },
    {
      "epoch": 0.5516206482593037,
      "grad_norm": 0.0009238467318937182,
      "learning_rate": 0.000198593085267237,
      "loss": 23.0,
      "step": 919
    },
    {
      "epoch": 0.5522208883553421,
      "grad_norm": 0.0011213804828003049,
      "learning_rate": 0.00019858992578831005,
      "loss": 23.0,
      "step": 920
    },
    {
      "epoch": 0.5528211284513805,
      "grad_norm": 0.0006154056172817945,
      "learning_rate": 0.0001985867627909622,
      "loss": 23.0,
      "step": 921
    },
    {
      "epoch": 0.553421368547419,
      "grad_norm": 0.000759779242798686,
      "learning_rate": 0.00019858359627530643,
      "loss": 23.0,
      "step": 922
    },
    {
      "epoch": 0.5540216086434574,
      "grad_norm": 0.0006431901128962636,
      "learning_rate": 0.00019858042624145566,
      "loss": 23.0,
      "step": 923
    },
    {
      "epoch": 0.5546218487394958,
      "grad_norm": 0.0006690253503620625,
      "learning_rate": 0.00019857725268952305,
      "loss": 23.0,
      "step": 924
    },
    {
      "epoch": 0.5552220888355343,
      "grad_norm": 0.0008324479567818344,
      "learning_rate": 0.0001985740756196219,
      "loss": 23.0,
      "step": 925
    },
    {
      "epoch": 0.5558223289315727,
      "grad_norm": 0.0006556141306646168,
      "learning_rate": 0.00019857089503186553,
      "loss": 23.0,
      "step": 926
    },
    {
      "epoch": 0.556422569027611,
      "grad_norm": 0.000579280371312052,
      "learning_rate": 0.00019856771092636747,
      "loss": 23.0,
      "step": 927
    },
    {
      "epoch": 0.5570228091236494,
      "grad_norm": 0.0006807451136410236,
      "learning_rate": 0.00019856452330324135,
      "loss": 23.0,
      "step": 928
    },
    {
      "epoch": 0.5576230492196879,
      "grad_norm": 0.0007975809858180583,
      "learning_rate": 0.00019856133216260096,
      "loss": 23.0,
      "step": 929
    },
    {
      "epoch": 0.5582232893157263,
      "grad_norm": 0.0006549794343300164,
      "learning_rate": 0.0001985581375045601,
      "loss": 23.0,
      "step": 930
    },
    {
      "epoch": 0.5588235294117647,
      "grad_norm": 0.0008131754002533853,
      "learning_rate": 0.00019855493932923287,
      "loss": 23.0,
      "step": 931
    },
    {
      "epoch": 0.5594237695078031,
      "grad_norm": 0.0004512158047873527,
      "learning_rate": 0.0001985517376367334,
      "loss": 23.0,
      "step": 932
    },
    {
      "epoch": 0.5600240096038416,
      "grad_norm": 0.0005497694946825504,
      "learning_rate": 0.00019854853242717589,
      "loss": 23.0,
      "step": 933
    },
    {
      "epoch": 0.56062424969988,
      "grad_norm": 0.0007320971344597638,
      "learning_rate": 0.00019854532370067472,
      "loss": 23.0,
      "step": 934
    },
    {
      "epoch": 0.5612244897959183,
      "grad_norm": 0.0007082552183419466,
      "learning_rate": 0.0001985421114573445,
      "loss": 23.0,
      "step": 935
    },
    {
      "epoch": 0.5618247298919568,
      "grad_norm": 0.0010231130290776491,
      "learning_rate": 0.00019853889569729978,
      "loss": 23.0,
      "step": 936
    },
    {
      "epoch": 0.5624249699879952,
      "grad_norm": 0.0009601555648259819,
      "learning_rate": 0.0001985356764206553,
      "loss": 23.0,
      "step": 937
    },
    {
      "epoch": 0.5630252100840336,
      "grad_norm": 0.0006255342741496861,
      "learning_rate": 0.00019853245362752605,
      "loss": 23.0,
      "step": 938
    },
    {
      "epoch": 0.563625450180072,
      "grad_norm": 0.0004980456433258951,
      "learning_rate": 0.00019852922731802695,
      "loss": 23.0,
      "step": 939
    },
    {
      "epoch": 0.5642256902761105,
      "grad_norm": 0.0007995213381946087,
      "learning_rate": 0.0001985259974922732,
      "loss": 23.0,
      "step": 940
    },
    {
      "epoch": 0.5648259303721489,
      "grad_norm": 0.00110067508649081,
      "learning_rate": 0.00019852276415038004,
      "loss": 23.0,
      "step": 941
    },
    {
      "epoch": 0.5654261704681873,
      "grad_norm": 0.001249002176336944,
      "learning_rate": 0.00019851952729246284,
      "loss": 23.0,
      "step": 942
    },
    {
      "epoch": 0.5660264105642256,
      "grad_norm": 0.0004925851826556027,
      "learning_rate": 0.00019851628691863713,
      "loss": 23.0,
      "step": 943
    },
    {
      "epoch": 0.5666266506602641,
      "grad_norm": 0.0005772880977019668,
      "learning_rate": 0.00019851304302901857,
      "loss": 23.0,
      "step": 944
    },
    {
      "epoch": 0.5672268907563025,
      "grad_norm": 0.0006800185656175017,
      "learning_rate": 0.0001985097956237229,
      "loss": 23.0,
      "step": 945
    },
    {
      "epoch": 0.5678271308523409,
      "grad_norm": 0.0006493727560155094,
      "learning_rate": 0.000198506544702866,
      "loss": 23.0,
      "step": 946
    },
    {
      "epoch": 0.5684273709483794,
      "grad_norm": 0.0012511987006291747,
      "learning_rate": 0.00019850329026656395,
      "loss": 23.0,
      "step": 947
    },
    {
      "epoch": 0.5690276110444178,
      "grad_norm": 0.0010138124926015735,
      "learning_rate": 0.00019850003231493285,
      "loss": 23.0,
      "step": 948
    },
    {
      "epoch": 0.5696278511404562,
      "grad_norm": 0.0007604805869050324,
      "learning_rate": 0.00019849677084808894,
      "loss": 23.0,
      "step": 949
    },
    {
      "epoch": 0.5702280912364946,
      "grad_norm": 0.0005830854061059654,
      "learning_rate": 0.00019849350586614866,
      "loss": 23.0,
      "step": 950
    },
    {
      "epoch": 0.570828331332533,
      "grad_norm": 0.0005887901061214507,
      "learning_rate": 0.00019849023736922853,
      "loss": 23.0,
      "step": 951
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.001255355542525649,
      "learning_rate": 0.00019848696535744513,
      "loss": 23.0,
      "step": 952
    },
    {
      "epoch": 0.5720288115246098,
      "grad_norm": 0.0004925208049826324,
      "learning_rate": 0.0001984836898309153,
      "loss": 23.0,
      "step": 953
    },
    {
      "epoch": 0.5726290516206483,
      "grad_norm": 0.0006673120078630745,
      "learning_rate": 0.0001984804107897559,
      "loss": 23.0,
      "step": 954
    },
    {
      "epoch": 0.5732292917166867,
      "grad_norm": 0.00044477067422121763,
      "learning_rate": 0.00019847712823408396,
      "loss": 23.0,
      "step": 955
    },
    {
      "epoch": 0.5738295318127251,
      "grad_norm": 0.0007815766730345786,
      "learning_rate": 0.00019847384216401665,
      "loss": 23.0,
      "step": 956
    },
    {
      "epoch": 0.5744297719087635,
      "grad_norm": 0.0005144941969774663,
      "learning_rate": 0.0001984705525796712,
      "loss": 23.0,
      "step": 957
    },
    {
      "epoch": 0.575030012004802,
      "grad_norm": 0.00047222990542650223,
      "learning_rate": 0.000198467259481165,
      "loss": 23.0,
      "step": 958
    },
    {
      "epoch": 0.5756302521008403,
      "grad_norm": 0.0008695271681062877,
      "learning_rate": 0.00019846396286861566,
      "loss": 23.0,
      "step": 959
    },
    {
      "epoch": 0.5762304921968787,
      "grad_norm": 0.0004319838772062212,
      "learning_rate": 0.00019846066274214073,
      "loss": 23.0,
      "step": 960
    },
    {
      "epoch": 0.5768307322929171,
      "grad_norm": 0.0007976375636644661,
      "learning_rate": 0.000198457359101858,
      "loss": 23.0,
      "step": 961
    },
    {
      "epoch": 0.5774309723889556,
      "grad_norm": 0.0004369649104773998,
      "learning_rate": 0.00019845405194788543,
      "loss": 23.0,
      "step": 962
    },
    {
      "epoch": 0.578031212484994,
      "grad_norm": 0.0010789295192807913,
      "learning_rate": 0.00019845074128034096,
      "loss": 23.0,
      "step": 963
    },
    {
      "epoch": 0.5786314525810324,
      "grad_norm": 0.00045877910451963544,
      "learning_rate": 0.0001984474270993428,
      "loss": 23.0,
      "step": 964
    },
    {
      "epoch": 0.5792316926770709,
      "grad_norm": 0.0005742406356148422,
      "learning_rate": 0.00019844410940500918,
      "loss": 23.0,
      "step": 965
    },
    {
      "epoch": 0.5798319327731093,
      "grad_norm": 0.0008709350368008018,
      "learning_rate": 0.00019844078819745856,
      "loss": 23.0,
      "step": 966
    },
    {
      "epoch": 0.5804321728691476,
      "grad_norm": 0.0005361419753171504,
      "learning_rate": 0.0001984374634768094,
      "loss": 23.0,
      "step": 967
    },
    {
      "epoch": 0.581032412965186,
      "grad_norm": 0.0003321192634757608,
      "learning_rate": 0.0001984341352431804,
      "loss": 23.0,
      "step": 968
    },
    {
      "epoch": 0.5816326530612245,
      "grad_norm": 0.0009687889833003283,
      "learning_rate": 0.0001984308034966903,
      "loss": 23.0,
      "step": 969
    },
    {
      "epoch": 0.5822328931572629,
      "grad_norm": 0.0008406118722632527,
      "learning_rate": 0.00019842746823745802,
      "loss": 23.0,
      "step": 970
    },
    {
      "epoch": 0.5828331332533013,
      "grad_norm": 0.0004772779066115618,
      "learning_rate": 0.00019842412946560257,
      "loss": 23.0,
      "step": 971
    },
    {
      "epoch": 0.5834333733493398,
      "grad_norm": 0.0014819619245827198,
      "learning_rate": 0.00019842078718124316,
      "loss": 23.0,
      "step": 972
    },
    {
      "epoch": 0.5840336134453782,
      "grad_norm": 0.001416862360201776,
      "learning_rate": 0.00019841744138449899,
      "loss": 23.0,
      "step": 973
    },
    {
      "epoch": 0.5846338535414166,
      "grad_norm": 0.0005799952778033912,
      "learning_rate": 0.00019841409207548951,
      "loss": 23.0,
      "step": 974
    },
    {
      "epoch": 0.5852340936374549,
      "grad_norm": 0.0013107365230098367,
      "learning_rate": 0.00019841073925433423,
      "loss": 23.0,
      "step": 975
    },
    {
      "epoch": 0.5858343337334934,
      "grad_norm": 0.0022494373843073845,
      "learning_rate": 0.0001984073829211528,
      "loss": 23.0,
      "step": 976
    },
    {
      "epoch": 0.5864345738295318,
      "grad_norm": 0.00047388195525854826,
      "learning_rate": 0.000198404023076065,
      "loss": 23.0,
      "step": 977
    },
    {
      "epoch": 0.5870348139255702,
      "grad_norm": 0.0004757082206197083,
      "learning_rate": 0.0001984006597191908,
      "loss": 23.0,
      "step": 978
    },
    {
      "epoch": 0.5876350540216086,
      "grad_norm": 0.0005135858664289117,
      "learning_rate": 0.00019839729285065012,
      "loss": 23.0,
      "step": 979
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.0009471832890994847,
      "learning_rate": 0.00019839392247056318,
      "loss": 23.0,
      "step": 980
    },
    {
      "epoch": 0.5888355342136855,
      "grad_norm": 0.0007709990604780614,
      "learning_rate": 0.00019839054857905022,
      "loss": 23.0,
      "step": 981
    },
    {
      "epoch": 0.5894357743097239,
      "grad_norm": 0.00041485187830403447,
      "learning_rate": 0.0001983871711762317,
      "loss": 23.0,
      "step": 982
    },
    {
      "epoch": 0.5900360144057623,
      "grad_norm": 0.0010062240762636065,
      "learning_rate": 0.0001983837902622281,
      "loss": 23.0,
      "step": 983
    },
    {
      "epoch": 0.5906362545018007,
      "grad_norm": 0.0022086831741034985,
      "learning_rate": 0.00019838040583716014,
      "loss": 23.0,
      "step": 984
    },
    {
      "epoch": 0.5912364945978391,
      "grad_norm": 0.000324814987834543,
      "learning_rate": 0.00019837701790114851,
      "loss": 23.0,
      "step": 985
    },
    {
      "epoch": 0.5918367346938775,
      "grad_norm": 0.0006208634586073458,
      "learning_rate": 0.0001983736264543142,
      "loss": 23.0,
      "step": 986
    },
    {
      "epoch": 0.592436974789916,
      "grad_norm": 0.000827543786726892,
      "learning_rate": 0.00019837023149677822,
      "loss": 23.0,
      "step": 987
    },
    {
      "epoch": 0.5930372148859544,
      "grad_norm": 0.0008125030435621738,
      "learning_rate": 0.0001983668330286617,
      "loss": 23.0,
      "step": 988
    },
    {
      "epoch": 0.5936374549819928,
      "grad_norm": 0.0006292484467849135,
      "learning_rate": 0.00019836343105008594,
      "loss": 23.0,
      "step": 989
    },
    {
      "epoch": 0.5942376950780313,
      "grad_norm": 0.0005044650752097368,
      "learning_rate": 0.00019836002556117234,
      "loss": 23.0,
      "step": 990
    },
    {
      "epoch": 0.5948379351740696,
      "grad_norm": 0.0005759747582487762,
      "learning_rate": 0.00019835661656204243,
      "loss": 23.0,
      "step": 991
    },
    {
      "epoch": 0.595438175270108,
      "grad_norm": 0.0010958920465782285,
      "learning_rate": 0.0001983532040528179,
      "loss": 23.0,
      "step": 992
    },
    {
      "epoch": 0.5960384153661464,
      "grad_norm": 0.0005467525334097445,
      "learning_rate": 0.0001983497880336205,
      "loss": 23.0,
      "step": 993
    },
    {
      "epoch": 0.5966386554621849,
      "grad_norm": 0.0010392843978479505,
      "learning_rate": 0.00019834636850457217,
      "loss": 23.0,
      "step": 994
    },
    {
      "epoch": 0.5972388955582233,
      "grad_norm": 0.0006499423179775476,
      "learning_rate": 0.00019834294546579488,
      "loss": 23.0,
      "step": 995
    },
    {
      "epoch": 0.5978391356542617,
      "grad_norm": 0.00035032519372180104,
      "learning_rate": 0.00019833951891741087,
      "loss": 23.0,
      "step": 996
    },
    {
      "epoch": 0.5984393757503002,
      "grad_norm": 0.0006970508839003742,
      "learning_rate": 0.00019833608885954238,
      "loss": 23.0,
      "step": 997
    },
    {
      "epoch": 0.5990396158463386,
      "grad_norm": 0.0005337274633347988,
      "learning_rate": 0.00019833265529231185,
      "loss": 23.0,
      "step": 998
    },
    {
      "epoch": 0.5996398559423769,
      "grad_norm": 0.0004186936712358147,
      "learning_rate": 0.0001983292182158418,
      "loss": 23.0,
      "step": 999
    },
    {
      "epoch": 0.6002400960384153,
      "grad_norm": 0.0005397711065597832,
      "learning_rate": 0.00019832577763025485,
      "loss": 23.0,
      "step": 1000
    },
    {
      "epoch": 0.6002400960384153,
      "eval_loss": 11.5,
      "eval_runtime": 5.4733,
      "eval_samples_per_second": 256.333,
      "eval_steps_per_second": 32.156,
      "step": 1000
    },
    {
      "epoch": 0.6008403361344538,
      "grad_norm": 0.0008261637412942946,
      "learning_rate": 0.00019832233353567381,
      "loss": 23.0,
      "step": 1001
    },
    {
      "epoch": 0.6014405762304922,
      "grad_norm": 0.00048322940710932016,
      "learning_rate": 0.00019831888593222163,
      "loss": 23.0,
      "step": 1002
    },
    {
      "epoch": 0.6020408163265306,
      "grad_norm": 0.0007284404709935188,
      "learning_rate": 0.0001983154348200213,
      "loss": 23.0,
      "step": 1003
    },
    {
      "epoch": 0.602641056422569,
      "grad_norm": 0.0010770062217488885,
      "learning_rate": 0.00019831198019919604,
      "loss": 23.0,
      "step": 1004
    },
    {
      "epoch": 0.6032412965186075,
      "grad_norm": 0.000601259816903621,
      "learning_rate": 0.00019830852206986907,
      "loss": 23.0,
      "step": 1005
    },
    {
      "epoch": 0.6038415366146459,
      "grad_norm": 0.0005424935952760279,
      "learning_rate": 0.00019830506043216382,
      "loss": 23.0,
      "step": 1006
    },
    {
      "epoch": 0.6044417767106842,
      "grad_norm": 0.0010248576290905476,
      "learning_rate": 0.00019830159528620383,
      "loss": 23.0,
      "step": 1007
    },
    {
      "epoch": 0.6050420168067226,
      "grad_norm": 0.0003622387012001127,
      "learning_rate": 0.00019829812663211275,
      "loss": 23.0,
      "step": 1008
    },
    {
      "epoch": 0.6056422569027611,
      "grad_norm": 0.0007084689568728209,
      "learning_rate": 0.00019829465447001443,
      "loss": 23.0,
      "step": 1009
    },
    {
      "epoch": 0.6062424969987995,
      "grad_norm": 0.0012313174083828926,
      "learning_rate": 0.00019829117880003268,
      "loss": 23.0,
      "step": 1010
    },
    {
      "epoch": 0.6068427370948379,
      "grad_norm": 0.0007114025647751987,
      "learning_rate": 0.00019828769962229164,
      "loss": 23.0,
      "step": 1011
    },
    {
      "epoch": 0.6074429771908764,
      "grad_norm": 0.0008022873662412167,
      "learning_rate": 0.0001982842169369154,
      "loss": 23.0,
      "step": 1012
    },
    {
      "epoch": 0.6080432172869148,
      "grad_norm": 0.00036245142109692097,
      "learning_rate": 0.0001982807307440283,
      "loss": 23.0,
      "step": 1013
    },
    {
      "epoch": 0.6086434573829532,
      "grad_norm": 0.0004387908265925944,
      "learning_rate": 0.00019827724104375474,
      "loss": 23.0,
      "step": 1014
    },
    {
      "epoch": 0.6092436974789915,
      "grad_norm": 0.0004411818808875978,
      "learning_rate": 0.0001982737478362192,
      "loss": 23.0,
      "step": 1015
    },
    {
      "epoch": 0.60984393757503,
      "grad_norm": 0.0005459621315822005,
      "learning_rate": 0.0001982702511215464,
      "loss": 23.0,
      "step": 1016
    },
    {
      "epoch": 0.6104441776710684,
      "grad_norm": 0.0004670721828006208,
      "learning_rate": 0.00019826675089986115,
      "loss": 23.0,
      "step": 1017
    },
    {
      "epoch": 0.6110444177671068,
      "grad_norm": 0.0008358248160220683,
      "learning_rate": 0.00019826324717128833,
      "loss": 23.0,
      "step": 1018
    },
    {
      "epoch": 0.6116446578631453,
      "grad_norm": 0.0003749076568055898,
      "learning_rate": 0.00019825973993595295,
      "loss": 23.0,
      "step": 1019
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 0.0014997563557699323,
      "learning_rate": 0.0001982562291939802,
      "loss": 23.0,
      "step": 1020
    },
    {
      "epoch": 0.6128451380552221,
      "grad_norm": 0.0007296651601791382,
      "learning_rate": 0.0001982527149454954,
      "loss": 23.0,
      "step": 1021
    },
    {
      "epoch": 0.6134453781512605,
      "grad_norm": 0.0010369520168751478,
      "learning_rate": 0.00019824919719062397,
      "loss": 23.0,
      "step": 1022
    },
    {
      "epoch": 0.614045618247299,
      "grad_norm": 0.0006567711825482547,
      "learning_rate": 0.00019824567592949136,
      "loss": 23.0,
      "step": 1023
    },
    {
      "epoch": 0.6146458583433373,
      "grad_norm": 0.0005550686037167907,
      "learning_rate": 0.0001982421511622233,
      "loss": 23.0,
      "step": 1024
    },
    {
      "epoch": 0.6152460984393757,
      "grad_norm": 0.0005852804169990122,
      "learning_rate": 0.0001982386228889456,
      "loss": 23.0,
      "step": 1025
    },
    {
      "epoch": 0.6158463385354142,
      "grad_norm": 0.0011896343203261495,
      "learning_rate": 0.00019823509110978416,
      "loss": 23.0,
      "step": 1026
    },
    {
      "epoch": 0.6164465786314526,
      "grad_norm": 0.00104072829708457,
      "learning_rate": 0.000198231555824865,
      "loss": 23.0,
      "step": 1027
    },
    {
      "epoch": 0.617046818727491,
      "grad_norm": 0.0006339827086776495,
      "learning_rate": 0.00019822801703431425,
      "loss": 23.0,
      "step": 1028
    },
    {
      "epoch": 0.6176470588235294,
      "grad_norm": 0.0007693166844546795,
      "learning_rate": 0.00019822447473825833,
      "loss": 23.0,
      "step": 1029
    },
    {
      "epoch": 0.6182472989195679,
      "grad_norm": 0.0009480812004767358,
      "learning_rate": 0.0001982209289368235,
      "loss": 23.0,
      "step": 1030
    },
    {
      "epoch": 0.6188475390156063,
      "grad_norm": 0.0013159072259441018,
      "learning_rate": 0.00019821737963013637,
      "loss": 23.0,
      "step": 1031
    },
    {
      "epoch": 0.6194477791116446,
      "grad_norm": 0.0008248360245488584,
      "learning_rate": 0.00019821382681832367,
      "loss": 23.0,
      "step": 1032
    },
    {
      "epoch": 0.620048019207683,
      "grad_norm": 0.0014143993612378836,
      "learning_rate": 0.00019821027050151205,
      "loss": 23.0,
      "step": 1033
    },
    {
      "epoch": 0.6206482593037215,
      "grad_norm": 0.0009804586879909039,
      "learning_rate": 0.00019820671067982855,
      "loss": 23.0,
      "step": 1034
    },
    {
      "epoch": 0.6212484993997599,
      "grad_norm": 0.0003080831083934754,
      "learning_rate": 0.00019820314735340017,
      "loss": 23.0,
      "step": 1035
    },
    {
      "epoch": 0.6218487394957983,
      "grad_norm": 0.000753725238610059,
      "learning_rate": 0.00019819958052235404,
      "loss": 23.0,
      "step": 1036
    },
    {
      "epoch": 0.6224489795918368,
      "grad_norm": 0.0003446524788159877,
      "learning_rate": 0.0001981960101868175,
      "loss": 23.0,
      "step": 1037
    },
    {
      "epoch": 0.6230492196878752,
      "grad_norm": 0.001190772163681686,
      "learning_rate": 0.0001981924363469179,
      "loss": 23.0,
      "step": 1038
    },
    {
      "epoch": 0.6236494597839136,
      "grad_norm": 0.0011959177209064364,
      "learning_rate": 0.00019818885900278288,
      "loss": 23.0,
      "step": 1039
    },
    {
      "epoch": 0.6242496998799519,
      "grad_norm": 0.0007084035314619541,
      "learning_rate": 0.00019818527815454004,
      "loss": 23.0,
      "step": 1040
    },
    {
      "epoch": 0.6248499399759904,
      "grad_norm": 0.0007333476096391678,
      "learning_rate": 0.00019818169380231718,
      "loss": 23.0,
      "step": 1041
    },
    {
      "epoch": 0.6254501800720288,
      "grad_norm": 0.0005077448440715671,
      "learning_rate": 0.00019817810594624223,
      "loss": 23.0,
      "step": 1042
    },
    {
      "epoch": 0.6260504201680672,
      "grad_norm": 0.001495958655141294,
      "learning_rate": 0.00019817451458644322,
      "loss": 23.0,
      "step": 1043
    },
    {
      "epoch": 0.6266506602641057,
      "grad_norm": 0.0008460984681732953,
      "learning_rate": 0.0001981709197230483,
      "loss": 23.0,
      "step": 1044
    },
    {
      "epoch": 0.6272509003601441,
      "grad_norm": 0.0012502010213211179,
      "learning_rate": 0.00019816732135618582,
      "loss": 23.0,
      "step": 1045
    },
    {
      "epoch": 0.6278511404561825,
      "grad_norm": 0.0005108893965370953,
      "learning_rate": 0.00019816371948598414,
      "loss": 23.0,
      "step": 1046
    },
    {
      "epoch": 0.6284513805522209,
      "grad_norm": 0.0008723620558157563,
      "learning_rate": 0.0001981601141125718,
      "loss": 23.0,
      "step": 1047
    },
    {
      "epoch": 0.6290516206482593,
      "grad_norm": 0.00039634722634218633,
      "learning_rate": 0.0001981565052360775,
      "loss": 23.0,
      "step": 1048
    },
    {
      "epoch": 0.6296518607442977,
      "grad_norm": 0.0007444315706379712,
      "learning_rate": 0.00019815289285663004,
      "loss": 23.0,
      "step": 1049
    },
    {
      "epoch": 0.6302521008403361,
      "grad_norm": 0.0008287147502414882,
      "learning_rate": 0.00019814927697435827,
      "loss": 23.0,
      "step": 1050
    },
    {
      "epoch": 0.6308523409363745,
      "grad_norm": 0.0009593195863999426,
      "learning_rate": 0.00019814565758939133,
      "loss": 23.0,
      "step": 1051
    },
    {
      "epoch": 0.631452581032413,
      "grad_norm": 0.0005941663403064013,
      "learning_rate": 0.00019814203470185828,
      "loss": 23.0,
      "step": 1052
    },
    {
      "epoch": 0.6320528211284514,
      "grad_norm": 0.0003965641662944108,
      "learning_rate": 0.00019813840831188847,
      "loss": 23.0,
      "step": 1053
    },
    {
      "epoch": 0.6326530612244898,
      "grad_norm": 0.0006090254755690694,
      "learning_rate": 0.00019813477841961133,
      "loss": 23.0,
      "step": 1054
    },
    {
      "epoch": 0.6332533013205283,
      "grad_norm": 0.0005143271991983056,
      "learning_rate": 0.00019813114502515634,
      "loss": 23.0,
      "step": 1055
    },
    {
      "epoch": 0.6338535414165666,
      "grad_norm": 0.0008223920594900846,
      "learning_rate": 0.00019812750812865326,
      "loss": 23.0,
      "step": 1056
    },
    {
      "epoch": 0.634453781512605,
      "grad_norm": 0.0006902985624037683,
      "learning_rate": 0.00019812386773023183,
      "loss": 23.0,
      "step": 1057
    },
    {
      "epoch": 0.6350540216086434,
      "grad_norm": 0.0007466610404662788,
      "learning_rate": 0.00019812022383002193,
      "loss": 23.0,
      "step": 1058
    },
    {
      "epoch": 0.6356542617046819,
      "grad_norm": 0.0008086318848654628,
      "learning_rate": 0.00019811657642815363,
      "loss": 23.0,
      "step": 1059
    },
    {
      "epoch": 0.6362545018007203,
      "grad_norm": 0.0014404834946617484,
      "learning_rate": 0.00019811292552475714,
      "loss": 23.0,
      "step": 1060
    },
    {
      "epoch": 0.6368547418967587,
      "grad_norm": 0.00042763081728480756,
      "learning_rate": 0.00019810927111996269,
      "loss": 23.0,
      "step": 1061
    },
    {
      "epoch": 0.6374549819927972,
      "grad_norm": 0.0004761995223816484,
      "learning_rate": 0.00019810561321390073,
      "loss": 23.0,
      "step": 1062
    },
    {
      "epoch": 0.6380552220888356,
      "grad_norm": 0.0008606476476415992,
      "learning_rate": 0.0001981019518067018,
      "loss": 23.0,
      "step": 1063
    },
    {
      "epoch": 0.6386554621848739,
      "grad_norm": 0.0010334079852327704,
      "learning_rate": 0.00019809828689849654,
      "loss": 23.0,
      "step": 1064
    },
    {
      "epoch": 0.6392557022809123,
      "grad_norm": 0.0007184183341450989,
      "learning_rate": 0.00019809461848941577,
      "loss": 23.0,
      "step": 1065
    },
    {
      "epoch": 0.6398559423769508,
      "grad_norm": 0.0006008398486301303,
      "learning_rate": 0.0001980909465795904,
      "loss": 23.0,
      "step": 1066
    },
    {
      "epoch": 0.6404561824729892,
      "grad_norm": 0.0009349811589345336,
      "learning_rate": 0.00019808727116915145,
      "loss": 23.0,
      "step": 1067
    },
    {
      "epoch": 0.6410564225690276,
      "grad_norm": 0.001079434179700911,
      "learning_rate": 0.0001980835922582301,
      "loss": 23.0,
      "step": 1068
    },
    {
      "epoch": 0.641656662665066,
      "grad_norm": 0.00046767835738137364,
      "learning_rate": 0.0001980799098469577,
      "loss": 23.0,
      "step": 1069
    },
    {
      "epoch": 0.6422569027611045,
      "grad_norm": 0.0007817531586624682,
      "learning_rate": 0.0001980762239354655,
      "loss": 23.0,
      "step": 1070
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.0009396449313499033,
      "learning_rate": 0.00019807253452388526,
      "loss": 23.0,
      "step": 1071
    },
    {
      "epoch": 0.6434573829531812,
      "grad_norm": 0.0006175727467052639,
      "learning_rate": 0.0001980688416123485,
      "loss": 23.0,
      "step": 1072
    },
    {
      "epoch": 0.6440576230492197,
      "grad_norm": 0.00031634268816560507,
      "learning_rate": 0.000198065145200987,
      "loss": 23.0,
      "step": 1073
    },
    {
      "epoch": 0.6446578631452581,
      "grad_norm": 0.00112092902418226,
      "learning_rate": 0.00019806144528993277,
      "loss": 23.0,
      "step": 1074
    },
    {
      "epoch": 0.6452581032412965,
      "grad_norm": 0.0004822269838768989,
      "learning_rate": 0.00019805774187931782,
      "loss": 23.0,
      "step": 1075
    },
    {
      "epoch": 0.6458583433373349,
      "grad_norm": 0.0007649666513316333,
      "learning_rate": 0.00019805403496927425,
      "loss": 23.0,
      "step": 1076
    },
    {
      "epoch": 0.6464585834333734,
      "grad_norm": 0.0004432103887666017,
      "learning_rate": 0.00019805032455993442,
      "loss": 23.0,
      "step": 1077
    },
    {
      "epoch": 0.6470588235294118,
      "grad_norm": 0.0005666416836902499,
      "learning_rate": 0.00019804661065143071,
      "loss": 23.0,
      "step": 1078
    },
    {
      "epoch": 0.6476590636254502,
      "grad_norm": 0.0007401452166959643,
      "learning_rate": 0.00019804289324389568,
      "loss": 23.0,
      "step": 1079
    },
    {
      "epoch": 0.6482593037214885,
      "grad_norm": 0.0005741413333453238,
      "learning_rate": 0.000198039172337462,
      "loss": 23.0,
      "step": 1080
    },
    {
      "epoch": 0.648859543817527,
      "grad_norm": 0.001217698329128325,
      "learning_rate": 0.0001980354479322624,
      "loss": 23.0,
      "step": 1081
    },
    {
      "epoch": 0.6494597839135654,
      "grad_norm": 0.0010777065763249993,
      "learning_rate": 0.00019803172002842988,
      "loss": 23.0,
      "step": 1082
    },
    {
      "epoch": 0.6500600240096038,
      "grad_norm": 0.0005868837470188737,
      "learning_rate": 0.00019802798862609744,
      "loss": 23.0,
      "step": 1083
    },
    {
      "epoch": 0.6506602641056423,
      "grad_norm": 0.0004865597002208233,
      "learning_rate": 0.00019802425372539824,
      "loss": 23.0,
      "step": 1084
    },
    {
      "epoch": 0.6512605042016807,
      "grad_norm": 0.0012064909096807241,
      "learning_rate": 0.00019802051532646556,
      "loss": 23.0,
      "step": 1085
    },
    {
      "epoch": 0.6518607442977191,
      "grad_norm": 0.000426503480412066,
      "learning_rate": 0.00019801677342943283,
      "loss": 23.0,
      "step": 1086
    },
    {
      "epoch": 0.6524609843937575,
      "grad_norm": 0.0006327254232019186,
      "learning_rate": 0.0001980130280344336,
      "loss": 23.0,
      "step": 1087
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 0.0007822368061169982,
      "learning_rate": 0.0001980092791416015,
      "loss": 23.0,
      "step": 1088
    },
    {
      "epoch": 0.6536614645858343,
      "grad_norm": 0.0005497146630659699,
      "learning_rate": 0.00019800552675107035,
      "loss": 23.0,
      "step": 1089
    },
    {
      "epoch": 0.6542617046818727,
      "grad_norm": 0.0010286032920703292,
      "learning_rate": 0.00019800177086297404,
      "loss": 23.0,
      "step": 1090
    },
    {
      "epoch": 0.6548619447779112,
      "grad_norm": 0.0011467989534139633,
      "learning_rate": 0.00019799801147744664,
      "loss": 23.0,
      "step": 1091
    },
    {
      "epoch": 0.6554621848739496,
      "grad_norm": 0.0007217766833491623,
      "learning_rate": 0.00019799424859462227,
      "loss": 23.0,
      "step": 1092
    },
    {
      "epoch": 0.656062424969988,
      "grad_norm": 0.0006906822673045099,
      "learning_rate": 0.00019799048221463524,
      "loss": 23.0,
      "step": 1093
    },
    {
      "epoch": 0.6566626650660264,
      "grad_norm": 0.0004381495527923107,
      "learning_rate": 0.00019798671233761997,
      "loss": 23.0,
      "step": 1094
    },
    {
      "epoch": 0.6572629051620649,
      "grad_norm": 0.000589704723097384,
      "learning_rate": 0.00019798293896371099,
      "loss": 23.0,
      "step": 1095
    },
    {
      "epoch": 0.6578631452581032,
      "grad_norm": 0.0009171977289952338,
      "learning_rate": 0.00019797916209304294,
      "loss": 23.0,
      "step": 1096
    },
    {
      "epoch": 0.6584633853541416,
      "grad_norm": 0.000955081544816494,
      "learning_rate": 0.00019797538172575067,
      "loss": 23.0,
      "step": 1097
    },
    {
      "epoch": 0.65906362545018,
      "grad_norm": 0.000602062267716974,
      "learning_rate": 0.00019797159786196902,
      "loss": 23.0,
      "step": 1098
    },
    {
      "epoch": 0.6596638655462185,
      "grad_norm": 0.0008331533172167838,
      "learning_rate": 0.00019796781050183306,
      "loss": 23.0,
      "step": 1099
    },
    {
      "epoch": 0.6602641056422569,
      "grad_norm": 0.0007971031591296196,
      "learning_rate": 0.00019796401964547794,
      "loss": 23.0,
      "step": 1100
    },
    {
      "epoch": 0.6608643457382953,
      "grad_norm": 0.0007683509029448032,
      "learning_rate": 0.00019796022529303895,
      "loss": 23.0,
      "step": 1101
    },
    {
      "epoch": 0.6614645858343338,
      "grad_norm": 0.0004775487759616226,
      "learning_rate": 0.00019795642744465153,
      "loss": 23.0,
      "step": 1102
    },
    {
      "epoch": 0.6620648259303722,
      "grad_norm": 0.0012434301897883415,
      "learning_rate": 0.00019795262610045118,
      "loss": 23.0,
      "step": 1103
    },
    {
      "epoch": 0.6626650660264105,
      "grad_norm": 0.0016068033874034882,
      "learning_rate": 0.00019794882126057353,
      "loss": 23.0,
      "step": 1104
    },
    {
      "epoch": 0.6632653061224489,
      "grad_norm": 0.0007158942753449082,
      "learning_rate": 0.00019794501292515443,
      "loss": 23.0,
      "step": 1105
    },
    {
      "epoch": 0.6638655462184874,
      "grad_norm": 0.0007577801588922739,
      "learning_rate": 0.00019794120109432978,
      "loss": 23.0,
      "step": 1106
    },
    {
      "epoch": 0.6644657863145258,
      "grad_norm": 0.0005971409264020622,
      "learning_rate": 0.00019793738576823558,
      "loss": 23.0,
      "step": 1107
    },
    {
      "epoch": 0.6650660264105642,
      "grad_norm": 0.0010119474027305841,
      "learning_rate": 0.00019793356694700802,
      "loss": 23.0,
      "step": 1108
    },
    {
      "epoch": 0.6656662665066027,
      "grad_norm": 0.0004232526698615402,
      "learning_rate": 0.00019792974463078337,
      "loss": 23.0,
      "step": 1109
    },
    {
      "epoch": 0.6662665066026411,
      "grad_norm": 0.0006399360718205571,
      "learning_rate": 0.00019792591881969803,
      "loss": 23.0,
      "step": 1110
    },
    {
      "epoch": 0.6668667466986795,
      "grad_norm": 0.0005233879201114178,
      "learning_rate": 0.0001979220895138885,
      "loss": 23.0,
      "step": 1111
    },
    {
      "epoch": 0.6674669867947179,
      "grad_norm": 0.0009274845360778272,
      "learning_rate": 0.00019791825671349153,
      "loss": 23.0,
      "step": 1112
    },
    {
      "epoch": 0.6680672268907563,
      "grad_norm": 0.00030104804318398237,
      "learning_rate": 0.00019791442041864387,
      "loss": 23.0,
      "step": 1113
    },
    {
      "epoch": 0.6686674669867947,
      "grad_norm": 0.0008154875249601901,
      "learning_rate": 0.0001979105806294824,
      "loss": 23.0,
      "step": 1114
    },
    {
      "epoch": 0.6692677070828331,
      "grad_norm": 0.0012951076496392488,
      "learning_rate": 0.00019790673734614413,
      "loss": 23.0,
      "step": 1115
    },
    {
      "epoch": 0.6698679471788715,
      "grad_norm": 0.0006664313841611147,
      "learning_rate": 0.00019790289056876626,
      "loss": 23.0,
      "step": 1116
    },
    {
      "epoch": 0.67046818727491,
      "grad_norm": 0.0006468769279308617,
      "learning_rate": 0.0001978990402974861,
      "loss": 23.0,
      "step": 1117
    },
    {
      "epoch": 0.6710684273709484,
      "grad_norm": 0.0006649359129369259,
      "learning_rate": 0.00019789518653244098,
      "loss": 23.0,
      "step": 1118
    },
    {
      "epoch": 0.6716686674669868,
      "grad_norm": 0.0008164861355908215,
      "learning_rate": 0.00019789132927376847,
      "loss": 23.0,
      "step": 1119
    },
    {
      "epoch": 0.6722689075630253,
      "grad_norm": 0.0004112444876227528,
      "learning_rate": 0.00019788746852160625,
      "loss": 23.0,
      "step": 1120
    },
    {
      "epoch": 0.6728691476590636,
      "grad_norm": 0.0005869813612662256,
      "learning_rate": 0.00019788360427609204,
      "loss": 23.0,
      "step": 1121
    },
    {
      "epoch": 0.673469387755102,
      "grad_norm": 0.0013158872025087476,
      "learning_rate": 0.00019787973653736382,
      "loss": 23.0,
      "step": 1122
    },
    {
      "epoch": 0.6740696278511404,
      "grad_norm": 0.0008600007859058678,
      "learning_rate": 0.00019787586530555957,
      "loss": 23.0,
      "step": 1123
    },
    {
      "epoch": 0.6746698679471789,
      "grad_norm": 0.0008138124831020832,
      "learning_rate": 0.00019787199058081745,
      "loss": 23.0,
      "step": 1124
    },
    {
      "epoch": 0.6752701080432173,
      "grad_norm": 0.0006357624079100788,
      "learning_rate": 0.00019786811236327574,
      "loss": 23.0,
      "step": 1125
    },
    {
      "epoch": 0.6758703481392557,
      "grad_norm": 0.0011706652585417032,
      "learning_rate": 0.00019786423065307283,
      "loss": 23.0,
      "step": 1126
    },
    {
      "epoch": 0.6764705882352942,
      "grad_norm": 0.0005258683231659234,
      "learning_rate": 0.0001978603454503473,
      "loss": 23.0,
      "step": 1127
    },
    {
      "epoch": 0.6770708283313326,
      "grad_norm": 0.0005510378978215158,
      "learning_rate": 0.00019785645675523774,
      "loss": 23.0,
      "step": 1128
    },
    {
      "epoch": 0.6776710684273709,
      "grad_norm": 0.000846723560243845,
      "learning_rate": 0.00019785256456788294,
      "loss": 23.0,
      "step": 1129
    },
    {
      "epoch": 0.6782713085234093,
      "grad_norm": 0.001075017498806119,
      "learning_rate": 0.00019784866888842187,
      "loss": 23.0,
      "step": 1130
    },
    {
      "epoch": 0.6788715486194478,
      "grad_norm": 0.0012707427376881242,
      "learning_rate": 0.00019784476971699346,
      "loss": 23.0,
      "step": 1131
    },
    {
      "epoch": 0.6794717887154862,
      "grad_norm": 0.0005327064427547157,
      "learning_rate": 0.00019784086705373693,
      "loss": 23.0,
      "step": 1132
    },
    {
      "epoch": 0.6800720288115246,
      "grad_norm": 0.0002214109554188326,
      "learning_rate": 0.0001978369608987915,
      "loss": 23.0,
      "step": 1133
    },
    {
      "epoch": 0.680672268907563,
      "grad_norm": 0.0012937859864905477,
      "learning_rate": 0.00019783305125229667,
      "loss": 23.0,
      "step": 1134
    },
    {
      "epoch": 0.6812725090036015,
      "grad_norm": 0.0005032792105339468,
      "learning_rate": 0.00019782913811439184,
      "loss": 23.0,
      "step": 1135
    },
    {
      "epoch": 0.6818727490996399,
      "grad_norm": 0.0009746074792928994,
      "learning_rate": 0.00019782522148521672,
      "loss": 23.0,
      "step": 1136
    },
    {
      "epoch": 0.6824729891956782,
      "grad_norm": 0.0006890950025990605,
      "learning_rate": 0.00019782130136491107,
      "loss": 23.0,
      "step": 1137
    },
    {
      "epoch": 0.6830732292917167,
      "grad_norm": 0.0012513356050476432,
      "learning_rate": 0.00019781737775361485,
      "loss": 23.0,
      "step": 1138
    },
    {
      "epoch": 0.6836734693877551,
      "grad_norm": 0.000591045303735882,
      "learning_rate": 0.000197813450651468,
      "loss": 23.0,
      "step": 1139
    },
    {
      "epoch": 0.6842737094837935,
      "grad_norm": 0.0003968993842136115,
      "learning_rate": 0.00019780952005861072,
      "loss": 23.0,
      "step": 1140
    },
    {
      "epoch": 0.6848739495798319,
      "grad_norm": 0.0014705831417813897,
      "learning_rate": 0.00019780558597518323,
      "loss": 23.0,
      "step": 1141
    },
    {
      "epoch": 0.6854741896758704,
      "grad_norm": 0.0006096931174397469,
      "learning_rate": 0.00019780164840132598,
      "loss": 23.0,
      "step": 1142
    },
    {
      "epoch": 0.6860744297719088,
      "grad_norm": 0.0011487213196232915,
      "learning_rate": 0.0001977977073371795,
      "loss": 23.0,
      "step": 1143
    },
    {
      "epoch": 0.6866746698679472,
      "grad_norm": 0.0006493202527053654,
      "learning_rate": 0.00019779376278288436,
      "loss": 23.0,
      "step": 1144
    },
    {
      "epoch": 0.6872749099639855,
      "grad_norm": 0.00033152487594634295,
      "learning_rate": 0.00019778981473858142,
      "loss": 23.0,
      "step": 1145
    },
    {
      "epoch": 0.687875150060024,
      "grad_norm": 0.0007980447262525558,
      "learning_rate": 0.00019778586320441153,
      "loss": 23.0,
      "step": 1146
    },
    {
      "epoch": 0.6884753901560624,
      "grad_norm": 0.000773175444919616,
      "learning_rate": 0.00019778190818051572,
      "loss": 23.0,
      "step": 1147
    },
    {
      "epoch": 0.6890756302521008,
      "grad_norm": 0.000656657968647778,
      "learning_rate": 0.00019777794966703514,
      "loss": 23.0,
      "step": 1148
    },
    {
      "epoch": 0.6896758703481393,
      "grad_norm": 0.0007515581091865897,
      "learning_rate": 0.00019777398766411105,
      "loss": 23.0,
      "step": 1149
    },
    {
      "epoch": 0.6902761104441777,
      "grad_norm": 0.000665062281768769,
      "learning_rate": 0.00019777002217188482,
      "loss": 23.0,
      "step": 1150
    },
    {
      "epoch": 0.6908763505402161,
      "grad_norm": 0.0005964938900433481,
      "learning_rate": 0.000197766053190498,
      "loss": 23.0,
      "step": 1151
    },
    {
      "epoch": 0.6914765906362546,
      "grad_norm": 0.0011970044579356909,
      "learning_rate": 0.00019776208072009227,
      "loss": 23.0,
      "step": 1152
    },
    {
      "epoch": 0.6920768307322929,
      "grad_norm": 0.0008634612313471735,
      "learning_rate": 0.0001977581047608093,
      "loss": 23.0,
      "step": 1153
    },
    {
      "epoch": 0.6926770708283313,
      "grad_norm": 0.0006760912365280092,
      "learning_rate": 0.00019775412531279106,
      "loss": 23.0,
      "step": 1154
    },
    {
      "epoch": 0.6932773109243697,
      "grad_norm": 0.000713443267159164,
      "learning_rate": 0.0001977501423761795,
      "loss": 23.0,
      "step": 1155
    },
    {
      "epoch": 0.6938775510204082,
      "grad_norm": 0.0007609796593897045,
      "learning_rate": 0.00019774615595111685,
      "loss": 23.0,
      "step": 1156
    },
    {
      "epoch": 0.6944777911164466,
      "grad_norm": 0.0005407094722613692,
      "learning_rate": 0.00019774216603774534,
      "loss": 23.0,
      "step": 1157
    },
    {
      "epoch": 0.695078031212485,
      "grad_norm": 0.0019383853068575263,
      "learning_rate": 0.00019773817263620731,
      "loss": 23.0,
      "step": 1158
    },
    {
      "epoch": 0.6956782713085234,
      "grad_norm": 0.0008223625482060015,
      "learning_rate": 0.00019773417574664532,
      "loss": 23.0,
      "step": 1159
    },
    {
      "epoch": 0.6962785114045619,
      "grad_norm": 0.00045506638707593083,
      "learning_rate": 0.000197730175369202,
      "loss": 23.0,
      "step": 1160
    },
    {
      "epoch": 0.6968787515006002,
      "grad_norm": 0.001282287179492414,
      "learning_rate": 0.00019772617150402012,
      "loss": 23.0,
      "step": 1161
    },
    {
      "epoch": 0.6974789915966386,
      "grad_norm": 0.0006550766411237419,
      "learning_rate": 0.00019772216415124257,
      "loss": 23.0,
      "step": 1162
    },
    {
      "epoch": 0.698079231692677,
      "grad_norm": 0.0007768716313876212,
      "learning_rate": 0.00019771815331101232,
      "loss": 23.0,
      "step": 1163
    },
    {
      "epoch": 0.6986794717887155,
      "grad_norm": 0.0004817126027774066,
      "learning_rate": 0.00019771413898347256,
      "loss": 23.0,
      "step": 1164
    },
    {
      "epoch": 0.6992797118847539,
      "grad_norm": 0.000473877094918862,
      "learning_rate": 0.0001977101211687665,
      "loss": 23.0,
      "step": 1165
    },
    {
      "epoch": 0.6998799519807923,
      "grad_norm": 0.0005641740863211453,
      "learning_rate": 0.00019770609986703762,
      "loss": 23.0,
      "step": 1166
    },
    {
      "epoch": 0.7004801920768308,
      "grad_norm": 0.0006742041441611946,
      "learning_rate": 0.0001977020750784293,
      "loss": 23.0,
      "step": 1167
    },
    {
      "epoch": 0.7010804321728692,
      "grad_norm": 0.0004474896704778075,
      "learning_rate": 0.0001976980468030853,
      "loss": 23.0,
      "step": 1168
    },
    {
      "epoch": 0.7016806722689075,
      "grad_norm": 0.0007633958011865616,
      "learning_rate": 0.00019769401504114926,
      "loss": 23.0,
      "step": 1169
    },
    {
      "epoch": 0.7022809123649459,
      "grad_norm": 0.0009618657059036195,
      "learning_rate": 0.00019768997979276515,
      "loss": 23.0,
      "step": 1170
    },
    {
      "epoch": 0.7028811524609844,
      "grad_norm": 0.0003534193674568087,
      "learning_rate": 0.00019768594105807697,
      "loss": 23.0,
      "step": 1171
    },
    {
      "epoch": 0.7034813925570228,
      "grad_norm": 0.0012746489373967052,
      "learning_rate": 0.0001976818988372288,
      "loss": 23.0,
      "step": 1172
    },
    {
      "epoch": 0.7040816326530612,
      "grad_norm": 0.0004915861063636839,
      "learning_rate": 0.00019767785313036496,
      "loss": 23.0,
      "step": 1173
    },
    {
      "epoch": 0.7046818727490997,
      "grad_norm": 0.001498072873800993,
      "learning_rate": 0.00019767380393762978,
      "loss": 23.0,
      "step": 1174
    },
    {
      "epoch": 0.7052821128451381,
      "grad_norm": 0.000703103025443852,
      "learning_rate": 0.00019766975125916778,
      "loss": 23.0,
      "step": 1175
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.0019263819558545947,
      "learning_rate": 0.00019766569509512358,
      "loss": 23.0,
      "step": 1176
    },
    {
      "epoch": 0.7064825930372148,
      "grad_norm": 0.0006543250638060272,
      "learning_rate": 0.00019766163544564198,
      "loss": 23.0,
      "step": 1177
    },
    {
      "epoch": 0.7070828331332533,
      "grad_norm": 0.0014025021810084581,
      "learning_rate": 0.00019765757231086784,
      "loss": 23.0,
      "step": 1178
    },
    {
      "epoch": 0.7076830732292917,
      "grad_norm": 0.0005559175042435527,
      "learning_rate": 0.0001976535056909461,
      "loss": 23.0,
      "step": 1179
    },
    {
      "epoch": 0.7082833133253301,
      "grad_norm": 0.0010201229015365243,
      "learning_rate": 0.00019764943558602197,
      "loss": 23.0,
      "step": 1180
    },
    {
      "epoch": 0.7088835534213686,
      "grad_norm": 0.0008555446984246373,
      "learning_rate": 0.00019764536199624066,
      "loss": 23.0,
      "step": 1181
    },
    {
      "epoch": 0.709483793517407,
      "grad_norm": 0.0009428702760487795,
      "learning_rate": 0.00019764128492174755,
      "loss": 23.0,
      "step": 1182
    },
    {
      "epoch": 0.7100840336134454,
      "grad_norm": 0.0007122241659089923,
      "learning_rate": 0.00019763720436268814,
      "loss": 23.0,
      "step": 1183
    },
    {
      "epoch": 0.7106842737094838,
      "grad_norm": 0.0006242154049687088,
      "learning_rate": 0.00019763312031920806,
      "loss": 23.0,
      "step": 1184
    },
    {
      "epoch": 0.7112845138055222,
      "grad_norm": 0.0005786659312434494,
      "learning_rate": 0.0001976290327914531,
      "loss": 23.0,
      "step": 1185
    },
    {
      "epoch": 0.7118847539015606,
      "grad_norm": 0.0006534986314363778,
      "learning_rate": 0.00019762494177956902,
      "loss": 23.0,
      "step": 1186
    },
    {
      "epoch": 0.712484993997599,
      "grad_norm": 0.0011145776370540261,
      "learning_rate": 0.00019762084728370193,
      "loss": 23.0,
      "step": 1187
    },
    {
      "epoch": 0.7130852340936374,
      "grad_norm": 0.0005813301540911198,
      "learning_rate": 0.0001976167493039979,
      "loss": 23.0,
      "step": 1188
    },
    {
      "epoch": 0.7136854741896759,
      "grad_norm": 0.0007640339899808168,
      "learning_rate": 0.00019761264784060322,
      "loss": 23.0,
      "step": 1189
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.0006791938212700188,
      "learning_rate": 0.00019760854289366418,
      "loss": 23.0,
      "step": 1190
    },
    {
      "epoch": 0.7148859543817527,
      "grad_norm": 0.0006645294488407671,
      "learning_rate": 0.00019760443446332734,
      "loss": 23.0,
      "step": 1191
    },
    {
      "epoch": 0.7154861944777912,
      "grad_norm": 0.000585117086302489,
      "learning_rate": 0.00019760032254973934,
      "loss": 23.0,
      "step": 1192
    },
    {
      "epoch": 0.7160864345738295,
      "grad_norm": 0.0005612806417047977,
      "learning_rate": 0.00019759620715304683,
      "loss": 23.0,
      "step": 1193
    },
    {
      "epoch": 0.7166866746698679,
      "grad_norm": 0.0006190498825162649,
      "learning_rate": 0.00019759208827339677,
      "loss": 23.0,
      "step": 1194
    },
    {
      "epoch": 0.7172869147659063,
      "grad_norm": 0.0007879780023358762,
      "learning_rate": 0.00019758796591093608,
      "loss": 23.0,
      "step": 1195
    },
    {
      "epoch": 0.7178871548619448,
      "grad_norm": 0.0006741132820025086,
      "learning_rate": 0.00019758384006581195,
      "loss": 23.0,
      "step": 1196
    },
    {
      "epoch": 0.7184873949579832,
      "grad_norm": 0.0004388502275105566,
      "learning_rate": 0.00019757971073817158,
      "loss": 23.0,
      "step": 1197
    },
    {
      "epoch": 0.7190876350540216,
      "grad_norm": 0.0004376680008135736,
      "learning_rate": 0.00019757557792816232,
      "loss": 23.0,
      "step": 1198
    },
    {
      "epoch": 0.71968787515006,
      "grad_norm": 0.0008469756576232612,
      "learning_rate": 0.0001975714416359317,
      "loss": 23.0,
      "step": 1199
    },
    {
      "epoch": 0.7202881152460985,
      "grad_norm": 0.0006075134151615202,
      "learning_rate": 0.0001975673018616273,
      "loss": 23.0,
      "step": 1200
    },
    {
      "epoch": 0.7208883553421368,
      "grad_norm": 0.0010701941791921854,
      "learning_rate": 0.00019756315860539687,
      "loss": 23.0,
      "step": 1201
    },
    {
      "epoch": 0.7214885954381752,
      "grad_norm": 0.0008353521698154509,
      "learning_rate": 0.00019755901186738827,
      "loss": 23.0,
      "step": 1202
    },
    {
      "epoch": 0.7220888355342137,
      "grad_norm": 0.00043628833373077214,
      "learning_rate": 0.00019755486164774947,
      "loss": 23.0,
      "step": 1203
    },
    {
      "epoch": 0.7226890756302521,
      "grad_norm": 0.0015184413641691208,
      "learning_rate": 0.00019755070794662865,
      "loss": 23.0,
      "step": 1204
    },
    {
      "epoch": 0.7232893157262905,
      "grad_norm": 0.0011512304190546274,
      "learning_rate": 0.00019754655076417396,
      "loss": 23.0,
      "step": 1205
    },
    {
      "epoch": 0.723889555822329,
      "grad_norm": 0.0009428293560631573,
      "learning_rate": 0.00019754239010053376,
      "loss": 23.0,
      "step": 1206
    },
    {
      "epoch": 0.7244897959183674,
      "grad_norm": 0.0006190779386088252,
      "learning_rate": 0.00019753822595585662,
      "loss": 23.0,
      "step": 1207
    },
    {
      "epoch": 0.7250900360144058,
      "grad_norm": 0.0004119576478842646,
      "learning_rate": 0.0001975340583302911,
      "loss": 23.0,
      "step": 1208
    },
    {
      "epoch": 0.7256902761104442,
      "grad_norm": 0.0005930104525759816,
      "learning_rate": 0.00019752988722398587,
      "loss": 23.0,
      "step": 1209
    },
    {
      "epoch": 0.7262905162064826,
      "grad_norm": 0.0004418524040374905,
      "learning_rate": 0.00019752571263708987,
      "loss": 23.0,
      "step": 1210
    },
    {
      "epoch": 0.726890756302521,
      "grad_norm": 0.0010499389609321952,
      "learning_rate": 0.00019752153456975208,
      "loss": 23.0,
      "step": 1211
    },
    {
      "epoch": 0.7274909963985594,
      "grad_norm": 0.0011717370944097638,
      "learning_rate": 0.00019751735302212154,
      "loss": 23.0,
      "step": 1212
    },
    {
      "epoch": 0.7280912364945978,
      "grad_norm": 0.00043891393579542637,
      "learning_rate": 0.0001975131679943475,
      "loss": 23.0,
      "step": 1213
    },
    {
      "epoch": 0.7286914765906363,
      "grad_norm": 0.0005633426480926573,
      "learning_rate": 0.00019750897948657938,
      "loss": 23.0,
      "step": 1214
    },
    {
      "epoch": 0.7292917166866747,
      "grad_norm": 0.0004441675846464932,
      "learning_rate": 0.00019750478749896657,
      "loss": 23.0,
      "step": 1215
    },
    {
      "epoch": 0.7298919567827131,
      "grad_norm": 0.0012788623571395874,
      "learning_rate": 0.00019750059203165873,
      "loss": 23.0,
      "step": 1216
    },
    {
      "epoch": 0.7304921968787516,
      "grad_norm": 0.0004025893867947161,
      "learning_rate": 0.00019749639308480553,
      "loss": 23.0,
      "step": 1217
    },
    {
      "epoch": 0.7310924369747899,
      "grad_norm": 0.000841877656057477,
      "learning_rate": 0.00019749219065855688,
      "loss": 23.0,
      "step": 1218
    },
    {
      "epoch": 0.7316926770708283,
      "grad_norm": 0.0007755985716357827,
      "learning_rate": 0.00019748798475306272,
      "loss": 23.0,
      "step": 1219
    },
    {
      "epoch": 0.7322929171668667,
      "grad_norm": 0.0006181802018545568,
      "learning_rate": 0.00019748377536847316,
      "loss": 23.0,
      "step": 1220
    },
    {
      "epoch": 0.7328931572629052,
      "grad_norm": 0.0007388837984763086,
      "learning_rate": 0.0001974795625049384,
      "loss": 23.0,
      "step": 1221
    },
    {
      "epoch": 0.7334933973589436,
      "grad_norm": 0.0009291544556617737,
      "learning_rate": 0.0001974753461626088,
      "loss": 23.0,
      "step": 1222
    },
    {
      "epoch": 0.734093637454982,
      "grad_norm": 0.00043915770947933197,
      "learning_rate": 0.00019747112634163486,
      "loss": 23.0,
      "step": 1223
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 0.002203888026997447,
      "learning_rate": 0.0001974669030421671,
      "loss": 23.0,
      "step": 1224
    },
    {
      "epoch": 0.7352941176470589,
      "grad_norm": 0.0006513252155855298,
      "learning_rate": 0.0001974626762643563,
      "loss": 23.0,
      "step": 1225
    },
    {
      "epoch": 0.7358943577430972,
      "grad_norm": 0.0007130668964236975,
      "learning_rate": 0.00019745844600835332,
      "loss": 23.0,
      "step": 1226
    },
    {
      "epoch": 0.7364945978391356,
      "grad_norm": 0.0014644530601799488,
      "learning_rate": 0.00019745421227430904,
      "loss": 23.0,
      "step": 1227
    },
    {
      "epoch": 0.737094837935174,
      "grad_norm": 0.0018409211188554764,
      "learning_rate": 0.00019744997506237466,
      "loss": 23.0,
      "step": 1228
    },
    {
      "epoch": 0.7376950780312125,
      "grad_norm": 0.0011605211766436696,
      "learning_rate": 0.00019744573437270133,
      "loss": 23.0,
      "step": 1229
    },
    {
      "epoch": 0.7382953181272509,
      "grad_norm": 0.0006024561589583755,
      "learning_rate": 0.00019744149020544036,
      "loss": 23.0,
      "step": 1230
    },
    {
      "epoch": 0.7388955582232893,
      "grad_norm": 0.00031817928538657725,
      "learning_rate": 0.0001974372425607433,
      "loss": 23.0,
      "step": 1231
    },
    {
      "epoch": 0.7394957983193278,
      "grad_norm": 0.0020129424519836903,
      "learning_rate": 0.00019743299143876167,
      "loss": 23.0,
      "step": 1232
    },
    {
      "epoch": 0.7400960384153662,
      "grad_norm": 0.0006386892055161297,
      "learning_rate": 0.0001974287368396472,
      "loss": 23.0,
      "step": 1233
    },
    {
      "epoch": 0.7406962785114045,
      "grad_norm": 0.000763615476898849,
      "learning_rate": 0.00019742447876355175,
      "loss": 23.0,
      "step": 1234
    },
    {
      "epoch": 0.741296518607443,
      "grad_norm": 0.0007875692099332809,
      "learning_rate": 0.00019742021721062724,
      "loss": 23.0,
      "step": 1235
    },
    {
      "epoch": 0.7418967587034814,
      "grad_norm": 0.0014706990914419293,
      "learning_rate": 0.0001974159521810258,
      "loss": 23.0,
      "step": 1236
    },
    {
      "epoch": 0.7424969987995198,
      "grad_norm": 0.0007930982974357903,
      "learning_rate": 0.00019741168367489958,
      "loss": 23.0,
      "step": 1237
    },
    {
      "epoch": 0.7430972388955582,
      "grad_norm": 0.0004145725688431412,
      "learning_rate": 0.00019740741169240096,
      "loss": 23.0,
      "step": 1238
    },
    {
      "epoch": 0.7436974789915967,
      "grad_norm": 0.0007471741992048919,
      "learning_rate": 0.00019740313623368234,
      "loss": 23.0,
      "step": 1239
    },
    {
      "epoch": 0.7442977190876351,
      "grad_norm": 0.001122286426834762,
      "learning_rate": 0.00019739885729889637,
      "loss": 23.0,
      "step": 1240
    },
    {
      "epoch": 0.7448979591836735,
      "grad_norm": 0.0006311690085567534,
      "learning_rate": 0.00019739457488819574,
      "loss": 23.0,
      "step": 1241
    },
    {
      "epoch": 0.7454981992797118,
      "grad_norm": 0.0005989365745335817,
      "learning_rate": 0.00019739028900173322,
      "loss": 23.0,
      "step": 1242
    },
    {
      "epoch": 0.7460984393757503,
      "grad_norm": 0.0006342297419905663,
      "learning_rate": 0.00019738599963966183,
      "loss": 23.0,
      "step": 1243
    },
    {
      "epoch": 0.7466986794717887,
      "grad_norm": 0.0006295668426901102,
      "learning_rate": 0.00019738170680213464,
      "loss": 23.0,
      "step": 1244
    },
    {
      "epoch": 0.7472989195678271,
      "grad_norm": 0.0005823764950037003,
      "learning_rate": 0.00019737741048930478,
      "loss": 23.0,
      "step": 1245
    },
    {
      "epoch": 0.7478991596638656,
      "grad_norm": 0.001335416454821825,
      "learning_rate": 0.00019737311070132564,
      "loss": 23.0,
      "step": 1246
    },
    {
      "epoch": 0.748499399759904,
      "grad_norm": 0.0008027239236980677,
      "learning_rate": 0.00019736880743835068,
      "loss": 23.0,
      "step": 1247
    },
    {
      "epoch": 0.7490996398559424,
      "grad_norm": 0.0004205139121040702,
      "learning_rate": 0.00019736450070053342,
      "loss": 23.0,
      "step": 1248
    },
    {
      "epoch": 0.7496998799519808,
      "grad_norm": 0.0005247509106993675,
      "learning_rate": 0.0001973601904880276,
      "loss": 23.0,
      "step": 1249
    },
    {
      "epoch": 0.7503001200480192,
      "grad_norm": 0.00047644192818552256,
      "learning_rate": 0.000197355876800987,
      "loss": 23.0,
      "step": 1250
    },
    {
      "epoch": 0.7509003601440576,
      "grad_norm": 0.0006163313519209623,
      "learning_rate": 0.0001973515596395656,
      "loss": 23.0,
      "step": 1251
    },
    {
      "epoch": 0.751500600240096,
      "grad_norm": 0.0009510073577985168,
      "learning_rate": 0.00019734723900391744,
      "loss": 23.0,
      "step": 1252
    },
    {
      "epoch": 0.7521008403361344,
      "grad_norm": 0.00038402751670219004,
      "learning_rate": 0.00019734291489419673,
      "loss": 23.0,
      "step": 1253
    },
    {
      "epoch": 0.7527010804321729,
      "grad_norm": 0.00068047852255404,
      "learning_rate": 0.0001973385873105578,
      "loss": 23.0,
      "step": 1254
    },
    {
      "epoch": 0.7533013205282113,
      "grad_norm": 0.00037462665932253003,
      "learning_rate": 0.00019733425625315508,
      "loss": 23.0,
      "step": 1255
    },
    {
      "epoch": 0.7539015606242497,
      "grad_norm": 0.0004069966671522707,
      "learning_rate": 0.0001973299217221431,
      "loss": 23.0,
      "step": 1256
    },
    {
      "epoch": 0.7545018007202882,
      "grad_norm": 0.001153744524344802,
      "learning_rate": 0.00019732558371767658,
      "loss": 23.0,
      "step": 1257
    },
    {
      "epoch": 0.7551020408163265,
      "grad_norm": 0.0005174583056941628,
      "learning_rate": 0.00019732124223991034,
      "loss": 23.0,
      "step": 1258
    },
    {
      "epoch": 0.7557022809123649,
      "grad_norm": 0.0006991420523263514,
      "learning_rate": 0.0001973168972889993,
      "loss": 23.0,
      "step": 1259
    },
    {
      "epoch": 0.7563025210084033,
      "grad_norm": 0.00046238224604167044,
      "learning_rate": 0.00019731254886509852,
      "loss": 23.0,
      "step": 1260
    },
    {
      "epoch": 0.7569027611044418,
      "grad_norm": 0.0011713410494849086,
      "learning_rate": 0.0001973081969683632,
      "loss": 23.0,
      "step": 1261
    },
    {
      "epoch": 0.7575030012004802,
      "grad_norm": 0.000421454751631245,
      "learning_rate": 0.00019730384159894864,
      "loss": 23.0,
      "step": 1262
    },
    {
      "epoch": 0.7581032412965186,
      "grad_norm": 0.0012959492160007358,
      "learning_rate": 0.00019729948275701027,
      "loss": 23.0,
      "step": 1263
    },
    {
      "epoch": 0.758703481392557,
      "grad_norm": 0.0007972400635480881,
      "learning_rate": 0.00019729512044270364,
      "loss": 23.0,
      "step": 1264
    },
    {
      "epoch": 0.7593037214885955,
      "grad_norm": 0.0007275702082552016,
      "learning_rate": 0.00019729075465618442,
      "loss": 23.0,
      "step": 1265
    },
    {
      "epoch": 0.7599039615846338,
      "grad_norm": 0.000372605980373919,
      "learning_rate": 0.00019728638539760843,
      "loss": 23.0,
      "step": 1266
    },
    {
      "epoch": 0.7605042016806722,
      "grad_norm": 0.0007662245770916343,
      "learning_rate": 0.0001972820126671316,
      "loss": 23.0,
      "step": 1267
    },
    {
      "epoch": 0.7611044417767107,
      "grad_norm": 0.0008609720971435308,
      "learning_rate": 0.00019727763646491,
      "loss": 23.0,
      "step": 1268
    },
    {
      "epoch": 0.7617046818727491,
      "grad_norm": 0.0013051877031102777,
      "learning_rate": 0.00019727325679109976,
      "loss": 23.0,
      "step": 1269
    },
    {
      "epoch": 0.7623049219687875,
      "grad_norm": 0.0011220271699130535,
      "learning_rate": 0.00019726887364585716,
      "loss": 23.0,
      "step": 1270
    },
    {
      "epoch": 0.762905162064826,
      "grad_norm": 0.0006207393016666174,
      "learning_rate": 0.00019726448702933876,
      "loss": 23.0,
      "step": 1271
    },
    {
      "epoch": 0.7635054021608644,
      "grad_norm": 0.0008961490821093321,
      "learning_rate": 0.00019726009694170094,
      "loss": 23.0,
      "step": 1272
    },
    {
      "epoch": 0.7641056422569028,
      "grad_norm": 0.0011103064753115177,
      "learning_rate": 0.00019725570338310046,
      "loss": 23.0,
      "step": 1273
    },
    {
      "epoch": 0.7647058823529411,
      "grad_norm": 0.0011056560324504972,
      "learning_rate": 0.0001972513063536941,
      "loss": 23.0,
      "step": 1274
    },
    {
      "epoch": 0.7653061224489796,
      "grad_norm": 0.0009583919891156256,
      "learning_rate": 0.00019724690585363878,
      "loss": 23.0,
      "step": 1275
    },
    {
      "epoch": 0.765906362545018,
      "grad_norm": 0.0007008077809587121,
      "learning_rate": 0.00019724250188309154,
      "loss": 23.0,
      "step": 1276
    },
    {
      "epoch": 0.7665066026410564,
      "grad_norm": 0.0014242903562262654,
      "learning_rate": 0.00019723809444220956,
      "loss": 23.0,
      "step": 1277
    },
    {
      "epoch": 0.7671068427370948,
      "grad_norm": 0.0008935600053519011,
      "learning_rate": 0.00019723368353115008,
      "loss": 23.0,
      "step": 1278
    },
    {
      "epoch": 0.7677070828331333,
      "grad_norm": 0.0012262960663065314,
      "learning_rate": 0.00019722926915007055,
      "loss": 23.0,
      "step": 1279
    },
    {
      "epoch": 0.7683073229291717,
      "grad_norm": 0.0005361196235753596,
      "learning_rate": 0.00019722485129912854,
      "loss": 23.0,
      "step": 1280
    },
    {
      "epoch": 0.7689075630252101,
      "grad_norm": 0.0010704381857067347,
      "learning_rate": 0.00019722042997848168,
      "loss": 23.0,
      "step": 1281
    },
    {
      "epoch": 0.7695078031212484,
      "grad_norm": 0.0009096910944208503,
      "learning_rate": 0.00019721600518828774,
      "loss": 23.0,
      "step": 1282
    },
    {
      "epoch": 0.7701080432172869,
      "grad_norm": 0.0007512019947171211,
      "learning_rate": 0.00019721157692870463,
      "loss": 23.0,
      "step": 1283
    },
    {
      "epoch": 0.7707082833133253,
      "grad_norm": 0.0009879972785711288,
      "learning_rate": 0.00019720714519989044,
      "loss": 23.0,
      "step": 1284
    },
    {
      "epoch": 0.7713085234093637,
      "grad_norm": 0.0007033760193735361,
      "learning_rate": 0.00019720271000200324,
      "loss": 23.0,
      "step": 1285
    },
    {
      "epoch": 0.7719087635054022,
      "grad_norm": 0.0007616998627781868,
      "learning_rate": 0.00019719827133520137,
      "loss": 23.0,
      "step": 1286
    },
    {
      "epoch": 0.7725090036014406,
      "grad_norm": 0.0008339902269653976,
      "learning_rate": 0.00019719382919964321,
      "loss": 23.0,
      "step": 1287
    },
    {
      "epoch": 0.773109243697479,
      "grad_norm": 0.0006231502629816532,
      "learning_rate": 0.0001971893835954873,
      "loss": 23.0,
      "step": 1288
    },
    {
      "epoch": 0.7737094837935174,
      "grad_norm": 0.0006061487947590649,
      "learning_rate": 0.0001971849345228923,
      "loss": 23.0,
      "step": 1289
    },
    {
      "epoch": 0.7743097238895558,
      "grad_norm": 0.0002856554929167032,
      "learning_rate": 0.00019718048198201697,
      "loss": 23.0,
      "step": 1290
    },
    {
      "epoch": 0.7749099639855942,
      "grad_norm": 0.0006068464717827737,
      "learning_rate": 0.00019717602597302025,
      "loss": 23.0,
      "step": 1291
    },
    {
      "epoch": 0.7755102040816326,
      "grad_norm": 0.00040136679308488965,
      "learning_rate": 0.00019717156649606108,
      "loss": 23.0,
      "step": 1292
    },
    {
      "epoch": 0.776110444177671,
      "grad_norm": 0.0006840699934400618,
      "learning_rate": 0.0001971671035512987,
      "loss": 23.0,
      "step": 1293
    },
    {
      "epoch": 0.7767106842737095,
      "grad_norm": 0.000576976512093097,
      "learning_rate": 0.00019716263713889228,
      "loss": 23.0,
      "step": 1294
    },
    {
      "epoch": 0.7773109243697479,
      "grad_norm": 0.000296157319098711,
      "learning_rate": 0.00019715816725900134,
      "loss": 23.0,
      "step": 1295
    },
    {
      "epoch": 0.7779111644657863,
      "grad_norm": 0.001230229390785098,
      "learning_rate": 0.00019715369391178528,
      "loss": 23.0,
      "step": 1296
    },
    {
      "epoch": 0.7785114045618248,
      "grad_norm": 0.00044690098729915917,
      "learning_rate": 0.0001971492170974038,
      "loss": 23.0,
      "step": 1297
    },
    {
      "epoch": 0.7791116446578632,
      "grad_norm": 0.0006561097106896341,
      "learning_rate": 0.0001971447368160167,
      "loss": 23.0,
      "step": 1298
    },
    {
      "epoch": 0.7797118847539015,
      "grad_norm": 0.0005576546536758542,
      "learning_rate": 0.0001971402530677838,
      "loss": 23.0,
      "step": 1299
    },
    {
      "epoch": 0.78031212484994,
      "grad_norm": 0.0013528415001928806,
      "learning_rate": 0.00019713576585286515,
      "loss": 23.0,
      "step": 1300
    },
    {
      "epoch": 0.7809123649459784,
      "grad_norm": 0.000892074138391763,
      "learning_rate": 0.00019713127517142088,
      "loss": 23.0,
      "step": 1301
    },
    {
      "epoch": 0.7815126050420168,
      "grad_norm": 0.001970496727153659,
      "learning_rate": 0.00019712678102361123,
      "loss": 23.0,
      "step": 1302
    },
    {
      "epoch": 0.7821128451380552,
      "grad_norm": 0.0010920679196715355,
      "learning_rate": 0.00019712228340959661,
      "loss": 23.0,
      "step": 1303
    },
    {
      "epoch": 0.7827130852340937,
      "grad_norm": 0.0005002216785214841,
      "learning_rate": 0.00019711778232953754,
      "loss": 23.0,
      "step": 1304
    },
    {
      "epoch": 0.7833133253301321,
      "grad_norm": 0.0010139576625078917,
      "learning_rate": 0.00019711327778359464,
      "loss": 23.0,
      "step": 1305
    },
    {
      "epoch": 0.7839135654261705,
      "grad_norm": 0.0016053385334089398,
      "learning_rate": 0.00019710876977192864,
      "loss": 23.0,
      "step": 1306
    },
    {
      "epoch": 0.7845138055222088,
      "grad_norm": 0.0008072099881246686,
      "learning_rate": 0.00019710425829470046,
      "loss": 23.0,
      "step": 1307
    },
    {
      "epoch": 0.7851140456182473,
      "grad_norm": 0.0008704765350557864,
      "learning_rate": 0.00019709974335207106,
      "loss": 23.0,
      "step": 1308
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 0.0007995392079465091,
      "learning_rate": 0.0001970952249442016,
      "loss": 23.0,
      "step": 1309
    },
    {
      "epoch": 0.7863145258103241,
      "grad_norm": 0.0007567350985482335,
      "learning_rate": 0.00019709070307125334,
      "loss": 23.0,
      "step": 1310
    },
    {
      "epoch": 0.7869147659063626,
      "grad_norm": 0.001349606434814632,
      "learning_rate": 0.00019708617773338762,
      "loss": 23.0,
      "step": 1311
    },
    {
      "epoch": 0.787515006002401,
      "grad_norm": 0.00046586417010985315,
      "learning_rate": 0.00019708164893076594,
      "loss": 23.0,
      "step": 1312
    },
    {
      "epoch": 0.7881152460984394,
      "grad_norm": 0.0007738059503026307,
      "learning_rate": 0.00019707711666354998,
      "loss": 23.0,
      "step": 1313
    },
    {
      "epoch": 0.7887154861944778,
      "grad_norm": 0.001490103080868721,
      "learning_rate": 0.0001970725809319014,
      "loss": 23.0,
      "step": 1314
    },
    {
      "epoch": 0.7893157262905162,
      "grad_norm": 0.0004968742723576725,
      "learning_rate": 0.00019706804173598208,
      "loss": 23.0,
      "step": 1315
    },
    {
      "epoch": 0.7899159663865546,
      "grad_norm": 0.0006271405727602541,
      "learning_rate": 0.00019706349907595407,
      "loss": 23.0,
      "step": 1316
    },
    {
      "epoch": 0.790516206482593,
      "grad_norm": 0.000878729042597115,
      "learning_rate": 0.00019705895295197946,
      "loss": 23.0,
      "step": 1317
    },
    {
      "epoch": 0.7911164465786314,
      "grad_norm": 0.000744846067391336,
      "learning_rate": 0.0001970544033642205,
      "loss": 23.0,
      "step": 1318
    },
    {
      "epoch": 0.7917166866746699,
      "grad_norm": 0.0011991016799584031,
      "learning_rate": 0.0001970498503128395,
      "loss": 23.0,
      "step": 1319
    },
    {
      "epoch": 0.7923169267707083,
      "grad_norm": 0.0007287758053280413,
      "learning_rate": 0.000197045293797999,
      "loss": 23.0,
      "step": 1320
    },
    {
      "epoch": 0.7929171668667467,
      "grad_norm": 0.00045763529487885535,
      "learning_rate": 0.00019704073381986162,
      "loss": 23.0,
      "step": 1321
    },
    {
      "epoch": 0.7935174069627852,
      "grad_norm": 0.00036194606218487024,
      "learning_rate": 0.00019703617037859002,
      "loss": 23.0,
      "step": 1322
    },
    {
      "epoch": 0.7941176470588235,
      "grad_norm": 0.0006920311716385186,
      "learning_rate": 0.00019703160347434712,
      "loss": 23.0,
      "step": 1323
    },
    {
      "epoch": 0.7947178871548619,
      "grad_norm": 0.0006445212638936937,
      "learning_rate": 0.0001970270331072959,
      "loss": 23.0,
      "step": 1324
    },
    {
      "epoch": 0.7953181272509003,
      "grad_norm": 0.001651501515880227,
      "learning_rate": 0.00019702245927759947,
      "loss": 23.0,
      "step": 1325
    },
    {
      "epoch": 0.7959183673469388,
      "grad_norm": 0.0009059353033080697,
      "learning_rate": 0.00019701788198542102,
      "loss": 23.0,
      "step": 1326
    },
    {
      "epoch": 0.7965186074429772,
      "grad_norm": 0.0016654802020639181,
      "learning_rate": 0.00019701330123092392,
      "loss": 23.0,
      "step": 1327
    },
    {
      "epoch": 0.7971188475390156,
      "grad_norm": 0.0012016391847282648,
      "learning_rate": 0.00019700871701427164,
      "loss": 23.0,
      "step": 1328
    },
    {
      "epoch": 0.7977190876350541,
      "grad_norm": 0.0009673786698840559,
      "learning_rate": 0.0001970041293356278,
      "loss": 23.0,
      "step": 1329
    },
    {
      "epoch": 0.7983193277310925,
      "grad_norm": 0.0007137698121368885,
      "learning_rate": 0.00019699953819515611,
      "loss": 23.0,
      "step": 1330
    },
    {
      "epoch": 0.7989195678271308,
      "grad_norm": 0.000576613936573267,
      "learning_rate": 0.00019699494359302037,
      "loss": 23.0,
      "step": 1331
    },
    {
      "epoch": 0.7995198079231692,
      "grad_norm": 0.00029468824504874647,
      "learning_rate": 0.00019699034552938467,
      "loss": 23.0,
      "step": 1332
    },
    {
      "epoch": 0.8001200480192077,
      "grad_norm": 0.0005525742890313268,
      "learning_rate": 0.00019698574400441298,
      "loss": 23.0,
      "step": 1333
    },
    {
      "epoch": 0.8007202881152461,
      "grad_norm": 0.0015778294764459133,
      "learning_rate": 0.00019698113901826957,
      "loss": 23.0,
      "step": 1334
    },
    {
      "epoch": 0.8013205282112845,
      "grad_norm": 0.0004467495309654623,
      "learning_rate": 0.00019697653057111877,
      "loss": 23.0,
      "step": 1335
    },
    {
      "epoch": 0.801920768307323,
      "grad_norm": 0.0006680064834654331,
      "learning_rate": 0.00019697191866312506,
      "loss": 23.0,
      "step": 1336
    },
    {
      "epoch": 0.8025210084033614,
      "grad_norm": 0.0011591626098379493,
      "learning_rate": 0.00019696730329445302,
      "loss": 23.0,
      "step": 1337
    },
    {
      "epoch": 0.8031212484993998,
      "grad_norm": 0.0004280972934793681,
      "learning_rate": 0.0001969626844652673,
      "loss": 23.0,
      "step": 1338
    },
    {
      "epoch": 0.8037214885954381,
      "grad_norm": 0.0008343479712493718,
      "learning_rate": 0.00019695806217573286,
      "loss": 23.0,
      "step": 1339
    },
    {
      "epoch": 0.8043217286914766,
      "grad_norm": 0.0013662086566910148,
      "learning_rate": 0.00019695343642601456,
      "loss": 23.0,
      "step": 1340
    },
    {
      "epoch": 0.804921968787515,
      "grad_norm": 0.0006845472962595522,
      "learning_rate": 0.00019694880721627748,
      "loss": 23.0,
      "step": 1341
    },
    {
      "epoch": 0.8055222088835534,
      "grad_norm": 0.0020524887368083,
      "learning_rate": 0.0001969441745466869,
      "loss": 23.0,
      "step": 1342
    },
    {
      "epoch": 0.8061224489795918,
      "grad_norm": 0.0011398709611967206,
      "learning_rate": 0.00019693953841740808,
      "loss": 23.0,
      "step": 1343
    },
    {
      "epoch": 0.8067226890756303,
      "grad_norm": 0.0005793363670818508,
      "learning_rate": 0.00019693489882860647,
      "loss": 23.0,
      "step": 1344
    },
    {
      "epoch": 0.8073229291716687,
      "grad_norm": 0.0007696032989770174,
      "learning_rate": 0.0001969302557804477,
      "loss": 23.0,
      "step": 1345
    },
    {
      "epoch": 0.8079231692677071,
      "grad_norm": 0.000903173116967082,
      "learning_rate": 0.00019692560927309742,
      "loss": 23.0,
      "step": 1346
    },
    {
      "epoch": 0.8085234093637454,
      "grad_norm": 0.0005450271419249475,
      "learning_rate": 0.00019692095930672147,
      "loss": 23.0,
      "step": 1347
    },
    {
      "epoch": 0.8091236494597839,
      "grad_norm": 0.0007734773098491132,
      "learning_rate": 0.0001969163058814858,
      "loss": 23.0,
      "step": 1348
    },
    {
      "epoch": 0.8097238895558223,
      "grad_norm": 0.0007161099347285926,
      "learning_rate": 0.0001969116489975565,
      "loss": 23.0,
      "step": 1349
    },
    {
      "epoch": 0.8103241296518607,
      "grad_norm": 0.00046699721133336425,
      "learning_rate": 0.00019690698865509966,
      "loss": 23.0,
      "step": 1350
    },
    {
      "epoch": 0.8109243697478992,
      "grad_norm": 0.0009711477323435247,
      "learning_rate": 0.0001969023248542817,
      "loss": 23.0,
      "step": 1351
    },
    {
      "epoch": 0.8115246098439376,
      "grad_norm": 0.001304179197177291,
      "learning_rate": 0.00019689765759526906,
      "loss": 23.0,
      "step": 1352
    },
    {
      "epoch": 0.812124849939976,
      "grad_norm": 0.0004178002418484539,
      "learning_rate": 0.00019689298687822822,
      "loss": 23.0,
      "step": 1353
    },
    {
      "epoch": 0.8127250900360145,
      "grad_norm": 0.00036371249007061124,
      "learning_rate": 0.00019688831270332595,
      "loss": 23.0,
      "step": 1354
    },
    {
      "epoch": 0.8133253301320528,
      "grad_norm": 0.0006969739333726466,
      "learning_rate": 0.00019688363507072904,
      "loss": 23.0,
      "step": 1355
    },
    {
      "epoch": 0.8139255702280912,
      "grad_norm": 0.001011022599413991,
      "learning_rate": 0.00019687895398060439,
      "loss": 23.0,
      "step": 1356
    },
    {
      "epoch": 0.8145258103241296,
      "grad_norm": 0.0005929060280323029,
      "learning_rate": 0.00019687426943311906,
      "loss": 23.0,
      "step": 1357
    },
    {
      "epoch": 0.8151260504201681,
      "grad_norm": 0.00029369836556725204,
      "learning_rate": 0.00019686958142844024,
      "loss": 23.0,
      "step": 1358
    },
    {
      "epoch": 0.8157262905162065,
      "grad_norm": 0.0005359333590604365,
      "learning_rate": 0.00019686488996673528,
      "loss": 23.0,
      "step": 1359
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 0.0005220797029323876,
      "learning_rate": 0.00019686019504817155,
      "loss": 23.0,
      "step": 1360
    },
    {
      "epoch": 0.8169267707082833,
      "grad_norm": 0.000851228425744921,
      "learning_rate": 0.0001968554966729166,
      "loss": 23.0,
      "step": 1361
    },
    {
      "epoch": 0.8175270108043218,
      "grad_norm": 0.0005501755513250828,
      "learning_rate": 0.0001968507948411381,
      "loss": 23.0,
      "step": 1362
    },
    {
      "epoch": 0.8181272509003601,
      "grad_norm": 0.0010599548695608974,
      "learning_rate": 0.0001968460895530039,
      "loss": 23.0,
      "step": 1363
    },
    {
      "epoch": 0.8187274909963985,
      "grad_norm": 0.0009824999142438173,
      "learning_rate": 0.00019684138080868185,
      "loss": 23.0,
      "step": 1364
    },
    {
      "epoch": 0.819327731092437,
      "grad_norm": 0.00047340954188257456,
      "learning_rate": 0.00019683666860834004,
      "loss": 23.0,
      "step": 1365
    },
    {
      "epoch": 0.8199279711884754,
      "grad_norm": 0.0014246002538129687,
      "learning_rate": 0.00019683195295214662,
      "loss": 23.0,
      "step": 1366
    },
    {
      "epoch": 0.8205282112845138,
      "grad_norm": 0.00044270779471844435,
      "learning_rate": 0.00019682723384026985,
      "loss": 23.0,
      "step": 1367
    },
    {
      "epoch": 0.8211284513805522,
      "grad_norm": 0.0005253563867881894,
      "learning_rate": 0.00019682251127287824,
      "loss": 23.0,
      "step": 1368
    },
    {
      "epoch": 0.8217286914765907,
      "grad_norm": 0.0022502955980598927,
      "learning_rate": 0.00019681778525014023,
      "loss": 23.0,
      "step": 1369
    },
    {
      "epoch": 0.8223289315726291,
      "grad_norm": 0.0013780895387753844,
      "learning_rate": 0.00019681305577222449,
      "loss": 23.0,
      "step": 1370
    },
    {
      "epoch": 0.8229291716686674,
      "grad_norm": 0.001034077606163919,
      "learning_rate": 0.00019680832283929984,
      "loss": 23.0,
      "step": 1371
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.000522664631716907,
      "learning_rate": 0.00019680358645153513,
      "loss": 23.0,
      "step": 1372
    },
    {
      "epoch": 0.8241296518607443,
      "grad_norm": 0.00048665585927665234,
      "learning_rate": 0.00019679884660909948,
      "loss": 23.0,
      "step": 1373
    },
    {
      "epoch": 0.8247298919567827,
      "grad_norm": 0.0013081665383651853,
      "learning_rate": 0.00019679410331216197,
      "loss": 23.0,
      "step": 1374
    },
    {
      "epoch": 0.8253301320528211,
      "grad_norm": 0.000508311262819916,
      "learning_rate": 0.0001967893565608919,
      "loss": 23.0,
      "step": 1375
    },
    {
      "epoch": 0.8259303721488596,
      "grad_norm": 0.0004545389674603939,
      "learning_rate": 0.0001967846063554587,
      "loss": 23.0,
      "step": 1376
    },
    {
      "epoch": 0.826530612244898,
      "grad_norm": 0.0004129679291509092,
      "learning_rate": 0.0001967798526960318,
      "loss": 23.0,
      "step": 1377
    },
    {
      "epoch": 0.8271308523409364,
      "grad_norm": 0.000469258549856022,
      "learning_rate": 0.00019677509558278092,
      "loss": 23.0,
      "step": 1378
    },
    {
      "epoch": 0.8277310924369747,
      "grad_norm": 0.0006527505465783179,
      "learning_rate": 0.00019677033501587583,
      "loss": 23.0,
      "step": 1379
    },
    {
      "epoch": 0.8283313325330132,
      "grad_norm": 0.0009122650953941047,
      "learning_rate": 0.0001967655709954864,
      "loss": 23.0,
      "step": 1380
    },
    {
      "epoch": 0.8289315726290516,
      "grad_norm": 0.000640625599771738,
      "learning_rate": 0.00019676080352178265,
      "loss": 23.0,
      "step": 1381
    },
    {
      "epoch": 0.82953181272509,
      "grad_norm": 0.0005764299421571195,
      "learning_rate": 0.00019675603259493472,
      "loss": 23.0,
      "step": 1382
    },
    {
      "epoch": 0.8301320528211285,
      "grad_norm": 0.0013179031666368246,
      "learning_rate": 0.00019675125821511287,
      "loss": 23.0,
      "step": 1383
    },
    {
      "epoch": 0.8307322929171669,
      "grad_norm": 0.0005996134132146835,
      "learning_rate": 0.00019674648038248748,
      "loss": 23.0,
      "step": 1384
    },
    {
      "epoch": 0.8313325330132053,
      "grad_norm": 0.0007975632906891406,
      "learning_rate": 0.00019674169909722911,
      "loss": 23.0,
      "step": 1385
    },
    {
      "epoch": 0.8319327731092437,
      "grad_norm": 0.0007109794532880187,
      "learning_rate": 0.00019673691435950828,
      "loss": 23.0,
      "step": 1386
    },
    {
      "epoch": 0.8325330132052821,
      "grad_norm": 0.0006750710308551788,
      "learning_rate": 0.00019673212616949588,
      "loss": 23.0,
      "step": 1387
    },
    {
      "epoch": 0.8331332533013205,
      "grad_norm": 0.00053409393876791,
      "learning_rate": 0.00019672733452736268,
      "loss": 23.0,
      "step": 1388
    },
    {
      "epoch": 0.8337334933973589,
      "grad_norm": 0.0008340748026967049,
      "learning_rate": 0.00019672253943327973,
      "loss": 23.0,
      "step": 1389
    },
    {
      "epoch": 0.8343337334933973,
      "grad_norm": 0.0005991861689835787,
      "learning_rate": 0.00019671774088741815,
      "loss": 23.0,
      "step": 1390
    },
    {
      "epoch": 0.8349339735894358,
      "grad_norm": 0.0009061013697646558,
      "learning_rate": 0.00019671293888994922,
      "loss": 23.0,
      "step": 1391
    },
    {
      "epoch": 0.8355342136854742,
      "grad_norm": 0.0011456649517640471,
      "learning_rate": 0.00019670813344104424,
      "loss": 23.0,
      "step": 1392
    },
    {
      "epoch": 0.8361344537815126,
      "grad_norm": 0.0008859516819939017,
      "learning_rate": 0.00019670332454087476,
      "loss": 23.0,
      "step": 1393
    },
    {
      "epoch": 0.8367346938775511,
      "grad_norm": 0.0008393028401769698,
      "learning_rate": 0.00019669851218961232,
      "loss": 23.0,
      "step": 1394
    },
    {
      "epoch": 0.8373349339735895,
      "grad_norm": 0.0008376188343390822,
      "learning_rate": 0.00019669369638742877,
      "loss": 23.0,
      "step": 1395
    },
    {
      "epoch": 0.8379351740696278,
      "grad_norm": 0.0004782997420988977,
      "learning_rate": 0.0001966888771344959,
      "loss": 23.0,
      "step": 1396
    },
    {
      "epoch": 0.8385354141656662,
      "grad_norm": 0.0008413901668973267,
      "learning_rate": 0.00019668405443098576,
      "loss": 23.0,
      "step": 1397
    },
    {
      "epoch": 0.8391356542617047,
      "grad_norm": 0.00035914263571612537,
      "learning_rate": 0.00019667922827707037,
      "loss": 23.0,
      "step": 1398
    },
    {
      "epoch": 0.8397358943577431,
      "grad_norm": 0.0005109699559397995,
      "learning_rate": 0.00019667439867292202,
      "loss": 23.0,
      "step": 1399
    },
    {
      "epoch": 0.8403361344537815,
      "grad_norm": 0.0008713309071026742,
      "learning_rate": 0.0001966695656187131,
      "loss": 23.0,
      "step": 1400
    },
    {
      "epoch": 0.84093637454982,
      "grad_norm": 0.0014358647167682648,
      "learning_rate": 0.00019666472911461603,
      "loss": 23.0,
      "step": 1401
    },
    {
      "epoch": 0.8415366146458584,
      "grad_norm": 0.0006467948551289737,
      "learning_rate": 0.0001966598891608034,
      "loss": 23.0,
      "step": 1402
    },
    {
      "epoch": 0.8421368547418968,
      "grad_norm": 0.0008227209909819067,
      "learning_rate": 0.000196655045757448,
      "loss": 23.0,
      "step": 1403
    },
    {
      "epoch": 0.8427370948379351,
      "grad_norm": 0.00041979513480328023,
      "learning_rate": 0.00019665019890472262,
      "loss": 23.0,
      "step": 1404
    },
    {
      "epoch": 0.8433373349339736,
      "grad_norm": 0.0006606088718399405,
      "learning_rate": 0.00019664534860280023,
      "loss": 23.0,
      "step": 1405
    },
    {
      "epoch": 0.843937575030012,
      "grad_norm": 0.00072106794686988,
      "learning_rate": 0.000196640494851854,
      "loss": 23.0,
      "step": 1406
    },
    {
      "epoch": 0.8445378151260504,
      "grad_norm": 0.0009728367440402508,
      "learning_rate": 0.00019663563765205707,
      "loss": 23.0,
      "step": 1407
    },
    {
      "epoch": 0.8451380552220888,
      "grad_norm": 0.001023156102746725,
      "learning_rate": 0.0001966307770035828,
      "loss": 23.0,
      "step": 1408
    },
    {
      "epoch": 0.8457382953181273,
      "grad_norm": 0.0005268081440590322,
      "learning_rate": 0.0001966259129066047,
      "loss": 23.0,
      "step": 1409
    },
    {
      "epoch": 0.8463385354141657,
      "grad_norm": 0.0008058066596277058,
      "learning_rate": 0.0001966210453612963,
      "loss": 23.0,
      "step": 1410
    },
    {
      "epoch": 0.8469387755102041,
      "grad_norm": 0.0006653359159827232,
      "learning_rate": 0.00019661617436783133,
      "loss": 23.0,
      "step": 1411
    },
    {
      "epoch": 0.8475390156062425,
      "grad_norm": 0.0004622488922905177,
      "learning_rate": 0.0001966112999263836,
      "loss": 23.0,
      "step": 1412
    },
    {
      "epoch": 0.8481392557022809,
      "grad_norm": 0.00042153761023655534,
      "learning_rate": 0.00019660642203712714,
      "loss": 23.0,
      "step": 1413
    },
    {
      "epoch": 0.8487394957983193,
      "grad_norm": 0.0008719123434275389,
      "learning_rate": 0.00019660154070023597,
      "loss": 23.0,
      "step": 1414
    },
    {
      "epoch": 0.8493397358943577,
      "grad_norm": 0.0012196196475997567,
      "learning_rate": 0.00019659665591588424,
      "loss": 23.0,
      "step": 1415
    },
    {
      "epoch": 0.8499399759903962,
      "grad_norm": 0.0006624732632189989,
      "learning_rate": 0.0001965917676842464,
      "loss": 23.0,
      "step": 1416
    },
    {
      "epoch": 0.8505402160864346,
      "grad_norm": 0.0004318116116337478,
      "learning_rate": 0.0001965868760054968,
      "loss": 23.0,
      "step": 1417
    },
    {
      "epoch": 0.851140456182473,
      "grad_norm": 0.0016290737548843026,
      "learning_rate": 0.00019658198087981008,
      "loss": 23.0,
      "step": 1418
    },
    {
      "epoch": 0.8517406962785115,
      "grad_norm": 0.0006357937818393111,
      "learning_rate": 0.00019657708230736085,
      "loss": 23.0,
      "step": 1419
    },
    {
      "epoch": 0.8523409363745498,
      "grad_norm": 0.0008247081423178315,
      "learning_rate": 0.00019657218028832404,
      "loss": 23.0,
      "step": 1420
    },
    {
      "epoch": 0.8529411764705882,
      "grad_norm": 0.0008558662375435233,
      "learning_rate": 0.0001965672748228745,
      "loss": 23.0,
      "step": 1421
    },
    {
      "epoch": 0.8535414165666266,
      "grad_norm": 0.0006650417344644666,
      "learning_rate": 0.00019656236591118732,
      "loss": 23.0,
      "step": 1422
    },
    {
      "epoch": 0.8541416566626651,
      "grad_norm": 0.0011802432127296925,
      "learning_rate": 0.00019655745355343765,
      "loss": 23.0,
      "step": 1423
    },
    {
      "epoch": 0.8547418967587035,
      "grad_norm": 0.0005101741407997906,
      "learning_rate": 0.0001965525377498009,
      "loss": 23.0,
      "step": 1424
    },
    {
      "epoch": 0.8553421368547419,
      "grad_norm": 0.0006552199483849108,
      "learning_rate": 0.0001965476185004524,
      "loss": 23.0,
      "step": 1425
    },
    {
      "epoch": 0.8559423769507803,
      "grad_norm": 0.0013779390137642622,
      "learning_rate": 0.0001965426958055678,
      "loss": 23.0,
      "step": 1426
    },
    {
      "epoch": 0.8565426170468188,
      "grad_norm": 0.0015933418180793524,
      "learning_rate": 0.00019653776966532265,
      "loss": 23.0,
      "step": 1427
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.0007609377498738468,
      "learning_rate": 0.00019653284007989288,
      "loss": 23.0,
      "step": 1428
    },
    {
      "epoch": 0.8577430972388955,
      "grad_norm": 0.0004458023759070784,
      "learning_rate": 0.00019652790704945434,
      "loss": 23.0,
      "step": 1429
    },
    {
      "epoch": 0.858343337334934,
      "grad_norm": 0.0006291582831181586,
      "learning_rate": 0.0001965229705741831,
      "loss": 23.0,
      "step": 1430
    },
    {
      "epoch": 0.8589435774309724,
      "grad_norm": 0.0006121039623394608,
      "learning_rate": 0.00019651803065425533,
      "loss": 23.0,
      "step": 1431
    },
    {
      "epoch": 0.8595438175270108,
      "grad_norm": 0.0012091121170669794,
      "learning_rate": 0.0001965130872898473,
      "loss": 23.0,
      "step": 1432
    },
    {
      "epoch": 0.8601440576230492,
      "grad_norm": 0.00046850915532559156,
      "learning_rate": 0.00019650814048113548,
      "loss": 23.0,
      "step": 1433
    },
    {
      "epoch": 0.8607442977190877,
      "grad_norm": 0.0009892003145068884,
      "learning_rate": 0.00019650319022829636,
      "loss": 23.0,
      "step": 1434
    },
    {
      "epoch": 0.8613445378151261,
      "grad_norm": 0.0005852010217495263,
      "learning_rate": 0.0001964982365315066,
      "loss": 23.0,
      "step": 1435
    },
    {
      "epoch": 0.8619447779111644,
      "grad_norm": 0.0006273521576076746,
      "learning_rate": 0.00019649327939094303,
      "loss": 23.0,
      "step": 1436
    },
    {
      "epoch": 0.8625450180072028,
      "grad_norm": 0.0006230876315385103,
      "learning_rate": 0.0001964883188067825,
      "loss": 23.0,
      "step": 1437
    },
    {
      "epoch": 0.8631452581032413,
      "grad_norm": 0.0003952436672989279,
      "learning_rate": 0.00019648335477920207,
      "loss": 23.0,
      "step": 1438
    },
    {
      "epoch": 0.8637454981992797,
      "grad_norm": 0.003755589248612523,
      "learning_rate": 0.0001964783873083789,
      "loss": 23.0,
      "step": 1439
    },
    {
      "epoch": 0.8643457382953181,
      "grad_norm": 0.0012677287450060248,
      "learning_rate": 0.00019647341639449027,
      "loss": 23.0,
      "step": 1440
    },
    {
      "epoch": 0.8649459783913566,
      "grad_norm": 0.0004233888175804168,
      "learning_rate": 0.00019646844203771358,
      "loss": 23.0,
      "step": 1441
    },
    {
      "epoch": 0.865546218487395,
      "grad_norm": 0.0010196702787652612,
      "learning_rate": 0.00019646346423822633,
      "loss": 23.0,
      "step": 1442
    },
    {
      "epoch": 0.8661464585834334,
      "grad_norm": 0.0009242582018487155,
      "learning_rate": 0.00019645848299620618,
      "loss": 23.0,
      "step": 1443
    },
    {
      "epoch": 0.8667466986794717,
      "grad_norm": 0.0009520480525679886,
      "learning_rate": 0.00019645349831183087,
      "loss": 23.0,
      "step": 1444
    },
    {
      "epoch": 0.8673469387755102,
      "grad_norm": 0.0017986897146329284,
      "learning_rate": 0.0001964485101852783,
      "loss": 23.0,
      "step": 1445
    },
    {
      "epoch": 0.8679471788715486,
      "grad_norm": 0.0005925934528931975,
      "learning_rate": 0.00019644351861672653,
      "loss": 23.0,
      "step": 1446
    },
    {
      "epoch": 0.868547418967587,
      "grad_norm": 0.0011633182875812054,
      "learning_rate": 0.00019643852360635365,
      "loss": 23.0,
      "step": 1447
    },
    {
      "epoch": 0.8691476590636255,
      "grad_norm": 0.0009133127168752253,
      "learning_rate": 0.00019643352515433794,
      "loss": 23.0,
      "step": 1448
    },
    {
      "epoch": 0.8697478991596639,
      "grad_norm": 0.001071081729605794,
      "learning_rate": 0.00019642852326085778,
      "loss": 23.0,
      "step": 1449
    },
    {
      "epoch": 0.8703481392557023,
      "grad_norm": 0.0005460986285470426,
      "learning_rate": 0.00019642351792609165,
      "loss": 23.0,
      "step": 1450
    },
    {
      "epoch": 0.8709483793517407,
      "grad_norm": 0.0007283551385626197,
      "learning_rate": 0.0001964185091502182,
      "loss": 23.0,
      "step": 1451
    },
    {
      "epoch": 0.8715486194477791,
      "grad_norm": 0.000346360175171867,
      "learning_rate": 0.0001964134969334162,
      "loss": 23.0,
      "step": 1452
    },
    {
      "epoch": 0.8721488595438175,
      "grad_norm": 0.0015917529817670584,
      "learning_rate": 0.00019640848127586445,
      "loss": 23.0,
      "step": 1453
    },
    {
      "epoch": 0.8727490996398559,
      "grad_norm": 0.0009000685531646013,
      "learning_rate": 0.00019640346217774204,
      "loss": 23.0,
      "step": 1454
    },
    {
      "epoch": 0.8733493397358943,
      "grad_norm": 0.00068512256257236,
      "learning_rate": 0.00019639843963922804,
      "loss": 23.0,
      "step": 1455
    },
    {
      "epoch": 0.8739495798319328,
      "grad_norm": 0.0012403937289491296,
      "learning_rate": 0.0001963934136605017,
      "loss": 23.0,
      "step": 1456
    },
    {
      "epoch": 0.8745498199279712,
      "grad_norm": 0.0012238931376487017,
      "learning_rate": 0.00019638838424174235,
      "loss": 23.0,
      "step": 1457
    },
    {
      "epoch": 0.8751500600240096,
      "grad_norm": 0.0005801822990179062,
      "learning_rate": 0.00019638335138312952,
      "loss": 23.0,
      "step": 1458
    },
    {
      "epoch": 0.8757503001200481,
      "grad_norm": 0.0006364347646012902,
      "learning_rate": 0.0001963783150848428,
      "loss": 23.0,
      "step": 1459
    },
    {
      "epoch": 0.8763505402160864,
      "grad_norm": 0.0006346119334921241,
      "learning_rate": 0.00019637327534706195,
      "loss": 23.0,
      "step": 1460
    },
    {
      "epoch": 0.8769507803121248,
      "grad_norm": 0.00040727152372710407,
      "learning_rate": 0.00019636823216996678,
      "loss": 23.0,
      "step": 1461
    },
    {
      "epoch": 0.8775510204081632,
      "grad_norm": 0.0008567664190195501,
      "learning_rate": 0.0001963631855537373,
      "loss": 23.0,
      "step": 1462
    },
    {
      "epoch": 0.8781512605042017,
      "grad_norm": 0.0006258795619942248,
      "learning_rate": 0.0001963581354985536,
      "loss": 23.0,
      "step": 1463
    },
    {
      "epoch": 0.8787515006002401,
      "grad_norm": 0.0005797443445771933,
      "learning_rate": 0.00019635308200459593,
      "loss": 23.0,
      "step": 1464
    },
    {
      "epoch": 0.8793517406962785,
      "grad_norm": 0.0007031543063931167,
      "learning_rate": 0.0001963480250720446,
      "loss": 23.0,
      "step": 1465
    },
    {
      "epoch": 0.879951980792317,
      "grad_norm": 0.0009526876965537667,
      "learning_rate": 0.0001963429647010801,
      "loss": 23.0,
      "step": 1466
    },
    {
      "epoch": 0.8805522208883554,
      "grad_norm": 0.000710951688233763,
      "learning_rate": 0.000196337900891883,
      "loss": 23.0,
      "step": 1467
    },
    {
      "epoch": 0.8811524609843937,
      "grad_norm": 0.0012169844703748822,
      "learning_rate": 0.00019633283364463403,
      "loss": 23.0,
      "step": 1468
    },
    {
      "epoch": 0.8817527010804321,
      "grad_norm": 0.0011517518432810903,
      "learning_rate": 0.00019632776295951403,
      "loss": 23.0,
      "step": 1469
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 0.0008190085063688457,
      "learning_rate": 0.00019632268883670393,
      "loss": 23.0,
      "step": 1470
    },
    {
      "epoch": 0.882953181272509,
      "grad_norm": 0.0016430607065558434,
      "learning_rate": 0.0001963176112763849,
      "loss": 23.0,
      "step": 1471
    },
    {
      "epoch": 0.8835534213685474,
      "grad_norm": 0.0006540712201967835,
      "learning_rate": 0.000196312530278738,
      "loss": 23.0,
      "step": 1472
    },
    {
      "epoch": 0.8841536614645858,
      "grad_norm": 0.0008266113582067192,
      "learning_rate": 0.00019630744584394467,
      "loss": 23.0,
      "step": 1473
    },
    {
      "epoch": 0.8847539015606243,
      "grad_norm": 0.000942007580306381,
      "learning_rate": 0.00019630235797218638,
      "loss": 23.0,
      "step": 1474
    },
    {
      "epoch": 0.8853541416566627,
      "grad_norm": 0.0007080623181536794,
      "learning_rate": 0.0001962972666636446,
      "loss": 23.0,
      "step": 1475
    },
    {
      "epoch": 0.885954381752701,
      "grad_norm": 0.0004341659077908844,
      "learning_rate": 0.0001962921719185011,
      "loss": 23.0,
      "step": 1476
    },
    {
      "epoch": 0.8865546218487395,
      "grad_norm": 0.0004911704454571009,
      "learning_rate": 0.00019628707373693763,
      "loss": 23.0,
      "step": 1477
    },
    {
      "epoch": 0.8871548619447779,
      "grad_norm": 0.00033043770235963166,
      "learning_rate": 0.00019628197211913625,
      "loss": 23.0,
      "step": 1478
    },
    {
      "epoch": 0.8877551020408163,
      "grad_norm": 0.0010875524021685123,
      "learning_rate": 0.0001962768670652789,
      "loss": 23.0,
      "step": 1479
    },
    {
      "epoch": 0.8883553421368547,
      "grad_norm": 0.0007014353759586811,
      "learning_rate": 0.00019627175857554784,
      "loss": 23.0,
      "step": 1480
    },
    {
      "epoch": 0.8889555822328932,
      "grad_norm": 0.0004790904640685767,
      "learning_rate": 0.00019626664665012535,
      "loss": 23.0,
      "step": 1481
    },
    {
      "epoch": 0.8895558223289316,
      "grad_norm": 0.00041154108475893736,
      "learning_rate": 0.00019626153128919386,
      "loss": 23.0,
      "step": 1482
    },
    {
      "epoch": 0.89015606242497,
      "grad_norm": 0.0009722684626467526,
      "learning_rate": 0.00019625641249293595,
      "loss": 23.0,
      "step": 1483
    },
    {
      "epoch": 0.8907563025210085,
      "grad_norm": 0.0011275166179984808,
      "learning_rate": 0.00019625129026153428,
      "loss": 23.0,
      "step": 1484
    },
    {
      "epoch": 0.8913565426170468,
      "grad_norm": 0.0010743034072220325,
      "learning_rate": 0.00019624616459517166,
      "loss": 23.0,
      "step": 1485
    },
    {
      "epoch": 0.8919567827130852,
      "grad_norm": 0.0004264815361239016,
      "learning_rate": 0.000196241035494031,
      "loss": 23.0,
      "step": 1486
    },
    {
      "epoch": 0.8925570228091236,
      "grad_norm": 0.0013276687823235989,
      "learning_rate": 0.0001962359029582953,
      "loss": 23.0,
      "step": 1487
    },
    {
      "epoch": 0.8931572629051621,
      "grad_norm": 0.000943381863180548,
      "learning_rate": 0.0001962307669881478,
      "loss": 23.0,
      "step": 1488
    },
    {
      "epoch": 0.8937575030012005,
      "grad_norm": 0.0006792533094994724,
      "learning_rate": 0.00019622562758377177,
      "loss": 23.0,
      "step": 1489
    },
    {
      "epoch": 0.8943577430972389,
      "grad_norm": 0.0022334514651447535,
      "learning_rate": 0.0001962204847453506,
      "loss": 23.0,
      "step": 1490
    },
    {
      "epoch": 0.8949579831932774,
      "grad_norm": 0.0005680921021848917,
      "learning_rate": 0.00019621533847306784,
      "loss": 23.0,
      "step": 1491
    },
    {
      "epoch": 0.8955582232893158,
      "grad_norm": 0.0007726846379227936,
      "learning_rate": 0.00019621018876710716,
      "loss": 23.0,
      "step": 1492
    },
    {
      "epoch": 0.8961584633853541,
      "grad_norm": 0.0007318484713323414,
      "learning_rate": 0.0001962050356276523,
      "loss": 23.0,
      "step": 1493
    },
    {
      "epoch": 0.8967587034813925,
      "grad_norm": 0.0006041128071956336,
      "learning_rate": 0.00019619987905488721,
      "loss": 23.0,
      "step": 1494
    },
    {
      "epoch": 0.897358943577431,
      "grad_norm": 0.0009723983239382505,
      "learning_rate": 0.00019619471904899587,
      "loss": 23.0,
      "step": 1495
    },
    {
      "epoch": 0.8979591836734694,
      "grad_norm": 0.0008362563094124198,
      "learning_rate": 0.00019618955561016247,
      "loss": 23.0,
      "step": 1496
    },
    {
      "epoch": 0.8985594237695078,
      "grad_norm": 0.0006792604108341038,
      "learning_rate": 0.00019618438873857124,
      "loss": 23.0,
      "step": 1497
    },
    {
      "epoch": 0.8991596638655462,
      "grad_norm": 0.0007895256276242435,
      "learning_rate": 0.0001961792184344066,
      "loss": 23.0,
      "step": 1498
    },
    {
      "epoch": 0.8997599039615847,
      "grad_norm": 0.0011160440044477582,
      "learning_rate": 0.00019617404469785305,
      "loss": 23.0,
      "step": 1499
    },
    {
      "epoch": 0.9003601440576231,
      "grad_norm": 0.0006315569626167417,
      "learning_rate": 0.00019616886752909526,
      "loss": 23.0,
      "step": 1500
    },
    {
      "epoch": 0.9009603841536614,
      "grad_norm": 0.0010441996855661273,
      "learning_rate": 0.00019616368692831795,
      "loss": 23.0,
      "step": 1501
    },
    {
      "epoch": 0.9015606242496998,
      "grad_norm": 0.001420605811290443,
      "learning_rate": 0.000196158502895706,
      "loss": 23.0,
      "step": 1502
    },
    {
      "epoch": 0.9021608643457383,
      "grad_norm": 0.0008463888661935925,
      "learning_rate": 0.00019615331543144446,
      "loss": 23.0,
      "step": 1503
    },
    {
      "epoch": 0.9027611044417767,
      "grad_norm": 0.0011006301501765847,
      "learning_rate": 0.0001961481245357184,
      "loss": 23.0,
      "step": 1504
    },
    {
      "epoch": 0.9033613445378151,
      "grad_norm": 0.000740750867407769,
      "learning_rate": 0.00019614293020871312,
      "loss": 23.0,
      "step": 1505
    },
    {
      "epoch": 0.9039615846338536,
      "grad_norm": 0.0004690020577982068,
      "learning_rate": 0.00019613773245061395,
      "loss": 23.0,
      "step": 1506
    },
    {
      "epoch": 0.904561824729892,
      "grad_norm": 0.0004144386912230402,
      "learning_rate": 0.00019613253126160641,
      "loss": 23.0,
      "step": 1507
    },
    {
      "epoch": 0.9051620648259304,
      "grad_norm": 0.0006933241384103894,
      "learning_rate": 0.00019612732664187613,
      "loss": 23.0,
      "step": 1508
    },
    {
      "epoch": 0.9057623049219687,
      "grad_norm": 0.00047544355038553476,
      "learning_rate": 0.0001961221185916088,
      "loss": 23.0,
      "step": 1509
    },
    {
      "epoch": 0.9063625450180072,
      "grad_norm": 0.0005907033337280154,
      "learning_rate": 0.00019611690711099033,
      "loss": 23.0,
      "step": 1510
    },
    {
      "epoch": 0.9069627851140456,
      "grad_norm": 0.000761785835493356,
      "learning_rate": 0.0001961116922002067,
      "loss": 23.0,
      "step": 1511
    },
    {
      "epoch": 0.907563025210084,
      "grad_norm": 0.000696085742674768,
      "learning_rate": 0.000196106473859444,
      "loss": 23.0,
      "step": 1512
    },
    {
      "epoch": 0.9081632653061225,
      "grad_norm": 0.0013306043110787868,
      "learning_rate": 0.00019610125208888843,
      "loss": 23.0,
      "step": 1513
    },
    {
      "epoch": 0.9087635054021609,
      "grad_norm": 0.0017198032001033425,
      "learning_rate": 0.00019609602688872642,
      "loss": 23.0,
      "step": 1514
    },
    {
      "epoch": 0.9093637454981993,
      "grad_norm": 0.0004617887025233358,
      "learning_rate": 0.00019609079825914436,
      "loss": 23.0,
      "step": 1515
    },
    {
      "epoch": 0.9099639855942377,
      "grad_norm": 0.0006450479850172997,
      "learning_rate": 0.00019608556620032892,
      "loss": 23.0,
      "step": 1516
    },
    {
      "epoch": 0.9105642256902761,
      "grad_norm": 0.000481103896163404,
      "learning_rate": 0.00019608033071246678,
      "loss": 23.0,
      "step": 1517
    },
    {
      "epoch": 0.9111644657863145,
      "grad_norm": 0.0014282638439908624,
      "learning_rate": 0.00019607509179574473,
      "loss": 23.0,
      "step": 1518
    },
    {
      "epoch": 0.9117647058823529,
      "grad_norm": 0.0011594934621825814,
      "learning_rate": 0.00019606984945034985,
      "loss": 23.0,
      "step": 1519
    },
    {
      "epoch": 0.9123649459783914,
      "grad_norm": 0.0007562115788459778,
      "learning_rate": 0.00019606460367646916,
      "loss": 23.0,
      "step": 1520
    },
    {
      "epoch": 0.9129651860744298,
      "grad_norm": 0.0005319854826666415,
      "learning_rate": 0.00019605935447428985,
      "loss": 23.0,
      "step": 1521
    },
    {
      "epoch": 0.9135654261704682,
      "grad_norm": 0.0007216419326141477,
      "learning_rate": 0.00019605410184399927,
      "loss": 23.0,
      "step": 1522
    },
    {
      "epoch": 0.9141656662665066,
      "grad_norm": 0.0007937728078104556,
      "learning_rate": 0.00019604884578578488,
      "loss": 23.0,
      "step": 1523
    },
    {
      "epoch": 0.9147659063625451,
      "grad_norm": 0.0007360963500104845,
      "learning_rate": 0.00019604358629983428,
      "loss": 23.0,
      "step": 1524
    },
    {
      "epoch": 0.9153661464585834,
      "grad_norm": 0.0009595039300620556,
      "learning_rate": 0.00019603832338633512,
      "loss": 23.0,
      "step": 1525
    },
    {
      "epoch": 0.9159663865546218,
      "grad_norm": 0.0013019670732319355,
      "learning_rate": 0.00019603305704547522,
      "loss": 23.0,
      "step": 1526
    },
    {
      "epoch": 0.9165666266506602,
      "grad_norm": 0.0011073811911046505,
      "learning_rate": 0.00019602778727744258,
      "loss": 23.0,
      "step": 1527
    },
    {
      "epoch": 0.9171668667466987,
      "grad_norm": 0.0009854474337771535,
      "learning_rate": 0.0001960225140824252,
      "loss": 23.0,
      "step": 1528
    },
    {
      "epoch": 0.9177671068427371,
      "grad_norm": 0.0018431590870022774,
      "learning_rate": 0.0001960172374606113,
      "loss": 23.0,
      "step": 1529
    },
    {
      "epoch": 0.9183673469387755,
      "grad_norm": 0.0007089927676133811,
      "learning_rate": 0.00019601195741218917,
      "loss": 23.0,
      "step": 1530
    },
    {
      "epoch": 0.918967587034814,
      "grad_norm": 0.0009421886061318219,
      "learning_rate": 0.0001960066739373472,
      "loss": 23.0,
      "step": 1531
    },
    {
      "epoch": 0.9195678271308524,
      "grad_norm": 0.0009645966929383576,
      "learning_rate": 0.00019600138703627408,
      "loss": 23.0,
      "step": 1532
    },
    {
      "epoch": 0.9201680672268907,
      "grad_norm": 0.00032688723877072334,
      "learning_rate": 0.00019599609670915838,
      "loss": 23.0,
      "step": 1533
    },
    {
      "epoch": 0.9207683073229291,
      "grad_norm": 0.0006491580861620605,
      "learning_rate": 0.0001959908029561889,
      "loss": 23.0,
      "step": 1534
    },
    {
      "epoch": 0.9213685474189676,
      "grad_norm": 0.0006321097607724369,
      "learning_rate": 0.00019598550577755458,
      "loss": 23.0,
      "step": 1535
    },
    {
      "epoch": 0.921968787515006,
      "grad_norm": 0.000790178484749049,
      "learning_rate": 0.0001959802051734445,
      "loss": 23.0,
      "step": 1536
    },
    {
      "epoch": 0.9225690276110444,
      "grad_norm": 0.0002988299529533833,
      "learning_rate": 0.00019597490114404775,
      "loss": 23.0,
      "step": 1537
    },
    {
      "epoch": 0.9231692677070829,
      "grad_norm": 0.00046298291999846697,
      "learning_rate": 0.00019596959368955368,
      "loss": 23.0,
      "step": 1538
    },
    {
      "epoch": 0.9237695078031213,
      "grad_norm": 0.0023733177222311497,
      "learning_rate": 0.00019596428281015168,
      "loss": 23.0,
      "step": 1539
    },
    {
      "epoch": 0.9243697478991597,
      "grad_norm": 0.001180400256998837,
      "learning_rate": 0.00019595896850603124,
      "loss": 23.0,
      "step": 1540
    },
    {
      "epoch": 0.924969987995198,
      "grad_norm": 0.0008325580274686217,
      "learning_rate": 0.00019595365077738208,
      "loss": 23.0,
      "step": 1541
    },
    {
      "epoch": 0.9255702280912365,
      "grad_norm": 0.000297403777949512,
      "learning_rate": 0.00019594832962439393,
      "loss": 23.0,
      "step": 1542
    },
    {
      "epoch": 0.9261704681872749,
      "grad_norm": 0.0009298890945501626,
      "learning_rate": 0.0001959430050472567,
      "loss": 23.0,
      "step": 1543
    },
    {
      "epoch": 0.9267707082833133,
      "grad_norm": 0.0006622796645388007,
      "learning_rate": 0.0001959376770461604,
      "loss": 23.0,
      "step": 1544
    },
    {
      "epoch": 0.9273709483793517,
      "grad_norm": 0.0008475258946418762,
      "learning_rate": 0.00019593234562129524,
      "loss": 23.0,
      "step": 1545
    },
    {
      "epoch": 0.9279711884753902,
      "grad_norm": 0.0005952999927103519,
      "learning_rate": 0.0001959270107728514,
      "loss": 23.0,
      "step": 1546
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 0.002795860404148698,
      "learning_rate": 0.0001959216725010193,
      "loss": 23.0,
      "step": 1547
    },
    {
      "epoch": 0.929171668667467,
      "grad_norm": 0.001094569219276309,
      "learning_rate": 0.00019591633080598948,
      "loss": 23.0,
      "step": 1548
    },
    {
      "epoch": 0.9297719087635054,
      "grad_norm": 0.0005409182049334049,
      "learning_rate": 0.0001959109856879525,
      "loss": 23.0,
      "step": 1549
    },
    {
      "epoch": 0.9303721488595438,
      "grad_norm": 0.00046397114056162536,
      "learning_rate": 0.00019590563714709918,
      "loss": 23.0,
      "step": 1550
    },
    {
      "epoch": 0.9309723889555822,
      "grad_norm": 0.001301168231293559,
      "learning_rate": 0.00019590028518362034,
      "loss": 23.0,
      "step": 1551
    },
    {
      "epoch": 0.9315726290516206,
      "grad_norm": 0.0005839786608703434,
      "learning_rate": 0.00019589492979770704,
      "loss": 23.0,
      "step": 1552
    },
    {
      "epoch": 0.9321728691476591,
      "grad_norm": 0.0005165533511899412,
      "learning_rate": 0.00019588957098955033,
      "loss": 23.0,
      "step": 1553
    },
    {
      "epoch": 0.9327731092436975,
      "grad_norm": 0.0005963983130641282,
      "learning_rate": 0.0001958842087593415,
      "loss": 23.0,
      "step": 1554
    },
    {
      "epoch": 0.9333733493397359,
      "grad_norm": 0.001547312829643488,
      "learning_rate": 0.0001958788431072719,
      "loss": 23.0,
      "step": 1555
    },
    {
      "epoch": 0.9339735894357744,
      "grad_norm": 0.000669624307192862,
      "learning_rate": 0.00019587347403353302,
      "loss": 23.0,
      "step": 1556
    },
    {
      "epoch": 0.9345738295318127,
      "grad_norm": 0.0019656869117170572,
      "learning_rate": 0.0001958681015383165,
      "loss": 23.0,
      "step": 1557
    },
    {
      "epoch": 0.9351740696278511,
      "grad_norm": 0.0012716340133920312,
      "learning_rate": 0.000195862725621814,
      "loss": 23.0,
      "step": 1558
    },
    {
      "epoch": 0.9357743097238895,
      "grad_norm": 0.0007798285805620253,
      "learning_rate": 0.00019585734628421744,
      "loss": 23.0,
      "step": 1559
    },
    {
      "epoch": 0.936374549819928,
      "grad_norm": 0.0011394252069294453,
      "learning_rate": 0.00019585196352571872,
      "loss": 23.0,
      "step": 1560
    },
    {
      "epoch": 0.9369747899159664,
      "grad_norm": 0.000518056214787066,
      "learning_rate": 0.00019584657734651005,
      "loss": 23.0,
      "step": 1561
    },
    {
      "epoch": 0.9375750300120048,
      "grad_norm": 0.0007476578466594219,
      "learning_rate": 0.00019584118774678352,
      "loss": 23.0,
      "step": 1562
    },
    {
      "epoch": 0.9381752701080432,
      "grad_norm": 0.0008059003739617765,
      "learning_rate": 0.00019583579472673158,
      "loss": 23.0,
      "step": 1563
    },
    {
      "epoch": 0.9387755102040817,
      "grad_norm": 0.0019434284185990691,
      "learning_rate": 0.00019583039828654662,
      "loss": 23.0,
      "step": 1564
    },
    {
      "epoch": 0.93937575030012,
      "grad_norm": 0.0008690199465490878,
      "learning_rate": 0.00019582499842642122,
      "loss": 23.0,
      "step": 1565
    },
    {
      "epoch": 0.9399759903961584,
      "grad_norm": 0.0008223998011089861,
      "learning_rate": 0.00019581959514654816,
      "loss": 23.0,
      "step": 1566
    },
    {
      "epoch": 0.9405762304921969,
      "grad_norm": 0.001912278588861227,
      "learning_rate": 0.00019581418844712023,
      "loss": 23.0,
      "step": 1567
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.0009676693007349968,
      "learning_rate": 0.00019580877832833037,
      "loss": 23.0,
      "step": 1568
    },
    {
      "epoch": 0.9417767106842737,
      "grad_norm": 0.0007918410701677203,
      "learning_rate": 0.00019580336479037164,
      "loss": 23.0,
      "step": 1569
    },
    {
      "epoch": 0.9423769507803121,
      "grad_norm": 0.0008247760706581175,
      "learning_rate": 0.00019579794783343728,
      "loss": 23.0,
      "step": 1570
    },
    {
      "epoch": 0.9429771908763506,
      "grad_norm": 0.0006936482386663556,
      "learning_rate": 0.0001957925274577206,
      "loss": 23.0,
      "step": 1571
    },
    {
      "epoch": 0.943577430972389,
      "grad_norm": 0.001130030956119299,
      "learning_rate": 0.00019578710366341498,
      "loss": 23.0,
      "step": 1572
    },
    {
      "epoch": 0.9441776710684273,
      "grad_norm": 0.0017003900138661265,
      "learning_rate": 0.00019578167645071403,
      "loss": 23.0,
      "step": 1573
    },
    {
      "epoch": 0.9447779111644657,
      "grad_norm": 0.00029980865656398237,
      "learning_rate": 0.00019577624581981144,
      "loss": 23.0,
      "step": 1574
    },
    {
      "epoch": 0.9453781512605042,
      "grad_norm": 0.0006558767054229975,
      "learning_rate": 0.000195770811770901,
      "loss": 23.0,
      "step": 1575
    },
    {
      "epoch": 0.9459783913565426,
      "grad_norm": 0.0006806926685385406,
      "learning_rate": 0.00019576537430417666,
      "loss": 23.0,
      "step": 1576
    },
    {
      "epoch": 0.946578631452581,
      "grad_norm": 0.000988876330666244,
      "learning_rate": 0.00019575993341983244,
      "loss": 23.0,
      "step": 1577
    },
    {
      "epoch": 0.9471788715486195,
      "grad_norm": 0.0013003622880205512,
      "learning_rate": 0.00019575448911806248,
      "loss": 23.0,
      "step": 1578
    },
    {
      "epoch": 0.9477791116446579,
      "grad_norm": 0.0005484142457135022,
      "learning_rate": 0.00019574904139906115,
      "loss": 23.0,
      "step": 1579
    },
    {
      "epoch": 0.9483793517406963,
      "grad_norm": 0.0021183365024626255,
      "learning_rate": 0.00019574359026302283,
      "loss": 23.0,
      "step": 1580
    },
    {
      "epoch": 0.9489795918367347,
      "grad_norm": 0.0008176713599823415,
      "learning_rate": 0.00019573813571014203,
      "loss": 23.0,
      "step": 1581
    },
    {
      "epoch": 0.9495798319327731,
      "grad_norm": 0.0007823684718459845,
      "learning_rate": 0.00019573267774061347,
      "loss": 23.0,
      "step": 1582
    },
    {
      "epoch": 0.9501800720288115,
      "grad_norm": 0.0012796282535418868,
      "learning_rate": 0.0001957272163546319,
      "loss": 23.0,
      "step": 1583
    },
    {
      "epoch": 0.9507803121248499,
      "grad_norm": 0.0009288830915465951,
      "learning_rate": 0.0001957217515523922,
      "loss": 23.0,
      "step": 1584
    },
    {
      "epoch": 0.9513805522208884,
      "grad_norm": 0.0009111070539802313,
      "learning_rate": 0.0001957162833340894,
      "loss": 23.0,
      "step": 1585
    },
    {
      "epoch": 0.9519807923169268,
      "grad_norm": 0.0007150248857215047,
      "learning_rate": 0.00019571081169991866,
      "loss": 23.0,
      "step": 1586
    },
    {
      "epoch": 0.9525810324129652,
      "grad_norm": 0.0008457883959636092,
      "learning_rate": 0.00019570533665007526,
      "loss": 23.0,
      "step": 1587
    },
    {
      "epoch": 0.9531812725090036,
      "grad_norm": 0.0007468207040801644,
      "learning_rate": 0.0001956998581847546,
      "loss": 23.0,
      "step": 1588
    },
    {
      "epoch": 0.9537815126050421,
      "grad_norm": 0.0007428768440149724,
      "learning_rate": 0.0001956943763041521,
      "loss": 23.0,
      "step": 1589
    },
    {
      "epoch": 0.9543817527010804,
      "grad_norm": 0.00031296900124289095,
      "learning_rate": 0.00019568889100846353,
      "loss": 23.0,
      "step": 1590
    },
    {
      "epoch": 0.9549819927971188,
      "grad_norm": 0.001030638930387795,
      "learning_rate": 0.00019568340229788456,
      "loss": 23.0,
      "step": 1591
    },
    {
      "epoch": 0.9555822328931572,
      "grad_norm": 0.0011427139397710562,
      "learning_rate": 0.00019567791017261108,
      "loss": 23.0,
      "step": 1592
    },
    {
      "epoch": 0.9561824729891957,
      "grad_norm": 0.0013832402182742953,
      "learning_rate": 0.0001956724146328391,
      "loss": 23.0,
      "step": 1593
    },
    {
      "epoch": 0.9567827130852341,
      "grad_norm": 0.001810853835195303,
      "learning_rate": 0.00019566691567876477,
      "loss": 23.0,
      "step": 1594
    },
    {
      "epoch": 0.9573829531812725,
      "grad_norm": 0.0009885139297693968,
      "learning_rate": 0.0001956614133105843,
      "loss": 23.0,
      "step": 1595
    },
    {
      "epoch": 0.957983193277311,
      "grad_norm": 0.0010269538033753633,
      "learning_rate": 0.00019565590752849404,
      "loss": 23.0,
      "step": 1596
    },
    {
      "epoch": 0.9585834333733494,
      "grad_norm": 0.0007400467875413597,
      "learning_rate": 0.0001956503983326905,
      "loss": 23.0,
      "step": 1597
    },
    {
      "epoch": 0.9591836734693877,
      "grad_norm": 0.0005352499429136515,
      "learning_rate": 0.00019564488572337027,
      "loss": 23.0,
      "step": 1598
    },
    {
      "epoch": 0.9597839135654261,
      "grad_norm": 0.0011886239517480135,
      "learning_rate": 0.0001956393697007301,
      "loss": 23.0,
      "step": 1599
    },
    {
      "epoch": 0.9603841536614646,
      "grad_norm": 0.0006831000209785998,
      "learning_rate": 0.00019563385026496688,
      "loss": 23.0,
      "step": 1600
    },
    {
      "epoch": 0.960984393757503,
      "grad_norm": 0.001316646346822381,
      "learning_rate": 0.0001956283274162775,
      "loss": 23.0,
      "step": 1601
    },
    {
      "epoch": 0.9615846338535414,
      "grad_norm": 0.0008040472748689353,
      "learning_rate": 0.00019562280115485915,
      "loss": 23.0,
      "step": 1602
    },
    {
      "epoch": 0.9621848739495799,
      "grad_norm": 0.0010138957295566797,
      "learning_rate": 0.00019561727148090894,
      "loss": 23.0,
      "step": 1603
    },
    {
      "epoch": 0.9627851140456183,
      "grad_norm": 0.0006073630647733808,
      "learning_rate": 0.00019561173839462427,
      "loss": 23.0,
      "step": 1604
    },
    {
      "epoch": 0.9633853541416567,
      "grad_norm": 0.000547137635294348,
      "learning_rate": 0.00019560620189620263,
      "loss": 23.0,
      "step": 1605
    },
    {
      "epoch": 0.963985594237695,
      "grad_norm": 0.000544386391993612,
      "learning_rate": 0.00019560066198584156,
      "loss": 23.0,
      "step": 1606
    },
    {
      "epoch": 0.9645858343337335,
      "grad_norm": 0.000897867837920785,
      "learning_rate": 0.00019559511866373882,
      "loss": 23.0,
      "step": 1607
    },
    {
      "epoch": 0.9651860744297719,
      "grad_norm": 0.0004369942471385002,
      "learning_rate": 0.00019558957193009213,
      "loss": 23.0,
      "step": 1608
    },
    {
      "epoch": 0.9657863145258103,
      "grad_norm": 0.0010517592309042811,
      "learning_rate": 0.00019558402178509952,
      "loss": 23.0,
      "step": 1609
    },
    {
      "epoch": 0.9663865546218487,
      "grad_norm": 0.00039016862865537405,
      "learning_rate": 0.00019557846822895907,
      "loss": 23.0,
      "step": 1610
    },
    {
      "epoch": 0.9669867947178872,
      "grad_norm": 0.0016062683425843716,
      "learning_rate": 0.00019557291126186895,
      "loss": 23.0,
      "step": 1611
    },
    {
      "epoch": 0.9675870348139256,
      "grad_norm": 0.0006835482781752944,
      "learning_rate": 0.00019556735088402743,
      "loss": 23.0,
      "step": 1612
    },
    {
      "epoch": 0.968187274909964,
      "grad_norm": 0.000899162725545466,
      "learning_rate": 0.000195561787095633,
      "loss": 23.0,
      "step": 1613
    },
    {
      "epoch": 0.9687875150060024,
      "grad_norm": 0.0005444439593702555,
      "learning_rate": 0.0001955562198968842,
      "loss": 23.0,
      "step": 1614
    },
    {
      "epoch": 0.9693877551020408,
      "grad_norm": 0.0008718172903172672,
      "learning_rate": 0.00019555064928797973,
      "loss": 23.0,
      "step": 1615
    },
    {
      "epoch": 0.9699879951980792,
      "grad_norm": 0.0003390471974853426,
      "learning_rate": 0.00019554507526911837,
      "loss": 23.0,
      "step": 1616
    },
    {
      "epoch": 0.9705882352941176,
      "grad_norm": 0.000555711449123919,
      "learning_rate": 0.00019553949784049902,
      "loss": 23.0,
      "step": 1617
    },
    {
      "epoch": 0.9711884753901561,
      "grad_norm": 0.0003842646547127515,
      "learning_rate": 0.00019553391700232077,
      "loss": 23.0,
      "step": 1618
    },
    {
      "epoch": 0.9717887154861945,
      "grad_norm": 0.0005270171677693725,
      "learning_rate": 0.00019552833275478278,
      "loss": 23.0,
      "step": 1619
    },
    {
      "epoch": 0.9723889555822329,
      "grad_norm": 0.0006583972717635334,
      "learning_rate": 0.00019552274509808428,
      "loss": 23.0,
      "step": 1620
    },
    {
      "epoch": 0.9729891956782714,
      "grad_norm": 0.0007377169677056372,
      "learning_rate": 0.00019551715403242476,
      "loss": 23.0,
      "step": 1621
    },
    {
      "epoch": 0.9735894357743097,
      "grad_norm": 0.0012161527993157506,
      "learning_rate": 0.0001955115595580037,
      "loss": 23.0,
      "step": 1622
    },
    {
      "epoch": 0.9741896758703481,
      "grad_norm": 0.00046388438204303384,
      "learning_rate": 0.00019550596167502076,
      "loss": 23.0,
      "step": 1623
    },
    {
      "epoch": 0.9747899159663865,
      "grad_norm": 0.0014881680253893137,
      "learning_rate": 0.0001955003603836757,
      "loss": 23.0,
      "step": 1624
    },
    {
      "epoch": 0.975390156062425,
      "grad_norm": 0.0010795058915391564,
      "learning_rate": 0.00019549475568416848,
      "loss": 23.0,
      "step": 1625
    },
    {
      "epoch": 0.9759903961584634,
      "grad_norm": 0.0007747529307380319,
      "learning_rate": 0.00019548914757669901,
      "loss": 23.0,
      "step": 1626
    },
    {
      "epoch": 0.9765906362545018,
      "grad_norm": 0.0010959148639813066,
      "learning_rate": 0.00019548353606146755,
      "loss": 23.0,
      "step": 1627
    },
    {
      "epoch": 0.9771908763505402,
      "grad_norm": 0.0006908348295837641,
      "learning_rate": 0.00019547792113867426,
      "loss": 23.0,
      "step": 1628
    },
    {
      "epoch": 0.9777911164465787,
      "grad_norm": 0.000946520478464663,
      "learning_rate": 0.0001954723028085196,
      "loss": 23.0,
      "step": 1629
    },
    {
      "epoch": 0.978391356542617,
      "grad_norm": 0.0008105000597424805,
      "learning_rate": 0.00019546668107120403,
      "loss": 23.0,
      "step": 1630
    },
    {
      "epoch": 0.9789915966386554,
      "grad_norm": 0.0005076247034594417,
      "learning_rate": 0.00019546105592692817,
      "loss": 23.0,
      "step": 1631
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 0.0004834830469917506,
      "learning_rate": 0.00019545542737589276,
      "loss": 23.0,
      "step": 1632
    },
    {
      "epoch": 0.9801920768307323,
      "grad_norm": 0.00032156877568922937,
      "learning_rate": 0.00019544979541829874,
      "loss": 23.0,
      "step": 1633
    },
    {
      "epoch": 0.9807923169267707,
      "grad_norm": 0.0005227511865086854,
      "learning_rate": 0.00019544416005434697,
      "loss": 23.0,
      "step": 1634
    },
    {
      "epoch": 0.9813925570228091,
      "grad_norm": 0.001222484279423952,
      "learning_rate": 0.00019543852128423868,
      "loss": 23.0,
      "step": 1635
    },
    {
      "epoch": 0.9819927971188476,
      "grad_norm": 0.0010409961687400937,
      "learning_rate": 0.00019543287910817506,
      "loss": 23.0,
      "step": 1636
    },
    {
      "epoch": 0.982593037214886,
      "grad_norm": 0.0007245879387483001,
      "learning_rate": 0.0001954272335263575,
      "loss": 23.0,
      "step": 1637
    },
    {
      "epoch": 0.9831932773109243,
      "grad_norm": 0.00042797072092071176,
      "learning_rate": 0.00019542158453898743,
      "loss": 23.0,
      "step": 1638
    },
    {
      "epoch": 0.9837935174069627,
      "grad_norm": 0.0005213615368120372,
      "learning_rate": 0.00019541593214626642,
      "loss": 23.0,
      "step": 1639
    },
    {
      "epoch": 0.9843937575030012,
      "grad_norm": 0.0005875686765648425,
      "learning_rate": 0.00019541027634839626,
      "loss": 23.0,
      "step": 1640
    },
    {
      "epoch": 0.9849939975990396,
      "grad_norm": 0.0010960171930491924,
      "learning_rate": 0.0001954046171455788,
      "loss": 23.0,
      "step": 1641
    },
    {
      "epoch": 0.985594237695078,
      "grad_norm": 0.002143792575225234,
      "learning_rate": 0.00019539895453801588,
      "loss": 23.0,
      "step": 1642
    },
    {
      "epoch": 0.9861944777911165,
      "grad_norm": 0.001718964776955545,
      "learning_rate": 0.0001953932885259097,
      "loss": 23.0,
      "step": 1643
    },
    {
      "epoch": 0.9867947178871549,
      "grad_norm": 0.0011779890628531575,
      "learning_rate": 0.00019538761910946247,
      "loss": 23.0,
      "step": 1644
    },
    {
      "epoch": 0.9873949579831933,
      "grad_norm": 0.0014262180775403976,
      "learning_rate": 0.00019538194628887645,
      "loss": 23.0,
      "step": 1645
    },
    {
      "epoch": 0.9879951980792316,
      "grad_norm": 0.0003844281018245965,
      "learning_rate": 0.00019537627006435413,
      "loss": 23.0,
      "step": 1646
    },
    {
      "epoch": 0.9885954381752701,
      "grad_norm": 0.0007693939842283726,
      "learning_rate": 0.00019537059043609807,
      "loss": 23.0,
      "step": 1647
    },
    {
      "epoch": 0.9891956782713085,
      "grad_norm": 0.0002765923854894936,
      "learning_rate": 0.00019536490740431096,
      "loss": 23.0,
      "step": 1648
    },
    {
      "epoch": 0.9897959183673469,
      "grad_norm": 0.00032293866388499737,
      "learning_rate": 0.0001953592209691956,
      "loss": 23.0,
      "step": 1649
    },
    {
      "epoch": 0.9903961584633854,
      "grad_norm": 0.0011859709629788995,
      "learning_rate": 0.00019535353113095494,
      "loss": 23.0,
      "step": 1650
    },
    {
      "epoch": 0.9909963985594238,
      "grad_norm": 0.0023132844362407923,
      "learning_rate": 0.00019534783788979204,
      "loss": 23.0,
      "step": 1651
    },
    {
      "epoch": 0.9915966386554622,
      "grad_norm": 0.0004042698710691184,
      "learning_rate": 0.00019534214124591006,
      "loss": 23.0,
      "step": 1652
    },
    {
      "epoch": 0.9921968787515006,
      "grad_norm": 0.0006533717969432473,
      "learning_rate": 0.0001953364411995123,
      "loss": 23.0,
      "step": 1653
    },
    {
      "epoch": 0.992797118847539,
      "grad_norm": 0.000819582084659487,
      "learning_rate": 0.00019533073775080222,
      "loss": 23.0,
      "step": 1654
    },
    {
      "epoch": 0.9933973589435774,
      "grad_norm": 0.00045290245907381177,
      "learning_rate": 0.0001953250308999833,
      "loss": 23.0,
      "step": 1655
    },
    {
      "epoch": 0.9939975990396158,
      "grad_norm": 0.0009432425140403211,
      "learning_rate": 0.00019531932064725924,
      "loss": 23.0,
      "step": 1656
    },
    {
      "epoch": 0.9945978391356542,
      "grad_norm": 0.0010298251872882247,
      "learning_rate": 0.0001953136069928338,
      "loss": 23.0,
      "step": 1657
    },
    {
      "epoch": 0.9951980792316927,
      "grad_norm": 0.00044269522186368704,
      "learning_rate": 0.00019530788993691095,
      "loss": 23.0,
      "step": 1658
    },
    {
      "epoch": 0.9957983193277311,
      "grad_norm": 0.0016579084331169724,
      "learning_rate": 0.00019530216947969464,
      "loss": 23.0,
      "step": 1659
    },
    {
      "epoch": 0.9963985594237695,
      "grad_norm": 0.0013545328984037042,
      "learning_rate": 0.00019529644562138904,
      "loss": 23.0,
      "step": 1660
    },
    {
      "epoch": 0.996998799519808,
      "grad_norm": 0.0006384035223163664,
      "learning_rate": 0.00019529071836219846,
      "loss": 23.0,
      "step": 1661
    },
    {
      "epoch": 0.9975990396158463,
      "grad_norm": 0.0019687407184392214,
      "learning_rate": 0.0001952849877023272,
      "loss": 23.0,
      "step": 1662
    },
    {
      "epoch": 0.9981992797118847,
      "grad_norm": 0.0010327474446967244,
      "learning_rate": 0.00019527925364197987,
      "loss": 23.0,
      "step": 1663
    },
    {
      "epoch": 0.9987995198079231,
      "grad_norm": 0.0003291741304565221,
      "learning_rate": 0.00019527351618136107,
      "loss": 23.0,
      "step": 1664
    },
    {
      "epoch": 0.9993997599039616,
      "grad_norm": 0.0008972626528702676,
      "learning_rate": 0.00019526777532067553,
      "loss": 23.0,
      "step": 1665
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0016940529458224773,
      "learning_rate": 0.00019526203106012816,
      "loss": 23.0,
      "step": 1666
    },
    {
      "epoch": 1.0006002400960383,
      "grad_norm": 0.00041232057265006006,
      "learning_rate": 0.00019525628339992393,
      "loss": 23.0,
      "step": 1667
    },
    {
      "epoch": 1.0012004801920769,
      "grad_norm": 0.0018668645061552525,
      "learning_rate": 0.00019525053234026803,
      "loss": 23.0,
      "step": 1668
    },
    {
      "epoch": 1.0018007202881152,
      "grad_norm": 0.0006821074057370424,
      "learning_rate": 0.00019524477788136557,
      "loss": 23.0,
      "step": 1669
    },
    {
      "epoch": 1.0024009603841537,
      "grad_norm": 0.0022220544051378965,
      "learning_rate": 0.000195239020023422,
      "loss": 23.0,
      "step": 1670
    },
    {
      "epoch": 1.003001200480192,
      "grad_norm": 0.0005048304446972907,
      "learning_rate": 0.00019523325876664283,
      "loss": 23.0,
      "step": 1671
    },
    {
      "epoch": 1.0036014405762306,
      "grad_norm": 0.0006994889699853957,
      "learning_rate": 0.0001952274941112336,
      "loss": 23.0,
      "step": 1672
    },
    {
      "epoch": 1.004201680672269,
      "grad_norm": 0.00142678152769804,
      "learning_rate": 0.00019522172605740008,
      "loss": 23.0,
      "step": 1673
    },
    {
      "epoch": 1.0048019207683074,
      "grad_norm": 0.0002718534669838846,
      "learning_rate": 0.00019521595460534808,
      "loss": 23.0,
      "step": 1674
    },
    {
      "epoch": 1.0054021608643458,
      "grad_norm": 0.0011870613088831306,
      "learning_rate": 0.00019521017975528357,
      "loss": 23.0,
      "step": 1675
    },
    {
      "epoch": 1.006002400960384,
      "grad_norm": 0.0005727344541810453,
      "learning_rate": 0.00019520440150741267,
      "loss": 23.0,
      "step": 1676
    },
    {
      "epoch": 1.0066026410564226,
      "grad_norm": 0.0013802223838865757,
      "learning_rate": 0.00019519861986194156,
      "loss": 23.0,
      "step": 1677
    },
    {
      "epoch": 1.007202881152461,
      "grad_norm": 0.0009242665255442262,
      "learning_rate": 0.0001951928348190766,
      "loss": 23.0,
      "step": 1678
    },
    {
      "epoch": 1.0078031212484995,
      "grad_norm": 0.000813992868643254,
      "learning_rate": 0.00019518704637902422,
      "loss": 23.0,
      "step": 1679
    },
    {
      "epoch": 1.0084033613445378,
      "grad_norm": 0.0006548427045345306,
      "learning_rate": 0.000195181254541991,
      "loss": 23.0,
      "step": 1680
    },
    {
      "epoch": 1.0090036014405763,
      "grad_norm": 0.00041556128417141736,
      "learning_rate": 0.00019517545930818368,
      "loss": 23.0,
      "step": 1681
    },
    {
      "epoch": 1.0096038415366146,
      "grad_norm": 0.0006826121243648231,
      "learning_rate": 0.00019516966067780898,
      "loss": 23.0,
      "step": 1682
    },
    {
      "epoch": 1.010204081632653,
      "grad_norm": 0.0007658922113478184,
      "learning_rate": 0.00019516385865107395,
      "loss": 23.0,
      "step": 1683
    },
    {
      "epoch": 1.0108043217286915,
      "grad_norm": 0.0010680507402867079,
      "learning_rate": 0.00019515805322818558,
      "loss": 23.0,
      "step": 1684
    },
    {
      "epoch": 1.0114045618247298,
      "grad_norm": 0.0007877147872932255,
      "learning_rate": 0.00019515224440935105,
      "loss": 23.0,
      "step": 1685
    },
    {
      "epoch": 1.0120048019207684,
      "grad_norm": 0.0010208765743300319,
      "learning_rate": 0.0001951464321947777,
      "loss": 23.0,
      "step": 1686
    },
    {
      "epoch": 1.0126050420168067,
      "grad_norm": 0.000991889159195125,
      "learning_rate": 0.00019514061658467292,
      "loss": 23.0,
      "step": 1687
    },
    {
      "epoch": 1.0132052821128452,
      "grad_norm": 0.0005913240602239966,
      "learning_rate": 0.00019513479757924426,
      "loss": 23.0,
      "step": 1688
    },
    {
      "epoch": 1.0138055222088835,
      "grad_norm": 0.0007095134351402521,
      "learning_rate": 0.00019512897517869938,
      "loss": 23.0,
      "step": 1689
    },
    {
      "epoch": 1.014405762304922,
      "grad_norm": 0.002312737051397562,
      "learning_rate": 0.0001951231493832461,
      "loss": 23.0,
      "step": 1690
    },
    {
      "epoch": 1.0150060024009604,
      "grad_norm": 0.0005010634195059538,
      "learning_rate": 0.0001951173201930923,
      "loss": 23.0,
      "step": 1691
    },
    {
      "epoch": 1.0156062424969987,
      "grad_norm": 0.000687493709847331,
      "learning_rate": 0.000195111487608446,
      "loss": 23.0,
      "step": 1692
    },
    {
      "epoch": 1.0162064825930373,
      "grad_norm": 0.0007613154011778533,
      "learning_rate": 0.00019510565162951537,
      "loss": 23.0,
      "step": 1693
    },
    {
      "epoch": 1.0168067226890756,
      "grad_norm": 0.0009327959851361811,
      "learning_rate": 0.00019509981225650867,
      "loss": 23.0,
      "step": 1694
    },
    {
      "epoch": 1.017406962785114,
      "grad_norm": 0.0010371070820838213,
      "learning_rate": 0.0001950939694896343,
      "loss": 23.0,
      "step": 1695
    },
    {
      "epoch": 1.0180072028811524,
      "grad_norm": 0.0004927398986183107,
      "learning_rate": 0.00019508812332910078,
      "loss": 23.0,
      "step": 1696
    },
    {
      "epoch": 1.018607442977191,
      "grad_norm": 0.001125394832342863,
      "learning_rate": 0.0001950822737751167,
      "loss": 23.0,
      "step": 1697
    },
    {
      "epoch": 1.0192076830732293,
      "grad_norm": 0.0010863045463338494,
      "learning_rate": 0.00019507642082789087,
      "loss": 23.0,
      "step": 1698
    },
    {
      "epoch": 1.0198079231692676,
      "grad_norm": 0.001297933398745954,
      "learning_rate": 0.0001950705644876322,
      "loss": 23.0,
      "step": 1699
    },
    {
      "epoch": 1.0204081632653061,
      "grad_norm": 0.0008561917347833514,
      "learning_rate": 0.00019506470475454956,
      "loss": 23.0,
      "step": 1700
    },
    {
      "epoch": 1.0210084033613445,
      "grad_norm": 0.000712038017809391,
      "learning_rate": 0.00019505884162885218,
      "loss": 23.0,
      "step": 1701
    },
    {
      "epoch": 1.021608643457383,
      "grad_norm": 0.0004501239163801074,
      "learning_rate": 0.00019505297511074926,
      "loss": 23.0,
      "step": 1702
    },
    {
      "epoch": 1.0222088835534213,
      "grad_norm": 0.0008231149986386299,
      "learning_rate": 0.00019504710520045016,
      "loss": 23.0,
      "step": 1703
    },
    {
      "epoch": 1.0228091236494599,
      "grad_norm": 0.0007951670559123158,
      "learning_rate": 0.00019504123189816435,
      "loss": 23.0,
      "step": 1704
    },
    {
      "epoch": 1.0234093637454982,
      "grad_norm": 0.0006479431758634746,
      "learning_rate": 0.00019503535520410146,
      "loss": 23.0,
      "step": 1705
    },
    {
      "epoch": 1.0240096038415367,
      "grad_norm": 0.0005281239282339811,
      "learning_rate": 0.0001950294751184712,
      "loss": 23.0,
      "step": 1706
    },
    {
      "epoch": 1.024609843937575,
      "grad_norm": 0.000501799862831831,
      "learning_rate": 0.0001950235916414834,
      "loss": 23.0,
      "step": 1707
    },
    {
      "epoch": 1.0252100840336134,
      "grad_norm": 0.0005454413476400077,
      "learning_rate": 0.0001950177047733481,
      "loss": 23.0,
      "step": 1708
    },
    {
      "epoch": 1.025810324129652,
      "grad_norm": 0.0004704859165940434,
      "learning_rate": 0.0001950118145142753,
      "loss": 23.0,
      "step": 1709
    },
    {
      "epoch": 1.0264105642256902,
      "grad_norm": 0.0006746204453520477,
      "learning_rate": 0.00019500592086447522,
      "loss": 23.0,
      "step": 1710
    },
    {
      "epoch": 1.0270108043217288,
      "grad_norm": 0.0018217653268948197,
      "learning_rate": 0.00019500002382415822,
      "loss": 23.0,
      "step": 1711
    },
    {
      "epoch": 1.027611044417767,
      "grad_norm": 0.0017384163802489638,
      "learning_rate": 0.00019499412339353477,
      "loss": 23.0,
      "step": 1712
    },
    {
      "epoch": 1.0282112845138056,
      "grad_norm": 0.001107073505409062,
      "learning_rate": 0.00019498821957281536,
      "loss": 23.0,
      "step": 1713
    },
    {
      "epoch": 1.028811524609844,
      "grad_norm": 0.001236962154507637,
      "learning_rate": 0.00019498231236221076,
      "loss": 23.0,
      "step": 1714
    },
    {
      "epoch": 1.0294117647058822,
      "grad_norm": 0.0008350252173841,
      "learning_rate": 0.00019497640176193175,
      "loss": 23.0,
      "step": 1715
    },
    {
      "epoch": 1.0300120048019208,
      "grad_norm": 0.0014757871394976974,
      "learning_rate": 0.00019497048777218927,
      "loss": 23.0,
      "step": 1716
    },
    {
      "epoch": 1.030612244897959,
      "grad_norm": 0.0012261768570169806,
      "learning_rate": 0.00019496457039319437,
      "loss": 23.0,
      "step": 1717
    },
    {
      "epoch": 1.0312124849939976,
      "grad_norm": 0.0005536626558750868,
      "learning_rate": 0.00019495864962515823,
      "loss": 23.0,
      "step": 1718
    },
    {
      "epoch": 1.031812725090036,
      "grad_norm": 0.000878203718457371,
      "learning_rate": 0.00019495272546829218,
      "loss": 23.0,
      "step": 1719
    },
    {
      "epoch": 1.0324129651860745,
      "grad_norm": 0.0019284028094261885,
      "learning_rate": 0.00019494679792280754,
      "loss": 23.0,
      "step": 1720
    },
    {
      "epoch": 1.0330132052821128,
      "grad_norm": 0.0012051918311044574,
      "learning_rate": 0.000194940866988916,
      "loss": 23.0,
      "step": 1721
    },
    {
      "epoch": 1.0336134453781514,
      "grad_norm": 0.0010601821122691035,
      "learning_rate": 0.00019493493266682906,
      "loss": 23.0,
      "step": 1722
    },
    {
      "epoch": 1.0342136854741897,
      "grad_norm": 0.00037417240673676133,
      "learning_rate": 0.0001949289949567586,
      "loss": 23.0,
      "step": 1723
    },
    {
      "epoch": 1.034813925570228,
      "grad_norm": 0.0007202907581813633,
      "learning_rate": 0.0001949230538589165,
      "loss": 23.0,
      "step": 1724
    },
    {
      "epoch": 1.0354141656662665,
      "grad_norm": 0.0011258620070293546,
      "learning_rate": 0.0001949171093735148,
      "loss": 23.0,
      "step": 1725
    },
    {
      "epoch": 1.0360144057623049,
      "grad_norm": 0.0006037678685970604,
      "learning_rate": 0.0001949111615007656,
      "loss": 23.0,
      "step": 1726
    },
    {
      "epoch": 1.0366146458583434,
      "grad_norm": 0.0015769102610647678,
      "learning_rate": 0.00019490521024088117,
      "loss": 23.0,
      "step": 1727
    },
    {
      "epoch": 1.0372148859543817,
      "grad_norm": 0.001167101669125259,
      "learning_rate": 0.00019489925559407393,
      "loss": 23.0,
      "step": 1728
    },
    {
      "epoch": 1.0378151260504203,
      "grad_norm": 0.0004130336456000805,
      "learning_rate": 0.00019489329756055637,
      "loss": 23.0,
      "step": 1729
    },
    {
      "epoch": 1.0384153661464586,
      "grad_norm": 0.0007463322835974395,
      "learning_rate": 0.0001948873361405411,
      "loss": 23.0,
      "step": 1730
    },
    {
      "epoch": 1.039015606242497,
      "grad_norm": 0.001030348939821124,
      "learning_rate": 0.0001948813713342409,
      "loss": 23.0,
      "step": 1731
    },
    {
      "epoch": 1.0396158463385354,
      "grad_norm": 0.0018405051669105887,
      "learning_rate": 0.00019487540314186864,
      "loss": 23.0,
      "step": 1732
    },
    {
      "epoch": 1.0402160864345738,
      "grad_norm": 0.0007769169751554728,
      "learning_rate": 0.00019486943156363725,
      "loss": 23.0,
      "step": 1733
    },
    {
      "epoch": 1.0408163265306123,
      "grad_norm": 0.0008031317847780883,
      "learning_rate": 0.00019486345659975988,
      "loss": 23.0,
      "step": 1734
    },
    {
      "epoch": 1.0414165666266506,
      "grad_norm": 0.0003373367653694004,
      "learning_rate": 0.0001948574782504498,
      "loss": 23.0,
      "step": 1735
    },
    {
      "epoch": 1.0420168067226891,
      "grad_norm": 0.001183771644718945,
      "learning_rate": 0.00019485149651592032,
      "loss": 23.0,
      "step": 1736
    },
    {
      "epoch": 1.0426170468187275,
      "grad_norm": 0.00049160688649863,
      "learning_rate": 0.00019484551139638487,
      "loss": 23.0,
      "step": 1737
    },
    {
      "epoch": 1.043217286914766,
      "grad_norm": 0.0005904583376832306,
      "learning_rate": 0.00019483952289205714,
      "loss": 23.0,
      "step": 1738
    },
    {
      "epoch": 1.0438175270108043,
      "grad_norm": 0.0012503239559009671,
      "learning_rate": 0.0001948335310031508,
      "loss": 23.0,
      "step": 1739
    },
    {
      "epoch": 1.0444177671068426,
      "grad_norm": 0.001702033099718392,
      "learning_rate": 0.00019482753572987963,
      "loss": 23.0,
      "step": 1740
    },
    {
      "epoch": 1.0450180072028812,
      "grad_norm": 0.0005251219845376909,
      "learning_rate": 0.00019482153707245766,
      "loss": 23.0,
      "step": 1741
    },
    {
      "epoch": 1.0456182472989195,
      "grad_norm": 0.0004746287304442376,
      "learning_rate": 0.00019481553503109892,
      "loss": 23.0,
      "step": 1742
    },
    {
      "epoch": 1.046218487394958,
      "grad_norm": 0.0006970618851482868,
      "learning_rate": 0.00019480952960601765,
      "loss": 23.0,
      "step": 1743
    },
    {
      "epoch": 1.0468187274909964,
      "grad_norm": 0.0005676356377080083,
      "learning_rate": 0.00019480352079742817,
      "loss": 23.0,
      "step": 1744
    },
    {
      "epoch": 1.047418967587035,
      "grad_norm": 0.000801927933935076,
      "learning_rate": 0.00019479750860554486,
      "loss": 23.0,
      "step": 1745
    },
    {
      "epoch": 1.0480192076830732,
      "grad_norm": 0.0004625970032066107,
      "learning_rate": 0.00019479149303058233,
      "loss": 23.0,
      "step": 1746
    },
    {
      "epoch": 1.0486194477791115,
      "grad_norm": 0.0007363795302808285,
      "learning_rate": 0.00019478547407275525,
      "loss": 23.0,
      "step": 1747
    },
    {
      "epoch": 1.04921968787515,
      "grad_norm": 0.0011515008518472314,
      "learning_rate": 0.00019477945173227844,
      "loss": 23.0,
      "step": 1748
    },
    {
      "epoch": 1.0498199279711884,
      "grad_norm": 0.0005580635624937713,
      "learning_rate": 0.00019477342600936675,
      "loss": 23.0,
      "step": 1749
    },
    {
      "epoch": 1.050420168067227,
      "grad_norm": 0.0010533389868214726,
      "learning_rate": 0.00019476739690423532,
      "loss": 23.0,
      "step": 1750
    },
    {
      "epoch": 1.0510204081632653,
      "grad_norm": 0.0008852750179357827,
      "learning_rate": 0.00019476136441709923,
      "loss": 23.0,
      "step": 1751
    },
    {
      "epoch": 1.0516206482593038,
      "grad_norm": 0.0007593745249323547,
      "learning_rate": 0.0001947553285481738,
      "loss": 23.0,
      "step": 1752
    },
    {
      "epoch": 1.052220888355342,
      "grad_norm": 0.00030658565810881555,
      "learning_rate": 0.00019474928929767444,
      "loss": 23.0,
      "step": 1753
    },
    {
      "epoch": 1.0528211284513807,
      "grad_norm": 0.0006529420497827232,
      "learning_rate": 0.0001947432466658167,
      "loss": 23.0,
      "step": 1754
    },
    {
      "epoch": 1.053421368547419,
      "grad_norm": 0.0011519229738041759,
      "learning_rate": 0.00019473720065281614,
      "loss": 23.0,
      "step": 1755
    },
    {
      "epoch": 1.0540216086434573,
      "grad_norm": 0.0003251166781410575,
      "learning_rate": 0.00019473115125888862,
      "loss": 23.0,
      "step": 1756
    },
    {
      "epoch": 1.0546218487394958,
      "grad_norm": 0.00099608872551471,
      "learning_rate": 0.00019472509848424997,
      "loss": 23.0,
      "step": 1757
    },
    {
      "epoch": 1.0552220888355341,
      "grad_norm": 0.0007561391685158014,
      "learning_rate": 0.00019471904232911622,
      "loss": 23.0,
      "step": 1758
    },
    {
      "epoch": 1.0558223289315727,
      "grad_norm": 0.0008905903669074178,
      "learning_rate": 0.00019471298279370353,
      "loss": 23.0,
      "step": 1759
    },
    {
      "epoch": 1.056422569027611,
      "grad_norm": 0.0009993897983804345,
      "learning_rate": 0.0001947069198782281,
      "loss": 23.0,
      "step": 1760
    },
    {
      "epoch": 1.0570228091236495,
      "grad_norm": 0.0006743833655491471,
      "learning_rate": 0.00019470085358290628,
      "loss": 23.0,
      "step": 1761
    },
    {
      "epoch": 1.0576230492196879,
      "grad_norm": 0.0009488465148024261,
      "learning_rate": 0.0001946947839079546,
      "loss": 23.0,
      "step": 1762
    },
    {
      "epoch": 1.0582232893157264,
      "grad_norm": 0.000561807188205421,
      "learning_rate": 0.0001946887108535897,
      "loss": 23.0,
      "step": 1763
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 0.0009906566701829433,
      "learning_rate": 0.0001946826344200283,
      "loss": 23.0,
      "step": 1764
    },
    {
      "epoch": 1.059423769507803,
      "grad_norm": 0.0008896515937522054,
      "learning_rate": 0.00019467655460748718,
      "loss": 23.0,
      "step": 1765
    },
    {
      "epoch": 1.0600240096038416,
      "grad_norm": 0.0005601485027000308,
      "learning_rate": 0.00019467047141618336,
      "loss": 23.0,
      "step": 1766
    },
    {
      "epoch": 1.06062424969988,
      "grad_norm": 0.0007029306143522263,
      "learning_rate": 0.00019466438484633396,
      "loss": 23.0,
      "step": 1767
    },
    {
      "epoch": 1.0612244897959184,
      "grad_norm": 0.0009415207314305007,
      "learning_rate": 0.00019465829489815614,
      "loss": 23.0,
      "step": 1768
    },
    {
      "epoch": 1.0618247298919568,
      "grad_norm": 0.0005582883604802191,
      "learning_rate": 0.0001946522015718673,
      "loss": 23.0,
      "step": 1769
    },
    {
      "epoch": 1.0624249699879953,
      "grad_norm": 0.0004074526659678668,
      "learning_rate": 0.00019464610486768485,
      "loss": 23.0,
      "step": 1770
    },
    {
      "epoch": 1.0630252100840336,
      "grad_norm": 0.0006803267169743776,
      "learning_rate": 0.00019464000478582638,
      "loss": 23.0,
      "step": 1771
    },
    {
      "epoch": 1.063625450180072,
      "grad_norm": 0.0007196204387582839,
      "learning_rate": 0.0001946339013265096,
      "loss": 23.0,
      "step": 1772
    },
    {
      "epoch": 1.0642256902761105,
      "grad_norm": 0.00023397163022309542,
      "learning_rate": 0.00019462779448995227,
      "loss": 23.0,
      "step": 1773
    },
    {
      "epoch": 1.0648259303721488,
      "grad_norm": 0.0004155267379246652,
      "learning_rate": 0.00019462168427637237,
      "loss": 23.0,
      "step": 1774
    },
    {
      "epoch": 1.0654261704681873,
      "grad_norm": 0.0004314890829846263,
      "learning_rate": 0.00019461557068598797,
      "loss": 23.0,
      "step": 1775
    },
    {
      "epoch": 1.0660264105642256,
      "grad_norm": 0.0005188965005800128,
      "learning_rate": 0.00019460945371901724,
      "loss": 23.0,
      "step": 1776
    },
    {
      "epoch": 1.0666266506602642,
      "grad_norm": 0.00037539261393249035,
      "learning_rate": 0.00019460333337567844,
      "loss": 23.0,
      "step": 1777
    },
    {
      "epoch": 1.0672268907563025,
      "grad_norm": 0.001011545886285603,
      "learning_rate": 0.00019459720965619005,
      "loss": 23.0,
      "step": 1778
    },
    {
      "epoch": 1.0678271308523408,
      "grad_norm": 0.0004321189771872014,
      "learning_rate": 0.0001945910825607706,
      "loss": 23.0,
      "step": 1779
    },
    {
      "epoch": 1.0684273709483794,
      "grad_norm": 0.0005004126578569412,
      "learning_rate": 0.0001945849520896387,
      "loss": 23.0,
      "step": 1780
    },
    {
      "epoch": 1.0690276110444177,
      "grad_norm": 0.0024395366199314594,
      "learning_rate": 0.00019457881824301313,
      "loss": 23.0,
      "step": 1781
    },
    {
      "epoch": 1.0696278511404562,
      "grad_norm": 0.0013693496584892273,
      "learning_rate": 0.00019457268102111288,
      "loss": 23.0,
      "step": 1782
    },
    {
      "epoch": 1.0702280912364945,
      "grad_norm": 0.0010166800348088145,
      "learning_rate": 0.00019456654042415686,
      "loss": 23.0,
      "step": 1783
    },
    {
      "epoch": 1.070828331332533,
      "grad_norm": 0.0006024075555615127,
      "learning_rate": 0.00019456039645236433,
      "loss": 23.0,
      "step": 1784
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 0.0007192663033492863,
      "learning_rate": 0.00019455424910595445,
      "loss": 23.0,
      "step": 1785
    },
    {
      "epoch": 1.07202881152461,
      "grad_norm": 0.0006474677356891334,
      "learning_rate": 0.00019454809838514664,
      "loss": 23.0,
      "step": 1786
    },
    {
      "epoch": 1.0726290516206483,
      "grad_norm": 0.0006114196730777621,
      "learning_rate": 0.00019454194429016042,
      "loss": 23.0,
      "step": 1787
    },
    {
      "epoch": 1.0732292917166866,
      "grad_norm": 0.0007615411304868758,
      "learning_rate": 0.00019453578682121538,
      "loss": 23.0,
      "step": 1788
    },
    {
      "epoch": 1.0738295318127251,
      "grad_norm": 0.0015807717572897673,
      "learning_rate": 0.00019452962597853133,
      "loss": 23.0,
      "step": 1789
    },
    {
      "epoch": 1.0744297719087634,
      "grad_norm": 0.000809942779596895,
      "learning_rate": 0.00019452346176232802,
      "loss": 23.0,
      "step": 1790
    },
    {
      "epoch": 1.075030012004802,
      "grad_norm": 0.0013856550212949514,
      "learning_rate": 0.00019451729417282557,
      "loss": 23.0,
      "step": 1791
    },
    {
      "epoch": 1.0756302521008403,
      "grad_norm": 0.0016725269379094243,
      "learning_rate": 0.00019451112321024398,
      "loss": 23.0,
      "step": 1792
    },
    {
      "epoch": 1.0762304921968788,
      "grad_norm": 0.0009614219306968153,
      "learning_rate": 0.0001945049488748035,
      "loss": 23.0,
      "step": 1793
    },
    {
      "epoch": 1.0768307322929171,
      "grad_norm": 0.0008319414919242263,
      "learning_rate": 0.00019449877116672453,
      "loss": 23.0,
      "step": 1794
    },
    {
      "epoch": 1.0774309723889557,
      "grad_norm": 0.0004205653676763177,
      "learning_rate": 0.00019449259008622746,
      "loss": 23.0,
      "step": 1795
    },
    {
      "epoch": 1.078031212484994,
      "grad_norm": 0.0007004413055256009,
      "learning_rate": 0.00019448640563353293,
      "loss": 23.0,
      "step": 1796
    },
    {
      "epoch": 1.0786314525810323,
      "grad_norm": 0.001981280744075775,
      "learning_rate": 0.00019448021780886165,
      "loss": 23.0,
      "step": 1797
    },
    {
      "epoch": 1.0792316926770709,
      "grad_norm": 0.0008109554764814675,
      "learning_rate": 0.0001944740266124344,
      "loss": 23.0,
      "step": 1798
    },
    {
      "epoch": 1.0798319327731092,
      "grad_norm": 0.0008627360220998526,
      "learning_rate": 0.00019446783204447217,
      "loss": 23.0,
      "step": 1799
    },
    {
      "epoch": 1.0804321728691477,
      "grad_norm": 0.00029498786898329854,
      "learning_rate": 0.00019446163410519603,
      "loss": 23.0,
      "step": 1800
    },
    {
      "epoch": 1.081032412965186,
      "grad_norm": 0.0010635429061949253,
      "learning_rate": 0.00019445543279482714,
      "loss": 23.0,
      "step": 1801
    },
    {
      "epoch": 1.0816326530612246,
      "grad_norm": 0.0012698029167950153,
      "learning_rate": 0.00019444922811358683,
      "loss": 23.0,
      "step": 1802
    },
    {
      "epoch": 1.082232893157263,
      "grad_norm": 0.0006329349125735462,
      "learning_rate": 0.0001944430200616965,
      "loss": 23.0,
      "step": 1803
    },
    {
      "epoch": 1.0828331332533012,
      "grad_norm": 0.0003994033904746175,
      "learning_rate": 0.0001944368086393777,
      "loss": 23.0,
      "step": 1804
    },
    {
      "epoch": 1.0834333733493398,
      "grad_norm": 0.0010559451766312122,
      "learning_rate": 0.00019443059384685217,
      "loss": 23.0,
      "step": 1805
    },
    {
      "epoch": 1.084033613445378,
      "grad_norm": 0.0007147497381083667,
      "learning_rate": 0.0001944243756843416,
      "loss": 23.0,
      "step": 1806
    },
    {
      "epoch": 1.0846338535414166,
      "grad_norm": 0.0006631044088862836,
      "learning_rate": 0.00019441815415206798,
      "loss": 23.0,
      "step": 1807
    },
    {
      "epoch": 1.085234093637455,
      "grad_norm": 0.0013852432603016496,
      "learning_rate": 0.00019441192925025334,
      "loss": 23.0,
      "step": 1808
    },
    {
      "epoch": 1.0858343337334935,
      "grad_norm": 0.0013341652229428291,
      "learning_rate": 0.00019440570097911975,
      "loss": 23.0,
      "step": 1809
    },
    {
      "epoch": 1.0864345738295318,
      "grad_norm": 0.000731373846065253,
      "learning_rate": 0.00019439946933888958,
      "loss": 23.0,
      "step": 1810
    },
    {
      "epoch": 1.0870348139255703,
      "grad_norm": 0.0006150943809188902,
      "learning_rate": 0.00019439323432978513,
      "loss": 23.0,
      "step": 1811
    },
    {
      "epoch": 1.0876350540216086,
      "grad_norm": 0.0007697598193772137,
      "learning_rate": 0.00019438699595202897,
      "loss": 23.0,
      "step": 1812
    },
    {
      "epoch": 1.088235294117647,
      "grad_norm": 0.000452876411145553,
      "learning_rate": 0.00019438075420584374,
      "loss": 23.0,
      "step": 1813
    },
    {
      "epoch": 1.0888355342136855,
      "grad_norm": 0.0010431616101413965,
      "learning_rate": 0.00019437450909145213,
      "loss": 23.0,
      "step": 1814
    },
    {
      "epoch": 1.0894357743097238,
      "grad_norm": 0.00121380016207695,
      "learning_rate": 0.0001943682606090771,
      "loss": 23.0,
      "step": 1815
    },
    {
      "epoch": 1.0900360144057624,
      "grad_norm": 0.0017278955783694983,
      "learning_rate": 0.00019436200875894155,
      "loss": 23.0,
      "step": 1816
    },
    {
      "epoch": 1.0906362545018007,
      "grad_norm": 0.0006101586041040719,
      "learning_rate": 0.00019435575354126864,
      "loss": 23.0,
      "step": 1817
    },
    {
      "epoch": 1.0912364945978392,
      "grad_norm": 0.000790750200394541,
      "learning_rate": 0.0001943494949562816,
      "loss": 23.0,
      "step": 1818
    },
    {
      "epoch": 1.0918367346938775,
      "grad_norm": 0.000308141577988863,
      "learning_rate": 0.00019434323300420377,
      "loss": 23.0,
      "step": 1819
    },
    {
      "epoch": 1.092436974789916,
      "grad_norm": 0.0013721545692533255,
      "learning_rate": 0.00019433696768525867,
      "loss": 23.0,
      "step": 1820
    },
    {
      "epoch": 1.0930372148859544,
      "grad_norm": 0.0009004747844301164,
      "learning_rate": 0.00019433069899966982,
      "loss": 23.0,
      "step": 1821
    },
    {
      "epoch": 1.0936374549819927,
      "grad_norm": 0.0007536576013080776,
      "learning_rate": 0.000194324426947661,
      "loss": 23.0,
      "step": 1822
    },
    {
      "epoch": 1.0942376950780313,
      "grad_norm": 0.0009083595359697938,
      "learning_rate": 0.000194318151529456,
      "loss": 23.0,
      "step": 1823
    },
    {
      "epoch": 1.0948379351740696,
      "grad_norm": 0.0006454298854805529,
      "learning_rate": 0.0001943118727452788,
      "loss": 23.0,
      "step": 1824
    },
    {
      "epoch": 1.0954381752701081,
      "grad_norm": 0.000571611279156059,
      "learning_rate": 0.00019430559059535344,
      "loss": 23.0,
      "step": 1825
    },
    {
      "epoch": 1.0960384153661464,
      "grad_norm": 0.000725729507394135,
      "learning_rate": 0.00019429930507990415,
      "loss": 23.0,
      "step": 1826
    },
    {
      "epoch": 1.096638655462185,
      "grad_norm": 0.0006706914282403886,
      "learning_rate": 0.00019429301619915522,
      "loss": 23.0,
      "step": 1827
    },
    {
      "epoch": 1.0972388955582233,
      "grad_norm": 0.0010118356440216303,
      "learning_rate": 0.00019428672395333112,
      "loss": 23.0,
      "step": 1828
    },
    {
      "epoch": 1.0978391356542616,
      "grad_norm": 0.0009120289469137788,
      "learning_rate": 0.00019428042834265634,
      "loss": 23.0,
      "step": 1829
    },
    {
      "epoch": 1.0984393757503002,
      "grad_norm": 0.0008227874641306698,
      "learning_rate": 0.00019427412936735562,
      "loss": 23.0,
      "step": 1830
    },
    {
      "epoch": 1.0990396158463385,
      "grad_norm": 0.00032032793387770653,
      "learning_rate": 0.00019426782702765366,
      "loss": 23.0,
      "step": 1831
    },
    {
      "epoch": 1.099639855942377,
      "grad_norm": 0.0006032648961991072,
      "learning_rate": 0.0001942615213237755,
      "loss": 23.0,
      "step": 1832
    },
    {
      "epoch": 1.1002400960384153,
      "grad_norm": 0.0007296554394997656,
      "learning_rate": 0.00019425521225594608,
      "loss": 23.0,
      "step": 1833
    },
    {
      "epoch": 1.1008403361344539,
      "grad_norm": 0.000931189744733274,
      "learning_rate": 0.00019424889982439062,
      "loss": 23.0,
      "step": 1834
    },
    {
      "epoch": 1.1014405762304922,
      "grad_norm": 0.001093988073989749,
      "learning_rate": 0.00019424258402933435,
      "loss": 23.0,
      "step": 1835
    },
    {
      "epoch": 1.1020408163265305,
      "grad_norm": 0.0012053363025188446,
      "learning_rate": 0.00019423626487100268,
      "loss": 23.0,
      "step": 1836
    },
    {
      "epoch": 1.102641056422569,
      "grad_norm": 0.0007283682352863252,
      "learning_rate": 0.00019422994234962113,
      "loss": 23.0,
      "step": 1837
    },
    {
      "epoch": 1.1032412965186074,
      "grad_norm": 0.0008196471026167274,
      "learning_rate": 0.00019422361646541532,
      "loss": 23.0,
      "step": 1838
    },
    {
      "epoch": 1.103841536614646,
      "grad_norm": 0.000663764018099755,
      "learning_rate": 0.00019421728721861098,
      "loss": 23.0,
      "step": 1839
    },
    {
      "epoch": 1.1044417767106842,
      "grad_norm": 0.00034105373197235167,
      "learning_rate": 0.00019421095460943406,
      "loss": 23.0,
      "step": 1840
    },
    {
      "epoch": 1.1050420168067228,
      "grad_norm": 0.00047400983748957515,
      "learning_rate": 0.00019420461863811048,
      "loss": 23.0,
      "step": 1841
    },
    {
      "epoch": 1.105642256902761,
      "grad_norm": 0.00045720263733528554,
      "learning_rate": 0.0001941982793048664,
      "loss": 23.0,
      "step": 1842
    },
    {
      "epoch": 1.1062424969987996,
      "grad_norm": 0.001074103289283812,
      "learning_rate": 0.00019419193660992801,
      "loss": 23.0,
      "step": 1843
    },
    {
      "epoch": 1.106842737094838,
      "grad_norm": 0.00114881107583642,
      "learning_rate": 0.00019418559055352176,
      "loss": 23.0,
      "step": 1844
    },
    {
      "epoch": 1.1074429771908763,
      "grad_norm": 0.00042918327380903065,
      "learning_rate": 0.000194179241135874,
      "loss": 23.0,
      "step": 1845
    },
    {
      "epoch": 1.1080432172869148,
      "grad_norm": 0.0016334144165739417,
      "learning_rate": 0.00019417288835721136,
      "loss": 23.0,
      "step": 1846
    },
    {
      "epoch": 1.1086434573829531,
      "grad_norm": 0.0006906944909133017,
      "learning_rate": 0.00019416653221776065,
      "loss": 23.0,
      "step": 1847
    },
    {
      "epoch": 1.1092436974789917,
      "grad_norm": 0.0012676967307925224,
      "learning_rate": 0.00019416017271774858,
      "loss": 23.0,
      "step": 1848
    },
    {
      "epoch": 1.10984393757503,
      "grad_norm": 0.0007337264833040535,
      "learning_rate": 0.00019415380985740214,
      "loss": 23.0,
      "step": 1849
    },
    {
      "epoch": 1.1104441776710685,
      "grad_norm": 0.0017202390590682626,
      "learning_rate": 0.00019414744363694845,
      "loss": 23.0,
      "step": 1850
    },
    {
      "epoch": 1.1110444177671068,
      "grad_norm": 0.0005216122954152524,
      "learning_rate": 0.00019414107405661468,
      "loss": 23.0,
      "step": 1851
    },
    {
      "epoch": 1.1116446578631454,
      "grad_norm": 0.0006443992024287581,
      "learning_rate": 0.00019413470111662813,
      "loss": 23.0,
      "step": 1852
    },
    {
      "epoch": 1.1122448979591837,
      "grad_norm": 0.0009258884820155799,
      "learning_rate": 0.00019412832481721624,
      "loss": 23.0,
      "step": 1853
    },
    {
      "epoch": 1.112845138055222,
      "grad_norm": 0.0015777413500472903,
      "learning_rate": 0.00019412194515860653,
      "loss": 23.0,
      "step": 1854
    },
    {
      "epoch": 1.1134453781512605,
      "grad_norm": 0.0007551502203568816,
      "learning_rate": 0.00019411556214102676,
      "loss": 23.0,
      "step": 1855
    },
    {
      "epoch": 1.1140456182472989,
      "grad_norm": 0.0004909979179501534,
      "learning_rate": 0.0001941091757647046,
      "loss": 23.0,
      "step": 1856
    },
    {
      "epoch": 1.1146458583433374,
      "grad_norm": 0.0002796545159071684,
      "learning_rate": 0.00019410278602986813,
      "loss": 23.0,
      "step": 1857
    },
    {
      "epoch": 1.1152460984393757,
      "grad_norm": 0.0010796081041917205,
      "learning_rate": 0.00019409639293674525,
      "loss": 23.0,
      "step": 1858
    },
    {
      "epoch": 1.1158463385354143,
      "grad_norm": 0.00046648201532661915,
      "learning_rate": 0.0001940899964855641,
      "loss": 23.0,
      "step": 1859
    },
    {
      "epoch": 1.1164465786314526,
      "grad_norm": 0.000706287391949445,
      "learning_rate": 0.00019408359667655306,
      "loss": 23.0,
      "step": 1860
    },
    {
      "epoch": 1.117046818727491,
      "grad_norm": 0.0010415187571197748,
      "learning_rate": 0.00019407719350994046,
      "loss": 23.0,
      "step": 1861
    },
    {
      "epoch": 1.1176470588235294,
      "grad_norm": 0.0007331138476729393,
      "learning_rate": 0.0001940707869859548,
      "loss": 23.0,
      "step": 1862
    },
    {
      "epoch": 1.1182472989195678,
      "grad_norm": 0.00044027238618582487,
      "learning_rate": 0.00019406437710482475,
      "loss": 23.0,
      "step": 1863
    },
    {
      "epoch": 1.1188475390156063,
      "grad_norm": 0.0006217750487849116,
      "learning_rate": 0.00019405796386677904,
      "loss": 23.0,
      "step": 1864
    },
    {
      "epoch": 1.1194477791116446,
      "grad_norm": 0.0009194968733936548,
      "learning_rate": 0.00019405154727204652,
      "loss": 23.0,
      "step": 1865
    },
    {
      "epoch": 1.1200480192076832,
      "grad_norm": 0.000699158466886729,
      "learning_rate": 0.00019404512732085625,
      "loss": 23.0,
      "step": 1866
    },
    {
      "epoch": 1.1206482593037215,
      "grad_norm": 0.0007176969083957374,
      "learning_rate": 0.00019403870401343726,
      "loss": 23.0,
      "step": 1867
    },
    {
      "epoch": 1.1212484993997598,
      "grad_norm": 0.0007284717867150903,
      "learning_rate": 0.00019403227735001886,
      "loss": 23.0,
      "step": 1868
    },
    {
      "epoch": 1.1218487394957983,
      "grad_norm": 0.0010808160295709968,
      "learning_rate": 0.00019402584733083032,
      "loss": 23.0,
      "step": 1869
    },
    {
      "epoch": 1.1224489795918366,
      "grad_norm": 0.0016216632211580873,
      "learning_rate": 0.00019401941395610117,
      "loss": 23.0,
      "step": 1870
    },
    {
      "epoch": 1.1230492196878752,
      "grad_norm": 0.0007159097003750503,
      "learning_rate": 0.00019401297722606098,
      "loss": 23.0,
      "step": 1871
    },
    {
      "epoch": 1.1236494597839135,
      "grad_norm": 0.0017748618265613914,
      "learning_rate": 0.00019400653714093946,
      "loss": 23.0,
      "step": 1872
    },
    {
      "epoch": 1.124249699879952,
      "grad_norm": 0.0015833675861358643,
      "learning_rate": 0.00019400009370096645,
      "loss": 23.0,
      "step": 1873
    },
    {
      "epoch": 1.1248499399759904,
      "grad_norm": 0.00064207857940346,
      "learning_rate": 0.0001939936469063719,
      "loss": 23.0,
      "step": 1874
    },
    {
      "epoch": 1.125450180072029,
      "grad_norm": 0.0004406536172609776,
      "learning_rate": 0.00019398719675738587,
      "loss": 23.0,
      "step": 1875
    },
    {
      "epoch": 1.1260504201680672,
      "grad_norm": 0.00028265893342904747,
      "learning_rate": 0.00019398074325423855,
      "loss": 23.0,
      "step": 1876
    },
    {
      "epoch": 1.1266506602641058,
      "grad_norm": 0.003184243803843856,
      "learning_rate": 0.00019397428639716022,
      "loss": 23.0,
      "step": 1877
    },
    {
      "epoch": 1.127250900360144,
      "grad_norm": 0.0008285728981718421,
      "learning_rate": 0.00019396782618638137,
      "loss": 23.0,
      "step": 1878
    },
    {
      "epoch": 1.1278511404561824,
      "grad_norm": 0.001333149615675211,
      "learning_rate": 0.00019396136262213253,
      "loss": 23.0,
      "step": 1879
    },
    {
      "epoch": 1.128451380552221,
      "grad_norm": 0.0005505660083144903,
      "learning_rate": 0.00019395489570464433,
      "loss": 23.0,
      "step": 1880
    },
    {
      "epoch": 1.1290516206482593,
      "grad_norm": 0.0003775360237341374,
      "learning_rate": 0.0001939484254341476,
      "loss": 23.0,
      "step": 1881
    },
    {
      "epoch": 1.1296518607442978,
      "grad_norm": 0.001226317137479782,
      "learning_rate": 0.0001939419518108732,
      "loss": 23.0,
      "step": 1882
    },
    {
      "epoch": 1.1302521008403361,
      "grad_norm": 0.0006789937033317983,
      "learning_rate": 0.00019393547483505222,
      "loss": 23.0,
      "step": 1883
    },
    {
      "epoch": 1.1308523409363747,
      "grad_norm": 0.001176654826849699,
      "learning_rate": 0.00019392899450691576,
      "loss": 23.0,
      "step": 1884
    },
    {
      "epoch": 1.131452581032413,
      "grad_norm": 0.0007635357324033976,
      "learning_rate": 0.00019392251082669512,
      "loss": 23.0,
      "step": 1885
    },
    {
      "epoch": 1.1320528211284513,
      "grad_norm": 0.001103962305933237,
      "learning_rate": 0.00019391602379462163,
      "loss": 23.0,
      "step": 1886
    },
    {
      "epoch": 1.1326530612244898,
      "grad_norm": 0.0004466601531021297,
      "learning_rate": 0.00019390953341092687,
      "loss": 23.0,
      "step": 1887
    },
    {
      "epoch": 1.1332533013205282,
      "grad_norm": 0.0006208834820427,
      "learning_rate": 0.0001939030396758424,
      "loss": 23.0,
      "step": 1888
    },
    {
      "epoch": 1.1338535414165667,
      "grad_norm": 0.0007072006119415164,
      "learning_rate": 0.0001938965425896,
      "loss": 23.0,
      "step": 1889
    },
    {
      "epoch": 1.134453781512605,
      "grad_norm": 0.0006387155153788626,
      "learning_rate": 0.00019389004215243153,
      "loss": 23.0,
      "step": 1890
    },
    {
      "epoch": 1.1350540216086435,
      "grad_norm": 0.0009332761983387172,
      "learning_rate": 0.00019388353836456896,
      "loss": 23.0,
      "step": 1891
    },
    {
      "epoch": 1.1356542617046819,
      "grad_norm": 0.0007608407759107649,
      "learning_rate": 0.00019387703122624442,
      "loss": 23.0,
      "step": 1892
    },
    {
      "epoch": 1.1362545018007202,
      "grad_norm": 0.00044079619692638516,
      "learning_rate": 0.0001938705207376901,
      "loss": 23.0,
      "step": 1893
    },
    {
      "epoch": 1.1368547418967587,
      "grad_norm": 0.0009360851254314184,
      "learning_rate": 0.00019386400689913838,
      "loss": 23.0,
      "step": 1894
    },
    {
      "epoch": 1.137454981992797,
      "grad_norm": 0.0009462125017307699,
      "learning_rate": 0.00019385748971082167,
      "loss": 23.0,
      "step": 1895
    },
    {
      "epoch": 1.1380552220888356,
      "grad_norm": 0.0005419541266746819,
      "learning_rate": 0.00019385096917297258,
      "loss": 23.0,
      "step": 1896
    },
    {
      "epoch": 1.138655462184874,
      "grad_norm": 0.0010163012193515897,
      "learning_rate": 0.00019384444528582385,
      "loss": 23.0,
      "step": 1897
    },
    {
      "epoch": 1.1392557022809124,
      "grad_norm": 0.0006951330578885972,
      "learning_rate": 0.00019383791804960824,
      "loss": 23.0,
      "step": 1898
    },
    {
      "epoch": 1.1398559423769508,
      "grad_norm": 0.0006850925274193287,
      "learning_rate": 0.00019383138746455872,
      "loss": 23.0,
      "step": 1899
    },
    {
      "epoch": 1.140456182472989,
      "grad_norm": 0.0007126698037609458,
      "learning_rate": 0.00019382485353090832,
      "loss": 23.0,
      "step": 1900
    },
    {
      "epoch": 1.1410564225690276,
      "grad_norm": 0.0005496939411386847,
      "learning_rate": 0.00019381831624889028,
      "loss": 23.0,
      "step": 1901
    },
    {
      "epoch": 1.141656662665066,
      "grad_norm": 0.0005483221611939371,
      "learning_rate": 0.00019381177561873786,
      "loss": 23.0,
      "step": 1902
    },
    {
      "epoch": 1.1422569027611045,
      "grad_norm": 0.000862023385707289,
      "learning_rate": 0.00019380523164068446,
      "loss": 23.0,
      "step": 1903
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.0019896167796105146,
      "learning_rate": 0.00019379868431496365,
      "loss": 23.0,
      "step": 1904
    },
    {
      "epoch": 1.1434573829531813,
      "grad_norm": 0.0011680411407724023,
      "learning_rate": 0.00019379213364180908,
      "loss": 23.0,
      "step": 1905
    },
    {
      "epoch": 1.1440576230492197,
      "grad_norm": 0.0008858213550411165,
      "learning_rate": 0.0001937855796214545,
      "loss": 23.0,
      "step": 1906
    },
    {
      "epoch": 1.1446578631452582,
      "grad_norm": 0.0010865547228604555,
      "learning_rate": 0.00019377902225413385,
      "loss": 23.0,
      "step": 1907
    },
    {
      "epoch": 1.1452581032412965,
      "grad_norm": 0.0010592964245006442,
      "learning_rate": 0.00019377246154008112,
      "loss": 23.0,
      "step": 1908
    },
    {
      "epoch": 1.145858343337335,
      "grad_norm": 0.000688974279910326,
      "learning_rate": 0.00019376589747953046,
      "loss": 23.0,
      "step": 1909
    },
    {
      "epoch": 1.1464585834333734,
      "grad_norm": 0.0005789738497696817,
      "learning_rate": 0.00019375933007271612,
      "loss": 23.0,
      "step": 1910
    },
    {
      "epoch": 1.1470588235294117,
      "grad_norm": 0.0003665656258817762,
      "learning_rate": 0.0001937527593198724,
      "loss": 23.0,
      "step": 1911
    },
    {
      "epoch": 1.1476590636254502,
      "grad_norm": 0.001227127038873732,
      "learning_rate": 0.00019374618522123395,
      "loss": 23.0,
      "step": 1912
    },
    {
      "epoch": 1.1482593037214885,
      "grad_norm": 0.0005941904382780194,
      "learning_rate": 0.00019373960777703525,
      "loss": 23.0,
      "step": 1913
    },
    {
      "epoch": 1.148859543817527,
      "grad_norm": 0.0007471052813343704,
      "learning_rate": 0.0001937330269875111,
      "loss": 23.0,
      "step": 1914
    },
    {
      "epoch": 1.1494597839135654,
      "grad_norm": 0.0004938453203067183,
      "learning_rate": 0.0001937264428528963,
      "loss": 23.0,
      "step": 1915
    },
    {
      "epoch": 1.150060024009604,
      "grad_norm": 0.0009274558979086578,
      "learning_rate": 0.00019371985537342586,
      "loss": 23.0,
      "step": 1916
    },
    {
      "epoch": 1.1506602641056423,
      "grad_norm": 0.0015852872747927904,
      "learning_rate": 0.00019371326454933484,
      "loss": 23.0,
      "step": 1917
    },
    {
      "epoch": 1.1512605042016806,
      "grad_norm": 0.0007189033203758299,
      "learning_rate": 0.00019370667038085847,
      "loss": 23.0,
      "step": 1918
    },
    {
      "epoch": 1.1518607442977191,
      "grad_norm": 0.0006547678494825959,
      "learning_rate": 0.0001937000728682321,
      "loss": 23.0,
      "step": 1919
    },
    {
      "epoch": 1.1524609843937574,
      "grad_norm": 0.0006517032161355019,
      "learning_rate": 0.00019369347201169114,
      "loss": 23.0,
      "step": 1920
    },
    {
      "epoch": 1.153061224489796,
      "grad_norm": 0.0009650959400460124,
      "learning_rate": 0.00019368686781147116,
      "loss": 23.0,
      "step": 1921
    },
    {
      "epoch": 1.1536614645858343,
      "grad_norm": 0.0007866573869250715,
      "learning_rate": 0.00019368026026780788,
      "loss": 23.0,
      "step": 1922
    },
    {
      "epoch": 1.1542617046818728,
      "grad_norm": 0.0005117384134791791,
      "learning_rate": 0.0001936736493809371,
      "loss": 23.0,
      "step": 1923
    },
    {
      "epoch": 1.1548619447779112,
      "grad_norm": 0.0005935249500907958,
      "learning_rate": 0.0001936670351510947,
      "loss": 23.0,
      "step": 1924
    },
    {
      "epoch": 1.1554621848739495,
      "grad_norm": 0.00046214647591114044,
      "learning_rate": 0.00019366041757851675,
      "loss": 23.0,
      "step": 1925
    },
    {
      "epoch": 1.156062424969988,
      "grad_norm": 0.0011201126035302877,
      "learning_rate": 0.00019365379666343944,
      "loss": 23.0,
      "step": 1926
    },
    {
      "epoch": 1.1566626650660263,
      "grad_norm": 0.000578420062083751,
      "learning_rate": 0.00019364717240609905,
      "loss": 23.0,
      "step": 1927
    },
    {
      "epoch": 1.1572629051620649,
      "grad_norm": 0.0009162966744042933,
      "learning_rate": 0.00019364054480673195,
      "loss": 23.0,
      "step": 1928
    },
    {
      "epoch": 1.1578631452581032,
      "grad_norm": 0.0007997804204933345,
      "learning_rate": 0.00019363391386557465,
      "loss": 23.0,
      "step": 1929
    },
    {
      "epoch": 1.1584633853541417,
      "grad_norm": 0.0005844663246534765,
      "learning_rate": 0.00019362727958286384,
      "loss": 23.0,
      "step": 1930
    },
    {
      "epoch": 1.15906362545018,
      "grad_norm": 0.0007434054859913886,
      "learning_rate": 0.00019362064195883627,
      "loss": 23.0,
      "step": 1931
    },
    {
      "epoch": 1.1596638655462184,
      "grad_norm": 0.001075836829841137,
      "learning_rate": 0.0001936140009937288,
      "loss": 23.0,
      "step": 1932
    },
    {
      "epoch": 1.160264105642257,
      "grad_norm": 0.0012545333011075854,
      "learning_rate": 0.00019360735668777843,
      "loss": 23.0,
      "step": 1933
    },
    {
      "epoch": 1.1608643457382952,
      "grad_norm": 0.000714300200343132,
      "learning_rate": 0.00019360070904122232,
      "loss": 23.0,
      "step": 1934
    },
    {
      "epoch": 1.1614645858343338,
      "grad_norm": 0.0012586622033268213,
      "learning_rate": 0.00019359405805429765,
      "loss": 23.0,
      "step": 1935
    },
    {
      "epoch": 1.162064825930372,
      "grad_norm": 0.0007662964635528624,
      "learning_rate": 0.0001935874037272418,
      "loss": 23.0,
      "step": 1936
    },
    {
      "epoch": 1.1626650660264106,
      "grad_norm": 0.0007426751544699073,
      "learning_rate": 0.00019358074606029227,
      "loss": 23.0,
      "step": 1937
    },
    {
      "epoch": 1.163265306122449,
      "grad_norm": 0.0007442353526130319,
      "learning_rate": 0.0001935740850536866,
      "loss": 23.0,
      "step": 1938
    },
    {
      "epoch": 1.1638655462184875,
      "grad_norm": 0.0003253078320994973,
      "learning_rate": 0.00019356742070766254,
      "loss": 23.0,
      "step": 1939
    },
    {
      "epoch": 1.1644657863145258,
      "grad_norm": 0.0004222916322760284,
      "learning_rate": 0.0001935607530224579,
      "loss": 23.0,
      "step": 1940
    },
    {
      "epoch": 1.1650660264105643,
      "grad_norm": 0.0008304398506879807,
      "learning_rate": 0.0001935540819983107,
      "loss": 23.0,
      "step": 1941
    },
    {
      "epoch": 1.1656662665066027,
      "grad_norm": 0.0008976273820735514,
      "learning_rate": 0.00019354740763545894,
      "loss": 23.0,
      "step": 1942
    },
    {
      "epoch": 1.166266506602641,
      "grad_norm": 0.0006581676425412297,
      "learning_rate": 0.00019354072993414082,
      "loss": 23.0,
      "step": 1943
    },
    {
      "epoch": 1.1668667466986795,
      "grad_norm": 0.0009220499778166413,
      "learning_rate": 0.0001935340488945947,
      "loss": 23.0,
      "step": 1944
    },
    {
      "epoch": 1.1674669867947178,
      "grad_norm": 0.0015496306587010622,
      "learning_rate": 0.00019352736451705896,
      "loss": 23.0,
      "step": 1945
    },
    {
      "epoch": 1.1680672268907564,
      "grad_norm": 0.001087090466171503,
      "learning_rate": 0.00019352067680177213,
      "loss": 23.0,
      "step": 1946
    },
    {
      "epoch": 1.1686674669867947,
      "grad_norm": 0.00046333111822605133,
      "learning_rate": 0.00019351398574897293,
      "loss": 23.0,
      "step": 1947
    },
    {
      "epoch": 1.1692677070828332,
      "grad_norm": 0.0005021764663979411,
      "learning_rate": 0.00019350729135890013,
      "loss": 23.0,
      "step": 1948
    },
    {
      "epoch": 1.1698679471788715,
      "grad_norm": 0.0011674378765746951,
      "learning_rate": 0.00019350059363179264,
      "loss": 23.0,
      "step": 1949
    },
    {
      "epoch": 1.1704681872749099,
      "grad_norm": 0.0008699194295331836,
      "learning_rate": 0.00019349389256788943,
      "loss": 23.0,
      "step": 1950
    },
    {
      "epoch": 1.1710684273709484,
      "grad_norm": 0.0007664512377232313,
      "learning_rate": 0.00019348718816742974,
      "loss": 23.0,
      "step": 1951
    },
    {
      "epoch": 1.1716686674669867,
      "grad_norm": 0.0016348707722499967,
      "learning_rate": 0.00019348048043065278,
      "loss": 23.0,
      "step": 1952
    },
    {
      "epoch": 1.1722689075630253,
      "grad_norm": 0.0011596891563385725,
      "learning_rate": 0.0001934737693577979,
      "loss": 23.0,
      "step": 1953
    },
    {
      "epoch": 1.1728691476590636,
      "grad_norm": 0.0011909161694347858,
      "learning_rate": 0.00019346705494910466,
      "loss": 23.0,
      "step": 1954
    },
    {
      "epoch": 1.1734693877551021,
      "grad_norm": 0.0013877099845558405,
      "learning_rate": 0.00019346033720481264,
      "loss": 23.0,
      "step": 1955
    },
    {
      "epoch": 1.1740696278511404,
      "grad_norm": 0.0008768134284764528,
      "learning_rate": 0.0001934536161251616,
      "loss": 23.0,
      "step": 1956
    },
    {
      "epoch": 1.1746698679471788,
      "grad_norm": 0.0006206629332154989,
      "learning_rate": 0.0001934468917103914,
      "loss": 23.0,
      "step": 1957
    },
    {
      "epoch": 1.1752701080432173,
      "grad_norm": 0.0009947314392775297,
      "learning_rate": 0.000193440163960742,
      "loss": 23.0,
      "step": 1958
    },
    {
      "epoch": 1.1758703481392556,
      "grad_norm": 0.0003389040066394955,
      "learning_rate": 0.0001934334328764535,
      "loss": 23.0,
      "step": 1959
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.0008335797465406358,
      "learning_rate": 0.00019342669845776614,
      "loss": 23.0,
      "step": 1960
    },
    {
      "epoch": 1.1770708283313325,
      "grad_norm": 0.0015772458864375949,
      "learning_rate": 0.00019341996070492022,
      "loss": 23.0,
      "step": 1961
    },
    {
      "epoch": 1.177671068427371,
      "grad_norm": 0.0009163774084299803,
      "learning_rate": 0.0001934132196181562,
      "loss": 23.0,
      "step": 1962
    },
    {
      "epoch": 1.1782713085234093,
      "grad_norm": 0.001706429524347186,
      "learning_rate": 0.00019340647519771467,
      "loss": 23.0,
      "step": 1963
    },
    {
      "epoch": 1.1788715486194479,
      "grad_norm": 0.0012178387260064483,
      "learning_rate": 0.0001933997274438363,
      "loss": 23.0,
      "step": 1964
    },
    {
      "epoch": 1.1794717887154862,
      "grad_norm": 0.00025741333956830204,
      "learning_rate": 0.00019339297635676195,
      "loss": 23.0,
      "step": 1965
    },
    {
      "epoch": 1.1800720288115247,
      "grad_norm": 0.002575822174549103,
      "learning_rate": 0.00019338622193673245,
      "loss": 23.0,
      "step": 1966
    },
    {
      "epoch": 1.180672268907563,
      "grad_norm": 0.0012099726591259241,
      "learning_rate": 0.00019337946418398897,
      "loss": 23.0,
      "step": 1967
    },
    {
      "epoch": 1.1812725090036014,
      "grad_norm": 0.000488203571876511,
      "learning_rate": 0.00019337270309877257,
      "loss": 23.0,
      "step": 1968
    },
    {
      "epoch": 1.18187274909964,
      "grad_norm": 0.0011559462873265147,
      "learning_rate": 0.00019336593868132457,
      "loss": 23.0,
      "step": 1969
    },
    {
      "epoch": 1.1824729891956782,
      "grad_norm": 0.001535662915557623,
      "learning_rate": 0.0001933591709318864,
      "loss": 23.0,
      "step": 1970
    },
    {
      "epoch": 1.1830732292917168,
      "grad_norm": 0.0009023676975630224,
      "learning_rate": 0.0001933523998506996,
      "loss": 23.0,
      "step": 1971
    },
    {
      "epoch": 1.183673469387755,
      "grad_norm": 0.0007497956976294518,
      "learning_rate": 0.00019334562543800573,
      "loss": 23.0,
      "step": 1972
    },
    {
      "epoch": 1.1842737094837936,
      "grad_norm": 0.0006753331399522722,
      "learning_rate": 0.00019333884769404661,
      "loss": 23.0,
      "step": 1973
    },
    {
      "epoch": 1.184873949579832,
      "grad_norm": 0.0019581643864512444,
      "learning_rate": 0.00019333206661906416,
      "loss": 23.0,
      "step": 1974
    },
    {
      "epoch": 1.1854741896758703,
      "grad_norm": 0.001169833354651928,
      "learning_rate": 0.00019332528221330028,
      "loss": 23.0,
      "step": 1975
    },
    {
      "epoch": 1.1860744297719088,
      "grad_norm": 0.0015272792661562562,
      "learning_rate": 0.00019331849447699716,
      "loss": 23.0,
      "step": 1976
    },
    {
      "epoch": 1.1866746698679471,
      "grad_norm": 0.0005761996726505458,
      "learning_rate": 0.000193311703410397,
      "loss": 23.0,
      "step": 1977
    },
    {
      "epoch": 1.1872749099639857,
      "grad_norm": 0.0010293567320331931,
      "learning_rate": 0.0001933049090137422,
      "loss": 23.0,
      "step": 1978
    },
    {
      "epoch": 1.187875150060024,
      "grad_norm": 0.0007244606385938823,
      "learning_rate": 0.0001932981112872752,
      "loss": 23.0,
      "step": 1979
    },
    {
      "epoch": 1.1884753901560625,
      "grad_norm": 0.00049425387987867,
      "learning_rate": 0.0001932913102312386,
      "loss": 23.0,
      "step": 1980
    },
    {
      "epoch": 1.1890756302521008,
      "grad_norm": 0.0006668847054243088,
      "learning_rate": 0.00019328450584587506,
      "loss": 23.0,
      "step": 1981
    },
    {
      "epoch": 1.1896758703481392,
      "grad_norm": 0.0004593547491822392,
      "learning_rate": 0.00019327769813142753,
      "loss": 23.0,
      "step": 1982
    },
    {
      "epoch": 1.1902761104441777,
      "grad_norm": 0.0003765747824218124,
      "learning_rate": 0.00019327088708813886,
      "loss": 23.0,
      "step": 1983
    },
    {
      "epoch": 1.190876350540216,
      "grad_norm": 0.0006263687973842025,
      "learning_rate": 0.00019326407271625217,
      "loss": 23.0,
      "step": 1984
    },
    {
      "epoch": 1.1914765906362546,
      "grad_norm": 0.0006472663953900337,
      "learning_rate": 0.0001932572550160106,
      "loss": 23.0,
      "step": 1985
    },
    {
      "epoch": 1.1920768307322929,
      "grad_norm": 0.0017092872876673937,
      "learning_rate": 0.0001932504339876575,
      "loss": 23.0,
      "step": 1986
    },
    {
      "epoch": 1.1926770708283314,
      "grad_norm": 0.000995015725493431,
      "learning_rate": 0.0001932436096314363,
      "loss": 23.0,
      "step": 1987
    },
    {
      "epoch": 1.1932773109243697,
      "grad_norm": 0.0012725737178698182,
      "learning_rate": 0.00019323678194759048,
      "loss": 23.0,
      "step": 1988
    },
    {
      "epoch": 1.193877551020408,
      "grad_norm": 0.0013512310106307268,
      "learning_rate": 0.00019322995093636375,
      "loss": 23.0,
      "step": 1989
    },
    {
      "epoch": 1.1944777911164466,
      "grad_norm": 0.0010033018188551068,
      "learning_rate": 0.0001932231165979999,
      "loss": 23.0,
      "step": 1990
    },
    {
      "epoch": 1.195078031212485,
      "grad_norm": 0.0009326778817921877,
      "learning_rate": 0.00019321627893274286,
      "loss": 23.0,
      "step": 1991
    },
    {
      "epoch": 1.1956782713085234,
      "grad_norm": 0.0008911388576962054,
      "learning_rate": 0.00019320943794083655,
      "loss": 23.0,
      "step": 1992
    },
    {
      "epoch": 1.1962785114045618,
      "grad_norm": 0.0008909364696592093,
      "learning_rate": 0.0001932025936225252,
      "loss": 23.0,
      "step": 1993
    },
    {
      "epoch": 1.1968787515006003,
      "grad_norm": 0.0005291649722494185,
      "learning_rate": 0.000193195745978053,
      "loss": 23.0,
      "step": 1994
    },
    {
      "epoch": 1.1974789915966386,
      "grad_norm": 0.0015328566078096628,
      "learning_rate": 0.00019318889500766437,
      "loss": 23.0,
      "step": 1995
    },
    {
      "epoch": 1.1980792316926772,
      "grad_norm": 0.0018019595881924033,
      "learning_rate": 0.0001931820407116038,
      "loss": 23.0,
      "step": 1996
    },
    {
      "epoch": 1.1986794717887155,
      "grad_norm": 0.0018812305061146617,
      "learning_rate": 0.00019317518309011587,
      "loss": 23.0,
      "step": 1997
    },
    {
      "epoch": 1.199279711884754,
      "grad_norm": 0.0010826725047081709,
      "learning_rate": 0.00019316832214344532,
      "loss": 23.0,
      "step": 1998
    },
    {
      "epoch": 1.1998799519807923,
      "grad_norm": 0.0004899448831565678,
      "learning_rate": 0.00019316145787183703,
      "loss": 23.0,
      "step": 1999
    },
    {
      "epoch": 1.2004801920768307,
      "grad_norm": 0.0009604006190784276,
      "learning_rate": 0.00019315459027553593,
      "loss": 23.0,
      "step": 2000
    },
    {
      "epoch": 1.2004801920768307,
      "eval_loss": 11.5,
      "eval_runtime": 5.481,
      "eval_samples_per_second": 255.976,
      "eval_steps_per_second": 32.111,
      "step": 2000
    },
    {
      "epoch": 1.2010804321728692,
      "grad_norm": 0.00026419636560603976,
      "learning_rate": 0.00019314771935478715,
      "loss": 23.0,
      "step": 2001
    },
    {
      "epoch": 1.2016806722689075,
      "grad_norm": 0.001335683511570096,
      "learning_rate": 0.00019314084510983586,
      "loss": 23.0,
      "step": 2002
    },
    {
      "epoch": 1.202280912364946,
      "grad_norm": 0.0013580817030742764,
      "learning_rate": 0.00019313396754092738,
      "loss": 23.0,
      "step": 2003
    },
    {
      "epoch": 1.2028811524609844,
      "grad_norm": 0.0006976149743422866,
      "learning_rate": 0.0001931270866483072,
      "loss": 23.0,
      "step": 2004
    },
    {
      "epoch": 1.203481392557023,
      "grad_norm": 0.0003821998252533376,
      "learning_rate": 0.00019312020243222083,
      "loss": 23.0,
      "step": 2005
    },
    {
      "epoch": 1.2040816326530612,
      "grad_norm": 0.0007348189828917384,
      "learning_rate": 0.00019311331489291398,
      "loss": 23.0,
      "step": 2006
    },
    {
      "epoch": 1.2046818727490995,
      "grad_norm": 0.0006483017932623625,
      "learning_rate": 0.00019310642403063243,
      "loss": 23.0,
      "step": 2007
    },
    {
      "epoch": 1.205282112845138,
      "grad_norm": 0.0008024135604500771,
      "learning_rate": 0.0001930995298456221,
      "loss": 23.0,
      "step": 2008
    },
    {
      "epoch": 1.2058823529411764,
      "grad_norm": 0.0016785354819148779,
      "learning_rate": 0.00019309263233812903,
      "loss": 23.0,
      "step": 2009
    },
    {
      "epoch": 1.206482593037215,
      "grad_norm": 0.0009075549314729869,
      "learning_rate": 0.00019308573150839936,
      "loss": 23.0,
      "step": 2010
    },
    {
      "epoch": 1.2070828331332533,
      "grad_norm": 0.0004748840001411736,
      "learning_rate": 0.00019307882735667938,
      "loss": 23.0,
      "step": 2011
    },
    {
      "epoch": 1.2076830732292918,
      "grad_norm": 0.0007885516388341784,
      "learning_rate": 0.0001930719198832155,
      "loss": 23.0,
      "step": 2012
    },
    {
      "epoch": 1.2082833133253301,
      "grad_norm": 0.0006227381527423859,
      "learning_rate": 0.0001930650090882542,
      "loss": 23.0,
      "step": 2013
    },
    {
      "epoch": 1.2088835534213684,
      "grad_norm": 0.0008232907857745886,
      "learning_rate": 0.00019305809497204216,
      "loss": 23.0,
      "step": 2014
    },
    {
      "epoch": 1.209483793517407,
      "grad_norm": 0.0010013766586780548,
      "learning_rate": 0.00019305117753482602,
      "loss": 23.0,
      "step": 2015
    },
    {
      "epoch": 1.2100840336134453,
      "grad_norm": 0.0007637909147888422,
      "learning_rate": 0.00019304425677685273,
      "loss": 23.0,
      "step": 2016
    },
    {
      "epoch": 1.2106842737094838,
      "grad_norm": 0.000717534392606467,
      "learning_rate": 0.00019303733269836925,
      "loss": 23.0,
      "step": 2017
    },
    {
      "epoch": 1.2112845138055222,
      "grad_norm": 0.0015243000816553831,
      "learning_rate": 0.0001930304052996227,
      "loss": 23.0,
      "step": 2018
    },
    {
      "epoch": 1.2118847539015607,
      "grad_norm": 0.0008644640329293907,
      "learning_rate": 0.0001930234745808603,
      "loss": 23.0,
      "step": 2019
    },
    {
      "epoch": 1.212484993997599,
      "grad_norm": 0.0006340988329611719,
      "learning_rate": 0.00019301654054232933,
      "loss": 23.0,
      "step": 2020
    },
    {
      "epoch": 1.2130852340936373,
      "grad_norm": 0.0020876338239759207,
      "learning_rate": 0.00019300960318427735,
      "loss": 23.0,
      "step": 2021
    },
    {
      "epoch": 1.2136854741896759,
      "grad_norm": 0.0013070470886304975,
      "learning_rate": 0.00019300266250695184,
      "loss": 23.0,
      "step": 2022
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 0.00044245924800634384,
      "learning_rate": 0.00019299571851060058,
      "loss": 23.0,
      "step": 2023
    },
    {
      "epoch": 1.2148859543817527,
      "grad_norm": 0.0024676304310560226,
      "learning_rate": 0.0001929887711954713,
      "loss": 23.0,
      "step": 2024
    },
    {
      "epoch": 1.215486194477791,
      "grad_norm": 0.0019125710241496563,
      "learning_rate": 0.00019298182056181197,
      "loss": 23.0,
      "step": 2025
    },
    {
      "epoch": 1.2160864345738296,
      "grad_norm": 0.001551152323372662,
      "learning_rate": 0.00019297486660987064,
      "loss": 23.0,
      "step": 2026
    },
    {
      "epoch": 1.216686674669868,
      "grad_norm": 0.0012455354444682598,
      "learning_rate": 0.00019296790933989554,
      "loss": 23.0,
      "step": 2027
    },
    {
      "epoch": 1.2172869147659064,
      "grad_norm": 0.0008667007205076516,
      "learning_rate": 0.00019296094875213485,
      "loss": 23.0,
      "step": 2028
    },
    {
      "epoch": 1.2178871548619448,
      "grad_norm": 0.0018421221757307649,
      "learning_rate": 0.000192953984846837,
      "loss": 23.0,
      "step": 2029
    },
    {
      "epoch": 1.2184873949579833,
      "grad_norm": 0.0005994903040118515,
      "learning_rate": 0.0001929470176242506,
      "loss": 23.0,
      "step": 2030
    },
    {
      "epoch": 1.2190876350540216,
      "grad_norm": 0.0008131373324431479,
      "learning_rate": 0.00019294004708462418,
      "loss": 23.0,
      "step": 2031
    },
    {
      "epoch": 1.21968787515006,
      "grad_norm": 0.0005514805088751018,
      "learning_rate": 0.00019293307322820654,
      "loss": 23.0,
      "step": 2032
    },
    {
      "epoch": 1.2202881152460985,
      "grad_norm": 0.002355314092710614,
      "learning_rate": 0.0001929260960552466,
      "loss": 23.0,
      "step": 2033
    },
    {
      "epoch": 1.2208883553421368,
      "grad_norm": 0.0011875891359522939,
      "learning_rate": 0.00019291911556599333,
      "loss": 23.0,
      "step": 2034
    },
    {
      "epoch": 1.2214885954381753,
      "grad_norm": 0.001237823162227869,
      "learning_rate": 0.00019291213176069586,
      "loss": 23.0,
      "step": 2035
    },
    {
      "epoch": 1.2220888355342137,
      "grad_norm": 0.0003888348874170333,
      "learning_rate": 0.00019290514463960335,
      "loss": 23.0,
      "step": 2036
    },
    {
      "epoch": 1.2226890756302522,
      "grad_norm": 0.0008330006385222077,
      "learning_rate": 0.00019289815420296522,
      "loss": 23.0,
      "step": 2037
    },
    {
      "epoch": 1.2232893157262905,
      "grad_norm": 0.0006106719956733286,
      "learning_rate": 0.00019289116045103098,
      "loss": 23.0,
      "step": 2038
    },
    {
      "epoch": 1.2238895558223288,
      "grad_norm": 0.0008156050462275743,
      "learning_rate": 0.0001928841633840501,
      "loss": 23.0,
      "step": 2039
    },
    {
      "epoch": 1.2244897959183674,
      "grad_norm": 0.0006298265652731061,
      "learning_rate": 0.00019287716300227242,
      "loss": 23.0,
      "step": 2040
    },
    {
      "epoch": 1.2250900360144057,
      "grad_norm": 0.0007098076166585088,
      "learning_rate": 0.00019287015930594765,
      "loss": 23.0,
      "step": 2041
    },
    {
      "epoch": 1.2256902761104442,
      "grad_norm": 0.001036522793583572,
      "learning_rate": 0.00019286315229532578,
      "loss": 23.0,
      "step": 2042
    },
    {
      "epoch": 1.2262905162064826,
      "grad_norm": 0.0007545838598161936,
      "learning_rate": 0.00019285614197065692,
      "loss": 23.0,
      "step": 2043
    },
    {
      "epoch": 1.226890756302521,
      "grad_norm": 0.0007426731754094362,
      "learning_rate": 0.0001928491283321912,
      "loss": 23.0,
      "step": 2044
    },
    {
      "epoch": 1.2274909963985594,
      "grad_norm": 0.0010281699942424893,
      "learning_rate": 0.0001928421113801789,
      "loss": 23.0,
      "step": 2045
    },
    {
      "epoch": 1.2280912364945977,
      "grad_norm": 0.002106086118146777,
      "learning_rate": 0.00019283509111487048,
      "loss": 23.0,
      "step": 2046
    },
    {
      "epoch": 1.2286914765906363,
      "grad_norm": 0.0012783302227035165,
      "learning_rate": 0.00019282806753651645,
      "loss": 23.0,
      "step": 2047
    },
    {
      "epoch": 1.2292917166866746,
      "grad_norm": 0.0009432644001208246,
      "learning_rate": 0.00019282104064536747,
      "loss": 23.0,
      "step": 2048
    },
    {
      "epoch": 1.2298919567827131,
      "grad_norm": 0.0011243056505918503,
      "learning_rate": 0.00019281401044167433,
      "loss": 23.0,
      "step": 2049
    },
    {
      "epoch": 1.2304921968787514,
      "grad_norm": 0.0022574688773602247,
      "learning_rate": 0.0001928069769256879,
      "loss": 23.0,
      "step": 2050
    },
    {
      "epoch": 1.23109243697479,
      "grad_norm": 0.0010988580761477351,
      "learning_rate": 0.00019279994009765918,
      "loss": 23.0,
      "step": 2051
    },
    {
      "epoch": 1.2316926770708283,
      "grad_norm": 0.0012815982336178422,
      "learning_rate": 0.00019279289995783932,
      "loss": 23.0,
      "step": 2052
    },
    {
      "epoch": 1.2322929171668668,
      "grad_norm": 0.0008450283785350621,
      "learning_rate": 0.00019278585650647957,
      "loss": 23.0,
      "step": 2053
    },
    {
      "epoch": 1.2328931572629052,
      "grad_norm": 0.0011646057246252894,
      "learning_rate": 0.00019277880974383124,
      "loss": 23.0,
      "step": 2054
    },
    {
      "epoch": 1.2334933973589437,
      "grad_norm": 0.00048065732698887587,
      "learning_rate": 0.00019277175967014586,
      "loss": 23.0,
      "step": 2055
    },
    {
      "epoch": 1.234093637454982,
      "grad_norm": 0.0006098438170738518,
      "learning_rate": 0.000192764706285675,
      "loss": 23.0,
      "step": 2056
    },
    {
      "epoch": 1.2346938775510203,
      "grad_norm": 0.0008195044356398284,
      "learning_rate": 0.00019275764959067042,
      "loss": 23.0,
      "step": 2057
    },
    {
      "epoch": 1.2352941176470589,
      "grad_norm": 0.000680581375490874,
      "learning_rate": 0.0001927505895853839,
      "loss": 23.0,
      "step": 2058
    },
    {
      "epoch": 1.2358943577430972,
      "grad_norm": 0.00044709868961945176,
      "learning_rate": 0.00019274352627006745,
      "loss": 23.0,
      "step": 2059
    },
    {
      "epoch": 1.2364945978391357,
      "grad_norm": 0.0006181981298141181,
      "learning_rate": 0.0001927364596449731,
      "loss": 23.0,
      "step": 2060
    },
    {
      "epoch": 1.237094837935174,
      "grad_norm": 0.0008112747455015779,
      "learning_rate": 0.00019272938971035304,
      "loss": 23.0,
      "step": 2061
    },
    {
      "epoch": 1.2376950780312126,
      "grad_norm": 0.001151968608610332,
      "learning_rate": 0.0001927223164664596,
      "loss": 23.0,
      "step": 2062
    },
    {
      "epoch": 1.238295318127251,
      "grad_norm": 0.000988475512713194,
      "learning_rate": 0.00019271523991354518,
      "loss": 23.0,
      "step": 2063
    },
    {
      "epoch": 1.2388955582232892,
      "grad_norm": 0.0005085468292236328,
      "learning_rate": 0.00019270816005186237,
      "loss": 23.0,
      "step": 2064
    },
    {
      "epoch": 1.2394957983193278,
      "grad_norm": 0.0015622277278453112,
      "learning_rate": 0.00019270107688166378,
      "loss": 23.0,
      "step": 2065
    },
    {
      "epoch": 1.240096038415366,
      "grad_norm": 0.00020089696045033634,
      "learning_rate": 0.0001926939904032022,
      "loss": 23.0,
      "step": 2066
    },
    {
      "epoch": 1.2406962785114046,
      "grad_norm": 0.0010697637917473912,
      "learning_rate": 0.00019268690061673057,
      "loss": 23.0,
      "step": 2067
    },
    {
      "epoch": 1.241296518607443,
      "grad_norm": 0.0005508419126272202,
      "learning_rate": 0.00019267980752250185,
      "loss": 23.0,
      "step": 2068
    },
    {
      "epoch": 1.2418967587034815,
      "grad_norm": 0.0016626521246507764,
      "learning_rate": 0.00019267271112076922,
      "loss": 23.0,
      "step": 2069
    },
    {
      "epoch": 1.2424969987995198,
      "grad_norm": 0.0008421835373155773,
      "learning_rate": 0.0001926656114117859,
      "loss": 23.0,
      "step": 2070
    },
    {
      "epoch": 1.2430972388955581,
      "grad_norm": 0.0008170657092705369,
      "learning_rate": 0.00019265850839580528,
      "loss": 23.0,
      "step": 2071
    },
    {
      "epoch": 1.2436974789915967,
      "grad_norm": 0.0004373532719910145,
      "learning_rate": 0.00019265140207308084,
      "loss": 23.0,
      "step": 2072
    },
    {
      "epoch": 1.244297719087635,
      "grad_norm": 0.0011121369898319244,
      "learning_rate": 0.0001926442924438662,
      "loss": 23.0,
      "step": 2073
    },
    {
      "epoch": 1.2448979591836735,
      "grad_norm": 0.002869250951334834,
      "learning_rate": 0.00019263717950841506,
      "loss": 23.0,
      "step": 2074
    },
    {
      "epoch": 1.2454981992797118,
      "grad_norm": 0.0016094837337732315,
      "learning_rate": 0.00019263006326698126,
      "loss": 23.0,
      "step": 2075
    },
    {
      "epoch": 1.2460984393757504,
      "grad_norm": 0.002325515728443861,
      "learning_rate": 0.00019262294371981878,
      "loss": 23.0,
      "step": 2076
    },
    {
      "epoch": 1.2466986794717887,
      "grad_norm": 0.0007525268592871726,
      "learning_rate": 0.0001926158208671817,
      "loss": 23.0,
      "step": 2077
    },
    {
      "epoch": 1.247298919567827,
      "grad_norm": 0.001686456729657948,
      "learning_rate": 0.0001926086947093242,
      "loss": 23.0,
      "step": 2078
    },
    {
      "epoch": 1.2478991596638656,
      "grad_norm": 0.0010857430752366781,
      "learning_rate": 0.00019260156524650063,
      "loss": 23.0,
      "step": 2079
    },
    {
      "epoch": 1.2484993997599039,
      "grad_norm": 0.002079520607367158,
      "learning_rate": 0.00019259443247896538,
      "loss": 23.0,
      "step": 2080
    },
    {
      "epoch": 1.2490996398559424,
      "grad_norm": 0.0006012477679178119,
      "learning_rate": 0.00019258729640697303,
      "loss": 23.0,
      "step": 2081
    },
    {
      "epoch": 1.2496998799519807,
      "grad_norm": 0.000783745723310858,
      "learning_rate": 0.0001925801570307782,
      "loss": 23.0,
      "step": 2082
    },
    {
      "epoch": 1.2503001200480193,
      "grad_norm": 0.0021253866143524647,
      "learning_rate": 0.00019257301435063573,
      "loss": 23.0,
      "step": 2083
    },
    {
      "epoch": 1.2509003601440576,
      "grad_norm": 0.0007235134835354984,
      "learning_rate": 0.00019256586836680054,
      "loss": 23.0,
      "step": 2084
    },
    {
      "epoch": 1.251500600240096,
      "grad_norm": 0.0009064192418009043,
      "learning_rate": 0.00019255871907952757,
      "loss": 23.0,
      "step": 2085
    },
    {
      "epoch": 1.2521008403361344,
      "grad_norm": 0.00042548036435618997,
      "learning_rate": 0.00019255156648907202,
      "loss": 23.0,
      "step": 2086
    },
    {
      "epoch": 1.252701080432173,
      "grad_norm": 0.0005528698093257844,
      "learning_rate": 0.0001925444105956891,
      "loss": 23.0,
      "step": 2087
    },
    {
      "epoch": 1.2533013205282113,
      "grad_norm": 0.00168812635820359,
      "learning_rate": 0.00019253725139963427,
      "loss": 23.0,
      "step": 2088
    },
    {
      "epoch": 1.2539015606242496,
      "grad_norm": 0.0014038488734513521,
      "learning_rate": 0.00019253008890116296,
      "loss": 23.0,
      "step": 2089
    },
    {
      "epoch": 1.2545018007202882,
      "grad_norm": 0.0008206113707274199,
      "learning_rate": 0.00019252292310053079,
      "loss": 23.0,
      "step": 2090
    },
    {
      "epoch": 1.2551020408163265,
      "grad_norm": 0.0004761457093991339,
      "learning_rate": 0.00019251575399799347,
      "loss": 23.0,
      "step": 2091
    },
    {
      "epoch": 1.255702280912365,
      "grad_norm": 0.001002299482934177,
      "learning_rate": 0.0001925085815938069,
      "loss": 23.0,
      "step": 2092
    },
    {
      "epoch": 1.2563025210084033,
      "grad_norm": 0.0009327147854492068,
      "learning_rate": 0.00019250140588822698,
      "loss": 23.0,
      "step": 2093
    },
    {
      "epoch": 1.2569027611044419,
      "grad_norm": 0.0012192872818559408,
      "learning_rate": 0.00019249422688150984,
      "loss": 23.0,
      "step": 2094
    },
    {
      "epoch": 1.2575030012004802,
      "grad_norm": 0.0009881231235340238,
      "learning_rate": 0.00019248704457391168,
      "loss": 23.0,
      "step": 2095
    },
    {
      "epoch": 1.2581032412965185,
      "grad_norm": 0.001701731001958251,
      "learning_rate": 0.0001924798589656888,
      "loss": 23.0,
      "step": 2096
    },
    {
      "epoch": 1.258703481392557,
      "grad_norm": 0.0018471120856702328,
      "learning_rate": 0.00019247267005709763,
      "loss": 23.0,
      "step": 2097
    },
    {
      "epoch": 1.2593037214885954,
      "grad_norm": 0.0005706116207875311,
      "learning_rate": 0.00019246547784839473,
      "loss": 23.0,
      "step": 2098
    },
    {
      "epoch": 1.259903961584634,
      "grad_norm": 0.002569433068856597,
      "learning_rate": 0.0001924582823398368,
      "loss": 23.0,
      "step": 2099
    },
    {
      "epoch": 1.2605042016806722,
      "grad_norm": 0.0016303047304973006,
      "learning_rate": 0.00019245108353168057,
      "loss": 23.0,
      "step": 2100
    },
    {
      "epoch": 1.2611044417767108,
      "grad_norm": 0.0008211960666812956,
      "learning_rate": 0.000192443881424183,
      "loss": 23.0,
      "step": 2101
    },
    {
      "epoch": 1.261704681872749,
      "grad_norm": 0.002193295396864414,
      "learning_rate": 0.0001924366760176011,
      "loss": 23.0,
      "step": 2102
    },
    {
      "epoch": 1.2623049219687874,
      "grad_norm": 0.0011547781759873033,
      "learning_rate": 0.000192429467312192,
      "loss": 23.0,
      "step": 2103
    },
    {
      "epoch": 1.262905162064826,
      "grad_norm": 0.0009378142422065139,
      "learning_rate": 0.00019242225530821297,
      "loss": 23.0,
      "step": 2104
    },
    {
      "epoch": 1.2635054021608643,
      "grad_norm": 0.0003709706070367247,
      "learning_rate": 0.00019241504000592138,
      "loss": 23.0,
      "step": 2105
    },
    {
      "epoch": 1.2641056422569028,
      "grad_norm": 0.0010277227265760303,
      "learning_rate": 0.00019240782140557476,
      "loss": 23.0,
      "step": 2106
    },
    {
      "epoch": 1.2647058823529411,
      "grad_norm": 0.001920362701639533,
      "learning_rate": 0.00019240059950743067,
      "loss": 23.0,
      "step": 2107
    },
    {
      "epoch": 1.2653061224489797,
      "grad_norm": 0.002360978629440069,
      "learning_rate": 0.00019239337431174687,
      "loss": 23.0,
      "step": 2108
    },
    {
      "epoch": 1.265906362545018,
      "grad_norm": 0.0010547456331551075,
      "learning_rate": 0.00019238614581878121,
      "loss": 23.0,
      "step": 2109
    },
    {
      "epoch": 1.2665066026410563,
      "grad_norm": 0.000743915035855025,
      "learning_rate": 0.0001923789140287916,
      "loss": 23.0,
      "step": 2110
    },
    {
      "epoch": 1.2671068427370948,
      "grad_norm": 0.001828330336138606,
      "learning_rate": 0.00019237167894203626,
      "loss": 23.0,
      "step": 2111
    },
    {
      "epoch": 1.2677070828331334,
      "grad_norm": 0.0018606131197884679,
      "learning_rate": 0.00019236444055877326,
      "loss": 23.0,
      "step": 2112
    },
    {
      "epoch": 1.2683073229291717,
      "grad_norm": 0.0015011877985671163,
      "learning_rate": 0.000192357198879261,
      "loss": 23.0,
      "step": 2113
    },
    {
      "epoch": 1.26890756302521,
      "grad_norm": 0.0007122366223484278,
      "learning_rate": 0.00019234995390375785,
      "loss": 23.0,
      "step": 2114
    },
    {
      "epoch": 1.2695078031212486,
      "grad_norm": 0.0007740426226519048,
      "learning_rate": 0.00019234270563252238,
      "loss": 23.0,
      "step": 2115
    },
    {
      "epoch": 1.2701080432172869,
      "grad_norm": 0.0010612767655402422,
      "learning_rate": 0.00019233545406581332,
      "loss": 23.0,
      "step": 2116
    },
    {
      "epoch": 1.2707082833133252,
      "grad_norm": 0.0016505690291523933,
      "learning_rate": 0.0001923281992038894,
      "loss": 23.0,
      "step": 2117
    },
    {
      "epoch": 1.2713085234093637,
      "grad_norm": 0.0023043276742100716,
      "learning_rate": 0.00019232094104700958,
      "loss": 23.0,
      "step": 2118
    },
    {
      "epoch": 1.2719087635054023,
      "grad_norm": 0.001076166401617229,
      "learning_rate": 0.00019231367959543283,
      "loss": 23.0,
      "step": 2119
    },
    {
      "epoch": 1.2725090036014406,
      "grad_norm": 0.0007836589938960969,
      "learning_rate": 0.0001923064148494183,
      "loss": 23.0,
      "step": 2120
    },
    {
      "epoch": 1.273109243697479,
      "grad_norm": 0.0007020245539024472,
      "learning_rate": 0.00019229914680922528,
      "loss": 23.0,
      "step": 2121
    },
    {
      "epoch": 1.2737094837935174,
      "grad_norm": 0.0008544962620362639,
      "learning_rate": 0.00019229187547511312,
      "loss": 23.0,
      "step": 2122
    },
    {
      "epoch": 1.2743097238895558,
      "grad_norm": 0.0005639759474433959,
      "learning_rate": 0.0001922846008473413,
      "loss": 23.0,
      "step": 2123
    },
    {
      "epoch": 1.2749099639855943,
      "grad_norm": 0.0004858477332163602,
      "learning_rate": 0.00019227732292616953,
      "loss": 23.0,
      "step": 2124
    },
    {
      "epoch": 1.2755102040816326,
      "grad_norm": 0.0009530950337648392,
      "learning_rate": 0.00019227004171185744,
      "loss": 23.0,
      "step": 2125
    },
    {
      "epoch": 1.2761104441776712,
      "grad_norm": 0.0008700921316631138,
      "learning_rate": 0.0001922627572046649,
      "loss": 23.0,
      "step": 2126
    },
    {
      "epoch": 1.2767106842737095,
      "grad_norm": 0.00051358831115067,
      "learning_rate": 0.00019225546940485186,
      "loss": 23.0,
      "step": 2127
    },
    {
      "epoch": 1.2773109243697478,
      "grad_norm": 0.0018087092321366072,
      "learning_rate": 0.00019224817831267849,
      "loss": 23.0,
      "step": 2128
    },
    {
      "epoch": 1.2779111644657863,
      "grad_norm": 0.0012789365136995912,
      "learning_rate": 0.00019224088392840488,
      "loss": 23.0,
      "step": 2129
    },
    {
      "epoch": 1.2785114045618247,
      "grad_norm": 0.0012055577244609594,
      "learning_rate": 0.0001922335862522914,
      "loss": 23.0,
      "step": 2130
    },
    {
      "epoch": 1.2791116446578632,
      "grad_norm": 0.0004859871114604175,
      "learning_rate": 0.00019222628528459847,
      "loss": 23.0,
      "step": 2131
    },
    {
      "epoch": 1.2797118847539015,
      "grad_norm": 0.002568140160292387,
      "learning_rate": 0.00019221898102558666,
      "loss": 23.0,
      "step": 2132
    },
    {
      "epoch": 1.28031212484994,
      "grad_norm": 0.0009791409829631448,
      "learning_rate": 0.00019221167347551664,
      "loss": 23.0,
      "step": 2133
    },
    {
      "epoch": 1.2809123649459784,
      "grad_norm": 0.0005456181825138628,
      "learning_rate": 0.00019220436263464922,
      "loss": 23.0,
      "step": 2134
    },
    {
      "epoch": 1.2815126050420167,
      "grad_norm": 0.000637693505268544,
      "learning_rate": 0.0001921970485032452,
      "loss": 23.0,
      "step": 2135
    },
    {
      "epoch": 1.2821128451380552,
      "grad_norm": 0.0009480795706622303,
      "learning_rate": 0.00019218973108156573,
      "loss": 23.0,
      "step": 2136
    },
    {
      "epoch": 1.2827130852340938,
      "grad_norm": 0.0003983963106293231,
      "learning_rate": 0.0001921824103698719,
      "loss": 23.0,
      "step": 2137
    },
    {
      "epoch": 1.283313325330132,
      "grad_norm": 0.0008890287135727704,
      "learning_rate": 0.00019217508636842494,
      "loss": 23.0,
      "step": 2138
    },
    {
      "epoch": 1.2839135654261704,
      "grad_norm": 0.0007761778542771935,
      "learning_rate": 0.00019216775907748628,
      "loss": 23.0,
      "step": 2139
    },
    {
      "epoch": 1.284513805522209,
      "grad_norm": 0.0007144435658119619,
      "learning_rate": 0.00019216042849731736,
      "loss": 23.0,
      "step": 2140
    },
    {
      "epoch": 1.2851140456182473,
      "grad_norm": 0.001242082566022873,
      "learning_rate": 0.00019215309462817983,
      "loss": 23.0,
      "step": 2141
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.0006997886812314391,
      "learning_rate": 0.00019214575747033538,
      "loss": 23.0,
      "step": 2142
    },
    {
      "epoch": 1.2863145258103241,
      "grad_norm": 0.0010159621015191078,
      "learning_rate": 0.0001921384170240459,
      "loss": 23.0,
      "step": 2143
    },
    {
      "epoch": 1.2869147659063627,
      "grad_norm": 0.0007167985895648599,
      "learning_rate": 0.00019213107328957328,
      "loss": 23.0,
      "step": 2144
    },
    {
      "epoch": 1.287515006002401,
      "grad_norm": 0.0008636672864668071,
      "learning_rate": 0.0001921237262671797,
      "loss": 23.0,
      "step": 2145
    },
    {
      "epoch": 1.2881152460984393,
      "grad_norm": 0.0005083580617792904,
      "learning_rate": 0.00019211637595712726,
      "loss": 23.0,
      "step": 2146
    },
    {
      "epoch": 1.2887154861944778,
      "grad_norm": 0.0004971363814547658,
      "learning_rate": 0.0001921090223596783,
      "loss": 23.0,
      "step": 2147
    },
    {
      "epoch": 1.2893157262905162,
      "grad_norm": 0.0008241168106906116,
      "learning_rate": 0.0001921016654750953,
      "loss": 23.0,
      "step": 2148
    },
    {
      "epoch": 1.2899159663865547,
      "grad_norm": 0.001977398758754134,
      "learning_rate": 0.00019209430530364076,
      "loss": 23.0,
      "step": 2149
    },
    {
      "epoch": 1.290516206482593,
      "grad_norm": 0.0014348250115290284,
      "learning_rate": 0.00019208694184557736,
      "loss": 23.0,
      "step": 2150
    },
    {
      "epoch": 1.2911164465786316,
      "grad_norm": 0.0009256337652914226,
      "learning_rate": 0.00019207957510116786,
      "loss": 23.0,
      "step": 2151
    },
    {
      "epoch": 1.2917166866746699,
      "grad_norm": 0.0022123123053461313,
      "learning_rate": 0.00019207220507067523,
      "loss": 23.0,
      "step": 2152
    },
    {
      "epoch": 1.2923169267707082,
      "grad_norm": 0.0014656918356195092,
      "learning_rate": 0.00019206483175436242,
      "loss": 23.0,
      "step": 2153
    },
    {
      "epoch": 1.2929171668667467,
      "grad_norm": 0.0016659768298268318,
      "learning_rate": 0.00019205745515249255,
      "loss": 23.0,
      "step": 2154
    },
    {
      "epoch": 1.293517406962785,
      "grad_norm": 0.000796042731963098,
      "learning_rate": 0.00019205007526532894,
      "loss": 23.0,
      "step": 2155
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 0.000784591306000948,
      "learning_rate": 0.00019204269209313493,
      "loss": 23.0,
      "step": 2156
    },
    {
      "epoch": 1.294717887154862,
      "grad_norm": 0.0017622907180339098,
      "learning_rate": 0.000192035305636174,
      "loss": 23.0,
      "step": 2157
    },
    {
      "epoch": 1.2953181272509005,
      "grad_norm": 0.0008178095449693501,
      "learning_rate": 0.0001920279158947097,
      "loss": 23.0,
      "step": 2158
    },
    {
      "epoch": 1.2959183673469388,
      "grad_norm": 0.0005034728674218059,
      "learning_rate": 0.00019202052286900587,
      "loss": 23.0,
      "step": 2159
    },
    {
      "epoch": 1.296518607442977,
      "grad_norm": 0.00046888680662959814,
      "learning_rate": 0.00019201312655932627,
      "loss": 23.0,
      "step": 2160
    },
    {
      "epoch": 1.2971188475390156,
      "grad_norm": 0.0005256042932160199,
      "learning_rate": 0.00019200572696593485,
      "loss": 23.0,
      "step": 2161
    },
    {
      "epoch": 1.297719087635054,
      "grad_norm": 0.0023607781622558832,
      "learning_rate": 0.00019199832408909573,
      "loss": 23.0,
      "step": 2162
    },
    {
      "epoch": 1.2983193277310925,
      "grad_norm": 0.0007398081361316144,
      "learning_rate": 0.00019199091792907304,
      "loss": 23.0,
      "step": 2163
    },
    {
      "epoch": 1.2989195678271308,
      "grad_norm": 0.0015953012043610215,
      "learning_rate": 0.00019198350848613114,
      "loss": 23.0,
      "step": 2164
    },
    {
      "epoch": 1.2995198079231693,
      "grad_norm": 0.000722983677405864,
      "learning_rate": 0.0001919760957605344,
      "loss": 23.0,
      "step": 2165
    },
    {
      "epoch": 1.3001200480192077,
      "grad_norm": 0.0008493372588418424,
      "learning_rate": 0.00019196867975254742,
      "loss": 23.0,
      "step": 2166
    },
    {
      "epoch": 1.300720288115246,
      "grad_norm": 0.0012202919460833073,
      "learning_rate": 0.00019196126046243484,
      "loss": 23.0,
      "step": 2167
    },
    {
      "epoch": 1.3013205282112845,
      "grad_norm": 0.0005322351935319602,
      "learning_rate": 0.0001919538378904614,
      "loss": 23.0,
      "step": 2168
    },
    {
      "epoch": 1.301920768307323,
      "grad_norm": 0.0007932167500257492,
      "learning_rate": 0.00019194641203689204,
      "loss": 23.0,
      "step": 2169
    },
    {
      "epoch": 1.3025210084033614,
      "grad_norm": 0.0009172199061140418,
      "learning_rate": 0.00019193898290199174,
      "loss": 23.0,
      "step": 2170
    },
    {
      "epoch": 1.3031212484993997,
      "grad_norm": 0.0008713841089047492,
      "learning_rate": 0.00019193155048602564,
      "loss": 23.0,
      "step": 2171
    },
    {
      "epoch": 1.3037214885954382,
      "grad_norm": 0.0030673607252538204,
      "learning_rate": 0.000191924114789259,
      "loss": 23.0,
      "step": 2172
    },
    {
      "epoch": 1.3043217286914766,
      "grad_norm": 0.0025866650976240635,
      "learning_rate": 0.00019191667581195713,
      "loss": 23.0,
      "step": 2173
    },
    {
      "epoch": 1.3049219687875149,
      "grad_norm": 0.000421585253207013,
      "learning_rate": 0.00019190923355438556,
      "loss": 23.0,
      "step": 2174
    },
    {
      "epoch": 1.3055222088835534,
      "grad_norm": 0.00038466870319098234,
      "learning_rate": 0.00019190178801680984,
      "loss": 23.0,
      "step": 2175
    },
    {
      "epoch": 1.306122448979592,
      "grad_norm": 0.001396143576130271,
      "learning_rate": 0.0001918943391994957,
      "loss": 23.0,
      "step": 2176
    },
    {
      "epoch": 1.3067226890756303,
      "grad_norm": 0.0006378819234669209,
      "learning_rate": 0.000191886887102709,
      "loss": 23.0,
      "step": 2177
    },
    {
      "epoch": 1.3073229291716686,
      "grad_norm": 0.0013305838219821453,
      "learning_rate": 0.00019187943172671563,
      "loss": 23.0,
      "step": 2178
    },
    {
      "epoch": 1.3079231692677071,
      "grad_norm": 0.0008714368450455368,
      "learning_rate": 0.00019187197307178171,
      "loss": 23.0,
      "step": 2179
    },
    {
      "epoch": 1.3085234093637454,
      "grad_norm": 0.0007764401962049305,
      "learning_rate": 0.00019186451113817338,
      "loss": 23.0,
      "step": 2180
    },
    {
      "epoch": 1.309123649459784,
      "grad_norm": 0.0008344615343958139,
      "learning_rate": 0.00019185704592615698,
      "loss": 23.0,
      "step": 2181
    },
    {
      "epoch": 1.3097238895558223,
      "grad_norm": 0.001748396665789187,
      "learning_rate": 0.00019184957743599888,
      "loss": 23.0,
      "step": 2182
    },
    {
      "epoch": 1.3103241296518608,
      "grad_norm": 0.0007260398706421256,
      "learning_rate": 0.0001918421056679656,
      "loss": 23.0,
      "step": 2183
    },
    {
      "epoch": 1.3109243697478992,
      "grad_norm": 0.0010761874727904797,
      "learning_rate": 0.00019183463062232386,
      "loss": 23.0,
      "step": 2184
    },
    {
      "epoch": 1.3115246098439375,
      "grad_norm": 0.0006633952725678682,
      "learning_rate": 0.0001918271522993403,
      "loss": 23.0,
      "step": 2185
    },
    {
      "epoch": 1.312124849939976,
      "grad_norm": 0.00033936891122721136,
      "learning_rate": 0.00019181967069928198,
      "loss": 23.0,
      "step": 2186
    },
    {
      "epoch": 1.3127250900360143,
      "grad_norm": 0.0006078967708162963,
      "learning_rate": 0.00019181218582241572,
      "loss": 23.0,
      "step": 2187
    },
    {
      "epoch": 1.3133253301320529,
      "grad_norm": 0.0008051611948758364,
      "learning_rate": 0.00019180469766900878,
      "loss": 23.0,
      "step": 2188
    },
    {
      "epoch": 1.3139255702280912,
      "grad_norm": 0.0008598083513788879,
      "learning_rate": 0.0001917972062393283,
      "loss": 23.0,
      "step": 2189
    },
    {
      "epoch": 1.3145258103241297,
      "grad_norm": 0.0005701205227524042,
      "learning_rate": 0.00019178971153364167,
      "loss": 23.0,
      "step": 2190
    },
    {
      "epoch": 1.315126050420168,
      "grad_norm": 0.0019546288531273603,
      "learning_rate": 0.00019178221355221634,
      "loss": 23.0,
      "step": 2191
    },
    {
      "epoch": 1.3157262905162064,
      "grad_norm": 0.0011223703622817993,
      "learning_rate": 0.00019177471229531992,
      "loss": 23.0,
      "step": 2192
    },
    {
      "epoch": 1.316326530612245,
      "grad_norm": 0.0010649703908711672,
      "learning_rate": 0.00019176720776322006,
      "loss": 23.0,
      "step": 2193
    },
    {
      "epoch": 1.3169267707082832,
      "grad_norm": 0.0003854198439512402,
      "learning_rate": 0.00019175969995618463,
      "loss": 23.0,
      "step": 2194
    },
    {
      "epoch": 1.3175270108043218,
      "grad_norm": 0.0007395111606456339,
      "learning_rate": 0.00019175218887448154,
      "loss": 23.0,
      "step": 2195
    },
    {
      "epoch": 1.31812725090036,
      "grad_norm": 0.000572826131246984,
      "learning_rate": 0.00019174467451837883,
      "loss": 23.0,
      "step": 2196
    },
    {
      "epoch": 1.3187274909963986,
      "grad_norm": 0.0016554270405322313,
      "learning_rate": 0.00019173715688814468,
      "loss": 23.0,
      "step": 2197
    },
    {
      "epoch": 1.319327731092437,
      "grad_norm": 0.001633728388696909,
      "learning_rate": 0.00019172963598404737,
      "loss": 23.0,
      "step": 2198
    },
    {
      "epoch": 1.3199279711884753,
      "grad_norm": 0.0005962157738395035,
      "learning_rate": 0.00019172211180635534,
      "loss": 23.0,
      "step": 2199
    },
    {
      "epoch": 1.3205282112845138,
      "grad_norm": 0.0007311934023164213,
      "learning_rate": 0.00019171458435533707,
      "loss": 23.0,
      "step": 2200
    },
    {
      "epoch": 1.3211284513805523,
      "grad_norm": 0.0023586212191730738,
      "learning_rate": 0.00019170705363126122,
      "loss": 23.0,
      "step": 2201
    },
    {
      "epoch": 1.3217286914765907,
      "grad_norm": 0.0025315319653600454,
      "learning_rate": 0.00019169951963439647,
      "loss": 23.0,
      "step": 2202
    },
    {
      "epoch": 1.322328931572629,
      "grad_norm": 0.0017554480582475662,
      "learning_rate": 0.00019169198236501178,
      "loss": 23.0,
      "step": 2203
    },
    {
      "epoch": 1.3229291716686675,
      "grad_norm": 0.0008356471662409604,
      "learning_rate": 0.00019168444182337608,
      "loss": 23.0,
      "step": 2204
    },
    {
      "epoch": 1.3235294117647058,
      "grad_norm": 0.0005913945497013628,
      "learning_rate": 0.0001916768980097585,
      "loss": 23.0,
      "step": 2205
    },
    {
      "epoch": 1.3241296518607442,
      "grad_norm": 0.0014620408182963729,
      "learning_rate": 0.0001916693509244283,
      "loss": 23.0,
      "step": 2206
    },
    {
      "epoch": 1.3247298919567827,
      "grad_norm": 0.000512272643391043,
      "learning_rate": 0.0001916618005676547,
      "loss": 23.0,
      "step": 2207
    },
    {
      "epoch": 1.3253301320528212,
      "grad_norm": 0.0008890106109902263,
      "learning_rate": 0.00019165424693970724,
      "loss": 23.0,
      "step": 2208
    },
    {
      "epoch": 1.3259303721488596,
      "grad_norm": 0.0010873951250687242,
      "learning_rate": 0.00019164669004085547,
      "loss": 23.0,
      "step": 2209
    },
    {
      "epoch": 1.3265306122448979,
      "grad_norm": 0.0007970140432007611,
      "learning_rate": 0.00019163912987136907,
      "loss": 23.0,
      "step": 2210
    },
    {
      "epoch": 1.3271308523409364,
      "grad_norm": 0.00045438751112669706,
      "learning_rate": 0.00019163156643151787,
      "loss": 23.0,
      "step": 2211
    },
    {
      "epoch": 1.3277310924369747,
      "grad_norm": 0.001197776640765369,
      "learning_rate": 0.00019162399972157178,
      "loss": 23.0,
      "step": 2212
    },
    {
      "epoch": 1.3283313325330133,
      "grad_norm": 0.0003519585879985243,
      "learning_rate": 0.0001916164297418008,
      "loss": 23.0,
      "step": 2213
    },
    {
      "epoch": 1.3289315726290516,
      "grad_norm": 0.0005266814259812236,
      "learning_rate": 0.00019160885649247507,
      "loss": 23.0,
      "step": 2214
    },
    {
      "epoch": 1.3295318127250901,
      "grad_norm": 0.0012946182396262884,
      "learning_rate": 0.00019160127997386496,
      "loss": 23.0,
      "step": 2215
    },
    {
      "epoch": 1.3301320528211285,
      "grad_norm": 0.001070786383934319,
      "learning_rate": 0.00019159370018624077,
      "loss": 23.0,
      "step": 2216
    },
    {
      "epoch": 1.3307322929171668,
      "grad_norm": 0.0010140641825273633,
      "learning_rate": 0.000191586117129873,
      "loss": 23.0,
      "step": 2217
    },
    {
      "epoch": 1.3313325330132053,
      "grad_norm": 0.0005159311112947762,
      "learning_rate": 0.00019157853080503233,
      "loss": 23.0,
      "step": 2218
    },
    {
      "epoch": 1.3319327731092436,
      "grad_norm": 0.00089635094627738,
      "learning_rate": 0.0001915709412119895,
      "loss": 23.0,
      "step": 2219
    },
    {
      "epoch": 1.3325330132052822,
      "grad_norm": 0.0012715085176751018,
      "learning_rate": 0.00019156334835101525,
      "loss": 23.0,
      "step": 2220
    },
    {
      "epoch": 1.3331332533013205,
      "grad_norm": 0.0008847500430420041,
      "learning_rate": 0.00019155575222238068,
      "loss": 23.0,
      "step": 2221
    },
    {
      "epoch": 1.333733493397359,
      "grad_norm": 0.0015832451172173023,
      "learning_rate": 0.00019154815282635678,
      "loss": 23.0,
      "step": 2222
    },
    {
      "epoch": 1.3343337334933973,
      "grad_norm": 0.0008942346321418881,
      "learning_rate": 0.00019154055016321484,
      "loss": 23.0,
      "step": 2223
    },
    {
      "epoch": 1.3349339735894357,
      "grad_norm": 0.0011502058478072286,
      "learning_rate": 0.00019153294423322612,
      "loss": 23.0,
      "step": 2224
    },
    {
      "epoch": 1.3355342136854742,
      "grad_norm": 0.0013056400930508971,
      "learning_rate": 0.00019152533503666205,
      "loss": 23.0,
      "step": 2225
    },
    {
      "epoch": 1.3361344537815127,
      "grad_norm": 0.0015896232798695564,
      "learning_rate": 0.00019151772257379424,
      "loss": 23.0,
      "step": 2226
    },
    {
      "epoch": 1.336734693877551,
      "grad_norm": 0.0015102779725566506,
      "learning_rate": 0.0001915101068448943,
      "loss": 23.0,
      "step": 2227
    },
    {
      "epoch": 1.3373349339735894,
      "grad_norm": 0.001040091272443533,
      "learning_rate": 0.00019150248785023406,
      "loss": 23.0,
      "step": 2228
    },
    {
      "epoch": 1.337935174069628,
      "grad_norm": 0.0003573865396901965,
      "learning_rate": 0.00019149486559008535,
      "loss": 23.0,
      "step": 2229
    },
    {
      "epoch": 1.3385354141656662,
      "grad_norm": 0.00047850568080320954,
      "learning_rate": 0.0001914872400647203,
      "loss": 23.0,
      "step": 2230
    },
    {
      "epoch": 1.3391356542617046,
      "grad_norm": 0.0013392203254625201,
      "learning_rate": 0.00019147961127441093,
      "loss": 23.0,
      "step": 2231
    },
    {
      "epoch": 1.339735894357743,
      "grad_norm": 0.001735447091050446,
      "learning_rate": 0.00019147197921942958,
      "loss": 23.0,
      "step": 2232
    },
    {
      "epoch": 1.3403361344537816,
      "grad_norm": 0.001040574279613793,
      "learning_rate": 0.0001914643439000486,
      "loss": 23.0,
      "step": 2233
    },
    {
      "epoch": 1.34093637454982,
      "grad_norm": 0.0008239164599217474,
      "learning_rate": 0.00019145670531654045,
      "loss": 23.0,
      "step": 2234
    },
    {
      "epoch": 1.3415366146458583,
      "grad_norm": 0.0007725712493993342,
      "learning_rate": 0.00019144906346917772,
      "loss": 23.0,
      "step": 2235
    },
    {
      "epoch": 1.3421368547418968,
      "grad_norm": 0.001513920957222581,
      "learning_rate": 0.00019144141835823316,
      "loss": 23.0,
      "step": 2236
    },
    {
      "epoch": 1.3427370948379351,
      "grad_norm": 0.0009453982929699123,
      "learning_rate": 0.00019143376998397962,
      "loss": 23.0,
      "step": 2237
    },
    {
      "epoch": 1.3433373349339737,
      "grad_norm": 0.000519758730661124,
      "learning_rate": 0.00019142611834668997,
      "loss": 23.0,
      "step": 2238
    },
    {
      "epoch": 1.343937575030012,
      "grad_norm": 0.0008691302500665188,
      "learning_rate": 0.00019141846344663738,
      "loss": 23.0,
      "step": 2239
    },
    {
      "epoch": 1.3445378151260505,
      "grad_norm": 0.00117309985216707,
      "learning_rate": 0.00019141080528409495,
      "loss": 23.0,
      "step": 2240
    },
    {
      "epoch": 1.3451380552220888,
      "grad_norm": 0.0016505197854712605,
      "learning_rate": 0.00019140314385933608,
      "loss": 23.0,
      "step": 2241
    },
    {
      "epoch": 1.3457382953181272,
      "grad_norm": 0.001228606328368187,
      "learning_rate": 0.00019139547917263405,
      "loss": 23.0,
      "step": 2242
    },
    {
      "epoch": 1.3463385354141657,
      "grad_norm": 0.0015320234233513474,
      "learning_rate": 0.0001913878112242625,
      "loss": 23.0,
      "step": 2243
    },
    {
      "epoch": 1.346938775510204,
      "grad_norm": 0.0024812333285808563,
      "learning_rate": 0.00019138014001449503,
      "loss": 23.0,
      "step": 2244
    },
    {
      "epoch": 1.3475390156062426,
      "grad_norm": 0.001015107030980289,
      "learning_rate": 0.00019137246554360547,
      "loss": 23.0,
      "step": 2245
    },
    {
      "epoch": 1.3481392557022809,
      "grad_norm": 0.0019902249332517385,
      "learning_rate": 0.0001913647878118676,
      "loss": 23.0,
      "step": 2246
    },
    {
      "epoch": 1.3487394957983194,
      "grad_norm": 0.0008530864142812788,
      "learning_rate": 0.00019135710681955547,
      "loss": 23.0,
      "step": 2247
    },
    {
      "epoch": 1.3493397358943577,
      "grad_norm": 0.0009168292162939906,
      "learning_rate": 0.00019134942256694322,
      "loss": 23.0,
      "step": 2248
    },
    {
      "epoch": 1.349939975990396,
      "grad_norm": 0.0006323324050754309,
      "learning_rate": 0.00019134173505430506,
      "loss": 23.0,
      "step": 2249
    },
    {
      "epoch": 1.3505402160864346,
      "grad_norm": 0.002496844856068492,
      "learning_rate": 0.00019133404428191533,
      "loss": 23.0,
      "step": 2250
    },
    {
      "epoch": 1.351140456182473,
      "grad_norm": 0.0013751224614679813,
      "learning_rate": 0.0001913263502500485,
      "loss": 23.0,
      "step": 2251
    },
    {
      "epoch": 1.3517406962785115,
      "grad_norm": 0.0009248317801393569,
      "learning_rate": 0.00019131865295897912,
      "loss": 23.0,
      "step": 2252
    },
    {
      "epoch": 1.3523409363745498,
      "grad_norm": 0.0009484643815085292,
      "learning_rate": 0.00019131095240898196,
      "loss": 23.0,
      "step": 2253
    },
    {
      "epoch": 1.3529411764705883,
      "grad_norm": 0.00116802251432091,
      "learning_rate": 0.00019130324860033176,
      "loss": 23.0,
      "step": 2254
    },
    {
      "epoch": 1.3535414165666266,
      "grad_norm": 0.000703223398886621,
      "learning_rate": 0.0001912955415333035,
      "loss": 23.0,
      "step": 2255
    },
    {
      "epoch": 1.354141656662665,
      "grad_norm": 0.0020953151397407055,
      "learning_rate": 0.00019128783120817214,
      "loss": 23.0,
      "step": 2256
    },
    {
      "epoch": 1.3547418967587035,
      "grad_norm": 0.0014630835503339767,
      "learning_rate": 0.00019128011762521298,
      "loss": 23.0,
      "step": 2257
    },
    {
      "epoch": 1.355342136854742,
      "grad_norm": 0.002706082770600915,
      "learning_rate": 0.0001912724007847012,
      "loss": 23.0,
      "step": 2258
    },
    {
      "epoch": 1.3559423769507803,
      "grad_norm": 0.0006396888638846576,
      "learning_rate": 0.00019126468068691218,
      "loss": 23.0,
      "step": 2259
    },
    {
      "epoch": 1.3565426170468187,
      "grad_norm": 0.0011111192870885134,
      "learning_rate": 0.0001912569573321215,
      "loss": 23.0,
      "step": 2260
    },
    {
      "epoch": 1.3571428571428572,
      "grad_norm": 0.001541930716484785,
      "learning_rate": 0.00019124923072060476,
      "loss": 23.0,
      "step": 2261
    },
    {
      "epoch": 1.3577430972388955,
      "grad_norm": 0.0007236685487441719,
      "learning_rate": 0.0001912415008526377,
      "loss": 23.0,
      "step": 2262
    },
    {
      "epoch": 1.3583433373349338,
      "grad_norm": 0.0009044542675837874,
      "learning_rate": 0.00019123376772849614,
      "loss": 23.0,
      "step": 2263
    },
    {
      "epoch": 1.3589435774309724,
      "grad_norm": 0.0010049508418887854,
      "learning_rate": 0.00019122603134845613,
      "loss": 23.0,
      "step": 2264
    },
    {
      "epoch": 1.359543817527011,
      "grad_norm": 0.000888896465767175,
      "learning_rate": 0.00019121829171279368,
      "loss": 23.0,
      "step": 2265
    },
    {
      "epoch": 1.3601440576230492,
      "grad_norm": 0.0020322557538747787,
      "learning_rate": 0.0001912105488217851,
      "loss": 23.0,
      "step": 2266
    },
    {
      "epoch": 1.3607442977190876,
      "grad_norm": 0.0015879786806181073,
      "learning_rate": 0.0001912028026757066,
      "loss": 23.0,
      "step": 2267
    },
    {
      "epoch": 1.361344537815126,
      "grad_norm": 0.001246255007572472,
      "learning_rate": 0.0001911950532748347,
      "loss": 23.0,
      "step": 2268
    },
    {
      "epoch": 1.3619447779111644,
      "grad_norm": 0.0009220580104738474,
      "learning_rate": 0.00019118730061944592,
      "loss": 23.0,
      "step": 2269
    },
    {
      "epoch": 1.362545018007203,
      "grad_norm": 0.0006278387154452503,
      "learning_rate": 0.00019117954470981697,
      "loss": 23.0,
      "step": 2270
    },
    {
      "epoch": 1.3631452581032413,
      "grad_norm": 0.001516466960310936,
      "learning_rate": 0.0001911717855462246,
      "loss": 23.0,
      "step": 2271
    },
    {
      "epoch": 1.3637454981992798,
      "grad_norm": 0.000532143865711987,
      "learning_rate": 0.0001911640231289457,
      "loss": 23.0,
      "step": 2272
    },
    {
      "epoch": 1.3643457382953181,
      "grad_norm": 0.0007709377678111196,
      "learning_rate": 0.00019115625745825732,
      "loss": 23.0,
      "step": 2273
    },
    {
      "epoch": 1.3649459783913565,
      "grad_norm": 0.0011233460390940309,
      "learning_rate": 0.0001911484885344366,
      "loss": 23.0,
      "step": 2274
    },
    {
      "epoch": 1.365546218487395,
      "grad_norm": 0.001685342169366777,
      "learning_rate": 0.0001911407163577608,
      "loss": 23.0,
      "step": 2275
    },
    {
      "epoch": 1.3661464585834333,
      "grad_norm": 0.0012172020506113768,
      "learning_rate": 0.00019113294092850725,
      "loss": 23.0,
      "step": 2276
    },
    {
      "epoch": 1.3667466986794718,
      "grad_norm": 0.0018125694477930665,
      "learning_rate": 0.00019112516224695347,
      "loss": 23.0,
      "step": 2277
    },
    {
      "epoch": 1.3673469387755102,
      "grad_norm": 0.0003331191255711019,
      "learning_rate": 0.0001911173803133771,
      "loss": 23.0,
      "step": 2278
    },
    {
      "epoch": 1.3679471788715487,
      "grad_norm": 0.0006346300942823291,
      "learning_rate": 0.00019110959512805574,
      "loss": 23.0,
      "step": 2279
    },
    {
      "epoch": 1.368547418967587,
      "grad_norm": 0.002104330575093627,
      "learning_rate": 0.0001911018066912673,
      "loss": 23.0,
      "step": 2280
    },
    {
      "epoch": 1.3691476590636253,
      "grad_norm": 0.0028025060892105103,
      "learning_rate": 0.00019109401500328973,
      "loss": 23.0,
      "step": 2281
    },
    {
      "epoch": 1.3697478991596639,
      "grad_norm": 0.0009631137363612652,
      "learning_rate": 0.00019108622006440112,
      "loss": 23.0,
      "step": 2282
    },
    {
      "epoch": 1.3703481392557022,
      "grad_norm": 0.0011770789278671145,
      "learning_rate": 0.0001910784218748796,
      "loss": 23.0,
      "step": 2283
    },
    {
      "epoch": 1.3709483793517407,
      "grad_norm": 0.0006025804905220866,
      "learning_rate": 0.00019107062043500345,
      "loss": 23.0,
      "step": 2284
    },
    {
      "epoch": 1.371548619447779,
      "grad_norm": 0.0008317616302520037,
      "learning_rate": 0.00019106281574505116,
      "loss": 23.0,
      "step": 2285
    },
    {
      "epoch": 1.3721488595438176,
      "grad_norm": 0.0026311827823519707,
      "learning_rate": 0.00019105500780530123,
      "loss": 23.0,
      "step": 2286
    },
    {
      "epoch": 1.372749099639856,
      "grad_norm": 0.0010507861152291298,
      "learning_rate": 0.00019104719661603223,
      "loss": 23.0,
      "step": 2287
    },
    {
      "epoch": 1.3733493397358942,
      "grad_norm": 0.0007436316227540374,
      "learning_rate": 0.00019103938217752303,
      "loss": 23.0,
      "step": 2288
    },
    {
      "epoch": 1.3739495798319328,
      "grad_norm": 0.000857638951856643,
      "learning_rate": 0.00019103156449005245,
      "loss": 23.0,
      "step": 2289
    },
    {
      "epoch": 1.3745498199279713,
      "grad_norm": 0.0014579903800040483,
      "learning_rate": 0.0001910237435538995,
      "loss": 23.0,
      "step": 2290
    },
    {
      "epoch": 1.3751500600240096,
      "grad_norm": 0.0011235218262299895,
      "learning_rate": 0.00019101591936934327,
      "loss": 23.0,
      "step": 2291
    },
    {
      "epoch": 1.375750300120048,
      "grad_norm": 0.0005367260309867561,
      "learning_rate": 0.00019100809193666303,
      "loss": 23.0,
      "step": 2292
    },
    {
      "epoch": 1.3763505402160865,
      "grad_norm": 0.0010245360899716616,
      "learning_rate": 0.00019100026125613804,
      "loss": 23.0,
      "step": 2293
    },
    {
      "epoch": 1.3769507803121248,
      "grad_norm": 0.0010933050652965903,
      "learning_rate": 0.00019099242732804786,
      "loss": 23.0,
      "step": 2294
    },
    {
      "epoch": 1.3775510204081631,
      "grad_norm": 0.0009408457553945482,
      "learning_rate": 0.00019098459015267198,
      "loss": 23.0,
      "step": 2295
    },
    {
      "epoch": 1.3781512605042017,
      "grad_norm": 0.0014276335714384913,
      "learning_rate": 0.00019097674973029012,
      "loss": 23.0,
      "step": 2296
    },
    {
      "epoch": 1.3787515006002402,
      "grad_norm": 0.0010626583825796843,
      "learning_rate": 0.00019096890606118207,
      "loss": 23.0,
      "step": 2297
    },
    {
      "epoch": 1.3793517406962785,
      "grad_norm": 0.003225801046937704,
      "learning_rate": 0.00019096105914562776,
      "loss": 23.0,
      "step": 2298
    },
    {
      "epoch": 1.3799519807923168,
      "grad_norm": 0.0023619416169822216,
      "learning_rate": 0.00019095320898390727,
      "loss": 23.0,
      "step": 2299
    },
    {
      "epoch": 1.3805522208883554,
      "grad_norm": 0.0009626576793380082,
      "learning_rate": 0.0001909453555763007,
      "loss": 23.0,
      "step": 2300
    },
    {
      "epoch": 1.3811524609843937,
      "grad_norm": 0.0005572536610998213,
      "learning_rate": 0.0001909374989230883,
      "loss": 23.0,
      "step": 2301
    },
    {
      "epoch": 1.3817527010804322,
      "grad_norm": 0.0014027627184987068,
      "learning_rate": 0.00019092963902455048,
      "loss": 23.0,
      "step": 2302
    },
    {
      "epoch": 1.3823529411764706,
      "grad_norm": 0.0008851447491906583,
      "learning_rate": 0.00019092177588096777,
      "loss": 23.0,
      "step": 2303
    },
    {
      "epoch": 1.382953181272509,
      "grad_norm": 0.0006001814617775381,
      "learning_rate": 0.00019091390949262075,
      "loss": 23.0,
      "step": 2304
    },
    {
      "epoch": 1.3835534213685474,
      "grad_norm": 0.0013794521801173687,
      "learning_rate": 0.00019090603985979013,
      "loss": 23.0,
      "step": 2305
    },
    {
      "epoch": 1.3841536614645857,
      "grad_norm": 0.0014090888435021043,
      "learning_rate": 0.00019089816698275683,
      "loss": 23.0,
      "step": 2306
    },
    {
      "epoch": 1.3847539015606243,
      "grad_norm": 0.0011951072374358773,
      "learning_rate": 0.00019089029086180178,
      "loss": 23.0,
      "step": 2307
    },
    {
      "epoch": 1.3853541416566626,
      "grad_norm": 0.000998032046481967,
      "learning_rate": 0.00019088241149720599,
      "loss": 23.0,
      "step": 2308
    },
    {
      "epoch": 1.3859543817527011,
      "grad_norm": 0.0004453418077901006,
      "learning_rate": 0.00019087452888925076,
      "loss": 23.0,
      "step": 2309
    },
    {
      "epoch": 1.3865546218487395,
      "grad_norm": 0.0009691806044429541,
      "learning_rate": 0.00019086664303821728,
      "loss": 23.0,
      "step": 2310
    },
    {
      "epoch": 1.387154861944778,
      "grad_norm": 0.0017116405069828033,
      "learning_rate": 0.00019085875394438711,
      "loss": 23.0,
      "step": 2311
    },
    {
      "epoch": 1.3877551020408163,
      "grad_norm": 0.0006677840137854218,
      "learning_rate": 0.00019085086160804173,
      "loss": 23.0,
      "step": 2312
    },
    {
      "epoch": 1.3883553421368546,
      "grad_norm": 0.0009493288234807551,
      "learning_rate": 0.00019084296602946276,
      "loss": 23.0,
      "step": 2313
    },
    {
      "epoch": 1.3889555822328932,
      "grad_norm": 0.00211167661473155,
      "learning_rate": 0.00019083506720893204,
      "loss": 23.0,
      "step": 2314
    },
    {
      "epoch": 1.3895558223289317,
      "grad_norm": 0.001380943926051259,
      "learning_rate": 0.0001908271651467314,
      "loss": 23.0,
      "step": 2315
    },
    {
      "epoch": 1.39015606242497,
      "grad_norm": 0.0011355826864019036,
      "learning_rate": 0.0001908192598431429,
      "loss": 23.0,
      "step": 2316
    },
    {
      "epoch": 1.3907563025210083,
      "grad_norm": 0.001229851390235126,
      "learning_rate": 0.00019081135129844862,
      "loss": 23.0,
      "step": 2317
    },
    {
      "epoch": 1.3913565426170469,
      "grad_norm": 0.0016962314257398248,
      "learning_rate": 0.0001908034395129308,
      "loss": 23.0,
      "step": 2318
    },
    {
      "epoch": 1.3919567827130852,
      "grad_norm": 0.0005502055282704532,
      "learning_rate": 0.00019079552448687183,
      "loss": 23.0,
      "step": 2319
    },
    {
      "epoch": 1.3925570228091235,
      "grad_norm": 0.0024172854609787464,
      "learning_rate": 0.00019078760622055408,
      "loss": 23.0,
      "step": 2320
    },
    {
      "epoch": 1.393157262905162,
      "grad_norm": 0.0010021852795034647,
      "learning_rate": 0.00019077968471426025,
      "loss": 23.0,
      "step": 2321
    },
    {
      "epoch": 1.3937575030012006,
      "grad_norm": 0.002354429569095373,
      "learning_rate": 0.00019077175996827297,
      "loss": 23.0,
      "step": 2322
    },
    {
      "epoch": 1.394357743097239,
      "grad_norm": 0.0012778837699443102,
      "learning_rate": 0.00019076383198287505,
      "loss": 23.0,
      "step": 2323
    },
    {
      "epoch": 1.3949579831932772,
      "grad_norm": 0.0016614958876743913,
      "learning_rate": 0.00019075590075834947,
      "loss": 23.0,
      "step": 2324
    },
    {
      "epoch": 1.3955582232893158,
      "grad_norm": 0.0007704552845098078,
      "learning_rate": 0.00019074796629497923,
      "loss": 23.0,
      "step": 2325
    },
    {
      "epoch": 1.396158463385354,
      "grad_norm": 0.002393638016656041,
      "learning_rate": 0.0001907400285930475,
      "loss": 23.0,
      "step": 2326
    },
    {
      "epoch": 1.3967587034813926,
      "grad_norm": 0.001964516704902053,
      "learning_rate": 0.00019073208765283756,
      "loss": 23.0,
      "step": 2327
    },
    {
      "epoch": 1.397358943577431,
      "grad_norm": 0.0004503531090449542,
      "learning_rate": 0.00019072414347463284,
      "loss": 23.0,
      "step": 2328
    },
    {
      "epoch": 1.3979591836734695,
      "grad_norm": 0.0006227733101695776,
      "learning_rate": 0.00019071619605871677,
      "loss": 23.0,
      "step": 2329
    },
    {
      "epoch": 1.3985594237695078,
      "grad_norm": 0.0005024244310334325,
      "learning_rate": 0.00019070824540537302,
      "loss": 23.0,
      "step": 2330
    },
    {
      "epoch": 1.3991596638655461,
      "grad_norm": 0.0006691348971799016,
      "learning_rate": 0.0001907002915148853,
      "loss": 23.0,
      "step": 2331
    },
    {
      "epoch": 1.3997599039615847,
      "grad_norm": 0.001576990936882794,
      "learning_rate": 0.00019069233438753753,
      "loss": 23.0,
      "step": 2332
    },
    {
      "epoch": 1.400360144057623,
      "grad_norm": 0.0007380506140179932,
      "learning_rate": 0.00019068437402361364,
      "loss": 23.0,
      "step": 2333
    },
    {
      "epoch": 1.4009603841536615,
      "grad_norm": 0.001869592466391623,
      "learning_rate": 0.00019067641042339766,
      "loss": 23.0,
      "step": 2334
    },
    {
      "epoch": 1.4015606242496998,
      "grad_norm": 0.0005642715259455144,
      "learning_rate": 0.00019066844358717387,
      "loss": 23.0,
      "step": 2335
    },
    {
      "epoch": 1.4021608643457384,
      "grad_norm": 0.0023026717826724052,
      "learning_rate": 0.00019066047351522656,
      "loss": 23.0,
      "step": 2336
    },
    {
      "epoch": 1.4027611044417767,
      "grad_norm": 0.0009191056014969945,
      "learning_rate": 0.00019065250020784013,
      "loss": 23.0,
      "step": 2337
    },
    {
      "epoch": 1.403361344537815,
      "grad_norm": 0.0015248807612806559,
      "learning_rate": 0.00019064452366529918,
      "loss": 23.0,
      "step": 2338
    },
    {
      "epoch": 1.4039615846338536,
      "grad_norm": 0.0011346536921337247,
      "learning_rate": 0.0001906365438878883,
      "loss": 23.0,
      "step": 2339
    },
    {
      "epoch": 1.4045618247298919,
      "grad_norm": 0.0008668985683470964,
      "learning_rate": 0.00019062856087589237,
      "loss": 23.0,
      "step": 2340
    },
    {
      "epoch": 1.4051620648259304,
      "grad_norm": 0.0007433283608406782,
      "learning_rate": 0.0001906205746295962,
      "loss": 23.0,
      "step": 2341
    },
    {
      "epoch": 1.4057623049219687,
      "grad_norm": 0.001420574146322906,
      "learning_rate": 0.00019061258514928482,
      "loss": 23.0,
      "step": 2342
    },
    {
      "epoch": 1.4063625450180073,
      "grad_norm": 0.0016676129307597876,
      "learning_rate": 0.00019060459243524337,
      "loss": 23.0,
      "step": 2343
    },
    {
      "epoch": 1.4069627851140456,
      "grad_norm": 0.001237909309566021,
      "learning_rate": 0.0001905965964877571,
      "loss": 23.0,
      "step": 2344
    },
    {
      "epoch": 1.407563025210084,
      "grad_norm": 0.0010071388678625226,
      "learning_rate": 0.00019058859730711132,
      "loss": 23.0,
      "step": 2345
    },
    {
      "epoch": 1.4081632653061225,
      "grad_norm": 0.00047392971464432776,
      "learning_rate": 0.00019058059489359154,
      "loss": 23.0,
      "step": 2346
    },
    {
      "epoch": 1.408763505402161,
      "grad_norm": 0.001075922162272036,
      "learning_rate": 0.00019057258924748326,
      "loss": 23.0,
      "step": 2347
    },
    {
      "epoch": 1.4093637454981993,
      "grad_norm": 0.00468486687168479,
      "learning_rate": 0.00019056458036907233,
      "loss": 23.0,
      "step": 2348
    },
    {
      "epoch": 1.4099639855942376,
      "grad_norm": 0.0004578742664307356,
      "learning_rate": 0.00019055656825864444,
      "loss": 23.0,
      "step": 2349
    },
    {
      "epoch": 1.4105642256902762,
      "grad_norm": 0.0006120658363215625,
      "learning_rate": 0.00019054855291648562,
      "loss": 23.0,
      "step": 2350
    },
    {
      "epoch": 1.4111644657863145,
      "grad_norm": 0.0007080400246195495,
      "learning_rate": 0.00019054053434288182,
      "loss": 23.0,
      "step": 2351
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 0.0007749073556624353,
      "learning_rate": 0.00019053251253811924,
      "loss": 23.0,
      "step": 2352
    },
    {
      "epoch": 1.4123649459783914,
      "grad_norm": 0.0013292570365592837,
      "learning_rate": 0.00019052448750248423,
      "loss": 23.0,
      "step": 2353
    },
    {
      "epoch": 1.41296518607443,
      "grad_norm": 0.0019230906618759036,
      "learning_rate": 0.00019051645923626306,
      "loss": 23.0,
      "step": 2354
    },
    {
      "epoch": 1.4135654261704682,
      "grad_norm": 0.004612374119460583,
      "learning_rate": 0.0001905084277397423,
      "loss": 23.0,
      "step": 2355
    },
    {
      "epoch": 1.4141656662665065,
      "grad_norm": 0.002344324951991439,
      "learning_rate": 0.0001905003930132086,
      "loss": 23.0,
      "step": 2356
    },
    {
      "epoch": 1.414765906362545,
      "grad_norm": 0.0015184414805844426,
      "learning_rate": 0.00019049235505694864,
      "loss": 23.0,
      "step": 2357
    },
    {
      "epoch": 1.4153661464585834,
      "grad_norm": 0.0006440380238927901,
      "learning_rate": 0.00019048431387124932,
      "loss": 23.0,
      "step": 2358
    },
    {
      "epoch": 1.415966386554622,
      "grad_norm": 0.0011973325163125992,
      "learning_rate": 0.00019047626945639757,
      "loss": 23.0,
      "step": 2359
    },
    {
      "epoch": 1.4165666266506602,
      "grad_norm": 0.0014662436442449689,
      "learning_rate": 0.00019046822181268048,
      "loss": 23.0,
      "step": 2360
    },
    {
      "epoch": 1.4171668667466988,
      "grad_norm": 0.0008078878745436668,
      "learning_rate": 0.00019046017094038532,
      "loss": 23.0,
      "step": 2361
    },
    {
      "epoch": 1.417767106842737,
      "grad_norm": 0.0005584595492109656,
      "learning_rate": 0.00019045211683979934,
      "loss": 23.0,
      "step": 2362
    },
    {
      "epoch": 1.4183673469387754,
      "grad_norm": 0.0008244646014645696,
      "learning_rate": 0.00019044405951120993,
      "loss": 23.0,
      "step": 2363
    },
    {
      "epoch": 1.418967587034814,
      "grad_norm": 0.0017755191074684262,
      "learning_rate": 0.00019043599895490475,
      "loss": 23.0,
      "step": 2364
    },
    {
      "epoch": 1.4195678271308523,
      "grad_norm": 0.0010308343917131424,
      "learning_rate": 0.00019042793517117133,
      "loss": 23.0,
      "step": 2365
    },
    {
      "epoch": 1.4201680672268908,
      "grad_norm": 0.0005713171558454633,
      "learning_rate": 0.00019041986816029755,
      "loss": 23.0,
      "step": 2366
    },
    {
      "epoch": 1.4207683073229291,
      "grad_norm": 0.0006858983542770147,
      "learning_rate": 0.00019041179792257128,
      "loss": 23.0,
      "step": 2367
    },
    {
      "epoch": 1.4213685474189677,
      "grad_norm": 0.0008377997437492013,
      "learning_rate": 0.0001904037244582805,
      "loss": 23.0,
      "step": 2368
    },
    {
      "epoch": 1.421968787515006,
      "grad_norm": 0.0011340173659846187,
      "learning_rate": 0.00019039564776771333,
      "loss": 23.0,
      "step": 2369
    },
    {
      "epoch": 1.4225690276110443,
      "grad_norm": 0.002067905617877841,
      "learning_rate": 0.000190387567851158,
      "loss": 23.0,
      "step": 2370
    },
    {
      "epoch": 1.4231692677070829,
      "grad_norm": 0.0007086987025104463,
      "learning_rate": 0.0001903794847089029,
      "loss": 23.0,
      "step": 2371
    },
    {
      "epoch": 1.4237695078031212,
      "grad_norm": 0.0014819285133853555,
      "learning_rate": 0.00019037139834123649,
      "loss": 23.0,
      "step": 2372
    },
    {
      "epoch": 1.4243697478991597,
      "grad_norm": 0.0010032330173999071,
      "learning_rate": 0.0001903633087484473,
      "loss": 23.0,
      "step": 2373
    },
    {
      "epoch": 1.424969987995198,
      "grad_norm": 0.00293930247426033,
      "learning_rate": 0.00019035521593082407,
      "loss": 23.0,
      "step": 2374
    },
    {
      "epoch": 1.4255702280912366,
      "grad_norm": 0.0007761257002130151,
      "learning_rate": 0.00019034711988865564,
      "loss": 23.0,
      "step": 2375
    },
    {
      "epoch": 1.4261704681872749,
      "grad_norm": 0.001004933612421155,
      "learning_rate": 0.00019033902062223085,
      "loss": 23.0,
      "step": 2376
    },
    {
      "epoch": 1.4267707082833132,
      "grad_norm": 0.0011718858731910586,
      "learning_rate": 0.00019033091813183884,
      "loss": 23.0,
      "step": 2377
    },
    {
      "epoch": 1.4273709483793517,
      "grad_norm": 0.0009318561642430723,
      "learning_rate": 0.00019032281241776868,
      "loss": 23.0,
      "step": 2378
    },
    {
      "epoch": 1.4279711884753903,
      "grad_norm": 0.001223216182552278,
      "learning_rate": 0.0001903147034803097,
      "loss": 23.0,
      "step": 2379
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.0014702687039971352,
      "learning_rate": 0.00019030659131975128,
      "loss": 23.0,
      "step": 2380
    },
    {
      "epoch": 1.429171668667467,
      "grad_norm": 0.0015447470359504223,
      "learning_rate": 0.0001902984759363829,
      "loss": 23.0,
      "step": 2381
    },
    {
      "epoch": 1.4297719087635055,
      "grad_norm": 0.0005654869601130486,
      "learning_rate": 0.0001902903573304942,
      "loss": 23.0,
      "step": 2382
    },
    {
      "epoch": 1.4303721488595438,
      "grad_norm": 0.0011179533321410418,
      "learning_rate": 0.0001902822355023749,
      "loss": 23.0,
      "step": 2383
    },
    {
      "epoch": 1.430972388955582,
      "grad_norm": 0.001682326546870172,
      "learning_rate": 0.00019027411045231482,
      "loss": 23.0,
      "step": 2384
    },
    {
      "epoch": 1.4315726290516206,
      "grad_norm": 0.0007868417887948453,
      "learning_rate": 0.000190265982180604,
      "loss": 23.0,
      "step": 2385
    },
    {
      "epoch": 1.4321728691476592,
      "grad_norm": 0.0008736375602893531,
      "learning_rate": 0.00019025785068753245,
      "loss": 23.0,
      "step": 2386
    },
    {
      "epoch": 1.4327731092436975,
      "grad_norm": 0.0012772384798154235,
      "learning_rate": 0.00019024971597339036,
      "loss": 23.0,
      "step": 2387
    },
    {
      "epoch": 1.4333733493397358,
      "grad_norm": 0.0024452575016766787,
      "learning_rate": 0.00019024157803846807,
      "loss": 23.0,
      "step": 2388
    },
    {
      "epoch": 1.4339735894357744,
      "grad_norm": 0.000944132509175688,
      "learning_rate": 0.00019023343688305598,
      "loss": 23.0,
      "step": 2389
    },
    {
      "epoch": 1.4345738295318127,
      "grad_norm": 0.0018963133916258812,
      "learning_rate": 0.00019022529250744465,
      "loss": 23.0,
      "step": 2390
    },
    {
      "epoch": 1.4351740696278512,
      "grad_norm": 0.0011875929776579142,
      "learning_rate": 0.00019021714491192473,
      "loss": 23.0,
      "step": 2391
    },
    {
      "epoch": 1.4357743097238895,
      "grad_norm": 0.0014995178207755089,
      "learning_rate": 0.00019020899409678694,
      "loss": 23.0,
      "step": 2392
    },
    {
      "epoch": 1.436374549819928,
      "grad_norm": 0.0005268016830086708,
      "learning_rate": 0.0001902008400623222,
      "loss": 23.0,
      "step": 2393
    },
    {
      "epoch": 1.4369747899159664,
      "grad_norm": 0.001868417952209711,
      "learning_rate": 0.00019019268280882153,
      "loss": 23.0,
      "step": 2394
    },
    {
      "epoch": 1.4375750300120047,
      "grad_norm": 0.0018561460310593247,
      "learning_rate": 0.000190184522336576,
      "loss": 23.0,
      "step": 2395
    },
    {
      "epoch": 1.4381752701080432,
      "grad_norm": 0.0015242897206917405,
      "learning_rate": 0.00019017635864587685,
      "loss": 23.0,
      "step": 2396
    },
    {
      "epoch": 1.4387755102040816,
      "grad_norm": 0.0020303060300648212,
      "learning_rate": 0.00019016819173701542,
      "loss": 23.0,
      "step": 2397
    },
    {
      "epoch": 1.43937575030012,
      "grad_norm": 0.0019102046499028802,
      "learning_rate": 0.00019016002161028315,
      "loss": 23.0,
      "step": 2398
    },
    {
      "epoch": 1.4399759903961584,
      "grad_norm": 0.0020126604940742254,
      "learning_rate": 0.00019015184826597165,
      "loss": 23.0,
      "step": 2399
    },
    {
      "epoch": 1.440576230492197,
      "grad_norm": 0.0019320209976285696,
      "learning_rate": 0.00019014367170437258,
      "loss": 23.0,
      "step": 2400
    },
    {
      "epoch": 1.4411764705882353,
      "grad_norm": 0.0005183393368497491,
      "learning_rate": 0.00019013549192577775,
      "loss": 23.0,
      "step": 2401
    },
    {
      "epoch": 1.4417767106842736,
      "grad_norm": 0.0018166369991376996,
      "learning_rate": 0.00019012730893047905,
      "loss": 23.0,
      "step": 2402
    },
    {
      "epoch": 1.4423769507803121,
      "grad_norm": 0.001152240438386798,
      "learning_rate": 0.00019011912271876854,
      "loss": 23.0,
      "step": 2403
    },
    {
      "epoch": 1.4429771908763505,
      "grad_norm": 0.0012412526411935687,
      "learning_rate": 0.00019011093329093835,
      "loss": 23.0,
      "step": 2404
    },
    {
      "epoch": 1.443577430972389,
      "grad_norm": 0.0005008511943742633,
      "learning_rate": 0.00019010274064728075,
      "loss": 23.0,
      "step": 2405
    },
    {
      "epoch": 1.4441776710684273,
      "grad_norm": 0.0009910052176564932,
      "learning_rate": 0.0001900945447880881,
      "loss": 23.0,
      "step": 2406
    },
    {
      "epoch": 1.4447779111644659,
      "grad_norm": 0.000912656367290765,
      "learning_rate": 0.00019008634571365292,
      "loss": 23.0,
      "step": 2407
    },
    {
      "epoch": 1.4453781512605042,
      "grad_norm": 0.0028049952816218138,
      "learning_rate": 0.00019007814342426776,
      "loss": 23.0,
      "step": 2408
    },
    {
      "epoch": 1.4459783913565425,
      "grad_norm": 0.0006766768638044596,
      "learning_rate": 0.0001900699379202254,
      "loss": 23.0,
      "step": 2409
    },
    {
      "epoch": 1.446578631452581,
      "grad_norm": 0.0007151064346544445,
      "learning_rate": 0.0001900617292018186,
      "loss": 23.0,
      "step": 2410
    },
    {
      "epoch": 1.4471788715486196,
      "grad_norm": 0.0006462062592618167,
      "learning_rate": 0.00019005351726934043,
      "loss": 23.0,
      "step": 2411
    },
    {
      "epoch": 1.447779111644658,
      "grad_norm": 0.0017846194095909595,
      "learning_rate": 0.0001900453021230838,
      "loss": 23.0,
      "step": 2412
    },
    {
      "epoch": 1.4483793517406962,
      "grad_norm": 0.0005367895937524736,
      "learning_rate": 0.00019003708376334196,
      "loss": 23.0,
      "step": 2413
    },
    {
      "epoch": 1.4489795918367347,
      "grad_norm": 0.0016319414135068655,
      "learning_rate": 0.00019002886219040825,
      "loss": 23.0,
      "step": 2414
    },
    {
      "epoch": 1.449579831932773,
      "grad_norm": 0.0011456223437562585,
      "learning_rate": 0.000190020637404576,
      "loss": 23.0,
      "step": 2415
    },
    {
      "epoch": 1.4501800720288116,
      "grad_norm": 0.0025063680950552225,
      "learning_rate": 0.00019001240940613877,
      "loss": 23.0,
      "step": 2416
    },
    {
      "epoch": 1.45078031212485,
      "grad_norm": 0.0013597876532003284,
      "learning_rate": 0.0001900041781953902,
      "loss": 23.0,
      "step": 2417
    },
    {
      "epoch": 1.4513805522208885,
      "grad_norm": 0.0005757425678893924,
      "learning_rate": 0.00018999594377262403,
      "loss": 23.0,
      "step": 2418
    },
    {
      "epoch": 1.4519807923169268,
      "grad_norm": 0.000643598148599267,
      "learning_rate": 0.0001899877061381341,
      "loss": 23.0,
      "step": 2419
    },
    {
      "epoch": 1.452581032412965,
      "grad_norm": 0.0012208611005917192,
      "learning_rate": 0.00018997946529221443,
      "loss": 23.0,
      "step": 2420
    },
    {
      "epoch": 1.4531812725090036,
      "grad_norm": 0.000552306417375803,
      "learning_rate": 0.00018997122123515909,
      "loss": 23.0,
      "step": 2421
    },
    {
      "epoch": 1.453781512605042,
      "grad_norm": 0.0013741564471274614,
      "learning_rate": 0.0001899629739672623,
      "loss": 23.0,
      "step": 2422
    },
    {
      "epoch": 1.4543817527010805,
      "grad_norm": 0.0015043517341837287,
      "learning_rate": 0.00018995472348881838,
      "loss": 23.0,
      "step": 2423
    },
    {
      "epoch": 1.4549819927971188,
      "grad_norm": 0.0006686780834570527,
      "learning_rate": 0.0001899464698001218,
      "loss": 23.0,
      "step": 2424
    },
    {
      "epoch": 1.4555822328931574,
      "grad_norm": 0.0020055018831044436,
      "learning_rate": 0.00018993821290146703,
      "loss": 23.0,
      "step": 2425
    },
    {
      "epoch": 1.4561824729891957,
      "grad_norm": 0.000782818766310811,
      "learning_rate": 0.00018992995279314883,
      "loss": 23.0,
      "step": 2426
    },
    {
      "epoch": 1.456782713085234,
      "grad_norm": 0.000715934787876904,
      "learning_rate": 0.00018992168947546193,
      "loss": 23.0,
      "step": 2427
    },
    {
      "epoch": 1.4573829531812725,
      "grad_norm": 0.002988057676702738,
      "learning_rate": 0.00018991342294870123,
      "loss": 23.0,
      "step": 2428
    },
    {
      "epoch": 1.4579831932773109,
      "grad_norm": 0.0012659791391342878,
      "learning_rate": 0.00018990515321316175,
      "loss": 23.0,
      "step": 2429
    },
    {
      "epoch": 1.4585834333733494,
      "grad_norm": 0.002724630059674382,
      "learning_rate": 0.00018989688026913864,
      "loss": 23.0,
      "step": 2430
    },
    {
      "epoch": 1.4591836734693877,
      "grad_norm": 0.0005784712266176939,
      "learning_rate": 0.0001898886041169271,
      "loss": 23.0,
      "step": 2431
    },
    {
      "epoch": 1.4597839135654262,
      "grad_norm": 0.0010768299689516425,
      "learning_rate": 0.00018988032475682247,
      "loss": 23.0,
      "step": 2432
    },
    {
      "epoch": 1.4603841536614646,
      "grad_norm": 0.0011558542028069496,
      "learning_rate": 0.00018987204218912028,
      "loss": 23.0,
      "step": 2433
    },
    {
      "epoch": 1.4609843937575029,
      "grad_norm": 0.0020028555300086737,
      "learning_rate": 0.0001898637564141161,
      "loss": 23.0,
      "step": 2434
    },
    {
      "epoch": 1.4615846338535414,
      "grad_norm": 0.0011177972191944718,
      "learning_rate": 0.00018985546743210559,
      "loss": 23.0,
      "step": 2435
    },
    {
      "epoch": 1.46218487394958,
      "grad_norm": 0.0016774655086919665,
      "learning_rate": 0.00018984717524338457,
      "loss": 23.0,
      "step": 2436
    },
    {
      "epoch": 1.4627851140456183,
      "grad_norm": 0.001039148191921413,
      "learning_rate": 0.00018983887984824898,
      "loss": 23.0,
      "step": 2437
    },
    {
      "epoch": 1.4633853541416566,
      "grad_norm": 0.0009601426427252591,
      "learning_rate": 0.0001898305812469949,
      "loss": 23.0,
      "step": 2438
    },
    {
      "epoch": 1.4639855942376951,
      "grad_norm": 0.00051392603199929,
      "learning_rate": 0.0001898222794399184,
      "loss": 23.0,
      "step": 2439
    },
    {
      "epoch": 1.4645858343337335,
      "grad_norm": 0.0011603968450799584,
      "learning_rate": 0.00018981397442731582,
      "loss": 23.0,
      "step": 2440
    },
    {
      "epoch": 1.4651860744297718,
      "grad_norm": 0.0007191347540356219,
      "learning_rate": 0.00018980566620948355,
      "loss": 23.0,
      "step": 2441
    },
    {
      "epoch": 1.4657863145258103,
      "grad_norm": 0.002068307949230075,
      "learning_rate": 0.000189797354786718,
      "loss": 23.0,
      "step": 2442
    },
    {
      "epoch": 1.4663865546218489,
      "grad_norm": 0.0010287434561178088,
      "learning_rate": 0.0001897890401593159,
      "loss": 23.0,
      "step": 2443
    },
    {
      "epoch": 1.4669867947178872,
      "grad_norm": 0.000688788655679673,
      "learning_rate": 0.00018978072232757388,
      "loss": 23.0,
      "step": 2444
    },
    {
      "epoch": 1.4675870348139255,
      "grad_norm": 0.0016002441989257932,
      "learning_rate": 0.00018977240129178883,
      "loss": 23.0,
      "step": 2445
    },
    {
      "epoch": 1.468187274909964,
      "grad_norm": 0.0014760290505364537,
      "learning_rate": 0.00018976407705225773,
      "loss": 23.0,
      "step": 2446
    },
    {
      "epoch": 1.4687875150060024,
      "grad_norm": 0.0016490549314767122,
      "learning_rate": 0.0001897557496092776,
      "loss": 23.0,
      "step": 2447
    },
    {
      "epoch": 1.469387755102041,
      "grad_norm": 0.0013884782092645764,
      "learning_rate": 0.00018974741896314565,
      "loss": 23.0,
      "step": 2448
    },
    {
      "epoch": 1.4699879951980792,
      "grad_norm": 0.0020489206071943045,
      "learning_rate": 0.00018973908511415918,
      "loss": 23.0,
      "step": 2449
    },
    {
      "epoch": 1.4705882352941178,
      "grad_norm": 0.0006343182176351547,
      "learning_rate": 0.00018973074806261558,
      "loss": 23.0,
      "step": 2450
    },
    {
      "epoch": 1.471188475390156,
      "grad_norm": 0.002966964850202203,
      "learning_rate": 0.0001897224078088124,
      "loss": 23.0,
      "step": 2451
    },
    {
      "epoch": 1.4717887154861944,
      "grad_norm": 0.0014962563291192055,
      "learning_rate": 0.0001897140643530473,
      "loss": 23.0,
      "step": 2452
    },
    {
      "epoch": 1.472388955582233,
      "grad_norm": 0.001116880332119763,
      "learning_rate": 0.000189705717695618,
      "loss": 23.0,
      "step": 2453
    },
    {
      "epoch": 1.4729891956782712,
      "grad_norm": 0.001895223744213581,
      "learning_rate": 0.0001896973678368224,
      "loss": 23.0,
      "step": 2454
    },
    {
      "epoch": 1.4735894357743098,
      "grad_norm": 0.0010942331282421947,
      "learning_rate": 0.00018968901477695846,
      "loss": 23.0,
      "step": 2455
    },
    {
      "epoch": 1.474189675870348,
      "grad_norm": 0.0024703119415789843,
      "learning_rate": 0.00018968065851632427,
      "loss": 23.0,
      "step": 2456
    },
    {
      "epoch": 1.4747899159663866,
      "grad_norm": 0.0033121961168944836,
      "learning_rate": 0.0001896722990552181,
      "loss": 23.0,
      "step": 2457
    },
    {
      "epoch": 1.475390156062425,
      "grad_norm": 0.0023775578010827303,
      "learning_rate": 0.00018966393639393824,
      "loss": 23.0,
      "step": 2458
    },
    {
      "epoch": 1.4759903961584633,
      "grad_norm": 0.0013918614713475108,
      "learning_rate": 0.0001896555705327831,
      "loss": 23.0,
      "step": 2459
    },
    {
      "epoch": 1.4765906362545018,
      "grad_norm": 0.0009230630821548402,
      "learning_rate": 0.00018964720147205132,
      "loss": 23.0,
      "step": 2460
    },
    {
      "epoch": 1.4771908763505401,
      "grad_norm": 0.0010851762490347028,
      "learning_rate": 0.00018963882921204145,
      "loss": 23.0,
      "step": 2461
    },
    {
      "epoch": 1.4777911164465787,
      "grad_norm": 0.0005464069545269012,
      "learning_rate": 0.0001896304537530524,
      "loss": 23.0,
      "step": 2462
    },
    {
      "epoch": 1.478391356542617,
      "grad_norm": 0.0014860890805721283,
      "learning_rate": 0.00018962207509538298,
      "loss": 23.0,
      "step": 2463
    },
    {
      "epoch": 1.4789915966386555,
      "grad_norm": 0.0019383092876523733,
      "learning_rate": 0.00018961369323933225,
      "loss": 23.0,
      "step": 2464
    },
    {
      "epoch": 1.4795918367346939,
      "grad_norm": 0.002337732119485736,
      "learning_rate": 0.00018960530818519934,
      "loss": 23.0,
      "step": 2465
    },
    {
      "epoch": 1.4801920768307322,
      "grad_norm": 0.0012965687783434987,
      "learning_rate": 0.00018959691993328343,
      "loss": 23.0,
      "step": 2466
    },
    {
      "epoch": 1.4807923169267707,
      "grad_norm": 0.0022145984694361687,
      "learning_rate": 0.00018958852848388396,
      "loss": 23.0,
      "step": 2467
    },
    {
      "epoch": 1.4813925570228093,
      "grad_norm": 0.0013813393888995051,
      "learning_rate": 0.00018958013383730036,
      "loss": 23.0,
      "step": 2468
    },
    {
      "epoch": 1.4819927971188476,
      "grad_norm": 0.0012145678047090769,
      "learning_rate": 0.00018957173599383218,
      "loss": 23.0,
      "step": 2469
    },
    {
      "epoch": 1.482593037214886,
      "grad_norm": 0.0006841164431534708,
      "learning_rate": 0.00018956333495377914,
      "loss": 23.0,
      "step": 2470
    },
    {
      "epoch": 1.4831932773109244,
      "grad_norm": 0.002038002945482731,
      "learning_rate": 0.00018955493071744109,
      "loss": 23.0,
      "step": 2471
    },
    {
      "epoch": 1.4837935174069627,
      "grad_norm": 0.0006899077561683953,
      "learning_rate": 0.00018954652328511792,
      "loss": 23.0,
      "step": 2472
    },
    {
      "epoch": 1.484393757503001,
      "grad_norm": 0.0032006518449634314,
      "learning_rate": 0.0001895381126571097,
      "loss": 23.0,
      "step": 2473
    },
    {
      "epoch": 1.4849939975990396,
      "grad_norm": 0.0013399770250543952,
      "learning_rate": 0.0001895296988337165,
      "loss": 23.0,
      "step": 2474
    },
    {
      "epoch": 1.4855942376950781,
      "grad_norm": 0.001352188060991466,
      "learning_rate": 0.00018952128181523868,
      "loss": 23.0,
      "step": 2475
    },
    {
      "epoch": 1.4861944777911165,
      "grad_norm": 0.0017339931800961494,
      "learning_rate": 0.00018951286160197656,
      "loss": 23.0,
      "step": 2476
    },
    {
      "epoch": 1.4867947178871548,
      "grad_norm": 0.0018044788157567382,
      "learning_rate": 0.00018950443819423072,
      "loss": 23.0,
      "step": 2477
    },
    {
      "epoch": 1.4873949579831933,
      "grad_norm": 0.002194973872974515,
      "learning_rate": 0.00018949601159230168,
      "loss": 23.0,
      "step": 2478
    },
    {
      "epoch": 1.4879951980792316,
      "grad_norm": 0.0009345563594251871,
      "learning_rate": 0.0001894875817964902,
      "loss": 23.0,
      "step": 2479
    },
    {
      "epoch": 1.4885954381752702,
      "grad_norm": 0.0019275409867987037,
      "learning_rate": 0.0001894791488070971,
      "loss": 23.0,
      "step": 2480
    },
    {
      "epoch": 1.4891956782713085,
      "grad_norm": 0.0015987679362297058,
      "learning_rate": 0.00018947071262442338,
      "loss": 23.0,
      "step": 2481
    },
    {
      "epoch": 1.489795918367347,
      "grad_norm": 0.0010282109724357724,
      "learning_rate": 0.00018946227324877004,
      "loss": 23.0,
      "step": 2482
    },
    {
      "epoch": 1.4903961584633854,
      "grad_norm": 0.0021582935005426407,
      "learning_rate": 0.00018945383068043832,
      "loss": 23.0,
      "step": 2483
    },
    {
      "epoch": 1.4909963985594237,
      "grad_norm": 0.0014552563661709428,
      "learning_rate": 0.00018944538491972947,
      "loss": 23.0,
      "step": 2484
    },
    {
      "epoch": 1.4915966386554622,
      "grad_norm": 0.003442388027906418,
      "learning_rate": 0.00018943693596694492,
      "loss": 23.0,
      "step": 2485
    },
    {
      "epoch": 1.4921968787515005,
      "grad_norm": 0.0013504648813977838,
      "learning_rate": 0.00018942848382238618,
      "loss": 23.0,
      "step": 2486
    },
    {
      "epoch": 1.492797118847539,
      "grad_norm": 0.0007023458019830287,
      "learning_rate": 0.0001894200284863549,
      "loss": 23.0,
      "step": 2487
    },
    {
      "epoch": 1.4933973589435774,
      "grad_norm": 0.0015389977488666773,
      "learning_rate": 0.0001894115699591528,
      "loss": 23.0,
      "step": 2488
    },
    {
      "epoch": 1.493997599039616,
      "grad_norm": 0.0009908635402098298,
      "learning_rate": 0.00018940310824108175,
      "loss": 23.0,
      "step": 2489
    },
    {
      "epoch": 1.4945978391356542,
      "grad_norm": 0.0017589518101885915,
      "learning_rate": 0.00018939464333244378,
      "loss": 23.0,
      "step": 2490
    },
    {
      "epoch": 1.4951980792316926,
      "grad_norm": 0.0017695450223982334,
      "learning_rate": 0.00018938617523354089,
      "loss": 23.0,
      "step": 2491
    },
    {
      "epoch": 1.495798319327731,
      "grad_norm": 0.0020960380788892508,
      "learning_rate": 0.0001893777039446754,
      "loss": 23.0,
      "step": 2492
    },
    {
      "epoch": 1.4963985594237694,
      "grad_norm": 0.0016196799697354436,
      "learning_rate": 0.0001893692294661495,
      "loss": 23.0,
      "step": 2493
    },
    {
      "epoch": 1.496998799519808,
      "grad_norm": 0.0021666022948920727,
      "learning_rate": 0.00018936075179826572,
      "loss": 23.0,
      "step": 2494
    },
    {
      "epoch": 1.4975990396158463,
      "grad_norm": 0.0011034191120415926,
      "learning_rate": 0.00018935227094132658,
      "loss": 23.0,
      "step": 2495
    },
    {
      "epoch": 1.4981992797118848,
      "grad_norm": 0.0014189224457368255,
      "learning_rate": 0.0001893437868956347,
      "loss": 23.0,
      "step": 2496
    },
    {
      "epoch": 1.4987995198079231,
      "grad_norm": 0.0015656520845368505,
      "learning_rate": 0.00018933529966149291,
      "loss": 23.0,
      "step": 2497
    },
    {
      "epoch": 1.4993997599039615,
      "grad_norm": 0.002596880542114377,
      "learning_rate": 0.00018932680923920407,
      "loss": 23.0,
      "step": 2498
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.001045036013238132,
      "learning_rate": 0.00018931831562907117,
      "loss": 23.0,
      "step": 2499
    },
    {
      "epoch": 1.5006002400960385,
      "grad_norm": 0.0023872291203588247,
      "learning_rate": 0.00018930981883139735,
      "loss": 23.0,
      "step": 2500
    },
    {
      "epoch": 1.5012004801920769,
      "grad_norm": 0.0013668040046468377,
      "learning_rate": 0.00018930131884648584,
      "loss": 23.0,
      "step": 2501
    },
    {
      "epoch": 1.5018007202881152,
      "grad_norm": 0.002747413469478488,
      "learning_rate": 0.00018929281567463995,
      "loss": 23.0,
      "step": 2502
    },
    {
      "epoch": 1.5024009603841537,
      "grad_norm": 0.001867507235147059,
      "learning_rate": 0.00018928430931616316,
      "loss": 23.0,
      "step": 2503
    },
    {
      "epoch": 1.503001200480192,
      "grad_norm": 0.0007478939369320869,
      "learning_rate": 0.00018927579977135906,
      "loss": 23.0,
      "step": 2504
    },
    {
      "epoch": 1.5036014405762304,
      "grad_norm": 0.0014916093787178397,
      "learning_rate": 0.0001892672870405313,
      "loss": 23.0,
      "step": 2505
    },
    {
      "epoch": 1.504201680672269,
      "grad_norm": 0.0011776410974562168,
      "learning_rate": 0.00018925877112398366,
      "loss": 23.0,
      "step": 2506
    },
    {
      "epoch": 1.5048019207683074,
      "grad_norm": 0.0019988473504781723,
      "learning_rate": 0.0001892502520220201,
      "loss": 23.0,
      "step": 2507
    },
    {
      "epoch": 1.5054021608643458,
      "grad_norm": 0.0010818892624229193,
      "learning_rate": 0.00018924172973494463,
      "loss": 23.0,
      "step": 2508
    },
    {
      "epoch": 1.506002400960384,
      "grad_norm": 0.0022928828839212656,
      "learning_rate": 0.0001892332042630614,
      "loss": 23.0,
      "step": 2509
    },
    {
      "epoch": 1.5066026410564226,
      "grad_norm": 0.001313739106990397,
      "learning_rate": 0.00018922467560667457,
      "loss": 23.0,
      "step": 2510
    },
    {
      "epoch": 1.5072028811524611,
      "grad_norm": 0.0003109007375314832,
      "learning_rate": 0.00018921614376608866,
      "loss": 23.0,
      "step": 2511
    },
    {
      "epoch": 1.5078031212484992,
      "grad_norm": 0.0010564455296844244,
      "learning_rate": 0.00018920760874160806,
      "loss": 23.0,
      "step": 2512
    },
    {
      "epoch": 1.5084033613445378,
      "grad_norm": 0.0011524189030751586,
      "learning_rate": 0.00018919907053353735,
      "loss": 23.0,
      "step": 2513
    },
    {
      "epoch": 1.5090036014405763,
      "grad_norm": 0.0017336489399895072,
      "learning_rate": 0.00018919052914218124,
      "loss": 23.0,
      "step": 2514
    },
    {
      "epoch": 1.5096038415366146,
      "grad_norm": 0.0009461884619668126,
      "learning_rate": 0.00018918198456784464,
      "loss": 23.0,
      "step": 2515
    },
    {
      "epoch": 1.510204081632653,
      "grad_norm": 0.0022306712344288826,
      "learning_rate": 0.00018917343681083235,
      "loss": 23.0,
      "step": 2516
    },
    {
      "epoch": 1.5108043217286915,
      "grad_norm": 0.0007099023787304759,
      "learning_rate": 0.0001891648858714495,
      "loss": 23.0,
      "step": 2517
    },
    {
      "epoch": 1.51140456182473,
      "grad_norm": 0.0007101186783984303,
      "learning_rate": 0.00018915633175000125,
      "loss": 23.0,
      "step": 2518
    },
    {
      "epoch": 1.5120048019207684,
      "grad_norm": 0.000460692128399387,
      "learning_rate": 0.00018914777444679282,
      "loss": 23.0,
      "step": 2519
    },
    {
      "epoch": 1.5126050420168067,
      "grad_norm": 0.0009436405962333083,
      "learning_rate": 0.00018913921396212968,
      "loss": 23.0,
      "step": 2520
    },
    {
      "epoch": 1.5132052821128452,
      "grad_norm": 0.0034887429792433977,
      "learning_rate": 0.00018913065029631726,
      "loss": 23.0,
      "step": 2521
    },
    {
      "epoch": 1.5138055222088835,
      "grad_norm": 0.001372122555039823,
      "learning_rate": 0.0001891220834496612,
      "loss": 23.0,
      "step": 2522
    },
    {
      "epoch": 1.5144057623049219,
      "grad_norm": 0.0013257608516141772,
      "learning_rate": 0.00018911351342246725,
      "loss": 23.0,
      "step": 2523
    },
    {
      "epoch": 1.5150060024009604,
      "grad_norm": 0.0008941548294387758,
      "learning_rate": 0.00018910494021504124,
      "loss": 23.0,
      "step": 2524
    },
    {
      "epoch": 1.515606242496999,
      "grad_norm": 0.0010371002135798335,
      "learning_rate": 0.0001890963638276891,
      "loss": 23.0,
      "step": 2525
    },
    {
      "epoch": 1.5162064825930373,
      "grad_norm": 0.002510013757273555,
      "learning_rate": 0.00018908778426071692,
      "loss": 23.0,
      "step": 2526
    },
    {
      "epoch": 1.5168067226890756,
      "grad_norm": 0.0019762797746807337,
      "learning_rate": 0.00018907920151443088,
      "loss": 23.0,
      "step": 2527
    },
    {
      "epoch": 1.517406962785114,
      "grad_norm": 0.0006245961994864047,
      "learning_rate": 0.00018907061558913726,
      "loss": 23.0,
      "step": 2528
    },
    {
      "epoch": 1.5180072028811524,
      "grad_norm": 0.001987320836633444,
      "learning_rate": 0.00018906202648514252,
      "loss": 23.0,
      "step": 2529
    },
    {
      "epoch": 1.5186074429771907,
      "grad_norm": 0.0013297735713422298,
      "learning_rate": 0.00018905343420275313,
      "loss": 23.0,
      "step": 2530
    },
    {
      "epoch": 1.5192076830732293,
      "grad_norm": 0.0017429828876629472,
      "learning_rate": 0.00018904483874227578,
      "loss": 23.0,
      "step": 2531
    },
    {
      "epoch": 1.5198079231692678,
      "grad_norm": 0.0008359710918739438,
      "learning_rate": 0.00018903624010401714,
      "loss": 23.0,
      "step": 2532
    },
    {
      "epoch": 1.5204081632653061,
      "grad_norm": 0.000509335775859654,
      "learning_rate": 0.0001890276382882841,
      "loss": 23.0,
      "step": 2533
    },
    {
      "epoch": 1.5210084033613445,
      "grad_norm": 0.0008946338784880936,
      "learning_rate": 0.00018901903329538372,
      "loss": 23.0,
      "step": 2534
    },
    {
      "epoch": 1.521608643457383,
      "grad_norm": 0.0009605719242244959,
      "learning_rate": 0.000189010425125623,
      "loss": 23.0,
      "step": 2535
    },
    {
      "epoch": 1.5222088835534213,
      "grad_norm": 0.0011078940005972981,
      "learning_rate": 0.00018900181377930916,
      "loss": 23.0,
      "step": 2536
    },
    {
      "epoch": 1.5228091236494596,
      "grad_norm": 0.0016285383608192205,
      "learning_rate": 0.0001889931992567495,
      "loss": 23.0,
      "step": 2537
    },
    {
      "epoch": 1.5234093637454982,
      "grad_norm": 0.0011555464006960392,
      "learning_rate": 0.00018898458155825154,
      "loss": 23.0,
      "step": 2538
    },
    {
      "epoch": 1.5240096038415367,
      "grad_norm": 0.0008873185142874718,
      "learning_rate": 0.00018897596068412272,
      "loss": 23.0,
      "step": 2539
    },
    {
      "epoch": 1.524609843937575,
      "grad_norm": 0.000611953844781965,
      "learning_rate": 0.00018896733663467075,
      "loss": 23.0,
      "step": 2540
    },
    {
      "epoch": 1.5252100840336134,
      "grad_norm": 0.001338244299404323,
      "learning_rate": 0.00018895870941020338,
      "loss": 23.0,
      "step": 2541
    },
    {
      "epoch": 1.525810324129652,
      "grad_norm": 0.0021093355026096106,
      "learning_rate": 0.00018895007901102853,
      "loss": 23.0,
      "step": 2542
    },
    {
      "epoch": 1.5264105642256904,
      "grad_norm": 0.0007958555361256003,
      "learning_rate": 0.0001889414454374541,
      "loss": 23.0,
      "step": 2543
    },
    {
      "epoch": 1.5270108043217285,
      "grad_norm": 0.001077108783647418,
      "learning_rate": 0.00018893280868978834,
      "loss": 23.0,
      "step": 2544
    },
    {
      "epoch": 1.527611044417767,
      "grad_norm": 0.003100859234109521,
      "learning_rate": 0.00018892416876833937,
      "loss": 23.0,
      "step": 2545
    },
    {
      "epoch": 1.5282112845138056,
      "grad_norm": 0.0009456438128836453,
      "learning_rate": 0.00018891552567341555,
      "loss": 23.0,
      "step": 2546
    },
    {
      "epoch": 1.528811524609844,
      "grad_norm": 0.0015903826570138335,
      "learning_rate": 0.00018890687940532533,
      "loss": 23.0,
      "step": 2547
    },
    {
      "epoch": 1.5294117647058822,
      "grad_norm": 0.0013777603162452579,
      "learning_rate": 0.0001888982299643773,
      "loss": 23.0,
      "step": 2548
    },
    {
      "epoch": 1.5300120048019208,
      "grad_norm": 0.002740132389590144,
      "learning_rate": 0.00018888957735088008,
      "loss": 23.0,
      "step": 2549
    },
    {
      "epoch": 1.5306122448979593,
      "grad_norm": 0.003908548038452864,
      "learning_rate": 0.00018888092156514255,
      "loss": 23.0,
      "step": 2550
    },
    {
      "epoch": 1.5312124849939976,
      "grad_norm": 0.0005045512807555497,
      "learning_rate": 0.00018887226260747353,
      "loss": 23.0,
      "step": 2551
    },
    {
      "epoch": 1.531812725090036,
      "grad_norm": 0.0009278635843656957,
      "learning_rate": 0.00018886360047818207,
      "loss": 23.0,
      "step": 2552
    },
    {
      "epoch": 1.5324129651860745,
      "grad_norm": 0.0005404682597145438,
      "learning_rate": 0.0001888549351775773,
      "loss": 23.0,
      "step": 2553
    },
    {
      "epoch": 1.5330132052821128,
      "grad_norm": 0.0007404520874843001,
      "learning_rate": 0.0001888462667059684,
      "loss": 23.0,
      "step": 2554
    },
    {
      "epoch": 1.5336134453781511,
      "grad_norm": 0.0019501621136441827,
      "learning_rate": 0.00018883759506366484,
      "loss": 23.0,
      "step": 2555
    },
    {
      "epoch": 1.5342136854741897,
      "grad_norm": 0.0011888896115124226,
      "learning_rate": 0.00018882892025097602,
      "loss": 23.0,
      "step": 2556
    },
    {
      "epoch": 1.5348139255702282,
      "grad_norm": 0.0018972912803292274,
      "learning_rate": 0.00018882024226821154,
      "loss": 23.0,
      "step": 2557
    },
    {
      "epoch": 1.5354141656662665,
      "grad_norm": 0.0017216479172930121,
      "learning_rate": 0.00018881156111568106,
      "loss": 23.0,
      "step": 2558
    },
    {
      "epoch": 1.5360144057623049,
      "grad_norm": 0.0018141475738957524,
      "learning_rate": 0.00018880287679369444,
      "loss": 23.0,
      "step": 2559
    },
    {
      "epoch": 1.5366146458583434,
      "grad_norm": 0.000642350350972265,
      "learning_rate": 0.00018879418930256155,
      "loss": 23.0,
      "step": 2560
    },
    {
      "epoch": 1.5372148859543817,
      "grad_norm": 0.0018314807675778866,
      "learning_rate": 0.00018878549864259248,
      "loss": 23.0,
      "step": 2561
    },
    {
      "epoch": 1.53781512605042,
      "grad_norm": 0.0006086727953515947,
      "learning_rate": 0.00018877680481409732,
      "loss": 23.0,
      "step": 2562
    },
    {
      "epoch": 1.5384153661464586,
      "grad_norm": 0.0024697300978004932,
      "learning_rate": 0.00018876810781738636,
      "loss": 23.0,
      "step": 2563
    },
    {
      "epoch": 1.5390156062424971,
      "grad_norm": 0.00328907766379416,
      "learning_rate": 0.00018875940765276998,
      "loss": 23.0,
      "step": 2564
    },
    {
      "epoch": 1.5396158463385354,
      "grad_norm": 0.0006023477180860937,
      "learning_rate": 0.00018875070432055867,
      "loss": 23.0,
      "step": 2565
    },
    {
      "epoch": 1.5402160864345738,
      "grad_norm": 0.001448491937480867,
      "learning_rate": 0.000188741997821063,
      "loss": 23.0,
      "step": 2566
    },
    {
      "epoch": 1.5408163265306123,
      "grad_norm": 0.005367580335587263,
      "learning_rate": 0.0001887332881545937,
      "loss": 23.0,
      "step": 2567
    },
    {
      "epoch": 1.5414165666266506,
      "grad_norm": 0.0038878999184817076,
      "learning_rate": 0.0001887245753214616,
      "loss": 23.0,
      "step": 2568
    },
    {
      "epoch": 1.542016806722689,
      "grad_norm": 0.0016524168895557523,
      "learning_rate": 0.00018871585932197764,
      "loss": 23.0,
      "step": 2569
    },
    {
      "epoch": 1.5426170468187275,
      "grad_norm": 0.0018641971983015537,
      "learning_rate": 0.00018870714015645286,
      "loss": 23.0,
      "step": 2570
    },
    {
      "epoch": 1.543217286914766,
      "grad_norm": 0.0016603025142103434,
      "learning_rate": 0.0001886984178251984,
      "loss": 23.0,
      "step": 2571
    },
    {
      "epoch": 1.5438175270108043,
      "grad_norm": 0.001292040222324431,
      "learning_rate": 0.0001886896923285256,
      "loss": 23.0,
      "step": 2572
    },
    {
      "epoch": 1.5444177671068426,
      "grad_norm": 0.0008465956198051572,
      "learning_rate": 0.0001886809636667458,
      "loss": 23.0,
      "step": 2573
    },
    {
      "epoch": 1.5450180072028812,
      "grad_norm": 0.000892527517862618,
      "learning_rate": 0.00018867223184017054,
      "loss": 23.0,
      "step": 2574
    },
    {
      "epoch": 1.5456182472989197,
      "grad_norm": 0.0008575652609579265,
      "learning_rate": 0.0001886634968491114,
      "loss": 23.0,
      "step": 2575
    },
    {
      "epoch": 1.5462184873949578,
      "grad_norm": 0.0013137779897078872,
      "learning_rate": 0.00018865475869388014,
      "loss": 23.0,
      "step": 2576
    },
    {
      "epoch": 1.5468187274909964,
      "grad_norm": 0.0005612425738945603,
      "learning_rate": 0.0001886460173747886,
      "loss": 23.0,
      "step": 2577
    },
    {
      "epoch": 1.547418967587035,
      "grad_norm": 0.0016055781161412597,
      "learning_rate": 0.0001886372728921487,
      "loss": 23.0,
      "step": 2578
    },
    {
      "epoch": 1.5480192076830732,
      "grad_norm": 0.0006941373576410115,
      "learning_rate": 0.00018862852524627254,
      "loss": 23.0,
      "step": 2579
    },
    {
      "epoch": 1.5486194477791115,
      "grad_norm": 0.0018642544746398926,
      "learning_rate": 0.0001886197744374723,
      "loss": 23.0,
      "step": 2580
    },
    {
      "epoch": 1.54921968787515,
      "grad_norm": 0.0013856175355613232,
      "learning_rate": 0.00018861102046606028,
      "loss": 23.0,
      "step": 2581
    },
    {
      "epoch": 1.5498199279711886,
      "grad_norm": 0.0014988569309934974,
      "learning_rate": 0.00018860226333234885,
      "loss": 23.0,
      "step": 2582
    },
    {
      "epoch": 1.550420168067227,
      "grad_norm": 0.0015163436764851213,
      "learning_rate": 0.00018859350303665058,
      "loss": 23.0,
      "step": 2583
    },
    {
      "epoch": 1.5510204081632653,
      "grad_norm": 0.002808369230479002,
      "learning_rate": 0.00018858473957927805,
      "loss": 23.0,
      "step": 2584
    },
    {
      "epoch": 1.5516206482593038,
      "grad_norm": 0.0011914572678506374,
      "learning_rate": 0.00018857597296054404,
      "loss": 23.0,
      "step": 2585
    },
    {
      "epoch": 1.552220888355342,
      "grad_norm": 0.0012025490868836641,
      "learning_rate": 0.00018856720318076139,
      "loss": 23.0,
      "step": 2586
    },
    {
      "epoch": 1.5528211284513804,
      "grad_norm": 0.001126796007156372,
      "learning_rate": 0.00018855843024024308,
      "loss": 23.0,
      "step": 2587
    },
    {
      "epoch": 1.553421368547419,
      "grad_norm": 0.0008894942584447563,
      "learning_rate": 0.00018854965413930223,
      "loss": 23.0,
      "step": 2588
    },
    {
      "epoch": 1.5540216086434575,
      "grad_norm": 0.0018061741720885038,
      "learning_rate": 0.00018854087487825196,
      "loss": 23.0,
      "step": 2589
    },
    {
      "epoch": 1.5546218487394958,
      "grad_norm": 0.0035426064860075712,
      "learning_rate": 0.00018853209245740565,
      "loss": 23.0,
      "step": 2590
    },
    {
      "epoch": 1.5552220888355341,
      "grad_norm": 0.0011812059674412012,
      "learning_rate": 0.00018852330687707669,
      "loss": 23.0,
      "step": 2591
    },
    {
      "epoch": 1.5558223289315727,
      "grad_norm": 0.001240815268829465,
      "learning_rate": 0.0001885145181375786,
      "loss": 23.0,
      "step": 2592
    },
    {
      "epoch": 1.556422569027611,
      "grad_norm": 0.0007779524312354624,
      "learning_rate": 0.00018850572623922503,
      "loss": 23.0,
      "step": 2593
    },
    {
      "epoch": 1.5570228091236493,
      "grad_norm": 0.0011602307204157114,
      "learning_rate": 0.00018849693118232978,
      "loss": 23.0,
      "step": 2594
    },
    {
      "epoch": 1.5576230492196879,
      "grad_norm": 0.002238318556919694,
      "learning_rate": 0.00018848813296720672,
      "loss": 23.0,
      "step": 2595
    },
    {
      "epoch": 1.5582232893157264,
      "grad_norm": 0.0019287511240690947,
      "learning_rate": 0.00018847933159416976,
      "loss": 23.0,
      "step": 2596
    },
    {
      "epoch": 1.5588235294117647,
      "grad_norm": 0.0006660337676294148,
      "learning_rate": 0.00018847052706353308,
      "loss": 23.0,
      "step": 2597
    },
    {
      "epoch": 1.559423769507803,
      "grad_norm": 0.0008473030175082386,
      "learning_rate": 0.00018846171937561085,
      "loss": 23.0,
      "step": 2598
    },
    {
      "epoch": 1.5600240096038416,
      "grad_norm": 0.001743235276080668,
      "learning_rate": 0.00018845290853071743,
      "loss": 23.0,
      "step": 2599
    },
    {
      "epoch": 1.5606242496998801,
      "grad_norm": 0.000990196131169796,
      "learning_rate": 0.0001884440945291672,
      "loss": 23.0,
      "step": 2600
    },
    {
      "epoch": 1.5612244897959182,
      "grad_norm": 0.0017959270626306534,
      "learning_rate": 0.00018843527737127475,
      "loss": 23.0,
      "step": 2601
    },
    {
      "epoch": 1.5618247298919568,
      "grad_norm": 0.0012067595962435007,
      "learning_rate": 0.00018842645705735475,
      "loss": 23.0,
      "step": 2602
    },
    {
      "epoch": 1.5624249699879953,
      "grad_norm": 0.0005739233456552029,
      "learning_rate": 0.00018841763358772194,
      "loss": 23.0,
      "step": 2603
    },
    {
      "epoch": 1.5630252100840336,
      "grad_norm": 0.0009168532560579479,
      "learning_rate": 0.00018840880696269127,
      "loss": 23.0,
      "step": 2604
    },
    {
      "epoch": 1.563625450180072,
      "grad_norm": 0.0008688515517860651,
      "learning_rate": 0.00018839997718257765,
      "loss": 23.0,
      "step": 2605
    },
    {
      "epoch": 1.5642256902761105,
      "grad_norm": 0.0036098710261285305,
      "learning_rate": 0.00018839114424769625,
      "loss": 23.0,
      "step": 2606
    },
    {
      "epoch": 1.564825930372149,
      "grad_norm": 0.0009449098724871874,
      "learning_rate": 0.00018838230815836226,
      "loss": 23.0,
      "step": 2607
    },
    {
      "epoch": 1.5654261704681873,
      "grad_norm": 0.0010306858457624912,
      "learning_rate": 0.00018837346891489105,
      "loss": 23.0,
      "step": 2608
    },
    {
      "epoch": 1.5660264105642256,
      "grad_norm": 0.0003999176260549575,
      "learning_rate": 0.00018836462651759806,
      "loss": 23.0,
      "step": 2609
    },
    {
      "epoch": 1.5666266506602642,
      "grad_norm": 0.0008212673128582537,
      "learning_rate": 0.00018835578096679885,
      "loss": 23.0,
      "step": 2610
    },
    {
      "epoch": 1.5672268907563025,
      "grad_norm": 0.0017521216068416834,
      "learning_rate": 0.0001883469322628091,
      "loss": 23.0,
      "step": 2611
    },
    {
      "epoch": 1.5678271308523408,
      "grad_norm": 0.0013428755337372422,
      "learning_rate": 0.0001883380804059446,
      "loss": 23.0,
      "step": 2612
    },
    {
      "epoch": 1.5684273709483794,
      "grad_norm": 0.0019537501502782106,
      "learning_rate": 0.00018832922539652124,
      "loss": 23.0,
      "step": 2613
    },
    {
      "epoch": 1.569027611044418,
      "grad_norm": 0.0013494847808033228,
      "learning_rate": 0.00018832036723485503,
      "loss": 23.0,
      "step": 2614
    },
    {
      "epoch": 1.5696278511404562,
      "grad_norm": 0.0013262125430628657,
      "learning_rate": 0.00018831150592126208,
      "loss": 23.0,
      "step": 2615
    },
    {
      "epoch": 1.5702280912364945,
      "grad_norm": 0.0012583464849740267,
      "learning_rate": 0.00018830264145605868,
      "loss": 23.0,
      "step": 2616
    },
    {
      "epoch": 1.570828331332533,
      "grad_norm": 0.0018956343410536647,
      "learning_rate": 0.00018829377383956114,
      "loss": 23.0,
      "step": 2617
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 0.0012951820390298963,
      "learning_rate": 0.0001882849030720859,
      "loss": 23.0,
      "step": 2618
    },
    {
      "epoch": 1.5720288115246097,
      "grad_norm": 0.0012970389798283577,
      "learning_rate": 0.0001882760291539496,
      "loss": 23.0,
      "step": 2619
    },
    {
      "epoch": 1.5726290516206483,
      "grad_norm": 0.001099150162190199,
      "learning_rate": 0.0001882671520854689,
      "loss": 23.0,
      "step": 2620
    },
    {
      "epoch": 1.5732292917166868,
      "grad_norm": 0.002676828298717737,
      "learning_rate": 0.00018825827186696057,
      "loss": 23.0,
      "step": 2621
    },
    {
      "epoch": 1.5738295318127251,
      "grad_norm": 0.001391171826981008,
      "learning_rate": 0.00018824938849874157,
      "loss": 23.0,
      "step": 2622
    },
    {
      "epoch": 1.5744297719087634,
      "grad_norm": 0.0007992578903213143,
      "learning_rate": 0.00018824050198112886,
      "loss": 23.0,
      "step": 2623
    },
    {
      "epoch": 1.575030012004802,
      "grad_norm": 0.0015663999365642667,
      "learning_rate": 0.00018823161231443965,
      "loss": 23.0,
      "step": 2624
    },
    {
      "epoch": 1.5756302521008403,
      "grad_norm": 0.00034730014158412814,
      "learning_rate": 0.00018822271949899114,
      "loss": 23.0,
      "step": 2625
    },
    {
      "epoch": 1.5762304921968786,
      "grad_norm": 0.0012912418460473418,
      "learning_rate": 0.00018821382353510072,
      "loss": 23.0,
      "step": 2626
    },
    {
      "epoch": 1.5768307322929171,
      "grad_norm": 0.001319359173066914,
      "learning_rate": 0.00018820492442308583,
      "loss": 23.0,
      "step": 2627
    },
    {
      "epoch": 1.5774309723889557,
      "grad_norm": 0.0020128735341131687,
      "learning_rate": 0.00018819602216326407,
      "loss": 23.0,
      "step": 2628
    },
    {
      "epoch": 1.578031212484994,
      "grad_norm": 0.0011375844478607178,
      "learning_rate": 0.00018818711675595316,
      "loss": 23.0,
      "step": 2629
    },
    {
      "epoch": 1.5786314525810323,
      "grad_norm": 0.001151563599705696,
      "learning_rate": 0.0001881782082014709,
      "loss": 23.0,
      "step": 2630
    },
    {
      "epoch": 1.5792316926770709,
      "grad_norm": 0.0022804574109613895,
      "learning_rate": 0.0001881692965001352,
      "loss": 23.0,
      "step": 2631
    },
    {
      "epoch": 1.5798319327731094,
      "grad_norm": 0.0008478496456518769,
      "learning_rate": 0.00018816038165226414,
      "loss": 23.0,
      "step": 2632
    },
    {
      "epoch": 1.5804321728691475,
      "grad_norm": 0.001113467151299119,
      "learning_rate": 0.0001881514636581758,
      "loss": 23.0,
      "step": 2633
    },
    {
      "epoch": 1.581032412965186,
      "grad_norm": 0.0009239333448931575,
      "learning_rate": 0.00018814254251818847,
      "loss": 23.0,
      "step": 2634
    },
    {
      "epoch": 1.5816326530612246,
      "grad_norm": 0.001530701294541359,
      "learning_rate": 0.00018813361823262056,
      "loss": 23.0,
      "step": 2635
    },
    {
      "epoch": 1.582232893157263,
      "grad_norm": 0.0008521588752046227,
      "learning_rate": 0.0001881246908017905,
      "loss": 23.0,
      "step": 2636
    },
    {
      "epoch": 1.5828331332533012,
      "grad_norm": 0.001877720351330936,
      "learning_rate": 0.00018811576022601691,
      "loss": 23.0,
      "step": 2637
    },
    {
      "epoch": 1.5834333733493398,
      "grad_norm": 0.0016532663721591234,
      "learning_rate": 0.0001881068265056185,
      "loss": 23.0,
      "step": 2638
    },
    {
      "epoch": 1.5840336134453783,
      "grad_norm": 0.0017709649400785565,
      "learning_rate": 0.0001880978896409141,
      "loss": 23.0,
      "step": 2639
    },
    {
      "epoch": 1.5846338535414166,
      "grad_norm": 0.0004166093422099948,
      "learning_rate": 0.00018808894963222262,
      "loss": 23.0,
      "step": 2640
    },
    {
      "epoch": 1.585234093637455,
      "grad_norm": 0.0013971725711598992,
      "learning_rate": 0.00018808000647986312,
      "loss": 23.0,
      "step": 2641
    },
    {
      "epoch": 1.5858343337334935,
      "grad_norm": 0.0002562387671787292,
      "learning_rate": 0.00018807106018415477,
      "loss": 23.0,
      "step": 2642
    },
    {
      "epoch": 1.5864345738295318,
      "grad_norm": 0.0017581944121047854,
      "learning_rate": 0.00018806211074541681,
      "loss": 23.0,
      "step": 2643
    },
    {
      "epoch": 1.58703481392557,
      "grad_norm": 0.0027277956251055002,
      "learning_rate": 0.00018805315816396867,
      "loss": 23.0,
      "step": 2644
    },
    {
      "epoch": 1.5876350540216086,
      "grad_norm": 0.0008474817150272429,
      "learning_rate": 0.0001880442024401298,
      "loss": 23.0,
      "step": 2645
    },
    {
      "epoch": 1.5882352941176472,
      "grad_norm": 0.0010701556457206607,
      "learning_rate": 0.00018803524357421982,
      "loss": 23.0,
      "step": 2646
    },
    {
      "epoch": 1.5888355342136855,
      "grad_norm": 0.001056216424331069,
      "learning_rate": 0.00018802628156655844,
      "loss": 23.0,
      "step": 2647
    },
    {
      "epoch": 1.5894357743097238,
      "grad_norm": 0.0007937379996292293,
      "learning_rate": 0.00018801731641746548,
      "loss": 23.0,
      "step": 2648
    },
    {
      "epoch": 1.5900360144057624,
      "grad_norm": 0.0021392793860286474,
      "learning_rate": 0.00018800834812726097,
      "loss": 23.0,
      "step": 2649
    },
    {
      "epoch": 1.5906362545018007,
      "grad_norm": 0.0005592108354903758,
      "learning_rate": 0.00018799937669626484,
      "loss": 23.0,
      "step": 2650
    },
    {
      "epoch": 1.591236494597839,
      "grad_norm": 0.0013419737806543708,
      "learning_rate": 0.00018799040212479735,
      "loss": 23.0,
      "step": 2651
    },
    {
      "epoch": 1.5918367346938775,
      "grad_norm": 0.0014628679491579533,
      "learning_rate": 0.00018798142441317874,
      "loss": 23.0,
      "step": 2652
    },
    {
      "epoch": 1.592436974789916,
      "grad_norm": 0.002074423711746931,
      "learning_rate": 0.0001879724435617294,
      "loss": 23.0,
      "step": 2653
    },
    {
      "epoch": 1.5930372148859544,
      "grad_norm": 0.001169572351500392,
      "learning_rate": 0.00018796345957076987,
      "loss": 23.0,
      "step": 2654
    },
    {
      "epoch": 1.5936374549819927,
      "grad_norm": 0.0019296786049380898,
      "learning_rate": 0.0001879544724406207,
      "loss": 23.0,
      "step": 2655
    },
    {
      "epoch": 1.5942376950780313,
      "grad_norm": 0.002079583238810301,
      "learning_rate": 0.00018794548217160268,
      "loss": 23.0,
      "step": 2656
    },
    {
      "epoch": 1.5948379351740696,
      "grad_norm": 0.0011784069938585162,
      "learning_rate": 0.0001879364887640366,
      "loss": 23.0,
      "step": 2657
    },
    {
      "epoch": 1.595438175270108,
      "grad_norm": 0.0019995667971670628,
      "learning_rate": 0.00018792749221824346,
      "loss": 23.0,
      "step": 2658
    },
    {
      "epoch": 1.5960384153661464,
      "grad_norm": 0.0017403881065547466,
      "learning_rate": 0.0001879184925345443,
      "loss": 23.0,
      "step": 2659
    },
    {
      "epoch": 1.596638655462185,
      "grad_norm": 0.002742258831858635,
      "learning_rate": 0.00018790948971326025,
      "loss": 23.0,
      "step": 2660
    },
    {
      "epoch": 1.5972388955582233,
      "grad_norm": 0.001577183953486383,
      "learning_rate": 0.0001879004837547127,
      "loss": 23.0,
      "step": 2661
    },
    {
      "epoch": 1.5978391356542616,
      "grad_norm": 0.003025846555829048,
      "learning_rate": 0.00018789147465922293,
      "loss": 23.0,
      "step": 2662
    },
    {
      "epoch": 1.5984393757503002,
      "grad_norm": 0.0008754048030823469,
      "learning_rate": 0.00018788246242711258,
      "loss": 23.0,
      "step": 2663
    },
    {
      "epoch": 1.5990396158463387,
      "grad_norm": 0.0012841924326494336,
      "learning_rate": 0.00018787344705870318,
      "loss": 23.0,
      "step": 2664
    },
    {
      "epoch": 1.5996398559423768,
      "grad_norm": 0.0024323989637196064,
      "learning_rate": 0.00018786442855431648,
      "loss": 23.0,
      "step": 2665
    },
    {
      "epoch": 1.6002400960384153,
      "grad_norm": 0.0006733940681442618,
      "learning_rate": 0.00018785540691427434,
      "loss": 23.0,
      "step": 2666
    },
    {
      "epoch": 1.6008403361344539,
      "grad_norm": 0.0016225153813138604,
      "learning_rate": 0.0001878463821388987,
      "loss": 23.0,
      "step": 2667
    },
    {
      "epoch": 1.6014405762304922,
      "grad_norm": 0.0007324088364839554,
      "learning_rate": 0.0001878373542285117,
      "loss": 23.0,
      "step": 2668
    },
    {
      "epoch": 1.6020408163265305,
      "grad_norm": 0.0010630915639922023,
      "learning_rate": 0.00018782832318343546,
      "loss": 23.0,
      "step": 2669
    },
    {
      "epoch": 1.602641056422569,
      "grad_norm": 0.0016140867955982685,
      "learning_rate": 0.00018781928900399227,
      "loss": 23.0,
      "step": 2670
    },
    {
      "epoch": 1.6032412965186076,
      "grad_norm": 0.0014620976289734244,
      "learning_rate": 0.00018781025169050454,
      "loss": 23.0,
      "step": 2671
    },
    {
      "epoch": 1.603841536614646,
      "grad_norm": 0.0013933505397289991,
      "learning_rate": 0.00018780121124329483,
      "loss": 23.0,
      "step": 2672
    },
    {
      "epoch": 1.6044417767106842,
      "grad_norm": 0.0016851526452228427,
      "learning_rate": 0.00018779216766268576,
      "loss": 23.0,
      "step": 2673
    },
    {
      "epoch": 1.6050420168067228,
      "grad_norm": 0.0022437949664890766,
      "learning_rate": 0.00018778312094900002,
      "loss": 23.0,
      "step": 2674
    },
    {
      "epoch": 1.605642256902761,
      "grad_norm": 0.0025705392472445965,
      "learning_rate": 0.00018777407110256048,
      "loss": 23.0,
      "step": 2675
    },
    {
      "epoch": 1.6062424969987994,
      "grad_norm": 0.0014031381579115987,
      "learning_rate": 0.00018776501812369016,
      "loss": 23.0,
      "step": 2676
    },
    {
      "epoch": 1.606842737094838,
      "grad_norm": 0.001516643795184791,
      "learning_rate": 0.0001877559620127121,
      "loss": 23.0,
      "step": 2677
    },
    {
      "epoch": 1.6074429771908765,
      "grad_norm": 0.0029964365530759096,
      "learning_rate": 0.00018774690276994948,
      "loss": 23.0,
      "step": 2678
    },
    {
      "epoch": 1.6080432172869148,
      "grad_norm": 0.0010419205063953996,
      "learning_rate": 0.0001877378403957256,
      "loss": 23.0,
      "step": 2679
    },
    {
      "epoch": 1.6086434573829531,
      "grad_norm": 0.0006046218913979828,
      "learning_rate": 0.0001877287748903639,
      "loss": 23.0,
      "step": 2680
    },
    {
      "epoch": 1.6092436974789917,
      "grad_norm": 0.0015536684077233076,
      "learning_rate": 0.0001877197062541879,
      "loss": 23.0,
      "step": 2681
    },
    {
      "epoch": 1.60984393757503,
      "grad_norm": 0.0012977086007595062,
      "learning_rate": 0.0001877106344875212,
      "loss": 23.0,
      "step": 2682
    },
    {
      "epoch": 1.6104441776710683,
      "grad_norm": 0.0015513448743149638,
      "learning_rate": 0.00018770155959068761,
      "loss": 23.0,
      "step": 2683
    },
    {
      "epoch": 1.6110444177671068,
      "grad_norm": 0.0021471292711794376,
      "learning_rate": 0.0001876924815640109,
      "loss": 23.0,
      "step": 2684
    },
    {
      "epoch": 1.6116446578631454,
      "grad_norm": 0.0021347152069211006,
      "learning_rate": 0.00018768340040781512,
      "loss": 23.0,
      "step": 2685
    },
    {
      "epoch": 1.6122448979591837,
      "grad_norm": 0.0017109934706240892,
      "learning_rate": 0.00018767431612242436,
      "loss": 23.0,
      "step": 2686
    },
    {
      "epoch": 1.612845138055222,
      "grad_norm": 0.0010965627152472734,
      "learning_rate": 0.00018766522870816273,
      "loss": 23.0,
      "step": 2687
    },
    {
      "epoch": 1.6134453781512605,
      "grad_norm": 0.0008786679827608168,
      "learning_rate": 0.00018765613816535464,
      "loss": 23.0,
      "step": 2688
    },
    {
      "epoch": 1.614045618247299,
      "grad_norm": 0.0014945195289328694,
      "learning_rate": 0.00018764704449432444,
      "loss": 23.0,
      "step": 2689
    },
    {
      "epoch": 1.6146458583433372,
      "grad_norm": 0.0020942450501024723,
      "learning_rate": 0.00018763794769539668,
      "loss": 23.0,
      "step": 2690
    },
    {
      "epoch": 1.6152460984393757,
      "grad_norm": 0.0010354813421145082,
      "learning_rate": 0.00018762884776889598,
      "loss": 23.0,
      "step": 2691
    },
    {
      "epoch": 1.6158463385354143,
      "grad_norm": 0.0007208411116153002,
      "learning_rate": 0.00018761974471514714,
      "loss": 23.0,
      "step": 2692
    },
    {
      "epoch": 1.6164465786314526,
      "grad_norm": 0.0014213203685358167,
      "learning_rate": 0.00018761063853447497,
      "loss": 23.0,
      "step": 2693
    },
    {
      "epoch": 1.617046818727491,
      "grad_norm": 0.0028033480048179626,
      "learning_rate": 0.00018760152922720445,
      "loss": 23.0,
      "step": 2694
    },
    {
      "epoch": 1.6176470588235294,
      "grad_norm": 0.0010669436305761337,
      "learning_rate": 0.00018759241679366072,
      "loss": 23.0,
      "step": 2695
    },
    {
      "epoch": 1.618247298919568,
      "grad_norm": 0.002347268396988511,
      "learning_rate": 0.00018758330123416897,
      "loss": 23.0,
      "step": 2696
    },
    {
      "epoch": 1.6188475390156063,
      "grad_norm": 0.00081385433441028,
      "learning_rate": 0.0001875741825490545,
      "loss": 23.0,
      "step": 2697
    },
    {
      "epoch": 1.6194477791116446,
      "grad_norm": 0.0014208892825990915,
      "learning_rate": 0.00018756506073864266,
      "loss": 23.0,
      "step": 2698
    },
    {
      "epoch": 1.6200480192076832,
      "grad_norm": 0.002604991663247347,
      "learning_rate": 0.00018755593580325908,
      "loss": 23.0,
      "step": 2699
    },
    {
      "epoch": 1.6206482593037215,
      "grad_norm": 0.0017933185445144773,
      "learning_rate": 0.00018754680774322937,
      "loss": 23.0,
      "step": 2700
    },
    {
      "epoch": 1.6212484993997598,
      "grad_norm": 0.001248373999260366,
      "learning_rate": 0.0001875376765588793,
      "loss": 23.0,
      "step": 2701
    },
    {
      "epoch": 1.6218487394957983,
      "grad_norm": 0.0012778202071785927,
      "learning_rate": 0.00018752854225053471,
      "loss": 23.0,
      "step": 2702
    },
    {
      "epoch": 1.6224489795918369,
      "grad_norm": 0.0027483454905450344,
      "learning_rate": 0.0001875194048185216,
      "loss": 23.0,
      "step": 2703
    },
    {
      "epoch": 1.6230492196878752,
      "grad_norm": 0.002060843864455819,
      "learning_rate": 0.00018751026426316604,
      "loss": 23.0,
      "step": 2704
    },
    {
      "epoch": 1.6236494597839135,
      "grad_norm": 0.0008097304962575436,
      "learning_rate": 0.00018750112058479429,
      "loss": 23.0,
      "step": 2705
    },
    {
      "epoch": 1.624249699879952,
      "grad_norm": 0.0013644036371260881,
      "learning_rate": 0.00018749197378373263,
      "loss": 23.0,
      "step": 2706
    },
    {
      "epoch": 1.6248499399759904,
      "grad_norm": 0.002363650593906641,
      "learning_rate": 0.00018748282386030747,
      "loss": 23.0,
      "step": 2707
    },
    {
      "epoch": 1.6254501800720287,
      "grad_norm": 0.001418265514075756,
      "learning_rate": 0.00018747367081484533,
      "loss": 23.0,
      "step": 2708
    },
    {
      "epoch": 1.6260504201680672,
      "grad_norm": 0.001415880979038775,
      "learning_rate": 0.00018746451464767293,
      "loss": 23.0,
      "step": 2709
    },
    {
      "epoch": 1.6266506602641058,
      "grad_norm": 0.0013812303077429533,
      "learning_rate": 0.00018745535535911697,
      "loss": 23.0,
      "step": 2710
    },
    {
      "epoch": 1.627250900360144,
      "grad_norm": 0.0014266427606344223,
      "learning_rate": 0.0001874461929495043,
      "loss": 23.0,
      "step": 2711
    },
    {
      "epoch": 1.6278511404561824,
      "grad_norm": 0.0015677702613174915,
      "learning_rate": 0.000187437027419162,
      "loss": 23.0,
      "step": 2712
    },
    {
      "epoch": 1.628451380552221,
      "grad_norm": 0.001671108417212963,
      "learning_rate": 0.00018742785876841708,
      "loss": 23.0,
      "step": 2713
    },
    {
      "epoch": 1.6290516206482593,
      "grad_norm": 0.002867297735065222,
      "learning_rate": 0.00018741868699759676,
      "loss": 23.0,
      "step": 2714
    },
    {
      "epoch": 1.6296518607442976,
      "grad_norm": 0.0007195291109383106,
      "learning_rate": 0.0001874095121070284,
      "loss": 23.0,
      "step": 2715
    },
    {
      "epoch": 1.6302521008403361,
      "grad_norm": 0.0005874624475836754,
      "learning_rate": 0.0001874003340970394,
      "loss": 23.0,
      "step": 2716
    },
    {
      "epoch": 1.6308523409363747,
      "grad_norm": 0.0006918995641171932,
      "learning_rate": 0.00018739115296795725,
      "loss": 23.0,
      "step": 2717
    },
    {
      "epoch": 1.631452581032413,
      "grad_norm": 0.0035936187487095594,
      "learning_rate": 0.0001873819687201097,
      "loss": 23.0,
      "step": 2718
    },
    {
      "epoch": 1.6320528211284513,
      "grad_norm": 0.0008181266020983458,
      "learning_rate": 0.0001873727813538244,
      "loss": 23.0,
      "step": 2719
    },
    {
      "epoch": 1.6326530612244898,
      "grad_norm": 0.0016008074162527919,
      "learning_rate": 0.00018736359086942933,
      "loss": 23.0,
      "step": 2720
    },
    {
      "epoch": 1.6332533013205284,
      "grad_norm": 0.00043132936116307974,
      "learning_rate": 0.00018735439726725242,
      "loss": 23.0,
      "step": 2721
    },
    {
      "epoch": 1.6338535414165665,
      "grad_norm": 0.001108235097490251,
      "learning_rate": 0.00018734520054762178,
      "loss": 23.0,
      "step": 2722
    },
    {
      "epoch": 1.634453781512605,
      "grad_norm": 0.0010154407937079668,
      "learning_rate": 0.00018733600071086558,
      "loss": 23.0,
      "step": 2723
    },
    {
      "epoch": 1.6350540216086435,
      "grad_norm": 0.0015051518566906452,
      "learning_rate": 0.00018732679775731222,
      "loss": 23.0,
      "step": 2724
    },
    {
      "epoch": 1.6356542617046819,
      "grad_norm": 0.0019470944534987211,
      "learning_rate": 0.00018731759168729003,
      "loss": 23.0,
      "step": 2725
    },
    {
      "epoch": 1.6362545018007202,
      "grad_norm": 0.0026996361557394266,
      "learning_rate": 0.00018730838250112763,
      "loss": 23.0,
      "step": 2726
    },
    {
      "epoch": 1.6368547418967587,
      "grad_norm": 0.0007813902338966727,
      "learning_rate": 0.00018729917019915364,
      "loss": 23.0,
      "step": 2727
    },
    {
      "epoch": 1.6374549819927973,
      "grad_norm": 0.0006077332654967904,
      "learning_rate": 0.0001872899547816968,
      "loss": 23.0,
      "step": 2728
    },
    {
      "epoch": 1.6380552220888356,
      "grad_norm": 0.001340494374744594,
      "learning_rate": 0.00018728073624908604,
      "loss": 23.0,
      "step": 2729
    },
    {
      "epoch": 1.638655462184874,
      "grad_norm": 0.001249544438906014,
      "learning_rate": 0.0001872715146016503,
      "loss": 23.0,
      "step": 2730
    },
    {
      "epoch": 1.6392557022809124,
      "grad_norm": 0.002360557671636343,
      "learning_rate": 0.0001872622898397187,
      "loss": 23.0,
      "step": 2731
    },
    {
      "epoch": 1.6398559423769508,
      "grad_norm": 0.0006911809323355556,
      "learning_rate": 0.00018725306196362045,
      "loss": 23.0,
      "step": 2732
    },
    {
      "epoch": 1.640456182472989,
      "grad_norm": 0.0007208830211311579,
      "learning_rate": 0.00018724383097368485,
      "loss": 23.0,
      "step": 2733
    },
    {
      "epoch": 1.6410564225690276,
      "grad_norm": 0.002218914683908224,
      "learning_rate": 0.00018723459687024135,
      "loss": 23.0,
      "step": 2734
    },
    {
      "epoch": 1.6416566626650662,
      "grad_norm": 0.00204529264010489,
      "learning_rate": 0.00018722535965361948,
      "loss": 23.0,
      "step": 2735
    },
    {
      "epoch": 1.6422569027611045,
      "grad_norm": 0.0008679120219312608,
      "learning_rate": 0.00018721611932414887,
      "loss": 23.0,
      "step": 2736
    },
    {
      "epoch": 1.6428571428571428,
      "grad_norm": 0.0013889900874346495,
      "learning_rate": 0.00018720687588215932,
      "loss": 23.0,
      "step": 2737
    },
    {
      "epoch": 1.6434573829531813,
      "grad_norm": 0.0008765452657826245,
      "learning_rate": 0.00018719762932798068,
      "loss": 23.0,
      "step": 2738
    },
    {
      "epoch": 1.6440576230492197,
      "grad_norm": 0.0008362770895473659,
      "learning_rate": 0.000187188379661943,
      "loss": 23.0,
      "step": 2739
    },
    {
      "epoch": 1.644657863145258,
      "grad_norm": 0.002404067199677229,
      "learning_rate": 0.0001871791268843763,
      "loss": 23.0,
      "step": 2740
    },
    {
      "epoch": 1.6452581032412965,
      "grad_norm": 0.0017013377510011196,
      "learning_rate": 0.00018716987099561078,
      "loss": 23.0,
      "step": 2741
    },
    {
      "epoch": 1.645858343337335,
      "grad_norm": 0.0016773076495155692,
      "learning_rate": 0.00018716061199597684,
      "loss": 23.0,
      "step": 2742
    },
    {
      "epoch": 1.6464585834333734,
      "grad_norm": 0.0012956123100593686,
      "learning_rate": 0.0001871513498858048,
      "loss": 23.0,
      "step": 2743
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 0.0014770814450457692,
      "learning_rate": 0.0001871420846654253,
      "loss": 23.0,
      "step": 2744
    },
    {
      "epoch": 1.6476590636254502,
      "grad_norm": 0.0028329326305538416,
      "learning_rate": 0.00018713281633516896,
      "loss": 23.0,
      "step": 2745
    },
    {
      "epoch": 1.6482593037214885,
      "grad_norm": 0.002407442545518279,
      "learning_rate": 0.00018712354489536654,
      "loss": 23.0,
      "step": 2746
    },
    {
      "epoch": 1.6488595438175269,
      "grad_norm": 0.0009911904344335198,
      "learning_rate": 0.00018711427034634893,
      "loss": 23.0,
      "step": 2747
    },
    {
      "epoch": 1.6494597839135654,
      "grad_norm": 0.0009113244013860822,
      "learning_rate": 0.00018710499268844704,
      "loss": 23.0,
      "step": 2748
    },
    {
      "epoch": 1.650060024009604,
      "grad_norm": 0.0010095303878188133,
      "learning_rate": 0.00018709571192199206,
      "loss": 23.0,
      "step": 2749
    },
    {
      "epoch": 1.6506602641056423,
      "grad_norm": 0.0011850446462631226,
      "learning_rate": 0.00018708642804731517,
      "loss": 23.0,
      "step": 2750
    },
    {
      "epoch": 1.6512605042016806,
      "grad_norm": 0.0017300506588071585,
      "learning_rate": 0.00018707714106474762,
      "loss": 23.0,
      "step": 2751
    },
    {
      "epoch": 1.6518607442977191,
      "grad_norm": 0.0003194924211129546,
      "learning_rate": 0.00018706785097462095,
      "loss": 23.0,
      "step": 2752
    },
    {
      "epoch": 1.6524609843937577,
      "grad_norm": 0.0020195262040942907,
      "learning_rate": 0.00018705855777726663,
      "loss": 23.0,
      "step": 2753
    },
    {
      "epoch": 1.6530612244897958,
      "grad_norm": 0.0017491821199655533,
      "learning_rate": 0.00018704926147301628,
      "loss": 23.0,
      "step": 2754
    },
    {
      "epoch": 1.6536614645858343,
      "grad_norm": 0.0022025639191269875,
      "learning_rate": 0.00018703996206220178,
      "loss": 23.0,
      "step": 2755
    },
    {
      "epoch": 1.6542617046818728,
      "grad_norm": 0.0026374238077551126,
      "learning_rate": 0.00018703065954515488,
      "loss": 23.0,
      "step": 2756
    },
    {
      "epoch": 1.6548619447779112,
      "grad_norm": 0.0011183744063600898,
      "learning_rate": 0.00018702135392220762,
      "loss": 23.0,
      "step": 2757
    },
    {
      "epoch": 1.6554621848739495,
      "grad_norm": 0.0015539663145318627,
      "learning_rate": 0.0001870120451936921,
      "loss": 23.0,
      "step": 2758
    },
    {
      "epoch": 1.656062424969988,
      "grad_norm": 0.0021634141448885202,
      "learning_rate": 0.00018700273335994048,
      "loss": 23.0,
      "step": 2759
    },
    {
      "epoch": 1.6566626650660266,
      "grad_norm": 0.0017279998864978552,
      "learning_rate": 0.0001869934184212851,
      "loss": 23.0,
      "step": 2760
    },
    {
      "epoch": 1.6572629051620649,
      "grad_norm": 0.0007326691993512213,
      "learning_rate": 0.00018698410037805842,
      "loss": 23.0,
      "step": 2761
    },
    {
      "epoch": 1.6578631452581032,
      "grad_norm": 0.0009592922870069742,
      "learning_rate": 0.00018697477923059294,
      "loss": 23.0,
      "step": 2762
    },
    {
      "epoch": 1.6584633853541417,
      "grad_norm": 0.001174344215542078,
      "learning_rate": 0.00018696545497922133,
      "loss": 23.0,
      "step": 2763
    },
    {
      "epoch": 1.65906362545018,
      "grad_norm": 0.0005253737326711416,
      "learning_rate": 0.0001869561276242763,
      "loss": 23.0,
      "step": 2764
    },
    {
      "epoch": 1.6596638655462184,
      "grad_norm": 0.002293418627232313,
      "learning_rate": 0.00018694679716609077,
      "loss": 23.0,
      "step": 2765
    },
    {
      "epoch": 1.660264105642257,
      "grad_norm": 0.001944358111359179,
      "learning_rate": 0.00018693746360499768,
      "loss": 23.0,
      "step": 2766
    },
    {
      "epoch": 1.6608643457382954,
      "grad_norm": 0.0005258253659121692,
      "learning_rate": 0.0001869281269413302,
      "loss": 23.0,
      "step": 2767
    },
    {
      "epoch": 1.6614645858343338,
      "grad_norm": 0.0025244082789868116,
      "learning_rate": 0.00018691878717542145,
      "loss": 23.0,
      "step": 2768
    },
    {
      "epoch": 1.662064825930372,
      "grad_norm": 0.003507652087137103,
      "learning_rate": 0.00018690944430760473,
      "loss": 23.0,
      "step": 2769
    },
    {
      "epoch": 1.6626650660264106,
      "grad_norm": 0.0008565374300815165,
      "learning_rate": 0.0001869000983382135,
      "loss": 23.0,
      "step": 2770
    },
    {
      "epoch": 1.663265306122449,
      "grad_norm": 0.0023754527792334557,
      "learning_rate": 0.00018689074926758138,
      "loss": 23.0,
      "step": 2771
    },
    {
      "epoch": 1.6638655462184873,
      "grad_norm": 0.0007233821670524776,
      "learning_rate": 0.00018688139709604186,
      "loss": 23.0,
      "step": 2772
    },
    {
      "epoch": 1.6644657863145258,
      "grad_norm": 0.0016606366261839867,
      "learning_rate": 0.00018687204182392874,
      "loss": 23.0,
      "step": 2773
    },
    {
      "epoch": 1.6650660264105643,
      "grad_norm": 0.000989391002804041,
      "learning_rate": 0.00018686268345157595,
      "loss": 23.0,
      "step": 2774
    },
    {
      "epoch": 1.6656662665066027,
      "grad_norm": 0.001483695930801332,
      "learning_rate": 0.00018685332197931743,
      "loss": 23.0,
      "step": 2775
    },
    {
      "epoch": 1.666266506602641,
      "grad_norm": 0.0011198918800801039,
      "learning_rate": 0.00018684395740748726,
      "loss": 23.0,
      "step": 2776
    },
    {
      "epoch": 1.6668667466986795,
      "grad_norm": 0.0007456332677975297,
      "learning_rate": 0.00018683458973641962,
      "loss": 23.0,
      "step": 2777
    },
    {
      "epoch": 1.667466986794718,
      "grad_norm": 0.001662615453824401,
      "learning_rate": 0.00018682521896644882,
      "loss": 23.0,
      "step": 2778
    },
    {
      "epoch": 1.6680672268907561,
      "grad_norm": 0.001044997712597251,
      "learning_rate": 0.00018681584509790936,
      "loss": 23.0,
      "step": 2779
    },
    {
      "epoch": 1.6686674669867947,
      "grad_norm": 0.0005202651955187321,
      "learning_rate": 0.00018680646813113564,
      "loss": 23.0,
      "step": 2780
    },
    {
      "epoch": 1.6692677070828332,
      "grad_norm": 0.0014490924077108502,
      "learning_rate": 0.0001867970880664624,
      "loss": 23.0,
      "step": 2781
    },
    {
      "epoch": 1.6698679471788715,
      "grad_norm": 0.0013336183037608862,
      "learning_rate": 0.00018678770490422437,
      "loss": 23.0,
      "step": 2782
    },
    {
      "epoch": 1.6704681872749099,
      "grad_norm": 0.0011176721891388297,
      "learning_rate": 0.00018677831864475635,
      "loss": 23.0,
      "step": 2783
    },
    {
      "epoch": 1.6710684273709484,
      "grad_norm": 0.0011959094554185867,
      "learning_rate": 0.0001867689292883934,
      "loss": 23.0,
      "step": 2784
    },
    {
      "epoch": 1.671668667466987,
      "grad_norm": 0.0013386125210672617,
      "learning_rate": 0.00018675953683547055,
      "loss": 23.0,
      "step": 2785
    },
    {
      "epoch": 1.6722689075630253,
      "grad_norm": 0.0008990365895442665,
      "learning_rate": 0.00018675014128632299,
      "loss": 23.0,
      "step": 2786
    },
    {
      "epoch": 1.6728691476590636,
      "grad_norm": 0.001523659797385335,
      "learning_rate": 0.00018674074264128602,
      "loss": 23.0,
      "step": 2787
    },
    {
      "epoch": 1.6734693877551021,
      "grad_norm": 0.0012871394865214825,
      "learning_rate": 0.0001867313409006951,
      "loss": 23.0,
      "step": 2788
    },
    {
      "epoch": 1.6740696278511404,
      "grad_norm": 0.0008513812208548188,
      "learning_rate": 0.0001867219360648857,
      "loss": 23.0,
      "step": 2789
    },
    {
      "epoch": 1.6746698679471788,
      "grad_norm": 0.0009285034029744565,
      "learning_rate": 0.0001867125281341935,
      "loss": 23.0,
      "step": 2790
    },
    {
      "epoch": 1.6752701080432173,
      "grad_norm": 0.0003783182764891535,
      "learning_rate": 0.0001867031171089542,
      "loss": 23.0,
      "step": 2791
    },
    {
      "epoch": 1.6758703481392558,
      "grad_norm": 0.0021908339112997055,
      "learning_rate": 0.00018669370298950372,
      "loss": 23.0,
      "step": 2792
    },
    {
      "epoch": 1.6764705882352942,
      "grad_norm": 0.002253185026347637,
      "learning_rate": 0.00018668428577617795,
      "loss": 23.0,
      "step": 2793
    },
    {
      "epoch": 1.6770708283313325,
      "grad_norm": 0.0012638723710551858,
      "learning_rate": 0.00018667486546931298,
      "loss": 23.0,
      "step": 2794
    },
    {
      "epoch": 1.677671068427371,
      "grad_norm": 0.0018435172969475389,
      "learning_rate": 0.00018666544206924506,
      "loss": 23.0,
      "step": 2795
    },
    {
      "epoch": 1.6782713085234093,
      "grad_norm": 0.0008160858415067196,
      "learning_rate": 0.0001866560155763104,
      "loss": 23.0,
      "step": 2796
    },
    {
      "epoch": 1.6788715486194477,
      "grad_norm": 0.001816743635572493,
      "learning_rate": 0.0001866465859908455,
      "loss": 23.0,
      "step": 2797
    },
    {
      "epoch": 1.6794717887154862,
      "grad_norm": 0.001861235941760242,
      "learning_rate": 0.00018663715331318682,
      "loss": 23.0,
      "step": 2798
    },
    {
      "epoch": 1.6800720288115247,
      "grad_norm": 0.0009585749357938766,
      "learning_rate": 0.00018662771754367094,
      "loss": 23.0,
      "step": 2799
    },
    {
      "epoch": 1.680672268907563,
      "grad_norm": 0.0016571884043514729,
      "learning_rate": 0.00018661827868263472,
      "loss": 23.0,
      "step": 2800
    },
    {
      "epoch": 1.6812725090036014,
      "grad_norm": 0.001067697536200285,
      "learning_rate": 0.00018660883673041488,
      "loss": 23.0,
      "step": 2801
    },
    {
      "epoch": 1.68187274909964,
      "grad_norm": 0.0011590105714276433,
      "learning_rate": 0.00018659939168734848,
      "loss": 23.0,
      "step": 2802
    },
    {
      "epoch": 1.6824729891956782,
      "grad_norm": 0.0012792943743988872,
      "learning_rate": 0.00018658994355377252,
      "loss": 23.0,
      "step": 2803
    },
    {
      "epoch": 1.6830732292917165,
      "grad_norm": 0.0009999220492318273,
      "learning_rate": 0.00018658049233002426,
      "loss": 23.0,
      "step": 2804
    },
    {
      "epoch": 1.683673469387755,
      "grad_norm": 0.0029290425591170788,
      "learning_rate": 0.00018657103801644092,
      "loss": 23.0,
      "step": 2805
    },
    {
      "epoch": 1.6842737094837936,
      "grad_norm": 0.0036934264935553074,
      "learning_rate": 0.0001865615806133599,
      "loss": 23.0,
      "step": 2806
    },
    {
      "epoch": 1.684873949579832,
      "grad_norm": 0.0015194531297311187,
      "learning_rate": 0.00018655212012111877,
      "loss": 23.0,
      "step": 2807
    },
    {
      "epoch": 1.6854741896758703,
      "grad_norm": 0.0013169290032237768,
      "learning_rate": 0.00018654265654005506,
      "loss": 23.0,
      "step": 2808
    },
    {
      "epoch": 1.6860744297719088,
      "grad_norm": 0.0026581338606774807,
      "learning_rate": 0.0001865331898705066,
      "loss": 23.0,
      "step": 2809
    },
    {
      "epoch": 1.6866746698679473,
      "grad_norm": 0.0007871690904721618,
      "learning_rate": 0.00018652372011281117,
      "loss": 23.0,
      "step": 2810
    },
    {
      "epoch": 1.6872749099639854,
      "grad_norm": 0.0032649554777890444,
      "learning_rate": 0.0001865142472673067,
      "loss": 23.0,
      "step": 2811
    },
    {
      "epoch": 1.687875150060024,
      "grad_norm": 0.0015880510909482837,
      "learning_rate": 0.0001865047713343313,
      "loss": 23.0,
      "step": 2812
    },
    {
      "epoch": 1.6884753901560625,
      "grad_norm": 0.002896341495215893,
      "learning_rate": 0.00018649529231422316,
      "loss": 23.0,
      "step": 2813
    },
    {
      "epoch": 1.6890756302521008,
      "grad_norm": 0.0030981532763689756,
      "learning_rate": 0.0001864858102073205,
      "loss": 23.0,
      "step": 2814
    },
    {
      "epoch": 1.6896758703481392,
      "grad_norm": 0.0014712786069139838,
      "learning_rate": 0.00018647632501396177,
      "loss": 23.0,
      "step": 2815
    },
    {
      "epoch": 1.6902761104441777,
      "grad_norm": 0.0037119940388947725,
      "learning_rate": 0.00018646683673448542,
      "loss": 23.0,
      "step": 2816
    },
    {
      "epoch": 1.6908763505402162,
      "grad_norm": 0.0017612777883186936,
      "learning_rate": 0.0001864573453692301,
      "loss": 23.0,
      "step": 2817
    },
    {
      "epoch": 1.6914765906362546,
      "grad_norm": 0.001388410571962595,
      "learning_rate": 0.0001864478509185345,
      "loss": 23.0,
      "step": 2818
    },
    {
      "epoch": 1.6920768307322929,
      "grad_norm": 0.0010633680503815413,
      "learning_rate": 0.0001864383533827375,
      "loss": 23.0,
      "step": 2819
    },
    {
      "epoch": 1.6926770708283314,
      "grad_norm": 0.004036045633256435,
      "learning_rate": 0.000186428852762178,
      "loss": 23.0,
      "step": 2820
    },
    {
      "epoch": 1.6932773109243697,
      "grad_norm": 0.0016166233690455556,
      "learning_rate": 0.00018641934905719505,
      "loss": 23.0,
      "step": 2821
    },
    {
      "epoch": 1.693877551020408,
      "grad_norm": 0.001802432001568377,
      "learning_rate": 0.00018640984226812787,
      "loss": 23.0,
      "step": 2822
    },
    {
      "epoch": 1.6944777911164466,
      "grad_norm": 0.0005948299658484757,
      "learning_rate": 0.00018640033239531565,
      "loss": 23.0,
      "step": 2823
    },
    {
      "epoch": 1.6950780312124851,
      "grad_norm": 0.0018249761778861284,
      "learning_rate": 0.00018639081943909783,
      "loss": 23.0,
      "step": 2824
    },
    {
      "epoch": 1.6956782713085234,
      "grad_norm": 0.001116413390263915,
      "learning_rate": 0.00018638130339981387,
      "loss": 23.0,
      "step": 2825
    },
    {
      "epoch": 1.6962785114045618,
      "grad_norm": 0.0015681873774155974,
      "learning_rate": 0.00018637178427780343,
      "loss": 23.0,
      "step": 2826
    },
    {
      "epoch": 1.6968787515006003,
      "grad_norm": 0.0022008116357028484,
      "learning_rate": 0.00018636226207340614,
      "loss": 23.0,
      "step": 2827
    },
    {
      "epoch": 1.6974789915966386,
      "grad_norm": 0.0006623781519010663,
      "learning_rate": 0.0001863527367869619,
      "loss": 23.0,
      "step": 2828
    },
    {
      "epoch": 1.698079231692677,
      "grad_norm": 0.002556073712185025,
      "learning_rate": 0.00018634320841881059,
      "loss": 23.0,
      "step": 2829
    },
    {
      "epoch": 1.6986794717887155,
      "grad_norm": 0.0012070855591446161,
      "learning_rate": 0.0001863336769692923,
      "loss": 23.0,
      "step": 2830
    },
    {
      "epoch": 1.699279711884754,
      "grad_norm": 0.0018783850828185678,
      "learning_rate": 0.00018632414243874713,
      "loss": 23.0,
      "step": 2831
    },
    {
      "epoch": 1.6998799519807923,
      "grad_norm": 0.0010801813332363963,
      "learning_rate": 0.00018631460482751536,
      "loss": 23.0,
      "step": 2832
    },
    {
      "epoch": 1.7004801920768307,
      "grad_norm": 0.0010861080372706056,
      "learning_rate": 0.0001863050641359374,
      "loss": 23.0,
      "step": 2833
    },
    {
      "epoch": 1.7010804321728692,
      "grad_norm": 0.0012606565142050385,
      "learning_rate": 0.00018629552036435372,
      "loss": 23.0,
      "step": 2834
    },
    {
      "epoch": 1.7016806722689075,
      "grad_norm": 0.002117091091349721,
      "learning_rate": 0.00018628597351310483,
      "loss": 23.0,
      "step": 2835
    },
    {
      "epoch": 1.7022809123649458,
      "grad_norm": 0.0013018603203818202,
      "learning_rate": 0.00018627642358253158,
      "loss": 23.0,
      "step": 2836
    },
    {
      "epoch": 1.7028811524609844,
      "grad_norm": 0.0006756331422366202,
      "learning_rate": 0.00018626687057297464,
      "loss": 23.0,
      "step": 2837
    },
    {
      "epoch": 1.703481392557023,
      "grad_norm": 0.0012766298605129123,
      "learning_rate": 0.000186257314484775,
      "loss": 23.0,
      "step": 2838
    },
    {
      "epoch": 1.7040816326530612,
      "grad_norm": 0.0006693665054626763,
      "learning_rate": 0.00018624775531827373,
      "loss": 23.0,
      "step": 2839
    },
    {
      "epoch": 1.7046818727490995,
      "grad_norm": 0.002290289616212249,
      "learning_rate": 0.00018623819307381188,
      "loss": 23.0,
      "step": 2840
    },
    {
      "epoch": 1.705282112845138,
      "grad_norm": 0.0008764312951825559,
      "learning_rate": 0.00018622862775173076,
      "loss": 23.0,
      "step": 2841
    },
    {
      "epoch": 1.7058823529411766,
      "grad_norm": 0.0008235902641899884,
      "learning_rate": 0.00018621905935237173,
      "loss": 23.0,
      "step": 2842
    },
    {
      "epoch": 1.7064825930372147,
      "grad_norm": 0.0006964323110878468,
      "learning_rate": 0.00018620948787607627,
      "loss": 23.0,
      "step": 2843
    },
    {
      "epoch": 1.7070828331332533,
      "grad_norm": 0.0005906512960791588,
      "learning_rate": 0.00018619991332318594,
      "loss": 23.0,
      "step": 2844
    },
    {
      "epoch": 1.7076830732292918,
      "grad_norm": 0.002924280008301139,
      "learning_rate": 0.0001861903356940424,
      "loss": 23.0,
      "step": 2845
    },
    {
      "epoch": 1.7082833133253301,
      "grad_norm": 0.000374676805222407,
      "learning_rate": 0.00018618075498898752,
      "loss": 23.0,
      "step": 2846
    },
    {
      "epoch": 1.7088835534213684,
      "grad_norm": 0.0012125116772949696,
      "learning_rate": 0.0001861711712083632,
      "loss": 23.0,
      "step": 2847
    },
    {
      "epoch": 1.709483793517407,
      "grad_norm": 0.00048197314026765525,
      "learning_rate": 0.00018616158435251137,
      "loss": 23.0,
      "step": 2848
    },
    {
      "epoch": 1.7100840336134455,
      "grad_norm": 0.0022257508244365454,
      "learning_rate": 0.00018615199442177427,
      "loss": 23.0,
      "step": 2849
    },
    {
      "epoch": 1.7106842737094838,
      "grad_norm": 0.0027254424057900906,
      "learning_rate": 0.0001861424014164941,
      "loss": 23.0,
      "step": 2850
    },
    {
      "epoch": 1.7112845138055222,
      "grad_norm": 0.002013017423450947,
      "learning_rate": 0.0001861328053370132,
      "loss": 23.0,
      "step": 2851
    },
    {
      "epoch": 1.7118847539015607,
      "grad_norm": 0.000818777596578002,
      "learning_rate": 0.00018612320618367404,
      "loss": 23.0,
      "step": 2852
    },
    {
      "epoch": 1.712484993997599,
      "grad_norm": 0.002587414812296629,
      "learning_rate": 0.0001861136039568192,
      "loss": 23.0,
      "step": 2853
    },
    {
      "epoch": 1.7130852340936373,
      "grad_norm": 0.001048891805112362,
      "learning_rate": 0.00018610399865679134,
      "loss": 23.0,
      "step": 2854
    },
    {
      "epoch": 1.7136854741896759,
      "grad_norm": 0.004490484017878771,
      "learning_rate": 0.00018609439028393325,
      "loss": 23.0,
      "step": 2855
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.0020678737200796604,
      "learning_rate": 0.00018608477883858782,
      "loss": 23.0,
      "step": 2856
    },
    {
      "epoch": 1.7148859543817527,
      "grad_norm": 0.0016980924410745502,
      "learning_rate": 0.00018607516432109805,
      "loss": 23.0,
      "step": 2857
    },
    {
      "epoch": 1.715486194477791,
      "grad_norm": 0.000979147502221167,
      "learning_rate": 0.00018606554673180714,
      "loss": 23.0,
      "step": 2858
    },
    {
      "epoch": 1.7160864345738296,
      "grad_norm": 0.0006189555278979242,
      "learning_rate": 0.00018605592607105818,
      "loss": 23.0,
      "step": 2859
    },
    {
      "epoch": 1.716686674669868,
      "grad_norm": 0.0011676882859319448,
      "learning_rate": 0.00018604630233919466,
      "loss": 23.0,
      "step": 2860
    },
    {
      "epoch": 1.7172869147659062,
      "grad_norm": 0.002290237694978714,
      "learning_rate": 0.0001860366755365599,
      "loss": 23.0,
      "step": 2861
    },
    {
      "epoch": 1.7178871548619448,
      "grad_norm": 0.0006044538458809257,
      "learning_rate": 0.00018602704566349752,
      "loss": 23.0,
      "step": 2862
    },
    {
      "epoch": 1.7184873949579833,
      "grad_norm": 0.0009377237292937934,
      "learning_rate": 0.00018601741272035116,
      "loss": 23.0,
      "step": 2863
    },
    {
      "epoch": 1.7190876350540216,
      "grad_norm": 0.0015994718996807933,
      "learning_rate": 0.00018600777670746462,
      "loss": 23.0,
      "step": 2864
    },
    {
      "epoch": 1.71968787515006,
      "grad_norm": 0.001292977947741747,
      "learning_rate": 0.00018599813762518172,
      "loss": 23.0,
      "step": 2865
    },
    {
      "epoch": 1.7202881152460985,
      "grad_norm": 0.002400642493739724,
      "learning_rate": 0.00018598849547384654,
      "loss": 23.0,
      "step": 2866
    },
    {
      "epoch": 1.7208883553421368,
      "grad_norm": 0.0013523305533453822,
      "learning_rate": 0.00018597885025380313,
      "loss": 23.0,
      "step": 2867
    },
    {
      "epoch": 1.7214885954381751,
      "grad_norm": 0.0012043274473398924,
      "learning_rate": 0.00018596920196539574,
      "loss": 23.0,
      "step": 2868
    },
    {
      "epoch": 1.7220888355342137,
      "grad_norm": 0.0021065007895231247,
      "learning_rate": 0.00018595955060896863,
      "loss": 23.0,
      "step": 2869
    },
    {
      "epoch": 1.7226890756302522,
      "grad_norm": 0.0005539829144254327,
      "learning_rate": 0.0001859498961848663,
      "loss": 23.0,
      "step": 2870
    },
    {
      "epoch": 1.7232893157262905,
      "grad_norm": 0.000724462850484997,
      "learning_rate": 0.0001859402386934333,
      "loss": 23.0,
      "step": 2871
    },
    {
      "epoch": 1.7238895558223288,
      "grad_norm": 0.001463893218897283,
      "learning_rate": 0.0001859305781350142,
      "loss": 23.0,
      "step": 2872
    },
    {
      "epoch": 1.7244897959183674,
      "grad_norm": 0.001887418096885085,
      "learning_rate": 0.0001859209145099538,
      "loss": 23.0,
      "step": 2873
    },
    {
      "epoch": 1.725090036014406,
      "grad_norm": 0.001024300348944962,
      "learning_rate": 0.000185911247818597,
      "loss": 23.0,
      "step": 2874
    },
    {
      "epoch": 1.7256902761104442,
      "grad_norm": 0.0024245274253189564,
      "learning_rate": 0.00018590157806128876,
      "loss": 23.0,
      "step": 2875
    },
    {
      "epoch": 1.7262905162064826,
      "grad_norm": 0.0017824191600084305,
      "learning_rate": 0.00018589190523837416,
      "loss": 23.0,
      "step": 2876
    },
    {
      "epoch": 1.726890756302521,
      "grad_norm": 0.0017186804907396436,
      "learning_rate": 0.00018588222935019837,
      "loss": 23.0,
      "step": 2877
    },
    {
      "epoch": 1.7274909963985594,
      "grad_norm": 0.001040443079546094,
      "learning_rate": 0.00018587255039710676,
      "loss": 23.0,
      "step": 2878
    },
    {
      "epoch": 1.7280912364945977,
      "grad_norm": 0.0011490770848467946,
      "learning_rate": 0.0001858628683794447,
      "loss": 23.0,
      "step": 2879
    },
    {
      "epoch": 1.7286914765906363,
      "grad_norm": 0.003117022570222616,
      "learning_rate": 0.00018585318329755775,
      "loss": 23.0,
      "step": 2880
    },
    {
      "epoch": 1.7292917166866748,
      "grad_norm": 0.0019710492342710495,
      "learning_rate": 0.00018584349515179152,
      "loss": 23.0,
      "step": 2881
    },
    {
      "epoch": 1.7298919567827131,
      "grad_norm": 0.0033849983010441065,
      "learning_rate": 0.00018583380394249178,
      "loss": 23.0,
      "step": 2882
    },
    {
      "epoch": 1.7304921968787514,
      "grad_norm": 0.0011864303378388286,
      "learning_rate": 0.00018582410967000438,
      "loss": 23.0,
      "step": 2883
    },
    {
      "epoch": 1.73109243697479,
      "grad_norm": 0.0011380382347851992,
      "learning_rate": 0.0001858144123346752,
      "loss": 23.0,
      "step": 2884
    },
    {
      "epoch": 1.7316926770708283,
      "grad_norm": 0.0026864493265748024,
      "learning_rate": 0.00018580471193685045,
      "loss": 23.0,
      "step": 2885
    },
    {
      "epoch": 1.7322929171668666,
      "grad_norm": 0.001236348645761609,
      "learning_rate": 0.00018579500847687623,
      "loss": 23.0,
      "step": 2886
    },
    {
      "epoch": 1.7328931572629052,
      "grad_norm": 0.0026629730127751827,
      "learning_rate": 0.00018578530195509883,
      "loss": 23.0,
      "step": 2887
    },
    {
      "epoch": 1.7334933973589437,
      "grad_norm": 0.003911031875759363,
      "learning_rate": 0.0001857755923718647,
      "loss": 23.0,
      "step": 2888
    },
    {
      "epoch": 1.734093637454982,
      "grad_norm": 0.0026813908480107784,
      "learning_rate": 0.0001857658797275203,
      "loss": 23.0,
      "step": 2889
    },
    {
      "epoch": 1.7346938775510203,
      "grad_norm": 0.001345346332527697,
      "learning_rate": 0.0001857561640224123,
      "loss": 23.0,
      "step": 2890
    },
    {
      "epoch": 1.7352941176470589,
      "grad_norm": 0.0010961355874314904,
      "learning_rate": 0.00018574644525688738,
      "loss": 23.0,
      "step": 2891
    },
    {
      "epoch": 1.7358943577430972,
      "grad_norm": 0.0021427536848932505,
      "learning_rate": 0.00018573672343129237,
      "loss": 23.0,
      "step": 2892
    },
    {
      "epoch": 1.7364945978391355,
      "grad_norm": 0.0031297134701162577,
      "learning_rate": 0.00018572699854597425,
      "loss": 23.0,
      "step": 2893
    },
    {
      "epoch": 1.737094837935174,
      "grad_norm": 0.0006432196823880076,
      "learning_rate": 0.0001857172706012801,
      "loss": 23.0,
      "step": 2894
    },
    {
      "epoch": 1.7376950780312126,
      "grad_norm": 0.0024129992816597223,
      "learning_rate": 0.00018570753959755703,
      "loss": 23.0,
      "step": 2895
    },
    {
      "epoch": 1.738295318127251,
      "grad_norm": 0.0017073117196559906,
      "learning_rate": 0.00018569780553515233,
      "loss": 23.0,
      "step": 2896
    },
    {
      "epoch": 1.7388955582232892,
      "grad_norm": 0.0020329838152974844,
      "learning_rate": 0.00018568806841441342,
      "loss": 23.0,
      "step": 2897
    },
    {
      "epoch": 1.7394957983193278,
      "grad_norm": 0.0009188112453557551,
      "learning_rate": 0.00018567832823568773,
      "loss": 23.0,
      "step": 2898
    },
    {
      "epoch": 1.7400960384153663,
      "grad_norm": 0.001637371489778161,
      "learning_rate": 0.00018566858499932288,
      "loss": 23.0,
      "step": 2899
    },
    {
      "epoch": 1.7406962785114044,
      "grad_norm": 0.0016789011424407363,
      "learning_rate": 0.00018565883870566666,
      "loss": 23.0,
      "step": 2900
    },
    {
      "epoch": 1.741296518607443,
      "grad_norm": 0.0010591614991426468,
      "learning_rate": 0.0001856490893550668,
      "loss": 23.0,
      "step": 2901
    },
    {
      "epoch": 1.7418967587034815,
      "grad_norm": 0.004590094089508057,
      "learning_rate": 0.00018563933694787125,
      "loss": 23.0,
      "step": 2902
    },
    {
      "epoch": 1.7424969987995198,
      "grad_norm": 0.0015041956212371588,
      "learning_rate": 0.00018562958148442803,
      "loss": 23.0,
      "step": 2903
    },
    {
      "epoch": 1.7430972388955581,
      "grad_norm": 0.00048696741578169167,
      "learning_rate": 0.00018561982296508536,
      "loss": 23.0,
      "step": 2904
    },
    {
      "epoch": 1.7436974789915967,
      "grad_norm": 0.0018151112599298358,
      "learning_rate": 0.00018561006139019144,
      "loss": 23.0,
      "step": 2905
    },
    {
      "epoch": 1.7442977190876352,
      "grad_norm": 0.00167025753762573,
      "learning_rate": 0.00018560029676009462,
      "loss": 23.0,
      "step": 2906
    },
    {
      "epoch": 1.7448979591836735,
      "grad_norm": 0.0013312215451151133,
      "learning_rate": 0.00018559052907514341,
      "loss": 23.0,
      "step": 2907
    },
    {
      "epoch": 1.7454981992797118,
      "grad_norm": 0.002203074749559164,
      "learning_rate": 0.00018558075833568637,
      "loss": 23.0,
      "step": 2908
    },
    {
      "epoch": 1.7460984393757504,
      "grad_norm": 0.0014002182288095355,
      "learning_rate": 0.00018557098454207225,
      "loss": 23.0,
      "step": 2909
    },
    {
      "epoch": 1.7466986794717887,
      "grad_norm": 0.000825204944703728,
      "learning_rate": 0.00018556120769464975,
      "loss": 23.0,
      "step": 2910
    },
    {
      "epoch": 1.747298919567827,
      "grad_norm": 0.001794505980797112,
      "learning_rate": 0.00018555142779376786,
      "loss": 23.0,
      "step": 2911
    },
    {
      "epoch": 1.7478991596638656,
      "grad_norm": 0.0010396147845312953,
      "learning_rate": 0.00018554164483977558,
      "loss": 23.0,
      "step": 2912
    },
    {
      "epoch": 1.748499399759904,
      "grad_norm": 0.0007051727152429521,
      "learning_rate": 0.00018553185883302207,
      "loss": 23.0,
      "step": 2913
    },
    {
      "epoch": 1.7490996398559424,
      "grad_norm": 0.001316645066253841,
      "learning_rate": 0.00018552206977385648,
      "loss": 23.0,
      "step": 2914
    },
    {
      "epoch": 1.7496998799519807,
      "grad_norm": 0.0025363380555063486,
      "learning_rate": 0.00018551227766262822,
      "loss": 23.0,
      "step": 2915
    },
    {
      "epoch": 1.7503001200480193,
      "grad_norm": 0.0016197512159124017,
      "learning_rate": 0.00018550248249968673,
      "loss": 23.0,
      "step": 2916
    },
    {
      "epoch": 1.7509003601440576,
      "grad_norm": 0.0016571108717471361,
      "learning_rate": 0.0001854926842853816,
      "loss": 23.0,
      "step": 2917
    },
    {
      "epoch": 1.751500600240096,
      "grad_norm": 0.002475859597325325,
      "learning_rate": 0.00018548288302006248,
      "loss": 23.0,
      "step": 2918
    },
    {
      "epoch": 1.7521008403361344,
      "grad_norm": 0.0017440817318856716,
      "learning_rate": 0.00018547307870407913,
      "loss": 23.0,
      "step": 2919
    },
    {
      "epoch": 1.752701080432173,
      "grad_norm": 0.0030700908973813057,
      "learning_rate": 0.00018546327133778148,
      "loss": 23.0,
      "step": 2920
    },
    {
      "epoch": 1.7533013205282113,
      "grad_norm": 0.0023228423669934273,
      "learning_rate": 0.00018545346092151949,
      "loss": 23.0,
      "step": 2921
    },
    {
      "epoch": 1.7539015606242496,
      "grad_norm": 0.0006581857451237738,
      "learning_rate": 0.00018544364745564333,
      "loss": 23.0,
      "step": 2922
    },
    {
      "epoch": 1.7545018007202882,
      "grad_norm": 0.0017514914507046342,
      "learning_rate": 0.00018543383094050318,
      "loss": 23.0,
      "step": 2923
    },
    {
      "epoch": 1.7551020408163265,
      "grad_norm": 0.0010599003871902823,
      "learning_rate": 0.00018542401137644935,
      "loss": 23.0,
      "step": 2924
    },
    {
      "epoch": 1.7557022809123648,
      "grad_norm": 0.0017716344445943832,
      "learning_rate": 0.00018541418876383227,
      "loss": 23.0,
      "step": 2925
    },
    {
      "epoch": 1.7563025210084033,
      "grad_norm": 0.0004864973889198154,
      "learning_rate": 0.00018540436310300252,
      "loss": 23.0,
      "step": 2926
    },
    {
      "epoch": 1.7569027611044419,
      "grad_norm": 0.001132338191382587,
      "learning_rate": 0.00018539453439431072,
      "loss": 23.0,
      "step": 2927
    },
    {
      "epoch": 1.7575030012004802,
      "grad_norm": 0.0017699715681374073,
      "learning_rate": 0.0001853847026381077,
      "loss": 23.0,
      "step": 2928
    },
    {
      "epoch": 1.7581032412965185,
      "grad_norm": 0.0012988667003810406,
      "learning_rate": 0.00018537486783474423,
      "loss": 23.0,
      "step": 2929
    },
    {
      "epoch": 1.758703481392557,
      "grad_norm": 0.0019157581264153123,
      "learning_rate": 0.00018536502998457138,
      "loss": 23.0,
      "step": 2930
    },
    {
      "epoch": 1.7593037214885956,
      "grad_norm": 0.0022223247215151787,
      "learning_rate": 0.00018535518908794013,
      "loss": 23.0,
      "step": 2931
    },
    {
      "epoch": 1.7599039615846337,
      "grad_norm": 0.002175229834392667,
      "learning_rate": 0.00018534534514520182,
      "loss": 23.0,
      "step": 2932
    },
    {
      "epoch": 1.7605042016806722,
      "grad_norm": 0.0016925784293562174,
      "learning_rate": 0.0001853354981567076,
      "loss": 23.0,
      "step": 2933
    },
    {
      "epoch": 1.7611044417767108,
      "grad_norm": 0.001816912554204464,
      "learning_rate": 0.00018532564812280905,
      "loss": 23.0,
      "step": 2934
    },
    {
      "epoch": 1.761704681872749,
      "grad_norm": 0.0013755745021626353,
      "learning_rate": 0.00018531579504385751,
      "loss": 23.0,
      "step": 2935
    },
    {
      "epoch": 1.7623049219687874,
      "grad_norm": 0.0008345983806066215,
      "learning_rate": 0.00018530593892020477,
      "loss": 23.0,
      "step": 2936
    },
    {
      "epoch": 1.762905162064826,
      "grad_norm": 0.0022048347163945436,
      "learning_rate": 0.00018529607975220245,
      "loss": 23.0,
      "step": 2937
    },
    {
      "epoch": 1.7635054021608645,
      "grad_norm": 0.001479812664911151,
      "learning_rate": 0.0001852862175402025,
      "loss": 23.0,
      "step": 2938
    },
    {
      "epoch": 1.7641056422569028,
      "grad_norm": 0.000645735184662044,
      "learning_rate": 0.00018527635228455678,
      "loss": 23.0,
      "step": 2939
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.0014726044610142708,
      "learning_rate": 0.00018526648398561744,
      "loss": 23.0,
      "step": 2940
    },
    {
      "epoch": 1.7653061224489797,
      "grad_norm": 0.00045085468445904553,
      "learning_rate": 0.0001852566126437366,
      "loss": 23.0,
      "step": 2941
    },
    {
      "epoch": 1.765906362545018,
      "grad_norm": 0.0018929915968328714,
      "learning_rate": 0.00018524673825926657,
      "loss": 23.0,
      "step": 2942
    },
    {
      "epoch": 1.7665066026410563,
      "grad_norm": 0.0010176091454923153,
      "learning_rate": 0.00018523686083255972,
      "loss": 23.0,
      "step": 2943
    },
    {
      "epoch": 1.7671068427370948,
      "grad_norm": 0.0013840744504705071,
      "learning_rate": 0.00018522698036396856,
      "loss": 23.0,
      "step": 2944
    },
    {
      "epoch": 1.7677070828331334,
      "grad_norm": 0.001135845435783267,
      "learning_rate": 0.00018521709685384572,
      "loss": 23.0,
      "step": 2945
    },
    {
      "epoch": 1.7683073229291717,
      "grad_norm": 0.0012144126230850816,
      "learning_rate": 0.0001852072103025439,
      "loss": 23.0,
      "step": 2946
    },
    {
      "epoch": 1.76890756302521,
      "grad_norm": 0.0015819761902093887,
      "learning_rate": 0.00018519732071041594,
      "loss": 23.0,
      "step": 2947
    },
    {
      "epoch": 1.7695078031212486,
      "grad_norm": 0.0015256191836670041,
      "learning_rate": 0.00018518742807781472,
      "loss": 23.0,
      "step": 2948
    },
    {
      "epoch": 1.7701080432172869,
      "grad_norm": 0.0010233953362330794,
      "learning_rate": 0.0001851775324050933,
      "loss": 23.0,
      "step": 2949
    },
    {
      "epoch": 1.7707082833133252,
      "grad_norm": 0.0009597190655767918,
      "learning_rate": 0.00018516763369260493,
      "loss": 23.0,
      "step": 2950
    },
    {
      "epoch": 1.7713085234093637,
      "grad_norm": 0.0007810408133082092,
      "learning_rate": 0.0001851577319407027,
      "loss": 23.0,
      "step": 2951
    },
    {
      "epoch": 1.7719087635054023,
      "grad_norm": 0.0017629596404731274,
      "learning_rate": 0.00018514782714974012,
      "loss": 23.0,
      "step": 2952
    },
    {
      "epoch": 1.7725090036014406,
      "grad_norm": 0.0036176845896989107,
      "learning_rate": 0.00018513791932007063,
      "loss": 23.0,
      "step": 2953
    },
    {
      "epoch": 1.773109243697479,
      "grad_norm": 0.0021507435012608767,
      "learning_rate": 0.0001851280084520478,
      "loss": 23.0,
      "step": 2954
    },
    {
      "epoch": 1.7737094837935174,
      "grad_norm": 0.0048975893296301365,
      "learning_rate": 0.00018511809454602526,
      "loss": 23.0,
      "step": 2955
    },
    {
      "epoch": 1.7743097238895558,
      "grad_norm": 0.0016850210959091783,
      "learning_rate": 0.00018510817760235694,
      "loss": 23.0,
      "step": 2956
    },
    {
      "epoch": 1.774909963985594,
      "grad_norm": 0.0007171302568167448,
      "learning_rate": 0.00018509825762139665,
      "loss": 23.0,
      "step": 2957
    },
    {
      "epoch": 1.7755102040816326,
      "grad_norm": 0.0011437806533649564,
      "learning_rate": 0.00018508833460349847,
      "loss": 23.0,
      "step": 2958
    },
    {
      "epoch": 1.7761104441776712,
      "grad_norm": 0.003833747934550047,
      "learning_rate": 0.00018507840854901648,
      "loss": 23.0,
      "step": 2959
    },
    {
      "epoch": 1.7767106842737095,
      "grad_norm": 0.0019063903018832207,
      "learning_rate": 0.00018506847945830494,
      "loss": 23.0,
      "step": 2960
    },
    {
      "epoch": 1.7773109243697478,
      "grad_norm": 0.0016151851741597056,
      "learning_rate": 0.0001850585473317182,
      "loss": 23.0,
      "step": 2961
    },
    {
      "epoch": 1.7779111644657863,
      "grad_norm": 0.004216872155666351,
      "learning_rate": 0.0001850486121696107,
      "loss": 23.0,
      "step": 2962
    },
    {
      "epoch": 1.7785114045618249,
      "grad_norm": 0.0011847265996038914,
      "learning_rate": 0.000185038673972337,
      "loss": 23.0,
      "step": 2963
    },
    {
      "epoch": 1.7791116446578632,
      "grad_norm": 0.0010568160796537995,
      "learning_rate": 0.00018502873274025178,
      "loss": 23.0,
      "step": 2964
    },
    {
      "epoch": 1.7797118847539015,
      "grad_norm": 0.001491167233325541,
      "learning_rate": 0.00018501878847370978,
      "loss": 23.0,
      "step": 2965
    },
    {
      "epoch": 1.78031212484994,
      "grad_norm": 0.000517088919878006,
      "learning_rate": 0.00018500884117306595,
      "loss": 23.0,
      "step": 2966
    },
    {
      "epoch": 1.7809123649459784,
      "grad_norm": 0.0025163828395307064,
      "learning_rate": 0.00018499889083867523,
      "loss": 23.0,
      "step": 2967
    },
    {
      "epoch": 1.7815126050420167,
      "grad_norm": 0.0018170212861150503,
      "learning_rate": 0.00018498893747089277,
      "loss": 23.0,
      "step": 2968
    },
    {
      "epoch": 1.7821128451380552,
      "grad_norm": 0.0015163187636062503,
      "learning_rate": 0.0001849789810700737,
      "loss": 23.0,
      "step": 2969
    },
    {
      "epoch": 1.7827130852340938,
      "grad_norm": 0.003132780082523823,
      "learning_rate": 0.00018496902163657342,
      "loss": 23.0,
      "step": 2970
    },
    {
      "epoch": 1.783313325330132,
      "grad_norm": 0.001669717370532453,
      "learning_rate": 0.00018495905917074732,
      "loss": 23.0,
      "step": 2971
    },
    {
      "epoch": 1.7839135654261704,
      "grad_norm": 0.0005561247817240655,
      "learning_rate": 0.00018494909367295091,
      "loss": 23.0,
      "step": 2972
    },
    {
      "epoch": 1.784513805522209,
      "grad_norm": 0.0017999939154833555,
      "learning_rate": 0.0001849391251435399,
      "loss": 23.0,
      "step": 2973
    },
    {
      "epoch": 1.7851140456182473,
      "grad_norm": 0.0021132882684469223,
      "learning_rate": 0.00018492915358286996,
      "loss": 23.0,
      "step": 2974
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 0.003159108804538846,
      "learning_rate": 0.00018491917899129704,
      "loss": 23.0,
      "step": 2975
    },
    {
      "epoch": 1.7863145258103241,
      "grad_norm": 0.0003376493405085057,
      "learning_rate": 0.00018490920136917706,
      "loss": 23.0,
      "step": 2976
    },
    {
      "epoch": 1.7869147659063627,
      "grad_norm": 0.001667432370595634,
      "learning_rate": 0.00018489922071686604,
      "loss": 23.0,
      "step": 2977
    },
    {
      "epoch": 1.787515006002401,
      "grad_norm": 0.0015298868529498577,
      "learning_rate": 0.0001848892370347203,
      "loss": 23.0,
      "step": 2978
    },
    {
      "epoch": 1.7881152460984393,
      "grad_norm": 0.0008003988186828792,
      "learning_rate": 0.00018487925032309598,
      "loss": 23.0,
      "step": 2979
    },
    {
      "epoch": 1.7887154861944778,
      "grad_norm": 0.00343932188116014,
      "learning_rate": 0.00018486926058234956,
      "loss": 23.0,
      "step": 2980
    },
    {
      "epoch": 1.7893157262905162,
      "grad_norm": 0.0029429318383336067,
      "learning_rate": 0.00018485926781283758,
      "loss": 23.0,
      "step": 2981
    },
    {
      "epoch": 1.7899159663865545,
      "grad_norm": 0.00047122538671828806,
      "learning_rate": 0.0001848492720149166,
      "loss": 23.0,
      "step": 2982
    },
    {
      "epoch": 1.790516206482593,
      "grad_norm": 0.0006141350604593754,
      "learning_rate": 0.00018483927318894337,
      "loss": 23.0,
      "step": 2983
    },
    {
      "epoch": 1.7911164465786316,
      "grad_norm": 0.0007275057141669095,
      "learning_rate": 0.00018482927133527469,
      "loss": 23.0,
      "step": 2984
    },
    {
      "epoch": 1.7917166866746699,
      "grad_norm": 0.0007060296484269202,
      "learning_rate": 0.00018481926645426752,
      "loss": 23.0,
      "step": 2985
    },
    {
      "epoch": 1.7923169267707082,
      "grad_norm": 0.0014618011191487312,
      "learning_rate": 0.00018480925854627892,
      "loss": 23.0,
      "step": 2986
    },
    {
      "epoch": 1.7929171668667467,
      "grad_norm": 0.0011972852516919374,
      "learning_rate": 0.00018479924761166604,
      "loss": 23.0,
      "step": 2987
    },
    {
      "epoch": 1.7935174069627853,
      "grad_norm": 0.002459679963067174,
      "learning_rate": 0.00018478923365078612,
      "loss": 23.0,
      "step": 2988
    },
    {
      "epoch": 1.7941176470588234,
      "grad_norm": 0.0012211776338517666,
      "learning_rate": 0.0001847792166639966,
      "loss": 23.0,
      "step": 2989
    },
    {
      "epoch": 1.794717887154862,
      "grad_norm": 0.000426774233346805,
      "learning_rate": 0.00018476919665165489,
      "loss": 23.0,
      "step": 2990
    },
    {
      "epoch": 1.7953181272509005,
      "grad_norm": 0.001523893210105598,
      "learning_rate": 0.00018475917361411858,
      "loss": 23.0,
      "step": 2991
    },
    {
      "epoch": 1.7959183673469388,
      "grad_norm": 0.001041338429786265,
      "learning_rate": 0.00018474914755174543,
      "loss": 23.0,
      "step": 2992
    },
    {
      "epoch": 1.796518607442977,
      "grad_norm": 0.0010472668800503016,
      "learning_rate": 0.00018473911846489318,
      "loss": 23.0,
      "step": 2993
    },
    {
      "epoch": 1.7971188475390156,
      "grad_norm": 0.0013296904508024454,
      "learning_rate": 0.0001847290863539198,
      "loss": 23.0,
      "step": 2994
    },
    {
      "epoch": 1.7977190876350542,
      "grad_norm": 0.001468316069804132,
      "learning_rate": 0.00018471905121918324,
      "loss": 23.0,
      "step": 2995
    },
    {
      "epoch": 1.7983193277310925,
      "grad_norm": 0.003528536995872855,
      "learning_rate": 0.0001847090130610417,
      "loss": 23.0,
      "step": 2996
    },
    {
      "epoch": 1.7989195678271308,
      "grad_norm": 0.0017783723305910826,
      "learning_rate": 0.00018469897187985336,
      "loss": 23.0,
      "step": 2997
    },
    {
      "epoch": 1.7995198079231693,
      "grad_norm": 0.005629006773233414,
      "learning_rate": 0.00018468892767597658,
      "loss": 23.0,
      "step": 2998
    },
    {
      "epoch": 1.8001200480192077,
      "grad_norm": 0.0033181149046868086,
      "learning_rate": 0.0001846788804497698,
      "loss": 23.0,
      "step": 2999
    },
    {
      "epoch": 1.800720288115246,
      "grad_norm": 0.001119964406825602,
      "learning_rate": 0.00018466883020159164,
      "loss": 23.0,
      "step": 3000
    },
    {
      "epoch": 1.800720288115246,
      "eval_loss": 11.5,
      "eval_runtime": 5.3647,
      "eval_samples_per_second": 261.522,
      "eval_steps_per_second": 32.807,
      "step": 3000
    },
    {
      "epoch": 1.8013205282112845,
      "grad_norm": 0.0011961712734773755,
      "learning_rate": 0.00018465877693180072,
      "loss": 23.0,
      "step": 3001
    },
    {
      "epoch": 1.801920768307323,
      "grad_norm": 0.0015885218745097518,
      "learning_rate": 0.00018464872064075577,
      "loss": 23.0,
      "step": 3002
    },
    {
      "epoch": 1.8025210084033614,
      "grad_norm": 0.0019562295638024807,
      "learning_rate": 0.00018463866132881578,
      "loss": 23.0,
      "step": 3003
    },
    {
      "epoch": 1.8031212484993997,
      "grad_norm": 0.003253971692174673,
      "learning_rate": 0.00018462859899633967,
      "loss": 23.0,
      "step": 3004
    },
    {
      "epoch": 1.8037214885954382,
      "grad_norm": 0.000821275869384408,
      "learning_rate": 0.00018461853364368653,
      "loss": 23.0,
      "step": 3005
    },
    {
      "epoch": 1.8043217286914766,
      "grad_norm": 0.0012803277932107449,
      "learning_rate": 0.0001846084652712156,
      "loss": 23.0,
      "step": 3006
    },
    {
      "epoch": 1.8049219687875149,
      "grad_norm": 0.0007911855354905128,
      "learning_rate": 0.00018459839387928619,
      "loss": 23.0,
      "step": 3007
    },
    {
      "epoch": 1.8055222088835534,
      "grad_norm": 0.0007396016153506935,
      "learning_rate": 0.0001845883194682577,
      "loss": 23.0,
      "step": 3008
    },
    {
      "epoch": 1.806122448979592,
      "grad_norm": 0.0014897441724315286,
      "learning_rate": 0.0001845782420384897,
      "loss": 23.0,
      "step": 3009
    },
    {
      "epoch": 1.8067226890756303,
      "grad_norm": 0.0004990187590010464,
      "learning_rate": 0.0001845681615903418,
      "loss": 23.0,
      "step": 3010
    },
    {
      "epoch": 1.8073229291716686,
      "grad_norm": 0.0008598102140240371,
      "learning_rate": 0.00018455807812417372,
      "loss": 23.0,
      "step": 3011
    },
    {
      "epoch": 1.8079231692677071,
      "grad_norm": 0.0011542581487447023,
      "learning_rate": 0.00018454799164034538,
      "loss": 23.0,
      "step": 3012
    },
    {
      "epoch": 1.8085234093637454,
      "grad_norm": 0.0013645098078995943,
      "learning_rate": 0.00018453790213921665,
      "loss": 23.0,
      "step": 3013
    },
    {
      "epoch": 1.8091236494597838,
      "grad_norm": 0.0008868533186614513,
      "learning_rate": 0.00018452780962114772,
      "loss": 23.0,
      "step": 3014
    },
    {
      "epoch": 1.8097238895558223,
      "grad_norm": 0.0018087849020957947,
      "learning_rate": 0.00018451771408649865,
      "loss": 23.0,
      "step": 3015
    },
    {
      "epoch": 1.8103241296518608,
      "grad_norm": 0.00035064207622781396,
      "learning_rate": 0.00018450761553562975,
      "loss": 23.0,
      "step": 3016
    },
    {
      "epoch": 1.8109243697478992,
      "grad_norm": 0.0014904963318258524,
      "learning_rate": 0.00018449751396890148,
      "loss": 23.0,
      "step": 3017
    },
    {
      "epoch": 1.8115246098439375,
      "grad_norm": 0.0019730026833713055,
      "learning_rate": 0.00018448740938667427,
      "loss": 23.0,
      "step": 3018
    },
    {
      "epoch": 1.812124849939976,
      "grad_norm": 0.0017877138452604413,
      "learning_rate": 0.00018447730178930874,
      "loss": 23.0,
      "step": 3019
    },
    {
      "epoch": 1.8127250900360146,
      "grad_norm": 0.0008809540886431932,
      "learning_rate": 0.0001844671911771656,
      "loss": 23.0,
      "step": 3020
    },
    {
      "epoch": 1.8133253301320527,
      "grad_norm": 0.0015870918286964297,
      "learning_rate": 0.0001844570775506057,
      "loss": 23.0,
      "step": 3021
    },
    {
      "epoch": 1.8139255702280912,
      "grad_norm": 0.001148207113146782,
      "learning_rate": 0.00018444696090998995,
      "loss": 23.0,
      "step": 3022
    },
    {
      "epoch": 1.8145258103241297,
      "grad_norm": 0.0025478298775851727,
      "learning_rate": 0.00018443684125567938,
      "loss": 23.0,
      "step": 3023
    },
    {
      "epoch": 1.815126050420168,
      "grad_norm": 0.0015544770285487175,
      "learning_rate": 0.00018442671858803514,
      "loss": 23.0,
      "step": 3024
    },
    {
      "epoch": 1.8157262905162064,
      "grad_norm": 0.001434922218322754,
      "learning_rate": 0.00018441659290741852,
      "loss": 23.0,
      "step": 3025
    },
    {
      "epoch": 1.816326530612245,
      "grad_norm": 0.0009723508264869452,
      "learning_rate": 0.00018440646421419085,
      "loss": 23.0,
      "step": 3026
    },
    {
      "epoch": 1.8169267707082835,
      "grad_norm": 0.0012221380602568388,
      "learning_rate": 0.00018439633250871354,
      "loss": 23.0,
      "step": 3027
    },
    {
      "epoch": 1.8175270108043218,
      "grad_norm": 0.0012381580891087651,
      "learning_rate": 0.00018438619779134822,
      "loss": 23.0,
      "step": 3028
    },
    {
      "epoch": 1.81812725090036,
      "grad_norm": 0.0011028312146663666,
      "learning_rate": 0.0001843760600624566,
      "loss": 23.0,
      "step": 3029
    },
    {
      "epoch": 1.8187274909963986,
      "grad_norm": 0.0011208378709852695,
      "learning_rate": 0.0001843659193224004,
      "loss": 23.0,
      "step": 3030
    },
    {
      "epoch": 1.819327731092437,
      "grad_norm": 0.001310601131990552,
      "learning_rate": 0.0001843557755715416,
      "loss": 23.0,
      "step": 3031
    },
    {
      "epoch": 1.8199279711884753,
      "grad_norm": 0.0017411686712875962,
      "learning_rate": 0.00018434562881024214,
      "loss": 23.0,
      "step": 3032
    },
    {
      "epoch": 1.8205282112845138,
      "grad_norm": 0.0012776413932442665,
      "learning_rate": 0.00018433547903886415,
      "loss": 23.0,
      "step": 3033
    },
    {
      "epoch": 1.8211284513805523,
      "grad_norm": 0.00228596362285316,
      "learning_rate": 0.00018432532625776987,
      "loss": 23.0,
      "step": 3034
    },
    {
      "epoch": 1.8217286914765907,
      "grad_norm": 0.0009363253484480083,
      "learning_rate": 0.00018431517046732157,
      "loss": 23.0,
      "step": 3035
    },
    {
      "epoch": 1.822328931572629,
      "grad_norm": 0.0019296659156680107,
      "learning_rate": 0.00018430501166788173,
      "loss": 23.0,
      "step": 3036
    },
    {
      "epoch": 1.8229291716686675,
      "grad_norm": 0.0037253268528729677,
      "learning_rate": 0.0001842948498598129,
      "loss": 23.0,
      "step": 3037
    },
    {
      "epoch": 1.8235294117647058,
      "grad_norm": 0.0003991881385445595,
      "learning_rate": 0.00018428468504347767,
      "loss": 23.0,
      "step": 3038
    },
    {
      "epoch": 1.8241296518607442,
      "grad_norm": 0.004006813745945692,
      "learning_rate": 0.00018427451721923888,
      "loss": 23.0,
      "step": 3039
    },
    {
      "epoch": 1.8247298919567827,
      "grad_norm": 0.0013116990448907018,
      "learning_rate": 0.00018426434638745933,
      "loss": 23.0,
      "step": 3040
    },
    {
      "epoch": 1.8253301320528212,
      "grad_norm": 0.0044224620796740055,
      "learning_rate": 0.000184254172548502,
      "loss": 23.0,
      "step": 3041
    },
    {
      "epoch": 1.8259303721488596,
      "grad_norm": 0.0010624263668432832,
      "learning_rate": 0.00018424399570273002,
      "loss": 23.0,
      "step": 3042
    },
    {
      "epoch": 1.8265306122448979,
      "grad_norm": 0.00032893690513446927,
      "learning_rate": 0.0001842338158505065,
      "loss": 23.0,
      "step": 3043
    },
    {
      "epoch": 1.8271308523409364,
      "grad_norm": 0.0007519045611843467,
      "learning_rate": 0.00018422363299219478,
      "loss": 23.0,
      "step": 3044
    },
    {
      "epoch": 1.8277310924369747,
      "grad_norm": 0.0011156450491398573,
      "learning_rate": 0.00018421344712815825,
      "loss": 23.0,
      "step": 3045
    },
    {
      "epoch": 1.828331332533013,
      "grad_norm": 0.0012068160576745868,
      "learning_rate": 0.0001842032582587604,
      "loss": 23.0,
      "step": 3046
    },
    {
      "epoch": 1.8289315726290516,
      "grad_norm": 0.00150686118286103,
      "learning_rate": 0.00018419306638436486,
      "loss": 23.0,
      "step": 3047
    },
    {
      "epoch": 1.8295318127250901,
      "grad_norm": 0.0015496856067329645,
      "learning_rate": 0.00018418287150533537,
      "loss": 23.0,
      "step": 3048
    },
    {
      "epoch": 1.8301320528211285,
      "grad_norm": 0.0019011717522516847,
      "learning_rate": 0.00018417267362203575,
      "loss": 23.0,
      "step": 3049
    },
    {
      "epoch": 1.8307322929171668,
      "grad_norm": 0.001324790413491428,
      "learning_rate": 0.00018416247273482988,
      "loss": 23.0,
      "step": 3050
    },
    {
      "epoch": 1.8313325330132053,
      "grad_norm": 0.0008508468163199723,
      "learning_rate": 0.0001841522688440819,
      "loss": 23.0,
      "step": 3051
    },
    {
      "epoch": 1.8319327731092439,
      "grad_norm": 0.0004634323122445494,
      "learning_rate": 0.00018414206195015587,
      "loss": 23.0,
      "step": 3052
    },
    {
      "epoch": 1.832533013205282,
      "grad_norm": 0.003283916972577572,
      "learning_rate": 0.0001841318520534161,
      "loss": 23.0,
      "step": 3053
    },
    {
      "epoch": 1.8331332533013205,
      "grad_norm": 0.0021354639902710915,
      "learning_rate": 0.00018412163915422695,
      "loss": 23.0,
      "step": 3054
    },
    {
      "epoch": 1.833733493397359,
      "grad_norm": 0.0010375322308391333,
      "learning_rate": 0.0001841114232529529,
      "loss": 23.0,
      "step": 3055
    },
    {
      "epoch": 1.8343337334933973,
      "grad_norm": 0.0026274758856743574,
      "learning_rate": 0.0001841012043499585,
      "loss": 23.0,
      "step": 3056
    },
    {
      "epoch": 1.8349339735894357,
      "grad_norm": 0.0008885942515917122,
      "learning_rate": 0.00018409098244560844,
      "loss": 23.0,
      "step": 3057
    },
    {
      "epoch": 1.8355342136854742,
      "grad_norm": 0.0019456876907497644,
      "learning_rate": 0.00018408075754026753,
      "loss": 23.0,
      "step": 3058
    },
    {
      "epoch": 1.8361344537815127,
      "grad_norm": 0.002114507369697094,
      "learning_rate": 0.00018407052963430068,
      "loss": 23.0,
      "step": 3059
    },
    {
      "epoch": 1.836734693877551,
      "grad_norm": 0.0008166294428519905,
      "learning_rate": 0.00018406029872807287,
      "loss": 23.0,
      "step": 3060
    },
    {
      "epoch": 1.8373349339735894,
      "grad_norm": 0.0016927891410887241,
      "learning_rate": 0.00018405006482194921,
      "loss": 23.0,
      "step": 3061
    },
    {
      "epoch": 1.837935174069628,
      "grad_norm": 0.0014067868469282985,
      "learning_rate": 0.00018403982791629497,
      "loss": 23.0,
      "step": 3062
    },
    {
      "epoch": 1.8385354141656662,
      "grad_norm": 0.0017834990285336971,
      "learning_rate": 0.00018402958801147542,
      "loss": 23.0,
      "step": 3063
    },
    {
      "epoch": 1.8391356542617046,
      "grad_norm": 0.0011745449155569077,
      "learning_rate": 0.00018401934510785606,
      "loss": 23.0,
      "step": 3064
    },
    {
      "epoch": 1.839735894357743,
      "grad_norm": 0.0009545880602672696,
      "learning_rate": 0.00018400909920580238,
      "loss": 23.0,
      "step": 3065
    },
    {
      "epoch": 1.8403361344537816,
      "grad_norm": 0.003715930739417672,
      "learning_rate": 0.00018399885030568003,
      "loss": 23.0,
      "step": 3066
    },
    {
      "epoch": 1.84093637454982,
      "grad_norm": 0.0006399541744031012,
      "learning_rate": 0.0001839885984078548,
      "loss": 23.0,
      "step": 3067
    },
    {
      "epoch": 1.8415366146458583,
      "grad_norm": 0.001763959531672299,
      "learning_rate": 0.00018397834351269255,
      "loss": 23.0,
      "step": 3068
    },
    {
      "epoch": 1.8421368547418968,
      "grad_norm": 0.0012634745799005032,
      "learning_rate": 0.0001839680856205592,
      "loss": 23.0,
      "step": 3069
    },
    {
      "epoch": 1.8427370948379351,
      "grad_norm": 0.0015287197893485427,
      "learning_rate": 0.0001839578247318209,
      "loss": 23.0,
      "step": 3070
    },
    {
      "epoch": 1.8433373349339734,
      "grad_norm": 0.0009792933706194162,
      "learning_rate": 0.00018394756084684377,
      "loss": 23.0,
      "step": 3071
    },
    {
      "epoch": 1.843937575030012,
      "grad_norm": 0.0009839729173108935,
      "learning_rate": 0.00018393729396599416,
      "loss": 23.0,
      "step": 3072
    },
    {
      "epoch": 1.8445378151260505,
      "grad_norm": 0.0008178153657354414,
      "learning_rate": 0.00018392702408963842,
      "loss": 23.0,
      "step": 3073
    },
    {
      "epoch": 1.8451380552220888,
      "grad_norm": 0.0018251874716952443,
      "learning_rate": 0.0001839167512181431,
      "loss": 23.0,
      "step": 3074
    },
    {
      "epoch": 1.8457382953181272,
      "grad_norm": 0.002089198911562562,
      "learning_rate": 0.00018390647535187476,
      "loss": 23.0,
      "step": 3075
    },
    {
      "epoch": 1.8463385354141657,
      "grad_norm": 0.0021562918554991484,
      "learning_rate": 0.00018389619649120017,
      "loss": 23.0,
      "step": 3076
    },
    {
      "epoch": 1.8469387755102042,
      "grad_norm": 0.0017833629390224814,
      "learning_rate": 0.00018388591463648616,
      "loss": 23.0,
      "step": 3077
    },
    {
      "epoch": 1.8475390156062423,
      "grad_norm": 0.000439735857071355,
      "learning_rate": 0.00018387562978809958,
      "loss": 23.0,
      "step": 3078
    },
    {
      "epoch": 1.8481392557022809,
      "grad_norm": 0.001050430117174983,
      "learning_rate": 0.00018386534194640754,
      "loss": 23.0,
      "step": 3079
    },
    {
      "epoch": 1.8487394957983194,
      "grad_norm": 0.001189610455185175,
      "learning_rate": 0.00018385505111177719,
      "loss": 23.0,
      "step": 3080
    },
    {
      "epoch": 1.8493397358943577,
      "grad_norm": 0.0010707725305110216,
      "learning_rate": 0.00018384475728457574,
      "loss": 23.0,
      "step": 3081
    },
    {
      "epoch": 1.849939975990396,
      "grad_norm": 0.002398673677816987,
      "learning_rate": 0.00018383446046517062,
      "loss": 23.0,
      "step": 3082
    },
    {
      "epoch": 1.8505402160864346,
      "grad_norm": 0.0018957408610731363,
      "learning_rate": 0.00018382416065392923,
      "loss": 23.0,
      "step": 3083
    },
    {
      "epoch": 1.8511404561824731,
      "grad_norm": 0.0022743563167750835,
      "learning_rate": 0.0001838138578512192,
      "loss": 23.0,
      "step": 3084
    },
    {
      "epoch": 1.8517406962785115,
      "grad_norm": 0.0014944793656468391,
      "learning_rate": 0.00018380355205740813,
      "loss": 23.0,
      "step": 3085
    },
    {
      "epoch": 1.8523409363745498,
      "grad_norm": 0.0011577775003388524,
      "learning_rate": 0.00018379324327286387,
      "loss": 23.0,
      "step": 3086
    },
    {
      "epoch": 1.8529411764705883,
      "grad_norm": 0.0023493997287005186,
      "learning_rate": 0.0001837829314979543,
      "loss": 23.0,
      "step": 3087
    },
    {
      "epoch": 1.8535414165666266,
      "grad_norm": 0.0017023790860548615,
      "learning_rate": 0.00018377261673304743,
      "loss": 23.0,
      "step": 3088
    },
    {
      "epoch": 1.854141656662665,
      "grad_norm": 0.001627809484489262,
      "learning_rate": 0.00018376229897851135,
      "loss": 23.0,
      "step": 3089
    },
    {
      "epoch": 1.8547418967587035,
      "grad_norm": 0.0016393931582570076,
      "learning_rate": 0.00018375197823471427,
      "loss": 23.0,
      "step": 3090
    },
    {
      "epoch": 1.855342136854742,
      "grad_norm": 0.0018370094476267695,
      "learning_rate": 0.00018374165450202456,
      "loss": 23.0,
      "step": 3091
    },
    {
      "epoch": 1.8559423769507803,
      "grad_norm": 0.0014342045178636909,
      "learning_rate": 0.0001837313277808106,
      "loss": 23.0,
      "step": 3092
    },
    {
      "epoch": 1.8565426170468187,
      "grad_norm": 0.0029582458082586527,
      "learning_rate": 0.0001837209980714409,
      "loss": 23.0,
      "step": 3093
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.0012816624948754907,
      "learning_rate": 0.00018371066537428417,
      "loss": 23.0,
      "step": 3094
    },
    {
      "epoch": 1.8577430972388955,
      "grad_norm": 0.002697428921237588,
      "learning_rate": 0.0001837003296897091,
      "loss": 23.0,
      "step": 3095
    },
    {
      "epoch": 1.8583433373349338,
      "grad_norm": 0.0037919096648693085,
      "learning_rate": 0.00018368999101808458,
      "loss": 23.0,
      "step": 3096
    },
    {
      "epoch": 1.8589435774309724,
      "grad_norm": 0.003335389541462064,
      "learning_rate": 0.00018367964935977953,
      "loss": 23.0,
      "step": 3097
    },
    {
      "epoch": 1.859543817527011,
      "grad_norm": 0.0011625085026025772,
      "learning_rate": 0.00018366930471516306,
      "loss": 23.0,
      "step": 3098
    },
    {
      "epoch": 1.8601440576230492,
      "grad_norm": 0.001312834327109158,
      "learning_rate": 0.00018365895708460434,
      "loss": 23.0,
      "step": 3099
    },
    {
      "epoch": 1.8607442977190876,
      "grad_norm": 0.0009217692422680557,
      "learning_rate": 0.00018364860646847265,
      "loss": 23.0,
      "step": 3100
    },
    {
      "epoch": 1.861344537815126,
      "grad_norm": 0.001049682148732245,
      "learning_rate": 0.00018363825286713735,
      "loss": 23.0,
      "step": 3101
    },
    {
      "epoch": 1.8619447779111644,
      "grad_norm": 0.0009720840607769787,
      "learning_rate": 0.00018362789628096794,
      "loss": 23.0,
      "step": 3102
    },
    {
      "epoch": 1.8625450180072027,
      "grad_norm": 0.0013249212643131614,
      "learning_rate": 0.000183617536710334,
      "loss": 23.0,
      "step": 3103
    },
    {
      "epoch": 1.8631452581032413,
      "grad_norm": 0.0006475619738921523,
      "learning_rate": 0.00018360717415560532,
      "loss": 23.0,
      "step": 3104
    },
    {
      "epoch": 1.8637454981992798,
      "grad_norm": 0.0009137241286225617,
      "learning_rate": 0.00018359680861715163,
      "loss": 23.0,
      "step": 3105
    },
    {
      "epoch": 1.8643457382953181,
      "grad_norm": 0.001901013427414,
      "learning_rate": 0.00018358644009534286,
      "loss": 23.0,
      "step": 3106
    },
    {
      "epoch": 1.8649459783913565,
      "grad_norm": 0.0014280707109719515,
      "learning_rate": 0.0001835760685905491,
      "loss": 23.0,
      "step": 3107
    },
    {
      "epoch": 1.865546218487395,
      "grad_norm": 0.0019309140043333173,
      "learning_rate": 0.00018356569410314042,
      "loss": 23.0,
      "step": 3108
    },
    {
      "epoch": 1.8661464585834335,
      "grad_norm": 0.0008051837212406099,
      "learning_rate": 0.00018355531663348705,
      "loss": 23.0,
      "step": 3109
    },
    {
      "epoch": 1.8667466986794716,
      "grad_norm": 0.0013398687588050961,
      "learning_rate": 0.00018354493618195935,
      "loss": 23.0,
      "step": 3110
    },
    {
      "epoch": 1.8673469387755102,
      "grad_norm": 0.000537229236215353,
      "learning_rate": 0.00018353455274892783,
      "loss": 23.0,
      "step": 3111
    },
    {
      "epoch": 1.8679471788715487,
      "grad_norm": 0.00452477065846324,
      "learning_rate": 0.00018352416633476298,
      "loss": 23.0,
      "step": 3112
    },
    {
      "epoch": 1.868547418967587,
      "grad_norm": 0.002200663322582841,
      "learning_rate": 0.0001835137769398355,
      "loss": 23.0,
      "step": 3113
    },
    {
      "epoch": 1.8691476590636253,
      "grad_norm": 0.0013940661447122693,
      "learning_rate": 0.0001835033845645161,
      "loss": 23.0,
      "step": 3114
    },
    {
      "epoch": 1.8697478991596639,
      "grad_norm": 0.002113936236128211,
      "learning_rate": 0.00018349298920917572,
      "loss": 23.0,
      "step": 3115
    },
    {
      "epoch": 1.8703481392557024,
      "grad_norm": 0.0012045325711369514,
      "learning_rate": 0.00018348259087418533,
      "loss": 23.0,
      "step": 3116
    },
    {
      "epoch": 1.8709483793517407,
      "grad_norm": 0.0033185603097081184,
      "learning_rate": 0.000183472189559916,
      "loss": 23.0,
      "step": 3117
    },
    {
      "epoch": 1.871548619447779,
      "grad_norm": 0.0011681788600981236,
      "learning_rate": 0.00018346178526673897,
      "loss": 23.0,
      "step": 3118
    },
    {
      "epoch": 1.8721488595438176,
      "grad_norm": 0.0022086468525230885,
      "learning_rate": 0.0001834513779950255,
      "loss": 23.0,
      "step": 3119
    },
    {
      "epoch": 1.872749099639856,
      "grad_norm": 0.0005794914322905242,
      "learning_rate": 0.000183440967745147,
      "loss": 23.0,
      "step": 3120
    },
    {
      "epoch": 1.8733493397358942,
      "grad_norm": 0.0011907683219760656,
      "learning_rate": 0.00018343055451747498,
      "loss": 23.0,
      "step": 3121
    },
    {
      "epoch": 1.8739495798319328,
      "grad_norm": 0.0005777979386039078,
      "learning_rate": 0.0001834201383123811,
      "loss": 23.0,
      "step": 3122
    },
    {
      "epoch": 1.8745498199279713,
      "grad_norm": 0.004292360506951809,
      "learning_rate": 0.00018340971913023707,
      "loss": 23.0,
      "step": 3123
    },
    {
      "epoch": 1.8751500600240096,
      "grad_norm": 0.0035806214436888695,
      "learning_rate": 0.00018339929697141474,
      "loss": 23.0,
      "step": 3124
    },
    {
      "epoch": 1.875750300120048,
      "grad_norm": 0.0015884626191109419,
      "learning_rate": 0.00018338887183628599,
      "loss": 23.0,
      "step": 3125
    },
    {
      "epoch": 1.8763505402160865,
      "grad_norm": 0.0011369887506589293,
      "learning_rate": 0.00018337844372522292,
      "loss": 23.0,
      "step": 3126
    },
    {
      "epoch": 1.8769507803121248,
      "grad_norm": 0.005162140820175409,
      "learning_rate": 0.00018336801263859768,
      "loss": 23.0,
      "step": 3127
    },
    {
      "epoch": 1.8775510204081631,
      "grad_norm": 0.0024860259145498276,
      "learning_rate": 0.00018335757857678253,
      "loss": 23.0,
      "step": 3128
    },
    {
      "epoch": 1.8781512605042017,
      "grad_norm": 0.0012651464203372598,
      "learning_rate": 0.0001833471415401498,
      "loss": 23.0,
      "step": 3129
    },
    {
      "epoch": 1.8787515006002402,
      "grad_norm": 0.00254456396214664,
      "learning_rate": 0.00018333670152907198,
      "loss": 23.0,
      "step": 3130
    },
    {
      "epoch": 1.8793517406962785,
      "grad_norm": 0.0015262291999533772,
      "learning_rate": 0.00018332625854392164,
      "loss": 23.0,
      "step": 3131
    },
    {
      "epoch": 1.8799519807923168,
      "grad_norm": 0.002018503611907363,
      "learning_rate": 0.0001833158125850715,
      "loss": 23.0,
      "step": 3132
    },
    {
      "epoch": 1.8805522208883554,
      "grad_norm": 0.00178581103682518,
      "learning_rate": 0.0001833053636528943,
      "loss": 23.0,
      "step": 3133
    },
    {
      "epoch": 1.8811524609843937,
      "grad_norm": 0.002688503125682473,
      "learning_rate": 0.00018329491174776295,
      "loss": 23.0,
      "step": 3134
    },
    {
      "epoch": 1.881752701080432,
      "grad_norm": 0.0027273113373667,
      "learning_rate": 0.0001832844568700505,
      "loss": 23.0,
      "step": 3135
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 0.0010961064836010337,
      "learning_rate": 0.00018327399902013,
      "loss": 23.0,
      "step": 3136
    },
    {
      "epoch": 1.882953181272509,
      "grad_norm": 0.0035878592170774937,
      "learning_rate": 0.00018326353819837467,
      "loss": 23.0,
      "step": 3137
    },
    {
      "epoch": 1.8835534213685474,
      "grad_norm": 0.0022766878828406334,
      "learning_rate": 0.00018325307440515785,
      "loss": 23.0,
      "step": 3138
    },
    {
      "epoch": 1.8841536614645857,
      "grad_norm": 0.0014452520990744233,
      "learning_rate": 0.00018324260764085294,
      "loss": 23.0,
      "step": 3139
    },
    {
      "epoch": 1.8847539015606243,
      "grad_norm": 0.001565837417729199,
      "learning_rate": 0.00018323213790583353,
      "loss": 23.0,
      "step": 3140
    },
    {
      "epoch": 1.8853541416566628,
      "grad_norm": 0.002488331636413932,
      "learning_rate": 0.0001832216652004732,
      "loss": 23.0,
      "step": 3141
    },
    {
      "epoch": 1.885954381752701,
      "grad_norm": 0.0023180756252259016,
      "learning_rate": 0.00018321118952514568,
      "loss": 23.0,
      "step": 3142
    },
    {
      "epoch": 1.8865546218487395,
      "grad_norm": 0.0013001611223444343,
      "learning_rate": 0.0001832007108802249,
      "loss": 23.0,
      "step": 3143
    },
    {
      "epoch": 1.887154861944778,
      "grad_norm": 0.00461295573040843,
      "learning_rate": 0.0001831902292660847,
      "loss": 23.0,
      "step": 3144
    },
    {
      "epoch": 1.8877551020408163,
      "grad_norm": 0.0016194535419344902,
      "learning_rate": 0.00018317974468309926,
      "loss": 23.0,
      "step": 3145
    },
    {
      "epoch": 1.8883553421368546,
      "grad_norm": 0.002495100488886237,
      "learning_rate": 0.00018316925713164267,
      "loss": 23.0,
      "step": 3146
    },
    {
      "epoch": 1.8889555822328932,
      "grad_norm": 0.0009624058147892356,
      "learning_rate": 0.00018315876661208926,
      "loss": 23.0,
      "step": 3147
    },
    {
      "epoch": 1.8895558223289317,
      "grad_norm": 0.002280160551890731,
      "learning_rate": 0.00018314827312481335,
      "loss": 23.0,
      "step": 3148
    },
    {
      "epoch": 1.89015606242497,
      "grad_norm": 0.001309713814407587,
      "learning_rate": 0.00018313777667018947,
      "loss": 23.0,
      "step": 3149
    },
    {
      "epoch": 1.8907563025210083,
      "grad_norm": 0.00454369792714715,
      "learning_rate": 0.0001831272772485922,
      "loss": 23.0,
      "step": 3150
    },
    {
      "epoch": 1.8913565426170469,
      "grad_norm": 0.0026175181847065687,
      "learning_rate": 0.0001831167748603962,
      "loss": 23.0,
      "step": 3151
    },
    {
      "epoch": 1.8919567827130852,
      "grad_norm": 0.0012701862724497914,
      "learning_rate": 0.00018310626950597637,
      "loss": 23.0,
      "step": 3152
    },
    {
      "epoch": 1.8925570228091235,
      "grad_norm": 0.0033725642133504152,
      "learning_rate": 0.0001830957611857075,
      "loss": 23.0,
      "step": 3153
    },
    {
      "epoch": 1.893157262905162,
      "grad_norm": 0.005127619951963425,
      "learning_rate": 0.0001830852498999647,
      "loss": 23.0,
      "step": 3154
    },
    {
      "epoch": 1.8937575030012006,
      "grad_norm": 0.0004975322517566383,
      "learning_rate": 0.00018307473564912303,
      "loss": 23.0,
      "step": 3155
    },
    {
      "epoch": 1.894357743097239,
      "grad_norm": 0.002670707879588008,
      "learning_rate": 0.00018306421843355773,
      "loss": 23.0,
      "step": 3156
    },
    {
      "epoch": 1.8949579831932772,
      "grad_norm": 0.003962348680943251,
      "learning_rate": 0.00018305369825364417,
      "loss": 23.0,
      "step": 3157
    },
    {
      "epoch": 1.8955582232893158,
      "grad_norm": 0.000557166466023773,
      "learning_rate": 0.00018304317510975775,
      "loss": 23.0,
      "step": 3158
    },
    {
      "epoch": 1.896158463385354,
      "grad_norm": 0.0021320811938494444,
      "learning_rate": 0.00018303264900227398,
      "loss": 23.0,
      "step": 3159
    },
    {
      "epoch": 1.8967587034813924,
      "grad_norm": 0.003195348661392927,
      "learning_rate": 0.0001830221199315686,
      "loss": 23.0,
      "step": 3160
    },
    {
      "epoch": 1.897358943577431,
      "grad_norm": 0.0013635127106681466,
      "learning_rate": 0.0001830115878980173,
      "loss": 23.0,
      "step": 3161
    },
    {
      "epoch": 1.8979591836734695,
      "grad_norm": 0.002584997797384858,
      "learning_rate": 0.00018300105290199595,
      "loss": 23.0,
      "step": 3162
    },
    {
      "epoch": 1.8985594237695078,
      "grad_norm": 0.0012520054588094354,
      "learning_rate": 0.00018299051494388054,
      "loss": 23.0,
      "step": 3163
    },
    {
      "epoch": 1.8991596638655461,
      "grad_norm": 0.0018940556328743696,
      "learning_rate": 0.00018297997402404715,
      "loss": 23.0,
      "step": 3164
    },
    {
      "epoch": 1.8997599039615847,
      "grad_norm": 0.0019348510541021824,
      "learning_rate": 0.00018296943014287193,
      "loss": 23.0,
      "step": 3165
    },
    {
      "epoch": 1.9003601440576232,
      "grad_norm": 0.0018156596925109625,
      "learning_rate": 0.00018295888330073113,
      "loss": 23.0,
      "step": 3166
    },
    {
      "epoch": 1.9009603841536613,
      "grad_norm": 0.0035436146426945925,
      "learning_rate": 0.0001829483334980012,
      "loss": 23.0,
      "step": 3167
    },
    {
      "epoch": 1.9015606242496998,
      "grad_norm": 0.0012572517152875662,
      "learning_rate": 0.00018293778073505865,
      "loss": 23.0,
      "step": 3168
    },
    {
      "epoch": 1.9021608643457384,
      "grad_norm": 0.0008448131266050041,
      "learning_rate": 0.00018292722501228004,
      "loss": 23.0,
      "step": 3169
    },
    {
      "epoch": 1.9027611044417767,
      "grad_norm": 0.0037472823169082403,
      "learning_rate": 0.00018291666633004204,
      "loss": 23.0,
      "step": 3170
    },
    {
      "epoch": 1.903361344537815,
      "grad_norm": 0.0018287114799022675,
      "learning_rate": 0.00018290610468872155,
      "loss": 23.0,
      "step": 3171
    },
    {
      "epoch": 1.9039615846338536,
      "grad_norm": 0.0022742743603885174,
      "learning_rate": 0.0001828955400886954,
      "loss": 23.0,
      "step": 3172
    },
    {
      "epoch": 1.904561824729892,
      "grad_norm": 0.0018581197364255786,
      "learning_rate": 0.0001828849725303407,
      "loss": 23.0,
      "step": 3173
    },
    {
      "epoch": 1.9051620648259304,
      "grad_norm": 0.001101133180782199,
      "learning_rate": 0.0001828744020140345,
      "loss": 23.0,
      "step": 3174
    },
    {
      "epoch": 1.9057623049219687,
      "grad_norm": 0.0020204149186611176,
      "learning_rate": 0.0001828638285401541,
      "loss": 23.0,
      "step": 3175
    },
    {
      "epoch": 1.9063625450180073,
      "grad_norm": 0.0028773958329111338,
      "learning_rate": 0.0001828532521090768,
      "loss": 23.0,
      "step": 3176
    },
    {
      "epoch": 1.9069627851140456,
      "grad_norm": 0.0018713538302108645,
      "learning_rate": 0.00018284267272118004,
      "loss": 23.0,
      "step": 3177
    },
    {
      "epoch": 1.907563025210084,
      "grad_norm": 0.0018324162811040878,
      "learning_rate": 0.00018283209037684141,
      "loss": 23.0,
      "step": 3178
    },
    {
      "epoch": 1.9081632653061225,
      "grad_norm": 0.002509085228666663,
      "learning_rate": 0.00018282150507643855,
      "loss": 23.0,
      "step": 3179
    },
    {
      "epoch": 1.908763505402161,
      "grad_norm": 0.002151481108739972,
      "learning_rate": 0.0001828109168203492,
      "loss": 23.0,
      "step": 3180
    },
    {
      "epoch": 1.9093637454981993,
      "grad_norm": 0.0014471776084974408,
      "learning_rate": 0.00018280032560895123,
      "loss": 23.0,
      "step": 3181
    },
    {
      "epoch": 1.9099639855942376,
      "grad_norm": 0.0023641546722501516,
      "learning_rate": 0.00018278973144262266,
      "loss": 23.0,
      "step": 3182
    },
    {
      "epoch": 1.9105642256902762,
      "grad_norm": 0.003392763901501894,
      "learning_rate": 0.0001827791343217415,
      "loss": 23.0,
      "step": 3183
    },
    {
      "epoch": 1.9111644657863145,
      "grad_norm": 0.003529018722474575,
      "learning_rate": 0.00018276853424668602,
      "loss": 23.0,
      "step": 3184
    },
    {
      "epoch": 1.9117647058823528,
      "grad_norm": 0.003656035289168358,
      "learning_rate": 0.0001827579312178344,
      "loss": 23.0,
      "step": 3185
    },
    {
      "epoch": 1.9123649459783914,
      "grad_norm": 0.0014189134817570448,
      "learning_rate": 0.00018274732523556513,
      "loss": 23.0,
      "step": 3186
    },
    {
      "epoch": 1.91296518607443,
      "grad_norm": 0.0013264564331620932,
      "learning_rate": 0.00018273671630025665,
      "loss": 23.0,
      "step": 3187
    },
    {
      "epoch": 1.9135654261704682,
      "grad_norm": 0.004117329139262438,
      "learning_rate": 0.00018272610441228763,
      "loss": 23.0,
      "step": 3188
    },
    {
      "epoch": 1.9141656662665065,
      "grad_norm": 0.0024092942476272583,
      "learning_rate": 0.0001827154895720367,
      "loss": 23.0,
      "step": 3189
    },
    {
      "epoch": 1.914765906362545,
      "grad_norm": 0.001247992506250739,
      "learning_rate": 0.00018270487177988272,
      "loss": 23.0,
      "step": 3190
    },
    {
      "epoch": 1.9153661464585834,
      "grad_norm": 0.006278992164880037,
      "learning_rate": 0.00018269425103620464,
      "loss": 23.0,
      "step": 3191
    },
    {
      "epoch": 1.9159663865546217,
      "grad_norm": 0.001128924312070012,
      "learning_rate": 0.00018268362734138142,
      "loss": 23.0,
      "step": 3192
    },
    {
      "epoch": 1.9165666266506602,
      "grad_norm": 0.004643549211323261,
      "learning_rate": 0.00018267300069579222,
      "loss": 23.0,
      "step": 3193
    },
    {
      "epoch": 1.9171668667466988,
      "grad_norm": 0.002494530752301216,
      "learning_rate": 0.00018266237109981632,
      "loss": 23.0,
      "step": 3194
    },
    {
      "epoch": 1.917767106842737,
      "grad_norm": 0.00441815285012126,
      "learning_rate": 0.000182651738553833,
      "loss": 23.0,
      "step": 3195
    },
    {
      "epoch": 1.9183673469387754,
      "grad_norm": 0.0028012660332024097,
      "learning_rate": 0.00018264110305822178,
      "loss": 23.0,
      "step": 3196
    },
    {
      "epoch": 1.918967587034814,
      "grad_norm": 0.0017156752292066813,
      "learning_rate": 0.00018263046461336214,
      "loss": 23.0,
      "step": 3197
    },
    {
      "epoch": 1.9195678271308525,
      "grad_norm": 0.002733529545366764,
      "learning_rate": 0.00018261982321963379,
      "loss": 23.0,
      "step": 3198
    },
    {
      "epoch": 1.9201680672268906,
      "grad_norm": 0.0011137155815958977,
      "learning_rate": 0.00018260917887741643,
      "loss": 23.0,
      "step": 3199
    },
    {
      "epoch": 1.9207683073229291,
      "grad_norm": 0.0014791876310482621,
      "learning_rate": 0.00018259853158708998,
      "loss": 23.0,
      "step": 3200
    },
    {
      "epoch": 1.9213685474189677,
      "grad_norm": 0.0009261185768991709,
      "learning_rate": 0.00018258788134903445,
      "loss": 23.0,
      "step": 3201
    },
    {
      "epoch": 1.921968787515006,
      "grad_norm": 0.0030771689489483833,
      "learning_rate": 0.00018257722816362985,
      "loss": 23.0,
      "step": 3202
    },
    {
      "epoch": 1.9225690276110443,
      "grad_norm": 0.003100763075053692,
      "learning_rate": 0.00018256657203125637,
      "loss": 23.0,
      "step": 3203
    },
    {
      "epoch": 1.9231692677070829,
      "grad_norm": 0.002674756571650505,
      "learning_rate": 0.00018255591295229437,
      "loss": 23.0,
      "step": 3204
    },
    {
      "epoch": 1.9237695078031214,
      "grad_norm": 0.0022848814260214567,
      "learning_rate": 0.00018254525092712415,
      "loss": 23.0,
      "step": 3205
    },
    {
      "epoch": 1.9243697478991597,
      "grad_norm": 0.0018523391336202621,
      "learning_rate": 0.00018253458595612627,
      "loss": 23.0,
      "step": 3206
    },
    {
      "epoch": 1.924969987995198,
      "grad_norm": 0.0009723806870169938,
      "learning_rate": 0.00018252391803968133,
      "loss": 23.0,
      "step": 3207
    },
    {
      "epoch": 1.9255702280912366,
      "grad_norm": 0.001458384795114398,
      "learning_rate": 0.00018251324717817005,
      "loss": 23.0,
      "step": 3208
    },
    {
      "epoch": 1.9261704681872749,
      "grad_norm": 0.0013265210436657071,
      "learning_rate": 0.0001825025733719732,
      "loss": 23.0,
      "step": 3209
    },
    {
      "epoch": 1.9267707082833132,
      "grad_norm": 0.0010744367027655244,
      "learning_rate": 0.00018249189662147171,
      "loss": 23.0,
      "step": 3210
    },
    {
      "epoch": 1.9273709483793517,
      "grad_norm": 0.0026470967568457127,
      "learning_rate": 0.00018248121692704668,
      "loss": 23.0,
      "step": 3211
    },
    {
      "epoch": 1.9279711884753903,
      "grad_norm": 0.0016062292270362377,
      "learning_rate": 0.00018247053428907913,
      "loss": 23.0,
      "step": 3212
    },
    {
      "epoch": 1.9285714285714286,
      "grad_norm": 0.000824375543743372,
      "learning_rate": 0.00018245984870795038,
      "loss": 23.0,
      "step": 3213
    },
    {
      "epoch": 1.929171668667467,
      "grad_norm": 0.002928749890998006,
      "learning_rate": 0.00018244916018404173,
      "loss": 23.0,
      "step": 3214
    },
    {
      "epoch": 1.9297719087635055,
      "grad_norm": 0.002207791665568948,
      "learning_rate": 0.00018243846871773465,
      "loss": 23.0,
      "step": 3215
    },
    {
      "epoch": 1.9303721488595438,
      "grad_norm": 0.0028601568192243576,
      "learning_rate": 0.00018242777430941068,
      "loss": 23.0,
      "step": 3216
    },
    {
      "epoch": 1.930972388955582,
      "grad_norm": 0.0021290185395628214,
      "learning_rate": 0.00018241707695945144,
      "loss": 23.0,
      "step": 3217
    },
    {
      "epoch": 1.9315726290516206,
      "grad_norm": 0.000839258951600641,
      "learning_rate": 0.00018240637666823876,
      "loss": 23.0,
      "step": 3218
    },
    {
      "epoch": 1.9321728691476592,
      "grad_norm": 0.002701072720810771,
      "learning_rate": 0.00018239567343615443,
      "loss": 23.0,
      "step": 3219
    },
    {
      "epoch": 1.9327731092436975,
      "grad_norm": 0.0020356529857963324,
      "learning_rate": 0.00018238496726358052,
      "loss": 23.0,
      "step": 3220
    },
    {
      "epoch": 1.9333733493397358,
      "grad_norm": 0.0014873469481244683,
      "learning_rate": 0.00018237425815089903,
      "loss": 23.0,
      "step": 3221
    },
    {
      "epoch": 1.9339735894357744,
      "grad_norm": 0.002182344440370798,
      "learning_rate": 0.00018236354609849214,
      "loss": 23.0,
      "step": 3222
    },
    {
      "epoch": 1.9345738295318127,
      "grad_norm": 0.0016528373816981912,
      "learning_rate": 0.00018235283110674218,
      "loss": 23.0,
      "step": 3223
    },
    {
      "epoch": 1.935174069627851,
      "grad_norm": 0.0012866542674601078,
      "learning_rate": 0.0001823421131760315,
      "loss": 23.0,
      "step": 3224
    },
    {
      "epoch": 1.9357743097238895,
      "grad_norm": 0.0019892966374754906,
      "learning_rate": 0.00018233139230674258,
      "loss": 23.0,
      "step": 3225
    },
    {
      "epoch": 1.936374549819928,
      "grad_norm": 0.002182098338380456,
      "learning_rate": 0.0001823206684992581,
      "loss": 23.0,
      "step": 3226
    },
    {
      "epoch": 1.9369747899159664,
      "grad_norm": 0.0021431762725114822,
      "learning_rate": 0.0001823099417539607,
      "loss": 23.0,
      "step": 3227
    },
    {
      "epoch": 1.9375750300120047,
      "grad_norm": 0.0024950047954916954,
      "learning_rate": 0.00018229921207123318,
      "loss": 23.0,
      "step": 3228
    },
    {
      "epoch": 1.9381752701080432,
      "grad_norm": 0.0017660352168604732,
      "learning_rate": 0.0001822884794514585,
      "loss": 23.0,
      "step": 3229
    },
    {
      "epoch": 1.9387755102040818,
      "grad_norm": 0.0065404558554291725,
      "learning_rate": 0.00018227774389501967,
      "loss": 23.0,
      "step": 3230
    },
    {
      "epoch": 1.9393757503001199,
      "grad_norm": 0.0011474026832729578,
      "learning_rate": 0.00018226700540229976,
      "loss": 23.0,
      "step": 3231
    },
    {
      "epoch": 1.9399759903961584,
      "grad_norm": 0.00047521217493340373,
      "learning_rate": 0.0001822562639736821,
      "loss": 23.0,
      "step": 3232
    },
    {
      "epoch": 1.940576230492197,
      "grad_norm": 0.0019478441681712866,
      "learning_rate": 0.00018224551960954995,
      "loss": 23.0,
      "step": 3233
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 0.0010433909483253956,
      "learning_rate": 0.00018223477231028677,
      "loss": 23.0,
      "step": 3234
    },
    {
      "epoch": 1.9417767106842736,
      "grad_norm": 0.0018290705047547817,
      "learning_rate": 0.00018222402207627606,
      "loss": 23.0,
      "step": 3235
    },
    {
      "epoch": 1.9423769507803121,
      "grad_norm": 0.0017392985755577683,
      "learning_rate": 0.00018221326890790153,
      "loss": 23.0,
      "step": 3236
    },
    {
      "epoch": 1.9429771908763507,
      "grad_norm": 0.002610771218314767,
      "learning_rate": 0.00018220251280554696,
      "loss": 23.0,
      "step": 3237
    },
    {
      "epoch": 1.943577430972389,
      "grad_norm": 0.0016300787683576345,
      "learning_rate": 0.00018219175376959606,
      "loss": 23.0,
      "step": 3238
    },
    {
      "epoch": 1.9441776710684273,
      "grad_norm": 0.0022878015879541636,
      "learning_rate": 0.00018218099180043295,
      "loss": 23.0,
      "step": 3239
    },
    {
      "epoch": 1.9447779111644659,
      "grad_norm": 0.0036289305426180363,
      "learning_rate": 0.00018217022689844163,
      "loss": 23.0,
      "step": 3240
    },
    {
      "epoch": 1.9453781512605042,
      "grad_norm": 0.0009557732264511287,
      "learning_rate": 0.00018215945906400628,
      "loss": 23.0,
      "step": 3241
    },
    {
      "epoch": 1.9459783913565425,
      "grad_norm": 0.0023918780498206615,
      "learning_rate": 0.0001821486882975112,
      "loss": 23.0,
      "step": 3242
    },
    {
      "epoch": 1.946578631452581,
      "grad_norm": 0.001407388597726822,
      "learning_rate": 0.0001821379145993407,
      "loss": 23.0,
      "step": 3243
    },
    {
      "epoch": 1.9471788715486196,
      "grad_norm": 0.0011877256911247969,
      "learning_rate": 0.00018212713796987936,
      "loss": 23.0,
      "step": 3244
    },
    {
      "epoch": 1.947779111644658,
      "grad_norm": 0.001021384377963841,
      "learning_rate": 0.0001821163584095117,
      "loss": 23.0,
      "step": 3245
    },
    {
      "epoch": 1.9483793517406962,
      "grad_norm": 0.0026115647051483393,
      "learning_rate": 0.0001821055759186224,
      "loss": 23.0,
      "step": 3246
    },
    {
      "epoch": 1.9489795918367347,
      "grad_norm": 0.001517315162345767,
      "learning_rate": 0.00018209479049759636,
      "loss": 23.0,
      "step": 3247
    },
    {
      "epoch": 1.949579831932773,
      "grad_norm": 0.003172236494719982,
      "learning_rate": 0.00018208400214681837,
      "loss": 23.0,
      "step": 3248
    },
    {
      "epoch": 1.9501800720288114,
      "grad_norm": 0.001906992169097066,
      "learning_rate": 0.00018207321086667353,
      "loss": 23.0,
      "step": 3249
    },
    {
      "epoch": 1.95078031212485,
      "grad_norm": 0.003104976611211896,
      "learning_rate": 0.00018206241665754688,
      "loss": 23.0,
      "step": 3250
    },
    {
      "epoch": 1.9513805522208885,
      "grad_norm": 0.0015054944669827819,
      "learning_rate": 0.00018205161951982367,
      "loss": 23.0,
      "step": 3251
    },
    {
      "epoch": 1.9519807923169268,
      "grad_norm": 0.0020006780978292227,
      "learning_rate": 0.00018204081945388922,
      "loss": 23.0,
      "step": 3252
    },
    {
      "epoch": 1.952581032412965,
      "grad_norm": 0.0018055624095723033,
      "learning_rate": 0.00018203001646012899,
      "loss": 23.0,
      "step": 3253
    },
    {
      "epoch": 1.9531812725090036,
      "grad_norm": 0.0017827142728492618,
      "learning_rate": 0.00018201921053892845,
      "loss": 23.0,
      "step": 3254
    },
    {
      "epoch": 1.9537815126050422,
      "grad_norm": 0.0020725359208881855,
      "learning_rate": 0.0001820084016906733,
      "loss": 23.0,
      "step": 3255
    },
    {
      "epoch": 1.9543817527010803,
      "grad_norm": 0.001384815899655223,
      "learning_rate": 0.0001819975899157492,
      "loss": 23.0,
      "step": 3256
    },
    {
      "epoch": 1.9549819927971188,
      "grad_norm": 0.002811410464346409,
      "learning_rate": 0.00018198677521454206,
      "loss": 23.0,
      "step": 3257
    },
    {
      "epoch": 1.9555822328931574,
      "grad_norm": 0.0008216790156438947,
      "learning_rate": 0.0001819759575874378,
      "loss": 23.0,
      "step": 3258
    },
    {
      "epoch": 1.9561824729891957,
      "grad_norm": 0.0014283070340752602,
      "learning_rate": 0.00018196513703482248,
      "loss": 23.0,
      "step": 3259
    },
    {
      "epoch": 1.956782713085234,
      "grad_norm": 0.0016851375112310052,
      "learning_rate": 0.00018195431355708225,
      "loss": 23.0,
      "step": 3260
    },
    {
      "epoch": 1.9573829531812725,
      "grad_norm": 0.0019182129763066769,
      "learning_rate": 0.0001819434871546034,
      "loss": 23.0,
      "step": 3261
    },
    {
      "epoch": 1.957983193277311,
      "grad_norm": 0.0008913344354368746,
      "learning_rate": 0.00018193265782777228,
      "loss": 23.0,
      "step": 3262
    },
    {
      "epoch": 1.9585834333733494,
      "grad_norm": 0.0013015334261581302,
      "learning_rate": 0.00018192182557697534,
      "loss": 23.0,
      "step": 3263
    },
    {
      "epoch": 1.9591836734693877,
      "grad_norm": 0.0009710061131045222,
      "learning_rate": 0.00018191099040259918,
      "loss": 23.0,
      "step": 3264
    },
    {
      "epoch": 1.9597839135654262,
      "grad_norm": 0.0013321270234882832,
      "learning_rate": 0.0001819001523050305,
      "loss": 23.0,
      "step": 3265
    },
    {
      "epoch": 1.9603841536614646,
      "grad_norm": 0.004230269230902195,
      "learning_rate": 0.00018188931128465603,
      "loss": 23.0,
      "step": 3266
    },
    {
      "epoch": 1.9609843937575029,
      "grad_norm": 0.0016738398699089885,
      "learning_rate": 0.00018187846734186267,
      "loss": 23.0,
      "step": 3267
    },
    {
      "epoch": 1.9615846338535414,
      "grad_norm": 0.0009718580404296517,
      "learning_rate": 0.00018186762047703745,
      "loss": 23.0,
      "step": 3268
    },
    {
      "epoch": 1.96218487394958,
      "grad_norm": 0.0022842460311949253,
      "learning_rate": 0.00018185677069056742,
      "loss": 23.0,
      "step": 3269
    },
    {
      "epoch": 1.9627851140456183,
      "grad_norm": 0.0014329022960737348,
      "learning_rate": 0.00018184591798283984,
      "loss": 23.0,
      "step": 3270
    },
    {
      "epoch": 1.9633853541416566,
      "grad_norm": 0.0033265934325754642,
      "learning_rate": 0.00018183506235424198,
      "loss": 23.0,
      "step": 3271
    },
    {
      "epoch": 1.9639855942376951,
      "grad_norm": 0.0007821978651918471,
      "learning_rate": 0.0001818242038051612,
      "loss": 23.0,
      "step": 3272
    },
    {
      "epoch": 1.9645858343337335,
      "grad_norm": 0.0015688884304836392,
      "learning_rate": 0.00018181334233598507,
      "loss": 23.0,
      "step": 3273
    },
    {
      "epoch": 1.9651860744297718,
      "grad_norm": 0.0008193565299734473,
      "learning_rate": 0.00018180247794710125,
      "loss": 23.0,
      "step": 3274
    },
    {
      "epoch": 1.9657863145258103,
      "grad_norm": 0.002213628264144063,
      "learning_rate": 0.0001817916106388974,
      "loss": 23.0,
      "step": 3275
    },
    {
      "epoch": 1.9663865546218489,
      "grad_norm": 0.0014024798292666674,
      "learning_rate": 0.00018178074041176135,
      "loss": 23.0,
      "step": 3276
    },
    {
      "epoch": 1.9669867947178872,
      "grad_norm": 0.0037618144415318966,
      "learning_rate": 0.00018176986726608103,
      "loss": 23.0,
      "step": 3277
    },
    {
      "epoch": 1.9675870348139255,
      "grad_norm": 0.0024038751143962145,
      "learning_rate": 0.00018175899120224449,
      "loss": 23.0,
      "step": 3278
    },
    {
      "epoch": 1.968187274909964,
      "grad_norm": 0.0033458671532571316,
      "learning_rate": 0.00018174811222063987,
      "loss": 23.0,
      "step": 3279
    },
    {
      "epoch": 1.9687875150060024,
      "grad_norm": 0.0019863531924784184,
      "learning_rate": 0.0001817372303216554,
      "loss": 23.0,
      "step": 3280
    },
    {
      "epoch": 1.9693877551020407,
      "grad_norm": 0.0015641896752640605,
      "learning_rate": 0.0001817263455056794,
      "loss": 23.0,
      "step": 3281
    },
    {
      "epoch": 1.9699879951980792,
      "grad_norm": 0.0012357387458905578,
      "learning_rate": 0.00018171545777310042,
      "loss": 23.0,
      "step": 3282
    },
    {
      "epoch": 1.9705882352941178,
      "grad_norm": 0.002113505732268095,
      "learning_rate": 0.00018170456712430691,
      "loss": 23.0,
      "step": 3283
    },
    {
      "epoch": 1.971188475390156,
      "grad_norm": 0.002042030915617943,
      "learning_rate": 0.0001816936735596876,
      "loss": 23.0,
      "step": 3284
    },
    {
      "epoch": 1.9717887154861944,
      "grad_norm": 0.0012755439383909106,
      "learning_rate": 0.00018168277707963118,
      "loss": 23.0,
      "step": 3285
    },
    {
      "epoch": 1.972388955582233,
      "grad_norm": 0.0019094778690487146,
      "learning_rate": 0.0001816718776845266,
      "loss": 23.0,
      "step": 3286
    },
    {
      "epoch": 1.9729891956782715,
      "grad_norm": 0.00413705175742507,
      "learning_rate": 0.00018166097537476277,
      "loss": 23.0,
      "step": 3287
    },
    {
      "epoch": 1.9735894357743096,
      "grad_norm": 0.0020690206438302994,
      "learning_rate": 0.00018165007015072876,
      "loss": 23.0,
      "step": 3288
    },
    {
      "epoch": 1.974189675870348,
      "grad_norm": 0.00169268692843616,
      "learning_rate": 0.00018163916201281382,
      "loss": 23.0,
      "step": 3289
    },
    {
      "epoch": 1.9747899159663866,
      "grad_norm": 0.001092193415388465,
      "learning_rate": 0.00018162825096140718,
      "loss": 23.0,
      "step": 3290
    },
    {
      "epoch": 1.975390156062425,
      "grad_norm": 0.0015564619097858667,
      "learning_rate": 0.0001816173369968982,
      "loss": 23.0,
      "step": 3291
    },
    {
      "epoch": 1.9759903961584633,
      "grad_norm": 0.0038381817284971476,
      "learning_rate": 0.00018160642011967644,
      "loss": 23.0,
      "step": 3292
    },
    {
      "epoch": 1.9765906362545018,
      "grad_norm": 0.0013099567731842399,
      "learning_rate": 0.00018159550033013144,
      "loss": 23.0,
      "step": 3293
    },
    {
      "epoch": 1.9771908763505404,
      "grad_norm": 0.00048246965161524713,
      "learning_rate": 0.00018158457762865296,
      "loss": 23.0,
      "step": 3294
    },
    {
      "epoch": 1.9777911164465787,
      "grad_norm": 0.0024292506277561188,
      "learning_rate": 0.00018157365201563073,
      "loss": 23.0,
      "step": 3295
    },
    {
      "epoch": 1.978391356542617,
      "grad_norm": 0.0020980066619813442,
      "learning_rate": 0.0001815627234914547,
      "loss": 23.0,
      "step": 3296
    },
    {
      "epoch": 1.9789915966386555,
      "grad_norm": 0.0009206788381561637,
      "learning_rate": 0.00018155179205651487,
      "loss": 23.0,
      "step": 3297
    },
    {
      "epoch": 1.9795918367346939,
      "grad_norm": 0.0014182684244588017,
      "learning_rate": 0.00018154085771120138,
      "loss": 23.0,
      "step": 3298
    },
    {
      "epoch": 1.9801920768307322,
      "grad_norm": 0.0010655620135366917,
      "learning_rate": 0.0001815299204559044,
      "loss": 23.0,
      "step": 3299
    },
    {
      "epoch": 1.9807923169267707,
      "grad_norm": 0.0018564078491181135,
      "learning_rate": 0.0001815189802910143,
      "loss": 23.0,
      "step": 3300
    },
    {
      "epoch": 1.9813925570228093,
      "grad_norm": 0.0016347630880773067,
      "learning_rate": 0.00018150803721692148,
      "loss": 23.0,
      "step": 3301
    },
    {
      "epoch": 1.9819927971188476,
      "grad_norm": 0.0021132517140358686,
      "learning_rate": 0.00018149709123401646,
      "loss": 23.0,
      "step": 3302
    },
    {
      "epoch": 1.982593037214886,
      "grad_norm": 0.002489116508513689,
      "learning_rate": 0.0001814861423426899,
      "loss": 23.0,
      "step": 3303
    },
    {
      "epoch": 1.9831932773109244,
      "grad_norm": 0.0014195957919582725,
      "learning_rate": 0.0001814751905433325,
      "loss": 23.0,
      "step": 3304
    },
    {
      "epoch": 1.9837935174069627,
      "grad_norm": 0.0011736804153770208,
      "learning_rate": 0.00018146423583633518,
      "loss": 23.0,
      "step": 3305
    },
    {
      "epoch": 1.984393757503001,
      "grad_norm": 0.002241331385448575,
      "learning_rate": 0.00018145327822208883,
      "loss": 23.0,
      "step": 3306
    },
    {
      "epoch": 1.9849939975990396,
      "grad_norm": 0.0017535320948809385,
      "learning_rate": 0.00018144231770098447,
      "loss": 23.0,
      "step": 3307
    },
    {
      "epoch": 1.9855942376950781,
      "grad_norm": 0.0021708435378968716,
      "learning_rate": 0.00018143135427341328,
      "loss": 23.0,
      "step": 3308
    },
    {
      "epoch": 1.9861944777911165,
      "grad_norm": 0.0023300815373659134,
      "learning_rate": 0.00018142038793976655,
      "loss": 23.0,
      "step": 3309
    },
    {
      "epoch": 1.9867947178871548,
      "grad_norm": 0.0014577904948964715,
      "learning_rate": 0.0001814094187004356,
      "loss": 23.0,
      "step": 3310
    },
    {
      "epoch": 1.9873949579831933,
      "grad_norm": 0.0024585998617112637,
      "learning_rate": 0.00018139844655581193,
      "loss": 23.0,
      "step": 3311
    },
    {
      "epoch": 1.9879951980792316,
      "grad_norm": 0.0038247143384069204,
      "learning_rate": 0.00018138747150628706,
      "loss": 23.0,
      "step": 3312
    },
    {
      "epoch": 1.98859543817527,
      "grad_norm": 0.0011696047149598598,
      "learning_rate": 0.0001813764935522527,
      "loss": 23.0,
      "step": 3313
    },
    {
      "epoch": 1.9891956782713085,
      "grad_norm": 0.0028556687757372856,
      "learning_rate": 0.00018136551269410062,
      "loss": 23.0,
      "step": 3314
    },
    {
      "epoch": 1.989795918367347,
      "grad_norm": 0.0023186022881418467,
      "learning_rate": 0.00018135452893222262,
      "loss": 23.0,
      "step": 3315
    },
    {
      "epoch": 1.9903961584633854,
      "grad_norm": 0.005861006677150726,
      "learning_rate": 0.0001813435422670108,
      "loss": 23.0,
      "step": 3316
    },
    {
      "epoch": 1.9909963985594237,
      "grad_norm": 0.0006705807754769921,
      "learning_rate": 0.00018133255269885723,
      "loss": 23.0,
      "step": 3317
    },
    {
      "epoch": 1.9915966386554622,
      "grad_norm": 0.0016647104639559984,
      "learning_rate": 0.00018132156022815404,
      "loss": 23.0,
      "step": 3318
    },
    {
      "epoch": 1.9921968787515008,
      "grad_norm": 0.0013882304774597287,
      "learning_rate": 0.00018131056485529352,
      "loss": 23.0,
      "step": 3319
    },
    {
      "epoch": 1.9927971188475389,
      "grad_norm": 0.0024001754354685545,
      "learning_rate": 0.00018129956658066812,
      "loss": 23.0,
      "step": 3320
    },
    {
      "epoch": 1.9933973589435774,
      "grad_norm": 0.0018285515252500772,
      "learning_rate": 0.0001812885654046703,
      "loss": 23.0,
      "step": 3321
    },
    {
      "epoch": 1.993997599039616,
      "grad_norm": 0.0015243416419252753,
      "learning_rate": 0.00018127756132769272,
      "loss": 23.0,
      "step": 3322
    },
    {
      "epoch": 1.9945978391356542,
      "grad_norm": 0.005187567323446274,
      "learning_rate": 0.00018126655435012803,
      "loss": 23.0,
      "step": 3323
    },
    {
      "epoch": 1.9951980792316926,
      "grad_norm": 0.0017248830990865827,
      "learning_rate": 0.00018125554447236906,
      "loss": 23.0,
      "step": 3324
    },
    {
      "epoch": 1.995798319327731,
      "grad_norm": 0.0024234713055193424,
      "learning_rate": 0.00018124453169480868,
      "loss": 23.0,
      "step": 3325
    },
    {
      "epoch": 1.9963985594237696,
      "grad_norm": 0.0025638325605541468,
      "learning_rate": 0.00018123351601784,
      "loss": 23.0,
      "step": 3326
    },
    {
      "epoch": 1.996998799519808,
      "grad_norm": 0.002679846016690135,
      "learning_rate": 0.00018122249744185605,
      "loss": 23.0,
      "step": 3327
    },
    {
      "epoch": 1.9975990396158463,
      "grad_norm": 0.003047887934371829,
      "learning_rate": 0.00018121147596725015,
      "loss": 23.0,
      "step": 3328
    },
    {
      "epoch": 1.9981992797118848,
      "grad_norm": 0.0017722995253279805,
      "learning_rate": 0.00018120045159441555,
      "loss": 23.0,
      "step": 3329
    },
    {
      "epoch": 1.9987995198079231,
      "grad_norm": 0.0012083728797733784,
      "learning_rate": 0.00018118942432374569,
      "loss": 23.0,
      "step": 3330
    },
    {
      "epoch": 1.9993997599039615,
      "grad_norm": 0.0022785505279898643,
      "learning_rate": 0.00018117839415563412,
      "loss": 23.0,
      "step": 3331
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.0014194271061569452,
      "learning_rate": 0.0001811673610904745,
      "loss": 23.0,
      "step": 3332
    },
    {
      "epoch": 2.0006002400960385,
      "grad_norm": 0.0032136687077581882,
      "learning_rate": 0.0001811563251286605,
      "loss": 23.0,
      "step": 3333
    },
    {
      "epoch": 2.0012004801920766,
      "grad_norm": 0.00217724172398448,
      "learning_rate": 0.00018114528627058605,
      "loss": 23.0,
      "step": 3334
    },
    {
      "epoch": 2.001800720288115,
      "grad_norm": 0.0017727838130667806,
      "learning_rate": 0.00018113424451664508,
      "loss": 23.0,
      "step": 3335
    },
    {
      "epoch": 2.0024009603841537,
      "grad_norm": 0.00423310836777091,
      "learning_rate": 0.00018112319986723162,
      "loss": 23.0,
      "step": 3336
    },
    {
      "epoch": 2.0030012004801923,
      "grad_norm": 0.0028058600146323442,
      "learning_rate": 0.00018111215232273983,
      "loss": 23.0,
      "step": 3337
    },
    {
      "epoch": 2.0036014405762304,
      "grad_norm": 0.0015503312461078167,
      "learning_rate": 0.00018110110188356395,
      "loss": 23.0,
      "step": 3338
    },
    {
      "epoch": 2.004201680672269,
      "grad_norm": 0.0017939833924174309,
      "learning_rate": 0.00018109004855009837,
      "loss": 23.0,
      "step": 3339
    },
    {
      "epoch": 2.0048019207683074,
      "grad_norm": 0.0033577047288417816,
      "learning_rate": 0.00018107899232273756,
      "loss": 23.0,
      "step": 3340
    },
    {
      "epoch": 2.0054021608643455,
      "grad_norm": 0.0008726062951609492,
      "learning_rate": 0.00018106793320187605,
      "loss": 23.0,
      "step": 3341
    },
    {
      "epoch": 2.006002400960384,
      "grad_norm": 0.0011513312347233295,
      "learning_rate": 0.00018105687118790854,
      "loss": 23.0,
      "step": 3342
    },
    {
      "epoch": 2.0066026410564226,
      "grad_norm": 0.001895427005365491,
      "learning_rate": 0.00018104580628122984,
      "loss": 23.0,
      "step": 3343
    },
    {
      "epoch": 2.007202881152461,
      "grad_norm": 0.0009158776374533772,
      "learning_rate": 0.0001810347384822347,
      "loss": 23.0,
      "step": 3344
    },
    {
      "epoch": 2.0078031212484992,
      "grad_norm": 0.0008634175173938274,
      "learning_rate": 0.00018102366779131827,
      "loss": 23.0,
      "step": 3345
    },
    {
      "epoch": 2.008403361344538,
      "grad_norm": 0.0019585357513278723,
      "learning_rate": 0.00018101259420887553,
      "loss": 23.0,
      "step": 3346
    },
    {
      "epoch": 2.0090036014405763,
      "grad_norm": 0.005161345470696688,
      "learning_rate": 0.00018100151773530168,
      "loss": 23.0,
      "step": 3347
    },
    {
      "epoch": 2.009603841536615,
      "grad_norm": 0.003942587878555059,
      "learning_rate": 0.00018099043837099203,
      "loss": 23.0,
      "step": 3348
    },
    {
      "epoch": 2.010204081632653,
      "grad_norm": 0.0013700701529160142,
      "learning_rate": 0.00018097935611634198,
      "loss": 23.0,
      "step": 3349
    },
    {
      "epoch": 2.0108043217286915,
      "grad_norm": 0.0007138294167816639,
      "learning_rate": 0.000180968270971747,
      "loss": 23.0,
      "step": 3350
    },
    {
      "epoch": 2.01140456182473,
      "grad_norm": 0.005530905444175005,
      "learning_rate": 0.0001809571829376027,
      "loss": 23.0,
      "step": 3351
    },
    {
      "epoch": 2.012004801920768,
      "grad_norm": 0.0032923435792326927,
      "learning_rate": 0.0001809460920143048,
      "loss": 23.0,
      "step": 3352
    },
    {
      "epoch": 2.0126050420168067,
      "grad_norm": 0.0005104127340018749,
      "learning_rate": 0.0001809349982022491,
      "loss": 23.0,
      "step": 3353
    },
    {
      "epoch": 2.013205282112845,
      "grad_norm": 0.001574002206325531,
      "learning_rate": 0.0001809239015018315,
      "loss": 23.0,
      "step": 3354
    },
    {
      "epoch": 2.0138055222088838,
      "grad_norm": 0.002212009159848094,
      "learning_rate": 0.000180912801913448,
      "loss": 23.0,
      "step": 3355
    },
    {
      "epoch": 2.014405762304922,
      "grad_norm": 0.0011559105478227139,
      "learning_rate": 0.00018090169943749476,
      "loss": 23.0,
      "step": 3356
    },
    {
      "epoch": 2.0150060024009604,
      "grad_norm": 0.0033464916050434113,
      "learning_rate": 0.00018089059407436794,
      "loss": 23.0,
      "step": 3357
    },
    {
      "epoch": 2.015606242496999,
      "grad_norm": 0.0011934455251321197,
      "learning_rate": 0.0001808794858244639,
      "loss": 23.0,
      "step": 3358
    },
    {
      "epoch": 2.016206482593037,
      "grad_norm": 0.0006701741367578506,
      "learning_rate": 0.00018086837468817907,
      "loss": 23.0,
      "step": 3359
    },
    {
      "epoch": 2.0168067226890756,
      "grad_norm": 0.0014794671442359686,
      "learning_rate": 0.00018085726066590998,
      "loss": 23.0,
      "step": 3360
    },
    {
      "epoch": 2.017406962785114,
      "grad_norm": 0.00047398850438185036,
      "learning_rate": 0.00018084614375805323,
      "loss": 23.0,
      "step": 3361
    },
    {
      "epoch": 2.0180072028811527,
      "grad_norm": 0.0023361060302704573,
      "learning_rate": 0.00018083502396500556,
      "loss": 23.0,
      "step": 3362
    },
    {
      "epoch": 2.0186074429771907,
      "grad_norm": 0.0012610971461981535,
      "learning_rate": 0.0001808239012871638,
      "loss": 23.0,
      "step": 3363
    },
    {
      "epoch": 2.0192076830732293,
      "grad_norm": 0.004134600982069969,
      "learning_rate": 0.00018081277572492494,
      "loss": 23.0,
      "step": 3364
    },
    {
      "epoch": 2.019807923169268,
      "grad_norm": 0.000841438421048224,
      "learning_rate": 0.000180801647278686,
      "loss": 23.0,
      "step": 3365
    },
    {
      "epoch": 2.020408163265306,
      "grad_norm": 0.002190615748986602,
      "learning_rate": 0.0001807905159488441,
      "loss": 23.0,
      "step": 3366
    },
    {
      "epoch": 2.0210084033613445,
      "grad_norm": 0.0028436193242669106,
      "learning_rate": 0.00018077938173579648,
      "loss": 23.0,
      "step": 3367
    },
    {
      "epoch": 2.021608643457383,
      "grad_norm": 0.003500116989016533,
      "learning_rate": 0.00018076824463994056,
      "loss": 23.0,
      "step": 3368
    },
    {
      "epoch": 2.0222088835534215,
      "grad_norm": 0.0026991087943315506,
      "learning_rate": 0.0001807571046616737,
      "loss": 23.0,
      "step": 3369
    },
    {
      "epoch": 2.0228091236494596,
      "grad_norm": 0.0010631488403305411,
      "learning_rate": 0.00018074596180139353,
      "loss": 23.0,
      "step": 3370
    },
    {
      "epoch": 2.023409363745498,
      "grad_norm": 0.0020518836099654436,
      "learning_rate": 0.00018073481605949772,
      "loss": 23.0,
      "step": 3371
    },
    {
      "epoch": 2.0240096038415367,
      "grad_norm": 0.006052622105926275,
      "learning_rate": 0.00018072366743638395,
      "loss": 23.0,
      "step": 3372
    },
    {
      "epoch": 2.024609843937575,
      "grad_norm": 0.0021911882795393467,
      "learning_rate": 0.00018071251593245016,
      "loss": 23.0,
      "step": 3373
    },
    {
      "epoch": 2.0252100840336134,
      "grad_norm": 0.0032586664892733097,
      "learning_rate": 0.00018070136154809427,
      "loss": 23.0,
      "step": 3374
    },
    {
      "epoch": 2.025810324129652,
      "grad_norm": 0.000785050040576607,
      "learning_rate": 0.0001806902042837144,
      "loss": 23.0,
      "step": 3375
    },
    {
      "epoch": 2.0264105642256904,
      "grad_norm": 0.0007825555512681603,
      "learning_rate": 0.00018067904413970868,
      "loss": 23.0,
      "step": 3376
    },
    {
      "epoch": 2.0270108043217285,
      "grad_norm": 0.003356239991262555,
      "learning_rate": 0.00018066788111647543,
      "loss": 23.0,
      "step": 3377
    },
    {
      "epoch": 2.027611044417767,
      "grad_norm": 0.0014766575768589973,
      "learning_rate": 0.00018065671521441302,
      "loss": 23.0,
      "step": 3378
    },
    {
      "epoch": 2.0282112845138056,
      "grad_norm": 0.0009158303728327155,
      "learning_rate": 0.00018064554643391987,
      "loss": 23.0,
      "step": 3379
    },
    {
      "epoch": 2.028811524609844,
      "grad_norm": 0.0009348996100015938,
      "learning_rate": 0.00018063437477539466,
      "loss": 23.0,
      "step": 3380
    },
    {
      "epoch": 2.0294117647058822,
      "grad_norm": 0.0008702090126462281,
      "learning_rate": 0.00018062320023923598,
      "loss": 23.0,
      "step": 3381
    },
    {
      "epoch": 2.030012004801921,
      "grad_norm": 0.002282114699482918,
      "learning_rate": 0.0001806120228258427,
      "loss": 23.0,
      "step": 3382
    },
    {
      "epoch": 2.0306122448979593,
      "grad_norm": 0.005010796710848808,
      "learning_rate": 0.00018060084253561368,
      "loss": 23.0,
      "step": 3383
    },
    {
      "epoch": 2.0312124849939974,
      "grad_norm": 0.0017042094841599464,
      "learning_rate": 0.0001805896593689479,
      "loss": 23.0,
      "step": 3384
    },
    {
      "epoch": 2.031812725090036,
      "grad_norm": 0.0037521664053201675,
      "learning_rate": 0.00018057847332624452,
      "loss": 23.0,
      "step": 3385
    },
    {
      "epoch": 2.0324129651860745,
      "grad_norm": 0.0015214249724522233,
      "learning_rate": 0.00018056728440790266,
      "loss": 23.0,
      "step": 3386
    },
    {
      "epoch": 2.033013205282113,
      "grad_norm": 0.0004935787292197347,
      "learning_rate": 0.00018055609261432169,
      "loss": 23.0,
      "step": 3387
    },
    {
      "epoch": 2.033613445378151,
      "grad_norm": 0.003463397966697812,
      "learning_rate": 0.00018054489794590097,
      "loss": 23.0,
      "step": 3388
    },
    {
      "epoch": 2.0342136854741897,
      "grad_norm": 0.002584519563242793,
      "learning_rate": 0.00018053370040304004,
      "loss": 23.0,
      "step": 3389
    },
    {
      "epoch": 2.034813925570228,
      "grad_norm": 0.0012157384771853685,
      "learning_rate": 0.0001805224999861385,
      "loss": 23.0,
      "step": 3390
    },
    {
      "epoch": 2.0354141656662663,
      "grad_norm": 0.0029177102260291576,
      "learning_rate": 0.00018051129669559605,
      "loss": 23.0,
      "step": 3391
    },
    {
      "epoch": 2.036014405762305,
      "grad_norm": 0.0015428938204422593,
      "learning_rate": 0.00018050009053181257,
      "loss": 23.0,
      "step": 3392
    },
    {
      "epoch": 2.0366146458583434,
      "grad_norm": 0.0007340401643887162,
      "learning_rate": 0.00018048888149518788,
      "loss": 23.0,
      "step": 3393
    },
    {
      "epoch": 2.037214885954382,
      "grad_norm": 0.0011957536917179823,
      "learning_rate": 0.00018047766958612208,
      "loss": 23.0,
      "step": 3394
    },
    {
      "epoch": 2.03781512605042,
      "grad_norm": 0.0017125644953921437,
      "learning_rate": 0.00018046645480501524,
      "loss": 23.0,
      "step": 3395
    },
    {
      "epoch": 2.0384153661464586,
      "grad_norm": 0.0023938282392919064,
      "learning_rate": 0.0001804552371522676,
      "loss": 23.0,
      "step": 3396
    },
    {
      "epoch": 2.039015606242497,
      "grad_norm": 0.0033557454589754343,
      "learning_rate": 0.00018044401662827955,
      "loss": 23.0,
      "step": 3397
    },
    {
      "epoch": 2.039615846338535,
      "grad_norm": 0.0025200413074344397,
      "learning_rate": 0.00018043279323345143,
      "loss": 23.0,
      "step": 3398
    },
    {
      "epoch": 2.0402160864345738,
      "grad_norm": 0.0025761276483535767,
      "learning_rate": 0.00018042156696818385,
      "loss": 23.0,
      "step": 3399
    },
    {
      "epoch": 2.0408163265306123,
      "grad_norm": 0.0009423838928341866,
      "learning_rate": 0.00018041033783287738,
      "loss": 23.0,
      "step": 3400
    },
    {
      "epoch": 2.041416566626651,
      "grad_norm": 0.004442538134753704,
      "learning_rate": 0.00018039910582793282,
      "loss": 23.0,
      "step": 3401
    },
    {
      "epoch": 2.042016806722689,
      "grad_norm": 0.002732986817136407,
      "learning_rate": 0.000180387870953751,
      "loss": 23.0,
      "step": 3402
    },
    {
      "epoch": 2.0426170468187275,
      "grad_norm": 0.003854157403111458,
      "learning_rate": 0.0001803766332107328,
      "loss": 23.0,
      "step": 3403
    },
    {
      "epoch": 2.043217286914766,
      "grad_norm": 0.00363258458673954,
      "learning_rate": 0.00018036539259927932,
      "loss": 23.0,
      "step": 3404
    },
    {
      "epoch": 2.0438175270108045,
      "grad_norm": 0.004020926542580128,
      "learning_rate": 0.00018035414911979177,
      "loss": 23.0,
      "step": 3405
    },
    {
      "epoch": 2.0444177671068426,
      "grad_norm": 0.0015982519835233688,
      "learning_rate": 0.00018034290277267126,
      "loss": 23.0,
      "step": 3406
    },
    {
      "epoch": 2.045018007202881,
      "grad_norm": 0.0038941625971347094,
      "learning_rate": 0.00018033165355831924,
      "loss": 23.0,
      "step": 3407
    },
    {
      "epoch": 2.0456182472989197,
      "grad_norm": 0.0026739637833088636,
      "learning_rate": 0.00018032040147713716,
      "loss": 23.0,
      "step": 3408
    },
    {
      "epoch": 2.046218487394958,
      "grad_norm": 0.0016816748538985848,
      "learning_rate": 0.00018030914652952652,
      "loss": 23.0,
      "step": 3409
    },
    {
      "epoch": 2.0468187274909964,
      "grad_norm": 0.001945959753356874,
      "learning_rate": 0.00018029788871588905,
      "loss": 23.0,
      "step": 3410
    },
    {
      "epoch": 2.047418967587035,
      "grad_norm": 0.002473194617778063,
      "learning_rate": 0.0001802866280366265,
      "loss": 23.0,
      "step": 3411
    },
    {
      "epoch": 2.0480192076830734,
      "grad_norm": 0.003011916531249881,
      "learning_rate": 0.00018027536449214066,
      "loss": 23.0,
      "step": 3412
    },
    {
      "epoch": 2.0486194477791115,
      "grad_norm": 0.0023058606311678886,
      "learning_rate": 0.0001802640980828336,
      "loss": 23.0,
      "step": 3413
    },
    {
      "epoch": 2.04921968787515,
      "grad_norm": 0.0021154494024813175,
      "learning_rate": 0.00018025282880910733,
      "loss": 23.0,
      "step": 3414
    },
    {
      "epoch": 2.0498199279711886,
      "grad_norm": 0.000943707418628037,
      "learning_rate": 0.00018024155667136405,
      "loss": 23.0,
      "step": 3415
    },
    {
      "epoch": 2.0504201680672267,
      "grad_norm": 0.0005733675789088011,
      "learning_rate": 0.000180230281670006,
      "loss": 23.0,
      "step": 3416
    },
    {
      "epoch": 2.0510204081632653,
      "grad_norm": 0.0012774687493219972,
      "learning_rate": 0.00018021900380543558,
      "loss": 23.0,
      "step": 3417
    },
    {
      "epoch": 2.051620648259304,
      "grad_norm": 0.0015681550139561296,
      "learning_rate": 0.00018020772307805524,
      "loss": 23.0,
      "step": 3418
    },
    {
      "epoch": 2.0522208883553423,
      "grad_norm": 0.002922586165368557,
      "learning_rate": 0.00018019643948826762,
      "loss": 23.0,
      "step": 3419
    },
    {
      "epoch": 2.0528211284513804,
      "grad_norm": 0.000941049656830728,
      "learning_rate": 0.00018018515303647533,
      "loss": 23.0,
      "step": 3420
    },
    {
      "epoch": 2.053421368547419,
      "grad_norm": 0.0013102737721055746,
      "learning_rate": 0.0001801738637230812,
      "loss": 23.0,
      "step": 3421
    },
    {
      "epoch": 2.0540216086434575,
      "grad_norm": 0.0010620380053296685,
      "learning_rate": 0.00018016257154848813,
      "loss": 23.0,
      "step": 3422
    },
    {
      "epoch": 2.0546218487394956,
      "grad_norm": 0.001468815840780735,
      "learning_rate": 0.00018015127651309907,
      "loss": 23.0,
      "step": 3423
    },
    {
      "epoch": 2.055222088835534,
      "grad_norm": 0.002112776506692171,
      "learning_rate": 0.0001801399786173171,
      "loss": 23.0,
      "step": 3424
    },
    {
      "epoch": 2.0558223289315727,
      "grad_norm": 0.0043845935724675655,
      "learning_rate": 0.00018012867786154545,
      "loss": 23.0,
      "step": 3425
    },
    {
      "epoch": 2.0564225690276112,
      "grad_norm": 0.0023064163979142904,
      "learning_rate": 0.0001801173742461874,
      "loss": 23.0,
      "step": 3426
    },
    {
      "epoch": 2.0570228091236493,
      "grad_norm": 0.0015078387223184109,
      "learning_rate": 0.0001801060677716464,
      "loss": 23.0,
      "step": 3427
    },
    {
      "epoch": 2.057623049219688,
      "grad_norm": 0.0014458864461630583,
      "learning_rate": 0.00018009475843832587,
      "loss": 23.0,
      "step": 3428
    },
    {
      "epoch": 2.0582232893157264,
      "grad_norm": 0.0032603242434561253,
      "learning_rate": 0.00018008344624662943,
      "loss": 23.0,
      "step": 3429
    },
    {
      "epoch": 2.0588235294117645,
      "grad_norm": 0.0012629261473193765,
      "learning_rate": 0.00018007213119696077,
      "loss": 23.0,
      "step": 3430
    },
    {
      "epoch": 2.059423769507803,
      "grad_norm": 0.0016698999097570777,
      "learning_rate": 0.00018006081328972374,
      "loss": 23.0,
      "step": 3431
    },
    {
      "epoch": 2.0600240096038416,
      "grad_norm": 0.001395948464050889,
      "learning_rate": 0.00018004949252532223,
      "loss": 23.0,
      "step": 3432
    },
    {
      "epoch": 2.06062424969988,
      "grad_norm": 0.004593064077198505,
      "learning_rate": 0.00018003816890416023,
      "loss": 23.0,
      "step": 3433
    },
    {
      "epoch": 2.061224489795918,
      "grad_norm": 0.0021219372283667326,
      "learning_rate": 0.00018002684242664186,
      "loss": 23.0,
      "step": 3434
    },
    {
      "epoch": 2.0618247298919568,
      "grad_norm": 0.0006292082252912223,
      "learning_rate": 0.00018001551309317136,
      "loss": 23.0,
      "step": 3435
    },
    {
      "epoch": 2.0624249699879953,
      "grad_norm": 0.0024436218664050102,
      "learning_rate": 0.000180004180904153,
      "loss": 23.0,
      "step": 3436
    },
    {
      "epoch": 2.0630252100840334,
      "grad_norm": 0.0022101947106420994,
      "learning_rate": 0.00017999284585999126,
      "loss": 23.0,
      "step": 3437
    },
    {
      "epoch": 2.063625450180072,
      "grad_norm": 0.0024126351345330477,
      "learning_rate": 0.00017998150796109057,
      "loss": 23.0,
      "step": 3438
    },
    {
      "epoch": 2.0642256902761105,
      "grad_norm": 0.00167564966250211,
      "learning_rate": 0.00017997016720785563,
      "loss": 23.0,
      "step": 3439
    },
    {
      "epoch": 2.064825930372149,
      "grad_norm": 0.001377441338263452,
      "learning_rate": 0.0001799588236006911,
      "loss": 23.0,
      "step": 3440
    },
    {
      "epoch": 2.065426170468187,
      "grad_norm": 0.002041054656729102,
      "learning_rate": 0.00017994747714000187,
      "loss": 23.0,
      "step": 3441
    },
    {
      "epoch": 2.0660264105642256,
      "grad_norm": 0.003891473403200507,
      "learning_rate": 0.0001799361278261928,
      "loss": 23.0,
      "step": 3442
    },
    {
      "epoch": 2.066626650660264,
      "grad_norm": 0.0010193546768277884,
      "learning_rate": 0.00017992477565966896,
      "loss": 23.0,
      "step": 3443
    },
    {
      "epoch": 2.0672268907563027,
      "grad_norm": 0.0011679233284667134,
      "learning_rate": 0.00017991342064083547,
      "loss": 23.0,
      "step": 3444
    },
    {
      "epoch": 2.067827130852341,
      "grad_norm": 0.0009892381494864821,
      "learning_rate": 0.00017990206277009756,
      "loss": 23.0,
      "step": 3445
    },
    {
      "epoch": 2.0684273709483794,
      "grad_norm": 0.0028641563840210438,
      "learning_rate": 0.00017989070204786053,
      "loss": 23.0,
      "step": 3446
    },
    {
      "epoch": 2.069027611044418,
      "grad_norm": 0.002391669899225235,
      "learning_rate": 0.0001798793384745299,
      "loss": 23.0,
      "step": 3447
    },
    {
      "epoch": 2.069627851140456,
      "grad_norm": 0.0008025117567740381,
      "learning_rate": 0.0001798679720505111,
      "loss": 23.0,
      "step": 3448
    },
    {
      "epoch": 2.0702280912364945,
      "grad_norm": 0.0011154050007462502,
      "learning_rate": 0.00017985660277620985,
      "loss": 23.0,
      "step": 3449
    },
    {
      "epoch": 2.070828331332533,
      "grad_norm": 0.001934365602210164,
      "learning_rate": 0.00017984523065203188,
      "loss": 23.0,
      "step": 3450
    },
    {
      "epoch": 2.0714285714285716,
      "grad_norm": 0.0007151182508096099,
      "learning_rate": 0.000179833855678383,
      "loss": 23.0,
      "step": 3451
    },
    {
      "epoch": 2.0720288115246097,
      "grad_norm": 0.001716459752060473,
      "learning_rate": 0.00017982247785566918,
      "loss": 23.0,
      "step": 3452
    },
    {
      "epoch": 2.0726290516206483,
      "grad_norm": 0.0010225954465568066,
      "learning_rate": 0.00017981109718429643,
      "loss": 23.0,
      "step": 3453
    },
    {
      "epoch": 2.073229291716687,
      "grad_norm": 0.0027839874383062124,
      "learning_rate": 0.00017979971366467094,
      "loss": 23.0,
      "step": 3454
    },
    {
      "epoch": 2.073829531812725,
      "grad_norm": 0.0011828261194750667,
      "learning_rate": 0.00017978832729719893,
      "loss": 23.0,
      "step": 3455
    },
    {
      "epoch": 2.0744297719087634,
      "grad_norm": 0.00159098906442523,
      "learning_rate": 0.00017977693808228677,
      "loss": 23.0,
      "step": 3456
    },
    {
      "epoch": 2.075030012004802,
      "grad_norm": 0.0014060731045901775,
      "learning_rate": 0.0001797655460203409,
      "loss": 23.0,
      "step": 3457
    },
    {
      "epoch": 2.0756302521008405,
      "grad_norm": 0.001484134467318654,
      "learning_rate": 0.00017975415111176783,
      "loss": 23.0,
      "step": 3458
    },
    {
      "epoch": 2.0762304921968786,
      "grad_norm": 0.0017137357499450445,
      "learning_rate": 0.0001797427533569743,
      "loss": 23.0,
      "step": 3459
    },
    {
      "epoch": 2.076830732292917,
      "grad_norm": 0.0030945942271500826,
      "learning_rate": 0.000179731352756367,
      "loss": 23.0,
      "step": 3460
    },
    {
      "epoch": 2.0774309723889557,
      "grad_norm": 0.0011171615915372968,
      "learning_rate": 0.00017971994931035283,
      "loss": 23.0,
      "step": 3461
    },
    {
      "epoch": 2.078031212484994,
      "grad_norm": 0.001267662737518549,
      "learning_rate": 0.00017970854301933874,
      "loss": 23.0,
      "step": 3462
    },
    {
      "epoch": 2.0786314525810323,
      "grad_norm": 0.0021587491501122713,
      "learning_rate": 0.0001796971338837318,
      "loss": 23.0,
      "step": 3463
    },
    {
      "epoch": 2.079231692677071,
      "grad_norm": 0.002123309997841716,
      "learning_rate": 0.00017968572190393912,
      "loss": 23.0,
      "step": 3464
    },
    {
      "epoch": 2.0798319327731094,
      "grad_norm": 0.0012843715958297253,
      "learning_rate": 0.000179674307080368,
      "loss": 23.0,
      "step": 3465
    },
    {
      "epoch": 2.0804321728691475,
      "grad_norm": 0.0021665177773684263,
      "learning_rate": 0.00017966288941342583,
      "loss": 23.0,
      "step": 3466
    },
    {
      "epoch": 2.081032412965186,
      "grad_norm": 0.0019804276525974274,
      "learning_rate": 0.00017965146890352006,
      "loss": 23.0,
      "step": 3467
    },
    {
      "epoch": 2.0816326530612246,
      "grad_norm": 0.0025594120379537344,
      "learning_rate": 0.00017964004555105824,
      "loss": 23.0,
      "step": 3468
    },
    {
      "epoch": 2.082232893157263,
      "grad_norm": 0.002886942122131586,
      "learning_rate": 0.00017962861935644805,
      "loss": 23.0,
      "step": 3469
    },
    {
      "epoch": 2.082833133253301,
      "grad_norm": 0.001857664086855948,
      "learning_rate": 0.0001796171903200973,
      "loss": 23.0,
      "step": 3470
    },
    {
      "epoch": 2.0834333733493398,
      "grad_norm": 0.002572208410128951,
      "learning_rate": 0.0001796057584424138,
      "loss": 23.0,
      "step": 3471
    },
    {
      "epoch": 2.0840336134453783,
      "grad_norm": 0.002026068978011608,
      "learning_rate": 0.00017959432372380554,
      "loss": 23.0,
      "step": 3472
    },
    {
      "epoch": 2.0846338535414164,
      "grad_norm": 0.0016905996017158031,
      "learning_rate": 0.0001795828861646806,
      "loss": 23.0,
      "step": 3473
    },
    {
      "epoch": 2.085234093637455,
      "grad_norm": 0.0011786259710788727,
      "learning_rate": 0.00017957144576544718,
      "loss": 23.0,
      "step": 3474
    },
    {
      "epoch": 2.0858343337334935,
      "grad_norm": 0.0007971758022904396,
      "learning_rate": 0.00017956000252651353,
      "loss": 23.0,
      "step": 3475
    },
    {
      "epoch": 2.086434573829532,
      "grad_norm": 0.0009448099881410599,
      "learning_rate": 0.00017954855644828808,
      "loss": 23.0,
      "step": 3476
    },
    {
      "epoch": 2.08703481392557,
      "grad_norm": 0.002794134197756648,
      "learning_rate": 0.00017953710753117923,
      "loss": 23.0,
      "step": 3477
    },
    {
      "epoch": 2.0876350540216086,
      "grad_norm": 0.0017918311059474945,
      "learning_rate": 0.00017952565577559562,
      "loss": 23.0,
      "step": 3478
    },
    {
      "epoch": 2.088235294117647,
      "grad_norm": 0.0032223749440163374,
      "learning_rate": 0.00017951420118194594,
      "loss": 23.0,
      "step": 3479
    },
    {
      "epoch": 2.0888355342136853,
      "grad_norm": 0.0009358105598948896,
      "learning_rate": 0.00017950274375063893,
      "loss": 23.0,
      "step": 3480
    },
    {
      "epoch": 2.089435774309724,
      "grad_norm": 0.004243810195475817,
      "learning_rate": 0.00017949128348208353,
      "loss": 23.0,
      "step": 3481
    },
    {
      "epoch": 2.0900360144057624,
      "grad_norm": 0.002074790420010686,
      "learning_rate": 0.00017947982037668867,
      "loss": 23.0,
      "step": 3482
    },
    {
      "epoch": 2.090636254501801,
      "grad_norm": 0.0004455752205103636,
      "learning_rate": 0.00017946835443486347,
      "loss": 23.0,
      "step": 3483
    },
    {
      "epoch": 2.091236494597839,
      "grad_norm": 0.005418194457888603,
      "learning_rate": 0.00017945688565701714,
      "loss": 23.0,
      "step": 3484
    },
    {
      "epoch": 2.0918367346938775,
      "grad_norm": 0.0012525448109954596,
      "learning_rate": 0.0001794454140435589,
      "loss": 23.0,
      "step": 3485
    },
    {
      "epoch": 2.092436974789916,
      "grad_norm": 0.0017173114465549588,
      "learning_rate": 0.00017943393959489825,
      "loss": 23.0,
      "step": 3486
    },
    {
      "epoch": 2.093037214885954,
      "grad_norm": 0.0008452086476609111,
      "learning_rate": 0.00017942246231144457,
      "loss": 23.0,
      "step": 3487
    },
    {
      "epoch": 2.0936374549819927,
      "grad_norm": 0.0011205211048945785,
      "learning_rate": 0.00017941098219360755,
      "loss": 23.0,
      "step": 3488
    },
    {
      "epoch": 2.0942376950780313,
      "grad_norm": 0.0008696285076439381,
      "learning_rate": 0.00017939949924179684,
      "loss": 23.0,
      "step": 3489
    },
    {
      "epoch": 2.09483793517407,
      "grad_norm": 0.0013399510644376278,
      "learning_rate": 0.00017938801345642223,
      "loss": 23.0,
      "step": 3490
    },
    {
      "epoch": 2.095438175270108,
      "grad_norm": 0.0006139983888715506,
      "learning_rate": 0.00017937652483789363,
      "loss": 23.0,
      "step": 3491
    },
    {
      "epoch": 2.0960384153661464,
      "grad_norm": 0.0014349283883348107,
      "learning_rate": 0.00017936503338662104,
      "loss": 23.0,
      "step": 3492
    },
    {
      "epoch": 2.096638655462185,
      "grad_norm": 0.0010579722002148628,
      "learning_rate": 0.00017935353910301455,
      "loss": 23.0,
      "step": 3493
    },
    {
      "epoch": 2.097238895558223,
      "grad_norm": 0.0010757379932329059,
      "learning_rate": 0.0001793420419874844,
      "loss": 23.0,
      "step": 3494
    },
    {
      "epoch": 2.0978391356542616,
      "grad_norm": 0.0016692588105797768,
      "learning_rate": 0.00017933054204044083,
      "loss": 23.0,
      "step": 3495
    },
    {
      "epoch": 2.0984393757503,
      "grad_norm": 0.002716292394325137,
      "learning_rate": 0.00017931903926229428,
      "loss": 23.0,
      "step": 3496
    },
    {
      "epoch": 2.0990396158463387,
      "grad_norm": 0.0011049581225961447,
      "learning_rate": 0.00017930753365345527,
      "loss": 23.0,
      "step": 3497
    },
    {
      "epoch": 2.099639855942377,
      "grad_norm": 0.001968407304957509,
      "learning_rate": 0.0001792960252143344,
      "loss": 23.0,
      "step": 3498
    },
    {
      "epoch": 2.1002400960384153,
      "grad_norm": 0.0006715644267387688,
      "learning_rate": 0.00017928451394534233,
      "loss": 23.0,
      "step": 3499
    },
    {
      "epoch": 2.100840336134454,
      "grad_norm": 0.0011009502923116088,
      "learning_rate": 0.00017927299984688992,
      "loss": 23.0,
      "step": 3500
    },
    {
      "epoch": 2.1014405762304924,
      "grad_norm": 0.002423146739602089,
      "learning_rate": 0.00017926148291938808,
      "loss": 23.0,
      "step": 3501
    },
    {
      "epoch": 2.1020408163265305,
      "grad_norm": 0.004653971176594496,
      "learning_rate": 0.00017924996316324778,
      "loss": 23.0,
      "step": 3502
    },
    {
      "epoch": 2.102641056422569,
      "grad_norm": 0.002340496750548482,
      "learning_rate": 0.00017923844057888012,
      "loss": 23.0,
      "step": 3503
    },
    {
      "epoch": 2.1032412965186076,
      "grad_norm": 0.0018698208732530475,
      "learning_rate": 0.00017922691516669635,
      "loss": 23.0,
      "step": 3504
    },
    {
      "epoch": 2.1038415366146457,
      "grad_norm": 0.0017265928909182549,
      "learning_rate": 0.00017921538692710782,
      "loss": 23.0,
      "step": 3505
    },
    {
      "epoch": 2.104441776710684,
      "grad_norm": 0.0035198975820094347,
      "learning_rate": 0.00017920385586052588,
      "loss": 23.0,
      "step": 3506
    },
    {
      "epoch": 2.1050420168067228,
      "grad_norm": 0.005279972683638334,
      "learning_rate": 0.00017919232196736203,
      "loss": 23.0,
      "step": 3507
    },
    {
      "epoch": 2.1056422569027613,
      "grad_norm": 0.0019218252273276448,
      "learning_rate": 0.00017918078524802795,
      "loss": 23.0,
      "step": 3508
    },
    {
      "epoch": 2.1062424969987994,
      "grad_norm": 0.0037660326343029737,
      "learning_rate": 0.00017916924570293528,
      "loss": 23.0,
      "step": 3509
    },
    {
      "epoch": 2.106842737094838,
      "grad_norm": 0.0016380378510802984,
      "learning_rate": 0.00017915770333249594,
      "loss": 23.0,
      "step": 3510
    },
    {
      "epoch": 2.1074429771908765,
      "grad_norm": 0.0015107912477105856,
      "learning_rate": 0.00017914615813712173,
      "loss": 23.0,
      "step": 3511
    },
    {
      "epoch": 2.1080432172869146,
      "grad_norm": 0.0029202804435044527,
      "learning_rate": 0.00017913461011722474,
      "loss": 23.0,
      "step": 3512
    },
    {
      "epoch": 2.108643457382953,
      "grad_norm": 0.0013868259266018867,
      "learning_rate": 0.0001791230592732171,
      "loss": 23.0,
      "step": 3513
    },
    {
      "epoch": 2.1092436974789917,
      "grad_norm": 0.0014497030060738325,
      "learning_rate": 0.000179111505605511,
      "loss": 23.0,
      "step": 3514
    },
    {
      "epoch": 2.10984393757503,
      "grad_norm": 0.0015206874813884497,
      "learning_rate": 0.00017909994911451872,
      "loss": 23.0,
      "step": 3515
    },
    {
      "epoch": 2.1104441776710683,
      "grad_norm": 0.003641763236373663,
      "learning_rate": 0.00017908838980065274,
      "loss": 23.0,
      "step": 3516
    },
    {
      "epoch": 2.111044417767107,
      "grad_norm": 0.0021101627498865128,
      "learning_rate": 0.0001790768276643256,
      "loss": 23.0,
      "step": 3517
    },
    {
      "epoch": 2.1116446578631454,
      "grad_norm": 0.0026833494193851948,
      "learning_rate": 0.00017906526270594988,
      "loss": 23.0,
      "step": 3518
    },
    {
      "epoch": 2.1122448979591835,
      "grad_norm": 0.0010312756057828665,
      "learning_rate": 0.0001790536949259383,
      "loss": 23.0,
      "step": 3519
    },
    {
      "epoch": 2.112845138055222,
      "grad_norm": 0.0033930472563952208,
      "learning_rate": 0.00017904212432470371,
      "loss": 23.0,
      "step": 3520
    },
    {
      "epoch": 2.1134453781512605,
      "grad_norm": 0.0020478384103626013,
      "learning_rate": 0.00017903055090265902,
      "loss": 23.0,
      "step": 3521
    },
    {
      "epoch": 2.114045618247299,
      "grad_norm": 0.0017036633798852563,
      "learning_rate": 0.00017901897466021725,
      "loss": 23.0,
      "step": 3522
    },
    {
      "epoch": 2.114645858343337,
      "grad_norm": 0.0015527663053944707,
      "learning_rate": 0.00017900739559779155,
      "loss": 23.0,
      "step": 3523
    },
    {
      "epoch": 2.1152460984393757,
      "grad_norm": 0.001124212285503745,
      "learning_rate": 0.00017899581371579512,
      "loss": 23.0,
      "step": 3524
    },
    {
      "epoch": 2.1158463385354143,
      "grad_norm": 0.0014042783295735717,
      "learning_rate": 0.0001789842290146413,
      "loss": 23.0,
      "step": 3525
    },
    {
      "epoch": 2.116446578631453,
      "grad_norm": 0.003444352652877569,
      "learning_rate": 0.0001789726414947435,
      "loss": 23.0,
      "step": 3526
    },
    {
      "epoch": 2.117046818727491,
      "grad_norm": 0.003073784988373518,
      "learning_rate": 0.00017896105115651533,
      "loss": 23.0,
      "step": 3527
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 0.0012429555645212531,
      "learning_rate": 0.0001789494580003703,
      "loss": 23.0,
      "step": 3528
    },
    {
      "epoch": 2.118247298919568,
      "grad_norm": 0.004864856600761414,
      "learning_rate": 0.0001789378620267222,
      "loss": 23.0,
      "step": 3529
    },
    {
      "epoch": 2.118847539015606,
      "grad_norm": 0.0022995832841843367,
      "learning_rate": 0.00017892626323598486,
      "loss": 23.0,
      "step": 3530
    },
    {
      "epoch": 2.1194477791116446,
      "grad_norm": 0.00140533153899014,
      "learning_rate": 0.00017891466162857223,
      "loss": 23.0,
      "step": 3531
    },
    {
      "epoch": 2.120048019207683,
      "grad_norm": 0.0009306234424002469,
      "learning_rate": 0.0001789030572048983,
      "loss": 23.0,
      "step": 3532
    },
    {
      "epoch": 2.1206482593037217,
      "grad_norm": 0.003682507202029228,
      "learning_rate": 0.00017889144996537725,
      "loss": 23.0,
      "step": 3533
    },
    {
      "epoch": 2.12124849939976,
      "grad_norm": 0.0009380745468661189,
      "learning_rate": 0.00017887983991042323,
      "loss": 23.0,
      "step": 3534
    },
    {
      "epoch": 2.1218487394957983,
      "grad_norm": 0.0013799188891425729,
      "learning_rate": 0.00017886822704045064,
      "loss": 23.0,
      "step": 3535
    },
    {
      "epoch": 2.122448979591837,
      "grad_norm": 0.0009243764216080308,
      "learning_rate": 0.00017885661135587393,
      "loss": 23.0,
      "step": 3536
    },
    {
      "epoch": 2.123049219687875,
      "grad_norm": 0.0008515262743458152,
      "learning_rate": 0.00017884499285710762,
      "loss": 23.0,
      "step": 3537
    },
    {
      "epoch": 2.1236494597839135,
      "grad_norm": 0.0015884919557720423,
      "learning_rate": 0.00017883337154456629,
      "loss": 23.0,
      "step": 3538
    },
    {
      "epoch": 2.124249699879952,
      "grad_norm": 0.0032588981557637453,
      "learning_rate": 0.0001788217474186647,
      "loss": 23.0,
      "step": 3539
    },
    {
      "epoch": 2.1248499399759906,
      "grad_norm": 0.0024941442534327507,
      "learning_rate": 0.00017881012047981772,
      "loss": 23.0,
      "step": 3540
    },
    {
      "epoch": 2.1254501800720287,
      "grad_norm": 0.0017060820246115327,
      "learning_rate": 0.00017879849072844028,
      "loss": 23.0,
      "step": 3541
    },
    {
      "epoch": 2.1260504201680672,
      "grad_norm": 0.003131990320980549,
      "learning_rate": 0.0001787868581649474,
      "loss": 23.0,
      "step": 3542
    },
    {
      "epoch": 2.1266506602641058,
      "grad_norm": 0.0024005863815546036,
      "learning_rate": 0.00017877522278975417,
      "loss": 23.0,
      "step": 3543
    },
    {
      "epoch": 2.127250900360144,
      "grad_norm": 0.0006045355694368482,
      "learning_rate": 0.0001787635846032759,
      "loss": 23.0,
      "step": 3544
    },
    {
      "epoch": 2.1278511404561824,
      "grad_norm": 0.0035146684385836124,
      "learning_rate": 0.0001787519436059279,
      "loss": 23.0,
      "step": 3545
    },
    {
      "epoch": 2.128451380552221,
      "grad_norm": 0.0014111412456259131,
      "learning_rate": 0.00017874029979812563,
      "loss": 23.0,
      "step": 3546
    },
    {
      "epoch": 2.1290516206482595,
      "grad_norm": 0.0018674216698855162,
      "learning_rate": 0.00017872865318028459,
      "loss": 23.0,
      "step": 3547
    },
    {
      "epoch": 2.1296518607442976,
      "grad_norm": 0.0012715421617031097,
      "learning_rate": 0.00017871700375282043,
      "loss": 23.0,
      "step": 3548
    },
    {
      "epoch": 2.130252100840336,
      "grad_norm": 0.003209817223250866,
      "learning_rate": 0.0001787053515161489,
      "loss": 23.0,
      "step": 3549
    },
    {
      "epoch": 2.1308523409363747,
      "grad_norm": 0.001961090601980686,
      "learning_rate": 0.0001786936964706858,
      "loss": 23.0,
      "step": 3550
    },
    {
      "epoch": 2.1314525810324128,
      "grad_norm": 0.0013860109029337764,
      "learning_rate": 0.0001786820386168471,
      "loss": 23.0,
      "step": 3551
    },
    {
      "epoch": 2.1320528211284513,
      "grad_norm": 0.00047536773490719497,
      "learning_rate": 0.00017867037795504887,
      "loss": 23.0,
      "step": 3552
    },
    {
      "epoch": 2.13265306122449,
      "grad_norm": 0.0019673663191497326,
      "learning_rate": 0.00017865871448570717,
      "loss": 23.0,
      "step": 3553
    },
    {
      "epoch": 2.1332533013205284,
      "grad_norm": 0.0021188133396208286,
      "learning_rate": 0.00017864704820923832,
      "loss": 23.0,
      "step": 3554
    },
    {
      "epoch": 2.1338535414165665,
      "grad_norm": 0.0027213292196393013,
      "learning_rate": 0.0001786353791260586,
      "loss": 23.0,
      "step": 3555
    },
    {
      "epoch": 2.134453781512605,
      "grad_norm": 0.0027741962112486362,
      "learning_rate": 0.0001786237072365845,
      "loss": 23.0,
      "step": 3556
    },
    {
      "epoch": 2.1350540216086435,
      "grad_norm": 0.002745026256889105,
      "learning_rate": 0.00017861203254123252,
      "loss": 23.0,
      "step": 3557
    },
    {
      "epoch": 2.1356542617046816,
      "grad_norm": 0.0020450023002922535,
      "learning_rate": 0.00017860035504041932,
      "loss": 23.0,
      "step": 3558
    },
    {
      "epoch": 2.13625450180072,
      "grad_norm": 0.0010032516438513994,
      "learning_rate": 0.00017858867473456163,
      "loss": 23.0,
      "step": 3559
    },
    {
      "epoch": 2.1368547418967587,
      "grad_norm": 0.0014476818032562733,
      "learning_rate": 0.00017857699162407626,
      "loss": 23.0,
      "step": 3560
    },
    {
      "epoch": 2.1374549819927973,
      "grad_norm": 0.001177955069579184,
      "learning_rate": 0.00017856530570938022,
      "loss": 23.0,
      "step": 3561
    },
    {
      "epoch": 2.1380552220888354,
      "grad_norm": 0.002862164517864585,
      "learning_rate": 0.0001785536169908905,
      "loss": 23.0,
      "step": 3562
    },
    {
      "epoch": 2.138655462184874,
      "grad_norm": 0.0010038964683189988,
      "learning_rate": 0.00017854192546902427,
      "loss": 23.0,
      "step": 3563
    },
    {
      "epoch": 2.1392557022809124,
      "grad_norm": 0.0017746913945302367,
      "learning_rate": 0.00017853023114419877,
      "loss": 23.0,
      "step": 3564
    },
    {
      "epoch": 2.139855942376951,
      "grad_norm": 0.0028854040428996086,
      "learning_rate": 0.00017851853401683126,
      "loss": 23.0,
      "step": 3565
    },
    {
      "epoch": 2.140456182472989,
      "grad_norm": 0.001586511847563088,
      "learning_rate": 0.00017850683408733928,
      "loss": 23.0,
      "step": 3566
    },
    {
      "epoch": 2.1410564225690276,
      "grad_norm": 0.003254214534536004,
      "learning_rate": 0.00017849513135614035,
      "loss": 23.0,
      "step": 3567
    },
    {
      "epoch": 2.141656662665066,
      "grad_norm": 0.0012385440059006214,
      "learning_rate": 0.00017848342582365206,
      "loss": 23.0,
      "step": 3568
    },
    {
      "epoch": 2.1422569027611043,
      "grad_norm": 0.0018727433634921908,
      "learning_rate": 0.0001784717174902922,
      "loss": 23.0,
      "step": 3569
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 0.004589486867189407,
      "learning_rate": 0.00017846000635647863,
      "loss": 23.0,
      "step": 3570
    },
    {
      "epoch": 2.1434573829531813,
      "grad_norm": 0.0021497050765901804,
      "learning_rate": 0.0001784482924226292,
      "loss": 23.0,
      "step": 3571
    },
    {
      "epoch": 2.14405762304922,
      "grad_norm": 0.0012627036776393652,
      "learning_rate": 0.00017843657568916206,
      "loss": 23.0,
      "step": 3572
    },
    {
      "epoch": 2.144657863145258,
      "grad_norm": 0.001609223079867661,
      "learning_rate": 0.00017842485615649528,
      "loss": 23.0,
      "step": 3573
    },
    {
      "epoch": 2.1452581032412965,
      "grad_norm": 0.001351055921986699,
      "learning_rate": 0.00017841313382504711,
      "loss": 23.0,
      "step": 3574
    },
    {
      "epoch": 2.145858343337335,
      "grad_norm": 0.004179363604635,
      "learning_rate": 0.00017840140869523587,
      "loss": 23.0,
      "step": 3575
    },
    {
      "epoch": 2.146458583433373,
      "grad_norm": 0.0016952658770605922,
      "learning_rate": 0.00017838968076748008,
      "loss": 23.0,
      "step": 3576
    },
    {
      "epoch": 2.1470588235294117,
      "grad_norm": 0.0015167961828410625,
      "learning_rate": 0.00017837795004219817,
      "loss": 23.0,
      "step": 3577
    },
    {
      "epoch": 2.1476590636254502,
      "grad_norm": 0.003133446676656604,
      "learning_rate": 0.00017836621651980887,
      "loss": 23.0,
      "step": 3578
    },
    {
      "epoch": 2.1482593037214888,
      "grad_norm": 0.00226042908616364,
      "learning_rate": 0.00017835448020073088,
      "loss": 23.0,
      "step": 3579
    },
    {
      "epoch": 2.148859543817527,
      "grad_norm": 0.0013810255331918597,
      "learning_rate": 0.00017834274108538304,
      "loss": 23.0,
      "step": 3580
    },
    {
      "epoch": 2.1494597839135654,
      "grad_norm": 0.001641986076720059,
      "learning_rate": 0.0001783309991741843,
      "loss": 23.0,
      "step": 3581
    },
    {
      "epoch": 2.150060024009604,
      "grad_norm": 0.002287675393745303,
      "learning_rate": 0.0001783192544675537,
      "loss": 23.0,
      "step": 3582
    },
    {
      "epoch": 2.1506602641056425,
      "grad_norm": 0.0009976879227906466,
      "learning_rate": 0.00017830750696591033,
      "loss": 23.0,
      "step": 3583
    },
    {
      "epoch": 2.1512605042016806,
      "grad_norm": 0.0018353014020249248,
      "learning_rate": 0.0001782957566696735,
      "loss": 23.0,
      "step": 3584
    },
    {
      "epoch": 2.151860744297719,
      "grad_norm": 0.0019918165635317564,
      "learning_rate": 0.00017828400357926253,
      "loss": 23.0,
      "step": 3585
    },
    {
      "epoch": 2.1524609843937577,
      "grad_norm": 0.0026910703163594007,
      "learning_rate": 0.0001782722476950968,
      "loss": 23.0,
      "step": 3586
    },
    {
      "epoch": 2.1530612244897958,
      "grad_norm": 0.002473343163728714,
      "learning_rate": 0.00017826048901759588,
      "loss": 23.0,
      "step": 3587
    },
    {
      "epoch": 2.1536614645858343,
      "grad_norm": 0.002732550958171487,
      "learning_rate": 0.00017824872754717945,
      "loss": 23.0,
      "step": 3588
    },
    {
      "epoch": 2.154261704681873,
      "grad_norm": 0.0010042079957202077,
      "learning_rate": 0.00017823696328426723,
      "loss": 23.0,
      "step": 3589
    },
    {
      "epoch": 2.1548619447779114,
      "grad_norm": 0.0027955600526183844,
      "learning_rate": 0.000178225196229279,
      "loss": 23.0,
      "step": 3590
    },
    {
      "epoch": 2.1554621848739495,
      "grad_norm": 0.004130184184759855,
      "learning_rate": 0.00017821342638263478,
      "loss": 23.0,
      "step": 3591
    },
    {
      "epoch": 2.156062424969988,
      "grad_norm": 0.003868487663567066,
      "learning_rate": 0.00017820165374475452,
      "loss": 23.0,
      "step": 3592
    },
    {
      "epoch": 2.1566626650660266,
      "grad_norm": 0.0011650120140984654,
      "learning_rate": 0.00017818987831605843,
      "loss": 23.0,
      "step": 3593
    },
    {
      "epoch": 2.1572629051620646,
      "grad_norm": 0.002393965143710375,
      "learning_rate": 0.00017817810009696666,
      "loss": 23.0,
      "step": 3594
    },
    {
      "epoch": 2.157863145258103,
      "grad_norm": 0.0035790205001831055,
      "learning_rate": 0.00017816631908789966,
      "loss": 23.0,
      "step": 3595
    },
    {
      "epoch": 2.1584633853541417,
      "grad_norm": 0.003235345706343651,
      "learning_rate": 0.00017815453528927778,
      "loss": 23.0,
      "step": 3596
    },
    {
      "epoch": 2.1590636254501803,
      "grad_norm": 0.0013717797119170427,
      "learning_rate": 0.00017814274870152155,
      "loss": 23.0,
      "step": 3597
    },
    {
      "epoch": 2.1596638655462184,
      "grad_norm": 0.0011841889936476946,
      "learning_rate": 0.0001781309593250516,
      "loss": 23.0,
      "step": 3598
    },
    {
      "epoch": 2.160264105642257,
      "grad_norm": 0.0035572301130741835,
      "learning_rate": 0.00017811916716028872,
      "loss": 23.0,
      "step": 3599
    },
    {
      "epoch": 2.1608643457382954,
      "grad_norm": 0.0009041174198500812,
      "learning_rate": 0.00017810737220765374,
      "loss": 23.0,
      "step": 3600
    },
    {
      "epoch": 2.1614645858343335,
      "grad_norm": 0.0011678171576932073,
      "learning_rate": 0.00017809557446756756,
      "loss": 23.0,
      "step": 3601
    },
    {
      "epoch": 2.162064825930372,
      "grad_norm": 0.0014870171435177326,
      "learning_rate": 0.0001780837739404512,
      "loss": 23.0,
      "step": 3602
    },
    {
      "epoch": 2.1626650660264106,
      "grad_norm": 0.0015179289039224386,
      "learning_rate": 0.00017807197062672577,
      "loss": 23.0,
      "step": 3603
    },
    {
      "epoch": 2.163265306122449,
      "grad_norm": 0.0020413794554769993,
      "learning_rate": 0.00017806016452681254,
      "loss": 23.0,
      "step": 3604
    },
    {
      "epoch": 2.1638655462184873,
      "grad_norm": 0.0032616599928587675,
      "learning_rate": 0.00017804835564113288,
      "loss": 23.0,
      "step": 3605
    },
    {
      "epoch": 2.164465786314526,
      "grad_norm": 0.0011516850208863616,
      "learning_rate": 0.00017803654397010815,
      "loss": 23.0,
      "step": 3606
    },
    {
      "epoch": 2.1650660264105643,
      "grad_norm": 0.003489278955385089,
      "learning_rate": 0.0001780247295141599,
      "loss": 23.0,
      "step": 3607
    },
    {
      "epoch": 2.1656662665066024,
      "grad_norm": 0.003842744044959545,
      "learning_rate": 0.00017801291227370974,
      "loss": 23.0,
      "step": 3608
    },
    {
      "epoch": 2.166266506602641,
      "grad_norm": 0.001340657938271761,
      "learning_rate": 0.0001780010922491794,
      "loss": 23.0,
      "step": 3609
    },
    {
      "epoch": 2.1668667466986795,
      "grad_norm": 0.0008930175099521875,
      "learning_rate": 0.00017798926944099077,
      "loss": 23.0,
      "step": 3610
    },
    {
      "epoch": 2.167466986794718,
      "grad_norm": 0.0036189949605613947,
      "learning_rate": 0.0001779774438495657,
      "loss": 23.0,
      "step": 3611
    },
    {
      "epoch": 2.168067226890756,
      "grad_norm": 0.003450029995292425,
      "learning_rate": 0.00017796561547532623,
      "loss": 23.0,
      "step": 3612
    },
    {
      "epoch": 2.1686674669867947,
      "grad_norm": 0.0013816222781315446,
      "learning_rate": 0.00017795378431869453,
      "loss": 23.0,
      "step": 3613
    },
    {
      "epoch": 2.1692677070828332,
      "grad_norm": 0.003153825644403696,
      "learning_rate": 0.00017794195038009275,
      "loss": 23.0,
      "step": 3614
    },
    {
      "epoch": 2.1698679471788713,
      "grad_norm": 0.001024477300234139,
      "learning_rate": 0.00017793011365994325,
      "loss": 23.0,
      "step": 3615
    },
    {
      "epoch": 2.17046818727491,
      "grad_norm": 0.003891037544235587,
      "learning_rate": 0.00017791827415866848,
      "loss": 23.0,
      "step": 3616
    },
    {
      "epoch": 2.1710684273709484,
      "grad_norm": 0.002924687694758177,
      "learning_rate": 0.00017790643187669093,
      "loss": 23.0,
      "step": 3617
    },
    {
      "epoch": 2.171668667466987,
      "grad_norm": 0.009289652109146118,
      "learning_rate": 0.00017789458681443323,
      "loss": 23.0,
      "step": 3618
    },
    {
      "epoch": 2.172268907563025,
      "grad_norm": 0.0021330954041332006,
      "learning_rate": 0.00017788273897231807,
      "loss": 23.0,
      "step": 3619
    },
    {
      "epoch": 2.1728691476590636,
      "grad_norm": 0.0019930656999349594,
      "learning_rate": 0.0001778708883507683,
      "loss": 23.0,
      "step": 3620
    },
    {
      "epoch": 2.173469387755102,
      "grad_norm": 0.00517197186127305,
      "learning_rate": 0.00017785903495020686,
      "loss": 23.0,
      "step": 3621
    },
    {
      "epoch": 2.1740696278511407,
      "grad_norm": 0.0015679982025176287,
      "learning_rate": 0.00017784717877105673,
      "loss": 23.0,
      "step": 3622
    },
    {
      "epoch": 2.1746698679471788,
      "grad_norm": 0.004141742363572121,
      "learning_rate": 0.00017783531981374102,
      "loss": 23.0,
      "step": 3623
    },
    {
      "epoch": 2.1752701080432173,
      "grad_norm": 0.00065791723318398,
      "learning_rate": 0.00017782345807868297,
      "loss": 23.0,
      "step": 3624
    },
    {
      "epoch": 2.175870348139256,
      "grad_norm": 0.004961501341313124,
      "learning_rate": 0.0001778115935663059,
      "loss": 23.0,
      "step": 3625
    },
    {
      "epoch": 2.176470588235294,
      "grad_norm": 0.0023005243856459856,
      "learning_rate": 0.0001777997262770332,
      "loss": 23.0,
      "step": 3626
    },
    {
      "epoch": 2.1770708283313325,
      "grad_norm": 0.00899723544716835,
      "learning_rate": 0.00017778785621128836,
      "loss": 23.0,
      "step": 3627
    },
    {
      "epoch": 2.177671068427371,
      "grad_norm": 0.0029707015492022038,
      "learning_rate": 0.0001777759833694951,
      "loss": 23.0,
      "step": 3628
    },
    {
      "epoch": 2.1782713085234096,
      "grad_norm": 0.0018353236373513937,
      "learning_rate": 0.000177764107752077,
      "loss": 23.0,
      "step": 3629
    },
    {
      "epoch": 2.1788715486194477,
      "grad_norm": 0.0010108323767781258,
      "learning_rate": 0.00017775222935945796,
      "loss": 23.0,
      "step": 3630
    },
    {
      "epoch": 2.179471788715486,
      "grad_norm": 0.0008860941743478179,
      "learning_rate": 0.00017774034819206181,
      "loss": 23.0,
      "step": 3631
    },
    {
      "epoch": 2.1800720288115247,
      "grad_norm": 0.003606445388868451,
      "learning_rate": 0.00017772846425031266,
      "loss": 23.0,
      "step": 3632
    },
    {
      "epoch": 2.180672268907563,
      "grad_norm": 0.0016308276681229472,
      "learning_rate": 0.00017771657753463453,
      "loss": 23.0,
      "step": 3633
    },
    {
      "epoch": 2.1812725090036014,
      "grad_norm": 0.0007045452948659658,
      "learning_rate": 0.00017770468804545169,
      "loss": 23.0,
      "step": 3634
    },
    {
      "epoch": 2.18187274909964,
      "grad_norm": 0.0013048743130639195,
      "learning_rate": 0.0001776927957831884,
      "loss": 23.0,
      "step": 3635
    },
    {
      "epoch": 2.1824729891956784,
      "grad_norm": 0.0012544161872938275,
      "learning_rate": 0.00017768090074826904,
      "loss": 23.0,
      "step": 3636
    },
    {
      "epoch": 2.1830732292917165,
      "grad_norm": 0.00237638084217906,
      "learning_rate": 0.0001776690029411182,
      "loss": 23.0,
      "step": 3637
    },
    {
      "epoch": 2.183673469387755,
      "grad_norm": 0.0048915608786046505,
      "learning_rate": 0.0001776571023621604,
      "loss": 23.0,
      "step": 3638
    },
    {
      "epoch": 2.1842737094837936,
      "grad_norm": 0.000844002643134445,
      "learning_rate": 0.0001776451990118204,
      "loss": 23.0,
      "step": 3639
    },
    {
      "epoch": 2.184873949579832,
      "grad_norm": 0.00372367724776268,
      "learning_rate": 0.00017763329289052296,
      "loss": 23.0,
      "step": 3640
    },
    {
      "epoch": 2.1854741896758703,
      "grad_norm": 0.0023307260125875473,
      "learning_rate": 0.000177621383998693,
      "loss": 23.0,
      "step": 3641
    },
    {
      "epoch": 2.186074429771909,
      "grad_norm": 0.0026479533407837152,
      "learning_rate": 0.0001776094723367555,
      "loss": 23.0,
      "step": 3642
    },
    {
      "epoch": 2.1866746698679473,
      "grad_norm": 0.00475787278264761,
      "learning_rate": 0.00017759755790513557,
      "loss": 23.0,
      "step": 3643
    },
    {
      "epoch": 2.1872749099639854,
      "grad_norm": 0.0008646330097690225,
      "learning_rate": 0.00017758564070425843,
      "loss": 23.0,
      "step": 3644
    },
    {
      "epoch": 2.187875150060024,
      "grad_norm": 0.000730241765268147,
      "learning_rate": 0.00017757372073454931,
      "loss": 23.0,
      "step": 3645
    },
    {
      "epoch": 2.1884753901560625,
      "grad_norm": 0.0020450761076062918,
      "learning_rate": 0.00017756179799643368,
      "loss": 23.0,
      "step": 3646
    },
    {
      "epoch": 2.189075630252101,
      "grad_norm": 0.001998509978875518,
      "learning_rate": 0.00017754987249033694,
      "loss": 23.0,
      "step": 3647
    },
    {
      "epoch": 2.189675870348139,
      "grad_norm": 0.0017022277461364865,
      "learning_rate": 0.00017753794421668478,
      "loss": 23.0,
      "step": 3648
    },
    {
      "epoch": 2.1902761104441777,
      "grad_norm": 0.0010228764731436968,
      "learning_rate": 0.0001775260131759028,
      "loss": 23.0,
      "step": 3649
    },
    {
      "epoch": 2.1908763505402162,
      "grad_norm": 0.0021690605208277702,
      "learning_rate": 0.00017751407936841688,
      "loss": 23.0,
      "step": 3650
    },
    {
      "epoch": 2.1914765906362543,
      "grad_norm": 0.0016400478780269623,
      "learning_rate": 0.00017750214279465282,
      "loss": 23.0,
      "step": 3651
    },
    {
      "epoch": 2.192076830732293,
      "grad_norm": 0.002070733578875661,
      "learning_rate": 0.00017749020345503666,
      "loss": 23.0,
      "step": 3652
    },
    {
      "epoch": 2.1926770708283314,
      "grad_norm": 0.0014095326187089086,
      "learning_rate": 0.00017747826134999445,
      "loss": 23.0,
      "step": 3653
    },
    {
      "epoch": 2.19327731092437,
      "grad_norm": 0.0019122892990708351,
      "learning_rate": 0.00017746631647995242,
      "loss": 23.0,
      "step": 3654
    },
    {
      "epoch": 2.193877551020408,
      "grad_norm": 0.0029271540697664022,
      "learning_rate": 0.00017745436884533683,
      "loss": 23.0,
      "step": 3655
    },
    {
      "epoch": 2.1944777911164466,
      "grad_norm": 0.002827174263074994,
      "learning_rate": 0.00017744241844657398,
      "loss": 23.0,
      "step": 3656
    },
    {
      "epoch": 2.195078031212485,
      "grad_norm": 0.0012498522410169244,
      "learning_rate": 0.00017743046528409049,
      "loss": 23.0,
      "step": 3657
    },
    {
      "epoch": 2.195678271308523,
      "grad_norm": 0.003829399822279811,
      "learning_rate": 0.00017741850935831287,
      "loss": 23.0,
      "step": 3658
    },
    {
      "epoch": 2.1962785114045618,
      "grad_norm": 0.0021660744678229094,
      "learning_rate": 0.00017740655066966773,
      "loss": 23.0,
      "step": 3659
    },
    {
      "epoch": 2.1968787515006003,
      "grad_norm": 0.002248313743621111,
      "learning_rate": 0.00017739458921858198,
      "loss": 23.0,
      "step": 3660
    },
    {
      "epoch": 2.197478991596639,
      "grad_norm": 0.0018800124526023865,
      "learning_rate": 0.00017738262500548236,
      "loss": 23.0,
      "step": 3661
    },
    {
      "epoch": 2.198079231692677,
      "grad_norm": 0.0030018959660083055,
      "learning_rate": 0.00017737065803079597,
      "loss": 23.0,
      "step": 3662
    },
    {
      "epoch": 2.1986794717887155,
      "grad_norm": 0.0009485651389695704,
      "learning_rate": 0.00017735868829494978,
      "loss": 23.0,
      "step": 3663
    },
    {
      "epoch": 2.199279711884754,
      "grad_norm": 0.002609929535537958,
      "learning_rate": 0.000177346715798371,
      "loss": 23.0,
      "step": 3664
    },
    {
      "epoch": 2.199879951980792,
      "grad_norm": 0.0013200175017118454,
      "learning_rate": 0.0001773347405414869,
      "loss": 23.0,
      "step": 3665
    },
    {
      "epoch": 2.2004801920768307,
      "grad_norm": 0.0007020602351985872,
      "learning_rate": 0.00017732276252472483,
      "loss": 23.0,
      "step": 3666
    },
    {
      "epoch": 2.201080432172869,
      "grad_norm": 0.0026193074882030487,
      "learning_rate": 0.00017731078174851227,
      "loss": 23.0,
      "step": 3667
    },
    {
      "epoch": 2.2016806722689077,
      "grad_norm": 0.001026253798045218,
      "learning_rate": 0.00017729879821327678,
      "loss": 23.0,
      "step": 3668
    },
    {
      "epoch": 2.202280912364946,
      "grad_norm": 0.0017351701389998198,
      "learning_rate": 0.000177286811919446,
      "loss": 23.0,
      "step": 3669
    },
    {
      "epoch": 2.2028811524609844,
      "grad_norm": 0.0014646050985902548,
      "learning_rate": 0.0001772748228674477,
      "loss": 23.0,
      "step": 3670
    },
    {
      "epoch": 2.203481392557023,
      "grad_norm": 0.0030125817283988,
      "learning_rate": 0.0001772628310577098,
      "loss": 23.0,
      "step": 3671
    },
    {
      "epoch": 2.204081632653061,
      "grad_norm": 0.004487418103963137,
      "learning_rate": 0.00017725083649066016,
      "loss": 23.0,
      "step": 3672
    },
    {
      "epoch": 2.2046818727490995,
      "grad_norm": 0.0021111650858074427,
      "learning_rate": 0.0001772388391667269,
      "loss": 23.0,
      "step": 3673
    },
    {
      "epoch": 2.205282112845138,
      "grad_norm": 0.00048810808220878243,
      "learning_rate": 0.00017722683908633812,
      "loss": 23.0,
      "step": 3674
    },
    {
      "epoch": 2.2058823529411766,
      "grad_norm": 0.003283157479017973,
      "learning_rate": 0.00017721483624992215,
      "loss": 23.0,
      "step": 3675
    },
    {
      "epoch": 2.2064825930372147,
      "grad_norm": 0.0016707901377230883,
      "learning_rate": 0.00017720283065790726,
      "loss": 23.0,
      "step": 3676
    },
    {
      "epoch": 2.2070828331332533,
      "grad_norm": 0.001358469482511282,
      "learning_rate": 0.00017719082231072193,
      "loss": 23.0,
      "step": 3677
    },
    {
      "epoch": 2.207683073229292,
      "grad_norm": 0.0013346460182219744,
      "learning_rate": 0.00017717881120879475,
      "loss": 23.0,
      "step": 3678
    },
    {
      "epoch": 2.20828331332533,
      "grad_norm": 0.001013239729218185,
      "learning_rate": 0.0001771667973525543,
      "loss": 23.0,
      "step": 3679
    },
    {
      "epoch": 2.2088835534213684,
      "grad_norm": 0.001471264404244721,
      "learning_rate": 0.00017715478074242932,
      "loss": 23.0,
      "step": 3680
    },
    {
      "epoch": 2.209483793517407,
      "grad_norm": 0.0015335743082687259,
      "learning_rate": 0.00017714276137884873,
      "loss": 23.0,
      "step": 3681
    },
    {
      "epoch": 2.2100840336134455,
      "grad_norm": 0.005060260649770498,
      "learning_rate": 0.0001771307392622414,
      "loss": 23.0,
      "step": 3682
    },
    {
      "epoch": 2.2106842737094836,
      "grad_norm": 0.0011684778146445751,
      "learning_rate": 0.00017711871439303638,
      "loss": 23.0,
      "step": 3683
    },
    {
      "epoch": 2.211284513805522,
      "grad_norm": 0.0035212927032262087,
      "learning_rate": 0.00017710668677166282,
      "loss": 23.0,
      "step": 3684
    },
    {
      "epoch": 2.2118847539015607,
      "grad_norm": 0.0023291881661862135,
      "learning_rate": 0.00017709465639854997,
      "loss": 23.0,
      "step": 3685
    },
    {
      "epoch": 2.2124849939975992,
      "grad_norm": 0.0015391027554869652,
      "learning_rate": 0.00017708262327412712,
      "loss": 23.0,
      "step": 3686
    },
    {
      "epoch": 2.2130852340936373,
      "grad_norm": 0.001002484350465238,
      "learning_rate": 0.00017707058739882374,
      "loss": 23.0,
      "step": 3687
    },
    {
      "epoch": 2.213685474189676,
      "grad_norm": 0.0014180615544319153,
      "learning_rate": 0.00017705854877306934,
      "loss": 23.0,
      "step": 3688
    },
    {
      "epoch": 2.2142857142857144,
      "grad_norm": 0.0008763477671891451,
      "learning_rate": 0.00017704650739729357,
      "loss": 23.0,
      "step": 3689
    },
    {
      "epoch": 2.2148859543817525,
      "grad_norm": 0.0022361702285706997,
      "learning_rate": 0.00017703446327192612,
      "loss": 23.0,
      "step": 3690
    },
    {
      "epoch": 2.215486194477791,
      "grad_norm": 0.0015827770112082362,
      "learning_rate": 0.0001770224163973968,
      "loss": 23.0,
      "step": 3691
    },
    {
      "epoch": 2.2160864345738296,
      "grad_norm": 0.001904232893139124,
      "learning_rate": 0.00017701036677413563,
      "loss": 23.0,
      "step": 3692
    },
    {
      "epoch": 2.216686674669868,
      "grad_norm": 0.0013081013457849622,
      "learning_rate": 0.00017699831440257251,
      "loss": 23.0,
      "step": 3693
    },
    {
      "epoch": 2.2172869147659062,
      "grad_norm": 0.004817170090973377,
      "learning_rate": 0.00017698625928313765,
      "loss": 23.0,
      "step": 3694
    },
    {
      "epoch": 2.2178871548619448,
      "grad_norm": 0.004000441636890173,
      "learning_rate": 0.0001769742014162612,
      "loss": 23.0,
      "step": 3695
    },
    {
      "epoch": 2.2184873949579833,
      "grad_norm": 0.0027444609440863132,
      "learning_rate": 0.00017696214080237352,
      "loss": 23.0,
      "step": 3696
    },
    {
      "epoch": 2.2190876350540214,
      "grad_norm": 0.002896189922466874,
      "learning_rate": 0.000176950077441905,
      "loss": 23.0,
      "step": 3697
    },
    {
      "epoch": 2.21968787515006,
      "grad_norm": 0.002128873486071825,
      "learning_rate": 0.00017693801133528616,
      "loss": 23.0,
      "step": 3698
    },
    {
      "epoch": 2.2202881152460985,
      "grad_norm": 0.0011179738212376833,
      "learning_rate": 0.0001769259424829476,
      "loss": 23.0,
      "step": 3699
    },
    {
      "epoch": 2.220888355342137,
      "grad_norm": 0.002058587037026882,
      "learning_rate": 0.00017691387088532003,
      "loss": 23.0,
      "step": 3700
    },
    {
      "epoch": 2.221488595438175,
      "grad_norm": 0.0022081804927438498,
      "learning_rate": 0.00017690179654283425,
      "loss": 23.0,
      "step": 3701
    },
    {
      "epoch": 2.2220888355342137,
      "grad_norm": 0.0023553965147584677,
      "learning_rate": 0.0001768897194559212,
      "loss": 23.0,
      "step": 3702
    },
    {
      "epoch": 2.222689075630252,
      "grad_norm": 0.001018383540213108,
      "learning_rate": 0.00017687763962501182,
      "loss": 23.0,
      "step": 3703
    },
    {
      "epoch": 2.2232893157262907,
      "grad_norm": 0.0007036729366518557,
      "learning_rate": 0.0001768655570505372,
      "loss": 23.0,
      "step": 3704
    },
    {
      "epoch": 2.223889555822329,
      "grad_norm": 0.0008182197925634682,
      "learning_rate": 0.0001768534717329286,
      "loss": 23.0,
      "step": 3705
    },
    {
      "epoch": 2.2244897959183674,
      "grad_norm": 0.0010957723716273904,
      "learning_rate": 0.00017684138367261731,
      "loss": 23.0,
      "step": 3706
    },
    {
      "epoch": 2.225090036014406,
      "grad_norm": 0.0017919333186000586,
      "learning_rate": 0.00017682929287003466,
      "loss": 23.0,
      "step": 3707
    },
    {
      "epoch": 2.225690276110444,
      "grad_norm": 0.003724980168044567,
      "learning_rate": 0.0001768171993256122,
      "loss": 23.0,
      "step": 3708
    },
    {
      "epoch": 2.2262905162064826,
      "grad_norm": 0.0013054338051006198,
      "learning_rate": 0.00017680510303978147,
      "loss": 23.0,
      "step": 3709
    },
    {
      "epoch": 2.226890756302521,
      "grad_norm": 0.0018630026606842875,
      "learning_rate": 0.0001767930040129742,
      "loss": 23.0,
      "step": 3710
    },
    {
      "epoch": 2.2274909963985596,
      "grad_norm": 0.0014727768721058965,
      "learning_rate": 0.00017678090224562213,
      "loss": 23.0,
      "step": 3711
    },
    {
      "epoch": 2.2280912364945977,
      "grad_norm": 0.0010810722596943378,
      "learning_rate": 0.0001767687977381572,
      "loss": 23.0,
      "step": 3712
    },
    {
      "epoch": 2.2286914765906363,
      "grad_norm": 0.001895120949484408,
      "learning_rate": 0.00017675669049101132,
      "loss": 23.0,
      "step": 3713
    },
    {
      "epoch": 2.229291716686675,
      "grad_norm": 0.0021297503262758255,
      "learning_rate": 0.0001767445805046166,
      "loss": 23.0,
      "step": 3714
    },
    {
      "epoch": 2.229891956782713,
      "grad_norm": 0.0009311452740803361,
      "learning_rate": 0.00017673246777940522,
      "loss": 23.0,
      "step": 3715
    },
    {
      "epoch": 2.2304921968787514,
      "grad_norm": 0.0026380294002592564,
      "learning_rate": 0.00017672035231580942,
      "loss": 23.0,
      "step": 3716
    },
    {
      "epoch": 2.23109243697479,
      "grad_norm": 0.0012547700898721814,
      "learning_rate": 0.00017670823411426162,
      "loss": 23.0,
      "step": 3717
    },
    {
      "epoch": 2.2316926770708285,
      "grad_norm": 0.0012008714256808162,
      "learning_rate": 0.0001766961131751943,
      "loss": 23.0,
      "step": 3718
    },
    {
      "epoch": 2.2322929171668666,
      "grad_norm": 0.0007726665935479105,
      "learning_rate": 0.00017668398949903993,
      "loss": 23.0,
      "step": 3719
    },
    {
      "epoch": 2.232893157262905,
      "grad_norm": 0.003182122251018882,
      "learning_rate": 0.00017667186308623122,
      "loss": 23.0,
      "step": 3720
    },
    {
      "epoch": 2.2334933973589437,
      "grad_norm": 0.0025740358978509903,
      "learning_rate": 0.000176659733937201,
      "loss": 23.0,
      "step": 3721
    },
    {
      "epoch": 2.234093637454982,
      "grad_norm": 0.0025537547189742327,
      "learning_rate": 0.00017664760205238203,
      "loss": 23.0,
      "step": 3722
    },
    {
      "epoch": 2.2346938775510203,
      "grad_norm": 0.0030393393244594336,
      "learning_rate": 0.00017663546743220733,
      "loss": 23.0,
      "step": 3723
    },
    {
      "epoch": 2.235294117647059,
      "grad_norm": 0.0024942343588918447,
      "learning_rate": 0.00017662333007710987,
      "loss": 23.0,
      "step": 3724
    },
    {
      "epoch": 2.2358943577430974,
      "grad_norm": 0.0016120598884299397,
      "learning_rate": 0.0001766111899875229,
      "loss": 23.0,
      "step": 3725
    },
    {
      "epoch": 2.2364945978391355,
      "grad_norm": 0.00033892018836922944,
      "learning_rate": 0.0001765990471638796,
      "loss": 23.0,
      "step": 3726
    },
    {
      "epoch": 2.237094837935174,
      "grad_norm": 0.002562824171036482,
      "learning_rate": 0.00017658690160661338,
      "loss": 23.0,
      "step": 3727
    },
    {
      "epoch": 2.2376950780312126,
      "grad_norm": 0.0022175584454089403,
      "learning_rate": 0.00017657475331615762,
      "loss": 23.0,
      "step": 3728
    },
    {
      "epoch": 2.2382953181272507,
      "grad_norm": 0.001756917918100953,
      "learning_rate": 0.0001765626022929459,
      "loss": 23.0,
      "step": 3729
    },
    {
      "epoch": 2.2388955582232892,
      "grad_norm": 0.003884568577632308,
      "learning_rate": 0.00017655044853741184,
      "loss": 23.0,
      "step": 3730
    },
    {
      "epoch": 2.2394957983193278,
      "grad_norm": 0.001167623675428331,
      "learning_rate": 0.00017653829204998918,
      "loss": 23.0,
      "step": 3731
    },
    {
      "epoch": 2.2400960384153663,
      "grad_norm": 0.0010216180235147476,
      "learning_rate": 0.00017652613283111176,
      "loss": 23.0,
      "step": 3732
    },
    {
      "epoch": 2.2406962785114044,
      "grad_norm": 0.003365274053066969,
      "learning_rate": 0.00017651397088121353,
      "loss": 23.0,
      "step": 3733
    },
    {
      "epoch": 2.241296518607443,
      "grad_norm": 0.0018816734664142132,
      "learning_rate": 0.00017650180620072846,
      "loss": 23.0,
      "step": 3734
    },
    {
      "epoch": 2.2418967587034815,
      "grad_norm": 0.0029328542295843363,
      "learning_rate": 0.00017648963879009075,
      "loss": 23.0,
      "step": 3735
    },
    {
      "epoch": 2.2424969987995196,
      "grad_norm": 0.0007874789298512042,
      "learning_rate": 0.00017647746864973458,
      "loss": 23.0,
      "step": 3736
    },
    {
      "epoch": 2.243097238895558,
      "grad_norm": 0.0030753707978874445,
      "learning_rate": 0.00017646529578009426,
      "loss": 23.0,
      "step": 3737
    },
    {
      "epoch": 2.2436974789915967,
      "grad_norm": 0.0010473758447915316,
      "learning_rate": 0.00017645312018160422,
      "loss": 23.0,
      "step": 3738
    },
    {
      "epoch": 2.244297719087635,
      "grad_norm": 0.0018478642450645566,
      "learning_rate": 0.00017644094185469898,
      "loss": 23.0,
      "step": 3739
    },
    {
      "epoch": 2.2448979591836733,
      "grad_norm": 0.0013350150547921658,
      "learning_rate": 0.0001764287607998132,
      "loss": 23.0,
      "step": 3740
    },
    {
      "epoch": 2.245498199279712,
      "grad_norm": 0.0012342286063358188,
      "learning_rate": 0.00017641657701738148,
      "loss": 23.0,
      "step": 3741
    },
    {
      "epoch": 2.2460984393757504,
      "grad_norm": 0.0013594189658761024,
      "learning_rate": 0.00017640439050783875,
      "loss": 23.0,
      "step": 3742
    },
    {
      "epoch": 2.246698679471789,
      "grad_norm": 0.0033450396731495857,
      "learning_rate": 0.00017639220127161983,
      "loss": 23.0,
      "step": 3743
    },
    {
      "epoch": 2.247298919567827,
      "grad_norm": 0.002889930969104171,
      "learning_rate": 0.00017638000930915974,
      "loss": 23.0,
      "step": 3744
    },
    {
      "epoch": 2.2478991596638656,
      "grad_norm": 0.0019997726194560528,
      "learning_rate": 0.00017636781462089364,
      "loss": 23.0,
      "step": 3745
    },
    {
      "epoch": 2.248499399759904,
      "grad_norm": 0.0012661672662943602,
      "learning_rate": 0.0001763556172072566,
      "loss": 23.0,
      "step": 3746
    },
    {
      "epoch": 2.249099639855942,
      "grad_norm": 0.007048753555864096,
      "learning_rate": 0.00017634341706868403,
      "loss": 23.0,
      "step": 3747
    },
    {
      "epoch": 2.2496998799519807,
      "grad_norm": 0.0015775463543832302,
      "learning_rate": 0.00017633121420561131,
      "loss": 23.0,
      "step": 3748
    },
    {
      "epoch": 2.2503001200480193,
      "grad_norm": 0.0020231371745467186,
      "learning_rate": 0.00017631900861847389,
      "loss": 23.0,
      "step": 3749
    },
    {
      "epoch": 2.250900360144058,
      "grad_norm": 0.0012863476295024157,
      "learning_rate": 0.00017630680030770735,
      "loss": 23.0,
      "step": 3750
    },
    {
      "epoch": 2.251500600240096,
      "grad_norm": 0.0016621587565168738,
      "learning_rate": 0.0001762945892737474,
      "loss": 23.0,
      "step": 3751
    },
    {
      "epoch": 2.2521008403361344,
      "grad_norm": 0.00226738303899765,
      "learning_rate": 0.00017628237551702982,
      "loss": 23.0,
      "step": 3752
    },
    {
      "epoch": 2.252701080432173,
      "grad_norm": 0.0014520992990583181,
      "learning_rate": 0.00017627015903799048,
      "loss": 23.0,
      "step": 3753
    },
    {
      "epoch": 2.2533013205282115,
      "grad_norm": 0.0013434399152174592,
      "learning_rate": 0.00017625793983706533,
      "loss": 23.0,
      "step": 3754
    },
    {
      "epoch": 2.2539015606242496,
      "grad_norm": 0.0007136189378798008,
      "learning_rate": 0.00017624571791469048,
      "loss": 23.0,
      "step": 3755
    },
    {
      "epoch": 2.254501800720288,
      "grad_norm": 0.002455267123878002,
      "learning_rate": 0.00017623349327130206,
      "loss": 23.0,
      "step": 3756
    },
    {
      "epoch": 2.2551020408163267,
      "grad_norm": 0.0009659494389779866,
      "learning_rate": 0.00017622126590733644,
      "loss": 23.0,
      "step": 3757
    },
    {
      "epoch": 2.255702280912365,
      "grad_norm": 0.002808001358062029,
      "learning_rate": 0.00017620903582322983,
      "loss": 23.0,
      "step": 3758
    },
    {
      "epoch": 2.2563025210084033,
      "grad_norm": 0.0018349566962569952,
      "learning_rate": 0.0001761968030194188,
      "loss": 23.0,
      "step": 3759
    },
    {
      "epoch": 2.256902761104442,
      "grad_norm": 0.002876591170206666,
      "learning_rate": 0.00017618456749633987,
      "loss": 23.0,
      "step": 3760
    },
    {
      "epoch": 2.2575030012004804,
      "grad_norm": 0.0012843102449551225,
      "learning_rate": 0.00017617232925442968,
      "loss": 23.0,
      "step": 3761
    },
    {
      "epoch": 2.2581032412965185,
      "grad_norm": 0.0011905563296750188,
      "learning_rate": 0.00017616008829412502,
      "loss": 23.0,
      "step": 3762
    },
    {
      "epoch": 2.258703481392557,
      "grad_norm": 0.000979992444626987,
      "learning_rate": 0.00017614784461586266,
      "loss": 23.0,
      "step": 3763
    },
    {
      "epoch": 2.2593037214885956,
      "grad_norm": 0.002167720813304186,
      "learning_rate": 0.00017613559822007963,
      "loss": 23.0,
      "step": 3764
    },
    {
      "epoch": 2.2599039615846337,
      "grad_norm": 0.0008691643597558141,
      "learning_rate": 0.00017612334910721298,
      "loss": 23.0,
      "step": 3765
    },
    {
      "epoch": 2.2605042016806722,
      "grad_norm": 0.0019571988377720118,
      "learning_rate": 0.00017611109727769977,
      "loss": 23.0,
      "step": 3766
    },
    {
      "epoch": 2.2611044417767108,
      "grad_norm": 0.0012904416071251035,
      "learning_rate": 0.00017609884273197732,
      "loss": 23.0,
      "step": 3767
    },
    {
      "epoch": 2.2617046818727493,
      "grad_norm": 0.0017439661314710975,
      "learning_rate": 0.00017608658547048288,
      "loss": 23.0,
      "step": 3768
    },
    {
      "epoch": 2.2623049219687874,
      "grad_norm": 0.003602192969992757,
      "learning_rate": 0.00017607432549365394,
      "loss": 23.0,
      "step": 3769
    },
    {
      "epoch": 2.262905162064826,
      "grad_norm": 0.002050132257863879,
      "learning_rate": 0.00017606206280192801,
      "loss": 23.0,
      "step": 3770
    },
    {
      "epoch": 2.2635054021608645,
      "grad_norm": 0.0030350079759955406,
      "learning_rate": 0.00017604979739574273,
      "loss": 23.0,
      "step": 3771
    },
    {
      "epoch": 2.2641056422569026,
      "grad_norm": 0.0011385931866243482,
      "learning_rate": 0.0001760375292755358,
      "loss": 23.0,
      "step": 3772
    },
    {
      "epoch": 2.264705882352941,
      "grad_norm": 0.0015834002988412976,
      "learning_rate": 0.000176025258441745,
      "loss": 23.0,
      "step": 3773
    },
    {
      "epoch": 2.2653061224489797,
      "grad_norm": 0.002169420477002859,
      "learning_rate": 0.00017601298489480834,
      "loss": 23.0,
      "step": 3774
    },
    {
      "epoch": 2.265906362545018,
      "grad_norm": 0.0020891001913696527,
      "learning_rate": 0.00017600070863516373,
      "loss": 23.0,
      "step": 3775
    },
    {
      "epoch": 2.2665066026410563,
      "grad_norm": 0.0015407392056658864,
      "learning_rate": 0.00017598842966324937,
      "loss": 23.0,
      "step": 3776
    },
    {
      "epoch": 2.267106842737095,
      "grad_norm": 0.0012298728106543422,
      "learning_rate": 0.0001759761479795034,
      "loss": 23.0,
      "step": 3777
    },
    {
      "epoch": 2.2677070828331334,
      "grad_norm": 0.0022670759353786707,
      "learning_rate": 0.00017596386358436413,
      "loss": 23.0,
      "step": 3778
    },
    {
      "epoch": 2.2683073229291715,
      "grad_norm": 0.0009981478797271848,
      "learning_rate": 0.00017595157647827,
      "loss": 23.0,
      "step": 3779
    },
    {
      "epoch": 2.26890756302521,
      "grad_norm": 0.0004938762285746634,
      "learning_rate": 0.00017593928666165945,
      "loss": 23.0,
      "step": 3780
    },
    {
      "epoch": 2.2695078031212486,
      "grad_norm": 0.0011169624049216509,
      "learning_rate": 0.00017592699413497114,
      "loss": 23.0,
      "step": 3781
    },
    {
      "epoch": 2.270108043217287,
      "grad_norm": 0.003080305876210332,
      "learning_rate": 0.00017591469889864368,
      "loss": 23.0,
      "step": 3782
    },
    {
      "epoch": 2.270708283313325,
      "grad_norm": 0.004471518564969301,
      "learning_rate": 0.00017590240095311587,
      "loss": 23.0,
      "step": 3783
    },
    {
      "epoch": 2.2713085234093637,
      "grad_norm": 0.0011711474508047104,
      "learning_rate": 0.00017589010029882664,
      "loss": 23.0,
      "step": 3784
    },
    {
      "epoch": 2.2719087635054023,
      "grad_norm": 0.0011285408399999142,
      "learning_rate": 0.00017587779693621495,
      "loss": 23.0,
      "step": 3785
    },
    {
      "epoch": 2.2725090036014404,
      "grad_norm": 0.0020412777084857225,
      "learning_rate": 0.00017586549086571988,
      "loss": 23.0,
      "step": 3786
    },
    {
      "epoch": 2.273109243697479,
      "grad_norm": 0.001386416843160987,
      "learning_rate": 0.00017585318208778058,
      "loss": 23.0,
      "step": 3787
    },
    {
      "epoch": 2.2737094837935174,
      "grad_norm": 0.0021522578317672014,
      "learning_rate": 0.00017584087060283632,
      "loss": 23.0,
      "step": 3788
    },
    {
      "epoch": 2.274309723889556,
      "grad_norm": 0.0017842737725004554,
      "learning_rate": 0.00017582855641132646,
      "loss": 23.0,
      "step": 3789
    },
    {
      "epoch": 2.274909963985594,
      "grad_norm": 0.0008319435291923583,
      "learning_rate": 0.0001758162395136905,
      "loss": 23.0,
      "step": 3790
    },
    {
      "epoch": 2.2755102040816326,
      "grad_norm": 0.0022286258172243834,
      "learning_rate": 0.000175803919910368,
      "loss": 23.0,
      "step": 3791
    },
    {
      "epoch": 2.276110444177671,
      "grad_norm": 0.0015450834762305021,
      "learning_rate": 0.00017579159760179854,
      "loss": 23.0,
      "step": 3792
    },
    {
      "epoch": 2.2767106842737093,
      "grad_norm": 0.0017403220990672708,
      "learning_rate": 0.00017577927258842193,
      "loss": 23.0,
      "step": 3793
    },
    {
      "epoch": 2.277310924369748,
      "grad_norm": 0.0021773509215563536,
      "learning_rate": 0.000175766944870678,
      "loss": 23.0,
      "step": 3794
    },
    {
      "epoch": 2.2779111644657863,
      "grad_norm": 0.0007699204725213349,
      "learning_rate": 0.00017575461444900676,
      "loss": 23.0,
      "step": 3795
    },
    {
      "epoch": 2.278511404561825,
      "grad_norm": 0.001452354365028441,
      "learning_rate": 0.00017574228132384814,
      "loss": 23.0,
      "step": 3796
    },
    {
      "epoch": 2.279111644657863,
      "grad_norm": 0.0027013334911316633,
      "learning_rate": 0.00017572994549564235,
      "loss": 23.0,
      "step": 3797
    },
    {
      "epoch": 2.2797118847539015,
      "grad_norm": 0.0016655053477734327,
      "learning_rate": 0.0001757176069648296,
      "loss": 23.0,
      "step": 3798
    },
    {
      "epoch": 2.28031212484994,
      "grad_norm": 0.0017022525426000357,
      "learning_rate": 0.00017570526573185022,
      "loss": 23.0,
      "step": 3799
    },
    {
      "epoch": 2.280912364945978,
      "grad_norm": 0.0011267677182331681,
      "learning_rate": 0.00017569292179714466,
      "loss": 23.0,
      "step": 3800
    },
    {
      "epoch": 2.2815126050420167,
      "grad_norm": 0.0012624160153791308,
      "learning_rate": 0.00017568057516115343,
      "loss": 23.0,
      "step": 3801
    },
    {
      "epoch": 2.2821128451380552,
      "grad_norm": 0.0012623511720448732,
      "learning_rate": 0.00017566822582431714,
      "loss": 23.0,
      "step": 3802
    },
    {
      "epoch": 2.2827130852340938,
      "grad_norm": 0.002993621863424778,
      "learning_rate": 0.0001756558737870765,
      "loss": 23.0,
      "step": 3803
    },
    {
      "epoch": 2.283313325330132,
      "grad_norm": 0.001764954999089241,
      "learning_rate": 0.00017564351904987232,
      "loss": 23.0,
      "step": 3804
    },
    {
      "epoch": 2.2839135654261704,
      "grad_norm": 0.0033577526919543743,
      "learning_rate": 0.00017563116161314557,
      "loss": 23.0,
      "step": 3805
    },
    {
      "epoch": 2.284513805522209,
      "grad_norm": 0.0030459025874733925,
      "learning_rate": 0.0001756188014773372,
      "loss": 23.0,
      "step": 3806
    },
    {
      "epoch": 2.2851140456182475,
      "grad_norm": 0.0020036788191646338,
      "learning_rate": 0.00017560643864288831,
      "loss": 23.0,
      "step": 3807
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 0.003118950640782714,
      "learning_rate": 0.00017559407311024008,
      "loss": 23.0,
      "step": 3808
    },
    {
      "epoch": 2.286314525810324,
      "grad_norm": 0.0023817415349185467,
      "learning_rate": 0.00017558170487983388,
      "loss": 23.0,
      "step": 3809
    },
    {
      "epoch": 2.2869147659063627,
      "grad_norm": 0.0020011558663100004,
      "learning_rate": 0.00017556933395211099,
      "loss": 23.0,
      "step": 3810
    },
    {
      "epoch": 2.287515006002401,
      "grad_norm": 0.0019172454485669732,
      "learning_rate": 0.000175556960327513,
      "loss": 23.0,
      "step": 3811
    },
    {
      "epoch": 2.2881152460984393,
      "grad_norm": 0.005320819094777107,
      "learning_rate": 0.00017554458400648146,
      "loss": 23.0,
      "step": 3812
    },
    {
      "epoch": 2.288715486194478,
      "grad_norm": 0.0011504385620355606,
      "learning_rate": 0.00017553220498945802,
      "loss": 23.0,
      "step": 3813
    },
    {
      "epoch": 2.2893157262905164,
      "grad_norm": 0.004282000940293074,
      "learning_rate": 0.00017551982327688447,
      "loss": 23.0,
      "step": 3814
    },
    {
      "epoch": 2.2899159663865545,
      "grad_norm": 0.0016033421270549297,
      "learning_rate": 0.0001755074388692027,
      "loss": 23.0,
      "step": 3815
    },
    {
      "epoch": 2.290516206482593,
      "grad_norm": 0.001392611302435398,
      "learning_rate": 0.00017549505176685468,
      "loss": 23.0,
      "step": 3816
    },
    {
      "epoch": 2.2911164465786316,
      "grad_norm": 0.0024148111697286367,
      "learning_rate": 0.00017548266197028243,
      "loss": 23.0,
      "step": 3817
    },
    {
      "epoch": 2.29171668667467,
      "grad_norm": 0.0022317797411233187,
      "learning_rate": 0.00017547026947992817,
      "loss": 23.0,
      "step": 3818
    },
    {
      "epoch": 2.292316926770708,
      "grad_norm": 0.00245701358653605,
      "learning_rate": 0.0001754578742962341,
      "loss": 23.0,
      "step": 3819
    },
    {
      "epoch": 2.2929171668667467,
      "grad_norm": 0.0017128176987171173,
      "learning_rate": 0.00017544547641964263,
      "loss": 23.0,
      "step": 3820
    },
    {
      "epoch": 2.2935174069627853,
      "grad_norm": 0.0031775077804923058,
      "learning_rate": 0.00017543307585059613,
      "loss": 23.0,
      "step": 3821
    },
    {
      "epoch": 2.2941176470588234,
      "grad_norm": 0.002050775336101651,
      "learning_rate": 0.00017542067258953723,
      "loss": 23.0,
      "step": 3822
    },
    {
      "epoch": 2.294717887154862,
      "grad_norm": 0.0014890647726133466,
      "learning_rate": 0.00017540826663690851,
      "loss": 23.0,
      "step": 3823
    },
    {
      "epoch": 2.2953181272509005,
      "grad_norm": 0.004480647388845682,
      "learning_rate": 0.00017539585799315272,
      "loss": 23.0,
      "step": 3824
    },
    {
      "epoch": 2.295918367346939,
      "grad_norm": 0.0015886913752183318,
      "learning_rate": 0.0001753834466587127,
      "loss": 23.0,
      "step": 3825
    },
    {
      "epoch": 2.296518607442977,
      "grad_norm": 0.0019695230294018984,
      "learning_rate": 0.0001753710326340314,
      "loss": 23.0,
      "step": 3826
    },
    {
      "epoch": 2.2971188475390156,
      "grad_norm": 0.0018076825654134154,
      "learning_rate": 0.0001753586159195518,
      "loss": 23.0,
      "step": 3827
    },
    {
      "epoch": 2.297719087635054,
      "grad_norm": 0.002728229621425271,
      "learning_rate": 0.00017534619651571705,
      "loss": 23.0,
      "step": 3828
    },
    {
      "epoch": 2.2983193277310923,
      "grad_norm": 0.0017432133900001645,
      "learning_rate": 0.0001753337744229704,
      "loss": 23.0,
      "step": 3829
    },
    {
      "epoch": 2.298919567827131,
      "grad_norm": 0.0018603568896651268,
      "learning_rate": 0.00017532134964175504,
      "loss": 23.0,
      "step": 3830
    },
    {
      "epoch": 2.2995198079231693,
      "grad_norm": 0.002048965310677886,
      "learning_rate": 0.0001753089221725145,
      "loss": 23.0,
      "step": 3831
    },
    {
      "epoch": 2.300120048019208,
      "grad_norm": 0.001226543914526701,
      "learning_rate": 0.00017529649201569226,
      "loss": 23.0,
      "step": 3832
    },
    {
      "epoch": 2.300720288115246,
      "grad_norm": 0.004013392608612776,
      "learning_rate": 0.0001752840591717319,
      "loss": 23.0,
      "step": 3833
    },
    {
      "epoch": 2.3013205282112845,
      "grad_norm": 0.0008342837681993842,
      "learning_rate": 0.00017527162364107713,
      "loss": 23.0,
      "step": 3834
    },
    {
      "epoch": 2.301920768307323,
      "grad_norm": 0.0007289492059499025,
      "learning_rate": 0.00017525918542417171,
      "loss": 23.0,
      "step": 3835
    },
    {
      "epoch": 2.302521008403361,
      "grad_norm": 0.0029447299893945456,
      "learning_rate": 0.00017524674452145956,
      "loss": 23.0,
      "step": 3836
    },
    {
      "epoch": 2.3031212484993997,
      "grad_norm": 0.002568695694208145,
      "learning_rate": 0.00017523430093338468,
      "loss": 23.0,
      "step": 3837
    },
    {
      "epoch": 2.3037214885954382,
      "grad_norm": 0.0012608342804014683,
      "learning_rate": 0.0001752218546603911,
      "loss": 23.0,
      "step": 3838
    },
    {
      "epoch": 2.304321728691477,
      "grad_norm": 0.00195410312153399,
      "learning_rate": 0.00017520940570292303,
      "loss": 23.0,
      "step": 3839
    },
    {
      "epoch": 2.304921968787515,
      "grad_norm": 0.003413090016692877,
      "learning_rate": 0.00017519695406142472,
      "loss": 23.0,
      "step": 3840
    },
    {
      "epoch": 2.3055222088835534,
      "grad_norm": 0.002465988975018263,
      "learning_rate": 0.00017518449973634055,
      "loss": 23.0,
      "step": 3841
    },
    {
      "epoch": 2.306122448979592,
      "grad_norm": 0.00117593037430197,
      "learning_rate": 0.00017517204272811501,
      "loss": 23.0,
      "step": 3842
    },
    {
      "epoch": 2.30672268907563,
      "grad_norm": 0.00045529394992627203,
      "learning_rate": 0.0001751595830371926,
      "loss": 23.0,
      "step": 3843
    },
    {
      "epoch": 2.3073229291716686,
      "grad_norm": 0.0013716558460146189,
      "learning_rate": 0.000175147120664018,
      "loss": 23.0,
      "step": 3844
    },
    {
      "epoch": 2.307923169267707,
      "grad_norm": 0.0009694201871752739,
      "learning_rate": 0.000175134655609036,
      "loss": 23.0,
      "step": 3845
    },
    {
      "epoch": 2.3085234093637457,
      "grad_norm": 0.002793167717754841,
      "learning_rate": 0.00017512218787269138,
      "loss": 23.0,
      "step": 3846
    },
    {
      "epoch": 2.3091236494597838,
      "grad_norm": 0.0012160228798165917,
      "learning_rate": 0.0001751097174554291,
      "loss": 23.0,
      "step": 3847
    },
    {
      "epoch": 2.3097238895558223,
      "grad_norm": 0.0041273715905845165,
      "learning_rate": 0.00017509724435769427,
      "loss": 23.0,
      "step": 3848
    },
    {
      "epoch": 2.310324129651861,
      "grad_norm": 0.0019413263071328402,
      "learning_rate": 0.00017508476857993192,
      "loss": 23.0,
      "step": 3849
    },
    {
      "epoch": 2.310924369747899,
      "grad_norm": 0.000679299992043525,
      "learning_rate": 0.00017507229012258732,
      "loss": 23.0,
      "step": 3850
    },
    {
      "epoch": 2.3115246098439375,
      "grad_norm": 0.0014612323138862848,
      "learning_rate": 0.0001750598089861058,
      "loss": 23.0,
      "step": 3851
    },
    {
      "epoch": 2.312124849939976,
      "grad_norm": 0.002256182488054037,
      "learning_rate": 0.00017504732517093278,
      "loss": 23.0,
      "step": 3852
    },
    {
      "epoch": 2.3127250900360146,
      "grad_norm": 0.0009005808969959617,
      "learning_rate": 0.00017503483867751377,
      "loss": 23.0,
      "step": 3853
    },
    {
      "epoch": 2.3133253301320527,
      "grad_norm": 0.0015893650706857443,
      "learning_rate": 0.00017502234950629437,
      "loss": 23.0,
      "step": 3854
    },
    {
      "epoch": 2.313925570228091,
      "grad_norm": 0.001747640548273921,
      "learning_rate": 0.0001750098576577203,
      "loss": 23.0,
      "step": 3855
    },
    {
      "epoch": 2.3145258103241297,
      "grad_norm": 0.0005708754761144519,
      "learning_rate": 0.00017499736313223737,
      "loss": 23.0,
      "step": 3856
    },
    {
      "epoch": 2.315126050420168,
      "grad_norm": 0.0028467399533838034,
      "learning_rate": 0.00017498486593029144,
      "loss": 23.0,
      "step": 3857
    },
    {
      "epoch": 2.3157262905162064,
      "grad_norm": 0.0014178903074935079,
      "learning_rate": 0.00017497236605232855,
      "loss": 23.0,
      "step": 3858
    },
    {
      "epoch": 2.316326530612245,
      "grad_norm": 0.005164925009012222,
      "learning_rate": 0.00017495986349879477,
      "loss": 23.0,
      "step": 3859
    },
    {
      "epoch": 2.3169267707082835,
      "grad_norm": 0.0027939407154917717,
      "learning_rate": 0.00017494735827013627,
      "loss": 23.0,
      "step": 3860
    },
    {
      "epoch": 2.3175270108043216,
      "grad_norm": 0.002403461141511798,
      "learning_rate": 0.00017493485036679935,
      "loss": 23.0,
      "step": 3861
    },
    {
      "epoch": 2.31812725090036,
      "grad_norm": 0.003365919226780534,
      "learning_rate": 0.00017492233978923035,
      "loss": 23.0,
      "step": 3862
    },
    {
      "epoch": 2.3187274909963986,
      "grad_norm": 0.001767502399161458,
      "learning_rate": 0.00017490982653787577,
      "loss": 23.0,
      "step": 3863
    },
    {
      "epoch": 2.3193277310924367,
      "grad_norm": 0.0037461065221577883,
      "learning_rate": 0.0001748973106131822,
      "loss": 23.0,
      "step": 3864
    },
    {
      "epoch": 2.3199279711884753,
      "grad_norm": 0.0054776170291006565,
      "learning_rate": 0.00017488479201559627,
      "loss": 23.0,
      "step": 3865
    },
    {
      "epoch": 2.320528211284514,
      "grad_norm": 0.0020852498710155487,
      "learning_rate": 0.00017487227074556474,
      "loss": 23.0,
      "step": 3866
    },
    {
      "epoch": 2.3211284513805523,
      "grad_norm": 0.001896170899271965,
      "learning_rate": 0.00017485974680353446,
      "loss": 23.0,
      "step": 3867
    },
    {
      "epoch": 2.3217286914765904,
      "grad_norm": 0.0022559871431440115,
      "learning_rate": 0.00017484722018995235,
      "loss": 23.0,
      "step": 3868
    },
    {
      "epoch": 2.322328931572629,
      "grad_norm": 0.0008233110420405865,
      "learning_rate": 0.00017483469090526552,
      "loss": 23.0,
      "step": 3869
    },
    {
      "epoch": 2.3229291716686675,
      "grad_norm": 0.003072496736422181,
      "learning_rate": 0.00017482215894992106,
      "loss": 23.0,
      "step": 3870
    },
    {
      "epoch": 2.323529411764706,
      "grad_norm": 0.0034991884604096413,
      "learning_rate": 0.00017480962432436618,
      "loss": 23.0,
      "step": 3871
    },
    {
      "epoch": 2.324129651860744,
      "grad_norm": 0.002439623698592186,
      "learning_rate": 0.00017479708702904827,
      "loss": 23.0,
      "step": 3872
    },
    {
      "epoch": 2.3247298919567827,
      "grad_norm": 0.0018866292666643858,
      "learning_rate": 0.00017478454706441472,
      "loss": 23.0,
      "step": 3873
    },
    {
      "epoch": 2.3253301320528212,
      "grad_norm": 0.00198563514277339,
      "learning_rate": 0.00017477200443091306,
      "loss": 23.0,
      "step": 3874
    },
    {
      "epoch": 2.32593037214886,
      "grad_norm": 0.005812112707644701,
      "learning_rate": 0.0001747594591289909,
      "loss": 23.0,
      "step": 3875
    },
    {
      "epoch": 2.326530612244898,
      "grad_norm": 0.0013709774939343333,
      "learning_rate": 0.00017474691115909594,
      "loss": 23.0,
      "step": 3876
    },
    {
      "epoch": 2.3271308523409364,
      "grad_norm": 0.0018185654189437628,
      "learning_rate": 0.00017473436052167597,
      "loss": 23.0,
      "step": 3877
    },
    {
      "epoch": 2.327731092436975,
      "grad_norm": 0.002282187808305025,
      "learning_rate": 0.00017472180721717895,
      "loss": 23.0,
      "step": 3878
    },
    {
      "epoch": 2.328331332533013,
      "grad_norm": 0.0013887138338759542,
      "learning_rate": 0.00017470925124605282,
      "loss": 23.0,
      "step": 3879
    },
    {
      "epoch": 2.3289315726290516,
      "grad_norm": 0.00287288217805326,
      "learning_rate": 0.0001746966926087457,
      "loss": 23.0,
      "step": 3880
    },
    {
      "epoch": 2.32953181272509,
      "grad_norm": 0.0033778510987758636,
      "learning_rate": 0.00017468413130570577,
      "loss": 23.0,
      "step": 3881
    },
    {
      "epoch": 2.3301320528211287,
      "grad_norm": 0.0016036947490647435,
      "learning_rate": 0.00017467156733738128,
      "loss": 23.0,
      "step": 3882
    },
    {
      "epoch": 2.3307322929171668,
      "grad_norm": 0.0013769504148513079,
      "learning_rate": 0.00017465900070422064,
      "loss": 23.0,
      "step": 3883
    },
    {
      "epoch": 2.3313325330132053,
      "grad_norm": 0.00230945716612041,
      "learning_rate": 0.0001746464314066723,
      "loss": 23.0,
      "step": 3884
    },
    {
      "epoch": 2.331932773109244,
      "grad_norm": 0.0031504861544817686,
      "learning_rate": 0.00017463385944518485,
      "loss": 23.0,
      "step": 3885
    },
    {
      "epoch": 2.332533013205282,
      "grad_norm": 0.0008833253523334861,
      "learning_rate": 0.00017462128482020694,
      "loss": 23.0,
      "step": 3886
    },
    {
      "epoch": 2.3331332533013205,
      "grad_norm": 0.002419440308585763,
      "learning_rate": 0.00017460870753218733,
      "loss": 23.0,
      "step": 3887
    },
    {
      "epoch": 2.333733493397359,
      "grad_norm": 0.00045076129026710987,
      "learning_rate": 0.00017459612758157484,
      "loss": 23.0,
      "step": 3888
    },
    {
      "epoch": 2.3343337334933976,
      "grad_norm": 0.0007832158589735627,
      "learning_rate": 0.00017458354496881846,
      "loss": 23.0,
      "step": 3889
    },
    {
      "epoch": 2.3349339735894357,
      "grad_norm": 0.0010271031642332673,
      "learning_rate": 0.00017457095969436722,
      "loss": 23.0,
      "step": 3890
    },
    {
      "epoch": 2.335534213685474,
      "grad_norm": 0.002968541579321027,
      "learning_rate": 0.0001745583717586702,
      "loss": 23.0,
      "step": 3891
    },
    {
      "epoch": 2.3361344537815127,
      "grad_norm": 0.005150226876139641,
      "learning_rate": 0.00017454578116217674,
      "loss": 23.0,
      "step": 3892
    },
    {
      "epoch": 2.336734693877551,
      "grad_norm": 0.0014614604879170656,
      "learning_rate": 0.00017453318790533606,
      "loss": 23.0,
      "step": 3893
    },
    {
      "epoch": 2.3373349339735894,
      "grad_norm": 0.003210582537576556,
      "learning_rate": 0.00017452059198859766,
      "loss": 23.0,
      "step": 3894
    },
    {
      "epoch": 2.337935174069628,
      "grad_norm": 0.0043360283598303795,
      "learning_rate": 0.000174507993412411,
      "loss": 23.0,
      "step": 3895
    },
    {
      "epoch": 2.3385354141656665,
      "grad_norm": 0.0008345422684215009,
      "learning_rate": 0.00017449539217722571,
      "loss": 23.0,
      "step": 3896
    },
    {
      "epoch": 2.3391356542617046,
      "grad_norm": 0.003543426748365164,
      "learning_rate": 0.00017448278828349152,
      "loss": 23.0,
      "step": 3897
    },
    {
      "epoch": 2.339735894357743,
      "grad_norm": 0.0006024944595992565,
      "learning_rate": 0.00017447018173165818,
      "loss": 23.0,
      "step": 3898
    },
    {
      "epoch": 2.3403361344537816,
      "grad_norm": 0.002525531454011798,
      "learning_rate": 0.00017445757252217562,
      "loss": 23.0,
      "step": 3899
    },
    {
      "epoch": 2.3409363745498197,
      "grad_norm": 0.0018678716151043773,
      "learning_rate": 0.00017444496065549385,
      "loss": 23.0,
      "step": 3900
    },
    {
      "epoch": 2.3415366146458583,
      "grad_norm": 0.002660698490217328,
      "learning_rate": 0.00017443234613206293,
      "loss": 23.0,
      "step": 3901
    },
    {
      "epoch": 2.342136854741897,
      "grad_norm": 0.0027779710944741964,
      "learning_rate": 0.000174419728952333,
      "loss": 23.0,
      "step": 3902
    },
    {
      "epoch": 2.3427370948379354,
      "grad_norm": 0.0036743993405252695,
      "learning_rate": 0.0001744071091167544,
      "loss": 23.0,
      "step": 3903
    },
    {
      "epoch": 2.3433373349339734,
      "grad_norm": 0.002380775986239314,
      "learning_rate": 0.00017439448662577749,
      "loss": 23.0,
      "step": 3904
    },
    {
      "epoch": 2.343937575030012,
      "grad_norm": 0.0030045825988054276,
      "learning_rate": 0.0001743818614798527,
      "loss": 23.0,
      "step": 3905
    },
    {
      "epoch": 2.3445378151260505,
      "grad_norm": 0.0010804989142343402,
      "learning_rate": 0.00017436923367943058,
      "loss": 23.0,
      "step": 3906
    },
    {
      "epoch": 2.3451380552220886,
      "grad_norm": 0.0031437580473721027,
      "learning_rate": 0.00017435660322496187,
      "loss": 23.0,
      "step": 3907
    },
    {
      "epoch": 2.345738295318127,
      "grad_norm": 0.002179528819397092,
      "learning_rate": 0.0001743439701168972,
      "loss": 23.0,
      "step": 3908
    },
    {
      "epoch": 2.3463385354141657,
      "grad_norm": 0.0047058421187102795,
      "learning_rate": 0.0001743313343556875,
      "loss": 23.0,
      "step": 3909
    },
    {
      "epoch": 2.3469387755102042,
      "grad_norm": 0.0017768297111615539,
      "learning_rate": 0.00017431869594178366,
      "loss": 23.0,
      "step": 3910
    },
    {
      "epoch": 2.3475390156062423,
      "grad_norm": 0.0016148180002346635,
      "learning_rate": 0.00017430605487563677,
      "loss": 23.0,
      "step": 3911
    },
    {
      "epoch": 2.348139255702281,
      "grad_norm": 0.001844774466007948,
      "learning_rate": 0.00017429341115769787,
      "loss": 23.0,
      "step": 3912
    },
    {
      "epoch": 2.3487394957983194,
      "grad_norm": 0.001695389044471085,
      "learning_rate": 0.0001742807647884183,
      "loss": 23.0,
      "step": 3913
    },
    {
      "epoch": 2.3493397358943575,
      "grad_norm": 0.0021679711062461138,
      "learning_rate": 0.00017426811576824923,
      "loss": 23.0,
      "step": 3914
    },
    {
      "epoch": 2.349939975990396,
      "grad_norm": 0.0033461309503763914,
      "learning_rate": 0.00017425546409764218,
      "loss": 23.0,
      "step": 3915
    },
    {
      "epoch": 2.3505402160864346,
      "grad_norm": 0.0019258579704910517,
      "learning_rate": 0.00017424280977704863,
      "loss": 23.0,
      "step": 3916
    },
    {
      "epoch": 2.351140456182473,
      "grad_norm": 0.0012793492060154676,
      "learning_rate": 0.0001742301528069202,
      "loss": 23.0,
      "step": 3917
    },
    {
      "epoch": 2.3517406962785112,
      "grad_norm": 0.0014299890026450157,
      "learning_rate": 0.00017421749318770853,
      "loss": 23.0,
      "step": 3918
    },
    {
      "epoch": 2.3523409363745498,
      "grad_norm": 0.0012088280636817217,
      "learning_rate": 0.00017420483091986543,
      "loss": 23.0,
      "step": 3919
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 0.0010071590077131987,
      "learning_rate": 0.0001741921660038428,
      "loss": 23.0,
      "step": 3920
    },
    {
      "epoch": 2.3535414165666264,
      "grad_norm": 0.0018569778185337782,
      "learning_rate": 0.0001741794984400926,
      "loss": 23.0,
      "step": 3921
    },
    {
      "epoch": 2.354141656662665,
      "grad_norm": 0.0018122204346582294,
      "learning_rate": 0.00017416682822906694,
      "loss": 23.0,
      "step": 3922
    },
    {
      "epoch": 2.3547418967587035,
      "grad_norm": 0.0012169646797701716,
      "learning_rate": 0.00017415415537121797,
      "loss": 23.0,
      "step": 3923
    },
    {
      "epoch": 2.355342136854742,
      "grad_norm": 0.0032757974695414305,
      "learning_rate": 0.00017414147986699792,
      "loss": 23.0,
      "step": 3924
    },
    {
      "epoch": 2.35594237695078,
      "grad_norm": 0.0015983044868335128,
      "learning_rate": 0.00017412880171685914,
      "loss": 23.0,
      "step": 3925
    },
    {
      "epoch": 2.3565426170468187,
      "grad_norm": 0.000568926683627069,
      "learning_rate": 0.00017411612092125417,
      "loss": 23.0,
      "step": 3926
    },
    {
      "epoch": 2.357142857142857,
      "grad_norm": 0.0018514172406867146,
      "learning_rate": 0.00017410343748063545,
      "loss": 23.0,
      "step": 3927
    },
    {
      "epoch": 2.3577430972388957,
      "grad_norm": 0.0015337765216827393,
      "learning_rate": 0.00017409075139545568,
      "loss": 23.0,
      "step": 3928
    },
    {
      "epoch": 2.358343337334934,
      "grad_norm": 0.0009689840371720493,
      "learning_rate": 0.00017407806266616754,
      "loss": 23.0,
      "step": 3929
    },
    {
      "epoch": 2.3589435774309724,
      "grad_norm": 0.002194127067923546,
      "learning_rate": 0.00017406537129322394,
      "loss": 23.0,
      "step": 3930
    },
    {
      "epoch": 2.359543817527011,
      "grad_norm": 0.0015414162771776319,
      "learning_rate": 0.00017405267727707774,
      "loss": 23.0,
      "step": 3931
    },
    {
      "epoch": 2.3601440576230495,
      "grad_norm": 0.00167924037668854,
      "learning_rate": 0.000174039980618182,
      "loss": 23.0,
      "step": 3932
    },
    {
      "epoch": 2.3607442977190876,
      "grad_norm": 0.001501681748777628,
      "learning_rate": 0.00017402728131698977,
      "loss": 23.0,
      "step": 3933
    },
    {
      "epoch": 2.361344537815126,
      "grad_norm": 0.0010487270774319768,
      "learning_rate": 0.0001740145793739543,
      "loss": 23.0,
      "step": 3934
    },
    {
      "epoch": 2.3619447779111646,
      "grad_norm": 0.0017754529835656285,
      "learning_rate": 0.0001740018747895289,
      "loss": 23.0,
      "step": 3935
    },
    {
      "epoch": 2.3625450180072027,
      "grad_norm": 0.00277283089235425,
      "learning_rate": 0.0001739891675641669,
      "loss": 23.0,
      "step": 3936
    },
    {
      "epoch": 2.3631452581032413,
      "grad_norm": 0.0016696704551577568,
      "learning_rate": 0.00017397645769832187,
      "loss": 23.0,
      "step": 3937
    },
    {
      "epoch": 2.36374549819928,
      "grad_norm": 0.0018485120963305235,
      "learning_rate": 0.00017396374519244737,
      "loss": 23.0,
      "step": 3938
    },
    {
      "epoch": 2.3643457382953184,
      "grad_norm": 0.0054781329818069935,
      "learning_rate": 0.000173951030046997,
      "loss": 23.0,
      "step": 3939
    },
    {
      "epoch": 2.3649459783913565,
      "grad_norm": 0.0019205682910978794,
      "learning_rate": 0.00017393831226242466,
      "loss": 23.0,
      "step": 3940
    },
    {
      "epoch": 2.365546218487395,
      "grad_norm": 0.003620882984250784,
      "learning_rate": 0.0001739255918391841,
      "loss": 23.0,
      "step": 3941
    },
    {
      "epoch": 2.3661464585834335,
      "grad_norm": 0.0012869867496192455,
      "learning_rate": 0.00017391286877772933,
      "loss": 23.0,
      "step": 3942
    },
    {
      "epoch": 2.3667466986794716,
      "grad_norm": 0.002010358963161707,
      "learning_rate": 0.00017390014307851442,
      "loss": 23.0,
      "step": 3943
    },
    {
      "epoch": 2.36734693877551,
      "grad_norm": 0.0021779662929475307,
      "learning_rate": 0.0001738874147419935,
      "loss": 23.0,
      "step": 3944
    },
    {
      "epoch": 2.3679471788715487,
      "grad_norm": 0.0021605216898024082,
      "learning_rate": 0.0001738746837686208,
      "loss": 23.0,
      "step": 3945
    },
    {
      "epoch": 2.3685474189675872,
      "grad_norm": 0.0006803186261095107,
      "learning_rate": 0.00017386195015885064,
      "loss": 23.0,
      "step": 3946
    },
    {
      "epoch": 2.3691476590636253,
      "grad_norm": 0.0011268210364505649,
      "learning_rate": 0.0001738492139131375,
      "loss": 23.0,
      "step": 3947
    },
    {
      "epoch": 2.369747899159664,
      "grad_norm": 0.0007069764542393386,
      "learning_rate": 0.0001738364750319359,
      "loss": 23.0,
      "step": 3948
    },
    {
      "epoch": 2.3703481392557024,
      "grad_norm": 0.007430940866470337,
      "learning_rate": 0.0001738237335157004,
      "loss": 23.0,
      "step": 3949
    },
    {
      "epoch": 2.3709483793517405,
      "grad_norm": 0.003919681068509817,
      "learning_rate": 0.00017381098936488574,
      "loss": 23.0,
      "step": 3950
    },
    {
      "epoch": 2.371548619447779,
      "grad_norm": 0.0008578869164921343,
      "learning_rate": 0.0001737982425799468,
      "loss": 23.0,
      "step": 3951
    },
    {
      "epoch": 2.3721488595438176,
      "grad_norm": 0.0018030635546892881,
      "learning_rate": 0.00017378549316133835,
      "loss": 23.0,
      "step": 3952
    },
    {
      "epoch": 2.372749099639856,
      "grad_norm": 0.0018941910238936543,
      "learning_rate": 0.00017377274110951543,
      "loss": 23.0,
      "step": 3953
    },
    {
      "epoch": 2.3733493397358942,
      "grad_norm": 0.0007153527112677693,
      "learning_rate": 0.00017375998642493322,
      "loss": 23.0,
      "step": 3954
    },
    {
      "epoch": 2.3739495798319328,
      "grad_norm": 0.0009323756094090641,
      "learning_rate": 0.00017374722910804677,
      "loss": 23.0,
      "step": 3955
    },
    {
      "epoch": 2.3745498199279713,
      "grad_norm": 0.0036982784513384104,
      "learning_rate": 0.00017373446915931143,
      "loss": 23.0,
      "step": 3956
    },
    {
      "epoch": 2.3751500600240094,
      "grad_norm": 0.0014185566687956452,
      "learning_rate": 0.00017372170657918256,
      "loss": 23.0,
      "step": 3957
    },
    {
      "epoch": 2.375750300120048,
      "grad_norm": 0.002080840989947319,
      "learning_rate": 0.00017370894136811563,
      "loss": 23.0,
      "step": 3958
    },
    {
      "epoch": 2.3763505402160865,
      "grad_norm": 0.0006996280862949789,
      "learning_rate": 0.00017369617352656618,
      "loss": 23.0,
      "step": 3959
    },
    {
      "epoch": 2.376950780312125,
      "grad_norm": 0.0018821045523509383,
      "learning_rate": 0.00017368340305498985,
      "loss": 23.0,
      "step": 3960
    },
    {
      "epoch": 2.377551020408163,
      "grad_norm": 0.0013187207514420152,
      "learning_rate": 0.0001736706299538424,
      "loss": 23.0,
      "step": 3961
    },
    {
      "epoch": 2.3781512605042017,
      "grad_norm": 0.001449449686333537,
      "learning_rate": 0.0001736578542235797,
      "loss": 23.0,
      "step": 3962
    },
    {
      "epoch": 2.37875150060024,
      "grad_norm": 0.002266083611175418,
      "learning_rate": 0.00017364507586465762,
      "loss": 23.0,
      "step": 3963
    },
    {
      "epoch": 2.3793517406962783,
      "grad_norm": 0.002582366345450282,
      "learning_rate": 0.00017363229487753223,
      "loss": 23.0,
      "step": 3964
    },
    {
      "epoch": 2.379951980792317,
      "grad_norm": 0.0003937427536584437,
      "learning_rate": 0.00017361951126265964,
      "loss": 23.0,
      "step": 3965
    },
    {
      "epoch": 2.3805522208883554,
      "grad_norm": 0.0017639965517446399,
      "learning_rate": 0.00017360672502049606,
      "loss": 23.0,
      "step": 3966
    },
    {
      "epoch": 2.381152460984394,
      "grad_norm": 0.004503778647631407,
      "learning_rate": 0.0001735939361514978,
      "loss": 23.0,
      "step": 3967
    },
    {
      "epoch": 2.381752701080432,
      "grad_norm": 0.0036358425859361887,
      "learning_rate": 0.0001735811446561213,
      "loss": 23.0,
      "step": 3968
    },
    {
      "epoch": 2.3823529411764706,
      "grad_norm": 0.0023689253721386194,
      "learning_rate": 0.00017356835053482299,
      "loss": 23.0,
      "step": 3969
    },
    {
      "epoch": 2.382953181272509,
      "grad_norm": 0.002466338686645031,
      "learning_rate": 0.0001735555537880595,
      "loss": 23.0,
      "step": 3970
    },
    {
      "epoch": 2.383553421368547,
      "grad_norm": 0.0021039212588220835,
      "learning_rate": 0.0001735427544162875,
      "loss": 23.0,
      "step": 3971
    },
    {
      "epoch": 2.3841536614645857,
      "grad_norm": 0.0027853043284267187,
      "learning_rate": 0.00017352995241996374,
      "loss": 23.0,
      "step": 3972
    },
    {
      "epoch": 2.3847539015606243,
      "grad_norm": 0.0033636244479566813,
      "learning_rate": 0.00017351714779954516,
      "loss": 23.0,
      "step": 3973
    },
    {
      "epoch": 2.385354141656663,
      "grad_norm": 0.002105874475091696,
      "learning_rate": 0.00017350434055548867,
      "loss": 23.0,
      "step": 3974
    },
    {
      "epoch": 2.385954381752701,
      "grad_norm": 0.0009028427302837372,
      "learning_rate": 0.00017349153068825133,
      "loss": 23.0,
      "step": 3975
    },
    {
      "epoch": 2.3865546218487395,
      "grad_norm": 0.0030058857519179583,
      "learning_rate": 0.00017347871819829033,
      "loss": 23.0,
      "step": 3976
    },
    {
      "epoch": 2.387154861944778,
      "grad_norm": 0.0017001151572912931,
      "learning_rate": 0.00017346590308606286,
      "loss": 23.0,
      "step": 3977
    },
    {
      "epoch": 2.387755102040816,
      "grad_norm": 0.0008253180421888828,
      "learning_rate": 0.0001734530853520263,
      "loss": 23.0,
      "step": 3978
    },
    {
      "epoch": 2.3883553421368546,
      "grad_norm": 0.0019861324690282345,
      "learning_rate": 0.00017344026499663807,
      "loss": 23.0,
      "step": 3979
    },
    {
      "epoch": 2.388955582232893,
      "grad_norm": 0.002510624472051859,
      "learning_rate": 0.0001734274420203557,
      "loss": 23.0,
      "step": 3980
    },
    {
      "epoch": 2.3895558223289317,
      "grad_norm": 0.0011581958970054984,
      "learning_rate": 0.00017341461642363676,
      "loss": 23.0,
      "step": 3981
    },
    {
      "epoch": 2.39015606242497,
      "grad_norm": 0.002012391574680805,
      "learning_rate": 0.00017340178820693906,
      "loss": 23.0,
      "step": 3982
    },
    {
      "epoch": 2.3907563025210083,
      "grad_norm": 0.002154824323952198,
      "learning_rate": 0.0001733889573707203,
      "loss": 23.0,
      "step": 3983
    },
    {
      "epoch": 2.391356542617047,
      "grad_norm": 0.0009000212885439396,
      "learning_rate": 0.00017337612391543847,
      "loss": 23.0,
      "step": 3984
    },
    {
      "epoch": 2.3919567827130854,
      "grad_norm": 0.003189866431057453,
      "learning_rate": 0.0001733632878415515,
      "loss": 23.0,
      "step": 3985
    },
    {
      "epoch": 2.3925570228091235,
      "grad_norm": 0.0023710643872618675,
      "learning_rate": 0.00017335044914951753,
      "loss": 23.0,
      "step": 3986
    },
    {
      "epoch": 2.393157262905162,
      "grad_norm": 0.0008908773888833821,
      "learning_rate": 0.00017333760783979466,
      "loss": 23.0,
      "step": 3987
    },
    {
      "epoch": 2.3937575030012006,
      "grad_norm": 0.0027625104412436485,
      "learning_rate": 0.00017332476391284128,
      "loss": 23.0,
      "step": 3988
    },
    {
      "epoch": 2.394357743097239,
      "grad_norm": 0.002077084966003895,
      "learning_rate": 0.00017331191736911564,
      "loss": 23.0,
      "step": 3989
    },
    {
      "epoch": 2.3949579831932772,
      "grad_norm": 0.0020401370711624622,
      "learning_rate": 0.00017329906820907627,
      "loss": 23.0,
      "step": 3990
    },
    {
      "epoch": 2.395558223289316,
      "grad_norm": 0.00204063905403018,
      "learning_rate": 0.0001732862164331817,
      "loss": 23.0,
      "step": 3991
    },
    {
      "epoch": 2.3961584633853543,
      "grad_norm": 0.0018513710238039494,
      "learning_rate": 0.00017327336204189055,
      "loss": 23.0,
      "step": 3992
    },
    {
      "epoch": 2.3967587034813924,
      "grad_norm": 0.0022755179088562727,
      "learning_rate": 0.00017326050503566162,
      "loss": 23.0,
      "step": 3993
    },
    {
      "epoch": 2.397358943577431,
      "grad_norm": 0.0027569227386265993,
      "learning_rate": 0.00017324764541495373,
      "loss": 23.0,
      "step": 3994
    },
    {
      "epoch": 2.3979591836734695,
      "grad_norm": 0.0031000536400824785,
      "learning_rate": 0.00017323478318022578,
      "loss": 23.0,
      "step": 3995
    },
    {
      "epoch": 2.398559423769508,
      "grad_norm": 0.0016642792616039515,
      "learning_rate": 0.00017322191833193674,
      "loss": 23.0,
      "step": 3996
    },
    {
      "epoch": 2.399159663865546,
      "grad_norm": 0.0015426806639879942,
      "learning_rate": 0.00017320905087054585,
      "loss": 23.0,
      "step": 3997
    },
    {
      "epoch": 2.3997599039615847,
      "grad_norm": 0.0016276773530989885,
      "learning_rate": 0.00017319618079651226,
      "loss": 23.0,
      "step": 3998
    },
    {
      "epoch": 2.400360144057623,
      "grad_norm": 0.00199799588881433,
      "learning_rate": 0.0001731833081102952,
      "loss": 23.0,
      "step": 3999
    },
    {
      "epoch": 2.4009603841536613,
      "grad_norm": 0.001323958276771009,
      "learning_rate": 0.00017317043281235418,
      "loss": 23.0,
      "step": 4000
    },
    {
      "epoch": 2.4009603841536613,
      "eval_loss": 11.5,
      "eval_runtime": 5.4819,
      "eval_samples_per_second": 255.931,
      "eval_steps_per_second": 32.105,
      "step": 4000
    }
  ],
  "logging_steps": 1,
  "max_steps": 16660,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 1000,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 80021001510912.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}