{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2899,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00034494653328734045,
      "grad_norm": 0.6609077453613281,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 10.3896,
      "step": 1
    },
    {
      "epoch": 0.0006898930665746809,
      "grad_norm": 0.6776272654533386,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 10.3747,
      "step": 2
    },
    {
      "epoch": 0.0010348395998620215,
      "grad_norm": 0.7447559833526611,
      "learning_rate": 3e-06,
      "loss": 10.3959,
      "step": 3
    },
    {
      "epoch": 0.0013797861331493618,
      "grad_norm": 0.7730655074119568,
      "learning_rate": 4.000000000000001e-06,
      "loss": 10.39,
      "step": 4
    },
    {
      "epoch": 0.0017247326664367024,
      "grad_norm": 0.7812376618385315,
      "learning_rate": 5e-06,
      "loss": 10.3765,
      "step": 5
    },
    {
      "epoch": 0.002069679199724043,
      "grad_norm": 0.854775607585907,
      "learning_rate": 6e-06,
      "loss": 10.3793,
      "step": 6
    },
    {
      "epoch": 0.002414625733011383,
      "grad_norm": 0.7408090233802795,
      "learning_rate": 7.000000000000001e-06,
      "loss": 10.3799,
      "step": 7
    },
    {
      "epoch": 0.0027595722662987236,
      "grad_norm": 0.7635711431503296,
      "learning_rate": 8.000000000000001e-06,
      "loss": 10.3856,
      "step": 8
    },
    {
      "epoch": 0.003104518799586064,
      "grad_norm": 0.7853497862815857,
      "learning_rate": 9e-06,
      "loss": 10.3786,
      "step": 9
    },
    {
      "epoch": 0.0034494653328734047,
      "grad_norm": 0.7585586905479431,
      "learning_rate": 1e-05,
      "loss": 10.3554,
      "step": 10
    },
    {
      "epoch": 0.0037944118661607453,
      "grad_norm": 0.8145893216133118,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 10.3792,
      "step": 11
    },
    {
      "epoch": 0.004139358399448086,
      "grad_norm": 0.8374782204627991,
      "learning_rate": 1.2e-05,
      "loss": 10.3807,
      "step": 12
    },
    {
      "epoch": 0.004484304932735426,
      "grad_norm": 0.7668830156326294,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 10.3674,
      "step": 13
    },
    {
      "epoch": 0.004829251466022766,
      "grad_norm": 0.7636863589286804,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 10.3887,
      "step": 14
    },
    {
      "epoch": 0.005174197999310107,
      "grad_norm": 0.8072705268859863,
      "learning_rate": 1.5e-05,
      "loss": 10.3833,
      "step": 15
    },
    {
      "epoch": 0.005519144532597447,
      "grad_norm": 0.8346720933914185,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 10.3675,
      "step": 16
    },
    {
      "epoch": 0.005864091065884788,
      "grad_norm": 0.764621376991272,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 10.3764,
      "step": 17
    },
    {
      "epoch": 0.006209037599172128,
      "grad_norm": 0.8266566395759583,
      "learning_rate": 1.8e-05,
      "loss": 10.3855,
      "step": 18
    },
    {
      "epoch": 0.006553984132459468,
      "grad_norm": 0.7850378751754761,
      "learning_rate": 1.9e-05,
      "loss": 10.3738,
      "step": 19
    },
    {
      "epoch": 0.006898930665746809,
      "grad_norm": 0.7893702387809753,
      "learning_rate": 2e-05,
      "loss": 10.3853,
      "step": 20
    },
    {
      "epoch": 0.0072438771990341495,
      "grad_norm": 0.8223876953125,
      "learning_rate": 2.1e-05,
      "loss": 10.3786,
      "step": 21
    },
    {
      "epoch": 0.0075888237323214905,
      "grad_norm": 0.8152451515197754,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 10.3782,
      "step": 22
    },
    {
      "epoch": 0.00793377026560883,
      "grad_norm": 0.8822249174118042,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 10.3668,
      "step": 23
    },
    {
      "epoch": 0.008278716798896172,
      "grad_norm": 1.0197700262069702,
      "learning_rate": 2.4e-05,
      "loss": 10.3814,
      "step": 24
    },
    {
      "epoch": 0.008623663332183512,
      "grad_norm": 0.8479687571525574,
      "learning_rate": 2.5e-05,
      "loss": 10.3559,
      "step": 25
    },
    {
      "epoch": 0.008968609865470852,
      "grad_norm": 0.8130446672439575,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 10.385,
      "step": 26
    },
    {
      "epoch": 0.009313556398758192,
      "grad_norm": 0.8617949485778809,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 10.3708,
      "step": 27
    },
    {
      "epoch": 0.009658502932045532,
      "grad_norm": 0.826323926448822,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 10.3719,
      "step": 28
    },
    {
      "epoch": 0.010003449465332874,
      "grad_norm": 0.9218830466270447,
      "learning_rate": 2.9e-05,
      "loss": 10.3632,
      "step": 29
    },
    {
      "epoch": 0.010348395998620214,
      "grad_norm": 0.8265900611877441,
      "learning_rate": 3e-05,
      "loss": 10.3511,
      "step": 30
    },
    {
      "epoch": 0.010693342531907554,
      "grad_norm": 0.869763970375061,
      "learning_rate": 3.1e-05,
      "loss": 10.3784,
      "step": 31
    },
    {
      "epoch": 0.011038289065194894,
      "grad_norm": 0.8748270273208618,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 10.3748,
      "step": 32
    },
    {
      "epoch": 0.011383235598482234,
      "grad_norm": 0.9415621757507324,
      "learning_rate": 3.3e-05,
      "loss": 10.3777,
      "step": 33
    },
    {
      "epoch": 0.011728182131769576,
      "grad_norm": 0.8795452117919922,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 10.3634,
      "step": 34
    },
    {
      "epoch": 0.012073128665056916,
      "grad_norm": 0.9584307074546814,
      "learning_rate": 3.5e-05,
      "loss": 10.3707,
      "step": 35
    },
    {
      "epoch": 0.012418075198344257,
      "grad_norm": 0.8180338740348816,
      "learning_rate": 3.6e-05,
      "loss": 10.3616,
      "step": 36
    },
    {
      "epoch": 0.012763021731631597,
      "grad_norm": 0.8298875093460083,
      "learning_rate": 3.7e-05,
      "loss": 10.3896,
      "step": 37
    },
    {
      "epoch": 0.013107968264918937,
      "grad_norm": 0.9823697209358215,
      "learning_rate": 3.8e-05,
      "loss": 10.3874,
      "step": 38
    },
    {
      "epoch": 0.013452914798206279,
      "grad_norm": 0.9253247976303101,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 10.3639,
      "step": 39
    },
    {
      "epoch": 0.013797861331493619,
      "grad_norm": 0.9450452327728271,
      "learning_rate": 4e-05,
      "loss": 10.3874,
      "step": 40
    },
    {
      "epoch": 0.014142807864780959,
      "grad_norm": 0.8728947639465332,
      "learning_rate": 4.1e-05,
      "loss": 10.3677,
      "step": 41
    },
    {
      "epoch": 0.014487754398068299,
      "grad_norm": 0.8550394177436829,
      "learning_rate": 4.2e-05,
      "loss": 10.3735,
      "step": 42
    },
    {
      "epoch": 0.01483270093135564,
      "grad_norm": 0.887096107006073,
      "learning_rate": 4.3e-05,
      "loss": 10.3676,
      "step": 43
    },
    {
      "epoch": 0.015177647464642981,
      "grad_norm": 1.0961228609085083,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 10.3681,
      "step": 44
    },
    {
      "epoch": 0.015522593997930321,
      "grad_norm": 0.9618821740150452,
      "learning_rate": 4.5e-05,
      "loss": 10.3789,
      "step": 45
    },
    {
      "epoch": 0.01586754053121766,
      "grad_norm": 0.9343708157539368,
      "learning_rate": 4.600000000000001e-05,
      "loss": 10.3825,
      "step": 46
    },
    {
      "epoch": 0.016212487064505003,
      "grad_norm": 0.9412921667098999,
      "learning_rate": 4.7e-05,
      "loss": 10.3817,
      "step": 47
    },
    {
      "epoch": 0.016557433597792343,
      "grad_norm": 1.0469144582748413,
      "learning_rate": 4.8e-05,
      "loss": 10.3506,
      "step": 48
    },
    {
      "epoch": 0.016902380131079683,
      "grad_norm": 1.068739891052246,
      "learning_rate": 4.9e-05,
      "loss": 10.3372,
      "step": 49
    },
    {
      "epoch": 0.017247326664367024,
      "grad_norm": 1.1528873443603516,
      "learning_rate": 5e-05,
      "loss": 10.3483,
      "step": 50
    },
    {
      "epoch": 0.017592273197654364,
      "grad_norm": 0.7057186961174011,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 10.3596,
      "step": 51
    },
    {
      "epoch": 0.017937219730941704,
      "grad_norm": 0.7338641285896301,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 10.3678,
      "step": 52
    },
    {
      "epoch": 0.018282166264229044,
      "grad_norm": 0.7243222594261169,
      "learning_rate": 5.300000000000001e-05,
      "loss": 10.3669,
      "step": 53
    },
    {
      "epoch": 0.018627112797516384,
      "grad_norm": 0.712495744228363,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 10.3632,
      "step": 54
    },
    {
      "epoch": 0.018972059330803724,
      "grad_norm": 0.7407525777816772,
      "learning_rate": 5.500000000000001e-05,
      "loss": 10.3715,
      "step": 55
    },
    {
      "epoch": 0.019317005864091064,
      "grad_norm": 0.7415438890457153,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 10.3801,
      "step": 56
    },
    {
      "epoch": 0.019661952397378408,
      "grad_norm": 0.8786624073982239,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 10.3781,
      "step": 57
    },
    {
      "epoch": 0.020006898930665748,
      "grad_norm": 0.7698060274124146,
      "learning_rate": 5.8e-05,
      "loss": 10.3719,
      "step": 58
    },
    {
      "epoch": 0.020351845463953088,
      "grad_norm": 0.7433401942253113,
      "learning_rate": 5.9e-05,
      "loss": 10.3633,
      "step": 59
    },
    {
      "epoch": 0.020696791997240428,
      "grad_norm": 0.8617119789123535,
      "learning_rate": 6e-05,
      "loss": 10.3708,
      "step": 60
    },
    {
      "epoch": 0.02104173853052777,
      "grad_norm": 0.9492126107215881,
      "learning_rate": 6.1e-05,
      "loss": 10.3524,
      "step": 61
    },
    {
      "epoch": 0.02138668506381511,
      "grad_norm": 0.8974522352218628,
      "learning_rate": 6.2e-05,
      "loss": 10.3708,
      "step": 62
    },
    {
      "epoch": 0.02173163159710245,
      "grad_norm": 0.7831613421440125,
      "learning_rate": 6.3e-05,
      "loss": 10.3636,
      "step": 63
    },
    {
      "epoch": 0.02207657813038979,
      "grad_norm": 0.8832221031188965,
      "learning_rate": 6.400000000000001e-05,
      "loss": 10.354,
      "step": 64
    },
    {
      "epoch": 0.02242152466367713,
      "grad_norm": 0.8677083253860474,
      "learning_rate": 6.500000000000001e-05,
      "loss": 10.3617,
      "step": 65
    },
    {
      "epoch": 0.02276647119696447,
      "grad_norm": 0.934693455696106,
      "learning_rate": 6.6e-05,
      "loss": 10.3556,
      "step": 66
    },
    {
      "epoch": 0.023111417730251813,
      "grad_norm": 0.9830888509750366,
      "learning_rate": 6.7e-05,
      "loss": 10.3612,
      "step": 67
    },
    {
      "epoch": 0.023456364263539153,
      "grad_norm": 1.0390836000442505,
      "learning_rate": 6.800000000000001e-05,
      "loss": 10.3761,
      "step": 68
    },
    {
      "epoch": 0.023801310796826493,
      "grad_norm": 0.9852373600006104,
      "learning_rate": 6.9e-05,
      "loss": 10.3501,
      "step": 69
    },
    {
      "epoch": 0.024146257330113833,
      "grad_norm": 0.8551622629165649,
      "learning_rate": 7e-05,
      "loss": 10.3612,
      "step": 70
    },
    {
      "epoch": 0.024491203863401173,
      "grad_norm": 0.9254797697067261,
      "learning_rate": 7.1e-05,
      "loss": 10.359,
      "step": 71
    },
    {
      "epoch": 0.024836150396688513,
      "grad_norm": 1.1132023334503174,
      "learning_rate": 7.2e-05,
      "loss": 10.3435,
      "step": 72
    },
    {
      "epoch": 0.025181096929975853,
      "grad_norm": 1.17877995967865,
      "learning_rate": 7.3e-05,
      "loss": 10.353,
      "step": 73
    },
    {
      "epoch": 0.025526043463263193,
      "grad_norm": 0.9523524641990662,
      "learning_rate": 7.4e-05,
      "loss": 10.36,
      "step": 74
    },
    {
      "epoch": 0.025870989996550534,
      "grad_norm": 0.991371214389801,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.3416,
      "step": 75
    },
    {
      "epoch": 0.026215936529837874,
      "grad_norm": 1.0377520322799683,
      "learning_rate": 7.6e-05,
      "loss": 10.3669,
      "step": 76
    },
    {
      "epoch": 0.026560883063125217,
      "grad_norm": 1.0984565019607544,
      "learning_rate": 7.7e-05,
      "loss": 10.322,
      "step": 77
    },
    {
      "epoch": 0.026905829596412557,
      "grad_norm": 1.147340178489685,
      "learning_rate": 7.800000000000001e-05,
      "loss": 10.3441,
      "step": 78
    },
    {
      "epoch": 0.027250776129699898,
      "grad_norm": 1.057822585105896,
      "learning_rate": 7.900000000000001e-05,
      "loss": 10.3325,
      "step": 79
    },
    {
      "epoch": 0.027595722662987238,
      "grad_norm": 1.1691458225250244,
      "learning_rate": 8e-05,
      "loss": 10.3458,
      "step": 80
    },
    {
      "epoch": 0.027940669196274578,
      "grad_norm": 1.2596869468688965,
      "learning_rate": 8.1e-05,
      "loss": 10.3158,
      "step": 81
    },
    {
      "epoch": 0.028285615729561918,
      "grad_norm": 1.1883245706558228,
      "learning_rate": 8.2e-05,
      "loss": 10.3267,
      "step": 82
    },
    {
      "epoch": 0.028630562262849258,
      "grad_norm": 1.2984602451324463,
      "learning_rate": 8.3e-05,
      "loss": 10.3389,
      "step": 83
    },
    {
      "epoch": 0.028975508796136598,
      "grad_norm": 1.2408357858657837,
      "learning_rate": 8.4e-05,
      "loss": 10.3261,
      "step": 84
    },
    {
      "epoch": 0.029320455329423938,
      "grad_norm": 1.3244764804840088,
      "learning_rate": 8.5e-05,
      "loss": 10.3211,
      "step": 85
    },
    {
      "epoch": 0.02966540186271128,
      "grad_norm": 1.4943519830703735,
      "learning_rate": 8.6e-05,
      "loss": 10.3072,
      "step": 86
    },
    {
      "epoch": 0.03001034839599862,
      "grad_norm": 1.3726887702941895,
      "learning_rate": 8.7e-05,
      "loss": 10.2841,
      "step": 87
    },
    {
      "epoch": 0.030355294929285962,
      "grad_norm": 1.5504422187805176,
      "learning_rate": 8.800000000000001e-05,
      "loss": 10.2865,
      "step": 88
    },
    {
      "epoch": 0.030700241462573302,
      "grad_norm": 1.499110221862793,
      "learning_rate": 8.900000000000001e-05,
      "loss": 10.2938,
      "step": 89
    },
    {
      "epoch": 0.031045187995860642,
      "grad_norm": 1.3083536624908447,
      "learning_rate": 9e-05,
      "loss": 10.2814,
      "step": 90
    },
    {
      "epoch": 0.03139013452914798,
      "grad_norm": 1.3920077085494995,
      "learning_rate": 9.1e-05,
      "loss": 10.2902,
      "step": 91
    },
    {
      "epoch": 0.03173508106243532,
      "grad_norm": 1.4502214193344116,
      "learning_rate": 9.200000000000001e-05,
      "loss": 10.2841,
      "step": 92
    },
    {
      "epoch": 0.03208002759572266,
      "grad_norm": 1.503482699394226,
      "learning_rate": 9.300000000000001e-05,
      "loss": 10.2558,
      "step": 93
    },
    {
      "epoch": 0.032424974129010006,
      "grad_norm": 1.4924733638763428,
      "learning_rate": 9.4e-05,
      "loss": 10.2587,
      "step": 94
    },
    {
      "epoch": 0.03276992066229734,
      "grad_norm": 1.4583185911178589,
      "learning_rate": 9.5e-05,
      "loss": 10.2477,
      "step": 95
    },
    {
      "epoch": 0.03311486719558469,
      "grad_norm": 1.5197653770446777,
      "learning_rate": 9.6e-05,
      "loss": 10.236,
      "step": 96
    },
    {
      "epoch": 0.03345981372887202,
      "grad_norm": 1.716496467590332,
      "learning_rate": 9.7e-05,
      "loss": 10.2214,
      "step": 97
    },
    {
      "epoch": 0.03380476026215937,
      "grad_norm": 1.5878716707229614,
      "learning_rate": 9.8e-05,
      "loss": 10.2318,
      "step": 98
    },
    {
      "epoch": 0.0341497067954467,
      "grad_norm": 1.5571258068084717,
      "learning_rate": 9.900000000000001e-05,
      "loss": 10.2292,
      "step": 99
    },
    {
      "epoch": 0.03449465332873405,
      "grad_norm": 1.6506520509719849,
      "learning_rate": 0.0001,
      "loss": 10.2108,
      "step": 100
    },
    {
      "epoch": 0.034839599862021384,
      "grad_norm": 1.0482630729675293,
      "learning_rate": 9.999996850555848e-05,
      "loss": 10.2815,
      "step": 101
    },
    {
      "epoch": 0.03518454639530873,
      "grad_norm": 1.0001797676086426,
      "learning_rate": 9.999987402227361e-05,
      "loss": 10.2609,
      "step": 102
    },
    {
      "epoch": 0.03552949292859607,
      "grad_norm": 1.1609498262405396,
      "learning_rate": 9.99997165502644e-05,
      "loss": 10.2727,
      "step": 103
    },
    {
      "epoch": 0.03587443946188341,
      "grad_norm": 1.1335071325302124,
      "learning_rate": 9.999949608972922e-05,
      "loss": 10.2701,
      "step": 104
    },
    {
      "epoch": 0.03621938599517075,
      "grad_norm": 1.193150520324707,
      "learning_rate": 9.999921264094586e-05,
      "loss": 10.2708,
      "step": 105
    },
    {
      "epoch": 0.03656433252845809,
      "grad_norm": 1.1749833822250366,
      "learning_rate": 9.999886620427133e-05,
      "loss": 10.2451,
      "step": 106
    },
    {
      "epoch": 0.03690927906174543,
      "grad_norm": 1.1283924579620361,
      "learning_rate": 9.99984567801421e-05,
      "loss": 10.2485,
      "step": 107
    },
    {
      "epoch": 0.03725422559503277,
      "grad_norm": 1.1589891910552979,
      "learning_rate": 9.999798436907395e-05,
      "loss": 10.2481,
      "step": 108
    },
    {
      "epoch": 0.03759917212832011,
      "grad_norm": 1.15898597240448,
      "learning_rate": 9.999744897166201e-05,
      "loss": 10.2464,
      "step": 109
    },
    {
      "epoch": 0.03794411866160745,
      "grad_norm": 1.2533750534057617,
      "learning_rate": 9.999685058858075e-05,
      "loss": 10.2044,
      "step": 110
    },
    {
      "epoch": 0.03828906519489479,
      "grad_norm": 1.2371543645858765,
      "learning_rate": 9.999618922058402e-05,
      "loss": 10.2436,
      "step": 111
    },
    {
      "epoch": 0.03863401172818213,
      "grad_norm": 1.1409327983856201,
      "learning_rate": 9.9995464868505e-05,
      "loss": 10.2129,
      "step": 112
    },
    {
      "epoch": 0.03897895826146947,
      "grad_norm": 1.3030234575271606,
      "learning_rate": 9.99946775332562e-05,
      "loss": 10.2095,
      "step": 113
    },
    {
      "epoch": 0.039323904794756816,
      "grad_norm": 1.194814682006836,
      "learning_rate": 9.999382721582948e-05,
      "loss": 10.2059,
      "step": 114
    },
    {
      "epoch": 0.03966885132804415,
      "grad_norm": 1.40117347240448,
      "learning_rate": 9.999291391729606e-05,
      "loss": 10.1761,
      "step": 115
    },
    {
      "epoch": 0.040013797861331496,
      "grad_norm": 1.1784946918487549,
      "learning_rate": 9.99919376388065e-05,
      "loss": 10.2012,
      "step": 116
    },
    {
      "epoch": 0.04035874439461883,
      "grad_norm": 1.2370185852050781,
      "learning_rate": 9.999089838159066e-05,
      "loss": 10.1933,
      "step": 117
    },
    {
      "epoch": 0.040703690927906176,
      "grad_norm": 1.2744057178497314,
      "learning_rate": 9.998979614695783e-05,
      "loss": 10.1672,
      "step": 118
    },
    {
      "epoch": 0.04104863746119351,
      "grad_norm": 1.2865217924118042,
      "learning_rate": 9.998863093629654e-05,
      "loss": 10.1898,
      "step": 119
    },
    {
      "epoch": 0.041393583994480856,
      "grad_norm": 1.2423728704452515,
      "learning_rate": 9.99874027510747e-05,
      "loss": 10.1716,
      "step": 120
    },
    {
      "epoch": 0.04173853052776819,
      "grad_norm": 1.341609001159668,
      "learning_rate": 9.998611159283954e-05,
      "loss": 10.1581,
      "step": 121
    },
    {
      "epoch": 0.04208347706105554,
      "grad_norm": 1.2851861715316772,
      "learning_rate": 9.998475746321769e-05,
      "loss": 10.1906,
      "step": 122
    },
    {
      "epoch": 0.04242842359434288,
      "grad_norm": 1.1925376653671265,
      "learning_rate": 9.998334036391498e-05,
      "loss": 10.1961,
      "step": 123
    },
    {
      "epoch": 0.04277337012763022,
      "grad_norm": 1.3623324632644653,
      "learning_rate": 9.99818602967167e-05,
      "loss": 10.1642,
      "step": 124
    },
    {
      "epoch": 0.04311831666091756,
      "grad_norm": 1.184838056564331,
      "learning_rate": 9.998031726348736e-05,
      "loss": 10.1793,
      "step": 125
    },
    {
      "epoch": 0.0434632631942049,
      "grad_norm": 1.369837760925293,
      "learning_rate": 9.997871126617085e-05,
      "loss": 10.1514,
      "step": 126
    },
    {
      "epoch": 0.04380820972749224,
      "grad_norm": 1.281457543373108,
      "learning_rate": 9.997704230679036e-05,
      "loss": 10.163,
      "step": 127
    },
    {
      "epoch": 0.04415315626077958,
      "grad_norm": 1.3965054750442505,
      "learning_rate": 9.997531038744845e-05,
      "loss": 10.1429,
      "step": 128
    },
    {
      "epoch": 0.04449810279406692,
      "grad_norm": 1.2722357511520386,
      "learning_rate": 9.997351551032692e-05,
      "loss": 10.138,
      "step": 129
    },
    {
      "epoch": 0.04484304932735426,
      "grad_norm": 1.3136990070343018,
      "learning_rate": 9.997165767768692e-05,
      "loss": 10.1314,
      "step": 130
    },
    {
      "epoch": 0.0451879958606416,
      "grad_norm": 1.252399206161499,
      "learning_rate": 9.99697368918689e-05,
      "loss": 10.1475,
      "step": 131
    },
    {
      "epoch": 0.04553294239392894,
      "grad_norm": 1.3466100692749023,
      "learning_rate": 9.996775315529265e-05,
      "loss": 10.1451,
      "step": 132
    },
    {
      "epoch": 0.04587788892721628,
      "grad_norm": 1.4946643114089966,
      "learning_rate": 9.99657064704572e-05,
      "loss": 10.12,
      "step": 133
    },
    {
      "epoch": 0.046222835460503625,
      "grad_norm": 1.42206609249115,
      "learning_rate": 9.996359683994096e-05,
      "loss": 10.1132,
      "step": 134
    },
    {
      "epoch": 0.04656778199379096,
      "grad_norm": 1.3316107988357544,
      "learning_rate": 9.996142426640155e-05,
      "loss": 10.1409,
      "step": 135
    },
    {
      "epoch": 0.046912728527078305,
      "grad_norm": 1.3772598505020142,
      "learning_rate": 9.995918875257594e-05,
      "loss": 10.1151,
      "step": 136
    },
    {
      "epoch": 0.04725767506036564,
      "grad_norm": 1.3657079935073853,
      "learning_rate": 9.995689030128041e-05,
      "loss": 10.1275,
      "step": 137
    },
    {
      "epoch": 0.047602621593652986,
      "grad_norm": 1.3499964475631714,
      "learning_rate": 9.995452891541048e-05,
      "loss": 10.1198,
      "step": 138
    },
    {
      "epoch": 0.04794756812694032,
      "grad_norm": 1.3011778593063354,
      "learning_rate": 9.995210459794098e-05,
      "loss": 10.1099,
      "step": 139
    },
    {
      "epoch": 0.048292514660227666,
      "grad_norm": 1.407361626625061,
      "learning_rate": 9.994961735192598e-05,
      "loss": 10.1072,
      "step": 140
    },
    {
      "epoch": 0.048637461193515,
      "grad_norm": 1.399038553237915,
      "learning_rate": 9.994706718049889e-05,
      "loss": 10.1009,
      "step": 141
    },
    {
      "epoch": 0.048982407726802346,
      "grad_norm": 1.3892508745193481,
      "learning_rate": 9.994445408687236e-05,
      "loss": 10.1054,
      "step": 142
    },
    {
      "epoch": 0.04932735426008968,
      "grad_norm": 1.4403347969055176,
      "learning_rate": 9.994177807433826e-05,
      "loss": 10.0781,
      "step": 143
    },
    {
      "epoch": 0.049672300793377026,
      "grad_norm": 1.472110629081726,
      "learning_rate": 9.993903914626783e-05,
      "loss": 10.0767,
      "step": 144
    },
    {
      "epoch": 0.05001724732666437,
      "grad_norm": 1.4474997520446777,
      "learning_rate": 9.993623730611147e-05,
      "loss": 10.0908,
      "step": 145
    },
    {
      "epoch": 0.05036219385995171,
      "grad_norm": 1.4426510334014893,
      "learning_rate": 9.993337255739892e-05,
      "loss": 10.0792,
      "step": 146
    },
    {
      "epoch": 0.05070714039323905,
      "grad_norm": 1.5582860708236694,
      "learning_rate": 9.993044490373907e-05,
      "loss": 10.0618,
      "step": 147
    },
    {
      "epoch": 0.05105208692652639,
      "grad_norm": 1.5666154623031616,
      "learning_rate": 9.992745434882014e-05,
      "loss": 10.0664,
      "step": 148
    },
    {
      "epoch": 0.05139703345981373,
      "grad_norm": 1.5812947750091553,
      "learning_rate": 9.992440089640958e-05,
      "loss": 10.0604,
      "step": 149
    },
    {
      "epoch": 0.05174197999310107,
      "grad_norm": 1.7215590476989746,
      "learning_rate": 9.992128455035403e-05,
      "loss": 10.0412,
      "step": 150
    },
    {
      "epoch": 0.05208692652638841,
      "grad_norm": 0.9909889101982117,
      "learning_rate": 9.991810531457943e-05,
      "loss": 10.1461,
      "step": 151
    },
    {
      "epoch": 0.05243187305967575,
      "grad_norm": 1.0242069959640503,
      "learning_rate": 9.991486319309086e-05,
      "loss": 10.1037,
      "step": 152
    },
    {
      "epoch": 0.05277681959296309,
      "grad_norm": 1.1299318075180054,
      "learning_rate": 9.991155818997273e-05,
      "loss": 10.1342,
      "step": 153
    },
    {
      "epoch": 0.053121766126250435,
      "grad_norm": 1.19455087184906,
      "learning_rate": 9.990819030938856e-05,
      "loss": 10.1224,
      "step": 154
    },
    {
      "epoch": 0.05346671265953777,
      "grad_norm": 1.1440625190734863,
      "learning_rate": 9.990475955558116e-05,
      "loss": 10.1076,
      "step": 155
    },
    {
      "epoch": 0.053811659192825115,
      "grad_norm": 1.1086006164550781,
      "learning_rate": 9.990126593287252e-05,
      "loss": 10.1221,
      "step": 156
    },
    {
      "epoch": 0.05415660572611245,
      "grad_norm": 1.0412251949310303,
      "learning_rate": 9.989770944566379e-05,
      "loss": 10.1014,
      "step": 157
    },
    {
      "epoch": 0.054501552259399795,
      "grad_norm": 1.1727021932601929,
      "learning_rate": 9.98940900984354e-05,
      "loss": 10.0921,
      "step": 158
    },
    {
      "epoch": 0.05484649879268713,
      "grad_norm": 1.1224405765533447,
      "learning_rate": 9.989040789574689e-05,
      "loss": 10.1181,
      "step": 159
    },
    {
      "epoch": 0.055191445325974475,
      "grad_norm": 1.233102798461914,
      "learning_rate": 9.988666284223703e-05,
      "loss": 10.0801,
      "step": 160
    },
    {
      "epoch": 0.05553639185926181,
      "grad_norm": 1.0544770956039429,
      "learning_rate": 9.988285494262377e-05,
      "loss": 10.1011,
      "step": 161
    },
    {
      "epoch": 0.055881338392549156,
      "grad_norm": 1.1918315887451172,
      "learning_rate": 9.987898420170418e-05,
      "loss": 10.0912,
      "step": 162
    },
    {
      "epoch": 0.05622628492583649,
      "grad_norm": 1.2117809057235718,
      "learning_rate": 9.987505062435455e-05,
      "loss": 10.0917,
      "step": 163
    },
    {
      "epoch": 0.056571231459123836,
      "grad_norm": 1.2464489936828613,
      "learning_rate": 9.987105421553032e-05,
      "loss": 10.0741,
      "step": 164
    },
    {
      "epoch": 0.05691617799241118,
      "grad_norm": 1.2690608501434326,
      "learning_rate": 9.986699498026609e-05,
      "loss": 10.0735,
      "step": 165
    },
    {
      "epoch": 0.057261124525698516,
      "grad_norm": 1.3744335174560547,
      "learning_rate": 9.986287292367555e-05,
      "loss": 10.0801,
      "step": 166
    },
    {
      "epoch": 0.05760607105898586,
      "grad_norm": 1.2143759727478027,
      "learning_rate": 9.985868805095163e-05,
      "loss": 10.1026,
      "step": 167
    },
    {
      "epoch": 0.057951017592273196,
      "grad_norm": 1.1724203824996948,
      "learning_rate": 9.985444036736628e-05,
      "loss": 10.0753,
      "step": 168
    },
    {
      "epoch": 0.05829596412556054,
      "grad_norm": 1.229285717010498,
      "learning_rate": 9.985012987827068e-05,
      "loss": 10.0444,
      "step": 169
    },
    {
      "epoch": 0.058640910658847877,
      "grad_norm": 1.280335545539856,
      "learning_rate": 9.984575658909508e-05,
      "loss": 10.0446,
      "step": 170
    },
    {
      "epoch": 0.05898585719213522,
      "grad_norm": 1.1907861232757568,
      "learning_rate": 9.984132050534885e-05,
      "loss": 10.0792,
      "step": 171
    },
    {
      "epoch": 0.05933080372542256,
      "grad_norm": 1.2323590517044067,
      "learning_rate": 9.983682163262044e-05,
      "loss": 10.0562,
      "step": 172
    },
    {
      "epoch": 0.0596757502587099,
      "grad_norm": 1.2697149515151978,
      "learning_rate": 9.983225997657749e-05,
      "loss": 10.0437,
      "step": 173
    },
    {
      "epoch": 0.06002069679199724,
      "grad_norm": 1.255582332611084,
      "learning_rate": 9.982763554296662e-05,
      "loss": 10.0482,
      "step": 174
    },
    {
      "epoch": 0.06036564332528458,
      "grad_norm": 1.2350727319717407,
      "learning_rate": 9.98229483376136e-05,
      "loss": 10.0387,
      "step": 175
    },
    {
      "epoch": 0.060710589858571924,
      "grad_norm": 1.3043851852416992,
      "learning_rate": 9.98181983664233e-05,
      "loss": 10.0554,
      "step": 176
    },
    {
      "epoch": 0.06105553639185926,
      "grad_norm": 1.152021884918213,
      "learning_rate": 9.981338563537959e-05,
      "loss": 10.0452,
      "step": 177
    },
    {
      "epoch": 0.061400482925146604,
      "grad_norm": 1.3231796026229858,
      "learning_rate": 9.980851015054543e-05,
      "loss": 10.036,
      "step": 178
    },
    {
      "epoch": 0.06174542945843394,
      "grad_norm": 1.3968857526779175,
      "learning_rate": 9.980357191806288e-05,
      "loss": 10.0244,
      "step": 179
    },
    {
      "epoch": 0.062090375991721285,
      "grad_norm": 1.324229121208191,
      "learning_rate": 9.979857094415301e-05,
      "loss": 10.0357,
      "step": 180
    },
    {
      "epoch": 0.06243532252500862,
      "grad_norm": 1.4936314821243286,
      "learning_rate": 9.979350723511594e-05,
      "loss": 10.0231,
      "step": 181
    },
    {
      "epoch": 0.06278026905829596,
      "grad_norm": 1.2559239864349365,
      "learning_rate": 9.978838079733078e-05,
      "loss": 10.0112,
      "step": 182
    },
    {
      "epoch": 0.06312521559158331,
      "grad_norm": 1.2411177158355713,
      "learning_rate": 9.978319163725574e-05,
      "loss": 10.0321,
      "step": 183
    },
    {
      "epoch": 0.06347016212487064,
      "grad_norm": 1.3505336046218872,
      "learning_rate": 9.9777939761428e-05,
      "loss": 10.0298,
      "step": 184
    },
    {
      "epoch": 0.06381510865815798,
      "grad_norm": 1.369786024093628,
      "learning_rate": 9.977262517646373e-05,
      "loss": 10.0121,
      "step": 185
    },
    {
      "epoch": 0.06416005519144533,
      "grad_norm": 1.348097801208496,
      "learning_rate": 9.976724788905817e-05,
      "loss": 10.0134,
      "step": 186
    },
    {
      "epoch": 0.06450500172473267,
      "grad_norm": 1.419002890586853,
      "learning_rate": 9.976180790598549e-05,
      "loss": 10.0246,
      "step": 187
    },
    {
      "epoch": 0.06484994825802001,
      "grad_norm": 1.3383148908615112,
      "learning_rate": 9.975630523409882e-05,
      "loss": 9.9983,
      "step": 188
    },
    {
      "epoch": 0.06519489479130734,
      "grad_norm": 1.3767681121826172,
      "learning_rate": 9.975073988033037e-05,
      "loss": 9.961,
      "step": 189
    },
    {
      "epoch": 0.06553984132459469,
      "grad_norm": 1.250262975692749,
      "learning_rate": 9.97451118516912e-05,
      "loss": 9.9955,
      "step": 190
    },
    {
      "epoch": 0.06588478785788203,
      "grad_norm": 1.399427056312561,
      "learning_rate": 9.973942115527137e-05,
      "loss": 10.0077,
      "step": 191
    },
    {
      "epoch": 0.06622973439116937,
      "grad_norm": 1.5137879848480225,
      "learning_rate": 9.973366779823994e-05,
      "loss": 9.9779,
      "step": 192
    },
    {
      "epoch": 0.0665746809244567,
      "grad_norm": 1.487107515335083,
      "learning_rate": 9.972785178784482e-05,
      "loss": 9.963,
      "step": 193
    },
    {
      "epoch": 0.06691962745774405,
      "grad_norm": 1.4277721643447876,
      "learning_rate": 9.972197313141291e-05,
      "loss": 9.9667,
      "step": 194
    },
    {
      "epoch": 0.06726457399103139,
      "grad_norm": 1.3553154468536377,
      "learning_rate": 9.971603183634998e-05,
      "loss": 9.977,
      "step": 195
    },
    {
      "epoch": 0.06760952052431873,
      "grad_norm": 1.3985073566436768,
      "learning_rate": 9.971002791014078e-05,
      "loss": 9.9725,
      "step": 196
    },
    {
      "epoch": 0.06795446705760608,
      "grad_norm": 1.6270498037338257,
      "learning_rate": 9.97039613603489e-05,
      "loss": 9.9501,
      "step": 197
    },
    {
      "epoch": 0.0682994135908934,
      "grad_norm": 1.6373891830444336,
      "learning_rate": 9.969783219461685e-05,
      "loss": 9.8966,
      "step": 198
    },
    {
      "epoch": 0.06864436012418075,
      "grad_norm": 1.5864276885986328,
      "learning_rate": 9.969164042066603e-05,
      "loss": 9.9174,
      "step": 199
    },
    {
      "epoch": 0.0689893066574681,
      "grad_norm": 1.659999966621399,
      "learning_rate": 9.968538604629667e-05,
      "loss": 9.9173,
      "step": 200
    },
    {
      "epoch": 0.06933425319075544,
      "grad_norm": 0.975853681564331,
      "learning_rate": 9.96790690793879e-05,
      "loss": 10.0666,
      "step": 201
    },
    {
      "epoch": 0.06967919972404277,
      "grad_norm": 1.065144658088684,
      "learning_rate": 9.96726895278977e-05,
      "loss": 10.0229,
      "step": 202
    },
    {
      "epoch": 0.07002414625733011,
      "grad_norm": 1.06117844581604,
      "learning_rate": 9.966624739986289e-05,
      "loss": 10.0301,
      "step": 203
    },
    {
      "epoch": 0.07036909279061745,
      "grad_norm": 1.0030990839004517,
      "learning_rate": 9.965974270339911e-05,
      "loss": 10.0408,
      "step": 204
    },
    {
      "epoch": 0.0707140393239048,
      "grad_norm": 1.122992753982544,
      "learning_rate": 9.965317544670083e-05,
      "loss": 10.0107,
      "step": 205
    },
    {
      "epoch": 0.07105898585719214,
      "grad_norm": 1.028238296508789,
      "learning_rate": 9.964654563804134e-05,
      "loss": 10.0143,
      "step": 206
    },
    {
      "epoch": 0.07140393239047947,
      "grad_norm": 1.094612717628479,
      "learning_rate": 9.963985328577273e-05,
      "loss": 9.9905,
      "step": 207
    },
    {
      "epoch": 0.07174887892376682,
      "grad_norm": 1.083109736442566,
      "learning_rate": 9.963309839832586e-05,
      "loss": 9.9978,
      "step": 208
    },
    {
      "epoch": 0.07209382545705416,
      "grad_norm": 1.2037516832351685,
      "learning_rate": 9.962628098421041e-05,
      "loss": 9.985,
      "step": 209
    },
    {
      "epoch": 0.0724387719903415,
      "grad_norm": 1.1501799821853638,
      "learning_rate": 9.961940105201477e-05,
      "loss": 9.9767,
      "step": 210
    },
    {
      "epoch": 0.07278371852362883,
      "grad_norm": 1.1958119869232178,
      "learning_rate": 9.961245861040617e-05,
      "loss": 10.0039,
      "step": 211
    },
    {
      "epoch": 0.07312866505691618,
      "grad_norm": 1.1737613677978516,
      "learning_rate": 9.96054536681305e-05,
      "loss": 10.0003,
      "step": 212
    },
    {
      "epoch": 0.07347361159020352,
      "grad_norm": 1.246880054473877,
      "learning_rate": 9.959838623401246e-05,
      "loss": 9.9803,
      "step": 213
    },
    {
      "epoch": 0.07381855812349086,
      "grad_norm": 1.1679513454437256,
      "learning_rate": 9.959125631695543e-05,
      "loss": 9.9684,
      "step": 214
    },
    {
      "epoch": 0.07416350465677819,
      "grad_norm": 1.2343090772628784,
      "learning_rate": 9.958406392594153e-05,
      "loss": 10.013,
      "step": 215
    },
    {
      "epoch": 0.07450845119006554,
      "grad_norm": 1.1205172538757324,
      "learning_rate": 9.957680907003155e-05,
      "loss": 10.0095,
      "step": 216
    },
    {
      "epoch": 0.07485339772335288,
      "grad_norm": 1.1602991819381714,
      "learning_rate": 9.956949175836503e-05,
      "loss": 9.9666,
      "step": 217
    },
    {
      "epoch": 0.07519834425664022,
      "grad_norm": 1.2786660194396973,
      "learning_rate": 9.956211200016013e-05,
      "loss": 9.9578,
      "step": 218
    },
    {
      "epoch": 0.07554329078992757,
      "grad_norm": 1.2090095281600952,
      "learning_rate": 9.955466980471373e-05,
      "loss": 9.9574,
      "step": 219
    },
    {
      "epoch": 0.0758882373232149,
      "grad_norm": 1.303739070892334,
      "learning_rate": 9.95471651814013e-05,
      "loss": 9.9349,
      "step": 220
    },
    {
      "epoch": 0.07623318385650224,
      "grad_norm": 1.188925862312317,
      "learning_rate": 9.953959813967704e-05,
      "loss": 9.9513,
      "step": 221
    },
    {
      "epoch": 0.07657813038978958,
      "grad_norm": 1.3651727437973022,
      "learning_rate": 9.953196868907373e-05,
      "loss": 9.9574,
      "step": 222
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 1.1526354551315308,
      "learning_rate": 9.952427683920276e-05,
      "loss": 9.9819,
      "step": 223
    },
    {
      "epoch": 0.07726802345636426,
      "grad_norm": 1.3217377662658691,
      "learning_rate": 9.951652259975418e-05,
      "loss": 9.961,
      "step": 224
    },
    {
      "epoch": 0.0776129699896516,
      "grad_norm": 1.2583246231079102,
      "learning_rate": 9.950870598049657e-05,
      "loss": 9.9224,
      "step": 225
    },
    {
      "epoch": 0.07795791652293894,
      "grad_norm": 1.2618720531463623,
      "learning_rate": 9.950082699127717e-05,
      "loss": 9.972,
      "step": 226
    },
    {
      "epoch": 0.07830286305622629,
      "grad_norm": 1.329683542251587,
      "learning_rate": 9.949288564202172e-05,
      "loss": 9.9146,
      "step": 227
    },
    {
      "epoch": 0.07864780958951363,
      "grad_norm": 1.3133656978607178,
      "learning_rate": 9.948488194273458e-05,
      "loss": 9.9109,
      "step": 228
    },
    {
      "epoch": 0.07899275612280096,
      "grad_norm": 1.4228748083114624,
      "learning_rate": 9.947681590349863e-05,
      "loss": 9.9242,
      "step": 229
    },
    {
      "epoch": 0.0793377026560883,
      "grad_norm": 1.2924089431762695,
      "learning_rate": 9.946868753447529e-05,
      "loss": 9.9271,
      "step": 230
    },
    {
      "epoch": 0.07968264918937565,
      "grad_norm": 1.463861107826233,
      "learning_rate": 9.946049684590448e-05,
      "loss": 9.8871,
      "step": 231
    },
    {
      "epoch": 0.08002759572266299,
      "grad_norm": 1.2419488430023193,
      "learning_rate": 9.945224384810465e-05,
      "loss": 9.9425,
      "step": 232
    },
    {
      "epoch": 0.08037254225595032,
      "grad_norm": 1.4464517831802368,
      "learning_rate": 9.944392855147275e-05,
      "loss": 9.8807,
      "step": 233
    },
    {
      "epoch": 0.08071748878923767,
      "grad_norm": 1.4393879175186157,
      "learning_rate": 9.94355509664842e-05,
      "loss": 9.9006,
      "step": 234
    },
    {
      "epoch": 0.08106243532252501,
      "grad_norm": 1.3730369806289673,
      "learning_rate": 9.942711110369292e-05,
      "loss": 9.9252,
      "step": 235
    },
    {
      "epoch": 0.08140738185581235,
      "grad_norm": 1.5222467184066772,
      "learning_rate": 9.941860897373121e-05,
      "loss": 9.8747,
      "step": 236
    },
    {
      "epoch": 0.0817523283890997,
      "grad_norm": 1.516811728477478,
      "learning_rate": 9.94100445873099e-05,
      "loss": 9.8623,
      "step": 237
    },
    {
      "epoch": 0.08209727492238703,
      "grad_norm": 1.3974412679672241,
      "learning_rate": 9.94014179552182e-05,
      "loss": 9.8805,
      "step": 238
    },
    {
      "epoch": 0.08244222145567437,
      "grad_norm": 1.4104135036468506,
      "learning_rate": 9.939272908832376e-05,
      "loss": 9.8955,
      "step": 239
    },
    {
      "epoch": 0.08278716798896171,
      "grad_norm": 1.3622193336486816,
      "learning_rate": 9.938397799757261e-05,
      "loss": 9.8692,
      "step": 240
    },
    {
      "epoch": 0.08313211452224906,
      "grad_norm": 1.4380640983581543,
      "learning_rate": 9.937516469398918e-05,
      "loss": 9.8608,
      "step": 241
    },
    {
      "epoch": 0.08347706105553639,
      "grad_norm": 1.483670711517334,
      "learning_rate": 9.936628918867627e-05,
      "loss": 9.8679,
      "step": 242
    },
    {
      "epoch": 0.08382200758882373,
      "grad_norm": 1.5402473211288452,
      "learning_rate": 9.935735149281504e-05,
      "loss": 9.8424,
      "step": 243
    },
    {
      "epoch": 0.08416695412211107,
      "grad_norm": 1.4765286445617676,
      "learning_rate": 9.934835161766502e-05,
      "loss": 9.8672,
      "step": 244
    },
    {
      "epoch": 0.08451190065539842,
      "grad_norm": 1.4256876707077026,
      "learning_rate": 9.933928957456404e-05,
      "loss": 9.8916,
      "step": 245
    },
    {
      "epoch": 0.08485684718868576,
      "grad_norm": 1.4095759391784668,
      "learning_rate": 9.933016537492826e-05,
      "loss": 9.8809,
      "step": 246
    },
    {
      "epoch": 0.08520179372197309,
      "grad_norm": 1.64491868019104,
      "learning_rate": 9.932097903025212e-05,
      "loss": 9.8161,
      "step": 247
    },
    {
      "epoch": 0.08554674025526043,
      "grad_norm": 1.608772873878479,
      "learning_rate": 9.93117305521084e-05,
      "loss": 9.8112,
      "step": 248
    },
    {
      "epoch": 0.08589168678854778,
      "grad_norm": 1.500699758529663,
      "learning_rate": 9.930241995214814e-05,
      "loss": 9.8622,
      "step": 249
    },
    {
      "epoch": 0.08623663332183512,
      "grad_norm": 1.723171591758728,
      "learning_rate": 9.929304724210058e-05,
      "loss": 9.7795,
      "step": 250
    },
    {
      "epoch": 0.08658157985512245,
      "grad_norm": 1.044055700302124,
      "learning_rate": 9.928361243377328e-05,
      "loss": 9.9278,
      "step": 251
    },
    {
      "epoch": 0.0869265263884098,
      "grad_norm": 0.9544603824615479,
      "learning_rate": 9.927411553905201e-05,
      "loss": 9.9693,
      "step": 252
    },
    {
      "epoch": 0.08727147292169714,
      "grad_norm": 1.0985561609268188,
      "learning_rate": 9.926455656990073e-05,
      "loss": 9.9014,
      "step": 253
    },
    {
      "epoch": 0.08761641945498448,
      "grad_norm": 1.0646491050720215,
      "learning_rate": 9.92549355383616e-05,
      "loss": 9.9267,
      "step": 254
    },
    {
      "epoch": 0.08796136598827181,
      "grad_norm": 1.0615615844726562,
      "learning_rate": 9.924525245655501e-05,
      "loss": 9.931,
      "step": 255
    },
    {
      "epoch": 0.08830631252155915,
      "grad_norm": 1.125076413154602,
      "learning_rate": 9.923550733667948e-05,
      "loss": 9.9225,
      "step": 256
    },
    {
      "epoch": 0.0886512590548465,
      "grad_norm": 1.1765639781951904,
      "learning_rate": 9.92257001910117e-05,
      "loss": 9.8822,
      "step": 257
    },
    {
      "epoch": 0.08899620558813384,
      "grad_norm": 1.0968784093856812,
      "learning_rate": 9.921583103190648e-05,
      "loss": 9.9152,
      "step": 258
    },
    {
      "epoch": 0.08934115212142119,
      "grad_norm": 1.183980941772461,
      "learning_rate": 9.920589987179676e-05,
      "loss": 9.9258,
      "step": 259
    },
    {
      "epoch": 0.08968609865470852,
      "grad_norm": 1.1285295486450195,
      "learning_rate": 9.919590672319361e-05,
      "loss": 9.8847,
      "step": 260
    },
    {
      "epoch": 0.09003104518799586,
      "grad_norm": 1.0491220951080322,
      "learning_rate": 9.918585159868617e-05,
      "loss": 9.9073,
      "step": 261
    },
    {
      "epoch": 0.0903759917212832,
      "grad_norm": 1.3533495664596558,
      "learning_rate": 9.917573451094168e-05,
      "loss": 9.8524,
      "step": 262
    },
    {
      "epoch": 0.09072093825457055,
      "grad_norm": 1.210127592086792,
      "learning_rate": 9.916555547270538e-05,
      "loss": 9.8654,
      "step": 263
    },
    {
      "epoch": 0.09106588478785788,
      "grad_norm": 1.2345919609069824,
      "learning_rate": 9.915531449680063e-05,
      "loss": 9.8847,
      "step": 264
    },
    {
      "epoch": 0.09141083132114522,
      "grad_norm": 1.298925757408142,
      "learning_rate": 9.914501159612876e-05,
      "loss": 9.8865,
      "step": 265
    },
    {
      "epoch": 0.09175577785443256,
      "grad_norm": 1.2874445915222168,
      "learning_rate": 9.913464678366917e-05,
      "loss": 9.8412,
      "step": 266
    },
    {
      "epoch": 0.0921007243877199,
      "grad_norm": 1.1336477994918823,
      "learning_rate": 9.912422007247917e-05,
      "loss": 9.8807,
      "step": 267
    },
    {
      "epoch": 0.09244567092100725,
      "grad_norm": 1.3368065357208252,
      "learning_rate": 9.911373147569414e-05,
      "loss": 9.8182,
      "step": 268
    },
    {
      "epoch": 0.09279061745429458,
      "grad_norm": 1.2617908716201782,
      "learning_rate": 9.910318100652736e-05,
      "loss": 9.8705,
      "step": 269
    },
    {
      "epoch": 0.09313556398758192,
      "grad_norm": 1.1415497064590454,
      "learning_rate": 9.909256867827006e-05,
      "loss": 9.8783,
      "step": 270
    },
    {
      "epoch": 0.09348051052086927,
      "grad_norm": 1.3616750240325928,
      "learning_rate": 9.908189450429143e-05,
      "loss": 9.8499,
      "step": 271
    },
    {
      "epoch": 0.09382545705415661,
      "grad_norm": 1.3078230619430542,
      "learning_rate": 9.907115849803859e-05,
      "loss": 9.8458,
      "step": 272
    },
    {
      "epoch": 0.09417040358744394,
      "grad_norm": 1.245226502418518,
      "learning_rate": 9.906036067303646e-05,
      "loss": 9.8787,
      "step": 273
    },
    {
      "epoch": 0.09451535012073128,
      "grad_norm": 1.316032886505127,
      "learning_rate": 9.904950104288793e-05,
      "loss": 9.8522,
      "step": 274
    },
    {
      "epoch": 0.09486029665401863,
      "grad_norm": 1.544091820716858,
      "learning_rate": 9.903857962127372e-05,
      "loss": 9.822,
      "step": 275
    },
    {
      "epoch": 0.09520524318730597,
      "grad_norm": 1.240692138671875,
      "learning_rate": 9.902759642195239e-05,
      "loss": 9.8909,
      "step": 276
    },
    {
      "epoch": 0.09555018972059331,
      "grad_norm": 1.4124935865402222,
      "learning_rate": 9.901655145876034e-05,
      "loss": 9.8022,
      "step": 277
    },
    {
      "epoch": 0.09589513625388064,
      "grad_norm": 1.238968849182129,
      "learning_rate": 9.900544474561174e-05,
      "loss": 9.8938,
      "step": 278
    },
    {
      "epoch": 0.09624008278716799,
      "grad_norm": 1.2817027568817139,
      "learning_rate": 9.89942762964986e-05,
      "loss": 9.8653,
      "step": 279
    },
    {
      "epoch": 0.09658502932045533,
      "grad_norm": 1.2441328763961792,
      "learning_rate": 9.898304612549067e-05,
      "loss": 9.8592,
      "step": 280
    },
    {
      "epoch": 0.09692997585374268,
      "grad_norm": 1.3548439741134644,
      "learning_rate": 9.89717542467355e-05,
      "loss": 9.8568,
      "step": 281
    },
    {
      "epoch": 0.09727492238703,
      "grad_norm": 1.3438184261322021,
      "learning_rate": 9.896040067445831e-05,
      "loss": 9.8484,
      "step": 282
    },
    {
      "epoch": 0.09761986892031735,
      "grad_norm": 1.4415910243988037,
      "learning_rate": 9.894898542296209e-05,
      "loss": 9.7968,
      "step": 283
    },
    {
      "epoch": 0.09796481545360469,
      "grad_norm": 1.273112177848816,
      "learning_rate": 9.893750850662752e-05,
      "loss": 9.8928,
      "step": 284
    },
    {
      "epoch": 0.09830976198689204,
      "grad_norm": 1.429869294166565,
      "learning_rate": 9.892596993991297e-05,
      "loss": 9.8333,
      "step": 285
    },
    {
      "epoch": 0.09865470852017937,
      "grad_norm": 1.5786771774291992,
      "learning_rate": 9.891436973735444e-05,
      "loss": 9.8043,
      "step": 286
    },
    {
      "epoch": 0.09899965505346671,
      "grad_norm": 1.3031753301620483,
      "learning_rate": 9.890270791356564e-05,
      "loss": 9.837,
      "step": 287
    },
    {
      "epoch": 0.09934460158675405,
      "grad_norm": 1.3726156949996948,
      "learning_rate": 9.889098448323786e-05,
      "loss": 9.8427,
      "step": 288
    },
    {
      "epoch": 0.0996895481200414,
      "grad_norm": 1.4839277267456055,
      "learning_rate": 9.887919946114001e-05,
      "loss": 9.7975,
      "step": 289
    },
    {
      "epoch": 0.10003449465332874,
      "grad_norm": 1.5329991579055786,
      "learning_rate": 9.886735286211861e-05,
      "loss": 9.7854,
      "step": 290
    },
    {
      "epoch": 0.10037944118661607,
      "grad_norm": 1.3865573406219482,
      "learning_rate": 9.885544470109774e-05,
      "loss": 9.7865,
      "step": 291
    },
    {
      "epoch": 0.10072438771990341,
      "grad_norm": 1.2633044719696045,
      "learning_rate": 9.884347499307902e-05,
      "loss": 9.8258,
      "step": 292
    },
    {
      "epoch": 0.10106933425319076,
      "grad_norm": 1.5327684879302979,
      "learning_rate": 9.883144375314164e-05,
      "loss": 9.7834,
      "step": 293
    },
    {
      "epoch": 0.1014142807864781,
      "grad_norm": 1.544948697090149,
      "learning_rate": 9.881935099644228e-05,
      "loss": 9.7626,
      "step": 294
    },
    {
      "epoch": 0.10175922731976543,
      "grad_norm": 1.3965122699737549,
      "learning_rate": 9.880719673821512e-05,
      "loss": 9.7893,
      "step": 295
    },
    {
      "epoch": 0.10210417385305277,
      "grad_norm": 1.5461406707763672,
      "learning_rate": 9.879498099377183e-05,
      "loss": 9.7662,
      "step": 296
    },
    {
      "epoch": 0.10244912038634012,
      "grad_norm": 1.5404688119888306,
      "learning_rate": 9.878270377850153e-05,
      "loss": 9.7516,
      "step": 297
    },
    {
      "epoch": 0.10279406691962746,
      "grad_norm": 1.6081411838531494,
      "learning_rate": 9.877036510787077e-05,
      "loss": 9.7487,
      "step": 298
    },
    {
      "epoch": 0.1031390134529148,
      "grad_norm": 1.6868000030517578,
      "learning_rate": 9.875796499742355e-05,
      "loss": 9.7492,
      "step": 299
    },
    {
      "epoch": 0.10348395998620213,
      "grad_norm": 1.6842550039291382,
      "learning_rate": 9.874550346278123e-05,
      "loss": 9.7523,
      "step": 300
    },
    {
      "epoch": 0.10382890651948948,
      "grad_norm": 0.9228402376174927,
      "learning_rate": 9.873298051964261e-05,
      "loss": 9.8752,
      "step": 301
    },
    {
      "epoch": 0.10417385305277682,
      "grad_norm": 1.069778561592102,
      "learning_rate": 9.872039618378378e-05,
      "loss": 9.8213,
      "step": 302
    },
    {
      "epoch": 0.10451879958606416,
      "grad_norm": 1.0447841882705688,
      "learning_rate": 9.870775047105822e-05,
      "loss": 9.8148,
      "step": 303
    },
    {
      "epoch": 0.1048637461193515,
      "grad_norm": 1.022067666053772,
      "learning_rate": 9.869504339739671e-05,
      "loss": 9.8546,
      "step": 304
    },
    {
      "epoch": 0.10520869265263884,
      "grad_norm": 1.168174386024475,
      "learning_rate": 9.868227497880733e-05,
      "loss": 9.8186,
      "step": 305
    },
    {
      "epoch": 0.10555363918592618,
      "grad_norm": 1.1317428350448608,
      "learning_rate": 9.866944523137546e-05,
      "loss": 9.8244,
      "step": 306
    },
    {
      "epoch": 0.10589858571921353,
      "grad_norm": 1.1827465295791626,
      "learning_rate": 9.865655417126374e-05,
      "loss": 9.8549,
      "step": 307
    },
    {
      "epoch": 0.10624353225250087,
      "grad_norm": 1.0269920825958252,
      "learning_rate": 9.864360181471202e-05,
      "loss": 9.8324,
      "step": 308
    },
    {
      "epoch": 0.1065884787857882,
      "grad_norm": 1.16238272190094,
      "learning_rate": 9.86305881780374e-05,
      "loss": 9.8568,
      "step": 309
    },
    {
      "epoch": 0.10693342531907554,
      "grad_norm": 1.1754971742630005,
      "learning_rate": 9.861751327763415e-05,
      "loss": 9.8146,
      "step": 310
    },
    {
      "epoch": 0.10727837185236289,
      "grad_norm": 1.247070550918579,
      "learning_rate": 9.860437712997378e-05,
      "loss": 9.8329,
      "step": 311
    },
    {
      "epoch": 0.10762331838565023,
      "grad_norm": 1.2488677501678467,
      "learning_rate": 9.859117975160488e-05,
      "loss": 9.7842,
      "step": 312
    },
    {
      "epoch": 0.10796826491893756,
      "grad_norm": 1.2628960609436035,
      "learning_rate": 9.85779211591532e-05,
      "loss": 9.8005,
      "step": 313
    },
    {
      "epoch": 0.1083132114522249,
      "grad_norm": 1.1880953311920166,
      "learning_rate": 9.856460136932166e-05,
      "loss": 9.8051,
      "step": 314
    },
    {
      "epoch": 0.10865815798551225,
      "grad_norm": 1.2296935319900513,
      "learning_rate": 9.85512203988902e-05,
      "loss": 9.7838,
      "step": 315
    },
    {
      "epoch": 0.10900310451879959,
      "grad_norm": 1.2373672723770142,
      "learning_rate": 9.853777826471589e-05,
      "loss": 9.7733,
      "step": 316
    },
    {
      "epoch": 0.10934805105208692,
      "grad_norm": 1.2010676860809326,
      "learning_rate": 9.852427498373283e-05,
      "loss": 9.8074,
      "step": 317
    },
    {
      "epoch": 0.10969299758537426,
      "grad_norm": 1.1413911581039429,
      "learning_rate": 9.851071057295213e-05,
      "loss": 9.8098,
      "step": 318
    },
    {
      "epoch": 0.11003794411866161,
      "grad_norm": 1.1600428819656372,
      "learning_rate": 9.849708504946193e-05,
      "loss": 9.8,
      "step": 319
    },
    {
      "epoch": 0.11038289065194895,
      "grad_norm": 1.2346547842025757,
      "learning_rate": 9.84833984304274e-05,
      "loss": 9.7597,
      "step": 320
    },
    {
      "epoch": 0.1107278371852363,
      "grad_norm": 1.3820494413375854,
      "learning_rate": 9.84696507330906e-05,
      "loss": 9.749,
      "step": 321
    },
    {
      "epoch": 0.11107278371852362,
      "grad_norm": 1.233407735824585,
      "learning_rate": 9.845584197477058e-05,
      "loss": 9.7642,
      "step": 322
    },
    {
      "epoch": 0.11141773025181097,
      "grad_norm": 1.3942359685897827,
      "learning_rate": 9.844197217286332e-05,
      "loss": 9.7944,
      "step": 323
    },
    {
      "epoch": 0.11176267678509831,
      "grad_norm": 1.2736047506332397,
      "learning_rate": 9.842804134484165e-05,
      "loss": 9.7716,
      "step": 324
    },
    {
      "epoch": 0.11210762331838565,
      "grad_norm": 1.3517143726348877,
      "learning_rate": 9.841404950825537e-05,
      "loss": 9.812,
      "step": 325
    },
    {
      "epoch": 0.11245256985167298,
      "grad_norm": 1.2531120777130127,
      "learning_rate": 9.839999668073105e-05,
      "loss": 9.7383,
      "step": 326
    },
    {
      "epoch": 0.11279751638496033,
      "grad_norm": 1.3106811046600342,
      "learning_rate": 9.838588287997212e-05,
      "loss": 9.7353,
      "step": 327
    },
    {
      "epoch": 0.11314246291824767,
      "grad_norm": 1.3290574550628662,
      "learning_rate": 9.837170812375885e-05,
      "loss": 9.7201,
      "step": 328
    },
    {
      "epoch": 0.11348740945153502,
      "grad_norm": 1.5268999338150024,
      "learning_rate": 9.835747242994828e-05,
      "loss": 9.7307,
      "step": 329
    },
    {
      "epoch": 0.11383235598482236,
      "grad_norm": 1.3327301740646362,
      "learning_rate": 9.83431758164742e-05,
      "loss": 9.7123,
      "step": 330
    },
    {
      "epoch": 0.11417730251810969,
      "grad_norm": 1.3089390993118286,
      "learning_rate": 9.832881830134719e-05,
      "loss": 9.7895,
      "step": 331
    },
    {
      "epoch": 0.11452224905139703,
      "grad_norm": 1.3884986639022827,
      "learning_rate": 9.83143999026545e-05,
      "loss": 9.7379,
      "step": 332
    },
    {
      "epoch": 0.11486719558468438,
      "grad_norm": 1.3758727312088013,
      "learning_rate": 9.829992063856013e-05,
      "loss": 9.7225,
      "step": 333
    },
    {
      "epoch": 0.11521214211797172,
      "grad_norm": 1.4772086143493652,
      "learning_rate": 9.828538052730473e-05,
      "loss": 9.7427,
      "step": 334
    },
    {
      "epoch": 0.11555708865125905,
      "grad_norm": 1.4043651819229126,
      "learning_rate": 9.82707795872056e-05,
      "loss": 9.7053,
      "step": 335
    },
    {
      "epoch": 0.11590203518454639,
      "grad_norm": 1.281566858291626,
      "learning_rate": 9.825611783665667e-05,
      "loss": 9.7229,
      "step": 336
    },
    {
      "epoch": 0.11624698171783374,
      "grad_norm": 1.4384312629699707,
      "learning_rate": 9.824139529412851e-05,
      "loss": 9.7285,
      "step": 337
    },
    {
      "epoch": 0.11659192825112108,
      "grad_norm": 1.391280174255371,
      "learning_rate": 9.822661197816823e-05,
      "loss": 9.7337,
      "step": 338
    },
    {
      "epoch": 0.11693687478440842,
      "grad_norm": 1.409494400024414,
      "learning_rate": 9.821176790739952e-05,
      "loss": 9.7505,
      "step": 339
    },
    {
      "epoch": 0.11728182131769575,
      "grad_norm": 1.260730266571045,
      "learning_rate": 9.819686310052263e-05,
      "loss": 9.7856,
      "step": 340
    },
    {
      "epoch": 0.1176267678509831,
      "grad_norm": 1.4997591972351074,
      "learning_rate": 9.818189757631429e-05,
      "loss": 9.6875,
      "step": 341
    },
    {
      "epoch": 0.11797171438427044,
      "grad_norm": 1.4658291339874268,
      "learning_rate": 9.816687135362771e-05,
      "loss": 9.7123,
      "step": 342
    },
    {
      "epoch": 0.11831666091755778,
      "grad_norm": 1.374866008758545,
      "learning_rate": 9.815178445139264e-05,
      "loss": 9.7303,
      "step": 343
    },
    {
      "epoch": 0.11866160745084511,
      "grad_norm": 1.6058346033096313,
      "learning_rate": 9.813663688861518e-05,
      "loss": 9.6672,
      "step": 344
    },
    {
      "epoch": 0.11900655398413246,
      "grad_norm": 1.5296977758407593,
      "learning_rate": 9.81214286843779e-05,
      "loss": 9.6622,
      "step": 345
    },
    {
      "epoch": 0.1193515005174198,
      "grad_norm": 1.4145252704620361,
      "learning_rate": 9.810615985783977e-05,
      "loss": 9.6812,
      "step": 346
    },
    {
      "epoch": 0.11969644705070714,
      "grad_norm": 1.5678027868270874,
      "learning_rate": 9.809083042823611e-05,
      "loss": 9.6396,
      "step": 347
    },
    {
      "epoch": 0.12004139358399447,
      "grad_norm": 1.5782277584075928,
      "learning_rate": 9.807544041487857e-05,
      "loss": 9.6638,
      "step": 348
    },
    {
      "epoch": 0.12038634011728182,
      "grad_norm": 1.6961196660995483,
      "learning_rate": 9.805998983715518e-05,
      "loss": 9.6102,
      "step": 349
    },
    {
      "epoch": 0.12073128665056916,
      "grad_norm": 1.6960567235946655,
      "learning_rate": 9.804447871453023e-05,
      "loss": 9.7084,
      "step": 350
    },
    {
      "epoch": 0.1210762331838565,
      "grad_norm": 0.730361819267273,
      "learning_rate": 9.802890706654426e-05,
      "loss": 9.884,
      "step": 351
    },
    {
      "epoch": 0.12142117971714385,
      "grad_norm": 1.076087474822998,
      "learning_rate": 9.801327491281411e-05,
      "loss": 9.7877,
      "step": 352
    },
    {
      "epoch": 0.12176612625043118,
      "grad_norm": 1.0796483755111694,
      "learning_rate": 9.79975822730328e-05,
      "loss": 9.8092,
      "step": 353
    },
    {
      "epoch": 0.12211107278371852,
      "grad_norm": 1.0125349760055542,
      "learning_rate": 9.798182916696956e-05,
      "loss": 9.7979,
      "step": 354
    },
    {
      "epoch": 0.12245601931700587,
      "grad_norm": 1.0231205224990845,
      "learning_rate": 9.796601561446983e-05,
      "loss": 9.7728,
      "step": 355
    },
    {
      "epoch": 0.12280096585029321,
      "grad_norm": 1.0823100805282593,
      "learning_rate": 9.795014163545515e-05,
      "loss": 9.7639,
      "step": 356
    },
    {
      "epoch": 0.12314591238358054,
      "grad_norm": 1.1489181518554688,
      "learning_rate": 9.793420724992321e-05,
      "loss": 9.7437,
      "step": 357
    },
    {
      "epoch": 0.12349085891686788,
      "grad_norm": 1.2228708267211914,
      "learning_rate": 9.79182124779478e-05,
      "loss": 9.7371,
      "step": 358
    },
    {
      "epoch": 0.12383580545015523,
      "grad_norm": 1.149304747581482,
      "learning_rate": 9.790215733967876e-05,
      "loss": 9.7673,
      "step": 359
    },
    {
      "epoch": 0.12418075198344257,
      "grad_norm": 1.167173147201538,
      "learning_rate": 9.7886041855342e-05,
      "loss": 9.7857,
      "step": 360
    },
    {
      "epoch": 0.12452569851672991,
      "grad_norm": 1.2003726959228516,
      "learning_rate": 9.786986604523946e-05,
      "loss": 9.7113,
      "step": 361
    },
    {
      "epoch": 0.12487064505001724,
      "grad_norm": 1.1671428680419922,
      "learning_rate": 9.785362992974904e-05,
      "loss": 9.7027,
      "step": 362
    },
    {
      "epoch": 0.1252155915833046,
      "grad_norm": 1.1911990642547607,
      "learning_rate": 9.783733352932467e-05,
      "loss": 9.783,
      "step": 363
    },
    {
      "epoch": 0.12556053811659193,
      "grad_norm": 1.1883963346481323,
      "learning_rate": 9.782097686449616e-05,
      "loss": 9.7556,
      "step": 364
    },
    {
      "epoch": 0.12590548464987927,
      "grad_norm": 1.1700190305709839,
      "learning_rate": 9.780455995586928e-05,
      "loss": 9.7978,
      "step": 365
    },
    {
      "epoch": 0.12625043118316662,
      "grad_norm": 1.2475215196609497,
      "learning_rate": 9.77880828241257e-05,
      "loss": 9.7012,
      "step": 366
    },
    {
      "epoch": 0.12659537771645396,
      "grad_norm": 1.3074865341186523,
      "learning_rate": 9.777154549002291e-05,
      "loss": 9.7239,
      "step": 367
    },
    {
      "epoch": 0.12694032424974128,
      "grad_norm": 1.0760914087295532,
      "learning_rate": 9.775494797439431e-05,
      "loss": 9.7418,
      "step": 368
    },
    {
      "epoch": 0.12728527078302862,
      "grad_norm": 1.2002527713775635,
      "learning_rate": 9.773829029814905e-05,
      "loss": 9.7252,
      "step": 369
    },
    {
      "epoch": 0.12763021731631596,
      "grad_norm": 1.2817260026931763,
      "learning_rate": 9.772157248227211e-05,
      "loss": 9.7005,
      "step": 370
    },
    {
      "epoch": 0.1279751638496033,
      "grad_norm": 1.2580500841140747,
      "learning_rate": 9.770479454782423e-05,
      "loss": 9.7851,
      "step": 371
    },
    {
      "epoch": 0.12832011038289065,
      "grad_norm": 1.1985312700271606,
      "learning_rate": 9.768795651594187e-05,
      "loss": 9.7286,
      "step": 372
    },
    {
      "epoch": 0.128665056916178,
      "grad_norm": 1.225277304649353,
      "learning_rate": 9.76710584078372e-05,
      "loss": 9.7375,
      "step": 373
    },
    {
      "epoch": 0.12901000344946534,
      "grad_norm": 1.5300991535186768,
      "learning_rate": 9.765410024479808e-05,
      "loss": 9.7107,
      "step": 374
    },
    {
      "epoch": 0.12935494998275268,
      "grad_norm": 1.3542684316635132,
      "learning_rate": 9.763708204818804e-05,
      "loss": 9.7217,
      "step": 375
    },
    {
      "epoch": 0.12969989651604003,
      "grad_norm": 1.2654287815093994,
      "learning_rate": 9.762000383944621e-05,
      "loss": 9.7224,
      "step": 376
    },
    {
      "epoch": 0.13004484304932734,
      "grad_norm": 1.3316888809204102,
      "learning_rate": 9.760286564008734e-05,
      "loss": 9.6934,
      "step": 377
    },
    {
      "epoch": 0.13038978958261468,
      "grad_norm": 1.4030239582061768,
      "learning_rate": 9.758566747170175e-05,
      "loss": 9.6717,
      "step": 378
    },
    {
      "epoch": 0.13073473611590203,
      "grad_norm": 1.372252345085144,
      "learning_rate": 9.75684093559553e-05,
      "loss": 9.6926,
      "step": 379
    },
    {
      "epoch": 0.13107968264918937,
      "grad_norm": 1.422454595565796,
      "learning_rate": 9.755109131458941e-05,
      "loss": 9.6567,
      "step": 380
    },
    {
      "epoch": 0.13142462918247672,
      "grad_norm": 1.433806300163269,
      "learning_rate": 9.753371336942092e-05,
      "loss": 9.6751,
      "step": 381
    },
    {
      "epoch": 0.13176957571576406,
      "grad_norm": 1.2416576147079468,
      "learning_rate": 9.751627554234219e-05,
      "loss": 9.7274,
      "step": 382
    },
    {
      "epoch": 0.1321145222490514,
      "grad_norm": 1.3043841123580933,
      "learning_rate": 9.749877785532103e-05,
      "loss": 9.7385,
      "step": 383
    },
    {
      "epoch": 0.13245946878233875,
      "grad_norm": 1.3732062578201294,
      "learning_rate": 9.748122033040061e-05,
      "loss": 9.6508,
      "step": 384
    },
    {
      "epoch": 0.1328044153156261,
      "grad_norm": 1.4883499145507812,
      "learning_rate": 9.746360298969951e-05,
      "loss": 9.635,
      "step": 385
    },
    {
      "epoch": 0.1331493618489134,
      "grad_norm": 1.3537193536758423,
      "learning_rate": 9.744592585541166e-05,
      "loss": 9.6717,
      "step": 386
    },
    {
      "epoch": 0.13349430838220075,
      "grad_norm": 1.470834732055664,
      "learning_rate": 9.742818894980634e-05,
      "loss": 9.6307,
      "step": 387
    },
    {
      "epoch": 0.1338392549154881,
      "grad_norm": 1.3867615461349487,
      "learning_rate": 9.741039229522809e-05,
      "loss": 9.6048,
      "step": 388
    },
    {
      "epoch": 0.13418420144877544,
      "grad_norm": 1.3195267915725708,
      "learning_rate": 9.739253591409673e-05,
      "loss": 9.6624,
      "step": 389
    },
    {
      "epoch": 0.13452914798206278,
      "grad_norm": 1.3774893283843994,
      "learning_rate": 9.737461982890734e-05,
      "loss": 9.63,
      "step": 390
    },
    {
      "epoch": 0.13487409451535012,
      "grad_norm": 1.541316270828247,
      "learning_rate": 9.735664406223022e-05,
      "loss": 9.6267,
      "step": 391
    },
    {
      "epoch": 0.13521904104863747,
      "grad_norm": 1.6289477348327637,
      "learning_rate": 9.733860863671081e-05,
      "loss": 9.6038,
      "step": 392
    },
    {
      "epoch": 0.1355639875819248,
      "grad_norm": 1.5535857677459717,
      "learning_rate": 9.732051357506974e-05,
      "loss": 9.6579,
      "step": 393
    },
    {
      "epoch": 0.13590893411521215,
      "grad_norm": 1.57676362991333,
      "learning_rate": 9.730235890010279e-05,
      "loss": 9.5814,
      "step": 394
    },
    {
      "epoch": 0.13625388064849947,
      "grad_norm": 1.4913179874420166,
      "learning_rate": 9.728414463468079e-05,
      "loss": 9.609,
      "step": 395
    },
    {
      "epoch": 0.1365988271817868,
      "grad_norm": 1.6802784204483032,
      "learning_rate": 9.726587080174968e-05,
      "loss": 9.5745,
      "step": 396
    },
    {
      "epoch": 0.13694377371507416,
      "grad_norm": 1.4532045125961304,
      "learning_rate": 9.724753742433042e-05,
      "loss": 9.688,
      "step": 397
    },
    {
      "epoch": 0.1372887202483615,
      "grad_norm": 1.6262314319610596,
      "learning_rate": 9.7229144525519e-05,
      "loss": 9.5843,
      "step": 398
    },
    {
      "epoch": 0.13763366678164884,
      "grad_norm": 1.7181068658828735,
      "learning_rate": 9.721069212848635e-05,
      "loss": 9.6142,
      "step": 399
    },
    {
      "epoch": 0.1379786133149362,
      "grad_norm": 1.8162275552749634,
      "learning_rate": 9.71921802564784e-05,
      "loss": 9.5156,
      "step": 400
    },
    {
      "epoch": 0.13832355984822353,
      "grad_norm": 0.8729615807533264,
      "learning_rate": 9.717360893281602e-05,
      "loss": 9.7578,
      "step": 401
    },
    {
      "epoch": 0.13866850638151088,
      "grad_norm": 1.138515591621399,
      "learning_rate": 9.715497818089492e-05,
      "loss": 9.6906,
      "step": 402
    },
    {
      "epoch": 0.13901345291479822,
      "grad_norm": 1.0682348012924194,
      "learning_rate": 9.713628802418573e-05,
      "loss": 9.7469,
      "step": 403
    },
    {
      "epoch": 0.13935839944808553,
      "grad_norm": 1.0238516330718994,
      "learning_rate": 9.711753848623389e-05,
      "loss": 9.7235,
      "step": 404
    },
    {
      "epoch": 0.13970334598137288,
      "grad_norm": 1.046028733253479,
      "learning_rate": 9.70987295906596e-05,
      "loss": 9.6959,
      "step": 405
    },
    {
      "epoch": 0.14004829251466022,
      "grad_norm": 1.1170198917388916,
      "learning_rate": 9.707986136115795e-05,
      "loss": 9.7301,
      "step": 406
    },
    {
      "epoch": 0.14039323904794757,
      "grad_norm": 1.259394645690918,
      "learning_rate": 9.706093382149868e-05,
      "loss": 9.6711,
      "step": 407
    },
    {
      "epoch": 0.1407381855812349,
      "grad_norm": 1.0471652746200562,
      "learning_rate": 9.704194699552628e-05,
      "loss": 9.6939,
      "step": 408
    },
    {
      "epoch": 0.14108313211452225,
      "grad_norm": 1.1050465106964111,
      "learning_rate": 9.702290090715996e-05,
      "loss": 9.7339,
      "step": 409
    },
    {
      "epoch": 0.1414280786478096,
      "grad_norm": 1.2006595134735107,
      "learning_rate": 9.700379558039351e-05,
      "loss": 9.6702,
      "step": 410
    },
    {
      "epoch": 0.14177302518109694,
      "grad_norm": 1.2607287168502808,
      "learning_rate": 9.698463103929542e-05,
      "loss": 9.6611,
      "step": 411
    },
    {
      "epoch": 0.14211797171438428,
      "grad_norm": 1.1868005990982056,
      "learning_rate": 9.696540730800876e-05,
      "loss": 9.7249,
      "step": 412
    },
    {
      "epoch": 0.1424629182476716,
      "grad_norm": 1.2063498497009277,
      "learning_rate": 9.694612441075115e-05,
      "loss": 9.6765,
      "step": 413
    },
    {
      "epoch": 0.14280786478095894,
      "grad_norm": 1.2330728769302368,
      "learning_rate": 9.692678237181474e-05,
      "loss": 9.7045,
      "step": 414
    },
    {
      "epoch": 0.1431528113142463,
      "grad_norm": 1.1720654964447021,
      "learning_rate": 9.690738121556622e-05,
      "loss": 9.6821,
      "step": 415
    },
    {
      "epoch": 0.14349775784753363,
      "grad_norm": 1.257949948310852,
      "learning_rate": 9.68879209664467e-05,
      "loss": 9.7019,
      "step": 416
    },
    {
      "epoch": 0.14384270438082097,
      "grad_norm": 1.2828700542449951,
      "learning_rate": 9.686840164897181e-05,
      "loss": 9.6634,
      "step": 417
    },
    {
      "epoch": 0.14418765091410832,
      "grad_norm": 1.3583167791366577,
      "learning_rate": 9.684882328773153e-05,
      "loss": 9.6357,
      "step": 418
    },
    {
      "epoch": 0.14453259744739566,
      "grad_norm": 1.24040687084198,
      "learning_rate": 9.682918590739022e-05,
      "loss": 9.7175,
      "step": 419
    },
    {
      "epoch": 0.144877543980683,
      "grad_norm": 1.2623732089996338,
      "learning_rate": 9.680948953268665e-05,
      "loss": 9.6945,
      "step": 420
    },
    {
      "epoch": 0.14522249051397035,
      "grad_norm": 1.2432429790496826,
      "learning_rate": 9.678973418843384e-05,
      "loss": 9.651,
      "step": 421
    },
    {
      "epoch": 0.14556743704725766,
      "grad_norm": 1.1109862327575684,
      "learning_rate": 9.676991989951917e-05,
      "loss": 9.6767,
      "step": 422
    },
    {
      "epoch": 0.145912383580545,
      "grad_norm": 1.350054383277893,
      "learning_rate": 9.67500466909042e-05,
      "loss": 9.6212,
      "step": 423
    },
    {
      "epoch": 0.14625733011383235,
      "grad_norm": 1.3729373216629028,
      "learning_rate": 9.673011458762479e-05,
      "loss": 9.6504,
      "step": 424
    },
    {
      "epoch": 0.1466022766471197,
      "grad_norm": 1.172640323638916,
      "learning_rate": 9.671012361479093e-05,
      "loss": 9.7148,
      "step": 425
    },
    {
      "epoch": 0.14694722318040704,
      "grad_norm": 1.4439337253570557,
      "learning_rate": 9.66900737975868e-05,
      "loss": 9.6306,
      "step": 426
    },
    {
      "epoch": 0.14729216971369438,
      "grad_norm": 1.2247321605682373,
      "learning_rate": 9.666996516127073e-05,
      "loss": 9.6421,
      "step": 427
    },
    {
      "epoch": 0.14763711624698173,
      "grad_norm": 1.4057327508926392,
      "learning_rate": 9.664979773117513e-05,
      "loss": 9.5413,
      "step": 428
    },
    {
      "epoch": 0.14798206278026907,
      "grad_norm": 1.3298213481903076,
      "learning_rate": 9.662957153270647e-05,
      "loss": 9.628,
      "step": 429
    },
    {
      "epoch": 0.14832700931355638,
      "grad_norm": 1.330397367477417,
      "learning_rate": 9.660928659134526e-05,
      "loss": 9.6531,
      "step": 430
    },
    {
      "epoch": 0.14867195584684373,
      "grad_norm": 1.2601134777069092,
      "learning_rate": 9.6588942932646e-05,
      "loss": 9.6714,
      "step": 431
    },
    {
      "epoch": 0.14901690238013107,
      "grad_norm": 1.3315186500549316,
      "learning_rate": 9.656854058223722e-05,
      "loss": 9.6103,
      "step": 432
    },
    {
      "epoch": 0.14936184891341842,
      "grad_norm": 1.4878896474838257,
      "learning_rate": 9.654807956582131e-05,
      "loss": 9.5996,
      "step": 433
    },
    {
      "epoch": 0.14970679544670576,
      "grad_norm": 1.3564176559448242,
      "learning_rate": 9.652755990917463e-05,
      "loss": 9.6551,
      "step": 434
    },
    {
      "epoch": 0.1500517419799931,
      "grad_norm": 1.5410569906234741,
      "learning_rate": 9.650698163814736e-05,
      "loss": 9.5719,
      "step": 435
    },
    {
      "epoch": 0.15039668851328045,
      "grad_norm": 1.3284558057785034,
      "learning_rate": 9.648634477866354e-05,
      "loss": 9.6312,
      "step": 436
    },
    {
      "epoch": 0.1507416350465678,
      "grad_norm": 1.3244214057922363,
      "learning_rate": 9.646564935672107e-05,
      "loss": 9.6812,
      "step": 437
    },
    {
      "epoch": 0.15108658157985513,
      "grad_norm": 1.446029782295227,
      "learning_rate": 9.644489539839153e-05,
      "loss": 9.6196,
      "step": 438
    },
    {
      "epoch": 0.15143152811314245,
      "grad_norm": 1.6580462455749512,
      "learning_rate": 9.64240829298203e-05,
      "loss": 9.519,
      "step": 439
    },
    {
      "epoch": 0.1517764746464298,
      "grad_norm": 1.436366319656372,
      "learning_rate": 9.64032119772265e-05,
      "loss": 9.6082,
      "step": 440
    },
    {
      "epoch": 0.15212142117971714,
      "grad_norm": 1.4805681705474854,
      "learning_rate": 9.638228256690285e-05,
      "loss": 9.6322,
      "step": 441
    },
    {
      "epoch": 0.15246636771300448,
      "grad_norm": 1.4746217727661133,
      "learning_rate": 9.636129472521577e-05,
      "loss": 9.6165,
      "step": 442
    },
    {
      "epoch": 0.15281131424629182,
      "grad_norm": 1.4036329984664917,
      "learning_rate": 9.634024847860527e-05,
      "loss": 9.6314,
      "step": 443
    },
    {
      "epoch": 0.15315626077957917,
      "grad_norm": 1.4564642906188965,
      "learning_rate": 9.631914385358495e-05,
      "loss": 9.5594,
      "step": 444
    },
    {
      "epoch": 0.1535012073128665,
      "grad_norm": 1.628414511680603,
      "learning_rate": 9.629798087674195e-05,
      "loss": 9.5453,
      "step": 445
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 1.4203163385391235,
      "learning_rate": 9.627675957473689e-05,
      "loss": 9.5673,
      "step": 446
    },
    {
      "epoch": 0.1541911003794412,
      "grad_norm": 1.4284478425979614,
      "learning_rate": 9.62554799743039e-05,
      "loss": 9.6228,
      "step": 447
    },
    {
      "epoch": 0.15453604691272851,
      "grad_norm": 1.5529612302780151,
      "learning_rate": 9.623414210225057e-05,
      "loss": 9.5866,
      "step": 448
    },
    {
      "epoch": 0.15488099344601586,
      "grad_norm": 1.741105556488037,
      "learning_rate": 9.621274598545785e-05,
      "loss": 9.5087,
      "step": 449
    },
    {
      "epoch": 0.1552259399793032,
      "grad_norm": 1.8047065734863281,
      "learning_rate": 9.61912916508801e-05,
      "loss": 9.5244,
      "step": 450
    },
    {
      "epoch": 0.15557088651259054,
      "grad_norm": 1.0891846418380737,
      "learning_rate": 9.6169779125545e-05,
      "loss": 9.6529,
      "step": 451
    },
    {
      "epoch": 0.1559158330458779,
      "grad_norm": 1.021998643875122,
      "learning_rate": 9.614820843655357e-05,
      "loss": 9.7033,
      "step": 452
    },
    {
      "epoch": 0.15626077957916523,
      "grad_norm": 1.0237458944320679,
      "learning_rate": 9.612657961108005e-05,
      "loss": 9.6524,
      "step": 453
    },
    {
      "epoch": 0.15660572611245258,
      "grad_norm": 1.085534691810608,
      "learning_rate": 9.610489267637199e-05,
      "loss": 9.6985,
      "step": 454
    },
    {
      "epoch": 0.15695067264573992,
      "grad_norm": 1.121556043624878,
      "learning_rate": 9.608314765975008e-05,
      "loss": 9.6726,
      "step": 455
    },
    {
      "epoch": 0.15729561917902726,
      "grad_norm": 1.1756961345672607,
      "learning_rate": 9.60613445886082e-05,
      "loss": 9.611,
      "step": 456
    },
    {
      "epoch": 0.15764056571231458,
      "grad_norm": 1.1245508193969727,
      "learning_rate": 9.60394834904134e-05,
      "loss": 9.6513,
      "step": 457
    },
    {
      "epoch": 0.15798551224560192,
      "grad_norm": 1.138670802116394,
      "learning_rate": 9.601756439270579e-05,
      "loss": 9.6294,
      "step": 458
    },
    {
      "epoch": 0.15833045877888927,
      "grad_norm": 1.0786954164505005,
      "learning_rate": 9.599558732309854e-05,
      "loss": 9.6677,
      "step": 459
    },
    {
      "epoch": 0.1586754053121766,
      "grad_norm": 1.1835142374038696,
      "learning_rate": 9.597355230927789e-05,
      "loss": 9.6656,
      "step": 460
    },
    {
      "epoch": 0.15902035184546395,
      "grad_norm": 1.2156816720962524,
      "learning_rate": 9.595145937900308e-05,
      "loss": 9.6536,
      "step": 461
    },
    {
      "epoch": 0.1593652983787513,
      "grad_norm": 1.2232426404953003,
      "learning_rate": 9.592930856010624e-05,
      "loss": 9.6273,
      "step": 462
    },
    {
      "epoch": 0.15971024491203864,
      "grad_norm": 1.159227967262268,
      "learning_rate": 9.590709988049251e-05,
      "loss": 9.6486,
      "step": 463
    },
    {
      "epoch": 0.16005519144532598,
      "grad_norm": 1.1979292631149292,
      "learning_rate": 9.58848333681399e-05,
      "loss": 9.6752,
      "step": 464
    },
    {
      "epoch": 0.16040013797861333,
      "grad_norm": 1.162811279296875,
      "learning_rate": 9.586250905109922e-05,
      "loss": 9.6441,
      "step": 465
    },
    {
      "epoch": 0.16074508451190064,
      "grad_norm": 1.238519549369812,
      "learning_rate": 9.584012695749418e-05,
      "loss": 9.6415,
      "step": 466
    },
    {
      "epoch": 0.161090031045188,
      "grad_norm": 1.3711762428283691,
      "learning_rate": 9.581768711552125e-05,
      "loss": 9.6418,
      "step": 467
    },
    {
      "epoch": 0.16143497757847533,
      "grad_norm": 1.235273838043213,
      "learning_rate": 9.579518955344961e-05,
      "loss": 9.6014,
      "step": 468
    },
    {
      "epoch": 0.16177992411176267,
      "grad_norm": 1.1799870729446411,
      "learning_rate": 9.57726342996212e-05,
      "loss": 9.664,
      "step": 469
    },
    {
      "epoch": 0.16212487064505002,
      "grad_norm": 1.305772066116333,
      "learning_rate": 9.575002138245064e-05,
      "loss": 9.6004,
      "step": 470
    },
    {
      "epoch": 0.16246981717833736,
      "grad_norm": 1.2880913019180298,
      "learning_rate": 9.572735083042516e-05,
      "loss": 9.6829,
      "step": 471
    },
    {
      "epoch": 0.1628147637116247,
      "grad_norm": 1.3265849351882935,
      "learning_rate": 9.570462267210461e-05,
      "loss": 9.6106,
      "step": 472
    },
    {
      "epoch": 0.16315971024491205,
      "grad_norm": 1.3995447158813477,
      "learning_rate": 9.568183693612143e-05,
      "loss": 9.6109,
      "step": 473
    },
    {
      "epoch": 0.1635046567781994,
      "grad_norm": 1.2574628591537476,
      "learning_rate": 9.565899365118058e-05,
      "loss": 9.6015,
      "step": 474
    },
    {
      "epoch": 0.1638496033114867,
      "grad_norm": 1.3726155757904053,
      "learning_rate": 9.563609284605953e-05,
      "loss": 9.5894,
      "step": 475
    },
    {
      "epoch": 0.16419454984477405,
      "grad_norm": 1.3234082460403442,
      "learning_rate": 9.561313454960818e-05,
      "loss": 9.5645,
      "step": 476
    },
    {
      "epoch": 0.1645394963780614,
      "grad_norm": 1.2199691534042358,
      "learning_rate": 9.559011879074888e-05,
      "loss": 9.6112,
      "step": 477
    },
    {
      "epoch": 0.16488444291134874,
      "grad_norm": 1.2691389322280884,
      "learning_rate": 9.556704559847639e-05,
      "loss": 9.6231,
      "step": 478
    },
    {
      "epoch": 0.16522938944463608,
      "grad_norm": 1.3256593942642212,
      "learning_rate": 9.554391500185777e-05,
      "loss": 9.6126,
      "step": 479
    },
    {
      "epoch": 0.16557433597792343,
      "grad_norm": 1.511549949645996,
      "learning_rate": 9.552072703003247e-05,
      "loss": 9.6111,
      "step": 480
    },
    {
      "epoch": 0.16591928251121077,
      "grad_norm": 1.3752213716506958,
      "learning_rate": 9.549748171221215e-05,
      "loss": 9.6003,
      "step": 481
    },
    {
      "epoch": 0.1662642290444981,
      "grad_norm": 1.3873335123062134,
      "learning_rate": 9.547417907768075e-05,
      "loss": 9.5684,
      "step": 482
    },
    {
      "epoch": 0.16660917557778546,
      "grad_norm": 1.3659286499023438,
      "learning_rate": 9.545081915579441e-05,
      "loss": 9.6035,
      "step": 483
    },
    {
      "epoch": 0.16695412211107277,
      "grad_norm": 1.575773000717163,
      "learning_rate": 9.542740197598144e-05,
      "loss": 9.6505,
      "step": 484
    },
    {
      "epoch": 0.16729906864436012,
      "grad_norm": 1.3886533975601196,
      "learning_rate": 9.540392756774225e-05,
      "loss": 9.5501,
      "step": 485
    },
    {
      "epoch": 0.16764401517764746,
      "grad_norm": 1.3898117542266846,
      "learning_rate": 9.538039596064943e-05,
      "loss": 9.6042,
      "step": 486
    },
    {
      "epoch": 0.1679889617109348,
      "grad_norm": 1.3748157024383545,
      "learning_rate": 9.535680718434755e-05,
      "loss": 9.5593,
      "step": 487
    },
    {
      "epoch": 0.16833390824422215,
      "grad_norm": 1.4047949314117432,
      "learning_rate": 9.533316126855319e-05,
      "loss": 9.5037,
      "step": 488
    },
    {
      "epoch": 0.1686788547775095,
      "grad_norm": 1.3030369281768799,
      "learning_rate": 9.530945824305498e-05,
      "loss": 9.5716,
      "step": 489
    },
    {
      "epoch": 0.16902380131079683,
      "grad_norm": 1.6138426065444946,
      "learning_rate": 9.528569813771346e-05,
      "loss": 9.4794,
      "step": 490
    },
    {
      "epoch": 0.16936874784408418,
      "grad_norm": 1.4896208047866821,
      "learning_rate": 9.526188098246108e-05,
      "loss": 9.5509,
      "step": 491
    },
    {
      "epoch": 0.16971369437737152,
      "grad_norm": 1.4523698091506958,
      "learning_rate": 9.523800680730214e-05,
      "loss": 9.5225,
      "step": 492
    },
    {
      "epoch": 0.17005864091065884,
      "grad_norm": 1.3837906122207642,
      "learning_rate": 9.521407564231283e-05,
      "loss": 9.5907,
      "step": 493
    },
    {
      "epoch": 0.17040358744394618,
      "grad_norm": 1.4787929058074951,
      "learning_rate": 9.519008751764107e-05,
      "loss": 9.4969,
      "step": 494
    },
    {
      "epoch": 0.17074853397723352,
      "grad_norm": 1.6119440793991089,
      "learning_rate": 9.516604246350658e-05,
      "loss": 9.4936,
      "step": 495
    },
    {
      "epoch": 0.17109348051052087,
      "grad_norm": 1.397890329360962,
      "learning_rate": 9.514194051020074e-05,
      "loss": 9.533,
      "step": 496
    },
    {
      "epoch": 0.1714384270438082,
      "grad_norm": 1.505786657333374,
      "learning_rate": 9.51177816880867e-05,
      "loss": 9.5489,
      "step": 497
    },
    {
      "epoch": 0.17178337357709555,
      "grad_norm": 1.5232138633728027,
      "learning_rate": 9.50935660275992e-05,
      "loss": 9.499,
      "step": 498
    },
    {
      "epoch": 0.1721283201103829,
      "grad_norm": 1.6428580284118652,
      "learning_rate": 9.506929355924456e-05,
      "loss": 9.4986,
      "step": 499
    },
    {
      "epoch": 0.17247326664367024,
      "grad_norm": 1.7673543691635132,
      "learning_rate": 9.504496431360071e-05,
      "loss": 9.4835,
      "step": 500
    },
    {
      "epoch": 0.17281821317695756,
      "grad_norm": 0.8733913898468018,
      "learning_rate": 9.50205783213171e-05,
      "loss": 9.6679,
      "step": 501
    },
    {
      "epoch": 0.1731631597102449,
      "grad_norm": 0.9659939408302307,
      "learning_rate": 9.499613561311465e-05,
      "loss": 9.647,
      "step": 502
    },
    {
      "epoch": 0.17350810624353225,
      "grad_norm": 1.0745805501937866,
      "learning_rate": 9.497163621978571e-05,
      "loss": 9.6277,
      "step": 503
    },
    {
      "epoch": 0.1738530527768196,
      "grad_norm": 1.1018327474594116,
      "learning_rate": 9.494708017219412e-05,
      "loss": 9.6958,
      "step": 504
    },
    {
      "epoch": 0.17419799931010693,
      "grad_norm": 0.9950918555259705,
      "learning_rate": 9.492246750127501e-05,
      "loss": 9.6551,
      "step": 505
    },
    {
      "epoch": 0.17454294584339428,
      "grad_norm": 1.1155328750610352,
      "learning_rate": 9.489779823803486e-05,
      "loss": 9.5963,
      "step": 506
    },
    {
      "epoch": 0.17488789237668162,
      "grad_norm": 1.182844877243042,
      "learning_rate": 9.487307241355148e-05,
      "loss": 9.6293,
      "step": 507
    },
    {
      "epoch": 0.17523283890996896,
      "grad_norm": 1.0601288080215454,
      "learning_rate": 9.484829005897392e-05,
      "loss": 9.6011,
      "step": 508
    },
    {
      "epoch": 0.1755777854432563,
      "grad_norm": 1.0896538496017456,
      "learning_rate": 9.482345120552242e-05,
      "loss": 9.6161,
      "step": 509
    },
    {
      "epoch": 0.17592273197654362,
      "grad_norm": 1.1434667110443115,
      "learning_rate": 9.47985558844884e-05,
      "loss": 9.6463,
      "step": 510
    },
    {
      "epoch": 0.17626767850983097,
      "grad_norm": 1.1508530378341675,
      "learning_rate": 9.477360412723447e-05,
      "loss": 9.6287,
      "step": 511
    },
    {
      "epoch": 0.1766126250431183,
      "grad_norm": 1.3060308694839478,
      "learning_rate": 9.474859596519424e-05,
      "loss": 9.59,
      "step": 512
    },
    {
      "epoch": 0.17695757157640565,
      "grad_norm": 1.1819398403167725,
      "learning_rate": 9.472353142987247e-05,
      "loss": 9.6123,
      "step": 513
    },
    {
      "epoch": 0.177302518109693,
      "grad_norm": 1.1345469951629639,
      "learning_rate": 9.46984105528449e-05,
      "loss": 9.5987,
      "step": 514
    },
    {
      "epoch": 0.17764746464298034,
      "grad_norm": 1.1536154747009277,
      "learning_rate": 9.467323336575826e-05,
      "loss": 9.5657,
      "step": 515
    },
    {
      "epoch": 0.17799241117626768,
      "grad_norm": 1.1551100015640259,
      "learning_rate": 9.464799990033018e-05,
      "loss": 9.6147,
      "step": 516
    },
    {
      "epoch": 0.17833735770955503,
      "grad_norm": 1.2950499057769775,
      "learning_rate": 9.462271018834923e-05,
      "loss": 9.5483,
      "step": 517
    },
    {
      "epoch": 0.17868230424284237,
      "grad_norm": 1.210809350013733,
      "learning_rate": 9.459736426167482e-05,
      "loss": 9.6072,
      "step": 518
    },
    {
      "epoch": 0.1790272507761297,
      "grad_norm": 1.1476471424102783,
      "learning_rate": 9.457196215223721e-05,
      "loss": 9.6183,
      "step": 519
    },
    {
      "epoch": 0.17937219730941703,
      "grad_norm": 1.2517257928848267,
      "learning_rate": 9.454650389203735e-05,
      "loss": 9.5937,
      "step": 520
    },
    {
      "epoch": 0.17971714384270437,
      "grad_norm": 1.2335177659988403,
      "learning_rate": 9.452098951314704e-05,
      "loss": 9.6255,
      "step": 521
    },
    {
      "epoch": 0.18006209037599172,
      "grad_norm": 1.3397551774978638,
      "learning_rate": 9.44954190477087e-05,
      "loss": 9.5745,
      "step": 522
    },
    {
      "epoch": 0.18040703690927906,
      "grad_norm": 1.2667393684387207,
      "learning_rate": 9.446979252793547e-05,
      "loss": 9.5642,
      "step": 523
    },
    {
      "epoch": 0.1807519834425664,
      "grad_norm": 1.269653081893921,
      "learning_rate": 9.444410998611099e-05,
      "loss": 9.5399,
      "step": 524
    },
    {
      "epoch": 0.18109692997585375,
      "grad_norm": 1.3198039531707764,
      "learning_rate": 9.441837145458962e-05,
      "loss": 9.5648,
      "step": 525
    },
    {
      "epoch": 0.1814418765091411,
      "grad_norm": 1.27424156665802,
      "learning_rate": 9.439257696579617e-05,
      "loss": 9.5742,
      "step": 526
    },
    {
      "epoch": 0.18178682304242844,
      "grad_norm": 1.2845133543014526,
      "learning_rate": 9.436672655222596e-05,
      "loss": 9.5993,
      "step": 527
    },
    {
      "epoch": 0.18213176957571575,
      "grad_norm": 1.2665520906448364,
      "learning_rate": 9.434082024644476e-05,
      "loss": 9.5995,
      "step": 528
    },
    {
      "epoch": 0.1824767161090031,
      "grad_norm": 1.2851994037628174,
      "learning_rate": 9.431485808108875e-05,
      "loss": 9.5827,
      "step": 529
    },
    {
      "epoch": 0.18282166264229044,
      "grad_norm": 1.4341552257537842,
      "learning_rate": 9.42888400888645e-05,
      "loss": 9.5422,
      "step": 530
    },
    {
      "epoch": 0.18316660917557778,
      "grad_norm": 1.3948688507080078,
      "learning_rate": 9.426276630254887e-05,
      "loss": 9.5377,
      "step": 531
    },
    {
      "epoch": 0.18351155570886513,
      "grad_norm": 1.3061037063598633,
      "learning_rate": 9.423663675498907e-05,
      "loss": 9.5961,
      "step": 532
    },
    {
      "epoch": 0.18385650224215247,
      "grad_norm": 1.6419034004211426,
      "learning_rate": 9.421045147910248e-05,
      "loss": 9.4838,
      "step": 533
    },
    {
      "epoch": 0.1842014487754398,
      "grad_norm": 1.4868634939193726,
      "learning_rate": 9.418421050787675e-05,
      "loss": 9.4782,
      "step": 534
    },
    {
      "epoch": 0.18454639530872716,
      "grad_norm": 1.3961265087127686,
      "learning_rate": 9.415791387436968e-05,
      "loss": 9.5505,
      "step": 535
    },
    {
      "epoch": 0.1848913418420145,
      "grad_norm": 1.2420654296875,
      "learning_rate": 9.413156161170913e-05,
      "loss": 9.5702,
      "step": 536
    },
    {
      "epoch": 0.18523628837530182,
      "grad_norm": 1.4459741115570068,
      "learning_rate": 9.410515375309316e-05,
      "loss": 9.5742,
      "step": 537
    },
    {
      "epoch": 0.18558123490858916,
      "grad_norm": 1.3767470121383667,
      "learning_rate": 9.407869033178978e-05,
      "loss": 9.5521,
      "step": 538
    },
    {
      "epoch": 0.1859261814418765,
      "grad_norm": 1.399837851524353,
      "learning_rate": 9.405217138113698e-05,
      "loss": 9.49,
      "step": 539
    },
    {
      "epoch": 0.18627112797516385,
      "grad_norm": 1.2332874536514282,
      "learning_rate": 9.402559693454278e-05,
      "loss": 9.5674,
      "step": 540
    },
    {
      "epoch": 0.1866160745084512,
      "grad_norm": 1.266040325164795,
      "learning_rate": 9.399896702548506e-05,
      "loss": 9.6078,
      "step": 541
    },
    {
      "epoch": 0.18696102104173853,
      "grad_norm": 1.3650012016296387,
      "learning_rate": 9.397228168751159e-05,
      "loss": 9.5375,
      "step": 542
    },
    {
      "epoch": 0.18730596757502588,
      "grad_norm": 1.4342894554138184,
      "learning_rate": 9.394554095423995e-05,
      "loss": 9.517,
      "step": 543
    },
    {
      "epoch": 0.18765091410831322,
      "grad_norm": 1.5103857517242432,
      "learning_rate": 9.391874485935753e-05,
      "loss": 9.4914,
      "step": 544
    },
    {
      "epoch": 0.18799586064160057,
      "grad_norm": 1.5459791421890259,
      "learning_rate": 9.389189343662146e-05,
      "loss": 9.5011,
      "step": 545
    },
    {
      "epoch": 0.18834080717488788,
      "grad_norm": 1.4449481964111328,
      "learning_rate": 9.386498671985852e-05,
      "loss": 9.4861,
      "step": 546
    },
    {
      "epoch": 0.18868575370817522,
      "grad_norm": 1.6004838943481445,
      "learning_rate": 9.383802474296526e-05,
      "loss": 9.4491,
      "step": 547
    },
    {
      "epoch": 0.18903070024146257,
      "grad_norm": 1.5383483171463013,
      "learning_rate": 9.381100753990773e-05,
      "loss": 9.5136,
      "step": 548
    },
    {
      "epoch": 0.1893756467747499,
      "grad_norm": 1.8076226711273193,
      "learning_rate": 9.37839351447216e-05,
      "loss": 9.4498,
      "step": 549
    },
    {
      "epoch": 0.18972059330803726,
      "grad_norm": 1.745660662651062,
      "learning_rate": 9.375680759151206e-05,
      "loss": 9.4381,
      "step": 550
    },
    {
      "epoch": 0.1900655398413246,
      "grad_norm": 0.9159271717071533,
      "learning_rate": 9.372962491445384e-05,
      "loss": 9.6563,
      "step": 551
    },
    {
      "epoch": 0.19041048637461194,
      "grad_norm": 0.9410980343818665,
      "learning_rate": 9.370238714779103e-05,
      "loss": 9.6111,
      "step": 552
    },
    {
      "epoch": 0.19075543290789929,
      "grad_norm": 1.0490546226501465,
      "learning_rate": 9.367509432583717e-05,
      "loss": 9.5987,
      "step": 553
    },
    {
      "epoch": 0.19110037944118663,
      "grad_norm": 1.0307831764221191,
      "learning_rate": 9.364774648297514e-05,
      "loss": 9.6333,
      "step": 554
    },
    {
      "epoch": 0.19144532597447395,
      "grad_norm": 1.148796796798706,
      "learning_rate": 9.362034365365717e-05,
      "loss": 9.6337,
      "step": 555
    },
    {
      "epoch": 0.1917902725077613,
      "grad_norm": 1.0962300300598145,
      "learning_rate": 9.35928858724047e-05,
      "loss": 9.605,
      "step": 556
    },
    {
      "epoch": 0.19213521904104863,
      "grad_norm": 1.1664516925811768,
      "learning_rate": 9.356537317380844e-05,
      "loss": 9.6228,
      "step": 557
    },
    {
      "epoch": 0.19248016557433598,
      "grad_norm": 1.0752300024032593,
      "learning_rate": 9.353780559252827e-05,
      "loss": 9.6528,
      "step": 558
    },
    {
      "epoch": 0.19282511210762332,
      "grad_norm": 1.3476428985595703,
      "learning_rate": 9.351018316329323e-05,
      "loss": 9.5958,
      "step": 559
    },
    {
      "epoch": 0.19317005864091066,
      "grad_norm": 1.0217633247375488,
      "learning_rate": 9.348250592090142e-05,
      "loss": 9.6622,
      "step": 560
    },
    {
      "epoch": 0.193515005174198,
      "grad_norm": 1.2857102155685425,
      "learning_rate": 9.345477390022002e-05,
      "loss": 9.5979,
      "step": 561
    },
    {
      "epoch": 0.19385995170748535,
      "grad_norm": 1.2263622283935547,
      "learning_rate": 9.342698713618521e-05,
      "loss": 9.5967,
      "step": 562
    },
    {
      "epoch": 0.19420489824077267,
      "grad_norm": 1.1881204843521118,
      "learning_rate": 9.339914566380212e-05,
      "loss": 9.5559,
      "step": 563
    },
    {
      "epoch": 0.19454984477406,
      "grad_norm": 1.288617730140686,
      "learning_rate": 9.337124951814485e-05,
      "loss": 9.5391,
      "step": 564
    },
    {
      "epoch": 0.19489479130734735,
      "grad_norm": 1.1224907636642456,
      "learning_rate": 9.33432987343563e-05,
      "loss": 9.59,
      "step": 565
    },
    {
      "epoch": 0.1952397378406347,
      "grad_norm": 1.3285510540008545,
      "learning_rate": 9.331529334764827e-05,
      "loss": 9.5726,
      "step": 566
    },
    {
      "epoch": 0.19558468437392204,
      "grad_norm": 1.209932804107666,
      "learning_rate": 9.328723339330132e-05,
      "loss": 9.585,
      "step": 567
    },
    {
      "epoch": 0.19592963090720938,
      "grad_norm": 1.1799026727676392,
      "learning_rate": 9.325911890666473e-05,
      "loss": 9.5934,
      "step": 568
    },
    {
      "epoch": 0.19627457744049673,
      "grad_norm": 1.3268376588821411,
      "learning_rate": 9.323094992315653e-05,
      "loss": 9.5686,
      "step": 569
    },
    {
      "epoch": 0.19661952397378407,
      "grad_norm": 1.270527958869934,
      "learning_rate": 9.320272647826336e-05,
      "loss": 9.5149,
      "step": 570
    },
    {
      "epoch": 0.19696447050707142,
      "grad_norm": 1.3913733959197998,
      "learning_rate": 9.31744486075405e-05,
      "loss": 9.5098,
      "step": 571
    },
    {
      "epoch": 0.19730941704035873,
      "grad_norm": 1.1947466135025024,
      "learning_rate": 9.314611634661177e-05,
      "loss": 9.6244,
      "step": 572
    },
    {
      "epoch": 0.19765436357364607,
      "grad_norm": 1.284228801727295,
      "learning_rate": 9.31177297311695e-05,
      "loss": 9.5676,
      "step": 573
    },
    {
      "epoch": 0.19799931010693342,
      "grad_norm": 1.313984751701355,
      "learning_rate": 9.308928879697456e-05,
      "loss": 9.5131,
      "step": 574
    },
    {
      "epoch": 0.19834425664022076,
      "grad_norm": 1.3125852346420288,
      "learning_rate": 9.306079357985616e-05,
      "loss": 9.5441,
      "step": 575
    },
    {
      "epoch": 0.1986892031735081,
      "grad_norm": 1.3541791439056396,
      "learning_rate": 9.303224411571195e-05,
      "loss": 9.506,
      "step": 576
    },
    {
      "epoch": 0.19903414970679545,
      "grad_norm": 1.4502359628677368,
      "learning_rate": 9.300364044050794e-05,
      "loss": 9.5573,
      "step": 577
    },
    {
      "epoch": 0.1993790962400828,
      "grad_norm": 1.2596427202224731,
      "learning_rate": 9.297498259027834e-05,
      "loss": 9.5027,
      "step": 578
    },
    {
      "epoch": 0.19972404277337014,
      "grad_norm": 1.1951390504837036,
      "learning_rate": 9.294627060112572e-05,
      "loss": 9.5988,
      "step": 579
    },
    {
      "epoch": 0.20006898930665748,
      "grad_norm": 1.3697572946548462,
      "learning_rate": 9.291750450922078e-05,
      "loss": 9.5958,
      "step": 580
    },
    {
      "epoch": 0.2004139358399448,
      "grad_norm": 1.2890172004699707,
      "learning_rate": 9.28886843508024e-05,
      "loss": 9.5881,
      "step": 581
    },
    {
      "epoch": 0.20075888237323214,
      "grad_norm": 1.365981936454773,
      "learning_rate": 9.285981016217759e-05,
      "loss": 9.5745,
      "step": 582
    },
    {
      "epoch": 0.20110382890651948,
      "grad_norm": 1.371512532234192,
      "learning_rate": 9.283088197972138e-05,
      "loss": 9.5246,
      "step": 583
    },
    {
      "epoch": 0.20144877543980683,
      "grad_norm": 1.427450180053711,
      "learning_rate": 9.280189983987688e-05,
      "loss": 9.555,
      "step": 584
    },
    {
      "epoch": 0.20179372197309417,
      "grad_norm": 1.450803518295288,
      "learning_rate": 9.277286377915512e-05,
      "loss": 9.5132,
      "step": 585
    },
    {
      "epoch": 0.2021386685063815,
      "grad_norm": 1.3853474855422974,
      "learning_rate": 9.274377383413508e-05,
      "loss": 9.5363,
      "step": 586
    },
    {
      "epoch": 0.20248361503966886,
      "grad_norm": 1.3772315979003906,
      "learning_rate": 9.271463004146363e-05,
      "loss": 9.5506,
      "step": 587
    },
    {
      "epoch": 0.2028285615729562,
      "grad_norm": 1.4147601127624512,
      "learning_rate": 9.268543243785547e-05,
      "loss": 9.4964,
      "step": 588
    },
    {
      "epoch": 0.20317350810624354,
      "grad_norm": 1.4551278352737427,
      "learning_rate": 9.265618106009308e-05,
      "loss": 9.5859,
      "step": 589
    },
    {
      "epoch": 0.20351845463953086,
      "grad_norm": 1.3521373271942139,
      "learning_rate": 9.262687594502671e-05,
      "loss": 9.5175,
      "step": 590
    },
    {
      "epoch": 0.2038634011728182,
      "grad_norm": 1.4414533376693726,
      "learning_rate": 9.259751712957428e-05,
      "loss": 9.454,
      "step": 591
    },
    {
      "epoch": 0.20420834770610555,
      "grad_norm": 1.4528087377548218,
      "learning_rate": 9.256810465072136e-05,
      "loss": 9.4803,
      "step": 592
    },
    {
      "epoch": 0.2045532942393929,
      "grad_norm": 1.6504982709884644,
      "learning_rate": 9.253863854552113e-05,
      "loss": 9.4405,
      "step": 593
    },
    {
      "epoch": 0.20489824077268023,
      "grad_norm": 1.3925122022628784,
      "learning_rate": 9.250911885109436e-05,
      "loss": 9.5588,
      "step": 594
    },
    {
      "epoch": 0.20524318730596758,
      "grad_norm": 1.4524890184402466,
      "learning_rate": 9.247954560462929e-05,
      "loss": 9.4648,
      "step": 595
    },
    {
      "epoch": 0.20558813383925492,
      "grad_norm": 1.4536553621292114,
      "learning_rate": 9.244991884338161e-05,
      "loss": 9.4829,
      "step": 596
    },
    {
      "epoch": 0.20593308037254227,
      "grad_norm": 1.6968779563903809,
      "learning_rate": 9.242023860467448e-05,
      "loss": 9.4031,
      "step": 597
    },
    {
      "epoch": 0.2062780269058296,
      "grad_norm": 1.6069278717041016,
      "learning_rate": 9.239050492589838e-05,
      "loss": 9.432,
      "step": 598
    },
    {
      "epoch": 0.20662297343911692,
      "grad_norm": 1.5400954484939575,
      "learning_rate": 9.236071784451117e-05,
      "loss": 9.4716,
      "step": 599
    },
    {
      "epoch": 0.20696791997240427,
      "grad_norm": 1.7635220289230347,
      "learning_rate": 9.233087739803792e-05,
      "loss": 9.3843,
      "step": 600
    },
    {
      "epoch": 0.2073128665056916,
      "grad_norm": 0.9172026515007019,
      "learning_rate": 9.230098362407095e-05,
      "loss": 9.6446,
      "step": 601
    },
    {
      "epoch": 0.20765781303897896,
      "grad_norm": 1.095590591430664,
      "learning_rate": 9.227103656026981e-05,
      "loss": 9.5951,
      "step": 602
    },
    {
      "epoch": 0.2080027595722663,
      "grad_norm": 1.0268243551254272,
      "learning_rate": 9.22410362443611e-05,
      "loss": 9.5898,
      "step": 603
    },
    {
      "epoch": 0.20834770610555364,
      "grad_norm": 1.0711814165115356,
      "learning_rate": 9.221098271413855e-05,
      "loss": 9.6209,
      "step": 604
    },
    {
      "epoch": 0.208692652638841,
      "grad_norm": 1.1392238140106201,
      "learning_rate": 9.218087600746297e-05,
      "loss": 9.5681,
      "step": 605
    },
    {
      "epoch": 0.20903759917212833,
      "grad_norm": 1.0311208963394165,
      "learning_rate": 9.215071616226207e-05,
      "loss": 9.6356,
      "step": 606
    },
    {
      "epoch": 0.20938254570541567,
      "grad_norm": 1.1314677000045776,
      "learning_rate": 9.212050321653056e-05,
      "loss": 9.5649,
      "step": 607
    },
    {
      "epoch": 0.209727492238703,
      "grad_norm": 1.0471137762069702,
      "learning_rate": 9.209023720833005e-05,
      "loss": 9.5673,
      "step": 608
    },
    {
      "epoch": 0.21007243877199033,
      "grad_norm": 1.2574738264083862,
      "learning_rate": 9.205991817578896e-05,
      "loss": 9.5615,
      "step": 609
    },
    {
      "epoch": 0.21041738530527768,
      "grad_norm": 1.195786476135254,
      "learning_rate": 9.202954615710256e-05,
      "loss": 9.5889,
      "step": 610
    },
    {
      "epoch": 0.21076233183856502,
      "grad_norm": 1.2348692417144775,
      "learning_rate": 9.19991211905328e-05,
      "loss": 9.5197,
      "step": 611
    },
    {
      "epoch": 0.21110727837185236,
      "grad_norm": 1.2200709581375122,
      "learning_rate": 9.19686433144084e-05,
      "loss": 9.6073,
      "step": 612
    },
    {
      "epoch": 0.2114522249051397,
      "grad_norm": 1.230139970779419,
      "learning_rate": 9.19381125671247e-05,
      "loss": 9.5837,
      "step": 613
    },
    {
      "epoch": 0.21179717143842705,
      "grad_norm": 1.3068007230758667,
      "learning_rate": 9.190752898714365e-05,
      "loss": 9.5385,
      "step": 614
    },
    {
      "epoch": 0.2121421179717144,
      "grad_norm": 1.3114107847213745,
      "learning_rate": 9.187689261299377e-05,
      "loss": 9.5578,
      "step": 615
    },
    {
      "epoch": 0.21248706450500174,
      "grad_norm": 1.1390831470489502,
      "learning_rate": 9.184620348327008e-05,
      "loss": 9.5274,
      "step": 616
    },
    {
      "epoch": 0.21283201103828905,
      "grad_norm": 1.1909688711166382,
      "learning_rate": 9.181546163663406e-05,
      "loss": 9.5915,
      "step": 617
    },
    {
      "epoch": 0.2131769575715764,
      "grad_norm": 1.1962287425994873,
      "learning_rate": 9.17846671118136e-05,
      "loss": 9.5722,
      "step": 618
    },
    {
      "epoch": 0.21352190410486374,
      "grad_norm": 1.2285292148590088,
      "learning_rate": 9.175381994760294e-05,
      "loss": 9.5933,
      "step": 619
    },
    {
      "epoch": 0.21386685063815108,
      "grad_norm": 1.3546727895736694,
      "learning_rate": 9.172292018286266e-05,
      "loss": 9.5063,
      "step": 620
    },
    {
      "epoch": 0.21421179717143843,
      "grad_norm": 1.3017284870147705,
      "learning_rate": 9.169196785651959e-05,
      "loss": 9.5152,
      "step": 621
    },
    {
      "epoch": 0.21455674370472577,
      "grad_norm": 1.3723574876785278,
      "learning_rate": 9.16609630075668e-05,
      "loss": 9.5051,
      "step": 622
    },
    {
      "epoch": 0.21490169023801312,
      "grad_norm": 1.2360388040542603,
      "learning_rate": 9.162990567506347e-05,
      "loss": 9.5724,
      "step": 623
    },
    {
      "epoch": 0.21524663677130046,
      "grad_norm": 1.2957721948623657,
      "learning_rate": 9.159879589813495e-05,
      "loss": 9.5246,
      "step": 624
    },
    {
      "epoch": 0.2155915833045878,
      "grad_norm": 1.430680751800537,
      "learning_rate": 9.156763371597267e-05,
      "loss": 9.5223,
      "step": 625
    },
    {
      "epoch": 0.21593652983787512,
      "grad_norm": 1.2834513187408447,
      "learning_rate": 9.1536419167834e-05,
      "loss": 9.539,
      "step": 626
    },
    {
      "epoch": 0.21628147637116246,
      "grad_norm": 1.2917267084121704,
      "learning_rate": 9.150515229304238e-05,
      "loss": 9.5258,
      "step": 627
    },
    {
      "epoch": 0.2166264229044498,
      "grad_norm": 1.3878052234649658,
      "learning_rate": 9.147383313098707e-05,
      "loss": 9.5223,
      "step": 628
    },
    {
      "epoch": 0.21697136943773715,
      "grad_norm": 1.3863571882247925,
      "learning_rate": 9.14424617211233e-05,
      "loss": 9.4875,
      "step": 629
    },
    {
      "epoch": 0.2173163159710245,
      "grad_norm": 1.3569782972335815,
      "learning_rate": 9.141103810297203e-05,
      "loss": 9.4716,
      "step": 630
    },
    {
      "epoch": 0.21766126250431184,
      "grad_norm": 1.3145636320114136,
      "learning_rate": 9.137956231612007e-05,
      "loss": 9.5432,
      "step": 631
    },
    {
      "epoch": 0.21800620903759918,
      "grad_norm": 1.2919973134994507,
      "learning_rate": 9.134803440021986e-05,
      "loss": 9.57,
      "step": 632
    },
    {
      "epoch": 0.21835115557088652,
      "grad_norm": 1.4465949535369873,
      "learning_rate": 9.131645439498963e-05,
      "loss": 9.4696,
      "step": 633
    },
    {
      "epoch": 0.21869610210417384,
      "grad_norm": 1.4450429677963257,
      "learning_rate": 9.128482234021311e-05,
      "loss": 9.4718,
      "step": 634
    },
    {
      "epoch": 0.21904104863746118,
      "grad_norm": 1.1756001710891724,
      "learning_rate": 9.12531382757397e-05,
      "loss": 9.6019,
      "step": 635
    },
    {
      "epoch": 0.21938599517074853,
      "grad_norm": 1.45283043384552,
      "learning_rate": 9.122140224148426e-05,
      "loss": 9.4854,
      "step": 636
    },
    {
      "epoch": 0.21973094170403587,
      "grad_norm": 1.4676939249038696,
      "learning_rate": 9.11896142774271e-05,
      "loss": 9.5498,
      "step": 637
    },
    {
      "epoch": 0.22007588823732321,
      "grad_norm": 1.3562549352645874,
      "learning_rate": 9.115777442361404e-05,
      "loss": 9.514,
      "step": 638
    },
    {
      "epoch": 0.22042083477061056,
      "grad_norm": 1.3062798976898193,
      "learning_rate": 9.11258827201562e-05,
      "loss": 9.4851,
      "step": 639
    },
    {
      "epoch": 0.2207657813038979,
      "grad_norm": 1.41761314868927,
      "learning_rate": 9.109393920723002e-05,
      "loss": 9.5105,
      "step": 640
    },
    {
      "epoch": 0.22111072783718524,
      "grad_norm": 1.4332202672958374,
      "learning_rate": 9.106194392507725e-05,
      "loss": 9.5448,
      "step": 641
    },
    {
      "epoch": 0.2214556743704726,
      "grad_norm": 1.3489359617233276,
      "learning_rate": 9.10298969140048e-05,
      "loss": 9.4664,
      "step": 642
    },
    {
      "epoch": 0.2218006209037599,
      "grad_norm": 1.4809479713439941,
      "learning_rate": 9.09977982143848e-05,
      "loss": 9.5064,
      "step": 643
    },
    {
      "epoch": 0.22214556743704725,
      "grad_norm": 1.3503596782684326,
      "learning_rate": 9.096564786665448e-05,
      "loss": 9.5502,
      "step": 644
    },
    {
      "epoch": 0.2224905139703346,
      "grad_norm": 1.4867632389068604,
      "learning_rate": 9.093344591131612e-05,
      "loss": 9.4481,
      "step": 645
    },
    {
      "epoch": 0.22283546050362193,
      "grad_norm": 1.5679457187652588,
      "learning_rate": 9.090119238893702e-05,
      "loss": 9.4593,
      "step": 646
    },
    {
      "epoch": 0.22318040703690928,
      "grad_norm": 1.50774085521698,
      "learning_rate": 9.086888734014945e-05,
      "loss": 9.4557,
      "step": 647
    },
    {
      "epoch": 0.22352535357019662,
      "grad_norm": 1.6264348030090332,
      "learning_rate": 9.083653080565058e-05,
      "loss": 9.4339,
      "step": 648
    },
    {
      "epoch": 0.22387030010348397,
      "grad_norm": 1.7029718160629272,
      "learning_rate": 9.080412282620248e-05,
      "loss": 9.416,
      "step": 649
    },
    {
      "epoch": 0.2242152466367713,
      "grad_norm": 1.7485153675079346,
      "learning_rate": 9.077166344263197e-05,
      "loss": 9.3769,
      "step": 650
    },
    {
      "epoch": 0.22456019317005865,
      "grad_norm": 0.9622684121131897,
      "learning_rate": 9.073915269583067e-05,
      "loss": 9.5603,
      "step": 651
    },
    {
      "epoch": 0.22490513970334597,
      "grad_norm": 1.1377094984054565,
      "learning_rate": 9.070659062675487e-05,
      "loss": 9.5554,
      "step": 652
    },
    {
      "epoch": 0.2252500862366333,
      "grad_norm": 0.9533647894859314,
      "learning_rate": 9.067397727642557e-05,
      "loss": 9.57,
      "step": 653
    },
    {
      "epoch": 0.22559503276992066,
      "grad_norm": 1.1317623853683472,
      "learning_rate": 9.064131268592831e-05,
      "loss": 9.5492,
      "step": 654
    },
    {
      "epoch": 0.225939979303208,
      "grad_norm": 1.1853982210159302,
      "learning_rate": 9.060859689641323e-05,
      "loss": 9.5454,
      "step": 655
    },
    {
      "epoch": 0.22628492583649534,
      "grad_norm": 1.0806140899658203,
      "learning_rate": 9.057582994909496e-05,
      "loss": 9.6048,
      "step": 656
    },
    {
      "epoch": 0.2266298723697827,
      "grad_norm": 1.0558366775512695,
      "learning_rate": 9.054301188525254e-05,
      "loss": 9.5815,
      "step": 657
    },
    {
      "epoch": 0.22697481890307003,
      "grad_norm": 1.124320387840271,
      "learning_rate": 9.051014274622947e-05,
      "loss": 9.5883,
      "step": 658
    },
    {
      "epoch": 0.22731976543635737,
      "grad_norm": 1.1820127964019775,
      "learning_rate": 9.04772225734335e-05,
      "loss": 9.5637,
      "step": 659
    },
    {
      "epoch": 0.22766471196964472,
      "grad_norm": 1.1382235288619995,
      "learning_rate": 9.044425140833679e-05,
      "loss": 9.5761,
      "step": 660
    },
    {
      "epoch": 0.22800965850293203,
      "grad_norm": 1.2138105630874634,
      "learning_rate": 9.041122929247566e-05,
      "loss": 9.5104,
      "step": 661
    },
    {
      "epoch": 0.22835460503621938,
      "grad_norm": 1.2376949787139893,
      "learning_rate": 9.037815626745059e-05,
      "loss": 9.538,
      "step": 662
    },
    {
      "epoch": 0.22869955156950672,
      "grad_norm": 1.0739209651947021,
      "learning_rate": 9.03450323749263e-05,
      "loss": 9.5949,
      "step": 663
    },
    {
      "epoch": 0.22904449810279406,
      "grad_norm": 1.1784121990203857,
      "learning_rate": 9.031185765663149e-05,
      "loss": 9.5563,
      "step": 664
    },
    {
      "epoch": 0.2293894446360814,
      "grad_norm": 1.3404618501663208,
      "learning_rate": 9.027863215435895e-05,
      "loss": 9.4811,
      "step": 665
    },
    {
      "epoch": 0.22973439116936875,
      "grad_norm": 1.2364481687545776,
      "learning_rate": 9.024535590996541e-05,
      "loss": 9.5474,
      "step": 666
    },
    {
      "epoch": 0.2300793377026561,
      "grad_norm": 1.3062660694122314,
      "learning_rate": 9.021202896537155e-05,
      "loss": 9.5248,
      "step": 667
    },
    {
      "epoch": 0.23042428423594344,
      "grad_norm": 1.19839346408844,
      "learning_rate": 9.017865136256191e-05,
      "loss": 9.5374,
      "step": 668
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 1.3458032608032227,
      "learning_rate": 9.014522314358484e-05,
      "loss": 9.4901,
      "step": 669
    },
    {
      "epoch": 0.2311141773025181,
      "grad_norm": 1.2809069156646729,
      "learning_rate": 9.011174435055247e-05,
      "loss": 9.5303,
      "step": 670
    },
    {
      "epoch": 0.23145912383580544,
      "grad_norm": 1.2737077474594116,
      "learning_rate": 9.007821502564064e-05,
      "loss": 9.5581,
      "step": 671
    },
    {
      "epoch": 0.23180407036909279,
      "grad_norm": 1.205348014831543,
      "learning_rate": 9.004463521108884e-05,
      "loss": 9.562,
      "step": 672
    },
    {
      "epoch": 0.23214901690238013,
      "grad_norm": 1.344906210899353,
      "learning_rate": 9.001100494920016e-05,
      "loss": 9.4788,
      "step": 673
    },
    {
      "epoch": 0.23249396343566747,
      "grad_norm": 1.2621089220046997,
      "learning_rate": 8.997732428234126e-05,
      "loss": 9.5728,
      "step": 674
    },
    {
      "epoch": 0.23283890996895482,
      "grad_norm": 1.2624657154083252,
      "learning_rate": 8.99435932529423e-05,
      "loss": 9.5437,
      "step": 675
    },
    {
      "epoch": 0.23318385650224216,
      "grad_norm": 1.3407676219940186,
      "learning_rate": 8.990981190349688e-05,
      "loss": 9.5087,
      "step": 676
    },
    {
      "epoch": 0.2335288030355295,
      "grad_norm": 1.365085244178772,
      "learning_rate": 8.987598027656197e-05,
      "loss": 9.4752,
      "step": 677
    },
    {
      "epoch": 0.23387374956881685,
      "grad_norm": 1.2666501998901367,
      "learning_rate": 8.984209841475791e-05,
      "loss": 9.5497,
      "step": 678
    },
    {
      "epoch": 0.23421869610210416,
      "grad_norm": 1.328727126121521,
      "learning_rate": 8.98081663607683e-05,
      "loss": 9.5032,
      "step": 679
    },
    {
      "epoch": 0.2345636426353915,
      "grad_norm": 1.334533929824829,
      "learning_rate": 8.977418415734002e-05,
      "loss": 9.5215,
      "step": 680
    },
    {
      "epoch": 0.23490858916867885,
      "grad_norm": 1.4917492866516113,
      "learning_rate": 8.974015184728304e-05,
      "loss": 9.4658,
      "step": 681
    },
    {
      "epoch": 0.2352535357019662,
      "grad_norm": 1.4826164245605469,
      "learning_rate": 8.970606947347054e-05,
      "loss": 9.4797,
      "step": 682
    },
    {
      "epoch": 0.23559848223525354,
      "grad_norm": 1.3176090717315674,
      "learning_rate": 8.967193707883872e-05,
      "loss": 9.4901,
      "step": 683
    },
    {
      "epoch": 0.23594342876854088,
      "grad_norm": 1.3593602180480957,
      "learning_rate": 8.963775470638682e-05,
      "loss": 9.459,
      "step": 684
    },
    {
      "epoch": 0.23628837530182822,
      "grad_norm": 1.395217776298523,
      "learning_rate": 8.960352239917699e-05,
      "loss": 9.5029,
      "step": 685
    },
    {
      "epoch": 0.23663332183511557,
      "grad_norm": 1.5413119792938232,
      "learning_rate": 8.956924020033439e-05,
      "loss": 9.4326,
      "step": 686
    },
    {
      "epoch": 0.2369782683684029,
      "grad_norm": 1.3983765840530396,
      "learning_rate": 8.953490815304692e-05,
      "loss": 9.5601,
      "step": 687
    },
    {
      "epoch": 0.23732321490169023,
      "grad_norm": 1.2415794134140015,
      "learning_rate": 8.950052630056534e-05,
      "loss": 9.5054,
      "step": 688
    },
    {
      "epoch": 0.23766816143497757,
      "grad_norm": 1.4755887985229492,
      "learning_rate": 8.946609468620313e-05,
      "loss": 9.5442,
      "step": 689
    },
    {
      "epoch": 0.23801310796826491,
      "grad_norm": 1.4634833335876465,
      "learning_rate": 8.94316133533365e-05,
      "loss": 9.4095,
      "step": 690
    },
    {
      "epoch": 0.23835805450155226,
      "grad_norm": 1.5690680742263794,
      "learning_rate": 8.939708234540423e-05,
      "loss": 9.4441,
      "step": 691
    },
    {
      "epoch": 0.2387030010348396,
      "grad_norm": 1.4769717454910278,
      "learning_rate": 8.936250170590772e-05,
      "loss": 9.4923,
      "step": 692
    },
    {
      "epoch": 0.23904794756812695,
      "grad_norm": 1.3412718772888184,
      "learning_rate": 8.932787147841089e-05,
      "loss": 9.5002,
      "step": 693
    },
    {
      "epoch": 0.2393928941014143,
      "grad_norm": 1.539059042930603,
      "learning_rate": 8.929319170654013e-05,
      "loss": 9.416,
      "step": 694
    },
    {
      "epoch": 0.23973784063470163,
      "grad_norm": 1.5159481763839722,
      "learning_rate": 8.925846243398423e-05,
      "loss": 9.4452,
      "step": 695
    },
    {
      "epoch": 0.24008278716798895,
      "grad_norm": 1.5319547653198242,
      "learning_rate": 8.922368370449438e-05,
      "loss": 9.4928,
      "step": 696
    },
    {
      "epoch": 0.2404277337012763,
      "grad_norm": 1.6070092916488647,
      "learning_rate": 8.918885556188401e-05,
      "loss": 9.4612,
      "step": 697
    },
    {
      "epoch": 0.24077268023456364,
      "grad_norm": 1.4909985065460205,
      "learning_rate": 8.915397805002887e-05,
      "loss": 9.4975,
      "step": 698
    },
    {
      "epoch": 0.24111762676785098,
      "grad_norm": 1.5712077617645264,
      "learning_rate": 8.911905121286684e-05,
      "loss": 9.4124,
      "step": 699
    },
    {
      "epoch": 0.24146257330113832,
      "grad_norm": 1.7502754926681519,
      "learning_rate": 8.9084075094398e-05,
      "loss": 9.3159,
      "step": 700
    },
    {
      "epoch": 0.24180751983442567,
      "grad_norm": 0.972281813621521,
      "learning_rate": 8.904904973868445e-05,
      "loss": 9.5847,
      "step": 701
    },
    {
      "epoch": 0.242152466367713,
      "grad_norm": 0.9288579821586609,
      "learning_rate": 8.901397518985038e-05,
      "loss": 9.6492,
      "step": 702
    },
    {
      "epoch": 0.24249741290100035,
      "grad_norm": 1.1141493320465088,
      "learning_rate": 8.89788514920819e-05,
      "loss": 9.55,
      "step": 703
    },
    {
      "epoch": 0.2428423594342877,
      "grad_norm": 1.0111804008483887,
      "learning_rate": 8.894367868962707e-05,
      "loss": 9.6488,
      "step": 704
    },
    {
      "epoch": 0.243187305967575,
      "grad_norm": 1.0936983823776245,
      "learning_rate": 8.890845682679579e-05,
      "loss": 9.5835,
      "step": 705
    },
    {
      "epoch": 0.24353225250086236,
      "grad_norm": 1.1165025234222412,
      "learning_rate": 8.887318594795982e-05,
      "loss": 9.5752,
      "step": 706
    },
    {
      "epoch": 0.2438771990341497,
      "grad_norm": 1.2050586938858032,
      "learning_rate": 8.883786609755255e-05,
      "loss": 9.5348,
      "step": 707
    },
    {
      "epoch": 0.24422214556743704,
      "grad_norm": 1.1180025339126587,
      "learning_rate": 8.88024973200692e-05,
      "loss": 9.6259,
      "step": 708
    },
    {
      "epoch": 0.2445670921007244,
      "grad_norm": 1.1547380685806274,
      "learning_rate": 8.876707966006656e-05,
      "loss": 9.5371,
      "step": 709
    },
    {
      "epoch": 0.24491203863401173,
      "grad_norm": 1.218729853630066,
      "learning_rate": 8.873161316216296e-05,
      "loss": 9.541,
      "step": 710
    },
    {
      "epoch": 0.24525698516729907,
      "grad_norm": 1.1403273344039917,
      "learning_rate": 8.869609787103836e-05,
      "loss": 9.5757,
      "step": 711
    },
    {
      "epoch": 0.24560193170058642,
      "grad_norm": 1.2201348543167114,
      "learning_rate": 8.86605338314341e-05,
      "loss": 9.5228,
      "step": 712
    },
    {
      "epoch": 0.24594687823387376,
      "grad_norm": 1.096950888633728,
      "learning_rate": 8.862492108815297e-05,
      "loss": 9.5398,
      "step": 713
    },
    {
      "epoch": 0.24629182476716108,
      "grad_norm": 0.9905299544334412,
      "learning_rate": 8.858925968605909e-05,
      "loss": 9.6028,
      "step": 714
    },
    {
      "epoch": 0.24663677130044842,
      "grad_norm": 1.2241541147232056,
      "learning_rate": 8.855354967007793e-05,
      "loss": 9.5068,
      "step": 715
    },
    {
      "epoch": 0.24698171783373576,
      "grad_norm": 1.3112493753433228,
      "learning_rate": 8.851779108519615e-05,
      "loss": 9.4628,
      "step": 716
    },
    {
      "epoch": 0.2473266643670231,
      "grad_norm": 1.1782866716384888,
      "learning_rate": 8.848198397646161e-05,
      "loss": 9.5325,
      "step": 717
    },
    {
      "epoch": 0.24767161090031045,
      "grad_norm": 1.3042865991592407,
      "learning_rate": 8.844612838898334e-05,
      "loss": 9.5256,
      "step": 718
    },
    {
      "epoch": 0.2480165574335978,
      "grad_norm": 1.2344504594802856,
      "learning_rate": 8.841022436793135e-05,
      "loss": 9.5835,
      "step": 719
    },
    {
      "epoch": 0.24836150396688514,
      "grad_norm": 1.2775154113769531,
      "learning_rate": 8.837427195853679e-05,
      "loss": 9.5256,
      "step": 720
    },
    {
      "epoch": 0.24870645050017248,
      "grad_norm": 1.3229973316192627,
      "learning_rate": 8.833827120609166e-05,
      "loss": 9.4894,
      "step": 721
    },
    {
      "epoch": 0.24905139703345983,
      "grad_norm": 1.1892383098602295,
      "learning_rate": 8.83022221559489e-05,
      "loss": 9.5601,
      "step": 722
    },
    {
      "epoch": 0.24939634356674714,
      "grad_norm": 1.3059921264648438,
      "learning_rate": 8.826612485352232e-05,
      "loss": 9.5064,
      "step": 723
    },
    {
      "epoch": 0.24974129010003449,
      "grad_norm": 1.1932693719863892,
      "learning_rate": 8.822997934428649e-05,
      "loss": 9.6054,
      "step": 724
    },
    {
      "epoch": 0.25008623663332186,
      "grad_norm": 1.3107818365097046,
      "learning_rate": 8.819378567377672e-05,
      "loss": 9.5453,
      "step": 725
    },
    {
      "epoch": 0.2504311831666092,
      "grad_norm": 1.4025793075561523,
      "learning_rate": 8.815754388758897e-05,
      "loss": 9.4636,
      "step": 726
    },
    {
      "epoch": 0.2507761296998965,
      "grad_norm": 1.3353124856948853,
      "learning_rate": 8.812125403137985e-05,
      "loss": 9.49,
      "step": 727
    },
    {
      "epoch": 0.25112107623318386,
      "grad_norm": 1.2831143140792847,
      "learning_rate": 8.808491615086649e-05,
      "loss": 9.5327,
      "step": 728
    },
    {
      "epoch": 0.2514660227664712,
      "grad_norm": 1.396451711654663,
      "learning_rate": 8.804853029182656e-05,
      "loss": 9.4794,
      "step": 729
    },
    {
      "epoch": 0.25181096929975855,
      "grad_norm": 1.2553220987319946,
      "learning_rate": 8.801209650009814e-05,
      "loss": 9.4996,
      "step": 730
    },
    {
      "epoch": 0.25215591583304586,
      "grad_norm": 1.3066761493682861,
      "learning_rate": 8.797561482157973e-05,
      "loss": 9.5276,
      "step": 731
    },
    {
      "epoch": 0.25250086236633323,
      "grad_norm": 1.443993091583252,
      "learning_rate": 8.79390853022301e-05,
      "loss": 9.4559,
      "step": 732
    },
    {
      "epoch": 0.25284580889962055,
      "grad_norm": 1.328656554222107,
      "learning_rate": 8.790250798806833e-05,
      "loss": 9.4936,
      "step": 733
    },
    {
      "epoch": 0.2531907554329079,
      "grad_norm": 1.3747369050979614,
      "learning_rate": 8.786588292517373e-05,
      "loss": 9.4605,
      "step": 734
    },
    {
      "epoch": 0.25353570196619524,
      "grad_norm": 1.3007593154907227,
      "learning_rate": 8.782921015968569e-05,
      "loss": 9.5227,
      "step": 735
    },
    {
      "epoch": 0.25388064849948255,
      "grad_norm": 1.508450984954834,
      "learning_rate": 8.779248973780378e-05,
      "loss": 9.4072,
      "step": 736
    },
    {
      "epoch": 0.2542255950327699,
      "grad_norm": 1.3896534442901611,
      "learning_rate": 8.775572170578757e-05,
      "loss": 9.4753,
      "step": 737
    },
    {
      "epoch": 0.25457054156605724,
      "grad_norm": 1.4728200435638428,
      "learning_rate": 8.771890610995659e-05,
      "loss": 9.4467,
      "step": 738
    },
    {
      "epoch": 0.2549154880993446,
      "grad_norm": 1.3297967910766602,
      "learning_rate": 8.768204299669029e-05,
      "loss": 9.5018,
      "step": 739
    },
    {
      "epoch": 0.2552604346326319,
      "grad_norm": 1.454687476158142,
      "learning_rate": 8.764513241242802e-05,
      "loss": 9.5259,
      "step": 740
    },
    {
      "epoch": 0.2556053811659193,
      "grad_norm": 1.4180147647857666,
      "learning_rate": 8.76081744036689e-05,
      "loss": 9.5011,
      "step": 741
    },
    {
      "epoch": 0.2559503276992066,
      "grad_norm": 1.5591294765472412,
      "learning_rate": 8.75711690169718e-05,
      "loss": 9.4629,
      "step": 742
    },
    {
      "epoch": 0.256295274232494,
      "grad_norm": 1.4959933757781982,
      "learning_rate": 8.75341162989553e-05,
      "loss": 9.4018,
      "step": 743
    },
    {
      "epoch": 0.2566402207657813,
      "grad_norm": 1.3860907554626465,
      "learning_rate": 8.749701629629756e-05,
      "loss": 9.4971,
      "step": 744
    },
    {
      "epoch": 0.2569851672990686,
      "grad_norm": 1.4601097106933594,
      "learning_rate": 8.745986905573634e-05,
      "loss": 9.4616,
      "step": 745
    },
    {
      "epoch": 0.257330113832356,
      "grad_norm": 1.5998722314834595,
      "learning_rate": 8.742267462406892e-05,
      "loss": 9.4879,
      "step": 746
    },
    {
      "epoch": 0.2576750603656433,
      "grad_norm": 1.7166414260864258,
      "learning_rate": 8.738543304815202e-05,
      "loss": 9.4445,
      "step": 747
    },
    {
      "epoch": 0.2580200068989307,
      "grad_norm": 1.5866833925247192,
      "learning_rate": 8.734814437490171e-05,
      "loss": 9.3825,
      "step": 748
    },
    {
      "epoch": 0.258364953432218,
      "grad_norm": 1.6864278316497803,
      "learning_rate": 8.731080865129344e-05,
      "loss": 9.3629,
      "step": 749
    },
    {
      "epoch": 0.25870989996550536,
      "grad_norm": 1.655298113822937,
      "learning_rate": 8.727342592436194e-05,
      "loss": 9.3556,
      "step": 750
    },
    {
      "epoch": 0.2590548464987927,
      "grad_norm": 0.8688000440597534,
      "learning_rate": 8.723599624120111e-05,
      "loss": 9.5912,
      "step": 751
    },
    {
      "epoch": 0.25939979303208005,
      "grad_norm": 1.0522035360336304,
      "learning_rate": 8.719851964896405e-05,
      "loss": 9.622,
      "step": 752
    },
    {
      "epoch": 0.25974473956536737,
      "grad_norm": 0.9646815657615662,
      "learning_rate": 8.716099619486293e-05,
      "loss": 9.6436,
      "step": 753
    },
    {
      "epoch": 0.2600896860986547,
      "grad_norm": 1.048831582069397,
      "learning_rate": 8.712342592616894e-05,
      "loss": 9.5588,
      "step": 754
    },
    {
      "epoch": 0.26043463263194205,
      "grad_norm": 1.0627691745758057,
      "learning_rate": 8.708580889021228e-05,
      "loss": 9.5313,
      "step": 755
    },
    {
      "epoch": 0.26077957916522937,
      "grad_norm": 1.06023371219635,
      "learning_rate": 8.704814513438206e-05,
      "loss": 9.532,
      "step": 756
    },
    {
      "epoch": 0.26112452569851674,
      "grad_norm": 1.085838794708252,
      "learning_rate": 8.701043470612622e-05,
      "loss": 9.5856,
      "step": 757
    },
    {
      "epoch": 0.26146947223180406,
      "grad_norm": 1.1486434936523438,
      "learning_rate": 8.697267765295153e-05,
      "loss": 9.6146,
      "step": 758
    },
    {
      "epoch": 0.26181441876509143,
      "grad_norm": 1.053144931793213,
      "learning_rate": 8.693487402242348e-05,
      "loss": 9.5736,
      "step": 759
    },
    {
      "epoch": 0.26215936529837874,
      "grad_norm": 1.2557427883148193,
      "learning_rate": 8.689702386216622e-05,
      "loss": 9.5062,
      "step": 760
    },
    {
      "epoch": 0.2625043118316661,
      "grad_norm": 1.2348942756652832,
      "learning_rate": 8.685912721986256e-05,
      "loss": 9.5179,
      "step": 761
    },
    {
      "epoch": 0.26284925836495343,
      "grad_norm": 1.2272906303405762,
      "learning_rate": 8.682118414325383e-05,
      "loss": 9.5305,
      "step": 762
    },
    {
      "epoch": 0.26319420489824075,
      "grad_norm": 1.2449849843978882,
      "learning_rate": 8.678319468013987e-05,
      "loss": 9.4919,
      "step": 763
    },
    {
      "epoch": 0.2635391514315281,
      "grad_norm": 1.264153003692627,
      "learning_rate": 8.674515887837897e-05,
      "loss": 9.5337,
      "step": 764
    },
    {
      "epoch": 0.26388409796481543,
      "grad_norm": 1.132380723953247,
      "learning_rate": 8.670707678588778e-05,
      "loss": 9.5491,
      "step": 765
    },
    {
      "epoch": 0.2642290444981028,
      "grad_norm": 1.2579922676086426,
      "learning_rate": 8.666894845064126e-05,
      "loss": 9.5501,
      "step": 766
    },
    {
      "epoch": 0.2645739910313901,
      "grad_norm": 1.2157853841781616,
      "learning_rate": 8.663077392067265e-05,
      "loss": 9.512,
      "step": 767
    },
    {
      "epoch": 0.2649189375646775,
      "grad_norm": 1.3099197149276733,
      "learning_rate": 8.659255324407334e-05,
      "loss": 9.4923,
      "step": 768
    },
    {
      "epoch": 0.2652638840979648,
      "grad_norm": 1.204965591430664,
      "learning_rate": 8.655428646899292e-05,
      "loss": 9.6055,
      "step": 769
    },
    {
      "epoch": 0.2656088306312522,
      "grad_norm": 1.3694841861724854,
      "learning_rate": 8.651597364363899e-05,
      "loss": 9.4878,
      "step": 770
    },
    {
      "epoch": 0.2659537771645395,
      "grad_norm": 1.263826847076416,
      "learning_rate": 8.647761481627721e-05,
      "loss": 9.5483,
      "step": 771
    },
    {
      "epoch": 0.2662987236978268,
      "grad_norm": 1.2380715608596802,
      "learning_rate": 8.643921003523117e-05,
      "loss": 9.563,
      "step": 772
    },
    {
      "epoch": 0.2666436702311142,
      "grad_norm": 1.3266398906707764,
      "learning_rate": 8.640075934888234e-05,
      "loss": 9.4974,
      "step": 773
    },
    {
      "epoch": 0.2669886167644015,
      "grad_norm": 1.2829941511154175,
      "learning_rate": 8.636226280567007e-05,
      "loss": 9.6534,
      "step": 774
    },
    {
      "epoch": 0.26733356329768887,
      "grad_norm": 1.2618184089660645,
      "learning_rate": 8.632372045409141e-05,
      "loss": 9.5707,
      "step": 775
    },
    {
      "epoch": 0.2676785098309762,
      "grad_norm": 1.2372167110443115,
      "learning_rate": 8.628513234270117e-05,
      "loss": 9.5143,
      "step": 776
    },
    {
      "epoch": 0.26802345636426356,
      "grad_norm": 1.3537949323654175,
      "learning_rate": 8.62464985201118e-05,
      "loss": 9.514,
      "step": 777
    },
    {
      "epoch": 0.2683684028975509,
      "grad_norm": 1.371895670890808,
      "learning_rate": 8.620781903499329e-05,
      "loss": 9.5151,
      "step": 778
    },
    {
      "epoch": 0.26871334943083824,
      "grad_norm": 1.3296464681625366,
      "learning_rate": 8.616909393607324e-05,
      "loss": 9.5469,
      "step": 779
    },
    {
      "epoch": 0.26905829596412556,
      "grad_norm": 1.4044578075408936,
      "learning_rate": 8.613032327213664e-05,
      "loss": 9.4461,
      "step": 780
    },
    {
      "epoch": 0.2694032424974129,
      "grad_norm": 1.4443117380142212,
      "learning_rate": 8.60915070920259e-05,
      "loss": 9.4868,
      "step": 781
    },
    {
      "epoch": 0.26974818903070025,
      "grad_norm": 1.3604815006256104,
      "learning_rate": 8.605264544464078e-05,
      "loss": 9.4855,
      "step": 782
    },
    {
      "epoch": 0.27009313556398756,
      "grad_norm": 1.2737566232681274,
      "learning_rate": 8.601373837893832e-05,
      "loss": 9.5417,
      "step": 783
    },
    {
      "epoch": 0.27043808209727493,
      "grad_norm": 1.444075345993042,
      "learning_rate": 8.597478594393275e-05,
      "loss": 9.5127,
      "step": 784
    },
    {
      "epoch": 0.27078302863056225,
      "grad_norm": 1.1705905199050903,
      "learning_rate": 8.593578818869553e-05,
      "loss": 9.5403,
      "step": 785
    },
    {
      "epoch": 0.2711279751638496,
      "grad_norm": 1.3495186567306519,
      "learning_rate": 8.589674516235513e-05,
      "loss": 9.5364,
      "step": 786
    },
    {
      "epoch": 0.27147292169713694,
      "grad_norm": 1.399301290512085,
      "learning_rate": 8.585765691409707e-05,
      "loss": 9.4696,
      "step": 787
    },
    {
      "epoch": 0.2718178682304243,
      "grad_norm": 1.4776448011398315,
      "learning_rate": 8.581852349316385e-05,
      "loss": 9.4584,
      "step": 788
    },
    {
      "epoch": 0.2721628147637116,
      "grad_norm": 1.3909804821014404,
      "learning_rate": 8.577934494885491e-05,
      "loss": 9.4188,
      "step": 789
    },
    {
      "epoch": 0.27250776129699894,
      "grad_norm": 1.6148669719696045,
      "learning_rate": 8.574012133052649e-05,
      "loss": 9.4471,
      "step": 790
    },
    {
      "epoch": 0.2728527078302863,
      "grad_norm": 1.453417420387268,
      "learning_rate": 8.570085268759163e-05,
      "loss": 9.4643,
      "step": 791
    },
    {
      "epoch": 0.2731976543635736,
      "grad_norm": 1.4808872938156128,
      "learning_rate": 8.566153906952007e-05,
      "loss": 9.3988,
      "step": 792
    },
    {
      "epoch": 0.273542600896861,
      "grad_norm": 1.3660118579864502,
      "learning_rate": 8.562218052583826e-05,
      "loss": 9.4487,
      "step": 793
    },
    {
      "epoch": 0.2738875474301483,
      "grad_norm": 1.6086071729660034,
      "learning_rate": 8.558277710612918e-05,
      "loss": 9.4475,
      "step": 794
    },
    {
      "epoch": 0.2742324939634357,
      "grad_norm": 1.5114543437957764,
      "learning_rate": 8.55433288600324e-05,
      "loss": 9.3927,
      "step": 795
    },
    {
      "epoch": 0.274577440496723,
      "grad_norm": 1.5037239789962769,
      "learning_rate": 8.550383583724394e-05,
      "loss": 9.4977,
      "step": 796
    },
    {
      "epoch": 0.2749223870300104,
      "grad_norm": 1.5446207523345947,
      "learning_rate": 8.546429808751621e-05,
      "loss": 9.4257,
      "step": 797
    },
    {
      "epoch": 0.2752673335632977,
      "grad_norm": 1.5671929121017456,
      "learning_rate": 8.542471566065801e-05,
      "loss": 9.4244,
      "step": 798
    },
    {
      "epoch": 0.275612280096585,
      "grad_norm": 1.8590857982635498,
      "learning_rate": 8.538508860653438e-05,
      "loss": 9.3651,
      "step": 799
    },
    {
      "epoch": 0.2759572266298724,
      "grad_norm": 1.864456057548523,
      "learning_rate": 8.534541697506661e-05,
      "loss": 9.3772,
      "step": 800
    },
    {
      "epoch": 0.2763021731631597,
      "grad_norm": 0.9062460660934448,
      "learning_rate": 8.53057008162321e-05,
      "loss": 9.6121,
      "step": 801
    },
    {
      "epoch": 0.27664711969644706,
      "grad_norm": 0.9802834987640381,
      "learning_rate": 8.526594018006443e-05,
      "loss": 9.5523,
      "step": 802
    },
    {
      "epoch": 0.2769920662297344,
      "grad_norm": 1.0618095397949219,
      "learning_rate": 8.522613511665314e-05,
      "loss": 9.5852,
      "step": 803
    },
    {
      "epoch": 0.27733701276302175,
      "grad_norm": 1.0446445941925049,
      "learning_rate": 8.518628567614373e-05,
      "loss": 9.5632,
      "step": 804
    },
    {
      "epoch": 0.27768195929630907,
      "grad_norm": 1.1162205934524536,
      "learning_rate": 8.514639190873768e-05,
      "loss": 9.5589,
      "step": 805
    },
    {
      "epoch": 0.27802690582959644,
      "grad_norm": 1.080688714981079,
      "learning_rate": 8.510645386469224e-05,
      "loss": 9.5131,
      "step": 806
    },
    {
      "epoch": 0.27837185236288375,
      "grad_norm": 1.093503475189209,
      "learning_rate": 8.506647159432047e-05,
      "loss": 9.6152,
      "step": 807
    },
    {
      "epoch": 0.27871679889617107,
      "grad_norm": 1.0839183330535889,
      "learning_rate": 8.502644514799116e-05,
      "loss": 9.584,
      "step": 808
    },
    {
      "epoch": 0.27906174542945844,
      "grad_norm": 1.2850983142852783,
      "learning_rate": 8.49863745761287e-05,
      "loss": 9.5141,
      "step": 809
    },
    {
      "epoch": 0.27940669196274576,
      "grad_norm": 1.2393872737884521,
      "learning_rate": 8.494625992921315e-05,
      "loss": 9.5024,
      "step": 810
    },
    {
      "epoch": 0.27975163849603313,
      "grad_norm": 1.2150235176086426,
      "learning_rate": 8.490610125777998e-05,
      "loss": 9.5355,
      "step": 811
    },
    {
      "epoch": 0.28009658502932044,
      "grad_norm": 1.0908310413360596,
      "learning_rate": 8.486589861242024e-05,
      "loss": 9.5201,
      "step": 812
    },
    {
      "epoch": 0.2804415315626078,
      "grad_norm": 1.2537291049957275,
      "learning_rate": 8.48256520437803e-05,
      "loss": 9.5507,
      "step": 813
    },
    {
      "epoch": 0.28078647809589513,
      "grad_norm": 1.3384590148925781,
      "learning_rate": 8.47853616025619e-05,
      "loss": 9.5322,
      "step": 814
    },
    {
      "epoch": 0.2811314246291825,
      "grad_norm": 1.163336992263794,
      "learning_rate": 8.474502733952203e-05,
      "loss": 9.5228,
      "step": 815
    },
    {
      "epoch": 0.2814763711624698,
      "grad_norm": 1.2593789100646973,
      "learning_rate": 8.470464930547291e-05,
      "loss": 9.489,
      "step": 816
    },
    {
      "epoch": 0.28182131769575713,
      "grad_norm": 1.0830525159835815,
      "learning_rate": 8.466422755128185e-05,
      "loss": 9.5628,
      "step": 817
    },
    {
      "epoch": 0.2821662642290445,
      "grad_norm": 1.3622268438339233,
      "learning_rate": 8.46237621278713e-05,
      "loss": 9.5389,
      "step": 818
    },
    {
      "epoch": 0.2825112107623318,
      "grad_norm": 1.237600564956665,
      "learning_rate": 8.45832530862187e-05,
      "loss": 9.4825,
      "step": 819
    },
    {
      "epoch": 0.2828561572956192,
      "grad_norm": 1.2572169303894043,
      "learning_rate": 8.454270047735644e-05,
      "loss": 9.5389,
      "step": 820
    },
    {
      "epoch": 0.2832011038289065,
      "grad_norm": 1.3330270051956177,
      "learning_rate": 8.450210435237174e-05,
      "loss": 9.5339,
      "step": 821
    },
    {
      "epoch": 0.2835460503621939,
      "grad_norm": 1.4301376342773438,
      "learning_rate": 8.446146476240675e-05,
      "loss": 9.4688,
      "step": 822
    },
    {
      "epoch": 0.2838909968954812,
      "grad_norm": 1.3089014291763306,
      "learning_rate": 8.44207817586583e-05,
      "loss": 9.531,
      "step": 823
    },
    {
      "epoch": 0.28423594342876857,
      "grad_norm": 1.1468137502670288,
      "learning_rate": 8.438005539237792e-05,
      "loss": 9.6279,
      "step": 824
    },
    {
      "epoch": 0.2845808899620559,
      "grad_norm": 1.2456337213516235,
      "learning_rate": 8.433928571487177e-05,
      "loss": 9.5898,
      "step": 825
    },
    {
      "epoch": 0.2849258364953432,
      "grad_norm": 1.2768372297286987,
      "learning_rate": 8.429847277750063e-05,
      "loss": 9.4724,
      "step": 826
    },
    {
      "epoch": 0.28527078302863057,
      "grad_norm": 1.3287653923034668,
      "learning_rate": 8.425761663167965e-05,
      "loss": 9.5066,
      "step": 827
    },
    {
      "epoch": 0.2856157295619179,
      "grad_norm": 1.379485845565796,
      "learning_rate": 8.421671732887854e-05,
      "loss": 9.4747,
      "step": 828
    },
    {
      "epoch": 0.28596067609520526,
      "grad_norm": 1.395017385482788,
      "learning_rate": 8.417577492062132e-05,
      "loss": 9.5308,
      "step": 829
    },
    {
      "epoch": 0.2863056226284926,
      "grad_norm": 1.3982423543930054,
      "learning_rate": 8.413478945848631e-05,
      "loss": 9.4779,
      "step": 830
    },
    {
      "epoch": 0.28665056916177994,
      "grad_norm": 1.3681056499481201,
      "learning_rate": 8.409376099410609e-05,
      "loss": 9.5195,
      "step": 831
    },
    {
      "epoch": 0.28699551569506726,
      "grad_norm": 1.401589274406433,
      "learning_rate": 8.405268957916738e-05,
      "loss": 9.4662,
      "step": 832
    },
    {
      "epoch": 0.28734046222835463,
      "grad_norm": 1.4695594310760498,
      "learning_rate": 8.401157526541108e-05,
      "loss": 9.4427,
      "step": 833
    },
    {
      "epoch": 0.28768540876164195,
      "grad_norm": 1.426550030708313,
      "learning_rate": 8.397041810463205e-05,
      "loss": 9.4904,
      "step": 834
    },
    {
      "epoch": 0.28803035529492926,
      "grad_norm": 1.5150681734085083,
      "learning_rate": 8.392921814867915e-05,
      "loss": 9.5391,
      "step": 835
    },
    {
      "epoch": 0.28837530182821663,
      "grad_norm": 1.4273500442504883,
      "learning_rate": 8.38879754494552e-05,
      "loss": 9.4018,
      "step": 836
    },
    {
      "epoch": 0.28872024836150395,
      "grad_norm": 1.4751886129379272,
      "learning_rate": 8.384669005891681e-05,
      "loss": 9.4735,
      "step": 837
    },
    {
      "epoch": 0.2890651948947913,
      "grad_norm": 1.4229878187179565,
      "learning_rate": 8.380536202907439e-05,
      "loss": 9.4534,
      "step": 838
    },
    {
      "epoch": 0.28941014142807864,
      "grad_norm": 1.3954477310180664,
      "learning_rate": 8.376399141199206e-05,
      "loss": 9.4548,
      "step": 839
    },
    {
      "epoch": 0.289755087961366,
      "grad_norm": 1.3925981521606445,
      "learning_rate": 8.372257825978762e-05,
      "loss": 9.4611,
      "step": 840
    },
    {
      "epoch": 0.2901000344946533,
      "grad_norm": 1.4793709516525269,
      "learning_rate": 8.368112262463244e-05,
      "loss": 9.4388,
      "step": 841
    },
    {
      "epoch": 0.2904449810279407,
      "grad_norm": 1.3460469245910645,
      "learning_rate": 8.363962455875136e-05,
      "loss": 9.5226,
      "step": 842
    },
    {
      "epoch": 0.290789927561228,
      "grad_norm": 1.399797797203064,
      "learning_rate": 8.359808411442277e-05,
      "loss": 9.485,
      "step": 843
    },
    {
      "epoch": 0.29113487409451533,
      "grad_norm": 1.6081780195236206,
      "learning_rate": 8.355650134397836e-05,
      "loss": 9.4068,
      "step": 844
    },
    {
      "epoch": 0.2914798206278027,
      "grad_norm": 1.4968883991241455,
      "learning_rate": 8.351487629980319e-05,
      "loss": 9.4218,
      "step": 845
    },
    {
      "epoch": 0.29182476716109,
      "grad_norm": 1.608202576637268,
      "learning_rate": 8.347320903433555e-05,
      "loss": 9.4663,
      "step": 846
    },
    {
      "epoch": 0.2921697136943774,
      "grad_norm": 1.6110312938690186,
      "learning_rate": 8.343149960006695e-05,
      "loss": 9.4781,
      "step": 847
    },
    {
      "epoch": 0.2925146602276647,
      "grad_norm": 1.6279586553573608,
      "learning_rate": 8.338974804954197e-05,
      "loss": 9.3999,
      "step": 848
    },
    {
      "epoch": 0.2928596067609521,
      "grad_norm": 1.801674723625183,
      "learning_rate": 8.33479544353583e-05,
      "loss": 9.4058,
      "step": 849
    },
    {
      "epoch": 0.2932045532942394,
      "grad_norm": 1.7849171161651611,
      "learning_rate": 8.330611881016661e-05,
      "loss": 9.3335,
      "step": 850
    },
    {
      "epoch": 0.29354949982752676,
      "grad_norm": 0.913504958152771,
      "learning_rate": 8.326424122667048e-05,
      "loss": 9.6455,
      "step": 851
    },
    {
      "epoch": 0.2938944463608141,
      "grad_norm": 1.0253827571868896,
      "learning_rate": 8.322232173762637e-05,
      "loss": 9.5369,
      "step": 852
    },
    {
      "epoch": 0.2942393928941014,
      "grad_norm": 0.8860166072845459,
      "learning_rate": 8.318036039584344e-05,
      "loss": 9.6341,
      "step": 853
    },
    {
      "epoch": 0.29458433942738876,
      "grad_norm": 1.1842451095581055,
      "learning_rate": 8.313835725418376e-05,
      "loss": 9.4935,
      "step": 854
    },
    {
      "epoch": 0.2949292859606761,
      "grad_norm": 1.193926215171814,
      "learning_rate": 8.309631236556189e-05,
      "loss": 9.5092,
      "step": 855
    },
    {
      "epoch": 0.29527423249396345,
      "grad_norm": 1.0631842613220215,
      "learning_rate": 8.305422578294504e-05,
      "loss": 9.6657,
      "step": 856
    },
    {
      "epoch": 0.29561917902725077,
      "grad_norm": 1.310213327407837,
      "learning_rate": 8.301209755935295e-05,
      "loss": 9.4685,
      "step": 857
    },
    {
      "epoch": 0.29596412556053814,
      "grad_norm": 1.3630036115646362,
      "learning_rate": 8.296992774785784e-05,
      "loss": 9.4808,
      "step": 858
    },
    {
      "epoch": 0.29630907209382545,
      "grad_norm": 1.0846993923187256,
      "learning_rate": 8.292771640158427e-05,
      "loss": 9.5889,
      "step": 859
    },
    {
      "epoch": 0.29665401862711277,
      "grad_norm": 1.2428042888641357,
      "learning_rate": 8.288546357370914e-05,
      "loss": 9.466,
      "step": 860
    },
    {
      "epoch": 0.29699896516040014,
      "grad_norm": 1.1466275453567505,
      "learning_rate": 8.284316931746166e-05,
      "loss": 9.5692,
      "step": 861
    },
    {
      "epoch": 0.29734391169368746,
      "grad_norm": 1.2471225261688232,
      "learning_rate": 8.280083368612315e-05,
      "loss": 9.5223,
      "step": 862
    },
    {
      "epoch": 0.29768885822697483,
      "grad_norm": 1.1912099123001099,
      "learning_rate": 8.275845673302713e-05,
      "loss": 9.5137,
      "step": 863
    },
    {
      "epoch": 0.29803380476026214,
      "grad_norm": 1.2651206254959106,
      "learning_rate": 8.271603851155911e-05,
      "loss": 9.4735,
      "step": 864
    },
    {
      "epoch": 0.2983787512935495,
      "grad_norm": 1.3276959657669067,
      "learning_rate": 8.267357907515661e-05,
      "loss": 9.4837,
      "step": 865
    },
    {
      "epoch": 0.29872369782683683,
      "grad_norm": 1.3006752729415894,
      "learning_rate": 8.263107847730911e-05,
      "loss": 9.5469,
      "step": 866
    },
    {
      "epoch": 0.2990686443601242,
      "grad_norm": 1.1855391263961792,
      "learning_rate": 8.25885367715579e-05,
      "loss": 9.5297,
      "step": 867
    },
    {
      "epoch": 0.2994135908934115,
      "grad_norm": 1.1859946250915527,
      "learning_rate": 8.254595401149605e-05,
      "loss": 9.5102,
      "step": 868
    },
    {
      "epoch": 0.29975853742669883,
      "grad_norm": 1.1786353588104248,
      "learning_rate": 8.250333025076842e-05,
      "loss": 9.5158,
      "step": 869
    },
    {
      "epoch": 0.3001034839599862,
      "grad_norm": 1.1845674514770508,
      "learning_rate": 8.246066554307141e-05,
      "loss": 9.5457,
      "step": 870
    },
    {
      "epoch": 0.3004484304932735,
      "grad_norm": 1.3296141624450684,
      "learning_rate": 8.241795994215312e-05,
      "loss": 9.5422,
      "step": 871
    },
    {
      "epoch": 0.3007933770265609,
      "grad_norm": 1.2888195514678955,
      "learning_rate": 8.237521350181308e-05,
      "loss": 9.4716,
      "step": 872
    },
    {
      "epoch": 0.3011383235598482,
      "grad_norm": 1.2534382343292236,
      "learning_rate": 8.233242627590232e-05,
      "loss": 9.4963,
      "step": 873
    },
    {
      "epoch": 0.3014832700931356,
      "grad_norm": 1.3657022714614868,
      "learning_rate": 8.22895983183232e-05,
      "loss": 9.5223,
      "step": 874
    },
    {
      "epoch": 0.3018282166264229,
      "grad_norm": 1.3497120141983032,
      "learning_rate": 8.224672968302946e-05,
      "loss": 9.5319,
      "step": 875
    },
    {
      "epoch": 0.30217316315971027,
      "grad_norm": 1.3509223461151123,
      "learning_rate": 8.220382042402604e-05,
      "loss": 9.4724,
      "step": 876
    },
    {
      "epoch": 0.3025181096929976,
      "grad_norm": 1.4304386377334595,
      "learning_rate": 8.216087059536903e-05,
      "loss": 9.5216,
      "step": 877
    },
    {
      "epoch": 0.3028630562262849,
      "grad_norm": 1.4425625801086426,
      "learning_rate": 8.211788025116571e-05,
      "loss": 9.5034,
      "step": 878
    },
    {
      "epoch": 0.30320800275957227,
      "grad_norm": 1.2742223739624023,
      "learning_rate": 8.207484944557436e-05,
      "loss": 9.5591,
      "step": 879
    },
    {
      "epoch": 0.3035529492928596,
      "grad_norm": 1.3797872066497803,
      "learning_rate": 8.20317782328042e-05,
      "loss": 9.5402,
      "step": 880
    },
    {
      "epoch": 0.30389789582614696,
      "grad_norm": 1.3795853853225708,
      "learning_rate": 8.198866666711535e-05,
      "loss": 9.5036,
      "step": 881
    },
    {
      "epoch": 0.3042428423594343,
      "grad_norm": 1.4435900449752808,
      "learning_rate": 8.194551480281888e-05,
      "loss": 9.4878,
      "step": 882
    },
    {
      "epoch": 0.30458778889272164,
      "grad_norm": 1.3728365898132324,
      "learning_rate": 8.19023226942765e-05,
      "loss": 9.5345,
      "step": 883
    },
    {
      "epoch": 0.30493273542600896,
      "grad_norm": 1.516736388206482,
      "learning_rate": 8.185909039590063e-05,
      "loss": 9.4699,
      "step": 884
    },
    {
      "epoch": 0.30527768195929633,
      "grad_norm": 1.4422507286071777,
      "learning_rate": 8.18158179621544e-05,
      "loss": 9.4944,
      "step": 885
    },
    {
      "epoch": 0.30562262849258365,
      "grad_norm": 1.424100637435913,
      "learning_rate": 8.177250544755146e-05,
      "loss": 9.4816,
      "step": 886
    },
    {
      "epoch": 0.30596757502587096,
      "grad_norm": 1.421716570854187,
      "learning_rate": 8.172915290665594e-05,
      "loss": 9.5022,
      "step": 887
    },
    {
      "epoch": 0.30631252155915834,
      "grad_norm": 1.4079738855361938,
      "learning_rate": 8.168576039408238e-05,
      "loss": 9.4492,
      "step": 888
    },
    {
      "epoch": 0.30665746809244565,
      "grad_norm": 1.3521206378936768,
      "learning_rate": 8.164232796449572e-05,
      "loss": 9.4899,
      "step": 889
    },
    {
      "epoch": 0.307002414625733,
      "grad_norm": 1.4199329614639282,
      "learning_rate": 8.159885567261114e-05,
      "loss": 9.4414,
      "step": 890
    },
    {
      "epoch": 0.30734736115902034,
      "grad_norm": 1.5063000917434692,
      "learning_rate": 8.155534357319408e-05,
      "loss": 9.4272,
      "step": 891
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 1.4825642108917236,
      "learning_rate": 8.151179172106012e-05,
      "loss": 9.4197,
      "step": 892
    },
    {
      "epoch": 0.308037254225595,
      "grad_norm": 1.5050302743911743,
      "learning_rate": 8.14682001710749e-05,
      "loss": 9.4654,
      "step": 893
    },
    {
      "epoch": 0.3083822007588824,
      "grad_norm": 1.4784218072891235,
      "learning_rate": 8.142456897815407e-05,
      "loss": 9.4737,
      "step": 894
    },
    {
      "epoch": 0.3087271472921697,
      "grad_norm": 1.6477797031402588,
      "learning_rate": 8.138089819726327e-05,
      "loss": 9.4824,
      "step": 895
    },
    {
      "epoch": 0.30907209382545703,
      "grad_norm": 1.5472941398620605,
      "learning_rate": 8.133718788341791e-05,
      "loss": 9.4556,
      "step": 896
    },
    {
      "epoch": 0.3094170403587444,
      "grad_norm": 1.5163496732711792,
      "learning_rate": 8.129343809168334e-05,
      "loss": 9.4168,
      "step": 897
    },
    {
      "epoch": 0.3097619868920317,
      "grad_norm": 1.538262128829956,
      "learning_rate": 8.12496488771745e-05,
      "loss": 9.437,
      "step": 898
    },
    {
      "epoch": 0.3101069334253191,
      "grad_norm": 1.7789909839630127,
      "learning_rate": 8.120582029505613e-05,
      "loss": 9.4236,
      "step": 899
    },
    {
      "epoch": 0.3104518799586064,
      "grad_norm": 1.8070483207702637,
      "learning_rate": 8.116195240054247e-05,
      "loss": 9.4368,
      "step": 900
    },
    {
      "epoch": 0.3107968264918938,
      "grad_norm": 0.9573423862457275,
      "learning_rate": 8.111804524889729e-05,
      "loss": 9.5679,
      "step": 901
    },
    {
      "epoch": 0.3111417730251811,
      "grad_norm": 1.020045518875122,
      "learning_rate": 8.107409889543386e-05,
      "loss": 9.5809,
      "step": 902
    },
    {
      "epoch": 0.31148671955846846,
      "grad_norm": 1.1108217239379883,
      "learning_rate": 8.103011339551482e-05,
      "loss": 9.5283,
      "step": 903
    },
    {
      "epoch": 0.3118316660917558,
      "grad_norm": 1.1079702377319336,
      "learning_rate": 8.098608880455213e-05,
      "loss": 9.5454,
      "step": 904
    },
    {
      "epoch": 0.3121766126250431,
      "grad_norm": 1.165088176727295,
      "learning_rate": 8.094202517800695e-05,
      "loss": 9.5713,
      "step": 905
    },
    {
      "epoch": 0.31252155915833046,
      "grad_norm": 1.1873505115509033,
      "learning_rate": 8.089792257138968e-05,
      "loss": 9.5299,
      "step": 906
    },
    {
      "epoch": 0.3128665056916178,
      "grad_norm": 1.1979339122772217,
      "learning_rate": 8.08537810402598e-05,
      "loss": 9.577,
      "step": 907
    },
    {
      "epoch": 0.31321145222490515,
      "grad_norm": 1.1789638996124268,
      "learning_rate": 8.080960064022581e-05,
      "loss": 9.6262,
      "step": 908
    },
    {
      "epoch": 0.31355639875819247,
      "grad_norm": 1.2188464403152466,
      "learning_rate": 8.076538142694518e-05,
      "loss": 9.553,
      "step": 909
    },
    {
      "epoch": 0.31390134529147984,
      "grad_norm": 1.1790921688079834,
      "learning_rate": 8.072112345612434e-05,
      "loss": 9.5114,
      "step": 910
    },
    {
      "epoch": 0.31424629182476715,
      "grad_norm": 1.131395697593689,
      "learning_rate": 8.067682678351845e-05,
      "loss": 9.5672,
      "step": 911
    },
    {
      "epoch": 0.3145912383580545,
      "grad_norm": 1.1688874959945679,
      "learning_rate": 8.063249146493145e-05,
      "loss": 9.5667,
      "step": 912
    },
    {
      "epoch": 0.31493618489134184,
      "grad_norm": 1.2182799577713013,
      "learning_rate": 8.058811755621603e-05,
      "loss": 9.5622,
      "step": 913
    },
    {
      "epoch": 0.31528113142462916,
      "grad_norm": 1.1979494094848633,
      "learning_rate": 8.054370511327342e-05,
      "loss": 9.4747,
      "step": 914
    },
    {
      "epoch": 0.31562607795791653,
      "grad_norm": 1.3254115581512451,
      "learning_rate": 8.049925419205344e-05,
      "loss": 9.5316,
      "step": 915
    },
    {
      "epoch": 0.31597102449120384,
      "grad_norm": 1.1189312934875488,
      "learning_rate": 8.045476484855434e-05,
      "loss": 9.5205,
      "step": 916
    },
    {
      "epoch": 0.3163159710244912,
      "grad_norm": 1.3546466827392578,
      "learning_rate": 8.041023713882286e-05,
      "loss": 9.5027,
      "step": 917
    },
    {
      "epoch": 0.31666091755777853,
      "grad_norm": 1.1889816522598267,
      "learning_rate": 8.036567111895394e-05,
      "loss": 9.5132,
      "step": 918
    },
    {
      "epoch": 0.3170058640910659,
      "grad_norm": 1.3260706663131714,
      "learning_rate": 8.032106684509091e-05,
      "loss": 9.4622,
      "step": 919
    },
    {
      "epoch": 0.3173508106243532,
      "grad_norm": 1.3710530996322632,
      "learning_rate": 8.027642437342521e-05,
      "loss": 9.5102,
      "step": 920
    },
    {
      "epoch": 0.3176957571576406,
      "grad_norm": 1.2274657487869263,
      "learning_rate": 8.023174376019644e-05,
      "loss": 9.5887,
      "step": 921
    },
    {
      "epoch": 0.3180407036909279,
      "grad_norm": 1.3517509698867798,
      "learning_rate": 8.018702506169223e-05,
      "loss": 9.4605,
      "step": 922
    },
    {
      "epoch": 0.3183856502242152,
      "grad_norm": 1.2345460653305054,
      "learning_rate": 8.014226833424823e-05,
      "loss": 9.5566,
      "step": 923
    },
    {
      "epoch": 0.3187305967575026,
      "grad_norm": 1.2032557725906372,
      "learning_rate": 8.009747363424791e-05,
      "loss": 9.5374,
      "step": 924
    },
    {
      "epoch": 0.3190755432907899,
      "grad_norm": 1.3755812644958496,
      "learning_rate": 8.005264101812268e-05,
      "loss": 9.4509,
      "step": 925
    },
    {
      "epoch": 0.3194204898240773,
      "grad_norm": 1.253843069076538,
      "learning_rate": 8.000777054235163e-05,
      "loss": 9.5665,
      "step": 926
    },
    {
      "epoch": 0.3197654363573646,
      "grad_norm": 1.4256430864334106,
      "learning_rate": 7.996286226346162e-05,
      "loss": 9.5317,
      "step": 927
    },
    {
      "epoch": 0.32011038289065197,
      "grad_norm": 1.4187809228897095,
      "learning_rate": 7.991791623802707e-05,
      "loss": 9.5435,
      "step": 928
    },
    {
      "epoch": 0.3204553294239393,
      "grad_norm": 1.3248265981674194,
      "learning_rate": 7.987293252266999e-05,
      "loss": 9.489,
      "step": 929
    },
    {
      "epoch": 0.32080027595722665,
      "grad_norm": 1.4751477241516113,
      "learning_rate": 7.982791117405986e-05,
      "loss": 9.4945,
      "step": 930
    },
    {
      "epoch": 0.32114522249051397,
      "grad_norm": 1.2307500839233398,
      "learning_rate": 7.978285224891356e-05,
      "loss": 9.538,
      "step": 931
    },
    {
      "epoch": 0.3214901690238013,
      "grad_norm": 1.5299421548843384,
      "learning_rate": 7.973775580399534e-05,
      "loss": 9.4755,
      "step": 932
    },
    {
      "epoch": 0.32183511555708866,
      "grad_norm": 1.2966649532318115,
      "learning_rate": 7.969262189611666e-05,
      "loss": 9.5272,
      "step": 933
    },
    {
      "epoch": 0.322180062090376,
      "grad_norm": 1.307608962059021,
      "learning_rate": 7.964745058213622e-05,
      "loss": 9.5482,
      "step": 934
    },
    {
      "epoch": 0.32252500862366335,
      "grad_norm": 1.349100947380066,
      "learning_rate": 7.960224191895987e-05,
      "loss": 9.4722,
      "step": 935
    },
    {
      "epoch": 0.32286995515695066,
      "grad_norm": 1.387831449508667,
      "learning_rate": 7.955699596354041e-05,
      "loss": 9.4829,
      "step": 936
    },
    {
      "epoch": 0.32321490169023803,
      "grad_norm": 1.3435271978378296,
      "learning_rate": 7.951171277287774e-05,
      "loss": 9.4139,
      "step": 937
    },
    {
      "epoch": 0.32355984822352535,
      "grad_norm": 1.3720812797546387,
      "learning_rate": 7.946639240401858e-05,
      "loss": 9.5294,
      "step": 938
    },
    {
      "epoch": 0.3239047947568127,
      "grad_norm": 1.3235864639282227,
      "learning_rate": 7.942103491405655e-05,
      "loss": 9.5735,
      "step": 939
    },
    {
      "epoch": 0.32424974129010004,
      "grad_norm": 1.387094259262085,
      "learning_rate": 7.937564036013194e-05,
      "loss": 9.5123,
      "step": 940
    },
    {
      "epoch": 0.32459468782338735,
      "grad_norm": 1.5132378339767456,
      "learning_rate": 7.933020879943185e-05,
      "loss": 9.4052,
      "step": 941
    },
    {
      "epoch": 0.3249396343566747,
      "grad_norm": 1.425180435180664,
      "learning_rate": 7.928474028918996e-05,
      "loss": 9.4817,
      "step": 942
    },
    {
      "epoch": 0.32528458088996204,
      "grad_norm": 1.4578827619552612,
      "learning_rate": 7.923923488668641e-05,
      "loss": 9.4453,
      "step": 943
    },
    {
      "epoch": 0.3256295274232494,
      "grad_norm": 1.6216663122177124,
      "learning_rate": 7.919369264924795e-05,
      "loss": 9.417,
      "step": 944
    },
    {
      "epoch": 0.3259744739565367,
      "grad_norm": 1.4772353172302246,
      "learning_rate": 7.914811363424765e-05,
      "loss": 9.512,
      "step": 945
    },
    {
      "epoch": 0.3263194204898241,
      "grad_norm": 1.6448122262954712,
      "learning_rate": 7.910249789910495e-05,
      "loss": 9.4389,
      "step": 946
    },
    {
      "epoch": 0.3266643670231114,
      "grad_norm": 1.4505155086517334,
      "learning_rate": 7.905684550128553e-05,
      "loss": 9.4267,
      "step": 947
    },
    {
      "epoch": 0.3270093135563988,
      "grad_norm": 1.6771926879882812,
      "learning_rate": 7.901115649830125e-05,
      "loss": 9.4435,
      "step": 948
    },
    {
      "epoch": 0.3273542600896861,
      "grad_norm": 1.53878915309906,
      "learning_rate": 7.896543094771011e-05,
      "loss": 9.4784,
      "step": 949
    },
    {
      "epoch": 0.3276992066229734,
      "grad_norm": 1.7891881465911865,
      "learning_rate": 7.89196689071161e-05,
      "loss": 9.4134,
      "step": 950
    },
    {
      "epoch": 0.3280441531562608,
      "grad_norm": 0.9297865033149719,
      "learning_rate": 7.887387043416927e-05,
      "loss": 9.5925,
      "step": 951
    },
    {
      "epoch": 0.3283890996895481,
      "grad_norm": 1.1696343421936035,
      "learning_rate": 7.882803558656549e-05,
      "loss": 9.5268,
      "step": 952
    },
    {
      "epoch": 0.3287340462228355,
      "grad_norm": 1.1735519170761108,
      "learning_rate": 7.878216442204647e-05,
      "loss": 9.4896,
      "step": 953
    },
    {
      "epoch": 0.3290789927561228,
      "grad_norm": 1.115681529045105,
      "learning_rate": 7.873625699839968e-05,
      "loss": 9.5907,
      "step": 954
    },
    {
      "epoch": 0.32942393928941016,
      "grad_norm": 1.0858581066131592,
      "learning_rate": 7.869031337345828e-05,
      "loss": 9.5537,
      "step": 955
    },
    {
      "epoch": 0.3297688858226975,
      "grad_norm": 1.0331652164459229,
      "learning_rate": 7.864433360510101e-05,
      "loss": 9.6112,
      "step": 956
    },
    {
      "epoch": 0.33011383235598485,
      "grad_norm": 1.1334502696990967,
      "learning_rate": 7.859831775125214e-05,
      "loss": 9.5681,
      "step": 957
    },
    {
      "epoch": 0.33045877888927216,
      "grad_norm": 1.0670603513717651,
      "learning_rate": 7.855226586988145e-05,
      "loss": 9.5747,
      "step": 958
    },
    {
      "epoch": 0.3308037254225595,
      "grad_norm": 1.1903663873672485,
      "learning_rate": 7.850617801900406e-05,
      "loss": 9.5893,
      "step": 959
    },
    {
      "epoch": 0.33114867195584685,
      "grad_norm": 1.1031749248504639,
      "learning_rate": 7.846005425668041e-05,
      "loss": 9.547,
      "step": 960
    },
    {
      "epoch": 0.33149361848913417,
      "grad_norm": 1.259435772895813,
      "learning_rate": 7.841389464101616e-05,
      "loss": 9.5292,
      "step": 961
    },
    {
      "epoch": 0.33183856502242154,
      "grad_norm": 1.1308194398880005,
      "learning_rate": 7.83676992301622e-05,
      "loss": 9.5453,
      "step": 962
    },
    {
      "epoch": 0.33218351155570885,
      "grad_norm": 1.3275343179702759,
      "learning_rate": 7.832146808231447e-05,
      "loss": 9.4979,
      "step": 963
    },
    {
      "epoch": 0.3325284580889962,
      "grad_norm": 1.1660116910934448,
      "learning_rate": 7.827520125571392e-05,
      "loss": 9.6073,
      "step": 964
    },
    {
      "epoch": 0.33287340462228354,
      "grad_norm": 1.347019076347351,
      "learning_rate": 7.822889880864649e-05,
      "loss": 9.5012,
      "step": 965
    },
    {
      "epoch": 0.3332183511555709,
      "grad_norm": 1.230370044708252,
      "learning_rate": 7.818256079944295e-05,
      "loss": 9.5268,
      "step": 966
    },
    {
      "epoch": 0.33356329768885823,
      "grad_norm": 1.2594881057739258,
      "learning_rate": 7.813618728647889e-05,
      "loss": 9.5,
      "step": 967
    },
    {
      "epoch": 0.33390824422214554,
      "grad_norm": 1.334810733795166,
      "learning_rate": 7.808977832817462e-05,
      "loss": 9.5172,
      "step": 968
    },
    {
      "epoch": 0.3342531907554329,
      "grad_norm": 1.1402246952056885,
      "learning_rate": 7.804333398299512e-05,
      "loss": 9.51,
      "step": 969
    },
    {
      "epoch": 0.33459813728872023,
      "grad_norm": 1.3126106262207031,
      "learning_rate": 7.799685430944995e-05,
      "loss": 9.5127,
      "step": 970
    },
    {
      "epoch": 0.3349430838220076,
      "grad_norm": 1.2847175598144531,
      "learning_rate": 7.795033936609314e-05,
      "loss": 9.5261,
      "step": 971
    },
    {
      "epoch": 0.3352880303552949,
      "grad_norm": 1.115410566329956,
      "learning_rate": 7.790378921152319e-05,
      "loss": 9.5918,
      "step": 972
    },
    {
      "epoch": 0.3356329768885823,
      "grad_norm": 1.4567341804504395,
      "learning_rate": 7.785720390438293e-05,
      "loss": 9.5061,
      "step": 973
    },
    {
      "epoch": 0.3359779234218696,
      "grad_norm": 1.3386863470077515,
      "learning_rate": 7.781058350335952e-05,
      "loss": 9.5209,
      "step": 974
    },
    {
      "epoch": 0.336322869955157,
      "grad_norm": 1.2178325653076172,
      "learning_rate": 7.776392806718427e-05,
      "loss": 9.582,
      "step": 975
    },
    {
      "epoch": 0.3366678164884443,
      "grad_norm": 1.3879817724227905,
      "learning_rate": 7.771723765463267e-05,
      "loss": 9.4433,
      "step": 976
    },
    {
      "epoch": 0.3370127630217316,
      "grad_norm": 1.2487505674362183,
      "learning_rate": 7.767051232452426e-05,
      "loss": 9.4916,
      "step": 977
    },
    {
      "epoch": 0.337357709555019,
      "grad_norm": 1.3047330379486084,
      "learning_rate": 7.762375213572254e-05,
      "loss": 9.5346,
      "step": 978
    },
    {
      "epoch": 0.3377026560883063,
      "grad_norm": 1.3577327728271484,
      "learning_rate": 7.7576957147135e-05,
      "loss": 9.4706,
      "step": 979
    },
    {
      "epoch": 0.33804760262159367,
      "grad_norm": 1.3883907794952393,
      "learning_rate": 7.753012741771288e-05,
      "loss": 9.4835,
      "step": 980
    },
    {
      "epoch": 0.338392549154881,
      "grad_norm": 1.4198276996612549,
      "learning_rate": 7.748326300645126e-05,
      "loss": 9.4446,
      "step": 981
    },
    {
      "epoch": 0.33873749568816836,
      "grad_norm": 1.4020169973373413,
      "learning_rate": 7.743636397238884e-05,
      "loss": 9.4671,
      "step": 982
    },
    {
      "epoch": 0.33908244222145567,
      "grad_norm": 1.375780463218689,
      "learning_rate": 7.738943037460802e-05,
      "loss": 9.4689,
      "step": 983
    },
    {
      "epoch": 0.33942738875474304,
      "grad_norm": 1.4482146501541138,
      "learning_rate": 7.734246227223464e-05,
      "loss": 9.451,
      "step": 984
    },
    {
      "epoch": 0.33977233528803036,
      "grad_norm": 1.3532198667526245,
      "learning_rate": 7.729545972443812e-05,
      "loss": 9.522,
      "step": 985
    },
    {
      "epoch": 0.3401172818213177,
      "grad_norm": 1.3163398504257202,
      "learning_rate": 7.724842279043121e-05,
      "loss": 9.4935,
      "step": 986
    },
    {
      "epoch": 0.34046222835460505,
      "grad_norm": 1.2999087572097778,
      "learning_rate": 7.720135152946997e-05,
      "loss": 9.5352,
      "step": 987
    },
    {
      "epoch": 0.34080717488789236,
      "grad_norm": 1.3928954601287842,
      "learning_rate": 7.715424600085373e-05,
      "loss": 9.4169,
      "step": 988
    },
    {
      "epoch": 0.34115212142117973,
      "grad_norm": 1.4478076696395874,
      "learning_rate": 7.710710626392499e-05,
      "loss": 9.4649,
      "step": 989
    },
    {
      "epoch": 0.34149706795446705,
      "grad_norm": 1.4046027660369873,
      "learning_rate": 7.705993237806932e-05,
      "loss": 9.4089,
      "step": 990
    },
    {
      "epoch": 0.3418420144877544,
      "grad_norm": 1.5680257081985474,
      "learning_rate": 7.701272440271534e-05,
      "loss": 9.4567,
      "step": 991
    },
    {
      "epoch": 0.34218696102104174,
      "grad_norm": 1.5575940608978271,
      "learning_rate": 7.696548239733461e-05,
      "loss": 9.4105,
      "step": 992
    },
    {
      "epoch": 0.34253190755432905,
      "grad_norm": 1.539071798324585,
      "learning_rate": 7.691820642144154e-05,
      "loss": 9.4895,
      "step": 993
    },
    {
      "epoch": 0.3428768540876164,
      "grad_norm": 1.5203800201416016,
      "learning_rate": 7.687089653459336e-05,
      "loss": 9.4102,
      "step": 994
    },
    {
      "epoch": 0.34322180062090374,
      "grad_norm": 1.6048343181610107,
      "learning_rate": 7.682355279639e-05,
      "loss": 9.4627,
      "step": 995
    },
    {
      "epoch": 0.3435667471541911,
      "grad_norm": 1.4530948400497437,
      "learning_rate": 7.677617526647403e-05,
      "loss": 9.4021,
      "step": 996
    },
    {
      "epoch": 0.3439116936874784,
      "grad_norm": 1.6160320043563843,
      "learning_rate": 7.672876400453063e-05,
      "loss": 9.3765,
      "step": 997
    },
    {
      "epoch": 0.3442566402207658,
      "grad_norm": 1.5577583312988281,
      "learning_rate": 7.668131907028746e-05,
      "loss": 9.396,
      "step": 998
    },
    {
      "epoch": 0.3446015867540531,
      "grad_norm": 1.5688533782958984,
      "learning_rate": 7.663384052351454e-05,
      "loss": 9.399,
      "step": 999
    },
    {
      "epoch": 0.3449465332873405,
      "grad_norm": 1.8562169075012207,
      "learning_rate": 7.658632842402432e-05,
      "loss": 9.336,
      "step": 1000
    },
    {
      "epoch": 0.3452914798206278,
      "grad_norm": 0.8915877342224121,
      "learning_rate": 7.653878283167148e-05,
      "loss": 9.5921,
      "step": 1001
    },
    {
      "epoch": 0.3456364263539151,
      "grad_norm": 1.0582910776138306,
      "learning_rate": 7.649120380635287e-05,
      "loss": 9.5403,
      "step": 1002
    },
    {
      "epoch": 0.3459813728872025,
      "grad_norm": 1.0659931898117065,
      "learning_rate": 7.64435914080075e-05,
      "loss": 9.4967,
      "step": 1003
    },
    {
      "epoch": 0.3463263194204898,
      "grad_norm": 1.1160966157913208,
      "learning_rate": 7.639594569661641e-05,
      "loss": 9.5211,
      "step": 1004
    },
    {
      "epoch": 0.3466712659537772,
      "grad_norm": 1.0769208669662476,
      "learning_rate": 7.634826673220259e-05,
      "loss": 9.5474,
      "step": 1005
    },
    {
      "epoch": 0.3470162124870645,
      "grad_norm": 1.1200584173202515,
      "learning_rate": 7.630055457483093e-05,
      "loss": 9.5184,
      "step": 1006
    },
    {
      "epoch": 0.34736115902035186,
      "grad_norm": 1.1214946508407593,
      "learning_rate": 7.625280928460819e-05,
      "loss": 9.5051,
      "step": 1007
    },
    {
      "epoch": 0.3477061055536392,
      "grad_norm": 1.2545450925827026,
      "learning_rate": 7.620503092168277e-05,
      "loss": 9.5243,
      "step": 1008
    },
    {
      "epoch": 0.34805105208692655,
      "grad_norm": 1.102694034576416,
      "learning_rate": 7.615721954624477e-05,
      "loss": 9.5558,
      "step": 1009
    },
    {
      "epoch": 0.34839599862021386,
      "grad_norm": 1.1054221391677856,
      "learning_rate": 7.610937521852594e-05,
      "loss": 9.5215,
      "step": 1010
    },
    {
      "epoch": 0.3487409451535012,
      "grad_norm": 1.1642522811889648,
      "learning_rate": 7.606149799879947e-05,
      "loss": 9.5085,
      "step": 1011
    },
    {
      "epoch": 0.34908589168678855,
      "grad_norm": 1.163087010383606,
      "learning_rate": 7.601358794738002e-05,
      "loss": 9.532,
      "step": 1012
    },
    {
      "epoch": 0.34943083822007587,
      "grad_norm": 1.2257179021835327,
      "learning_rate": 7.59656451246236e-05,
      "loss": 9.5124,
      "step": 1013
    },
    {
      "epoch": 0.34977578475336324,
      "grad_norm": 1.1685041189193726,
      "learning_rate": 7.59176695909275e-05,
      "loss": 9.5394,
      "step": 1014
    },
    {
      "epoch": 0.35012073128665055,
      "grad_norm": 1.3486417531967163,
      "learning_rate": 7.586966140673024e-05,
      "loss": 9.5435,
      "step": 1015
    },
    {
      "epoch": 0.3504656778199379,
      "grad_norm": 1.178091049194336,
      "learning_rate": 7.582162063251144e-05,
      "loss": 9.4923,
      "step": 1016
    },
    {
      "epoch": 0.35081062435322524,
      "grad_norm": 1.2380973100662231,
      "learning_rate": 7.57735473287918e-05,
      "loss": 9.5371,
      "step": 1017
    },
    {
      "epoch": 0.3511555708865126,
      "grad_norm": 1.2036174535751343,
      "learning_rate": 7.572544155613299e-05,
      "loss": 9.5282,
      "step": 1018
    },
    {
      "epoch": 0.35150051741979993,
      "grad_norm": 1.240722417831421,
      "learning_rate": 7.567730337513762e-05,
      "loss": 9.5513,
      "step": 1019
    },
    {
      "epoch": 0.35184546395308725,
      "grad_norm": 1.3504445552825928,
      "learning_rate": 7.562913284644906e-05,
      "loss": 9.5439,
      "step": 1020
    },
    {
      "epoch": 0.3521904104863746,
      "grad_norm": 1.3790074586868286,
      "learning_rate": 7.558093003075146e-05,
      "loss": 9.4826,
      "step": 1021
    },
    {
      "epoch": 0.35253535701966193,
      "grad_norm": 1.3520622253417969,
      "learning_rate": 7.55326949887697e-05,
      "loss": 9.4962,
      "step": 1022
    },
    {
      "epoch": 0.3528803035529493,
      "grad_norm": 1.0791302919387817,
      "learning_rate": 7.548442778126916e-05,
      "loss": 9.5365,
      "step": 1023
    },
    {
      "epoch": 0.3532252500862366,
      "grad_norm": 1.5069971084594727,
      "learning_rate": 7.543612846905579e-05,
      "loss": 9.5134,
      "step": 1024
    },
    {
      "epoch": 0.353570196619524,
      "grad_norm": 1.3267474174499512,
      "learning_rate": 7.538779711297601e-05,
      "loss": 9.4633,
      "step": 1025
    },
    {
      "epoch": 0.3539151431528113,
      "grad_norm": 1.2925893068313599,
      "learning_rate": 7.533943377391658e-05,
      "loss": 9.512,
      "step": 1026
    },
    {
      "epoch": 0.3542600896860987,
      "grad_norm": 1.2562297582626343,
      "learning_rate": 7.529103851280454e-05,
      "loss": 9.5156,
      "step": 1027
    },
    {
      "epoch": 0.354605036219386,
      "grad_norm": 1.2134735584259033,
      "learning_rate": 7.524261139060715e-05,
      "loss": 9.5536,
      "step": 1028
    },
    {
      "epoch": 0.3549499827526733,
      "grad_norm": 1.3246090412139893,
      "learning_rate": 7.519415246833186e-05,
      "loss": 9.4857,
      "step": 1029
    },
    {
      "epoch": 0.3552949292859607,
      "grad_norm": 1.234519362449646,
      "learning_rate": 7.514566180702609e-05,
      "loss": 9.4969,
      "step": 1030
    },
    {
      "epoch": 0.355639875819248,
      "grad_norm": 1.3051451444625854,
      "learning_rate": 7.509713946777731e-05,
      "loss": 9.5441,
      "step": 1031
    },
    {
      "epoch": 0.35598482235253537,
      "grad_norm": 1.420652985572815,
      "learning_rate": 7.504858551171289e-05,
      "loss": 9.4662,
      "step": 1032
    },
    {
      "epoch": 0.3563297688858227,
      "grad_norm": 1.3639212846755981,
      "learning_rate": 7.500000000000001e-05,
      "loss": 9.4872,
      "step": 1033
    },
    {
      "epoch": 0.35667471541911006,
      "grad_norm": 1.3951948881149292,
      "learning_rate": 7.49513829938456e-05,
      "loss": 9.4484,
      "step": 1034
    },
    {
      "epoch": 0.35701966195239737,
      "grad_norm": 1.2805373668670654,
      "learning_rate": 7.49027345544963e-05,
      "loss": 9.4841,
      "step": 1035
    },
    {
      "epoch": 0.35736460848568474,
      "grad_norm": 1.3282732963562012,
      "learning_rate": 7.485405474323832e-05,
      "loss": 9.4856,
      "step": 1036
    },
    {
      "epoch": 0.35770955501897206,
      "grad_norm": 1.301007866859436,
      "learning_rate": 7.48053436213974e-05,
      "loss": 9.4881,
      "step": 1037
    },
    {
      "epoch": 0.3580545015522594,
      "grad_norm": 1.2560900449752808,
      "learning_rate": 7.47566012503387e-05,
      "loss": 9.5065,
      "step": 1038
    },
    {
      "epoch": 0.35839944808554675,
      "grad_norm": 1.4078426361083984,
      "learning_rate": 7.470782769146679e-05,
      "loss": 9.4497,
      "step": 1039
    },
    {
      "epoch": 0.35874439461883406,
      "grad_norm": 1.4189677238464355,
      "learning_rate": 7.465902300622552e-05,
      "loss": 9.5039,
      "step": 1040
    },
    {
      "epoch": 0.35908934115212143,
      "grad_norm": 1.4221440553665161,
      "learning_rate": 7.461018725609794e-05,
      "loss": 9.4685,
      "step": 1041
    },
    {
      "epoch": 0.35943428768540875,
      "grad_norm": 1.4293962717056274,
      "learning_rate": 7.456132050260622e-05,
      "loss": 9.4399,
      "step": 1042
    },
    {
      "epoch": 0.3597792342186961,
      "grad_norm": 1.5681432485580444,
      "learning_rate": 7.451242280731162e-05,
      "loss": 9.4055,
      "step": 1043
    },
    {
      "epoch": 0.36012418075198344,
      "grad_norm": 1.488181471824646,
      "learning_rate": 7.446349423181435e-05,
      "loss": 9.4348,
      "step": 1044
    },
    {
      "epoch": 0.3604691272852708,
      "grad_norm": 1.6025547981262207,
      "learning_rate": 7.441453483775354e-05,
      "loss": 9.4014,
      "step": 1045
    },
    {
      "epoch": 0.3608140738185581,
      "grad_norm": 1.429567575454712,
      "learning_rate": 7.436554468680714e-05,
      "loss": 9.4213,
      "step": 1046
    },
    {
      "epoch": 0.36115902035184544,
      "grad_norm": 1.5469809770584106,
      "learning_rate": 7.431652384069186e-05,
      "loss": 9.4063,
      "step": 1047
    },
    {
      "epoch": 0.3615039668851328,
      "grad_norm": 1.6328766345977783,
      "learning_rate": 7.426747236116305e-05,
      "loss": 9.3885,
      "step": 1048
    },
    {
      "epoch": 0.3618489134184201,
      "grad_norm": 1.6448739767074585,
      "learning_rate": 7.42183903100147e-05,
      "loss": 9.4563,
      "step": 1049
    },
    {
      "epoch": 0.3621938599517075,
      "grad_norm": 1.7313497066497803,
      "learning_rate": 7.416927774907923e-05,
      "loss": 9.3958,
      "step": 1050
    },
    {
      "epoch": 0.3625388064849948,
      "grad_norm": 0.9497975707054138,
      "learning_rate": 7.41201347402276e-05,
      "loss": 9.6031,
      "step": 1051
    },
    {
      "epoch": 0.3628837530182822,
      "grad_norm": 0.9865735173225403,
      "learning_rate": 7.407096134536903e-05,
      "loss": 9.5727,
      "step": 1052
    },
    {
      "epoch": 0.3632286995515695,
      "grad_norm": 0.9498660564422607,
      "learning_rate": 7.40217576264511e-05,
      "loss": 9.6308,
      "step": 1053
    },
    {
      "epoch": 0.36357364608485687,
      "grad_norm": 1.051511287689209,
      "learning_rate": 7.397252364545954e-05,
      "loss": 9.6518,
      "step": 1054
    },
    {
      "epoch": 0.3639185926181442,
      "grad_norm": 1.1738197803497314,
      "learning_rate": 7.392325946441822e-05,
      "loss": 9.4939,
      "step": 1055
    },
    {
      "epoch": 0.3642635391514315,
      "grad_norm": 1.02397882938385,
      "learning_rate": 7.387396514538904e-05,
      "loss": 9.608,
      "step": 1056
    },
    {
      "epoch": 0.3646084856847189,
      "grad_norm": 1.2825227975845337,
      "learning_rate": 7.382464075047192e-05,
      "loss": 9.5337,
      "step": 1057
    },
    {
      "epoch": 0.3649534322180062,
      "grad_norm": 1.1852742433547974,
      "learning_rate": 7.37752863418046e-05,
      "loss": 9.5027,
      "step": 1058
    },
    {
      "epoch": 0.36529837875129356,
      "grad_norm": 1.1019452810287476,
      "learning_rate": 7.372590198156266e-05,
      "loss": 9.4872,
      "step": 1059
    },
    {
      "epoch": 0.3656433252845809,
      "grad_norm": 1.1270240545272827,
      "learning_rate": 7.367648773195944e-05,
      "loss": 9.6016,
      "step": 1060
    },
    {
      "epoch": 0.36598827181786825,
      "grad_norm": 1.2325217723846436,
      "learning_rate": 7.362704365524586e-05,
      "loss": 9.4971,
      "step": 1061
    },
    {
      "epoch": 0.36633321835115557,
      "grad_norm": 1.1359938383102417,
      "learning_rate": 7.357756981371052e-05,
      "loss": 9.4867,
      "step": 1062
    },
    {
      "epoch": 0.36667816488444294,
      "grad_norm": 1.264073133468628,
      "learning_rate": 7.352806626967943e-05,
      "loss": 9.4759,
      "step": 1063
    },
    {
      "epoch": 0.36702311141773025,
      "grad_norm": 1.2164510488510132,
      "learning_rate": 7.347853308551604e-05,
      "loss": 9.5165,
      "step": 1064
    },
    {
      "epoch": 0.36736805795101757,
      "grad_norm": 1.2314850091934204,
      "learning_rate": 7.342897032362118e-05,
      "loss": 9.5585,
      "step": 1065
    },
    {
      "epoch": 0.36771300448430494,
      "grad_norm": 1.1831406354904175,
      "learning_rate": 7.33793780464329e-05,
      "loss": 9.5417,
      "step": 1066
    },
    {
      "epoch": 0.36805795101759226,
      "grad_norm": 1.194286823272705,
      "learning_rate": 7.332975631642641e-05,
      "loss": 9.5301,
      "step": 1067
    },
    {
      "epoch": 0.3684028975508796,
      "grad_norm": 1.2886695861816406,
      "learning_rate": 7.32801051961141e-05,
      "loss": 9.4868,
      "step": 1068
    },
    {
      "epoch": 0.36874784408416694,
      "grad_norm": 1.3282010555267334,
      "learning_rate": 7.323042474804532e-05,
      "loss": 9.5042,
      "step": 1069
    },
    {
      "epoch": 0.3690927906174543,
      "grad_norm": 1.2334239482879639,
      "learning_rate": 7.318071503480639e-05,
      "loss": 9.5492,
      "step": 1070
    },
    {
      "epoch": 0.36943773715074163,
      "grad_norm": 1.3581160306930542,
      "learning_rate": 7.31309761190205e-05,
      "loss": 9.5154,
      "step": 1071
    },
    {
      "epoch": 0.369782683684029,
      "grad_norm": 1.441379427909851,
      "learning_rate": 7.308120806334761e-05,
      "loss": 9.4175,
      "step": 1072
    },
    {
      "epoch": 0.3701276302173163,
      "grad_norm": 1.3497183322906494,
      "learning_rate": 7.303141093048445e-05,
      "loss": 9.5024,
      "step": 1073
    },
    {
      "epoch": 0.37047257675060363,
      "grad_norm": 1.3548775911331177,
      "learning_rate": 7.298158478316429e-05,
      "loss": 9.5069,
      "step": 1074
    },
    {
      "epoch": 0.370817523283891,
      "grad_norm": 1.3406723737716675,
      "learning_rate": 7.2931729684157e-05,
      "loss": 9.5365,
      "step": 1075
    },
    {
      "epoch": 0.3711624698171783,
      "grad_norm": 1.3262616395950317,
      "learning_rate": 7.288184569626895e-05,
      "loss": 9.4802,
      "step": 1076
    },
    {
      "epoch": 0.3715074163504657,
      "grad_norm": 1.2792712450027466,
      "learning_rate": 7.283193288234286e-05,
      "loss": 9.4721,
      "step": 1077
    },
    {
      "epoch": 0.371852362883753,
      "grad_norm": 1.2492791414260864,
      "learning_rate": 7.278199130525776e-05,
      "loss": 9.5401,
      "step": 1078
    },
    {
      "epoch": 0.3721973094170404,
      "grad_norm": 1.3007467985153198,
      "learning_rate": 7.273202102792894e-05,
      "loss": 9.4928,
      "step": 1079
    },
    {
      "epoch": 0.3725422559503277,
      "grad_norm": 1.2172558307647705,
      "learning_rate": 7.268202211330787e-05,
      "loss": 9.4789,
      "step": 1080
    },
    {
      "epoch": 0.37288720248361507,
      "grad_norm": 1.348846673965454,
      "learning_rate": 7.263199462438203e-05,
      "loss": 9.5034,
      "step": 1081
    },
    {
      "epoch": 0.3732321490169024,
      "grad_norm": 1.4215086698532104,
      "learning_rate": 7.258193862417494e-05,
      "loss": 9.4715,
      "step": 1082
    },
    {
      "epoch": 0.3735770955501897,
      "grad_norm": 1.390318751335144,
      "learning_rate": 7.253185417574604e-05,
      "loss": 9.5005,
      "step": 1083
    },
    {
      "epoch": 0.37392204208347707,
      "grad_norm": 1.3486655950546265,
      "learning_rate": 7.248174134219058e-05,
      "loss": 9.4772,
      "step": 1084
    },
    {
      "epoch": 0.3742669886167644,
      "grad_norm": 1.4781957864761353,
      "learning_rate": 7.243160018663962e-05,
      "loss": 9.4457,
      "step": 1085
    },
    {
      "epoch": 0.37461193515005176,
      "grad_norm": 1.3520734310150146,
      "learning_rate": 7.238143077225983e-05,
      "loss": 9.5074,
      "step": 1086
    },
    {
      "epoch": 0.37495688168333907,
      "grad_norm": 1.4090464115142822,
      "learning_rate": 7.233123316225357e-05,
      "loss": 9.425,
      "step": 1087
    },
    {
      "epoch": 0.37530182821662644,
      "grad_norm": 1.2668166160583496,
      "learning_rate": 7.228100741985861e-05,
      "loss": 9.4516,
      "step": 1088
    },
    {
      "epoch": 0.37564677474991376,
      "grad_norm": 1.4521373510360718,
      "learning_rate": 7.223075360834827e-05,
      "loss": 9.4834,
      "step": 1089
    },
    {
      "epoch": 0.37599172128320113,
      "grad_norm": 1.2660518884658813,
      "learning_rate": 7.218047179103112e-05,
      "loss": 9.5091,
      "step": 1090
    },
    {
      "epoch": 0.37633666781648845,
      "grad_norm": 1.4445570707321167,
      "learning_rate": 7.213016203125113e-05,
      "loss": 9.5,
      "step": 1091
    },
    {
      "epoch": 0.37668161434977576,
      "grad_norm": 1.3990899324417114,
      "learning_rate": 7.207982439238739e-05,
      "loss": 9.4546,
      "step": 1092
    },
    {
      "epoch": 0.37702656088306313,
      "grad_norm": 1.2153910398483276,
      "learning_rate": 7.20294589378541e-05,
      "loss": 9.5494,
      "step": 1093
    },
    {
      "epoch": 0.37737150741635045,
      "grad_norm": 1.382103681564331,
      "learning_rate": 7.197906573110059e-05,
      "loss": 9.419,
      "step": 1094
    },
    {
      "epoch": 0.3777164539496378,
      "grad_norm": 1.4809362888336182,
      "learning_rate": 7.192864483561103e-05,
      "loss": 9.4376,
      "step": 1095
    },
    {
      "epoch": 0.37806140048292514,
      "grad_norm": 1.5676347017288208,
      "learning_rate": 7.18781963149046e-05,
      "loss": 9.4146,
      "step": 1096
    },
    {
      "epoch": 0.3784063470162125,
      "grad_norm": 1.551276683807373,
      "learning_rate": 7.18277202325352e-05,
      "loss": 9.4692,
      "step": 1097
    },
    {
      "epoch": 0.3787512935494998,
      "grad_norm": 1.5128159523010254,
      "learning_rate": 7.177721665209146e-05,
      "loss": 9.4816,
      "step": 1098
    },
    {
      "epoch": 0.3790962400827872,
      "grad_norm": 1.534871220588684,
      "learning_rate": 7.172668563719667e-05,
      "loss": 9.4842,
      "step": 1099
    },
    {
      "epoch": 0.3794411866160745,
      "grad_norm": 1.7613797187805176,
      "learning_rate": 7.167612725150867e-05,
      "loss": 9.4056,
      "step": 1100
    },
    {
      "epoch": 0.3797861331493618,
      "grad_norm": 0.8986926674842834,
      "learning_rate": 7.162554155871979e-05,
      "loss": 9.6347,
      "step": 1101
    },
    {
      "epoch": 0.3801310796826492,
      "grad_norm": 1.0825634002685547,
      "learning_rate": 7.157492862255676e-05,
      "loss": 9.5473,
      "step": 1102
    },
    {
      "epoch": 0.3804760262159365,
      "grad_norm": 1.0305083990097046,
      "learning_rate": 7.152428850678061e-05,
      "loss": 9.5789,
      "step": 1103
    },
    {
      "epoch": 0.3808209727492239,
      "grad_norm": 1.0649081468582153,
      "learning_rate": 7.147362127518664e-05,
      "loss": 9.5759,
      "step": 1104
    },
    {
      "epoch": 0.3811659192825112,
      "grad_norm": 1.0573383569717407,
      "learning_rate": 7.14229269916043e-05,
      "loss": 9.5686,
      "step": 1105
    },
    {
      "epoch": 0.38151086581579857,
      "grad_norm": 1.024278998374939,
      "learning_rate": 7.137220571989708e-05,
      "loss": 9.5416,
      "step": 1106
    },
    {
      "epoch": 0.3818558123490859,
      "grad_norm": 1.1951700448989868,
      "learning_rate": 7.132145752396257e-05,
      "loss": 9.5353,
      "step": 1107
    },
    {
      "epoch": 0.38220075888237326,
      "grad_norm": 1.1564604043960571,
      "learning_rate": 7.127068246773215e-05,
      "loss": 9.5338,
      "step": 1108
    },
    {
      "epoch": 0.3825457054156606,
      "grad_norm": 1.0238999128341675,
      "learning_rate": 7.121988061517115e-05,
      "loss": 9.6087,
      "step": 1109
    },
    {
      "epoch": 0.3828906519489479,
      "grad_norm": 1.205392599105835,
      "learning_rate": 7.116905203027859e-05,
      "loss": 9.4998,
      "step": 1110
    },
    {
      "epoch": 0.38323559848223526,
      "grad_norm": 1.1162025928497314,
      "learning_rate": 7.111819677708716e-05,
      "loss": 9.5259,
      "step": 1111
    },
    {
      "epoch": 0.3835805450155226,
      "grad_norm": 1.2404072284698486,
      "learning_rate": 7.106731491966321e-05,
      "loss": 9.525,
      "step": 1112
    },
    {
      "epoch": 0.38392549154880995,
      "grad_norm": 1.1695785522460938,
      "learning_rate": 7.101640652210654e-05,
      "loss": 9.4944,
      "step": 1113
    },
    {
      "epoch": 0.38427043808209727,
      "grad_norm": 1.082716703414917,
      "learning_rate": 7.096547164855044e-05,
      "loss": 9.5486,
      "step": 1114
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 1.2770994901657104,
      "learning_rate": 7.09145103631615e-05,
      "loss": 9.4949,
      "step": 1115
    },
    {
      "epoch": 0.38496033114867195,
      "grad_norm": 1.373692512512207,
      "learning_rate": 7.086352273013963e-05,
      "loss": 9.4809,
      "step": 1116
    },
    {
      "epoch": 0.3853052776819593,
      "grad_norm": 1.1916162967681885,
      "learning_rate": 7.08125088137179e-05,
      "loss": 9.5082,
      "step": 1117
    },
    {
      "epoch": 0.38565022421524664,
      "grad_norm": 1.142784833908081,
      "learning_rate": 7.076146867816249e-05,
      "loss": 9.5339,
      "step": 1118
    },
    {
      "epoch": 0.38599517074853396,
      "grad_norm": 1.2644896507263184,
      "learning_rate": 7.071040238777264e-05,
      "loss": 9.5302,
      "step": 1119
    },
    {
      "epoch": 0.3863401172818213,
      "grad_norm": 1.3028452396392822,
      "learning_rate": 7.065931000688053e-05,
      "loss": 9.4775,
      "step": 1120
    },
    {
      "epoch": 0.38668506381510864,
      "grad_norm": 1.2342731952667236,
      "learning_rate": 7.060819159985116e-05,
      "loss": 9.5798,
      "step": 1121
    },
    {
      "epoch": 0.387030010348396,
      "grad_norm": 1.328894019126892,
      "learning_rate": 7.05570472310824e-05,
      "loss": 9.5172,
      "step": 1122
    },
    {
      "epoch": 0.38737495688168333,
      "grad_norm": 1.1947658061981201,
      "learning_rate": 7.05058769650048e-05,
      "loss": 9.4893,
      "step": 1123
    },
    {
      "epoch": 0.3877199034149707,
      "grad_norm": 1.2407337427139282,
      "learning_rate": 7.045468086608145e-05,
      "loss": 9.506,
      "step": 1124
    },
    {
      "epoch": 0.388064849948258,
      "grad_norm": 1.3401503562927246,
      "learning_rate": 7.040345899880809e-05,
      "loss": 9.4722,
      "step": 1125
    },
    {
      "epoch": 0.38840979648154533,
      "grad_norm": 1.6716581583023071,
      "learning_rate": 7.035221142771288e-05,
      "loss": 9.3779,
      "step": 1126
    },
    {
      "epoch": 0.3887547430148327,
      "grad_norm": 1.2554329633712769,
      "learning_rate": 7.030093821735639e-05,
      "loss": 9.5142,
      "step": 1127
    },
    {
      "epoch": 0.38909968954812,
      "grad_norm": 1.3145281076431274,
      "learning_rate": 7.024963943233141e-05,
      "loss": 9.4608,
      "step": 1128
    },
    {
      "epoch": 0.3894446360814074,
      "grad_norm": 1.3281456232070923,
      "learning_rate": 7.019831513726305e-05,
      "loss": 9.4716,
      "step": 1129
    },
    {
      "epoch": 0.3897895826146947,
      "grad_norm": 1.3885881900787354,
      "learning_rate": 7.014696539680849e-05,
      "loss": 9.5253,
      "step": 1130
    },
    {
      "epoch": 0.3901345291479821,
      "grad_norm": 1.44611394405365,
      "learning_rate": 7.009559027565699e-05,
      "loss": 9.508,
      "step": 1131
    },
    {
      "epoch": 0.3904794756812694,
      "grad_norm": 1.4117335081100464,
      "learning_rate": 7.004418983852979e-05,
      "loss": 9.4963,
      "step": 1132
    },
    {
      "epoch": 0.39082442221455677,
      "grad_norm": 1.3524430990219116,
      "learning_rate": 6.999276415018001e-05,
      "loss": 9.5305,
      "step": 1133
    },
    {
      "epoch": 0.3911693687478441,
      "grad_norm": 1.4730901718139648,
      "learning_rate": 6.994131327539258e-05,
      "loss": 9.4665,
      "step": 1134
    },
    {
      "epoch": 0.3915143152811314,
      "grad_norm": 1.216262698173523,
      "learning_rate": 6.988983727898414e-05,
      "loss": 9.5145,
      "step": 1135
    },
    {
      "epoch": 0.39185926181441877,
      "grad_norm": 1.3632173538208008,
      "learning_rate": 6.983833622580304e-05,
      "loss": 9.4898,
      "step": 1136
    },
    {
      "epoch": 0.3922042083477061,
      "grad_norm": 1.3551784753799438,
      "learning_rate": 6.978681018072912e-05,
      "loss": 9.5128,
      "step": 1137
    },
    {
      "epoch": 0.39254915488099346,
      "grad_norm": 1.3956425189971924,
      "learning_rate": 6.973525920867378e-05,
      "loss": 9.4895,
      "step": 1138
    },
    {
      "epoch": 0.39289410141428077,
      "grad_norm": 1.3603055477142334,
      "learning_rate": 6.968368337457973e-05,
      "loss": 9.4716,
      "step": 1139
    },
    {
      "epoch": 0.39323904794756814,
      "grad_norm": 1.5052947998046875,
      "learning_rate": 6.96320827434211e-05,
      "loss": 9.4741,
      "step": 1140
    },
    {
      "epoch": 0.39358399448085546,
      "grad_norm": 1.5152782201766968,
      "learning_rate": 6.95804573802032e-05,
      "loss": 9.441,
      "step": 1141
    },
    {
      "epoch": 0.39392894101414283,
      "grad_norm": 1.5431822538375854,
      "learning_rate": 6.95288073499625e-05,
      "loss": 9.4526,
      "step": 1142
    },
    {
      "epoch": 0.39427388754743015,
      "grad_norm": 1.4436509609222412,
      "learning_rate": 6.947713271776656e-05,
      "loss": 9.4532,
      "step": 1143
    },
    {
      "epoch": 0.39461883408071746,
      "grad_norm": 1.528663158416748,
      "learning_rate": 6.942543354871393e-05,
      "loss": 9.4697,
      "step": 1144
    },
    {
      "epoch": 0.39496378061400483,
      "grad_norm": 1.472640037536621,
      "learning_rate": 6.937370990793407e-05,
      "loss": 9.4878,
      "step": 1145
    },
    {
      "epoch": 0.39530872714729215,
      "grad_norm": 1.5636851787567139,
      "learning_rate": 6.932196186058725e-05,
      "loss": 9.4147,
      "step": 1146
    },
    {
      "epoch": 0.3956536736805795,
      "grad_norm": 1.5114054679870605,
      "learning_rate": 6.927018947186453e-05,
      "loss": 9.4385,
      "step": 1147
    },
    {
      "epoch": 0.39599862021386684,
      "grad_norm": 1.6888611316680908,
      "learning_rate": 6.921839280698758e-05,
      "loss": 9.3557,
      "step": 1148
    },
    {
      "epoch": 0.3963435667471542,
      "grad_norm": 1.7319715023040771,
      "learning_rate": 6.916657193120871e-05,
      "loss": 9.3959,
      "step": 1149
    },
    {
      "epoch": 0.3966885132804415,
      "grad_norm": 1.6487864255905151,
      "learning_rate": 6.91147269098107e-05,
      "loss": 9.3655,
      "step": 1150
    },
    {
      "epoch": 0.3970334598137289,
      "grad_norm": 0.8318474292755127,
      "learning_rate": 6.90628578081067e-05,
      "loss": 9.5905,
      "step": 1151
    },
    {
      "epoch": 0.3973784063470162,
      "grad_norm": 0.986283540725708,
      "learning_rate": 6.901096469144032e-05,
      "loss": 9.5246,
      "step": 1152
    },
    {
      "epoch": 0.3977233528803035,
      "grad_norm": 1.0239535570144653,
      "learning_rate": 6.895904762518528e-05,
      "loss": 9.5762,
      "step": 1153
    },
    {
      "epoch": 0.3980682994135909,
      "grad_norm": 1.1255637407302856,
      "learning_rate": 6.89071066747456e-05,
      "loss": 9.4697,
      "step": 1154
    },
    {
      "epoch": 0.3984132459468782,
      "grad_norm": 1.1051015853881836,
      "learning_rate": 6.88551419055553e-05,
      "loss": 9.5427,
      "step": 1155
    },
    {
      "epoch": 0.3987581924801656,
      "grad_norm": 1.1070098876953125,
      "learning_rate": 6.880315338307843e-05,
      "loss": 9.5184,
      "step": 1156
    },
    {
      "epoch": 0.3991031390134529,
      "grad_norm": 1.0810296535491943,
      "learning_rate": 6.875114117280898e-05,
      "loss": 9.5697,
      "step": 1157
    },
    {
      "epoch": 0.3994480855467403,
      "grad_norm": 1.0810821056365967,
      "learning_rate": 6.869910534027077e-05,
      "loss": 9.5222,
      "step": 1158
    },
    {
      "epoch": 0.3997930320800276,
      "grad_norm": 1.1185795068740845,
      "learning_rate": 6.864704595101737e-05,
      "loss": 9.5386,
      "step": 1159
    },
    {
      "epoch": 0.40013797861331496,
      "grad_norm": 1.0730277299880981,
      "learning_rate": 6.859496307063205e-05,
      "loss": 9.5331,
      "step": 1160
    },
    {
      "epoch": 0.4004829251466023,
      "grad_norm": 1.2169557809829712,
      "learning_rate": 6.854285676472764e-05,
      "loss": 9.5272,
      "step": 1161
    },
    {
      "epoch": 0.4008278716798896,
      "grad_norm": 1.1956943273544312,
      "learning_rate": 6.849072709894651e-05,
      "loss": 9.4893,
      "step": 1162
    },
    {
      "epoch": 0.40117281821317696,
      "grad_norm": 1.1670726537704468,
      "learning_rate": 6.843857413896047e-05,
      "loss": 9.5798,
      "step": 1163
    },
    {
      "epoch": 0.4015177647464643,
      "grad_norm": 1.2046582698822021,
      "learning_rate": 6.838639795047061e-05,
      "loss": 9.5015,
      "step": 1164
    },
    {
      "epoch": 0.40186271127975165,
      "grad_norm": 1.232411503791809,
      "learning_rate": 6.833419859920736e-05,
      "loss": 9.5492,
      "step": 1165
    },
    {
      "epoch": 0.40220765781303897,
      "grad_norm": 1.1485507488250732,
      "learning_rate": 6.82819761509303e-05,
      "loss": 9.5267,
      "step": 1166
    },
    {
      "epoch": 0.40255260434632634,
      "grad_norm": 1.2545028924942017,
      "learning_rate": 6.822973067142808e-05,
      "loss": 9.5099,
      "step": 1167
    },
    {
      "epoch": 0.40289755087961365,
      "grad_norm": 1.3489080667495728,
      "learning_rate": 6.81774622265184e-05,
      "loss": 9.4798,
      "step": 1168
    },
    {
      "epoch": 0.403242497412901,
      "grad_norm": 1.2620747089385986,
      "learning_rate": 6.812517088204788e-05,
      "loss": 9.4896,
      "step": 1169
    },
    {
      "epoch": 0.40358744394618834,
      "grad_norm": 1.1811059713363647,
      "learning_rate": 6.807285670389199e-05,
      "loss": 9.5281,
      "step": 1170
    },
    {
      "epoch": 0.40393239047947566,
      "grad_norm": 1.2100896835327148,
      "learning_rate": 6.802051975795495e-05,
      "loss": 9.5105,
      "step": 1171
    },
    {
      "epoch": 0.404277337012763,
      "grad_norm": 1.364280104637146,
      "learning_rate": 6.79681601101697e-05,
      "loss": 9.4185,
      "step": 1172
    },
    {
      "epoch": 0.40462228354605034,
      "grad_norm": 1.2773628234863281,
      "learning_rate": 6.791577782649771e-05,
      "loss": 9.5374,
      "step": 1173
    },
    {
      "epoch": 0.4049672300793377,
      "grad_norm": 1.2328367233276367,
      "learning_rate": 6.786337297292907e-05,
      "loss": 9.5104,
      "step": 1174
    },
    {
      "epoch": 0.40531217661262503,
      "grad_norm": 1.275637149810791,
      "learning_rate": 6.781094561548219e-05,
      "loss": 9.4765,
      "step": 1175
    },
    {
      "epoch": 0.4056571231459124,
      "grad_norm": 1.2586758136749268,
      "learning_rate": 6.775849582020393e-05,
      "loss": 9.5087,
      "step": 1176
    },
    {
      "epoch": 0.4060020696791997,
      "grad_norm": 1.2569774389266968,
      "learning_rate": 6.770602365316931e-05,
      "loss": 9.5354,
      "step": 1177
    },
    {
      "epoch": 0.4063470162124871,
      "grad_norm": 1.2480967044830322,
      "learning_rate": 6.765352918048167e-05,
      "loss": 9.4769,
      "step": 1178
    },
    {
      "epoch": 0.4066919627457744,
      "grad_norm": 1.3023992776870728,
      "learning_rate": 6.760101246827232e-05,
      "loss": 9.494,
      "step": 1179
    },
    {
      "epoch": 0.4070369092790617,
      "grad_norm": 1.3120083808898926,
      "learning_rate": 6.754847358270067e-05,
      "loss": 9.4736,
      "step": 1180
    },
    {
      "epoch": 0.4073818558123491,
      "grad_norm": 1.1869686841964722,
      "learning_rate": 6.749591258995401e-05,
      "loss": 9.4962,
      "step": 1181
    },
    {
      "epoch": 0.4077268023456364,
      "grad_norm": 1.373629093170166,
      "learning_rate": 6.744332955624751e-05,
      "loss": 9.4913,
      "step": 1182
    },
    {
      "epoch": 0.4080717488789238,
      "grad_norm": 1.3023265600204468,
      "learning_rate": 6.739072454782411e-05,
      "loss": 9.5136,
      "step": 1183
    },
    {
      "epoch": 0.4084166954122111,
      "grad_norm": 1.282807469367981,
      "learning_rate": 6.733809763095443e-05,
      "loss": 9.5072,
      "step": 1184
    },
    {
      "epoch": 0.40876164194549847,
      "grad_norm": 1.3017750978469849,
      "learning_rate": 6.728544887193667e-05,
      "loss": 9.5657,
      "step": 1185
    },
    {
      "epoch": 0.4091065884787858,
      "grad_norm": 1.6445789337158203,
      "learning_rate": 6.723277833709655e-05,
      "loss": 9.435,
      "step": 1186
    },
    {
      "epoch": 0.40945153501207315,
      "grad_norm": 1.4415967464447021,
      "learning_rate": 6.718008609278727e-05,
      "loss": 9.4222,
      "step": 1187
    },
    {
      "epoch": 0.40979648154536047,
      "grad_norm": 1.3269176483154297,
      "learning_rate": 6.71273722053893e-05,
      "loss": 9.471,
      "step": 1188
    },
    {
      "epoch": 0.4101414280786478,
      "grad_norm": 1.4464130401611328,
      "learning_rate": 6.707463674131045e-05,
      "loss": 9.4682,
      "step": 1189
    },
    {
      "epoch": 0.41048637461193516,
      "grad_norm": 1.451816201210022,
      "learning_rate": 6.702187976698567e-05,
      "loss": 9.4768,
      "step": 1190
    },
    {
      "epoch": 0.4108313211452225,
      "grad_norm": 1.4212086200714111,
      "learning_rate": 6.696910134887702e-05,
      "loss": 9.4485,
      "step": 1191
    },
    {
      "epoch": 0.41117626767850984,
      "grad_norm": 1.429534912109375,
      "learning_rate": 6.691630155347357e-05,
      "loss": 9.4584,
      "step": 1192
    },
    {
      "epoch": 0.41152121421179716,
      "grad_norm": 1.5786341428756714,
      "learning_rate": 6.686348044729131e-05,
      "loss": 9.455,
      "step": 1193
    },
    {
      "epoch": 0.41186616074508453,
      "grad_norm": 1.4174039363861084,
      "learning_rate": 6.681063809687312e-05,
      "loss": 9.4575,
      "step": 1194
    },
    {
      "epoch": 0.41221110727837185,
      "grad_norm": 1.4542423486709595,
      "learning_rate": 6.675777456878856e-05,
      "loss": 9.4186,
      "step": 1195
    },
    {
      "epoch": 0.4125560538116592,
      "grad_norm": 1.4207855463027954,
      "learning_rate": 6.670488992963396e-05,
      "loss": 9.4742,
      "step": 1196
    },
    {
      "epoch": 0.41290100034494653,
      "grad_norm": 1.489596962928772,
      "learning_rate": 6.665198424603223e-05,
      "loss": 9.4153,
      "step": 1197
    },
    {
      "epoch": 0.41324594687823385,
      "grad_norm": 1.5289177894592285,
      "learning_rate": 6.65990575846327e-05,
      "loss": 9.4263,
      "step": 1198
    },
    {
      "epoch": 0.4135908934115212,
      "grad_norm": 1.4849228858947754,
      "learning_rate": 6.654611001211127e-05,
      "loss": 9.4248,
      "step": 1199
    },
    {
      "epoch": 0.41393583994480854,
      "grad_norm": 1.7787175178527832,
      "learning_rate": 6.649314159517007e-05,
      "loss": 9.336,
      "step": 1200
    },
    {
      "epoch": 0.4142807864780959,
      "grad_norm": 0.8549869656562805,
      "learning_rate": 6.644015240053754e-05,
      "loss": 9.5802,
      "step": 1201
    },
    {
      "epoch": 0.4146257330113832,
      "grad_norm": 1.0080626010894775,
      "learning_rate": 6.638714249496824e-05,
      "loss": 9.5279,
      "step": 1202
    },
    {
      "epoch": 0.4149706795446706,
      "grad_norm": 0.9619107842445374,
      "learning_rate": 6.633411194524294e-05,
      "loss": 9.5521,
      "step": 1203
    },
    {
      "epoch": 0.4153156260779579,
      "grad_norm": 1.1162559986114502,
      "learning_rate": 6.628106081816829e-05,
      "loss": 9.5364,
      "step": 1204
    },
    {
      "epoch": 0.4156605726112453,
      "grad_norm": 1.0343070030212402,
      "learning_rate": 6.622798918057693e-05,
      "loss": 9.5726,
      "step": 1205
    },
    {
      "epoch": 0.4160055191445326,
      "grad_norm": 1.1603822708129883,
      "learning_rate": 6.617489709932733e-05,
      "loss": 9.4792,
      "step": 1206
    },
    {
      "epoch": 0.4163504656778199,
      "grad_norm": 1.2021888494491577,
      "learning_rate": 6.612178464130367e-05,
      "loss": 9.5337,
      "step": 1207
    },
    {
      "epoch": 0.4166954122111073,
      "grad_norm": 1.0984165668487549,
      "learning_rate": 6.60686518734159e-05,
      "loss": 9.527,
      "step": 1208
    },
    {
      "epoch": 0.4170403587443946,
      "grad_norm": 1.1222621202468872,
      "learning_rate": 6.601549886259945e-05,
      "loss": 9.4886,
      "step": 1209
    },
    {
      "epoch": 0.417385305277682,
      "grad_norm": 1.0295212268829346,
      "learning_rate": 6.596232567581531e-05,
      "loss": 9.6241,
      "step": 1210
    },
    {
      "epoch": 0.4177302518109693,
      "grad_norm": 1.2314414978027344,
      "learning_rate": 6.590913238004987e-05,
      "loss": 9.4848,
      "step": 1211
    },
    {
      "epoch": 0.41807519834425666,
      "grad_norm": 1.137915015220642,
      "learning_rate": 6.585591904231484e-05,
      "loss": 9.5238,
      "step": 1212
    },
    {
      "epoch": 0.418420144877544,
      "grad_norm": 1.2005376815795898,
      "learning_rate": 6.580268572964722e-05,
      "loss": 9.553,
      "step": 1213
    },
    {
      "epoch": 0.41876509141083135,
      "grad_norm": 1.156784176826477,
      "learning_rate": 6.574943250910916e-05,
      "loss": 9.5773,
      "step": 1214
    },
    {
      "epoch": 0.41911003794411866,
      "grad_norm": 1.1334538459777832,
      "learning_rate": 6.569615944778785e-05,
      "loss": 9.5779,
      "step": 1215
    },
    {
      "epoch": 0.419454984477406,
      "grad_norm": 1.3424021005630493,
      "learning_rate": 6.564286661279549e-05,
      "loss": 9.5571,
      "step": 1216
    },
    {
      "epoch": 0.41979993101069335,
      "grad_norm": 1.2306984663009644,
      "learning_rate": 6.558955407126922e-05,
      "loss": 9.4602,
      "step": 1217
    },
    {
      "epoch": 0.42014487754398067,
      "grad_norm": 1.1734577417373657,
      "learning_rate": 6.553622189037099e-05,
      "loss": 9.516,
      "step": 1218
    },
    {
      "epoch": 0.42048982407726804,
      "grad_norm": 1.301382064819336,
      "learning_rate": 6.548287013728751e-05,
      "loss": 9.5368,
      "step": 1219
    },
    {
      "epoch": 0.42083477061055535,
      "grad_norm": 1.2797003984451294,
      "learning_rate": 6.54294988792301e-05,
      "loss": 9.5108,
      "step": 1220
    },
    {
      "epoch": 0.4211797171438427,
      "grad_norm": 1.4130553007125854,
      "learning_rate": 6.537610818343467e-05,
      "loss": 9.4416,
      "step": 1221
    },
    {
      "epoch": 0.42152466367713004,
      "grad_norm": 1.3803390264511108,
      "learning_rate": 6.532269811716165e-05,
      "loss": 9.4939,
      "step": 1222
    },
    {
      "epoch": 0.4218696102104174,
      "grad_norm": 1.276152491569519,
      "learning_rate": 6.526926874769584e-05,
      "loss": 9.5169,
      "step": 1223
    },
    {
      "epoch": 0.4222145567437047,
      "grad_norm": 1.2787901163101196,
      "learning_rate": 6.521582014234638e-05,
      "loss": 9.5251,
      "step": 1224
    },
    {
      "epoch": 0.42255950327699204,
      "grad_norm": 1.3120802640914917,
      "learning_rate": 6.51623523684466e-05,
      "loss": 9.4306,
      "step": 1225
    },
    {
      "epoch": 0.4229044498102794,
      "grad_norm": 1.2330889701843262,
      "learning_rate": 6.510886549335402e-05,
      "loss": 9.4904,
      "step": 1226
    },
    {
      "epoch": 0.42324939634356673,
      "grad_norm": 1.3258132934570312,
      "learning_rate": 6.505535958445023e-05,
      "loss": 9.4623,
      "step": 1227
    },
    {
      "epoch": 0.4235943428768541,
      "grad_norm": 1.3963433504104614,
      "learning_rate": 6.500183470914075e-05,
      "loss": 9.4231,
      "step": 1228
    },
    {
      "epoch": 0.4239392894101414,
      "grad_norm": 1.3144148588180542,
      "learning_rate": 6.494829093485504e-05,
      "loss": 9.4466,
      "step": 1229
    },
    {
      "epoch": 0.4242842359434288,
      "grad_norm": 1.3022875785827637,
      "learning_rate": 6.489472832904634e-05,
      "loss": 9.5031,
      "step": 1230
    },
    {
      "epoch": 0.4246291824767161,
      "grad_norm": 1.3690828084945679,
      "learning_rate": 6.484114695919161e-05,
      "loss": 9.5058,
      "step": 1231
    },
    {
      "epoch": 0.4249741290100035,
      "grad_norm": 1.3271464109420776,
      "learning_rate": 6.478754689279151e-05,
      "loss": 9.4568,
      "step": 1232
    },
    {
      "epoch": 0.4253190755432908,
      "grad_norm": 1.3607510328292847,
      "learning_rate": 6.473392819737017e-05,
      "loss": 9.5171,
      "step": 1233
    },
    {
      "epoch": 0.4256640220765781,
      "grad_norm": 1.3704109191894531,
      "learning_rate": 6.468029094047521e-05,
      "loss": 9.5053,
      "step": 1234
    },
    {
      "epoch": 0.4260089686098655,
      "grad_norm": 1.5395640134811401,
      "learning_rate": 6.46266351896777e-05,
      "loss": 9.4434,
      "step": 1235
    },
    {
      "epoch": 0.4263539151431528,
      "grad_norm": 1.3000792264938354,
      "learning_rate": 6.457296101257192e-05,
      "loss": 9.4975,
      "step": 1236
    },
    {
      "epoch": 0.42669886167644017,
      "grad_norm": 1.3487573862075806,
      "learning_rate": 6.45192684767754e-05,
      "loss": 9.5217,
      "step": 1237
    },
    {
      "epoch": 0.4270438082097275,
      "grad_norm": 1.372479796409607,
      "learning_rate": 6.44655576499288e-05,
      "loss": 9.4771,
      "step": 1238
    },
    {
      "epoch": 0.42738875474301485,
      "grad_norm": 1.3763235807418823,
      "learning_rate": 6.441182859969583e-05,
      "loss": 9.434,
      "step": 1239
    },
    {
      "epoch": 0.42773370127630217,
      "grad_norm": 1.475024938583374,
      "learning_rate": 6.435808139376313e-05,
      "loss": 9.3964,
      "step": 1240
    },
    {
      "epoch": 0.42807864780958954,
      "grad_norm": 1.5142197608947754,
      "learning_rate": 6.430431609984025e-05,
      "loss": 9.4465,
      "step": 1241
    },
    {
      "epoch": 0.42842359434287686,
      "grad_norm": 1.2984328269958496,
      "learning_rate": 6.425053278565949e-05,
      "loss": 9.4826,
      "step": 1242
    },
    {
      "epoch": 0.4287685408761642,
      "grad_norm": 1.619287371635437,
      "learning_rate": 6.419673151897587e-05,
      "loss": 9.4733,
      "step": 1243
    },
    {
      "epoch": 0.42911348740945154,
      "grad_norm": 1.4279025793075562,
      "learning_rate": 6.414291236756702e-05,
      "loss": 9.4634,
      "step": 1244
    },
    {
      "epoch": 0.42945843394273886,
      "grad_norm": 1.4935935735702515,
      "learning_rate": 6.408907539923314e-05,
      "loss": 9.4086,
      "step": 1245
    },
    {
      "epoch": 0.42980338047602623,
      "grad_norm": 1.606303095817566,
      "learning_rate": 6.403522068179679e-05,
      "loss": 9.4784,
      "step": 1246
    },
    {
      "epoch": 0.43014832700931355,
      "grad_norm": 1.5438969135284424,
      "learning_rate": 6.398134828310296e-05,
      "loss": 9.4186,
      "step": 1247
    },
    {
      "epoch": 0.4304932735426009,
      "grad_norm": 1.450445294380188,
      "learning_rate": 6.39274582710189e-05,
      "loss": 9.4346,
      "step": 1248
    },
    {
      "epoch": 0.43083822007588823,
      "grad_norm": 1.7464522123336792,
      "learning_rate": 6.387355071343405e-05,
      "loss": 9.3591,
      "step": 1249
    },
    {
      "epoch": 0.4311831666091756,
      "grad_norm": 1.8599432706832886,
      "learning_rate": 6.381962567825991e-05,
      "loss": 9.3323,
      "step": 1250
    },
    {
      "epoch": 0.4315281131424629,
      "grad_norm": 0.9142783284187317,
      "learning_rate": 6.37656832334301e-05,
      "loss": 9.5449,
      "step": 1251
    },
    {
      "epoch": 0.43187305967575024,
      "grad_norm": 1.044974446296692,
      "learning_rate": 6.371172344690004e-05,
      "loss": 9.5463,
      "step": 1252
    },
    {
      "epoch": 0.4322180062090376,
      "grad_norm": 1.0773589611053467,
      "learning_rate": 6.36577463866471e-05,
      "loss": 9.5608,
      "step": 1253
    },
    {
      "epoch": 0.4325629527423249,
      "grad_norm": 1.068940281867981,
      "learning_rate": 6.360375212067036e-05,
      "loss": 9.5209,
      "step": 1254
    },
    {
      "epoch": 0.4329078992756123,
      "grad_norm": 1.03683602809906,
      "learning_rate": 6.35497407169906e-05,
      "loss": 9.5599,
      "step": 1255
    },
    {
      "epoch": 0.4332528458088996,
      "grad_norm": 1.0739017724990845,
      "learning_rate": 6.349571224365018e-05,
      "loss": 9.5328,
      "step": 1256
    },
    {
      "epoch": 0.433597792342187,
      "grad_norm": 0.9906735420227051,
      "learning_rate": 6.344166676871295e-05,
      "loss": 9.5635,
      "step": 1257
    },
    {
      "epoch": 0.4339427388754743,
      "grad_norm": 1.1994469165802002,
      "learning_rate": 6.338760436026422e-05,
      "loss": 9.5827,
      "step": 1258
    },
    {
      "epoch": 0.4342876854087616,
      "grad_norm": 1.1381406784057617,
      "learning_rate": 6.333352508641058e-05,
      "loss": 9.5336,
      "step": 1259
    },
    {
      "epoch": 0.434632631942049,
      "grad_norm": 1.1975433826446533,
      "learning_rate": 6.327942901527989e-05,
      "loss": 9.5102,
      "step": 1260
    },
    {
      "epoch": 0.4349775784753363,
      "grad_norm": 1.1926690340042114,
      "learning_rate": 6.322531621502117e-05,
      "loss": 9.5124,
      "step": 1261
    },
    {
      "epoch": 0.4353225250086237,
      "grad_norm": 1.0483787059783936,
      "learning_rate": 6.317118675380455e-05,
      "loss": 9.5291,
      "step": 1262
    },
    {
      "epoch": 0.435667471541911,
      "grad_norm": 1.3115662336349487,
      "learning_rate": 6.311704069982108e-05,
      "loss": 9.5245,
      "step": 1263
    },
    {
      "epoch": 0.43601241807519836,
      "grad_norm": 1.1819570064544678,
      "learning_rate": 6.306287812128276e-05,
      "loss": 9.4733,
      "step": 1264
    },
    {
      "epoch": 0.4363573646084857,
      "grad_norm": 1.2581042051315308,
      "learning_rate": 6.30086990864224e-05,
      "loss": 9.5113,
      "step": 1265
    },
    {
      "epoch": 0.43670231114177305,
      "grad_norm": 1.2800958156585693,
      "learning_rate": 6.295450366349354e-05,
      "loss": 9.5554,
      "step": 1266
    },
    {
      "epoch": 0.43704725767506036,
      "grad_norm": 1.0594321489334106,
      "learning_rate": 6.290029192077035e-05,
      "loss": 9.5154,
      "step": 1267
    },
    {
      "epoch": 0.4373922042083477,
      "grad_norm": 1.2820416688919067,
      "learning_rate": 6.284606392654758e-05,
      "loss": 9.4685,
      "step": 1268
    },
    {
      "epoch": 0.43773715074163505,
      "grad_norm": 1.1895943880081177,
      "learning_rate": 6.279181974914045e-05,
      "loss": 9.4791,
      "step": 1269
    },
    {
      "epoch": 0.43808209727492237,
      "grad_norm": 1.3060154914855957,
      "learning_rate": 6.273755945688458e-05,
      "loss": 9.5066,
      "step": 1270
    },
    {
      "epoch": 0.43842704380820974,
      "grad_norm": 1.3100395202636719,
      "learning_rate": 6.268328311813582e-05,
      "loss": 9.4843,
      "step": 1271
    },
    {
      "epoch": 0.43877199034149705,
      "grad_norm": 1.3740859031677246,
      "learning_rate": 6.262899080127034e-05,
      "loss": 9.4552,
      "step": 1272
    },
    {
      "epoch": 0.4391169368747844,
      "grad_norm": 1.3276293277740479,
      "learning_rate": 6.257468257468438e-05,
      "loss": 9.5196,
      "step": 1273
    },
    {
      "epoch": 0.43946188340807174,
      "grad_norm": 1.3813064098358154,
      "learning_rate": 6.25203585067942e-05,
      "loss": 9.4719,
      "step": 1274
    },
    {
      "epoch": 0.4398068299413591,
      "grad_norm": 1.360579252243042,
      "learning_rate": 6.246601866603609e-05,
      "loss": 9.4658,
      "step": 1275
    },
    {
      "epoch": 0.44015177647464643,
      "grad_norm": 1.1985294818878174,
      "learning_rate": 6.241166312086614e-05,
      "loss": 9.4809,
      "step": 1276
    },
    {
      "epoch": 0.44049672300793374,
      "grad_norm": 1.4840415716171265,
      "learning_rate": 6.235729193976026e-05,
      "loss": 9.4197,
      "step": 1277
    },
    {
      "epoch": 0.4408416695412211,
      "grad_norm": 1.2561899423599243,
      "learning_rate": 6.230290519121405e-05,
      "loss": 9.5185,
      "step": 1278
    },
    {
      "epoch": 0.44118661607450843,
      "grad_norm": 1.4068700075149536,
      "learning_rate": 6.22485029437427e-05,
      "loss": 9.4594,
      "step": 1279
    },
    {
      "epoch": 0.4415315626077958,
      "grad_norm": 1.3950265645980835,
      "learning_rate": 6.219408526588098e-05,
      "loss": 9.4549,
      "step": 1280
    },
    {
      "epoch": 0.4418765091410831,
      "grad_norm": 1.3500123023986816,
      "learning_rate": 6.213965222618305e-05,
      "loss": 9.4758,
      "step": 1281
    },
    {
      "epoch": 0.4422214556743705,
      "grad_norm": 1.347732424736023,
      "learning_rate": 6.208520389322243e-05,
      "loss": 9.492,
      "step": 1282
    },
    {
      "epoch": 0.4425664022076578,
      "grad_norm": 1.3567559719085693,
      "learning_rate": 6.203074033559193e-05,
      "loss": 9.5828,
      "step": 1283
    },
    {
      "epoch": 0.4429113487409452,
      "grad_norm": 1.2305982112884521,
      "learning_rate": 6.197626162190351e-05,
      "loss": 9.5299,
      "step": 1284
    },
    {
      "epoch": 0.4432562952742325,
      "grad_norm": 1.3164910078048706,
      "learning_rate": 6.192176782078822e-05,
      "loss": 9.4837,
      "step": 1285
    },
    {
      "epoch": 0.4436012418075198,
      "grad_norm": 1.2801580429077148,
      "learning_rate": 6.186725900089617e-05,
      "loss": 9.4811,
      "step": 1286
    },
    {
      "epoch": 0.4439461883408072,
      "grad_norm": 1.5440186262130737,
      "learning_rate": 6.181273523089632e-05,
      "loss": 9.4587,
      "step": 1287
    },
    {
      "epoch": 0.4442911348740945,
      "grad_norm": 1.3695796728134155,
      "learning_rate": 6.175819657947655e-05,
      "loss": 9.5089,
      "step": 1288
    },
    {
      "epoch": 0.44463608140738187,
      "grad_norm": 1.256692886352539,
      "learning_rate": 6.170364311534336e-05,
      "loss": 9.4668,
      "step": 1289
    },
    {
      "epoch": 0.4449810279406692,
      "grad_norm": 1.450398564338684,
      "learning_rate": 6.164907490722205e-05,
      "loss": 9.4356,
      "step": 1290
    },
    {
      "epoch": 0.44532597447395655,
      "grad_norm": 1.5067511796951294,
      "learning_rate": 6.15944920238564e-05,
      "loss": 9.4279,
      "step": 1291
    },
    {
      "epoch": 0.44567092100724387,
      "grad_norm": 1.4862086772918701,
      "learning_rate": 6.15398945340087e-05,
      "loss": 9.4324,
      "step": 1292
    },
    {
      "epoch": 0.44601586754053124,
      "grad_norm": 1.4513452053070068,
      "learning_rate": 6.148528250645966e-05,
      "loss": 9.4995,
      "step": 1293
    },
    {
      "epoch": 0.44636081407381856,
      "grad_norm": 1.5199989080429077,
      "learning_rate": 6.14306560100083e-05,
      "loss": 9.44,
      "step": 1294
    },
    {
      "epoch": 0.4467057606071059,
      "grad_norm": 1.5350712537765503,
      "learning_rate": 6.137601511347185e-05,
      "loss": 9.4288,
      "step": 1295
    },
    {
      "epoch": 0.44705070714039324,
      "grad_norm": 1.4825901985168457,
      "learning_rate": 6.132135988568568e-05,
      "loss": 9.4414,
      "step": 1296
    },
    {
      "epoch": 0.44739565367368056,
      "grad_norm": 1.6072896718978882,
      "learning_rate": 6.126669039550325e-05,
      "loss": 9.4124,
      "step": 1297
    },
    {
      "epoch": 0.44774060020696793,
      "grad_norm": 1.5674636363983154,
      "learning_rate": 6.121200671179595e-05,
      "loss": 9.3757,
      "step": 1298
    },
    {
      "epoch": 0.44808554674025525,
      "grad_norm": 1.6910454034805298,
      "learning_rate": 6.115730890345305e-05,
      "loss": 9.3629,
      "step": 1299
    },
    {
      "epoch": 0.4484304932735426,
      "grad_norm": 1.6481505632400513,
      "learning_rate": 6.110259703938165e-05,
      "loss": 9.387,
      "step": 1300
    },
    {
      "epoch": 0.44877543980682993,
      "grad_norm": 0.8783053755760193,
      "learning_rate": 6.104787118850652e-05,
      "loss": 9.5823,
      "step": 1301
    },
    {
      "epoch": 0.4491203863401173,
      "grad_norm": 1.075404167175293,
      "learning_rate": 6.0993131419770076e-05,
      "loss": 9.5247,
      "step": 1302
    },
    {
      "epoch": 0.4494653328734046,
      "grad_norm": 1.0360864400863647,
      "learning_rate": 6.093837780213224e-05,
      "loss": 9.5537,
      "step": 1303
    },
    {
      "epoch": 0.44981027940669194,
      "grad_norm": 1.0971293449401855,
      "learning_rate": 6.08836104045704e-05,
      "loss": 9.5563,
      "step": 1304
    },
    {
      "epoch": 0.4501552259399793,
      "grad_norm": 1.074342966079712,
      "learning_rate": 6.082882929607931e-05,
      "loss": 9.5333,
      "step": 1305
    },
    {
      "epoch": 0.4505001724732666,
      "grad_norm": 1.1947624683380127,
      "learning_rate": 6.077403454567099e-05,
      "loss": 9.5231,
      "step": 1306
    },
    {
      "epoch": 0.450845119006554,
      "grad_norm": 1.0972341299057007,
      "learning_rate": 6.071922622237463e-05,
      "loss": 9.4951,
      "step": 1307
    },
    {
      "epoch": 0.4511900655398413,
      "grad_norm": 1.1745643615722656,
      "learning_rate": 6.066440439523653e-05,
      "loss": 9.4876,
      "step": 1308
    },
    {
      "epoch": 0.4515350120731287,
      "grad_norm": 1.2012512683868408,
      "learning_rate": 6.0609569133319996e-05,
      "loss": 9.5364,
      "step": 1309
    },
    {
      "epoch": 0.451879958606416,
      "grad_norm": 1.1683628559112549,
      "learning_rate": 6.0554720505705286e-05,
      "loss": 9.5097,
      "step": 1310
    },
    {
      "epoch": 0.45222490513970337,
      "grad_norm": 1.1659014225006104,
      "learning_rate": 6.049985858148948e-05,
      "loss": 9.5134,
      "step": 1311
    },
    {
      "epoch": 0.4525698516729907,
      "grad_norm": 1.2198443412780762,
      "learning_rate": 6.044498342978638e-05,
      "loss": 9.5175,
      "step": 1312
    },
    {
      "epoch": 0.452914798206278,
      "grad_norm": 1.1924607753753662,
      "learning_rate": 6.0390095119726484e-05,
      "loss": 9.5343,
      "step": 1313
    },
    {
      "epoch": 0.4532597447395654,
      "grad_norm": 1.3247007131576538,
      "learning_rate": 6.0335193720456876e-05,
      "loss": 9.5194,
      "step": 1314
    },
    {
      "epoch": 0.4536046912728527,
      "grad_norm": 1.2382389307022095,
      "learning_rate": 6.02802793011411e-05,
      "loss": 9.5108,
      "step": 1315
    },
    {
      "epoch": 0.45394963780614006,
      "grad_norm": 1.2551473379135132,
      "learning_rate": 6.0225351930959107e-05,
      "loss": 9.5181,
      "step": 1316
    },
    {
      "epoch": 0.4542945843394274,
      "grad_norm": 1.1353877782821655,
      "learning_rate": 6.0170411679107174e-05,
      "loss": 9.5048,
      "step": 1317
    },
    {
      "epoch": 0.45463953087271475,
      "grad_norm": 1.1790921688079834,
      "learning_rate": 6.011545861479783e-05,
      "loss": 9.5132,
      "step": 1318
    },
    {
      "epoch": 0.45498447740600206,
      "grad_norm": 1.091915249824524,
      "learning_rate": 6.006049280725966e-05,
      "loss": 9.5422,
      "step": 1319
    },
    {
      "epoch": 0.45532942393928943,
      "grad_norm": 1.1751124858856201,
      "learning_rate": 6.000551432573742e-05,
      "loss": 9.5443,
      "step": 1320
    },
    {
      "epoch": 0.45567437047257675,
      "grad_norm": 1.18860924243927,
      "learning_rate": 5.995052323949174e-05,
      "loss": 9.4933,
      "step": 1321
    },
    {
      "epoch": 0.45601931700586407,
      "grad_norm": 1.313167691230774,
      "learning_rate": 5.9895519617799175e-05,
      "loss": 9.5229,
      "step": 1322
    },
    {
      "epoch": 0.45636426353915144,
      "grad_norm": 1.3344849348068237,
      "learning_rate": 5.984050352995205e-05,
      "loss": 9.502,
      "step": 1323
    },
    {
      "epoch": 0.45670921007243875,
      "grad_norm": 1.2585493326187134,
      "learning_rate": 5.978547504525841e-05,
      "loss": 9.452,
      "step": 1324
    },
    {
      "epoch": 0.4570541566057261,
      "grad_norm": 1.2542619705200195,
      "learning_rate": 5.973043423304191e-05,
      "loss": 9.5105,
      "step": 1325
    },
    {
      "epoch": 0.45739910313901344,
      "grad_norm": 1.2945995330810547,
      "learning_rate": 5.967538116264174e-05,
      "loss": 9.4899,
      "step": 1326
    },
    {
      "epoch": 0.4577440496723008,
      "grad_norm": 1.3108856678009033,
      "learning_rate": 5.962031590341251e-05,
      "loss": 9.4645,
      "step": 1327
    },
    {
      "epoch": 0.45808899620558813,
      "grad_norm": 1.3480335474014282,
      "learning_rate": 5.956523852472422e-05,
      "loss": 9.5071,
      "step": 1328
    },
    {
      "epoch": 0.4584339427388755,
      "grad_norm": 1.3959459066390991,
      "learning_rate": 5.951014909596212e-05,
      "loss": 9.4515,
      "step": 1329
    },
    {
      "epoch": 0.4587788892721628,
      "grad_norm": 1.3636053800582886,
      "learning_rate": 5.9455047686526635e-05,
      "loss": 9.4722,
      "step": 1330
    },
    {
      "epoch": 0.45912383580545013,
      "grad_norm": 1.3342995643615723,
      "learning_rate": 5.93999343658333e-05,
      "loss": 9.4706,
      "step": 1331
    },
    {
      "epoch": 0.4594687823387375,
      "grad_norm": 1.3717411756515503,
      "learning_rate": 5.9344809203312626e-05,
      "loss": 9.4213,
      "step": 1332
    },
    {
      "epoch": 0.4598137288720248,
      "grad_norm": 1.3053313493728638,
      "learning_rate": 5.9289672268410076e-05,
      "loss": 9.4657,
      "step": 1333
    },
    {
      "epoch": 0.4601586754053122,
      "grad_norm": 1.3955119848251343,
      "learning_rate": 5.9234523630585936e-05,
      "loss": 9.5157,
      "step": 1334
    },
    {
      "epoch": 0.4605036219385995,
      "grad_norm": 1.3672175407409668,
      "learning_rate": 5.917936335931521e-05,
      "loss": 9.5097,
      "step": 1335
    },
    {
      "epoch": 0.4608485684718869,
      "grad_norm": 1.417806625366211,
      "learning_rate": 5.91241915240876e-05,
      "loss": 9.4413,
      "step": 1336
    },
    {
      "epoch": 0.4611935150051742,
      "grad_norm": 1.343322992324829,
      "learning_rate": 5.9069008194407324e-05,
      "loss": 9.48,
      "step": 1337
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 1.343604326248169,
      "learning_rate": 5.901381343979312e-05,
      "loss": 9.4752,
      "step": 1338
    },
    {
      "epoch": 0.4618834080717489,
      "grad_norm": 1.279022216796875,
      "learning_rate": 5.89586073297781e-05,
      "loss": 9.5199,
      "step": 1339
    },
    {
      "epoch": 0.4622283546050362,
      "grad_norm": 1.3533319234848022,
      "learning_rate": 5.8903389933909714e-05,
      "loss": 9.4999,
      "step": 1340
    },
    {
      "epoch": 0.46257330113832357,
      "grad_norm": 1.4877336025238037,
      "learning_rate": 5.884816132174956e-05,
      "loss": 9.4258,
      "step": 1341
    },
    {
      "epoch": 0.4629182476716109,
      "grad_norm": 1.4366827011108398,
      "learning_rate": 5.879292156287346e-05,
      "loss": 9.4265,
      "step": 1342
    },
    {
      "epoch": 0.46326319420489825,
      "grad_norm": 1.5778945684432983,
      "learning_rate": 5.8737670726871176e-05,
      "loss": 9.4417,
      "step": 1343
    },
    {
      "epoch": 0.46360814073818557,
      "grad_norm": 1.4845643043518066,
      "learning_rate": 5.868240888334653e-05,
      "loss": 9.4152,
      "step": 1344
    },
    {
      "epoch": 0.46395308727147294,
      "grad_norm": 1.4498002529144287,
      "learning_rate": 5.86271361019171e-05,
      "loss": 9.3937,
      "step": 1345
    },
    {
      "epoch": 0.46429803380476026,
      "grad_norm": 1.688463807106018,
      "learning_rate": 5.857185245221436e-05,
      "loss": 9.3969,
      "step": 1346
    },
    {
      "epoch": 0.46464298033804763,
      "grad_norm": 1.6296204328536987,
      "learning_rate": 5.851655800388338e-05,
      "loss": 9.3736,
      "step": 1347
    },
    {
      "epoch": 0.46498792687133494,
      "grad_norm": 1.6956126689910889,
      "learning_rate": 5.846125282658288e-05,
      "loss": 9.3915,
      "step": 1348
    },
    {
      "epoch": 0.46533287340462226,
      "grad_norm": 1.7350283861160278,
      "learning_rate": 5.8405936989985086e-05,
      "loss": 9.3848,
      "step": 1349
    },
    {
      "epoch": 0.46567781993790963,
      "grad_norm": 1.6955797672271729,
      "learning_rate": 5.835061056377567e-05,
      "loss": 9.401,
      "step": 1350
    },
    {
      "epoch": 0.46602276647119695,
      "grad_norm": 0.9491857290267944,
      "learning_rate": 5.82952736176536e-05,
      "loss": 9.5889,
      "step": 1351
    },
    {
      "epoch": 0.4663677130044843,
      "grad_norm": 1.0396230220794678,
      "learning_rate": 5.8239926221331134e-05,
      "loss": 9.5876,
      "step": 1352
    },
    {
      "epoch": 0.46671265953777163,
      "grad_norm": 0.9977867603302002,
      "learning_rate": 5.8184568444533697e-05,
      "loss": 9.5805,
      "step": 1353
    },
    {
      "epoch": 0.467057606071059,
      "grad_norm": 1.1935267448425293,
      "learning_rate": 5.8129200356999755e-05,
      "loss": 9.5101,
      "step": 1354
    },
    {
      "epoch": 0.4674025526043463,
      "grad_norm": 1.0658153295516968,
      "learning_rate": 5.8073822028480805e-05,
      "loss": 9.497,
      "step": 1355
    },
    {
      "epoch": 0.4677474991376337,
      "grad_norm": 1.0787256956100464,
      "learning_rate": 5.801843352874123e-05,
      "loss": 9.6134,
      "step": 1356
    },
    {
      "epoch": 0.468092445670921,
      "grad_norm": 1.1341609954833984,
      "learning_rate": 5.7963034927558235e-05,
      "loss": 9.5077,
      "step": 1357
    },
    {
      "epoch": 0.4684373922042083,
      "grad_norm": 1.1322368383407593,
      "learning_rate": 5.790762629472173e-05,
      "loss": 9.5009,
      "step": 1358
    },
    {
      "epoch": 0.4687823387374957,
      "grad_norm": 1.1739046573638916,
      "learning_rate": 5.785220770003424e-05,
      "loss": 9.5338,
      "step": 1359
    },
    {
      "epoch": 0.469127285270783,
      "grad_norm": 1.1165406703948975,
      "learning_rate": 5.7796779213310934e-05,
      "loss": 9.5175,
      "step": 1360
    },
    {
      "epoch": 0.4694722318040704,
      "grad_norm": 1.1558291912078857,
      "learning_rate": 5.774134090437934e-05,
      "loss": 9.5053,
      "step": 1361
    },
    {
      "epoch": 0.4698171783373577,
      "grad_norm": 1.215736746788025,
      "learning_rate": 5.768589284307939e-05,
      "loss": 9.4972,
      "step": 1362
    },
    {
      "epoch": 0.47016212487064507,
      "grad_norm": 1.3127442598342896,
      "learning_rate": 5.7630435099263356e-05,
      "loss": 9.514,
      "step": 1363
    },
    {
      "epoch": 0.4705070714039324,
      "grad_norm": 1.2573809623718262,
      "learning_rate": 5.757496774279565e-05,
      "loss": 9.5645,
      "step": 1364
    },
    {
      "epoch": 0.47085201793721976,
      "grad_norm": 1.3243647813796997,
      "learning_rate": 5.751949084355277e-05,
      "loss": 9.4822,
      "step": 1365
    },
    {
      "epoch": 0.4711969644705071,
      "grad_norm": 1.22898268699646,
      "learning_rate": 5.746400447142332e-05,
      "loss": 9.5129,
      "step": 1366
    },
    {
      "epoch": 0.4715419110037944,
      "grad_norm": 1.2787483930587769,
      "learning_rate": 5.7408508696307784e-05,
      "loss": 9.5467,
      "step": 1367
    },
    {
      "epoch": 0.47188685753708176,
      "grad_norm": 1.3485304117202759,
      "learning_rate": 5.7353003588118495e-05,
      "loss": 9.4727,
      "step": 1368
    },
    {
      "epoch": 0.4722318040703691,
      "grad_norm": 1.2806965112686157,
      "learning_rate": 5.729748921677953e-05,
      "loss": 9.5296,
      "step": 1369
    },
    {
      "epoch": 0.47257675060365645,
      "grad_norm": 1.2300639152526855,
      "learning_rate": 5.724196565222668e-05,
      "loss": 9.5279,
      "step": 1370
    },
    {
      "epoch": 0.47292169713694376,
      "grad_norm": 1.451492428779602,
      "learning_rate": 5.718643296440729e-05,
      "loss": 9.4898,
      "step": 1371
    },
    {
      "epoch": 0.47326664367023114,
      "grad_norm": 1.3249592781066895,
      "learning_rate": 5.713089122328019e-05,
      "loss": 9.5028,
      "step": 1372
    },
    {
      "epoch": 0.47361159020351845,
      "grad_norm": 1.3282641172409058,
      "learning_rate": 5.707534049881561e-05,
      "loss": 9.5387,
      "step": 1373
    },
    {
      "epoch": 0.4739565367368058,
      "grad_norm": 1.2661629915237427,
      "learning_rate": 5.7019780860995145e-05,
      "loss": 9.5194,
      "step": 1374
    },
    {
      "epoch": 0.47430148327009314,
      "grad_norm": 1.3013981580734253,
      "learning_rate": 5.696421237981156e-05,
      "loss": 9.4936,
      "step": 1375
    },
    {
      "epoch": 0.47464642980338045,
      "grad_norm": 1.4271470308303833,
      "learning_rate": 5.690863512526878e-05,
      "loss": 9.525,
      "step": 1376
    },
    {
      "epoch": 0.4749913763366678,
      "grad_norm": 1.246364712715149,
      "learning_rate": 5.685304916738182e-05,
      "loss": 9.4834,
      "step": 1377
    },
    {
      "epoch": 0.47533632286995514,
      "grad_norm": 1.4177308082580566,
      "learning_rate": 5.679745457617661e-05,
      "loss": 9.46,
      "step": 1378
    },
    {
      "epoch": 0.4756812694032425,
      "grad_norm": 1.1870930194854736,
      "learning_rate": 5.674185142168997e-05,
      "loss": 9.5407,
      "step": 1379
    },
    {
      "epoch": 0.47602621593652983,
      "grad_norm": 1.2858004570007324,
      "learning_rate": 5.668623977396952e-05,
      "loss": 9.5336,
      "step": 1380
    },
    {
      "epoch": 0.4763711624698172,
      "grad_norm": 1.3497929573059082,
      "learning_rate": 5.663061970307357e-05,
      "loss": 9.4715,
      "step": 1381
    },
    {
      "epoch": 0.4767161090031045,
      "grad_norm": 1.3225263357162476,
      "learning_rate": 5.6574991279071046e-05,
      "loss": 9.4959,
      "step": 1382
    },
    {
      "epoch": 0.4770610555363919,
      "grad_norm": 1.3493469953536987,
      "learning_rate": 5.6519354572041384e-05,
      "loss": 9.4836,
      "step": 1383
    },
    {
      "epoch": 0.4774060020696792,
      "grad_norm": 1.3390860557556152,
      "learning_rate": 5.646370965207447e-05,
      "loss": 9.5075,
      "step": 1384
    },
    {
      "epoch": 0.4777509486029665,
      "grad_norm": 1.4489787817001343,
      "learning_rate": 5.640805658927053e-05,
      "loss": 9.4673,
      "step": 1385
    },
    {
      "epoch": 0.4780958951362539,
      "grad_norm": 1.428164005279541,
      "learning_rate": 5.6352395453740045e-05,
      "loss": 9.4675,
      "step": 1386
    },
    {
      "epoch": 0.4784408416695412,
      "grad_norm": 1.3007261753082275,
      "learning_rate": 5.629672631560369e-05,
      "loss": 9.4735,
      "step": 1387
    },
    {
      "epoch": 0.4787857882028286,
      "grad_norm": 1.4828238487243652,
      "learning_rate": 5.6241049244992164e-05,
      "loss": 9.4761,
      "step": 1388
    },
    {
      "epoch": 0.4791307347361159,
      "grad_norm": 1.3790254592895508,
      "learning_rate": 5.618536431204624e-05,
      "loss": 9.5257,
      "step": 1389
    },
    {
      "epoch": 0.47947568126940326,
      "grad_norm": 1.4181640148162842,
      "learning_rate": 5.612967158691652e-05,
      "loss": 9.5113,
      "step": 1390
    },
    {
      "epoch": 0.4798206278026906,
      "grad_norm": 1.5841997861862183,
      "learning_rate": 5.607397113976347e-05,
      "loss": 9.4038,
      "step": 1391
    },
    {
      "epoch": 0.4801655743359779,
      "grad_norm": 1.487927794456482,
      "learning_rate": 5.601826304075728e-05,
      "loss": 9.4257,
      "step": 1392
    },
    {
      "epoch": 0.48051052086926527,
      "grad_norm": 1.5204216241836548,
      "learning_rate": 5.5962547360077745e-05,
      "loss": 9.3959,
      "step": 1393
    },
    {
      "epoch": 0.4808554674025526,
      "grad_norm": 1.4272792339324951,
      "learning_rate": 5.590682416791424e-05,
      "loss": 9.4667,
      "step": 1394
    },
    {
      "epoch": 0.48120041393583995,
      "grad_norm": 1.5055564641952515,
      "learning_rate": 5.5851093534465605e-05,
      "loss": 9.3947,
      "step": 1395
    },
    {
      "epoch": 0.48154536046912727,
      "grad_norm": 1.5691437721252441,
      "learning_rate": 5.579535552994005e-05,
      "loss": 9.4853,
      "step": 1396
    },
    {
      "epoch": 0.48189030700241464,
      "grad_norm": 1.537042260169983,
      "learning_rate": 5.573961022455505e-05,
      "loss": 9.421,
      "step": 1397
    },
    {
      "epoch": 0.48223525353570196,
      "grad_norm": 1.6870508193969727,
      "learning_rate": 5.568385768853732e-05,
      "loss": 9.4025,
      "step": 1398
    },
    {
      "epoch": 0.48258020006898933,
      "grad_norm": 1.7952243089675903,
      "learning_rate": 5.562809799212264e-05,
      "loss": 9.3092,
      "step": 1399
    },
    {
      "epoch": 0.48292514660227664,
      "grad_norm": 1.8603969812393188,
      "learning_rate": 5.5572331205555836e-05,
      "loss": 9.3861,
      "step": 1400
    },
    {
      "epoch": 0.48327009313556396,
      "grad_norm": 0.943221926689148,
      "learning_rate": 5.5516557399090665e-05,
      "loss": 9.5419,
      "step": 1401
    },
    {
      "epoch": 0.48361503966885133,
      "grad_norm": 0.949111819267273,
      "learning_rate": 5.546077664298972e-05,
      "loss": 9.545,
      "step": 1402
    },
    {
      "epoch": 0.48395998620213865,
      "grad_norm": 1.0729368925094604,
      "learning_rate": 5.5404989007524355e-05,
      "loss": 9.5456,
      "step": 1403
    },
    {
      "epoch": 0.484304932735426,
      "grad_norm": 1.070397138595581,
      "learning_rate": 5.534919456297457e-05,
      "loss": 9.5503,
      "step": 1404
    },
    {
      "epoch": 0.48464987926871334,
      "grad_norm": 1.0518831014633179,
      "learning_rate": 5.5293393379628975e-05,
      "loss": 9.5389,
      "step": 1405
    },
    {
      "epoch": 0.4849948258020007,
      "grad_norm": 1.099482536315918,
      "learning_rate": 5.5237585527784664e-05,
      "loss": 9.5556,
      "step": 1406
    },
    {
      "epoch": 0.485339772335288,
      "grad_norm": 1.1868844032287598,
      "learning_rate": 5.51817710777471e-05,
      "loss": 9.4733,
      "step": 1407
    },
    {
      "epoch": 0.4856847188685754,
      "grad_norm": 1.234670877456665,
      "learning_rate": 5.512595009983008e-05,
      "loss": 9.5143,
      "step": 1408
    },
    {
      "epoch": 0.4860296654018627,
      "grad_norm": 1.1355990171432495,
      "learning_rate": 5.507012266435564e-05,
      "loss": 9.5251,
      "step": 1409
    },
    {
      "epoch": 0.48637461193515,
      "grad_norm": 1.1389458179473877,
      "learning_rate": 5.501428884165393e-05,
      "loss": 9.5322,
      "step": 1410
    },
    {
      "epoch": 0.4867195584684374,
      "grad_norm": 1.123797059059143,
      "learning_rate": 5.495844870206315e-05,
      "loss": 9.5602,
      "step": 1411
    },
    {
      "epoch": 0.4870645050017247,
      "grad_norm": 1.1804533004760742,
      "learning_rate": 5.490260231592946e-05,
      "loss": 9.5132,
      "step": 1412
    },
    {
      "epoch": 0.4874094515350121,
      "grad_norm": 1.3132667541503906,
      "learning_rate": 5.4846749753606906e-05,
      "loss": 9.5615,
      "step": 1413
    },
    {
      "epoch": 0.4877543980682994,
      "grad_norm": 1.2434678077697754,
      "learning_rate": 5.4790891085457276e-05,
      "loss": 9.538,
      "step": 1414
    },
    {
      "epoch": 0.48809934460158677,
      "grad_norm": 1.2329378128051758,
      "learning_rate": 5.47350263818501e-05,
      "loss": 9.5578,
      "step": 1415
    },
    {
      "epoch": 0.4884442911348741,
      "grad_norm": 1.2569046020507812,
      "learning_rate": 5.4679155713162446e-05,
      "loss": 9.4817,
      "step": 1416
    },
    {
      "epoch": 0.48878923766816146,
      "grad_norm": 1.2540076971054077,
      "learning_rate": 5.462327914977896e-05,
      "loss": 9.605,
      "step": 1417
    },
    {
      "epoch": 0.4891341842014488,
      "grad_norm": 1.140594482421875,
      "learning_rate": 5.456739676209168e-05,
      "loss": 9.5552,
      "step": 1418
    },
    {
      "epoch": 0.4894791307347361,
      "grad_norm": 1.2399282455444336,
      "learning_rate": 5.451150862049999e-05,
      "loss": 9.5012,
      "step": 1419
    },
    {
      "epoch": 0.48982407726802346,
      "grad_norm": 1.388474941253662,
      "learning_rate": 5.445561479541054e-05,
      "loss": 9.4309,
      "step": 1420
    },
    {
      "epoch": 0.4901690238013108,
      "grad_norm": 1.2384589910507202,
      "learning_rate": 5.439971535723708e-05,
      "loss": 9.5436,
      "step": 1421
    },
    {
      "epoch": 0.49051397033459815,
      "grad_norm": 1.2547194957733154,
      "learning_rate": 5.4343810376400503e-05,
      "loss": 9.5297,
      "step": 1422
    },
    {
      "epoch": 0.49085891686788546,
      "grad_norm": 1.33177649974823,
      "learning_rate": 5.4287899923328664e-05,
      "loss": 9.4787,
      "step": 1423
    },
    {
      "epoch": 0.49120386340117284,
      "grad_norm": 1.2320342063903809,
      "learning_rate": 5.423198406845629e-05,
      "loss": 9.5077,
      "step": 1424
    },
    {
      "epoch": 0.49154880993446015,
      "grad_norm": 1.1713165044784546,
      "learning_rate": 5.417606288222491e-05,
      "loss": 9.5643,
      "step": 1425
    },
    {
      "epoch": 0.4918937564677475,
      "grad_norm": 1.2229870557785034,
      "learning_rate": 5.4120136435082826e-05,
      "loss": 9.516,
      "step": 1426
    },
    {
      "epoch": 0.49223870300103484,
      "grad_norm": 1.3608933687210083,
      "learning_rate": 5.406420479748489e-05,
      "loss": 9.4398,
      "step": 1427
    },
    {
      "epoch": 0.49258364953432215,
      "grad_norm": 1.3051090240478516,
      "learning_rate": 5.400826803989253e-05,
      "loss": 9.5392,
      "step": 1428
    },
    {
      "epoch": 0.4929285960676095,
      "grad_norm": 1.2316997051239014,
      "learning_rate": 5.3952326232773646e-05,
      "loss": 9.4952,
      "step": 1429
    },
    {
      "epoch": 0.49327354260089684,
      "grad_norm": 1.303128719329834,
      "learning_rate": 5.3896379446602476e-05,
      "loss": 9.4482,
      "step": 1430
    },
    {
      "epoch": 0.4936184891341842,
      "grad_norm": 1.3697431087493896,
      "learning_rate": 5.384042775185951e-05,
      "loss": 9.475,
      "step": 1431
    },
    {
      "epoch": 0.49396343566747153,
      "grad_norm": 1.3772562742233276,
      "learning_rate": 5.378447121903145e-05,
      "loss": 9.5066,
      "step": 1432
    },
    {
      "epoch": 0.4943083822007589,
      "grad_norm": 1.229060173034668,
      "learning_rate": 5.372850991861109e-05,
      "loss": 9.5655,
      "step": 1433
    },
    {
      "epoch": 0.4946533287340462,
      "grad_norm": 1.2789677381515503,
      "learning_rate": 5.3672543921097226e-05,
      "loss": 9.4798,
      "step": 1434
    },
    {
      "epoch": 0.4949982752673336,
      "grad_norm": 1.3277839422225952,
      "learning_rate": 5.3616573296994577e-05,
      "loss": 9.527,
      "step": 1435
    },
    {
      "epoch": 0.4953432218006209,
      "grad_norm": 1.2170019149780273,
      "learning_rate": 5.356059811681369e-05,
      "loss": 9.5078,
      "step": 1436
    },
    {
      "epoch": 0.4956881683339082,
      "grad_norm": 1.353895902633667,
      "learning_rate": 5.3504618451070833e-05,
      "loss": 9.4486,
      "step": 1437
    },
    {
      "epoch": 0.4960331148671956,
      "grad_norm": 1.350578784942627,
      "learning_rate": 5.344863437028794e-05,
      "loss": 9.4945,
      "step": 1438
    },
    {
      "epoch": 0.4963780614004829,
      "grad_norm": 1.6105198860168457,
      "learning_rate": 5.3392645944992495e-05,
      "loss": 9.4206,
      "step": 1439
    },
    {
      "epoch": 0.4967230079337703,
      "grad_norm": 1.5112483501434326,
      "learning_rate": 5.3336653245717494e-05,
      "loss": 9.3976,
      "step": 1440
    },
    {
      "epoch": 0.4970679544670576,
      "grad_norm": 1.5440597534179688,
      "learning_rate": 5.328065634300127e-05,
      "loss": 9.4787,
      "step": 1441
    },
    {
      "epoch": 0.49741290100034496,
      "grad_norm": 1.568060040473938,
      "learning_rate": 5.322465530738746e-05,
      "loss": 9.3947,
      "step": 1442
    },
    {
      "epoch": 0.4977578475336323,
      "grad_norm": 1.365394949913025,
      "learning_rate": 5.316865020942494e-05,
      "loss": 9.4426,
      "step": 1443
    },
    {
      "epoch": 0.49810279406691965,
      "grad_norm": 1.5042157173156738,
      "learning_rate": 5.311264111966767e-05,
      "loss": 9.4702,
      "step": 1444
    },
    {
      "epoch": 0.49844774060020697,
      "grad_norm": 1.4479701519012451,
      "learning_rate": 5.305662810867466e-05,
      "loss": 9.374,
      "step": 1445
    },
    {
      "epoch": 0.4987926871334943,
      "grad_norm": 1.5126973390579224,
      "learning_rate": 5.300061124700982e-05,
      "loss": 9.3831,
      "step": 1446
    },
    {
      "epoch": 0.49913763366678165,
      "grad_norm": 1.536891222000122,
      "learning_rate": 5.2944590605241986e-05,
      "loss": 9.4704,
      "step": 1447
    },
    {
      "epoch": 0.49948258020006897,
      "grad_norm": 1.661973476409912,
      "learning_rate": 5.288856625394467e-05,
      "loss": 9.3696,
      "step": 1448
    },
    {
      "epoch": 0.49982752673335634,
      "grad_norm": 1.7247661352157593,
      "learning_rate": 5.283253826369612e-05,
      "loss": 9.3604,
      "step": 1449
    },
    {
      "epoch": 0.5001724732666437,
      "grad_norm": 1.7713525295257568,
      "learning_rate": 5.277650670507915e-05,
      "loss": 9.3608,
      "step": 1450
    },
    {
      "epoch": 0.500517419799931,
      "grad_norm": 0.9689469337463379,
      "learning_rate": 5.272047164868106e-05,
      "loss": 9.5352,
      "step": 1451
    },
    {
      "epoch": 0.5008623663332183,
      "grad_norm": 1.0381879806518555,
      "learning_rate": 5.266443316509355e-05,
      "loss": 9.5648,
      "step": 1452
    },
    {
      "epoch": 0.5012073128665057,
      "grad_norm": 1.0764495134353638,
      "learning_rate": 5.260839132491266e-05,
      "loss": 9.5591,
      "step": 1453
    },
    {
      "epoch": 0.501552259399793,
      "grad_norm": 1.1121872663497925,
      "learning_rate": 5.255234619873866e-05,
      "loss": 9.5613,
      "step": 1454
    },
    {
      "epoch": 0.5018972059330804,
      "grad_norm": 1.153185486793518,
      "learning_rate": 5.249629785717592e-05,
      "loss": 9.5407,
      "step": 1455
    },
    {
      "epoch": 0.5022421524663677,
      "grad_norm": 1.0321553945541382,
      "learning_rate": 5.2440246370832914e-05,
      "loss": 9.5957,
      "step": 1456
    },
    {
      "epoch": 0.502587098999655,
      "grad_norm": 1.1872321367263794,
      "learning_rate": 5.238419181032204e-05,
      "loss": 9.4708,
      "step": 1457
    },
    {
      "epoch": 0.5029320455329424,
      "grad_norm": 1.1325219869613647,
      "learning_rate": 5.2328134246259594e-05,
      "loss": 9.5841,
      "step": 1458
    },
    {
      "epoch": 0.5032769920662298,
      "grad_norm": 1.1462233066558838,
      "learning_rate": 5.227207374926563e-05,
      "loss": 9.5935,
      "step": 1459
    },
    {
      "epoch": 0.5036219385995171,
      "grad_norm": 1.1690250635147095,
      "learning_rate": 5.2216010389963896e-05,
      "loss": 9.4704,
      "step": 1460
    },
    {
      "epoch": 0.5039668851328044,
      "grad_norm": 1.1512292623519897,
      "learning_rate": 5.2159944238981803e-05,
      "loss": 9.5424,
      "step": 1461
    },
    {
      "epoch": 0.5043118316660917,
      "grad_norm": 1.2307171821594238,
      "learning_rate": 5.2103875366950196e-05,
      "loss": 9.5497,
      "step": 1462
    },
    {
      "epoch": 0.504656778199379,
      "grad_norm": 1.2325085401535034,
      "learning_rate": 5.20478038445034e-05,
      "loss": 9.5028,
      "step": 1463
    },
    {
      "epoch": 0.5050017247326665,
      "grad_norm": 1.056069016456604,
      "learning_rate": 5.199172974227907e-05,
      "loss": 9.57,
      "step": 1464
    },
    {
      "epoch": 0.5053466712659538,
      "grad_norm": 1.256806492805481,
      "learning_rate": 5.1935653130918106e-05,
      "loss": 9.4998,
      "step": 1465
    },
    {
      "epoch": 0.5056916177992411,
      "grad_norm": 1.1562812328338623,
      "learning_rate": 5.187957408106456e-05,
      "loss": 9.4816,
      "step": 1466
    },
    {
      "epoch": 0.5060365643325284,
      "grad_norm": 1.1788362264633179,
      "learning_rate": 5.182349266336558e-05,
      "loss": 9.5193,
      "step": 1467
    },
    {
      "epoch": 0.5063815108658158,
      "grad_norm": 1.1777877807617188,
      "learning_rate": 5.1767408948471286e-05,
      "loss": 9.5475,
      "step": 1468
    },
    {
      "epoch": 0.5067264573991032,
      "grad_norm": 1.248210072517395,
      "learning_rate": 5.171132300703467e-05,
      "loss": 9.4653,
      "step": 1469
    },
    {
      "epoch": 0.5070714039323905,
      "grad_norm": 1.3637944459915161,
      "learning_rate": 5.1655234909711555e-05,
      "loss": 9.5111,
      "step": 1470
    },
    {
      "epoch": 0.5074163504656778,
      "grad_norm": 1.1688549518585205,
      "learning_rate": 5.15991447271605e-05,
      "loss": 9.5297,
      "step": 1471
    },
    {
      "epoch": 0.5077612969989651,
      "grad_norm": 1.4187387228012085,
      "learning_rate": 5.154305253004264e-05,
      "loss": 9.4879,
      "step": 1472
    },
    {
      "epoch": 0.5081062435322525,
      "grad_norm": 1.2831348180770874,
      "learning_rate": 5.1486958389021655e-05,
      "loss": 9.5302,
      "step": 1473
    },
    {
      "epoch": 0.5084511900655398,
      "grad_norm": 1.333694577217102,
      "learning_rate": 5.143086237476371e-05,
      "loss": 9.4627,
      "step": 1474
    },
    {
      "epoch": 0.5087961365988272,
      "grad_norm": 1.2403720617294312,
      "learning_rate": 5.137476455793732e-05,
      "loss": 9.5265,
      "step": 1475
    },
    {
      "epoch": 0.5091410831321145,
      "grad_norm": 1.301084041595459,
      "learning_rate": 5.131866500921325e-05,
      "loss": 9.5289,
      "step": 1476
    },
    {
      "epoch": 0.5094860296654019,
      "grad_norm": 1.443691372871399,
      "learning_rate": 5.1262563799264455e-05,
      "loss": 9.4393,
      "step": 1477
    },
    {
      "epoch": 0.5098309761986892,
      "grad_norm": 1.3539116382598877,
      "learning_rate": 5.120646099876598e-05,
      "loss": 9.4564,
      "step": 1478
    },
    {
      "epoch": 0.5101759227319765,
      "grad_norm": 1.3923438787460327,
      "learning_rate": 5.1150356678394905e-05,
      "loss": 9.4671,
      "step": 1479
    },
    {
      "epoch": 0.5105208692652639,
      "grad_norm": 1.301438331604004,
      "learning_rate": 5.109425090883019e-05,
      "loss": 9.493,
      "step": 1480
    },
    {
      "epoch": 0.5108658157985512,
      "grad_norm": 1.3289682865142822,
      "learning_rate": 5.10381437607526e-05,
      "loss": 9.5038,
      "step": 1481
    },
    {
      "epoch": 0.5112107623318386,
      "grad_norm": 1.3028764724731445,
      "learning_rate": 5.098203530484471e-05,
      "loss": 9.5781,
      "step": 1482
    },
    {
      "epoch": 0.5115557088651259,
      "grad_norm": 1.3720993995666504,
      "learning_rate": 5.092592561179069e-05,
      "loss": 9.4409,
      "step": 1483
    },
    {
      "epoch": 0.5119006553984132,
      "grad_norm": 1.3018594980239868,
      "learning_rate": 5.086981475227624e-05,
      "loss": 9.5002,
      "step": 1484
    },
    {
      "epoch": 0.5122456019317005,
      "grad_norm": 1.3004437685012817,
      "learning_rate": 5.0813702796988624e-05,
      "loss": 9.4889,
      "step": 1485
    },
    {
      "epoch": 0.512590548464988,
      "grad_norm": 1.32923424243927,
      "learning_rate": 5.07575898166164e-05,
      "loss": 9.4854,
      "step": 1486
    },
    {
      "epoch": 0.5129354949982753,
      "grad_norm": 1.6009782552719116,
      "learning_rate": 5.0701475881849444e-05,
      "loss": 9.4702,
      "step": 1487
    },
    {
      "epoch": 0.5132804415315626,
      "grad_norm": 1.4030383825302124,
      "learning_rate": 5.064536106337883e-05,
      "loss": 9.4459,
      "step": 1488
    },
    {
      "epoch": 0.5136253880648499,
      "grad_norm": 1.4751638174057007,
      "learning_rate": 5.058924543189677e-05,
      "loss": 9.3985,
      "step": 1489
    },
    {
      "epoch": 0.5139703345981372,
      "grad_norm": 1.4806643724441528,
      "learning_rate": 5.0533129058096484e-05,
      "loss": 9.4081,
      "step": 1490
    },
    {
      "epoch": 0.5143152811314247,
      "grad_norm": 1.6850110292434692,
      "learning_rate": 5.047701201267211e-05,
      "loss": 9.3762,
      "step": 1491
    },
    {
      "epoch": 0.514660227664712,
      "grad_norm": 1.448501467704773,
      "learning_rate": 5.042089436631867e-05,
      "loss": 9.3982,
      "step": 1492
    },
    {
      "epoch": 0.5150051741979993,
      "grad_norm": 1.3358256816864014,
      "learning_rate": 5.036477618973191e-05,
      "loss": 9.4275,
      "step": 1493
    },
    {
      "epoch": 0.5153501207312866,
      "grad_norm": 1.3124507665634155,
      "learning_rate": 5.030865755360826e-05,
      "loss": 9.4639,
      "step": 1494
    },
    {
      "epoch": 0.515695067264574,
      "grad_norm": 1.502692461013794,
      "learning_rate": 5.025253852864471e-05,
      "loss": 9.4533,
      "step": 1495
    },
    {
      "epoch": 0.5160400137978614,
      "grad_norm": 1.5912710428237915,
      "learning_rate": 5.019641918553877e-05,
      "loss": 9.3773,
      "step": 1496
    },
    {
      "epoch": 0.5163849603311487,
      "grad_norm": 1.5063811540603638,
      "learning_rate": 5.0140299594988315e-05,
      "loss": 9.4607,
      "step": 1497
    },
    {
      "epoch": 0.516729906864436,
      "grad_norm": 1.5693217515945435,
      "learning_rate": 5.0084179827691566e-05,
      "loss": 9.4344,
      "step": 1498
    },
    {
      "epoch": 0.5170748533977233,
      "grad_norm": 1.6469818353652954,
      "learning_rate": 5.0028059954346964e-05,
      "loss": 9.3941,
      "step": 1499
    },
    {
      "epoch": 0.5174197999310107,
      "grad_norm": 1.855391025543213,
      "learning_rate": 4.997194004565304e-05,
      "loss": 9.3555,
      "step": 1500
    },
    {
      "epoch": 0.517764746464298,
      "grad_norm": 0.9066408276557922,
      "learning_rate": 4.991582017230843e-05,
      "loss": 9.608,
      "step": 1501
    },
    {
      "epoch": 0.5181096929975854,
      "grad_norm": 0.9518599510192871,
      "learning_rate": 4.985970040501169e-05,
      "loss": 9.5632,
      "step": 1502
    },
    {
      "epoch": 0.5184546395308727,
      "grad_norm": 1.1453851461410522,
      "learning_rate": 4.980358081446125e-05,
      "loss": 9.5027,
      "step": 1503
    },
    {
      "epoch": 0.5187995860641601,
      "grad_norm": 1.0721421241760254,
      "learning_rate": 4.9747461471355306e-05,
      "loss": 9.5719,
      "step": 1504
    },
    {
      "epoch": 0.5191445325974474,
      "grad_norm": 1.0002890825271606,
      "learning_rate": 4.969134244639176e-05,
      "loss": 9.5454,
      "step": 1505
    },
    {
      "epoch": 0.5194894791307347,
      "grad_norm": 1.0482701063156128,
      "learning_rate": 4.9635223810268104e-05,
      "loss": 9.4886,
      "step": 1506
    },
    {
      "epoch": 0.519834425664022,
      "grad_norm": 1.034656286239624,
      "learning_rate": 4.957910563368133e-05,
      "loss": 9.6031,
      "step": 1507
    },
    {
      "epoch": 0.5201793721973094,
      "grad_norm": 1.1927002668380737,
      "learning_rate": 4.9522987987327886e-05,
      "loss": 9.5221,
      "step": 1508
    },
    {
      "epoch": 0.5205243187305968,
      "grad_norm": 1.0671268701553345,
      "learning_rate": 4.946687094190353e-05,
      "loss": 9.5753,
      "step": 1509
    },
    {
      "epoch": 0.5208692652638841,
      "grad_norm": 1.1840550899505615,
      "learning_rate": 4.941075456810324e-05,
      "loss": 9.5308,
      "step": 1510
    },
    {
      "epoch": 0.5212142117971714,
      "grad_norm": 1.0830540657043457,
      "learning_rate": 4.9354638936621177e-05,
      "loss": 9.5537,
      "step": 1511
    },
    {
      "epoch": 0.5215591583304587,
      "grad_norm": 1.1524994373321533,
      "learning_rate": 4.929852411815058e-05,
      "loss": 9.5055,
      "step": 1512
    },
    {
      "epoch": 0.5219041048637462,
      "grad_norm": 1.2321434020996094,
      "learning_rate": 4.9242410183383624e-05,
      "loss": 9.478,
      "step": 1513
    },
    {
      "epoch": 0.5222490513970335,
      "grad_norm": 1.2240982055664062,
      "learning_rate": 4.9186297203011374e-05,
      "loss": 9.4934,
      "step": 1514
    },
    {
      "epoch": 0.5225939979303208,
      "grad_norm": 1.1383882761001587,
      "learning_rate": 4.913018524772375e-05,
      "loss": 9.5314,
      "step": 1515
    },
    {
      "epoch": 0.5229389444636081,
      "grad_norm": 1.1765910387039185,
      "learning_rate": 4.907407438820932e-05,
      "loss": 9.5152,
      "step": 1516
    },
    {
      "epoch": 0.5232838909968954,
      "grad_norm": 1.3054953813552856,
      "learning_rate": 4.90179646951553e-05,
      "loss": 9.4555,
      "step": 1517
    },
    {
      "epoch": 0.5236288375301829,
      "grad_norm": 1.2543056011199951,
      "learning_rate": 4.8961856239247406e-05,
      "loss": 9.5529,
      "step": 1518
    },
    {
      "epoch": 0.5239737840634702,
      "grad_norm": 1.239845633506775,
      "learning_rate": 4.890574909116984e-05,
      "loss": 9.5519,
      "step": 1519
    },
    {
      "epoch": 0.5243187305967575,
      "grad_norm": 1.1929110288619995,
      "learning_rate": 4.8849643321605107e-05,
      "loss": 9.526,
      "step": 1520
    },
    {
      "epoch": 0.5246636771300448,
      "grad_norm": 1.391808032989502,
      "learning_rate": 4.8793539001234025e-05,
      "loss": 9.4912,
      "step": 1521
    },
    {
      "epoch": 0.5250086236633322,
      "grad_norm": 1.3413889408111572,
      "learning_rate": 4.8737436200735564e-05,
      "loss": 9.4721,
      "step": 1522
    },
    {
      "epoch": 0.5253535701966195,
      "grad_norm": 1.3500343561172485,
      "learning_rate": 4.868133499078676e-05,
      "loss": 9.4887,
      "step": 1523
    },
    {
      "epoch": 0.5256985167299069,
      "grad_norm": 1.2600687742233276,
      "learning_rate": 4.8625235442062686e-05,
      "loss": 9.551,
      "step": 1524
    },
    {
      "epoch": 0.5260434632631942,
      "grad_norm": 1.4446165561676025,
      "learning_rate": 4.85691376252363e-05,
      "loss": 9.4995,
      "step": 1525
    },
    {
      "epoch": 0.5263884097964815,
      "grad_norm": 1.2388883829116821,
      "learning_rate": 4.851304161097836e-05,
      "loss": 9.5027,
      "step": 1526
    },
    {
      "epoch": 0.5267333563297689,
      "grad_norm": 1.4213436841964722,
      "learning_rate": 4.8456947469957395e-05,
      "loss": 9.4535,
      "step": 1527
    },
    {
      "epoch": 0.5270783028630562,
      "grad_norm": 1.3508599996566772,
      "learning_rate": 4.84008552728395e-05,
      "loss": 9.4191,
      "step": 1528
    },
    {
      "epoch": 0.5274232493963436,
      "grad_norm": 1.305248737335205,
      "learning_rate": 4.8344765090288436e-05,
      "loss": 9.4776,
      "step": 1529
    },
    {
      "epoch": 0.5277681959296309,
      "grad_norm": 1.4581810235977173,
      "learning_rate": 4.828867699296534e-05,
      "loss": 9.4615,
      "step": 1530
    },
    {
      "epoch": 0.5281131424629183,
      "grad_norm": 1.2803398370742798,
      "learning_rate": 4.823259105152873e-05,
      "loss": 9.5474,
      "step": 1531
    },
    {
      "epoch": 0.5284580889962056,
      "grad_norm": 1.3474336862564087,
      "learning_rate": 4.8176507336634435e-05,
      "loss": 9.4794,
      "step": 1532
    },
    {
      "epoch": 0.5288030355294929,
      "grad_norm": 1.5129001140594482,
      "learning_rate": 4.812042591893545e-05,
      "loss": 9.3918,
      "step": 1533
    },
    {
      "epoch": 0.5291479820627802,
      "grad_norm": 1.2152923345565796,
      "learning_rate": 4.806434686908191e-05,
      "loss": 9.5461,
      "step": 1534
    },
    {
      "epoch": 0.5294929285960676,
      "grad_norm": 1.4519068002700806,
      "learning_rate": 4.800827025772094e-05,
      "loss": 9.4692,
      "step": 1535
    },
    {
      "epoch": 0.529837875129355,
      "grad_norm": 1.441679835319519,
      "learning_rate": 4.79521961554966e-05,
      "loss": 9.4283,
      "step": 1536
    },
    {
      "epoch": 0.5301828216626423,
      "grad_norm": 1.451649785041809,
      "learning_rate": 4.789612463304981e-05,
      "loss": 9.3841,
      "step": 1537
    },
    {
      "epoch": 0.5305277681959296,
      "grad_norm": 1.3632525205612183,
      "learning_rate": 4.784005576101821e-05,
      "loss": 9.5185,
      "step": 1538
    },
    {
      "epoch": 0.5308727147292169,
      "grad_norm": 1.319543719291687,
      "learning_rate": 4.778398961003611e-05,
      "loss": 9.5255,
      "step": 1539
    },
    {
      "epoch": 0.5312176612625044,
      "grad_norm": 1.4514024257659912,
      "learning_rate": 4.7727926250734396e-05,
      "loss": 9.5116,
      "step": 1540
    },
    {
      "epoch": 0.5315626077957917,
      "grad_norm": 1.3405132293701172,
      "learning_rate": 4.767186575374043e-05,
      "loss": 9.4577,
      "step": 1541
    },
    {
      "epoch": 0.531907554329079,
      "grad_norm": 1.5354069471359253,
      "learning_rate": 4.761580818967796e-05,
      "loss": 9.4914,
      "step": 1542
    },
    {
      "epoch": 0.5322525008623663,
      "grad_norm": 1.3511700630187988,
      "learning_rate": 4.755975362916709e-05,
      "loss": 9.4602,
      "step": 1543
    },
    {
      "epoch": 0.5325974473956536,
      "grad_norm": 1.4528660774230957,
      "learning_rate": 4.750370214282409e-05,
      "loss": 9.403,
      "step": 1544
    },
    {
      "epoch": 0.532942393928941,
      "grad_norm": 1.4760255813598633,
      "learning_rate": 4.7447653801261355e-05,
      "loss": 9.402,
      "step": 1545
    },
    {
      "epoch": 0.5332873404622284,
      "grad_norm": 1.5060619115829468,
      "learning_rate": 4.739160867508735e-05,
      "loss": 9.3647,
      "step": 1546
    },
    {
      "epoch": 0.5336322869955157,
      "grad_norm": 1.6048104763031006,
      "learning_rate": 4.7335566834906466e-05,
      "loss": 9.3865,
      "step": 1547
    },
    {
      "epoch": 0.533977233528803,
      "grad_norm": 1.4714221954345703,
      "learning_rate": 4.727952835131897e-05,
      "loss": 9.455,
      "step": 1548
    },
    {
      "epoch": 0.5343221800620904,
      "grad_norm": 1.62465500831604,
      "learning_rate": 4.7223493294920853e-05,
      "loss": 9.3758,
      "step": 1549
    },
    {
      "epoch": 0.5346671265953777,
      "grad_norm": 1.764082908630371,
      "learning_rate": 4.716746173630388e-05,
      "loss": 9.3348,
      "step": 1550
    },
    {
      "epoch": 0.535012073128665,
      "grad_norm": 0.9597568511962891,
      "learning_rate": 4.711143374605534e-05,
      "loss": 9.5354,
      "step": 1551
    },
    {
      "epoch": 0.5353570196619524,
      "grad_norm": 1.048038125038147,
      "learning_rate": 4.705540939475803e-05,
      "loss": 9.5423,
      "step": 1552
    },
    {
      "epoch": 0.5357019661952397,
      "grad_norm": 1.1174081563949585,
      "learning_rate": 4.699938875299019e-05,
      "loss": 9.4938,
      "step": 1553
    },
    {
      "epoch": 0.5360469127285271,
      "grad_norm": 1.1701761484146118,
      "learning_rate": 4.694337189132537e-05,
      "loss": 9.5567,
      "step": 1554
    },
    {
      "epoch": 0.5363918592618144,
      "grad_norm": 1.0438191890716553,
      "learning_rate": 4.6887358880332346e-05,
      "loss": 9.5207,
      "step": 1555
    },
    {
      "epoch": 0.5367368057951017,
      "grad_norm": 1.1834765672683716,
      "learning_rate": 4.683134979057507e-05,
      "loss": 9.4953,
      "step": 1556
    },
    {
      "epoch": 0.5370817523283891,
      "grad_norm": 1.191499948501587,
      "learning_rate": 4.677534469261254e-05,
      "loss": 9.5511,
      "step": 1557
    },
    {
      "epoch": 0.5374266988616765,
      "grad_norm": 1.1501351594924927,
      "learning_rate": 4.671934365699874e-05,
      "loss": 9.5603,
      "step": 1558
    },
    {
      "epoch": 0.5377716453949638,
      "grad_norm": 1.171834111213684,
      "learning_rate": 4.666334675428252e-05,
      "loss": 9.5079,
      "step": 1559
    },
    {
      "epoch": 0.5381165919282511,
      "grad_norm": 1.2038573026657104,
      "learning_rate": 4.660735405500751e-05,
      "loss": 9.5096,
      "step": 1560
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 1.138796091079712,
      "learning_rate": 4.655136562971208e-05,
      "loss": 9.5425,
      "step": 1561
    },
    {
      "epoch": 0.5388064849948258,
      "grad_norm": 1.2308675050735474,
      "learning_rate": 4.649538154892919e-05,
      "loss": 9.5467,
      "step": 1562
    },
    {
      "epoch": 0.5391514315281132,
      "grad_norm": 1.1142117977142334,
      "learning_rate": 4.643940188318631e-05,
      "loss": 9.5166,
      "step": 1563
    },
    {
      "epoch": 0.5394963780614005,
      "grad_norm": 1.2303515672683716,
      "learning_rate": 4.638342670300542e-05,
      "loss": 9.5361,
      "step": 1564
    },
    {
      "epoch": 0.5398413245946878,
      "grad_norm": 1.2311829328536987,
      "learning_rate": 4.632745607890278e-05,
      "loss": 9.4819,
      "step": 1565
    },
    {
      "epoch": 0.5401862711279751,
      "grad_norm": 1.3217933177947998,
      "learning_rate": 4.627149008138892e-05,
      "loss": 9.5043,
      "step": 1566
    },
    {
      "epoch": 0.5405312176612626,
      "grad_norm": 1.1946293115615845,
      "learning_rate": 4.621552878096857e-05,
      "loss": 9.5427,
      "step": 1567
    },
    {
      "epoch": 0.5408761641945499,
      "grad_norm": 1.326340675354004,
      "learning_rate": 4.615957224814051e-05,
      "loss": 9.4889,
      "step": 1568
    },
    {
      "epoch": 0.5412211107278372,
      "grad_norm": 1.1464662551879883,
      "learning_rate": 4.6103620553397556e-05,
      "loss": 9.6259,
      "step": 1569
    },
    {
      "epoch": 0.5415660572611245,
      "grad_norm": 1.2796909809112549,
      "learning_rate": 4.604767376722635e-05,
      "loss": 9.4796,
      "step": 1570
    },
    {
      "epoch": 0.5419110037944118,
      "grad_norm": 1.395712971687317,
      "learning_rate": 4.599173196010747e-05,
      "loss": 9.507,
      "step": 1571
    },
    {
      "epoch": 0.5422559503276992,
      "grad_norm": 1.3465749025344849,
      "learning_rate": 4.593579520251512e-05,
      "loss": 9.556,
      "step": 1572
    },
    {
      "epoch": 0.5426008968609866,
      "grad_norm": 1.215400218963623,
      "learning_rate": 4.587986356491719e-05,
      "loss": 9.5338,
      "step": 1573
    },
    {
      "epoch": 0.5429458433942739,
      "grad_norm": 1.1242035627365112,
      "learning_rate": 4.58239371177751e-05,
      "loss": 9.564,
      "step": 1574
    },
    {
      "epoch": 0.5432907899275612,
      "grad_norm": 1.254112720489502,
      "learning_rate": 4.5768015931543736e-05,
      "loss": 9.5347,
      "step": 1575
    },
    {
      "epoch": 0.5436357364608486,
      "grad_norm": 1.2248470783233643,
      "learning_rate": 4.571210007667135e-05,
      "loss": 9.4812,
      "step": 1576
    },
    {
      "epoch": 0.5439806829941359,
      "grad_norm": 1.480251669883728,
      "learning_rate": 4.56561896235995e-05,
      "loss": 9.4632,
      "step": 1577
    },
    {
      "epoch": 0.5443256295274232,
      "grad_norm": 1.379051685333252,
      "learning_rate": 4.560028464276293e-05,
      "loss": 9.4717,
      "step": 1578
    },
    {
      "epoch": 0.5446705760607106,
      "grad_norm": 1.3023099899291992,
      "learning_rate": 4.554438520458948e-05,
      "loss": 9.5061,
      "step": 1579
    },
    {
      "epoch": 0.5450155225939979,
      "grad_norm": 1.3963764905929565,
      "learning_rate": 4.548849137950001e-05,
      "loss": 9.5005,
      "step": 1580
    },
    {
      "epoch": 0.5453604691272853,
      "grad_norm": 1.3574894666671753,
      "learning_rate": 4.543260323790833e-05,
      "loss": 9.4728,
      "step": 1581
    },
    {
      "epoch": 0.5457054156605726,
      "grad_norm": 1.3623203039169312,
      "learning_rate": 4.5376720850221054e-05,
      "loss": 9.451,
      "step": 1582
    },
    {
      "epoch": 0.5460503621938599,
      "grad_norm": 1.5018421411514282,
      "learning_rate": 4.532084428683757e-05,
      "loss": 9.4535,
      "step": 1583
    },
    {
      "epoch": 0.5463953087271473,
      "grad_norm": 1.3744196891784668,
      "learning_rate": 4.526497361814991e-05,
      "loss": 9.4854,
      "step": 1584
    },
    {
      "epoch": 0.5467402552604347,
      "grad_norm": 1.177968144416809,
      "learning_rate": 4.520910891454272e-05,
      "loss": 9.5429,
      "step": 1585
    },
    {
      "epoch": 0.547085201793722,
      "grad_norm": 1.4899221658706665,
      "learning_rate": 4.51532502463931e-05,
      "loss": 9.4118,
      "step": 1586
    },
    {
      "epoch": 0.5474301483270093,
      "grad_norm": 1.3041064739227295,
      "learning_rate": 4.509739768407054e-05,
      "loss": 9.484,
      "step": 1587
    },
    {
      "epoch": 0.5477750948602966,
      "grad_norm": 1.3999149799346924,
      "learning_rate": 4.5041551297936865e-05,
      "loss": 9.4015,
      "step": 1588
    },
    {
      "epoch": 0.548120041393584,
      "grad_norm": 1.323775053024292,
      "learning_rate": 4.498571115834608e-05,
      "loss": 9.4503,
      "step": 1589
    },
    {
      "epoch": 0.5484649879268714,
      "grad_norm": 1.481468915939331,
      "learning_rate": 4.4929877335644376e-05,
      "loss": 9.4589,
      "step": 1590
    },
    {
      "epoch": 0.5488099344601587,
      "grad_norm": 1.4510349035263062,
      "learning_rate": 4.487404990016993e-05,
      "loss": 9.4337,
      "step": 1591
    },
    {
      "epoch": 0.549154880993446,
      "grad_norm": 1.4892961978912354,
      "learning_rate": 4.4818228922252916e-05,
      "loss": 9.4147,
      "step": 1592
    },
    {
      "epoch": 0.5494998275267333,
      "grad_norm": 1.451235055923462,
      "learning_rate": 4.4762414472215354e-05,
      "loss": 9.4324,
      "step": 1593
    },
    {
      "epoch": 0.5498447740600207,
      "grad_norm": 1.5764260292053223,
      "learning_rate": 4.4706606620371037e-05,
      "loss": 9.4358,
      "step": 1594
    },
    {
      "epoch": 0.5501897205933081,
      "grad_norm": 1.3623017072677612,
      "learning_rate": 4.4650805437025446e-05,
      "loss": 9.4655,
      "step": 1595
    },
    {
      "epoch": 0.5505346671265954,
      "grad_norm": 1.5152086019515991,
      "learning_rate": 4.459501099247567e-05,
      "loss": 9.4347,
      "step": 1596
    },
    {
      "epoch": 0.5508796136598827,
      "grad_norm": 1.5371122360229492,
      "learning_rate": 4.4539223357010304e-05,
      "loss": 9.3836,
      "step": 1597
    },
    {
      "epoch": 0.55122456019317,
      "grad_norm": 1.666603922843933,
      "learning_rate": 4.4483442600909333e-05,
      "loss": 9.3984,
      "step": 1598
    },
    {
      "epoch": 0.5515695067264574,
      "grad_norm": 1.5026121139526367,
      "learning_rate": 4.442766879444417e-05,
      "loss": 9.4533,
      "step": 1599
    },
    {
      "epoch": 0.5519144532597448,
      "grad_norm": 1.6777715682983398,
      "learning_rate": 4.437190200787737e-05,
      "loss": 9.3604,
      "step": 1600
    },
    {
      "epoch": 0.5522593997930321,
      "grad_norm": 1.0869591236114502,
      "learning_rate": 4.431614231146269e-05,
      "loss": 9.5396,
      "step": 1601
    },
    {
      "epoch": 0.5526043463263194,
      "grad_norm": 1.0521719455718994,
      "learning_rate": 4.426038977544496e-05,
      "loss": 9.5658,
      "step": 1602
    },
    {
      "epoch": 0.5529492928596068,
      "grad_norm": 1.0579216480255127,
      "learning_rate": 4.420464447005997e-05,
      "loss": 9.5042,
      "step": 1603
    },
    {
      "epoch": 0.5532942393928941,
      "grad_norm": 1.0266507863998413,
      "learning_rate": 4.414890646553442e-05,
      "loss": 9.5387,
      "step": 1604
    },
    {
      "epoch": 0.5536391859261814,
      "grad_norm": 1.0023735761642456,
      "learning_rate": 4.409317583208576e-05,
      "loss": 9.5934,
      "step": 1605
    },
    {
      "epoch": 0.5539841324594688,
      "grad_norm": 1.1149392127990723,
      "learning_rate": 4.403745263992227e-05,
      "loss": 9.563,
      "step": 1606
    },
    {
      "epoch": 0.5543290789927561,
      "grad_norm": 1.165576457977295,
      "learning_rate": 4.3981736959242734e-05,
      "loss": 9.5337,
      "step": 1607
    },
    {
      "epoch": 0.5546740255260435,
      "grad_norm": 1.1478943824768066,
      "learning_rate": 4.392602886023653e-05,
      "loss": 9.5396,
      "step": 1608
    },
    {
      "epoch": 0.5550189720593308,
      "grad_norm": 1.1021052598953247,
      "learning_rate": 4.38703284130835e-05,
      "loss": 9.5555,
      "step": 1609
    },
    {
      "epoch": 0.5553639185926181,
      "grad_norm": 1.2310773134231567,
      "learning_rate": 4.381463568795378e-05,
      "loss": 9.4717,
      "step": 1610
    },
    {
      "epoch": 0.5557088651259054,
      "grad_norm": 1.1044448614120483,
      "learning_rate": 4.375895075500784e-05,
      "loss": 9.5253,
      "step": 1611
    },
    {
      "epoch": 0.5560538116591929,
      "grad_norm": 1.3454209566116333,
      "learning_rate": 4.370327368439633e-05,
      "loss": 9.48,
      "step": 1612
    },
    {
      "epoch": 0.5563987581924802,
      "grad_norm": 1.3490056991577148,
      "learning_rate": 4.364760454625997e-05,
      "loss": 9.5315,
      "step": 1613
    },
    {
      "epoch": 0.5567437047257675,
      "grad_norm": 1.1780604124069214,
      "learning_rate": 4.359194341072948e-05,
      "loss": 9.5037,
      "step": 1614
    },
    {
      "epoch": 0.5570886512590548,
      "grad_norm": 1.2844969034194946,
      "learning_rate": 4.3536290347925544e-05,
      "loss": 9.4968,
      "step": 1615
    },
    {
      "epoch": 0.5574335977923421,
      "grad_norm": 1.2029293775558472,
      "learning_rate": 4.3480645427958634e-05,
      "loss": 9.5103,
      "step": 1616
    },
    {
      "epoch": 0.5577785443256296,
      "grad_norm": 1.1009153127670288,
      "learning_rate": 4.342500872092897e-05,
      "loss": 9.5644,
      "step": 1617
    },
    {
      "epoch": 0.5581234908589169,
      "grad_norm": 1.240302562713623,
      "learning_rate": 4.336938029692645e-05,
      "loss": 9.546,
      "step": 1618
    },
    {
      "epoch": 0.5584684373922042,
      "grad_norm": 1.3886947631835938,
      "learning_rate": 4.331376022603048e-05,
      "loss": 9.4193,
      "step": 1619
    },
    {
      "epoch": 0.5588133839254915,
      "grad_norm": 1.3278820514678955,
      "learning_rate": 4.3258148578310033e-05,
      "loss": 9.5184,
      "step": 1620
    },
    {
      "epoch": 0.5591583304587789,
      "grad_norm": 1.2009872198104858,
      "learning_rate": 4.3202545423823406e-05,
      "loss": 9.5146,
      "step": 1621
    },
    {
      "epoch": 0.5595032769920663,
      "grad_norm": 1.2812772989273071,
      "learning_rate": 4.31469508326182e-05,
      "loss": 9.5334,
      "step": 1622
    },
    {
      "epoch": 0.5598482235253536,
      "grad_norm": 1.3486807346343994,
      "learning_rate": 4.309136487473122e-05,
      "loss": 9.4973,
      "step": 1623
    },
    {
      "epoch": 0.5601931700586409,
      "grad_norm": 1.192238211631775,
      "learning_rate": 4.303578762018846e-05,
      "loss": 9.5171,
      "step": 1624
    },
    {
      "epoch": 0.5605381165919282,
      "grad_norm": 1.329584002494812,
      "learning_rate": 4.298021913900488e-05,
      "loss": 9.5528,
      "step": 1625
    },
    {
      "epoch": 0.5608830631252156,
      "grad_norm": 1.3967713117599487,
      "learning_rate": 4.292465950118439e-05,
      "loss": 9.5038,
      "step": 1626
    },
    {
      "epoch": 0.561228009658503,
      "grad_norm": 1.1745870113372803,
      "learning_rate": 4.286910877671982e-05,
      "loss": 9.5121,
      "step": 1627
    },
    {
      "epoch": 0.5615729561917903,
      "grad_norm": 1.3389042615890503,
      "learning_rate": 4.281356703559271e-05,
      "loss": 9.4378,
      "step": 1628
    },
    {
      "epoch": 0.5619179027250776,
      "grad_norm": 1.3878090381622314,
      "learning_rate": 4.275803434777332e-05,
      "loss": 9.4766,
      "step": 1629
    },
    {
      "epoch": 0.562262849258365,
      "grad_norm": 1.4438650608062744,
      "learning_rate": 4.270251078322048e-05,
      "loss": 9.3935,
      "step": 1630
    },
    {
      "epoch": 0.5626077957916523,
      "grad_norm": 1.3266236782073975,
      "learning_rate": 4.264699641188153e-05,
      "loss": 9.5028,
      "step": 1631
    },
    {
      "epoch": 0.5629527423249396,
      "grad_norm": 1.347325086593628,
      "learning_rate": 4.259149130369224e-05,
      "loss": 9.4892,
      "step": 1632
    },
    {
      "epoch": 0.563297688858227,
      "grad_norm": 1.3031338453292847,
      "learning_rate": 4.253599552857668e-05,
      "loss": 9.5652,
      "step": 1633
    },
    {
      "epoch": 0.5636426353915143,
      "grad_norm": 1.4257103204727173,
      "learning_rate": 4.2480509156447235e-05,
      "loss": 9.4456,
      "step": 1634
    },
    {
      "epoch": 0.5639875819248017,
      "grad_norm": 1.4598137140274048,
      "learning_rate": 4.242503225720437e-05,
      "loss": 9.4028,
      "step": 1635
    },
    {
      "epoch": 0.564332528458089,
      "grad_norm": 1.364773154258728,
      "learning_rate": 4.236956490073665e-05,
      "loss": 9.4691,
      "step": 1636
    },
    {
      "epoch": 0.5646774749913763,
      "grad_norm": 1.4265398979187012,
      "learning_rate": 4.2314107156920613e-05,
      "loss": 9.519,
      "step": 1637
    },
    {
      "epoch": 0.5650224215246636,
      "grad_norm": 1.3014031648635864,
      "learning_rate": 4.2258659095620686e-05,
      "loss": 9.4917,
      "step": 1638
    },
    {
      "epoch": 0.5653673680579511,
      "grad_norm": 1.33185613155365,
      "learning_rate": 4.220322078668909e-05,
      "loss": 9.4324,
      "step": 1639
    },
    {
      "epoch": 0.5657123145912384,
      "grad_norm": 1.449734091758728,
      "learning_rate": 4.214779229996575e-05,
      "loss": 9.4531,
      "step": 1640
    },
    {
      "epoch": 0.5660572611245257,
      "grad_norm": 1.4314883947372437,
      "learning_rate": 4.209237370527828e-05,
      "loss": 9.4829,
      "step": 1641
    },
    {
      "epoch": 0.566402207657813,
      "grad_norm": 1.4500685930252075,
      "learning_rate": 4.203696507244177e-05,
      "loss": 9.4823,
      "step": 1642
    },
    {
      "epoch": 0.5667471541911003,
      "grad_norm": 1.4503802061080933,
      "learning_rate": 4.198156647125877e-05,
      "loss": 9.4211,
      "step": 1643
    },
    {
      "epoch": 0.5670921007243878,
      "grad_norm": 1.5186033248901367,
      "learning_rate": 4.192617797151921e-05,
      "loss": 9.3657,
      "step": 1644
    },
    {
      "epoch": 0.5674370472576751,
      "grad_norm": 1.5139539241790771,
      "learning_rate": 4.187079964300026e-05,
      "loss": 9.4758,
      "step": 1645
    },
    {
      "epoch": 0.5677819937909624,
      "grad_norm": 1.5969244241714478,
      "learning_rate": 4.181543155546633e-05,
      "loss": 9.4532,
      "step": 1646
    },
    {
      "epoch": 0.5681269403242497,
      "grad_norm": 1.6448180675506592,
      "learning_rate": 4.176007377866888e-05,
      "loss": 9.3931,
      "step": 1647
    },
    {
      "epoch": 0.5684718868575371,
      "grad_norm": 1.6749922037124634,
      "learning_rate": 4.170472638234641e-05,
      "loss": 9.3962,
      "step": 1648
    },
    {
      "epoch": 0.5688168333908245,
      "grad_norm": 1.5745540857315063,
      "learning_rate": 4.164938943622434e-05,
      "loss": 9.4104,
      "step": 1649
    },
    {
      "epoch": 0.5691617799241118,
      "grad_norm": 1.7174612283706665,
      "learning_rate": 4.159406301001492e-05,
      "loss": 9.441,
      "step": 1650
    },
    {
      "epoch": 0.5695067264573991,
      "grad_norm": 0.7641019821166992,
      "learning_rate": 4.153874717341713e-05,
      "loss": 9.5972,
      "step": 1651
    },
    {
      "epoch": 0.5698516729906864,
      "grad_norm": 0.890247642993927,
      "learning_rate": 4.148344199611664e-05,
      "loss": 9.6203,
      "step": 1652
    },
    {
      "epoch": 0.5701966195239738,
      "grad_norm": 1.0681867599487305,
      "learning_rate": 4.142814754778566e-05,
      "loss": 9.5632,
      "step": 1653
    },
    {
      "epoch": 0.5705415660572611,
      "grad_norm": 1.0469846725463867,
      "learning_rate": 4.1372863898082895e-05,
      "loss": 9.533,
      "step": 1654
    },
    {
      "epoch": 0.5708865125905485,
      "grad_norm": 1.002163052558899,
      "learning_rate": 4.131759111665349e-05,
      "loss": 9.5788,
      "step": 1655
    },
    {
      "epoch": 0.5712314591238358,
      "grad_norm": 1.1065441370010376,
      "learning_rate": 4.126232927312883e-05,
      "loss": 9.5626,
      "step": 1656
    },
    {
      "epoch": 0.5715764056571232,
      "grad_norm": 1.079107642173767,
      "learning_rate": 4.120707843712656e-05,
      "loss": 9.4588,
      "step": 1657
    },
    {
      "epoch": 0.5719213521904105,
      "grad_norm": 1.0924365520477295,
      "learning_rate": 4.1151838678250443e-05,
      "loss": 9.5678,
      "step": 1658
    },
    {
      "epoch": 0.5722662987236978,
      "grad_norm": 1.1700769662857056,
      "learning_rate": 4.1096610066090304e-05,
      "loss": 9.5389,
      "step": 1659
    },
    {
      "epoch": 0.5726112452569851,
      "grad_norm": 1.1450729370117188,
      "learning_rate": 4.104139267022191e-05,
      "loss": 9.4952,
      "step": 1660
    },
    {
      "epoch": 0.5729561917902725,
      "grad_norm": 1.2342902421951294,
      "learning_rate": 4.098618656020688e-05,
      "loss": 9.4941,
      "step": 1661
    },
    {
      "epoch": 0.5733011383235599,
      "grad_norm": 1.2648087739944458,
      "learning_rate": 4.093099180559268e-05,
      "loss": 9.537,
      "step": 1662
    },
    {
      "epoch": 0.5736460848568472,
      "grad_norm": 1.2142956256866455,
      "learning_rate": 4.0875808475912415e-05,
      "loss": 9.5433,
      "step": 1663
    },
    {
      "epoch": 0.5739910313901345,
      "grad_norm": 1.1451528072357178,
      "learning_rate": 4.0820636640684797e-05,
      "loss": 9.5745,
      "step": 1664
    },
    {
      "epoch": 0.5743359779234218,
      "grad_norm": 1.2990100383758545,
      "learning_rate": 4.076547636941408e-05,
      "loss": 9.5221,
      "step": 1665
    },
    {
      "epoch": 0.5746809244567093,
      "grad_norm": 1.1534736156463623,
      "learning_rate": 4.0710327731589935e-05,
      "loss": 9.5126,
      "step": 1666
    },
    {
      "epoch": 0.5750258709899966,
      "grad_norm": 1.3099054098129272,
      "learning_rate": 4.0655190796687385e-05,
      "loss": 9.5155,
      "step": 1667
    },
    {
      "epoch": 0.5753708175232839,
      "grad_norm": 1.176540732383728,
      "learning_rate": 4.060006563416672e-05,
      "loss": 9.5128,
      "step": 1668
    },
    {
      "epoch": 0.5757157640565712,
      "grad_norm": 1.3951212167739868,
      "learning_rate": 4.0544952313473376e-05,
      "loss": 9.4966,
      "step": 1669
    },
    {
      "epoch": 0.5760607105898585,
      "grad_norm": 1.2419620752334595,
      "learning_rate": 4.048985090403788e-05,
      "loss": 9.4978,
      "step": 1670
    },
    {
      "epoch": 0.576405657123146,
      "grad_norm": 1.2248680591583252,
      "learning_rate": 4.043476147527579e-05,
      "loss": 9.531,
      "step": 1671
    },
    {
      "epoch": 0.5767506036564333,
      "grad_norm": 1.3265471458435059,
      "learning_rate": 4.0379684096587504e-05,
      "loss": 9.4789,
      "step": 1672
    },
    {
      "epoch": 0.5770955501897206,
      "grad_norm": 1.2454609870910645,
      "learning_rate": 4.0324618837358285e-05,
      "loss": 9.5451,
      "step": 1673
    },
    {
      "epoch": 0.5774404967230079,
      "grad_norm": 1.2326449155807495,
      "learning_rate": 4.026956576695811e-05,
      "loss": 9.563,
      "step": 1674
    },
    {
      "epoch": 0.5777854432562953,
      "grad_norm": 1.3096765279769897,
      "learning_rate": 4.021452495474159e-05,
      "loss": 9.4312,
      "step": 1675
    },
    {
      "epoch": 0.5781303897895826,
      "grad_norm": 1.3253740072250366,
      "learning_rate": 4.0159496470047954e-05,
      "loss": 9.4891,
      "step": 1676
    },
    {
      "epoch": 0.57847533632287,
      "grad_norm": 1.3036460876464844,
      "learning_rate": 4.0104480382200836e-05,
      "loss": 9.5537,
      "step": 1677
    },
    {
      "epoch": 0.5788202828561573,
      "grad_norm": 1.1840957403182983,
      "learning_rate": 4.004947676050828e-05,
      "loss": 9.5231,
      "step": 1678
    },
    {
      "epoch": 0.5791652293894446,
      "grad_norm": 1.319153904914856,
      "learning_rate": 3.999448567426259e-05,
      "loss": 9.4613,
      "step": 1679
    },
    {
      "epoch": 0.579510175922732,
      "grad_norm": 1.3345825672149658,
      "learning_rate": 3.993950719274035e-05,
      "loss": 9.4576,
      "step": 1680
    },
    {
      "epoch": 0.5798551224560193,
      "grad_norm": 1.373167634010315,
      "learning_rate": 3.988454138520219e-05,
      "loss": 9.4671,
      "step": 1681
    },
    {
      "epoch": 0.5802000689893066,
      "grad_norm": 1.2741271257400513,
      "learning_rate": 3.982958832089282e-05,
      "loss": 9.5347,
      "step": 1682
    },
    {
      "epoch": 0.580545015522594,
      "grad_norm": 1.4847259521484375,
      "learning_rate": 3.97746480690409e-05,
      "loss": 9.4661,
      "step": 1683
    },
    {
      "epoch": 0.5808899620558814,
      "grad_norm": 1.3877465724945068,
      "learning_rate": 3.971972069885891e-05,
      "loss": 9.469,
      "step": 1684
    },
    {
      "epoch": 0.5812349085891687,
      "grad_norm": 1.5072605609893799,
      "learning_rate": 3.9664806279543136e-05,
      "loss": 9.4667,
      "step": 1685
    },
    {
      "epoch": 0.581579855122456,
      "grad_norm": 1.4354532957077026,
      "learning_rate": 3.960990488027353e-05,
      "loss": 9.4154,
      "step": 1686
    },
    {
      "epoch": 0.5819248016557433,
      "grad_norm": 1.4752713441848755,
      "learning_rate": 3.955501657021364e-05,
      "loss": 9.4524,
      "step": 1687
    },
    {
      "epoch": 0.5822697481890307,
      "grad_norm": 1.3264256715774536,
      "learning_rate": 3.9500141418510526e-05,
      "loss": 9.5238,
      "step": 1688
    },
    {
      "epoch": 0.5826146947223181,
      "grad_norm": 1.3095241785049438,
      "learning_rate": 3.944527949429472e-05,
      "loss": 9.4962,
      "step": 1689
    },
    {
      "epoch": 0.5829596412556054,
      "grad_norm": 1.382723093032837,
      "learning_rate": 3.9390430866680016e-05,
      "loss": 9.4848,
      "step": 1690
    },
    {
      "epoch": 0.5833045877888927,
      "grad_norm": 1.508972406387329,
      "learning_rate": 3.933559560476349e-05,
      "loss": 9.379,
      "step": 1691
    },
    {
      "epoch": 0.58364953432218,
      "grad_norm": 1.3861052989959717,
      "learning_rate": 3.928077377762539e-05,
      "loss": 9.4771,
      "step": 1692
    },
    {
      "epoch": 0.5839944808554675,
      "grad_norm": 1.8806065320968628,
      "learning_rate": 3.922596545432903e-05,
      "loss": 9.348,
      "step": 1693
    },
    {
      "epoch": 0.5843394273887548,
      "grad_norm": 1.6042461395263672,
      "learning_rate": 3.91711707039207e-05,
      "loss": 9.4361,
      "step": 1694
    },
    {
      "epoch": 0.5846843739220421,
      "grad_norm": 1.6209865808486938,
      "learning_rate": 3.911638959542959e-05,
      "loss": 9.4388,
      "step": 1695
    },
    {
      "epoch": 0.5850293204553294,
      "grad_norm": 1.7686316967010498,
      "learning_rate": 3.906162219786776e-05,
      "loss": 9.3907,
      "step": 1696
    },
    {
      "epoch": 0.5853742669886167,
      "grad_norm": 1.5446085929870605,
      "learning_rate": 3.9006868580229936e-05,
      "loss": 9.4078,
      "step": 1697
    },
    {
      "epoch": 0.5857192135219041,
      "grad_norm": 1.5874037742614746,
      "learning_rate": 3.895212881149349e-05,
      "loss": 9.3253,
      "step": 1698
    },
    {
      "epoch": 0.5860641600551915,
      "grad_norm": 1.5663777589797974,
      "learning_rate": 3.889740296061836e-05,
      "loss": 9.4409,
      "step": 1699
    },
    {
      "epoch": 0.5864091065884788,
      "grad_norm": 1.5761209726333618,
      "learning_rate": 3.8842691096546965e-05,
      "loss": 9.4123,
      "step": 1700
    },
    {
      "epoch": 0.5867540531217661,
      "grad_norm": 0.9454447031021118,
      "learning_rate": 3.878799328820407e-05,
      "loss": 9.5656,
      "step": 1701
    },
    {
      "epoch": 0.5870989996550535,
      "grad_norm": 1.008521556854248,
      "learning_rate": 3.8733309604496754e-05,
      "loss": 9.5158,
      "step": 1702
    },
    {
      "epoch": 0.5874439461883408,
      "grad_norm": 0.9412047266960144,
      "learning_rate": 3.867864011431432e-05,
      "loss": 9.6026,
      "step": 1703
    },
    {
      "epoch": 0.5877888927216282,
      "grad_norm": 1.052085041999817,
      "learning_rate": 3.862398488652816e-05,
      "loss": 9.4797,
      "step": 1704
    },
    {
      "epoch": 0.5881338392549155,
      "grad_norm": 1.0003331899642944,
      "learning_rate": 3.8569343989991707e-05,
      "loss": 9.599,
      "step": 1705
    },
    {
      "epoch": 0.5884787857882028,
      "grad_norm": 1.1032466888427734,
      "learning_rate": 3.851471749354035e-05,
      "loss": 9.5337,
      "step": 1706
    },
    {
      "epoch": 0.5888237323214902,
      "grad_norm": 1.0652117729187012,
      "learning_rate": 3.8460105465991315e-05,
      "loss": 9.5674,
      "step": 1707
    },
    {
      "epoch": 0.5891686788547775,
      "grad_norm": 1.25304114818573,
      "learning_rate": 3.840550797614363e-05,
      "loss": 9.4834,
      "step": 1708
    },
    {
      "epoch": 0.5895136253880648,
      "grad_norm": 1.2567071914672852,
      "learning_rate": 3.835092509277796e-05,
      "loss": 9.4938,
      "step": 1709
    },
    {
      "epoch": 0.5898585719213522,
      "grad_norm": 1.0688841342926025,
      "learning_rate": 3.8296356884656634e-05,
      "loss": 9.5581,
      "step": 1710
    },
    {
      "epoch": 0.5902035184546396,
      "grad_norm": 1.3249194622039795,
      "learning_rate": 3.824180342052347e-05,
      "loss": 9.468,
      "step": 1711
    },
    {
      "epoch": 0.5905484649879269,
      "grad_norm": 1.0625578165054321,
      "learning_rate": 3.818726476910368e-05,
      "loss": 9.5566,
      "step": 1712
    },
    {
      "epoch": 0.5908934115212142,
      "grad_norm": 1.1471552848815918,
      "learning_rate": 3.813274099910384e-05,
      "loss": 9.5621,
      "step": 1713
    },
    {
      "epoch": 0.5912383580545015,
      "grad_norm": 1.0972129106521606,
      "learning_rate": 3.8078232179211794e-05,
      "loss": 9.4839,
      "step": 1714
    },
    {
      "epoch": 0.5915833045877888,
      "grad_norm": 1.187414526939392,
      "learning_rate": 3.802373837809652e-05,
      "loss": 9.5577,
      "step": 1715
    },
    {
      "epoch": 0.5919282511210763,
      "grad_norm": 1.2144163846969604,
      "learning_rate": 3.7969259664408074e-05,
      "loss": 9.4984,
      "step": 1716
    },
    {
      "epoch": 0.5922731976543636,
      "grad_norm": 1.3942620754241943,
      "learning_rate": 3.791479610677757e-05,
      "loss": 9.4702,
      "step": 1717
    },
    {
      "epoch": 0.5926181441876509,
      "grad_norm": 1.2334468364715576,
      "learning_rate": 3.786034777381695e-05,
      "loss": 9.4928,
      "step": 1718
    },
    {
      "epoch": 0.5929630907209382,
      "grad_norm": 1.1848747730255127,
      "learning_rate": 3.7805914734119025e-05,
      "loss": 9.5711,
      "step": 1719
    },
    {
      "epoch": 0.5933080372542255,
      "grad_norm": 1.3229336738586426,
      "learning_rate": 3.7751497056257304e-05,
      "loss": 9.5629,
      "step": 1720
    },
    {
      "epoch": 0.593652983787513,
      "grad_norm": 1.3299323320388794,
      "learning_rate": 3.769709480878597e-05,
      "loss": 9.4933,
      "step": 1721
    },
    {
      "epoch": 0.5939979303208003,
      "grad_norm": 1.19021737575531,
      "learning_rate": 3.764270806023976e-05,
      "loss": 9.5504,
      "step": 1722
    },
    {
      "epoch": 0.5943428768540876,
      "grad_norm": 1.2337749004364014,
      "learning_rate": 3.7588336879133855e-05,
      "loss": 9.4946,
      "step": 1723
    },
    {
      "epoch": 0.5946878233873749,
      "grad_norm": 1.1160210371017456,
      "learning_rate": 3.753398133396391e-05,
      "loss": 9.5584,
      "step": 1724
    },
    {
      "epoch": 0.5950327699206623,
      "grad_norm": 1.3497905731201172,
      "learning_rate": 3.7479641493205796e-05,
      "loss": 9.5159,
      "step": 1725
    },
    {
      "epoch": 0.5953777164539497,
      "grad_norm": 1.304688811302185,
      "learning_rate": 3.742531742531562e-05,
      "loss": 9.491,
      "step": 1726
    },
    {
      "epoch": 0.595722662987237,
      "grad_norm": 1.2296416759490967,
      "learning_rate": 3.7371009198729654e-05,
      "loss": 9.584,
      "step": 1727
    },
    {
      "epoch": 0.5960676095205243,
      "grad_norm": 1.2551003694534302,
      "learning_rate": 3.731671688186418e-05,
      "loss": 9.5121,
      "step": 1728
    },
    {
      "epoch": 0.5964125560538116,
      "grad_norm": 1.3699496984481812,
      "learning_rate": 3.7262440543115446e-05,
      "loss": 9.4163,
      "step": 1729
    },
    {
      "epoch": 0.596757502587099,
      "grad_norm": 1.4183300733566284,
      "learning_rate": 3.720818025085954e-05,
      "loss": 9.5081,
      "step": 1730
    },
    {
      "epoch": 0.5971024491203863,
      "grad_norm": 1.3562133312225342,
      "learning_rate": 3.715393607345242e-05,
      "loss": 9.4844,
      "step": 1731
    },
    {
      "epoch": 0.5974473956536737,
      "grad_norm": 1.2870327234268188,
      "learning_rate": 3.7099708079229654e-05,
      "loss": 9.5323,
      "step": 1732
    },
    {
      "epoch": 0.597792342186961,
      "grad_norm": 1.2621005773544312,
      "learning_rate": 3.704549633650648e-05,
      "loss": 9.5163,
      "step": 1733
    },
    {
      "epoch": 0.5981372887202484,
      "grad_norm": 1.3484461307525635,
      "learning_rate": 3.699130091357762e-05,
      "loss": 9.4726,
      "step": 1734
    },
    {
      "epoch": 0.5984822352535357,
      "grad_norm": 1.372406244277954,
      "learning_rate": 3.693712187871725e-05,
      "loss": 9.4711,
      "step": 1735
    },
    {
      "epoch": 0.598827181786823,
      "grad_norm": 1.4271361827850342,
      "learning_rate": 3.6882959300178936e-05,
      "loss": 9.5137,
      "step": 1736
    },
    {
      "epoch": 0.5991721283201104,
      "grad_norm": 1.3963810205459595,
      "learning_rate": 3.682881324619546e-05,
      "loss": 9.5354,
      "step": 1737
    },
    {
      "epoch": 0.5995170748533977,
      "grad_norm": 1.3229588270187378,
      "learning_rate": 3.6774683784978825e-05,
      "loss": 9.475,
      "step": 1738
    },
    {
      "epoch": 0.5998620213866851,
      "grad_norm": 1.5664173364639282,
      "learning_rate": 3.672057098472012e-05,
      "loss": 9.4822,
      "step": 1739
    },
    {
      "epoch": 0.6002069679199724,
      "grad_norm": 1.4833637475967407,
      "learning_rate": 3.666647491358943e-05,
      "loss": 9.4125,
      "step": 1740
    },
    {
      "epoch": 0.6005519144532597,
      "grad_norm": 1.571965217590332,
      "learning_rate": 3.66123956397358e-05,
      "loss": 9.4406,
      "step": 1741
    },
    {
      "epoch": 0.600896860986547,
      "grad_norm": 1.3770232200622559,
      "learning_rate": 3.655833323128706e-05,
      "loss": 9.5445,
      "step": 1742
    },
    {
      "epoch": 0.6012418075198345,
      "grad_norm": 1.4384090900421143,
      "learning_rate": 3.6504287756349844e-05,
      "loss": 9.4861,
      "step": 1743
    },
    {
      "epoch": 0.6015867540531218,
      "grad_norm": 1.48284113407135,
      "learning_rate": 3.6450259283009395e-05,
      "loss": 9.5147,
      "step": 1744
    },
    {
      "epoch": 0.6019317005864091,
      "grad_norm": 1.5759750604629517,
      "learning_rate": 3.639624787932964e-05,
      "loss": 9.4079,
      "step": 1745
    },
    {
      "epoch": 0.6022766471196964,
      "grad_norm": 1.4510525465011597,
      "learning_rate": 3.634225361335291e-05,
      "loss": 9.4536,
      "step": 1746
    },
    {
      "epoch": 0.6026215936529837,
      "grad_norm": 1.5437071323394775,
      "learning_rate": 3.6288276553099976e-05,
      "loss": 9.4269,
      "step": 1747
    },
    {
      "epoch": 0.6029665401862712,
      "grad_norm": 1.4966944456100464,
      "learning_rate": 3.6234316766569917e-05,
      "loss": 9.3802,
      "step": 1748
    },
    {
      "epoch": 0.6033114867195585,
      "grad_norm": 1.8180314302444458,
      "learning_rate": 3.6180374321740084e-05,
      "loss": 9.3866,
      "step": 1749
    },
    {
      "epoch": 0.6036564332528458,
      "grad_norm": 1.7780247926712036,
      "learning_rate": 3.6126449286565965e-05,
      "loss": 9.3133,
      "step": 1750
    },
    {
      "epoch": 0.6040013797861331,
      "grad_norm": 0.9698929786682129,
      "learning_rate": 3.6072541728981094e-05,
      "loss": 9.5623,
      "step": 1751
    },
    {
      "epoch": 0.6043463263194205,
      "grad_norm": 0.9983956813812256,
      "learning_rate": 3.601865171689704e-05,
      "loss": 9.5587,
      "step": 1752
    },
    {
      "epoch": 0.6046912728527079,
      "grad_norm": 1.0467181205749512,
      "learning_rate": 3.596477931820322e-05,
      "loss": 9.5551,
      "step": 1753
    },
    {
      "epoch": 0.6050362193859952,
      "grad_norm": 1.2166588306427002,
      "learning_rate": 3.591092460076688e-05,
      "loss": 9.5129,
      "step": 1754
    },
    {
      "epoch": 0.6053811659192825,
      "grad_norm": 0.9980295896530151,
      "learning_rate": 3.5857087632432986e-05,
      "loss": 9.5484,
      "step": 1755
    },
    {
      "epoch": 0.6057261124525698,
      "grad_norm": 1.0149705410003662,
      "learning_rate": 3.580326848102415e-05,
      "loss": 9.5623,
      "step": 1756
    },
    {
      "epoch": 0.6060710589858572,
      "grad_norm": 1.147386908531189,
      "learning_rate": 3.5749467214340525e-05,
      "loss": 9.5167,
      "step": 1757
    },
    {
      "epoch": 0.6064160055191445,
      "grad_norm": 1.0991644859313965,
      "learning_rate": 3.569568390015976e-05,
      "loss": 9.5704,
      "step": 1758
    },
    {
      "epoch": 0.6067609520524319,
      "grad_norm": 1.0665550231933594,
      "learning_rate": 3.564191860623688e-05,
      "loss": 9.535,
      "step": 1759
    },
    {
      "epoch": 0.6071058985857192,
      "grad_norm": 1.2261543273925781,
      "learning_rate": 3.558817140030418e-05,
      "loss": 9.5013,
      "step": 1760
    },
    {
      "epoch": 0.6074508451190066,
      "grad_norm": 1.1674590110778809,
      "learning_rate": 3.553444235007121e-05,
      "loss": 9.5437,
      "step": 1761
    },
    {
      "epoch": 0.6077957916522939,
      "grad_norm": 1.214137077331543,
      "learning_rate": 3.548073152322462e-05,
      "loss": 9.5122,
      "step": 1762
    },
    {
      "epoch": 0.6081407381855812,
      "grad_norm": 1.1316503286361694,
      "learning_rate": 3.5427038987428105e-05,
      "loss": 9.5394,
      "step": 1763
    },
    {
      "epoch": 0.6084856847188685,
      "grad_norm": 1.1909269094467163,
      "learning_rate": 3.537336481032232e-05,
      "loss": 9.5682,
      "step": 1764
    },
    {
      "epoch": 0.6088306312521559,
      "grad_norm": 1.2326979637145996,
      "learning_rate": 3.531970905952478e-05,
      "loss": 9.4657,
      "step": 1765
    },
    {
      "epoch": 0.6091755777854433,
      "grad_norm": 1.2010674476623535,
      "learning_rate": 3.526607180262984e-05,
      "loss": 9.5197,
      "step": 1766
    },
    {
      "epoch": 0.6095205243187306,
      "grad_norm": 1.1736875772476196,
      "learning_rate": 3.5212453107208506e-05,
      "loss": 9.4788,
      "step": 1767
    },
    {
      "epoch": 0.6098654708520179,
      "grad_norm": 1.167720079421997,
      "learning_rate": 3.515885304080839e-05,
      "loss": 9.5067,
      "step": 1768
    },
    {
      "epoch": 0.6102104173853052,
      "grad_norm": 1.3078930377960205,
      "learning_rate": 3.5105271670953674e-05,
      "loss": 9.4946,
      "step": 1769
    },
    {
      "epoch": 0.6105553639185927,
      "grad_norm": 1.2530713081359863,
      "learning_rate": 3.505170906514498e-05,
      "loss": 9.535,
      "step": 1770
    },
    {
      "epoch": 0.61090031045188,
      "grad_norm": 1.2267308235168457,
      "learning_rate": 3.4998165290859266e-05,
      "loss": 9.4742,
      "step": 1771
    },
    {
      "epoch": 0.6112452569851673,
      "grad_norm": 1.2809817790985107,
      "learning_rate": 3.494464041554977e-05,
      "loss": 9.5116,
      "step": 1772
    },
    {
      "epoch": 0.6115902035184546,
      "grad_norm": 1.1434053182601929,
      "learning_rate": 3.489113450664597e-05,
      "loss": 9.5531,
      "step": 1773
    },
    {
      "epoch": 0.6119351500517419,
      "grad_norm": 1.3463351726531982,
      "learning_rate": 3.4837647631553405e-05,
      "loss": 9.5608,
      "step": 1774
    },
    {
      "epoch": 0.6122800965850294,
      "grad_norm": 1.5135565996170044,
      "learning_rate": 3.478417985765363e-05,
      "loss": 9.4173,
      "step": 1775
    },
    {
      "epoch": 0.6126250431183167,
      "grad_norm": 1.2766389846801758,
      "learning_rate": 3.473073125230417e-05,
      "loss": 9.5591,
      "step": 1776
    },
    {
      "epoch": 0.612969989651604,
      "grad_norm": 1.29940664768219,
      "learning_rate": 3.467730188283836e-05,
      "loss": 9.5021,
      "step": 1777
    },
    {
      "epoch": 0.6133149361848913,
      "grad_norm": 1.322685956954956,
      "learning_rate": 3.462389181656535e-05,
      "loss": 9.4625,
      "step": 1778
    },
    {
      "epoch": 0.6136598827181787,
      "grad_norm": 1.272139310836792,
      "learning_rate": 3.457050112076992e-05,
      "loss": 9.4416,
      "step": 1779
    },
    {
      "epoch": 0.614004829251466,
      "grad_norm": 1.449051856994629,
      "learning_rate": 3.4517129862712504e-05,
      "loss": 9.443,
      "step": 1780
    },
    {
      "epoch": 0.6143497757847534,
      "grad_norm": 1.3246345520019531,
      "learning_rate": 3.446377810962902e-05,
      "loss": 9.4954,
      "step": 1781
    },
    {
      "epoch": 0.6146947223180407,
      "grad_norm": 1.449554443359375,
      "learning_rate": 3.441044592873079e-05,
      "loss": 9.4129,
      "step": 1782
    },
    {
      "epoch": 0.615039668851328,
      "grad_norm": 1.2014191150665283,
      "learning_rate": 3.435713338720453e-05,
      "loss": 9.5974,
      "step": 1783
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 1.471830129623413,
      "learning_rate": 3.4303840552212184e-05,
      "loss": 9.4773,
      "step": 1784
    },
    {
      "epoch": 0.6157295619179027,
      "grad_norm": 1.3256584405899048,
      "learning_rate": 3.425056749089086e-05,
      "loss": 9.5171,
      "step": 1785
    },
    {
      "epoch": 0.61607450845119,
      "grad_norm": 1.4362033605575562,
      "learning_rate": 3.419731427035277e-05,
      "loss": 9.4557,
      "step": 1786
    },
    {
      "epoch": 0.6164194549844774,
      "grad_norm": 1.4328278303146362,
      "learning_rate": 3.414408095768516e-05,
      "loss": 9.4515,
      "step": 1787
    },
    {
      "epoch": 0.6167644015177648,
      "grad_norm": 1.360702395439148,
      "learning_rate": 3.409086761995015e-05,
      "loss": 9.4638,
      "step": 1788
    },
    {
      "epoch": 0.6171093480510521,
      "grad_norm": 1.3730055093765259,
      "learning_rate": 3.4037674324184706e-05,
      "loss": 9.4274,
      "step": 1789
    },
    {
      "epoch": 0.6174542945843394,
      "grad_norm": 1.568692684173584,
      "learning_rate": 3.398450113740057e-05,
      "loss": 9.4313,
      "step": 1790
    },
    {
      "epoch": 0.6177992411176267,
      "grad_norm": 1.3866729736328125,
      "learning_rate": 3.393134812658411e-05,
      "loss": 9.4285,
      "step": 1791
    },
    {
      "epoch": 0.6181441876509141,
      "grad_norm": 1.4505038261413574,
      "learning_rate": 3.3878215358696333e-05,
      "loss": 9.4724,
      "step": 1792
    },
    {
      "epoch": 0.6184891341842015,
      "grad_norm": 1.435584306716919,
      "learning_rate": 3.382510290067269e-05,
      "loss": 9.4448,
      "step": 1793
    },
    {
      "epoch": 0.6188340807174888,
      "grad_norm": 1.4031693935394287,
      "learning_rate": 3.377201081942307e-05,
      "loss": 9.4456,
      "step": 1794
    },
    {
      "epoch": 0.6191790272507761,
      "grad_norm": 1.3541998863220215,
      "learning_rate": 3.371893918183171e-05,
      "loss": 9.4522,
      "step": 1795
    },
    {
      "epoch": 0.6195239737840634,
      "grad_norm": 1.6531020402908325,
      "learning_rate": 3.366588805475707e-05,
      "loss": 9.3786,
      "step": 1796
    },
    {
      "epoch": 0.6198689203173509,
      "grad_norm": 1.5427093505859375,
      "learning_rate": 3.361285750503176e-05,
      "loss": 9.4252,
      "step": 1797
    },
    {
      "epoch": 0.6202138668506382,
      "grad_norm": 1.6081955432891846,
      "learning_rate": 3.355984759946249e-05,
      "loss": 9.4544,
      "step": 1798
    },
    {
      "epoch": 0.6205588133839255,
      "grad_norm": 1.565117359161377,
      "learning_rate": 3.350685840482995e-05,
      "loss": 9.3773,
      "step": 1799
    },
    {
      "epoch": 0.6209037599172128,
      "grad_norm": 1.8025498390197754,
      "learning_rate": 3.3453889987888724e-05,
      "loss": 9.361,
      "step": 1800
    },
    {
      "epoch": 0.6212487064505001,
      "grad_norm": 0.9393291473388672,
      "learning_rate": 3.340094241536729e-05,
      "loss": 9.5986,
      "step": 1801
    },
    {
      "epoch": 0.6215936529837875,
      "grad_norm": 0.9784131050109863,
      "learning_rate": 3.3348015753967785e-05,
      "loss": 9.5321,
      "step": 1802
    },
    {
      "epoch": 0.6219385995170749,
      "grad_norm": 1.0297746658325195,
      "learning_rate": 3.329511007036604e-05,
      "loss": 9.5718,
      "step": 1803
    },
    {
      "epoch": 0.6222835460503622,
      "grad_norm": 1.0639649629592896,
      "learning_rate": 3.324222543121145e-05,
      "loss": 9.537,
      "step": 1804
    },
    {
      "epoch": 0.6226284925836495,
      "grad_norm": 1.1060718297958374,
      "learning_rate": 3.3189361903126916e-05,
      "loss": 9.5329,
      "step": 1805
    },
    {
      "epoch": 0.6229734391169369,
      "grad_norm": 1.1022459268569946,
      "learning_rate": 3.313651955270871e-05,
      "loss": 9.513,
      "step": 1806
    },
    {
      "epoch": 0.6233183856502242,
      "grad_norm": 1.0694283246994019,
      "learning_rate": 3.308369844652643e-05,
      "loss": 9.5594,
      "step": 1807
    },
    {
      "epoch": 0.6236633321835116,
      "grad_norm": 1.255054235458374,
      "learning_rate": 3.3030898651122985e-05,
      "loss": 9.4931,
      "step": 1808
    },
    {
      "epoch": 0.6240082787167989,
      "grad_norm": 1.1304607391357422,
      "learning_rate": 3.297812023301433e-05,
      "loss": 9.546,
      "step": 1809
    },
    {
      "epoch": 0.6243532252500862,
      "grad_norm": 1.214964747428894,
      "learning_rate": 3.2925363258689555e-05,
      "loss": 9.5152,
      "step": 1810
    },
    {
      "epoch": 0.6246981717833736,
      "grad_norm": 1.1067923307418823,
      "learning_rate": 3.287262779461071e-05,
      "loss": 9.5478,
      "step": 1811
    },
    {
      "epoch": 0.6250431183166609,
      "grad_norm": 1.1942600011825562,
      "learning_rate": 3.281991390721276e-05,
      "loss": 9.5681,
      "step": 1812
    },
    {
      "epoch": 0.6253880648499482,
      "grad_norm": 1.2584760189056396,
      "learning_rate": 3.276722166290346e-05,
      "loss": 9.4977,
      "step": 1813
    },
    {
      "epoch": 0.6257330113832356,
      "grad_norm": 1.0992047786712646,
      "learning_rate": 3.2714551128063344e-05,
      "loss": 9.5753,
      "step": 1814
    },
    {
      "epoch": 0.626077957916523,
      "grad_norm": 1.187728762626648,
      "learning_rate": 3.2661902369045586e-05,
      "loss": 9.5477,
      "step": 1815
    },
    {
      "epoch": 0.6264229044498103,
      "grad_norm": 1.3243029117584229,
      "learning_rate": 3.260927545217589e-05,
      "loss": 9.4887,
      "step": 1816
    },
    {
      "epoch": 0.6267678509830976,
      "grad_norm": 1.2621115446090698,
      "learning_rate": 3.25566704437525e-05,
      "loss": 9.4954,
      "step": 1817
    },
    {
      "epoch": 0.6271127975163849,
      "grad_norm": 1.326181173324585,
      "learning_rate": 3.2504087410046004e-05,
      "loss": 9.4583,
      "step": 1818
    },
    {
      "epoch": 0.6274577440496723,
      "grad_norm": 1.183527946472168,
      "learning_rate": 3.245152641729935e-05,
      "loss": 9.559,
      "step": 1819
    },
    {
      "epoch": 0.6278026905829597,
      "grad_norm": 1.1669021844863892,
      "learning_rate": 3.2398987531727694e-05,
      "loss": 9.4858,
      "step": 1820
    },
    {
      "epoch": 0.628147637116247,
      "grad_norm": 1.3547872304916382,
      "learning_rate": 3.2346470819518326e-05,
      "loss": 9.4709,
      "step": 1821
    },
    {
      "epoch": 0.6284925836495343,
      "grad_norm": 1.2788245677947998,
      "learning_rate": 3.229397634683068e-05,
      "loss": 9.4746,
      "step": 1822
    },
    {
      "epoch": 0.6288375301828216,
      "grad_norm": 1.2765026092529297,
      "learning_rate": 3.224150417979609e-05,
      "loss": 9.5076,
      "step": 1823
    },
    {
      "epoch": 0.629182476716109,
      "grad_norm": 1.3237452507019043,
      "learning_rate": 3.218905438451782e-05,
      "loss": 9.5053,
      "step": 1824
    },
    {
      "epoch": 0.6295274232493964,
      "grad_norm": 1.1867791414260864,
      "learning_rate": 3.213662702707094e-05,
      "loss": 9.4948,
      "step": 1825
    },
    {
      "epoch": 0.6298723697826837,
      "grad_norm": 1.2228574752807617,
      "learning_rate": 3.20842221735023e-05,
      "loss": 9.5036,
      "step": 1826
    },
    {
      "epoch": 0.630217316315971,
      "grad_norm": 1.3094220161437988,
      "learning_rate": 3.203183988983033e-05,
      "loss": 9.4877,
      "step": 1827
    },
    {
      "epoch": 0.6305622628492583,
      "grad_norm": 1.2245163917541504,
      "learning_rate": 3.1979480242045045e-05,
      "loss": 9.5359,
      "step": 1828
    },
    {
      "epoch": 0.6309072093825457,
      "grad_norm": 1.34917151927948,
      "learning_rate": 3.192714329610802e-05,
      "loss": 9.5315,
      "step": 1829
    },
    {
      "epoch": 0.6312521559158331,
      "grad_norm": 1.3427605628967285,
      "learning_rate": 3.1874829117952124e-05,
      "loss": 9.4754,
      "step": 1830
    },
    {
      "epoch": 0.6315971024491204,
      "grad_norm": 1.4418407678604126,
      "learning_rate": 3.182253777348161e-05,
      "loss": 9.4585,
      "step": 1831
    },
    {
      "epoch": 0.6319420489824077,
      "grad_norm": 1.387801170349121,
      "learning_rate": 3.177026932857193e-05,
      "loss": 9.4647,
      "step": 1832
    },
    {
      "epoch": 0.6322869955156951,
      "grad_norm": 1.41670823097229,
      "learning_rate": 3.171802384906972e-05,
      "loss": 9.5347,
      "step": 1833
    },
    {
      "epoch": 0.6326319420489824,
      "grad_norm": 1.417445182800293,
      "learning_rate": 3.166580140079265e-05,
      "loss": 9.5112,
      "step": 1834
    },
    {
      "epoch": 0.6329768885822697,
      "grad_norm": 1.3484708070755005,
      "learning_rate": 3.161360204952939e-05,
      "loss": 9.5273,
      "step": 1835
    },
    {
      "epoch": 0.6333218351155571,
      "grad_norm": 1.4183062314987183,
      "learning_rate": 3.1561425861039546e-05,
      "loss": 9.4721,
      "step": 1836
    },
    {
      "epoch": 0.6336667816488444,
      "grad_norm": 1.4750018119812012,
      "learning_rate": 3.15092729010535e-05,
      "loss": 9.4265,
      "step": 1837
    },
    {
      "epoch": 0.6340117281821318,
      "grad_norm": 1.3089995384216309,
      "learning_rate": 3.145714323527237e-05,
      "loss": 9.438,
      "step": 1838
    },
    {
      "epoch": 0.6343566747154191,
      "grad_norm": 1.4804081916809082,
      "learning_rate": 3.140503692936797e-05,
      "loss": 9.4808,
      "step": 1839
    },
    {
      "epoch": 0.6347016212487064,
      "grad_norm": 1.3244197368621826,
      "learning_rate": 3.135295404898265e-05,
      "loss": 9.4566,
      "step": 1840
    },
    {
      "epoch": 0.6350465677819938,
      "grad_norm": 1.364261269569397,
      "learning_rate": 3.130089465972926e-05,
      "loss": 9.4726,
      "step": 1841
    },
    {
      "epoch": 0.6353915143152812,
      "grad_norm": 1.5104689598083496,
      "learning_rate": 3.124885882719102e-05,
      "loss": 9.4577,
      "step": 1842
    },
    {
      "epoch": 0.6357364608485685,
      "grad_norm": 1.30232572555542,
      "learning_rate": 3.119684661692158e-05,
      "loss": 9.4958,
      "step": 1843
    },
    {
      "epoch": 0.6360814073818558,
      "grad_norm": 1.457231879234314,
      "learning_rate": 3.1144858094444715e-05,
      "loss": 9.5086,
      "step": 1844
    },
    {
      "epoch": 0.6364263539151431,
      "grad_norm": 1.4896906614303589,
      "learning_rate": 3.1092893325254413e-05,
      "loss": 9.4611,
      "step": 1845
    },
    {
      "epoch": 0.6367713004484304,
      "grad_norm": 1.5810197591781616,
      "learning_rate": 3.104095237481473e-05,
      "loss": 9.4457,
      "step": 1846
    },
    {
      "epoch": 0.6371162469817179,
      "grad_norm": 1.4873422384262085,
      "learning_rate": 3.0989035308559696e-05,
      "loss": 9.4359,
      "step": 1847
    },
    {
      "epoch": 0.6374611935150052,
      "grad_norm": 1.7242629528045654,
      "learning_rate": 3.093714219189331e-05,
      "loss": 9.3872,
      "step": 1848
    },
    {
      "epoch": 0.6378061400482925,
      "grad_norm": 1.7722023725509644,
      "learning_rate": 3.0885273090189324e-05,
      "loss": 9.3504,
      "step": 1849
    },
    {
      "epoch": 0.6381510865815798,
      "grad_norm": 1.7719429731369019,
      "learning_rate": 3.083342806879129e-05,
      "loss": 9.3643,
      "step": 1850
    },
    {
      "epoch": 0.6384960331148672,
      "grad_norm": 0.9851622581481934,
      "learning_rate": 3.078160719301242e-05,
      "loss": 9.5158,
      "step": 1851
    },
    {
      "epoch": 0.6388409796481546,
      "grad_norm": 1.0696388483047485,
      "learning_rate": 3.0729810528135484e-05,
      "loss": 9.4771,
      "step": 1852
    },
    {
      "epoch": 0.6391859261814419,
      "grad_norm": 1.0862210988998413,
      "learning_rate": 3.067803813941276e-05,
      "loss": 9.535,
      "step": 1853
    },
    {
      "epoch": 0.6395308727147292,
      "grad_norm": 0.9991084933280945,
      "learning_rate": 3.062629009206595e-05,
      "loss": 9.5883,
      "step": 1854
    },
    {
      "epoch": 0.6398758192480165,
      "grad_norm": 1.1714235544204712,
      "learning_rate": 3.0574566451286094e-05,
      "loss": 9.527,
      "step": 1855
    },
    {
      "epoch": 0.6402207657813039,
      "grad_norm": 1.2183157205581665,
      "learning_rate": 3.052286728223343e-05,
      "loss": 9.4489,
      "step": 1856
    },
    {
      "epoch": 0.6405657123145913,
      "grad_norm": 1.1056588888168335,
      "learning_rate": 3.0471192650037504e-05,
      "loss": 9.5884,
      "step": 1857
    },
    {
      "epoch": 0.6409106588478786,
      "grad_norm": 1.1708348989486694,
      "learning_rate": 3.041954261979681e-05,
      "loss": 9.5137,
      "step": 1858
    },
    {
      "epoch": 0.6412556053811659,
      "grad_norm": 1.127313494682312,
      "learning_rate": 3.0367917256578908e-05,
      "loss": 9.5326,
      "step": 1859
    },
    {
      "epoch": 0.6416005519144533,
      "grad_norm": 1.2033101320266724,
      "learning_rate": 3.0316316625420272e-05,
      "loss": 9.4886,
      "step": 1860
    },
    {
      "epoch": 0.6419454984477406,
      "grad_norm": 1.183809757232666,
      "learning_rate": 3.026474079132624e-05,
      "loss": 9.4902,
      "step": 1861
    },
    {
      "epoch": 0.6422904449810279,
      "grad_norm": 1.071576714515686,
      "learning_rate": 3.0213189819270894e-05,
      "loss": 9.5036,
      "step": 1862
    },
    {
      "epoch": 0.6426353915143153,
      "grad_norm": 1.2379518747329712,
      "learning_rate": 3.0161663774196962e-05,
      "loss": 9.4938,
      "step": 1863
    },
    {
      "epoch": 0.6429803380476026,
      "grad_norm": 1.231271505355835,
      "learning_rate": 3.0110162721015856e-05,
      "loss": 9.5153,
      "step": 1864
    },
    {
      "epoch": 0.64332528458089,
      "grad_norm": 1.246327519416809,
      "learning_rate": 3.0058686724607432e-05,
      "loss": 9.5067,
      "step": 1865
    },
    {
      "epoch": 0.6436702311141773,
      "grad_norm": 1.328060507774353,
      "learning_rate": 3.000723584982e-05,
      "loss": 9.452,
      "step": 1866
    },
    {
      "epoch": 0.6440151776474646,
      "grad_norm": 1.3291081190109253,
      "learning_rate": 2.995581016147021e-05,
      "loss": 9.4638,
      "step": 1867
    },
    {
      "epoch": 0.644360124180752,
      "grad_norm": 1.3064149618148804,
      "learning_rate": 2.990440972434302e-05,
      "loss": 9.5566,
      "step": 1868
    },
    {
      "epoch": 0.6447050707140394,
      "grad_norm": 1.1765854358673096,
      "learning_rate": 2.985303460319152e-05,
      "loss": 9.5289,
      "step": 1869
    },
    {
      "epoch": 0.6450500172473267,
      "grad_norm": 1.3379801511764526,
      "learning_rate": 2.9801684862736958e-05,
      "loss": 9.5201,
      "step": 1870
    },
    {
      "epoch": 0.645394963780614,
      "grad_norm": 1.2638524770736694,
      "learning_rate": 2.97503605676686e-05,
      "loss": 9.5207,
      "step": 1871
    },
    {
      "epoch": 0.6457399103139013,
      "grad_norm": 1.3365275859832764,
      "learning_rate": 2.9699061782643632e-05,
      "loss": 9.4655,
      "step": 1872
    },
    {
      "epoch": 0.6460848568471886,
      "grad_norm": 1.2336562871932983,
      "learning_rate": 2.9647788572287126e-05,
      "loss": 9.5254,
      "step": 1873
    },
    {
      "epoch": 0.6464298033804761,
      "grad_norm": 1.1859928369522095,
      "learning_rate": 2.9596541001191924e-05,
      "loss": 9.461,
      "step": 1874
    },
    {
      "epoch": 0.6467747499137634,
      "grad_norm": 1.2857304811477661,
      "learning_rate": 2.954531913391857e-05,
      "loss": 9.4632,
      "step": 1875
    },
    {
      "epoch": 0.6471196964470507,
      "grad_norm": 1.1990352869033813,
      "learning_rate": 2.9494123034995236e-05,
      "loss": 9.5014,
      "step": 1876
    },
    {
      "epoch": 0.647464642980338,
      "grad_norm": 1.325065016746521,
      "learning_rate": 2.9442952768917588e-05,
      "loss": 9.5199,
      "step": 1877
    },
    {
      "epoch": 0.6478095895136254,
      "grad_norm": 1.4335227012634277,
      "learning_rate": 2.9391808400148834e-05,
      "loss": 9.4413,
      "step": 1878
    },
    {
      "epoch": 0.6481545360469128,
      "grad_norm": 1.347676157951355,
      "learning_rate": 2.9340689993119485e-05,
      "loss": 9.5028,
      "step": 1879
    },
    {
      "epoch": 0.6484994825802001,
      "grad_norm": 1.3796898126602173,
      "learning_rate": 2.928959761222737e-05,
      "loss": 9.4798,
      "step": 1880
    },
    {
      "epoch": 0.6488444291134874,
      "grad_norm": 1.427625298500061,
      "learning_rate": 2.9238531321837513e-05,
      "loss": 9.5211,
      "step": 1881
    },
    {
      "epoch": 0.6491893756467747,
      "grad_norm": 1.3672127723693848,
      "learning_rate": 2.918749118628213e-05,
      "loss": 9.5062,
      "step": 1882
    },
    {
      "epoch": 0.6495343221800621,
      "grad_norm": 1.348329782485962,
      "learning_rate": 2.9136477269860386e-05,
      "loss": 9.4853,
      "step": 1883
    },
    {
      "epoch": 0.6498792687133494,
      "grad_norm": 1.4505221843719482,
      "learning_rate": 2.9085489636838504e-05,
      "loss": 9.4153,
      "step": 1884
    },
    {
      "epoch": 0.6502242152466368,
      "grad_norm": 1.262218713760376,
      "learning_rate": 2.9034528351449563e-05,
      "loss": 9.4583,
      "step": 1885
    },
    {
      "epoch": 0.6505691617799241,
      "grad_norm": 1.4502121210098267,
      "learning_rate": 2.8983593477893474e-05,
      "loss": 9.4859,
      "step": 1886
    },
    {
      "epoch": 0.6509141083132115,
      "grad_norm": 1.3023953437805176,
      "learning_rate": 2.8932685080336807e-05,
      "loss": 9.498,
      "step": 1887
    },
    {
      "epoch": 0.6512590548464988,
      "grad_norm": 1.2097197771072388,
      "learning_rate": 2.8881803222912844e-05,
      "loss": 9.5218,
      "step": 1888
    },
    {
      "epoch": 0.6516040013797861,
      "grad_norm": 1.571702003479004,
      "learning_rate": 2.8830947969721445e-05,
      "loss": 9.4553,
      "step": 1889
    },
    {
      "epoch": 0.6519489479130735,
      "grad_norm": 1.5287089347839355,
      "learning_rate": 2.878011938482886e-05,
      "loss": 9.3639,
      "step": 1890
    },
    {
      "epoch": 0.6522938944463608,
      "grad_norm": 1.3860400915145874,
      "learning_rate": 2.8729317532267846e-05,
      "loss": 9.5053,
      "step": 1891
    },
    {
      "epoch": 0.6526388409796482,
      "grad_norm": 1.4050540924072266,
      "learning_rate": 2.8678542476037427e-05,
      "loss": 9.4575,
      "step": 1892
    },
    {
      "epoch": 0.6529837875129355,
      "grad_norm": 1.5032589435577393,
      "learning_rate": 2.862779428010292e-05,
      "loss": 9.3963,
      "step": 1893
    },
    {
      "epoch": 0.6533287340462228,
      "grad_norm": 1.5438631772994995,
      "learning_rate": 2.8577073008395717e-05,
      "loss": 9.4169,
      "step": 1894
    },
    {
      "epoch": 0.6536736805795101,
      "grad_norm": 1.5366860628128052,
      "learning_rate": 2.852637872481338e-05,
      "loss": 9.3983,
      "step": 1895
    },
    {
      "epoch": 0.6540186271127976,
      "grad_norm": 1.5280556678771973,
      "learning_rate": 2.8475711493219404e-05,
      "loss": 9.4123,
      "step": 1896
    },
    {
      "epoch": 0.6543635736460849,
      "grad_norm": 1.4505887031555176,
      "learning_rate": 2.8425071377443245e-05,
      "loss": 9.473,
      "step": 1897
    },
    {
      "epoch": 0.6547085201793722,
      "grad_norm": 1.694767713546753,
      "learning_rate": 2.8374458441280204e-05,
      "loss": 9.3827,
      "step": 1898
    },
    {
      "epoch": 0.6550534667126595,
      "grad_norm": 1.6835027933120728,
      "learning_rate": 2.832387274849134e-05,
      "loss": 9.4103,
      "step": 1899
    },
    {
      "epoch": 0.6553984132459468,
      "grad_norm": 1.7232531309127808,
      "learning_rate": 2.8273314362803338e-05,
      "loss": 9.3244,
      "step": 1900
    },
    {
      "epoch": 0.6557433597792343,
      "grad_norm": 0.9157406091690063,
      "learning_rate": 2.8222783347908545e-05,
      "loss": 9.5555,
      "step": 1901
    },
    {
      "epoch": 0.6560883063125216,
      "grad_norm": 1.0032362937927246,
      "learning_rate": 2.8172279767464814e-05,
      "loss": 9.6001,
      "step": 1902
    },
    {
      "epoch": 0.6564332528458089,
      "grad_norm": 1.1082993745803833,
      "learning_rate": 2.8121803685095403e-05,
      "loss": 9.5102,
      "step": 1903
    },
    {
      "epoch": 0.6567781993790962,
      "grad_norm": 0.9890560507774353,
      "learning_rate": 2.807135516438899e-05,
      "loss": 9.5794,
      "step": 1904
    },
    {
      "epoch": 0.6571231459123836,
      "grad_norm": 1.0277185440063477,
      "learning_rate": 2.8020934268899423e-05,
      "loss": 9.5428,
      "step": 1905
    },
    {
      "epoch": 0.657468092445671,
      "grad_norm": 1.018535852432251,
      "learning_rate": 2.7970541062145918e-05,
      "loss": 9.5505,
      "step": 1906
    },
    {
      "epoch": 0.6578130389789583,
      "grad_norm": 1.0695736408233643,
      "learning_rate": 2.792017560761263e-05,
      "loss": 9.5001,
      "step": 1907
    },
    {
      "epoch": 0.6581579855122456,
      "grad_norm": 1.1729811429977417,
      "learning_rate": 2.786983796874889e-05,
      "loss": 9.4728,
      "step": 1908
    },
    {
      "epoch": 0.6585029320455329,
      "grad_norm": 1.1934943199157715,
      "learning_rate": 2.7819528208968883e-05,
      "loss": 9.4998,
      "step": 1909
    },
    {
      "epoch": 0.6588478785788203,
      "grad_norm": 1.0921857357025146,
      "learning_rate": 2.7769246391651742e-05,
      "loss": 9.5279,
      "step": 1910
    },
    {
      "epoch": 0.6591928251121076,
      "grad_norm": 1.2350205183029175,
      "learning_rate": 2.7718992580141402e-05,
      "loss": 9.4769,
      "step": 1911
    },
    {
      "epoch": 0.659537771645395,
      "grad_norm": 1.2308374643325806,
      "learning_rate": 2.7668766837746422e-05,
      "loss": 9.5153,
      "step": 1912
    },
    {
      "epoch": 0.6598827181786823,
      "grad_norm": 1.287536859512329,
      "learning_rate": 2.7618569227740165e-05,
      "loss": 9.4465,
      "step": 1913
    },
    {
      "epoch": 0.6602276647119697,
      "grad_norm": 1.2224645614624023,
      "learning_rate": 2.7568399813360378e-05,
      "loss": 9.5243,
      "step": 1914
    },
    {
      "epoch": 0.660572611245257,
      "grad_norm": 1.2047133445739746,
      "learning_rate": 2.751825865780943e-05,
      "loss": 9.4819,
      "step": 1915
    },
    {
      "epoch": 0.6609175577785443,
      "grad_norm": 1.182023525238037,
      "learning_rate": 2.7468145824253977e-05,
      "loss": 9.5053,
      "step": 1916
    },
    {
      "epoch": 0.6612625043118316,
      "grad_norm": 1.197792887687683,
      "learning_rate": 2.7418061375825087e-05,
      "loss": 9.556,
      "step": 1917
    },
    {
      "epoch": 0.661607450845119,
      "grad_norm": 1.1292563676834106,
      "learning_rate": 2.7368005375617994e-05,
      "loss": 9.5674,
      "step": 1918
    },
    {
      "epoch": 0.6619523973784064,
      "grad_norm": 1.3293133974075317,
      "learning_rate": 2.7317977886692147e-05,
      "loss": 9.4748,
      "step": 1919
    },
    {
      "epoch": 0.6622973439116937,
      "grad_norm": 1.1616480350494385,
      "learning_rate": 2.7267978972071057e-05,
      "loss": 9.5377,
      "step": 1920
    },
    {
      "epoch": 0.662642290444981,
      "grad_norm": 1.3261762857437134,
      "learning_rate": 2.721800869474226e-05,
      "loss": 9.5438,
      "step": 1921
    },
    {
      "epoch": 0.6629872369782683,
      "grad_norm": 1.2228319644927979,
      "learning_rate": 2.716806711765716e-05,
      "loss": 9.4966,
      "step": 1922
    },
    {
      "epoch": 0.6633321835115558,
      "grad_norm": 1.2406237125396729,
      "learning_rate": 2.7118154303731048e-05,
      "loss": 9.4417,
      "step": 1923
    },
    {
      "epoch": 0.6636771300448431,
      "grad_norm": 1.3249214887619019,
      "learning_rate": 2.706827031584301e-05,
      "loss": 9.4701,
      "step": 1924
    },
    {
      "epoch": 0.6640220765781304,
      "grad_norm": 1.22321617603302,
      "learning_rate": 2.7018415216835725e-05,
      "loss": 9.4863,
      "step": 1925
    },
    {
      "epoch": 0.6643670231114177,
      "grad_norm": 1.1897952556610107,
      "learning_rate": 2.6968589069515558e-05,
      "loss": 9.5445,
      "step": 1926
    },
    {
      "epoch": 0.664711969644705,
      "grad_norm": 1.3240059614181519,
      "learning_rate": 2.6918791936652376e-05,
      "loss": 9.4969,
      "step": 1927
    },
    {
      "epoch": 0.6650569161779925,
      "grad_norm": 1.255070447921753,
      "learning_rate": 2.6869023880979516e-05,
      "loss": 9.5387,
      "step": 1928
    },
    {
      "epoch": 0.6654018627112798,
      "grad_norm": 1.351804494857788,
      "learning_rate": 2.6819284965193625e-05,
      "loss": 9.5075,
      "step": 1929
    },
    {
      "epoch": 0.6657468092445671,
      "grad_norm": 1.2300724983215332,
      "learning_rate": 2.6769575251954704e-05,
      "loss": 9.5479,
      "step": 1930
    },
    {
      "epoch": 0.6660917557778544,
      "grad_norm": 1.231567144393921,
      "learning_rate": 2.671989480388592e-05,
      "loss": 9.4898,
      "step": 1931
    },
    {
      "epoch": 0.6664367023111418,
      "grad_norm": 1.3243263959884644,
      "learning_rate": 2.6670243683573598e-05,
      "loss": 9.4612,
      "step": 1932
    },
    {
      "epoch": 0.6667816488444291,
      "grad_norm": 1.426049828529358,
      "learning_rate": 2.6620621953567115e-05,
      "loss": 9.4235,
      "step": 1933
    },
    {
      "epoch": 0.6671265953777165,
      "grad_norm": 1.2795472145080566,
      "learning_rate": 2.657102967637881e-05,
      "loss": 9.5161,
      "step": 1934
    },
    {
      "epoch": 0.6674715419110038,
      "grad_norm": 1.3263260126113892,
      "learning_rate": 2.652146691448396e-05,
      "loss": 9.4876,
      "step": 1935
    },
    {
      "epoch": 0.6678164884442911,
      "grad_norm": 1.2887285947799683,
      "learning_rate": 2.6471933730320576e-05,
      "loss": 9.5039,
      "step": 1936
    },
    {
      "epoch": 0.6681614349775785,
      "grad_norm": 1.3971636295318604,
      "learning_rate": 2.64224301862895e-05,
      "loss": 9.4347,
      "step": 1937
    },
    {
      "epoch": 0.6685063815108658,
      "grad_norm": 1.395354151725769,
      "learning_rate": 2.6372956344754142e-05,
      "loss": 9.519,
      "step": 1938
    },
    {
      "epoch": 0.6688513280441531,
      "grad_norm": 1.3385522365570068,
      "learning_rate": 2.6323512268040597e-05,
      "loss": 9.4831,
      "step": 1939
    },
    {
      "epoch": 0.6691962745774405,
      "grad_norm": 1.4280192852020264,
      "learning_rate": 2.6274098018437343e-05,
      "loss": 9.471,
      "step": 1940
    },
    {
      "epoch": 0.6695412211107279,
      "grad_norm": 1.4592801332473755,
      "learning_rate": 2.622471365819542e-05,
      "loss": 9.4051,
      "step": 1941
    },
    {
      "epoch": 0.6698861676440152,
      "grad_norm": 1.3051594495773315,
      "learning_rate": 2.6175359249528088e-05,
      "loss": 9.497,
      "step": 1942
    },
    {
      "epoch": 0.6702311141773025,
      "grad_norm": 1.4578258991241455,
      "learning_rate": 2.612603485461097e-05,
      "loss": 9.4316,
      "step": 1943
    },
    {
      "epoch": 0.6705760607105898,
      "grad_norm": 1.4668503999710083,
      "learning_rate": 2.6076740535581802e-05,
      "loss": 9.4014,
      "step": 1944
    },
    {
      "epoch": 0.6709210072438772,
      "grad_norm": 1.718100666999817,
      "learning_rate": 2.602747635454047e-05,
      "loss": 9.3969,
      "step": 1945
    },
    {
      "epoch": 0.6712659537771646,
      "grad_norm": 1.4831199645996094,
      "learning_rate": 2.5978242373548915e-05,
      "loss": 9.4129,
      "step": 1946
    },
    {
      "epoch": 0.6716109003104519,
      "grad_norm": 1.6233813762664795,
      "learning_rate": 2.5929038654630954e-05,
      "loss": 9.4001,
      "step": 1947
    },
    {
      "epoch": 0.6719558468437392,
      "grad_norm": 1.645786166191101,
      "learning_rate": 2.587986525977241e-05,
      "loss": 9.3544,
      "step": 1948
    },
    {
      "epoch": 0.6723007933770265,
      "grad_norm": 1.6554784774780273,
      "learning_rate": 2.5830722250920768e-05,
      "loss": 9.4351,
      "step": 1949
    },
    {
      "epoch": 0.672645739910314,
      "grad_norm": 1.8901984691619873,
      "learning_rate": 2.578160968998532e-05,
      "loss": 9.2865,
      "step": 1950
    },
    {
      "epoch": 0.6729906864436013,
      "grad_norm": 1.0118002891540527,
      "learning_rate": 2.5732527638836957e-05,
      "loss": 9.5655,
      "step": 1951
    },
    {
      "epoch": 0.6733356329768886,
      "grad_norm": 0.9816641807556152,
      "learning_rate": 2.568347615930816e-05,
      "loss": 9.538,
      "step": 1952
    },
    {
      "epoch": 0.6736805795101759,
      "grad_norm": 1.0502429008483887,
      "learning_rate": 2.5634455313192872e-05,
      "loss": 9.5611,
      "step": 1953
    },
    {
      "epoch": 0.6740255260434632,
      "grad_norm": 1.091073989868164,
      "learning_rate": 2.558546516224648e-05,
      "loss": 9.501,
      "step": 1954
    },
    {
      "epoch": 0.6743704725767506,
      "grad_norm": 1.0108824968338013,
      "learning_rate": 2.5536505768185664e-05,
      "loss": 9.6199,
      "step": 1955
    },
    {
      "epoch": 0.674715419110038,
      "grad_norm": 1.1202064752578735,
      "learning_rate": 2.5487577192688388e-05,
      "loss": 9.5031,
      "step": 1956
    },
    {
      "epoch": 0.6750603656433253,
      "grad_norm": 1.0962861776351929,
      "learning_rate": 2.5438679497393793e-05,
      "loss": 9.5629,
      "step": 1957
    },
    {
      "epoch": 0.6754053121766126,
      "grad_norm": 1.1996490955352783,
      "learning_rate": 2.5389812743902063e-05,
      "loss": 9.4618,
      "step": 1958
    },
    {
      "epoch": 0.6757502587099,
      "grad_norm": 1.098293662071228,
      "learning_rate": 2.534097699377449e-05,
      "loss": 9.5814,
      "step": 1959
    },
    {
      "epoch": 0.6760952052431873,
      "grad_norm": 1.176307201385498,
      "learning_rate": 2.5292172308533213e-05,
      "loss": 9.4972,
      "step": 1960
    },
    {
      "epoch": 0.6764401517764747,
      "grad_norm": 1.2546685934066772,
      "learning_rate": 2.5243398749661307e-05,
      "loss": 9.5163,
      "step": 1961
    },
    {
      "epoch": 0.676785098309762,
      "grad_norm": 1.1058201789855957,
      "learning_rate": 2.519465637860261e-05,
      "loss": 9.5297,
      "step": 1962
    },
    {
      "epoch": 0.6771300448430493,
      "grad_norm": 1.1880666017532349,
      "learning_rate": 2.5145945256761693e-05,
      "loss": 9.5706,
      "step": 1963
    },
    {
      "epoch": 0.6774749913763367,
      "grad_norm": 1.3704041242599487,
      "learning_rate": 2.5097265445503703e-05,
      "loss": 9.4795,
      "step": 1964
    },
    {
      "epoch": 0.677819937909624,
      "grad_norm": 1.386462688446045,
      "learning_rate": 2.504861700615442e-05,
      "loss": 9.5004,
      "step": 1965
    },
    {
      "epoch": 0.6781648844429113,
      "grad_norm": 1.2360806465148926,
      "learning_rate": 2.500000000000001e-05,
      "loss": 9.5056,
      "step": 1966
    },
    {
      "epoch": 0.6785098309761987,
      "grad_norm": 1.2533848285675049,
      "learning_rate": 2.4951414488287123e-05,
      "loss": 9.4878,
      "step": 1967
    },
    {
      "epoch": 0.6788547775094861,
      "grad_norm": 1.3033040761947632,
      "learning_rate": 2.4902860532222692e-05,
      "loss": 9.4286,
      "step": 1968
    },
    {
      "epoch": 0.6791997240427734,
      "grad_norm": 1.2532498836517334,
      "learning_rate": 2.4854338192973913e-05,
      "loss": 9.4829,
      "step": 1969
    },
    {
      "epoch": 0.6795446705760607,
      "grad_norm": 1.265669345855713,
      "learning_rate": 2.480584753166816e-05,
      "loss": 9.4645,
      "step": 1970
    },
    {
      "epoch": 0.679889617109348,
      "grad_norm": 1.2609977722167969,
      "learning_rate": 2.475738860939285e-05,
      "loss": 9.4758,
      "step": 1971
    },
    {
      "epoch": 0.6802345636426353,
      "grad_norm": 1.324218511581421,
      "learning_rate": 2.4708961487195486e-05,
      "loss": 9.452,
      "step": 1972
    },
    {
      "epoch": 0.6805795101759228,
      "grad_norm": 1.5088634490966797,
      "learning_rate": 2.4660566226083444e-05,
      "loss": 9.4525,
      "step": 1973
    },
    {
      "epoch": 0.6809244567092101,
      "grad_norm": 1.3118107318878174,
      "learning_rate": 2.4612202887024016e-05,
      "loss": 9.5058,
      "step": 1974
    },
    {
      "epoch": 0.6812694032424974,
      "grad_norm": 1.4367679357528687,
      "learning_rate": 2.456387153094421e-05,
      "loss": 9.4298,
      "step": 1975
    },
    {
      "epoch": 0.6816143497757847,
      "grad_norm": 1.1999366283416748,
      "learning_rate": 2.4515572218730866e-05,
      "loss": 9.5317,
      "step": 1976
    },
    {
      "epoch": 0.681959296309072,
      "grad_norm": 1.3008366823196411,
      "learning_rate": 2.4467305011230318e-05,
      "loss": 9.498,
      "step": 1977
    },
    {
      "epoch": 0.6823042428423595,
      "grad_norm": 1.4255015850067139,
      "learning_rate": 2.4419069969248533e-05,
      "loss": 9.5161,
      "step": 1978
    },
    {
      "epoch": 0.6826491893756468,
      "grad_norm": 1.217320203781128,
      "learning_rate": 2.4370867153550956e-05,
      "loss": 9.4937,
      "step": 1979
    },
    {
      "epoch": 0.6829941359089341,
      "grad_norm": 1.2340617179870605,
      "learning_rate": 2.432269662486239e-05,
      "loss": 9.5439,
      "step": 1980
    },
    {
      "epoch": 0.6833390824422214,
      "grad_norm": 1.2029873132705688,
      "learning_rate": 2.4274558443867024e-05,
      "loss": 9.54,
      "step": 1981
    },
    {
      "epoch": 0.6836840289755088,
      "grad_norm": 1.3623865842819214,
      "learning_rate": 2.42264526712082e-05,
      "loss": 9.4325,
      "step": 1982
    },
    {
      "epoch": 0.6840289755087962,
      "grad_norm": 1.3263287544250488,
      "learning_rate": 2.417837936748858e-05,
      "loss": 9.4755,
      "step": 1983
    },
    {
      "epoch": 0.6843739220420835,
      "grad_norm": 1.3248276710510254,
      "learning_rate": 2.4130338593269773e-05,
      "loss": 9.4778,
      "step": 1984
    },
    {
      "epoch": 0.6847188685753708,
      "grad_norm": 1.3979350328445435,
      "learning_rate": 2.408233040907252e-05,
      "loss": 9.4865,
      "step": 1985
    },
    {
      "epoch": 0.6850638151086581,
      "grad_norm": 1.253639817237854,
      "learning_rate": 2.4034354875376414e-05,
      "loss": 9.5268,
      "step": 1986
    },
    {
      "epoch": 0.6854087616419455,
      "grad_norm": 1.254916787147522,
      "learning_rate": 2.3986412052619985e-05,
      "loss": 9.4905,
      "step": 1987
    },
    {
      "epoch": 0.6857537081752328,
      "grad_norm": 1.5778090953826904,
      "learning_rate": 2.393850200120054e-05,
      "loss": 9.4679,
      "step": 1988
    },
    {
      "epoch": 0.6860986547085202,
      "grad_norm": 1.5365502834320068,
      "learning_rate": 2.3890624781474068e-05,
      "loss": 9.4484,
      "step": 1989
    },
    {
      "epoch": 0.6864436012418075,
      "grad_norm": 1.519955039024353,
      "learning_rate": 2.384278045375523e-05,
      "loss": 9.4295,
      "step": 1990
    },
    {
      "epoch": 0.6867885477750949,
      "grad_norm": 1.3635841608047485,
      "learning_rate": 2.3794969078317243e-05,
      "loss": 9.4853,
      "step": 1991
    },
    {
      "epoch": 0.6871334943083822,
      "grad_norm": 1.5128055810928345,
      "learning_rate": 2.3747190715391824e-05,
      "loss": 9.4405,
      "step": 1992
    },
    {
      "epoch": 0.6874784408416695,
      "grad_norm": 1.3982264995574951,
      "learning_rate": 2.369944542516906e-05,
      "loss": 9.4502,
      "step": 1993
    },
    {
      "epoch": 0.6878233873749569,
      "grad_norm": 1.3720015287399292,
      "learning_rate": 2.365173326779743e-05,
      "loss": 9.4478,
      "step": 1994
    },
    {
      "epoch": 0.6881683339082442,
      "grad_norm": 1.5373377799987793,
      "learning_rate": 2.360405430338361e-05,
      "loss": 9.3671,
      "step": 1995
    },
    {
      "epoch": 0.6885132804415316,
      "grad_norm": 1.6050822734832764,
      "learning_rate": 2.355640859199251e-05,
      "loss": 9.4249,
      "step": 1996
    },
    {
      "epoch": 0.6888582269748189,
      "grad_norm": 1.7167375087738037,
      "learning_rate": 2.3508796193647138e-05,
      "loss": 9.4011,
      "step": 1997
    },
    {
      "epoch": 0.6892031735081062,
      "grad_norm": 1.7736046314239502,
      "learning_rate": 2.346121716832855e-05,
      "loss": 9.4303,
      "step": 1998
    },
    {
      "epoch": 0.6895481200413935,
      "grad_norm": 1.566192626953125,
      "learning_rate": 2.341367157597569e-05,
      "loss": 9.3759,
      "step": 1999
    },
    {
      "epoch": 0.689893066574681,
      "grad_norm": 1.855802059173584,
      "learning_rate": 2.336615947648546e-05,
      "loss": 9.3965,
      "step": 2000
    },
    {
      "epoch": 0.6902380131079683,
      "grad_norm": 1.0500324964523315,
      "learning_rate": 2.331868092971256e-05,
      "loss": 9.5872,
      "step": 2001
    },
    {
      "epoch": 0.6905829596412556,
      "grad_norm": 0.9782236814498901,
      "learning_rate": 2.3271235995469363e-05,
      "loss": 9.6103,
      "step": 2002
    },
    {
      "epoch": 0.6909279061745429,
      "grad_norm": 0.9627702236175537,
      "learning_rate": 2.322382473352596e-05,
      "loss": 9.498,
      "step": 2003
    },
    {
      "epoch": 0.6912728527078302,
      "grad_norm": 1.0011677742004395,
      "learning_rate": 2.3176447203610003e-05,
      "loss": 9.5546,
      "step": 2004
    },
    {
      "epoch": 0.6916177992411177,
      "grad_norm": 1.1541500091552734,
      "learning_rate": 2.3129103465406654e-05,
      "loss": 9.5231,
      "step": 2005
    },
    {
      "epoch": 0.691962745774405,
      "grad_norm": 1.135690450668335,
      "learning_rate": 2.3081793578558462e-05,
      "loss": 9.6356,
      "step": 2006
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 1.2155224084854126,
      "learning_rate": 2.3034517602665407e-05,
      "loss": 9.5496,
      "step": 2007
    },
    {
      "epoch": 0.6926526388409796,
      "grad_norm": 1.1832019090652466,
      "learning_rate": 2.298727559728467e-05,
      "loss": 9.4695,
      "step": 2008
    },
    {
      "epoch": 0.692997585374267,
      "grad_norm": 1.1747037172317505,
      "learning_rate": 2.294006762193069e-05,
      "loss": 9.5464,
      "step": 2009
    },
    {
      "epoch": 0.6933425319075543,
      "grad_norm": 1.2343512773513794,
      "learning_rate": 2.2892893736075016e-05,
      "loss": 9.506,
      "step": 2010
    },
    {
      "epoch": 0.6936874784408417,
      "grad_norm": 1.1944860219955444,
      "learning_rate": 2.2845753999146284e-05,
      "loss": 9.5046,
      "step": 2011
    },
    {
      "epoch": 0.694032424974129,
      "grad_norm": 1.2822582721710205,
      "learning_rate": 2.2798648470530038e-05,
      "loss": 9.5136,
      "step": 2012
    },
    {
      "epoch": 0.6943773715074163,
      "grad_norm": 1.2347842454910278,
      "learning_rate": 2.2751577209568793e-05,
      "loss": 9.5564,
      "step": 2013
    },
    {
      "epoch": 0.6947223180407037,
      "grad_norm": 1.1867071390151978,
      "learning_rate": 2.2704540275561887e-05,
      "loss": 9.4912,
      "step": 2014
    },
    {
      "epoch": 0.695067264573991,
      "grad_norm": 1.163027048110962,
      "learning_rate": 2.265753772776536e-05,
      "loss": 9.5166,
      "step": 2015
    },
    {
      "epoch": 0.6954122111072784,
      "grad_norm": 1.1979963779449463,
      "learning_rate": 2.261056962539201e-05,
      "loss": 9.603,
      "step": 2016
    },
    {
      "epoch": 0.6957571576405657,
      "grad_norm": 1.2587945461273193,
      "learning_rate": 2.256363602761115e-05,
      "loss": 9.4699,
      "step": 2017
    },
    {
      "epoch": 0.6961021041738531,
      "grad_norm": 1.2772116661071777,
      "learning_rate": 2.2516736993548755e-05,
      "loss": 9.5098,
      "step": 2018
    },
    {
      "epoch": 0.6964470507071404,
      "grad_norm": 1.1604729890823364,
      "learning_rate": 2.2469872582287118e-05,
      "loss": 9.5239,
      "step": 2019
    },
    {
      "epoch": 0.6967919972404277,
      "grad_norm": 1.2303416728973389,
      "learning_rate": 2.2423042852865013e-05,
      "loss": 9.5669,
      "step": 2020
    },
    {
      "epoch": 0.697136943773715,
      "grad_norm": 1.3527498245239258,
      "learning_rate": 2.237624786427746e-05,
      "loss": 9.4579,
      "step": 2021
    },
    {
      "epoch": 0.6974818903070024,
      "grad_norm": 1.2704267501831055,
      "learning_rate": 2.2329487675475753e-05,
      "loss": 9.4686,
      "step": 2022
    },
    {
      "epoch": 0.6978268368402898,
      "grad_norm": 1.1141988039016724,
      "learning_rate": 2.2282762345367346e-05,
      "loss": 9.5626,
      "step": 2023
    },
    {
      "epoch": 0.6981717833735771,
      "grad_norm": 1.3489370346069336,
      "learning_rate": 2.223607193281572e-05,
      "loss": 9.4736,
      "step": 2024
    },
    {
      "epoch": 0.6985167299068644,
      "grad_norm": 1.3955328464508057,
      "learning_rate": 2.2189416496640488e-05,
      "loss": 9.4638,
      "step": 2025
    },
    {
      "epoch": 0.6988616764401517,
      "grad_norm": 1.254169225692749,
      "learning_rate": 2.2142796095617063e-05,
      "loss": 9.5615,
      "step": 2026
    },
    {
      "epoch": 0.6992066229734392,
      "grad_norm": 1.3857367038726807,
      "learning_rate": 2.2096210788476822e-05,
      "loss": 9.5208,
      "step": 2027
    },
    {
      "epoch": 0.6995515695067265,
      "grad_norm": 1.374439001083374,
      "learning_rate": 2.2049660633906865e-05,
      "loss": 9.5285,
      "step": 2028
    },
    {
      "epoch": 0.6998965160400138,
      "grad_norm": 1.1612699031829834,
      "learning_rate": 2.200314569055007e-05,
      "loss": 9.5181,
      "step": 2029
    },
    {
      "epoch": 0.7002414625733011,
      "grad_norm": 1.3591151237487793,
      "learning_rate": 2.1956666017004867e-05,
      "loss": 9.4199,
      "step": 2030
    },
    {
      "epoch": 0.7005864091065884,
      "grad_norm": 1.3390651941299438,
      "learning_rate": 2.1910221671825388e-05,
      "loss": 9.4853,
      "step": 2031
    },
    {
      "epoch": 0.7009313556398759,
      "grad_norm": 1.390390396118164,
      "learning_rate": 2.1863812713521115e-05,
      "loss": 9.4814,
      "step": 2032
    },
    {
      "epoch": 0.7012763021731632,
      "grad_norm": 1.3022857904434204,
      "learning_rate": 2.1817439200557065e-05,
      "loss": 9.466,
      "step": 2033
    },
    {
      "epoch": 0.7016212487064505,
      "grad_norm": 1.3257852792739868,
      "learning_rate": 2.177110119135352e-05,
      "loss": 9.4746,
      "step": 2034
    },
    {
      "epoch": 0.7019661952397378,
      "grad_norm": 1.3743329048156738,
      "learning_rate": 2.1724798744286072e-05,
      "loss": 9.4532,
      "step": 2035
    },
    {
      "epoch": 0.7023111417730252,
      "grad_norm": 1.3872151374816895,
      "learning_rate": 2.1678531917685545e-05,
      "loss": 9.5759,
      "step": 2036
    },
    {
      "epoch": 0.7026560883063125,
      "grad_norm": 1.346116542816162,
      "learning_rate": 2.163230076983779e-05,
      "loss": 9.4619,
      "step": 2037
    },
    {
      "epoch": 0.7030010348395999,
      "grad_norm": 1.4963953495025635,
      "learning_rate": 2.158610535898384e-05,
      "loss": 9.4303,
      "step": 2038
    },
    {
      "epoch": 0.7033459813728872,
      "grad_norm": 1.4644328355789185,
      "learning_rate": 2.15399457433196e-05,
      "loss": 9.4146,
      "step": 2039
    },
    {
      "epoch": 0.7036909279061745,
      "grad_norm": 1.5116710662841797,
      "learning_rate": 2.1493821980995953e-05,
      "loss": 9.4338,
      "step": 2040
    },
    {
      "epoch": 0.7040358744394619,
      "grad_norm": 1.473137378692627,
      "learning_rate": 2.1447734130118546e-05,
      "loss": 9.4159,
      "step": 2041
    },
    {
      "epoch": 0.7043808209727492,
      "grad_norm": 1.3014733791351318,
      "learning_rate": 2.1401682248747867e-05,
      "loss": 9.4929,
      "step": 2042
    },
    {
      "epoch": 0.7047257675060365,
      "grad_norm": 1.5376137495040894,
      "learning_rate": 2.135566639489901e-05,
      "loss": 9.4507,
      "step": 2043
    },
    {
      "epoch": 0.7050707140393239,
      "grad_norm": 1.567149043083191,
      "learning_rate": 2.130968662654173e-05,
      "loss": 9.4694,
      "step": 2044
    },
    {
      "epoch": 0.7054156605726113,
      "grad_norm": 1.4803193807601929,
      "learning_rate": 2.126374300160032e-05,
      "loss": 9.388,
      "step": 2045
    },
    {
      "epoch": 0.7057606071058986,
      "grad_norm": 1.4357073307037354,
      "learning_rate": 2.1217835577953525e-05,
      "loss": 9.4049,
      "step": 2046
    },
    {
      "epoch": 0.7061055536391859,
      "grad_norm": 1.5136798620224,
      "learning_rate": 2.117196441343452e-05,
      "loss": 9.4743,
      "step": 2047
    },
    {
      "epoch": 0.7064505001724732,
      "grad_norm": 1.4890515804290771,
      "learning_rate": 2.1126129565830728e-05,
      "loss": 9.4143,
      "step": 2048
    },
    {
      "epoch": 0.7067954467057606,
      "grad_norm": 1.567327618598938,
      "learning_rate": 2.1080331092883905e-05,
      "loss": 9.4024,
      "step": 2049
    },
    {
      "epoch": 0.707140393239048,
      "grad_norm": 1.809950590133667,
      "learning_rate": 2.103456905228991e-05,
      "loss": 9.4008,
      "step": 2050
    },
    {
      "epoch": 0.7074853397723353,
      "grad_norm": 0.894708514213562,
      "learning_rate": 2.098884350169876e-05,
      "loss": 9.5418,
      "step": 2051
    },
    {
      "epoch": 0.7078302863056226,
      "grad_norm": 1.0703632831573486,
      "learning_rate": 2.0943154498714467e-05,
      "loss": 9.5756,
      "step": 2052
    },
    {
      "epoch": 0.7081752328389099,
      "grad_norm": 1.0689934492111206,
      "learning_rate": 2.0897502100895054e-05,
      "loss": 9.5163,
      "step": 2053
    },
    {
      "epoch": 0.7085201793721974,
      "grad_norm": 1.0289816856384277,
      "learning_rate": 2.0851886365752348e-05,
      "loss": 9.5832,
      "step": 2054
    },
    {
      "epoch": 0.7088651259054847,
      "grad_norm": 1.028753638267517,
      "learning_rate": 2.0806307350752068e-05,
      "loss": 9.5606,
      "step": 2055
    },
    {
      "epoch": 0.709210072438772,
      "grad_norm": 1.2290129661560059,
      "learning_rate": 2.07607651133136e-05,
      "loss": 9.5328,
      "step": 2056
    },
    {
      "epoch": 0.7095550189720593,
      "grad_norm": 1.2330552339553833,
      "learning_rate": 2.0715259710810065e-05,
      "loss": 9.4955,
      "step": 2057
    },
    {
      "epoch": 0.7098999655053466,
      "grad_norm": 1.059729814529419,
      "learning_rate": 2.0669791200568144e-05,
      "loss": 9.5306,
      "step": 2058
    },
    {
      "epoch": 0.710244912038634,
      "grad_norm": 1.1995996236801147,
      "learning_rate": 2.062435963986805e-05,
      "loss": 9.5008,
      "step": 2059
    },
    {
      "epoch": 0.7105898585719214,
      "grad_norm": 1.1978609561920166,
      "learning_rate": 2.0578965085943475e-05,
      "loss": 9.5495,
      "step": 2060
    },
    {
      "epoch": 0.7109348051052087,
      "grad_norm": 1.2156633138656616,
      "learning_rate": 2.053360759598142e-05,
      "loss": 9.4956,
      "step": 2061
    },
    {
      "epoch": 0.711279751638496,
      "grad_norm": 1.2758194208145142,
      "learning_rate": 2.0488287227122272e-05,
      "loss": 9.5299,
      "step": 2062
    },
    {
      "epoch": 0.7116246981717834,
      "grad_norm": 1.1374874114990234,
      "learning_rate": 2.044300403645959e-05,
      "loss": 9.5721,
      "step": 2063
    },
    {
      "epoch": 0.7119696447050707,
      "grad_norm": 1.1768476963043213,
      "learning_rate": 2.0397758081040157e-05,
      "loss": 9.4924,
      "step": 2064
    },
    {
      "epoch": 0.712314591238358,
      "grad_norm": 1.3491202592849731,
      "learning_rate": 2.035254941786377e-05,
      "loss": 9.4696,
      "step": 2065
    },
    {
      "epoch": 0.7126595377716454,
      "grad_norm": 1.257044792175293,
      "learning_rate": 2.0307378103883358e-05,
      "loss": 9.5058,
      "step": 2066
    },
    {
      "epoch": 0.7130044843049327,
      "grad_norm": 1.2255409955978394,
      "learning_rate": 2.0262244196004675e-05,
      "loss": 9.4967,
      "step": 2067
    },
    {
      "epoch": 0.7133494308382201,
      "grad_norm": 1.139023780822754,
      "learning_rate": 2.021714775108644e-05,
      "loss": 9.6305,
      "step": 2068
    },
    {
      "epoch": 0.7136943773715074,
      "grad_norm": 1.3005746603012085,
      "learning_rate": 2.0172088825940157e-05,
      "loss": 9.4542,
      "step": 2069
    },
    {
      "epoch": 0.7140393239047947,
      "grad_norm": 1.28018319606781,
      "learning_rate": 2.012706747733002e-05,
      "loss": 9.4719,
      "step": 2070
    },
    {
      "epoch": 0.7143842704380821,
      "grad_norm": 1.32435941696167,
      "learning_rate": 2.0082083761972952e-05,
      "loss": 9.4833,
      "step": 2071
    },
    {
      "epoch": 0.7147292169713695,
      "grad_norm": 1.2384377717971802,
      "learning_rate": 2.003713773653838e-05,
      "loss": 9.5325,
      "step": 2072
    },
    {
      "epoch": 0.7150741635046568,
      "grad_norm": 1.3312957286834717,
      "learning_rate": 1.9992229457648375e-05,
      "loss": 9.4717,
      "step": 2073
    },
    {
      "epoch": 0.7154191100379441,
      "grad_norm": 1.3630597591400146,
      "learning_rate": 1.994735898187733e-05,
      "loss": 9.4864,
      "step": 2074
    },
    {
      "epoch": 0.7157640565712314,
      "grad_norm": 1.2554008960723877,
      "learning_rate": 1.9902526365752105e-05,
      "loss": 9.4386,
      "step": 2075
    },
    {
      "epoch": 0.7161090031045187,
      "grad_norm": 1.365967035293579,
      "learning_rate": 1.985773166575179e-05,
      "loss": 9.5193,
      "step": 2076
    },
    {
      "epoch": 0.7164539496378062,
      "grad_norm": 1.196242332458496,
      "learning_rate": 1.981297493830778e-05,
      "loss": 9.5456,
      "step": 2077
    },
    {
      "epoch": 0.7167988961710935,
      "grad_norm": 1.3467918634414673,
      "learning_rate": 1.9768256239803574e-05,
      "loss": 9.4929,
      "step": 2078
    },
    {
      "epoch": 0.7171438427043808,
      "grad_norm": 1.429176688194275,
      "learning_rate": 1.97235756265748e-05,
      "loss": 9.478,
      "step": 2079
    },
    {
      "epoch": 0.7174887892376681,
      "grad_norm": 1.2556706666946411,
      "learning_rate": 1.9678933154909095e-05,
      "loss": 9.5675,
      "step": 2080
    },
    {
      "epoch": 0.7178337357709555,
      "grad_norm": 1.2309414148330688,
      "learning_rate": 1.963432888104606e-05,
      "loss": 9.5355,
      "step": 2081
    },
    {
      "epoch": 0.7181786823042429,
      "grad_norm": 1.325849175453186,
      "learning_rate": 1.9589762861177164e-05,
      "loss": 9.5337,
      "step": 2082
    },
    {
      "epoch": 0.7185236288375302,
      "grad_norm": 1.3904569149017334,
      "learning_rate": 1.9545235151445655e-05,
      "loss": 9.4365,
      "step": 2083
    },
    {
      "epoch": 0.7188685753708175,
      "grad_norm": 1.3091845512390137,
      "learning_rate": 1.9500745807946586e-05,
      "loss": 9.4856,
      "step": 2084
    },
    {
      "epoch": 0.7192135219041048,
      "grad_norm": 1.2799763679504395,
      "learning_rate": 1.9456294886726594e-05,
      "loss": 9.4987,
      "step": 2085
    },
    {
      "epoch": 0.7195584684373922,
      "grad_norm": 1.484225869178772,
      "learning_rate": 1.9411882443783984e-05,
      "loss": 9.4233,
      "step": 2086
    },
    {
      "epoch": 0.7199034149706796,
      "grad_norm": 1.3886871337890625,
      "learning_rate": 1.936750853506855e-05,
      "loss": 9.5053,
      "step": 2087
    },
    {
      "epoch": 0.7202483615039669,
      "grad_norm": 1.4728593826293945,
      "learning_rate": 1.932317321648158e-05,
      "loss": 9.5088,
      "step": 2088
    },
    {
      "epoch": 0.7205933080372542,
      "grad_norm": 1.5038708448410034,
      "learning_rate": 1.9278876543875675e-05,
      "loss": 9.4059,
      "step": 2089
    },
    {
      "epoch": 0.7209382545705416,
      "grad_norm": 1.4496952295303345,
      "learning_rate": 1.9234618573054807e-05,
      "loss": 9.44,
      "step": 2090
    },
    {
      "epoch": 0.7212832011038289,
      "grad_norm": 1.4279743432998657,
      "learning_rate": 1.9190399359774208e-05,
      "loss": 9.5069,
      "step": 2091
    },
    {
      "epoch": 0.7216281476371162,
      "grad_norm": 1.6038273572921753,
      "learning_rate": 1.9146218959740214e-05,
      "loss": 9.5013,
      "step": 2092
    },
    {
      "epoch": 0.7219730941704036,
      "grad_norm": 1.551633358001709,
      "learning_rate": 1.910207742861032e-05,
      "loss": 9.3824,
      "step": 2093
    },
    {
      "epoch": 0.7223180407036909,
      "grad_norm": 1.510208249092102,
      "learning_rate": 1.9057974821993046e-05,
      "loss": 9.375,
      "step": 2094
    },
    {
      "epoch": 0.7226629872369783,
      "grad_norm": 1.4527089595794678,
      "learning_rate": 1.9013911195447887e-05,
      "loss": 9.4562,
      "step": 2095
    },
    {
      "epoch": 0.7230079337702656,
      "grad_norm": 1.4268805980682373,
      "learning_rate": 1.8969886604485176e-05,
      "loss": 9.447,
      "step": 2096
    },
    {
      "epoch": 0.7233528803035529,
      "grad_norm": 1.4859588146209717,
      "learning_rate": 1.892590110456615e-05,
      "loss": 9.3738,
      "step": 2097
    },
    {
      "epoch": 0.7236978268368403,
      "grad_norm": 1.5741686820983887,
      "learning_rate": 1.888195475110272e-05,
      "loss": 9.4351,
      "step": 2098
    },
    {
      "epoch": 0.7240427733701277,
      "grad_norm": 1.651714563369751,
      "learning_rate": 1.8838047599457563e-05,
      "loss": 9.3773,
      "step": 2099
    },
    {
      "epoch": 0.724387719903415,
      "grad_norm": 1.6480320692062378,
      "learning_rate": 1.8794179704943865e-05,
      "loss": 9.3666,
      "step": 2100
    },
    {
      "epoch": 0.7247326664367023,
      "grad_norm": 0.9026673436164856,
      "learning_rate": 1.87503511228255e-05,
      "loss": 9.6036,
      "step": 2101
    },
    {
      "epoch": 0.7250776129699896,
      "grad_norm": 0.9937422275543213,
      "learning_rate": 1.8706561908316673e-05,
      "loss": 9.5324,
      "step": 2102
    },
    {
      "epoch": 0.7254225595032769,
      "grad_norm": 0.9917021989822388,
      "learning_rate": 1.8662812116582077e-05,
      "loss": 9.5648,
      "step": 2103
    },
    {
      "epoch": 0.7257675060365644,
      "grad_norm": 1.0386805534362793,
      "learning_rate": 1.8619101802736755e-05,
      "loss": 9.5532,
      "step": 2104
    },
    {
      "epoch": 0.7261124525698517,
      "grad_norm": 1.044515609741211,
      "learning_rate": 1.8575431021845928e-05,
      "loss": 9.5103,
      "step": 2105
    },
    {
      "epoch": 0.726457399103139,
      "grad_norm": 1.216599941253662,
      "learning_rate": 1.8531799828925123e-05,
      "loss": 9.466,
      "step": 2106
    },
    {
      "epoch": 0.7268023456364263,
      "grad_norm": 1.0359028577804565,
      "learning_rate": 1.8488208278939878e-05,
      "loss": 9.5026,
      "step": 2107
    },
    {
      "epoch": 0.7271472921697137,
      "grad_norm": 1.160091519355774,
      "learning_rate": 1.844465642680593e-05,
      "loss": 9.512,
      "step": 2108
    },
    {
      "epoch": 0.7274922387030011,
      "grad_norm": 1.1458526849746704,
      "learning_rate": 1.840114432738887e-05,
      "loss": 9.4816,
      "step": 2109
    },
    {
      "epoch": 0.7278371852362884,
      "grad_norm": 1.1700782775878906,
      "learning_rate": 1.8357672035504313e-05,
      "loss": 9.5054,
      "step": 2110
    },
    {
      "epoch": 0.7281821317695757,
      "grad_norm": 1.1910396814346313,
      "learning_rate": 1.8314239605917638e-05,
      "loss": 9.533,
      "step": 2111
    },
    {
      "epoch": 0.728527078302863,
      "grad_norm": 1.1200579404830933,
      "learning_rate": 1.8270847093344074e-05,
      "loss": 9.5353,
      "step": 2112
    },
    {
      "epoch": 0.7288720248361504,
      "grad_norm": 1.1751587390899658,
      "learning_rate": 1.8227494552448547e-05,
      "loss": 9.5049,
      "step": 2113
    },
    {
      "epoch": 0.7292169713694377,
      "grad_norm": 1.2288830280303955,
      "learning_rate": 1.81841820378456e-05,
      "loss": 9.4904,
      "step": 2114
    },
    {
      "epoch": 0.7295619179027251,
      "grad_norm": 1.1687408685684204,
      "learning_rate": 1.814090960409937e-05,
      "loss": 9.4727,
      "step": 2115
    },
    {
      "epoch": 0.7299068644360124,
      "grad_norm": 1.3285386562347412,
      "learning_rate": 1.8097677305723517e-05,
      "loss": 9.4991,
      "step": 2116
    },
    {
      "epoch": 0.7302518109692998,
      "grad_norm": 1.1721177101135254,
      "learning_rate": 1.8054485197181136e-05,
      "loss": 9.4992,
      "step": 2117
    },
    {
      "epoch": 0.7305967575025871,
      "grad_norm": 1.2847938537597656,
      "learning_rate": 1.8011333332884645e-05,
      "loss": 9.4851,
      "step": 2118
    },
    {
      "epoch": 0.7309417040358744,
      "grad_norm": 1.2144776582717896,
      "learning_rate": 1.7968221767195836e-05,
      "loss": 9.5051,
      "step": 2119
    },
    {
      "epoch": 0.7312866505691618,
      "grad_norm": 1.2324657440185547,
      "learning_rate": 1.7925150554425662e-05,
      "loss": 9.4642,
      "step": 2120
    },
    {
      "epoch": 0.7316315971024491,
      "grad_norm": 1.3017010688781738,
      "learning_rate": 1.788211974883429e-05,
      "loss": 9.5029,
      "step": 2121
    },
    {
      "epoch": 0.7319765436357365,
      "grad_norm": 1.2795159816741943,
      "learning_rate": 1.7839129404630965e-05,
      "loss": 9.5278,
      "step": 2122
    },
    {
      "epoch": 0.7323214901690238,
      "grad_norm": 1.1670211553573608,
      "learning_rate": 1.7796179575973988e-05,
      "loss": 9.5435,
      "step": 2123
    },
    {
      "epoch": 0.7326664367023111,
      "grad_norm": 1.2223438024520874,
      "learning_rate": 1.775327031697055e-05,
      "loss": 9.5277,
      "step": 2124
    },
    {
      "epoch": 0.7330113832355984,
      "grad_norm": 1.2785682678222656,
      "learning_rate": 1.7710401681676803e-05,
      "loss": 9.5291,
      "step": 2125
    },
    {
      "epoch": 0.7333563297688859,
      "grad_norm": 1.332833170890808,
      "learning_rate": 1.7667573724097702e-05,
      "loss": 9.5411,
      "step": 2126
    },
    {
      "epoch": 0.7337012763021732,
      "grad_norm": 1.4342169761657715,
      "learning_rate": 1.762478649818693e-05,
      "loss": 9.4796,
      "step": 2127
    },
    {
      "epoch": 0.7340462228354605,
      "grad_norm": 1.2753576040267944,
      "learning_rate": 1.7582040057846883e-05,
      "loss": 9.5234,
      "step": 2128
    },
    {
      "epoch": 0.7343911693687478,
      "grad_norm": 1.2632288932800293,
      "learning_rate": 1.753933445692858e-05,
      "loss": 9.4742,
      "step": 2129
    },
    {
      "epoch": 0.7347361159020351,
      "grad_norm": 1.3039580583572388,
      "learning_rate": 1.74966697492316e-05,
      "loss": 9.4876,
      "step": 2130
    },
    {
      "epoch": 0.7350810624353226,
      "grad_norm": 1.1701034307479858,
      "learning_rate": 1.745404598850395e-05,
      "loss": 9.5418,
      "step": 2131
    },
    {
      "epoch": 0.7354260089686099,
      "grad_norm": 1.2938871383666992,
      "learning_rate": 1.7411463228442122e-05,
      "loss": 9.4901,
      "step": 2132
    },
    {
      "epoch": 0.7357709555018972,
      "grad_norm": 1.3119385242462158,
      "learning_rate": 1.73689215226909e-05,
      "loss": 9.5219,
      "step": 2133
    },
    {
      "epoch": 0.7361159020351845,
      "grad_norm": 1.5491797924041748,
      "learning_rate": 1.7326420924843395e-05,
      "loss": 9.4397,
      "step": 2134
    },
    {
      "epoch": 0.7364608485684719,
      "grad_norm": 1.3859881162643433,
      "learning_rate": 1.7283961488440904e-05,
      "loss": 9.4493,
      "step": 2135
    },
    {
      "epoch": 0.7368057951017593,
      "grad_norm": 1.3091398477554321,
      "learning_rate": 1.7241543266972888e-05,
      "loss": 9.4656,
      "step": 2136
    },
    {
      "epoch": 0.7371507416350466,
      "grad_norm": 1.3951692581176758,
      "learning_rate": 1.719916631387685e-05,
      "loss": 9.534,
      "step": 2137
    },
    {
      "epoch": 0.7374956881683339,
      "grad_norm": 1.4812074899673462,
      "learning_rate": 1.7156830682538343e-05,
      "loss": 9.4484,
      "step": 2138
    },
    {
      "epoch": 0.7378406347016212,
      "grad_norm": 1.377474069595337,
      "learning_rate": 1.7114536426290868e-05,
      "loss": 9.4458,
      "step": 2139
    },
    {
      "epoch": 0.7381855812349086,
      "grad_norm": 1.3391168117523193,
      "learning_rate": 1.707228359841575e-05,
      "loss": 9.5431,
      "step": 2140
    },
    {
      "epoch": 0.7385305277681959,
      "grad_norm": 1.4583009481430054,
      "learning_rate": 1.7030072252142188e-05,
      "loss": 9.4805,
      "step": 2141
    },
    {
      "epoch": 0.7388754743014833,
      "grad_norm": 1.4434447288513184,
      "learning_rate": 1.6987902440647047e-05,
      "loss": 9.4072,
      "step": 2142
    },
    {
      "epoch": 0.7392204208347706,
      "grad_norm": 1.5743590593338013,
      "learning_rate": 1.6945774217054973e-05,
      "loss": 9.4339,
      "step": 2143
    },
    {
      "epoch": 0.739565367368058,
      "grad_norm": 1.3587831258773804,
      "learning_rate": 1.690368763443812e-05,
      "loss": 9.4421,
      "step": 2144
    },
    {
      "epoch": 0.7399103139013453,
      "grad_norm": 1.3895113468170166,
      "learning_rate": 1.6861642745816253e-05,
      "loss": 9.3958,
      "step": 2145
    },
    {
      "epoch": 0.7402552604346326,
      "grad_norm": 1.43183434009552,
      "learning_rate": 1.6819639604156557e-05,
      "loss": 9.4746,
      "step": 2146
    },
    {
      "epoch": 0.74060020696792,
      "grad_norm": 1.5698809623718262,
      "learning_rate": 1.6777678262373653e-05,
      "loss": 9.3977,
      "step": 2147
    },
    {
      "epoch": 0.7409451535012073,
      "grad_norm": 1.4829764366149902,
      "learning_rate": 1.6735758773329536e-05,
      "loss": 9.4524,
      "step": 2148
    },
    {
      "epoch": 0.7412901000344947,
      "grad_norm": 1.6046053171157837,
      "learning_rate": 1.6693881189833376e-05,
      "loss": 9.4178,
      "step": 2149
    },
    {
      "epoch": 0.741635046567782,
      "grad_norm": 1.7703441381454468,
      "learning_rate": 1.66520455646417e-05,
      "loss": 9.357,
      "step": 2150
    },
    {
      "epoch": 0.7419799931010693,
      "grad_norm": 0.9205633401870728,
      "learning_rate": 1.6610251950458033e-05,
      "loss": 9.5522,
      "step": 2151
    },
    {
      "epoch": 0.7423249396343566,
      "grad_norm": 1.0296859741210938,
      "learning_rate": 1.6568500399933073e-05,
      "loss": 9.5601,
      "step": 2152
    },
    {
      "epoch": 0.7426698861676441,
      "grad_norm": 1.0968191623687744,
      "learning_rate": 1.6526790965664457e-05,
      "loss": 9.5359,
      "step": 2153
    },
    {
      "epoch": 0.7430148327009314,
      "grad_norm": 1.110319972038269,
      "learning_rate": 1.648512370019683e-05,
      "loss": 9.5107,
      "step": 2154
    },
    {
      "epoch": 0.7433597792342187,
      "grad_norm": 1.0326136350631714,
      "learning_rate": 1.644349865602165e-05,
      "loss": 9.5391,
      "step": 2155
    },
    {
      "epoch": 0.743704725767506,
      "grad_norm": 1.1610885858535767,
      "learning_rate": 1.640191588557724e-05,
      "loss": 9.5459,
      "step": 2156
    },
    {
      "epoch": 0.7440496723007933,
      "grad_norm": 1.0827651023864746,
      "learning_rate": 1.636037544124864e-05,
      "loss": 9.5505,
      "step": 2157
    },
    {
      "epoch": 0.7443946188340808,
      "grad_norm": 1.101487398147583,
      "learning_rate": 1.6318877375367568e-05,
      "loss": 9.5195,
      "step": 2158
    },
    {
      "epoch": 0.7447395653673681,
      "grad_norm": 1.18477201461792,
      "learning_rate": 1.627742174021239e-05,
      "loss": 9.5322,
      "step": 2159
    },
    {
      "epoch": 0.7450845119006554,
      "grad_norm": 1.223777174949646,
      "learning_rate": 1.6236008588007945e-05,
      "loss": 9.5771,
      "step": 2160
    },
    {
      "epoch": 0.7454294584339427,
      "grad_norm": 1.2236958742141724,
      "learning_rate": 1.6194637970925634e-05,
      "loss": 9.5204,
      "step": 2161
    },
    {
      "epoch": 0.7457744049672301,
      "grad_norm": 1.1944174766540527,
      "learning_rate": 1.6153309941083204e-05,
      "loss": 9.5285,
      "step": 2162
    },
    {
      "epoch": 0.7461193515005174,
      "grad_norm": 1.1148663759231567,
      "learning_rate": 1.611202455054481e-05,
      "loss": 9.5056,
      "step": 2163
    },
    {
      "epoch": 0.7464642980338048,
      "grad_norm": 1.2774313688278198,
      "learning_rate": 1.607078185132084e-05,
      "loss": 9.5108,
      "step": 2164
    },
    {
      "epoch": 0.7468092445670921,
      "grad_norm": 1.3021870851516724,
      "learning_rate": 1.6029581895367962e-05,
      "loss": 9.5348,
      "step": 2165
    },
    {
      "epoch": 0.7471541911003794,
      "grad_norm": 1.3037549257278442,
      "learning_rate": 1.598842473458892e-05,
      "loss": 9.4827,
      "step": 2166
    },
    {
      "epoch": 0.7474991376336668,
      "grad_norm": 1.0977771282196045,
      "learning_rate": 1.5947310420832622e-05,
      "loss": 9.5269,
      "step": 2167
    },
    {
      "epoch": 0.7478440841669541,
      "grad_norm": 1.2924681901931763,
      "learning_rate": 1.5906239005893923e-05,
      "loss": 9.4714,
      "step": 2168
    },
    {
      "epoch": 0.7481890307002415,
      "grad_norm": 1.2541073560714722,
      "learning_rate": 1.58652105415137e-05,
      "loss": 9.5176,
      "step": 2169
    },
    {
      "epoch": 0.7485339772335288,
      "grad_norm": 1.374686360359192,
      "learning_rate": 1.5824225079378685e-05,
      "loss": 9.5176,
      "step": 2170
    },
    {
      "epoch": 0.7488789237668162,
      "grad_norm": 1.0905579328536987,
      "learning_rate": 1.578328267112146e-05,
      "loss": 9.5887,
      "step": 2171
    },
    {
      "epoch": 0.7492238703001035,
      "grad_norm": 1.2786760330200195,
      "learning_rate": 1.5742383368320358e-05,
      "loss": 9.5006,
      "step": 2172
    },
    {
      "epoch": 0.7495688168333908,
      "grad_norm": 1.3505619764328003,
      "learning_rate": 1.5701527222499386e-05,
      "loss": 9.4938,
      "step": 2173
    },
    {
      "epoch": 0.7499137633666781,
      "grad_norm": 1.2414374351501465,
      "learning_rate": 1.566071428512823e-05,
      "loss": 9.5063,
      "step": 2174
    },
    {
      "epoch": 0.7502587098999655,
      "grad_norm": 1.2529624700546265,
      "learning_rate": 1.561994460762209e-05,
      "loss": 9.5985,
      "step": 2175
    },
    {
      "epoch": 0.7506036564332529,
      "grad_norm": 1.2772570848464966,
      "learning_rate": 1.557921824134172e-05,
      "loss": 9.5573,
      "step": 2176
    },
    {
      "epoch": 0.7509486029665402,
      "grad_norm": 1.3001856803894043,
      "learning_rate": 1.5538535237593243e-05,
      "loss": 9.4594,
      "step": 2177
    },
    {
      "epoch": 0.7512935494998275,
      "grad_norm": 1.2496637105941772,
      "learning_rate": 1.5497895647628265e-05,
      "loss": 9.4983,
      "step": 2178
    },
    {
      "epoch": 0.7516384960331148,
      "grad_norm": 1.2094224691390991,
      "learning_rate": 1.545729952264358e-05,
      "loss": 9.4784,
      "step": 2179
    },
    {
      "epoch": 0.7519834425664023,
      "grad_norm": 1.3114935159683228,
      "learning_rate": 1.54167469137813e-05,
      "loss": 9.5164,
      "step": 2180
    },
    {
      "epoch": 0.7523283890996896,
      "grad_norm": 1.2864598035812378,
      "learning_rate": 1.5376237872128705e-05,
      "loss": 9.5024,
      "step": 2181
    },
    {
      "epoch": 0.7526733356329769,
      "grad_norm": 1.303964376449585,
      "learning_rate": 1.5335772448718157e-05,
      "loss": 9.5344,
      "step": 2182
    },
    {
      "epoch": 0.7530182821662642,
      "grad_norm": 1.4014822244644165,
      "learning_rate": 1.5295350694527117e-05,
      "loss": 9.4536,
      "step": 2183
    },
    {
      "epoch": 0.7533632286995515,
      "grad_norm": 1.2853933572769165,
      "learning_rate": 1.5254972660477961e-05,
      "loss": 9.5274,
      "step": 2184
    },
    {
      "epoch": 0.753708175232839,
      "grad_norm": 1.4304496049880981,
      "learning_rate": 1.5214638397438109e-05,
      "loss": 9.4957,
      "step": 2185
    },
    {
      "epoch": 0.7540531217661263,
      "grad_norm": 1.3712221384048462,
      "learning_rate": 1.5174347956219697e-05,
      "loss": 9.4559,
      "step": 2186
    },
    {
      "epoch": 0.7543980682994136,
      "grad_norm": 1.3548405170440674,
      "learning_rate": 1.5134101387579775e-05,
      "loss": 9.4248,
      "step": 2187
    },
    {
      "epoch": 0.7547430148327009,
      "grad_norm": 1.3276457786560059,
      "learning_rate": 1.5093898742220026e-05,
      "loss": 9.4374,
      "step": 2188
    },
    {
      "epoch": 0.7550879613659883,
      "grad_norm": 1.4717036485671997,
      "learning_rate": 1.5053740070786881e-05,
      "loss": 9.4696,
      "step": 2189
    },
    {
      "epoch": 0.7554329078992756,
      "grad_norm": 1.605868935585022,
      "learning_rate": 1.5013625423871307e-05,
      "loss": 9.3633,
      "step": 2190
    },
    {
      "epoch": 0.755777854432563,
      "grad_norm": 1.4265213012695312,
      "learning_rate": 1.4973554852008853e-05,
      "loss": 9.4659,
      "step": 2191
    },
    {
      "epoch": 0.7561228009658503,
      "grad_norm": 1.370707631111145,
      "learning_rate": 1.493352840567953e-05,
      "loss": 9.4668,
      "step": 2192
    },
    {
      "epoch": 0.7564677474991376,
      "grad_norm": 1.4738413095474243,
      "learning_rate": 1.4893546135307762e-05,
      "loss": 9.4277,
      "step": 2193
    },
    {
      "epoch": 0.756812694032425,
      "grad_norm": 1.393288016319275,
      "learning_rate": 1.4853608091262333e-05,
      "loss": 9.4523,
      "step": 2194
    },
    {
      "epoch": 0.7571576405657123,
      "grad_norm": 1.4397128820419312,
      "learning_rate": 1.4813714323856276e-05,
      "loss": 9.497,
      "step": 2195
    },
    {
      "epoch": 0.7575025870989996,
      "grad_norm": 1.4494057893753052,
      "learning_rate": 1.4773864883346889e-05,
      "loss": 9.4915,
      "step": 2196
    },
    {
      "epoch": 0.757847533632287,
      "grad_norm": 1.5768532752990723,
      "learning_rate": 1.473405981993558e-05,
      "loss": 9.4324,
      "step": 2197
    },
    {
      "epoch": 0.7581924801655744,
      "grad_norm": 1.572800874710083,
      "learning_rate": 1.4694299183767896e-05,
      "loss": 9.3683,
      "step": 2198
    },
    {
      "epoch": 0.7585374266988617,
      "grad_norm": 1.6129246950149536,
      "learning_rate": 1.46545830249334e-05,
      "loss": 9.4134,
      "step": 2199
    },
    {
      "epoch": 0.758882373232149,
      "grad_norm": 1.646519422531128,
      "learning_rate": 1.461491139346563e-05,
      "loss": 9.3784,
      "step": 2200
    },
    {
      "epoch": 0.7592273197654363,
      "grad_norm": 0.8638380765914917,
      "learning_rate": 1.4575284339341994e-05,
      "loss": 9.5856,
      "step": 2201
    },
    {
      "epoch": 0.7595722662987237,
      "grad_norm": 0.9636580348014832,
      "learning_rate": 1.4535701912483784e-05,
      "loss": 9.5805,
      "step": 2202
    },
    {
      "epoch": 0.7599172128320111,
      "grad_norm": 0.9889575839042664,
      "learning_rate": 1.4496164162756071e-05,
      "loss": 9.5958,
      "step": 2203
    },
    {
      "epoch": 0.7602621593652984,
      "grad_norm": 1.0838682651519775,
      "learning_rate": 1.4456671139967604e-05,
      "loss": 9.5557,
      "step": 2204
    },
    {
      "epoch": 0.7606071058985857,
      "grad_norm": 1.0837624073028564,
      "learning_rate": 1.4417222893870825e-05,
      "loss": 9.5078,
      "step": 2205
    },
    {
      "epoch": 0.760952052431873,
      "grad_norm": 1.0023666620254517,
      "learning_rate": 1.4377819474161746e-05,
      "loss": 9.5269,
      "step": 2206
    },
    {
      "epoch": 0.7612969989651605,
      "grad_norm": 1.0206482410430908,
      "learning_rate": 1.4338460930479935e-05,
      "loss": 9.5391,
      "step": 2207
    },
    {
      "epoch": 0.7616419454984478,
      "grad_norm": 1.030287265777588,
      "learning_rate": 1.4299147312408379e-05,
      "loss": 9.5096,
      "step": 2208
    },
    {
      "epoch": 0.7619868920317351,
      "grad_norm": 1.1585311889648438,
      "learning_rate": 1.4259878669473525e-05,
      "loss": 9.5347,
      "step": 2209
    },
    {
      "epoch": 0.7623318385650224,
      "grad_norm": 1.1257652044296265,
      "learning_rate": 1.4220655051145094e-05,
      "loss": 9.539,
      "step": 2210
    },
    {
      "epoch": 0.7626767850983097,
      "grad_norm": 1.0882645845413208,
      "learning_rate": 1.4181476506836161e-05,
      "loss": 9.5286,
      "step": 2211
    },
    {
      "epoch": 0.7630217316315971,
      "grad_norm": 1.1985927820205688,
      "learning_rate": 1.4142343085902937e-05,
      "loss": 9.5612,
      "step": 2212
    },
    {
      "epoch": 0.7633666781648845,
      "grad_norm": 1.2014905214309692,
      "learning_rate": 1.4103254837644892e-05,
      "loss": 9.4562,
      "step": 2213
    },
    {
      "epoch": 0.7637116246981718,
      "grad_norm": 1.1860039234161377,
      "learning_rate": 1.4064211811304472e-05,
      "loss": 9.5269,
      "step": 2214
    },
    {
      "epoch": 0.7640565712314591,
      "grad_norm": 1.2806318998336792,
      "learning_rate": 1.4025214056067238e-05,
      "loss": 9.4719,
      "step": 2215
    },
    {
      "epoch": 0.7644015177647465,
      "grad_norm": 1.3235222101211548,
      "learning_rate": 1.3986261621061703e-05,
      "loss": 9.523,
      "step": 2216
    },
    {
      "epoch": 0.7647464642980338,
      "grad_norm": 1.3476524353027344,
      "learning_rate": 1.3947354555359238e-05,
      "loss": 9.4753,
      "step": 2217
    },
    {
      "epoch": 0.7650914108313212,
      "grad_norm": 1.1621500253677368,
      "learning_rate": 1.390849290797413e-05,
      "loss": 9.5493,
      "step": 2218
    },
    {
      "epoch": 0.7654363573646085,
      "grad_norm": 1.3728376626968384,
      "learning_rate": 1.3869676727863368e-05,
      "loss": 9.4585,
      "step": 2219
    },
    {
      "epoch": 0.7657813038978958,
      "grad_norm": 1.260619878768921,
      "learning_rate": 1.3830906063926769e-05,
      "loss": 9.495,
      "step": 2220
    },
    {
      "epoch": 0.7661262504311832,
      "grad_norm": 1.451232671737671,
      "learning_rate": 1.3792180965006707e-05,
      "loss": 9.4501,
      "step": 2221
    },
    {
      "epoch": 0.7664711969644705,
      "grad_norm": 1.2228461503982544,
      "learning_rate": 1.375350147988822e-05,
      "loss": 9.4814,
      "step": 2222
    },
    {
      "epoch": 0.7668161434977578,
      "grad_norm": 1.3036428689956665,
      "learning_rate": 1.3714867657298835e-05,
      "loss": 9.4188,
      "step": 2223
    },
    {
      "epoch": 0.7671610900310452,
      "grad_norm": 1.3564863204956055,
      "learning_rate": 1.3676279545908593e-05,
      "loss": 9.4896,
      "step": 2224
    },
    {
      "epoch": 0.7675060365643326,
      "grad_norm": 1.2337634563446045,
      "learning_rate": 1.3637737194329947e-05,
      "loss": 9.5008,
      "step": 2225
    },
    {
      "epoch": 0.7678509830976199,
      "grad_norm": 1.2691717147827148,
      "learning_rate": 1.3599240651117662e-05,
      "loss": 9.5147,
      "step": 2226
    },
    {
      "epoch": 0.7681959296309072,
      "grad_norm": 1.3650051355361938,
      "learning_rate": 1.3560789964768839e-05,
      "loss": 9.4761,
      "step": 2227
    },
    {
      "epoch": 0.7685408761641945,
      "grad_norm": 1.2378038167953491,
      "learning_rate": 1.3522385183722791e-05,
      "loss": 9.5646,
      "step": 2228
    },
    {
      "epoch": 0.7688858226974818,
      "grad_norm": 1.2806237936019897,
      "learning_rate": 1.3484026356361018e-05,
      "loss": 9.4591,
      "step": 2229
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 1.3981233835220337,
      "learning_rate": 1.3445713531007092e-05,
      "loss": 9.5678,
      "step": 2230
    },
    {
      "epoch": 0.7695757157640566,
      "grad_norm": 1.347752571105957,
      "learning_rate": 1.3407446755926672e-05,
      "loss": 9.4931,
      "step": 2231
    },
    {
      "epoch": 0.7699206622973439,
      "grad_norm": 1.3737890720367432,
      "learning_rate": 1.3369226079327368e-05,
      "loss": 9.5142,
      "step": 2232
    },
    {
      "epoch": 0.7702656088306312,
      "grad_norm": 1.4244921207427979,
      "learning_rate": 1.3331051549358742e-05,
      "loss": 9.5011,
      "step": 2233
    },
    {
      "epoch": 0.7706105553639186,
      "grad_norm": 1.2368680238723755,
      "learning_rate": 1.3292923214112218e-05,
      "loss": 9.4674,
      "step": 2234
    },
    {
      "epoch": 0.770955501897206,
      "grad_norm": 1.3025181293487549,
      "learning_rate": 1.3254841121621037e-05,
      "loss": 9.5319,
      "step": 2235
    },
    {
      "epoch": 0.7713004484304933,
      "grad_norm": 1.3030591011047363,
      "learning_rate": 1.3216805319860131e-05,
      "loss": 9.484,
      "step": 2236
    },
    {
      "epoch": 0.7716453949637806,
      "grad_norm": 1.5651847124099731,
      "learning_rate": 1.3178815856746173e-05,
      "loss": 9.4354,
      "step": 2237
    },
    {
      "epoch": 0.7719903414970679,
      "grad_norm": 1.34073007106781,
      "learning_rate": 1.3140872780137458e-05,
      "loss": 9.4937,
      "step": 2238
    },
    {
      "epoch": 0.7723352880303553,
      "grad_norm": 1.3797394037246704,
      "learning_rate": 1.310297613783379e-05,
      "loss": 9.4472,
      "step": 2239
    },
    {
      "epoch": 0.7726802345636427,
      "grad_norm": 1.5019760131835938,
      "learning_rate": 1.3065125977576531e-05,
      "loss": 9.4539,
      "step": 2240
    },
    {
      "epoch": 0.77302518109693,
      "grad_norm": 1.3732333183288574,
      "learning_rate": 1.302732234704847e-05,
      "loss": 9.486,
      "step": 2241
    },
    {
      "epoch": 0.7733701276302173,
      "grad_norm": 1.3739562034606934,
      "learning_rate": 1.298956529387379e-05,
      "loss": 9.4486,
      "step": 2242
    },
    {
      "epoch": 0.7737150741635046,
      "grad_norm": 1.497048020362854,
      "learning_rate": 1.2951854865617946e-05,
      "loss": 9.4217,
      "step": 2243
    },
    {
      "epoch": 0.774060020696792,
      "grad_norm": 1.4363867044448853,
      "learning_rate": 1.2914191109787733e-05,
      "loss": 9.436,
      "step": 2244
    },
    {
      "epoch": 0.7744049672300793,
      "grad_norm": 1.5675686597824097,
      "learning_rate": 1.2876574073831071e-05,
      "loss": 9.4108,
      "step": 2245
    },
    {
      "epoch": 0.7747499137633667,
      "grad_norm": 1.644961953163147,
      "learning_rate": 1.2839003805137085e-05,
      "loss": 9.343,
      "step": 2246
    },
    {
      "epoch": 0.775094860296654,
      "grad_norm": 1.659659743309021,
      "learning_rate": 1.2801480351035955e-05,
      "loss": 9.3956,
      "step": 2247
    },
    {
      "epoch": 0.7754398068299414,
      "grad_norm": 1.6075783967971802,
      "learning_rate": 1.2764003758798904e-05,
      "loss": 9.3805,
      "step": 2248
    },
    {
      "epoch": 0.7757847533632287,
      "grad_norm": 1.7635935544967651,
      "learning_rate": 1.2726574075638075e-05,
      "loss": 9.3577,
      "step": 2249
    },
    {
      "epoch": 0.776129699896516,
      "grad_norm": 1.8175138235092163,
      "learning_rate": 1.268919134870657e-05,
      "loss": 9.3385,
      "step": 2250
    },
    {
      "epoch": 0.7764746464298034,
      "grad_norm": 0.8989494442939758,
      "learning_rate": 1.265185562509832e-05,
      "loss": 9.5623,
      "step": 2251
    },
    {
      "epoch": 0.7768195929630907,
      "grad_norm": 1.050932765007019,
      "learning_rate": 1.2614566951848006e-05,
      "loss": 9.4873,
      "step": 2252
    },
    {
      "epoch": 0.7771645394963781,
      "grad_norm": 1.1079734563827515,
      "learning_rate": 1.2577325375931099e-05,
      "loss": 9.5309,
      "step": 2253
    },
    {
      "epoch": 0.7775094860296654,
      "grad_norm": 1.1610482931137085,
      "learning_rate": 1.2540130944263657e-05,
      "loss": 9.4721,
      "step": 2254
    },
    {
      "epoch": 0.7778544325629527,
      "grad_norm": 1.0923736095428467,
      "learning_rate": 1.2502983703702454e-05,
      "loss": 9.4681,
      "step": 2255
    },
    {
      "epoch": 0.77819937909624,
      "grad_norm": 1.096974492073059,
      "learning_rate": 1.2465883701044712e-05,
      "loss": 9.5229,
      "step": 2256
    },
    {
      "epoch": 0.7785443256295275,
      "grad_norm": 1.0986400842666626,
      "learning_rate": 1.2428830983028211e-05,
      "loss": 9.5748,
      "step": 2257
    },
    {
      "epoch": 0.7788892721628148,
      "grad_norm": 1.1701014041900635,
      "learning_rate": 1.2391825596331114e-05,
      "loss": 9.5348,
      "step": 2258
    },
    {
      "epoch": 0.7792342186961021,
      "grad_norm": 1.0841007232666016,
      "learning_rate": 1.2354867587571989e-05,
      "loss": 9.5687,
      "step": 2259
    },
    {
      "epoch": 0.7795791652293894,
      "grad_norm": 1.0948131084442139,
      "learning_rate": 1.2317957003309727e-05,
      "loss": 9.5516,
      "step": 2260
    },
    {
      "epoch": 0.7799241117626767,
      "grad_norm": 1.104779601097107,
      "learning_rate": 1.2281093890043416e-05,
      "loss": 9.5223,
      "step": 2261
    },
    {
      "epoch": 0.7802690582959642,
      "grad_norm": 1.276573896408081,
      "learning_rate": 1.2244278294212436e-05,
      "loss": 9.5379,
      "step": 2262
    },
    {
      "epoch": 0.7806140048292515,
      "grad_norm": 1.25575590133667,
      "learning_rate": 1.2207510262196215e-05,
      "loss": 9.5144,
      "step": 2263
    },
    {
      "epoch": 0.7809589513625388,
      "grad_norm": 1.231367826461792,
      "learning_rate": 1.2170789840314318e-05,
      "loss": 9.5943,
      "step": 2264
    },
    {
      "epoch": 0.7813038978958261,
      "grad_norm": 1.329979419708252,
      "learning_rate": 1.213411707482629e-05,
      "loss": 9.458,
      "step": 2265
    },
    {
      "epoch": 0.7816488444291135,
      "grad_norm": 1.1375129222869873,
      "learning_rate": 1.2097492011931688e-05,
      "loss": 9.515,
      "step": 2266
    },
    {
      "epoch": 0.7819937909624008,
      "grad_norm": 1.3155901432037354,
      "learning_rate": 1.2060914697769926e-05,
      "loss": 9.4361,
      "step": 2267
    },
    {
      "epoch": 0.7823387374956882,
      "grad_norm": 1.2770370244979858,
      "learning_rate": 1.2024385178420289e-05,
      "loss": 9.5143,
      "step": 2268
    },
    {
      "epoch": 0.7826836840289755,
      "grad_norm": 1.18695068359375,
      "learning_rate": 1.198790349990186e-05,
      "loss": 9.494,
      "step": 2269
    },
    {
      "epoch": 0.7830286305622628,
      "grad_norm": 1.2336689233779907,
      "learning_rate": 1.1951469708173457e-05,
      "loss": 9.5422,
      "step": 2270
    },
    {
      "epoch": 0.7833735770955502,
      "grad_norm": 1.1328778266906738,
      "learning_rate": 1.1915083849133518e-05,
      "loss": 9.5457,
      "step": 2271
    },
    {
      "epoch": 0.7837185236288375,
      "grad_norm": 1.2695673704147339,
      "learning_rate": 1.1878745968620158e-05,
      "loss": 9.4853,
      "step": 2272
    },
    {
      "epoch": 0.7840634701621249,
      "grad_norm": 1.348298192024231,
      "learning_rate": 1.1842456112411044e-05,
      "loss": 9.445,
      "step": 2273
    },
    {
      "epoch": 0.7844084166954122,
      "grad_norm": 1.4354599714279175,
      "learning_rate": 1.1806214326223292e-05,
      "loss": 9.4896,
      "step": 2274
    },
    {
      "epoch": 0.7847533632286996,
      "grad_norm": 1.4190089702606201,
      "learning_rate": 1.177002065571351e-05,
      "loss": 9.4625,
      "step": 2275
    },
    {
      "epoch": 0.7850983097619869,
      "grad_norm": 1.3022481203079224,
      "learning_rate": 1.1733875146477675e-05,
      "loss": 9.568,
      "step": 2276
    },
    {
      "epoch": 0.7854432562952742,
      "grad_norm": 1.257018804550171,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 9.5252,
      "step": 2277
    },
    {
      "epoch": 0.7857882028285615,
      "grad_norm": 1.2993141412734985,
      "learning_rate": 1.1661728793908349e-05,
      "loss": 9.4571,
      "step": 2278
    },
    {
      "epoch": 0.7861331493618489,
      "grad_norm": 1.1839405298233032,
      "learning_rate": 1.162572804146323e-05,
      "loss": 9.5178,
      "step": 2279
    },
    {
      "epoch": 0.7864780958951363,
      "grad_norm": 1.3165513277053833,
      "learning_rate": 1.158977563206865e-05,
      "loss": 9.4607,
      "step": 2280
    },
    {
      "epoch": 0.7868230424284236,
      "grad_norm": 1.5338356494903564,
      "learning_rate": 1.1553871611016675e-05,
      "loss": 9.3831,
      "step": 2281
    },
    {
      "epoch": 0.7871679889617109,
      "grad_norm": 1.392738938331604,
      "learning_rate": 1.1518016023538386e-05,
      "loss": 9.4592,
      "step": 2282
    },
    {
      "epoch": 0.7875129354949982,
      "grad_norm": 1.4205015897750854,
      "learning_rate": 1.1482208914803849e-05,
      "loss": 9.4043,
      "step": 2283
    },
    {
      "epoch": 0.7878578820282857,
      "grad_norm": 1.3894630670547485,
      "learning_rate": 1.1446450329922077e-05,
      "loss": 9.4858,
      "step": 2284
    },
    {
      "epoch": 0.788202828561573,
      "grad_norm": 1.3243451118469238,
      "learning_rate": 1.141074031394091e-05,
      "loss": 9.5666,
      "step": 2285
    },
    {
      "epoch": 0.7885477750948603,
      "grad_norm": 1.3860193490982056,
      "learning_rate": 1.1375078911847048e-05,
      "loss": 9.4606,
      "step": 2286
    },
    {
      "epoch": 0.7888927216281476,
      "grad_norm": 1.3805482387542725,
      "learning_rate": 1.133946616856591e-05,
      "loss": 9.5212,
      "step": 2287
    },
    {
      "epoch": 0.7892376681614349,
      "grad_norm": 1.4170877933502197,
      "learning_rate": 1.1303902128961658e-05,
      "loss": 9.4076,
      "step": 2288
    },
    {
      "epoch": 0.7895826146947224,
      "grad_norm": 1.3475642204284668,
      "learning_rate": 1.1268386837837031e-05,
      "loss": 9.4137,
      "step": 2289
    },
    {
      "epoch": 0.7899275612280097,
      "grad_norm": 1.3942818641662598,
      "learning_rate": 1.1232920339933461e-05,
      "loss": 9.4543,
      "step": 2290
    },
    {
      "epoch": 0.790272507761297,
      "grad_norm": 1.3341909646987915,
      "learning_rate": 1.1197502679930799e-05,
      "loss": 9.4859,
      "step": 2291
    },
    {
      "epoch": 0.7906174542945843,
      "grad_norm": 1.537868618965149,
      "learning_rate": 1.1162133902447447e-05,
      "loss": 9.4129,
      "step": 2292
    },
    {
      "epoch": 0.7909624008278717,
      "grad_norm": 1.4403679370880127,
      "learning_rate": 1.1126814052040208e-05,
      "loss": 9.4867,
      "step": 2293
    },
    {
      "epoch": 0.791307347361159,
      "grad_norm": 1.6439192295074463,
      "learning_rate": 1.109154317320421e-05,
      "loss": 9.4521,
      "step": 2294
    },
    {
      "epoch": 0.7916522938944464,
      "grad_norm": 1.4819450378417969,
      "learning_rate": 1.1056321310372952e-05,
      "loss": 9.3779,
      "step": 2295
    },
    {
      "epoch": 0.7919972404277337,
      "grad_norm": 1.452518105506897,
      "learning_rate": 1.1021148507918106e-05,
      "loss": 9.4669,
      "step": 2296
    },
    {
      "epoch": 0.792342186961021,
      "grad_norm": 1.6845139265060425,
      "learning_rate": 1.0986024810149636e-05,
      "loss": 9.4524,
      "step": 2297
    },
    {
      "epoch": 0.7926871334943084,
      "grad_norm": 1.6107673645019531,
      "learning_rate": 1.0950950261315557e-05,
      "loss": 9.4202,
      "step": 2298
    },
    {
      "epoch": 0.7930320800275957,
      "grad_norm": 1.7000176906585693,
      "learning_rate": 1.091592490560202e-05,
      "loss": 9.4115,
      "step": 2299
    },
    {
      "epoch": 0.793377026560883,
      "grad_norm": 1.7845960855484009,
      "learning_rate": 1.0880948787133167e-05,
      "loss": 9.3256,
      "step": 2300
    },
    {
      "epoch": 0.7937219730941704,
      "grad_norm": 0.9648402333259583,
      "learning_rate": 1.0846021949971152e-05,
      "loss": 9.5441,
      "step": 2301
    },
    {
      "epoch": 0.7940669196274578,
      "grad_norm": 1.0515005588531494,
      "learning_rate": 1.0811144438116e-05,
      "loss": 9.5474,
      "step": 2302
    },
    {
      "epoch": 0.7944118661607451,
      "grad_norm": 0.9792898893356323,
      "learning_rate": 1.0776316295505634e-05,
      "loss": 9.535,
      "step": 2303
    },
    {
      "epoch": 0.7947568126940324,
      "grad_norm": 1.0286474227905273,
      "learning_rate": 1.0741537566015769e-05,
      "loss": 9.5707,
      "step": 2304
    },
    {
      "epoch": 0.7951017592273197,
      "grad_norm": 1.0761648416519165,
      "learning_rate": 1.0706808293459875e-05,
      "loss": 9.5962,
      "step": 2305
    },
    {
      "epoch": 0.795446705760607,
      "grad_norm": 1.0304803848266602,
      "learning_rate": 1.0672128521589125e-05,
      "loss": 9.5236,
      "step": 2306
    },
    {
      "epoch": 0.7957916522938945,
      "grad_norm": 1.0607376098632812,
      "learning_rate": 1.0637498294092291e-05,
      "loss": 9.5741,
      "step": 2307
    },
    {
      "epoch": 0.7961365988271818,
      "grad_norm": 1.1150785684585571,
      "learning_rate": 1.0602917654595796e-05,
      "loss": 9.493,
      "step": 2308
    },
    {
      "epoch": 0.7964815453604691,
      "grad_norm": 1.2552889585494995,
      "learning_rate": 1.056838664666352e-05,
      "loss": 9.5416,
      "step": 2309
    },
    {
      "epoch": 0.7968264918937564,
      "grad_norm": 1.281883955001831,
      "learning_rate": 1.0533905313796872e-05,
      "loss": 9.4997,
      "step": 2310
    },
    {
      "epoch": 0.7971714384270439,
      "grad_norm": 1.1472063064575195,
      "learning_rate": 1.0499473699434664e-05,
      "loss": 9.5945,
      "step": 2311
    },
    {
      "epoch": 0.7975163849603312,
      "grad_norm": 1.0823895931243896,
      "learning_rate": 1.0465091846953095e-05,
      "loss": 9.5251,
      "step": 2312
    },
    {
      "epoch": 0.7978613314936185,
      "grad_norm": 1.2799855470657349,
      "learning_rate": 1.0430759799665618e-05,
      "loss": 9.5017,
      "step": 2313
    },
    {
      "epoch": 0.7982062780269058,
      "grad_norm": 1.2302875518798828,
      "learning_rate": 1.0396477600823002e-05,
      "loss": 9.4611,
      "step": 2314
    },
    {
      "epoch": 0.7985512245601931,
      "grad_norm": 1.2777308225631714,
      "learning_rate": 1.0362245293613199e-05,
      "loss": 9.4718,
      "step": 2315
    },
    {
      "epoch": 0.7988961710934805,
      "grad_norm": 1.1929059028625488,
      "learning_rate": 1.0328062921161286e-05,
      "loss": 9.5054,
      "step": 2316
    },
    {
      "epoch": 0.7992411176267679,
      "grad_norm": 1.2410634756088257,
      "learning_rate": 1.029393052652946e-05,
      "loss": 9.4943,
      "step": 2317
    },
    {
      "epoch": 0.7995860641600552,
      "grad_norm": 1.2318724393844604,
      "learning_rate": 1.0259848152716955e-05,
      "loss": 9.4966,
      "step": 2318
    },
    {
      "epoch": 0.7999310106933425,
      "grad_norm": 1.230173110961914,
      "learning_rate": 1.0225815842659996e-05,
      "loss": 9.3985,
      "step": 2319
    },
    {
      "epoch": 0.8002759572266299,
      "grad_norm": 1.107927680015564,
      "learning_rate": 1.0191833639231696e-05,
      "loss": 9.4939,
      "step": 2320
    },
    {
      "epoch": 0.8006209037599172,
      "grad_norm": 1.3488962650299072,
      "learning_rate": 1.0157901585242101e-05,
      "loss": 9.464,
      "step": 2321
    },
    {
      "epoch": 0.8009658502932046,
      "grad_norm": 1.299417495727539,
      "learning_rate": 1.0124019723438038e-05,
      "loss": 9.4867,
      "step": 2322
    },
    {
      "epoch": 0.8013107968264919,
      "grad_norm": 1.3014909029006958,
      "learning_rate": 1.0090188096503139e-05,
      "loss": 9.5483,
      "step": 2323
    },
    {
      "epoch": 0.8016557433597792,
      "grad_norm": 1.4022812843322754,
      "learning_rate": 1.0056406747057695e-05,
      "loss": 9.4322,
      "step": 2324
    },
    {
      "epoch": 0.8020006898930666,
      "grad_norm": 1.2554450035095215,
      "learning_rate": 1.0022675717658742e-05,
      "loss": 9.6187,
      "step": 2325
    },
    {
      "epoch": 0.8023456364263539,
      "grad_norm": 1.3391486406326294,
      "learning_rate": 9.988995050799844e-06,
      "loss": 9.5109,
      "step": 2326
    },
    {
      "epoch": 0.8026905829596412,
      "grad_norm": 1.3482129573822021,
      "learning_rate": 9.955364788911164e-06,
      "loss": 9.5367,
      "step": 2327
    },
    {
      "epoch": 0.8030355294929286,
      "grad_norm": 1.2753441333770752,
      "learning_rate": 9.921784974359372e-06,
      "loss": 9.4852,
      "step": 2328
    },
    {
      "epoch": 0.803380476026216,
      "grad_norm": 1.3399298191070557,
      "learning_rate": 9.88825564944753e-06,
      "loss": 9.5077,
      "step": 2329
    },
    {
      "epoch": 0.8037254225595033,
      "grad_norm": 1.2231972217559814,
      "learning_rate": 9.85477685641517e-06,
      "loss": 9.4774,
      "step": 2330
    },
    {
      "epoch": 0.8040703690927906,
      "grad_norm": 1.4416093826293945,
      "learning_rate": 9.821348637438088e-06,
      "loss": 9.4557,
      "step": 2331
    },
    {
      "epoch": 0.8044153156260779,
      "grad_norm": 1.282536268234253,
      "learning_rate": 9.787971034628458e-06,
      "loss": 9.5142,
      "step": 2332
    },
    {
      "epoch": 0.8047602621593652,
      "grad_norm": 1.3632969856262207,
      "learning_rate": 9.754644090034593e-06,
      "loss": 9.4723,
      "step": 2333
    },
    {
      "epoch": 0.8051052086926527,
      "grad_norm": 1.2392255067825317,
      "learning_rate": 9.721367845641067e-06,
      "loss": 9.478,
      "step": 2334
    },
    {
      "epoch": 0.80545015522594,
      "grad_norm": 1.2770322561264038,
      "learning_rate": 9.688142343368517e-06,
      "loss": 9.437,
      "step": 2335
    },
    {
      "epoch": 0.8057951017592273,
      "grad_norm": 1.3481700420379639,
      "learning_rate": 9.654967625073708e-06,
      "loss": 9.4381,
      "step": 2336
    },
    {
      "epoch": 0.8061400482925146,
      "grad_norm": 1.5370646715164185,
      "learning_rate": 9.621843732549418e-06,
      "loss": 9.4208,
      "step": 2337
    },
    {
      "epoch": 0.806484994825802,
      "grad_norm": 1.4297363758087158,
      "learning_rate": 9.588770707524364e-06,
      "loss": 9.5563,
      "step": 2338
    },
    {
      "epoch": 0.8068299413590894,
      "grad_norm": 1.436538577079773,
      "learning_rate": 9.555748591663215e-06,
      "loss": 9.3967,
      "step": 2339
    },
    {
      "epoch": 0.8071748878923767,
      "grad_norm": 1.4317678213119507,
      "learning_rate": 9.522777426566492e-06,
      "loss": 9.5017,
      "step": 2340
    },
    {
      "epoch": 0.807519834425664,
      "grad_norm": 1.4889994859695435,
      "learning_rate": 9.48985725377055e-06,
      "loss": 9.5296,
      "step": 2341
    },
    {
      "epoch": 0.8078647809589513,
      "grad_norm": 1.5667929649353027,
      "learning_rate": 9.456988114747462e-06,
      "loss": 9.4448,
      "step": 2342
    },
    {
      "epoch": 0.8082097274922387,
      "grad_norm": 1.4982324838638306,
      "learning_rate": 9.424170050905057e-06,
      "loss": 9.4171,
      "step": 2343
    },
    {
      "epoch": 0.808554674025526,
      "grad_norm": 1.699577808380127,
      "learning_rate": 9.39140310358677e-06,
      "loss": 9.3494,
      "step": 2344
    },
    {
      "epoch": 0.8088996205588134,
      "grad_norm": 1.545198917388916,
      "learning_rate": 9.358687314071696e-06,
      "loss": 9.4545,
      "step": 2345
    },
    {
      "epoch": 0.8092445670921007,
      "grad_norm": 1.724341869354248,
      "learning_rate": 9.32602272357444e-06,
      "loss": 9.3616,
      "step": 2346
    },
    {
      "epoch": 0.8095895136253881,
      "grad_norm": 1.5193202495574951,
      "learning_rate": 9.293409373245144e-06,
      "loss": 9.4388,
      "step": 2347
    },
    {
      "epoch": 0.8099344601586754,
      "grad_norm": 1.5527263879776,
      "learning_rate": 9.260847304169345e-06,
      "loss": 9.3641,
      "step": 2348
    },
    {
      "epoch": 0.8102794066919627,
      "grad_norm": 1.701203465461731,
      "learning_rate": 9.22833655736804e-06,
      "loss": 9.3475,
      "step": 2349
    },
    {
      "epoch": 0.8106243532252501,
      "grad_norm": 1.9347988367080688,
      "learning_rate": 9.195877173797535e-06,
      "loss": 9.3568,
      "step": 2350
    },
    {
      "epoch": 0.8109692997585374,
      "grad_norm": 0.9528728127479553,
      "learning_rate": 9.163469194349423e-06,
      "loss": 9.5367,
      "step": 2351
    },
    {
      "epoch": 0.8113142462918248,
      "grad_norm": 0.9608731269836426,
      "learning_rate": 9.13111265985056e-06,
      "loss": 9.5382,
      "step": 2352
    },
    {
      "epoch": 0.8116591928251121,
      "grad_norm": 1.016716480255127,
      "learning_rate": 9.098807611062988e-06,
      "loss": 9.5135,
      "step": 2353
    },
    {
      "epoch": 0.8120041393583994,
      "grad_norm": 0.9165623784065247,
      "learning_rate": 9.066554088683894e-06,
      "loss": 9.6206,
      "step": 2354
    },
    {
      "epoch": 0.8123490858916868,
      "grad_norm": 1.1385338306427002,
      "learning_rate": 9.034352133345525e-06,
      "loss": 9.4904,
      "step": 2355
    },
    {
      "epoch": 0.8126940324249742,
      "grad_norm": 1.1017615795135498,
      "learning_rate": 9.002201785615211e-06,
      "loss": 9.5393,
      "step": 2356
    },
    {
      "epoch": 0.8130389789582615,
      "grad_norm": 1.0922075510025024,
      "learning_rate": 8.970103085995207e-06,
      "loss": 9.4825,
      "step": 2357
    },
    {
      "epoch": 0.8133839254915488,
      "grad_norm": 1.123779535293579,
      "learning_rate": 8.938056074922762e-06,
      "loss": 9.5213,
      "step": 2358
    },
    {
      "epoch": 0.8137288720248361,
      "grad_norm": 1.0766289234161377,
      "learning_rate": 8.906060792769982e-06,
      "loss": 9.5497,
      "step": 2359
    },
    {
      "epoch": 0.8140738185581234,
      "grad_norm": 1.142089605331421,
      "learning_rate": 8.874117279843813e-06,
      "loss": 9.5189,
      "step": 2360
    },
    {
      "epoch": 0.8144187650914109,
      "grad_norm": 1.2554246187210083,
      "learning_rate": 8.842225576385964e-06,
      "loss": 9.5115,
      "step": 2361
    },
    {
      "epoch": 0.8147637116246982,
      "grad_norm": 1.347861886024475,
      "learning_rate": 8.810385722572901e-06,
      "loss": 9.5216,
      "step": 2362
    },
    {
      "epoch": 0.8151086581579855,
      "grad_norm": 1.1705132722854614,
      "learning_rate": 8.778597758515766e-06,
      "loss": 9.4948,
      "step": 2363
    },
    {
      "epoch": 0.8154536046912728,
      "grad_norm": 1.1871190071105957,
      "learning_rate": 8.746861724260308e-06,
      "loss": 9.5249,
      "step": 2364
    },
    {
      "epoch": 0.8157985512245602,
      "grad_norm": 1.1689561605453491,
      "learning_rate": 8.7151776597869e-06,
      "loss": 9.5313,
      "step": 2365
    },
    {
      "epoch": 0.8161434977578476,
      "grad_norm": 1.2277796268463135,
      "learning_rate": 8.683545605010373e-06,
      "loss": 9.5339,
      "step": 2366
    },
    {
      "epoch": 0.8164884442911349,
      "grad_norm": 1.2542256116867065,
      "learning_rate": 8.651965599780143e-06,
      "loss": 9.493,
      "step": 2367
    },
    {
      "epoch": 0.8168333908244222,
      "grad_norm": 1.2161953449249268,
      "learning_rate": 8.62043768387995e-06,
      "loss": 9.5418,
      "step": 2368
    },
    {
      "epoch": 0.8171783373577095,
      "grad_norm": 1.3967869281768799,
      "learning_rate": 8.588961897027986e-06,
      "loss": 9.5238,
      "step": 2369
    },
    {
      "epoch": 0.8175232838909969,
      "grad_norm": 1.2551013231277466,
      "learning_rate": 8.55753827887672e-06,
      "loss": 9.4706,
      "step": 2370
    },
    {
      "epoch": 0.8178682304242842,
      "grad_norm": 1.3102620840072632,
      "learning_rate": 8.526166869012936e-06,
      "loss": 9.4533,
      "step": 2371
    },
    {
      "epoch": 0.8182131769575716,
      "grad_norm": 1.2320739030838013,
      "learning_rate": 8.494847706957642e-06,
      "loss": 9.4959,
      "step": 2372
    },
    {
      "epoch": 0.8185581234908589,
      "grad_norm": 1.4043902158737183,
      "learning_rate": 8.463580832165996e-06,
      "loss": 9.5451,
      "step": 2373
    },
    {
      "epoch": 0.8189030700241463,
      "grad_norm": 1.2317163944244385,
      "learning_rate": 8.432366284027343e-06,
      "loss": 9.4723,
      "step": 2374
    },
    {
      "epoch": 0.8192480165574336,
      "grad_norm": 1.389200210571289,
      "learning_rate": 8.401204101865046e-06,
      "loss": 9.5072,
      "step": 2375
    },
    {
      "epoch": 0.8195929630907209,
      "grad_norm": 1.277854561805725,
      "learning_rate": 8.370094324936545e-06,
      "loss": 9.4576,
      "step": 2376
    },
    {
      "epoch": 0.8199379096240083,
      "grad_norm": 1.2297642230987549,
      "learning_rate": 8.33903699243322e-06,
      "loss": 9.5131,
      "step": 2377
    },
    {
      "epoch": 0.8202828561572956,
      "grad_norm": 1.2601749897003174,
      "learning_rate": 8.308032143480427e-06,
      "loss": 9.46,
      "step": 2378
    },
    {
      "epoch": 0.820627802690583,
      "grad_norm": 1.3500021696090698,
      "learning_rate": 8.277079817137345e-06,
      "loss": 9.4438,
      "step": 2379
    },
    {
      "epoch": 0.8209727492238703,
      "grad_norm": 1.278239130973816,
      "learning_rate": 8.24618005239708e-06,
      "loss": 9.5169,
      "step": 2380
    },
    {
      "epoch": 0.8213176957571576,
      "grad_norm": 1.4504845142364502,
      "learning_rate": 8.215332888186417e-06,
      "loss": 9.4441,
      "step": 2381
    },
    {
      "epoch": 0.821662642290445,
      "grad_norm": 1.3500051498413086,
      "learning_rate": 8.184538363365957e-06,
      "loss": 9.4842,
      "step": 2382
    },
    {
      "epoch": 0.8220075888237324,
      "grad_norm": 1.5992237329483032,
      "learning_rate": 8.153796516729928e-06,
      "loss": 9.4223,
      "step": 2383
    },
    {
      "epoch": 0.8223525353570197,
      "grad_norm": 1.363759994506836,
      "learning_rate": 8.123107387006229e-06,
      "loss": 9.4813,
      "step": 2384
    },
    {
      "epoch": 0.822697481890307,
      "grad_norm": 1.442180871963501,
      "learning_rate": 8.092471012856362e-06,
      "loss": 9.4769,
      "step": 2385
    },
    {
      "epoch": 0.8230424284235943,
      "grad_norm": 1.4553319215774536,
      "learning_rate": 8.061887432875304e-06,
      "loss": 9.4806,
      "step": 2386
    },
    {
      "epoch": 0.8233873749568816,
      "grad_norm": 1.4236608743667603,
      "learning_rate": 8.031356685591612e-06,
      "loss": 9.4485,
      "step": 2387
    },
    {
      "epoch": 0.8237323214901691,
      "grad_norm": 1.4737285375595093,
      "learning_rate": 8.00087880946721e-06,
      "loss": 9.4346,
      "step": 2388
    },
    {
      "epoch": 0.8240772680234564,
      "grad_norm": 1.4963711500167847,
      "learning_rate": 7.970453842897462e-06,
      "loss": 9.4678,
      "step": 2389
    },
    {
      "epoch": 0.8244222145567437,
      "grad_norm": 1.4410032033920288,
      "learning_rate": 7.940081824211049e-06,
      "loss": 9.4932,
      "step": 2390
    },
    {
      "epoch": 0.824767161090031,
      "grad_norm": 1.51363205909729,
      "learning_rate": 7.909762791669972e-06,
      "loss": 9.4573,
      "step": 2391
    },
    {
      "epoch": 0.8251121076233184,
      "grad_norm": 1.5144532918930054,
      "learning_rate": 7.879496783469454e-06,
      "loss": 9.3881,
      "step": 2392
    },
    {
      "epoch": 0.8254570541566058,
      "grad_norm": 1.5119389295578003,
      "learning_rate": 7.849283837737941e-06,
      "loss": 9.4617,
      "step": 2393
    },
    {
      "epoch": 0.8258020006898931,
      "grad_norm": 1.6774712800979614,
      "learning_rate": 7.819123992537042e-06,
      "loss": 9.4475,
      "step": 2394
    },
    {
      "epoch": 0.8261469472231804,
      "grad_norm": 1.540063500404358,
      "learning_rate": 7.789017285861439e-06,
      "loss": 9.4618,
      "step": 2395
    },
    {
      "epoch": 0.8264918937564677,
      "grad_norm": 1.6762762069702148,
      "learning_rate": 7.758963755638916e-06,
      "loss": 9.436,
      "step": 2396
    },
    {
      "epoch": 0.8268368402897551,
      "grad_norm": 1.5456700325012207,
      "learning_rate": 7.728963439730203e-06,
      "loss": 9.3776,
      "step": 2397
    },
    {
      "epoch": 0.8271817868230424,
      "grad_norm": 1.513611912727356,
      "learning_rate": 7.699016375929057e-06,
      "loss": 9.4647,
      "step": 2398
    },
    {
      "epoch": 0.8275267333563298,
      "grad_norm": 1.805281400680542,
      "learning_rate": 7.669122601962098e-06,
      "loss": 9.3213,
      "step": 2399
    },
    {
      "epoch": 0.8278716798896171,
      "grad_norm": 1.686733603477478,
      "learning_rate": 7.639282155488836e-06,
      "loss": 9.3407,
      "step": 2400
    },
    {
      "epoch": 0.8282166264229045,
      "grad_norm": 0.8331398963928223,
      "learning_rate": 7.609495074101614e-06,
      "loss": 9.5749,
      "step": 2401
    },
    {
      "epoch": 0.8285615729561918,
      "grad_norm": 0.9920496940612793,
      "learning_rate": 7.579761395325536e-06,
      "loss": 9.5205,
      "step": 2402
    },
    {
      "epoch": 0.8289065194894791,
      "grad_norm": 1.0967357158660889,
      "learning_rate": 7.550081156618399e-06,
      "loss": 9.4806,
      "step": 2403
    },
    {
      "epoch": 0.8292514660227664,
      "grad_norm": 1.0771796703338623,
      "learning_rate": 7.520454395370724e-06,
      "loss": 9.5243,
      "step": 2404
    },
    {
      "epoch": 0.8295964125560538,
      "grad_norm": 1.102107286453247,
      "learning_rate": 7.490881148905654e-06,
      "loss": 9.4947,
      "step": 2405
    },
    {
      "epoch": 0.8299413590893412,
      "grad_norm": 0.9728934168815613,
      "learning_rate": 7.461361454478871e-06,
      "loss": 9.5951,
      "step": 2406
    },
    {
      "epoch": 0.8302863056226285,
      "grad_norm": 1.1681954860687256,
      "learning_rate": 7.43189534927865e-06,
      "loss": 9.5276,
      "step": 2407
    },
    {
      "epoch": 0.8306312521559158,
      "grad_norm": 1.1862436532974243,
      "learning_rate": 7.402482870425725e-06,
      "loss": 9.524,
      "step": 2408
    },
    {
      "epoch": 0.8309761986892031,
      "grad_norm": 1.2328039407730103,
      "learning_rate": 7.373124054973296e-06,
      "loss": 9.5313,
      "step": 2409
    },
    {
      "epoch": 0.8313211452224906,
      "grad_norm": 1.20083749294281,
      "learning_rate": 7.343818939906916e-06,
      "loss": 9.5427,
      "step": 2410
    },
    {
      "epoch": 0.8316660917557779,
      "grad_norm": 1.2234563827514648,
      "learning_rate": 7.314567562144542e-06,
      "loss": 9.52,
      "step": 2411
    },
    {
      "epoch": 0.8320110382890652,
      "grad_norm": 1.3493642807006836,
      "learning_rate": 7.285369958536375e-06,
      "loss": 9.4352,
      "step": 2412
    },
    {
      "epoch": 0.8323559848223525,
      "grad_norm": 1.0981404781341553,
      "learning_rate": 7.2562261658649325e-06,
      "loss": 9.5559,
      "step": 2413
    },
    {
      "epoch": 0.8327009313556398,
      "grad_norm": 1.130433201789856,
      "learning_rate": 7.227136220844883e-06,
      "loss": 9.5719,
      "step": 2414
    },
    {
      "epoch": 0.8330458778889273,
      "grad_norm": 1.3468449115753174,
      "learning_rate": 7.1981001601231276e-06,
      "loss": 9.4805,
      "step": 2415
    },
    {
      "epoch": 0.8333908244222146,
      "grad_norm": 1.3730062246322632,
      "learning_rate": 7.169118020278615e-06,
      "loss": 9.4373,
      "step": 2416
    },
    {
      "epoch": 0.8337357709555019,
      "grad_norm": 1.30331289768219,
      "learning_rate": 7.140189837822409e-06,
      "loss": 9.4876,
      "step": 2417
    },
    {
      "epoch": 0.8340807174887892,
      "grad_norm": 1.2246065139770508,
      "learning_rate": 7.111315649197603e-06,
      "loss": 9.523,
      "step": 2418
    },
    {
      "epoch": 0.8344256640220766,
      "grad_norm": 1.301967740058899,
      "learning_rate": 7.082495490779229e-06,
      "loss": 9.4755,
      "step": 2419
    },
    {
      "epoch": 0.834770610555364,
      "grad_norm": 1.2529209852218628,
      "learning_rate": 7.053729398874298e-06,
      "loss": 9.5024,
      "step": 2420
    },
    {
      "epoch": 0.8351155570886513,
      "grad_norm": 1.276577115058899,
      "learning_rate": 7.025017409721657e-06,
      "loss": 9.6069,
      "step": 2421
    },
    {
      "epoch": 0.8354605036219386,
      "grad_norm": 1.323262095451355,
      "learning_rate": 6.996359559492083e-06,
      "loss": 9.4286,
      "step": 2422
    },
    {
      "epoch": 0.8358054501552259,
      "grad_norm": 1.3153644800186157,
      "learning_rate": 6.967755884288046e-06,
      "loss": 9.5094,
      "step": 2423
    },
    {
      "epoch": 0.8361503966885133,
      "grad_norm": 1.23281991481781,
      "learning_rate": 6.939206420143857e-06,
      "loss": 9.4647,
      "step": 2424
    },
    {
      "epoch": 0.8364953432218006,
      "grad_norm": 1.269282341003418,
      "learning_rate": 6.910711203025455e-06,
      "loss": 9.5021,
      "step": 2425
    },
    {
      "epoch": 0.836840289755088,
      "grad_norm": 1.2297919988632202,
      "learning_rate": 6.882270268830498e-06,
      "loss": 9.4929,
      "step": 2426
    },
    {
      "epoch": 0.8371852362883753,
      "grad_norm": 1.2677528858184814,
      "learning_rate": 6.853883653388249e-06,
      "loss": 9.4232,
      "step": 2427
    },
    {
      "epoch": 0.8375301828216627,
      "grad_norm": 1.2528631687164307,
      "learning_rate": 6.825551392459511e-06,
      "loss": 9.4827,
      "step": 2428
    },
    {
      "epoch": 0.83787512935495,
      "grad_norm": 1.3022857904434204,
      "learning_rate": 6.797273521736641e-06,
      "loss": 9.4533,
      "step": 2429
    },
    {
      "epoch": 0.8382200758882373,
      "grad_norm": 1.4493368864059448,
      "learning_rate": 6.769050076843469e-06,
      "loss": 9.4251,
      "step": 2430
    },
    {
      "epoch": 0.8385650224215246,
      "grad_norm": 1.4479137659072876,
      "learning_rate": 6.740881093335278e-06,
      "loss": 9.4296,
      "step": 2431
    },
    {
      "epoch": 0.838909968954812,
      "grad_norm": 1.3461737632751465,
      "learning_rate": 6.712766606698689e-06,
      "loss": 9.5194,
      "step": 2432
    },
    {
      "epoch": 0.8392549154880994,
      "grad_norm": 1.3478329181671143,
      "learning_rate": 6.684706652351741e-06,
      "loss": 9.4739,
      "step": 2433
    },
    {
      "epoch": 0.8395998620213867,
      "grad_norm": 1.5345542430877686,
      "learning_rate": 6.656701265643711e-06,
      "loss": 9.4211,
      "step": 2434
    },
    {
      "epoch": 0.839944808554674,
      "grad_norm": 1.2493342161178589,
      "learning_rate": 6.628750481855167e-06,
      "loss": 9.4707,
      "step": 2435
    },
    {
      "epoch": 0.8402897550879613,
      "grad_norm": 1.3178075551986694,
      "learning_rate": 6.6008543361978814e-06,
      "loss": 9.4852,
      "step": 2436
    },
    {
      "epoch": 0.8406347016212488,
      "grad_norm": 1.4194328784942627,
      "learning_rate": 6.5730128638148095e-06,
      "loss": 9.4392,
      "step": 2437
    },
    {
      "epoch": 0.8409796481545361,
      "grad_norm": 1.4278353452682495,
      "learning_rate": 6.545226099779994e-06,
      "loss": 9.4369,
      "step": 2438
    },
    {
      "epoch": 0.8413245946878234,
      "grad_norm": 1.4180926084518433,
      "learning_rate": 6.517494079098585e-06,
      "loss": 9.4777,
      "step": 2439
    },
    {
      "epoch": 0.8416695412211107,
      "grad_norm": 1.3950679302215576,
      "learning_rate": 6.489816836706786e-06,
      "loss": 9.4472,
      "step": 2440
    },
    {
      "epoch": 0.842014487754398,
      "grad_norm": 1.5117658376693726,
      "learning_rate": 6.462194407471733e-06,
      "loss": 9.4362,
      "step": 2441
    },
    {
      "epoch": 0.8423594342876854,
      "grad_norm": 1.452319622039795,
      "learning_rate": 6.43462682619157e-06,
      "loss": 9.443,
      "step": 2442
    },
    {
      "epoch": 0.8427043808209728,
      "grad_norm": 1.3729323148727417,
      "learning_rate": 6.407114127595304e-06,
      "loss": 9.4496,
      "step": 2443
    },
    {
      "epoch": 0.8430493273542601,
      "grad_norm": 1.5410650968551636,
      "learning_rate": 6.379656346342844e-06,
      "loss": 9.4024,
      "step": 2444
    },
    {
      "epoch": 0.8433942738875474,
      "grad_norm": 1.4286586046218872,
      "learning_rate": 6.352253517024859e-06,
      "loss": 9.4946,
      "step": 2445
    },
    {
      "epoch": 0.8437392204208348,
      "grad_norm": 1.544579267501831,
      "learning_rate": 6.324905674162846e-06,
      "loss": 9.3807,
      "step": 2446
    },
    {
      "epoch": 0.8440841669541221,
      "grad_norm": 1.5438258647918701,
      "learning_rate": 6.297612852208978e-06,
      "loss": 9.4389,
      "step": 2447
    },
    {
      "epoch": 0.8444291134874095,
      "grad_norm": 1.5225231647491455,
      "learning_rate": 6.2703750855461654e-06,
      "loss": 9.3885,
      "step": 2448
    },
    {
      "epoch": 0.8447740600206968,
      "grad_norm": 1.6549159288406372,
      "learning_rate": 6.24319240848793e-06,
      "loss": 9.4022,
      "step": 2449
    },
    {
      "epoch": 0.8451190065539841,
      "grad_norm": 1.7702971696853638,
      "learning_rate": 6.216064855278414e-06,
      "loss": 9.3768,
      "step": 2450
    },
    {
      "epoch": 0.8454639530872715,
      "grad_norm": 0.8622440695762634,
      "learning_rate": 6.188992460092286e-06,
      "loss": 9.5452,
      "step": 2451
    },
    {
      "epoch": 0.8458088996205588,
      "grad_norm": 1.0240706205368042,
      "learning_rate": 6.161975257034741e-06,
      "loss": 9.5128,
      "step": 2452
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 1.021040678024292,
      "learning_rate": 6.135013280141477e-06,
      "loss": 9.5117,
      "step": 2453
    },
    {
      "epoch": 0.8464987926871335,
      "grad_norm": 1.160355806350708,
      "learning_rate": 6.108106563378557e-06,
      "loss": 9.5123,
      "step": 2454
    },
    {
      "epoch": 0.8468437392204209,
      "grad_norm": 1.145014762878418,
      "learning_rate": 6.081255140642483e-06,
      "loss": 9.4774,
      "step": 2455
    },
    {
      "epoch": 0.8471886857537082,
      "grad_norm": 1.0997885465621948,
      "learning_rate": 6.054459045760053e-06,
      "loss": 9.531,
      "step": 2456
    },
    {
      "epoch": 0.8475336322869955,
      "grad_norm": 1.057318925857544,
      "learning_rate": 6.027718312488423e-06,
      "loss": 9.5878,
      "step": 2457
    },
    {
      "epoch": 0.8478785788202828,
      "grad_norm": 1.2089539766311646,
      "learning_rate": 6.001032974514947e-06,
      "loss": 9.4941,
      "step": 2458
    },
    {
      "epoch": 0.8482235253535702,
      "grad_norm": 1.2328418493270874,
      "learning_rate": 5.974403065457235e-06,
      "loss": 9.4992,
      "step": 2459
    },
    {
      "epoch": 0.8485684718868576,
      "grad_norm": 1.1309876441955566,
      "learning_rate": 5.947828618863027e-06,
      "loss": 9.544,
      "step": 2460
    },
    {
      "epoch": 0.8489134184201449,
      "grad_norm": 1.2640806436538696,
      "learning_rate": 5.921309668210234e-06,
      "loss": 9.4976,
      "step": 2461
    },
    {
      "epoch": 0.8492583649534322,
      "grad_norm": 1.1987804174423218,
      "learning_rate": 5.894846246906843e-06,
      "loss": 9.5373,
      "step": 2462
    },
    {
      "epoch": 0.8496033114867195,
      "grad_norm": 1.2217447757720947,
      "learning_rate": 5.868438388290854e-06,
      "loss": 9.4895,
      "step": 2463
    },
    {
      "epoch": 0.849948258020007,
      "grad_norm": 1.1472078561782837,
      "learning_rate": 5.8420861256303415e-06,
      "loss": 9.5564,
      "step": 2464
    },
    {
      "epoch": 0.8502932045532943,
      "grad_norm": 1.3079042434692383,
      "learning_rate": 5.815789492123258e-06,
      "loss": 9.4928,
      "step": 2465
    },
    {
      "epoch": 0.8506381510865816,
      "grad_norm": 1.388771891593933,
      "learning_rate": 5.7895485208975365e-06,
      "loss": 9.5433,
      "step": 2466
    },
    {
      "epoch": 0.8509830976198689,
      "grad_norm": 1.3080216646194458,
      "learning_rate": 5.76336324501095e-06,
      "loss": 9.4568,
      "step": 2467
    },
    {
      "epoch": 0.8513280441531562,
      "grad_norm": 1.1682881116867065,
      "learning_rate": 5.737233697451145e-06,
      "loss": 9.5277,
      "step": 2468
    },
    {
      "epoch": 0.8516729906864436,
      "grad_norm": 1.3328065872192383,
      "learning_rate": 5.7111599111355215e-06,
      "loss": 9.4417,
      "step": 2469
    },
    {
      "epoch": 0.852017937219731,
      "grad_norm": 1.3096672296524048,
      "learning_rate": 5.685141918911257e-06,
      "loss": 9.4382,
      "step": 2470
    },
    {
      "epoch": 0.8523628837530183,
      "grad_norm": 1.3736581802368164,
      "learning_rate": 5.659179753555244e-06,
      "loss": 9.4432,
      "step": 2471
    },
    {
      "epoch": 0.8527078302863056,
      "grad_norm": 1.2530180215835571,
      "learning_rate": 5.633273447774046e-06,
      "loss": 9.4777,
      "step": 2472
    },
    {
      "epoch": 0.853052776819593,
      "grad_norm": 1.279069423675537,
      "learning_rate": 5.607423034203829e-06,
      "loss": 9.538,
      "step": 2473
    },
    {
      "epoch": 0.8533977233528803,
      "grad_norm": 1.3018614053726196,
      "learning_rate": 5.581628545410372e-06,
      "loss": 9.4599,
      "step": 2474
    },
    {
      "epoch": 0.8537426698861676,
      "grad_norm": 1.3087834119796753,
      "learning_rate": 5.555890013889009e-06,
      "loss": 9.503,
      "step": 2475
    },
    {
      "epoch": 0.854087616419455,
      "grad_norm": 1.349788784980774,
      "learning_rate": 5.530207472064552e-06,
      "loss": 9.5043,
      "step": 2476
    },
    {
      "epoch": 0.8544325629527423,
      "grad_norm": 1.2353582382202148,
      "learning_rate": 5.504580952291294e-06,
      "loss": 9.5006,
      "step": 2477
    },
    {
      "epoch": 0.8547775094860297,
      "grad_norm": 1.3644788265228271,
      "learning_rate": 5.479010486852959e-06,
      "loss": 9.4729,
      "step": 2478
    },
    {
      "epoch": 0.855122456019317,
      "grad_norm": 1.2602345943450928,
      "learning_rate": 5.453496107962658e-06,
      "loss": 9.5259,
      "step": 2479
    },
    {
      "epoch": 0.8554674025526043,
      "grad_norm": 1.4060320854187012,
      "learning_rate": 5.428037847762813e-06,
      "loss": 9.5535,
      "step": 2480
    },
    {
      "epoch": 0.8558123490858917,
      "grad_norm": 1.5353844165802002,
      "learning_rate": 5.40263573832519e-06,
      "loss": 9.4855,
      "step": 2481
    },
    {
      "epoch": 0.8561572956191791,
      "grad_norm": 1.502706527709961,
      "learning_rate": 5.377289811650782e-06,
      "loss": 9.3855,
      "step": 2482
    },
    {
      "epoch": 0.8565022421524664,
      "grad_norm": 1.2772150039672852,
      "learning_rate": 5.35200009966983e-06,
      "loss": 9.5104,
      "step": 2483
    },
    {
      "epoch": 0.8568471886857537,
      "grad_norm": 1.388257622718811,
      "learning_rate": 5.326766634241748e-06,
      "loss": 9.4841,
      "step": 2484
    },
    {
      "epoch": 0.857192135219041,
      "grad_norm": 1.3631497621536255,
      "learning_rate": 5.301589447155092e-06,
      "loss": 9.4448,
      "step": 2485
    },
    {
      "epoch": 0.8575370817523283,
      "grad_norm": 1.4563467502593994,
      "learning_rate": 5.27646857012753e-06,
      "loss": 9.4469,
      "step": 2486
    },
    {
      "epoch": 0.8578820282856158,
      "grad_norm": 1.3632886409759521,
      "learning_rate": 5.251404034805768e-06,
      "loss": 9.4865,
      "step": 2487
    },
    {
      "epoch": 0.8582269748189031,
      "grad_norm": 1.3538519144058228,
      "learning_rate": 5.226395872765555e-06,
      "loss": 9.4439,
      "step": 2488
    },
    {
      "epoch": 0.8585719213521904,
      "grad_norm": 1.3481531143188477,
      "learning_rate": 5.201444115511605e-06,
      "loss": 9.52,
      "step": 2489
    },
    {
      "epoch": 0.8589168678854777,
      "grad_norm": 1.370755910873413,
      "learning_rate": 5.176548794477598e-06,
      "loss": 9.4518,
      "step": 2490
    },
    {
      "epoch": 0.8592618144187651,
      "grad_norm": 1.4279519319534302,
      "learning_rate": 5.151709941026078e-06,
      "loss": 9.4622,
      "step": 2491
    },
    {
      "epoch": 0.8596067609520525,
      "grad_norm": 1.441394329071045,
      "learning_rate": 5.126927586448516e-06,
      "loss": 9.4812,
      "step": 2492
    },
    {
      "epoch": 0.8599517074853398,
      "grad_norm": 1.4180115461349487,
      "learning_rate": 5.1022017619651415e-06,
      "loss": 9.4861,
      "step": 2493
    },
    {
      "epoch": 0.8602966540186271,
      "grad_norm": 1.5045257806777954,
      "learning_rate": 5.077532498725013e-06,
      "loss": 9.4093,
      "step": 2494
    },
    {
      "epoch": 0.8606416005519144,
      "grad_norm": 1.4450771808624268,
      "learning_rate": 5.052919827805891e-06,
      "loss": 9.4047,
      "step": 2495
    },
    {
      "epoch": 0.8609865470852018,
      "grad_norm": 1.6171228885650635,
      "learning_rate": 5.02836378021429e-06,
      "loss": 9.3855,
      "step": 2496
    },
    {
      "epoch": 0.8613314936184892,
      "grad_norm": 1.4856090545654297,
      "learning_rate": 5.003864386885376e-06,
      "loss": 9.3877,
      "step": 2497
    },
    {
      "epoch": 0.8616764401517765,
      "grad_norm": 1.6896843910217285,
      "learning_rate": 4.979421678682905e-06,
      "loss": 9.4596,
      "step": 2498
    },
    {
      "epoch": 0.8620213866850638,
      "grad_norm": 1.7836569547653198,
      "learning_rate": 4.95503568639929e-06,
      "loss": 9.325,
      "step": 2499
    },
    {
      "epoch": 0.8623663332183512,
      "grad_norm": 1.8124544620513916,
      "learning_rate": 4.930706440755445e-06,
      "loss": 9.3828,
      "step": 2500
    },
    {
      "epoch": 0.8627112797516385,
      "grad_norm": 0.9383218884468079,
      "learning_rate": 4.9064339724008144e-06,
      "loss": 9.5376,
      "step": 2501
    },
    {
      "epoch": 0.8630562262849258,
      "grad_norm": 1.0528539419174194,
      "learning_rate": 4.8822183119133e-06,
      "loss": 9.5191,
      "step": 2502
    },
    {
      "epoch": 0.8634011728182132,
      "grad_norm": 1.0615488290786743,
      "learning_rate": 4.858059489799266e-06,
      "loss": 9.5424,
      "step": 2503
    },
    {
      "epoch": 0.8637461193515005,
      "grad_norm": 1.0702091455459595,
      "learning_rate": 4.833957536493439e-06,
      "loss": 9.5417,
      "step": 2504
    },
    {
      "epoch": 0.8640910658847879,
      "grad_norm": 1.1862672567367554,
      "learning_rate": 4.809912482358936e-06,
      "loss": 9.5302,
      "step": 2505
    },
    {
      "epoch": 0.8644360124180752,
      "grad_norm": 1.08174729347229,
      "learning_rate": 4.785924357687166e-06,
      "loss": 9.5423,
      "step": 2506
    },
    {
      "epoch": 0.8647809589513625,
      "grad_norm": 1.0673178434371948,
      "learning_rate": 4.761993192697844e-06,
      "loss": 9.5344,
      "step": 2507
    },
    {
      "epoch": 0.8651259054846498,
      "grad_norm": 1.2792860269546509,
      "learning_rate": 4.7381190175389275e-06,
      "loss": 9.5388,
      "step": 2508
    },
    {
      "epoch": 0.8654708520179372,
      "grad_norm": 1.148232340812683,
      "learning_rate": 4.714301862286541e-06,
      "loss": 9.46,
      "step": 2509
    },
    {
      "epoch": 0.8658157985512246,
      "grad_norm": 1.2341820001602173,
      "learning_rate": 4.6905417569450275e-06,
      "loss": 9.5178,
      "step": 2510
    },
    {
      "epoch": 0.8661607450845119,
      "grad_norm": 1.114518642425537,
      "learning_rate": 4.666838731446821e-06,
      "loss": 9.5216,
      "step": 2511
    },
    {
      "epoch": 0.8665056916177992,
      "grad_norm": 1.1133232116699219,
      "learning_rate": 4.643192815652469e-06,
      "loss": 9.49,
      "step": 2512
    },
    {
      "epoch": 0.8668506381510865,
      "grad_norm": 1.1808667182922363,
      "learning_rate": 4.619604039350572e-06,
      "loss": 9.5987,
      "step": 2513
    },
    {
      "epoch": 0.867195584684374,
      "grad_norm": 1.2320611476898193,
      "learning_rate": 4.596072432257748e-06,
      "loss": 9.4902,
      "step": 2514
    },
    {
      "epoch": 0.8675405312176613,
      "grad_norm": 1.1610668897628784,
      "learning_rate": 4.572598024018571e-06,
      "loss": 9.5405,
      "step": 2515
    },
    {
      "epoch": 0.8678854777509486,
      "grad_norm": 1.1383475065231323,
      "learning_rate": 4.549180844205603e-06,
      "loss": 9.5299,
      "step": 2516
    },
    {
      "epoch": 0.8682304242842359,
      "grad_norm": 1.1785327196121216,
      "learning_rate": 4.525820922319257e-06,
      "loss": 9.5377,
      "step": 2517
    },
    {
      "epoch": 0.8685753708175232,
      "grad_norm": 1.3649024963378906,
      "learning_rate": 4.502518287787855e-06,
      "loss": 9.4778,
      "step": 2518
    },
    {
      "epoch": 0.8689203173508107,
      "grad_norm": 1.3480148315429688,
      "learning_rate": 4.4792729699675294e-06,
      "loss": 9.4929,
      "step": 2519
    },
    {
      "epoch": 0.869265263884098,
      "grad_norm": 1.2999801635742188,
      "learning_rate": 4.456084998142224e-06,
      "loss": 9.5344,
      "step": 2520
    },
    {
      "epoch": 0.8696102104173853,
      "grad_norm": 1.2308170795440674,
      "learning_rate": 4.43295440152362e-06,
      "loss": 9.478,
      "step": 2521
    },
    {
      "epoch": 0.8699551569506726,
      "grad_norm": 1.3496099710464478,
      "learning_rate": 4.409881209251121e-06,
      "loss": 9.49,
      "step": 2522
    },
    {
      "epoch": 0.87030010348396,
      "grad_norm": 1.218358039855957,
      "learning_rate": 4.386865450391836e-06,
      "loss": 9.5208,
      "step": 2523
    },
    {
      "epoch": 0.8706450500172473,
      "grad_norm": 1.460325002670288,
      "learning_rate": 4.3639071539404775e-06,
      "loss": 9.4209,
      "step": 2524
    },
    {
      "epoch": 0.8709899965505347,
      "grad_norm": 1.4746004343032837,
      "learning_rate": 4.341006348819421e-06,
      "loss": 9.5211,
      "step": 2525
    },
    {
      "epoch": 0.871334943083822,
      "grad_norm": 1.2816509008407593,
      "learning_rate": 4.318163063878561e-06,
      "loss": 9.4756,
      "step": 2526
    },
    {
      "epoch": 0.8716798896171093,
      "grad_norm": 1.3259309530258179,
      "learning_rate": 4.295377327895389e-06,
      "loss": 9.5018,
      "step": 2527
    },
    {
      "epoch": 0.8720248361503967,
      "grad_norm": 1.4273511171340942,
      "learning_rate": 4.272649169574849e-06,
      "loss": 9.4362,
      "step": 2528
    },
    {
      "epoch": 0.872369782683684,
      "grad_norm": 1.3017966747283936,
      "learning_rate": 4.249978617549361e-06,
      "loss": 9.4516,
      "step": 2529
    },
    {
      "epoch": 0.8727147292169714,
      "grad_norm": 1.2315176725387573,
      "learning_rate": 4.227365700378799e-06,
      "loss": 9.5207,
      "step": 2530
    },
    {
      "epoch": 0.8730596757502587,
      "grad_norm": 1.395295262336731,
      "learning_rate": 4.204810446550394e-06,
      "loss": 9.4892,
      "step": 2531
    },
    {
      "epoch": 0.8734046222835461,
      "grad_norm": 1.3410911560058594,
      "learning_rate": 4.182312884478767e-06,
      "loss": 9.5657,
      "step": 2532
    },
    {
      "epoch": 0.8737495688168334,
      "grad_norm": 1.4816246032714844,
      "learning_rate": 4.159873042505813e-06,
      "loss": 9.4206,
      "step": 2533
    },
    {
      "epoch": 0.8740945153501207,
      "grad_norm": 1.3325238227844238,
      "learning_rate": 4.137490948900785e-06,
      "loss": 9.484,
      "step": 2534
    },
    {
      "epoch": 0.874439461883408,
      "grad_norm": 1.395811676979065,
      "learning_rate": 4.115166631860113e-06,
      "loss": 9.5019,
      "step": 2535
    },
    {
      "epoch": 0.8747844084166954,
      "grad_norm": 1.3411831855773926,
      "learning_rate": 4.092900119507498e-06,
      "loss": 9.5089,
      "step": 2536
    },
    {
      "epoch": 0.8751293549499828,
      "grad_norm": 1.315796971321106,
      "learning_rate": 4.07069143989377e-06,
      "loss": 9.471,
      "step": 2537
    },
    {
      "epoch": 0.8754743014832701,
      "grad_norm": 1.3734862804412842,
      "learning_rate": 4.048540620996932e-06,
      "loss": 9.545,
      "step": 2538
    },
    {
      "epoch": 0.8758192480165574,
      "grad_norm": 1.5111862421035767,
      "learning_rate": 4.026447690722113e-06,
      "loss": 9.402,
      "step": 2539
    },
    {
      "epoch": 0.8761641945498447,
      "grad_norm": 1.3808101415634155,
      "learning_rate": 4.00441267690147e-06,
      "loss": 9.4543,
      "step": 2540
    },
    {
      "epoch": 0.8765091410831322,
      "grad_norm": 1.574131727218628,
      "learning_rate": 3.982435607294227e-06,
      "loss": 9.436,
      "step": 2541
    },
    {
      "epoch": 0.8768540876164195,
      "grad_norm": 1.2923246622085571,
      "learning_rate": 3.9605165095866034e-06,
      "loss": 9.5278,
      "step": 2542
    },
    {
      "epoch": 0.8771990341497068,
      "grad_norm": 1.4172004461288452,
      "learning_rate": 3.938655411391806e-06,
      "loss": 9.4752,
      "step": 2543
    },
    {
      "epoch": 0.8775439806829941,
      "grad_norm": 1.5436758995056152,
      "learning_rate": 3.916852340249932e-06,
      "loss": 9.4072,
      "step": 2544
    },
    {
      "epoch": 0.8778889272162814,
      "grad_norm": 1.5451159477233887,
      "learning_rate": 3.895107323628022e-06,
      "loss": 9.4346,
      "step": 2545
    },
    {
      "epoch": 0.8782338737495688,
      "grad_norm": 1.6603964567184448,
      "learning_rate": 3.873420388919951e-06,
      "loss": 9.3515,
      "step": 2546
    },
    {
      "epoch": 0.8785788202828562,
      "grad_norm": 1.6220159530639648,
      "learning_rate": 3.851791563446444e-06,
      "loss": 9.354,
      "step": 2547
    },
    {
      "epoch": 0.8789237668161435,
      "grad_norm": 1.769572138786316,
      "learning_rate": 3.8302208744549985e-06,
      "loss": 9.3154,
      "step": 2548
    },
    {
      "epoch": 0.8792687133494308,
      "grad_norm": 1.58255136013031,
      "learning_rate": 3.8087083491199104e-06,
      "loss": 9.4663,
      "step": 2549
    },
    {
      "epoch": 0.8796136598827182,
      "grad_norm": 1.7701119184494019,
      "learning_rate": 3.7872540145421574e-06,
      "loss": 9.3808,
      "step": 2550
    },
    {
      "epoch": 0.8799586064160055,
      "grad_norm": 0.9108375310897827,
      "learning_rate": 3.765857897749431e-06,
      "loss": 9.5752,
      "step": 2551
    },
    {
      "epoch": 0.8803035529492929,
      "grad_norm": 1.0239676237106323,
      "learning_rate": 3.7445200256961023e-06,
      "loss": 9.6122,
      "step": 2552
    },
    {
      "epoch": 0.8806484994825802,
      "grad_norm": 1.0186281204223633,
      "learning_rate": 3.723240425263119e-06,
      "loss": 9.5595,
      "step": 2553
    },
    {
      "epoch": 0.8809934460158675,
      "grad_norm": 1.052709698677063,
      "learning_rate": 3.7020191232580603e-06,
      "loss": 9.5816,
      "step": 2554
    },
    {
      "epoch": 0.8813383925491549,
      "grad_norm": 1.0933276414871216,
      "learning_rate": 3.680856146415046e-06,
      "loss": 9.538,
      "step": 2555
    },
    {
      "epoch": 0.8816833390824422,
      "grad_norm": 1.128589153289795,
      "learning_rate": 3.659751521394733e-06,
      "loss": 9.5611,
      "step": 2556
    },
    {
      "epoch": 0.8820282856157295,
      "grad_norm": 1.0991203784942627,
      "learning_rate": 3.6387052747842376e-06,
      "loss": 9.601,
      "step": 2557
    },
    {
      "epoch": 0.8823732321490169,
      "grad_norm": 1.1087894439697266,
      "learning_rate": 3.6177174330971675e-06,
      "loss": 9.4994,
      "step": 2558
    },
    {
      "epoch": 0.8827181786823043,
      "grad_norm": 1.183724045753479,
      "learning_rate": 3.5967880227735172e-06,
      "loss": 9.553,
      "step": 2559
    },
    {
      "epoch": 0.8830631252155916,
      "grad_norm": 1.090376615524292,
      "learning_rate": 3.5759170701797017e-06,
      "loss": 9.5626,
      "step": 2560
    },
    {
      "epoch": 0.8834080717488789,
      "grad_norm": 1.3279502391815186,
      "learning_rate": 3.555104601608483e-06,
      "loss": 9.5579,
      "step": 2561
    },
    {
      "epoch": 0.8837530182821662,
      "grad_norm": 1.1904933452606201,
      "learning_rate": 3.5343506432789498e-06,
      "loss": 9.5454,
      "step": 2562
    },
    {
      "epoch": 0.8840979648154536,
      "grad_norm": 1.2588059902191162,
      "learning_rate": 3.5136552213364593e-06,
      "loss": 9.5167,
      "step": 2563
    },
    {
      "epoch": 0.884442911348741,
      "grad_norm": 1.151564598083496,
      "learning_rate": 3.493018361852651e-06,
      "loss": 9.5044,
      "step": 2564
    },
    {
      "epoch": 0.8847878578820283,
      "grad_norm": 1.164340853691101,
      "learning_rate": 3.4724400908253853e-06,
      "loss": 9.5034,
      "step": 2565
    },
    {
      "epoch": 0.8851328044153156,
      "grad_norm": 1.254442572593689,
      "learning_rate": 3.4519204341786902e-06,
      "loss": 9.558,
      "step": 2566
    },
    {
      "epoch": 0.8854777509486029,
      "grad_norm": 1.2330467700958252,
      "learning_rate": 3.4314594177627944e-06,
      "loss": 9.4793,
      "step": 2567
    },
    {
      "epoch": 0.8858226974818904,
      "grad_norm": 1.302596092224121,
      "learning_rate": 3.4110570673539955e-06,
      "loss": 9.5275,
      "step": 2568
    },
    {
      "epoch": 0.8861676440151777,
      "grad_norm": 1.3285911083221436,
      "learning_rate": 3.390713408654761e-06,
      "loss": 9.5662,
      "step": 2569
    },
    {
      "epoch": 0.886512590548465,
      "grad_norm": 1.2535372972488403,
      "learning_rate": 3.370428467293546e-06,
      "loss": 9.5002,
      "step": 2570
    },
    {
      "epoch": 0.8868575370817523,
      "grad_norm": 1.2314890623092651,
      "learning_rate": 3.3502022688248867e-06,
      "loss": 9.5106,
      "step": 2571
    },
    {
      "epoch": 0.8872024836150396,
      "grad_norm": 1.347776174545288,
      "learning_rate": 3.330034838729279e-06,
      "loss": 9.4378,
      "step": 2572
    },
    {
      "epoch": 0.887547430148327,
      "grad_norm": 1.3018256425857544,
      "learning_rate": 3.309926202413205e-06,
      "loss": 9.5894,
      "step": 2573
    },
    {
      "epoch": 0.8878923766816144,
      "grad_norm": 1.2346081733703613,
      "learning_rate": 3.2898763852090895e-06,
      "loss": 9.498,
      "step": 2574
    },
    {
      "epoch": 0.8882373232149017,
      "grad_norm": 1.3945900201797485,
      "learning_rate": 3.269885412375223e-06,
      "loss": 9.46,
      "step": 2575
    },
    {
      "epoch": 0.888582269748189,
      "grad_norm": 1.4205291271209717,
      "learning_rate": 3.249953309095799e-06,
      "loss": 9.5027,
      "step": 2576
    },
    {
      "epoch": 0.8889272162814764,
      "grad_norm": 1.2638293504714966,
      "learning_rate": 3.2300801004808314e-06,
      "loss": 9.491,
      "step": 2577
    },
    {
      "epoch": 0.8892721628147637,
      "grad_norm": 1.215364694595337,
      "learning_rate": 3.21026581156616e-06,
      "loss": 9.5241,
      "step": 2578
    },
    {
      "epoch": 0.889617109348051,
      "grad_norm": 1.4178768396377563,
      "learning_rate": 3.1905104673133625e-06,
      "loss": 9.5186,
      "step": 2579
    },
    {
      "epoch": 0.8899620558813384,
      "grad_norm": 1.3347620964050293,
      "learning_rate": 3.170814092609792e-06,
      "loss": 9.4986,
      "step": 2580
    },
    {
      "epoch": 0.8903070024146257,
      "grad_norm": 1.4867733716964722,
      "learning_rate": 3.151176712268489e-06,
      "loss": 9.497,
      "step": 2581
    },
    {
      "epoch": 0.8906519489479131,
      "grad_norm": 1.451385259628296,
      "learning_rate": 3.1315983510281976e-06,
      "loss": 9.4539,
      "step": 2582
    },
    {
      "epoch": 0.8909968954812004,
      "grad_norm": 1.2720712423324585,
      "learning_rate": 3.1120790335533e-06,
      "loss": 9.5447,
      "step": 2583
    },
    {
      "epoch": 0.8913418420144877,
      "grad_norm": 1.3472886085510254,
      "learning_rate": 3.0926187844337984e-06,
      "loss": 9.4754,
      "step": 2584
    },
    {
      "epoch": 0.891686788547775,
      "grad_norm": 1.3972887992858887,
      "learning_rate": 3.0732176281852652e-06,
      "loss": 9.4664,
      "step": 2585
    },
    {
      "epoch": 0.8920317350810625,
      "grad_norm": 1.3255772590637207,
      "learning_rate": 3.053875589248861e-06,
      "loss": 9.5324,
      "step": 2586
    },
    {
      "epoch": 0.8923766816143498,
      "grad_norm": 1.3736553192138672,
      "learning_rate": 3.0345926919912505e-06,
      "loss": 9.4713,
      "step": 2587
    },
    {
      "epoch": 0.8927216281476371,
      "grad_norm": 1.5358455181121826,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 9.3817,
      "step": 2588
    },
    {
      "epoch": 0.8930665746809244,
      "grad_norm": 1.3966903686523438,
      "learning_rate": 2.996204419606502e-06,
      "loss": 9.4691,
      "step": 2589
    },
    {
      "epoch": 0.8934115212142117,
      "grad_norm": 1.3497977256774902,
      "learning_rate": 2.9770990928400576e-06,
      "loss": 9.5051,
      "step": 2590
    },
    {
      "epoch": 0.8937564677474992,
      "grad_norm": 1.4420788288116455,
      "learning_rate": 2.9580530044737263e-06,
      "loss": 9.5041,
      "step": 2591
    },
    {
      "epoch": 0.8941014142807865,
      "grad_norm": 1.465009331703186,
      "learning_rate": 2.939066178501332e-06,
      "loss": 9.4508,
      "step": 2592
    },
    {
      "epoch": 0.8944463608140738,
      "grad_norm": 1.221989631652832,
      "learning_rate": 2.920138638842068e-06,
      "loss": 9.469,
      "step": 2593
    },
    {
      "epoch": 0.8947913073473611,
      "grad_norm": 1.483028769493103,
      "learning_rate": 2.9012704093404062e-06,
      "loss": 9.4044,
      "step": 2594
    },
    {
      "epoch": 0.8951362538806485,
      "grad_norm": 1.5430452823638916,
      "learning_rate": 2.882461513766133e-06,
      "loss": 9.4083,
      "step": 2595
    },
    {
      "epoch": 0.8954812004139359,
      "grad_norm": 1.4477607011795044,
      "learning_rate": 2.8637119758142707e-06,
      "loss": 9.4623,
      "step": 2596
    },
    {
      "epoch": 0.8958261469472232,
      "grad_norm": 1.4406673908233643,
      "learning_rate": 2.8450218191050705e-06,
      "loss": 9.4275,
      "step": 2597
    },
    {
      "epoch": 0.8961710934805105,
      "grad_norm": 1.5870102643966675,
      "learning_rate": 2.8263910671839866e-06,
      "loss": 9.4317,
      "step": 2598
    },
    {
      "epoch": 0.8965160400137978,
      "grad_norm": 1.490122675895691,
      "learning_rate": 2.8078197435216e-06,
      "loss": 9.4802,
      "step": 2599
    },
    {
      "epoch": 0.8968609865470852,
      "grad_norm": 1.7018402814865112,
      "learning_rate": 2.7893078715136687e-06,
      "loss": 9.3888,
      "step": 2600
    },
    {
      "epoch": 0.8972059330803726,
      "grad_norm": 0.8538911938667297,
      "learning_rate": 2.770855474481021e-06,
      "loss": 9.6266,
      "step": 2601
    },
    {
      "epoch": 0.8975508796136599,
      "grad_norm": 1.0329569578170776,
      "learning_rate": 2.7524625756695954e-06,
      "loss": 9.5183,
      "step": 2602
    },
    {
      "epoch": 0.8978958261469472,
      "grad_norm": 1.030372142791748,
      "learning_rate": 2.734129198250318e-06,
      "loss": 9.5589,
      "step": 2603
    },
    {
      "epoch": 0.8982407726802346,
      "grad_norm": 1.0216647386550903,
      "learning_rate": 2.7158553653192144e-06,
      "loss": 9.5562,
      "step": 2604
    },
    {
      "epoch": 0.8985857192135219,
      "grad_norm": 1.0627166032791138,
      "learning_rate": 2.6976410998972136e-06,
      "loss": 9.5075,
      "step": 2605
    },
    {
      "epoch": 0.8989306657468092,
      "grad_norm": 1.1064709424972534,
      "learning_rate": 2.6794864249302664e-06,
      "loss": 9.5467,
      "step": 2606
    },
    {
      "epoch": 0.8992756122800966,
      "grad_norm": 1.0749237537384033,
      "learning_rate": 2.6613913632892064e-06,
      "loss": 9.5038,
      "step": 2607
    },
    {
      "epoch": 0.8996205588133839,
      "grad_norm": 1.1983023881912231,
      "learning_rate": 2.6433559377697925e-06,
      "loss": 9.4699,
      "step": 2608
    },
    {
      "epoch": 0.8999655053466713,
      "grad_norm": 1.2673431634902954,
      "learning_rate": 2.625380171092667e-06,
      "loss": 9.482,
      "step": 2609
    },
    {
      "epoch": 0.9003104518799586,
      "grad_norm": 1.168282389640808,
      "learning_rate": 2.6074640859032718e-06,
      "loss": 9.5551,
      "step": 2610
    },
    {
      "epoch": 0.9006553984132459,
      "grad_norm": 1.100938320159912,
      "learning_rate": 2.5896077047719237e-06,
      "loss": 9.6096,
      "step": 2611
    },
    {
      "epoch": 0.9010003449465332,
      "grad_norm": 1.2417744398117065,
      "learning_rate": 2.5718110501936675e-06,
      "loss": 9.5024,
      "step": 2612
    },
    {
      "epoch": 0.9013452914798207,
      "grad_norm": 1.1761174201965332,
      "learning_rate": 2.554074144588342e-06,
      "loss": 9.5431,
      "step": 2613
    },
    {
      "epoch": 0.901690238013108,
      "grad_norm": 1.2780194282531738,
      "learning_rate": 2.5363970103004955e-06,
      "loss": 9.5324,
      "step": 2614
    },
    {
      "epoch": 0.9020351845463953,
      "grad_norm": 1.1857703924179077,
      "learning_rate": 2.5187796695994026e-06,
      "loss": 9.4997,
      "step": 2615
    },
    {
      "epoch": 0.9023801310796826,
      "grad_norm": 1.2250052690505981,
      "learning_rate": 2.5012221446789775e-06,
      "loss": 9.4995,
      "step": 2616
    },
    {
      "epoch": 0.9027250776129699,
      "grad_norm": 1.1668322086334229,
      "learning_rate": 2.4837244576578047e-06,
      "loss": 9.4859,
      "step": 2617
    },
    {
      "epoch": 0.9030700241462574,
      "grad_norm": 1.3255215883255005,
      "learning_rate": 2.4662866305790842e-06,
      "loss": 9.4968,
      "step": 2618
    },
    {
      "epoch": 0.9034149706795447,
      "grad_norm": 1.371646761894226,
      "learning_rate": 2.4489086854105946e-06,
      "loss": 9.5009,
      "step": 2619
    },
    {
      "epoch": 0.903759917212832,
      "grad_norm": 1.182887315750122,
      "learning_rate": 2.4315906440446955e-06,
      "loss": 9.5169,
      "step": 2620
    },
    {
      "epoch": 0.9041048637461193,
      "grad_norm": 1.215737223625183,
      "learning_rate": 2.414332528298252e-06,
      "loss": 9.5355,
      "step": 2621
    },
    {
      "epoch": 0.9044498102794067,
      "grad_norm": 1.2013062238693237,
      "learning_rate": 2.397134359912667e-06,
      "loss": 9.4352,
      "step": 2622
    },
    {
      "epoch": 0.9047947568126941,
      "grad_norm": 1.2379329204559326,
      "learning_rate": 2.3799961605537937e-06,
      "loss": 9.4499,
      "step": 2623
    },
    {
      "epoch": 0.9051397033459814,
      "grad_norm": 1.1866499185562134,
      "learning_rate": 2.3629179518119606e-06,
      "loss": 9.5595,
      "step": 2624
    },
    {
      "epoch": 0.9054846498792687,
      "grad_norm": 1.3945666551589966,
      "learning_rate": 2.345899755201919e-06,
      "loss": 9.4705,
      "step": 2625
    },
    {
      "epoch": 0.905829596412556,
      "grad_norm": 1.3572181463241577,
      "learning_rate": 2.328941592162809e-06,
      "loss": 9.4464,
      "step": 2626
    },
    {
      "epoch": 0.9061745429458434,
      "grad_norm": 1.3635412454605103,
      "learning_rate": 2.3120434840581406e-06,
      "loss": 9.5151,
      "step": 2627
    },
    {
      "epoch": 0.9065194894791307,
      "grad_norm": 1.2851320505142212,
      "learning_rate": 2.2952054521757804e-06,
      "loss": 9.5445,
      "step": 2628
    },
    {
      "epoch": 0.9068644360124181,
      "grad_norm": 1.4047993421554565,
      "learning_rate": 2.2784275177278934e-06,
      "loss": 9.3997,
      "step": 2629
    },
    {
      "epoch": 0.9072093825457054,
      "grad_norm": 1.3021107912063599,
      "learning_rate": 2.2617097018509613e-06,
      "loss": 9.5147,
      "step": 2630
    },
    {
      "epoch": 0.9075543290789928,
      "grad_norm": 1.280267357826233,
      "learning_rate": 2.2450520256057038e-06,
      "loss": 9.5056,
      "step": 2631
    },
    {
      "epoch": 0.9078992756122801,
      "grad_norm": 1.3371871709823608,
      "learning_rate": 2.22845450997709e-06,
      "loss": 9.521,
      "step": 2632
    },
    {
      "epoch": 0.9082442221455674,
      "grad_norm": 1.3242119550704956,
      "learning_rate": 2.2119171758743117e-06,
      "loss": 9.4992,
      "step": 2633
    },
    {
      "epoch": 0.9085891686788548,
      "grad_norm": 1.357699990272522,
      "learning_rate": 2.19544004413072e-06,
      "loss": 9.4417,
      "step": 2634
    },
    {
      "epoch": 0.9089341152121421,
      "grad_norm": 1.3349027633666992,
      "learning_rate": 2.1790231355038495e-06,
      "loss": 9.462,
      "step": 2635
    },
    {
      "epoch": 0.9092790617454295,
      "grad_norm": 1.4189931154251099,
      "learning_rate": 2.162666470675334e-06,
      "loss": 9.4083,
      "step": 2636
    },
    {
      "epoch": 0.9096240082787168,
      "grad_norm": 1.5067139863967896,
      "learning_rate": 2.146370070250958e-06,
      "loss": 9.4424,
      "step": 2637
    },
    {
      "epoch": 0.9099689548120041,
      "grad_norm": 1.5032390356063843,
      "learning_rate": 2.130133954760538e-06,
      "loss": 9.4019,
      "step": 2638
    },
    {
      "epoch": 0.9103139013452914,
      "grad_norm": 1.3871209621429443,
      "learning_rate": 2.1139581446580017e-06,
      "loss": 9.5032,
      "step": 2639
    },
    {
      "epoch": 0.9106588478785789,
      "grad_norm": 1.5128953456878662,
      "learning_rate": 2.097842660321242e-06,
      "loss": 9.3989,
      "step": 2640
    },
    {
      "epoch": 0.9110037944118662,
      "grad_norm": 1.3251484632492065,
      "learning_rate": 2.081787522052203e-06,
      "loss": 9.5056,
      "step": 2641
    },
    {
      "epoch": 0.9113487409451535,
      "grad_norm": 1.3176518678665161,
      "learning_rate": 2.0657927500767894e-06,
      "loss": 9.4636,
      "step": 2642
    },
    {
      "epoch": 0.9116936874784408,
      "grad_norm": 1.5699663162231445,
      "learning_rate": 2.0498583645448487e-06,
      "loss": 9.4639,
      "step": 2643
    },
    {
      "epoch": 0.9120386340117281,
      "grad_norm": 1.4477311372756958,
      "learning_rate": 2.0339843855301744e-06,
      "loss": 9.4008,
      "step": 2644
    },
    {
      "epoch": 0.9123835805450156,
      "grad_norm": 1.6083184480667114,
      "learning_rate": 2.018170833030436e-06,
      "loss": 9.4396,
      "step": 2645
    },
    {
      "epoch": 0.9127285270783029,
      "grad_norm": 1.6080248355865479,
      "learning_rate": 2.0024177269672094e-06,
      "loss": 9.4432,
      "step": 2646
    },
    {
      "epoch": 0.9130734736115902,
      "grad_norm": 1.4727885723114014,
      "learning_rate": 1.986725087185898e-06,
      "loss": 9.4662,
      "step": 2647
    },
    {
      "epoch": 0.9134184201448775,
      "grad_norm": 1.6938202381134033,
      "learning_rate": 1.9710929334557484e-06,
      "loss": 9.332,
      "step": 2648
    },
    {
      "epoch": 0.9137633666781649,
      "grad_norm": 1.6540645360946655,
      "learning_rate": 1.9555212854697803e-06,
      "loss": 9.3841,
      "step": 2649
    },
    {
      "epoch": 0.9141083132114523,
      "grad_norm": 1.7224019765853882,
      "learning_rate": 1.940010162844824e-06,
      "loss": 9.3694,
      "step": 2650
    },
    {
      "epoch": 0.9144532597447396,
      "grad_norm": 0.7851664423942566,
      "learning_rate": 1.9245595851214328e-06,
      "loss": 9.6529,
      "step": 2651
    },
    {
      "epoch": 0.9147982062780269,
      "grad_norm": 0.9760227203369141,
      "learning_rate": 1.909169571763908e-06,
      "loss": 9.5619,
      "step": 2652
    },
    {
      "epoch": 0.9151431528113142,
      "grad_norm": 1.0906972885131836,
      "learning_rate": 1.8938401421602359e-06,
      "loss": 9.4804,
      "step": 2653
    },
    {
      "epoch": 0.9154880993446016,
      "grad_norm": 1.1999458074569702,
      "learning_rate": 1.8785713156221018e-06,
      "loss": 9.527,
      "step": 2654
    },
    {
      "epoch": 0.9158330458778889,
      "grad_norm": 1.1160448789596558,
      "learning_rate": 1.863363111384836e-06,
      "loss": 9.5069,
      "step": 2655
    },
    {
      "epoch": 0.9161779924111763,
      "grad_norm": 1.0647850036621094,
      "learning_rate": 1.8482155486073739e-06,
      "loss": 9.5693,
      "step": 2656
    },
    {
      "epoch": 0.9165229389444636,
      "grad_norm": 1.2221078872680664,
      "learning_rate": 1.8331286463722951e-06,
      "loss": 9.4765,
      "step": 2657
    },
    {
      "epoch": 0.916867885477751,
      "grad_norm": 1.1700892448425293,
      "learning_rate": 1.8181024236857246e-06,
      "loss": 9.5458,
      "step": 2658
    },
    {
      "epoch": 0.9172128320110383,
      "grad_norm": 1.2416577339172363,
      "learning_rate": 1.8031368994773756e-06,
      "loss": 9.5379,
      "step": 2659
    },
    {
      "epoch": 0.9175577785443256,
      "grad_norm": 1.1829198598861694,
      "learning_rate": 1.788232092600478e-06,
      "loss": 9.5421,
      "step": 2660
    },
    {
      "epoch": 0.917902725077613,
      "grad_norm": 1.1688567399978638,
      "learning_rate": 1.7733880218317788e-06,
      "loss": 9.4909,
      "step": 2661
    },
    {
      "epoch": 0.9182476716109003,
      "grad_norm": 1.1145144701004028,
      "learning_rate": 1.7586047058714972e-06,
      "loss": 9.5701,
      "step": 2662
    },
    {
      "epoch": 0.9185926181441877,
      "grad_norm": 1.1775919198989868,
      "learning_rate": 1.74388216334333e-06,
      "loss": 9.5376,
      "step": 2663
    },
    {
      "epoch": 0.918937564677475,
      "grad_norm": 1.2331610918045044,
      "learning_rate": 1.7292204127944134e-06,
      "loss": 9.4995,
      "step": 2664
    },
    {
      "epoch": 0.9192825112107623,
      "grad_norm": 1.2776384353637695,
      "learning_rate": 1.714619472695278e-06,
      "loss": 9.5589,
      "step": 2665
    },
    {
      "epoch": 0.9196274577440496,
      "grad_norm": 1.1205222606658936,
      "learning_rate": 1.7000793614398714e-06,
      "loss": 9.586,
      "step": 2666
    },
    {
      "epoch": 0.9199724042773371,
      "grad_norm": 1.172733187675476,
      "learning_rate": 1.6856000973455022e-06,
      "loss": 9.5626,
      "step": 2667
    },
    {
      "epoch": 0.9203173508106244,
      "grad_norm": 1.1725226640701294,
      "learning_rate": 1.6711816986528238e-06,
      "loss": 9.5409,
      "step": 2668
    },
    {
      "epoch": 0.9206622973439117,
      "grad_norm": 1.3246572017669678,
      "learning_rate": 1.6568241835258068e-06,
      "loss": 9.5028,
      "step": 2669
    },
    {
      "epoch": 0.921007243877199,
      "grad_norm": 1.4405431747436523,
      "learning_rate": 1.6425275700517385e-06,
      "loss": 9.4395,
      "step": 2670
    },
    {
      "epoch": 0.9213521904104863,
      "grad_norm": 1.2822954654693604,
      "learning_rate": 1.6282918762411614e-06,
      "loss": 9.5234,
      "step": 2671
    },
    {
      "epoch": 0.9216971369437738,
      "grad_norm": 1.3265615701675415,
      "learning_rate": 1.614117120027886e-06,
      "loss": 9.4913,
      "step": 2672
    },
    {
      "epoch": 0.9220420834770611,
      "grad_norm": 1.2790005207061768,
      "learning_rate": 1.6000033192689611e-06,
      "loss": 9.4907,
      "step": 2673
    },
    {
      "epoch": 0.9223870300103484,
      "grad_norm": 1.3700817823410034,
      "learning_rate": 1.5859504917446366e-06,
      "loss": 9.48,
      "step": 2674
    },
    {
      "epoch": 0.9227319765436357,
      "grad_norm": 1.3636354207992554,
      "learning_rate": 1.5719586551583454e-06,
      "loss": 9.4867,
      "step": 2675
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 1.2950481176376343,
      "learning_rate": 1.5580278271366878e-06,
      "loss": 9.4936,
      "step": 2676
    },
    {
      "epoch": 0.9234218696102104,
      "grad_norm": 1.3804364204406738,
      "learning_rate": 1.5441580252294253e-06,
      "loss": 9.4613,
      "step": 2677
    },
    {
      "epoch": 0.9237668161434978,
      "grad_norm": 1.369942307472229,
      "learning_rate": 1.5303492669094089e-06,
      "loss": 9.5052,
      "step": 2678
    },
    {
      "epoch": 0.9241117626767851,
      "grad_norm": 1.2750253677368164,
      "learning_rate": 1.5166015695726122e-06,
      "loss": 9.4562,
      "step": 2679
    },
    {
      "epoch": 0.9244567092100724,
      "grad_norm": 1.348370909690857,
      "learning_rate": 1.5029149505380646e-06,
      "loss": 9.4602,
      "step": 2680
    },
    {
      "epoch": 0.9248016557433598,
      "grad_norm": 1.2539187669754028,
      "learning_rate": 1.4892894270478853e-06,
      "loss": 9.525,
      "step": 2681
    },
    {
      "epoch": 0.9251466022766471,
      "grad_norm": 1.474637508392334,
      "learning_rate": 1.4757250162671822e-06,
      "loss": 9.449,
      "step": 2682
    },
    {
      "epoch": 0.9254915488099345,
      "grad_norm": 1.484521508216858,
      "learning_rate": 1.4622217352841138e-06,
      "loss": 9.3875,
      "step": 2683
    },
    {
      "epoch": 0.9258364953432218,
      "grad_norm": 1.2464600801467896,
      "learning_rate": 1.448779601109801e-06,
      "loss": 9.4846,
      "step": 2684
    },
    {
      "epoch": 0.9261814418765092,
      "grad_norm": 1.404441237449646,
      "learning_rate": 1.4353986306783418e-06,
      "loss": 9.516,
      "step": 2685
    },
    {
      "epoch": 0.9265263884097965,
      "grad_norm": 1.3945449590682983,
      "learning_rate": 1.4220788408468021e-06,
      "loss": 9.4727,
      "step": 2686
    },
    {
      "epoch": 0.9268713349430838,
      "grad_norm": 1.3472886085510254,
      "learning_rate": 1.4088202483951374e-06,
      "loss": 9.4367,
      "step": 2687
    },
    {
      "epoch": 0.9272162814763711,
      "grad_norm": 1.377700686454773,
      "learning_rate": 1.3956228700262252e-06,
      "loss": 9.5139,
      "step": 2688
    },
    {
      "epoch": 0.9275612280096585,
      "grad_norm": 1.389609456062317,
      "learning_rate": 1.3824867223658388e-06,
      "loss": 9.4322,
      "step": 2689
    },
    {
      "epoch": 0.9279061745429459,
      "grad_norm": 1.4232558012008667,
      "learning_rate": 1.3694118219626074e-06,
      "loss": 9.5109,
      "step": 2690
    },
    {
      "epoch": 0.9282511210762332,
      "grad_norm": 1.3743571043014526,
      "learning_rate": 1.3563981852879827e-06,
      "loss": 9.4921,
      "step": 2691
    },
    {
      "epoch": 0.9285960676095205,
      "grad_norm": 1.4875822067260742,
      "learning_rate": 1.3434458287362672e-06,
      "loss": 9.463,
      "step": 2692
    },
    {
      "epoch": 0.9289410141428078,
      "grad_norm": 1.3445839881896973,
      "learning_rate": 1.3305547686245422e-06,
      "loss": 9.4212,
      "step": 2693
    },
    {
      "epoch": 0.9292859606760953,
      "grad_norm": 1.6044926643371582,
      "learning_rate": 1.3177250211926728e-06,
      "loss": 9.4379,
      "step": 2694
    },
    {
      "epoch": 0.9296309072093826,
      "grad_norm": 1.5688241720199585,
      "learning_rate": 1.3049566026033023e-06,
      "loss": 9.4778,
      "step": 2695
    },
    {
      "epoch": 0.9299758537426699,
      "grad_norm": 1.6058199405670166,
      "learning_rate": 1.2922495289417913e-06,
      "loss": 9.421,
      "step": 2696
    },
    {
      "epoch": 0.9303208002759572,
      "grad_norm": 1.693510890007019,
      "learning_rate": 1.2796038162162239e-06,
      "loss": 9.354,
      "step": 2697
    },
    {
      "epoch": 0.9306657468092445,
      "grad_norm": 1.5772984027862549,
      "learning_rate": 1.2670194803573954e-06,
      "loss": 9.3735,
      "step": 2698
    },
    {
      "epoch": 0.931010693342532,
      "grad_norm": 1.614292860031128,
      "learning_rate": 1.2544965372187635e-06,
      "loss": 9.3702,
      "step": 2699
    },
    {
      "epoch": 0.9313556398758193,
      "grad_norm": 1.9061168432235718,
      "learning_rate": 1.2420350025764528e-06,
      "loss": 9.2773,
      "step": 2700
    },
    {
      "epoch": 0.9317005864091066,
      "grad_norm": 0.8620436787605286,
      "learning_rate": 1.2296348921292333e-06,
      "loss": 9.6332,
      "step": 2701
    },
    {
      "epoch": 0.9320455329423939,
      "grad_norm": 0.8850879073143005,
      "learning_rate": 1.2172962214984763e-06,
      "loss": 9.6099,
      "step": 2702
    },
    {
      "epoch": 0.9323904794756813,
      "grad_norm": 1.0009554624557495,
      "learning_rate": 1.2050190062281752e-06,
      "loss": 9.5471,
      "step": 2703
    },
    {
      "epoch": 0.9327354260089686,
      "grad_norm": 1.0762373208999634,
      "learning_rate": 1.1928032617848805e-06,
      "loss": 9.5482,
      "step": 2704
    },
    {
      "epoch": 0.933080372542256,
      "grad_norm": 1.0936650037765503,
      "learning_rate": 1.1806490035577267e-06,
      "loss": 9.5399,
      "step": 2705
    },
    {
      "epoch": 0.9334253190755433,
      "grad_norm": 1.043621301651001,
      "learning_rate": 1.16855624685836e-06,
      "loss": 9.5344,
      "step": 2706
    },
    {
      "epoch": 0.9337702656088306,
      "grad_norm": 1.2308648824691772,
      "learning_rate": 1.1565250069209776e-06,
      "loss": 9.5064,
      "step": 2707
    },
    {
      "epoch": 0.934115212142118,
      "grad_norm": 1.162726640701294,
      "learning_rate": 1.1445552989022668e-06,
      "loss": 9.5275,
      "step": 2708
    },
    {
      "epoch": 0.9344601586754053,
      "grad_norm": 1.1094176769256592,
      "learning_rate": 1.132647137881393e-06,
      "loss": 9.5538,
      "step": 2709
    },
    {
      "epoch": 0.9348051052086926,
      "grad_norm": 1.255376935005188,
      "learning_rate": 1.120800538859995e-06,
      "loss": 9.463,
      "step": 2710
    },
    {
      "epoch": 0.93515005174198,
      "grad_norm": 1.1070146560668945,
      "learning_rate": 1.1090155167621518e-06,
      "loss": 9.5253,
      "step": 2711
    },
    {
      "epoch": 0.9354949982752674,
      "grad_norm": 1.3558119535446167,
      "learning_rate": 1.0972920864343705e-06,
      "loss": 9.4959,
      "step": 2712
    },
    {
      "epoch": 0.9358399448085547,
      "grad_norm": 1.2252329587936401,
      "learning_rate": 1.085630262645565e-06,
      "loss": 9.5362,
      "step": 2713
    },
    {
      "epoch": 0.936184891341842,
      "grad_norm": 1.1986970901489258,
      "learning_rate": 1.07403006008705e-06,
      "loss": 9.4938,
      "step": 2714
    },
    {
      "epoch": 0.9365298378751293,
      "grad_norm": 1.1853275299072266,
      "learning_rate": 1.062491493372486e-06,
      "loss": 9.5321,
      "step": 2715
    },
    {
      "epoch": 0.9368747844084166,
      "grad_norm": 1.169337511062622,
      "learning_rate": 1.0510145770379177e-06,
      "loss": 9.5066,
      "step": 2716
    },
    {
      "epoch": 0.9372197309417041,
      "grad_norm": 1.2148215770721436,
      "learning_rate": 1.0395993255416957e-06,
      "loss": 9.5333,
      "step": 2717
    },
    {
      "epoch": 0.9375646774749914,
      "grad_norm": 1.3632956743240356,
      "learning_rate": 1.0282457532645119e-06,
      "loss": 9.4924,
      "step": 2718
    },
    {
      "epoch": 0.9379096240082787,
      "grad_norm": 1.1406636238098145,
      "learning_rate": 1.0169538745093242e-06,
      "loss": 9.5274,
      "step": 2719
    },
    {
      "epoch": 0.938254570541566,
      "grad_norm": 1.1914732456207275,
      "learning_rate": 1.0057237035014044e-06,
      "loss": 9.4695,
      "step": 2720
    },
    {
      "epoch": 0.9385995170748535,
      "grad_norm": 1.4516353607177734,
      "learning_rate": 9.945552543882685e-07,
      "loss": 9.4253,
      "step": 2721
    },
    {
      "epoch": 0.9389444636081408,
      "grad_norm": 1.3653576374053955,
      "learning_rate": 9.834485412396677e-07,
      "loss": 9.4781,
      "step": 2722
    },
    {
      "epoch": 0.9392894101414281,
      "grad_norm": 1.1982650756835938,
      "learning_rate": 9.724035780476092e-07,
      "loss": 9.599,
      "step": 2723
    },
    {
      "epoch": 0.9396343566747154,
      "grad_norm": 1.244680643081665,
      "learning_rate": 9.6142037872628e-07,
      "loss": 9.5415,
      "step": 2724
    },
    {
      "epoch": 0.9399793032080027,
      "grad_norm": 1.288313388824463,
      "learning_rate": 9.504989571120726e-07,
      "loss": 9.5385,
      "step": 2725
    },
    {
      "epoch": 0.9403242497412901,
      "grad_norm": 1.185012698173523,
      "learning_rate": 9.396393269635484e-07,
      "loss": 9.5646,
      "step": 2726
    },
    {
      "epoch": 0.9406691962745775,
      "grad_norm": 1.2785496711730957,
      "learning_rate": 9.28841501961425e-07,
      "loss": 9.496,
      "step": 2727
    },
    {
      "epoch": 0.9410141428078648,
      "grad_norm": 1.2989834547042847,
      "learning_rate": 9.1810549570856e-07,
      "loss": 9.5301,
      "step": 2728
    },
    {
      "epoch": 0.9413590893411521,
      "grad_norm": 1.2136149406433105,
      "learning_rate": 9.074313217299457e-07,
      "loss": 9.5197,
      "step": 2729
    },
    {
      "epoch": 0.9417040358744395,
      "grad_norm": 1.4056475162506104,
      "learning_rate": 8.968189934726534e-07,
      "loss": 9.5294,
      "step": 2730
    },
    {
      "epoch": 0.9420489824077268,
      "grad_norm": 1.2778812646865845,
      "learning_rate": 8.862685243058666e-07,
      "loss": 9.455,
      "step": 2731
    },
    {
      "epoch": 0.9423939289410141,
      "grad_norm": 1.3463318347930908,
      "learning_rate": 8.757799275208311e-07,
      "loss": 9.5346,
      "step": 2732
    },
    {
      "epoch": 0.9427388754743015,
      "grad_norm": 1.2876229286193848,
      "learning_rate": 8.653532163308387e-07,
      "loss": 9.5036,
      "step": 2733
    },
    {
      "epoch": 0.9430838220075888,
      "grad_norm": 1.3723164796829224,
      "learning_rate": 8.549884038712375e-07,
      "loss": 9.5081,
      "step": 2734
    },
    {
      "epoch": 0.9434287685408762,
      "grad_norm": 1.4172887802124023,
      "learning_rate": 8.446855031993717e-07,
      "loss": 9.4852,
      "step": 2735
    },
    {
      "epoch": 0.9437737150741635,
      "grad_norm": 1.4235061407089233,
      "learning_rate": 8.344445272946199e-07,
      "loss": 9.486,
      "step": 2736
    },
    {
      "epoch": 0.9441186616074508,
      "grad_norm": 1.449661374092102,
      "learning_rate": 8.24265489058329e-07,
      "loss": 9.4349,
      "step": 2737
    },
    {
      "epoch": 0.9444636081407382,
      "grad_norm": 1.4169197082519531,
      "learning_rate": 8.1414840131383e-07,
      "loss": 9.4753,
      "step": 2738
    },
    {
      "epoch": 0.9448085546740256,
      "grad_norm": 1.4273600578308105,
      "learning_rate": 8.040932768063947e-07,
      "loss": 9.4446,
      "step": 2739
    },
    {
      "epoch": 0.9451535012073129,
      "grad_norm": 1.4585797786712646,
      "learning_rate": 7.941001282032512e-07,
      "loss": 9.4683,
      "step": 2740
    },
    {
      "epoch": 0.9454984477406002,
      "grad_norm": 1.433382272720337,
      "learning_rate": 7.841689680935349e-07,
      "loss": 9.4656,
      "step": 2741
    },
    {
      "epoch": 0.9458433942738875,
      "grad_norm": 1.298423409461975,
      "learning_rate": 7.742998089883102e-07,
      "loss": 9.4483,
      "step": 2742
    },
    {
      "epoch": 0.9461883408071748,
      "grad_norm": 1.4828376770019531,
      "learning_rate": 7.644926633205208e-07,
      "loss": 9.4068,
      "step": 2743
    },
    {
      "epoch": 0.9465332873404623,
      "grad_norm": 1.4661272764205933,
      "learning_rate": 7.547475434449835e-07,
      "loss": 9.4218,
      "step": 2744
    },
    {
      "epoch": 0.9468782338737496,
      "grad_norm": 1.4254372119903564,
      "learning_rate": 7.450644616383951e-07,
      "loss": 9.4599,
      "step": 2745
    },
    {
      "epoch": 0.9472231804070369,
      "grad_norm": 1.6096229553222656,
      "learning_rate": 7.354434300992752e-07,
      "loss": 9.4529,
      "step": 2746
    },
    {
      "epoch": 0.9475681269403242,
      "grad_norm": 1.6057524681091309,
      "learning_rate": 7.258844609479953e-07,
      "loss": 9.3971,
      "step": 2747
    },
    {
      "epoch": 0.9479130734736116,
      "grad_norm": 1.5125082731246948,
      "learning_rate": 7.163875662267117e-07,
      "loss": 9.417,
      "step": 2748
    },
    {
      "epoch": 0.948258020006899,
      "grad_norm": 1.7600480318069458,
      "learning_rate": 7.069527578994151e-07,
      "loss": 9.2774,
      "step": 2749
    },
    {
      "epoch": 0.9486029665401863,
      "grad_norm": 1.6905750036239624,
      "learning_rate": 6.975800478518646e-07,
      "loss": 9.3521,
      "step": 2750
    },
    {
      "epoch": 0.9489479130734736,
      "grad_norm": 0.99737948179245,
      "learning_rate": 6.88269447891593e-07,
      "loss": 9.533,
      "step": 2751
    },
    {
      "epoch": 0.9492928596067609,
      "grad_norm": 0.9953950643539429,
      "learning_rate": 6.790209697478789e-07,
      "loss": 9.5993,
      "step": 2752
    },
    {
      "epoch": 0.9496378061400483,
      "grad_norm": 1.0281240940093994,
      "learning_rate": 6.698346250717524e-07,
      "loss": 9.5267,
      "step": 2753
    },
    {
      "epoch": 0.9499827526733357,
      "grad_norm": 1.020925760269165,
      "learning_rate": 6.607104254359675e-07,
      "loss": 9.5643,
      "step": 2754
    },
    {
      "epoch": 0.950327699206623,
      "grad_norm": 1.0711194276809692,
      "learning_rate": 6.516483823349795e-07,
      "loss": 9.5639,
      "step": 2755
    },
    {
      "epoch": 0.9506726457399103,
      "grad_norm": 1.0857421159744263,
      "learning_rate": 6.426485071849564e-07,
      "loss": 9.4995,
      "step": 2756
    },
    {
      "epoch": 0.9510175922731977,
      "grad_norm": 1.262204885482788,
      "learning_rate": 6.337108113237344e-07,
      "loss": 9.4864,
      "step": 2757
    },
    {
      "epoch": 0.951362538806485,
      "grad_norm": 1.0766891241073608,
      "learning_rate": 6.248353060108292e-07,
      "loss": 9.5252,
      "step": 2758
    },
    {
      "epoch": 0.9517074853397723,
      "grad_norm": 1.153833031654358,
      "learning_rate": 6.160220024273966e-07,
      "loss": 9.5178,
      "step": 2759
    },
    {
      "epoch": 0.9520524318730597,
      "grad_norm": 1.1806145906448364,
      "learning_rate": 6.072709116762442e-07,
      "loss": 9.538,
      "step": 2760
    },
    {
      "epoch": 0.952397378406347,
      "grad_norm": 1.146377682685852,
      "learning_rate": 5.98582044781798e-07,
      "loss": 9.534,
      "step": 2761
    },
    {
      "epoch": 0.9527423249396344,
      "grad_norm": 1.1945569515228271,
      "learning_rate": 5.899554126901075e-07,
      "loss": 9.5208,
      "step": 2762
    },
    {
      "epoch": 0.9530872714729217,
      "grad_norm": 1.286086916923523,
      "learning_rate": 5.813910262687905e-07,
      "loss": 9.4976,
      "step": 2763
    },
    {
      "epoch": 0.953432218006209,
      "grad_norm": 1.2750720977783203,
      "learning_rate": 5.728888963070945e-07,
      "loss": 9.5186,
      "step": 2764
    },
    {
      "epoch": 0.9537771645394963,
      "grad_norm": 1.2140814065933228,
      "learning_rate": 5.644490335157959e-07,
      "loss": 9.4896,
      "step": 2765
    },
    {
      "epoch": 0.9541221110727838,
      "grad_norm": 1.1700353622436523,
      "learning_rate": 5.560714485272512e-07,
      "loss": 9.6212,
      "step": 2766
    },
    {
      "epoch": 0.9544670576060711,
      "grad_norm": 1.3271950483322144,
      "learning_rate": 5.477561518953566e-07,
      "loss": 9.4967,
      "step": 2767
    },
    {
      "epoch": 0.9548120041393584,
      "grad_norm": 1.1459087133407593,
      "learning_rate": 5.395031540955275e-07,
      "loss": 9.5539,
      "step": 2768
    },
    {
      "epoch": 0.9551569506726457,
      "grad_norm": 1.122616171836853,
      "learning_rate": 5.313124655247192e-07,
      "loss": 9.5894,
      "step": 2769
    },
    {
      "epoch": 0.955501897205933,
      "grad_norm": 1.3789931535720825,
      "learning_rate": 5.231840965013668e-07,
      "loss": 9.4521,
      "step": 2770
    },
    {
      "epoch": 0.9558468437392205,
      "grad_norm": 1.2291425466537476,
      "learning_rate": 5.151180572654235e-07,
      "loss": 9.5169,
      "step": 2771
    },
    {
      "epoch": 0.9561917902725078,
      "grad_norm": 1.1856980323791504,
      "learning_rate": 5.071143579782889e-07,
      "loss": 9.4942,
      "step": 2772
    },
    {
      "epoch": 0.9565367368057951,
      "grad_norm": 1.2218326330184937,
      "learning_rate": 4.99173008722853e-07,
      "loss": 9.55,
      "step": 2773
    },
    {
      "epoch": 0.9568816833390824,
      "grad_norm": 1.2583118677139282,
      "learning_rate": 4.91294019503441e-07,
      "loss": 9.538,
      "step": 2774
    },
    {
      "epoch": 0.9572266298723697,
      "grad_norm": 1.3726310729980469,
      "learning_rate": 4.834774002458409e-07,
      "loss": 9.4831,
      "step": 2775
    },
    {
      "epoch": 0.9575715764056572,
      "grad_norm": 1.3480925559997559,
      "learning_rate": 4.757231607972534e-07,
      "loss": 9.4674,
      "step": 2776
    },
    {
      "epoch": 0.9579165229389445,
      "grad_norm": 1.2078763246536255,
      "learning_rate": 4.680313109262813e-07,
      "loss": 9.4641,
      "step": 2777
    },
    {
      "epoch": 0.9582614694722318,
      "grad_norm": 1.2556449174880981,
      "learning_rate": 4.6040186032296206e-07,
      "loss": 9.4801,
      "step": 2778
    },
    {
      "epoch": 0.9586064160055191,
      "grad_norm": 1.2832894325256348,
      "learning_rate": 4.5283481859869635e-07,
      "loss": 9.5116,
      "step": 2779
    },
    {
      "epoch": 0.9589513625388065,
      "grad_norm": 1.3231313228607178,
      "learning_rate": 4.4533019528628093e-07,
      "loss": 9.4913,
      "step": 2780
    },
    {
      "epoch": 0.9592963090720938,
      "grad_norm": 1.2611724138259888,
      "learning_rate": 4.3788799983986997e-07,
      "loss": 9.521,
      "step": 2781
    },
    {
      "epoch": 0.9596412556053812,
      "grad_norm": 1.2805002927780151,
      "learning_rate": 4.305082416349804e-07,
      "loss": 9.502,
      "step": 2782
    },
    {
      "epoch": 0.9599862021386685,
      "grad_norm": 1.3639451265335083,
      "learning_rate": 4.231909299684533e-07,
      "loss": 9.4736,
      "step": 2783
    },
    {
      "epoch": 0.9603311486719558,
      "grad_norm": 1.4248766899108887,
      "learning_rate": 4.159360740584817e-07,
      "loss": 9.5154,
      "step": 2784
    },
    {
      "epoch": 0.9606760952052432,
      "grad_norm": 1.5050346851348877,
      "learning_rate": 4.0874368304457676e-07,
      "loss": 9.4806,
      "step": 2785
    },
    {
      "epoch": 0.9610210417385305,
      "grad_norm": 1.28929603099823,
      "learning_rate": 4.016137659875463e-07,
      "loss": 9.5005,
      "step": 2786
    },
    {
      "epoch": 0.9613659882718179,
      "grad_norm": 1.1755974292755127,
      "learning_rate": 3.945463318695053e-07,
      "loss": 9.4845,
      "step": 2787
    },
    {
      "epoch": 0.9617109348051052,
      "grad_norm": 1.4888191223144531,
      "learning_rate": 3.8754138959383733e-07,
      "loss": 9.4333,
      "step": 2788
    },
    {
      "epoch": 0.9620558813383926,
      "grad_norm": 1.3838120698928833,
      "learning_rate": 3.805989479852279e-07,
      "loss": 9.4622,
      "step": 2789
    },
    {
      "epoch": 0.9624008278716799,
      "grad_norm": 1.3965057134628296,
      "learning_rate": 3.7371901578959756e-07,
      "loss": 9.4135,
      "step": 2790
    },
    {
      "epoch": 0.9627457744049672,
      "grad_norm": 1.3354548215866089,
      "learning_rate": 3.6690160167413554e-07,
      "loss": 9.4748,
      "step": 2791
    },
    {
      "epoch": 0.9630907209382545,
      "grad_norm": 1.4212675094604492,
      "learning_rate": 3.6014671422727185e-07,
      "loss": 9.4848,
      "step": 2792
    },
    {
      "epoch": 0.9634356674715419,
      "grad_norm": 1.4828577041625977,
      "learning_rate": 3.5345436195866053e-07,
      "loss": 9.4245,
      "step": 2793
    },
    {
      "epoch": 0.9637806140048293,
      "grad_norm": 1.4509541988372803,
      "learning_rate": 3.468245532991743e-07,
      "loss": 9.4091,
      "step": 2794
    },
    {
      "epoch": 0.9641255605381166,
      "grad_norm": 1.5361698865890503,
      "learning_rate": 3.4025729660089877e-07,
      "loss": 9.4698,
      "step": 2795
    },
    {
      "epoch": 0.9644705070714039,
      "grad_norm": 1.5707025527954102,
      "learning_rate": 3.3375260013711604e-07,
      "loss": 9.4206,
      "step": 2796
    },
    {
      "epoch": 0.9648154536046912,
      "grad_norm": 1.47184157371521,
      "learning_rate": 3.273104721023046e-07,
      "loss": 9.3935,
      "step": 2797
    },
    {
      "epoch": 0.9651604001379787,
      "grad_norm": 1.5686029195785522,
      "learning_rate": 3.209309206121058e-07,
      "loss": 9.3636,
      "step": 2798
    },
    {
      "epoch": 0.965505346671266,
      "grad_norm": 1.5438534021377563,
      "learning_rate": 3.1461395370334104e-07,
      "loss": 9.3861,
      "step": 2799
    },
    {
      "epoch": 0.9658502932045533,
      "grad_norm": 1.6539479494094849,
      "learning_rate": 3.0835957933397773e-07,
      "loss": 9.398,
      "step": 2800
    },
    {
      "epoch": 0.9661952397378406,
      "grad_norm": 0.774260401725769,
      "learning_rate": 3.0216780538314116e-07,
      "loss": 9.6004,
      "step": 2801
    },
    {
      "epoch": 0.9665401862711279,
      "grad_norm": 0.9886148571968079,
      "learning_rate": 2.960386396510972e-07,
      "loss": 9.5417,
      "step": 2802
    },
    {
      "epoch": 0.9668851328044153,
      "grad_norm": 1.0178803205490112,
      "learning_rate": 2.8997208985921953e-07,
      "loss": 9.5666,
      "step": 2803
    },
    {
      "epoch": 0.9672300793377027,
      "grad_norm": 1.0676816701889038,
      "learning_rate": 2.8396816365001687e-07,
      "loss": 9.499,
      "step": 2804
    },
    {
      "epoch": 0.96757502587099,
      "grad_norm": 1.0291966199874878,
      "learning_rate": 2.7802686858710016e-07,
      "loss": 9.5364,
      "step": 2805
    },
    {
      "epoch": 0.9679199724042773,
      "grad_norm": 1.100340723991394,
      "learning_rate": 2.7214821215518214e-07,
      "loss": 9.5105,
      "step": 2806
    },
    {
      "epoch": 0.9682649189375647,
      "grad_norm": 1.1685614585876465,
      "learning_rate": 2.6633220176006667e-07,
      "loss": 9.5038,
      "step": 2807
    },
    {
      "epoch": 0.968609865470852,
      "grad_norm": 1.1775288581848145,
      "learning_rate": 2.6057884472862617e-07,
      "loss": 9.5383,
      "step": 2808
    },
    {
      "epoch": 0.9689548120041394,
      "grad_norm": 1.2399282455444336,
      "learning_rate": 2.548881483088128e-07,
      "loss": 9.4643,
      "step": 2809
    },
    {
      "epoch": 0.9692997585374267,
      "grad_norm": 1.2756638526916504,
      "learning_rate": 2.49260119669642e-07,
      "loss": 9.4916,
      "step": 2810
    },
    {
      "epoch": 0.969644705070714,
      "grad_norm": 1.3724730014801025,
      "learning_rate": 2.4369476590118123e-07,
      "loss": 9.4872,
      "step": 2811
    },
    {
      "epoch": 0.9699896516040014,
      "grad_norm": 1.1623852252960205,
      "learning_rate": 2.381920940145277e-07,
      "loss": 9.5659,
      "step": 2812
    },
    {
      "epoch": 0.9703345981372887,
      "grad_norm": 1.3232759237289429,
      "learning_rate": 2.3275211094183623e-07,
      "loss": 9.4726,
      "step": 2813
    },
    {
      "epoch": 0.970679544670576,
      "grad_norm": 1.105460286140442,
      "learning_rate": 2.2737482353626937e-07,
      "loss": 9.5364,
      "step": 2814
    },
    {
      "epoch": 0.9710244912038634,
      "grad_norm": 1.130436897277832,
      "learning_rate": 2.2206023857201385e-07,
      "loss": 9.5565,
      "step": 2815
    },
    {
      "epoch": 0.9713694377371508,
      "grad_norm": 1.2147849798202515,
      "learning_rate": 2.1680836274426962e-07,
      "loss": 9.5194,
      "step": 2816
    },
    {
      "epoch": 0.9717143842704381,
      "grad_norm": 1.2784674167633057,
      "learning_rate": 2.1161920266922763e-07,
      "loss": 9.4549,
      "step": 2817
    },
    {
      "epoch": 0.9720593308037254,
      "grad_norm": 1.3010627031326294,
      "learning_rate": 2.0649276488408086e-07,
      "loss": 9.4602,
      "step": 2818
    },
    {
      "epoch": 0.9724042773370127,
      "grad_norm": 1.2807101011276245,
      "learning_rate": 2.014290558469911e-07,
      "loss": 9.5456,
      "step": 2819
    },
    {
      "epoch": 0.9727492238703,
      "grad_norm": 1.301635503768921,
      "learning_rate": 1.964280819371167e-07,
      "loss": 9.5044,
      "step": 2820
    },
    {
      "epoch": 0.9730941704035875,
      "grad_norm": 1.1405683755874634,
      "learning_rate": 1.914898494545736e-07,
      "loss": 9.5604,
      "step": 2821
    },
    {
      "epoch": 0.9734391169368748,
      "grad_norm": 1.261732816696167,
      "learning_rate": 1.8661436462042437e-07,
      "loss": 9.4851,
      "step": 2822
    },
    {
      "epoch": 0.9737840634701621,
      "grad_norm": 1.2781423330307007,
      "learning_rate": 1.8180163357671143e-07,
      "loss": 9.5201,
      "step": 2823
    },
    {
      "epoch": 0.9741290100034494,
      "grad_norm": 1.2555526494979858,
      "learning_rate": 1.7705166238639047e-07,
      "loss": 9.4532,
      "step": 2824
    },
    {
      "epoch": 0.9744739565367369,
      "grad_norm": 1.2804704904556274,
      "learning_rate": 1.7236445703338044e-07,
      "loss": 9.5293,
      "step": 2825
    },
    {
      "epoch": 0.9748189030700242,
      "grad_norm": 1.3036941289901733,
      "learning_rate": 1.677400234225135e-07,
      "loss": 9.536,
      "step": 2826
    },
    {
      "epoch": 0.9751638496033115,
      "grad_norm": 1.2297347784042358,
      "learning_rate": 1.6317836737955172e-07,
      "loss": 9.496,
      "step": 2827
    },
    {
      "epoch": 0.9755087961365988,
      "grad_norm": 1.2707699537277222,
      "learning_rate": 1.586794946511594e-07,
      "loss": 9.4632,
      "step": 2828
    },
    {
      "epoch": 0.9758537426698861,
      "grad_norm": 1.3694789409637451,
      "learning_rate": 1.542434109049251e-07,
      "loss": 9.4874,
      "step": 2829
    },
    {
      "epoch": 0.9761986892031735,
      "grad_norm": 1.3027503490447998,
      "learning_rate": 1.4987012172932302e-07,
      "loss": 9.4417,
      "step": 2830
    },
    {
      "epoch": 0.9765436357364609,
      "grad_norm": 1.300711750984192,
      "learning_rate": 1.4555963263372385e-07,
      "loss": 9.5223,
      "step": 2831
    },
    {
      "epoch": 0.9768885822697482,
      "grad_norm": 1.29239821434021,
      "learning_rate": 1.413119490483894e-07,
      "loss": 9.4713,
      "step": 2832
    },
    {
      "epoch": 0.9772335288030355,
      "grad_norm": 1.3773186206817627,
      "learning_rate": 1.3712707632445032e-07,
      "loss": 9.4819,
      "step": 2833
    },
    {
      "epoch": 0.9775784753363229,
      "grad_norm": 1.4483076333999634,
      "learning_rate": 1.3300501973392277e-07,
      "loss": 9.4047,
      "step": 2834
    },
    {
      "epoch": 0.9779234218696102,
      "grad_norm": 1.4170414209365845,
      "learning_rate": 1.2894578446968065e-07,
      "loss": 9.5096,
      "step": 2835
    },
    {
      "epoch": 0.9782683684028975,
      "grad_norm": 1.5087085962295532,
      "learning_rate": 1.2494937564545562e-07,
      "loss": 9.4874,
      "step": 2836
    },
    {
      "epoch": 0.9786133149361849,
      "grad_norm": 1.5026702880859375,
      "learning_rate": 1.2101579829583154e-07,
      "loss": 9.4319,
      "step": 2837
    },
    {
      "epoch": 0.9789582614694722,
      "grad_norm": 1.3315801620483398,
      "learning_rate": 1.1714505737625004e-07,
      "loss": 9.4616,
      "step": 2838
    },
    {
      "epoch": 0.9793032080027596,
      "grad_norm": 1.5373655557632446,
      "learning_rate": 1.133371577629716e-07,
      "loss": 9.4924,
      "step": 2839
    },
    {
      "epoch": 0.9796481545360469,
      "grad_norm": 1.4027314186096191,
      "learning_rate": 1.095921042531145e-07,
      "loss": 9.3959,
      "step": 2840
    },
    {
      "epoch": 0.9799931010693342,
      "grad_norm": 1.4827989339828491,
      "learning_rate": 1.0590990156461034e-07,
      "loss": 9.4222,
      "step": 2841
    },
    {
      "epoch": 0.9803380476026216,
      "grad_norm": 1.365110158920288,
      "learning_rate": 1.022905543362096e-07,
      "loss": 9.4457,
      "step": 2842
    },
    {
      "epoch": 0.980682994135909,
      "grad_norm": 1.4800422191619873,
      "learning_rate": 9.873406712749279e-08,
      "loss": 9.4845,
      "step": 2843
    },
    {
      "epoch": 0.9810279406691963,
      "grad_norm": 1.482553243637085,
      "learning_rate": 9.524044441883706e-08,
      "loss": 9.4586,
      "step": 2844
    },
    {
      "epoch": 0.9813728872024836,
      "grad_norm": 1.5155751705169678,
      "learning_rate": 9.180969061143852e-08,
      "loss": 9.4402,
      "step": 2845
    },
    {
      "epoch": 0.9817178337357709,
      "grad_norm": 1.5355010032653809,
      "learning_rate": 8.844181002727325e-08,
      "loss": 9.4685,
      "step": 2846
    },
    {
      "epoch": 0.9820627802690582,
      "grad_norm": 1.5655204057693481,
      "learning_rate": 8.513680690913073e-08,
      "loss": 9.4197,
      "step": 2847
    },
    {
      "epoch": 0.9824077268023457,
      "grad_norm": 1.5435409545898438,
      "learning_rate": 8.189468542057488e-08,
      "loss": 9.4517,
      "step": 2848
    },
    {
      "epoch": 0.982752673335633,
      "grad_norm": 1.6485852003097534,
      "learning_rate": 7.871544964596633e-08,
      "loss": 9.4181,
      "step": 2849
    },
    {
      "epoch": 0.9830976198689203,
      "grad_norm": 1.6750664710998535,
      "learning_rate": 7.559910359042355e-08,
      "loss": 9.361,
      "step": 2850
    },
    {
      "epoch": 0.9834425664022076,
      "grad_norm": 0.9129282832145691,
      "learning_rate": 7.254565117985613e-08,
      "loss": 9.5804,
      "step": 2851
    },
    {
      "epoch": 0.983787512935495,
      "grad_norm": 1.0424609184265137,
      "learning_rate": 6.955509626093703e-08,
      "loss": 9.5371,
      "step": 2852
    },
    {
      "epoch": 0.9841324594687824,
      "grad_norm": 0.9472970366477966,
      "learning_rate": 6.662744260109155e-08,
      "loss": 9.6361,
      "step": 2853
    },
    {
      "epoch": 0.9844774060020697,
      "grad_norm": 1.1693047285079956,
      "learning_rate": 6.376269388852496e-08,
      "loss": 9.4819,
      "step": 2854
    },
    {
      "epoch": 0.984822352535357,
      "grad_norm": 0.99554044008255,
      "learning_rate": 6.096085373217264e-08,
      "loss": 9.5079,
      "step": 2855
    },
    {
      "epoch": 0.9851672990686443,
      "grad_norm": 0.9872909188270569,
      "learning_rate": 5.822192566173334e-08,
      "loss": 9.5455,
      "step": 2856
    },
    {
      "epoch": 0.9855122456019317,
      "grad_norm": 1.08110773563385,
      "learning_rate": 5.554591312765811e-08,
      "loss": 9.5288,
      "step": 2857
    },
    {
      "epoch": 0.985857192135219,
      "grad_norm": 1.1271926164627075,
      "learning_rate": 5.2932819501111395e-08,
      "loss": 9.5112,
      "step": 2858
    },
    {
      "epoch": 0.9862021386685064,
      "grad_norm": 1.2304562330245972,
      "learning_rate": 5.038264807402105e-08,
      "loss": 9.5027,
      "step": 2859
    },
    {
      "epoch": 0.9865470852017937,
      "grad_norm": 1.0815626382827759,
      "learning_rate": 4.789540205902832e-08,
      "loss": 9.5418,
      "step": 2860
    },
    {
      "epoch": 0.9868920317350811,
      "grad_norm": 1.279129147529602,
      "learning_rate": 4.547108458951566e-08,
      "loss": 9.5028,
      "step": 2861
    },
    {
      "epoch": 0.9872369782683684,
      "grad_norm": 1.1527676582336426,
      "learning_rate": 4.310969871958448e-08,
      "loss": 9.541,
      "step": 2862
    },
    {
      "epoch": 0.9875819248016557,
      "grad_norm": 1.260764241218567,
      "learning_rate": 4.0811247424049625e-08,
      "loss": 9.4835,
      "step": 2863
    },
    {
      "epoch": 0.9879268713349431,
      "grad_norm": 1.153011679649353,
      "learning_rate": 3.857573359845601e-08,
      "loss": 9.5011,
      "step": 2864
    },
    {
      "epoch": 0.9882718178682304,
      "grad_norm": 1.124871015548706,
      "learning_rate": 3.6403160059050865e-08,
      "loss": 9.5657,
      "step": 2865
    },
    {
      "epoch": 0.9886167644015178,
      "grad_norm": 1.1694023609161377,
      "learning_rate": 3.4293529542800406e-08,
      "loss": 9.4759,
      "step": 2866
    },
    {
      "epoch": 0.9889617109348051,
      "grad_norm": 1.1712218523025513,
      "learning_rate": 3.224684470735651e-08,
      "loss": 9.5351,
      "step": 2867
    },
    {
      "epoch": 0.9893066574680924,
      "grad_norm": 1.2313039302825928,
      "learning_rate": 3.0263108131095566e-08,
      "loss": 9.5316,
      "step": 2868
    },
    {
      "epoch": 0.9896516040013797,
      "grad_norm": 1.1785590648651123,
      "learning_rate": 2.8342322313085202e-08,
      "loss": 9.4722,
      "step": 2869
    },
    {
      "epoch": 0.9899965505346672,
      "grad_norm": 1.2561894655227661,
      "learning_rate": 2.6484489673084257e-08,
      "loss": 9.5575,
      "step": 2870
    },
    {
      "epoch": 0.9903414970679545,
      "grad_norm": 1.1631916761398315,
      "learning_rate": 2.4689612551553888e-08,
      "loss": 9.5365,
      "step": 2871
    },
    {
      "epoch": 0.9906864436012418,
      "grad_norm": 1.2572925090789795,
      "learning_rate": 2.2957693209635368e-08,
      "loss": 9.5596,
      "step": 2872
    },
    {
      "epoch": 0.9910313901345291,
      "grad_norm": 1.324566125869751,
      "learning_rate": 2.1288733829161188e-08,
      "loss": 9.4521,
      "step": 2873
    },
    {
      "epoch": 0.9913763366678164,
      "grad_norm": 1.3486089706420898,
      "learning_rate": 1.9682736512660617e-08,
      "loss": 9.4974,
      "step": 2874
    },
    {
      "epoch": 0.9917212832011039,
      "grad_norm": 1.174851417541504,
      "learning_rate": 1.813970328331527e-08,
      "loss": 9.5033,
      "step": 2875
    },
    {
      "epoch": 0.9920662297343912,
      "grad_norm": 1.2707513570785522,
      "learning_rate": 1.6659636085020192e-08,
      "loss": 9.4723,
      "step": 2876
    },
    {
      "epoch": 0.9924111762676785,
      "grad_norm": 1.474165678024292,
      "learning_rate": 1.5242536782317242e-08,
      "loss": 9.4423,
      "step": 2877
    },
    {
      "epoch": 0.9927561228009658,
      "grad_norm": 1.3797553777694702,
      "learning_rate": 1.388840716045059e-08,
      "loss": 9.4966,
      "step": 2878
    },
    {
      "epoch": 0.9931010693342532,
      "grad_norm": 1.326752781867981,
      "learning_rate": 1.2597248925311222e-08,
      "loss": 9.5313,
      "step": 2879
    },
    {
      "epoch": 0.9934460158675406,
      "grad_norm": 1.3259997367858887,
      "learning_rate": 1.1369063703475791e-08,
      "loss": 9.4992,
      "step": 2880
    },
    {
      "epoch": 0.9937909624008279,
      "grad_norm": 1.3481284379959106,
      "learning_rate": 1.0203853042184407e-08,
      "loss": 9.4431,
      "step": 2881
    },
    {
      "epoch": 0.9941359089341152,
      "grad_norm": 1.3297460079193115,
      "learning_rate": 9.101618409340651e-09,
      "loss": 9.4599,
      "step": 2882
    },
    {
      "epoch": 0.9944808554674025,
      "grad_norm": 1.3250812292099,
      "learning_rate": 8.06236119351711e-09,
      "loss": 9.4787,
      "step": 2883
    },
    {
      "epoch": 0.9948258020006899,
      "grad_norm": 1.3159713745117188,
      "learning_rate": 7.086082703949837e-09,
      "loss": 9.4653,
      "step": 2884
    },
    {
      "epoch": 0.9951707485339772,
      "grad_norm": 1.3725025653839111,
      "learning_rate": 6.172784170532797e-09,
      "loss": 9.5064,
      "step": 2885
    },
    {
      "epoch": 0.9955156950672646,
      "grad_norm": 1.3856096267700195,
      "learning_rate": 5.3224667438123155e-09,
      "loss": 9.4516,
      "step": 2886
    },
    {
      "epoch": 0.9958606416005519,
      "grad_norm": 1.391958475112915,
      "learning_rate": 4.53513149500373e-09,
      "loss": 9.4719,
      "step": 2887
    },
    {
      "epoch": 0.9962055881338393,
      "grad_norm": 1.5442167520523071,
      "learning_rate": 3.810779415974741e-09,
      "loss": 9.4641,
      "step": 2888
    },
    {
      "epoch": 0.9965505346671266,
      "grad_norm": 1.4803158044815063,
      "learning_rate": 3.1494114192509585e-09,
      "loss": 9.4186,
      "step": 2889
    },
    {
      "epoch": 0.9968954812004139,
      "grad_norm": 1.4997535943984985,
      "learning_rate": 2.5510283379992504e-09,
      "loss": 9.4915,
      "step": 2890
    },
    {
      "epoch": 0.9972404277337013,
      "grad_norm": 1.5729633569717407,
      "learning_rate": 2.0156309260610517e-09,
      "loss": 9.423,
      "step": 2891
    },
    {
      "epoch": 0.9975853742669886,
      "grad_norm": 1.605699062347412,
      "learning_rate": 1.5432198579079516e-09,
      "loss": 9.4781,
      "step": 2892
    },
    {
      "epoch": 0.997930320800276,
      "grad_norm": 1.4182220697402954,
      "learning_rate": 1.1337957286805534e-09,
      "loss": 9.453,
      "step": 2893
    },
    {
      "epoch": 0.9982752673335633,
      "grad_norm": 1.5460253953933716,
      "learning_rate": 7.873590541551679e-10,
      "loss": 9.5107,
      "step": 2894
    },
    {
      "epoch": 0.9986202138668506,
      "grad_norm": 1.456726312637329,
      "learning_rate": 5.039102707715682e-10,
      "loss": 9.3826,
      "step": 2895
    },
    {
      "epoch": 0.9989651604001379,
      "grad_norm": 1.55360746383667,
      "learning_rate": 2.8344973560523456e-10,
      "loss": 9.394,
      "step": 2896
    },
    {
      "epoch": 0.9993101069334254,
      "grad_norm": 1.6202263832092285,
      "learning_rate": 1.2597772639511006e-10,
      "loss": 9.4164,
      "step": 2897
    },
    {
      "epoch": 0.9996550534667127,
      "grad_norm": 1.6138488054275513,
      "learning_rate": 3.1494441515844684e-11,
      "loss": 9.3771,
      "step": 2898
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.670608639717102,
      "learning_rate": 0.0,
      "loss": 9.3193,
      "step": 2899
    }
  ],
  "logging_steps": 1,
  "max_steps": 2899,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 239,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 601758459101184.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}