{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2082,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004803073967339097,
      "grad_norm": 12.295453677898383,
      "learning_rate": 1.6e-08,
      "loss": 0.3404,
      "step": 1
    },
    {
      "epoch": 0.0009606147934678194,
      "grad_norm": 11.115325503679177,
      "learning_rate": 3.2e-08,
      "loss": 0.2956,
      "step": 2
    },
    {
      "epoch": 0.001440922190201729,
      "grad_norm": 12.134299471519068,
      "learning_rate": 4.8e-08,
      "loss": 0.3497,
      "step": 3
    },
    {
      "epoch": 0.0019212295869356388,
      "grad_norm": 11.889325768094226,
      "learning_rate": 6.4e-08,
      "loss": 0.3239,
      "step": 4
    },
    {
      "epoch": 0.0024015369836695487,
      "grad_norm": 11.017686747111874,
      "learning_rate": 8e-08,
      "loss": 0.2895,
      "step": 5
    },
    {
      "epoch": 0.002881844380403458,
      "grad_norm": 10.764449822783085,
      "learning_rate": 9.6e-08,
      "loss": 0.3062,
      "step": 6
    },
    {
      "epoch": 0.0033621517771373678,
      "grad_norm": 9.810863684605177,
      "learning_rate": 1.1200000000000001e-07,
      "loss": 0.3056,
      "step": 7
    },
    {
      "epoch": 0.0038424591738712775,
      "grad_norm": 11.438972361023298,
      "learning_rate": 1.28e-07,
      "loss": 0.294,
      "step": 8
    },
    {
      "epoch": 0.004322766570605188,
      "grad_norm": 11.696011568247082,
      "learning_rate": 1.4400000000000002e-07,
      "loss": 0.3285,
      "step": 9
    },
    {
      "epoch": 0.004803073967339097,
      "grad_norm": 11.507512344037782,
      "learning_rate": 1.6e-07,
      "loss": 0.314,
      "step": 10
    },
    {
      "epoch": 0.005283381364073006,
      "grad_norm": 10.684581565163377,
      "learning_rate": 1.7600000000000001e-07,
      "loss": 0.2883,
      "step": 11
    },
    {
      "epoch": 0.005763688760806916,
      "grad_norm": 11.24558403809558,
      "learning_rate": 1.92e-07,
      "loss": 0.3116,
      "step": 12
    },
    {
      "epoch": 0.006243996157540826,
      "grad_norm": 10.790135146175551,
      "learning_rate": 2.08e-07,
      "loss": 0.3244,
      "step": 13
    },
    {
      "epoch": 0.0067243035542747355,
      "grad_norm": 10.608527245093223,
      "learning_rate": 2.2400000000000002e-07,
      "loss": 0.2747,
      "step": 14
    },
    {
      "epoch": 0.007204610951008645,
      "grad_norm": 10.265000231723027,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 0.2674,
      "step": 15
    },
    {
      "epoch": 0.007684918347742555,
      "grad_norm": 11.298317377682416,
      "learning_rate": 2.56e-07,
      "loss": 0.2679,
      "step": 16
    },
    {
      "epoch": 0.008165225744476465,
      "grad_norm": 9.86248788668308,
      "learning_rate": 2.72e-07,
      "loss": 0.3099,
      "step": 17
    },
    {
      "epoch": 0.008645533141210375,
      "grad_norm": 9.0577072692478,
      "learning_rate": 2.8800000000000004e-07,
      "loss": 0.2674,
      "step": 18
    },
    {
      "epoch": 0.009125840537944284,
      "grad_norm": 7.238412780645614,
      "learning_rate": 3.04e-07,
      "loss": 0.2082,
      "step": 19
    },
    {
      "epoch": 0.009606147934678195,
      "grad_norm": 9.490442029212941,
      "learning_rate": 3.2e-07,
      "loss": 0.2548,
      "step": 20
    },
    {
      "epoch": 0.010086455331412104,
      "grad_norm": 7.936036846955943,
      "learning_rate": 3.36e-07,
      "loss": 0.2487,
      "step": 21
    },
    {
      "epoch": 0.010566762728146013,
      "grad_norm": 8.384911819682703,
      "learning_rate": 3.5200000000000003e-07,
      "loss": 0.2578,
      "step": 22
    },
    {
      "epoch": 0.011047070124879923,
      "grad_norm": 6.667251467184058,
      "learning_rate": 3.68e-07,
      "loss": 0.2315,
      "step": 23
    },
    {
      "epoch": 0.011527377521613832,
      "grad_norm": 5.388532747106631,
      "learning_rate": 3.84e-07,
      "loss": 0.2153,
      "step": 24
    },
    {
      "epoch": 0.012007684918347743,
      "grad_norm": 5.076276759500522,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.1773,
      "step": 25
    },
    {
      "epoch": 0.012487992315081652,
      "grad_norm": 5.044901367454851,
      "learning_rate": 4.16e-07,
      "loss": 0.1981,
      "step": 26
    },
    {
      "epoch": 0.012968299711815562,
      "grad_norm": 4.6484570213541145,
      "learning_rate": 4.3200000000000006e-07,
      "loss": 0.1962,
      "step": 27
    },
    {
      "epoch": 0.013448607108549471,
      "grad_norm": 4.820579032408368,
      "learning_rate": 4.4800000000000004e-07,
      "loss": 0.1923,
      "step": 28
    },
    {
      "epoch": 0.013928914505283382,
      "grad_norm": 3.6954293876191535,
      "learning_rate": 4.64e-07,
      "loss": 0.1827,
      "step": 29
    },
    {
      "epoch": 0.01440922190201729,
      "grad_norm": 3.8040876250915083,
      "learning_rate": 4.800000000000001e-07,
      "loss": 0.1959,
      "step": 30
    },
    {
      "epoch": 0.014889529298751201,
      "grad_norm": 3.0166885970679043,
      "learning_rate": 4.96e-07,
      "loss": 0.1652,
      "step": 31
    },
    {
      "epoch": 0.01536983669548511,
      "grad_norm": 2.2910049629070497,
      "learning_rate": 5.12e-07,
      "loss": 0.1723,
      "step": 32
    },
    {
      "epoch": 0.01585014409221902,
      "grad_norm": 2.7933136537228744,
      "learning_rate": 5.280000000000001e-07,
      "loss": 0.1663,
      "step": 33
    },
    {
      "epoch": 0.01633045148895293,
      "grad_norm": 2.5494412148634487,
      "learning_rate": 5.44e-07,
      "loss": 0.1593,
      "step": 34
    },
    {
      "epoch": 0.01681075888568684,
      "grad_norm": 2.7365853076508504,
      "learning_rate": 5.6e-07,
      "loss": 0.1845,
      "step": 35
    },
    {
      "epoch": 0.01729106628242075,
      "grad_norm": 2.6326300548480956,
      "learning_rate": 5.760000000000001e-07,
      "loss": 0.1734,
      "step": 36
    },
    {
      "epoch": 0.01777137367915466,
      "grad_norm": 1.9178697505579982,
      "learning_rate": 5.920000000000001e-07,
      "loss": 0.1249,
      "step": 37
    },
    {
      "epoch": 0.01825168107588857,
      "grad_norm": 2.5724433031106853,
      "learning_rate": 6.08e-07,
      "loss": 0.1308,
      "step": 38
    },
    {
      "epoch": 0.018731988472622477,
      "grad_norm": 2.6912073654452016,
      "learning_rate": 6.24e-07,
      "loss": 0.1527,
      "step": 39
    },
    {
      "epoch": 0.01921229586935639,
      "grad_norm": 2.167890593402019,
      "learning_rate": 6.4e-07,
      "loss": 0.1551,
      "step": 40
    },
    {
      "epoch": 0.0196926032660903,
      "grad_norm": 1.9023821581489182,
      "learning_rate": 6.560000000000002e-07,
      "loss": 0.1264,
      "step": 41
    },
    {
      "epoch": 0.020172910662824207,
      "grad_norm": 3.3946961363245363,
      "learning_rate": 6.72e-07,
      "loss": 0.183,
      "step": 42
    },
    {
      "epoch": 0.020653218059558116,
      "grad_norm": 2.2864808587177783,
      "learning_rate": 6.88e-07,
      "loss": 0.1473,
      "step": 43
    },
    {
      "epoch": 0.021133525456292025,
      "grad_norm": 2.0274847089097765,
      "learning_rate": 7.040000000000001e-07,
      "loss": 0.125,
      "step": 44
    },
    {
      "epoch": 0.021613832853025938,
      "grad_norm": 2.315770203936131,
      "learning_rate": 7.2e-07,
      "loss": 0.1362,
      "step": 45
    },
    {
      "epoch": 0.022094140249759846,
      "grad_norm": 2.305110403527537,
      "learning_rate": 7.36e-07,
      "loss": 0.1515,
      "step": 46
    },
    {
      "epoch": 0.022574447646493755,
      "grad_norm": 2.3692793672422985,
      "learning_rate": 7.520000000000001e-07,
      "loss": 0.1427,
      "step": 47
    },
    {
      "epoch": 0.023054755043227664,
      "grad_norm": 1.853538426035806,
      "learning_rate": 7.68e-07,
      "loss": 0.13,
      "step": 48
    },
    {
      "epoch": 0.023535062439961577,
      "grad_norm": 1.7993480609748818,
      "learning_rate": 7.84e-07,
      "loss": 0.1384,
      "step": 49
    },
    {
      "epoch": 0.024015369836695485,
      "grad_norm": 1.2624888010807611,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.0812,
      "step": 50
    },
    {
      "epoch": 0.024495677233429394,
      "grad_norm": 1.654775529490419,
      "learning_rate": 8.160000000000001e-07,
      "loss": 0.1289,
      "step": 51
    },
    {
      "epoch": 0.024975984630163303,
      "grad_norm": 1.9053266889486429,
      "learning_rate": 8.32e-07,
      "loss": 0.128,
      "step": 52
    },
    {
      "epoch": 0.025456292026897216,
      "grad_norm": 1.3131685994032491,
      "learning_rate": 8.480000000000001e-07,
      "loss": 0.1099,
      "step": 53
    },
    {
      "epoch": 0.025936599423631124,
      "grad_norm": 1.8073567315806054,
      "learning_rate": 8.640000000000001e-07,
      "loss": 0.1209,
      "step": 54
    },
    {
      "epoch": 0.026416906820365033,
      "grad_norm": 1.5555744040575845,
      "learning_rate": 8.8e-07,
      "loss": 0.1268,
      "step": 55
    },
    {
      "epoch": 0.026897214217098942,
      "grad_norm": 1.1507622935499833,
      "learning_rate": 8.960000000000001e-07,
      "loss": 0.0939,
      "step": 56
    },
    {
      "epoch": 0.027377521613832854,
      "grad_norm": 1.8426592777636386,
      "learning_rate": 9.120000000000001e-07,
      "loss": 0.1267,
      "step": 57
    },
    {
      "epoch": 0.027857829010566763,
      "grad_norm": 1.8172160960989603,
      "learning_rate": 9.28e-07,
      "loss": 0.1419,
      "step": 58
    },
    {
      "epoch": 0.028338136407300672,
      "grad_norm": 1.734138032657134,
      "learning_rate": 9.440000000000001e-07,
      "loss": 0.1063,
      "step": 59
    },
    {
      "epoch": 0.02881844380403458,
      "grad_norm": 1.9904099384917207,
      "learning_rate": 9.600000000000001e-07,
      "loss": 0.1547,
      "step": 60
    },
    {
      "epoch": 0.029298751200768493,
      "grad_norm": 1.431824839460655,
      "learning_rate": 9.76e-07,
      "loss": 0.1297,
      "step": 61
    },
    {
      "epoch": 0.029779058597502402,
      "grad_norm": 1.3047846711226216,
      "learning_rate": 9.92e-07,
      "loss": 0.0861,
      "step": 62
    },
    {
      "epoch": 0.03025936599423631,
      "grad_norm": 1.9570883092765483,
      "learning_rate": 1.0080000000000001e-06,
      "loss": 0.1552,
      "step": 63
    },
    {
      "epoch": 0.03073967339097022,
      "grad_norm": 1.2651593114993294,
      "learning_rate": 1.024e-06,
      "loss": 0.0973,
      "step": 64
    },
    {
      "epoch": 0.03121998078770413,
      "grad_norm": 1.576842206277053,
      "learning_rate": 1.04e-06,
      "loss": 0.1015,
      "step": 65
    },
    {
      "epoch": 0.03170028818443804,
      "grad_norm": 1.7388488234803576,
      "learning_rate": 1.0560000000000001e-06,
      "loss": 0.1098,
      "step": 66
    },
    {
      "epoch": 0.03218059558117195,
      "grad_norm": 1.5418141689606337,
      "learning_rate": 1.072e-06,
      "loss": 0.0982,
      "step": 67
    },
    {
      "epoch": 0.03266090297790586,
      "grad_norm": 1.7260585222683362,
      "learning_rate": 1.088e-06,
      "loss": 0.1348,
      "step": 68
    },
    {
      "epoch": 0.03314121037463977,
      "grad_norm": 1.4436799175309636,
      "learning_rate": 1.1040000000000001e-06,
      "loss": 0.1192,
      "step": 69
    },
    {
      "epoch": 0.03362151777137368,
      "grad_norm": 3.222503996824014,
      "learning_rate": 1.12e-06,
      "loss": 0.1234,
      "step": 70
    },
    {
      "epoch": 0.034101825168107586,
      "grad_norm": 1.6939823082128094,
      "learning_rate": 1.1360000000000002e-06,
      "loss": 0.1165,
      "step": 71
    },
    {
      "epoch": 0.0345821325648415,
      "grad_norm": 1.8379294842379774,
      "learning_rate": 1.1520000000000002e-06,
      "loss": 0.1185,
      "step": 72
    },
    {
      "epoch": 0.03506243996157541,
      "grad_norm": 1.5569252017032666,
      "learning_rate": 1.168e-06,
      "loss": 0.1312,
      "step": 73
    },
    {
      "epoch": 0.03554274735830932,
      "grad_norm": 1.714684617371849,
      "learning_rate": 1.1840000000000002e-06,
      "loss": 0.0876,
      "step": 74
    },
    {
      "epoch": 0.03602305475504323,
      "grad_norm": 1.3225072536309823,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.0957,
      "step": 75
    },
    {
      "epoch": 0.03650336215177714,
      "grad_norm": 1.8975516746309764,
      "learning_rate": 1.216e-06,
      "loss": 0.1211,
      "step": 76
    },
    {
      "epoch": 0.036983669548511046,
      "grad_norm": 1.5164480525596415,
      "learning_rate": 1.2320000000000002e-06,
      "loss": 0.1117,
      "step": 77
    },
    {
      "epoch": 0.037463976945244955,
      "grad_norm": 1.8598013368901916,
      "learning_rate": 1.248e-06,
      "loss": 0.0974,
      "step": 78
    },
    {
      "epoch": 0.037944284341978864,
      "grad_norm": 1.6096443930527105,
      "learning_rate": 1.2640000000000003e-06,
      "loss": 0.1066,
      "step": 79
    },
    {
      "epoch": 0.03842459173871278,
      "grad_norm": 1.1438512252152493,
      "learning_rate": 1.28e-06,
      "loss": 0.096,
      "step": 80
    },
    {
      "epoch": 0.03890489913544669,
      "grad_norm": 1.464888807639377,
      "learning_rate": 1.296e-06,
      "loss": 0.0982,
      "step": 81
    },
    {
      "epoch": 0.0393852065321806,
      "grad_norm": 1.81847293552993,
      "learning_rate": 1.3120000000000003e-06,
      "loss": 0.1419,
      "step": 82
    },
    {
      "epoch": 0.039865513928914506,
      "grad_norm": 1.7194540925734543,
      "learning_rate": 1.328e-06,
      "loss": 0.1054,
      "step": 83
    },
    {
      "epoch": 0.040345821325648415,
      "grad_norm": 1.46339302106847,
      "learning_rate": 1.344e-06,
      "loss": 0.1246,
      "step": 84
    },
    {
      "epoch": 0.040826128722382324,
      "grad_norm": 1.6380410081431438,
      "learning_rate": 1.3600000000000001e-06,
      "loss": 0.0806,
      "step": 85
    },
    {
      "epoch": 0.04130643611911623,
      "grad_norm": 1.8756328408381668,
      "learning_rate": 1.376e-06,
      "loss": 0.1188,
      "step": 86
    },
    {
      "epoch": 0.04178674351585014,
      "grad_norm": 1.4464326251601627,
      "learning_rate": 1.392e-06,
      "loss": 0.113,
      "step": 87
    },
    {
      "epoch": 0.04226705091258405,
      "grad_norm": 1.240058007444482,
      "learning_rate": 1.4080000000000001e-06,
      "loss": 0.0934,
      "step": 88
    },
    {
      "epoch": 0.042747358309317966,
      "grad_norm": 2.3178241893688645,
      "learning_rate": 1.424e-06,
      "loss": 0.1126,
      "step": 89
    },
    {
      "epoch": 0.043227665706051875,
      "grad_norm": 1.7177471908214919,
      "learning_rate": 1.44e-06,
      "loss": 0.1345,
      "step": 90
    },
    {
      "epoch": 0.043707973102785784,
      "grad_norm": 2.7314130913960812,
      "learning_rate": 1.4560000000000001e-06,
      "loss": 0.1069,
      "step": 91
    },
    {
      "epoch": 0.04418828049951969,
      "grad_norm": 1.0979463668294076,
      "learning_rate": 1.472e-06,
      "loss": 0.0966,
      "step": 92
    },
    {
      "epoch": 0.0446685878962536,
      "grad_norm": 1.4821760121201706,
      "learning_rate": 1.488e-06,
      "loss": 0.1048,
      "step": 93
    },
    {
      "epoch": 0.04514889529298751,
      "grad_norm": 1.6721548051611312,
      "learning_rate": 1.5040000000000001e-06,
      "loss": 0.1065,
      "step": 94
    },
    {
      "epoch": 0.04562920268972142,
      "grad_norm": 1.4046486532313425,
      "learning_rate": 1.52e-06,
      "loss": 0.1121,
      "step": 95
    },
    {
      "epoch": 0.04610951008645533,
      "grad_norm": 1.8920282885751272,
      "learning_rate": 1.536e-06,
      "loss": 0.093,
      "step": 96
    },
    {
      "epoch": 0.046589817483189244,
      "grad_norm": 1.3117971937414172,
      "learning_rate": 1.5520000000000001e-06,
      "loss": 0.0763,
      "step": 97
    },
    {
      "epoch": 0.04707012487992315,
      "grad_norm": 1.2226066484010771,
      "learning_rate": 1.568e-06,
      "loss": 0.0925,
      "step": 98
    },
    {
      "epoch": 0.04755043227665706,
      "grad_norm": 1.2523216029904312,
      "learning_rate": 1.5840000000000002e-06,
      "loss": 0.0886,
      "step": 99
    },
    {
      "epoch": 0.04803073967339097,
      "grad_norm": 1.2711895556124746,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.0871,
      "step": 100
    },
    {
      "epoch": 0.04851104707012488,
      "grad_norm": 3.040299010535842,
      "learning_rate": 1.616e-06,
      "loss": 0.0745,
      "step": 101
    },
    {
      "epoch": 0.04899135446685879,
      "grad_norm": 1.6389584028051778,
      "learning_rate": 1.6320000000000002e-06,
      "loss": 0.1278,
      "step": 102
    },
    {
      "epoch": 0.0494716618635927,
      "grad_norm": 1.5552388728614976,
      "learning_rate": 1.6480000000000001e-06,
      "loss": 0.0993,
      "step": 103
    },
    {
      "epoch": 0.049951969260326606,
      "grad_norm": 1.3887590517997233,
      "learning_rate": 1.664e-06,
      "loss": 0.0846,
      "step": 104
    },
    {
      "epoch": 0.05043227665706052,
      "grad_norm": 1.5458592978622285,
      "learning_rate": 1.6800000000000002e-06,
      "loss": 0.1046,
      "step": 105
    },
    {
      "epoch": 0.05091258405379443,
      "grad_norm": 2.9179450722754434,
      "learning_rate": 1.6960000000000002e-06,
      "loss": 0.1242,
      "step": 106
    },
    {
      "epoch": 0.05139289145052834,
      "grad_norm": 2.7394754636725693,
      "learning_rate": 1.712e-06,
      "loss": 0.0819,
      "step": 107
    },
    {
      "epoch": 0.05187319884726225,
      "grad_norm": 2.077429221247708,
      "learning_rate": 1.7280000000000002e-06,
      "loss": 0.1007,
      "step": 108
    },
    {
      "epoch": 0.05235350624399616,
      "grad_norm": 1.2362929358102204,
      "learning_rate": 1.7440000000000002e-06,
      "loss": 0.0975,
      "step": 109
    },
    {
      "epoch": 0.052833813640730067,
      "grad_norm": 1.5896189455542367,
      "learning_rate": 1.76e-06,
      "loss": 0.1118,
      "step": 110
    },
    {
      "epoch": 0.053314121037463975,
      "grad_norm": 1.8111208272150052,
      "learning_rate": 1.7760000000000002e-06,
      "loss": 0.1173,
      "step": 111
    },
    {
      "epoch": 0.053794428434197884,
      "grad_norm": 1.1388547833141291,
      "learning_rate": 1.7920000000000002e-06,
      "loss": 0.0715,
      "step": 112
    },
    {
      "epoch": 0.05427473583093179,
      "grad_norm": 1.637903840464677,
      "learning_rate": 1.808e-06,
      "loss": 0.1148,
      "step": 113
    },
    {
      "epoch": 0.05475504322766571,
      "grad_norm": 1.6505805474404067,
      "learning_rate": 1.8240000000000002e-06,
      "loss": 0.1086,
      "step": 114
    },
    {
      "epoch": 0.05523535062439962,
      "grad_norm": 2.027298880252341,
      "learning_rate": 1.8400000000000002e-06,
      "loss": 0.1137,
      "step": 115
    },
    {
      "epoch": 0.05571565802113353,
      "grad_norm": 1.9867450378269047,
      "learning_rate": 1.856e-06,
      "loss": 0.0949,
      "step": 116
    },
    {
      "epoch": 0.056195965417867436,
      "grad_norm": 1.0287961975246656,
      "learning_rate": 1.8720000000000002e-06,
      "loss": 0.0993,
      "step": 117
    },
    {
      "epoch": 0.056676272814601344,
      "grad_norm": 3.6886267076116726,
      "learning_rate": 1.8880000000000002e-06,
      "loss": 0.0734,
      "step": 118
    },
    {
      "epoch": 0.05715658021133525,
      "grad_norm": 1.4209403055185024,
      "learning_rate": 1.9040000000000003e-06,
      "loss": 0.088,
      "step": 119
    },
    {
      "epoch": 0.05763688760806916,
      "grad_norm": 1.5252469546994676,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 0.1195,
      "step": 120
    },
    {
      "epoch": 0.05811719500480307,
      "grad_norm": 1.2043288229655742,
      "learning_rate": 1.936e-06,
      "loss": 0.0929,
      "step": 121
    },
    {
      "epoch": 0.05859750240153699,
      "grad_norm": 4.538691137978197,
      "learning_rate": 1.952e-06,
      "loss": 0.0773,
      "step": 122
    },
    {
      "epoch": 0.059077809798270896,
      "grad_norm": 1.6211192048515701,
      "learning_rate": 1.968e-06,
      "loss": 0.1006,
      "step": 123
    },
    {
      "epoch": 0.059558117195004805,
      "grad_norm": 1.1742642998496757,
      "learning_rate": 1.984e-06,
      "loss": 0.0682,
      "step": 124
    },
    {
      "epoch": 0.060038424591738714,
      "grad_norm": 1.3758629673989613,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0732,
      "step": 125
    },
    {
      "epoch": 0.06051873198847262,
      "grad_norm": 1.4595665452207671,
      "learning_rate": 2.0160000000000003e-06,
      "loss": 0.0885,
      "step": 126
    },
    {
      "epoch": 0.06099903938520653,
      "grad_norm": 1.2343001806608365,
      "learning_rate": 2.032e-06,
      "loss": 0.0871,
      "step": 127
    },
    {
      "epoch": 0.06147934678194044,
      "grad_norm": 1.4944430814568856,
      "learning_rate": 2.048e-06,
      "loss": 0.0987,
      "step": 128
    },
    {
      "epoch": 0.06195965417867435,
      "grad_norm": 1.1669826735751045,
      "learning_rate": 2.064e-06,
      "loss": 0.0828,
      "step": 129
    },
    {
      "epoch": 0.06243996157540826,
      "grad_norm": 2.050815392469096,
      "learning_rate": 2.08e-06,
      "loss": 0.1436,
      "step": 130
    },
    {
      "epoch": 0.06292026897214217,
      "grad_norm": 2.6189886490073837,
      "learning_rate": 2.0960000000000003e-06,
      "loss": 0.1121,
      "step": 131
    },
    {
      "epoch": 0.06340057636887608,
      "grad_norm": 1.465151382295218,
      "learning_rate": 2.1120000000000003e-06,
      "loss": 0.1569,
      "step": 132
    },
    {
      "epoch": 0.06388088376560999,
      "grad_norm": 1.6059405590984388,
      "learning_rate": 2.128e-06,
      "loss": 0.0891,
      "step": 133
    },
    {
      "epoch": 0.0643611911623439,
      "grad_norm": 1.3606090814821632,
      "learning_rate": 2.144e-06,
      "loss": 0.104,
      "step": 134
    },
    {
      "epoch": 0.06484149855907781,
      "grad_norm": 1.1881719471401748,
      "learning_rate": 2.16e-06,
      "loss": 0.0775,
      "step": 135
    },
    {
      "epoch": 0.06532180595581172,
      "grad_norm": 1.205747071113525,
      "learning_rate": 2.176e-06,
      "loss": 0.084,
      "step": 136
    },
    {
      "epoch": 0.06580211335254563,
      "grad_norm": 1.2301202707311236,
      "learning_rate": 2.1920000000000004e-06,
      "loss": 0.0647,
      "step": 137
    },
    {
      "epoch": 0.06628242074927954,
      "grad_norm": 1.154403895526943,
      "learning_rate": 2.2080000000000003e-06,
      "loss": 0.0884,
      "step": 138
    },
    {
      "epoch": 0.06676272814601344,
      "grad_norm": 1.1013298512945302,
      "learning_rate": 2.2240000000000002e-06,
      "loss": 0.0643,
      "step": 139
    },
    {
      "epoch": 0.06724303554274735,
      "grad_norm": 1.0932013932642193,
      "learning_rate": 2.24e-06,
      "loss": 0.092,
      "step": 140
    },
    {
      "epoch": 0.06772334293948126,
      "grad_norm": 1.0454452002652364,
      "learning_rate": 2.256e-06,
      "loss": 0.0799,
      "step": 141
    },
    {
      "epoch": 0.06820365033621517,
      "grad_norm": 1.4743566119826792,
      "learning_rate": 2.2720000000000004e-06,
      "loss": 0.1136,
      "step": 142
    },
    {
      "epoch": 0.0686839577329491,
      "grad_norm": 1.4039096246861968,
      "learning_rate": 2.2880000000000004e-06,
      "loss": 0.0909,
      "step": 143
    },
    {
      "epoch": 0.069164265129683,
      "grad_norm": 1.2657755183130568,
      "learning_rate": 2.3040000000000003e-06,
      "loss": 0.1117,
      "step": 144
    },
    {
      "epoch": 0.06964457252641691,
      "grad_norm": 1.5851847672258597,
      "learning_rate": 2.3200000000000002e-06,
      "loss": 0.0951,
      "step": 145
    },
    {
      "epoch": 0.07012487992315082,
      "grad_norm": 1.2581561491117204,
      "learning_rate": 2.336e-06,
      "loss": 0.0878,
      "step": 146
    },
    {
      "epoch": 0.07060518731988473,
      "grad_norm": 1.420979749511269,
      "learning_rate": 2.352e-06,
      "loss": 0.0936,
      "step": 147
    },
    {
      "epoch": 0.07108549471661864,
      "grad_norm": 1.3031371111764456,
      "learning_rate": 2.3680000000000005e-06,
      "loss": 0.0909,
      "step": 148
    },
    {
      "epoch": 0.07156580211335255,
      "grad_norm": 1.4573325610550967,
      "learning_rate": 2.3840000000000004e-06,
      "loss": 0.0827,
      "step": 149
    },
    {
      "epoch": 0.07204610951008646,
      "grad_norm": 1.3047911271077164,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.0851,
      "step": 150
    },
    {
      "epoch": 0.07252641690682037,
      "grad_norm": 1.4176849309242068,
      "learning_rate": 2.4160000000000002e-06,
      "loss": 0.0831,
      "step": 151
    },
    {
      "epoch": 0.07300672430355427,
      "grad_norm": 2.5041223053166903,
      "learning_rate": 2.432e-06,
      "loss": 0.0963,
      "step": 152
    },
    {
      "epoch": 0.07348703170028818,
      "grad_norm": 1.2243042228153846,
      "learning_rate": 2.448e-06,
      "loss": 0.092,
      "step": 153
    },
    {
      "epoch": 0.07396733909702209,
      "grad_norm": 1.3595133894472542,
      "learning_rate": 2.4640000000000005e-06,
      "loss": 0.1088,
      "step": 154
    },
    {
      "epoch": 0.074447646493756,
      "grad_norm": 1.8046494765641843,
      "learning_rate": 2.4800000000000004e-06,
      "loss": 0.1067,
      "step": 155
    },
    {
      "epoch": 0.07492795389048991,
      "grad_norm": 1.3961366152512769,
      "learning_rate": 2.496e-06,
      "loss": 0.1066,
      "step": 156
    },
    {
      "epoch": 0.07540826128722382,
      "grad_norm": 1.569757837719253,
      "learning_rate": 2.512e-06,
      "loss": 0.116,
      "step": 157
    },
    {
      "epoch": 0.07588856868395773,
      "grad_norm": 1.2274453659822118,
      "learning_rate": 2.5280000000000006e-06,
      "loss": 0.0888,
      "step": 158
    },
    {
      "epoch": 0.07636887608069164,
      "grad_norm": 1.3105009460198556,
      "learning_rate": 2.5440000000000005e-06,
      "loss": 0.109,
      "step": 159
    },
    {
      "epoch": 0.07684918347742556,
      "grad_norm": 1.1721145729988776,
      "learning_rate": 2.56e-06,
      "loss": 0.0994,
      "step": 160
    },
    {
      "epoch": 0.07732949087415947,
      "grad_norm": 1.2096805266515924,
      "learning_rate": 2.576e-06,
      "loss": 0.0887,
      "step": 161
    },
    {
      "epoch": 0.07780979827089338,
      "grad_norm": 2.25466141210399,
      "learning_rate": 2.592e-06,
      "loss": 0.0977,
      "step": 162
    },
    {
      "epoch": 0.07829010566762729,
      "grad_norm": 1.2284935839699218,
      "learning_rate": 2.608e-06,
      "loss": 0.0991,
      "step": 163
    },
    {
      "epoch": 0.0787704130643612,
      "grad_norm": 1.1932140220390712,
      "learning_rate": 2.6240000000000006e-06,
      "loss": 0.0866,
      "step": 164
    },
    {
      "epoch": 0.0792507204610951,
      "grad_norm": 1.0789509823335703,
      "learning_rate": 2.64e-06,
      "loss": 0.0996,
      "step": 165
    },
    {
      "epoch": 0.07973102785782901,
      "grad_norm": 0.9371763943071141,
      "learning_rate": 2.656e-06,
      "loss": 0.0582,
      "step": 166
    },
    {
      "epoch": 0.08021133525456292,
      "grad_norm": 1.7368430972984232,
      "learning_rate": 2.672e-06,
      "loss": 0.1086,
      "step": 167
    },
    {
      "epoch": 0.08069164265129683,
      "grad_norm": 1.6755808286564804,
      "learning_rate": 2.688e-06,
      "loss": 0.0873,
      "step": 168
    },
    {
      "epoch": 0.08117195004803074,
      "grad_norm": 1.3121940443924218,
      "learning_rate": 2.704e-06,
      "loss": 0.0805,
      "step": 169
    },
    {
      "epoch": 0.08165225744476465,
      "grad_norm": 1.3271980163235704,
      "learning_rate": 2.7200000000000002e-06,
      "loss": 0.0813,
      "step": 170
    },
    {
      "epoch": 0.08213256484149856,
      "grad_norm": 1.1932639416175121,
      "learning_rate": 2.736e-06,
      "loss": 0.0823,
      "step": 171
    },
    {
      "epoch": 0.08261287223823247,
      "grad_norm": 1.1588476048524006,
      "learning_rate": 2.752e-06,
      "loss": 0.0987,
      "step": 172
    },
    {
      "epoch": 0.08309317963496637,
      "grad_norm": 0.9072215194037728,
      "learning_rate": 2.768e-06,
      "loss": 0.067,
      "step": 173
    },
    {
      "epoch": 0.08357348703170028,
      "grad_norm": 1.356010402076568,
      "learning_rate": 2.784e-06,
      "loss": 0.0868,
      "step": 174
    },
    {
      "epoch": 0.08405379442843419,
      "grad_norm": 1.2650989878632668,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.0812,
      "step": 175
    },
    {
      "epoch": 0.0845341018251681,
      "grad_norm": 1.4802205189991515,
      "learning_rate": 2.8160000000000002e-06,
      "loss": 0.099,
      "step": 176
    },
    {
      "epoch": 0.08501440922190202,
      "grad_norm": 1.1417060742721667,
      "learning_rate": 2.832e-06,
      "loss": 0.0815,
      "step": 177
    },
    {
      "epoch": 0.08549471661863593,
      "grad_norm": 1.3705511538151176,
      "learning_rate": 2.848e-06,
      "loss": 0.1102,
      "step": 178
    },
    {
      "epoch": 0.08597502401536984,
      "grad_norm": 1.3163532047784325,
      "learning_rate": 2.864e-06,
      "loss": 0.0755,
      "step": 179
    },
    {
      "epoch": 0.08645533141210375,
      "grad_norm": 1.47238236129405,
      "learning_rate": 2.88e-06,
      "loss": 0.0949,
      "step": 180
    },
    {
      "epoch": 0.08693563880883766,
      "grad_norm": 0.9260065978895016,
      "learning_rate": 2.8960000000000003e-06,
      "loss": 0.0586,
      "step": 181
    },
    {
      "epoch": 0.08741594620557157,
      "grad_norm": 1.311717165904964,
      "learning_rate": 2.9120000000000002e-06,
      "loss": 0.0816,
      "step": 182
    },
    {
      "epoch": 0.08789625360230548,
      "grad_norm": 1.3185427170913515,
      "learning_rate": 2.928e-06,
      "loss": 0.1061,
      "step": 183
    },
    {
      "epoch": 0.08837656099903939,
      "grad_norm": 1.0037072726769718,
      "learning_rate": 2.944e-06,
      "loss": 0.062,
      "step": 184
    },
    {
      "epoch": 0.0888568683957733,
      "grad_norm": 1.0985735941649346,
      "learning_rate": 2.96e-06,
      "loss": 0.0805,
      "step": 185
    },
    {
      "epoch": 0.0893371757925072,
      "grad_norm": 1.3125670658691977,
      "learning_rate": 2.976e-06,
      "loss": 0.0931,
      "step": 186
    },
    {
      "epoch": 0.08981748318924111,
      "grad_norm": 1.0838144485116263,
      "learning_rate": 2.9920000000000003e-06,
      "loss": 0.0834,
      "step": 187
    },
    {
      "epoch": 0.09029779058597502,
      "grad_norm": 1.799900406869807,
      "learning_rate": 3.0080000000000003e-06,
      "loss": 0.0919,
      "step": 188
    },
    {
      "epoch": 0.09077809798270893,
      "grad_norm": 1.173117411096857,
      "learning_rate": 3.024e-06,
      "loss": 0.0721,
      "step": 189
    },
    {
      "epoch": 0.09125840537944284,
      "grad_norm": 1.373335051247715,
      "learning_rate": 3.04e-06,
      "loss": 0.0846,
      "step": 190
    },
    {
      "epoch": 0.09173871277617675,
      "grad_norm": 1.3981230024483795,
      "learning_rate": 3.056e-06,
      "loss": 0.1008,
      "step": 191
    },
    {
      "epoch": 0.09221902017291066,
      "grad_norm": 1.4513236326035954,
      "learning_rate": 3.072e-06,
      "loss": 0.1059,
      "step": 192
    },
    {
      "epoch": 0.09269932756964457,
      "grad_norm": 1.2186053621072894,
      "learning_rate": 3.0880000000000003e-06,
      "loss": 0.09,
      "step": 193
    },
    {
      "epoch": 0.09317963496637849,
      "grad_norm": 1.1068806010368781,
      "learning_rate": 3.1040000000000003e-06,
      "loss": 0.0718,
      "step": 194
    },
    {
      "epoch": 0.0936599423631124,
      "grad_norm": 1.237536933577925,
      "learning_rate": 3.12e-06,
      "loss": 0.0768,
      "step": 195
    },
    {
      "epoch": 0.0941402497598463,
      "grad_norm": 1.2585614650979184,
      "learning_rate": 3.136e-06,
      "loss": 0.0757,
      "step": 196
    },
    {
      "epoch": 0.09462055715658022,
      "grad_norm": 0.9595246063304191,
      "learning_rate": 3.152e-06,
      "loss": 0.0696,
      "step": 197
    },
    {
      "epoch": 0.09510086455331412,
      "grad_norm": 1.1929913387812179,
      "learning_rate": 3.1680000000000004e-06,
      "loss": 0.0867,
      "step": 198
    },
    {
      "epoch": 0.09558117195004803,
      "grad_norm": 1.0572272850541955,
      "learning_rate": 3.1840000000000003e-06,
      "loss": 0.0816,
      "step": 199
    },
    {
      "epoch": 0.09606147934678194,
      "grad_norm": 1.160822305103066,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.085,
      "step": 200
    },
    {
      "epoch": 0.09654178674351585,
      "grad_norm": 1.3295668411317592,
      "learning_rate": 3.216e-06,
      "loss": 0.1003,
      "step": 201
    },
    {
      "epoch": 0.09702209414024976,
      "grad_norm": 1.3117059163085465,
      "learning_rate": 3.232e-06,
      "loss": 0.0814,
      "step": 202
    },
    {
      "epoch": 0.09750240153698367,
      "grad_norm": 1.4613138653883198,
      "learning_rate": 3.248e-06,
      "loss": 0.1041,
      "step": 203
    },
    {
      "epoch": 0.09798270893371758,
      "grad_norm": 1.3385478082564994,
      "learning_rate": 3.2640000000000004e-06,
      "loss": 0.0846,
      "step": 204
    },
    {
      "epoch": 0.09846301633045149,
      "grad_norm": 1.271829712055238,
      "learning_rate": 3.2800000000000004e-06,
      "loss": 0.0986,
      "step": 205
    },
    {
      "epoch": 0.0989433237271854,
      "grad_norm": 1.1382933856150044,
      "learning_rate": 3.2960000000000003e-06,
      "loss": 0.0662,
      "step": 206
    },
    {
      "epoch": 0.0994236311239193,
      "grad_norm": 1.2777612246188284,
      "learning_rate": 3.3120000000000002e-06,
      "loss": 0.0944,
      "step": 207
    },
    {
      "epoch": 0.09990393852065321,
      "grad_norm": 1.2569659513450397,
      "learning_rate": 3.328e-06,
      "loss": 0.0688,
      "step": 208
    },
    {
      "epoch": 0.10038424591738712,
      "grad_norm": 1.2019771035760043,
      "learning_rate": 3.344e-06,
      "loss": 0.092,
      "step": 209
    },
    {
      "epoch": 0.10086455331412104,
      "grad_norm": 1.0705863361886445,
      "learning_rate": 3.3600000000000004e-06,
      "loss": 0.085,
      "step": 210
    },
    {
      "epoch": 0.10134486071085495,
      "grad_norm": 1.1235244108380247,
      "learning_rate": 3.3760000000000004e-06,
      "loss": 0.0798,
      "step": 211
    },
    {
      "epoch": 0.10182516810758886,
      "grad_norm": 1.0830560530059452,
      "learning_rate": 3.3920000000000003e-06,
      "loss": 0.0745,
      "step": 212
    },
    {
      "epoch": 0.10230547550432277,
      "grad_norm": 1.2281697980226327,
      "learning_rate": 3.4080000000000002e-06,
      "loss": 0.0822,
      "step": 213
    },
    {
      "epoch": 0.10278578290105668,
      "grad_norm": 1.0293384591155774,
      "learning_rate": 3.424e-06,
      "loss": 0.0699,
      "step": 214
    },
    {
      "epoch": 0.10326609029779059,
      "grad_norm": 1.1168765812283847,
      "learning_rate": 3.44e-06,
      "loss": 0.0903,
      "step": 215
    },
    {
      "epoch": 0.1037463976945245,
      "grad_norm": 1.2232503580307106,
      "learning_rate": 3.4560000000000005e-06,
      "loss": 0.073,
      "step": 216
    },
    {
      "epoch": 0.1042267050912584,
      "grad_norm": 1.1853245402764723,
      "learning_rate": 3.4720000000000004e-06,
      "loss": 0.0804,
      "step": 217
    },
    {
      "epoch": 0.10470701248799232,
      "grad_norm": 1.0349173303797323,
      "learning_rate": 3.4880000000000003e-06,
      "loss": 0.0791,
      "step": 218
    },
    {
      "epoch": 0.10518731988472622,
      "grad_norm": 0.9123748906672083,
      "learning_rate": 3.5040000000000002e-06,
      "loss": 0.0846,
      "step": 219
    },
    {
      "epoch": 0.10566762728146013,
      "grad_norm": 1.1862034175014786,
      "learning_rate": 3.52e-06,
      "loss": 0.0968,
      "step": 220
    },
    {
      "epoch": 0.10614793467819404,
      "grad_norm": 1.2703272165951818,
      "learning_rate": 3.5360000000000005e-06,
      "loss": 0.106,
      "step": 221
    },
    {
      "epoch": 0.10662824207492795,
      "grad_norm": 0.825924476735835,
      "learning_rate": 3.5520000000000005e-06,
      "loss": 0.0722,
      "step": 222
    },
    {
      "epoch": 0.10710854947166186,
      "grad_norm": 1.0687515686341,
      "learning_rate": 3.5680000000000004e-06,
      "loss": 0.0716,
      "step": 223
    },
    {
      "epoch": 0.10758885686839577,
      "grad_norm": 1.0805753001019858,
      "learning_rate": 3.5840000000000003e-06,
      "loss": 0.0691,
      "step": 224
    },
    {
      "epoch": 0.10806916426512968,
      "grad_norm": 1.6287239686502346,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.067,
      "step": 225
    },
    {
      "epoch": 0.10854947166186359,
      "grad_norm": 1.9814573884606366,
      "learning_rate": 3.616e-06,
      "loss": 0.0752,
      "step": 226
    },
    {
      "epoch": 0.10902977905859751,
      "grad_norm": 1.3374463525520308,
      "learning_rate": 3.6320000000000005e-06,
      "loss": 0.0876,
      "step": 227
    },
    {
      "epoch": 0.10951008645533142,
      "grad_norm": 1.452473041749981,
      "learning_rate": 3.6480000000000005e-06,
      "loss": 0.0788,
      "step": 228
    },
    {
      "epoch": 0.10999039385206533,
      "grad_norm": 0.9767761821599609,
      "learning_rate": 3.6640000000000004e-06,
      "loss": 0.059,
      "step": 229
    },
    {
      "epoch": 0.11047070124879924,
      "grad_norm": 2.0951330953014837,
      "learning_rate": 3.6800000000000003e-06,
      "loss": 0.1257,
      "step": 230
    },
    {
      "epoch": 0.11095100864553314,
      "grad_norm": 1.4760857161454637,
      "learning_rate": 3.6960000000000003e-06,
      "loss": 0.0694,
      "step": 231
    },
    {
      "epoch": 0.11143131604226705,
      "grad_norm": 1.426490577687766,
      "learning_rate": 3.712e-06,
      "loss": 0.078,
      "step": 232
    },
    {
      "epoch": 0.11191162343900096,
      "grad_norm": 1.049787450994003,
      "learning_rate": 3.7280000000000006e-06,
      "loss": 0.0679,
      "step": 233
    },
    {
      "epoch": 0.11239193083573487,
      "grad_norm": 0.8796776572576731,
      "learning_rate": 3.7440000000000005e-06,
      "loss": 0.0607,
      "step": 234
    },
    {
      "epoch": 0.11287223823246878,
      "grad_norm": 1.3931219386969078,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 0.102,
      "step": 235
    },
    {
      "epoch": 0.11335254562920269,
      "grad_norm": 1.393636110259964,
      "learning_rate": 3.7760000000000004e-06,
      "loss": 0.111,
      "step": 236
    },
    {
      "epoch": 0.1138328530259366,
      "grad_norm": 0.9780090176809716,
      "learning_rate": 3.7920000000000003e-06,
      "loss": 0.0947,
      "step": 237
    },
    {
      "epoch": 0.1143131604226705,
      "grad_norm": 1.3598074761781376,
      "learning_rate": 3.8080000000000006e-06,
      "loss": 0.1047,
      "step": 238
    },
    {
      "epoch": 0.11479346781940442,
      "grad_norm": 1.413503721034128,
      "learning_rate": 3.824e-06,
      "loss": 0.0731,
      "step": 239
    },
    {
      "epoch": 0.11527377521613832,
      "grad_norm": 1.0636156845849574,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.0947,
      "step": 240
    },
    {
      "epoch": 0.11575408261287223,
      "grad_norm": 1.214860626930102,
      "learning_rate": 3.856e-06,
      "loss": 0.0652,
      "step": 241
    },
    {
      "epoch": 0.11623439000960614,
      "grad_norm": 1.203838132160498,
      "learning_rate": 3.872e-06,
      "loss": 0.0772,
      "step": 242
    },
    {
      "epoch": 0.11671469740634005,
      "grad_norm": 1.4337623413545681,
      "learning_rate": 3.888e-06,
      "loss": 0.1038,
      "step": 243
    },
    {
      "epoch": 0.11719500480307397,
      "grad_norm": 1.8506896303135723,
      "learning_rate": 3.904e-06,
      "loss": 0.0974,
      "step": 244
    },
    {
      "epoch": 0.11767531219980788,
      "grad_norm": 1.0209624715713772,
      "learning_rate": 3.920000000000001e-06,
      "loss": 0.0668,
      "step": 245
    },
    {
      "epoch": 0.11815561959654179,
      "grad_norm": 1.0430831232311266,
      "learning_rate": 3.936e-06,
      "loss": 0.0693,
      "step": 246
    },
    {
      "epoch": 0.1186359269932757,
      "grad_norm": 0.9781386213920611,
      "learning_rate": 3.9520000000000004e-06,
      "loss": 0.0922,
      "step": 247
    },
    {
      "epoch": 0.11911623439000961,
      "grad_norm": 1.3090670400806688,
      "learning_rate": 3.968e-06,
      "loss": 0.0697,
      "step": 248
    },
    {
      "epoch": 0.11959654178674352,
      "grad_norm": 1.6565002477397985,
      "learning_rate": 3.984e-06,
      "loss": 0.1017,
      "step": 249
    },
    {
      "epoch": 0.12007684918347743,
      "grad_norm": 1.013045916484816,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0601,
      "step": 250
    },
    {
      "epoch": 0.12055715658021134,
      "grad_norm": 0.9922526880061665,
      "learning_rate": 4.016e-06,
      "loss": 0.0567,
      "step": 251
    },
    {
      "epoch": 0.12103746397694524,
      "grad_norm": 1.1410675081208677,
      "learning_rate": 4.0320000000000005e-06,
      "loss": 0.0766,
      "step": 252
    },
    {
      "epoch": 0.12151777137367915,
      "grad_norm": 1.3175828458287309,
      "learning_rate": 4.048e-06,
      "loss": 0.1039,
      "step": 253
    },
    {
      "epoch": 0.12199807877041306,
      "grad_norm": 1.3696186687199499,
      "learning_rate": 4.064e-06,
      "loss": 0.068,
      "step": 254
    },
    {
      "epoch": 0.12247838616714697,
      "grad_norm": 1.5050717359950978,
      "learning_rate": 4.08e-06,
      "loss": 0.0797,
      "step": 255
    },
    {
      "epoch": 0.12295869356388088,
      "grad_norm": 1.3400957977202135,
      "learning_rate": 4.096e-06,
      "loss": 0.0999,
      "step": 256
    },
    {
      "epoch": 0.12343900096061479,
      "grad_norm": 1.2617471427133966,
      "learning_rate": 4.112000000000001e-06,
      "loss": 0.0822,
      "step": 257
    },
    {
      "epoch": 0.1239193083573487,
      "grad_norm": 1.267747053897931,
      "learning_rate": 4.128e-06,
      "loss": 0.0773,
      "step": 258
    },
    {
      "epoch": 0.12439961575408261,
      "grad_norm": 1.2676946090954653,
      "learning_rate": 4.1440000000000005e-06,
      "loss": 0.0754,
      "step": 259
    },
    {
      "epoch": 0.12487992315081652,
      "grad_norm": 1.2292713163126525,
      "learning_rate": 4.16e-06,
      "loss": 0.09,
      "step": 260
    },
    {
      "epoch": 0.12536023054755044,
      "grad_norm": 1.0012628385996274,
      "learning_rate": 4.176e-06,
      "loss": 0.0697,
      "step": 261
    },
    {
      "epoch": 0.12584053794428435,
      "grad_norm": 1.3874865197803345,
      "learning_rate": 4.192000000000001e-06,
      "loss": 0.0932,
      "step": 262
    },
    {
      "epoch": 0.12632084534101826,
      "grad_norm": 1.177205882683919,
      "learning_rate": 4.208e-06,
      "loss": 0.0664,
      "step": 263
    },
    {
      "epoch": 0.12680115273775217,
      "grad_norm": 1.2350489439494747,
      "learning_rate": 4.2240000000000006e-06,
      "loss": 0.0898,
      "step": 264
    },
    {
      "epoch": 0.12728146013448607,
      "grad_norm": 1.1559020122519823,
      "learning_rate": 4.24e-06,
      "loss": 0.0786,
      "step": 265
    },
    {
      "epoch": 0.12776176753121998,
      "grad_norm": 1.3539373299195447,
      "learning_rate": 4.256e-06,
      "loss": 0.0728,
      "step": 266
    },
    {
      "epoch": 0.1282420749279539,
      "grad_norm": 1.2165821416393527,
      "learning_rate": 4.272000000000001e-06,
      "loss": 0.0665,
      "step": 267
    },
    {
      "epoch": 0.1287223823246878,
      "grad_norm": 0.8998352503566164,
      "learning_rate": 4.288e-06,
      "loss": 0.0572,
      "step": 268
    },
    {
      "epoch": 0.1292026897214217,
      "grad_norm": 1.332170024000591,
      "learning_rate": 4.304000000000001e-06,
      "loss": 0.0896,
      "step": 269
    },
    {
      "epoch": 0.12968299711815562,
      "grad_norm": 0.8702695667096485,
      "learning_rate": 4.32e-06,
      "loss": 0.0579,
      "step": 270
    },
    {
      "epoch": 0.13016330451488953,
      "grad_norm": 1.1346194333159956,
      "learning_rate": 4.3360000000000005e-06,
      "loss": 0.0809,
      "step": 271
    },
    {
      "epoch": 0.13064361191162344,
      "grad_norm": 1.3242195848254044,
      "learning_rate": 4.352e-06,
      "loss": 0.0953,
      "step": 272
    },
    {
      "epoch": 0.13112391930835735,
      "grad_norm": 0.947922780884019,
      "learning_rate": 4.368e-06,
      "loss": 0.0726,
      "step": 273
    },
    {
      "epoch": 0.13160422670509125,
      "grad_norm": 0.8590229026058185,
      "learning_rate": 4.384000000000001e-06,
      "loss": 0.0859,
      "step": 274
    },
    {
      "epoch": 0.13208453410182516,
      "grad_norm": 2.0364307950864897,
      "learning_rate": 4.4e-06,
      "loss": 0.1065,
      "step": 275
    },
    {
      "epoch": 0.13256484149855907,
      "grad_norm": 0.9495154556473616,
      "learning_rate": 4.416000000000001e-06,
      "loss": 0.0539,
      "step": 276
    },
    {
      "epoch": 0.13304514889529298,
      "grad_norm": 1.274982611643209,
      "learning_rate": 4.432e-06,
      "loss": 0.0847,
      "step": 277
    },
    {
      "epoch": 0.1335254562920269,
      "grad_norm": 1.0179278688325417,
      "learning_rate": 4.4480000000000004e-06,
      "loss": 0.0781,
      "step": 278
    },
    {
      "epoch": 0.1340057636887608,
      "grad_norm": 1.0777338379659434,
      "learning_rate": 4.464000000000001e-06,
      "loss": 0.0823,
      "step": 279
    },
    {
      "epoch": 0.1344860710854947,
      "grad_norm": 1.7837833704627275,
      "learning_rate": 4.48e-06,
      "loss": 0.0809,
      "step": 280
    },
    {
      "epoch": 0.13496637848222862,
      "grad_norm": 1.1491936699966703,
      "learning_rate": 4.496000000000001e-06,
      "loss": 0.0726,
      "step": 281
    },
    {
      "epoch": 0.13544668587896252,
      "grad_norm": 1.1891782145869991,
      "learning_rate": 4.512e-06,
      "loss": 0.0872,
      "step": 282
    },
    {
      "epoch": 0.13592699327569643,
      "grad_norm": 0.8682236854456676,
      "learning_rate": 4.5280000000000005e-06,
      "loss": 0.0739,
      "step": 283
    },
    {
      "epoch": 0.13640730067243034,
      "grad_norm": 2.941844018714875,
      "learning_rate": 4.544000000000001e-06,
      "loss": 0.0862,
      "step": 284
    },
    {
      "epoch": 0.13688760806916425,
      "grad_norm": 1.1758035570138663,
      "learning_rate": 4.56e-06,
      "loss": 0.094,
      "step": 285
    },
    {
      "epoch": 0.1373679154658982,
      "grad_norm": 1.213798829069968,
      "learning_rate": 4.576000000000001e-06,
      "loss": 0.0582,
      "step": 286
    },
    {
      "epoch": 0.1378482228626321,
      "grad_norm": 1.1521708031630107,
      "learning_rate": 4.592e-06,
      "loss": 0.0961,
      "step": 287
    },
    {
      "epoch": 0.138328530259366,
      "grad_norm": 1.5167984093098619,
      "learning_rate": 4.608000000000001e-06,
      "loss": 0.0886,
      "step": 288
    },
    {
      "epoch": 0.13880883765609991,
      "grad_norm": 1.334195306364946,
      "learning_rate": 4.624e-06,
      "loss": 0.0926,
      "step": 289
    },
    {
      "epoch": 0.13928914505283382,
      "grad_norm": 1.201794873892803,
      "learning_rate": 4.6400000000000005e-06,
      "loss": 0.0714,
      "step": 290
    },
    {
      "epoch": 0.13976945244956773,
      "grad_norm": 1.1161034977945354,
      "learning_rate": 4.656000000000001e-06,
      "loss": 0.0835,
      "step": 291
    },
    {
      "epoch": 0.14024975984630164,
      "grad_norm": 1.1187497922543908,
      "learning_rate": 4.672e-06,
      "loss": 0.0768,
      "step": 292
    },
    {
      "epoch": 0.14073006724303555,
      "grad_norm": 1.135592941087255,
      "learning_rate": 4.688000000000001e-06,
      "loss": 0.0987,
      "step": 293
    },
    {
      "epoch": 0.14121037463976946,
      "grad_norm": 1.0116456228936113,
      "learning_rate": 4.704e-06,
      "loss": 0.0868,
      "step": 294
    },
    {
      "epoch": 0.14169068203650337,
      "grad_norm": 0.8988345290651037,
      "learning_rate": 4.7200000000000005e-06,
      "loss": 0.0729,
      "step": 295
    },
    {
      "epoch": 0.14217098943323728,
      "grad_norm": 1.4920548715651283,
      "learning_rate": 4.736000000000001e-06,
      "loss": 0.0689,
      "step": 296
    },
    {
      "epoch": 0.14265129682997119,
      "grad_norm": 1.1491177326584943,
      "learning_rate": 4.752e-06,
      "loss": 0.0822,
      "step": 297
    },
    {
      "epoch": 0.1431316042267051,
      "grad_norm": 1.3710614031545738,
      "learning_rate": 4.768000000000001e-06,
      "loss": 0.1054,
      "step": 298
    },
    {
      "epoch": 0.143611911623439,
      "grad_norm": 0.798935770934357,
      "learning_rate": 4.784e-06,
      "loss": 0.0609,
      "step": 299
    },
    {
      "epoch": 0.1440922190201729,
      "grad_norm": 1.0553443465463967,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.0588,
      "step": 300
    },
    {
      "epoch": 0.14457252641690682,
      "grad_norm": 0.862660611904342,
      "learning_rate": 4.816e-06,
      "loss": 0.0607,
      "step": 301
    },
    {
      "epoch": 0.14505283381364073,
      "grad_norm": 1.0888178319055062,
      "learning_rate": 4.8320000000000005e-06,
      "loss": 0.0888,
      "step": 302
    },
    {
      "epoch": 0.14553314121037464,
      "grad_norm": 0.8383673824701234,
      "learning_rate": 4.848000000000001e-06,
      "loss": 0.0474,
      "step": 303
    },
    {
      "epoch": 0.14601344860710855,
      "grad_norm": 0.946622673839637,
      "learning_rate": 4.864e-06,
      "loss": 0.0644,
      "step": 304
    },
    {
      "epoch": 0.14649375600384246,
      "grad_norm": 0.9082218680745726,
      "learning_rate": 4.880000000000001e-06,
      "loss": 0.0639,
      "step": 305
    },
    {
      "epoch": 0.14697406340057637,
      "grad_norm": 1.3359634218439083,
      "learning_rate": 4.896e-06,
      "loss": 0.0808,
      "step": 306
    },
    {
      "epoch": 0.14745437079731027,
      "grad_norm": 2.4941362749992577,
      "learning_rate": 4.9120000000000006e-06,
      "loss": 0.1029,
      "step": 307
    },
    {
      "epoch": 0.14793467819404418,
      "grad_norm": 0.8768474072222495,
      "learning_rate": 4.928000000000001e-06,
      "loss": 0.0603,
      "step": 308
    },
    {
      "epoch": 0.1484149855907781,
      "grad_norm": 1.0191777482177848,
      "learning_rate": 4.9440000000000004e-06,
      "loss": 0.1195,
      "step": 309
    },
    {
      "epoch": 0.148895292987512,
      "grad_norm": 1.3776289338364824,
      "learning_rate": 4.960000000000001e-06,
      "loss": 0.0929,
      "step": 310
    },
    {
      "epoch": 0.1493756003842459,
      "grad_norm": 1.118324563424117,
      "learning_rate": 4.976e-06,
      "loss": 0.0994,
      "step": 311
    },
    {
      "epoch": 0.14985590778097982,
      "grad_norm": 0.94374520838527,
      "learning_rate": 4.992e-06,
      "loss": 0.0641,
      "step": 312
    },
    {
      "epoch": 0.15033621517771373,
      "grad_norm": 1.2295166440217464,
      "learning_rate": 5.008000000000001e-06,
      "loss": 0.0764,
      "step": 313
    },
    {
      "epoch": 0.15081652257444764,
      "grad_norm": 1.0038813920972378,
      "learning_rate": 5.024e-06,
      "loss": 0.0894,
      "step": 314
    },
    {
      "epoch": 0.15129682997118155,
      "grad_norm": 0.9564388991346412,
      "learning_rate": 5.04e-06,
      "loss": 0.0646,
      "step": 315
    },
    {
      "epoch": 0.15177713736791545,
      "grad_norm": 1.001140276754152,
      "learning_rate": 5.056000000000001e-06,
      "loss": 0.0665,
      "step": 316
    },
    {
      "epoch": 0.15225744476464936,
      "grad_norm": 1.6640587209098594,
      "learning_rate": 5.072e-06,
      "loss": 0.0996,
      "step": 317
    },
    {
      "epoch": 0.15273775216138327,
      "grad_norm": 0.9225089106833559,
      "learning_rate": 5.088000000000001e-06,
      "loss": 0.0636,
      "step": 318
    },
    {
      "epoch": 0.15321805955811718,
      "grad_norm": 0.8945005100522845,
      "learning_rate": 5.104e-06,
      "loss": 0.0769,
      "step": 319
    },
    {
      "epoch": 0.15369836695485112,
      "grad_norm": 1.7987667845062436,
      "learning_rate": 5.12e-06,
      "loss": 0.1081,
      "step": 320
    },
    {
      "epoch": 0.15417867435158503,
      "grad_norm": 1.255685447274647,
      "learning_rate": 5.136e-06,
      "loss": 0.0614,
      "step": 321
    },
    {
      "epoch": 0.15465898174831894,
      "grad_norm": 0.9074627196357872,
      "learning_rate": 5.152e-06,
      "loss": 0.0713,
      "step": 322
    },
    {
      "epoch": 0.15513928914505284,
      "grad_norm": 0.8841836941406669,
      "learning_rate": 5.168000000000001e-06,
      "loss": 0.0644,
      "step": 323
    },
    {
      "epoch": 0.15561959654178675,
      "grad_norm": 1.119142780438519,
      "learning_rate": 5.184e-06,
      "loss": 0.1067,
      "step": 324
    },
    {
      "epoch": 0.15609990393852066,
      "grad_norm": 0.7376552219638878,
      "learning_rate": 5.2e-06,
      "loss": 0.0584,
      "step": 325
    },
    {
      "epoch": 0.15658021133525457,
      "grad_norm": 0.7093197189610518,
      "learning_rate": 5.216e-06,
      "loss": 0.067,
      "step": 326
    },
    {
      "epoch": 0.15706051873198848,
      "grad_norm": 0.9146977097269845,
      "learning_rate": 5.232e-06,
      "loss": 0.0878,
      "step": 327
    },
    {
      "epoch": 0.1575408261287224,
      "grad_norm": 0.893881736147892,
      "learning_rate": 5.248000000000001e-06,
      "loss": 0.0749,
      "step": 328
    },
    {
      "epoch": 0.1580211335254563,
      "grad_norm": 1.0133364596640484,
      "learning_rate": 5.264e-06,
      "loss": 0.0669,
      "step": 329
    },
    {
      "epoch": 0.1585014409221902,
      "grad_norm": 1.0188513118607696,
      "learning_rate": 5.28e-06,
      "loss": 0.0744,
      "step": 330
    },
    {
      "epoch": 0.15898174831892412,
      "grad_norm": 1.3769922031431308,
      "learning_rate": 5.296e-06,
      "loss": 0.0804,
      "step": 331
    },
    {
      "epoch": 0.15946205571565802,
      "grad_norm": 1.0924350131956813,
      "learning_rate": 5.312e-06,
      "loss": 0.073,
      "step": 332
    },
    {
      "epoch": 0.15994236311239193,
      "grad_norm": 1.1474380032242861,
      "learning_rate": 5.328000000000001e-06,
      "loss": 0.0748,
      "step": 333
    },
    {
      "epoch": 0.16042267050912584,
      "grad_norm": 1.2007135376136377,
      "learning_rate": 5.344e-06,
      "loss": 0.0633,
      "step": 334
    },
    {
      "epoch": 0.16090297790585975,
      "grad_norm": 0.8615599372907458,
      "learning_rate": 5.36e-06,
      "loss": 0.0565,
      "step": 335
    },
    {
      "epoch": 0.16138328530259366,
      "grad_norm": 1.239088822730045,
      "learning_rate": 5.376e-06,
      "loss": 0.0774,
      "step": 336
    },
    {
      "epoch": 0.16186359269932757,
      "grad_norm": 1.2276765585158387,
      "learning_rate": 5.392e-06,
      "loss": 0.0798,
      "step": 337
    },
    {
      "epoch": 0.16234390009606148,
      "grad_norm": 1.4151473611051535,
      "learning_rate": 5.408e-06,
      "loss": 0.0946,
      "step": 338
    },
    {
      "epoch": 0.1628242074927954,
      "grad_norm": 1.55213536433148,
      "learning_rate": 5.424e-06,
      "loss": 0.081,
      "step": 339
    },
    {
      "epoch": 0.1633045148895293,
      "grad_norm": 1.2190236388499174,
      "learning_rate": 5.4400000000000004e-06,
      "loss": 0.1009,
      "step": 340
    },
    {
      "epoch": 0.1637848222862632,
      "grad_norm": 0.9398509713441388,
      "learning_rate": 5.456e-06,
      "loss": 0.0597,
      "step": 341
    },
    {
      "epoch": 0.1642651296829971,
      "grad_norm": 2.110512888312876,
      "learning_rate": 5.472e-06,
      "loss": 0.1204,
      "step": 342
    },
    {
      "epoch": 0.16474543707973102,
      "grad_norm": 1.05455486733766,
      "learning_rate": 5.488e-06,
      "loss": 0.0927,
      "step": 343
    },
    {
      "epoch": 0.16522574447646493,
      "grad_norm": 0.6714424760276676,
      "learning_rate": 5.504e-06,
      "loss": 0.0605,
      "step": 344
    },
    {
      "epoch": 0.16570605187319884,
      "grad_norm": 0.8868751693597045,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 0.0592,
      "step": 345
    },
    {
      "epoch": 0.16618635926993275,
      "grad_norm": 1.3695330733027493,
      "learning_rate": 5.536e-06,
      "loss": 0.0836,
      "step": 346
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.6064272072499131,
      "learning_rate": 5.552e-06,
      "loss": 0.051,
      "step": 347
    },
    {
      "epoch": 0.16714697406340057,
      "grad_norm": 1.1286927064113377,
      "learning_rate": 5.568e-06,
      "loss": 0.0813,
      "step": 348
    },
    {
      "epoch": 0.16762728146013448,
      "grad_norm": 2.9709430641912284,
      "learning_rate": 5.584e-06,
      "loss": 0.0718,
      "step": 349
    },
    {
      "epoch": 0.16810758885686838,
      "grad_norm": 1.1789706938875752,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.0912,
      "step": 350
    },
    {
      "epoch": 0.1685878962536023,
      "grad_norm": 1.0699393776208648,
      "learning_rate": 5.616e-06,
      "loss": 0.0832,
      "step": 351
    },
    {
      "epoch": 0.1690682036503362,
      "grad_norm": 1.4723575935028883,
      "learning_rate": 5.6320000000000005e-06,
      "loss": 0.0807,
      "step": 352
    },
    {
      "epoch": 0.16954851104707014,
      "grad_norm": 0.9417323601320572,
      "learning_rate": 5.648e-06,
      "loss": 0.0745,
      "step": 353
    },
    {
      "epoch": 0.17002881844380405,
      "grad_norm": 1.0015388420248077,
      "learning_rate": 5.664e-06,
      "loss": 0.0742,
      "step": 354
    },
    {
      "epoch": 0.17050912584053796,
      "grad_norm": 0.846437738871243,
      "learning_rate": 5.68e-06,
      "loss": 0.0738,
      "step": 355
    },
    {
      "epoch": 0.17098943323727187,
      "grad_norm": 1.0340220342143194,
      "learning_rate": 5.696e-06,
      "loss": 0.1003,
      "step": 356
    },
    {
      "epoch": 0.17146974063400577,
      "grad_norm": 1.050210067042245,
      "learning_rate": 5.7120000000000005e-06,
      "loss": 0.0944,
      "step": 357
    },
    {
      "epoch": 0.17195004803073968,
      "grad_norm": 1.1441788701822333,
      "learning_rate": 5.728e-06,
      "loss": 0.0935,
      "step": 358
    },
    {
      "epoch": 0.1724303554274736,
      "grad_norm": 0.8578595357010815,
      "learning_rate": 5.744e-06,
      "loss": 0.065,
      "step": 359
    },
    {
      "epoch": 0.1729106628242075,
      "grad_norm": 1.0827485015540093,
      "learning_rate": 5.76e-06,
      "loss": 0.061,
      "step": 360
    },
    {
      "epoch": 0.1733909702209414,
      "grad_norm": 0.9247919383729161,
      "learning_rate": 5.776e-06,
      "loss": 0.0566,
      "step": 361
    },
    {
      "epoch": 0.17387127761767532,
      "grad_norm": 0.9359918484086721,
      "learning_rate": 5.792000000000001e-06,
      "loss": 0.0918,
      "step": 362
    },
    {
      "epoch": 0.17435158501440923,
      "grad_norm": 0.6461989395520733,
      "learning_rate": 5.808e-06,
      "loss": 0.0596,
      "step": 363
    },
    {
      "epoch": 0.17483189241114314,
      "grad_norm": 1.2658049954240356,
      "learning_rate": 5.8240000000000005e-06,
      "loss": 0.0921,
      "step": 364
    },
    {
      "epoch": 0.17531219980787704,
      "grad_norm": 1.3185639772784372,
      "learning_rate": 5.84e-06,
      "loss": 0.0949,
      "step": 365
    },
    {
      "epoch": 0.17579250720461095,
      "grad_norm": 1.147542350202481,
      "learning_rate": 5.856e-06,
      "loss": 0.0644,
      "step": 366
    },
    {
      "epoch": 0.17627281460134486,
      "grad_norm": 1.2428899445451909,
      "learning_rate": 5.872000000000001e-06,
      "loss": 0.0923,
      "step": 367
    },
    {
      "epoch": 0.17675312199807877,
      "grad_norm": 1.0123606541011576,
      "learning_rate": 5.888e-06,
      "loss": 0.062,
      "step": 368
    },
    {
      "epoch": 0.17723342939481268,
      "grad_norm": 1.2676040664006363,
      "learning_rate": 5.9040000000000006e-06,
      "loss": 0.0993,
      "step": 369
    },
    {
      "epoch": 0.1777137367915466,
      "grad_norm": 1.7674197429016132,
      "learning_rate": 5.92e-06,
      "loss": 0.1028,
      "step": 370
    },
    {
      "epoch": 0.1781940441882805,
      "grad_norm": 1.3206360765520475,
      "learning_rate": 5.9360000000000004e-06,
      "loss": 0.0909,
      "step": 371
    },
    {
      "epoch": 0.1786743515850144,
      "grad_norm": 1.397633543771018,
      "learning_rate": 5.952e-06,
      "loss": 0.0922,
      "step": 372
    },
    {
      "epoch": 0.17915465898174832,
      "grad_norm": 1.4356961352059268,
      "learning_rate": 5.968e-06,
      "loss": 0.0789,
      "step": 373
    },
    {
      "epoch": 0.17963496637848222,
      "grad_norm": 1.0525284488409845,
      "learning_rate": 5.984000000000001e-06,
      "loss": 0.0883,
      "step": 374
    },
    {
      "epoch": 0.18011527377521613,
      "grad_norm": 1.6818878970224407,
      "learning_rate": 6e-06,
      "loss": 0.0957,
      "step": 375
    },
    {
      "epoch": 0.18059558117195004,
      "grad_norm": 1.0922291599097236,
      "learning_rate": 6.0160000000000005e-06,
      "loss": 0.0648,
      "step": 376
    },
    {
      "epoch": 0.18107588856868395,
      "grad_norm": 0.9542123025634616,
      "learning_rate": 6.032e-06,
      "loss": 0.0799,
      "step": 377
    },
    {
      "epoch": 0.18155619596541786,
      "grad_norm": 1.173846664966484,
      "learning_rate": 6.048e-06,
      "loss": 0.0663,
      "step": 378
    },
    {
      "epoch": 0.18203650336215177,
      "grad_norm": 0.8431515012151173,
      "learning_rate": 6.064000000000001e-06,
      "loss": 0.0746,
      "step": 379
    },
    {
      "epoch": 0.18251681075888568,
      "grad_norm": 0.8101682237804423,
      "learning_rate": 6.08e-06,
      "loss": 0.0692,
      "step": 380
    },
    {
      "epoch": 0.1829971181556196,
      "grad_norm": 0.8310188133668261,
      "learning_rate": 6.096000000000001e-06,
      "loss": 0.072,
      "step": 381
    },
    {
      "epoch": 0.1834774255523535,
      "grad_norm": 1.275506422840803,
      "learning_rate": 6.112e-06,
      "loss": 0.1017,
      "step": 382
    },
    {
      "epoch": 0.1839577329490874,
      "grad_norm": 1.2320868444882946,
      "learning_rate": 6.1280000000000005e-06,
      "loss": 0.0851,
      "step": 383
    },
    {
      "epoch": 0.1844380403458213,
      "grad_norm": 0.9892285232343796,
      "learning_rate": 6.144e-06,
      "loss": 0.0902,
      "step": 384
    },
    {
      "epoch": 0.18491834774255522,
      "grad_norm": 1.2195514742095586,
      "learning_rate": 6.16e-06,
      "loss": 0.0784,
      "step": 385
    },
    {
      "epoch": 0.18539865513928913,
      "grad_norm": 0.7206845127032119,
      "learning_rate": 6.176000000000001e-06,
      "loss": 0.0486,
      "step": 386
    },
    {
      "epoch": 0.18587896253602307,
      "grad_norm": 0.9671084909417331,
      "learning_rate": 6.192e-06,
      "loss": 0.0772,
      "step": 387
    },
    {
      "epoch": 0.18635926993275698,
      "grad_norm": 1.0054776144188855,
      "learning_rate": 6.2080000000000005e-06,
      "loss": 0.0693,
      "step": 388
    },
    {
      "epoch": 0.18683957732949089,
      "grad_norm": 0.9245309107680109,
      "learning_rate": 6.224e-06,
      "loss": 0.0739,
      "step": 389
    },
    {
      "epoch": 0.1873198847262248,
      "grad_norm": 1.0103439605117195,
      "learning_rate": 6.24e-06,
      "loss": 0.0853,
      "step": 390
    },
    {
      "epoch": 0.1878001921229587,
      "grad_norm": 0.7676096813994673,
      "learning_rate": 6.256000000000001e-06,
      "loss": 0.0765,
      "step": 391
    },
    {
      "epoch": 0.1882804995196926,
      "grad_norm": 0.9670272222035949,
      "learning_rate": 6.272e-06,
      "loss": 0.073,
      "step": 392
    },
    {
      "epoch": 0.18876080691642652,
      "grad_norm": 1.5221890168980567,
      "learning_rate": 6.288000000000001e-06,
      "loss": 0.0796,
      "step": 393
    },
    {
      "epoch": 0.18924111431316043,
      "grad_norm": 0.7006607860700146,
      "learning_rate": 6.304e-06,
      "loss": 0.0626,
      "step": 394
    },
    {
      "epoch": 0.18972142170989434,
      "grad_norm": 1.5246564977183437,
      "learning_rate": 6.3200000000000005e-06,
      "loss": 0.0722,
      "step": 395
    },
    {
      "epoch": 0.19020172910662825,
      "grad_norm": 1.9379386031947312,
      "learning_rate": 6.336000000000001e-06,
      "loss": 0.0807,
      "step": 396
    },
    {
      "epoch": 0.19068203650336216,
      "grad_norm": 1.1381971297694942,
      "learning_rate": 6.352e-06,
      "loss": 0.0882,
      "step": 397
    },
    {
      "epoch": 0.19116234390009607,
      "grad_norm": 1.051013291042444,
      "learning_rate": 6.368000000000001e-06,
      "loss": 0.0742,
      "step": 398
    },
    {
      "epoch": 0.19164265129682997,
      "grad_norm": 0.8171391283247772,
      "learning_rate": 6.384e-06,
      "loss": 0.055,
      "step": 399
    },
    {
      "epoch": 0.19212295869356388,
      "grad_norm": 0.9833424794189063,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.0681,
      "step": 400
    },
    {
      "epoch": 0.1926032660902978,
      "grad_norm": 1.6364622173487873,
      "learning_rate": 6.416e-06,
      "loss": 0.0785,
      "step": 401
    },
    {
      "epoch": 0.1930835734870317,
      "grad_norm": 1.0615403955397404,
      "learning_rate": 6.432e-06,
      "loss": 0.0991,
      "step": 402
    },
    {
      "epoch": 0.1935638808837656,
      "grad_norm": 0.7691979544920202,
      "learning_rate": 6.448000000000001e-06,
      "loss": 0.0748,
      "step": 403
    },
    {
      "epoch": 0.19404418828049952,
      "grad_norm": 1.1926569318031923,
      "learning_rate": 6.464e-06,
      "loss": 0.0644,
      "step": 404
    },
    {
      "epoch": 0.19452449567723343,
      "grad_norm": 0.5496991194443854,
      "learning_rate": 6.480000000000001e-06,
      "loss": 0.0531,
      "step": 405
    },
    {
      "epoch": 0.19500480307396734,
      "grad_norm": 0.8326919307353674,
      "learning_rate": 6.496e-06,
      "loss": 0.0725,
      "step": 406
    },
    {
      "epoch": 0.19548511047070125,
      "grad_norm": 1.0612929395385298,
      "learning_rate": 6.5120000000000005e-06,
      "loss": 0.0786,
      "step": 407
    },
    {
      "epoch": 0.19596541786743515,
      "grad_norm": 0.718075701725119,
      "learning_rate": 6.528000000000001e-06,
      "loss": 0.0585,
      "step": 408
    },
    {
      "epoch": 0.19644572526416906,
      "grad_norm": 1.2693919034433871,
      "learning_rate": 6.544e-06,
      "loss": 0.0737,
      "step": 409
    },
    {
      "epoch": 0.19692603266090297,
      "grad_norm": 0.9667230149007086,
      "learning_rate": 6.560000000000001e-06,
      "loss": 0.0939,
      "step": 410
    },
    {
      "epoch": 0.19740634005763688,
      "grad_norm": 0.7618960818418478,
      "learning_rate": 6.576e-06,
      "loss": 0.0451,
      "step": 411
    },
    {
      "epoch": 0.1978866474543708,
      "grad_norm": 0.9832215410415592,
      "learning_rate": 6.592000000000001e-06,
      "loss": 0.0671,
      "step": 412
    },
    {
      "epoch": 0.1983669548511047,
      "grad_norm": 0.757140075019698,
      "learning_rate": 6.608000000000001e-06,
      "loss": 0.072,
      "step": 413
    },
    {
      "epoch": 0.1988472622478386,
      "grad_norm": 0.8925384905186106,
      "learning_rate": 6.6240000000000004e-06,
      "loss": 0.0603,
      "step": 414
    },
    {
      "epoch": 0.19932756964457252,
      "grad_norm": 1.1216976394709526,
      "learning_rate": 6.640000000000001e-06,
      "loss": 0.073,
      "step": 415
    },
    {
      "epoch": 0.19980787704130643,
      "grad_norm": 0.635084003555822,
      "learning_rate": 6.656e-06,
      "loss": 0.05,
      "step": 416
    },
    {
      "epoch": 0.20028818443804033,
      "grad_norm": 0.7700738336691185,
      "learning_rate": 6.672000000000001e-06,
      "loss": 0.0723,
      "step": 417
    },
    {
      "epoch": 0.20076849183477424,
      "grad_norm": 1.2073339284961622,
      "learning_rate": 6.688e-06,
      "loss": 0.0789,
      "step": 418
    },
    {
      "epoch": 0.20124879923150815,
      "grad_norm": 1.0303414284152617,
      "learning_rate": 6.7040000000000005e-06,
      "loss": 0.0942,
      "step": 419
    },
    {
      "epoch": 0.2017291066282421,
      "grad_norm": 1.4496720069236764,
      "learning_rate": 6.720000000000001e-06,
      "loss": 0.0934,
      "step": 420
    },
    {
      "epoch": 0.202209414024976,
      "grad_norm": 1.39926810164135,
      "learning_rate": 6.736e-06,
      "loss": 0.0943,
      "step": 421
    },
    {
      "epoch": 0.2026897214217099,
      "grad_norm": 0.9605472208895933,
      "learning_rate": 6.752000000000001e-06,
      "loss": 0.07,
      "step": 422
    },
    {
      "epoch": 0.20317002881844382,
      "grad_norm": 1.1676905060892429,
      "learning_rate": 6.768e-06,
      "loss": 0.0957,
      "step": 423
    },
    {
      "epoch": 0.20365033621517772,
      "grad_norm": 1.0226460242714408,
      "learning_rate": 6.784000000000001e-06,
      "loss": 0.0924,
      "step": 424
    },
    {
      "epoch": 0.20413064361191163,
      "grad_norm": 0.7672962042725741,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.0731,
      "step": 425
    },
    {
      "epoch": 0.20461095100864554,
      "grad_norm": 1.1567284889374434,
      "learning_rate": 6.8160000000000005e-06,
      "loss": 0.0783,
      "step": 426
    },
    {
      "epoch": 0.20509125840537945,
      "grad_norm": 1.2828166133488417,
      "learning_rate": 6.832000000000001e-06,
      "loss": 0.085,
      "step": 427
    },
    {
      "epoch": 0.20557156580211336,
      "grad_norm": 0.9948993350398787,
      "learning_rate": 6.848e-06,
      "loss": 0.0712,
      "step": 428
    },
    {
      "epoch": 0.20605187319884727,
      "grad_norm": 1.0013503870526177,
      "learning_rate": 6.864000000000001e-06,
      "loss": 0.0725,
      "step": 429
    },
    {
      "epoch": 0.20653218059558118,
      "grad_norm": 0.828738359422558,
      "learning_rate": 6.88e-06,
      "loss": 0.0793,
      "step": 430
    },
    {
      "epoch": 0.2070124879923151,
      "grad_norm": 1.2375082026111115,
      "learning_rate": 6.8960000000000006e-06,
      "loss": 0.0868,
      "step": 431
    },
    {
      "epoch": 0.207492795389049,
      "grad_norm": 0.9272139309771383,
      "learning_rate": 6.912000000000001e-06,
      "loss": 0.0498,
      "step": 432
    },
    {
      "epoch": 0.2079731027857829,
      "grad_norm": 1.1885661004953512,
      "learning_rate": 6.928e-06,
      "loss": 0.0644,
      "step": 433
    },
    {
      "epoch": 0.2084534101825168,
      "grad_norm": 0.9603791498315314,
      "learning_rate": 6.944000000000001e-06,
      "loss": 0.0652,
      "step": 434
    },
    {
      "epoch": 0.20893371757925072,
      "grad_norm": 1.9606166639543452,
      "learning_rate": 6.96e-06,
      "loss": 0.0883,
      "step": 435
    },
    {
      "epoch": 0.20941402497598463,
      "grad_norm": 1.042036621430657,
      "learning_rate": 6.976000000000001e-06,
      "loss": 0.0773,
      "step": 436
    },
    {
      "epoch": 0.20989433237271854,
      "grad_norm": 0.807705104762043,
      "learning_rate": 6.992000000000001e-06,
      "loss": 0.0663,
      "step": 437
    },
    {
      "epoch": 0.21037463976945245,
      "grad_norm": 0.644829498274957,
      "learning_rate": 7.0080000000000005e-06,
      "loss": 0.0532,
      "step": 438
    },
    {
      "epoch": 0.21085494716618636,
      "grad_norm": 1.3697814725304436,
      "learning_rate": 7.024000000000001e-06,
      "loss": 0.0971,
      "step": 439
    },
    {
      "epoch": 0.21133525456292027,
      "grad_norm": 0.8499179196327713,
      "learning_rate": 7.04e-06,
      "loss": 0.06,
      "step": 440
    },
    {
      "epoch": 0.21181556195965417,
      "grad_norm": 1.1863374964467173,
      "learning_rate": 7.056000000000001e-06,
      "loss": 0.0835,
      "step": 441
    },
    {
      "epoch": 0.21229586935638808,
      "grad_norm": 1.025816763411157,
      "learning_rate": 7.072000000000001e-06,
      "loss": 0.0663,
      "step": 442
    },
    {
      "epoch": 0.212776176753122,
      "grad_norm": 0.7405994529015164,
      "learning_rate": 7.088000000000001e-06,
      "loss": 0.0598,
      "step": 443
    },
    {
      "epoch": 0.2132564841498559,
      "grad_norm": 0.6945613328918526,
      "learning_rate": 7.104000000000001e-06,
      "loss": 0.0455,
      "step": 444
    },
    {
      "epoch": 0.2137367915465898,
      "grad_norm": 1.2263272059064698,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 0.0892,
      "step": 445
    },
    {
      "epoch": 0.21421709894332372,
      "grad_norm": 0.8174840860440725,
      "learning_rate": 7.136000000000001e-06,
      "loss": 0.1028,
      "step": 446
    },
    {
      "epoch": 0.21469740634005763,
      "grad_norm": 0.8623153464251558,
      "learning_rate": 7.152e-06,
      "loss": 0.0851,
      "step": 447
    },
    {
      "epoch": 0.21517771373679154,
      "grad_norm": 0.8772215697627417,
      "learning_rate": 7.168000000000001e-06,
      "loss": 0.0631,
      "step": 448
    },
    {
      "epoch": 0.21565802113352545,
      "grad_norm": 1.1867765756821822,
      "learning_rate": 7.184000000000001e-06,
      "loss": 0.071,
      "step": 449
    },
    {
      "epoch": 0.21613832853025935,
      "grad_norm": 1.100610239534095,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.0819,
      "step": 450
    },
    {
      "epoch": 0.21661863592699326,
      "grad_norm": 1.0692001780240707,
      "learning_rate": 7.216000000000001e-06,
      "loss": 0.0752,
      "step": 451
    },
    {
      "epoch": 0.21709894332372717,
      "grad_norm": 1.8257869421442983,
      "learning_rate": 7.232e-06,
      "loss": 0.0961,
      "step": 452
    },
    {
      "epoch": 0.21757925072046108,
      "grad_norm": 0.8717548117040241,
      "learning_rate": 7.248000000000001e-06,
      "loss": 0.0705,
      "step": 453
    },
    {
      "epoch": 0.21805955811719502,
      "grad_norm": 0.679515897488406,
      "learning_rate": 7.264000000000001e-06,
      "loss": 0.0595,
      "step": 454
    },
    {
      "epoch": 0.21853986551392893,
      "grad_norm": 0.895857495998295,
      "learning_rate": 7.280000000000001e-06,
      "loss": 0.0897,
      "step": 455
    },
    {
      "epoch": 0.21902017291066284,
      "grad_norm": 0.8679817923298461,
      "learning_rate": 7.296000000000001e-06,
      "loss": 0.0767,
      "step": 456
    },
    {
      "epoch": 0.21950048030739674,
      "grad_norm": 0.9327609488192921,
      "learning_rate": 7.3120000000000005e-06,
      "loss": 0.0806,
      "step": 457
    },
    {
      "epoch": 0.21998078770413065,
      "grad_norm": 0.5822697098791232,
      "learning_rate": 7.328000000000001e-06,
      "loss": 0.0568,
      "step": 458
    },
    {
      "epoch": 0.22046109510086456,
      "grad_norm": 0.8136477286314635,
      "learning_rate": 7.344000000000001e-06,
      "loss": 0.0794,
      "step": 459
    },
    {
      "epoch": 0.22094140249759847,
      "grad_norm": 0.6332136039248887,
      "learning_rate": 7.360000000000001e-06,
      "loss": 0.0576,
      "step": 460
    },
    {
      "epoch": 0.22142170989433238,
      "grad_norm": 0.9815300986796249,
      "learning_rate": 7.376000000000001e-06,
      "loss": 0.0643,
      "step": 461
    },
    {
      "epoch": 0.2219020172910663,
      "grad_norm": 0.724954981826669,
      "learning_rate": 7.3920000000000005e-06,
      "loss": 0.0788,
      "step": 462
    },
    {
      "epoch": 0.2223823246878002,
      "grad_norm": 0.9480652345961487,
      "learning_rate": 7.408000000000001e-06,
      "loss": 0.0818,
      "step": 463
    },
    {
      "epoch": 0.2228626320845341,
      "grad_norm": 1.3054685818951983,
      "learning_rate": 7.424e-06,
      "loss": 0.0919,
      "step": 464
    },
    {
      "epoch": 0.22334293948126802,
      "grad_norm": 1.0576795770417882,
      "learning_rate": 7.440000000000001e-06,
      "loss": 0.0724,
      "step": 465
    },
    {
      "epoch": 0.22382324687800192,
      "grad_norm": 0.7169105130347482,
      "learning_rate": 7.456000000000001e-06,
      "loss": 0.0516,
      "step": 466
    },
    {
      "epoch": 0.22430355427473583,
      "grad_norm": 0.9174559479694644,
      "learning_rate": 7.472000000000001e-06,
      "loss": 0.0465,
      "step": 467
    },
    {
      "epoch": 0.22478386167146974,
      "grad_norm": 1.2190080584481207,
      "learning_rate": 7.488000000000001e-06,
      "loss": 0.0813,
      "step": 468
    },
    {
      "epoch": 0.22526416906820365,
      "grad_norm": 0.8257736438539969,
      "learning_rate": 7.5040000000000005e-06,
      "loss": 0.0665,
      "step": 469
    },
    {
      "epoch": 0.22574447646493756,
      "grad_norm": 1.005516446834955,
      "learning_rate": 7.520000000000001e-06,
      "loss": 0.0591,
      "step": 470
    },
    {
      "epoch": 0.22622478386167147,
      "grad_norm": 0.9928080490901438,
      "learning_rate": 7.536000000000001e-06,
      "loss": 0.0864,
      "step": 471
    },
    {
      "epoch": 0.22670509125840538,
      "grad_norm": 1.4816159445977666,
      "learning_rate": 7.552000000000001e-06,
      "loss": 0.0794,
      "step": 472
    },
    {
      "epoch": 0.2271853986551393,
      "grad_norm": 1.758118832520925,
      "learning_rate": 7.568000000000001e-06,
      "loss": 0.0671,
      "step": 473
    },
    {
      "epoch": 0.2276657060518732,
      "grad_norm": 1.2359986745074605,
      "learning_rate": 7.5840000000000006e-06,
      "loss": 0.0862,
      "step": 474
    },
    {
      "epoch": 0.2281460134486071,
      "grad_norm": 1.065041005575724,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.0867,
      "step": 475
    },
    {
      "epoch": 0.228626320845341,
      "grad_norm": 1.1386518653125688,
      "learning_rate": 7.616000000000001e-06,
      "loss": 0.0601,
      "step": 476
    },
    {
      "epoch": 0.22910662824207492,
      "grad_norm": 0.9844587250229394,
      "learning_rate": 7.632e-06,
      "loss": 0.0646,
      "step": 477
    },
    {
      "epoch": 0.22958693563880883,
      "grad_norm": 1.2058332109509622,
      "learning_rate": 7.648e-06,
      "loss": 0.0708,
      "step": 478
    },
    {
      "epoch": 0.23006724303554274,
      "grad_norm": 1.081013183286517,
      "learning_rate": 7.664e-06,
      "loss": 0.0701,
      "step": 479
    },
    {
      "epoch": 0.23054755043227665,
      "grad_norm": 0.9951838717376329,
      "learning_rate": 7.680000000000001e-06,
      "loss": 0.0769,
      "step": 480
    },
    {
      "epoch": 0.23102785782901056,
      "grad_norm": 0.6593691281860705,
      "learning_rate": 7.696e-06,
      "loss": 0.0624,
      "step": 481
    },
    {
      "epoch": 0.23150816522574447,
      "grad_norm": 1.1041552206575123,
      "learning_rate": 7.712e-06,
      "loss": 0.0786,
      "step": 482
    },
    {
      "epoch": 0.23198847262247838,
      "grad_norm": 1.1024525564632526,
      "learning_rate": 7.728000000000001e-06,
      "loss": 0.0825,
      "step": 483
    },
    {
      "epoch": 0.23246878001921228,
      "grad_norm": 0.7562432101013394,
      "learning_rate": 7.744e-06,
      "loss": 0.0626,
      "step": 484
    },
    {
      "epoch": 0.2329490874159462,
      "grad_norm": 0.7159892284435388,
      "learning_rate": 7.76e-06,
      "loss": 0.0666,
      "step": 485
    },
    {
      "epoch": 0.2334293948126801,
      "grad_norm": 0.7637031784203081,
      "learning_rate": 7.776e-06,
      "loss": 0.0535,
      "step": 486
    },
    {
      "epoch": 0.23390970220941404,
      "grad_norm": 1.341218149155002,
      "learning_rate": 7.792000000000001e-06,
      "loss": 0.0841,
      "step": 487
    },
    {
      "epoch": 0.23439000960614795,
      "grad_norm": 1.0381162643143098,
      "learning_rate": 7.808e-06,
      "loss": 0.0522,
      "step": 488
    },
    {
      "epoch": 0.23487031700288186,
      "grad_norm": 1.1102483912786174,
      "learning_rate": 7.824e-06,
      "loss": 0.0839,
      "step": 489
    },
    {
      "epoch": 0.23535062439961577,
      "grad_norm": 1.1434606793180935,
      "learning_rate": 7.840000000000001e-06,
      "loss": 0.0691,
      "step": 490
    },
    {
      "epoch": 0.23583093179634967,
      "grad_norm": 2.1962165048575533,
      "learning_rate": 7.856e-06,
      "loss": 0.0763,
      "step": 491
    },
    {
      "epoch": 0.23631123919308358,
      "grad_norm": 0.890647664698546,
      "learning_rate": 7.872e-06,
      "loss": 0.0718,
      "step": 492
    },
    {
      "epoch": 0.2367915465898175,
      "grad_norm": 0.8203501888126637,
      "learning_rate": 7.888e-06,
      "loss": 0.0726,
      "step": 493
    },
    {
      "epoch": 0.2372718539865514,
      "grad_norm": 1.2395973157128906,
      "learning_rate": 7.904000000000001e-06,
      "loss": 0.0898,
      "step": 494
    },
    {
      "epoch": 0.2377521613832853,
      "grad_norm": 0.9214838558224117,
      "learning_rate": 7.92e-06,
      "loss": 0.0657,
      "step": 495
    },
    {
      "epoch": 0.23823246878001922,
      "grad_norm": 1.0269834431751539,
      "learning_rate": 7.936e-06,
      "loss": 0.0807,
      "step": 496
    },
    {
      "epoch": 0.23871277617675313,
      "grad_norm": 1.020546331564402,
      "learning_rate": 7.952000000000001e-06,
      "loss": 0.0756,
      "step": 497
    },
    {
      "epoch": 0.23919308357348704,
      "grad_norm": 0.8476975296985142,
      "learning_rate": 7.968e-06,
      "loss": 0.0643,
      "step": 498
    },
    {
      "epoch": 0.23967339097022095,
      "grad_norm": 1.1388832811626284,
      "learning_rate": 7.984e-06,
      "loss": 0.0849,
      "step": 499
    },
    {
      "epoch": 0.24015369836695485,
      "grad_norm": 1.0929897122086174,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0807,
      "step": 500
    },
    {
      "epoch": 0.24063400576368876,
      "grad_norm": 1.1481077522909218,
      "learning_rate": 8.016e-06,
      "loss": 0.0809,
      "step": 501
    },
    {
      "epoch": 0.24111431316042267,
      "grad_norm": 0.65682707360465,
      "learning_rate": 8.032e-06,
      "loss": 0.0654,
      "step": 502
    },
    {
      "epoch": 0.24159462055715658,
      "grad_norm": 0.8221527194893445,
      "learning_rate": 8.048e-06,
      "loss": 0.0723,
      "step": 503
    },
    {
      "epoch": 0.2420749279538905,
      "grad_norm": 1.0897467045485394,
      "learning_rate": 8.064000000000001e-06,
      "loss": 0.0678,
      "step": 504
    },
    {
      "epoch": 0.2425552353506244,
      "grad_norm": 0.7475058575087895,
      "learning_rate": 8.08e-06,
      "loss": 0.0516,
      "step": 505
    },
    {
      "epoch": 0.2430355427473583,
      "grad_norm": 0.7036037103827971,
      "learning_rate": 8.096e-06,
      "loss": 0.0538,
      "step": 506
    },
    {
      "epoch": 0.24351585014409222,
      "grad_norm": 0.8406021606293904,
      "learning_rate": 8.112000000000001e-06,
      "loss": 0.0588,
      "step": 507
    },
    {
      "epoch": 0.24399615754082613,
      "grad_norm": 0.9956392385750449,
      "learning_rate": 8.128e-06,
      "loss": 0.0529,
      "step": 508
    },
    {
      "epoch": 0.24447646493756003,
      "grad_norm": 1.1472735203009934,
      "learning_rate": 8.144e-06,
      "loss": 0.0935,
      "step": 509
    },
    {
      "epoch": 0.24495677233429394,
      "grad_norm": 1.024997413494147,
      "learning_rate": 8.16e-06,
      "loss": 0.0776,
      "step": 510
    },
    {
      "epoch": 0.24543707973102785,
      "grad_norm": 0.6571287147481466,
      "learning_rate": 8.176000000000001e-06,
      "loss": 0.0628,
      "step": 511
    },
    {
      "epoch": 0.24591738712776176,
      "grad_norm": 1.3063911540302608,
      "learning_rate": 8.192e-06,
      "loss": 0.0717,
      "step": 512
    },
    {
      "epoch": 0.24639769452449567,
      "grad_norm": 1.4230401674267745,
      "learning_rate": 8.208e-06,
      "loss": 0.0764,
      "step": 513
    },
    {
      "epoch": 0.24687800192122958,
      "grad_norm": 0.8048952076569829,
      "learning_rate": 8.224000000000001e-06,
      "loss": 0.0571,
      "step": 514
    },
    {
      "epoch": 0.2473583093179635,
      "grad_norm": 0.6650682219974673,
      "learning_rate": 8.24e-06,
      "loss": 0.0565,
      "step": 515
    },
    {
      "epoch": 0.2478386167146974,
      "grad_norm": 1.2997870431175331,
      "learning_rate": 8.256e-06,
      "loss": 0.0837,
      "step": 516
    },
    {
      "epoch": 0.2483189241114313,
      "grad_norm": 4.337586159697626,
      "learning_rate": 8.272000000000001e-06,
      "loss": 0.0467,
      "step": 517
    },
    {
      "epoch": 0.24879923150816521,
      "grad_norm": 0.8239655704265273,
      "learning_rate": 8.288000000000001e-06,
      "loss": 0.0647,
      "step": 518
    },
    {
      "epoch": 0.24927953890489912,
      "grad_norm": 0.8494250080431497,
      "learning_rate": 8.304e-06,
      "loss": 0.0526,
      "step": 519
    },
    {
      "epoch": 0.24975984630163303,
      "grad_norm": 0.7032539553571172,
      "learning_rate": 8.32e-06,
      "loss": 0.0504,
      "step": 520
    },
    {
      "epoch": 0.25024015369836694,
      "grad_norm": 0.7697835483160133,
      "learning_rate": 8.336000000000001e-06,
      "loss": 0.0508,
      "step": 521
    },
    {
      "epoch": 0.2507204610951009,
      "grad_norm": 0.8341875541080136,
      "learning_rate": 8.352e-06,
      "loss": 0.0807,
      "step": 522
    },
    {
      "epoch": 0.25120076849183476,
      "grad_norm": 0.8726308431077691,
      "learning_rate": 8.368e-06,
      "loss": 0.0636,
      "step": 523
    },
    {
      "epoch": 0.2516810758885687,
      "grad_norm": 0.9286195806386074,
      "learning_rate": 8.384000000000001e-06,
      "loss": 0.1007,
      "step": 524
    },
    {
      "epoch": 0.2521613832853026,
      "grad_norm": 1.0477568055323367,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.0653,
      "step": 525
    },
    {
      "epoch": 0.2526416906820365,
      "grad_norm": 0.907504117294368,
      "learning_rate": 8.416e-06,
      "loss": 0.0687,
      "step": 526
    },
    {
      "epoch": 0.2531219980787704,
      "grad_norm": 0.6672546812595275,
      "learning_rate": 8.432e-06,
      "loss": 0.0833,
      "step": 527
    },
    {
      "epoch": 0.25360230547550433,
      "grad_norm": 0.8481460299512423,
      "learning_rate": 8.448000000000001e-06,
      "loss": 0.069,
      "step": 528
    },
    {
      "epoch": 0.2540826128722382,
      "grad_norm": 0.7153223957448147,
      "learning_rate": 8.464e-06,
      "loss": 0.0685,
      "step": 529
    },
    {
      "epoch": 0.25456292026897215,
      "grad_norm": 0.6977196547557283,
      "learning_rate": 8.48e-06,
      "loss": 0.0642,
      "step": 530
    },
    {
      "epoch": 0.25504322766570603,
      "grad_norm": 0.8261907792245994,
      "learning_rate": 8.496000000000001e-06,
      "loss": 0.0675,
      "step": 531
    },
    {
      "epoch": 0.25552353506243997,
      "grad_norm": 0.7198555590025666,
      "learning_rate": 8.512e-06,
      "loss": 0.0851,
      "step": 532
    },
    {
      "epoch": 0.25600384245917385,
      "grad_norm": 0.6637599827890456,
      "learning_rate": 8.528e-06,
      "loss": 0.0899,
      "step": 533
    },
    {
      "epoch": 0.2564841498559078,
      "grad_norm": 0.7512929164073144,
      "learning_rate": 8.544000000000002e-06,
      "loss": 0.0865,
      "step": 534
    },
    {
      "epoch": 0.25696445725264166,
      "grad_norm": 1.8750413633075549,
      "learning_rate": 8.560000000000001e-06,
      "loss": 0.0611,
      "step": 535
    },
    {
      "epoch": 0.2574447646493756,
      "grad_norm": 0.8162254344386675,
      "learning_rate": 8.576e-06,
      "loss": 0.082,
      "step": 536
    },
    {
      "epoch": 0.2579250720461095,
      "grad_norm": 0.7407860981655993,
      "learning_rate": 8.592e-06,
      "loss": 0.0659,
      "step": 537
    },
    {
      "epoch": 0.2584053794428434,
      "grad_norm": 0.8614219120120522,
      "learning_rate": 8.608000000000001e-06,
      "loss": 0.1007,
      "step": 538
    },
    {
      "epoch": 0.25888568683957736,
      "grad_norm": 1.2916018437852648,
      "learning_rate": 8.624e-06,
      "loss": 0.0711,
      "step": 539
    },
    {
      "epoch": 0.25936599423631124,
      "grad_norm": 1.0968982870807809,
      "learning_rate": 8.64e-06,
      "loss": 0.0494,
      "step": 540
    },
    {
      "epoch": 0.2598463016330452,
      "grad_norm": 0.8333925089939908,
      "learning_rate": 8.656000000000001e-06,
      "loss": 0.0718,
      "step": 541
    },
    {
      "epoch": 0.26032660902977905,
      "grad_norm": 1.5580346095928794,
      "learning_rate": 8.672000000000001e-06,
      "loss": 0.0728,
      "step": 542
    },
    {
      "epoch": 0.260806916426513,
      "grad_norm": 1.3177727190392534,
      "learning_rate": 8.688e-06,
      "loss": 0.068,
      "step": 543
    },
    {
      "epoch": 0.2612872238232469,
      "grad_norm": 1.1607948509472454,
      "learning_rate": 8.704e-06,
      "loss": 0.0691,
      "step": 544
    },
    {
      "epoch": 0.2617675312199808,
      "grad_norm": 0.9590351041286705,
      "learning_rate": 8.720000000000001e-06,
      "loss": 0.0613,
      "step": 545
    },
    {
      "epoch": 0.2622478386167147,
      "grad_norm": 0.830353272020343,
      "learning_rate": 8.736e-06,
      "loss": 0.0661,
      "step": 546
    },
    {
      "epoch": 0.2627281460134486,
      "grad_norm": 0.8307084029819353,
      "learning_rate": 8.752e-06,
      "loss": 0.0557,
      "step": 547
    },
    {
      "epoch": 0.2632084534101825,
      "grad_norm": 0.8137293430136122,
      "learning_rate": 8.768000000000001e-06,
      "loss": 0.0625,
      "step": 548
    },
    {
      "epoch": 0.26368876080691644,
      "grad_norm": 1.6678230574606077,
      "learning_rate": 8.784000000000001e-06,
      "loss": 0.0663,
      "step": 549
    },
    {
      "epoch": 0.2641690682036503,
      "grad_norm": 0.8131640866998433,
      "learning_rate": 8.8e-06,
      "loss": 0.0758,
      "step": 550
    },
    {
      "epoch": 0.26464937560038426,
      "grad_norm": 0.746830009216251,
      "learning_rate": 8.816000000000002e-06,
      "loss": 0.0594,
      "step": 551
    },
    {
      "epoch": 0.26512968299711814,
      "grad_norm": 1.0612071437391581,
      "learning_rate": 8.832000000000001e-06,
      "loss": 0.1148,
      "step": 552
    },
    {
      "epoch": 0.2656099903938521,
      "grad_norm": 1.0736052555532878,
      "learning_rate": 8.848e-06,
      "loss": 0.0795,
      "step": 553
    },
    {
      "epoch": 0.26609029779058596,
      "grad_norm": 1.7811521568482815,
      "learning_rate": 8.864e-06,
      "loss": 0.0708,
      "step": 554
    },
    {
      "epoch": 0.2665706051873199,
      "grad_norm": 1.1601873615680562,
      "learning_rate": 8.880000000000001e-06,
      "loss": 0.0632,
      "step": 555
    },
    {
      "epoch": 0.2670509125840538,
      "grad_norm": 0.9443323090818715,
      "learning_rate": 8.896000000000001e-06,
      "loss": 0.0726,
      "step": 556
    },
    {
      "epoch": 0.2675312199807877,
      "grad_norm": 0.751021625854935,
      "learning_rate": 8.912e-06,
      "loss": 0.0496,
      "step": 557
    },
    {
      "epoch": 0.2680115273775216,
      "grad_norm": 0.93106805297644,
      "learning_rate": 8.928000000000002e-06,
      "loss": 0.0777,
      "step": 558
    },
    {
      "epoch": 0.26849183477425553,
      "grad_norm": 0.635298715287655,
      "learning_rate": 8.944000000000001e-06,
      "loss": 0.059,
      "step": 559
    },
    {
      "epoch": 0.2689721421709894,
      "grad_norm": 0.8466300533119754,
      "learning_rate": 8.96e-06,
      "loss": 0.0788,
      "step": 560
    },
    {
      "epoch": 0.26945244956772335,
      "grad_norm": 0.7748167130306874,
      "learning_rate": 8.976e-06,
      "loss": 0.0599,
      "step": 561
    },
    {
      "epoch": 0.26993275696445723,
      "grad_norm": 0.7427736105339172,
      "learning_rate": 8.992000000000001e-06,
      "loss": 0.0614,
      "step": 562
    },
    {
      "epoch": 0.27041306436119117,
      "grad_norm": 0.8320001022304129,
      "learning_rate": 9.008e-06,
      "loss": 0.0674,
      "step": 563
    },
    {
      "epoch": 0.27089337175792505,
      "grad_norm": 0.7704318557146501,
      "learning_rate": 9.024e-06,
      "loss": 0.0702,
      "step": 564
    },
    {
      "epoch": 0.271373679154659,
      "grad_norm": 0.7407765317427476,
      "learning_rate": 9.040000000000002e-06,
      "loss": 0.0744,
      "step": 565
    },
    {
      "epoch": 0.27185398655139287,
      "grad_norm": 0.7206956721264949,
      "learning_rate": 9.056000000000001e-06,
      "loss": 0.0543,
      "step": 566
    },
    {
      "epoch": 0.2723342939481268,
      "grad_norm": 0.9451040246693897,
      "learning_rate": 9.072e-06,
      "loss": 0.0807,
      "step": 567
    },
    {
      "epoch": 0.2728146013448607,
      "grad_norm": 0.9009649199487116,
      "learning_rate": 9.088000000000002e-06,
      "loss": 0.062,
      "step": 568
    },
    {
      "epoch": 0.2732949087415946,
      "grad_norm": 1.0113937709014764,
      "learning_rate": 9.104000000000001e-06,
      "loss": 0.0648,
      "step": 569
    },
    {
      "epoch": 0.2737752161383285,
      "grad_norm": 1.0501699409202674,
      "learning_rate": 9.12e-06,
      "loss": 0.0537,
      "step": 570
    },
    {
      "epoch": 0.27425552353506244,
      "grad_norm": 0.5693860840353209,
      "learning_rate": 9.136e-06,
      "loss": 0.0448,
      "step": 571
    },
    {
      "epoch": 0.2747358309317964,
      "grad_norm": 1.0055693738637874,
      "learning_rate": 9.152000000000001e-06,
      "loss": 0.0747,
      "step": 572
    },
    {
      "epoch": 0.27521613832853026,
      "grad_norm": 1.2257150323649562,
      "learning_rate": 9.168000000000001e-06,
      "loss": 0.0971,
      "step": 573
    },
    {
      "epoch": 0.2756964457252642,
      "grad_norm": 0.8489461117348059,
      "learning_rate": 9.184e-06,
      "loss": 0.067,
      "step": 574
    },
    {
      "epoch": 0.2761767531219981,
      "grad_norm": 0.8112579026025699,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.0682,
      "step": 575
    },
    {
      "epoch": 0.276657060518732,
      "grad_norm": 0.8857859004688043,
      "learning_rate": 9.216000000000001e-06,
      "loss": 0.0799,
      "step": 576
    },
    {
      "epoch": 0.2771373679154659,
      "grad_norm": 0.7523380576301403,
      "learning_rate": 9.232e-06,
      "loss": 0.0682,
      "step": 577
    },
    {
      "epoch": 0.27761767531219983,
      "grad_norm": 0.8161110604162131,
      "learning_rate": 9.248e-06,
      "loss": 0.0713,
      "step": 578
    },
    {
      "epoch": 0.2780979827089337,
      "grad_norm": 1.2335737488024285,
      "learning_rate": 9.264000000000001e-06,
      "loss": 0.0976,
      "step": 579
    },
    {
      "epoch": 0.27857829010566765,
      "grad_norm": 0.8454243600862699,
      "learning_rate": 9.280000000000001e-06,
      "loss": 0.0606,
      "step": 580
    },
    {
      "epoch": 0.27905859750240153,
      "grad_norm": 1.2912157131289221,
      "learning_rate": 9.296e-06,
      "loss": 0.0864,
      "step": 581
    },
    {
      "epoch": 0.27953890489913547,
      "grad_norm": 1.0229081590491909,
      "learning_rate": 9.312000000000002e-06,
      "loss": 0.0914,
      "step": 582
    },
    {
      "epoch": 0.28001921229586935,
      "grad_norm": 1.206253943899943,
      "learning_rate": 9.328000000000001e-06,
      "loss": 0.0862,
      "step": 583
    },
    {
      "epoch": 0.2804995196926033,
      "grad_norm": 0.9742697094872371,
      "learning_rate": 9.344e-06,
      "loss": 0.0534,
      "step": 584
    },
    {
      "epoch": 0.28097982708933716,
      "grad_norm": 1.3994880281576707,
      "learning_rate": 9.360000000000002e-06,
      "loss": 0.0942,
      "step": 585
    },
    {
      "epoch": 0.2814601344860711,
      "grad_norm": 0.7863832094391792,
      "learning_rate": 9.376000000000001e-06,
      "loss": 0.0651,
      "step": 586
    },
    {
      "epoch": 0.281940441882805,
      "grad_norm": 1.790391699352207,
      "learning_rate": 9.392000000000001e-06,
      "loss": 0.0837,
      "step": 587
    },
    {
      "epoch": 0.2824207492795389,
      "grad_norm": 0.7167809240107046,
      "learning_rate": 9.408e-06,
      "loss": 0.0641,
      "step": 588
    },
    {
      "epoch": 0.2829010566762728,
      "grad_norm": 1.3108752667323915,
      "learning_rate": 9.424000000000002e-06,
      "loss": 0.0711,
      "step": 589
    },
    {
      "epoch": 0.28338136407300674,
      "grad_norm": 0.752914709706173,
      "learning_rate": 9.440000000000001e-06,
      "loss": 0.0584,
      "step": 590
    },
    {
      "epoch": 0.2838616714697406,
      "grad_norm": 0.8764860102695597,
      "learning_rate": 9.456e-06,
      "loss": 0.0477,
      "step": 591
    },
    {
      "epoch": 0.28434197886647455,
      "grad_norm": 0.6356131868385272,
      "learning_rate": 9.472000000000002e-06,
      "loss": 0.0522,
      "step": 592
    },
    {
      "epoch": 0.28482228626320844,
      "grad_norm": 1.263517402410837,
      "learning_rate": 9.488000000000001e-06,
      "loss": 0.0873,
      "step": 593
    },
    {
      "epoch": 0.28530259365994237,
      "grad_norm": 0.6481285314183274,
      "learning_rate": 9.504e-06,
      "loss": 0.0594,
      "step": 594
    },
    {
      "epoch": 0.28578290105667625,
      "grad_norm": 0.6520090713036953,
      "learning_rate": 9.52e-06,
      "loss": 0.0574,
      "step": 595
    },
    {
      "epoch": 0.2862632084534102,
      "grad_norm": 1.2576547381797791,
      "learning_rate": 9.536000000000002e-06,
      "loss": 0.0782,
      "step": 596
    },
    {
      "epoch": 0.28674351585014407,
      "grad_norm": 1.0281958680840946,
      "learning_rate": 9.552000000000001e-06,
      "loss": 0.0884,
      "step": 597
    },
    {
      "epoch": 0.287223823246878,
      "grad_norm": 2.0780498256230904,
      "learning_rate": 9.568e-06,
      "loss": 0.0686,
      "step": 598
    },
    {
      "epoch": 0.2877041306436119,
      "grad_norm": 0.7251320029199013,
      "learning_rate": 9.584000000000002e-06,
      "loss": 0.0667,
      "step": 599
    },
    {
      "epoch": 0.2881844380403458,
      "grad_norm": 0.9843387438340672,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.0696,
      "step": 600
    },
    {
      "epoch": 0.2886647454370797,
      "grad_norm": 1.5091693125190424,
      "learning_rate": 9.616e-06,
      "loss": 0.11,
      "step": 601
    },
    {
      "epoch": 0.28914505283381364,
      "grad_norm": 0.7606667303666816,
      "learning_rate": 9.632e-06,
      "loss": 0.0645,
      "step": 602
    },
    {
      "epoch": 0.2896253602305475,
      "grad_norm": 0.8702218966486639,
      "learning_rate": 9.648000000000001e-06,
      "loss": 0.0695,
      "step": 603
    },
    {
      "epoch": 0.29010566762728146,
      "grad_norm": 0.9692656471462159,
      "learning_rate": 9.664000000000001e-06,
      "loss": 0.071,
      "step": 604
    },
    {
      "epoch": 0.2905859750240154,
      "grad_norm": 1.152104986956867,
      "learning_rate": 9.68e-06,
      "loss": 0.0621,
      "step": 605
    },
    {
      "epoch": 0.2910662824207493,
      "grad_norm": 0.7344517356138976,
      "learning_rate": 9.696000000000002e-06,
      "loss": 0.0485,
      "step": 606
    },
    {
      "epoch": 0.2915465898174832,
      "grad_norm": 0.9928918365561824,
      "learning_rate": 9.712e-06,
      "loss": 0.0719,
      "step": 607
    },
    {
      "epoch": 0.2920268972142171,
      "grad_norm": 0.7667129001690043,
      "learning_rate": 9.728e-06,
      "loss": 0.0679,
      "step": 608
    },
    {
      "epoch": 0.29250720461095103,
      "grad_norm": 0.9950000199696363,
      "learning_rate": 9.744000000000002e-06,
      "loss": 0.0723,
      "step": 609
    },
    {
      "epoch": 0.2929875120076849,
      "grad_norm": 1.4085105451496738,
      "learning_rate": 9.760000000000001e-06,
      "loss": 0.0798,
      "step": 610
    },
    {
      "epoch": 0.29346781940441885,
      "grad_norm": 0.849429025717193,
      "learning_rate": 9.776000000000001e-06,
      "loss": 0.0573,
      "step": 611
    },
    {
      "epoch": 0.29394812680115273,
      "grad_norm": 0.8950057496274839,
      "learning_rate": 9.792e-06,
      "loss": 0.0718,
      "step": 612
    },
    {
      "epoch": 0.29442843419788667,
      "grad_norm": 0.728848993249414,
      "learning_rate": 9.808000000000002e-06,
      "loss": 0.0397,
      "step": 613
    },
    {
      "epoch": 0.29490874159462055,
      "grad_norm": 0.9156226984501676,
      "learning_rate": 9.824000000000001e-06,
      "loss": 0.0722,
      "step": 614
    },
    {
      "epoch": 0.2953890489913545,
      "grad_norm": 1.3199838058869244,
      "learning_rate": 9.84e-06,
      "loss": 0.0663,
      "step": 615
    },
    {
      "epoch": 0.29586935638808837,
      "grad_norm": 0.6241326568020441,
      "learning_rate": 9.856000000000002e-06,
      "loss": 0.0477,
      "step": 616
    },
    {
      "epoch": 0.2963496637848223,
      "grad_norm": 0.8937661400674575,
      "learning_rate": 9.872e-06,
      "loss": 0.0655,
      "step": 617
    },
    {
      "epoch": 0.2968299711815562,
      "grad_norm": 0.7018639448793483,
      "learning_rate": 9.888000000000001e-06,
      "loss": 0.0666,
      "step": 618
    },
    {
      "epoch": 0.2973102785782901,
      "grad_norm": 0.8792158375314773,
      "learning_rate": 9.904e-06,
      "loss": 0.0806,
      "step": 619
    },
    {
      "epoch": 0.297790585975024,
      "grad_norm": 0.8101543945686708,
      "learning_rate": 9.920000000000002e-06,
      "loss": 0.0628,
      "step": 620
    },
    {
      "epoch": 0.29827089337175794,
      "grad_norm": 0.6958139936512852,
      "learning_rate": 9.936000000000001e-06,
      "loss": 0.0586,
      "step": 621
    },
    {
      "epoch": 0.2987512007684918,
      "grad_norm": 0.8118776422093879,
      "learning_rate": 9.952e-06,
      "loss": 0.0696,
      "step": 622
    },
    {
      "epoch": 0.29923150816522576,
      "grad_norm": 0.801044484279383,
      "learning_rate": 9.968000000000002e-06,
      "loss": 0.0587,
      "step": 623
    },
    {
      "epoch": 0.29971181556195964,
      "grad_norm": 0.732515461824863,
      "learning_rate": 9.984e-06,
      "loss": 0.0564,
      "step": 624
    },
    {
      "epoch": 0.3001921229586936,
      "grad_norm": 0.8322604839391531,
      "learning_rate": 1e-05,
      "loss": 0.0812,
      "step": 625
    },
    {
      "epoch": 0.30067243035542746,
      "grad_norm": 0.7657461534156353,
      "learning_rate": 9.999999219069164e-06,
      "loss": 0.0553,
      "step": 626
    },
    {
      "epoch": 0.3011527377521614,
      "grad_norm": 0.96087209630369,
      "learning_rate": 9.999996876276899e-06,
      "loss": 0.0722,
      "step": 627
    },
    {
      "epoch": 0.3016330451488953,
      "grad_norm": 0.8664306047674389,
      "learning_rate": 9.999992971623935e-06,
      "loss": 0.0699,
      "step": 628
    },
    {
      "epoch": 0.3021133525456292,
      "grad_norm": 0.7586378661950982,
      "learning_rate": 9.999987505111493e-06,
      "loss": 0.0547,
      "step": 629
    },
    {
      "epoch": 0.3025936599423631,
      "grad_norm": 0.8770352678016571,
      "learning_rate": 9.999980476741282e-06,
      "loss": 0.0544,
      "step": 630
    },
    {
      "epoch": 0.30307396733909703,
      "grad_norm": 2.875494497735063,
      "learning_rate": 9.999971886515496e-06,
      "loss": 0.0706,
      "step": 631
    },
    {
      "epoch": 0.3035542747358309,
      "grad_norm": 1.2573657771022355,
      "learning_rate": 9.999961734436818e-06,
      "loss": 0.0687,
      "step": 632
    },
    {
      "epoch": 0.30403458213256485,
      "grad_norm": 0.683322107768531,
      "learning_rate": 9.99995002050842e-06,
      "loss": 0.0598,
      "step": 633
    },
    {
      "epoch": 0.3045148895292987,
      "grad_norm": 0.8277715966275165,
      "learning_rate": 9.99993674473396e-06,
      "loss": 0.0723,
      "step": 634
    },
    {
      "epoch": 0.30499519692603266,
      "grad_norm": 2.8948604349281584,
      "learning_rate": 9.999921907117588e-06,
      "loss": 0.1244,
      "step": 635
    },
    {
      "epoch": 0.30547550432276654,
      "grad_norm": 0.8638445515605259,
      "learning_rate": 9.999905507663936e-06,
      "loss": 0.0752,
      "step": 636
    },
    {
      "epoch": 0.3059558117195005,
      "grad_norm": 0.6727302040122436,
      "learning_rate": 9.999887546378127e-06,
      "loss": 0.0621,
      "step": 637
    },
    {
      "epoch": 0.30643611911623436,
      "grad_norm": 1.0952285566093234,
      "learning_rate": 9.999868023265772e-06,
      "loss": 0.095,
      "step": 638
    },
    {
      "epoch": 0.3069164265129683,
      "grad_norm": 0.957687462950951,
      "learning_rate": 9.99984693833297e-06,
      "loss": 0.0802,
      "step": 639
    },
    {
      "epoch": 0.30739673390970224,
      "grad_norm": 1.0874976971451398,
      "learning_rate": 9.999824291586306e-06,
      "loss": 0.0964,
      "step": 640
    },
    {
      "epoch": 0.3078770413064361,
      "grad_norm": 0.826343746073476,
      "learning_rate": 9.999800083032856e-06,
      "loss": 0.0553,
      "step": 641
    },
    {
      "epoch": 0.30835734870317005,
      "grad_norm": 0.8615917787952223,
      "learning_rate": 9.999774312680182e-06,
      "loss": 0.0674,
      "step": 642
    },
    {
      "epoch": 0.30883765609990393,
      "grad_norm": 0.9536454074667529,
      "learning_rate": 9.999746980536332e-06,
      "loss": 0.0754,
      "step": 643
    },
    {
      "epoch": 0.30931796349663787,
      "grad_norm": 0.8678398758854321,
      "learning_rate": 9.999718086609845e-06,
      "loss": 0.0536,
      "step": 644
    },
    {
      "epoch": 0.30979827089337175,
      "grad_norm": 0.8195920103808025,
      "learning_rate": 9.999687630909748e-06,
      "loss": 0.0676,
      "step": 645
    },
    {
      "epoch": 0.3102785782901057,
      "grad_norm": 0.833916110606186,
      "learning_rate": 9.999655613445552e-06,
      "loss": 0.0662,
      "step": 646
    },
    {
      "epoch": 0.31075888568683957,
      "grad_norm": 0.8552019334840019,
      "learning_rate": 9.99962203422726e-06,
      "loss": 0.0601,
      "step": 647
    },
    {
      "epoch": 0.3112391930835735,
      "grad_norm": 1.1244393216373334,
      "learning_rate": 9.99958689326536e-06,
      "loss": 0.0835,
      "step": 648
    },
    {
      "epoch": 0.3117195004803074,
      "grad_norm": 1.0921918261230938,
      "learning_rate": 9.99955019057083e-06,
      "loss": 0.0735,
      "step": 649
    },
    {
      "epoch": 0.3121998078770413,
      "grad_norm": 0.9476055236594563,
      "learning_rate": 9.999511926155135e-06,
      "loss": 0.0568,
      "step": 650
    },
    {
      "epoch": 0.3126801152737752,
      "grad_norm": 0.955427263476322,
      "learning_rate": 9.999472100030227e-06,
      "loss": 0.0682,
      "step": 651
    },
    {
      "epoch": 0.31316042267050914,
      "grad_norm": 0.6227107477762361,
      "learning_rate": 9.999430712208548e-06,
      "loss": 0.0515,
      "step": 652
    },
    {
      "epoch": 0.313640730067243,
      "grad_norm": 0.8819662452007495,
      "learning_rate": 9.999387762703025e-06,
      "loss": 0.0708,
      "step": 653
    },
    {
      "epoch": 0.31412103746397696,
      "grad_norm": 0.7140416262289825,
      "learning_rate": 9.999343251527076e-06,
      "loss": 0.0546,
      "step": 654
    },
    {
      "epoch": 0.31460134486071084,
      "grad_norm": 0.7721756355018631,
      "learning_rate": 9.999297178694603e-06,
      "loss": 0.0819,
      "step": 655
    },
    {
      "epoch": 0.3150816522574448,
      "grad_norm": 0.6792124284969571,
      "learning_rate": 9.999249544219998e-06,
      "loss": 0.0626,
      "step": 656
    },
    {
      "epoch": 0.31556195965417866,
      "grad_norm": 0.8965139505972938,
      "learning_rate": 9.999200348118142e-06,
      "loss": 0.0793,
      "step": 657
    },
    {
      "epoch": 0.3160422670509126,
      "grad_norm": 0.6875270287873301,
      "learning_rate": 9.9991495904044e-06,
      "loss": 0.0841,
      "step": 658
    },
    {
      "epoch": 0.3165225744476465,
      "grad_norm": 0.9499599493614728,
      "learning_rate": 9.999097271094632e-06,
      "loss": 0.0746,
      "step": 659
    },
    {
      "epoch": 0.3170028818443804,
      "grad_norm": 0.9142388374355066,
      "learning_rate": 9.999043390205176e-06,
      "loss": 0.0789,
      "step": 660
    },
    {
      "epoch": 0.3174831892411143,
      "grad_norm": 2.344335044128447,
      "learning_rate": 9.998987947752866e-06,
      "loss": 0.0992,
      "step": 661
    },
    {
      "epoch": 0.31796349663784823,
      "grad_norm": 1.318190298500077,
      "learning_rate": 9.99893094375502e-06,
      "loss": 0.0673,
      "step": 662
    },
    {
      "epoch": 0.3184438040345821,
      "grad_norm": 0.7219015180259919,
      "learning_rate": 9.998872378229444e-06,
      "loss": 0.0557,
      "step": 663
    },
    {
      "epoch": 0.31892411143131605,
      "grad_norm": 0.9108960437724677,
      "learning_rate": 9.998812251194432e-06,
      "loss": 0.0701,
      "step": 664
    },
    {
      "epoch": 0.31940441882804993,
      "grad_norm": 0.8017801457442163,
      "learning_rate": 9.998750562668767e-06,
      "loss": 0.0607,
      "step": 665
    },
    {
      "epoch": 0.31988472622478387,
      "grad_norm": 0.9355161886203067,
      "learning_rate": 9.99868731267172e-06,
      "loss": 0.0838,
      "step": 666
    },
    {
      "epoch": 0.32036503362151775,
      "grad_norm": 1.1545057767987545,
      "learning_rate": 9.998622501223045e-06,
      "loss": 0.0935,
      "step": 667
    },
    {
      "epoch": 0.3208453410182517,
      "grad_norm": 0.6776024404099317,
      "learning_rate": 9.998556128342989e-06,
      "loss": 0.0527,
      "step": 668
    },
    {
      "epoch": 0.32132564841498557,
      "grad_norm": 0.9245439985854373,
      "learning_rate": 9.998488194052287e-06,
      "loss": 0.0815,
      "step": 669
    },
    {
      "epoch": 0.3218059558117195,
      "grad_norm": 0.9108820872007813,
      "learning_rate": 9.998418698372156e-06,
      "loss": 0.0806,
      "step": 670
    },
    {
      "epoch": 0.3222862632084534,
      "grad_norm": 0.6413003064759893,
      "learning_rate": 9.998347641324309e-06,
      "loss": 0.0567,
      "step": 671
    },
    {
      "epoch": 0.3227665706051873,
      "grad_norm": 0.7124154469417349,
      "learning_rate": 9.998275022930937e-06,
      "loss": 0.0622,
      "step": 672
    },
    {
      "epoch": 0.32324687800192126,
      "grad_norm": 0.729795364251075,
      "learning_rate": 9.99820084321473e-06,
      "loss": 0.0549,
      "step": 673
    },
    {
      "epoch": 0.32372718539865514,
      "grad_norm": 0.7072881688429468,
      "learning_rate": 9.998125102198855e-06,
      "loss": 0.057,
      "step": 674
    },
    {
      "epoch": 0.3242074927953891,
      "grad_norm": 0.9404196385002046,
      "learning_rate": 9.998047799906972e-06,
      "loss": 0.0731,
      "step": 675
    },
    {
      "epoch": 0.32468780019212296,
      "grad_norm": 2.2283889506644803,
      "learning_rate": 9.99796893636323e-06,
      "loss": 0.0611,
      "step": 676
    },
    {
      "epoch": 0.3251681075888569,
      "grad_norm": 0.9898738372811073,
      "learning_rate": 9.997888511592262e-06,
      "loss": 0.075,
      "step": 677
    },
    {
      "epoch": 0.3256484149855908,
      "grad_norm": 0.5669213992706328,
      "learning_rate": 9.997806525619191e-06,
      "loss": 0.0462,
      "step": 678
    },
    {
      "epoch": 0.3261287223823247,
      "grad_norm": 0.7769207156534598,
      "learning_rate": 9.997722978469629e-06,
      "loss": 0.0598,
      "step": 679
    },
    {
      "epoch": 0.3266090297790586,
      "grad_norm": 1.14913919174941,
      "learning_rate": 9.997637870169673e-06,
      "loss": 0.0802,
      "step": 680
    },
    {
      "epoch": 0.3270893371757925,
      "grad_norm": 0.5579221498365345,
      "learning_rate": 9.997551200745905e-06,
      "loss": 0.0513,
      "step": 681
    },
    {
      "epoch": 0.3275696445725264,
      "grad_norm": 1.1622979766569141,
      "learning_rate": 9.997462970225402e-06,
      "loss": 0.0831,
      "step": 682
    },
    {
      "epoch": 0.32804995196926034,
      "grad_norm": 0.5843385711909048,
      "learning_rate": 9.997373178635723e-06,
      "loss": 0.0463,
      "step": 683
    },
    {
      "epoch": 0.3285302593659942,
      "grad_norm": 0.6003237420367146,
      "learning_rate": 9.997281826004919e-06,
      "loss": 0.0463,
      "step": 684
    },
    {
      "epoch": 0.32901056676272816,
      "grad_norm": 0.707379073392421,
      "learning_rate": 9.997188912361522e-06,
      "loss": 0.0743,
      "step": 685
    },
    {
      "epoch": 0.32949087415946204,
      "grad_norm": 0.6053436647816546,
      "learning_rate": 9.997094437734558e-06,
      "loss": 0.0586,
      "step": 686
    },
    {
      "epoch": 0.329971181556196,
      "grad_norm": 0.6165862274607972,
      "learning_rate": 9.99699840215354e-06,
      "loss": 0.0615,
      "step": 687
    },
    {
      "epoch": 0.33045148895292986,
      "grad_norm": 0.7162733421968076,
      "learning_rate": 9.996900805648462e-06,
      "loss": 0.0664,
      "step": 688
    },
    {
      "epoch": 0.3309317963496638,
      "grad_norm": 0.9782090778876608,
      "learning_rate": 9.996801648249815e-06,
      "loss": 0.0694,
      "step": 689
    },
    {
      "epoch": 0.3314121037463977,
      "grad_norm": 0.9345924479490764,
      "learning_rate": 9.996700929988571e-06,
      "loss": 0.0703,
      "step": 690
    },
    {
      "epoch": 0.3318924111431316,
      "grad_norm": 0.4641992659191584,
      "learning_rate": 9.996598650896191e-06,
      "loss": 0.0581,
      "step": 691
    },
    {
      "epoch": 0.3323727185398655,
      "grad_norm": 0.730369918229153,
      "learning_rate": 9.996494811004626e-06,
      "loss": 0.0671,
      "step": 692
    },
    {
      "epoch": 0.33285302593659943,
      "grad_norm": 0.7429226769702214,
      "learning_rate": 9.996389410346312e-06,
      "loss": 0.0603,
      "step": 693
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.8374173727738586,
      "learning_rate": 9.996282448954173e-06,
      "loss": 0.0686,
      "step": 694
    },
    {
      "epoch": 0.33381364073006725,
      "grad_norm": 0.7712716340055917,
      "learning_rate": 9.99617392686162e-06,
      "loss": 0.0606,
      "step": 695
    },
    {
      "epoch": 0.33429394812680113,
      "grad_norm": 0.5623336642825997,
      "learning_rate": 9.996063844102555e-06,
      "loss": 0.069,
      "step": 696
    },
    {
      "epoch": 0.33477425552353507,
      "grad_norm": 0.6995521178496045,
      "learning_rate": 9.995952200711361e-06,
      "loss": 0.0621,
      "step": 697
    },
    {
      "epoch": 0.33525456292026895,
      "grad_norm": 0.9184035660471785,
      "learning_rate": 9.995838996722916e-06,
      "loss": 0.0693,
      "step": 698
    },
    {
      "epoch": 0.3357348703170029,
      "grad_norm": 0.9083685686418375,
      "learning_rate": 9.995724232172578e-06,
      "loss": 0.0619,
      "step": 699
    },
    {
      "epoch": 0.33621517771373677,
      "grad_norm": 0.7795460332437362,
      "learning_rate": 9.995607907096198e-06,
      "loss": 0.0579,
      "step": 700
    },
    {
      "epoch": 0.3366954851104707,
      "grad_norm": 0.9597351909663773,
      "learning_rate": 9.995490021530116e-06,
      "loss": 0.0635,
      "step": 701
    },
    {
      "epoch": 0.3371757925072046,
      "grad_norm": 0.777710780224125,
      "learning_rate": 9.995370575511151e-06,
      "loss": 0.0644,
      "step": 702
    },
    {
      "epoch": 0.3376560999039385,
      "grad_norm": 0.8524311441817263,
      "learning_rate": 9.995249569076617e-06,
      "loss": 0.0649,
      "step": 703
    },
    {
      "epoch": 0.3381364073006724,
      "grad_norm": 0.9788665595689388,
      "learning_rate": 9.995127002264313e-06,
      "loss": 0.0483,
      "step": 704
    },
    {
      "epoch": 0.33861671469740634,
      "grad_norm": 0.9043257189131158,
      "learning_rate": 9.995002875112525e-06,
      "loss": 0.0802,
      "step": 705
    },
    {
      "epoch": 0.3390970220941403,
      "grad_norm": 1.0447948509065825,
      "learning_rate": 9.994877187660028e-06,
      "loss": 0.1116,
      "step": 706
    },
    {
      "epoch": 0.33957732949087416,
      "grad_norm": 0.6843801841988756,
      "learning_rate": 9.994749939946082e-06,
      "loss": 0.0445,
      "step": 707
    },
    {
      "epoch": 0.3400576368876081,
      "grad_norm": 0.8386099307338503,
      "learning_rate": 9.994621132010439e-06,
      "loss": 0.0761,
      "step": 708
    },
    {
      "epoch": 0.340537944284342,
      "grad_norm": 0.8431393706231558,
      "learning_rate": 9.994490763893328e-06,
      "loss": 0.0719,
      "step": 709
    },
    {
      "epoch": 0.3410182516810759,
      "grad_norm": 0.6771024567973309,
      "learning_rate": 9.994358835635477e-06,
      "loss": 0.0568,
      "step": 710
    },
    {
      "epoch": 0.3414985590778098,
      "grad_norm": 0.8961034176669026,
      "learning_rate": 9.9942253472781e-06,
      "loss": 0.0688,
      "step": 711
    },
    {
      "epoch": 0.34197886647454373,
      "grad_norm": 1.0453002901140704,
      "learning_rate": 9.99409029886289e-06,
      "loss": 0.1101,
      "step": 712
    },
    {
      "epoch": 0.3424591738712776,
      "grad_norm": 0.786975549589651,
      "learning_rate": 9.993953690432032e-06,
      "loss": 0.0554,
      "step": 713
    },
    {
      "epoch": 0.34293948126801155,
      "grad_norm": 0.7042757673435847,
      "learning_rate": 9.993815522028203e-06,
      "loss": 0.0617,
      "step": 714
    },
    {
      "epoch": 0.34341978866474543,
      "grad_norm": 0.9449988070860224,
      "learning_rate": 9.993675793694558e-06,
      "loss": 0.0615,
      "step": 715
    },
    {
      "epoch": 0.34390009606147937,
      "grad_norm": 1.2419899707910662,
      "learning_rate": 9.99353450547475e-06,
      "loss": 0.0724,
      "step": 716
    },
    {
      "epoch": 0.34438040345821325,
      "grad_norm": 1.1846845186388448,
      "learning_rate": 9.993391657412908e-06,
      "loss": 0.0892,
      "step": 717
    },
    {
      "epoch": 0.3448607108549472,
      "grad_norm": 0.9397604288615065,
      "learning_rate": 9.993247249553656e-06,
      "loss": 0.0561,
      "step": 718
    },
    {
      "epoch": 0.34534101825168106,
      "grad_norm": 0.6412293473440506,
      "learning_rate": 9.993101281942103e-06,
      "loss": 0.0506,
      "step": 719
    },
    {
      "epoch": 0.345821325648415,
      "grad_norm": 0.6410026176064426,
      "learning_rate": 9.992953754623847e-06,
      "loss": 0.0556,
      "step": 720
    },
    {
      "epoch": 0.3463016330451489,
      "grad_norm": 0.9400115622092615,
      "learning_rate": 9.992804667644969e-06,
      "loss": 0.0713,
      "step": 721
    },
    {
      "epoch": 0.3467819404418828,
      "grad_norm": 0.8307344024344647,
      "learning_rate": 9.99265402105204e-06,
      "loss": 0.0749,
      "step": 722
    },
    {
      "epoch": 0.3472622478386167,
      "grad_norm": 0.6264072104068256,
      "learning_rate": 9.992501814892118e-06,
      "loss": 0.0572,
      "step": 723
    },
    {
      "epoch": 0.34774255523535064,
      "grad_norm": 0.8873301693627772,
      "learning_rate": 9.99234804921275e-06,
      "loss": 0.0636,
      "step": 724
    },
    {
      "epoch": 0.3482228626320845,
      "grad_norm": 0.5551682372717668,
      "learning_rate": 9.992192724061965e-06,
      "loss": 0.0614,
      "step": 725
    },
    {
      "epoch": 0.34870317002881845,
      "grad_norm": 0.6741323301483185,
      "learning_rate": 9.992035839488283e-06,
      "loss": 0.0598,
      "step": 726
    },
    {
      "epoch": 0.34918347742555234,
      "grad_norm": 0.7478055633816404,
      "learning_rate": 9.991877395540714e-06,
      "loss": 0.0698,
      "step": 727
    },
    {
      "epoch": 0.34966378482228627,
      "grad_norm": 0.7204195910532974,
      "learning_rate": 9.991717392268747e-06,
      "loss": 0.0684,
      "step": 728
    },
    {
      "epoch": 0.35014409221902015,
      "grad_norm": 0.680949486584263,
      "learning_rate": 9.991555829722363e-06,
      "loss": 0.0723,
      "step": 729
    },
    {
      "epoch": 0.3506243996157541,
      "grad_norm": 0.6473589893605443,
      "learning_rate": 9.991392707952032e-06,
      "loss": 0.0595,
      "step": 730
    },
    {
      "epoch": 0.35110470701248797,
      "grad_norm": 0.9261456172897032,
      "learning_rate": 9.991228027008708e-06,
      "loss": 0.0842,
      "step": 731
    },
    {
      "epoch": 0.3515850144092219,
      "grad_norm": 1.0662113215148616,
      "learning_rate": 9.991061786943832e-06,
      "loss": 0.0724,
      "step": 732
    },
    {
      "epoch": 0.3520653218059558,
      "grad_norm": 0.9090716830219397,
      "learning_rate": 9.990893987809334e-06,
      "loss": 0.0574,
      "step": 733
    },
    {
      "epoch": 0.3525456292026897,
      "grad_norm": 0.5908937589147102,
      "learning_rate": 9.990724629657628e-06,
      "loss": 0.0514,
      "step": 734
    },
    {
      "epoch": 0.3530259365994236,
      "grad_norm": 1.0249840050472094,
      "learning_rate": 9.990553712541617e-06,
      "loss": 0.0838,
      "step": 735
    },
    {
      "epoch": 0.35350624399615754,
      "grad_norm": 0.7254263684691535,
      "learning_rate": 9.990381236514694e-06,
      "loss": 0.0568,
      "step": 736
    },
    {
      "epoch": 0.3539865513928914,
      "grad_norm": 0.7557297240778922,
      "learning_rate": 9.99020720163073e-06,
      "loss": 0.0613,
      "step": 737
    },
    {
      "epoch": 0.35446685878962536,
      "grad_norm": 0.7212710346277188,
      "learning_rate": 9.990031607944095e-06,
      "loss": 0.0615,
      "step": 738
    },
    {
      "epoch": 0.3549471661863593,
      "grad_norm": 0.5429849169379788,
      "learning_rate": 9.989854455509636e-06,
      "loss": 0.0508,
      "step": 739
    },
    {
      "epoch": 0.3554274735830932,
      "grad_norm": 0.790313789150935,
      "learning_rate": 9.98967574438269e-06,
      "loss": 0.0607,
      "step": 740
    },
    {
      "epoch": 0.3559077809798271,
      "grad_norm": 0.6588326074450415,
      "learning_rate": 9.989495474619084e-06,
      "loss": 0.0501,
      "step": 741
    },
    {
      "epoch": 0.356388088376561,
      "grad_norm": 1.0216588961310469,
      "learning_rate": 9.989313646275127e-06,
      "loss": 0.0817,
      "step": 742
    },
    {
      "epoch": 0.35686839577329493,
      "grad_norm": 1.2144514747608683,
      "learning_rate": 9.989130259407617e-06,
      "loss": 0.095,
      "step": 743
    },
    {
      "epoch": 0.3573487031700288,
      "grad_norm": 0.8230958178941028,
      "learning_rate": 9.988945314073842e-06,
      "loss": 0.0675,
      "step": 744
    },
    {
      "epoch": 0.35782901056676275,
      "grad_norm": 0.5442331145018258,
      "learning_rate": 9.988758810331572e-06,
      "loss": 0.0472,
      "step": 745
    },
    {
      "epoch": 0.35830931796349663,
      "grad_norm": 0.9420681725327574,
      "learning_rate": 9.988570748239062e-06,
      "loss": 0.0609,
      "step": 746
    },
    {
      "epoch": 0.35878962536023057,
      "grad_norm": 0.7063636013049791,
      "learning_rate": 9.988381127855063e-06,
      "loss": 0.0612,
      "step": 747
    },
    {
      "epoch": 0.35926993275696445,
      "grad_norm": 0.8868618317047221,
      "learning_rate": 9.988189949238804e-06,
      "loss": 0.0604,
      "step": 748
    },
    {
      "epoch": 0.3597502401536984,
      "grad_norm": 0.7054240383568051,
      "learning_rate": 9.987997212450007e-06,
      "loss": 0.0641,
      "step": 749
    },
    {
      "epoch": 0.36023054755043227,
      "grad_norm": 0.7475702380405793,
      "learning_rate": 9.987802917548874e-06,
      "loss": 0.0554,
      "step": 750
    },
    {
      "epoch": 0.3607108549471662,
      "grad_norm": 0.8420433761309497,
      "learning_rate": 9.9876070645961e-06,
      "loss": 0.0646,
      "step": 751
    },
    {
      "epoch": 0.3611911623439001,
      "grad_norm": 1.151439713731852,
      "learning_rate": 9.98740965365286e-06,
      "loss": 0.0894,
      "step": 752
    },
    {
      "epoch": 0.361671469740634,
      "grad_norm": 0.9861736223093036,
      "learning_rate": 9.987210684780826e-06,
      "loss": 0.0653,
      "step": 753
    },
    {
      "epoch": 0.3621517771373679,
      "grad_norm": 1.023603091918796,
      "learning_rate": 9.987010158042145e-06,
      "loss": 0.0716,
      "step": 754
    },
    {
      "epoch": 0.36263208453410184,
      "grad_norm": 0.7427517829264801,
      "learning_rate": 9.986808073499459e-06,
      "loss": 0.0588,
      "step": 755
    },
    {
      "epoch": 0.3631123919308357,
      "grad_norm": 1.0002779227618477,
      "learning_rate": 9.98660443121589e-06,
      "loss": 0.0601,
      "step": 756
    },
    {
      "epoch": 0.36359269932756966,
      "grad_norm": 0.7758536271834264,
      "learning_rate": 9.986399231255057e-06,
      "loss": 0.0629,
      "step": 757
    },
    {
      "epoch": 0.36407300672430354,
      "grad_norm": 0.5275682172040984,
      "learning_rate": 9.98619247368105e-06,
      "loss": 0.0591,
      "step": 758
    },
    {
      "epoch": 0.3645533141210375,
      "grad_norm": 0.8592188357318262,
      "learning_rate": 9.985984158558462e-06,
      "loss": 0.0686,
      "step": 759
    },
    {
      "epoch": 0.36503362151777136,
      "grad_norm": 0.6429739999465831,
      "learning_rate": 9.985774285952362e-06,
      "loss": 0.0875,
      "step": 760
    },
    {
      "epoch": 0.3655139289145053,
      "grad_norm": 0.9068989203363063,
      "learning_rate": 9.985562855928309e-06,
      "loss": 0.0728,
      "step": 761
    },
    {
      "epoch": 0.3659942363112392,
      "grad_norm": 0.8958890702630246,
      "learning_rate": 9.985349868552343e-06,
      "loss": 0.0762,
      "step": 762
    },
    {
      "epoch": 0.3664745437079731,
      "grad_norm": 0.9260397876251805,
      "learning_rate": 9.985135323891002e-06,
      "loss": 0.0639,
      "step": 763
    },
    {
      "epoch": 0.366954851104707,
      "grad_norm": 0.46145582448125044,
      "learning_rate": 9.984919222011301e-06,
      "loss": 0.0392,
      "step": 764
    },
    {
      "epoch": 0.36743515850144093,
      "grad_norm": 0.7368863607093822,
      "learning_rate": 9.984701562980745e-06,
      "loss": 0.0675,
      "step": 765
    },
    {
      "epoch": 0.3679154658981748,
      "grad_norm": 0.8374165642976674,
      "learning_rate": 9.984482346867325e-06,
      "loss": 0.0724,
      "step": 766
    },
    {
      "epoch": 0.36839577329490875,
      "grad_norm": 0.5004955641611937,
      "learning_rate": 9.984261573739515e-06,
      "loss": 0.0559,
      "step": 767
    },
    {
      "epoch": 0.3688760806916426,
      "grad_norm": 0.6896996348098048,
      "learning_rate": 9.984039243666284e-06,
      "loss": 0.0683,
      "step": 768
    },
    {
      "epoch": 0.36935638808837656,
      "grad_norm": 0.4917654383061967,
      "learning_rate": 9.983815356717075e-06,
      "loss": 0.0509,
      "step": 769
    },
    {
      "epoch": 0.36983669548511044,
      "grad_norm": 0.6841919810633339,
      "learning_rate": 9.983589912961828e-06,
      "loss": 0.0693,
      "step": 770
    },
    {
      "epoch": 0.3703170028818444,
      "grad_norm": 0.6557223578797912,
      "learning_rate": 9.983362912470967e-06,
      "loss": 0.064,
      "step": 771
    },
    {
      "epoch": 0.37079731027857826,
      "grad_norm": 0.6242477698953485,
      "learning_rate": 9.983134355315397e-06,
      "loss": 0.0587,
      "step": 772
    },
    {
      "epoch": 0.3712776176753122,
      "grad_norm": 0.5988595426851467,
      "learning_rate": 9.982904241566515e-06,
      "loss": 0.0634,
      "step": 773
    },
    {
      "epoch": 0.37175792507204614,
      "grad_norm": 0.6754592703701289,
      "learning_rate": 9.982672571296201e-06,
      "loss": 0.0629,
      "step": 774
    },
    {
      "epoch": 0.37223823246878,
      "grad_norm": 0.6078844744556268,
      "learning_rate": 9.982439344576824e-06,
      "loss": 0.0508,
      "step": 775
    },
    {
      "epoch": 0.37271853986551395,
      "grad_norm": 0.6857139355602675,
      "learning_rate": 9.982204561481237e-06,
      "loss": 0.0466,
      "step": 776
    },
    {
      "epoch": 0.37319884726224783,
      "grad_norm": 0.6033713609538075,
      "learning_rate": 9.981968222082778e-06,
      "loss": 0.0537,
      "step": 777
    },
    {
      "epoch": 0.37367915465898177,
      "grad_norm": 0.5844207725067214,
      "learning_rate": 9.981730326455275e-06,
      "loss": 0.0434,
      "step": 778
    },
    {
      "epoch": 0.37415946205571565,
      "grad_norm": 0.9540307943992555,
      "learning_rate": 9.98149087467304e-06,
      "loss": 0.0525,
      "step": 779
    },
    {
      "epoch": 0.3746397694524496,
      "grad_norm": 0.7214661073713216,
      "learning_rate": 9.98124986681087e-06,
      "loss": 0.0743,
      "step": 780
    },
    {
      "epoch": 0.37512007684918347,
      "grad_norm": 0.6813082935520749,
      "learning_rate": 9.981007302944048e-06,
      "loss": 0.0826,
      "step": 781
    },
    {
      "epoch": 0.3756003842459174,
      "grad_norm": 0.5582899213242017,
      "learning_rate": 9.980763183148347e-06,
      "loss": 0.0603,
      "step": 782
    },
    {
      "epoch": 0.3760806916426513,
      "grad_norm": 0.5713120406122318,
      "learning_rate": 9.980517507500023e-06,
      "loss": 0.0418,
      "step": 783
    },
    {
      "epoch": 0.3765609990393852,
      "grad_norm": 0.5631392163416006,
      "learning_rate": 9.980270276075816e-06,
      "loss": 0.076,
      "step": 784
    },
    {
      "epoch": 0.3770413064361191,
      "grad_norm": 0.6842884389848988,
      "learning_rate": 9.980021488952957e-06,
      "loss": 0.0555,
      "step": 785
    },
    {
      "epoch": 0.37752161383285304,
      "grad_norm": 0.6012427638288268,
      "learning_rate": 9.979771146209159e-06,
      "loss": 0.0581,
      "step": 786
    },
    {
      "epoch": 0.3780019212295869,
      "grad_norm": 0.6380907947348216,
      "learning_rate": 9.97951924792262e-06,
      "loss": 0.0558,
      "step": 787
    },
    {
      "epoch": 0.37848222862632086,
      "grad_norm": 0.4499455933552752,
      "learning_rate": 9.979265794172029e-06,
      "loss": 0.045,
      "step": 788
    },
    {
      "epoch": 0.37896253602305474,
      "grad_norm": 0.9207107334074435,
      "learning_rate": 9.979010785036557e-06,
      "loss": 0.0746,
      "step": 789
    },
    {
      "epoch": 0.3794428434197887,
      "grad_norm": 0.6001083473477673,
      "learning_rate": 9.978754220595861e-06,
      "loss": 0.0578,
      "step": 790
    },
    {
      "epoch": 0.37992315081652256,
      "grad_norm": 0.5596962433628107,
      "learning_rate": 9.978496100930086e-06,
      "loss": 0.0584,
      "step": 791
    },
    {
      "epoch": 0.3804034582132565,
      "grad_norm": 0.5021012713607019,
      "learning_rate": 9.978236426119862e-06,
      "loss": 0.0538,
      "step": 792
    },
    {
      "epoch": 0.3808837656099904,
      "grad_norm": 0.5102988447573241,
      "learning_rate": 9.977975196246302e-06,
      "loss": 0.0498,
      "step": 793
    },
    {
      "epoch": 0.3813640730067243,
      "grad_norm": 0.6861671569949817,
      "learning_rate": 9.97771241139101e-06,
      "loss": 0.0644,
      "step": 794
    },
    {
      "epoch": 0.3818443804034582,
      "grad_norm": 0.6357301541893875,
      "learning_rate": 9.977448071636068e-06,
      "loss": 0.0601,
      "step": 795
    },
    {
      "epoch": 0.38232468780019213,
      "grad_norm": 0.4024807592618827,
      "learning_rate": 9.977182177064053e-06,
      "loss": 0.0379,
      "step": 796
    },
    {
      "epoch": 0.382804995196926,
      "grad_norm": 1.1288610789705666,
      "learning_rate": 9.97691472775802e-06,
      "loss": 0.071,
      "step": 797
    },
    {
      "epoch": 0.38328530259365995,
      "grad_norm": 0.6469760760858952,
      "learning_rate": 9.976645723801515e-06,
      "loss": 0.0654,
      "step": 798
    },
    {
      "epoch": 0.38376560999039383,
      "grad_norm": 0.5560215918537325,
      "learning_rate": 9.976375165278567e-06,
      "loss": 0.0512,
      "step": 799
    },
    {
      "epoch": 0.38424591738712777,
      "grad_norm": 0.7975603385140326,
      "learning_rate": 9.976103052273689e-06,
      "loss": 0.0883,
      "step": 800
    },
    {
      "epoch": 0.38472622478386165,
      "grad_norm": 0.7728034217886376,
      "learning_rate": 9.975829384871884e-06,
      "loss": 0.072,
      "step": 801
    },
    {
      "epoch": 0.3852065321805956,
      "grad_norm": 0.7767874847251652,
      "learning_rate": 9.975554163158636e-06,
      "loss": 0.0464,
      "step": 802
    },
    {
      "epoch": 0.38568683957732947,
      "grad_norm": 0.6266192997429625,
      "learning_rate": 9.975277387219919e-06,
      "loss": 0.0528,
      "step": 803
    },
    {
      "epoch": 0.3861671469740634,
      "grad_norm": 0.7250311644638058,
      "learning_rate": 9.97499905714219e-06,
      "loss": 0.0686,
      "step": 804
    },
    {
      "epoch": 0.3866474543707973,
      "grad_norm": 1.1253149275559908,
      "learning_rate": 9.974719173012388e-06,
      "loss": 0.049,
      "step": 805
    },
    {
      "epoch": 0.3871277617675312,
      "grad_norm": 0.6222049434177256,
      "learning_rate": 9.974437734917945e-06,
      "loss": 0.0653,
      "step": 806
    },
    {
      "epoch": 0.38760806916426516,
      "grad_norm": 0.6636209039456584,
      "learning_rate": 9.974154742946775e-06,
      "loss": 0.0619,
      "step": 807
    },
    {
      "epoch": 0.38808837656099904,
      "grad_norm": 1.347074867858616,
      "learning_rate": 9.973870197187272e-06,
      "loss": 0.0659,
      "step": 808
    },
    {
      "epoch": 0.388568683957733,
      "grad_norm": 0.685156022786215,
      "learning_rate": 9.973584097728325e-06,
      "loss": 0.0545,
      "step": 809
    },
    {
      "epoch": 0.38904899135446686,
      "grad_norm": 0.8480519439599661,
      "learning_rate": 9.973296444659301e-06,
      "loss": 0.0659,
      "step": 810
    },
    {
      "epoch": 0.3895292987512008,
      "grad_norm": 0.4931591852346543,
      "learning_rate": 9.973007238070057e-06,
      "loss": 0.0543,
      "step": 811
    },
    {
      "epoch": 0.3900096061479347,
      "grad_norm": 0.410538917518561,
      "learning_rate": 9.97271647805093e-06,
      "loss": 0.0356,
      "step": 812
    },
    {
      "epoch": 0.3904899135446686,
      "grad_norm": 0.6183835853418933,
      "learning_rate": 9.972424164692748e-06,
      "loss": 0.0548,
      "step": 813
    },
    {
      "epoch": 0.3909702209414025,
      "grad_norm": 0.5518914943760818,
      "learning_rate": 9.972130298086821e-06,
      "loss": 0.0615,
      "step": 814
    },
    {
      "epoch": 0.3914505283381364,
      "grad_norm": 0.48864613714652044,
      "learning_rate": 9.971834878324944e-06,
      "loss": 0.0505,
      "step": 815
    },
    {
      "epoch": 0.3919308357348703,
      "grad_norm": 0.5320681967319949,
      "learning_rate": 9.971537905499397e-06,
      "loss": 0.0501,
      "step": 816
    },
    {
      "epoch": 0.39241114313160425,
      "grad_norm": 0.45876250806739416,
      "learning_rate": 9.971239379702951e-06,
      "loss": 0.0385,
      "step": 817
    },
    {
      "epoch": 0.3928914505283381,
      "grad_norm": 0.8069704061567374,
      "learning_rate": 9.970939301028853e-06,
      "loss": 0.0805,
      "step": 818
    },
    {
      "epoch": 0.39337175792507206,
      "grad_norm": 0.8977429261025726,
      "learning_rate": 9.970637669570838e-06,
      "loss": 0.0594,
      "step": 819
    },
    {
      "epoch": 0.39385206532180594,
      "grad_norm": 0.509233161828096,
      "learning_rate": 9.97033448542313e-06,
      "loss": 0.0577,
      "step": 820
    },
    {
      "epoch": 0.3943323727185399,
      "grad_norm": 0.5471204223282794,
      "learning_rate": 9.970029748680437e-06,
      "loss": 0.0433,
      "step": 821
    },
    {
      "epoch": 0.39481268011527376,
      "grad_norm": 0.8632034689344821,
      "learning_rate": 9.969723459437945e-06,
      "loss": 0.0597,
      "step": 822
    },
    {
      "epoch": 0.3952929875120077,
      "grad_norm": 0.9564865645949854,
      "learning_rate": 9.969415617791336e-06,
      "loss": 0.0627,
      "step": 823
    },
    {
      "epoch": 0.3957732949087416,
      "grad_norm": 0.7450452763817546,
      "learning_rate": 9.969106223836766e-06,
      "loss": 0.0547,
      "step": 824
    },
    {
      "epoch": 0.3962536023054755,
      "grad_norm": 0.5891613173133876,
      "learning_rate": 9.968795277670886e-06,
      "loss": 0.0482,
      "step": 825
    },
    {
      "epoch": 0.3967339097022094,
      "grad_norm": 0.5741708509441278,
      "learning_rate": 9.968482779390824e-06,
      "loss": 0.0699,
      "step": 826
    },
    {
      "epoch": 0.39721421709894333,
      "grad_norm": 0.5752534750406278,
      "learning_rate": 9.968168729094197e-06,
      "loss": 0.0513,
      "step": 827
    },
    {
      "epoch": 0.3976945244956772,
      "grad_norm": 0.5909076992409328,
      "learning_rate": 9.967853126879103e-06,
      "loss": 0.0584,
      "step": 828
    },
    {
      "epoch": 0.39817483189241115,
      "grad_norm": 0.5478980970759338,
      "learning_rate": 9.967535972844131e-06,
      "loss": 0.0578,
      "step": 829
    },
    {
      "epoch": 0.39865513928914503,
      "grad_norm": 0.43802378334815034,
      "learning_rate": 9.96721726708835e-06,
      "loss": 0.047,
      "step": 830
    },
    {
      "epoch": 0.39913544668587897,
      "grad_norm": 0.5964278773959507,
      "learning_rate": 9.966897009711314e-06,
      "loss": 0.0553,
      "step": 831
    },
    {
      "epoch": 0.39961575408261285,
      "grad_norm": 0.8075938997085979,
      "learning_rate": 9.966575200813064e-06,
      "loss": 0.0671,
      "step": 832
    },
    {
      "epoch": 0.4000960614793468,
      "grad_norm": 0.5724293589628411,
      "learning_rate": 9.966251840494123e-06,
      "loss": 0.0565,
      "step": 833
    },
    {
      "epoch": 0.40057636887608067,
      "grad_norm": 0.8710191021054293,
      "learning_rate": 9.965926928855498e-06,
      "loss": 0.0723,
      "step": 834
    },
    {
      "epoch": 0.4010566762728146,
      "grad_norm": 0.5363225318409294,
      "learning_rate": 9.965600465998686e-06,
      "loss": 0.0458,
      "step": 835
    },
    {
      "epoch": 0.4015369836695485,
      "grad_norm": 0.7851083772033061,
      "learning_rate": 9.965272452025666e-06,
      "loss": 0.044,
      "step": 836
    },
    {
      "epoch": 0.4020172910662824,
      "grad_norm": 0.7693177422501187,
      "learning_rate": 9.964942887038893e-06,
      "loss": 0.0475,
      "step": 837
    },
    {
      "epoch": 0.4024975984630163,
      "grad_norm": 0.7987401635767456,
      "learning_rate": 9.964611771141322e-06,
      "loss": 0.0651,
      "step": 838
    },
    {
      "epoch": 0.40297790585975024,
      "grad_norm": 0.8276640756002739,
      "learning_rate": 9.96427910443638e-06,
      "loss": 0.0785,
      "step": 839
    },
    {
      "epoch": 0.4034582132564842,
      "grad_norm": 0.6809367893973843,
      "learning_rate": 9.963944887027985e-06,
      "loss": 0.0715,
      "step": 840
    },
    {
      "epoch": 0.40393852065321806,
      "grad_norm": 0.6804973064252685,
      "learning_rate": 9.963609119020538e-06,
      "loss": 0.0585,
      "step": 841
    },
    {
      "epoch": 0.404418828049952,
      "grad_norm": 0.5534861907234978,
      "learning_rate": 9.963271800518921e-06,
      "loss": 0.0535,
      "step": 842
    },
    {
      "epoch": 0.4048991354466859,
      "grad_norm": 1.0547311716146044,
      "learning_rate": 9.962932931628504e-06,
      "loss": 0.0651,
      "step": 843
    },
    {
      "epoch": 0.4053794428434198,
      "grad_norm": 0.6295639846864554,
      "learning_rate": 9.96259251245514e-06,
      "loss": 0.0568,
      "step": 844
    },
    {
      "epoch": 0.4058597502401537,
      "grad_norm": 0.5719452618102057,
      "learning_rate": 9.962250543105167e-06,
      "loss": 0.0418,
      "step": 845
    },
    {
      "epoch": 0.40634005763688763,
      "grad_norm": 0.7395158167833543,
      "learning_rate": 9.961907023685407e-06,
      "loss": 0.0567,
      "step": 846
    },
    {
      "epoch": 0.4068203650336215,
      "grad_norm": 0.5391369878497289,
      "learning_rate": 9.961561954303164e-06,
      "loss": 0.0506,
      "step": 847
    },
    {
      "epoch": 0.40730067243035545,
      "grad_norm": 0.8066572923047143,
      "learning_rate": 9.961215335066232e-06,
      "loss": 0.0883,
      "step": 848
    },
    {
      "epoch": 0.40778097982708933,
      "grad_norm": 0.5151043209208244,
      "learning_rate": 9.960867166082884e-06,
      "loss": 0.044,
      "step": 849
    },
    {
      "epoch": 0.40826128722382327,
      "grad_norm": 0.6317828023573119,
      "learning_rate": 9.960517447461875e-06,
      "loss": 0.0468,
      "step": 850
    },
    {
      "epoch": 0.40874159462055715,
      "grad_norm": 0.5968432172723486,
      "learning_rate": 9.96016617931245e-06,
      "loss": 0.05,
      "step": 851
    },
    {
      "epoch": 0.4092219020172911,
      "grad_norm": 0.48351796440258127,
      "learning_rate": 9.959813361744337e-06,
      "loss": 0.0422,
      "step": 852
    },
    {
      "epoch": 0.40970220941402496,
      "grad_norm": 0.45514792871791687,
      "learning_rate": 9.959458994867744e-06,
      "loss": 0.052,
      "step": 853
    },
    {
      "epoch": 0.4101825168107589,
      "grad_norm": 0.8035467442104323,
      "learning_rate": 9.959103078793364e-06,
      "loss": 0.0897,
      "step": 854
    },
    {
      "epoch": 0.4106628242074928,
      "grad_norm": 0.7097960300705127,
      "learning_rate": 9.95874561363238e-06,
      "loss": 0.0551,
      "step": 855
    },
    {
      "epoch": 0.4111431316042267,
      "grad_norm": 0.40391671652511296,
      "learning_rate": 9.95838659949645e-06,
      "loss": 0.0414,
      "step": 856
    },
    {
      "epoch": 0.4116234390009606,
      "grad_norm": 0.6170051420564876,
      "learning_rate": 9.958026036497723e-06,
      "loss": 0.0559,
      "step": 857
    },
    {
      "epoch": 0.41210374639769454,
      "grad_norm": 0.7115787557160136,
      "learning_rate": 9.957663924748828e-06,
      "loss": 0.0565,
      "step": 858
    },
    {
      "epoch": 0.4125840537944284,
      "grad_norm": 0.8705757145217116,
      "learning_rate": 9.957300264362878e-06,
      "loss": 0.0674,
      "step": 859
    },
    {
      "epoch": 0.41306436119116235,
      "grad_norm": 0.5500919289906383,
      "learning_rate": 9.95693505545347e-06,
      "loss": 0.0621,
      "step": 860
    },
    {
      "epoch": 0.41354466858789624,
      "grad_norm": 0.660836487601902,
      "learning_rate": 9.956568298134687e-06,
      "loss": 0.0632,
      "step": 861
    },
    {
      "epoch": 0.4140249759846302,
      "grad_norm": 0.9931232470867721,
      "learning_rate": 9.956199992521092e-06,
      "loss": 0.0653,
      "step": 862
    },
    {
      "epoch": 0.41450528338136405,
      "grad_norm": 0.7024061823998546,
      "learning_rate": 9.955830138727736e-06,
      "loss": 0.0593,
      "step": 863
    },
    {
      "epoch": 0.414985590778098,
      "grad_norm": 0.4764259054817133,
      "learning_rate": 9.955458736870148e-06,
      "loss": 0.0526,
      "step": 864
    },
    {
      "epoch": 0.41546589817483187,
      "grad_norm": 0.7066580129944023,
      "learning_rate": 9.955085787064344e-06,
      "loss": 0.0541,
      "step": 865
    },
    {
      "epoch": 0.4159462055715658,
      "grad_norm": 0.665310242989133,
      "learning_rate": 9.954711289426826e-06,
      "loss": 0.061,
      "step": 866
    },
    {
      "epoch": 0.4164265129682997,
      "grad_norm": 0.5484058455441252,
      "learning_rate": 9.954335244074575e-06,
      "loss": 0.0459,
      "step": 867
    },
    {
      "epoch": 0.4169068203650336,
      "grad_norm": 0.5666272682660084,
      "learning_rate": 9.953957651125056e-06,
      "loss": 0.0557,
      "step": 868
    },
    {
      "epoch": 0.4173871277617675,
      "grad_norm": 0.8668633540680836,
      "learning_rate": 9.95357851069622e-06,
      "loss": 0.0768,
      "step": 869
    },
    {
      "epoch": 0.41786743515850144,
      "grad_norm": 0.5410731748495422,
      "learning_rate": 9.9531978229065e-06,
      "loss": 0.0627,
      "step": 870
    },
    {
      "epoch": 0.4183477425552353,
      "grad_norm": 0.5775107068931661,
      "learning_rate": 9.952815587874811e-06,
      "loss": 0.0496,
      "step": 871
    },
    {
      "epoch": 0.41882804995196926,
      "grad_norm": 0.690239624690725,
      "learning_rate": 9.952431805720555e-06,
      "loss": 0.0751,
      "step": 872
    },
    {
      "epoch": 0.41930835734870314,
      "grad_norm": 0.49376413841588845,
      "learning_rate": 9.952046476563614e-06,
      "loss": 0.0552,
      "step": 873
    },
    {
      "epoch": 0.4197886647454371,
      "grad_norm": 0.6948681019679253,
      "learning_rate": 9.951659600524353e-06,
      "loss": 0.0527,
      "step": 874
    },
    {
      "epoch": 0.420268972142171,
      "grad_norm": 0.49290241630164605,
      "learning_rate": 9.951271177723623e-06,
      "loss": 0.0506,
      "step": 875
    },
    {
      "epoch": 0.4207492795389049,
      "grad_norm": 0.8721248981892464,
      "learning_rate": 9.950881208282755e-06,
      "loss": 0.0569,
      "step": 876
    },
    {
      "epoch": 0.42122958693563883,
      "grad_norm": 0.5497841692019572,
      "learning_rate": 9.950489692323564e-06,
      "loss": 0.0633,
      "step": 877
    },
    {
      "epoch": 0.4217098943323727,
      "grad_norm": 0.6397458240250244,
      "learning_rate": 9.950096629968353e-06,
      "loss": 0.0537,
      "step": 878
    },
    {
      "epoch": 0.42219020172910665,
      "grad_norm": 0.5682428472860261,
      "learning_rate": 9.949702021339897e-06,
      "loss": 0.0558,
      "step": 879
    },
    {
      "epoch": 0.42267050912584053,
      "grad_norm": 1.0280188654825122,
      "learning_rate": 9.949305866561468e-06,
      "loss": 0.0732,
      "step": 880
    },
    {
      "epoch": 0.42315081652257447,
      "grad_norm": 0.6451938425896641,
      "learning_rate": 9.94890816575681e-06,
      "loss": 0.0668,
      "step": 881
    },
    {
      "epoch": 0.42363112391930835,
      "grad_norm": 0.45108905962672585,
      "learning_rate": 9.948508919050153e-06,
      "loss": 0.0309,
      "step": 882
    },
    {
      "epoch": 0.4241114313160423,
      "grad_norm": 0.7378292399931741,
      "learning_rate": 9.948108126566213e-06,
      "loss": 0.0805,
      "step": 883
    },
    {
      "epoch": 0.42459173871277617,
      "grad_norm": 0.5519105751376474,
      "learning_rate": 9.947705788430185e-06,
      "loss": 0.0521,
      "step": 884
    },
    {
      "epoch": 0.4250720461095101,
      "grad_norm": 0.6856313077407358,
      "learning_rate": 9.94730190476775e-06,
      "loss": 0.0644,
      "step": 885
    },
    {
      "epoch": 0.425552353506244,
      "grad_norm": 0.5969546888876337,
      "learning_rate": 9.946896475705067e-06,
      "loss": 0.0434,
      "step": 886
    },
    {
      "epoch": 0.4260326609029779,
      "grad_norm": 0.45619128279304955,
      "learning_rate": 9.946489501368783e-06,
      "loss": 0.051,
      "step": 887
    },
    {
      "epoch": 0.4265129682997118,
      "grad_norm": 0.6592474450888789,
      "learning_rate": 9.946080981886025e-06,
      "loss": 0.0482,
      "step": 888
    },
    {
      "epoch": 0.42699327569644574,
      "grad_norm": 0.8000680158848036,
      "learning_rate": 9.945670917384404e-06,
      "loss": 0.0779,
      "step": 889
    },
    {
      "epoch": 0.4274735830931796,
      "grad_norm": 0.5334860843524303,
      "learning_rate": 9.94525930799201e-06,
      "loss": 0.0489,
      "step": 890
    },
    {
      "epoch": 0.42795389048991356,
      "grad_norm": 0.5254575930785399,
      "learning_rate": 9.944846153837423e-06,
      "loss": 0.0397,
      "step": 891
    },
    {
      "epoch": 0.42843419788664744,
      "grad_norm": 0.5660738613547326,
      "learning_rate": 9.944431455049697e-06,
      "loss": 0.0529,
      "step": 892
    },
    {
      "epoch": 0.4289145052833814,
      "grad_norm": 0.7532191005217946,
      "learning_rate": 9.944015211758375e-06,
      "loss": 0.0567,
      "step": 893
    },
    {
      "epoch": 0.42939481268011526,
      "grad_norm": 0.6751729764036046,
      "learning_rate": 9.943597424093477e-06,
      "loss": 0.0536,
      "step": 894
    },
    {
      "epoch": 0.4298751200768492,
      "grad_norm": 0.7399414072106839,
      "learning_rate": 9.943178092185511e-06,
      "loss": 0.0627,
      "step": 895
    },
    {
      "epoch": 0.4303554274735831,
      "grad_norm": 0.5319921340444789,
      "learning_rate": 9.942757216165464e-06,
      "loss": 0.0526,
      "step": 896
    },
    {
      "epoch": 0.430835734870317,
      "grad_norm": 0.9231172091302492,
      "learning_rate": 9.942334796164805e-06,
      "loss": 0.0611,
      "step": 897
    },
    {
      "epoch": 0.4313160422670509,
      "grad_norm": 0.7768175094389944,
      "learning_rate": 9.941910832315488e-06,
      "loss": 0.0639,
      "step": 898
    },
    {
      "epoch": 0.43179634966378483,
      "grad_norm": 0.4825476064702085,
      "learning_rate": 9.941485324749947e-06,
      "loss": 0.0518,
      "step": 899
    },
    {
      "epoch": 0.4322766570605187,
      "grad_norm": 0.7021828010660213,
      "learning_rate": 9.941058273601097e-06,
      "loss": 0.0593,
      "step": 900
    },
    {
      "epoch": 0.43275696445725265,
      "grad_norm": 0.5871945041167091,
      "learning_rate": 9.94062967900234e-06,
      "loss": 0.0574,
      "step": 901
    },
    {
      "epoch": 0.4332372718539865,
      "grad_norm": 0.7091892488086937,
      "learning_rate": 9.940199541087554e-06,
      "loss": 0.0709,
      "step": 902
    },
    {
      "epoch": 0.43371757925072046,
      "grad_norm": 0.5164796945559618,
      "learning_rate": 9.939767859991104e-06,
      "loss": 0.0567,
      "step": 903
    },
    {
      "epoch": 0.43419788664745435,
      "grad_norm": 1.0053328543134505,
      "learning_rate": 9.939334635847834e-06,
      "loss": 0.0623,
      "step": 904
    },
    {
      "epoch": 0.4346781940441883,
      "grad_norm": 1.014246583976263,
      "learning_rate": 9.938899868793074e-06,
      "loss": 0.0678,
      "step": 905
    },
    {
      "epoch": 0.43515850144092216,
      "grad_norm": 0.620672410400972,
      "learning_rate": 9.93846355896263e-06,
      "loss": 0.0499,
      "step": 906
    },
    {
      "epoch": 0.4356388088376561,
      "grad_norm": 0.6018522052869041,
      "learning_rate": 9.938025706492796e-06,
      "loss": 0.0633,
      "step": 907
    },
    {
      "epoch": 0.43611911623439004,
      "grad_norm": 0.7802965838379325,
      "learning_rate": 9.937586311520342e-06,
      "loss": 0.0542,
      "step": 908
    },
    {
      "epoch": 0.4365994236311239,
      "grad_norm": 0.7890750457426956,
      "learning_rate": 9.937145374182523e-06,
      "loss": 0.0674,
      "step": 909
    },
    {
      "epoch": 0.43707973102785785,
      "grad_norm": 0.6932375338768142,
      "learning_rate": 9.936702894617081e-06,
      "loss": 0.0705,
      "step": 910
    },
    {
      "epoch": 0.43756003842459174,
      "grad_norm": 0.6257587333406103,
      "learning_rate": 9.936258872962229e-06,
      "loss": 0.0628,
      "step": 911
    },
    {
      "epoch": 0.43804034582132567,
      "grad_norm": 0.5607233207353541,
      "learning_rate": 9.935813309356666e-06,
      "loss": 0.0544,
      "step": 912
    },
    {
      "epoch": 0.43852065321805955,
      "grad_norm": 0.6321356409882269,
      "learning_rate": 9.935366203939579e-06,
      "loss": 0.0491,
      "step": 913
    },
    {
      "epoch": 0.4390009606147935,
      "grad_norm": 0.6757767770961596,
      "learning_rate": 9.934917556850625e-06,
      "loss": 0.0434,
      "step": 914
    },
    {
      "epoch": 0.43948126801152737,
      "grad_norm": 0.7906338628745093,
      "learning_rate": 9.934467368229955e-06,
      "loss": 0.0638,
      "step": 915
    },
    {
      "epoch": 0.4399615754082613,
      "grad_norm": 0.4652920811833901,
      "learning_rate": 9.934015638218193e-06,
      "loss": 0.0519,
      "step": 916
    },
    {
      "epoch": 0.4404418828049952,
      "grad_norm": 0.5470520415253206,
      "learning_rate": 9.933562366956445e-06,
      "loss": 0.0574,
      "step": 917
    },
    {
      "epoch": 0.4409221902017291,
      "grad_norm": 0.7607179274563025,
      "learning_rate": 9.933107554586303e-06,
      "loss": 0.0675,
      "step": 918
    },
    {
      "epoch": 0.441402497598463,
      "grad_norm": 0.5532199345436009,
      "learning_rate": 9.93265120124984e-06,
      "loss": 0.0446,
      "step": 919
    },
    {
      "epoch": 0.44188280499519694,
      "grad_norm": 0.501284318665828,
      "learning_rate": 9.932193307089602e-06,
      "loss": 0.0518,
      "step": 920
    },
    {
      "epoch": 0.4423631123919308,
      "grad_norm": 0.6013902688897093,
      "learning_rate": 9.931733872248626e-06,
      "loss": 0.0584,
      "step": 921
    },
    {
      "epoch": 0.44284341978866476,
      "grad_norm": 0.5972118740317204,
      "learning_rate": 9.931272896870427e-06,
      "loss": 0.0476,
      "step": 922
    },
    {
      "epoch": 0.44332372718539864,
      "grad_norm": 2.253298791198679,
      "learning_rate": 9.930810381098999e-06,
      "loss": 0.0561,
      "step": 923
    },
    {
      "epoch": 0.4438040345821326,
      "grad_norm": 0.6102253348891266,
      "learning_rate": 9.93034632507882e-06,
      "loss": 0.0482,
      "step": 924
    },
    {
      "epoch": 0.44428434197886646,
      "grad_norm": 0.4713494577341782,
      "learning_rate": 9.929880728954853e-06,
      "loss": 0.0371,
      "step": 925
    },
    {
      "epoch": 0.4447646493756004,
      "grad_norm": 0.6627531547942727,
      "learning_rate": 9.92941359287253e-06,
      "loss": 0.0509,
      "step": 926
    },
    {
      "epoch": 0.4452449567723343,
      "grad_norm": 0.5428810469885857,
      "learning_rate": 9.928944916977775e-06,
      "loss": 0.051,
      "step": 927
    },
    {
      "epoch": 0.4457252641690682,
      "grad_norm": 0.6643297934290499,
      "learning_rate": 9.92847470141699e-06,
      "loss": 0.0765,
      "step": 928
    },
    {
      "epoch": 0.4462055715658021,
      "grad_norm": 0.5681242371551549,
      "learning_rate": 9.928002946337055e-06,
      "loss": 0.0487,
      "step": 929
    },
    {
      "epoch": 0.44668587896253603,
      "grad_norm": 0.45349350065521055,
      "learning_rate": 9.927529651885334e-06,
      "loss": 0.0548,
      "step": 930
    },
    {
      "epoch": 0.4471661863592699,
      "grad_norm": 0.42679372029366797,
      "learning_rate": 9.92705481820967e-06,
      "loss": 0.0529,
      "step": 931
    },
    {
      "epoch": 0.44764649375600385,
      "grad_norm": 0.4695476492659101,
      "learning_rate": 9.926578445458393e-06,
      "loss": 0.0414,
      "step": 932
    },
    {
      "epoch": 0.44812680115273773,
      "grad_norm": 0.5300464019739584,
      "learning_rate": 9.926100533780304e-06,
      "loss": 0.0439,
      "step": 933
    },
    {
      "epoch": 0.44860710854947167,
      "grad_norm": 0.4180296006214648,
      "learning_rate": 9.92562108332469e-06,
      "loss": 0.0391,
      "step": 934
    },
    {
      "epoch": 0.44908741594620555,
      "grad_norm": 0.40241583405442505,
      "learning_rate": 9.92514009424132e-06,
      "loss": 0.0523,
      "step": 935
    },
    {
      "epoch": 0.4495677233429395,
      "grad_norm": 0.4954138376589334,
      "learning_rate": 9.924657566680438e-06,
      "loss": 0.058,
      "step": 936
    },
    {
      "epoch": 0.45004803073967337,
      "grad_norm": 0.6273768947654529,
      "learning_rate": 9.924173500792775e-06,
      "loss": 0.055,
      "step": 937
    },
    {
      "epoch": 0.4505283381364073,
      "grad_norm": 0.3724637804837822,
      "learning_rate": 9.92368789672954e-06,
      "loss": 0.0334,
      "step": 938
    },
    {
      "epoch": 0.4510086455331412,
      "grad_norm": 0.5849978243981226,
      "learning_rate": 9.923200754642422e-06,
      "loss": 0.0443,
      "step": 939
    },
    {
      "epoch": 0.4514889529298751,
      "grad_norm": 0.49650880356005744,
      "learning_rate": 9.92271207468359e-06,
      "loss": 0.0436,
      "step": 940
    },
    {
      "epoch": 0.45196926032660906,
      "grad_norm": 0.44519643938812636,
      "learning_rate": 9.922221857005693e-06,
      "loss": 0.0523,
      "step": 941
    },
    {
      "epoch": 0.45244956772334294,
      "grad_norm": 0.983957220828581,
      "learning_rate": 9.921730101761865e-06,
      "loss": 0.0945,
      "step": 942
    },
    {
      "epoch": 0.4529298751200769,
      "grad_norm": 0.6035634372006827,
      "learning_rate": 9.921236809105711e-06,
      "loss": 0.0522,
      "step": 943
    },
    {
      "epoch": 0.45341018251681076,
      "grad_norm": 0.9095761086634716,
      "learning_rate": 9.92074197919133e-06,
      "loss": 0.0482,
      "step": 944
    },
    {
      "epoch": 0.4538904899135447,
      "grad_norm": 0.7038000923358203,
      "learning_rate": 9.920245612173288e-06,
      "loss": 0.06,
      "step": 945
    },
    {
      "epoch": 0.4543707973102786,
      "grad_norm": 0.5317259635381724,
      "learning_rate": 9.919747708206635e-06,
      "loss": 0.0513,
      "step": 946
    },
    {
      "epoch": 0.4548511047070125,
      "grad_norm": 1.612873760960674,
      "learning_rate": 9.919248267446904e-06,
      "loss": 0.0555,
      "step": 947
    },
    {
      "epoch": 0.4553314121037464,
      "grad_norm": 0.46423671806106315,
      "learning_rate": 9.918747290050108e-06,
      "loss": 0.0368,
      "step": 948
    },
    {
      "epoch": 0.45581171950048033,
      "grad_norm": 0.7137648813885012,
      "learning_rate": 9.918244776172739e-06,
      "loss": 0.058,
      "step": 949
    },
    {
      "epoch": 0.4562920268972142,
      "grad_norm": 0.677167631389359,
      "learning_rate": 9.917740725971765e-06,
      "loss": 0.0476,
      "step": 950
    },
    {
      "epoch": 0.45677233429394815,
      "grad_norm": 0.6324916873646828,
      "learning_rate": 9.91723513960464e-06,
      "loss": 0.0524,
      "step": 951
    },
    {
      "epoch": 0.457252641690682,
      "grad_norm": 0.5324488356285513,
      "learning_rate": 9.916728017229293e-06,
      "loss": 0.0398,
      "step": 952
    },
    {
      "epoch": 0.45773294908741596,
      "grad_norm": 0.6324777151535353,
      "learning_rate": 9.916219359004137e-06,
      "loss": 0.0571,
      "step": 953
    },
    {
      "epoch": 0.45821325648414984,
      "grad_norm": 0.6435220215655985,
      "learning_rate": 9.915709165088063e-06,
      "loss": 0.0696,
      "step": 954
    },
    {
      "epoch": 0.4586935638808838,
      "grad_norm": 0.8164681729134681,
      "learning_rate": 9.91519743564044e-06,
      "loss": 0.0611,
      "step": 955
    },
    {
      "epoch": 0.45917387127761766,
      "grad_norm": 0.5313310873883075,
      "learning_rate": 9.914684170821119e-06,
      "loss": 0.0476,
      "step": 956
    },
    {
      "epoch": 0.4596541786743516,
      "grad_norm": 0.5973052937068528,
      "learning_rate": 9.91416937079043e-06,
      "loss": 0.0757,
      "step": 957
    },
    {
      "epoch": 0.4601344860710855,
      "grad_norm": 0.4440073895198674,
      "learning_rate": 9.91365303570918e-06,
      "loss": 0.0435,
      "step": 958
    },
    {
      "epoch": 0.4606147934678194,
      "grad_norm": 0.5635682614371156,
      "learning_rate": 9.913135165738661e-06,
      "loss": 0.0601,
      "step": 959
    },
    {
      "epoch": 0.4610951008645533,
      "grad_norm": 0.6205543687338353,
      "learning_rate": 9.91261576104064e-06,
      "loss": 0.0607,
      "step": 960
    },
    {
      "epoch": 0.46157540826128723,
      "grad_norm": 0.8474866942319256,
      "learning_rate": 9.912094821777362e-06,
      "loss": 0.0508,
      "step": 961
    },
    {
      "epoch": 0.4620557156580211,
      "grad_norm": 0.4415856929796406,
      "learning_rate": 9.91157234811156e-06,
      "loss": 0.0426,
      "step": 962
    },
    {
      "epoch": 0.46253602305475505,
      "grad_norm": 0.4965072565036624,
      "learning_rate": 9.911048340206435e-06,
      "loss": 0.0485,
      "step": 963
    },
    {
      "epoch": 0.46301633045148893,
      "grad_norm": 0.7207884825617963,
      "learning_rate": 9.910522798225673e-06,
      "loss": 0.0767,
      "step": 964
    },
    {
      "epoch": 0.46349663784822287,
      "grad_norm": 0.6441736297504665,
      "learning_rate": 9.909995722333442e-06,
      "loss": 0.0582,
      "step": 965
    },
    {
      "epoch": 0.46397694524495675,
      "grad_norm": 0.6229851666491163,
      "learning_rate": 9.909467112694385e-06,
      "loss": 0.0649,
      "step": 966
    },
    {
      "epoch": 0.4644572526416907,
      "grad_norm": 0.7004494943555968,
      "learning_rate": 9.908936969473621e-06,
      "loss": 0.0479,
      "step": 967
    },
    {
      "epoch": 0.46493756003842457,
      "grad_norm": 0.7544245132592255,
      "learning_rate": 9.908405292836758e-06,
      "loss": 0.0604,
      "step": 968
    },
    {
      "epoch": 0.4654178674351585,
      "grad_norm": 0.7931595109408737,
      "learning_rate": 9.907872082949873e-06,
      "loss": 0.0822,
      "step": 969
    },
    {
      "epoch": 0.4658981748318924,
      "grad_norm": 0.6826313680933557,
      "learning_rate": 9.907337339979525e-06,
      "loss": 0.0481,
      "step": 970
    },
    {
      "epoch": 0.4663784822286263,
      "grad_norm": 0.5041754294571876,
      "learning_rate": 9.90680106409276e-06,
      "loss": 0.0417,
      "step": 971
    },
    {
      "epoch": 0.4668587896253602,
      "grad_norm": 0.8823164058501882,
      "learning_rate": 9.906263255457087e-06,
      "loss": 0.0652,
      "step": 972
    },
    {
      "epoch": 0.46733909702209414,
      "grad_norm": 0.7871751423538097,
      "learning_rate": 9.905723914240507e-06,
      "loss": 0.0654,
      "step": 973
    },
    {
      "epoch": 0.4678194044188281,
      "grad_norm": 0.8027977963955859,
      "learning_rate": 9.905183040611498e-06,
      "loss": 0.0545,
      "step": 974
    },
    {
      "epoch": 0.46829971181556196,
      "grad_norm": 0.8567181353492188,
      "learning_rate": 9.904640634739007e-06,
      "loss": 0.0728,
      "step": 975
    },
    {
      "epoch": 0.4687800192122959,
      "grad_norm": 0.4735301698046212,
      "learning_rate": 9.904096696792472e-06,
      "loss": 0.0402,
      "step": 976
    },
    {
      "epoch": 0.4692603266090298,
      "grad_norm": 0.9898206363808187,
      "learning_rate": 9.903551226941801e-06,
      "loss": 0.078,
      "step": 977
    },
    {
      "epoch": 0.4697406340057637,
      "grad_norm": 0.6741220652957427,
      "learning_rate": 9.903004225357387e-06,
      "loss": 0.0793,
      "step": 978
    },
    {
      "epoch": 0.4702209414024976,
      "grad_norm": 0.44391076772330806,
      "learning_rate": 9.902455692210094e-06,
      "loss": 0.0519,
      "step": 979
    },
    {
      "epoch": 0.47070124879923153,
      "grad_norm": 0.8124544269575636,
      "learning_rate": 9.901905627671273e-06,
      "loss": 0.0534,
      "step": 980
    },
    {
      "epoch": 0.4711815561959654,
      "grad_norm": 0.7481718577639421,
      "learning_rate": 9.901354031912746e-06,
      "loss": 0.0819,
      "step": 981
    },
    {
      "epoch": 0.47166186359269935,
      "grad_norm": 1.8659551510872119,
      "learning_rate": 9.900800905106817e-06,
      "loss": 0.0675,
      "step": 982
    },
    {
      "epoch": 0.47214217098943323,
      "grad_norm": 0.695535075758422,
      "learning_rate": 9.900246247426269e-06,
      "loss": 0.0622,
      "step": 983
    },
    {
      "epoch": 0.47262247838616717,
      "grad_norm": 0.49991906676132497,
      "learning_rate": 9.899690059044358e-06,
      "loss": 0.0573,
      "step": 984
    },
    {
      "epoch": 0.47310278578290105,
      "grad_norm": 0.8200411105348173,
      "learning_rate": 9.899132340134825e-06,
      "loss": 0.0867,
      "step": 985
    },
    {
      "epoch": 0.473583093179635,
      "grad_norm": 0.6775421926604714,
      "learning_rate": 9.898573090871885e-06,
      "loss": 0.05,
      "step": 986
    },
    {
      "epoch": 0.47406340057636887,
      "grad_norm": 0.5492047748590122,
      "learning_rate": 9.898012311430232e-06,
      "loss": 0.0538,
      "step": 987
    },
    {
      "epoch": 0.4745437079731028,
      "grad_norm": 0.6656920927770863,
      "learning_rate": 9.897450001985038e-06,
      "loss": 0.0677,
      "step": 988
    },
    {
      "epoch": 0.4750240153698367,
      "grad_norm": 1.072363462917926,
      "learning_rate": 9.896886162711955e-06,
      "loss": 0.0523,
      "step": 989
    },
    {
      "epoch": 0.4755043227665706,
      "grad_norm": 0.627671736207023,
      "learning_rate": 9.896320793787106e-06,
      "loss": 0.0499,
      "step": 990
    },
    {
      "epoch": 0.4759846301633045,
      "grad_norm": 0.8567917544335966,
      "learning_rate": 9.895753895387101e-06,
      "loss": 0.0564,
      "step": 991
    },
    {
      "epoch": 0.47646493756003844,
      "grad_norm": 1.0192507334666507,
      "learning_rate": 9.895185467689022e-06,
      "loss": 0.0674,
      "step": 992
    },
    {
      "epoch": 0.4769452449567723,
      "grad_norm": 0.7961715978827638,
      "learning_rate": 9.894615510870429e-06,
      "loss": 0.0831,
      "step": 993
    },
    {
      "epoch": 0.47742555235350626,
      "grad_norm": 0.5483648725578345,
      "learning_rate": 9.894044025109363e-06,
      "loss": 0.0485,
      "step": 994
    },
    {
      "epoch": 0.47790585975024014,
      "grad_norm": 0.8078314585601833,
      "learning_rate": 9.893471010584337e-06,
      "loss": 0.0625,
      "step": 995
    },
    {
      "epoch": 0.4783861671469741,
      "grad_norm": 0.6306541062019995,
      "learning_rate": 9.892896467474348e-06,
      "loss": 0.0678,
      "step": 996
    },
    {
      "epoch": 0.47886647454370795,
      "grad_norm": 0.4985055781130739,
      "learning_rate": 9.892320395958865e-06,
      "loss": 0.0508,
      "step": 997
    },
    {
      "epoch": 0.4793467819404419,
      "grad_norm": 0.5440005750071006,
      "learning_rate": 9.89174279621784e-06,
      "loss": 0.0441,
      "step": 998
    },
    {
      "epoch": 0.47982708933717577,
      "grad_norm": 0.7286303002598773,
      "learning_rate": 9.891163668431696e-06,
      "loss": 0.0555,
      "step": 999
    },
    {
      "epoch": 0.4803073967339097,
      "grad_norm": 0.5690411831803118,
      "learning_rate": 9.890583012781338e-06,
      "loss": 0.0458,
      "step": 1000
    },
    {
      "epoch": 0.4807877041306436,
      "grad_norm": 0.46362095698553785,
      "learning_rate": 9.890000829448145e-06,
      "loss": 0.0385,
      "step": 1001
    },
    {
      "epoch": 0.4812680115273775,
      "grad_norm": 0.6543145086812504,
      "learning_rate": 9.889417118613978e-06,
      "loss": 0.0683,
      "step": 1002
    },
    {
      "epoch": 0.4817483189241114,
      "grad_norm": 0.556537310472474,
      "learning_rate": 9.888831880461171e-06,
      "loss": 0.0454,
      "step": 1003
    },
    {
      "epoch": 0.48222862632084534,
      "grad_norm": 0.6944965802448689,
      "learning_rate": 9.888245115172535e-06,
      "loss": 0.0499,
      "step": 1004
    },
    {
      "epoch": 0.4827089337175792,
      "grad_norm": 1.9875426185604483,
      "learning_rate": 9.88765682293136e-06,
      "loss": 0.0623,
      "step": 1005
    },
    {
      "epoch": 0.48318924111431316,
      "grad_norm": 0.5589465167484113,
      "learning_rate": 9.887067003921412e-06,
      "loss": 0.0549,
      "step": 1006
    },
    {
      "epoch": 0.48366954851104704,
      "grad_norm": 0.5825022233011566,
      "learning_rate": 9.886475658326935e-06,
      "loss": 0.0415,
      "step": 1007
    },
    {
      "epoch": 0.484149855907781,
      "grad_norm": 0.8085519967345373,
      "learning_rate": 9.885882786332647e-06,
      "loss": 0.0618,
      "step": 1008
    },
    {
      "epoch": 0.4846301633045149,
      "grad_norm": 0.6678730905959983,
      "learning_rate": 9.885288388123748e-06,
      "loss": 0.0466,
      "step": 1009
    },
    {
      "epoch": 0.4851104707012488,
      "grad_norm": 0.6528204330501064,
      "learning_rate": 9.88469246388591e-06,
      "loss": 0.0462,
      "step": 1010
    },
    {
      "epoch": 0.48559077809798273,
      "grad_norm": 0.50393839362882,
      "learning_rate": 9.884095013805282e-06,
      "loss": 0.0418,
      "step": 1011
    },
    {
      "epoch": 0.4860710854947166,
      "grad_norm": 0.6274523697170888,
      "learning_rate": 9.88349603806849e-06,
      "loss": 0.0687,
      "step": 1012
    },
    {
      "epoch": 0.48655139289145055,
      "grad_norm": 0.6366989175368812,
      "learning_rate": 9.882895536862643e-06,
      "loss": 0.0485,
      "step": 1013
    },
    {
      "epoch": 0.48703170028818443,
      "grad_norm": 0.48888844573115503,
      "learning_rate": 9.882293510375314e-06,
      "loss": 0.0512,
      "step": 1014
    },
    {
      "epoch": 0.48751200768491837,
      "grad_norm": 0.5932068169061537,
      "learning_rate": 9.881689958794564e-06,
      "loss": 0.0488,
      "step": 1015
    },
    {
      "epoch": 0.48799231508165225,
      "grad_norm": 1.2627441745695316,
      "learning_rate": 9.881084882308924e-06,
      "loss": 0.0714,
      "step": 1016
    },
    {
      "epoch": 0.4884726224783862,
      "grad_norm": 0.9627073875348847,
      "learning_rate": 9.880478281107404e-06,
      "loss": 0.0577,
      "step": 1017
    },
    {
      "epoch": 0.48895292987512007,
      "grad_norm": 0.7088884846326223,
      "learning_rate": 9.87987015537949e-06,
      "loss": 0.0874,
      "step": 1018
    },
    {
      "epoch": 0.489433237271854,
      "grad_norm": 0.7529185355526793,
      "learning_rate": 9.879260505315143e-06,
      "loss": 0.0497,
      "step": 1019
    },
    {
      "epoch": 0.4899135446685879,
      "grad_norm": 0.6165894758520094,
      "learning_rate": 9.878649331104798e-06,
      "loss": 0.0604,
      "step": 1020
    },
    {
      "epoch": 0.4903938520653218,
      "grad_norm": 0.5042776271300216,
      "learning_rate": 9.878036632939374e-06,
      "loss": 0.047,
      "step": 1021
    },
    {
      "epoch": 0.4908741594620557,
      "grad_norm": 0.44643658788918444,
      "learning_rate": 9.877422411010257e-06,
      "loss": 0.0498,
      "step": 1022
    },
    {
      "epoch": 0.49135446685878964,
      "grad_norm": 0.5634531390244911,
      "learning_rate": 9.876806665509314e-06,
      "loss": 0.0569,
      "step": 1023
    },
    {
      "epoch": 0.4918347742555235,
      "grad_norm": 0.5619776061410058,
      "learning_rate": 9.876189396628889e-06,
      "loss": 0.0446,
      "step": 1024
    },
    {
      "epoch": 0.49231508165225746,
      "grad_norm": 0.9390023063566659,
      "learning_rate": 9.875570604561796e-06,
      "loss": 0.0792,
      "step": 1025
    },
    {
      "epoch": 0.49279538904899134,
      "grad_norm": 0.5013365650894346,
      "learning_rate": 9.874950289501332e-06,
      "loss": 0.0519,
      "step": 1026
    },
    {
      "epoch": 0.4932756964457253,
      "grad_norm": 0.46104032503551207,
      "learning_rate": 9.874328451641264e-06,
      "loss": 0.0458,
      "step": 1027
    },
    {
      "epoch": 0.49375600384245916,
      "grad_norm": 0.46235258731993134,
      "learning_rate": 9.873705091175838e-06,
      "loss": 0.0417,
      "step": 1028
    },
    {
      "epoch": 0.4942363112391931,
      "grad_norm": 0.49971536792567545,
      "learning_rate": 9.873080208299773e-06,
      "loss": 0.045,
      "step": 1029
    },
    {
      "epoch": 0.494716618635927,
      "grad_norm": 0.5340542754950781,
      "learning_rate": 9.872453803208268e-06,
      "loss": 0.0455,
      "step": 1030
    },
    {
      "epoch": 0.4951969260326609,
      "grad_norm": 0.5727266302637889,
      "learning_rate": 9.871825876096992e-06,
      "loss": 0.0461,
      "step": 1031
    },
    {
      "epoch": 0.4956772334293948,
      "grad_norm": 0.6471275463338492,
      "learning_rate": 9.871196427162094e-06,
      "loss": 0.0531,
      "step": 1032
    },
    {
      "epoch": 0.49615754082612873,
      "grad_norm": 0.8462618075680615,
      "learning_rate": 9.870565456600194e-06,
      "loss": 0.0913,
      "step": 1033
    },
    {
      "epoch": 0.4966378482228626,
      "grad_norm": 0.6248139766098908,
      "learning_rate": 9.869932964608392e-06,
      "loss": 0.0488,
      "step": 1034
    },
    {
      "epoch": 0.49711815561959655,
      "grad_norm": 0.5031224522406227,
      "learning_rate": 9.86929895138426e-06,
      "loss": 0.0463,
      "step": 1035
    },
    {
      "epoch": 0.49759846301633043,
      "grad_norm": 0.6015714100591678,
      "learning_rate": 9.868663417125849e-06,
      "loss": 0.0543,
      "step": 1036
    },
    {
      "epoch": 0.49807877041306436,
      "grad_norm": 0.7015166961244664,
      "learning_rate": 9.868026362031676e-06,
      "loss": 0.0577,
      "step": 1037
    },
    {
      "epoch": 0.49855907780979825,
      "grad_norm": 0.5811321733944121,
      "learning_rate": 9.867387786300743e-06,
      "loss": 0.046,
      "step": 1038
    },
    {
      "epoch": 0.4990393852065322,
      "grad_norm": 0.9085434522653888,
      "learning_rate": 9.866747690132527e-06,
      "loss": 0.073,
      "step": 1039
    },
    {
      "epoch": 0.49951969260326606,
      "grad_norm": 0.6591822101096578,
      "learning_rate": 9.866106073726971e-06,
      "loss": 0.0567,
      "step": 1040
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.835558730775927,
      "learning_rate": 9.865462937284501e-06,
      "loss": 0.0535,
      "step": 1041
    },
    {
      "epoch": 0.5004803073967339,
      "grad_norm": 0.5892727698062462,
      "learning_rate": 9.864818281006013e-06,
      "loss": 0.0439,
      "step": 1042
    },
    {
      "epoch": 0.5009606147934679,
      "grad_norm": 0.8395833865727974,
      "learning_rate": 9.86417210509288e-06,
      "loss": 0.0718,
      "step": 1043
    },
    {
      "epoch": 0.5014409221902018,
      "grad_norm": 1.1531297612195082,
      "learning_rate": 9.86352440974695e-06,
      "loss": 0.0701,
      "step": 1044
    },
    {
      "epoch": 0.5019212295869356,
      "grad_norm": 0.5682920886173015,
      "learning_rate": 9.862875195170547e-06,
      "loss": 0.0478,
      "step": 1045
    },
    {
      "epoch": 0.5024015369836695,
      "grad_norm": 0.5479631332409276,
      "learning_rate": 9.862224461566467e-06,
      "loss": 0.0481,
      "step": 1046
    },
    {
      "epoch": 0.5028818443804035,
      "grad_norm": 0.6686914490763103,
      "learning_rate": 9.861572209137978e-06,
      "loss": 0.054,
      "step": 1047
    },
    {
      "epoch": 0.5033621517771374,
      "grad_norm": 0.5548966562906734,
      "learning_rate": 9.860918438088828e-06,
      "loss": 0.058,
      "step": 1048
    },
    {
      "epoch": 0.5038424591738713,
      "grad_norm": 0.6221834543978872,
      "learning_rate": 9.860263148623238e-06,
      "loss": 0.0663,
      "step": 1049
    },
    {
      "epoch": 0.5043227665706052,
      "grad_norm": 0.4096266159270642,
      "learning_rate": 9.859606340945904e-06,
      "loss": 0.0339,
      "step": 1050
    },
    {
      "epoch": 0.5048030739673391,
      "grad_norm": 1.091503069790824,
      "learning_rate": 9.858948015261988e-06,
      "loss": 0.0628,
      "step": 1051
    },
    {
      "epoch": 0.505283381364073,
      "grad_norm": 0.6683717006752189,
      "learning_rate": 9.858288171777137e-06,
      "loss": 0.0588,
      "step": 1052
    },
    {
      "epoch": 0.5057636887608069,
      "grad_norm": 0.6905695109286866,
      "learning_rate": 9.857626810697468e-06,
      "loss": 0.0633,
      "step": 1053
    },
    {
      "epoch": 0.5062439961575408,
      "grad_norm": 0.5075936442073656,
      "learning_rate": 9.85696393222957e-06,
      "loss": 0.0475,
      "step": 1054
    },
    {
      "epoch": 0.5067243035542748,
      "grad_norm": 0.59460142658833,
      "learning_rate": 9.856299536580511e-06,
      "loss": 0.0494,
      "step": 1055
    },
    {
      "epoch": 0.5072046109510087,
      "grad_norm": 1.6512164384609662,
      "learning_rate": 9.855633623957828e-06,
      "loss": 0.0727,
      "step": 1056
    },
    {
      "epoch": 0.5076849183477425,
      "grad_norm": 0.5580012014124549,
      "learning_rate": 9.854966194569533e-06,
      "loss": 0.0441,
      "step": 1057
    },
    {
      "epoch": 0.5081652257444764,
      "grad_norm": 0.5763621029835471,
      "learning_rate": 9.854297248624113e-06,
      "loss": 0.0477,
      "step": 1058
    },
    {
      "epoch": 0.5086455331412104,
      "grad_norm": 0.8992511423911591,
      "learning_rate": 9.853626786330529e-06,
      "loss": 0.0812,
      "step": 1059
    },
    {
      "epoch": 0.5091258405379443,
      "grad_norm": 0.5368922969181873,
      "learning_rate": 9.852954807898212e-06,
      "loss": 0.0467,
      "step": 1060
    },
    {
      "epoch": 0.5096061479346782,
      "grad_norm": 0.44033960156979046,
      "learning_rate": 9.852281313537074e-06,
      "loss": 0.0426,
      "step": 1061
    },
    {
      "epoch": 0.5100864553314121,
      "grad_norm": 0.46569533955278264,
      "learning_rate": 9.851606303457492e-06,
      "loss": 0.0423,
      "step": 1062
    },
    {
      "epoch": 0.510566762728146,
      "grad_norm": 0.6709478469641016,
      "learning_rate": 9.850929777870324e-06,
      "loss": 0.0635,
      "step": 1063
    },
    {
      "epoch": 0.5110470701248799,
      "grad_norm": 0.7589418223076211,
      "learning_rate": 9.850251736986895e-06,
      "loss": 0.0485,
      "step": 1064
    },
    {
      "epoch": 0.5115273775216138,
      "grad_norm": 0.5039707034476073,
      "learning_rate": 9.849572181019008e-06,
      "loss": 0.0396,
      "step": 1065
    },
    {
      "epoch": 0.5120076849183477,
      "grad_norm": 0.665169242046722,
      "learning_rate": 9.848891110178936e-06,
      "loss": 0.056,
      "step": 1066
    },
    {
      "epoch": 0.5124879923150817,
      "grad_norm": 0.7006135018748972,
      "learning_rate": 9.848208524679426e-06,
      "loss": 0.0545,
      "step": 1067
    },
    {
      "epoch": 0.5129682997118156,
      "grad_norm": 1.371433470057797,
      "learning_rate": 9.847524424733701e-06,
      "loss": 0.0592,
      "step": 1068
    },
    {
      "epoch": 0.5134486071085494,
      "grad_norm": 0.5371431774356002,
      "learning_rate": 9.846838810555454e-06,
      "loss": 0.0481,
      "step": 1069
    },
    {
      "epoch": 0.5139289145052833,
      "grad_norm": 0.9546711896225505,
      "learning_rate": 9.846151682358853e-06,
      "loss": 0.0605,
      "step": 1070
    },
    {
      "epoch": 0.5144092219020173,
      "grad_norm": 0.4469628087936052,
      "learning_rate": 9.845463040358538e-06,
      "loss": 0.0461,
      "step": 1071
    },
    {
      "epoch": 0.5148895292987512,
      "grad_norm": 0.7337259088697539,
      "learning_rate": 9.84477288476962e-06,
      "loss": 0.0344,
      "step": 1072
    },
    {
      "epoch": 0.5153698366954851,
      "grad_norm": 0.6176921441355193,
      "learning_rate": 9.844081215807684e-06,
      "loss": 0.0554,
      "step": 1073
    },
    {
      "epoch": 0.515850144092219,
      "grad_norm": 0.7254549945035398,
      "learning_rate": 9.843388033688789e-06,
      "loss": 0.0406,
      "step": 1074
    },
    {
      "epoch": 0.516330451488953,
      "grad_norm": 0.6081358712579048,
      "learning_rate": 9.842693338629468e-06,
      "loss": 0.0607,
      "step": 1075
    },
    {
      "epoch": 0.5168107588856868,
      "grad_norm": 0.5697699264079668,
      "learning_rate": 9.84199713084672e-06,
      "loss": 0.0693,
      "step": 1076
    },
    {
      "epoch": 0.5172910662824207,
      "grad_norm": 0.3687520833244058,
      "learning_rate": 9.841299410558026e-06,
      "loss": 0.035,
      "step": 1077
    },
    {
      "epoch": 0.5177713736791547,
      "grad_norm": 0.6188868823780803,
      "learning_rate": 9.840600177981331e-06,
      "loss": 0.0378,
      "step": 1078
    },
    {
      "epoch": 0.5182516810758886,
      "grad_norm": 0.44919065810406017,
      "learning_rate": 9.839899433335059e-06,
      "loss": 0.0569,
      "step": 1079
    },
    {
      "epoch": 0.5187319884726225,
      "grad_norm": 0.525069381664302,
      "learning_rate": 9.839197176838102e-06,
      "loss": 0.0534,
      "step": 1080
    },
    {
      "epoch": 0.5192122958693564,
      "grad_norm": 0.5212151097899781,
      "learning_rate": 9.838493408709823e-06,
      "loss": 0.051,
      "step": 1081
    },
    {
      "epoch": 0.5196926032660903,
      "grad_norm": 0.6605149345379034,
      "learning_rate": 9.837788129170063e-06,
      "loss": 0.076,
      "step": 1082
    },
    {
      "epoch": 0.5201729106628242,
      "grad_norm": 0.7131108392835402,
      "learning_rate": 9.83708133843913e-06,
      "loss": 0.0723,
      "step": 1083
    },
    {
      "epoch": 0.5206532180595581,
      "grad_norm": 0.5238463836985972,
      "learning_rate": 9.836373036737805e-06,
      "loss": 0.0476,
      "step": 1084
    },
    {
      "epoch": 0.521133525456292,
      "grad_norm": 0.4394705568289204,
      "learning_rate": 9.835663224287343e-06,
      "loss": 0.052,
      "step": 1085
    },
    {
      "epoch": 0.521613832853026,
      "grad_norm": 0.6627060321697689,
      "learning_rate": 9.834951901309473e-06,
      "loss": 0.0514,
      "step": 1086
    },
    {
      "epoch": 0.5220941402497599,
      "grad_norm": 0.6047813561651657,
      "learning_rate": 9.834239068026388e-06,
      "loss": 0.0662,
      "step": 1087
    },
    {
      "epoch": 0.5225744476464937,
      "grad_norm": 0.608811144599817,
      "learning_rate": 9.83352472466076e-06,
      "loss": 0.046,
      "step": 1088
    },
    {
      "epoch": 0.5230547550432276,
      "grad_norm": 1.158801190471405,
      "learning_rate": 9.832808871435728e-06,
      "loss": 0.0524,
      "step": 1089
    },
    {
      "epoch": 0.5235350624399616,
      "grad_norm": 0.46635581913051444,
      "learning_rate": 9.832091508574906e-06,
      "loss": 0.0449,
      "step": 1090
    },
    {
      "epoch": 0.5240153698366955,
      "grad_norm": 0.48367373014821347,
      "learning_rate": 9.831372636302379e-06,
      "loss": 0.0458,
      "step": 1091
    },
    {
      "epoch": 0.5244956772334294,
      "grad_norm": 0.361218140616094,
      "learning_rate": 9.8306522548427e-06,
      "loss": 0.04,
      "step": 1092
    },
    {
      "epoch": 0.5249759846301633,
      "grad_norm": 0.655690358593338,
      "learning_rate": 9.829930364420902e-06,
      "loss": 0.0554,
      "step": 1093
    },
    {
      "epoch": 0.5254562920268973,
      "grad_norm": 0.7505215320060055,
      "learning_rate": 9.829206965262477e-06,
      "loss": 0.0743,
      "step": 1094
    },
    {
      "epoch": 0.5259365994236311,
      "grad_norm": 0.5750818638699632,
      "learning_rate": 9.828482057593397e-06,
      "loss": 0.0576,
      "step": 1095
    },
    {
      "epoch": 0.526416906820365,
      "grad_norm": 0.47388206139707323,
      "learning_rate": 9.827755641640105e-06,
      "loss": 0.0379,
      "step": 1096
    },
    {
      "epoch": 0.5268972142170989,
      "grad_norm": 0.5077311880425335,
      "learning_rate": 9.827027717629511e-06,
      "loss": 0.0623,
      "step": 1097
    },
    {
      "epoch": 0.5273775216138329,
      "grad_norm": 0.45204043027410235,
      "learning_rate": 9.826298285789002e-06,
      "loss": 0.0398,
      "step": 1098
    },
    {
      "epoch": 0.5278578290105668,
      "grad_norm": 0.7965027074583484,
      "learning_rate": 9.825567346346427e-06,
      "loss": 0.0678,
      "step": 1099
    },
    {
      "epoch": 0.5283381364073007,
      "grad_norm": 0.6917329847032074,
      "learning_rate": 9.824834899530116e-06,
      "loss": 0.0619,
      "step": 1100
    },
    {
      "epoch": 0.5288184438040345,
      "grad_norm": 0.5524043158252445,
      "learning_rate": 9.824100945568862e-06,
      "loss": 0.0465,
      "step": 1101
    },
    {
      "epoch": 0.5292987512007685,
      "grad_norm": 0.5638562567517995,
      "learning_rate": 9.823365484691933e-06,
      "loss": 0.0466,
      "step": 1102
    },
    {
      "epoch": 0.5297790585975024,
      "grad_norm": 0.8316442556169921,
      "learning_rate": 9.822628517129067e-06,
      "loss": 0.0881,
      "step": 1103
    },
    {
      "epoch": 0.5302593659942363,
      "grad_norm": 0.5548828049736547,
      "learning_rate": 9.821890043110471e-06,
      "loss": 0.0518,
      "step": 1104
    },
    {
      "epoch": 0.5307396733909702,
      "grad_norm": 0.41550027200381046,
      "learning_rate": 9.821150062866826e-06,
      "loss": 0.0408,
      "step": 1105
    },
    {
      "epoch": 0.5312199807877042,
      "grad_norm": 0.5895290692704226,
      "learning_rate": 9.82040857662928e-06,
      "loss": 0.051,
      "step": 1106
    },
    {
      "epoch": 0.531700288184438,
      "grad_norm": 0.5248740479596763,
      "learning_rate": 9.819665584629453e-06,
      "loss": 0.0561,
      "step": 1107
    },
    {
      "epoch": 0.5321805955811719,
      "grad_norm": 0.6505572311350228,
      "learning_rate": 9.818921087099435e-06,
      "loss": 0.059,
      "step": 1108
    },
    {
      "epoch": 0.5326609029779059,
      "grad_norm": 0.6556567838840428,
      "learning_rate": 9.818175084271786e-06,
      "loss": 0.0559,
      "step": 1109
    },
    {
      "epoch": 0.5331412103746398,
      "grad_norm": 0.5382916731099718,
      "learning_rate": 9.817427576379536e-06,
      "loss": 0.0485,
      "step": 1110
    },
    {
      "epoch": 0.5336215177713737,
      "grad_norm": 0.3758870057448877,
      "learning_rate": 9.81667856365619e-06,
      "loss": 0.0496,
      "step": 1111
    },
    {
      "epoch": 0.5341018251681076,
      "grad_norm": 0.6578831277841793,
      "learning_rate": 9.815928046335713e-06,
      "loss": 0.0557,
      "step": 1112
    },
    {
      "epoch": 0.5345821325648416,
      "grad_norm": 1.060321357406775,
      "learning_rate": 9.81517602465255e-06,
      "loss": 0.0532,
      "step": 1113
    },
    {
      "epoch": 0.5350624399615754,
      "grad_norm": 0.5103225144908645,
      "learning_rate": 9.81442249884161e-06,
      "loss": 0.035,
      "step": 1114
    },
    {
      "epoch": 0.5355427473583093,
      "grad_norm": 0.37825548058600067,
      "learning_rate": 9.813667469138273e-06,
      "loss": 0.0429,
      "step": 1115
    },
    {
      "epoch": 0.5360230547550432,
      "grad_norm": 0.805237319756677,
      "learning_rate": 9.812910935778393e-06,
      "loss": 0.068,
      "step": 1116
    },
    {
      "epoch": 0.5365033621517772,
      "grad_norm": 0.42363680651332974,
      "learning_rate": 9.812152898998286e-06,
      "loss": 0.0469,
      "step": 1117
    },
    {
      "epoch": 0.5369836695485111,
      "grad_norm": 0.6106525825949898,
      "learning_rate": 9.811393359034742e-06,
      "loss": 0.0542,
      "step": 1118
    },
    {
      "epoch": 0.537463976945245,
      "grad_norm": 0.8001537869444254,
      "learning_rate": 9.810632316125023e-06,
      "loss": 0.0565,
      "step": 1119
    },
    {
      "epoch": 0.5379442843419788,
      "grad_norm": 0.7262847802024673,
      "learning_rate": 9.809869770506855e-06,
      "loss": 0.0695,
      "step": 1120
    },
    {
      "epoch": 0.5384245917387128,
      "grad_norm": 1.1087961916090308,
      "learning_rate": 9.80910572241844e-06,
      "loss": 0.0657,
      "step": 1121
    },
    {
      "epoch": 0.5389048991354467,
      "grad_norm": 0.41628123944641576,
      "learning_rate": 9.80834017209844e-06,
      "loss": 0.0463,
      "step": 1122
    },
    {
      "epoch": 0.5393852065321806,
      "grad_norm": 0.5808409195734693,
      "learning_rate": 9.807573119785995e-06,
      "loss": 0.0606,
      "step": 1123
    },
    {
      "epoch": 0.5398655139289145,
      "grad_norm": 0.49580631030323374,
      "learning_rate": 9.806804565720712e-06,
      "loss": 0.0302,
      "step": 1124
    },
    {
      "epoch": 0.5403458213256485,
      "grad_norm": 1.0814853755104163,
      "learning_rate": 9.806034510142664e-06,
      "loss": 0.0675,
      "step": 1125
    },
    {
      "epoch": 0.5408261287223823,
      "grad_norm": 0.549003216277453,
      "learning_rate": 9.805262953292395e-06,
      "loss": 0.0533,
      "step": 1126
    },
    {
      "epoch": 0.5413064361191162,
      "grad_norm": 0.5609756437216896,
      "learning_rate": 9.80448989541092e-06,
      "loss": 0.057,
      "step": 1127
    },
    {
      "epoch": 0.5417867435158501,
      "grad_norm": 0.9338428086220139,
      "learning_rate": 9.803715336739717e-06,
      "loss": 0.0834,
      "step": 1128
    },
    {
      "epoch": 0.5422670509125841,
      "grad_norm": 0.5122018899106414,
      "learning_rate": 9.802939277520742e-06,
      "loss": 0.0484,
      "step": 1129
    },
    {
      "epoch": 0.542747358309318,
      "grad_norm": 0.5809571628506907,
      "learning_rate": 9.80216171799641e-06,
      "loss": 0.0446,
      "step": 1130
    },
    {
      "epoch": 0.5432276657060519,
      "grad_norm": 0.5453610787253219,
      "learning_rate": 9.801382658409611e-06,
      "loss": 0.0605,
      "step": 1131
    },
    {
      "epoch": 0.5437079731027857,
      "grad_norm": 0.45815751598603865,
      "learning_rate": 9.800602099003702e-06,
      "loss": 0.0441,
      "step": 1132
    },
    {
      "epoch": 0.5441882804995197,
      "grad_norm": 0.618974171564184,
      "learning_rate": 9.799820040022507e-06,
      "loss": 0.0565,
      "step": 1133
    },
    {
      "epoch": 0.5446685878962536,
      "grad_norm": 0.577552914326065,
      "learning_rate": 9.79903648171032e-06,
      "loss": 0.0625,
      "step": 1134
    },
    {
      "epoch": 0.5451488952929875,
      "grad_norm": 0.674317185574733,
      "learning_rate": 9.798251424311904e-06,
      "loss": 0.0601,
      "step": 1135
    },
    {
      "epoch": 0.5456292026897214,
      "grad_norm": 0.5368692747640629,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.0425,
      "step": 1136
    },
    {
      "epoch": 0.5461095100864554,
      "grad_norm": 0.8059063068692716,
      "learning_rate": 9.79667681323777e-06,
      "loss": 0.0607,
      "step": 1137
    },
    {
      "epoch": 0.5465898174831892,
      "grad_norm": 0.44591141730781386,
      "learning_rate": 9.795887260053918e-06,
      "loss": 0.0458,
      "step": 1138
    },
    {
      "epoch": 0.5470701248799231,
      "grad_norm": 0.49579824081441304,
      "learning_rate": 9.795096208767565e-06,
      "loss": 0.0417,
      "step": 1139
    },
    {
      "epoch": 0.547550432276657,
      "grad_norm": 0.9974120235484584,
      "learning_rate": 9.794303659625815e-06,
      "loss": 0.067,
      "step": 1140
    },
    {
      "epoch": 0.548030739673391,
      "grad_norm": 0.6293616673344423,
      "learning_rate": 9.793509612876237e-06,
      "loss": 0.0574,
      "step": 1141
    },
    {
      "epoch": 0.5485110470701249,
      "grad_norm": 0.5069087987447661,
      "learning_rate": 9.792714068766872e-06,
      "loss": 0.0507,
      "step": 1142
    },
    {
      "epoch": 0.5489913544668588,
      "grad_norm": 0.44270498114612894,
      "learning_rate": 9.791917027546223e-06,
      "loss": 0.0511,
      "step": 1143
    },
    {
      "epoch": 0.5494716618635928,
      "grad_norm": 0.5614291657476101,
      "learning_rate": 9.791118489463265e-06,
      "loss": 0.0482,
      "step": 1144
    },
    {
      "epoch": 0.5499519692603266,
      "grad_norm": 0.6022614849565968,
      "learning_rate": 9.790318454767438e-06,
      "loss": 0.0433,
      "step": 1145
    },
    {
      "epoch": 0.5504322766570605,
      "grad_norm": 0.4960237823747353,
      "learning_rate": 9.78951692370865e-06,
      "loss": 0.0618,
      "step": 1146
    },
    {
      "epoch": 0.5509125840537944,
      "grad_norm": 0.7042914518317633,
      "learning_rate": 9.78871389653728e-06,
      "loss": 0.058,
      "step": 1147
    },
    {
      "epoch": 0.5513928914505284,
      "grad_norm": 0.6312566271413287,
      "learning_rate": 9.787909373504172e-06,
      "loss": 0.0429,
      "step": 1148
    },
    {
      "epoch": 0.5518731988472623,
      "grad_norm": 0.39432463738714213,
      "learning_rate": 9.787103354860633e-06,
      "loss": 0.0346,
      "step": 1149
    },
    {
      "epoch": 0.5523535062439962,
      "grad_norm": 0.9311799513518504,
      "learning_rate": 9.786295840858444e-06,
      "loss": 0.0658,
      "step": 1150
    },
    {
      "epoch": 0.55283381364073,
      "grad_norm": 0.5007348827746324,
      "learning_rate": 9.785486831749847e-06,
      "loss": 0.0452,
      "step": 1151
    },
    {
      "epoch": 0.553314121037464,
      "grad_norm": 0.49962024985272985,
      "learning_rate": 9.784676327787557e-06,
      "loss": 0.0568,
      "step": 1152
    },
    {
      "epoch": 0.5537944284341979,
      "grad_norm": 0.3691640957902528,
      "learning_rate": 9.783864329224752e-06,
      "loss": 0.0451,
      "step": 1153
    },
    {
      "epoch": 0.5542747358309318,
      "grad_norm": 0.413567800244177,
      "learning_rate": 9.783050836315078e-06,
      "loss": 0.0472,
      "step": 1154
    },
    {
      "epoch": 0.5547550432276657,
      "grad_norm": 0.39518046681000146,
      "learning_rate": 9.782235849312647e-06,
      "loss": 0.041,
      "step": 1155
    },
    {
      "epoch": 0.5552353506243997,
      "grad_norm": 0.49523548669900735,
      "learning_rate": 9.781419368472039e-06,
      "loss": 0.0441,
      "step": 1156
    },
    {
      "epoch": 0.5557156580211335,
      "grad_norm": 0.6024044401609281,
      "learning_rate": 9.7806013940483e-06,
      "loss": 0.0536,
      "step": 1157
    },
    {
      "epoch": 0.5561959654178674,
      "grad_norm": 1.2088324636184875,
      "learning_rate": 9.779781926296942e-06,
      "loss": 0.0708,
      "step": 1158
    },
    {
      "epoch": 0.5566762728146013,
      "grad_norm": 0.6166764403933446,
      "learning_rate": 9.778960965473945e-06,
      "loss": 0.0482,
      "step": 1159
    },
    {
      "epoch": 0.5571565802113353,
      "grad_norm": 0.9288628974217886,
      "learning_rate": 9.778138511835753e-06,
      "loss": 0.0672,
      "step": 1160
    },
    {
      "epoch": 0.5576368876080692,
      "grad_norm": 1.2330701366219725,
      "learning_rate": 9.77731456563928e-06,
      "loss": 0.0558,
      "step": 1161
    },
    {
      "epoch": 0.5581171950048031,
      "grad_norm": 0.4791648541731939,
      "learning_rate": 9.776489127141902e-06,
      "loss": 0.043,
      "step": 1162
    },
    {
      "epoch": 0.5585975024015369,
      "grad_norm": 0.5709504036661929,
      "learning_rate": 9.775662196601464e-06,
      "loss": 0.0481,
      "step": 1163
    },
    {
      "epoch": 0.5590778097982709,
      "grad_norm": 0.6468021810734045,
      "learning_rate": 9.774833774276278e-06,
      "loss": 0.0541,
      "step": 1164
    },
    {
      "epoch": 0.5595581171950048,
      "grad_norm": 0.46892414876617317,
      "learning_rate": 9.774003860425116e-06,
      "loss": 0.0408,
      "step": 1165
    },
    {
      "epoch": 0.5600384245917387,
      "grad_norm": 0.7110924147202355,
      "learning_rate": 9.773172455307223e-06,
      "loss": 0.0869,
      "step": 1166
    },
    {
      "epoch": 0.5605187319884726,
      "grad_norm": 0.5925406633330682,
      "learning_rate": 9.772339559182307e-06,
      "loss": 0.038,
      "step": 1167
    },
    {
      "epoch": 0.5609990393852066,
      "grad_norm": 0.4344853797907911,
      "learning_rate": 9.77150517231054e-06,
      "loss": 0.0448,
      "step": 1168
    },
    {
      "epoch": 0.5614793467819404,
      "grad_norm": 1.5378334777798277,
      "learning_rate": 9.770669294952562e-06,
      "loss": 0.0455,
      "step": 1169
    },
    {
      "epoch": 0.5619596541786743,
      "grad_norm": 0.661380367146787,
      "learning_rate": 9.76983192736948e-06,
      "loss": 0.0646,
      "step": 1170
    },
    {
      "epoch": 0.5624399615754082,
      "grad_norm": 0.6292936111066385,
      "learning_rate": 9.768993069822862e-06,
      "loss": 0.0543,
      "step": 1171
    },
    {
      "epoch": 0.5629202689721422,
      "grad_norm": 0.5168700137660031,
      "learning_rate": 9.768152722574747e-06,
      "loss": 0.0482,
      "step": 1172
    },
    {
      "epoch": 0.5634005763688761,
      "grad_norm": 0.48793915527548615,
      "learning_rate": 9.76731088588763e-06,
      "loss": 0.052,
      "step": 1173
    },
    {
      "epoch": 0.56388088376561,
      "grad_norm": 0.8671102690781609,
      "learning_rate": 9.766467560024485e-06,
      "loss": 0.07,
      "step": 1174
    },
    {
      "epoch": 0.5643611911623438,
      "grad_norm": 0.5202999857616358,
      "learning_rate": 9.765622745248739e-06,
      "loss": 0.0541,
      "step": 1175
    },
    {
      "epoch": 0.5648414985590778,
      "grad_norm": 1.2213889314313011,
      "learning_rate": 9.76477644182429e-06,
      "loss": 0.045,
      "step": 1176
    },
    {
      "epoch": 0.5653218059558117,
      "grad_norm": 0.5437848268666584,
      "learning_rate": 9.7639286500155e-06,
      "loss": 0.0434,
      "step": 1177
    },
    {
      "epoch": 0.5658021133525456,
      "grad_norm": 0.46403176390550605,
      "learning_rate": 9.763079370087196e-06,
      "loss": 0.0352,
      "step": 1178
    },
    {
      "epoch": 0.5662824207492796,
      "grad_norm": 0.7171751460163477,
      "learning_rate": 9.762228602304667e-06,
      "loss": 0.0578,
      "step": 1179
    },
    {
      "epoch": 0.5667627281460135,
      "grad_norm": 0.48339364673648727,
      "learning_rate": 9.761376346933672e-06,
      "loss": 0.0464,
      "step": 1180
    },
    {
      "epoch": 0.5672430355427474,
      "grad_norm": 0.4414131140482316,
      "learning_rate": 9.760522604240434e-06,
      "loss": 0.0356,
      "step": 1181
    },
    {
      "epoch": 0.5677233429394812,
      "grad_norm": 0.995263931495515,
      "learning_rate": 9.759667374491632e-06,
      "loss": 0.064,
      "step": 1182
    },
    {
      "epoch": 0.5682036503362152,
      "grad_norm": 0.6310705676300887,
      "learning_rate": 9.758810657954424e-06,
      "loss": 0.0643,
      "step": 1183
    },
    {
      "epoch": 0.5686839577329491,
      "grad_norm": 0.8865712295477853,
      "learning_rate": 9.757952454896418e-06,
      "loss": 0.0707,
      "step": 1184
    },
    {
      "epoch": 0.569164265129683,
      "grad_norm": 0.44330618752194106,
      "learning_rate": 9.757092765585695e-06,
      "loss": 0.0427,
      "step": 1185
    },
    {
      "epoch": 0.5696445725264169,
      "grad_norm": 0.5101911660662402,
      "learning_rate": 9.7562315902908e-06,
      "loss": 0.0612,
      "step": 1186
    },
    {
      "epoch": 0.5701248799231509,
      "grad_norm": 0.3634609856460933,
      "learning_rate": 9.755368929280738e-06,
      "loss": 0.0359,
      "step": 1187
    },
    {
      "epoch": 0.5706051873198847,
      "grad_norm": 0.8929703488376083,
      "learning_rate": 9.754504782824982e-06,
      "loss": 0.0608,
      "step": 1188
    },
    {
      "epoch": 0.5710854947166186,
      "grad_norm": 0.7621720458573135,
      "learning_rate": 9.753639151193468e-06,
      "loss": 0.0747,
      "step": 1189
    },
    {
      "epoch": 0.5715658021133525,
      "grad_norm": 0.4562290573979324,
      "learning_rate": 9.752772034656593e-06,
      "loss": 0.0453,
      "step": 1190
    },
    {
      "epoch": 0.5720461095100865,
      "grad_norm": 0.454374886984673,
      "learning_rate": 9.75190343348522e-06,
      "loss": 0.046,
      "step": 1191
    },
    {
      "epoch": 0.5725264169068204,
      "grad_norm": 0.4985067047730674,
      "learning_rate": 9.75103334795068e-06,
      "loss": 0.0465,
      "step": 1192
    },
    {
      "epoch": 0.5730067243035543,
      "grad_norm": 0.62518531188208,
      "learning_rate": 9.750161778324759e-06,
      "loss": 0.0546,
      "step": 1193
    },
    {
      "epoch": 0.5734870317002881,
      "grad_norm": 0.5218968728591356,
      "learning_rate": 9.749288724879716e-06,
      "loss": 0.0451,
      "step": 1194
    },
    {
      "epoch": 0.5739673390970221,
      "grad_norm": 0.520699846944625,
      "learning_rate": 9.748414187888262e-06,
      "loss": 0.0511,
      "step": 1195
    },
    {
      "epoch": 0.574447646493756,
      "grad_norm": 0.7212870071975475,
      "learning_rate": 9.747538167623585e-06,
      "loss": 0.0539,
      "step": 1196
    },
    {
      "epoch": 0.5749279538904899,
      "grad_norm": 0.41354709356039293,
      "learning_rate": 9.746660664359326e-06,
      "loss": 0.0348,
      "step": 1197
    },
    {
      "epoch": 0.5754082612872238,
      "grad_norm": 0.47323694276958017,
      "learning_rate": 9.745781678369594e-06,
      "loss": 0.0342,
      "step": 1198
    },
    {
      "epoch": 0.5758885686839578,
      "grad_norm": 0.7798319365982702,
      "learning_rate": 9.744901209928959e-06,
      "loss": 0.0386,
      "step": 1199
    },
    {
      "epoch": 0.5763688760806917,
      "grad_norm": 0.41388948108418533,
      "learning_rate": 9.744019259312454e-06,
      "loss": 0.0404,
      "step": 1200
    },
    {
      "epoch": 0.5768491834774255,
      "grad_norm": 0.45590776454332793,
      "learning_rate": 9.74313582679558e-06,
      "loss": 0.059,
      "step": 1201
    },
    {
      "epoch": 0.5773294908741594,
      "grad_norm": 0.8624611717777299,
      "learning_rate": 9.742250912654292e-06,
      "loss": 0.0573,
      "step": 1202
    },
    {
      "epoch": 0.5778097982708934,
      "grad_norm": 0.4105792475406539,
      "learning_rate": 9.741364517165017e-06,
      "loss": 0.0474,
      "step": 1203
    },
    {
      "epoch": 0.5782901056676273,
      "grad_norm": 0.4030691281347931,
      "learning_rate": 9.740476640604637e-06,
      "loss": 0.0533,
      "step": 1204
    },
    {
      "epoch": 0.5787704130643612,
      "grad_norm": 0.8208559639247031,
      "learning_rate": 9.7395872832505e-06,
      "loss": 0.0491,
      "step": 1205
    },
    {
      "epoch": 0.579250720461095,
      "grad_norm": 0.6089308840844816,
      "learning_rate": 9.73869644538042e-06,
      "loss": 0.069,
      "step": 1206
    },
    {
      "epoch": 0.579731027857829,
      "grad_norm": 0.6462057166604767,
      "learning_rate": 9.737804127272668e-06,
      "loss": 0.051,
      "step": 1207
    },
    {
      "epoch": 0.5802113352545629,
      "grad_norm": 0.4269852827384148,
      "learning_rate": 9.73691032920598e-06,
      "loss": 0.0434,
      "step": 1208
    },
    {
      "epoch": 0.5806916426512968,
      "grad_norm": 0.4245254269716441,
      "learning_rate": 9.736015051459551e-06,
      "loss": 0.0372,
      "step": 1209
    },
    {
      "epoch": 0.5811719500480308,
      "grad_norm": 0.3532506765163264,
      "learning_rate": 9.735118294313045e-06,
      "loss": 0.0466,
      "step": 1210
    },
    {
      "epoch": 0.5816522574447647,
      "grad_norm": 0.34262000331545467,
      "learning_rate": 9.734220058046582e-06,
      "loss": 0.0356,
      "step": 1211
    },
    {
      "epoch": 0.5821325648414986,
      "grad_norm": 0.5754600784602188,
      "learning_rate": 9.733320342940747e-06,
      "loss": 0.0543,
      "step": 1212
    },
    {
      "epoch": 0.5826128722382324,
      "grad_norm": 0.5170774377794336,
      "learning_rate": 9.732419149276586e-06,
      "loss": 0.0558,
      "step": 1213
    },
    {
      "epoch": 0.5830931796349664,
      "grad_norm": 0.5109210528826162,
      "learning_rate": 9.731516477335607e-06,
      "loss": 0.062,
      "step": 1214
    },
    {
      "epoch": 0.5835734870317003,
      "grad_norm": 0.4985819768058242,
      "learning_rate": 9.73061232739978e-06,
      "loss": 0.0529,
      "step": 1215
    },
    {
      "epoch": 0.5840537944284342,
      "grad_norm": 0.5498108762960605,
      "learning_rate": 9.729706699751535e-06,
      "loss": 0.0517,
      "step": 1216
    },
    {
      "epoch": 0.5845341018251681,
      "grad_norm": 0.7920509732214022,
      "learning_rate": 9.728799594673766e-06,
      "loss": 0.0478,
      "step": 1217
    },
    {
      "epoch": 0.5850144092219021,
      "grad_norm": 0.5652672810126951,
      "learning_rate": 9.727891012449827e-06,
      "loss": 0.0525,
      "step": 1218
    },
    {
      "epoch": 0.585494716618636,
      "grad_norm": 0.5691398549888063,
      "learning_rate": 9.726980953363536e-06,
      "loss": 0.0496,
      "step": 1219
    },
    {
      "epoch": 0.5859750240153698,
      "grad_norm": 0.5253745870416723,
      "learning_rate": 9.726069417699167e-06,
      "loss": 0.0483,
      "step": 1220
    },
    {
      "epoch": 0.5864553314121037,
      "grad_norm": 0.6773929868480058,
      "learning_rate": 9.725156405741461e-06,
      "loss": 0.0455,
      "step": 1221
    },
    {
      "epoch": 0.5869356388088377,
      "grad_norm": 0.8365763274612497,
      "learning_rate": 9.724241917775616e-06,
      "loss": 0.0619,
      "step": 1222
    },
    {
      "epoch": 0.5874159462055716,
      "grad_norm": 0.6630891954815319,
      "learning_rate": 9.723325954087294e-06,
      "loss": 0.064,
      "step": 1223
    },
    {
      "epoch": 0.5878962536023055,
      "grad_norm": 0.34729186105181487,
      "learning_rate": 9.722408514962619e-06,
      "loss": 0.0353,
      "step": 1224
    },
    {
      "epoch": 0.5883765609990393,
      "grad_norm": 0.44671842700761955,
      "learning_rate": 9.721489600688168e-06,
      "loss": 0.0442,
      "step": 1225
    },
    {
      "epoch": 0.5888568683957733,
      "grad_norm": 0.44380993259781293,
      "learning_rate": 9.720569211550988e-06,
      "loss": 0.0401,
      "step": 1226
    },
    {
      "epoch": 0.5893371757925072,
      "grad_norm": 0.7426688381698872,
      "learning_rate": 9.719647347838584e-06,
      "loss": 0.0536,
      "step": 1227
    },
    {
      "epoch": 0.5898174831892411,
      "grad_norm": 0.6909672290678518,
      "learning_rate": 9.718724009838917e-06,
      "loss": 0.0488,
      "step": 1228
    },
    {
      "epoch": 0.590297790585975,
      "grad_norm": 0.5801124250465916,
      "learning_rate": 9.717799197840416e-06,
      "loss": 0.0559,
      "step": 1229
    },
    {
      "epoch": 0.590778097982709,
      "grad_norm": 0.38585748356811145,
      "learning_rate": 9.716872912131964e-06,
      "loss": 0.0387,
      "step": 1230
    },
    {
      "epoch": 0.5912584053794429,
      "grad_norm": 0.5037471388123007,
      "learning_rate": 9.715945153002908e-06,
      "loss": 0.0478,
      "step": 1231
    },
    {
      "epoch": 0.5917387127761767,
      "grad_norm": 0.42301309786103813,
      "learning_rate": 9.715015920743056e-06,
      "loss": 0.0441,
      "step": 1232
    },
    {
      "epoch": 0.5922190201729106,
      "grad_norm": 0.5895965832076508,
      "learning_rate": 9.714085215642672e-06,
      "loss": 0.0442,
      "step": 1233
    },
    {
      "epoch": 0.5926993275696446,
      "grad_norm": 0.6111729772870694,
      "learning_rate": 9.713153037992484e-06,
      "loss": 0.0494,
      "step": 1234
    },
    {
      "epoch": 0.5931796349663785,
      "grad_norm": 0.6116675844026525,
      "learning_rate": 9.712219388083676e-06,
      "loss": 0.0578,
      "step": 1235
    },
    {
      "epoch": 0.5936599423631124,
      "grad_norm": 0.46491825617527227,
      "learning_rate": 9.711284266207899e-06,
      "loss": 0.0564,
      "step": 1236
    },
    {
      "epoch": 0.5941402497598463,
      "grad_norm": 0.952450984803784,
      "learning_rate": 9.710347672657254e-06,
      "loss": 0.046,
      "step": 1237
    },
    {
      "epoch": 0.5946205571565802,
      "grad_norm": 0.6313323004545888,
      "learning_rate": 9.70940960772431e-06,
      "loss": 0.0452,
      "step": 1238
    },
    {
      "epoch": 0.5951008645533141,
      "grad_norm": 0.5509216833295507,
      "learning_rate": 9.708470071702094e-06,
      "loss": 0.0517,
      "step": 1239
    },
    {
      "epoch": 0.595581171950048,
      "grad_norm": 1.043696226103975,
      "learning_rate": 9.707529064884087e-06,
      "loss": 0.0586,
      "step": 1240
    },
    {
      "epoch": 0.5960614793467819,
      "grad_norm": 0.5486854474154073,
      "learning_rate": 9.706586587564236e-06,
      "loss": 0.057,
      "step": 1241
    },
    {
      "epoch": 0.5965417867435159,
      "grad_norm": 0.633733102374295,
      "learning_rate": 9.705642640036945e-06,
      "loss": 0.0548,
      "step": 1242
    },
    {
      "epoch": 0.5970220941402498,
      "grad_norm": 0.4966097609578802,
      "learning_rate": 9.704697222597074e-06,
      "loss": 0.0428,
      "step": 1243
    },
    {
      "epoch": 0.5975024015369836,
      "grad_norm": 0.5824655077330811,
      "learning_rate": 9.703750335539952e-06,
      "loss": 0.0565,
      "step": 1244
    },
    {
      "epoch": 0.5979827089337176,
      "grad_norm": 0.563546323378018,
      "learning_rate": 9.702801979161353e-06,
      "loss": 0.0543,
      "step": 1245
    },
    {
      "epoch": 0.5984630163304515,
      "grad_norm": 0.5563049946399633,
      "learning_rate": 9.70185215375752e-06,
      "loss": 0.049,
      "step": 1246
    },
    {
      "epoch": 0.5989433237271854,
      "grad_norm": 0.3744869072214263,
      "learning_rate": 9.700900859625155e-06,
      "loss": 0.032,
      "step": 1247
    },
    {
      "epoch": 0.5994236311239193,
      "grad_norm": 0.5977503365912346,
      "learning_rate": 9.699948097061412e-06,
      "loss": 0.0598,
      "step": 1248
    },
    {
      "epoch": 0.5999039385206533,
      "grad_norm": 0.7030336814456123,
      "learning_rate": 9.69899386636391e-06,
      "loss": 0.0446,
      "step": 1249
    },
    {
      "epoch": 0.6003842459173871,
      "grad_norm": 0.4669080970568829,
      "learning_rate": 9.698038167830722e-06,
      "loss": 0.0515,
      "step": 1250
    },
    {
      "epoch": 0.600864553314121,
      "grad_norm": 0.49830925851066343,
      "learning_rate": 9.697081001760384e-06,
      "loss": 0.0576,
      "step": 1251
    },
    {
      "epoch": 0.6013448607108549,
      "grad_norm": 0.6686926230527453,
      "learning_rate": 9.696122368451887e-06,
      "loss": 0.0399,
      "step": 1252
    },
    {
      "epoch": 0.6018251681075889,
      "grad_norm": 0.6340142308992163,
      "learning_rate": 9.695162268204681e-06,
      "loss": 0.0792,
      "step": 1253
    },
    {
      "epoch": 0.6023054755043228,
      "grad_norm": 0.532128653279977,
      "learning_rate": 9.694200701318679e-06,
      "loss": 0.041,
      "step": 1254
    },
    {
      "epoch": 0.6027857829010567,
      "grad_norm": 0.5065554349402721,
      "learning_rate": 9.693237668094242e-06,
      "loss": 0.0466,
      "step": 1255
    },
    {
      "epoch": 0.6032660902977905,
      "grad_norm": 0.5153614448612733,
      "learning_rate": 9.692273168832198e-06,
      "loss": 0.054,
      "step": 1256
    },
    {
      "epoch": 0.6037463976945245,
      "grad_norm": 0.5353390841759837,
      "learning_rate": 9.69130720383383e-06,
      "loss": 0.0374,
      "step": 1257
    },
    {
      "epoch": 0.6042267050912584,
      "grad_norm": 0.639129347628516,
      "learning_rate": 9.690339773400876e-06,
      "loss": 0.0432,
      "step": 1258
    },
    {
      "epoch": 0.6047070124879923,
      "grad_norm": 0.58928331069562,
      "learning_rate": 9.689370877835538e-06,
      "loss": 0.0357,
      "step": 1259
    },
    {
      "epoch": 0.6051873198847262,
      "grad_norm": 0.5336886529718406,
      "learning_rate": 9.688400517440471e-06,
      "loss": 0.056,
      "step": 1260
    },
    {
      "epoch": 0.6056676272814602,
      "grad_norm": 0.5656089550860497,
      "learning_rate": 9.687428692518789e-06,
      "loss": 0.0539,
      "step": 1261
    },
    {
      "epoch": 0.6061479346781941,
      "grad_norm": 0.4162011114169869,
      "learning_rate": 9.686455403374062e-06,
      "loss": 0.0354,
      "step": 1262
    },
    {
      "epoch": 0.6066282420749279,
      "grad_norm": 0.6906145746227967,
      "learning_rate": 9.685480650310319e-06,
      "loss": 0.0526,
      "step": 1263
    },
    {
      "epoch": 0.6071085494716618,
      "grad_norm": 1.028325898856233,
      "learning_rate": 9.684504433632049e-06,
      "loss": 0.0588,
      "step": 1264
    },
    {
      "epoch": 0.6075888568683958,
      "grad_norm": 0.5167645947728937,
      "learning_rate": 9.68352675364419e-06,
      "loss": 0.0354,
      "step": 1265
    },
    {
      "epoch": 0.6080691642651297,
      "grad_norm": 0.6196435075019047,
      "learning_rate": 9.682547610652145e-06,
      "loss": 0.05,
      "step": 1266
    },
    {
      "epoch": 0.6085494716618636,
      "grad_norm": 0.47414855740871636,
      "learning_rate": 9.681567004961769e-06,
      "loss": 0.0495,
      "step": 1267
    },
    {
      "epoch": 0.6090297790585975,
      "grad_norm": 0.4592587792291843,
      "learning_rate": 9.68058493687938e-06,
      "loss": 0.0386,
      "step": 1268
    },
    {
      "epoch": 0.6095100864553314,
      "grad_norm": 0.46426133154926164,
      "learning_rate": 9.679601406711746e-06,
      "loss": 0.0631,
      "step": 1269
    },
    {
      "epoch": 0.6099903938520653,
      "grad_norm": 0.4772199148883563,
      "learning_rate": 9.678616414766096e-06,
      "loss": 0.056,
      "step": 1270
    },
    {
      "epoch": 0.6104707012487992,
      "grad_norm": 0.7108506602194041,
      "learning_rate": 9.677629961350113e-06,
      "loss": 0.0658,
      "step": 1271
    },
    {
      "epoch": 0.6109510086455331,
      "grad_norm": 0.6231876402263944,
      "learning_rate": 9.676642046771938e-06,
      "loss": 0.0433,
      "step": 1272
    },
    {
      "epoch": 0.6114313160422671,
      "grad_norm": 0.47557579581945064,
      "learning_rate": 9.675652671340169e-06,
      "loss": 0.04,
      "step": 1273
    },
    {
      "epoch": 0.611911623439001,
      "grad_norm": 0.7107171113991735,
      "learning_rate": 9.67466183536386e-06,
      "loss": 0.0628,
      "step": 1274
    },
    {
      "epoch": 0.6123919308357348,
      "grad_norm": 0.9476315212332672,
      "learning_rate": 9.673669539152518e-06,
      "loss": 0.0443,
      "step": 1275
    },
    {
      "epoch": 0.6128722382324687,
      "grad_norm": 0.7168398810301742,
      "learning_rate": 9.67267578301611e-06,
      "loss": 0.0744,
      "step": 1276
    },
    {
      "epoch": 0.6133525456292027,
      "grad_norm": 0.5729632454572932,
      "learning_rate": 9.67168056726506e-06,
      "loss": 0.0514,
      "step": 1277
    },
    {
      "epoch": 0.6138328530259366,
      "grad_norm": 0.9643183924220915,
      "learning_rate": 9.670683892210245e-06,
      "loss": 0.0622,
      "step": 1278
    },
    {
      "epoch": 0.6143131604226705,
      "grad_norm": 0.6683536482508268,
      "learning_rate": 9.669685758162996e-06,
      "loss": 0.0764,
      "step": 1279
    },
    {
      "epoch": 0.6147934678194045,
      "grad_norm": 0.4503574435017507,
      "learning_rate": 9.668686165435106e-06,
      "loss": 0.0494,
      "step": 1280
    },
    {
      "epoch": 0.6152737752161384,
      "grad_norm": 0.4920339006116487,
      "learning_rate": 9.667685114338819e-06,
      "loss": 0.0478,
      "step": 1281
    },
    {
      "epoch": 0.6157540826128722,
      "grad_norm": 0.5067811483748804,
      "learning_rate": 9.666682605186834e-06,
      "loss": 0.0502,
      "step": 1282
    },
    {
      "epoch": 0.6162343900096061,
      "grad_norm": 0.4640157427837885,
      "learning_rate": 9.66567863829231e-06,
      "loss": 0.0317,
      "step": 1283
    },
    {
      "epoch": 0.6167146974063401,
      "grad_norm": 0.7574081208925442,
      "learning_rate": 9.664673213968856e-06,
      "loss": 0.0478,
      "step": 1284
    },
    {
      "epoch": 0.617195004803074,
      "grad_norm": 0.6090301906896739,
      "learning_rate": 9.663666332530541e-06,
      "loss": 0.0532,
      "step": 1285
    },
    {
      "epoch": 0.6176753121998079,
      "grad_norm": 0.5041461690169365,
      "learning_rate": 9.662657994291884e-06,
      "loss": 0.059,
      "step": 1286
    },
    {
      "epoch": 0.6181556195965417,
      "grad_norm": 0.3750500141302966,
      "learning_rate": 9.661648199567866e-06,
      "loss": 0.0367,
      "step": 1287
    },
    {
      "epoch": 0.6186359269932757,
      "grad_norm": 0.5062593535367501,
      "learning_rate": 9.660636948673913e-06,
      "loss": 0.0353,
      "step": 1288
    },
    {
      "epoch": 0.6191162343900096,
      "grad_norm": 0.639235731952125,
      "learning_rate": 9.659624241925917e-06,
      "loss": 0.0596,
      "step": 1289
    },
    {
      "epoch": 0.6195965417867435,
      "grad_norm": 0.5049024822490057,
      "learning_rate": 9.65861007964022e-06,
      "loss": 0.0447,
      "step": 1290
    },
    {
      "epoch": 0.6200768491834774,
      "grad_norm": 0.4296613053659539,
      "learning_rate": 9.657594462133614e-06,
      "loss": 0.044,
      "step": 1291
    },
    {
      "epoch": 0.6205571565802114,
      "grad_norm": 0.5138878968910088,
      "learning_rate": 9.656577389723353e-06,
      "loss": 0.0493,
      "step": 1292
    },
    {
      "epoch": 0.6210374639769453,
      "grad_norm": 0.46036782082872985,
      "learning_rate": 9.655558862727141e-06,
      "loss": 0.037,
      "step": 1293
    },
    {
      "epoch": 0.6215177713736791,
      "grad_norm": 0.4704621930317916,
      "learning_rate": 9.654538881463139e-06,
      "loss": 0.048,
      "step": 1294
    },
    {
      "epoch": 0.621998078770413,
      "grad_norm": 0.48737100335382305,
      "learning_rate": 9.653517446249955e-06,
      "loss": 0.0396,
      "step": 1295
    },
    {
      "epoch": 0.622478386167147,
      "grad_norm": 0.49092045898074815,
      "learning_rate": 9.652494557406666e-06,
      "loss": 0.0456,
      "step": 1296
    },
    {
      "epoch": 0.6229586935638809,
      "grad_norm": 0.5835467675400211,
      "learning_rate": 9.65147021525279e-06,
      "loss": 0.0378,
      "step": 1297
    },
    {
      "epoch": 0.6234390009606148,
      "grad_norm": 0.42343488555568504,
      "learning_rate": 9.650444420108303e-06,
      "loss": 0.0378,
      "step": 1298
    },
    {
      "epoch": 0.6239193083573487,
      "grad_norm": 0.49168464423068825,
      "learning_rate": 9.649417172293636e-06,
      "loss": 0.0522,
      "step": 1299
    },
    {
      "epoch": 0.6243996157540826,
      "grad_norm": 0.5154918146605992,
      "learning_rate": 9.648388472129671e-06,
      "loss": 0.044,
      "step": 1300
    },
    {
      "epoch": 0.6248799231508165,
      "grad_norm": 0.3503174379219431,
      "learning_rate": 9.647358319937746e-06,
      "loss": 0.0347,
      "step": 1301
    },
    {
      "epoch": 0.6253602305475504,
      "grad_norm": 0.4137072426795727,
      "learning_rate": 9.646326716039653e-06,
      "loss": 0.0296,
      "step": 1302
    },
    {
      "epoch": 0.6258405379442843,
      "grad_norm": 0.5333253731829095,
      "learning_rate": 9.645293660757637e-06,
      "loss": 0.0504,
      "step": 1303
    },
    {
      "epoch": 0.6263208453410183,
      "grad_norm": 0.45225710559548166,
      "learning_rate": 9.644259154414396e-06,
      "loss": 0.0433,
      "step": 1304
    },
    {
      "epoch": 0.6268011527377522,
      "grad_norm": 0.5129776997920263,
      "learning_rate": 9.643223197333078e-06,
      "loss": 0.0555,
      "step": 1305
    },
    {
      "epoch": 0.627281460134486,
      "grad_norm": 0.6635014880318021,
      "learning_rate": 9.64218578983729e-06,
      "loss": 0.0663,
      "step": 1306
    },
    {
      "epoch": 0.6277617675312199,
      "grad_norm": 0.40151280112857785,
      "learning_rate": 9.641146932251088e-06,
      "loss": 0.0449,
      "step": 1307
    },
    {
      "epoch": 0.6282420749279539,
      "grad_norm": 0.4750922210262119,
      "learning_rate": 9.640106624898985e-06,
      "loss": 0.0458,
      "step": 1308
    },
    {
      "epoch": 0.6287223823246878,
      "grad_norm": 0.534233777229496,
      "learning_rate": 9.639064868105943e-06,
      "loss": 0.0522,
      "step": 1309
    },
    {
      "epoch": 0.6292026897214217,
      "grad_norm": 0.48289449635982123,
      "learning_rate": 9.638021662197376e-06,
      "loss": 0.0373,
      "step": 1310
    },
    {
      "epoch": 0.6296829971181557,
      "grad_norm": 0.5462852993159404,
      "learning_rate": 9.636977007499153e-06,
      "loss": 0.0742,
      "step": 1311
    },
    {
      "epoch": 0.6301633045148896,
      "grad_norm": 0.5378449863960542,
      "learning_rate": 9.6359309043376e-06,
      "loss": 0.0493,
      "step": 1312
    },
    {
      "epoch": 0.6306436119116234,
      "grad_norm": 0.7788750925559146,
      "learning_rate": 9.634883353039484e-06,
      "loss": 0.0855,
      "step": 1313
    },
    {
      "epoch": 0.6311239193083573,
      "grad_norm": 0.4117391703483272,
      "learning_rate": 9.633834353932035e-06,
      "loss": 0.0424,
      "step": 1314
    },
    {
      "epoch": 0.6316042267050913,
      "grad_norm": 0.7547438553021464,
      "learning_rate": 9.63278390734293e-06,
      "loss": 0.0534,
      "step": 1315
    },
    {
      "epoch": 0.6320845341018252,
      "grad_norm": 0.6007646020799848,
      "learning_rate": 9.631732013600302e-06,
      "loss": 0.0403,
      "step": 1316
    },
    {
      "epoch": 0.6325648414985591,
      "grad_norm": 0.589813047694365,
      "learning_rate": 9.63067867303273e-06,
      "loss": 0.0536,
      "step": 1317
    },
    {
      "epoch": 0.633045148895293,
      "grad_norm": 0.5824782124675916,
      "learning_rate": 9.62962388596925e-06,
      "loss": 0.0588,
      "step": 1318
    },
    {
      "epoch": 0.633525456292027,
      "grad_norm": 0.5558162647237488,
      "learning_rate": 9.628567652739348e-06,
      "loss": 0.0459,
      "step": 1319
    },
    {
      "epoch": 0.6340057636887608,
      "grad_norm": 0.43466194972289696,
      "learning_rate": 9.627509973672962e-06,
      "loss": 0.0504,
      "step": 1320
    },
    {
      "epoch": 0.6344860710854947,
      "grad_norm": 0.5311930203126808,
      "learning_rate": 9.626450849100483e-06,
      "loss": 0.0493,
      "step": 1321
    },
    {
      "epoch": 0.6349663784822286,
      "grad_norm": 0.4875635622342151,
      "learning_rate": 9.62539027935275e-06,
      "loss": 0.0502,
      "step": 1322
    },
    {
      "epoch": 0.6354466858789626,
      "grad_norm": 0.7400810886912117,
      "learning_rate": 9.624328264761056e-06,
      "loss": 0.0834,
      "step": 1323
    },
    {
      "epoch": 0.6359269932756965,
      "grad_norm": 0.48011574918186833,
      "learning_rate": 9.623264805657146e-06,
      "loss": 0.0461,
      "step": 1324
    },
    {
      "epoch": 0.6364073006724303,
      "grad_norm": 0.36881464066476766,
      "learning_rate": 9.622199902373218e-06,
      "loss": 0.0344,
      "step": 1325
    },
    {
      "epoch": 0.6368876080691642,
      "grad_norm": 0.4130880239031028,
      "learning_rate": 9.621133555241912e-06,
      "loss": 0.0326,
      "step": 1326
    },
    {
      "epoch": 0.6373679154658982,
      "grad_norm": 0.4479060787169812,
      "learning_rate": 9.620065764596328e-06,
      "loss": 0.0393,
      "step": 1327
    },
    {
      "epoch": 0.6378482228626321,
      "grad_norm": 0.5942961946492873,
      "learning_rate": 9.618996530770018e-06,
      "loss": 0.0432,
      "step": 1328
    },
    {
      "epoch": 0.638328530259366,
      "grad_norm": 0.5638567374914406,
      "learning_rate": 9.617925854096975e-06,
      "loss": 0.0501,
      "step": 1329
    },
    {
      "epoch": 0.6388088376560999,
      "grad_norm": 0.42880488417099755,
      "learning_rate": 9.616853734911653e-06,
      "loss": 0.0374,
      "step": 1330
    },
    {
      "epoch": 0.6392891450528339,
      "grad_norm": 0.5463743835118613,
      "learning_rate": 9.615780173548952e-06,
      "loss": 0.0516,
      "step": 1331
    },
    {
      "epoch": 0.6397694524495677,
      "grad_norm": 0.691161000342225,
      "learning_rate": 9.614705170344221e-06,
      "loss": 0.0848,
      "step": 1332
    },
    {
      "epoch": 0.6402497598463016,
      "grad_norm": 0.6156410304491652,
      "learning_rate": 9.613628725633262e-06,
      "loss": 0.056,
      "step": 1333
    },
    {
      "epoch": 0.6407300672430355,
      "grad_norm": 0.45322396316185093,
      "learning_rate": 9.612550839752326e-06,
      "loss": 0.044,
      "step": 1334
    },
    {
      "epoch": 0.6412103746397695,
      "grad_norm": 0.6507298412844991,
      "learning_rate": 9.611471513038115e-06,
      "loss": 0.0687,
      "step": 1335
    },
    {
      "epoch": 0.6416906820365034,
      "grad_norm": 0.6325291411421659,
      "learning_rate": 9.610390745827783e-06,
      "loss": 0.0608,
      "step": 1336
    },
    {
      "epoch": 0.6421709894332372,
      "grad_norm": 0.60388801983135,
      "learning_rate": 9.60930853845893e-06,
      "loss": 0.0402,
      "step": 1337
    },
    {
      "epoch": 0.6426512968299711,
      "grad_norm": 0.4995528775460853,
      "learning_rate": 9.608224891269607e-06,
      "loss": 0.0381,
      "step": 1338
    },
    {
      "epoch": 0.6431316042267051,
      "grad_norm": 0.3700705272643591,
      "learning_rate": 9.607139804598316e-06,
      "loss": 0.0349,
      "step": 1339
    },
    {
      "epoch": 0.643611911623439,
      "grad_norm": 0.4362560597879333,
      "learning_rate": 9.606053278784009e-06,
      "loss": 0.0403,
      "step": 1340
    },
    {
      "epoch": 0.6440922190201729,
      "grad_norm": 0.54879971648486,
      "learning_rate": 9.604965314166085e-06,
      "loss": 0.0541,
      "step": 1341
    },
    {
      "epoch": 0.6445725264169068,
      "grad_norm": 0.4929293793701462,
      "learning_rate": 9.603875911084394e-06,
      "loss": 0.0365,
      "step": 1342
    },
    {
      "epoch": 0.6450528338136408,
      "grad_norm": 0.5628523668339881,
      "learning_rate": 9.602785069879239e-06,
      "loss": 0.0598,
      "step": 1343
    },
    {
      "epoch": 0.6455331412103746,
      "grad_norm": 0.467411924570006,
      "learning_rate": 9.601692790891363e-06,
      "loss": 0.0521,
      "step": 1344
    },
    {
      "epoch": 0.6460134486071085,
      "grad_norm": 0.5210675000514947,
      "learning_rate": 9.600599074461967e-06,
      "loss": 0.0637,
      "step": 1345
    },
    {
      "epoch": 0.6464937560038425,
      "grad_norm": 0.6153647716384278,
      "learning_rate": 9.599503920932698e-06,
      "loss": 0.0479,
      "step": 1346
    },
    {
      "epoch": 0.6469740634005764,
      "grad_norm": 0.5616593537777448,
      "learning_rate": 9.59840733064565e-06,
      "loss": 0.0364,
      "step": 1347
    },
    {
      "epoch": 0.6474543707973103,
      "grad_norm": 0.3344274743801641,
      "learning_rate": 9.59730930394337e-06,
      "loss": 0.0314,
      "step": 1348
    },
    {
      "epoch": 0.6479346781940442,
      "grad_norm": 0.6754093193995799,
      "learning_rate": 9.59620984116885e-06,
      "loss": 0.0438,
      "step": 1349
    },
    {
      "epoch": 0.6484149855907781,
      "grad_norm": 0.8533549419575339,
      "learning_rate": 9.595108942665528e-06,
      "loss": 0.0513,
      "step": 1350
    },
    {
      "epoch": 0.648895292987512,
      "grad_norm": 0.3620854442522442,
      "learning_rate": 9.5940066087773e-06,
      "loss": 0.0423,
      "step": 1351
    },
    {
      "epoch": 0.6493756003842459,
      "grad_norm": 0.5893844408390839,
      "learning_rate": 9.592902839848502e-06,
      "loss": 0.0377,
      "step": 1352
    },
    {
      "epoch": 0.6498559077809798,
      "grad_norm": 0.4092201290840598,
      "learning_rate": 9.591797636223921e-06,
      "loss": 0.0411,
      "step": 1353
    },
    {
      "epoch": 0.6503362151777138,
      "grad_norm": 0.47322297966868715,
      "learning_rate": 9.590690998248791e-06,
      "loss": 0.043,
      "step": 1354
    },
    {
      "epoch": 0.6508165225744477,
      "grad_norm": 0.4636748975157291,
      "learning_rate": 9.589582926268798e-06,
      "loss": 0.0429,
      "step": 1355
    },
    {
      "epoch": 0.6512968299711815,
      "grad_norm": 0.4349692458727795,
      "learning_rate": 9.588473420630071e-06,
      "loss": 0.0444,
      "step": 1356
    },
    {
      "epoch": 0.6517771373679154,
      "grad_norm": 0.5677476836765938,
      "learning_rate": 9.587362481679187e-06,
      "loss": 0.0562,
      "step": 1357
    },
    {
      "epoch": 0.6522574447646494,
      "grad_norm": 0.4371170679582839,
      "learning_rate": 9.586250109763176e-06,
      "loss": 0.0371,
      "step": 1358
    },
    {
      "epoch": 0.6527377521613833,
      "grad_norm": 0.6019100000609227,
      "learning_rate": 9.585136305229513e-06,
      "loss": 0.0488,
      "step": 1359
    },
    {
      "epoch": 0.6532180595581172,
      "grad_norm": 0.38363888520290607,
      "learning_rate": 9.584021068426114e-06,
      "loss": 0.0338,
      "step": 1360
    },
    {
      "epoch": 0.6536983669548511,
      "grad_norm": 0.4993078534835986,
      "learning_rate": 9.582904399701353e-06,
      "loss": 0.0431,
      "step": 1361
    },
    {
      "epoch": 0.654178674351585,
      "grad_norm": 0.5671412832634853,
      "learning_rate": 9.581786299404046e-06,
      "loss": 0.0561,
      "step": 1362
    },
    {
      "epoch": 0.6546589817483189,
      "grad_norm": 0.7306005927500124,
      "learning_rate": 9.580666767883456e-06,
      "loss": 0.042,
      "step": 1363
    },
    {
      "epoch": 0.6551392891450528,
      "grad_norm": 0.5032575865518208,
      "learning_rate": 9.579545805489292e-06,
      "loss": 0.0518,
      "step": 1364
    },
    {
      "epoch": 0.6556195965417867,
      "grad_norm": 0.6343768547939731,
      "learning_rate": 9.578423412571713e-06,
      "loss": 0.055,
      "step": 1365
    },
    {
      "epoch": 0.6560999039385207,
      "grad_norm": 0.6313252829889849,
      "learning_rate": 9.577299589481325e-06,
      "loss": 0.0499,
      "step": 1366
    },
    {
      "epoch": 0.6565802113352546,
      "grad_norm": 0.5874538450840524,
      "learning_rate": 9.576174336569177e-06,
      "loss": 0.0546,
      "step": 1367
    },
    {
      "epoch": 0.6570605187319885,
      "grad_norm": 0.4208422595696595,
      "learning_rate": 9.575047654186768e-06,
      "loss": 0.0427,
      "step": 1368
    },
    {
      "epoch": 0.6575408261287223,
      "grad_norm": 0.6079831399082672,
      "learning_rate": 9.57391954268604e-06,
      "loss": 0.051,
      "step": 1369
    },
    {
      "epoch": 0.6580211335254563,
      "grad_norm": 0.5767257558258905,
      "learning_rate": 9.57279000241939e-06,
      "loss": 0.0402,
      "step": 1370
    },
    {
      "epoch": 0.6585014409221902,
      "grad_norm": 0.4733503070820277,
      "learning_rate": 9.571659033739648e-06,
      "loss": 0.048,
      "step": 1371
    },
    {
      "epoch": 0.6589817483189241,
      "grad_norm": 0.4524785331372919,
      "learning_rate": 9.570526637000102e-06,
      "loss": 0.04,
      "step": 1372
    },
    {
      "epoch": 0.659462055715658,
      "grad_norm": 0.6668276001778005,
      "learning_rate": 9.56939281255448e-06,
      "loss": 0.0498,
      "step": 1373
    },
    {
      "epoch": 0.659942363112392,
      "grad_norm": 0.49259152852847804,
      "learning_rate": 9.568257560756955e-06,
      "loss": 0.0386,
      "step": 1374
    },
    {
      "epoch": 0.6604226705091258,
      "grad_norm": 0.8583845386777227,
      "learning_rate": 9.567120881962152e-06,
      "loss": 0.0498,
      "step": 1375
    },
    {
      "epoch": 0.6609029779058597,
      "grad_norm": 0.533492536983965,
      "learning_rate": 9.565982776525136e-06,
      "loss": 0.0472,
      "step": 1376
    },
    {
      "epoch": 0.6613832853025937,
      "grad_norm": 0.6027948215201636,
      "learning_rate": 9.56484324480142e-06,
      "loss": 0.047,
      "step": 1377
    },
    {
      "epoch": 0.6618635926993276,
      "grad_norm": 0.3382733587601467,
      "learning_rate": 9.563702287146963e-06,
      "loss": 0.0423,
      "step": 1378
    },
    {
      "epoch": 0.6623439000960615,
      "grad_norm": 0.40078317285361365,
      "learning_rate": 9.562559903918167e-06,
      "loss": 0.0437,
      "step": 1379
    },
    {
      "epoch": 0.6628242074927954,
      "grad_norm": 0.4152293792382375,
      "learning_rate": 9.561416095471882e-06,
      "loss": 0.0487,
      "step": 1380
    },
    {
      "epoch": 0.6633045148895294,
      "grad_norm": 0.37525899009119723,
      "learning_rate": 9.560270862165401e-06,
      "loss": 0.0445,
      "step": 1381
    },
    {
      "epoch": 0.6637848222862632,
      "grad_norm": 0.4239007996552649,
      "learning_rate": 9.559124204356465e-06,
      "loss": 0.0365,
      "step": 1382
    },
    {
      "epoch": 0.6642651296829971,
      "grad_norm": 0.44856667071995515,
      "learning_rate": 9.557976122403259e-06,
      "loss": 0.0432,
      "step": 1383
    },
    {
      "epoch": 0.664745437079731,
      "grad_norm": 0.40105877291250286,
      "learning_rate": 9.556826616664408e-06,
      "loss": 0.0375,
      "step": 1384
    },
    {
      "epoch": 0.665225744476465,
      "grad_norm": 0.6413226916161751,
      "learning_rate": 9.555675687498988e-06,
      "loss": 0.0491,
      "step": 1385
    },
    {
      "epoch": 0.6657060518731989,
      "grad_norm": 0.4994127702541754,
      "learning_rate": 9.554523335266519e-06,
      "loss": 0.0362,
      "step": 1386
    },
    {
      "epoch": 0.6661863592699327,
      "grad_norm": 0.43217572367178236,
      "learning_rate": 9.553369560326961e-06,
      "loss": 0.041,
      "step": 1387
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.5576149940821006,
      "learning_rate": 9.552214363040725e-06,
      "loss": 0.0609,
      "step": 1388
    },
    {
      "epoch": 0.6671469740634006,
      "grad_norm": 0.4612462479554566,
      "learning_rate": 9.551057743768658e-06,
      "loss": 0.0326,
      "step": 1389
    },
    {
      "epoch": 0.6676272814601345,
      "grad_norm": 0.7707155411339915,
      "learning_rate": 9.54989970287206e-06,
      "loss": 0.0515,
      "step": 1390
    },
    {
      "epoch": 0.6681075888568684,
      "grad_norm": 0.6183305732020524,
      "learning_rate": 9.54874024071267e-06,
      "loss": 0.0573,
      "step": 1391
    },
    {
      "epoch": 0.6685878962536023,
      "grad_norm": 0.7355014675302118,
      "learning_rate": 9.54757935765267e-06,
      "loss": 0.0509,
      "step": 1392
    },
    {
      "epoch": 0.6690682036503363,
      "grad_norm": 0.41942626169957864,
      "learning_rate": 9.54641705405469e-06,
      "loss": 0.0346,
      "step": 1393
    },
    {
      "epoch": 0.6695485110470701,
      "grad_norm": 0.46998738583164973,
      "learning_rate": 9.5452533302818e-06,
      "loss": 0.0441,
      "step": 1394
    },
    {
      "epoch": 0.670028818443804,
      "grad_norm": 0.8255715566434525,
      "learning_rate": 9.544088186697515e-06,
      "loss": 0.0762,
      "step": 1395
    },
    {
      "epoch": 0.6705091258405379,
      "grad_norm": 0.742976741485616,
      "learning_rate": 9.542921623665796e-06,
      "loss": 0.056,
      "step": 1396
    },
    {
      "epoch": 0.6709894332372719,
      "grad_norm": 0.7975592021603152,
      "learning_rate": 9.541753641551042e-06,
      "loss": 0.0618,
      "step": 1397
    },
    {
      "epoch": 0.6714697406340058,
      "grad_norm": 0.7133686245791877,
      "learning_rate": 9.540584240718098e-06,
      "loss": 0.0359,
      "step": 1398
    },
    {
      "epoch": 0.6719500480307397,
      "grad_norm": 0.6417332791114704,
      "learning_rate": 9.539413421532256e-06,
      "loss": 0.067,
      "step": 1399
    },
    {
      "epoch": 0.6724303554274735,
      "grad_norm": 0.6512453356740282,
      "learning_rate": 9.538241184359245e-06,
      "loss": 0.057,
      "step": 1400
    },
    {
      "epoch": 0.6729106628242075,
      "grad_norm": 0.6011163829953513,
      "learning_rate": 9.537067529565241e-06,
      "loss": 0.0433,
      "step": 1401
    },
    {
      "epoch": 0.6733909702209414,
      "grad_norm": 0.5605867258554449,
      "learning_rate": 9.535892457516858e-06,
      "loss": 0.07,
      "step": 1402
    },
    {
      "epoch": 0.6738712776176753,
      "grad_norm": 0.44843028713428207,
      "learning_rate": 9.53471596858116e-06,
      "loss": 0.0346,
      "step": 1403
    },
    {
      "epoch": 0.6743515850144092,
      "grad_norm": 0.45306372964448993,
      "learning_rate": 9.533538063125649e-06,
      "loss": 0.0395,
      "step": 1404
    },
    {
      "epoch": 0.6748318924111432,
      "grad_norm": 0.665610180113429,
      "learning_rate": 9.532358741518265e-06,
      "loss": 0.0727,
      "step": 1405
    },
    {
      "epoch": 0.675312199807877,
      "grad_norm": 0.40211030378348633,
      "learning_rate": 9.531178004127404e-06,
      "loss": 0.0415,
      "step": 1406
    },
    {
      "epoch": 0.6757925072046109,
      "grad_norm": 0.4867082233781477,
      "learning_rate": 9.52999585132189e-06,
      "loss": 0.0413,
      "step": 1407
    },
    {
      "epoch": 0.6762728146013448,
      "grad_norm": 0.7741469232795118,
      "learning_rate": 9.528812283470994e-06,
      "loss": 0.0717,
      "step": 1408
    },
    {
      "epoch": 0.6767531219980788,
      "grad_norm": 0.42760299424580006,
      "learning_rate": 9.527627300944434e-06,
      "loss": 0.0378,
      "step": 1409
    },
    {
      "epoch": 0.6772334293948127,
      "grad_norm": 0.6198122845932187,
      "learning_rate": 9.52644090411236e-06,
      "loss": 0.0644,
      "step": 1410
    },
    {
      "epoch": 0.6777137367915466,
      "grad_norm": 0.7010269392103107,
      "learning_rate": 9.525253093345377e-06,
      "loss": 0.0742,
      "step": 1411
    },
    {
      "epoch": 0.6781940441882806,
      "grad_norm": 0.46098176739925134,
      "learning_rate": 9.524063869014517e-06,
      "loss": 0.0539,
      "step": 1412
    },
    {
      "epoch": 0.6786743515850144,
      "grad_norm": 0.45626576421316317,
      "learning_rate": 9.522873231491268e-06,
      "loss": 0.0459,
      "step": 1413
    },
    {
      "epoch": 0.6791546589817483,
      "grad_norm": 0.5235232885723231,
      "learning_rate": 9.521681181147544e-06,
      "loss": 0.0354,
      "step": 1414
    },
    {
      "epoch": 0.6796349663784822,
      "grad_norm": 0.4083701335457131,
      "learning_rate": 9.520487718355716e-06,
      "loss": 0.0376,
      "step": 1415
    },
    {
      "epoch": 0.6801152737752162,
      "grad_norm": 0.4519589464495528,
      "learning_rate": 9.519292843488584e-06,
      "loss": 0.0435,
      "step": 1416
    },
    {
      "epoch": 0.6805955811719501,
      "grad_norm": 0.5270239622507114,
      "learning_rate": 9.518096556919396e-06,
      "loss": 0.0385,
      "step": 1417
    },
    {
      "epoch": 0.681075888568684,
      "grad_norm": 0.49655054870938087,
      "learning_rate": 9.51689885902184e-06,
      "loss": 0.0418,
      "step": 1418
    },
    {
      "epoch": 0.6815561959654178,
      "grad_norm": 0.48390295607618405,
      "learning_rate": 9.51569975017004e-06,
      "loss": 0.0473,
      "step": 1419
    },
    {
      "epoch": 0.6820365033621518,
      "grad_norm": 0.4134572167278166,
      "learning_rate": 9.514499230738567e-06,
      "loss": 0.0356,
      "step": 1420
    },
    {
      "epoch": 0.6825168107588857,
      "grad_norm": 0.42968236699716256,
      "learning_rate": 9.51329730110243e-06,
      "loss": 0.0396,
      "step": 1421
    },
    {
      "epoch": 0.6829971181556196,
      "grad_norm": 0.507732504462261,
      "learning_rate": 9.512093961637077e-06,
      "loss": 0.044,
      "step": 1422
    },
    {
      "epoch": 0.6834774255523535,
      "grad_norm": 0.513408290011866,
      "learning_rate": 9.510889212718398e-06,
      "loss": 0.0539,
      "step": 1423
    },
    {
      "epoch": 0.6839577329490875,
      "grad_norm": 0.5815081646240737,
      "learning_rate": 9.509683054722726e-06,
      "loss": 0.0398,
      "step": 1424
    },
    {
      "epoch": 0.6844380403458213,
      "grad_norm": 0.4078417115979057,
      "learning_rate": 9.508475488026829e-06,
      "loss": 0.0332,
      "step": 1425
    },
    {
      "epoch": 0.6849183477425552,
      "grad_norm": 1.0586820403034098,
      "learning_rate": 9.507266513007918e-06,
      "loss": 0.0763,
      "step": 1426
    },
    {
      "epoch": 0.6853986551392891,
      "grad_norm": 0.564383510484416,
      "learning_rate": 9.506056130043644e-06,
      "loss": 0.0434,
      "step": 1427
    },
    {
      "epoch": 0.6858789625360231,
      "grad_norm": 0.3768335964210522,
      "learning_rate": 9.504844339512096e-06,
      "loss": 0.0326,
      "step": 1428
    },
    {
      "epoch": 0.686359269932757,
      "grad_norm": 0.5437219091200941,
      "learning_rate": 9.503631141791804e-06,
      "loss": 0.0551,
      "step": 1429
    },
    {
      "epoch": 0.6868395773294909,
      "grad_norm": 0.39882576370499795,
      "learning_rate": 9.502416537261739e-06,
      "loss": 0.0394,
      "step": 1430
    },
    {
      "epoch": 0.6873198847262247,
      "grad_norm": 0.5298522285045902,
      "learning_rate": 9.50120052630131e-06,
      "loss": 0.0424,
      "step": 1431
    },
    {
      "epoch": 0.6878001921229587,
      "grad_norm": 0.7335489985769658,
      "learning_rate": 9.499983109290361e-06,
      "loss": 0.0695,
      "step": 1432
    },
    {
      "epoch": 0.6882804995196926,
      "grad_norm": 0.7767384589812387,
      "learning_rate": 9.498764286609183e-06,
      "loss": 0.0684,
      "step": 1433
    },
    {
      "epoch": 0.6887608069164265,
      "grad_norm": 0.4485179955778425,
      "learning_rate": 9.4975440586385e-06,
      "loss": 0.0545,
      "step": 1434
    },
    {
      "epoch": 0.6892411143131604,
      "grad_norm": 0.7644156832335937,
      "learning_rate": 9.496322425759484e-06,
      "loss": 0.056,
      "step": 1435
    },
    {
      "epoch": 0.6897214217098944,
      "grad_norm": 0.6682437861904863,
      "learning_rate": 9.495099388353731e-06,
      "loss": 0.0413,
      "step": 1436
    },
    {
      "epoch": 0.6902017291066282,
      "grad_norm": 0.2782387488882061,
      "learning_rate": 9.493874946803287e-06,
      "loss": 0.0307,
      "step": 1437
    },
    {
      "epoch": 0.6906820365033621,
      "grad_norm": 0.7196277741729624,
      "learning_rate": 9.492649101490636e-06,
      "loss": 0.0605,
      "step": 1438
    },
    {
      "epoch": 0.691162343900096,
      "grad_norm": 0.43324754881869854,
      "learning_rate": 9.491421852798695e-06,
      "loss": 0.0475,
      "step": 1439
    },
    {
      "epoch": 0.69164265129683,
      "grad_norm": 0.43738740093767214,
      "learning_rate": 9.490193201110825e-06,
      "loss": 0.0375,
      "step": 1440
    },
    {
      "epoch": 0.6921229586935639,
      "grad_norm": 0.5488539442142931,
      "learning_rate": 9.48896314681082e-06,
      "loss": 0.0728,
      "step": 1441
    },
    {
      "epoch": 0.6926032660902978,
      "grad_norm": 1.1540239102800764,
      "learning_rate": 9.48773169028292e-06,
      "loss": 0.0586,
      "step": 1442
    },
    {
      "epoch": 0.6930835734870316,
      "grad_norm": 0.6283641043852788,
      "learning_rate": 9.486498831911792e-06,
      "loss": 0.0736,
      "step": 1443
    },
    {
      "epoch": 0.6935638808837656,
      "grad_norm": 1.0108486453877055,
      "learning_rate": 9.485264572082551e-06,
      "loss": 0.0918,
      "step": 1444
    },
    {
      "epoch": 0.6940441882804995,
      "grad_norm": 0.40880665155636625,
      "learning_rate": 9.484028911180742e-06,
      "loss": 0.0389,
      "step": 1445
    },
    {
      "epoch": 0.6945244956772334,
      "grad_norm": 0.47431051486539744,
      "learning_rate": 9.482791849592354e-06,
      "loss": 0.0597,
      "step": 1446
    },
    {
      "epoch": 0.6950048030739674,
      "grad_norm": 0.8329629013113664,
      "learning_rate": 9.48155338770381e-06,
      "loss": 0.0727,
      "step": 1447
    },
    {
      "epoch": 0.6954851104707013,
      "grad_norm": 0.5434685409654171,
      "learning_rate": 9.480313525901973e-06,
      "loss": 0.0505,
      "step": 1448
    },
    {
      "epoch": 0.6959654178674352,
      "grad_norm": 0.6327014782848,
      "learning_rate": 9.479072264574138e-06,
      "loss": 0.0524,
      "step": 1449
    },
    {
      "epoch": 0.696445725264169,
      "grad_norm": 0.4768500817588462,
      "learning_rate": 9.477829604108044e-06,
      "loss": 0.0479,
      "step": 1450
    },
    {
      "epoch": 0.696926032660903,
      "grad_norm": 0.33637687220332657,
      "learning_rate": 9.476585544891862e-06,
      "loss": 0.0335,
      "step": 1451
    },
    {
      "epoch": 0.6974063400576369,
      "grad_norm": 0.6410159305875692,
      "learning_rate": 9.475340087314203e-06,
      "loss": 0.0457,
      "step": 1452
    },
    {
      "epoch": 0.6978866474543708,
      "grad_norm": 0.7561772611383961,
      "learning_rate": 9.47409323176411e-06,
      "loss": 0.0625,
      "step": 1453
    },
    {
      "epoch": 0.6983669548511047,
      "grad_norm": 0.3920818511938983,
      "learning_rate": 9.472844978631071e-06,
      "loss": 0.0282,
      "step": 1454
    },
    {
      "epoch": 0.6988472622478387,
      "grad_norm": 0.7746401992023322,
      "learning_rate": 9.471595328305002e-06,
      "loss": 0.0577,
      "step": 1455
    },
    {
      "epoch": 0.6993275696445725,
      "grad_norm": 1.1432462417277984,
      "learning_rate": 9.470344281176261e-06,
      "loss": 0.0503,
      "step": 1456
    },
    {
      "epoch": 0.6998078770413064,
      "grad_norm": 0.5579207608396545,
      "learning_rate": 9.469091837635641e-06,
      "loss": 0.0436,
      "step": 1457
    },
    {
      "epoch": 0.7002881844380403,
      "grad_norm": 0.4293775018518884,
      "learning_rate": 9.467837998074369e-06,
      "loss": 0.0426,
      "step": 1458
    },
    {
      "epoch": 0.7007684918347743,
      "grad_norm": 0.5654128796038869,
      "learning_rate": 9.466582762884111e-06,
      "loss": 0.0515,
      "step": 1459
    },
    {
      "epoch": 0.7012487992315082,
      "grad_norm": 0.6275264278723893,
      "learning_rate": 9.465326132456966e-06,
      "loss": 0.0314,
      "step": 1460
    },
    {
      "epoch": 0.7017291066282421,
      "grad_norm": 0.7608816446615059,
      "learning_rate": 9.464068107185476e-06,
      "loss": 0.0793,
      "step": 1461
    },
    {
      "epoch": 0.7022094140249759,
      "grad_norm": 0.588693970208091,
      "learning_rate": 9.462808687462606e-06,
      "loss": 0.0451,
      "step": 1462
    },
    {
      "epoch": 0.7026897214217099,
      "grad_norm": 0.6771566307850208,
      "learning_rate": 9.461547873681767e-06,
      "loss": 0.0515,
      "step": 1463
    },
    {
      "epoch": 0.7031700288184438,
      "grad_norm": 0.526932323467287,
      "learning_rate": 9.460285666236804e-06,
      "loss": 0.057,
      "step": 1464
    },
    {
      "epoch": 0.7036503362151777,
      "grad_norm": 0.6547665214657098,
      "learning_rate": 9.459022065521994e-06,
      "loss": 0.0469,
      "step": 1465
    },
    {
      "epoch": 0.7041306436119116,
      "grad_norm": 1.1306126178130151,
      "learning_rate": 9.45775707193205e-06,
      "loss": 0.0437,
      "step": 1466
    },
    {
      "epoch": 0.7046109510086456,
      "grad_norm": 0.6656984791797591,
      "learning_rate": 9.456490685862123e-06,
      "loss": 0.0657,
      "step": 1467
    },
    {
      "epoch": 0.7050912584053795,
      "grad_norm": 0.5262955218143904,
      "learning_rate": 9.455222907707795e-06,
      "loss": 0.0405,
      "step": 1468
    },
    {
      "epoch": 0.7055715658021133,
      "grad_norm": 0.4113212330571846,
      "learning_rate": 9.453953737865087e-06,
      "loss": 0.0331,
      "step": 1469
    },
    {
      "epoch": 0.7060518731988472,
      "grad_norm": 0.4687499190476505,
      "learning_rate": 9.452683176730452e-06,
      "loss": 0.0424,
      "step": 1470
    },
    {
      "epoch": 0.7065321805955812,
      "grad_norm": 0.5572677332289074,
      "learning_rate": 9.451411224700776e-06,
      "loss": 0.0612,
      "step": 1471
    },
    {
      "epoch": 0.7070124879923151,
      "grad_norm": 0.39073970748339487,
      "learning_rate": 9.450137882173385e-06,
      "loss": 0.0279,
      "step": 1472
    },
    {
      "epoch": 0.707492795389049,
      "grad_norm": 0.8716017403152045,
      "learning_rate": 9.448863149546032e-06,
      "loss": 0.0612,
      "step": 1473
    },
    {
      "epoch": 0.7079731027857828,
      "grad_norm": 0.8087490747385843,
      "learning_rate": 9.447587027216912e-06,
      "loss": 0.0754,
      "step": 1474
    },
    {
      "epoch": 0.7084534101825168,
      "grad_norm": 0.40033610599261016,
      "learning_rate": 9.446309515584648e-06,
      "loss": 0.0516,
      "step": 1475
    },
    {
      "epoch": 0.7089337175792507,
      "grad_norm": 0.3828791531186985,
      "learning_rate": 9.445030615048301e-06,
      "loss": 0.0308,
      "step": 1476
    },
    {
      "epoch": 0.7094140249759846,
      "grad_norm": 0.558762243566099,
      "learning_rate": 9.443750326007361e-06,
      "loss": 0.046,
      "step": 1477
    },
    {
      "epoch": 0.7098943323727186,
      "grad_norm": 0.46666384505267333,
      "learning_rate": 9.44246864886176e-06,
      "loss": 0.0463,
      "step": 1478
    },
    {
      "epoch": 0.7103746397694525,
      "grad_norm": 0.4943805321936899,
      "learning_rate": 9.441185584011854e-06,
      "loss": 0.0412,
      "step": 1479
    },
    {
      "epoch": 0.7108549471661864,
      "grad_norm": 0.4595133347133602,
      "learning_rate": 9.439901131858437e-06,
      "loss": 0.0457,
      "step": 1480
    },
    {
      "epoch": 0.7113352545629202,
      "grad_norm": 0.43379618763244315,
      "learning_rate": 9.43861529280274e-06,
      "loss": 0.0357,
      "step": 1481
    },
    {
      "epoch": 0.7118155619596542,
      "grad_norm": 0.58990377026948,
      "learning_rate": 9.43732806724642e-06,
      "loss": 0.0533,
      "step": 1482
    },
    {
      "epoch": 0.7122958693563881,
      "grad_norm": 0.31473504698760496,
      "learning_rate": 9.436039455591574e-06,
      "loss": 0.0309,
      "step": 1483
    },
    {
      "epoch": 0.712776176753122,
      "grad_norm": 0.45683837690648205,
      "learning_rate": 9.434749458240723e-06,
      "loss": 0.047,
      "step": 1484
    },
    {
      "epoch": 0.7132564841498559,
      "grad_norm": 0.3777015877291396,
      "learning_rate": 9.433458075596834e-06,
      "loss": 0.0434,
      "step": 1485
    },
    {
      "epoch": 0.7137367915465899,
      "grad_norm": 0.5110934466400615,
      "learning_rate": 9.432165308063293e-06,
      "loss": 0.0603,
      "step": 1486
    },
    {
      "epoch": 0.7142170989433237,
      "grad_norm": 0.4887572837652129,
      "learning_rate": 9.430871156043929e-06,
      "loss": 0.0508,
      "step": 1487
    },
    {
      "epoch": 0.7146974063400576,
      "grad_norm": 0.5164375394535271,
      "learning_rate": 9.429575619942996e-06,
      "loss": 0.0357,
      "step": 1488
    },
    {
      "epoch": 0.7151777137367915,
      "grad_norm": 0.4860839779256668,
      "learning_rate": 9.428278700165185e-06,
      "loss": 0.0496,
      "step": 1489
    },
    {
      "epoch": 0.7156580211335255,
      "grad_norm": 0.44939726645063366,
      "learning_rate": 9.426980397115619e-06,
      "loss": 0.0392,
      "step": 1490
    },
    {
      "epoch": 0.7161383285302594,
      "grad_norm": 0.4491948208101559,
      "learning_rate": 9.42568071119985e-06,
      "loss": 0.0418,
      "step": 1491
    },
    {
      "epoch": 0.7166186359269933,
      "grad_norm": 0.4413729317811157,
      "learning_rate": 9.424379642823864e-06,
      "loss": 0.0418,
      "step": 1492
    },
    {
      "epoch": 0.7170989433237271,
      "grad_norm": 0.6050908044633452,
      "learning_rate": 9.423077192394081e-06,
      "loss": 0.054,
      "step": 1493
    },
    {
      "epoch": 0.7175792507204611,
      "grad_norm": 0.34829153690830816,
      "learning_rate": 9.421773360317348e-06,
      "loss": 0.0455,
      "step": 1494
    },
    {
      "epoch": 0.718059558117195,
      "grad_norm": 0.6479436968237636,
      "learning_rate": 9.420468147000947e-06,
      "loss": 0.0591,
      "step": 1495
    },
    {
      "epoch": 0.7185398655139289,
      "grad_norm": 0.5935139475806216,
      "learning_rate": 9.419161552852592e-06,
      "loss": 0.0592,
      "step": 1496
    },
    {
      "epoch": 0.7190201729106628,
      "grad_norm": 0.35365850589490483,
      "learning_rate": 9.417853578280425e-06,
      "loss": 0.035,
      "step": 1497
    },
    {
      "epoch": 0.7195004803073968,
      "grad_norm": 0.5335048893582442,
      "learning_rate": 9.41654422369302e-06,
      "loss": 0.0343,
      "step": 1498
    },
    {
      "epoch": 0.7199807877041307,
      "grad_norm": 0.47006252573172974,
      "learning_rate": 9.415233489499388e-06,
      "loss": 0.0466,
      "step": 1499
    },
    {
      "epoch": 0.7204610951008645,
      "grad_norm": 0.4138903520224807,
      "learning_rate": 9.413921376108958e-06,
      "loss": 0.0337,
      "step": 1500
    },
    {
      "epoch": 0.7209414024975984,
      "grad_norm": 0.5302837596305445,
      "learning_rate": 9.412607883931608e-06,
      "loss": 0.0355,
      "step": 1501
    },
    {
      "epoch": 0.7214217098943324,
      "grad_norm": 0.46188113895356303,
      "learning_rate": 9.411293013377628e-06,
      "loss": 0.0464,
      "step": 1502
    },
    {
      "epoch": 0.7219020172910663,
      "grad_norm": 0.5048308160692421,
      "learning_rate": 9.409976764857752e-06,
      "loss": 0.0659,
      "step": 1503
    },
    {
      "epoch": 0.7223823246878002,
      "grad_norm": 0.42557944797976405,
      "learning_rate": 9.408659138783138e-06,
      "loss": 0.0476,
      "step": 1504
    },
    {
      "epoch": 0.722862632084534,
      "grad_norm": 0.43337734311918524,
      "learning_rate": 9.407340135565375e-06,
      "loss": 0.0473,
      "step": 1505
    },
    {
      "epoch": 0.723342939481268,
      "grad_norm": 0.43466454870999904,
      "learning_rate": 9.406019755616484e-06,
      "loss": 0.0456,
      "step": 1506
    },
    {
      "epoch": 0.7238232468780019,
      "grad_norm": 0.44882261533196544,
      "learning_rate": 9.404697999348917e-06,
      "loss": 0.0545,
      "step": 1507
    },
    {
      "epoch": 0.7243035542747358,
      "grad_norm": 0.4627444515845815,
      "learning_rate": 9.403374867175553e-06,
      "loss": 0.039,
      "step": 1508
    },
    {
      "epoch": 0.7247838616714697,
      "grad_norm": 0.43987793983099877,
      "learning_rate": 9.4020503595097e-06,
      "loss": 0.0315,
      "step": 1509
    },
    {
      "epoch": 0.7252641690682037,
      "grad_norm": 0.4866538331222698,
      "learning_rate": 9.400724476765099e-06,
      "loss": 0.0451,
      "step": 1510
    },
    {
      "epoch": 0.7257444764649376,
      "grad_norm": 0.35933139183080276,
      "learning_rate": 9.39939721935592e-06,
      "loss": 0.0341,
      "step": 1511
    },
    {
      "epoch": 0.7262247838616714,
      "grad_norm": 0.36877632658996023,
      "learning_rate": 9.398068587696758e-06,
      "loss": 0.0382,
      "step": 1512
    },
    {
      "epoch": 0.7267050912584054,
      "grad_norm": 0.9738122717404578,
      "learning_rate": 9.396738582202645e-06,
      "loss": 0.0742,
      "step": 1513
    },
    {
      "epoch": 0.7271853986551393,
      "grad_norm": 0.3887278676876787,
      "learning_rate": 9.395407203289036e-06,
      "loss": 0.0376,
      "step": 1514
    },
    {
      "epoch": 0.7276657060518732,
      "grad_norm": 0.6994372693236012,
      "learning_rate": 9.394074451371817e-06,
      "loss": 0.0394,
      "step": 1515
    },
    {
      "epoch": 0.7281460134486071,
      "grad_norm": 0.45247145506590497,
      "learning_rate": 9.392740326867304e-06,
      "loss": 0.0364,
      "step": 1516
    },
    {
      "epoch": 0.7286263208453411,
      "grad_norm": 0.5967559115342186,
      "learning_rate": 9.391404830192239e-06,
      "loss": 0.0333,
      "step": 1517
    },
    {
      "epoch": 0.729106628242075,
      "grad_norm": 0.6079539868350357,
      "learning_rate": 9.390067961763795e-06,
      "loss": 0.0472,
      "step": 1518
    },
    {
      "epoch": 0.7295869356388088,
      "grad_norm": 0.4845564720247174,
      "learning_rate": 9.388729721999573e-06,
      "loss": 0.0399,
      "step": 1519
    },
    {
      "epoch": 0.7300672430355427,
      "grad_norm": 0.41693125786890695,
      "learning_rate": 9.387390111317599e-06,
      "loss": 0.0413,
      "step": 1520
    },
    {
      "epoch": 0.7305475504322767,
      "grad_norm": 0.550874709305461,
      "learning_rate": 9.386049130136335e-06,
      "loss": 0.0527,
      "step": 1521
    },
    {
      "epoch": 0.7310278578290106,
      "grad_norm": 0.7548778248070213,
      "learning_rate": 9.384706778874664e-06,
      "loss": 0.0521,
      "step": 1522
    },
    {
      "epoch": 0.7315081652257445,
      "grad_norm": 0.5908031992669268,
      "learning_rate": 9.3833630579519e-06,
      "loss": 0.0407,
      "step": 1523
    },
    {
      "epoch": 0.7319884726224783,
      "grad_norm": 0.6721817113672193,
      "learning_rate": 9.382017967787783e-06,
      "loss": 0.0456,
      "step": 1524
    },
    {
      "epoch": 0.7324687800192123,
      "grad_norm": 0.412185975988658,
      "learning_rate": 9.380671508802483e-06,
      "loss": 0.0392,
      "step": 1525
    },
    {
      "epoch": 0.7329490874159462,
      "grad_norm": 0.46229352306795535,
      "learning_rate": 9.379323681416596e-06,
      "loss": 0.0357,
      "step": 1526
    },
    {
      "epoch": 0.7334293948126801,
      "grad_norm": 0.5289989908411948,
      "learning_rate": 9.377974486051149e-06,
      "loss": 0.0548,
      "step": 1527
    },
    {
      "epoch": 0.733909702209414,
      "grad_norm": 0.5146883009854686,
      "learning_rate": 9.376623923127588e-06,
      "loss": 0.0505,
      "step": 1528
    },
    {
      "epoch": 0.734390009606148,
      "grad_norm": 0.5889961825894928,
      "learning_rate": 9.375271993067794e-06,
      "loss": 0.0442,
      "step": 1529
    },
    {
      "epoch": 0.7348703170028819,
      "grad_norm": 0.492353320981019,
      "learning_rate": 9.373918696294074e-06,
      "loss": 0.045,
      "step": 1530
    },
    {
      "epoch": 0.7353506243996157,
      "grad_norm": 0.39252458109011074,
      "learning_rate": 9.372564033229159e-06,
      "loss": 0.0357,
      "step": 1531
    },
    {
      "epoch": 0.7358309317963496,
      "grad_norm": 0.6428195670677652,
      "learning_rate": 9.371208004296208e-06,
      "loss": 0.0644,
      "step": 1532
    },
    {
      "epoch": 0.7363112391930836,
      "grad_norm": 0.6817577459803879,
      "learning_rate": 9.36985060991881e-06,
      "loss": 0.0804,
      "step": 1533
    },
    {
      "epoch": 0.7367915465898175,
      "grad_norm": 0.7456675871856415,
      "learning_rate": 9.368491850520972e-06,
      "loss": 0.0476,
      "step": 1534
    },
    {
      "epoch": 0.7372718539865514,
      "grad_norm": 0.45828290446709025,
      "learning_rate": 9.367131726527137e-06,
      "loss": 0.0387,
      "step": 1535
    },
    {
      "epoch": 0.7377521613832853,
      "grad_norm": 0.5469514522999743,
      "learning_rate": 9.365770238362168e-06,
      "loss": 0.0401,
      "step": 1536
    },
    {
      "epoch": 0.7382324687800192,
      "grad_norm": 0.6196800460825531,
      "learning_rate": 9.364407386451358e-06,
      "loss": 0.0382,
      "step": 1537
    },
    {
      "epoch": 0.7387127761767531,
      "grad_norm": 0.8185923332401619,
      "learning_rate": 9.363043171220423e-06,
      "loss": 0.0439,
      "step": 1538
    },
    {
      "epoch": 0.739193083573487,
      "grad_norm": 0.5914692890542824,
      "learning_rate": 9.361677593095506e-06,
      "loss": 0.0616,
      "step": 1539
    },
    {
      "epoch": 0.7396733909702209,
      "grad_norm": 0.42464742265876826,
      "learning_rate": 9.360310652503176e-06,
      "loss": 0.0454,
      "step": 1540
    },
    {
      "epoch": 0.7401536983669549,
      "grad_norm": 0.3717412618511177,
      "learning_rate": 9.358942349870428e-06,
      "loss": 0.0335,
      "step": 1541
    },
    {
      "epoch": 0.7406340057636888,
      "grad_norm": 0.47502141185790275,
      "learning_rate": 9.35757268562468e-06,
      "loss": 0.0488,
      "step": 1542
    },
    {
      "epoch": 0.7411143131604226,
      "grad_norm": 0.3846823017408531,
      "learning_rate": 9.356201660193782e-06,
      "loss": 0.0345,
      "step": 1543
    },
    {
      "epoch": 0.7415946205571565,
      "grad_norm": 0.5650933228577771,
      "learning_rate": 9.354829274005999e-06,
      "loss": 0.0383,
      "step": 1544
    },
    {
      "epoch": 0.7420749279538905,
      "grad_norm": 0.5690246647765329,
      "learning_rate": 9.35345552749003e-06,
      "loss": 0.0569,
      "step": 1545
    },
    {
      "epoch": 0.7425552353506244,
      "grad_norm": 0.38731794835615024,
      "learning_rate": 9.352080421074993e-06,
      "loss": 0.0419,
      "step": 1546
    },
    {
      "epoch": 0.7430355427473583,
      "grad_norm": 0.40974758691143354,
      "learning_rate": 9.350703955190435e-06,
      "loss": 0.0376,
      "step": 1547
    },
    {
      "epoch": 0.7435158501440923,
      "grad_norm": 0.46283701258349547,
      "learning_rate": 9.349326130266323e-06,
      "loss": 0.0328,
      "step": 1548
    },
    {
      "epoch": 0.7439961575408262,
      "grad_norm": 0.40124018251646276,
      "learning_rate": 9.347946946733055e-06,
      "loss": 0.0387,
      "step": 1549
    },
    {
      "epoch": 0.74447646493756,
      "grad_norm": 0.4121164368713994,
      "learning_rate": 9.346566405021448e-06,
      "loss": 0.0387,
      "step": 1550
    },
    {
      "epoch": 0.7449567723342939,
      "grad_norm": 0.48335845270026145,
      "learning_rate": 9.345184505562747e-06,
      "loss": 0.0368,
      "step": 1551
    },
    {
      "epoch": 0.7454370797310279,
      "grad_norm": 0.45269768346922323,
      "learning_rate": 9.343801248788616e-06,
      "loss": 0.052,
      "step": 1552
    },
    {
      "epoch": 0.7459173871277618,
      "grad_norm": 0.4610035020071121,
      "learning_rate": 9.342416635131148e-06,
      "loss": 0.0614,
      "step": 1553
    },
    {
      "epoch": 0.7463976945244957,
      "grad_norm": 0.4443227466310018,
      "learning_rate": 9.341030665022856e-06,
      "loss": 0.0447,
      "step": 1554
    },
    {
      "epoch": 0.7468780019212296,
      "grad_norm": 0.7642793611261786,
      "learning_rate": 9.339643338896682e-06,
      "loss": 0.0635,
      "step": 1555
    },
    {
      "epoch": 0.7473583093179635,
      "grad_norm": 0.4923735731312003,
      "learning_rate": 9.338254657185988e-06,
      "loss": 0.0507,
      "step": 1556
    },
    {
      "epoch": 0.7478386167146974,
      "grad_norm": 0.630683703114263,
      "learning_rate": 9.336864620324555e-06,
      "loss": 0.0621,
      "step": 1557
    },
    {
      "epoch": 0.7483189241114313,
      "grad_norm": 0.5090396267143328,
      "learning_rate": 9.335473228746599e-06,
      "loss": 0.0406,
      "step": 1558
    },
    {
      "epoch": 0.7487992315081652,
      "grad_norm": 0.367559997986288,
      "learning_rate": 9.334080482886746e-06,
      "loss": 0.0384,
      "step": 1559
    },
    {
      "epoch": 0.7492795389048992,
      "grad_norm": 0.4446659468934462,
      "learning_rate": 9.332686383180055e-06,
      "loss": 0.0361,
      "step": 1560
    },
    {
      "epoch": 0.7497598463016331,
      "grad_norm": 0.38856445414113083,
      "learning_rate": 9.331290930062003e-06,
      "loss": 0.0259,
      "step": 1561
    },
    {
      "epoch": 0.7502401536983669,
      "grad_norm": 0.5997908771571422,
      "learning_rate": 9.32989412396849e-06,
      "loss": 0.0471,
      "step": 1562
    },
    {
      "epoch": 0.7507204610951008,
      "grad_norm": 0.3393607659537801,
      "learning_rate": 9.328495965335844e-06,
      "loss": 0.0394,
      "step": 1563
    },
    {
      "epoch": 0.7512007684918348,
      "grad_norm": 0.37942787002008155,
      "learning_rate": 9.327096454600804e-06,
      "loss": 0.0273,
      "step": 1564
    },
    {
      "epoch": 0.7516810758885687,
      "grad_norm": 0.5089526163454388,
      "learning_rate": 9.325695592200545e-06,
      "loss": 0.0632,
      "step": 1565
    },
    {
      "epoch": 0.7521613832853026,
      "grad_norm": 0.5121396909775255,
      "learning_rate": 9.324293378572654e-06,
      "loss": 0.0551,
      "step": 1566
    },
    {
      "epoch": 0.7526416906820365,
      "grad_norm": 0.38135878200286283,
      "learning_rate": 9.322889814155143e-06,
      "loss": 0.0345,
      "step": 1567
    },
    {
      "epoch": 0.7531219980787704,
      "grad_norm": 0.6071199641753309,
      "learning_rate": 9.32148489938645e-06,
      "loss": 0.0424,
      "step": 1568
    },
    {
      "epoch": 0.7536023054755043,
      "grad_norm": 0.6157559078057468,
      "learning_rate": 9.32007863470543e-06,
      "loss": 0.045,
      "step": 1569
    },
    {
      "epoch": 0.7540826128722382,
      "grad_norm": 0.48273199109262704,
      "learning_rate": 9.31867102055136e-06,
      "loss": 0.0406,
      "step": 1570
    },
    {
      "epoch": 0.7545629202689721,
      "grad_norm": 0.4593296987544262,
      "learning_rate": 9.31726205736394e-06,
      "loss": 0.0436,
      "step": 1571
    },
    {
      "epoch": 0.7550432276657061,
      "grad_norm": 0.4500045712808899,
      "learning_rate": 9.315851745583294e-06,
      "loss": 0.0269,
      "step": 1572
    },
    {
      "epoch": 0.75552353506244,
      "grad_norm": 0.4454538603150306,
      "learning_rate": 9.31444008564996e-06,
      "loss": 0.0366,
      "step": 1573
    },
    {
      "epoch": 0.7560038424591738,
      "grad_norm": 0.8394007155144771,
      "learning_rate": 9.313027078004903e-06,
      "loss": 0.0503,
      "step": 1574
    },
    {
      "epoch": 0.7564841498559077,
      "grad_norm": 0.5712174753843666,
      "learning_rate": 9.311612723089511e-06,
      "loss": 0.0547,
      "step": 1575
    },
    {
      "epoch": 0.7569644572526417,
      "grad_norm": 0.46645965469348327,
      "learning_rate": 9.310197021345586e-06,
      "loss": 0.0374,
      "step": 1576
    },
    {
      "epoch": 0.7574447646493756,
      "grad_norm": 0.4740884325639232,
      "learning_rate": 9.308779973215355e-06,
      "loss": 0.0548,
      "step": 1577
    },
    {
      "epoch": 0.7579250720461095,
      "grad_norm": 0.5888223818640053,
      "learning_rate": 9.307361579141461e-06,
      "loss": 0.0581,
      "step": 1578
    },
    {
      "epoch": 0.7584053794428435,
      "grad_norm": 0.5714740123256156,
      "learning_rate": 9.305941839566978e-06,
      "loss": 0.0479,
      "step": 1579
    },
    {
      "epoch": 0.7588856868395774,
      "grad_norm": 0.6224308689788665,
      "learning_rate": 9.304520754935387e-06,
      "loss": 0.0418,
      "step": 1580
    },
    {
      "epoch": 0.7593659942363112,
      "grad_norm": 0.5273832414104769,
      "learning_rate": 9.303098325690601e-06,
      "loss": 0.0534,
      "step": 1581
    },
    {
      "epoch": 0.7598463016330451,
      "grad_norm": 0.4100845821662855,
      "learning_rate": 9.301674552276942e-06,
      "loss": 0.0393,
      "step": 1582
    },
    {
      "epoch": 0.7603266090297791,
      "grad_norm": 0.5233760618757851,
      "learning_rate": 9.300249435139162e-06,
      "loss": 0.0567,
      "step": 1583
    },
    {
      "epoch": 0.760806916426513,
      "grad_norm": 0.4470067504003617,
      "learning_rate": 9.298822974722425e-06,
      "loss": 0.039,
      "step": 1584
    },
    {
      "epoch": 0.7612872238232469,
      "grad_norm": 0.6788279010789742,
      "learning_rate": 9.297395171472321e-06,
      "loss": 0.045,
      "step": 1585
    },
    {
      "epoch": 0.7617675312199808,
      "grad_norm": 0.5433301733942123,
      "learning_rate": 9.295966025834853e-06,
      "loss": 0.0564,
      "step": 1586
    },
    {
      "epoch": 0.7622478386167147,
      "grad_norm": 0.49450333266253477,
      "learning_rate": 9.294535538256447e-06,
      "loss": 0.0531,
      "step": 1587
    },
    {
      "epoch": 0.7627281460134486,
      "grad_norm": 0.5471927312457198,
      "learning_rate": 9.29310370918395e-06,
      "loss": 0.0669,
      "step": 1588
    },
    {
      "epoch": 0.7632084534101825,
      "grad_norm": 0.4715635212373278,
      "learning_rate": 9.291670539064623e-06,
      "loss": 0.0405,
      "step": 1589
    },
    {
      "epoch": 0.7636887608069164,
      "grad_norm": 0.4111372690921772,
      "learning_rate": 9.290236028346152e-06,
      "loss": 0.0379,
      "step": 1590
    },
    {
      "epoch": 0.7641690682036504,
      "grad_norm": 0.7088243704037053,
      "learning_rate": 9.288800177476636e-06,
      "loss": 0.049,
      "step": 1591
    },
    {
      "epoch": 0.7646493756003843,
      "grad_norm": 0.38910744032185673,
      "learning_rate": 9.287362986904595e-06,
      "loss": 0.0377,
      "step": 1592
    },
    {
      "epoch": 0.7651296829971181,
      "grad_norm": 0.47900015103136745,
      "learning_rate": 9.28592445707897e-06,
      "loss": 0.0406,
      "step": 1593
    },
    {
      "epoch": 0.765609990393852,
      "grad_norm": 1.0851740572976578,
      "learning_rate": 9.284484588449115e-06,
      "loss": 0.0444,
      "step": 1594
    },
    {
      "epoch": 0.766090297790586,
      "grad_norm": 0.5147230458014541,
      "learning_rate": 9.283043381464806e-06,
      "loss": 0.0387,
      "step": 1595
    },
    {
      "epoch": 0.7665706051873199,
      "grad_norm": 0.46123024101584026,
      "learning_rate": 9.281600836576237e-06,
      "loss": 0.0451,
      "step": 1596
    },
    {
      "epoch": 0.7670509125840538,
      "grad_norm": 0.4441738388860228,
      "learning_rate": 9.280156954234017e-06,
      "loss": 0.0388,
      "step": 1597
    },
    {
      "epoch": 0.7675312199807877,
      "grad_norm": 0.5563354966449449,
      "learning_rate": 9.278711734889178e-06,
      "loss": 0.0407,
      "step": 1598
    },
    {
      "epoch": 0.7680115273775217,
      "grad_norm": 0.6141961922794104,
      "learning_rate": 9.277265178993164e-06,
      "loss": 0.0398,
      "step": 1599
    },
    {
      "epoch": 0.7684918347742555,
      "grad_norm": 0.5225559365038454,
      "learning_rate": 9.275817286997843e-06,
      "loss": 0.0371,
      "step": 1600
    },
    {
      "epoch": 0.7689721421709894,
      "grad_norm": 0.5371833677807129,
      "learning_rate": 9.274368059355491e-06,
      "loss": 0.0369,
      "step": 1601
    },
    {
      "epoch": 0.7694524495677233,
      "grad_norm": 0.4509358847362159,
      "learning_rate": 9.27291749651881e-06,
      "loss": 0.0348,
      "step": 1602
    },
    {
      "epoch": 0.7699327569644573,
      "grad_norm": 0.3790301116583118,
      "learning_rate": 9.271465598940914e-06,
      "loss": 0.0341,
      "step": 1603
    },
    {
      "epoch": 0.7704130643611912,
      "grad_norm": 0.5646089029750233,
      "learning_rate": 9.270012367075337e-06,
      "loss": 0.0458,
      "step": 1604
    },
    {
      "epoch": 0.770893371757925,
      "grad_norm": 0.6621291150481707,
      "learning_rate": 9.268557801376027e-06,
      "loss": 0.0673,
      "step": 1605
    },
    {
      "epoch": 0.7713736791546589,
      "grad_norm": 0.4824676654847145,
      "learning_rate": 9.267101902297354e-06,
      "loss": 0.0605,
      "step": 1606
    },
    {
      "epoch": 0.7718539865513929,
      "grad_norm": 0.47867737571924573,
      "learning_rate": 9.265644670294094e-06,
      "loss": 0.0402,
      "step": 1607
    },
    {
      "epoch": 0.7723342939481268,
      "grad_norm": 0.7521404239285508,
      "learning_rate": 9.26418610582145e-06,
      "loss": 0.0437,
      "step": 1608
    },
    {
      "epoch": 0.7728146013448607,
      "grad_norm": 0.4949872678032156,
      "learning_rate": 9.262726209335038e-06,
      "loss": 0.0471,
      "step": 1609
    },
    {
      "epoch": 0.7732949087415946,
      "grad_norm": 0.8146436171510608,
      "learning_rate": 9.261264981290887e-06,
      "loss": 0.0499,
      "step": 1610
    },
    {
      "epoch": 0.7737752161383286,
      "grad_norm": 0.5261485997463431,
      "learning_rate": 9.259802422145445e-06,
      "loss": 0.0499,
      "step": 1611
    },
    {
      "epoch": 0.7742555235350624,
      "grad_norm": 0.37722938922014426,
      "learning_rate": 9.258338532355575e-06,
      "loss": 0.0406,
      "step": 1612
    },
    {
      "epoch": 0.7747358309317963,
      "grad_norm": 0.5137949267423024,
      "learning_rate": 9.256873312378559e-06,
      "loss": 0.0371,
      "step": 1613
    },
    {
      "epoch": 0.7752161383285303,
      "grad_norm": 0.46384787647828507,
      "learning_rate": 9.255406762672085e-06,
      "loss": 0.0436,
      "step": 1614
    },
    {
      "epoch": 0.7756964457252642,
      "grad_norm": 0.42501919449374165,
      "learning_rate": 9.253938883694266e-06,
      "loss": 0.0319,
      "step": 1615
    },
    {
      "epoch": 0.7761767531219981,
      "grad_norm": 0.48825545396901937,
      "learning_rate": 9.252469675903627e-06,
      "loss": 0.0438,
      "step": 1616
    },
    {
      "epoch": 0.776657060518732,
      "grad_norm": 0.45045012320002853,
      "learning_rate": 9.250999139759107e-06,
      "loss": 0.0507,
      "step": 1617
    },
    {
      "epoch": 0.777137367915466,
      "grad_norm": 0.47986993527995575,
      "learning_rate": 9.249527275720062e-06,
      "loss": 0.0438,
      "step": 1618
    },
    {
      "epoch": 0.7776176753121998,
      "grad_norm": 0.6897438756676713,
      "learning_rate": 9.248054084246259e-06,
      "loss": 0.0711,
      "step": 1619
    },
    {
      "epoch": 0.7780979827089337,
      "grad_norm": 0.33151049726562337,
      "learning_rate": 9.246579565797886e-06,
      "loss": 0.0357,
      "step": 1620
    },
    {
      "epoch": 0.7785782901056676,
      "grad_norm": 0.43010301611246965,
      "learning_rate": 9.245103720835538e-06,
      "loss": 0.0451,
      "step": 1621
    },
    {
      "epoch": 0.7790585975024016,
      "grad_norm": 0.40143671990795116,
      "learning_rate": 9.24362654982023e-06,
      "loss": 0.0389,
      "step": 1622
    },
    {
      "epoch": 0.7795389048991355,
      "grad_norm": 0.4379641653718776,
      "learning_rate": 9.24214805321339e-06,
      "loss": 0.0422,
      "step": 1623
    },
    {
      "epoch": 0.7800192122958693,
      "grad_norm": 0.7421035915776883,
      "learning_rate": 9.24066823147686e-06,
      "loss": 0.0477,
      "step": 1624
    },
    {
      "epoch": 0.7804995196926032,
      "grad_norm": 0.6044451650392826,
      "learning_rate": 9.239187085072891e-06,
      "loss": 0.0482,
      "step": 1625
    },
    {
      "epoch": 0.7809798270893372,
      "grad_norm": 0.5596451678457982,
      "learning_rate": 9.237704614464157e-06,
      "loss": 0.0418,
      "step": 1626
    },
    {
      "epoch": 0.7814601344860711,
      "grad_norm": 0.6592034808792037,
      "learning_rate": 9.236220820113738e-06,
      "loss": 0.0546,
      "step": 1627
    },
    {
      "epoch": 0.781940441882805,
      "grad_norm": 0.5465739249129324,
      "learning_rate": 9.234735702485132e-06,
      "loss": 0.0442,
      "step": 1628
    },
    {
      "epoch": 0.7824207492795389,
      "grad_norm": 0.4911854758209489,
      "learning_rate": 9.233249262042247e-06,
      "loss": 0.0398,
      "step": 1629
    },
    {
      "epoch": 0.7829010566762729,
      "grad_norm": 0.5748292545758572,
      "learning_rate": 9.231761499249407e-06,
      "loss": 0.0508,
      "step": 1630
    },
    {
      "epoch": 0.7833813640730067,
      "grad_norm": 0.6346041474004851,
      "learning_rate": 9.230272414571349e-06,
      "loss": 0.0439,
      "step": 1631
    },
    {
      "epoch": 0.7838616714697406,
      "grad_norm": 0.5634165757189631,
      "learning_rate": 9.22878200847322e-06,
      "loss": 0.0539,
      "step": 1632
    },
    {
      "epoch": 0.7843419788664745,
      "grad_norm": 0.5027607179661531,
      "learning_rate": 9.227290281420583e-06,
      "loss": 0.045,
      "step": 1633
    },
    {
      "epoch": 0.7848222862632085,
      "grad_norm": 0.3037912293882595,
      "learning_rate": 9.22579723387941e-06,
      "loss": 0.0369,
      "step": 1634
    },
    {
      "epoch": 0.7853025936599424,
      "grad_norm": 0.8410699245876824,
      "learning_rate": 9.22430286631609e-06,
      "loss": 0.0356,
      "step": 1635
    },
    {
      "epoch": 0.7857829010566763,
      "grad_norm": 0.5368787438929216,
      "learning_rate": 9.222807179197421e-06,
      "loss": 0.043,
      "step": 1636
    },
    {
      "epoch": 0.7862632084534101,
      "grad_norm": 0.4032303101187688,
      "learning_rate": 9.221310172990616e-06,
      "loss": 0.0368,
      "step": 1637
    },
    {
      "epoch": 0.7867435158501441,
      "grad_norm": 0.5409648779344621,
      "learning_rate": 9.219811848163295e-06,
      "loss": 0.0629,
      "step": 1638
    },
    {
      "epoch": 0.787223823246878,
      "grad_norm": 0.6171811747268313,
      "learning_rate": 9.218312205183497e-06,
      "loss": 0.0531,
      "step": 1639
    },
    {
      "epoch": 0.7877041306436119,
      "grad_norm": 0.4216070952232712,
      "learning_rate": 9.216811244519667e-06,
      "loss": 0.0449,
      "step": 1640
    },
    {
      "epoch": 0.7881844380403458,
      "grad_norm": 0.5763421855553458,
      "learning_rate": 9.215308966640662e-06,
      "loss": 0.0368,
      "step": 1641
    },
    {
      "epoch": 0.7886647454370798,
      "grad_norm": 0.4870149354831771,
      "learning_rate": 9.213805372015756e-06,
      "loss": 0.0451,
      "step": 1642
    },
    {
      "epoch": 0.7891450528338136,
      "grad_norm": 0.393109877651519,
      "learning_rate": 9.212300461114626e-06,
      "loss": 0.0335,
      "step": 1643
    },
    {
      "epoch": 0.7896253602305475,
      "grad_norm": 0.5879082191240319,
      "learning_rate": 9.210794234407368e-06,
      "loss": 0.0451,
      "step": 1644
    },
    {
      "epoch": 0.7901056676272814,
      "grad_norm": 0.5295234190274382,
      "learning_rate": 9.209286692364484e-06,
      "loss": 0.061,
      "step": 1645
    },
    {
      "epoch": 0.7905859750240154,
      "grad_norm": 0.9676479560008302,
      "learning_rate": 9.207777835456887e-06,
      "loss": 0.0593,
      "step": 1646
    },
    {
      "epoch": 0.7910662824207493,
      "grad_norm": 0.41367077682405545,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.0286,
      "step": 1647
    },
    {
      "epoch": 0.7915465898174832,
      "grad_norm": 0.3879540531683163,
      "learning_rate": 9.204756178933274e-06,
      "loss": 0.0377,
      "step": 1648
    },
    {
      "epoch": 0.7920268972142172,
      "grad_norm": 0.4431616617976934,
      "learning_rate": 9.203243380261138e-06,
      "loss": 0.0558,
      "step": 1649
    },
    {
      "epoch": 0.792507204610951,
      "grad_norm": 0.3784767168841202,
      "learning_rate": 9.201729268612054e-06,
      "loss": 0.0539,
      "step": 1650
    },
    {
      "epoch": 0.7929875120076849,
      "grad_norm": 0.45445078560396285,
      "learning_rate": 9.20021384445899e-06,
      "loss": 0.0406,
      "step": 1651
    },
    {
      "epoch": 0.7934678194044188,
      "grad_norm": 0.40977728429479643,
      "learning_rate": 9.198697108275318e-06,
      "loss": 0.0337,
      "step": 1652
    },
    {
      "epoch": 0.7939481268011528,
      "grad_norm": 0.36463474145779456,
      "learning_rate": 9.19717906053483e-06,
      "loss": 0.0377,
      "step": 1653
    },
    {
      "epoch": 0.7944284341978867,
      "grad_norm": 0.36067865022176565,
      "learning_rate": 9.19565970171172e-06,
      "loss": 0.032,
      "step": 1654
    },
    {
      "epoch": 0.7949087415946205,
      "grad_norm": 0.5087941975761743,
      "learning_rate": 9.194139032280594e-06,
      "loss": 0.0363,
      "step": 1655
    },
    {
      "epoch": 0.7953890489913544,
      "grad_norm": 0.48198473655680424,
      "learning_rate": 9.192617052716463e-06,
      "loss": 0.0458,
      "step": 1656
    },
    {
      "epoch": 0.7958693563880884,
      "grad_norm": 0.5456323999363246,
      "learning_rate": 9.19109376349476e-06,
      "loss": 0.0546,
      "step": 1657
    },
    {
      "epoch": 0.7963496637848223,
      "grad_norm": 0.4698008133059572,
      "learning_rate": 9.18956916509131e-06,
      "loss": 0.041,
      "step": 1658
    },
    {
      "epoch": 0.7968299711815562,
      "grad_norm": 0.48858912139881877,
      "learning_rate": 9.18804325798236e-06,
      "loss": 0.0482,
      "step": 1659
    },
    {
      "epoch": 0.7973102785782901,
      "grad_norm": 0.45224694401197335,
      "learning_rate": 9.18651604264456e-06,
      "loss": 0.0432,
      "step": 1660
    },
    {
      "epoch": 0.7977905859750241,
      "grad_norm": 0.5291111116543427,
      "learning_rate": 9.184987519554969e-06,
      "loss": 0.0392,
      "step": 1661
    },
    {
      "epoch": 0.7982708933717579,
      "grad_norm": 0.39953149277001193,
      "learning_rate": 9.183457689191055e-06,
      "loss": 0.0392,
      "step": 1662
    },
    {
      "epoch": 0.7987512007684918,
      "grad_norm": 0.49701247579632074,
      "learning_rate": 9.181926552030698e-06,
      "loss": 0.0374,
      "step": 1663
    },
    {
      "epoch": 0.7992315081652257,
      "grad_norm": 0.38365739480522254,
      "learning_rate": 9.18039410855218e-06,
      "loss": 0.043,
      "step": 1664
    },
    {
      "epoch": 0.7997118155619597,
      "grad_norm": 0.37429467352808293,
      "learning_rate": 9.178860359234193e-06,
      "loss": 0.0407,
      "step": 1665
    },
    {
      "epoch": 0.8001921229586936,
      "grad_norm": 0.4091686599510648,
      "learning_rate": 9.17732530455584e-06,
      "loss": 0.038,
      "step": 1666
    },
    {
      "epoch": 0.8006724303554275,
      "grad_norm": 0.7195325093951213,
      "learning_rate": 9.175788944996629e-06,
      "loss": 0.0741,
      "step": 1667
    },
    {
      "epoch": 0.8011527377521613,
      "grad_norm": 0.3820712024818586,
      "learning_rate": 9.174251281036478e-06,
      "loss": 0.0516,
      "step": 1668
    },
    {
      "epoch": 0.8016330451488953,
      "grad_norm": 0.6407457249834841,
      "learning_rate": 9.172712313155708e-06,
      "loss": 0.0434,
      "step": 1669
    },
    {
      "epoch": 0.8021133525456292,
      "grad_norm": 0.4969915345518172,
      "learning_rate": 9.171172041835048e-06,
      "loss": 0.0385,
      "step": 1670
    },
    {
      "epoch": 0.8025936599423631,
      "grad_norm": 0.8727857658628531,
      "learning_rate": 9.169630467555643e-06,
      "loss": 0.0548,
      "step": 1671
    },
    {
      "epoch": 0.803073967339097,
      "grad_norm": 0.5478163807092746,
      "learning_rate": 9.168087590799034e-06,
      "loss": 0.0417,
      "step": 1672
    },
    {
      "epoch": 0.803554274735831,
      "grad_norm": 0.5827507091175965,
      "learning_rate": 9.16654341204717e-06,
      "loss": 0.0568,
      "step": 1673
    },
    {
      "epoch": 0.8040345821325648,
      "grad_norm": 0.5177032542659947,
      "learning_rate": 9.164997931782415e-06,
      "loss": 0.0492,
      "step": 1674
    },
    {
      "epoch": 0.8045148895292987,
      "grad_norm": 0.4309298356028421,
      "learning_rate": 9.163451150487531e-06,
      "loss": 0.0454,
      "step": 1675
    },
    {
      "epoch": 0.8049951969260326,
      "grad_norm": 0.4057663676388221,
      "learning_rate": 9.161903068645692e-06,
      "loss": 0.0422,
      "step": 1676
    },
    {
      "epoch": 0.8054755043227666,
      "grad_norm": 0.47646062877439854,
      "learning_rate": 9.160353686740476e-06,
      "loss": 0.0351,
      "step": 1677
    },
    {
      "epoch": 0.8059558117195005,
      "grad_norm": 0.45558821865778054,
      "learning_rate": 9.158803005255862e-06,
      "loss": 0.0416,
      "step": 1678
    },
    {
      "epoch": 0.8064361191162344,
      "grad_norm": 0.536210255513318,
      "learning_rate": 9.157251024676249e-06,
      "loss": 0.0499,
      "step": 1679
    },
    {
      "epoch": 0.8069164265129684,
      "grad_norm": 0.6111498216460901,
      "learning_rate": 9.155697745486424e-06,
      "loss": 0.0342,
      "step": 1680
    },
    {
      "epoch": 0.8073967339097022,
      "grad_norm": 0.4833245316704844,
      "learning_rate": 9.154143168171594e-06,
      "loss": 0.0447,
      "step": 1681
    },
    {
      "epoch": 0.8078770413064361,
      "grad_norm": 0.7125662823152267,
      "learning_rate": 9.152587293217362e-06,
      "loss": 0.0475,
      "step": 1682
    },
    {
      "epoch": 0.80835734870317,
      "grad_norm": 0.42403591379119365,
      "learning_rate": 9.151030121109745e-06,
      "loss": 0.0391,
      "step": 1683
    },
    {
      "epoch": 0.808837656099904,
      "grad_norm": 0.5205159350136663,
      "learning_rate": 9.149471652335155e-06,
      "loss": 0.0512,
      "step": 1684
    },
    {
      "epoch": 0.8093179634966379,
      "grad_norm": 0.4957469692515393,
      "learning_rate": 9.14791188738042e-06,
      "loss": 0.0406,
      "step": 1685
    },
    {
      "epoch": 0.8097982708933718,
      "grad_norm": 0.7665309194587624,
      "learning_rate": 9.146350826732762e-06,
      "loss": 0.0549,
      "step": 1686
    },
    {
      "epoch": 0.8102785782901056,
      "grad_norm": 0.43844120238867723,
      "learning_rate": 9.144788470879818e-06,
      "loss": 0.0429,
      "step": 1687
    },
    {
      "epoch": 0.8107588856868396,
      "grad_norm": 0.5211050949999975,
      "learning_rate": 9.143224820309622e-06,
      "loss": 0.0479,
      "step": 1688
    },
    {
      "epoch": 0.8112391930835735,
      "grad_norm": 0.45269831596656984,
      "learning_rate": 9.141659875510615e-06,
      "loss": 0.0527,
      "step": 1689
    },
    {
      "epoch": 0.8117195004803074,
      "grad_norm": 0.4724243903083948,
      "learning_rate": 9.140093636971646e-06,
      "loss": 0.0353,
      "step": 1690
    },
    {
      "epoch": 0.8121998078770413,
      "grad_norm": 0.6109510239550704,
      "learning_rate": 9.13852610518196e-06,
      "loss": 0.0466,
      "step": 1691
    },
    {
      "epoch": 0.8126801152737753,
      "grad_norm": 0.41144558793142044,
      "learning_rate": 9.136957280631212e-06,
      "loss": 0.0363,
      "step": 1692
    },
    {
      "epoch": 0.8131604226705091,
      "grad_norm": 0.4886106331868101,
      "learning_rate": 9.135387163809462e-06,
      "loss": 0.0442,
      "step": 1693
    },
    {
      "epoch": 0.813640730067243,
      "grad_norm": 0.486941943939095,
      "learning_rate": 9.133815755207168e-06,
      "loss": 0.0533,
      "step": 1694
    },
    {
      "epoch": 0.8141210374639769,
      "grad_norm": 0.34872099800574413,
      "learning_rate": 9.132243055315193e-06,
      "loss": 0.0314,
      "step": 1695
    },
    {
      "epoch": 0.8146013448607109,
      "grad_norm": 0.44324684859322067,
      "learning_rate": 9.130669064624811e-06,
      "loss": 0.0532,
      "step": 1696
    },
    {
      "epoch": 0.8150816522574448,
      "grad_norm": 0.44545944868455956,
      "learning_rate": 9.129093783627687e-06,
      "loss": 0.037,
      "step": 1697
    },
    {
      "epoch": 0.8155619596541787,
      "grad_norm": 0.45798246599733317,
      "learning_rate": 9.1275172128159e-06,
      "loss": 0.0481,
      "step": 1698
    },
    {
      "epoch": 0.8160422670509125,
      "grad_norm": 0.44349109742245846,
      "learning_rate": 9.125939352681922e-06,
      "loss": 0.0517,
      "step": 1699
    },
    {
      "epoch": 0.8165225744476465,
      "grad_norm": 0.6001166233671226,
      "learning_rate": 9.124360203718638e-06,
      "loss": 0.0461,
      "step": 1700
    },
    {
      "epoch": 0.8170028818443804,
      "grad_norm": 0.6831035265294242,
      "learning_rate": 9.122779766419329e-06,
      "loss": 0.0591,
      "step": 1701
    },
    {
      "epoch": 0.8174831892411143,
      "grad_norm": 0.38077445809771976,
      "learning_rate": 9.121198041277677e-06,
      "loss": 0.0329,
      "step": 1702
    },
    {
      "epoch": 0.8179634966378482,
      "grad_norm": 0.41140076728447145,
      "learning_rate": 9.119615028787771e-06,
      "loss": 0.0485,
      "step": 1703
    },
    {
      "epoch": 0.8184438040345822,
      "grad_norm": 0.5189348552493814,
      "learning_rate": 9.118030729444103e-06,
      "loss": 0.0448,
      "step": 1704
    },
    {
      "epoch": 0.818924111431316,
      "grad_norm": 0.465577166008072,
      "learning_rate": 9.11644514374156e-06,
      "loss": 0.0514,
      "step": 1705
    },
    {
      "epoch": 0.8194044188280499,
      "grad_norm": 0.5621851173772174,
      "learning_rate": 9.114858272175438e-06,
      "loss": 0.0392,
      "step": 1706
    },
    {
      "epoch": 0.8198847262247838,
      "grad_norm": 0.6859692616790941,
      "learning_rate": 9.113270115241429e-06,
      "loss": 0.0705,
      "step": 1707
    },
    {
      "epoch": 0.8203650336215178,
      "grad_norm": 0.47502833358038377,
      "learning_rate": 9.111680673435632e-06,
      "loss": 0.0441,
      "step": 1708
    },
    {
      "epoch": 0.8208453410182517,
      "grad_norm": 0.46706802163027805,
      "learning_rate": 9.110089947254544e-06,
      "loss": 0.0391,
      "step": 1709
    },
    {
      "epoch": 0.8213256484149856,
      "grad_norm": 0.31719462369803303,
      "learning_rate": 9.108497937195064e-06,
      "loss": 0.0289,
      "step": 1710
    },
    {
      "epoch": 0.8218059558117194,
      "grad_norm": 0.40385158643378627,
      "learning_rate": 9.106904643754491e-06,
      "loss": 0.0381,
      "step": 1711
    },
    {
      "epoch": 0.8222862632084534,
      "grad_norm": 0.5545734057584185,
      "learning_rate": 9.105310067430526e-06,
      "loss": 0.0471,
      "step": 1712
    },
    {
      "epoch": 0.8227665706051873,
      "grad_norm": 0.4461054981605916,
      "learning_rate": 9.10371420872127e-06,
      "loss": 0.0508,
      "step": 1713
    },
    {
      "epoch": 0.8232468780019212,
      "grad_norm": 0.47654544400356025,
      "learning_rate": 9.102117068125227e-06,
      "loss": 0.0475,
      "step": 1714
    },
    {
      "epoch": 0.8237271853986552,
      "grad_norm": 0.39911477246208293,
      "learning_rate": 9.100518646141299e-06,
      "loss": 0.0404,
      "step": 1715
    },
    {
      "epoch": 0.8242074927953891,
      "grad_norm": 0.7172841710639825,
      "learning_rate": 9.098918943268786e-06,
      "loss": 0.046,
      "step": 1716
    },
    {
      "epoch": 0.824687800192123,
      "grad_norm": 0.43088188833750957,
      "learning_rate": 9.097317960007395e-06,
      "loss": 0.0403,
      "step": 1717
    },
    {
      "epoch": 0.8251681075888568,
      "grad_norm": 0.5012218482080577,
      "learning_rate": 9.095715696857225e-06,
      "loss": 0.045,
      "step": 1718
    },
    {
      "epoch": 0.8256484149855908,
      "grad_norm": 0.46478416389541866,
      "learning_rate": 9.094112154318784e-06,
      "loss": 0.033,
      "step": 1719
    },
    {
      "epoch": 0.8261287223823247,
      "grad_norm": 0.4130084158371063,
      "learning_rate": 9.092507332892968e-06,
      "loss": 0.0348,
      "step": 1720
    },
    {
      "epoch": 0.8266090297790586,
      "grad_norm": 0.6684505123844774,
      "learning_rate": 9.090901233081082e-06,
      "loss": 0.0629,
      "step": 1721
    },
    {
      "epoch": 0.8270893371757925,
      "grad_norm": 0.4323599984449867,
      "learning_rate": 9.089293855384828e-06,
      "loss": 0.0437,
      "step": 1722
    },
    {
      "epoch": 0.8275696445725265,
      "grad_norm": 0.32774506981480916,
      "learning_rate": 9.087685200306306e-06,
      "loss": 0.0334,
      "step": 1723
    },
    {
      "epoch": 0.8280499519692603,
      "grad_norm": 0.5191345063331478,
      "learning_rate": 9.086075268348014e-06,
      "loss": 0.0433,
      "step": 1724
    },
    {
      "epoch": 0.8285302593659942,
      "grad_norm": 0.33509417066826364,
      "learning_rate": 9.084464060012849e-06,
      "loss": 0.0402,
      "step": 1725
    },
    {
      "epoch": 0.8290105667627281,
      "grad_norm": 0.3142718339340029,
      "learning_rate": 9.082851575804112e-06,
      "loss": 0.0331,
      "step": 1726
    },
    {
      "epoch": 0.8294908741594621,
      "grad_norm": 0.4748408271731178,
      "learning_rate": 9.081237816225497e-06,
      "loss": 0.0516,
      "step": 1727
    },
    {
      "epoch": 0.829971181556196,
      "grad_norm": 0.37444079243147355,
      "learning_rate": 9.079622781781094e-06,
      "loss": 0.037,
      "step": 1728
    },
    {
      "epoch": 0.8304514889529299,
      "grad_norm": 0.5592237440039568,
      "learning_rate": 9.0780064729754e-06,
      "loss": 0.0452,
      "step": 1729
    },
    {
      "epoch": 0.8309317963496637,
      "grad_norm": 0.5481405856733301,
      "learning_rate": 9.076388890313304e-06,
      "loss": 0.0446,
      "step": 1730
    },
    {
      "epoch": 0.8314121037463977,
      "grad_norm": 0.3932969307248321,
      "learning_rate": 9.07477003430009e-06,
      "loss": 0.0488,
      "step": 1731
    },
    {
      "epoch": 0.8318924111431316,
      "grad_norm": 0.3702785209776025,
      "learning_rate": 9.073149905441451e-06,
      "loss": 0.0334,
      "step": 1732
    },
    {
      "epoch": 0.8323727185398655,
      "grad_norm": 1.4773631914590435,
      "learning_rate": 9.071528504243465e-06,
      "loss": 0.0347,
      "step": 1733
    },
    {
      "epoch": 0.8328530259365994,
      "grad_norm": 0.4936274544739719,
      "learning_rate": 9.069905831212616e-06,
      "loss": 0.081,
      "step": 1734
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.5882359938166744,
      "learning_rate": 9.068281886855778e-06,
      "loss": 0.0437,
      "step": 1735
    },
    {
      "epoch": 0.8338136407300673,
      "grad_norm": 0.46989224834772747,
      "learning_rate": 9.066656671680231e-06,
      "loss": 0.0439,
      "step": 1736
    },
    {
      "epoch": 0.8342939481268011,
      "grad_norm": 0.4534004273658601,
      "learning_rate": 9.065030186193643e-06,
      "loss": 0.033,
      "step": 1737
    },
    {
      "epoch": 0.834774255523535,
      "grad_norm": 1.2265014670233068,
      "learning_rate": 9.063402430904087e-06,
      "loss": 0.0525,
      "step": 1738
    },
    {
      "epoch": 0.835254562920269,
      "grad_norm": 0.5422498828003771,
      "learning_rate": 9.061773406320027e-06,
      "loss": 0.0521,
      "step": 1739
    },
    {
      "epoch": 0.8357348703170029,
      "grad_norm": 0.4195808712209068,
      "learning_rate": 9.060143112950326e-06,
      "loss": 0.0316,
      "step": 1740
    },
    {
      "epoch": 0.8362151777137368,
      "grad_norm": 0.4855308450574598,
      "learning_rate": 9.058511551304241e-06,
      "loss": 0.0503,
      "step": 1741
    },
    {
      "epoch": 0.8366954851104706,
      "grad_norm": 0.5075843680608557,
      "learning_rate": 9.056878721891427e-06,
      "loss": 0.0425,
      "step": 1742
    },
    {
      "epoch": 0.8371757925072046,
      "grad_norm": 0.5482382504823542,
      "learning_rate": 9.055244625221935e-06,
      "loss": 0.0428,
      "step": 1743
    },
    {
      "epoch": 0.8376560999039385,
      "grad_norm": 0.6695595892233471,
      "learning_rate": 9.053609261806214e-06,
      "loss": 0.0587,
      "step": 1744
    },
    {
      "epoch": 0.8381364073006724,
      "grad_norm": 0.4581918394935273,
      "learning_rate": 9.0519726321551e-06,
      "loss": 0.0361,
      "step": 1745
    },
    {
      "epoch": 0.8386167146974063,
      "grad_norm": 0.3952408558855935,
      "learning_rate": 9.050334736779837e-06,
      "loss": 0.0381,
      "step": 1746
    },
    {
      "epoch": 0.8390970220941403,
      "grad_norm": 0.4109873080252859,
      "learning_rate": 9.048695576192058e-06,
      "loss": 0.0442,
      "step": 1747
    },
    {
      "epoch": 0.8395773294908742,
      "grad_norm": 0.5616982246438386,
      "learning_rate": 9.047055150903787e-06,
      "loss": 0.0584,
      "step": 1748
    },
    {
      "epoch": 0.840057636887608,
      "grad_norm": 0.37826080721339767,
      "learning_rate": 9.045413461427453e-06,
      "loss": 0.0409,
      "step": 1749
    },
    {
      "epoch": 0.840537944284342,
      "grad_norm": 0.40088324296361066,
      "learning_rate": 9.043770508275868e-06,
      "loss": 0.0487,
      "step": 1750
    },
    {
      "epoch": 0.8410182516810759,
      "grad_norm": 0.44718093929332914,
      "learning_rate": 9.04212629196225e-06,
      "loss": 0.0352,
      "step": 1751
    },
    {
      "epoch": 0.8414985590778098,
      "grad_norm": 0.4706059052244275,
      "learning_rate": 9.040480813000205e-06,
      "loss": 0.0358,
      "step": 1752
    },
    {
      "epoch": 0.8419788664745437,
      "grad_norm": 0.3670084341616506,
      "learning_rate": 9.038834071903736e-06,
      "loss": 0.0286,
      "step": 1753
    },
    {
      "epoch": 0.8424591738712777,
      "grad_norm": 0.5139890566947024,
      "learning_rate": 9.037186069187239e-06,
      "loss": 0.0471,
      "step": 1754
    },
    {
      "epoch": 0.8429394812680115,
      "grad_norm": 0.6580821574923913,
      "learning_rate": 9.035536805365503e-06,
      "loss": 0.0404,
      "step": 1755
    },
    {
      "epoch": 0.8434197886647454,
      "grad_norm": 0.3488569658059193,
      "learning_rate": 9.033886280953714e-06,
      "loss": 0.0332,
      "step": 1756
    },
    {
      "epoch": 0.8439000960614793,
      "grad_norm": 0.459104475456851,
      "learning_rate": 9.032234496467448e-06,
      "loss": 0.0457,
      "step": 1757
    },
    {
      "epoch": 0.8443804034582133,
      "grad_norm": 0.5555684583259805,
      "learning_rate": 9.03058145242268e-06,
      "loss": 0.0471,
      "step": 1758
    },
    {
      "epoch": 0.8448607108549472,
      "grad_norm": 0.4615657333941676,
      "learning_rate": 9.028927149335773e-06,
      "loss": 0.0412,
      "step": 1759
    },
    {
      "epoch": 0.8453410182516811,
      "grad_norm": 0.45856078963996716,
      "learning_rate": 9.027271587723487e-06,
      "loss": 0.0374,
      "step": 1760
    },
    {
      "epoch": 0.845821325648415,
      "grad_norm": 0.6912129790142436,
      "learning_rate": 9.025614768102972e-06,
      "loss": 0.0693,
      "step": 1761
    },
    {
      "epoch": 0.8463016330451489,
      "grad_norm": 0.5122989554497435,
      "learning_rate": 9.023956690991775e-06,
      "loss": 0.0465,
      "step": 1762
    },
    {
      "epoch": 0.8467819404418828,
      "grad_norm": 0.36126845802883945,
      "learning_rate": 9.02229735690783e-06,
      "loss": 0.0296,
      "step": 1763
    },
    {
      "epoch": 0.8472622478386167,
      "grad_norm": 0.6087551399045577,
      "learning_rate": 9.020636766369471e-06,
      "loss": 0.0506,
      "step": 1764
    },
    {
      "epoch": 0.8477425552353506,
      "grad_norm": 0.38525060421145424,
      "learning_rate": 9.018974919895418e-06,
      "loss": 0.0355,
      "step": 1765
    },
    {
      "epoch": 0.8482228626320846,
      "grad_norm": 0.346018375745237,
      "learning_rate": 9.017311818004785e-06,
      "loss": 0.0341,
      "step": 1766
    },
    {
      "epoch": 0.8487031700288185,
      "grad_norm": 0.6260615640367205,
      "learning_rate": 9.01564746121708e-06,
      "loss": 0.0443,
      "step": 1767
    },
    {
      "epoch": 0.8491834774255523,
      "grad_norm": 0.5187278606147561,
      "learning_rate": 9.013981850052205e-06,
      "loss": 0.0433,
      "step": 1768
    },
    {
      "epoch": 0.8496637848222862,
      "grad_norm": 0.5503187612040098,
      "learning_rate": 9.012314985030445e-06,
      "loss": 0.0358,
      "step": 1769
    },
    {
      "epoch": 0.8501440922190202,
      "grad_norm": 0.6316117180141888,
      "learning_rate": 9.010646866672488e-06,
      "loss": 0.0384,
      "step": 1770
    },
    {
      "epoch": 0.8506243996157541,
      "grad_norm": 0.5212663878288761,
      "learning_rate": 9.008977495499405e-06,
      "loss": 0.0361,
      "step": 1771
    },
    {
      "epoch": 0.851104707012488,
      "grad_norm": 0.5547948672955962,
      "learning_rate": 9.007306872032663e-06,
      "loss": 0.0446,
      "step": 1772
    },
    {
      "epoch": 0.8515850144092219,
      "grad_norm": 0.42396510808899085,
      "learning_rate": 9.005634996794117e-06,
      "loss": 0.032,
      "step": 1773
    },
    {
      "epoch": 0.8520653218059558,
      "grad_norm": 0.7725093468772292,
      "learning_rate": 9.003961870306015e-06,
      "loss": 0.0549,
      "step": 1774
    },
    {
      "epoch": 0.8525456292026897,
      "grad_norm": 0.4308369316438503,
      "learning_rate": 9.002287493090996e-06,
      "loss": 0.0499,
      "step": 1775
    },
    {
      "epoch": 0.8530259365994236,
      "grad_norm": 0.2780784001577288,
      "learning_rate": 9.000611865672088e-06,
      "loss": 0.0287,
      "step": 1776
    },
    {
      "epoch": 0.8535062439961575,
      "grad_norm": 0.5180718543954558,
      "learning_rate": 8.998934988572713e-06,
      "loss": 0.0329,
      "step": 1777
    },
    {
      "epoch": 0.8539865513928915,
      "grad_norm": 0.5057214443895521,
      "learning_rate": 8.997256862316678e-06,
      "loss": 0.0481,
      "step": 1778
    },
    {
      "epoch": 0.8544668587896254,
      "grad_norm": 0.3751078985501239,
      "learning_rate": 8.995577487428187e-06,
      "loss": 0.0321,
      "step": 1779
    },
    {
      "epoch": 0.8549471661863592,
      "grad_norm": 0.8942585028670543,
      "learning_rate": 8.993896864431825e-06,
      "loss": 0.0442,
      "step": 1780
    },
    {
      "epoch": 0.8554274735830932,
      "grad_norm": 0.537990863386714,
      "learning_rate": 8.992214993852576e-06,
      "loss": 0.0385,
      "step": 1781
    },
    {
      "epoch": 0.8559077809798271,
      "grad_norm": 0.3368917295945931,
      "learning_rate": 8.99053187621581e-06,
      "loss": 0.0358,
      "step": 1782
    },
    {
      "epoch": 0.856388088376561,
      "grad_norm": 0.3625643306114823,
      "learning_rate": 8.988847512047285e-06,
      "loss": 0.0359,
      "step": 1783
    },
    {
      "epoch": 0.8568683957732949,
      "grad_norm": 0.5392108809967152,
      "learning_rate": 8.98716190187315e-06,
      "loss": 0.0445,
      "step": 1784
    },
    {
      "epoch": 0.8573487031700289,
      "grad_norm": 0.5998276931275567,
      "learning_rate": 8.985475046219942e-06,
      "loss": 0.0443,
      "step": 1785
    },
    {
      "epoch": 0.8578290105667628,
      "grad_norm": 0.5131152759715736,
      "learning_rate": 8.983786945614589e-06,
      "loss": 0.0387,
      "step": 1786
    },
    {
      "epoch": 0.8583093179634966,
      "grad_norm": 0.6910401242120892,
      "learning_rate": 8.98209760058441e-06,
      "loss": 0.0469,
      "step": 1787
    },
    {
      "epoch": 0.8587896253602305,
      "grad_norm": 0.4194901862760443,
      "learning_rate": 8.980407011657103e-06,
      "loss": 0.0403,
      "step": 1788
    },
    {
      "epoch": 0.8592699327569645,
      "grad_norm": 0.46623807550964225,
      "learning_rate": 8.978715179360767e-06,
      "loss": 0.0392,
      "step": 1789
    },
    {
      "epoch": 0.8597502401536984,
      "grad_norm": 0.44859136364628743,
      "learning_rate": 8.97702210422388e-06,
      "loss": 0.0458,
      "step": 1790
    },
    {
      "epoch": 0.8602305475504323,
      "grad_norm": 0.6257891336546498,
      "learning_rate": 8.975327786775316e-06,
      "loss": 0.0366,
      "step": 1791
    },
    {
      "epoch": 0.8607108549471661,
      "grad_norm": 0.4736431571435201,
      "learning_rate": 8.973632227544326e-06,
      "loss": 0.0433,
      "step": 1792
    },
    {
      "epoch": 0.8611911623439001,
      "grad_norm": 0.4152452493682378,
      "learning_rate": 8.971935427060563e-06,
      "loss": 0.0405,
      "step": 1793
    },
    {
      "epoch": 0.861671469740634,
      "grad_norm": 0.9662287838754176,
      "learning_rate": 8.970237385854059e-06,
      "loss": 0.0583,
      "step": 1794
    },
    {
      "epoch": 0.8621517771373679,
      "grad_norm": 0.5056226800498823,
      "learning_rate": 8.96853810445523e-06,
      "loss": 0.0476,
      "step": 1795
    },
    {
      "epoch": 0.8626320845341018,
      "grad_norm": 0.5166750113936333,
      "learning_rate": 8.966837583394891e-06,
      "loss": 0.0434,
      "step": 1796
    },
    {
      "epoch": 0.8631123919308358,
      "grad_norm": 0.5477224778262247,
      "learning_rate": 8.965135823204232e-06,
      "loss": 0.0416,
      "step": 1797
    },
    {
      "epoch": 0.8635926993275697,
      "grad_norm": 0.3837910069912173,
      "learning_rate": 8.963432824414842e-06,
      "loss": 0.0268,
      "step": 1798
    },
    {
      "epoch": 0.8640730067243035,
      "grad_norm": 0.750187104754067,
      "learning_rate": 8.961728587558684e-06,
      "loss": 0.0514,
      "step": 1799
    },
    {
      "epoch": 0.8645533141210374,
      "grad_norm": 0.5116240081446706,
      "learning_rate": 8.96002311316812e-06,
      "loss": 0.0362,
      "step": 1800
    },
    {
      "epoch": 0.8650336215177714,
      "grad_norm": 0.5452866982844962,
      "learning_rate": 8.95831640177589e-06,
      "loss": 0.0412,
      "step": 1801
    },
    {
      "epoch": 0.8655139289145053,
      "grad_norm": 0.3893421475161622,
      "learning_rate": 8.956608453915126e-06,
      "loss": 0.0422,
      "step": 1802
    },
    {
      "epoch": 0.8659942363112392,
      "grad_norm": 0.400744042269121,
      "learning_rate": 8.954899270119339e-06,
      "loss": 0.0442,
      "step": 1803
    },
    {
      "epoch": 0.866474543707973,
      "grad_norm": 0.8459931991973275,
      "learning_rate": 8.953188850922436e-06,
      "loss": 0.0319,
      "step": 1804
    },
    {
      "epoch": 0.866954851104707,
      "grad_norm": 0.4769183729256473,
      "learning_rate": 8.951477196858703e-06,
      "loss": 0.0392,
      "step": 1805
    },
    {
      "epoch": 0.8674351585014409,
      "grad_norm": 0.5961258504118155,
      "learning_rate": 8.94976430846281e-06,
      "loss": 0.044,
      "step": 1806
    },
    {
      "epoch": 0.8679154658981748,
      "grad_norm": 0.7889199748787529,
      "learning_rate": 8.94805018626982e-06,
      "loss": 0.0554,
      "step": 1807
    },
    {
      "epoch": 0.8683957732949087,
      "grad_norm": 0.6776357232960094,
      "learning_rate": 8.946334830815176e-06,
      "loss": 0.0474,
      "step": 1808
    },
    {
      "epoch": 0.8688760806916427,
      "grad_norm": 0.5294621498199231,
      "learning_rate": 8.944618242634707e-06,
      "loss": 0.0441,
      "step": 1809
    },
    {
      "epoch": 0.8693563880883766,
      "grad_norm": 0.40665428772649925,
      "learning_rate": 8.942900422264627e-06,
      "loss": 0.0293,
      "step": 1810
    },
    {
      "epoch": 0.8698366954851104,
      "grad_norm": 0.5854731121778896,
      "learning_rate": 8.941181370241538e-06,
      "loss": 0.0472,
      "step": 1811
    },
    {
      "epoch": 0.8703170028818443,
      "grad_norm": 0.514195838759812,
      "learning_rate": 8.939461087102424e-06,
      "loss": 0.0396,
      "step": 1812
    },
    {
      "epoch": 0.8707973102785783,
      "grad_norm": 0.6456083251358632,
      "learning_rate": 8.937739573384653e-06,
      "loss": 0.048,
      "step": 1813
    },
    {
      "epoch": 0.8712776176753122,
      "grad_norm": 0.40027049247918695,
      "learning_rate": 8.936016829625977e-06,
      "loss": 0.0373,
      "step": 1814
    },
    {
      "epoch": 0.8717579250720461,
      "grad_norm": 1.4781590167762066,
      "learning_rate": 8.934292856364535e-06,
      "loss": 0.0706,
      "step": 1815
    },
    {
      "epoch": 0.8722382324687801,
      "grad_norm": 0.5618715192739092,
      "learning_rate": 8.932567654138849e-06,
      "loss": 0.0523,
      "step": 1816
    },
    {
      "epoch": 0.872718539865514,
      "grad_norm": 0.43422019108974075,
      "learning_rate": 8.930841223487823e-06,
      "loss": 0.0464,
      "step": 1817
    },
    {
      "epoch": 0.8731988472622478,
      "grad_norm": 0.3876967201171116,
      "learning_rate": 8.929113564950746e-06,
      "loss": 0.0278,
      "step": 1818
    },
    {
      "epoch": 0.8736791546589817,
      "grad_norm": 0.46720060258300933,
      "learning_rate": 8.927384679067293e-06,
      "loss": 0.0402,
      "step": 1819
    },
    {
      "epoch": 0.8741594620557157,
      "grad_norm": 0.519410545548104,
      "learning_rate": 8.925654566377519e-06,
      "loss": 0.0541,
      "step": 1820
    },
    {
      "epoch": 0.8746397694524496,
      "grad_norm": 1.022650569670746,
      "learning_rate": 8.923923227421862e-06,
      "loss": 0.0564,
      "step": 1821
    },
    {
      "epoch": 0.8751200768491835,
      "grad_norm": 0.358444548188737,
      "learning_rate": 8.922190662741146e-06,
      "loss": 0.0352,
      "step": 1822
    },
    {
      "epoch": 0.8756003842459174,
      "grad_norm": 0.6535408000392346,
      "learning_rate": 8.920456872876575e-06,
      "loss": 0.0679,
      "step": 1823
    },
    {
      "epoch": 0.8760806916426513,
      "grad_norm": 0.3811923244253041,
      "learning_rate": 8.918721858369738e-06,
      "loss": 0.0495,
      "step": 1824
    },
    {
      "epoch": 0.8765609990393852,
      "grad_norm": 0.502122776772645,
      "learning_rate": 8.916985619762605e-06,
      "loss": 0.0371,
      "step": 1825
    },
    {
      "epoch": 0.8770413064361191,
      "grad_norm": 1.0483402982818049,
      "learning_rate": 8.915248157597529e-06,
      "loss": 0.0655,
      "step": 1826
    },
    {
      "epoch": 0.877521613832853,
      "grad_norm": 0.45223694971335204,
      "learning_rate": 8.913509472417246e-06,
      "loss": 0.0311,
      "step": 1827
    },
    {
      "epoch": 0.878001921229587,
      "grad_norm": 0.4508039587720194,
      "learning_rate": 8.91176956476487e-06,
      "loss": 0.0436,
      "step": 1828
    },
    {
      "epoch": 0.8784822286263209,
      "grad_norm": 0.4562648581626756,
      "learning_rate": 8.910028435183906e-06,
      "loss": 0.0358,
      "step": 1829
    },
    {
      "epoch": 0.8789625360230547,
      "grad_norm": 0.3635207577181714,
      "learning_rate": 8.90828608421823e-06,
      "loss": 0.0357,
      "step": 1830
    },
    {
      "epoch": 0.8794428434197886,
      "grad_norm": 0.5209884592558325,
      "learning_rate": 8.906542512412105e-06,
      "loss": 0.0478,
      "step": 1831
    },
    {
      "epoch": 0.8799231508165226,
      "grad_norm": 0.37857114644619005,
      "learning_rate": 8.904797720310176e-06,
      "loss": 0.045,
      "step": 1832
    },
    {
      "epoch": 0.8804034582132565,
      "grad_norm": 0.8142054557917969,
      "learning_rate": 8.903051708457465e-06,
      "loss": 0.0376,
      "step": 1833
    },
    {
      "epoch": 0.8808837656099904,
      "grad_norm": 0.7396298749741087,
      "learning_rate": 8.90130447739938e-06,
      "loss": 0.0589,
      "step": 1834
    },
    {
      "epoch": 0.8813640730067243,
      "grad_norm": 0.33453688614785887,
      "learning_rate": 8.899556027681708e-06,
      "loss": 0.0231,
      "step": 1835
    },
    {
      "epoch": 0.8818443804034583,
      "grad_norm": 0.4457358953894517,
      "learning_rate": 8.897806359850614e-06,
      "loss": 0.0317,
      "step": 1836
    },
    {
      "epoch": 0.8823246878001921,
      "grad_norm": 1.3515020556940829,
      "learning_rate": 8.896055474452649e-06,
      "loss": 0.0427,
      "step": 1837
    },
    {
      "epoch": 0.882804995196926,
      "grad_norm": 0.4580165993160693,
      "learning_rate": 8.894303372034738e-06,
      "loss": 0.0433,
      "step": 1838
    },
    {
      "epoch": 0.8832853025936599,
      "grad_norm": 0.4503178993927309,
      "learning_rate": 8.89255005314419e-06,
      "loss": 0.0373,
      "step": 1839
    },
    {
      "epoch": 0.8837656099903939,
      "grad_norm": 0.4368170227289143,
      "learning_rate": 8.890795518328698e-06,
      "loss": 0.0461,
      "step": 1840
    },
    {
      "epoch": 0.8842459173871278,
      "grad_norm": 0.5359261995070854,
      "learning_rate": 8.889039768136323e-06,
      "loss": 0.0633,
      "step": 1841
    },
    {
      "epoch": 0.8847262247838616,
      "grad_norm": 0.32753604552230753,
      "learning_rate": 8.887282803115518e-06,
      "loss": 0.031,
      "step": 1842
    },
    {
      "epoch": 0.8852065321805955,
      "grad_norm": 0.561984541895264,
      "learning_rate": 8.885524623815107e-06,
      "loss": 0.061,
      "step": 1843
    },
    {
      "epoch": 0.8856868395773295,
      "grad_norm": 0.4954923412229821,
      "learning_rate": 8.883765230784298e-06,
      "loss": 0.0376,
      "step": 1844
    },
    {
      "epoch": 0.8861671469740634,
      "grad_norm": 0.40554156267027097,
      "learning_rate": 8.882004624572676e-06,
      "loss": 0.0286,
      "step": 1845
    },
    {
      "epoch": 0.8866474543707973,
      "grad_norm": 0.5113849488787681,
      "learning_rate": 8.880242805730208e-06,
      "loss": 0.0426,
      "step": 1846
    },
    {
      "epoch": 0.8871277617675313,
      "grad_norm": 0.37027022381367386,
      "learning_rate": 8.878479774807235e-06,
      "loss": 0.0295,
      "step": 1847
    },
    {
      "epoch": 0.8876080691642652,
      "grad_norm": 0.6769968021961212,
      "learning_rate": 8.876715532354478e-06,
      "loss": 0.0452,
      "step": 1848
    },
    {
      "epoch": 0.888088376560999,
      "grad_norm": 0.48737302282911404,
      "learning_rate": 8.87495007892304e-06,
      "loss": 0.0445,
      "step": 1849
    },
    {
      "epoch": 0.8885686839577329,
      "grad_norm": 0.3386523878498041,
      "learning_rate": 8.873183415064401e-06,
      "loss": 0.0333,
      "step": 1850
    },
    {
      "epoch": 0.8890489913544669,
      "grad_norm": 0.6658697028186209,
      "learning_rate": 8.871415541330417e-06,
      "loss": 0.0343,
      "step": 1851
    },
    {
      "epoch": 0.8895292987512008,
      "grad_norm": 0.4919802717601218,
      "learning_rate": 8.86964645827332e-06,
      "loss": 0.0455,
      "step": 1852
    },
    {
      "epoch": 0.8900096061479347,
      "grad_norm": 0.4263226706660005,
      "learning_rate": 8.867876166445724e-06,
      "loss": 0.0418,
      "step": 1853
    },
    {
      "epoch": 0.8904899135446686,
      "grad_norm": 0.3214256914507718,
      "learning_rate": 8.86610466640062e-06,
      "loss": 0.0327,
      "step": 1854
    },
    {
      "epoch": 0.8909702209414025,
      "grad_norm": 0.6264841255471314,
      "learning_rate": 8.864331958691377e-06,
      "loss": 0.0305,
      "step": 1855
    },
    {
      "epoch": 0.8914505283381364,
      "grad_norm": 1.2829456922872837,
      "learning_rate": 8.862558043871737e-06,
      "loss": 0.0585,
      "step": 1856
    },
    {
      "epoch": 0.8919308357348703,
      "grad_norm": 0.5015853830955843,
      "learning_rate": 8.860782922495821e-06,
      "loss": 0.0479,
      "step": 1857
    },
    {
      "epoch": 0.8924111431316042,
      "grad_norm": 0.6658604832983488,
      "learning_rate": 8.859006595118133e-06,
      "loss": 0.0485,
      "step": 1858
    },
    {
      "epoch": 0.8928914505283382,
      "grad_norm": 0.47639098590814116,
      "learning_rate": 8.857229062293544e-06,
      "loss": 0.0388,
      "step": 1859
    },
    {
      "epoch": 0.8933717579250721,
      "grad_norm": 0.4302098869706598,
      "learning_rate": 8.855450324577308e-06,
      "loss": 0.0482,
      "step": 1860
    },
    {
      "epoch": 0.8938520653218059,
      "grad_norm": 0.6482456313500351,
      "learning_rate": 8.853670382525052e-06,
      "loss": 0.0436,
      "step": 1861
    },
    {
      "epoch": 0.8943323727185398,
      "grad_norm": 0.47660614333387596,
      "learning_rate": 8.851889236692783e-06,
      "loss": 0.046,
      "step": 1862
    },
    {
      "epoch": 0.8948126801152738,
      "grad_norm": 0.6952319224454124,
      "learning_rate": 8.85010688763688e-06,
      "loss": 0.0495,
      "step": 1863
    },
    {
      "epoch": 0.8952929875120077,
      "grad_norm": 0.4901487858015449,
      "learning_rate": 8.8483233359141e-06,
      "loss": 0.0358,
      "step": 1864
    },
    {
      "epoch": 0.8957732949087416,
      "grad_norm": 0.5728646775595035,
      "learning_rate": 8.846538582081575e-06,
      "loss": 0.0435,
      "step": 1865
    },
    {
      "epoch": 0.8962536023054755,
      "grad_norm": 0.44725468456391415,
      "learning_rate": 8.84475262669681e-06,
      "loss": 0.0443,
      "step": 1866
    },
    {
      "epoch": 0.8967339097022095,
      "grad_norm": 0.5773271593291803,
      "learning_rate": 8.842965470317694e-06,
      "loss": 0.0424,
      "step": 1867
    },
    {
      "epoch": 0.8972142170989433,
      "grad_norm": 0.7198849042715714,
      "learning_rate": 8.84117711350248e-06,
      "loss": 0.0468,
      "step": 1868
    },
    {
      "epoch": 0.8976945244956772,
      "grad_norm": 0.5957419746315502,
      "learning_rate": 8.839387556809805e-06,
      "loss": 0.0468,
      "step": 1869
    },
    {
      "epoch": 0.8981748318924111,
      "grad_norm": 0.588750525644505,
      "learning_rate": 8.837596800798674e-06,
      "loss": 0.0464,
      "step": 1870
    },
    {
      "epoch": 0.8986551392891451,
      "grad_norm": 0.4505825678549945,
      "learning_rate": 8.835804846028473e-06,
      "loss": 0.0428,
      "step": 1871
    },
    {
      "epoch": 0.899135446685879,
      "grad_norm": 0.6871356390036418,
      "learning_rate": 8.834011693058955e-06,
      "loss": 0.0693,
      "step": 1872
    },
    {
      "epoch": 0.8996157540826129,
      "grad_norm": 0.4898107165484671,
      "learning_rate": 8.832217342450253e-06,
      "loss": 0.0425,
      "step": 1873
    },
    {
      "epoch": 0.9000960614793467,
      "grad_norm": 0.5624738413790371,
      "learning_rate": 8.830421794762873e-06,
      "loss": 0.044,
      "step": 1874
    },
    {
      "epoch": 0.9005763688760807,
      "grad_norm": 0.5195088744781344,
      "learning_rate": 8.828625050557695e-06,
      "loss": 0.0606,
      "step": 1875
    },
    {
      "epoch": 0.9010566762728146,
      "grad_norm": 0.38938511927268654,
      "learning_rate": 8.826827110395973e-06,
      "loss": 0.0336,
      "step": 1876
    },
    {
      "epoch": 0.9015369836695485,
      "grad_norm": 0.38684793123655187,
      "learning_rate": 8.82502797483933e-06,
      "loss": 0.0375,
      "step": 1877
    },
    {
      "epoch": 0.9020172910662824,
      "grad_norm": 0.7363107702433455,
      "learning_rate": 8.823227644449767e-06,
      "loss": 0.0465,
      "step": 1878
    },
    {
      "epoch": 0.9024975984630164,
      "grad_norm": 0.48895156850885646,
      "learning_rate": 8.821426119789662e-06,
      "loss": 0.0483,
      "step": 1879
    },
    {
      "epoch": 0.9029779058597502,
      "grad_norm": 0.660241990808151,
      "learning_rate": 8.819623401421757e-06,
      "loss": 0.0476,
      "step": 1880
    },
    {
      "epoch": 0.9034582132564841,
      "grad_norm": 0.46543854311143706,
      "learning_rate": 8.817819489909172e-06,
      "loss": 0.0403,
      "step": 1881
    },
    {
      "epoch": 0.9039385206532181,
      "grad_norm": 0.6017118184027497,
      "learning_rate": 8.8160143858154e-06,
      "loss": 0.0478,
      "step": 1882
    },
    {
      "epoch": 0.904418828049952,
      "grad_norm": 0.39226105718412113,
      "learning_rate": 8.814208089704306e-06,
      "loss": 0.0418,
      "step": 1883
    },
    {
      "epoch": 0.9048991354466859,
      "grad_norm": 0.4599436630675356,
      "learning_rate": 8.812400602140125e-06,
      "loss": 0.0468,
      "step": 1884
    },
    {
      "epoch": 0.9053794428434198,
      "grad_norm": 0.8681683906823245,
      "learning_rate": 8.810591923687468e-06,
      "loss": 0.0473,
      "step": 1885
    },
    {
      "epoch": 0.9058597502401537,
      "grad_norm": 0.7322132140172354,
      "learning_rate": 8.808782054911315e-06,
      "loss": 0.0662,
      "step": 1886
    },
    {
      "epoch": 0.9063400576368876,
      "grad_norm": 0.6055607958778498,
      "learning_rate": 8.806970996377018e-06,
      "loss": 0.0461,
      "step": 1887
    },
    {
      "epoch": 0.9068203650336215,
      "grad_norm": 0.4352135953438638,
      "learning_rate": 8.805158748650304e-06,
      "loss": 0.0443,
      "step": 1888
    },
    {
      "epoch": 0.9073006724303554,
      "grad_norm": 0.8207849567629364,
      "learning_rate": 8.803345312297269e-06,
      "loss": 0.0686,
      "step": 1889
    },
    {
      "epoch": 0.9077809798270894,
      "grad_norm": 0.4467690737669225,
      "learning_rate": 8.801530687884378e-06,
      "loss": 0.0532,
      "step": 1890
    },
    {
      "epoch": 0.9082612872238233,
      "grad_norm": 0.5198478158932659,
      "learning_rate": 8.799714875978472e-06,
      "loss": 0.0419,
      "step": 1891
    },
    {
      "epoch": 0.9087415946205571,
      "grad_norm": 1.0896879533925696,
      "learning_rate": 8.797897877146757e-06,
      "loss": 0.0734,
      "step": 1892
    },
    {
      "epoch": 0.909221902017291,
      "grad_norm": 0.3617708509543011,
      "learning_rate": 8.796079691956818e-06,
      "loss": 0.0327,
      "step": 1893
    },
    {
      "epoch": 0.909702209414025,
      "grad_norm": 0.7301142094909887,
      "learning_rate": 8.794260320976602e-06,
      "loss": 0.0392,
      "step": 1894
    },
    {
      "epoch": 0.9101825168107589,
      "grad_norm": 0.4068531645359285,
      "learning_rate": 8.79243976477443e-06,
      "loss": 0.0469,
      "step": 1895
    },
    {
      "epoch": 0.9106628242074928,
      "grad_norm": 0.48618429718856376,
      "learning_rate": 8.790618023918995e-06,
      "loss": 0.0558,
      "step": 1896
    },
    {
      "epoch": 0.9111431316042267,
      "grad_norm": 0.5431200818159274,
      "learning_rate": 8.788795098979358e-06,
      "loss": 0.0515,
      "step": 1897
    },
    {
      "epoch": 0.9116234390009607,
      "grad_norm": 0.6076681250674426,
      "learning_rate": 8.786970990524952e-06,
      "loss": 0.043,
      "step": 1898
    },
    {
      "epoch": 0.9121037463976945,
      "grad_norm": 0.4574042135859862,
      "learning_rate": 8.785145699125577e-06,
      "loss": 0.0468,
      "step": 1899
    },
    {
      "epoch": 0.9125840537944284,
      "grad_norm": 0.7075435238401119,
      "learning_rate": 8.783319225351401e-06,
      "loss": 0.0464,
      "step": 1900
    },
    {
      "epoch": 0.9130643611911623,
      "grad_norm": 0.534633281625552,
      "learning_rate": 8.781491569772966e-06,
      "loss": 0.0426,
      "step": 1901
    },
    {
      "epoch": 0.9135446685878963,
      "grad_norm": 0.3921046587973985,
      "learning_rate": 8.77966273296118e-06,
      "loss": 0.0358,
      "step": 1902
    },
    {
      "epoch": 0.9140249759846302,
      "grad_norm": 0.41990313919226124,
      "learning_rate": 8.777832715487325e-06,
      "loss": 0.0285,
      "step": 1903
    },
    {
      "epoch": 0.914505283381364,
      "grad_norm": 0.41717379390801274,
      "learning_rate": 8.776001517923042e-06,
      "loss": 0.0363,
      "step": 1904
    },
    {
      "epoch": 0.9149855907780979,
      "grad_norm": 0.3366938255497245,
      "learning_rate": 8.774169140840349e-06,
      "loss": 0.0374,
      "step": 1905
    },
    {
      "epoch": 0.9154658981748319,
      "grad_norm": 0.36457948244141064,
      "learning_rate": 8.772335584811631e-06,
      "loss": 0.0299,
      "step": 1906
    },
    {
      "epoch": 0.9159462055715658,
      "grad_norm": 0.4828707338456646,
      "learning_rate": 8.770500850409641e-06,
      "loss": 0.0331,
      "step": 1907
    },
    {
      "epoch": 0.9164265129682997,
      "grad_norm": 0.5966763216669441,
      "learning_rate": 8.768664938207494e-06,
      "loss": 0.0498,
      "step": 1908
    },
    {
      "epoch": 0.9169068203650336,
      "grad_norm": 0.3927338383436385,
      "learning_rate": 8.766827848778683e-06,
      "loss": 0.0338,
      "step": 1909
    },
    {
      "epoch": 0.9173871277617676,
      "grad_norm": 0.5893155428418285,
      "learning_rate": 8.764989582697065e-06,
      "loss": 0.0411,
      "step": 1910
    },
    {
      "epoch": 0.9178674351585014,
      "grad_norm": 0.5356691788194019,
      "learning_rate": 8.763150140536858e-06,
      "loss": 0.0398,
      "step": 1911
    },
    {
      "epoch": 0.9183477425552353,
      "grad_norm": 0.4306294724075641,
      "learning_rate": 8.761309522872657e-06,
      "loss": 0.0403,
      "step": 1912
    },
    {
      "epoch": 0.9188280499519692,
      "grad_norm": 0.4226012125799819,
      "learning_rate": 8.75946773027942e-06,
      "loss": 0.0353,
      "step": 1913
    },
    {
      "epoch": 0.9193083573487032,
      "grad_norm": 0.4136981494543946,
      "learning_rate": 8.75762476333247e-06,
      "loss": 0.0362,
      "step": 1914
    },
    {
      "epoch": 0.9197886647454371,
      "grad_norm": 0.6346827624212021,
      "learning_rate": 8.755780622607499e-06,
      "loss": 0.0445,
      "step": 1915
    },
    {
      "epoch": 0.920268972142171,
      "grad_norm": 0.558450568203092,
      "learning_rate": 8.753935308680568e-06,
      "loss": 0.043,
      "step": 1916
    },
    {
      "epoch": 0.920749279538905,
      "grad_norm": 0.43965358503774565,
      "learning_rate": 8.7520888221281e-06,
      "loss": 0.0495,
      "step": 1917
    },
    {
      "epoch": 0.9212295869356388,
      "grad_norm": 0.5476700265664696,
      "learning_rate": 8.750241163526887e-06,
      "loss": 0.0347,
      "step": 1918
    },
    {
      "epoch": 0.9217098943323727,
      "grad_norm": 0.39304076802918375,
      "learning_rate": 8.748392333454085e-06,
      "loss": 0.0349,
      "step": 1919
    },
    {
      "epoch": 0.9221902017291066,
      "grad_norm": 0.4971949825044665,
      "learning_rate": 8.74654233248722e-06,
      "loss": 0.0395,
      "step": 1920
    },
    {
      "epoch": 0.9226705091258406,
      "grad_norm": 0.5186534483519105,
      "learning_rate": 8.74469116120418e-06,
      "loss": 0.0307,
      "step": 1921
    },
    {
      "epoch": 0.9231508165225745,
      "grad_norm": 0.3224287987538711,
      "learning_rate": 8.742838820183218e-06,
      "loss": 0.024,
      "step": 1922
    },
    {
      "epoch": 0.9236311239193083,
      "grad_norm": 0.4280132539326167,
      "learning_rate": 8.740985310002956e-06,
      "loss": 0.0417,
      "step": 1923
    },
    {
      "epoch": 0.9241114313160422,
      "grad_norm": 0.41045791987815555,
      "learning_rate": 8.739130631242379e-06,
      "loss": 0.0359,
      "step": 1924
    },
    {
      "epoch": 0.9245917387127762,
      "grad_norm": 0.5184741851396398,
      "learning_rate": 8.737274784480839e-06,
      "loss": 0.0563,
      "step": 1925
    },
    {
      "epoch": 0.9250720461095101,
      "grad_norm": 0.31914533082950014,
      "learning_rate": 8.735417770298046e-06,
      "loss": 0.0299,
      "step": 1926
    },
    {
      "epoch": 0.925552353506244,
      "grad_norm": 0.5115763990615543,
      "learning_rate": 8.733559589274086e-06,
      "loss": 0.0495,
      "step": 1927
    },
    {
      "epoch": 0.9260326609029779,
      "grad_norm": 0.4761196913567422,
      "learning_rate": 8.731700241989398e-06,
      "loss": 0.0445,
      "step": 1928
    },
    {
      "epoch": 0.9265129682997119,
      "grad_norm": 0.551401916157282,
      "learning_rate": 8.729839729024794e-06,
      "loss": 0.036,
      "step": 1929
    },
    {
      "epoch": 0.9269932756964457,
      "grad_norm": 0.3446484842092706,
      "learning_rate": 8.727978050961446e-06,
      "loss": 0.036,
      "step": 1930
    },
    {
      "epoch": 0.9274735830931796,
      "grad_norm": 0.3994863651754855,
      "learning_rate": 8.726115208380892e-06,
      "loss": 0.0445,
      "step": 1931
    },
    {
      "epoch": 0.9279538904899135,
      "grad_norm": 0.39412495064642333,
      "learning_rate": 8.724251201865029e-06,
      "loss": 0.0495,
      "step": 1932
    },
    {
      "epoch": 0.9284341978866475,
      "grad_norm": 0.6031610926197576,
      "learning_rate": 8.722386031996124e-06,
      "loss": 0.0532,
      "step": 1933
    },
    {
      "epoch": 0.9289145052833814,
      "grad_norm": 0.4130746229906913,
      "learning_rate": 8.720519699356804e-06,
      "loss": 0.0438,
      "step": 1934
    },
    {
      "epoch": 0.9293948126801153,
      "grad_norm": 0.4945763714239433,
      "learning_rate": 8.71865220453006e-06,
      "loss": 0.0537,
      "step": 1935
    },
    {
      "epoch": 0.9298751200768491,
      "grad_norm": 0.6642151852115892,
      "learning_rate": 8.716783548099243e-06,
      "loss": 0.0305,
      "step": 1936
    },
    {
      "epoch": 0.9303554274735831,
      "grad_norm": 0.4808430941047988,
      "learning_rate": 8.714913730648073e-06,
      "loss": 0.0383,
      "step": 1937
    },
    {
      "epoch": 0.930835734870317,
      "grad_norm": 0.38784696307406447,
      "learning_rate": 8.713042752760629e-06,
      "loss": 0.0427,
      "step": 1938
    },
    {
      "epoch": 0.9313160422670509,
      "grad_norm": 0.46691293249094495,
      "learning_rate": 8.71117061502135e-06,
      "loss": 0.037,
      "step": 1939
    },
    {
      "epoch": 0.9317963496637848,
      "grad_norm": 0.528772802563039,
      "learning_rate": 8.709297318015042e-06,
      "loss": 0.0413,
      "step": 1940
    },
    {
      "epoch": 0.9322766570605188,
      "grad_norm": 0.49227721232250315,
      "learning_rate": 8.707422862326872e-06,
      "loss": 0.0359,
      "step": 1941
    },
    {
      "epoch": 0.9327569644572526,
      "grad_norm": 0.47166989196463194,
      "learning_rate": 8.705547248542366e-06,
      "loss": 0.0362,
      "step": 1942
    },
    {
      "epoch": 0.9332372718539865,
      "grad_norm": 0.5256984649221889,
      "learning_rate": 8.703670477247415e-06,
      "loss": 0.0483,
      "step": 1943
    },
    {
      "epoch": 0.9337175792507204,
      "grad_norm": 0.4307455296849523,
      "learning_rate": 8.701792549028269e-06,
      "loss": 0.0487,
      "step": 1944
    },
    {
      "epoch": 0.9341978866474544,
      "grad_norm": 0.46947997670245323,
      "learning_rate": 8.699913464471543e-06,
      "loss": 0.0441,
      "step": 1945
    },
    {
      "epoch": 0.9346781940441883,
      "grad_norm": 0.5890864790257363,
      "learning_rate": 8.69803322416421e-06,
      "loss": 0.0629,
      "step": 1946
    },
    {
      "epoch": 0.9351585014409222,
      "grad_norm": 0.380486982939282,
      "learning_rate": 8.696151828693606e-06,
      "loss": 0.0367,
      "step": 1947
    },
    {
      "epoch": 0.9356388088376562,
      "grad_norm": 0.42021576660117066,
      "learning_rate": 8.694269278647425e-06,
      "loss": 0.0379,
      "step": 1948
    },
    {
      "epoch": 0.93611911623439,
      "grad_norm": 0.41391650206618286,
      "learning_rate": 8.692385574613725e-06,
      "loss": 0.0365,
      "step": 1949
    },
    {
      "epoch": 0.9365994236311239,
      "grad_norm": 0.6058118454984688,
      "learning_rate": 8.690500717180924e-06,
      "loss": 0.065,
      "step": 1950
    },
    {
      "epoch": 0.9370797310278578,
      "grad_norm": 0.4144008472750772,
      "learning_rate": 8.688614706937794e-06,
      "loss": 0.0337,
      "step": 1951
    },
    {
      "epoch": 0.9375600384245918,
      "grad_norm": 0.46837940872225015,
      "learning_rate": 8.68672754447348e-06,
      "loss": 0.0539,
      "step": 1952
    },
    {
      "epoch": 0.9380403458213257,
      "grad_norm": 0.4192145916135169,
      "learning_rate": 8.684839230377475e-06,
      "loss": 0.0324,
      "step": 1953
    },
    {
      "epoch": 0.9385206532180596,
      "grad_norm": 0.5489777561757881,
      "learning_rate": 8.682949765239636e-06,
      "loss": 0.0357,
      "step": 1954
    },
    {
      "epoch": 0.9390009606147934,
      "grad_norm": 0.34093817269965004,
      "learning_rate": 8.681059149650181e-06,
      "loss": 0.0268,
      "step": 1955
    },
    {
      "epoch": 0.9394812680115274,
      "grad_norm": 0.395654696038833,
      "learning_rate": 8.679167384199686e-06,
      "loss": 0.0286,
      "step": 1956
    },
    {
      "epoch": 0.9399615754082613,
      "grad_norm": 0.2944331212675771,
      "learning_rate": 8.677274469479083e-06,
      "loss": 0.0263,
      "step": 1957
    },
    {
      "epoch": 0.9404418828049952,
      "grad_norm": 0.6932183759828536,
      "learning_rate": 8.67538040607967e-06,
      "loss": 0.0459,
      "step": 1958
    },
    {
      "epoch": 0.9409221902017291,
      "grad_norm": 0.9103586040099304,
      "learning_rate": 8.6734851945931e-06,
      "loss": 0.046,
      "step": 1959
    },
    {
      "epoch": 0.9414024975984631,
      "grad_norm": 0.4687709996696501,
      "learning_rate": 8.671588835611381e-06,
      "loss": 0.0448,
      "step": 1960
    },
    {
      "epoch": 0.9418828049951969,
      "grad_norm": 0.9039414492489983,
      "learning_rate": 8.669691329726888e-06,
      "loss": 0.0412,
      "step": 1961
    },
    {
      "epoch": 0.9423631123919308,
      "grad_norm": 0.5555653981562542,
      "learning_rate": 8.667792677532346e-06,
      "loss": 0.0339,
      "step": 1962
    },
    {
      "epoch": 0.9428434197886647,
      "grad_norm": 0.444243469483319,
      "learning_rate": 8.665892879620843e-06,
      "loss": 0.0427,
      "step": 1963
    },
    {
      "epoch": 0.9433237271853987,
      "grad_norm": 0.342635961775897,
      "learning_rate": 8.663991936585821e-06,
      "loss": 0.0337,
      "step": 1964
    },
    {
      "epoch": 0.9438040345821326,
      "grad_norm": 0.6397758038332509,
      "learning_rate": 8.662089849021086e-06,
      "loss": 0.0505,
      "step": 1965
    },
    {
      "epoch": 0.9442843419788665,
      "grad_norm": 0.5964279866861497,
      "learning_rate": 8.660186617520792e-06,
      "loss": 0.0358,
      "step": 1966
    },
    {
      "epoch": 0.9447646493756003,
      "grad_norm": 0.5659109590321414,
      "learning_rate": 8.658282242679461e-06,
      "loss": 0.0418,
      "step": 1967
    },
    {
      "epoch": 0.9452449567723343,
      "grad_norm": 0.506925916793818,
      "learning_rate": 8.656376725091965e-06,
      "loss": 0.0417,
      "step": 1968
    },
    {
      "epoch": 0.9457252641690682,
      "grad_norm": 0.48404249293897045,
      "learning_rate": 8.654470065353535e-06,
      "loss": 0.0429,
      "step": 1969
    },
    {
      "epoch": 0.9462055715658021,
      "grad_norm": 1.6891950338293646,
      "learning_rate": 8.652562264059758e-06,
      "loss": 0.0376,
      "step": 1970
    },
    {
      "epoch": 0.946685878962536,
      "grad_norm": 1.4094881449650138,
      "learning_rate": 8.65065332180658e-06,
      "loss": 0.0351,
      "step": 1971
    },
    {
      "epoch": 0.94716618635927,
      "grad_norm": 0.5177625328838332,
      "learning_rate": 8.6487432391903e-06,
      "loss": 0.0343,
      "step": 1972
    },
    {
      "epoch": 0.9476464937560038,
      "grad_norm": 0.5743763929907509,
      "learning_rate": 8.646832016807576e-06,
      "loss": 0.0366,
      "step": 1973
    },
    {
      "epoch": 0.9481268011527377,
      "grad_norm": 1.2258062310893443,
      "learning_rate": 8.644919655255421e-06,
      "loss": 0.0508,
      "step": 1974
    },
    {
      "epoch": 0.9486071085494716,
      "grad_norm": 0.9221165067101089,
      "learning_rate": 8.643006155131204e-06,
      "loss": 0.0673,
      "step": 1975
    },
    {
      "epoch": 0.9490874159462056,
      "grad_norm": 0.6966238615893148,
      "learning_rate": 8.641091517032648e-06,
      "loss": 0.0738,
      "step": 1976
    },
    {
      "epoch": 0.9495677233429395,
      "grad_norm": 0.4344930032517037,
      "learning_rate": 8.639175741557835e-06,
      "loss": 0.0326,
      "step": 1977
    },
    {
      "epoch": 0.9500480307396734,
      "grad_norm": 0.7423186859550626,
      "learning_rate": 8.6372588293052e-06,
      "loss": 0.0446,
      "step": 1978
    },
    {
      "epoch": 0.9505283381364072,
      "grad_norm": 0.4731476879098025,
      "learning_rate": 8.635340780873531e-06,
      "loss": 0.0353,
      "step": 1979
    },
    {
      "epoch": 0.9510086455331412,
      "grad_norm": 0.6600870526226297,
      "learning_rate": 8.633421596861977e-06,
      "loss": 0.0569,
      "step": 1980
    },
    {
      "epoch": 0.9514889529298751,
      "grad_norm": 0.5036533651424044,
      "learning_rate": 8.631501277870034e-06,
      "loss": 0.0406,
      "step": 1981
    },
    {
      "epoch": 0.951969260326609,
      "grad_norm": 1.3169404498677184,
      "learning_rate": 8.62957982449756e-06,
      "loss": 0.0786,
      "step": 1982
    },
    {
      "epoch": 0.952449567723343,
      "grad_norm": 0.7672144324242122,
      "learning_rate": 8.627657237344762e-06,
      "loss": 0.0437,
      "step": 1983
    },
    {
      "epoch": 0.9529298751200769,
      "grad_norm": 0.41972977550479074,
      "learning_rate": 8.625733517012202e-06,
      "loss": 0.0477,
      "step": 1984
    },
    {
      "epoch": 0.9534101825168108,
      "grad_norm": 0.507878350787633,
      "learning_rate": 8.6238086641008e-06,
      "loss": 0.0367,
      "step": 1985
    },
    {
      "epoch": 0.9538904899135446,
      "grad_norm": 0.516358838359933,
      "learning_rate": 8.621882679211826e-06,
      "loss": 0.0478,
      "step": 1986
    },
    {
      "epoch": 0.9543707973102786,
      "grad_norm": 0.3726975644915236,
      "learning_rate": 8.619955562946902e-06,
      "loss": 0.0341,
      "step": 1987
    },
    {
      "epoch": 0.9548511047070125,
      "grad_norm": 0.5774338343855023,
      "learning_rate": 8.618027315908009e-06,
      "loss": 0.0454,
      "step": 1988
    },
    {
      "epoch": 0.9553314121037464,
      "grad_norm": 0.710615004605156,
      "learning_rate": 8.616097938697476e-06,
      "loss": 0.0448,
      "step": 1989
    },
    {
      "epoch": 0.9558117195004803,
      "grad_norm": 0.9510665106194851,
      "learning_rate": 8.614167431917986e-06,
      "loss": 0.0543,
      "step": 1990
    },
    {
      "epoch": 0.9562920268972143,
      "grad_norm": 0.4147522768624343,
      "learning_rate": 8.612235796172579e-06,
      "loss": 0.0328,
      "step": 1991
    },
    {
      "epoch": 0.9567723342939481,
      "grad_norm": 0.5983476792554373,
      "learning_rate": 8.610303032064642e-06,
      "loss": 0.0606,
      "step": 1992
    },
    {
      "epoch": 0.957252641690682,
      "grad_norm": 0.3768025246044265,
      "learning_rate": 8.60836914019792e-06,
      "loss": 0.0323,
      "step": 1993
    },
    {
      "epoch": 0.9577329490874159,
      "grad_norm": 0.5553521298460753,
      "learning_rate": 8.606434121176504e-06,
      "loss": 0.0502,
      "step": 1994
    },
    {
      "epoch": 0.9582132564841499,
      "grad_norm": 0.6171063653286382,
      "learning_rate": 8.60449797560484e-06,
      "loss": 0.0505,
      "step": 1995
    },
    {
      "epoch": 0.9586935638808838,
      "grad_norm": 0.5729885075449906,
      "learning_rate": 8.60256070408773e-06,
      "loss": 0.0534,
      "step": 1996
    },
    {
      "epoch": 0.9591738712776177,
      "grad_norm": 0.4954254466131748,
      "learning_rate": 8.600622307230323e-06,
      "loss": 0.0339,
      "step": 1997
    },
    {
      "epoch": 0.9596541786743515,
      "grad_norm": 0.36497637328284854,
      "learning_rate": 8.598682785638119e-06,
      "loss": 0.0293,
      "step": 1998
    },
    {
      "epoch": 0.9601344860710855,
      "grad_norm": 0.5588614544226611,
      "learning_rate": 8.59674213991697e-06,
      "loss": 0.0497,
      "step": 1999
    },
    {
      "epoch": 0.9606147934678194,
      "grad_norm": 0.34528031432849365,
      "learning_rate": 8.594800370673083e-06,
      "loss": 0.032,
      "step": 2000
    },
    {
      "epoch": 0.9610951008645533,
      "grad_norm": 0.6906177073677481,
      "learning_rate": 8.592857478513011e-06,
      "loss": 0.0485,
      "step": 2001
    },
    {
      "epoch": 0.9615754082612872,
      "grad_norm": 0.3546432685965284,
      "learning_rate": 8.590913464043661e-06,
      "loss": 0.0297,
      "step": 2002
    },
    {
      "epoch": 0.9620557156580212,
      "grad_norm": 0.7379863818566214,
      "learning_rate": 8.58896832787229e-06,
      "loss": 0.0551,
      "step": 2003
    },
    {
      "epoch": 0.962536023054755,
      "grad_norm": 0.43001383506184815,
      "learning_rate": 8.5870220706065e-06,
      "loss": 0.0345,
      "step": 2004
    },
    {
      "epoch": 0.9630163304514889,
      "grad_norm": 0.40952514169054416,
      "learning_rate": 8.585074692854254e-06,
      "loss": 0.0387,
      "step": 2005
    },
    {
      "epoch": 0.9634966378482228,
      "grad_norm": 0.5621926001572276,
      "learning_rate": 8.583126195223854e-06,
      "loss": 0.0453,
      "step": 2006
    },
    {
      "epoch": 0.9639769452449568,
      "grad_norm": 0.6202940152268431,
      "learning_rate": 8.581176578323962e-06,
      "loss": 0.0455,
      "step": 2007
    },
    {
      "epoch": 0.9644572526416907,
      "grad_norm": 0.4253832733284014,
      "learning_rate": 8.579225842763578e-06,
      "loss": 0.038,
      "step": 2008
    },
    {
      "epoch": 0.9649375600384246,
      "grad_norm": 0.5301193530668448,
      "learning_rate": 8.577273989152063e-06,
      "loss": 0.045,
      "step": 2009
    },
    {
      "epoch": 0.9654178674351584,
      "grad_norm": 0.4447740781906591,
      "learning_rate": 8.575321018099122e-06,
      "loss": 0.0463,
      "step": 2010
    },
    {
      "epoch": 0.9658981748318924,
      "grad_norm": 0.5647765842461724,
      "learning_rate": 8.573366930214807e-06,
      "loss": 0.0382,
      "step": 2011
    },
    {
      "epoch": 0.9663784822286263,
      "grad_norm": 0.3880514422764617,
      "learning_rate": 8.571411726109518e-06,
      "loss": 0.0406,
      "step": 2012
    },
    {
      "epoch": 0.9668587896253602,
      "grad_norm": 0.49001392319362364,
      "learning_rate": 8.569455406394013e-06,
      "loss": 0.0379,
      "step": 2013
    },
    {
      "epoch": 0.9673390970220941,
      "grad_norm": 0.5021370712990999,
      "learning_rate": 8.567497971679387e-06,
      "loss": 0.0422,
      "step": 2014
    },
    {
      "epoch": 0.9678194044188281,
      "grad_norm": 0.46323226363095604,
      "learning_rate": 8.565539422577093e-06,
      "loss": 0.0404,
      "step": 2015
    },
    {
      "epoch": 0.968299711815562,
      "grad_norm": 0.41870207201506,
      "learning_rate": 8.563579759698925e-06,
      "loss": 0.0374,
      "step": 2016
    },
    {
      "epoch": 0.9687800192122958,
      "grad_norm": 0.45414111344706365,
      "learning_rate": 8.561618983657028e-06,
      "loss": 0.0418,
      "step": 2017
    },
    {
      "epoch": 0.9692603266090298,
      "grad_norm": 0.5305044501921113,
      "learning_rate": 8.559657095063893e-06,
      "loss": 0.0416,
      "step": 2018
    },
    {
      "epoch": 0.9697406340057637,
      "grad_norm": 0.5045574789523771,
      "learning_rate": 8.557694094532361e-06,
      "loss": 0.0324,
      "step": 2019
    },
    {
      "epoch": 0.9702209414024976,
      "grad_norm": 0.4492848331956523,
      "learning_rate": 8.555729982675619e-06,
      "loss": 0.0434,
      "step": 2020
    },
    {
      "epoch": 0.9707012487992315,
      "grad_norm": 0.4461437770479501,
      "learning_rate": 8.5537647601072e-06,
      "loss": 0.0363,
      "step": 2021
    },
    {
      "epoch": 0.9711815561959655,
      "grad_norm": 0.3915155679851884,
      "learning_rate": 8.551798427440985e-06,
      "loss": 0.0325,
      "step": 2022
    },
    {
      "epoch": 0.9716618635926993,
      "grad_norm": 0.4611826440449828,
      "learning_rate": 8.549830985291206e-06,
      "loss": 0.0485,
      "step": 2023
    },
    {
      "epoch": 0.9721421709894332,
      "grad_norm": 0.37428792596353483,
      "learning_rate": 8.547862434272431e-06,
      "loss": 0.0351,
      "step": 2024
    },
    {
      "epoch": 0.9726224783861671,
      "grad_norm": 0.323502628978833,
      "learning_rate": 8.545892774999589e-06,
      "loss": 0.0274,
      "step": 2025
    },
    {
      "epoch": 0.9731027857829011,
      "grad_norm": 0.6817640871809518,
      "learning_rate": 8.543922008087938e-06,
      "loss": 0.0429,
      "step": 2026
    },
    {
      "epoch": 0.973583093179635,
      "grad_norm": 0.503635291249707,
      "learning_rate": 8.541950134153099e-06,
      "loss": 0.0365,
      "step": 2027
    },
    {
      "epoch": 0.9740634005763689,
      "grad_norm": 0.5092392383162542,
      "learning_rate": 8.539977153811024e-06,
      "loss": 0.0412,
      "step": 2028
    },
    {
      "epoch": 0.9745437079731027,
      "grad_norm": 0.5178003304544473,
      "learning_rate": 8.538003067678022e-06,
      "loss": 0.0554,
      "step": 2029
    },
    {
      "epoch": 0.9750240153698367,
      "grad_norm": 0.35350741688529513,
      "learning_rate": 8.536027876370743e-06,
      "loss": 0.0346,
      "step": 2030
    },
    {
      "epoch": 0.9755043227665706,
      "grad_norm": 0.36618272757164555,
      "learning_rate": 8.53405158050618e-06,
      "loss": 0.0351,
      "step": 2031
    },
    {
      "epoch": 0.9759846301633045,
      "grad_norm": 0.4057826094482221,
      "learning_rate": 8.532074180701674e-06,
      "loss": 0.0394,
      "step": 2032
    },
    {
      "epoch": 0.9764649375600384,
      "grad_norm": 0.5037133903672764,
      "learning_rate": 8.53009567757491e-06,
      "loss": 0.0353,
      "step": 2033
    },
    {
      "epoch": 0.9769452449567724,
      "grad_norm": 0.45708751529472025,
      "learning_rate": 8.528116071743917e-06,
      "loss": 0.039,
      "step": 2034
    },
    {
      "epoch": 0.9774255523535063,
      "grad_norm": 0.4474137964593103,
      "learning_rate": 8.52613536382707e-06,
      "loss": 0.0482,
      "step": 2035
    },
    {
      "epoch": 0.9779058597502401,
      "grad_norm": 0.8437380174457212,
      "learning_rate": 8.524153554443088e-06,
      "loss": 0.0457,
      "step": 2036
    },
    {
      "epoch": 0.978386167146974,
      "grad_norm": 0.4317132378185624,
      "learning_rate": 8.522170644211032e-06,
      "loss": 0.0433,
      "step": 2037
    },
    {
      "epoch": 0.978866474543708,
      "grad_norm": 0.5306878906801935,
      "learning_rate": 8.520186633750309e-06,
      "loss": 0.0442,
      "step": 2038
    },
    {
      "epoch": 0.9793467819404419,
      "grad_norm": 0.4430347167684453,
      "learning_rate": 8.518201523680668e-06,
      "loss": 0.0472,
      "step": 2039
    },
    {
      "epoch": 0.9798270893371758,
      "grad_norm": 0.40367863591910896,
      "learning_rate": 8.516215314622203e-06,
      "loss": 0.0547,
      "step": 2040
    },
    {
      "epoch": 0.9803073967339097,
      "grad_norm": 1.0144384311645978,
      "learning_rate": 8.514228007195354e-06,
      "loss": 0.0632,
      "step": 2041
    },
    {
      "epoch": 0.9807877041306436,
      "grad_norm": 0.6255894136401294,
      "learning_rate": 8.512239602020894e-06,
      "loss": 0.0374,
      "step": 2042
    },
    {
      "epoch": 0.9812680115273775,
      "grad_norm": 0.29213416072835724,
      "learning_rate": 8.510250099719953e-06,
      "loss": 0.0302,
      "step": 2043
    },
    {
      "epoch": 0.9817483189241114,
      "grad_norm": 0.48617017713788496,
      "learning_rate": 8.50825950091399e-06,
      "loss": 0.049,
      "step": 2044
    },
    {
      "epoch": 0.9822286263208453,
      "grad_norm": 0.5170970045891808,
      "learning_rate": 8.506267806224817e-06,
      "loss": 0.0418,
      "step": 2045
    },
    {
      "epoch": 0.9827089337175793,
      "grad_norm": 0.5619222767720121,
      "learning_rate": 8.504275016274584e-06,
      "loss": 0.0477,
      "step": 2046
    },
    {
      "epoch": 0.9831892411143132,
      "grad_norm": 0.34210045056579325,
      "learning_rate": 8.502281131685783e-06,
      "loss": 0.0352,
      "step": 2047
    },
    {
      "epoch": 0.983669548511047,
      "grad_norm": 0.4991446492244405,
      "learning_rate": 8.500286153081248e-06,
      "loss": 0.0301,
      "step": 2048
    },
    {
      "epoch": 0.984149855907781,
      "grad_norm": 0.44922768509123856,
      "learning_rate": 8.498290081084156e-06,
      "loss": 0.0341,
      "step": 2049
    },
    {
      "epoch": 0.9846301633045149,
      "grad_norm": 0.307944880192032,
      "learning_rate": 8.496292916318023e-06,
      "loss": 0.025,
      "step": 2050
    },
    {
      "epoch": 0.9851104707012488,
      "grad_norm": 0.5664099644492291,
      "learning_rate": 8.494294659406709e-06,
      "loss": 0.0487,
      "step": 2051
    },
    {
      "epoch": 0.9855907780979827,
      "grad_norm": 0.7989706431633349,
      "learning_rate": 8.492295310974416e-06,
      "loss": 0.0515,
      "step": 2052
    },
    {
      "epoch": 0.9860710854947167,
      "grad_norm": 0.4172068934478134,
      "learning_rate": 8.490294871645681e-06,
      "loss": 0.037,
      "step": 2053
    },
    {
      "epoch": 0.9865513928914506,
      "grad_norm": 0.44711071451716344,
      "learning_rate": 8.488293342045391e-06,
      "loss": 0.0358,
      "step": 2054
    },
    {
      "epoch": 0.9870317002881844,
      "grad_norm": 0.394037791565088,
      "learning_rate": 8.486290722798765e-06,
      "loss": 0.0352,
      "step": 2055
    },
    {
      "epoch": 0.9875120076849183,
      "grad_norm": 0.44258059109076087,
      "learning_rate": 8.484287014531366e-06,
      "loss": 0.05,
      "step": 2056
    },
    {
      "epoch": 0.9879923150816523,
      "grad_norm": 0.4913903621371682,
      "learning_rate": 8.482282217869096e-06,
      "loss": 0.0418,
      "step": 2057
    },
    {
      "epoch": 0.9884726224783862,
      "grad_norm": 0.5964000018471872,
      "learning_rate": 8.480276333438203e-06,
      "loss": 0.0394,
      "step": 2058
    },
    {
      "epoch": 0.9889529298751201,
      "grad_norm": 0.42025505935398855,
      "learning_rate": 8.478269361865264e-06,
      "loss": 0.0315,
      "step": 2059
    },
    {
      "epoch": 0.989433237271854,
      "grad_norm": 0.3825968875041374,
      "learning_rate": 8.476261303777205e-06,
      "loss": 0.0267,
      "step": 2060
    },
    {
      "epoch": 0.9899135446685879,
      "grad_norm": 0.5266897373376677,
      "learning_rate": 8.474252159801287e-06,
      "loss": 0.0542,
      "step": 2061
    },
    {
      "epoch": 0.9903938520653218,
      "grad_norm": 0.9138199723571926,
      "learning_rate": 8.472241930565108e-06,
      "loss": 0.043,
      "step": 2062
    },
    {
      "epoch": 0.9908741594620557,
      "grad_norm": 0.4491153125193229,
      "learning_rate": 8.470230616696613e-06,
      "loss": 0.0416,
      "step": 2063
    },
    {
      "epoch": 0.9913544668587896,
      "grad_norm": 0.39648825911272,
      "learning_rate": 8.468218218824078e-06,
      "loss": 0.0308,
      "step": 2064
    },
    {
      "epoch": 0.9918347742555236,
      "grad_norm": 0.4266101634870432,
      "learning_rate": 8.46620473757612e-06,
      "loss": 0.0398,
      "step": 2065
    },
    {
      "epoch": 0.9923150816522575,
      "grad_norm": 0.46728827118397764,
      "learning_rate": 8.464190173581698e-06,
      "loss": 0.0332,
      "step": 2066
    },
    {
      "epoch": 0.9927953890489913,
      "grad_norm": 0.402026280201352,
      "learning_rate": 8.462174527470102e-06,
      "loss": 0.0306,
      "step": 2067
    },
    {
      "epoch": 0.9932756964457252,
      "grad_norm": 1.2753342168137747,
      "learning_rate": 8.460157799870967e-06,
      "loss": 0.0422,
      "step": 2068
    },
    {
      "epoch": 0.9937560038424592,
      "grad_norm": 0.48784901997970864,
      "learning_rate": 8.45813999141426e-06,
      "loss": 0.045,
      "step": 2069
    },
    {
      "epoch": 0.9942363112391931,
      "grad_norm": 0.4598639642272694,
      "learning_rate": 8.456121102730293e-06,
      "loss": 0.0339,
      "step": 2070
    },
    {
      "epoch": 0.994716618635927,
      "grad_norm": 0.43585915181157464,
      "learning_rate": 8.454101134449706e-06,
      "loss": 0.0495,
      "step": 2071
    },
    {
      "epoch": 0.9951969260326609,
      "grad_norm": 0.47450866187493235,
      "learning_rate": 8.452080087203484e-06,
      "loss": 0.0454,
      "step": 2072
    },
    {
      "epoch": 0.9956772334293948,
      "grad_norm": 0.40100519344897884,
      "learning_rate": 8.450057961622945e-06,
      "loss": 0.032,
      "step": 2073
    },
    {
      "epoch": 0.9961575408261287,
      "grad_norm": 0.3834770013645444,
      "learning_rate": 8.448034758339747e-06,
      "loss": 0.0366,
      "step": 2074
    },
    {
      "epoch": 0.9966378482228626,
      "grad_norm": 0.4018989482709451,
      "learning_rate": 8.446010477985882e-06,
      "loss": 0.0434,
      "step": 2075
    },
    {
      "epoch": 0.9971181556195965,
      "grad_norm": 0.44809689260265556,
      "learning_rate": 8.443985121193679e-06,
      "loss": 0.0464,
      "step": 2076
    },
    {
      "epoch": 0.9975984630163305,
      "grad_norm": 0.2905562947958189,
      "learning_rate": 8.441958688595802e-06,
      "loss": 0.0282,
      "step": 2077
    },
    {
      "epoch": 0.9980787704130644,
      "grad_norm": 0.7881415875093454,
      "learning_rate": 8.439931180825253e-06,
      "loss": 0.037,
      "step": 2078
    },
    {
      "epoch": 0.9985590778097982,
      "grad_norm": 0.38631353813613173,
      "learning_rate": 8.437902598515371e-06,
      "loss": 0.0406,
      "step": 2079
    },
    {
      "epoch": 0.9990393852065321,
      "grad_norm": 0.38950921399372224,
      "learning_rate": 8.435872942299827e-06,
      "loss": 0.0348,
      "step": 2080
    },
    {
      "epoch": 0.9995196926032661,
      "grad_norm": 0.5056529616402811,
      "learning_rate": 8.433842212812632e-06,
      "loss": 0.0509,
      "step": 2081
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5020057590352653,
      "learning_rate": 8.431810410688126e-06,
      "loss": 0.0373,
      "step": 2082
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.03837773576378822,
      "eval_runtime": 509.5894,
      "eval_samples_per_second": 32.828,
      "eval_steps_per_second": 1.026,
      "step": 2082
    }
  ],
  "logging_steps": 1,
  "max_steps": 6246,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 7.075959801947095e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}