{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 18531,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.3963628514381306e-05,
      "grad_norm": 20.62471154119197,
      "learning_rate": 9.999999935332727e-06,
      "loss": 1.1326,
      "step": 1
    },
    {
      "epoch": 5.3963628514381306e-05,
      "eval_loss": 1.2850651741027832,
      "eval_runtime": 163.9451,
      "eval_samples_per_second": 20.977,
      "eval_steps_per_second": 0.878,
      "step": 1
    },
    {
      "epoch": 0.00010792725702876261,
      "grad_norm": 11.978215884570838,
      "learning_rate": 9.99999974133091e-06,
      "loss": 0.69,
      "step": 2
    },
    {
      "epoch": 0.00016189088554314392,
      "grad_norm": 2.7920195200003564,
      "learning_rate": 9.999999417994552e-06,
      "loss": 0.8313,
      "step": 3
    },
    {
      "epoch": 0.00021585451405752522,
      "grad_norm": 2.495653632936459,
      "learning_rate": 9.999998965323662e-06,
      "loss": 0.7776,
      "step": 4
    },
    {
      "epoch": 0.00026981814257190655,
      "grad_norm": 2.9616137141489753,
      "learning_rate": 9.999998383318257e-06,
      "loss": 0.8333,
      "step": 5
    },
    {
      "epoch": 0.00032378177108628785,
      "grad_norm": 1.6624487327415252,
      "learning_rate": 9.999997671978352e-06,
      "loss": 0.6244,
      "step": 6
    },
    {
      "epoch": 0.00037774539960066915,
      "grad_norm": 1.6137268553442474,
      "learning_rate": 9.999996831303965e-06,
      "loss": 0.6916,
      "step": 7
    },
    {
      "epoch": 0.00043170902811505045,
      "grad_norm": 1.4937969414472936,
      "learning_rate": 9.999995861295124e-06,
      "loss": 0.7256,
      "step": 8
    },
    {
      "epoch": 0.00048567265662943174,
      "grad_norm": 1.5617292166348637,
      "learning_rate": 9.999994761951853e-06,
      "loss": 0.682,
      "step": 9
    },
    {
      "epoch": 0.0005396362851438131,
      "grad_norm": 1.378846924064355,
      "learning_rate": 9.999993533274188e-06,
      "loss": 0.5825,
      "step": 10
    },
    {
      "epoch": 0.0005935999136581943,
      "grad_norm": 1.5976562369786855,
      "learning_rate": 9.999992175262161e-06,
      "loss": 0.5589,
      "step": 11
    },
    {
      "epoch": 0.0006475635421725757,
      "grad_norm": 1.6012385037210424,
      "learning_rate": 9.999990687915811e-06,
      "loss": 0.6812,
      "step": 12
    },
    {
      "epoch": 0.0007015271706869569,
      "grad_norm": 1.2509010209304539,
      "learning_rate": 9.999989071235184e-06,
      "loss": 0.6066,
      "step": 13
    },
    {
      "epoch": 0.0007554907992013383,
      "grad_norm": 1.8035441373911127,
      "learning_rate": 9.999987325220323e-06,
      "loss": 0.7727,
      "step": 14
    },
    {
      "epoch": 0.0008094544277157196,
      "grad_norm": 1.3921621567950277,
      "learning_rate": 9.999985449871278e-06,
      "loss": 0.4671,
      "step": 15
    },
    {
      "epoch": 0.0008634180562301009,
      "grad_norm": 1.6855287366327114,
      "learning_rate": 9.999983445188106e-06,
      "loss": 0.7438,
      "step": 16
    },
    {
      "epoch": 0.0009173816847444822,
      "grad_norm": 1.3435022323213992,
      "learning_rate": 9.999981311170862e-06,
      "loss": 0.6278,
      "step": 17
    },
    {
      "epoch": 0.0009713453132588635,
      "grad_norm": 1.2677017155543135,
      "learning_rate": 9.99997904781961e-06,
      "loss": 0.5984,
      "step": 18
    },
    {
      "epoch": 0.0010253089417732448,
      "grad_norm": 1.6430046461611512,
      "learning_rate": 9.99997665513441e-06,
      "loss": 0.88,
      "step": 19
    },
    {
      "epoch": 0.0010792725702876262,
      "grad_norm": 1.4463352162823673,
      "learning_rate": 9.999974133115337e-06,
      "loss": 0.6748,
      "step": 20
    },
    {
      "epoch": 0.0011332361988020075,
      "grad_norm": 1.5050398589101073,
      "learning_rate": 9.999971481762459e-06,
      "loss": 0.6264,
      "step": 21
    },
    {
      "epoch": 0.0011871998273163887,
      "grad_norm": 1.5644834623203105,
      "learning_rate": 9.999968701075853e-06,
      "loss": 0.6322,
      "step": 22
    },
    {
      "epoch": 0.00124116345583077,
      "grad_norm": 1.3403191435319328,
      "learning_rate": 9.999965791055601e-06,
      "loss": 0.5249,
      "step": 23
    },
    {
      "epoch": 0.0012951270843451514,
      "grad_norm": 1.493128153009394,
      "learning_rate": 9.999962751701786e-06,
      "loss": 0.7377,
      "step": 24
    },
    {
      "epoch": 0.0013490907128595327,
      "grad_norm": 1.1182061816796247,
      "learning_rate": 9.999959583014494e-06,
      "loss": 0.5741,
      "step": 25
    },
    {
      "epoch": 0.0014030543413739139,
      "grad_norm": 1.2693796235233346,
      "learning_rate": 9.999956284993815e-06,
      "loss": 0.5187,
      "step": 26
    },
    {
      "epoch": 0.0014570179698882952,
      "grad_norm": 1.0740755829937843,
      "learning_rate": 9.999952857639847e-06,
      "loss": 0.4692,
      "step": 27
    },
    {
      "epoch": 0.0015109815984026766,
      "grad_norm": 1.4554543012598844,
      "learning_rate": 9.999949300952688e-06,
      "loss": 0.6432,
      "step": 28
    },
    {
      "epoch": 0.001564945226917058,
      "grad_norm": 1.5263483734100178,
      "learning_rate": 9.999945614932437e-06,
      "loss": 0.5959,
      "step": 29
    },
    {
      "epoch": 0.0016189088554314393,
      "grad_norm": 1.3596137256673293,
      "learning_rate": 9.999941799579203e-06,
      "loss": 0.754,
      "step": 30
    },
    {
      "epoch": 0.0016728724839458204,
      "grad_norm": 0.9196760705015845,
      "learning_rate": 9.999937854893096e-06,
      "loss": 0.4175,
      "step": 31
    },
    {
      "epoch": 0.0017268361124602018,
      "grad_norm": 1.2698619328595582,
      "learning_rate": 9.999933780874227e-06,
      "loss": 0.5936,
      "step": 32
    },
    {
      "epoch": 0.0017807997409745831,
      "grad_norm": 1.1624589243802812,
      "learning_rate": 9.999929577522715e-06,
      "loss": 0.6055,
      "step": 33
    },
    {
      "epoch": 0.0018347633694889645,
      "grad_norm": 1.1733684912023161,
      "learning_rate": 9.999925244838679e-06,
      "loss": 0.7958,
      "step": 34
    },
    {
      "epoch": 0.0018887269980033458,
      "grad_norm": 1.4988701157981699,
      "learning_rate": 9.999920782822246e-06,
      "loss": 0.7344,
      "step": 35
    },
    {
      "epoch": 0.001942690626517727,
      "grad_norm": 1.1064293874678428,
      "learning_rate": 9.999916191473542e-06,
      "loss": 0.4257,
      "step": 36
    },
    {
      "epoch": 0.0019966542550321083,
      "grad_norm": 1.098798309417731,
      "learning_rate": 9.9999114707927e-06,
      "loss": 0.4599,
      "step": 37
    },
    {
      "epoch": 0.0020506178835464897,
      "grad_norm": 1.2564662535885602,
      "learning_rate": 9.999906620779854e-06,
      "loss": 0.5248,
      "step": 38
    },
    {
      "epoch": 0.002104581512060871,
      "grad_norm": 1.1966810881700105,
      "learning_rate": 9.999901641435147e-06,
      "loss": 0.5687,
      "step": 39
    },
    {
      "epoch": 0.0021585451405752524,
      "grad_norm": 1.1032986453241542,
      "learning_rate": 9.99989653275872e-06,
      "loss": 0.5577,
      "step": 40
    },
    {
      "epoch": 0.0022125087690896337,
      "grad_norm": 1.1920229090432144,
      "learning_rate": 9.999891294750719e-06,
      "loss": 0.6813,
      "step": 41
    },
    {
      "epoch": 0.002266472397604015,
      "grad_norm": 1.5572389614879893,
      "learning_rate": 9.999885927411296e-06,
      "loss": 0.6559,
      "step": 42
    },
    {
      "epoch": 0.002320436026118396,
      "grad_norm": 1.4067941217710171,
      "learning_rate": 9.999880430740603e-06,
      "loss": 0.6079,
      "step": 43
    },
    {
      "epoch": 0.0023743996546327774,
      "grad_norm": 1.2737559465331796,
      "learning_rate": 9.999874804738801e-06,
      "loss": 0.6533,
      "step": 44
    },
    {
      "epoch": 0.0024283632831471587,
      "grad_norm": 1.1590656997016004,
      "learning_rate": 9.99986904940605e-06,
      "loss": 0.5842,
      "step": 45
    },
    {
      "epoch": 0.00248232691166154,
      "grad_norm": 1.3696445997560607,
      "learning_rate": 9.999863164742515e-06,
      "loss": 0.6552,
      "step": 46
    },
    {
      "epoch": 0.0025362905401759214,
      "grad_norm": 1.5172040261890776,
      "learning_rate": 9.999857150748368e-06,
      "loss": 0.7312,
      "step": 47
    },
    {
      "epoch": 0.0025902541686903028,
      "grad_norm": 1.2528565188917922,
      "learning_rate": 9.99985100742378e-06,
      "loss": 0.615,
      "step": 48
    },
    {
      "epoch": 0.002644217797204684,
      "grad_norm": 1.589497912926135,
      "learning_rate": 9.999844734768924e-06,
      "loss": 0.5708,
      "step": 49
    },
    {
      "epoch": 0.0026981814257190655,
      "grad_norm": 1.2507827369389508,
      "learning_rate": 9.999838332783986e-06,
      "loss": 0.5464,
      "step": 50
    },
    {
      "epoch": 0.002752145054233447,
      "grad_norm": 1.1368103015379805,
      "learning_rate": 9.999831801469148e-06,
      "loss": 0.5777,
      "step": 51
    },
    {
      "epoch": 0.0028061086827478278,
      "grad_norm": 1.414047683271421,
      "learning_rate": 9.999825140824596e-06,
      "loss": 0.4937,
      "step": 52
    },
    {
      "epoch": 0.002860072311262209,
      "grad_norm": 1.5281045248145013,
      "learning_rate": 9.999818350850521e-06,
      "loss": 0.5942,
      "step": 53
    },
    {
      "epoch": 0.0029140359397765905,
      "grad_norm": 1.2491596458976133,
      "learning_rate": 9.999811431547122e-06,
      "loss": 0.5425,
      "step": 54
    },
    {
      "epoch": 0.002967999568290972,
      "grad_norm": 1.2783137365804016,
      "learning_rate": 9.999804382914596e-06,
      "loss": 0.5798,
      "step": 55
    },
    {
      "epoch": 0.003021963196805353,
      "grad_norm": 1.3064064194461447,
      "learning_rate": 9.999797204953144e-06,
      "loss": 0.5842,
      "step": 56
    },
    {
      "epoch": 0.0030759268253197345,
      "grad_norm": 1.137432305904066,
      "learning_rate": 9.999789897662975e-06,
      "loss": 0.4564,
      "step": 57
    },
    {
      "epoch": 0.003129890453834116,
      "grad_norm": 1.3204701904729959,
      "learning_rate": 9.999782461044297e-06,
      "loss": 0.5736,
      "step": 58
    },
    {
      "epoch": 0.0031838540823484972,
      "grad_norm": 1.3948398219985885,
      "learning_rate": 9.999774895097325e-06,
      "loss": 0.7644,
      "step": 59
    },
    {
      "epoch": 0.0032378177108628786,
      "grad_norm": 1.174379552130365,
      "learning_rate": 9.999767199822274e-06,
      "loss": 0.4547,
      "step": 60
    },
    {
      "epoch": 0.00329178133937726,
      "grad_norm": 1.1842108746350566,
      "learning_rate": 9.999759375219368e-06,
      "loss": 0.5331,
      "step": 61
    },
    {
      "epoch": 0.003345744967891641,
      "grad_norm": 1.2289334165755277,
      "learning_rate": 9.999751421288831e-06,
      "loss": 0.6269,
      "step": 62
    },
    {
      "epoch": 0.003399708596406022,
      "grad_norm": 1.2482406798625927,
      "learning_rate": 9.999743338030893e-06,
      "loss": 0.5734,
      "step": 63
    },
    {
      "epoch": 0.0034536722249204036,
      "grad_norm": 1.2028190488134864,
      "learning_rate": 9.999735125445784e-06,
      "loss": 0.6226,
      "step": 64
    },
    {
      "epoch": 0.003507635853434785,
      "grad_norm": 1.3563974287565117,
      "learning_rate": 9.99972678353374e-06,
      "loss": 0.6946,
      "step": 65
    },
    {
      "epoch": 0.0035615994819491663,
      "grad_norm": 1.2969861284073565,
      "learning_rate": 9.999718312295001e-06,
      "loss": 0.608,
      "step": 66
    },
    {
      "epoch": 0.0036155631104635476,
      "grad_norm": 1.109704224926767,
      "learning_rate": 9.999709711729813e-06,
      "loss": 0.5284,
      "step": 67
    },
    {
      "epoch": 0.003669526738977929,
      "grad_norm": 1.0926354182967197,
      "learning_rate": 9.999700981838419e-06,
      "loss": 0.5231,
      "step": 68
    },
    {
      "epoch": 0.0037234903674923103,
      "grad_norm": 1.316966278875892,
      "learning_rate": 9.999692122621074e-06,
      "loss": 0.6912,
      "step": 69
    },
    {
      "epoch": 0.0037774539960066917,
      "grad_norm": 1.1028038098378286,
      "learning_rate": 9.99968313407803e-06,
      "loss": 0.5388,
      "step": 70
    },
    {
      "epoch": 0.0038314176245210726,
      "grad_norm": 1.4592761701618011,
      "learning_rate": 9.999674016209545e-06,
      "loss": 0.691,
      "step": 71
    },
    {
      "epoch": 0.003885381253035454,
      "grad_norm": 1.2055763924979754,
      "learning_rate": 9.999664769015885e-06,
      "loss": 0.8268,
      "step": 72
    },
    {
      "epoch": 0.003939344881549836,
      "grad_norm": 1.5208705766562787,
      "learning_rate": 9.999655392497311e-06,
      "loss": 0.5616,
      "step": 73
    },
    {
      "epoch": 0.003993308510064217,
      "grad_norm": 1.5467569289119585,
      "learning_rate": 9.999645886654094e-06,
      "loss": 0.6716,
      "step": 74
    },
    {
      "epoch": 0.0040472721385785984,
      "grad_norm": 1.1406202607003082,
      "learning_rate": 9.999636251486508e-06,
      "loss": 0.5915,
      "step": 75
    },
    {
      "epoch": 0.004101235767092979,
      "grad_norm": 1.2502422721883082,
      "learning_rate": 9.999626486994831e-06,
      "loss": 0.5715,
      "step": 76
    },
    {
      "epoch": 0.00415519939560736,
      "grad_norm": 1.24141874973768,
      "learning_rate": 9.999616593179342e-06,
      "loss": 0.6205,
      "step": 77
    },
    {
      "epoch": 0.004209163024121742,
      "grad_norm": 1.074528911526831,
      "learning_rate": 9.999606570040324e-06,
      "loss": 0.5283,
      "step": 78
    },
    {
      "epoch": 0.004263126652636123,
      "grad_norm": 1.183869723262245,
      "learning_rate": 9.999596417578067e-06,
      "loss": 0.6861,
      "step": 79
    },
    {
      "epoch": 0.004317090281150505,
      "grad_norm": 1.3348217393736452,
      "learning_rate": 9.999586135792863e-06,
      "loss": 0.7591,
      "step": 80
    },
    {
      "epoch": 0.004371053909664886,
      "grad_norm": 1.1321295692058997,
      "learning_rate": 9.999575724685008e-06,
      "loss": 0.5951,
      "step": 81
    },
    {
      "epoch": 0.0044250175381792675,
      "grad_norm": 1.1306837011159268,
      "learning_rate": 9.999565184254798e-06,
      "loss": 0.5639,
      "step": 82
    },
    {
      "epoch": 0.004478981166693648,
      "grad_norm": 1.382248266536327,
      "learning_rate": 9.99955451450254e-06,
      "loss": 0.6085,
      "step": 83
    },
    {
      "epoch": 0.00453294479520803,
      "grad_norm": 1.0991967140033378,
      "learning_rate": 9.99954371542854e-06,
      "loss": 0.6118,
      "step": 84
    },
    {
      "epoch": 0.004586908423722411,
      "grad_norm": 1.0619382352967455,
      "learning_rate": 9.999532787033103e-06,
      "loss": 0.6348,
      "step": 85
    },
    {
      "epoch": 0.004640872052236792,
      "grad_norm": 1.2339151264107624,
      "learning_rate": 9.999521729316549e-06,
      "loss": 0.5824,
      "step": 86
    },
    {
      "epoch": 0.004694835680751174,
      "grad_norm": 1.3638922454106583,
      "learning_rate": 9.999510542279196e-06,
      "loss": 0.6507,
      "step": 87
    },
    {
      "epoch": 0.004748799309265555,
      "grad_norm": 1.0353403520223565,
      "learning_rate": 9.999499225921362e-06,
      "loss": 0.4242,
      "step": 88
    },
    {
      "epoch": 0.0048027629377799365,
      "grad_norm": 1.1388152200198804,
      "learning_rate": 9.999487780243374e-06,
      "loss": 0.6553,
      "step": 89
    },
    {
      "epoch": 0.0048567265662943174,
      "grad_norm": 1.0088999050018588,
      "learning_rate": 9.999476205245559e-06,
      "loss": 0.5447,
      "step": 90
    },
    {
      "epoch": 0.004910690194808699,
      "grad_norm": 1.3343741994178577,
      "learning_rate": 9.999464500928253e-06,
      "loss": 0.7178,
      "step": 91
    },
    {
      "epoch": 0.00496465382332308,
      "grad_norm": 1.2193554718190278,
      "learning_rate": 9.999452667291792e-06,
      "loss": 0.6227,
      "step": 92
    },
    {
      "epoch": 0.005018617451837462,
      "grad_norm": 1.2707108112552563,
      "learning_rate": 9.999440704336514e-06,
      "loss": 0.503,
      "step": 93
    },
    {
      "epoch": 0.005072581080351843,
      "grad_norm": 0.8699897258652027,
      "learning_rate": 9.999428612062763e-06,
      "loss": 0.3978,
      "step": 94
    },
    {
      "epoch": 0.005126544708866224,
      "grad_norm": 1.0285133671896063,
      "learning_rate": 9.99941639047089e-06,
      "loss": 0.5334,
      "step": 95
    },
    {
      "epoch": 0.0051805083373806056,
      "grad_norm": 1.3935214679247594,
      "learning_rate": 9.99940403956124e-06,
      "loss": 0.6722,
      "step": 96
    },
    {
      "epoch": 0.0052344719658949865,
      "grad_norm": 1.1390441173917967,
      "learning_rate": 9.999391559334175e-06,
      "loss": 0.6349,
      "step": 97
    },
    {
      "epoch": 0.005288435594409368,
      "grad_norm": 1.215342377859159,
      "learning_rate": 9.999378949790047e-06,
      "loss": 0.5232,
      "step": 98
    },
    {
      "epoch": 0.005342399222923749,
      "grad_norm": 1.2510327002090098,
      "learning_rate": 9.999366210929225e-06,
      "loss": 0.5196,
      "step": 99
    },
    {
      "epoch": 0.005396362851438131,
      "grad_norm": 1.2511226145939793,
      "learning_rate": 9.99935334275207e-06,
      "loss": 0.4622,
      "step": 100
    },
    {
      "epoch": 0.005450326479952512,
      "grad_norm": 1.2557162363186618,
      "learning_rate": 9.999340345258954e-06,
      "loss": 0.7242,
      "step": 101
    },
    {
      "epoch": 0.005504290108466894,
      "grad_norm": 1.361854651875888,
      "learning_rate": 9.99932721845025e-06,
      "loss": 0.5739,
      "step": 102
    },
    {
      "epoch": 0.005558253736981275,
      "grad_norm": 1.0151654094784182,
      "learning_rate": 9.999313962326335e-06,
      "loss": 0.4747,
      "step": 103
    },
    {
      "epoch": 0.0056122173654956555,
      "grad_norm": 1.307395363961812,
      "learning_rate": 9.999300576887592e-06,
      "loss": 0.5408,
      "step": 104
    },
    {
      "epoch": 0.005666180994010037,
      "grad_norm": 0.9999351567265667,
      "learning_rate": 9.999287062134402e-06,
      "loss": 0.4409,
      "step": 105
    },
    {
      "epoch": 0.005720144622524418,
      "grad_norm": 1.4198227189438999,
      "learning_rate": 9.999273418067157e-06,
      "loss": 0.6053,
      "step": 106
    },
    {
      "epoch": 0.0057741082510388,
      "grad_norm": 1.0283706964326431,
      "learning_rate": 9.999259644686248e-06,
      "loss": 0.4966,
      "step": 107
    },
    {
      "epoch": 0.005828071879553181,
      "grad_norm": 1.3975065339436565,
      "learning_rate": 9.99924574199207e-06,
      "loss": 0.5233,
      "step": 108
    },
    {
      "epoch": 0.005882035508067563,
      "grad_norm": 1.564248936330875,
      "learning_rate": 9.999231709985024e-06,
      "loss": 0.7572,
      "step": 109
    },
    {
      "epoch": 0.005935999136581944,
      "grad_norm": 0.953825958818951,
      "learning_rate": 9.999217548665513e-06,
      "loss": 0.5019,
      "step": 110
    },
    {
      "epoch": 0.005989962765096325,
      "grad_norm": 1.0989677737430716,
      "learning_rate": 9.999203258033943e-06,
      "loss": 0.5178,
      "step": 111
    },
    {
      "epoch": 0.006043926393610706,
      "grad_norm": 1.0873136252986095,
      "learning_rate": 9.999188838090723e-06,
      "loss": 0.5897,
      "step": 112
    },
    {
      "epoch": 0.006097890022125087,
      "grad_norm": 1.29257268513247,
      "learning_rate": 9.999174288836271e-06,
      "loss": 0.5261,
      "step": 113
    },
    {
      "epoch": 0.006151853650639469,
      "grad_norm": 1.4362780268116098,
      "learning_rate": 9.999159610271006e-06,
      "loss": 0.6261,
      "step": 114
    },
    {
      "epoch": 0.00620581727915385,
      "grad_norm": 1.4334912621392732,
      "learning_rate": 9.999144802395347e-06,
      "loss": 0.6802,
      "step": 115
    },
    {
      "epoch": 0.006259780907668232,
      "grad_norm": 1.3026939446003196,
      "learning_rate": 9.999129865209719e-06,
      "loss": 0.6078,
      "step": 116
    },
    {
      "epoch": 0.006313744536182613,
      "grad_norm": 1.3023261900733292,
      "learning_rate": 9.999114798714553e-06,
      "loss": 0.6115,
      "step": 117
    },
    {
      "epoch": 0.0063677081646969945,
      "grad_norm": 1.1621921082069646,
      "learning_rate": 9.999099602910282e-06,
      "loss": 0.6119,
      "step": 118
    },
    {
      "epoch": 0.006421671793211375,
      "grad_norm": 1.1321372544966917,
      "learning_rate": 9.99908427779734e-06,
      "loss": 0.518,
      "step": 119
    },
    {
      "epoch": 0.006475635421725757,
      "grad_norm": 1.1943533030251596,
      "learning_rate": 9.999068823376174e-06,
      "loss": 0.6093,
      "step": 120
    },
    {
      "epoch": 0.006529599050240138,
      "grad_norm": 1.240932756137203,
      "learning_rate": 9.99905323964722e-06,
      "loss": 0.5732,
      "step": 121
    },
    {
      "epoch": 0.00658356267875452,
      "grad_norm": 1.2798199696856134,
      "learning_rate": 9.999037526610934e-06,
      "loss": 0.6628,
      "step": 122
    },
    {
      "epoch": 0.006637526307268901,
      "grad_norm": 1.0920364951156052,
      "learning_rate": 9.999021684267761e-06,
      "loss": 0.4364,
      "step": 123
    },
    {
      "epoch": 0.006691489935783282,
      "grad_norm": 0.8643131495764302,
      "learning_rate": 9.999005712618159e-06,
      "loss": 0.3643,
      "step": 124
    },
    {
      "epoch": 0.0067454535642976635,
      "grad_norm": 1.2001944603703765,
      "learning_rate": 9.998989611662588e-06,
      "loss": 0.5291,
      "step": 125
    },
    {
      "epoch": 0.006799417192812044,
      "grad_norm": 1.191811116474948,
      "learning_rate": 9.998973381401509e-06,
      "loss": 0.5764,
      "step": 126
    },
    {
      "epoch": 0.006853380821326426,
      "grad_norm": 1.054067689722635,
      "learning_rate": 9.998957021835389e-06,
      "loss": 0.4844,
      "step": 127
    },
    {
      "epoch": 0.006907344449840807,
      "grad_norm": 1.1270242027136157,
      "learning_rate": 9.998940532964699e-06,
      "loss": 0.4923,
      "step": 128
    },
    {
      "epoch": 0.006961308078355189,
      "grad_norm": 1.0726357001289062,
      "learning_rate": 9.998923914789911e-06,
      "loss": 0.5963,
      "step": 129
    },
    {
      "epoch": 0.00701527170686957,
      "grad_norm": 1.1435268473899969,
      "learning_rate": 9.998907167311504e-06,
      "loss": 0.5464,
      "step": 130
    },
    {
      "epoch": 0.007069235335383952,
      "grad_norm": 1.2686625720572489,
      "learning_rate": 9.99889029052996e-06,
      "loss": 0.7449,
      "step": 131
    },
    {
      "epoch": 0.0071231989638983325,
      "grad_norm": 1.0793489291384164,
      "learning_rate": 9.998873284445762e-06,
      "loss": 0.4533,
      "step": 132
    },
    {
      "epoch": 0.0071771625924127135,
      "grad_norm": 1.3885398932409267,
      "learning_rate": 9.998856149059403e-06,
      "loss": 0.7217,
      "step": 133
    },
    {
      "epoch": 0.007231126220927095,
      "grad_norm": 1.285567221125925,
      "learning_rate": 9.99883888437137e-06,
      "loss": 0.4857,
      "step": 134
    },
    {
      "epoch": 0.007285089849441476,
      "grad_norm": 1.0822792865405872,
      "learning_rate": 9.998821490382161e-06,
      "loss": 0.4783,
      "step": 135
    },
    {
      "epoch": 0.007339053477955858,
      "grad_norm": 1.0501515400470738,
      "learning_rate": 9.99880396709228e-06,
      "loss": 0.4415,
      "step": 136
    },
    {
      "epoch": 0.007393017106470239,
      "grad_norm": 1.1488759595832994,
      "learning_rate": 9.998786314502224e-06,
      "loss": 0.4706,
      "step": 137
    },
    {
      "epoch": 0.007446980734984621,
      "grad_norm": 1.103982892567204,
      "learning_rate": 9.998768532612505e-06,
      "loss": 0.5226,
      "step": 138
    },
    {
      "epoch": 0.007500944363499002,
      "grad_norm": 0.9421778278661262,
      "learning_rate": 9.998750621423631e-06,
      "loss": 0.4667,
      "step": 139
    },
    {
      "epoch": 0.007554907992013383,
      "grad_norm": 1.0143824011867635,
      "learning_rate": 9.998732580936121e-06,
      "loss": 0.5049,
      "step": 140
    },
    {
      "epoch": 0.007608871620527764,
      "grad_norm": 1.1633847473650814,
      "learning_rate": 9.99871441115049e-06,
      "loss": 0.502,
      "step": 141
    },
    {
      "epoch": 0.007662835249042145,
      "grad_norm": 1.3041350676189563,
      "learning_rate": 9.99869611206726e-06,
      "loss": 0.7025,
      "step": 142
    },
    {
      "epoch": 0.007716798877556527,
      "grad_norm": 1.3908993686152158,
      "learning_rate": 9.99867768368696e-06,
      "loss": 0.6131,
      "step": 143
    },
    {
      "epoch": 0.007770762506070908,
      "grad_norm": 1.0207539270593207,
      "learning_rate": 9.998659126010118e-06,
      "loss": 0.5482,
      "step": 144
    },
    {
      "epoch": 0.00782472613458529,
      "grad_norm": 1.3608943167624925,
      "learning_rate": 9.998640439037265e-06,
      "loss": 0.7621,
      "step": 145
    },
    {
      "epoch": 0.007878689763099671,
      "grad_norm": 1.2548745897722715,
      "learning_rate": 9.998621622768942e-06,
      "loss": 0.5751,
      "step": 146
    },
    {
      "epoch": 0.007932653391614052,
      "grad_norm": 1.3718821597112052,
      "learning_rate": 9.998602677205687e-06,
      "loss": 0.6654,
      "step": 147
    },
    {
      "epoch": 0.007986617020128433,
      "grad_norm": 1.4261535885004726,
      "learning_rate": 9.998583602348047e-06,
      "loss": 0.6473,
      "step": 148
    },
    {
      "epoch": 0.008040580648642815,
      "grad_norm": 1.1217430186640438,
      "learning_rate": 9.99856439819657e-06,
      "loss": 0.4908,
      "step": 149
    },
    {
      "epoch": 0.008094544277157197,
      "grad_norm": 1.034071060097991,
      "learning_rate": 9.998545064751804e-06,
      "loss": 0.5102,
      "step": 150
    },
    {
      "epoch": 0.008148507905671577,
      "grad_norm": 1.0843017956342151,
      "learning_rate": 9.998525602014306e-06,
      "loss": 0.5048,
      "step": 151
    },
    {
      "epoch": 0.008202471534185959,
      "grad_norm": 1.4186154439889314,
      "learning_rate": 9.998506009984639e-06,
      "loss": 0.5596,
      "step": 152
    },
    {
      "epoch": 0.00825643516270034,
      "grad_norm": 1.012367711029462,
      "learning_rate": 9.998486288663364e-06,
      "loss": 0.603,
      "step": 153
    },
    {
      "epoch": 0.00831039879121472,
      "grad_norm": 1.09990662038871,
      "learning_rate": 9.998466438051047e-06,
      "loss": 0.5676,
      "step": 154
    },
    {
      "epoch": 0.008364362419729102,
      "grad_norm": 1.6328102766295634,
      "learning_rate": 9.99844645814826e-06,
      "loss": 0.6774,
      "step": 155
    },
    {
      "epoch": 0.008418326048243484,
      "grad_norm": 1.3320551667138403,
      "learning_rate": 9.998426348955575e-06,
      "loss": 0.6241,
      "step": 156
    },
    {
      "epoch": 0.008472289676757866,
      "grad_norm": 1.1788922269729352,
      "learning_rate": 9.998406110473573e-06,
      "loss": 0.6458,
      "step": 157
    },
    {
      "epoch": 0.008526253305272246,
      "grad_norm": 0.888558931597322,
      "learning_rate": 9.998385742702833e-06,
      "loss": 0.4062,
      "step": 158
    },
    {
      "epoch": 0.008580216933786628,
      "grad_norm": 1.0140068397136668,
      "learning_rate": 9.998365245643942e-06,
      "loss": 0.5355,
      "step": 159
    },
    {
      "epoch": 0.00863418056230101,
      "grad_norm": 1.2011625074559469,
      "learning_rate": 9.998344619297488e-06,
      "loss": 0.4706,
      "step": 160
    },
    {
      "epoch": 0.00868814419081539,
      "grad_norm": 1.1389715522250872,
      "learning_rate": 9.998323863664065e-06,
      "loss": 0.5874,
      "step": 161
    },
    {
      "epoch": 0.008742107819329771,
      "grad_norm": 1.2187759337676876,
      "learning_rate": 9.99830297874427e-06,
      "loss": 0.6923,
      "step": 162
    },
    {
      "epoch": 0.008796071447844153,
      "grad_norm": 1.6551973215397675,
      "learning_rate": 9.9982819645387e-06,
      "loss": 0.5792,
      "step": 163
    },
    {
      "epoch": 0.008850035076358535,
      "grad_norm": 0.8574771873796844,
      "learning_rate": 9.998260821047963e-06,
      "loss": 0.372,
      "step": 164
    },
    {
      "epoch": 0.008903998704872915,
      "grad_norm": 1.6701912055897754,
      "learning_rate": 9.998239548272663e-06,
      "loss": 0.6745,
      "step": 165
    },
    {
      "epoch": 0.008957962333387297,
      "grad_norm": 1.2714762488887172,
      "learning_rate": 9.998218146213416e-06,
      "loss": 0.6262,
      "step": 166
    },
    {
      "epoch": 0.009011925961901679,
      "grad_norm": 1.1103967321141592,
      "learning_rate": 9.998196614870831e-06,
      "loss": 0.5723,
      "step": 167
    },
    {
      "epoch": 0.00906588959041606,
      "grad_norm": 1.2833257398994573,
      "learning_rate": 9.998174954245532e-06,
      "loss": 0.583,
      "step": 168
    },
    {
      "epoch": 0.00911985321893044,
      "grad_norm": 1.2738967003285746,
      "learning_rate": 9.998153164338139e-06,
      "loss": 0.5546,
      "step": 169
    },
    {
      "epoch": 0.009173816847444822,
      "grad_norm": 1.2287790353967385,
      "learning_rate": 9.99813124514928e-06,
      "loss": 0.5358,
      "step": 170
    },
    {
      "epoch": 0.009227780475959204,
      "grad_norm": 1.203846588721628,
      "learning_rate": 9.998109196679583e-06,
      "loss": 0.6192,
      "step": 171
    },
    {
      "epoch": 0.009281744104473584,
      "grad_norm": 1.1397105672979877,
      "learning_rate": 9.998087018929685e-06,
      "loss": 0.5762,
      "step": 172
    },
    {
      "epoch": 0.009335707732987966,
      "grad_norm": 1.191304309249943,
      "learning_rate": 9.998064711900218e-06,
      "loss": 0.5048,
      "step": 173
    },
    {
      "epoch": 0.009389671361502348,
      "grad_norm": 1.4096998782102477,
      "learning_rate": 9.998042275591827e-06,
      "loss": 0.6585,
      "step": 174
    },
    {
      "epoch": 0.00944363499001673,
      "grad_norm": 1.3498881860364034,
      "learning_rate": 9.998019710005156e-06,
      "loss": 0.6566,
      "step": 175
    },
    {
      "epoch": 0.00949759861853111,
      "grad_norm": 1.0220711731954275,
      "learning_rate": 9.997997015140854e-06,
      "loss": 0.3949,
      "step": 176
    },
    {
      "epoch": 0.009551562247045491,
      "grad_norm": 1.122898194797035,
      "learning_rate": 9.997974190999574e-06,
      "loss": 0.573,
      "step": 177
    },
    {
      "epoch": 0.009605525875559873,
      "grad_norm": 1.1424439408128342,
      "learning_rate": 9.99795123758197e-06,
      "loss": 0.635,
      "step": 178
    },
    {
      "epoch": 0.009659489504074253,
      "grad_norm": 1.401557558251473,
      "learning_rate": 9.997928154888702e-06,
      "loss": 0.6309,
      "step": 179
    },
    {
      "epoch": 0.009713453132588635,
      "grad_norm": 1.222168741639048,
      "learning_rate": 9.997904942920435e-06,
      "loss": 0.7144,
      "step": 180
    },
    {
      "epoch": 0.009767416761103017,
      "grad_norm": 1.1875438927738955,
      "learning_rate": 9.997881601677836e-06,
      "loss": 0.6016,
      "step": 181
    },
    {
      "epoch": 0.009821380389617398,
      "grad_norm": 1.006927343303014,
      "learning_rate": 9.997858131161573e-06,
      "loss": 0.4256,
      "step": 182
    },
    {
      "epoch": 0.009875344018131779,
      "grad_norm": 1.0639553067807195,
      "learning_rate": 9.997834531372322e-06,
      "loss": 0.5162,
      "step": 183
    },
    {
      "epoch": 0.00992930764664616,
      "grad_norm": 1.537529849129402,
      "learning_rate": 9.997810802310763e-06,
      "loss": 0.9302,
      "step": 184
    },
    {
      "epoch": 0.009983271275160542,
      "grad_norm": 0.9034875159543039,
      "learning_rate": 9.997786943977578e-06,
      "loss": 0.3756,
      "step": 185
    },
    {
      "epoch": 0.010037234903674924,
      "grad_norm": 1.2064675194157035,
      "learning_rate": 9.997762956373451e-06,
      "loss": 0.5162,
      "step": 186
    },
    {
      "epoch": 0.010091198532189304,
      "grad_norm": 1.088112796798825,
      "learning_rate": 9.99773883949907e-06,
      "loss": 0.4671,
      "step": 187
    },
    {
      "epoch": 0.010145162160703686,
      "grad_norm": 1.1473843801769765,
      "learning_rate": 9.997714593355132e-06,
      "loss": 0.6305,
      "step": 188
    },
    {
      "epoch": 0.010199125789218067,
      "grad_norm": 1.2458415600123915,
      "learning_rate": 9.997690217942331e-06,
      "loss": 0.594,
      "step": 189
    },
    {
      "epoch": 0.010253089417732448,
      "grad_norm": 1.0425721869778186,
      "learning_rate": 9.99766571326137e-06,
      "loss": 0.6579,
      "step": 190
    },
    {
      "epoch": 0.01030705304624683,
      "grad_norm": 1.201821517106067,
      "learning_rate": 9.997641079312951e-06,
      "loss": 0.6153,
      "step": 191
    },
    {
      "epoch": 0.010361016674761211,
      "grad_norm": 1.2309229809549613,
      "learning_rate": 9.997616316097782e-06,
      "loss": 0.6137,
      "step": 192
    },
    {
      "epoch": 0.010414980303275593,
      "grad_norm": 1.2235037143833754,
      "learning_rate": 9.997591423616575e-06,
      "loss": 0.6649,
      "step": 193
    },
    {
      "epoch": 0.010468943931789973,
      "grad_norm": 1.2498932392147601,
      "learning_rate": 9.99756640187005e-06,
      "loss": 0.46,
      "step": 194
    },
    {
      "epoch": 0.010522907560304355,
      "grad_norm": 1.4750830159701147,
      "learning_rate": 9.997541250858917e-06,
      "loss": 0.5676,
      "step": 195
    },
    {
      "epoch": 0.010576871188818737,
      "grad_norm": 1.0791524250500986,
      "learning_rate": 9.997515970583907e-06,
      "loss": 0.4956,
      "step": 196
    },
    {
      "epoch": 0.010630834817333118,
      "grad_norm": 1.2643503979391109,
      "learning_rate": 9.997490561045744e-06,
      "loss": 0.5893,
      "step": 197
    },
    {
      "epoch": 0.010684798445847498,
      "grad_norm": 1.1432182544177327,
      "learning_rate": 9.997465022245157e-06,
      "loss": 0.5914,
      "step": 198
    },
    {
      "epoch": 0.01073876207436188,
      "grad_norm": 1.573535653351988,
      "learning_rate": 9.99743935418288e-06,
      "loss": 0.7028,
      "step": 199
    },
    {
      "epoch": 0.010792725702876262,
      "grad_norm": 1.3389910777902252,
      "learning_rate": 9.997413556859654e-06,
      "loss": 0.6192,
      "step": 200
    },
    {
      "epoch": 0.010846689331390642,
      "grad_norm": 0.8941336546169296,
      "learning_rate": 9.997387630276214e-06,
      "loss": 0.3799,
      "step": 201
    },
    {
      "epoch": 0.010900652959905024,
      "grad_norm": 1.1648849081985506,
      "learning_rate": 9.997361574433313e-06,
      "loss": 0.5682,
      "step": 202
    },
    {
      "epoch": 0.010954616588419406,
      "grad_norm": 1.053113316445482,
      "learning_rate": 9.997335389331693e-06,
      "loss": 0.6232,
      "step": 203
    },
    {
      "epoch": 0.011008580216933787,
      "grad_norm": 1.0761303152202273,
      "learning_rate": 9.997309074972112e-06,
      "loss": 0.4745,
      "step": 204
    },
    {
      "epoch": 0.011062543845448167,
      "grad_norm": 1.1473439558487417,
      "learning_rate": 9.997282631355325e-06,
      "loss": 0.4957,
      "step": 205
    },
    {
      "epoch": 0.01111650747396255,
      "grad_norm": 1.2223679691873086,
      "learning_rate": 9.99725605848209e-06,
      "loss": 0.5827,
      "step": 206
    },
    {
      "epoch": 0.011170471102476931,
      "grad_norm": 1.0981509552274973,
      "learning_rate": 9.99722935635317e-06,
      "loss": 0.5869,
      "step": 207
    },
    {
      "epoch": 0.011224434730991311,
      "grad_norm": 1.4704546867499864,
      "learning_rate": 9.997202524969337e-06,
      "loss": 0.6123,
      "step": 208
    },
    {
      "epoch": 0.011278398359505693,
      "grad_norm": 1.0645843133318582,
      "learning_rate": 9.997175564331359e-06,
      "loss": 0.4807,
      "step": 209
    },
    {
      "epoch": 0.011332361988020075,
      "grad_norm": 1.1274735657508994,
      "learning_rate": 9.997148474440009e-06,
      "loss": 0.5622,
      "step": 210
    },
    {
      "epoch": 0.011386325616534456,
      "grad_norm": 1.2358385027704084,
      "learning_rate": 9.997121255296068e-06,
      "loss": 0.6144,
      "step": 211
    },
    {
      "epoch": 0.011440289245048836,
      "grad_norm": 1.319621430143422,
      "learning_rate": 9.99709390690032e-06,
      "loss": 0.7473,
      "step": 212
    },
    {
      "epoch": 0.011494252873563218,
      "grad_norm": 1.2811889357342763,
      "learning_rate": 9.997066429253546e-06,
      "loss": 0.4787,
      "step": 213
    },
    {
      "epoch": 0.0115482165020776,
      "grad_norm": 1.171717692968343,
      "learning_rate": 9.997038822356543e-06,
      "loss": 0.4688,
      "step": 214
    },
    {
      "epoch": 0.011602180130591982,
      "grad_norm": 0.9580409637170038,
      "learning_rate": 9.997011086210096e-06,
      "loss": 0.3909,
      "step": 215
    },
    {
      "epoch": 0.011656143759106362,
      "grad_norm": 1.3138056995523446,
      "learning_rate": 9.996983220815009e-06,
      "loss": 0.6829,
      "step": 216
    },
    {
      "epoch": 0.011710107387620744,
      "grad_norm": 1.458926745441639,
      "learning_rate": 9.99695522617208e-06,
      "loss": 0.8704,
      "step": 217
    },
    {
      "epoch": 0.011764071016135125,
      "grad_norm": 1.3303970112972667,
      "learning_rate": 9.996927102282115e-06,
      "loss": 0.6338,
      "step": 218
    },
    {
      "epoch": 0.011818034644649505,
      "grad_norm": 1.2138386431991741,
      "learning_rate": 9.99689884914592e-06,
      "loss": 0.6906,
      "step": 219
    },
    {
      "epoch": 0.011871998273163887,
      "grad_norm": 1.1858395508401343,
      "learning_rate": 9.99687046676431e-06,
      "loss": 0.5059,
      "step": 220
    },
    {
      "epoch": 0.011925961901678269,
      "grad_norm": 1.1115608879719827,
      "learning_rate": 9.996841955138098e-06,
      "loss": 0.5521,
      "step": 221
    },
    {
      "epoch": 0.01197992553019265,
      "grad_norm": 1.0733546674988221,
      "learning_rate": 9.996813314268103e-06,
      "loss": 0.4899,
      "step": 222
    },
    {
      "epoch": 0.012033889158707031,
      "grad_norm": 1.4498359325116648,
      "learning_rate": 9.99678454415515e-06,
      "loss": 1.0193,
      "step": 223
    },
    {
      "epoch": 0.012087852787221413,
      "grad_norm": 1.2563070702223789,
      "learning_rate": 9.996755644800068e-06,
      "loss": 0.6354,
      "step": 224
    },
    {
      "epoch": 0.012141816415735794,
      "grad_norm": 1.0268052194602335,
      "learning_rate": 9.996726616203682e-06,
      "loss": 0.4692,
      "step": 225
    },
    {
      "epoch": 0.012195780044250175,
      "grad_norm": 1.8204979456323878,
      "learning_rate": 9.996697458366832e-06,
      "loss": 0.5616,
      "step": 226
    },
    {
      "epoch": 0.012249743672764556,
      "grad_norm": 1.2717677288653928,
      "learning_rate": 9.996668171290353e-06,
      "loss": 0.6308,
      "step": 227
    },
    {
      "epoch": 0.012303707301278938,
      "grad_norm": 1.1932236773195832,
      "learning_rate": 9.996638754975085e-06,
      "loss": 0.5354,
      "step": 228
    },
    {
      "epoch": 0.01235767092979332,
      "grad_norm": 0.9449431584018725,
      "learning_rate": 9.996609209421877e-06,
      "loss": 0.3974,
      "step": 229
    },
    {
      "epoch": 0.0124116345583077,
      "grad_norm": 1.161763024048585,
      "learning_rate": 9.996579534631578e-06,
      "loss": 0.4988,
      "step": 230
    },
    {
      "epoch": 0.012465598186822082,
      "grad_norm": 0.8690161676114153,
      "learning_rate": 9.996549730605039e-06,
      "loss": 0.3913,
      "step": 231
    },
    {
      "epoch": 0.012519561815336464,
      "grad_norm": 1.247403134820183,
      "learning_rate": 9.996519797343117e-06,
      "loss": 0.5993,
      "step": 232
    },
    {
      "epoch": 0.012573525443850845,
      "grad_norm": 1.11321460397835,
      "learning_rate": 9.996489734846672e-06,
      "loss": 0.5337,
      "step": 233
    },
    {
      "epoch": 0.012627489072365225,
      "grad_norm": 1.0978134832882454,
      "learning_rate": 9.99645954311657e-06,
      "loss": 0.6005,
      "step": 234
    },
    {
      "epoch": 0.012681452700879607,
      "grad_norm": 1.0541623598870151,
      "learning_rate": 9.996429222153676e-06,
      "loss": 0.4935,
      "step": 235
    },
    {
      "epoch": 0.012735416329393989,
      "grad_norm": 1.1543669864396497,
      "learning_rate": 9.996398771958865e-06,
      "loss": 0.5047,
      "step": 236
    },
    {
      "epoch": 0.012789379957908369,
      "grad_norm": 1.0640833183002085,
      "learning_rate": 9.996368192533009e-06,
      "loss": 0.4428,
      "step": 237
    },
    {
      "epoch": 0.01284334358642275,
      "grad_norm": 1.3678945497868689,
      "learning_rate": 9.996337483876987e-06,
      "loss": 0.7162,
      "step": 238
    },
    {
      "epoch": 0.012897307214937133,
      "grad_norm": 1.0984548994354322,
      "learning_rate": 9.996306645991682e-06,
      "loss": 0.5046,
      "step": 239
    },
    {
      "epoch": 0.012951270843451514,
      "grad_norm": 1.3848200621596305,
      "learning_rate": 9.996275678877982e-06,
      "loss": 0.7497,
      "step": 240
    },
    {
      "epoch": 0.013005234471965894,
      "grad_norm": 1.0761524958808466,
      "learning_rate": 9.996244582536775e-06,
      "loss": 0.5328,
      "step": 241
    },
    {
      "epoch": 0.013059198100480276,
      "grad_norm": 0.9775089979906237,
      "learning_rate": 9.996213356968957e-06,
      "loss": 0.5706,
      "step": 242
    },
    {
      "epoch": 0.013113161728994658,
      "grad_norm": 1.5040696987581101,
      "learning_rate": 9.996182002175423e-06,
      "loss": 0.6973,
      "step": 243
    },
    {
      "epoch": 0.01316712535750904,
      "grad_norm": 0.9630029498630243,
      "learning_rate": 9.996150518157077e-06,
      "loss": 0.3413,
      "step": 244
    },
    {
      "epoch": 0.01322108898602342,
      "grad_norm": 1.2080131338079794,
      "learning_rate": 9.996118904914818e-06,
      "loss": 0.6624,
      "step": 245
    },
    {
      "epoch": 0.013275052614537802,
      "grad_norm": 1.0703602760499134,
      "learning_rate": 9.996087162449562e-06,
      "loss": 0.5848,
      "step": 246
    },
    {
      "epoch": 0.013329016243052183,
      "grad_norm": 1.2659525460953978,
      "learning_rate": 9.996055290762217e-06,
      "loss": 0.5972,
      "step": 247
    },
    {
      "epoch": 0.013382979871566563,
      "grad_norm": 1.3476130462115583,
      "learning_rate": 9.9960232898537e-06,
      "loss": 0.5393,
      "step": 248
    },
    {
      "epoch": 0.013436943500080945,
      "grad_norm": 1.2763433889992966,
      "learning_rate": 9.995991159724933e-06,
      "loss": 0.6175,
      "step": 249
    },
    {
      "epoch": 0.013490907128595327,
      "grad_norm": 1.3606932218582706,
      "learning_rate": 9.995958900376835e-06,
      "loss": 0.7005,
      "step": 250
    },
    {
      "epoch": 0.013544870757109709,
      "grad_norm": 1.4461851927365759,
      "learning_rate": 9.995926511810336e-06,
      "loss": 0.558,
      "step": 251
    },
    {
      "epoch": 0.013598834385624089,
      "grad_norm": 1.192890069735771,
      "learning_rate": 9.995893994026365e-06,
      "loss": 0.6835,
      "step": 252
    },
    {
      "epoch": 0.01365279801413847,
      "grad_norm": 0.9060194145350097,
      "learning_rate": 9.995861347025858e-06,
      "loss": 0.4069,
      "step": 253
    },
    {
      "epoch": 0.013706761642652852,
      "grad_norm": 1.4397082880115948,
      "learning_rate": 9.995828570809753e-06,
      "loss": 0.6054,
      "step": 254
    },
    {
      "epoch": 0.013760725271167232,
      "grad_norm": 1.1939353182809131,
      "learning_rate": 9.995795665378993e-06,
      "loss": 0.621,
      "step": 255
    },
    {
      "epoch": 0.013814688899681614,
      "grad_norm": 1.273747209553719,
      "learning_rate": 9.995762630734524e-06,
      "loss": 0.67,
      "step": 256
    },
    {
      "epoch": 0.013868652528195996,
      "grad_norm": 1.2463987479425098,
      "learning_rate": 9.995729466877294e-06,
      "loss": 0.6794,
      "step": 257
    },
    {
      "epoch": 0.013922616156710378,
      "grad_norm": 1.3382261672022895,
      "learning_rate": 9.995696173808254e-06,
      "loss": 0.5532,
      "step": 258
    },
    {
      "epoch": 0.013976579785224758,
      "grad_norm": 1.1127761869896213,
      "learning_rate": 9.995662751528366e-06,
      "loss": 0.4956,
      "step": 259
    },
    {
      "epoch": 0.01403054341373914,
      "grad_norm": 1.3641595580702635,
      "learning_rate": 9.995629200038586e-06,
      "loss": 0.6197,
      "step": 260
    },
    {
      "epoch": 0.014084507042253521,
      "grad_norm": 1.123741530834796,
      "learning_rate": 9.995595519339882e-06,
      "loss": 0.6066,
      "step": 261
    },
    {
      "epoch": 0.014138470670767903,
      "grad_norm": 1.2068617191567075,
      "learning_rate": 9.995561709433219e-06,
      "loss": 0.5622,
      "step": 262
    },
    {
      "epoch": 0.014192434299282283,
      "grad_norm": 1.346334566713258,
      "learning_rate": 9.99552777031957e-06,
      "loss": 0.6559,
      "step": 263
    },
    {
      "epoch": 0.014246397927796665,
      "grad_norm": 1.0053081925345104,
      "learning_rate": 9.995493701999912e-06,
      "loss": 0.516,
      "step": 264
    },
    {
      "epoch": 0.014300361556311047,
      "grad_norm": 1.2112570428487228,
      "learning_rate": 9.995459504475223e-06,
      "loss": 0.4491,
      "step": 265
    },
    {
      "epoch": 0.014354325184825427,
      "grad_norm": 1.2986392377260056,
      "learning_rate": 9.995425177746484e-06,
      "loss": 0.6181,
      "step": 266
    },
    {
      "epoch": 0.014408288813339809,
      "grad_norm": 1.0653526638799027,
      "learning_rate": 9.995390721814683e-06,
      "loss": 0.5026,
      "step": 267
    },
    {
      "epoch": 0.01446225244185419,
      "grad_norm": 1.259812320370971,
      "learning_rate": 9.995356136680811e-06,
      "loss": 0.4447,
      "step": 268
    },
    {
      "epoch": 0.014516216070368572,
      "grad_norm": 0.9736190445696172,
      "learning_rate": 9.995321422345862e-06,
      "loss": 0.3902,
      "step": 269
    },
    {
      "epoch": 0.014570179698882952,
      "grad_norm": 1.2958778367226536,
      "learning_rate": 9.995286578810833e-06,
      "loss": 0.5193,
      "step": 270
    },
    {
      "epoch": 0.014624143327397334,
      "grad_norm": 1.164966261479915,
      "learning_rate": 9.995251606076724e-06,
      "loss": 0.5022,
      "step": 271
    },
    {
      "epoch": 0.014678106955911716,
      "grad_norm": 0.9666983690245742,
      "learning_rate": 9.995216504144543e-06,
      "loss": 0.5106,
      "step": 272
    },
    {
      "epoch": 0.014732070584426096,
      "grad_norm": 1.1069920244729308,
      "learning_rate": 9.995181273015297e-06,
      "loss": 0.5576,
      "step": 273
    },
    {
      "epoch": 0.014786034212940478,
      "grad_norm": 1.1128100553232345,
      "learning_rate": 9.995145912689998e-06,
      "loss": 0.5319,
      "step": 274
    },
    {
      "epoch": 0.01483999784145486,
      "grad_norm": 1.2769181693810012,
      "learning_rate": 9.995110423169665e-06,
      "loss": 0.6883,
      "step": 275
    },
    {
      "epoch": 0.014893961469969241,
      "grad_norm": 1.255845242513406,
      "learning_rate": 9.995074804455317e-06,
      "loss": 0.5589,
      "step": 276
    },
    {
      "epoch": 0.014947925098483621,
      "grad_norm": 1.4030259725520333,
      "learning_rate": 9.995039056547977e-06,
      "loss": 0.7846,
      "step": 277
    },
    {
      "epoch": 0.015001888726998003,
      "grad_norm": 1.09429455887684,
      "learning_rate": 9.995003179448673e-06,
      "loss": 0.5616,
      "step": 278
    },
    {
      "epoch": 0.015055852355512385,
      "grad_norm": 0.9757421804231383,
      "learning_rate": 9.994967173158434e-06,
      "loss": 0.5016,
      "step": 279
    },
    {
      "epoch": 0.015109815984026767,
      "grad_norm": 0.8847690502095292,
      "learning_rate": 9.994931037678297e-06,
      "loss": 0.3401,
      "step": 280
    },
    {
      "epoch": 0.015163779612541147,
      "grad_norm": 1.1057166785274695,
      "learning_rate": 9.994894773009301e-06,
      "loss": 0.5028,
      "step": 281
    },
    {
      "epoch": 0.015217743241055529,
      "grad_norm": 1.101812150891309,
      "learning_rate": 9.994858379152486e-06,
      "loss": 0.5727,
      "step": 282
    },
    {
      "epoch": 0.01527170686956991,
      "grad_norm": 1.2747397536257468,
      "learning_rate": 9.994821856108902e-06,
      "loss": 0.6635,
      "step": 283
    },
    {
      "epoch": 0.01532567049808429,
      "grad_norm": 1.1942069829536837,
      "learning_rate": 9.994785203879594e-06,
      "loss": 0.6384,
      "step": 284
    },
    {
      "epoch": 0.015379634126598672,
      "grad_norm": 1.1015489903141056,
      "learning_rate": 9.994748422465618e-06,
      "loss": 0.7011,
      "step": 285
    },
    {
      "epoch": 0.015433597755113054,
      "grad_norm": 1.0161278765443331,
      "learning_rate": 9.994711511868032e-06,
      "loss": 0.4708,
      "step": 286
    },
    {
      "epoch": 0.015487561383627436,
      "grad_norm": 1.121874126321933,
      "learning_rate": 9.994674472087896e-06,
      "loss": 0.4531,
      "step": 287
    },
    {
      "epoch": 0.015541525012141816,
      "grad_norm": 1.0492910723138273,
      "learning_rate": 9.994637303126272e-06,
      "loss": 0.398,
      "step": 288
    },
    {
      "epoch": 0.015595488640656198,
      "grad_norm": 1.1045915167009837,
      "learning_rate": 9.99460000498423e-06,
      "loss": 0.5987,
      "step": 289
    },
    {
      "epoch": 0.01564945226917058,
      "grad_norm": 1.2438915845902245,
      "learning_rate": 9.994562577662845e-06,
      "loss": 0.6903,
      "step": 290
    },
    {
      "epoch": 0.01570341589768496,
      "grad_norm": 0.8592788112987058,
      "learning_rate": 9.994525021163188e-06,
      "loss": 0.3996,
      "step": 291
    },
    {
      "epoch": 0.015757379526199343,
      "grad_norm": 0.9456141115471028,
      "learning_rate": 9.994487335486342e-06,
      "loss": 0.5333,
      "step": 292
    },
    {
      "epoch": 0.015811343154713723,
      "grad_norm": 1.4482688719577594,
      "learning_rate": 9.994449520633386e-06,
      "loss": 0.5056,
      "step": 293
    },
    {
      "epoch": 0.015865306783228103,
      "grad_norm": 1.0433010803417053,
      "learning_rate": 9.994411576605413e-06,
      "loss": 0.5134,
      "step": 294
    },
    {
      "epoch": 0.015919270411742487,
      "grad_norm": 1.0792885939124144,
      "learning_rate": 9.994373503403507e-06,
      "loss": 0.5134,
      "step": 295
    },
    {
      "epoch": 0.015973234040256867,
      "grad_norm": 1.1533137276573562,
      "learning_rate": 9.994335301028768e-06,
      "loss": 0.7336,
      "step": 296
    },
    {
      "epoch": 0.016027197668771247,
      "grad_norm": 1.2103948179761086,
      "learning_rate": 9.99429696948229e-06,
      "loss": 0.6977,
      "step": 297
    },
    {
      "epoch": 0.01608116129728563,
      "grad_norm": 1.2714543342806681,
      "learning_rate": 9.994258508765175e-06,
      "loss": 0.5472,
      "step": 298
    },
    {
      "epoch": 0.01613512492580001,
      "grad_norm": 1.64434725699464,
      "learning_rate": 9.994219918878531e-06,
      "loss": 0.772,
      "step": 299
    },
    {
      "epoch": 0.016189088554314394,
      "grad_norm": 1.1441854407927474,
      "learning_rate": 9.994181199823463e-06,
      "loss": 0.5516,
      "step": 300
    },
    {
      "epoch": 0.016243052182828774,
      "grad_norm": 1.2754927387713395,
      "learning_rate": 9.994142351601089e-06,
      "loss": 0.5164,
      "step": 301
    },
    {
      "epoch": 0.016297015811343154,
      "grad_norm": 1.0388992215157398,
      "learning_rate": 9.994103374212522e-06,
      "loss": 0.3441,
      "step": 302
    },
    {
      "epoch": 0.016350979439857537,
      "grad_norm": 0.9307976971500482,
      "learning_rate": 9.994064267658882e-06,
      "loss": 0.4018,
      "step": 303
    },
    {
      "epoch": 0.016404943068371917,
      "grad_norm": 1.434324345445989,
      "learning_rate": 9.994025031941295e-06,
      "loss": 0.5894,
      "step": 304
    },
    {
      "epoch": 0.016458906696886298,
      "grad_norm": 1.3557578833374386,
      "learning_rate": 9.993985667060887e-06,
      "loss": 0.6722,
      "step": 305
    },
    {
      "epoch": 0.01651287032540068,
      "grad_norm": 1.4210356035802283,
      "learning_rate": 9.993946173018792e-06,
      "loss": 0.7326,
      "step": 306
    },
    {
      "epoch": 0.01656683395391506,
      "grad_norm": 1.3176899298059377,
      "learning_rate": 9.993906549816141e-06,
      "loss": 0.7488,
      "step": 307
    },
    {
      "epoch": 0.01662079758242944,
      "grad_norm": 1.1605402034219734,
      "learning_rate": 9.993866797454076e-06,
      "loss": 0.5744,
      "step": 308
    },
    {
      "epoch": 0.016674761210943825,
      "grad_norm": 1.2348787683587534,
      "learning_rate": 9.993826915933738e-06,
      "loss": 0.5132,
      "step": 309
    },
    {
      "epoch": 0.016728724839458205,
      "grad_norm": 1.3295254195522013,
      "learning_rate": 9.993786905256273e-06,
      "loss": 0.7401,
      "step": 310
    },
    {
      "epoch": 0.016782688467972585,
      "grad_norm": 1.1704290073676298,
      "learning_rate": 9.993746765422834e-06,
      "loss": 0.5894,
      "step": 311
    },
    {
      "epoch": 0.01683665209648697,
      "grad_norm": 1.3489950408119658,
      "learning_rate": 9.99370649643457e-06,
      "loss": 0.5088,
      "step": 312
    },
    {
      "epoch": 0.01689061572500135,
      "grad_norm": 1.1663551477633096,
      "learning_rate": 9.993666098292643e-06,
      "loss": 0.5308,
      "step": 313
    },
    {
      "epoch": 0.016944579353515732,
      "grad_norm": 1.2444978826388653,
      "learning_rate": 9.99362557099821e-06,
      "loss": 0.5979,
      "step": 314
    },
    {
      "epoch": 0.016998542982030112,
      "grad_norm": 1.1669848128935618,
      "learning_rate": 9.993584914552437e-06,
      "loss": 0.5185,
      "step": 315
    },
    {
      "epoch": 0.017052506610544492,
      "grad_norm": 1.1785288436941468,
      "learning_rate": 9.993544128956494e-06,
      "loss": 0.6184,
      "step": 316
    },
    {
      "epoch": 0.017106470239058875,
      "grad_norm": 1.1287095546326185,
      "learning_rate": 9.993503214211556e-06,
      "loss": 0.6079,
      "step": 317
    },
    {
      "epoch": 0.017160433867573256,
      "grad_norm": 1.3208306869926496,
      "learning_rate": 9.993462170318791e-06,
      "loss": 0.6181,
      "step": 318
    },
    {
      "epoch": 0.017214397496087636,
      "grad_norm": 1.2687226372856646,
      "learning_rate": 9.993420997279383e-06,
      "loss": 0.607,
      "step": 319
    },
    {
      "epoch": 0.01726836112460202,
      "grad_norm": 1.0532544589435897,
      "learning_rate": 9.993379695094516e-06,
      "loss": 0.5164,
      "step": 320
    },
    {
      "epoch": 0.0173223247531164,
      "grad_norm": 1.146514029063011,
      "learning_rate": 9.993338263765377e-06,
      "loss": 0.5759,
      "step": 321
    },
    {
      "epoch": 0.01737628838163078,
      "grad_norm": 1.2409108189053872,
      "learning_rate": 9.993296703293156e-06,
      "loss": 0.7701,
      "step": 322
    },
    {
      "epoch": 0.017430252010145163,
      "grad_norm": 1.239396038545281,
      "learning_rate": 9.993255013679047e-06,
      "loss": 0.6736,
      "step": 323
    },
    {
      "epoch": 0.017484215638659543,
      "grad_norm": 1.0987468594332794,
      "learning_rate": 9.99321319492425e-06,
      "loss": 0.5417,
      "step": 324
    },
    {
      "epoch": 0.017538179267173926,
      "grad_norm": 1.2087645651398047,
      "learning_rate": 9.993171247029963e-06,
      "loss": 0.5748,
      "step": 325
    },
    {
      "epoch": 0.017592142895688306,
      "grad_norm": 1.0571439326868024,
      "learning_rate": 9.993129169997397e-06,
      "loss": 0.4707,
      "step": 326
    },
    {
      "epoch": 0.017646106524202686,
      "grad_norm": 1.2947374398426923,
      "learning_rate": 9.993086963827758e-06,
      "loss": 0.5796,
      "step": 327
    },
    {
      "epoch": 0.01770007015271707,
      "grad_norm": 1.157335974063005,
      "learning_rate": 9.993044628522258e-06,
      "loss": 0.5345,
      "step": 328
    },
    {
      "epoch": 0.01775403378123145,
      "grad_norm": 0.8450354570160381,
      "learning_rate": 9.993002164082118e-06,
      "loss": 0.4465,
      "step": 329
    },
    {
      "epoch": 0.01780799740974583,
      "grad_norm": 1.0798109763088657,
      "learning_rate": 9.992959570508554e-06,
      "loss": 0.5593,
      "step": 330
    },
    {
      "epoch": 0.017861961038260214,
      "grad_norm": 1.4764884866302763,
      "learning_rate": 9.992916847802793e-06,
      "loss": 0.7197,
      "step": 331
    },
    {
      "epoch": 0.017915924666774594,
      "grad_norm": 1.029910839549723,
      "learning_rate": 9.992873995966063e-06,
      "loss": 0.5725,
      "step": 332
    },
    {
      "epoch": 0.017969888295288974,
      "grad_norm": 1.1273273151709464,
      "learning_rate": 9.992831014999595e-06,
      "loss": 0.4723,
      "step": 333
    },
    {
      "epoch": 0.018023851923803357,
      "grad_norm": 1.1154601701917475,
      "learning_rate": 9.992787904904621e-06,
      "loss": 0.4921,
      "step": 334
    },
    {
      "epoch": 0.018077815552317737,
      "grad_norm": 1.155569195738189,
      "learning_rate": 9.992744665682386e-06,
      "loss": 0.6977,
      "step": 335
    },
    {
      "epoch": 0.01813177918083212,
      "grad_norm": 1.1784027045666452,
      "learning_rate": 9.992701297334127e-06,
      "loss": 0.4818,
      "step": 336
    },
    {
      "epoch": 0.0181857428093465,
      "grad_norm": 1.155068223585959,
      "learning_rate": 9.992657799861095e-06,
      "loss": 0.6108,
      "step": 337
    },
    {
      "epoch": 0.01823970643786088,
      "grad_norm": 1.1407259509958554,
      "learning_rate": 9.992614173264538e-06,
      "loss": 0.5003,
      "step": 338
    },
    {
      "epoch": 0.018293670066375264,
      "grad_norm": 1.0176430995940269,
      "learning_rate": 9.99257041754571e-06,
      "loss": 0.478,
      "step": 339
    },
    {
      "epoch": 0.018347633694889644,
      "grad_norm": 1.1118635127507424,
      "learning_rate": 9.992526532705869e-06,
      "loss": 0.5654,
      "step": 340
    },
    {
      "epoch": 0.018401597323404024,
      "grad_norm": 1.2135320201288273,
      "learning_rate": 9.992482518746274e-06,
      "loss": 0.5657,
      "step": 341
    },
    {
      "epoch": 0.018455560951918408,
      "grad_norm": 1.085537399883959,
      "learning_rate": 9.992438375668196e-06,
      "loss": 0.5541,
      "step": 342
    },
    {
      "epoch": 0.018509524580432788,
      "grad_norm": 1.0052941103448916,
      "learning_rate": 9.992394103472899e-06,
      "loss": 0.5616,
      "step": 343
    },
    {
      "epoch": 0.018563488208947168,
      "grad_norm": 1.2665821633476784,
      "learning_rate": 9.992349702161653e-06,
      "loss": 0.5954,
      "step": 344
    },
    {
      "epoch": 0.01861745183746155,
      "grad_norm": 1.0918213076130114,
      "learning_rate": 9.99230517173574e-06,
      "loss": 0.4426,
      "step": 345
    },
    {
      "epoch": 0.01867141546597593,
      "grad_norm": 1.073841582140135,
      "learning_rate": 9.992260512196437e-06,
      "loss": 0.5025,
      "step": 346
    },
    {
      "epoch": 0.018725379094490315,
      "grad_norm": 1.0383813865163982,
      "learning_rate": 9.992215723545029e-06,
      "loss": 0.5631,
      "step": 347
    },
    {
      "epoch": 0.018779342723004695,
      "grad_norm": 1.3164985062021122,
      "learning_rate": 9.992170805782799e-06,
      "loss": 0.6187,
      "step": 348
    },
    {
      "epoch": 0.018833306351519075,
      "grad_norm": 1.1021367368461157,
      "learning_rate": 9.992125758911043e-06,
      "loss": 0.5302,
      "step": 349
    },
    {
      "epoch": 0.01888726998003346,
      "grad_norm": 0.8934010962199753,
      "learning_rate": 9.992080582931052e-06,
      "loss": 0.3645,
      "step": 350
    },
    {
      "epoch": 0.01894123360854784,
      "grad_norm": 1.1996360550617136,
      "learning_rate": 9.99203527784413e-06,
      "loss": 0.4724,
      "step": 351
    },
    {
      "epoch": 0.01899519723706222,
      "grad_norm": 1.1280992512564252,
      "learning_rate": 9.99198984365157e-06,
      "loss": 0.5557,
      "step": 352
    },
    {
      "epoch": 0.019049160865576602,
      "grad_norm": 1.2136948086761201,
      "learning_rate": 9.991944280354685e-06,
      "loss": 0.6769,
      "step": 353
    },
    {
      "epoch": 0.019103124494090983,
      "grad_norm": 1.1039408042652477,
      "learning_rate": 9.991898587954784e-06,
      "loss": 0.442,
      "step": 354
    },
    {
      "epoch": 0.019157088122605363,
      "grad_norm": 1.356495134931273,
      "learning_rate": 9.991852766453179e-06,
      "loss": 0.6489,
      "step": 355
    },
    {
      "epoch": 0.019211051751119746,
      "grad_norm": 1.0219343512193042,
      "learning_rate": 9.991806815851186e-06,
      "loss": 0.4407,
      "step": 356
    },
    {
      "epoch": 0.019265015379634126,
      "grad_norm": 1.1804707429959662,
      "learning_rate": 9.991760736150126e-06,
      "loss": 0.5527,
      "step": 357
    },
    {
      "epoch": 0.019318979008148506,
      "grad_norm": 1.1591357203925168,
      "learning_rate": 9.991714527351323e-06,
      "loss": 0.5369,
      "step": 358
    },
    {
      "epoch": 0.01937294263666289,
      "grad_norm": 0.9582203943450978,
      "learning_rate": 9.99166818945611e-06,
      "loss": 0.414,
      "step": 359
    },
    {
      "epoch": 0.01942690626517727,
      "grad_norm": 1.0443032853878906,
      "learning_rate": 9.991621722465812e-06,
      "loss": 0.4966,
      "step": 360
    },
    {
      "epoch": 0.019480869893691653,
      "grad_norm": 1.2341283036013393,
      "learning_rate": 9.991575126381767e-06,
      "loss": 0.488,
      "step": 361
    },
    {
      "epoch": 0.019534833522206033,
      "grad_norm": 1.1218387784437682,
      "learning_rate": 9.991528401205315e-06,
      "loss": 0.5758,
      "step": 362
    },
    {
      "epoch": 0.019588797150720413,
      "grad_norm": 0.9356431766914127,
      "learning_rate": 9.991481546937797e-06,
      "loss": 0.3837,
      "step": 363
    },
    {
      "epoch": 0.019642760779234797,
      "grad_norm": 1.1757396086902396,
      "learning_rate": 9.991434563580561e-06,
      "loss": 0.5802,
      "step": 364
    },
    {
      "epoch": 0.019696724407749177,
      "grad_norm": 1.2136740074861603,
      "learning_rate": 9.99138745113496e-06,
      "loss": 0.545,
      "step": 365
    },
    {
      "epoch": 0.019750688036263557,
      "grad_norm": 1.5243756054514035,
      "learning_rate": 9.991340209602343e-06,
      "loss": 0.6434,
      "step": 366
    },
    {
      "epoch": 0.01980465166477794,
      "grad_norm": 0.9236937535852299,
      "learning_rate": 9.991292838984071e-06,
      "loss": 0.3782,
      "step": 367
    },
    {
      "epoch": 0.01985861529329232,
      "grad_norm": 1.131405185565812,
      "learning_rate": 9.991245339281507e-06,
      "loss": 0.6485,
      "step": 368
    },
    {
      "epoch": 0.0199125789218067,
      "grad_norm": 1.039846342455159,
      "learning_rate": 9.991197710496012e-06,
      "loss": 0.5499,
      "step": 369
    },
    {
      "epoch": 0.019966542550321084,
      "grad_norm": 1.0289320772557986,
      "learning_rate": 9.991149952628954e-06,
      "loss": 0.4486,
      "step": 370
    },
    {
      "epoch": 0.020020506178835464,
      "grad_norm": 0.893959457797538,
      "learning_rate": 9.991102065681713e-06,
      "loss": 0.4515,
      "step": 371
    },
    {
      "epoch": 0.020074469807349848,
      "grad_norm": 1.0074706151635824,
      "learning_rate": 9.991054049655657e-06,
      "loss": 0.5068,
      "step": 372
    },
    {
      "epoch": 0.020128433435864228,
      "grad_norm": 1.2604886614806916,
      "learning_rate": 9.991005904552171e-06,
      "loss": 0.5376,
      "step": 373
    },
    {
      "epoch": 0.020182397064378608,
      "grad_norm": 1.288867864182396,
      "learning_rate": 9.99095763037264e-06,
      "loss": 0.6698,
      "step": 374
    },
    {
      "epoch": 0.02023636069289299,
      "grad_norm": 1.0604201301316702,
      "learning_rate": 9.990909227118445e-06,
      "loss": 0.6035,
      "step": 375
    },
    {
      "epoch": 0.02029032432140737,
      "grad_norm": 1.4511241524576817,
      "learning_rate": 9.990860694790984e-06,
      "loss": 0.515,
      "step": 376
    },
    {
      "epoch": 0.02034428794992175,
      "grad_norm": 0.9691096093259752,
      "learning_rate": 9.990812033391647e-06,
      "loss": 0.5259,
      "step": 377
    },
    {
      "epoch": 0.020398251578436135,
      "grad_norm": 1.1933954800722981,
      "learning_rate": 9.990763242921836e-06,
      "loss": 0.4883,
      "step": 378
    },
    {
      "epoch": 0.020452215206950515,
      "grad_norm": 0.9890971821504816,
      "learning_rate": 9.99071432338295e-06,
      "loss": 0.5103,
      "step": 379
    },
    {
      "epoch": 0.020506178835464895,
      "grad_norm": 1.6161190384179327,
      "learning_rate": 9.990665274776397e-06,
      "loss": 0.7775,
      "step": 380
    },
    {
      "epoch": 0.02056014246397928,
      "grad_norm": 1.1588996284220572,
      "learning_rate": 9.990616097103589e-06,
      "loss": 0.6452,
      "step": 381
    },
    {
      "epoch": 0.02061410609249366,
      "grad_norm": 1.6387354365251323,
      "learning_rate": 9.990566790365934e-06,
      "loss": 0.538,
      "step": 382
    },
    {
      "epoch": 0.020668069721008042,
      "grad_norm": 1.181503408307723,
      "learning_rate": 9.990517354564854e-06,
      "loss": 0.6847,
      "step": 383
    },
    {
      "epoch": 0.020722033349522422,
      "grad_norm": 1.1210325924067448,
      "learning_rate": 9.990467789701768e-06,
      "loss": 0.4307,
      "step": 384
    },
    {
      "epoch": 0.020775996978036802,
      "grad_norm": 1.2351073551458729,
      "learning_rate": 9.990418095778099e-06,
      "loss": 0.5608,
      "step": 385
    },
    {
      "epoch": 0.020829960606551186,
      "grad_norm": 0.980133901174093,
      "learning_rate": 9.990368272795278e-06,
      "loss": 0.4729,
      "step": 386
    },
    {
      "epoch": 0.020883924235065566,
      "grad_norm": 0.9846342161427473,
      "learning_rate": 9.990318320754735e-06,
      "loss": 0.4843,
      "step": 387
    },
    {
      "epoch": 0.020937887863579946,
      "grad_norm": 1.0602643921251473,
      "learning_rate": 9.990268239657905e-06,
      "loss": 0.6411,
      "step": 388
    },
    {
      "epoch": 0.02099185149209433,
      "grad_norm": 1.152417197857347,
      "learning_rate": 9.99021802950623e-06,
      "loss": 0.6283,
      "step": 389
    },
    {
      "epoch": 0.02104581512060871,
      "grad_norm": 1.0971472320188196,
      "learning_rate": 9.990167690301153e-06,
      "loss": 0.5246,
      "step": 390
    },
    {
      "epoch": 0.02109977874912309,
      "grad_norm": 1.1513246271661204,
      "learning_rate": 9.99011722204412e-06,
      "loss": 0.6266,
      "step": 391
    },
    {
      "epoch": 0.021153742377637473,
      "grad_norm": 0.8725236933810684,
      "learning_rate": 9.990066624736577e-06,
      "loss": 0.3411,
      "step": 392
    },
    {
      "epoch": 0.021207706006151853,
      "grad_norm": 1.1007994871284874,
      "learning_rate": 9.990015898379985e-06,
      "loss": 0.603,
      "step": 393
    },
    {
      "epoch": 0.021261669634666237,
      "grad_norm": 0.981791162436866,
      "learning_rate": 9.9899650429758e-06,
      "loss": 0.5486,
      "step": 394
    },
    {
      "epoch": 0.021315633263180617,
      "grad_norm": 1.0712790034863355,
      "learning_rate": 9.989914058525483e-06,
      "loss": 0.482,
      "step": 395
    },
    {
      "epoch": 0.021369596891694997,
      "grad_norm": 1.281567362686101,
      "learning_rate": 9.989862945030497e-06,
      "loss": 0.6381,
      "step": 396
    },
    {
      "epoch": 0.02142356052020938,
      "grad_norm": 1.4066313108261324,
      "learning_rate": 9.989811702492315e-06,
      "loss": 0.7477,
      "step": 397
    },
    {
      "epoch": 0.02147752414872376,
      "grad_norm": 0.9009563673648116,
      "learning_rate": 9.989760330912407e-06,
      "loss": 0.4479,
      "step": 398
    },
    {
      "epoch": 0.02153148777723814,
      "grad_norm": 1.220952938845963,
      "learning_rate": 9.989708830292251e-06,
      "loss": 0.6219,
      "step": 399
    },
    {
      "epoch": 0.021585451405752524,
      "grad_norm": 1.2806449811219345,
      "learning_rate": 9.989657200633327e-06,
      "loss": 0.7035,
      "step": 400
    },
    {
      "epoch": 0.021639415034266904,
      "grad_norm": 1.2115281442409713,
      "learning_rate": 9.98960544193712e-06,
      "loss": 0.5445,
      "step": 401
    },
    {
      "epoch": 0.021693378662781284,
      "grad_norm": 0.9687200143125931,
      "learning_rate": 9.989553554205114e-06,
      "loss": 0.3712,
      "step": 402
    },
    {
      "epoch": 0.021747342291295668,
      "grad_norm": 0.9159610194814729,
      "learning_rate": 9.989501537438803e-06,
      "loss": 0.3589,
      "step": 403
    },
    {
      "epoch": 0.021801305919810048,
      "grad_norm": 1.1767781408873939,
      "learning_rate": 9.989449391639682e-06,
      "loss": 0.6396,
      "step": 404
    },
    {
      "epoch": 0.021855269548324428,
      "grad_norm": 1.08831950478531,
      "learning_rate": 9.989397116809248e-06,
      "loss": 0.6823,
      "step": 405
    },
    {
      "epoch": 0.02190923317683881,
      "grad_norm": 1.1423996719973566,
      "learning_rate": 9.989344712949007e-06,
      "loss": 0.6034,
      "step": 406
    },
    {
      "epoch": 0.02196319680535319,
      "grad_norm": 1.1634817089721479,
      "learning_rate": 9.98929218006046e-06,
      "loss": 0.5996,
      "step": 407
    },
    {
      "epoch": 0.022017160433867575,
      "grad_norm": 1.27239761726615,
      "learning_rate": 9.98923951814512e-06,
      "loss": 0.7378,
      "step": 408
    },
    {
      "epoch": 0.022071124062381955,
      "grad_norm": 1.2099833306386703,
      "learning_rate": 9.989186727204502e-06,
      "loss": 0.5914,
      "step": 409
    },
    {
      "epoch": 0.022125087690896335,
      "grad_norm": 1.142430265696326,
      "learning_rate": 9.989133807240121e-06,
      "loss": 0.5911,
      "step": 410
    },
    {
      "epoch": 0.02217905131941072,
      "grad_norm": 1.098549992682177,
      "learning_rate": 9.989080758253497e-06,
      "loss": 0.4004,
      "step": 411
    },
    {
      "epoch": 0.0222330149479251,
      "grad_norm": 1.182378957105608,
      "learning_rate": 9.989027580246157e-06,
      "loss": 0.5299,
      "step": 412
    },
    {
      "epoch": 0.02228697857643948,
      "grad_norm": 1.025952821533961,
      "learning_rate": 9.98897427321963e-06,
      "loss": 0.3932,
      "step": 413
    },
    {
      "epoch": 0.022340942204953862,
      "grad_norm": 1.2688654595356164,
      "learning_rate": 9.988920837175445e-06,
      "loss": 0.551,
      "step": 414
    },
    {
      "epoch": 0.022394905833468242,
      "grad_norm": 1.186693883564988,
      "learning_rate": 9.98886727211514e-06,
      "loss": 0.6896,
      "step": 415
    },
    {
      "epoch": 0.022448869461982622,
      "grad_norm": 1.1981808509414456,
      "learning_rate": 9.988813578040254e-06,
      "loss": 0.5842,
      "step": 416
    },
    {
      "epoch": 0.022502833090497006,
      "grad_norm": 1.3802870957789928,
      "learning_rate": 9.988759754952329e-06,
      "loss": 0.6488,
      "step": 417
    },
    {
      "epoch": 0.022556796719011386,
      "grad_norm": 1.323892653382377,
      "learning_rate": 9.988705802852913e-06,
      "loss": 0.5748,
      "step": 418
    },
    {
      "epoch": 0.02261076034752577,
      "grad_norm": 2.2002603319561653,
      "learning_rate": 9.988651721743558e-06,
      "loss": 0.6829,
      "step": 419
    },
    {
      "epoch": 0.02266472397604015,
      "grad_norm": 1.2480406558189578,
      "learning_rate": 9.988597511625818e-06,
      "loss": 0.6835,
      "step": 420
    },
    {
      "epoch": 0.02271868760455453,
      "grad_norm": 1.4122743308123062,
      "learning_rate": 9.988543172501247e-06,
      "loss": 0.7582,
      "step": 421
    },
    {
      "epoch": 0.022772651233068913,
      "grad_norm": 1.277262217326017,
      "learning_rate": 9.988488704371413e-06,
      "loss": 0.6366,
      "step": 422
    },
    {
      "epoch": 0.022826614861583293,
      "grad_norm": 0.9723536348145589,
      "learning_rate": 9.988434107237877e-06,
      "loss": 0.413,
      "step": 423
    },
    {
      "epoch": 0.022880578490097673,
      "grad_norm": 0.9223259724942151,
      "learning_rate": 9.98837938110221e-06,
      "loss": 0.3605,
      "step": 424
    },
    {
      "epoch": 0.022934542118612056,
      "grad_norm": 1.3905536665566678,
      "learning_rate": 9.988324525965984e-06,
      "loss": 0.6113,
      "step": 425
    },
    {
      "epoch": 0.022988505747126436,
      "grad_norm": 1.2123117468654485,
      "learning_rate": 9.988269541830775e-06,
      "loss": 0.6777,
      "step": 426
    },
    {
      "epoch": 0.023042469375640817,
      "grad_norm": 1.2493484390535818,
      "learning_rate": 9.988214428698166e-06,
      "loss": 0.4866,
      "step": 427
    },
    {
      "epoch": 0.0230964330041552,
      "grad_norm": 1.1509258716763837,
      "learning_rate": 9.988159186569738e-06,
      "loss": 0.4522,
      "step": 428
    },
    {
      "epoch": 0.02315039663266958,
      "grad_norm": 1.0837610111659282,
      "learning_rate": 9.988103815447082e-06,
      "loss": 0.4903,
      "step": 429
    },
    {
      "epoch": 0.023204360261183964,
      "grad_norm": 1.1704103386781648,
      "learning_rate": 9.988048315331784e-06,
      "loss": 0.546,
      "step": 430
    },
    {
      "epoch": 0.023258323889698344,
      "grad_norm": 1.089398629817117,
      "learning_rate": 9.987992686225444e-06,
      "loss": 0.5504,
      "step": 431
    },
    {
      "epoch": 0.023312287518212724,
      "grad_norm": 1.2721274937374005,
      "learning_rate": 9.987936928129662e-06,
      "loss": 0.539,
      "step": 432
    },
    {
      "epoch": 0.023366251146727107,
      "grad_norm": 0.9155705808375275,
      "learning_rate": 9.987881041046035e-06,
      "loss": 0.4581,
      "step": 433
    },
    {
      "epoch": 0.023420214775241487,
      "grad_norm": 1.0098221179095268,
      "learning_rate": 9.987825024976174e-06,
      "loss": 0.4936,
      "step": 434
    },
    {
      "epoch": 0.023474178403755867,
      "grad_norm": 1.1719022207004701,
      "learning_rate": 9.987768879921685e-06,
      "loss": 0.514,
      "step": 435
    },
    {
      "epoch": 0.02352814203227025,
      "grad_norm": 1.0511233620834823,
      "learning_rate": 9.987712605884185e-06,
      "loss": 0.5475,
      "step": 436
    },
    {
      "epoch": 0.02358210566078463,
      "grad_norm": 1.3894231984902898,
      "learning_rate": 9.987656202865291e-06,
      "loss": 0.5608,
      "step": 437
    },
    {
      "epoch": 0.02363606928929901,
      "grad_norm": 1.394301736069319,
      "learning_rate": 9.987599670866623e-06,
      "loss": 0.6223,
      "step": 438
    },
    {
      "epoch": 0.023690032917813394,
      "grad_norm": 1.1258535515648678,
      "learning_rate": 9.987543009889806e-06,
      "loss": 0.4578,
      "step": 439
    },
    {
      "epoch": 0.023743996546327775,
      "grad_norm": 1.0492990611438835,
      "learning_rate": 9.987486219936468e-06,
      "loss": 0.5358,
      "step": 440
    },
    {
      "epoch": 0.023797960174842158,
      "grad_norm": 0.9301331576169605,
      "learning_rate": 9.987429301008243e-06,
      "loss": 0.5961,
      "step": 441
    },
    {
      "epoch": 0.023851923803356538,
      "grad_norm": 1.147881652309137,
      "learning_rate": 9.987372253106764e-06,
      "loss": 0.5741,
      "step": 442
    },
    {
      "epoch": 0.023905887431870918,
      "grad_norm": 1.2789401882768119,
      "learning_rate": 9.987315076233673e-06,
      "loss": 0.5482,
      "step": 443
    },
    {
      "epoch": 0.0239598510603853,
      "grad_norm": 0.9678701947635667,
      "learning_rate": 9.987257770390612e-06,
      "loss": 0.5344,
      "step": 444
    },
    {
      "epoch": 0.024013814688899682,
      "grad_norm": 1.519705404305563,
      "learning_rate": 9.987200335579229e-06,
      "loss": 0.5431,
      "step": 445
    },
    {
      "epoch": 0.024067778317414062,
      "grad_norm": 1.3675015921979972,
      "learning_rate": 9.987142771801174e-06,
      "loss": 0.5168,
      "step": 446
    },
    {
      "epoch": 0.024121741945928445,
      "grad_norm": 1.0776624481236126,
      "learning_rate": 9.987085079058105e-06,
      "loss": 0.4961,
      "step": 447
    },
    {
      "epoch": 0.024175705574442825,
      "grad_norm": 1.1015060691686396,
      "learning_rate": 9.987027257351674e-06,
      "loss": 0.4721,
      "step": 448
    },
    {
      "epoch": 0.024229669202957205,
      "grad_norm": 0.9745202617892904,
      "learning_rate": 9.986969306683545e-06,
      "loss": 0.4596,
      "step": 449
    },
    {
      "epoch": 0.02428363283147159,
      "grad_norm": 1.4453594364528202,
      "learning_rate": 9.986911227055385e-06,
      "loss": 0.6479,
      "step": 450
    },
    {
      "epoch": 0.02433759645998597,
      "grad_norm": 1.3303851080342572,
      "learning_rate": 9.986853018468865e-06,
      "loss": 0.5806,
      "step": 451
    },
    {
      "epoch": 0.02439156008850035,
      "grad_norm": 1.1128854103209724,
      "learning_rate": 9.986794680925654e-06,
      "loss": 0.4633,
      "step": 452
    },
    {
      "epoch": 0.024445523717014733,
      "grad_norm": 1.382592561502497,
      "learning_rate": 9.986736214427428e-06,
      "loss": 0.5454,
      "step": 453
    },
    {
      "epoch": 0.024499487345529113,
      "grad_norm": 1.2571146851520059,
      "learning_rate": 9.986677618975875e-06,
      "loss": 0.6085,
      "step": 454
    },
    {
      "epoch": 0.024553450974043496,
      "grad_norm": 1.3180036253007132,
      "learning_rate": 9.986618894572667e-06,
      "loss": 0.602,
      "step": 455
    },
    {
      "epoch": 0.024607414602557876,
      "grad_norm": 0.9924928677849718,
      "learning_rate": 9.986560041219504e-06,
      "loss": 0.5012,
      "step": 456
    },
    {
      "epoch": 0.024661378231072256,
      "grad_norm": 1.0916553984789978,
      "learning_rate": 9.98650105891807e-06,
      "loss": 0.6007,
      "step": 457
    },
    {
      "epoch": 0.02471534185958664,
      "grad_norm": 1.1282895420477115,
      "learning_rate": 9.986441947670064e-06,
      "loss": 0.4772,
      "step": 458
    },
    {
      "epoch": 0.02476930548810102,
      "grad_norm": 1.042439146813708,
      "learning_rate": 9.98638270747718e-06,
      "loss": 0.5285,
      "step": 459
    },
    {
      "epoch": 0.0248232691166154,
      "grad_norm": 1.0563715968159886,
      "learning_rate": 9.986323338341127e-06,
      "loss": 0.5221,
      "step": 460
    },
    {
      "epoch": 0.024877232745129783,
      "grad_norm": 1.2986207979017148,
      "learning_rate": 9.986263840263606e-06,
      "loss": 0.6796,
      "step": 461
    },
    {
      "epoch": 0.024931196373644163,
      "grad_norm": 1.1084671828072887,
      "learning_rate": 9.98620421324633e-06,
      "loss": 0.6311,
      "step": 462
    },
    {
      "epoch": 0.024985160002158543,
      "grad_norm": 1.1879179337387835,
      "learning_rate": 9.986144457291012e-06,
      "loss": 0.6492,
      "step": 463
    },
    {
      "epoch": 0.025039123630672927,
      "grad_norm": 1.1349278759199277,
      "learning_rate": 9.986084572399369e-06,
      "loss": 0.4431,
      "step": 464
    },
    {
      "epoch": 0.025093087259187307,
      "grad_norm": 1.3227190707778786,
      "learning_rate": 9.986024558573122e-06,
      "loss": 0.5417,
      "step": 465
    },
    {
      "epoch": 0.02514705088770169,
      "grad_norm": 1.2593625285913357,
      "learning_rate": 9.985964415813997e-06,
      "loss": 0.5563,
      "step": 466
    },
    {
      "epoch": 0.02520101451621607,
      "grad_norm": 1.3550316587819038,
      "learning_rate": 9.985904144123723e-06,
      "loss": 0.4973,
      "step": 467
    },
    {
      "epoch": 0.02525497814473045,
      "grad_norm": 1.238208407663588,
      "learning_rate": 9.985843743504031e-06,
      "loss": 0.6036,
      "step": 468
    },
    {
      "epoch": 0.025308941773244834,
      "grad_norm": 1.3082385934445693,
      "learning_rate": 9.985783213956655e-06,
      "loss": 0.5714,
      "step": 469
    },
    {
      "epoch": 0.025362905401759214,
      "grad_norm": 1.0957314862399292,
      "learning_rate": 9.98572255548334e-06,
      "loss": 0.4711,
      "step": 470
    },
    {
      "epoch": 0.025416869030273594,
      "grad_norm": 1.3106289250592238,
      "learning_rate": 9.985661768085824e-06,
      "loss": 0.6982,
      "step": 471
    },
    {
      "epoch": 0.025470832658787978,
      "grad_norm": 1.1484042676375412,
      "learning_rate": 9.985600851765857e-06,
      "loss": 0.5221,
      "step": 472
    },
    {
      "epoch": 0.025524796287302358,
      "grad_norm": 1.0584262230980137,
      "learning_rate": 9.985539806525189e-06,
      "loss": 0.5805,
      "step": 473
    },
    {
      "epoch": 0.025578759915816738,
      "grad_norm": 1.0453924181359033,
      "learning_rate": 9.985478632365575e-06,
      "loss": 0.407,
      "step": 474
    },
    {
      "epoch": 0.02563272354433112,
      "grad_norm": 1.1058111944764053,
      "learning_rate": 9.985417329288774e-06,
      "loss": 0.516,
      "step": 475
    },
    {
      "epoch": 0.0256866871728455,
      "grad_norm": 1.2577401678807718,
      "learning_rate": 9.985355897296544e-06,
      "loss": 0.5843,
      "step": 476
    },
    {
      "epoch": 0.025740650801359885,
      "grad_norm": 1.07662485373891,
      "learning_rate": 9.985294336390656e-06,
      "loss": 0.5407,
      "step": 477
    },
    {
      "epoch": 0.025794614429874265,
      "grad_norm": 1.0695165179724182,
      "learning_rate": 9.985232646572874e-06,
      "loss": 0.4783,
      "step": 478
    },
    {
      "epoch": 0.025848578058388645,
      "grad_norm": 1.1276310322838505,
      "learning_rate": 9.985170827844976e-06,
      "loss": 0.7169,
      "step": 479
    },
    {
      "epoch": 0.02590254168690303,
      "grad_norm": 1.2047066021553443,
      "learning_rate": 9.985108880208736e-06,
      "loss": 0.4976,
      "step": 480
    },
    {
      "epoch": 0.02595650531541741,
      "grad_norm": 1.2593515997905331,
      "learning_rate": 9.985046803665934e-06,
      "loss": 0.5791,
      "step": 481
    },
    {
      "epoch": 0.02601046894393179,
      "grad_norm": 1.2016432572228684,
      "learning_rate": 9.984984598218357e-06,
      "loss": 0.5673,
      "step": 482
    },
    {
      "epoch": 0.026064432572446172,
      "grad_norm": 1.0189816790974007,
      "learning_rate": 9.984922263867789e-06,
      "loss": 0.4651,
      "step": 483
    },
    {
      "epoch": 0.026118396200960552,
      "grad_norm": 0.8988387572382215,
      "learning_rate": 9.984859800616024e-06,
      "loss": 0.3962,
      "step": 484
    },
    {
      "epoch": 0.026172359829474932,
      "grad_norm": 1.324787047734092,
      "learning_rate": 9.984797208464856e-06,
      "loss": 0.468,
      "step": 485
    },
    {
      "epoch": 0.026226323457989316,
      "grad_norm": 1.2617512630988965,
      "learning_rate": 9.984734487416086e-06,
      "loss": 0.5054,
      "step": 486
    },
    {
      "epoch": 0.026280287086503696,
      "grad_norm": 1.16735705539683,
      "learning_rate": 9.984671637471514e-06,
      "loss": 0.7776,
      "step": 487
    },
    {
      "epoch": 0.02633425071501808,
      "grad_norm": 1.004076616605721,
      "learning_rate": 9.98460865863295e-06,
      "loss": 0.4508,
      "step": 488
    },
    {
      "epoch": 0.02638821434353246,
      "grad_norm": 1.1462526187827538,
      "learning_rate": 9.9845455509022e-06,
      "loss": 0.4452,
      "step": 489
    },
    {
      "epoch": 0.02644217797204684,
      "grad_norm": 1.117487523543565,
      "learning_rate": 9.98448231428108e-06,
      "loss": 0.5695,
      "step": 490
    },
    {
      "epoch": 0.026496141600561223,
      "grad_norm": 0.9761634834798999,
      "learning_rate": 9.984418948771406e-06,
      "loss": 0.4342,
      "step": 491
    },
    {
      "epoch": 0.026550105229075603,
      "grad_norm": 1.0872276274350663,
      "learning_rate": 9.984355454375003e-06,
      "loss": 0.5526,
      "step": 492
    },
    {
      "epoch": 0.026604068857589983,
      "grad_norm": 1.0285732840486945,
      "learning_rate": 9.98429183109369e-06,
      "loss": 0.4997,
      "step": 493
    },
    {
      "epoch": 0.026658032486104367,
      "grad_norm": 1.5225246708669666,
      "learning_rate": 9.9842280789293e-06,
      "loss": 0.761,
      "step": 494
    },
    {
      "epoch": 0.026711996114618747,
      "grad_norm": 1.074960375956489,
      "learning_rate": 9.984164197883664e-06,
      "loss": 0.6129,
      "step": 495
    },
    {
      "epoch": 0.026765959743133127,
      "grad_norm": 1.006262347022148,
      "learning_rate": 9.984100187958618e-06,
      "loss": 0.4775,
      "step": 496
    },
    {
      "epoch": 0.02681992337164751,
      "grad_norm": 1.0262174107545028,
      "learning_rate": 9.984036049156002e-06,
      "loss": 0.4773,
      "step": 497
    },
    {
      "epoch": 0.02687388700016189,
      "grad_norm": 1.414132776550392,
      "learning_rate": 9.98397178147766e-06,
      "loss": 0.6144,
      "step": 498
    },
    {
      "epoch": 0.02692785062867627,
      "grad_norm": 1.1644426924260391,
      "learning_rate": 9.983907384925436e-06,
      "loss": 0.6346,
      "step": 499
    },
    {
      "epoch": 0.026981814257190654,
      "grad_norm": 1.0919694577033208,
      "learning_rate": 9.983842859501185e-06,
      "loss": 0.4418,
      "step": 500
    },
    {
      "epoch": 0.026981814257190654,
      "eval_loss": 0.6459314227104187,
      "eval_runtime": 164.1726,
      "eval_samples_per_second": 20.947,
      "eval_steps_per_second": 0.877,
      "step": 500
    },
    {
      "epoch": 0.027035777885705034,
      "grad_norm": 1.110109343155162,
      "learning_rate": 9.98377820520676e-06,
      "loss": 0.476,
      "step": 501
    },
    {
      "epoch": 0.027089741514219418,
      "grad_norm": 0.964460134527613,
      "learning_rate": 9.983713422044016e-06,
      "loss": 0.5242,
      "step": 502
    },
    {
      "epoch": 0.027143705142733798,
      "grad_norm": 1.2029588344530528,
      "learning_rate": 9.983648510014821e-06,
      "loss": 0.5624,
      "step": 503
    },
    {
      "epoch": 0.027197668771248178,
      "grad_norm": 1.0453470072965703,
      "learning_rate": 9.983583469121035e-06,
      "loss": 0.492,
      "step": 504
    },
    {
      "epoch": 0.02725163239976256,
      "grad_norm": 1.1008645815352225,
      "learning_rate": 9.983518299364531e-06,
      "loss": 0.5241,
      "step": 505
    },
    {
      "epoch": 0.02730559602827694,
      "grad_norm": 0.9139297808075197,
      "learning_rate": 9.983453000747181e-06,
      "loss": 0.4855,
      "step": 506
    },
    {
      "epoch": 0.02735955965679132,
      "grad_norm": 1.1774237738671027,
      "learning_rate": 9.983387573270861e-06,
      "loss": 0.4794,
      "step": 507
    },
    {
      "epoch": 0.027413523285305705,
      "grad_norm": 1.018400989814044,
      "learning_rate": 9.983322016937452e-06,
      "loss": 0.5955,
      "step": 508
    },
    {
      "epoch": 0.027467486913820085,
      "grad_norm": 1.1984477181782833,
      "learning_rate": 9.983256331748839e-06,
      "loss": 0.6075,
      "step": 509
    },
    {
      "epoch": 0.027521450542334465,
      "grad_norm": 1.2314614349710196,
      "learning_rate": 9.983190517706907e-06,
      "loss": 0.4281,
      "step": 510
    },
    {
      "epoch": 0.02757541417084885,
      "grad_norm": 1.1892243260952917,
      "learning_rate": 9.983124574813551e-06,
      "loss": 0.4865,
      "step": 511
    },
    {
      "epoch": 0.02762937779936323,
      "grad_norm": 1.3971079333132348,
      "learning_rate": 9.983058503070665e-06,
      "loss": 0.6548,
      "step": 512
    },
    {
      "epoch": 0.027683341427877612,
      "grad_norm": 1.233557591521009,
      "learning_rate": 9.982992302480148e-06,
      "loss": 0.5943,
      "step": 513
    },
    {
      "epoch": 0.027737305056391992,
      "grad_norm": 1.2284263058598675,
      "learning_rate": 9.982925973043902e-06,
      "loss": 0.5194,
      "step": 514
    },
    {
      "epoch": 0.027791268684906372,
      "grad_norm": 1.178338326028012,
      "learning_rate": 9.982859514763835e-06,
      "loss": 0.6166,
      "step": 515
    },
    {
      "epoch": 0.027845232313420756,
      "grad_norm": 0.9951496679963399,
      "learning_rate": 9.982792927641854e-06,
      "loss": 0.4512,
      "step": 516
    },
    {
      "epoch": 0.027899195941935136,
      "grad_norm": 0.8887784595612234,
      "learning_rate": 9.982726211679876e-06,
      "loss": 0.3754,
      "step": 517
    },
    {
      "epoch": 0.027953159570449516,
      "grad_norm": 1.24765799032227,
      "learning_rate": 9.982659366879817e-06,
      "loss": 0.548,
      "step": 518
    },
    {
      "epoch": 0.0280071231989639,
      "grad_norm": 1.2782875460704954,
      "learning_rate": 9.982592393243599e-06,
      "loss": 0.5232,
      "step": 519
    },
    {
      "epoch": 0.02806108682747828,
      "grad_norm": 0.9872821160150816,
      "learning_rate": 9.982525290773146e-06,
      "loss": 0.452,
      "step": 520
    },
    {
      "epoch": 0.02811505045599266,
      "grad_norm": 1.2844856524366668,
      "learning_rate": 9.982458059470386e-06,
      "loss": 0.7114,
      "step": 521
    },
    {
      "epoch": 0.028169014084507043,
      "grad_norm": 1.0825546900075842,
      "learning_rate": 9.982390699337253e-06,
      "loss": 0.5508,
      "step": 522
    },
    {
      "epoch": 0.028222977713021423,
      "grad_norm": 1.1895834434933414,
      "learning_rate": 9.982323210375681e-06,
      "loss": 0.5057,
      "step": 523
    },
    {
      "epoch": 0.028276941341535806,
      "grad_norm": 1.0794046514082432,
      "learning_rate": 9.982255592587612e-06,
      "loss": 0.4881,
      "step": 524
    },
    {
      "epoch": 0.028330904970050187,
      "grad_norm": 0.7589128492820196,
      "learning_rate": 9.982187845974988e-06,
      "loss": 0.3514,
      "step": 525
    },
    {
      "epoch": 0.028384868598564567,
      "grad_norm": 0.9777096703164789,
      "learning_rate": 9.982119970539757e-06,
      "loss": 0.4498,
      "step": 526
    },
    {
      "epoch": 0.02843883222707895,
      "grad_norm": 0.8718822369283625,
      "learning_rate": 9.98205196628387e-06,
      "loss": 0.405,
      "step": 527
    },
    {
      "epoch": 0.02849279585559333,
      "grad_norm": 1.116196074637517,
      "learning_rate": 9.981983833209279e-06,
      "loss": 0.6034,
      "step": 528
    },
    {
      "epoch": 0.02854675948410771,
      "grad_norm": 1.1676009518710115,
      "learning_rate": 9.981915571317942e-06,
      "loss": 0.4965,
      "step": 529
    },
    {
      "epoch": 0.028600723112622094,
      "grad_norm": 1.4052539130464754,
      "learning_rate": 9.981847180611826e-06,
      "loss": 0.6058,
      "step": 530
    },
    {
      "epoch": 0.028654686741136474,
      "grad_norm": 1.2334131614113848,
      "learning_rate": 9.981778661092893e-06,
      "loss": 0.6446,
      "step": 531
    },
    {
      "epoch": 0.028708650369650854,
      "grad_norm": 1.2741530240332686,
      "learning_rate": 9.981710012763111e-06,
      "loss": 0.5937,
      "step": 532
    },
    {
      "epoch": 0.028762613998165237,
      "grad_norm": 1.0247503718233553,
      "learning_rate": 9.981641235624457e-06,
      "loss": 0.6278,
      "step": 533
    },
    {
      "epoch": 0.028816577626679617,
      "grad_norm": 0.6977777911839362,
      "learning_rate": 9.981572329678904e-06,
      "loss": 0.3395,
      "step": 534
    },
    {
      "epoch": 0.028870541255194,
      "grad_norm": 1.1374177735247473,
      "learning_rate": 9.981503294928435e-06,
      "loss": 0.5862,
      "step": 535
    },
    {
      "epoch": 0.02892450488370838,
      "grad_norm": 1.1837132664625118,
      "learning_rate": 9.981434131375032e-06,
      "loss": 0.5048,
      "step": 536
    },
    {
      "epoch": 0.02897846851222276,
      "grad_norm": 1.110877117304451,
      "learning_rate": 9.981364839020685e-06,
      "loss": 0.6427,
      "step": 537
    },
    {
      "epoch": 0.029032432140737145,
      "grad_norm": 1.0662793460128148,
      "learning_rate": 9.981295417867384e-06,
      "loss": 0.4295,
      "step": 538
    },
    {
      "epoch": 0.029086395769251525,
      "grad_norm": 1.1779090743849445,
      "learning_rate": 9.981225867917124e-06,
      "loss": 0.5002,
      "step": 539
    },
    {
      "epoch": 0.029140359397765905,
      "grad_norm": 1.3172880780435805,
      "learning_rate": 9.981156189171907e-06,
      "loss": 0.6471,
      "step": 540
    },
    {
      "epoch": 0.029194323026280288,
      "grad_norm": 0.9266136440174673,
      "learning_rate": 9.98108638163373e-06,
      "loss": 0.3623,
      "step": 541
    },
    {
      "epoch": 0.029248286654794668,
      "grad_norm": 1.2667509480810353,
      "learning_rate": 9.981016445304604e-06,
      "loss": 0.7001,
      "step": 542
    },
    {
      "epoch": 0.02930225028330905,
      "grad_norm": 1.097025919017542,
      "learning_rate": 9.980946380186538e-06,
      "loss": 0.4448,
      "step": 543
    },
    {
      "epoch": 0.029356213911823432,
      "grad_norm": 1.0759727430485406,
      "learning_rate": 9.980876186281545e-06,
      "loss": 0.5169,
      "step": 544
    },
    {
      "epoch": 0.029410177540337812,
      "grad_norm": 0.9975899107011369,
      "learning_rate": 9.980805863591643e-06,
      "loss": 0.5351,
      "step": 545
    },
    {
      "epoch": 0.029464141168852192,
      "grad_norm": 1.1942104104934133,
      "learning_rate": 9.980735412118852e-06,
      "loss": 0.6671,
      "step": 546
    },
    {
      "epoch": 0.029518104797366575,
      "grad_norm": 0.9218625952339685,
      "learning_rate": 9.9806648318652e-06,
      "loss": 0.3542,
      "step": 547
    },
    {
      "epoch": 0.029572068425880955,
      "grad_norm": 1.1310218531366758,
      "learning_rate": 9.98059412283271e-06,
      "loss": 0.4743,
      "step": 548
    },
    {
      "epoch": 0.02962603205439534,
      "grad_norm": 1.0386891504578748,
      "learning_rate": 9.98052328502342e-06,
      "loss": 0.5505,
      "step": 549
    },
    {
      "epoch": 0.02967999568290972,
      "grad_norm": 1.3161231300084628,
      "learning_rate": 9.980452318439364e-06,
      "loss": 0.653,
      "step": 550
    },
    {
      "epoch": 0.0297339593114241,
      "grad_norm": 1.1787767975767434,
      "learning_rate": 9.980381223082581e-06,
      "loss": 0.6018,
      "step": 551
    },
    {
      "epoch": 0.029787922939938483,
      "grad_norm": 1.2327431095876908,
      "learning_rate": 9.980309998955113e-06,
      "loss": 0.4534,
      "step": 552
    },
    {
      "epoch": 0.029841886568452863,
      "grad_norm": 1.0123075863963638,
      "learning_rate": 9.980238646059009e-06,
      "loss": 0.5781,
      "step": 553
    },
    {
      "epoch": 0.029895850196967243,
      "grad_norm": 1.2074894084933527,
      "learning_rate": 9.980167164396321e-06,
      "loss": 0.6345,
      "step": 554
    },
    {
      "epoch": 0.029949813825481626,
      "grad_norm": 1.2016242071003427,
      "learning_rate": 9.980095553969101e-06,
      "loss": 0.5373,
      "step": 555
    },
    {
      "epoch": 0.030003777453996006,
      "grad_norm": 0.8842356569430259,
      "learning_rate": 9.980023814779407e-06,
      "loss": 0.3865,
      "step": 556
    },
    {
      "epoch": 0.030057741082510386,
      "grad_norm": 1.0238042871637816,
      "learning_rate": 9.979951946829302e-06,
      "loss": 0.3351,
      "step": 557
    },
    {
      "epoch": 0.03011170471102477,
      "grad_norm": 1.4603793159419332,
      "learning_rate": 9.979879950120852e-06,
      "loss": 0.7998,
      "step": 558
    },
    {
      "epoch": 0.03016566833953915,
      "grad_norm": 1.1312441024107023,
      "learning_rate": 9.979807824656125e-06,
      "loss": 0.435,
      "step": 559
    },
    {
      "epoch": 0.030219631968053533,
      "grad_norm": 1.1573531913038548,
      "learning_rate": 9.979735570437196e-06,
      "loss": 0.53,
      "step": 560
    },
    {
      "epoch": 0.030273595596567913,
      "grad_norm": 1.0057770230043936,
      "learning_rate": 9.979663187466139e-06,
      "loss": 0.4392,
      "step": 561
    },
    {
      "epoch": 0.030327559225082294,
      "grad_norm": 1.2583548279674015,
      "learning_rate": 9.979590675745036e-06,
      "loss": 0.5539,
      "step": 562
    },
    {
      "epoch": 0.030381522853596677,
      "grad_norm": 1.3151013253146489,
      "learning_rate": 9.979518035275972e-06,
      "loss": 0.4692,
      "step": 563
    },
    {
      "epoch": 0.030435486482111057,
      "grad_norm": 1.1230285182334558,
      "learning_rate": 9.97944526606103e-06,
      "loss": 0.6521,
      "step": 564
    },
    {
      "epoch": 0.030489450110625437,
      "grad_norm": 1.2629477082608564,
      "learning_rate": 9.979372368102307e-06,
      "loss": 0.5995,
      "step": 565
    },
    {
      "epoch": 0.03054341373913982,
      "grad_norm": 0.9839501803153108,
      "learning_rate": 9.979299341401896e-06,
      "loss": 0.4172,
      "step": 566
    },
    {
      "epoch": 0.0305973773676542,
      "grad_norm": 1.0221909418443713,
      "learning_rate": 9.979226185961898e-06,
      "loss": 0.4144,
      "step": 567
    },
    {
      "epoch": 0.03065134099616858,
      "grad_norm": 1.2836373848293925,
      "learning_rate": 9.979152901784411e-06,
      "loss": 0.6641,
      "step": 568
    },
    {
      "epoch": 0.030705304624682964,
      "grad_norm": 0.9175894057112968,
      "learning_rate": 9.979079488871544e-06,
      "loss": 0.4045,
      "step": 569
    },
    {
      "epoch": 0.030759268253197344,
      "grad_norm": 0.7862773944221269,
      "learning_rate": 9.979005947225407e-06,
      "loss": 0.3637,
      "step": 570
    },
    {
      "epoch": 0.030813231881711728,
      "grad_norm": 1.1386824914638112,
      "learning_rate": 9.978932276848112e-06,
      "loss": 0.4089,
      "step": 571
    },
    {
      "epoch": 0.030867195510226108,
      "grad_norm": 1.1261748321601806,
      "learning_rate": 9.97885847774178e-06,
      "loss": 0.5087,
      "step": 572
    },
    {
      "epoch": 0.030921159138740488,
      "grad_norm": 1.106878664327847,
      "learning_rate": 9.97878454990853e-06,
      "loss": 0.5337,
      "step": 573
    },
    {
      "epoch": 0.03097512276725487,
      "grad_norm": 1.2136879322490512,
      "learning_rate": 9.978710493350485e-06,
      "loss": 0.5178,
      "step": 574
    },
    {
      "epoch": 0.03102908639576925,
      "grad_norm": 0.9820022755592402,
      "learning_rate": 9.978636308069776e-06,
      "loss": 0.487,
      "step": 575
    },
    {
      "epoch": 0.03108305002428363,
      "grad_norm": 1.0470547617291561,
      "learning_rate": 9.978561994068532e-06,
      "loss": 0.4949,
      "step": 576
    },
    {
      "epoch": 0.031137013652798015,
      "grad_norm": 1.055441311740176,
      "learning_rate": 9.978487551348894e-06,
      "loss": 0.5426,
      "step": 577
    },
    {
      "epoch": 0.031190977281312395,
      "grad_norm": 1.2417993655060524,
      "learning_rate": 9.978412979912996e-06,
      "loss": 0.6439,
      "step": 578
    },
    {
      "epoch": 0.031244940909826775,
      "grad_norm": 1.1683808146319994,
      "learning_rate": 9.978338279762986e-06,
      "loss": 0.4765,
      "step": 579
    },
    {
      "epoch": 0.03129890453834116,
      "grad_norm": 1.1795114220484655,
      "learning_rate": 9.978263450901008e-06,
      "loss": 0.7299,
      "step": 580
    },
    {
      "epoch": 0.03135286816685554,
      "grad_norm": 1.1473697552121351,
      "learning_rate": 9.978188493329215e-06,
      "loss": 0.4675,
      "step": 581
    },
    {
      "epoch": 0.03140683179536992,
      "grad_norm": 0.9443838275186192,
      "learning_rate": 9.978113407049756e-06,
      "loss": 0.5326,
      "step": 582
    },
    {
      "epoch": 0.0314607954238843,
      "grad_norm": 1.1037380205963083,
      "learning_rate": 9.978038192064794e-06,
      "loss": 0.4849,
      "step": 583
    },
    {
      "epoch": 0.031514759052398686,
      "grad_norm": 1.1646219496312937,
      "learning_rate": 9.977962848376492e-06,
      "loss": 0.5496,
      "step": 584
    },
    {
      "epoch": 0.03156872268091306,
      "grad_norm": 1.0951060593829312,
      "learning_rate": 9.977887375987011e-06,
      "loss": 0.5345,
      "step": 585
    },
    {
      "epoch": 0.031622686309427446,
      "grad_norm": 1.3235756308444222,
      "learning_rate": 9.977811774898523e-06,
      "loss": 0.5481,
      "step": 586
    },
    {
      "epoch": 0.03167664993794183,
      "grad_norm": 0.9793840181262841,
      "learning_rate": 9.977736045113201e-06,
      "loss": 0.5435,
      "step": 587
    },
    {
      "epoch": 0.031730613566456206,
      "grad_norm": 1.105382311223432,
      "learning_rate": 9.977660186633219e-06,
      "loss": 0.5132,
      "step": 588
    },
    {
      "epoch": 0.03178457719497059,
      "grad_norm": 1.1200007439249184,
      "learning_rate": 9.977584199460762e-06,
      "loss": 0.4973,
      "step": 589
    },
    {
      "epoch": 0.03183854082348497,
      "grad_norm": 1.249298460297479,
      "learning_rate": 9.977508083598006e-06,
      "loss": 0.6439,
      "step": 590
    },
    {
      "epoch": 0.03189250445199935,
      "grad_norm": 1.2530124122559925,
      "learning_rate": 9.977431839047146e-06,
      "loss": 0.8403,
      "step": 591
    },
    {
      "epoch": 0.03194646808051373,
      "grad_norm": 0.997271518388178,
      "learning_rate": 9.977355465810371e-06,
      "loss": 0.4812,
      "step": 592
    },
    {
      "epoch": 0.03200043170902812,
      "grad_norm": 1.3074725702897192,
      "learning_rate": 9.977278963889875e-06,
      "loss": 0.6025,
      "step": 593
    },
    {
      "epoch": 0.03205439533754249,
      "grad_norm": 1.038322668837078,
      "learning_rate": 9.97720233328786e-06,
      "loss": 0.4901,
      "step": 594
    },
    {
      "epoch": 0.03210835896605688,
      "grad_norm": 1.1451343706709212,
      "learning_rate": 9.977125574006525e-06,
      "loss": 0.5426,
      "step": 595
    },
    {
      "epoch": 0.03216232259457126,
      "grad_norm": 1.2168745905748297,
      "learning_rate": 9.977048686048075e-06,
      "loss": 0.5768,
      "step": 596
    },
    {
      "epoch": 0.03221628622308564,
      "grad_norm": 1.1127259232538744,
      "learning_rate": 9.976971669414724e-06,
      "loss": 0.5462,
      "step": 597
    },
    {
      "epoch": 0.03227024985160002,
      "grad_norm": 1.1520816202174538,
      "learning_rate": 9.976894524108683e-06,
      "loss": 0.6138,
      "step": 598
    },
    {
      "epoch": 0.032324213480114404,
      "grad_norm": 1.0405659237347167,
      "learning_rate": 9.97681725013217e-06,
      "loss": 0.4405,
      "step": 599
    },
    {
      "epoch": 0.03237817710862879,
      "grad_norm": 1.2517486504367097,
      "learning_rate": 9.976739847487405e-06,
      "loss": 0.4688,
      "step": 600
    },
    {
      "epoch": 0.032432140737143164,
      "grad_norm": 0.9766098049663949,
      "learning_rate": 9.976662316176615e-06,
      "loss": 0.4981,
      "step": 601
    },
    {
      "epoch": 0.03248610436565755,
      "grad_norm": 1.3559931862898909,
      "learning_rate": 9.976584656202025e-06,
      "loss": 0.5543,
      "step": 602
    },
    {
      "epoch": 0.03254006799417193,
      "grad_norm": 1.4593350177312336,
      "learning_rate": 9.976506867565871e-06,
      "loss": 0.7446,
      "step": 603
    },
    {
      "epoch": 0.03259403162268631,
      "grad_norm": 1.0917633627665384,
      "learning_rate": 9.976428950270383e-06,
      "loss": 0.3551,
      "step": 604
    },
    {
      "epoch": 0.03264799525120069,
      "grad_norm": 1.283019691463763,
      "learning_rate": 9.976350904317806e-06,
      "loss": 0.5902,
      "step": 605
    },
    {
      "epoch": 0.032701958879715075,
      "grad_norm": 1.1034219477728264,
      "learning_rate": 9.976272729710383e-06,
      "loss": 0.4989,
      "step": 606
    },
    {
      "epoch": 0.03275592250822945,
      "grad_norm": 1.424922751570215,
      "learning_rate": 9.976194426450355e-06,
      "loss": 0.7038,
      "step": 607
    },
    {
      "epoch": 0.032809886136743835,
      "grad_norm": 1.1740696710472487,
      "learning_rate": 9.97611599453998e-06,
      "loss": 0.4829,
      "step": 608
    },
    {
      "epoch": 0.03286384976525822,
      "grad_norm": 1.1535403703450355,
      "learning_rate": 9.976037433981505e-06,
      "loss": 0.4847,
      "step": 609
    },
    {
      "epoch": 0.032917813393772595,
      "grad_norm": 0.9846926316613003,
      "learning_rate": 9.975958744777193e-06,
      "loss": 0.448,
      "step": 610
    },
    {
      "epoch": 0.03297177702228698,
      "grad_norm": 1.101924282582547,
      "learning_rate": 9.975879926929303e-06,
      "loss": 0.4891,
      "step": 611
    },
    {
      "epoch": 0.03302574065080136,
      "grad_norm": 1.3461251105948324,
      "learning_rate": 9.975800980440103e-06,
      "loss": 0.5654,
      "step": 612
    },
    {
      "epoch": 0.03307970427931574,
      "grad_norm": 1.6082354103507113,
      "learning_rate": 9.97572190531186e-06,
      "loss": 0.738,
      "step": 613
    },
    {
      "epoch": 0.03313366790783012,
      "grad_norm": 0.8235800550886619,
      "learning_rate": 9.975642701546846e-06,
      "loss": 0.3482,
      "step": 614
    },
    {
      "epoch": 0.033187631536344506,
      "grad_norm": 0.9766243693482498,
      "learning_rate": 9.97556336914734e-06,
      "loss": 0.4409,
      "step": 615
    },
    {
      "epoch": 0.03324159516485888,
      "grad_norm": 1.089966889925386,
      "learning_rate": 9.97548390811562e-06,
      "loss": 0.5387,
      "step": 616
    },
    {
      "epoch": 0.033295558793373266,
      "grad_norm": 1.1246783363316304,
      "learning_rate": 9.975404318453968e-06,
      "loss": 0.4865,
      "step": 617
    },
    {
      "epoch": 0.03334952242188765,
      "grad_norm": 1.180499738036913,
      "learning_rate": 9.975324600164678e-06,
      "loss": 0.5316,
      "step": 618
    },
    {
      "epoch": 0.033403486050402026,
      "grad_norm": 0.9249326514007677,
      "learning_rate": 9.975244753250035e-06,
      "loss": 0.4847,
      "step": 619
    },
    {
      "epoch": 0.03345744967891641,
      "grad_norm": 0.9733414663728588,
      "learning_rate": 9.975164777712335e-06,
      "loss": 0.3706,
      "step": 620
    },
    {
      "epoch": 0.03351141330743079,
      "grad_norm": 1.2304692361003597,
      "learning_rate": 9.975084673553878e-06,
      "loss": 0.5161,
      "step": 621
    },
    {
      "epoch": 0.03356537693594517,
      "grad_norm": 1.1947034439345585,
      "learning_rate": 9.975004440776965e-06,
      "loss": 0.5345,
      "step": 622
    },
    {
      "epoch": 0.03361934056445955,
      "grad_norm": 0.9441431259346416,
      "learning_rate": 9.974924079383905e-06,
      "loss": 0.4389,
      "step": 623
    },
    {
      "epoch": 0.03367330419297394,
      "grad_norm": 1.2286766556983493,
      "learning_rate": 9.974843589377003e-06,
      "loss": 0.4402,
      "step": 624
    },
    {
      "epoch": 0.03372726782148832,
      "grad_norm": 1.000082075862358,
      "learning_rate": 9.974762970758576e-06,
      "loss": 0.4443,
      "step": 625
    },
    {
      "epoch": 0.0337812314500027,
      "grad_norm": 1.1755997199423103,
      "learning_rate": 9.97468222353094e-06,
      "loss": 0.4008,
      "step": 626
    },
    {
      "epoch": 0.03383519507851708,
      "grad_norm": 0.9723235074841852,
      "learning_rate": 9.974601347696414e-06,
      "loss": 0.5516,
      "step": 627
    },
    {
      "epoch": 0.033889158707031464,
      "grad_norm": 0.9221374364490146,
      "learning_rate": 9.974520343257328e-06,
      "loss": 0.5571,
      "step": 628
    },
    {
      "epoch": 0.03394312233554584,
      "grad_norm": 1.0773764898622797,
      "learning_rate": 9.974439210216002e-06,
      "loss": 0.4685,
      "step": 629
    },
    {
      "epoch": 0.033997085964060224,
      "grad_norm": 1.107310253089235,
      "learning_rate": 9.974357948574774e-06,
      "loss": 0.4057,
      "step": 630
    },
    {
      "epoch": 0.03405104959257461,
      "grad_norm": 0.9162615412650591,
      "learning_rate": 9.974276558335975e-06,
      "loss": 0.3731,
      "step": 631
    },
    {
      "epoch": 0.034105013221088984,
      "grad_norm": 1.2586936932631536,
      "learning_rate": 9.974195039501949e-06,
      "loss": 0.5618,
      "step": 632
    },
    {
      "epoch": 0.03415897684960337,
      "grad_norm": 1.005459526986536,
      "learning_rate": 9.974113392075036e-06,
      "loss": 0.469,
      "step": 633
    },
    {
      "epoch": 0.03421294047811775,
      "grad_norm": 1.2129018180995976,
      "learning_rate": 9.974031616057584e-06,
      "loss": 0.5849,
      "step": 634
    },
    {
      "epoch": 0.03426690410663213,
      "grad_norm": 1.1055894264113686,
      "learning_rate": 9.973949711451943e-06,
      "loss": 0.5218,
      "step": 635
    },
    {
      "epoch": 0.03432086773514651,
      "grad_norm": 1.0244879733632453,
      "learning_rate": 9.973867678260466e-06,
      "loss": 0.4557,
      "step": 636
    },
    {
      "epoch": 0.034374831363660895,
      "grad_norm": 0.8427496611135044,
      "learning_rate": 9.97378551648551e-06,
      "loss": 0.3866,
      "step": 637
    },
    {
      "epoch": 0.03442879499217527,
      "grad_norm": 0.9940139937032199,
      "learning_rate": 9.97370322612944e-06,
      "loss": 0.4317,
      "step": 638
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 0.9907925129356866,
      "learning_rate": 9.97362080719462e-06,
      "loss": 0.4346,
      "step": 639
    },
    {
      "epoch": 0.03453672224920404,
      "grad_norm": 1.044847070827823,
      "learning_rate": 9.973538259683414e-06,
      "loss": 0.5501,
      "step": 640
    },
    {
      "epoch": 0.034590685877718415,
      "grad_norm": 1.2963142234244194,
      "learning_rate": 9.973455583598202e-06,
      "loss": 0.5729,
      "step": 641
    },
    {
      "epoch": 0.0346446495062328,
      "grad_norm": 1.1338078291252127,
      "learning_rate": 9.973372778941355e-06,
      "loss": 0.5105,
      "step": 642
    },
    {
      "epoch": 0.03469861313474718,
      "grad_norm": 1.2530568643852007,
      "learning_rate": 9.973289845715253e-06,
      "loss": 0.6123,
      "step": 643
    },
    {
      "epoch": 0.03475257676326156,
      "grad_norm": 1.0265232539412,
      "learning_rate": 9.973206783922282e-06,
      "loss": 0.4767,
      "step": 644
    },
    {
      "epoch": 0.03480654039177594,
      "grad_norm": 1.1588369685666042,
      "learning_rate": 9.973123593564828e-06,
      "loss": 0.6759,
      "step": 645
    },
    {
      "epoch": 0.034860504020290325,
      "grad_norm": 1.0746359596862989,
      "learning_rate": 9.973040274645282e-06,
      "loss": 0.4395,
      "step": 646
    },
    {
      "epoch": 0.03491446764880471,
      "grad_norm": 1.162708447503128,
      "learning_rate": 9.97295682716604e-06,
      "loss": 0.6355,
      "step": 647
    },
    {
      "epoch": 0.034968431277319086,
      "grad_norm": 1.1489117358412222,
      "learning_rate": 9.972873251129497e-06,
      "loss": 0.5061,
      "step": 648
    },
    {
      "epoch": 0.03502239490583347,
      "grad_norm": 1.052277830082588,
      "learning_rate": 9.972789546538059e-06,
      "loss": 0.5428,
      "step": 649
    },
    {
      "epoch": 0.03507635853434785,
      "grad_norm": 1.382702133255439,
      "learning_rate": 9.97270571339413e-06,
      "loss": 0.6205,
      "step": 650
    },
    {
      "epoch": 0.03513032216286223,
      "grad_norm": 1.1370874959665158,
      "learning_rate": 9.972621751700118e-06,
      "loss": 0.528,
      "step": 651
    },
    {
      "epoch": 0.03518428579137661,
      "grad_norm": 1.204968053967647,
      "learning_rate": 9.97253766145844e-06,
      "loss": 0.4759,
      "step": 652
    },
    {
      "epoch": 0.035238249419890996,
      "grad_norm": 1.1342280326210048,
      "learning_rate": 9.972453442671508e-06,
      "loss": 0.5389,
      "step": 653
    },
    {
      "epoch": 0.03529221304840537,
      "grad_norm": 0.8431360675819526,
      "learning_rate": 9.972369095341745e-06,
      "loss": 0.2807,
      "step": 654
    },
    {
      "epoch": 0.035346176676919756,
      "grad_norm": 1.185401144989261,
      "learning_rate": 9.972284619471577e-06,
      "loss": 0.5785,
      "step": 655
    },
    {
      "epoch": 0.03540014030543414,
      "grad_norm": 1.1749968169977327,
      "learning_rate": 9.97220001506343e-06,
      "loss": 0.6228,
      "step": 656
    },
    {
      "epoch": 0.035454103933948516,
      "grad_norm": 1.4371790572942826,
      "learning_rate": 9.972115282119736e-06,
      "loss": 0.6856,
      "step": 657
    },
    {
      "epoch": 0.0355080675624629,
      "grad_norm": 0.9043149055984723,
      "learning_rate": 9.972030420642927e-06,
      "loss": 0.4869,
      "step": 658
    },
    {
      "epoch": 0.035562031190977283,
      "grad_norm": 1.2276827475819863,
      "learning_rate": 9.971945430635448e-06,
      "loss": 0.6236,
      "step": 659
    },
    {
      "epoch": 0.03561599481949166,
      "grad_norm": 1.164631647536771,
      "learning_rate": 9.971860312099737e-06,
      "loss": 0.5166,
      "step": 660
    },
    {
      "epoch": 0.035669958448006044,
      "grad_norm": 0.9230512322782697,
      "learning_rate": 9.971775065038246e-06,
      "loss": 0.3973,
      "step": 661
    },
    {
      "epoch": 0.03572392207652043,
      "grad_norm": 1.053792932336806,
      "learning_rate": 9.971689689453416e-06,
      "loss": 0.4266,
      "step": 662
    },
    {
      "epoch": 0.035777885705034804,
      "grad_norm": 1.100343683454817,
      "learning_rate": 9.97160418534771e-06,
      "loss": 0.5455,
      "step": 663
    },
    {
      "epoch": 0.03583184933354919,
      "grad_norm": 1.1305454234247014,
      "learning_rate": 9.971518552723578e-06,
      "loss": 0.5087,
      "step": 664
    },
    {
      "epoch": 0.03588581296206357,
      "grad_norm": 1.3512641711109625,
      "learning_rate": 9.971432791583487e-06,
      "loss": 0.7354,
      "step": 665
    },
    {
      "epoch": 0.03593977659057795,
      "grad_norm": 1.44580082437797,
      "learning_rate": 9.971346901929898e-06,
      "loss": 0.731,
      "step": 666
    },
    {
      "epoch": 0.03599374021909233,
      "grad_norm": 1.1961213600145169,
      "learning_rate": 9.971260883765284e-06,
      "loss": 0.5212,
      "step": 667
    },
    {
      "epoch": 0.036047703847606714,
      "grad_norm": 1.1945660289743327,
      "learning_rate": 9.97117473709211e-06,
      "loss": 0.6593,
      "step": 668
    },
    {
      "epoch": 0.03610166747612109,
      "grad_norm": 1.254461644769536,
      "learning_rate": 9.97108846191286e-06,
      "loss": 0.6091,
      "step": 669
    },
    {
      "epoch": 0.036155631104635474,
      "grad_norm": 1.1047742408465668,
      "learning_rate": 9.971002058230009e-06,
      "loss": 0.5427,
      "step": 670
    },
    {
      "epoch": 0.03620959473314986,
      "grad_norm": 1.0630864865456484,
      "learning_rate": 9.970915526046042e-06,
      "loss": 0.5027,
      "step": 671
    },
    {
      "epoch": 0.03626355836166424,
      "grad_norm": 1.0951737585047432,
      "learning_rate": 9.970828865363443e-06,
      "loss": 0.6427,
      "step": 672
    },
    {
      "epoch": 0.03631752199017862,
      "grad_norm": 0.9025619571489237,
      "learning_rate": 9.970742076184706e-06,
      "loss": 0.4295,
      "step": 673
    },
    {
      "epoch": 0.036371485618693,
      "grad_norm": 1.1537471806161166,
      "learning_rate": 9.970655158512324e-06,
      "loss": 0.7454,
      "step": 674
    },
    {
      "epoch": 0.036425449247207385,
      "grad_norm": 1.2074926990006456,
      "learning_rate": 9.970568112348795e-06,
      "loss": 0.6205,
      "step": 675
    },
    {
      "epoch": 0.03647941287572176,
      "grad_norm": 1.1626190471402271,
      "learning_rate": 9.970480937696622e-06,
      "loss": 0.6302,
      "step": 676
    },
    {
      "epoch": 0.036533376504236145,
      "grad_norm": 1.135494630396891,
      "learning_rate": 9.970393634558311e-06,
      "loss": 0.6574,
      "step": 677
    },
    {
      "epoch": 0.03658734013275053,
      "grad_norm": 0.8561688553054742,
      "learning_rate": 9.970306202936368e-06,
      "loss": 0.3822,
      "step": 678
    },
    {
      "epoch": 0.036641303761264905,
      "grad_norm": 1.1306487387899604,
      "learning_rate": 9.970218642833308e-06,
      "loss": 0.5182,
      "step": 679
    },
    {
      "epoch": 0.03669526738977929,
      "grad_norm": 1.2058304980077048,
      "learning_rate": 9.970130954251648e-06,
      "loss": 0.6108,
      "step": 680
    },
    {
      "epoch": 0.03674923101829367,
      "grad_norm": 1.3117141478546073,
      "learning_rate": 9.970043137193908e-06,
      "loss": 0.6338,
      "step": 681
    },
    {
      "epoch": 0.03680319464680805,
      "grad_norm": 0.9791418578661547,
      "learning_rate": 9.969955191662612e-06,
      "loss": 0.427,
      "step": 682
    },
    {
      "epoch": 0.03685715827532243,
      "grad_norm": 1.069373859691777,
      "learning_rate": 9.969867117660287e-06,
      "loss": 0.4667,
      "step": 683
    },
    {
      "epoch": 0.036911121903836816,
      "grad_norm": 1.0868409328214808,
      "learning_rate": 9.969778915189462e-06,
      "loss": 0.5161,
      "step": 684
    },
    {
      "epoch": 0.03696508553235119,
      "grad_norm": 1.3856610061553871,
      "learning_rate": 9.969690584252678e-06,
      "loss": 0.6275,
      "step": 685
    },
    {
      "epoch": 0.037019049160865576,
      "grad_norm": 1.0561004318363176,
      "learning_rate": 9.969602124852469e-06,
      "loss": 0.503,
      "step": 686
    },
    {
      "epoch": 0.03707301278937996,
      "grad_norm": 1.0766005898443531,
      "learning_rate": 9.96951353699138e-06,
      "loss": 0.4514,
      "step": 687
    },
    {
      "epoch": 0.037126976417894336,
      "grad_norm": 1.1592447461780662,
      "learning_rate": 9.969424820671953e-06,
      "loss": 0.7777,
      "step": 688
    },
    {
      "epoch": 0.03718094004640872,
      "grad_norm": 1.4019311171946947,
      "learning_rate": 9.969335975896742e-06,
      "loss": 0.6162,
      "step": 689
    },
    {
      "epoch": 0.0372349036749231,
      "grad_norm": 1.2303440577817262,
      "learning_rate": 9.969247002668299e-06,
      "loss": 0.5196,
      "step": 690
    },
    {
      "epoch": 0.03728886730343748,
      "grad_norm": 1.154785159243188,
      "learning_rate": 9.969157900989182e-06,
      "loss": 0.5386,
      "step": 691
    },
    {
      "epoch": 0.03734283093195186,
      "grad_norm": 1.0959239657347508,
      "learning_rate": 9.96906867086195e-06,
      "loss": 0.4789,
      "step": 692
    },
    {
      "epoch": 0.03739679456046625,
      "grad_norm": 1.1183022099067816,
      "learning_rate": 9.96897931228917e-06,
      "loss": 0.5246,
      "step": 693
    },
    {
      "epoch": 0.03745075818898063,
      "grad_norm": 1.357914733647718,
      "learning_rate": 9.96888982527341e-06,
      "loss": 0.6085,
      "step": 694
    },
    {
      "epoch": 0.03750472181749501,
      "grad_norm": 1.3160612315164668,
      "learning_rate": 9.968800209817237e-06,
      "loss": 0.6066,
      "step": 695
    },
    {
      "epoch": 0.03755868544600939,
      "grad_norm": 1.0354159921439854,
      "learning_rate": 9.968710465923233e-06,
      "loss": 0.5688,
      "step": 696
    },
    {
      "epoch": 0.037612649074523774,
      "grad_norm": 1.0373869100624307,
      "learning_rate": 9.968620593593975e-06,
      "loss": 0.7545,
      "step": 697
    },
    {
      "epoch": 0.03766661270303815,
      "grad_norm": 1.2213722833832241,
      "learning_rate": 9.968530592832046e-06,
      "loss": 0.604,
      "step": 698
    },
    {
      "epoch": 0.037720576331552534,
      "grad_norm": 1.3194654548763127,
      "learning_rate": 9.968440463640032e-06,
      "loss": 0.5765,
      "step": 699
    },
    {
      "epoch": 0.03777453996006692,
      "grad_norm": 1.0516278070602276,
      "learning_rate": 9.968350206020526e-06,
      "loss": 0.4966,
      "step": 700
    },
    {
      "epoch": 0.037828503588581294,
      "grad_norm": 1.1148408206527374,
      "learning_rate": 9.968259819976118e-06,
      "loss": 0.4844,
      "step": 701
    },
    {
      "epoch": 0.03788246721709568,
      "grad_norm": 1.0103582606632104,
      "learning_rate": 9.96816930550941e-06,
      "loss": 0.5215,
      "step": 702
    },
    {
      "epoch": 0.03793643084561006,
      "grad_norm": 1.0529674441321135,
      "learning_rate": 9.968078662622999e-06,
      "loss": 0.5016,
      "step": 703
    },
    {
      "epoch": 0.03799039447412444,
      "grad_norm": 1.157328780634197,
      "learning_rate": 9.967987891319494e-06,
      "loss": 0.6078,
      "step": 704
    },
    {
      "epoch": 0.03804435810263882,
      "grad_norm": 1.5097333201657452,
      "learning_rate": 9.9678969916015e-06,
      "loss": 0.659,
      "step": 705
    },
    {
      "epoch": 0.038098321731153205,
      "grad_norm": 1.1056360268423069,
      "learning_rate": 9.967805963471634e-06,
      "loss": 0.5458,
      "step": 706
    },
    {
      "epoch": 0.03815228535966758,
      "grad_norm": 1.0179121258300226,
      "learning_rate": 9.967714806932511e-06,
      "loss": 0.4811,
      "step": 707
    },
    {
      "epoch": 0.038206248988181965,
      "grad_norm": 1.0966970010341872,
      "learning_rate": 9.96762352198675e-06,
      "loss": 0.5329,
      "step": 708
    },
    {
      "epoch": 0.03826021261669635,
      "grad_norm": 0.9904963679221083,
      "learning_rate": 9.967532108636973e-06,
      "loss": 0.5471,
      "step": 709
    },
    {
      "epoch": 0.038314176245210725,
      "grad_norm": 0.9967960124460453,
      "learning_rate": 9.96744056688581e-06,
      "loss": 0.4234,
      "step": 710
    },
    {
      "epoch": 0.03836813987372511,
      "grad_norm": 1.0362773978965538,
      "learning_rate": 9.967348896735892e-06,
      "loss": 0.4708,
      "step": 711
    },
    {
      "epoch": 0.03842210350223949,
      "grad_norm": 0.982630399151348,
      "learning_rate": 9.967257098189853e-06,
      "loss": 0.5004,
      "step": 712
    },
    {
      "epoch": 0.03847606713075387,
      "grad_norm": 1.0962644362782163,
      "learning_rate": 9.967165171250329e-06,
      "loss": 0.5692,
      "step": 713
    },
    {
      "epoch": 0.03853003075926825,
      "grad_norm": 0.8983170214180445,
      "learning_rate": 9.967073115919965e-06,
      "loss": 0.4079,
      "step": 714
    },
    {
      "epoch": 0.038583994387782636,
      "grad_norm": 1.0356706644538067,
      "learning_rate": 9.966980932201407e-06,
      "loss": 0.5331,
      "step": 715
    },
    {
      "epoch": 0.03863795801629701,
      "grad_norm": 1.1588322063890113,
      "learning_rate": 9.966888620097303e-06,
      "loss": 0.5561,
      "step": 716
    },
    {
      "epoch": 0.038691921644811396,
      "grad_norm": 1.1961146208660638,
      "learning_rate": 9.966796179610306e-06,
      "loss": 0.5388,
      "step": 717
    },
    {
      "epoch": 0.03874588527332578,
      "grad_norm": 1.1051479235994306,
      "learning_rate": 9.966703610743073e-06,
      "loss": 0.4948,
      "step": 718
    },
    {
      "epoch": 0.03879984890184016,
      "grad_norm": 0.7661323142531153,
      "learning_rate": 9.966610913498267e-06,
      "loss": 0.3183,
      "step": 719
    },
    {
      "epoch": 0.03885381253035454,
      "grad_norm": 1.083168886419266,
      "learning_rate": 9.96651808787855e-06,
      "loss": 0.4921,
      "step": 720
    },
    {
      "epoch": 0.03890777615886892,
      "grad_norm": 1.298634486799853,
      "learning_rate": 9.966425133886588e-06,
      "loss": 0.6332,
      "step": 721
    },
    {
      "epoch": 0.03896173978738331,
      "grad_norm": 1.2791251606448744,
      "learning_rate": 9.966332051525055e-06,
      "loss": 0.5252,
      "step": 722
    },
    {
      "epoch": 0.03901570341589768,
      "grad_norm": 1.1249022204155936,
      "learning_rate": 9.966238840796628e-06,
      "loss": 0.5595,
      "step": 723
    },
    {
      "epoch": 0.03906966704441207,
      "grad_norm": 0.8190816465698044,
      "learning_rate": 9.966145501703982e-06,
      "loss": 0.4424,
      "step": 724
    },
    {
      "epoch": 0.03912363067292645,
      "grad_norm": 1.1047061045742799,
      "learning_rate": 9.966052034249802e-06,
      "loss": 0.4675,
      "step": 725
    },
    {
      "epoch": 0.03917759430144083,
      "grad_norm": 1.0200768627096775,
      "learning_rate": 9.965958438436775e-06,
      "loss": 0.389,
      "step": 726
    },
    {
      "epoch": 0.03923155792995521,
      "grad_norm": 1.161695423127608,
      "learning_rate": 9.965864714267587e-06,
      "loss": 0.5093,
      "step": 727
    },
    {
      "epoch": 0.039285521558469594,
      "grad_norm": 1.0126559050060262,
      "learning_rate": 9.965770861744937e-06,
      "loss": 0.5076,
      "step": 728
    },
    {
      "epoch": 0.03933948518698397,
      "grad_norm": 1.168264868095318,
      "learning_rate": 9.965676880871522e-06,
      "loss": 0.5194,
      "step": 729
    },
    {
      "epoch": 0.039393448815498354,
      "grad_norm": 1.0322137644895344,
      "learning_rate": 9.965582771650038e-06,
      "loss": 0.5804,
      "step": 730
    },
    {
      "epoch": 0.03944741244401274,
      "grad_norm": 0.9355836663779755,
      "learning_rate": 9.965488534083195e-06,
      "loss": 0.469,
      "step": 731
    },
    {
      "epoch": 0.039501376072527114,
      "grad_norm": 1.0586181045442185,
      "learning_rate": 9.965394168173697e-06,
      "loss": 0.5508,
      "step": 732
    },
    {
      "epoch": 0.0395553397010415,
      "grad_norm": 1.0999258315100529,
      "learning_rate": 9.965299673924261e-06,
      "loss": 0.5656,
      "step": 733
    },
    {
      "epoch": 0.03960930332955588,
      "grad_norm": 1.0945696751619651,
      "learning_rate": 9.965205051337599e-06,
      "loss": 0.4721,
      "step": 734
    },
    {
      "epoch": 0.03966326695807026,
      "grad_norm": 1.2661034122021886,
      "learning_rate": 9.965110300416435e-06,
      "loss": 0.5925,
      "step": 735
    },
    {
      "epoch": 0.03971723058658464,
      "grad_norm": 1.1482636546647562,
      "learning_rate": 9.965015421163485e-06,
      "loss": 0.5825,
      "step": 736
    },
    {
      "epoch": 0.039771194215099025,
      "grad_norm": 1.1320438218871105,
      "learning_rate": 9.964920413581483e-06,
      "loss": 0.5023,
      "step": 737
    },
    {
      "epoch": 0.0398251578436134,
      "grad_norm": 0.8789083858187371,
      "learning_rate": 9.964825277673157e-06,
      "loss": 0.3935,
      "step": 738
    },
    {
      "epoch": 0.039879121472127785,
      "grad_norm": 1.0207605324573932,
      "learning_rate": 9.964730013441239e-06,
      "loss": 0.503,
      "step": 739
    },
    {
      "epoch": 0.03993308510064217,
      "grad_norm": 1.0698855451377445,
      "learning_rate": 9.96463462088847e-06,
      "loss": 0.4421,
      "step": 740
    },
    {
      "epoch": 0.03998704872915655,
      "grad_norm": 1.1635708748769076,
      "learning_rate": 9.964539100017592e-06,
      "loss": 0.4286,
      "step": 741
    },
    {
      "epoch": 0.04004101235767093,
      "grad_norm": 0.9524480092919881,
      "learning_rate": 9.964443450831348e-06,
      "loss": 0.4072,
      "step": 742
    },
    {
      "epoch": 0.04009497598618531,
      "grad_norm": 1.2787711364777035,
      "learning_rate": 9.964347673332488e-06,
      "loss": 0.6577,
      "step": 743
    },
    {
      "epoch": 0.040148939614699695,
      "grad_norm": 1.096966389940386,
      "learning_rate": 9.964251767523767e-06,
      "loss": 0.4192,
      "step": 744
    },
    {
      "epoch": 0.04020290324321407,
      "grad_norm": 1.1309243953832508,
      "learning_rate": 9.964155733407938e-06,
      "loss": 0.5705,
      "step": 745
    },
    {
      "epoch": 0.040256866871728456,
      "grad_norm": 1.2166427205383552,
      "learning_rate": 9.964059570987762e-06,
      "loss": 0.6032,
      "step": 746
    },
    {
      "epoch": 0.04031083050024284,
      "grad_norm": 1.0858889845400743,
      "learning_rate": 9.963963280266004e-06,
      "loss": 0.4877,
      "step": 747
    },
    {
      "epoch": 0.040364794128757216,
      "grad_norm": 1.275153474652414,
      "learning_rate": 9.96386686124543e-06,
      "loss": 0.5672,
      "step": 748
    },
    {
      "epoch": 0.0404187577572716,
      "grad_norm": 1.1052013607832605,
      "learning_rate": 9.963770313928814e-06,
      "loss": 0.4938,
      "step": 749
    },
    {
      "epoch": 0.04047272138578598,
      "grad_norm": 1.2679158762662448,
      "learning_rate": 9.963673638318927e-06,
      "loss": 0.5668,
      "step": 750
    },
    {
      "epoch": 0.04052668501430036,
      "grad_norm": 1.013042529904683,
      "learning_rate": 9.96357683441855e-06,
      "loss": 0.5224,
      "step": 751
    },
    {
      "epoch": 0.04058064864281474,
      "grad_norm": 1.208600292875119,
      "learning_rate": 9.963479902230466e-06,
      "loss": 0.5954,
      "step": 752
    },
    {
      "epoch": 0.040634612271329126,
      "grad_norm": 1.3402670358806734,
      "learning_rate": 9.963382841757456e-06,
      "loss": 0.6935,
      "step": 753
    },
    {
      "epoch": 0.0406885758998435,
      "grad_norm": 1.069309414095788,
      "learning_rate": 9.963285653002316e-06,
      "loss": 0.5321,
      "step": 754
    },
    {
      "epoch": 0.040742539528357886,
      "grad_norm": 1.1890978890324924,
      "learning_rate": 9.963188335967835e-06,
      "loss": 0.6441,
      "step": 755
    },
    {
      "epoch": 0.04079650315687227,
      "grad_norm": 1.1666412611464825,
      "learning_rate": 9.963090890656813e-06,
      "loss": 0.4805,
      "step": 756
    },
    {
      "epoch": 0.04085046678538665,
      "grad_norm": 1.126455463257009,
      "learning_rate": 9.962993317072049e-06,
      "loss": 0.6256,
      "step": 757
    },
    {
      "epoch": 0.04090443041390103,
      "grad_norm": 1.0874174072404954,
      "learning_rate": 9.962895615216347e-06,
      "loss": 0.6864,
      "step": 758
    },
    {
      "epoch": 0.040958394042415414,
      "grad_norm": 0.8814123576436613,
      "learning_rate": 9.962797785092515e-06,
      "loss": 0.4101,
      "step": 759
    },
    {
      "epoch": 0.04101235767092979,
      "grad_norm": 1.1746721637441901,
      "learning_rate": 9.962699826703366e-06,
      "loss": 0.6292,
      "step": 760
    },
    {
      "epoch": 0.041066321299444174,
      "grad_norm": 1.1348673779038585,
      "learning_rate": 9.962601740051717e-06,
      "loss": 0.4951,
      "step": 761
    },
    {
      "epoch": 0.04112028492795856,
      "grad_norm": 1.1290437096486725,
      "learning_rate": 9.962503525140381e-06,
      "loss": 0.4875,
      "step": 762
    },
    {
      "epoch": 0.041174248556472934,
      "grad_norm": 1.3367491564369545,
      "learning_rate": 9.962405181972188e-06,
      "loss": 0.5412,
      "step": 763
    },
    {
      "epoch": 0.04122821218498732,
      "grad_norm": 0.9932825015552385,
      "learning_rate": 9.962306710549958e-06,
      "loss": 0.4933,
      "step": 764
    },
    {
      "epoch": 0.0412821758135017,
      "grad_norm": 1.0256004888977786,
      "learning_rate": 9.962208110876528e-06,
      "loss": 0.6563,
      "step": 765
    },
    {
      "epoch": 0.041336139442016084,
      "grad_norm": 0.9104868137756835,
      "learning_rate": 9.962109382954724e-06,
      "loss": 0.3745,
      "step": 766
    },
    {
      "epoch": 0.04139010307053046,
      "grad_norm": 1.1415664441224562,
      "learning_rate": 9.962010526787389e-06,
      "loss": 0.5475,
      "step": 767
    },
    {
      "epoch": 0.041444066699044844,
      "grad_norm": 0.9550361610308175,
      "learning_rate": 9.961911542377365e-06,
      "loss": 0.3728,
      "step": 768
    },
    {
      "epoch": 0.04149803032755923,
      "grad_norm": 1.0722487816760142,
      "learning_rate": 9.961812429727493e-06,
      "loss": 0.6561,
      "step": 769
    },
    {
      "epoch": 0.041551993956073605,
      "grad_norm": 1.0537783222149142,
      "learning_rate": 9.961713188840624e-06,
      "loss": 0.5233,
      "step": 770
    },
    {
      "epoch": 0.04160595758458799,
      "grad_norm": 0.9343983623875682,
      "learning_rate": 9.96161381971961e-06,
      "loss": 0.3844,
      "step": 771
    },
    {
      "epoch": 0.04165992121310237,
      "grad_norm": 1.1496691780489372,
      "learning_rate": 9.961514322367305e-06,
      "loss": 0.6206,
      "step": 772
    },
    {
      "epoch": 0.04171388484161675,
      "grad_norm": 1.1983814688496892,
      "learning_rate": 9.961414696786573e-06,
      "loss": 0.5378,
      "step": 773
    },
    {
      "epoch": 0.04176784847013113,
      "grad_norm": 1.1836631645660005,
      "learning_rate": 9.961314942980273e-06,
      "loss": 0.5352,
      "step": 774
    },
    {
      "epoch": 0.041821812098645515,
      "grad_norm": 1.1869502593578356,
      "learning_rate": 9.961215060951274e-06,
      "loss": 0.5422,
      "step": 775
    },
    {
      "epoch": 0.04187577572715989,
      "grad_norm": 1.2992153032934948,
      "learning_rate": 9.961115050702445e-06,
      "loss": 0.5854,
      "step": 776
    },
    {
      "epoch": 0.041929739355674275,
      "grad_norm": 1.2240265961873171,
      "learning_rate": 9.961014912236664e-06,
      "loss": 0.5855,
      "step": 777
    },
    {
      "epoch": 0.04198370298418866,
      "grad_norm": 1.1310116533800991,
      "learning_rate": 9.960914645556806e-06,
      "loss": 0.5042,
      "step": 778
    },
    {
      "epoch": 0.042037666612703035,
      "grad_norm": 1.305660290961825,
      "learning_rate": 9.960814250665755e-06,
      "loss": 0.543,
      "step": 779
    },
    {
      "epoch": 0.04209163024121742,
      "grad_norm": 0.9109453758486583,
      "learning_rate": 9.960713727566392e-06,
      "loss": 0.4382,
      "step": 780
    },
    {
      "epoch": 0.0421455938697318,
      "grad_norm": 0.9582772793026495,
      "learning_rate": 9.960613076261612e-06,
      "loss": 0.4755,
      "step": 781
    },
    {
      "epoch": 0.04219955749824618,
      "grad_norm": 1.3905042558023346,
      "learning_rate": 9.960512296754305e-06,
      "loss": 0.6492,
      "step": 782
    },
    {
      "epoch": 0.04225352112676056,
      "grad_norm": 1.4237769289679167,
      "learning_rate": 9.960411389047366e-06,
      "loss": 0.8267,
      "step": 783
    },
    {
      "epoch": 0.042307484755274946,
      "grad_norm": 0.9865989322565573,
      "learning_rate": 9.960310353143697e-06,
      "loss": 0.4543,
      "step": 784
    },
    {
      "epoch": 0.04236144838378932,
      "grad_norm": 1.3675667607957243,
      "learning_rate": 9.960209189046203e-06,
      "loss": 0.504,
      "step": 785
    },
    {
      "epoch": 0.042415412012303706,
      "grad_norm": 1.3955889883374388,
      "learning_rate": 9.96010789675779e-06,
      "loss": 0.6794,
      "step": 786
    },
    {
      "epoch": 0.04246937564081809,
      "grad_norm": 1.3628001778666408,
      "learning_rate": 9.960006476281369e-06,
      "loss": 0.7429,
      "step": 787
    },
    {
      "epoch": 0.04252333926933247,
      "grad_norm": 0.9853353125060482,
      "learning_rate": 9.959904927619855e-06,
      "loss": 0.476,
      "step": 788
    },
    {
      "epoch": 0.04257730289784685,
      "grad_norm": 1.1642060412638036,
      "learning_rate": 9.959803250776168e-06,
      "loss": 0.5423,
      "step": 789
    },
    {
      "epoch": 0.04263126652636123,
      "grad_norm": 0.9657027265109951,
      "learning_rate": 9.959701445753228e-06,
      "loss": 0.4265,
      "step": 790
    },
    {
      "epoch": 0.04268523015487562,
      "grad_norm": 1.283198805129021,
      "learning_rate": 9.959599512553962e-06,
      "loss": 0.6696,
      "step": 791
    },
    {
      "epoch": 0.042739193783389993,
      "grad_norm": 1.141233984693726,
      "learning_rate": 9.959497451181302e-06,
      "loss": 0.6223,
      "step": 792
    },
    {
      "epoch": 0.04279315741190438,
      "grad_norm": 1.1750742000585968,
      "learning_rate": 9.959395261638175e-06,
      "loss": 0.573,
      "step": 793
    },
    {
      "epoch": 0.04284712104041876,
      "grad_norm": 1.1161770824778332,
      "learning_rate": 9.959292943927527e-06,
      "loss": 0.5537,
      "step": 794
    },
    {
      "epoch": 0.04290108466893314,
      "grad_norm": 1.0599726030986878,
      "learning_rate": 9.959190498052292e-06,
      "loss": 0.5516,
      "step": 795
    },
    {
      "epoch": 0.04295504829744752,
      "grad_norm": 1.0998115533102952,
      "learning_rate": 9.959087924015417e-06,
      "loss": 0.4421,
      "step": 796
    },
    {
      "epoch": 0.043009011925961904,
      "grad_norm": 0.8207845985207481,
      "learning_rate": 9.958985221819848e-06,
      "loss": 0.3529,
      "step": 797
    },
    {
      "epoch": 0.04306297555447628,
      "grad_norm": 1.170251041300329,
      "learning_rate": 9.958882391468541e-06,
      "loss": 0.5737,
      "step": 798
    },
    {
      "epoch": 0.043116939182990664,
      "grad_norm": 1.3622069799585852,
      "learning_rate": 9.958779432964446e-06,
      "loss": 0.6869,
      "step": 799
    },
    {
      "epoch": 0.04317090281150505,
      "grad_norm": 1.0832448392828693,
      "learning_rate": 9.958676346310526e-06,
      "loss": 0.4923,
      "step": 800
    },
    {
      "epoch": 0.043224866440019424,
      "grad_norm": 0.9694841309202773,
      "learning_rate": 9.958573131509742e-06,
      "loss": 0.4431,
      "step": 801
    },
    {
      "epoch": 0.04327883006853381,
      "grad_norm": 1.1173316212145825,
      "learning_rate": 9.958469788565061e-06,
      "loss": 0.5131,
      "step": 802
    },
    {
      "epoch": 0.04333279369704819,
      "grad_norm": 1.2816631455841192,
      "learning_rate": 9.958366317479456e-06,
      "loss": 0.6544,
      "step": 803
    },
    {
      "epoch": 0.04338675732556257,
      "grad_norm": 1.1648494294915082,
      "learning_rate": 9.958262718255896e-06,
      "loss": 0.6186,
      "step": 804
    },
    {
      "epoch": 0.04344072095407695,
      "grad_norm": 1.1711850844075506,
      "learning_rate": 9.958158990897359e-06,
      "loss": 0.4915,
      "step": 805
    },
    {
      "epoch": 0.043494684582591335,
      "grad_norm": 1.0606035622789363,
      "learning_rate": 9.95805513540683e-06,
      "loss": 0.4366,
      "step": 806
    },
    {
      "epoch": 0.04354864821110571,
      "grad_norm": 1.037981123602516,
      "learning_rate": 9.957951151787292e-06,
      "loss": 0.4313,
      "step": 807
    },
    {
      "epoch": 0.043602611839620095,
      "grad_norm": 1.3303904934753212,
      "learning_rate": 9.957847040041734e-06,
      "loss": 0.6739,
      "step": 808
    },
    {
      "epoch": 0.04365657546813448,
      "grad_norm": 1.0767940870560473,
      "learning_rate": 9.957742800173146e-06,
      "loss": 0.4243,
      "step": 809
    },
    {
      "epoch": 0.043710539096648855,
      "grad_norm": 1.1296467372940135,
      "learning_rate": 9.957638432184528e-06,
      "loss": 0.5154,
      "step": 810
    },
    {
      "epoch": 0.04376450272516324,
      "grad_norm": 0.969125918397227,
      "learning_rate": 9.957533936078874e-06,
      "loss": 0.4604,
      "step": 811
    },
    {
      "epoch": 0.04381846635367762,
      "grad_norm": 1.3247012212267448,
      "learning_rate": 9.957429311859193e-06,
      "loss": 0.5916,
      "step": 812
    },
    {
      "epoch": 0.043872429982192006,
      "grad_norm": 1.2946218747124218,
      "learning_rate": 9.95732455952849e-06,
      "loss": 0.7409,
      "step": 813
    },
    {
      "epoch": 0.04392639361070638,
      "grad_norm": 1.0921285610742266,
      "learning_rate": 9.957219679089774e-06,
      "loss": 0.6463,
      "step": 814
    },
    {
      "epoch": 0.043980357239220766,
      "grad_norm": 1.0250104908913187,
      "learning_rate": 9.957114670546062e-06,
      "loss": 0.4082,
      "step": 815
    },
    {
      "epoch": 0.04403432086773515,
      "grad_norm": 1.24791909525768,
      "learning_rate": 9.95700953390037e-06,
      "loss": 0.5775,
      "step": 816
    },
    {
      "epoch": 0.044088284496249526,
      "grad_norm": 1.1199302467618155,
      "learning_rate": 9.956904269155722e-06,
      "loss": 0.6594,
      "step": 817
    },
    {
      "epoch": 0.04414224812476391,
      "grad_norm": 1.0729047403541216,
      "learning_rate": 9.95679887631514e-06,
      "loss": 0.5091,
      "step": 818
    },
    {
      "epoch": 0.04419621175327829,
      "grad_norm": 0.9236947214435037,
      "learning_rate": 9.956693355381657e-06,
      "loss": 0.4626,
      "step": 819
    },
    {
      "epoch": 0.04425017538179267,
      "grad_norm": 1.1600202430517659,
      "learning_rate": 9.9565877063583e-06,
      "loss": 0.5589,
      "step": 820
    },
    {
      "epoch": 0.04430413901030705,
      "grad_norm": 1.23908090319664,
      "learning_rate": 9.956481929248113e-06,
      "loss": 0.6001,
      "step": 821
    },
    {
      "epoch": 0.04435810263882144,
      "grad_norm": 1.2081352033935002,
      "learning_rate": 9.95637602405413e-06,
      "loss": 0.5244,
      "step": 822
    },
    {
      "epoch": 0.04441206626733581,
      "grad_norm": 1.1158389757429827,
      "learning_rate": 9.956269990779399e-06,
      "loss": 0.6365,
      "step": 823
    },
    {
      "epoch": 0.0444660298958502,
      "grad_norm": 1.0228014766103806,
      "learning_rate": 9.956163829426964e-06,
      "loss": 0.5234,
      "step": 824
    },
    {
      "epoch": 0.04451999352436458,
      "grad_norm": 0.9379362357714647,
      "learning_rate": 9.956057539999879e-06,
      "loss": 0.5101,
      "step": 825
    },
    {
      "epoch": 0.04457395715287896,
      "grad_norm": 1.0419729226602952,
      "learning_rate": 9.955951122501198e-06,
      "loss": 0.5367,
      "step": 826
    },
    {
      "epoch": 0.04462792078139334,
      "grad_norm": 0.9449896021359578,
      "learning_rate": 9.955844576933979e-06,
      "loss": 0.4264,
      "step": 827
    },
    {
      "epoch": 0.044681884409907724,
      "grad_norm": 1.2890742429392263,
      "learning_rate": 9.955737903301285e-06,
      "loss": 0.5736,
      "step": 828
    },
    {
      "epoch": 0.0447358480384221,
      "grad_norm": 1.2929235394842682,
      "learning_rate": 9.955631101606181e-06,
      "loss": 0.5301,
      "step": 829
    },
    {
      "epoch": 0.044789811666936484,
      "grad_norm": 1.2910754231569337,
      "learning_rate": 9.955524171851737e-06,
      "loss": 0.5764,
      "step": 830
    },
    {
      "epoch": 0.04484377529545087,
      "grad_norm": 1.1522987846466173,
      "learning_rate": 9.955417114041027e-06,
      "loss": 0.5704,
      "step": 831
    },
    {
      "epoch": 0.044897738923965244,
      "grad_norm": 1.1825395933693907,
      "learning_rate": 9.955309928177125e-06,
      "loss": 0.6493,
      "step": 832
    },
    {
      "epoch": 0.04495170255247963,
      "grad_norm": 1.0917887930299823,
      "learning_rate": 9.955202614263118e-06,
      "loss": 0.5403,
      "step": 833
    },
    {
      "epoch": 0.04500566618099401,
      "grad_norm": 1.1539025758259183,
      "learning_rate": 9.955095172302082e-06,
      "loss": 0.5588,
      "step": 834
    },
    {
      "epoch": 0.045059629809508395,
      "grad_norm": 0.9618472043347563,
      "learning_rate": 9.954987602297112e-06,
      "loss": 0.39,
      "step": 835
    },
    {
      "epoch": 0.04511359343802277,
      "grad_norm": 1.0470025305708133,
      "learning_rate": 9.954879904251296e-06,
      "loss": 0.6432,
      "step": 836
    },
    {
      "epoch": 0.045167557066537155,
      "grad_norm": 1.0480055933972139,
      "learning_rate": 9.95477207816773e-06,
      "loss": 0.5512,
      "step": 837
    },
    {
      "epoch": 0.04522152069505154,
      "grad_norm": 0.999995877405877,
      "learning_rate": 9.954664124049513e-06,
      "loss": 0.4842,
      "step": 838
    },
    {
      "epoch": 0.045275484323565915,
      "grad_norm": 0.942848391250373,
      "learning_rate": 9.95455604189975e-06,
      "loss": 0.3936,
      "step": 839
    },
    {
      "epoch": 0.0453294479520803,
      "grad_norm": 1.2739432987117487,
      "learning_rate": 9.954447831721544e-06,
      "loss": 0.6475,
      "step": 840
    },
    {
      "epoch": 0.04538341158059468,
      "grad_norm": 0.9509442896722112,
      "learning_rate": 9.954339493518004e-06,
      "loss": 0.398,
      "step": 841
    },
    {
      "epoch": 0.04543737520910906,
      "grad_norm": 0.9667800908523281,
      "learning_rate": 9.95423102729225e-06,
      "loss": 0.4467,
      "step": 842
    },
    {
      "epoch": 0.04549133883762344,
      "grad_norm": 1.1650107026895582,
      "learning_rate": 9.954122433047394e-06,
      "loss": 0.6361,
      "step": 843
    },
    {
      "epoch": 0.045545302466137826,
      "grad_norm": 0.8457595407140776,
      "learning_rate": 9.954013710786558e-06,
      "loss": 0.4393,
      "step": 844
    },
    {
      "epoch": 0.0455992660946522,
      "grad_norm": 1.0185578443042687,
      "learning_rate": 9.953904860512869e-06,
      "loss": 0.5241,
      "step": 845
    },
    {
      "epoch": 0.045653229723166586,
      "grad_norm": 1.011919273653372,
      "learning_rate": 9.953795882229454e-06,
      "loss": 0.553,
      "step": 846
    },
    {
      "epoch": 0.04570719335168097,
      "grad_norm": 0.9421095555057688,
      "learning_rate": 9.953686775939443e-06,
      "loss": 0.4638,
      "step": 847
    },
    {
      "epoch": 0.045761156980195346,
      "grad_norm": 1.0513831834952192,
      "learning_rate": 9.953577541645975e-06,
      "loss": 0.4686,
      "step": 848
    },
    {
      "epoch": 0.04581512060870973,
      "grad_norm": 1.3461692569628683,
      "learning_rate": 9.953468179352189e-06,
      "loss": 0.6178,
      "step": 849
    },
    {
      "epoch": 0.04586908423722411,
      "grad_norm": 1.2294571450271108,
      "learning_rate": 9.953358689061228e-06,
      "loss": 0.6248,
      "step": 850
    },
    {
      "epoch": 0.04592304786573849,
      "grad_norm": 1.207253596824993,
      "learning_rate": 9.953249070776236e-06,
      "loss": 0.613,
      "step": 851
    },
    {
      "epoch": 0.04597701149425287,
      "grad_norm": 1.0742117125075181,
      "learning_rate": 9.95313932450037e-06,
      "loss": 0.5171,
      "step": 852
    },
    {
      "epoch": 0.046030975122767256,
      "grad_norm": 1.024532321451586,
      "learning_rate": 9.953029450236777e-06,
      "loss": 0.6241,
      "step": 853
    },
    {
      "epoch": 0.04608493875128163,
      "grad_norm": 1.3082001301235946,
      "learning_rate": 9.952919447988619e-06,
      "loss": 0.5491,
      "step": 854
    },
    {
      "epoch": 0.04613890237979602,
      "grad_norm": 1.0039499261884828,
      "learning_rate": 9.952809317759057e-06,
      "loss": 0.4348,
      "step": 855
    },
    {
      "epoch": 0.0461928660083104,
      "grad_norm": 1.0606995065463638,
      "learning_rate": 9.952699059551255e-06,
      "loss": 0.4873,
      "step": 856
    },
    {
      "epoch": 0.04624682963682478,
      "grad_norm": 1.1013389386979189,
      "learning_rate": 9.952588673368385e-06,
      "loss": 0.5586,
      "step": 857
    },
    {
      "epoch": 0.04630079326533916,
      "grad_norm": 1.1326071564789608,
      "learning_rate": 9.952478159213615e-06,
      "loss": 0.558,
      "step": 858
    },
    {
      "epoch": 0.046354756893853544,
      "grad_norm": 1.1829065474044214,
      "learning_rate": 9.952367517090126e-06,
      "loss": 0.5743,
      "step": 859
    },
    {
      "epoch": 0.04640872052236793,
      "grad_norm": 1.335348778037312,
      "learning_rate": 9.952256747001095e-06,
      "loss": 0.7002,
      "step": 860
    },
    {
      "epoch": 0.046462684150882304,
      "grad_norm": 1.0657595400051647,
      "learning_rate": 9.952145848949705e-06,
      "loss": 0.4487,
      "step": 861
    },
    {
      "epoch": 0.04651664777939669,
      "grad_norm": 0.9456656967812963,
      "learning_rate": 9.952034822939146e-06,
      "loss": 0.4797,
      "step": 862
    },
    {
      "epoch": 0.04657061140791107,
      "grad_norm": 1.1449853335406464,
      "learning_rate": 9.95192366897261e-06,
      "loss": 0.5472,
      "step": 863
    },
    {
      "epoch": 0.04662457503642545,
      "grad_norm": 0.9749603834609615,
      "learning_rate": 9.951812387053285e-06,
      "loss": 0.4366,
      "step": 864
    },
    {
      "epoch": 0.04667853866493983,
      "grad_norm": 1.2354880999088815,
      "learning_rate": 9.951700977184376e-06,
      "loss": 0.5315,
      "step": 865
    },
    {
      "epoch": 0.046732502293454214,
      "grad_norm": 1.1462689644973072,
      "learning_rate": 9.951589439369085e-06,
      "loss": 0.5435,
      "step": 866
    },
    {
      "epoch": 0.04678646592196859,
      "grad_norm": 1.0732309054100402,
      "learning_rate": 9.951477773610612e-06,
      "loss": 0.4782,
      "step": 867
    },
    {
      "epoch": 0.046840429550482975,
      "grad_norm": 1.172253476056852,
      "learning_rate": 9.951365979912171e-06,
      "loss": 0.6239,
      "step": 868
    },
    {
      "epoch": 0.04689439317899736,
      "grad_norm": 1.169417301170583,
      "learning_rate": 9.951254058276973e-06,
      "loss": 0.6685,
      "step": 869
    },
    {
      "epoch": 0.046948356807511735,
      "grad_norm": 1.0336422807383343,
      "learning_rate": 9.951142008708238e-06,
      "loss": 0.5499,
      "step": 870
    },
    {
      "epoch": 0.04700232043602612,
      "grad_norm": 1.0172000053778143,
      "learning_rate": 9.951029831209182e-06,
      "loss": 0.5051,
      "step": 871
    },
    {
      "epoch": 0.0470562840645405,
      "grad_norm": 1.1388258783549963,
      "learning_rate": 9.950917525783032e-06,
      "loss": 0.4595,
      "step": 872
    },
    {
      "epoch": 0.04711024769305488,
      "grad_norm": 1.173881312158791,
      "learning_rate": 9.950805092433016e-06,
      "loss": 0.648,
      "step": 873
    },
    {
      "epoch": 0.04716421132156926,
      "grad_norm": 1.1788809034420238,
      "learning_rate": 9.950692531162363e-06,
      "loss": 0.6048,
      "step": 874
    },
    {
      "epoch": 0.047218174950083645,
      "grad_norm": 1.0858016137565945,
      "learning_rate": 9.95057984197431e-06,
      "loss": 0.5727,
      "step": 875
    },
    {
      "epoch": 0.04727213857859802,
      "grad_norm": 1.1040578102124057,
      "learning_rate": 9.950467024872096e-06,
      "loss": 0.5268,
      "step": 876
    },
    {
      "epoch": 0.047326102207112405,
      "grad_norm": 1.0572361901994995,
      "learning_rate": 9.950354079858962e-06,
      "loss": 0.5294,
      "step": 877
    },
    {
      "epoch": 0.04738006583562679,
      "grad_norm": 1.4501507525450223,
      "learning_rate": 9.950241006938154e-06,
      "loss": 0.533,
      "step": 878
    },
    {
      "epoch": 0.047434029464141166,
      "grad_norm": 1.1775183486804905,
      "learning_rate": 9.950127806112927e-06,
      "loss": 0.6693,
      "step": 879
    },
    {
      "epoch": 0.04748799309265555,
      "grad_norm": 1.033222834721493,
      "learning_rate": 9.950014477386527e-06,
      "loss": 0.5367,
      "step": 880
    },
    {
      "epoch": 0.04754195672116993,
      "grad_norm": 1.226964582999013,
      "learning_rate": 9.949901020762215e-06,
      "loss": 0.6071,
      "step": 881
    },
    {
      "epoch": 0.047595920349684316,
      "grad_norm": 1.2851448583804694,
      "learning_rate": 9.94978743624325e-06,
      "loss": 0.6728,
      "step": 882
    },
    {
      "epoch": 0.04764988397819869,
      "grad_norm": 1.0985996569725616,
      "learning_rate": 9.9496737238329e-06,
      "loss": 0.6255,
      "step": 883
    },
    {
      "epoch": 0.047703847606713076,
      "grad_norm": 1.278570103286649,
      "learning_rate": 9.94955988353443e-06,
      "loss": 0.4899,
      "step": 884
    },
    {
      "epoch": 0.04775781123522746,
      "grad_norm": 1.0622605455885163,
      "learning_rate": 9.949445915351114e-06,
      "loss": 0.5208,
      "step": 885
    },
    {
      "epoch": 0.047811774863741836,
      "grad_norm": 1.14813735808111,
      "learning_rate": 9.949331819286224e-06,
      "loss": 0.6805,
      "step": 886
    },
    {
      "epoch": 0.04786573849225622,
      "grad_norm": 1.0978744242379226,
      "learning_rate": 9.949217595343044e-06,
      "loss": 0.445,
      "step": 887
    },
    {
      "epoch": 0.0479197021207706,
      "grad_norm": 1.3050579388952461,
      "learning_rate": 9.949103243524855e-06,
      "loss": 0.7052,
      "step": 888
    },
    {
      "epoch": 0.04797366574928498,
      "grad_norm": 1.1568123728537694,
      "learning_rate": 9.948988763834943e-06,
      "loss": 0.6378,
      "step": 889
    },
    {
      "epoch": 0.048027629377799363,
      "grad_norm": 1.0555064994304,
      "learning_rate": 9.948874156276599e-06,
      "loss": 0.5383,
      "step": 890
    },
    {
      "epoch": 0.04808159300631375,
      "grad_norm": 1.3732364076772197,
      "learning_rate": 9.948759420853114e-06,
      "loss": 0.4914,
      "step": 891
    },
    {
      "epoch": 0.048135556634828124,
      "grad_norm": 1.3621135636183506,
      "learning_rate": 9.94864455756779e-06,
      "loss": 0.5587,
      "step": 892
    },
    {
      "epoch": 0.04818952026334251,
      "grad_norm": 1.2161033354683788,
      "learning_rate": 9.948529566423926e-06,
      "loss": 0.5355,
      "step": 893
    },
    {
      "epoch": 0.04824348389185689,
      "grad_norm": 1.2347943493873368,
      "learning_rate": 9.948414447424828e-06,
      "loss": 0.6741,
      "step": 894
    },
    {
      "epoch": 0.04829744752037127,
      "grad_norm": 1.0990730119464749,
      "learning_rate": 9.948299200573804e-06,
      "loss": 0.5162,
      "step": 895
    },
    {
      "epoch": 0.04835141114888565,
      "grad_norm": 1.1517453969953158,
      "learning_rate": 9.948183825874165e-06,
      "loss": 0.6907,
      "step": 896
    },
    {
      "epoch": 0.048405374777400034,
      "grad_norm": 1.0916894811709865,
      "learning_rate": 9.94806832332923e-06,
      "loss": 0.6028,
      "step": 897
    },
    {
      "epoch": 0.04845933840591441,
      "grad_norm": 1.131941499588161,
      "learning_rate": 9.947952692942315e-06,
      "loss": 0.4923,
      "step": 898
    },
    {
      "epoch": 0.048513302034428794,
      "grad_norm": 1.0827248735353647,
      "learning_rate": 9.947836934716748e-06,
      "loss": 0.5119,
      "step": 899
    },
    {
      "epoch": 0.04856726566294318,
      "grad_norm": 1.142551533747074,
      "learning_rate": 9.947721048655851e-06,
      "loss": 0.5733,
      "step": 900
    },
    {
      "epoch": 0.048621229291457554,
      "grad_norm": 1.1691018272197893,
      "learning_rate": 9.947605034762957e-06,
      "loss": 0.5001,
      "step": 901
    },
    {
      "epoch": 0.04867519291997194,
      "grad_norm": 1.2452394132365128,
      "learning_rate": 9.9474888930414e-06,
      "loss": 0.5879,
      "step": 902
    },
    {
      "epoch": 0.04872915654848632,
      "grad_norm": 1.121845133387402,
      "learning_rate": 9.947372623494521e-06,
      "loss": 0.5128,
      "step": 903
    },
    {
      "epoch": 0.0487831201770007,
      "grad_norm": 0.8206533386720382,
      "learning_rate": 9.947256226125655e-06,
      "loss": 0.3144,
      "step": 904
    },
    {
      "epoch": 0.04883708380551508,
      "grad_norm": 1.0988307642935975,
      "learning_rate": 9.947139700938155e-06,
      "loss": 0.4608,
      "step": 905
    },
    {
      "epoch": 0.048891047434029465,
      "grad_norm": 1.2311889179252746,
      "learning_rate": 9.947023047935366e-06,
      "loss": 0.5923,
      "step": 906
    },
    {
      "epoch": 0.04894501106254385,
      "grad_norm": 1.1869626886398155,
      "learning_rate": 9.94690626712064e-06,
      "loss": 0.616,
      "step": 907
    },
    {
      "epoch": 0.048998974691058225,
      "grad_norm": 1.1377227457111447,
      "learning_rate": 9.946789358497334e-06,
      "loss": 0.5944,
      "step": 908
    },
    {
      "epoch": 0.04905293831957261,
      "grad_norm": 0.8850516222750494,
      "learning_rate": 9.94667232206881e-06,
      "loss": 0.3872,
      "step": 909
    },
    {
      "epoch": 0.04910690194808699,
      "grad_norm": 1.1855855414506806,
      "learning_rate": 9.946555157838431e-06,
      "loss": 0.592,
      "step": 910
    },
    {
      "epoch": 0.04916086557660137,
      "grad_norm": 1.043740924519021,
      "learning_rate": 9.946437865809563e-06,
      "loss": 0.3715,
      "step": 911
    },
    {
      "epoch": 0.04921482920511575,
      "grad_norm": 1.1200607490411842,
      "learning_rate": 9.946320445985577e-06,
      "loss": 0.609,
      "step": 912
    },
    {
      "epoch": 0.049268792833630136,
      "grad_norm": 1.1435492781265595,
      "learning_rate": 9.94620289836985e-06,
      "loss": 0.6849,
      "step": 913
    },
    {
      "epoch": 0.04932275646214451,
      "grad_norm": 1.1800112011495618,
      "learning_rate": 9.946085222965759e-06,
      "loss": 0.597,
      "step": 914
    },
    {
      "epoch": 0.049376720090658896,
      "grad_norm": 1.2704387357312605,
      "learning_rate": 9.945967419776687e-06,
      "loss": 0.5696,
      "step": 915
    },
    {
      "epoch": 0.04943068371917328,
      "grad_norm": 1.2884289142012513,
      "learning_rate": 9.945849488806017e-06,
      "loss": 0.5663,
      "step": 916
    },
    {
      "epoch": 0.049484647347687656,
      "grad_norm": 1.175951397470401,
      "learning_rate": 9.945731430057142e-06,
      "loss": 0.5097,
      "step": 917
    },
    {
      "epoch": 0.04953861097620204,
      "grad_norm": 1.1128480787952366,
      "learning_rate": 9.945613243533454e-06,
      "loss": 0.5986,
      "step": 918
    },
    {
      "epoch": 0.04959257460471642,
      "grad_norm": 1.0409962664966061,
      "learning_rate": 9.945494929238347e-06,
      "loss": 0.5551,
      "step": 919
    },
    {
      "epoch": 0.0496465382332308,
      "grad_norm": 1.3161197113986414,
      "learning_rate": 9.945376487175225e-06,
      "loss": 0.6711,
      "step": 920
    },
    {
      "epoch": 0.04970050186174518,
      "grad_norm": 1.0603301510247114,
      "learning_rate": 9.945257917347494e-06,
      "loss": 0.498,
      "step": 921
    },
    {
      "epoch": 0.04975446549025957,
      "grad_norm": 1.2689361906824057,
      "learning_rate": 9.945139219758557e-06,
      "loss": 0.7067,
      "step": 922
    },
    {
      "epoch": 0.04980842911877394,
      "grad_norm": 1.1863340445933492,
      "learning_rate": 9.945020394411825e-06,
      "loss": 0.5514,
      "step": 923
    },
    {
      "epoch": 0.04986239274728833,
      "grad_norm": 1.1305097743943155,
      "learning_rate": 9.944901441310718e-06,
      "loss": 0.4715,
      "step": 924
    },
    {
      "epoch": 0.04991635637580271,
      "grad_norm": 1.182687493870884,
      "learning_rate": 9.944782360458651e-06,
      "loss": 0.534,
      "step": 925
    },
    {
      "epoch": 0.04997032000431709,
      "grad_norm": 1.2140043040050221,
      "learning_rate": 9.944663151859047e-06,
      "loss": 0.7507,
      "step": 926
    },
    {
      "epoch": 0.05002428363283147,
      "grad_norm": 0.8890439120387297,
      "learning_rate": 9.944543815515334e-06,
      "loss": 0.3346,
      "step": 927
    },
    {
      "epoch": 0.050078247261345854,
      "grad_norm": 0.9995037436744633,
      "learning_rate": 9.944424351430942e-06,
      "loss": 0.4698,
      "step": 928
    },
    {
      "epoch": 0.05013221088986024,
      "grad_norm": 1.0680249722463102,
      "learning_rate": 9.944304759609303e-06,
      "loss": 0.4948,
      "step": 929
    },
    {
      "epoch": 0.050186174518374614,
      "grad_norm": 1.0249911843876578,
      "learning_rate": 9.944185040053855e-06,
      "loss": 0.4454,
      "step": 930
    },
    {
      "epoch": 0.050240138146889,
      "grad_norm": 1.1014612078908064,
      "learning_rate": 9.944065192768037e-06,
      "loss": 0.4797,
      "step": 931
    },
    {
      "epoch": 0.05029410177540338,
      "grad_norm": 0.9924841059203763,
      "learning_rate": 9.943945217755296e-06,
      "loss": 0.4721,
      "step": 932
    },
    {
      "epoch": 0.05034806540391776,
      "grad_norm": 1.1356030242662922,
      "learning_rate": 9.94382511501908e-06,
      "loss": 0.5683,
      "step": 933
    },
    {
      "epoch": 0.05040202903243214,
      "grad_norm": 1.2249284315308386,
      "learning_rate": 9.943704884562838e-06,
      "loss": 0.6563,
      "step": 934
    },
    {
      "epoch": 0.050455992660946525,
      "grad_norm": 1.0910310936758012,
      "learning_rate": 9.943584526390028e-06,
      "loss": 0.4433,
      "step": 935
    },
    {
      "epoch": 0.0505099562894609,
      "grad_norm": 1.06334702100724,
      "learning_rate": 9.94346404050411e-06,
      "loss": 0.5281,
      "step": 936
    },
    {
      "epoch": 0.050563919917975285,
      "grad_norm": 0.9941522802311998,
      "learning_rate": 9.943343426908544e-06,
      "loss": 0.5305,
      "step": 937
    },
    {
      "epoch": 0.05061788354648967,
      "grad_norm": 1.1969860609642315,
      "learning_rate": 9.943222685606799e-06,
      "loss": 0.4892,
      "step": 938
    },
    {
      "epoch": 0.050671847175004045,
      "grad_norm": 0.9532988061705191,
      "learning_rate": 9.943101816602344e-06,
      "loss": 0.3916,
      "step": 939
    },
    {
      "epoch": 0.05072581080351843,
      "grad_norm": 1.073260374045587,
      "learning_rate": 9.942980819898655e-06,
      "loss": 0.5817,
      "step": 940
    },
    {
      "epoch": 0.05077977443203281,
      "grad_norm": 1.1386205153888997,
      "learning_rate": 9.942859695499206e-06,
      "loss": 0.491,
      "step": 941
    },
    {
      "epoch": 0.05083373806054719,
      "grad_norm": 1.4006842128178485,
      "learning_rate": 9.94273844340748e-06,
      "loss": 0.5655,
      "step": 942
    },
    {
      "epoch": 0.05088770168906157,
      "grad_norm": 1.2238228760547045,
      "learning_rate": 9.942617063626964e-06,
      "loss": 0.5998,
      "step": 943
    },
    {
      "epoch": 0.050941665317575956,
      "grad_norm": 1.3009739056071936,
      "learning_rate": 9.942495556161143e-06,
      "loss": 0.7357,
      "step": 944
    },
    {
      "epoch": 0.05099562894609033,
      "grad_norm": 1.2108375479876154,
      "learning_rate": 9.942373921013512e-06,
      "loss": 0.5261,
      "step": 945
    },
    {
      "epoch": 0.051049592574604716,
      "grad_norm": 1.190659206324907,
      "learning_rate": 9.942252158187567e-06,
      "loss": 0.7053,
      "step": 946
    },
    {
      "epoch": 0.0511035562031191,
      "grad_norm": 0.8601414145935143,
      "learning_rate": 9.942130267686804e-06,
      "loss": 0.3742,
      "step": 947
    },
    {
      "epoch": 0.051157519831633476,
      "grad_norm": 1.1685961538246128,
      "learning_rate": 9.94200824951473e-06,
      "loss": 0.4824,
      "step": 948
    },
    {
      "epoch": 0.05121148346014786,
      "grad_norm": 1.1359198702540192,
      "learning_rate": 9.94188610367485e-06,
      "loss": 0.5011,
      "step": 949
    },
    {
      "epoch": 0.05126544708866224,
      "grad_norm": 1.1763716493388692,
      "learning_rate": 9.941763830170675e-06,
      "loss": 0.4977,
      "step": 950
    },
    {
      "epoch": 0.05131941071717662,
      "grad_norm": 1.2452579498304215,
      "learning_rate": 9.94164142900572e-06,
      "loss": 0.5505,
      "step": 951
    },
    {
      "epoch": 0.051373374345691,
      "grad_norm": 0.9287950381384263,
      "learning_rate": 9.941518900183501e-06,
      "loss": 0.3928,
      "step": 952
    },
    {
      "epoch": 0.05142733797420539,
      "grad_norm": 0.9733887430579281,
      "learning_rate": 9.941396243707544e-06,
      "loss": 0.4204,
      "step": 953
    },
    {
      "epoch": 0.05148130160271977,
      "grad_norm": 1.210708109561921,
      "learning_rate": 9.941273459581368e-06,
      "loss": 0.6957,
      "step": 954
    },
    {
      "epoch": 0.05153526523123415,
      "grad_norm": 1.2437245782495712,
      "learning_rate": 9.941150547808508e-06,
      "loss": 0.6411,
      "step": 955
    },
    {
      "epoch": 0.05158922885974853,
      "grad_norm": 1.339490879085528,
      "learning_rate": 9.941027508392493e-06,
      "loss": 0.6383,
      "step": 956
    },
    {
      "epoch": 0.051643192488262914,
      "grad_norm": 0.9581110984262579,
      "learning_rate": 9.940904341336859e-06,
      "loss": 0.4195,
      "step": 957
    },
    {
      "epoch": 0.05169715611677729,
      "grad_norm": 1.2973297179797718,
      "learning_rate": 9.940781046645147e-06,
      "loss": 0.642,
      "step": 958
    },
    {
      "epoch": 0.051751119745291674,
      "grad_norm": 1.1188470742790184,
      "learning_rate": 9.940657624320901e-06,
      "loss": 0.607,
      "step": 959
    },
    {
      "epoch": 0.05180508337380606,
      "grad_norm": 1.0749905459353266,
      "learning_rate": 9.940534074367667e-06,
      "loss": 0.5667,
      "step": 960
    },
    {
      "epoch": 0.051859047002320434,
      "grad_norm": 1.1126112661538172,
      "learning_rate": 9.940410396788997e-06,
      "loss": 0.5524,
      "step": 961
    },
    {
      "epoch": 0.05191301063083482,
      "grad_norm": 1.068142493588087,
      "learning_rate": 9.940286591588448e-06,
      "loss": 0.4731,
      "step": 962
    },
    {
      "epoch": 0.0519669742593492,
      "grad_norm": 0.9837912435942615,
      "learning_rate": 9.940162658769574e-06,
      "loss": 0.5821,
      "step": 963
    },
    {
      "epoch": 0.05202093788786358,
      "grad_norm": 1.2557086886793707,
      "learning_rate": 9.94003859833594e-06,
      "loss": 0.5706,
      "step": 964
    },
    {
      "epoch": 0.05207490151637796,
      "grad_norm": 1.0808312491617713,
      "learning_rate": 9.939914410291109e-06,
      "loss": 0.4327,
      "step": 965
    },
    {
      "epoch": 0.052128865144892345,
      "grad_norm": 0.9920629003973672,
      "learning_rate": 9.939790094638651e-06,
      "loss": 0.4885,
      "step": 966
    },
    {
      "epoch": 0.05218282877340672,
      "grad_norm": 1.0616517555397011,
      "learning_rate": 9.93966565138214e-06,
      "loss": 0.4787,
      "step": 967
    },
    {
      "epoch": 0.052236792401921105,
      "grad_norm": 1.1129504896846398,
      "learning_rate": 9.939541080525153e-06,
      "loss": 0.4747,
      "step": 968
    },
    {
      "epoch": 0.05229075603043549,
      "grad_norm": 1.2973136398948457,
      "learning_rate": 9.939416382071269e-06,
      "loss": 0.6651,
      "step": 969
    },
    {
      "epoch": 0.052344719658949865,
      "grad_norm": 1.124001588908291,
      "learning_rate": 9.939291556024072e-06,
      "loss": 0.5799,
      "step": 970
    },
    {
      "epoch": 0.05239868328746425,
      "grad_norm": 1.1290480849471018,
      "learning_rate": 9.939166602387152e-06,
      "loss": 0.7056,
      "step": 971
    },
    {
      "epoch": 0.05245264691597863,
      "grad_norm": 1.0055457271969237,
      "learning_rate": 9.939041521164097e-06,
      "loss": 0.5228,
      "step": 972
    },
    {
      "epoch": 0.05250661054449301,
      "grad_norm": 1.2953131890566674,
      "learning_rate": 9.938916312358505e-06,
      "loss": 0.619,
      "step": 973
    },
    {
      "epoch": 0.05256057417300739,
      "grad_norm": 1.1553552381650756,
      "learning_rate": 9.93879097597397e-06,
      "loss": 0.5775,
      "step": 974
    },
    {
      "epoch": 0.052614537801521775,
      "grad_norm": 1.0619771952527604,
      "learning_rate": 9.938665512014098e-06,
      "loss": 0.5347,
      "step": 975
    },
    {
      "epoch": 0.05266850143003616,
      "grad_norm": 1.265294361991696,
      "learning_rate": 9.938539920482494e-06,
      "loss": 0.6542,
      "step": 976
    },
    {
      "epoch": 0.052722465058550536,
      "grad_norm": 1.2430706697411082,
      "learning_rate": 9.93841420138277e-06,
      "loss": 0.5138,
      "step": 977
    },
    {
      "epoch": 0.05277642868706492,
      "grad_norm": 1.1635300732565532,
      "learning_rate": 9.938288354718536e-06,
      "loss": 0.4974,
      "step": 978
    },
    {
      "epoch": 0.0528303923155793,
      "grad_norm": 1.1917326046310424,
      "learning_rate": 9.93816238049341e-06,
      "loss": 0.5582,
      "step": 979
    },
    {
      "epoch": 0.05288435594409368,
      "grad_norm": 0.8240484699433677,
      "learning_rate": 9.938036278711015e-06,
      "loss": 0.3748,
      "step": 980
    },
    {
      "epoch": 0.05293831957260806,
      "grad_norm": 1.1330558016273184,
      "learning_rate": 9.937910049374971e-06,
      "loss": 0.4984,
      "step": 981
    },
    {
      "epoch": 0.052992283201122446,
      "grad_norm": 1.3269596197489946,
      "learning_rate": 9.937783692488906e-06,
      "loss": 0.5885,
      "step": 982
    },
    {
      "epoch": 0.05304624682963682,
      "grad_norm": 0.9910824435900697,
      "learning_rate": 9.937657208056458e-06,
      "loss": 0.4172,
      "step": 983
    },
    {
      "epoch": 0.053100210458151206,
      "grad_norm": 1.0741608027785567,
      "learning_rate": 9.937530596081254e-06,
      "loss": 0.6688,
      "step": 984
    },
    {
      "epoch": 0.05315417408666559,
      "grad_norm": 1.0619712769687613,
      "learning_rate": 9.937403856566939e-06,
      "loss": 0.4817,
      "step": 985
    },
    {
      "epoch": 0.053208137715179966,
      "grad_norm": 1.2688578788517515,
      "learning_rate": 9.937276989517153e-06,
      "loss": 0.6519,
      "step": 986
    },
    {
      "epoch": 0.05326210134369435,
      "grad_norm": 0.9383043202362142,
      "learning_rate": 9.937149994935543e-06,
      "loss": 0.405,
      "step": 987
    },
    {
      "epoch": 0.053316064972208733,
      "grad_norm": 1.001995972414624,
      "learning_rate": 9.937022872825756e-06,
      "loss": 0.5951,
      "step": 988
    },
    {
      "epoch": 0.05337002860072311,
      "grad_norm": 1.016055997196146,
      "learning_rate": 9.936895623191453e-06,
      "loss": 0.4707,
      "step": 989
    },
    {
      "epoch": 0.053423992229237494,
      "grad_norm": 1.0738252874343075,
      "learning_rate": 9.936768246036283e-06,
      "loss": 0.5661,
      "step": 990
    },
    {
      "epoch": 0.05347795585775188,
      "grad_norm": 1.0950234445594653,
      "learning_rate": 9.936640741363913e-06,
      "loss": 0.4569,
      "step": 991
    },
    {
      "epoch": 0.053531919486266254,
      "grad_norm": 0.8663372520293344,
      "learning_rate": 9.936513109178003e-06,
      "loss": 0.423,
      "step": 992
    },
    {
      "epoch": 0.05358588311478064,
      "grad_norm": 1.143248129276903,
      "learning_rate": 9.936385349482226e-06,
      "loss": 0.6686,
      "step": 993
    },
    {
      "epoch": 0.05363984674329502,
      "grad_norm": 1.159089527874994,
      "learning_rate": 9.936257462280251e-06,
      "loss": 0.5181,
      "step": 994
    },
    {
      "epoch": 0.0536938103718094,
      "grad_norm": 0.9020277036544208,
      "learning_rate": 9.936129447575752e-06,
      "loss": 0.3769,
      "step": 995
    },
    {
      "epoch": 0.05374777400032378,
      "grad_norm": 1.320910263474217,
      "learning_rate": 9.936001305372412e-06,
      "loss": 0.6481,
      "step": 996
    },
    {
      "epoch": 0.053801737628838164,
      "grad_norm": 1.2050958636270876,
      "learning_rate": 9.935873035673911e-06,
      "loss": 0.5827,
      "step": 997
    },
    {
      "epoch": 0.05385570125735254,
      "grad_norm": 0.9171114944046602,
      "learning_rate": 9.935744638483938e-06,
      "loss": 0.4028,
      "step": 998
    },
    {
      "epoch": 0.053909664885866924,
      "grad_norm": 0.7184313624066699,
      "learning_rate": 9.935616113806182e-06,
      "loss": 0.3021,
      "step": 999
    },
    {
      "epoch": 0.05396362851438131,
      "grad_norm": 1.0547786803586519,
      "learning_rate": 9.935487461644336e-06,
      "loss": 0.5216,
      "step": 1000
    },
    {
      "epoch": 0.05396362851438131,
      "eval_loss": 0.6226067543029785,
      "eval_runtime": 164.0493,
      "eval_samples_per_second": 20.963,
      "eval_steps_per_second": 0.878,
      "step": 1000
    },
    {
      "epoch": 0.05401759214289569,
      "grad_norm": 1.2023994644950398,
      "learning_rate": 9.9353586820021e-06,
      "loss": 0.6233,
      "step": 1001
    },
    {
      "epoch": 0.05407155577141007,
      "grad_norm": 0.8612062687703272,
      "learning_rate": 9.935229774883172e-06,
      "loss": 0.4073,
      "step": 1002
    },
    {
      "epoch": 0.05412551939992445,
      "grad_norm": 1.0617764289252274,
      "learning_rate": 9.935100740291259e-06,
      "loss": 0.5222,
      "step": 1003
    },
    {
      "epoch": 0.054179483028438835,
      "grad_norm": 1.3426345089131284,
      "learning_rate": 9.934971578230072e-06,
      "loss": 0.6215,
      "step": 1004
    },
    {
      "epoch": 0.05423344665695321,
      "grad_norm": 1.0878461397701575,
      "learning_rate": 9.934842288703318e-06,
      "loss": 0.614,
      "step": 1005
    },
    {
      "epoch": 0.054287410285467595,
      "grad_norm": 1.3384591563841215,
      "learning_rate": 9.934712871714716e-06,
      "loss": 0.5782,
      "step": 1006
    },
    {
      "epoch": 0.05434137391398198,
      "grad_norm": 1.607785271003181,
      "learning_rate": 9.934583327267986e-06,
      "loss": 0.7857,
      "step": 1007
    },
    {
      "epoch": 0.054395337542496355,
      "grad_norm": 1.3370854430986108,
      "learning_rate": 9.93445365536685e-06,
      "loss": 0.5012,
      "step": 1008
    },
    {
      "epoch": 0.05444930117101074,
      "grad_norm": 0.8279419905356157,
      "learning_rate": 9.934323856015036e-06,
      "loss": 0.5149,
      "step": 1009
    },
    {
      "epoch": 0.05450326479952512,
      "grad_norm": 1.130515250980697,
      "learning_rate": 9.934193929216271e-06,
      "loss": 0.5142,
      "step": 1010
    },
    {
      "epoch": 0.0545572284280395,
      "grad_norm": 0.8609342972741308,
      "learning_rate": 9.934063874974295e-06,
      "loss": 0.4632,
      "step": 1011
    },
    {
      "epoch": 0.05461119205655388,
      "grad_norm": 1.0774058847067693,
      "learning_rate": 9.93393369329284e-06,
      "loss": 0.5084,
      "step": 1012
    },
    {
      "epoch": 0.054665155685068266,
      "grad_norm": 1.0401120193620315,
      "learning_rate": 9.933803384175654e-06,
      "loss": 0.5945,
      "step": 1013
    },
    {
      "epoch": 0.05471911931358264,
      "grad_norm": 1.1362958751041685,
      "learning_rate": 9.933672947626476e-06,
      "loss": 0.5032,
      "step": 1014
    },
    {
      "epoch": 0.054773082942097026,
      "grad_norm": 1.1736921606921518,
      "learning_rate": 9.93354238364906e-06,
      "loss": 0.5598,
      "step": 1015
    },
    {
      "epoch": 0.05482704657061141,
      "grad_norm": 0.8712858333084482,
      "learning_rate": 9.933411692247155e-06,
      "loss": 0.3491,
      "step": 1016
    },
    {
      "epoch": 0.054881010199125786,
      "grad_norm": 0.90698567107981,
      "learning_rate": 9.933280873424517e-06,
      "loss": 0.3474,
      "step": 1017
    },
    {
      "epoch": 0.05493497382764017,
      "grad_norm": 0.9719069159174657,
      "learning_rate": 9.93314992718491e-06,
      "loss": 0.4207,
      "step": 1018
    },
    {
      "epoch": 0.05498893745615455,
      "grad_norm": 1.1279934819881443,
      "learning_rate": 9.933018853532093e-06,
      "loss": 0.5144,
      "step": 1019
    },
    {
      "epoch": 0.05504290108466893,
      "grad_norm": 1.0669557952489312,
      "learning_rate": 9.932887652469836e-06,
      "loss": 0.5522,
      "step": 1020
    },
    {
      "epoch": 0.05509686471318331,
      "grad_norm": 1.3518367262557713,
      "learning_rate": 9.932756324001907e-06,
      "loss": 0.5376,
      "step": 1021
    },
    {
      "epoch": 0.0551508283416977,
      "grad_norm": 0.970990148943271,
      "learning_rate": 9.932624868132082e-06,
      "loss": 0.4878,
      "step": 1022
    },
    {
      "epoch": 0.05520479197021208,
      "grad_norm": 1.495257366842139,
      "learning_rate": 9.93249328486414e-06,
      "loss": 0.6202,
      "step": 1023
    },
    {
      "epoch": 0.05525875559872646,
      "grad_norm": 1.4963237547364954,
      "learning_rate": 9.932361574201865e-06,
      "loss": 0.9286,
      "step": 1024
    },
    {
      "epoch": 0.05531271922724084,
      "grad_norm": 1.093292082047647,
      "learning_rate": 9.932229736149037e-06,
      "loss": 0.591,
      "step": 1025
    },
    {
      "epoch": 0.055366682855755224,
      "grad_norm": 1.073237667282872,
      "learning_rate": 9.932097770709447e-06,
      "loss": 0.4965,
      "step": 1026
    },
    {
      "epoch": 0.0554206464842696,
      "grad_norm": 1.1836114874130435,
      "learning_rate": 9.931965677886892e-06,
      "loss": 0.6085,
      "step": 1027
    },
    {
      "epoch": 0.055474610112783984,
      "grad_norm": 1.0930939128299604,
      "learning_rate": 9.931833457685163e-06,
      "loss": 0.4859,
      "step": 1028
    },
    {
      "epoch": 0.05552857374129837,
      "grad_norm": 1.063142817823407,
      "learning_rate": 9.931701110108064e-06,
      "loss": 0.5838,
      "step": 1029
    },
    {
      "epoch": 0.055582537369812744,
      "grad_norm": 1.1531640236787277,
      "learning_rate": 9.931568635159396e-06,
      "loss": 0.7143,
      "step": 1030
    },
    {
      "epoch": 0.05563650099832713,
      "grad_norm": 0.9688223805285074,
      "learning_rate": 9.931436032842965e-06,
      "loss": 0.4661,
      "step": 1031
    },
    {
      "epoch": 0.05569046462684151,
      "grad_norm": 1.1774047790315187,
      "learning_rate": 9.93130330316259e-06,
      "loss": 0.5187,
      "step": 1032
    },
    {
      "epoch": 0.05574442825535589,
      "grad_norm": 1.1309853070142364,
      "learning_rate": 9.931170446122077e-06,
      "loss": 0.4226,
      "step": 1033
    },
    {
      "epoch": 0.05579839188387027,
      "grad_norm": 1.1410310957879708,
      "learning_rate": 9.931037461725249e-06,
      "loss": 0.6308,
      "step": 1034
    },
    {
      "epoch": 0.055852355512384655,
      "grad_norm": 1.2339220973976475,
      "learning_rate": 9.930904349975927e-06,
      "loss": 0.7152,
      "step": 1035
    },
    {
      "epoch": 0.05590631914089903,
      "grad_norm": 1.218639895042501,
      "learning_rate": 9.930771110877938e-06,
      "loss": 0.6497,
      "step": 1036
    },
    {
      "epoch": 0.055960282769413415,
      "grad_norm": 1.2044668967273855,
      "learning_rate": 9.930637744435108e-06,
      "loss": 0.7125,
      "step": 1037
    },
    {
      "epoch": 0.0560142463979278,
      "grad_norm": 1.1617520760101143,
      "learning_rate": 9.930504250651274e-06,
      "loss": 0.5038,
      "step": 1038
    },
    {
      "epoch": 0.056068210026442175,
      "grad_norm": 0.9601996092976851,
      "learning_rate": 9.930370629530272e-06,
      "loss": 0.5533,
      "step": 1039
    },
    {
      "epoch": 0.05612217365495656,
      "grad_norm": 1.0620232727884185,
      "learning_rate": 9.93023688107594e-06,
      "loss": 0.4111,
      "step": 1040
    },
    {
      "epoch": 0.05617613728347094,
      "grad_norm": 0.9838885960634436,
      "learning_rate": 9.930103005292125e-06,
      "loss": 0.4681,
      "step": 1041
    },
    {
      "epoch": 0.05623010091198532,
      "grad_norm": 1.041853751086941,
      "learning_rate": 9.929969002182672e-06,
      "loss": 0.486,
      "step": 1042
    },
    {
      "epoch": 0.0562840645404997,
      "grad_norm": 1.3353148178216587,
      "learning_rate": 9.929834871751434e-06,
      "loss": 0.5567,
      "step": 1043
    },
    {
      "epoch": 0.056338028169014086,
      "grad_norm": 1.014180967498552,
      "learning_rate": 9.929700614002265e-06,
      "loss": 0.5128,
      "step": 1044
    },
    {
      "epoch": 0.05639199179752846,
      "grad_norm": 0.7958715730878317,
      "learning_rate": 9.929566228939025e-06,
      "loss": 0.4399,
      "step": 1045
    },
    {
      "epoch": 0.056445955426042846,
      "grad_norm": 1.0517660060868628,
      "learning_rate": 9.929431716565577e-06,
      "loss": 0.557,
      "step": 1046
    },
    {
      "epoch": 0.05649991905455723,
      "grad_norm": 1.012651541210173,
      "learning_rate": 9.929297076885785e-06,
      "loss": 0.4803,
      "step": 1047
    },
    {
      "epoch": 0.05655388268307161,
      "grad_norm": 0.9239130151281298,
      "learning_rate": 9.92916230990352e-06,
      "loss": 0.4523,
      "step": 1048
    },
    {
      "epoch": 0.05660784631158599,
      "grad_norm": 1.1115792548533128,
      "learning_rate": 9.929027415622655e-06,
      "loss": 0.5136,
      "step": 1049
    },
    {
      "epoch": 0.05666180994010037,
      "grad_norm": 1.14445002946903,
      "learning_rate": 9.928892394047066e-06,
      "loss": 0.652,
      "step": 1050
    },
    {
      "epoch": 0.05671577356861476,
      "grad_norm": 1.080993708601679,
      "learning_rate": 9.928757245180635e-06,
      "loss": 0.5891,
      "step": 1051
    },
    {
      "epoch": 0.05676973719712913,
      "grad_norm": 1.0509488370260447,
      "learning_rate": 9.928621969027246e-06,
      "loss": 0.4039,
      "step": 1052
    },
    {
      "epoch": 0.05682370082564352,
      "grad_norm": 1.2389372245481627,
      "learning_rate": 9.928486565590785e-06,
      "loss": 0.6045,
      "step": 1053
    },
    {
      "epoch": 0.0568776644541579,
      "grad_norm": 0.9911170868183955,
      "learning_rate": 9.928351034875149e-06,
      "loss": 0.5899,
      "step": 1054
    },
    {
      "epoch": 0.05693162808267228,
      "grad_norm": 1.0892073043968464,
      "learning_rate": 9.928215376884227e-06,
      "loss": 0.6116,
      "step": 1055
    },
    {
      "epoch": 0.05698559171118666,
      "grad_norm": 1.4176933196564943,
      "learning_rate": 9.928079591621922e-06,
      "loss": 0.57,
      "step": 1056
    },
    {
      "epoch": 0.057039555339701044,
      "grad_norm": 1.076250954270727,
      "learning_rate": 9.927943679092134e-06,
      "loss": 0.5164,
      "step": 1057
    },
    {
      "epoch": 0.05709351896821542,
      "grad_norm": 0.8958215243377342,
      "learning_rate": 9.927807639298772e-06,
      "loss": 0.3483,
      "step": 1058
    },
    {
      "epoch": 0.057147482596729804,
      "grad_norm": 1.3218711661544011,
      "learning_rate": 9.927671472245743e-06,
      "loss": 0.6363,
      "step": 1059
    },
    {
      "epoch": 0.05720144622524419,
      "grad_norm": 1.215084343070576,
      "learning_rate": 9.927535177936964e-06,
      "loss": 0.5085,
      "step": 1060
    },
    {
      "epoch": 0.057255409853758564,
      "grad_norm": 1.172103929714746,
      "learning_rate": 9.927398756376348e-06,
      "loss": 0.6634,
      "step": 1061
    },
    {
      "epoch": 0.05730937348227295,
      "grad_norm": 0.8655345151576931,
      "learning_rate": 9.927262207567817e-06,
      "loss": 0.3874,
      "step": 1062
    },
    {
      "epoch": 0.05736333711078733,
      "grad_norm": 1.046629513682674,
      "learning_rate": 9.9271255315153e-06,
      "loss": 0.5689,
      "step": 1063
    },
    {
      "epoch": 0.05741730073930171,
      "grad_norm": 1.3747233185738972,
      "learning_rate": 9.926988728222721e-06,
      "loss": 0.7255,
      "step": 1064
    },
    {
      "epoch": 0.05747126436781609,
      "grad_norm": 1.215027122982026,
      "learning_rate": 9.926851797694012e-06,
      "loss": 0.5023,
      "step": 1065
    },
    {
      "epoch": 0.057525227996330475,
      "grad_norm": 1.0386643954071433,
      "learning_rate": 9.926714739933111e-06,
      "loss": 0.4577,
      "step": 1066
    },
    {
      "epoch": 0.05757919162484485,
      "grad_norm": 0.9931422641125638,
      "learning_rate": 9.926577554943955e-06,
      "loss": 0.3957,
      "step": 1067
    },
    {
      "epoch": 0.057633155253359235,
      "grad_norm": 1.0768245900535862,
      "learning_rate": 9.926440242730486e-06,
      "loss": 0.4253,
      "step": 1068
    },
    {
      "epoch": 0.05768711888187362,
      "grad_norm": 1.075005360701322,
      "learning_rate": 9.926302803296654e-06,
      "loss": 0.5134,
      "step": 1069
    },
    {
      "epoch": 0.057741082510388,
      "grad_norm": 1.1832985481611653,
      "learning_rate": 9.926165236646405e-06,
      "loss": 0.5456,
      "step": 1070
    },
    {
      "epoch": 0.05779504613890238,
      "grad_norm": 1.2279499461384902,
      "learning_rate": 9.926027542783695e-06,
      "loss": 0.4491,
      "step": 1071
    },
    {
      "epoch": 0.05784900976741676,
      "grad_norm": 1.0539094085718188,
      "learning_rate": 9.925889721712482e-06,
      "loss": 0.4674,
      "step": 1072
    },
    {
      "epoch": 0.057902973395931145,
      "grad_norm": 1.036909349965767,
      "learning_rate": 9.925751773436726e-06,
      "loss": 0.5682,
      "step": 1073
    },
    {
      "epoch": 0.05795693702444552,
      "grad_norm": 1.3448201955446006,
      "learning_rate": 9.925613697960394e-06,
      "loss": 0.5136,
      "step": 1074
    },
    {
      "epoch": 0.058010900652959906,
      "grad_norm": 1.3956994166612555,
      "learning_rate": 9.925475495287449e-06,
      "loss": 0.697,
      "step": 1075
    },
    {
      "epoch": 0.05806486428147429,
      "grad_norm": 1.2027719043061338,
      "learning_rate": 9.92533716542187e-06,
      "loss": 0.5785,
      "step": 1076
    },
    {
      "epoch": 0.058118827909988666,
      "grad_norm": 1.1998103977282748,
      "learning_rate": 9.925198708367627e-06,
      "loss": 0.5562,
      "step": 1077
    },
    {
      "epoch": 0.05817279153850305,
      "grad_norm": 1.0382680257462724,
      "learning_rate": 9.925060124128702e-06,
      "loss": 0.5405,
      "step": 1078
    },
    {
      "epoch": 0.05822675516701743,
      "grad_norm": 0.7110051637436826,
      "learning_rate": 9.924921412709078e-06,
      "loss": 0.2886,
      "step": 1079
    },
    {
      "epoch": 0.05828071879553181,
      "grad_norm": 1.0381054017020952,
      "learning_rate": 9.924782574112743e-06,
      "loss": 0.4543,
      "step": 1080
    },
    {
      "epoch": 0.05833468242404619,
      "grad_norm": 1.3173854726163965,
      "learning_rate": 9.924643608343685e-06,
      "loss": 0.5803,
      "step": 1081
    },
    {
      "epoch": 0.058388646052560576,
      "grad_norm": 0.9498448360556373,
      "learning_rate": 9.9245045154059e-06,
      "loss": 0.4187,
      "step": 1082
    },
    {
      "epoch": 0.05844260968107495,
      "grad_norm": 0.8521959964681866,
      "learning_rate": 9.924365295303382e-06,
      "loss": 0.392,
      "step": 1083
    },
    {
      "epoch": 0.058496573309589336,
      "grad_norm": 1.0725999110184812,
      "learning_rate": 9.924225948040136e-06,
      "loss": 0.5742,
      "step": 1084
    },
    {
      "epoch": 0.05855053693810372,
      "grad_norm": 1.075339973980148,
      "learning_rate": 9.924086473620165e-06,
      "loss": 0.6659,
      "step": 1085
    },
    {
      "epoch": 0.0586045005666181,
      "grad_norm": 1.0862911255040697,
      "learning_rate": 9.923946872047481e-06,
      "loss": 0.6123,
      "step": 1086
    },
    {
      "epoch": 0.05865846419513248,
      "grad_norm": 1.2162331075707398,
      "learning_rate": 9.923807143326092e-06,
      "loss": 0.4996,
      "step": 1087
    },
    {
      "epoch": 0.058712427823646864,
      "grad_norm": 1.172245830967354,
      "learning_rate": 9.923667287460017e-06,
      "loss": 0.5834,
      "step": 1088
    },
    {
      "epoch": 0.05876639145216124,
      "grad_norm": 1.1463887060631397,
      "learning_rate": 9.923527304453273e-06,
      "loss": 0.5358,
      "step": 1089
    },
    {
      "epoch": 0.058820355080675624,
      "grad_norm": 1.1476867291157467,
      "learning_rate": 9.923387194309887e-06,
      "loss": 0.8617,
      "step": 1090
    },
    {
      "epoch": 0.05887431870919001,
      "grad_norm": 1.3333360690245066,
      "learning_rate": 9.923246957033881e-06,
      "loss": 0.4891,
      "step": 1091
    },
    {
      "epoch": 0.058928282337704384,
      "grad_norm": 1.169477520921293,
      "learning_rate": 9.92310659262929e-06,
      "loss": 0.5805,
      "step": 1092
    },
    {
      "epoch": 0.05898224596621877,
      "grad_norm": 1.3279530837610412,
      "learning_rate": 9.922966101100143e-06,
      "loss": 0.585,
      "step": 1093
    },
    {
      "epoch": 0.05903620959473315,
      "grad_norm": 1.2730111755866398,
      "learning_rate": 9.922825482450484e-06,
      "loss": 0.5245,
      "step": 1094
    },
    {
      "epoch": 0.059090173223247534,
      "grad_norm": 0.851589004597301,
      "learning_rate": 9.922684736684351e-06,
      "loss": 0.365,
      "step": 1095
    },
    {
      "epoch": 0.05914413685176191,
      "grad_norm": 0.9334940177851558,
      "learning_rate": 9.92254386380579e-06,
      "loss": 0.4499,
      "step": 1096
    },
    {
      "epoch": 0.059198100480276294,
      "grad_norm": 1.3891174706133667,
      "learning_rate": 9.92240286381885e-06,
      "loss": 0.4977,
      "step": 1097
    },
    {
      "epoch": 0.05925206410879068,
      "grad_norm": 1.1099315485360044,
      "learning_rate": 9.922261736727583e-06,
      "loss": 0.4729,
      "step": 1098
    },
    {
      "epoch": 0.059306027737305055,
      "grad_norm": 1.207761323144064,
      "learning_rate": 9.922120482536045e-06,
      "loss": 0.6132,
      "step": 1099
    },
    {
      "epoch": 0.05935999136581944,
      "grad_norm": 1.0804922418361151,
      "learning_rate": 9.921979101248296e-06,
      "loss": 0.4625,
      "step": 1100
    },
    {
      "epoch": 0.05941395499433382,
      "grad_norm": 1.2669076847139964,
      "learning_rate": 9.9218375928684e-06,
      "loss": 0.5528,
      "step": 1101
    },
    {
      "epoch": 0.0594679186228482,
      "grad_norm": 1.0411547516200543,
      "learning_rate": 9.921695957400425e-06,
      "loss": 0.4875,
      "step": 1102
    },
    {
      "epoch": 0.05952188225136258,
      "grad_norm": 0.8224949030006246,
      "learning_rate": 9.921554194848438e-06,
      "loss": 0.3581,
      "step": 1103
    },
    {
      "epoch": 0.059575845879876965,
      "grad_norm": 1.2174325731156257,
      "learning_rate": 9.921412305216516e-06,
      "loss": 0.5318,
      "step": 1104
    },
    {
      "epoch": 0.05962980950839134,
      "grad_norm": 1.176216394943767,
      "learning_rate": 9.921270288508738e-06,
      "loss": 0.4839,
      "step": 1105
    },
    {
      "epoch": 0.059683773136905725,
      "grad_norm": 1.173592501411679,
      "learning_rate": 9.921128144729185e-06,
      "loss": 0.5715,
      "step": 1106
    },
    {
      "epoch": 0.05973773676542011,
      "grad_norm": 0.9408335407170557,
      "learning_rate": 9.92098587388194e-06,
      "loss": 0.3462,
      "step": 1107
    },
    {
      "epoch": 0.059791700393934485,
      "grad_norm": 1.2356074106048929,
      "learning_rate": 9.920843475971096e-06,
      "loss": 0.7519,
      "step": 1108
    },
    {
      "epoch": 0.05984566402244887,
      "grad_norm": 1.256223513423086,
      "learning_rate": 9.92070095100074e-06,
      "loss": 0.615,
      "step": 1109
    },
    {
      "epoch": 0.05989962765096325,
      "grad_norm": 0.9682461083343752,
      "learning_rate": 9.920558298974974e-06,
      "loss": 0.4698,
      "step": 1110
    },
    {
      "epoch": 0.05995359127947763,
      "grad_norm": 1.112808887345549,
      "learning_rate": 9.920415519897898e-06,
      "loss": 0.5708,
      "step": 1111
    },
    {
      "epoch": 0.06000755490799201,
      "grad_norm": 1.294263814030072,
      "learning_rate": 9.920272613773611e-06,
      "loss": 0.5738,
      "step": 1112
    },
    {
      "epoch": 0.060061518536506396,
      "grad_norm": 1.1806926711202252,
      "learning_rate": 9.920129580606224e-06,
      "loss": 0.4584,
      "step": 1113
    },
    {
      "epoch": 0.06011548216502077,
      "grad_norm": 1.2492989748049381,
      "learning_rate": 9.919986420399846e-06,
      "loss": 0.5246,
      "step": 1114
    },
    {
      "epoch": 0.060169445793535156,
      "grad_norm": 1.149865129144684,
      "learning_rate": 9.919843133158592e-06,
      "loss": 0.4929,
      "step": 1115
    },
    {
      "epoch": 0.06022340942204954,
      "grad_norm": 1.0603190437917798,
      "learning_rate": 9.91969971888658e-06,
      "loss": 0.5278,
      "step": 1116
    },
    {
      "epoch": 0.06027737305056392,
      "grad_norm": 1.221355498686458,
      "learning_rate": 9.919556177587932e-06,
      "loss": 0.5234,
      "step": 1117
    },
    {
      "epoch": 0.0603313366790783,
      "grad_norm": 1.3173457859269384,
      "learning_rate": 9.919412509266775e-06,
      "loss": 0.6905,
      "step": 1118
    },
    {
      "epoch": 0.06038530030759268,
      "grad_norm": 0.976047658477216,
      "learning_rate": 9.919268713927236e-06,
      "loss": 0.445,
      "step": 1119
    },
    {
      "epoch": 0.06043926393610707,
      "grad_norm": 1.1178463171268915,
      "learning_rate": 9.91912479157345e-06,
      "loss": 0.5306,
      "step": 1120
    },
    {
      "epoch": 0.06049322756462144,
      "grad_norm": 2.1512471198640797,
      "learning_rate": 9.918980742209552e-06,
      "loss": 0.6731,
      "step": 1121
    },
    {
      "epoch": 0.06054719119313583,
      "grad_norm": 1.1287806516156254,
      "learning_rate": 9.918836565839682e-06,
      "loss": 0.5513,
      "step": 1122
    },
    {
      "epoch": 0.06060115482165021,
      "grad_norm": 1.2546453911339908,
      "learning_rate": 9.918692262467985e-06,
      "loss": 0.5559,
      "step": 1123
    },
    {
      "epoch": 0.06065511845016459,
      "grad_norm": 1.1685790377528955,
      "learning_rate": 9.918547832098606e-06,
      "loss": 0.6592,
      "step": 1124
    },
    {
      "epoch": 0.06070908207867897,
      "grad_norm": 1.2586584645164742,
      "learning_rate": 9.9184032747357e-06,
      "loss": 0.5594,
      "step": 1125
    },
    {
      "epoch": 0.060763045707193354,
      "grad_norm": 1.020357620807058,
      "learning_rate": 9.918258590383417e-06,
      "loss": 0.4455,
      "step": 1126
    },
    {
      "epoch": 0.06081700933570773,
      "grad_norm": 0.8355037912170709,
      "learning_rate": 9.91811377904592e-06,
      "loss": 0.3881,
      "step": 1127
    },
    {
      "epoch": 0.060870972964222114,
      "grad_norm": 1.0330667095735402,
      "learning_rate": 9.917968840727366e-06,
      "loss": 0.4268,
      "step": 1128
    },
    {
      "epoch": 0.0609249365927365,
      "grad_norm": 1.1430983240252517,
      "learning_rate": 9.917823775431927e-06,
      "loss": 0.4948,
      "step": 1129
    },
    {
      "epoch": 0.060978900221250874,
      "grad_norm": 1.0255848846305782,
      "learning_rate": 9.917678583163766e-06,
      "loss": 0.5068,
      "step": 1130
    },
    {
      "epoch": 0.06103286384976526,
      "grad_norm": 1.5068935437787216,
      "learning_rate": 9.91753326392706e-06,
      "loss": 0.6864,
      "step": 1131
    },
    {
      "epoch": 0.06108682747827964,
      "grad_norm": 0.9076958075822859,
      "learning_rate": 9.917387817725984e-06,
      "loss": 0.4973,
      "step": 1132
    },
    {
      "epoch": 0.06114079110679402,
      "grad_norm": 1.082157768190573,
      "learning_rate": 9.91724224456472e-06,
      "loss": 0.6103,
      "step": 1133
    },
    {
      "epoch": 0.0611947547353084,
      "grad_norm": 0.8786983823623391,
      "learning_rate": 9.917096544447448e-06,
      "loss": 0.392,
      "step": 1134
    },
    {
      "epoch": 0.061248718363822785,
      "grad_norm": 1.25416503663267,
      "learning_rate": 9.916950717378361e-06,
      "loss": 0.5387,
      "step": 1135
    },
    {
      "epoch": 0.06130268199233716,
      "grad_norm": 0.8202764755749875,
      "learning_rate": 9.916804763361644e-06,
      "loss": 0.3534,
      "step": 1136
    },
    {
      "epoch": 0.061356645620851545,
      "grad_norm": 1.2845513020235255,
      "learning_rate": 9.916658682401497e-06,
      "loss": 0.6599,
      "step": 1137
    },
    {
      "epoch": 0.06141060924936593,
      "grad_norm": 1.1322518063569047,
      "learning_rate": 9.91651247450212e-06,
      "loss": 0.6219,
      "step": 1138
    },
    {
      "epoch": 0.061464572877880305,
      "grad_norm": 1.215808317793313,
      "learning_rate": 9.916366139667707e-06,
      "loss": 0.5506,
      "step": 1139
    },
    {
      "epoch": 0.06151853650639469,
      "grad_norm": 0.9393399404357899,
      "learning_rate": 9.91621967790247e-06,
      "loss": 0.5155,
      "step": 1140
    },
    {
      "epoch": 0.06157250013490907,
      "grad_norm": 1.1820679116000006,
      "learning_rate": 9.916073089210619e-06,
      "loss": 0.5404,
      "step": 1141
    },
    {
      "epoch": 0.061626463763423456,
      "grad_norm": 1.239971889166963,
      "learning_rate": 9.915926373596364e-06,
      "loss": 0.6447,
      "step": 1142
    },
    {
      "epoch": 0.06168042739193783,
      "grad_norm": 0.944595241054684,
      "learning_rate": 9.915779531063923e-06,
      "loss": 0.4077,
      "step": 1143
    },
    {
      "epoch": 0.061734391020452216,
      "grad_norm": 1.0904850277250273,
      "learning_rate": 9.915632561617518e-06,
      "loss": 0.4909,
      "step": 1144
    },
    {
      "epoch": 0.0617883546489666,
      "grad_norm": 0.7567115234235193,
      "learning_rate": 9.91548546526137e-06,
      "loss": 0.3562,
      "step": 1145
    },
    {
      "epoch": 0.061842318277480976,
      "grad_norm": 1.21617298381758,
      "learning_rate": 9.915338241999707e-06,
      "loss": 0.6208,
      "step": 1146
    },
    {
      "epoch": 0.06189628190599536,
      "grad_norm": 1.0513148380621817,
      "learning_rate": 9.915190891836764e-06,
      "loss": 0.4853,
      "step": 1147
    },
    {
      "epoch": 0.06195024553450974,
      "grad_norm": 1.1441813281399302,
      "learning_rate": 9.91504341477677e-06,
      "loss": 0.6087,
      "step": 1148
    },
    {
      "epoch": 0.06200420916302412,
      "grad_norm": 1.1407880484823956,
      "learning_rate": 9.914895810823971e-06,
      "loss": 0.5844,
      "step": 1149
    },
    {
      "epoch": 0.0620581727915385,
      "grad_norm": 1.4385825400129781,
      "learning_rate": 9.914748079982605e-06,
      "loss": 0.6704,
      "step": 1150
    },
    {
      "epoch": 0.06211213642005289,
      "grad_norm": 1.0480631466226744,
      "learning_rate": 9.914600222256919e-06,
      "loss": 0.4473,
      "step": 1151
    },
    {
      "epoch": 0.06216610004856726,
      "grad_norm": 1.1797892665469236,
      "learning_rate": 9.914452237651158e-06,
      "loss": 0.5103,
      "step": 1152
    },
    {
      "epoch": 0.06222006367708165,
      "grad_norm": 1.5032582818757396,
      "learning_rate": 9.914304126169583e-06,
      "loss": 0.6047,
      "step": 1153
    },
    {
      "epoch": 0.06227402730559603,
      "grad_norm": 0.8684947886284589,
      "learning_rate": 9.914155887816446e-06,
      "loss": 0.3546,
      "step": 1154
    },
    {
      "epoch": 0.06232799093411041,
      "grad_norm": 1.5666880896107755,
      "learning_rate": 9.914007522596007e-06,
      "loss": 0.5919,
      "step": 1155
    },
    {
      "epoch": 0.06238195456262479,
      "grad_norm": 1.0510489293032053,
      "learning_rate": 9.913859030512534e-06,
      "loss": 0.3881,
      "step": 1156
    },
    {
      "epoch": 0.062435918191139174,
      "grad_norm": 1.4114247599651213,
      "learning_rate": 9.913710411570292e-06,
      "loss": 0.6325,
      "step": 1157
    },
    {
      "epoch": 0.06248988181965355,
      "grad_norm": 1.2126095428117198,
      "learning_rate": 9.913561665773554e-06,
      "loss": 0.5722,
      "step": 1158
    },
    {
      "epoch": 0.06254384544816793,
      "grad_norm": 1.057885219383531,
      "learning_rate": 9.913412793126592e-06,
      "loss": 0.5153,
      "step": 1159
    },
    {
      "epoch": 0.06259780907668232,
      "grad_norm": 1.2375173887892084,
      "learning_rate": 9.913263793633687e-06,
      "loss": 0.5986,
      "step": 1160
    },
    {
      "epoch": 0.0626517727051967,
      "grad_norm": 1.0036003195251175,
      "learning_rate": 9.913114667299122e-06,
      "loss": 0.5251,
      "step": 1161
    },
    {
      "epoch": 0.06270573633371108,
      "grad_norm": 1.0323354816185977,
      "learning_rate": 9.912965414127182e-06,
      "loss": 0.6265,
      "step": 1162
    },
    {
      "epoch": 0.06275969996222545,
      "grad_norm": 1.1574169313900733,
      "learning_rate": 9.91281603412216e-06,
      "loss": 0.6769,
      "step": 1163
    },
    {
      "epoch": 0.06281366359073984,
      "grad_norm": 1.0494706011749473,
      "learning_rate": 9.912666527288341e-06,
      "loss": 0.4362,
      "step": 1164
    },
    {
      "epoch": 0.06286762721925422,
      "grad_norm": 1.2761413413548277,
      "learning_rate": 9.912516893630031e-06,
      "loss": 0.6058,
      "step": 1165
    },
    {
      "epoch": 0.0629215908477686,
      "grad_norm": 1.4476528384056986,
      "learning_rate": 9.912367133151528e-06,
      "loss": 0.5537,
      "step": 1166
    },
    {
      "epoch": 0.06297555447628299,
      "grad_norm": 1.152500597590952,
      "learning_rate": 9.912217245857132e-06,
      "loss": 0.3886,
      "step": 1167
    },
    {
      "epoch": 0.06302951810479737,
      "grad_norm": 1.480417601860406,
      "learning_rate": 9.912067231751156e-06,
      "loss": 0.5057,
      "step": 1168
    },
    {
      "epoch": 0.06308348173331174,
      "grad_norm": 1.194157368716829,
      "learning_rate": 9.91191709083791e-06,
      "loss": 0.4916,
      "step": 1169
    },
    {
      "epoch": 0.06313744536182612,
      "grad_norm": 0.8836345252791998,
      "learning_rate": 9.911766823121706e-06,
      "loss": 0.3916,
      "step": 1170
    },
    {
      "epoch": 0.06319140899034051,
      "grad_norm": 1.3154308440943532,
      "learning_rate": 9.91161642860687e-06,
      "loss": 0.5987,
      "step": 1171
    },
    {
      "epoch": 0.06324537261885489,
      "grad_norm": 1.0400054157092624,
      "learning_rate": 9.911465907297718e-06,
      "loss": 0.4697,
      "step": 1172
    },
    {
      "epoch": 0.06329933624736928,
      "grad_norm": 1.1416142832602578,
      "learning_rate": 9.911315259198579e-06,
      "loss": 0.4946,
      "step": 1173
    },
    {
      "epoch": 0.06335329987588366,
      "grad_norm": 0.9914967202066232,
      "learning_rate": 9.91116448431378e-06,
      "loss": 0.4809,
      "step": 1174
    },
    {
      "epoch": 0.06340726350439804,
      "grad_norm": 1.1495280753890167,
      "learning_rate": 9.911013582647658e-06,
      "loss": 0.5652,
      "step": 1175
    },
    {
      "epoch": 0.06346122713291241,
      "grad_norm": 1.2240616377358424,
      "learning_rate": 9.910862554204548e-06,
      "loss": 0.6197,
      "step": 1176
    },
    {
      "epoch": 0.0635151907614268,
      "grad_norm": 1.209973008453103,
      "learning_rate": 9.910711398988793e-06,
      "loss": 0.6588,
      "step": 1177
    },
    {
      "epoch": 0.06356915438994118,
      "grad_norm": 1.1091934374707177,
      "learning_rate": 9.910560117004734e-06,
      "loss": 0.5144,
      "step": 1178
    },
    {
      "epoch": 0.06362311801845556,
      "grad_norm": 1.0043435373715903,
      "learning_rate": 9.910408708256722e-06,
      "loss": 0.4508,
      "step": 1179
    },
    {
      "epoch": 0.06367708164696995,
      "grad_norm": 1.3952575583295528,
      "learning_rate": 9.910257172749106e-06,
      "loss": 0.6002,
      "step": 1180
    },
    {
      "epoch": 0.06373104527548433,
      "grad_norm": 0.8719066868090879,
      "learning_rate": 9.910105510486244e-06,
      "loss": 0.3516,
      "step": 1181
    },
    {
      "epoch": 0.0637850089039987,
      "grad_norm": 1.1934117753541411,
      "learning_rate": 9.909953721472495e-06,
      "loss": 0.7167,
      "step": 1182
    },
    {
      "epoch": 0.06383897253251308,
      "grad_norm": 0.8587043230756479,
      "learning_rate": 9.909801805712218e-06,
      "loss": 0.3118,
      "step": 1183
    },
    {
      "epoch": 0.06389293616102747,
      "grad_norm": 1.1854522015607856,
      "learning_rate": 9.90964976320978e-06,
      "loss": 0.5166,
      "step": 1184
    },
    {
      "epoch": 0.06394689978954185,
      "grad_norm": 1.0099459697692823,
      "learning_rate": 9.909497593969554e-06,
      "loss": 0.5383,
      "step": 1185
    },
    {
      "epoch": 0.06400086341805623,
      "grad_norm": 1.293744297887391,
      "learning_rate": 9.90934529799591e-06,
      "loss": 0.6095,
      "step": 1186
    },
    {
      "epoch": 0.06405482704657062,
      "grad_norm": 1.0666799915816103,
      "learning_rate": 9.909192875293229e-06,
      "loss": 0.5046,
      "step": 1187
    },
    {
      "epoch": 0.06410879067508499,
      "grad_norm": 1.1739432941456882,
      "learning_rate": 9.909040325865888e-06,
      "loss": 0.621,
      "step": 1188
    },
    {
      "epoch": 0.06416275430359937,
      "grad_norm": 1.1147396812816694,
      "learning_rate": 9.908887649718274e-06,
      "loss": 0.6104,
      "step": 1189
    },
    {
      "epoch": 0.06421671793211375,
      "grad_norm": 1.1497117814066808,
      "learning_rate": 9.908734846854774e-06,
      "loss": 0.4546,
      "step": 1190
    },
    {
      "epoch": 0.06427068156062814,
      "grad_norm": 1.23405205033504,
      "learning_rate": 9.90858191727978e-06,
      "loss": 0.5482,
      "step": 1191
    },
    {
      "epoch": 0.06432464518914252,
      "grad_norm": 1.273379325808436,
      "learning_rate": 9.908428860997686e-06,
      "loss": 0.5062,
      "step": 1192
    },
    {
      "epoch": 0.0643786088176569,
      "grad_norm": 1.1045224742364133,
      "learning_rate": 9.908275678012892e-06,
      "loss": 0.548,
      "step": 1193
    },
    {
      "epoch": 0.06443257244617127,
      "grad_norm": 1.0930908233723715,
      "learning_rate": 9.908122368329801e-06,
      "loss": 0.6525,
      "step": 1194
    },
    {
      "epoch": 0.06448653607468566,
      "grad_norm": 1.0660427665845988,
      "learning_rate": 9.907968931952821e-06,
      "loss": 0.6039,
      "step": 1195
    },
    {
      "epoch": 0.06454049970320004,
      "grad_norm": 0.8476648078949227,
      "learning_rate": 9.90781536888636e-06,
      "loss": 0.432,
      "step": 1196
    },
    {
      "epoch": 0.06459446333171442,
      "grad_norm": 1.3671380001878286,
      "learning_rate": 9.907661679134829e-06,
      "loss": 0.8111,
      "step": 1197
    },
    {
      "epoch": 0.06464842696022881,
      "grad_norm": 0.9528680484148464,
      "learning_rate": 9.907507862702649e-06,
      "loss": 0.4975,
      "step": 1198
    },
    {
      "epoch": 0.06470239058874319,
      "grad_norm": 1.2964981045499149,
      "learning_rate": 9.90735391959424e-06,
      "loss": 0.7153,
      "step": 1199
    },
    {
      "epoch": 0.06475635421725758,
      "grad_norm": 1.0355378456042084,
      "learning_rate": 9.907199849814025e-06,
      "loss": 0.5623,
      "step": 1200
    },
    {
      "epoch": 0.06481031784577194,
      "grad_norm": 1.2537173730438371,
      "learning_rate": 9.907045653366435e-06,
      "loss": 0.5,
      "step": 1201
    },
    {
      "epoch": 0.06486428147428633,
      "grad_norm": 1.212170226608698,
      "learning_rate": 9.906891330255898e-06,
      "loss": 0.5577,
      "step": 1202
    },
    {
      "epoch": 0.06491824510280071,
      "grad_norm": 1.1121775734259214,
      "learning_rate": 9.906736880486852e-06,
      "loss": 0.5889,
      "step": 1203
    },
    {
      "epoch": 0.0649722087313151,
      "grad_norm": 1.038469497238206,
      "learning_rate": 9.906582304063735e-06,
      "loss": 0.5965,
      "step": 1204
    },
    {
      "epoch": 0.06502617235982948,
      "grad_norm": 1.1056631997683497,
      "learning_rate": 9.90642760099099e-06,
      "loss": 0.5221,
      "step": 1205
    },
    {
      "epoch": 0.06508013598834386,
      "grad_norm": 1.0720565748716335,
      "learning_rate": 9.906272771273064e-06,
      "loss": 0.4575,
      "step": 1206
    },
    {
      "epoch": 0.06513409961685823,
      "grad_norm": 1.1773842941732493,
      "learning_rate": 9.906117814914406e-06,
      "loss": 0.5201,
      "step": 1207
    },
    {
      "epoch": 0.06518806324537262,
      "grad_norm": 1.1483471120084654,
      "learning_rate": 9.905962731919471e-06,
      "loss": 0.5771,
      "step": 1208
    },
    {
      "epoch": 0.065242026873887,
      "grad_norm": 0.9350855019275369,
      "learning_rate": 9.905807522292714e-06,
      "loss": 0.4265,
      "step": 1209
    },
    {
      "epoch": 0.06529599050240138,
      "grad_norm": 1.0033984360258812,
      "learning_rate": 9.905652186038597e-06,
      "loss": 0.4818,
      "step": 1210
    },
    {
      "epoch": 0.06534995413091577,
      "grad_norm": 0.9918315308051188,
      "learning_rate": 9.905496723161585e-06,
      "loss": 0.511,
      "step": 1211
    },
    {
      "epoch": 0.06540391775943015,
      "grad_norm": 1.0145604266526105,
      "learning_rate": 9.905341133666145e-06,
      "loss": 0.4,
      "step": 1212
    },
    {
      "epoch": 0.06545788138794452,
      "grad_norm": 1.008670119335899,
      "learning_rate": 9.90518541755675e-06,
      "loss": 0.5809,
      "step": 1213
    },
    {
      "epoch": 0.0655118450164589,
      "grad_norm": 1.0720431660775398,
      "learning_rate": 9.905029574837876e-06,
      "loss": 0.4742,
      "step": 1214
    },
    {
      "epoch": 0.06556580864497329,
      "grad_norm": 1.1780792982451493,
      "learning_rate": 9.904873605514001e-06,
      "loss": 0.5215,
      "step": 1215
    },
    {
      "epoch": 0.06561977227348767,
      "grad_norm": 1.056284730122944,
      "learning_rate": 9.904717509589607e-06,
      "loss": 0.5105,
      "step": 1216
    },
    {
      "epoch": 0.06567373590200205,
      "grad_norm": 1.1381611853573148,
      "learning_rate": 9.90456128706918e-06,
      "loss": 0.4028,
      "step": 1217
    },
    {
      "epoch": 0.06572769953051644,
      "grad_norm": 1.20677167314594,
      "learning_rate": 9.904404937957213e-06,
      "loss": 0.5369,
      "step": 1218
    },
    {
      "epoch": 0.0657816631590308,
      "grad_norm": 1.1081183549244946,
      "learning_rate": 9.904248462258197e-06,
      "loss": 0.5269,
      "step": 1219
    },
    {
      "epoch": 0.06583562678754519,
      "grad_norm": 1.2323146476458697,
      "learning_rate": 9.90409185997663e-06,
      "loss": 0.55,
      "step": 1220
    },
    {
      "epoch": 0.06588959041605957,
      "grad_norm": 0.9112362945863107,
      "learning_rate": 9.903935131117011e-06,
      "loss": 0.4685,
      "step": 1221
    },
    {
      "epoch": 0.06594355404457396,
      "grad_norm": 1.0296930429924567,
      "learning_rate": 9.90377827568385e-06,
      "loss": 0.4595,
      "step": 1222
    },
    {
      "epoch": 0.06599751767308834,
      "grad_norm": 0.9943557725840793,
      "learning_rate": 9.903621293681651e-06,
      "loss": 0.469,
      "step": 1223
    },
    {
      "epoch": 0.06605148130160272,
      "grad_norm": 1.130129501746787,
      "learning_rate": 9.903464185114925e-06,
      "loss": 0.4656,
      "step": 1224
    },
    {
      "epoch": 0.06610544493011711,
      "grad_norm": 1.038727764660228,
      "learning_rate": 9.90330694998819e-06,
      "loss": 0.5105,
      "step": 1225
    },
    {
      "epoch": 0.06615940855863148,
      "grad_norm": 1.1432536959078887,
      "learning_rate": 9.903149588305962e-06,
      "loss": 0.4604,
      "step": 1226
    },
    {
      "epoch": 0.06621337218714586,
      "grad_norm": 1.1401384528976566,
      "learning_rate": 9.902992100072768e-06,
      "loss": 0.5932,
      "step": 1227
    },
    {
      "epoch": 0.06626733581566024,
      "grad_norm": 0.9822748241981794,
      "learning_rate": 9.902834485293132e-06,
      "loss": 0.4737,
      "step": 1228
    },
    {
      "epoch": 0.06632129944417463,
      "grad_norm": 1.3974625908209215,
      "learning_rate": 9.902676743971584e-06,
      "loss": 0.6035,
      "step": 1229
    },
    {
      "epoch": 0.06637526307268901,
      "grad_norm": 0.7875303962578376,
      "learning_rate": 9.902518876112657e-06,
      "loss": 0.3551,
      "step": 1230
    },
    {
      "epoch": 0.0664292267012034,
      "grad_norm": 0.8107934174695123,
      "learning_rate": 9.902360881720889e-06,
      "loss": 0.3788,
      "step": 1231
    },
    {
      "epoch": 0.06648319032971776,
      "grad_norm": 1.2018556053868974,
      "learning_rate": 9.902202760800823e-06,
      "loss": 0.5138,
      "step": 1232
    },
    {
      "epoch": 0.06653715395823215,
      "grad_norm": 1.0767159731328886,
      "learning_rate": 9.902044513357e-06,
      "loss": 0.5205,
      "step": 1233
    },
    {
      "epoch": 0.06659111758674653,
      "grad_norm": 1.0604471707474723,
      "learning_rate": 9.901886139393968e-06,
      "loss": 0.4008,
      "step": 1234
    },
    {
      "epoch": 0.06664508121526092,
      "grad_norm": 1.0346428995736863,
      "learning_rate": 9.901727638916284e-06,
      "loss": 0.5544,
      "step": 1235
    },
    {
      "epoch": 0.0666990448437753,
      "grad_norm": 1.0109933551344352,
      "learning_rate": 9.9015690119285e-06,
      "loss": 0.568,
      "step": 1236
    },
    {
      "epoch": 0.06675300847228968,
      "grad_norm": 1.077339599258623,
      "learning_rate": 9.901410258435172e-06,
      "loss": 0.5509,
      "step": 1237
    },
    {
      "epoch": 0.06680697210080405,
      "grad_norm": 1.0272052979550161,
      "learning_rate": 9.901251378440867e-06,
      "loss": 0.5267,
      "step": 1238
    },
    {
      "epoch": 0.06686093572931844,
      "grad_norm": 0.6809533432630285,
      "learning_rate": 9.90109237195015e-06,
      "loss": 0.2503,
      "step": 1239
    },
    {
      "epoch": 0.06691489935783282,
      "grad_norm": 1.196197606909191,
      "learning_rate": 9.900933238967594e-06,
      "loss": 0.5416,
      "step": 1240
    },
    {
      "epoch": 0.0669688629863472,
      "grad_norm": 0.9891962346539578,
      "learning_rate": 9.900773979497766e-06,
      "loss": 0.3866,
      "step": 1241
    },
    {
      "epoch": 0.06702282661486159,
      "grad_norm": 1.09445159942569,
      "learning_rate": 9.900614593545249e-06,
      "loss": 0.5561,
      "step": 1242
    },
    {
      "epoch": 0.06707679024337597,
      "grad_norm": 1.1695187770888185,
      "learning_rate": 9.900455081114623e-06,
      "loss": 0.5888,
      "step": 1243
    },
    {
      "epoch": 0.06713075387189034,
      "grad_norm": 1.0641905539721739,
      "learning_rate": 9.90029544221047e-06,
      "loss": 0.4525,
      "step": 1244
    },
    {
      "epoch": 0.06718471750040472,
      "grad_norm": 0.8808208480544224,
      "learning_rate": 9.900135676837382e-06,
      "loss": 0.4773,
      "step": 1245
    },
    {
      "epoch": 0.0672386811289191,
      "grad_norm": 1.0536364789447765,
      "learning_rate": 9.899975784999946e-06,
      "loss": 0.507,
      "step": 1246
    },
    {
      "epoch": 0.06729264475743349,
      "grad_norm": 1.049070033875011,
      "learning_rate": 9.899815766702762e-06,
      "loss": 0.522,
      "step": 1247
    },
    {
      "epoch": 0.06734660838594787,
      "grad_norm": 1.0865736584788848,
      "learning_rate": 9.899655621950426e-06,
      "loss": 0.8252,
      "step": 1248
    },
    {
      "epoch": 0.06740057201446226,
      "grad_norm": 1.1332885240116637,
      "learning_rate": 9.899495350747543e-06,
      "loss": 0.559,
      "step": 1249
    },
    {
      "epoch": 0.06745453564297664,
      "grad_norm": 0.998592122689888,
      "learning_rate": 9.899334953098718e-06,
      "loss": 0.5819,
      "step": 1250
    },
    {
      "epoch": 0.06750849927149101,
      "grad_norm": 0.9424013288908004,
      "learning_rate": 9.89917442900856e-06,
      "loss": 0.5392,
      "step": 1251
    },
    {
      "epoch": 0.0675624629000054,
      "grad_norm": 1.0570605717238244,
      "learning_rate": 9.899013778481685e-06,
      "loss": 0.4872,
      "step": 1252
    },
    {
      "epoch": 0.06761642652851978,
      "grad_norm": 1.3281893454625193,
      "learning_rate": 9.898853001522709e-06,
      "loss": 0.6775,
      "step": 1253
    },
    {
      "epoch": 0.06767039015703416,
      "grad_norm": 1.2270893852637632,
      "learning_rate": 9.898692098136252e-06,
      "loss": 0.6019,
      "step": 1254
    },
    {
      "epoch": 0.06772435378554854,
      "grad_norm": 0.9933541028848147,
      "learning_rate": 9.898531068326941e-06,
      "loss": 0.4468,
      "step": 1255
    },
    {
      "epoch": 0.06777831741406293,
      "grad_norm": 1.2519848674573097,
      "learning_rate": 9.898369912099401e-06,
      "loss": 0.6209,
      "step": 1256
    },
    {
      "epoch": 0.0678322810425773,
      "grad_norm": 1.022355968046646,
      "learning_rate": 9.898208629458267e-06,
      "loss": 0.4097,
      "step": 1257
    },
    {
      "epoch": 0.06788624467109168,
      "grad_norm": 1.3561032280506533,
      "learning_rate": 9.898047220408174e-06,
      "loss": 0.4646,
      "step": 1258
    },
    {
      "epoch": 0.06794020829960606,
      "grad_norm": 1.5410274124922365,
      "learning_rate": 9.897885684953758e-06,
      "loss": 0.5176,
      "step": 1259
    },
    {
      "epoch": 0.06799417192812045,
      "grad_norm": 1.0242001978910307,
      "learning_rate": 9.897724023099663e-06,
      "loss": 0.511,
      "step": 1260
    },
    {
      "epoch": 0.06804813555663483,
      "grad_norm": 0.808023448397742,
      "learning_rate": 9.897562234850538e-06,
      "loss": 0.3101,
      "step": 1261
    },
    {
      "epoch": 0.06810209918514921,
      "grad_norm": 1.2318290141400206,
      "learning_rate": 9.89740032021103e-06,
      "loss": 0.7173,
      "step": 1262
    },
    {
      "epoch": 0.06815606281366358,
      "grad_norm": 1.5815542248607608,
      "learning_rate": 9.897238279185793e-06,
      "loss": 0.7235,
      "step": 1263
    },
    {
      "epoch": 0.06821002644217797,
      "grad_norm": 1.0770135176453377,
      "learning_rate": 9.897076111779487e-06,
      "loss": 0.4103,
      "step": 1264
    },
    {
      "epoch": 0.06826399007069235,
      "grad_norm": 0.9576938412338435,
      "learning_rate": 9.896913817996769e-06,
      "loss": 0.4226,
      "step": 1265
    },
    {
      "epoch": 0.06831795369920673,
      "grad_norm": 1.3209514863119294,
      "learning_rate": 9.896751397842304e-06,
      "loss": 0.6787,
      "step": 1266
    },
    {
      "epoch": 0.06837191732772112,
      "grad_norm": 1.0564972787844833,
      "learning_rate": 9.896588851320763e-06,
      "loss": 0.5264,
      "step": 1267
    },
    {
      "epoch": 0.0684258809562355,
      "grad_norm": 0.8508933352273291,
      "learning_rate": 9.896426178436815e-06,
      "loss": 0.45,
      "step": 1268
    },
    {
      "epoch": 0.06847984458474989,
      "grad_norm": 0.9507977363069192,
      "learning_rate": 9.896263379195137e-06,
      "loss": 0.4189,
      "step": 1269
    },
    {
      "epoch": 0.06853380821326426,
      "grad_norm": 0.8244573927480834,
      "learning_rate": 9.896100453600407e-06,
      "loss": 0.4949,
      "step": 1270
    },
    {
      "epoch": 0.06858777184177864,
      "grad_norm": 1.064085097786398,
      "learning_rate": 9.895937401657309e-06,
      "loss": 0.489,
      "step": 1271
    },
    {
      "epoch": 0.06864173547029302,
      "grad_norm": 1.0549572332879322,
      "learning_rate": 9.895774223370526e-06,
      "loss": 0.5162,
      "step": 1272
    },
    {
      "epoch": 0.0686956990988074,
      "grad_norm": 1.120855272304418,
      "learning_rate": 9.895610918744752e-06,
      "loss": 0.4999,
      "step": 1273
    },
    {
      "epoch": 0.06874966272732179,
      "grad_norm": 1.1686633508697848,
      "learning_rate": 9.895447487784679e-06,
      "loss": 0.6084,
      "step": 1274
    },
    {
      "epoch": 0.06880362635583617,
      "grad_norm": 1.0405055309000513,
      "learning_rate": 9.895283930495003e-06,
      "loss": 0.4964,
      "step": 1275
    },
    {
      "epoch": 0.06885758998435054,
      "grad_norm": 1.2960881877758363,
      "learning_rate": 9.895120246880427e-06,
      "loss": 0.506,
      "step": 1276
    },
    {
      "epoch": 0.06891155361286493,
      "grad_norm": 1.0686435472564517,
      "learning_rate": 9.894956436945653e-06,
      "loss": 0.5458,
      "step": 1277
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 1.0905883560152692,
      "learning_rate": 9.89479250069539e-06,
      "loss": 0.4494,
      "step": 1278
    },
    {
      "epoch": 0.06901948086989369,
      "grad_norm": 1.0044385912776697,
      "learning_rate": 9.894628438134351e-06,
      "loss": 0.4121,
      "step": 1279
    },
    {
      "epoch": 0.06907344449840808,
      "grad_norm": 1.388799715247707,
      "learning_rate": 9.89446424926725e-06,
      "loss": 0.6794,
      "step": 1280
    },
    {
      "epoch": 0.06912740812692246,
      "grad_norm": 1.2259726622717861,
      "learning_rate": 9.894299934098805e-06,
      "loss": 0.5546,
      "step": 1281
    },
    {
      "epoch": 0.06918137175543683,
      "grad_norm": 1.090031038346463,
      "learning_rate": 9.894135492633742e-06,
      "loss": 0.491,
      "step": 1282
    },
    {
      "epoch": 0.06923533538395121,
      "grad_norm": 1.1025694813959253,
      "learning_rate": 9.893970924876783e-06,
      "loss": 0.4197,
      "step": 1283
    },
    {
      "epoch": 0.0692892990124656,
      "grad_norm": 1.1436354480458757,
      "learning_rate": 9.89380623083266e-06,
      "loss": 0.7405,
      "step": 1284
    },
    {
      "epoch": 0.06934326264097998,
      "grad_norm": 1.137637109686756,
      "learning_rate": 9.893641410506107e-06,
      "loss": 0.4212,
      "step": 1285
    },
    {
      "epoch": 0.06939722626949436,
      "grad_norm": 1.3106430304823857,
      "learning_rate": 9.89347646390186e-06,
      "loss": 0.541,
      "step": 1286
    },
    {
      "epoch": 0.06945118989800875,
      "grad_norm": 1.438737980772488,
      "learning_rate": 9.89331139102466e-06,
      "loss": 0.523,
      "step": 1287
    },
    {
      "epoch": 0.06950515352652312,
      "grad_norm": 1.0443079539185955,
      "learning_rate": 9.893146191879254e-06,
      "loss": 0.4721,
      "step": 1288
    },
    {
      "epoch": 0.0695591171550375,
      "grad_norm": 1.083102361040646,
      "learning_rate": 9.892980866470383e-06,
      "loss": 0.4348,
      "step": 1289
    },
    {
      "epoch": 0.06961308078355188,
      "grad_norm": 1.0899998628729872,
      "learning_rate": 9.892815414802807e-06,
      "loss": 0.5119,
      "step": 1290
    },
    {
      "epoch": 0.06966704441206627,
      "grad_norm": 1.061828024703141,
      "learning_rate": 9.892649836881276e-06,
      "loss": 0.4667,
      "step": 1291
    },
    {
      "epoch": 0.06972100804058065,
      "grad_norm": 0.9373877736100461,
      "learning_rate": 9.892484132710549e-06,
      "loss": 0.4201,
      "step": 1292
    },
    {
      "epoch": 0.06977497166909503,
      "grad_norm": 1.0883307848605492,
      "learning_rate": 9.892318302295392e-06,
      "loss": 0.5018,
      "step": 1293
    },
    {
      "epoch": 0.06982893529760942,
      "grad_norm": 0.9985812999678902,
      "learning_rate": 9.892152345640568e-06,
      "loss": 0.4986,
      "step": 1294
    },
    {
      "epoch": 0.06988289892612379,
      "grad_norm": 1.0702767801569917,
      "learning_rate": 9.891986262750846e-06,
      "loss": 0.4875,
      "step": 1295
    },
    {
      "epoch": 0.06993686255463817,
      "grad_norm": 1.3495093881975602,
      "learning_rate": 9.891820053631003e-06,
      "loss": 0.8506,
      "step": 1296
    },
    {
      "epoch": 0.06999082618315255,
      "grad_norm": 1.2873538896620391,
      "learning_rate": 9.891653718285815e-06,
      "loss": 0.5292,
      "step": 1297
    },
    {
      "epoch": 0.07004478981166694,
      "grad_norm": 1.0949458463462969,
      "learning_rate": 9.891487256720058e-06,
      "loss": 0.4176,
      "step": 1298
    },
    {
      "epoch": 0.07009875344018132,
      "grad_norm": 1.0192077829341488,
      "learning_rate": 9.891320668938523e-06,
      "loss": 0.5131,
      "step": 1299
    },
    {
      "epoch": 0.0701527170686957,
      "grad_norm": 1.217215663047037,
      "learning_rate": 9.891153954945994e-06,
      "loss": 0.5815,
      "step": 1300
    },
    {
      "epoch": 0.07020668069721007,
      "grad_norm": 1.0709972056787778,
      "learning_rate": 9.890987114747262e-06,
      "loss": 0.6182,
      "step": 1301
    },
    {
      "epoch": 0.07026064432572446,
      "grad_norm": 1.0689849256355781,
      "learning_rate": 9.890820148347126e-06,
      "loss": 0.4765,
      "step": 1302
    },
    {
      "epoch": 0.07031460795423884,
      "grad_norm": 1.2774043846790137,
      "learning_rate": 9.89065305575038e-06,
      "loss": 0.6815,
      "step": 1303
    },
    {
      "epoch": 0.07036857158275323,
      "grad_norm": 0.8352990073885136,
      "learning_rate": 9.89048583696183e-06,
      "loss": 0.408,
      "step": 1304
    },
    {
      "epoch": 0.07042253521126761,
      "grad_norm": 1.0086871424985875,
      "learning_rate": 9.890318491986282e-06,
      "loss": 0.4671,
      "step": 1305
    },
    {
      "epoch": 0.07047649883978199,
      "grad_norm": 0.8656607341025481,
      "learning_rate": 9.890151020828542e-06,
      "loss": 0.3313,
      "step": 1306
    },
    {
      "epoch": 0.07053046246829636,
      "grad_norm": 1.0708131139642576,
      "learning_rate": 9.889983423493428e-06,
      "loss": 0.6434,
      "step": 1307
    },
    {
      "epoch": 0.07058442609681075,
      "grad_norm": 1.0731101149836713,
      "learning_rate": 9.889815699985755e-06,
      "loss": 0.6455,
      "step": 1308
    },
    {
      "epoch": 0.07063838972532513,
      "grad_norm": 1.1210721798942302,
      "learning_rate": 9.88964785031034e-06,
      "loss": 0.6771,
      "step": 1309
    },
    {
      "epoch": 0.07069235335383951,
      "grad_norm": 1.170011729357879,
      "learning_rate": 9.889479874472012e-06,
      "loss": 0.5006,
      "step": 1310
    },
    {
      "epoch": 0.0707463169823539,
      "grad_norm": 1.1552290898873112,
      "learning_rate": 9.889311772475597e-06,
      "loss": 0.5545,
      "step": 1311
    },
    {
      "epoch": 0.07080028061086828,
      "grad_norm": 1.3156333314460178,
      "learning_rate": 9.889143544325926e-06,
      "loss": 0.601,
      "step": 1312
    },
    {
      "epoch": 0.07085424423938265,
      "grad_norm": 1.1998178584109462,
      "learning_rate": 9.888975190027837e-06,
      "loss": 0.7065,
      "step": 1313
    },
    {
      "epoch": 0.07090820786789703,
      "grad_norm": 1.1722387328498052,
      "learning_rate": 9.888806709586163e-06,
      "loss": 0.6057,
      "step": 1314
    },
    {
      "epoch": 0.07096217149641142,
      "grad_norm": 0.953568028644597,
      "learning_rate": 9.888638103005753e-06,
      "loss": 0.5316,
      "step": 1315
    },
    {
      "epoch": 0.0710161351249258,
      "grad_norm": 0.9464147533806352,
      "learning_rate": 9.888469370291448e-06,
      "loss": 0.4641,
      "step": 1316
    },
    {
      "epoch": 0.07107009875344018,
      "grad_norm": 1.220563680429435,
      "learning_rate": 9.888300511448099e-06,
      "loss": 0.6344,
      "step": 1317
    },
    {
      "epoch": 0.07112406238195457,
      "grad_norm": 0.9093703810615675,
      "learning_rate": 9.88813152648056e-06,
      "loss": 0.4236,
      "step": 1318
    },
    {
      "epoch": 0.07117802601046895,
      "grad_norm": 0.9583875339446923,
      "learning_rate": 9.887962415393685e-06,
      "loss": 0.466,
      "step": 1319
    },
    {
      "epoch": 0.07123198963898332,
      "grad_norm": 1.032828488882854,
      "learning_rate": 9.887793178192338e-06,
      "loss": 0.5059,
      "step": 1320
    },
    {
      "epoch": 0.0712859532674977,
      "grad_norm": 1.227250122945056,
      "learning_rate": 9.887623814881382e-06,
      "loss": 0.5929,
      "step": 1321
    },
    {
      "epoch": 0.07133991689601209,
      "grad_norm": 1.20230556777309,
      "learning_rate": 9.887454325465683e-06,
      "loss": 0.5909,
      "step": 1322
    },
    {
      "epoch": 0.07139388052452647,
      "grad_norm": 1.472206886931071,
      "learning_rate": 9.887284709950113e-06,
      "loss": 0.6516,
      "step": 1323
    },
    {
      "epoch": 0.07144784415304085,
      "grad_norm": 1.019024353052126,
      "learning_rate": 9.88711496833955e-06,
      "loss": 0.4596,
      "step": 1324
    },
    {
      "epoch": 0.07150180778155524,
      "grad_norm": 1.1060622738391286,
      "learning_rate": 9.886945100638868e-06,
      "loss": 0.4561,
      "step": 1325
    },
    {
      "epoch": 0.07155577141006961,
      "grad_norm": 1.164882065672562,
      "learning_rate": 9.88677510685295e-06,
      "loss": 0.421,
      "step": 1326
    },
    {
      "epoch": 0.07160973503858399,
      "grad_norm": 1.0693042439763076,
      "learning_rate": 9.886604986986685e-06,
      "loss": 0.5095,
      "step": 1327
    },
    {
      "epoch": 0.07166369866709837,
      "grad_norm": 1.3070559497267162,
      "learning_rate": 9.886434741044962e-06,
      "loss": 0.6779,
      "step": 1328
    },
    {
      "epoch": 0.07171766229561276,
      "grad_norm": 1.2486835781892738,
      "learning_rate": 9.886264369032669e-06,
      "loss": 0.6945,
      "step": 1329
    },
    {
      "epoch": 0.07177162592412714,
      "grad_norm": 1.007144743360955,
      "learning_rate": 9.886093870954707e-06,
      "loss": 0.4336,
      "step": 1330
    },
    {
      "epoch": 0.07182558955264153,
      "grad_norm": 1.277452180819719,
      "learning_rate": 9.885923246815974e-06,
      "loss": 0.6454,
      "step": 1331
    },
    {
      "epoch": 0.0718795531811559,
      "grad_norm": 1.0270566050532783,
      "learning_rate": 9.885752496621378e-06,
      "loss": 0.4727,
      "step": 1332
    },
    {
      "epoch": 0.07193351680967028,
      "grad_norm": 1.6895702394247003,
      "learning_rate": 9.885581620375823e-06,
      "loss": 0.4697,
      "step": 1333
    },
    {
      "epoch": 0.07198748043818466,
      "grad_norm": 1.142009293051695,
      "learning_rate": 9.885410618084218e-06,
      "loss": 0.4569,
      "step": 1334
    },
    {
      "epoch": 0.07204144406669905,
      "grad_norm": 1.024199754538245,
      "learning_rate": 9.885239489751485e-06,
      "loss": 0.5457,
      "step": 1335
    },
    {
      "epoch": 0.07209540769521343,
      "grad_norm": 1.0721261665930917,
      "learning_rate": 9.885068235382535e-06,
      "loss": 0.7341,
      "step": 1336
    },
    {
      "epoch": 0.07214937132372781,
      "grad_norm": 0.9571357893767014,
      "learning_rate": 9.884896854982295e-06,
      "loss": 0.4335,
      "step": 1337
    },
    {
      "epoch": 0.07220333495224218,
      "grad_norm": 1.0376508475427155,
      "learning_rate": 9.88472534855569e-06,
      "loss": 0.5813,
      "step": 1338
    },
    {
      "epoch": 0.07225729858075657,
      "grad_norm": 1.1844641911457416,
      "learning_rate": 9.884553716107646e-06,
      "loss": 0.5705,
      "step": 1339
    },
    {
      "epoch": 0.07231126220927095,
      "grad_norm": 1.22886962162792,
      "learning_rate": 9.884381957643096e-06,
      "loss": 0.5847,
      "step": 1340
    },
    {
      "epoch": 0.07236522583778533,
      "grad_norm": 1.1919558046026268,
      "learning_rate": 9.884210073166982e-06,
      "loss": 0.5849,
      "step": 1341
    },
    {
      "epoch": 0.07241918946629972,
      "grad_norm": 1.2204931291727565,
      "learning_rate": 9.884038062684239e-06,
      "loss": 0.5792,
      "step": 1342
    },
    {
      "epoch": 0.0724731530948141,
      "grad_norm": 0.8214518194672884,
      "learning_rate": 9.883865926199812e-06,
      "loss": 0.4572,
      "step": 1343
    },
    {
      "epoch": 0.07252711672332848,
      "grad_norm": 1.025309764037712,
      "learning_rate": 9.883693663718648e-06,
      "loss": 0.5484,
      "step": 1344
    },
    {
      "epoch": 0.07258108035184285,
      "grad_norm": 1.1530513234583,
      "learning_rate": 9.8835212752457e-06,
      "loss": 0.6541,
      "step": 1345
    },
    {
      "epoch": 0.07263504398035724,
      "grad_norm": 1.1307589246480472,
      "learning_rate": 9.883348760785922e-06,
      "loss": 0.5462,
      "step": 1346
    },
    {
      "epoch": 0.07268900760887162,
      "grad_norm": 1.2814630177058313,
      "learning_rate": 9.88317612034427e-06,
      "loss": 0.6591,
      "step": 1347
    },
    {
      "epoch": 0.072742971237386,
      "grad_norm": 1.0184193120221992,
      "learning_rate": 9.883003353925709e-06,
      "loss": 0.4958,
      "step": 1348
    },
    {
      "epoch": 0.07279693486590039,
      "grad_norm": 1.0652729315050362,
      "learning_rate": 9.882830461535202e-06,
      "loss": 0.4572,
      "step": 1349
    },
    {
      "epoch": 0.07285089849441477,
      "grad_norm": 0.8468286121757141,
      "learning_rate": 9.88265744317772e-06,
      "loss": 0.3354,
      "step": 1350
    },
    {
      "epoch": 0.07290486212292914,
      "grad_norm": 1.122813364663986,
      "learning_rate": 9.882484298858232e-06,
      "loss": 0.5723,
      "step": 1351
    },
    {
      "epoch": 0.07295882575144352,
      "grad_norm": 0.8808273744649087,
      "learning_rate": 9.88231102858172e-06,
      "loss": 0.3491,
      "step": 1352
    },
    {
      "epoch": 0.07301278937995791,
      "grad_norm": 1.100059421113247,
      "learning_rate": 9.88213763235316e-06,
      "loss": 0.5995,
      "step": 1353
    },
    {
      "epoch": 0.07306675300847229,
      "grad_norm": 1.0491102575485138,
      "learning_rate": 9.881964110177537e-06,
      "loss": 0.4412,
      "step": 1354
    },
    {
      "epoch": 0.07312071663698667,
      "grad_norm": 1.025894748978225,
      "learning_rate": 9.881790462059836e-06,
      "loss": 0.4851,
      "step": 1355
    },
    {
      "epoch": 0.07317468026550106,
      "grad_norm": 1.0081592749449473,
      "learning_rate": 9.881616688005052e-06,
      "loss": 0.6059,
      "step": 1356
    },
    {
      "epoch": 0.07322864389401543,
      "grad_norm": 1.3253708635397425,
      "learning_rate": 9.881442788018176e-06,
      "loss": 0.4249,
      "step": 1357
    },
    {
      "epoch": 0.07328260752252981,
      "grad_norm": 1.0151580121964023,
      "learning_rate": 9.881268762104207e-06,
      "loss": 0.4598,
      "step": 1358
    },
    {
      "epoch": 0.0733365711510442,
      "grad_norm": 1.2406743002990386,
      "learning_rate": 9.881094610268146e-06,
      "loss": 0.5223,
      "step": 1359
    },
    {
      "epoch": 0.07339053477955858,
      "grad_norm": 1.3562233124710563,
      "learning_rate": 9.880920332515001e-06,
      "loss": 0.6148,
      "step": 1360
    },
    {
      "epoch": 0.07344449840807296,
      "grad_norm": 1.0973597451727162,
      "learning_rate": 9.880745928849777e-06,
      "loss": 0.48,
      "step": 1361
    },
    {
      "epoch": 0.07349846203658734,
      "grad_norm": 0.9897650962413922,
      "learning_rate": 9.880571399277491e-06,
      "loss": 0.5059,
      "step": 1362
    },
    {
      "epoch": 0.07355242566510173,
      "grad_norm": 1.1310911829283203,
      "learning_rate": 9.880396743803155e-06,
      "loss": 0.4973,
      "step": 1363
    },
    {
      "epoch": 0.0736063892936161,
      "grad_norm": 1.131130288405654,
      "learning_rate": 9.880221962431793e-06,
      "loss": 0.5682,
      "step": 1364
    },
    {
      "epoch": 0.07366035292213048,
      "grad_norm": 0.9595916027242318,
      "learning_rate": 9.880047055168422e-06,
      "loss": 0.4736,
      "step": 1365
    },
    {
      "epoch": 0.07371431655064486,
      "grad_norm": 0.9826388128317063,
      "learning_rate": 9.879872022018076e-06,
      "loss": 0.4291,
      "step": 1366
    },
    {
      "epoch": 0.07376828017915925,
      "grad_norm": 0.8814962570745393,
      "learning_rate": 9.87969686298578e-06,
      "loss": 0.3894,
      "step": 1367
    },
    {
      "epoch": 0.07382224380767363,
      "grad_norm": 1.410852632537319,
      "learning_rate": 9.879521578076571e-06,
      "loss": 0.7539,
      "step": 1368
    },
    {
      "epoch": 0.07387620743618802,
      "grad_norm": 1.2642213593429998,
      "learning_rate": 9.87934616729549e-06,
      "loss": 0.6153,
      "step": 1369
    },
    {
      "epoch": 0.07393017106470239,
      "grad_norm": 1.0651239703821658,
      "learning_rate": 9.879170630647573e-06,
      "loss": 0.5788,
      "step": 1370
    },
    {
      "epoch": 0.07398413469321677,
      "grad_norm": 1.2538256270353412,
      "learning_rate": 9.878994968137867e-06,
      "loss": 0.6711,
      "step": 1371
    },
    {
      "epoch": 0.07403809832173115,
      "grad_norm": 1.3448755346794439,
      "learning_rate": 9.878819179771421e-06,
      "loss": 0.7628,
      "step": 1372
    },
    {
      "epoch": 0.07409206195024554,
      "grad_norm": 1.0957245802380502,
      "learning_rate": 9.878643265553289e-06,
      "loss": 0.4997,
      "step": 1373
    },
    {
      "epoch": 0.07414602557875992,
      "grad_norm": 1.0756860387728229,
      "learning_rate": 9.878467225488524e-06,
      "loss": 0.5338,
      "step": 1374
    },
    {
      "epoch": 0.0741999892072743,
      "grad_norm": 1.3450133524875882,
      "learning_rate": 9.878291059582185e-06,
      "loss": 0.6876,
      "step": 1375
    },
    {
      "epoch": 0.07425395283578867,
      "grad_norm": 1.0086571871272343,
      "learning_rate": 9.878114767839339e-06,
      "loss": 0.5276,
      "step": 1376
    },
    {
      "epoch": 0.07430791646430306,
      "grad_norm": 1.2376565219085747,
      "learning_rate": 9.87793835026505e-06,
      "loss": 0.4949,
      "step": 1377
    },
    {
      "epoch": 0.07436188009281744,
      "grad_norm": 0.9668329278997928,
      "learning_rate": 9.87776180686439e-06,
      "loss": 0.4691,
      "step": 1378
    },
    {
      "epoch": 0.07441584372133182,
      "grad_norm": 0.9407464409393771,
      "learning_rate": 9.877585137642431e-06,
      "loss": 0.3982,
      "step": 1379
    },
    {
      "epoch": 0.0744698073498462,
      "grad_norm": 0.9930611373186313,
      "learning_rate": 9.877408342604254e-06,
      "loss": 0.3368,
      "step": 1380
    },
    {
      "epoch": 0.07452377097836059,
      "grad_norm": 1.1122083373628897,
      "learning_rate": 9.877231421754939e-06,
      "loss": 0.5944,
      "step": 1381
    },
    {
      "epoch": 0.07457773460687496,
      "grad_norm": 1.3592230765430233,
      "learning_rate": 9.877054375099565e-06,
      "loss": 0.4754,
      "step": 1382
    },
    {
      "epoch": 0.07463169823538934,
      "grad_norm": 1.231853515584301,
      "learning_rate": 9.87687720264323e-06,
      "loss": 0.5812,
      "step": 1383
    },
    {
      "epoch": 0.07468566186390373,
      "grad_norm": 1.0627488331679096,
      "learning_rate": 9.876699904391019e-06,
      "loss": 0.4706,
      "step": 1384
    },
    {
      "epoch": 0.07473962549241811,
      "grad_norm": 1.2249040447217507,
      "learning_rate": 9.876522480348034e-06,
      "loss": 0.622,
      "step": 1385
    },
    {
      "epoch": 0.0747935891209325,
      "grad_norm": 1.0469375712052675,
      "learning_rate": 9.876344930519368e-06,
      "loss": 0.5364,
      "step": 1386
    },
    {
      "epoch": 0.07484755274944688,
      "grad_norm": 1.0517305301550146,
      "learning_rate": 9.876167254910127e-06,
      "loss": 0.6213,
      "step": 1387
    },
    {
      "epoch": 0.07490151637796126,
      "grad_norm": 1.1405961334518535,
      "learning_rate": 9.87598945352542e-06,
      "loss": 0.4842,
      "step": 1388
    },
    {
      "epoch": 0.07495548000647563,
      "grad_norm": 1.0798106593091088,
      "learning_rate": 9.875811526370349e-06,
      "loss": 0.5571,
      "step": 1389
    },
    {
      "epoch": 0.07500944363499001,
      "grad_norm": 0.9299544024296335,
      "learning_rate": 9.875633473450038e-06,
      "loss": 0.4723,
      "step": 1390
    },
    {
      "epoch": 0.0750634072635044,
      "grad_norm": 1.0266914149986648,
      "learning_rate": 9.875455294769598e-06,
      "loss": 0.5307,
      "step": 1391
    },
    {
      "epoch": 0.07511737089201878,
      "grad_norm": 0.8777907072326566,
      "learning_rate": 9.87527699033415e-06,
      "loss": 0.3752,
      "step": 1392
    },
    {
      "epoch": 0.07517133452053316,
      "grad_norm": 1.0296573583662345,
      "learning_rate": 9.875098560148823e-06,
      "loss": 0.6315,
      "step": 1393
    },
    {
      "epoch": 0.07522529814904755,
      "grad_norm": 1.1710635386858284,
      "learning_rate": 9.874920004218741e-06,
      "loss": 0.5093,
      "step": 1394
    },
    {
      "epoch": 0.07527926177756192,
      "grad_norm": 1.3032163605375449,
      "learning_rate": 9.87474132254904e-06,
      "loss": 0.6582,
      "step": 1395
    },
    {
      "epoch": 0.0753332254060763,
      "grad_norm": 1.0891684767103764,
      "learning_rate": 9.87456251514485e-06,
      "loss": 0.5775,
      "step": 1396
    },
    {
      "epoch": 0.07538718903459068,
      "grad_norm": 1.1123695657048627,
      "learning_rate": 9.874383582011314e-06,
      "loss": 0.5737,
      "step": 1397
    },
    {
      "epoch": 0.07544115266310507,
      "grad_norm": 1.1437853422300601,
      "learning_rate": 9.874204523153576e-06,
      "loss": 0.5532,
      "step": 1398
    },
    {
      "epoch": 0.07549511629161945,
      "grad_norm": 1.221134660243962,
      "learning_rate": 9.874025338576778e-06,
      "loss": 0.5409,
      "step": 1399
    },
    {
      "epoch": 0.07554907992013384,
      "grad_norm": 1.2284744208021552,
      "learning_rate": 9.873846028286073e-06,
      "loss": 0.6292,
      "step": 1400
    },
    {
      "epoch": 0.0756030435486482,
      "grad_norm": 1.1128867258560233,
      "learning_rate": 9.873666592286613e-06,
      "loss": 0.4633,
      "step": 1401
    },
    {
      "epoch": 0.07565700717716259,
      "grad_norm": 1.0580769400216088,
      "learning_rate": 9.873487030583557e-06,
      "loss": 0.472,
      "step": 1402
    },
    {
      "epoch": 0.07571097080567697,
      "grad_norm": 1.1255150435270471,
      "learning_rate": 9.873307343182065e-06,
      "loss": 0.5454,
      "step": 1403
    },
    {
      "epoch": 0.07576493443419136,
      "grad_norm": 1.1817209302522385,
      "learning_rate": 9.8731275300873e-06,
      "loss": 0.5515,
      "step": 1404
    },
    {
      "epoch": 0.07581889806270574,
      "grad_norm": 0.9668033613228058,
      "learning_rate": 9.87294759130443e-06,
      "loss": 0.4567,
      "step": 1405
    },
    {
      "epoch": 0.07587286169122012,
      "grad_norm": 1.367242744903834,
      "learning_rate": 9.87276752683863e-06,
      "loss": 0.7216,
      "step": 1406
    },
    {
      "epoch": 0.07592682531973449,
      "grad_norm": 0.9028418658167503,
      "learning_rate": 9.87258733669507e-06,
      "loss": 0.4066,
      "step": 1407
    },
    {
      "epoch": 0.07598078894824888,
      "grad_norm": 1.0117044010494998,
      "learning_rate": 9.872407020878935e-06,
      "loss": 0.497,
      "step": 1408
    },
    {
      "epoch": 0.07603475257676326,
      "grad_norm": 1.0982062937724875,
      "learning_rate": 9.872226579395403e-06,
      "loss": 0.4744,
      "step": 1409
    },
    {
      "epoch": 0.07608871620527764,
      "grad_norm": 1.4153922946353177,
      "learning_rate": 9.87204601224966e-06,
      "loss": 0.5274,
      "step": 1410
    },
    {
      "epoch": 0.07614267983379203,
      "grad_norm": 1.2327376197321842,
      "learning_rate": 9.871865319446898e-06,
      "loss": 0.8642,
      "step": 1411
    },
    {
      "epoch": 0.07619664346230641,
      "grad_norm": 1.083491103854235,
      "learning_rate": 9.871684500992311e-06,
      "loss": 0.4422,
      "step": 1412
    },
    {
      "epoch": 0.0762506070908208,
      "grad_norm": 1.160377295489073,
      "learning_rate": 9.871503556891091e-06,
      "loss": 0.5039,
      "step": 1413
    },
    {
      "epoch": 0.07630457071933516,
      "grad_norm": 0.930025103853537,
      "learning_rate": 9.871322487148443e-06,
      "loss": 0.5091,
      "step": 1414
    },
    {
      "epoch": 0.07635853434784955,
      "grad_norm": 1.115305167417295,
      "learning_rate": 9.87114129176957e-06,
      "loss": 0.5114,
      "step": 1415
    },
    {
      "epoch": 0.07641249797636393,
      "grad_norm": 1.0751915996349761,
      "learning_rate": 9.87095997075968e-06,
      "loss": 0.4455,
      "step": 1416
    },
    {
      "epoch": 0.07646646160487831,
      "grad_norm": 0.7236439217019938,
      "learning_rate": 9.870778524123984e-06,
      "loss": 0.3525,
      "step": 1417
    },
    {
      "epoch": 0.0765204252333927,
      "grad_norm": 1.1651640949299422,
      "learning_rate": 9.870596951867695e-06,
      "loss": 0.6346,
      "step": 1418
    },
    {
      "epoch": 0.07657438886190708,
      "grad_norm": 1.1150536829259694,
      "learning_rate": 9.870415253996035e-06,
      "loss": 0.4391,
      "step": 1419
    },
    {
      "epoch": 0.07662835249042145,
      "grad_norm": 0.898592522230811,
      "learning_rate": 9.870233430514224e-06,
      "loss": 0.427,
      "step": 1420
    },
    {
      "epoch": 0.07668231611893583,
      "grad_norm": 0.9342003491093377,
      "learning_rate": 9.87005148142749e-06,
      "loss": 0.4328,
      "step": 1421
    },
    {
      "epoch": 0.07673627974745022,
      "grad_norm": 1.2016248332588375,
      "learning_rate": 9.86986940674106e-06,
      "loss": 0.5819,
      "step": 1422
    },
    {
      "epoch": 0.0767902433759646,
      "grad_norm": 1.191696133549967,
      "learning_rate": 9.869687206460167e-06,
      "loss": 0.5651,
      "step": 1423
    },
    {
      "epoch": 0.07684420700447898,
      "grad_norm": 1.0476512389761121,
      "learning_rate": 9.869504880590048e-06,
      "loss": 0.5231,
      "step": 1424
    },
    {
      "epoch": 0.07689817063299337,
      "grad_norm": 1.3408752448458274,
      "learning_rate": 9.869322429135945e-06,
      "loss": 0.5502,
      "step": 1425
    },
    {
      "epoch": 0.07695213426150774,
      "grad_norm": 0.9776976465401988,
      "learning_rate": 9.8691398521031e-06,
      "loss": 0.5854,
      "step": 1426
    },
    {
      "epoch": 0.07700609789002212,
      "grad_norm": 1.0394366221764808,
      "learning_rate": 9.86895714949676e-06,
      "loss": 0.3746,
      "step": 1427
    },
    {
      "epoch": 0.0770600615185365,
      "grad_norm": 1.1964174509006205,
      "learning_rate": 9.868774321322179e-06,
      "loss": 0.5364,
      "step": 1428
    },
    {
      "epoch": 0.07711402514705089,
      "grad_norm": 0.9831195754750282,
      "learning_rate": 9.868591367584609e-06,
      "loss": 0.4765,
      "step": 1429
    },
    {
      "epoch": 0.07716798877556527,
      "grad_norm": 0.9867185352555774,
      "learning_rate": 9.868408288289309e-06,
      "loss": 0.5155,
      "step": 1430
    },
    {
      "epoch": 0.07722195240407966,
      "grad_norm": 0.8251961683724085,
      "learning_rate": 9.86822508344154e-06,
      "loss": 0.4613,
      "step": 1431
    },
    {
      "epoch": 0.07727591603259402,
      "grad_norm": 1.0345335763712429,
      "learning_rate": 9.86804175304657e-06,
      "loss": 0.4671,
      "step": 1432
    },
    {
      "epoch": 0.07732987966110841,
      "grad_norm": 0.9341757938183023,
      "learning_rate": 9.867858297109663e-06,
      "loss": 0.4752,
      "step": 1433
    },
    {
      "epoch": 0.07738384328962279,
      "grad_norm": 1.160587148930441,
      "learning_rate": 9.867674715636098e-06,
      "loss": 0.491,
      "step": 1434
    },
    {
      "epoch": 0.07743780691813718,
      "grad_norm": 1.0836427319372677,
      "learning_rate": 9.867491008631148e-06,
      "loss": 0.434,
      "step": 1435
    },
    {
      "epoch": 0.07749177054665156,
      "grad_norm": 1.0559453818687794,
      "learning_rate": 9.867307176100093e-06,
      "loss": 0.507,
      "step": 1436
    },
    {
      "epoch": 0.07754573417516594,
      "grad_norm": 1.2529640389859145,
      "learning_rate": 9.867123218048217e-06,
      "loss": 0.596,
      "step": 1437
    },
    {
      "epoch": 0.07759969780368033,
      "grad_norm": 1.0323130023185692,
      "learning_rate": 9.866939134480809e-06,
      "loss": 0.4515,
      "step": 1438
    },
    {
      "epoch": 0.0776536614321947,
      "grad_norm": 0.9039489075926891,
      "learning_rate": 9.866754925403155e-06,
      "loss": 0.4464,
      "step": 1439
    },
    {
      "epoch": 0.07770762506070908,
      "grad_norm": 0.9353282870534332,
      "learning_rate": 9.866570590820552e-06,
      "loss": 0.4157,
      "step": 1440
    },
    {
      "epoch": 0.07776158868922346,
      "grad_norm": 1.1436730301977664,
      "learning_rate": 9.8663861307383e-06,
      "loss": 0.4728,
      "step": 1441
    },
    {
      "epoch": 0.07781555231773785,
      "grad_norm": 1.205298017870878,
      "learning_rate": 9.866201545161697e-06,
      "loss": 0.6542,
      "step": 1442
    },
    {
      "epoch": 0.07786951594625223,
      "grad_norm": 1.2026611316019373,
      "learning_rate": 9.866016834096051e-06,
      "loss": 0.5678,
      "step": 1443
    },
    {
      "epoch": 0.07792347957476661,
      "grad_norm": 1.2245263511227187,
      "learning_rate": 9.865831997546671e-06,
      "loss": 0.6857,
      "step": 1444
    },
    {
      "epoch": 0.07797744320328098,
      "grad_norm": 1.2355437471738135,
      "learning_rate": 9.865647035518864e-06,
      "loss": 0.689,
      "step": 1445
    },
    {
      "epoch": 0.07803140683179537,
      "grad_norm": 1.2277745201727839,
      "learning_rate": 9.865461948017954e-06,
      "loss": 0.5433,
      "step": 1446
    },
    {
      "epoch": 0.07808537046030975,
      "grad_norm": 1.186068147842787,
      "learning_rate": 9.865276735049257e-06,
      "loss": 0.543,
      "step": 1447
    },
    {
      "epoch": 0.07813933408882413,
      "grad_norm": 1.1277496567219691,
      "learning_rate": 9.865091396618094e-06,
      "loss": 0.5915,
      "step": 1448
    },
    {
      "epoch": 0.07819329771733852,
      "grad_norm": 1.2531167963137233,
      "learning_rate": 9.864905932729793e-06,
      "loss": 0.6918,
      "step": 1449
    },
    {
      "epoch": 0.0782472613458529,
      "grad_norm": 1.0245060141670959,
      "learning_rate": 9.864720343389685e-06,
      "loss": 0.4987,
      "step": 1450
    },
    {
      "epoch": 0.07830122497436727,
      "grad_norm": 1.1206027549310897,
      "learning_rate": 9.864534628603104e-06,
      "loss": 0.5332,
      "step": 1451
    },
    {
      "epoch": 0.07835518860288165,
      "grad_norm": 1.0691952766003536,
      "learning_rate": 9.86434878837539e-06,
      "loss": 0.5743,
      "step": 1452
    },
    {
      "epoch": 0.07840915223139604,
      "grad_norm": 1.2801306728083763,
      "learning_rate": 9.86416282271188e-06,
      "loss": 0.591,
      "step": 1453
    },
    {
      "epoch": 0.07846311585991042,
      "grad_norm": 0.8782631910423312,
      "learning_rate": 9.863976731617922e-06,
      "loss": 0.5008,
      "step": 1454
    },
    {
      "epoch": 0.0785170794884248,
      "grad_norm": 1.5073536444498166,
      "learning_rate": 9.863790515098863e-06,
      "loss": 0.6005,
      "step": 1455
    },
    {
      "epoch": 0.07857104311693919,
      "grad_norm": 0.9509012874588001,
      "learning_rate": 9.863604173160054e-06,
      "loss": 0.395,
      "step": 1456
    },
    {
      "epoch": 0.07862500674545357,
      "grad_norm": 0.992038118489577,
      "learning_rate": 9.863417705806854e-06,
      "loss": 0.4284,
      "step": 1457
    },
    {
      "epoch": 0.07867897037396794,
      "grad_norm": 0.9145494405449114,
      "learning_rate": 9.86323111304462e-06,
      "loss": 0.3893,
      "step": 1458
    },
    {
      "epoch": 0.07873293400248232,
      "grad_norm": 0.8142662446402474,
      "learning_rate": 9.863044394878712e-06,
      "loss": 0.427,
      "step": 1459
    },
    {
      "epoch": 0.07878689763099671,
      "grad_norm": 0.9737968481837906,
      "learning_rate": 9.862857551314503e-06,
      "loss": 0.4625,
      "step": 1460
    },
    {
      "epoch": 0.07884086125951109,
      "grad_norm": 1.05966516755037,
      "learning_rate": 9.862670582357358e-06,
      "loss": 0.4998,
      "step": 1461
    },
    {
      "epoch": 0.07889482488802547,
      "grad_norm": 1.237120469588189,
      "learning_rate": 9.862483488012653e-06,
      "loss": 0.6764,
      "step": 1462
    },
    {
      "epoch": 0.07894878851653986,
      "grad_norm": 1.2112161188587183,
      "learning_rate": 9.862296268285764e-06,
      "loss": 0.5315,
      "step": 1463
    },
    {
      "epoch": 0.07900275214505423,
      "grad_norm": 0.9788383146859465,
      "learning_rate": 9.862108923182073e-06,
      "loss": 0.51,
      "step": 1464
    },
    {
      "epoch": 0.07905671577356861,
      "grad_norm": 1.2412483063590298,
      "learning_rate": 9.861921452706964e-06,
      "loss": 0.5567,
      "step": 1465
    },
    {
      "epoch": 0.079110679402083,
      "grad_norm": 1.265692075720178,
      "learning_rate": 9.861733856865825e-06,
      "loss": 0.5197,
      "step": 1466
    },
    {
      "epoch": 0.07916464303059738,
      "grad_norm": 1.206037172647099,
      "learning_rate": 9.861546135664047e-06,
      "loss": 0.5381,
      "step": 1467
    },
    {
      "epoch": 0.07921860665911176,
      "grad_norm": 0.6921957406159033,
      "learning_rate": 9.861358289107028e-06,
      "loss": 0.3112,
      "step": 1468
    },
    {
      "epoch": 0.07927257028762615,
      "grad_norm": 1.08216762357684,
      "learning_rate": 9.861170317200164e-06,
      "loss": 0.4626,
      "step": 1469
    },
    {
      "epoch": 0.07932653391614052,
      "grad_norm": 1.2075259022925382,
      "learning_rate": 9.860982219948856e-06,
      "loss": 0.5026,
      "step": 1470
    },
    {
      "epoch": 0.0793804975446549,
      "grad_norm": 1.2446012654277088,
      "learning_rate": 9.860793997358515e-06,
      "loss": 0.4357,
      "step": 1471
    },
    {
      "epoch": 0.07943446117316928,
      "grad_norm": 0.9854572307082262,
      "learning_rate": 9.860605649434547e-06,
      "loss": 0.3771,
      "step": 1472
    },
    {
      "epoch": 0.07948842480168367,
      "grad_norm": 1.1193974224597572,
      "learning_rate": 9.860417176182368e-06,
      "loss": 0.651,
      "step": 1473
    },
    {
      "epoch": 0.07954238843019805,
      "grad_norm": 1.0916857462805774,
      "learning_rate": 9.860228577607392e-06,
      "loss": 0.486,
      "step": 1474
    },
    {
      "epoch": 0.07959635205871243,
      "grad_norm": 1.2815082610516313,
      "learning_rate": 9.860039853715043e-06,
      "loss": 0.5562,
      "step": 1475
    },
    {
      "epoch": 0.0796503156872268,
      "grad_norm": 1.1651273032580152,
      "learning_rate": 9.859851004510741e-06,
      "loss": 0.6071,
      "step": 1476
    },
    {
      "epoch": 0.07970427931574119,
      "grad_norm": 1.0956430590425383,
      "learning_rate": 9.859662029999917e-06,
      "loss": 0.4904,
      "step": 1477
    },
    {
      "epoch": 0.07975824294425557,
      "grad_norm": 0.9472708300017778,
      "learning_rate": 9.859472930188002e-06,
      "loss": 0.4158,
      "step": 1478
    },
    {
      "epoch": 0.07981220657276995,
      "grad_norm": 1.0615490832922667,
      "learning_rate": 9.85928370508043e-06,
      "loss": 0.5284,
      "step": 1479
    },
    {
      "epoch": 0.07986617020128434,
      "grad_norm": 1.0237575294712045,
      "learning_rate": 9.859094354682636e-06,
      "loss": 0.4561,
      "step": 1480
    },
    {
      "epoch": 0.07992013382979872,
      "grad_norm": 1.1954722898403902,
      "learning_rate": 9.858904879000068e-06,
      "loss": 0.6271,
      "step": 1481
    },
    {
      "epoch": 0.0799740974583131,
      "grad_norm": 0.9246314156408526,
      "learning_rate": 9.85871527803817e-06,
      "loss": 0.3628,
      "step": 1482
    },
    {
      "epoch": 0.08002806108682747,
      "grad_norm": 1.1535051229274245,
      "learning_rate": 9.85852555180239e-06,
      "loss": 0.5989,
      "step": 1483
    },
    {
      "epoch": 0.08008202471534186,
      "grad_norm": 1.0107034488730695,
      "learning_rate": 9.858335700298183e-06,
      "loss": 0.4543,
      "step": 1484
    },
    {
      "epoch": 0.08013598834385624,
      "grad_norm": 1.0933195350337488,
      "learning_rate": 9.858145723531003e-06,
      "loss": 0.5148,
      "step": 1485
    },
    {
      "epoch": 0.08018995197237062,
      "grad_norm": 1.2817658042371736,
      "learning_rate": 9.857955621506312e-06,
      "loss": 0.5432,
      "step": 1486
    },
    {
      "epoch": 0.08024391560088501,
      "grad_norm": 1.1124611024398199,
      "learning_rate": 9.857765394229573e-06,
      "loss": 0.5106,
      "step": 1487
    },
    {
      "epoch": 0.08029787922939939,
      "grad_norm": 1.090046228360641,
      "learning_rate": 9.857575041706253e-06,
      "loss": 0.5011,
      "step": 1488
    },
    {
      "epoch": 0.08035184285791376,
      "grad_norm": 0.9952267743209943,
      "learning_rate": 9.857384563941822e-06,
      "loss": 0.4861,
      "step": 1489
    },
    {
      "epoch": 0.08040580648642814,
      "grad_norm": 1.1146891235017142,
      "learning_rate": 9.857193960941757e-06,
      "loss": 0.5959,
      "step": 1490
    },
    {
      "epoch": 0.08045977011494253,
      "grad_norm": 0.9742156618232539,
      "learning_rate": 9.857003232711535e-06,
      "loss": 0.4979,
      "step": 1491
    },
    {
      "epoch": 0.08051373374345691,
      "grad_norm": 1.2594637760359897,
      "learning_rate": 9.856812379256637e-06,
      "loss": 0.5615,
      "step": 1492
    },
    {
      "epoch": 0.0805676973719713,
      "grad_norm": 1.024781234168918,
      "learning_rate": 9.85662140058255e-06,
      "loss": 0.481,
      "step": 1493
    },
    {
      "epoch": 0.08062166100048568,
      "grad_norm": 1.1244158600069667,
      "learning_rate": 9.856430296694762e-06,
      "loss": 0.6746,
      "step": 1494
    },
    {
      "epoch": 0.08067562462900005,
      "grad_norm": 1.1677602531457634,
      "learning_rate": 9.856239067598766e-06,
      "loss": 0.6207,
      "step": 1495
    },
    {
      "epoch": 0.08072958825751443,
      "grad_norm": 1.0726082817996365,
      "learning_rate": 9.856047713300055e-06,
      "loss": 0.5905,
      "step": 1496
    },
    {
      "epoch": 0.08078355188602881,
      "grad_norm": 1.2114995837635878,
      "learning_rate": 9.855856233804134e-06,
      "loss": 0.5323,
      "step": 1497
    },
    {
      "epoch": 0.0808375155145432,
      "grad_norm": 1.16000317153991,
      "learning_rate": 9.855664629116502e-06,
      "loss": 0.6126,
      "step": 1498
    },
    {
      "epoch": 0.08089147914305758,
      "grad_norm": 1.0492276381754382,
      "learning_rate": 9.855472899242668e-06,
      "loss": 0.4683,
      "step": 1499
    },
    {
      "epoch": 0.08094544277157197,
      "grad_norm": 1.1663522259630936,
      "learning_rate": 9.855281044188142e-06,
      "loss": 0.6185,
      "step": 1500
    },
    {
      "epoch": 0.08094544277157197,
      "eval_loss": 0.6146706938743591,
      "eval_runtime": 164.3247,
      "eval_samples_per_second": 20.928,
      "eval_steps_per_second": 0.876,
      "step": 1500
    },
    {
      "epoch": 0.08099940640008634,
      "grad_norm": 0.9798140207891484,
      "learning_rate": 9.855089063958438e-06,
      "loss": 0.5013,
      "step": 1501
    },
    {
      "epoch": 0.08105337002860072,
      "grad_norm": 1.2582772065254066,
      "learning_rate": 9.854896958559075e-06,
      "loss": 0.5821,
      "step": 1502
    },
    {
      "epoch": 0.0811073336571151,
      "grad_norm": 1.2520645620663025,
      "learning_rate": 9.854704727995572e-06,
      "loss": 0.453,
      "step": 1503
    },
    {
      "epoch": 0.08116129728562949,
      "grad_norm": 1.2157887560299088,
      "learning_rate": 9.854512372273453e-06,
      "loss": 0.5886,
      "step": 1504
    },
    {
      "epoch": 0.08121526091414387,
      "grad_norm": 1.1146875166182604,
      "learning_rate": 9.854319891398251e-06,
      "loss": 0.4069,
      "step": 1505
    },
    {
      "epoch": 0.08126922454265825,
      "grad_norm": 1.4202003350134589,
      "learning_rate": 9.854127285375495e-06,
      "loss": 0.6642,
      "step": 1506
    },
    {
      "epoch": 0.08132318817117264,
      "grad_norm": 1.3259499177089509,
      "learning_rate": 9.853934554210722e-06,
      "loss": 0.5552,
      "step": 1507
    },
    {
      "epoch": 0.081377151799687,
      "grad_norm": 1.1060516920685122,
      "learning_rate": 9.853741697909469e-06,
      "loss": 0.5061,
      "step": 1508
    },
    {
      "epoch": 0.08143111542820139,
      "grad_norm": 1.0258360711480912,
      "learning_rate": 9.853548716477282e-06,
      "loss": 0.6109,
      "step": 1509
    },
    {
      "epoch": 0.08148507905671577,
      "grad_norm": 1.2655816403044782,
      "learning_rate": 9.853355609919704e-06,
      "loss": 0.5719,
      "step": 1510
    },
    {
      "epoch": 0.08153904268523016,
      "grad_norm": 1.1343385881609889,
      "learning_rate": 9.853162378242288e-06,
      "loss": 0.5053,
      "step": 1511
    },
    {
      "epoch": 0.08159300631374454,
      "grad_norm": 1.0328876512863703,
      "learning_rate": 9.852969021450588e-06,
      "loss": 0.5016,
      "step": 1512
    },
    {
      "epoch": 0.08164696994225892,
      "grad_norm": 0.878102636532819,
      "learning_rate": 9.852775539550158e-06,
      "loss": 0.4054,
      "step": 1513
    },
    {
      "epoch": 0.0817009335707733,
      "grad_norm": 1.0514038860420063,
      "learning_rate": 9.852581932546564e-06,
      "loss": 0.3864,
      "step": 1514
    },
    {
      "epoch": 0.08175489719928768,
      "grad_norm": 1.0362282012694082,
      "learning_rate": 9.852388200445364e-06,
      "loss": 0.4653,
      "step": 1515
    },
    {
      "epoch": 0.08180886082780206,
      "grad_norm": 1.0185811148611448,
      "learning_rate": 9.85219434325213e-06,
      "loss": 0.424,
      "step": 1516
    },
    {
      "epoch": 0.08186282445631644,
      "grad_norm": 1.1444059859700202,
      "learning_rate": 9.852000360972435e-06,
      "loss": 0.5455,
      "step": 1517
    },
    {
      "epoch": 0.08191678808483083,
      "grad_norm": 1.199478908214755,
      "learning_rate": 9.85180625361185e-06,
      "loss": 0.6757,
      "step": 1518
    },
    {
      "epoch": 0.08197075171334521,
      "grad_norm": 1.115159861386223,
      "learning_rate": 9.851612021175958e-06,
      "loss": 0.5406,
      "step": 1519
    },
    {
      "epoch": 0.08202471534185958,
      "grad_norm": 0.9312085379169391,
      "learning_rate": 9.851417663670338e-06,
      "loss": 0.4855,
      "step": 1520
    },
    {
      "epoch": 0.08207867897037396,
      "grad_norm": 1.0896347948181393,
      "learning_rate": 9.851223181100577e-06,
      "loss": 0.451,
      "step": 1521
    },
    {
      "epoch": 0.08213264259888835,
      "grad_norm": 1.0466600772463077,
      "learning_rate": 9.851028573472267e-06,
      "loss": 0.5355,
      "step": 1522
    },
    {
      "epoch": 0.08218660622740273,
      "grad_norm": 1.1596608974361473,
      "learning_rate": 9.850833840790996e-06,
      "loss": 0.4676,
      "step": 1523
    },
    {
      "epoch": 0.08224056985591711,
      "grad_norm": 1.0780443075418005,
      "learning_rate": 9.850638983062367e-06,
      "loss": 0.5934,
      "step": 1524
    },
    {
      "epoch": 0.0822945334844315,
      "grad_norm": 1.0496587614329123,
      "learning_rate": 9.850444000291977e-06,
      "loss": 0.4462,
      "step": 1525
    },
    {
      "epoch": 0.08234849711294587,
      "grad_norm": 0.9084719891997122,
      "learning_rate": 9.850248892485434e-06,
      "loss": 0.3869,
      "step": 1526
    },
    {
      "epoch": 0.08240246074146025,
      "grad_norm": 1.0413229017852348,
      "learning_rate": 9.850053659648338e-06,
      "loss": 0.5009,
      "step": 1527
    },
    {
      "epoch": 0.08245642436997463,
      "grad_norm": 1.1587227563029678,
      "learning_rate": 9.849858301786305e-06,
      "loss": 0.5335,
      "step": 1528
    },
    {
      "epoch": 0.08251038799848902,
      "grad_norm": 1.1680125356628455,
      "learning_rate": 9.84966281890495e-06,
      "loss": 0.573,
      "step": 1529
    },
    {
      "epoch": 0.0825643516270034,
      "grad_norm": 0.8598853255457625,
      "learning_rate": 9.84946721100989e-06,
      "loss": 0.2967,
      "step": 1530
    },
    {
      "epoch": 0.08261831525551779,
      "grad_norm": 0.9157247685416676,
      "learning_rate": 9.849271478106747e-06,
      "loss": 0.4403,
      "step": 1531
    },
    {
      "epoch": 0.08267227888403217,
      "grad_norm": 1.0124714028200428,
      "learning_rate": 9.849075620201148e-06,
      "loss": 0.4885,
      "step": 1532
    },
    {
      "epoch": 0.08272624251254654,
      "grad_norm": 0.9958019600890563,
      "learning_rate": 9.848879637298724e-06,
      "loss": 0.5238,
      "step": 1533
    },
    {
      "epoch": 0.08278020614106092,
      "grad_norm": 1.3366909708225285,
      "learning_rate": 9.8486835294051e-06,
      "loss": 0.6829,
      "step": 1534
    },
    {
      "epoch": 0.0828341697695753,
      "grad_norm": 0.9181762422893593,
      "learning_rate": 9.84848729652592e-06,
      "loss": 0.4989,
      "step": 1535
    },
    {
      "epoch": 0.08288813339808969,
      "grad_norm": 1.1826804758244975,
      "learning_rate": 9.848290938666821e-06,
      "loss": 0.5855,
      "step": 1536
    },
    {
      "epoch": 0.08294209702660407,
      "grad_norm": 1.0940651721566208,
      "learning_rate": 9.848094455833447e-06,
      "loss": 0.5753,
      "step": 1537
    },
    {
      "epoch": 0.08299606065511846,
      "grad_norm": 1.2331370009972131,
      "learning_rate": 9.847897848031444e-06,
      "loss": 0.656,
      "step": 1538
    },
    {
      "epoch": 0.08305002428363283,
      "grad_norm": 1.2762993941501477,
      "learning_rate": 9.847701115266464e-06,
      "loss": 0.7644,
      "step": 1539
    },
    {
      "epoch": 0.08310398791214721,
      "grad_norm": 0.9726644290765585,
      "learning_rate": 9.847504257544163e-06,
      "loss": 0.5011,
      "step": 1540
    },
    {
      "epoch": 0.08315795154066159,
      "grad_norm": 0.9584812503039151,
      "learning_rate": 9.847307274870193e-06,
      "loss": 0.4976,
      "step": 1541
    },
    {
      "epoch": 0.08321191516917598,
      "grad_norm": 1.1750148556944287,
      "learning_rate": 9.847110167250223e-06,
      "loss": 0.58,
      "step": 1542
    },
    {
      "epoch": 0.08326587879769036,
      "grad_norm": 1.222355056247674,
      "learning_rate": 9.84691293468991e-06,
      "loss": 0.6011,
      "step": 1543
    },
    {
      "epoch": 0.08331984242620474,
      "grad_norm": 1.1254245067873054,
      "learning_rate": 9.846715577194931e-06,
      "loss": 0.4862,
      "step": 1544
    },
    {
      "epoch": 0.08337380605471911,
      "grad_norm": 1.0455939198311845,
      "learning_rate": 9.846518094770951e-06,
      "loss": 0.4624,
      "step": 1545
    },
    {
      "epoch": 0.0834277696832335,
      "grad_norm": 1.228633884409697,
      "learning_rate": 9.846320487423653e-06,
      "loss": 0.5733,
      "step": 1546
    },
    {
      "epoch": 0.08348173331174788,
      "grad_norm": 0.7365752934107501,
      "learning_rate": 9.84612275515871e-06,
      "loss": 0.2979,
      "step": 1547
    },
    {
      "epoch": 0.08353569694026226,
      "grad_norm": 1.297037613261306,
      "learning_rate": 9.845924897981809e-06,
      "loss": 0.5991,
      "step": 1548
    },
    {
      "epoch": 0.08358966056877665,
      "grad_norm": 1.0307260290793485,
      "learning_rate": 9.845726915898634e-06,
      "loss": 0.438,
      "step": 1549
    },
    {
      "epoch": 0.08364362419729103,
      "grad_norm": 1.083529896053623,
      "learning_rate": 9.845528808914876e-06,
      "loss": 0.5046,
      "step": 1550
    },
    {
      "epoch": 0.08369758782580541,
      "grad_norm": 1.0815254116768536,
      "learning_rate": 9.845330577036229e-06,
      "loss": 0.515,
      "step": 1551
    },
    {
      "epoch": 0.08375155145431978,
      "grad_norm": 0.942989789903542,
      "learning_rate": 9.845132220268393e-06,
      "loss": 0.4579,
      "step": 1552
    },
    {
      "epoch": 0.08380551508283417,
      "grad_norm": 1.1382890688730474,
      "learning_rate": 9.844933738617063e-06,
      "loss": 0.4499,
      "step": 1553
    },
    {
      "epoch": 0.08385947871134855,
      "grad_norm": 0.8690473800277624,
      "learning_rate": 9.84473513208795e-06,
      "loss": 0.4089,
      "step": 1554
    },
    {
      "epoch": 0.08391344233986293,
      "grad_norm": 1.1505086404814973,
      "learning_rate": 9.84453640068676e-06,
      "loss": 0.5531,
      "step": 1555
    },
    {
      "epoch": 0.08396740596837732,
      "grad_norm": 1.0483287103181118,
      "learning_rate": 9.844337544419201e-06,
      "loss": 0.4522,
      "step": 1556
    },
    {
      "epoch": 0.0840213695968917,
      "grad_norm": 1.1589430216527066,
      "learning_rate": 9.844138563290993e-06,
      "loss": 0.6265,
      "step": 1557
    },
    {
      "epoch": 0.08407533322540607,
      "grad_norm": 1.262231200769264,
      "learning_rate": 9.843939457307853e-06,
      "loss": 0.6673,
      "step": 1558
    },
    {
      "epoch": 0.08412929685392045,
      "grad_norm": 0.8240812365831286,
      "learning_rate": 9.843740226475505e-06,
      "loss": 0.3645,
      "step": 1559
    },
    {
      "epoch": 0.08418326048243484,
      "grad_norm": 1.2755777293978097,
      "learning_rate": 9.843540870799671e-06,
      "loss": 0.5795,
      "step": 1560
    },
    {
      "epoch": 0.08423722411094922,
      "grad_norm": 1.1559880831674183,
      "learning_rate": 9.843341390286087e-06,
      "loss": 0.7272,
      "step": 1561
    },
    {
      "epoch": 0.0842911877394636,
      "grad_norm": 1.3779131757746121,
      "learning_rate": 9.84314178494048e-06,
      "loss": 0.6434,
      "step": 1562
    },
    {
      "epoch": 0.08434515136797799,
      "grad_norm": 1.346690560850989,
      "learning_rate": 9.84294205476859e-06,
      "loss": 0.7369,
      "step": 1563
    },
    {
      "epoch": 0.08439911499649236,
      "grad_norm": 1.2208802990487055,
      "learning_rate": 9.84274219977616e-06,
      "loss": 0.5447,
      "step": 1564
    },
    {
      "epoch": 0.08445307862500674,
      "grad_norm": 1.156660916184976,
      "learning_rate": 9.842542219968926e-06,
      "loss": 0.5326,
      "step": 1565
    },
    {
      "epoch": 0.08450704225352113,
      "grad_norm": 1.1089401599190514,
      "learning_rate": 9.842342115352647e-06,
      "loss": 0.5314,
      "step": 1566
    },
    {
      "epoch": 0.08456100588203551,
      "grad_norm": 0.9395095432669703,
      "learning_rate": 9.842141885933063e-06,
      "loss": 0.4176,
      "step": 1567
    },
    {
      "epoch": 0.08461496951054989,
      "grad_norm": 1.1210758235682878,
      "learning_rate": 9.841941531715936e-06,
      "loss": 0.5447,
      "step": 1568
    },
    {
      "epoch": 0.08466893313906428,
      "grad_norm": 1.2527745425116128,
      "learning_rate": 9.841741052707021e-06,
      "loss": 0.6707,
      "step": 1569
    },
    {
      "epoch": 0.08472289676757865,
      "grad_norm": 0.9981604489122831,
      "learning_rate": 9.841540448912083e-06,
      "loss": 0.4276,
      "step": 1570
    },
    {
      "epoch": 0.08477686039609303,
      "grad_norm": 1.0319580285317156,
      "learning_rate": 9.841339720336885e-06,
      "loss": 0.5513,
      "step": 1571
    },
    {
      "epoch": 0.08483082402460741,
      "grad_norm": 1.0105271934793305,
      "learning_rate": 9.841138866987196e-06,
      "loss": 0.5267,
      "step": 1572
    },
    {
      "epoch": 0.0848847876531218,
      "grad_norm": 0.9073355854440462,
      "learning_rate": 9.84093788886879e-06,
      "loss": 0.4137,
      "step": 1573
    },
    {
      "epoch": 0.08493875128163618,
      "grad_norm": 1.1546972256070236,
      "learning_rate": 9.840736785987443e-06,
      "loss": 0.515,
      "step": 1574
    },
    {
      "epoch": 0.08499271491015056,
      "grad_norm": 1.22179328513042,
      "learning_rate": 9.840535558348934e-06,
      "loss": 0.5654,
      "step": 1575
    },
    {
      "epoch": 0.08504667853866495,
      "grad_norm": 1.2223317208394844,
      "learning_rate": 9.840334205959047e-06,
      "loss": 0.5655,
      "step": 1576
    },
    {
      "epoch": 0.08510064216717932,
      "grad_norm": 1.1658447829794572,
      "learning_rate": 9.84013272882357e-06,
      "loss": 0.5585,
      "step": 1577
    },
    {
      "epoch": 0.0851546057956937,
      "grad_norm": 1.1713083436248977,
      "learning_rate": 9.839931126948294e-06,
      "loss": 0.5375,
      "step": 1578
    },
    {
      "epoch": 0.08520856942420808,
      "grad_norm": 1.0074625828735708,
      "learning_rate": 9.83972940033901e-06,
      "loss": 0.4389,
      "step": 1579
    },
    {
      "epoch": 0.08526253305272247,
      "grad_norm": 1.0141735897654298,
      "learning_rate": 9.839527549001522e-06,
      "loss": 0.3648,
      "step": 1580
    },
    {
      "epoch": 0.08531649668123685,
      "grad_norm": 1.086980057179789,
      "learning_rate": 9.839325572941624e-06,
      "loss": 0.4758,
      "step": 1581
    },
    {
      "epoch": 0.08537046030975123,
      "grad_norm": 1.0944083611542792,
      "learning_rate": 9.839123472165125e-06,
      "loss": 0.5433,
      "step": 1582
    },
    {
      "epoch": 0.0854244239382656,
      "grad_norm": 0.9484114757413327,
      "learning_rate": 9.838921246677833e-06,
      "loss": 0.529,
      "step": 1583
    },
    {
      "epoch": 0.08547838756677999,
      "grad_norm": 1.249388415303616,
      "learning_rate": 9.838718896485562e-06,
      "loss": 0.5792,
      "step": 1584
    },
    {
      "epoch": 0.08553235119529437,
      "grad_norm": 1.2069786134825093,
      "learning_rate": 9.838516421594123e-06,
      "loss": 0.611,
      "step": 1585
    },
    {
      "epoch": 0.08558631482380875,
      "grad_norm": 1.1367159383841587,
      "learning_rate": 9.838313822009342e-06,
      "loss": 0.5743,
      "step": 1586
    },
    {
      "epoch": 0.08564027845232314,
      "grad_norm": 1.0690548361561774,
      "learning_rate": 9.838111097737034e-06,
      "loss": 0.5715,
      "step": 1587
    },
    {
      "epoch": 0.08569424208083752,
      "grad_norm": 1.0807623232506285,
      "learning_rate": 9.837908248783033e-06,
      "loss": 0.6421,
      "step": 1588
    },
    {
      "epoch": 0.08574820570935189,
      "grad_norm": 1.0783708764592845,
      "learning_rate": 9.837705275153163e-06,
      "loss": 0.4799,
      "step": 1589
    },
    {
      "epoch": 0.08580216933786627,
      "grad_norm": 0.9061602416524287,
      "learning_rate": 9.837502176853263e-06,
      "loss": 0.51,
      "step": 1590
    },
    {
      "epoch": 0.08585613296638066,
      "grad_norm": 1.151846880949086,
      "learning_rate": 9.837298953889167e-06,
      "loss": 0.5705,
      "step": 1591
    },
    {
      "epoch": 0.08591009659489504,
      "grad_norm": 1.0876547236078546,
      "learning_rate": 9.837095606266716e-06,
      "loss": 0.57,
      "step": 1592
    },
    {
      "epoch": 0.08596406022340942,
      "grad_norm": 1.1138372740908353,
      "learning_rate": 9.836892133991756e-06,
      "loss": 0.5763,
      "step": 1593
    },
    {
      "epoch": 0.08601802385192381,
      "grad_norm": 0.852474862855587,
      "learning_rate": 9.836688537070133e-06,
      "loss": 0.4211,
      "step": 1594
    },
    {
      "epoch": 0.08607198748043818,
      "grad_norm": 1.152205613848411,
      "learning_rate": 9.8364848155077e-06,
      "loss": 0.436,
      "step": 1595
    },
    {
      "epoch": 0.08612595110895256,
      "grad_norm": 0.9890111297566225,
      "learning_rate": 9.836280969310311e-06,
      "loss": 0.4744,
      "step": 1596
    },
    {
      "epoch": 0.08617991473746694,
      "grad_norm": 1.1267327430720884,
      "learning_rate": 9.836076998483827e-06,
      "loss": 0.5306,
      "step": 1597
    },
    {
      "epoch": 0.08623387836598133,
      "grad_norm": 1.256551240729594,
      "learning_rate": 9.835872903034107e-06,
      "loss": 0.6612,
      "step": 1598
    },
    {
      "epoch": 0.08628784199449571,
      "grad_norm": 1.1490930821353862,
      "learning_rate": 9.83566868296702e-06,
      "loss": 0.5736,
      "step": 1599
    },
    {
      "epoch": 0.0863418056230101,
      "grad_norm": 0.8801770895212956,
      "learning_rate": 9.835464338288433e-06,
      "loss": 0.337,
      "step": 1600
    },
    {
      "epoch": 0.08639576925152448,
      "grad_norm": 1.1206498024505585,
      "learning_rate": 9.83525986900422e-06,
      "loss": 0.6459,
      "step": 1601
    },
    {
      "epoch": 0.08644973288003885,
      "grad_norm": 1.0631613107392668,
      "learning_rate": 9.83505527512026e-06,
      "loss": 0.5481,
      "step": 1602
    },
    {
      "epoch": 0.08650369650855323,
      "grad_norm": 0.787002373070203,
      "learning_rate": 9.834850556642429e-06,
      "loss": 0.365,
      "step": 1603
    },
    {
      "epoch": 0.08655766013706762,
      "grad_norm": 1.2890055639787825,
      "learning_rate": 9.834645713576615e-06,
      "loss": 0.6179,
      "step": 1604
    },
    {
      "epoch": 0.086611623765582,
      "grad_norm": 1.2914020715362726,
      "learning_rate": 9.8344407459287e-06,
      "loss": 0.667,
      "step": 1605
    },
    {
      "epoch": 0.08666558739409638,
      "grad_norm": 1.1843440352892518,
      "learning_rate": 9.834235653704583e-06,
      "loss": 0.4882,
      "step": 1606
    },
    {
      "epoch": 0.08671955102261077,
      "grad_norm": 0.9121622957994678,
      "learning_rate": 9.83403043691015e-06,
      "loss": 0.4114,
      "step": 1607
    },
    {
      "epoch": 0.08677351465112514,
      "grad_norm": 1.1316613787339864,
      "learning_rate": 9.833825095551304e-06,
      "loss": 0.5457,
      "step": 1608
    },
    {
      "epoch": 0.08682747827963952,
      "grad_norm": 1.3354579259543042,
      "learning_rate": 9.833619629633946e-06,
      "loss": 0.6108,
      "step": 1609
    },
    {
      "epoch": 0.0868814419081539,
      "grad_norm": 1.004378439975859,
      "learning_rate": 9.833414039163983e-06,
      "loss": 0.5821,
      "step": 1610
    },
    {
      "epoch": 0.08693540553666829,
      "grad_norm": 0.9257516759832259,
      "learning_rate": 9.833208324147318e-06,
      "loss": 0.4459,
      "step": 1611
    },
    {
      "epoch": 0.08698936916518267,
      "grad_norm": 1.3658895076135296,
      "learning_rate": 9.833002484589871e-06,
      "loss": 0.617,
      "step": 1612
    },
    {
      "epoch": 0.08704333279369705,
      "grad_norm": 1.3484458340651433,
      "learning_rate": 9.832796520497552e-06,
      "loss": 0.5209,
      "step": 1613
    },
    {
      "epoch": 0.08709729642221142,
      "grad_norm": 1.3788447896712182,
      "learning_rate": 9.832590431876285e-06,
      "loss": 0.7221,
      "step": 1614
    },
    {
      "epoch": 0.0871512600507258,
      "grad_norm": 1.0761047530954517,
      "learning_rate": 9.83238421873199e-06,
      "loss": 0.6012,
      "step": 1615
    },
    {
      "epoch": 0.08720522367924019,
      "grad_norm": 0.8347598100449255,
      "learning_rate": 9.832177881070594e-06,
      "loss": 0.289,
      "step": 1616
    },
    {
      "epoch": 0.08725918730775457,
      "grad_norm": 1.240941394104531,
      "learning_rate": 9.831971418898029e-06,
      "loss": 0.5491,
      "step": 1617
    },
    {
      "epoch": 0.08731315093626896,
      "grad_norm": 1.12464094020021,
      "learning_rate": 9.83176483222023e-06,
      "loss": 0.5414,
      "step": 1618
    },
    {
      "epoch": 0.08736711456478334,
      "grad_norm": 1.2160853220993917,
      "learning_rate": 9.831558121043132e-06,
      "loss": 0.5913,
      "step": 1619
    },
    {
      "epoch": 0.08742107819329771,
      "grad_norm": 1.1736920426391768,
      "learning_rate": 9.831351285372677e-06,
      "loss": 0.4621,
      "step": 1620
    },
    {
      "epoch": 0.0874750418218121,
      "grad_norm": 0.8947489008974683,
      "learning_rate": 9.831144325214807e-06,
      "loss": 0.4323,
      "step": 1621
    },
    {
      "epoch": 0.08752900545032648,
      "grad_norm": 0.8576290615011672,
      "learning_rate": 9.830937240575476e-06,
      "loss": 0.3433,
      "step": 1622
    },
    {
      "epoch": 0.08758296907884086,
      "grad_norm": 0.8202390883169676,
      "learning_rate": 9.830730031460631e-06,
      "loss": 0.3983,
      "step": 1623
    },
    {
      "epoch": 0.08763693270735524,
      "grad_norm": 1.0073211066754821,
      "learning_rate": 9.83052269787623e-06,
      "loss": 0.4526,
      "step": 1624
    },
    {
      "epoch": 0.08769089633586963,
      "grad_norm": 1.2100917935625481,
      "learning_rate": 9.83031523982823e-06,
      "loss": 0.5639,
      "step": 1625
    },
    {
      "epoch": 0.08774485996438401,
      "grad_norm": 1.0518192301087639,
      "learning_rate": 9.830107657322595e-06,
      "loss": 0.5762,
      "step": 1626
    },
    {
      "epoch": 0.08779882359289838,
      "grad_norm": 1.4607934577686827,
      "learning_rate": 9.829899950365291e-06,
      "loss": 0.7201,
      "step": 1627
    },
    {
      "epoch": 0.08785278722141276,
      "grad_norm": 1.1780400437548577,
      "learning_rate": 9.829692118962288e-06,
      "loss": 0.5046,
      "step": 1628
    },
    {
      "epoch": 0.08790675084992715,
      "grad_norm": 1.0305028798455964,
      "learning_rate": 9.829484163119558e-06,
      "loss": 0.555,
      "step": 1629
    },
    {
      "epoch": 0.08796071447844153,
      "grad_norm": 0.9483324996319525,
      "learning_rate": 9.829276082843077e-06,
      "loss": 0.3648,
      "step": 1630
    },
    {
      "epoch": 0.08801467810695592,
      "grad_norm": 1.0524614300174187,
      "learning_rate": 9.829067878138828e-06,
      "loss": 0.4713,
      "step": 1631
    },
    {
      "epoch": 0.0880686417354703,
      "grad_norm": 0.9599722735210796,
      "learning_rate": 9.828859549012795e-06,
      "loss": 0.4018,
      "step": 1632
    },
    {
      "epoch": 0.08812260536398467,
      "grad_norm": 1.0558785725193331,
      "learning_rate": 9.828651095470963e-06,
      "loss": 0.6398,
      "step": 1633
    },
    {
      "epoch": 0.08817656899249905,
      "grad_norm": 1.2490202671484132,
      "learning_rate": 9.828442517519325e-06,
      "loss": 0.6621,
      "step": 1634
    },
    {
      "epoch": 0.08823053262101344,
      "grad_norm": 0.9597501357336855,
      "learning_rate": 9.828233815163877e-06,
      "loss": 0.4455,
      "step": 1635
    },
    {
      "epoch": 0.08828449624952782,
      "grad_norm": 1.1436511229580923,
      "learning_rate": 9.828024988410614e-06,
      "loss": 0.5793,
      "step": 1636
    },
    {
      "epoch": 0.0883384598780422,
      "grad_norm": 1.1270627241489257,
      "learning_rate": 9.827816037265541e-06,
      "loss": 0.5021,
      "step": 1637
    },
    {
      "epoch": 0.08839242350655659,
      "grad_norm": 1.4219463515828699,
      "learning_rate": 9.827606961734661e-06,
      "loss": 0.735,
      "step": 1638
    },
    {
      "epoch": 0.08844638713507096,
      "grad_norm": 0.9714128992279043,
      "learning_rate": 9.827397761823984e-06,
      "loss": 0.3808,
      "step": 1639
    },
    {
      "epoch": 0.08850035076358534,
      "grad_norm": 1.4904570373380337,
      "learning_rate": 9.827188437539524e-06,
      "loss": 0.7883,
      "step": 1640
    },
    {
      "epoch": 0.08855431439209972,
      "grad_norm": 1.1838036727180905,
      "learning_rate": 9.826978988887297e-06,
      "loss": 0.5839,
      "step": 1641
    },
    {
      "epoch": 0.0886082780206141,
      "grad_norm": 1.1617384767629275,
      "learning_rate": 9.82676941587332e-06,
      "loss": 0.4723,
      "step": 1642
    },
    {
      "epoch": 0.08866224164912849,
      "grad_norm": 1.3683706335596868,
      "learning_rate": 9.826559718503617e-06,
      "loss": 0.5093,
      "step": 1643
    },
    {
      "epoch": 0.08871620527764287,
      "grad_norm": 1.2675392424392362,
      "learning_rate": 9.826349896784219e-06,
      "loss": 0.6712,
      "step": 1644
    },
    {
      "epoch": 0.08877016890615726,
      "grad_norm": 1.3289530387490138,
      "learning_rate": 9.826139950721151e-06,
      "loss": 0.6452,
      "step": 1645
    },
    {
      "epoch": 0.08882413253467163,
      "grad_norm": 1.2228801552837718,
      "learning_rate": 9.825929880320452e-06,
      "loss": 0.5753,
      "step": 1646
    },
    {
      "epoch": 0.08887809616318601,
      "grad_norm": 1.0343104073204008,
      "learning_rate": 9.825719685588156e-06,
      "loss": 0.6086,
      "step": 1647
    },
    {
      "epoch": 0.0889320597917004,
      "grad_norm": 1.1870806049082987,
      "learning_rate": 9.825509366530303e-06,
      "loss": 0.5365,
      "step": 1648
    },
    {
      "epoch": 0.08898602342021478,
      "grad_norm": 1.0046447938937908,
      "learning_rate": 9.825298923152944e-06,
      "loss": 0.5849,
      "step": 1649
    },
    {
      "epoch": 0.08903998704872916,
      "grad_norm": 0.9978182875348294,
      "learning_rate": 9.825088355462122e-06,
      "loss": 0.5223,
      "step": 1650
    },
    {
      "epoch": 0.08909395067724354,
      "grad_norm": 1.044125844081303,
      "learning_rate": 9.82487766346389e-06,
      "loss": 0.482,
      "step": 1651
    },
    {
      "epoch": 0.08914791430575791,
      "grad_norm": 1.225174758023279,
      "learning_rate": 9.824666847164305e-06,
      "loss": 0.6918,
      "step": 1652
    },
    {
      "epoch": 0.0892018779342723,
      "grad_norm": 1.06438626007324,
      "learning_rate": 9.824455906569423e-06,
      "loss": 0.5024,
      "step": 1653
    },
    {
      "epoch": 0.08925584156278668,
      "grad_norm": 1.245934185643498,
      "learning_rate": 9.824244841685312e-06,
      "loss": 0.6606,
      "step": 1654
    },
    {
      "epoch": 0.08930980519130106,
      "grad_norm": 1.1185088785196011,
      "learning_rate": 9.824033652518031e-06,
      "loss": 0.482,
      "step": 1655
    },
    {
      "epoch": 0.08936376881981545,
      "grad_norm": 1.0840397608763517,
      "learning_rate": 9.823822339073656e-06,
      "loss": 0.5517,
      "step": 1656
    },
    {
      "epoch": 0.08941773244832983,
      "grad_norm": 1.329610598092162,
      "learning_rate": 9.823610901358258e-06,
      "loss": 0.6059,
      "step": 1657
    },
    {
      "epoch": 0.0894716960768442,
      "grad_norm": 1.3229738100958877,
      "learning_rate": 9.823399339377913e-06,
      "loss": 0.591,
      "step": 1658
    },
    {
      "epoch": 0.08952565970535858,
      "grad_norm": 1.0775612463918058,
      "learning_rate": 9.823187653138704e-06,
      "loss": 0.5562,
      "step": 1659
    },
    {
      "epoch": 0.08957962333387297,
      "grad_norm": 0.9785007886055614,
      "learning_rate": 9.822975842646713e-06,
      "loss": 0.4405,
      "step": 1660
    },
    {
      "epoch": 0.08963358696238735,
      "grad_norm": 0.8586064196390926,
      "learning_rate": 9.822763907908028e-06,
      "loss": 0.3278,
      "step": 1661
    },
    {
      "epoch": 0.08968755059090174,
      "grad_norm": 0.9333122172315669,
      "learning_rate": 9.822551848928742e-06,
      "loss": 0.3912,
      "step": 1662
    },
    {
      "epoch": 0.08974151421941612,
      "grad_norm": 0.9869739468486504,
      "learning_rate": 9.822339665714946e-06,
      "loss": 0.4953,
      "step": 1663
    },
    {
      "epoch": 0.08979547784793049,
      "grad_norm": 1.1252225806440233,
      "learning_rate": 9.822127358272742e-06,
      "loss": 0.5658,
      "step": 1664
    },
    {
      "epoch": 0.08984944147644487,
      "grad_norm": 1.3715820146447948,
      "learning_rate": 9.82191492660823e-06,
      "loss": 0.6715,
      "step": 1665
    },
    {
      "epoch": 0.08990340510495926,
      "grad_norm": 1.2153187702701098,
      "learning_rate": 9.821702370727517e-06,
      "loss": 0.6509,
      "step": 1666
    },
    {
      "epoch": 0.08995736873347364,
      "grad_norm": 0.8146975489692445,
      "learning_rate": 9.82148969063671e-06,
      "loss": 0.3417,
      "step": 1667
    },
    {
      "epoch": 0.09001133236198802,
      "grad_norm": 1.25428516083054,
      "learning_rate": 9.821276886341921e-06,
      "loss": 0.652,
      "step": 1668
    },
    {
      "epoch": 0.0900652959905024,
      "grad_norm": 0.7877677891714017,
      "learning_rate": 9.82106395784927e-06,
      "loss": 0.2937,
      "step": 1669
    },
    {
      "epoch": 0.09011925961901679,
      "grad_norm": 1.1984374000287379,
      "learning_rate": 9.820850905164875e-06,
      "loss": 0.5934,
      "step": 1670
    },
    {
      "epoch": 0.09017322324753116,
      "grad_norm": 1.2228934987459703,
      "learning_rate": 9.820637728294859e-06,
      "loss": 0.5393,
      "step": 1671
    },
    {
      "epoch": 0.09022718687604554,
      "grad_norm": 1.2555563442989714,
      "learning_rate": 9.820424427245349e-06,
      "loss": 0.5592,
      "step": 1672
    },
    {
      "epoch": 0.09028115050455993,
      "grad_norm": 1.0774177619090028,
      "learning_rate": 9.820211002022475e-06,
      "loss": 0.6326,
      "step": 1673
    },
    {
      "epoch": 0.09033511413307431,
      "grad_norm": 0.924511641561153,
      "learning_rate": 9.819997452632372e-06,
      "loss": 0.3981,
      "step": 1674
    },
    {
      "epoch": 0.0903890777615887,
      "grad_norm": 1.08463739646336,
      "learning_rate": 9.819783779081176e-06,
      "loss": 0.4948,
      "step": 1675
    },
    {
      "epoch": 0.09044304139010308,
      "grad_norm": 1.2105462433132683,
      "learning_rate": 9.819569981375031e-06,
      "loss": 0.4794,
      "step": 1676
    },
    {
      "epoch": 0.09049700501861745,
      "grad_norm": 1.0756202429582895,
      "learning_rate": 9.81935605952008e-06,
      "loss": 0.5709,
      "step": 1677
    },
    {
      "epoch": 0.09055096864713183,
      "grad_norm": 0.8921964399721621,
      "learning_rate": 9.81914201352247e-06,
      "loss": 0.466,
      "step": 1678
    },
    {
      "epoch": 0.09060493227564621,
      "grad_norm": 1.069290093281084,
      "learning_rate": 9.818927843388353e-06,
      "loss": 0.59,
      "step": 1679
    },
    {
      "epoch": 0.0906588959041606,
      "grad_norm": 1.1030831464156623,
      "learning_rate": 9.81871354912389e-06,
      "loss": 0.5554,
      "step": 1680
    },
    {
      "epoch": 0.09071285953267498,
      "grad_norm": 1.0161444169948883,
      "learning_rate": 9.818499130735236e-06,
      "loss": 0.421,
      "step": 1681
    },
    {
      "epoch": 0.09076682316118936,
      "grad_norm": 1.0531489691120555,
      "learning_rate": 9.818284588228549e-06,
      "loss": 0.6346,
      "step": 1682
    },
    {
      "epoch": 0.09082078678970373,
      "grad_norm": 0.927684292184937,
      "learning_rate": 9.818069921610002e-06,
      "loss": 0.4583,
      "step": 1683
    },
    {
      "epoch": 0.09087475041821812,
      "grad_norm": 1.073151799884988,
      "learning_rate": 9.817855130885762e-06,
      "loss": 0.498,
      "step": 1684
    },
    {
      "epoch": 0.0909287140467325,
      "grad_norm": 1.0582736106515125,
      "learning_rate": 9.817640216062003e-06,
      "loss": 0.4456,
      "step": 1685
    },
    {
      "epoch": 0.09098267767524688,
      "grad_norm": 1.025772844411493,
      "learning_rate": 9.817425177144903e-06,
      "loss": 0.3691,
      "step": 1686
    },
    {
      "epoch": 0.09103664130376127,
      "grad_norm": 1.0891343214577762,
      "learning_rate": 9.81721001414064e-06,
      "loss": 0.4927,
      "step": 1687
    },
    {
      "epoch": 0.09109060493227565,
      "grad_norm": 0.9504154617191033,
      "learning_rate": 9.816994727055397e-06,
      "loss": 0.4328,
      "step": 1688
    },
    {
      "epoch": 0.09114456856079002,
      "grad_norm": 1.0943679321142035,
      "learning_rate": 9.816779315895366e-06,
      "loss": 0.4348,
      "step": 1689
    },
    {
      "epoch": 0.0911985321893044,
      "grad_norm": 1.1139241373970679,
      "learning_rate": 9.816563780666734e-06,
      "loss": 0.464,
      "step": 1690
    },
    {
      "epoch": 0.09125249581781879,
      "grad_norm": 1.137816870929604,
      "learning_rate": 9.816348121375697e-06,
      "loss": 0.4801,
      "step": 1691
    },
    {
      "epoch": 0.09130645944633317,
      "grad_norm": 1.1393581216763444,
      "learning_rate": 9.816132338028455e-06,
      "loss": 0.5652,
      "step": 1692
    },
    {
      "epoch": 0.09136042307484755,
      "grad_norm": 1.0317802011950623,
      "learning_rate": 9.815916430631209e-06,
      "loss": 0.5314,
      "step": 1693
    },
    {
      "epoch": 0.09141438670336194,
      "grad_norm": 1.029407606259921,
      "learning_rate": 9.815700399190162e-06,
      "loss": 0.4722,
      "step": 1694
    },
    {
      "epoch": 0.09146835033187632,
      "grad_norm": 1.140417592765902,
      "learning_rate": 9.815484243711525e-06,
      "loss": 0.4671,
      "step": 1695
    },
    {
      "epoch": 0.09152231396039069,
      "grad_norm": 1.178258194091602,
      "learning_rate": 9.815267964201511e-06,
      "loss": 0.4784,
      "step": 1696
    },
    {
      "epoch": 0.09157627758890508,
      "grad_norm": 1.2934630824706141,
      "learning_rate": 9.815051560666337e-06,
      "loss": 0.5439,
      "step": 1697
    },
    {
      "epoch": 0.09163024121741946,
      "grad_norm": 1.1753392952512376,
      "learning_rate": 9.81483503311222e-06,
      "loss": 0.587,
      "step": 1698
    },
    {
      "epoch": 0.09168420484593384,
      "grad_norm": 1.0874961980852107,
      "learning_rate": 9.814618381545385e-06,
      "loss": 0.5716,
      "step": 1699
    },
    {
      "epoch": 0.09173816847444823,
      "grad_norm": 1.1789266039726995,
      "learning_rate": 9.814401605972056e-06,
      "loss": 0.6019,
      "step": 1700
    },
    {
      "epoch": 0.09179213210296261,
      "grad_norm": 0.813923817098091,
      "learning_rate": 9.814184706398467e-06,
      "loss": 0.3539,
      "step": 1701
    },
    {
      "epoch": 0.09184609573147698,
      "grad_norm": 1.3132245363138273,
      "learning_rate": 9.813967682830851e-06,
      "loss": 0.5825,
      "step": 1702
    },
    {
      "epoch": 0.09190005935999136,
      "grad_norm": 1.0948781294439975,
      "learning_rate": 9.813750535275443e-06,
      "loss": 0.3776,
      "step": 1703
    },
    {
      "epoch": 0.09195402298850575,
      "grad_norm": 1.3664642869615802,
      "learning_rate": 9.813533263738486e-06,
      "loss": 0.7115,
      "step": 1704
    },
    {
      "epoch": 0.09200798661702013,
      "grad_norm": 1.079512338213839,
      "learning_rate": 9.813315868226226e-06,
      "loss": 0.6042,
      "step": 1705
    },
    {
      "epoch": 0.09206195024553451,
      "grad_norm": 1.2216389626565454,
      "learning_rate": 9.813098348744908e-06,
      "loss": 0.6331,
      "step": 1706
    },
    {
      "epoch": 0.0921159138740489,
      "grad_norm": 0.9557344063718698,
      "learning_rate": 9.812880705300788e-06,
      "loss": 0.5468,
      "step": 1707
    },
    {
      "epoch": 0.09216987750256327,
      "grad_norm": 1.0944854324097417,
      "learning_rate": 9.812662937900115e-06,
      "loss": 0.6354,
      "step": 1708
    },
    {
      "epoch": 0.09222384113107765,
      "grad_norm": 1.021843422721648,
      "learning_rate": 9.812445046549156e-06,
      "loss": 0.5327,
      "step": 1709
    },
    {
      "epoch": 0.09227780475959203,
      "grad_norm": 0.9678859433516638,
      "learning_rate": 9.812227031254165e-06,
      "loss": 0.4132,
      "step": 1710
    },
    {
      "epoch": 0.09233176838810642,
      "grad_norm": 0.8988119562999242,
      "learning_rate": 9.812008892021414e-06,
      "loss": 0.3923,
      "step": 1711
    },
    {
      "epoch": 0.0923857320166208,
      "grad_norm": 1.1646460322989367,
      "learning_rate": 9.81179062885717e-06,
      "loss": 0.4816,
      "step": 1712
    },
    {
      "epoch": 0.09243969564513518,
      "grad_norm": 1.1162727452001793,
      "learning_rate": 9.811572241767706e-06,
      "loss": 0.5885,
      "step": 1713
    },
    {
      "epoch": 0.09249365927364955,
      "grad_norm": 1.0996732397513724,
      "learning_rate": 9.8113537307593e-06,
      "loss": 0.4349,
      "step": 1714
    },
    {
      "epoch": 0.09254762290216394,
      "grad_norm": 0.999500261909367,
      "learning_rate": 9.811135095838231e-06,
      "loss": 0.4081,
      "step": 1715
    },
    {
      "epoch": 0.09260158653067832,
      "grad_norm": 0.8847185037684455,
      "learning_rate": 9.810916337010783e-06,
      "loss": 0.4378,
      "step": 1716
    },
    {
      "epoch": 0.0926555501591927,
      "grad_norm": 1.0378386908800896,
      "learning_rate": 9.810697454283244e-06,
      "loss": 0.4723,
      "step": 1717
    },
    {
      "epoch": 0.09270951378770709,
      "grad_norm": 1.0533433981723221,
      "learning_rate": 9.810478447661905e-06,
      "loss": 0.4544,
      "step": 1718
    },
    {
      "epoch": 0.09276347741622147,
      "grad_norm": 0.9962979316321182,
      "learning_rate": 9.81025931715306e-06,
      "loss": 0.4935,
      "step": 1719
    },
    {
      "epoch": 0.09281744104473585,
      "grad_norm": 1.093176694811763,
      "learning_rate": 9.810040062763007e-06,
      "loss": 0.5344,
      "step": 1720
    },
    {
      "epoch": 0.09287140467325022,
      "grad_norm": 1.0962559073863924,
      "learning_rate": 9.809820684498047e-06,
      "loss": 0.5263,
      "step": 1721
    },
    {
      "epoch": 0.09292536830176461,
      "grad_norm": 1.1360375754944454,
      "learning_rate": 9.809601182364484e-06,
      "loss": 0.5281,
      "step": 1722
    },
    {
      "epoch": 0.09297933193027899,
      "grad_norm": 0.9765547124984795,
      "learning_rate": 9.809381556368632e-06,
      "loss": 0.3518,
      "step": 1723
    },
    {
      "epoch": 0.09303329555879337,
      "grad_norm": 1.0779994570172466,
      "learning_rate": 9.809161806516797e-06,
      "loss": 0.4334,
      "step": 1724
    },
    {
      "epoch": 0.09308725918730776,
      "grad_norm": 1.003737619795514,
      "learning_rate": 9.8089419328153e-06,
      "loss": 0.4246,
      "step": 1725
    },
    {
      "epoch": 0.09314122281582214,
      "grad_norm": 1.2274996988254032,
      "learning_rate": 9.808721935270457e-06,
      "loss": 0.6325,
      "step": 1726
    },
    {
      "epoch": 0.09319518644433651,
      "grad_norm": 1.0822542723169275,
      "learning_rate": 9.80850181388859e-06,
      "loss": 0.5053,
      "step": 1727
    },
    {
      "epoch": 0.0932491500728509,
      "grad_norm": 1.120562931452158,
      "learning_rate": 9.808281568676032e-06,
      "loss": 0.4591,
      "step": 1728
    },
    {
      "epoch": 0.09330311370136528,
      "grad_norm": 1.0021283275466906,
      "learning_rate": 9.808061199639105e-06,
      "loss": 0.5276,
      "step": 1729
    },
    {
      "epoch": 0.09335707732987966,
      "grad_norm": 1.2798593177691098,
      "learning_rate": 9.807840706784146e-06,
      "loss": 0.5406,
      "step": 1730
    },
    {
      "epoch": 0.09341104095839405,
      "grad_norm": 0.8934191089316476,
      "learning_rate": 9.807620090117494e-06,
      "loss": 0.4272,
      "step": 1731
    },
    {
      "epoch": 0.09346500458690843,
      "grad_norm": 1.1813582621611651,
      "learning_rate": 9.807399349645486e-06,
      "loss": 0.6701,
      "step": 1732
    },
    {
      "epoch": 0.0935189682154228,
      "grad_norm": 1.1172849750615685,
      "learning_rate": 9.80717848537447e-06,
      "loss": 0.4658,
      "step": 1733
    },
    {
      "epoch": 0.09357293184393718,
      "grad_norm": 1.1189582016106767,
      "learning_rate": 9.806957497310792e-06,
      "loss": 0.4672,
      "step": 1734
    },
    {
      "epoch": 0.09362689547245157,
      "grad_norm": 1.038513728809273,
      "learning_rate": 9.806736385460804e-06,
      "loss": 0.4507,
      "step": 1735
    },
    {
      "epoch": 0.09368085910096595,
      "grad_norm": 0.9949872424989441,
      "learning_rate": 9.80651514983086e-06,
      "loss": 0.5313,
      "step": 1736
    },
    {
      "epoch": 0.09373482272948033,
      "grad_norm": 1.268630327038093,
      "learning_rate": 9.80629379042732e-06,
      "loss": 0.6811,
      "step": 1737
    },
    {
      "epoch": 0.09378878635799472,
      "grad_norm": 0.9852084071710385,
      "learning_rate": 9.806072307256544e-06,
      "loss": 0.4773,
      "step": 1738
    },
    {
      "epoch": 0.0938427499865091,
      "grad_norm": 1.3832159594107338,
      "learning_rate": 9.8058507003249e-06,
      "loss": 0.5529,
      "step": 1739
    },
    {
      "epoch": 0.09389671361502347,
      "grad_norm": 1.1398019688137881,
      "learning_rate": 9.805628969638757e-06,
      "loss": 0.4677,
      "step": 1740
    },
    {
      "epoch": 0.09395067724353785,
      "grad_norm": 1.2312960713727539,
      "learning_rate": 9.805407115204485e-06,
      "loss": 0.6004,
      "step": 1741
    },
    {
      "epoch": 0.09400464087205224,
      "grad_norm": 1.3078251924291642,
      "learning_rate": 9.805185137028462e-06,
      "loss": 0.5637,
      "step": 1742
    },
    {
      "epoch": 0.09405860450056662,
      "grad_norm": 1.273966544786765,
      "learning_rate": 9.804963035117068e-06,
      "loss": 0.652,
      "step": 1743
    },
    {
      "epoch": 0.094112568129081,
      "grad_norm": 1.1369745280783805,
      "learning_rate": 9.804740809476688e-06,
      "loss": 0.6,
      "step": 1744
    },
    {
      "epoch": 0.09416653175759539,
      "grad_norm": 0.8609721359305004,
      "learning_rate": 9.804518460113706e-06,
      "loss": 0.4077,
      "step": 1745
    },
    {
      "epoch": 0.09422049538610976,
      "grad_norm": 1.139629018203808,
      "learning_rate": 9.804295987034515e-06,
      "loss": 0.5478,
      "step": 1746
    },
    {
      "epoch": 0.09427445901462414,
      "grad_norm": 0.9120199284557523,
      "learning_rate": 9.804073390245508e-06,
      "loss": 0.4141,
      "step": 1747
    },
    {
      "epoch": 0.09432842264313852,
      "grad_norm": 0.9657490366756944,
      "learning_rate": 9.803850669753083e-06,
      "loss": 0.413,
      "step": 1748
    },
    {
      "epoch": 0.09438238627165291,
      "grad_norm": 1.0892512503475946,
      "learning_rate": 9.803627825563641e-06,
      "loss": 0.5968,
      "step": 1749
    },
    {
      "epoch": 0.09443634990016729,
      "grad_norm": 1.3626845502378782,
      "learning_rate": 9.803404857683588e-06,
      "loss": 0.516,
      "step": 1750
    },
    {
      "epoch": 0.09449031352868167,
      "grad_norm": 1.1209065797502615,
      "learning_rate": 9.80318176611933e-06,
      "loss": 0.5044,
      "step": 1751
    },
    {
      "epoch": 0.09454427715719604,
      "grad_norm": 1.2686221369868365,
      "learning_rate": 9.80295855087728e-06,
      "loss": 0.7517,
      "step": 1752
    },
    {
      "epoch": 0.09459824078571043,
      "grad_norm": 1.3084350909515148,
      "learning_rate": 9.802735211963853e-06,
      "loss": 0.6232,
      "step": 1753
    },
    {
      "epoch": 0.09465220441422481,
      "grad_norm": 1.261863390994182,
      "learning_rate": 9.802511749385468e-06,
      "loss": 0.572,
      "step": 1754
    },
    {
      "epoch": 0.0947061680427392,
      "grad_norm": 0.9056923726438948,
      "learning_rate": 9.80228816314855e-06,
      "loss": 0.4114,
      "step": 1755
    },
    {
      "epoch": 0.09476013167125358,
      "grad_norm": 0.9738461539492286,
      "learning_rate": 9.802064453259521e-06,
      "loss": 0.3676,
      "step": 1756
    },
    {
      "epoch": 0.09481409529976796,
      "grad_norm": 1.1412149410123702,
      "learning_rate": 9.801840619724813e-06,
      "loss": 0.5079,
      "step": 1757
    },
    {
      "epoch": 0.09486805892828233,
      "grad_norm": 1.1537521167338853,
      "learning_rate": 9.80161666255086e-06,
      "loss": 0.5674,
      "step": 1758
    },
    {
      "epoch": 0.09492202255679671,
      "grad_norm": 1.123774222484853,
      "learning_rate": 9.801392581744097e-06,
      "loss": 0.5131,
      "step": 1759
    },
    {
      "epoch": 0.0949759861853111,
      "grad_norm": 1.0388352692224978,
      "learning_rate": 9.801168377310965e-06,
      "loss": 0.5121,
      "step": 1760
    },
    {
      "epoch": 0.09502994981382548,
      "grad_norm": 0.9909389700218928,
      "learning_rate": 9.800944049257908e-06,
      "loss": 0.4373,
      "step": 1761
    },
    {
      "epoch": 0.09508391344233987,
      "grad_norm": 0.9891729319816386,
      "learning_rate": 9.800719597591371e-06,
      "loss": 0.519,
      "step": 1762
    },
    {
      "epoch": 0.09513787707085425,
      "grad_norm": 1.0633666047469092,
      "learning_rate": 9.80049502231781e-06,
      "loss": 0.499,
      "step": 1763
    },
    {
      "epoch": 0.09519184069936863,
      "grad_norm": 1.0431196568476266,
      "learning_rate": 9.800270323443678e-06,
      "loss": 0.4682,
      "step": 1764
    },
    {
      "epoch": 0.095245804327883,
      "grad_norm": 1.4035536849589416,
      "learning_rate": 9.800045500975428e-06,
      "loss": 0.688,
      "step": 1765
    },
    {
      "epoch": 0.09529976795639739,
      "grad_norm": 0.9686498850327151,
      "learning_rate": 9.799820554919528e-06,
      "loss": 0.4384,
      "step": 1766
    },
    {
      "epoch": 0.09535373158491177,
      "grad_norm": 0.9894101258175626,
      "learning_rate": 9.79959548528244e-06,
      "loss": 0.4117,
      "step": 1767
    },
    {
      "epoch": 0.09540769521342615,
      "grad_norm": 1.3496947995128765,
      "learning_rate": 9.799370292070634e-06,
      "loss": 0.5401,
      "step": 1768
    },
    {
      "epoch": 0.09546165884194054,
      "grad_norm": 0.9816714609340523,
      "learning_rate": 9.799144975290579e-06,
      "loss": 0.4971,
      "step": 1769
    },
    {
      "epoch": 0.09551562247045492,
      "grad_norm": 1.2409106331323252,
      "learning_rate": 9.798919534948756e-06,
      "loss": 0.5532,
      "step": 1770
    },
    {
      "epoch": 0.09556958609896929,
      "grad_norm": 1.0021526498451414,
      "learning_rate": 9.798693971051641e-06,
      "loss": 0.4412,
      "step": 1771
    },
    {
      "epoch": 0.09562354972748367,
      "grad_norm": 1.0864988722263433,
      "learning_rate": 9.798468283605716e-06,
      "loss": 0.4165,
      "step": 1772
    },
    {
      "epoch": 0.09567751335599806,
      "grad_norm": 1.0066569326563153,
      "learning_rate": 9.798242472617471e-06,
      "loss": 0.4913,
      "step": 1773
    },
    {
      "epoch": 0.09573147698451244,
      "grad_norm": 1.2266450087530774,
      "learning_rate": 9.798016538093396e-06,
      "loss": 0.562,
      "step": 1774
    },
    {
      "epoch": 0.09578544061302682,
      "grad_norm": 1.237983520389527,
      "learning_rate": 9.79779048003998e-06,
      "loss": 0.651,
      "step": 1775
    },
    {
      "epoch": 0.0958394042415412,
      "grad_norm": 1.299106086132227,
      "learning_rate": 9.797564298463725e-06,
      "loss": 0.5381,
      "step": 1776
    },
    {
      "epoch": 0.09589336787005558,
      "grad_norm": 1.0839563005024957,
      "learning_rate": 9.797337993371128e-06,
      "loss": 0.517,
      "step": 1777
    },
    {
      "epoch": 0.09594733149856996,
      "grad_norm": 1.1913820795935592,
      "learning_rate": 9.797111564768696e-06,
      "loss": 0.453,
      "step": 1778
    },
    {
      "epoch": 0.09600129512708434,
      "grad_norm": 1.1079006742861335,
      "learning_rate": 9.796885012662936e-06,
      "loss": 0.5147,
      "step": 1779
    },
    {
      "epoch": 0.09605525875559873,
      "grad_norm": 1.129216268516609,
      "learning_rate": 9.796658337060357e-06,
      "loss": 0.4553,
      "step": 1780
    },
    {
      "epoch": 0.09610922238411311,
      "grad_norm": 1.0722575770751104,
      "learning_rate": 9.796431537967477e-06,
      "loss": 0.5327,
      "step": 1781
    },
    {
      "epoch": 0.0961631860126275,
      "grad_norm": 1.1813500952193647,
      "learning_rate": 9.796204615390814e-06,
      "loss": 0.6856,
      "step": 1782
    },
    {
      "epoch": 0.09621714964114186,
      "grad_norm": 1.0913983396043994,
      "learning_rate": 9.795977569336891e-06,
      "loss": 0.4578,
      "step": 1783
    },
    {
      "epoch": 0.09627111326965625,
      "grad_norm": 1.301097327416357,
      "learning_rate": 9.79575039981223e-06,
      "loss": 0.6079,
      "step": 1784
    },
    {
      "epoch": 0.09632507689817063,
      "grad_norm": 1.0548437808049047,
      "learning_rate": 9.79552310682336e-06,
      "loss": 0.4871,
      "step": 1785
    },
    {
      "epoch": 0.09637904052668501,
      "grad_norm": 1.0807952374555794,
      "learning_rate": 9.795295690376818e-06,
      "loss": 0.4061,
      "step": 1786
    },
    {
      "epoch": 0.0964330041551994,
      "grad_norm": 0.9448667901122797,
      "learning_rate": 9.79506815047914e-06,
      "loss": 0.3571,
      "step": 1787
    },
    {
      "epoch": 0.09648696778371378,
      "grad_norm": 0.9529909949888399,
      "learning_rate": 9.794840487136858e-06,
      "loss": 0.4449,
      "step": 1788
    },
    {
      "epoch": 0.09654093141222816,
      "grad_norm": 1.1814576992176162,
      "learning_rate": 9.794612700356525e-06,
      "loss": 0.4849,
      "step": 1789
    },
    {
      "epoch": 0.09659489504074253,
      "grad_norm": 0.9458414707473647,
      "learning_rate": 9.794384790144682e-06,
      "loss": 0.4459,
      "step": 1790
    },
    {
      "epoch": 0.09664885866925692,
      "grad_norm": 1.062069498421552,
      "learning_rate": 9.794156756507881e-06,
      "loss": 0.5639,
      "step": 1791
    },
    {
      "epoch": 0.0967028222977713,
      "grad_norm": 1.123282884208523,
      "learning_rate": 9.793928599452675e-06,
      "loss": 0.71,
      "step": 1792
    },
    {
      "epoch": 0.09675678592628568,
      "grad_norm": 1.217075319920651,
      "learning_rate": 9.793700318985624e-06,
      "loss": 0.5692,
      "step": 1793
    },
    {
      "epoch": 0.09681074955480007,
      "grad_norm": 1.35123143598289,
      "learning_rate": 9.793471915113287e-06,
      "loss": 0.7978,
      "step": 1794
    },
    {
      "epoch": 0.09686471318331445,
      "grad_norm": 1.019110242222147,
      "learning_rate": 9.79324338784223e-06,
      "loss": 0.5335,
      "step": 1795
    },
    {
      "epoch": 0.09691867681182882,
      "grad_norm": 1.0913769935948132,
      "learning_rate": 9.79301473717902e-06,
      "loss": 0.564,
      "step": 1796
    },
    {
      "epoch": 0.0969726404403432,
      "grad_norm": 1.1513368941302295,
      "learning_rate": 9.792785963130225e-06,
      "loss": 0.5794,
      "step": 1797
    },
    {
      "epoch": 0.09702660406885759,
      "grad_norm": 1.0586796465639843,
      "learning_rate": 9.792557065702427e-06,
      "loss": 0.4919,
      "step": 1798
    },
    {
      "epoch": 0.09708056769737197,
      "grad_norm": 1.0502276852068373,
      "learning_rate": 9.792328044902202e-06,
      "loss": 0.4535,
      "step": 1799
    },
    {
      "epoch": 0.09713453132588636,
      "grad_norm": 1.063088844539688,
      "learning_rate": 9.79209890073613e-06,
      "loss": 0.5,
      "step": 1800
    },
    {
      "epoch": 0.09718849495440074,
      "grad_norm": 0.9636081820731449,
      "learning_rate": 9.791869633210798e-06,
      "loss": 0.4989,
      "step": 1801
    },
    {
      "epoch": 0.09724245858291511,
      "grad_norm": 1.1209434214124236,
      "learning_rate": 9.7916402423328e-06,
      "loss": 0.4977,
      "step": 1802
    },
    {
      "epoch": 0.09729642221142949,
      "grad_norm": 1.1209376618342108,
      "learning_rate": 9.791410728108722e-06,
      "loss": 0.5485,
      "step": 1803
    },
    {
      "epoch": 0.09735038583994388,
      "grad_norm": 1.0658658566139085,
      "learning_rate": 9.791181090545164e-06,
      "loss": 0.5307,
      "step": 1804
    },
    {
      "epoch": 0.09740434946845826,
      "grad_norm": 0.9498123257113005,
      "learning_rate": 9.790951329648725e-06,
      "loss": 0.4237,
      "step": 1805
    },
    {
      "epoch": 0.09745831309697264,
      "grad_norm": 1.444550404742815,
      "learning_rate": 9.790721445426011e-06,
      "loss": 0.6454,
      "step": 1806
    },
    {
      "epoch": 0.09751227672548703,
      "grad_norm": 1.101732025832489,
      "learning_rate": 9.790491437883625e-06,
      "loss": 0.5342,
      "step": 1807
    },
    {
      "epoch": 0.0975662403540014,
      "grad_norm": 1.1787976384381782,
      "learning_rate": 9.790261307028183e-06,
      "loss": 0.5147,
      "step": 1808
    },
    {
      "epoch": 0.09762020398251578,
      "grad_norm": 0.951082357310187,
      "learning_rate": 9.790031052866293e-06,
      "loss": 0.5273,
      "step": 1809
    },
    {
      "epoch": 0.09767416761103016,
      "grad_norm": 0.9810628269929993,
      "learning_rate": 9.789800675404578e-06,
      "loss": 0.4761,
      "step": 1810
    },
    {
      "epoch": 0.09772813123954455,
      "grad_norm": 0.8623410505750264,
      "learning_rate": 9.789570174649655e-06,
      "loss": 0.4157,
      "step": 1811
    },
    {
      "epoch": 0.09778209486805893,
      "grad_norm": 0.9726672898715044,
      "learning_rate": 9.789339550608153e-06,
      "loss": 0.472,
      "step": 1812
    },
    {
      "epoch": 0.09783605849657331,
      "grad_norm": 0.9353190594260716,
      "learning_rate": 9.789108803286695e-06,
      "loss": 0.4506,
      "step": 1813
    },
    {
      "epoch": 0.0978900221250877,
      "grad_norm": 1.1331545689286646,
      "learning_rate": 9.78887793269192e-06,
      "loss": 0.6281,
      "step": 1814
    },
    {
      "epoch": 0.09794398575360207,
      "grad_norm": 1.2319173037194096,
      "learning_rate": 9.788646938830457e-06,
      "loss": 0.6923,
      "step": 1815
    },
    {
      "epoch": 0.09799794938211645,
      "grad_norm": 1.0209871537627047,
      "learning_rate": 9.78841582170895e-06,
      "loss": 0.392,
      "step": 1816
    },
    {
      "epoch": 0.09805191301063083,
      "grad_norm": 0.9867241653850197,
      "learning_rate": 9.788184581334036e-06,
      "loss": 0.496,
      "step": 1817
    },
    {
      "epoch": 0.09810587663914522,
      "grad_norm": 1.0244230279437918,
      "learning_rate": 9.787953217712367e-06,
      "loss": 0.4196,
      "step": 1818
    },
    {
      "epoch": 0.0981598402676596,
      "grad_norm": 1.0899912281690927,
      "learning_rate": 9.787721730850588e-06,
      "loss": 0.4899,
      "step": 1819
    },
    {
      "epoch": 0.09821380389617398,
      "grad_norm": 1.0335985348672776,
      "learning_rate": 9.787490120755355e-06,
      "loss": 0.4097,
      "step": 1820
    },
    {
      "epoch": 0.09826776752468835,
      "grad_norm": 1.2996907616305213,
      "learning_rate": 9.787258387433323e-06,
      "loss": 0.6559,
      "step": 1821
    },
    {
      "epoch": 0.09832173115320274,
      "grad_norm": 0.9206905533173715,
      "learning_rate": 9.787026530891153e-06,
      "loss": 0.4209,
      "step": 1822
    },
    {
      "epoch": 0.09837569478171712,
      "grad_norm": 1.0649630561016947,
      "learning_rate": 9.786794551135509e-06,
      "loss": 0.4777,
      "step": 1823
    },
    {
      "epoch": 0.0984296584102315,
      "grad_norm": 1.1307178076231366,
      "learning_rate": 9.786562448173059e-06,
      "loss": 0.6167,
      "step": 1824
    },
    {
      "epoch": 0.09848362203874589,
      "grad_norm": 1.0260615660667647,
      "learning_rate": 9.786330222010471e-06,
      "loss": 0.4616,
      "step": 1825
    },
    {
      "epoch": 0.09853758566726027,
      "grad_norm": 0.8048563765739971,
      "learning_rate": 9.786097872654423e-06,
      "loss": 0.4392,
      "step": 1826
    },
    {
      "epoch": 0.09859154929577464,
      "grad_norm": 1.3138791829106728,
      "learning_rate": 9.785865400111593e-06,
      "loss": 0.5068,
      "step": 1827
    },
    {
      "epoch": 0.09864551292428902,
      "grad_norm": 1.1264502419904723,
      "learning_rate": 9.785632804388658e-06,
      "loss": 0.4967,
      "step": 1828
    },
    {
      "epoch": 0.09869947655280341,
      "grad_norm": 0.9930704575281883,
      "learning_rate": 9.785400085492307e-06,
      "loss": 0.512,
      "step": 1829
    },
    {
      "epoch": 0.09875344018131779,
      "grad_norm": 1.2054757116428756,
      "learning_rate": 9.785167243429226e-06,
      "loss": 0.8699,
      "step": 1830
    },
    {
      "epoch": 0.09880740380983218,
      "grad_norm": 1.2464665362339837,
      "learning_rate": 9.78493427820611e-06,
      "loss": 0.5111,
      "step": 1831
    },
    {
      "epoch": 0.09886136743834656,
      "grad_norm": 0.995354033237861,
      "learning_rate": 9.784701189829654e-06,
      "loss": 0.3865,
      "step": 1832
    },
    {
      "epoch": 0.09891533106686094,
      "grad_norm": 0.9982282218056565,
      "learning_rate": 9.784467978306556e-06,
      "loss": 0.5704,
      "step": 1833
    },
    {
      "epoch": 0.09896929469537531,
      "grad_norm": 1.1208261196725686,
      "learning_rate": 9.784234643643518e-06,
      "loss": 0.6302,
      "step": 1834
    },
    {
      "epoch": 0.0990232583238897,
      "grad_norm": 0.9558653700993097,
      "learning_rate": 9.78400118584725e-06,
      "loss": 0.5035,
      "step": 1835
    },
    {
      "epoch": 0.09907722195240408,
      "grad_norm": 1.0012625298386546,
      "learning_rate": 9.783767604924455e-06,
      "loss": 0.6002,
      "step": 1836
    },
    {
      "epoch": 0.09913118558091846,
      "grad_norm": 1.6742097593036334,
      "learning_rate": 9.783533900881853e-06,
      "loss": 0.7644,
      "step": 1837
    },
    {
      "epoch": 0.09918514920943285,
      "grad_norm": 1.059317947976064,
      "learning_rate": 9.78330007372616e-06,
      "loss": 0.3876,
      "step": 1838
    },
    {
      "epoch": 0.09923911283794723,
      "grad_norm": 1.1821808255801756,
      "learning_rate": 9.783066123464091e-06,
      "loss": 0.5948,
      "step": 1839
    },
    {
      "epoch": 0.0992930764664616,
      "grad_norm": 1.238947540278007,
      "learning_rate": 9.782832050102378e-06,
      "loss": 0.6476,
      "step": 1840
    },
    {
      "epoch": 0.09934704009497598,
      "grad_norm": 0.9430134847310206,
      "learning_rate": 9.782597853647743e-06,
      "loss": 0.4044,
      "step": 1841
    },
    {
      "epoch": 0.09940100372349037,
      "grad_norm": 1.2301054647503713,
      "learning_rate": 9.782363534106917e-06,
      "loss": 0.6224,
      "step": 1842
    },
    {
      "epoch": 0.09945496735200475,
      "grad_norm": 1.178245806928214,
      "learning_rate": 9.782129091486638e-06,
      "loss": 0.5336,
      "step": 1843
    },
    {
      "epoch": 0.09950893098051913,
      "grad_norm": 1.0221590079033773,
      "learning_rate": 9.781894525793641e-06,
      "loss": 0.494,
      "step": 1844
    },
    {
      "epoch": 0.09956289460903352,
      "grad_norm": 1.1996440833919009,
      "learning_rate": 9.78165983703467e-06,
      "loss": 0.6876,
      "step": 1845
    },
    {
      "epoch": 0.09961685823754789,
      "grad_norm": 1.081663977146522,
      "learning_rate": 9.781425025216467e-06,
      "loss": 0.4538,
      "step": 1846
    },
    {
      "epoch": 0.09967082186606227,
      "grad_norm": 1.2547613743094457,
      "learning_rate": 9.781190090345784e-06,
      "loss": 0.6449,
      "step": 1847
    },
    {
      "epoch": 0.09972478549457665,
      "grad_norm": 1.087594038997108,
      "learning_rate": 9.780955032429372e-06,
      "loss": 0.5822,
      "step": 1848
    },
    {
      "epoch": 0.09977874912309104,
      "grad_norm": 1.1864787589616799,
      "learning_rate": 9.780719851473985e-06,
      "loss": 0.5884,
      "step": 1849
    },
    {
      "epoch": 0.09983271275160542,
      "grad_norm": 1.1043706902386765,
      "learning_rate": 9.780484547486388e-06,
      "loss": 0.5356,
      "step": 1850
    },
    {
      "epoch": 0.0998866763801198,
      "grad_norm": 1.1004756822246937,
      "learning_rate": 9.780249120473336e-06,
      "loss": 0.4644,
      "step": 1851
    },
    {
      "epoch": 0.09994064000863417,
      "grad_norm": 1.0306662951937564,
      "learning_rate": 9.780013570441601e-06,
      "loss": 0.5103,
      "step": 1852
    },
    {
      "epoch": 0.09999460363714856,
      "grad_norm": 1.172975462899251,
      "learning_rate": 9.779777897397951e-06,
      "loss": 0.6517,
      "step": 1853
    },
    {
      "epoch": 0.10004856726566294,
      "grad_norm": 1.2606032253971604,
      "learning_rate": 9.779542101349159e-06,
      "loss": 0.6306,
      "step": 1854
    },
    {
      "epoch": 0.10010253089417732,
      "grad_norm": 1.0079193323931124,
      "learning_rate": 9.779306182302007e-06,
      "loss": 0.4927,
      "step": 1855
    },
    {
      "epoch": 0.10015649452269171,
      "grad_norm": 1.1727876550411127,
      "learning_rate": 9.779070140263268e-06,
      "loss": 0.4622,
      "step": 1856
    },
    {
      "epoch": 0.10021045815120609,
      "grad_norm": 0.8888248268363738,
      "learning_rate": 9.77883397523973e-06,
      "loss": 0.4503,
      "step": 1857
    },
    {
      "epoch": 0.10026442177972048,
      "grad_norm": 1.0650611495111002,
      "learning_rate": 9.77859768723818e-06,
      "loss": 0.5226,
      "step": 1858
    },
    {
      "epoch": 0.10031838540823484,
      "grad_norm": 1.1361345648201215,
      "learning_rate": 9.77836127626541e-06,
      "loss": 0.4927,
      "step": 1859
    },
    {
      "epoch": 0.10037234903674923,
      "grad_norm": 0.9875975064934887,
      "learning_rate": 9.778124742328212e-06,
      "loss": 0.5089,
      "step": 1860
    },
    {
      "epoch": 0.10042631266526361,
      "grad_norm": 1.208188369414098,
      "learning_rate": 9.777888085433387e-06,
      "loss": 0.5662,
      "step": 1861
    },
    {
      "epoch": 0.100480276293778,
      "grad_norm": 0.9986593616226938,
      "learning_rate": 9.777651305587736e-06,
      "loss": 0.4445,
      "step": 1862
    },
    {
      "epoch": 0.10053423992229238,
      "grad_norm": 0.9914849452851272,
      "learning_rate": 9.777414402798065e-06,
      "loss": 0.4619,
      "step": 1863
    },
    {
      "epoch": 0.10058820355080676,
      "grad_norm": 1.1413348890840276,
      "learning_rate": 9.777177377071183e-06,
      "loss": 0.4622,
      "step": 1864
    },
    {
      "epoch": 0.10064216717932113,
      "grad_norm": 1.1994138704593589,
      "learning_rate": 9.776940228413899e-06,
      "loss": 0.5304,
      "step": 1865
    },
    {
      "epoch": 0.10069613080783552,
      "grad_norm": 1.0976185142941253,
      "learning_rate": 9.776702956833032e-06,
      "loss": 0.5702,
      "step": 1866
    },
    {
      "epoch": 0.1007500944363499,
      "grad_norm": 0.8346531393376027,
      "learning_rate": 9.7764655623354e-06,
      "loss": 0.3277,
      "step": 1867
    },
    {
      "epoch": 0.10080405806486428,
      "grad_norm": 1.121980705541515,
      "learning_rate": 9.776228044927828e-06,
      "loss": 0.6302,
      "step": 1868
    },
    {
      "epoch": 0.10085802169337867,
      "grad_norm": 1.2308518065437721,
      "learning_rate": 9.775990404617141e-06,
      "loss": 0.6476,
      "step": 1869
    },
    {
      "epoch": 0.10091198532189305,
      "grad_norm": 1.2447969601492699,
      "learning_rate": 9.775752641410169e-06,
      "loss": 0.6093,
      "step": 1870
    },
    {
      "epoch": 0.10096594895040742,
      "grad_norm": 0.9467125711371354,
      "learning_rate": 9.775514755313746e-06,
      "loss": 0.458,
      "step": 1871
    },
    {
      "epoch": 0.1010199125789218,
      "grad_norm": 1.208223856424701,
      "learning_rate": 9.775276746334708e-06,
      "loss": 0.5962,
      "step": 1872
    },
    {
      "epoch": 0.10107387620743619,
      "grad_norm": 0.9845535196052614,
      "learning_rate": 9.775038614479898e-06,
      "loss": 0.4954,
      "step": 1873
    },
    {
      "epoch": 0.10112783983595057,
      "grad_norm": 1.1984163519840867,
      "learning_rate": 9.774800359756158e-06,
      "loss": 0.6807,
      "step": 1874
    },
    {
      "epoch": 0.10118180346446495,
      "grad_norm": 1.126261603044235,
      "learning_rate": 9.774561982170337e-06,
      "loss": 0.6708,
      "step": 1875
    },
    {
      "epoch": 0.10123576709297934,
      "grad_norm": 1.4660249265947256,
      "learning_rate": 9.774323481729284e-06,
      "loss": 0.7612,
      "step": 1876
    },
    {
      "epoch": 0.1012897307214937,
      "grad_norm": 1.3514992175069376,
      "learning_rate": 9.774084858439857e-06,
      "loss": 0.7213,
      "step": 1877
    },
    {
      "epoch": 0.10134369435000809,
      "grad_norm": 1.165358703418103,
      "learning_rate": 9.77384611230891e-06,
      "loss": 0.5559,
      "step": 1878
    },
    {
      "epoch": 0.10139765797852247,
      "grad_norm": 0.9486946037174541,
      "learning_rate": 9.773607243343308e-06,
      "loss": 0.4351,
      "step": 1879
    },
    {
      "epoch": 0.10145162160703686,
      "grad_norm": 0.6860129533340775,
      "learning_rate": 9.773368251549918e-06,
      "loss": 0.2685,
      "step": 1880
    },
    {
      "epoch": 0.10150558523555124,
      "grad_norm": 1.0110395434502522,
      "learning_rate": 9.773129136935602e-06,
      "loss": 0.4514,
      "step": 1881
    },
    {
      "epoch": 0.10155954886406562,
      "grad_norm": 0.9933075677480542,
      "learning_rate": 9.772889899507242e-06,
      "loss": 0.5124,
      "step": 1882
    },
    {
      "epoch": 0.10161351249258001,
      "grad_norm": 1.237437058665746,
      "learning_rate": 9.772650539271705e-06,
      "loss": 0.4625,
      "step": 1883
    },
    {
      "epoch": 0.10166747612109438,
      "grad_norm": 0.9749338352717853,
      "learning_rate": 9.772411056235878e-06,
      "loss": 0.5221,
      "step": 1884
    },
    {
      "epoch": 0.10172143974960876,
      "grad_norm": 1.1970607217938785,
      "learning_rate": 9.772171450406637e-06,
      "loss": 0.6625,
      "step": 1885
    },
    {
      "epoch": 0.10177540337812314,
      "grad_norm": 1.169662906668592,
      "learning_rate": 9.771931721790874e-06,
      "loss": 0.5919,
      "step": 1886
    },
    {
      "epoch": 0.10182936700663753,
      "grad_norm": 1.2846094887357278,
      "learning_rate": 9.771691870395475e-06,
      "loss": 0.449,
      "step": 1887
    },
    {
      "epoch": 0.10188333063515191,
      "grad_norm": 1.0579844204100555,
      "learning_rate": 9.771451896227336e-06,
      "loss": 0.4532,
      "step": 1888
    },
    {
      "epoch": 0.1019372942636663,
      "grad_norm": 1.1597314171062565,
      "learning_rate": 9.771211799293354e-06,
      "loss": 0.5575,
      "step": 1889
    },
    {
      "epoch": 0.10199125789218066,
      "grad_norm": 0.9659196655786471,
      "learning_rate": 9.77097157960043e-06,
      "loss": 0.375,
      "step": 1890
    },
    {
      "epoch": 0.10204522152069505,
      "grad_norm": 0.8475788025608892,
      "learning_rate": 9.770731237155465e-06,
      "loss": 0.3533,
      "step": 1891
    },
    {
      "epoch": 0.10209918514920943,
      "grad_norm": 0.9958470031529646,
      "learning_rate": 9.77049077196537e-06,
      "loss": 0.4248,
      "step": 1892
    },
    {
      "epoch": 0.10215314877772382,
      "grad_norm": 1.1857783460869353,
      "learning_rate": 9.770250184037056e-06,
      "loss": 0.4717,
      "step": 1893
    },
    {
      "epoch": 0.1022071124062382,
      "grad_norm": 1.0591714113278927,
      "learning_rate": 9.770009473377437e-06,
      "loss": 0.5194,
      "step": 1894
    },
    {
      "epoch": 0.10226107603475258,
      "grad_norm": 1.122486941608562,
      "learning_rate": 9.76976863999343e-06,
      "loss": 0.5574,
      "step": 1895
    },
    {
      "epoch": 0.10231503966326695,
      "grad_norm": 1.2220364464661535,
      "learning_rate": 9.769527683891958e-06,
      "loss": 0.6314,
      "step": 1896
    },
    {
      "epoch": 0.10236900329178134,
      "grad_norm": 1.0533765017318564,
      "learning_rate": 9.769286605079947e-06,
      "loss": 0.5347,
      "step": 1897
    },
    {
      "epoch": 0.10242296692029572,
      "grad_norm": 1.2544918180915918,
      "learning_rate": 9.769045403564325e-06,
      "loss": 0.5127,
      "step": 1898
    },
    {
      "epoch": 0.1024769305488101,
      "grad_norm": 1.1509187331144999,
      "learning_rate": 9.768804079352025e-06,
      "loss": 0.5742,
      "step": 1899
    },
    {
      "epoch": 0.10253089417732449,
      "grad_norm": 1.218467989454776,
      "learning_rate": 9.768562632449982e-06,
      "loss": 0.6117,
      "step": 1900
    },
    {
      "epoch": 0.10258485780583887,
      "grad_norm": 0.9654140422632717,
      "learning_rate": 9.768321062865135e-06,
      "loss": 0.4095,
      "step": 1901
    },
    {
      "epoch": 0.10263882143435324,
      "grad_norm": 0.9060808909925768,
      "learning_rate": 9.768079370604431e-06,
      "loss": 0.457,
      "step": 1902
    },
    {
      "epoch": 0.10269278506286762,
      "grad_norm": 1.0915297243852264,
      "learning_rate": 9.76783755567481e-06,
      "loss": 0.5095,
      "step": 1903
    },
    {
      "epoch": 0.102746748691382,
      "grad_norm": 0.854957363160912,
      "learning_rate": 9.767595618083227e-06,
      "loss": 0.4444,
      "step": 1904
    },
    {
      "epoch": 0.10280071231989639,
      "grad_norm": 1.3057005954878795,
      "learning_rate": 9.767353557836633e-06,
      "loss": 0.7067,
      "step": 1905
    },
    {
      "epoch": 0.10285467594841077,
      "grad_norm": 1.2888731693275481,
      "learning_rate": 9.767111374941987e-06,
      "loss": 0.5779,
      "step": 1906
    },
    {
      "epoch": 0.10290863957692516,
      "grad_norm": 1.0462384817391857,
      "learning_rate": 9.766869069406245e-06,
      "loss": 0.5162,
      "step": 1907
    },
    {
      "epoch": 0.10296260320543954,
      "grad_norm": 0.868789715317468,
      "learning_rate": 9.766626641236379e-06,
      "loss": 0.3558,
      "step": 1908
    },
    {
      "epoch": 0.10301656683395391,
      "grad_norm": 1.0344716985952944,
      "learning_rate": 9.76638409043935e-06,
      "loss": 0.395,
      "step": 1909
    },
    {
      "epoch": 0.1030705304624683,
      "grad_norm": 1.1302954093544293,
      "learning_rate": 9.766141417022131e-06,
      "loss": 0.5591,
      "step": 1910
    },
    {
      "epoch": 0.10312449409098268,
      "grad_norm": 1.1115187130104567,
      "learning_rate": 9.765898620991699e-06,
      "loss": 0.5296,
      "step": 1911
    },
    {
      "epoch": 0.10317845771949706,
      "grad_norm": 1.0707296622789542,
      "learning_rate": 9.765655702355028e-06,
      "loss": 0.5228,
      "step": 1912
    },
    {
      "epoch": 0.10323242134801144,
      "grad_norm": 1.135206198426863,
      "learning_rate": 9.765412661119103e-06,
      "loss": 0.5881,
      "step": 1913
    },
    {
      "epoch": 0.10328638497652583,
      "grad_norm": 1.0137421601106908,
      "learning_rate": 9.765169497290908e-06,
      "loss": 0.4427,
      "step": 1914
    },
    {
      "epoch": 0.1033403486050402,
      "grad_norm": 1.041405489898531,
      "learning_rate": 9.764926210877434e-06,
      "loss": 0.5667,
      "step": 1915
    },
    {
      "epoch": 0.10339431223355458,
      "grad_norm": 0.8015744955615476,
      "learning_rate": 9.76468280188567e-06,
      "loss": 0.4133,
      "step": 1916
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 1.2761945552458638,
      "learning_rate": 9.764439270322612e-06,
      "loss": 0.5874,
      "step": 1917
    },
    {
      "epoch": 0.10350223949058335,
      "grad_norm": 0.8768767769631532,
      "learning_rate": 9.764195616195261e-06,
      "loss": 0.5022,
      "step": 1918
    },
    {
      "epoch": 0.10355620311909773,
      "grad_norm": 1.1827522028036292,
      "learning_rate": 9.76395183951062e-06,
      "loss": 0.6599,
      "step": 1919
    },
    {
      "epoch": 0.10361016674761211,
      "grad_norm": 1.0709601727285045,
      "learning_rate": 9.763707940275695e-06,
      "loss": 0.5559,
      "step": 1920
    },
    {
      "epoch": 0.10366413037612648,
      "grad_norm": 0.8464960943503373,
      "learning_rate": 9.763463918497494e-06,
      "loss": 0.4622,
      "step": 1921
    },
    {
      "epoch": 0.10371809400464087,
      "grad_norm": 1.0747608486689184,
      "learning_rate": 9.763219774183034e-06,
      "loss": 0.5414,
      "step": 1922
    },
    {
      "epoch": 0.10377205763315525,
      "grad_norm": 1.0538094625918362,
      "learning_rate": 9.762975507339328e-06,
      "loss": 0.4669,
      "step": 1923
    },
    {
      "epoch": 0.10382602126166963,
      "grad_norm": 1.0218935444697155,
      "learning_rate": 9.762731117973399e-06,
      "loss": 0.4382,
      "step": 1924
    },
    {
      "epoch": 0.10387998489018402,
      "grad_norm": 1.2765613911370342,
      "learning_rate": 9.76248660609227e-06,
      "loss": 0.6027,
      "step": 1925
    },
    {
      "epoch": 0.1039339485186984,
      "grad_norm": 1.1135633334890616,
      "learning_rate": 9.76224197170297e-06,
      "loss": 0.4015,
      "step": 1926
    },
    {
      "epoch": 0.10398791214721279,
      "grad_norm": 1.188900621126394,
      "learning_rate": 9.76199721481253e-06,
      "loss": 0.6685,
      "step": 1927
    },
    {
      "epoch": 0.10404187577572716,
      "grad_norm": 1.3723524816982617,
      "learning_rate": 9.761752335427984e-06,
      "loss": 0.6583,
      "step": 1928
    },
    {
      "epoch": 0.10409583940424154,
      "grad_norm": 1.051500901366465,
      "learning_rate": 9.761507333556368e-06,
      "loss": 0.5768,
      "step": 1929
    },
    {
      "epoch": 0.10414980303275592,
      "grad_norm": 1.2811555752558006,
      "learning_rate": 9.761262209204724e-06,
      "loss": 0.6703,
      "step": 1930
    },
    {
      "epoch": 0.1042037666612703,
      "grad_norm": 1.2065967281972392,
      "learning_rate": 9.7610169623801e-06,
      "loss": 0.79,
      "step": 1931
    },
    {
      "epoch": 0.10425773028978469,
      "grad_norm": 1.0255104229522647,
      "learning_rate": 9.760771593089544e-06,
      "loss": 0.4396,
      "step": 1932
    },
    {
      "epoch": 0.10431169391829907,
      "grad_norm": 0.9753122444393113,
      "learning_rate": 9.760526101340106e-06,
      "loss": 0.3077,
      "step": 1933
    },
    {
      "epoch": 0.10436565754681344,
      "grad_norm": 1.369651160929907,
      "learning_rate": 9.760280487138845e-06,
      "loss": 0.5699,
      "step": 1934
    },
    {
      "epoch": 0.10441962117532783,
      "grad_norm": 1.0337622269568871,
      "learning_rate": 9.760034750492815e-06,
      "loss": 0.4813,
      "step": 1935
    },
    {
      "epoch": 0.10447358480384221,
      "grad_norm": 1.0734827658009616,
      "learning_rate": 9.759788891409083e-06,
      "loss": 0.5315,
      "step": 1936
    },
    {
      "epoch": 0.10452754843235659,
      "grad_norm": 0.9120105406121684,
      "learning_rate": 9.759542909894717e-06,
      "loss": 0.4425,
      "step": 1937
    },
    {
      "epoch": 0.10458151206087098,
      "grad_norm": 1.1344171256380333,
      "learning_rate": 9.759296805956778e-06,
      "loss": 0.5243,
      "step": 1938
    },
    {
      "epoch": 0.10463547568938536,
      "grad_norm": 1.3691033689696215,
      "learning_rate": 9.75905057960235e-06,
      "loss": 0.6494,
      "step": 1939
    },
    {
      "epoch": 0.10468943931789973,
      "grad_norm": 1.0077312221045187,
      "learning_rate": 9.758804230838502e-06,
      "loss": 0.4312,
      "step": 1940
    },
    {
      "epoch": 0.10474340294641411,
      "grad_norm": 1.151571797197468,
      "learning_rate": 9.758557759672318e-06,
      "loss": 0.5536,
      "step": 1941
    },
    {
      "epoch": 0.1047973665749285,
      "grad_norm": 1.366696536457912,
      "learning_rate": 9.758311166110882e-06,
      "loss": 0.6511,
      "step": 1942
    },
    {
      "epoch": 0.10485133020344288,
      "grad_norm": 1.1269916955119221,
      "learning_rate": 9.758064450161278e-06,
      "loss": 0.4899,
      "step": 1943
    },
    {
      "epoch": 0.10490529383195726,
      "grad_norm": 0.8255439351284474,
      "learning_rate": 9.757817611830602e-06,
      "loss": 0.3823,
      "step": 1944
    },
    {
      "epoch": 0.10495925746047165,
      "grad_norm": 1.1549598155843972,
      "learning_rate": 9.757570651125945e-06,
      "loss": 0.5465,
      "step": 1945
    },
    {
      "epoch": 0.10501322108898602,
      "grad_norm": 1.2461128575807885,
      "learning_rate": 9.757323568054405e-06,
      "loss": 0.6161,
      "step": 1946
    },
    {
      "epoch": 0.1050671847175004,
      "grad_norm": 1.1585262251951964,
      "learning_rate": 9.757076362623083e-06,
      "loss": 0.5056,
      "step": 1947
    },
    {
      "epoch": 0.10512114834601478,
      "grad_norm": 1.3014714051098684,
      "learning_rate": 9.756829034839086e-06,
      "loss": 0.7011,
      "step": 1948
    },
    {
      "epoch": 0.10517511197452917,
      "grad_norm": 1.0012839196414296,
      "learning_rate": 9.75658158470952e-06,
      "loss": 0.6167,
      "step": 1949
    },
    {
      "epoch": 0.10522907560304355,
      "grad_norm": 1.1769901577350173,
      "learning_rate": 9.756334012241501e-06,
      "loss": 0.5975,
      "step": 1950
    },
    {
      "epoch": 0.10528303923155793,
      "grad_norm": 1.1265232575338653,
      "learning_rate": 9.756086317442141e-06,
      "loss": 0.4979,
      "step": 1951
    },
    {
      "epoch": 0.10533700286007232,
      "grad_norm": 1.3404783447215725,
      "learning_rate": 9.755838500318558e-06,
      "loss": 0.5928,
      "step": 1952
    },
    {
      "epoch": 0.10539096648858669,
      "grad_norm": 1.0660806744304923,
      "learning_rate": 9.755590560877878e-06,
      "loss": 0.4715,
      "step": 1953
    },
    {
      "epoch": 0.10544493011710107,
      "grad_norm": 1.3687208201551049,
      "learning_rate": 9.755342499127223e-06,
      "loss": 0.7336,
      "step": 1954
    },
    {
      "epoch": 0.10549889374561545,
      "grad_norm": 1.0976752123329738,
      "learning_rate": 9.755094315073728e-06,
      "loss": 0.5251,
      "step": 1955
    },
    {
      "epoch": 0.10555285737412984,
      "grad_norm": 1.0175485014124146,
      "learning_rate": 9.754846008724522e-06,
      "loss": 0.5214,
      "step": 1956
    },
    {
      "epoch": 0.10560682100264422,
      "grad_norm": 1.0757956040202814,
      "learning_rate": 9.754597580086741e-06,
      "loss": 0.6126,
      "step": 1957
    },
    {
      "epoch": 0.1056607846311586,
      "grad_norm": 1.3058368333925987,
      "learning_rate": 9.754349029167529e-06,
      "loss": 0.7183,
      "step": 1958
    },
    {
      "epoch": 0.10571474825967297,
      "grad_norm": 1.0926702304581917,
      "learning_rate": 9.754100355974024e-06,
      "loss": 0.5207,
      "step": 1959
    },
    {
      "epoch": 0.10576871188818736,
      "grad_norm": 1.2346418114399282,
      "learning_rate": 9.75385156051338e-06,
      "loss": 0.4418,
      "step": 1960
    },
    {
      "epoch": 0.10582267551670174,
      "grad_norm": 1.1853217431163343,
      "learning_rate": 9.753602642792739e-06,
      "loss": 0.6107,
      "step": 1961
    },
    {
      "epoch": 0.10587663914521613,
      "grad_norm": 1.0503762219815762,
      "learning_rate": 9.753353602819264e-06,
      "loss": 0.5852,
      "step": 1962
    },
    {
      "epoch": 0.10593060277373051,
      "grad_norm": 1.1059707897708044,
      "learning_rate": 9.753104440600107e-06,
      "loss": 0.4826,
      "step": 1963
    },
    {
      "epoch": 0.10598456640224489,
      "grad_norm": 1.0704021597905693,
      "learning_rate": 9.752855156142433e-06,
      "loss": 0.5172,
      "step": 1964
    },
    {
      "epoch": 0.10603853003075926,
      "grad_norm": 1.231331718989751,
      "learning_rate": 9.752605749453404e-06,
      "loss": 0.5151,
      "step": 1965
    },
    {
      "epoch": 0.10609249365927365,
      "grad_norm": 1.1303527860448364,
      "learning_rate": 9.752356220540188e-06,
      "loss": 0.5823,
      "step": 1966
    },
    {
      "epoch": 0.10614645728778803,
      "grad_norm": 0.9725785999061783,
      "learning_rate": 9.752106569409958e-06,
      "loss": 0.3771,
      "step": 1967
    },
    {
      "epoch": 0.10620042091630241,
      "grad_norm": 1.2282665862506168,
      "learning_rate": 9.751856796069887e-06,
      "loss": 0.507,
      "step": 1968
    },
    {
      "epoch": 0.1062543845448168,
      "grad_norm": 1.227974916413366,
      "learning_rate": 9.751606900527157e-06,
      "loss": 0.5988,
      "step": 1969
    },
    {
      "epoch": 0.10630834817333118,
      "grad_norm": 0.9658033899837036,
      "learning_rate": 9.75135688278895e-06,
      "loss": 0.4305,
      "step": 1970
    },
    {
      "epoch": 0.10636231180184555,
      "grad_norm": 0.9774777307316999,
      "learning_rate": 9.751106742862449e-06,
      "loss": 0.4212,
      "step": 1971
    },
    {
      "epoch": 0.10641627543035993,
      "grad_norm": 1.1506168800102283,
      "learning_rate": 9.750856480754845e-06,
      "loss": 0.6729,
      "step": 1972
    },
    {
      "epoch": 0.10647023905887432,
      "grad_norm": 1.0171715937156907,
      "learning_rate": 9.75060609647333e-06,
      "loss": 0.4997,
      "step": 1973
    },
    {
      "epoch": 0.1065242026873887,
      "grad_norm": 1.23918714731053,
      "learning_rate": 9.750355590025102e-06,
      "loss": 0.5971,
      "step": 1974
    },
    {
      "epoch": 0.10657816631590308,
      "grad_norm": 1.1331798037533234,
      "learning_rate": 9.75010496141736e-06,
      "loss": 0.6696,
      "step": 1975
    },
    {
      "epoch": 0.10663212994441747,
      "grad_norm": 0.862378184423155,
      "learning_rate": 9.749854210657304e-06,
      "loss": 0.4031,
      "step": 1976
    },
    {
      "epoch": 0.10668609357293185,
      "grad_norm": 1.3031839012483875,
      "learning_rate": 9.749603337752147e-06,
      "loss": 0.4922,
      "step": 1977
    },
    {
      "epoch": 0.10674005720144622,
      "grad_norm": 0.834720570968053,
      "learning_rate": 9.749352342709094e-06,
      "loss": 0.3966,
      "step": 1978
    },
    {
      "epoch": 0.1067940208299606,
      "grad_norm": 0.9881052096592996,
      "learning_rate": 9.749101225535364e-06,
      "loss": 0.4956,
      "step": 1979
    },
    {
      "epoch": 0.10684798445847499,
      "grad_norm": 1.1064283541599667,
      "learning_rate": 9.748849986238168e-06,
      "loss": 0.4747,
      "step": 1980
    },
    {
      "epoch": 0.10690194808698937,
      "grad_norm": 1.255985580021028,
      "learning_rate": 9.748598624824732e-06,
      "loss": 0.6021,
      "step": 1981
    },
    {
      "epoch": 0.10695591171550375,
      "grad_norm": 1.4403620330298401,
      "learning_rate": 9.748347141302278e-06,
      "loss": 0.8397,
      "step": 1982
    },
    {
      "epoch": 0.10700987534401814,
      "grad_norm": 1.0841775932535316,
      "learning_rate": 9.748095535678034e-06,
      "loss": 0.4786,
      "step": 1983
    },
    {
      "epoch": 0.10706383897253251,
      "grad_norm": 1.2428417221318295,
      "learning_rate": 9.747843807959234e-06,
      "loss": 0.5827,
      "step": 1984
    },
    {
      "epoch": 0.10711780260104689,
      "grad_norm": 0.971418060580997,
      "learning_rate": 9.74759195815311e-06,
      "loss": 0.4969,
      "step": 1985
    },
    {
      "epoch": 0.10717176622956127,
      "grad_norm": 1.304176071243054,
      "learning_rate": 9.7473399862669e-06,
      "loss": 0.6962,
      "step": 1986
    },
    {
      "epoch": 0.10722572985807566,
      "grad_norm": 1.2448690984846835,
      "learning_rate": 9.747087892307849e-06,
      "loss": 0.5538,
      "step": 1987
    },
    {
      "epoch": 0.10727969348659004,
      "grad_norm": 1.0846650476547828,
      "learning_rate": 9.746835676283198e-06,
      "loss": 0.5185,
      "step": 1988
    },
    {
      "epoch": 0.10733365711510442,
      "grad_norm": 1.100617112690641,
      "learning_rate": 9.746583338200202e-06,
      "loss": 0.6433,
      "step": 1989
    },
    {
      "epoch": 0.1073876207436188,
      "grad_norm": 0.8669020274209454,
      "learning_rate": 9.746330878066107e-06,
      "loss": 0.3392,
      "step": 1990
    },
    {
      "epoch": 0.10744158437213318,
      "grad_norm": 1.0775369463586923,
      "learning_rate": 9.746078295888173e-06,
      "loss": 0.5139,
      "step": 1991
    },
    {
      "epoch": 0.10749554800064756,
      "grad_norm": 0.8529980238729221,
      "learning_rate": 9.745825591673658e-06,
      "loss": 0.2807,
      "step": 1992
    },
    {
      "epoch": 0.10754951162916195,
      "grad_norm": 0.925742204592247,
      "learning_rate": 9.745572765429827e-06,
      "loss": 0.483,
      "step": 1993
    },
    {
      "epoch": 0.10760347525767633,
      "grad_norm": 1.0341754172552462,
      "learning_rate": 9.745319817163945e-06,
      "loss": 0.5378,
      "step": 1994
    },
    {
      "epoch": 0.10765743888619071,
      "grad_norm": 1.1535429085742888,
      "learning_rate": 9.74506674688328e-06,
      "loss": 0.4762,
      "step": 1995
    },
    {
      "epoch": 0.10771140251470508,
      "grad_norm": 1.09618826383207,
      "learning_rate": 9.744813554595108e-06,
      "loss": 0.5686,
      "step": 1996
    },
    {
      "epoch": 0.10776536614321947,
      "grad_norm": 1.2670251559217747,
      "learning_rate": 9.744560240306703e-06,
      "loss": 0.6233,
      "step": 1997
    },
    {
      "epoch": 0.10781932977173385,
      "grad_norm": 1.1273646205289356,
      "learning_rate": 9.744306804025351e-06,
      "loss": 0.5205,
      "step": 1998
    },
    {
      "epoch": 0.10787329340024823,
      "grad_norm": 1.2114931709774968,
      "learning_rate": 9.744053245758334e-06,
      "loss": 0.5368,
      "step": 1999
    },
    {
      "epoch": 0.10792725702876262,
      "grad_norm": 0.9709901278473052,
      "learning_rate": 9.743799565512937e-06,
      "loss": 0.3807,
      "step": 2000
    },
    {
      "epoch": 0.10792725702876262,
      "eval_loss": 0.6009196639060974,
      "eval_runtime": 165.0692,
      "eval_samples_per_second": 20.834,
      "eval_steps_per_second": 0.872,
      "step": 2000
    },
    {
      "epoch": 0.107981220657277,
      "grad_norm": 1.086327057488277,
      "learning_rate": 9.743545763296451e-06,
      "loss": 0.5048,
      "step": 2001
    },
    {
      "epoch": 0.10803518428579138,
      "grad_norm": 0.9618518192303469,
      "learning_rate": 9.743291839116173e-06,
      "loss": 0.4861,
      "step": 2002
    },
    {
      "epoch": 0.10808914791430575,
      "grad_norm": 1.0349094577675364,
      "learning_rate": 9.7430377929794e-06,
      "loss": 0.5862,
      "step": 2003
    },
    {
      "epoch": 0.10814311154282014,
      "grad_norm": 1.0807198401468234,
      "learning_rate": 9.742783624893433e-06,
      "loss": 0.5229,
      "step": 2004
    },
    {
      "epoch": 0.10819707517133452,
      "grad_norm": 1.1473711659554733,
      "learning_rate": 9.74252933486558e-06,
      "loss": 0.4916,
      "step": 2005
    },
    {
      "epoch": 0.1082510387998489,
      "grad_norm": 1.1290787695001119,
      "learning_rate": 9.742274922903144e-06,
      "loss": 0.5355,
      "step": 2006
    },
    {
      "epoch": 0.10830500242836329,
      "grad_norm": 0.9942238419896946,
      "learning_rate": 9.742020389013443e-06,
      "loss": 0.3975,
      "step": 2007
    },
    {
      "epoch": 0.10835896605687767,
      "grad_norm": 0.7861395908615773,
      "learning_rate": 9.741765733203787e-06,
      "loss": 0.3258,
      "step": 2008
    },
    {
      "epoch": 0.10841292968539204,
      "grad_norm": 1.033835244951,
      "learning_rate": 9.7415109554815e-06,
      "loss": 0.4824,
      "step": 2009
    },
    {
      "epoch": 0.10846689331390642,
      "grad_norm": 1.0998800748785262,
      "learning_rate": 9.7412560558539e-06,
      "loss": 0.45,
      "step": 2010
    },
    {
      "epoch": 0.1085208569424208,
      "grad_norm": 1.0364260881550968,
      "learning_rate": 9.741001034328317e-06,
      "loss": 0.466,
      "step": 2011
    },
    {
      "epoch": 0.10857482057093519,
      "grad_norm": 1.558558225493801,
      "learning_rate": 9.74074589091208e-06,
      "loss": 0.6723,
      "step": 2012
    },
    {
      "epoch": 0.10862878419944957,
      "grad_norm": 1.0105016772140298,
      "learning_rate": 9.740490625612521e-06,
      "loss": 0.5403,
      "step": 2013
    },
    {
      "epoch": 0.10868274782796396,
      "grad_norm": 0.8278394022293022,
      "learning_rate": 9.740235238436975e-06,
      "loss": 0.4206,
      "step": 2014
    },
    {
      "epoch": 0.10873671145647833,
      "grad_norm": 1.1417752499540412,
      "learning_rate": 9.739979729392786e-06,
      "loss": 0.5724,
      "step": 2015
    },
    {
      "epoch": 0.10879067508499271,
      "grad_norm": 0.8586195525028376,
      "learning_rate": 9.739724098487294e-06,
      "loss": 0.4163,
      "step": 2016
    },
    {
      "epoch": 0.1088446387135071,
      "grad_norm": 1.1747285478241036,
      "learning_rate": 9.739468345727847e-06,
      "loss": 0.5618,
      "step": 2017
    },
    {
      "epoch": 0.10889860234202148,
      "grad_norm": 1.2407179989122474,
      "learning_rate": 9.739212471121796e-06,
      "loss": 0.603,
      "step": 2018
    },
    {
      "epoch": 0.10895256597053586,
      "grad_norm": 1.275554684918799,
      "learning_rate": 9.738956474676497e-06,
      "loss": 0.6539,
      "step": 2019
    },
    {
      "epoch": 0.10900652959905024,
      "grad_norm": 1.280343135592382,
      "learning_rate": 9.738700356399305e-06,
      "loss": 0.5992,
      "step": 2020
    },
    {
      "epoch": 0.10906049322756463,
      "grad_norm": 1.019487658172093,
      "learning_rate": 9.738444116297581e-06,
      "loss": 0.4199,
      "step": 2021
    },
    {
      "epoch": 0.109114456856079,
      "grad_norm": 0.9937539966186173,
      "learning_rate": 9.738187754378692e-06,
      "loss": 0.4821,
      "step": 2022
    },
    {
      "epoch": 0.10916842048459338,
      "grad_norm": 0.7000724513408018,
      "learning_rate": 9.737931270650002e-06,
      "loss": 0.2971,
      "step": 2023
    },
    {
      "epoch": 0.10922238411310776,
      "grad_norm": 1.4876554142497398,
      "learning_rate": 9.737674665118887e-06,
      "loss": 0.7589,
      "step": 2024
    },
    {
      "epoch": 0.10927634774162215,
      "grad_norm": 0.9254853080874313,
      "learning_rate": 9.73741793779272e-06,
      "loss": 0.3833,
      "step": 2025
    },
    {
      "epoch": 0.10933031137013653,
      "grad_norm": 0.9243946785957957,
      "learning_rate": 9.73716108867888e-06,
      "loss": 0.3512,
      "step": 2026
    },
    {
      "epoch": 0.10938427499865092,
      "grad_norm": 1.071068732509766,
      "learning_rate": 9.736904117784748e-06,
      "loss": 0.5512,
      "step": 2027
    },
    {
      "epoch": 0.10943823862716529,
      "grad_norm": 1.2049337134569917,
      "learning_rate": 9.736647025117711e-06,
      "loss": 0.7301,
      "step": 2028
    },
    {
      "epoch": 0.10949220225567967,
      "grad_norm": 1.3019665514348748,
      "learning_rate": 9.736389810685158e-06,
      "loss": 0.6059,
      "step": 2029
    },
    {
      "epoch": 0.10954616588419405,
      "grad_norm": 1.0494434705119964,
      "learning_rate": 9.736132474494481e-06,
      "loss": 0.4897,
      "step": 2030
    },
    {
      "epoch": 0.10960012951270844,
      "grad_norm": 1.0506933083696381,
      "learning_rate": 9.735875016553078e-06,
      "loss": 0.5215,
      "step": 2031
    },
    {
      "epoch": 0.10965409314122282,
      "grad_norm": 1.110877362099513,
      "learning_rate": 9.735617436868346e-06,
      "loss": 0.5188,
      "step": 2032
    },
    {
      "epoch": 0.1097080567697372,
      "grad_norm": 1.1617723752266298,
      "learning_rate": 9.735359735447688e-06,
      "loss": 0.4617,
      "step": 2033
    },
    {
      "epoch": 0.10976202039825157,
      "grad_norm": 1.2478676094178631,
      "learning_rate": 9.735101912298513e-06,
      "loss": 0.7055,
      "step": 2034
    },
    {
      "epoch": 0.10981598402676596,
      "grad_norm": 0.9557792821728467,
      "learning_rate": 9.73484396742823e-06,
      "loss": 0.3877,
      "step": 2035
    },
    {
      "epoch": 0.10986994765528034,
      "grad_norm": 1.1573665753897102,
      "learning_rate": 9.734585900844253e-06,
      "loss": 0.4788,
      "step": 2036
    },
    {
      "epoch": 0.10992391128379472,
      "grad_norm": 1.0370546997934993,
      "learning_rate": 9.734327712553995e-06,
      "loss": 0.573,
      "step": 2037
    },
    {
      "epoch": 0.1099778749123091,
      "grad_norm": 0.980348856038039,
      "learning_rate": 9.734069402564882e-06,
      "loss": 0.4304,
      "step": 2038
    },
    {
      "epoch": 0.11003183854082349,
      "grad_norm": 1.1168487005494403,
      "learning_rate": 9.733810970884337e-06,
      "loss": 0.6007,
      "step": 2039
    },
    {
      "epoch": 0.11008580216933786,
      "grad_norm": 0.8303364341660308,
      "learning_rate": 9.733552417519787e-06,
      "loss": 0.3367,
      "step": 2040
    },
    {
      "epoch": 0.11013976579785224,
      "grad_norm": 1.0739726817212272,
      "learning_rate": 9.733293742478661e-06,
      "loss": 0.3936,
      "step": 2041
    },
    {
      "epoch": 0.11019372942636663,
      "grad_norm": 1.1208176391013152,
      "learning_rate": 9.733034945768396e-06,
      "loss": 0.4646,
      "step": 2042
    },
    {
      "epoch": 0.11024769305488101,
      "grad_norm": 0.9835334674808709,
      "learning_rate": 9.73277602739643e-06,
      "loss": 0.4459,
      "step": 2043
    },
    {
      "epoch": 0.1103016566833954,
      "grad_norm": 0.8932043869611919,
      "learning_rate": 9.732516987370204e-06,
      "loss": 0.4296,
      "step": 2044
    },
    {
      "epoch": 0.11035562031190978,
      "grad_norm": 1.1156003941281467,
      "learning_rate": 9.732257825697164e-06,
      "loss": 0.5599,
      "step": 2045
    },
    {
      "epoch": 0.11040958394042416,
      "grad_norm": 1.2569202916685973,
      "learning_rate": 9.731998542384755e-06,
      "loss": 0.6855,
      "step": 2046
    },
    {
      "epoch": 0.11046354756893853,
      "grad_norm": 1.0683697151603757,
      "learning_rate": 9.731739137440435e-06,
      "loss": 0.4774,
      "step": 2047
    },
    {
      "epoch": 0.11051751119745291,
      "grad_norm": 1.1041546828753652,
      "learning_rate": 9.731479610871653e-06,
      "loss": 0.6182,
      "step": 2048
    },
    {
      "epoch": 0.1105714748259673,
      "grad_norm": 1.0594516865997277,
      "learning_rate": 9.731219962685875e-06,
      "loss": 0.4786,
      "step": 2049
    },
    {
      "epoch": 0.11062543845448168,
      "grad_norm": 0.9481132721538373,
      "learning_rate": 9.730960192890557e-06,
      "loss": 0.4332,
      "step": 2050
    },
    {
      "epoch": 0.11067940208299606,
      "grad_norm": 1.2386574979502512,
      "learning_rate": 9.73070030149317e-06,
      "loss": 0.6605,
      "step": 2051
    },
    {
      "epoch": 0.11073336571151045,
      "grad_norm": 1.1749002189703048,
      "learning_rate": 9.73044028850118e-06,
      "loss": 0.74,
      "step": 2052
    },
    {
      "epoch": 0.11078732934002482,
      "grad_norm": 1.0390648996291156,
      "learning_rate": 9.730180153922064e-06,
      "loss": 0.4984,
      "step": 2053
    },
    {
      "epoch": 0.1108412929685392,
      "grad_norm": 1.2089202836836257,
      "learning_rate": 9.729919897763292e-06,
      "loss": 0.6758,
      "step": 2054
    },
    {
      "epoch": 0.11089525659705358,
      "grad_norm": 1.0501131657131468,
      "learning_rate": 9.729659520032351e-06,
      "loss": 0.4171,
      "step": 2055
    },
    {
      "epoch": 0.11094922022556797,
      "grad_norm": 0.8490220409725224,
      "learning_rate": 9.729399020736721e-06,
      "loss": 0.3382,
      "step": 2056
    },
    {
      "epoch": 0.11100318385408235,
      "grad_norm": 1.2036475796245694,
      "learning_rate": 9.729138399883889e-06,
      "loss": 0.5028,
      "step": 2057
    },
    {
      "epoch": 0.11105714748259674,
      "grad_norm": 0.9829412932223817,
      "learning_rate": 9.728877657481348e-06,
      "loss": 0.4841,
      "step": 2058
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 1.2777568068937704,
      "learning_rate": 9.728616793536588e-06,
      "loss": 0.6849,
      "step": 2059
    },
    {
      "epoch": 0.11116507473962549,
      "grad_norm": 1.172701257470385,
      "learning_rate": 9.72835580805711e-06,
      "loss": 0.4159,
      "step": 2060
    },
    {
      "epoch": 0.11121903836813987,
      "grad_norm": 1.0582032856621537,
      "learning_rate": 9.728094701050413e-06,
      "loss": 0.4911,
      "step": 2061
    },
    {
      "epoch": 0.11127300199665426,
      "grad_norm": 1.3071440638814404,
      "learning_rate": 9.727833472524003e-06,
      "loss": 0.6866,
      "step": 2062
    },
    {
      "epoch": 0.11132696562516864,
      "grad_norm": 1.0590361342197327,
      "learning_rate": 9.727572122485386e-06,
      "loss": 0.497,
      "step": 2063
    },
    {
      "epoch": 0.11138092925368302,
      "grad_norm": 1.4589507959826196,
      "learning_rate": 9.727310650942075e-06,
      "loss": 0.6057,
      "step": 2064
    },
    {
      "epoch": 0.11143489288219739,
      "grad_norm": 1.0155104005182347,
      "learning_rate": 9.727049057901586e-06,
      "loss": 0.4684,
      "step": 2065
    },
    {
      "epoch": 0.11148885651071178,
      "grad_norm": 1.0711181466199533,
      "learning_rate": 9.726787343371433e-06,
      "loss": 0.5111,
      "step": 2066
    },
    {
      "epoch": 0.11154282013922616,
      "grad_norm": 1.0975477574969559,
      "learning_rate": 9.726525507359143e-06,
      "loss": 0.5015,
      "step": 2067
    },
    {
      "epoch": 0.11159678376774054,
      "grad_norm": 1.0752889361973546,
      "learning_rate": 9.726263549872238e-06,
      "loss": 0.4686,
      "step": 2068
    },
    {
      "epoch": 0.11165074739625493,
      "grad_norm": 1.1687453568198447,
      "learning_rate": 9.726001470918249e-06,
      "loss": 0.4929,
      "step": 2069
    },
    {
      "epoch": 0.11170471102476931,
      "grad_norm": 1.3647900262835757,
      "learning_rate": 9.725739270504707e-06,
      "loss": 0.7172,
      "step": 2070
    },
    {
      "epoch": 0.1117586746532837,
      "grad_norm": 1.0630016687011967,
      "learning_rate": 9.725476948639152e-06,
      "loss": 0.6157,
      "step": 2071
    },
    {
      "epoch": 0.11181263828179806,
      "grad_norm": 1.0252247452586676,
      "learning_rate": 9.725214505329114e-06,
      "loss": 0.5517,
      "step": 2072
    },
    {
      "epoch": 0.11186660191031245,
      "grad_norm": 1.1792647687740614,
      "learning_rate": 9.724951940582146e-06,
      "loss": 0.5353,
      "step": 2073
    },
    {
      "epoch": 0.11192056553882683,
      "grad_norm": 1.0774365003462847,
      "learning_rate": 9.724689254405789e-06,
      "loss": 0.481,
      "step": 2074
    },
    {
      "epoch": 0.11197452916734121,
      "grad_norm": 1.124972102210081,
      "learning_rate": 9.724426446807595e-06,
      "loss": 0.57,
      "step": 2075
    },
    {
      "epoch": 0.1120284927958556,
      "grad_norm": 0.9853076749628312,
      "learning_rate": 9.724163517795116e-06,
      "loss": 0.4656,
      "step": 2076
    },
    {
      "epoch": 0.11208245642436998,
      "grad_norm": 0.8760299877697736,
      "learning_rate": 9.72390046737591e-06,
      "loss": 0.3396,
      "step": 2077
    },
    {
      "epoch": 0.11213642005288435,
      "grad_norm": 0.8879348514567834,
      "learning_rate": 9.723637295557535e-06,
      "loss": 0.445,
      "step": 2078
    },
    {
      "epoch": 0.11219038368139873,
      "grad_norm": 1.024961794933285,
      "learning_rate": 9.723374002347557e-06,
      "loss": 0.4248,
      "step": 2079
    },
    {
      "epoch": 0.11224434730991312,
      "grad_norm": 0.9135338012918697,
      "learning_rate": 9.723110587753544e-06,
      "loss": 0.4054,
      "step": 2080
    },
    {
      "epoch": 0.1122983109384275,
      "grad_norm": 1.019269099574995,
      "learning_rate": 9.722847051783062e-06,
      "loss": 0.3603,
      "step": 2081
    },
    {
      "epoch": 0.11235227456694188,
      "grad_norm": 1.2011374791691507,
      "learning_rate": 9.722583394443692e-06,
      "loss": 0.6786,
      "step": 2082
    },
    {
      "epoch": 0.11240623819545627,
      "grad_norm": 1.1884726055708354,
      "learning_rate": 9.722319615743009e-06,
      "loss": 0.5652,
      "step": 2083
    },
    {
      "epoch": 0.11246020182397064,
      "grad_norm": 1.0785529997291525,
      "learning_rate": 9.722055715688593e-06,
      "loss": 0.5207,
      "step": 2084
    },
    {
      "epoch": 0.11251416545248502,
      "grad_norm": 1.026048563046313,
      "learning_rate": 9.721791694288029e-06,
      "loss": 0.5358,
      "step": 2085
    },
    {
      "epoch": 0.1125681290809994,
      "grad_norm": 1.0613392585949002,
      "learning_rate": 9.721527551548908e-06,
      "loss": 0.5013,
      "step": 2086
    },
    {
      "epoch": 0.11262209270951379,
      "grad_norm": 1.245909574567919,
      "learning_rate": 9.721263287478817e-06,
      "loss": 0.5602,
      "step": 2087
    },
    {
      "epoch": 0.11267605633802817,
      "grad_norm": 1.0464217842666608,
      "learning_rate": 9.720998902085354e-06,
      "loss": 0.5457,
      "step": 2088
    },
    {
      "epoch": 0.11273001996654256,
      "grad_norm": 1.1439479559145347,
      "learning_rate": 9.72073439537612e-06,
      "loss": 0.4971,
      "step": 2089
    },
    {
      "epoch": 0.11278398359505692,
      "grad_norm": 1.182967193168024,
      "learning_rate": 9.720469767358712e-06,
      "loss": 0.5949,
      "step": 2090
    },
    {
      "epoch": 0.11283794722357131,
      "grad_norm": 1.154401640187526,
      "learning_rate": 9.720205018040739e-06,
      "loss": 0.6422,
      "step": 2091
    },
    {
      "epoch": 0.11289191085208569,
      "grad_norm": 1.119996773183585,
      "learning_rate": 9.71994014742981e-06,
      "loss": 0.5825,
      "step": 2092
    },
    {
      "epoch": 0.11294587448060008,
      "grad_norm": 1.0596990776268982,
      "learning_rate": 9.719675155533538e-06,
      "loss": 0.5041,
      "step": 2093
    },
    {
      "epoch": 0.11299983810911446,
      "grad_norm": 0.8844227592405739,
      "learning_rate": 9.719410042359538e-06,
      "loss": 0.3153,
      "step": 2094
    },
    {
      "epoch": 0.11305380173762884,
      "grad_norm": 1.010530722002409,
      "learning_rate": 9.71914480791543e-06,
      "loss": 0.5955,
      "step": 2095
    },
    {
      "epoch": 0.11310776536614323,
      "grad_norm": 1.2464321068180946,
      "learning_rate": 9.718879452208836e-06,
      "loss": 0.5403,
      "step": 2096
    },
    {
      "epoch": 0.1131617289946576,
      "grad_norm": 1.0799884153596904,
      "learning_rate": 9.718613975247385e-06,
      "loss": 0.6415,
      "step": 2097
    },
    {
      "epoch": 0.11321569262317198,
      "grad_norm": 1.0544805841691753,
      "learning_rate": 9.718348377038705e-06,
      "loss": 0.4599,
      "step": 2098
    },
    {
      "epoch": 0.11326965625168636,
      "grad_norm": 1.0942206196361584,
      "learning_rate": 9.71808265759043e-06,
      "loss": 0.615,
      "step": 2099
    },
    {
      "epoch": 0.11332361988020075,
      "grad_norm": 1.178566236604381,
      "learning_rate": 9.7178168169102e-06,
      "loss": 0.4576,
      "step": 2100
    },
    {
      "epoch": 0.11337758350871513,
      "grad_norm": 1.1328586071696003,
      "learning_rate": 9.71755085500565e-06,
      "loss": 0.468,
      "step": 2101
    },
    {
      "epoch": 0.11343154713722951,
      "grad_norm": 0.7794337763447532,
      "learning_rate": 9.717284771884427e-06,
      "loss": 0.4533,
      "step": 2102
    },
    {
      "epoch": 0.11348551076574388,
      "grad_norm": 1.0031295924330543,
      "learning_rate": 9.71701856755418e-06,
      "loss": 0.4203,
      "step": 2103
    },
    {
      "epoch": 0.11353947439425827,
      "grad_norm": 0.966836137429084,
      "learning_rate": 9.716752242022557e-06,
      "loss": 0.4889,
      "step": 2104
    },
    {
      "epoch": 0.11359343802277265,
      "grad_norm": 1.1146599231197085,
      "learning_rate": 9.716485795297213e-06,
      "loss": 0.4987,
      "step": 2105
    },
    {
      "epoch": 0.11364740165128703,
      "grad_norm": 1.1189504959809826,
      "learning_rate": 9.716219227385809e-06,
      "loss": 0.4945,
      "step": 2106
    },
    {
      "epoch": 0.11370136527980142,
      "grad_norm": 1.1303991596120908,
      "learning_rate": 9.715952538296004e-06,
      "loss": 0.6319,
      "step": 2107
    },
    {
      "epoch": 0.1137553289083158,
      "grad_norm": 0.911771599076188,
      "learning_rate": 9.715685728035462e-06,
      "loss": 0.4308,
      "step": 2108
    },
    {
      "epoch": 0.11380929253683017,
      "grad_norm": 1.0914598689834962,
      "learning_rate": 9.715418796611854e-06,
      "loss": 0.4169,
      "step": 2109
    },
    {
      "epoch": 0.11386325616534455,
      "grad_norm": 0.8205296786086751,
      "learning_rate": 9.715151744032849e-06,
      "loss": 0.3976,
      "step": 2110
    },
    {
      "epoch": 0.11391721979385894,
      "grad_norm": 1.2377533912758019,
      "learning_rate": 9.714884570306123e-06,
      "loss": 0.493,
      "step": 2111
    },
    {
      "epoch": 0.11397118342237332,
      "grad_norm": 1.1933932254179174,
      "learning_rate": 9.714617275439358e-06,
      "loss": 0.4986,
      "step": 2112
    },
    {
      "epoch": 0.1140251470508877,
      "grad_norm": 1.094748515201122,
      "learning_rate": 9.714349859440232e-06,
      "loss": 0.5111,
      "step": 2113
    },
    {
      "epoch": 0.11407911067940209,
      "grad_norm": 1.1550967078214478,
      "learning_rate": 9.714082322316434e-06,
      "loss": 0.5567,
      "step": 2114
    },
    {
      "epoch": 0.11413307430791647,
      "grad_norm": 1.0381901687122022,
      "learning_rate": 9.713814664075652e-06,
      "loss": 0.4062,
      "step": 2115
    },
    {
      "epoch": 0.11418703793643084,
      "grad_norm": 0.8926215387923336,
      "learning_rate": 9.713546884725578e-06,
      "loss": 0.4334,
      "step": 2116
    },
    {
      "epoch": 0.11424100156494522,
      "grad_norm": 1.121271114812557,
      "learning_rate": 9.71327898427391e-06,
      "loss": 0.6181,
      "step": 2117
    },
    {
      "epoch": 0.11429496519345961,
      "grad_norm": 1.1486350849731022,
      "learning_rate": 9.713010962728346e-06,
      "loss": 0.549,
      "step": 2118
    },
    {
      "epoch": 0.11434892882197399,
      "grad_norm": 1.0531303113515242,
      "learning_rate": 9.71274282009659e-06,
      "loss": 0.5114,
      "step": 2119
    },
    {
      "epoch": 0.11440289245048837,
      "grad_norm": 1.0070433807681014,
      "learning_rate": 9.712474556386349e-06,
      "loss": 0.468,
      "step": 2120
    },
    {
      "epoch": 0.11445685607900276,
      "grad_norm": 0.8818546760986574,
      "learning_rate": 9.712206171605333e-06,
      "loss": 0.4355,
      "step": 2121
    },
    {
      "epoch": 0.11451081970751713,
      "grad_norm": 0.9920858563948186,
      "learning_rate": 9.711937665761255e-06,
      "loss": 0.4312,
      "step": 2122
    },
    {
      "epoch": 0.11456478333603151,
      "grad_norm": 1.0192576805920608,
      "learning_rate": 9.711669038861833e-06,
      "loss": 0.5495,
      "step": 2123
    },
    {
      "epoch": 0.1146187469645459,
      "grad_norm": 0.9418997920985216,
      "learning_rate": 9.711400290914787e-06,
      "loss": 0.3956,
      "step": 2124
    },
    {
      "epoch": 0.11467271059306028,
      "grad_norm": 1.1400140711658797,
      "learning_rate": 9.71113142192784e-06,
      "loss": 0.6297,
      "step": 2125
    },
    {
      "epoch": 0.11472667422157466,
      "grad_norm": 0.955892471460207,
      "learning_rate": 9.710862431908723e-06,
      "loss": 0.425,
      "step": 2126
    },
    {
      "epoch": 0.11478063785008905,
      "grad_norm": 1.2328722045887308,
      "learning_rate": 9.710593320865165e-06,
      "loss": 0.7923,
      "step": 2127
    },
    {
      "epoch": 0.11483460147860342,
      "grad_norm": 1.2988420613874465,
      "learning_rate": 9.710324088804899e-06,
      "loss": 0.653,
      "step": 2128
    },
    {
      "epoch": 0.1148885651071178,
      "grad_norm": 1.094077090228835,
      "learning_rate": 9.710054735735665e-06,
      "loss": 0.6028,
      "step": 2129
    },
    {
      "epoch": 0.11494252873563218,
      "grad_norm": 1.2222961951220337,
      "learning_rate": 9.709785261665205e-06,
      "loss": 0.4907,
      "step": 2130
    },
    {
      "epoch": 0.11499649236414657,
      "grad_norm": 0.9443231244877854,
      "learning_rate": 9.709515666601262e-06,
      "loss": 0.4208,
      "step": 2131
    },
    {
      "epoch": 0.11505045599266095,
      "grad_norm": 1.0563208397163193,
      "learning_rate": 9.709245950551586e-06,
      "loss": 0.5139,
      "step": 2132
    },
    {
      "epoch": 0.11510441962117533,
      "grad_norm": 0.928467590525105,
      "learning_rate": 9.708976113523929e-06,
      "loss": 0.4348,
      "step": 2133
    },
    {
      "epoch": 0.1151583832496897,
      "grad_norm": 1.0240504443384908,
      "learning_rate": 9.708706155526042e-06,
      "loss": 0.4389,
      "step": 2134
    },
    {
      "epoch": 0.11521234687820409,
      "grad_norm": 1.1788583005461306,
      "learning_rate": 9.70843607656569e-06,
      "loss": 0.4491,
      "step": 2135
    },
    {
      "epoch": 0.11526631050671847,
      "grad_norm": 1.0255206433579533,
      "learning_rate": 9.708165876650634e-06,
      "loss": 0.5015,
      "step": 2136
    },
    {
      "epoch": 0.11532027413523285,
      "grad_norm": 1.100632821414947,
      "learning_rate": 9.707895555788638e-06,
      "loss": 0.6681,
      "step": 2137
    },
    {
      "epoch": 0.11537423776374724,
      "grad_norm": 1.2760326420241366,
      "learning_rate": 9.707625113987468e-06,
      "loss": 0.7037,
      "step": 2138
    },
    {
      "epoch": 0.11542820139226162,
      "grad_norm": 1.184046177513834,
      "learning_rate": 9.707354551254906e-06,
      "loss": 0.5489,
      "step": 2139
    },
    {
      "epoch": 0.115482165020776,
      "grad_norm": 0.941891360003267,
      "learning_rate": 9.707083867598722e-06,
      "loss": 0.4095,
      "step": 2140
    },
    {
      "epoch": 0.11553612864929037,
      "grad_norm": 0.9592418910260238,
      "learning_rate": 9.706813063026694e-06,
      "loss": 0.5687,
      "step": 2141
    },
    {
      "epoch": 0.11559009227780476,
      "grad_norm": 1.216384524866234,
      "learning_rate": 9.70654213754661e-06,
      "loss": 0.6112,
      "step": 2142
    },
    {
      "epoch": 0.11564405590631914,
      "grad_norm": 1.1271086372560108,
      "learning_rate": 9.706271091166254e-06,
      "loss": 0.4728,
      "step": 2143
    },
    {
      "epoch": 0.11569801953483352,
      "grad_norm": 1.035078179931676,
      "learning_rate": 9.705999923893417e-06,
      "loss": 0.558,
      "step": 2144
    },
    {
      "epoch": 0.11575198316334791,
      "grad_norm": 1.1459829936066768,
      "learning_rate": 9.705728635735892e-06,
      "loss": 0.5734,
      "step": 2145
    },
    {
      "epoch": 0.11580594679186229,
      "grad_norm": 1.025290354986629,
      "learning_rate": 9.705457226701476e-06,
      "loss": 0.4987,
      "step": 2146
    },
    {
      "epoch": 0.11585991042037666,
      "grad_norm": 1.1494116639809646,
      "learning_rate": 9.705185696797972e-06,
      "loss": 0.5385,
      "step": 2147
    },
    {
      "epoch": 0.11591387404889104,
      "grad_norm": 0.9120287383364258,
      "learning_rate": 9.70491404603318e-06,
      "loss": 0.6142,
      "step": 2148
    },
    {
      "epoch": 0.11596783767740543,
      "grad_norm": 1.0465135255117215,
      "learning_rate": 9.70464227441491e-06,
      "loss": 0.4155,
      "step": 2149
    },
    {
      "epoch": 0.11602180130591981,
      "grad_norm": 1.0130069439592695,
      "learning_rate": 9.704370381950974e-06,
      "loss": 0.4597,
      "step": 2150
    },
    {
      "epoch": 0.1160757649344342,
      "grad_norm": 1.1512564818300701,
      "learning_rate": 9.704098368649181e-06,
      "loss": 0.6123,
      "step": 2151
    },
    {
      "epoch": 0.11612972856294858,
      "grad_norm": 1.2423992686953171,
      "learning_rate": 9.703826234517358e-06,
      "loss": 0.5082,
      "step": 2152
    },
    {
      "epoch": 0.11618369219146295,
      "grad_norm": 0.8925834830733531,
      "learning_rate": 9.70355397956332e-06,
      "loss": 0.4523,
      "step": 2153
    },
    {
      "epoch": 0.11623765581997733,
      "grad_norm": 0.967594413798547,
      "learning_rate": 9.70328160379489e-06,
      "loss": 0.443,
      "step": 2154
    },
    {
      "epoch": 0.11629161944849171,
      "grad_norm": 0.8453519442799124,
      "learning_rate": 9.703009107219904e-06,
      "loss": 0.3905,
      "step": 2155
    },
    {
      "epoch": 0.1163455830770061,
      "grad_norm": 1.1059504958176312,
      "learning_rate": 9.702736489846187e-06,
      "loss": 0.5401,
      "step": 2156
    },
    {
      "epoch": 0.11639954670552048,
      "grad_norm": 1.3918652504879416,
      "learning_rate": 9.702463751681577e-06,
      "loss": 0.6284,
      "step": 2157
    },
    {
      "epoch": 0.11645351033403487,
      "grad_norm": 0.9582342909149959,
      "learning_rate": 9.702190892733913e-06,
      "loss": 0.4074,
      "step": 2158
    },
    {
      "epoch": 0.11650747396254924,
      "grad_norm": 1.035499039427595,
      "learning_rate": 9.701917913011035e-06,
      "loss": 0.3915,
      "step": 2159
    },
    {
      "epoch": 0.11656143759106362,
      "grad_norm": 1.2348681022612513,
      "learning_rate": 9.701644812520792e-06,
      "loss": 0.5226,
      "step": 2160
    },
    {
      "epoch": 0.116615401219578,
      "grad_norm": 1.0316289082892915,
      "learning_rate": 9.701371591271031e-06,
      "loss": 0.4473,
      "step": 2161
    },
    {
      "epoch": 0.11666936484809239,
      "grad_norm": 1.0642023489563224,
      "learning_rate": 9.701098249269606e-06,
      "loss": 0.4486,
      "step": 2162
    },
    {
      "epoch": 0.11672332847660677,
      "grad_norm": 1.026526047463095,
      "learning_rate": 9.700824786524372e-06,
      "loss": 0.5221,
      "step": 2163
    },
    {
      "epoch": 0.11677729210512115,
      "grad_norm": 1.135013428447993,
      "learning_rate": 9.70055120304319e-06,
      "loss": 0.4675,
      "step": 2164
    },
    {
      "epoch": 0.11683125573363554,
      "grad_norm": 1.0201577017599577,
      "learning_rate": 9.700277498833921e-06,
      "loss": 0.5379,
      "step": 2165
    },
    {
      "epoch": 0.1168852193621499,
      "grad_norm": 0.9371695966850042,
      "learning_rate": 9.700003673904433e-06,
      "loss": 0.5225,
      "step": 2166
    },
    {
      "epoch": 0.11693918299066429,
      "grad_norm": 1.0880857939485777,
      "learning_rate": 9.699729728262596e-06,
      "loss": 0.529,
      "step": 2167
    },
    {
      "epoch": 0.11699314661917867,
      "grad_norm": 1.0726276715929168,
      "learning_rate": 9.699455661916281e-06,
      "loss": 0.5636,
      "step": 2168
    },
    {
      "epoch": 0.11704711024769306,
      "grad_norm": 0.9398092630930535,
      "learning_rate": 9.69918147487337e-06,
      "loss": 0.4371,
      "step": 2169
    },
    {
      "epoch": 0.11710107387620744,
      "grad_norm": 0.8844183672272802,
      "learning_rate": 9.698907167141739e-06,
      "loss": 0.4259,
      "step": 2170
    },
    {
      "epoch": 0.11715503750472182,
      "grad_norm": 1.140494840944878,
      "learning_rate": 9.698632738729273e-06,
      "loss": 0.5548,
      "step": 2171
    },
    {
      "epoch": 0.1172090011332362,
      "grad_norm": 1.099954541875123,
      "learning_rate": 9.69835818964386e-06,
      "loss": 0.4904,
      "step": 2172
    },
    {
      "epoch": 0.11726296476175058,
      "grad_norm": 1.0900031468786004,
      "learning_rate": 9.69808351989339e-06,
      "loss": 0.6291,
      "step": 2173
    },
    {
      "epoch": 0.11731692839026496,
      "grad_norm": 1.0851603119621536,
      "learning_rate": 9.69780872948576e-06,
      "loss": 0.4439,
      "step": 2174
    },
    {
      "epoch": 0.11737089201877934,
      "grad_norm": 0.9631078378601368,
      "learning_rate": 9.697533818428863e-06,
      "loss": 0.4114,
      "step": 2175
    },
    {
      "epoch": 0.11742485564729373,
      "grad_norm": 0.9950717121723265,
      "learning_rate": 9.697258786730603e-06,
      "loss": 0.5193,
      "step": 2176
    },
    {
      "epoch": 0.11747881927580811,
      "grad_norm": 1.22702270470133,
      "learning_rate": 9.696983634398886e-06,
      "loss": 0.6386,
      "step": 2177
    },
    {
      "epoch": 0.11753278290432248,
      "grad_norm": 1.0893143520594046,
      "learning_rate": 9.696708361441617e-06,
      "loss": 0.5014,
      "step": 2178
    },
    {
      "epoch": 0.11758674653283686,
      "grad_norm": 0.9031315969572481,
      "learning_rate": 9.696432967866712e-06,
      "loss": 0.4056,
      "step": 2179
    },
    {
      "epoch": 0.11764071016135125,
      "grad_norm": 0.9875513301486594,
      "learning_rate": 9.69615745368208e-06,
      "loss": 0.5105,
      "step": 2180
    },
    {
      "epoch": 0.11769467378986563,
      "grad_norm": 1.0112005447768027,
      "learning_rate": 9.695881818895642e-06,
      "loss": 0.4693,
      "step": 2181
    },
    {
      "epoch": 0.11774863741838001,
      "grad_norm": 0.9654037146706186,
      "learning_rate": 9.695606063515325e-06,
      "loss": 0.3687,
      "step": 2182
    },
    {
      "epoch": 0.1178026010468944,
      "grad_norm": 1.2173417540977598,
      "learning_rate": 9.695330187549048e-06,
      "loss": 0.5959,
      "step": 2183
    },
    {
      "epoch": 0.11785656467540877,
      "grad_norm": 0.8912592311311957,
      "learning_rate": 9.695054191004743e-06,
      "loss": 0.4056,
      "step": 2184
    },
    {
      "epoch": 0.11791052830392315,
      "grad_norm": 1.0473782442352688,
      "learning_rate": 9.694778073890342e-06,
      "loss": 0.4809,
      "step": 2185
    },
    {
      "epoch": 0.11796449193243753,
      "grad_norm": 0.9326417039400774,
      "learning_rate": 9.69450183621378e-06,
      "loss": 0.4508,
      "step": 2186
    },
    {
      "epoch": 0.11801845556095192,
      "grad_norm": 1.04928607430521,
      "learning_rate": 9.694225477982996e-06,
      "loss": 0.5075,
      "step": 2187
    },
    {
      "epoch": 0.1180724191894663,
      "grad_norm": 1.0011047276138676,
      "learning_rate": 9.693948999205935e-06,
      "loss": 0.7541,
      "step": 2188
    },
    {
      "epoch": 0.11812638281798069,
      "grad_norm": 1.3456540891225421,
      "learning_rate": 9.693672399890543e-06,
      "loss": 0.7273,
      "step": 2189
    },
    {
      "epoch": 0.11818034644649507,
      "grad_norm": 1.2161558121391915,
      "learning_rate": 9.693395680044766e-06,
      "loss": 0.5821,
      "step": 2190
    },
    {
      "epoch": 0.11823431007500944,
      "grad_norm": 0.9572900643636869,
      "learning_rate": 9.693118839676562e-06,
      "loss": 0.4236,
      "step": 2191
    },
    {
      "epoch": 0.11828827370352382,
      "grad_norm": 1.0412734889470079,
      "learning_rate": 9.692841878793888e-06,
      "loss": 0.6252,
      "step": 2192
    },
    {
      "epoch": 0.1183422373320382,
      "grad_norm": 1.122067561259171,
      "learning_rate": 9.692564797404698e-06,
      "loss": 0.565,
      "step": 2193
    },
    {
      "epoch": 0.11839620096055259,
      "grad_norm": 1.0317664850900574,
      "learning_rate": 9.69228759551696e-06,
      "loss": 0.4377,
      "step": 2194
    },
    {
      "epoch": 0.11845016458906697,
      "grad_norm": 1.2601597990560485,
      "learning_rate": 9.692010273138642e-06,
      "loss": 0.5287,
      "step": 2195
    },
    {
      "epoch": 0.11850412821758136,
      "grad_norm": 1.08477682890788,
      "learning_rate": 9.691732830277712e-06,
      "loss": 0.6646,
      "step": 2196
    },
    {
      "epoch": 0.11855809184609573,
      "grad_norm": 1.0696842005743334,
      "learning_rate": 9.691455266942146e-06,
      "loss": 0.4496,
      "step": 2197
    },
    {
      "epoch": 0.11861205547461011,
      "grad_norm": 1.026353876095435,
      "learning_rate": 9.69117758313992e-06,
      "loss": 0.5071,
      "step": 2198
    },
    {
      "epoch": 0.11866601910312449,
      "grad_norm": 1.051444886850701,
      "learning_rate": 9.690899778879016e-06,
      "loss": 0.4464,
      "step": 2199
    },
    {
      "epoch": 0.11871998273163888,
      "grad_norm": 1.1875435174597502,
      "learning_rate": 9.690621854167415e-06,
      "loss": 0.5663,
      "step": 2200
    },
    {
      "epoch": 0.11877394636015326,
      "grad_norm": 1.131346145204395,
      "learning_rate": 9.690343809013111e-06,
      "loss": 0.6198,
      "step": 2201
    },
    {
      "epoch": 0.11882790998866764,
      "grad_norm": 1.1486586422763523,
      "learning_rate": 9.69006564342409e-06,
      "loss": 0.427,
      "step": 2202
    },
    {
      "epoch": 0.11888187361718201,
      "grad_norm": 1.0837775288585423,
      "learning_rate": 9.689787357408348e-06,
      "loss": 0.4634,
      "step": 2203
    },
    {
      "epoch": 0.1189358372456964,
      "grad_norm": 1.3740114987384053,
      "learning_rate": 9.689508950973888e-06,
      "loss": 0.72,
      "step": 2204
    },
    {
      "epoch": 0.11898980087421078,
      "grad_norm": 1.162623860963007,
      "learning_rate": 9.689230424128703e-06,
      "loss": 0.5259,
      "step": 2205
    },
    {
      "epoch": 0.11904376450272516,
      "grad_norm": 1.1643884469495673,
      "learning_rate": 9.688951776880805e-06,
      "loss": 0.5399,
      "step": 2206
    },
    {
      "epoch": 0.11909772813123955,
      "grad_norm": 1.059434341566082,
      "learning_rate": 9.688673009238199e-06,
      "loss": 0.5182,
      "step": 2207
    },
    {
      "epoch": 0.11915169175975393,
      "grad_norm": 1.1165082545626912,
      "learning_rate": 9.688394121208899e-06,
      "loss": 0.6969,
      "step": 2208
    },
    {
      "epoch": 0.11920565538826831,
      "grad_norm": 1.0883483656850015,
      "learning_rate": 9.688115112800922e-06,
      "loss": 0.6342,
      "step": 2209
    },
    {
      "epoch": 0.11925961901678268,
      "grad_norm": 0.9494426199177266,
      "learning_rate": 9.687835984022281e-06,
      "loss": 0.6165,
      "step": 2210
    },
    {
      "epoch": 0.11931358264529707,
      "grad_norm": 0.9686385992805834,
      "learning_rate": 9.687556734881006e-06,
      "loss": 0.4665,
      "step": 2211
    },
    {
      "epoch": 0.11936754627381145,
      "grad_norm": 1.157418220068164,
      "learning_rate": 9.687277365385116e-06,
      "loss": 0.6218,
      "step": 2212
    },
    {
      "epoch": 0.11942150990232583,
      "grad_norm": 1.0293779911615994,
      "learning_rate": 9.686997875542644e-06,
      "loss": 0.4685,
      "step": 2213
    },
    {
      "epoch": 0.11947547353084022,
      "grad_norm": 1.5861511364682184,
      "learning_rate": 9.686718265361625e-06,
      "loss": 0.7018,
      "step": 2214
    },
    {
      "epoch": 0.1195294371593546,
      "grad_norm": 1.0054865831088418,
      "learning_rate": 9.68643853485009e-06,
      "loss": 0.3506,
      "step": 2215
    },
    {
      "epoch": 0.11958340078786897,
      "grad_norm": 1.2206746111948432,
      "learning_rate": 9.686158684016084e-06,
      "loss": 0.6254,
      "step": 2216
    },
    {
      "epoch": 0.11963736441638335,
      "grad_norm": 0.9718927206405142,
      "learning_rate": 9.685878712867644e-06,
      "loss": 0.4597,
      "step": 2217
    },
    {
      "epoch": 0.11969132804489774,
      "grad_norm": 1.0444441606983377,
      "learning_rate": 9.685598621412824e-06,
      "loss": 0.5662,
      "step": 2218
    },
    {
      "epoch": 0.11974529167341212,
      "grad_norm": 0.9501010885692707,
      "learning_rate": 9.685318409659668e-06,
      "loss": 0.4915,
      "step": 2219
    },
    {
      "epoch": 0.1197992553019265,
      "grad_norm": 1.137605788339061,
      "learning_rate": 9.685038077616231e-06,
      "loss": 0.5529,
      "step": 2220
    },
    {
      "epoch": 0.11985321893044089,
      "grad_norm": 1.0397386718597894,
      "learning_rate": 9.684757625290574e-06,
      "loss": 0.6095,
      "step": 2221
    },
    {
      "epoch": 0.11990718255895526,
      "grad_norm": 0.9318530954776885,
      "learning_rate": 9.684477052690754e-06,
      "loss": 0.3826,
      "step": 2222
    },
    {
      "epoch": 0.11996114618746964,
      "grad_norm": 1.0119627906820465,
      "learning_rate": 9.684196359824833e-06,
      "loss": 0.569,
      "step": 2223
    },
    {
      "epoch": 0.12001510981598403,
      "grad_norm": 1.1090213373768198,
      "learning_rate": 9.683915546700883e-06,
      "loss": 0.623,
      "step": 2224
    },
    {
      "epoch": 0.12006907344449841,
      "grad_norm": 0.9471935628760784,
      "learning_rate": 9.683634613326972e-06,
      "loss": 0.4909,
      "step": 2225
    },
    {
      "epoch": 0.12012303707301279,
      "grad_norm": 1.3845759877807065,
      "learning_rate": 9.683353559711175e-06,
      "loss": 0.598,
      "step": 2226
    },
    {
      "epoch": 0.12017700070152718,
      "grad_norm": 0.8221055747147173,
      "learning_rate": 9.683072385861569e-06,
      "loss": 0.4061,
      "step": 2227
    },
    {
      "epoch": 0.12023096433004155,
      "grad_norm": 0.9664110823160382,
      "learning_rate": 9.682791091786237e-06,
      "loss": 0.504,
      "step": 2228
    },
    {
      "epoch": 0.12028492795855593,
      "grad_norm": 1.079246691711793,
      "learning_rate": 9.68250967749326e-06,
      "loss": 0.456,
      "step": 2229
    },
    {
      "epoch": 0.12033889158707031,
      "grad_norm": 1.1104342349265683,
      "learning_rate": 9.682228142990732e-06,
      "loss": 0.5795,
      "step": 2230
    },
    {
      "epoch": 0.1203928552155847,
      "grad_norm": 1.2027124158240485,
      "learning_rate": 9.68194648828674e-06,
      "loss": 0.7313,
      "step": 2231
    },
    {
      "epoch": 0.12044681884409908,
      "grad_norm": 1.0009290953727872,
      "learning_rate": 9.681664713389382e-06,
      "loss": 0.4615,
      "step": 2232
    },
    {
      "epoch": 0.12050078247261346,
      "grad_norm": 0.9520443193405584,
      "learning_rate": 9.681382818306754e-06,
      "loss": 0.433,
      "step": 2233
    },
    {
      "epoch": 0.12055474610112785,
      "grad_norm": 0.8669540435783489,
      "learning_rate": 9.681100803046956e-06,
      "loss": 0.4159,
      "step": 2234
    },
    {
      "epoch": 0.12060870972964222,
      "grad_norm": 1.4765173908952733,
      "learning_rate": 9.680818667618101e-06,
      "loss": 0.6846,
      "step": 2235
    },
    {
      "epoch": 0.1206626733581566,
      "grad_norm": 1.0214499606258602,
      "learning_rate": 9.680536412028291e-06,
      "loss": 0.4474,
      "step": 2236
    },
    {
      "epoch": 0.12071663698667098,
      "grad_norm": 1.2052665544356185,
      "learning_rate": 9.680254036285638e-06,
      "loss": 0.7253,
      "step": 2237
    },
    {
      "epoch": 0.12077060061518537,
      "grad_norm": 1.1073765270459122,
      "learning_rate": 9.679971540398263e-06,
      "loss": 0.7594,
      "step": 2238
    },
    {
      "epoch": 0.12082456424369975,
      "grad_norm": 0.8110268777075951,
      "learning_rate": 9.679688924374283e-06,
      "loss": 0.4053,
      "step": 2239
    },
    {
      "epoch": 0.12087852787221413,
      "grad_norm": 1.1120168263411978,
      "learning_rate": 9.679406188221818e-06,
      "loss": 0.5394,
      "step": 2240
    },
    {
      "epoch": 0.1209324915007285,
      "grad_norm": 1.3392086023213707,
      "learning_rate": 9.679123331948995e-06,
      "loss": 0.5211,
      "step": 2241
    },
    {
      "epoch": 0.12098645512924289,
      "grad_norm": 1.0364699707537088,
      "learning_rate": 9.678840355563947e-06,
      "loss": 0.5329,
      "step": 2242
    },
    {
      "epoch": 0.12104041875775727,
      "grad_norm": 1.157960590526774,
      "learning_rate": 9.678557259074805e-06,
      "loss": 0.5674,
      "step": 2243
    },
    {
      "epoch": 0.12109438238627165,
      "grad_norm": 1.1101877591724363,
      "learning_rate": 9.678274042489704e-06,
      "loss": 0.4706,
      "step": 2244
    },
    {
      "epoch": 0.12114834601478604,
      "grad_norm": 1.1488828078726339,
      "learning_rate": 9.677990705816787e-06,
      "loss": 0.496,
      "step": 2245
    },
    {
      "epoch": 0.12120230964330042,
      "grad_norm": 0.9298744766980935,
      "learning_rate": 9.677707249064193e-06,
      "loss": 0.3834,
      "step": 2246
    },
    {
      "epoch": 0.12125627327181479,
      "grad_norm": 1.2496939382762586,
      "learning_rate": 9.677423672240073e-06,
      "loss": 0.5499,
      "step": 2247
    },
    {
      "epoch": 0.12131023690032917,
      "grad_norm": 1.1698730037388965,
      "learning_rate": 9.677139975352577e-06,
      "loss": 0.5242,
      "step": 2248
    },
    {
      "epoch": 0.12136420052884356,
      "grad_norm": 1.142238733423812,
      "learning_rate": 9.676856158409854e-06,
      "loss": 0.3603,
      "step": 2249
    },
    {
      "epoch": 0.12141816415735794,
      "grad_norm": 1.229077501052302,
      "learning_rate": 9.676572221420067e-06,
      "loss": 0.5999,
      "step": 2250
    },
    {
      "epoch": 0.12147212778587232,
      "grad_norm": 1.1711008649042807,
      "learning_rate": 9.676288164391373e-06,
      "loss": 0.552,
      "step": 2251
    },
    {
      "epoch": 0.12152609141438671,
      "grad_norm": 1.062514411032166,
      "learning_rate": 9.676003987331939e-06,
      "loss": 0.5048,
      "step": 2252
    },
    {
      "epoch": 0.12158005504290108,
      "grad_norm": 1.1474293897922299,
      "learning_rate": 9.67571969024993e-06,
      "loss": 0.4124,
      "step": 2253
    },
    {
      "epoch": 0.12163401867141546,
      "grad_norm": 1.0463006830783068,
      "learning_rate": 9.675435273153515e-06,
      "loss": 0.4837,
      "step": 2254
    },
    {
      "epoch": 0.12168798229992984,
      "grad_norm": 1.1629971896909328,
      "learning_rate": 9.675150736050875e-06,
      "loss": 0.6153,
      "step": 2255
    },
    {
      "epoch": 0.12174194592844423,
      "grad_norm": 1.185342591314412,
      "learning_rate": 9.674866078950184e-06,
      "loss": 0.5341,
      "step": 2256
    },
    {
      "epoch": 0.12179590955695861,
      "grad_norm": 1.2347685339359427,
      "learning_rate": 9.674581301859624e-06,
      "loss": 0.656,
      "step": 2257
    },
    {
      "epoch": 0.121849873185473,
      "grad_norm": 0.831253140334588,
      "learning_rate": 9.674296404787376e-06,
      "loss": 0.3719,
      "step": 2258
    },
    {
      "epoch": 0.12190383681398738,
      "grad_norm": 0.9277944807722457,
      "learning_rate": 9.674011387741635e-06,
      "loss": 0.4361,
      "step": 2259
    },
    {
      "epoch": 0.12195780044250175,
      "grad_norm": 1.0897314234304791,
      "learning_rate": 9.673726250730587e-06,
      "loss": 0.4018,
      "step": 2260
    },
    {
      "epoch": 0.12201176407101613,
      "grad_norm": 0.8230387588679382,
      "learning_rate": 9.673440993762429e-06,
      "loss": 0.3782,
      "step": 2261
    },
    {
      "epoch": 0.12206572769953052,
      "grad_norm": 0.9421259183686265,
      "learning_rate": 9.673155616845362e-06,
      "loss": 0.5124,
      "step": 2262
    },
    {
      "epoch": 0.1221196913280449,
      "grad_norm": 0.907467768225828,
      "learning_rate": 9.672870119987584e-06,
      "loss": 0.428,
      "step": 2263
    },
    {
      "epoch": 0.12217365495655928,
      "grad_norm": 1.5072659227187655,
      "learning_rate": 9.672584503197303e-06,
      "loss": 0.5897,
      "step": 2264
    },
    {
      "epoch": 0.12222761858507367,
      "grad_norm": 0.9317382001346334,
      "learning_rate": 9.672298766482728e-06,
      "loss": 0.4433,
      "step": 2265
    },
    {
      "epoch": 0.12228158221358804,
      "grad_norm": 1.0146027636011954,
      "learning_rate": 9.67201290985207e-06,
      "loss": 0.5124,
      "step": 2266
    },
    {
      "epoch": 0.12233554584210242,
      "grad_norm": 1.0785171894259167,
      "learning_rate": 9.671726933313545e-06,
      "loss": 0.647,
      "step": 2267
    },
    {
      "epoch": 0.1223895094706168,
      "grad_norm": 1.1040007523715156,
      "learning_rate": 9.671440836875374e-06,
      "loss": 0.5698,
      "step": 2268
    },
    {
      "epoch": 0.12244347309913119,
      "grad_norm": 1.0678578174289854,
      "learning_rate": 9.671154620545775e-06,
      "loss": 0.5571,
      "step": 2269
    },
    {
      "epoch": 0.12249743672764557,
      "grad_norm": 1.0381194634940218,
      "learning_rate": 9.670868284332982e-06,
      "loss": 0.46,
      "step": 2270
    },
    {
      "epoch": 0.12255140035615995,
      "grad_norm": 1.1021253867203855,
      "learning_rate": 9.670581828245218e-06,
      "loss": 0.5695,
      "step": 2271
    },
    {
      "epoch": 0.12260536398467432,
      "grad_norm": 1.0731139164555994,
      "learning_rate": 9.670295252290717e-06,
      "loss": 0.5801,
      "step": 2272
    },
    {
      "epoch": 0.1226593276131887,
      "grad_norm": 0.866065270712914,
      "learning_rate": 9.67000855647772e-06,
      "loss": 0.3971,
      "step": 2273
    },
    {
      "epoch": 0.12271329124170309,
      "grad_norm": 0.9131932290987935,
      "learning_rate": 9.669721740814462e-06,
      "loss": 0.4184,
      "step": 2274
    },
    {
      "epoch": 0.12276725487021747,
      "grad_norm": 0.9333520435832217,
      "learning_rate": 9.669434805309187e-06,
      "loss": 0.4704,
      "step": 2275
    },
    {
      "epoch": 0.12282121849873186,
      "grad_norm": 0.9112828751784572,
      "learning_rate": 9.669147749970143e-06,
      "loss": 0.4396,
      "step": 2276
    },
    {
      "epoch": 0.12287518212724624,
      "grad_norm": 1.252045603114486,
      "learning_rate": 9.668860574805581e-06,
      "loss": 0.5901,
      "step": 2277
    },
    {
      "epoch": 0.12292914575576061,
      "grad_norm": 1.040656827069334,
      "learning_rate": 9.668573279823753e-06,
      "loss": 0.5295,
      "step": 2278
    },
    {
      "epoch": 0.122983109384275,
      "grad_norm": 1.1830336186386003,
      "learning_rate": 9.668285865032917e-06,
      "loss": 0.5786,
      "step": 2279
    },
    {
      "epoch": 0.12303707301278938,
      "grad_norm": 1.2300312025901774,
      "learning_rate": 9.667998330441335e-06,
      "loss": 0.5892,
      "step": 2280
    },
    {
      "epoch": 0.12309103664130376,
      "grad_norm": 1.0161061153877486,
      "learning_rate": 9.667710676057268e-06,
      "loss": 0.4717,
      "step": 2281
    },
    {
      "epoch": 0.12314500026981814,
      "grad_norm": 0.9500036499035054,
      "learning_rate": 9.667422901888985e-06,
      "loss": 0.5285,
      "step": 2282
    },
    {
      "epoch": 0.12319896389833253,
      "grad_norm": 1.082961516705248,
      "learning_rate": 9.667135007944757e-06,
      "loss": 0.611,
      "step": 2283
    },
    {
      "epoch": 0.12325292752684691,
      "grad_norm": 1.5818763667705031,
      "learning_rate": 9.66684699423286e-06,
      "loss": 0.8236,
      "step": 2284
    },
    {
      "epoch": 0.12330689115536128,
      "grad_norm": 1.091196762263724,
      "learning_rate": 9.666558860761567e-06,
      "loss": 0.5574,
      "step": 2285
    },
    {
      "epoch": 0.12336085478387566,
      "grad_norm": 0.8856862786818642,
      "learning_rate": 9.666270607539163e-06,
      "loss": 0.424,
      "step": 2286
    },
    {
      "epoch": 0.12341481841239005,
      "grad_norm": 1.1437200014543105,
      "learning_rate": 9.665982234573936e-06,
      "loss": 0.5005,
      "step": 2287
    },
    {
      "epoch": 0.12346878204090443,
      "grad_norm": 1.1362864378908073,
      "learning_rate": 9.665693741874166e-06,
      "loss": 0.4611,
      "step": 2288
    },
    {
      "epoch": 0.12352274566941882,
      "grad_norm": 1.1231989414190497,
      "learning_rate": 9.665405129448148e-06,
      "loss": 0.5942,
      "step": 2289
    },
    {
      "epoch": 0.1235767092979332,
      "grad_norm": 0.8697400029417068,
      "learning_rate": 9.66511639730418e-06,
      "loss": 0.4915,
      "step": 2290
    },
    {
      "epoch": 0.12363067292644757,
      "grad_norm": 1.3202433195580263,
      "learning_rate": 9.664827545450558e-06,
      "loss": 0.6408,
      "step": 2291
    },
    {
      "epoch": 0.12368463655496195,
      "grad_norm": 1.108800565583756,
      "learning_rate": 9.664538573895584e-06,
      "loss": 0.4425,
      "step": 2292
    },
    {
      "epoch": 0.12373860018347634,
      "grad_norm": 0.8900464895758888,
      "learning_rate": 9.664249482647563e-06,
      "loss": 0.4278,
      "step": 2293
    },
    {
      "epoch": 0.12379256381199072,
      "grad_norm": 1.0069899345340934,
      "learning_rate": 9.663960271714805e-06,
      "loss": 0.3956,
      "step": 2294
    },
    {
      "epoch": 0.1238465274405051,
      "grad_norm": 1.0908276370743775,
      "learning_rate": 9.66367094110562e-06,
      "loss": 0.5027,
      "step": 2295
    },
    {
      "epoch": 0.12390049106901949,
      "grad_norm": 1.2590867700383381,
      "learning_rate": 9.663381490828326e-06,
      "loss": 0.4648,
      "step": 2296
    },
    {
      "epoch": 0.12395445469753386,
      "grad_norm": 0.9943299838390417,
      "learning_rate": 9.663091920891241e-06,
      "loss": 0.5348,
      "step": 2297
    },
    {
      "epoch": 0.12400841832604824,
      "grad_norm": 0.895696595881996,
      "learning_rate": 9.662802231302687e-06,
      "loss": 0.4343,
      "step": 2298
    },
    {
      "epoch": 0.12406238195456262,
      "grad_norm": 0.988818066946634,
      "learning_rate": 9.662512422070994e-06,
      "loss": 0.4817,
      "step": 2299
    },
    {
      "epoch": 0.124116345583077,
      "grad_norm": 1.0628598546050112,
      "learning_rate": 9.662222493204485e-06,
      "loss": 0.4844,
      "step": 2300
    },
    {
      "epoch": 0.12417030921159139,
      "grad_norm": 1.0788173755181631,
      "learning_rate": 9.661932444711495e-06,
      "loss": 0.4939,
      "step": 2301
    },
    {
      "epoch": 0.12422427284010577,
      "grad_norm": 1.130407900785564,
      "learning_rate": 9.661642276600365e-06,
      "loss": 0.5054,
      "step": 2302
    },
    {
      "epoch": 0.12427823646862016,
      "grad_norm": 1.0700600279021275,
      "learning_rate": 9.661351988879427e-06,
      "loss": 0.4695,
      "step": 2303
    },
    {
      "epoch": 0.12433220009713453,
      "grad_norm": 1.2969314259022702,
      "learning_rate": 9.661061581557031e-06,
      "loss": 0.6024,
      "step": 2304
    },
    {
      "epoch": 0.12438616372564891,
      "grad_norm": 1.1191176049592504,
      "learning_rate": 9.66077105464152e-06,
      "loss": 0.4106,
      "step": 2305
    },
    {
      "epoch": 0.1244401273541633,
      "grad_norm": 0.971310521576742,
      "learning_rate": 9.660480408141244e-06,
      "loss": 0.4564,
      "step": 2306
    },
    {
      "epoch": 0.12449409098267768,
      "grad_norm": 1.0429586913915552,
      "learning_rate": 9.660189642064558e-06,
      "loss": 0.4254,
      "step": 2307
    },
    {
      "epoch": 0.12454805461119206,
      "grad_norm": 1.1586945413085292,
      "learning_rate": 9.659898756419817e-06,
      "loss": 0.4639,
      "step": 2308
    },
    {
      "epoch": 0.12460201823970644,
      "grad_norm": 1.230196555922898,
      "learning_rate": 9.659607751215383e-06,
      "loss": 0.7129,
      "step": 2309
    },
    {
      "epoch": 0.12465598186822081,
      "grad_norm": 1.0697726836246857,
      "learning_rate": 9.65931662645962e-06,
      "loss": 0.4951,
      "step": 2310
    },
    {
      "epoch": 0.1247099454967352,
      "grad_norm": 1.1048557328586497,
      "learning_rate": 9.659025382160893e-06,
      "loss": 0.4879,
      "step": 2311
    },
    {
      "epoch": 0.12476390912524958,
      "grad_norm": 1.0084590359615,
      "learning_rate": 9.658734018327573e-06,
      "loss": 0.3901,
      "step": 2312
    },
    {
      "epoch": 0.12481787275376396,
      "grad_norm": 1.2009920220889132,
      "learning_rate": 9.65844253496804e-06,
      "loss": 0.7065,
      "step": 2313
    },
    {
      "epoch": 0.12487183638227835,
      "grad_norm": 0.986576097524222,
      "learning_rate": 9.658150932090663e-06,
      "loss": 0.4532,
      "step": 2314
    },
    {
      "epoch": 0.12492580001079273,
      "grad_norm": 0.8220129343918802,
      "learning_rate": 9.657859209703827e-06,
      "loss": 0.3863,
      "step": 2315
    },
    {
      "epoch": 0.1249797636393071,
      "grad_norm": 1.0604531856680535,
      "learning_rate": 9.657567367815916e-06,
      "loss": 0.5317,
      "step": 2316
    },
    {
      "epoch": 0.1250337272678215,
      "grad_norm": 1.0088592003408487,
      "learning_rate": 9.657275406435319e-06,
      "loss": 0.5537,
      "step": 2317
    },
    {
      "epoch": 0.12508769089633587,
      "grad_norm": 1.1099986590988968,
      "learning_rate": 9.656983325570426e-06,
      "loss": 0.5502,
      "step": 2318
    },
    {
      "epoch": 0.12514165452485024,
      "grad_norm": 0.8910825962336899,
      "learning_rate": 9.65669112522963e-06,
      "loss": 0.4911,
      "step": 2319
    },
    {
      "epoch": 0.12519561815336464,
      "grad_norm": 1.121673789514824,
      "learning_rate": 9.656398805421335e-06,
      "loss": 0.5092,
      "step": 2320
    },
    {
      "epoch": 0.125249581781879,
      "grad_norm": 1.0019397725685486,
      "learning_rate": 9.656106366153936e-06,
      "loss": 0.5038,
      "step": 2321
    },
    {
      "epoch": 0.1253035454103934,
      "grad_norm": 1.0513763789138886,
      "learning_rate": 9.655813807435841e-06,
      "loss": 0.5735,
      "step": 2322
    },
    {
      "epoch": 0.12535750903890777,
      "grad_norm": 0.9652003909575069,
      "learning_rate": 9.655521129275458e-06,
      "loss": 0.4019,
      "step": 2323
    },
    {
      "epoch": 0.12541147266742217,
      "grad_norm": 1.166351544480196,
      "learning_rate": 9.655228331681197e-06,
      "loss": 0.5495,
      "step": 2324
    },
    {
      "epoch": 0.12546543629593654,
      "grad_norm": 1.109953569517171,
      "learning_rate": 9.654935414661478e-06,
      "loss": 0.5282,
      "step": 2325
    },
    {
      "epoch": 0.1255193999244509,
      "grad_norm": 0.7984846021363996,
      "learning_rate": 9.654642378224716e-06,
      "loss": 0.3519,
      "step": 2326
    },
    {
      "epoch": 0.1255733635529653,
      "grad_norm": 1.059791634150416,
      "learning_rate": 9.654349222379334e-06,
      "loss": 0.6065,
      "step": 2327
    },
    {
      "epoch": 0.12562732718147968,
      "grad_norm": 1.1047908768219616,
      "learning_rate": 9.654055947133758e-06,
      "loss": 0.511,
      "step": 2328
    },
    {
      "epoch": 0.12568129080999407,
      "grad_norm": 1.2578559335080715,
      "learning_rate": 9.653762552496417e-06,
      "loss": 0.7344,
      "step": 2329
    },
    {
      "epoch": 0.12573525443850844,
      "grad_norm": 1.3406460761544932,
      "learning_rate": 9.653469038475742e-06,
      "loss": 0.5372,
      "step": 2330
    },
    {
      "epoch": 0.12578921806702284,
      "grad_norm": 0.9756130876637941,
      "learning_rate": 9.65317540508017e-06,
      "loss": 0.4718,
      "step": 2331
    },
    {
      "epoch": 0.1258431816955372,
      "grad_norm": 1.0531367259468962,
      "learning_rate": 9.652881652318142e-06,
      "loss": 0.5154,
      "step": 2332
    },
    {
      "epoch": 0.12589714532405158,
      "grad_norm": 0.9901447002548144,
      "learning_rate": 9.652587780198098e-06,
      "loss": 0.502,
      "step": 2333
    },
    {
      "epoch": 0.12595110895256598,
      "grad_norm": 1.1715081027693313,
      "learning_rate": 9.652293788728486e-06,
      "loss": 0.6447,
      "step": 2334
    },
    {
      "epoch": 0.12600507258108035,
      "grad_norm": 1.0931870998648412,
      "learning_rate": 9.651999677917753e-06,
      "loss": 0.6276,
      "step": 2335
    },
    {
      "epoch": 0.12605903620959474,
      "grad_norm": 1.1298716728195484,
      "learning_rate": 9.651705447774357e-06,
      "loss": 0.5572,
      "step": 2336
    },
    {
      "epoch": 0.1261129998381091,
      "grad_norm": 1.1339630199520756,
      "learning_rate": 9.65141109830675e-06,
      "loss": 0.6943,
      "step": 2337
    },
    {
      "epoch": 0.12616696346662348,
      "grad_norm": 1.045219148018875,
      "learning_rate": 9.651116629523393e-06,
      "loss": 0.5688,
      "step": 2338
    },
    {
      "epoch": 0.12622092709513788,
      "grad_norm": 1.123672007235184,
      "learning_rate": 9.65082204143275e-06,
      "loss": 0.5997,
      "step": 2339
    },
    {
      "epoch": 0.12627489072365225,
      "grad_norm": 0.9582358570183779,
      "learning_rate": 9.650527334043286e-06,
      "loss": 0.38,
      "step": 2340
    },
    {
      "epoch": 0.12632885435216665,
      "grad_norm": 0.9553825726036964,
      "learning_rate": 9.650232507363474e-06,
      "loss": 0.4944,
      "step": 2341
    },
    {
      "epoch": 0.12638281798068102,
      "grad_norm": 1.0644758202658042,
      "learning_rate": 9.649937561401786e-06,
      "loss": 0.5437,
      "step": 2342
    },
    {
      "epoch": 0.12643678160919541,
      "grad_norm": 0.8911858578923589,
      "learning_rate": 9.6496424961667e-06,
      "loss": 0.3247,
      "step": 2343
    },
    {
      "epoch": 0.12649074523770978,
      "grad_norm": 1.0732386497563131,
      "learning_rate": 9.649347311666694e-06,
      "loss": 0.5302,
      "step": 2344
    },
    {
      "epoch": 0.12654470886622415,
      "grad_norm": 1.2324268362974626,
      "learning_rate": 9.649052007910255e-06,
      "loss": 0.5527,
      "step": 2345
    },
    {
      "epoch": 0.12659867249473855,
      "grad_norm": 0.9658054703646195,
      "learning_rate": 9.64875658490587e-06,
      "loss": 0.4225,
      "step": 2346
    },
    {
      "epoch": 0.12665263612325292,
      "grad_norm": 0.9332139358562981,
      "learning_rate": 9.648461042662027e-06,
      "loss": 0.3635,
      "step": 2347
    },
    {
      "epoch": 0.12670659975176732,
      "grad_norm": 1.0599404283870741,
      "learning_rate": 9.648165381187223e-06,
      "loss": 0.5378,
      "step": 2348
    },
    {
      "epoch": 0.1267605633802817,
      "grad_norm": 1.2145549993630358,
      "learning_rate": 9.647869600489954e-06,
      "loss": 0.6625,
      "step": 2349
    },
    {
      "epoch": 0.12681452700879609,
      "grad_norm": 0.9751897129335274,
      "learning_rate": 9.647573700578721e-06,
      "loss": 0.3854,
      "step": 2350
    },
    {
      "epoch": 0.12686849063731045,
      "grad_norm": 0.9768421948855404,
      "learning_rate": 9.64727768146203e-06,
      "loss": 0.4006,
      "step": 2351
    },
    {
      "epoch": 0.12692245426582482,
      "grad_norm": 1.1018301961283798,
      "learning_rate": 9.646981543148387e-06,
      "loss": 0.421,
      "step": 2352
    },
    {
      "epoch": 0.12697641789433922,
      "grad_norm": 1.286981143639693,
      "learning_rate": 9.646685285646304e-06,
      "loss": 0.6644,
      "step": 2353
    },
    {
      "epoch": 0.1270303815228536,
      "grad_norm": 1.1121912385109616,
      "learning_rate": 9.646388908964297e-06,
      "loss": 0.5332,
      "step": 2354
    },
    {
      "epoch": 0.127084345151368,
      "grad_norm": 1.0553933835029896,
      "learning_rate": 9.646092413110883e-06,
      "loss": 0.5645,
      "step": 2355
    },
    {
      "epoch": 0.12713830877988236,
      "grad_norm": 0.7894738988967905,
      "learning_rate": 9.645795798094583e-06,
      "loss": 0.4628,
      "step": 2356
    },
    {
      "epoch": 0.12719227240839673,
      "grad_norm": 0.8986074065583881,
      "learning_rate": 9.645499063923925e-06,
      "loss": 0.4192,
      "step": 2357
    },
    {
      "epoch": 0.12724623603691113,
      "grad_norm": 0.9779527614674771,
      "learning_rate": 9.645202210607433e-06,
      "loss": 0.4687,
      "step": 2358
    },
    {
      "epoch": 0.1273001996654255,
      "grad_norm": 1.2292497155955948,
      "learning_rate": 9.644905238153642e-06,
      "loss": 0.4825,
      "step": 2359
    },
    {
      "epoch": 0.1273541632939399,
      "grad_norm": 0.9467758623028063,
      "learning_rate": 9.644608146571084e-06,
      "loss": 0.4671,
      "step": 2360
    },
    {
      "epoch": 0.12740812692245426,
      "grad_norm": 1.091059325211132,
      "learning_rate": 9.644310935868303e-06,
      "loss": 0.4729,
      "step": 2361
    },
    {
      "epoch": 0.12746209055096866,
      "grad_norm": 1.0600944026062862,
      "learning_rate": 9.644013606053836e-06,
      "loss": 0.4672,
      "step": 2362
    },
    {
      "epoch": 0.12751605417948303,
      "grad_norm": 0.7840662638831936,
      "learning_rate": 9.643716157136232e-06,
      "loss": 0.3388,
      "step": 2363
    },
    {
      "epoch": 0.1275700178079974,
      "grad_norm": 1.076131696848735,
      "learning_rate": 9.643418589124038e-06,
      "loss": 0.4295,
      "step": 2364
    },
    {
      "epoch": 0.1276239814365118,
      "grad_norm": 1.299056985877537,
      "learning_rate": 9.643120902025808e-06,
      "loss": 0.6662,
      "step": 2365
    },
    {
      "epoch": 0.12767794506502617,
      "grad_norm": 1.1212532029125037,
      "learning_rate": 9.642823095850097e-06,
      "loss": 0.5579,
      "step": 2366
    },
    {
      "epoch": 0.12773190869354056,
      "grad_norm": 1.090655243646464,
      "learning_rate": 9.642525170605464e-06,
      "loss": 0.5574,
      "step": 2367
    },
    {
      "epoch": 0.12778587232205493,
      "grad_norm": 0.9412617905299603,
      "learning_rate": 9.642227126300472e-06,
      "loss": 0.4045,
      "step": 2368
    },
    {
      "epoch": 0.1278398359505693,
      "grad_norm": 1.0765261510123068,
      "learning_rate": 9.641928962943687e-06,
      "loss": 0.3827,
      "step": 2369
    },
    {
      "epoch": 0.1278937995790837,
      "grad_norm": 1.130923581847816,
      "learning_rate": 9.64163068054368e-06,
      "loss": 0.5618,
      "step": 2370
    },
    {
      "epoch": 0.12794776320759807,
      "grad_norm": 0.9371509506883278,
      "learning_rate": 9.64133227910902e-06,
      "loss": 0.451,
      "step": 2371
    },
    {
      "epoch": 0.12800172683611247,
      "grad_norm": 1.0830884506795813,
      "learning_rate": 9.641033758648286e-06,
      "loss": 0.5312,
      "step": 2372
    },
    {
      "epoch": 0.12805569046462684,
      "grad_norm": 1.0213862469568282,
      "learning_rate": 9.640735119170058e-06,
      "loss": 0.4496,
      "step": 2373
    },
    {
      "epoch": 0.12810965409314123,
      "grad_norm": 1.0960032223707379,
      "learning_rate": 9.640436360682918e-06,
      "loss": 0.4521,
      "step": 2374
    },
    {
      "epoch": 0.1281636177216556,
      "grad_norm": 0.9606648476859436,
      "learning_rate": 9.640137483195454e-06,
      "loss": 0.4681,
      "step": 2375
    },
    {
      "epoch": 0.12821758135016997,
      "grad_norm": 1.1172880473254505,
      "learning_rate": 9.639838486716256e-06,
      "loss": 0.5794,
      "step": 2376
    },
    {
      "epoch": 0.12827154497868437,
      "grad_norm": 1.1058131600254175,
      "learning_rate": 9.639539371253917e-06,
      "loss": 0.5061,
      "step": 2377
    },
    {
      "epoch": 0.12832550860719874,
      "grad_norm": 1.0313060732410173,
      "learning_rate": 9.639240136817033e-06,
      "loss": 0.4649,
      "step": 2378
    },
    {
      "epoch": 0.12837947223571314,
      "grad_norm": 1.1953392544536947,
      "learning_rate": 9.638940783414205e-06,
      "loss": 0.6473,
      "step": 2379
    },
    {
      "epoch": 0.1284334358642275,
      "grad_norm": 0.7873834199318798,
      "learning_rate": 9.638641311054037e-06,
      "loss": 0.4563,
      "step": 2380
    },
    {
      "epoch": 0.1284873994927419,
      "grad_norm": 1.103509483071724,
      "learning_rate": 9.638341719745135e-06,
      "loss": 0.5284,
      "step": 2381
    },
    {
      "epoch": 0.12854136312125627,
      "grad_norm": 1.17896761435768,
      "learning_rate": 9.638042009496113e-06,
      "loss": 0.4917,
      "step": 2382
    },
    {
      "epoch": 0.12859532674977064,
      "grad_norm": 1.042597902144271,
      "learning_rate": 9.63774218031558e-06,
      "loss": 0.5803,
      "step": 2383
    },
    {
      "epoch": 0.12864929037828504,
      "grad_norm": 1.2876964421340653,
      "learning_rate": 9.637442232212156e-06,
      "loss": 0.742,
      "step": 2384
    },
    {
      "epoch": 0.1287032540067994,
      "grad_norm": 1.0124103943886325,
      "learning_rate": 9.637142165194463e-06,
      "loss": 0.4369,
      "step": 2385
    },
    {
      "epoch": 0.1287572176353138,
      "grad_norm": 1.0195095112655648,
      "learning_rate": 9.636841979271124e-06,
      "loss": 0.4574,
      "step": 2386
    },
    {
      "epoch": 0.12881118126382818,
      "grad_norm": 1.2009140099321645,
      "learning_rate": 9.636541674450766e-06,
      "loss": 0.493,
      "step": 2387
    },
    {
      "epoch": 0.12886514489234255,
      "grad_norm": 0.9056379203867875,
      "learning_rate": 9.63624125074202e-06,
      "loss": 0.4328,
      "step": 2388
    },
    {
      "epoch": 0.12891910852085695,
      "grad_norm": 1.1481425038859074,
      "learning_rate": 9.635940708153522e-06,
      "loss": 0.5056,
      "step": 2389
    },
    {
      "epoch": 0.12897307214937132,
      "grad_norm": 0.9157110951050635,
      "learning_rate": 9.635640046693908e-06,
      "loss": 0.5054,
      "step": 2390
    },
    {
      "epoch": 0.1290270357778857,
      "grad_norm": 1.07460102493212,
      "learning_rate": 9.635339266371822e-06,
      "loss": 0.5474,
      "step": 2391
    },
    {
      "epoch": 0.12908099940640008,
      "grad_norm": 1.0030390726620408,
      "learning_rate": 9.635038367195905e-06,
      "loss": 0.4663,
      "step": 2392
    },
    {
      "epoch": 0.12913496303491448,
      "grad_norm": 1.117900712629355,
      "learning_rate": 9.634737349174808e-06,
      "loss": 0.5975,
      "step": 2393
    },
    {
      "epoch": 0.12918892666342885,
      "grad_norm": 1.3555245158028257,
      "learning_rate": 9.634436212317183e-06,
      "loss": 0.666,
      "step": 2394
    },
    {
      "epoch": 0.12924289029194322,
      "grad_norm": 0.6534616139106463,
      "learning_rate": 9.634134956631682e-06,
      "loss": 0.2691,
      "step": 2395
    },
    {
      "epoch": 0.12929685392045762,
      "grad_norm": 0.9775394153258076,
      "learning_rate": 9.633833582126965e-06,
      "loss": 0.4342,
      "step": 2396
    },
    {
      "epoch": 0.12935081754897199,
      "grad_norm": 1.0146337512552532,
      "learning_rate": 9.633532088811695e-06,
      "loss": 0.4693,
      "step": 2397
    },
    {
      "epoch": 0.12940478117748638,
      "grad_norm": 0.967699387726096,
      "learning_rate": 9.633230476694537e-06,
      "loss": 0.4306,
      "step": 2398
    },
    {
      "epoch": 0.12945874480600075,
      "grad_norm": 0.857916673328048,
      "learning_rate": 9.632928745784155e-06,
      "loss": 0.3807,
      "step": 2399
    },
    {
      "epoch": 0.12951270843451515,
      "grad_norm": 1.0676377315763095,
      "learning_rate": 9.632626896089227e-06,
      "loss": 0.5414,
      "step": 2400
    },
    {
      "epoch": 0.12956667206302952,
      "grad_norm": 1.2251926894520226,
      "learning_rate": 9.632324927618427e-06,
      "loss": 0.5675,
      "step": 2401
    },
    {
      "epoch": 0.1296206356915439,
      "grad_norm": 0.8927103427909798,
      "learning_rate": 9.632022840380434e-06,
      "loss": 0.4772,
      "step": 2402
    },
    {
      "epoch": 0.1296745993200583,
      "grad_norm": 0.8765908472064589,
      "learning_rate": 9.631720634383928e-06,
      "loss": 0.4286,
      "step": 2403
    },
    {
      "epoch": 0.12972856294857266,
      "grad_norm": 1.1782956295449456,
      "learning_rate": 9.631418309637596e-06,
      "loss": 0.5105,
      "step": 2404
    },
    {
      "epoch": 0.12978252657708705,
      "grad_norm": 0.9034348128422381,
      "learning_rate": 9.631115866150128e-06,
      "loss": 0.4183,
      "step": 2405
    },
    {
      "epoch": 0.12983649020560142,
      "grad_norm": 1.0964704607944813,
      "learning_rate": 9.630813303930216e-06,
      "loss": 0.5026,
      "step": 2406
    },
    {
      "epoch": 0.1298904538341158,
      "grad_norm": 1.1136694357283998,
      "learning_rate": 9.630510622986554e-06,
      "loss": 0.6531,
      "step": 2407
    },
    {
      "epoch": 0.1299444174626302,
      "grad_norm": 1.4270243400488787,
      "learning_rate": 9.630207823327846e-06,
      "loss": 0.675,
      "step": 2408
    },
    {
      "epoch": 0.12999838109114456,
      "grad_norm": 1.2708352520950916,
      "learning_rate": 9.629904904962791e-06,
      "loss": 0.7248,
      "step": 2409
    },
    {
      "epoch": 0.13005234471965896,
      "grad_norm": 1.2031617042791496,
      "learning_rate": 9.629601867900093e-06,
      "loss": 0.5103,
      "step": 2410
    },
    {
      "epoch": 0.13010630834817333,
      "grad_norm": 1.0085558137015336,
      "learning_rate": 9.629298712148469e-06,
      "loss": 0.4638,
      "step": 2411
    },
    {
      "epoch": 0.13016027197668772,
      "grad_norm": 0.8869487714830846,
      "learning_rate": 9.628995437716625e-06,
      "loss": 0.2859,
      "step": 2412
    },
    {
      "epoch": 0.1302142356052021,
      "grad_norm": 0.9736740855142443,
      "learning_rate": 9.62869204461328e-06,
      "loss": 0.5188,
      "step": 2413
    },
    {
      "epoch": 0.13026819923371646,
      "grad_norm": 1.1797904952668532,
      "learning_rate": 9.628388532847154e-06,
      "loss": 0.5881,
      "step": 2414
    },
    {
      "epoch": 0.13032216286223086,
      "grad_norm": 1.166877476662343,
      "learning_rate": 9.628084902426972e-06,
      "loss": 0.5259,
      "step": 2415
    },
    {
      "epoch": 0.13037612649074523,
      "grad_norm": 1.3834053765911938,
      "learning_rate": 9.627781153361457e-06,
      "loss": 0.6617,
      "step": 2416
    },
    {
      "epoch": 0.13043009011925963,
      "grad_norm": 1.2555975142595714,
      "learning_rate": 9.627477285659342e-06,
      "loss": 0.4746,
      "step": 2417
    },
    {
      "epoch": 0.130484053747774,
      "grad_norm": 1.0448532464614437,
      "learning_rate": 9.627173299329357e-06,
      "loss": 0.5222,
      "step": 2418
    },
    {
      "epoch": 0.13053801737628837,
      "grad_norm": 0.9829225057644765,
      "learning_rate": 9.626869194380244e-06,
      "loss": 0.5243,
      "step": 2419
    },
    {
      "epoch": 0.13059198100480277,
      "grad_norm": 0.957158270450429,
      "learning_rate": 9.62656497082074e-06,
      "loss": 0.4233,
      "step": 2420
    },
    {
      "epoch": 0.13064594463331713,
      "grad_norm": 1.0449644453801614,
      "learning_rate": 9.626260628659588e-06,
      "loss": 0.4826,
      "step": 2421
    },
    {
      "epoch": 0.13069990826183153,
      "grad_norm": 1.0207092747101252,
      "learning_rate": 9.625956167905537e-06,
      "loss": 0.4251,
      "step": 2422
    },
    {
      "epoch": 0.1307538718903459,
      "grad_norm": 0.8998483236584914,
      "learning_rate": 9.625651588567336e-06,
      "loss": 0.4228,
      "step": 2423
    },
    {
      "epoch": 0.1308078355188603,
      "grad_norm": 1.2680256486807358,
      "learning_rate": 9.62534689065374e-06,
      "loss": 0.5302,
      "step": 2424
    },
    {
      "epoch": 0.13086179914737467,
      "grad_norm": 1.2111756956184165,
      "learning_rate": 9.625042074173508e-06,
      "loss": 0.5867,
      "step": 2425
    },
    {
      "epoch": 0.13091576277588904,
      "grad_norm": 1.069740453486259,
      "learning_rate": 9.624737139135394e-06,
      "loss": 0.4858,
      "step": 2426
    },
    {
      "epoch": 0.13096972640440344,
      "grad_norm": 1.3376218797658184,
      "learning_rate": 9.624432085548171e-06,
      "loss": 0.4978,
      "step": 2427
    },
    {
      "epoch": 0.1310236900329178,
      "grad_norm": 1.1430438302576718,
      "learning_rate": 9.624126913420602e-06,
      "loss": 0.6198,
      "step": 2428
    },
    {
      "epoch": 0.1310776536614322,
      "grad_norm": 1.0004736159347272,
      "learning_rate": 9.623821622761457e-06,
      "loss": 0.4199,
      "step": 2429
    },
    {
      "epoch": 0.13113161728994657,
      "grad_norm": 1.2692414954178206,
      "learning_rate": 9.623516213579513e-06,
      "loss": 0.4773,
      "step": 2430
    },
    {
      "epoch": 0.13118558091846097,
      "grad_norm": 0.9002773963320035,
      "learning_rate": 9.623210685883545e-06,
      "loss": 0.4066,
      "step": 2431
    },
    {
      "epoch": 0.13123954454697534,
      "grad_norm": 0.9869864656682837,
      "learning_rate": 9.622905039682336e-06,
      "loss": 0.393,
      "step": 2432
    },
    {
      "epoch": 0.1312935081754897,
      "grad_norm": 0.9590246192881277,
      "learning_rate": 9.622599274984671e-06,
      "loss": 0.4118,
      "step": 2433
    },
    {
      "epoch": 0.1313474718040041,
      "grad_norm": 1.1492526357596056,
      "learning_rate": 9.622293391799337e-06,
      "loss": 0.5548,
      "step": 2434
    },
    {
      "epoch": 0.13140143543251848,
      "grad_norm": 0.9370903017822045,
      "learning_rate": 9.621987390135125e-06,
      "loss": 0.4195,
      "step": 2435
    },
    {
      "epoch": 0.13145539906103287,
      "grad_norm": 1.0136113905246822,
      "learning_rate": 9.621681270000833e-06,
      "loss": 0.4666,
      "step": 2436
    },
    {
      "epoch": 0.13150936268954724,
      "grad_norm": 1.085792518089485,
      "learning_rate": 9.621375031405255e-06,
      "loss": 0.4646,
      "step": 2437
    },
    {
      "epoch": 0.1315633263180616,
      "grad_norm": 1.2352015013851452,
      "learning_rate": 9.621068674357193e-06,
      "loss": 0.5272,
      "step": 2438
    },
    {
      "epoch": 0.131617289946576,
      "grad_norm": 0.8028213251695808,
      "learning_rate": 9.620762198865454e-06,
      "loss": 0.3269,
      "step": 2439
    },
    {
      "epoch": 0.13167125357509038,
      "grad_norm": 0.9976372090632596,
      "learning_rate": 9.620455604938848e-06,
      "loss": 0.3987,
      "step": 2440
    },
    {
      "epoch": 0.13172521720360478,
      "grad_norm": 1.1127761942622467,
      "learning_rate": 9.620148892586182e-06,
      "loss": 0.5511,
      "step": 2441
    },
    {
      "epoch": 0.13177918083211915,
      "grad_norm": 0.8015034349488216,
      "learning_rate": 9.619842061816274e-06,
      "loss": 0.336,
      "step": 2442
    },
    {
      "epoch": 0.13183314446063354,
      "grad_norm": 1.1426894223164443,
      "learning_rate": 9.61953511263794e-06,
      "loss": 0.5223,
      "step": 2443
    },
    {
      "epoch": 0.13188710808914791,
      "grad_norm": 1.0900535494826618,
      "learning_rate": 9.619228045060009e-06,
      "loss": 0.481,
      "step": 2444
    },
    {
      "epoch": 0.13194107171766228,
      "grad_norm": 1.1322871389096443,
      "learning_rate": 9.618920859091302e-06,
      "loss": 0.5074,
      "step": 2445
    },
    {
      "epoch": 0.13199503534617668,
      "grad_norm": 1.043328564956112,
      "learning_rate": 9.618613554740643e-06,
      "loss": 0.4115,
      "step": 2446
    },
    {
      "epoch": 0.13204899897469105,
      "grad_norm": 1.1264655513708715,
      "learning_rate": 9.61830613201687e-06,
      "loss": 0.5749,
      "step": 2447
    },
    {
      "epoch": 0.13210296260320545,
      "grad_norm": 1.0118079909092637,
      "learning_rate": 9.61799859092882e-06,
      "loss": 0.5687,
      "step": 2448
    },
    {
      "epoch": 0.13215692623171982,
      "grad_norm": 1.020229301501136,
      "learning_rate": 9.61769093148533e-06,
      "loss": 0.5671,
      "step": 2449
    },
    {
      "epoch": 0.13221088986023422,
      "grad_norm": 1.0828895202750095,
      "learning_rate": 9.617383153695239e-06,
      "loss": 0.3592,
      "step": 2450
    },
    {
      "epoch": 0.13226485348874858,
      "grad_norm": 1.0596406078708909,
      "learning_rate": 9.617075257567396e-06,
      "loss": 0.6218,
      "step": 2451
    },
    {
      "epoch": 0.13231881711726295,
      "grad_norm": 0.9752104492561213,
      "learning_rate": 9.616767243110652e-06,
      "loss": 0.5265,
      "step": 2452
    },
    {
      "epoch": 0.13237278074577735,
      "grad_norm": 1.3085855838016145,
      "learning_rate": 9.616459110333857e-06,
      "loss": 0.6442,
      "step": 2453
    },
    {
      "epoch": 0.13242674437429172,
      "grad_norm": 1.121650074806892,
      "learning_rate": 9.61615085924587e-06,
      "loss": 0.5288,
      "step": 2454
    },
    {
      "epoch": 0.13248070800280612,
      "grad_norm": 1.1055303019320688,
      "learning_rate": 9.615842489855544e-06,
      "loss": 0.5232,
      "step": 2455
    },
    {
      "epoch": 0.1325346716313205,
      "grad_norm": 1.2129562679317283,
      "learning_rate": 9.615534002171748e-06,
      "loss": 0.6321,
      "step": 2456
    },
    {
      "epoch": 0.13258863525983486,
      "grad_norm": 1.1957399176974686,
      "learning_rate": 9.615225396203346e-06,
      "loss": 0.5964,
      "step": 2457
    },
    {
      "epoch": 0.13264259888834926,
      "grad_norm": 1.2377423166940396,
      "learning_rate": 9.61491667195921e-06,
      "loss": 0.5543,
      "step": 2458
    },
    {
      "epoch": 0.13269656251686363,
      "grad_norm": 1.346362147879752,
      "learning_rate": 9.614607829448208e-06,
      "loss": 0.7215,
      "step": 2459
    },
    {
      "epoch": 0.13275052614537802,
      "grad_norm": 0.9535335356999718,
      "learning_rate": 9.61429886867922e-06,
      "loss": 0.4411,
      "step": 2460
    },
    {
      "epoch": 0.1328044897738924,
      "grad_norm": 1.0247754056072995,
      "learning_rate": 9.613989789661129e-06,
      "loss": 0.5801,
      "step": 2461
    },
    {
      "epoch": 0.1328584534024068,
      "grad_norm": 0.977434445340742,
      "learning_rate": 9.613680592402811e-06,
      "loss": 0.4674,
      "step": 2462
    },
    {
      "epoch": 0.13291241703092116,
      "grad_norm": 1.04626538900482,
      "learning_rate": 9.613371276913157e-06,
      "loss": 0.6238,
      "step": 2463
    },
    {
      "epoch": 0.13296638065943553,
      "grad_norm": 0.9885610191677416,
      "learning_rate": 9.613061843201056e-06,
      "loss": 0.6203,
      "step": 2464
    },
    {
      "epoch": 0.13302034428794993,
      "grad_norm": 1.2455341685581247,
      "learning_rate": 9.612752291275402e-06,
      "loss": 0.4998,
      "step": 2465
    },
    {
      "epoch": 0.1330743079164643,
      "grad_norm": 1.169939118486903,
      "learning_rate": 9.61244262114509e-06,
      "loss": 0.5575,
      "step": 2466
    },
    {
      "epoch": 0.1331282715449787,
      "grad_norm": 1.0664988788314986,
      "learning_rate": 9.612132832819026e-06,
      "loss": 0.4612,
      "step": 2467
    },
    {
      "epoch": 0.13318223517349306,
      "grad_norm": 1.0206933584213325,
      "learning_rate": 9.611822926306105e-06,
      "loss": 0.4886,
      "step": 2468
    },
    {
      "epoch": 0.13323619880200746,
      "grad_norm": 0.9988512482641646,
      "learning_rate": 9.61151290161524e-06,
      "loss": 0.5729,
      "step": 2469
    },
    {
      "epoch": 0.13329016243052183,
      "grad_norm": 1.1047423093010345,
      "learning_rate": 9.61120275875534e-06,
      "loss": 0.7915,
      "step": 2470
    },
    {
      "epoch": 0.1333441260590362,
      "grad_norm": 0.988310547692114,
      "learning_rate": 9.610892497735318e-06,
      "loss": 0.4449,
      "step": 2471
    },
    {
      "epoch": 0.1333980896875506,
      "grad_norm": 1.0445904806882997,
      "learning_rate": 9.610582118564094e-06,
      "loss": 0.4446,
      "step": 2472
    },
    {
      "epoch": 0.13345205331606497,
      "grad_norm": 1.0821206944813198,
      "learning_rate": 9.610271621250583e-06,
      "loss": 0.5476,
      "step": 2473
    },
    {
      "epoch": 0.13350601694457936,
      "grad_norm": 1.2224905940279551,
      "learning_rate": 9.609961005803717e-06,
      "loss": 0.6291,
      "step": 2474
    },
    {
      "epoch": 0.13355998057309373,
      "grad_norm": 0.9004469412710865,
      "learning_rate": 9.609650272232414e-06,
      "loss": 0.4127,
      "step": 2475
    },
    {
      "epoch": 0.1336139442016081,
      "grad_norm": 1.1188922327563204,
      "learning_rate": 9.609339420545612e-06,
      "loss": 0.6003,
      "step": 2476
    },
    {
      "epoch": 0.1336679078301225,
      "grad_norm": 0.8726946473185105,
      "learning_rate": 9.609028450752244e-06,
      "loss": 0.3706,
      "step": 2477
    },
    {
      "epoch": 0.13372187145863687,
      "grad_norm": 0.8963312001084753,
      "learning_rate": 9.608717362861244e-06,
      "loss": 0.4538,
      "step": 2478
    },
    {
      "epoch": 0.13377583508715127,
      "grad_norm": 0.987510092117334,
      "learning_rate": 9.608406156881556e-06,
      "loss": 0.4215,
      "step": 2479
    },
    {
      "epoch": 0.13382979871566564,
      "grad_norm": 0.951825455480239,
      "learning_rate": 9.608094832822125e-06,
      "loss": 0.4191,
      "step": 2480
    },
    {
      "epoch": 0.13388376234418004,
      "grad_norm": 1.194530000304039,
      "learning_rate": 9.607783390691897e-06,
      "loss": 0.6759,
      "step": 2481
    },
    {
      "epoch": 0.1339377259726944,
      "grad_norm": 0.8948112319041206,
      "learning_rate": 9.607471830499824e-06,
      "loss": 0.414,
      "step": 2482
    },
    {
      "epoch": 0.13399168960120877,
      "grad_norm": 1.0094061253921351,
      "learning_rate": 9.607160152254859e-06,
      "loss": 0.4218,
      "step": 2483
    },
    {
      "epoch": 0.13404565322972317,
      "grad_norm": 0.7844190787087654,
      "learning_rate": 9.606848355965964e-06,
      "loss": 0.3715,
      "step": 2484
    },
    {
      "epoch": 0.13409961685823754,
      "grad_norm": 0.7606854328208149,
      "learning_rate": 9.606536441642095e-06,
      "loss": 0.3128,
      "step": 2485
    },
    {
      "epoch": 0.13415358048675194,
      "grad_norm": 1.071083198019202,
      "learning_rate": 9.60622440929222e-06,
      "loss": 0.6381,
      "step": 2486
    },
    {
      "epoch": 0.1342075441152663,
      "grad_norm": 1.040828527890293,
      "learning_rate": 9.605912258925307e-06,
      "loss": 0.5335,
      "step": 2487
    },
    {
      "epoch": 0.13426150774378068,
      "grad_norm": 1.174359419468511,
      "learning_rate": 9.605599990550328e-06,
      "loss": 0.5561,
      "step": 2488
    },
    {
      "epoch": 0.13431547137229508,
      "grad_norm": 0.9748030315859836,
      "learning_rate": 9.605287604176254e-06,
      "loss": 0.4449,
      "step": 2489
    },
    {
      "epoch": 0.13436943500080945,
      "grad_norm": 1.0200785089753444,
      "learning_rate": 9.604975099812067e-06,
      "loss": 0.5766,
      "step": 2490
    },
    {
      "epoch": 0.13442339862932384,
      "grad_norm": 1.1279461252828833,
      "learning_rate": 9.604662477466749e-06,
      "loss": 0.4621,
      "step": 2491
    },
    {
      "epoch": 0.1344773622578382,
      "grad_norm": 1.1348280737827194,
      "learning_rate": 9.604349737149284e-06,
      "loss": 0.5523,
      "step": 2492
    },
    {
      "epoch": 0.1345313258863526,
      "grad_norm": 1.2243609099980748,
      "learning_rate": 9.604036878868659e-06,
      "loss": 0.6548,
      "step": 2493
    },
    {
      "epoch": 0.13458528951486698,
      "grad_norm": 1.24533581395501,
      "learning_rate": 9.60372390263387e-06,
      "loss": 0.6079,
      "step": 2494
    },
    {
      "epoch": 0.13463925314338135,
      "grad_norm": 0.9909986314270336,
      "learning_rate": 9.603410808453908e-06,
      "loss": 0.367,
      "step": 2495
    },
    {
      "epoch": 0.13469321677189575,
      "grad_norm": 1.0652539883547631,
      "learning_rate": 9.603097596337771e-06,
      "loss": 0.4058,
      "step": 2496
    },
    {
      "epoch": 0.13474718040041012,
      "grad_norm": 1.1147590269355505,
      "learning_rate": 9.602784266294465e-06,
      "loss": 0.4668,
      "step": 2497
    },
    {
      "epoch": 0.1348011440289245,
      "grad_norm": 1.2660133946669825,
      "learning_rate": 9.602470818332994e-06,
      "loss": 0.6954,
      "step": 2498
    },
    {
      "epoch": 0.13485510765743888,
      "grad_norm": 1.015002525451692,
      "learning_rate": 9.602157252462366e-06,
      "loss": 0.4105,
      "step": 2499
    },
    {
      "epoch": 0.13490907128595328,
      "grad_norm": 0.9577233129333887,
      "learning_rate": 9.601843568691593e-06,
      "loss": 0.4719,
      "step": 2500
    },
    {
      "epoch": 0.13490907128595328,
      "eval_loss": 0.5979181528091431,
      "eval_runtime": 164.2476,
      "eval_samples_per_second": 20.938,
      "eval_steps_per_second": 0.877,
      "step": 2500
    },
    {
      "epoch": 0.13496303491446765,
      "grad_norm": 1.029910066759321,
      "learning_rate": 9.601529767029693e-06,
      "loss": 0.528,
      "step": 2501
    },
    {
      "epoch": 0.13501699854298202,
      "grad_norm": 1.134005129744382,
      "learning_rate": 9.601215847485683e-06,
      "loss": 0.4481,
      "step": 2502
    },
    {
      "epoch": 0.13507096217149642,
      "grad_norm": 0.9967761863757987,
      "learning_rate": 9.600901810068583e-06,
      "loss": 0.4424,
      "step": 2503
    },
    {
      "epoch": 0.1351249258000108,
      "grad_norm": 0.9506240119650282,
      "learning_rate": 9.600587654787424e-06,
      "loss": 0.4962,
      "step": 2504
    },
    {
      "epoch": 0.13517888942852518,
      "grad_norm": 1.0109301091251528,
      "learning_rate": 9.600273381651231e-06,
      "loss": 0.4526,
      "step": 2505
    },
    {
      "epoch": 0.13523285305703955,
      "grad_norm": 1.1042680615407552,
      "learning_rate": 9.599958990669038e-06,
      "loss": 0.5816,
      "step": 2506
    },
    {
      "epoch": 0.13528681668555392,
      "grad_norm": 1.0954231731115147,
      "learning_rate": 9.599644481849882e-06,
      "loss": 0.494,
      "step": 2507
    },
    {
      "epoch": 0.13534078031406832,
      "grad_norm": 1.0758162220892895,
      "learning_rate": 9.5993298552028e-06,
      "loss": 0.7047,
      "step": 2508
    },
    {
      "epoch": 0.1353947439425827,
      "grad_norm": 0.8056014486717537,
      "learning_rate": 9.599015110736837e-06,
      "loss": 0.3388,
      "step": 2509
    },
    {
      "epoch": 0.1354487075710971,
      "grad_norm": 1.0099862056561784,
      "learning_rate": 9.598700248461037e-06,
      "loss": 0.5323,
      "step": 2510
    },
    {
      "epoch": 0.13550267119961146,
      "grad_norm": 1.0855549940230655,
      "learning_rate": 9.598385268384451e-06,
      "loss": 0.4472,
      "step": 2511
    },
    {
      "epoch": 0.13555663482812585,
      "grad_norm": 1.0834810297140982,
      "learning_rate": 9.59807017051613e-06,
      "loss": 0.4654,
      "step": 2512
    },
    {
      "epoch": 0.13561059845664022,
      "grad_norm": 0.9839433055733857,
      "learning_rate": 9.597754954865135e-06,
      "loss": 0.4587,
      "step": 2513
    },
    {
      "epoch": 0.1356645620851546,
      "grad_norm": 1.0058060991557032,
      "learning_rate": 9.597439621440518e-06,
      "loss": 0.4803,
      "step": 2514
    },
    {
      "epoch": 0.135718525713669,
      "grad_norm": 0.9913408751687934,
      "learning_rate": 9.597124170251348e-06,
      "loss": 0.4609,
      "step": 2515
    },
    {
      "epoch": 0.13577248934218336,
      "grad_norm": 1.1227426972215355,
      "learning_rate": 9.59680860130669e-06,
      "loss": 0.567,
      "step": 2516
    },
    {
      "epoch": 0.13582645297069776,
      "grad_norm": 0.9683478453412216,
      "learning_rate": 9.596492914615612e-06,
      "loss": 0.4141,
      "step": 2517
    },
    {
      "epoch": 0.13588041659921213,
      "grad_norm": 1.0552723142140177,
      "learning_rate": 9.596177110187189e-06,
      "loss": 0.5502,
      "step": 2518
    },
    {
      "epoch": 0.13593438022772653,
      "grad_norm": 1.141275200682066,
      "learning_rate": 9.595861188030497e-06,
      "loss": 0.6188,
      "step": 2519
    },
    {
      "epoch": 0.1359883438562409,
      "grad_norm": 0.9973201886089194,
      "learning_rate": 9.595545148154615e-06,
      "loss": 0.5388,
      "step": 2520
    },
    {
      "epoch": 0.13604230748475526,
      "grad_norm": 1.340008910568637,
      "learning_rate": 9.595228990568629e-06,
      "loss": 0.6626,
      "step": 2521
    },
    {
      "epoch": 0.13609627111326966,
      "grad_norm": 1.2491490711296749,
      "learning_rate": 9.594912715281624e-06,
      "loss": 0.5527,
      "step": 2522
    },
    {
      "epoch": 0.13615023474178403,
      "grad_norm": 0.9329716158670116,
      "learning_rate": 9.594596322302688e-06,
      "loss": 0.4582,
      "step": 2523
    },
    {
      "epoch": 0.13620419837029843,
      "grad_norm": 0.8732251108470945,
      "learning_rate": 9.594279811640919e-06,
      "loss": 0.4118,
      "step": 2524
    },
    {
      "epoch": 0.1362581619988128,
      "grad_norm": 1.1890024672951138,
      "learning_rate": 9.593963183305408e-06,
      "loss": 0.4772,
      "step": 2525
    },
    {
      "epoch": 0.13631212562732717,
      "grad_norm": 1.3836628478930972,
      "learning_rate": 9.59364643730526e-06,
      "loss": 0.4739,
      "step": 2526
    },
    {
      "epoch": 0.13636608925584157,
      "grad_norm": 1.2902259612881681,
      "learning_rate": 9.593329573649578e-06,
      "loss": 0.6125,
      "step": 2527
    },
    {
      "epoch": 0.13642005288435594,
      "grad_norm": 0.8087477486409129,
      "learning_rate": 9.593012592347468e-06,
      "loss": 0.3189,
      "step": 2528
    },
    {
      "epoch": 0.13647401651287033,
      "grad_norm": 1.1602530983014696,
      "learning_rate": 9.592695493408039e-06,
      "loss": 0.4625,
      "step": 2529
    },
    {
      "epoch": 0.1365279801413847,
      "grad_norm": 0.9521072564287097,
      "learning_rate": 9.592378276840406e-06,
      "loss": 0.3513,
      "step": 2530
    },
    {
      "epoch": 0.1365819437698991,
      "grad_norm": 1.0132130018012742,
      "learning_rate": 9.592060942653687e-06,
      "loss": 0.5234,
      "step": 2531
    },
    {
      "epoch": 0.13663590739841347,
      "grad_norm": 1.165906448192873,
      "learning_rate": 9.591743490857002e-06,
      "loss": 0.5985,
      "step": 2532
    },
    {
      "epoch": 0.13668987102692784,
      "grad_norm": 0.7884327482710408,
      "learning_rate": 9.591425921459475e-06,
      "loss": 0.3133,
      "step": 2533
    },
    {
      "epoch": 0.13674383465544224,
      "grad_norm": 0.9240095470179405,
      "learning_rate": 9.591108234470232e-06,
      "loss": 0.4745,
      "step": 2534
    },
    {
      "epoch": 0.1367977982839566,
      "grad_norm": 1.065547895464235,
      "learning_rate": 9.590790429898405e-06,
      "loss": 0.5158,
      "step": 2535
    },
    {
      "epoch": 0.136851761912471,
      "grad_norm": 0.9777372636408309,
      "learning_rate": 9.590472507753127e-06,
      "loss": 0.4382,
      "step": 2536
    },
    {
      "epoch": 0.13690572554098537,
      "grad_norm": 1.1308187633334466,
      "learning_rate": 9.590154468043536e-06,
      "loss": 0.4877,
      "step": 2537
    },
    {
      "epoch": 0.13695968916949977,
      "grad_norm": 0.7940184553753229,
      "learning_rate": 9.589836310778775e-06,
      "loss": 0.3455,
      "step": 2538
    },
    {
      "epoch": 0.13701365279801414,
      "grad_norm": 1.102230633105096,
      "learning_rate": 9.589518035967985e-06,
      "loss": 0.7786,
      "step": 2539
    },
    {
      "epoch": 0.1370676164265285,
      "grad_norm": 1.304275919149733,
      "learning_rate": 9.589199643620313e-06,
      "loss": 0.6737,
      "step": 2540
    },
    {
      "epoch": 0.1371215800550429,
      "grad_norm": 1.0210457107753166,
      "learning_rate": 9.588881133744912e-06,
      "loss": 0.4063,
      "step": 2541
    },
    {
      "epoch": 0.13717554368355728,
      "grad_norm": 1.0386254495498621,
      "learning_rate": 9.588562506350935e-06,
      "loss": 0.6125,
      "step": 2542
    },
    {
      "epoch": 0.13722950731207167,
      "grad_norm": 0.8895917731810747,
      "learning_rate": 9.588243761447542e-06,
      "loss": 0.408,
      "step": 2543
    },
    {
      "epoch": 0.13728347094058604,
      "grad_norm": 1.067530374667688,
      "learning_rate": 9.587924899043891e-06,
      "loss": 0.4107,
      "step": 2544
    },
    {
      "epoch": 0.13733743456910041,
      "grad_norm": 1.0277261987723012,
      "learning_rate": 9.58760591914915e-06,
      "loss": 0.686,
      "step": 2545
    },
    {
      "epoch": 0.1373913981976148,
      "grad_norm": 1.0365401907144827,
      "learning_rate": 9.587286821772482e-06,
      "loss": 0.5248,
      "step": 2546
    },
    {
      "epoch": 0.13744536182612918,
      "grad_norm": 1.0368307252113131,
      "learning_rate": 9.586967606923064e-06,
      "loss": 0.393,
      "step": 2547
    },
    {
      "epoch": 0.13749932545464358,
      "grad_norm": 1.0273634170984998,
      "learning_rate": 9.586648274610064e-06,
      "loss": 0.3996,
      "step": 2548
    },
    {
      "epoch": 0.13755328908315795,
      "grad_norm": 1.2538092172866357,
      "learning_rate": 9.586328824842666e-06,
      "loss": 0.6681,
      "step": 2549
    },
    {
      "epoch": 0.13760725271167235,
      "grad_norm": 1.302772807289807,
      "learning_rate": 9.586009257630047e-06,
      "loss": 0.4649,
      "step": 2550
    },
    {
      "epoch": 0.13766121634018672,
      "grad_norm": 1.1219474209095346,
      "learning_rate": 9.585689572981394e-06,
      "loss": 0.4485,
      "step": 2551
    },
    {
      "epoch": 0.13771517996870108,
      "grad_norm": 1.0831766812714303,
      "learning_rate": 9.585369770905893e-06,
      "loss": 0.6332,
      "step": 2552
    },
    {
      "epoch": 0.13776914359721548,
      "grad_norm": 1.0076020834204702,
      "learning_rate": 9.585049851412739e-06,
      "loss": 0.5317,
      "step": 2553
    },
    {
      "epoch": 0.13782310722572985,
      "grad_norm": 1.0280097172550269,
      "learning_rate": 9.584729814511123e-06,
      "loss": 0.4816,
      "step": 2554
    },
    {
      "epoch": 0.13787707085424425,
      "grad_norm": 1.1147073858086427,
      "learning_rate": 9.584409660210246e-06,
      "loss": 0.4072,
      "step": 2555
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 1.1536095536285746,
      "learning_rate": 9.584089388519307e-06,
      "loss": 0.5672,
      "step": 2556
    },
    {
      "epoch": 0.137984998111273,
      "grad_norm": 1.3147134998710537,
      "learning_rate": 9.583768999447514e-06,
      "loss": 0.6541,
      "step": 2557
    },
    {
      "epoch": 0.13803896173978739,
      "grad_norm": 0.7982761535559738,
      "learning_rate": 9.583448493004072e-06,
      "loss": 0.3368,
      "step": 2558
    },
    {
      "epoch": 0.13809292536830176,
      "grad_norm": 1.0077510203567732,
      "learning_rate": 9.583127869198196e-06,
      "loss": 0.4421,
      "step": 2559
    },
    {
      "epoch": 0.13814688899681615,
      "grad_norm": 1.3678041966263175,
      "learning_rate": 9.5828071280391e-06,
      "loss": 0.5334,
      "step": 2560
    },
    {
      "epoch": 0.13820085262533052,
      "grad_norm": 1.1255253080048273,
      "learning_rate": 9.582486269535998e-06,
      "loss": 0.5211,
      "step": 2561
    },
    {
      "epoch": 0.13825481625384492,
      "grad_norm": 1.0549558665271186,
      "learning_rate": 9.582165293698117e-06,
      "loss": 0.6559,
      "step": 2562
    },
    {
      "epoch": 0.1383087798823593,
      "grad_norm": 1.0888693478731601,
      "learning_rate": 9.581844200534682e-06,
      "loss": 0.5551,
      "step": 2563
    },
    {
      "epoch": 0.13836274351087366,
      "grad_norm": 1.016183020598562,
      "learning_rate": 9.58152299005492e-06,
      "loss": 0.4505,
      "step": 2564
    },
    {
      "epoch": 0.13841670713938806,
      "grad_norm": 1.110625805188764,
      "learning_rate": 9.581201662268063e-06,
      "loss": 0.4914,
      "step": 2565
    },
    {
      "epoch": 0.13847067076790243,
      "grad_norm": 0.9435012255574498,
      "learning_rate": 9.580880217183346e-06,
      "loss": 0.4038,
      "step": 2566
    },
    {
      "epoch": 0.13852463439641682,
      "grad_norm": 1.085540641750587,
      "learning_rate": 9.580558654810009e-06,
      "loss": 0.444,
      "step": 2567
    },
    {
      "epoch": 0.1385785980249312,
      "grad_norm": 0.8813751323089457,
      "learning_rate": 9.580236975157292e-06,
      "loss": 0.3689,
      "step": 2568
    },
    {
      "epoch": 0.1386325616534456,
      "grad_norm": 0.820317559867008,
      "learning_rate": 9.579915178234442e-06,
      "loss": 0.4265,
      "step": 2569
    },
    {
      "epoch": 0.13868652528195996,
      "grad_norm": 0.8693620741878239,
      "learning_rate": 9.579593264050708e-06,
      "loss": 0.3753,
      "step": 2570
    },
    {
      "epoch": 0.13874048891047433,
      "grad_norm": 1.2816812160027693,
      "learning_rate": 9.57927123261534e-06,
      "loss": 0.6314,
      "step": 2571
    },
    {
      "epoch": 0.13879445253898873,
      "grad_norm": 0.9904263251854922,
      "learning_rate": 9.578949083937597e-06,
      "loss": 0.4331,
      "step": 2572
    },
    {
      "epoch": 0.1388484161675031,
      "grad_norm": 0.9953179851933146,
      "learning_rate": 9.578626818026734e-06,
      "loss": 0.476,
      "step": 2573
    },
    {
      "epoch": 0.1389023797960175,
      "grad_norm": 1.1534774527531024,
      "learning_rate": 9.578304434892015e-06,
      "loss": 0.4492,
      "step": 2574
    },
    {
      "epoch": 0.13895634342453186,
      "grad_norm": 1.0394417402054497,
      "learning_rate": 9.577981934542708e-06,
      "loss": 0.5753,
      "step": 2575
    },
    {
      "epoch": 0.13901030705304623,
      "grad_norm": 1.1484559018496852,
      "learning_rate": 9.577659316988078e-06,
      "loss": 0.5677,
      "step": 2576
    },
    {
      "epoch": 0.13906427068156063,
      "grad_norm": 1.009247268310256,
      "learning_rate": 9.5773365822374e-06,
      "loss": 0.5041,
      "step": 2577
    },
    {
      "epoch": 0.139118234310075,
      "grad_norm": 1.072964181643425,
      "learning_rate": 9.57701373029995e-06,
      "loss": 0.4502,
      "step": 2578
    },
    {
      "epoch": 0.1391721979385894,
      "grad_norm": 0.9549320934435183,
      "learning_rate": 9.576690761185003e-06,
      "loss": 0.4482,
      "step": 2579
    },
    {
      "epoch": 0.13922616156710377,
      "grad_norm": 1.1089248740062063,
      "learning_rate": 9.576367674901844e-06,
      "loss": 0.6669,
      "step": 2580
    },
    {
      "epoch": 0.13928012519561817,
      "grad_norm": 1.0193315557908955,
      "learning_rate": 9.57604447145976e-06,
      "loss": 0.5047,
      "step": 2581
    },
    {
      "epoch": 0.13933408882413253,
      "grad_norm": 0.9607659260039592,
      "learning_rate": 9.57572115086804e-06,
      "loss": 0.4033,
      "step": 2582
    },
    {
      "epoch": 0.1393880524526469,
      "grad_norm": 1.14485580605853,
      "learning_rate": 9.575397713135976e-06,
      "loss": 0.4549,
      "step": 2583
    },
    {
      "epoch": 0.1394420160811613,
      "grad_norm": 1.0256920449242288,
      "learning_rate": 9.575074158272863e-06,
      "loss": 0.5008,
      "step": 2584
    },
    {
      "epoch": 0.13949597970967567,
      "grad_norm": 0.9360501234966311,
      "learning_rate": 9.574750486288002e-06,
      "loss": 0.4252,
      "step": 2585
    },
    {
      "epoch": 0.13954994333819007,
      "grad_norm": 1.2979261413733383,
      "learning_rate": 9.574426697190693e-06,
      "loss": 0.5747,
      "step": 2586
    },
    {
      "epoch": 0.13960390696670444,
      "grad_norm": 1.0418664474978445,
      "learning_rate": 9.574102790990248e-06,
      "loss": 0.4653,
      "step": 2587
    },
    {
      "epoch": 0.13965787059521884,
      "grad_norm": 1.045132092266482,
      "learning_rate": 9.57377876769597e-06,
      "loss": 0.4415,
      "step": 2588
    },
    {
      "epoch": 0.1397118342237332,
      "grad_norm": 0.9916717423858653,
      "learning_rate": 9.573454627317173e-06,
      "loss": 0.314,
      "step": 2589
    },
    {
      "epoch": 0.13976579785224758,
      "grad_norm": 0.9970491284596702,
      "learning_rate": 9.573130369863174e-06,
      "loss": 0.4225,
      "step": 2590
    },
    {
      "epoch": 0.13981976148076197,
      "grad_norm": 1.0390133110786588,
      "learning_rate": 9.572805995343295e-06,
      "loss": 0.4093,
      "step": 2591
    },
    {
      "epoch": 0.13987372510927634,
      "grad_norm": 0.9727906022994752,
      "learning_rate": 9.572481503766853e-06,
      "loss": 0.4695,
      "step": 2592
    },
    {
      "epoch": 0.13992768873779074,
      "grad_norm": 0.9022336383234881,
      "learning_rate": 9.57215689514318e-06,
      "loss": 0.4672,
      "step": 2593
    },
    {
      "epoch": 0.1399816523663051,
      "grad_norm": 1.0385106267425408,
      "learning_rate": 9.571832169481602e-06,
      "loss": 0.5557,
      "step": 2594
    },
    {
      "epoch": 0.14003561599481948,
      "grad_norm": 0.9654756810000235,
      "learning_rate": 9.571507326791453e-06,
      "loss": 0.4752,
      "step": 2595
    },
    {
      "epoch": 0.14008957962333388,
      "grad_norm": 1.1548396564453938,
      "learning_rate": 9.57118236708207e-06,
      "loss": 0.4955,
      "step": 2596
    },
    {
      "epoch": 0.14014354325184825,
      "grad_norm": 1.186801450845665,
      "learning_rate": 9.570857290362794e-06,
      "loss": 0.5206,
      "step": 2597
    },
    {
      "epoch": 0.14019750688036264,
      "grad_norm": 0.8664592693871139,
      "learning_rate": 9.570532096642963e-06,
      "loss": 0.3991,
      "step": 2598
    },
    {
      "epoch": 0.140251470508877,
      "grad_norm": 0.969692784429654,
      "learning_rate": 9.570206785931927e-06,
      "loss": 0.4325,
      "step": 2599
    },
    {
      "epoch": 0.1403054341373914,
      "grad_norm": 1.0492571386861889,
      "learning_rate": 9.569881358239037e-06,
      "loss": 0.5411,
      "step": 2600
    },
    {
      "epoch": 0.14035939776590578,
      "grad_norm": 1.0060693669054226,
      "learning_rate": 9.569555813573644e-06,
      "loss": 0.5672,
      "step": 2601
    },
    {
      "epoch": 0.14041336139442015,
      "grad_norm": 1.143113838901137,
      "learning_rate": 9.569230151945105e-06,
      "loss": 0.5534,
      "step": 2602
    },
    {
      "epoch": 0.14046732502293455,
      "grad_norm": 1.087133662747898,
      "learning_rate": 9.568904373362779e-06,
      "loss": 0.4858,
      "step": 2603
    },
    {
      "epoch": 0.14052128865144892,
      "grad_norm": 1.1522554951874138,
      "learning_rate": 9.56857847783603e-06,
      "loss": 0.5182,
      "step": 2604
    },
    {
      "epoch": 0.14057525227996331,
      "grad_norm": 1.0480853831318047,
      "learning_rate": 9.568252465374224e-06,
      "loss": 0.506,
      "step": 2605
    },
    {
      "epoch": 0.14062921590847768,
      "grad_norm": 0.9186438405095925,
      "learning_rate": 9.567926335986733e-06,
      "loss": 0.4158,
      "step": 2606
    },
    {
      "epoch": 0.14068317953699205,
      "grad_norm": 0.8678569034327193,
      "learning_rate": 9.567600089682928e-06,
      "loss": 0.3834,
      "step": 2607
    },
    {
      "epoch": 0.14073714316550645,
      "grad_norm": 1.091831426426634,
      "learning_rate": 9.567273726472187e-06,
      "loss": 0.5875,
      "step": 2608
    },
    {
      "epoch": 0.14079110679402082,
      "grad_norm": 0.8421450490830088,
      "learning_rate": 9.566947246363888e-06,
      "loss": 0.4066,
      "step": 2609
    },
    {
      "epoch": 0.14084507042253522,
      "grad_norm": 0.9666356438661695,
      "learning_rate": 9.566620649367418e-06,
      "loss": 0.4077,
      "step": 2610
    },
    {
      "epoch": 0.1408990340510496,
      "grad_norm": 1.2187214965749216,
      "learning_rate": 9.566293935492162e-06,
      "loss": 0.8097,
      "step": 2611
    },
    {
      "epoch": 0.14095299767956398,
      "grad_norm": 0.9571714197283464,
      "learning_rate": 9.565967104747507e-06,
      "loss": 0.3968,
      "step": 2612
    },
    {
      "epoch": 0.14100696130807835,
      "grad_norm": 1.028097354324081,
      "learning_rate": 9.56564015714285e-06,
      "loss": 0.5724,
      "step": 2613
    },
    {
      "epoch": 0.14106092493659272,
      "grad_norm": 1.1586571950425648,
      "learning_rate": 9.565313092687587e-06,
      "loss": 0.5813,
      "step": 2614
    },
    {
      "epoch": 0.14111488856510712,
      "grad_norm": 1.0418364353844511,
      "learning_rate": 9.564985911391117e-06,
      "loss": 0.4542,
      "step": 2615
    },
    {
      "epoch": 0.1411688521936215,
      "grad_norm": 1.0598796307897886,
      "learning_rate": 9.564658613262845e-06,
      "loss": 0.5188,
      "step": 2616
    },
    {
      "epoch": 0.1412228158221359,
      "grad_norm": 1.133300172474042,
      "learning_rate": 9.564331198312177e-06,
      "loss": 0.4732,
      "step": 2617
    },
    {
      "epoch": 0.14127677945065026,
      "grad_norm": 1.0259005484867296,
      "learning_rate": 9.564003666548526e-06,
      "loss": 0.5847,
      "step": 2618
    },
    {
      "epoch": 0.14133074307916466,
      "grad_norm": 0.7893556267682196,
      "learning_rate": 9.563676017981302e-06,
      "loss": 0.334,
      "step": 2619
    },
    {
      "epoch": 0.14138470670767903,
      "grad_norm": 1.070962047424158,
      "learning_rate": 9.563348252619923e-06,
      "loss": 0.5642,
      "step": 2620
    },
    {
      "epoch": 0.1414386703361934,
      "grad_norm": 1.0480827546547675,
      "learning_rate": 9.56302037047381e-06,
      "loss": 0.4962,
      "step": 2621
    },
    {
      "epoch": 0.1414926339647078,
      "grad_norm": 1.1513529929024682,
      "learning_rate": 9.562692371552385e-06,
      "loss": 0.5157,
      "step": 2622
    },
    {
      "epoch": 0.14154659759322216,
      "grad_norm": 1.0422364332615788,
      "learning_rate": 9.562364255865075e-06,
      "loss": 0.5357,
      "step": 2623
    },
    {
      "epoch": 0.14160056122173656,
      "grad_norm": 0.9789437352776942,
      "learning_rate": 9.562036023421314e-06,
      "loss": 0.454,
      "step": 2624
    },
    {
      "epoch": 0.14165452485025093,
      "grad_norm": 1.2250638756945975,
      "learning_rate": 9.561707674230534e-06,
      "loss": 0.5736,
      "step": 2625
    },
    {
      "epoch": 0.1417084884787653,
      "grad_norm": 1.0655496105126427,
      "learning_rate": 9.56137920830217e-06,
      "loss": 0.514,
      "step": 2626
    },
    {
      "epoch": 0.1417624521072797,
      "grad_norm": 0.9673476574120283,
      "learning_rate": 9.561050625645664e-06,
      "loss": 0.4301,
      "step": 2627
    },
    {
      "epoch": 0.14181641573579407,
      "grad_norm": 1.0647956149616595,
      "learning_rate": 9.56072192627046e-06,
      "loss": 0.475,
      "step": 2628
    },
    {
      "epoch": 0.14187037936430846,
      "grad_norm": 0.9021195591978254,
      "learning_rate": 9.560393110186003e-06,
      "loss": 0.4324,
      "step": 2629
    },
    {
      "epoch": 0.14192434299282283,
      "grad_norm": 1.0802933818558718,
      "learning_rate": 9.560064177401747e-06,
      "loss": 0.6331,
      "step": 2630
    },
    {
      "epoch": 0.14197830662133723,
      "grad_norm": 1.2837754742114944,
      "learning_rate": 9.559735127927144e-06,
      "loss": 0.6139,
      "step": 2631
    },
    {
      "epoch": 0.1420322702498516,
      "grad_norm": 0.9536782547287448,
      "learning_rate": 9.55940596177165e-06,
      "loss": 0.4511,
      "step": 2632
    },
    {
      "epoch": 0.14208623387836597,
      "grad_norm": 1.11044933185833,
      "learning_rate": 9.559076678944728e-06,
      "loss": 0.5647,
      "step": 2633
    },
    {
      "epoch": 0.14214019750688037,
      "grad_norm": 0.8873046372163033,
      "learning_rate": 9.558747279455842e-06,
      "loss": 0.4095,
      "step": 2634
    },
    {
      "epoch": 0.14219416113539474,
      "grad_norm": 1.199580699078788,
      "learning_rate": 9.558417763314457e-06,
      "loss": 0.6209,
      "step": 2635
    },
    {
      "epoch": 0.14224812476390913,
      "grad_norm": 0.8584095858866477,
      "learning_rate": 9.558088130530044e-06,
      "loss": 0.3892,
      "step": 2636
    },
    {
      "epoch": 0.1423020883924235,
      "grad_norm": 1.1019214451448518,
      "learning_rate": 9.557758381112078e-06,
      "loss": 0.4975,
      "step": 2637
    },
    {
      "epoch": 0.1423560520209379,
      "grad_norm": 1.2253871231919182,
      "learning_rate": 9.557428515070038e-06,
      "loss": 0.5796,
      "step": 2638
    },
    {
      "epoch": 0.14241001564945227,
      "grad_norm": 1.0227896510673953,
      "learning_rate": 9.557098532413401e-06,
      "loss": 0.4383,
      "step": 2639
    },
    {
      "epoch": 0.14246397927796664,
      "grad_norm": 1.0502765388406958,
      "learning_rate": 9.556768433151653e-06,
      "loss": 0.4901,
      "step": 2640
    },
    {
      "epoch": 0.14251794290648104,
      "grad_norm": 0.9921377963357235,
      "learning_rate": 9.556438217294281e-06,
      "loss": 0.4664,
      "step": 2641
    },
    {
      "epoch": 0.1425719065349954,
      "grad_norm": 1.0081820587697776,
      "learning_rate": 9.556107884850775e-06,
      "loss": 0.5231,
      "step": 2642
    },
    {
      "epoch": 0.1426258701635098,
      "grad_norm": 1.2791367590435578,
      "learning_rate": 9.555777435830631e-06,
      "loss": 0.4995,
      "step": 2643
    },
    {
      "epoch": 0.14267983379202417,
      "grad_norm": 0.9611821180144038,
      "learning_rate": 9.555446870243347e-06,
      "loss": 0.3968,
      "step": 2644
    },
    {
      "epoch": 0.14273379742053854,
      "grad_norm": 1.0454212039989443,
      "learning_rate": 9.55511618809842e-06,
      "loss": 0.5337,
      "step": 2645
    },
    {
      "epoch": 0.14278776104905294,
      "grad_norm": 0.8810747894593064,
      "learning_rate": 9.554785389405358e-06,
      "loss": 0.3838,
      "step": 2646
    },
    {
      "epoch": 0.1428417246775673,
      "grad_norm": 0.8947913990143506,
      "learning_rate": 9.554454474173666e-06,
      "loss": 0.3778,
      "step": 2647
    },
    {
      "epoch": 0.1428956883060817,
      "grad_norm": 1.0221665550766166,
      "learning_rate": 9.554123442412854e-06,
      "loss": 0.4618,
      "step": 2648
    },
    {
      "epoch": 0.14294965193459608,
      "grad_norm": 1.038492341392835,
      "learning_rate": 9.55379229413244e-06,
      "loss": 0.5345,
      "step": 2649
    },
    {
      "epoch": 0.14300361556311048,
      "grad_norm": 1.0760012700412676,
      "learning_rate": 9.553461029341939e-06,
      "loss": 0.5385,
      "step": 2650
    },
    {
      "epoch": 0.14305757919162485,
      "grad_norm": 1.16707773826301,
      "learning_rate": 9.553129648050874e-06,
      "loss": 0.6774,
      "step": 2651
    },
    {
      "epoch": 0.14311154282013921,
      "grad_norm": 1.0638068705902184,
      "learning_rate": 9.552798150268765e-06,
      "loss": 0.6202,
      "step": 2652
    },
    {
      "epoch": 0.1431655064486536,
      "grad_norm": 1.1093648067862005,
      "learning_rate": 9.552466536005142e-06,
      "loss": 0.6886,
      "step": 2653
    },
    {
      "epoch": 0.14321947007716798,
      "grad_norm": 0.8265117177854048,
      "learning_rate": 9.552134805269536e-06,
      "loss": 0.3481,
      "step": 2654
    },
    {
      "epoch": 0.14327343370568238,
      "grad_norm": 1.0110327824272152,
      "learning_rate": 9.551802958071483e-06,
      "loss": 0.4439,
      "step": 2655
    },
    {
      "epoch": 0.14332739733419675,
      "grad_norm": 0.8972082467465926,
      "learning_rate": 9.551470994420517e-06,
      "loss": 0.4069,
      "step": 2656
    },
    {
      "epoch": 0.14338136096271115,
      "grad_norm": 0.927206713277317,
      "learning_rate": 9.551138914326184e-06,
      "loss": 0.426,
      "step": 2657
    },
    {
      "epoch": 0.14343532459122552,
      "grad_norm": 0.8414110153102673,
      "learning_rate": 9.550806717798022e-06,
      "loss": 0.3912,
      "step": 2658
    },
    {
      "epoch": 0.14348928821973989,
      "grad_norm": 1.1585750829838957,
      "learning_rate": 9.550474404845584e-06,
      "loss": 0.4648,
      "step": 2659
    },
    {
      "epoch": 0.14354325184825428,
      "grad_norm": 1.1449548531018088,
      "learning_rate": 9.550141975478418e-06,
      "loss": 0.5446,
      "step": 2660
    },
    {
      "epoch": 0.14359721547676865,
      "grad_norm": 0.8992677296285637,
      "learning_rate": 9.54980942970608e-06,
      "loss": 0.3789,
      "step": 2661
    },
    {
      "epoch": 0.14365117910528305,
      "grad_norm": 1.152628082428434,
      "learning_rate": 9.549476767538126e-06,
      "loss": 0.6134,
      "step": 2662
    },
    {
      "epoch": 0.14370514273379742,
      "grad_norm": 0.9587653189423988,
      "learning_rate": 9.549143988984118e-06,
      "loss": 0.404,
      "step": 2663
    },
    {
      "epoch": 0.1437591063623118,
      "grad_norm": 1.0115456045337243,
      "learning_rate": 9.548811094053621e-06,
      "loss": 0.6834,
      "step": 2664
    },
    {
      "epoch": 0.1438130699908262,
      "grad_norm": 1.1208233546789934,
      "learning_rate": 9.548478082756201e-06,
      "loss": 0.5224,
      "step": 2665
    },
    {
      "epoch": 0.14386703361934056,
      "grad_norm": 1.142640644520483,
      "learning_rate": 9.548144955101432e-06,
      "loss": 0.4811,
      "step": 2666
    },
    {
      "epoch": 0.14392099724785495,
      "grad_norm": 1.0974440880571412,
      "learning_rate": 9.547811711098884e-06,
      "loss": 0.5431,
      "step": 2667
    },
    {
      "epoch": 0.14397496087636932,
      "grad_norm": 1.09323246984495,
      "learning_rate": 9.547478350758139e-06,
      "loss": 0.5424,
      "step": 2668
    },
    {
      "epoch": 0.14402892450488372,
      "grad_norm": 0.9954539985184983,
      "learning_rate": 9.547144874088778e-06,
      "loss": 0.4651,
      "step": 2669
    },
    {
      "epoch": 0.1440828881333981,
      "grad_norm": 1.1383665457547127,
      "learning_rate": 9.546811281100382e-06,
      "loss": 0.5504,
      "step": 2670
    },
    {
      "epoch": 0.14413685176191246,
      "grad_norm": 1.3613893937646355,
      "learning_rate": 9.546477571802543e-06,
      "loss": 0.6844,
      "step": 2671
    },
    {
      "epoch": 0.14419081539042686,
      "grad_norm": 1.0474214759758598,
      "learning_rate": 9.546143746204846e-06,
      "loss": 0.3626,
      "step": 2672
    },
    {
      "epoch": 0.14424477901894123,
      "grad_norm": 1.0865055636534702,
      "learning_rate": 9.545809804316892e-06,
      "loss": 0.6026,
      "step": 2673
    },
    {
      "epoch": 0.14429874264745562,
      "grad_norm": 1.1863800958005426,
      "learning_rate": 9.545475746148276e-06,
      "loss": 0.5424,
      "step": 2674
    },
    {
      "epoch": 0.14435270627597,
      "grad_norm": 1.0058743943550288,
      "learning_rate": 9.5451415717086e-06,
      "loss": 0.6409,
      "step": 2675
    },
    {
      "epoch": 0.14440666990448436,
      "grad_norm": 1.0251103526293064,
      "learning_rate": 9.544807281007468e-06,
      "loss": 0.4847,
      "step": 2676
    },
    {
      "epoch": 0.14446063353299876,
      "grad_norm": 0.9777966961140083,
      "learning_rate": 9.544472874054486e-06,
      "loss": 0.3921,
      "step": 2677
    },
    {
      "epoch": 0.14451459716151313,
      "grad_norm": 1.099678643169673,
      "learning_rate": 9.544138350859268e-06,
      "loss": 0.6029,
      "step": 2678
    },
    {
      "epoch": 0.14456856079002753,
      "grad_norm": 1.0037004095857347,
      "learning_rate": 9.543803711431426e-06,
      "loss": 0.4527,
      "step": 2679
    },
    {
      "epoch": 0.1446225244185419,
      "grad_norm": 1.437348677265869,
      "learning_rate": 9.543468955780582e-06,
      "loss": 0.5544,
      "step": 2680
    },
    {
      "epoch": 0.1446764880470563,
      "grad_norm": 1.0707494506682007,
      "learning_rate": 9.543134083916352e-06,
      "loss": 0.5283,
      "step": 2681
    },
    {
      "epoch": 0.14473045167557066,
      "grad_norm": 0.9294263595550656,
      "learning_rate": 9.542799095848365e-06,
      "loss": 0.4903,
      "step": 2682
    },
    {
      "epoch": 0.14478441530408503,
      "grad_norm": 0.9520122378533792,
      "learning_rate": 9.542463991586248e-06,
      "loss": 0.4262,
      "step": 2683
    },
    {
      "epoch": 0.14483837893259943,
      "grad_norm": 1.122554724878602,
      "learning_rate": 9.54212877113963e-06,
      "loss": 0.4245,
      "step": 2684
    },
    {
      "epoch": 0.1448923425611138,
      "grad_norm": 1.133346608105541,
      "learning_rate": 9.541793434518146e-06,
      "loss": 0.4565,
      "step": 2685
    },
    {
      "epoch": 0.1449463061896282,
      "grad_norm": 1.0526754529488411,
      "learning_rate": 9.541457981731434e-06,
      "loss": 0.4691,
      "step": 2686
    },
    {
      "epoch": 0.14500026981814257,
      "grad_norm": 1.208048125468102,
      "learning_rate": 9.54112241278914e-06,
      "loss": 0.4247,
      "step": 2687
    },
    {
      "epoch": 0.14505423344665697,
      "grad_norm": 1.0733951365423606,
      "learning_rate": 9.540786727700901e-06,
      "loss": 0.6435,
      "step": 2688
    },
    {
      "epoch": 0.14510819707517134,
      "grad_norm": 1.1257436806259096,
      "learning_rate": 9.540450926476368e-06,
      "loss": 0.492,
      "step": 2689
    },
    {
      "epoch": 0.1451621607036857,
      "grad_norm": 1.006434784909184,
      "learning_rate": 9.540115009125196e-06,
      "loss": 0.4799,
      "step": 2690
    },
    {
      "epoch": 0.1452161243322001,
      "grad_norm": 0.9722214088182579,
      "learning_rate": 9.539778975657034e-06,
      "loss": 0.4375,
      "step": 2691
    },
    {
      "epoch": 0.14527008796071447,
      "grad_norm": 1.1070342600110088,
      "learning_rate": 9.539442826081544e-06,
      "loss": 0.6341,
      "step": 2692
    },
    {
      "epoch": 0.14532405158922887,
      "grad_norm": 1.132535510797178,
      "learning_rate": 9.539106560408383e-06,
      "loss": 0.5306,
      "step": 2693
    },
    {
      "epoch": 0.14537801521774324,
      "grad_norm": 1.195577648501966,
      "learning_rate": 9.538770178647221e-06,
      "loss": 0.5398,
      "step": 2694
    },
    {
      "epoch": 0.1454319788462576,
      "grad_norm": 0.9729369863913087,
      "learning_rate": 9.538433680807722e-06,
      "loss": 0.4369,
      "step": 2695
    },
    {
      "epoch": 0.145485942474772,
      "grad_norm": 1.039248128363371,
      "learning_rate": 9.53809706689956e-06,
      "loss": 0.5098,
      "step": 2696
    },
    {
      "epoch": 0.14553990610328638,
      "grad_norm": 1.0501415495755502,
      "learning_rate": 9.537760336932406e-06,
      "loss": 0.7323,
      "step": 2697
    },
    {
      "epoch": 0.14559386973180077,
      "grad_norm": 1.0263063462553053,
      "learning_rate": 9.53742349091594e-06,
      "loss": 0.5112,
      "step": 2698
    },
    {
      "epoch": 0.14564783336031514,
      "grad_norm": 1.2392996840627537,
      "learning_rate": 9.537086528859844e-06,
      "loss": 0.5922,
      "step": 2699
    },
    {
      "epoch": 0.14570179698882954,
      "grad_norm": 1.157408317157222,
      "learning_rate": 9.536749450773803e-06,
      "loss": 0.5631,
      "step": 2700
    },
    {
      "epoch": 0.1457557606173439,
      "grad_norm": 1.0618491489102444,
      "learning_rate": 9.536412256667502e-06,
      "loss": 0.5086,
      "step": 2701
    },
    {
      "epoch": 0.14580972424585828,
      "grad_norm": 1.1903306678334087,
      "learning_rate": 9.536074946550637e-06,
      "loss": 0.5743,
      "step": 2702
    },
    {
      "epoch": 0.14586368787437268,
      "grad_norm": 0.9564412282544611,
      "learning_rate": 9.535737520432898e-06,
      "loss": 0.4714,
      "step": 2703
    },
    {
      "epoch": 0.14591765150288705,
      "grad_norm": 1.0997757318815997,
      "learning_rate": 9.535399978323983e-06,
      "loss": 0.5966,
      "step": 2704
    },
    {
      "epoch": 0.14597161513140144,
      "grad_norm": 1.1125981078194969,
      "learning_rate": 9.535062320233598e-06,
      "loss": 0.5394,
      "step": 2705
    },
    {
      "epoch": 0.14602557875991581,
      "grad_norm": 1.026991788916191,
      "learning_rate": 9.534724546171443e-06,
      "loss": 0.4997,
      "step": 2706
    },
    {
      "epoch": 0.1460795423884302,
      "grad_norm": 1.4259261434919575,
      "learning_rate": 9.53438665614723e-06,
      "loss": 0.4821,
      "step": 2707
    },
    {
      "epoch": 0.14613350601694458,
      "grad_norm": 1.1792280050864627,
      "learning_rate": 9.534048650170666e-06,
      "loss": 0.5849,
      "step": 2708
    },
    {
      "epoch": 0.14618746964545895,
      "grad_norm": 1.1801925982652528,
      "learning_rate": 9.533710528251468e-06,
      "loss": 0.5223,
      "step": 2709
    },
    {
      "epoch": 0.14624143327397335,
      "grad_norm": 0.8929005822487396,
      "learning_rate": 9.533372290399355e-06,
      "loss": 0.4584,
      "step": 2710
    },
    {
      "epoch": 0.14629539690248772,
      "grad_norm": 0.9546866661356682,
      "learning_rate": 9.533033936624046e-06,
      "loss": 0.5108,
      "step": 2711
    },
    {
      "epoch": 0.14634936053100212,
      "grad_norm": 0.8599338529564818,
      "learning_rate": 9.532695466935268e-06,
      "loss": 0.3805,
      "step": 2712
    },
    {
      "epoch": 0.14640332415951648,
      "grad_norm": 1.2149036956078132,
      "learning_rate": 9.532356881342745e-06,
      "loss": 0.5312,
      "step": 2713
    },
    {
      "epoch": 0.14645728778803085,
      "grad_norm": 1.1915843959350059,
      "learning_rate": 9.532018179856212e-06,
      "loss": 0.5733,
      "step": 2714
    },
    {
      "epoch": 0.14651125141654525,
      "grad_norm": 1.024591019475785,
      "learning_rate": 9.531679362485403e-06,
      "loss": 0.4889,
      "step": 2715
    },
    {
      "epoch": 0.14656521504505962,
      "grad_norm": 1.0559808657112495,
      "learning_rate": 9.531340429240054e-06,
      "loss": 0.5174,
      "step": 2716
    },
    {
      "epoch": 0.14661917867357402,
      "grad_norm": 0.9567789831665745,
      "learning_rate": 9.531001380129908e-06,
      "loss": 0.5371,
      "step": 2717
    },
    {
      "epoch": 0.1466731423020884,
      "grad_norm": 0.8765303698997371,
      "learning_rate": 9.530662215164711e-06,
      "loss": 0.4249,
      "step": 2718
    },
    {
      "epoch": 0.14672710593060279,
      "grad_norm": 0.8809714435190494,
      "learning_rate": 9.530322934354207e-06,
      "loss": 0.4663,
      "step": 2719
    },
    {
      "epoch": 0.14678106955911716,
      "grad_norm": 0.9975744852914546,
      "learning_rate": 9.529983537708152e-06,
      "loss": 0.5255,
      "step": 2720
    },
    {
      "epoch": 0.14683503318763153,
      "grad_norm": 1.168151205435896,
      "learning_rate": 9.529644025236296e-06,
      "loss": 0.4823,
      "step": 2721
    },
    {
      "epoch": 0.14688899681614592,
      "grad_norm": 0.910044789382668,
      "learning_rate": 9.5293043969484e-06,
      "loss": 0.4541,
      "step": 2722
    },
    {
      "epoch": 0.1469429604446603,
      "grad_norm": 1.1770842834344528,
      "learning_rate": 9.528964652854225e-06,
      "loss": 0.5935,
      "step": 2723
    },
    {
      "epoch": 0.1469969240731747,
      "grad_norm": 1.0066274310321626,
      "learning_rate": 9.528624792963532e-06,
      "loss": 0.4019,
      "step": 2724
    },
    {
      "epoch": 0.14705088770168906,
      "grad_norm": 1.0085593712731897,
      "learning_rate": 9.528284817286095e-06,
      "loss": 0.5454,
      "step": 2725
    },
    {
      "epoch": 0.14710485133020346,
      "grad_norm": 0.8915287296790081,
      "learning_rate": 9.52794472583168e-06,
      "loss": 0.4547,
      "step": 2726
    },
    {
      "epoch": 0.14715881495871783,
      "grad_norm": 1.0002889739111178,
      "learning_rate": 9.527604518610065e-06,
      "loss": 0.5321,
      "step": 2727
    },
    {
      "epoch": 0.1472127785872322,
      "grad_norm": 1.1038141629037825,
      "learning_rate": 9.527264195631027e-06,
      "loss": 0.5067,
      "step": 2728
    },
    {
      "epoch": 0.1472667422157466,
      "grad_norm": 1.196890746349286,
      "learning_rate": 9.526923756904345e-06,
      "loss": 0.5626,
      "step": 2729
    },
    {
      "epoch": 0.14732070584426096,
      "grad_norm": 1.0503680774394206,
      "learning_rate": 9.526583202439807e-06,
      "loss": 0.5281,
      "step": 2730
    },
    {
      "epoch": 0.14737466947277536,
      "grad_norm": 1.1897915063209588,
      "learning_rate": 9.526242532247198e-06,
      "loss": 0.5781,
      "step": 2731
    },
    {
      "epoch": 0.14742863310128973,
      "grad_norm": 0.8366426721531985,
      "learning_rate": 9.525901746336311e-06,
      "loss": 0.5152,
      "step": 2732
    },
    {
      "epoch": 0.1474825967298041,
      "grad_norm": 1.12820813958084,
      "learning_rate": 9.525560844716939e-06,
      "loss": 0.675,
      "step": 2733
    },
    {
      "epoch": 0.1475365603583185,
      "grad_norm": 1.310594039995791,
      "learning_rate": 9.525219827398883e-06,
      "loss": 0.6133,
      "step": 2734
    },
    {
      "epoch": 0.14759052398683287,
      "grad_norm": 1.0766343103037874,
      "learning_rate": 9.524878694391941e-06,
      "loss": 0.4979,
      "step": 2735
    },
    {
      "epoch": 0.14764448761534726,
      "grad_norm": 0.7882782885328782,
      "learning_rate": 9.524537445705916e-06,
      "loss": 0.3136,
      "step": 2736
    },
    {
      "epoch": 0.14769845124386163,
      "grad_norm": 1.3486679751539103,
      "learning_rate": 9.52419608135062e-06,
      "loss": 0.6084,
      "step": 2737
    },
    {
      "epoch": 0.14775241487237603,
      "grad_norm": 1.0556444452364526,
      "learning_rate": 9.523854601335863e-06,
      "loss": 0.6846,
      "step": 2738
    },
    {
      "epoch": 0.1478063785008904,
      "grad_norm": 1.3341109686927861,
      "learning_rate": 9.523513005671458e-06,
      "loss": 0.5441,
      "step": 2739
    },
    {
      "epoch": 0.14786034212940477,
      "grad_norm": 1.0711440183458931,
      "learning_rate": 9.523171294367227e-06,
      "loss": 0.5149,
      "step": 2740
    },
    {
      "epoch": 0.14791430575791917,
      "grad_norm": 1.0223612217807727,
      "learning_rate": 9.522829467432983e-06,
      "loss": 0.5387,
      "step": 2741
    },
    {
      "epoch": 0.14796826938643354,
      "grad_norm": 1.0822431884422958,
      "learning_rate": 9.522487524878559e-06,
      "loss": 0.5001,
      "step": 2742
    },
    {
      "epoch": 0.14802223301494793,
      "grad_norm": 1.2800366587387562,
      "learning_rate": 9.522145466713776e-06,
      "loss": 0.5449,
      "step": 2743
    },
    {
      "epoch": 0.1480761966434623,
      "grad_norm": 1.1838853836813952,
      "learning_rate": 9.521803292948472e-06,
      "loss": 0.6643,
      "step": 2744
    },
    {
      "epoch": 0.14813016027197667,
      "grad_norm": 0.9141166364874875,
      "learning_rate": 9.521461003592475e-06,
      "loss": 0.4307,
      "step": 2745
    },
    {
      "epoch": 0.14818412390049107,
      "grad_norm": 1.1705889006255623,
      "learning_rate": 9.521118598655625e-06,
      "loss": 0.597,
      "step": 2746
    },
    {
      "epoch": 0.14823808752900544,
      "grad_norm": 0.7804647602362148,
      "learning_rate": 9.520776078147764e-06,
      "loss": 0.3506,
      "step": 2747
    },
    {
      "epoch": 0.14829205115751984,
      "grad_norm": 0.8681293886515719,
      "learning_rate": 9.520433442078736e-06,
      "loss": 0.4139,
      "step": 2748
    },
    {
      "epoch": 0.1483460147860342,
      "grad_norm": 1.4558325911129746,
      "learning_rate": 9.52009069045839e-06,
      "loss": 0.7303,
      "step": 2749
    },
    {
      "epoch": 0.1483999784145486,
      "grad_norm": 1.3023893387614927,
      "learning_rate": 9.519747823296575e-06,
      "loss": 0.6067,
      "step": 2750
    },
    {
      "epoch": 0.14845394204306298,
      "grad_norm": 0.8390630174981729,
      "learning_rate": 9.519404840603144e-06,
      "loss": 0.4123,
      "step": 2751
    },
    {
      "epoch": 0.14850790567157734,
      "grad_norm": 1.0628666991988718,
      "learning_rate": 9.519061742387958e-06,
      "loss": 0.5666,
      "step": 2752
    },
    {
      "epoch": 0.14856186930009174,
      "grad_norm": 1.2775771207586568,
      "learning_rate": 9.518718528660876e-06,
      "loss": 0.5672,
      "step": 2753
    },
    {
      "epoch": 0.1486158329286061,
      "grad_norm": 1.2759656925981926,
      "learning_rate": 9.518375199431765e-06,
      "loss": 0.5179,
      "step": 2754
    },
    {
      "epoch": 0.1486697965571205,
      "grad_norm": 0.744565339672643,
      "learning_rate": 9.518031754710487e-06,
      "loss": 0.3222,
      "step": 2755
    },
    {
      "epoch": 0.14872376018563488,
      "grad_norm": 1.267383341314063,
      "learning_rate": 9.51768819450692e-06,
      "loss": 0.4921,
      "step": 2756
    },
    {
      "epoch": 0.14877772381414928,
      "grad_norm": 1.3700116211689632,
      "learning_rate": 9.517344518830933e-06,
      "loss": 0.6167,
      "step": 2757
    },
    {
      "epoch": 0.14883168744266365,
      "grad_norm": 1.0318331241769707,
      "learning_rate": 9.517000727692404e-06,
      "loss": 0.4929,
      "step": 2758
    },
    {
      "epoch": 0.14888565107117802,
      "grad_norm": 1.0691213584611883,
      "learning_rate": 9.516656821101216e-06,
      "loss": 0.4211,
      "step": 2759
    },
    {
      "epoch": 0.1489396146996924,
      "grad_norm": 0.872281370383322,
      "learning_rate": 9.516312799067252e-06,
      "loss": 0.3697,
      "step": 2760
    },
    {
      "epoch": 0.14899357832820678,
      "grad_norm": 1.0177898649819634,
      "learning_rate": 9.5159686616004e-06,
      "loss": 0.5554,
      "step": 2761
    },
    {
      "epoch": 0.14904754195672118,
      "grad_norm": 0.9670687956150409,
      "learning_rate": 9.515624408710553e-06,
      "loss": 0.4102,
      "step": 2762
    },
    {
      "epoch": 0.14910150558523555,
      "grad_norm": 1.3021185340492363,
      "learning_rate": 9.5152800404076e-06,
      "loss": 0.6789,
      "step": 2763
    },
    {
      "epoch": 0.14915546921374992,
      "grad_norm": 0.9597878396484694,
      "learning_rate": 9.514935556701444e-06,
      "loss": 0.4563,
      "step": 2764
    },
    {
      "epoch": 0.14920943284226432,
      "grad_norm": 1.0462466572931999,
      "learning_rate": 9.514590957601982e-06,
      "loss": 0.5444,
      "step": 2765
    },
    {
      "epoch": 0.1492633964707787,
      "grad_norm": 0.9962500727661605,
      "learning_rate": 9.51424624311912e-06,
      "loss": 0.4202,
      "step": 2766
    },
    {
      "epoch": 0.14931736009929308,
      "grad_norm": 1.1550706051308917,
      "learning_rate": 9.513901413262765e-06,
      "loss": 0.5264,
      "step": 2767
    },
    {
      "epoch": 0.14937132372780745,
      "grad_norm": 1.068799516294129,
      "learning_rate": 9.513556468042827e-06,
      "loss": 0.4578,
      "step": 2768
    },
    {
      "epoch": 0.14942528735632185,
      "grad_norm": 1.2645756935412595,
      "learning_rate": 9.51321140746922e-06,
      "loss": 0.6239,
      "step": 2769
    },
    {
      "epoch": 0.14947925098483622,
      "grad_norm": 1.244606082613979,
      "learning_rate": 9.512866231551864e-06,
      "loss": 0.5557,
      "step": 2770
    },
    {
      "epoch": 0.1495332146133506,
      "grad_norm": 1.1317309432732894,
      "learning_rate": 9.512520940300677e-06,
      "loss": 0.5896,
      "step": 2771
    },
    {
      "epoch": 0.149587178241865,
      "grad_norm": 1.1140591289980817,
      "learning_rate": 9.512175533725584e-06,
      "loss": 0.4529,
      "step": 2772
    },
    {
      "epoch": 0.14964114187037936,
      "grad_norm": 1.0242381038144122,
      "learning_rate": 9.511830011836511e-06,
      "loss": 0.5195,
      "step": 2773
    },
    {
      "epoch": 0.14969510549889375,
      "grad_norm": 1.0701669212234728,
      "learning_rate": 9.51148437464339e-06,
      "loss": 0.4555,
      "step": 2774
    },
    {
      "epoch": 0.14974906912740812,
      "grad_norm": 1.2178655764423278,
      "learning_rate": 9.511138622156156e-06,
      "loss": 0.6977,
      "step": 2775
    },
    {
      "epoch": 0.14980303275592252,
      "grad_norm": 0.9112326183885575,
      "learning_rate": 9.510792754384746e-06,
      "loss": 0.4007,
      "step": 2776
    },
    {
      "epoch": 0.1498569963844369,
      "grad_norm": 0.9579132603705647,
      "learning_rate": 9.510446771339097e-06,
      "loss": 0.4015,
      "step": 2777
    },
    {
      "epoch": 0.14991096001295126,
      "grad_norm": 1.0872432074395484,
      "learning_rate": 9.510100673029157e-06,
      "loss": 0.4934,
      "step": 2778
    },
    {
      "epoch": 0.14996492364146566,
      "grad_norm": 1.2052227289733362,
      "learning_rate": 9.509754459464871e-06,
      "loss": 0.6154,
      "step": 2779
    },
    {
      "epoch": 0.15001888726998003,
      "grad_norm": 1.002968709589042,
      "learning_rate": 9.50940813065619e-06,
      "loss": 0.4652,
      "step": 2780
    },
    {
      "epoch": 0.15007285089849443,
      "grad_norm": 0.8737905108910214,
      "learning_rate": 9.50906168661307e-06,
      "loss": 0.4181,
      "step": 2781
    },
    {
      "epoch": 0.1501268145270088,
      "grad_norm": 0.9193251672944583,
      "learning_rate": 9.508715127345463e-06,
      "loss": 0.3952,
      "step": 2782
    },
    {
      "epoch": 0.15018077815552316,
      "grad_norm": 1.0084630576533118,
      "learning_rate": 9.508368452863337e-06,
      "loss": 0.4859,
      "step": 2783
    },
    {
      "epoch": 0.15023474178403756,
      "grad_norm": 1.1291265525635383,
      "learning_rate": 9.508021663176648e-06,
      "loss": 0.4592,
      "step": 2784
    },
    {
      "epoch": 0.15028870541255193,
      "grad_norm": 1.4834595867164335,
      "learning_rate": 9.50767475829537e-06,
      "loss": 0.715,
      "step": 2785
    },
    {
      "epoch": 0.15034266904106633,
      "grad_norm": 1.4135220165690807,
      "learning_rate": 9.507327738229466e-06,
      "loss": 0.6709,
      "step": 2786
    },
    {
      "epoch": 0.1503966326695807,
      "grad_norm": 1.1418754550040022,
      "learning_rate": 9.506980602988916e-06,
      "loss": 0.4757,
      "step": 2787
    },
    {
      "epoch": 0.1504505962980951,
      "grad_norm": 1.0410461761137528,
      "learning_rate": 9.506633352583693e-06,
      "loss": 0.5803,
      "step": 2788
    },
    {
      "epoch": 0.15050455992660947,
      "grad_norm": 1.0915138686218968,
      "learning_rate": 9.506285987023781e-06,
      "loss": 0.5661,
      "step": 2789
    },
    {
      "epoch": 0.15055852355512384,
      "grad_norm": 0.997682727949255,
      "learning_rate": 9.50593850631916e-06,
      "loss": 0.4109,
      "step": 2790
    },
    {
      "epoch": 0.15061248718363823,
      "grad_norm": 0.9804741684481599,
      "learning_rate": 9.50559091047982e-06,
      "loss": 0.481,
      "step": 2791
    },
    {
      "epoch": 0.1506664508121526,
      "grad_norm": 1.0043907499911815,
      "learning_rate": 9.50524319951575e-06,
      "loss": 0.5298,
      "step": 2792
    },
    {
      "epoch": 0.150720414440667,
      "grad_norm": 1.1952438444683178,
      "learning_rate": 9.504895373436944e-06,
      "loss": 0.5577,
      "step": 2793
    },
    {
      "epoch": 0.15077437806918137,
      "grad_norm": 1.1238283573149006,
      "learning_rate": 9.504547432253397e-06,
      "loss": 0.472,
      "step": 2794
    },
    {
      "epoch": 0.15082834169769577,
      "grad_norm": 1.374206317281408,
      "learning_rate": 9.50419937597511e-06,
      "loss": 0.8451,
      "step": 2795
    },
    {
      "epoch": 0.15088230532621014,
      "grad_norm": 0.9650478445037453,
      "learning_rate": 9.503851204612087e-06,
      "loss": 0.4587,
      "step": 2796
    },
    {
      "epoch": 0.1509362689547245,
      "grad_norm": 0.9839652655826362,
      "learning_rate": 9.503502918174335e-06,
      "loss": 0.3269,
      "step": 2797
    },
    {
      "epoch": 0.1509902325832389,
      "grad_norm": 0.9964073919264328,
      "learning_rate": 9.503154516671865e-06,
      "loss": 0.4895,
      "step": 2798
    },
    {
      "epoch": 0.15104419621175327,
      "grad_norm": 1.0898911423807338,
      "learning_rate": 9.50280600011469e-06,
      "loss": 0.4066,
      "step": 2799
    },
    {
      "epoch": 0.15109815984026767,
      "grad_norm": 1.330056718842153,
      "learning_rate": 9.502457368512822e-06,
      "loss": 0.601,
      "step": 2800
    },
    {
      "epoch": 0.15115212346878204,
      "grad_norm": 1.0027926633446427,
      "learning_rate": 9.50210862187629e-06,
      "loss": 0.4215,
      "step": 2801
    },
    {
      "epoch": 0.1512060870972964,
      "grad_norm": 1.1353000132097928,
      "learning_rate": 9.50175976021511e-06,
      "loss": 0.5901,
      "step": 2802
    },
    {
      "epoch": 0.1512600507258108,
      "grad_norm": 0.9208197119539769,
      "learning_rate": 9.501410783539311e-06,
      "loss": 0.483,
      "step": 2803
    },
    {
      "epoch": 0.15131401435432518,
      "grad_norm": 1.1516905524126346,
      "learning_rate": 9.501061691858923e-06,
      "loss": 0.5956,
      "step": 2804
    },
    {
      "epoch": 0.15136797798283957,
      "grad_norm": 1.4497308701953389,
      "learning_rate": 9.500712485183979e-06,
      "loss": 0.6845,
      "step": 2805
    },
    {
      "epoch": 0.15142194161135394,
      "grad_norm": 1.0025938544902167,
      "learning_rate": 9.500363163524517e-06,
      "loss": 0.5178,
      "step": 2806
    },
    {
      "epoch": 0.15147590523986834,
      "grad_norm": 0.9873272024920466,
      "learning_rate": 9.500013726890573e-06,
      "loss": 0.4225,
      "step": 2807
    },
    {
      "epoch": 0.1515298688683827,
      "grad_norm": 0.9196890987899318,
      "learning_rate": 9.499664175292194e-06,
      "loss": 0.3996,
      "step": 2808
    },
    {
      "epoch": 0.15158383249689708,
      "grad_norm": 1.1451173550985723,
      "learning_rate": 9.499314508739427e-06,
      "loss": 0.4811,
      "step": 2809
    },
    {
      "epoch": 0.15163779612541148,
      "grad_norm": 1.0659462029621256,
      "learning_rate": 9.498964727242319e-06,
      "loss": 0.5329,
      "step": 2810
    },
    {
      "epoch": 0.15169175975392585,
      "grad_norm": 1.0703593982816626,
      "learning_rate": 9.498614830810922e-06,
      "loss": 0.4612,
      "step": 2811
    },
    {
      "epoch": 0.15174572338244025,
      "grad_norm": 0.9695327974457811,
      "learning_rate": 9.498264819455298e-06,
      "loss": 0.4831,
      "step": 2812
    },
    {
      "epoch": 0.15179968701095461,
      "grad_norm": 1.0082917882854028,
      "learning_rate": 9.497914693185502e-06,
      "loss": 0.5948,
      "step": 2813
    },
    {
      "epoch": 0.15185365063946898,
      "grad_norm": 1.0637152304507682,
      "learning_rate": 9.497564452011596e-06,
      "loss": 0.4635,
      "step": 2814
    },
    {
      "epoch": 0.15190761426798338,
      "grad_norm": 1.1357602928253683,
      "learning_rate": 9.49721409594365e-06,
      "loss": 0.5427,
      "step": 2815
    },
    {
      "epoch": 0.15196157789649775,
      "grad_norm": 1.1644111266702526,
      "learning_rate": 9.496863624991731e-06,
      "loss": 0.5459,
      "step": 2816
    },
    {
      "epoch": 0.15201554152501215,
      "grad_norm": 0.981129528489329,
      "learning_rate": 9.496513039165914e-06,
      "loss": 0.5778,
      "step": 2817
    },
    {
      "epoch": 0.15206950515352652,
      "grad_norm": 0.9531871700703887,
      "learning_rate": 9.496162338476273e-06,
      "loss": 0.412,
      "step": 2818
    },
    {
      "epoch": 0.15212346878204092,
      "grad_norm": 0.8599717573928912,
      "learning_rate": 9.495811522932891e-06,
      "loss": 0.3252,
      "step": 2819
    },
    {
      "epoch": 0.15217743241055529,
      "grad_norm": 1.2000203757639407,
      "learning_rate": 9.495460592545847e-06,
      "loss": 0.6232,
      "step": 2820
    },
    {
      "epoch": 0.15223139603906966,
      "grad_norm": 1.1921194369363222,
      "learning_rate": 9.495109547325226e-06,
      "loss": 0.6013,
      "step": 2821
    },
    {
      "epoch": 0.15228535966758405,
      "grad_norm": 1.2644511454902831,
      "learning_rate": 9.494758387281124e-06,
      "loss": 0.615,
      "step": 2822
    },
    {
      "epoch": 0.15233932329609842,
      "grad_norm": 0.9013605252302545,
      "learning_rate": 9.494407112423626e-06,
      "loss": 0.3422,
      "step": 2823
    },
    {
      "epoch": 0.15239328692461282,
      "grad_norm": 1.086334142432782,
      "learning_rate": 9.494055722762832e-06,
      "loss": 0.4352,
      "step": 2824
    },
    {
      "epoch": 0.1524472505531272,
      "grad_norm": 1.1042704671775259,
      "learning_rate": 9.493704218308844e-06,
      "loss": 0.4663,
      "step": 2825
    },
    {
      "epoch": 0.1525012141816416,
      "grad_norm": 0.9857749382345511,
      "learning_rate": 9.493352599071758e-06,
      "loss": 0.4665,
      "step": 2826
    },
    {
      "epoch": 0.15255517781015596,
      "grad_norm": 1.0179170188940228,
      "learning_rate": 9.493000865061685e-06,
      "loss": 0.4057,
      "step": 2827
    },
    {
      "epoch": 0.15260914143867033,
      "grad_norm": 1.024508144899749,
      "learning_rate": 9.492649016288734e-06,
      "loss": 0.487,
      "step": 2828
    },
    {
      "epoch": 0.15266310506718472,
      "grad_norm": 1.0224996333437304,
      "learning_rate": 9.492297052763013e-06,
      "loss": 0.4659,
      "step": 2829
    },
    {
      "epoch": 0.1527170686956991,
      "grad_norm": 1.1015211610625972,
      "learning_rate": 9.491944974494643e-06,
      "loss": 0.5576,
      "step": 2830
    },
    {
      "epoch": 0.1527710323242135,
      "grad_norm": 1.0702420403191497,
      "learning_rate": 9.49159278149374e-06,
      "loss": 0.5733,
      "step": 2831
    },
    {
      "epoch": 0.15282499595272786,
      "grad_norm": 0.9954699410982839,
      "learning_rate": 9.491240473770426e-06,
      "loss": 0.5624,
      "step": 2832
    },
    {
      "epoch": 0.15287895958124223,
      "grad_norm": 0.9127593904418583,
      "learning_rate": 9.49088805133483e-06,
      "loss": 0.4448,
      "step": 2833
    },
    {
      "epoch": 0.15293292320975663,
      "grad_norm": 0.9246399420247385,
      "learning_rate": 9.490535514197077e-06,
      "loss": 0.4123,
      "step": 2834
    },
    {
      "epoch": 0.152986886838271,
      "grad_norm": 1.0703769462733244,
      "learning_rate": 9.490182862367302e-06,
      "loss": 0.5527,
      "step": 2835
    },
    {
      "epoch": 0.1530408504667854,
      "grad_norm": 0.9900364888738452,
      "learning_rate": 9.48983009585564e-06,
      "loss": 0.5052,
      "step": 2836
    },
    {
      "epoch": 0.15309481409529976,
      "grad_norm": 1.4593715585423634,
      "learning_rate": 9.48947721467223e-06,
      "loss": 0.6111,
      "step": 2837
    },
    {
      "epoch": 0.15314877772381416,
      "grad_norm": 1.104102787868028,
      "learning_rate": 9.489124218827212e-06,
      "loss": 0.3644,
      "step": 2838
    },
    {
      "epoch": 0.15320274135232853,
      "grad_norm": 0.987638030896048,
      "learning_rate": 9.488771108330734e-06,
      "loss": 0.4787,
      "step": 2839
    },
    {
      "epoch": 0.1532567049808429,
      "grad_norm": 1.0613157413275411,
      "learning_rate": 9.488417883192945e-06,
      "loss": 0.6394,
      "step": 2840
    },
    {
      "epoch": 0.1533106686093573,
      "grad_norm": 0.9458738190414954,
      "learning_rate": 9.488064543423996e-06,
      "loss": 0.4025,
      "step": 2841
    },
    {
      "epoch": 0.15336463223787167,
      "grad_norm": 1.2169479780082078,
      "learning_rate": 9.487711089034041e-06,
      "loss": 0.7105,
      "step": 2842
    },
    {
      "epoch": 0.15341859586638606,
      "grad_norm": 0.9601533896935032,
      "learning_rate": 9.48735752003324e-06,
      "loss": 0.494,
      "step": 2843
    },
    {
      "epoch": 0.15347255949490043,
      "grad_norm": 1.1641318249446895,
      "learning_rate": 9.487003836431756e-06,
      "loss": 0.5663,
      "step": 2844
    },
    {
      "epoch": 0.15352652312341483,
      "grad_norm": 0.9563745737604212,
      "learning_rate": 9.486650038239753e-06,
      "loss": 0.3602,
      "step": 2845
    },
    {
      "epoch": 0.1535804867519292,
      "grad_norm": 0.9180793400155437,
      "learning_rate": 9.4862961254674e-06,
      "loss": 0.4173,
      "step": 2846
    },
    {
      "epoch": 0.15363445038044357,
      "grad_norm": 1.1258785878347464,
      "learning_rate": 9.485942098124869e-06,
      "loss": 0.5589,
      "step": 2847
    },
    {
      "epoch": 0.15368841400895797,
      "grad_norm": 0.883033392317904,
      "learning_rate": 9.485587956222334e-06,
      "loss": 0.4165,
      "step": 2848
    },
    {
      "epoch": 0.15374237763747234,
      "grad_norm": 0.8361668081627076,
      "learning_rate": 9.485233699769973e-06,
      "loss": 0.3216,
      "step": 2849
    },
    {
      "epoch": 0.15379634126598674,
      "grad_norm": 1.1900547827422676,
      "learning_rate": 9.48487932877797e-06,
      "loss": 0.5849,
      "step": 2850
    },
    {
      "epoch": 0.1538503048945011,
      "grad_norm": 1.3838262025952157,
      "learning_rate": 9.484524843256508e-06,
      "loss": 0.7638,
      "step": 2851
    },
    {
      "epoch": 0.15390426852301548,
      "grad_norm": 1.083019812992553,
      "learning_rate": 9.484170243215776e-06,
      "loss": 0.6214,
      "step": 2852
    },
    {
      "epoch": 0.15395823215152987,
      "grad_norm": 1.0219788986930345,
      "learning_rate": 9.483815528665965e-06,
      "loss": 0.4615,
      "step": 2853
    },
    {
      "epoch": 0.15401219578004424,
      "grad_norm": 1.6232248513350636,
      "learning_rate": 9.48346069961727e-06,
      "loss": 0.5991,
      "step": 2854
    },
    {
      "epoch": 0.15406615940855864,
      "grad_norm": 1.0951075975276294,
      "learning_rate": 9.48310575607989e-06,
      "loss": 0.3977,
      "step": 2855
    },
    {
      "epoch": 0.154120123037073,
      "grad_norm": 1.1375374670957732,
      "learning_rate": 9.482750698064028e-06,
      "loss": 0.5911,
      "step": 2856
    },
    {
      "epoch": 0.1541740866655874,
      "grad_norm": 1.0645224308179708,
      "learning_rate": 9.482395525579884e-06,
      "loss": 0.6209,
      "step": 2857
    },
    {
      "epoch": 0.15422805029410178,
      "grad_norm": 0.9923636920012383,
      "learning_rate": 9.48204023863767e-06,
      "loss": 0.5374,
      "step": 2858
    },
    {
      "epoch": 0.15428201392261615,
      "grad_norm": 0.9609218034965911,
      "learning_rate": 9.481684837247595e-06,
      "loss": 0.4595,
      "step": 2859
    },
    {
      "epoch": 0.15433597755113054,
      "grad_norm": 0.7647594926047586,
      "learning_rate": 9.481329321419877e-06,
      "loss": 0.301,
      "step": 2860
    },
    {
      "epoch": 0.1543899411796449,
      "grad_norm": 1.0551851841784197,
      "learning_rate": 9.480973691164728e-06,
      "loss": 0.4641,
      "step": 2861
    },
    {
      "epoch": 0.1544439048081593,
      "grad_norm": 0.8545558320980838,
      "learning_rate": 9.480617946492374e-06,
      "loss": 0.3856,
      "step": 2862
    },
    {
      "epoch": 0.15449786843667368,
      "grad_norm": 1.011789944460799,
      "learning_rate": 9.480262087413039e-06,
      "loss": 0.4297,
      "step": 2863
    },
    {
      "epoch": 0.15455183206518805,
      "grad_norm": 1.2599715393855264,
      "learning_rate": 9.47990611393695e-06,
      "loss": 0.4941,
      "step": 2864
    },
    {
      "epoch": 0.15460579569370245,
      "grad_norm": 0.9677200898396942,
      "learning_rate": 9.479550026074335e-06,
      "loss": 0.403,
      "step": 2865
    },
    {
      "epoch": 0.15465975932221682,
      "grad_norm": 1.0377728072893408,
      "learning_rate": 9.479193823835434e-06,
      "loss": 0.5074,
      "step": 2866
    },
    {
      "epoch": 0.15471372295073121,
      "grad_norm": 1.025915069495186,
      "learning_rate": 9.47883750723048e-06,
      "loss": 0.4828,
      "step": 2867
    },
    {
      "epoch": 0.15476768657924558,
      "grad_norm": 1.1834059300264819,
      "learning_rate": 9.478481076269719e-06,
      "loss": 0.5575,
      "step": 2868
    },
    {
      "epoch": 0.15482165020775998,
      "grad_norm": 0.9636221667891641,
      "learning_rate": 9.47812453096339e-06,
      "loss": 0.4809,
      "step": 2869
    },
    {
      "epoch": 0.15487561383627435,
      "grad_norm": 0.9704198580116432,
      "learning_rate": 9.477767871321741e-06,
      "loss": 0.5914,
      "step": 2870
    },
    {
      "epoch": 0.15492957746478872,
      "grad_norm": 0.9750063028919201,
      "learning_rate": 9.477411097355025e-06,
      "loss": 0.4519,
      "step": 2871
    },
    {
      "epoch": 0.15498354109330312,
      "grad_norm": 1.0979218847502932,
      "learning_rate": 9.477054209073497e-06,
      "loss": 0.4913,
      "step": 2872
    },
    {
      "epoch": 0.1550375047218175,
      "grad_norm": 1.0876237145767753,
      "learning_rate": 9.47669720648741e-06,
      "loss": 0.4799,
      "step": 2873
    },
    {
      "epoch": 0.15509146835033188,
      "grad_norm": 1.0078010239957849,
      "learning_rate": 9.476340089607028e-06,
      "loss": 0.5526,
      "step": 2874
    },
    {
      "epoch": 0.15514543197884625,
      "grad_norm": 1.1930517622787165,
      "learning_rate": 9.475982858442614e-06,
      "loss": 0.5519,
      "step": 2875
    },
    {
      "epoch": 0.15519939560736065,
      "grad_norm": 1.1500906430899376,
      "learning_rate": 9.475625513004437e-06,
      "loss": 0.4688,
      "step": 2876
    },
    {
      "epoch": 0.15525335923587502,
      "grad_norm": 1.0790049906878143,
      "learning_rate": 9.475268053302763e-06,
      "loss": 0.5366,
      "step": 2877
    },
    {
      "epoch": 0.1553073228643894,
      "grad_norm": 1.2458981329388064,
      "learning_rate": 9.47491047934787e-06,
      "loss": 0.5673,
      "step": 2878
    },
    {
      "epoch": 0.1553612864929038,
      "grad_norm": 1.064254343830126,
      "learning_rate": 9.474552791150033e-06,
      "loss": 0.5475,
      "step": 2879
    },
    {
      "epoch": 0.15541525012141816,
      "grad_norm": 0.9097563265657234,
      "learning_rate": 9.474194988719533e-06,
      "loss": 0.3253,
      "step": 2880
    },
    {
      "epoch": 0.15546921374993256,
      "grad_norm": 1.3085861653582225,
      "learning_rate": 9.473837072066651e-06,
      "loss": 0.485,
      "step": 2881
    },
    {
      "epoch": 0.15552317737844693,
      "grad_norm": 1.1867553902854862,
      "learning_rate": 9.473479041201679e-06,
      "loss": 0.398,
      "step": 2882
    },
    {
      "epoch": 0.1555771410069613,
      "grad_norm": 0.824277977218826,
      "learning_rate": 9.473120896134904e-06,
      "loss": 0.3572,
      "step": 2883
    },
    {
      "epoch": 0.1556311046354757,
      "grad_norm": 0.9281228279094402,
      "learning_rate": 9.47276263687662e-06,
      "loss": 0.4196,
      "step": 2884
    },
    {
      "epoch": 0.15568506826399006,
      "grad_norm": 1.1900765997057472,
      "learning_rate": 9.472404263437123e-06,
      "loss": 0.6778,
      "step": 2885
    },
    {
      "epoch": 0.15573903189250446,
      "grad_norm": 0.9890690930687212,
      "learning_rate": 9.472045775826714e-06,
      "loss": 0.5263,
      "step": 2886
    },
    {
      "epoch": 0.15579299552101883,
      "grad_norm": 1.1837927520805074,
      "learning_rate": 9.471687174055696e-06,
      "loss": 0.6636,
      "step": 2887
    },
    {
      "epoch": 0.15584695914953323,
      "grad_norm": 1.2344221976148766,
      "learning_rate": 9.471328458134375e-06,
      "loss": 0.5035,
      "step": 2888
    },
    {
      "epoch": 0.1559009227780476,
      "grad_norm": 1.0464913022132651,
      "learning_rate": 9.470969628073063e-06,
      "loss": 0.461,
      "step": 2889
    },
    {
      "epoch": 0.15595488640656197,
      "grad_norm": 1.1958548996745026,
      "learning_rate": 9.470610683882072e-06,
      "loss": 0.6025,
      "step": 2890
    },
    {
      "epoch": 0.15600885003507636,
      "grad_norm": 0.9076619095644475,
      "learning_rate": 9.470251625571715e-06,
      "loss": 0.4786,
      "step": 2891
    },
    {
      "epoch": 0.15606281366359073,
      "grad_norm": 1.0783035884236594,
      "learning_rate": 9.469892453152315e-06,
      "loss": 0.6129,
      "step": 2892
    },
    {
      "epoch": 0.15611677729210513,
      "grad_norm": 0.9887525022195978,
      "learning_rate": 9.469533166634196e-06,
      "loss": 0.532,
      "step": 2893
    },
    {
      "epoch": 0.1561707409206195,
      "grad_norm": 0.8576644728034248,
      "learning_rate": 9.469173766027682e-06,
      "loss": 0.3744,
      "step": 2894
    },
    {
      "epoch": 0.1562247045491339,
      "grad_norm": 0.9829558603837744,
      "learning_rate": 9.468814251343103e-06,
      "loss": 0.4242,
      "step": 2895
    },
    {
      "epoch": 0.15627866817764827,
      "grad_norm": 1.193911322653794,
      "learning_rate": 9.468454622590794e-06,
      "loss": 0.4999,
      "step": 2896
    },
    {
      "epoch": 0.15633263180616264,
      "grad_norm": 1.103571413898659,
      "learning_rate": 9.46809487978109e-06,
      "loss": 0.4886,
      "step": 2897
    },
    {
      "epoch": 0.15638659543467703,
      "grad_norm": 1.1004416436398254,
      "learning_rate": 9.467735022924327e-06,
      "loss": 0.626,
      "step": 2898
    },
    {
      "epoch": 0.1564405590631914,
      "grad_norm": 1.0489516331262132,
      "learning_rate": 9.467375052030852e-06,
      "loss": 0.419,
      "step": 2899
    },
    {
      "epoch": 0.1564945226917058,
      "grad_norm": 1.2073312628495325,
      "learning_rate": 9.467014967111009e-06,
      "loss": 0.5705,
      "step": 2900
    },
    {
      "epoch": 0.15654848632022017,
      "grad_norm": 1.0119178368994157,
      "learning_rate": 9.466654768175147e-06,
      "loss": 0.4512,
      "step": 2901
    },
    {
      "epoch": 0.15660244994873454,
      "grad_norm": 1.0875765578620804,
      "learning_rate": 9.466294455233619e-06,
      "loss": 0.5958,
      "step": 2902
    },
    {
      "epoch": 0.15665641357724894,
      "grad_norm": 1.0478843241285938,
      "learning_rate": 9.465934028296779e-06,
      "loss": 0.4958,
      "step": 2903
    },
    {
      "epoch": 0.1567103772057633,
      "grad_norm": 0.9884385864044685,
      "learning_rate": 9.46557348737499e-06,
      "loss": 0.3717,
      "step": 2904
    },
    {
      "epoch": 0.1567643408342777,
      "grad_norm": 1.0627591161496852,
      "learning_rate": 9.465212832478612e-06,
      "loss": 0.4862,
      "step": 2905
    },
    {
      "epoch": 0.15681830446279207,
      "grad_norm": 1.1562972027559206,
      "learning_rate": 9.46485206361801e-06,
      "loss": 0.5356,
      "step": 2906
    },
    {
      "epoch": 0.15687226809130647,
      "grad_norm": 0.9603524675841937,
      "learning_rate": 9.464491180803554e-06,
      "loss": 0.5413,
      "step": 2907
    },
    {
      "epoch": 0.15692623171982084,
      "grad_norm": 1.1836173381727348,
      "learning_rate": 9.464130184045616e-06,
      "loss": 0.571,
      "step": 2908
    },
    {
      "epoch": 0.1569801953483352,
      "grad_norm": 1.313629758337978,
      "learning_rate": 9.46376907335457e-06,
      "loss": 0.8012,
      "step": 2909
    },
    {
      "epoch": 0.1570341589768496,
      "grad_norm": 1.0209304454351344,
      "learning_rate": 9.463407848740795e-06,
      "loss": 0.4758,
      "step": 2910
    },
    {
      "epoch": 0.15708812260536398,
      "grad_norm": 0.8563920806117512,
      "learning_rate": 9.463046510214675e-06,
      "loss": 0.3656,
      "step": 2911
    },
    {
      "epoch": 0.15714208623387838,
      "grad_norm": 0.9523887574688876,
      "learning_rate": 9.462685057786596e-06,
      "loss": 0.4095,
      "step": 2912
    },
    {
      "epoch": 0.15719604986239274,
      "grad_norm": 1.025706722008397,
      "learning_rate": 9.462323491466941e-06,
      "loss": 0.5064,
      "step": 2913
    },
    {
      "epoch": 0.15725001349090714,
      "grad_norm": 1.0188317608498416,
      "learning_rate": 9.461961811266105e-06,
      "loss": 0.5028,
      "step": 2914
    },
    {
      "epoch": 0.1573039771194215,
      "grad_norm": 0.9845470517114103,
      "learning_rate": 9.461600017194487e-06,
      "loss": 0.3829,
      "step": 2915
    },
    {
      "epoch": 0.15735794074793588,
      "grad_norm": 0.8352989680949157,
      "learning_rate": 9.461238109262479e-06,
      "loss": 0.3657,
      "step": 2916
    },
    {
      "epoch": 0.15741190437645028,
      "grad_norm": 1.0666792908952116,
      "learning_rate": 9.460876087480486e-06,
      "loss": 0.5231,
      "step": 2917
    },
    {
      "epoch": 0.15746586800496465,
      "grad_norm": 1.2502735000746397,
      "learning_rate": 9.46051395185891e-06,
      "loss": 0.473,
      "step": 2918
    },
    {
      "epoch": 0.15751983163347905,
      "grad_norm": 1.0384826580623279,
      "learning_rate": 9.460151702408164e-06,
      "loss": 0.5164,
      "step": 2919
    },
    {
      "epoch": 0.15757379526199342,
      "grad_norm": 1.2383654112918259,
      "learning_rate": 9.459789339138655e-06,
      "loss": 0.526,
      "step": 2920
    },
    {
      "epoch": 0.15762775889050779,
      "grad_norm": 1.1991652787562606,
      "learning_rate": 9.4594268620608e-06,
      "loss": 0.6141,
      "step": 2921
    },
    {
      "epoch": 0.15768172251902218,
      "grad_norm": 1.1661750393832249,
      "learning_rate": 9.459064271185015e-06,
      "loss": 0.5834,
      "step": 2922
    },
    {
      "epoch": 0.15773568614753655,
      "grad_norm": 1.0844068000511828,
      "learning_rate": 9.458701566521725e-06,
      "loss": 0.6405,
      "step": 2923
    },
    {
      "epoch": 0.15778964977605095,
      "grad_norm": 1.02503241827222,
      "learning_rate": 9.45833874808135e-06,
      "loss": 0.5292,
      "step": 2924
    },
    {
      "epoch": 0.15784361340456532,
      "grad_norm": 0.9831051027170661,
      "learning_rate": 9.45797581587432e-06,
      "loss": 0.4583,
      "step": 2925
    },
    {
      "epoch": 0.15789757703307972,
      "grad_norm": 0.9056946053694395,
      "learning_rate": 9.457612769911067e-06,
      "loss": 0.3339,
      "step": 2926
    },
    {
      "epoch": 0.1579515406615941,
      "grad_norm": 0.8803807335797599,
      "learning_rate": 9.457249610202022e-06,
      "loss": 0.3508,
      "step": 2927
    },
    {
      "epoch": 0.15800550429010846,
      "grad_norm": 0.938112782940209,
      "learning_rate": 9.456886336757627e-06,
      "loss": 0.473,
      "step": 2928
    },
    {
      "epoch": 0.15805946791862285,
      "grad_norm": 0.8830492335405095,
      "learning_rate": 9.456522949588319e-06,
      "loss": 0.4094,
      "step": 2929
    },
    {
      "epoch": 0.15811343154713722,
      "grad_norm": 1.1349872542128703,
      "learning_rate": 9.456159448704544e-06,
      "loss": 0.5457,
      "step": 2930
    },
    {
      "epoch": 0.15816739517565162,
      "grad_norm": 1.0307333397601042,
      "learning_rate": 9.455795834116748e-06,
      "loss": 0.5877,
      "step": 2931
    },
    {
      "epoch": 0.158221358804166,
      "grad_norm": 0.7006549112109498,
      "learning_rate": 9.455432105835385e-06,
      "loss": 0.3124,
      "step": 2932
    },
    {
      "epoch": 0.15827532243268036,
      "grad_norm": 0.9668862621859664,
      "learning_rate": 9.455068263870905e-06,
      "loss": 0.4285,
      "step": 2933
    },
    {
      "epoch": 0.15832928606119476,
      "grad_norm": 0.9029724476936948,
      "learning_rate": 9.454704308233767e-06,
      "loss": 0.4115,
      "step": 2934
    },
    {
      "epoch": 0.15838324968970913,
      "grad_norm": 1.1087508953689431,
      "learning_rate": 9.454340238934431e-06,
      "loss": 0.5999,
      "step": 2935
    },
    {
      "epoch": 0.15843721331822352,
      "grad_norm": 1.075836463843238,
      "learning_rate": 9.453976055983363e-06,
      "loss": 0.5349,
      "step": 2936
    },
    {
      "epoch": 0.1584911769467379,
      "grad_norm": 1.20229994550125,
      "learning_rate": 9.453611759391026e-06,
      "loss": 0.5615,
      "step": 2937
    },
    {
      "epoch": 0.1585451405752523,
      "grad_norm": 1.226899527228933,
      "learning_rate": 9.453247349167892e-06,
      "loss": 0.5455,
      "step": 2938
    },
    {
      "epoch": 0.15859910420376666,
      "grad_norm": 1.0544148396871993,
      "learning_rate": 9.452882825324436e-06,
      "loss": 0.4789,
      "step": 2939
    },
    {
      "epoch": 0.15865306783228103,
      "grad_norm": 0.9006553287205337,
      "learning_rate": 9.452518187871132e-06,
      "loss": 0.4425,
      "step": 2940
    },
    {
      "epoch": 0.15870703146079543,
      "grad_norm": 1.0720855721461386,
      "learning_rate": 9.452153436818462e-06,
      "loss": 0.6386,
      "step": 2941
    },
    {
      "epoch": 0.1587609950893098,
      "grad_norm": 1.0565794148654033,
      "learning_rate": 9.45178857217691e-06,
      "loss": 0.5664,
      "step": 2942
    },
    {
      "epoch": 0.1588149587178242,
      "grad_norm": 1.007972009815946,
      "learning_rate": 9.45142359395696e-06,
      "loss": 0.4532,
      "step": 2943
    },
    {
      "epoch": 0.15886892234633856,
      "grad_norm": 1.0495827968349112,
      "learning_rate": 9.451058502169106e-06,
      "loss": 0.4482,
      "step": 2944
    },
    {
      "epoch": 0.15892288597485296,
      "grad_norm": 0.8590267904173257,
      "learning_rate": 9.450693296823835e-06,
      "loss": 0.4679,
      "step": 2945
    },
    {
      "epoch": 0.15897684960336733,
      "grad_norm": 1.1121114511714107,
      "learning_rate": 9.450327977931647e-06,
      "loss": 0.4488,
      "step": 2946
    },
    {
      "epoch": 0.1590308132318817,
      "grad_norm": 0.978526513636428,
      "learning_rate": 9.449962545503045e-06,
      "loss": 0.4382,
      "step": 2947
    },
    {
      "epoch": 0.1590847768603961,
      "grad_norm": 1.1629246545472256,
      "learning_rate": 9.449596999548525e-06,
      "loss": 0.6222,
      "step": 2948
    },
    {
      "epoch": 0.15913874048891047,
      "grad_norm": 1.071501081712184,
      "learning_rate": 9.449231340078598e-06,
      "loss": 0.545,
      "step": 2949
    },
    {
      "epoch": 0.15919270411742487,
      "grad_norm": 1.2254608374115032,
      "learning_rate": 9.44886556710377e-06,
      "loss": 0.7286,
      "step": 2950
    },
    {
      "epoch": 0.15924666774593924,
      "grad_norm": 1.001361155889579,
      "learning_rate": 9.448499680634556e-06,
      "loss": 0.5595,
      "step": 2951
    },
    {
      "epoch": 0.1593006313744536,
      "grad_norm": 1.3320681292760825,
      "learning_rate": 9.448133680681474e-06,
      "loss": 0.5593,
      "step": 2952
    },
    {
      "epoch": 0.159354595002968,
      "grad_norm": 1.0290498009540576,
      "learning_rate": 9.447767567255038e-06,
      "loss": 0.5075,
      "step": 2953
    },
    {
      "epoch": 0.15940855863148237,
      "grad_norm": 0.9937671870003342,
      "learning_rate": 9.447401340365773e-06,
      "loss": 0.4641,
      "step": 2954
    },
    {
      "epoch": 0.15946252225999677,
      "grad_norm": 1.1828711994574164,
      "learning_rate": 9.447035000024208e-06,
      "loss": 0.5113,
      "step": 2955
    },
    {
      "epoch": 0.15951648588851114,
      "grad_norm": 1.1963924723036305,
      "learning_rate": 9.446668546240866e-06,
      "loss": 0.4672,
      "step": 2956
    },
    {
      "epoch": 0.15957044951702554,
      "grad_norm": 1.0002823889427979,
      "learning_rate": 9.446301979026284e-06,
      "loss": 0.4434,
      "step": 2957
    },
    {
      "epoch": 0.1596244131455399,
      "grad_norm": 1.302124221685511,
      "learning_rate": 9.445935298390994e-06,
      "loss": 0.5243,
      "step": 2958
    },
    {
      "epoch": 0.15967837677405428,
      "grad_norm": 1.1380623406917816,
      "learning_rate": 9.445568504345537e-06,
      "loss": 0.627,
      "step": 2959
    },
    {
      "epoch": 0.15973234040256867,
      "grad_norm": 1.1150746968545358,
      "learning_rate": 9.445201596900455e-06,
      "loss": 0.5856,
      "step": 2960
    },
    {
      "epoch": 0.15978630403108304,
      "grad_norm": 0.9819504343339145,
      "learning_rate": 9.444834576066294e-06,
      "loss": 0.5137,
      "step": 2961
    },
    {
      "epoch": 0.15984026765959744,
      "grad_norm": 0.980496432471336,
      "learning_rate": 9.444467441853599e-06,
      "loss": 0.3801,
      "step": 2962
    },
    {
      "epoch": 0.1598942312881118,
      "grad_norm": 1.1309617074388112,
      "learning_rate": 9.444100194272925e-06,
      "loss": 0.4412,
      "step": 2963
    },
    {
      "epoch": 0.1599481949166262,
      "grad_norm": 0.9049356981254731,
      "learning_rate": 9.443732833334827e-06,
      "loss": 0.357,
      "step": 2964
    },
    {
      "epoch": 0.16000215854514058,
      "grad_norm": 1.1552843340026835,
      "learning_rate": 9.443365359049861e-06,
      "loss": 0.6825,
      "step": 2965
    },
    {
      "epoch": 0.16005612217365495,
      "grad_norm": 1.2585909640919055,
      "learning_rate": 9.442997771428592e-06,
      "loss": 0.6187,
      "step": 2966
    },
    {
      "epoch": 0.16011008580216934,
      "grad_norm": 0.9453560526947975,
      "learning_rate": 9.442630070481581e-06,
      "loss": 0.4266,
      "step": 2967
    },
    {
      "epoch": 0.1601640494306837,
      "grad_norm": 1.1399775819290876,
      "learning_rate": 9.442262256219398e-06,
      "loss": 0.4922,
      "step": 2968
    },
    {
      "epoch": 0.1602180130591981,
      "grad_norm": 1.1522101107397504,
      "learning_rate": 9.441894328652616e-06,
      "loss": 0.6737,
      "step": 2969
    },
    {
      "epoch": 0.16027197668771248,
      "grad_norm": 1.0093176766089809,
      "learning_rate": 9.441526287791808e-06,
      "loss": 0.5261,
      "step": 2970
    },
    {
      "epoch": 0.16032594031622685,
      "grad_norm": 1.1617219249125292,
      "learning_rate": 9.44115813364755e-06,
      "loss": 0.5208,
      "step": 2971
    },
    {
      "epoch": 0.16037990394474125,
      "grad_norm": 0.9555423481831041,
      "learning_rate": 9.440789866230425e-06,
      "loss": 0.5283,
      "step": 2972
    },
    {
      "epoch": 0.16043386757325562,
      "grad_norm": 1.38965754978105,
      "learning_rate": 9.44042148555102e-06,
      "loss": 0.673,
      "step": 2973
    },
    {
      "epoch": 0.16048783120177001,
      "grad_norm": 1.3841161533500563,
      "learning_rate": 9.440052991619916e-06,
      "loss": 0.5446,
      "step": 2974
    },
    {
      "epoch": 0.16054179483028438,
      "grad_norm": 0.9949161992789052,
      "learning_rate": 9.43968438444771e-06,
      "loss": 0.4264,
      "step": 2975
    },
    {
      "epoch": 0.16059575845879878,
      "grad_norm": 0.9895256678105773,
      "learning_rate": 9.439315664044995e-06,
      "loss": 0.3891,
      "step": 2976
    },
    {
      "epoch": 0.16064972208731315,
      "grad_norm": 1.1226751756829116,
      "learning_rate": 9.438946830422368e-06,
      "loss": 0.5229,
      "step": 2977
    },
    {
      "epoch": 0.16070368571582752,
      "grad_norm": 0.9613734894365971,
      "learning_rate": 9.438577883590429e-06,
      "loss": 0.4265,
      "step": 2978
    },
    {
      "epoch": 0.16075764934434192,
      "grad_norm": 0.8707514694948728,
      "learning_rate": 9.43820882355978e-06,
      "loss": 0.3897,
      "step": 2979
    },
    {
      "epoch": 0.1608116129728563,
      "grad_norm": 0.9865680178703112,
      "learning_rate": 9.437839650341029e-06,
      "loss": 0.5033,
      "step": 2980
    },
    {
      "epoch": 0.16086557660137069,
      "grad_norm": 0.8215037060126451,
      "learning_rate": 9.43747036394479e-06,
      "loss": 0.3681,
      "step": 2981
    },
    {
      "epoch": 0.16091954022988506,
      "grad_norm": 1.037724137563405,
      "learning_rate": 9.437100964381674e-06,
      "loss": 0.4976,
      "step": 2982
    },
    {
      "epoch": 0.16097350385839945,
      "grad_norm": 1.3051493987429608,
      "learning_rate": 9.436731451662298e-06,
      "loss": 0.6589,
      "step": 2983
    },
    {
      "epoch": 0.16102746748691382,
      "grad_norm": 1.1884775347865255,
      "learning_rate": 9.436361825797282e-06,
      "loss": 0.482,
      "step": 2984
    },
    {
      "epoch": 0.1610814311154282,
      "grad_norm": 1.0206626874955227,
      "learning_rate": 9.43599208679725e-06,
      "loss": 0.383,
      "step": 2985
    },
    {
      "epoch": 0.1611353947439426,
      "grad_norm": 1.1000631040821316,
      "learning_rate": 9.435622234672828e-06,
      "loss": 0.6137,
      "step": 2986
    },
    {
      "epoch": 0.16118935837245696,
      "grad_norm": 1.028320014277096,
      "learning_rate": 9.435252269434648e-06,
      "loss": 0.4819,
      "step": 2987
    },
    {
      "epoch": 0.16124332200097136,
      "grad_norm": 0.9466939472657221,
      "learning_rate": 9.43488219109334e-06,
      "loss": 0.5253,
      "step": 2988
    },
    {
      "epoch": 0.16129728562948573,
      "grad_norm": 1.1445087675353023,
      "learning_rate": 9.43451199965954e-06,
      "loss": 0.6027,
      "step": 2989
    },
    {
      "epoch": 0.1613512492580001,
      "grad_norm": 1.0650579875920114,
      "learning_rate": 9.434141695143892e-06,
      "loss": 0.4571,
      "step": 2990
    },
    {
      "epoch": 0.1614052128865145,
      "grad_norm": 0.979517265359635,
      "learning_rate": 9.433771277557037e-06,
      "loss": 0.3932,
      "step": 2991
    },
    {
      "epoch": 0.16145917651502886,
      "grad_norm": 1.0275475366138263,
      "learning_rate": 9.43340074690962e-06,
      "loss": 0.5097,
      "step": 2992
    },
    {
      "epoch": 0.16151314014354326,
      "grad_norm": 1.452466258311937,
      "learning_rate": 9.43303010321229e-06,
      "loss": 0.7083,
      "step": 2993
    },
    {
      "epoch": 0.16156710377205763,
      "grad_norm": 1.184735503574892,
      "learning_rate": 9.432659346475701e-06,
      "loss": 0.6073,
      "step": 2994
    },
    {
      "epoch": 0.16162106740057203,
      "grad_norm": 1.1704524150563005,
      "learning_rate": 9.43228847671051e-06,
      "loss": 0.5536,
      "step": 2995
    },
    {
      "epoch": 0.1616750310290864,
      "grad_norm": 1.177285702112136,
      "learning_rate": 9.431917493927375e-06,
      "loss": 0.5353,
      "step": 2996
    },
    {
      "epoch": 0.16172899465760077,
      "grad_norm": 1.0310662876272836,
      "learning_rate": 9.431546398136958e-06,
      "loss": 0.493,
      "step": 2997
    },
    {
      "epoch": 0.16178295828611516,
      "grad_norm": 0.9884204856095337,
      "learning_rate": 9.431175189349923e-06,
      "loss": 0.3979,
      "step": 2998
    },
    {
      "epoch": 0.16183692191462953,
      "grad_norm": 1.1622571114156945,
      "learning_rate": 9.430803867576943e-06,
      "loss": 0.6345,
      "step": 2999
    },
    {
      "epoch": 0.16189088554314393,
      "grad_norm": 0.817843255602589,
      "learning_rate": 9.430432432828689e-06,
      "loss": 0.437,
      "step": 3000
    },
    {
      "epoch": 0.16189088554314393,
      "eval_loss": 0.5924049019813538,
      "eval_runtime": 164.6583,
      "eval_samples_per_second": 20.886,
      "eval_steps_per_second": 0.875,
      "step": 3000
    },
    {
      "epoch": 0.1619448491716583,
      "grad_norm": 0.9556694643284128,
      "learning_rate": 9.430060885115833e-06,
      "loss": 0.3971,
      "step": 3001
    },
    {
      "epoch": 0.16199881280017267,
      "grad_norm": 1.0818664172169097,
      "learning_rate": 9.429689224449057e-06,
      "loss": 0.3972,
      "step": 3002
    },
    {
      "epoch": 0.16205277642868707,
      "grad_norm": 1.273605535210558,
      "learning_rate": 9.429317450839042e-06,
      "loss": 0.6418,
      "step": 3003
    },
    {
      "epoch": 0.16210674005720144,
      "grad_norm": 1.2352653813107772,
      "learning_rate": 9.428945564296475e-06,
      "loss": 0.4909,
      "step": 3004
    },
    {
      "epoch": 0.16216070368571583,
      "grad_norm": 1.1533149620424703,
      "learning_rate": 9.42857356483204e-06,
      "loss": 0.6047,
      "step": 3005
    },
    {
      "epoch": 0.1622146673142302,
      "grad_norm": 1.171398421840585,
      "learning_rate": 9.428201452456432e-06,
      "loss": 0.566,
      "step": 3006
    },
    {
      "epoch": 0.1622686309427446,
      "grad_norm": 0.9813872132490498,
      "learning_rate": 9.427829227180344e-06,
      "loss": 0.4373,
      "step": 3007
    },
    {
      "epoch": 0.16232259457125897,
      "grad_norm": 1.1626122944699004,
      "learning_rate": 9.427456889014477e-06,
      "loss": 0.5867,
      "step": 3008
    },
    {
      "epoch": 0.16237655819977334,
      "grad_norm": 0.9626729888386892,
      "learning_rate": 9.427084437969532e-06,
      "loss": 0.4243,
      "step": 3009
    },
    {
      "epoch": 0.16243052182828774,
      "grad_norm": 1.2500473323898016,
      "learning_rate": 9.42671187405621e-06,
      "loss": 0.5294,
      "step": 3010
    },
    {
      "epoch": 0.1624844854568021,
      "grad_norm": 0.9623809037424444,
      "learning_rate": 9.426339197285222e-06,
      "loss": 0.5611,
      "step": 3011
    },
    {
      "epoch": 0.1625384490853165,
      "grad_norm": 0.9289052195571385,
      "learning_rate": 9.425966407667277e-06,
      "loss": 0.3921,
      "step": 3012
    },
    {
      "epoch": 0.16259241271383088,
      "grad_norm": 1.0527435465638553,
      "learning_rate": 9.42559350521309e-06,
      "loss": 0.664,
      "step": 3013
    },
    {
      "epoch": 0.16264637634234527,
      "grad_norm": 0.9429453793611975,
      "learning_rate": 9.425220489933381e-06,
      "loss": 0.4107,
      "step": 3014
    },
    {
      "epoch": 0.16270033997085964,
      "grad_norm": 0.9151679554357353,
      "learning_rate": 9.424847361838869e-06,
      "loss": 0.4721,
      "step": 3015
    },
    {
      "epoch": 0.162754303599374,
      "grad_norm": 1.037592736754966,
      "learning_rate": 9.424474120940277e-06,
      "loss": 0.5925,
      "step": 3016
    },
    {
      "epoch": 0.1628082672278884,
      "grad_norm": 1.2248237437559515,
      "learning_rate": 9.424100767248334e-06,
      "loss": 0.6128,
      "step": 3017
    },
    {
      "epoch": 0.16286223085640278,
      "grad_norm": 1.1470604338214256,
      "learning_rate": 9.423727300773768e-06,
      "loss": 0.5027,
      "step": 3018
    },
    {
      "epoch": 0.16291619448491718,
      "grad_norm": 1.0431866760864006,
      "learning_rate": 9.423353721527318e-06,
      "loss": 0.6309,
      "step": 3019
    },
    {
      "epoch": 0.16297015811343155,
      "grad_norm": 0.6933722126647592,
      "learning_rate": 9.422980029519715e-06,
      "loss": 0.2494,
      "step": 3020
    },
    {
      "epoch": 0.16302412174194592,
      "grad_norm": 1.1536956453898288,
      "learning_rate": 9.422606224761704e-06,
      "loss": 0.6348,
      "step": 3021
    },
    {
      "epoch": 0.1630780853704603,
      "grad_norm": 1.13266758935705,
      "learning_rate": 9.422232307264026e-06,
      "loss": 0.4723,
      "step": 3022
    },
    {
      "epoch": 0.16313204899897468,
      "grad_norm": 1.2221156784475118,
      "learning_rate": 9.421858277037428e-06,
      "loss": 0.4799,
      "step": 3023
    },
    {
      "epoch": 0.16318601262748908,
      "grad_norm": 0.9748698185272858,
      "learning_rate": 9.42148413409266e-06,
      "loss": 0.5042,
      "step": 3024
    },
    {
      "epoch": 0.16323997625600345,
      "grad_norm": 1.0014204569606169,
      "learning_rate": 9.421109878440477e-06,
      "loss": 0.4705,
      "step": 3025
    },
    {
      "epoch": 0.16329393988451785,
      "grad_norm": 1.1399509315661114,
      "learning_rate": 9.420735510091632e-06,
      "loss": 0.5868,
      "step": 3026
    },
    {
      "epoch": 0.16334790351303222,
      "grad_norm": 0.9899289293266449,
      "learning_rate": 9.420361029056888e-06,
      "loss": 0.4035,
      "step": 3027
    },
    {
      "epoch": 0.1634018671415466,
      "grad_norm": 1.01127412264611,
      "learning_rate": 9.419986435347008e-06,
      "loss": 0.4397,
      "step": 3028
    },
    {
      "epoch": 0.16345583077006098,
      "grad_norm": 0.9588630463431623,
      "learning_rate": 9.419611728972755e-06,
      "loss": 0.4449,
      "step": 3029
    },
    {
      "epoch": 0.16350979439857535,
      "grad_norm": 1.2380666327172114,
      "learning_rate": 9.419236909944901e-06,
      "loss": 0.5602,
      "step": 3030
    },
    {
      "epoch": 0.16356375802708975,
      "grad_norm": 1.127776180023369,
      "learning_rate": 9.418861978274218e-06,
      "loss": 0.5521,
      "step": 3031
    },
    {
      "epoch": 0.16361772165560412,
      "grad_norm": 0.9306796501593076,
      "learning_rate": 9.418486933971484e-06,
      "loss": 0.3872,
      "step": 3032
    },
    {
      "epoch": 0.16367168528411852,
      "grad_norm": 1.0351015360996014,
      "learning_rate": 9.418111777047474e-06,
      "loss": 0.3687,
      "step": 3033
    },
    {
      "epoch": 0.1637256489126329,
      "grad_norm": 0.8924275562872201,
      "learning_rate": 9.417736507512973e-06,
      "loss": 0.3921,
      "step": 3034
    },
    {
      "epoch": 0.16377961254114726,
      "grad_norm": 1.0661561176176064,
      "learning_rate": 9.417361125378766e-06,
      "loss": 0.6482,
      "step": 3035
    },
    {
      "epoch": 0.16383357616966165,
      "grad_norm": 1.241335436368178,
      "learning_rate": 9.416985630655644e-06,
      "loss": 0.5584,
      "step": 3036
    },
    {
      "epoch": 0.16388753979817602,
      "grad_norm": 0.8888631290440462,
      "learning_rate": 9.416610023354394e-06,
      "loss": 0.4033,
      "step": 3037
    },
    {
      "epoch": 0.16394150342669042,
      "grad_norm": 0.8972331457001452,
      "learning_rate": 9.416234303485818e-06,
      "loss": 0.4195,
      "step": 3038
    },
    {
      "epoch": 0.1639954670552048,
      "grad_norm": 1.054331208898059,
      "learning_rate": 9.415858471060708e-06,
      "loss": 0.4508,
      "step": 3039
    },
    {
      "epoch": 0.16404943068371916,
      "grad_norm": 0.7954476759868314,
      "learning_rate": 9.41548252608987e-06,
      "loss": 0.4296,
      "step": 3040
    },
    {
      "epoch": 0.16410339431223356,
      "grad_norm": 1.030451652512724,
      "learning_rate": 9.415106468584107e-06,
      "loss": 0.5327,
      "step": 3041
    },
    {
      "epoch": 0.16415735794074793,
      "grad_norm": 1.015956215361609,
      "learning_rate": 9.414730298554228e-06,
      "loss": 0.4974,
      "step": 3042
    },
    {
      "epoch": 0.16421132156926233,
      "grad_norm": 0.9272709245301843,
      "learning_rate": 9.414354016011047e-06,
      "loss": 0.3483,
      "step": 3043
    },
    {
      "epoch": 0.1642652851977767,
      "grad_norm": 1.0150350771222634,
      "learning_rate": 9.413977620965375e-06,
      "loss": 0.5064,
      "step": 3044
    },
    {
      "epoch": 0.1643192488262911,
      "grad_norm": 1.19288810364433,
      "learning_rate": 9.413601113428032e-06,
      "loss": 0.6861,
      "step": 3045
    },
    {
      "epoch": 0.16437321245480546,
      "grad_norm": 0.9624948077538739,
      "learning_rate": 9.413224493409838e-06,
      "loss": 0.4252,
      "step": 3046
    },
    {
      "epoch": 0.16442717608331983,
      "grad_norm": 1.2253308895360053,
      "learning_rate": 9.412847760921619e-06,
      "loss": 0.6518,
      "step": 3047
    },
    {
      "epoch": 0.16448113971183423,
      "grad_norm": 1.1671457288957277,
      "learning_rate": 9.412470915974199e-06,
      "loss": 0.651,
      "step": 3048
    },
    {
      "epoch": 0.1645351033403486,
      "grad_norm": 0.8765505935801314,
      "learning_rate": 9.412093958578415e-06,
      "loss": 0.3962,
      "step": 3049
    },
    {
      "epoch": 0.164589066968863,
      "grad_norm": 1.1450536790754162,
      "learning_rate": 9.411716888745095e-06,
      "loss": 0.5284,
      "step": 3050
    },
    {
      "epoch": 0.16464303059737737,
      "grad_norm": 1.2483592298579325,
      "learning_rate": 9.411339706485081e-06,
      "loss": 0.6087,
      "step": 3051
    },
    {
      "epoch": 0.16469699422589174,
      "grad_norm": 1.0439490842009356,
      "learning_rate": 9.41096241180921e-06,
      "loss": 0.4902,
      "step": 3052
    },
    {
      "epoch": 0.16475095785440613,
      "grad_norm": 1.056495166148305,
      "learning_rate": 9.41058500472833e-06,
      "loss": 0.5415,
      "step": 3053
    },
    {
      "epoch": 0.1648049214829205,
      "grad_norm": 1.2145790304172641,
      "learning_rate": 9.410207485253282e-06,
      "loss": 0.5508,
      "step": 3054
    },
    {
      "epoch": 0.1648588851114349,
      "grad_norm": 0.8904792645399217,
      "learning_rate": 9.409829853394922e-06,
      "loss": 0.3888,
      "step": 3055
    },
    {
      "epoch": 0.16491284873994927,
      "grad_norm": 1.0552352552400153,
      "learning_rate": 9.409452109164103e-06,
      "loss": 0.4298,
      "step": 3056
    },
    {
      "epoch": 0.16496681236846367,
      "grad_norm": 1.106455965800329,
      "learning_rate": 9.409074252571679e-06,
      "loss": 0.5506,
      "step": 3057
    },
    {
      "epoch": 0.16502077599697804,
      "grad_norm": 1.0744145272574606,
      "learning_rate": 9.40869628362851e-06,
      "loss": 0.5517,
      "step": 3058
    },
    {
      "epoch": 0.1650747396254924,
      "grad_norm": 0.9307517810607085,
      "learning_rate": 9.408318202345461e-06,
      "loss": 0.4244,
      "step": 3059
    },
    {
      "epoch": 0.1651287032540068,
      "grad_norm": 1.2019725571847832,
      "learning_rate": 9.407940008733399e-06,
      "loss": 0.575,
      "step": 3060
    },
    {
      "epoch": 0.16518266688252117,
      "grad_norm": 1.1416811803351057,
      "learning_rate": 9.40756170280319e-06,
      "loss": 0.5289,
      "step": 3061
    },
    {
      "epoch": 0.16523663051103557,
      "grad_norm": 1.2665498059748768,
      "learning_rate": 9.407183284565712e-06,
      "loss": 0.5472,
      "step": 3062
    },
    {
      "epoch": 0.16529059413954994,
      "grad_norm": 0.9739350308640315,
      "learning_rate": 9.406804754031837e-06,
      "loss": 0.3614,
      "step": 3063
    },
    {
      "epoch": 0.16534455776806434,
      "grad_norm": 0.8335953100771376,
      "learning_rate": 9.406426111212446e-06,
      "loss": 0.3219,
      "step": 3064
    },
    {
      "epoch": 0.1653985213965787,
      "grad_norm": 0.9032177672871006,
      "learning_rate": 9.40604735611842e-06,
      "loss": 0.4387,
      "step": 3065
    },
    {
      "epoch": 0.16545248502509308,
      "grad_norm": 1.0115050926061735,
      "learning_rate": 9.40566848876065e-06,
      "loss": 0.5073,
      "step": 3066
    },
    {
      "epoch": 0.16550644865360747,
      "grad_norm": 1.086252264671102,
      "learning_rate": 9.405289509150017e-06,
      "loss": 0.5555,
      "step": 3067
    },
    {
      "epoch": 0.16556041228212184,
      "grad_norm": 0.9018130363882,
      "learning_rate": 9.40491041729742e-06,
      "loss": 0.4076,
      "step": 3068
    },
    {
      "epoch": 0.16561437591063624,
      "grad_norm": 1.439400994384993,
      "learning_rate": 9.404531213213752e-06,
      "loss": 0.5329,
      "step": 3069
    },
    {
      "epoch": 0.1656683395391506,
      "grad_norm": 0.7145592529461156,
      "learning_rate": 9.404151896909911e-06,
      "loss": 0.3082,
      "step": 3070
    },
    {
      "epoch": 0.16572230316766498,
      "grad_norm": 0.8870963461821364,
      "learning_rate": 9.4037724683968e-06,
      "loss": 0.3478,
      "step": 3071
    },
    {
      "epoch": 0.16577626679617938,
      "grad_norm": 1.182121552084783,
      "learning_rate": 9.403392927685322e-06,
      "loss": 0.5312,
      "step": 3072
    },
    {
      "epoch": 0.16583023042469375,
      "grad_norm": 1.1421941915077434,
      "learning_rate": 9.40301327478639e-06,
      "loss": 0.572,
      "step": 3073
    },
    {
      "epoch": 0.16588419405320814,
      "grad_norm": 1.38019449890495,
      "learning_rate": 9.402633509710913e-06,
      "loss": 0.6512,
      "step": 3074
    },
    {
      "epoch": 0.16593815768172251,
      "grad_norm": 0.8797995630826165,
      "learning_rate": 9.402253632469803e-06,
      "loss": 0.4058,
      "step": 3075
    },
    {
      "epoch": 0.1659921213102369,
      "grad_norm": 1.1216965841314825,
      "learning_rate": 9.401873643073981e-06,
      "loss": 0.4823,
      "step": 3076
    },
    {
      "epoch": 0.16604608493875128,
      "grad_norm": 1.071196255031116,
      "learning_rate": 9.40149354153437e-06,
      "loss": 0.4646,
      "step": 3077
    },
    {
      "epoch": 0.16610004856726565,
      "grad_norm": 1.1689014808512714,
      "learning_rate": 9.401113327861891e-06,
      "loss": 0.5937,
      "step": 3078
    },
    {
      "epoch": 0.16615401219578005,
      "grad_norm": 1.1496105290643719,
      "learning_rate": 9.400733002067472e-06,
      "loss": 0.634,
      "step": 3079
    },
    {
      "epoch": 0.16620797582429442,
      "grad_norm": 0.9086245552587248,
      "learning_rate": 9.400352564162047e-06,
      "loss": 0.4068,
      "step": 3080
    },
    {
      "epoch": 0.16626193945280882,
      "grad_norm": 1.23963551141849,
      "learning_rate": 9.399972014156548e-06,
      "loss": 0.5891,
      "step": 3081
    },
    {
      "epoch": 0.16631590308132319,
      "grad_norm": 0.8077414700341824,
      "learning_rate": 9.399591352061914e-06,
      "loss": 0.3298,
      "step": 3082
    },
    {
      "epoch": 0.16636986670983758,
      "grad_norm": 0.9904344852496164,
      "learning_rate": 9.39921057788908e-06,
      "loss": 0.4816,
      "step": 3083
    },
    {
      "epoch": 0.16642383033835195,
      "grad_norm": 0.8854066282235407,
      "learning_rate": 9.398829691648997e-06,
      "loss": 0.4646,
      "step": 3084
    },
    {
      "epoch": 0.16647779396686632,
      "grad_norm": 1.2972689095176384,
      "learning_rate": 9.39844869335261e-06,
      "loss": 0.5955,
      "step": 3085
    },
    {
      "epoch": 0.16653175759538072,
      "grad_norm": 0.9793100068946493,
      "learning_rate": 9.398067583010867e-06,
      "loss": 0.3875,
      "step": 3086
    },
    {
      "epoch": 0.1665857212238951,
      "grad_norm": 1.145494148297733,
      "learning_rate": 9.397686360634723e-06,
      "loss": 0.6392,
      "step": 3087
    },
    {
      "epoch": 0.1666396848524095,
      "grad_norm": 1.1219676551495839,
      "learning_rate": 9.397305026235136e-06,
      "loss": 0.5467,
      "step": 3088
    },
    {
      "epoch": 0.16669364848092386,
      "grad_norm": 1.4293147408556335,
      "learning_rate": 9.396923579823063e-06,
      "loss": 0.5576,
      "step": 3089
    },
    {
      "epoch": 0.16674761210943823,
      "grad_norm": 1.0875718719882101,
      "learning_rate": 9.39654202140947e-06,
      "loss": 0.6202,
      "step": 3090
    },
    {
      "epoch": 0.16680157573795262,
      "grad_norm": 1.1120292049738836,
      "learning_rate": 9.39616035100532e-06,
      "loss": 0.5581,
      "step": 3091
    },
    {
      "epoch": 0.166855539366467,
      "grad_norm": 1.0647139724354546,
      "learning_rate": 9.395778568621587e-06,
      "loss": 0.5451,
      "step": 3092
    },
    {
      "epoch": 0.1669095029949814,
      "grad_norm": 1.1375409615676562,
      "learning_rate": 9.395396674269242e-06,
      "loss": 0.4928,
      "step": 3093
    },
    {
      "epoch": 0.16696346662349576,
      "grad_norm": 0.815365525349067,
      "learning_rate": 9.39501466795926e-06,
      "loss": 0.396,
      "step": 3094
    },
    {
      "epoch": 0.16701743025201016,
      "grad_norm": 0.9776969430339892,
      "learning_rate": 9.39463254970262e-06,
      "loss": 0.5518,
      "step": 3095
    },
    {
      "epoch": 0.16707139388052453,
      "grad_norm": 1.0601102856547375,
      "learning_rate": 9.394250319510303e-06,
      "loss": 0.4114,
      "step": 3096
    },
    {
      "epoch": 0.1671253575090389,
      "grad_norm": 1.5050886559918186,
      "learning_rate": 9.393867977393301e-06,
      "loss": 0.516,
      "step": 3097
    },
    {
      "epoch": 0.1671793211375533,
      "grad_norm": 0.8597350697318071,
      "learning_rate": 9.393485523362598e-06,
      "loss": 0.435,
      "step": 3098
    },
    {
      "epoch": 0.16723328476606766,
      "grad_norm": 0.9872186768310999,
      "learning_rate": 9.393102957429186e-06,
      "loss": 0.4943,
      "step": 3099
    },
    {
      "epoch": 0.16728724839458206,
      "grad_norm": 1.0445663431609618,
      "learning_rate": 9.39272027960406e-06,
      "loss": 0.4823,
      "step": 3100
    },
    {
      "epoch": 0.16734121202309643,
      "grad_norm": 1.0634782719924891,
      "learning_rate": 9.392337489898221e-06,
      "loss": 0.5365,
      "step": 3101
    },
    {
      "epoch": 0.16739517565161083,
      "grad_norm": 0.998589561500629,
      "learning_rate": 9.391954588322671e-06,
      "loss": 0.4343,
      "step": 3102
    },
    {
      "epoch": 0.1674491392801252,
      "grad_norm": 0.9660567173279985,
      "learning_rate": 9.391571574888413e-06,
      "loss": 0.4845,
      "step": 3103
    },
    {
      "epoch": 0.16750310290863957,
      "grad_norm": 1.1680840656489786,
      "learning_rate": 9.391188449606455e-06,
      "loss": 0.5011,
      "step": 3104
    },
    {
      "epoch": 0.16755706653715396,
      "grad_norm": 0.8028264854819163,
      "learning_rate": 9.390805212487812e-06,
      "loss": 0.3304,
      "step": 3105
    },
    {
      "epoch": 0.16761103016566833,
      "grad_norm": 0.9638020102365698,
      "learning_rate": 9.390421863543493e-06,
      "loss": 0.4739,
      "step": 3106
    },
    {
      "epoch": 0.16766499379418273,
      "grad_norm": 1.2420489557987215,
      "learning_rate": 9.390038402784518e-06,
      "loss": 0.583,
      "step": 3107
    },
    {
      "epoch": 0.1677189574226971,
      "grad_norm": 0.8764612160942299,
      "learning_rate": 9.38965483022191e-06,
      "loss": 0.396,
      "step": 3108
    },
    {
      "epoch": 0.16777292105121147,
      "grad_norm": 0.9298058848908223,
      "learning_rate": 9.389271145866692e-06,
      "loss": 0.4538,
      "step": 3109
    },
    {
      "epoch": 0.16782688467972587,
      "grad_norm": 0.8843831186291994,
      "learning_rate": 9.388887349729892e-06,
      "loss": 0.3868,
      "step": 3110
    },
    {
      "epoch": 0.16788084830824024,
      "grad_norm": 0.9683556653072294,
      "learning_rate": 9.388503441822538e-06,
      "loss": 0.4548,
      "step": 3111
    },
    {
      "epoch": 0.16793481193675464,
      "grad_norm": 1.1594149310781166,
      "learning_rate": 9.388119422155669e-06,
      "loss": 0.5885,
      "step": 3112
    },
    {
      "epoch": 0.167988775565269,
      "grad_norm": 1.0731272354600097,
      "learning_rate": 9.387735290740318e-06,
      "loss": 0.5017,
      "step": 3113
    },
    {
      "epoch": 0.1680427391937834,
      "grad_norm": 1.2248377523260037,
      "learning_rate": 9.387351047587526e-06,
      "loss": 0.6311,
      "step": 3114
    },
    {
      "epoch": 0.16809670282229777,
      "grad_norm": 0.9878766630350391,
      "learning_rate": 9.386966692708337e-06,
      "loss": 0.5302,
      "step": 3115
    },
    {
      "epoch": 0.16815066645081214,
      "grad_norm": 1.1054680257046703,
      "learning_rate": 9.386582226113797e-06,
      "loss": 0.6424,
      "step": 3116
    },
    {
      "epoch": 0.16820463007932654,
      "grad_norm": 1.3494022635303744,
      "learning_rate": 9.386197647814957e-06,
      "loss": 0.6652,
      "step": 3117
    },
    {
      "epoch": 0.1682585937078409,
      "grad_norm": 0.908662956904971,
      "learning_rate": 9.385812957822872e-06,
      "loss": 0.282,
      "step": 3118
    },
    {
      "epoch": 0.1683125573363553,
      "grad_norm": 0.9769440317855215,
      "learning_rate": 9.385428156148594e-06,
      "loss": 0.4228,
      "step": 3119
    },
    {
      "epoch": 0.16836652096486968,
      "grad_norm": 1.3345610696828134,
      "learning_rate": 9.385043242803185e-06,
      "loss": 0.6465,
      "step": 3120
    },
    {
      "epoch": 0.16842048459338405,
      "grad_norm": 1.0283701558408973,
      "learning_rate": 9.384658217797709e-06,
      "loss": 0.5418,
      "step": 3121
    },
    {
      "epoch": 0.16847444822189844,
      "grad_norm": 1.0473900974859198,
      "learning_rate": 9.384273081143231e-06,
      "loss": 0.4831,
      "step": 3122
    },
    {
      "epoch": 0.1685284118504128,
      "grad_norm": 1.1231703724221165,
      "learning_rate": 9.383887832850818e-06,
      "loss": 0.4373,
      "step": 3123
    },
    {
      "epoch": 0.1685823754789272,
      "grad_norm": 0.9920505908283784,
      "learning_rate": 9.383502472931544e-06,
      "loss": 0.466,
      "step": 3124
    },
    {
      "epoch": 0.16863633910744158,
      "grad_norm": 0.9248699283439118,
      "learning_rate": 9.383117001396486e-06,
      "loss": 0.439,
      "step": 3125
    },
    {
      "epoch": 0.16869030273595598,
      "grad_norm": 1.1028877674318893,
      "learning_rate": 9.382731418256722e-06,
      "loss": 0.6322,
      "step": 3126
    },
    {
      "epoch": 0.16874426636447035,
      "grad_norm": 0.9265599504493932,
      "learning_rate": 9.382345723523334e-06,
      "loss": 0.4509,
      "step": 3127
    },
    {
      "epoch": 0.16879822999298472,
      "grad_norm": 1.074225159766606,
      "learning_rate": 9.381959917207406e-06,
      "loss": 0.4589,
      "step": 3128
    },
    {
      "epoch": 0.1688521936214991,
      "grad_norm": 1.3126964215060857,
      "learning_rate": 9.38157399932003e-06,
      "loss": 0.5964,
      "step": 3129
    },
    {
      "epoch": 0.16890615725001348,
      "grad_norm": 1.130087571507967,
      "learning_rate": 9.381187969872293e-06,
      "loss": 0.4258,
      "step": 3130
    },
    {
      "epoch": 0.16896012087852788,
      "grad_norm": 1.051620216315125,
      "learning_rate": 9.380801828875293e-06,
      "loss": 0.498,
      "step": 3131
    },
    {
      "epoch": 0.16901408450704225,
      "grad_norm": 1.0424098212459374,
      "learning_rate": 9.380415576340127e-06,
      "loss": 0.4507,
      "step": 3132
    },
    {
      "epoch": 0.16906804813555665,
      "grad_norm": 1.2055373278127448,
      "learning_rate": 9.380029212277894e-06,
      "loss": 0.4558,
      "step": 3133
    },
    {
      "epoch": 0.16912201176407102,
      "grad_norm": 1.1492780036121855,
      "learning_rate": 9.379642736699705e-06,
      "loss": 0.5091,
      "step": 3134
    },
    {
      "epoch": 0.1691759753925854,
      "grad_norm": 0.8874085665919307,
      "learning_rate": 9.37925614961666e-06,
      "loss": 0.4329,
      "step": 3135
    },
    {
      "epoch": 0.16922993902109978,
      "grad_norm": 1.0191130000583988,
      "learning_rate": 9.378869451039875e-06,
      "loss": 0.546,
      "step": 3136
    },
    {
      "epoch": 0.16928390264961415,
      "grad_norm": 1.1253211184489675,
      "learning_rate": 9.378482640980464e-06,
      "loss": 0.5971,
      "step": 3137
    },
    {
      "epoch": 0.16933786627812855,
      "grad_norm": 0.7600441046962375,
      "learning_rate": 9.378095719449542e-06,
      "loss": 0.3519,
      "step": 3138
    },
    {
      "epoch": 0.16939182990664292,
      "grad_norm": 0.9521796236362152,
      "learning_rate": 9.377708686458232e-06,
      "loss": 0.399,
      "step": 3139
    },
    {
      "epoch": 0.1694457935351573,
      "grad_norm": 1.050674325423393,
      "learning_rate": 9.377321542017653e-06,
      "loss": 0.5083,
      "step": 3140
    },
    {
      "epoch": 0.1694997571636717,
      "grad_norm": 1.1269299485204087,
      "learning_rate": 9.376934286138937e-06,
      "loss": 0.6998,
      "step": 3141
    },
    {
      "epoch": 0.16955372079218606,
      "grad_norm": 1.1621490340719962,
      "learning_rate": 9.376546918833213e-06,
      "loss": 0.5044,
      "step": 3142
    },
    {
      "epoch": 0.16960768442070046,
      "grad_norm": 1.0469693139614664,
      "learning_rate": 9.376159440111612e-06,
      "loss": 0.5744,
      "step": 3143
    },
    {
      "epoch": 0.16966164804921482,
      "grad_norm": 1.192464924568567,
      "learning_rate": 9.375771849985274e-06,
      "loss": 0.5223,
      "step": 3144
    },
    {
      "epoch": 0.16971561167772922,
      "grad_norm": 1.1163127645661057,
      "learning_rate": 9.375384148465336e-06,
      "loss": 0.4891,
      "step": 3145
    },
    {
      "epoch": 0.1697695753062436,
      "grad_norm": 1.102258566582523,
      "learning_rate": 9.37499633556294e-06,
      "loss": 0.5244,
      "step": 3146
    },
    {
      "epoch": 0.16982353893475796,
      "grad_norm": 1.0329631513717354,
      "learning_rate": 9.374608411289235e-06,
      "loss": 0.5739,
      "step": 3147
    },
    {
      "epoch": 0.16987750256327236,
      "grad_norm": 1.1048982274665755,
      "learning_rate": 9.37422037565537e-06,
      "loss": 0.5065,
      "step": 3148
    },
    {
      "epoch": 0.16993146619178673,
      "grad_norm": 1.2623232069609536,
      "learning_rate": 9.373832228672496e-06,
      "loss": 0.6486,
      "step": 3149
    },
    {
      "epoch": 0.16998542982030113,
      "grad_norm": 1.0598765701628676,
      "learning_rate": 9.37344397035177e-06,
      "loss": 0.5006,
      "step": 3150
    },
    {
      "epoch": 0.1700393934488155,
      "grad_norm": 1.1317557180272588,
      "learning_rate": 9.37305560070435e-06,
      "loss": 0.5813,
      "step": 3151
    },
    {
      "epoch": 0.1700933570773299,
      "grad_norm": 1.1754520574515648,
      "learning_rate": 9.3726671197414e-06,
      "loss": 0.6186,
      "step": 3152
    },
    {
      "epoch": 0.17014732070584426,
      "grad_norm": 0.8678252653715242,
      "learning_rate": 9.372278527474082e-06,
      "loss": 0.441,
      "step": 3153
    },
    {
      "epoch": 0.17020128433435863,
      "grad_norm": 1.08115063063288,
      "learning_rate": 9.371889823913568e-06,
      "loss": 0.598,
      "step": 3154
    },
    {
      "epoch": 0.17025524796287303,
      "grad_norm": 0.950660372804506,
      "learning_rate": 9.371501009071028e-06,
      "loss": 0.3981,
      "step": 3155
    },
    {
      "epoch": 0.1703092115913874,
      "grad_norm": 1.4665065497095973,
      "learning_rate": 9.371112082957636e-06,
      "loss": 0.6937,
      "step": 3156
    },
    {
      "epoch": 0.1703631752199018,
      "grad_norm": 0.7324722046788382,
      "learning_rate": 9.370723045584572e-06,
      "loss": 0.3634,
      "step": 3157
    },
    {
      "epoch": 0.17041713884841617,
      "grad_norm": 0.8367632956957777,
      "learning_rate": 9.370333896963016e-06,
      "loss": 0.311,
      "step": 3158
    },
    {
      "epoch": 0.17047110247693054,
      "grad_norm": 0.9664147070769433,
      "learning_rate": 9.369944637104154e-06,
      "loss": 0.3895,
      "step": 3159
    },
    {
      "epoch": 0.17052506610544493,
      "grad_norm": 0.7592322372672324,
      "learning_rate": 9.369555266019172e-06,
      "loss": 0.2704,
      "step": 3160
    },
    {
      "epoch": 0.1705790297339593,
      "grad_norm": 1.0911157977038604,
      "learning_rate": 9.369165783719263e-06,
      "loss": 0.5379,
      "step": 3161
    },
    {
      "epoch": 0.1706329933624737,
      "grad_norm": 1.066173691287015,
      "learning_rate": 9.368776190215621e-06,
      "loss": 0.5272,
      "step": 3162
    },
    {
      "epoch": 0.17068695699098807,
      "grad_norm": 1.3363042425935967,
      "learning_rate": 9.36838648551944e-06,
      "loss": 0.6508,
      "step": 3163
    },
    {
      "epoch": 0.17074092061950247,
      "grad_norm": 1.2487725767035265,
      "learning_rate": 9.367996669641922e-06,
      "loss": 0.6781,
      "step": 3164
    },
    {
      "epoch": 0.17079488424801684,
      "grad_norm": 0.9304481930527686,
      "learning_rate": 9.367606742594272e-06,
      "loss": 0.4372,
      "step": 3165
    },
    {
      "epoch": 0.1708488478765312,
      "grad_norm": 1.047154252405805,
      "learning_rate": 9.367216704387697e-06,
      "loss": 0.4792,
      "step": 3166
    },
    {
      "epoch": 0.1709028115050456,
      "grad_norm": 1.0416008138735509,
      "learning_rate": 9.366826555033407e-06,
      "loss": 0.4857,
      "step": 3167
    },
    {
      "epoch": 0.17095677513355997,
      "grad_norm": 1.069034862353067,
      "learning_rate": 9.366436294542613e-06,
      "loss": 0.5767,
      "step": 3168
    },
    {
      "epoch": 0.17101073876207437,
      "grad_norm": 0.9678460143389742,
      "learning_rate": 9.366045922926533e-06,
      "loss": 0.5385,
      "step": 3169
    },
    {
      "epoch": 0.17106470239058874,
      "grad_norm": 1.06093944632542,
      "learning_rate": 9.36565544019639e-06,
      "loss": 0.5412,
      "step": 3170
    },
    {
      "epoch": 0.17111866601910314,
      "grad_norm": 1.1966492709091994,
      "learning_rate": 9.3652648463634e-06,
      "loss": 0.5807,
      "step": 3171
    },
    {
      "epoch": 0.1711726296476175,
      "grad_norm": 0.9414369136681316,
      "learning_rate": 9.364874141438794e-06,
      "loss": 0.445,
      "step": 3172
    },
    {
      "epoch": 0.17122659327613188,
      "grad_norm": 0.9007917942082062,
      "learning_rate": 9.3644833254338e-06,
      "loss": 0.4046,
      "step": 3173
    },
    {
      "epoch": 0.17128055690464628,
      "grad_norm": 0.8691328597730115,
      "learning_rate": 9.36409239835965e-06,
      "loss": 0.4779,
      "step": 3174
    },
    {
      "epoch": 0.17133452053316064,
      "grad_norm": 0.98268329044992,
      "learning_rate": 9.363701360227582e-06,
      "loss": 0.4573,
      "step": 3175
    },
    {
      "epoch": 0.17138848416167504,
      "grad_norm": 1.1831323615552372,
      "learning_rate": 9.363310211048831e-06,
      "loss": 0.5779,
      "step": 3176
    },
    {
      "epoch": 0.1714424477901894,
      "grad_norm": 1.0182781491082997,
      "learning_rate": 9.36291895083464e-06,
      "loss": 0.44,
      "step": 3177
    },
    {
      "epoch": 0.17149641141870378,
      "grad_norm": 1.4296176623865242,
      "learning_rate": 9.36252757959626e-06,
      "loss": 0.7297,
      "step": 3178
    },
    {
      "epoch": 0.17155037504721818,
      "grad_norm": 0.9030071143021786,
      "learning_rate": 9.36213609734493e-06,
      "loss": 0.4115,
      "step": 3179
    },
    {
      "epoch": 0.17160433867573255,
      "grad_norm": 1.111632939999548,
      "learning_rate": 9.361744504091908e-06,
      "loss": 0.6168,
      "step": 3180
    },
    {
      "epoch": 0.17165830230424695,
      "grad_norm": 1.0346813646290283,
      "learning_rate": 9.361352799848445e-06,
      "loss": 0.5056,
      "step": 3181
    },
    {
      "epoch": 0.17171226593276132,
      "grad_norm": 1.2234205017832493,
      "learning_rate": 9.360960984625804e-06,
      "loss": 0.6338,
      "step": 3182
    },
    {
      "epoch": 0.1717662295612757,
      "grad_norm": 1.3361862240278448,
      "learning_rate": 9.360569058435241e-06,
      "loss": 0.4717,
      "step": 3183
    },
    {
      "epoch": 0.17182019318979008,
      "grad_norm": 1.0484649300735198,
      "learning_rate": 9.360177021288023e-06,
      "loss": 0.4227,
      "step": 3184
    },
    {
      "epoch": 0.17187415681830445,
      "grad_norm": 1.0655346388463542,
      "learning_rate": 9.359784873195416e-06,
      "loss": 0.464,
      "step": 3185
    },
    {
      "epoch": 0.17192812044681885,
      "grad_norm": 1.0355144351054362,
      "learning_rate": 9.359392614168693e-06,
      "loss": 0.5299,
      "step": 3186
    },
    {
      "epoch": 0.17198208407533322,
      "grad_norm": 0.9740925109337184,
      "learning_rate": 9.359000244219126e-06,
      "loss": 0.5065,
      "step": 3187
    },
    {
      "epoch": 0.17203604770384762,
      "grad_norm": 1.0953196260892126,
      "learning_rate": 9.358607763357993e-06,
      "loss": 0.5905,
      "step": 3188
    },
    {
      "epoch": 0.172090011332362,
      "grad_norm": 1.0523016671119958,
      "learning_rate": 9.358215171596573e-06,
      "loss": 0.5661,
      "step": 3189
    },
    {
      "epoch": 0.17214397496087636,
      "grad_norm": 1.2276682040883806,
      "learning_rate": 9.357822468946151e-06,
      "loss": 0.4918,
      "step": 3190
    },
    {
      "epoch": 0.17219793858939075,
      "grad_norm": 0.9251643008960653,
      "learning_rate": 9.357429655418015e-06,
      "loss": 0.4246,
      "step": 3191
    },
    {
      "epoch": 0.17225190221790512,
      "grad_norm": 0.8470788817677216,
      "learning_rate": 9.357036731023453e-06,
      "loss": 0.3816,
      "step": 3192
    },
    {
      "epoch": 0.17230586584641952,
      "grad_norm": 1.026408059475465,
      "learning_rate": 9.356643695773756e-06,
      "loss": 0.417,
      "step": 3193
    },
    {
      "epoch": 0.1723598294749339,
      "grad_norm": 0.9647376766991798,
      "learning_rate": 9.356250549680222e-06,
      "loss": 0.4904,
      "step": 3194
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.6809438894547291,
      "learning_rate": 9.355857292754152e-06,
      "loss": 0.2818,
      "step": 3195
    },
    {
      "epoch": 0.17246775673196266,
      "grad_norm": 1.2558224240508546,
      "learning_rate": 9.355463925006847e-06,
      "loss": 0.6028,
      "step": 3196
    },
    {
      "epoch": 0.17252172036047703,
      "grad_norm": 1.0383872508159173,
      "learning_rate": 9.355070446449614e-06,
      "loss": 0.4985,
      "step": 3197
    },
    {
      "epoch": 0.17257568398899142,
      "grad_norm": 1.0932739056654472,
      "learning_rate": 9.35467685709376e-06,
      "loss": 0.5205,
      "step": 3198
    },
    {
      "epoch": 0.1726296476175058,
      "grad_norm": 1.0351938552964597,
      "learning_rate": 9.354283156950599e-06,
      "loss": 0.514,
      "step": 3199
    },
    {
      "epoch": 0.1726836112460202,
      "grad_norm": 1.0586882652954654,
      "learning_rate": 9.353889346031445e-06,
      "loss": 0.4492,
      "step": 3200
    },
    {
      "epoch": 0.17273757487453456,
      "grad_norm": 1.0500731483979102,
      "learning_rate": 9.353495424347619e-06,
      "loss": 0.4969,
      "step": 3201
    },
    {
      "epoch": 0.17279153850304896,
      "grad_norm": 0.8389163707053308,
      "learning_rate": 9.353101391910438e-06,
      "loss": 0.3551,
      "step": 3202
    },
    {
      "epoch": 0.17284550213156333,
      "grad_norm": 1.304678420501069,
      "learning_rate": 9.352707248731232e-06,
      "loss": 0.6063,
      "step": 3203
    },
    {
      "epoch": 0.1728994657600777,
      "grad_norm": 1.2252522932989551,
      "learning_rate": 9.352312994821326e-06,
      "loss": 0.6369,
      "step": 3204
    },
    {
      "epoch": 0.1729534293885921,
      "grad_norm": 0.8893561599290669,
      "learning_rate": 9.35191863019205e-06,
      "loss": 0.3713,
      "step": 3205
    },
    {
      "epoch": 0.17300739301710646,
      "grad_norm": 1.1692360566313433,
      "learning_rate": 9.351524154854744e-06,
      "loss": 0.4404,
      "step": 3206
    },
    {
      "epoch": 0.17306135664562086,
      "grad_norm": 1.1258113219579153,
      "learning_rate": 9.351129568820741e-06,
      "loss": 0.7105,
      "step": 3207
    },
    {
      "epoch": 0.17311532027413523,
      "grad_norm": 1.1096574469198752,
      "learning_rate": 9.350734872101382e-06,
      "loss": 0.4757,
      "step": 3208
    },
    {
      "epoch": 0.1731692839026496,
      "grad_norm": 0.9013096492377883,
      "learning_rate": 9.350340064708012e-06,
      "loss": 0.4479,
      "step": 3209
    },
    {
      "epoch": 0.173223247531164,
      "grad_norm": 1.0969597487443385,
      "learning_rate": 9.34994514665198e-06,
      "loss": 0.4586,
      "step": 3210
    },
    {
      "epoch": 0.17327721115967837,
      "grad_norm": 1.2034394522788074,
      "learning_rate": 9.34955011794463e-06,
      "loss": 0.5529,
      "step": 3211
    },
    {
      "epoch": 0.17333117478819277,
      "grad_norm": 1.2147045614543763,
      "learning_rate": 9.349154978597323e-06,
      "loss": 0.6469,
      "step": 3212
    },
    {
      "epoch": 0.17338513841670714,
      "grad_norm": 1.1006007247622636,
      "learning_rate": 9.348759728621414e-06,
      "loss": 0.4802,
      "step": 3213
    },
    {
      "epoch": 0.17343910204522153,
      "grad_norm": 0.7835335407997517,
      "learning_rate": 9.348364368028258e-06,
      "loss": 0.3803,
      "step": 3214
    },
    {
      "epoch": 0.1734930656737359,
      "grad_norm": 0.9533931856040292,
      "learning_rate": 9.347968896829223e-06,
      "loss": 0.441,
      "step": 3215
    },
    {
      "epoch": 0.17354702930225027,
      "grad_norm": 1.0442213999689582,
      "learning_rate": 9.347573315035674e-06,
      "loss": 0.462,
      "step": 3216
    },
    {
      "epoch": 0.17360099293076467,
      "grad_norm": 1.303094685739711,
      "learning_rate": 9.347177622658981e-06,
      "loss": 0.4358,
      "step": 3217
    },
    {
      "epoch": 0.17365495655927904,
      "grad_norm": 1.0106174444243956,
      "learning_rate": 9.346781819710517e-06,
      "loss": 0.52,
      "step": 3218
    },
    {
      "epoch": 0.17370892018779344,
      "grad_norm": 1.1849587960670875,
      "learning_rate": 9.346385906201653e-06,
      "loss": 0.6035,
      "step": 3219
    },
    {
      "epoch": 0.1737628838163078,
      "grad_norm": 0.9121007311677226,
      "learning_rate": 9.345989882143774e-06,
      "loss": 0.4114,
      "step": 3220
    },
    {
      "epoch": 0.1738168474448222,
      "grad_norm": 1.138535168778012,
      "learning_rate": 9.34559374754826e-06,
      "loss": 0.643,
      "step": 3221
    },
    {
      "epoch": 0.17387081107333657,
      "grad_norm": 1.1697686250851338,
      "learning_rate": 9.345197502426493e-06,
      "loss": 0.5988,
      "step": 3222
    },
    {
      "epoch": 0.17392477470185094,
      "grad_norm": 1.0428752617830082,
      "learning_rate": 9.344801146789868e-06,
      "loss": 0.5793,
      "step": 3223
    },
    {
      "epoch": 0.17397873833036534,
      "grad_norm": 1.2675938613389781,
      "learning_rate": 9.344404680649772e-06,
      "loss": 0.4874,
      "step": 3224
    },
    {
      "epoch": 0.1740327019588797,
      "grad_norm": 1.06273226312279,
      "learning_rate": 9.344008104017602e-06,
      "loss": 0.5549,
      "step": 3225
    },
    {
      "epoch": 0.1740866655873941,
      "grad_norm": 0.9938776312782399,
      "learning_rate": 9.343611416904753e-06,
      "loss": 0.4367,
      "step": 3226
    },
    {
      "epoch": 0.17414062921590848,
      "grad_norm": 1.0033742631460751,
      "learning_rate": 9.34321461932263e-06,
      "loss": 0.5087,
      "step": 3227
    },
    {
      "epoch": 0.17419459284442285,
      "grad_norm": 1.0527137173871501,
      "learning_rate": 9.342817711282632e-06,
      "loss": 0.5159,
      "step": 3228
    },
    {
      "epoch": 0.17424855647293724,
      "grad_norm": 0.9823574897939077,
      "learning_rate": 9.342420692796174e-06,
      "loss": 0.5646,
      "step": 3229
    },
    {
      "epoch": 0.1743025201014516,
      "grad_norm": 0.8467193379999528,
      "learning_rate": 9.342023563874662e-06,
      "loss": 0.366,
      "step": 3230
    },
    {
      "epoch": 0.174356483729966,
      "grad_norm": 0.9368740858494746,
      "learning_rate": 9.34162632452951e-06,
      "loss": 0.3576,
      "step": 3231
    },
    {
      "epoch": 0.17441044735848038,
      "grad_norm": 1.099694877484971,
      "learning_rate": 9.341228974772136e-06,
      "loss": 0.5787,
      "step": 3232
    },
    {
      "epoch": 0.17446441098699478,
      "grad_norm": 1.2123322009480553,
      "learning_rate": 9.34083151461396e-06,
      "loss": 0.5488,
      "step": 3233
    },
    {
      "epoch": 0.17451837461550915,
      "grad_norm": 1.2148609589199066,
      "learning_rate": 9.340433944066404e-06,
      "loss": 0.569,
      "step": 3234
    },
    {
      "epoch": 0.17457233824402352,
      "grad_norm": 0.917629070811702,
      "learning_rate": 9.340036263140897e-06,
      "loss": 0.4794,
      "step": 3235
    },
    {
      "epoch": 0.17462630187253791,
      "grad_norm": 1.114222829729946,
      "learning_rate": 9.339638471848868e-06,
      "loss": 0.5525,
      "step": 3236
    },
    {
      "epoch": 0.17468026550105228,
      "grad_norm": 1.1922208820225446,
      "learning_rate": 9.339240570201751e-06,
      "loss": 0.5042,
      "step": 3237
    },
    {
      "epoch": 0.17473422912956668,
      "grad_norm": 0.9856681273461215,
      "learning_rate": 9.338842558210978e-06,
      "loss": 0.4067,
      "step": 3238
    },
    {
      "epoch": 0.17478819275808105,
      "grad_norm": 0.9974510264838814,
      "learning_rate": 9.338444435887992e-06,
      "loss": 0.4181,
      "step": 3239
    },
    {
      "epoch": 0.17484215638659542,
      "grad_norm": 1.247648327619207,
      "learning_rate": 9.338046203244236e-06,
      "loss": 0.6015,
      "step": 3240
    },
    {
      "epoch": 0.17489612001510982,
      "grad_norm": 0.9566208872313153,
      "learning_rate": 9.337647860291153e-06,
      "loss": 0.4391,
      "step": 3241
    },
    {
      "epoch": 0.1749500836436242,
      "grad_norm": 1.1677003247093316,
      "learning_rate": 9.337249407040192e-06,
      "loss": 0.7163,
      "step": 3242
    },
    {
      "epoch": 0.17500404727213859,
      "grad_norm": 1.045418805789118,
      "learning_rate": 9.336850843502806e-06,
      "loss": 0.5056,
      "step": 3243
    },
    {
      "epoch": 0.17505801090065295,
      "grad_norm": 0.9652699574270031,
      "learning_rate": 9.336452169690452e-06,
      "loss": 0.534,
      "step": 3244
    },
    {
      "epoch": 0.17511197452916735,
      "grad_norm": 1.047371902013245,
      "learning_rate": 9.336053385614583e-06,
      "loss": 0.4443,
      "step": 3245
    },
    {
      "epoch": 0.17516593815768172,
      "grad_norm": 1.0320307181082093,
      "learning_rate": 9.335654491286665e-06,
      "loss": 0.5416,
      "step": 3246
    },
    {
      "epoch": 0.1752199017861961,
      "grad_norm": 0.9316780206979535,
      "learning_rate": 9.335255486718163e-06,
      "loss": 0.4266,
      "step": 3247
    },
    {
      "epoch": 0.1752738654147105,
      "grad_norm": 1.0811059928942819,
      "learning_rate": 9.334856371920541e-06,
      "loss": 0.5733,
      "step": 3248
    },
    {
      "epoch": 0.17532782904322486,
      "grad_norm": 1.014106130214839,
      "learning_rate": 9.334457146905272e-06,
      "loss": 0.5464,
      "step": 3249
    },
    {
      "epoch": 0.17538179267173926,
      "grad_norm": 1.0217909285725855,
      "learning_rate": 9.33405781168383e-06,
      "loss": 0.5105,
      "step": 3250
    },
    {
      "epoch": 0.17543575630025363,
      "grad_norm": 1.0955810019076024,
      "learning_rate": 9.333658366267696e-06,
      "loss": 0.5502,
      "step": 3251
    },
    {
      "epoch": 0.17548971992876802,
      "grad_norm": 1.1791534620278106,
      "learning_rate": 9.333258810668342e-06,
      "loss": 0.6343,
      "step": 3252
    },
    {
      "epoch": 0.1755436835572824,
      "grad_norm": 0.9547441940443747,
      "learning_rate": 9.332859144897259e-06,
      "loss": 0.5532,
      "step": 3253
    },
    {
      "epoch": 0.17559764718579676,
      "grad_norm": 0.9940651715911237,
      "learning_rate": 9.332459368965931e-06,
      "loss": 0.4437,
      "step": 3254
    },
    {
      "epoch": 0.17565161081431116,
      "grad_norm": 1.1651945156910053,
      "learning_rate": 9.332059482885848e-06,
      "loss": 0.5651,
      "step": 3255
    },
    {
      "epoch": 0.17570557444282553,
      "grad_norm": 0.9455896428110837,
      "learning_rate": 9.331659486668504e-06,
      "loss": 0.5146,
      "step": 3256
    },
    {
      "epoch": 0.17575953807133993,
      "grad_norm": 1.1119095488510455,
      "learning_rate": 9.331259380325394e-06,
      "loss": 0.7488,
      "step": 3257
    },
    {
      "epoch": 0.1758135016998543,
      "grad_norm": 1.2603621376809295,
      "learning_rate": 9.33085916386802e-06,
      "loss": 0.5016,
      "step": 3258
    },
    {
      "epoch": 0.17586746532836867,
      "grad_norm": 0.9792258642789261,
      "learning_rate": 9.33045883730788e-06,
      "loss": 0.41,
      "step": 3259
    },
    {
      "epoch": 0.17592142895688306,
      "grad_norm": 0.9719434036938298,
      "learning_rate": 9.330058400656485e-06,
      "loss": 0.4073,
      "step": 3260
    },
    {
      "epoch": 0.17597539258539743,
      "grad_norm": 1.0244334264908586,
      "learning_rate": 9.329657853925341e-06,
      "loss": 0.4301,
      "step": 3261
    },
    {
      "epoch": 0.17602935621391183,
      "grad_norm": 1.2994254823114377,
      "learning_rate": 9.32925719712596e-06,
      "loss": 0.6954,
      "step": 3262
    },
    {
      "epoch": 0.1760833198424262,
      "grad_norm": 1.1513503786063108,
      "learning_rate": 9.32885643026986e-06,
      "loss": 0.4671,
      "step": 3263
    },
    {
      "epoch": 0.1761372834709406,
      "grad_norm": 1.0002628255651296,
      "learning_rate": 9.328455553368555e-06,
      "loss": 0.5593,
      "step": 3264
    },
    {
      "epoch": 0.17619124709945497,
      "grad_norm": 1.26962893102264,
      "learning_rate": 9.32805456643357e-06,
      "loss": 0.7464,
      "step": 3265
    },
    {
      "epoch": 0.17624521072796934,
      "grad_norm": 0.7824678197452039,
      "learning_rate": 9.327653469476429e-06,
      "loss": 0.3755,
      "step": 3266
    },
    {
      "epoch": 0.17629917435648373,
      "grad_norm": 1.051999440894643,
      "learning_rate": 9.327252262508658e-06,
      "loss": 0.4842,
      "step": 3267
    },
    {
      "epoch": 0.1763531379849981,
      "grad_norm": 1.15630611433821,
      "learning_rate": 9.326850945541792e-06,
      "loss": 0.4028,
      "step": 3268
    },
    {
      "epoch": 0.1764071016135125,
      "grad_norm": 1.1440553563911282,
      "learning_rate": 9.326449518587362e-06,
      "loss": 0.4883,
      "step": 3269
    },
    {
      "epoch": 0.17646106524202687,
      "grad_norm": 1.0617842800495336,
      "learning_rate": 9.326047981656907e-06,
      "loss": 0.676,
      "step": 3270
    },
    {
      "epoch": 0.17651502887054127,
      "grad_norm": 1.5971815549102086,
      "learning_rate": 9.325646334761968e-06,
      "loss": 0.683,
      "step": 3271
    },
    {
      "epoch": 0.17656899249905564,
      "grad_norm": 1.1659009036449632,
      "learning_rate": 9.325244577914087e-06,
      "loss": 0.5659,
      "step": 3272
    },
    {
      "epoch": 0.17662295612757,
      "grad_norm": 0.9452383221006792,
      "learning_rate": 9.32484271112481e-06,
      "loss": 0.4219,
      "step": 3273
    },
    {
      "epoch": 0.1766769197560844,
      "grad_norm": 0.9412400911991763,
      "learning_rate": 9.324440734405691e-06,
      "loss": 0.551,
      "step": 3274
    },
    {
      "epoch": 0.17673088338459877,
      "grad_norm": 0.8808292727968727,
      "learning_rate": 9.32403864776828e-06,
      "loss": 0.4037,
      "step": 3275
    },
    {
      "epoch": 0.17678484701311317,
      "grad_norm": 1.039745305177311,
      "learning_rate": 9.323636451224134e-06,
      "loss": 0.5608,
      "step": 3276
    },
    {
      "epoch": 0.17683881064162754,
      "grad_norm": 0.9620416422151222,
      "learning_rate": 9.323234144784813e-06,
      "loss": 0.4995,
      "step": 3277
    },
    {
      "epoch": 0.1768927742701419,
      "grad_norm": 0.815618409367972,
      "learning_rate": 9.32283172846188e-06,
      "loss": 0.3628,
      "step": 3278
    },
    {
      "epoch": 0.1769467378986563,
      "grad_norm": 1.0547044649589403,
      "learning_rate": 9.3224292022669e-06,
      "loss": 0.51,
      "step": 3279
    },
    {
      "epoch": 0.17700070152717068,
      "grad_norm": 1.3070943342158574,
      "learning_rate": 9.322026566211445e-06,
      "loss": 0.5071,
      "step": 3280
    },
    {
      "epoch": 0.17705466515568508,
      "grad_norm": 1.1735337901435394,
      "learning_rate": 9.321623820307083e-06,
      "loss": 0.5709,
      "step": 3281
    },
    {
      "epoch": 0.17710862878419945,
      "grad_norm": 1.2701511872697282,
      "learning_rate": 9.321220964565389e-06,
      "loss": 0.4199,
      "step": 3282
    },
    {
      "epoch": 0.17716259241271384,
      "grad_norm": 1.1539412480526907,
      "learning_rate": 9.320817998997945e-06,
      "loss": 0.6511,
      "step": 3283
    },
    {
      "epoch": 0.1772165560412282,
      "grad_norm": 1.0263610283883908,
      "learning_rate": 9.32041492361633e-06,
      "loss": 0.4635,
      "step": 3284
    },
    {
      "epoch": 0.17727051966974258,
      "grad_norm": 0.9856630254617735,
      "learning_rate": 9.320011738432131e-06,
      "loss": 0.4157,
      "step": 3285
    },
    {
      "epoch": 0.17732448329825698,
      "grad_norm": 0.8543579872459929,
      "learning_rate": 9.319608443456935e-06,
      "loss": 0.3369,
      "step": 3286
    },
    {
      "epoch": 0.17737844692677135,
      "grad_norm": 1.2337477370201717,
      "learning_rate": 9.319205038702332e-06,
      "loss": 0.7126,
      "step": 3287
    },
    {
      "epoch": 0.17743241055528575,
      "grad_norm": 0.9419621395036354,
      "learning_rate": 9.318801524179916e-06,
      "loss": 0.3209,
      "step": 3288
    },
    {
      "epoch": 0.17748637418380012,
      "grad_norm": 1.0582641777403659,
      "learning_rate": 9.318397899901288e-06,
      "loss": 0.633,
      "step": 3289
    },
    {
      "epoch": 0.1775403378123145,
      "grad_norm": 0.9910102928148292,
      "learning_rate": 9.317994165878047e-06,
      "loss": 0.4484,
      "step": 3290
    },
    {
      "epoch": 0.17759430144082888,
      "grad_norm": 0.9121037577456016,
      "learning_rate": 9.317590322121793e-06,
      "loss": 0.404,
      "step": 3291
    },
    {
      "epoch": 0.17764826506934325,
      "grad_norm": 0.9437541509326917,
      "learning_rate": 9.317186368644136e-06,
      "loss": 0.3371,
      "step": 3292
    },
    {
      "epoch": 0.17770222869785765,
      "grad_norm": 1.0328304649893163,
      "learning_rate": 9.316782305456687e-06,
      "loss": 0.6038,
      "step": 3293
    },
    {
      "epoch": 0.17775619232637202,
      "grad_norm": 0.9688883352092765,
      "learning_rate": 9.316378132571057e-06,
      "loss": 0.531,
      "step": 3294
    },
    {
      "epoch": 0.17781015595488642,
      "grad_norm": 1.105327546125664,
      "learning_rate": 9.315973849998865e-06,
      "loss": 0.5082,
      "step": 3295
    },
    {
      "epoch": 0.1778641195834008,
      "grad_norm": 0.8824029983111507,
      "learning_rate": 9.315569457751727e-06,
      "loss": 0.3592,
      "step": 3296
    },
    {
      "epoch": 0.17791808321191516,
      "grad_norm": 0.9852205024300851,
      "learning_rate": 9.31516495584127e-06,
      "loss": 0.5855,
      "step": 3297
    },
    {
      "epoch": 0.17797204684042955,
      "grad_norm": 0.9786839663225962,
      "learning_rate": 9.314760344279115e-06,
      "loss": 0.3922,
      "step": 3298
    },
    {
      "epoch": 0.17802601046894392,
      "grad_norm": 1.0900601031427941,
      "learning_rate": 9.314355623076893e-06,
      "loss": 0.4363,
      "step": 3299
    },
    {
      "epoch": 0.17807997409745832,
      "grad_norm": 1.3321222386509908,
      "learning_rate": 9.313950792246235e-06,
      "loss": 0.6845,
      "step": 3300
    },
    {
      "epoch": 0.1781339377259727,
      "grad_norm": 1.0227741303943747,
      "learning_rate": 9.313545851798781e-06,
      "loss": 0.4957,
      "step": 3301
    },
    {
      "epoch": 0.1781879013544871,
      "grad_norm": 1.0989740181221477,
      "learning_rate": 9.313140801746163e-06,
      "loss": 0.5653,
      "step": 3302
    },
    {
      "epoch": 0.17824186498300146,
      "grad_norm": 1.170586723336589,
      "learning_rate": 9.312735642100028e-06,
      "loss": 0.7054,
      "step": 3303
    },
    {
      "epoch": 0.17829582861151583,
      "grad_norm": 1.1674203575265656,
      "learning_rate": 9.312330372872017e-06,
      "loss": 0.4465,
      "step": 3304
    },
    {
      "epoch": 0.17834979224003022,
      "grad_norm": 0.9738439055009164,
      "learning_rate": 9.311924994073778e-06,
      "loss": 0.3496,
      "step": 3305
    },
    {
      "epoch": 0.1784037558685446,
      "grad_norm": 1.0672067509862484,
      "learning_rate": 9.311519505716963e-06,
      "loss": 0.5504,
      "step": 3306
    },
    {
      "epoch": 0.178457719497059,
      "grad_norm": 1.041829780251987,
      "learning_rate": 9.31111390781323e-06,
      "loss": 0.4337,
      "step": 3307
    },
    {
      "epoch": 0.17851168312557336,
      "grad_norm": 1.075398756699929,
      "learning_rate": 9.310708200374228e-06,
      "loss": 0.62,
      "step": 3308
    },
    {
      "epoch": 0.17856564675408773,
      "grad_norm": 1.02371527646056,
      "learning_rate": 9.310302383411622e-06,
      "loss": 0.4891,
      "step": 3309
    },
    {
      "epoch": 0.17861961038260213,
      "grad_norm": 0.9560526469992675,
      "learning_rate": 9.309896456937078e-06,
      "loss": 0.357,
      "step": 3310
    },
    {
      "epoch": 0.1786735740111165,
      "grad_norm": 0.9730731129826901,
      "learning_rate": 9.309490420962258e-06,
      "loss": 0.4397,
      "step": 3311
    },
    {
      "epoch": 0.1787275376396309,
      "grad_norm": 1.0152941707725756,
      "learning_rate": 9.309084275498835e-06,
      "loss": 0.44,
      "step": 3312
    },
    {
      "epoch": 0.17878150126814527,
      "grad_norm": 1.0142110520354706,
      "learning_rate": 9.308678020558483e-06,
      "loss": 0.6049,
      "step": 3313
    },
    {
      "epoch": 0.17883546489665966,
      "grad_norm": 1.0297795741479252,
      "learning_rate": 9.308271656152874e-06,
      "loss": 0.5356,
      "step": 3314
    },
    {
      "epoch": 0.17888942852517403,
      "grad_norm": 0.998483510934037,
      "learning_rate": 9.30786518229369e-06,
      "loss": 0.4465,
      "step": 3315
    },
    {
      "epoch": 0.1789433921536884,
      "grad_norm": 1.2140340542509656,
      "learning_rate": 9.307458598992614e-06,
      "loss": 0.6528,
      "step": 3316
    },
    {
      "epoch": 0.1789973557822028,
      "grad_norm": 1.1416732007239871,
      "learning_rate": 9.30705190626133e-06,
      "loss": 0.6565,
      "step": 3317
    },
    {
      "epoch": 0.17905131941071717,
      "grad_norm": 1.0686693098573046,
      "learning_rate": 9.306645104111526e-06,
      "loss": 0.5676,
      "step": 3318
    },
    {
      "epoch": 0.17910528303923157,
      "grad_norm": 1.1185008636101514,
      "learning_rate": 9.306238192554897e-06,
      "loss": 0.5591,
      "step": 3319
    },
    {
      "epoch": 0.17915924666774594,
      "grad_norm": 1.2855477621968296,
      "learning_rate": 9.305831171603137e-06,
      "loss": 0.6093,
      "step": 3320
    },
    {
      "epoch": 0.17921321029626033,
      "grad_norm": 1.1158796433255014,
      "learning_rate": 9.305424041267943e-06,
      "loss": 0.4844,
      "step": 3321
    },
    {
      "epoch": 0.1792671739247747,
      "grad_norm": 0.7429794435644391,
      "learning_rate": 9.305016801561017e-06,
      "loss": 0.3228,
      "step": 3322
    },
    {
      "epoch": 0.17932113755328907,
      "grad_norm": 1.3104756570822056,
      "learning_rate": 9.304609452494064e-06,
      "loss": 0.6458,
      "step": 3323
    },
    {
      "epoch": 0.17937510118180347,
      "grad_norm": 1.2936074729727534,
      "learning_rate": 9.30420199407879e-06,
      "loss": 0.7339,
      "step": 3324
    },
    {
      "epoch": 0.17942906481031784,
      "grad_norm": 0.8958662939192786,
      "learning_rate": 9.303794426326906e-06,
      "loss": 0.4413,
      "step": 3325
    },
    {
      "epoch": 0.17948302843883224,
      "grad_norm": 1.1710054067687343,
      "learning_rate": 9.303386749250129e-06,
      "loss": 0.5735,
      "step": 3326
    },
    {
      "epoch": 0.1795369920673466,
      "grad_norm": 1.0309293850514063,
      "learning_rate": 9.302978962860172e-06,
      "loss": 0.559,
      "step": 3327
    },
    {
      "epoch": 0.17959095569586098,
      "grad_norm": 1.0674542284716562,
      "learning_rate": 9.302571067168758e-06,
      "loss": 0.4708,
      "step": 3328
    },
    {
      "epoch": 0.17964491932437537,
      "grad_norm": 1.2177004632352904,
      "learning_rate": 9.30216306218761e-06,
      "loss": 0.6017,
      "step": 3329
    },
    {
      "epoch": 0.17969888295288974,
      "grad_norm": 1.007711145174143,
      "learning_rate": 9.301754947928451e-06,
      "loss": 0.4493,
      "step": 3330
    },
    {
      "epoch": 0.17975284658140414,
      "grad_norm": 0.9700419388779866,
      "learning_rate": 9.301346724403015e-06,
      "loss": 0.5282,
      "step": 3331
    },
    {
      "epoch": 0.1798068102099185,
      "grad_norm": 0.8812928901911473,
      "learning_rate": 9.300938391623034e-06,
      "loss": 0.3274,
      "step": 3332
    },
    {
      "epoch": 0.1798607738384329,
      "grad_norm": 0.960260917232155,
      "learning_rate": 9.300529949600243e-06,
      "loss": 0.4369,
      "step": 3333
    },
    {
      "epoch": 0.17991473746694728,
      "grad_norm": 0.9199512596536168,
      "learning_rate": 9.30012139834638e-06,
      "loss": 0.4801,
      "step": 3334
    },
    {
      "epoch": 0.17996870109546165,
      "grad_norm": 1.1603865245724476,
      "learning_rate": 9.299712737873188e-06,
      "loss": 0.5125,
      "step": 3335
    },
    {
      "epoch": 0.18002266472397604,
      "grad_norm": 0.8271989275867807,
      "learning_rate": 9.299303968192414e-06,
      "loss": 0.3391,
      "step": 3336
    },
    {
      "epoch": 0.18007662835249041,
      "grad_norm": 1.1428907308232006,
      "learning_rate": 9.298895089315802e-06,
      "loss": 0.4528,
      "step": 3337
    },
    {
      "epoch": 0.1801305919810048,
      "grad_norm": 1.0306078452013545,
      "learning_rate": 9.29848610125511e-06,
      "loss": 0.5307,
      "step": 3338
    },
    {
      "epoch": 0.18018455560951918,
      "grad_norm": 1.0699440464244176,
      "learning_rate": 9.298077004022089e-06,
      "loss": 0.5289,
      "step": 3339
    },
    {
      "epoch": 0.18023851923803358,
      "grad_norm": 1.2736183734576514,
      "learning_rate": 9.297667797628498e-06,
      "loss": 0.4943,
      "step": 3340
    },
    {
      "epoch": 0.18029248286654795,
      "grad_norm": 1.0752536850350756,
      "learning_rate": 9.297258482086095e-06,
      "loss": 0.5378,
      "step": 3341
    },
    {
      "epoch": 0.18034644649506232,
      "grad_norm": 1.0861044149618735,
      "learning_rate": 9.296849057406646e-06,
      "loss": 0.5236,
      "step": 3342
    },
    {
      "epoch": 0.18040041012357672,
      "grad_norm": 0.880314867656759,
      "learning_rate": 9.29643952360192e-06,
      "loss": 0.4368,
      "step": 3343
    },
    {
      "epoch": 0.18045437375209109,
      "grad_norm": 1.108225132249348,
      "learning_rate": 9.296029880683687e-06,
      "loss": 0.5206,
      "step": 3344
    },
    {
      "epoch": 0.18050833738060548,
      "grad_norm": 0.9241104486220661,
      "learning_rate": 9.295620128663718e-06,
      "loss": 0.4938,
      "step": 3345
    },
    {
      "epoch": 0.18056230100911985,
      "grad_norm": 1.2673764791630482,
      "learning_rate": 9.29521026755379e-06,
      "loss": 0.4529,
      "step": 3346
    },
    {
      "epoch": 0.18061626463763422,
      "grad_norm": 0.9680492789836844,
      "learning_rate": 9.294800297365686e-06,
      "loss": 0.5137,
      "step": 3347
    },
    {
      "epoch": 0.18067022826614862,
      "grad_norm": 0.8640926526322925,
      "learning_rate": 9.294390218111189e-06,
      "loss": 0.4024,
      "step": 3348
    },
    {
      "epoch": 0.180724191894663,
      "grad_norm": 0.8898384559968857,
      "learning_rate": 9.293980029802081e-06,
      "loss": 0.3496,
      "step": 3349
    },
    {
      "epoch": 0.1807781555231774,
      "grad_norm": 1.0346616866166647,
      "learning_rate": 9.293569732450152e-06,
      "loss": 0.4422,
      "step": 3350
    },
    {
      "epoch": 0.18083211915169176,
      "grad_norm": 0.9168741296014665,
      "learning_rate": 9.293159326067199e-06,
      "loss": 0.4061,
      "step": 3351
    },
    {
      "epoch": 0.18088608278020615,
      "grad_norm": 1.1574232540369924,
      "learning_rate": 9.292748810665011e-06,
      "loss": 0.7215,
      "step": 3352
    },
    {
      "epoch": 0.18094004640872052,
      "grad_norm": 1.0864359202445553,
      "learning_rate": 9.292338186255392e-06,
      "loss": 0.489,
      "step": 3353
    },
    {
      "epoch": 0.1809940100372349,
      "grad_norm": 1.1430777170980695,
      "learning_rate": 9.291927452850142e-06,
      "loss": 0.6622,
      "step": 3354
    },
    {
      "epoch": 0.1810479736657493,
      "grad_norm": 0.9433634004684675,
      "learning_rate": 9.291516610461064e-06,
      "loss": 0.6419,
      "step": 3355
    },
    {
      "epoch": 0.18110193729426366,
      "grad_norm": 0.8835197466182724,
      "learning_rate": 9.291105659099969e-06,
      "loss": 0.4394,
      "step": 3356
    },
    {
      "epoch": 0.18115590092277806,
      "grad_norm": 1.0789301874473527,
      "learning_rate": 9.290694598778665e-06,
      "loss": 0.4438,
      "step": 3357
    },
    {
      "epoch": 0.18120986455129243,
      "grad_norm": 0.9060626523725631,
      "learning_rate": 9.29028342950897e-06,
      "loss": 0.4452,
      "step": 3358
    },
    {
      "epoch": 0.18126382817980682,
      "grad_norm": 1.1660051766039203,
      "learning_rate": 9.2898721513027e-06,
      "loss": 0.5056,
      "step": 3359
    },
    {
      "epoch": 0.1813177918083212,
      "grad_norm": 1.1512519747094312,
      "learning_rate": 9.289460764171672e-06,
      "loss": 0.5044,
      "step": 3360
    },
    {
      "epoch": 0.18137175543683556,
      "grad_norm": 0.9380121697826385,
      "learning_rate": 9.289049268127713e-06,
      "loss": 0.4698,
      "step": 3361
    },
    {
      "epoch": 0.18142571906534996,
      "grad_norm": 1.0702939795518407,
      "learning_rate": 9.288637663182651e-06,
      "loss": 0.4988,
      "step": 3362
    },
    {
      "epoch": 0.18147968269386433,
      "grad_norm": 1.1465321957186811,
      "learning_rate": 9.288225949348314e-06,
      "loss": 0.6344,
      "step": 3363
    },
    {
      "epoch": 0.18153364632237873,
      "grad_norm": 1.162711795318182,
      "learning_rate": 9.287814126636536e-06,
      "loss": 0.7037,
      "step": 3364
    },
    {
      "epoch": 0.1815876099508931,
      "grad_norm": 0.913961290755787,
      "learning_rate": 9.287402195059152e-06,
      "loss": 0.4598,
      "step": 3365
    },
    {
      "epoch": 0.18164157357940747,
      "grad_norm": 0.9974714743613253,
      "learning_rate": 9.286990154628003e-06,
      "loss": 0.409,
      "step": 3366
    },
    {
      "epoch": 0.18169553720792186,
      "grad_norm": 0.938876778683415,
      "learning_rate": 9.286578005354929e-06,
      "loss": 0.3952,
      "step": 3367
    },
    {
      "epoch": 0.18174950083643623,
      "grad_norm": 1.2622041805409798,
      "learning_rate": 9.286165747251777e-06,
      "loss": 0.4929,
      "step": 3368
    },
    {
      "epoch": 0.18180346446495063,
      "grad_norm": 1.0862165460584956,
      "learning_rate": 9.285753380330397e-06,
      "loss": 0.4817,
      "step": 3369
    },
    {
      "epoch": 0.181857428093465,
      "grad_norm": 1.1038101367234663,
      "learning_rate": 9.28534090460264e-06,
      "loss": 0.4966,
      "step": 3370
    },
    {
      "epoch": 0.1819113917219794,
      "grad_norm": 1.1529182834691796,
      "learning_rate": 9.284928320080359e-06,
      "loss": 0.4643,
      "step": 3371
    },
    {
      "epoch": 0.18196535535049377,
      "grad_norm": 0.9137565597669305,
      "learning_rate": 9.284515626775416e-06,
      "loss": 0.3574,
      "step": 3372
    },
    {
      "epoch": 0.18201931897900814,
      "grad_norm": 0.8673277395531713,
      "learning_rate": 9.284102824699667e-06,
      "loss": 0.3566,
      "step": 3373
    },
    {
      "epoch": 0.18207328260752254,
      "grad_norm": 1.1428203569066189,
      "learning_rate": 9.283689913864982e-06,
      "loss": 0.7193,
      "step": 3374
    },
    {
      "epoch": 0.1821272462360369,
      "grad_norm": 1.1846749187720897,
      "learning_rate": 9.283276894283224e-06,
      "loss": 0.6534,
      "step": 3375
    },
    {
      "epoch": 0.1821812098645513,
      "grad_norm": 1.2627483587706738,
      "learning_rate": 9.282863765966264e-06,
      "loss": 0.6325,
      "step": 3376
    },
    {
      "epoch": 0.18223517349306567,
      "grad_norm": 1.0622289099828244,
      "learning_rate": 9.28245052892598e-06,
      "loss": 0.4054,
      "step": 3377
    },
    {
      "epoch": 0.18228913712158004,
      "grad_norm": 0.8909687306982086,
      "learning_rate": 9.282037183174246e-06,
      "loss": 0.3957,
      "step": 3378
    },
    {
      "epoch": 0.18234310075009444,
      "grad_norm": 0.9890775774933492,
      "learning_rate": 9.28162372872294e-06,
      "loss": 0.4631,
      "step": 3379
    },
    {
      "epoch": 0.1823970643786088,
      "grad_norm": 0.9572318660208106,
      "learning_rate": 9.281210165583948e-06,
      "loss": 0.4878,
      "step": 3380
    },
    {
      "epoch": 0.1824510280071232,
      "grad_norm": 0.7574305009864738,
      "learning_rate": 9.280796493769154e-06,
      "loss": 0.3604,
      "step": 3381
    },
    {
      "epoch": 0.18250499163563758,
      "grad_norm": 1.0412049330231905,
      "learning_rate": 9.280382713290449e-06,
      "loss": 0.6203,
      "step": 3382
    },
    {
      "epoch": 0.18255895526415197,
      "grad_norm": 0.8475356447430414,
      "learning_rate": 9.279968824159726e-06,
      "loss": 0.4097,
      "step": 3383
    },
    {
      "epoch": 0.18261291889266634,
      "grad_norm": 1.0736785644586544,
      "learning_rate": 9.279554826388878e-06,
      "loss": 0.5533,
      "step": 3384
    },
    {
      "epoch": 0.1826668825211807,
      "grad_norm": 1.1067516300997777,
      "learning_rate": 9.279140719989806e-06,
      "loss": 0.4882,
      "step": 3385
    },
    {
      "epoch": 0.1827208461496951,
      "grad_norm": 1.0654879691357955,
      "learning_rate": 9.278726504974412e-06,
      "loss": 0.5979,
      "step": 3386
    },
    {
      "epoch": 0.18277480977820948,
      "grad_norm": 0.9850064813014754,
      "learning_rate": 9.278312181354599e-06,
      "loss": 0.5414,
      "step": 3387
    },
    {
      "epoch": 0.18282877340672388,
      "grad_norm": 0.8460700886497378,
      "learning_rate": 9.277897749142276e-06,
      "loss": 0.3356,
      "step": 3388
    },
    {
      "epoch": 0.18288273703523825,
      "grad_norm": 0.9617044534574388,
      "learning_rate": 9.277483208349357e-06,
      "loss": 0.5169,
      "step": 3389
    },
    {
      "epoch": 0.18293670066375264,
      "grad_norm": 1.023612753160458,
      "learning_rate": 9.27706855898775e-06,
      "loss": 0.4592,
      "step": 3390
    },
    {
      "epoch": 0.182990664292267,
      "grad_norm": 0.8381025289763231,
      "learning_rate": 9.276653801069379e-06,
      "loss": 0.3695,
      "step": 3391
    },
    {
      "epoch": 0.18304462792078138,
      "grad_norm": 1.0997135090772479,
      "learning_rate": 9.27623893460616e-06,
      "loss": 0.4523,
      "step": 3392
    },
    {
      "epoch": 0.18309859154929578,
      "grad_norm": 0.8530527387299963,
      "learning_rate": 9.275823959610019e-06,
      "loss": 0.425,
      "step": 3393
    },
    {
      "epoch": 0.18315255517781015,
      "grad_norm": 0.9607476891833553,
      "learning_rate": 9.275408876092884e-06,
      "loss": 0.4553,
      "step": 3394
    },
    {
      "epoch": 0.18320651880632455,
      "grad_norm": 1.1152480720624944,
      "learning_rate": 9.27499368406668e-06,
      "loss": 0.5299,
      "step": 3395
    },
    {
      "epoch": 0.18326048243483892,
      "grad_norm": 0.9176787252004119,
      "learning_rate": 9.274578383543346e-06,
      "loss": 0.4425,
      "step": 3396
    },
    {
      "epoch": 0.1833144460633533,
      "grad_norm": 1.0642023660642137,
      "learning_rate": 9.274162974534814e-06,
      "loss": 0.4749,
      "step": 3397
    },
    {
      "epoch": 0.18336840969186768,
      "grad_norm": 1.2651195464192453,
      "learning_rate": 9.273747457053025e-06,
      "loss": 0.6713,
      "step": 3398
    },
    {
      "epoch": 0.18342237332038205,
      "grad_norm": 1.0580599975443903,
      "learning_rate": 9.27333183110992e-06,
      "loss": 0.48,
      "step": 3399
    },
    {
      "epoch": 0.18347633694889645,
      "grad_norm": 1.0845136587764994,
      "learning_rate": 9.272916096717447e-06,
      "loss": 0.481,
      "step": 3400
    },
    {
      "epoch": 0.18353030057741082,
      "grad_norm": 1.2201654816547105,
      "learning_rate": 9.272500253887551e-06,
      "loss": 0.5575,
      "step": 3401
    },
    {
      "epoch": 0.18358426420592522,
      "grad_norm": 1.025715500242841,
      "learning_rate": 9.272084302632187e-06,
      "loss": 0.4312,
      "step": 3402
    },
    {
      "epoch": 0.1836382278344396,
      "grad_norm": 1.2405464753238307,
      "learning_rate": 9.27166824296331e-06,
      "loss": 0.6212,
      "step": 3403
    },
    {
      "epoch": 0.18369219146295396,
      "grad_norm": 0.9546881179283732,
      "learning_rate": 9.271252074892875e-06,
      "loss": 0.4591,
      "step": 3404
    },
    {
      "epoch": 0.18374615509146836,
      "grad_norm": 1.4233247038148675,
      "learning_rate": 9.270835798432845e-06,
      "loss": 0.6821,
      "step": 3405
    },
    {
      "epoch": 0.18380011871998272,
      "grad_norm": 0.9455876618440167,
      "learning_rate": 9.270419413595184e-06,
      "loss": 0.3976,
      "step": 3406
    },
    {
      "epoch": 0.18385408234849712,
      "grad_norm": 0.9898702111549852,
      "learning_rate": 9.27000292039186e-06,
      "loss": 0.5434,
      "step": 3407
    },
    {
      "epoch": 0.1839080459770115,
      "grad_norm": 0.9388878687183496,
      "learning_rate": 9.269586318834841e-06,
      "loss": 0.4107,
      "step": 3408
    },
    {
      "epoch": 0.1839620096055259,
      "grad_norm": 1.2529233911758717,
      "learning_rate": 9.269169608936105e-06,
      "loss": 0.6721,
      "step": 3409
    },
    {
      "epoch": 0.18401597323404026,
      "grad_norm": 0.9736496451066581,
      "learning_rate": 9.268752790707625e-06,
      "loss": 0.4169,
      "step": 3410
    },
    {
      "epoch": 0.18406993686255463,
      "grad_norm": 1.0346357610468684,
      "learning_rate": 9.268335864161381e-06,
      "loss": 0.6374,
      "step": 3411
    },
    {
      "epoch": 0.18412390049106903,
      "grad_norm": 1.0285352222222677,
      "learning_rate": 9.267918829309356e-06,
      "loss": 0.4478,
      "step": 3412
    },
    {
      "epoch": 0.1841778641195834,
      "grad_norm": 1.1646391593992762,
      "learning_rate": 9.267501686163539e-06,
      "loss": 0.5021,
      "step": 3413
    },
    {
      "epoch": 0.1842318277480978,
      "grad_norm": 1.1542285800713197,
      "learning_rate": 9.267084434735916e-06,
      "loss": 0.5561,
      "step": 3414
    },
    {
      "epoch": 0.18428579137661216,
      "grad_norm": 0.8891687949404169,
      "learning_rate": 9.26666707503848e-06,
      "loss": 0.4995,
      "step": 3415
    },
    {
      "epoch": 0.18433975500512653,
      "grad_norm": 1.0160714571981209,
      "learning_rate": 9.266249607083225e-06,
      "loss": 0.5065,
      "step": 3416
    },
    {
      "epoch": 0.18439371863364093,
      "grad_norm": 1.0374493742540898,
      "learning_rate": 9.265832030882152e-06,
      "loss": 0.566,
      "step": 3417
    },
    {
      "epoch": 0.1844476822621553,
      "grad_norm": 0.8579524403058835,
      "learning_rate": 9.265414346447261e-06,
      "loss": 0.4901,
      "step": 3418
    },
    {
      "epoch": 0.1845016458906697,
      "grad_norm": 0.9946558535269783,
      "learning_rate": 9.264996553790557e-06,
      "loss": 0.6282,
      "step": 3419
    },
    {
      "epoch": 0.18455560951918407,
      "grad_norm": 1.0222891000111585,
      "learning_rate": 9.264578652924047e-06,
      "loss": 0.3936,
      "step": 3420
    },
    {
      "epoch": 0.18460957314769846,
      "grad_norm": 1.2257043053142351,
      "learning_rate": 9.264160643859744e-06,
      "loss": 0.5604,
      "step": 3421
    },
    {
      "epoch": 0.18466353677621283,
      "grad_norm": 1.2365215955352395,
      "learning_rate": 9.263742526609659e-06,
      "loss": 0.6103,
      "step": 3422
    },
    {
      "epoch": 0.1847175004047272,
      "grad_norm": 1.4488132129875555,
      "learning_rate": 9.263324301185813e-06,
      "loss": 0.5988,
      "step": 3423
    },
    {
      "epoch": 0.1847714640332416,
      "grad_norm": 1.1252588567062876,
      "learning_rate": 9.262905967600223e-06,
      "loss": 0.5053,
      "step": 3424
    },
    {
      "epoch": 0.18482542766175597,
      "grad_norm": 1.0981007806889382,
      "learning_rate": 9.262487525864912e-06,
      "loss": 0.5106,
      "step": 3425
    },
    {
      "epoch": 0.18487939129027037,
      "grad_norm": 1.1648618486236735,
      "learning_rate": 9.262068975991909e-06,
      "loss": 0.4392,
      "step": 3426
    },
    {
      "epoch": 0.18493335491878474,
      "grad_norm": 1.0314005729561397,
      "learning_rate": 9.261650317993243e-06,
      "loss": 0.5458,
      "step": 3427
    },
    {
      "epoch": 0.1849873185472991,
      "grad_norm": 0.8779380437821133,
      "learning_rate": 9.261231551880944e-06,
      "loss": 0.4519,
      "step": 3428
    },
    {
      "epoch": 0.1850412821758135,
      "grad_norm": 1.325374382885057,
      "learning_rate": 9.260812677667053e-06,
      "loss": 0.6434,
      "step": 3429
    },
    {
      "epoch": 0.18509524580432787,
      "grad_norm": 1.130735575226392,
      "learning_rate": 9.260393695363603e-06,
      "loss": 0.5674,
      "step": 3430
    },
    {
      "epoch": 0.18514920943284227,
      "grad_norm": 1.159351325546032,
      "learning_rate": 9.259974604982639e-06,
      "loss": 0.557,
      "step": 3431
    },
    {
      "epoch": 0.18520317306135664,
      "grad_norm": 0.9846036727499488,
      "learning_rate": 9.259555406536206e-06,
      "loss": 0.4268,
      "step": 3432
    },
    {
      "epoch": 0.18525713668987104,
      "grad_norm": 0.7502301814664805,
      "learning_rate": 9.259136100036352e-06,
      "loss": 0.3467,
      "step": 3433
    },
    {
      "epoch": 0.1853111003183854,
      "grad_norm": 0.9692974823179602,
      "learning_rate": 9.258716685495128e-06,
      "loss": 0.4271,
      "step": 3434
    },
    {
      "epoch": 0.18536506394689978,
      "grad_norm": 0.7814698153113552,
      "learning_rate": 9.25829716292459e-06,
      "loss": 0.3268,
      "step": 3435
    },
    {
      "epoch": 0.18541902757541417,
      "grad_norm": 1.3395346973117894,
      "learning_rate": 9.25787753233679e-06,
      "loss": 0.6795,
      "step": 3436
    },
    {
      "epoch": 0.18547299120392854,
      "grad_norm": 0.9789329771624914,
      "learning_rate": 9.257457793743797e-06,
      "loss": 0.5786,
      "step": 3437
    },
    {
      "epoch": 0.18552695483244294,
      "grad_norm": 1.0635472203588612,
      "learning_rate": 9.257037947157669e-06,
      "loss": 0.4254,
      "step": 3438
    },
    {
      "epoch": 0.1855809184609573,
      "grad_norm": 1.060849791958406,
      "learning_rate": 9.256617992590474e-06,
      "loss": 0.4779,
      "step": 3439
    },
    {
      "epoch": 0.1856348820894717,
      "grad_norm": 1.115484162825859,
      "learning_rate": 9.256197930054282e-06,
      "loss": 0.4942,
      "step": 3440
    },
    {
      "epoch": 0.18568884571798608,
      "grad_norm": 1.1556526716139666,
      "learning_rate": 9.255777759561165e-06,
      "loss": 0.5941,
      "step": 3441
    },
    {
      "epoch": 0.18574280934650045,
      "grad_norm": 1.1886439767462937,
      "learning_rate": 9.255357481123202e-06,
      "loss": 0.4939,
      "step": 3442
    },
    {
      "epoch": 0.18579677297501485,
      "grad_norm": 1.283475024261839,
      "learning_rate": 9.25493709475247e-06,
      "loss": 0.7041,
      "step": 3443
    },
    {
      "epoch": 0.18585073660352922,
      "grad_norm": 0.9722056209296598,
      "learning_rate": 9.254516600461051e-06,
      "loss": 0.4517,
      "step": 3444
    },
    {
      "epoch": 0.1859047002320436,
      "grad_norm": 1.1458358473171781,
      "learning_rate": 9.254095998261032e-06,
      "loss": 0.5273,
      "step": 3445
    },
    {
      "epoch": 0.18595866386055798,
      "grad_norm": 0.8864873146178459,
      "learning_rate": 9.2536752881645e-06,
      "loss": 0.4115,
      "step": 3446
    },
    {
      "epoch": 0.18601262748907235,
      "grad_norm": 1.1084416753074882,
      "learning_rate": 9.25325447018355e-06,
      "loss": 0.5521,
      "step": 3447
    },
    {
      "epoch": 0.18606659111758675,
      "grad_norm": 0.8723446745150307,
      "learning_rate": 9.252833544330271e-06,
      "loss": 0.4044,
      "step": 3448
    },
    {
      "epoch": 0.18612055474610112,
      "grad_norm": 1.00134020577514,
      "learning_rate": 9.252412510616766e-06,
      "loss": 0.4214,
      "step": 3449
    },
    {
      "epoch": 0.18617451837461552,
      "grad_norm": 1.2697332778957287,
      "learning_rate": 9.251991369055133e-06,
      "loss": 0.5776,
      "step": 3450
    },
    {
      "epoch": 0.18622848200312989,
      "grad_norm": 1.0148362788993683,
      "learning_rate": 9.251570119657477e-06,
      "loss": 0.3954,
      "step": 3451
    },
    {
      "epoch": 0.18628244563164428,
      "grad_norm": 0.8625814751685347,
      "learning_rate": 9.251148762435904e-06,
      "loss": 0.4109,
      "step": 3452
    },
    {
      "epoch": 0.18633640926015865,
      "grad_norm": 0.7929076942918725,
      "learning_rate": 9.250727297402526e-06,
      "loss": 0.3682,
      "step": 3453
    },
    {
      "epoch": 0.18639037288867302,
      "grad_norm": 1.1025806886945726,
      "learning_rate": 9.250305724569458e-06,
      "loss": 0.5482,
      "step": 3454
    },
    {
      "epoch": 0.18644433651718742,
      "grad_norm": 0.9318397616702896,
      "learning_rate": 9.24988404394881e-06,
      "loss": 0.4039,
      "step": 3455
    },
    {
      "epoch": 0.1864983001457018,
      "grad_norm": 1.3202777018261782,
      "learning_rate": 9.249462255552708e-06,
      "loss": 0.718,
      "step": 3456
    },
    {
      "epoch": 0.1865522637742162,
      "grad_norm": 1.0447643179725168,
      "learning_rate": 9.249040359393273e-06,
      "loss": 0.4465,
      "step": 3457
    },
    {
      "epoch": 0.18660622740273056,
      "grad_norm": 1.1410350321319078,
      "learning_rate": 9.248618355482629e-06,
      "loss": 0.6849,
      "step": 3458
    },
    {
      "epoch": 0.18666019103124495,
      "grad_norm": 0.9879405716650461,
      "learning_rate": 9.248196243832906e-06,
      "loss": 0.4913,
      "step": 3459
    },
    {
      "epoch": 0.18671415465975932,
      "grad_norm": 1.1763099788805969,
      "learning_rate": 9.247774024456235e-06,
      "loss": 0.4552,
      "step": 3460
    },
    {
      "epoch": 0.1867681182882737,
      "grad_norm": 0.9573549291367229,
      "learning_rate": 9.247351697364751e-06,
      "loss": 0.3575,
      "step": 3461
    },
    {
      "epoch": 0.1868220819167881,
      "grad_norm": 1.2663126514944159,
      "learning_rate": 9.246929262570594e-06,
      "loss": 0.5593,
      "step": 3462
    },
    {
      "epoch": 0.18687604554530246,
      "grad_norm": 1.0076383687905381,
      "learning_rate": 9.246506720085904e-06,
      "loss": 0.5205,
      "step": 3463
    },
    {
      "epoch": 0.18693000917381686,
      "grad_norm": 1.1711432966567077,
      "learning_rate": 9.246084069922825e-06,
      "loss": 0.5753,
      "step": 3464
    },
    {
      "epoch": 0.18698397280233123,
      "grad_norm": 1.1069830558264135,
      "learning_rate": 9.245661312093504e-06,
      "loss": 0.4602,
      "step": 3465
    },
    {
      "epoch": 0.1870379364308456,
      "grad_norm": 0.9064463942479322,
      "learning_rate": 9.245238446610093e-06,
      "loss": 0.4034,
      "step": 3466
    },
    {
      "epoch": 0.18709190005936,
      "grad_norm": 1.0574835197248882,
      "learning_rate": 9.244815473484746e-06,
      "loss": 0.4813,
      "step": 3467
    },
    {
      "epoch": 0.18714586368787436,
      "grad_norm": 0.9738614849331148,
      "learning_rate": 9.244392392729617e-06,
      "loss": 0.4542,
      "step": 3468
    },
    {
      "epoch": 0.18719982731638876,
      "grad_norm": 1.189935112463329,
      "learning_rate": 9.243969204356866e-06,
      "loss": 0.675,
      "step": 3469
    },
    {
      "epoch": 0.18725379094490313,
      "grad_norm": 1.1330631411943373,
      "learning_rate": 9.24354590837866e-06,
      "loss": 0.43,
      "step": 3470
    },
    {
      "epoch": 0.18730775457341753,
      "grad_norm": 0.9281664275659937,
      "learning_rate": 9.24312250480716e-06,
      "loss": 0.55,
      "step": 3471
    },
    {
      "epoch": 0.1873617182019319,
      "grad_norm": 1.0399880070461784,
      "learning_rate": 9.242698993654537e-06,
      "loss": 0.5427,
      "step": 3472
    },
    {
      "epoch": 0.18741568183044627,
      "grad_norm": 1.1990096032698496,
      "learning_rate": 9.242275374932964e-06,
      "loss": 0.6653,
      "step": 3473
    },
    {
      "epoch": 0.18746964545896067,
      "grad_norm": 1.0844353409882246,
      "learning_rate": 9.241851648654615e-06,
      "loss": 0.4249,
      "step": 3474
    },
    {
      "epoch": 0.18752360908747503,
      "grad_norm": 0.9047991933503821,
      "learning_rate": 9.24142781483167e-06,
      "loss": 0.4016,
      "step": 3475
    },
    {
      "epoch": 0.18757757271598943,
      "grad_norm": 1.029046545861315,
      "learning_rate": 9.241003873476308e-06,
      "loss": 0.5697,
      "step": 3476
    },
    {
      "epoch": 0.1876315363445038,
      "grad_norm": 1.0062966383079572,
      "learning_rate": 9.240579824600715e-06,
      "loss": 0.494,
      "step": 3477
    },
    {
      "epoch": 0.1876854999730182,
      "grad_norm": 1.186940306943247,
      "learning_rate": 9.240155668217077e-06,
      "loss": 0.5438,
      "step": 3478
    },
    {
      "epoch": 0.18773946360153257,
      "grad_norm": 1.2879449774601435,
      "learning_rate": 9.239731404337586e-06,
      "loss": 0.5318,
      "step": 3479
    },
    {
      "epoch": 0.18779342723004694,
      "grad_norm": 1.0613821417380622,
      "learning_rate": 9.239307032974438e-06,
      "loss": 0.5963,
      "step": 3480
    },
    {
      "epoch": 0.18784739085856134,
      "grad_norm": 1.6503328933051795,
      "learning_rate": 9.238882554139826e-06,
      "loss": 0.6752,
      "step": 3481
    },
    {
      "epoch": 0.1879013544870757,
      "grad_norm": 1.2321961575999703,
      "learning_rate": 9.238457967845952e-06,
      "loss": 0.5486,
      "step": 3482
    },
    {
      "epoch": 0.1879553181155901,
      "grad_norm": 1.2075999319593684,
      "learning_rate": 9.238033274105019e-06,
      "loss": 0.6074,
      "step": 3483
    },
    {
      "epoch": 0.18800928174410447,
      "grad_norm": 1.1141759103653237,
      "learning_rate": 9.23760847292923e-06,
      "loss": 0.5654,
      "step": 3484
    },
    {
      "epoch": 0.18806324537261884,
      "grad_norm": 1.0834347698686992,
      "learning_rate": 9.237183564330799e-06,
      "loss": 0.4681,
      "step": 3485
    },
    {
      "epoch": 0.18811720900113324,
      "grad_norm": 1.2143611832361731,
      "learning_rate": 9.236758548321936e-06,
      "loss": 0.5697,
      "step": 3486
    },
    {
      "epoch": 0.1881711726296476,
      "grad_norm": 1.0168337874864963,
      "learning_rate": 9.236333424914858e-06,
      "loss": 0.5645,
      "step": 3487
    },
    {
      "epoch": 0.188225136258162,
      "grad_norm": 1.0190346442508489,
      "learning_rate": 9.23590819412178e-06,
      "loss": 0.6447,
      "step": 3488
    },
    {
      "epoch": 0.18827909988667638,
      "grad_norm": 1.05238358321972,
      "learning_rate": 9.235482855954927e-06,
      "loss": 0.5029,
      "step": 3489
    },
    {
      "epoch": 0.18833306351519077,
      "grad_norm": 0.777758048009598,
      "learning_rate": 9.235057410426523e-06,
      "loss": 0.3958,
      "step": 3490
    },
    {
      "epoch": 0.18838702714370514,
      "grad_norm": 1.0656339621105695,
      "learning_rate": 9.234631857548793e-06,
      "loss": 0.4746,
      "step": 3491
    },
    {
      "epoch": 0.1884409907722195,
      "grad_norm": 1.1436061916244185,
      "learning_rate": 9.23420619733397e-06,
      "loss": 0.5674,
      "step": 3492
    },
    {
      "epoch": 0.1884949544007339,
      "grad_norm": 1.1530082096168304,
      "learning_rate": 9.233780429794289e-06,
      "loss": 0.5282,
      "step": 3493
    },
    {
      "epoch": 0.18854891802924828,
      "grad_norm": 1.0219645271003759,
      "learning_rate": 9.233354554941985e-06,
      "loss": 0.5054,
      "step": 3494
    },
    {
      "epoch": 0.18860288165776268,
      "grad_norm": 1.1820011605223422,
      "learning_rate": 9.232928572789298e-06,
      "loss": 0.4161,
      "step": 3495
    },
    {
      "epoch": 0.18865684528627705,
      "grad_norm": 1.007723844492105,
      "learning_rate": 9.232502483348473e-06,
      "loss": 0.4088,
      "step": 3496
    },
    {
      "epoch": 0.18871080891479142,
      "grad_norm": 1.359710262049616,
      "learning_rate": 9.232076286631755e-06,
      "loss": 0.5925,
      "step": 3497
    },
    {
      "epoch": 0.18876477254330581,
      "grad_norm": 0.967431021953732,
      "learning_rate": 9.231649982651395e-06,
      "loss": 0.4017,
      "step": 3498
    },
    {
      "epoch": 0.18881873617182018,
      "grad_norm": 0.9829602598227183,
      "learning_rate": 9.231223571419641e-06,
      "loss": 0.4106,
      "step": 3499
    },
    {
      "epoch": 0.18887269980033458,
      "grad_norm": 1.099857209487639,
      "learning_rate": 9.230797052948753e-06,
      "loss": 0.5176,
      "step": 3500
    },
    {
      "epoch": 0.18887269980033458,
      "eval_loss": 0.5863323211669922,
      "eval_runtime": 160.241,
      "eval_samples_per_second": 21.461,
      "eval_steps_per_second": 0.899,
      "step": 3500
    },
    {
      "epoch": 0.18892666342884895,
      "grad_norm": 0.9413599538002425,
      "learning_rate": 9.23037042725099e-06,
      "loss": 0.425,
      "step": 3501
    },
    {
      "epoch": 0.18898062705736335,
      "grad_norm": 1.0803720114828943,
      "learning_rate": 9.22994369433861e-06,
      "loss": 0.4157,
      "step": 3502
    },
    {
      "epoch": 0.18903459068587772,
      "grad_norm": 0.978064266852157,
      "learning_rate": 9.229516854223878e-06,
      "loss": 0.5308,
      "step": 3503
    },
    {
      "epoch": 0.1890885543143921,
      "grad_norm": 1.236389627655464,
      "learning_rate": 9.229089906919064e-06,
      "loss": 0.6252,
      "step": 3504
    },
    {
      "epoch": 0.18914251794290649,
      "grad_norm": 1.0322789006066662,
      "learning_rate": 9.228662852436438e-06,
      "loss": 0.5367,
      "step": 3505
    },
    {
      "epoch": 0.18919648157142085,
      "grad_norm": 1.1344734389597244,
      "learning_rate": 9.228235690788274e-06,
      "loss": 0.4481,
      "step": 3506
    },
    {
      "epoch": 0.18925044519993525,
      "grad_norm": 0.9318181110759796,
      "learning_rate": 9.227808421986851e-06,
      "loss": 0.4044,
      "step": 3507
    },
    {
      "epoch": 0.18930440882844962,
      "grad_norm": 0.939628546952394,
      "learning_rate": 9.227381046044444e-06,
      "loss": 0.4319,
      "step": 3508
    },
    {
      "epoch": 0.18935837245696402,
      "grad_norm": 0.8636356009954431,
      "learning_rate": 9.226953562973343e-06,
      "loss": 0.4298,
      "step": 3509
    },
    {
      "epoch": 0.1894123360854784,
      "grad_norm": 1.3868254043795416,
      "learning_rate": 9.226525972785827e-06,
      "loss": 0.5882,
      "step": 3510
    },
    {
      "epoch": 0.18946629971399276,
      "grad_norm": 1.1375186306338791,
      "learning_rate": 9.226098275494192e-06,
      "loss": 0.3922,
      "step": 3511
    },
    {
      "epoch": 0.18952026334250716,
      "grad_norm": 1.0638640024274246,
      "learning_rate": 9.225670471110728e-06,
      "loss": 0.6124,
      "step": 3512
    },
    {
      "epoch": 0.18957422697102153,
      "grad_norm": 1.1454161836759171,
      "learning_rate": 9.225242559647729e-06,
      "loss": 0.5908,
      "step": 3513
    },
    {
      "epoch": 0.18962819059953592,
      "grad_norm": 0.9042197694399127,
      "learning_rate": 9.224814541117493e-06,
      "loss": 0.4233,
      "step": 3514
    },
    {
      "epoch": 0.1896821542280503,
      "grad_norm": 1.0858707555162825,
      "learning_rate": 9.224386415532327e-06,
      "loss": 0.4924,
      "step": 3515
    },
    {
      "epoch": 0.18973611785656466,
      "grad_norm": 0.9007210257822292,
      "learning_rate": 9.22395818290453e-06,
      "loss": 0.4306,
      "step": 3516
    },
    {
      "epoch": 0.18979008148507906,
      "grad_norm": 1.2072767463676732,
      "learning_rate": 9.223529843246412e-06,
      "loss": 0.5953,
      "step": 3517
    },
    {
      "epoch": 0.18984404511359343,
      "grad_norm": 1.18181194335848,
      "learning_rate": 9.223101396570285e-06,
      "loss": 0.6356,
      "step": 3518
    },
    {
      "epoch": 0.18989800874210783,
      "grad_norm": 0.8823631791466998,
      "learning_rate": 9.22267284288846e-06,
      "loss": 0.4222,
      "step": 3519
    },
    {
      "epoch": 0.1899519723706222,
      "grad_norm": 1.1492767388930047,
      "learning_rate": 9.222244182213258e-06,
      "loss": 0.5605,
      "step": 3520
    },
    {
      "epoch": 0.1900059359991366,
      "grad_norm": 0.915733559599143,
      "learning_rate": 9.221815414556998e-06,
      "loss": 0.4578,
      "step": 3521
    },
    {
      "epoch": 0.19005989962765096,
      "grad_norm": 0.8558759151395682,
      "learning_rate": 9.221386539932e-06,
      "loss": 0.3731,
      "step": 3522
    },
    {
      "epoch": 0.19011386325616533,
      "grad_norm": 1.2633267791968665,
      "learning_rate": 9.220957558350594e-06,
      "loss": 0.6015,
      "step": 3523
    },
    {
      "epoch": 0.19016782688467973,
      "grad_norm": 0.9069104546218806,
      "learning_rate": 9.220528469825107e-06,
      "loss": 0.3843,
      "step": 3524
    },
    {
      "epoch": 0.1902217905131941,
      "grad_norm": 1.100535321502579,
      "learning_rate": 9.220099274367874e-06,
      "loss": 0.6306,
      "step": 3525
    },
    {
      "epoch": 0.1902757541417085,
      "grad_norm": 1.0841256832675252,
      "learning_rate": 9.219669971991228e-06,
      "loss": 0.6069,
      "step": 3526
    },
    {
      "epoch": 0.19032971777022287,
      "grad_norm": 0.8915221394550601,
      "learning_rate": 9.219240562707508e-06,
      "loss": 0.4238,
      "step": 3527
    },
    {
      "epoch": 0.19038368139873726,
      "grad_norm": 1.0889425069056702,
      "learning_rate": 9.218811046529055e-06,
      "loss": 0.5184,
      "step": 3528
    },
    {
      "epoch": 0.19043764502725163,
      "grad_norm": 0.9314305189264818,
      "learning_rate": 9.218381423468218e-06,
      "loss": 0.4617,
      "step": 3529
    },
    {
      "epoch": 0.190491608655766,
      "grad_norm": 1.0642877726546343,
      "learning_rate": 9.217951693537341e-06,
      "loss": 0.442,
      "step": 3530
    },
    {
      "epoch": 0.1905455722842804,
      "grad_norm": 1.1320889404173977,
      "learning_rate": 9.217521856748776e-06,
      "loss": 0.4866,
      "step": 3531
    },
    {
      "epoch": 0.19059953591279477,
      "grad_norm": 1.3647397692872159,
      "learning_rate": 9.217091913114874e-06,
      "loss": 0.6245,
      "step": 3532
    },
    {
      "epoch": 0.19065349954130917,
      "grad_norm": 1.117233877126864,
      "learning_rate": 9.216661862647997e-06,
      "loss": 0.5378,
      "step": 3533
    },
    {
      "epoch": 0.19070746316982354,
      "grad_norm": 1.2285993461638316,
      "learning_rate": 9.216231705360502e-06,
      "loss": 0.5282,
      "step": 3534
    },
    {
      "epoch": 0.1907614267983379,
      "grad_norm": 1.1024656763318308,
      "learning_rate": 9.215801441264752e-06,
      "loss": 0.6157,
      "step": 3535
    },
    {
      "epoch": 0.1908153904268523,
      "grad_norm": 1.0410446994674691,
      "learning_rate": 9.215371070373114e-06,
      "loss": 0.5525,
      "step": 3536
    },
    {
      "epoch": 0.19086935405536667,
      "grad_norm": 0.818203398417902,
      "learning_rate": 9.214940592697959e-06,
      "loss": 0.3322,
      "step": 3537
    },
    {
      "epoch": 0.19092331768388107,
      "grad_norm": 1.2003426936065618,
      "learning_rate": 9.214510008251656e-06,
      "loss": 0.4921,
      "step": 3538
    },
    {
      "epoch": 0.19097728131239544,
      "grad_norm": 0.9449314471918646,
      "learning_rate": 9.214079317046583e-06,
      "loss": 0.5402,
      "step": 3539
    },
    {
      "epoch": 0.19103124494090984,
      "grad_norm": 1.0881166789556225,
      "learning_rate": 9.213648519095118e-06,
      "loss": 0.5106,
      "step": 3540
    },
    {
      "epoch": 0.1910852085694242,
      "grad_norm": 1.258125117877621,
      "learning_rate": 9.213217614409641e-06,
      "loss": 0.4694,
      "step": 3541
    },
    {
      "epoch": 0.19113917219793858,
      "grad_norm": 1.1953205087009697,
      "learning_rate": 9.21278660300254e-06,
      "loss": 0.6117,
      "step": 3542
    },
    {
      "epoch": 0.19119313582645298,
      "grad_norm": 1.0545429903101766,
      "learning_rate": 9.212355484886198e-06,
      "loss": 0.57,
      "step": 3543
    },
    {
      "epoch": 0.19124709945496735,
      "grad_norm": 1.2301276625917152,
      "learning_rate": 9.21192426007301e-06,
      "loss": 0.563,
      "step": 3544
    },
    {
      "epoch": 0.19130106308348174,
      "grad_norm": 1.2361063244687553,
      "learning_rate": 9.211492928575367e-06,
      "loss": 0.4705,
      "step": 3545
    },
    {
      "epoch": 0.1913550267119961,
      "grad_norm": 1.157708055320251,
      "learning_rate": 9.211061490405667e-06,
      "loss": 0.5523,
      "step": 3546
    },
    {
      "epoch": 0.1914089903405105,
      "grad_norm": 1.111581194563244,
      "learning_rate": 9.21062994557631e-06,
      "loss": 0.4602,
      "step": 3547
    },
    {
      "epoch": 0.19146295396902488,
      "grad_norm": 1.2308133479713206,
      "learning_rate": 9.2101982940997e-06,
      "loss": 0.6038,
      "step": 3548
    },
    {
      "epoch": 0.19151691759753925,
      "grad_norm": 1.4173235606196728,
      "learning_rate": 9.209766535988242e-06,
      "loss": 0.6588,
      "step": 3549
    },
    {
      "epoch": 0.19157088122605365,
      "grad_norm": 1.1828597239765997,
      "learning_rate": 9.209334671254346e-06,
      "loss": 0.5733,
      "step": 3550
    },
    {
      "epoch": 0.19162484485456802,
      "grad_norm": 1.019645518792412,
      "learning_rate": 9.208902699910422e-06,
      "loss": 0.524,
      "step": 3551
    },
    {
      "epoch": 0.1916788084830824,
      "grad_norm": 1.4321370889472942,
      "learning_rate": 9.208470621968888e-06,
      "loss": 0.5143,
      "step": 3552
    },
    {
      "epoch": 0.19173277211159678,
      "grad_norm": 0.9771196529761064,
      "learning_rate": 9.20803843744216e-06,
      "loss": 0.4762,
      "step": 3553
    },
    {
      "epoch": 0.19178673574011115,
      "grad_norm": 0.8453750798422196,
      "learning_rate": 9.207606146342661e-06,
      "loss": 0.3185,
      "step": 3554
    },
    {
      "epoch": 0.19184069936862555,
      "grad_norm": 1.1337413271824397,
      "learning_rate": 9.207173748682815e-06,
      "loss": 0.5927,
      "step": 3555
    },
    {
      "epoch": 0.19189466299713992,
      "grad_norm": 1.1851828234013277,
      "learning_rate": 9.20674124447505e-06,
      "loss": 0.5292,
      "step": 3556
    },
    {
      "epoch": 0.19194862662565432,
      "grad_norm": 1.130782551043283,
      "learning_rate": 9.206308633731796e-06,
      "loss": 0.5354,
      "step": 3557
    },
    {
      "epoch": 0.1920025902541687,
      "grad_norm": 1.0285656058686468,
      "learning_rate": 9.205875916465487e-06,
      "loss": 0.498,
      "step": 3558
    },
    {
      "epoch": 0.19205655388268308,
      "grad_norm": 0.8044955352133234,
      "learning_rate": 9.20544309268856e-06,
      "loss": 0.3501,
      "step": 3559
    },
    {
      "epoch": 0.19211051751119745,
      "grad_norm": 1.130479066625325,
      "learning_rate": 9.205010162413453e-06,
      "loss": 0.604,
      "step": 3560
    },
    {
      "epoch": 0.19216448113971182,
      "grad_norm": 1.183288302034953,
      "learning_rate": 9.204577125652613e-06,
      "loss": 0.5448,
      "step": 3561
    },
    {
      "epoch": 0.19221844476822622,
      "grad_norm": 1.1491433880773687,
      "learning_rate": 9.204143982418482e-06,
      "loss": 0.4885,
      "step": 3562
    },
    {
      "epoch": 0.1922724083967406,
      "grad_norm": 1.05871317408975,
      "learning_rate": 9.20371073272351e-06,
      "loss": 0.5711,
      "step": 3563
    },
    {
      "epoch": 0.192326372025255,
      "grad_norm": 0.9612177767402478,
      "learning_rate": 9.203277376580148e-06,
      "loss": 0.4386,
      "step": 3564
    },
    {
      "epoch": 0.19238033565376936,
      "grad_norm": 1.1435755817938,
      "learning_rate": 9.202843914000855e-06,
      "loss": 0.5742,
      "step": 3565
    },
    {
      "epoch": 0.19243429928228373,
      "grad_norm": 1.086469307128442,
      "learning_rate": 9.202410344998084e-06,
      "loss": 0.5575,
      "step": 3566
    },
    {
      "epoch": 0.19248826291079812,
      "grad_norm": 1.1501097092990864,
      "learning_rate": 9.201976669584299e-06,
      "loss": 0.4186,
      "step": 3567
    },
    {
      "epoch": 0.1925422265393125,
      "grad_norm": 0.9696458262977752,
      "learning_rate": 9.201542887771964e-06,
      "loss": 0.4917,
      "step": 3568
    },
    {
      "epoch": 0.1925961901678269,
      "grad_norm": 0.9993446749816595,
      "learning_rate": 9.201108999573546e-06,
      "loss": 0.536,
      "step": 3569
    },
    {
      "epoch": 0.19265015379634126,
      "grad_norm": 1.025394678729257,
      "learning_rate": 9.200675005001518e-06,
      "loss": 0.496,
      "step": 3570
    },
    {
      "epoch": 0.19270411742485566,
      "grad_norm": 1.2479100996140824,
      "learning_rate": 9.20024090406835e-06,
      "loss": 0.8365,
      "step": 3571
    },
    {
      "epoch": 0.19275808105337003,
      "grad_norm": 1.0274313489907139,
      "learning_rate": 9.199806696786517e-06,
      "loss": 0.5111,
      "step": 3572
    },
    {
      "epoch": 0.1928120446818844,
      "grad_norm": 1.0007183310603083,
      "learning_rate": 9.199372383168505e-06,
      "loss": 0.4077,
      "step": 3573
    },
    {
      "epoch": 0.1928660083103988,
      "grad_norm": 0.9281619648660502,
      "learning_rate": 9.198937963226791e-06,
      "loss": 0.3626,
      "step": 3574
    },
    {
      "epoch": 0.19291997193891317,
      "grad_norm": 1.141218285658685,
      "learning_rate": 9.198503436973862e-06,
      "loss": 0.5597,
      "step": 3575
    },
    {
      "epoch": 0.19297393556742756,
      "grad_norm": 1.1108387673604245,
      "learning_rate": 9.198068804422206e-06,
      "loss": 0.524,
      "step": 3576
    },
    {
      "epoch": 0.19302789919594193,
      "grad_norm": 1.0749912164667301,
      "learning_rate": 9.197634065584319e-06,
      "loss": 0.3801,
      "step": 3577
    },
    {
      "epoch": 0.19308186282445633,
      "grad_norm": 1.0637333940611702,
      "learning_rate": 9.197199220472691e-06,
      "loss": 0.455,
      "step": 3578
    },
    {
      "epoch": 0.1931358264529707,
      "grad_norm": 1.2942680509329512,
      "learning_rate": 9.19676426909982e-06,
      "loss": 0.5516,
      "step": 3579
    },
    {
      "epoch": 0.19318979008148507,
      "grad_norm": 1.1753712397057503,
      "learning_rate": 9.196329211478212e-06,
      "loss": 0.5056,
      "step": 3580
    },
    {
      "epoch": 0.19324375370999947,
      "grad_norm": 1.0742807121379698,
      "learning_rate": 9.195894047620367e-06,
      "loss": 0.6193,
      "step": 3581
    },
    {
      "epoch": 0.19329771733851384,
      "grad_norm": 1.3563544006951858,
      "learning_rate": 9.195458777538792e-06,
      "loss": 0.5342,
      "step": 3582
    },
    {
      "epoch": 0.19335168096702823,
      "grad_norm": 1.0839770191043454,
      "learning_rate": 9.195023401245998e-06,
      "loss": 0.4762,
      "step": 3583
    },
    {
      "epoch": 0.1934056445955426,
      "grad_norm": 1.0275590327005384,
      "learning_rate": 9.194587918754496e-06,
      "loss": 0.439,
      "step": 3584
    },
    {
      "epoch": 0.19345960822405697,
      "grad_norm": 0.9458721337676708,
      "learning_rate": 9.194152330076806e-06,
      "loss": 0.4961,
      "step": 3585
    },
    {
      "epoch": 0.19351357185257137,
      "grad_norm": 0.9496046857940559,
      "learning_rate": 9.193716635225445e-06,
      "loss": 0.4236,
      "step": 3586
    },
    {
      "epoch": 0.19356753548108574,
      "grad_norm": 0.9427693237687016,
      "learning_rate": 9.193280834212935e-06,
      "loss": 0.4386,
      "step": 3587
    },
    {
      "epoch": 0.19362149910960014,
      "grad_norm": 1.0891050792957426,
      "learning_rate": 9.192844927051802e-06,
      "loss": 0.5124,
      "step": 3588
    },
    {
      "epoch": 0.1936754627381145,
      "grad_norm": 0.9779880912946277,
      "learning_rate": 9.192408913754575e-06,
      "loss": 0.4483,
      "step": 3589
    },
    {
      "epoch": 0.1937294263666289,
      "grad_norm": 1.1842951511195103,
      "learning_rate": 9.191972794333786e-06,
      "loss": 0.5986,
      "step": 3590
    },
    {
      "epoch": 0.19378338999514327,
      "grad_norm": 1.060208841482625,
      "learning_rate": 9.191536568801967e-06,
      "loss": 0.4817,
      "step": 3591
    },
    {
      "epoch": 0.19383735362365764,
      "grad_norm": 1.0208953406081516,
      "learning_rate": 9.191100237171657e-06,
      "loss": 0.4607,
      "step": 3592
    },
    {
      "epoch": 0.19389131725217204,
      "grad_norm": 1.0014919263055355,
      "learning_rate": 9.190663799455397e-06,
      "loss": 0.4642,
      "step": 3593
    },
    {
      "epoch": 0.1939452808806864,
      "grad_norm": 1.0442691432330509,
      "learning_rate": 9.19022725566573e-06,
      "loss": 0.4865,
      "step": 3594
    },
    {
      "epoch": 0.1939992445092008,
      "grad_norm": 1.018641699783968,
      "learning_rate": 9.189790605815203e-06,
      "loss": 0.3655,
      "step": 3595
    },
    {
      "epoch": 0.19405320813771518,
      "grad_norm": 1.1244596123138388,
      "learning_rate": 9.189353849916366e-06,
      "loss": 0.5873,
      "step": 3596
    },
    {
      "epoch": 0.19410717176622957,
      "grad_norm": 1.1316079909347732,
      "learning_rate": 9.188916987981772e-06,
      "loss": 0.6233,
      "step": 3597
    },
    {
      "epoch": 0.19416113539474394,
      "grad_norm": 0.850623357354391,
      "learning_rate": 9.188480020023976e-06,
      "loss": 0.3491,
      "step": 3598
    },
    {
      "epoch": 0.19421509902325831,
      "grad_norm": 1.1388085581199436,
      "learning_rate": 9.188042946055536e-06,
      "loss": 0.5394,
      "step": 3599
    },
    {
      "epoch": 0.1942690626517727,
      "grad_norm": 1.328353140314693,
      "learning_rate": 9.187605766089017e-06,
      "loss": 0.673,
      "step": 3600
    },
    {
      "epoch": 0.19432302628028708,
      "grad_norm": 0.9615930435843016,
      "learning_rate": 9.187168480136982e-06,
      "loss": 0.3963,
      "step": 3601
    },
    {
      "epoch": 0.19437698990880148,
      "grad_norm": 0.9483627916873354,
      "learning_rate": 9.186731088212e-06,
      "loss": 0.4975,
      "step": 3602
    },
    {
      "epoch": 0.19443095353731585,
      "grad_norm": 0.8542570093009577,
      "learning_rate": 9.186293590326639e-06,
      "loss": 0.3762,
      "step": 3603
    },
    {
      "epoch": 0.19448491716583022,
      "grad_norm": 0.9211920276186257,
      "learning_rate": 9.185855986493478e-06,
      "loss": 0.4036,
      "step": 3604
    },
    {
      "epoch": 0.19453888079434462,
      "grad_norm": 1.1779246821180303,
      "learning_rate": 9.185418276725089e-06,
      "loss": 0.661,
      "step": 3605
    },
    {
      "epoch": 0.19459284442285898,
      "grad_norm": 0.9507352115524731,
      "learning_rate": 9.184980461034056e-06,
      "loss": 0.352,
      "step": 3606
    },
    {
      "epoch": 0.19464680805137338,
      "grad_norm": 1.4180710691701253,
      "learning_rate": 9.184542539432963e-06,
      "loss": 0.6853,
      "step": 3607
    },
    {
      "epoch": 0.19470077167988775,
      "grad_norm": 1.025037776645239,
      "learning_rate": 9.184104511934392e-06,
      "loss": 0.5463,
      "step": 3608
    },
    {
      "epoch": 0.19475473530840215,
      "grad_norm": 0.8767518913857991,
      "learning_rate": 9.183666378550935e-06,
      "loss": 0.3942,
      "step": 3609
    },
    {
      "epoch": 0.19480869893691652,
      "grad_norm": 1.1743405552100048,
      "learning_rate": 9.183228139295185e-06,
      "loss": 0.4834,
      "step": 3610
    },
    {
      "epoch": 0.1948626625654309,
      "grad_norm": 1.198131092729544,
      "learning_rate": 9.182789794179737e-06,
      "loss": 0.619,
      "step": 3611
    },
    {
      "epoch": 0.19491662619394529,
      "grad_norm": 1.338304467082019,
      "learning_rate": 9.18235134321719e-06,
      "loss": 0.6273,
      "step": 3612
    },
    {
      "epoch": 0.19497058982245966,
      "grad_norm": 1.1673223970546338,
      "learning_rate": 9.181912786420142e-06,
      "loss": 0.5269,
      "step": 3613
    },
    {
      "epoch": 0.19502455345097405,
      "grad_norm": 0.7890082003084596,
      "learning_rate": 9.181474123801201e-06,
      "loss": 0.4055,
      "step": 3614
    },
    {
      "epoch": 0.19507851707948842,
      "grad_norm": 1.132574254869667,
      "learning_rate": 9.181035355372974e-06,
      "loss": 0.5497,
      "step": 3615
    },
    {
      "epoch": 0.1951324807080028,
      "grad_norm": 0.9038162642220137,
      "learning_rate": 9.180596481148072e-06,
      "loss": 0.4027,
      "step": 3616
    },
    {
      "epoch": 0.1951864443365172,
      "grad_norm": 0.8315942463006982,
      "learning_rate": 9.180157501139108e-06,
      "loss": 0.416,
      "step": 3617
    },
    {
      "epoch": 0.19524040796503156,
      "grad_norm": 1.2392969593251206,
      "learning_rate": 9.179718415358698e-06,
      "loss": 0.5767,
      "step": 3618
    },
    {
      "epoch": 0.19529437159354596,
      "grad_norm": 1.0479830495919433,
      "learning_rate": 9.179279223819464e-06,
      "loss": 0.5535,
      "step": 3619
    },
    {
      "epoch": 0.19534833522206033,
      "grad_norm": 0.9311288840124303,
      "learning_rate": 9.178839926534027e-06,
      "loss": 0.4557,
      "step": 3620
    },
    {
      "epoch": 0.19540229885057472,
      "grad_norm": 1.0557026472005977,
      "learning_rate": 9.178400523515013e-06,
      "loss": 0.5382,
      "step": 3621
    },
    {
      "epoch": 0.1954562624790891,
      "grad_norm": 1.1854025420247465,
      "learning_rate": 9.177961014775052e-06,
      "loss": 0.6454,
      "step": 3622
    },
    {
      "epoch": 0.19551022610760346,
      "grad_norm": 0.8185365233147655,
      "learning_rate": 9.177521400326774e-06,
      "loss": 0.4338,
      "step": 3623
    },
    {
      "epoch": 0.19556418973611786,
      "grad_norm": 1.1220356407791416,
      "learning_rate": 9.177081680182817e-06,
      "loss": 0.5525,
      "step": 3624
    },
    {
      "epoch": 0.19561815336463223,
      "grad_norm": 0.9482150235546478,
      "learning_rate": 9.176641854355815e-06,
      "loss": 0.4341,
      "step": 3625
    },
    {
      "epoch": 0.19567211699314663,
      "grad_norm": 0.7004963440572141,
      "learning_rate": 9.176201922858411e-06,
      "loss": 0.2866,
      "step": 3626
    },
    {
      "epoch": 0.195726080621661,
      "grad_norm": 1.0806331857400004,
      "learning_rate": 9.175761885703253e-06,
      "loss": 0.4915,
      "step": 3627
    },
    {
      "epoch": 0.1957800442501754,
      "grad_norm": 1.1965538613778017,
      "learning_rate": 9.17532174290298e-06,
      "loss": 0.5701,
      "step": 3628
    },
    {
      "epoch": 0.19583400787868976,
      "grad_norm": 1.0450893480707388,
      "learning_rate": 9.174881494470247e-06,
      "loss": 0.4841,
      "step": 3629
    },
    {
      "epoch": 0.19588797150720413,
      "grad_norm": 0.9204543684344223,
      "learning_rate": 9.174441140417709e-06,
      "loss": 0.4216,
      "step": 3630
    },
    {
      "epoch": 0.19594193513571853,
      "grad_norm": 0.8022516175723758,
      "learning_rate": 9.174000680758019e-06,
      "loss": 0.399,
      "step": 3631
    },
    {
      "epoch": 0.1959958987642329,
      "grad_norm": 1.19860940407008,
      "learning_rate": 9.173560115503836e-06,
      "loss": 0.5936,
      "step": 3632
    },
    {
      "epoch": 0.1960498623927473,
      "grad_norm": 0.8835650103042798,
      "learning_rate": 9.173119444667822e-06,
      "loss": 0.3686,
      "step": 3633
    },
    {
      "epoch": 0.19610382602126167,
      "grad_norm": 1.0362649541666233,
      "learning_rate": 9.172678668262648e-06,
      "loss": 0.4747,
      "step": 3634
    },
    {
      "epoch": 0.19615778964977604,
      "grad_norm": 1.1694835300634574,
      "learning_rate": 9.172237786300976e-06,
      "loss": 0.6331,
      "step": 3635
    },
    {
      "epoch": 0.19621175327829043,
      "grad_norm": 1.134872845981011,
      "learning_rate": 9.17179679879548e-06,
      "loss": 0.4929,
      "step": 3636
    },
    {
      "epoch": 0.1962657169068048,
      "grad_norm": 1.0970020471593487,
      "learning_rate": 9.171355705758833e-06,
      "loss": 0.5916,
      "step": 3637
    },
    {
      "epoch": 0.1963196805353192,
      "grad_norm": 1.1800775629307798,
      "learning_rate": 9.170914507203714e-06,
      "loss": 0.591,
      "step": 3638
    },
    {
      "epoch": 0.19637364416383357,
      "grad_norm": 1.1359360906745266,
      "learning_rate": 9.170473203142801e-06,
      "loss": 0.5648,
      "step": 3639
    },
    {
      "epoch": 0.19642760779234797,
      "grad_norm": 0.9354847798201783,
      "learning_rate": 9.17003179358878e-06,
      "loss": 0.4316,
      "step": 3640
    },
    {
      "epoch": 0.19648157142086234,
      "grad_norm": 0.9907227227255224,
      "learning_rate": 9.16959027855434e-06,
      "loss": 0.4445,
      "step": 3641
    },
    {
      "epoch": 0.1965355350493767,
      "grad_norm": 1.0098236455436445,
      "learning_rate": 9.169148658052165e-06,
      "loss": 0.5278,
      "step": 3642
    },
    {
      "epoch": 0.1965894986778911,
      "grad_norm": 1.0996804161678086,
      "learning_rate": 9.16870693209495e-06,
      "loss": 0.6128,
      "step": 3643
    },
    {
      "epoch": 0.19664346230640548,
      "grad_norm": 1.1627519776013067,
      "learning_rate": 9.168265100695391e-06,
      "loss": 0.5903,
      "step": 3644
    },
    {
      "epoch": 0.19669742593491987,
      "grad_norm": 1.1378521650086106,
      "learning_rate": 9.167823163866188e-06,
      "loss": 0.6362,
      "step": 3645
    },
    {
      "epoch": 0.19675138956343424,
      "grad_norm": 1.1276788386610797,
      "learning_rate": 9.16738112162004e-06,
      "loss": 0.5274,
      "step": 3646
    },
    {
      "epoch": 0.19680535319194864,
      "grad_norm": 0.9694403167256228,
      "learning_rate": 9.166938973969653e-06,
      "loss": 0.4906,
      "step": 3647
    },
    {
      "epoch": 0.196859316820463,
      "grad_norm": 1.154591828039086,
      "learning_rate": 9.166496720927734e-06,
      "loss": 0.5508,
      "step": 3648
    },
    {
      "epoch": 0.19691328044897738,
      "grad_norm": 1.125946973321343,
      "learning_rate": 9.166054362506995e-06,
      "loss": 0.4575,
      "step": 3649
    },
    {
      "epoch": 0.19696724407749178,
      "grad_norm": 1.0707590538363734,
      "learning_rate": 9.16561189872015e-06,
      "loss": 0.4795,
      "step": 3650
    },
    {
      "epoch": 0.19702120770600615,
      "grad_norm": 1.049676985488393,
      "learning_rate": 9.165169329579917e-06,
      "loss": 0.5877,
      "step": 3651
    },
    {
      "epoch": 0.19707517133452054,
      "grad_norm": 1.0865584386024731,
      "learning_rate": 9.16472665509901e-06,
      "loss": 0.5218,
      "step": 3652
    },
    {
      "epoch": 0.1971291349630349,
      "grad_norm": 0.9581409334255403,
      "learning_rate": 9.16428387529016e-06,
      "loss": 0.4987,
      "step": 3653
    },
    {
      "epoch": 0.19718309859154928,
      "grad_norm": 0.9707156414098648,
      "learning_rate": 9.163840990166085e-06,
      "loss": 0.4199,
      "step": 3654
    },
    {
      "epoch": 0.19723706222006368,
      "grad_norm": 1.1806341562036526,
      "learning_rate": 9.16339799973952e-06,
      "loss": 0.64,
      "step": 3655
    },
    {
      "epoch": 0.19729102584857805,
      "grad_norm": 1.026361694002647,
      "learning_rate": 9.162954904023195e-06,
      "loss": 0.5245,
      "step": 3656
    },
    {
      "epoch": 0.19734498947709245,
      "grad_norm": 1.1930545809305801,
      "learning_rate": 9.162511703029842e-06,
      "loss": 0.5413,
      "step": 3657
    },
    {
      "epoch": 0.19739895310560682,
      "grad_norm": 0.9907684552810251,
      "learning_rate": 9.162068396772207e-06,
      "loss": 0.389,
      "step": 3658
    },
    {
      "epoch": 0.19745291673412121,
      "grad_norm": 1.319194534790042,
      "learning_rate": 9.161624985263023e-06,
      "loss": 0.5117,
      "step": 3659
    },
    {
      "epoch": 0.19750688036263558,
      "grad_norm": 1.0702382399427064,
      "learning_rate": 9.161181468515037e-06,
      "loss": 0.5431,
      "step": 3660
    },
    {
      "epoch": 0.19756084399114995,
      "grad_norm": 1.0315151508177454,
      "learning_rate": 9.160737846540997e-06,
      "loss": 0.4605,
      "step": 3661
    },
    {
      "epoch": 0.19761480761966435,
      "grad_norm": 1.1267962159886362,
      "learning_rate": 9.160294119353652e-06,
      "loss": 0.5176,
      "step": 3662
    },
    {
      "epoch": 0.19766877124817872,
      "grad_norm": 0.8535401806708524,
      "learning_rate": 9.159850286965757e-06,
      "loss": 0.4492,
      "step": 3663
    },
    {
      "epoch": 0.19772273487669312,
      "grad_norm": 0.8655100307829648,
      "learning_rate": 9.159406349390065e-06,
      "loss": 0.4324,
      "step": 3664
    },
    {
      "epoch": 0.1977766985052075,
      "grad_norm": 0.8172992160695234,
      "learning_rate": 9.158962306639338e-06,
      "loss": 0.4176,
      "step": 3665
    },
    {
      "epoch": 0.19783066213372189,
      "grad_norm": 0.993936509723193,
      "learning_rate": 9.158518158726335e-06,
      "loss": 0.5083,
      "step": 3666
    },
    {
      "epoch": 0.19788462576223625,
      "grad_norm": 1.2083528942114232,
      "learning_rate": 9.158073905663827e-06,
      "loss": 0.6151,
      "step": 3667
    },
    {
      "epoch": 0.19793858939075062,
      "grad_norm": 1.130946843712377,
      "learning_rate": 9.157629547464576e-06,
      "loss": 0.5618,
      "step": 3668
    },
    {
      "epoch": 0.19799255301926502,
      "grad_norm": 0.9998668629923408,
      "learning_rate": 9.157185084141356e-06,
      "loss": 0.4444,
      "step": 3669
    },
    {
      "epoch": 0.1980465166477794,
      "grad_norm": 1.0889537463800498,
      "learning_rate": 9.156740515706943e-06,
      "loss": 0.4777,
      "step": 3670
    },
    {
      "epoch": 0.1981004802762938,
      "grad_norm": 1.1091655375607095,
      "learning_rate": 9.156295842174113e-06,
      "loss": 0.5302,
      "step": 3671
    },
    {
      "epoch": 0.19815444390480816,
      "grad_norm": 1.1412984108149076,
      "learning_rate": 9.155851063555644e-06,
      "loss": 0.496,
      "step": 3672
    },
    {
      "epoch": 0.19820840753332253,
      "grad_norm": 0.9304990469459504,
      "learning_rate": 9.155406179864323e-06,
      "loss": 0.4858,
      "step": 3673
    },
    {
      "epoch": 0.19826237116183693,
      "grad_norm": 1.214694080711274,
      "learning_rate": 9.154961191112935e-06,
      "loss": 0.6366,
      "step": 3674
    },
    {
      "epoch": 0.1983163347903513,
      "grad_norm": 0.8526081098641014,
      "learning_rate": 9.15451609731427e-06,
      "loss": 0.3443,
      "step": 3675
    },
    {
      "epoch": 0.1983702984188657,
      "grad_norm": 1.138549151178086,
      "learning_rate": 9.154070898481116e-06,
      "loss": 0.5303,
      "step": 3676
    },
    {
      "epoch": 0.19842426204738006,
      "grad_norm": 1.1186069512379517,
      "learning_rate": 9.153625594626275e-06,
      "loss": 0.6459,
      "step": 3677
    },
    {
      "epoch": 0.19847822567589446,
      "grad_norm": 1.0135586168488384,
      "learning_rate": 9.153180185762543e-06,
      "loss": 0.4083,
      "step": 3678
    },
    {
      "epoch": 0.19853218930440883,
      "grad_norm": 1.0903371526228016,
      "learning_rate": 9.15273467190272e-06,
      "loss": 0.4743,
      "step": 3679
    },
    {
      "epoch": 0.1985861529329232,
      "grad_norm": 1.3198972853086282,
      "learning_rate": 9.152289053059613e-06,
      "loss": 0.487,
      "step": 3680
    },
    {
      "epoch": 0.1986401165614376,
      "grad_norm": 0.9482080660441373,
      "learning_rate": 9.151843329246027e-06,
      "loss": 0.5716,
      "step": 3681
    },
    {
      "epoch": 0.19869408018995197,
      "grad_norm": 0.9466516083637744,
      "learning_rate": 9.151397500474774e-06,
      "loss": 0.4856,
      "step": 3682
    },
    {
      "epoch": 0.19874804381846636,
      "grad_norm": 1.067625054948517,
      "learning_rate": 9.15095156675867e-06,
      "loss": 0.4292,
      "step": 3683
    },
    {
      "epoch": 0.19880200744698073,
      "grad_norm": 1.0892470711529207,
      "learning_rate": 9.150505528110525e-06,
      "loss": 0.4628,
      "step": 3684
    },
    {
      "epoch": 0.1988559710754951,
      "grad_norm": 1.2898357736097448,
      "learning_rate": 9.150059384543165e-06,
      "loss": 0.5822,
      "step": 3685
    },
    {
      "epoch": 0.1989099347040095,
      "grad_norm": 1.1066441917713825,
      "learning_rate": 9.149613136069409e-06,
      "loss": 0.519,
      "step": 3686
    },
    {
      "epoch": 0.19896389833252387,
      "grad_norm": 1.101692089019711,
      "learning_rate": 9.149166782702085e-06,
      "loss": 0.663,
      "step": 3687
    },
    {
      "epoch": 0.19901786196103827,
      "grad_norm": 0.8307843789271808,
      "learning_rate": 9.14872032445402e-06,
      "loss": 0.374,
      "step": 3688
    },
    {
      "epoch": 0.19907182558955264,
      "grad_norm": 1.0642949265532295,
      "learning_rate": 9.148273761338046e-06,
      "loss": 0.6425,
      "step": 3689
    },
    {
      "epoch": 0.19912578921806703,
      "grad_norm": 1.096181684833071,
      "learning_rate": 9.147827093366996e-06,
      "loss": 0.4325,
      "step": 3690
    },
    {
      "epoch": 0.1991797528465814,
      "grad_norm": 1.1574670603263386,
      "learning_rate": 9.147380320553712e-06,
      "loss": 0.4986,
      "step": 3691
    },
    {
      "epoch": 0.19923371647509577,
      "grad_norm": 1.005501035949866,
      "learning_rate": 9.146933442911032e-06,
      "loss": 0.5354,
      "step": 3692
    },
    {
      "epoch": 0.19928768010361017,
      "grad_norm": 1.0195661926515458,
      "learning_rate": 9.146486460451797e-06,
      "loss": 0.4688,
      "step": 3693
    },
    {
      "epoch": 0.19934164373212454,
      "grad_norm": 1.1124136235004647,
      "learning_rate": 9.14603937318886e-06,
      "loss": 0.4182,
      "step": 3694
    },
    {
      "epoch": 0.19939560736063894,
      "grad_norm": 1.234936762727358,
      "learning_rate": 9.145592181135066e-06,
      "loss": 0.5513,
      "step": 3695
    },
    {
      "epoch": 0.1994495709891533,
      "grad_norm": 1.0893680973489885,
      "learning_rate": 9.145144884303268e-06,
      "loss": 0.5407,
      "step": 3696
    },
    {
      "epoch": 0.1995035346176677,
      "grad_norm": 1.3464603067962544,
      "learning_rate": 9.144697482706324e-06,
      "loss": 0.6722,
      "step": 3697
    },
    {
      "epoch": 0.19955749824618207,
      "grad_norm": 1.0284302168979524,
      "learning_rate": 9.144249976357093e-06,
      "loss": 0.4518,
      "step": 3698
    },
    {
      "epoch": 0.19961146187469644,
      "grad_norm": 0.9949188187012509,
      "learning_rate": 9.143802365268435e-06,
      "loss": 0.5139,
      "step": 3699
    },
    {
      "epoch": 0.19966542550321084,
      "grad_norm": 1.0823165027009942,
      "learning_rate": 9.143354649453212e-06,
      "loss": 0.5281,
      "step": 3700
    },
    {
      "epoch": 0.1997193891317252,
      "grad_norm": 1.0907409762550577,
      "learning_rate": 9.1429068289243e-06,
      "loss": 0.4633,
      "step": 3701
    },
    {
      "epoch": 0.1997733527602396,
      "grad_norm": 1.0132207005497866,
      "learning_rate": 9.14245890369456e-06,
      "loss": 0.5515,
      "step": 3702
    },
    {
      "epoch": 0.19982731638875398,
      "grad_norm": 0.8478118733129804,
      "learning_rate": 9.142010873776875e-06,
      "loss": 0.4438,
      "step": 3703
    },
    {
      "epoch": 0.19988128001726835,
      "grad_norm": 0.9427127130661139,
      "learning_rate": 9.141562739184115e-06,
      "loss": 0.4157,
      "step": 3704
    },
    {
      "epoch": 0.19993524364578275,
      "grad_norm": 1.1250699601499659,
      "learning_rate": 9.141114499929162e-06,
      "loss": 0.7275,
      "step": 3705
    },
    {
      "epoch": 0.19998920727429711,
      "grad_norm": 0.9680254469744685,
      "learning_rate": 9.1406661560249e-06,
      "loss": 0.5907,
      "step": 3706
    },
    {
      "epoch": 0.2000431709028115,
      "grad_norm": 0.9195594819190348,
      "learning_rate": 9.140217707484214e-06,
      "loss": 0.4476,
      "step": 3707
    },
    {
      "epoch": 0.20009713453132588,
      "grad_norm": 0.8535184568247356,
      "learning_rate": 9.139769154319993e-06,
      "loss": 0.413,
      "step": 3708
    },
    {
      "epoch": 0.20015109815984028,
      "grad_norm": 1.0798688658374924,
      "learning_rate": 9.139320496545129e-06,
      "loss": 0.4226,
      "step": 3709
    },
    {
      "epoch": 0.20020506178835465,
      "grad_norm": 1.0475628489345057,
      "learning_rate": 9.138871734172516e-06,
      "loss": 0.4836,
      "step": 3710
    },
    {
      "epoch": 0.20025902541686902,
      "grad_norm": 1.1049357465146818,
      "learning_rate": 9.138422867215053e-06,
      "loss": 0.5274,
      "step": 3711
    },
    {
      "epoch": 0.20031298904538342,
      "grad_norm": 1.3199057348444236,
      "learning_rate": 9.137973895685641e-06,
      "loss": 0.59,
      "step": 3712
    },
    {
      "epoch": 0.20036695267389779,
      "grad_norm": 0.809125376913214,
      "learning_rate": 9.137524819597182e-06,
      "loss": 0.3552,
      "step": 3713
    },
    {
      "epoch": 0.20042091630241218,
      "grad_norm": 0.9454948767301519,
      "learning_rate": 9.137075638962584e-06,
      "loss": 0.3995,
      "step": 3714
    },
    {
      "epoch": 0.20047487993092655,
      "grad_norm": 1.111817631093611,
      "learning_rate": 9.13662635379476e-06,
      "loss": 0.521,
      "step": 3715
    },
    {
      "epoch": 0.20052884355944095,
      "grad_norm": 1.051126541186184,
      "learning_rate": 9.136176964106617e-06,
      "loss": 0.5015,
      "step": 3716
    },
    {
      "epoch": 0.20058280718795532,
      "grad_norm": 1.4695960205959748,
      "learning_rate": 9.135727469911075e-06,
      "loss": 0.645,
      "step": 3717
    },
    {
      "epoch": 0.2006367708164697,
      "grad_norm": 0.9996250279813591,
      "learning_rate": 9.135277871221051e-06,
      "loss": 0.4357,
      "step": 3718
    },
    {
      "epoch": 0.2006907344449841,
      "grad_norm": 1.11234064852628,
      "learning_rate": 9.134828168049469e-06,
      "loss": 0.5853,
      "step": 3719
    },
    {
      "epoch": 0.20074469807349846,
      "grad_norm": 0.9939452332084993,
      "learning_rate": 9.134378360409255e-06,
      "loss": 0.452,
      "step": 3720
    },
    {
      "epoch": 0.20079866170201285,
      "grad_norm": 1.0526604119645906,
      "learning_rate": 9.133928448313332e-06,
      "loss": 0.5914,
      "step": 3721
    },
    {
      "epoch": 0.20085262533052722,
      "grad_norm": 1.00049085068388,
      "learning_rate": 9.133478431774631e-06,
      "loss": 0.4602,
      "step": 3722
    },
    {
      "epoch": 0.2009065889590416,
      "grad_norm": 0.8405873770395913,
      "learning_rate": 9.133028310806093e-06,
      "loss": 0.4055,
      "step": 3723
    },
    {
      "epoch": 0.200960552587556,
      "grad_norm": 1.1341216314451776,
      "learning_rate": 9.132578085420648e-06,
      "loss": 0.564,
      "step": 3724
    },
    {
      "epoch": 0.20101451621607036,
      "grad_norm": 1.0759309560977517,
      "learning_rate": 9.132127755631239e-06,
      "loss": 0.397,
      "step": 3725
    },
    {
      "epoch": 0.20106847984458476,
      "grad_norm": 0.9968363196507504,
      "learning_rate": 9.131677321450809e-06,
      "loss": 0.4783,
      "step": 3726
    },
    {
      "epoch": 0.20112244347309913,
      "grad_norm": 1.195439300642246,
      "learning_rate": 9.131226782892303e-06,
      "loss": 0.6821,
      "step": 3727
    },
    {
      "epoch": 0.20117640710161352,
      "grad_norm": 1.0593748786844126,
      "learning_rate": 9.13077613996867e-06,
      "loss": 0.5606,
      "step": 3728
    },
    {
      "epoch": 0.2012303707301279,
      "grad_norm": 1.2347747200174293,
      "learning_rate": 9.130325392692861e-06,
      "loss": 0.5657,
      "step": 3729
    },
    {
      "epoch": 0.20128433435864226,
      "grad_norm": 1.1766522706375906,
      "learning_rate": 9.129874541077832e-06,
      "loss": 0.5685,
      "step": 3730
    },
    {
      "epoch": 0.20133829798715666,
      "grad_norm": 0.6688025576594917,
      "learning_rate": 9.129423585136542e-06,
      "loss": 0.2463,
      "step": 3731
    },
    {
      "epoch": 0.20139226161567103,
      "grad_norm": 0.9700431086726674,
      "learning_rate": 9.128972524881952e-06,
      "loss": 0.3616,
      "step": 3732
    },
    {
      "epoch": 0.20144622524418543,
      "grad_norm": 1.1510830981224824,
      "learning_rate": 9.128521360327025e-06,
      "loss": 0.4724,
      "step": 3733
    },
    {
      "epoch": 0.2015001888726998,
      "grad_norm": 1.1139481495240269,
      "learning_rate": 9.128070091484725e-06,
      "loss": 0.525,
      "step": 3734
    },
    {
      "epoch": 0.2015541525012142,
      "grad_norm": 1.0892795807328075,
      "learning_rate": 9.127618718368026e-06,
      "loss": 0.546,
      "step": 3735
    },
    {
      "epoch": 0.20160811612972857,
      "grad_norm": 1.21382346680645,
      "learning_rate": 9.127167240989899e-06,
      "loss": 0.681,
      "step": 3736
    },
    {
      "epoch": 0.20166207975824293,
      "grad_norm": 1.0981012544491835,
      "learning_rate": 9.126715659363322e-06,
      "loss": 0.5568,
      "step": 3737
    },
    {
      "epoch": 0.20171604338675733,
      "grad_norm": 1.2338161400663619,
      "learning_rate": 9.126263973501271e-06,
      "loss": 0.4916,
      "step": 3738
    },
    {
      "epoch": 0.2017700070152717,
      "grad_norm": 1.2021458283340636,
      "learning_rate": 9.12581218341673e-06,
      "loss": 0.5919,
      "step": 3739
    },
    {
      "epoch": 0.2018239706437861,
      "grad_norm": 0.8029228772857078,
      "learning_rate": 9.125360289122685e-06,
      "loss": 0.3895,
      "step": 3740
    },
    {
      "epoch": 0.20187793427230047,
      "grad_norm": 1.1617322613679533,
      "learning_rate": 9.124908290632119e-06,
      "loss": 0.5522,
      "step": 3741
    },
    {
      "epoch": 0.20193189790081484,
      "grad_norm": 1.11771296420502,
      "learning_rate": 9.124456187958027e-06,
      "loss": 0.6752,
      "step": 3742
    },
    {
      "epoch": 0.20198586152932924,
      "grad_norm": 0.7245114922881668,
      "learning_rate": 9.124003981113402e-06,
      "loss": 0.4101,
      "step": 3743
    },
    {
      "epoch": 0.2020398251578436,
      "grad_norm": 0.898860249771672,
      "learning_rate": 9.123551670111242e-06,
      "loss": 0.4668,
      "step": 3744
    },
    {
      "epoch": 0.202093788786358,
      "grad_norm": 1.2234294778727197,
      "learning_rate": 9.123099254964544e-06,
      "loss": 0.6372,
      "step": 3745
    },
    {
      "epoch": 0.20214775241487237,
      "grad_norm": 1.1270426971777125,
      "learning_rate": 9.122646735686313e-06,
      "loss": 0.5213,
      "step": 3746
    },
    {
      "epoch": 0.20220171604338677,
      "grad_norm": 1.204079728903731,
      "learning_rate": 9.122194112289554e-06,
      "loss": 0.6473,
      "step": 3747
    },
    {
      "epoch": 0.20225567967190114,
      "grad_norm": 0.9090023866417571,
      "learning_rate": 9.121741384787278e-06,
      "loss": 0.4942,
      "step": 3748
    },
    {
      "epoch": 0.2023096433004155,
      "grad_norm": 0.9556497317556221,
      "learning_rate": 9.121288553192494e-06,
      "loss": 0.4741,
      "step": 3749
    },
    {
      "epoch": 0.2023636069289299,
      "grad_norm": 0.9240768386839322,
      "learning_rate": 9.120835617518216e-06,
      "loss": 0.4037,
      "step": 3750
    },
    {
      "epoch": 0.20241757055744428,
      "grad_norm": 1.233303339592849,
      "learning_rate": 9.120382577777468e-06,
      "loss": 0.5834,
      "step": 3751
    },
    {
      "epoch": 0.20247153418595867,
      "grad_norm": 1.024209839920656,
      "learning_rate": 9.119929433983263e-06,
      "loss": 0.5692,
      "step": 3752
    },
    {
      "epoch": 0.20252549781447304,
      "grad_norm": 1.0479240752985408,
      "learning_rate": 9.11947618614863e-06,
      "loss": 0.6407,
      "step": 3753
    },
    {
      "epoch": 0.2025794614429874,
      "grad_norm": 0.9362282824222296,
      "learning_rate": 9.119022834286594e-06,
      "loss": 0.4625,
      "step": 3754
    },
    {
      "epoch": 0.2026334250715018,
      "grad_norm": 1.238717836371045,
      "learning_rate": 9.118569378410183e-06,
      "loss": 0.6564,
      "step": 3755
    },
    {
      "epoch": 0.20268738870001618,
      "grad_norm": 0.9541062575767985,
      "learning_rate": 9.118115818532434e-06,
      "loss": 0.4446,
      "step": 3756
    },
    {
      "epoch": 0.20274135232853058,
      "grad_norm": 1.157429130169837,
      "learning_rate": 9.117662154666382e-06,
      "loss": 0.4296,
      "step": 3757
    },
    {
      "epoch": 0.20279531595704495,
      "grad_norm": 1.2658198841988768,
      "learning_rate": 9.117208386825061e-06,
      "loss": 0.5239,
      "step": 3758
    },
    {
      "epoch": 0.20284927958555934,
      "grad_norm": 0.9807121924602964,
      "learning_rate": 9.116754515021519e-06,
      "loss": 0.3702,
      "step": 3759
    },
    {
      "epoch": 0.20290324321407371,
      "grad_norm": 1.169113994911858,
      "learning_rate": 9.116300539268794e-06,
      "loss": 0.4832,
      "step": 3760
    },
    {
      "epoch": 0.20295720684258808,
      "grad_norm": 1.112474941085719,
      "learning_rate": 9.11584645957994e-06,
      "loss": 0.4955,
      "step": 3761
    },
    {
      "epoch": 0.20301117047110248,
      "grad_norm": 0.8078690227852009,
      "learning_rate": 9.115392275968005e-06,
      "loss": 0.329,
      "step": 3762
    },
    {
      "epoch": 0.20306513409961685,
      "grad_norm": 0.9659226500099185,
      "learning_rate": 9.114937988446044e-06,
      "loss": 0.5848,
      "step": 3763
    },
    {
      "epoch": 0.20311909772813125,
      "grad_norm": 1.1254703567857467,
      "learning_rate": 9.114483597027111e-06,
      "loss": 0.4772,
      "step": 3764
    },
    {
      "epoch": 0.20317306135664562,
      "grad_norm": 1.0774208147264799,
      "learning_rate": 9.114029101724268e-06,
      "loss": 0.5636,
      "step": 3765
    },
    {
      "epoch": 0.20322702498516002,
      "grad_norm": 0.8925182485383784,
      "learning_rate": 9.113574502550576e-06,
      "loss": 0.4924,
      "step": 3766
    },
    {
      "epoch": 0.20328098861367438,
      "grad_norm": 1.1340742932385532,
      "learning_rate": 9.113119799519103e-06,
      "loss": 0.6247,
      "step": 3767
    },
    {
      "epoch": 0.20333495224218875,
      "grad_norm": 1.0360772061554622,
      "learning_rate": 9.112664992642915e-06,
      "loss": 0.5195,
      "step": 3768
    },
    {
      "epoch": 0.20338891587070315,
      "grad_norm": 0.9709576731864566,
      "learning_rate": 9.112210081935084e-06,
      "loss": 0.5441,
      "step": 3769
    },
    {
      "epoch": 0.20344287949921752,
      "grad_norm": 0.8208956677653393,
      "learning_rate": 9.111755067408688e-06,
      "loss": 0.3147,
      "step": 3770
    },
    {
      "epoch": 0.20349684312773192,
      "grad_norm": 0.9670784941304759,
      "learning_rate": 9.111299949076801e-06,
      "loss": 0.4656,
      "step": 3771
    },
    {
      "epoch": 0.2035508067562463,
      "grad_norm": 1.0733556297819917,
      "learning_rate": 9.110844726952503e-06,
      "loss": 0.4475,
      "step": 3772
    },
    {
      "epoch": 0.20360477038476066,
      "grad_norm": 1.1590222183027303,
      "learning_rate": 9.110389401048879e-06,
      "loss": 0.5622,
      "step": 3773
    },
    {
      "epoch": 0.20365873401327506,
      "grad_norm": 0.9668901144273859,
      "learning_rate": 9.109933971379016e-06,
      "loss": 0.4204,
      "step": 3774
    },
    {
      "epoch": 0.20371269764178943,
      "grad_norm": 1.080854489945085,
      "learning_rate": 9.109478437956005e-06,
      "loss": 0.4893,
      "step": 3775
    },
    {
      "epoch": 0.20376666127030382,
      "grad_norm": 1.2927781961112563,
      "learning_rate": 9.109022800792935e-06,
      "loss": 0.7542,
      "step": 3776
    },
    {
      "epoch": 0.2038206248988182,
      "grad_norm": 0.9185852179216725,
      "learning_rate": 9.108567059902903e-06,
      "loss": 0.3752,
      "step": 3777
    },
    {
      "epoch": 0.2038745885273326,
      "grad_norm": 1.1452906267976544,
      "learning_rate": 9.108111215299008e-06,
      "loss": 0.7748,
      "step": 3778
    },
    {
      "epoch": 0.20392855215584696,
      "grad_norm": 1.0824200291017472,
      "learning_rate": 9.10765526699435e-06,
      "loss": 0.4636,
      "step": 3779
    },
    {
      "epoch": 0.20398251578436133,
      "grad_norm": 0.8368675507202236,
      "learning_rate": 9.107199215002036e-06,
      "loss": 0.3172,
      "step": 3780
    },
    {
      "epoch": 0.20403647941287573,
      "grad_norm": 1.0780568229109078,
      "learning_rate": 9.106743059335171e-06,
      "loss": 0.4018,
      "step": 3781
    },
    {
      "epoch": 0.2040904430413901,
      "grad_norm": 0.9992955101218016,
      "learning_rate": 9.106286800006867e-06,
      "loss": 0.4154,
      "step": 3782
    },
    {
      "epoch": 0.2041444066699045,
      "grad_norm": 1.1579744631904958,
      "learning_rate": 9.105830437030234e-06,
      "loss": 0.6817,
      "step": 3783
    },
    {
      "epoch": 0.20419837029841886,
      "grad_norm": 0.8295782847520651,
      "learning_rate": 9.105373970418392e-06,
      "loss": 0.3972,
      "step": 3784
    },
    {
      "epoch": 0.20425233392693326,
      "grad_norm": 1.127411411720756,
      "learning_rate": 9.10491740018446e-06,
      "loss": 0.6368,
      "step": 3785
    },
    {
      "epoch": 0.20430629755544763,
      "grad_norm": 0.937387790429257,
      "learning_rate": 9.104460726341558e-06,
      "loss": 0.457,
      "step": 3786
    },
    {
      "epoch": 0.204360261183962,
      "grad_norm": 1.3483683723351014,
      "learning_rate": 9.104003948902814e-06,
      "loss": 0.6207,
      "step": 3787
    },
    {
      "epoch": 0.2044142248124764,
      "grad_norm": 1.238710290776146,
      "learning_rate": 9.103547067881355e-06,
      "loss": 0.5052,
      "step": 3788
    },
    {
      "epoch": 0.20446818844099077,
      "grad_norm": 1.0984899629588436,
      "learning_rate": 9.103090083290312e-06,
      "loss": 0.6517,
      "step": 3789
    },
    {
      "epoch": 0.20452215206950516,
      "grad_norm": 1.084297066461451,
      "learning_rate": 9.102632995142818e-06,
      "loss": 0.4235,
      "step": 3790
    },
    {
      "epoch": 0.20457611569801953,
      "grad_norm": 0.9898888746164872,
      "learning_rate": 9.102175803452012e-06,
      "loss": 0.4578,
      "step": 3791
    },
    {
      "epoch": 0.2046300793265339,
      "grad_norm": 1.0459420712389298,
      "learning_rate": 9.101718508231034e-06,
      "loss": 0.5295,
      "step": 3792
    },
    {
      "epoch": 0.2046840429550483,
      "grad_norm": 0.8450037866062065,
      "learning_rate": 9.101261109493029e-06,
      "loss": 0.363,
      "step": 3793
    },
    {
      "epoch": 0.20473800658356267,
      "grad_norm": 1.0126135952629696,
      "learning_rate": 9.100803607251138e-06,
      "loss": 0.4806,
      "step": 3794
    },
    {
      "epoch": 0.20479197021207707,
      "grad_norm": 1.1794469538180645,
      "learning_rate": 9.100346001518512e-06,
      "loss": 0.5338,
      "step": 3795
    },
    {
      "epoch": 0.20484593384059144,
      "grad_norm": 0.9852862596143186,
      "learning_rate": 9.099888292308308e-06,
      "loss": 0.4918,
      "step": 3796
    },
    {
      "epoch": 0.20489989746910584,
      "grad_norm": 1.1002683158656994,
      "learning_rate": 9.099430479633674e-06,
      "loss": 0.4877,
      "step": 3797
    },
    {
      "epoch": 0.2049538610976202,
      "grad_norm": 1.3273032700084622,
      "learning_rate": 9.098972563507771e-06,
      "loss": 0.5508,
      "step": 3798
    },
    {
      "epoch": 0.20500782472613457,
      "grad_norm": 0.9520863986741406,
      "learning_rate": 9.098514543943762e-06,
      "loss": 0.373,
      "step": 3799
    },
    {
      "epoch": 0.20506178835464897,
      "grad_norm": 1.0555605223861957,
      "learning_rate": 9.098056420954807e-06,
      "loss": 0.4654,
      "step": 3800
    },
    {
      "epoch": 0.20511575198316334,
      "grad_norm": 0.8038614412779561,
      "learning_rate": 9.097598194554076e-06,
      "loss": 0.4105,
      "step": 3801
    },
    {
      "epoch": 0.20516971561167774,
      "grad_norm": 1.0295001955289393,
      "learning_rate": 9.097139864754738e-06,
      "loss": 0.5183,
      "step": 3802
    },
    {
      "epoch": 0.2052236792401921,
      "grad_norm": 1.23551425545782,
      "learning_rate": 9.096681431569965e-06,
      "loss": 0.5704,
      "step": 3803
    },
    {
      "epoch": 0.20527764286870648,
      "grad_norm": 1.148326710491872,
      "learning_rate": 9.096222895012935e-06,
      "loss": 0.6574,
      "step": 3804
    },
    {
      "epoch": 0.20533160649722088,
      "grad_norm": 1.021426779912693,
      "learning_rate": 9.095764255096823e-06,
      "loss": 0.4965,
      "step": 3805
    },
    {
      "epoch": 0.20538557012573525,
      "grad_norm": 1.0937830868746914,
      "learning_rate": 9.095305511834813e-06,
      "loss": 0.5664,
      "step": 3806
    },
    {
      "epoch": 0.20543953375424964,
      "grad_norm": 0.8990240022806519,
      "learning_rate": 9.094846665240092e-06,
      "loss": 0.467,
      "step": 3807
    },
    {
      "epoch": 0.205493497382764,
      "grad_norm": 1.2601528761778267,
      "learning_rate": 9.094387715325844e-06,
      "loss": 0.6459,
      "step": 3808
    },
    {
      "epoch": 0.2055474610112784,
      "grad_norm": 1.0765649235795685,
      "learning_rate": 9.093928662105261e-06,
      "loss": 0.5058,
      "step": 3809
    },
    {
      "epoch": 0.20560142463979278,
      "grad_norm": 1.1583687805552179,
      "learning_rate": 9.093469505591537e-06,
      "loss": 0.581,
      "step": 3810
    },
    {
      "epoch": 0.20565538826830715,
      "grad_norm": 0.8588458958621885,
      "learning_rate": 9.09301024579787e-06,
      "loss": 0.4083,
      "step": 3811
    },
    {
      "epoch": 0.20570935189682155,
      "grad_norm": 1.0205325869096444,
      "learning_rate": 9.092550882737458e-06,
      "loss": 0.529,
      "step": 3812
    },
    {
      "epoch": 0.20576331552533592,
      "grad_norm": 1.1514347715487017,
      "learning_rate": 9.092091416423502e-06,
      "loss": 0.5511,
      "step": 3813
    },
    {
      "epoch": 0.2058172791538503,
      "grad_norm": 0.9946967484657464,
      "learning_rate": 9.091631846869208e-06,
      "loss": 0.5226,
      "step": 3814
    },
    {
      "epoch": 0.20587124278236468,
      "grad_norm": 1.1848641096765617,
      "learning_rate": 9.091172174087789e-06,
      "loss": 0.5819,
      "step": 3815
    },
    {
      "epoch": 0.20592520641087908,
      "grad_norm": 0.9366671692202229,
      "learning_rate": 9.090712398092451e-06,
      "loss": 0.4026,
      "step": 3816
    },
    {
      "epoch": 0.20597917003939345,
      "grad_norm": 0.7898139235707033,
      "learning_rate": 9.090252518896412e-06,
      "loss": 0.4523,
      "step": 3817
    },
    {
      "epoch": 0.20603313366790782,
      "grad_norm": 0.6649074646820992,
      "learning_rate": 9.089792536512887e-06,
      "loss": 0.3197,
      "step": 3818
    },
    {
      "epoch": 0.20608709729642222,
      "grad_norm": 1.3778380184118229,
      "learning_rate": 9.089332450955097e-06,
      "loss": 0.6503,
      "step": 3819
    },
    {
      "epoch": 0.2061410609249366,
      "grad_norm": 1.0365057247961833,
      "learning_rate": 9.088872262236266e-06,
      "loss": 0.4165,
      "step": 3820
    },
    {
      "epoch": 0.20619502455345098,
      "grad_norm": 1.2670053733517477,
      "learning_rate": 9.088411970369619e-06,
      "loss": 0.6734,
      "step": 3821
    },
    {
      "epoch": 0.20624898818196535,
      "grad_norm": 0.9117451318292406,
      "learning_rate": 9.087951575368386e-06,
      "loss": 0.3714,
      "step": 3822
    },
    {
      "epoch": 0.20630295181047972,
      "grad_norm": 1.0465414362507464,
      "learning_rate": 9.087491077245801e-06,
      "loss": 0.4708,
      "step": 3823
    },
    {
      "epoch": 0.20635691543899412,
      "grad_norm": 1.2517457139784924,
      "learning_rate": 9.087030476015096e-06,
      "loss": 0.4552,
      "step": 3824
    },
    {
      "epoch": 0.2064108790675085,
      "grad_norm": 1.257565526312639,
      "learning_rate": 9.086569771689511e-06,
      "loss": 0.6989,
      "step": 3825
    },
    {
      "epoch": 0.2064648426960229,
      "grad_norm": 1.0479561616606654,
      "learning_rate": 9.086108964282288e-06,
      "loss": 0.3842,
      "step": 3826
    },
    {
      "epoch": 0.20651880632453726,
      "grad_norm": 0.7239587247970184,
      "learning_rate": 9.08564805380667e-06,
      "loss": 0.2905,
      "step": 3827
    },
    {
      "epoch": 0.20657276995305165,
      "grad_norm": 0.7892006314927505,
      "learning_rate": 9.085187040275903e-06,
      "loss": 0.4043,
      "step": 3828
    },
    {
      "epoch": 0.20662673358156602,
      "grad_norm": 1.0232482133900491,
      "learning_rate": 9.084725923703239e-06,
      "loss": 0.4761,
      "step": 3829
    },
    {
      "epoch": 0.2066806972100804,
      "grad_norm": 1.076150120569057,
      "learning_rate": 9.084264704101928e-06,
      "loss": 0.3932,
      "step": 3830
    },
    {
      "epoch": 0.2067346608385948,
      "grad_norm": 0.9726391786688343,
      "learning_rate": 9.083803381485228e-06,
      "loss": 0.5401,
      "step": 3831
    },
    {
      "epoch": 0.20678862446710916,
      "grad_norm": 0.9680419362008221,
      "learning_rate": 9.083341955866397e-06,
      "loss": 0.4337,
      "step": 3832
    },
    {
      "epoch": 0.20684258809562356,
      "grad_norm": 1.152372370389519,
      "learning_rate": 9.0828804272587e-06,
      "loss": 0.4794,
      "step": 3833
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 1.1488019150465112,
      "learning_rate": 9.082418795675397e-06,
      "loss": 0.4996,
      "step": 3834
    },
    {
      "epoch": 0.20695051535265233,
      "grad_norm": 1.1613562490404625,
      "learning_rate": 9.081957061129759e-06,
      "loss": 0.5326,
      "step": 3835
    },
    {
      "epoch": 0.2070044789811667,
      "grad_norm": 1.0321620061465928,
      "learning_rate": 9.081495223635056e-06,
      "loss": 0.5818,
      "step": 3836
    },
    {
      "epoch": 0.20705844260968106,
      "grad_norm": 1.1031430448277395,
      "learning_rate": 9.081033283204562e-06,
      "loss": 0.5467,
      "step": 3837
    },
    {
      "epoch": 0.20711240623819546,
      "grad_norm": 1.17344392101849,
      "learning_rate": 9.080571239851553e-06,
      "loss": 0.5775,
      "step": 3838
    },
    {
      "epoch": 0.20716636986670983,
      "grad_norm": 1.0929116623317745,
      "learning_rate": 9.080109093589307e-06,
      "loss": 0.4895,
      "step": 3839
    },
    {
      "epoch": 0.20722033349522423,
      "grad_norm": 1.1104152489632335,
      "learning_rate": 9.079646844431109e-06,
      "loss": 0.5181,
      "step": 3840
    },
    {
      "epoch": 0.2072742971237386,
      "grad_norm": 1.172165218288601,
      "learning_rate": 9.079184492390243e-06,
      "loss": 0.7755,
      "step": 3841
    },
    {
      "epoch": 0.20732826075225297,
      "grad_norm": 0.862232851144642,
      "learning_rate": 9.07872203748e-06,
      "loss": 0.3866,
      "step": 3842
    },
    {
      "epoch": 0.20738222438076737,
      "grad_norm": 1.145180630667375,
      "learning_rate": 9.078259479713668e-06,
      "loss": 0.4741,
      "step": 3843
    },
    {
      "epoch": 0.20743618800928174,
      "grad_norm": 0.8201913840696834,
      "learning_rate": 9.077796819104543e-06,
      "loss": 0.323,
      "step": 3844
    },
    {
      "epoch": 0.20749015163779613,
      "grad_norm": 0.9465618237701664,
      "learning_rate": 9.077334055665924e-06,
      "loss": 0.51,
      "step": 3845
    },
    {
      "epoch": 0.2075441152663105,
      "grad_norm": 0.730568131899897,
      "learning_rate": 9.076871189411109e-06,
      "loss": 0.262,
      "step": 3846
    },
    {
      "epoch": 0.2075980788948249,
      "grad_norm": 1.3392263889868612,
      "learning_rate": 9.0764082203534e-06,
      "loss": 0.4794,
      "step": 3847
    },
    {
      "epoch": 0.20765204252333927,
      "grad_norm": 1.1405824253588246,
      "learning_rate": 9.075945148506108e-06,
      "loss": 0.8523,
      "step": 3848
    },
    {
      "epoch": 0.20770600615185364,
      "grad_norm": 1.121804347175139,
      "learning_rate": 9.075481973882536e-06,
      "loss": 0.6648,
      "step": 3849
    },
    {
      "epoch": 0.20775996978036804,
      "grad_norm": 0.8206467960411512,
      "learning_rate": 9.075018696496002e-06,
      "loss": 0.3791,
      "step": 3850
    },
    {
      "epoch": 0.2078139334088824,
      "grad_norm": 0.9918088821101608,
      "learning_rate": 9.074555316359817e-06,
      "loss": 0.4499,
      "step": 3851
    },
    {
      "epoch": 0.2078678970373968,
      "grad_norm": 0.9970567092088943,
      "learning_rate": 9.0740918334873e-06,
      "loss": 0.4116,
      "step": 3852
    },
    {
      "epoch": 0.20792186066591117,
      "grad_norm": 0.9813827498885892,
      "learning_rate": 9.073628247891775e-06,
      "loss": 0.2799,
      "step": 3853
    },
    {
      "epoch": 0.20797582429442557,
      "grad_norm": 1.1529450533520045,
      "learning_rate": 9.07316455958656e-06,
      "loss": 0.5303,
      "step": 3854
    },
    {
      "epoch": 0.20802978792293994,
      "grad_norm": 1.0205694942858632,
      "learning_rate": 9.072700768584986e-06,
      "loss": 0.4358,
      "step": 3855
    },
    {
      "epoch": 0.2080837515514543,
      "grad_norm": 1.0648678275716918,
      "learning_rate": 9.072236874900385e-06,
      "loss": 0.5501,
      "step": 3856
    },
    {
      "epoch": 0.2081377151799687,
      "grad_norm": 1.3816538057096799,
      "learning_rate": 9.071772878546083e-06,
      "loss": 0.6802,
      "step": 3857
    },
    {
      "epoch": 0.20819167880848308,
      "grad_norm": 1.0495137518250075,
      "learning_rate": 9.07130877953542e-06,
      "loss": 0.4967,
      "step": 3858
    },
    {
      "epoch": 0.20824564243699747,
      "grad_norm": 1.1863997644347717,
      "learning_rate": 9.070844577881735e-06,
      "loss": 0.5231,
      "step": 3859
    },
    {
      "epoch": 0.20829960606551184,
      "grad_norm": 0.9326985312300379,
      "learning_rate": 9.070380273598368e-06,
      "loss": 0.3829,
      "step": 3860
    },
    {
      "epoch": 0.2083535696940262,
      "grad_norm": 1.1410302104902033,
      "learning_rate": 9.069915866698666e-06,
      "loss": 0.5768,
      "step": 3861
    },
    {
      "epoch": 0.2084075333225406,
      "grad_norm": 1.0875319487257407,
      "learning_rate": 9.069451357195974e-06,
      "loss": 0.6122,
      "step": 3862
    },
    {
      "epoch": 0.20846149695105498,
      "grad_norm": 0.731919192222664,
      "learning_rate": 9.068986745103643e-06,
      "loss": 0.3615,
      "step": 3863
    },
    {
      "epoch": 0.20851546057956938,
      "grad_norm": 1.108033990239275,
      "learning_rate": 9.068522030435025e-06,
      "loss": 0.6173,
      "step": 3864
    },
    {
      "epoch": 0.20856942420808375,
      "grad_norm": 1.126197723277647,
      "learning_rate": 9.068057213203481e-06,
      "loss": 0.5498,
      "step": 3865
    },
    {
      "epoch": 0.20862338783659815,
      "grad_norm": 1.2839405504606491,
      "learning_rate": 9.067592293422366e-06,
      "loss": 0.6349,
      "step": 3866
    },
    {
      "epoch": 0.20867735146511251,
      "grad_norm": 0.9764652265971667,
      "learning_rate": 9.067127271105043e-06,
      "loss": 0.4433,
      "step": 3867
    },
    {
      "epoch": 0.20873131509362688,
      "grad_norm": 0.7594458240336066,
      "learning_rate": 9.06666214626488e-06,
      "loss": 0.4252,
      "step": 3868
    },
    {
      "epoch": 0.20878527872214128,
      "grad_norm": 1.0964259034579231,
      "learning_rate": 9.066196918915243e-06,
      "loss": 0.5883,
      "step": 3869
    },
    {
      "epoch": 0.20883924235065565,
      "grad_norm": 1.1874433087104141,
      "learning_rate": 9.0657315890695e-06,
      "loss": 0.6546,
      "step": 3870
    },
    {
      "epoch": 0.20889320597917005,
      "grad_norm": 1.2360349584605888,
      "learning_rate": 9.065266156741031e-06,
      "loss": 0.5418,
      "step": 3871
    },
    {
      "epoch": 0.20894716960768442,
      "grad_norm": 0.9329975163968138,
      "learning_rate": 9.06480062194321e-06,
      "loss": 0.4156,
      "step": 3872
    },
    {
      "epoch": 0.2090011332361988,
      "grad_norm": 1.4863437214203792,
      "learning_rate": 9.064334984689417e-06,
      "loss": 0.6885,
      "step": 3873
    },
    {
      "epoch": 0.20905509686471319,
      "grad_norm": 0.8371105586740093,
      "learning_rate": 9.063869244993035e-06,
      "loss": 0.3983,
      "step": 3874
    },
    {
      "epoch": 0.20910906049322756,
      "grad_norm": 1.0338699673628178,
      "learning_rate": 9.06340340286745e-06,
      "loss": 0.49,
      "step": 3875
    },
    {
      "epoch": 0.20916302412174195,
      "grad_norm": 0.9841241600143728,
      "learning_rate": 9.06293745832605e-06,
      "loss": 0.4871,
      "step": 3876
    },
    {
      "epoch": 0.20921698775025632,
      "grad_norm": 1.1415178268292643,
      "learning_rate": 9.062471411382228e-06,
      "loss": 0.517,
      "step": 3877
    },
    {
      "epoch": 0.20927095137877072,
      "grad_norm": 1.1482227277060528,
      "learning_rate": 9.062005262049377e-06,
      "loss": 0.4614,
      "step": 3878
    },
    {
      "epoch": 0.2093249150072851,
      "grad_norm": 1.1838484878353295,
      "learning_rate": 9.061539010340898e-06,
      "loss": 0.6213,
      "step": 3879
    },
    {
      "epoch": 0.20937887863579946,
      "grad_norm": 1.2359974630529107,
      "learning_rate": 9.061072656270186e-06,
      "loss": 0.7294,
      "step": 3880
    },
    {
      "epoch": 0.20943284226431386,
      "grad_norm": 1.3210752697964867,
      "learning_rate": 9.060606199850648e-06,
      "loss": 0.6458,
      "step": 3881
    },
    {
      "epoch": 0.20948680589282823,
      "grad_norm": 1.0389529672318072,
      "learning_rate": 9.060139641095694e-06,
      "loss": 0.3766,
      "step": 3882
    },
    {
      "epoch": 0.20954076952134262,
      "grad_norm": 1.0025935314030558,
      "learning_rate": 9.059672980018725e-06,
      "loss": 0.5016,
      "step": 3883
    },
    {
      "epoch": 0.209594733149857,
      "grad_norm": 1.1218859341736258,
      "learning_rate": 9.059206216633161e-06,
      "loss": 0.5728,
      "step": 3884
    },
    {
      "epoch": 0.2096486967783714,
      "grad_norm": 1.202372198238103,
      "learning_rate": 9.058739350952411e-06,
      "loss": 0.5637,
      "step": 3885
    },
    {
      "epoch": 0.20970266040688576,
      "grad_norm": 1.0351766674514313,
      "learning_rate": 9.058272382989899e-06,
      "loss": 0.5758,
      "step": 3886
    },
    {
      "epoch": 0.20975662403540013,
      "grad_norm": 1.135669097182712,
      "learning_rate": 9.057805312759042e-06,
      "loss": 0.6708,
      "step": 3887
    },
    {
      "epoch": 0.20981058766391453,
      "grad_norm": 1.1728584558035258,
      "learning_rate": 9.057338140273263e-06,
      "loss": 0.7437,
      "step": 3888
    },
    {
      "epoch": 0.2098645512924289,
      "grad_norm": 0.9670312473946563,
      "learning_rate": 9.056870865545995e-06,
      "loss": 0.4158,
      "step": 3889
    },
    {
      "epoch": 0.2099185149209433,
      "grad_norm": 1.0169606853053872,
      "learning_rate": 9.056403488590664e-06,
      "loss": 0.4104,
      "step": 3890
    },
    {
      "epoch": 0.20997247854945766,
      "grad_norm": 1.0513006911578175,
      "learning_rate": 9.055936009420701e-06,
      "loss": 0.4837,
      "step": 3891
    },
    {
      "epoch": 0.21002644217797203,
      "grad_norm": 0.994266635767907,
      "learning_rate": 9.055468428049547e-06,
      "loss": 0.5132,
      "step": 3892
    },
    {
      "epoch": 0.21008040580648643,
      "grad_norm": 1.3176735817535101,
      "learning_rate": 9.055000744490636e-06,
      "loss": 0.7013,
      "step": 3893
    },
    {
      "epoch": 0.2101343694350008,
      "grad_norm": 1.0985426132081926,
      "learning_rate": 9.054532958757413e-06,
      "loss": 0.6176,
      "step": 3894
    },
    {
      "epoch": 0.2101883330635152,
      "grad_norm": 0.9023904960073221,
      "learning_rate": 9.05406507086332e-06,
      "loss": 0.4917,
      "step": 3895
    },
    {
      "epoch": 0.21024229669202957,
      "grad_norm": 0.9832362482417746,
      "learning_rate": 9.053597080821806e-06,
      "loss": 0.4066,
      "step": 3896
    },
    {
      "epoch": 0.21029626032054397,
      "grad_norm": 1.021488711386533,
      "learning_rate": 9.05312898864632e-06,
      "loss": 0.4585,
      "step": 3897
    },
    {
      "epoch": 0.21035022394905833,
      "grad_norm": 1.2145369172421092,
      "learning_rate": 9.052660794350318e-06,
      "loss": 0.5546,
      "step": 3898
    },
    {
      "epoch": 0.2104041875775727,
      "grad_norm": 0.9983674272525982,
      "learning_rate": 9.052192497947255e-06,
      "loss": 0.4014,
      "step": 3899
    },
    {
      "epoch": 0.2104581512060871,
      "grad_norm": 1.0745283458527848,
      "learning_rate": 9.051724099450591e-06,
      "loss": 0.5116,
      "step": 3900
    },
    {
      "epoch": 0.21051211483460147,
      "grad_norm": 0.9298405422290709,
      "learning_rate": 9.051255598873787e-06,
      "loss": 0.5071,
      "step": 3901
    },
    {
      "epoch": 0.21056607846311587,
      "grad_norm": 1.1694964336381588,
      "learning_rate": 9.050786996230308e-06,
      "loss": 0.425,
      "step": 3902
    },
    {
      "epoch": 0.21062004209163024,
      "grad_norm": 1.125314301489721,
      "learning_rate": 9.050318291533624e-06,
      "loss": 0.6144,
      "step": 3903
    },
    {
      "epoch": 0.21067400572014464,
      "grad_norm": 0.9432318833764584,
      "learning_rate": 9.049849484797205e-06,
      "loss": 0.4255,
      "step": 3904
    },
    {
      "epoch": 0.210727969348659,
      "grad_norm": 1.1475244897572214,
      "learning_rate": 9.049380576034524e-06,
      "loss": 0.5721,
      "step": 3905
    },
    {
      "epoch": 0.21078193297717338,
      "grad_norm": 1.0482920789783552,
      "learning_rate": 9.04891156525906e-06,
      "loss": 0.4578,
      "step": 3906
    },
    {
      "epoch": 0.21083589660568777,
      "grad_norm": 0.9043257434772238,
      "learning_rate": 9.04844245248429e-06,
      "loss": 0.4027,
      "step": 3907
    },
    {
      "epoch": 0.21088986023420214,
      "grad_norm": 1.2417123813265774,
      "learning_rate": 9.0479732377237e-06,
      "loss": 0.7197,
      "step": 3908
    },
    {
      "epoch": 0.21094382386271654,
      "grad_norm": 0.9912213476465899,
      "learning_rate": 9.047503920990773e-06,
      "loss": 0.4283,
      "step": 3909
    },
    {
      "epoch": 0.2109977874912309,
      "grad_norm": 1.0625242799866899,
      "learning_rate": 9.047034502299001e-06,
      "loss": 0.4694,
      "step": 3910
    },
    {
      "epoch": 0.21105175111974528,
      "grad_norm": 1.184741449080488,
      "learning_rate": 9.046564981661872e-06,
      "loss": 0.5428,
      "step": 3911
    },
    {
      "epoch": 0.21110571474825968,
      "grad_norm": 1.0232591537901663,
      "learning_rate": 9.046095359092882e-06,
      "loss": 0.6114,
      "step": 3912
    },
    {
      "epoch": 0.21115967837677405,
      "grad_norm": 1.0483303416013794,
      "learning_rate": 9.045625634605528e-06,
      "loss": 0.4922,
      "step": 3913
    },
    {
      "epoch": 0.21121364200528844,
      "grad_norm": 1.0152066138835125,
      "learning_rate": 9.04515580821331e-06,
      "loss": 0.4821,
      "step": 3914
    },
    {
      "epoch": 0.2112676056338028,
      "grad_norm": 0.9005491964594909,
      "learning_rate": 9.044685879929734e-06,
      "loss": 0.5389,
      "step": 3915
    },
    {
      "epoch": 0.2113215692623172,
      "grad_norm": 0.8736380425750268,
      "learning_rate": 9.044215849768301e-06,
      "loss": 0.3748,
      "step": 3916
    },
    {
      "epoch": 0.21137553289083158,
      "grad_norm": 1.1130456743586536,
      "learning_rate": 9.043745717742527e-06,
      "loss": 0.4542,
      "step": 3917
    },
    {
      "epoch": 0.21142949651934595,
      "grad_norm": 0.9087707234184306,
      "learning_rate": 9.043275483865918e-06,
      "loss": 0.3466,
      "step": 3918
    },
    {
      "epoch": 0.21148346014786035,
      "grad_norm": 1.1589649440131629,
      "learning_rate": 9.042805148151991e-06,
      "loss": 0.5031,
      "step": 3919
    },
    {
      "epoch": 0.21153742377637472,
      "grad_norm": 0.916612921848723,
      "learning_rate": 9.042334710614267e-06,
      "loss": 0.4077,
      "step": 3920
    },
    {
      "epoch": 0.21159138740488911,
      "grad_norm": 1.09943628477085,
      "learning_rate": 9.041864171266261e-06,
      "loss": 0.6119,
      "step": 3921
    },
    {
      "epoch": 0.21164535103340348,
      "grad_norm": 1.0573910691570063,
      "learning_rate": 9.041393530121503e-06,
      "loss": 0.5843,
      "step": 3922
    },
    {
      "epoch": 0.21169931466191788,
      "grad_norm": 1.0216860887252937,
      "learning_rate": 9.040922787193514e-06,
      "loss": 0.4922,
      "step": 3923
    },
    {
      "epoch": 0.21175327829043225,
      "grad_norm": 1.0388331986792314,
      "learning_rate": 9.040451942495827e-06,
      "loss": 0.377,
      "step": 3924
    },
    {
      "epoch": 0.21180724191894662,
      "grad_norm": 1.3289374662872535,
      "learning_rate": 9.039980996041974e-06,
      "loss": 0.5038,
      "step": 3925
    },
    {
      "epoch": 0.21186120554746102,
      "grad_norm": 1.2046519157742162,
      "learning_rate": 9.03950994784549e-06,
      "loss": 0.628,
      "step": 3926
    },
    {
      "epoch": 0.2119151691759754,
      "grad_norm": 1.0317793064865266,
      "learning_rate": 9.039038797919915e-06,
      "loss": 0.4974,
      "step": 3927
    },
    {
      "epoch": 0.21196913280448978,
      "grad_norm": 0.8901791870227522,
      "learning_rate": 9.038567546278788e-06,
      "loss": 0.3548,
      "step": 3928
    },
    {
      "epoch": 0.21202309643300415,
      "grad_norm": 1.0571825549646474,
      "learning_rate": 9.038096192935655e-06,
      "loss": 0.5301,
      "step": 3929
    },
    {
      "epoch": 0.21207706006151852,
      "grad_norm": 0.9965957672642536,
      "learning_rate": 9.03762473790406e-06,
      "loss": 0.4422,
      "step": 3930
    },
    {
      "epoch": 0.21213102369003292,
      "grad_norm": 1.135320265719766,
      "learning_rate": 9.037153181197558e-06,
      "loss": 0.5513,
      "step": 3931
    },
    {
      "epoch": 0.2121849873185473,
      "grad_norm": 0.891997805803228,
      "learning_rate": 9.036681522829698e-06,
      "loss": 0.5171,
      "step": 3932
    },
    {
      "epoch": 0.2122389509470617,
      "grad_norm": 0.7622237444844877,
      "learning_rate": 9.036209762814038e-06,
      "loss": 0.3067,
      "step": 3933
    },
    {
      "epoch": 0.21229291457557606,
      "grad_norm": 1.0378832163556038,
      "learning_rate": 9.035737901164138e-06,
      "loss": 0.5174,
      "step": 3934
    },
    {
      "epoch": 0.21234687820409046,
      "grad_norm": 1.155870327987508,
      "learning_rate": 9.035265937893556e-06,
      "loss": 0.6494,
      "step": 3935
    },
    {
      "epoch": 0.21240084183260483,
      "grad_norm": 1.1442332615468283,
      "learning_rate": 9.03479387301586e-06,
      "loss": 0.513,
      "step": 3936
    },
    {
      "epoch": 0.2124548054611192,
      "grad_norm": 0.874162254995752,
      "learning_rate": 9.034321706544617e-06,
      "loss": 0.44,
      "step": 3937
    },
    {
      "epoch": 0.2125087690896336,
      "grad_norm": 1.184770792385275,
      "learning_rate": 9.033849438493396e-06,
      "loss": 0.5672,
      "step": 3938
    },
    {
      "epoch": 0.21256273271814796,
      "grad_norm": 1.063807350405684,
      "learning_rate": 9.03337706887577e-06,
      "loss": 0.4988,
      "step": 3939
    },
    {
      "epoch": 0.21261669634666236,
      "grad_norm": 1.1456144386108473,
      "learning_rate": 9.032904597705318e-06,
      "loss": 0.5399,
      "step": 3940
    },
    {
      "epoch": 0.21267065997517673,
      "grad_norm": 0.9093594370494846,
      "learning_rate": 9.032432024995618e-06,
      "loss": 0.3614,
      "step": 3941
    },
    {
      "epoch": 0.2127246236036911,
      "grad_norm": 1.1817771519574174,
      "learning_rate": 9.031959350760254e-06,
      "loss": 0.6147,
      "step": 3942
    },
    {
      "epoch": 0.2127785872322055,
      "grad_norm": 0.9428467533589016,
      "learning_rate": 9.031486575012809e-06,
      "loss": 0.4093,
      "step": 3943
    },
    {
      "epoch": 0.21283255086071987,
      "grad_norm": 0.7599893495075416,
      "learning_rate": 9.031013697766871e-06,
      "loss": 0.4031,
      "step": 3944
    },
    {
      "epoch": 0.21288651448923426,
      "grad_norm": 1.1440116340229138,
      "learning_rate": 9.03054071903603e-06,
      "loss": 0.515,
      "step": 3945
    },
    {
      "epoch": 0.21294047811774863,
      "grad_norm": 0.8962400584161349,
      "learning_rate": 9.030067638833882e-06,
      "loss": 0.3129,
      "step": 3946
    },
    {
      "epoch": 0.21299444174626303,
      "grad_norm": 1.1504491610991356,
      "learning_rate": 9.029594457174025e-06,
      "loss": 0.4481,
      "step": 3947
    },
    {
      "epoch": 0.2130484053747774,
      "grad_norm": 1.0946426863870848,
      "learning_rate": 9.029121174070055e-06,
      "loss": 0.4452,
      "step": 3948
    },
    {
      "epoch": 0.21310236900329177,
      "grad_norm": 0.9903173583548343,
      "learning_rate": 9.028647789535575e-06,
      "loss": 0.5246,
      "step": 3949
    },
    {
      "epoch": 0.21315633263180617,
      "grad_norm": 1.0607844823404882,
      "learning_rate": 9.028174303584198e-06,
      "loss": 0.5039,
      "step": 3950
    },
    {
      "epoch": 0.21321029626032054,
      "grad_norm": 1.0008862319530052,
      "learning_rate": 9.027700716229522e-06,
      "loss": 0.4335,
      "step": 3951
    },
    {
      "epoch": 0.21326425988883493,
      "grad_norm": 1.2022074823755349,
      "learning_rate": 9.027227027485163e-06,
      "loss": 0.4486,
      "step": 3952
    },
    {
      "epoch": 0.2133182235173493,
      "grad_norm": 0.9284403411512231,
      "learning_rate": 9.026753237364735e-06,
      "loss": 0.3795,
      "step": 3953
    },
    {
      "epoch": 0.2133721871458637,
      "grad_norm": 0.6324710945038569,
      "learning_rate": 9.026279345881858e-06,
      "loss": 0.2983,
      "step": 3954
    },
    {
      "epoch": 0.21342615077437807,
      "grad_norm": 1.1952451858348805,
      "learning_rate": 9.025805353050149e-06,
      "loss": 0.7813,
      "step": 3955
    },
    {
      "epoch": 0.21348011440289244,
      "grad_norm": 1.260142705979331,
      "learning_rate": 9.02533125888323e-06,
      "loss": 0.6504,
      "step": 3956
    },
    {
      "epoch": 0.21353407803140684,
      "grad_norm": 1.2540320144076724,
      "learning_rate": 9.024857063394729e-06,
      "loss": 0.6934,
      "step": 3957
    },
    {
      "epoch": 0.2135880416599212,
      "grad_norm": 1.1477533983017443,
      "learning_rate": 9.024382766598274e-06,
      "loss": 0.5293,
      "step": 3958
    },
    {
      "epoch": 0.2136420052884356,
      "grad_norm": 0.9493108751403639,
      "learning_rate": 9.023908368507498e-06,
      "loss": 0.5305,
      "step": 3959
    },
    {
      "epoch": 0.21369596891694997,
      "grad_norm": 0.9524559818208181,
      "learning_rate": 9.023433869136034e-06,
      "loss": 0.4008,
      "step": 3960
    },
    {
      "epoch": 0.21374993254546434,
      "grad_norm": 0.7820605858019544,
      "learning_rate": 9.022959268497522e-06,
      "loss": 0.3139,
      "step": 3961
    },
    {
      "epoch": 0.21380389617397874,
      "grad_norm": 1.1350732499094303,
      "learning_rate": 9.022484566605599e-06,
      "loss": 0.5454,
      "step": 3962
    },
    {
      "epoch": 0.2138578598024931,
      "grad_norm": 0.9023840038650534,
      "learning_rate": 9.022009763473912e-06,
      "loss": 0.4492,
      "step": 3963
    },
    {
      "epoch": 0.2139118234310075,
      "grad_norm": 0.9966316783480835,
      "learning_rate": 9.021534859116104e-06,
      "loss": 0.5936,
      "step": 3964
    },
    {
      "epoch": 0.21396578705952188,
      "grad_norm": 1.3261019470858544,
      "learning_rate": 9.021059853545828e-06,
      "loss": 0.7163,
      "step": 3965
    },
    {
      "epoch": 0.21401975068803628,
      "grad_norm": 0.8726886947641488,
      "learning_rate": 9.020584746776731e-06,
      "loss": 0.4816,
      "step": 3966
    },
    {
      "epoch": 0.21407371431655065,
      "grad_norm": 1.2910453457306585,
      "learning_rate": 9.020109538822474e-06,
      "loss": 0.7472,
      "step": 3967
    },
    {
      "epoch": 0.21412767794506501,
      "grad_norm": 1.2231730028898598,
      "learning_rate": 9.019634229696711e-06,
      "loss": 0.8247,
      "step": 3968
    },
    {
      "epoch": 0.2141816415735794,
      "grad_norm": 1.1891277143936283,
      "learning_rate": 9.019158819413104e-06,
      "loss": 0.6231,
      "step": 3969
    },
    {
      "epoch": 0.21423560520209378,
      "grad_norm": 1.0118545804911891,
      "learning_rate": 9.018683307985317e-06,
      "loss": 0.5139,
      "step": 3970
    },
    {
      "epoch": 0.21428956883060818,
      "grad_norm": 0.9973902877485001,
      "learning_rate": 9.018207695427015e-06,
      "loss": 0.4673,
      "step": 3971
    },
    {
      "epoch": 0.21434353245912255,
      "grad_norm": 0.8424684420498934,
      "learning_rate": 9.01773198175187e-06,
      "loss": 0.364,
      "step": 3972
    },
    {
      "epoch": 0.21439749608763695,
      "grad_norm": 1.0674028067795285,
      "learning_rate": 9.017256166973554e-06,
      "loss": 0.523,
      "step": 3973
    },
    {
      "epoch": 0.21445145971615132,
      "grad_norm": 1.1139550831641658,
      "learning_rate": 9.01678025110574e-06,
      "loss": 0.4778,
      "step": 3974
    },
    {
      "epoch": 0.21450542334466569,
      "grad_norm": 0.8965690917412549,
      "learning_rate": 9.01630423416211e-06,
      "loss": 0.369,
      "step": 3975
    },
    {
      "epoch": 0.21455938697318008,
      "grad_norm": 0.8215376518556305,
      "learning_rate": 9.015828116156343e-06,
      "loss": 0.3582,
      "step": 3976
    },
    {
      "epoch": 0.21461335060169445,
      "grad_norm": 0.8693451111709126,
      "learning_rate": 9.015351897102123e-06,
      "loss": 0.4038,
      "step": 3977
    },
    {
      "epoch": 0.21466731423020885,
      "grad_norm": 1.4012206145220893,
      "learning_rate": 9.014875577013137e-06,
      "loss": 0.5622,
      "step": 3978
    },
    {
      "epoch": 0.21472127785872322,
      "grad_norm": 1.0657626753983134,
      "learning_rate": 9.014399155903077e-06,
      "loss": 0.482,
      "step": 3979
    },
    {
      "epoch": 0.2147752414872376,
      "grad_norm": 1.129379227255306,
      "learning_rate": 9.013922633785632e-06,
      "loss": 0.5782,
      "step": 3980
    },
    {
      "epoch": 0.214829205115752,
      "grad_norm": 1.2337238539095285,
      "learning_rate": 9.0134460106745e-06,
      "loss": 0.6213,
      "step": 3981
    },
    {
      "epoch": 0.21488316874426636,
      "grad_norm": 1.0357518292975791,
      "learning_rate": 9.012969286583381e-06,
      "loss": 0.5107,
      "step": 3982
    },
    {
      "epoch": 0.21493713237278075,
      "grad_norm": 1.0196893526290927,
      "learning_rate": 9.012492461525976e-06,
      "loss": 0.4394,
      "step": 3983
    },
    {
      "epoch": 0.21499109600129512,
      "grad_norm": 0.8946379944661457,
      "learning_rate": 9.012015535515987e-06,
      "loss": 0.4383,
      "step": 3984
    },
    {
      "epoch": 0.21504505962980952,
      "grad_norm": 1.0568577099659149,
      "learning_rate": 9.011538508567122e-06,
      "loss": 0.5242,
      "step": 3985
    },
    {
      "epoch": 0.2150990232583239,
      "grad_norm": 0.9451234441609712,
      "learning_rate": 9.011061380693094e-06,
      "loss": 0.426,
      "step": 3986
    },
    {
      "epoch": 0.21515298688683826,
      "grad_norm": 1.0153537833073967,
      "learning_rate": 9.010584151907615e-06,
      "loss": 0.5366,
      "step": 3987
    },
    {
      "epoch": 0.21520695051535266,
      "grad_norm": 1.1497605468617007,
      "learning_rate": 9.010106822224398e-06,
      "loss": 0.6683,
      "step": 3988
    },
    {
      "epoch": 0.21526091414386703,
      "grad_norm": 1.212212990329426,
      "learning_rate": 9.009629391657166e-06,
      "loss": 0.596,
      "step": 3989
    },
    {
      "epoch": 0.21531487777238142,
      "grad_norm": 0.950751734965433,
      "learning_rate": 9.009151860219637e-06,
      "loss": 0.4963,
      "step": 3990
    },
    {
      "epoch": 0.2153688414008958,
      "grad_norm": 1.0972075578089489,
      "learning_rate": 9.008674227925539e-06,
      "loss": 0.4146,
      "step": 3991
    },
    {
      "epoch": 0.21542280502941016,
      "grad_norm": 1.1645867041581603,
      "learning_rate": 9.0081964947886e-06,
      "loss": 0.6042,
      "step": 3992
    },
    {
      "epoch": 0.21547676865792456,
      "grad_norm": 1.4023136307320923,
      "learning_rate": 9.007718660822547e-06,
      "loss": 0.6124,
      "step": 3993
    },
    {
      "epoch": 0.21553073228643893,
      "grad_norm": 1.0124618801017364,
      "learning_rate": 9.007240726041115e-06,
      "loss": 0.4454,
      "step": 3994
    },
    {
      "epoch": 0.21558469591495333,
      "grad_norm": 0.9208581619967927,
      "learning_rate": 9.00676269045804e-06,
      "loss": 0.4426,
      "step": 3995
    },
    {
      "epoch": 0.2156386595434677,
      "grad_norm": 0.8666223155897976,
      "learning_rate": 9.006284554087065e-06,
      "loss": 0.4424,
      "step": 3996
    },
    {
      "epoch": 0.2156926231719821,
      "grad_norm": 1.3599242310249426,
      "learning_rate": 9.005806316941927e-06,
      "loss": 0.6218,
      "step": 3997
    },
    {
      "epoch": 0.21574658680049646,
      "grad_norm": 0.8411577808261556,
      "learning_rate": 9.005327979036373e-06,
      "loss": 0.3873,
      "step": 3998
    },
    {
      "epoch": 0.21580055042901083,
      "grad_norm": 0.9657131027277541,
      "learning_rate": 9.004849540384153e-06,
      "loss": 0.5001,
      "step": 3999
    },
    {
      "epoch": 0.21585451405752523,
      "grad_norm": 1.292862194383677,
      "learning_rate": 9.004371000999013e-06,
      "loss": 0.6114,
      "step": 4000
    },
    {
      "epoch": 0.21585451405752523,
      "eval_loss": 0.5817294120788574,
      "eval_runtime": 160.8162,
      "eval_samples_per_second": 21.385,
      "eval_steps_per_second": 0.895,
      "step": 4000
    },
    {
      "epoch": 0.2159084776860396,
      "grad_norm": 1.0563048429126018,
      "learning_rate": 9.003892360894711e-06,
      "loss": 0.5492,
      "step": 4001
    },
    {
      "epoch": 0.215962441314554,
      "grad_norm": 1.0515773777035606,
      "learning_rate": 9.003413620085002e-06,
      "loss": 0.448,
      "step": 4002
    },
    {
      "epoch": 0.21601640494306837,
      "grad_norm": 1.1816263709952683,
      "learning_rate": 9.002934778583647e-06,
      "loss": 0.5911,
      "step": 4003
    },
    {
      "epoch": 0.21607036857158277,
      "grad_norm": 0.9884789111703723,
      "learning_rate": 9.002455836404404e-06,
      "loss": 0.5374,
      "step": 4004
    },
    {
      "epoch": 0.21612433220009714,
      "grad_norm": 0.9484853864647014,
      "learning_rate": 9.001976793561044e-06,
      "loss": 0.5617,
      "step": 4005
    },
    {
      "epoch": 0.2161782958286115,
      "grad_norm": 1.18031054644485,
      "learning_rate": 9.001497650067331e-06,
      "loss": 0.4511,
      "step": 4006
    },
    {
      "epoch": 0.2162322594571259,
      "grad_norm": 1.252263560155836,
      "learning_rate": 9.001018405937038e-06,
      "loss": 0.5487,
      "step": 4007
    },
    {
      "epoch": 0.21628622308564027,
      "grad_norm": 1.0061238804944843,
      "learning_rate": 9.000539061183939e-06,
      "loss": 0.5622,
      "step": 4008
    },
    {
      "epoch": 0.21634018671415467,
      "grad_norm": 0.8942974252079775,
      "learning_rate": 9.00005961582181e-06,
      "loss": 0.3772,
      "step": 4009
    },
    {
      "epoch": 0.21639415034266904,
      "grad_norm": 0.9514468082825852,
      "learning_rate": 8.999580069864429e-06,
      "loss": 0.4783,
      "step": 4010
    },
    {
      "epoch": 0.2164481139711834,
      "grad_norm": 1.0411101974513266,
      "learning_rate": 8.999100423325585e-06,
      "loss": 0.4902,
      "step": 4011
    },
    {
      "epoch": 0.2165020775996978,
      "grad_norm": 1.0168004115480387,
      "learning_rate": 8.998620676219057e-06,
      "loss": 0.5075,
      "step": 4012
    },
    {
      "epoch": 0.21655604122821218,
      "grad_norm": 1.2457042345560991,
      "learning_rate": 8.998140828558634e-06,
      "loss": 0.5782,
      "step": 4013
    },
    {
      "epoch": 0.21661000485672657,
      "grad_norm": 0.8665809394188911,
      "learning_rate": 8.997660880358111e-06,
      "loss": 0.4262,
      "step": 4014
    },
    {
      "epoch": 0.21666396848524094,
      "grad_norm": 1.1878243764904397,
      "learning_rate": 8.997180831631279e-06,
      "loss": 0.5639,
      "step": 4015
    },
    {
      "epoch": 0.21671793211375534,
      "grad_norm": 1.0888283580903304,
      "learning_rate": 8.996700682391938e-06,
      "loss": 0.3735,
      "step": 4016
    },
    {
      "epoch": 0.2167718957422697,
      "grad_norm": 1.2367023549717704,
      "learning_rate": 8.996220432653884e-06,
      "loss": 0.6938,
      "step": 4017
    },
    {
      "epoch": 0.21682585937078408,
      "grad_norm": 0.9808439328255408,
      "learning_rate": 8.995740082430924e-06,
      "loss": 0.4821,
      "step": 4018
    },
    {
      "epoch": 0.21687982299929848,
      "grad_norm": 1.0366699148700493,
      "learning_rate": 8.995259631736862e-06,
      "loss": 0.5017,
      "step": 4019
    },
    {
      "epoch": 0.21693378662781285,
      "grad_norm": 0.9273114940176643,
      "learning_rate": 8.994779080585503e-06,
      "loss": 0.3972,
      "step": 4020
    },
    {
      "epoch": 0.21698775025632724,
      "grad_norm": 1.2573622007462064,
      "learning_rate": 8.994298428990667e-06,
      "loss": 0.6585,
      "step": 4021
    },
    {
      "epoch": 0.2170417138848416,
      "grad_norm": 1.0937169778287301,
      "learning_rate": 8.99381767696616e-06,
      "loss": 0.5648,
      "step": 4022
    },
    {
      "epoch": 0.217095677513356,
      "grad_norm": 1.33401456577074,
      "learning_rate": 8.993336824525804e-06,
      "loss": 0.6267,
      "step": 4023
    },
    {
      "epoch": 0.21714964114187038,
      "grad_norm": 1.0501639380316055,
      "learning_rate": 8.992855871683417e-06,
      "loss": 0.4715,
      "step": 4024
    },
    {
      "epoch": 0.21720360477038475,
      "grad_norm": 1.2312575956203569,
      "learning_rate": 8.992374818452823e-06,
      "loss": 0.5575,
      "step": 4025
    },
    {
      "epoch": 0.21725756839889915,
      "grad_norm": 1.0149235200835898,
      "learning_rate": 8.99189366484785e-06,
      "loss": 0.4831,
      "step": 4026
    },
    {
      "epoch": 0.21731153202741352,
      "grad_norm": 1.000839083245443,
      "learning_rate": 8.991412410882323e-06,
      "loss": 0.451,
      "step": 4027
    },
    {
      "epoch": 0.21736549565592791,
      "grad_norm": 1.0214543211395493,
      "learning_rate": 8.990931056570075e-06,
      "loss": 0.5052,
      "step": 4028
    },
    {
      "epoch": 0.21741945928444228,
      "grad_norm": 0.967045541897537,
      "learning_rate": 8.990449601924943e-06,
      "loss": 0.4166,
      "step": 4029
    },
    {
      "epoch": 0.21747342291295665,
      "grad_norm": 1.3024119976346453,
      "learning_rate": 8.989968046960761e-06,
      "loss": 0.6361,
      "step": 4030
    },
    {
      "epoch": 0.21752738654147105,
      "grad_norm": 1.3366669988509314,
      "learning_rate": 8.989486391691372e-06,
      "loss": 0.5269,
      "step": 4031
    },
    {
      "epoch": 0.21758135016998542,
      "grad_norm": 1.029736233840357,
      "learning_rate": 8.989004636130618e-06,
      "loss": 0.4635,
      "step": 4032
    },
    {
      "epoch": 0.21763531379849982,
      "grad_norm": 1.1642415151600105,
      "learning_rate": 8.988522780292345e-06,
      "loss": 0.5301,
      "step": 4033
    },
    {
      "epoch": 0.2176892774270142,
      "grad_norm": 1.3044939567494471,
      "learning_rate": 8.988040824190402e-06,
      "loss": 0.5727,
      "step": 4034
    },
    {
      "epoch": 0.21774324105552859,
      "grad_norm": 1.243176583591493,
      "learning_rate": 8.987558767838642e-06,
      "loss": 0.6035,
      "step": 4035
    },
    {
      "epoch": 0.21779720468404296,
      "grad_norm": 1.2813814586958956,
      "learning_rate": 8.987076611250918e-06,
      "loss": 0.5033,
      "step": 4036
    },
    {
      "epoch": 0.21785116831255733,
      "grad_norm": 1.0621116910036337,
      "learning_rate": 8.98659435444109e-06,
      "loss": 0.3963,
      "step": 4037
    },
    {
      "epoch": 0.21790513194107172,
      "grad_norm": 1.0541490348535043,
      "learning_rate": 8.986111997423017e-06,
      "loss": 0.4538,
      "step": 4038
    },
    {
      "epoch": 0.2179590955695861,
      "grad_norm": 0.8236509031673342,
      "learning_rate": 8.985629540210561e-06,
      "loss": 0.4094,
      "step": 4039
    },
    {
      "epoch": 0.2180130591981005,
      "grad_norm": 1.2004135326224785,
      "learning_rate": 8.985146982817593e-06,
      "loss": 0.4932,
      "step": 4040
    },
    {
      "epoch": 0.21806702282661486,
      "grad_norm": 1.1167471145002856,
      "learning_rate": 8.984664325257977e-06,
      "loss": 0.4979,
      "step": 4041
    },
    {
      "epoch": 0.21812098645512926,
      "grad_norm": 1.0271973268916812,
      "learning_rate": 8.984181567545591e-06,
      "loss": 0.5463,
      "step": 4042
    },
    {
      "epoch": 0.21817495008364363,
      "grad_norm": 1.3530530212199736,
      "learning_rate": 8.983698709694301e-06,
      "loss": 0.631,
      "step": 4043
    },
    {
      "epoch": 0.218228913712158,
      "grad_norm": 1.1438034775101846,
      "learning_rate": 8.983215751717993e-06,
      "loss": 0.5026,
      "step": 4044
    },
    {
      "epoch": 0.2182828773406724,
      "grad_norm": 1.1642893761609479,
      "learning_rate": 8.982732693630544e-06,
      "loss": 0.7373,
      "step": 4045
    },
    {
      "epoch": 0.21833684096918676,
      "grad_norm": 1.0935518027083104,
      "learning_rate": 8.982249535445839e-06,
      "loss": 0.3831,
      "step": 4046
    },
    {
      "epoch": 0.21839080459770116,
      "grad_norm": 1.2980427621761101,
      "learning_rate": 8.981766277177764e-06,
      "loss": 0.6428,
      "step": 4047
    },
    {
      "epoch": 0.21844476822621553,
      "grad_norm": 1.017452541800854,
      "learning_rate": 8.981282918840207e-06,
      "loss": 0.5347,
      "step": 4048
    },
    {
      "epoch": 0.2184987318547299,
      "grad_norm": 0.9050196178244304,
      "learning_rate": 8.980799460447062e-06,
      "loss": 0.3752,
      "step": 4049
    },
    {
      "epoch": 0.2185526954832443,
      "grad_norm": 0.9787900219809772,
      "learning_rate": 8.980315902012225e-06,
      "loss": 0.5528,
      "step": 4050
    },
    {
      "epoch": 0.21860665911175867,
      "grad_norm": 0.9616656827000798,
      "learning_rate": 8.97983224354959e-06,
      "loss": 0.4803,
      "step": 4051
    },
    {
      "epoch": 0.21866062274027306,
      "grad_norm": 0.9220009905140023,
      "learning_rate": 8.97934848507306e-06,
      "loss": 0.5308,
      "step": 4052
    },
    {
      "epoch": 0.21871458636878743,
      "grad_norm": 1.161256216281666,
      "learning_rate": 8.978864626596539e-06,
      "loss": 0.6176,
      "step": 4053
    },
    {
      "epoch": 0.21876854999730183,
      "grad_norm": 1.0634278657312266,
      "learning_rate": 8.978380668133935e-06,
      "loss": 0.39,
      "step": 4054
    },
    {
      "epoch": 0.2188225136258162,
      "grad_norm": 1.030157959467724,
      "learning_rate": 8.977896609699155e-06,
      "loss": 0.3982,
      "step": 4055
    },
    {
      "epoch": 0.21887647725433057,
      "grad_norm": 0.8235031171575286,
      "learning_rate": 8.977412451306113e-06,
      "loss": 0.3665,
      "step": 4056
    },
    {
      "epoch": 0.21893044088284497,
      "grad_norm": 1.1368618754906603,
      "learning_rate": 8.976928192968724e-06,
      "loss": 0.4891,
      "step": 4057
    },
    {
      "epoch": 0.21898440451135934,
      "grad_norm": 1.0438595482066209,
      "learning_rate": 8.976443834700903e-06,
      "loss": 0.6424,
      "step": 4058
    },
    {
      "epoch": 0.21903836813987373,
      "grad_norm": 0.9972526266260556,
      "learning_rate": 8.975959376516575e-06,
      "loss": 0.4432,
      "step": 4059
    },
    {
      "epoch": 0.2190923317683881,
      "grad_norm": 0.957352258519993,
      "learning_rate": 8.975474818429664e-06,
      "loss": 0.5539,
      "step": 4060
    },
    {
      "epoch": 0.21914629539690247,
      "grad_norm": 1.3009896276290596,
      "learning_rate": 8.974990160454094e-06,
      "loss": 0.5961,
      "step": 4061
    },
    {
      "epoch": 0.21920025902541687,
      "grad_norm": 1.2429791798238328,
      "learning_rate": 8.974505402603793e-06,
      "loss": 0.5173,
      "step": 4062
    },
    {
      "epoch": 0.21925422265393124,
      "grad_norm": 1.0283813761778038,
      "learning_rate": 8.974020544892699e-06,
      "loss": 0.5579,
      "step": 4063
    },
    {
      "epoch": 0.21930818628244564,
      "grad_norm": 0.8985381319945738,
      "learning_rate": 8.973535587334743e-06,
      "loss": 0.4627,
      "step": 4064
    },
    {
      "epoch": 0.21936214991096,
      "grad_norm": 0.9537275172497061,
      "learning_rate": 8.973050529943863e-06,
      "loss": 0.4152,
      "step": 4065
    },
    {
      "epoch": 0.2194161135394744,
      "grad_norm": 0.865533992664969,
      "learning_rate": 8.972565372734004e-06,
      "loss": 0.4835,
      "step": 4066
    },
    {
      "epoch": 0.21947007716798878,
      "grad_norm": 0.8934062523831986,
      "learning_rate": 8.972080115719107e-06,
      "loss": 0.3278,
      "step": 4067
    },
    {
      "epoch": 0.21952404079650314,
      "grad_norm": 1.0477904992543636,
      "learning_rate": 8.971594758913119e-06,
      "loss": 0.5012,
      "step": 4068
    },
    {
      "epoch": 0.21957800442501754,
      "grad_norm": 0.9616828304255225,
      "learning_rate": 8.97110930232999e-06,
      "loss": 0.5428,
      "step": 4069
    },
    {
      "epoch": 0.2196319680535319,
      "grad_norm": 1.1313341683599356,
      "learning_rate": 8.970623745983673e-06,
      "loss": 0.5242,
      "step": 4070
    },
    {
      "epoch": 0.2196859316820463,
      "grad_norm": 1.0418498418774882,
      "learning_rate": 8.97013808988812e-06,
      "loss": 0.5005,
      "step": 4071
    },
    {
      "epoch": 0.21973989531056068,
      "grad_norm": 0.7414784823001763,
      "learning_rate": 8.969652334057296e-06,
      "loss": 0.2434,
      "step": 4072
    },
    {
      "epoch": 0.21979385893907508,
      "grad_norm": 1.072182239328063,
      "learning_rate": 8.969166478505155e-06,
      "loss": 0.5964,
      "step": 4073
    },
    {
      "epoch": 0.21984782256758945,
      "grad_norm": 1.1612831183380452,
      "learning_rate": 8.968680523245665e-06,
      "loss": 0.3883,
      "step": 4074
    },
    {
      "epoch": 0.21990178619610382,
      "grad_norm": 1.0297712400615584,
      "learning_rate": 8.968194468292792e-06,
      "loss": 0.4257,
      "step": 4075
    },
    {
      "epoch": 0.2199557498246182,
      "grad_norm": 0.7961173237206042,
      "learning_rate": 8.967708313660504e-06,
      "loss": 0.3833,
      "step": 4076
    },
    {
      "epoch": 0.22000971345313258,
      "grad_norm": 1.167388854994612,
      "learning_rate": 8.967222059362777e-06,
      "loss": 0.6036,
      "step": 4077
    },
    {
      "epoch": 0.22006367708164698,
      "grad_norm": 1.0281012597884969,
      "learning_rate": 8.966735705413584e-06,
      "loss": 0.5783,
      "step": 4078
    },
    {
      "epoch": 0.22011764071016135,
      "grad_norm": 0.9551695348081268,
      "learning_rate": 8.966249251826905e-06,
      "loss": 0.4604,
      "step": 4079
    },
    {
      "epoch": 0.22017160433867572,
      "grad_norm": 1.124765000567545,
      "learning_rate": 8.96576269861672e-06,
      "loss": 0.571,
      "step": 4080
    },
    {
      "epoch": 0.22022556796719012,
      "grad_norm": 0.9112201646827748,
      "learning_rate": 8.965276045797011e-06,
      "loss": 0.437,
      "step": 4081
    },
    {
      "epoch": 0.2202795315957045,
      "grad_norm": 1.0292561958525943,
      "learning_rate": 8.96478929338177e-06,
      "loss": 0.4876,
      "step": 4082
    },
    {
      "epoch": 0.22033349522421888,
      "grad_norm": 0.9877788416416108,
      "learning_rate": 8.964302441384983e-06,
      "loss": 0.4398,
      "step": 4083
    },
    {
      "epoch": 0.22038745885273325,
      "grad_norm": 0.9322745236600732,
      "learning_rate": 8.96381548982064e-06,
      "loss": 0.4167,
      "step": 4084
    },
    {
      "epoch": 0.22044142248124765,
      "grad_norm": 0.9877958398889292,
      "learning_rate": 8.963328438702744e-06,
      "loss": 0.4249,
      "step": 4085
    },
    {
      "epoch": 0.22049538610976202,
      "grad_norm": 1.1734113771186048,
      "learning_rate": 8.962841288045287e-06,
      "loss": 0.524,
      "step": 4086
    },
    {
      "epoch": 0.2205493497382764,
      "grad_norm": 1.0822743828770462,
      "learning_rate": 8.962354037862274e-06,
      "loss": 0.5177,
      "step": 4087
    },
    {
      "epoch": 0.2206033133667908,
      "grad_norm": 0.9996964330874292,
      "learning_rate": 8.961866688167708e-06,
      "loss": 0.4903,
      "step": 4088
    },
    {
      "epoch": 0.22065727699530516,
      "grad_norm": 1.6177500774623486,
      "learning_rate": 8.961379238975594e-06,
      "loss": 0.5319,
      "step": 4089
    },
    {
      "epoch": 0.22071124062381955,
      "grad_norm": 0.9509287108513386,
      "learning_rate": 8.960891690299944e-06,
      "loss": 0.4636,
      "step": 4090
    },
    {
      "epoch": 0.22076520425233392,
      "grad_norm": 1.031238015015512,
      "learning_rate": 8.960404042154767e-06,
      "loss": 0.6424,
      "step": 4091
    },
    {
      "epoch": 0.22081916788084832,
      "grad_norm": 1.0106914192316385,
      "learning_rate": 8.959916294554084e-06,
      "loss": 0.3962,
      "step": 4092
    },
    {
      "epoch": 0.2208731315093627,
      "grad_norm": 1.0910875891908924,
      "learning_rate": 8.959428447511912e-06,
      "loss": 0.5329,
      "step": 4093
    },
    {
      "epoch": 0.22092709513787706,
      "grad_norm": 1.0385840012897831,
      "learning_rate": 8.95894050104227e-06,
      "loss": 0.4289,
      "step": 4094
    },
    {
      "epoch": 0.22098105876639146,
      "grad_norm": 1.0045327230079815,
      "learning_rate": 8.95845245515918e-06,
      "loss": 0.5755,
      "step": 4095
    },
    {
      "epoch": 0.22103502239490583,
      "grad_norm": 0.8823817879051734,
      "learning_rate": 8.957964309876675e-06,
      "loss": 0.4123,
      "step": 4096
    },
    {
      "epoch": 0.22108898602342023,
      "grad_norm": 0.8540134158441888,
      "learning_rate": 8.957476065208782e-06,
      "loss": 0.379,
      "step": 4097
    },
    {
      "epoch": 0.2211429496519346,
      "grad_norm": 1.1624954516311459,
      "learning_rate": 8.956987721169532e-06,
      "loss": 0.594,
      "step": 4098
    },
    {
      "epoch": 0.22119691328044896,
      "grad_norm": 1.13859785979404,
      "learning_rate": 8.956499277772961e-06,
      "loss": 0.4572,
      "step": 4099
    },
    {
      "epoch": 0.22125087690896336,
      "grad_norm": 0.9655043430426752,
      "learning_rate": 8.956010735033109e-06,
      "loss": 0.4984,
      "step": 4100
    },
    {
      "epoch": 0.22130484053747773,
      "grad_norm": 1.1517416895404495,
      "learning_rate": 8.955522092964017e-06,
      "loss": 0.5561,
      "step": 4101
    },
    {
      "epoch": 0.22135880416599213,
      "grad_norm": 1.0374749142356428,
      "learning_rate": 8.955033351579729e-06,
      "loss": 0.4836,
      "step": 4102
    },
    {
      "epoch": 0.2214127677945065,
      "grad_norm": 1.4497411288788788,
      "learning_rate": 8.954544510894291e-06,
      "loss": 0.7018,
      "step": 4103
    },
    {
      "epoch": 0.2214667314230209,
      "grad_norm": 0.8015117899857264,
      "learning_rate": 8.954055570921751e-06,
      "loss": 0.3448,
      "step": 4104
    },
    {
      "epoch": 0.22152069505153527,
      "grad_norm": 0.9796133860791367,
      "learning_rate": 8.953566531676166e-06,
      "loss": 0.4345,
      "step": 4105
    },
    {
      "epoch": 0.22157465868004964,
      "grad_norm": 1.2846425814363263,
      "learning_rate": 8.95307739317159e-06,
      "loss": 0.6509,
      "step": 4106
    },
    {
      "epoch": 0.22162862230856403,
      "grad_norm": 0.9288885421021263,
      "learning_rate": 8.952588155422079e-06,
      "loss": 0.4174,
      "step": 4107
    },
    {
      "epoch": 0.2216825859370784,
      "grad_norm": 0.9458164143901633,
      "learning_rate": 8.952098818441697e-06,
      "loss": 0.3676,
      "step": 4108
    },
    {
      "epoch": 0.2217365495655928,
      "grad_norm": 1.1121638402023792,
      "learning_rate": 8.951609382244506e-06,
      "loss": 0.5679,
      "step": 4109
    },
    {
      "epoch": 0.22179051319410717,
      "grad_norm": 0.9729036549505502,
      "learning_rate": 8.951119846844576e-06,
      "loss": 0.4568,
      "step": 4110
    },
    {
      "epoch": 0.22184447682262157,
      "grad_norm": 1.171064719184763,
      "learning_rate": 8.950630212255971e-06,
      "loss": 0.4814,
      "step": 4111
    },
    {
      "epoch": 0.22189844045113594,
      "grad_norm": 0.9611435363428809,
      "learning_rate": 8.95014047849277e-06,
      "loss": 0.4274,
      "step": 4112
    },
    {
      "epoch": 0.2219524040796503,
      "grad_norm": 1.1424155034886319,
      "learning_rate": 8.949650645569042e-06,
      "loss": 0.5637,
      "step": 4113
    },
    {
      "epoch": 0.2220063677081647,
      "grad_norm": 0.967686409838488,
      "learning_rate": 8.949160713498872e-06,
      "loss": 0.411,
      "step": 4114
    },
    {
      "epoch": 0.22206033133667907,
      "grad_norm": 0.8086050482496685,
      "learning_rate": 8.948670682296338e-06,
      "loss": 0.3656,
      "step": 4115
    },
    {
      "epoch": 0.22211429496519347,
      "grad_norm": 1.0702450413453835,
      "learning_rate": 8.948180551975522e-06,
      "loss": 0.6477,
      "step": 4116
    },
    {
      "epoch": 0.22216825859370784,
      "grad_norm": 1.01536331206658,
      "learning_rate": 8.947690322550514e-06,
      "loss": 0.439,
      "step": 4117
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.903582903828541,
      "learning_rate": 8.947199994035402e-06,
      "loss": 0.4243,
      "step": 4118
    },
    {
      "epoch": 0.2222761858507366,
      "grad_norm": 0.9374751026841127,
      "learning_rate": 8.94670956644428e-06,
      "loss": 0.4227,
      "step": 4119
    },
    {
      "epoch": 0.22233014947925098,
      "grad_norm": 1.0596942250130763,
      "learning_rate": 8.94621903979124e-06,
      "loss": 0.62,
      "step": 4120
    },
    {
      "epoch": 0.22238411310776537,
      "grad_norm": 0.8436998136427647,
      "learning_rate": 8.945728414090384e-06,
      "loss": 0.4039,
      "step": 4121
    },
    {
      "epoch": 0.22243807673627974,
      "grad_norm": 0.7628293416739909,
      "learning_rate": 8.945237689355814e-06,
      "loss": 0.3252,
      "step": 4122
    },
    {
      "epoch": 0.22249204036479414,
      "grad_norm": 1.0350417374084566,
      "learning_rate": 8.944746865601629e-06,
      "loss": 0.5435,
      "step": 4123
    },
    {
      "epoch": 0.2225460039933085,
      "grad_norm": 0.9698398203088243,
      "learning_rate": 8.94425594284194e-06,
      "loss": 0.4395,
      "step": 4124
    },
    {
      "epoch": 0.22259996762182288,
      "grad_norm": 1.0930077205524755,
      "learning_rate": 8.943764921090854e-06,
      "loss": 0.5986,
      "step": 4125
    },
    {
      "epoch": 0.22265393125033728,
      "grad_norm": 1.1529034604248085,
      "learning_rate": 8.943273800362485e-06,
      "loss": 0.622,
      "step": 4126
    },
    {
      "epoch": 0.22270789487885165,
      "grad_norm": 0.980949974799224,
      "learning_rate": 8.942782580670946e-06,
      "loss": 0.5569,
      "step": 4127
    },
    {
      "epoch": 0.22276185850736605,
      "grad_norm": 1.2002415814334453,
      "learning_rate": 8.94229126203036e-06,
      "loss": 0.7676,
      "step": 4128
    },
    {
      "epoch": 0.22281582213588041,
      "grad_norm": 1.1664303687141202,
      "learning_rate": 8.941799844454842e-06,
      "loss": 0.5861,
      "step": 4129
    },
    {
      "epoch": 0.22286978576439478,
      "grad_norm": 1.0338148026822138,
      "learning_rate": 8.94130832795852e-06,
      "loss": 0.3795,
      "step": 4130
    },
    {
      "epoch": 0.22292374939290918,
      "grad_norm": 1.1187286175572075,
      "learning_rate": 8.94081671255552e-06,
      "loss": 0.4514,
      "step": 4131
    },
    {
      "epoch": 0.22297771302142355,
      "grad_norm": 1.046668632098277,
      "learning_rate": 8.94032499825997e-06,
      "loss": 0.5655,
      "step": 4132
    },
    {
      "epoch": 0.22303167664993795,
      "grad_norm": 0.9836688524422647,
      "learning_rate": 8.939833185086005e-06,
      "loss": 0.3575,
      "step": 4133
    },
    {
      "epoch": 0.22308564027845232,
      "grad_norm": 0.9791946066732348,
      "learning_rate": 8.939341273047758e-06,
      "loss": 0.4513,
      "step": 4134
    },
    {
      "epoch": 0.22313960390696672,
      "grad_norm": 0.8850210780279401,
      "learning_rate": 8.938849262159368e-06,
      "loss": 0.4186,
      "step": 4135
    },
    {
      "epoch": 0.22319356753548109,
      "grad_norm": 1.2294477585170056,
      "learning_rate": 8.938357152434975e-06,
      "loss": 0.6458,
      "step": 4136
    },
    {
      "epoch": 0.22324753116399546,
      "grad_norm": 1.241116771816486,
      "learning_rate": 8.937864943888721e-06,
      "loss": 0.6709,
      "step": 4137
    },
    {
      "epoch": 0.22330149479250985,
      "grad_norm": 1.0100196945962354,
      "learning_rate": 8.937372636534757e-06,
      "loss": 0.5567,
      "step": 4138
    },
    {
      "epoch": 0.22335545842102422,
      "grad_norm": 1.1201647378160644,
      "learning_rate": 8.93688023038723e-06,
      "loss": 0.5282,
      "step": 4139
    },
    {
      "epoch": 0.22340942204953862,
      "grad_norm": 1.2610779113955102,
      "learning_rate": 8.936387725460293e-06,
      "loss": 0.5776,
      "step": 4140
    },
    {
      "epoch": 0.223463385678053,
      "grad_norm": 1.0957387229292392,
      "learning_rate": 8.9358951217681e-06,
      "loss": 0.5212,
      "step": 4141
    },
    {
      "epoch": 0.2235173493065674,
      "grad_norm": 1.1782102401401422,
      "learning_rate": 8.935402419324809e-06,
      "loss": 0.5886,
      "step": 4142
    },
    {
      "epoch": 0.22357131293508176,
      "grad_norm": 1.1713213594844538,
      "learning_rate": 8.934909618144583e-06,
      "loss": 0.5134,
      "step": 4143
    },
    {
      "epoch": 0.22362527656359613,
      "grad_norm": 1.4572245958079553,
      "learning_rate": 8.934416718241582e-06,
      "loss": 0.58,
      "step": 4144
    },
    {
      "epoch": 0.22367924019211052,
      "grad_norm": 0.8982852214507034,
      "learning_rate": 8.933923719629975e-06,
      "loss": 0.4228,
      "step": 4145
    },
    {
      "epoch": 0.2237332038206249,
      "grad_norm": 1.0367242252408408,
      "learning_rate": 8.93343062232393e-06,
      "loss": 0.36,
      "step": 4146
    },
    {
      "epoch": 0.2237871674491393,
      "grad_norm": 1.106136579679524,
      "learning_rate": 8.932937426337619e-06,
      "loss": 0.4767,
      "step": 4147
    },
    {
      "epoch": 0.22384113107765366,
      "grad_norm": 0.9708779660580348,
      "learning_rate": 8.93244413168522e-06,
      "loss": 0.4327,
      "step": 4148
    },
    {
      "epoch": 0.22389509470616803,
      "grad_norm": 0.8729816428325439,
      "learning_rate": 8.931950738380907e-06,
      "loss": 0.4061,
      "step": 4149
    },
    {
      "epoch": 0.22394905833468243,
      "grad_norm": 1.1169144939160642,
      "learning_rate": 8.931457246438862e-06,
      "loss": 0.4832,
      "step": 4150
    },
    {
      "epoch": 0.2240030219631968,
      "grad_norm": 0.8734765413553331,
      "learning_rate": 8.930963655873268e-06,
      "loss": 0.3442,
      "step": 4151
    },
    {
      "epoch": 0.2240569855917112,
      "grad_norm": 1.1580977068059468,
      "learning_rate": 8.930469966698313e-06,
      "loss": 0.5714,
      "step": 4152
    },
    {
      "epoch": 0.22411094922022556,
      "grad_norm": 1.3200087552695383,
      "learning_rate": 8.929976178928185e-06,
      "loss": 0.4888,
      "step": 4153
    },
    {
      "epoch": 0.22416491284873996,
      "grad_norm": 1.0242800604911402,
      "learning_rate": 8.929482292577075e-06,
      "loss": 0.4936,
      "step": 4154
    },
    {
      "epoch": 0.22421887647725433,
      "grad_norm": 1.200200188353733,
      "learning_rate": 8.928988307659178e-06,
      "loss": 0.5812,
      "step": 4155
    },
    {
      "epoch": 0.2242728401057687,
      "grad_norm": 0.9934708027308374,
      "learning_rate": 8.928494224188694e-06,
      "loss": 0.4851,
      "step": 4156
    },
    {
      "epoch": 0.2243268037342831,
      "grad_norm": 0.8781530264094205,
      "learning_rate": 8.92800004217982e-06,
      "loss": 0.4091,
      "step": 4157
    },
    {
      "epoch": 0.22438076736279747,
      "grad_norm": 1.0636847590192327,
      "learning_rate": 8.927505761646764e-06,
      "loss": 0.6683,
      "step": 4158
    },
    {
      "epoch": 0.22443473099131186,
      "grad_norm": 0.8592389729553446,
      "learning_rate": 8.927011382603725e-06,
      "loss": 0.3564,
      "step": 4159
    },
    {
      "epoch": 0.22448869461982623,
      "grad_norm": 1.1971357086116088,
      "learning_rate": 8.92651690506492e-06,
      "loss": 0.5926,
      "step": 4160
    },
    {
      "epoch": 0.22454265824834063,
      "grad_norm": 0.8433865543220043,
      "learning_rate": 8.926022329044555e-06,
      "loss": 0.4982,
      "step": 4161
    },
    {
      "epoch": 0.224596621876855,
      "grad_norm": 1.2121082482377088,
      "learning_rate": 8.925527654556846e-06,
      "loss": 0.6006,
      "step": 4162
    },
    {
      "epoch": 0.22465058550536937,
      "grad_norm": 1.0063795255508543,
      "learning_rate": 8.925032881616013e-06,
      "loss": 0.4056,
      "step": 4163
    },
    {
      "epoch": 0.22470454913388377,
      "grad_norm": 1.1077478268428798,
      "learning_rate": 8.924538010236273e-06,
      "loss": 0.4408,
      "step": 4164
    },
    {
      "epoch": 0.22475851276239814,
      "grad_norm": 1.0763533584744858,
      "learning_rate": 8.92404304043185e-06,
      "loss": 0.6446,
      "step": 4165
    },
    {
      "epoch": 0.22481247639091254,
      "grad_norm": 1.17910620580634,
      "learning_rate": 8.92354797221697e-06,
      "loss": 0.5861,
      "step": 4166
    },
    {
      "epoch": 0.2248664400194269,
      "grad_norm": 1.1481134849077221,
      "learning_rate": 8.923052805605861e-06,
      "loss": 0.4669,
      "step": 4167
    },
    {
      "epoch": 0.22492040364794127,
      "grad_norm": 0.9608982268793681,
      "learning_rate": 8.922557540612759e-06,
      "loss": 0.3779,
      "step": 4168
    },
    {
      "epoch": 0.22497436727645567,
      "grad_norm": 1.018488835116828,
      "learning_rate": 8.922062177251893e-06,
      "loss": 0.53,
      "step": 4169
    },
    {
      "epoch": 0.22502833090497004,
      "grad_norm": 1.2949182836445468,
      "learning_rate": 8.921566715537502e-06,
      "loss": 0.6546,
      "step": 4170
    },
    {
      "epoch": 0.22508229453348444,
      "grad_norm": 1.1440041555810876,
      "learning_rate": 8.921071155483827e-06,
      "loss": 0.5664,
      "step": 4171
    },
    {
      "epoch": 0.2251362581619988,
      "grad_norm": 0.9680233122609794,
      "learning_rate": 8.920575497105111e-06,
      "loss": 0.3881,
      "step": 4172
    },
    {
      "epoch": 0.2251902217905132,
      "grad_norm": 0.9276079948297263,
      "learning_rate": 8.920079740415598e-06,
      "loss": 0.4471,
      "step": 4173
    },
    {
      "epoch": 0.22524418541902758,
      "grad_norm": 1.1584316767595777,
      "learning_rate": 8.919583885429537e-06,
      "loss": 0.5973,
      "step": 4174
    },
    {
      "epoch": 0.22529814904754195,
      "grad_norm": 1.0531476263513175,
      "learning_rate": 8.919087932161182e-06,
      "loss": 0.4471,
      "step": 4175
    },
    {
      "epoch": 0.22535211267605634,
      "grad_norm": 1.1209996820580306,
      "learning_rate": 8.918591880624783e-06,
      "loss": 0.6233,
      "step": 4176
    },
    {
      "epoch": 0.2254060763045707,
      "grad_norm": 0.9819268238797572,
      "learning_rate": 8.9180957308346e-06,
      "loss": 0.4733,
      "step": 4177
    },
    {
      "epoch": 0.2254600399330851,
      "grad_norm": 0.970423572480629,
      "learning_rate": 8.917599482804892e-06,
      "loss": 0.4026,
      "step": 4178
    },
    {
      "epoch": 0.22551400356159948,
      "grad_norm": 0.9536961469082397,
      "learning_rate": 8.91710313654992e-06,
      "loss": 0.4697,
      "step": 4179
    },
    {
      "epoch": 0.22556796719011385,
      "grad_norm": 1.0970101624574227,
      "learning_rate": 8.916606692083953e-06,
      "loss": 0.554,
      "step": 4180
    },
    {
      "epoch": 0.22562193081862825,
      "grad_norm": 1.2337050362657989,
      "learning_rate": 8.916110149421257e-06,
      "loss": 0.5846,
      "step": 4181
    },
    {
      "epoch": 0.22567589444714262,
      "grad_norm": 0.9530236030275984,
      "learning_rate": 8.915613508576104e-06,
      "loss": 0.4349,
      "step": 4182
    },
    {
      "epoch": 0.22572985807565701,
      "grad_norm": 1.0512358300529536,
      "learning_rate": 8.915116769562768e-06,
      "loss": 0.5618,
      "step": 4183
    },
    {
      "epoch": 0.22578382170417138,
      "grad_norm": 0.7197642291378159,
      "learning_rate": 8.914619932395523e-06,
      "loss": 0.2788,
      "step": 4184
    },
    {
      "epoch": 0.22583778533268578,
      "grad_norm": 1.165800358170503,
      "learning_rate": 8.914122997088652e-06,
      "loss": 0.71,
      "step": 4185
    },
    {
      "epoch": 0.22589174896120015,
      "grad_norm": 1.1494566248230678,
      "learning_rate": 8.913625963656437e-06,
      "loss": 0.6532,
      "step": 4186
    },
    {
      "epoch": 0.22594571258971452,
      "grad_norm": 1.0426722431275415,
      "learning_rate": 8.913128832113164e-06,
      "loss": 0.4893,
      "step": 4187
    },
    {
      "epoch": 0.22599967621822892,
      "grad_norm": 0.9716829600960734,
      "learning_rate": 8.912631602473117e-06,
      "loss": 0.585,
      "step": 4188
    },
    {
      "epoch": 0.2260536398467433,
      "grad_norm": 0.8563201914162332,
      "learning_rate": 8.91213427475059e-06,
      "loss": 0.3973,
      "step": 4189
    },
    {
      "epoch": 0.22610760347525768,
      "grad_norm": 1.2341370433452616,
      "learning_rate": 8.911636848959876e-06,
      "loss": 0.6115,
      "step": 4190
    },
    {
      "epoch": 0.22616156710377205,
      "grad_norm": 1.1409364542777995,
      "learning_rate": 8.911139325115274e-06,
      "loss": 0.4954,
      "step": 4191
    },
    {
      "epoch": 0.22621553073228645,
      "grad_norm": 1.0698605080022052,
      "learning_rate": 8.910641703231079e-06,
      "loss": 0.5339,
      "step": 4192
    },
    {
      "epoch": 0.22626949436080082,
      "grad_norm": 0.9940865210791319,
      "learning_rate": 8.910143983321596e-06,
      "loss": 0.5827,
      "step": 4193
    },
    {
      "epoch": 0.2263234579893152,
      "grad_norm": 1.2022944062679384,
      "learning_rate": 8.909646165401128e-06,
      "loss": 0.5125,
      "step": 4194
    },
    {
      "epoch": 0.2263774216178296,
      "grad_norm": 1.1178263358501914,
      "learning_rate": 8.909148249483984e-06,
      "loss": 0.5641,
      "step": 4195
    },
    {
      "epoch": 0.22643138524634396,
      "grad_norm": 0.7443849723704626,
      "learning_rate": 8.908650235584477e-06,
      "loss": 0.2891,
      "step": 4196
    },
    {
      "epoch": 0.22648534887485836,
      "grad_norm": 1.040331495967735,
      "learning_rate": 8.908152123716917e-06,
      "loss": 0.4695,
      "step": 4197
    },
    {
      "epoch": 0.22653931250337273,
      "grad_norm": 1.0893988743850729,
      "learning_rate": 8.907653913895622e-06,
      "loss": 0.4992,
      "step": 4198
    },
    {
      "epoch": 0.2265932761318871,
      "grad_norm": 1.0320143366137184,
      "learning_rate": 8.907155606134907e-06,
      "loss": 0.5432,
      "step": 4199
    },
    {
      "epoch": 0.2266472397604015,
      "grad_norm": 1.0771158856277476,
      "learning_rate": 8.9066572004491e-06,
      "loss": 0.6623,
      "step": 4200
    },
    {
      "epoch": 0.22670120338891586,
      "grad_norm": 1.3125658385317989,
      "learning_rate": 8.906158696852523e-06,
      "loss": 0.5609,
      "step": 4201
    },
    {
      "epoch": 0.22675516701743026,
      "grad_norm": 0.7501022545065628,
      "learning_rate": 8.905660095359502e-06,
      "loss": 0.3314,
      "step": 4202
    },
    {
      "epoch": 0.22680913064594463,
      "grad_norm": 1.0513668045645406,
      "learning_rate": 8.90516139598437e-06,
      "loss": 0.38,
      "step": 4203
    },
    {
      "epoch": 0.22686309427445903,
      "grad_norm": 1.1045705465964784,
      "learning_rate": 8.904662598741457e-06,
      "loss": 0.4299,
      "step": 4204
    },
    {
      "epoch": 0.2269170579029734,
      "grad_norm": 1.0976583189423483,
      "learning_rate": 8.904163703645102e-06,
      "loss": 0.6709,
      "step": 4205
    },
    {
      "epoch": 0.22697102153148777,
      "grad_norm": 0.9951642942765463,
      "learning_rate": 8.903664710709643e-06,
      "loss": 0.5125,
      "step": 4206
    },
    {
      "epoch": 0.22702498516000216,
      "grad_norm": 0.9617935680681392,
      "learning_rate": 8.90316561994942e-06,
      "loss": 0.4374,
      "step": 4207
    },
    {
      "epoch": 0.22707894878851653,
      "grad_norm": 0.998701680744596,
      "learning_rate": 8.90266643137878e-06,
      "loss": 0.4762,
      "step": 4208
    },
    {
      "epoch": 0.22713291241703093,
      "grad_norm": 0.949788765590773,
      "learning_rate": 8.902167145012067e-06,
      "loss": 0.4375,
      "step": 4209
    },
    {
      "epoch": 0.2271868760455453,
      "grad_norm": 1.1922421756692756,
      "learning_rate": 8.901667760863633e-06,
      "loss": 0.5052,
      "step": 4210
    },
    {
      "epoch": 0.2272408396740597,
      "grad_norm": 1.0026072902164263,
      "learning_rate": 8.901168278947832e-06,
      "loss": 0.4409,
      "step": 4211
    },
    {
      "epoch": 0.22729480330257407,
      "grad_norm": 1.1952331852131484,
      "learning_rate": 8.900668699279015e-06,
      "loss": 0.4889,
      "step": 4212
    },
    {
      "epoch": 0.22734876693108844,
      "grad_norm": 0.955925820758912,
      "learning_rate": 8.900169021871544e-06,
      "loss": 0.4665,
      "step": 4213
    },
    {
      "epoch": 0.22740273055960283,
      "grad_norm": 1.1559229761283465,
      "learning_rate": 8.899669246739783e-06,
      "loss": 0.7371,
      "step": 4214
    },
    {
      "epoch": 0.2274566941881172,
      "grad_norm": 1.1060319091781599,
      "learning_rate": 8.899169373898091e-06,
      "loss": 0.5395,
      "step": 4215
    },
    {
      "epoch": 0.2275106578166316,
      "grad_norm": 1.0297372939628304,
      "learning_rate": 8.898669403360835e-06,
      "loss": 0.4694,
      "step": 4216
    },
    {
      "epoch": 0.22756462144514597,
      "grad_norm": 0.913239694210688,
      "learning_rate": 8.898169335142388e-06,
      "loss": 0.4383,
      "step": 4217
    },
    {
      "epoch": 0.22761858507366034,
      "grad_norm": 1.0180784321261644,
      "learning_rate": 8.89766916925712e-06,
      "loss": 0.4424,
      "step": 4218
    },
    {
      "epoch": 0.22767254870217474,
      "grad_norm": 0.8786052806762851,
      "learning_rate": 8.897168905719406e-06,
      "loss": 0.4493,
      "step": 4219
    },
    {
      "epoch": 0.2277265123306891,
      "grad_norm": 1.171917696377168,
      "learning_rate": 8.896668544543628e-06,
      "loss": 0.5056,
      "step": 4220
    },
    {
      "epoch": 0.2277804759592035,
      "grad_norm": 1.1776391001404922,
      "learning_rate": 8.89616808574416e-06,
      "loss": 0.5048,
      "step": 4221
    },
    {
      "epoch": 0.22783443958771787,
      "grad_norm": 1.012316150941073,
      "learning_rate": 8.895667529335393e-06,
      "loss": 0.4847,
      "step": 4222
    },
    {
      "epoch": 0.22788840321623227,
      "grad_norm": 0.9242215709219594,
      "learning_rate": 8.89516687533171e-06,
      "loss": 0.3184,
      "step": 4223
    },
    {
      "epoch": 0.22794236684474664,
      "grad_norm": 1.0644696411350414,
      "learning_rate": 8.894666123747497e-06,
      "loss": 0.5301,
      "step": 4224
    },
    {
      "epoch": 0.227996330473261,
      "grad_norm": 1.2414733290516247,
      "learning_rate": 8.894165274597153e-06,
      "loss": 0.5959,
      "step": 4225
    },
    {
      "epoch": 0.2280502941017754,
      "grad_norm": 1.2311808232261539,
      "learning_rate": 8.893664327895067e-06,
      "loss": 0.5504,
      "step": 4226
    },
    {
      "epoch": 0.22810425773028978,
      "grad_norm": 1.1023302476738075,
      "learning_rate": 8.893163283655642e-06,
      "loss": 0.4164,
      "step": 4227
    },
    {
      "epoch": 0.22815822135880418,
      "grad_norm": 1.1397767435811703,
      "learning_rate": 8.892662141893274e-06,
      "loss": 0.6805,
      "step": 4228
    },
    {
      "epoch": 0.22821218498731854,
      "grad_norm": 1.003397468731344,
      "learning_rate": 8.892160902622368e-06,
      "loss": 0.5144,
      "step": 4229
    },
    {
      "epoch": 0.22826614861583294,
      "grad_norm": 1.0697723309582954,
      "learning_rate": 8.891659565857331e-06,
      "loss": 0.4131,
      "step": 4230
    },
    {
      "epoch": 0.2283201122443473,
      "grad_norm": 1.0861622113310254,
      "learning_rate": 8.891158131612568e-06,
      "loss": 0.4786,
      "step": 4231
    },
    {
      "epoch": 0.22837407587286168,
      "grad_norm": 0.9903712633444866,
      "learning_rate": 8.890656599902497e-06,
      "loss": 0.4743,
      "step": 4232
    },
    {
      "epoch": 0.22842803950137608,
      "grad_norm": 1.0875230942155125,
      "learning_rate": 8.890154970741527e-06,
      "loss": 0.5137,
      "step": 4233
    },
    {
      "epoch": 0.22848200312989045,
      "grad_norm": 0.7760738260224667,
      "learning_rate": 8.88965324414408e-06,
      "loss": 0.2923,
      "step": 4234
    },
    {
      "epoch": 0.22853596675840485,
      "grad_norm": 1.1025028646817099,
      "learning_rate": 8.889151420124573e-06,
      "loss": 0.5875,
      "step": 4235
    },
    {
      "epoch": 0.22858993038691922,
      "grad_norm": 0.9491971020061925,
      "learning_rate": 8.88864949869743e-06,
      "loss": 0.5046,
      "step": 4236
    },
    {
      "epoch": 0.22864389401543359,
      "grad_norm": 0.8726322996835082,
      "learning_rate": 8.888147479877076e-06,
      "loss": 0.3897,
      "step": 4237
    },
    {
      "epoch": 0.22869785764394798,
      "grad_norm": 1.0318043402707135,
      "learning_rate": 8.887645363677939e-06,
      "loss": 0.4022,
      "step": 4238
    },
    {
      "epoch": 0.22875182127246235,
      "grad_norm": 0.9883519579100625,
      "learning_rate": 8.887143150114453e-06,
      "loss": 0.4923,
      "step": 4239
    },
    {
      "epoch": 0.22880578490097675,
      "grad_norm": 0.8780799273560774,
      "learning_rate": 8.88664083920105e-06,
      "loss": 0.4429,
      "step": 4240
    },
    {
      "epoch": 0.22885974852949112,
      "grad_norm": 1.0400042114873702,
      "learning_rate": 8.886138430952166e-06,
      "loss": 0.3872,
      "step": 4241
    },
    {
      "epoch": 0.22891371215800552,
      "grad_norm": 1.0524801245914936,
      "learning_rate": 8.885635925382244e-06,
      "loss": 0.4657,
      "step": 4242
    },
    {
      "epoch": 0.2289676757865199,
      "grad_norm": 0.8991194025114374,
      "learning_rate": 8.885133322505724e-06,
      "loss": 0.5374,
      "step": 4243
    },
    {
      "epoch": 0.22902163941503426,
      "grad_norm": 0.9400100698539551,
      "learning_rate": 8.88463062233705e-06,
      "loss": 0.3829,
      "step": 4244
    },
    {
      "epoch": 0.22907560304354865,
      "grad_norm": 0.9265237855648651,
      "learning_rate": 8.884127824890673e-06,
      "loss": 0.3829,
      "step": 4245
    },
    {
      "epoch": 0.22912956667206302,
      "grad_norm": 1.1374307795545215,
      "learning_rate": 8.883624930181045e-06,
      "loss": 0.5186,
      "step": 4246
    },
    {
      "epoch": 0.22918353030057742,
      "grad_norm": 1.3715941049808282,
      "learning_rate": 8.883121938222614e-06,
      "loss": 0.6195,
      "step": 4247
    },
    {
      "epoch": 0.2292374939290918,
      "grad_norm": 0.962725685851421,
      "learning_rate": 8.882618849029844e-06,
      "loss": 0.4638,
      "step": 4248
    },
    {
      "epoch": 0.22929145755760616,
      "grad_norm": 0.9344335651635133,
      "learning_rate": 8.882115662617188e-06,
      "loss": 0.4889,
      "step": 4249
    },
    {
      "epoch": 0.22934542118612056,
      "grad_norm": 1.21679266699377,
      "learning_rate": 8.88161237899911e-06,
      "loss": 0.5355,
      "step": 4250
    },
    {
      "epoch": 0.22939938481463493,
      "grad_norm": 0.952017146765123,
      "learning_rate": 8.881108998190077e-06,
      "loss": 0.523,
      "step": 4251
    },
    {
      "epoch": 0.22945334844314932,
      "grad_norm": 1.053957002530713,
      "learning_rate": 8.880605520204554e-06,
      "loss": 0.4929,
      "step": 4252
    },
    {
      "epoch": 0.2295073120716637,
      "grad_norm": 1.0988128107147797,
      "learning_rate": 8.880101945057012e-06,
      "loss": 0.452,
      "step": 4253
    },
    {
      "epoch": 0.2295612757001781,
      "grad_norm": 1.1092556005888314,
      "learning_rate": 8.879598272761927e-06,
      "loss": 0.5861,
      "step": 4254
    },
    {
      "epoch": 0.22961523932869246,
      "grad_norm": 1.184420475611664,
      "learning_rate": 8.879094503333772e-06,
      "loss": 0.5513,
      "step": 4255
    },
    {
      "epoch": 0.22966920295720683,
      "grad_norm": 1.3435321256644301,
      "learning_rate": 8.878590636787024e-06,
      "loss": 0.8189,
      "step": 4256
    },
    {
      "epoch": 0.22972316658572123,
      "grad_norm": 0.9108882913517538,
      "learning_rate": 8.878086673136169e-06,
      "loss": 0.3758,
      "step": 4257
    },
    {
      "epoch": 0.2297771302142356,
      "grad_norm": 1.1369536192820615,
      "learning_rate": 8.87758261239569e-06,
      "loss": 0.493,
      "step": 4258
    },
    {
      "epoch": 0.22983109384275,
      "grad_norm": 0.8503592174469613,
      "learning_rate": 8.877078454580075e-06,
      "loss": 0.3754,
      "step": 4259
    },
    {
      "epoch": 0.22988505747126436,
      "grad_norm": 0.8446263629628106,
      "learning_rate": 8.87657419970381e-06,
      "loss": 0.3653,
      "step": 4260
    },
    {
      "epoch": 0.22993902109977876,
      "grad_norm": 1.209461872833154,
      "learning_rate": 8.876069847781394e-06,
      "loss": 0.7052,
      "step": 4261
    },
    {
      "epoch": 0.22999298472829313,
      "grad_norm": 1.356339812298453,
      "learning_rate": 8.875565398827318e-06,
      "loss": 0.5134,
      "step": 4262
    },
    {
      "epoch": 0.2300469483568075,
      "grad_norm": 1.1153141163887883,
      "learning_rate": 8.875060852856082e-06,
      "loss": 0.471,
      "step": 4263
    },
    {
      "epoch": 0.2301009119853219,
      "grad_norm": 0.9865368783059812,
      "learning_rate": 8.874556209882185e-06,
      "loss": 0.3982,
      "step": 4264
    },
    {
      "epoch": 0.23015487561383627,
      "grad_norm": 0.9855722530215885,
      "learning_rate": 8.874051469920135e-06,
      "loss": 0.525,
      "step": 4265
    },
    {
      "epoch": 0.23020883924235067,
      "grad_norm": 1.0158313877708975,
      "learning_rate": 8.873546632984435e-06,
      "loss": 0.4927,
      "step": 4266
    },
    {
      "epoch": 0.23026280287086504,
      "grad_norm": 0.9871122763443364,
      "learning_rate": 8.873041699089597e-06,
      "loss": 0.3966,
      "step": 4267
    },
    {
      "epoch": 0.2303167664993794,
      "grad_norm": 0.91906101940436,
      "learning_rate": 8.872536668250132e-06,
      "loss": 0.488,
      "step": 4268
    },
    {
      "epoch": 0.2303707301278938,
      "grad_norm": 0.8603050805399844,
      "learning_rate": 8.872031540480557e-06,
      "loss": 0.4352,
      "step": 4269
    },
    {
      "epoch": 0.23042469375640817,
      "grad_norm": 1.0133474465620715,
      "learning_rate": 8.871526315795386e-06,
      "loss": 0.4464,
      "step": 4270
    },
    {
      "epoch": 0.23047865738492257,
      "grad_norm": 1.111747218373013,
      "learning_rate": 8.871020994209143e-06,
      "loss": 0.4935,
      "step": 4271
    },
    {
      "epoch": 0.23053262101343694,
      "grad_norm": 1.0149561846708304,
      "learning_rate": 8.87051557573635e-06,
      "loss": 0.5624,
      "step": 4272
    },
    {
      "epoch": 0.23058658464195134,
      "grad_norm": 0.9442732151882096,
      "learning_rate": 8.870010060391534e-06,
      "loss": 0.4153,
      "step": 4273
    },
    {
      "epoch": 0.2306405482704657,
      "grad_norm": 1.0096363296590476,
      "learning_rate": 8.869504448189223e-06,
      "loss": 0.495,
      "step": 4274
    },
    {
      "epoch": 0.23069451189898008,
      "grad_norm": 0.90175713182598,
      "learning_rate": 8.86899873914395e-06,
      "loss": 0.4138,
      "step": 4275
    },
    {
      "epoch": 0.23074847552749447,
      "grad_norm": 1.1139798751347338,
      "learning_rate": 8.86849293327025e-06,
      "loss": 0.5741,
      "step": 4276
    },
    {
      "epoch": 0.23080243915600884,
      "grad_norm": 1.108189972750971,
      "learning_rate": 8.86798703058266e-06,
      "loss": 0.549,
      "step": 4277
    },
    {
      "epoch": 0.23085640278452324,
      "grad_norm": 0.9335382908852994,
      "learning_rate": 8.867481031095719e-06,
      "loss": 0.5081,
      "step": 4278
    },
    {
      "epoch": 0.2309103664130376,
      "grad_norm": 1.0813560804343791,
      "learning_rate": 8.866974934823969e-06,
      "loss": 0.4865,
      "step": 4279
    },
    {
      "epoch": 0.230964330041552,
      "grad_norm": 1.0954516980452589,
      "learning_rate": 8.86646874178196e-06,
      "loss": 0.5032,
      "step": 4280
    },
    {
      "epoch": 0.23101829367006638,
      "grad_norm": 0.8311954357144007,
      "learning_rate": 8.865962451984235e-06,
      "loss": 0.3464,
      "step": 4281
    },
    {
      "epoch": 0.23107225729858075,
      "grad_norm": 0.8966321243707828,
      "learning_rate": 8.865456065445348e-06,
      "loss": 0.3913,
      "step": 4282
    },
    {
      "epoch": 0.23112622092709514,
      "grad_norm": 0.9846125816863257,
      "learning_rate": 8.864949582179854e-06,
      "loss": 0.4771,
      "step": 4283
    },
    {
      "epoch": 0.2311801845556095,
      "grad_norm": 0.8726933707142843,
      "learning_rate": 8.86444300220231e-06,
      "loss": 0.5076,
      "step": 4284
    },
    {
      "epoch": 0.2312341481841239,
      "grad_norm": 1.3004201856524287,
      "learning_rate": 8.863936325527274e-06,
      "loss": 0.5173,
      "step": 4285
    },
    {
      "epoch": 0.23128811181263828,
      "grad_norm": 1.0130975329242455,
      "learning_rate": 8.86342955216931e-06,
      "loss": 0.5646,
      "step": 4286
    },
    {
      "epoch": 0.23134207544115265,
      "grad_norm": 1.030676721432551,
      "learning_rate": 8.862922682142981e-06,
      "loss": 0.4488,
      "step": 4287
    },
    {
      "epoch": 0.23139603906966705,
      "grad_norm": 1.008906818299179,
      "learning_rate": 8.862415715462855e-06,
      "loss": 0.4991,
      "step": 4288
    },
    {
      "epoch": 0.23145000269818142,
      "grad_norm": 1.0543323730433567,
      "learning_rate": 8.861908652143506e-06,
      "loss": 0.5383,
      "step": 4289
    },
    {
      "epoch": 0.23150396632669581,
      "grad_norm": 1.1115893777788908,
      "learning_rate": 8.861401492199505e-06,
      "loss": 0.415,
      "step": 4290
    },
    {
      "epoch": 0.23155792995521018,
      "grad_norm": 0.8637530526451926,
      "learning_rate": 8.860894235645427e-06,
      "loss": 0.3965,
      "step": 4291
    },
    {
      "epoch": 0.23161189358372458,
      "grad_norm": 1.1014743917742287,
      "learning_rate": 8.860386882495854e-06,
      "loss": 0.6244,
      "step": 4292
    },
    {
      "epoch": 0.23166585721223895,
      "grad_norm": 0.8478111687240644,
      "learning_rate": 8.859879432765366e-06,
      "loss": 0.3925,
      "step": 4293
    },
    {
      "epoch": 0.23171982084075332,
      "grad_norm": 1.08294437995867,
      "learning_rate": 8.859371886468549e-06,
      "loss": 0.5316,
      "step": 4294
    },
    {
      "epoch": 0.23177378446926772,
      "grad_norm": 0.9591093469183052,
      "learning_rate": 8.85886424361999e-06,
      "loss": 0.3865,
      "step": 4295
    },
    {
      "epoch": 0.2318277480977821,
      "grad_norm": 0.9580500099307673,
      "learning_rate": 8.858356504234278e-06,
      "loss": 0.4226,
      "step": 4296
    },
    {
      "epoch": 0.23188171172629649,
      "grad_norm": 1.2238897863570526,
      "learning_rate": 8.857848668326005e-06,
      "loss": 0.6211,
      "step": 4297
    },
    {
      "epoch": 0.23193567535481086,
      "grad_norm": 1.15164731754067,
      "learning_rate": 8.857340735909772e-06,
      "loss": 0.4269,
      "step": 4298
    },
    {
      "epoch": 0.23198963898332525,
      "grad_norm": 0.9037249263569062,
      "learning_rate": 8.856832707000172e-06,
      "loss": 0.3706,
      "step": 4299
    },
    {
      "epoch": 0.23204360261183962,
      "grad_norm": 1.1237870327045316,
      "learning_rate": 8.856324581611808e-06,
      "loss": 0.421,
      "step": 4300
    },
    {
      "epoch": 0.232097566240354,
      "grad_norm": 0.8781357554467456,
      "learning_rate": 8.855816359759283e-06,
      "loss": 0.3878,
      "step": 4301
    },
    {
      "epoch": 0.2321515298688684,
      "grad_norm": 0.8677405088798672,
      "learning_rate": 8.855308041457208e-06,
      "loss": 0.3667,
      "step": 4302
    },
    {
      "epoch": 0.23220549349738276,
      "grad_norm": 1.1553070055635584,
      "learning_rate": 8.854799626720187e-06,
      "loss": 0.6579,
      "step": 4303
    },
    {
      "epoch": 0.23225945712589716,
      "grad_norm": 0.9748291144533766,
      "learning_rate": 8.854291115562837e-06,
      "loss": 0.4439,
      "step": 4304
    },
    {
      "epoch": 0.23231342075441153,
      "grad_norm": 1.029576453505428,
      "learning_rate": 8.853782507999769e-06,
      "loss": 0.5533,
      "step": 4305
    },
    {
      "epoch": 0.2323673843829259,
      "grad_norm": 1.19724482777032,
      "learning_rate": 8.853273804045605e-06,
      "loss": 0.4949,
      "step": 4306
    },
    {
      "epoch": 0.2324213480114403,
      "grad_norm": 1.035621633122897,
      "learning_rate": 8.852765003714963e-06,
      "loss": 0.495,
      "step": 4307
    },
    {
      "epoch": 0.23247531163995466,
      "grad_norm": 1.1270690519523168,
      "learning_rate": 8.852256107022465e-06,
      "loss": 0.4842,
      "step": 4308
    },
    {
      "epoch": 0.23252927526846906,
      "grad_norm": 1.0166842657456197,
      "learning_rate": 8.85174711398274e-06,
      "loss": 0.4985,
      "step": 4309
    },
    {
      "epoch": 0.23258323889698343,
      "grad_norm": 1.2902067126071957,
      "learning_rate": 8.851238024610418e-06,
      "loss": 0.5769,
      "step": 4310
    },
    {
      "epoch": 0.23263720252549783,
      "grad_norm": 0.9294636511265644,
      "learning_rate": 8.85072883892013e-06,
      "loss": 0.5055,
      "step": 4311
    },
    {
      "epoch": 0.2326911661540122,
      "grad_norm": 1.0175295390551642,
      "learning_rate": 8.850219556926506e-06,
      "loss": 0.5916,
      "step": 4312
    },
    {
      "epoch": 0.23274512978252657,
      "grad_norm": 0.960058267579395,
      "learning_rate": 8.84971017864419e-06,
      "loss": 0.4849,
      "step": 4313
    },
    {
      "epoch": 0.23279909341104096,
      "grad_norm": 0.8607329467182275,
      "learning_rate": 8.849200704087818e-06,
      "loss": 0.453,
      "step": 4314
    },
    {
      "epoch": 0.23285305703955533,
      "grad_norm": 1.1464748530021955,
      "learning_rate": 8.848691133272033e-06,
      "loss": 0.6183,
      "step": 4315
    },
    {
      "epoch": 0.23290702066806973,
      "grad_norm": 1.066488444254917,
      "learning_rate": 8.84818146621148e-06,
      "loss": 0.4407,
      "step": 4316
    },
    {
      "epoch": 0.2329609842965841,
      "grad_norm": 1.4024568753450393,
      "learning_rate": 8.84767170292081e-06,
      "loss": 0.7498,
      "step": 4317
    },
    {
      "epoch": 0.23301494792509847,
      "grad_norm": 0.9403023759037108,
      "learning_rate": 8.847161843414674e-06,
      "loss": 0.4197,
      "step": 4318
    },
    {
      "epoch": 0.23306891155361287,
      "grad_norm": 1.1335302063864279,
      "learning_rate": 8.84665188770772e-06,
      "loss": 0.4275,
      "step": 4319
    },
    {
      "epoch": 0.23312287518212724,
      "grad_norm": 1.1692816942166677,
      "learning_rate": 8.846141835814613e-06,
      "loss": 0.4992,
      "step": 4320
    },
    {
      "epoch": 0.23317683881064163,
      "grad_norm": 0.9359651896370513,
      "learning_rate": 8.845631687750008e-06,
      "loss": 0.5537,
      "step": 4321
    },
    {
      "epoch": 0.233230802439156,
      "grad_norm": 1.0666729337528817,
      "learning_rate": 8.845121443528567e-06,
      "loss": 0.5709,
      "step": 4322
    },
    {
      "epoch": 0.2332847660676704,
      "grad_norm": 0.9650158325398492,
      "learning_rate": 8.844611103164957e-06,
      "loss": 0.3949,
      "step": 4323
    },
    {
      "epoch": 0.23333872969618477,
      "grad_norm": 0.8499119303607209,
      "learning_rate": 8.844100666673845e-06,
      "loss": 0.3177,
      "step": 4324
    },
    {
      "epoch": 0.23339269332469914,
      "grad_norm": 0.8397536315262232,
      "learning_rate": 8.8435901340699e-06,
      "loss": 0.4382,
      "step": 4325
    },
    {
      "epoch": 0.23344665695321354,
      "grad_norm": 1.1845849999537605,
      "learning_rate": 8.843079505367794e-06,
      "loss": 0.5236,
      "step": 4326
    },
    {
      "epoch": 0.2335006205817279,
      "grad_norm": 0.8372986391265121,
      "learning_rate": 8.842568780582206e-06,
      "loss": 0.4106,
      "step": 4327
    },
    {
      "epoch": 0.2335545842102423,
      "grad_norm": 1.259611438903873,
      "learning_rate": 8.842057959727815e-06,
      "loss": 0.5597,
      "step": 4328
    },
    {
      "epoch": 0.23360854783875667,
      "grad_norm": 1.0067055960977618,
      "learning_rate": 8.841547042819301e-06,
      "loss": 0.4928,
      "step": 4329
    },
    {
      "epoch": 0.23366251146727107,
      "grad_norm": 0.8551285691669201,
      "learning_rate": 8.84103602987135e-06,
      "loss": 0.4835,
      "step": 4330
    },
    {
      "epoch": 0.23371647509578544,
      "grad_norm": 1.1137524682225863,
      "learning_rate": 8.840524920898645e-06,
      "loss": 0.6417,
      "step": 4331
    },
    {
      "epoch": 0.2337704387242998,
      "grad_norm": 1.191704192076781,
      "learning_rate": 8.84001371591588e-06,
      "loss": 0.6284,
      "step": 4332
    },
    {
      "epoch": 0.2338244023528142,
      "grad_norm": 1.208408802379505,
      "learning_rate": 8.839502414937745e-06,
      "loss": 0.5869,
      "step": 4333
    },
    {
      "epoch": 0.23387836598132858,
      "grad_norm": 1.1256126131622404,
      "learning_rate": 8.838991017978937e-06,
      "loss": 0.5215,
      "step": 4334
    },
    {
      "epoch": 0.23393232960984298,
      "grad_norm": 1.3315455106979754,
      "learning_rate": 8.838479525054153e-06,
      "loss": 0.5281,
      "step": 4335
    },
    {
      "epoch": 0.23398629323835735,
      "grad_norm": 0.9910674903674576,
      "learning_rate": 8.837967936178095e-06,
      "loss": 0.4993,
      "step": 4336
    },
    {
      "epoch": 0.23404025686687172,
      "grad_norm": 1.12609720932742,
      "learning_rate": 8.837456251365466e-06,
      "loss": 0.6089,
      "step": 4337
    },
    {
      "epoch": 0.2340942204953861,
      "grad_norm": 0.9407906525347404,
      "learning_rate": 8.83694447063097e-06,
      "loss": 0.3663,
      "step": 4338
    },
    {
      "epoch": 0.23414818412390048,
      "grad_norm": 1.2233821566603786,
      "learning_rate": 8.83643259398932e-06,
      "loss": 0.6363,
      "step": 4339
    },
    {
      "epoch": 0.23420214775241488,
      "grad_norm": 0.9711851058494575,
      "learning_rate": 8.835920621455225e-06,
      "loss": 0.4572,
      "step": 4340
    },
    {
      "epoch": 0.23425611138092925,
      "grad_norm": 0.8615266852364176,
      "learning_rate": 8.8354085530434e-06,
      "loss": 0.4759,
      "step": 4341
    },
    {
      "epoch": 0.23431007500944365,
      "grad_norm": 1.1759350800963684,
      "learning_rate": 8.834896388768565e-06,
      "loss": 0.574,
      "step": 4342
    },
    {
      "epoch": 0.23436403863795802,
      "grad_norm": 1.184254392667908,
      "learning_rate": 8.834384128645439e-06,
      "loss": 0.5895,
      "step": 4343
    },
    {
      "epoch": 0.2344180022664724,
      "grad_norm": 1.132207594382214,
      "learning_rate": 8.833871772688741e-06,
      "loss": 0.5023,
      "step": 4344
    },
    {
      "epoch": 0.23447196589498678,
      "grad_norm": 0.9776216068914209,
      "learning_rate": 8.8333593209132e-06,
      "loss": 0.4647,
      "step": 4345
    },
    {
      "epoch": 0.23452592952350115,
      "grad_norm": 0.8597875662842117,
      "learning_rate": 8.832846773333544e-06,
      "loss": 0.383,
      "step": 4346
    },
    {
      "epoch": 0.23457989315201555,
      "grad_norm": 1.059448780233623,
      "learning_rate": 8.832334129964505e-06,
      "loss": 0.4694,
      "step": 4347
    },
    {
      "epoch": 0.23463385678052992,
      "grad_norm": 1.0831799127979564,
      "learning_rate": 8.831821390820815e-06,
      "loss": 0.4467,
      "step": 4348
    },
    {
      "epoch": 0.23468782040904432,
      "grad_norm": 1.0980445228647002,
      "learning_rate": 8.831308555917213e-06,
      "loss": 0.5461,
      "step": 4349
    },
    {
      "epoch": 0.2347417840375587,
      "grad_norm": 0.8204431415290188,
      "learning_rate": 8.830795625268437e-06,
      "loss": 0.3559,
      "step": 4350
    },
    {
      "epoch": 0.23479574766607306,
      "grad_norm": 0.9757006411880733,
      "learning_rate": 8.83028259888923e-06,
      "loss": 0.5887,
      "step": 4351
    },
    {
      "epoch": 0.23484971129458745,
      "grad_norm": 1.0592479496790406,
      "learning_rate": 8.829769476794334e-06,
      "loss": 0.4722,
      "step": 4352
    },
    {
      "epoch": 0.23490367492310182,
      "grad_norm": 0.9104820507918483,
      "learning_rate": 8.8292562589985e-06,
      "loss": 0.3461,
      "step": 4353
    },
    {
      "epoch": 0.23495763855161622,
      "grad_norm": 0.9501502804441881,
      "learning_rate": 8.828742945516478e-06,
      "loss": 0.3982,
      "step": 4354
    },
    {
      "epoch": 0.2350116021801306,
      "grad_norm": 1.0975843306868056,
      "learning_rate": 8.828229536363018e-06,
      "loss": 0.4222,
      "step": 4355
    },
    {
      "epoch": 0.23506556580864496,
      "grad_norm": 1.0185578537571258,
      "learning_rate": 8.82771603155288e-06,
      "loss": 0.4954,
      "step": 4356
    },
    {
      "epoch": 0.23511952943715936,
      "grad_norm": 1.1985899142667398,
      "learning_rate": 8.82720243110082e-06,
      "loss": 0.7365,
      "step": 4357
    },
    {
      "epoch": 0.23517349306567373,
      "grad_norm": 1.038119897581883,
      "learning_rate": 8.8266887350216e-06,
      "loss": 0.4322,
      "step": 4358
    },
    {
      "epoch": 0.23522745669418813,
      "grad_norm": 1.1311527530084275,
      "learning_rate": 8.826174943329986e-06,
      "loss": 0.6467,
      "step": 4359
    },
    {
      "epoch": 0.2352814203227025,
      "grad_norm": 1.2843945716059986,
      "learning_rate": 8.82566105604074e-06,
      "loss": 0.6104,
      "step": 4360
    },
    {
      "epoch": 0.2353353839512169,
      "grad_norm": 1.0773562189818182,
      "learning_rate": 8.825147073168639e-06,
      "loss": 0.506,
      "step": 4361
    },
    {
      "epoch": 0.23538934757973126,
      "grad_norm": 1.046791650762785,
      "learning_rate": 8.824632994728449e-06,
      "loss": 0.5448,
      "step": 4362
    },
    {
      "epoch": 0.23544331120824563,
      "grad_norm": 1.23133381745459,
      "learning_rate": 8.824118820734948e-06,
      "loss": 0.5516,
      "step": 4363
    },
    {
      "epoch": 0.23549727483676003,
      "grad_norm": 0.8857200985938232,
      "learning_rate": 8.823604551202912e-06,
      "loss": 0.3874,
      "step": 4364
    },
    {
      "epoch": 0.2355512384652744,
      "grad_norm": 1.0780822095268483,
      "learning_rate": 8.823090186147125e-06,
      "loss": 0.5135,
      "step": 4365
    },
    {
      "epoch": 0.2356052020937888,
      "grad_norm": 1.1562755635990478,
      "learning_rate": 8.822575725582367e-06,
      "loss": 0.4561,
      "step": 4366
    },
    {
      "epoch": 0.23565916572230317,
      "grad_norm": 0.8888984401583951,
      "learning_rate": 8.822061169523426e-06,
      "loss": 0.361,
      "step": 4367
    },
    {
      "epoch": 0.23571312935081754,
      "grad_norm": 1.15908386920342,
      "learning_rate": 8.82154651798509e-06,
      "loss": 0.6063,
      "step": 4368
    },
    {
      "epoch": 0.23576709297933193,
      "grad_norm": 1.0772012280512686,
      "learning_rate": 8.82103177098215e-06,
      "loss": 0.5387,
      "step": 4369
    },
    {
      "epoch": 0.2358210566078463,
      "grad_norm": 0.969742164426776,
      "learning_rate": 8.820516928529402e-06,
      "loss": 0.4737,
      "step": 4370
    },
    {
      "epoch": 0.2358750202363607,
      "grad_norm": 1.0789167882819684,
      "learning_rate": 8.820001990641644e-06,
      "loss": 0.5025,
      "step": 4371
    },
    {
      "epoch": 0.23592898386487507,
      "grad_norm": 0.7671262894132683,
      "learning_rate": 8.819486957333672e-06,
      "loss": 0.333,
      "step": 4372
    },
    {
      "epoch": 0.23598294749338947,
      "grad_norm": 1.0675713687322526,
      "learning_rate": 8.818971828620291e-06,
      "loss": 0.3698,
      "step": 4373
    },
    {
      "epoch": 0.23603691112190384,
      "grad_norm": 1.0901986807445843,
      "learning_rate": 8.818456604516305e-06,
      "loss": 0.5537,
      "step": 4374
    },
    {
      "epoch": 0.2360908747504182,
      "grad_norm": 1.1726107244228667,
      "learning_rate": 8.817941285036525e-06,
      "loss": 0.5997,
      "step": 4375
    },
    {
      "epoch": 0.2361448383789326,
      "grad_norm": 1.0352167424115302,
      "learning_rate": 8.81742587019576e-06,
      "loss": 0.4877,
      "step": 4376
    },
    {
      "epoch": 0.23619880200744697,
      "grad_norm": 0.9938193562936292,
      "learning_rate": 8.816910360008824e-06,
      "loss": 0.4993,
      "step": 4377
    },
    {
      "epoch": 0.23625276563596137,
      "grad_norm": 0.9389130512500483,
      "learning_rate": 8.816394754490531e-06,
      "loss": 0.3606,
      "step": 4378
    },
    {
      "epoch": 0.23630672926447574,
      "grad_norm": 0.9404745506794373,
      "learning_rate": 8.815879053655702e-06,
      "loss": 0.3905,
      "step": 4379
    },
    {
      "epoch": 0.23636069289299014,
      "grad_norm": 0.7739994838801398,
      "learning_rate": 8.81536325751916e-06,
      "loss": 0.3042,
      "step": 4380
    },
    {
      "epoch": 0.2364146565215045,
      "grad_norm": 0.797544898263942,
      "learning_rate": 8.814847366095728e-06,
      "loss": 0.4998,
      "step": 4381
    },
    {
      "epoch": 0.23646862015001888,
      "grad_norm": 0.9841069057124918,
      "learning_rate": 8.814331379400232e-06,
      "loss": 0.5548,
      "step": 4382
    },
    {
      "epoch": 0.23652258377853327,
      "grad_norm": 1.1407222679650029,
      "learning_rate": 8.813815297447505e-06,
      "loss": 0.5294,
      "step": 4383
    },
    {
      "epoch": 0.23657654740704764,
      "grad_norm": 1.191495804750151,
      "learning_rate": 8.813299120252376e-06,
      "loss": 0.6453,
      "step": 4384
    },
    {
      "epoch": 0.23663051103556204,
      "grad_norm": 1.150917484385976,
      "learning_rate": 8.812782847829682e-06,
      "loss": 0.5789,
      "step": 4385
    },
    {
      "epoch": 0.2366844746640764,
      "grad_norm": 1.223642259170739,
      "learning_rate": 8.812266480194263e-06,
      "loss": 0.6294,
      "step": 4386
    },
    {
      "epoch": 0.23673843829259078,
      "grad_norm": 0.8941445824384119,
      "learning_rate": 8.811750017360957e-06,
      "loss": 0.4205,
      "step": 4387
    },
    {
      "epoch": 0.23679240192110518,
      "grad_norm": 1.052619234823713,
      "learning_rate": 8.811233459344612e-06,
      "loss": 0.4018,
      "step": 4388
    },
    {
      "epoch": 0.23684636554961955,
      "grad_norm": 1.084935333204135,
      "learning_rate": 8.810716806160071e-06,
      "loss": 0.5192,
      "step": 4389
    },
    {
      "epoch": 0.23690032917813394,
      "grad_norm": 0.9903791009425426,
      "learning_rate": 8.810200057822183e-06,
      "loss": 0.5643,
      "step": 4390
    },
    {
      "epoch": 0.23695429280664831,
      "grad_norm": 1.1959885157959225,
      "learning_rate": 8.8096832143458e-06,
      "loss": 0.6143,
      "step": 4391
    },
    {
      "epoch": 0.2370082564351627,
      "grad_norm": 1.2361495594681793,
      "learning_rate": 8.809166275745778e-06,
      "loss": 0.4855,
      "step": 4392
    },
    {
      "epoch": 0.23706222006367708,
      "grad_norm": 1.4232033412954477,
      "learning_rate": 8.808649242036973e-06,
      "loss": 0.7707,
      "step": 4393
    },
    {
      "epoch": 0.23711618369219145,
      "grad_norm": 0.8590172392863666,
      "learning_rate": 8.808132113234246e-06,
      "loss": 0.366,
      "step": 4394
    },
    {
      "epoch": 0.23717014732070585,
      "grad_norm": 1.1060641175115506,
      "learning_rate": 8.80761488935246e-06,
      "loss": 0.5636,
      "step": 4395
    },
    {
      "epoch": 0.23722411094922022,
      "grad_norm": 1.4626189751077419,
      "learning_rate": 8.807097570406479e-06,
      "loss": 0.592,
      "step": 4396
    },
    {
      "epoch": 0.23727807457773462,
      "grad_norm": 0.9107326919099172,
      "learning_rate": 8.806580156411171e-06,
      "loss": 0.4456,
      "step": 4397
    },
    {
      "epoch": 0.23733203820624899,
      "grad_norm": 0.9756624810973178,
      "learning_rate": 8.80606264738141e-06,
      "loss": 0.5083,
      "step": 4398
    },
    {
      "epoch": 0.23738600183476338,
      "grad_norm": 1.041742857273947,
      "learning_rate": 8.805545043332068e-06,
      "loss": 0.4233,
      "step": 4399
    },
    {
      "epoch": 0.23743996546327775,
      "grad_norm": 1.0038686305036868,
      "learning_rate": 8.805027344278022e-06,
      "loss": 0.4605,
      "step": 4400
    },
    {
      "epoch": 0.23749392909179212,
      "grad_norm": 1.1723838648696772,
      "learning_rate": 8.80450955023415e-06,
      "loss": 0.7379,
      "step": 4401
    },
    {
      "epoch": 0.23754789272030652,
      "grad_norm": 1.0051713261276964,
      "learning_rate": 8.803991661215334e-06,
      "loss": 0.4267,
      "step": 4402
    },
    {
      "epoch": 0.2376018563488209,
      "grad_norm": 1.0785563032732313,
      "learning_rate": 8.803473677236461e-06,
      "loss": 0.585,
      "step": 4403
    },
    {
      "epoch": 0.2376558199773353,
      "grad_norm": 1.2915591474910588,
      "learning_rate": 8.802955598312416e-06,
      "loss": 0.5827,
      "step": 4404
    },
    {
      "epoch": 0.23770978360584966,
      "grad_norm": 0.9870151625822382,
      "learning_rate": 8.80243742445809e-06,
      "loss": 0.4676,
      "step": 4405
    },
    {
      "epoch": 0.23776374723436403,
      "grad_norm": 1.059162165622893,
      "learning_rate": 8.801919155688374e-06,
      "loss": 0.4648,
      "step": 4406
    },
    {
      "epoch": 0.23781771086287842,
      "grad_norm": 0.844387260639048,
      "learning_rate": 8.801400792018166e-06,
      "loss": 0.4754,
      "step": 4407
    },
    {
      "epoch": 0.2378716744913928,
      "grad_norm": 1.1646843032869052,
      "learning_rate": 8.800882333462361e-06,
      "loss": 0.4835,
      "step": 4408
    },
    {
      "epoch": 0.2379256381199072,
      "grad_norm": 1.082992591250399,
      "learning_rate": 8.800363780035865e-06,
      "loss": 0.6019,
      "step": 4409
    },
    {
      "epoch": 0.23797960174842156,
      "grad_norm": 1.0856228009803985,
      "learning_rate": 8.799845131753581e-06,
      "loss": 0.5902,
      "step": 4410
    },
    {
      "epoch": 0.23803356537693596,
      "grad_norm": 0.9463257680748529,
      "learning_rate": 8.799326388630409e-06,
      "loss": 0.3682,
      "step": 4411
    },
    {
      "epoch": 0.23808752900545033,
      "grad_norm": 1.231707749218174,
      "learning_rate": 8.798807550681266e-06,
      "loss": 0.634,
      "step": 4412
    },
    {
      "epoch": 0.2381414926339647,
      "grad_norm": 0.9902154080387326,
      "learning_rate": 8.798288617921061e-06,
      "loss": 0.5658,
      "step": 4413
    },
    {
      "epoch": 0.2381954562624791,
      "grad_norm": 0.956782362443882,
      "learning_rate": 8.797769590364706e-06,
      "loss": 0.5058,
      "step": 4414
    },
    {
      "epoch": 0.23824941989099346,
      "grad_norm": 0.9829207728761068,
      "learning_rate": 8.797250468027124e-06,
      "loss": 0.5784,
      "step": 4415
    },
    {
      "epoch": 0.23830338351950786,
      "grad_norm": 0.9314631121396891,
      "learning_rate": 8.796731250923229e-06,
      "loss": 0.4859,
      "step": 4416
    },
    {
      "epoch": 0.23835734714802223,
      "grad_norm": 1.1093704822387043,
      "learning_rate": 8.796211939067948e-06,
      "loss": 0.506,
      "step": 4417
    },
    {
      "epoch": 0.23841131077653663,
      "grad_norm": 1.118440513473199,
      "learning_rate": 8.795692532476206e-06,
      "loss": 0.5183,
      "step": 4418
    },
    {
      "epoch": 0.238465274405051,
      "grad_norm": 1.0125155068683966,
      "learning_rate": 8.79517303116293e-06,
      "loss": 0.5391,
      "step": 4419
    },
    {
      "epoch": 0.23851923803356537,
      "grad_norm": 1.0554181451704647,
      "learning_rate": 8.794653435143052e-06,
      "loss": 0.4988,
      "step": 4420
    },
    {
      "epoch": 0.23857320166207976,
      "grad_norm": 0.9053544133653306,
      "learning_rate": 8.794133744431504e-06,
      "loss": 0.5167,
      "step": 4421
    },
    {
      "epoch": 0.23862716529059413,
      "grad_norm": 1.2086001400434754,
      "learning_rate": 8.793613959043224e-06,
      "loss": 0.618,
      "step": 4422
    },
    {
      "epoch": 0.23868112891910853,
      "grad_norm": 1.2116479905549473,
      "learning_rate": 8.793094078993152e-06,
      "loss": 0.5234,
      "step": 4423
    },
    {
      "epoch": 0.2387350925476229,
      "grad_norm": 0.9732075073145292,
      "learning_rate": 8.792574104296227e-06,
      "loss": 0.4703,
      "step": 4424
    },
    {
      "epoch": 0.23878905617613727,
      "grad_norm": 1.0514097967596792,
      "learning_rate": 8.792054034967396e-06,
      "loss": 0.6827,
      "step": 4425
    },
    {
      "epoch": 0.23884301980465167,
      "grad_norm": 1.1053612680069402,
      "learning_rate": 8.791533871021607e-06,
      "loss": 0.5606,
      "step": 4426
    },
    {
      "epoch": 0.23889698343316604,
      "grad_norm": 1.0476676490188268,
      "learning_rate": 8.791013612473808e-06,
      "loss": 0.4679,
      "step": 4427
    },
    {
      "epoch": 0.23895094706168044,
      "grad_norm": 1.194465980594172,
      "learning_rate": 8.790493259338951e-06,
      "loss": 0.5016,
      "step": 4428
    },
    {
      "epoch": 0.2390049106901948,
      "grad_norm": 0.9391728881187966,
      "learning_rate": 8.789972811631995e-06,
      "loss": 0.5013,
      "step": 4429
    },
    {
      "epoch": 0.2390588743187092,
      "grad_norm": 1.2862266063235424,
      "learning_rate": 8.789452269367895e-06,
      "loss": 0.6144,
      "step": 4430
    },
    {
      "epoch": 0.23911283794722357,
      "grad_norm": 0.9844588027101957,
      "learning_rate": 8.788931632561614e-06,
      "loss": 0.4269,
      "step": 4431
    },
    {
      "epoch": 0.23916680157573794,
      "grad_norm": 1.1532586292136673,
      "learning_rate": 8.788410901228112e-06,
      "loss": 0.4794,
      "step": 4432
    },
    {
      "epoch": 0.23922076520425234,
      "grad_norm": 0.8392037506548712,
      "learning_rate": 8.787890075382362e-06,
      "loss": 0.307,
      "step": 4433
    },
    {
      "epoch": 0.2392747288327667,
      "grad_norm": 0.9301486269190737,
      "learning_rate": 8.787369155039327e-06,
      "loss": 0.4579,
      "step": 4434
    },
    {
      "epoch": 0.2393286924612811,
      "grad_norm": 1.0157799882379444,
      "learning_rate": 8.78684814021398e-06,
      "loss": 0.4332,
      "step": 4435
    },
    {
      "epoch": 0.23938265608979548,
      "grad_norm": 1.0209303728763566,
      "learning_rate": 8.786327030921297e-06,
      "loss": 0.6182,
      "step": 4436
    },
    {
      "epoch": 0.23943661971830985,
      "grad_norm": 0.8414072933796239,
      "learning_rate": 8.785805827176256e-06,
      "loss": 0.3736,
      "step": 4437
    },
    {
      "epoch": 0.23949058334682424,
      "grad_norm": 1.114723735839323,
      "learning_rate": 8.785284528993832e-06,
      "loss": 0.684,
      "step": 4438
    },
    {
      "epoch": 0.2395445469753386,
      "grad_norm": 1.2288926229374904,
      "learning_rate": 8.784763136389015e-06,
      "loss": 0.5597,
      "step": 4439
    },
    {
      "epoch": 0.239598510603853,
      "grad_norm": 1.1283941129174084,
      "learning_rate": 8.784241649376784e-06,
      "loss": 0.5406,
      "step": 4440
    },
    {
      "epoch": 0.23965247423236738,
      "grad_norm": 1.1032518288304136,
      "learning_rate": 8.78372006797213e-06,
      "loss": 0.501,
      "step": 4441
    },
    {
      "epoch": 0.23970643786088178,
      "grad_norm": 0.8231821667735801,
      "learning_rate": 8.783198392190043e-06,
      "loss": 0.4031,
      "step": 4442
    },
    {
      "epoch": 0.23976040148939615,
      "grad_norm": 0.9304145448977094,
      "learning_rate": 8.782676622045517e-06,
      "loss": 0.4174,
      "step": 4443
    },
    {
      "epoch": 0.23981436511791052,
      "grad_norm": 0.9457067954719116,
      "learning_rate": 8.78215475755355e-06,
      "loss": 0.3959,
      "step": 4444
    },
    {
      "epoch": 0.2398683287464249,
      "grad_norm": 0.9999599936551482,
      "learning_rate": 8.781632798729137e-06,
      "loss": 0.4854,
      "step": 4445
    },
    {
      "epoch": 0.23992229237493928,
      "grad_norm": 1.0319051381283428,
      "learning_rate": 8.781110745587281e-06,
      "loss": 0.5168,
      "step": 4446
    },
    {
      "epoch": 0.23997625600345368,
      "grad_norm": 1.0042353296128097,
      "learning_rate": 8.780588598142987e-06,
      "loss": 0.4266,
      "step": 4447
    },
    {
      "epoch": 0.24003021963196805,
      "grad_norm": 0.994443631715621,
      "learning_rate": 8.780066356411262e-06,
      "loss": 0.5309,
      "step": 4448
    },
    {
      "epoch": 0.24008418326048245,
      "grad_norm": 1.0916162812550392,
      "learning_rate": 8.779544020407116e-06,
      "loss": 0.4831,
      "step": 4449
    },
    {
      "epoch": 0.24013814688899682,
      "grad_norm": 1.0945821112302112,
      "learning_rate": 8.779021590145562e-06,
      "loss": 0.4311,
      "step": 4450
    },
    {
      "epoch": 0.2401921105175112,
      "grad_norm": 1.0857189120778294,
      "learning_rate": 8.778499065641614e-06,
      "loss": 0.6337,
      "step": 4451
    },
    {
      "epoch": 0.24024607414602558,
      "grad_norm": 0.8548613471329187,
      "learning_rate": 8.77797644691029e-06,
      "loss": 0.3708,
      "step": 4452
    },
    {
      "epoch": 0.24030003777453995,
      "grad_norm": 0.8836519630135827,
      "learning_rate": 8.777453733966609e-06,
      "loss": 0.4169,
      "step": 4453
    },
    {
      "epoch": 0.24035400140305435,
      "grad_norm": 1.0071023958163081,
      "learning_rate": 8.776930926825598e-06,
      "loss": 0.6848,
      "step": 4454
    },
    {
      "epoch": 0.24040796503156872,
      "grad_norm": 0.7260458908256008,
      "learning_rate": 8.776408025502281e-06,
      "loss": 0.3105,
      "step": 4455
    },
    {
      "epoch": 0.2404619286600831,
      "grad_norm": 0.8954039160270562,
      "learning_rate": 8.775885030011686e-06,
      "loss": 0.4436,
      "step": 4456
    },
    {
      "epoch": 0.2405158922885975,
      "grad_norm": 0.7670009622812443,
      "learning_rate": 8.775361940368845e-06,
      "loss": 0.2746,
      "step": 4457
    },
    {
      "epoch": 0.24056985591711186,
      "grad_norm": 1.176189635313819,
      "learning_rate": 8.774838756588794e-06,
      "loss": 0.5389,
      "step": 4458
    },
    {
      "epoch": 0.24062381954562626,
      "grad_norm": 0.9860281371505387,
      "learning_rate": 8.774315478686566e-06,
      "loss": 0.3606,
      "step": 4459
    },
    {
      "epoch": 0.24067778317414062,
      "grad_norm": 0.9569033458670732,
      "learning_rate": 8.773792106677203e-06,
      "loss": 0.4682,
      "step": 4460
    },
    {
      "epoch": 0.24073174680265502,
      "grad_norm": 0.852850415670999,
      "learning_rate": 8.773268640575748e-06,
      "loss": 0.3728,
      "step": 4461
    },
    {
      "epoch": 0.2407857104311694,
      "grad_norm": 0.9681582402641141,
      "learning_rate": 8.772745080397245e-06,
      "loss": 0.5061,
      "step": 4462
    },
    {
      "epoch": 0.24083967405968376,
      "grad_norm": 0.8825902441889901,
      "learning_rate": 8.772221426156741e-06,
      "loss": 0.5365,
      "step": 4463
    },
    {
      "epoch": 0.24089363768819816,
      "grad_norm": 1.0585990052253709,
      "learning_rate": 8.771697677869286e-06,
      "loss": 0.5712,
      "step": 4464
    },
    {
      "epoch": 0.24094760131671253,
      "grad_norm": 1.2143698230657496,
      "learning_rate": 8.771173835549936e-06,
      "loss": 0.6284,
      "step": 4465
    },
    {
      "epoch": 0.24100156494522693,
      "grad_norm": 0.9470806233844631,
      "learning_rate": 8.770649899213744e-06,
      "loss": 0.3699,
      "step": 4466
    },
    {
      "epoch": 0.2410555285737413,
      "grad_norm": 1.2449869724174918,
      "learning_rate": 8.770125868875768e-06,
      "loss": 0.5524,
      "step": 4467
    },
    {
      "epoch": 0.2411094922022557,
      "grad_norm": 1.1135492611520865,
      "learning_rate": 8.769601744551072e-06,
      "loss": 0.4846,
      "step": 4468
    },
    {
      "epoch": 0.24116345583077006,
      "grad_norm": 0.893999334963022,
      "learning_rate": 8.769077526254715e-06,
      "loss": 0.3236,
      "step": 4469
    },
    {
      "epoch": 0.24121741945928443,
      "grad_norm": 0.9806389429710375,
      "learning_rate": 8.768553214001769e-06,
      "loss": 0.3784,
      "step": 4470
    },
    {
      "epoch": 0.24127138308779883,
      "grad_norm": 0.971481123607927,
      "learning_rate": 8.7680288078073e-06,
      "loss": 0.5386,
      "step": 4471
    },
    {
      "epoch": 0.2413253467163132,
      "grad_norm": 0.8908321580481223,
      "learning_rate": 8.767504307686383e-06,
      "loss": 0.3698,
      "step": 4472
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 1.1808912837909107,
      "learning_rate": 8.76697971365409e-06,
      "loss": 0.604,
      "step": 4473
    },
    {
      "epoch": 0.24143327397334197,
      "grad_norm": 0.9058774534773671,
      "learning_rate": 8.766455025725497e-06,
      "loss": 0.4287,
      "step": 4474
    },
    {
      "epoch": 0.24148723760185634,
      "grad_norm": 1.0613054771463337,
      "learning_rate": 8.765930243915689e-06,
      "loss": 0.585,
      "step": 4475
    },
    {
      "epoch": 0.24154120123037073,
      "grad_norm": 1.0406556690078919,
      "learning_rate": 8.765405368239743e-06,
      "loss": 0.5328,
      "step": 4476
    },
    {
      "epoch": 0.2415951648588851,
      "grad_norm": 1.1153393510559482,
      "learning_rate": 8.76488039871275e-06,
      "loss": 0.4759,
      "step": 4477
    },
    {
      "epoch": 0.2416491284873995,
      "grad_norm": 1.0594136475555267,
      "learning_rate": 8.764355335349794e-06,
      "loss": 0.5356,
      "step": 4478
    },
    {
      "epoch": 0.24170309211591387,
      "grad_norm": 1.0543849984150937,
      "learning_rate": 8.763830178165967e-06,
      "loss": 0.5712,
      "step": 4479
    },
    {
      "epoch": 0.24175705574442827,
      "grad_norm": 1.1520437234916578,
      "learning_rate": 8.763304927176363e-06,
      "loss": 0.5626,
      "step": 4480
    },
    {
      "epoch": 0.24181101937294264,
      "grad_norm": 0.9355460169629677,
      "learning_rate": 8.762779582396078e-06,
      "loss": 0.4356,
      "step": 4481
    },
    {
      "epoch": 0.241864983001457,
      "grad_norm": 0.8855893023183755,
      "learning_rate": 8.762254143840211e-06,
      "loss": 0.3247,
      "step": 4482
    },
    {
      "epoch": 0.2419189466299714,
      "grad_norm": 1.1672813772507327,
      "learning_rate": 8.761728611523864e-06,
      "loss": 0.4471,
      "step": 4483
    },
    {
      "epoch": 0.24197291025848577,
      "grad_norm": 1.231238592597526,
      "learning_rate": 8.761202985462141e-06,
      "loss": 0.6042,
      "step": 4484
    },
    {
      "epoch": 0.24202687388700017,
      "grad_norm": 1.05754937086084,
      "learning_rate": 8.760677265670149e-06,
      "loss": 0.6493,
      "step": 4485
    },
    {
      "epoch": 0.24208083751551454,
      "grad_norm": 1.0607517925350525,
      "learning_rate": 8.760151452162996e-06,
      "loss": 0.4747,
      "step": 4486
    },
    {
      "epoch": 0.24213480114402894,
      "grad_norm": 1.123298156123676,
      "learning_rate": 8.759625544955797e-06,
      "loss": 0.5499,
      "step": 4487
    },
    {
      "epoch": 0.2421887647725433,
      "grad_norm": 0.9238540280104023,
      "learning_rate": 8.759099544063668e-06,
      "loss": 0.4266,
      "step": 4488
    },
    {
      "epoch": 0.24224272840105768,
      "grad_norm": 0.9390164836135448,
      "learning_rate": 8.758573449501724e-06,
      "loss": 0.4411,
      "step": 4489
    },
    {
      "epoch": 0.24229669202957207,
      "grad_norm": 1.0727487274331275,
      "learning_rate": 8.758047261285085e-06,
      "loss": 0.5513,
      "step": 4490
    },
    {
      "epoch": 0.24235065565808644,
      "grad_norm": 0.9476608451235209,
      "learning_rate": 8.757520979428878e-06,
      "loss": 0.458,
      "step": 4491
    },
    {
      "epoch": 0.24240461928660084,
      "grad_norm": 1.1311171234811987,
      "learning_rate": 8.756994603948226e-06,
      "loss": 0.4286,
      "step": 4492
    },
    {
      "epoch": 0.2424585829151152,
      "grad_norm": 1.1529072357283119,
      "learning_rate": 8.756468134858258e-06,
      "loss": 0.58,
      "step": 4493
    },
    {
      "epoch": 0.24251254654362958,
      "grad_norm": 1.073373146549898,
      "learning_rate": 8.755941572174104e-06,
      "loss": 0.4255,
      "step": 4494
    },
    {
      "epoch": 0.24256651017214398,
      "grad_norm": 1.1113101286638392,
      "learning_rate": 8.755414915910901e-06,
      "loss": 0.5512,
      "step": 4495
    },
    {
      "epoch": 0.24262047380065835,
      "grad_norm": 1.277170819662614,
      "learning_rate": 8.754888166083783e-06,
      "loss": 0.5204,
      "step": 4496
    },
    {
      "epoch": 0.24267443742917275,
      "grad_norm": 0.7686999413496634,
      "learning_rate": 8.754361322707891e-06,
      "loss": 0.3238,
      "step": 4497
    },
    {
      "epoch": 0.24272840105768712,
      "grad_norm": 1.0838427561996915,
      "learning_rate": 8.753834385798365e-06,
      "loss": 0.4976,
      "step": 4498
    },
    {
      "epoch": 0.2427823646862015,
      "grad_norm": 0.9685891805430745,
      "learning_rate": 8.753307355370353e-06,
      "loss": 0.46,
      "step": 4499
    },
    {
      "epoch": 0.24283632831471588,
      "grad_norm": 1.0452559920421545,
      "learning_rate": 8.752780231438999e-06,
      "loss": 0.4608,
      "step": 4500
    },
    {
      "epoch": 0.24283632831471588,
      "eval_loss": 0.5786745548248291,
      "eval_runtime": 159.742,
      "eval_samples_per_second": 21.528,
      "eval_steps_per_second": 0.901,
      "step": 4500
    },
    {
      "epoch": 0.24289029194323025,
      "grad_norm": 1.0347496393922702,
      "learning_rate": 8.752253014019455e-06,
      "loss": 0.4134,
      "step": 4501
    },
    {
      "epoch": 0.24294425557174465,
      "grad_norm": 0.9911194331172879,
      "learning_rate": 8.751725703126872e-06,
      "loss": 0.499,
      "step": 4502
    },
    {
      "epoch": 0.24299821920025902,
      "grad_norm": 1.1719826253193577,
      "learning_rate": 8.751198298776408e-06,
      "loss": 0.6023,
      "step": 4503
    },
    {
      "epoch": 0.24305218282877342,
      "grad_norm": 1.0240929812232697,
      "learning_rate": 8.75067080098322e-06,
      "loss": 0.5047,
      "step": 4504
    },
    {
      "epoch": 0.2431061464572878,
      "grad_norm": 1.0952239543049223,
      "learning_rate": 8.750143209762468e-06,
      "loss": 0.5975,
      "step": 4505
    },
    {
      "epoch": 0.24316011008580216,
      "grad_norm": 0.9423099116349676,
      "learning_rate": 8.749615525129317e-06,
      "loss": 0.4691,
      "step": 4506
    },
    {
      "epoch": 0.24321407371431655,
      "grad_norm": 0.9685305262234355,
      "learning_rate": 8.74908774709893e-06,
      "loss": 0.4043,
      "step": 4507
    },
    {
      "epoch": 0.24326803734283092,
      "grad_norm": 0.9591719677662082,
      "learning_rate": 8.74855987568648e-06,
      "loss": 0.4351,
      "step": 4508
    },
    {
      "epoch": 0.24332200097134532,
      "grad_norm": 1.3645078583975547,
      "learning_rate": 8.748031910907136e-06,
      "loss": 0.392,
      "step": 4509
    },
    {
      "epoch": 0.2433759645998597,
      "grad_norm": 1.1078272797837547,
      "learning_rate": 8.747503852776073e-06,
      "loss": 0.4798,
      "step": 4510
    },
    {
      "epoch": 0.2434299282283741,
      "grad_norm": 1.0606201467634875,
      "learning_rate": 8.746975701308468e-06,
      "loss": 0.485,
      "step": 4511
    },
    {
      "epoch": 0.24348389185688846,
      "grad_norm": 1.0712520787198745,
      "learning_rate": 8.746447456519501e-06,
      "loss": 0.5411,
      "step": 4512
    },
    {
      "epoch": 0.24353785548540283,
      "grad_norm": 1.1683380252992273,
      "learning_rate": 8.745919118424355e-06,
      "loss": 0.4768,
      "step": 4513
    },
    {
      "epoch": 0.24359181911391722,
      "grad_norm": 1.2297470620274826,
      "learning_rate": 8.74539068703821e-06,
      "loss": 0.6862,
      "step": 4514
    },
    {
      "epoch": 0.2436457827424316,
      "grad_norm": 1.0616846827148587,
      "learning_rate": 8.744862162376263e-06,
      "loss": 0.5352,
      "step": 4515
    },
    {
      "epoch": 0.243699746370946,
      "grad_norm": 1.3467223574754694,
      "learning_rate": 8.744333544453695e-06,
      "loss": 0.6953,
      "step": 4516
    },
    {
      "epoch": 0.24375370999946036,
      "grad_norm": 1.1024770718147714,
      "learning_rate": 8.743804833285701e-06,
      "loss": 0.5203,
      "step": 4517
    },
    {
      "epoch": 0.24380767362797476,
      "grad_norm": 0.9266411634138806,
      "learning_rate": 8.743276028887482e-06,
      "loss": 0.4766,
      "step": 4518
    },
    {
      "epoch": 0.24386163725648913,
      "grad_norm": 1.0790471917638484,
      "learning_rate": 8.742747131274232e-06,
      "loss": 0.5278,
      "step": 4519
    },
    {
      "epoch": 0.2439156008850035,
      "grad_norm": 1.2317009842365523,
      "learning_rate": 8.742218140461153e-06,
      "loss": 0.5706,
      "step": 4520
    },
    {
      "epoch": 0.2439695645135179,
      "grad_norm": 0.9959228615098635,
      "learning_rate": 8.741689056463448e-06,
      "loss": 0.5652,
      "step": 4521
    },
    {
      "epoch": 0.24402352814203226,
      "grad_norm": 1.0977625070963575,
      "learning_rate": 8.741159879296324e-06,
      "loss": 0.5457,
      "step": 4522
    },
    {
      "epoch": 0.24407749177054666,
      "grad_norm": 1.0317165271240092,
      "learning_rate": 8.74063060897499e-06,
      "loss": 0.5447,
      "step": 4523
    },
    {
      "epoch": 0.24413145539906103,
      "grad_norm": 0.9779113411044444,
      "learning_rate": 8.740101245514659e-06,
      "loss": 0.4748,
      "step": 4524
    },
    {
      "epoch": 0.2441854190275754,
      "grad_norm": 0.9710308012077197,
      "learning_rate": 8.739571788930543e-06,
      "loss": 0.47,
      "step": 4525
    },
    {
      "epoch": 0.2442393826560898,
      "grad_norm": 1.113005808511701,
      "learning_rate": 8.73904223923786e-06,
      "loss": 0.638,
      "step": 4526
    },
    {
      "epoch": 0.24429334628460417,
      "grad_norm": 1.075946744567272,
      "learning_rate": 8.738512596451832e-06,
      "loss": 0.4959,
      "step": 4527
    },
    {
      "epoch": 0.24434730991311857,
      "grad_norm": 1.0972062178771205,
      "learning_rate": 8.73798286058768e-06,
      "loss": 0.5097,
      "step": 4528
    },
    {
      "epoch": 0.24440127354163294,
      "grad_norm": 1.154828910206549,
      "learning_rate": 8.737453031660628e-06,
      "loss": 0.523,
      "step": 4529
    },
    {
      "epoch": 0.24445523717014733,
      "grad_norm": 0.9821061456345133,
      "learning_rate": 8.736923109685904e-06,
      "loss": 0.4468,
      "step": 4530
    },
    {
      "epoch": 0.2445092007986617,
      "grad_norm": 0.6205791818245048,
      "learning_rate": 8.736393094678738e-06,
      "loss": 0.2753,
      "step": 4531
    },
    {
      "epoch": 0.24456316442717607,
      "grad_norm": 0.8305769708879828,
      "learning_rate": 8.735862986654368e-06,
      "loss": 0.4028,
      "step": 4532
    },
    {
      "epoch": 0.24461712805569047,
      "grad_norm": 1.0737823363226737,
      "learning_rate": 8.735332785628024e-06,
      "loss": 0.4925,
      "step": 4533
    },
    {
      "epoch": 0.24467109168420484,
      "grad_norm": 1.203826068989072,
      "learning_rate": 8.734802491614945e-06,
      "loss": 0.6011,
      "step": 4534
    },
    {
      "epoch": 0.24472505531271924,
      "grad_norm": 0.9569006456756168,
      "learning_rate": 8.734272104630375e-06,
      "loss": 0.5013,
      "step": 4535
    },
    {
      "epoch": 0.2447790189412336,
      "grad_norm": 1.1281118489319313,
      "learning_rate": 8.73374162468956e-06,
      "loss": 0.4843,
      "step": 4536
    },
    {
      "epoch": 0.244832982569748,
      "grad_norm": 0.9571279540509768,
      "learning_rate": 8.733211051807738e-06,
      "loss": 0.428,
      "step": 4537
    },
    {
      "epoch": 0.24488694619826237,
      "grad_norm": 1.3644121347277292,
      "learning_rate": 8.732680386000167e-06,
      "loss": 0.7089,
      "step": 4538
    },
    {
      "epoch": 0.24494090982677674,
      "grad_norm": 1.221640813412554,
      "learning_rate": 8.732149627282095e-06,
      "loss": 0.7131,
      "step": 4539
    },
    {
      "epoch": 0.24499487345529114,
      "grad_norm": 1.2587239234793959,
      "learning_rate": 8.731618775668775e-06,
      "loss": 0.6507,
      "step": 4540
    },
    {
      "epoch": 0.2450488370838055,
      "grad_norm": 0.9690081162761357,
      "learning_rate": 8.731087831175467e-06,
      "loss": 0.4993,
      "step": 4541
    },
    {
      "epoch": 0.2451028007123199,
      "grad_norm": 0.9919115195721688,
      "learning_rate": 8.73055679381743e-06,
      "loss": 0.4952,
      "step": 4542
    },
    {
      "epoch": 0.24515676434083428,
      "grad_norm": 1.1416151141268795,
      "learning_rate": 8.730025663609927e-06,
      "loss": 0.6717,
      "step": 4543
    },
    {
      "epoch": 0.24521072796934865,
      "grad_norm": 1.0119859741074868,
      "learning_rate": 8.729494440568223e-06,
      "loss": 0.5031,
      "step": 4544
    },
    {
      "epoch": 0.24526469159786304,
      "grad_norm": 1.1496931830843466,
      "learning_rate": 8.728963124707586e-06,
      "loss": 0.4927,
      "step": 4545
    },
    {
      "epoch": 0.2453186552263774,
      "grad_norm": 0.9607133828245882,
      "learning_rate": 8.728431716043286e-06,
      "loss": 0.531,
      "step": 4546
    },
    {
      "epoch": 0.2453726188548918,
      "grad_norm": 1.0663621337918092,
      "learning_rate": 8.727900214590593e-06,
      "loss": 0.5069,
      "step": 4547
    },
    {
      "epoch": 0.24542658248340618,
      "grad_norm": 1.1490887573087747,
      "learning_rate": 8.72736862036479e-06,
      "loss": 0.6,
      "step": 4548
    },
    {
      "epoch": 0.24548054611192058,
      "grad_norm": 0.8979682376756449,
      "learning_rate": 8.72683693338115e-06,
      "loss": 0.4038,
      "step": 4549
    },
    {
      "epoch": 0.24553450974043495,
      "grad_norm": 1.201568651160385,
      "learning_rate": 8.726305153654957e-06,
      "loss": 0.5398,
      "step": 4550
    },
    {
      "epoch": 0.24558847336894932,
      "grad_norm": 1.1372986400999774,
      "learning_rate": 8.725773281201494e-06,
      "loss": 0.4798,
      "step": 4551
    },
    {
      "epoch": 0.24564243699746371,
      "grad_norm": 1.085641349381679,
      "learning_rate": 8.725241316036047e-06,
      "loss": 0.4917,
      "step": 4552
    },
    {
      "epoch": 0.24569640062597808,
      "grad_norm": 1.0872764441269733,
      "learning_rate": 8.724709258173907e-06,
      "loss": 0.5352,
      "step": 4553
    },
    {
      "epoch": 0.24575036425449248,
      "grad_norm": 1.1493950975859577,
      "learning_rate": 8.724177107630361e-06,
      "loss": 0.4649,
      "step": 4554
    },
    {
      "epoch": 0.24580432788300685,
      "grad_norm": 1.0032706957121922,
      "learning_rate": 8.723644864420712e-06,
      "loss": 0.5106,
      "step": 4555
    },
    {
      "epoch": 0.24585829151152122,
      "grad_norm": 1.058277303548084,
      "learning_rate": 8.72311252856025e-06,
      "loss": 0.6369,
      "step": 4556
    },
    {
      "epoch": 0.24591225514003562,
      "grad_norm": 1.2471526354021933,
      "learning_rate": 8.722580100064277e-06,
      "loss": 0.5429,
      "step": 4557
    },
    {
      "epoch": 0.24596621876855,
      "grad_norm": 1.1521586280422114,
      "learning_rate": 8.722047578948096e-06,
      "loss": 0.5377,
      "step": 4558
    },
    {
      "epoch": 0.24602018239706439,
      "grad_norm": 0.9485490723422761,
      "learning_rate": 8.721514965227011e-06,
      "loss": 0.4998,
      "step": 4559
    },
    {
      "epoch": 0.24607414602557875,
      "grad_norm": 1.156746842364887,
      "learning_rate": 8.720982258916333e-06,
      "loss": 0.5515,
      "step": 4560
    },
    {
      "epoch": 0.24612810965409315,
      "grad_norm": 0.9616277434630006,
      "learning_rate": 8.720449460031367e-06,
      "loss": 0.4737,
      "step": 4561
    },
    {
      "epoch": 0.24618207328260752,
      "grad_norm": 1.1271548662016844,
      "learning_rate": 8.719916568587434e-06,
      "loss": 0.5528,
      "step": 4562
    },
    {
      "epoch": 0.2462360369111219,
      "grad_norm": 1.033726686748811,
      "learning_rate": 8.719383584599841e-06,
      "loss": 0.4057,
      "step": 4563
    },
    {
      "epoch": 0.2462900005396363,
      "grad_norm": 0.862506593182918,
      "learning_rate": 8.718850508083913e-06,
      "loss": 0.4427,
      "step": 4564
    },
    {
      "epoch": 0.24634396416815066,
      "grad_norm": 1.2206997004035272,
      "learning_rate": 8.718317339054969e-06,
      "loss": 0.5024,
      "step": 4565
    },
    {
      "epoch": 0.24639792779666506,
      "grad_norm": 1.2122725844275872,
      "learning_rate": 8.717784077528332e-06,
      "loss": 0.5288,
      "step": 4566
    },
    {
      "epoch": 0.24645189142517943,
      "grad_norm": 1.1507799811112978,
      "learning_rate": 8.71725072351933e-06,
      "loss": 0.5805,
      "step": 4567
    },
    {
      "epoch": 0.24650585505369382,
      "grad_norm": 0.9667649730452603,
      "learning_rate": 8.71671727704329e-06,
      "loss": 0.377,
      "step": 4568
    },
    {
      "epoch": 0.2465598186822082,
      "grad_norm": 1.0151565570693313,
      "learning_rate": 8.716183738115549e-06,
      "loss": 0.5779,
      "step": 4569
    },
    {
      "epoch": 0.24661378231072256,
      "grad_norm": 1.1091655036166475,
      "learning_rate": 8.715650106751434e-06,
      "loss": 0.4461,
      "step": 4570
    },
    {
      "epoch": 0.24666774593923696,
      "grad_norm": 0.8237104050226594,
      "learning_rate": 8.715116382966285e-06,
      "loss": 0.4064,
      "step": 4571
    },
    {
      "epoch": 0.24672170956775133,
      "grad_norm": 1.1276838481599933,
      "learning_rate": 8.714582566775444e-06,
      "loss": 0.5253,
      "step": 4572
    },
    {
      "epoch": 0.24677567319626573,
      "grad_norm": 0.933199669287803,
      "learning_rate": 8.714048658194253e-06,
      "loss": 0.3671,
      "step": 4573
    },
    {
      "epoch": 0.2468296368247801,
      "grad_norm": 1.196799895364356,
      "learning_rate": 8.713514657238055e-06,
      "loss": 0.5556,
      "step": 4574
    },
    {
      "epoch": 0.24688360045329447,
      "grad_norm": 1.1835107715234925,
      "learning_rate": 8.7129805639222e-06,
      "loss": 0.5963,
      "step": 4575
    },
    {
      "epoch": 0.24693756408180886,
      "grad_norm": 1.1221259317842405,
      "learning_rate": 8.712446378262035e-06,
      "loss": 0.5739,
      "step": 4576
    },
    {
      "epoch": 0.24699152771032323,
      "grad_norm": 0.9202137034395859,
      "learning_rate": 8.711912100272917e-06,
      "loss": 0.4501,
      "step": 4577
    },
    {
      "epoch": 0.24704549133883763,
      "grad_norm": 1.0304341995620017,
      "learning_rate": 8.711377729970198e-06,
      "loss": 0.4529,
      "step": 4578
    },
    {
      "epoch": 0.247099454967352,
      "grad_norm": 1.2867382828846046,
      "learning_rate": 8.71084326736924e-06,
      "loss": 0.6419,
      "step": 4579
    },
    {
      "epoch": 0.2471534185958664,
      "grad_norm": 1.162557532593156,
      "learning_rate": 8.710308712485401e-06,
      "loss": 0.4967,
      "step": 4580
    },
    {
      "epoch": 0.24720738222438077,
      "grad_norm": 0.8662074185769372,
      "learning_rate": 8.709774065334046e-06,
      "loss": 0.4078,
      "step": 4581
    },
    {
      "epoch": 0.24726134585289514,
      "grad_norm": 1.1568262463563586,
      "learning_rate": 8.709239325930543e-06,
      "loss": 0.4416,
      "step": 4582
    },
    {
      "epoch": 0.24731530948140953,
      "grad_norm": 1.120356437351626,
      "learning_rate": 8.708704494290257e-06,
      "loss": 0.5596,
      "step": 4583
    },
    {
      "epoch": 0.2473692731099239,
      "grad_norm": 1.083889146639063,
      "learning_rate": 8.708169570428563e-06,
      "loss": 0.4399,
      "step": 4584
    },
    {
      "epoch": 0.2474232367384383,
      "grad_norm": 1.060589039276693,
      "learning_rate": 8.707634554360832e-06,
      "loss": 0.7136,
      "step": 4585
    },
    {
      "epoch": 0.24747720036695267,
      "grad_norm": 1.0301517348437907,
      "learning_rate": 8.707099446102443e-06,
      "loss": 0.5168,
      "step": 4586
    },
    {
      "epoch": 0.24753116399546707,
      "grad_norm": 1.0932143106425458,
      "learning_rate": 8.706564245668778e-06,
      "loss": 0.5495,
      "step": 4587
    },
    {
      "epoch": 0.24758512762398144,
      "grad_norm": 0.9755221315096876,
      "learning_rate": 8.706028953075214e-06,
      "loss": 0.4375,
      "step": 4588
    },
    {
      "epoch": 0.2476390912524958,
      "grad_norm": 1.002573067407256,
      "learning_rate": 8.705493568337138e-06,
      "loss": 0.4836,
      "step": 4589
    },
    {
      "epoch": 0.2476930548810102,
      "grad_norm": 1.0374628231733365,
      "learning_rate": 8.70495809146994e-06,
      "loss": 0.4863,
      "step": 4590
    },
    {
      "epoch": 0.24774701850952457,
      "grad_norm": 0.9067463974134781,
      "learning_rate": 8.704422522489005e-06,
      "loss": 0.5309,
      "step": 4591
    },
    {
      "epoch": 0.24780098213803897,
      "grad_norm": 0.8930549565284348,
      "learning_rate": 8.703886861409732e-06,
      "loss": 0.4249,
      "step": 4592
    },
    {
      "epoch": 0.24785494576655334,
      "grad_norm": 1.0995963555824488,
      "learning_rate": 8.703351108247514e-06,
      "loss": 0.4133,
      "step": 4593
    },
    {
      "epoch": 0.2479089093950677,
      "grad_norm": 1.0436639421212672,
      "learning_rate": 8.702815263017747e-06,
      "loss": 0.3836,
      "step": 4594
    },
    {
      "epoch": 0.2479628730235821,
      "grad_norm": 0.8309231500220468,
      "learning_rate": 8.702279325735831e-06,
      "loss": 0.3313,
      "step": 4595
    },
    {
      "epoch": 0.24801683665209648,
      "grad_norm": 0.7904761241845633,
      "learning_rate": 8.701743296417175e-06,
      "loss": 0.2885,
      "step": 4596
    },
    {
      "epoch": 0.24807080028061088,
      "grad_norm": 0.9934532964507723,
      "learning_rate": 8.70120717507718e-06,
      "loss": 0.447,
      "step": 4597
    },
    {
      "epoch": 0.24812476390912525,
      "grad_norm": 1.0000756396733694,
      "learning_rate": 8.700670961731257e-06,
      "loss": 0.477,
      "step": 4598
    },
    {
      "epoch": 0.24817872753763964,
      "grad_norm": 0.9746253116354061,
      "learning_rate": 8.700134656394814e-06,
      "loss": 0.4913,
      "step": 4599
    },
    {
      "epoch": 0.248232691166154,
      "grad_norm": 1.2629598790773646,
      "learning_rate": 8.699598259083269e-06,
      "loss": 0.5343,
      "step": 4600
    },
    {
      "epoch": 0.24828665479466838,
      "grad_norm": 0.9645849442478929,
      "learning_rate": 8.699061769812038e-06,
      "loss": 0.3381,
      "step": 4601
    },
    {
      "epoch": 0.24834061842318278,
      "grad_norm": 1.1471534748000067,
      "learning_rate": 8.698525188596537e-06,
      "loss": 0.4834,
      "step": 4602
    },
    {
      "epoch": 0.24839458205169715,
      "grad_norm": 1.205853159750565,
      "learning_rate": 8.697988515452194e-06,
      "loss": 0.5552,
      "step": 4603
    },
    {
      "epoch": 0.24844854568021155,
      "grad_norm": 1.0358730504018554,
      "learning_rate": 8.697451750394425e-06,
      "loss": 0.4595,
      "step": 4604
    },
    {
      "epoch": 0.24850250930872592,
      "grad_norm": 1.02513956247496,
      "learning_rate": 8.696914893438665e-06,
      "loss": 0.3858,
      "step": 4605
    },
    {
      "epoch": 0.2485564729372403,
      "grad_norm": 0.8755438204696631,
      "learning_rate": 8.696377944600342e-06,
      "loss": 0.4549,
      "step": 4606
    },
    {
      "epoch": 0.24861043656575468,
      "grad_norm": 0.7804218915302031,
      "learning_rate": 8.695840903894886e-06,
      "loss": 0.3907,
      "step": 4607
    },
    {
      "epoch": 0.24866440019426905,
      "grad_norm": 1.0136461016388192,
      "learning_rate": 8.69530377133773e-06,
      "loss": 0.5114,
      "step": 4608
    },
    {
      "epoch": 0.24871836382278345,
      "grad_norm": 1.0196432768227743,
      "learning_rate": 8.69476654694432e-06,
      "loss": 0.3728,
      "step": 4609
    },
    {
      "epoch": 0.24877232745129782,
      "grad_norm": 1.0685559132464502,
      "learning_rate": 8.694229230730087e-06,
      "loss": 0.4677,
      "step": 4610
    },
    {
      "epoch": 0.24882629107981222,
      "grad_norm": 1.3060421254073247,
      "learning_rate": 8.69369182271048e-06,
      "loss": 0.5106,
      "step": 4611
    },
    {
      "epoch": 0.2488802547083266,
      "grad_norm": 0.9558738119319248,
      "learning_rate": 8.693154322900944e-06,
      "loss": 0.4536,
      "step": 4612
    },
    {
      "epoch": 0.24893421833684096,
      "grad_norm": 1.0111618166878278,
      "learning_rate": 8.692616731316924e-06,
      "loss": 0.5893,
      "step": 4613
    },
    {
      "epoch": 0.24898818196535535,
      "grad_norm": 1.0162019025276867,
      "learning_rate": 8.692079047973876e-06,
      "loss": 0.4364,
      "step": 4614
    },
    {
      "epoch": 0.24904214559386972,
      "grad_norm": 1.0202567031837155,
      "learning_rate": 8.691541272887251e-06,
      "loss": 0.4729,
      "step": 4615
    },
    {
      "epoch": 0.24909610922238412,
      "grad_norm": 1.0198511187883077,
      "learning_rate": 8.691003406072503e-06,
      "loss": 0.4855,
      "step": 4616
    },
    {
      "epoch": 0.2491500728508985,
      "grad_norm": 1.0879680664225109,
      "learning_rate": 8.690465447545093e-06,
      "loss": 0.5632,
      "step": 4617
    },
    {
      "epoch": 0.2492040364794129,
      "grad_norm": 1.1209853007557333,
      "learning_rate": 8.689927397320484e-06,
      "loss": 0.4848,
      "step": 4618
    },
    {
      "epoch": 0.24925800010792726,
      "grad_norm": 1.0108392411398617,
      "learning_rate": 8.689389255414137e-06,
      "loss": 0.5416,
      "step": 4619
    },
    {
      "epoch": 0.24931196373644163,
      "grad_norm": 1.1903438858376374,
      "learning_rate": 8.68885102184152e-06,
      "loss": 0.7676,
      "step": 4620
    },
    {
      "epoch": 0.24936592736495602,
      "grad_norm": 1.178054306411386,
      "learning_rate": 8.688312696618103e-06,
      "loss": 0.5978,
      "step": 4621
    },
    {
      "epoch": 0.2494198909934704,
      "grad_norm": 1.0194057410781583,
      "learning_rate": 8.687774279759357e-06,
      "loss": 0.4899,
      "step": 4622
    },
    {
      "epoch": 0.2494738546219848,
      "grad_norm": 1.1247838937534858,
      "learning_rate": 8.687235771280757e-06,
      "loss": 0.5165,
      "step": 4623
    },
    {
      "epoch": 0.24952781825049916,
      "grad_norm": 0.8413363716071782,
      "learning_rate": 8.686697171197782e-06,
      "loss": 0.4829,
      "step": 4624
    },
    {
      "epoch": 0.24958178187901353,
      "grad_norm": 1.4362665728065536,
      "learning_rate": 8.68615847952591e-06,
      "loss": 0.6991,
      "step": 4625
    },
    {
      "epoch": 0.24963574550752793,
      "grad_norm": 1.3413767496700388,
      "learning_rate": 8.685619696280624e-06,
      "loss": 0.4819,
      "step": 4626
    },
    {
      "epoch": 0.2496897091360423,
      "grad_norm": 1.0329409385564143,
      "learning_rate": 8.685080821477407e-06,
      "loss": 0.4053,
      "step": 4627
    },
    {
      "epoch": 0.2497436727645567,
      "grad_norm": 0.7942215860910685,
      "learning_rate": 8.684541855131751e-06,
      "loss": 0.4179,
      "step": 4628
    },
    {
      "epoch": 0.24979763639307107,
      "grad_norm": 0.9339792277485138,
      "learning_rate": 8.684002797259144e-06,
      "loss": 0.531,
      "step": 4629
    },
    {
      "epoch": 0.24985160002158546,
      "grad_norm": 1.119468146350275,
      "learning_rate": 8.683463647875079e-06,
      "loss": 0.5996,
      "step": 4630
    },
    {
      "epoch": 0.24990556365009983,
      "grad_norm": 1.0354026908687988,
      "learning_rate": 8.682924406995052e-06,
      "loss": 0.5106,
      "step": 4631
    },
    {
      "epoch": 0.2499595272786142,
      "grad_norm": 1.1310648875565443,
      "learning_rate": 8.682385074634562e-06,
      "loss": 0.4588,
      "step": 4632
    },
    {
      "epoch": 0.2500134909071286,
      "grad_norm": 0.834884580377105,
      "learning_rate": 8.681845650809109e-06,
      "loss": 0.384,
      "step": 4633
    },
    {
      "epoch": 0.250067454535643,
      "grad_norm": 1.1593628471354747,
      "learning_rate": 8.681306135534198e-06,
      "loss": 0.597,
      "step": 4634
    },
    {
      "epoch": 0.25012141816415734,
      "grad_norm": 0.8096407218154917,
      "learning_rate": 8.680766528825332e-06,
      "loss": 0.3914,
      "step": 4635
    },
    {
      "epoch": 0.25017538179267174,
      "grad_norm": 1.167017215959199,
      "learning_rate": 8.680226830698023e-06,
      "loss": 0.6153,
      "step": 4636
    },
    {
      "epoch": 0.25022934542118613,
      "grad_norm": 1.1686417031431657,
      "learning_rate": 8.679687041167782e-06,
      "loss": 0.5213,
      "step": 4637
    },
    {
      "epoch": 0.2502833090497005,
      "grad_norm": 1.1654903952526496,
      "learning_rate": 8.679147160250122e-06,
      "loss": 0.564,
      "step": 4638
    },
    {
      "epoch": 0.2503372726782149,
      "grad_norm": 1.0893147217891122,
      "learning_rate": 8.678607187960561e-06,
      "loss": 0.5024,
      "step": 4639
    },
    {
      "epoch": 0.25039123630672927,
      "grad_norm": 1.0939666921570177,
      "learning_rate": 8.678067124314618e-06,
      "loss": 0.7687,
      "step": 4640
    },
    {
      "epoch": 0.25044519993524367,
      "grad_norm": 1.30004042268799,
      "learning_rate": 8.677526969327814e-06,
      "loss": 0.4862,
      "step": 4641
    },
    {
      "epoch": 0.250499163563758,
      "grad_norm": 1.051924153650951,
      "learning_rate": 8.676986723015674e-06,
      "loss": 0.6033,
      "step": 4642
    },
    {
      "epoch": 0.2505531271922724,
      "grad_norm": 0.9541341161489371,
      "learning_rate": 8.676446385393726e-06,
      "loss": 0.3822,
      "step": 4643
    },
    {
      "epoch": 0.2506070908207868,
      "grad_norm": 1.176884088381866,
      "learning_rate": 8.675905956477498e-06,
      "loss": 0.5733,
      "step": 4644
    },
    {
      "epoch": 0.25066105444930115,
      "grad_norm": 1.092383829590621,
      "learning_rate": 8.675365436282527e-06,
      "loss": 0.7483,
      "step": 4645
    },
    {
      "epoch": 0.25071501807781554,
      "grad_norm": 0.9064241608921764,
      "learning_rate": 8.674824824824342e-06,
      "loss": 0.5765,
      "step": 4646
    },
    {
      "epoch": 0.25076898170632994,
      "grad_norm": 1.2936825788828394,
      "learning_rate": 8.674284122118484e-06,
      "loss": 0.6663,
      "step": 4647
    },
    {
      "epoch": 0.25082294533484434,
      "grad_norm": 0.9757480576532195,
      "learning_rate": 8.673743328180494e-06,
      "loss": 0.3793,
      "step": 4648
    },
    {
      "epoch": 0.2508769089633587,
      "grad_norm": 1.1495317543832295,
      "learning_rate": 8.673202443025913e-06,
      "loss": 0.585,
      "step": 4649
    },
    {
      "epoch": 0.2509308725918731,
      "grad_norm": 0.9352192046083353,
      "learning_rate": 8.672661466670287e-06,
      "loss": 0.4025,
      "step": 4650
    },
    {
      "epoch": 0.2509848362203875,
      "grad_norm": 1.0526747709530804,
      "learning_rate": 8.672120399129164e-06,
      "loss": 0.4262,
      "step": 4651
    },
    {
      "epoch": 0.2510387998489018,
      "grad_norm": 1.045492351445694,
      "learning_rate": 8.671579240418099e-06,
      "loss": 0.4032,
      "step": 4652
    },
    {
      "epoch": 0.2510927634774162,
      "grad_norm": 1.2557011290679319,
      "learning_rate": 8.67103799055264e-06,
      "loss": 0.806,
      "step": 4653
    },
    {
      "epoch": 0.2511467271059306,
      "grad_norm": 0.9694591174202163,
      "learning_rate": 8.670496649548344e-06,
      "loss": 0.3825,
      "step": 4654
    },
    {
      "epoch": 0.251200690734445,
      "grad_norm": 0.9899254989407549,
      "learning_rate": 8.669955217420771e-06,
      "loss": 0.3894,
      "step": 4655
    },
    {
      "epoch": 0.25125465436295935,
      "grad_norm": 1.0420059948494929,
      "learning_rate": 8.669413694185485e-06,
      "loss": 0.5142,
      "step": 4656
    },
    {
      "epoch": 0.25130861799147375,
      "grad_norm": 1.0823639015600264,
      "learning_rate": 8.668872079858042e-06,
      "loss": 0.4501,
      "step": 4657
    },
    {
      "epoch": 0.25136258161998815,
      "grad_norm": 1.0829709161739571,
      "learning_rate": 8.668330374454016e-06,
      "loss": 0.4438,
      "step": 4658
    },
    {
      "epoch": 0.2514165452485025,
      "grad_norm": 1.1079061031720523,
      "learning_rate": 8.667788577988975e-06,
      "loss": 0.6563,
      "step": 4659
    },
    {
      "epoch": 0.2514705088770169,
      "grad_norm": 1.062552492316104,
      "learning_rate": 8.66724669047849e-06,
      "loss": 0.4529,
      "step": 4660
    },
    {
      "epoch": 0.2515244725055313,
      "grad_norm": 1.133973701737392,
      "learning_rate": 8.666704711938133e-06,
      "loss": 0.5176,
      "step": 4661
    },
    {
      "epoch": 0.2515784361340457,
      "grad_norm": 1.1103955260711251,
      "learning_rate": 8.666162642383481e-06,
      "loss": 0.5229,
      "step": 4662
    },
    {
      "epoch": 0.25163239976256,
      "grad_norm": 1.1537072699447342,
      "learning_rate": 8.66562048183012e-06,
      "loss": 0.5537,
      "step": 4663
    },
    {
      "epoch": 0.2516863633910744,
      "grad_norm": 1.1018935525465061,
      "learning_rate": 8.665078230293626e-06,
      "loss": 0.5703,
      "step": 4664
    },
    {
      "epoch": 0.2517403270195888,
      "grad_norm": 1.2030832666004732,
      "learning_rate": 8.664535887789583e-06,
      "loss": 0.5521,
      "step": 4665
    },
    {
      "epoch": 0.25179429064810316,
      "grad_norm": 0.9748328118325305,
      "learning_rate": 8.663993454333583e-06,
      "loss": 0.494,
      "step": 4666
    },
    {
      "epoch": 0.25184825427661756,
      "grad_norm": 0.9875485350776795,
      "learning_rate": 8.663450929941214e-06,
      "loss": 0.4388,
      "step": 4667
    },
    {
      "epoch": 0.25190221790513195,
      "grad_norm": 1.338492272955238,
      "learning_rate": 8.662908314628069e-06,
      "loss": 0.6516,
      "step": 4668
    },
    {
      "epoch": 0.2519561815336463,
      "grad_norm": 0.8767492182970065,
      "learning_rate": 8.662365608409744e-06,
      "loss": 0.3735,
      "step": 4669
    },
    {
      "epoch": 0.2520101451621607,
      "grad_norm": 0.9947005162480036,
      "learning_rate": 8.661822811301834e-06,
      "loss": 0.4555,
      "step": 4670
    },
    {
      "epoch": 0.2520641087906751,
      "grad_norm": 1.1429969419517736,
      "learning_rate": 8.661279923319944e-06,
      "loss": 0.5985,
      "step": 4671
    },
    {
      "epoch": 0.2521180724191895,
      "grad_norm": 1.008991925784486,
      "learning_rate": 8.660736944479673e-06,
      "loss": 0.5753,
      "step": 4672
    },
    {
      "epoch": 0.25217203604770383,
      "grad_norm": 0.9489067867810032,
      "learning_rate": 8.660193874796629e-06,
      "loss": 0.4668,
      "step": 4673
    },
    {
      "epoch": 0.2522259996762182,
      "grad_norm": 0.994584421856532,
      "learning_rate": 8.659650714286419e-06,
      "loss": 0.5281,
      "step": 4674
    },
    {
      "epoch": 0.2522799633047326,
      "grad_norm": 0.8639435658018252,
      "learning_rate": 8.659107462964657e-06,
      "loss": 0.3761,
      "step": 4675
    },
    {
      "epoch": 0.25233392693324697,
      "grad_norm": 1.2739342995905965,
      "learning_rate": 8.658564120846953e-06,
      "loss": 0.7129,
      "step": 4676
    },
    {
      "epoch": 0.25238789056176136,
      "grad_norm": 1.1252140282275422,
      "learning_rate": 8.658020687948925e-06,
      "loss": 0.5598,
      "step": 4677
    },
    {
      "epoch": 0.25244185419027576,
      "grad_norm": 0.8989640921268353,
      "learning_rate": 8.65747716428619e-06,
      "loss": 0.3795,
      "step": 4678
    },
    {
      "epoch": 0.25249581781879016,
      "grad_norm": 1.0563550262086794,
      "learning_rate": 8.656933549874374e-06,
      "loss": 0.4486,
      "step": 4679
    },
    {
      "epoch": 0.2525497814473045,
      "grad_norm": 1.0427272650172887,
      "learning_rate": 8.656389844729094e-06,
      "loss": 0.4763,
      "step": 4680
    },
    {
      "epoch": 0.2526037450758189,
      "grad_norm": 1.2025475470600222,
      "learning_rate": 8.655846048865983e-06,
      "loss": 0.5987,
      "step": 4681
    },
    {
      "epoch": 0.2526577087043333,
      "grad_norm": 0.911868503447582,
      "learning_rate": 8.655302162300668e-06,
      "loss": 0.3402,
      "step": 4682
    },
    {
      "epoch": 0.25271167233284764,
      "grad_norm": 1.0489916125764895,
      "learning_rate": 8.654758185048781e-06,
      "loss": 0.4968,
      "step": 4683
    },
    {
      "epoch": 0.25276563596136203,
      "grad_norm": 1.0519211908890327,
      "learning_rate": 8.654214117125957e-06,
      "loss": 0.4823,
      "step": 4684
    },
    {
      "epoch": 0.25281959958987643,
      "grad_norm": 1.0858183812793807,
      "learning_rate": 8.65366995854783e-06,
      "loss": 0.5445,
      "step": 4685
    },
    {
      "epoch": 0.25287356321839083,
      "grad_norm": 1.0512473453056037,
      "learning_rate": 8.65312570933004e-06,
      "loss": 0.4689,
      "step": 4686
    },
    {
      "epoch": 0.25292752684690517,
      "grad_norm": 1.243521561722281,
      "learning_rate": 8.652581369488235e-06,
      "loss": 0.5978,
      "step": 4687
    },
    {
      "epoch": 0.25298149047541957,
      "grad_norm": 1.0360413851599632,
      "learning_rate": 8.652036939038055e-06,
      "loss": 0.5051,
      "step": 4688
    },
    {
      "epoch": 0.25303545410393397,
      "grad_norm": 1.0659968738774683,
      "learning_rate": 8.651492417995147e-06,
      "loss": 0.5013,
      "step": 4689
    },
    {
      "epoch": 0.2530894177324483,
      "grad_norm": 1.220151833199879,
      "learning_rate": 8.650947806375165e-06,
      "loss": 0.5658,
      "step": 4690
    },
    {
      "epoch": 0.2531433813609627,
      "grad_norm": 1.1767838769975547,
      "learning_rate": 8.650403104193758e-06,
      "loss": 0.5583,
      "step": 4691
    },
    {
      "epoch": 0.2531973449894771,
      "grad_norm": 1.0289736242325367,
      "learning_rate": 8.649858311466581e-06,
      "loss": 0.4434,
      "step": 4692
    },
    {
      "epoch": 0.2532513086179915,
      "grad_norm": 1.001042167713907,
      "learning_rate": 8.649313428209295e-06,
      "loss": 0.4333,
      "step": 4693
    },
    {
      "epoch": 0.25330527224650584,
      "grad_norm": 0.9721962469303661,
      "learning_rate": 8.648768454437559e-06,
      "loss": 0.5783,
      "step": 4694
    },
    {
      "epoch": 0.25335923587502024,
      "grad_norm": 1.0639881155303819,
      "learning_rate": 8.648223390167036e-06,
      "loss": 0.606,
      "step": 4695
    },
    {
      "epoch": 0.25341319950353464,
      "grad_norm": 0.9781002664198847,
      "learning_rate": 8.647678235413392e-06,
      "loss": 0.3875,
      "step": 4696
    },
    {
      "epoch": 0.253467163132049,
      "grad_norm": 1.0792283403050487,
      "learning_rate": 8.647132990192294e-06,
      "loss": 0.4838,
      "step": 4697
    },
    {
      "epoch": 0.2535211267605634,
      "grad_norm": 0.9225755664128706,
      "learning_rate": 8.646587654519413e-06,
      "loss": 0.394,
      "step": 4698
    },
    {
      "epoch": 0.2535750903890778,
      "grad_norm": 0.983300300991792,
      "learning_rate": 8.646042228410427e-06,
      "loss": 0.5214,
      "step": 4699
    },
    {
      "epoch": 0.25362905401759217,
      "grad_norm": 1.1094239308384897,
      "learning_rate": 8.645496711881005e-06,
      "loss": 0.4959,
      "step": 4700
    },
    {
      "epoch": 0.2536830176461065,
      "grad_norm": 1.0289085413727432,
      "learning_rate": 8.64495110494683e-06,
      "loss": 0.4549,
      "step": 4701
    },
    {
      "epoch": 0.2537369812746209,
      "grad_norm": 1.1405343903525262,
      "learning_rate": 8.644405407623584e-06,
      "loss": 0.6822,
      "step": 4702
    },
    {
      "epoch": 0.2537909449031353,
      "grad_norm": 1.067754758540559,
      "learning_rate": 8.64385961992695e-06,
      "loss": 0.5095,
      "step": 4703
    },
    {
      "epoch": 0.25384490853164965,
      "grad_norm": 1.126055992031456,
      "learning_rate": 8.64331374187261e-06,
      "loss": 0.6047,
      "step": 4704
    },
    {
      "epoch": 0.25389887216016405,
      "grad_norm": 1.2390534775340674,
      "learning_rate": 8.642767773476258e-06,
      "loss": 0.7254,
      "step": 4705
    },
    {
      "epoch": 0.25395283578867844,
      "grad_norm": 0.84808052360508,
      "learning_rate": 8.642221714753588e-06,
      "loss": 0.4104,
      "step": 4706
    },
    {
      "epoch": 0.2540067994171928,
      "grad_norm": 1.1264051857222837,
      "learning_rate": 8.641675565720289e-06,
      "loss": 0.5063,
      "step": 4707
    },
    {
      "epoch": 0.2540607630457072,
      "grad_norm": 1.0694037360849165,
      "learning_rate": 8.64112932639206e-06,
      "loss": 0.4464,
      "step": 4708
    },
    {
      "epoch": 0.2541147266742216,
      "grad_norm": 1.110202349014479,
      "learning_rate": 8.640582996784597e-06,
      "loss": 0.5092,
      "step": 4709
    },
    {
      "epoch": 0.254168690302736,
      "grad_norm": 1.0926368279848881,
      "learning_rate": 8.640036576913609e-06,
      "loss": 0.6214,
      "step": 4710
    },
    {
      "epoch": 0.2542226539312503,
      "grad_norm": 1.0403056060885083,
      "learning_rate": 8.639490066794797e-06,
      "loss": 0.649,
      "step": 4711
    },
    {
      "epoch": 0.2542766175597647,
      "grad_norm": 1.1521916394916742,
      "learning_rate": 8.638943466443867e-06,
      "loss": 0.654,
      "step": 4712
    },
    {
      "epoch": 0.2543305811882791,
      "grad_norm": 1.1973875911827312,
      "learning_rate": 8.638396775876531e-06,
      "loss": 0.4643,
      "step": 4713
    },
    {
      "epoch": 0.25438454481679346,
      "grad_norm": 0.9985086186284929,
      "learning_rate": 8.6378499951085e-06,
      "loss": 0.4789,
      "step": 4714
    },
    {
      "epoch": 0.25443850844530785,
      "grad_norm": 1.0028378786621082,
      "learning_rate": 8.63730312415549e-06,
      "loss": 0.4989,
      "step": 4715
    },
    {
      "epoch": 0.25449247207382225,
      "grad_norm": 1.1766990006039972,
      "learning_rate": 8.636756163033218e-06,
      "loss": 0.5806,
      "step": 4716
    },
    {
      "epoch": 0.25454643570233665,
      "grad_norm": 0.8926750708551197,
      "learning_rate": 8.636209111757405e-06,
      "loss": 0.4293,
      "step": 4717
    },
    {
      "epoch": 0.254600399330851,
      "grad_norm": 0.9527827225045191,
      "learning_rate": 8.635661970343773e-06,
      "loss": 0.3837,
      "step": 4718
    },
    {
      "epoch": 0.2546543629593654,
      "grad_norm": 0.9513594112110718,
      "learning_rate": 8.635114738808045e-06,
      "loss": 0.3435,
      "step": 4719
    },
    {
      "epoch": 0.2547083265878798,
      "grad_norm": 1.0428734444886056,
      "learning_rate": 8.634567417165956e-06,
      "loss": 0.4809,
      "step": 4720
    },
    {
      "epoch": 0.2547622902163941,
      "grad_norm": 1.1991795804777174,
      "learning_rate": 8.634020005433229e-06,
      "loss": 0.5313,
      "step": 4721
    },
    {
      "epoch": 0.2548162538449085,
      "grad_norm": 0.8578801078833512,
      "learning_rate": 8.633472503625602e-06,
      "loss": 0.3417,
      "step": 4722
    },
    {
      "epoch": 0.2548702174734229,
      "grad_norm": 1.0048990019304103,
      "learning_rate": 8.632924911758808e-06,
      "loss": 0.4366,
      "step": 4723
    },
    {
      "epoch": 0.2549241811019373,
      "grad_norm": 0.9515931558967206,
      "learning_rate": 8.632377229848586e-06,
      "loss": 0.4103,
      "step": 4724
    },
    {
      "epoch": 0.25497814473045166,
      "grad_norm": 0.9422457322772619,
      "learning_rate": 8.631829457910679e-06,
      "loss": 0.3797,
      "step": 4725
    },
    {
      "epoch": 0.25503210835896606,
      "grad_norm": 1.1299191544604263,
      "learning_rate": 8.63128159596083e-06,
      "loss": 0.5568,
      "step": 4726
    },
    {
      "epoch": 0.25508607198748046,
      "grad_norm": 1.174319083036384,
      "learning_rate": 8.630733644014781e-06,
      "loss": 0.5016,
      "step": 4727
    },
    {
      "epoch": 0.2551400356159948,
      "grad_norm": 1.1703259551196623,
      "learning_rate": 8.630185602088285e-06,
      "loss": 0.6325,
      "step": 4728
    },
    {
      "epoch": 0.2551939992445092,
      "grad_norm": 1.065845328081628,
      "learning_rate": 8.62963747019709e-06,
      "loss": 0.4913,
      "step": 4729
    },
    {
      "epoch": 0.2552479628730236,
      "grad_norm": 1.0061706344992347,
      "learning_rate": 8.629089248356955e-06,
      "loss": 0.5379,
      "step": 4730
    },
    {
      "epoch": 0.255301926501538,
      "grad_norm": 1.1611418906501025,
      "learning_rate": 8.628540936583632e-06,
      "loss": 0.6365,
      "step": 4731
    },
    {
      "epoch": 0.25535589013005233,
      "grad_norm": 0.8567499291371931,
      "learning_rate": 8.62799253489288e-06,
      "loss": 0.5439,
      "step": 4732
    },
    {
      "epoch": 0.25540985375856673,
      "grad_norm": 1.3126388618857745,
      "learning_rate": 8.627444043300463e-06,
      "loss": 0.6789,
      "step": 4733
    },
    {
      "epoch": 0.2554638173870811,
      "grad_norm": 0.9697156031059926,
      "learning_rate": 8.626895461822146e-06,
      "loss": 0.4409,
      "step": 4734
    },
    {
      "epoch": 0.25551778101559547,
      "grad_norm": 1.2984711627583192,
      "learning_rate": 8.626346790473693e-06,
      "loss": 0.4832,
      "step": 4735
    },
    {
      "epoch": 0.25557174464410987,
      "grad_norm": 1.2373761000460908,
      "learning_rate": 8.625798029270873e-06,
      "loss": 0.6429,
      "step": 4736
    },
    {
      "epoch": 0.25562570827262426,
      "grad_norm": 1.0110518775283,
      "learning_rate": 8.62524917822946e-06,
      "loss": 0.418,
      "step": 4737
    },
    {
      "epoch": 0.2556796719011386,
      "grad_norm": 0.9960210432015533,
      "learning_rate": 8.624700237365226e-06,
      "loss": 0.4677,
      "step": 4738
    },
    {
      "epoch": 0.255733635529653,
      "grad_norm": 0.8954507144754973,
      "learning_rate": 8.624151206693953e-06,
      "loss": 0.3791,
      "step": 4739
    },
    {
      "epoch": 0.2557875991581674,
      "grad_norm": 1.0543805923328142,
      "learning_rate": 8.623602086231417e-06,
      "loss": 0.4617,
      "step": 4740
    },
    {
      "epoch": 0.2558415627866818,
      "grad_norm": 1.2261070394313822,
      "learning_rate": 8.6230528759934e-06,
      "loss": 0.5016,
      "step": 4741
    },
    {
      "epoch": 0.25589552641519614,
      "grad_norm": 1.193618167482218,
      "learning_rate": 8.622503575995686e-06,
      "loss": 0.6098,
      "step": 4742
    },
    {
      "epoch": 0.25594949004371054,
      "grad_norm": 1.1007036449198757,
      "learning_rate": 8.621954186254067e-06,
      "loss": 0.5304,
      "step": 4743
    },
    {
      "epoch": 0.25600345367222493,
      "grad_norm": 1.1212114280309335,
      "learning_rate": 8.621404706784328e-06,
      "loss": 0.5112,
      "step": 4744
    },
    {
      "epoch": 0.2560574173007393,
      "grad_norm": 1.0832913545722138,
      "learning_rate": 8.620855137602267e-06,
      "loss": 0.5702,
      "step": 4745
    },
    {
      "epoch": 0.2561113809292537,
      "grad_norm": 1.3125030567538019,
      "learning_rate": 8.620305478723674e-06,
      "loss": 0.642,
      "step": 4746
    },
    {
      "epoch": 0.25616534455776807,
      "grad_norm": 1.1424828112775594,
      "learning_rate": 8.619755730164346e-06,
      "loss": 0.6617,
      "step": 4747
    },
    {
      "epoch": 0.25621930818628247,
      "grad_norm": 0.9609140987008827,
      "learning_rate": 8.619205891940089e-06,
      "loss": 0.6349,
      "step": 4748
    },
    {
      "epoch": 0.2562732718147968,
      "grad_norm": 1.0533856788269076,
      "learning_rate": 8.618655964066701e-06,
      "loss": 0.5189,
      "step": 4749
    },
    {
      "epoch": 0.2563272354433112,
      "grad_norm": 0.9012173701100783,
      "learning_rate": 8.618105946559993e-06,
      "loss": 0.4672,
      "step": 4750
    },
    {
      "epoch": 0.2563811990718256,
      "grad_norm": 1.0927629992181256,
      "learning_rate": 8.617555839435767e-06,
      "loss": 0.5131,
      "step": 4751
    },
    {
      "epoch": 0.25643516270033995,
      "grad_norm": 1.1157358564048347,
      "learning_rate": 8.617005642709836e-06,
      "loss": 0.5583,
      "step": 4752
    },
    {
      "epoch": 0.25648912632885434,
      "grad_norm": 1.3540776232728797,
      "learning_rate": 8.616455356398015e-06,
      "loss": 0.6633,
      "step": 4753
    },
    {
      "epoch": 0.25654308995736874,
      "grad_norm": 1.0385672344907089,
      "learning_rate": 8.615904980516116e-06,
      "loss": 0.6293,
      "step": 4754
    },
    {
      "epoch": 0.25659705358588314,
      "grad_norm": 1.13875050764832,
      "learning_rate": 8.61535451507996e-06,
      "loss": 0.509,
      "step": 4755
    },
    {
      "epoch": 0.2566510172143975,
      "grad_norm": 1.1093511562642127,
      "learning_rate": 8.614803960105368e-06,
      "loss": 0.5437,
      "step": 4756
    },
    {
      "epoch": 0.2567049808429119,
      "grad_norm": 1.0687098996725628,
      "learning_rate": 8.614253315608164e-06,
      "loss": 0.4725,
      "step": 4757
    },
    {
      "epoch": 0.2567589444714263,
      "grad_norm": 0.902088830126422,
      "learning_rate": 8.61370258160417e-06,
      "loss": 0.3781,
      "step": 4758
    },
    {
      "epoch": 0.2568129080999406,
      "grad_norm": 1.0420752523569332,
      "learning_rate": 8.613151758109221e-06,
      "loss": 0.4699,
      "step": 4759
    },
    {
      "epoch": 0.256866871728455,
      "grad_norm": 0.9266623243198918,
      "learning_rate": 8.612600845139144e-06,
      "loss": 0.4729,
      "step": 4760
    },
    {
      "epoch": 0.2569208353569694,
      "grad_norm": 1.180925513821991,
      "learning_rate": 8.612049842709771e-06,
      "loss": 0.5895,
      "step": 4761
    },
    {
      "epoch": 0.2569747989854838,
      "grad_norm": 1.237327064292216,
      "learning_rate": 8.611498750836942e-06,
      "loss": 0.5759,
      "step": 4762
    },
    {
      "epoch": 0.25702876261399815,
      "grad_norm": 1.2130067591257077,
      "learning_rate": 8.610947569536498e-06,
      "loss": 0.5098,
      "step": 4763
    },
    {
      "epoch": 0.25708272624251255,
      "grad_norm": 1.066143716042951,
      "learning_rate": 8.610396298824275e-06,
      "loss": 0.6153,
      "step": 4764
    },
    {
      "epoch": 0.25713668987102695,
      "grad_norm": 0.7871548833746083,
      "learning_rate": 8.60984493871612e-06,
      "loss": 0.3356,
      "step": 4765
    },
    {
      "epoch": 0.2571906534995413,
      "grad_norm": 1.0239917491078074,
      "learning_rate": 8.609293489227878e-06,
      "loss": 0.4657,
      "step": 4766
    },
    {
      "epoch": 0.2572446171280557,
      "grad_norm": 1.1054470581966795,
      "learning_rate": 8.6087419503754e-06,
      "loss": 0.4631,
      "step": 4767
    },
    {
      "epoch": 0.2572985807565701,
      "grad_norm": 0.997525205037016,
      "learning_rate": 8.608190322174539e-06,
      "loss": 0.536,
      "step": 4768
    },
    {
      "epoch": 0.2573525443850845,
      "grad_norm": 1.0420573831908493,
      "learning_rate": 8.607638604641145e-06,
      "loss": 0.5231,
      "step": 4769
    },
    {
      "epoch": 0.2574065080135988,
      "grad_norm": 1.0683902052910277,
      "learning_rate": 8.60708679779108e-06,
      "loss": 0.5861,
      "step": 4770
    },
    {
      "epoch": 0.2574604716421132,
      "grad_norm": 0.8923409518475283,
      "learning_rate": 8.606534901640198e-06,
      "loss": 0.3813,
      "step": 4771
    },
    {
      "epoch": 0.2575144352706276,
      "grad_norm": 1.0450148437435072,
      "learning_rate": 8.605982916204366e-06,
      "loss": 0.4895,
      "step": 4772
    },
    {
      "epoch": 0.25756839889914196,
      "grad_norm": 0.8725860693353078,
      "learning_rate": 8.605430841499446e-06,
      "loss": 0.4524,
      "step": 4773
    },
    {
      "epoch": 0.25762236252765636,
      "grad_norm": 1.1051453773295155,
      "learning_rate": 8.604878677541307e-06,
      "loss": 0.5623,
      "step": 4774
    },
    {
      "epoch": 0.25767632615617075,
      "grad_norm": 1.158119773584772,
      "learning_rate": 8.604326424345817e-06,
      "loss": 0.4481,
      "step": 4775
    },
    {
      "epoch": 0.2577302897846851,
      "grad_norm": 1.2949127443596018,
      "learning_rate": 8.603774081928848e-06,
      "loss": 0.66,
      "step": 4776
    },
    {
      "epoch": 0.2577842534131995,
      "grad_norm": 1.2503810331559044,
      "learning_rate": 8.603221650306276e-06,
      "loss": 0.5866,
      "step": 4777
    },
    {
      "epoch": 0.2578382170417139,
      "grad_norm": 0.8804330828311822,
      "learning_rate": 8.602669129493979e-06,
      "loss": 0.3406,
      "step": 4778
    },
    {
      "epoch": 0.2578921806702283,
      "grad_norm": 1.0764854027936386,
      "learning_rate": 8.602116519507835e-06,
      "loss": 0.5813,
      "step": 4779
    },
    {
      "epoch": 0.25794614429874263,
      "grad_norm": 1.0327333828383412,
      "learning_rate": 8.601563820363728e-06,
      "loss": 0.5389,
      "step": 4780
    },
    {
      "epoch": 0.258000107927257,
      "grad_norm": 1.0117585824266941,
      "learning_rate": 8.601011032077544e-06,
      "loss": 0.4145,
      "step": 4781
    },
    {
      "epoch": 0.2580540715557714,
      "grad_norm": 0.9219203401855592,
      "learning_rate": 8.600458154665167e-06,
      "loss": 0.3668,
      "step": 4782
    },
    {
      "epoch": 0.25810803518428577,
      "grad_norm": 1.0590268948442758,
      "learning_rate": 8.599905188142491e-06,
      "loss": 0.6278,
      "step": 4783
    },
    {
      "epoch": 0.25816199881280016,
      "grad_norm": 1.1117320646419748,
      "learning_rate": 8.59935213252541e-06,
      "loss": 0.4273,
      "step": 4784
    },
    {
      "epoch": 0.25821596244131456,
      "grad_norm": 1.021085077853354,
      "learning_rate": 8.598798987829816e-06,
      "loss": 0.4567,
      "step": 4785
    },
    {
      "epoch": 0.25826992606982896,
      "grad_norm": 1.2565843972287152,
      "learning_rate": 8.598245754071606e-06,
      "loss": 0.6665,
      "step": 4786
    },
    {
      "epoch": 0.2583238896983433,
      "grad_norm": 1.0343178498932173,
      "learning_rate": 8.597692431266685e-06,
      "loss": 0.551,
      "step": 4787
    },
    {
      "epoch": 0.2583778533268577,
      "grad_norm": 0.877592881943391,
      "learning_rate": 8.597139019430954e-06,
      "loss": 0.3768,
      "step": 4788
    },
    {
      "epoch": 0.2584318169553721,
      "grad_norm": 0.9277158610719495,
      "learning_rate": 8.596585518580315e-06,
      "loss": 0.3807,
      "step": 4789
    },
    {
      "epoch": 0.25848578058388644,
      "grad_norm": 1.067034470540038,
      "learning_rate": 8.59603192873068e-06,
      "loss": 0.5563,
      "step": 4790
    },
    {
      "epoch": 0.25853974421240083,
      "grad_norm": 1.1430864644634144,
      "learning_rate": 8.595478249897961e-06,
      "loss": 0.5265,
      "step": 4791
    },
    {
      "epoch": 0.25859370784091523,
      "grad_norm": 0.9795650958628277,
      "learning_rate": 8.594924482098069e-06,
      "loss": 0.4726,
      "step": 4792
    },
    {
      "epoch": 0.25864767146942963,
      "grad_norm": 1.6141125782572077,
      "learning_rate": 8.59437062534692e-06,
      "loss": 0.6229,
      "step": 4793
    },
    {
      "epoch": 0.25870163509794397,
      "grad_norm": 1.09011336030748,
      "learning_rate": 8.59381667966043e-06,
      "loss": 0.5511,
      "step": 4794
    },
    {
      "epoch": 0.25875559872645837,
      "grad_norm": 0.962058958335601,
      "learning_rate": 8.593262645054525e-06,
      "loss": 0.4118,
      "step": 4795
    },
    {
      "epoch": 0.25880956235497277,
      "grad_norm": 1.0616623335983002,
      "learning_rate": 8.592708521545125e-06,
      "loss": 0.5242,
      "step": 4796
    },
    {
      "epoch": 0.2588635259834871,
      "grad_norm": 0.7732391387288949,
      "learning_rate": 8.592154309148159e-06,
      "loss": 0.4085,
      "step": 4797
    },
    {
      "epoch": 0.2589174896120015,
      "grad_norm": 1.0836440835062717,
      "learning_rate": 8.591600007879552e-06,
      "loss": 0.7179,
      "step": 4798
    },
    {
      "epoch": 0.2589714532405159,
      "grad_norm": 0.7692170737726116,
      "learning_rate": 8.591045617755238e-06,
      "loss": 0.329,
      "step": 4799
    },
    {
      "epoch": 0.2590254168690303,
      "grad_norm": 1.1732359820706788,
      "learning_rate": 8.590491138791148e-06,
      "loss": 0.6607,
      "step": 4800
    },
    {
      "epoch": 0.25907938049754464,
      "grad_norm": 1.0737690690773984,
      "learning_rate": 8.589936571003223e-06,
      "loss": 0.5566,
      "step": 4801
    },
    {
      "epoch": 0.25913334412605904,
      "grad_norm": 1.2648455408968318,
      "learning_rate": 8.589381914407397e-06,
      "loss": 0.6435,
      "step": 4802
    },
    {
      "epoch": 0.25918730775457344,
      "grad_norm": 1.1154547488812638,
      "learning_rate": 8.588827169019613e-06,
      "loss": 0.6349,
      "step": 4803
    },
    {
      "epoch": 0.2592412713830878,
      "grad_norm": 1.1333387201788943,
      "learning_rate": 8.588272334855814e-06,
      "loss": 0.5413,
      "step": 4804
    },
    {
      "epoch": 0.2592952350116022,
      "grad_norm": 2.2278715805878457,
      "learning_rate": 8.58771741193195e-06,
      "loss": 0.4389,
      "step": 4805
    },
    {
      "epoch": 0.2593491986401166,
      "grad_norm": 1.0932450822569113,
      "learning_rate": 8.587162400263968e-06,
      "loss": 0.547,
      "step": 4806
    },
    {
      "epoch": 0.2594031622686309,
      "grad_norm": 1.228222949934109,
      "learning_rate": 8.586607299867816e-06,
      "loss": 0.4077,
      "step": 4807
    },
    {
      "epoch": 0.2594571258971453,
      "grad_norm": 1.1545562886297172,
      "learning_rate": 8.586052110759452e-06,
      "loss": 0.5296,
      "step": 4808
    },
    {
      "epoch": 0.2595110895256597,
      "grad_norm": 1.3216791437503053,
      "learning_rate": 8.585496832954835e-06,
      "loss": 0.628,
      "step": 4809
    },
    {
      "epoch": 0.2595650531541741,
      "grad_norm": 1.0028265697163208,
      "learning_rate": 8.58494146646992e-06,
      "loss": 0.5562,
      "step": 4810
    },
    {
      "epoch": 0.25961901678268845,
      "grad_norm": 1.2758895808057582,
      "learning_rate": 8.584386011320669e-06,
      "loss": 0.4651,
      "step": 4811
    },
    {
      "epoch": 0.25967298041120285,
      "grad_norm": 1.1479876864300085,
      "learning_rate": 8.583830467523048e-06,
      "loss": 0.4849,
      "step": 4812
    },
    {
      "epoch": 0.25972694403971724,
      "grad_norm": 1.0633183976589873,
      "learning_rate": 8.583274835093024e-06,
      "loss": 0.5501,
      "step": 4813
    },
    {
      "epoch": 0.2597809076682316,
      "grad_norm": 1.1512892941996538,
      "learning_rate": 8.582719114046566e-06,
      "loss": 0.6046,
      "step": 4814
    },
    {
      "epoch": 0.259834871296746,
      "grad_norm": 1.2058438850158357,
      "learning_rate": 8.582163304399646e-06,
      "loss": 0.3819,
      "step": 4815
    },
    {
      "epoch": 0.2598888349252604,
      "grad_norm": 0.8939949582031429,
      "learning_rate": 8.581607406168238e-06,
      "loss": 0.3644,
      "step": 4816
    },
    {
      "epoch": 0.2599427985537748,
      "grad_norm": 1.031048229794335,
      "learning_rate": 8.581051419368319e-06,
      "loss": 0.4676,
      "step": 4817
    },
    {
      "epoch": 0.2599967621822891,
      "grad_norm": 0.8493255738632824,
      "learning_rate": 8.580495344015869e-06,
      "loss": 0.4403,
      "step": 4818
    },
    {
      "epoch": 0.2600507258108035,
      "grad_norm": 0.9576952536351344,
      "learning_rate": 8.57993918012687e-06,
      "loss": 0.3935,
      "step": 4819
    },
    {
      "epoch": 0.2601046894393179,
      "grad_norm": 1.0039936739113335,
      "learning_rate": 8.579382927717306e-06,
      "loss": 0.4725,
      "step": 4820
    },
    {
      "epoch": 0.26015865306783226,
      "grad_norm": 1.3889602150207494,
      "learning_rate": 8.578826586803168e-06,
      "loss": 0.525,
      "step": 4821
    },
    {
      "epoch": 0.26021261669634665,
      "grad_norm": 0.9748389840729356,
      "learning_rate": 8.578270157400441e-06,
      "loss": 0.4622,
      "step": 4822
    },
    {
      "epoch": 0.26026658032486105,
      "grad_norm": 1.0615486641873189,
      "learning_rate": 8.57771363952512e-06,
      "loss": 0.4935,
      "step": 4823
    },
    {
      "epoch": 0.26032054395337545,
      "grad_norm": 0.973205392136642,
      "learning_rate": 8.577157033193197e-06,
      "loss": 0.4625,
      "step": 4824
    },
    {
      "epoch": 0.2603745075818898,
      "grad_norm": 0.9057937014557815,
      "learning_rate": 8.576600338420676e-06,
      "loss": 0.4211,
      "step": 4825
    },
    {
      "epoch": 0.2604284712104042,
      "grad_norm": 1.0158299482663928,
      "learning_rate": 8.57604355522355e-06,
      "loss": 0.4789,
      "step": 4826
    },
    {
      "epoch": 0.2604824348389186,
      "grad_norm": 0.9249741783503165,
      "learning_rate": 8.575486683617823e-06,
      "loss": 0.3642,
      "step": 4827
    },
    {
      "epoch": 0.26053639846743293,
      "grad_norm": 0.8584313182640777,
      "learning_rate": 8.574929723619502e-06,
      "loss": 0.4919,
      "step": 4828
    },
    {
      "epoch": 0.2605903620959473,
      "grad_norm": 1.115599877706248,
      "learning_rate": 8.574372675244595e-06,
      "loss": 0.5763,
      "step": 4829
    },
    {
      "epoch": 0.2606443257244617,
      "grad_norm": 1.308816181793228,
      "learning_rate": 8.57381553850911e-06,
      "loss": 0.7704,
      "step": 4830
    },
    {
      "epoch": 0.2606982893529761,
      "grad_norm": 1.128873806299612,
      "learning_rate": 8.57325831342906e-06,
      "loss": 0.5451,
      "step": 4831
    },
    {
      "epoch": 0.26075225298149046,
      "grad_norm": 1.2670500738382617,
      "learning_rate": 8.572701000020462e-06,
      "loss": 0.5222,
      "step": 4832
    },
    {
      "epoch": 0.26080621661000486,
      "grad_norm": 1.1234382018255646,
      "learning_rate": 8.572143598299332e-06,
      "loss": 0.547,
      "step": 4833
    },
    {
      "epoch": 0.26086018023851926,
      "grad_norm": 1.125368324949549,
      "learning_rate": 8.57158610828169e-06,
      "loss": 0.5177,
      "step": 4834
    },
    {
      "epoch": 0.2609141438670336,
      "grad_norm": 1.072537057266257,
      "learning_rate": 8.571028529983561e-06,
      "loss": 0.5756,
      "step": 4835
    },
    {
      "epoch": 0.260968107495548,
      "grad_norm": 1.0296411196191309,
      "learning_rate": 8.570470863420968e-06,
      "loss": 0.4859,
      "step": 4836
    },
    {
      "epoch": 0.2610220711240624,
      "grad_norm": 0.9138756588227304,
      "learning_rate": 8.56991310860994e-06,
      "loss": 0.459,
      "step": 4837
    },
    {
      "epoch": 0.26107603475257674,
      "grad_norm": 0.9511752396065486,
      "learning_rate": 8.569355265566508e-06,
      "loss": 0.4063,
      "step": 4838
    },
    {
      "epoch": 0.26112999838109113,
      "grad_norm": 1.104098685877078,
      "learning_rate": 8.568797334306703e-06,
      "loss": 0.4338,
      "step": 4839
    },
    {
      "epoch": 0.26118396200960553,
      "grad_norm": 1.1786502798592364,
      "learning_rate": 8.568239314846563e-06,
      "loss": 0.4616,
      "step": 4840
    },
    {
      "epoch": 0.2612379256381199,
      "grad_norm": 1.244208070747673,
      "learning_rate": 8.567681207202124e-06,
      "loss": 0.4382,
      "step": 4841
    },
    {
      "epoch": 0.26129188926663427,
      "grad_norm": 1.0455108342913553,
      "learning_rate": 8.56712301138943e-06,
      "loss": 0.4487,
      "step": 4842
    },
    {
      "epoch": 0.26134585289514867,
      "grad_norm": 1.0602898811387886,
      "learning_rate": 8.566564727424518e-06,
      "loss": 0.4714,
      "step": 4843
    },
    {
      "epoch": 0.26139981652366306,
      "grad_norm": 1.083045654232036,
      "learning_rate": 8.566006355323439e-06,
      "loss": 0.5861,
      "step": 4844
    },
    {
      "epoch": 0.2614537801521774,
      "grad_norm": 0.9407909075359852,
      "learning_rate": 8.56544789510224e-06,
      "loss": 0.5161,
      "step": 4845
    },
    {
      "epoch": 0.2615077437806918,
      "grad_norm": 1.2272369157506438,
      "learning_rate": 8.56488934677697e-06,
      "loss": 0.6047,
      "step": 4846
    },
    {
      "epoch": 0.2615617074092062,
      "grad_norm": 0.7988983948925199,
      "learning_rate": 8.564330710363682e-06,
      "loss": 0.2992,
      "step": 4847
    },
    {
      "epoch": 0.2616156710377206,
      "grad_norm": 1.1346296053853278,
      "learning_rate": 8.563771985878437e-06,
      "loss": 0.54,
      "step": 4848
    },
    {
      "epoch": 0.26166963466623494,
      "grad_norm": 1.113461239886182,
      "learning_rate": 8.563213173337287e-06,
      "loss": 0.4907,
      "step": 4849
    },
    {
      "epoch": 0.26172359829474934,
      "grad_norm": 0.9232650349284862,
      "learning_rate": 8.562654272756294e-06,
      "loss": 0.463,
      "step": 4850
    },
    {
      "epoch": 0.26177756192326374,
      "grad_norm": 1.2781360042073813,
      "learning_rate": 8.562095284151524e-06,
      "loss": 0.5849,
      "step": 4851
    },
    {
      "epoch": 0.2618315255517781,
      "grad_norm": 1.103956884495305,
      "learning_rate": 8.561536207539041e-06,
      "loss": 0.5006,
      "step": 4852
    },
    {
      "epoch": 0.2618854891802925,
      "grad_norm": 1.1024423561607866,
      "learning_rate": 8.560977042934915e-06,
      "loss": 0.5186,
      "step": 4853
    },
    {
      "epoch": 0.26193945280880687,
      "grad_norm": 0.8615639776757269,
      "learning_rate": 8.560417790355215e-06,
      "loss": 0.4259,
      "step": 4854
    },
    {
      "epoch": 0.26199341643732127,
      "grad_norm": 1.050811366327978,
      "learning_rate": 8.559858449816016e-06,
      "loss": 0.4822,
      "step": 4855
    },
    {
      "epoch": 0.2620473800658356,
      "grad_norm": 0.9780810424185326,
      "learning_rate": 8.559299021333392e-06,
      "loss": 0.4573,
      "step": 4856
    },
    {
      "epoch": 0.26210134369435,
      "grad_norm": 1.0230735501133428,
      "learning_rate": 8.558739504923424e-06,
      "loss": 0.4432,
      "step": 4857
    },
    {
      "epoch": 0.2621553073228644,
      "grad_norm": 1.0485820287888148,
      "learning_rate": 8.558179900602192e-06,
      "loss": 0.4724,
      "step": 4858
    },
    {
      "epoch": 0.26220927095137875,
      "grad_norm": 1.0497026283300308,
      "learning_rate": 8.55762020838578e-06,
      "loss": 0.4598,
      "step": 4859
    },
    {
      "epoch": 0.26226323457989315,
      "grad_norm": 1.0007377310284702,
      "learning_rate": 8.557060428290274e-06,
      "loss": 0.415,
      "step": 4860
    },
    {
      "epoch": 0.26231719820840754,
      "grad_norm": 1.3460855932211668,
      "learning_rate": 8.556500560331761e-06,
      "loss": 0.5424,
      "step": 4861
    },
    {
      "epoch": 0.26237116183692194,
      "grad_norm": 1.158346629858631,
      "learning_rate": 8.555940604526334e-06,
      "loss": 0.5684,
      "step": 4862
    },
    {
      "epoch": 0.2624251254654363,
      "grad_norm": 0.9854158072195656,
      "learning_rate": 8.555380560890086e-06,
      "loss": 0.4751,
      "step": 4863
    },
    {
      "epoch": 0.2624790890939507,
      "grad_norm": 1.1011620411044138,
      "learning_rate": 8.554820429439114e-06,
      "loss": 0.4901,
      "step": 4864
    },
    {
      "epoch": 0.2625330527224651,
      "grad_norm": 1.0275198858264731,
      "learning_rate": 8.554260210189516e-06,
      "loss": 0.5032,
      "step": 4865
    },
    {
      "epoch": 0.2625870163509794,
      "grad_norm": 0.8822803516546975,
      "learning_rate": 8.553699903157394e-06,
      "loss": 0.4158,
      "step": 4866
    },
    {
      "epoch": 0.2626409799794938,
      "grad_norm": 0.7976567262389546,
      "learning_rate": 8.55313950835885e-06,
      "loss": 0.3918,
      "step": 4867
    },
    {
      "epoch": 0.2626949436080082,
      "grad_norm": 1.2190675538317952,
      "learning_rate": 8.552579025809991e-06,
      "loss": 0.6225,
      "step": 4868
    },
    {
      "epoch": 0.2627489072365226,
      "grad_norm": 0.9706520185451039,
      "learning_rate": 8.552018455526928e-06,
      "loss": 0.4973,
      "step": 4869
    },
    {
      "epoch": 0.26280287086503695,
      "grad_norm": 1.1317902498235322,
      "learning_rate": 8.551457797525772e-06,
      "loss": 0.5281,
      "step": 4870
    },
    {
      "epoch": 0.26285683449355135,
      "grad_norm": 0.8883466169361715,
      "learning_rate": 8.550897051822633e-06,
      "loss": 0.3693,
      "step": 4871
    },
    {
      "epoch": 0.26291079812206575,
      "grad_norm": 1.0641957214042368,
      "learning_rate": 8.550336218433631e-06,
      "loss": 0.5204,
      "step": 4872
    },
    {
      "epoch": 0.2629647617505801,
      "grad_norm": 1.1442668294680263,
      "learning_rate": 8.549775297374884e-06,
      "loss": 0.5386,
      "step": 4873
    },
    {
      "epoch": 0.2630187253790945,
      "grad_norm": 1.2115103196298687,
      "learning_rate": 8.549214288662514e-06,
      "loss": 0.6344,
      "step": 4874
    },
    {
      "epoch": 0.2630726890076089,
      "grad_norm": 0.8010888174486609,
      "learning_rate": 8.548653192312644e-06,
      "loss": 0.4189,
      "step": 4875
    },
    {
      "epoch": 0.2631266526361232,
      "grad_norm": 1.262382806077851,
      "learning_rate": 8.548092008341401e-06,
      "loss": 0.627,
      "step": 4876
    },
    {
      "epoch": 0.2631806162646376,
      "grad_norm": 1.0041215661238831,
      "learning_rate": 8.547530736764914e-06,
      "loss": 0.5993,
      "step": 4877
    },
    {
      "epoch": 0.263234579893152,
      "grad_norm": 0.8825581149445243,
      "learning_rate": 8.546969377599312e-06,
      "loss": 0.3771,
      "step": 4878
    },
    {
      "epoch": 0.2632885435216664,
      "grad_norm": 1.1668402127016324,
      "learning_rate": 8.546407930860733e-06,
      "loss": 0.498,
      "step": 4879
    },
    {
      "epoch": 0.26334250715018076,
      "grad_norm": 0.8934903740322111,
      "learning_rate": 8.545846396565311e-06,
      "loss": 0.4427,
      "step": 4880
    },
    {
      "epoch": 0.26339647077869516,
      "grad_norm": 1.3561539850324098,
      "learning_rate": 8.545284774729187e-06,
      "loss": 0.522,
      "step": 4881
    },
    {
      "epoch": 0.26345043440720955,
      "grad_norm": 1.002778313222618,
      "learning_rate": 8.544723065368502e-06,
      "loss": 0.5016,
      "step": 4882
    },
    {
      "epoch": 0.2635043980357239,
      "grad_norm": 1.0891355731923946,
      "learning_rate": 8.544161268499399e-06,
      "loss": 0.4248,
      "step": 4883
    },
    {
      "epoch": 0.2635583616642383,
      "grad_norm": 1.0929498516502842,
      "learning_rate": 8.543599384138025e-06,
      "loss": 0.5241,
      "step": 4884
    },
    {
      "epoch": 0.2636123252927527,
      "grad_norm": 0.9937245367859349,
      "learning_rate": 8.54303741230053e-06,
      "loss": 0.5438,
      "step": 4885
    },
    {
      "epoch": 0.2636662889212671,
      "grad_norm": 0.945441882544221,
      "learning_rate": 8.542475353003064e-06,
      "loss": 0.4658,
      "step": 4886
    },
    {
      "epoch": 0.26372025254978143,
      "grad_norm": 1.2119461221591072,
      "learning_rate": 8.541913206261783e-06,
      "loss": 0.507,
      "step": 4887
    },
    {
      "epoch": 0.26377421617829583,
      "grad_norm": 0.9805697269663831,
      "learning_rate": 8.541350972092843e-06,
      "loss": 0.423,
      "step": 4888
    },
    {
      "epoch": 0.2638281798068102,
      "grad_norm": 1.3733588413033384,
      "learning_rate": 8.540788650512401e-06,
      "loss": 0.4798,
      "step": 4889
    },
    {
      "epoch": 0.26388214343532457,
      "grad_norm": 0.9058825153051512,
      "learning_rate": 8.540226241536625e-06,
      "loss": 0.4633,
      "step": 4890
    },
    {
      "epoch": 0.26393610706383897,
      "grad_norm": 1.1885466969877745,
      "learning_rate": 8.539663745181671e-06,
      "loss": 0.4967,
      "step": 4891
    },
    {
      "epoch": 0.26399007069235336,
      "grad_norm": 1.3401219680281002,
      "learning_rate": 8.539101161463713e-06,
      "loss": 0.7331,
      "step": 4892
    },
    {
      "epoch": 0.26404403432086776,
      "grad_norm": 1.24458237950023,
      "learning_rate": 8.538538490398912e-06,
      "loss": 0.6674,
      "step": 4893
    },
    {
      "epoch": 0.2640979979493821,
      "grad_norm": 1.1992251325676055,
      "learning_rate": 8.537975732003447e-06,
      "loss": 0.5936,
      "step": 4894
    },
    {
      "epoch": 0.2641519615778965,
      "grad_norm": 1.1983170242143621,
      "learning_rate": 8.53741288629349e-06,
      "loss": 0.5659,
      "step": 4895
    },
    {
      "epoch": 0.2642059252064109,
      "grad_norm": 1.0960972870316203,
      "learning_rate": 8.53684995328522e-06,
      "loss": 0.5348,
      "step": 4896
    },
    {
      "epoch": 0.26425988883492524,
      "grad_norm": 1.1468152081448304,
      "learning_rate": 8.536286932994811e-06,
      "loss": 0.5875,
      "step": 4897
    },
    {
      "epoch": 0.26431385246343964,
      "grad_norm": 0.8837263576493469,
      "learning_rate": 8.535723825438447e-06,
      "loss": 0.3534,
      "step": 4898
    },
    {
      "epoch": 0.26436781609195403,
      "grad_norm": 1.083306136706758,
      "learning_rate": 8.535160630632312e-06,
      "loss": 0.5039,
      "step": 4899
    },
    {
      "epoch": 0.26442177972046843,
      "grad_norm": 0.8487465289892849,
      "learning_rate": 8.534597348592597e-06,
      "loss": 0.3825,
      "step": 4900
    },
    {
      "epoch": 0.2644757433489828,
      "grad_norm": 0.8917678155121717,
      "learning_rate": 8.534033979335487e-06,
      "loss": 0.4749,
      "step": 4901
    },
    {
      "epoch": 0.26452970697749717,
      "grad_norm": 1.063237531771167,
      "learning_rate": 8.533470522877174e-06,
      "loss": 0.425,
      "step": 4902
    },
    {
      "epoch": 0.26458367060601157,
      "grad_norm": 1.022171581995924,
      "learning_rate": 8.532906979233852e-06,
      "loss": 0.4709,
      "step": 4903
    },
    {
      "epoch": 0.2646376342345259,
      "grad_norm": 1.1751425070862658,
      "learning_rate": 8.532343348421721e-06,
      "loss": 0.598,
      "step": 4904
    },
    {
      "epoch": 0.2646915978630403,
      "grad_norm": 0.9413060174357715,
      "learning_rate": 8.531779630456978e-06,
      "loss": 0.4074,
      "step": 4905
    },
    {
      "epoch": 0.2647455614915547,
      "grad_norm": 1.0362937557240024,
      "learning_rate": 8.531215825355824e-06,
      "loss": 0.5182,
      "step": 4906
    },
    {
      "epoch": 0.26479952512006905,
      "grad_norm": 1.082178090386478,
      "learning_rate": 8.530651933134465e-06,
      "loss": 0.4217,
      "step": 4907
    },
    {
      "epoch": 0.26485348874858344,
      "grad_norm": 1.1357642393200462,
      "learning_rate": 8.530087953809107e-06,
      "loss": 0.5889,
      "step": 4908
    },
    {
      "epoch": 0.26490745237709784,
      "grad_norm": 0.7936709247551555,
      "learning_rate": 8.529523887395961e-06,
      "loss": 0.3453,
      "step": 4909
    },
    {
      "epoch": 0.26496141600561224,
      "grad_norm": 0.9507990614044645,
      "learning_rate": 8.528959733911235e-06,
      "loss": 0.4072,
      "step": 4910
    },
    {
      "epoch": 0.2650153796341266,
      "grad_norm": 0.8993053600341796,
      "learning_rate": 8.528395493371146e-06,
      "loss": 0.3692,
      "step": 4911
    },
    {
      "epoch": 0.265069343262641,
      "grad_norm": 1.114215692651555,
      "learning_rate": 8.527831165791912e-06,
      "loss": 0.5838,
      "step": 4912
    },
    {
      "epoch": 0.2651233068911554,
      "grad_norm": 1.039676739180002,
      "learning_rate": 8.52726675118975e-06,
      "loss": 0.5503,
      "step": 4913
    },
    {
      "epoch": 0.2651772705196697,
      "grad_norm": 0.9297622857246778,
      "learning_rate": 8.526702249580883e-06,
      "loss": 0.4542,
      "step": 4914
    },
    {
      "epoch": 0.2652312341481841,
      "grad_norm": 1.0250481983396156,
      "learning_rate": 8.526137660981534e-06,
      "loss": 0.4836,
      "step": 4915
    },
    {
      "epoch": 0.2652851977766985,
      "grad_norm": 1.0766404804137226,
      "learning_rate": 8.525572985407933e-06,
      "loss": 0.5208,
      "step": 4916
    },
    {
      "epoch": 0.2653391614052129,
      "grad_norm": 0.9764260789777678,
      "learning_rate": 8.525008222876306e-06,
      "loss": 0.391,
      "step": 4917
    },
    {
      "epoch": 0.26539312503372725,
      "grad_norm": 0.9706112110198597,
      "learning_rate": 8.524443373402886e-06,
      "loss": 0.5071,
      "step": 4918
    },
    {
      "epoch": 0.26544708866224165,
      "grad_norm": 1.0054135618310376,
      "learning_rate": 8.52387843700391e-06,
      "loss": 0.3903,
      "step": 4919
    },
    {
      "epoch": 0.26550105229075605,
      "grad_norm": 1.0689764702778293,
      "learning_rate": 8.523313413695607e-06,
      "loss": 0.5215,
      "step": 4920
    },
    {
      "epoch": 0.2655550159192704,
      "grad_norm": 1.1136459293285068,
      "learning_rate": 8.522748303494224e-06,
      "loss": 0.4378,
      "step": 4921
    },
    {
      "epoch": 0.2656089795477848,
      "grad_norm": 1.0620473495105778,
      "learning_rate": 8.522183106416003e-06,
      "loss": 0.514,
      "step": 4922
    },
    {
      "epoch": 0.2656629431762992,
      "grad_norm": 1.0800005742134127,
      "learning_rate": 8.521617822477184e-06,
      "loss": 0.5729,
      "step": 4923
    },
    {
      "epoch": 0.2657169068048136,
      "grad_norm": 1.29501546969231,
      "learning_rate": 8.521052451694016e-06,
      "loss": 0.6134,
      "step": 4924
    },
    {
      "epoch": 0.2657708704333279,
      "grad_norm": 1.0374424261155721,
      "learning_rate": 8.520486994082747e-06,
      "loss": 0.4423,
      "step": 4925
    },
    {
      "epoch": 0.2658248340618423,
      "grad_norm": 0.9416477634223471,
      "learning_rate": 8.519921449659631e-06,
      "loss": 0.4271,
      "step": 4926
    },
    {
      "epoch": 0.2658787976903567,
      "grad_norm": 1.0619787232161133,
      "learning_rate": 8.51935581844092e-06,
      "loss": 0.5613,
      "step": 4927
    },
    {
      "epoch": 0.26593276131887106,
      "grad_norm": 1.261604275249425,
      "learning_rate": 8.518790100442873e-06,
      "loss": 0.538,
      "step": 4928
    },
    {
      "epoch": 0.26598672494738546,
      "grad_norm": 0.8690823222648841,
      "learning_rate": 8.518224295681748e-06,
      "loss": 0.4226,
      "step": 4929
    },
    {
      "epoch": 0.26604068857589985,
      "grad_norm": 1.1368489490418927,
      "learning_rate": 8.517658404173807e-06,
      "loss": 0.5564,
      "step": 4930
    },
    {
      "epoch": 0.26609465220441425,
      "grad_norm": 0.8633800490488659,
      "learning_rate": 8.517092425935313e-06,
      "loss": 0.2955,
      "step": 4931
    },
    {
      "epoch": 0.2661486158329286,
      "grad_norm": 0.9210270128927458,
      "learning_rate": 8.516526360982535e-06,
      "loss": 0.4041,
      "step": 4932
    },
    {
      "epoch": 0.266202579461443,
      "grad_norm": 1.0228828297817194,
      "learning_rate": 8.515960209331743e-06,
      "loss": 0.4931,
      "step": 4933
    },
    {
      "epoch": 0.2662565430899574,
      "grad_norm": 1.1669014263107473,
      "learning_rate": 8.515393970999206e-06,
      "loss": 0.6052,
      "step": 4934
    },
    {
      "epoch": 0.26631050671847173,
      "grad_norm": 1.2023129409372522,
      "learning_rate": 8.5148276460012e-06,
      "loss": 0.573,
      "step": 4935
    },
    {
      "epoch": 0.2663644703469861,
      "grad_norm": 0.867230737750647,
      "learning_rate": 8.514261234354e-06,
      "loss": 0.4794,
      "step": 4936
    },
    {
      "epoch": 0.2664184339755005,
      "grad_norm": 1.0197875658726432,
      "learning_rate": 8.513694736073886e-06,
      "loss": 0.5603,
      "step": 4937
    },
    {
      "epoch": 0.2664723976040149,
      "grad_norm": 1.4296621744182947,
      "learning_rate": 8.513128151177141e-06,
      "loss": 0.6773,
      "step": 4938
    },
    {
      "epoch": 0.26652636123252926,
      "grad_norm": 0.9917595113619254,
      "learning_rate": 8.512561479680049e-06,
      "loss": 0.4913,
      "step": 4939
    },
    {
      "epoch": 0.26658032486104366,
      "grad_norm": 1.0137842238279766,
      "learning_rate": 8.511994721598897e-06,
      "loss": 0.5145,
      "step": 4940
    },
    {
      "epoch": 0.26663428848955806,
      "grad_norm": 1.1369499223294355,
      "learning_rate": 8.511427876949971e-06,
      "loss": 0.4622,
      "step": 4941
    },
    {
      "epoch": 0.2666882521180724,
      "grad_norm": 1.047667903117249,
      "learning_rate": 8.510860945749566e-06,
      "loss": 0.4462,
      "step": 4942
    },
    {
      "epoch": 0.2667422157465868,
      "grad_norm": 1.132218455917715,
      "learning_rate": 8.510293928013976e-06,
      "loss": 0.6244,
      "step": 4943
    },
    {
      "epoch": 0.2667961793751012,
      "grad_norm": 0.9104460672372381,
      "learning_rate": 8.509726823759496e-06,
      "loss": 0.4047,
      "step": 4944
    },
    {
      "epoch": 0.26685014300361554,
      "grad_norm": 1.0510498449981627,
      "learning_rate": 8.509159633002427e-06,
      "loss": 0.5043,
      "step": 4945
    },
    {
      "epoch": 0.26690410663212993,
      "grad_norm": 1.2238303298889028,
      "learning_rate": 8.508592355759067e-06,
      "loss": 0.7582,
      "step": 4946
    },
    {
      "epoch": 0.26695807026064433,
      "grad_norm": 1.1010395345598394,
      "learning_rate": 8.508024992045724e-06,
      "loss": 0.6039,
      "step": 4947
    },
    {
      "epoch": 0.26701203388915873,
      "grad_norm": 0.8863813520161247,
      "learning_rate": 8.507457541878701e-06,
      "loss": 0.4679,
      "step": 4948
    },
    {
      "epoch": 0.26706599751767307,
      "grad_norm": 0.889051889445057,
      "learning_rate": 8.506890005274314e-06,
      "loss": 0.3708,
      "step": 4949
    },
    {
      "epoch": 0.26711996114618747,
      "grad_norm": 1.104902030318065,
      "learning_rate": 8.506322382248868e-06,
      "loss": 0.4639,
      "step": 4950
    },
    {
      "epoch": 0.26717392477470187,
      "grad_norm": 1.043100441457956,
      "learning_rate": 8.50575467281868e-06,
      "loss": 0.4699,
      "step": 4951
    },
    {
      "epoch": 0.2672278884032162,
      "grad_norm": 1.273542753520947,
      "learning_rate": 8.505186877000064e-06,
      "loss": 0.6534,
      "step": 4952
    },
    {
      "epoch": 0.2672818520317306,
      "grad_norm": 1.0827495330451984,
      "learning_rate": 8.504618994809342e-06,
      "loss": 0.4986,
      "step": 4953
    },
    {
      "epoch": 0.267335815660245,
      "grad_norm": 1.0283460024200892,
      "learning_rate": 8.504051026262832e-06,
      "loss": 0.4184,
      "step": 4954
    },
    {
      "epoch": 0.2673897792887594,
      "grad_norm": 0.8188646051556691,
      "learning_rate": 8.503482971376862e-06,
      "loss": 0.3102,
      "step": 4955
    },
    {
      "epoch": 0.26744374291727374,
      "grad_norm": 1.2076085310113696,
      "learning_rate": 8.502914830167757e-06,
      "loss": 0.5662,
      "step": 4956
    },
    {
      "epoch": 0.26749770654578814,
      "grad_norm": 1.3076179903454865,
      "learning_rate": 8.502346602651844e-06,
      "loss": 0.5597,
      "step": 4957
    },
    {
      "epoch": 0.26755167017430254,
      "grad_norm": 1.0339943961687594,
      "learning_rate": 8.501778288845459e-06,
      "loss": 0.5177,
      "step": 4958
    },
    {
      "epoch": 0.2676056338028169,
      "grad_norm": 1.1157164637839228,
      "learning_rate": 8.501209888764928e-06,
      "loss": 0.6153,
      "step": 4959
    },
    {
      "epoch": 0.2676595974313313,
      "grad_norm": 1.400569285351183,
      "learning_rate": 8.500641402426597e-06,
      "loss": 0.572,
      "step": 4960
    },
    {
      "epoch": 0.2677135610598457,
      "grad_norm": 0.886495899557425,
      "learning_rate": 8.500072829846796e-06,
      "loss": 0.4799,
      "step": 4961
    },
    {
      "epoch": 0.26776752468836007,
      "grad_norm": 1.0416668842599197,
      "learning_rate": 8.499504171041873e-06,
      "loss": 0.5245,
      "step": 4962
    },
    {
      "epoch": 0.2678214883168744,
      "grad_norm": 0.9161240387089902,
      "learning_rate": 8.49893542602817e-06,
      "loss": 0.4196,
      "step": 4963
    },
    {
      "epoch": 0.2678754519453888,
      "grad_norm": 0.7864410285310075,
      "learning_rate": 8.498366594822031e-06,
      "loss": 0.3162,
      "step": 4964
    },
    {
      "epoch": 0.2679294155739032,
      "grad_norm": 0.8805952860698942,
      "learning_rate": 8.497797677439808e-06,
      "loss": 0.3687,
      "step": 4965
    },
    {
      "epoch": 0.26798337920241755,
      "grad_norm": 1.1513061744265791,
      "learning_rate": 8.49722867389785e-06,
      "loss": 0.4061,
      "step": 4966
    },
    {
      "epoch": 0.26803734283093195,
      "grad_norm": 1.1342884541900962,
      "learning_rate": 8.496659584212512e-06,
      "loss": 0.506,
      "step": 4967
    },
    {
      "epoch": 0.26809130645944634,
      "grad_norm": 1.038042241661417,
      "learning_rate": 8.496090408400148e-06,
      "loss": 0.3968,
      "step": 4968
    },
    {
      "epoch": 0.26814527008796074,
      "grad_norm": 1.1320287049013158,
      "learning_rate": 8.495521146477119e-06,
      "loss": 0.4365,
      "step": 4969
    },
    {
      "epoch": 0.2681992337164751,
      "grad_norm": 0.8811255769842016,
      "learning_rate": 8.494951798459785e-06,
      "loss": 0.5295,
      "step": 4970
    },
    {
      "epoch": 0.2682531973449895,
      "grad_norm": 1.0109772481397439,
      "learning_rate": 8.494382364364512e-06,
      "loss": 0.4931,
      "step": 4971
    },
    {
      "epoch": 0.2683071609735039,
      "grad_norm": 1.0770758110378873,
      "learning_rate": 8.493812844207663e-06,
      "loss": 0.4884,
      "step": 4972
    },
    {
      "epoch": 0.2683611246020182,
      "grad_norm": 0.9012339892556664,
      "learning_rate": 8.493243238005607e-06,
      "loss": 0.4217,
      "step": 4973
    },
    {
      "epoch": 0.2684150882305326,
      "grad_norm": 0.987062628624061,
      "learning_rate": 8.492673545774718e-06,
      "loss": 0.5145,
      "step": 4974
    },
    {
      "epoch": 0.268469051859047,
      "grad_norm": 0.9595980027413639,
      "learning_rate": 8.492103767531366e-06,
      "loss": 0.4872,
      "step": 4975
    },
    {
      "epoch": 0.26852301548756136,
      "grad_norm": 0.8470756611748547,
      "learning_rate": 8.491533903291929e-06,
      "loss": 0.3724,
      "step": 4976
    },
    {
      "epoch": 0.26857697911607575,
      "grad_norm": 0.9357610969520691,
      "learning_rate": 8.490963953072784e-06,
      "loss": 0.4369,
      "step": 4977
    },
    {
      "epoch": 0.26863094274459015,
      "grad_norm": 1.100253710923987,
      "learning_rate": 8.490393916890313e-06,
      "loss": 0.4602,
      "step": 4978
    },
    {
      "epoch": 0.26868490637310455,
      "grad_norm": 0.9279529459673218,
      "learning_rate": 8.489823794760898e-06,
      "loss": 0.384,
      "step": 4979
    },
    {
      "epoch": 0.2687388700016189,
      "grad_norm": 1.3728200765506169,
      "learning_rate": 8.48925358670093e-06,
      "loss": 0.4891,
      "step": 4980
    },
    {
      "epoch": 0.2687928336301333,
      "grad_norm": 0.8957614056233371,
      "learning_rate": 8.488683292726791e-06,
      "loss": 0.4558,
      "step": 4981
    },
    {
      "epoch": 0.2688467972586477,
      "grad_norm": 0.727349194462971,
      "learning_rate": 8.488112912854875e-06,
      "loss": 0.3004,
      "step": 4982
    },
    {
      "epoch": 0.268900760887162,
      "grad_norm": 1.1094459118875597,
      "learning_rate": 8.487542447101574e-06,
      "loss": 0.5764,
      "step": 4983
    },
    {
      "epoch": 0.2689547245156764,
      "grad_norm": 0.8575342359738434,
      "learning_rate": 8.486971895483285e-06,
      "loss": 0.3176,
      "step": 4984
    },
    {
      "epoch": 0.2690086881441908,
      "grad_norm": 1.016538916087091,
      "learning_rate": 8.486401258016404e-06,
      "loss": 0.6115,
      "step": 4985
    },
    {
      "epoch": 0.2690626517727052,
      "grad_norm": 0.9891547486858964,
      "learning_rate": 8.485830534717334e-06,
      "loss": 0.5239,
      "step": 4986
    },
    {
      "epoch": 0.26911661540121956,
      "grad_norm": 1.0167122264955455,
      "learning_rate": 8.485259725602478e-06,
      "loss": 0.4932,
      "step": 4987
    },
    {
      "epoch": 0.26917057902973396,
      "grad_norm": 1.1717447412479178,
      "learning_rate": 8.484688830688243e-06,
      "loss": 0.4245,
      "step": 4988
    },
    {
      "epoch": 0.26922454265824836,
      "grad_norm": 1.2847463695021393,
      "learning_rate": 8.48411784999103e-06,
      "loss": 0.5168,
      "step": 4989
    },
    {
      "epoch": 0.2692785062867627,
      "grad_norm": 1.3156048681390915,
      "learning_rate": 8.48354678352726e-06,
      "loss": 0.498,
      "step": 4990
    },
    {
      "epoch": 0.2693324699152771,
      "grad_norm": 1.0090518804458701,
      "learning_rate": 8.48297563131334e-06,
      "loss": 0.4793,
      "step": 4991
    },
    {
      "epoch": 0.2693864335437915,
      "grad_norm": 1.0475593233037548,
      "learning_rate": 8.482404393365685e-06,
      "loss": 0.5675,
      "step": 4992
    },
    {
      "epoch": 0.2694403971723059,
      "grad_norm": 1.0197142873335316,
      "learning_rate": 8.481833069700714e-06,
      "loss": 0.5549,
      "step": 4993
    },
    {
      "epoch": 0.26949436080082023,
      "grad_norm": 0.9922306095291865,
      "learning_rate": 8.481261660334849e-06,
      "loss": 0.405,
      "step": 4994
    },
    {
      "epoch": 0.26954832442933463,
      "grad_norm": 0.7935425579329741,
      "learning_rate": 8.48069016528451e-06,
      "loss": 0.344,
      "step": 4995
    },
    {
      "epoch": 0.269602288057849,
      "grad_norm": 0.9424635511999723,
      "learning_rate": 8.480118584566124e-06,
      "loss": 0.3767,
      "step": 4996
    },
    {
      "epoch": 0.26965625168636337,
      "grad_norm": 1.1325714535020714,
      "learning_rate": 8.479546918196119e-06,
      "loss": 0.5869,
      "step": 4997
    },
    {
      "epoch": 0.26971021531487777,
      "grad_norm": 1.0151959178975154,
      "learning_rate": 8.478975166190925e-06,
      "loss": 0.4681,
      "step": 4998
    },
    {
      "epoch": 0.26976417894339216,
      "grad_norm": 1.0493906335144532,
      "learning_rate": 8.478403328566974e-06,
      "loss": 0.4868,
      "step": 4999
    },
    {
      "epoch": 0.26981814257190656,
      "grad_norm": 1.0784442292445506,
      "learning_rate": 8.477831405340704e-06,
      "loss": 0.4453,
      "step": 5000
    },
    {
      "epoch": 0.26981814257190656,
      "eval_loss": 0.574369490146637,
      "eval_runtime": 160.0872,
      "eval_samples_per_second": 21.482,
      "eval_steps_per_second": 0.9,
      "step": 5000
    },
    {
      "epoch": 0.2698721062004209,
      "grad_norm": 1.0659093442740304,
      "learning_rate": 8.477259396528548e-06,
      "loss": 0.4954,
      "step": 5001
    },
    {
      "epoch": 0.2699260698289353,
      "grad_norm": 1.0070598494129597,
      "learning_rate": 8.47668730214695e-06,
      "loss": 0.4619,
      "step": 5002
    },
    {
      "epoch": 0.2699800334574497,
      "grad_norm": 1.2496494051013718,
      "learning_rate": 8.476115122212352e-06,
      "loss": 0.604,
      "step": 5003
    },
    {
      "epoch": 0.27003399708596404,
      "grad_norm": 0.9105106750094834,
      "learning_rate": 8.475542856741197e-06,
      "loss": 0.6233,
      "step": 5004
    },
    {
      "epoch": 0.27008796071447844,
      "grad_norm": 1.0788889917140985,
      "learning_rate": 8.474970505749934e-06,
      "loss": 0.4729,
      "step": 5005
    },
    {
      "epoch": 0.27014192434299283,
      "grad_norm": 1.057682448366313,
      "learning_rate": 8.474398069255012e-06,
      "loss": 0.5027,
      "step": 5006
    },
    {
      "epoch": 0.27019588797150723,
      "grad_norm": 1.0392111686279433,
      "learning_rate": 8.473825547272887e-06,
      "loss": 0.4613,
      "step": 5007
    },
    {
      "epoch": 0.2702498516000216,
      "grad_norm": 1.090733174890004,
      "learning_rate": 8.473252939820007e-06,
      "loss": 0.563,
      "step": 5008
    },
    {
      "epoch": 0.27030381522853597,
      "grad_norm": 1.1095414170467692,
      "learning_rate": 8.472680246912836e-06,
      "loss": 0.5478,
      "step": 5009
    },
    {
      "epoch": 0.27035777885705037,
      "grad_norm": 0.7964808135127515,
      "learning_rate": 8.472107468567829e-06,
      "loss": 0.399,
      "step": 5010
    },
    {
      "epoch": 0.2704117424855647,
      "grad_norm": 1.184984786508031,
      "learning_rate": 8.471534604801451e-06,
      "loss": 0.5999,
      "step": 5011
    },
    {
      "epoch": 0.2704657061140791,
      "grad_norm": 1.0160100527368403,
      "learning_rate": 8.470961655630166e-06,
      "loss": 0.4221,
      "step": 5012
    },
    {
      "epoch": 0.2705196697425935,
      "grad_norm": 1.1686602202508165,
      "learning_rate": 8.47038862107044e-06,
      "loss": 0.5626,
      "step": 5013
    },
    {
      "epoch": 0.27057363337110785,
      "grad_norm": 1.2751461168384453,
      "learning_rate": 8.469815501138747e-06,
      "loss": 0.5128,
      "step": 5014
    },
    {
      "epoch": 0.27062759699962224,
      "grad_norm": 1.0325140761141676,
      "learning_rate": 8.469242295851553e-06,
      "loss": 0.4823,
      "step": 5015
    },
    {
      "epoch": 0.27068156062813664,
      "grad_norm": 0.9454161180952948,
      "learning_rate": 8.468669005225335e-06,
      "loss": 0.6885,
      "step": 5016
    },
    {
      "epoch": 0.27073552425665104,
      "grad_norm": 0.9153203827656389,
      "learning_rate": 8.468095629276571e-06,
      "loss": 0.4928,
      "step": 5017
    },
    {
      "epoch": 0.2707894878851654,
      "grad_norm": 0.8143847071242002,
      "learning_rate": 8.467522168021741e-06,
      "loss": 0.3772,
      "step": 5018
    },
    {
      "epoch": 0.2708434515136798,
      "grad_norm": 1.1255830015338906,
      "learning_rate": 8.466948621477323e-06,
      "loss": 0.4239,
      "step": 5019
    },
    {
      "epoch": 0.2708974151421942,
      "grad_norm": 0.9930946754694855,
      "learning_rate": 8.466374989659805e-06,
      "loss": 0.5661,
      "step": 5020
    },
    {
      "epoch": 0.2709513787707085,
      "grad_norm": 1.061281546571973,
      "learning_rate": 8.465801272585672e-06,
      "loss": 0.493,
      "step": 5021
    },
    {
      "epoch": 0.2710053423992229,
      "grad_norm": 1.116821178570163,
      "learning_rate": 8.465227470271414e-06,
      "loss": 0.673,
      "step": 5022
    },
    {
      "epoch": 0.2710593060277373,
      "grad_norm": 0.8850789010441871,
      "learning_rate": 8.464653582733521e-06,
      "loss": 0.4216,
      "step": 5023
    },
    {
      "epoch": 0.2711132696562517,
      "grad_norm": 1.2158504254998508,
      "learning_rate": 8.464079609988492e-06,
      "loss": 0.6665,
      "step": 5024
    },
    {
      "epoch": 0.27116723328476605,
      "grad_norm": 1.1467165349665343,
      "learning_rate": 8.463505552052819e-06,
      "loss": 0.5846,
      "step": 5025
    },
    {
      "epoch": 0.27122119691328045,
      "grad_norm": 1.1285256474554934,
      "learning_rate": 8.462931408942999e-06,
      "loss": 0.4812,
      "step": 5026
    },
    {
      "epoch": 0.27127516054179485,
      "grad_norm": 1.0025213875598828,
      "learning_rate": 8.462357180675537e-06,
      "loss": 0.4575,
      "step": 5027
    },
    {
      "epoch": 0.2713291241703092,
      "grad_norm": 1.0703177228345229,
      "learning_rate": 8.46178286726694e-06,
      "loss": 0.5122,
      "step": 5028
    },
    {
      "epoch": 0.2713830877988236,
      "grad_norm": 0.911516789226813,
      "learning_rate": 8.461208468733708e-06,
      "loss": 0.3802,
      "step": 5029
    },
    {
      "epoch": 0.271437051427338,
      "grad_norm": 1.0006674963119548,
      "learning_rate": 8.460633985092352e-06,
      "loss": 0.5366,
      "step": 5030
    },
    {
      "epoch": 0.2714910150558524,
      "grad_norm": 1.1789945161060549,
      "learning_rate": 8.460059416359386e-06,
      "loss": 0.4999,
      "step": 5031
    },
    {
      "epoch": 0.2715449786843667,
      "grad_norm": 1.0395320282108502,
      "learning_rate": 8.45948476255132e-06,
      "loss": 0.3344,
      "step": 5032
    },
    {
      "epoch": 0.2715989423128811,
      "grad_norm": 0.9791130688822701,
      "learning_rate": 8.458910023684673e-06,
      "loss": 0.4367,
      "step": 5033
    },
    {
      "epoch": 0.2716529059413955,
      "grad_norm": 0.9840433360157309,
      "learning_rate": 8.45833519977596e-06,
      "loss": 0.3535,
      "step": 5034
    },
    {
      "epoch": 0.27170686956990986,
      "grad_norm": 0.9739610477492691,
      "learning_rate": 8.457760290841705e-06,
      "loss": 0.4588,
      "step": 5035
    },
    {
      "epoch": 0.27176083319842426,
      "grad_norm": 1.2253675091163532,
      "learning_rate": 8.457185296898431e-06,
      "loss": 0.6463,
      "step": 5036
    },
    {
      "epoch": 0.27181479682693865,
      "grad_norm": 1.1277861928319362,
      "learning_rate": 8.456610217962662e-06,
      "loss": 0.4692,
      "step": 5037
    },
    {
      "epoch": 0.27186876045545305,
      "grad_norm": 1.143112260972578,
      "learning_rate": 8.45603505405093e-06,
      "loss": 0.5268,
      "step": 5038
    },
    {
      "epoch": 0.2719227240839674,
      "grad_norm": 1.0815323573744626,
      "learning_rate": 8.455459805179764e-06,
      "loss": 0.5321,
      "step": 5039
    },
    {
      "epoch": 0.2719766877124818,
      "grad_norm": 1.197674826231189,
      "learning_rate": 8.454884471365694e-06,
      "loss": 0.6476,
      "step": 5040
    },
    {
      "epoch": 0.2720306513409962,
      "grad_norm": 0.9576522395300567,
      "learning_rate": 8.45430905262526e-06,
      "loss": 0.5474,
      "step": 5041
    },
    {
      "epoch": 0.27208461496951053,
      "grad_norm": 0.9074578102020335,
      "learning_rate": 8.453733548975001e-06,
      "loss": 0.4108,
      "step": 5042
    },
    {
      "epoch": 0.2721385785980249,
      "grad_norm": 0.964708093340206,
      "learning_rate": 8.453157960431453e-06,
      "loss": 0.451,
      "step": 5043
    },
    {
      "epoch": 0.2721925422265393,
      "grad_norm": 1.1017540635767715,
      "learning_rate": 8.452582287011161e-06,
      "loss": 0.739,
      "step": 5044
    },
    {
      "epoch": 0.27224650585505367,
      "grad_norm": 1.3695125860384991,
      "learning_rate": 8.452006528730672e-06,
      "loss": 0.6403,
      "step": 5045
    },
    {
      "epoch": 0.27230046948356806,
      "grad_norm": 1.146474040394263,
      "learning_rate": 8.451430685606532e-06,
      "loss": 0.5609,
      "step": 5046
    },
    {
      "epoch": 0.27235443311208246,
      "grad_norm": 0.8505454320237756,
      "learning_rate": 8.450854757655291e-06,
      "loss": 0.3811,
      "step": 5047
    },
    {
      "epoch": 0.27240839674059686,
      "grad_norm": 0.9784521947218713,
      "learning_rate": 8.450278744893504e-06,
      "loss": 0.4416,
      "step": 5048
    },
    {
      "epoch": 0.2724623603691112,
      "grad_norm": 1.295507548389036,
      "learning_rate": 8.449702647337724e-06,
      "loss": 0.6001,
      "step": 5049
    },
    {
      "epoch": 0.2725163239976256,
      "grad_norm": 0.9381827749944345,
      "learning_rate": 8.44912646500451e-06,
      "loss": 0.4663,
      "step": 5050
    },
    {
      "epoch": 0.27257028762614,
      "grad_norm": 0.7350919808304873,
      "learning_rate": 8.448550197910421e-06,
      "loss": 0.3603,
      "step": 5051
    },
    {
      "epoch": 0.27262425125465434,
      "grad_norm": 1.126397535695852,
      "learning_rate": 8.447973846072021e-06,
      "loss": 0.6336,
      "step": 5052
    },
    {
      "epoch": 0.27267821488316873,
      "grad_norm": 0.9048991170752444,
      "learning_rate": 8.447397409505874e-06,
      "loss": 0.4232,
      "step": 5053
    },
    {
      "epoch": 0.27273217851168313,
      "grad_norm": 1.1225133092607755,
      "learning_rate": 8.446820888228547e-06,
      "loss": 0.596,
      "step": 5054
    },
    {
      "epoch": 0.27278614214019753,
      "grad_norm": 1.2472138112996929,
      "learning_rate": 8.446244282256609e-06,
      "loss": 0.5805,
      "step": 5055
    },
    {
      "epoch": 0.27284010576871187,
      "grad_norm": 0.9917401274010875,
      "learning_rate": 8.445667591606634e-06,
      "loss": 0.4142,
      "step": 5056
    },
    {
      "epoch": 0.27289406939722627,
      "grad_norm": 0.902867282072962,
      "learning_rate": 8.445090816295195e-06,
      "loss": 0.4012,
      "step": 5057
    },
    {
      "epoch": 0.27294803302574067,
      "grad_norm": 1.1800769062465017,
      "learning_rate": 8.44451395633887e-06,
      "loss": 0.7055,
      "step": 5058
    },
    {
      "epoch": 0.273001996654255,
      "grad_norm": 1.10792888884941,
      "learning_rate": 8.443937011754242e-06,
      "loss": 0.4622,
      "step": 5059
    },
    {
      "epoch": 0.2730559602827694,
      "grad_norm": 1.173149378108911,
      "learning_rate": 8.443359982557887e-06,
      "loss": 0.5471,
      "step": 5060
    },
    {
      "epoch": 0.2731099239112838,
      "grad_norm": 1.0168270021926529,
      "learning_rate": 8.442782868766394e-06,
      "loss": 0.433,
      "step": 5061
    },
    {
      "epoch": 0.2731638875397982,
      "grad_norm": 1.1966712446255392,
      "learning_rate": 8.442205670396344e-06,
      "loss": 0.4853,
      "step": 5062
    },
    {
      "epoch": 0.27321785116831254,
      "grad_norm": 1.0627865811473995,
      "learning_rate": 8.441628387464335e-06,
      "loss": 0.5376,
      "step": 5063
    },
    {
      "epoch": 0.27327181479682694,
      "grad_norm": 0.9919371942135345,
      "learning_rate": 8.44105101998695e-06,
      "loss": 0.4975,
      "step": 5064
    },
    {
      "epoch": 0.27332577842534134,
      "grad_norm": 1.1146194685456778,
      "learning_rate": 8.44047356798079e-06,
      "loss": 0.6138,
      "step": 5065
    },
    {
      "epoch": 0.2733797420538557,
      "grad_norm": 1.168554625383551,
      "learning_rate": 8.439896031462449e-06,
      "loss": 0.4881,
      "step": 5066
    },
    {
      "epoch": 0.2734337056823701,
      "grad_norm": 0.8791749401725523,
      "learning_rate": 8.439318410448522e-06,
      "loss": 0.4225,
      "step": 5067
    },
    {
      "epoch": 0.2734876693108845,
      "grad_norm": 1.2825232490679803,
      "learning_rate": 8.438740704955615e-06,
      "loss": 0.6178,
      "step": 5068
    },
    {
      "epoch": 0.27354163293939887,
      "grad_norm": 1.3052020244565279,
      "learning_rate": 8.43816291500033e-06,
      "loss": 0.5441,
      "step": 5069
    },
    {
      "epoch": 0.2735955965679132,
      "grad_norm": 0.9120281730842009,
      "learning_rate": 8.437585040599276e-06,
      "loss": 0.4658,
      "step": 5070
    },
    {
      "epoch": 0.2736495601964276,
      "grad_norm": 0.9253218427223191,
      "learning_rate": 8.437007081769059e-06,
      "loss": 0.4107,
      "step": 5071
    },
    {
      "epoch": 0.273703523824942,
      "grad_norm": 0.99747864890601,
      "learning_rate": 8.436429038526291e-06,
      "loss": 0.5365,
      "step": 5072
    },
    {
      "epoch": 0.27375748745345635,
      "grad_norm": 0.8795267358098525,
      "learning_rate": 8.435850910887585e-06,
      "loss": 0.3622,
      "step": 5073
    },
    {
      "epoch": 0.27381145108197075,
      "grad_norm": 0.9495593184306,
      "learning_rate": 8.435272698869558e-06,
      "loss": 0.4213,
      "step": 5074
    },
    {
      "epoch": 0.27386541471048514,
      "grad_norm": 0.9732769515514071,
      "learning_rate": 8.434694402488828e-06,
      "loss": 0.5532,
      "step": 5075
    },
    {
      "epoch": 0.27391937833899954,
      "grad_norm": 1.238460596029063,
      "learning_rate": 8.434116021762013e-06,
      "loss": 0.4655,
      "step": 5076
    },
    {
      "epoch": 0.2739733419675139,
      "grad_norm": 1.023957344398195,
      "learning_rate": 8.433537556705741e-06,
      "loss": 0.467,
      "step": 5077
    },
    {
      "epoch": 0.2740273055960283,
      "grad_norm": 1.176393334647037,
      "learning_rate": 8.432959007336637e-06,
      "loss": 0.5832,
      "step": 5078
    },
    {
      "epoch": 0.2740812692245427,
      "grad_norm": 1.1472160680536063,
      "learning_rate": 8.432380373671325e-06,
      "loss": 0.7125,
      "step": 5079
    },
    {
      "epoch": 0.274135232853057,
      "grad_norm": 0.9315799581981671,
      "learning_rate": 8.43180165572644e-06,
      "loss": 0.4408,
      "step": 5080
    },
    {
      "epoch": 0.2741891964815714,
      "grad_norm": 1.0141928686721873,
      "learning_rate": 8.43122285351861e-06,
      "loss": 0.3821,
      "step": 5081
    },
    {
      "epoch": 0.2742431601100858,
      "grad_norm": 1.1766900985608466,
      "learning_rate": 8.430643967064476e-06,
      "loss": 0.5704,
      "step": 5082
    },
    {
      "epoch": 0.27429712373860016,
      "grad_norm": 0.9988370693113201,
      "learning_rate": 8.430064996380672e-06,
      "loss": 0.6988,
      "step": 5083
    },
    {
      "epoch": 0.27435108736711455,
      "grad_norm": 0.9876301173428249,
      "learning_rate": 8.42948594148384e-06,
      "loss": 0.3864,
      "step": 5084
    },
    {
      "epoch": 0.27440505099562895,
      "grad_norm": 1.0257953218399134,
      "learning_rate": 8.428906802390623e-06,
      "loss": 0.488,
      "step": 5085
    },
    {
      "epoch": 0.27445901462414335,
      "grad_norm": 1.0362404675287735,
      "learning_rate": 8.428327579117664e-06,
      "loss": 0.4953,
      "step": 5086
    },
    {
      "epoch": 0.2745129782526577,
      "grad_norm": 1.168402919507919,
      "learning_rate": 8.427748271681612e-06,
      "loss": 0.5544,
      "step": 5087
    },
    {
      "epoch": 0.2745669418811721,
      "grad_norm": 0.9326917561398205,
      "learning_rate": 8.427168880099115e-06,
      "loss": 0.4308,
      "step": 5088
    },
    {
      "epoch": 0.2746209055096865,
      "grad_norm": 0.9266602203628659,
      "learning_rate": 8.426589404386827e-06,
      "loss": 0.4007,
      "step": 5089
    },
    {
      "epoch": 0.27467486913820083,
      "grad_norm": 0.9349769053192981,
      "learning_rate": 8.426009844561403e-06,
      "loss": 0.6142,
      "step": 5090
    },
    {
      "epoch": 0.2747288327667152,
      "grad_norm": 1.112542268307128,
      "learning_rate": 8.4254302006395e-06,
      "loss": 0.5465,
      "step": 5091
    },
    {
      "epoch": 0.2747827963952296,
      "grad_norm": 1.1530111627123463,
      "learning_rate": 8.424850472637777e-06,
      "loss": 0.6653,
      "step": 5092
    },
    {
      "epoch": 0.274836760023744,
      "grad_norm": 1.1455839179672693,
      "learning_rate": 8.424270660572896e-06,
      "loss": 0.635,
      "step": 5093
    },
    {
      "epoch": 0.27489072365225836,
      "grad_norm": 1.259626763088531,
      "learning_rate": 8.423690764461522e-06,
      "loss": 0.6676,
      "step": 5094
    },
    {
      "epoch": 0.27494468728077276,
      "grad_norm": 0.9358174837673061,
      "learning_rate": 8.42311078432032e-06,
      "loss": 0.4808,
      "step": 5095
    },
    {
      "epoch": 0.27499865090928716,
      "grad_norm": 0.9240641023458724,
      "learning_rate": 8.422530720165962e-06,
      "loss": 0.3619,
      "step": 5096
    },
    {
      "epoch": 0.2750526145378015,
      "grad_norm": 1.2744223508868635,
      "learning_rate": 8.421950572015115e-06,
      "loss": 0.6097,
      "step": 5097
    },
    {
      "epoch": 0.2751065781663159,
      "grad_norm": 0.8359664883577728,
      "learning_rate": 8.421370339884459e-06,
      "loss": 0.4016,
      "step": 5098
    },
    {
      "epoch": 0.2751605417948303,
      "grad_norm": 1.0883275352335537,
      "learning_rate": 8.420790023790665e-06,
      "loss": 0.5547,
      "step": 5099
    },
    {
      "epoch": 0.2752145054233447,
      "grad_norm": 1.1598786792810918,
      "learning_rate": 8.420209623750418e-06,
      "loss": 0.6056,
      "step": 5100
    },
    {
      "epoch": 0.27526846905185903,
      "grad_norm": 1.028744452882345,
      "learning_rate": 8.419629139780393e-06,
      "loss": 0.4726,
      "step": 5101
    },
    {
      "epoch": 0.27532243268037343,
      "grad_norm": 0.9337552247886279,
      "learning_rate": 8.419048571897278e-06,
      "loss": 0.4362,
      "step": 5102
    },
    {
      "epoch": 0.2753763963088878,
      "grad_norm": 1.0963768947238157,
      "learning_rate": 8.418467920117756e-06,
      "loss": 0.5055,
      "step": 5103
    },
    {
      "epoch": 0.27543035993740217,
      "grad_norm": 0.893469145270453,
      "learning_rate": 8.41788718445852e-06,
      "loss": 0.4225,
      "step": 5104
    },
    {
      "epoch": 0.27548432356591657,
      "grad_norm": 1.176830963694303,
      "learning_rate": 8.417306364936256e-06,
      "loss": 0.5374,
      "step": 5105
    },
    {
      "epoch": 0.27553828719443096,
      "grad_norm": 1.0025504708369914,
      "learning_rate": 8.416725461567659e-06,
      "loss": 0.6486,
      "step": 5106
    },
    {
      "epoch": 0.27559225082294536,
      "grad_norm": 1.0959336492329297,
      "learning_rate": 8.416144474369426e-06,
      "loss": 0.4322,
      "step": 5107
    },
    {
      "epoch": 0.2756462144514597,
      "grad_norm": 1.1516850452897203,
      "learning_rate": 8.415563403358253e-06,
      "loss": 0.4025,
      "step": 5108
    },
    {
      "epoch": 0.2757001780799741,
      "grad_norm": 1.0731944077722275,
      "learning_rate": 8.414982248550843e-06,
      "loss": 0.4932,
      "step": 5109
    },
    {
      "epoch": 0.2757541417084885,
      "grad_norm": 0.7721074221055555,
      "learning_rate": 8.414401009963897e-06,
      "loss": 0.2987,
      "step": 5110
    },
    {
      "epoch": 0.27580810533700284,
      "grad_norm": 1.2119601217712772,
      "learning_rate": 8.413819687614122e-06,
      "loss": 0.5232,
      "step": 5111
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 1.3612455101183176,
      "learning_rate": 8.413238281518225e-06,
      "loss": 0.5678,
      "step": 5112
    },
    {
      "epoch": 0.27591603259403163,
      "grad_norm": 1.4390519145365697,
      "learning_rate": 8.412656791692917e-06,
      "loss": 0.7584,
      "step": 5113
    },
    {
      "epoch": 0.275969996222546,
      "grad_norm": 1.0156598127384096,
      "learning_rate": 8.412075218154908e-06,
      "loss": 0.4622,
      "step": 5114
    },
    {
      "epoch": 0.2760239598510604,
      "grad_norm": 1.0791792903533781,
      "learning_rate": 8.411493560920915e-06,
      "loss": 0.5272,
      "step": 5115
    },
    {
      "epoch": 0.27607792347957477,
      "grad_norm": 1.045566604854827,
      "learning_rate": 8.410911820007656e-06,
      "loss": 0.5111,
      "step": 5116
    },
    {
      "epoch": 0.27613188710808917,
      "grad_norm": 1.007578660937908,
      "learning_rate": 8.410329995431848e-06,
      "loss": 0.3793,
      "step": 5117
    },
    {
      "epoch": 0.2761858507366035,
      "grad_norm": 1.1568028031999509,
      "learning_rate": 8.409748087210217e-06,
      "loss": 0.4382,
      "step": 5118
    },
    {
      "epoch": 0.2762398143651179,
      "grad_norm": 0.9806255099862317,
      "learning_rate": 8.409166095359487e-06,
      "loss": 0.48,
      "step": 5119
    },
    {
      "epoch": 0.2762937779936323,
      "grad_norm": 0.9619326671059396,
      "learning_rate": 8.408584019896383e-06,
      "loss": 0.4683,
      "step": 5120
    },
    {
      "epoch": 0.27634774162214665,
      "grad_norm": 1.020673975996857,
      "learning_rate": 8.408001860837635e-06,
      "loss": 0.4197,
      "step": 5121
    },
    {
      "epoch": 0.27640170525066105,
      "grad_norm": 0.8747233951471021,
      "learning_rate": 8.407419618199975e-06,
      "loss": 0.3709,
      "step": 5122
    },
    {
      "epoch": 0.27645566887917544,
      "grad_norm": 0.8181535948024199,
      "learning_rate": 8.406837292000138e-06,
      "loss": 0.3481,
      "step": 5123
    },
    {
      "epoch": 0.27650963250768984,
      "grad_norm": 1.2639556375780803,
      "learning_rate": 8.40625488225486e-06,
      "loss": 0.5555,
      "step": 5124
    },
    {
      "epoch": 0.2765635961362042,
      "grad_norm": 1.0061299410825792,
      "learning_rate": 8.405672388980879e-06,
      "loss": 0.5004,
      "step": 5125
    },
    {
      "epoch": 0.2766175597647186,
      "grad_norm": 1.0941971069456888,
      "learning_rate": 8.405089812194938e-06,
      "loss": 0.4235,
      "step": 5126
    },
    {
      "epoch": 0.276671523393233,
      "grad_norm": 1.2281585729837157,
      "learning_rate": 8.404507151913782e-06,
      "loss": 0.6125,
      "step": 5127
    },
    {
      "epoch": 0.2767254870217473,
      "grad_norm": 0.9635223137437686,
      "learning_rate": 8.403924408154154e-06,
      "loss": 0.4063,
      "step": 5128
    },
    {
      "epoch": 0.2767794506502617,
      "grad_norm": 1.0648375165815833,
      "learning_rate": 8.403341580932805e-06,
      "loss": 0.5049,
      "step": 5129
    },
    {
      "epoch": 0.2768334142787761,
      "grad_norm": 0.998859144410337,
      "learning_rate": 8.402758670266486e-06,
      "loss": 0.3863,
      "step": 5130
    },
    {
      "epoch": 0.2768873779072905,
      "grad_norm": 0.964514918225284,
      "learning_rate": 8.402175676171947e-06,
      "loss": 0.6031,
      "step": 5131
    },
    {
      "epoch": 0.27694134153580485,
      "grad_norm": 0.9582730367230397,
      "learning_rate": 8.401592598665949e-06,
      "loss": 0.5754,
      "step": 5132
    },
    {
      "epoch": 0.27699530516431925,
      "grad_norm": 1.097272178854074,
      "learning_rate": 8.401009437765248e-06,
      "loss": 0.5147,
      "step": 5133
    },
    {
      "epoch": 0.27704926879283365,
      "grad_norm": 0.9315169631393253,
      "learning_rate": 8.400426193486606e-06,
      "loss": 0.4174,
      "step": 5134
    },
    {
      "epoch": 0.277103232421348,
      "grad_norm": 1.0025438596871863,
      "learning_rate": 8.399842865846781e-06,
      "loss": 0.5341,
      "step": 5135
    },
    {
      "epoch": 0.2771571960498624,
      "grad_norm": 1.1752054780841596,
      "learning_rate": 8.399259454862545e-06,
      "loss": 0.518,
      "step": 5136
    },
    {
      "epoch": 0.2772111596783768,
      "grad_norm": 1.0799194183602907,
      "learning_rate": 8.39867596055066e-06,
      "loss": 0.5096,
      "step": 5137
    },
    {
      "epoch": 0.2772651233068912,
      "grad_norm": 1.2059157867343997,
      "learning_rate": 8.3980923829279e-06,
      "loss": 0.5018,
      "step": 5138
    },
    {
      "epoch": 0.2773190869354055,
      "grad_norm": 1.0835478518157242,
      "learning_rate": 8.397508722011037e-06,
      "loss": 0.4718,
      "step": 5139
    },
    {
      "epoch": 0.2773730505639199,
      "grad_norm": 1.0264742609630149,
      "learning_rate": 8.396924977816845e-06,
      "loss": 0.5005,
      "step": 5140
    },
    {
      "epoch": 0.2774270141924343,
      "grad_norm": 0.804563495151556,
      "learning_rate": 8.396341150362102e-06,
      "loss": 0.3667,
      "step": 5141
    },
    {
      "epoch": 0.27748097782094866,
      "grad_norm": 0.9031802363689143,
      "learning_rate": 8.395757239663589e-06,
      "loss": 0.3818,
      "step": 5142
    },
    {
      "epoch": 0.27753494144946306,
      "grad_norm": 1.020885677895989,
      "learning_rate": 8.395173245738087e-06,
      "loss": 0.5764,
      "step": 5143
    },
    {
      "epoch": 0.27758890507797745,
      "grad_norm": 1.2462240574241217,
      "learning_rate": 8.39458916860238e-06,
      "loss": 0.5994,
      "step": 5144
    },
    {
      "epoch": 0.27764286870649185,
      "grad_norm": 1.0495763160945715,
      "learning_rate": 8.394005008273255e-06,
      "loss": 0.4174,
      "step": 5145
    },
    {
      "epoch": 0.2776968323350062,
      "grad_norm": 1.2908751391208328,
      "learning_rate": 8.393420764767503e-06,
      "loss": 0.5662,
      "step": 5146
    },
    {
      "epoch": 0.2777507959635206,
      "grad_norm": 1.0671278977322058,
      "learning_rate": 8.392836438101914e-06,
      "loss": 0.5382,
      "step": 5147
    },
    {
      "epoch": 0.277804759592035,
      "grad_norm": 0.9417373730120165,
      "learning_rate": 8.39225202829328e-06,
      "loss": 0.3589,
      "step": 5148
    },
    {
      "epoch": 0.27785872322054933,
      "grad_norm": 1.0548223592855124,
      "learning_rate": 8.391667535358405e-06,
      "loss": 0.4992,
      "step": 5149
    },
    {
      "epoch": 0.27791268684906373,
      "grad_norm": 1.1374598782576733,
      "learning_rate": 8.39108295931408e-06,
      "loss": 0.5503,
      "step": 5150
    },
    {
      "epoch": 0.2779666504775781,
      "grad_norm": 1.0293922911470434,
      "learning_rate": 8.390498300177111e-06,
      "loss": 0.4669,
      "step": 5151
    },
    {
      "epoch": 0.27802061410609247,
      "grad_norm": 1.0088674529971597,
      "learning_rate": 8.3899135579643e-06,
      "loss": 0.486,
      "step": 5152
    },
    {
      "epoch": 0.27807457773460686,
      "grad_norm": 1.1140643532631689,
      "learning_rate": 8.389328732692454e-06,
      "loss": 0.5382,
      "step": 5153
    },
    {
      "epoch": 0.27812854136312126,
      "grad_norm": 1.098818075820503,
      "learning_rate": 8.38874382437838e-06,
      "loss": 0.5075,
      "step": 5154
    },
    {
      "epoch": 0.27818250499163566,
      "grad_norm": 1.1217444561658938,
      "learning_rate": 8.388158833038888e-06,
      "loss": 0.5473,
      "step": 5155
    },
    {
      "epoch": 0.27823646862015,
      "grad_norm": 1.0302331305246388,
      "learning_rate": 8.387573758690796e-06,
      "loss": 0.5617,
      "step": 5156
    },
    {
      "epoch": 0.2782904322486644,
      "grad_norm": 0.9602298764945383,
      "learning_rate": 8.386988601350916e-06,
      "loss": 0.4704,
      "step": 5157
    },
    {
      "epoch": 0.2783443958771788,
      "grad_norm": 1.1842450604403707,
      "learning_rate": 8.386403361036064e-06,
      "loss": 0.6777,
      "step": 5158
    },
    {
      "epoch": 0.27839835950569314,
      "grad_norm": 1.0848056302635525,
      "learning_rate": 8.385818037763064e-06,
      "loss": 0.6154,
      "step": 5159
    },
    {
      "epoch": 0.27845232313420754,
      "grad_norm": 1.321408288417881,
      "learning_rate": 8.38523263154874e-06,
      "loss": 0.5374,
      "step": 5160
    },
    {
      "epoch": 0.27850628676272193,
      "grad_norm": 0.9098792536009153,
      "learning_rate": 8.384647142409911e-06,
      "loss": 0.4813,
      "step": 5161
    },
    {
      "epoch": 0.27856025039123633,
      "grad_norm": 1.0456677386888666,
      "learning_rate": 8.384061570363411e-06,
      "loss": 0.5235,
      "step": 5162
    },
    {
      "epoch": 0.2786142140197507,
      "grad_norm": 1.1096475108821384,
      "learning_rate": 8.383475915426067e-06,
      "loss": 0.5031,
      "step": 5163
    },
    {
      "epoch": 0.27866817764826507,
      "grad_norm": 1.023470968276582,
      "learning_rate": 8.382890177614711e-06,
      "loss": 0.4926,
      "step": 5164
    },
    {
      "epoch": 0.27872214127677947,
      "grad_norm": 1.1851831795830645,
      "learning_rate": 8.382304356946178e-06,
      "loss": 0.6504,
      "step": 5165
    },
    {
      "epoch": 0.2787761049052938,
      "grad_norm": 1.079702018321339,
      "learning_rate": 8.381718453437307e-06,
      "loss": 0.5236,
      "step": 5166
    },
    {
      "epoch": 0.2788300685338082,
      "grad_norm": 0.9268666736659805,
      "learning_rate": 8.381132467104936e-06,
      "loss": 0.5269,
      "step": 5167
    },
    {
      "epoch": 0.2788840321623226,
      "grad_norm": 0.8786112763719611,
      "learning_rate": 8.380546397965904e-06,
      "loss": 0.4765,
      "step": 5168
    },
    {
      "epoch": 0.278937995790837,
      "grad_norm": 0.9877041493944586,
      "learning_rate": 8.37996024603706e-06,
      "loss": 0.4075,
      "step": 5169
    },
    {
      "epoch": 0.27899195941935134,
      "grad_norm": 0.8463161806803607,
      "learning_rate": 8.379374011335248e-06,
      "loss": 0.4173,
      "step": 5170
    },
    {
      "epoch": 0.27904592304786574,
      "grad_norm": 1.0933164236708315,
      "learning_rate": 8.378787693877318e-06,
      "loss": 0.6097,
      "step": 5171
    },
    {
      "epoch": 0.27909988667638014,
      "grad_norm": 0.9657070377376511,
      "learning_rate": 8.378201293680121e-06,
      "loss": 0.4661,
      "step": 5172
    },
    {
      "epoch": 0.2791538503048945,
      "grad_norm": 0.9071057576735806,
      "learning_rate": 8.377614810760509e-06,
      "loss": 0.3571,
      "step": 5173
    },
    {
      "epoch": 0.2792078139334089,
      "grad_norm": 0.9614238337293085,
      "learning_rate": 8.377028245135343e-06,
      "loss": 0.5366,
      "step": 5174
    },
    {
      "epoch": 0.2792617775619233,
      "grad_norm": 1.0032450035532046,
      "learning_rate": 8.376441596821475e-06,
      "loss": 0.5753,
      "step": 5175
    },
    {
      "epoch": 0.27931574119043767,
      "grad_norm": 0.9096085950586701,
      "learning_rate": 8.375854865835769e-06,
      "loss": 0.4688,
      "step": 5176
    },
    {
      "epoch": 0.279369704818952,
      "grad_norm": 0.9313417626631812,
      "learning_rate": 8.375268052195092e-06,
      "loss": 0.4133,
      "step": 5177
    },
    {
      "epoch": 0.2794236684474664,
      "grad_norm": 0.8858556868687976,
      "learning_rate": 8.374681155916302e-06,
      "loss": 0.4179,
      "step": 5178
    },
    {
      "epoch": 0.2794776320759808,
      "grad_norm": 1.241038069763822,
      "learning_rate": 8.374094177016271e-06,
      "loss": 0.71,
      "step": 5179
    },
    {
      "epoch": 0.27953159570449515,
      "grad_norm": 0.726303800177769,
      "learning_rate": 8.373507115511871e-06,
      "loss": 0.2853,
      "step": 5180
    },
    {
      "epoch": 0.27958555933300955,
      "grad_norm": 1.2834513512672878,
      "learning_rate": 8.372919971419974e-06,
      "loss": 0.6339,
      "step": 5181
    },
    {
      "epoch": 0.27963952296152395,
      "grad_norm": 1.2907785633539508,
      "learning_rate": 8.372332744757451e-06,
      "loss": 0.7897,
      "step": 5182
    },
    {
      "epoch": 0.2796934865900383,
      "grad_norm": 1.0138731266807657,
      "learning_rate": 8.371745435541187e-06,
      "loss": 0.4844,
      "step": 5183
    },
    {
      "epoch": 0.2797474502185527,
      "grad_norm": 0.9742985069461777,
      "learning_rate": 8.371158043788053e-06,
      "loss": 0.365,
      "step": 5184
    },
    {
      "epoch": 0.2798014138470671,
      "grad_norm": 1.0211065872616585,
      "learning_rate": 8.370570569514939e-06,
      "loss": 0.5078,
      "step": 5185
    },
    {
      "epoch": 0.2798553774755815,
      "grad_norm": 0.8583967322819118,
      "learning_rate": 8.369983012738725e-06,
      "loss": 0.3169,
      "step": 5186
    },
    {
      "epoch": 0.2799093411040958,
      "grad_norm": 0.9737478527592637,
      "learning_rate": 8.3693953734763e-06,
      "loss": 0.528,
      "step": 5187
    },
    {
      "epoch": 0.2799633047326102,
      "grad_norm": 1.1991064662352786,
      "learning_rate": 8.36880765174455e-06,
      "loss": 0.6277,
      "step": 5188
    },
    {
      "epoch": 0.2800172683611246,
      "grad_norm": 0.9292688528088241,
      "learning_rate": 8.368219847560372e-06,
      "loss": 0.4335,
      "step": 5189
    },
    {
      "epoch": 0.28007123198963896,
      "grad_norm": 1.0233726886781838,
      "learning_rate": 8.367631960940656e-06,
      "loss": 0.495,
      "step": 5190
    },
    {
      "epoch": 0.28012519561815336,
      "grad_norm": 1.1341116880415643,
      "learning_rate": 8.367043991902302e-06,
      "loss": 0.3519,
      "step": 5191
    },
    {
      "epoch": 0.28017915924666775,
      "grad_norm": 1.1029131077407095,
      "learning_rate": 8.366455940462205e-06,
      "loss": 0.4744,
      "step": 5192
    },
    {
      "epoch": 0.28023312287518215,
      "grad_norm": 1.3243149742718867,
      "learning_rate": 8.365867806637267e-06,
      "loss": 0.6258,
      "step": 5193
    },
    {
      "epoch": 0.2802870865036965,
      "grad_norm": 0.861426740239744,
      "learning_rate": 8.365279590444396e-06,
      "loss": 0.3822,
      "step": 5194
    },
    {
      "epoch": 0.2803410501322109,
      "grad_norm": 1.0789063539814836,
      "learning_rate": 8.364691291900491e-06,
      "loss": 0.5185,
      "step": 5195
    },
    {
      "epoch": 0.2803950137607253,
      "grad_norm": 1.1925559505285184,
      "learning_rate": 8.364102911022465e-06,
      "loss": 0.5673,
      "step": 5196
    },
    {
      "epoch": 0.28044897738923963,
      "grad_norm": 0.9768618989750608,
      "learning_rate": 8.36351444782723e-06,
      "loss": 0.5069,
      "step": 5197
    },
    {
      "epoch": 0.280502941017754,
      "grad_norm": 1.0007996356663025,
      "learning_rate": 8.362925902331692e-06,
      "loss": 0.5036,
      "step": 5198
    },
    {
      "epoch": 0.2805569046462684,
      "grad_norm": 0.9449320212842772,
      "learning_rate": 8.362337274552773e-06,
      "loss": 0.3695,
      "step": 5199
    },
    {
      "epoch": 0.2806108682747828,
      "grad_norm": 0.9884088893361652,
      "learning_rate": 8.361748564507387e-06,
      "loss": 0.5037,
      "step": 5200
    },
    {
      "epoch": 0.28066483190329716,
      "grad_norm": 1.2831977828391317,
      "learning_rate": 8.361159772212457e-06,
      "loss": 0.5981,
      "step": 5201
    },
    {
      "epoch": 0.28071879553181156,
      "grad_norm": 1.1509976665718125,
      "learning_rate": 8.360570897684905e-06,
      "loss": 0.6116,
      "step": 5202
    },
    {
      "epoch": 0.28077275916032596,
      "grad_norm": 0.7157367843923668,
      "learning_rate": 8.359981940941653e-06,
      "loss": 0.2681,
      "step": 5203
    },
    {
      "epoch": 0.2808267227888403,
      "grad_norm": 1.129770776732526,
      "learning_rate": 8.35939290199963e-06,
      "loss": 0.62,
      "step": 5204
    },
    {
      "epoch": 0.2808806864173547,
      "grad_norm": 1.0754692553439553,
      "learning_rate": 8.358803780875768e-06,
      "loss": 0.4591,
      "step": 5205
    },
    {
      "epoch": 0.2809346500458691,
      "grad_norm": 0.9795256363186552,
      "learning_rate": 8.358214577586995e-06,
      "loss": 0.4919,
      "step": 5206
    },
    {
      "epoch": 0.2809886136743835,
      "grad_norm": 0.9203708557747434,
      "learning_rate": 8.357625292150248e-06,
      "loss": 0.433,
      "step": 5207
    },
    {
      "epoch": 0.28104257730289783,
      "grad_norm": 1.2695843481038314,
      "learning_rate": 8.357035924582463e-06,
      "loss": 0.4863,
      "step": 5208
    },
    {
      "epoch": 0.28109654093141223,
      "grad_norm": 1.3246282664478144,
      "learning_rate": 8.356446474900578e-06,
      "loss": 0.4565,
      "step": 5209
    },
    {
      "epoch": 0.28115050455992663,
      "grad_norm": 1.143089835735994,
      "learning_rate": 8.355856943121536e-06,
      "loss": 0.4816,
      "step": 5210
    },
    {
      "epoch": 0.28120446818844097,
      "grad_norm": 1.0187325172334678,
      "learning_rate": 8.355267329262278e-06,
      "loss": 0.4471,
      "step": 5211
    },
    {
      "epoch": 0.28125843181695537,
      "grad_norm": 1.305157118218297,
      "learning_rate": 8.354677633339754e-06,
      "loss": 0.6088,
      "step": 5212
    },
    {
      "epoch": 0.28131239544546977,
      "grad_norm": 0.9146035159197472,
      "learning_rate": 8.354087855370909e-06,
      "loss": 0.3738,
      "step": 5213
    },
    {
      "epoch": 0.2813663590739841,
      "grad_norm": 0.8960768025370934,
      "learning_rate": 8.353497995372696e-06,
      "loss": 0.359,
      "step": 5214
    },
    {
      "epoch": 0.2814203227024985,
      "grad_norm": 1.0725420479656078,
      "learning_rate": 8.352908053362067e-06,
      "loss": 0.4419,
      "step": 5215
    },
    {
      "epoch": 0.2814742863310129,
      "grad_norm": 0.8507771988656984,
      "learning_rate": 8.352318029355977e-06,
      "loss": 0.4253,
      "step": 5216
    },
    {
      "epoch": 0.2815282499595273,
      "grad_norm": 0.9314833281648784,
      "learning_rate": 8.351727923371384e-06,
      "loss": 0.57,
      "step": 5217
    },
    {
      "epoch": 0.28158221358804164,
      "grad_norm": 1.1495705960788871,
      "learning_rate": 8.351137735425252e-06,
      "loss": 0.4345,
      "step": 5218
    },
    {
      "epoch": 0.28163617721655604,
      "grad_norm": 0.9791286271581207,
      "learning_rate": 8.350547465534538e-06,
      "loss": 0.5251,
      "step": 5219
    },
    {
      "epoch": 0.28169014084507044,
      "grad_norm": 1.0515167392180507,
      "learning_rate": 8.349957113716213e-06,
      "loss": 0.4363,
      "step": 5220
    },
    {
      "epoch": 0.2817441044735848,
      "grad_norm": 0.8349613592534751,
      "learning_rate": 8.349366679987237e-06,
      "loss": 0.3719,
      "step": 5221
    },
    {
      "epoch": 0.2817980681020992,
      "grad_norm": 0.8292161500472444,
      "learning_rate": 8.348776164364587e-06,
      "loss": 0.3255,
      "step": 5222
    },
    {
      "epoch": 0.2818520317306136,
      "grad_norm": 1.2565793912489867,
      "learning_rate": 8.34818556686523e-06,
      "loss": 0.6643,
      "step": 5223
    },
    {
      "epoch": 0.28190599535912797,
      "grad_norm": 0.9101156071186645,
      "learning_rate": 8.347594887506141e-06,
      "loss": 0.4416,
      "step": 5224
    },
    {
      "epoch": 0.2819599589876423,
      "grad_norm": 1.1524709155773445,
      "learning_rate": 8.347004126304297e-06,
      "loss": 0.5683,
      "step": 5225
    },
    {
      "epoch": 0.2820139226161567,
      "grad_norm": 0.8341321018357162,
      "learning_rate": 8.34641328327668e-06,
      "loss": 0.3497,
      "step": 5226
    },
    {
      "epoch": 0.2820678862446711,
      "grad_norm": 1.2668434836325178,
      "learning_rate": 8.345822358440268e-06,
      "loss": 0.6403,
      "step": 5227
    },
    {
      "epoch": 0.28212184987318545,
      "grad_norm": 1.050822214182496,
      "learning_rate": 8.345231351812045e-06,
      "loss": 0.4323,
      "step": 5228
    },
    {
      "epoch": 0.28217581350169985,
      "grad_norm": 1.1273374653670862,
      "learning_rate": 8.344640263408999e-06,
      "loss": 0.5613,
      "step": 5229
    },
    {
      "epoch": 0.28222977713021424,
      "grad_norm": 1.11068001568026,
      "learning_rate": 8.344049093248118e-06,
      "loss": 0.5278,
      "step": 5230
    },
    {
      "epoch": 0.28228374075872864,
      "grad_norm": 1.1057827306191248,
      "learning_rate": 8.343457841346391e-06,
      "loss": 0.6577,
      "step": 5231
    },
    {
      "epoch": 0.282337704387243,
      "grad_norm": 1.1927247126790186,
      "learning_rate": 8.342866507720817e-06,
      "loss": 0.503,
      "step": 5232
    },
    {
      "epoch": 0.2823916680157574,
      "grad_norm": 1.0893226374327662,
      "learning_rate": 8.342275092388382e-06,
      "loss": 0.5403,
      "step": 5233
    },
    {
      "epoch": 0.2824456316442718,
      "grad_norm": 1.2062931197595659,
      "learning_rate": 8.341683595366091e-06,
      "loss": 0.5266,
      "step": 5234
    },
    {
      "epoch": 0.2824995952727861,
      "grad_norm": 1.088510171170987,
      "learning_rate": 8.341092016670942e-06,
      "loss": 0.4691,
      "step": 5235
    },
    {
      "epoch": 0.2825535589013005,
      "grad_norm": 1.1393829029002434,
      "learning_rate": 8.340500356319938e-06,
      "loss": 0.4648,
      "step": 5236
    },
    {
      "epoch": 0.2826075225298149,
      "grad_norm": 1.2383087402811706,
      "learning_rate": 8.33990861433008e-06,
      "loss": 0.5759,
      "step": 5237
    },
    {
      "epoch": 0.2826614861583293,
      "grad_norm": 1.158388393443371,
      "learning_rate": 8.339316790718384e-06,
      "loss": 0.6915,
      "step": 5238
    },
    {
      "epoch": 0.28271544978684365,
      "grad_norm": 1.0733591859590972,
      "learning_rate": 8.338724885501852e-06,
      "loss": 0.5101,
      "step": 5239
    },
    {
      "epoch": 0.28276941341535805,
      "grad_norm": 1.0831872317622955,
      "learning_rate": 8.338132898697498e-06,
      "loss": 0.6607,
      "step": 5240
    },
    {
      "epoch": 0.28282337704387245,
      "grad_norm": 1.1098006949879207,
      "learning_rate": 8.337540830322338e-06,
      "loss": 0.4747,
      "step": 5241
    },
    {
      "epoch": 0.2828773406723868,
      "grad_norm": 0.8594101318684795,
      "learning_rate": 8.336948680393389e-06,
      "loss": 0.4554,
      "step": 5242
    },
    {
      "epoch": 0.2829313043009012,
      "grad_norm": 0.8475147965094816,
      "learning_rate": 8.336356448927665e-06,
      "loss": 0.4089,
      "step": 5243
    },
    {
      "epoch": 0.2829852679294156,
      "grad_norm": 1.1360751967876888,
      "learning_rate": 8.335764135942192e-06,
      "loss": 0.5592,
      "step": 5244
    },
    {
      "epoch": 0.28303923155793,
      "grad_norm": 1.0307022365279201,
      "learning_rate": 8.335171741453994e-06,
      "loss": 0.5353,
      "step": 5245
    },
    {
      "epoch": 0.2830931951864443,
      "grad_norm": 1.1730224545012737,
      "learning_rate": 8.334579265480094e-06,
      "loss": 0.5233,
      "step": 5246
    },
    {
      "epoch": 0.2831471588149587,
      "grad_norm": 0.7916927050990835,
      "learning_rate": 8.33398670803752e-06,
      "loss": 0.3586,
      "step": 5247
    },
    {
      "epoch": 0.2832011224434731,
      "grad_norm": 0.9436581499927891,
      "learning_rate": 8.333394069143309e-06,
      "loss": 0.3952,
      "step": 5248
    },
    {
      "epoch": 0.28325508607198746,
      "grad_norm": 1.070323017967436,
      "learning_rate": 8.332801348814485e-06,
      "loss": 0.4616,
      "step": 5249
    },
    {
      "epoch": 0.28330904970050186,
      "grad_norm": 0.9573372477518666,
      "learning_rate": 8.33220854706809e-06,
      "loss": 0.5162,
      "step": 5250
    },
    {
      "epoch": 0.28336301332901626,
      "grad_norm": 1.1585513949342028,
      "learning_rate": 8.331615663921159e-06,
      "loss": 0.6672,
      "step": 5251
    },
    {
      "epoch": 0.2834169769575306,
      "grad_norm": 0.8834631878106044,
      "learning_rate": 8.331022699390734e-06,
      "loss": 0.4705,
      "step": 5252
    },
    {
      "epoch": 0.283470940586045,
      "grad_norm": 1.2864485291397298,
      "learning_rate": 8.330429653493856e-06,
      "loss": 0.6313,
      "step": 5253
    },
    {
      "epoch": 0.2835249042145594,
      "grad_norm": 0.8733286481763205,
      "learning_rate": 8.329836526247568e-06,
      "loss": 0.4253,
      "step": 5254
    },
    {
      "epoch": 0.2835788678430738,
      "grad_norm": 1.0304861908664968,
      "learning_rate": 8.32924331766892e-06,
      "loss": 0.5078,
      "step": 5255
    },
    {
      "epoch": 0.28363283147158813,
      "grad_norm": 0.9393371350873889,
      "learning_rate": 8.328650027774961e-06,
      "loss": 0.4085,
      "step": 5256
    },
    {
      "epoch": 0.28368679510010253,
      "grad_norm": 1.2698276204761005,
      "learning_rate": 8.328056656582741e-06,
      "loss": 0.5969,
      "step": 5257
    },
    {
      "epoch": 0.2837407587286169,
      "grad_norm": 0.9069514263330534,
      "learning_rate": 8.327463204109316e-06,
      "loss": 0.4597,
      "step": 5258
    },
    {
      "epoch": 0.28379472235713127,
      "grad_norm": 0.9513637608508044,
      "learning_rate": 8.32686967037174e-06,
      "loss": 0.5246,
      "step": 5259
    },
    {
      "epoch": 0.28384868598564567,
      "grad_norm": 0.9858167083018621,
      "learning_rate": 8.326276055387074e-06,
      "loss": 0.5333,
      "step": 5260
    },
    {
      "epoch": 0.28390264961416006,
      "grad_norm": 1.252218898042466,
      "learning_rate": 8.32568235917238e-06,
      "loss": 0.5273,
      "step": 5261
    },
    {
      "epoch": 0.28395661324267446,
      "grad_norm": 1.1096524984315819,
      "learning_rate": 8.325088581744718e-06,
      "loss": 0.4854,
      "step": 5262
    },
    {
      "epoch": 0.2840105768711888,
      "grad_norm": 1.1571064917502285,
      "learning_rate": 8.324494723121156e-06,
      "loss": 0.4816,
      "step": 5263
    },
    {
      "epoch": 0.2840645404997032,
      "grad_norm": 1.0214729625428483,
      "learning_rate": 8.323900783318763e-06,
      "loss": 0.5225,
      "step": 5264
    },
    {
      "epoch": 0.2841185041282176,
      "grad_norm": 0.9602134731411627,
      "learning_rate": 8.323306762354606e-06,
      "loss": 0.4346,
      "step": 5265
    },
    {
      "epoch": 0.28417246775673194,
      "grad_norm": 1.003491878466424,
      "learning_rate": 8.32271266024576e-06,
      "loss": 0.4838,
      "step": 5266
    },
    {
      "epoch": 0.28422643138524634,
      "grad_norm": 0.9961159876873469,
      "learning_rate": 8.322118477009302e-06,
      "loss": 0.5509,
      "step": 5267
    },
    {
      "epoch": 0.28428039501376073,
      "grad_norm": 1.10443576954379,
      "learning_rate": 8.321524212662306e-06,
      "loss": 0.5677,
      "step": 5268
    },
    {
      "epoch": 0.28433435864227513,
      "grad_norm": 1.059302300848515,
      "learning_rate": 8.320929867221853e-06,
      "loss": 0.4434,
      "step": 5269
    },
    {
      "epoch": 0.2843883222707895,
      "grad_norm": 1.116302503945931,
      "learning_rate": 8.320335440705027e-06,
      "loss": 0.6942,
      "step": 5270
    },
    {
      "epoch": 0.28444228589930387,
      "grad_norm": 1.1731409931129784,
      "learning_rate": 8.319740933128909e-06,
      "loss": 0.655,
      "step": 5271
    },
    {
      "epoch": 0.28449624952781827,
      "grad_norm": 0.9802973559467687,
      "learning_rate": 8.319146344510587e-06,
      "loss": 0.4073,
      "step": 5272
    },
    {
      "epoch": 0.2845502131563326,
      "grad_norm": 0.9818750488104417,
      "learning_rate": 8.318551674867154e-06,
      "loss": 0.4637,
      "step": 5273
    },
    {
      "epoch": 0.284604176784847,
      "grad_norm": 0.9684884559716106,
      "learning_rate": 8.317956924215694e-06,
      "loss": 0.5124,
      "step": 5274
    },
    {
      "epoch": 0.2846581404133614,
      "grad_norm": 1.153653840275477,
      "learning_rate": 8.317362092573305e-06,
      "loss": 0.4813,
      "step": 5275
    },
    {
      "epoch": 0.2847121040418758,
      "grad_norm": 1.227678657381375,
      "learning_rate": 8.316767179957086e-06,
      "loss": 0.5169,
      "step": 5276
    },
    {
      "epoch": 0.28476606767039014,
      "grad_norm": 1.1822483721550656,
      "learning_rate": 8.316172186384128e-06,
      "loss": 0.5857,
      "step": 5277
    },
    {
      "epoch": 0.28482003129890454,
      "grad_norm": 1.107453794051031,
      "learning_rate": 8.315577111871539e-06,
      "loss": 0.4236,
      "step": 5278
    },
    {
      "epoch": 0.28487399492741894,
      "grad_norm": 1.055904068297378,
      "learning_rate": 8.314981956436418e-06,
      "loss": 0.4489,
      "step": 5279
    },
    {
      "epoch": 0.2849279585559333,
      "grad_norm": 1.0862839078490334,
      "learning_rate": 8.314386720095871e-06,
      "loss": 0.5057,
      "step": 5280
    },
    {
      "epoch": 0.2849819221844477,
      "grad_norm": 1.1746623265796625,
      "learning_rate": 8.313791402867005e-06,
      "loss": 0.4907,
      "step": 5281
    },
    {
      "epoch": 0.2850358858129621,
      "grad_norm": 0.8413264516204407,
      "learning_rate": 8.313196004766931e-06,
      "loss": 0.3666,
      "step": 5282
    },
    {
      "epoch": 0.2850898494414764,
      "grad_norm": 1.2320785224116908,
      "learning_rate": 8.312600525812764e-06,
      "loss": 0.6733,
      "step": 5283
    },
    {
      "epoch": 0.2851438130699908,
      "grad_norm": 1.0121220179431036,
      "learning_rate": 8.312004966021614e-06,
      "loss": 0.5218,
      "step": 5284
    },
    {
      "epoch": 0.2851977766985052,
      "grad_norm": 0.8601793472793967,
      "learning_rate": 8.311409325410598e-06,
      "loss": 0.4177,
      "step": 5285
    },
    {
      "epoch": 0.2852517403270196,
      "grad_norm": 1.17044729092144,
      "learning_rate": 8.31081360399684e-06,
      "loss": 0.6768,
      "step": 5286
    },
    {
      "epoch": 0.28530570395553395,
      "grad_norm": 0.9818135941724351,
      "learning_rate": 8.31021780179746e-06,
      "loss": 0.4669,
      "step": 5287
    },
    {
      "epoch": 0.28535966758404835,
      "grad_norm": 1.109802723527431,
      "learning_rate": 8.309621918829577e-06,
      "loss": 0.5982,
      "step": 5288
    },
    {
      "epoch": 0.28541363121256275,
      "grad_norm": 1.1965973490015482,
      "learning_rate": 8.309025955110322e-06,
      "loss": 0.5572,
      "step": 5289
    },
    {
      "epoch": 0.2854675948410771,
      "grad_norm": 0.8545864074578304,
      "learning_rate": 8.308429910656825e-06,
      "loss": 0.3617,
      "step": 5290
    },
    {
      "epoch": 0.2855215584695915,
      "grad_norm": 1.0131251750944372,
      "learning_rate": 8.307833785486214e-06,
      "loss": 0.4626,
      "step": 5291
    },
    {
      "epoch": 0.2855755220981059,
      "grad_norm": 0.8344219881157394,
      "learning_rate": 8.307237579615622e-06,
      "loss": 0.4027,
      "step": 5292
    },
    {
      "epoch": 0.2856294857266203,
      "grad_norm": 1.0832303553142841,
      "learning_rate": 8.306641293062185e-06,
      "loss": 0.5916,
      "step": 5293
    },
    {
      "epoch": 0.2856834493551346,
      "grad_norm": 1.102988197371885,
      "learning_rate": 8.306044925843044e-06,
      "loss": 0.7076,
      "step": 5294
    },
    {
      "epoch": 0.285737412983649,
      "grad_norm": 1.2008423341904468,
      "learning_rate": 8.305448477975334e-06,
      "loss": 0.5923,
      "step": 5295
    },
    {
      "epoch": 0.2857913766121634,
      "grad_norm": 0.8749593249745591,
      "learning_rate": 8.304851949476202e-06,
      "loss": 0.4231,
      "step": 5296
    },
    {
      "epoch": 0.28584534024067776,
      "grad_norm": 1.0504811931287372,
      "learning_rate": 8.30425534036279e-06,
      "loss": 0.4763,
      "step": 5297
    },
    {
      "epoch": 0.28589930386919216,
      "grad_norm": 1.084806656253144,
      "learning_rate": 8.303658650652245e-06,
      "loss": 0.5237,
      "step": 5298
    },
    {
      "epoch": 0.28595326749770655,
      "grad_norm": 1.0845437186622442,
      "learning_rate": 8.303061880361717e-06,
      "loss": 0.5559,
      "step": 5299
    },
    {
      "epoch": 0.28600723112622095,
      "grad_norm": 1.0155568524574485,
      "learning_rate": 8.30246502950836e-06,
      "loss": 0.5147,
      "step": 5300
    },
    {
      "epoch": 0.2860611947547353,
      "grad_norm": 1.012945340281663,
      "learning_rate": 8.301868098109326e-06,
      "loss": 0.5269,
      "step": 5301
    },
    {
      "epoch": 0.2861151583832497,
      "grad_norm": 1.0600868776244337,
      "learning_rate": 8.30127108618177e-06,
      "loss": 0.6879,
      "step": 5302
    },
    {
      "epoch": 0.2861691220117641,
      "grad_norm": 0.8685829709623949,
      "learning_rate": 8.300673993742856e-06,
      "loss": 0.3714,
      "step": 5303
    },
    {
      "epoch": 0.28622308564027843,
      "grad_norm": 0.9758221399855582,
      "learning_rate": 8.300076820809739e-06,
      "loss": 0.4791,
      "step": 5304
    },
    {
      "epoch": 0.2862770492687928,
      "grad_norm": 0.9391237851130257,
      "learning_rate": 8.299479567399585e-06,
      "loss": 0.4553,
      "step": 5305
    },
    {
      "epoch": 0.2863310128973072,
      "grad_norm": 1.058592126971488,
      "learning_rate": 8.29888223352956e-06,
      "loss": 0.5271,
      "step": 5306
    },
    {
      "epoch": 0.2863849765258216,
      "grad_norm": 1.0226100985913054,
      "learning_rate": 8.29828481921683e-06,
      "loss": 0.5826,
      "step": 5307
    },
    {
      "epoch": 0.28643894015433596,
      "grad_norm": 0.9555447593355231,
      "learning_rate": 8.297687324478572e-06,
      "loss": 0.4977,
      "step": 5308
    },
    {
      "epoch": 0.28649290378285036,
      "grad_norm": 1.0506232493996632,
      "learning_rate": 8.297089749331948e-06,
      "loss": 0.5216,
      "step": 5309
    },
    {
      "epoch": 0.28654686741136476,
      "grad_norm": 1.0382295158113921,
      "learning_rate": 8.29649209379414e-06,
      "loss": 0.3467,
      "step": 5310
    },
    {
      "epoch": 0.2866008310398791,
      "grad_norm": 1.0318329523797682,
      "learning_rate": 8.295894357882325e-06,
      "loss": 0.5723,
      "step": 5311
    },
    {
      "epoch": 0.2866547946683935,
      "grad_norm": 0.8639993145164058,
      "learning_rate": 8.29529654161368e-06,
      "loss": 0.4378,
      "step": 5312
    },
    {
      "epoch": 0.2867087582969079,
      "grad_norm": 0.9627058230658225,
      "learning_rate": 8.294698645005388e-06,
      "loss": 0.4302,
      "step": 5313
    },
    {
      "epoch": 0.2867627219254223,
      "grad_norm": 1.0104795783410847,
      "learning_rate": 8.294100668074632e-06,
      "loss": 0.4711,
      "step": 5314
    },
    {
      "epoch": 0.28681668555393663,
      "grad_norm": 0.9163214770613735,
      "learning_rate": 8.293502610838603e-06,
      "loss": 0.3796,
      "step": 5315
    },
    {
      "epoch": 0.28687064918245103,
      "grad_norm": 0.7381929135575839,
      "learning_rate": 8.292904473314482e-06,
      "loss": 0.3027,
      "step": 5316
    },
    {
      "epoch": 0.28692461281096543,
      "grad_norm": 1.009542838191959,
      "learning_rate": 8.292306255519467e-06,
      "loss": 0.4176,
      "step": 5317
    },
    {
      "epoch": 0.28697857643947977,
      "grad_norm": 1.1070272269681232,
      "learning_rate": 8.291707957470748e-06,
      "loss": 0.6255,
      "step": 5318
    },
    {
      "epoch": 0.28703254006799417,
      "grad_norm": 1.0066582367907981,
      "learning_rate": 8.29110957918552e-06,
      "loss": 0.5236,
      "step": 5319
    },
    {
      "epoch": 0.28708650369650857,
      "grad_norm": 1.2580977754001146,
      "learning_rate": 8.290511120680983e-06,
      "loss": 0.4868,
      "step": 5320
    },
    {
      "epoch": 0.2871404673250229,
      "grad_norm": 0.9549151537489389,
      "learning_rate": 8.28991258197434e-06,
      "loss": 0.4468,
      "step": 5321
    },
    {
      "epoch": 0.2871944309535373,
      "grad_norm": 1.0200049538453886,
      "learning_rate": 8.28931396308279e-06,
      "loss": 0.4361,
      "step": 5322
    },
    {
      "epoch": 0.2872483945820517,
      "grad_norm": 1.16405211470237,
      "learning_rate": 8.288715264023534e-06,
      "loss": 0.4582,
      "step": 5323
    },
    {
      "epoch": 0.2873023582105661,
      "grad_norm": 1.005309631343437,
      "learning_rate": 8.288116484813786e-06,
      "loss": 0.4996,
      "step": 5324
    },
    {
      "epoch": 0.28735632183908044,
      "grad_norm": 0.8889590951230237,
      "learning_rate": 8.287517625470754e-06,
      "loss": 0.4026,
      "step": 5325
    },
    {
      "epoch": 0.28741028546759484,
      "grad_norm": 0.8720753981019168,
      "learning_rate": 8.28691868601165e-06,
      "loss": 0.3754,
      "step": 5326
    },
    {
      "epoch": 0.28746424909610924,
      "grad_norm": 1.2623302572009456,
      "learning_rate": 8.286319666453682e-06,
      "loss": 0.7516,
      "step": 5327
    },
    {
      "epoch": 0.2875182127246236,
      "grad_norm": 1.0459545899110336,
      "learning_rate": 8.285720566814076e-06,
      "loss": 0.4001,
      "step": 5328
    },
    {
      "epoch": 0.287572176353138,
      "grad_norm": 1.0079595754939419,
      "learning_rate": 8.285121387110044e-06,
      "loss": 0.4526,
      "step": 5329
    },
    {
      "epoch": 0.2876261399816524,
      "grad_norm": 1.1201170004147303,
      "learning_rate": 8.284522127358809e-06,
      "loss": 0.5641,
      "step": 5330
    },
    {
      "epoch": 0.28768010361016677,
      "grad_norm": 0.9532457026941401,
      "learning_rate": 8.283922787577596e-06,
      "loss": 0.4008,
      "step": 5331
    },
    {
      "epoch": 0.2877340672386811,
      "grad_norm": 1.1028198846853186,
      "learning_rate": 8.283323367783627e-06,
      "loss": 0.5664,
      "step": 5332
    },
    {
      "epoch": 0.2877880308671955,
      "grad_norm": 1.144328468829172,
      "learning_rate": 8.282723867994135e-06,
      "loss": 0.5574,
      "step": 5333
    },
    {
      "epoch": 0.2878419944957099,
      "grad_norm": 1.0425593906311137,
      "learning_rate": 8.282124288226346e-06,
      "loss": 0.3982,
      "step": 5334
    },
    {
      "epoch": 0.28789595812422425,
      "grad_norm": 0.9455468236766429,
      "learning_rate": 8.281524628497492e-06,
      "loss": 0.5002,
      "step": 5335
    },
    {
      "epoch": 0.28794992175273865,
      "grad_norm": 1.2031298857041257,
      "learning_rate": 8.280924888824812e-06,
      "loss": 0.6162,
      "step": 5336
    },
    {
      "epoch": 0.28800388538125304,
      "grad_norm": 1.1849757761009654,
      "learning_rate": 8.28032506922554e-06,
      "loss": 0.5512,
      "step": 5337
    },
    {
      "epoch": 0.28805784900976744,
      "grad_norm": 1.0288548564280657,
      "learning_rate": 8.279725169716917e-06,
      "loss": 0.3496,
      "step": 5338
    },
    {
      "epoch": 0.2881118126382818,
      "grad_norm": 0.9502486230723481,
      "learning_rate": 8.279125190316182e-06,
      "loss": 0.4609,
      "step": 5339
    },
    {
      "epoch": 0.2881657762667962,
      "grad_norm": 1.212778276609102,
      "learning_rate": 8.278525131040582e-06,
      "loss": 0.6149,
      "step": 5340
    },
    {
      "epoch": 0.2882197398953106,
      "grad_norm": 1.1084392221827315,
      "learning_rate": 8.277924991907363e-06,
      "loss": 0.3774,
      "step": 5341
    },
    {
      "epoch": 0.2882737035238249,
      "grad_norm": 1.0635753482369519,
      "learning_rate": 8.277324772933772e-06,
      "loss": 0.4498,
      "step": 5342
    },
    {
      "epoch": 0.2883276671523393,
      "grad_norm": 0.8876907625015934,
      "learning_rate": 8.27672447413706e-06,
      "loss": 0.3878,
      "step": 5343
    },
    {
      "epoch": 0.2883816307808537,
      "grad_norm": 0.9340236997187584,
      "learning_rate": 8.27612409553448e-06,
      "loss": 0.4761,
      "step": 5344
    },
    {
      "epoch": 0.2884355944093681,
      "grad_norm": 1.2588573342806322,
      "learning_rate": 8.275523637143292e-06,
      "loss": 0.6165,
      "step": 5345
    },
    {
      "epoch": 0.28848955803788245,
      "grad_norm": 0.8950565095279595,
      "learning_rate": 8.274923098980749e-06,
      "loss": 0.4173,
      "step": 5346
    },
    {
      "epoch": 0.28854352166639685,
      "grad_norm": 1.2283973344435488,
      "learning_rate": 8.274322481064112e-06,
      "loss": 0.4533,
      "step": 5347
    },
    {
      "epoch": 0.28859748529491125,
      "grad_norm": 1.1358369546547467,
      "learning_rate": 8.273721783410643e-06,
      "loss": 0.5539,
      "step": 5348
    },
    {
      "epoch": 0.2886514489234256,
      "grad_norm": 1.0353221302404354,
      "learning_rate": 8.273121006037608e-06,
      "loss": 0.5357,
      "step": 5349
    },
    {
      "epoch": 0.28870541255194,
      "grad_norm": 1.0787186213548108,
      "learning_rate": 8.272520148962272e-06,
      "loss": 0.5179,
      "step": 5350
    },
    {
      "epoch": 0.2887593761804544,
      "grad_norm": 0.9430812535413787,
      "learning_rate": 8.271919212201907e-06,
      "loss": 0.4222,
      "step": 5351
    },
    {
      "epoch": 0.2888133398089687,
      "grad_norm": 0.9956531377766678,
      "learning_rate": 8.271318195773782e-06,
      "loss": 0.4742,
      "step": 5352
    },
    {
      "epoch": 0.2888673034374831,
      "grad_norm": 1.2357996457291027,
      "learning_rate": 8.270717099695173e-06,
      "loss": 0.4661,
      "step": 5353
    },
    {
      "epoch": 0.2889212670659975,
      "grad_norm": 0.986131967300041,
      "learning_rate": 8.270115923983355e-06,
      "loss": 0.3932,
      "step": 5354
    },
    {
      "epoch": 0.2889752306945119,
      "grad_norm": 1.2507282945597666,
      "learning_rate": 8.269514668655606e-06,
      "loss": 0.534,
      "step": 5355
    },
    {
      "epoch": 0.28902919432302626,
      "grad_norm": 0.9649999800689361,
      "learning_rate": 8.268913333729206e-06,
      "loss": 0.6083,
      "step": 5356
    },
    {
      "epoch": 0.28908315795154066,
      "grad_norm": 1.0600703490750132,
      "learning_rate": 8.268311919221439e-06,
      "loss": 0.4797,
      "step": 5357
    },
    {
      "epoch": 0.28913712158005506,
      "grad_norm": 1.002010915110171,
      "learning_rate": 8.267710425149592e-06,
      "loss": 0.438,
      "step": 5358
    },
    {
      "epoch": 0.2891910852085694,
      "grad_norm": 1.0761274338600197,
      "learning_rate": 8.267108851530949e-06,
      "loss": 0.6289,
      "step": 5359
    },
    {
      "epoch": 0.2892450488370838,
      "grad_norm": 1.0051417189596616,
      "learning_rate": 8.266507198382801e-06,
      "loss": 0.4645,
      "step": 5360
    },
    {
      "epoch": 0.2892990124655982,
      "grad_norm": 1.255253736523476,
      "learning_rate": 8.265905465722442e-06,
      "loss": 0.4766,
      "step": 5361
    },
    {
      "epoch": 0.2893529760941126,
      "grad_norm": 1.2243599995429308,
      "learning_rate": 8.265303653567166e-06,
      "loss": 0.6498,
      "step": 5362
    },
    {
      "epoch": 0.28940693972262693,
      "grad_norm": 0.9891044353067284,
      "learning_rate": 8.264701761934268e-06,
      "loss": 0.4538,
      "step": 5363
    },
    {
      "epoch": 0.28946090335114133,
      "grad_norm": 0.9263801064862729,
      "learning_rate": 8.264099790841047e-06,
      "loss": 0.4086,
      "step": 5364
    },
    {
      "epoch": 0.2895148669796557,
      "grad_norm": 1.3073434674959985,
      "learning_rate": 8.263497740304807e-06,
      "loss": 0.6906,
      "step": 5365
    },
    {
      "epoch": 0.28956883060817007,
      "grad_norm": 1.0404513363794408,
      "learning_rate": 8.26289561034285e-06,
      "loss": 0.5747,
      "step": 5366
    },
    {
      "epoch": 0.28962279423668447,
      "grad_norm": 1.427748007810944,
      "learning_rate": 8.262293400972478e-06,
      "loss": 0.5797,
      "step": 5367
    },
    {
      "epoch": 0.28967675786519886,
      "grad_norm": 0.9544937010317941,
      "learning_rate": 8.261691112211006e-06,
      "loss": 0.401,
      "step": 5368
    },
    {
      "epoch": 0.28973072149371326,
      "grad_norm": 1.0175496589820658,
      "learning_rate": 8.261088744075739e-06,
      "loss": 0.4157,
      "step": 5369
    },
    {
      "epoch": 0.2897846851222276,
      "grad_norm": 1.1014040030809185,
      "learning_rate": 8.260486296583992e-06,
      "loss": 0.4536,
      "step": 5370
    },
    {
      "epoch": 0.289838648750742,
      "grad_norm": 1.116412236204789,
      "learning_rate": 8.25988376975308e-06,
      "loss": 0.5688,
      "step": 5371
    },
    {
      "epoch": 0.2898926123792564,
      "grad_norm": 1.1399711376563233,
      "learning_rate": 8.25928116360032e-06,
      "loss": 0.5436,
      "step": 5372
    },
    {
      "epoch": 0.28994657600777074,
      "grad_norm": 0.847831458932473,
      "learning_rate": 8.258678478143032e-06,
      "loss": 0.4155,
      "step": 5373
    },
    {
      "epoch": 0.29000053963628514,
      "grad_norm": 1.2379818308535322,
      "learning_rate": 8.258075713398536e-06,
      "loss": 0.4429,
      "step": 5374
    },
    {
      "epoch": 0.29005450326479953,
      "grad_norm": 1.2381041974942528,
      "learning_rate": 8.257472869384156e-06,
      "loss": 0.5696,
      "step": 5375
    },
    {
      "epoch": 0.29010846689331393,
      "grad_norm": 1.1872922720443846,
      "learning_rate": 8.25686994611722e-06,
      "loss": 0.6659,
      "step": 5376
    },
    {
      "epoch": 0.2901624305218283,
      "grad_norm": 1.040378457305383,
      "learning_rate": 8.256266943615059e-06,
      "loss": 0.5299,
      "step": 5377
    },
    {
      "epoch": 0.29021639415034267,
      "grad_norm": 0.9826808545351431,
      "learning_rate": 8.255663861894998e-06,
      "loss": 0.4029,
      "step": 5378
    },
    {
      "epoch": 0.29027035777885707,
      "grad_norm": 0.9791381762969836,
      "learning_rate": 8.255060700974374e-06,
      "loss": 0.4387,
      "step": 5379
    },
    {
      "epoch": 0.2903243214073714,
      "grad_norm": 1.0324158777726515,
      "learning_rate": 8.254457460870523e-06,
      "loss": 0.5707,
      "step": 5380
    },
    {
      "epoch": 0.2903782850358858,
      "grad_norm": 1.0881804681399938,
      "learning_rate": 8.25385414160078e-06,
      "loss": 0.5091,
      "step": 5381
    },
    {
      "epoch": 0.2904322486644002,
      "grad_norm": 0.9469636009164442,
      "learning_rate": 8.253250743182486e-06,
      "loss": 0.4769,
      "step": 5382
    },
    {
      "epoch": 0.2904862122929146,
      "grad_norm": 0.7687497521535107,
      "learning_rate": 8.252647265632983e-06,
      "loss": 0.2819,
      "step": 5383
    },
    {
      "epoch": 0.29054017592142894,
      "grad_norm": 0.9433582375627291,
      "learning_rate": 8.252043708969619e-06,
      "loss": 0.4617,
      "step": 5384
    },
    {
      "epoch": 0.29059413954994334,
      "grad_norm": 0.827458323168548,
      "learning_rate": 8.251440073209735e-06,
      "loss": 0.3072,
      "step": 5385
    },
    {
      "epoch": 0.29064810317845774,
      "grad_norm": 0.9467999625777371,
      "learning_rate": 8.250836358370686e-06,
      "loss": 0.4146,
      "step": 5386
    },
    {
      "epoch": 0.2907020668069721,
      "grad_norm": 0.9899421326637341,
      "learning_rate": 8.250232564469819e-06,
      "loss": 0.536,
      "step": 5387
    },
    {
      "epoch": 0.2907560304354865,
      "grad_norm": 1.1909411469299798,
      "learning_rate": 8.24962869152449e-06,
      "loss": 0.5406,
      "step": 5388
    },
    {
      "epoch": 0.2908099940640009,
      "grad_norm": 1.146237262181166,
      "learning_rate": 8.249024739552053e-06,
      "loss": 0.5784,
      "step": 5389
    },
    {
      "epoch": 0.2908639576925152,
      "grad_norm": 1.1087033144287817,
      "learning_rate": 8.248420708569867e-06,
      "loss": 0.5568,
      "step": 5390
    },
    {
      "epoch": 0.2909179213210296,
      "grad_norm": 0.967375010073956,
      "learning_rate": 8.247816598595295e-06,
      "loss": 0.4035,
      "step": 5391
    },
    {
      "epoch": 0.290971884949544,
      "grad_norm": 1.041482864774634,
      "learning_rate": 8.247212409645695e-06,
      "loss": 0.4772,
      "step": 5392
    },
    {
      "epoch": 0.2910258485780584,
      "grad_norm": 1.073923591731376,
      "learning_rate": 8.246608141738436e-06,
      "loss": 0.5403,
      "step": 5393
    },
    {
      "epoch": 0.29107981220657275,
      "grad_norm": 1.0275768764893314,
      "learning_rate": 8.246003794890885e-06,
      "loss": 0.543,
      "step": 5394
    },
    {
      "epoch": 0.29113377583508715,
      "grad_norm": 1.2735076082564012,
      "learning_rate": 8.245399369120407e-06,
      "loss": 0.4239,
      "step": 5395
    },
    {
      "epoch": 0.29118773946360155,
      "grad_norm": 0.7554750894448626,
      "learning_rate": 8.244794864444378e-06,
      "loss": 0.3589,
      "step": 5396
    },
    {
      "epoch": 0.2912417030921159,
      "grad_norm": 1.045449726593084,
      "learning_rate": 8.244190280880175e-06,
      "loss": 0.4992,
      "step": 5397
    },
    {
      "epoch": 0.2912956667206303,
      "grad_norm": 1.1965004638833823,
      "learning_rate": 8.243585618445166e-06,
      "loss": 0.4952,
      "step": 5398
    },
    {
      "epoch": 0.2913496303491447,
      "grad_norm": 0.9199266289175159,
      "learning_rate": 8.242980877156735e-06,
      "loss": 0.4529,
      "step": 5399
    },
    {
      "epoch": 0.2914035939776591,
      "grad_norm": 0.9442601209321192,
      "learning_rate": 8.242376057032263e-06,
      "loss": 0.3833,
      "step": 5400
    },
    {
      "epoch": 0.2914575576061734,
      "grad_norm": 0.9376888997010562,
      "learning_rate": 8.241771158089132e-06,
      "loss": 0.4161,
      "step": 5401
    },
    {
      "epoch": 0.2915115212346878,
      "grad_norm": 1.0500417413452352,
      "learning_rate": 8.241166180344728e-06,
      "loss": 0.4168,
      "step": 5402
    },
    {
      "epoch": 0.2915654848632022,
      "grad_norm": 1.1573073174924724,
      "learning_rate": 8.240561123816437e-06,
      "loss": 0.62,
      "step": 5403
    },
    {
      "epoch": 0.29161944849171656,
      "grad_norm": 1.0446495661094266,
      "learning_rate": 8.23995598852165e-06,
      "loss": 0.467,
      "step": 5404
    },
    {
      "epoch": 0.29167341212023096,
      "grad_norm": 0.9583283215429682,
      "learning_rate": 8.239350774477761e-06,
      "loss": 0.3966,
      "step": 5405
    },
    {
      "epoch": 0.29172737574874535,
      "grad_norm": 0.8797972135789057,
      "learning_rate": 8.238745481702161e-06,
      "loss": 0.3979,
      "step": 5406
    },
    {
      "epoch": 0.29178133937725975,
      "grad_norm": 1.1298912186093513,
      "learning_rate": 8.238140110212249e-06,
      "loss": 0.4898,
      "step": 5407
    },
    {
      "epoch": 0.2918353030057741,
      "grad_norm": 1.0661438980055717,
      "learning_rate": 8.237534660025425e-06,
      "loss": 0.6284,
      "step": 5408
    },
    {
      "epoch": 0.2918892666342885,
      "grad_norm": 1.215853836598387,
      "learning_rate": 8.236929131159088e-06,
      "loss": 0.6196,
      "step": 5409
    },
    {
      "epoch": 0.2919432302628029,
      "grad_norm": 1.1409039838798734,
      "learning_rate": 8.236323523630642e-06,
      "loss": 0.3982,
      "step": 5410
    },
    {
      "epoch": 0.29199719389131723,
      "grad_norm": 0.8400040183169017,
      "learning_rate": 8.235717837457493e-06,
      "loss": 0.3328,
      "step": 5411
    },
    {
      "epoch": 0.29205115751983163,
      "grad_norm": 1.0488940463657699,
      "learning_rate": 8.23511207265705e-06,
      "loss": 0.5608,
      "step": 5412
    },
    {
      "epoch": 0.292105121148346,
      "grad_norm": 1.1366563299138206,
      "learning_rate": 8.234506229246721e-06,
      "loss": 0.4682,
      "step": 5413
    },
    {
      "epoch": 0.2921590847768604,
      "grad_norm": 1.1224182857202638,
      "learning_rate": 8.233900307243921e-06,
      "loss": 0.4539,
      "step": 5414
    },
    {
      "epoch": 0.29221304840537476,
      "grad_norm": 0.9959704626024016,
      "learning_rate": 8.233294306666063e-06,
      "loss": 0.5272,
      "step": 5415
    },
    {
      "epoch": 0.29226701203388916,
      "grad_norm": 1.0323317438566149,
      "learning_rate": 8.232688227530567e-06,
      "loss": 0.5355,
      "step": 5416
    },
    {
      "epoch": 0.29232097566240356,
      "grad_norm": 1.1367329705569469,
      "learning_rate": 8.232082069854847e-06,
      "loss": 0.4943,
      "step": 5417
    },
    {
      "epoch": 0.2923749392909179,
      "grad_norm": 1.246354004340486,
      "learning_rate": 8.23147583365633e-06,
      "loss": 0.4472,
      "step": 5418
    },
    {
      "epoch": 0.2924289029194323,
      "grad_norm": 1.1123869751782323,
      "learning_rate": 8.230869518952436e-06,
      "loss": 0.4595,
      "step": 5419
    },
    {
      "epoch": 0.2924828665479467,
      "grad_norm": 0.9773400274911657,
      "learning_rate": 8.230263125760593e-06,
      "loss": 0.4384,
      "step": 5420
    },
    {
      "epoch": 0.29253683017646104,
      "grad_norm": 0.9407495478043133,
      "learning_rate": 8.22965665409823e-06,
      "loss": 0.5721,
      "step": 5421
    },
    {
      "epoch": 0.29259079380497544,
      "grad_norm": 0.8399569173836842,
      "learning_rate": 8.229050103982776e-06,
      "loss": 0.3927,
      "step": 5422
    },
    {
      "epoch": 0.29264475743348983,
      "grad_norm": 1.1422291009629884,
      "learning_rate": 8.228443475431665e-06,
      "loss": 0.5209,
      "step": 5423
    },
    {
      "epoch": 0.29269872106200423,
      "grad_norm": 1.1079426577120186,
      "learning_rate": 8.22783676846233e-06,
      "loss": 0.51,
      "step": 5424
    },
    {
      "epoch": 0.29275268469051857,
      "grad_norm": 0.9478205754385052,
      "learning_rate": 8.227229983092211e-06,
      "loss": 0.4824,
      "step": 5425
    },
    {
      "epoch": 0.29280664831903297,
      "grad_norm": 1.0131658723271872,
      "learning_rate": 8.22662311933875e-06,
      "loss": 0.6823,
      "step": 5426
    },
    {
      "epoch": 0.29286061194754737,
      "grad_norm": 0.9083996947801121,
      "learning_rate": 8.22601617721938e-06,
      "loss": 0.4977,
      "step": 5427
    },
    {
      "epoch": 0.2929145755760617,
      "grad_norm": 1.0545415853905382,
      "learning_rate": 8.225409156751554e-06,
      "loss": 0.5247,
      "step": 5428
    },
    {
      "epoch": 0.2929685392045761,
      "grad_norm": 0.9476851606799219,
      "learning_rate": 8.224802057952714e-06,
      "loss": 0.6508,
      "step": 5429
    },
    {
      "epoch": 0.2930225028330905,
      "grad_norm": 1.0718557642674063,
      "learning_rate": 8.22419488084031e-06,
      "loss": 0.5035,
      "step": 5430
    },
    {
      "epoch": 0.2930764664616049,
      "grad_norm": 1.1004390861485833,
      "learning_rate": 8.223587625431791e-06,
      "loss": 0.5303,
      "step": 5431
    },
    {
      "epoch": 0.29313043009011924,
      "grad_norm": 0.9707714179839401,
      "learning_rate": 8.222980291744613e-06,
      "loss": 0.3874,
      "step": 5432
    },
    {
      "epoch": 0.29318439371863364,
      "grad_norm": 1.1599962859974324,
      "learning_rate": 8.22237287979623e-06,
      "loss": 0.5409,
      "step": 5433
    },
    {
      "epoch": 0.29323835734714804,
      "grad_norm": 1.319992450834645,
      "learning_rate": 8.221765389604099e-06,
      "loss": 0.6826,
      "step": 5434
    },
    {
      "epoch": 0.2932923209756624,
      "grad_norm": 1.1222617387727523,
      "learning_rate": 8.221157821185682e-06,
      "loss": 0.5894,
      "step": 5435
    },
    {
      "epoch": 0.2933462846041768,
      "grad_norm": 1.0258580884477058,
      "learning_rate": 8.220550174558439e-06,
      "loss": 0.4643,
      "step": 5436
    },
    {
      "epoch": 0.2934002482326912,
      "grad_norm": 0.8929660844363109,
      "learning_rate": 8.219942449739835e-06,
      "loss": 0.3826,
      "step": 5437
    },
    {
      "epoch": 0.29345421186120557,
      "grad_norm": 1.260848110419985,
      "learning_rate": 8.219334646747337e-06,
      "loss": 0.6333,
      "step": 5438
    },
    {
      "epoch": 0.2935081754897199,
      "grad_norm": 0.916291206183131,
      "learning_rate": 8.218726765598415e-06,
      "loss": 0.486,
      "step": 5439
    },
    {
      "epoch": 0.2935621391182343,
      "grad_norm": 0.9959265291854632,
      "learning_rate": 8.218118806310535e-06,
      "loss": 0.4151,
      "step": 5440
    },
    {
      "epoch": 0.2936161027467487,
      "grad_norm": 1.1356767435419513,
      "learning_rate": 8.217510768901174e-06,
      "loss": 0.4802,
      "step": 5441
    },
    {
      "epoch": 0.29367006637526305,
      "grad_norm": 0.9158152730978059,
      "learning_rate": 8.216902653387812e-06,
      "loss": 0.3979,
      "step": 5442
    },
    {
      "epoch": 0.29372403000377745,
      "grad_norm": 0.8560654957978295,
      "learning_rate": 8.216294459787918e-06,
      "loss": 0.3533,
      "step": 5443
    },
    {
      "epoch": 0.29377799363229185,
      "grad_norm": 1.1253049266915858,
      "learning_rate": 8.215686188118979e-06,
      "loss": 0.6517,
      "step": 5444
    },
    {
      "epoch": 0.29383195726080624,
      "grad_norm": 0.9451847467936215,
      "learning_rate": 8.215077838398475e-06,
      "loss": 0.5125,
      "step": 5445
    },
    {
      "epoch": 0.2938859208893206,
      "grad_norm": 1.1226813490154475,
      "learning_rate": 8.214469410643891e-06,
      "loss": 0.484,
      "step": 5446
    },
    {
      "epoch": 0.293939884517835,
      "grad_norm": 0.9629203593706895,
      "learning_rate": 8.21386090487271e-06,
      "loss": 0.3967,
      "step": 5447
    },
    {
      "epoch": 0.2939938481463494,
      "grad_norm": 1.1345575066789746,
      "learning_rate": 8.213252321102428e-06,
      "loss": 0.5399,
      "step": 5448
    },
    {
      "epoch": 0.2940478117748637,
      "grad_norm": 1.174680487968514,
      "learning_rate": 8.212643659350532e-06,
      "loss": 0.6315,
      "step": 5449
    },
    {
      "epoch": 0.2941017754033781,
      "grad_norm": 0.970722312137542,
      "learning_rate": 8.212034919634516e-06,
      "loss": 0.4666,
      "step": 5450
    },
    {
      "epoch": 0.2941557390318925,
      "grad_norm": 1.1106574122583126,
      "learning_rate": 8.211426101971876e-06,
      "loss": 0.6176,
      "step": 5451
    },
    {
      "epoch": 0.2942097026604069,
      "grad_norm": 0.9572675762355692,
      "learning_rate": 8.21081720638011e-06,
      "loss": 0.6453,
      "step": 5452
    },
    {
      "epoch": 0.29426366628892126,
      "grad_norm": 1.0082633080405732,
      "learning_rate": 8.210208232876721e-06,
      "loss": 0.4474,
      "step": 5453
    },
    {
      "epoch": 0.29431762991743565,
      "grad_norm": 0.9692325946468063,
      "learning_rate": 8.209599181479205e-06,
      "loss": 0.3964,
      "step": 5454
    },
    {
      "epoch": 0.29437159354595005,
      "grad_norm": 0.9185319853993519,
      "learning_rate": 8.208990052205073e-06,
      "loss": 0.4206,
      "step": 5455
    },
    {
      "epoch": 0.2944255571744644,
      "grad_norm": 1.0465981560266164,
      "learning_rate": 8.208380845071828e-06,
      "loss": 0.5051,
      "step": 5456
    },
    {
      "epoch": 0.2944795208029788,
      "grad_norm": 1.1622069843671032,
      "learning_rate": 8.207771560096982e-06,
      "loss": 0.5873,
      "step": 5457
    },
    {
      "epoch": 0.2945334844314932,
      "grad_norm": 1.2042948373974285,
      "learning_rate": 8.207162197298045e-06,
      "loss": 0.5492,
      "step": 5458
    },
    {
      "epoch": 0.29458744806000753,
      "grad_norm": 1.0334129981830786,
      "learning_rate": 8.206552756692532e-06,
      "loss": 0.5539,
      "step": 5459
    },
    {
      "epoch": 0.2946414116885219,
      "grad_norm": 0.967767523028003,
      "learning_rate": 8.205943238297956e-06,
      "loss": 0.3979,
      "step": 5460
    },
    {
      "epoch": 0.2946953753170363,
      "grad_norm": 1.030113174711838,
      "learning_rate": 8.205333642131838e-06,
      "loss": 0.4857,
      "step": 5461
    },
    {
      "epoch": 0.2947493389455507,
      "grad_norm": 1.0032352160325175,
      "learning_rate": 8.204723968211699e-06,
      "loss": 0.443,
      "step": 5462
    },
    {
      "epoch": 0.29480330257406506,
      "grad_norm": 1.2518514833718986,
      "learning_rate": 8.204114216555059e-06,
      "loss": 0.6762,
      "step": 5463
    },
    {
      "epoch": 0.29485726620257946,
      "grad_norm": 0.9502072659825652,
      "learning_rate": 8.203504387179446e-06,
      "loss": 0.4451,
      "step": 5464
    },
    {
      "epoch": 0.29491122983109386,
      "grad_norm": 1.308942790203988,
      "learning_rate": 8.202894480102382e-06,
      "loss": 0.6533,
      "step": 5465
    },
    {
      "epoch": 0.2949651934596082,
      "grad_norm": 1.2274345047975126,
      "learning_rate": 8.202284495341402e-06,
      "loss": 0.5171,
      "step": 5466
    },
    {
      "epoch": 0.2950191570881226,
      "grad_norm": 1.2400624548111532,
      "learning_rate": 8.201674432914033e-06,
      "loss": 0.6303,
      "step": 5467
    },
    {
      "epoch": 0.295073120716637,
      "grad_norm": 1.124512448773544,
      "learning_rate": 8.201064292837814e-06,
      "loss": 0.5204,
      "step": 5468
    },
    {
      "epoch": 0.2951270843451514,
      "grad_norm": 1.037354415023125,
      "learning_rate": 8.200454075130275e-06,
      "loss": 0.5452,
      "step": 5469
    },
    {
      "epoch": 0.29518104797366573,
      "grad_norm": 0.8969797021299134,
      "learning_rate": 8.19984377980896e-06,
      "loss": 0.3452,
      "step": 5470
    },
    {
      "epoch": 0.29523501160218013,
      "grad_norm": 1.0063473326281613,
      "learning_rate": 8.199233406891404e-06,
      "loss": 0.5309,
      "step": 5471
    },
    {
      "epoch": 0.29528897523069453,
      "grad_norm": 1.4009649960631787,
      "learning_rate": 8.198622956395153e-06,
      "loss": 0.686,
      "step": 5472
    },
    {
      "epoch": 0.29534293885920887,
      "grad_norm": 1.168596336176328,
      "learning_rate": 8.198012428337753e-06,
      "loss": 0.502,
      "step": 5473
    },
    {
      "epoch": 0.29539690248772327,
      "grad_norm": 1.0552223572901842,
      "learning_rate": 8.197401822736749e-06,
      "loss": 0.6253,
      "step": 5474
    },
    {
      "epoch": 0.29545086611623766,
      "grad_norm": 0.9684397244117144,
      "learning_rate": 8.196791139609691e-06,
      "loss": 0.4244,
      "step": 5475
    },
    {
      "epoch": 0.29550482974475206,
      "grad_norm": 1.0450748823856952,
      "learning_rate": 8.19618037897413e-06,
      "loss": 0.5186,
      "step": 5476
    },
    {
      "epoch": 0.2955587933732664,
      "grad_norm": 1.068329761333297,
      "learning_rate": 8.195569540847622e-06,
      "loss": 0.4829,
      "step": 5477
    },
    {
      "epoch": 0.2956127570017808,
      "grad_norm": 0.8647067709651073,
      "learning_rate": 8.194958625247721e-06,
      "loss": 0.4079,
      "step": 5478
    },
    {
      "epoch": 0.2956667206302952,
      "grad_norm": 1.0098411927804525,
      "learning_rate": 8.194347632191987e-06,
      "loss": 0.4356,
      "step": 5479
    },
    {
      "epoch": 0.29572068425880954,
      "grad_norm": 0.9538474646247859,
      "learning_rate": 8.193736561697979e-06,
      "loss": 0.3655,
      "step": 5480
    },
    {
      "epoch": 0.29577464788732394,
      "grad_norm": 1.0720353565271512,
      "learning_rate": 8.19312541378326e-06,
      "loss": 0.4534,
      "step": 5481
    },
    {
      "epoch": 0.29582861151583834,
      "grad_norm": 1.0095604090945147,
      "learning_rate": 8.192514188465395e-06,
      "loss": 0.5597,
      "step": 5482
    },
    {
      "epoch": 0.29588257514435273,
      "grad_norm": 0.935768151951977,
      "learning_rate": 8.19190288576195e-06,
      "loss": 0.3941,
      "step": 5483
    },
    {
      "epoch": 0.2959365387728671,
      "grad_norm": 1.3040973182569031,
      "learning_rate": 8.1912915056905e-06,
      "loss": 0.6695,
      "step": 5484
    },
    {
      "epoch": 0.2959905024013815,
      "grad_norm": 0.9114292938686618,
      "learning_rate": 8.190680048268611e-06,
      "loss": 0.4794,
      "step": 5485
    },
    {
      "epoch": 0.29604446602989587,
      "grad_norm": 1.3142024446146092,
      "learning_rate": 8.190068513513859e-06,
      "loss": 0.5565,
      "step": 5486
    },
    {
      "epoch": 0.2960984296584102,
      "grad_norm": 0.8929360891297731,
      "learning_rate": 8.189456901443821e-06,
      "loss": 0.4014,
      "step": 5487
    },
    {
      "epoch": 0.2961523932869246,
      "grad_norm": 0.9409169018175859,
      "learning_rate": 8.18884521207607e-06,
      "loss": 0.5017,
      "step": 5488
    },
    {
      "epoch": 0.296206356915439,
      "grad_norm": 1.1118146943914398,
      "learning_rate": 8.188233445428194e-06,
      "loss": 0.4724,
      "step": 5489
    },
    {
      "epoch": 0.29626032054395335,
      "grad_norm": 1.0800866486821898,
      "learning_rate": 8.187621601517773e-06,
      "loss": 0.4814,
      "step": 5490
    },
    {
      "epoch": 0.29631428417246775,
      "grad_norm": 1.0933208759476125,
      "learning_rate": 8.18700968036239e-06,
      "loss": 0.4988,
      "step": 5491
    },
    {
      "epoch": 0.29636824780098214,
      "grad_norm": 1.030853753670057,
      "learning_rate": 8.186397681979635e-06,
      "loss": 0.3688,
      "step": 5492
    },
    {
      "epoch": 0.29642221142949654,
      "grad_norm": 1.0572947431436854,
      "learning_rate": 8.185785606387095e-06,
      "loss": 0.6066,
      "step": 5493
    },
    {
      "epoch": 0.2964761750580109,
      "grad_norm": 0.9663742219454755,
      "learning_rate": 8.185173453602363e-06,
      "loss": 0.4927,
      "step": 5494
    },
    {
      "epoch": 0.2965301386865253,
      "grad_norm": 0.8451379195775102,
      "learning_rate": 8.184561223643034e-06,
      "loss": 0.4113,
      "step": 5495
    },
    {
      "epoch": 0.2965841023150397,
      "grad_norm": 1.1458128831754122,
      "learning_rate": 8.183948916526701e-06,
      "loss": 0.5873,
      "step": 5496
    },
    {
      "epoch": 0.296638065943554,
      "grad_norm": 0.8250334071608266,
      "learning_rate": 8.183336532270966e-06,
      "loss": 0.2951,
      "step": 5497
    },
    {
      "epoch": 0.2966920295720684,
      "grad_norm": 1.075284745846835,
      "learning_rate": 8.182724070893427e-06,
      "loss": 0.4702,
      "step": 5498
    },
    {
      "epoch": 0.2967459932005828,
      "grad_norm": 0.8906695467443344,
      "learning_rate": 8.182111532411686e-06,
      "loss": 0.3468,
      "step": 5499
    },
    {
      "epoch": 0.2967999568290972,
      "grad_norm": 0.9297036636633297,
      "learning_rate": 8.181498916843352e-06,
      "loss": 0.4743,
      "step": 5500
    },
    {
      "epoch": 0.2967999568290972,
      "eval_loss": 0.5707113742828369,
      "eval_runtime": 163.793,
      "eval_samples_per_second": 20.996,
      "eval_steps_per_second": 0.879,
      "step": 5500
    },
    {
      "epoch": 0.29685392045761155,
      "grad_norm": 0.9712918253502731,
      "learning_rate": 8.180886224206028e-06,
      "loss": 0.5744,
      "step": 5501
    },
    {
      "epoch": 0.29690788408612595,
      "grad_norm": 1.1044244633896194,
      "learning_rate": 8.180273454517325e-06,
      "loss": 0.3832,
      "step": 5502
    },
    {
      "epoch": 0.29696184771464035,
      "grad_norm": 1.273284977741108,
      "learning_rate": 8.179660607794855e-06,
      "loss": 0.5419,
      "step": 5503
    },
    {
      "epoch": 0.2970158113431547,
      "grad_norm": 0.9816134931221421,
      "learning_rate": 8.179047684056233e-06,
      "loss": 0.4513,
      "step": 5504
    },
    {
      "epoch": 0.2970697749716691,
      "grad_norm": 1.0999858716250308,
      "learning_rate": 8.178434683319072e-06,
      "loss": 0.6246,
      "step": 5505
    },
    {
      "epoch": 0.2971237386001835,
      "grad_norm": 1.1085406786154242,
      "learning_rate": 8.177821605600992e-06,
      "loss": 0.5914,
      "step": 5506
    },
    {
      "epoch": 0.2971777022286979,
      "grad_norm": 0.9654091907524764,
      "learning_rate": 8.177208450919613e-06,
      "loss": 0.4483,
      "step": 5507
    },
    {
      "epoch": 0.2972316658572122,
      "grad_norm": 1.5915707278993687,
      "learning_rate": 8.17659521929256e-06,
      "loss": 0.604,
      "step": 5508
    },
    {
      "epoch": 0.2972856294857266,
      "grad_norm": 1.1974455303308587,
      "learning_rate": 8.175981910737454e-06,
      "loss": 0.5802,
      "step": 5509
    },
    {
      "epoch": 0.297339593114241,
      "grad_norm": 0.9642831386101156,
      "learning_rate": 8.175368525271924e-06,
      "loss": 0.6693,
      "step": 5510
    },
    {
      "epoch": 0.29739355674275536,
      "grad_norm": 1.290045038065177,
      "learning_rate": 8.174755062913601e-06,
      "loss": 0.4657,
      "step": 5511
    },
    {
      "epoch": 0.29744752037126976,
      "grad_norm": 1.0461644271114532,
      "learning_rate": 8.174141523680114e-06,
      "loss": 0.5876,
      "step": 5512
    },
    {
      "epoch": 0.29750148399978416,
      "grad_norm": 0.8306402278676918,
      "learning_rate": 8.173527907589097e-06,
      "loss": 0.4327,
      "step": 5513
    },
    {
      "epoch": 0.29755544762829855,
      "grad_norm": 1.5404685765421697,
      "learning_rate": 8.172914214658187e-06,
      "loss": 0.7321,
      "step": 5514
    },
    {
      "epoch": 0.2976094112568129,
      "grad_norm": 0.9319422185521044,
      "learning_rate": 8.172300444905023e-06,
      "loss": 0.4465,
      "step": 5515
    },
    {
      "epoch": 0.2976633748853273,
      "grad_norm": 1.017641275949194,
      "learning_rate": 8.17168659834724e-06,
      "loss": 0.5397,
      "step": 5516
    },
    {
      "epoch": 0.2977173385138417,
      "grad_norm": 1.1637338596536795,
      "learning_rate": 8.171072675002487e-06,
      "loss": 0.4962,
      "step": 5517
    },
    {
      "epoch": 0.29777130214235603,
      "grad_norm": 1.1100342430534302,
      "learning_rate": 8.170458674888408e-06,
      "loss": 0.4538,
      "step": 5518
    },
    {
      "epoch": 0.29782526577087043,
      "grad_norm": 1.0310449234586758,
      "learning_rate": 8.169844598022647e-06,
      "loss": 0.4852,
      "step": 5519
    },
    {
      "epoch": 0.2978792293993848,
      "grad_norm": 1.1038785231301893,
      "learning_rate": 8.169230444422854e-06,
      "loss": 0.4856,
      "step": 5520
    },
    {
      "epoch": 0.2979331930278992,
      "grad_norm": 0.8231013447904444,
      "learning_rate": 8.168616214106682e-06,
      "loss": 0.4012,
      "step": 5521
    },
    {
      "epoch": 0.29798715665641357,
      "grad_norm": 1.1562261800992588,
      "learning_rate": 8.168001907091783e-06,
      "loss": 0.4114,
      "step": 5522
    },
    {
      "epoch": 0.29804112028492796,
      "grad_norm": 1.1682632849547983,
      "learning_rate": 8.167387523395814e-06,
      "loss": 0.5034,
      "step": 5523
    },
    {
      "epoch": 0.29809508391344236,
      "grad_norm": 1.0879205718093867,
      "learning_rate": 8.166773063036431e-06,
      "loss": 0.4802,
      "step": 5524
    },
    {
      "epoch": 0.2981490475419567,
      "grad_norm": 0.8309307951049498,
      "learning_rate": 8.166158526031296e-06,
      "loss": 0.3696,
      "step": 5525
    },
    {
      "epoch": 0.2982030111704711,
      "grad_norm": 1.1950457863844937,
      "learning_rate": 8.165543912398073e-06,
      "loss": 0.6335,
      "step": 5526
    },
    {
      "epoch": 0.2982569747989855,
      "grad_norm": 0.990844990881177,
      "learning_rate": 8.16492922215442e-06,
      "loss": 0.4283,
      "step": 5527
    },
    {
      "epoch": 0.29831093842749984,
      "grad_norm": 1.28814764784975,
      "learning_rate": 8.164314455318013e-06,
      "loss": 0.4542,
      "step": 5528
    },
    {
      "epoch": 0.29836490205601424,
      "grad_norm": 0.9699348030926074,
      "learning_rate": 8.163699611906517e-06,
      "loss": 0.4362,
      "step": 5529
    },
    {
      "epoch": 0.29841886568452863,
      "grad_norm": 1.0363943659220578,
      "learning_rate": 8.163084691937599e-06,
      "loss": 0.5889,
      "step": 5530
    },
    {
      "epoch": 0.29847282931304303,
      "grad_norm": 1.2902372375169315,
      "learning_rate": 8.162469695428938e-06,
      "loss": 0.6224,
      "step": 5531
    },
    {
      "epoch": 0.2985267929415574,
      "grad_norm": 1.143048644818585,
      "learning_rate": 8.161854622398207e-06,
      "loss": 0.5849,
      "step": 5532
    },
    {
      "epoch": 0.29858075657007177,
      "grad_norm": 0.7877927090924,
      "learning_rate": 8.161239472863086e-06,
      "loss": 0.3388,
      "step": 5533
    },
    {
      "epoch": 0.29863472019858617,
      "grad_norm": 1.1883476167480753,
      "learning_rate": 8.16062424684125e-06,
      "loss": 0.5812,
      "step": 5534
    },
    {
      "epoch": 0.2986886838271005,
      "grad_norm": 0.9611499252872551,
      "learning_rate": 8.160008944350389e-06,
      "loss": 0.5783,
      "step": 5535
    },
    {
      "epoch": 0.2987426474556149,
      "grad_norm": 1.167194486148674,
      "learning_rate": 8.159393565408181e-06,
      "loss": 0.6517,
      "step": 5536
    },
    {
      "epoch": 0.2987966110841293,
      "grad_norm": 0.8945729996759143,
      "learning_rate": 8.158778110032316e-06,
      "loss": 0.418,
      "step": 5537
    },
    {
      "epoch": 0.2988505747126437,
      "grad_norm": 0.938910285460627,
      "learning_rate": 8.158162578240479e-06,
      "loss": 0.502,
      "step": 5538
    },
    {
      "epoch": 0.29890453834115804,
      "grad_norm": 1.0090214275223737,
      "learning_rate": 8.157546970050368e-06,
      "loss": 0.4514,
      "step": 5539
    },
    {
      "epoch": 0.29895850196967244,
      "grad_norm": 0.9330970410632513,
      "learning_rate": 8.15693128547967e-06,
      "loss": 0.4529,
      "step": 5540
    },
    {
      "epoch": 0.29901246559818684,
      "grad_norm": 1.0695322642703895,
      "learning_rate": 8.15631552454608e-06,
      "loss": 0.4586,
      "step": 5541
    },
    {
      "epoch": 0.2990664292267012,
      "grad_norm": 0.8922299103968803,
      "learning_rate": 8.155699687267302e-06,
      "loss": 0.462,
      "step": 5542
    },
    {
      "epoch": 0.2991203928552156,
      "grad_norm": 1.1305598644689097,
      "learning_rate": 8.155083773661027e-06,
      "loss": 0.5908,
      "step": 5543
    },
    {
      "epoch": 0.29917435648373,
      "grad_norm": 1.4083299711571309,
      "learning_rate": 8.154467783744962e-06,
      "loss": 0.5023,
      "step": 5544
    },
    {
      "epoch": 0.2992283201122444,
      "grad_norm": 0.9540100168050423,
      "learning_rate": 8.153851717536813e-06,
      "loss": 0.4728,
      "step": 5545
    },
    {
      "epoch": 0.2992822837407587,
      "grad_norm": 1.0334761589473818,
      "learning_rate": 8.153235575054284e-06,
      "loss": 0.4552,
      "step": 5546
    },
    {
      "epoch": 0.2993362473692731,
      "grad_norm": 1.1799571760495575,
      "learning_rate": 8.152619356315083e-06,
      "loss": 0.6491,
      "step": 5547
    },
    {
      "epoch": 0.2993902109977875,
      "grad_norm": 0.9253563342651993,
      "learning_rate": 8.152003061336923e-06,
      "loss": 0.4588,
      "step": 5548
    },
    {
      "epoch": 0.29944417462630185,
      "grad_norm": 0.9901137242563935,
      "learning_rate": 8.151386690137513e-06,
      "loss": 0.4987,
      "step": 5549
    },
    {
      "epoch": 0.29949813825481625,
      "grad_norm": 0.8852548257952214,
      "learning_rate": 8.150770242734572e-06,
      "loss": 0.4499,
      "step": 5550
    },
    {
      "epoch": 0.29955210188333065,
      "grad_norm": 0.7695439621827244,
      "learning_rate": 8.150153719145816e-06,
      "loss": 0.291,
      "step": 5551
    },
    {
      "epoch": 0.29960606551184504,
      "grad_norm": 0.9319945764255112,
      "learning_rate": 8.149537119388962e-06,
      "loss": 0.4422,
      "step": 5552
    },
    {
      "epoch": 0.2996600291403594,
      "grad_norm": 1.4321809363268323,
      "learning_rate": 8.148920443481735e-06,
      "loss": 0.827,
      "step": 5553
    },
    {
      "epoch": 0.2997139927688738,
      "grad_norm": 0.9170146925411133,
      "learning_rate": 8.14830369144186e-06,
      "loss": 0.4488,
      "step": 5554
    },
    {
      "epoch": 0.2997679563973882,
      "grad_norm": 1.099364464947338,
      "learning_rate": 8.147686863287059e-06,
      "loss": 0.4642,
      "step": 5555
    },
    {
      "epoch": 0.2998219200259025,
      "grad_norm": 1.1007866771080732,
      "learning_rate": 8.147069959035064e-06,
      "loss": 0.5789,
      "step": 5556
    },
    {
      "epoch": 0.2998758836544169,
      "grad_norm": 1.0343223268387993,
      "learning_rate": 8.146452978703602e-06,
      "loss": 0.5172,
      "step": 5557
    },
    {
      "epoch": 0.2999298472829313,
      "grad_norm": 1.190795450972479,
      "learning_rate": 8.145835922310408e-06,
      "loss": 0.5474,
      "step": 5558
    },
    {
      "epoch": 0.29998381091144566,
      "grad_norm": 1.163337483113725,
      "learning_rate": 8.145218789873215e-06,
      "loss": 0.5431,
      "step": 5559
    },
    {
      "epoch": 0.30003777453996006,
      "grad_norm": 0.8093379823966488,
      "learning_rate": 8.144601581409764e-06,
      "loss": 0.3599,
      "step": 5560
    },
    {
      "epoch": 0.30009173816847445,
      "grad_norm": 1.0080482557997457,
      "learning_rate": 8.143984296937788e-06,
      "loss": 0.5517,
      "step": 5561
    },
    {
      "epoch": 0.30014570179698885,
      "grad_norm": 1.0964024267343018,
      "learning_rate": 8.143366936475033e-06,
      "loss": 0.5297,
      "step": 5562
    },
    {
      "epoch": 0.3001996654255032,
      "grad_norm": 1.290066255719801,
      "learning_rate": 8.142749500039243e-06,
      "loss": 0.6599,
      "step": 5563
    },
    {
      "epoch": 0.3002536290540176,
      "grad_norm": 1.0838978780428834,
      "learning_rate": 8.14213198764816e-06,
      "loss": 0.5452,
      "step": 5564
    },
    {
      "epoch": 0.300307592682532,
      "grad_norm": 0.9728809230325576,
      "learning_rate": 8.141514399319534e-06,
      "loss": 0.4632,
      "step": 5565
    },
    {
      "epoch": 0.30036155631104633,
      "grad_norm": 0.813989758108876,
      "learning_rate": 8.140896735071117e-06,
      "loss": 0.3638,
      "step": 5566
    },
    {
      "epoch": 0.3004155199395607,
      "grad_norm": 1.1119372025840815,
      "learning_rate": 8.140278994920657e-06,
      "loss": 0.4794,
      "step": 5567
    },
    {
      "epoch": 0.3004694835680751,
      "grad_norm": 1.0935860759829532,
      "learning_rate": 8.139661178885912e-06,
      "loss": 0.4132,
      "step": 5568
    },
    {
      "epoch": 0.3005234471965895,
      "grad_norm": 1.0107179224779197,
      "learning_rate": 8.139043286984638e-06,
      "loss": 0.4807,
      "step": 5569
    },
    {
      "epoch": 0.30057741082510386,
      "grad_norm": 0.8825258336629854,
      "learning_rate": 8.138425319234593e-06,
      "loss": 0.4398,
      "step": 5570
    },
    {
      "epoch": 0.30063137445361826,
      "grad_norm": 1.1179407831170807,
      "learning_rate": 8.137807275653538e-06,
      "loss": 0.4667,
      "step": 5571
    },
    {
      "epoch": 0.30068533808213266,
      "grad_norm": 1.1469055847717473,
      "learning_rate": 8.137189156259238e-06,
      "loss": 0.5493,
      "step": 5572
    },
    {
      "epoch": 0.300739301710647,
      "grad_norm": 1.084633207072137,
      "learning_rate": 8.136570961069456e-06,
      "loss": 0.441,
      "step": 5573
    },
    {
      "epoch": 0.3007932653391614,
      "grad_norm": 1.0297921058476203,
      "learning_rate": 8.13595269010196e-06,
      "loss": 0.589,
      "step": 5574
    },
    {
      "epoch": 0.3008472289676758,
      "grad_norm": 0.8372349074843124,
      "learning_rate": 8.135334343374522e-06,
      "loss": 0.3748,
      "step": 5575
    },
    {
      "epoch": 0.3009011925961902,
      "grad_norm": 1.0258839611266815,
      "learning_rate": 8.13471592090491e-06,
      "loss": 0.5307,
      "step": 5576
    },
    {
      "epoch": 0.30095515622470453,
      "grad_norm": 0.8096973955557034,
      "learning_rate": 8.1340974227109e-06,
      "loss": 0.3455,
      "step": 5577
    },
    {
      "epoch": 0.30100911985321893,
      "grad_norm": 1.1399273209511402,
      "learning_rate": 8.13347884881027e-06,
      "loss": 0.5883,
      "step": 5578
    },
    {
      "epoch": 0.30106308348173333,
      "grad_norm": 0.9137459273293648,
      "learning_rate": 8.132860199220796e-06,
      "loss": 0.4103,
      "step": 5579
    },
    {
      "epoch": 0.30111704711024767,
      "grad_norm": 0.7986444779728575,
      "learning_rate": 8.132241473960262e-06,
      "loss": 0.4176,
      "step": 5580
    },
    {
      "epoch": 0.30117101073876207,
      "grad_norm": 1.023032533506991,
      "learning_rate": 8.131622673046446e-06,
      "loss": 0.4219,
      "step": 5581
    },
    {
      "epoch": 0.30122497436727647,
      "grad_norm": 1.441354694703945,
      "learning_rate": 8.131003796497136e-06,
      "loss": 0.6018,
      "step": 5582
    },
    {
      "epoch": 0.30127893799579086,
      "grad_norm": 0.9453804870701896,
      "learning_rate": 8.130384844330118e-06,
      "loss": 0.3586,
      "step": 5583
    },
    {
      "epoch": 0.3013329016243052,
      "grad_norm": 1.0246365104445057,
      "learning_rate": 8.129765816563182e-06,
      "loss": 0.5022,
      "step": 5584
    },
    {
      "epoch": 0.3013868652528196,
      "grad_norm": 1.1052074177657991,
      "learning_rate": 8.129146713214119e-06,
      "loss": 0.5767,
      "step": 5585
    },
    {
      "epoch": 0.301440828881334,
      "grad_norm": 1.1818970903401833,
      "learning_rate": 8.128527534300723e-06,
      "loss": 0.6273,
      "step": 5586
    },
    {
      "epoch": 0.30149479250984834,
      "grad_norm": 1.170423927567819,
      "learning_rate": 8.127908279840791e-06,
      "loss": 0.5209,
      "step": 5587
    },
    {
      "epoch": 0.30154875613836274,
      "grad_norm": 1.320802168194481,
      "learning_rate": 8.127288949852118e-06,
      "loss": 0.5008,
      "step": 5588
    },
    {
      "epoch": 0.30160271976687714,
      "grad_norm": 1.0899490896311732,
      "learning_rate": 8.126669544352506e-06,
      "loss": 0.5052,
      "step": 5589
    },
    {
      "epoch": 0.30165668339539153,
      "grad_norm": 0.9903482526565055,
      "learning_rate": 8.126050063359757e-06,
      "loss": 0.5444,
      "step": 5590
    },
    {
      "epoch": 0.3017106470239059,
      "grad_norm": 0.9486053527463052,
      "learning_rate": 8.125430506891677e-06,
      "loss": 0.4294,
      "step": 5591
    },
    {
      "epoch": 0.3017646106524203,
      "grad_norm": 1.1565609846487668,
      "learning_rate": 8.12481087496607e-06,
      "loss": 0.5562,
      "step": 5592
    },
    {
      "epoch": 0.30181857428093467,
      "grad_norm": 1.2788610653093737,
      "learning_rate": 8.124191167600749e-06,
      "loss": 0.5142,
      "step": 5593
    },
    {
      "epoch": 0.301872537909449,
      "grad_norm": 1.0357093273374591,
      "learning_rate": 8.123571384813521e-06,
      "loss": 0.4588,
      "step": 5594
    },
    {
      "epoch": 0.3019265015379634,
      "grad_norm": 1.2197215911674228,
      "learning_rate": 8.122951526622199e-06,
      "loss": 0.5764,
      "step": 5595
    },
    {
      "epoch": 0.3019804651664778,
      "grad_norm": 1.1537585243467803,
      "learning_rate": 8.1223315930446e-06,
      "loss": 0.4942,
      "step": 5596
    },
    {
      "epoch": 0.30203442879499215,
      "grad_norm": 0.9162902118186534,
      "learning_rate": 8.121711584098544e-06,
      "loss": 0.4406,
      "step": 5597
    },
    {
      "epoch": 0.30208839242350655,
      "grad_norm": 0.8782032433844676,
      "learning_rate": 8.121091499801846e-06,
      "loss": 0.589,
      "step": 5598
    },
    {
      "epoch": 0.30214235605202094,
      "grad_norm": 0.9464663188097919,
      "learning_rate": 8.120471340172328e-06,
      "loss": 0.4401,
      "step": 5599
    },
    {
      "epoch": 0.30219631968053534,
      "grad_norm": 1.006411203481918,
      "learning_rate": 8.119851105227819e-06,
      "loss": 0.3438,
      "step": 5600
    },
    {
      "epoch": 0.3022502833090497,
      "grad_norm": 0.953975744019338,
      "learning_rate": 8.119230794986143e-06,
      "loss": 0.4899,
      "step": 5601
    },
    {
      "epoch": 0.3023042469375641,
      "grad_norm": 1.0974408589153004,
      "learning_rate": 8.118610409465125e-06,
      "loss": 0.4289,
      "step": 5602
    },
    {
      "epoch": 0.3023582105660785,
      "grad_norm": 1.0855741626603244,
      "learning_rate": 8.117989948682597e-06,
      "loss": 0.4042,
      "step": 5603
    },
    {
      "epoch": 0.3024121741945928,
      "grad_norm": 0.9991161230718788,
      "learning_rate": 8.117369412656395e-06,
      "loss": 0.4602,
      "step": 5604
    },
    {
      "epoch": 0.3024661378231072,
      "grad_norm": 0.8678952774258104,
      "learning_rate": 8.11674880140435e-06,
      "loss": 0.3918,
      "step": 5605
    },
    {
      "epoch": 0.3025201014516216,
      "grad_norm": 0.9224038609131077,
      "learning_rate": 8.1161281149443e-06,
      "loss": 0.4366,
      "step": 5606
    },
    {
      "epoch": 0.302574065080136,
      "grad_norm": 1.2580085546909257,
      "learning_rate": 8.115507353294086e-06,
      "loss": 0.5465,
      "step": 5607
    },
    {
      "epoch": 0.30262802870865035,
      "grad_norm": 1.010569757245223,
      "learning_rate": 8.114886516471547e-06,
      "loss": 0.4961,
      "step": 5608
    },
    {
      "epoch": 0.30268199233716475,
      "grad_norm": 0.9199854508735003,
      "learning_rate": 8.114265604494528e-06,
      "loss": 0.3388,
      "step": 5609
    },
    {
      "epoch": 0.30273595596567915,
      "grad_norm": 1.1129787015246886,
      "learning_rate": 8.113644617380874e-06,
      "loss": 0.4383,
      "step": 5610
    },
    {
      "epoch": 0.3027899195941935,
      "grad_norm": 1.041937101636486,
      "learning_rate": 8.113023555148432e-06,
      "loss": 0.5354,
      "step": 5611
    },
    {
      "epoch": 0.3028438832227079,
      "grad_norm": 1.0270096053696323,
      "learning_rate": 8.112402417815053e-06,
      "loss": 0.5185,
      "step": 5612
    },
    {
      "epoch": 0.3028978468512223,
      "grad_norm": 0.9702692240517283,
      "learning_rate": 8.11178120539859e-06,
      "loss": 0.3952,
      "step": 5613
    },
    {
      "epoch": 0.3029518104797367,
      "grad_norm": 1.1345670728996977,
      "learning_rate": 8.111159917916895e-06,
      "loss": 0.7042,
      "step": 5614
    },
    {
      "epoch": 0.303005774108251,
      "grad_norm": 0.9973945848772792,
      "learning_rate": 8.110538555387825e-06,
      "loss": 0.5149,
      "step": 5615
    },
    {
      "epoch": 0.3030597377367654,
      "grad_norm": 0.9103133840872114,
      "learning_rate": 8.10991711782924e-06,
      "loss": 0.4237,
      "step": 5616
    },
    {
      "epoch": 0.3031137013652798,
      "grad_norm": 0.9735867666074121,
      "learning_rate": 8.109295605258999e-06,
      "loss": 0.4926,
      "step": 5617
    },
    {
      "epoch": 0.30316766499379416,
      "grad_norm": 0.9424365079311553,
      "learning_rate": 8.108674017694967e-06,
      "loss": 0.4675,
      "step": 5618
    },
    {
      "epoch": 0.30322162862230856,
      "grad_norm": 0.9558619225225495,
      "learning_rate": 8.108052355155009e-06,
      "loss": 0.4793,
      "step": 5619
    },
    {
      "epoch": 0.30327559225082296,
      "grad_norm": 0.8643693298393155,
      "learning_rate": 8.107430617656988e-06,
      "loss": 0.3747,
      "step": 5620
    },
    {
      "epoch": 0.30332955587933735,
      "grad_norm": 1.2971067720772373,
      "learning_rate": 8.106808805218778e-06,
      "loss": 0.5317,
      "step": 5621
    },
    {
      "epoch": 0.3033835195078517,
      "grad_norm": 1.3828155170950338,
      "learning_rate": 8.10618691785825e-06,
      "loss": 0.6234,
      "step": 5622
    },
    {
      "epoch": 0.3034374831363661,
      "grad_norm": 1.1002875177925207,
      "learning_rate": 8.105564955593276e-06,
      "loss": 0.5742,
      "step": 5623
    },
    {
      "epoch": 0.3034914467648805,
      "grad_norm": 1.0621698177276115,
      "learning_rate": 8.10494291844173e-06,
      "loss": 0.5223,
      "step": 5624
    },
    {
      "epoch": 0.30354541039339483,
      "grad_norm": 1.115907113000123,
      "learning_rate": 8.104320806421495e-06,
      "loss": 0.5872,
      "step": 5625
    },
    {
      "epoch": 0.30359937402190923,
      "grad_norm": 0.9397455022294577,
      "learning_rate": 8.103698619550448e-06,
      "loss": 0.5793,
      "step": 5626
    },
    {
      "epoch": 0.3036533376504236,
      "grad_norm": 0.9175872459607314,
      "learning_rate": 8.103076357846473e-06,
      "loss": 0.4196,
      "step": 5627
    },
    {
      "epoch": 0.30370730127893797,
      "grad_norm": 0.628694421975127,
      "learning_rate": 8.102454021327453e-06,
      "loss": 0.2351,
      "step": 5628
    },
    {
      "epoch": 0.30376126490745237,
      "grad_norm": 0.9911911831146212,
      "learning_rate": 8.101831610011273e-06,
      "loss": 0.4012,
      "step": 5629
    },
    {
      "epoch": 0.30381522853596676,
      "grad_norm": 1.0363664495500855,
      "learning_rate": 8.101209123915824e-06,
      "loss": 0.5631,
      "step": 5630
    },
    {
      "epoch": 0.30386919216448116,
      "grad_norm": 1.0672446721283464,
      "learning_rate": 8.100586563058997e-06,
      "loss": 0.5057,
      "step": 5631
    },
    {
      "epoch": 0.3039231557929955,
      "grad_norm": 0.8957773164484549,
      "learning_rate": 8.099963927458684e-06,
      "loss": 0.4355,
      "step": 5632
    },
    {
      "epoch": 0.3039771194215099,
      "grad_norm": 1.6643713377153944,
      "learning_rate": 8.09934121713278e-06,
      "loss": 0.7048,
      "step": 5633
    },
    {
      "epoch": 0.3040310830500243,
      "grad_norm": 0.8934258465842406,
      "learning_rate": 8.098718432099186e-06,
      "loss": 0.3854,
      "step": 5634
    },
    {
      "epoch": 0.30408504667853864,
      "grad_norm": 1.0817116820246995,
      "learning_rate": 8.098095572375796e-06,
      "loss": 0.7578,
      "step": 5635
    },
    {
      "epoch": 0.30413901030705304,
      "grad_norm": 1.2359778806944424,
      "learning_rate": 8.097472637980515e-06,
      "loss": 0.6524,
      "step": 5636
    },
    {
      "epoch": 0.30419297393556743,
      "grad_norm": 1.3710073335821784,
      "learning_rate": 8.096849628931244e-06,
      "loss": 0.643,
      "step": 5637
    },
    {
      "epoch": 0.30424693756408183,
      "grad_norm": 1.1954399048586752,
      "learning_rate": 8.096226545245893e-06,
      "loss": 0.5407,
      "step": 5638
    },
    {
      "epoch": 0.3043009011925962,
      "grad_norm": 1.0643216815620422,
      "learning_rate": 8.095603386942366e-06,
      "loss": 0.5492,
      "step": 5639
    },
    {
      "epoch": 0.30435486482111057,
      "grad_norm": 0.9259441105607429,
      "learning_rate": 8.094980154038576e-06,
      "loss": 0.4543,
      "step": 5640
    },
    {
      "epoch": 0.30440882844962497,
      "grad_norm": 0.9137327181493116,
      "learning_rate": 8.094356846552435e-06,
      "loss": 0.3863,
      "step": 5641
    },
    {
      "epoch": 0.3044627920781393,
      "grad_norm": 1.0378723688587277,
      "learning_rate": 8.093733464501856e-06,
      "loss": 0.4317,
      "step": 5642
    },
    {
      "epoch": 0.3045167557066537,
      "grad_norm": 1.130225795149892,
      "learning_rate": 8.093110007904758e-06,
      "loss": 0.6817,
      "step": 5643
    },
    {
      "epoch": 0.3045707193351681,
      "grad_norm": 0.9528413177732721,
      "learning_rate": 8.092486476779055e-06,
      "loss": 0.4801,
      "step": 5644
    },
    {
      "epoch": 0.3046246829636825,
      "grad_norm": 0.8584354188463329,
      "learning_rate": 8.091862871142673e-06,
      "loss": 0.416,
      "step": 5645
    },
    {
      "epoch": 0.30467864659219684,
      "grad_norm": 1.0655929187142261,
      "learning_rate": 8.091239191013534e-06,
      "loss": 0.5465,
      "step": 5646
    },
    {
      "epoch": 0.30473261022071124,
      "grad_norm": 1.1479349997764827,
      "learning_rate": 8.09061543640956e-06,
      "loss": 0.5168,
      "step": 5647
    },
    {
      "epoch": 0.30478657384922564,
      "grad_norm": 1.1849258972780432,
      "learning_rate": 8.089991607348682e-06,
      "loss": 0.5216,
      "step": 5648
    },
    {
      "epoch": 0.30484053747774,
      "grad_norm": 1.1107529981419964,
      "learning_rate": 8.08936770384883e-06,
      "loss": 0.5037,
      "step": 5649
    },
    {
      "epoch": 0.3048945011062544,
      "grad_norm": 1.0657859810291685,
      "learning_rate": 8.088743725927932e-06,
      "loss": 0.5348,
      "step": 5650
    },
    {
      "epoch": 0.3049484647347688,
      "grad_norm": 1.0764146558525167,
      "learning_rate": 8.088119673603924e-06,
      "loss": 0.3931,
      "step": 5651
    },
    {
      "epoch": 0.3050024283632832,
      "grad_norm": 1.1273693100484672,
      "learning_rate": 8.087495546894742e-06,
      "loss": 0.5869,
      "step": 5652
    },
    {
      "epoch": 0.3050563919917975,
      "grad_norm": 1.3430042820025332,
      "learning_rate": 8.086871345818322e-06,
      "loss": 0.5854,
      "step": 5653
    },
    {
      "epoch": 0.3051103556203119,
      "grad_norm": 0.9747383968458961,
      "learning_rate": 8.086247070392606e-06,
      "loss": 0.4819,
      "step": 5654
    },
    {
      "epoch": 0.3051643192488263,
      "grad_norm": 1.0013086931490989,
      "learning_rate": 8.085622720635536e-06,
      "loss": 0.3576,
      "step": 5655
    },
    {
      "epoch": 0.30521828287734065,
      "grad_norm": 1.2214685092822675,
      "learning_rate": 8.084998296565057e-06,
      "loss": 0.5054,
      "step": 5656
    },
    {
      "epoch": 0.30527224650585505,
      "grad_norm": 0.982081155987322,
      "learning_rate": 8.084373798199117e-06,
      "loss": 0.4717,
      "step": 5657
    },
    {
      "epoch": 0.30532621013436945,
      "grad_norm": 1.0092086351766365,
      "learning_rate": 8.083749225555659e-06,
      "loss": 0.4944,
      "step": 5658
    },
    {
      "epoch": 0.3053801737628838,
      "grad_norm": 1.1491669676617895,
      "learning_rate": 8.08312457865264e-06,
      "loss": 0.4769,
      "step": 5659
    },
    {
      "epoch": 0.3054341373913982,
      "grad_norm": 1.305334787842396,
      "learning_rate": 8.082499857508009e-06,
      "loss": 0.5999,
      "step": 5660
    },
    {
      "epoch": 0.3054881010199126,
      "grad_norm": 0.9042370450695271,
      "learning_rate": 8.081875062139724e-06,
      "loss": 0.428,
      "step": 5661
    },
    {
      "epoch": 0.305542064648427,
      "grad_norm": 0.9052514722664241,
      "learning_rate": 8.081250192565739e-06,
      "loss": 0.4168,
      "step": 5662
    },
    {
      "epoch": 0.3055960282769413,
      "grad_norm": 1.2640739692276923,
      "learning_rate": 8.080625248804018e-06,
      "loss": 0.684,
      "step": 5663
    },
    {
      "epoch": 0.3056499919054557,
      "grad_norm": 1.1584083082491656,
      "learning_rate": 8.080000230872518e-06,
      "loss": 0.5252,
      "step": 5664
    },
    {
      "epoch": 0.3057039555339701,
      "grad_norm": 0.9232452297622722,
      "learning_rate": 8.079375138789204e-06,
      "loss": 0.4807,
      "step": 5665
    },
    {
      "epoch": 0.30575791916248446,
      "grad_norm": 0.8430881472269192,
      "learning_rate": 8.078749972572044e-06,
      "loss": 0.4157,
      "step": 5666
    },
    {
      "epoch": 0.30581188279099886,
      "grad_norm": 0.8897413254190745,
      "learning_rate": 8.078124732239004e-06,
      "loss": 0.3522,
      "step": 5667
    },
    {
      "epoch": 0.30586584641951325,
      "grad_norm": 0.9793038600260184,
      "learning_rate": 8.077499417808053e-06,
      "loss": 0.4852,
      "step": 5668
    },
    {
      "epoch": 0.30591981004802765,
      "grad_norm": 1.1763457571185545,
      "learning_rate": 8.076874029297165e-06,
      "loss": 0.6048,
      "step": 5669
    },
    {
      "epoch": 0.305973773676542,
      "grad_norm": 1.0377660944345446,
      "learning_rate": 8.076248566724314e-06,
      "loss": 0.561,
      "step": 5670
    },
    {
      "epoch": 0.3060277373050564,
      "grad_norm": 1.012156394474549,
      "learning_rate": 8.075623030107475e-06,
      "loss": 0.5114,
      "step": 5671
    },
    {
      "epoch": 0.3060817009335708,
      "grad_norm": 1.1404360054140812,
      "learning_rate": 8.074997419464629e-06,
      "loss": 0.5797,
      "step": 5672
    },
    {
      "epoch": 0.30613566456208513,
      "grad_norm": 0.9314242918976164,
      "learning_rate": 8.074371734813754e-06,
      "loss": 0.5464,
      "step": 5673
    },
    {
      "epoch": 0.3061896281905995,
      "grad_norm": 1.0005502443048095,
      "learning_rate": 8.073745976172834e-06,
      "loss": 0.4868,
      "step": 5674
    },
    {
      "epoch": 0.3062435918191139,
      "grad_norm": 1.3320044308351207,
      "learning_rate": 8.073120143559855e-06,
      "loss": 0.546,
      "step": 5675
    },
    {
      "epoch": 0.3062975554476283,
      "grad_norm": 0.9117754759949876,
      "learning_rate": 8.072494236992803e-06,
      "loss": 0.3933,
      "step": 5676
    },
    {
      "epoch": 0.30635151907614266,
      "grad_norm": 1.0396727281547622,
      "learning_rate": 8.071868256489668e-06,
      "loss": 0.4479,
      "step": 5677
    },
    {
      "epoch": 0.30640548270465706,
      "grad_norm": 1.0364374186814083,
      "learning_rate": 8.071242202068439e-06,
      "loss": 0.3505,
      "step": 5678
    },
    {
      "epoch": 0.30645944633317146,
      "grad_norm": 0.9642933101022922,
      "learning_rate": 8.070616073747112e-06,
      "loss": 0.4248,
      "step": 5679
    },
    {
      "epoch": 0.3065134099616858,
      "grad_norm": 1.2408825193579636,
      "learning_rate": 8.069989871543683e-06,
      "loss": 0.6152,
      "step": 5680
    },
    {
      "epoch": 0.3065673735902002,
      "grad_norm": 0.9698010325738592,
      "learning_rate": 8.069363595476146e-06,
      "loss": 0.4464,
      "step": 5681
    },
    {
      "epoch": 0.3066213372187146,
      "grad_norm": 0.9346970151768461,
      "learning_rate": 8.068737245562504e-06,
      "loss": 0.4185,
      "step": 5682
    },
    {
      "epoch": 0.306675300847229,
      "grad_norm": 1.121652419064162,
      "learning_rate": 8.068110821820758e-06,
      "loss": 0.5193,
      "step": 5683
    },
    {
      "epoch": 0.30672926447574334,
      "grad_norm": 0.8499846399476231,
      "learning_rate": 8.067484324268912e-06,
      "loss": 0.375,
      "step": 5684
    },
    {
      "epoch": 0.30678322810425773,
      "grad_norm": 1.0615598606914582,
      "learning_rate": 8.066857752924973e-06,
      "loss": 0.5325,
      "step": 5685
    },
    {
      "epoch": 0.30683719173277213,
      "grad_norm": 0.864465587817257,
      "learning_rate": 8.066231107806947e-06,
      "loss": 0.4728,
      "step": 5686
    },
    {
      "epoch": 0.30689115536128647,
      "grad_norm": 0.9238729385419961,
      "learning_rate": 8.065604388932847e-06,
      "loss": 0.4502,
      "step": 5687
    },
    {
      "epoch": 0.30694511898980087,
      "grad_norm": 0.704653921444429,
      "learning_rate": 8.064977596320684e-06,
      "loss": 0.2941,
      "step": 5688
    },
    {
      "epoch": 0.30699908261831527,
      "grad_norm": 0.7367501687721607,
      "learning_rate": 8.064350729988474e-06,
      "loss": 0.3706,
      "step": 5689
    },
    {
      "epoch": 0.30705304624682966,
      "grad_norm": 0.8970047055803194,
      "learning_rate": 8.063723789954233e-06,
      "loss": 0.5277,
      "step": 5690
    },
    {
      "epoch": 0.307107009875344,
      "grad_norm": 1.2819597049044584,
      "learning_rate": 8.063096776235978e-06,
      "loss": 0.6626,
      "step": 5691
    },
    {
      "epoch": 0.3071609735038584,
      "grad_norm": 1.1492900280504672,
      "learning_rate": 8.062469688851734e-06,
      "loss": 0.6614,
      "step": 5692
    },
    {
      "epoch": 0.3072149371323728,
      "grad_norm": 1.056764534893476,
      "learning_rate": 8.06184252781952e-06,
      "loss": 0.5453,
      "step": 5693
    },
    {
      "epoch": 0.30726890076088714,
      "grad_norm": 0.7326550091123926,
      "learning_rate": 8.061215293157365e-06,
      "loss": 0.2709,
      "step": 5694
    },
    {
      "epoch": 0.30732286438940154,
      "grad_norm": 0.9771661138204013,
      "learning_rate": 8.060587984883294e-06,
      "loss": 0.4315,
      "step": 5695
    },
    {
      "epoch": 0.30737682801791594,
      "grad_norm": 0.9582993041205412,
      "learning_rate": 8.059960603015337e-06,
      "loss": 0.3735,
      "step": 5696
    },
    {
      "epoch": 0.3074307916464303,
      "grad_norm": 0.8858920976016671,
      "learning_rate": 8.059333147571525e-06,
      "loss": 0.4375,
      "step": 5697
    },
    {
      "epoch": 0.3074847552749447,
      "grad_norm": 1.060543980001729,
      "learning_rate": 8.058705618569893e-06,
      "loss": 0.6025,
      "step": 5698
    },
    {
      "epoch": 0.3075387189034591,
      "grad_norm": 1.140007371132824,
      "learning_rate": 8.058078016028475e-06,
      "loss": 0.5694,
      "step": 5699
    },
    {
      "epoch": 0.30759268253197347,
      "grad_norm": 1.1097039745244839,
      "learning_rate": 8.057450339965313e-06,
      "loss": 0.5164,
      "step": 5700
    },
    {
      "epoch": 0.3076466461604878,
      "grad_norm": 1.064130736033157,
      "learning_rate": 8.056822590398441e-06,
      "loss": 0.6087,
      "step": 5701
    },
    {
      "epoch": 0.3077006097890022,
      "grad_norm": 0.8300452785622948,
      "learning_rate": 8.056194767345905e-06,
      "loss": 0.3429,
      "step": 5702
    },
    {
      "epoch": 0.3077545734175166,
      "grad_norm": 1.1591852350902305,
      "learning_rate": 8.05556687082575e-06,
      "loss": 0.5618,
      "step": 5703
    },
    {
      "epoch": 0.30780853704603095,
      "grad_norm": 1.0442871038567847,
      "learning_rate": 8.05493890085602e-06,
      "loss": 0.4645,
      "step": 5704
    },
    {
      "epoch": 0.30786250067454535,
      "grad_norm": 0.9316574993115146,
      "learning_rate": 8.054310857454764e-06,
      "loss": 0.4848,
      "step": 5705
    },
    {
      "epoch": 0.30791646430305974,
      "grad_norm": 1.0489283846846142,
      "learning_rate": 8.053682740640033e-06,
      "loss": 0.4811,
      "step": 5706
    },
    {
      "epoch": 0.30797042793157414,
      "grad_norm": 1.0731036279912236,
      "learning_rate": 8.05305455042988e-06,
      "loss": 0.508,
      "step": 5707
    },
    {
      "epoch": 0.3080243915600885,
      "grad_norm": 0.958058172229073,
      "learning_rate": 8.05242628684236e-06,
      "loss": 0.4211,
      "step": 5708
    },
    {
      "epoch": 0.3080783551886029,
      "grad_norm": 1.2974062314193937,
      "learning_rate": 8.051797949895529e-06,
      "loss": 0.6721,
      "step": 5709
    },
    {
      "epoch": 0.3081323188171173,
      "grad_norm": 1.0019057625137746,
      "learning_rate": 8.051169539607447e-06,
      "loss": 0.3736,
      "step": 5710
    },
    {
      "epoch": 0.3081862824456316,
      "grad_norm": 1.0763100277308877,
      "learning_rate": 8.050541055996173e-06,
      "loss": 0.4442,
      "step": 5711
    },
    {
      "epoch": 0.308240246074146,
      "grad_norm": 0.9031026163772028,
      "learning_rate": 8.049912499079773e-06,
      "loss": 0.4676,
      "step": 5712
    },
    {
      "epoch": 0.3082942097026604,
      "grad_norm": 1.083523042534364,
      "learning_rate": 8.049283868876312e-06,
      "loss": 0.4232,
      "step": 5713
    },
    {
      "epoch": 0.3083481733311748,
      "grad_norm": 1.0477609441170574,
      "learning_rate": 8.048655165403856e-06,
      "loss": 0.4302,
      "step": 5714
    },
    {
      "epoch": 0.30840213695968915,
      "grad_norm": 1.0982322460224874,
      "learning_rate": 8.048026388680476e-06,
      "loss": 0.4446,
      "step": 5715
    },
    {
      "epoch": 0.30845610058820355,
      "grad_norm": 1.1789833701804686,
      "learning_rate": 8.04739753872424e-06,
      "loss": 0.56,
      "step": 5716
    },
    {
      "epoch": 0.30851006421671795,
      "grad_norm": 0.8298565856289954,
      "learning_rate": 8.046768615553229e-06,
      "loss": 0.4233,
      "step": 5717
    },
    {
      "epoch": 0.3085640278452323,
      "grad_norm": 0.99642188653052,
      "learning_rate": 8.046139619185514e-06,
      "loss": 0.3782,
      "step": 5718
    },
    {
      "epoch": 0.3086179914737467,
      "grad_norm": 0.9188234053634301,
      "learning_rate": 8.045510549639173e-06,
      "loss": 0.5027,
      "step": 5719
    },
    {
      "epoch": 0.3086719551022611,
      "grad_norm": 0.9858820941218008,
      "learning_rate": 8.044881406932285e-06,
      "loss": 0.5159,
      "step": 5720
    },
    {
      "epoch": 0.3087259187307755,
      "grad_norm": 1.1112577772595984,
      "learning_rate": 8.044252191082936e-06,
      "loss": 0.4953,
      "step": 5721
    },
    {
      "epoch": 0.3087798823592898,
      "grad_norm": 1.1428043491063755,
      "learning_rate": 8.043622902109207e-06,
      "loss": 0.5777,
      "step": 5722
    },
    {
      "epoch": 0.3088338459878042,
      "grad_norm": 0.9696183711473637,
      "learning_rate": 8.042993540029186e-06,
      "loss": 0.4286,
      "step": 5723
    },
    {
      "epoch": 0.3088878096163186,
      "grad_norm": 0.777400507112986,
      "learning_rate": 8.042364104860962e-06,
      "loss": 0.3256,
      "step": 5724
    },
    {
      "epoch": 0.30894177324483296,
      "grad_norm": 1.0089457247572042,
      "learning_rate": 8.041734596622623e-06,
      "loss": 0.4298,
      "step": 5725
    },
    {
      "epoch": 0.30899573687334736,
      "grad_norm": 0.987856301123255,
      "learning_rate": 8.041105015332264e-06,
      "loss": 0.3837,
      "step": 5726
    },
    {
      "epoch": 0.30904970050186176,
      "grad_norm": 1.1344529722271766,
      "learning_rate": 8.040475361007982e-06,
      "loss": 0.4597,
      "step": 5727
    },
    {
      "epoch": 0.3091036641303761,
      "grad_norm": 1.1778670888542118,
      "learning_rate": 8.039845633667867e-06,
      "loss": 0.5681,
      "step": 5728
    },
    {
      "epoch": 0.3091576277588905,
      "grad_norm": 1.0456648659823868,
      "learning_rate": 8.039215833330026e-06,
      "loss": 0.5533,
      "step": 5729
    },
    {
      "epoch": 0.3092115913874049,
      "grad_norm": 1.0425636848769562,
      "learning_rate": 8.038585960012553e-06,
      "loss": 0.4264,
      "step": 5730
    },
    {
      "epoch": 0.3092655550159193,
      "grad_norm": 0.9565466919863763,
      "learning_rate": 8.037956013733556e-06,
      "loss": 0.5255,
      "step": 5731
    },
    {
      "epoch": 0.30931951864443363,
      "grad_norm": 1.0693905086280673,
      "learning_rate": 8.03732599451114e-06,
      "loss": 0.4671,
      "step": 5732
    },
    {
      "epoch": 0.30937348227294803,
      "grad_norm": 1.0379982879624805,
      "learning_rate": 8.036695902363408e-06,
      "loss": 0.4308,
      "step": 5733
    },
    {
      "epoch": 0.30942744590146243,
      "grad_norm": 0.8924010759329344,
      "learning_rate": 8.036065737308476e-06,
      "loss": 0.3995,
      "step": 5734
    },
    {
      "epoch": 0.30948140952997677,
      "grad_norm": 0.9129497072366776,
      "learning_rate": 8.03543549936445e-06,
      "loss": 0.4324,
      "step": 5735
    },
    {
      "epoch": 0.30953537315849117,
      "grad_norm": 0.9611986766029539,
      "learning_rate": 8.034805188549446e-06,
      "loss": 0.5475,
      "step": 5736
    },
    {
      "epoch": 0.30958933678700556,
      "grad_norm": 0.8381527660725501,
      "learning_rate": 8.03417480488158e-06,
      "loss": 0.3226,
      "step": 5737
    },
    {
      "epoch": 0.30964330041551996,
      "grad_norm": 1.1066767969846116,
      "learning_rate": 8.03354434837897e-06,
      "loss": 0.5505,
      "step": 5738
    },
    {
      "epoch": 0.3096972640440343,
      "grad_norm": 0.9499330874896346,
      "learning_rate": 8.032913819059734e-06,
      "loss": 0.4768,
      "step": 5739
    },
    {
      "epoch": 0.3097512276725487,
      "grad_norm": 0.8730288590704799,
      "learning_rate": 8.032283216941997e-06,
      "loss": 0.4084,
      "step": 5740
    },
    {
      "epoch": 0.3098051913010631,
      "grad_norm": 0.952662245864132,
      "learning_rate": 8.031652542043881e-06,
      "loss": 0.3509,
      "step": 5741
    },
    {
      "epoch": 0.30985915492957744,
      "grad_norm": 1.0508404580333337,
      "learning_rate": 8.031021794383513e-06,
      "loss": 0.4786,
      "step": 5742
    },
    {
      "epoch": 0.30991311855809184,
      "grad_norm": 1.0277263668851035,
      "learning_rate": 8.030390973979021e-06,
      "loss": 0.4977,
      "step": 5743
    },
    {
      "epoch": 0.30996708218660624,
      "grad_norm": 0.9290790689098543,
      "learning_rate": 8.029760080848537e-06,
      "loss": 0.462,
      "step": 5744
    },
    {
      "epoch": 0.31002104581512063,
      "grad_norm": 0.9266918610806855,
      "learning_rate": 8.029129115010191e-06,
      "loss": 0.3753,
      "step": 5745
    },
    {
      "epoch": 0.310075009443635,
      "grad_norm": 0.9818755824028417,
      "learning_rate": 8.028498076482119e-06,
      "loss": 0.6027,
      "step": 5746
    },
    {
      "epoch": 0.31012897307214937,
      "grad_norm": 0.9790688257925357,
      "learning_rate": 8.027866965282457e-06,
      "loss": 0.5445,
      "step": 5747
    },
    {
      "epoch": 0.31018293670066377,
      "grad_norm": 0.921062702338916,
      "learning_rate": 8.027235781429345e-06,
      "loss": 0.4154,
      "step": 5748
    },
    {
      "epoch": 0.3102369003291781,
      "grad_norm": 0.94225339249066,
      "learning_rate": 8.026604524940922e-06,
      "loss": 0.487,
      "step": 5749
    },
    {
      "epoch": 0.3102908639576925,
      "grad_norm": 0.9280899582904373,
      "learning_rate": 8.025973195835332e-06,
      "loss": 0.5227,
      "step": 5750
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 1.2554964797386488,
      "learning_rate": 8.025341794130722e-06,
      "loss": 0.5629,
      "step": 5751
    },
    {
      "epoch": 0.3103987912147213,
      "grad_norm": 0.914519474507173,
      "learning_rate": 8.024710319845233e-06,
      "loss": 0.5197,
      "step": 5752
    },
    {
      "epoch": 0.31045275484323565,
      "grad_norm": 1.1169964808535187,
      "learning_rate": 8.024078772997024e-06,
      "loss": 0.4897,
      "step": 5753
    },
    {
      "epoch": 0.31050671847175004,
      "grad_norm": 1.0009252970778322,
      "learning_rate": 8.023447153604236e-06,
      "loss": 0.5467,
      "step": 5754
    },
    {
      "epoch": 0.31056068210026444,
      "grad_norm": 0.7320710509581911,
      "learning_rate": 8.02281546168503e-06,
      "loss": 0.3095,
      "step": 5755
    },
    {
      "epoch": 0.3106146457287788,
      "grad_norm": 1.1227087692522757,
      "learning_rate": 8.022183697257557e-06,
      "loss": 0.5622,
      "step": 5756
    },
    {
      "epoch": 0.3106686093572932,
      "grad_norm": 0.9435790618574552,
      "learning_rate": 8.021551860339976e-06,
      "loss": 0.4308,
      "step": 5757
    },
    {
      "epoch": 0.3107225729858076,
      "grad_norm": 1.1943472855606505,
      "learning_rate": 8.020919950950448e-06,
      "loss": 0.5189,
      "step": 5758
    },
    {
      "epoch": 0.310776536614322,
      "grad_norm": 1.1136469037947343,
      "learning_rate": 8.020287969107132e-06,
      "loss": 0.5543,
      "step": 5759
    },
    {
      "epoch": 0.3108305002428363,
      "grad_norm": 1.0289136561858603,
      "learning_rate": 8.019655914828195e-06,
      "loss": 0.4599,
      "step": 5760
    },
    {
      "epoch": 0.3108844638713507,
      "grad_norm": 0.9797790273514165,
      "learning_rate": 8.0190237881318e-06,
      "loss": 0.4706,
      "step": 5761
    },
    {
      "epoch": 0.3109384274998651,
      "grad_norm": 0.7853063307729679,
      "learning_rate": 8.018391589036115e-06,
      "loss": 0.3983,
      "step": 5762
    },
    {
      "epoch": 0.31099239112837945,
      "grad_norm": 1.1835599872882585,
      "learning_rate": 8.017759317559312e-06,
      "loss": 0.5284,
      "step": 5763
    },
    {
      "epoch": 0.31104635475689385,
      "grad_norm": 1.0717012281299545,
      "learning_rate": 8.017126973719561e-06,
      "loss": 0.5832,
      "step": 5764
    },
    {
      "epoch": 0.31110031838540825,
      "grad_norm": 0.915749955156974,
      "learning_rate": 8.01649455753504e-06,
      "loss": 0.3854,
      "step": 5765
    },
    {
      "epoch": 0.3111542820139226,
      "grad_norm": 0.8535666670822716,
      "learning_rate": 8.01586206902392e-06,
      "loss": 0.4165,
      "step": 5766
    },
    {
      "epoch": 0.311208245642437,
      "grad_norm": 0.9842008297824502,
      "learning_rate": 8.015229508204383e-06,
      "loss": 0.4358,
      "step": 5767
    },
    {
      "epoch": 0.3112622092709514,
      "grad_norm": 1.076046957118243,
      "learning_rate": 8.014596875094609e-06,
      "loss": 0.507,
      "step": 5768
    },
    {
      "epoch": 0.3113161728994658,
      "grad_norm": 0.9890596293772697,
      "learning_rate": 8.01396416971278e-06,
      "loss": 0.5824,
      "step": 5769
    },
    {
      "epoch": 0.3113701365279801,
      "grad_norm": 1.2345176238394864,
      "learning_rate": 8.01333139207708e-06,
      "loss": 0.5423,
      "step": 5770
    },
    {
      "epoch": 0.3114241001564945,
      "grad_norm": 1.1293622033199444,
      "learning_rate": 8.012698542205698e-06,
      "loss": 0.5652,
      "step": 5771
    },
    {
      "epoch": 0.3114780637850089,
      "grad_norm": 1.0954738166785163,
      "learning_rate": 8.012065620116819e-06,
      "loss": 0.538,
      "step": 5772
    },
    {
      "epoch": 0.31153202741352326,
      "grad_norm": 0.7255140756272943,
      "learning_rate": 8.011432625828636e-06,
      "loss": 0.2899,
      "step": 5773
    },
    {
      "epoch": 0.31158599104203766,
      "grad_norm": 0.8835809701839762,
      "learning_rate": 8.010799559359342e-06,
      "loss": 0.4119,
      "step": 5774
    },
    {
      "epoch": 0.31163995467055206,
      "grad_norm": 0.9444065081302631,
      "learning_rate": 8.010166420727132e-06,
      "loss": 0.4164,
      "step": 5775
    },
    {
      "epoch": 0.31169391829906645,
      "grad_norm": 1.251770748116836,
      "learning_rate": 8.009533209950201e-06,
      "loss": 0.7404,
      "step": 5776
    },
    {
      "epoch": 0.3117478819275808,
      "grad_norm": 1.0206917175047399,
      "learning_rate": 8.008899927046752e-06,
      "loss": 0.4548,
      "step": 5777
    },
    {
      "epoch": 0.3118018455560952,
      "grad_norm": 0.6878741713820212,
      "learning_rate": 8.008266572034984e-06,
      "loss": 0.2496,
      "step": 5778
    },
    {
      "epoch": 0.3118558091846096,
      "grad_norm": 1.025049048254685,
      "learning_rate": 8.0076331449331e-06,
      "loss": 0.5123,
      "step": 5779
    },
    {
      "epoch": 0.31190977281312393,
      "grad_norm": 1.2347249780374974,
      "learning_rate": 8.006999645759305e-06,
      "loss": 0.6378,
      "step": 5780
    },
    {
      "epoch": 0.31196373644163833,
      "grad_norm": 0.9175210376702057,
      "learning_rate": 8.006366074531808e-06,
      "loss": 0.5526,
      "step": 5781
    },
    {
      "epoch": 0.3120177000701527,
      "grad_norm": 1.0859465031888946,
      "learning_rate": 8.005732431268817e-06,
      "loss": 0.604,
      "step": 5782
    },
    {
      "epoch": 0.3120716636986671,
      "grad_norm": 1.2249113840135557,
      "learning_rate": 8.005098715988543e-06,
      "loss": 0.477,
      "step": 5783
    },
    {
      "epoch": 0.31212562732718147,
      "grad_norm": 1.0952431674957677,
      "learning_rate": 8.004464928709203e-06,
      "loss": 0.4421,
      "step": 5784
    },
    {
      "epoch": 0.31217959095569586,
      "grad_norm": 0.9588434052193681,
      "learning_rate": 8.003831069449008e-06,
      "loss": 0.3682,
      "step": 5785
    },
    {
      "epoch": 0.31223355458421026,
      "grad_norm": 1.2124834730959386,
      "learning_rate": 8.003197138226178e-06,
      "loss": 0.5892,
      "step": 5786
    },
    {
      "epoch": 0.3122875182127246,
      "grad_norm": 1.0599732871955534,
      "learning_rate": 8.002563135058935e-06,
      "loss": 0.5058,
      "step": 5787
    },
    {
      "epoch": 0.312341481841239,
      "grad_norm": 1.1721815324702174,
      "learning_rate": 8.001929059965499e-06,
      "loss": 0.6736,
      "step": 5788
    },
    {
      "epoch": 0.3123954454697534,
      "grad_norm": 0.9858698274947841,
      "learning_rate": 8.001294912964092e-06,
      "loss": 0.4858,
      "step": 5789
    },
    {
      "epoch": 0.3124494090982678,
      "grad_norm": 1.0448021397704002,
      "learning_rate": 8.000660694072944e-06,
      "loss": 0.6486,
      "step": 5790
    },
    {
      "epoch": 0.31250337272678214,
      "grad_norm": 0.929609520581377,
      "learning_rate": 8.000026403310278e-06,
      "loss": 0.4042,
      "step": 5791
    },
    {
      "epoch": 0.31255733635529653,
      "grad_norm": 0.9196163368492284,
      "learning_rate": 7.999392040694329e-06,
      "loss": 0.3877,
      "step": 5792
    },
    {
      "epoch": 0.31261129998381093,
      "grad_norm": 0.9868209275203024,
      "learning_rate": 7.998757606243326e-06,
      "loss": 0.4454,
      "step": 5793
    },
    {
      "epoch": 0.3126652636123253,
      "grad_norm": 1.0883217893689185,
      "learning_rate": 7.998123099975507e-06,
      "loss": 0.6294,
      "step": 5794
    },
    {
      "epoch": 0.31271922724083967,
      "grad_norm": 0.9490428151289455,
      "learning_rate": 7.997488521909103e-06,
      "loss": 0.5898,
      "step": 5795
    },
    {
      "epoch": 0.31277319086935407,
      "grad_norm": 1.1350095340638253,
      "learning_rate": 7.996853872062359e-06,
      "loss": 0.5844,
      "step": 5796
    },
    {
      "epoch": 0.3128271544978684,
      "grad_norm": 0.8904157067709089,
      "learning_rate": 7.996219150453509e-06,
      "loss": 0.4268,
      "step": 5797
    },
    {
      "epoch": 0.3128811181263828,
      "grad_norm": 0.906666002328019,
      "learning_rate": 7.995584357100798e-06,
      "loss": 0.4028,
      "step": 5798
    },
    {
      "epoch": 0.3129350817548972,
      "grad_norm": 1.0366802775180255,
      "learning_rate": 7.994949492022472e-06,
      "loss": 0.4525,
      "step": 5799
    },
    {
      "epoch": 0.3129890453834116,
      "grad_norm": 0.9826857575165051,
      "learning_rate": 7.994314555236776e-06,
      "loss": 0.4226,
      "step": 5800
    },
    {
      "epoch": 0.31304300901192594,
      "grad_norm": 0.7873717905009159,
      "learning_rate": 7.993679546761962e-06,
      "loss": 0.3608,
      "step": 5801
    },
    {
      "epoch": 0.31309697264044034,
      "grad_norm": 0.8321285685663172,
      "learning_rate": 7.993044466616275e-06,
      "loss": 0.3378,
      "step": 5802
    },
    {
      "epoch": 0.31315093626895474,
      "grad_norm": 0.8949440503810001,
      "learning_rate": 7.99240931481797e-06,
      "loss": 0.3047,
      "step": 5803
    },
    {
      "epoch": 0.3132048998974691,
      "grad_norm": 1.257705208503103,
      "learning_rate": 7.991774091385306e-06,
      "loss": 0.6254,
      "step": 5804
    },
    {
      "epoch": 0.3132588635259835,
      "grad_norm": 1.399848691701927,
      "learning_rate": 7.991138796336537e-06,
      "loss": 0.5634,
      "step": 5805
    },
    {
      "epoch": 0.3133128271544979,
      "grad_norm": 1.0304363885464278,
      "learning_rate": 7.99050342968992e-06,
      "loss": 0.4374,
      "step": 5806
    },
    {
      "epoch": 0.3133667907830123,
      "grad_norm": 1.1844254310899136,
      "learning_rate": 7.989867991463717e-06,
      "loss": 0.6405,
      "step": 5807
    },
    {
      "epoch": 0.3134207544115266,
      "grad_norm": 1.000691911414819,
      "learning_rate": 7.989232481676196e-06,
      "loss": 0.429,
      "step": 5808
    },
    {
      "epoch": 0.313474718040041,
      "grad_norm": 1.2173707285541955,
      "learning_rate": 7.988596900345615e-06,
      "loss": 0.552,
      "step": 5809
    },
    {
      "epoch": 0.3135286816685554,
      "grad_norm": 1.0927331513298872,
      "learning_rate": 7.987961247490246e-06,
      "loss": 0.5113,
      "step": 5810
    },
    {
      "epoch": 0.31358264529706975,
      "grad_norm": 0.9120528394989482,
      "learning_rate": 7.987325523128357e-06,
      "loss": 0.4621,
      "step": 5811
    },
    {
      "epoch": 0.31363660892558415,
      "grad_norm": 1.0221986062073876,
      "learning_rate": 7.986689727278219e-06,
      "loss": 0.4781,
      "step": 5812
    },
    {
      "epoch": 0.31369057255409855,
      "grad_norm": 1.1210470670955046,
      "learning_rate": 7.986053859958104e-06,
      "loss": 0.652,
      "step": 5813
    },
    {
      "epoch": 0.31374453618261294,
      "grad_norm": 0.9254124183949032,
      "learning_rate": 7.985417921186292e-06,
      "loss": 0.4828,
      "step": 5814
    },
    {
      "epoch": 0.3137984998111273,
      "grad_norm": 1.044871074608017,
      "learning_rate": 7.984781910981055e-06,
      "loss": 0.5252,
      "step": 5815
    },
    {
      "epoch": 0.3138524634396417,
      "grad_norm": 1.1903180790776156,
      "learning_rate": 7.984145829360676e-06,
      "loss": 0.5789,
      "step": 5816
    },
    {
      "epoch": 0.3139064270681561,
      "grad_norm": 1.1783099741528913,
      "learning_rate": 7.983509676343435e-06,
      "loss": 0.4649,
      "step": 5817
    },
    {
      "epoch": 0.3139603906966704,
      "grad_norm": 1.1873672591705628,
      "learning_rate": 7.98287345194762e-06,
      "loss": 0.4212,
      "step": 5818
    },
    {
      "epoch": 0.3140143543251848,
      "grad_norm": 1.0455420534572923,
      "learning_rate": 7.98223715619151e-06,
      "loss": 0.5026,
      "step": 5819
    },
    {
      "epoch": 0.3140683179536992,
      "grad_norm": 0.874271130430988,
      "learning_rate": 7.981600789093395e-06,
      "loss": 0.3778,
      "step": 5820
    },
    {
      "epoch": 0.3141222815822136,
      "grad_norm": 0.9671985515972982,
      "learning_rate": 7.98096435067157e-06,
      "loss": 0.4618,
      "step": 5821
    },
    {
      "epoch": 0.31417624521072796,
      "grad_norm": 1.0907235402612039,
      "learning_rate": 7.980327840944318e-06,
      "loss": 0.5757,
      "step": 5822
    },
    {
      "epoch": 0.31423020883924235,
      "grad_norm": 0.9567141576071382,
      "learning_rate": 7.979691259929942e-06,
      "loss": 0.4432,
      "step": 5823
    },
    {
      "epoch": 0.31428417246775675,
      "grad_norm": 0.973129899444177,
      "learning_rate": 7.97905460764673e-06,
      "loss": 0.4555,
      "step": 5824
    },
    {
      "epoch": 0.3143381360962711,
      "grad_norm": 1.265399572361619,
      "learning_rate": 7.978417884112986e-06,
      "loss": 0.5079,
      "step": 5825
    },
    {
      "epoch": 0.3143920997247855,
      "grad_norm": 1.0640899919627675,
      "learning_rate": 7.977781089347007e-06,
      "loss": 0.4992,
      "step": 5826
    },
    {
      "epoch": 0.3144460633532999,
      "grad_norm": 0.9683520723134889,
      "learning_rate": 7.977144223367097e-06,
      "loss": 0.387,
      "step": 5827
    },
    {
      "epoch": 0.3145000269818143,
      "grad_norm": 0.930444249927146,
      "learning_rate": 7.976507286191557e-06,
      "loss": 0.4317,
      "step": 5828
    },
    {
      "epoch": 0.3145539906103286,
      "grad_norm": 0.9596739957461282,
      "learning_rate": 7.975870277838695e-06,
      "loss": 0.5599,
      "step": 5829
    },
    {
      "epoch": 0.314607954238843,
      "grad_norm": 1.0553082050145277,
      "learning_rate": 7.97523319832682e-06,
      "loss": 0.4343,
      "step": 5830
    },
    {
      "epoch": 0.3146619178673574,
      "grad_norm": 0.835048279175396,
      "learning_rate": 7.974596047674244e-06,
      "loss": 0.3504,
      "step": 5831
    },
    {
      "epoch": 0.31471588149587176,
      "grad_norm": 1.120678880375561,
      "learning_rate": 7.973958825899276e-06,
      "loss": 0.6657,
      "step": 5832
    },
    {
      "epoch": 0.31476984512438616,
      "grad_norm": 0.9853529486491105,
      "learning_rate": 7.973321533020232e-06,
      "loss": 0.5031,
      "step": 5833
    },
    {
      "epoch": 0.31482380875290056,
      "grad_norm": 0.8893062074578372,
      "learning_rate": 7.972684169055427e-06,
      "loss": 0.369,
      "step": 5834
    },
    {
      "epoch": 0.3148777723814149,
      "grad_norm": 1.0920588924263088,
      "learning_rate": 7.97204673402318e-06,
      "loss": 0.4106,
      "step": 5835
    },
    {
      "epoch": 0.3149317360099293,
      "grad_norm": 1.1086154064300722,
      "learning_rate": 7.971409227941814e-06,
      "loss": 0.4241,
      "step": 5836
    },
    {
      "epoch": 0.3149856996384437,
      "grad_norm": 0.9581381485147547,
      "learning_rate": 7.97077165082965e-06,
      "loss": 0.4289,
      "step": 5837
    },
    {
      "epoch": 0.3150396632669581,
      "grad_norm": 1.22657202320478,
      "learning_rate": 7.97013400270501e-06,
      "loss": 0.5311,
      "step": 5838
    },
    {
      "epoch": 0.31509362689547243,
      "grad_norm": 0.9912941421355445,
      "learning_rate": 7.969496283586224e-06,
      "loss": 0.4534,
      "step": 5839
    },
    {
      "epoch": 0.31514759052398683,
      "grad_norm": 1.15814048669185,
      "learning_rate": 7.968858493491618e-06,
      "loss": 0.5488,
      "step": 5840
    },
    {
      "epoch": 0.31520155415250123,
      "grad_norm": 1.1456935878843448,
      "learning_rate": 7.968220632439524e-06,
      "loss": 0.4011,
      "step": 5841
    },
    {
      "epoch": 0.31525551778101557,
      "grad_norm": 1.2962206486990293,
      "learning_rate": 7.967582700448279e-06,
      "loss": 0.5436,
      "step": 5842
    },
    {
      "epoch": 0.31530948140952997,
      "grad_norm": 0.8289831987761082,
      "learning_rate": 7.966944697536211e-06,
      "loss": 0.3543,
      "step": 5843
    },
    {
      "epoch": 0.31536344503804437,
      "grad_norm": 0.8939314937567485,
      "learning_rate": 7.96630662372166e-06,
      "loss": 0.3883,
      "step": 5844
    },
    {
      "epoch": 0.31541740866655876,
      "grad_norm": 0.9971532050688644,
      "learning_rate": 7.965668479022964e-06,
      "loss": 0.4185,
      "step": 5845
    },
    {
      "epoch": 0.3154713722950731,
      "grad_norm": 1.0901568973577713,
      "learning_rate": 7.965030263458466e-06,
      "loss": 0.4931,
      "step": 5846
    },
    {
      "epoch": 0.3155253359235875,
      "grad_norm": 1.259989299005883,
      "learning_rate": 7.964391977046508e-06,
      "loss": 0.523,
      "step": 5847
    },
    {
      "epoch": 0.3155792995521019,
      "grad_norm": 1.0859776653563968,
      "learning_rate": 7.96375361980543e-06,
      "loss": 0.5298,
      "step": 5848
    },
    {
      "epoch": 0.31563326318061624,
      "grad_norm": 1.1906997893829152,
      "learning_rate": 7.96311519175359e-06,
      "loss": 0.6941,
      "step": 5849
    },
    {
      "epoch": 0.31568722680913064,
      "grad_norm": 0.9914111111225924,
      "learning_rate": 7.962476692909327e-06,
      "loss": 0.5361,
      "step": 5850
    },
    {
      "epoch": 0.31574119043764504,
      "grad_norm": 1.1997176849686149,
      "learning_rate": 7.961838123290997e-06,
      "loss": 0.5156,
      "step": 5851
    },
    {
      "epoch": 0.31579515406615943,
      "grad_norm": 1.0835697909731707,
      "learning_rate": 7.96119948291695e-06,
      "loss": 0.4638,
      "step": 5852
    },
    {
      "epoch": 0.3158491176946738,
      "grad_norm": 1.298205417606924,
      "learning_rate": 7.960560771805544e-06,
      "loss": 0.5146,
      "step": 5853
    },
    {
      "epoch": 0.3159030813231882,
      "grad_norm": 0.9745858262658678,
      "learning_rate": 7.959921989975136e-06,
      "loss": 0.4748,
      "step": 5854
    },
    {
      "epoch": 0.31595704495170257,
      "grad_norm": 1.1012545910573555,
      "learning_rate": 7.959283137444083e-06,
      "loss": 0.5379,
      "step": 5855
    },
    {
      "epoch": 0.3160110085802169,
      "grad_norm": 0.9030981716785355,
      "learning_rate": 7.958644214230749e-06,
      "loss": 0.4077,
      "step": 5856
    },
    {
      "epoch": 0.3160649722087313,
      "grad_norm": 0.9730020436697281,
      "learning_rate": 7.958005220353495e-06,
      "loss": 0.4376,
      "step": 5857
    },
    {
      "epoch": 0.3161189358372457,
      "grad_norm": 1.055955379773911,
      "learning_rate": 7.957366155830687e-06,
      "loss": 0.5935,
      "step": 5858
    },
    {
      "epoch": 0.3161728994657601,
      "grad_norm": 1.0181651174423476,
      "learning_rate": 7.956727020680692e-06,
      "loss": 0.6437,
      "step": 5859
    },
    {
      "epoch": 0.31622686309427445,
      "grad_norm": 0.9700126992750109,
      "learning_rate": 7.956087814921884e-06,
      "loss": 0.4049,
      "step": 5860
    },
    {
      "epoch": 0.31628082672278884,
      "grad_norm": 1.0385669158530928,
      "learning_rate": 7.955448538572626e-06,
      "loss": 0.5072,
      "step": 5861
    },
    {
      "epoch": 0.31633479035130324,
      "grad_norm": 1.2194739998610478,
      "learning_rate": 7.954809191651298e-06,
      "loss": 0.8048,
      "step": 5862
    },
    {
      "epoch": 0.3163887539798176,
      "grad_norm": 1.0953804244036105,
      "learning_rate": 7.954169774176275e-06,
      "loss": 0.4738,
      "step": 5863
    },
    {
      "epoch": 0.316442717608332,
      "grad_norm": 0.9178850994733903,
      "learning_rate": 7.95353028616593e-06,
      "loss": 0.4045,
      "step": 5864
    },
    {
      "epoch": 0.3164966812368464,
      "grad_norm": 1.073957676644468,
      "learning_rate": 7.952890727638646e-06,
      "loss": 0.4289,
      "step": 5865
    },
    {
      "epoch": 0.3165506448653607,
      "grad_norm": 1.1100709617649702,
      "learning_rate": 7.952251098612807e-06,
      "loss": 0.4919,
      "step": 5866
    },
    {
      "epoch": 0.3166046084938751,
      "grad_norm": 0.9488450181157873,
      "learning_rate": 7.951611399106792e-06,
      "loss": 0.3607,
      "step": 5867
    },
    {
      "epoch": 0.3166585721223895,
      "grad_norm": 1.0707686511755923,
      "learning_rate": 7.950971629138986e-06,
      "loss": 0.6324,
      "step": 5868
    },
    {
      "epoch": 0.3167125357509039,
      "grad_norm": 1.3027139485558845,
      "learning_rate": 7.95033178872778e-06,
      "loss": 0.7684,
      "step": 5869
    },
    {
      "epoch": 0.31676649937941825,
      "grad_norm": 1.1559163865572664,
      "learning_rate": 7.949691877891564e-06,
      "loss": 0.6084,
      "step": 5870
    },
    {
      "epoch": 0.31682046300793265,
      "grad_norm": 1.154749064820439,
      "learning_rate": 7.949051896648726e-06,
      "loss": 0.5162,
      "step": 5871
    },
    {
      "epoch": 0.31687442663644705,
      "grad_norm": 0.8885038138718508,
      "learning_rate": 7.948411845017665e-06,
      "loss": 0.3846,
      "step": 5872
    },
    {
      "epoch": 0.3169283902649614,
      "grad_norm": 1.277908554544182,
      "learning_rate": 7.94777172301677e-06,
      "loss": 0.5957,
      "step": 5873
    },
    {
      "epoch": 0.3169823538934758,
      "grad_norm": 0.9126303217641203,
      "learning_rate": 7.947131530664444e-06,
      "loss": 0.4912,
      "step": 5874
    },
    {
      "epoch": 0.3170363175219902,
      "grad_norm": 0.9861993525521239,
      "learning_rate": 7.946491267979084e-06,
      "loss": 0.5479,
      "step": 5875
    },
    {
      "epoch": 0.3170902811505046,
      "grad_norm": 1.1387184460532889,
      "learning_rate": 7.945850934979095e-06,
      "loss": 0.5,
      "step": 5876
    },
    {
      "epoch": 0.3171442447790189,
      "grad_norm": 0.8190446527054122,
      "learning_rate": 7.945210531682878e-06,
      "loss": 0.3435,
      "step": 5877
    },
    {
      "epoch": 0.3171982084075333,
      "grad_norm": 0.9781052880177178,
      "learning_rate": 7.94457005810884e-06,
      "loss": 0.4093,
      "step": 5878
    },
    {
      "epoch": 0.3172521720360477,
      "grad_norm": 0.8602296661362321,
      "learning_rate": 7.943929514275388e-06,
      "loss": 0.3782,
      "step": 5879
    },
    {
      "epoch": 0.31730613566456206,
      "grad_norm": 0.9073202524499124,
      "learning_rate": 7.943288900200932e-06,
      "loss": 0.3531,
      "step": 5880
    },
    {
      "epoch": 0.31736009929307646,
      "grad_norm": 0.8346163380267516,
      "learning_rate": 7.942648215903886e-06,
      "loss": 0.3661,
      "step": 5881
    },
    {
      "epoch": 0.31741406292159086,
      "grad_norm": 1.009301338151871,
      "learning_rate": 7.94200746140266e-06,
      "loss": 0.454,
      "step": 5882
    },
    {
      "epoch": 0.31746802655010525,
      "grad_norm": 0.777601415227095,
      "learning_rate": 7.941366636715673e-06,
      "loss": 0.2918,
      "step": 5883
    },
    {
      "epoch": 0.3175219901786196,
      "grad_norm": 0.9679503802474361,
      "learning_rate": 7.940725741861342e-06,
      "loss": 0.363,
      "step": 5884
    },
    {
      "epoch": 0.317575953807134,
      "grad_norm": 1.024693736225454,
      "learning_rate": 7.940084776858089e-06,
      "loss": 0.4917,
      "step": 5885
    },
    {
      "epoch": 0.3176299174356484,
      "grad_norm": 1.1321107069169487,
      "learning_rate": 7.939443741724329e-06,
      "loss": 0.4444,
      "step": 5886
    },
    {
      "epoch": 0.31768388106416273,
      "grad_norm": 1.0918332090350371,
      "learning_rate": 7.938802636478496e-06,
      "loss": 0.3264,
      "step": 5887
    },
    {
      "epoch": 0.31773784469267713,
      "grad_norm": 1.0715956475824344,
      "learning_rate": 7.938161461139009e-06,
      "loss": 0.4277,
      "step": 5888
    },
    {
      "epoch": 0.3177918083211915,
      "grad_norm": 1.1311215407989392,
      "learning_rate": 7.937520215724297e-06,
      "loss": 0.6258,
      "step": 5889
    },
    {
      "epoch": 0.3178457719497059,
      "grad_norm": 1.1986855642251864,
      "learning_rate": 7.936878900252793e-06,
      "loss": 0.627,
      "step": 5890
    },
    {
      "epoch": 0.31789973557822027,
      "grad_norm": 1.0196362554573277,
      "learning_rate": 7.936237514742927e-06,
      "loss": 0.5161,
      "step": 5891
    },
    {
      "epoch": 0.31795369920673466,
      "grad_norm": 1.074218132442038,
      "learning_rate": 7.93559605921313e-06,
      "loss": 0.5126,
      "step": 5892
    },
    {
      "epoch": 0.31800766283524906,
      "grad_norm": 0.9410662748087897,
      "learning_rate": 7.934954533681843e-06,
      "loss": 0.4155,
      "step": 5893
    },
    {
      "epoch": 0.3180616264637634,
      "grad_norm": 0.9072156683166208,
      "learning_rate": 7.934312938167505e-06,
      "loss": 0.4293,
      "step": 5894
    },
    {
      "epoch": 0.3181155900922778,
      "grad_norm": 0.9187281987882211,
      "learning_rate": 7.93367127268855e-06,
      "loss": 0.4486,
      "step": 5895
    },
    {
      "epoch": 0.3181695537207922,
      "grad_norm": 1.1444334076752136,
      "learning_rate": 7.933029537263426e-06,
      "loss": 0.5998,
      "step": 5896
    },
    {
      "epoch": 0.3182235173493066,
      "grad_norm": 0.9043559026228211,
      "learning_rate": 7.932387731910573e-06,
      "loss": 0.3689,
      "step": 5897
    },
    {
      "epoch": 0.31827748097782094,
      "grad_norm": 0.9624484762918142,
      "learning_rate": 7.93174585664844e-06,
      "loss": 0.4615,
      "step": 5898
    },
    {
      "epoch": 0.31833144460633533,
      "grad_norm": 1.126440355890838,
      "learning_rate": 7.931103911495474e-06,
      "loss": 0.5642,
      "step": 5899
    },
    {
      "epoch": 0.31838540823484973,
      "grad_norm": 1.272228930320947,
      "learning_rate": 7.930461896470125e-06,
      "loss": 0.5955,
      "step": 5900
    },
    {
      "epoch": 0.3184393718633641,
      "grad_norm": 0.8470531449091266,
      "learning_rate": 7.929819811590844e-06,
      "loss": 0.3538,
      "step": 5901
    },
    {
      "epoch": 0.31849333549187847,
      "grad_norm": 0.8624163210271384,
      "learning_rate": 7.929177656876088e-06,
      "loss": 0.3359,
      "step": 5902
    },
    {
      "epoch": 0.31854729912039287,
      "grad_norm": 0.9890881209901963,
      "learning_rate": 7.928535432344311e-06,
      "loss": 0.4398,
      "step": 5903
    },
    {
      "epoch": 0.3186012627489072,
      "grad_norm": 0.9508911334110701,
      "learning_rate": 7.927893138013971e-06,
      "loss": 0.3783,
      "step": 5904
    },
    {
      "epoch": 0.3186552263774216,
      "grad_norm": 1.0773084056504816,
      "learning_rate": 7.92725077390353e-06,
      "loss": 0.5059,
      "step": 5905
    },
    {
      "epoch": 0.318709190005936,
      "grad_norm": 1.1008550347381039,
      "learning_rate": 7.92660834003145e-06,
      "loss": 0.5392,
      "step": 5906
    },
    {
      "epoch": 0.3187631536344504,
      "grad_norm": 1.0224588585450411,
      "learning_rate": 7.925965836416193e-06,
      "loss": 0.3786,
      "step": 5907
    },
    {
      "epoch": 0.31881711726296474,
      "grad_norm": 0.948906860440352,
      "learning_rate": 7.925323263076227e-06,
      "loss": 0.4795,
      "step": 5908
    },
    {
      "epoch": 0.31887108089147914,
      "grad_norm": 0.9097690473329768,
      "learning_rate": 7.924680620030019e-06,
      "loss": 0.4127,
      "step": 5909
    },
    {
      "epoch": 0.31892504451999354,
      "grad_norm": 1.214642090980527,
      "learning_rate": 7.924037907296042e-06,
      "loss": 0.5129,
      "step": 5910
    },
    {
      "epoch": 0.3189790081485079,
      "grad_norm": 0.9667265393100248,
      "learning_rate": 7.923395124892764e-06,
      "loss": 0.3679,
      "step": 5911
    },
    {
      "epoch": 0.3190329717770223,
      "grad_norm": 0.9335198752441802,
      "learning_rate": 7.922752272838664e-06,
      "loss": 0.3908,
      "step": 5912
    },
    {
      "epoch": 0.3190869354055367,
      "grad_norm": 1.1824299136461927,
      "learning_rate": 7.922109351152214e-06,
      "loss": 0.4857,
      "step": 5913
    },
    {
      "epoch": 0.3191408990340511,
      "grad_norm": 1.002311810325618,
      "learning_rate": 7.921466359851892e-06,
      "loss": 0.5593,
      "step": 5914
    },
    {
      "epoch": 0.3191948626625654,
      "grad_norm": 0.8861196534846978,
      "learning_rate": 7.920823298956184e-06,
      "loss": 0.4333,
      "step": 5915
    },
    {
      "epoch": 0.3192488262910798,
      "grad_norm": 1.1235941111438703,
      "learning_rate": 7.920180168483565e-06,
      "loss": 0.5012,
      "step": 5916
    },
    {
      "epoch": 0.3193027899195942,
      "grad_norm": 1.1314726926058671,
      "learning_rate": 7.919536968452524e-06,
      "loss": 0.5409,
      "step": 5917
    },
    {
      "epoch": 0.31935675354810855,
      "grad_norm": 0.9919398993156224,
      "learning_rate": 7.918893698881547e-06,
      "loss": 0.4717,
      "step": 5918
    },
    {
      "epoch": 0.31941071717662295,
      "grad_norm": 0.8342868306808415,
      "learning_rate": 7.91825035978912e-06,
      "loss": 0.3469,
      "step": 5919
    },
    {
      "epoch": 0.31946468080513735,
      "grad_norm": 0.9287318330248803,
      "learning_rate": 7.917606951193734e-06,
      "loss": 0.4345,
      "step": 5920
    },
    {
      "epoch": 0.31951864443365174,
      "grad_norm": 1.028241427927597,
      "learning_rate": 7.916963473113882e-06,
      "loss": 0.5754,
      "step": 5921
    },
    {
      "epoch": 0.3195726080621661,
      "grad_norm": 1.1402164802305959,
      "learning_rate": 7.916319925568056e-06,
      "loss": 0.428,
      "step": 5922
    },
    {
      "epoch": 0.3196265716906805,
      "grad_norm": 1.0302098615496027,
      "learning_rate": 7.915676308574755e-06,
      "loss": 0.457,
      "step": 5923
    },
    {
      "epoch": 0.3196805353191949,
      "grad_norm": 0.6309800117100535,
      "learning_rate": 7.915032622152475e-06,
      "loss": 0.2743,
      "step": 5924
    },
    {
      "epoch": 0.3197344989477092,
      "grad_norm": 1.1632934329006708,
      "learning_rate": 7.91438886631972e-06,
      "loss": 0.4602,
      "step": 5925
    },
    {
      "epoch": 0.3197884625762236,
      "grad_norm": 0.913815986530333,
      "learning_rate": 7.913745041094987e-06,
      "loss": 0.4002,
      "step": 5926
    },
    {
      "epoch": 0.319842426204738,
      "grad_norm": 0.9356763597533835,
      "learning_rate": 7.913101146496783e-06,
      "loss": 0.5539,
      "step": 5927
    },
    {
      "epoch": 0.3198963898332524,
      "grad_norm": 1.1499240588052162,
      "learning_rate": 7.912457182543613e-06,
      "loss": 0.4764,
      "step": 5928
    },
    {
      "epoch": 0.31995035346176676,
      "grad_norm": 1.11358676688239,
      "learning_rate": 7.91181314925399e-06,
      "loss": 0.4975,
      "step": 5929
    },
    {
      "epoch": 0.32000431709028115,
      "grad_norm": 0.9132441045556481,
      "learning_rate": 7.911169046646416e-06,
      "loss": 0.4274,
      "step": 5930
    },
    {
      "epoch": 0.32005828071879555,
      "grad_norm": 0.9228528240488291,
      "learning_rate": 7.910524874739407e-06,
      "loss": 0.5065,
      "step": 5931
    },
    {
      "epoch": 0.3201122443473099,
      "grad_norm": 0.9091138510326043,
      "learning_rate": 7.90988063355148e-06,
      "loss": 0.5389,
      "step": 5932
    },
    {
      "epoch": 0.3201662079758243,
      "grad_norm": 0.9745784632652947,
      "learning_rate": 7.909236323101147e-06,
      "loss": 0.4594,
      "step": 5933
    },
    {
      "epoch": 0.3202201716043387,
      "grad_norm": 1.1059827753247802,
      "learning_rate": 7.908591943406927e-06,
      "loss": 0.6131,
      "step": 5934
    },
    {
      "epoch": 0.32027413523285303,
      "grad_norm": 0.9785017759485038,
      "learning_rate": 7.907947494487343e-06,
      "loss": 0.4735,
      "step": 5935
    },
    {
      "epoch": 0.3203280988613674,
      "grad_norm": 1.000623208514124,
      "learning_rate": 7.907302976360913e-06,
      "loss": 0.378,
      "step": 5936
    },
    {
      "epoch": 0.3203820624898818,
      "grad_norm": 0.9689126532603618,
      "learning_rate": 7.906658389046165e-06,
      "loss": 0.3999,
      "step": 5937
    },
    {
      "epoch": 0.3204360261183962,
      "grad_norm": 0.9401649502626862,
      "learning_rate": 7.906013732561622e-06,
      "loss": 0.4357,
      "step": 5938
    },
    {
      "epoch": 0.32048998974691056,
      "grad_norm": 0.935819315183388,
      "learning_rate": 7.905369006925813e-06,
      "loss": 0.5898,
      "step": 5939
    },
    {
      "epoch": 0.32054395337542496,
      "grad_norm": 1.2529673858211547,
      "learning_rate": 7.904724212157267e-06,
      "loss": 0.527,
      "step": 5940
    },
    {
      "epoch": 0.32059791700393936,
      "grad_norm": 1.3005659348505847,
      "learning_rate": 7.90407934827452e-06,
      "loss": 0.5526,
      "step": 5941
    },
    {
      "epoch": 0.3206518806324537,
      "grad_norm": 0.91918447169636,
      "learning_rate": 7.903434415296103e-06,
      "loss": 0.4816,
      "step": 5942
    },
    {
      "epoch": 0.3207058442609681,
      "grad_norm": 1.3140619574175858,
      "learning_rate": 7.902789413240551e-06,
      "loss": 0.6179,
      "step": 5943
    },
    {
      "epoch": 0.3207598078894825,
      "grad_norm": 0.8075932533080238,
      "learning_rate": 7.902144342126405e-06,
      "loss": 0.3865,
      "step": 5944
    },
    {
      "epoch": 0.3208137715179969,
      "grad_norm": 1.0526313229442101,
      "learning_rate": 7.901499201972202e-06,
      "loss": 0.5472,
      "step": 5945
    },
    {
      "epoch": 0.32086773514651123,
      "grad_norm": 1.1604395833397465,
      "learning_rate": 7.900853992796486e-06,
      "loss": 0.4839,
      "step": 5946
    },
    {
      "epoch": 0.32092169877502563,
      "grad_norm": 0.9962635402597405,
      "learning_rate": 7.9002087146178e-06,
      "loss": 0.4912,
      "step": 5947
    },
    {
      "epoch": 0.32097566240354003,
      "grad_norm": 1.0304913103294047,
      "learning_rate": 7.899563367454692e-06,
      "loss": 0.4947,
      "step": 5948
    },
    {
      "epoch": 0.32102962603205437,
      "grad_norm": 1.20271684228534,
      "learning_rate": 7.898917951325706e-06,
      "loss": 0.636,
      "step": 5949
    },
    {
      "epoch": 0.32108358966056877,
      "grad_norm": 1.0036392943660768,
      "learning_rate": 7.898272466249396e-06,
      "loss": 0.4575,
      "step": 5950
    },
    {
      "epoch": 0.32113755328908317,
      "grad_norm": 1.0462348981071277,
      "learning_rate": 7.897626912244311e-06,
      "loss": 0.4681,
      "step": 5951
    },
    {
      "epoch": 0.32119151691759756,
      "grad_norm": 1.034491830458183,
      "learning_rate": 7.896981289329005e-06,
      "loss": 0.4538,
      "step": 5952
    },
    {
      "epoch": 0.3212454805461119,
      "grad_norm": 1.102045428683058,
      "learning_rate": 7.896335597522037e-06,
      "loss": 0.5755,
      "step": 5953
    },
    {
      "epoch": 0.3212994441746263,
      "grad_norm": 1.0231528289320657,
      "learning_rate": 7.895689836841961e-06,
      "loss": 0.6004,
      "step": 5954
    },
    {
      "epoch": 0.3213534078031407,
      "grad_norm": 1.0053977450325944,
      "learning_rate": 7.895044007307338e-06,
      "loss": 0.4682,
      "step": 5955
    },
    {
      "epoch": 0.32140737143165504,
      "grad_norm": 1.0194904388710258,
      "learning_rate": 7.894398108936733e-06,
      "loss": 0.4243,
      "step": 5956
    },
    {
      "epoch": 0.32146133506016944,
      "grad_norm": 0.8586913189830022,
      "learning_rate": 7.893752141748705e-06,
      "loss": 0.4316,
      "step": 5957
    },
    {
      "epoch": 0.32151529868868384,
      "grad_norm": 1.1836899440149393,
      "learning_rate": 7.89310610576182e-06,
      "loss": 0.5903,
      "step": 5958
    },
    {
      "epoch": 0.32156926231719823,
      "grad_norm": 1.0327946598185154,
      "learning_rate": 7.892460000994649e-06,
      "loss": 0.5217,
      "step": 5959
    },
    {
      "epoch": 0.3216232259457126,
      "grad_norm": 0.9772371785492189,
      "learning_rate": 7.891813827465761e-06,
      "loss": 0.4047,
      "step": 5960
    },
    {
      "epoch": 0.321677189574227,
      "grad_norm": 0.928707106540232,
      "learning_rate": 7.891167585193724e-06,
      "loss": 0.3613,
      "step": 5961
    },
    {
      "epoch": 0.32173115320274137,
      "grad_norm": 0.9004139349383272,
      "learning_rate": 7.890521274197119e-06,
      "loss": 0.3914,
      "step": 5962
    },
    {
      "epoch": 0.3217851168312557,
      "grad_norm": 0.9258773042001399,
      "learning_rate": 7.889874894494515e-06,
      "loss": 0.3772,
      "step": 5963
    },
    {
      "epoch": 0.3218390804597701,
      "grad_norm": 1.2108257169204986,
      "learning_rate": 7.889228446104492e-06,
      "loss": 0.474,
      "step": 5964
    },
    {
      "epoch": 0.3218930440882845,
      "grad_norm": 1.0170126463214018,
      "learning_rate": 7.888581929045629e-06,
      "loss": 0.5159,
      "step": 5965
    },
    {
      "epoch": 0.3219470077167989,
      "grad_norm": 1.0522850283683047,
      "learning_rate": 7.887935343336506e-06,
      "loss": 0.5003,
      "step": 5966
    },
    {
      "epoch": 0.32200097134531325,
      "grad_norm": 1.2083278056572362,
      "learning_rate": 7.88728868899571e-06,
      "loss": 0.5081,
      "step": 5967
    },
    {
      "epoch": 0.32205493497382764,
      "grad_norm": 1.1254941612702527,
      "learning_rate": 7.886641966041826e-06,
      "loss": 0.4811,
      "step": 5968
    },
    {
      "epoch": 0.32210889860234204,
      "grad_norm": 0.895796721462946,
      "learning_rate": 7.88599517449344e-06,
      "loss": 0.4238,
      "step": 5969
    },
    {
      "epoch": 0.3221628622308564,
      "grad_norm": 0.8962440626420475,
      "learning_rate": 7.885348314369141e-06,
      "loss": 0.3583,
      "step": 5970
    },
    {
      "epoch": 0.3222168258593708,
      "grad_norm": 1.0338502518591037,
      "learning_rate": 7.884701385687522e-06,
      "loss": 0.4516,
      "step": 5971
    },
    {
      "epoch": 0.3222707894878852,
      "grad_norm": 1.0770589366193566,
      "learning_rate": 7.884054388467176e-06,
      "loss": 0.4435,
      "step": 5972
    },
    {
      "epoch": 0.3223247531163995,
      "grad_norm": 1.0027776555865757,
      "learning_rate": 7.883407322726696e-06,
      "loss": 0.5537,
      "step": 5973
    },
    {
      "epoch": 0.3223787167449139,
      "grad_norm": 1.0257954816733534,
      "learning_rate": 7.882760188484683e-06,
      "loss": 0.3707,
      "step": 5974
    },
    {
      "epoch": 0.3224326803734283,
      "grad_norm": 0.9372059086577426,
      "learning_rate": 7.882112985759734e-06,
      "loss": 0.4735,
      "step": 5975
    },
    {
      "epoch": 0.3224866440019427,
      "grad_norm": 0.7959259272351011,
      "learning_rate": 7.88146571457045e-06,
      "loss": 0.343,
      "step": 5976
    },
    {
      "epoch": 0.32254060763045705,
      "grad_norm": 0.8314271617331521,
      "learning_rate": 7.880818374935436e-06,
      "loss": 0.3982,
      "step": 5977
    },
    {
      "epoch": 0.32259457125897145,
      "grad_norm": 1.1371707843625738,
      "learning_rate": 7.880170966873296e-06,
      "loss": 0.45,
      "step": 5978
    },
    {
      "epoch": 0.32264853488748585,
      "grad_norm": 1.0329817177978784,
      "learning_rate": 7.879523490402638e-06,
      "loss": 0.4057,
      "step": 5979
    },
    {
      "epoch": 0.3227024985160002,
      "grad_norm": 0.9584914980517119,
      "learning_rate": 7.878875945542071e-06,
      "loss": 0.4011,
      "step": 5980
    },
    {
      "epoch": 0.3227564621445146,
      "grad_norm": 1.053067823833615,
      "learning_rate": 7.878228332310204e-06,
      "loss": 0.5175,
      "step": 5981
    },
    {
      "epoch": 0.322810425773029,
      "grad_norm": 0.9883802204666124,
      "learning_rate": 7.877580650725653e-06,
      "loss": 0.4732,
      "step": 5982
    },
    {
      "epoch": 0.3228643894015434,
      "grad_norm": 0.9003743700416751,
      "learning_rate": 7.876932900807029e-06,
      "loss": 0.5378,
      "step": 5983
    },
    {
      "epoch": 0.3229183530300577,
      "grad_norm": 1.0234876987748245,
      "learning_rate": 7.876285082572956e-06,
      "loss": 0.4916,
      "step": 5984
    },
    {
      "epoch": 0.3229723166585721,
      "grad_norm": 0.8167169409750263,
      "learning_rate": 7.875637196042046e-06,
      "loss": 0.3551,
      "step": 5985
    },
    {
      "epoch": 0.3230262802870865,
      "grad_norm": 1.0531019347989066,
      "learning_rate": 7.874989241232923e-06,
      "loss": 0.4865,
      "step": 5986
    },
    {
      "epoch": 0.32308024391560086,
      "grad_norm": 1.0819642828732705,
      "learning_rate": 7.87434121816421e-06,
      "loss": 0.4772,
      "step": 5987
    },
    {
      "epoch": 0.32313420754411526,
      "grad_norm": 0.9951657783778652,
      "learning_rate": 7.87369312685453e-06,
      "loss": 0.5682,
      "step": 5988
    },
    {
      "epoch": 0.32318817117262966,
      "grad_norm": 1.0192453957095002,
      "learning_rate": 7.873044967322511e-06,
      "loss": 0.5713,
      "step": 5989
    },
    {
      "epoch": 0.32324213480114405,
      "grad_norm": 1.0250968739478206,
      "learning_rate": 7.872396739586784e-06,
      "loss": 0.4743,
      "step": 5990
    },
    {
      "epoch": 0.3232960984296584,
      "grad_norm": 1.1481343947313774,
      "learning_rate": 7.871748443665977e-06,
      "loss": 0.5565,
      "step": 5991
    },
    {
      "epoch": 0.3233500620581728,
      "grad_norm": 0.9749919528663178,
      "learning_rate": 7.871100079578722e-06,
      "loss": 0.3781,
      "step": 5992
    },
    {
      "epoch": 0.3234040256866872,
      "grad_norm": 1.2675135877431165,
      "learning_rate": 7.870451647343656e-06,
      "loss": 0.5373,
      "step": 5993
    },
    {
      "epoch": 0.32345798931520153,
      "grad_norm": 0.8214021154524622,
      "learning_rate": 7.869803146979414e-06,
      "loss": 0.3419,
      "step": 5994
    },
    {
      "epoch": 0.32351195294371593,
      "grad_norm": 1.0231918047672879,
      "learning_rate": 7.869154578504634e-06,
      "loss": 0.5955,
      "step": 5995
    },
    {
      "epoch": 0.3235659165722303,
      "grad_norm": 1.0119235350260083,
      "learning_rate": 7.868505941937959e-06,
      "loss": 0.3894,
      "step": 5996
    },
    {
      "epoch": 0.3236198802007447,
      "grad_norm": 1.1171954318871131,
      "learning_rate": 7.86785723729803e-06,
      "loss": 0.4735,
      "step": 5997
    },
    {
      "epoch": 0.32367384382925907,
      "grad_norm": 0.9953878269664985,
      "learning_rate": 7.86720846460349e-06,
      "loss": 0.4938,
      "step": 5998
    },
    {
      "epoch": 0.32372780745777346,
      "grad_norm": 0.7900486873321353,
      "learning_rate": 7.86655962387299e-06,
      "loss": 0.3387,
      "step": 5999
    },
    {
      "epoch": 0.32378177108628786,
      "grad_norm": 1.3181226899486023,
      "learning_rate": 7.865910715125172e-06,
      "loss": 0.5042,
      "step": 6000
    },
    {
      "epoch": 0.32378177108628786,
      "eval_loss": 0.5695191025733948,
      "eval_runtime": 163.9743,
      "eval_samples_per_second": 20.973,
      "eval_steps_per_second": 0.878,
      "step": 6000
    },
    {
      "epoch": 0.3238357347148022,
      "grad_norm": 0.8016413255225768,
      "learning_rate": 7.86526173837869e-06,
      "loss": 0.2859,
      "step": 6001
    },
    {
      "epoch": 0.3238896983433166,
      "grad_norm": 1.0263554092244853,
      "learning_rate": 7.864612693652198e-06,
      "loss": 0.4303,
      "step": 6002
    },
    {
      "epoch": 0.323943661971831,
      "grad_norm": 1.0507886807859228,
      "learning_rate": 7.863963580964344e-06,
      "loss": 0.4679,
      "step": 6003
    },
    {
      "epoch": 0.32399762560034534,
      "grad_norm": 0.8509495752549467,
      "learning_rate": 7.863314400333793e-06,
      "loss": 0.4705,
      "step": 6004
    },
    {
      "epoch": 0.32405158922885974,
      "grad_norm": 1.1381288010712423,
      "learning_rate": 7.862665151779195e-06,
      "loss": 0.5968,
      "step": 6005
    },
    {
      "epoch": 0.32410555285737414,
      "grad_norm": 1.1859335784464307,
      "learning_rate": 7.862015835319213e-06,
      "loss": 0.5179,
      "step": 6006
    },
    {
      "epoch": 0.32415951648588853,
      "grad_norm": 1.2668019825465817,
      "learning_rate": 7.861366450972509e-06,
      "loss": 0.4522,
      "step": 6007
    },
    {
      "epoch": 0.3242134801144029,
      "grad_norm": 1.1165754504194687,
      "learning_rate": 7.86071699875775e-06,
      "loss": 0.4782,
      "step": 6008
    },
    {
      "epoch": 0.32426744374291727,
      "grad_norm": 1.1151598077623261,
      "learning_rate": 7.860067478693597e-06,
      "loss": 0.6063,
      "step": 6009
    },
    {
      "epoch": 0.32432140737143167,
      "grad_norm": 1.0404140266536035,
      "learning_rate": 7.85941789079872e-06,
      "loss": 0.3934,
      "step": 6010
    },
    {
      "epoch": 0.324375370999946,
      "grad_norm": 1.0403130950851807,
      "learning_rate": 7.858768235091788e-06,
      "loss": 0.3591,
      "step": 6011
    },
    {
      "epoch": 0.3244293346284604,
      "grad_norm": 1.0106328791158328,
      "learning_rate": 7.858118511591476e-06,
      "loss": 0.5218,
      "step": 6012
    },
    {
      "epoch": 0.3244832982569748,
      "grad_norm": 0.9074220419301285,
      "learning_rate": 7.857468720316454e-06,
      "loss": 0.4859,
      "step": 6013
    },
    {
      "epoch": 0.3245372618854892,
      "grad_norm": 1.1941177222021893,
      "learning_rate": 7.856818861285398e-06,
      "loss": 0.5575,
      "step": 6014
    },
    {
      "epoch": 0.32459122551400355,
      "grad_norm": 1.2394127047785763,
      "learning_rate": 7.856168934516987e-06,
      "loss": 0.6421,
      "step": 6015
    },
    {
      "epoch": 0.32464518914251794,
      "grad_norm": 1.037366397795698,
      "learning_rate": 7.8555189400299e-06,
      "loss": 0.4506,
      "step": 6016
    },
    {
      "epoch": 0.32469915277103234,
      "grad_norm": 1.0801336652199232,
      "learning_rate": 7.85486887784282e-06,
      "loss": 0.5151,
      "step": 6017
    },
    {
      "epoch": 0.3247531163995467,
      "grad_norm": 0.9538346787866813,
      "learning_rate": 7.854218747974429e-06,
      "loss": 0.2954,
      "step": 6018
    },
    {
      "epoch": 0.3248070800280611,
      "grad_norm": 0.9276513311268766,
      "learning_rate": 7.853568550443412e-06,
      "loss": 0.3824,
      "step": 6019
    },
    {
      "epoch": 0.3248610436565755,
      "grad_norm": 0.9627758705059813,
      "learning_rate": 7.852918285268455e-06,
      "loss": 0.4308,
      "step": 6020
    },
    {
      "epoch": 0.3249150072850899,
      "grad_norm": 1.1566862255646715,
      "learning_rate": 7.85226795246825e-06,
      "loss": 0.6689,
      "step": 6021
    },
    {
      "epoch": 0.3249689709136042,
      "grad_norm": 0.9065902930994093,
      "learning_rate": 7.851617552061488e-06,
      "loss": 0.4199,
      "step": 6022
    },
    {
      "epoch": 0.3250229345421186,
      "grad_norm": 1.201369124323388,
      "learning_rate": 7.850967084066861e-06,
      "loss": 0.6766,
      "step": 6023
    },
    {
      "epoch": 0.325076898170633,
      "grad_norm": 1.1384224649745283,
      "learning_rate": 7.850316548503063e-06,
      "loss": 0.5562,
      "step": 6024
    },
    {
      "epoch": 0.32513086179914735,
      "grad_norm": 1.0720609338241982,
      "learning_rate": 7.849665945388797e-06,
      "loss": 0.5318,
      "step": 6025
    },
    {
      "epoch": 0.32518482542766175,
      "grad_norm": 1.00911565792164,
      "learning_rate": 7.849015274742753e-06,
      "loss": 0.4561,
      "step": 6026
    },
    {
      "epoch": 0.32523878905617615,
      "grad_norm": 0.8873024972829716,
      "learning_rate": 7.848364536583638e-06,
      "loss": 0.3531,
      "step": 6027
    },
    {
      "epoch": 0.32529275268469054,
      "grad_norm": 1.1154447628880586,
      "learning_rate": 7.847713730930155e-06,
      "loss": 0.5626,
      "step": 6028
    },
    {
      "epoch": 0.3253467163132049,
      "grad_norm": 1.1178920203049403,
      "learning_rate": 7.847062857801005e-06,
      "loss": 0.5527,
      "step": 6029
    },
    {
      "epoch": 0.3254006799417193,
      "grad_norm": 0.9609055520142648,
      "learning_rate": 7.846411917214897e-06,
      "loss": 0.4658,
      "step": 6030
    },
    {
      "epoch": 0.3254546435702337,
      "grad_norm": 1.0578684828601208,
      "learning_rate": 7.845760909190542e-06,
      "loss": 0.5102,
      "step": 6031
    },
    {
      "epoch": 0.325508607198748,
      "grad_norm": 1.0343317965987715,
      "learning_rate": 7.845109833746644e-06,
      "loss": 0.4733,
      "step": 6032
    },
    {
      "epoch": 0.3255625708272624,
      "grad_norm": 1.0830611706693334,
      "learning_rate": 7.844458690901923e-06,
      "loss": 0.5699,
      "step": 6033
    },
    {
      "epoch": 0.3256165344557768,
      "grad_norm": 0.9023550897494328,
      "learning_rate": 7.84380748067509e-06,
      "loss": 0.4296,
      "step": 6034
    },
    {
      "epoch": 0.32567049808429116,
      "grad_norm": 1.2816201584222446,
      "learning_rate": 7.843156203084861e-06,
      "loss": 0.604,
      "step": 6035
    },
    {
      "epoch": 0.32572446171280556,
      "grad_norm": 1.036893186826935,
      "learning_rate": 7.842504858149956e-06,
      "loss": 0.412,
      "step": 6036
    },
    {
      "epoch": 0.32577842534131995,
      "grad_norm": 0.9973101259191083,
      "learning_rate": 7.841853445889096e-06,
      "loss": 0.3588,
      "step": 6037
    },
    {
      "epoch": 0.32583238896983435,
      "grad_norm": 1.1612846766152949,
      "learning_rate": 7.841201966320999e-06,
      "loss": 0.4536,
      "step": 6038
    },
    {
      "epoch": 0.3258863525983487,
      "grad_norm": 1.010176772537534,
      "learning_rate": 7.84055041946439e-06,
      "loss": 0.3987,
      "step": 6039
    },
    {
      "epoch": 0.3259403162268631,
      "grad_norm": 1.2053397749336856,
      "learning_rate": 7.839898805338002e-06,
      "loss": 0.6834,
      "step": 6040
    },
    {
      "epoch": 0.3259942798553775,
      "grad_norm": 1.1000928622467565,
      "learning_rate": 7.839247123960554e-06,
      "loss": 0.6532,
      "step": 6041
    },
    {
      "epoch": 0.32604824348389183,
      "grad_norm": 0.8847699387158673,
      "learning_rate": 7.838595375350781e-06,
      "loss": 0.4449,
      "step": 6042
    },
    {
      "epoch": 0.32610220711240623,
      "grad_norm": 1.1879070708546324,
      "learning_rate": 7.837943559527416e-06,
      "loss": 0.611,
      "step": 6043
    },
    {
      "epoch": 0.3261561707409206,
      "grad_norm": 0.9433047923938608,
      "learning_rate": 7.837291676509188e-06,
      "loss": 0.3984,
      "step": 6044
    },
    {
      "epoch": 0.326210134369435,
      "grad_norm": 0.9425278924284124,
      "learning_rate": 7.836639726314837e-06,
      "loss": 0.4435,
      "step": 6045
    },
    {
      "epoch": 0.32626409799794936,
      "grad_norm": 0.9889509525049538,
      "learning_rate": 7.8359877089631e-06,
      "loss": 0.3967,
      "step": 6046
    },
    {
      "epoch": 0.32631806162646376,
      "grad_norm": 0.9562498165914874,
      "learning_rate": 7.835335624472714e-06,
      "loss": 0.4304,
      "step": 6047
    },
    {
      "epoch": 0.32637202525497816,
      "grad_norm": 0.8245261955777404,
      "learning_rate": 7.834683472862424e-06,
      "loss": 0.3383,
      "step": 6048
    },
    {
      "epoch": 0.3264259888834925,
      "grad_norm": 0.9901776927061094,
      "learning_rate": 7.834031254150972e-06,
      "loss": 0.6345,
      "step": 6049
    },
    {
      "epoch": 0.3264799525120069,
      "grad_norm": 0.8124534871056996,
      "learning_rate": 7.833378968357103e-06,
      "loss": 0.4048,
      "step": 6050
    },
    {
      "epoch": 0.3265339161405213,
      "grad_norm": 1.1018172149430634,
      "learning_rate": 7.832726615499566e-06,
      "loss": 0.5347,
      "step": 6051
    },
    {
      "epoch": 0.3265878797690357,
      "grad_norm": 0.8540654018758913,
      "learning_rate": 7.832074195597107e-06,
      "loss": 0.3897,
      "step": 6052
    },
    {
      "epoch": 0.32664184339755004,
      "grad_norm": 1.0156765117065754,
      "learning_rate": 7.831421708668479e-06,
      "loss": 0.5805,
      "step": 6053
    },
    {
      "epoch": 0.32669580702606443,
      "grad_norm": 0.9635602139293212,
      "learning_rate": 7.830769154732436e-06,
      "loss": 0.5176,
      "step": 6054
    },
    {
      "epoch": 0.32674977065457883,
      "grad_norm": 1.0171674463810585,
      "learning_rate": 7.830116533807733e-06,
      "loss": 0.4853,
      "step": 6055
    },
    {
      "epoch": 0.3268037342830932,
      "grad_norm": 1.1800036474157283,
      "learning_rate": 7.829463845913126e-06,
      "loss": 0.5722,
      "step": 6056
    },
    {
      "epoch": 0.32685769791160757,
      "grad_norm": 1.2229143633185457,
      "learning_rate": 7.828811091067374e-06,
      "loss": 0.5646,
      "step": 6057
    },
    {
      "epoch": 0.32691166154012197,
      "grad_norm": 0.7801234341026698,
      "learning_rate": 7.828158269289239e-06,
      "loss": 0.3733,
      "step": 6058
    },
    {
      "epoch": 0.32696562516863636,
      "grad_norm": 1.005833340571049,
      "learning_rate": 7.827505380597481e-06,
      "loss": 0.4422,
      "step": 6059
    },
    {
      "epoch": 0.3270195887971507,
      "grad_norm": 0.9441850532543788,
      "learning_rate": 7.826852425010868e-06,
      "loss": 0.489,
      "step": 6060
    },
    {
      "epoch": 0.3270735524256651,
      "grad_norm": 0.8921613351754532,
      "learning_rate": 7.826199402548167e-06,
      "loss": 0.367,
      "step": 6061
    },
    {
      "epoch": 0.3271275160541795,
      "grad_norm": 0.8861007731079703,
      "learning_rate": 7.825546313228141e-06,
      "loss": 0.3417,
      "step": 6062
    },
    {
      "epoch": 0.32718147968269384,
      "grad_norm": 1.1268070936083874,
      "learning_rate": 7.824893157069568e-06,
      "loss": 0.6046,
      "step": 6063
    },
    {
      "epoch": 0.32723544331120824,
      "grad_norm": 1.026040299302559,
      "learning_rate": 7.824239934091214e-06,
      "loss": 0.4673,
      "step": 6064
    },
    {
      "epoch": 0.32728940693972264,
      "grad_norm": 1.0507323097998134,
      "learning_rate": 7.823586644311857e-06,
      "loss": 0.5246,
      "step": 6065
    },
    {
      "epoch": 0.32734337056823704,
      "grad_norm": 0.9128044276434362,
      "learning_rate": 7.82293328775027e-06,
      "loss": 0.3925,
      "step": 6066
    },
    {
      "epoch": 0.3273973341967514,
      "grad_norm": 0.8892358679105018,
      "learning_rate": 7.822279864425235e-06,
      "loss": 0.3187,
      "step": 6067
    },
    {
      "epoch": 0.3274512978252658,
      "grad_norm": 1.0746749303298049,
      "learning_rate": 7.82162637435553e-06,
      "loss": 0.5598,
      "step": 6068
    },
    {
      "epoch": 0.32750526145378017,
      "grad_norm": 1.0253778098937847,
      "learning_rate": 7.820972817559938e-06,
      "loss": 0.5352,
      "step": 6069
    },
    {
      "epoch": 0.3275592250822945,
      "grad_norm": 1.1504190017318505,
      "learning_rate": 7.820319194057241e-06,
      "loss": 0.6271,
      "step": 6070
    },
    {
      "epoch": 0.3276131887108089,
      "grad_norm": 1.1438704935625597,
      "learning_rate": 7.819665503866226e-06,
      "loss": 0.4766,
      "step": 6071
    },
    {
      "epoch": 0.3276671523393233,
      "grad_norm": 0.9461197678211211,
      "learning_rate": 7.819011747005679e-06,
      "loss": 0.5585,
      "step": 6072
    },
    {
      "epoch": 0.32772111596783765,
      "grad_norm": 1.010602382751191,
      "learning_rate": 7.818357923494393e-06,
      "loss": 0.4191,
      "step": 6073
    },
    {
      "epoch": 0.32777507959635205,
      "grad_norm": 0.9575444124062902,
      "learning_rate": 7.817704033351157e-06,
      "loss": 0.4572,
      "step": 6074
    },
    {
      "epoch": 0.32782904322486645,
      "grad_norm": 0.8752066227157896,
      "learning_rate": 7.817050076594767e-06,
      "loss": 0.373,
      "step": 6075
    },
    {
      "epoch": 0.32788300685338084,
      "grad_norm": 1.1749051834514366,
      "learning_rate": 7.816396053244014e-06,
      "loss": 0.4921,
      "step": 6076
    },
    {
      "epoch": 0.3279369704818952,
      "grad_norm": 0.91179918091204,
      "learning_rate": 7.8157419633177e-06,
      "loss": 0.3591,
      "step": 6077
    },
    {
      "epoch": 0.3279909341104096,
      "grad_norm": 1.0242596440257616,
      "learning_rate": 7.815087806834622e-06,
      "loss": 0.4468,
      "step": 6078
    },
    {
      "epoch": 0.328044897738924,
      "grad_norm": 1.1730953189174727,
      "learning_rate": 7.814433583813579e-06,
      "loss": 0.5415,
      "step": 6079
    },
    {
      "epoch": 0.3280988613674383,
      "grad_norm": 1.1579758203193515,
      "learning_rate": 7.813779294273378e-06,
      "loss": 0.562,
      "step": 6080
    },
    {
      "epoch": 0.3281528249959527,
      "grad_norm": 0.8083202886741409,
      "learning_rate": 7.813124938232824e-06,
      "loss": 0.3884,
      "step": 6081
    },
    {
      "epoch": 0.3282067886244671,
      "grad_norm": 1.0111330168355748,
      "learning_rate": 7.812470515710719e-06,
      "loss": 0.4727,
      "step": 6082
    },
    {
      "epoch": 0.3282607522529815,
      "grad_norm": 0.9842406225630427,
      "learning_rate": 7.811816026725874e-06,
      "loss": 0.422,
      "step": 6083
    },
    {
      "epoch": 0.32831471588149586,
      "grad_norm": 1.2312340463962488,
      "learning_rate": 7.811161471297105e-06,
      "loss": 0.5857,
      "step": 6084
    },
    {
      "epoch": 0.32836867951001025,
      "grad_norm": 1.0622988280755192,
      "learning_rate": 7.810506849443216e-06,
      "loss": 0.4229,
      "step": 6085
    },
    {
      "epoch": 0.32842264313852465,
      "grad_norm": 1.0680878961896851,
      "learning_rate": 7.809852161183026e-06,
      "loss": 0.4025,
      "step": 6086
    },
    {
      "epoch": 0.328476606767039,
      "grad_norm": 1.1132893641940282,
      "learning_rate": 7.809197406535352e-06,
      "loss": 0.5829,
      "step": 6087
    },
    {
      "epoch": 0.3285305703955534,
      "grad_norm": 1.1023019765738784,
      "learning_rate": 7.808542585519009e-06,
      "loss": 0.6425,
      "step": 6088
    },
    {
      "epoch": 0.3285845340240678,
      "grad_norm": 0.940321954427669,
      "learning_rate": 7.807887698152822e-06,
      "loss": 0.4212,
      "step": 6089
    },
    {
      "epoch": 0.3286384976525822,
      "grad_norm": 1.031814900473182,
      "learning_rate": 7.80723274445561e-06,
      "loss": 0.4511,
      "step": 6090
    },
    {
      "epoch": 0.3286924612810965,
      "grad_norm": 0.908503840395287,
      "learning_rate": 7.806577724446198e-06,
      "loss": 0.4227,
      "step": 6091
    },
    {
      "epoch": 0.3287464249096109,
      "grad_norm": 0.763864642759267,
      "learning_rate": 7.80592263814341e-06,
      "loss": 0.3654,
      "step": 6092
    },
    {
      "epoch": 0.3288003885381253,
      "grad_norm": 0.9987776396446228,
      "learning_rate": 7.805267485566078e-06,
      "loss": 0.3385,
      "step": 6093
    },
    {
      "epoch": 0.32885435216663966,
      "grad_norm": 0.8743003862823229,
      "learning_rate": 7.804612266733027e-06,
      "loss": 0.3894,
      "step": 6094
    },
    {
      "epoch": 0.32890831579515406,
      "grad_norm": 1.0113020870084648,
      "learning_rate": 7.803956981663093e-06,
      "loss": 0.5101,
      "step": 6095
    },
    {
      "epoch": 0.32896227942366846,
      "grad_norm": 0.9141782010831855,
      "learning_rate": 7.803301630375106e-06,
      "loss": 0.3546,
      "step": 6096
    },
    {
      "epoch": 0.32901624305218286,
      "grad_norm": 0.9479588818707817,
      "learning_rate": 7.802646212887902e-06,
      "loss": 0.4038,
      "step": 6097
    },
    {
      "epoch": 0.3290702066806972,
      "grad_norm": 0.8217898268444614,
      "learning_rate": 7.80199072922032e-06,
      "loss": 0.3269,
      "step": 6098
    },
    {
      "epoch": 0.3291241703092116,
      "grad_norm": 0.8950533932566802,
      "learning_rate": 7.801335179391199e-06,
      "loss": 0.4485,
      "step": 6099
    },
    {
      "epoch": 0.329178133937726,
      "grad_norm": 1.1121206135623902,
      "learning_rate": 7.800679563419378e-06,
      "loss": 0.5339,
      "step": 6100
    },
    {
      "epoch": 0.32923209756624033,
      "grad_norm": 0.8633564223702286,
      "learning_rate": 7.800023881323705e-06,
      "loss": 0.5134,
      "step": 6101
    },
    {
      "epoch": 0.32928606119475473,
      "grad_norm": 0.810312885318976,
      "learning_rate": 7.79936813312302e-06,
      "loss": 0.5086,
      "step": 6102
    },
    {
      "epoch": 0.32934002482326913,
      "grad_norm": 1.0229570710582174,
      "learning_rate": 7.79871231883617e-06,
      "loss": 0.3955,
      "step": 6103
    },
    {
      "epoch": 0.32939398845178347,
      "grad_norm": 1.0115095598305568,
      "learning_rate": 7.798056438482006e-06,
      "loss": 0.513,
      "step": 6104
    },
    {
      "epoch": 0.32944795208029787,
      "grad_norm": 0.7328385130147109,
      "learning_rate": 7.797400492079379e-06,
      "loss": 0.3241,
      "step": 6105
    },
    {
      "epoch": 0.32950191570881227,
      "grad_norm": 1.1570114396302806,
      "learning_rate": 7.796744479647138e-06,
      "loss": 0.5975,
      "step": 6106
    },
    {
      "epoch": 0.32955587933732666,
      "grad_norm": 1.1896064684835053,
      "learning_rate": 7.796088401204142e-06,
      "loss": 0.6235,
      "step": 6107
    },
    {
      "epoch": 0.329609842965841,
      "grad_norm": 1.0584511717228382,
      "learning_rate": 7.795432256769247e-06,
      "loss": 0.5006,
      "step": 6108
    },
    {
      "epoch": 0.3296638065943554,
      "grad_norm": 1.139777008966188,
      "learning_rate": 7.794776046361306e-06,
      "loss": 0.5461,
      "step": 6109
    },
    {
      "epoch": 0.3297177702228698,
      "grad_norm": 0.9015283740802661,
      "learning_rate": 7.794119769999183e-06,
      "loss": 0.4175,
      "step": 6110
    },
    {
      "epoch": 0.32977173385138414,
      "grad_norm": 1.0879177556968282,
      "learning_rate": 7.793463427701741e-06,
      "loss": 0.5745,
      "step": 6111
    },
    {
      "epoch": 0.32982569747989854,
      "grad_norm": 1.0406410240543096,
      "learning_rate": 7.792807019487845e-06,
      "loss": 0.5017,
      "step": 6112
    },
    {
      "epoch": 0.32987966110841294,
      "grad_norm": 1.2146470574222767,
      "learning_rate": 7.792150545376354e-06,
      "loss": 0.5486,
      "step": 6113
    },
    {
      "epoch": 0.32993362473692733,
      "grad_norm": 0.9175806778670064,
      "learning_rate": 7.791494005386144e-06,
      "loss": 0.4245,
      "step": 6114
    },
    {
      "epoch": 0.3299875883654417,
      "grad_norm": 1.0663173678806595,
      "learning_rate": 7.79083739953608e-06,
      "loss": 0.4961,
      "step": 6115
    },
    {
      "epoch": 0.3300415519939561,
      "grad_norm": 1.1251536064299132,
      "learning_rate": 7.790180727845033e-06,
      "loss": 0.5283,
      "step": 6116
    },
    {
      "epoch": 0.33009551562247047,
      "grad_norm": 0.7970967079000394,
      "learning_rate": 7.789523990331877e-06,
      "loss": 0.3398,
      "step": 6117
    },
    {
      "epoch": 0.3301494792509848,
      "grad_norm": 1.469769274472929,
      "learning_rate": 7.78886718701549e-06,
      "loss": 0.6187,
      "step": 6118
    },
    {
      "epoch": 0.3302034428794992,
      "grad_norm": 0.721677327935083,
      "learning_rate": 7.788210317914747e-06,
      "loss": 0.3028,
      "step": 6119
    },
    {
      "epoch": 0.3302574065080136,
      "grad_norm": 0.9598269193925923,
      "learning_rate": 7.787553383048529e-06,
      "loss": 0.5038,
      "step": 6120
    },
    {
      "epoch": 0.330311370136528,
      "grad_norm": 0.8890415062612552,
      "learning_rate": 7.786896382435712e-06,
      "loss": 0.51,
      "step": 6121
    },
    {
      "epoch": 0.33036533376504235,
      "grad_norm": 0.8767350977759248,
      "learning_rate": 7.786239316095186e-06,
      "loss": 0.3517,
      "step": 6122
    },
    {
      "epoch": 0.33041929739355674,
      "grad_norm": 0.9129863341236272,
      "learning_rate": 7.78558218404583e-06,
      "loss": 0.49,
      "step": 6123
    },
    {
      "epoch": 0.33047326102207114,
      "grad_norm": 0.8351938611984346,
      "learning_rate": 7.78492498630653e-06,
      "loss": 0.3829,
      "step": 6124
    },
    {
      "epoch": 0.3305272246505855,
      "grad_norm": 0.9214872936628962,
      "learning_rate": 7.784267722896182e-06,
      "loss": 0.4818,
      "step": 6125
    },
    {
      "epoch": 0.3305811882790999,
      "grad_norm": 0.9772387074633102,
      "learning_rate": 7.78361039383367e-06,
      "loss": 0.665,
      "step": 6126
    },
    {
      "epoch": 0.3306351519076143,
      "grad_norm": 0.8771553690406717,
      "learning_rate": 7.782952999137888e-06,
      "loss": 0.3572,
      "step": 6127
    },
    {
      "epoch": 0.3306891155361287,
      "grad_norm": 1.0081758631176496,
      "learning_rate": 7.782295538827727e-06,
      "loss": 0.5799,
      "step": 6128
    },
    {
      "epoch": 0.330743079164643,
      "grad_norm": 1.1598085416065091,
      "learning_rate": 7.78163801292209e-06,
      "loss": 0.5301,
      "step": 6129
    },
    {
      "epoch": 0.3307970427931574,
      "grad_norm": 1.0784440956454608,
      "learning_rate": 7.780980421439868e-06,
      "loss": 0.6295,
      "step": 6130
    },
    {
      "epoch": 0.3308510064216718,
      "grad_norm": 1.0730411088684984,
      "learning_rate": 7.780322764399966e-06,
      "loss": 0.5644,
      "step": 6131
    },
    {
      "epoch": 0.33090497005018615,
      "grad_norm": 0.9585971642264112,
      "learning_rate": 7.779665041821283e-06,
      "loss": 0.4031,
      "step": 6132
    },
    {
      "epoch": 0.33095893367870055,
      "grad_norm": 0.9090169181178056,
      "learning_rate": 7.779007253722724e-06,
      "loss": 0.3768,
      "step": 6133
    },
    {
      "epoch": 0.33101289730721495,
      "grad_norm": 0.9027988457670335,
      "learning_rate": 7.778349400123193e-06,
      "loss": 0.4037,
      "step": 6134
    },
    {
      "epoch": 0.33106686093572935,
      "grad_norm": 1.2268859048647611,
      "learning_rate": 7.777691481041597e-06,
      "loss": 0.6937,
      "step": 6135
    },
    {
      "epoch": 0.3311208245642437,
      "grad_norm": 1.1372995277245583,
      "learning_rate": 7.77703349649685e-06,
      "loss": 0.6176,
      "step": 6136
    },
    {
      "epoch": 0.3311747881927581,
      "grad_norm": 1.0567387595288629,
      "learning_rate": 7.776375446507857e-06,
      "loss": 0.5809,
      "step": 6137
    },
    {
      "epoch": 0.3312287518212725,
      "grad_norm": 1.081897335836043,
      "learning_rate": 7.775717331093533e-06,
      "loss": 0.5538,
      "step": 6138
    },
    {
      "epoch": 0.3312827154497868,
      "grad_norm": 1.1047635189905596,
      "learning_rate": 7.775059150272795e-06,
      "loss": 0.6363,
      "step": 6139
    },
    {
      "epoch": 0.3313366790783012,
      "grad_norm": 1.2515148565500125,
      "learning_rate": 7.774400904064556e-06,
      "loss": 0.4816,
      "step": 6140
    },
    {
      "epoch": 0.3313906427068156,
      "grad_norm": 0.9158682287962409,
      "learning_rate": 7.773742592487739e-06,
      "loss": 0.3639,
      "step": 6141
    },
    {
      "epoch": 0.33144460633532996,
      "grad_norm": 0.8884769213525272,
      "learning_rate": 7.773084215561262e-06,
      "loss": 0.3785,
      "step": 6142
    },
    {
      "epoch": 0.33149856996384436,
      "grad_norm": 1.1432520493471308,
      "learning_rate": 7.772425773304047e-06,
      "loss": 0.5964,
      "step": 6143
    },
    {
      "epoch": 0.33155253359235876,
      "grad_norm": 1.1356593622416467,
      "learning_rate": 7.771767265735019e-06,
      "loss": 0.467,
      "step": 6144
    },
    {
      "epoch": 0.33160649722087315,
      "grad_norm": 0.9889063966056117,
      "learning_rate": 7.771108692873106e-06,
      "loss": 0.4611,
      "step": 6145
    },
    {
      "epoch": 0.3316604608493875,
      "grad_norm": 1.175240397130465,
      "learning_rate": 7.770450054737234e-06,
      "loss": 0.6055,
      "step": 6146
    },
    {
      "epoch": 0.3317144244779019,
      "grad_norm": 0.9047121406246202,
      "learning_rate": 7.769791351346332e-06,
      "loss": 0.3788,
      "step": 6147
    },
    {
      "epoch": 0.3317683881064163,
      "grad_norm": 0.8664363420739913,
      "learning_rate": 7.769132582719334e-06,
      "loss": 0.3846,
      "step": 6148
    },
    {
      "epoch": 0.33182235173493063,
      "grad_norm": 1.2441924006650231,
      "learning_rate": 7.768473748875174e-06,
      "loss": 0.6713,
      "step": 6149
    },
    {
      "epoch": 0.33187631536344503,
      "grad_norm": 1.2550275352176894,
      "learning_rate": 7.767814849832785e-06,
      "loss": 0.6879,
      "step": 6150
    },
    {
      "epoch": 0.3319302789919594,
      "grad_norm": 0.7888016212071025,
      "learning_rate": 7.767155885611105e-06,
      "loss": 0.3727,
      "step": 6151
    },
    {
      "epoch": 0.3319842426204738,
      "grad_norm": 0.8147802624082152,
      "learning_rate": 7.766496856229076e-06,
      "loss": 0.3738,
      "step": 6152
    },
    {
      "epoch": 0.33203820624898817,
      "grad_norm": 1.0351181504952358,
      "learning_rate": 7.765837761705636e-06,
      "loss": 0.5501,
      "step": 6153
    },
    {
      "epoch": 0.33209216987750256,
      "grad_norm": 1.2782201053461395,
      "learning_rate": 7.76517860205973e-06,
      "loss": 0.6329,
      "step": 6154
    },
    {
      "epoch": 0.33214613350601696,
      "grad_norm": 0.8739673044433726,
      "learning_rate": 7.7645193773103e-06,
      "loss": 0.3975,
      "step": 6155
    },
    {
      "epoch": 0.3322000971345313,
      "grad_norm": 1.0526898003005154,
      "learning_rate": 7.763860087476299e-06,
      "loss": 0.4815,
      "step": 6156
    },
    {
      "epoch": 0.3322540607630457,
      "grad_norm": 1.0634299284524442,
      "learning_rate": 7.763200732576667e-06,
      "loss": 0.5085,
      "step": 6157
    },
    {
      "epoch": 0.3323080243915601,
      "grad_norm": 1.0657653731444394,
      "learning_rate": 7.762541312630362e-06,
      "loss": 0.5456,
      "step": 6158
    },
    {
      "epoch": 0.3323619880200745,
      "grad_norm": 1.0837744262351066,
      "learning_rate": 7.76188182765633e-06,
      "loss": 0.5756,
      "step": 6159
    },
    {
      "epoch": 0.33241595164858884,
      "grad_norm": 0.9058608486351204,
      "learning_rate": 7.761222277673534e-06,
      "loss": 0.5253,
      "step": 6160
    },
    {
      "epoch": 0.33246991527710323,
      "grad_norm": 0.9383867067102744,
      "learning_rate": 7.760562662700922e-06,
      "loss": 0.393,
      "step": 6161
    },
    {
      "epoch": 0.33252387890561763,
      "grad_norm": 1.0754609082770223,
      "learning_rate": 7.759902982757455e-06,
      "loss": 0.5065,
      "step": 6162
    },
    {
      "epoch": 0.332577842534132,
      "grad_norm": 1.1052932132644804,
      "learning_rate": 7.759243237862094e-06,
      "loss": 0.5199,
      "step": 6163
    },
    {
      "epoch": 0.33263180616264637,
      "grad_norm": 0.9241645044762838,
      "learning_rate": 7.758583428033798e-06,
      "loss": 0.3702,
      "step": 6164
    },
    {
      "epoch": 0.33268576979116077,
      "grad_norm": 1.1045685198886126,
      "learning_rate": 7.757923553291534e-06,
      "loss": 0.7153,
      "step": 6165
    },
    {
      "epoch": 0.33273973341967517,
      "grad_norm": 1.0452794996703634,
      "learning_rate": 7.757263613654264e-06,
      "loss": 0.5346,
      "step": 6166
    },
    {
      "epoch": 0.3327936970481895,
      "grad_norm": 0.7905622623354926,
      "learning_rate": 7.756603609140958e-06,
      "loss": 0.4622,
      "step": 6167
    },
    {
      "epoch": 0.3328476606767039,
      "grad_norm": 1.1683529243915605,
      "learning_rate": 7.755943539770583e-06,
      "loss": 0.5473,
      "step": 6168
    },
    {
      "epoch": 0.3329016243052183,
      "grad_norm": 0.9113725120770961,
      "learning_rate": 7.755283405562111e-06,
      "loss": 0.5397,
      "step": 6169
    },
    {
      "epoch": 0.33295558793373264,
      "grad_norm": 0.9219700550871036,
      "learning_rate": 7.754623206534517e-06,
      "loss": 0.4165,
      "step": 6170
    },
    {
      "epoch": 0.33300955156224704,
      "grad_norm": 0.9884210014034804,
      "learning_rate": 7.753962942706774e-06,
      "loss": 0.466,
      "step": 6171
    },
    {
      "epoch": 0.33306351519076144,
      "grad_norm": 1.168377542711775,
      "learning_rate": 7.753302614097856e-06,
      "loss": 0.5802,
      "step": 6172
    },
    {
      "epoch": 0.3331174788192758,
      "grad_norm": 0.9469792470858159,
      "learning_rate": 7.752642220726746e-06,
      "loss": 0.4147,
      "step": 6173
    },
    {
      "epoch": 0.3331714424477902,
      "grad_norm": 1.0928192095907618,
      "learning_rate": 7.751981762612422e-06,
      "loss": 0.4672,
      "step": 6174
    },
    {
      "epoch": 0.3332254060763046,
      "grad_norm": 1.053552205707948,
      "learning_rate": 7.751321239773867e-06,
      "loss": 0.552,
      "step": 6175
    },
    {
      "epoch": 0.333279369704819,
      "grad_norm": 1.151135806094273,
      "learning_rate": 7.750660652230066e-06,
      "loss": 0.5742,
      "step": 6176
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.9427588102285339,
      "learning_rate": 7.75e-06,
      "loss": 0.4847,
      "step": 6177
    },
    {
      "epoch": 0.3333872969618477,
      "grad_norm": 0.9781882049791273,
      "learning_rate": 7.749339283102664e-06,
      "loss": 0.3721,
      "step": 6178
    },
    {
      "epoch": 0.3334412605903621,
      "grad_norm": 1.1147208162108093,
      "learning_rate": 7.748678501557042e-06,
      "loss": 0.5947,
      "step": 6179
    },
    {
      "epoch": 0.33349522421887645,
      "grad_norm": 0.9711244454710616,
      "learning_rate": 7.74801765538213e-06,
      "loss": 0.4781,
      "step": 6180
    },
    {
      "epoch": 0.33354918784739085,
      "grad_norm": 1.1561837661608232,
      "learning_rate": 7.747356744596917e-06,
      "loss": 0.5519,
      "step": 6181
    },
    {
      "epoch": 0.33360315147590525,
      "grad_norm": 0.9588564490847111,
      "learning_rate": 7.7466957692204e-06,
      "loss": 0.3841,
      "step": 6182
    },
    {
      "epoch": 0.33365711510441964,
      "grad_norm": 1.0385495945361292,
      "learning_rate": 7.74603472927158e-06,
      "loss": 0.444,
      "step": 6183
    },
    {
      "epoch": 0.333711078732934,
      "grad_norm": 1.1460838471311547,
      "learning_rate": 7.74537362476945e-06,
      "loss": 0.5644,
      "step": 6184
    },
    {
      "epoch": 0.3337650423614484,
      "grad_norm": 1.162062961537598,
      "learning_rate": 7.744712455733013e-06,
      "loss": 0.5541,
      "step": 6185
    },
    {
      "epoch": 0.3338190059899628,
      "grad_norm": 1.1094366700974185,
      "learning_rate": 7.74405122218127e-06,
      "loss": 0.548,
      "step": 6186
    },
    {
      "epoch": 0.3338729696184771,
      "grad_norm": 0.8991752029372594,
      "learning_rate": 7.743389924133233e-06,
      "loss": 0.4114,
      "step": 6187
    },
    {
      "epoch": 0.3339269332469915,
      "grad_norm": 0.9009820680520623,
      "learning_rate": 7.742728561607899e-06,
      "loss": 0.4457,
      "step": 6188
    },
    {
      "epoch": 0.3339808968755059,
      "grad_norm": 1.0331677867944369,
      "learning_rate": 7.74206713462428e-06,
      "loss": 0.4025,
      "step": 6189
    },
    {
      "epoch": 0.3340348605040203,
      "grad_norm": 1.047441014607124,
      "learning_rate": 7.741405643201386e-06,
      "loss": 0.5374,
      "step": 6190
    },
    {
      "epoch": 0.33408882413253466,
      "grad_norm": 0.9205806518078077,
      "learning_rate": 7.74074408735823e-06,
      "loss": 0.4519,
      "step": 6191
    },
    {
      "epoch": 0.33414278776104905,
      "grad_norm": 1.1531118003245446,
      "learning_rate": 7.740082467113825e-06,
      "loss": 0.5055,
      "step": 6192
    },
    {
      "epoch": 0.33419675138956345,
      "grad_norm": 1.2584658364447374,
      "learning_rate": 7.739420782487186e-06,
      "loss": 0.6092,
      "step": 6193
    },
    {
      "epoch": 0.3342507150180778,
      "grad_norm": 1.0165848053520155,
      "learning_rate": 7.73875903349733e-06,
      "loss": 0.5802,
      "step": 6194
    },
    {
      "epoch": 0.3343046786465922,
      "grad_norm": 0.8266691740490584,
      "learning_rate": 7.738097220163277e-06,
      "loss": 0.4005,
      "step": 6195
    },
    {
      "epoch": 0.3343586422751066,
      "grad_norm": 0.9807061306459408,
      "learning_rate": 7.737435342504051e-06,
      "loss": 0.5693,
      "step": 6196
    },
    {
      "epoch": 0.334412605903621,
      "grad_norm": 1.1439539392576559,
      "learning_rate": 7.73677340053867e-06,
      "loss": 0.5825,
      "step": 6197
    },
    {
      "epoch": 0.3344665695321353,
      "grad_norm": 1.3212221441915841,
      "learning_rate": 7.736111394286163e-06,
      "loss": 0.7168,
      "step": 6198
    },
    {
      "epoch": 0.3345205331606497,
      "grad_norm": 1.1349413721010104,
      "learning_rate": 7.735449323765555e-06,
      "loss": 0.4282,
      "step": 6199
    },
    {
      "epoch": 0.3345744967891641,
      "grad_norm": 0.8921006412308018,
      "learning_rate": 7.734787188995873e-06,
      "loss": 0.3258,
      "step": 6200
    },
    {
      "epoch": 0.33462846041767846,
      "grad_norm": 1.0736174761664117,
      "learning_rate": 7.73412498999615e-06,
      "loss": 0.5256,
      "step": 6201
    },
    {
      "epoch": 0.33468242404619286,
      "grad_norm": 1.0787289276179206,
      "learning_rate": 7.73346272678542e-06,
      "loss": 0.6061,
      "step": 6202
    },
    {
      "epoch": 0.33473638767470726,
      "grad_norm": 1.0814675129678235,
      "learning_rate": 7.73280039938271e-06,
      "loss": 0.5207,
      "step": 6203
    },
    {
      "epoch": 0.33479035130322166,
      "grad_norm": 1.131049104851575,
      "learning_rate": 7.732138007807064e-06,
      "loss": 0.5389,
      "step": 6204
    },
    {
      "epoch": 0.334844314931736,
      "grad_norm": 1.2451629938135993,
      "learning_rate": 7.731475552077515e-06,
      "loss": 0.5252,
      "step": 6205
    },
    {
      "epoch": 0.3348982785602504,
      "grad_norm": 0.7658844056184926,
      "learning_rate": 7.730813032213103e-06,
      "loss": 0.3439,
      "step": 6206
    },
    {
      "epoch": 0.3349522421887648,
      "grad_norm": 0.9075539313989543,
      "learning_rate": 7.730150448232872e-06,
      "loss": 0.524,
      "step": 6207
    },
    {
      "epoch": 0.33500620581727913,
      "grad_norm": 1.0886275514523147,
      "learning_rate": 7.729487800155864e-06,
      "loss": 0.4539,
      "step": 6208
    },
    {
      "epoch": 0.33506016944579353,
      "grad_norm": 1.1341538757109026,
      "learning_rate": 7.728825088001122e-06,
      "loss": 0.536,
      "step": 6209
    },
    {
      "epoch": 0.33511413307430793,
      "grad_norm": 0.9472875223922648,
      "learning_rate": 7.728162311787696e-06,
      "loss": 0.4728,
      "step": 6210
    },
    {
      "epoch": 0.33516809670282227,
      "grad_norm": 1.1595884962346827,
      "learning_rate": 7.727499471534635e-06,
      "loss": 0.3942,
      "step": 6211
    },
    {
      "epoch": 0.33522206033133667,
      "grad_norm": 1.0454994808364215,
      "learning_rate": 7.726836567260986e-06,
      "loss": 0.4795,
      "step": 6212
    },
    {
      "epoch": 0.33527602395985107,
      "grad_norm": 0.9659529030000801,
      "learning_rate": 7.726173598985803e-06,
      "loss": 0.4252,
      "step": 6213
    },
    {
      "epoch": 0.33532998758836546,
      "grad_norm": 0.9990540779486091,
      "learning_rate": 7.725510566728145e-06,
      "loss": 0.4897,
      "step": 6214
    },
    {
      "epoch": 0.3353839512168798,
      "grad_norm": 1.1385907335869565,
      "learning_rate": 7.724847470507062e-06,
      "loss": 0.5397,
      "step": 6215
    },
    {
      "epoch": 0.3354379148453942,
      "grad_norm": 0.9090975220667553,
      "learning_rate": 7.724184310341614e-06,
      "loss": 0.4305,
      "step": 6216
    },
    {
      "epoch": 0.3354918784739086,
      "grad_norm": 0.9399110094263099,
      "learning_rate": 7.723521086250865e-06,
      "loss": 0.4619,
      "step": 6217
    },
    {
      "epoch": 0.33554584210242294,
      "grad_norm": 1.2387712828669781,
      "learning_rate": 7.72285779825387e-06,
      "loss": 0.5123,
      "step": 6218
    },
    {
      "epoch": 0.33559980573093734,
      "grad_norm": 1.0545356541682946,
      "learning_rate": 7.722194446369698e-06,
      "loss": 0.5675,
      "step": 6219
    },
    {
      "epoch": 0.33565376935945174,
      "grad_norm": 1.0968888939082861,
      "learning_rate": 7.721531030617412e-06,
      "loss": 0.5025,
      "step": 6220
    },
    {
      "epoch": 0.33570773298796613,
      "grad_norm": 1.0857646422151292,
      "learning_rate": 7.720867551016078e-06,
      "loss": 0.4928,
      "step": 6221
    },
    {
      "epoch": 0.3357616966164805,
      "grad_norm": 0.9409480017119642,
      "learning_rate": 7.720204007584769e-06,
      "loss": 0.4365,
      "step": 6222
    },
    {
      "epoch": 0.3358156602449949,
      "grad_norm": 1.1023841549887154,
      "learning_rate": 7.71954040034255e-06,
      "loss": 0.4462,
      "step": 6223
    },
    {
      "epoch": 0.33586962387350927,
      "grad_norm": 0.8566536236688128,
      "learning_rate": 7.7188767293085e-06,
      "loss": 0.3889,
      "step": 6224
    },
    {
      "epoch": 0.3359235875020236,
      "grad_norm": 0.9221969152580449,
      "learning_rate": 7.71821299450169e-06,
      "loss": 0.4176,
      "step": 6225
    },
    {
      "epoch": 0.335977551130538,
      "grad_norm": 1.131641950374379,
      "learning_rate": 7.717549195941197e-06,
      "loss": 0.5448,
      "step": 6226
    },
    {
      "epoch": 0.3360315147590524,
      "grad_norm": 1.0508744107354993,
      "learning_rate": 7.7168853336461e-06,
      "loss": 0.5044,
      "step": 6227
    },
    {
      "epoch": 0.3360854783875668,
      "grad_norm": 1.1458849771886326,
      "learning_rate": 7.716221407635476e-06,
      "loss": 0.794,
      "step": 6228
    },
    {
      "epoch": 0.33613944201608115,
      "grad_norm": 1.0531669851335395,
      "learning_rate": 7.715557417928414e-06,
      "loss": 0.483,
      "step": 6229
    },
    {
      "epoch": 0.33619340564459554,
      "grad_norm": 0.9396037936660145,
      "learning_rate": 7.714893364543989e-06,
      "loss": 0.4818,
      "step": 6230
    },
    {
      "epoch": 0.33624736927310994,
      "grad_norm": 1.0367994153004367,
      "learning_rate": 7.714229247501293e-06,
      "loss": 0.4946,
      "step": 6231
    },
    {
      "epoch": 0.3363013329016243,
      "grad_norm": 0.8949548810558738,
      "learning_rate": 7.713565066819411e-06,
      "loss": 0.3695,
      "step": 6232
    },
    {
      "epoch": 0.3363552965301387,
      "grad_norm": 1.1761031374673392,
      "learning_rate": 7.712900822517432e-06,
      "loss": 0.5905,
      "step": 6233
    },
    {
      "epoch": 0.3364092601586531,
      "grad_norm": 0.9355424164463529,
      "learning_rate": 7.712236514614448e-06,
      "loss": 0.4164,
      "step": 6234
    },
    {
      "epoch": 0.3364632237871675,
      "grad_norm": 1.1632342365180592,
      "learning_rate": 7.71157214312955e-06,
      "loss": 0.4868,
      "step": 6235
    },
    {
      "epoch": 0.3365171874156818,
      "grad_norm": 0.7976187222577481,
      "learning_rate": 7.710907708081836e-06,
      "loss": 0.2694,
      "step": 6236
    },
    {
      "epoch": 0.3365711510441962,
      "grad_norm": 0.8439531299283709,
      "learning_rate": 7.710243209490398e-06,
      "loss": 0.3382,
      "step": 6237
    },
    {
      "epoch": 0.3366251146727106,
      "grad_norm": 1.0469436978608435,
      "learning_rate": 7.709578647374337e-06,
      "loss": 0.4593,
      "step": 6238
    },
    {
      "epoch": 0.33667907830122495,
      "grad_norm": 0.920572205891937,
      "learning_rate": 7.708914021752753e-06,
      "loss": 0.3301,
      "step": 6239
    },
    {
      "epoch": 0.33673304192973935,
      "grad_norm": 1.546360424269349,
      "learning_rate": 7.70824933264475e-06,
      "loss": 0.6702,
      "step": 6240
    },
    {
      "epoch": 0.33678700555825375,
      "grad_norm": 0.9578930665605552,
      "learning_rate": 7.70758458006943e-06,
      "loss": 0.4949,
      "step": 6241
    },
    {
      "epoch": 0.3368409691867681,
      "grad_norm": 0.8931781657743159,
      "learning_rate": 7.706919764045896e-06,
      "loss": 0.3517,
      "step": 6242
    },
    {
      "epoch": 0.3368949328152825,
      "grad_norm": 1.0050605057702025,
      "learning_rate": 7.70625488459326e-06,
      "loss": 0.4211,
      "step": 6243
    },
    {
      "epoch": 0.3369488964437969,
      "grad_norm": 0.8907736225700217,
      "learning_rate": 7.705589941730627e-06,
      "loss": 0.4586,
      "step": 6244
    },
    {
      "epoch": 0.3370028600723113,
      "grad_norm": 1.1853935600242616,
      "learning_rate": 7.704924935477114e-06,
      "loss": 0.4748,
      "step": 6245
    },
    {
      "epoch": 0.3370568237008256,
      "grad_norm": 1.1082499456181796,
      "learning_rate": 7.704259865851828e-06,
      "loss": 0.5586,
      "step": 6246
    },
    {
      "epoch": 0.33711078732934,
      "grad_norm": 0.8452663073810047,
      "learning_rate": 7.703594732873887e-06,
      "loss": 0.3875,
      "step": 6247
    },
    {
      "epoch": 0.3371647509578544,
      "grad_norm": 1.2142050814123375,
      "learning_rate": 7.702929536562409e-06,
      "loss": 0.5466,
      "step": 6248
    },
    {
      "epoch": 0.33721871458636876,
      "grad_norm": 1.078643778358689,
      "learning_rate": 7.702264276936508e-06,
      "loss": 0.4407,
      "step": 6249
    },
    {
      "epoch": 0.33727267821488316,
      "grad_norm": 1.1167831486414013,
      "learning_rate": 7.701598954015306e-06,
      "loss": 0.4881,
      "step": 6250
    },
    {
      "epoch": 0.33732664184339756,
      "grad_norm": 1.2160464689199002,
      "learning_rate": 7.700933567817926e-06,
      "loss": 0.7051,
      "step": 6251
    },
    {
      "epoch": 0.33738060547191195,
      "grad_norm": 1.1510221384976584,
      "learning_rate": 7.70026811836349e-06,
      "loss": 0.5096,
      "step": 6252
    },
    {
      "epoch": 0.3374345691004263,
      "grad_norm": 1.2141072527163788,
      "learning_rate": 7.699602605671128e-06,
      "loss": 0.5092,
      "step": 6253
    },
    {
      "epoch": 0.3374885327289407,
      "grad_norm": 0.9539107797251729,
      "learning_rate": 7.698937029759964e-06,
      "loss": 0.6115,
      "step": 6254
    },
    {
      "epoch": 0.3375424963574551,
      "grad_norm": 1.072742963942722,
      "learning_rate": 7.698271390649126e-06,
      "loss": 0.6071,
      "step": 6255
    },
    {
      "epoch": 0.33759645998596943,
      "grad_norm": 1.289193193092255,
      "learning_rate": 7.697605688357748e-06,
      "loss": 0.647,
      "step": 6256
    },
    {
      "epoch": 0.33765042361448383,
      "grad_norm": 1.0025469314040807,
      "learning_rate": 7.696939922904962e-06,
      "loss": 0.4288,
      "step": 6257
    },
    {
      "epoch": 0.3377043872429982,
      "grad_norm": 1.1559997467984333,
      "learning_rate": 7.696274094309902e-06,
      "loss": 0.5281,
      "step": 6258
    },
    {
      "epoch": 0.3377583508715126,
      "grad_norm": 1.1146118335200494,
      "learning_rate": 7.695608202591705e-06,
      "loss": 0.5448,
      "step": 6259
    },
    {
      "epoch": 0.33781231450002697,
      "grad_norm": 1.2992741138294905,
      "learning_rate": 7.694942247769514e-06,
      "loss": 0.6642,
      "step": 6260
    },
    {
      "epoch": 0.33786627812854136,
      "grad_norm": 1.4254929804782697,
      "learning_rate": 7.694276229862461e-06,
      "loss": 0.6635,
      "step": 6261
    },
    {
      "epoch": 0.33792024175705576,
      "grad_norm": 0.9549017920870384,
      "learning_rate": 7.693610148889692e-06,
      "loss": 0.388,
      "step": 6262
    },
    {
      "epoch": 0.3379742053855701,
      "grad_norm": 1.138202996844142,
      "learning_rate": 7.692944004870352e-06,
      "loss": 0.5313,
      "step": 6263
    },
    {
      "epoch": 0.3380281690140845,
      "grad_norm": 0.9567849978964882,
      "learning_rate": 7.692277797823585e-06,
      "loss": 0.3199,
      "step": 6264
    },
    {
      "epoch": 0.3380821326425989,
      "grad_norm": 1.024072452779663,
      "learning_rate": 7.69161152776854e-06,
      "loss": 0.4674,
      "step": 6265
    },
    {
      "epoch": 0.3381360962711133,
      "grad_norm": 0.7157583320499319,
      "learning_rate": 7.690945194724365e-06,
      "loss": 0.2732,
      "step": 6266
    },
    {
      "epoch": 0.33819005989962764,
      "grad_norm": 0.8954539961358388,
      "learning_rate": 7.690278798710209e-06,
      "loss": 0.3866,
      "step": 6267
    },
    {
      "epoch": 0.33824402352814203,
      "grad_norm": 0.8744872246341053,
      "learning_rate": 7.689612339745228e-06,
      "loss": 0.4,
      "step": 6268
    },
    {
      "epoch": 0.33829798715665643,
      "grad_norm": 0.7116845345196735,
      "learning_rate": 7.688945817848578e-06,
      "loss": 0.2993,
      "step": 6269
    },
    {
      "epoch": 0.3383519507851708,
      "grad_norm": 0.7845956396475425,
      "learning_rate": 7.688279233039413e-06,
      "loss": 0.2657,
      "step": 6270
    },
    {
      "epoch": 0.33840591441368517,
      "grad_norm": 1.1770705891807658,
      "learning_rate": 7.68761258533689e-06,
      "loss": 0.6383,
      "step": 6271
    },
    {
      "epoch": 0.33845987804219957,
      "grad_norm": 0.9761786362237939,
      "learning_rate": 7.686945874760172e-06,
      "loss": 0.418,
      "step": 6272
    },
    {
      "epoch": 0.33851384167071397,
      "grad_norm": 0.7805130628555754,
      "learning_rate": 7.686279101328419e-06,
      "loss": 0.4591,
      "step": 6273
    },
    {
      "epoch": 0.3385678052992283,
      "grad_norm": 0.9493394770658493,
      "learning_rate": 7.685612265060798e-06,
      "loss": 0.3988,
      "step": 6274
    },
    {
      "epoch": 0.3386217689277427,
      "grad_norm": 1.125565518828023,
      "learning_rate": 7.68494536597647e-06,
      "loss": 0.4915,
      "step": 6275
    },
    {
      "epoch": 0.3386757325562571,
      "grad_norm": 0.9511463352002476,
      "learning_rate": 7.684278404094605e-06,
      "loss": 0.4452,
      "step": 6276
    },
    {
      "epoch": 0.33872969618477144,
      "grad_norm": 1.1563436443726094,
      "learning_rate": 7.68361137943437e-06,
      "loss": 0.5193,
      "step": 6277
    },
    {
      "epoch": 0.33878365981328584,
      "grad_norm": 1.0792499628518062,
      "learning_rate": 7.682944292014938e-06,
      "loss": 0.5096,
      "step": 6278
    },
    {
      "epoch": 0.33883762344180024,
      "grad_norm": 1.0937024864006741,
      "learning_rate": 7.682277141855484e-06,
      "loss": 0.4247,
      "step": 6279
    },
    {
      "epoch": 0.3388915870703146,
      "grad_norm": 0.8723439174311844,
      "learning_rate": 7.681609928975178e-06,
      "loss": 0.441,
      "step": 6280
    },
    {
      "epoch": 0.338945550698829,
      "grad_norm": 0.9670619914367017,
      "learning_rate": 7.680942653393199e-06,
      "loss": 0.4241,
      "step": 6281
    },
    {
      "epoch": 0.3389995143273434,
      "grad_norm": 0.9456894080266762,
      "learning_rate": 7.680275315128723e-06,
      "loss": 0.5151,
      "step": 6282
    },
    {
      "epoch": 0.3390534779558578,
      "grad_norm": 0.914106705642595,
      "learning_rate": 7.679607914200933e-06,
      "loss": 0.378,
      "step": 6283
    },
    {
      "epoch": 0.3391074415843721,
      "grad_norm": 1.0857173479311941,
      "learning_rate": 7.678940450629009e-06,
      "loss": 0.5901,
      "step": 6284
    },
    {
      "epoch": 0.3391614052128865,
      "grad_norm": 1.043204465634035,
      "learning_rate": 7.678272924432134e-06,
      "loss": 0.5068,
      "step": 6285
    },
    {
      "epoch": 0.3392153688414009,
      "grad_norm": 0.9517043860576055,
      "learning_rate": 7.677605335629495e-06,
      "loss": 0.4279,
      "step": 6286
    },
    {
      "epoch": 0.33926933246991525,
      "grad_norm": 0.9713022561636512,
      "learning_rate": 7.676937684240278e-06,
      "loss": 0.5305,
      "step": 6287
    },
    {
      "epoch": 0.33932329609842965,
      "grad_norm": 0.8537366096349396,
      "learning_rate": 7.676269970283671e-06,
      "loss": 0.4337,
      "step": 6288
    },
    {
      "epoch": 0.33937725972694405,
      "grad_norm": 0.9047815139704588,
      "learning_rate": 7.675602193778868e-06,
      "loss": 0.3887,
      "step": 6289
    },
    {
      "epoch": 0.33943122335545844,
      "grad_norm": 0.973979252588054,
      "learning_rate": 7.67493435474506e-06,
      "loss": 0.5091,
      "step": 6290
    },
    {
      "epoch": 0.3394851869839728,
      "grad_norm": 1.0036115990364909,
      "learning_rate": 7.674266453201439e-06,
      "loss": 0.595,
      "step": 6291
    },
    {
      "epoch": 0.3395391506124872,
      "grad_norm": 0.910612619538208,
      "learning_rate": 7.673598489167204e-06,
      "loss": 0.4023,
      "step": 6292
    },
    {
      "epoch": 0.3395931142410016,
      "grad_norm": 0.9692480446749643,
      "learning_rate": 7.672930462661552e-06,
      "loss": 0.4907,
      "step": 6293
    },
    {
      "epoch": 0.3396470778695159,
      "grad_norm": 1.0176595291929587,
      "learning_rate": 7.672262373703684e-06,
      "loss": 0.5278,
      "step": 6294
    },
    {
      "epoch": 0.3397010414980303,
      "grad_norm": 0.8389376156128697,
      "learning_rate": 7.6715942223128e-06,
      "loss": 0.3492,
      "step": 6295
    },
    {
      "epoch": 0.3397550051265447,
      "grad_norm": 1.0074580725478366,
      "learning_rate": 7.670926008508103e-06,
      "loss": 0.6242,
      "step": 6296
    },
    {
      "epoch": 0.3398089687550591,
      "grad_norm": 1.2622812273116983,
      "learning_rate": 7.670257732308798e-06,
      "loss": 0.591,
      "step": 6297
    },
    {
      "epoch": 0.33986293238357346,
      "grad_norm": 0.8862328220889185,
      "learning_rate": 7.669589393734096e-06,
      "loss": 0.5103,
      "step": 6298
    },
    {
      "epoch": 0.33991689601208785,
      "grad_norm": 1.0370404965270081,
      "learning_rate": 7.668920992803201e-06,
      "loss": 0.4119,
      "step": 6299
    },
    {
      "epoch": 0.33997085964060225,
      "grad_norm": 0.9787020411238342,
      "learning_rate": 7.668252529535325e-06,
      "loss": 0.5462,
      "step": 6300
    },
    {
      "epoch": 0.3400248232691166,
      "grad_norm": 1.0101759048242935,
      "learning_rate": 7.667584003949681e-06,
      "loss": 0.454,
      "step": 6301
    },
    {
      "epoch": 0.340078786897631,
      "grad_norm": 1.0867862169794416,
      "learning_rate": 7.666915416065483e-06,
      "loss": 0.6773,
      "step": 6302
    },
    {
      "epoch": 0.3401327505261454,
      "grad_norm": 1.0849707605236367,
      "learning_rate": 7.666246765901945e-06,
      "loss": 0.508,
      "step": 6303
    },
    {
      "epoch": 0.3401867141546598,
      "grad_norm": 0.9020795056785497,
      "learning_rate": 7.665578053478287e-06,
      "loss": 0.4128,
      "step": 6304
    },
    {
      "epoch": 0.34024067778317413,
      "grad_norm": 1.0632992691286802,
      "learning_rate": 7.664909278813727e-06,
      "loss": 0.4166,
      "step": 6305
    },
    {
      "epoch": 0.3402946414116885,
      "grad_norm": 1.1723284530761957,
      "learning_rate": 7.664240441927487e-06,
      "loss": 0.5434,
      "step": 6306
    },
    {
      "epoch": 0.3403486050402029,
      "grad_norm": 0.8624393610562399,
      "learning_rate": 7.663571542838789e-06,
      "loss": 0.5296,
      "step": 6307
    },
    {
      "epoch": 0.34040256866871726,
      "grad_norm": 1.0564669912287514,
      "learning_rate": 7.662902581566863e-06,
      "loss": 0.4807,
      "step": 6308
    },
    {
      "epoch": 0.34045653229723166,
      "grad_norm": 1.112544418472094,
      "learning_rate": 7.662233558130927e-06,
      "loss": 0.4677,
      "step": 6309
    },
    {
      "epoch": 0.34051049592574606,
      "grad_norm": 1.0163153145371064,
      "learning_rate": 7.661564472550214e-06,
      "loss": 0.527,
      "step": 6310
    },
    {
      "epoch": 0.3405644595542604,
      "grad_norm": 0.9668505684434294,
      "learning_rate": 7.660895324843957e-06,
      "loss": 0.4929,
      "step": 6311
    },
    {
      "epoch": 0.3406184231827748,
      "grad_norm": 1.2603419245206888,
      "learning_rate": 7.660226115031382e-06,
      "loss": 0.6693,
      "step": 6312
    },
    {
      "epoch": 0.3406723868112892,
      "grad_norm": 0.9117661213439325,
      "learning_rate": 7.659556843131725e-06,
      "loss": 0.4289,
      "step": 6313
    },
    {
      "epoch": 0.3407263504398036,
      "grad_norm": 1.3583767620233498,
      "learning_rate": 7.658887509164224e-06,
      "loss": 0.6548,
      "step": 6314
    },
    {
      "epoch": 0.34078031406831794,
      "grad_norm": 1.1418352331171917,
      "learning_rate": 7.658218113148114e-06,
      "loss": 0.4595,
      "step": 6315
    },
    {
      "epoch": 0.34083427769683233,
      "grad_norm": 1.0193413813616137,
      "learning_rate": 7.657548655102636e-06,
      "loss": 0.5823,
      "step": 6316
    },
    {
      "epoch": 0.34088824132534673,
      "grad_norm": 0.7149267136920753,
      "learning_rate": 7.656879135047027e-06,
      "loss": 0.3041,
      "step": 6317
    },
    {
      "epoch": 0.34094220495386107,
      "grad_norm": 1.046913567743191,
      "learning_rate": 7.656209553000534e-06,
      "loss": 0.4253,
      "step": 6318
    },
    {
      "epoch": 0.34099616858237547,
      "grad_norm": 1.3113168800793509,
      "learning_rate": 7.6555399089824e-06,
      "loss": 0.4963,
      "step": 6319
    },
    {
      "epoch": 0.34105013221088987,
      "grad_norm": 1.138870550778005,
      "learning_rate": 7.654870203011871e-06,
      "loss": 0.5497,
      "step": 6320
    },
    {
      "epoch": 0.34110409583940426,
      "grad_norm": 0.8557430901305412,
      "learning_rate": 7.654200435108194e-06,
      "loss": 0.4814,
      "step": 6321
    },
    {
      "epoch": 0.3411580594679186,
      "grad_norm": 0.8848039623278529,
      "learning_rate": 7.653530605290623e-06,
      "loss": 0.4062,
      "step": 6322
    },
    {
      "epoch": 0.341212023096433,
      "grad_norm": 1.4748674041503242,
      "learning_rate": 7.652860713578402e-06,
      "loss": 0.6646,
      "step": 6323
    },
    {
      "epoch": 0.3412659867249474,
      "grad_norm": 1.0496727986508256,
      "learning_rate": 7.65219075999079e-06,
      "loss": 0.6126,
      "step": 6324
    },
    {
      "epoch": 0.34131995035346174,
      "grad_norm": 1.071841800651816,
      "learning_rate": 7.651520744547044e-06,
      "loss": 0.546,
      "step": 6325
    },
    {
      "epoch": 0.34137391398197614,
      "grad_norm": 0.9689676452905024,
      "learning_rate": 7.650850667266414e-06,
      "loss": 0.5725,
      "step": 6326
    },
    {
      "epoch": 0.34142787761049054,
      "grad_norm": 1.0099071982322663,
      "learning_rate": 7.650180528168164e-06,
      "loss": 0.5473,
      "step": 6327
    },
    {
      "epoch": 0.34148184123900494,
      "grad_norm": 1.1909987987018726,
      "learning_rate": 7.649510327271553e-06,
      "loss": 0.5555,
      "step": 6328
    },
    {
      "epoch": 0.3415358048675193,
      "grad_norm": 0.9131746471027308,
      "learning_rate": 7.648840064595844e-06,
      "loss": 0.3918,
      "step": 6329
    },
    {
      "epoch": 0.3415897684960337,
      "grad_norm": 0.7974986153292442,
      "learning_rate": 7.6481697401603e-06,
      "loss": 0.3634,
      "step": 6330
    },
    {
      "epoch": 0.34164373212454807,
      "grad_norm": 0.8314613240049529,
      "learning_rate": 7.647499353984188e-06,
      "loss": 0.4014,
      "step": 6331
    },
    {
      "epoch": 0.3416976957530624,
      "grad_norm": 1.0361843510210693,
      "learning_rate": 7.646828906086772e-06,
      "loss": 0.4472,
      "step": 6332
    },
    {
      "epoch": 0.3417516593815768,
      "grad_norm": 0.9235053291665339,
      "learning_rate": 7.646158396487327e-06,
      "loss": 0.4443,
      "step": 6333
    },
    {
      "epoch": 0.3418056230100912,
      "grad_norm": 1.00072126004096,
      "learning_rate": 7.645487825205118e-06,
      "loss": 0.4256,
      "step": 6334
    },
    {
      "epoch": 0.3418595866386056,
      "grad_norm": 1.142601042215496,
      "learning_rate": 7.644817192259423e-06,
      "loss": 0.4915,
      "step": 6335
    },
    {
      "epoch": 0.34191355026711995,
      "grad_norm": 0.8576856261709185,
      "learning_rate": 7.644146497669515e-06,
      "loss": 0.5452,
      "step": 6336
    },
    {
      "epoch": 0.34196751389563435,
      "grad_norm": 1.149481291781249,
      "learning_rate": 7.643475741454668e-06,
      "loss": 0.5219,
      "step": 6337
    },
    {
      "epoch": 0.34202147752414874,
      "grad_norm": 1.0423935225361187,
      "learning_rate": 7.642804923634165e-06,
      "loss": 0.4498,
      "step": 6338
    },
    {
      "epoch": 0.3420754411526631,
      "grad_norm": 0.9969054322548437,
      "learning_rate": 7.642134044227282e-06,
      "loss": 0.3759,
      "step": 6339
    },
    {
      "epoch": 0.3421294047811775,
      "grad_norm": 0.987731582957206,
      "learning_rate": 7.641463103253304e-06,
      "loss": 0.4883,
      "step": 6340
    },
    {
      "epoch": 0.3421833684096919,
      "grad_norm": 1.0382894479850697,
      "learning_rate": 7.640792100731509e-06,
      "loss": 0.5929,
      "step": 6341
    },
    {
      "epoch": 0.3422373320382063,
      "grad_norm": 1.1907527047385358,
      "learning_rate": 7.640121036681188e-06,
      "loss": 0.5691,
      "step": 6342
    },
    {
      "epoch": 0.3422912956667206,
      "grad_norm": 1.0946507105809948,
      "learning_rate": 7.639449911121626e-06,
      "loss": 0.504,
      "step": 6343
    },
    {
      "epoch": 0.342345259295235,
      "grad_norm": 0.8565754923633495,
      "learning_rate": 7.638778724072112e-06,
      "loss": 0.3902,
      "step": 6344
    },
    {
      "epoch": 0.3423992229237494,
      "grad_norm": 1.0617283960807233,
      "learning_rate": 7.638107475551935e-06,
      "loss": 0.4494,
      "step": 6345
    },
    {
      "epoch": 0.34245318655226376,
      "grad_norm": 0.910302382480241,
      "learning_rate": 7.63743616558039e-06,
      "loss": 0.3844,
      "step": 6346
    },
    {
      "epoch": 0.34250715018077815,
      "grad_norm": 0.9198534408101244,
      "learning_rate": 7.63676479417677e-06,
      "loss": 0.5034,
      "step": 6347
    },
    {
      "epoch": 0.34256111380929255,
      "grad_norm": 1.0720400159709258,
      "learning_rate": 7.63609336136037e-06,
      "loss": 0.568,
      "step": 6348
    },
    {
      "epoch": 0.3426150774378069,
      "grad_norm": 1.0838660416382222,
      "learning_rate": 7.63542186715049e-06,
      "loss": 0.5605,
      "step": 6349
    },
    {
      "epoch": 0.3426690410663213,
      "grad_norm": 1.0181251805640879,
      "learning_rate": 7.634750311566426e-06,
      "loss": 0.5865,
      "step": 6350
    },
    {
      "epoch": 0.3427230046948357,
      "grad_norm": 1.2847340280195056,
      "learning_rate": 7.634078694627483e-06,
      "loss": 0.6857,
      "step": 6351
    },
    {
      "epoch": 0.3427769683233501,
      "grad_norm": 0.9259508363518663,
      "learning_rate": 7.63340701635296e-06,
      "loss": 0.4335,
      "step": 6352
    },
    {
      "epoch": 0.3428309319518644,
      "grad_norm": 1.0443402344006814,
      "learning_rate": 7.632735276762166e-06,
      "loss": 0.5123,
      "step": 6353
    },
    {
      "epoch": 0.3428848955803788,
      "grad_norm": 0.8426121279675487,
      "learning_rate": 7.632063475874402e-06,
      "loss": 0.3425,
      "step": 6354
    },
    {
      "epoch": 0.3429388592088932,
      "grad_norm": 1.1210732642770516,
      "learning_rate": 7.631391613708982e-06,
      "loss": 0.5385,
      "step": 6355
    },
    {
      "epoch": 0.34299282283740756,
      "grad_norm": 0.9400363009568219,
      "learning_rate": 7.630719690285213e-06,
      "loss": 0.4866,
      "step": 6356
    },
    {
      "epoch": 0.34304678646592196,
      "grad_norm": 0.9403605898886146,
      "learning_rate": 7.630047705622406e-06,
      "loss": 0.4823,
      "step": 6357
    },
    {
      "epoch": 0.34310075009443636,
      "grad_norm": 1.1084938714531694,
      "learning_rate": 7.629375659739875e-06,
      "loss": 0.4763,
      "step": 6358
    },
    {
      "epoch": 0.34315471372295075,
      "grad_norm": 0.981265325152567,
      "learning_rate": 7.628703552656937e-06,
      "loss": 0.4047,
      "step": 6359
    },
    {
      "epoch": 0.3432086773514651,
      "grad_norm": 1.0064003155348655,
      "learning_rate": 7.628031384392909e-06,
      "loss": 0.4409,
      "step": 6360
    },
    {
      "epoch": 0.3432626409799795,
      "grad_norm": 1.1013551358252252,
      "learning_rate": 7.627359154967109e-06,
      "loss": 0.4745,
      "step": 6361
    },
    {
      "epoch": 0.3433166046084939,
      "grad_norm": 0.8141790756362646,
      "learning_rate": 7.626686864398856e-06,
      "loss": 0.4019,
      "step": 6362
    },
    {
      "epoch": 0.34337056823700823,
      "grad_norm": 1.3989751451926509,
      "learning_rate": 7.6260145127074746e-06,
      "loss": 0.5176,
      "step": 6363
    },
    {
      "epoch": 0.34342453186552263,
      "grad_norm": 0.874110896221824,
      "learning_rate": 7.6253420999122875e-06,
      "loss": 0.3832,
      "step": 6364
    },
    {
      "epoch": 0.34347849549403703,
      "grad_norm": 1.0114588233995112,
      "learning_rate": 7.624669626032621e-06,
      "loss": 0.5306,
      "step": 6365
    },
    {
      "epoch": 0.3435324591225514,
      "grad_norm": 0.982987149687826,
      "learning_rate": 7.623997091087802e-06,
      "loss": 0.4903,
      "step": 6366
    },
    {
      "epoch": 0.34358642275106577,
      "grad_norm": 0.8828418394548824,
      "learning_rate": 7.623324495097162e-06,
      "loss": 0.5846,
      "step": 6367
    },
    {
      "epoch": 0.34364038637958016,
      "grad_norm": 0.9420449875319705,
      "learning_rate": 7.6226518380800275e-06,
      "loss": 0.4316,
      "step": 6368
    },
    {
      "epoch": 0.34369435000809456,
      "grad_norm": 0.9188101349918704,
      "learning_rate": 7.621979120055737e-06,
      "loss": 0.5859,
      "step": 6369
    },
    {
      "epoch": 0.3437483136366089,
      "grad_norm": 1.0419763229538284,
      "learning_rate": 7.6213063410436216e-06,
      "loss": 0.4107,
      "step": 6370
    },
    {
      "epoch": 0.3438022772651233,
      "grad_norm": 1.199786812127711,
      "learning_rate": 7.620633501063018e-06,
      "loss": 0.626,
      "step": 6371
    },
    {
      "epoch": 0.3438562408936377,
      "grad_norm": 1.0553500629029329,
      "learning_rate": 7.619960600133266e-06,
      "loss": 0.4565,
      "step": 6372
    },
    {
      "epoch": 0.3439102045221521,
      "grad_norm": 0.794081552326649,
      "learning_rate": 7.619287638273704e-06,
      "loss": 0.4156,
      "step": 6373
    },
    {
      "epoch": 0.34396416815066644,
      "grad_norm": 0.997361412584346,
      "learning_rate": 7.618614615503674e-06,
      "loss": 0.3859,
      "step": 6374
    },
    {
      "epoch": 0.34401813177918084,
      "grad_norm": 0.98300758544729,
      "learning_rate": 7.6179415318425195e-06,
      "loss": 0.4705,
      "step": 6375
    },
    {
      "epoch": 0.34407209540769523,
      "grad_norm": 1.079058765860522,
      "learning_rate": 7.617268387309584e-06,
      "loss": 0.4356,
      "step": 6376
    },
    {
      "epoch": 0.3441260590362096,
      "grad_norm": 0.9422696275862578,
      "learning_rate": 7.616595181924218e-06,
      "loss": 0.4377,
      "step": 6377
    },
    {
      "epoch": 0.344180022664724,
      "grad_norm": 1.0578269437497625,
      "learning_rate": 7.615921915705765e-06,
      "loss": 0.4188,
      "step": 6378
    },
    {
      "epoch": 0.34423398629323837,
      "grad_norm": 1.0310118964689827,
      "learning_rate": 7.61524858867358e-06,
      "loss": 0.4343,
      "step": 6379
    },
    {
      "epoch": 0.3442879499217527,
      "grad_norm": 1.0829446227000492,
      "learning_rate": 7.614575200847014e-06,
      "loss": 0.5215,
      "step": 6380
    },
    {
      "epoch": 0.3443419135502671,
      "grad_norm": 0.9737894662501605,
      "learning_rate": 7.61390175224542e-06,
      "loss": 0.4392,
      "step": 6381
    },
    {
      "epoch": 0.3443958771787815,
      "grad_norm": 0.9360616719228899,
      "learning_rate": 7.613228242888151e-06,
      "loss": 0.4848,
      "step": 6382
    },
    {
      "epoch": 0.3444498408072959,
      "grad_norm": 0.8024323938970329,
      "learning_rate": 7.612554672794569e-06,
      "loss": 0.3335,
      "step": 6383
    },
    {
      "epoch": 0.34450380443581025,
      "grad_norm": 0.920052733755465,
      "learning_rate": 7.6118810419840304e-06,
      "loss": 0.4222,
      "step": 6384
    },
    {
      "epoch": 0.34455776806432464,
      "grad_norm": 1.117902105427638,
      "learning_rate": 7.611207350475898e-06,
      "loss": 0.7098,
      "step": 6385
    },
    {
      "epoch": 0.34461173169283904,
      "grad_norm": 1.2803140830960997,
      "learning_rate": 7.6105335982895315e-06,
      "loss": 0.619,
      "step": 6386
    },
    {
      "epoch": 0.3446656953213534,
      "grad_norm": 1.0674378230252821,
      "learning_rate": 7.609859785444299e-06,
      "loss": 0.6925,
      "step": 6387
    },
    {
      "epoch": 0.3447196589498678,
      "grad_norm": 1.0580062888092039,
      "learning_rate": 7.609185911959562e-06,
      "loss": 0.4609,
      "step": 6388
    },
    {
      "epoch": 0.3447736225783822,
      "grad_norm": 0.8912327249496358,
      "learning_rate": 7.608511977854692e-06,
      "loss": 0.4097,
      "step": 6389
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.8336328877984187,
      "learning_rate": 7.607837983149057e-06,
      "loss": 0.4029,
      "step": 6390
    },
    {
      "epoch": 0.3448815498354109,
      "grad_norm": 0.8578264416157988,
      "learning_rate": 7.607163927862029e-06,
      "loss": 0.3493,
      "step": 6391
    },
    {
      "epoch": 0.3449355134639253,
      "grad_norm": 1.167348057195879,
      "learning_rate": 7.606489812012979e-06,
      "loss": 0.4624,
      "step": 6392
    },
    {
      "epoch": 0.3449894770924397,
      "grad_norm": 0.8676766903105042,
      "learning_rate": 7.605815635621285e-06,
      "loss": 0.3989,
      "step": 6393
    },
    {
      "epoch": 0.34504344072095405,
      "grad_norm": 1.0999557762317755,
      "learning_rate": 7.60514139870632e-06,
      "loss": 0.5148,
      "step": 6394
    },
    {
      "epoch": 0.34509740434946845,
      "grad_norm": 0.952408679345257,
      "learning_rate": 7.604467101287466e-06,
      "loss": 0.4578,
      "step": 6395
    },
    {
      "epoch": 0.34515136797798285,
      "grad_norm": 1.2131888393093284,
      "learning_rate": 7.603792743384101e-06,
      "loss": 0.5792,
      "step": 6396
    },
    {
      "epoch": 0.34520533160649725,
      "grad_norm": 1.063457761280014,
      "learning_rate": 7.603118325015605e-06,
      "loss": 0.5566,
      "step": 6397
    },
    {
      "epoch": 0.3452592952350116,
      "grad_norm": 1.1188837544843955,
      "learning_rate": 7.602443846201366e-06,
      "loss": 0.4646,
      "step": 6398
    },
    {
      "epoch": 0.345313258863526,
      "grad_norm": 0.9044692449588837,
      "learning_rate": 7.601769306960765e-06,
      "loss": 0.5402,
      "step": 6399
    },
    {
      "epoch": 0.3453672224920404,
      "grad_norm": 1.0051518446811119,
      "learning_rate": 7.60109470731319e-06,
      "loss": 0.5386,
      "step": 6400
    },
    {
      "epoch": 0.3454211861205547,
      "grad_norm": 1.0566452822436465,
      "learning_rate": 7.600420047278031e-06,
      "loss": 0.5006,
      "step": 6401
    },
    {
      "epoch": 0.3454751497490691,
      "grad_norm": 0.9832971256714372,
      "learning_rate": 7.599745326874678e-06,
      "loss": 0.4828,
      "step": 6402
    },
    {
      "epoch": 0.3455291133775835,
      "grad_norm": 0.9721892342294046,
      "learning_rate": 7.599070546122521e-06,
      "loss": 0.4232,
      "step": 6403
    },
    {
      "epoch": 0.3455830770060979,
      "grad_norm": 0.8665740092486355,
      "learning_rate": 7.5983957050409565e-06,
      "loss": 0.3887,
      "step": 6404
    },
    {
      "epoch": 0.34563704063461226,
      "grad_norm": 0.9207762282099392,
      "learning_rate": 7.59772080364938e-06,
      "loss": 0.3645,
      "step": 6405
    },
    {
      "epoch": 0.34569100426312666,
      "grad_norm": 1.1279087747897294,
      "learning_rate": 7.597045841967186e-06,
      "loss": 0.4083,
      "step": 6406
    },
    {
      "epoch": 0.34574496789164105,
      "grad_norm": 1.0425945935703738,
      "learning_rate": 7.596370820013777e-06,
      "loss": 0.3998,
      "step": 6407
    },
    {
      "epoch": 0.3457989315201554,
      "grad_norm": 1.232002556156866,
      "learning_rate": 7.595695737808553e-06,
      "loss": 0.68,
      "step": 6408
    },
    {
      "epoch": 0.3458528951486698,
      "grad_norm": 1.156549679973464,
      "learning_rate": 7.595020595370915e-06,
      "loss": 0.4332,
      "step": 6409
    },
    {
      "epoch": 0.3459068587771842,
      "grad_norm": 0.940999975926922,
      "learning_rate": 7.594345392720268e-06,
      "loss": 0.4218,
      "step": 6410
    },
    {
      "epoch": 0.34596082240569853,
      "grad_norm": 1.0924012627886959,
      "learning_rate": 7.5936701298760195e-06,
      "loss": 0.5578,
      "step": 6411
    },
    {
      "epoch": 0.34601478603421293,
      "grad_norm": 1.0841953500309107,
      "learning_rate": 7.592994806857575e-06,
      "loss": 0.4731,
      "step": 6412
    },
    {
      "epoch": 0.3460687496627273,
      "grad_norm": 0.8754921314138775,
      "learning_rate": 7.592319423684344e-06,
      "loss": 0.3733,
      "step": 6413
    },
    {
      "epoch": 0.3461227132912417,
      "grad_norm": 1.2619835290026082,
      "learning_rate": 7.591643980375739e-06,
      "loss": 0.6186,
      "step": 6414
    },
    {
      "epoch": 0.34617667691975607,
      "grad_norm": 1.1960137828326032,
      "learning_rate": 7.590968476951175e-06,
      "loss": 0.6281,
      "step": 6415
    },
    {
      "epoch": 0.34623064054827046,
      "grad_norm": 1.1911410052819897,
      "learning_rate": 7.5902929134300615e-06,
      "loss": 0.5291,
      "step": 6416
    },
    {
      "epoch": 0.34628460417678486,
      "grad_norm": 1.0711483248649838,
      "learning_rate": 7.589617289831817e-06,
      "loss": 0.5239,
      "step": 6417
    },
    {
      "epoch": 0.3463385678052992,
      "grad_norm": 0.8714906748261642,
      "learning_rate": 7.5889416061758625e-06,
      "loss": 0.4281,
      "step": 6418
    },
    {
      "epoch": 0.3463925314338136,
      "grad_norm": 1.0642068525464896,
      "learning_rate": 7.588265862481614e-06,
      "loss": 0.5423,
      "step": 6419
    },
    {
      "epoch": 0.346446495062328,
      "grad_norm": 0.9575506429811761,
      "learning_rate": 7.587590058768497e-06,
      "loss": 0.4753,
      "step": 6420
    },
    {
      "epoch": 0.3465004586908424,
      "grad_norm": 1.00335644490642,
      "learning_rate": 7.58691419505593e-06,
      "loss": 0.4925,
      "step": 6421
    },
    {
      "epoch": 0.34655442231935674,
      "grad_norm": 1.0086906323913811,
      "learning_rate": 7.586238271363341e-06,
      "loss": 0.503,
      "step": 6422
    },
    {
      "epoch": 0.34660838594787113,
      "grad_norm": 1.0230368435058932,
      "learning_rate": 7.585562287710156e-06,
      "loss": 0.596,
      "step": 6423
    },
    {
      "epoch": 0.34666234957638553,
      "grad_norm": 1.0827476034878034,
      "learning_rate": 7.584886244115803e-06,
      "loss": 0.5902,
      "step": 6424
    },
    {
      "epoch": 0.3467163132048999,
      "grad_norm": 0.8223604910409316,
      "learning_rate": 7.584210140599714e-06,
      "loss": 0.3812,
      "step": 6425
    },
    {
      "epoch": 0.34677027683341427,
      "grad_norm": 0.8966546820624178,
      "learning_rate": 7.583533977181319e-06,
      "loss": 0.3258,
      "step": 6426
    },
    {
      "epoch": 0.34682424046192867,
      "grad_norm": 0.8394177718062036,
      "learning_rate": 7.5828577538800505e-06,
      "loss": 0.5146,
      "step": 6427
    },
    {
      "epoch": 0.34687820409044307,
      "grad_norm": 1.0897835582347766,
      "learning_rate": 7.582181470715347e-06,
      "loss": 0.5358,
      "step": 6428
    },
    {
      "epoch": 0.3469321677189574,
      "grad_norm": 1.11215251980922,
      "learning_rate": 7.5815051277066455e-06,
      "loss": 0.5131,
      "step": 6429
    },
    {
      "epoch": 0.3469861313474718,
      "grad_norm": 0.9644162958511308,
      "learning_rate": 7.580828724873382e-06,
      "loss": 0.5649,
      "step": 6430
    },
    {
      "epoch": 0.3470400949759862,
      "grad_norm": 1.0695343312091405,
      "learning_rate": 7.580152262234999e-06,
      "loss": 0.4032,
      "step": 6431
    },
    {
      "epoch": 0.34709405860450054,
      "grad_norm": 1.1788988180911955,
      "learning_rate": 7.579475739810938e-06,
      "loss": 0.4835,
      "step": 6432
    },
    {
      "epoch": 0.34714802223301494,
      "grad_norm": 1.0003558963080261,
      "learning_rate": 7.5787991576206435e-06,
      "loss": 0.5289,
      "step": 6433
    },
    {
      "epoch": 0.34720198586152934,
      "grad_norm": 1.1524254463321606,
      "learning_rate": 7.578122515683559e-06,
      "loss": 0.5713,
      "step": 6434
    },
    {
      "epoch": 0.34725594949004374,
      "grad_norm": 0.9615286333214317,
      "learning_rate": 7.577445814019135e-06,
      "loss": 0.3906,
      "step": 6435
    },
    {
      "epoch": 0.3473099131185581,
      "grad_norm": 1.101815021687922,
      "learning_rate": 7.576769052646819e-06,
      "loss": 0.5202,
      "step": 6436
    },
    {
      "epoch": 0.3473638767470725,
      "grad_norm": 1.063094103031705,
      "learning_rate": 7.576092231586062e-06,
      "loss": 0.5565,
      "step": 6437
    },
    {
      "epoch": 0.3474178403755869,
      "grad_norm": 1.1340573369368097,
      "learning_rate": 7.575415350856316e-06,
      "loss": 0.5966,
      "step": 6438
    },
    {
      "epoch": 0.3474718040041012,
      "grad_norm": 1.0566919540456226,
      "learning_rate": 7.574738410477037e-06,
      "loss": 0.4258,
      "step": 6439
    },
    {
      "epoch": 0.3475257676326156,
      "grad_norm": 0.9569617990190707,
      "learning_rate": 7.57406141046768e-06,
      "loss": 0.5104,
      "step": 6440
    },
    {
      "epoch": 0.34757973126113,
      "grad_norm": 1.0831590881197488,
      "learning_rate": 7.573384350847701e-06,
      "loss": 0.5075,
      "step": 6441
    },
    {
      "epoch": 0.3476336948896444,
      "grad_norm": 1.0824118263068179,
      "learning_rate": 7.5727072316365605e-06,
      "loss": 0.4955,
      "step": 6442
    },
    {
      "epoch": 0.34768765851815875,
      "grad_norm": 0.9063023030182364,
      "learning_rate": 7.572030052853722e-06,
      "loss": 0.467,
      "step": 6443
    },
    {
      "epoch": 0.34774162214667315,
      "grad_norm": 1.1201981532855316,
      "learning_rate": 7.5713528145186435e-06,
      "loss": 0.4068,
      "step": 6444
    },
    {
      "epoch": 0.34779558577518754,
      "grad_norm": 1.1609110774948603,
      "learning_rate": 7.570675516650793e-06,
      "loss": 0.5264,
      "step": 6445
    },
    {
      "epoch": 0.3478495494037019,
      "grad_norm": 1.1637518139264371,
      "learning_rate": 7.5699981592696345e-06,
      "loss": 0.6107,
      "step": 6446
    },
    {
      "epoch": 0.3479035130322163,
      "grad_norm": 0.9272526193771339,
      "learning_rate": 7.56932074239464e-06,
      "loss": 0.4033,
      "step": 6447
    },
    {
      "epoch": 0.3479574766607307,
      "grad_norm": 1.2123822935211714,
      "learning_rate": 7.568643266045275e-06,
      "loss": 0.7548,
      "step": 6448
    },
    {
      "epoch": 0.348011440289245,
      "grad_norm": 0.9616844197235982,
      "learning_rate": 7.567965730241012e-06,
      "loss": 0.3641,
      "step": 6449
    },
    {
      "epoch": 0.3480654039177594,
      "grad_norm": 0.9304163656923773,
      "learning_rate": 7.567288135001326e-06,
      "loss": 0.4787,
      "step": 6450
    },
    {
      "epoch": 0.3481193675462738,
      "grad_norm": 1.043286908052124,
      "learning_rate": 7.56661048034569e-06,
      "loss": 0.545,
      "step": 6451
    },
    {
      "epoch": 0.3481733311747882,
      "grad_norm": 1.069173449791837,
      "learning_rate": 7.5659327662935786e-06,
      "loss": 0.5335,
      "step": 6452
    },
    {
      "epoch": 0.34822729480330256,
      "grad_norm": 1.0317317417452494,
      "learning_rate": 7.565254992864474e-06,
      "loss": 0.4982,
      "step": 6453
    },
    {
      "epoch": 0.34828125843181695,
      "grad_norm": 0.8975768981766608,
      "learning_rate": 7.564577160077854e-06,
      "loss": 0.3754,
      "step": 6454
    },
    {
      "epoch": 0.34833522206033135,
      "grad_norm": 0.9326442682545582,
      "learning_rate": 7.5638992679532e-06,
      "loss": 0.3907,
      "step": 6455
    },
    {
      "epoch": 0.3483891856888457,
      "grad_norm": 1.1291538707618298,
      "learning_rate": 7.563221316509996e-06,
      "loss": 0.7164,
      "step": 6456
    },
    {
      "epoch": 0.3484431493173601,
      "grad_norm": 1.044709528421894,
      "learning_rate": 7.562543305767726e-06,
      "loss": 0.486,
      "step": 6457
    },
    {
      "epoch": 0.3484971129458745,
      "grad_norm": 0.9766970730675506,
      "learning_rate": 7.561865235745876e-06,
      "loss": 0.5857,
      "step": 6458
    },
    {
      "epoch": 0.3485510765743889,
      "grad_norm": 1.1383732920925034,
      "learning_rate": 7.561187106463936e-06,
      "loss": 0.6222,
      "step": 6459
    },
    {
      "epoch": 0.3486050402029032,
      "grad_norm": 0.9393773196610555,
      "learning_rate": 7.560508917941399e-06,
      "loss": 0.4116,
      "step": 6460
    },
    {
      "epoch": 0.3486590038314176,
      "grad_norm": 1.0125907231067426,
      "learning_rate": 7.5598306701977505e-06,
      "loss": 0.5227,
      "step": 6461
    },
    {
      "epoch": 0.348712967459932,
      "grad_norm": 1.2114878822826867,
      "learning_rate": 7.559152363252489e-06,
      "loss": 0.472,
      "step": 6462
    },
    {
      "epoch": 0.34876693108844636,
      "grad_norm": 1.0400701747371557,
      "learning_rate": 7.558473997125108e-06,
      "loss": 0.6635,
      "step": 6463
    },
    {
      "epoch": 0.34882089471696076,
      "grad_norm": 0.8836531776509655,
      "learning_rate": 7.557795571835103e-06,
      "loss": 0.4867,
      "step": 6464
    },
    {
      "epoch": 0.34887485834547516,
      "grad_norm": 1.141639891334631,
      "learning_rate": 7.5571170874019754e-06,
      "loss": 0.4431,
      "step": 6465
    },
    {
      "epoch": 0.34892882197398956,
      "grad_norm": 1.0156102734384294,
      "learning_rate": 7.556438543845223e-06,
      "loss": 0.4458,
      "step": 6466
    },
    {
      "epoch": 0.3489827856025039,
      "grad_norm": 1.0374516517955577,
      "learning_rate": 7.555759941184351e-06,
      "loss": 0.467,
      "step": 6467
    },
    {
      "epoch": 0.3490367492310183,
      "grad_norm": 1.115149400148132,
      "learning_rate": 7.55508127943886e-06,
      "loss": 0.5661,
      "step": 6468
    },
    {
      "epoch": 0.3490907128595327,
      "grad_norm": 1.0427405465174675,
      "learning_rate": 7.5544025586282556e-06,
      "loss": 0.5215,
      "step": 6469
    },
    {
      "epoch": 0.34914467648804703,
      "grad_norm": 0.9750814615296937,
      "learning_rate": 7.553723778772045e-06,
      "loss": 0.4367,
      "step": 6470
    },
    {
      "epoch": 0.34919864011656143,
      "grad_norm": 0.9036776672247029,
      "learning_rate": 7.5530449398897395e-06,
      "loss": 0.435,
      "step": 6471
    },
    {
      "epoch": 0.34925260374507583,
      "grad_norm": 1.1634490762494756,
      "learning_rate": 7.552366042000847e-06,
      "loss": 0.649,
      "step": 6472
    },
    {
      "epoch": 0.3493065673735902,
      "grad_norm": 0.7996603303044308,
      "learning_rate": 7.551687085124884e-06,
      "loss": 0.3644,
      "step": 6473
    },
    {
      "epoch": 0.34936053100210457,
      "grad_norm": 0.9898006183722865,
      "learning_rate": 7.551008069281357e-06,
      "loss": 0.5369,
      "step": 6474
    },
    {
      "epoch": 0.34941449463061897,
      "grad_norm": 0.99000172000474,
      "learning_rate": 7.550328994489786e-06,
      "loss": 0.4508,
      "step": 6475
    },
    {
      "epoch": 0.34946845825913336,
      "grad_norm": 1.0784946671790563,
      "learning_rate": 7.54964986076969e-06,
      "loss": 0.473,
      "step": 6476
    },
    {
      "epoch": 0.3495224218876477,
      "grad_norm": 0.9638909696345807,
      "learning_rate": 7.548970668140585e-06,
      "loss": 0.4623,
      "step": 6477
    },
    {
      "epoch": 0.3495763855161621,
      "grad_norm": 1.1759166298808386,
      "learning_rate": 7.548291416621994e-06,
      "loss": 0.4774,
      "step": 6478
    },
    {
      "epoch": 0.3496303491446765,
      "grad_norm": 0.9478636487886638,
      "learning_rate": 7.547612106233436e-06,
      "loss": 0.402,
      "step": 6479
    },
    {
      "epoch": 0.34968431277319084,
      "grad_norm": 1.2501544237995892,
      "learning_rate": 7.546932736994438e-06,
      "loss": 0.5943,
      "step": 6480
    },
    {
      "epoch": 0.34973827640170524,
      "grad_norm": 1.0661834044753822,
      "learning_rate": 7.546253308924525e-06,
      "loss": 0.6138,
      "step": 6481
    },
    {
      "epoch": 0.34979224003021964,
      "grad_norm": 1.236486177431717,
      "learning_rate": 7.545573822043224e-06,
      "loss": 0.5332,
      "step": 6482
    },
    {
      "epoch": 0.34984620365873403,
      "grad_norm": 0.8428806868503281,
      "learning_rate": 7.544894276370065e-06,
      "loss": 0.4421,
      "step": 6483
    },
    {
      "epoch": 0.3499001672872484,
      "grad_norm": 0.9053618769874671,
      "learning_rate": 7.544214671924579e-06,
      "loss": 0.3339,
      "step": 6484
    },
    {
      "epoch": 0.3499541309157628,
      "grad_norm": 0.9496733088647022,
      "learning_rate": 7.543535008726296e-06,
      "loss": 0.4169,
      "step": 6485
    },
    {
      "epoch": 0.35000809454427717,
      "grad_norm": 0.9239762575429569,
      "learning_rate": 7.542855286794752e-06,
      "loss": 0.3784,
      "step": 6486
    },
    {
      "epoch": 0.3500620581727915,
      "grad_norm": 1.2124299029352064,
      "learning_rate": 7.542175506149484e-06,
      "loss": 0.506,
      "step": 6487
    },
    {
      "epoch": 0.3501160218013059,
      "grad_norm": 0.9059247553173704,
      "learning_rate": 7.541495666810029e-06,
      "loss": 0.4454,
      "step": 6488
    },
    {
      "epoch": 0.3501699854298203,
      "grad_norm": 1.0510207747074805,
      "learning_rate": 7.540815768795924e-06,
      "loss": 0.418,
      "step": 6489
    },
    {
      "epoch": 0.3502239490583347,
      "grad_norm": 1.0799165418452552,
      "learning_rate": 7.540135812126712e-06,
      "loss": 0.5981,
      "step": 6490
    },
    {
      "epoch": 0.35027791268684905,
      "grad_norm": 1.1378964315927171,
      "learning_rate": 7.539455796821936e-06,
      "loss": 0.5029,
      "step": 6491
    },
    {
      "epoch": 0.35033187631536344,
      "grad_norm": 0.9716436585584425,
      "learning_rate": 7.53877572290114e-06,
      "loss": 0.3808,
      "step": 6492
    },
    {
      "epoch": 0.35038583994387784,
      "grad_norm": 0.9631044257974124,
      "learning_rate": 7.53809559038387e-06,
      "loss": 0.4207,
      "step": 6493
    },
    {
      "epoch": 0.3504398035723922,
      "grad_norm": 1.1221484059369171,
      "learning_rate": 7.537415399289672e-06,
      "loss": 0.697,
      "step": 6494
    },
    {
      "epoch": 0.3504937672009066,
      "grad_norm": 0.9437282279109885,
      "learning_rate": 7.536735149638098e-06,
      "loss": 0.3959,
      "step": 6495
    },
    {
      "epoch": 0.350547730829421,
      "grad_norm": 0.9971717406194858,
      "learning_rate": 7.536054841448698e-06,
      "loss": 0.4324,
      "step": 6496
    },
    {
      "epoch": 0.3506016944579354,
      "grad_norm": 0.9334461101355187,
      "learning_rate": 7.535374474741023e-06,
      "loss": 0.4975,
      "step": 6497
    },
    {
      "epoch": 0.3506556580864497,
      "grad_norm": 1.089678497834821,
      "learning_rate": 7.53469404953463e-06,
      "loss": 0.3305,
      "step": 6498
    },
    {
      "epoch": 0.3507096217149641,
      "grad_norm": 1.112337886551743,
      "learning_rate": 7.534013565849073e-06,
      "loss": 0.5749,
      "step": 6499
    },
    {
      "epoch": 0.3507635853434785,
      "grad_norm": 0.9452904016541749,
      "learning_rate": 7.533333023703912e-06,
      "loss": 0.3516,
      "step": 6500
    },
    {
      "epoch": 0.3507635853434785,
      "eval_loss": 0.5656527876853943,
      "eval_runtime": 163.6851,
      "eval_samples_per_second": 21.01,
      "eval_steps_per_second": 0.88,
      "step": 6500
    },
    {
      "epoch": 0.35081754897199285,
      "grad_norm": 0.9767187745997931,
      "learning_rate": 7.532652423118705e-06,
      "loss": 0.4503,
      "step": 6501
    },
    {
      "epoch": 0.35087151260050725,
      "grad_norm": 1.0899514245120523,
      "learning_rate": 7.531971764113015e-06,
      "loss": 0.5035,
      "step": 6502
    },
    {
      "epoch": 0.35092547622902165,
      "grad_norm": 1.2373739896553027,
      "learning_rate": 7.531291046706402e-06,
      "loss": 0.5972,
      "step": 6503
    },
    {
      "epoch": 0.35097943985753605,
      "grad_norm": 0.9063542039805706,
      "learning_rate": 7.53061027091843e-06,
      "loss": 0.492,
      "step": 6504
    },
    {
      "epoch": 0.3510334034860504,
      "grad_norm": 0.8782547005794519,
      "learning_rate": 7.529929436768671e-06,
      "loss": 0.3857,
      "step": 6505
    },
    {
      "epoch": 0.3510873671145648,
      "grad_norm": 1.2906422581490622,
      "learning_rate": 7.5292485442766865e-06,
      "loss": 0.5738,
      "step": 6506
    },
    {
      "epoch": 0.3511413307430792,
      "grad_norm": 0.8534822401540496,
      "learning_rate": 7.528567593462049e-06,
      "loss": 0.4218,
      "step": 6507
    },
    {
      "epoch": 0.3511952943715935,
      "grad_norm": 0.7167377293175881,
      "learning_rate": 7.527886584344329e-06,
      "loss": 0.3629,
      "step": 6508
    },
    {
      "epoch": 0.3512492580001079,
      "grad_norm": 1.0704166342047348,
      "learning_rate": 7.527205516943099e-06,
      "loss": 0.5124,
      "step": 6509
    },
    {
      "epoch": 0.3513032216286223,
      "grad_norm": 0.9980188363034932,
      "learning_rate": 7.526524391277935e-06,
      "loss": 0.4384,
      "step": 6510
    },
    {
      "epoch": 0.3513571852571367,
      "grad_norm": 1.0428380111100113,
      "learning_rate": 7.525843207368411e-06,
      "loss": 0.4504,
      "step": 6511
    },
    {
      "epoch": 0.35141114888565106,
      "grad_norm": 1.2156763880472938,
      "learning_rate": 7.525161965234108e-06,
      "loss": 0.6225,
      "step": 6512
    },
    {
      "epoch": 0.35146511251416546,
      "grad_norm": 0.9182257561093974,
      "learning_rate": 7.524480664894602e-06,
      "loss": 0.3537,
      "step": 6513
    },
    {
      "epoch": 0.35151907614267985,
      "grad_norm": 0.9104483078398737,
      "learning_rate": 7.523799306369477e-06,
      "loss": 0.4191,
      "step": 6514
    },
    {
      "epoch": 0.3515730397711942,
      "grad_norm": 0.8672700304103346,
      "learning_rate": 7.523117889678317e-06,
      "loss": 0.4061,
      "step": 6515
    },
    {
      "epoch": 0.3516270033997086,
      "grad_norm": 1.0509974525092312,
      "learning_rate": 7.5224364148407026e-06,
      "loss": 0.4303,
      "step": 6516
    },
    {
      "epoch": 0.351680967028223,
      "grad_norm": 0.9847922725703373,
      "learning_rate": 7.521754881876221e-06,
      "loss": 0.441,
      "step": 6517
    },
    {
      "epoch": 0.35173493065673733,
      "grad_norm": 0.9925256941585913,
      "learning_rate": 7.521073290804464e-06,
      "loss": 0.4132,
      "step": 6518
    },
    {
      "epoch": 0.35178889428525173,
      "grad_norm": 1.0322812676550408,
      "learning_rate": 7.520391641645016e-06,
      "loss": 0.5567,
      "step": 6519
    },
    {
      "epoch": 0.3518428579137661,
      "grad_norm": 1.0933715668745543,
      "learning_rate": 7.519709934417472e-06,
      "loss": 0.4357,
      "step": 6520
    },
    {
      "epoch": 0.3518968215422805,
      "grad_norm": 1.0906053014846768,
      "learning_rate": 7.5190281691414235e-06,
      "loss": 0.4322,
      "step": 6521
    },
    {
      "epoch": 0.35195078517079487,
      "grad_norm": 0.9217831467895617,
      "learning_rate": 7.518346345836465e-06,
      "loss": 0.4451,
      "step": 6522
    },
    {
      "epoch": 0.35200474879930926,
      "grad_norm": 1.0339158278913738,
      "learning_rate": 7.517664464522193e-06,
      "loss": 0.5398,
      "step": 6523
    },
    {
      "epoch": 0.35205871242782366,
      "grad_norm": 0.9235251329321243,
      "learning_rate": 7.516982525218206e-06,
      "loss": 0.4561,
      "step": 6524
    },
    {
      "epoch": 0.352112676056338,
      "grad_norm": 1.0244854001814672,
      "learning_rate": 7.516300527944104e-06,
      "loss": 0.3849,
      "step": 6525
    },
    {
      "epoch": 0.3521666396848524,
      "grad_norm": 1.012666720232384,
      "learning_rate": 7.515618472719487e-06,
      "loss": 0.5758,
      "step": 6526
    },
    {
      "epoch": 0.3522206033133668,
      "grad_norm": 0.8983934466627281,
      "learning_rate": 7.514936359563958e-06,
      "loss": 0.4833,
      "step": 6527
    },
    {
      "epoch": 0.3522745669418812,
      "grad_norm": 1.0900704030007031,
      "learning_rate": 7.514254188497122e-06,
      "loss": 0.5407,
      "step": 6528
    },
    {
      "epoch": 0.35232853057039554,
      "grad_norm": 0.9486340976612919,
      "learning_rate": 7.513571959538586e-06,
      "loss": 0.4836,
      "step": 6529
    },
    {
      "epoch": 0.35238249419890993,
      "grad_norm": 0.8820742830405071,
      "learning_rate": 7.512889672707957e-06,
      "loss": 0.3309,
      "step": 6530
    },
    {
      "epoch": 0.35243645782742433,
      "grad_norm": 1.075369047919462,
      "learning_rate": 7.5122073280248445e-06,
      "loss": 0.5763,
      "step": 6531
    },
    {
      "epoch": 0.3524904214559387,
      "grad_norm": 1.1301400030523072,
      "learning_rate": 7.5115249255088615e-06,
      "loss": 0.6124,
      "step": 6532
    },
    {
      "epoch": 0.35254438508445307,
      "grad_norm": 1.1899868255866837,
      "learning_rate": 7.510842465179619e-06,
      "loss": 0.4622,
      "step": 6533
    },
    {
      "epoch": 0.35259834871296747,
      "grad_norm": 1.283437097140067,
      "learning_rate": 7.510159947056731e-06,
      "loss": 0.4292,
      "step": 6534
    },
    {
      "epoch": 0.35265231234148187,
      "grad_norm": 1.2186542485318304,
      "learning_rate": 7.509477371159816e-06,
      "loss": 0.7543,
      "step": 6535
    },
    {
      "epoch": 0.3527062759699962,
      "grad_norm": 0.9148757560334622,
      "learning_rate": 7.5087947375084915e-06,
      "loss": 0.4527,
      "step": 6536
    },
    {
      "epoch": 0.3527602395985106,
      "grad_norm": 1.1039072354820263,
      "learning_rate": 7.508112046122377e-06,
      "loss": 0.5021,
      "step": 6537
    },
    {
      "epoch": 0.352814203227025,
      "grad_norm": 1.1977852556720494,
      "learning_rate": 7.507429297021091e-06,
      "loss": 0.4306,
      "step": 6538
    },
    {
      "epoch": 0.35286816685553934,
      "grad_norm": 1.0543272646655182,
      "learning_rate": 7.506746490224261e-06,
      "loss": 0.4738,
      "step": 6539
    },
    {
      "epoch": 0.35292213048405374,
      "grad_norm": 1.1317986055465181,
      "learning_rate": 7.506063625751508e-06,
      "loss": 0.601,
      "step": 6540
    },
    {
      "epoch": 0.35297609411256814,
      "grad_norm": 0.8227484742377368,
      "learning_rate": 7.505380703622459e-06,
      "loss": 0.3848,
      "step": 6541
    },
    {
      "epoch": 0.35303005774108254,
      "grad_norm": 1.0516969749419087,
      "learning_rate": 7.504697723856742e-06,
      "loss": 0.5662,
      "step": 6542
    },
    {
      "epoch": 0.3530840213695969,
      "grad_norm": 0.8543176655609518,
      "learning_rate": 7.504014686473988e-06,
      "loss": 0.3906,
      "step": 6543
    },
    {
      "epoch": 0.3531379849981113,
      "grad_norm": 0.9048863952231032,
      "learning_rate": 7.503331591493827e-06,
      "loss": 0.4213,
      "step": 6544
    },
    {
      "epoch": 0.3531919486266257,
      "grad_norm": 1.034883559470357,
      "learning_rate": 7.5026484389358915e-06,
      "loss": 0.5135,
      "step": 6545
    },
    {
      "epoch": 0.35324591225514,
      "grad_norm": 0.8891547021946267,
      "learning_rate": 7.501965228819815e-06,
      "loss": 0.3996,
      "step": 6546
    },
    {
      "epoch": 0.3532998758836544,
      "grad_norm": 1.1161633245858142,
      "learning_rate": 7.501281961165236e-06,
      "loss": 0.4659,
      "step": 6547
    },
    {
      "epoch": 0.3533538395121688,
      "grad_norm": 0.9282455990120431,
      "learning_rate": 7.500598635991793e-06,
      "loss": 0.458,
      "step": 6548
    },
    {
      "epoch": 0.35340780314068315,
      "grad_norm": 1.126714993253324,
      "learning_rate": 7.499915253319121e-06,
      "loss": 0.6586,
      "step": 6549
    },
    {
      "epoch": 0.35346176676919755,
      "grad_norm": 1.1556134500078927,
      "learning_rate": 7.499231813166865e-06,
      "loss": 0.5005,
      "step": 6550
    },
    {
      "epoch": 0.35351573039771195,
      "grad_norm": 1.0527378431484553,
      "learning_rate": 7.498548315554667e-06,
      "loss": 0.5111,
      "step": 6551
    },
    {
      "epoch": 0.35356969402622634,
      "grad_norm": 0.9064509540699975,
      "learning_rate": 7.49786476050217e-06,
      "loss": 0.4669,
      "step": 6552
    },
    {
      "epoch": 0.3536236576547407,
      "grad_norm": 1.0730732019894063,
      "learning_rate": 7.4971811480290225e-06,
      "loss": 0.4403,
      "step": 6553
    },
    {
      "epoch": 0.3536776212832551,
      "grad_norm": 0.9676578458527063,
      "learning_rate": 7.49649747815487e-06,
      "loss": 0.3726,
      "step": 6554
    },
    {
      "epoch": 0.3537315849117695,
      "grad_norm": 1.1235517391637238,
      "learning_rate": 7.495813750899363e-06,
      "loss": 0.5071,
      "step": 6555
    },
    {
      "epoch": 0.3537855485402838,
      "grad_norm": 0.8700223500916819,
      "learning_rate": 7.495129966282153e-06,
      "loss": 0.4991,
      "step": 6556
    },
    {
      "epoch": 0.3538395121687982,
      "grad_norm": 0.8530719531778566,
      "learning_rate": 7.4944461243228904e-06,
      "loss": 0.3444,
      "step": 6557
    },
    {
      "epoch": 0.3538934757973126,
      "grad_norm": 0.9832266779014823,
      "learning_rate": 7.4937622250412325e-06,
      "loss": 0.431,
      "step": 6558
    },
    {
      "epoch": 0.353947439425827,
      "grad_norm": 0.976178697446358,
      "learning_rate": 7.493078268456832e-06,
      "loss": 0.4827,
      "step": 6559
    },
    {
      "epoch": 0.35400140305434136,
      "grad_norm": 1.3005500017645097,
      "learning_rate": 7.492394254589348e-06,
      "loss": 0.6174,
      "step": 6560
    },
    {
      "epoch": 0.35405536668285575,
      "grad_norm": 0.7978554418033181,
      "learning_rate": 7.491710183458441e-06,
      "loss": 0.3386,
      "step": 6561
    },
    {
      "epoch": 0.35410933031137015,
      "grad_norm": 0.9614706703767442,
      "learning_rate": 7.491026055083771e-06,
      "loss": 0.505,
      "step": 6562
    },
    {
      "epoch": 0.3541632939398845,
      "grad_norm": 0.7489083952812767,
      "learning_rate": 7.4903418694849995e-06,
      "loss": 0.3307,
      "step": 6563
    },
    {
      "epoch": 0.3542172575683989,
      "grad_norm": 1.3067046354755372,
      "learning_rate": 7.489657626681792e-06,
      "loss": 0.6683,
      "step": 6564
    },
    {
      "epoch": 0.3542712211969133,
      "grad_norm": 1.0315985869882875,
      "learning_rate": 7.488973326693815e-06,
      "loss": 0.4483,
      "step": 6565
    },
    {
      "epoch": 0.3543251848254277,
      "grad_norm": 1.0352344968906375,
      "learning_rate": 7.488288969540732e-06,
      "loss": 0.4214,
      "step": 6566
    },
    {
      "epoch": 0.35437914845394203,
      "grad_norm": 1.029079623392603,
      "learning_rate": 7.487604555242218e-06,
      "loss": 0.4392,
      "step": 6567
    },
    {
      "epoch": 0.3544331120824564,
      "grad_norm": 1.0465136385765714,
      "learning_rate": 7.48692008381794e-06,
      "loss": 0.4642,
      "step": 6568
    },
    {
      "epoch": 0.3544870757109708,
      "grad_norm": 1.0752119066245482,
      "learning_rate": 7.48623555528757e-06,
      "loss": 0.5067,
      "step": 6569
    },
    {
      "epoch": 0.35454103933948516,
      "grad_norm": 1.0812256978833386,
      "learning_rate": 7.485550969670785e-06,
      "loss": 0.4937,
      "step": 6570
    },
    {
      "epoch": 0.35459500296799956,
      "grad_norm": 0.9426732062988696,
      "learning_rate": 7.484866326987258e-06,
      "loss": 0.523,
      "step": 6571
    },
    {
      "epoch": 0.35464896659651396,
      "grad_norm": 0.9659435840074309,
      "learning_rate": 7.484181627256667e-06,
      "loss": 0.5122,
      "step": 6572
    },
    {
      "epoch": 0.35470293022502836,
      "grad_norm": 0.8643323539680438,
      "learning_rate": 7.48349687049869e-06,
      "loss": 0.5218,
      "step": 6573
    },
    {
      "epoch": 0.3547568938535427,
      "grad_norm": 1.017844928597624,
      "learning_rate": 7.482812056733011e-06,
      "loss": 0.418,
      "step": 6574
    },
    {
      "epoch": 0.3548108574820571,
      "grad_norm": 0.9180222678695006,
      "learning_rate": 7.482127185979308e-06,
      "loss": 0.3677,
      "step": 6575
    },
    {
      "epoch": 0.3548648211105715,
      "grad_norm": 1.128070416482334,
      "learning_rate": 7.4814422582572675e-06,
      "loss": 0.5085,
      "step": 6576
    },
    {
      "epoch": 0.35491878473908584,
      "grad_norm": 1.0541985998890318,
      "learning_rate": 7.480757273586575e-06,
      "loss": 0.5239,
      "step": 6577
    },
    {
      "epoch": 0.35497274836760023,
      "grad_norm": 1.4431963200329356,
      "learning_rate": 7.4800722319869166e-06,
      "loss": 0.5252,
      "step": 6578
    },
    {
      "epoch": 0.35502671199611463,
      "grad_norm": 0.835595730971848,
      "learning_rate": 7.479387133477982e-06,
      "loss": 0.4226,
      "step": 6579
    },
    {
      "epoch": 0.355080675624629,
      "grad_norm": 1.1275770036973884,
      "learning_rate": 7.478701978079461e-06,
      "loss": 0.4153,
      "step": 6580
    },
    {
      "epoch": 0.35513463925314337,
      "grad_norm": 1.0814275394069335,
      "learning_rate": 7.478016765811047e-06,
      "loss": 0.501,
      "step": 6581
    },
    {
      "epoch": 0.35518860288165777,
      "grad_norm": 1.0176986314889893,
      "learning_rate": 7.477331496692429e-06,
      "loss": 0.412,
      "step": 6582
    },
    {
      "epoch": 0.35524256651017216,
      "grad_norm": 1.091184424651562,
      "learning_rate": 7.476646170743309e-06,
      "loss": 0.4449,
      "step": 6583
    },
    {
      "epoch": 0.3552965301386865,
      "grad_norm": 1.1409242080403468,
      "learning_rate": 7.475960787983381e-06,
      "loss": 0.5783,
      "step": 6584
    },
    {
      "epoch": 0.3553504937672009,
      "grad_norm": 1.0044724139352164,
      "learning_rate": 7.475275348432342e-06,
      "loss": 0.5442,
      "step": 6585
    },
    {
      "epoch": 0.3554044573957153,
      "grad_norm": 0.9760728680467642,
      "learning_rate": 7.474589852109893e-06,
      "loss": 0.4575,
      "step": 6586
    },
    {
      "epoch": 0.35545842102422964,
      "grad_norm": 0.9342750166931196,
      "learning_rate": 7.473904299035738e-06,
      "loss": 0.4684,
      "step": 6587
    },
    {
      "epoch": 0.35551238465274404,
      "grad_norm": 1.1136080273960873,
      "learning_rate": 7.4732186892295775e-06,
      "loss": 0.4052,
      "step": 6588
    },
    {
      "epoch": 0.35556634828125844,
      "grad_norm": 0.8539770458239477,
      "learning_rate": 7.4725330227111195e-06,
      "loss": 0.3837,
      "step": 6589
    },
    {
      "epoch": 0.35562031190977283,
      "grad_norm": 1.0565883288543376,
      "learning_rate": 7.471847299500068e-06,
      "loss": 0.6871,
      "step": 6590
    },
    {
      "epoch": 0.3556742755382872,
      "grad_norm": 1.0309164332032599,
      "learning_rate": 7.471161519616135e-06,
      "loss": 0.4717,
      "step": 6591
    },
    {
      "epoch": 0.3557282391668016,
      "grad_norm": 1.077510731868254,
      "learning_rate": 7.470475683079026e-06,
      "loss": 0.4438,
      "step": 6592
    },
    {
      "epoch": 0.35578220279531597,
      "grad_norm": 0.8440084373183798,
      "learning_rate": 7.469789789908455e-06,
      "loss": 0.3238,
      "step": 6593
    },
    {
      "epoch": 0.3558361664238303,
      "grad_norm": 1.2947619627388036,
      "learning_rate": 7.469103840124137e-06,
      "loss": 0.841,
      "step": 6594
    },
    {
      "epoch": 0.3558901300523447,
      "grad_norm": 0.8581337425178368,
      "learning_rate": 7.468417833745782e-06,
      "loss": 0.4336,
      "step": 6595
    },
    {
      "epoch": 0.3559440936808591,
      "grad_norm": 0.9937365637669544,
      "learning_rate": 7.467731770793112e-06,
      "loss": 0.4611,
      "step": 6596
    },
    {
      "epoch": 0.3559980573093735,
      "grad_norm": 0.9230737949764661,
      "learning_rate": 7.46704565128584e-06,
      "loss": 0.4621,
      "step": 6597
    },
    {
      "epoch": 0.35605202093788785,
      "grad_norm": 1.0466849447499893,
      "learning_rate": 7.466359475243692e-06,
      "loss": 0.4781,
      "step": 6598
    },
    {
      "epoch": 0.35610598456640224,
      "grad_norm": 1.0625913204688138,
      "learning_rate": 7.4656732426863835e-06,
      "loss": 0.6118,
      "step": 6599
    },
    {
      "epoch": 0.35615994819491664,
      "grad_norm": 0.9835674576466229,
      "learning_rate": 7.464986953633639e-06,
      "loss": 0.3728,
      "step": 6600
    },
    {
      "epoch": 0.356213911823431,
      "grad_norm": 1.3895064200829803,
      "learning_rate": 7.464300608105186e-06,
      "loss": 0.5864,
      "step": 6601
    },
    {
      "epoch": 0.3562678754519454,
      "grad_norm": 0.8891016292741093,
      "learning_rate": 7.463614206120748e-06,
      "loss": 0.4361,
      "step": 6602
    },
    {
      "epoch": 0.3563218390804598,
      "grad_norm": 1.0638876528766092,
      "learning_rate": 7.462927747700054e-06,
      "loss": 0.5618,
      "step": 6603
    },
    {
      "epoch": 0.3563758027089742,
      "grad_norm": 1.1073684957876997,
      "learning_rate": 7.462241232862833e-06,
      "loss": 0.4606,
      "step": 6604
    },
    {
      "epoch": 0.3564297663374885,
      "grad_norm": 1.276653625763433,
      "learning_rate": 7.461554661628816e-06,
      "loss": 0.4649,
      "step": 6605
    },
    {
      "epoch": 0.3564837299660029,
      "grad_norm": 1.1282786489095455,
      "learning_rate": 7.460868034017737e-06,
      "loss": 0.5665,
      "step": 6606
    },
    {
      "epoch": 0.3565376935945173,
      "grad_norm": 0.9956975657041539,
      "learning_rate": 7.460181350049328e-06,
      "loss": 0.5908,
      "step": 6607
    },
    {
      "epoch": 0.35659165722303166,
      "grad_norm": 1.1835863813987963,
      "learning_rate": 7.459494609743327e-06,
      "loss": 0.5879,
      "step": 6608
    },
    {
      "epoch": 0.35664562085154605,
      "grad_norm": 1.014259400012214,
      "learning_rate": 7.4588078131194705e-06,
      "loss": 0.476,
      "step": 6609
    },
    {
      "epoch": 0.35669958448006045,
      "grad_norm": 1.1079682667691544,
      "learning_rate": 7.458120960197498e-06,
      "loss": 0.5974,
      "step": 6610
    },
    {
      "epoch": 0.35675354810857485,
      "grad_norm": 1.0329480755026565,
      "learning_rate": 7.457434050997152e-06,
      "loss": 0.563,
      "step": 6611
    },
    {
      "epoch": 0.3568075117370892,
      "grad_norm": 1.1903914532945965,
      "learning_rate": 7.456747085538173e-06,
      "loss": 0.6955,
      "step": 6612
    },
    {
      "epoch": 0.3568614753656036,
      "grad_norm": 1.063247988350127,
      "learning_rate": 7.456060063840304e-06,
      "loss": 0.4794,
      "step": 6613
    },
    {
      "epoch": 0.356915438994118,
      "grad_norm": 0.9356931714980239,
      "learning_rate": 7.455372985923295e-06,
      "loss": 0.4093,
      "step": 6614
    },
    {
      "epoch": 0.3569694026226323,
      "grad_norm": 0.9886556058041656,
      "learning_rate": 7.45468585180689e-06,
      "loss": 0.5079,
      "step": 6615
    },
    {
      "epoch": 0.3570233662511467,
      "grad_norm": 1.0649377861656881,
      "learning_rate": 7.4539986615108365e-06,
      "loss": 0.6654,
      "step": 6616
    },
    {
      "epoch": 0.3570773298796611,
      "grad_norm": 1.0171607866143828,
      "learning_rate": 7.453311415054889e-06,
      "loss": 0.3463,
      "step": 6617
    },
    {
      "epoch": 0.35713129350817546,
      "grad_norm": 0.9053378066952775,
      "learning_rate": 7.452624112458797e-06,
      "loss": 0.3593,
      "step": 6618
    },
    {
      "epoch": 0.35718525713668986,
      "grad_norm": 1.0969382588410779,
      "learning_rate": 7.451936753742316e-06,
      "loss": 0.463,
      "step": 6619
    },
    {
      "epoch": 0.35723922076520426,
      "grad_norm": 1.5047068387374998,
      "learning_rate": 7.4512493389252e-06,
      "loss": 0.6974,
      "step": 6620
    },
    {
      "epoch": 0.35729318439371865,
      "grad_norm": 1.142019670340147,
      "learning_rate": 7.450561868027204e-06,
      "loss": 0.535,
      "step": 6621
    },
    {
      "epoch": 0.357347148022233,
      "grad_norm": 1.0439734278783264,
      "learning_rate": 7.449874341068092e-06,
      "loss": 0.5284,
      "step": 6622
    },
    {
      "epoch": 0.3574011116507474,
      "grad_norm": 1.33118939664702,
      "learning_rate": 7.4491867580676206e-06,
      "loss": 0.5786,
      "step": 6623
    },
    {
      "epoch": 0.3574550752792618,
      "grad_norm": 0.8699417205292298,
      "learning_rate": 7.4484991190455515e-06,
      "loss": 0.4434,
      "step": 6624
    },
    {
      "epoch": 0.35750903890777613,
      "grad_norm": 1.0501206693270315,
      "learning_rate": 7.447811424021649e-06,
      "loss": 0.5014,
      "step": 6625
    },
    {
      "epoch": 0.35756300253629053,
      "grad_norm": 0.920471822603437,
      "learning_rate": 7.447123673015678e-06,
      "loss": 0.4456,
      "step": 6626
    },
    {
      "epoch": 0.35761696616480493,
      "grad_norm": 0.9494119732148438,
      "learning_rate": 7.446435866047405e-06,
      "loss": 0.3606,
      "step": 6627
    },
    {
      "epoch": 0.3576709297933193,
      "grad_norm": 0.9767816035143059,
      "learning_rate": 7.445748003136599e-06,
      "loss": 0.4785,
      "step": 6628
    },
    {
      "epoch": 0.35772489342183367,
      "grad_norm": 0.9626289423710078,
      "learning_rate": 7.44506008430303e-06,
      "loss": 0.4139,
      "step": 6629
    },
    {
      "epoch": 0.35777885705034806,
      "grad_norm": 1.1833292100571229,
      "learning_rate": 7.4443721095664674e-06,
      "loss": 0.5452,
      "step": 6630
    },
    {
      "epoch": 0.35783282067886246,
      "grad_norm": 0.8705835170467068,
      "learning_rate": 7.443684078946687e-06,
      "loss": 0.4015,
      "step": 6631
    },
    {
      "epoch": 0.3578867843073768,
      "grad_norm": 0.9627277422019352,
      "learning_rate": 7.442995992463463e-06,
      "loss": 0.5252,
      "step": 6632
    },
    {
      "epoch": 0.3579407479358912,
      "grad_norm": 0.898825001323243,
      "learning_rate": 7.442307850136569e-06,
      "loss": 0.4017,
      "step": 6633
    },
    {
      "epoch": 0.3579947115644056,
      "grad_norm": 0.9257550988748362,
      "learning_rate": 7.441619651985786e-06,
      "loss": 0.3955,
      "step": 6634
    },
    {
      "epoch": 0.35804867519292,
      "grad_norm": 0.9279711635707943,
      "learning_rate": 7.440931398030892e-06,
      "loss": 0.4296,
      "step": 6635
    },
    {
      "epoch": 0.35810263882143434,
      "grad_norm": 0.915673266293192,
      "learning_rate": 7.440243088291667e-06,
      "loss": 0.4268,
      "step": 6636
    },
    {
      "epoch": 0.35815660244994874,
      "grad_norm": 1.0158044843888405,
      "learning_rate": 7.439554722787898e-06,
      "loss": 0.437,
      "step": 6637
    },
    {
      "epoch": 0.35821056607846313,
      "grad_norm": 0.8455530392246313,
      "learning_rate": 7.438866301539364e-06,
      "loss": 0.4003,
      "step": 6638
    },
    {
      "epoch": 0.3582645297069775,
      "grad_norm": 1.0440569166205138,
      "learning_rate": 7.4381778245658555e-06,
      "loss": 0.5095,
      "step": 6639
    },
    {
      "epoch": 0.35831849333549187,
      "grad_norm": 1.0839259295131853,
      "learning_rate": 7.437489291887157e-06,
      "loss": 0.5261,
      "step": 6640
    },
    {
      "epoch": 0.35837245696400627,
      "grad_norm": 1.0241333524541356,
      "learning_rate": 7.436800703523058e-06,
      "loss": 0.4799,
      "step": 6641
    },
    {
      "epoch": 0.35842642059252067,
      "grad_norm": 1.2342266809693268,
      "learning_rate": 7.436112059493349e-06,
      "loss": 0.4906,
      "step": 6642
    },
    {
      "epoch": 0.358480384221035,
      "grad_norm": 1.1129063157727925,
      "learning_rate": 7.435423359817826e-06,
      "loss": 0.7385,
      "step": 6643
    },
    {
      "epoch": 0.3585343478495494,
      "grad_norm": 0.9509114101784311,
      "learning_rate": 7.434734604516278e-06,
      "loss": 0.4911,
      "step": 6644
    },
    {
      "epoch": 0.3585883114780638,
      "grad_norm": 1.151058866627913,
      "learning_rate": 7.4340457936085016e-06,
      "loss": 0.5691,
      "step": 6645
    },
    {
      "epoch": 0.35864227510657815,
      "grad_norm": 0.8654342396644933,
      "learning_rate": 7.433356927114298e-06,
      "loss": 0.442,
      "step": 6646
    },
    {
      "epoch": 0.35869623873509254,
      "grad_norm": 1.0237116974661722,
      "learning_rate": 7.43266800505346e-06,
      "loss": 0.4589,
      "step": 6647
    },
    {
      "epoch": 0.35875020236360694,
      "grad_norm": 1.2367458811592502,
      "learning_rate": 7.431979027445792e-06,
      "loss": 0.6021,
      "step": 6648
    },
    {
      "epoch": 0.35880416599212134,
      "grad_norm": 1.0690113091558766,
      "learning_rate": 7.431289994311094e-06,
      "loss": 0.45,
      "step": 6649
    },
    {
      "epoch": 0.3588581296206357,
      "grad_norm": 1.317585001034018,
      "learning_rate": 7.4306009056691694e-06,
      "loss": 0.7059,
      "step": 6650
    },
    {
      "epoch": 0.3589120932491501,
      "grad_norm": 0.81493175067679,
      "learning_rate": 7.429911761539826e-06,
      "loss": 0.4105,
      "step": 6651
    },
    {
      "epoch": 0.3589660568776645,
      "grad_norm": 1.1712616130160447,
      "learning_rate": 7.429222561942867e-06,
      "loss": 0.6172,
      "step": 6652
    },
    {
      "epoch": 0.3590200205061788,
      "grad_norm": 0.9000987095954782,
      "learning_rate": 7.4285333068981035e-06,
      "loss": 0.3895,
      "step": 6653
    },
    {
      "epoch": 0.3590739841346932,
      "grad_norm": 1.1259739074898663,
      "learning_rate": 7.427843996425344e-06,
      "loss": 0.5202,
      "step": 6654
    },
    {
      "epoch": 0.3591279477632076,
      "grad_norm": 0.850591060161571,
      "learning_rate": 7.427154630544399e-06,
      "loss": 0.341,
      "step": 6655
    },
    {
      "epoch": 0.35918191139172195,
      "grad_norm": 0.9268074658266506,
      "learning_rate": 7.4264652092750836e-06,
      "loss": 0.3238,
      "step": 6656
    },
    {
      "epoch": 0.35923587502023635,
      "grad_norm": 0.8237024535448539,
      "learning_rate": 7.425775732637211e-06,
      "loss": 0.4207,
      "step": 6657
    },
    {
      "epoch": 0.35928983864875075,
      "grad_norm": 1.2149992370561,
      "learning_rate": 7.425086200650599e-06,
      "loss": 0.6533,
      "step": 6658
    },
    {
      "epoch": 0.35934380227726515,
      "grad_norm": 0.9600226648529733,
      "learning_rate": 7.424396613335062e-06,
      "loss": 0.5061,
      "step": 6659
    },
    {
      "epoch": 0.3593977659057795,
      "grad_norm": 0.9437161543975005,
      "learning_rate": 7.423706970710424e-06,
      "loss": 0.4779,
      "step": 6660
    },
    {
      "epoch": 0.3594517295342939,
      "grad_norm": 0.950617017309305,
      "learning_rate": 7.423017272796503e-06,
      "loss": 0.3685,
      "step": 6661
    },
    {
      "epoch": 0.3595056931628083,
      "grad_norm": 1.1192587061868684,
      "learning_rate": 7.422327519613123e-06,
      "loss": 0.5192,
      "step": 6662
    },
    {
      "epoch": 0.3595596567913226,
      "grad_norm": 1.0468027724457487,
      "learning_rate": 7.421637711180108e-06,
      "loss": 0.4558,
      "step": 6663
    },
    {
      "epoch": 0.359613620419837,
      "grad_norm": 1.0207220348239847,
      "learning_rate": 7.420947847517284e-06,
      "loss": 0.5137,
      "step": 6664
    },
    {
      "epoch": 0.3596675840483514,
      "grad_norm": 1.0022578132574254,
      "learning_rate": 7.420257928644477e-06,
      "loss": 0.4817,
      "step": 6665
    },
    {
      "epoch": 0.3597215476768658,
      "grad_norm": 1.0066828029688761,
      "learning_rate": 7.419567954581516e-06,
      "loss": 0.5416,
      "step": 6666
    },
    {
      "epoch": 0.35977551130538016,
      "grad_norm": 1.0927096085989094,
      "learning_rate": 7.418877925348234e-06,
      "loss": 0.5381,
      "step": 6667
    },
    {
      "epoch": 0.35982947493389456,
      "grad_norm": 0.936941111777973,
      "learning_rate": 7.41818784096446e-06,
      "loss": 0.5296,
      "step": 6668
    },
    {
      "epoch": 0.35988343856240895,
      "grad_norm": 1.163979528349895,
      "learning_rate": 7.41749770145003e-06,
      "loss": 0.4771,
      "step": 6669
    },
    {
      "epoch": 0.3599374021909233,
      "grad_norm": 1.0830612309611383,
      "learning_rate": 7.41680750682478e-06,
      "loss": 0.468,
      "step": 6670
    },
    {
      "epoch": 0.3599913658194377,
      "grad_norm": 0.8573546730914839,
      "learning_rate": 7.416117257108544e-06,
      "loss": 0.3908,
      "step": 6671
    },
    {
      "epoch": 0.3600453294479521,
      "grad_norm": 0.883265865964376,
      "learning_rate": 7.4154269523211616e-06,
      "loss": 0.5056,
      "step": 6672
    },
    {
      "epoch": 0.3600992930764665,
      "grad_norm": 1.0043717075995802,
      "learning_rate": 7.4147365924824735e-06,
      "loss": 0.445,
      "step": 6673
    },
    {
      "epoch": 0.36015325670498083,
      "grad_norm": 0.9320814636889937,
      "learning_rate": 7.414046177612321e-06,
      "loss": 0.4516,
      "step": 6674
    },
    {
      "epoch": 0.3602072203334952,
      "grad_norm": 1.0804595867182527,
      "learning_rate": 7.413355707730548e-06,
      "loss": 0.4971,
      "step": 6675
    },
    {
      "epoch": 0.3602611839620096,
      "grad_norm": 1.0582082211768769,
      "learning_rate": 7.412665182856999e-06,
      "loss": 0.4123,
      "step": 6676
    },
    {
      "epoch": 0.36031514759052397,
      "grad_norm": 0.6598923968975566,
      "learning_rate": 7.411974603011518e-06,
      "loss": 0.3254,
      "step": 6677
    },
    {
      "epoch": 0.36036911121903836,
      "grad_norm": 1.0818196534636064,
      "learning_rate": 7.411283968213956e-06,
      "loss": 0.5554,
      "step": 6678
    },
    {
      "epoch": 0.36042307484755276,
      "grad_norm": 0.9761849826331294,
      "learning_rate": 7.410593278484162e-06,
      "loss": 0.501,
      "step": 6679
    },
    {
      "epoch": 0.36047703847606716,
      "grad_norm": 1.1007499939513266,
      "learning_rate": 7.409902533841986e-06,
      "loss": 0.4952,
      "step": 6680
    },
    {
      "epoch": 0.3605310021045815,
      "grad_norm": 1.3077730535378476,
      "learning_rate": 7.409211734307282e-06,
      "loss": 0.742,
      "step": 6681
    },
    {
      "epoch": 0.3605849657330959,
      "grad_norm": 0.7882348467555511,
      "learning_rate": 7.408520879899903e-06,
      "loss": 0.4254,
      "step": 6682
    },
    {
      "epoch": 0.3606389293616103,
      "grad_norm": 1.0117600772958226,
      "learning_rate": 7.407829970639706e-06,
      "loss": 0.4548,
      "step": 6683
    },
    {
      "epoch": 0.36069289299012464,
      "grad_norm": 0.8551459933241918,
      "learning_rate": 7.407139006546547e-06,
      "loss": 0.3204,
      "step": 6684
    },
    {
      "epoch": 0.36074685661863903,
      "grad_norm": 0.9810011601361532,
      "learning_rate": 7.406447987640289e-06,
      "loss": 0.4861,
      "step": 6685
    },
    {
      "epoch": 0.36080082024715343,
      "grad_norm": 1.040858574485072,
      "learning_rate": 7.405756913940786e-06,
      "loss": 0.583,
      "step": 6686
    },
    {
      "epoch": 0.3608547838756678,
      "grad_norm": 1.0847896698081008,
      "learning_rate": 7.405065785467906e-06,
      "loss": 0.4994,
      "step": 6687
    },
    {
      "epoch": 0.36090874750418217,
      "grad_norm": 1.2445319816296931,
      "learning_rate": 7.404374602241508e-06,
      "loss": 0.4734,
      "step": 6688
    },
    {
      "epoch": 0.36096271113269657,
      "grad_norm": 1.054148207895995,
      "learning_rate": 7.403683364281461e-06,
      "loss": 0.5537,
      "step": 6689
    },
    {
      "epoch": 0.36101667476121097,
      "grad_norm": 1.1023535643928675,
      "learning_rate": 7.402992071607631e-06,
      "loss": 0.5372,
      "step": 6690
    },
    {
      "epoch": 0.3610706383897253,
      "grad_norm": 1.0589661849690708,
      "learning_rate": 7.402300724239887e-06,
      "loss": 0.4729,
      "step": 6691
    },
    {
      "epoch": 0.3611246020182397,
      "grad_norm": 0.9399510942983993,
      "learning_rate": 7.401609322198096e-06,
      "loss": 0.4094,
      "step": 6692
    },
    {
      "epoch": 0.3611785656467541,
      "grad_norm": 1.0725266899455026,
      "learning_rate": 7.400917865502132e-06,
      "loss": 0.419,
      "step": 6693
    },
    {
      "epoch": 0.36123252927526844,
      "grad_norm": 1.0747433337379437,
      "learning_rate": 7.4002263541718685e-06,
      "loss": 0.5058,
      "step": 6694
    },
    {
      "epoch": 0.36128649290378284,
      "grad_norm": 1.0717118611584169,
      "learning_rate": 7.39953478822718e-06,
      "loss": 0.4893,
      "step": 6695
    },
    {
      "epoch": 0.36134045653229724,
      "grad_norm": 0.9782585599158999,
      "learning_rate": 7.3988431676879405e-06,
      "loss": 0.5225,
      "step": 6696
    },
    {
      "epoch": 0.36139442016081164,
      "grad_norm": 1.0822589882995608,
      "learning_rate": 7.39815149257403e-06,
      "loss": 0.4938,
      "step": 6697
    },
    {
      "epoch": 0.361448383789326,
      "grad_norm": 1.0983818378486014,
      "learning_rate": 7.397459762905329e-06,
      "loss": 0.4546,
      "step": 6698
    },
    {
      "epoch": 0.3615023474178404,
      "grad_norm": 1.0132096328113913,
      "learning_rate": 7.3967679787017166e-06,
      "loss": 0.6245,
      "step": 6699
    },
    {
      "epoch": 0.3615563110463548,
      "grad_norm": 0.9391089371362723,
      "learning_rate": 7.396076139983076e-06,
      "loss": 0.4106,
      "step": 6700
    },
    {
      "epoch": 0.3616102746748691,
      "grad_norm": 0.9183376458775473,
      "learning_rate": 7.395384246769292e-06,
      "loss": 0.322,
      "step": 6701
    },
    {
      "epoch": 0.3616642383033835,
      "grad_norm": 1.1309536239002644,
      "learning_rate": 7.394692299080247e-06,
      "loss": 0.4504,
      "step": 6702
    },
    {
      "epoch": 0.3617182019318979,
      "grad_norm": 1.0681396062355577,
      "learning_rate": 7.394000296935833e-06,
      "loss": 0.5795,
      "step": 6703
    },
    {
      "epoch": 0.3617721655604123,
      "grad_norm": 0.9164680790865626,
      "learning_rate": 7.393308240355936e-06,
      "loss": 0.5164,
      "step": 6704
    },
    {
      "epoch": 0.36182612918892665,
      "grad_norm": 0.9206541736296995,
      "learning_rate": 7.392616129360448e-06,
      "loss": 0.4118,
      "step": 6705
    },
    {
      "epoch": 0.36188009281744105,
      "grad_norm": 1.215608484834527,
      "learning_rate": 7.39192396396926e-06,
      "loss": 0.5381,
      "step": 6706
    },
    {
      "epoch": 0.36193405644595544,
      "grad_norm": 0.9448301818952969,
      "learning_rate": 7.391231744202265e-06,
      "loss": 0.5094,
      "step": 6707
    },
    {
      "epoch": 0.3619880200744698,
      "grad_norm": 0.8599721981162372,
      "learning_rate": 7.390539470079361e-06,
      "loss": 0.4683,
      "step": 6708
    },
    {
      "epoch": 0.3620419837029842,
      "grad_norm": 1.0573677053051795,
      "learning_rate": 7.389847141620439e-06,
      "loss": 0.4512,
      "step": 6709
    },
    {
      "epoch": 0.3620959473314986,
      "grad_norm": 0.8777779451716963,
      "learning_rate": 7.3891547588454015e-06,
      "loss": 0.4748,
      "step": 6710
    },
    {
      "epoch": 0.362149910960013,
      "grad_norm": 0.8271504574280402,
      "learning_rate": 7.388462321774148e-06,
      "loss": 0.2948,
      "step": 6711
    },
    {
      "epoch": 0.3622038745885273,
      "grad_norm": 0.9427055373983112,
      "learning_rate": 7.387769830426579e-06,
      "loss": 0.4516,
      "step": 6712
    },
    {
      "epoch": 0.3622578382170417,
      "grad_norm": 1.0110359234713924,
      "learning_rate": 7.387077284822598e-06,
      "loss": 0.5316,
      "step": 6713
    },
    {
      "epoch": 0.3623118018455561,
      "grad_norm": 0.9097815869809164,
      "learning_rate": 7.3863846849821085e-06,
      "loss": 0.349,
      "step": 6714
    },
    {
      "epoch": 0.36236576547407046,
      "grad_norm": 1.1034829534612796,
      "learning_rate": 7.385692030925018e-06,
      "loss": 0.549,
      "step": 6715
    },
    {
      "epoch": 0.36241972910258485,
      "grad_norm": 1.197758370180457,
      "learning_rate": 7.384999322671232e-06,
      "loss": 0.4943,
      "step": 6716
    },
    {
      "epoch": 0.36247369273109925,
      "grad_norm": 0.9404653769179567,
      "learning_rate": 7.384306560240661e-06,
      "loss": 0.457,
      "step": 6717
    },
    {
      "epoch": 0.36252765635961365,
      "grad_norm": 1.0665030339637867,
      "learning_rate": 7.383613743653216e-06,
      "loss": 0.4754,
      "step": 6718
    },
    {
      "epoch": 0.362581619988128,
      "grad_norm": 1.2037917709838055,
      "learning_rate": 7.382920872928809e-06,
      "loss": 0.5744,
      "step": 6719
    },
    {
      "epoch": 0.3626355836166424,
      "grad_norm": 1.0682383961243649,
      "learning_rate": 7.382227948087354e-06,
      "loss": 0.6927,
      "step": 6720
    },
    {
      "epoch": 0.3626895472451568,
      "grad_norm": 0.9922792020421781,
      "learning_rate": 7.381534969148764e-06,
      "loss": 0.4456,
      "step": 6721
    },
    {
      "epoch": 0.3627435108736711,
      "grad_norm": 1.0411597229247016,
      "learning_rate": 7.38084193613296e-06,
      "loss": 0.4731,
      "step": 6722
    },
    {
      "epoch": 0.3627974745021855,
      "grad_norm": 0.9100601383430312,
      "learning_rate": 7.380148849059856e-06,
      "loss": 0.4141,
      "step": 6723
    },
    {
      "epoch": 0.3628514381306999,
      "grad_norm": 1.1851920921373118,
      "learning_rate": 7.379455707949376e-06,
      "loss": 0.5014,
      "step": 6724
    },
    {
      "epoch": 0.36290540175921426,
      "grad_norm": 1.4700881879327952,
      "learning_rate": 7.378762512821439e-06,
      "loss": 0.6874,
      "step": 6725
    },
    {
      "epoch": 0.36295936538772866,
      "grad_norm": 1.099857309494144,
      "learning_rate": 7.37806926369597e-06,
      "loss": 0.5948,
      "step": 6726
    },
    {
      "epoch": 0.36301332901624306,
      "grad_norm": 0.8385107764540245,
      "learning_rate": 7.377375960592892e-06,
      "loss": 0.3613,
      "step": 6727
    },
    {
      "epoch": 0.36306729264475746,
      "grad_norm": 1.1439698893188082,
      "learning_rate": 7.376682603532131e-06,
      "loss": 0.6483,
      "step": 6728
    },
    {
      "epoch": 0.3631212562732718,
      "grad_norm": 0.8628354532109298,
      "learning_rate": 7.375989192533619e-06,
      "loss": 0.4436,
      "step": 6729
    },
    {
      "epoch": 0.3631752199017862,
      "grad_norm": 1.0303970011329506,
      "learning_rate": 7.3752957276172784e-06,
      "loss": 0.4842,
      "step": 6730
    },
    {
      "epoch": 0.3632291835303006,
      "grad_norm": 1.0660396660635163,
      "learning_rate": 7.374602208803046e-06,
      "loss": 0.5123,
      "step": 6731
    },
    {
      "epoch": 0.36328314715881493,
      "grad_norm": 1.1877683390870848,
      "learning_rate": 7.3739086361108514e-06,
      "loss": 0.4013,
      "step": 6732
    },
    {
      "epoch": 0.36333711078732933,
      "grad_norm": 1.0748515356608208,
      "learning_rate": 7.373215009560628e-06,
      "loss": 0.5335,
      "step": 6733
    },
    {
      "epoch": 0.36339107441584373,
      "grad_norm": 0.9536344040334294,
      "learning_rate": 7.372521329172314e-06,
      "loss": 0.4339,
      "step": 6734
    },
    {
      "epoch": 0.3634450380443581,
      "grad_norm": 0.9361802437442717,
      "learning_rate": 7.371827594965845e-06,
      "loss": 0.3478,
      "step": 6735
    },
    {
      "epoch": 0.36349900167287247,
      "grad_norm": 1.051282707297924,
      "learning_rate": 7.371133806961159e-06,
      "loss": 0.4401,
      "step": 6736
    },
    {
      "epoch": 0.36355296530138687,
      "grad_norm": 1.526781427329277,
      "learning_rate": 7.370439965178197e-06,
      "loss": 0.6135,
      "step": 6737
    },
    {
      "epoch": 0.36360692892990126,
      "grad_norm": 0.9731838856606003,
      "learning_rate": 7.369746069636899e-06,
      "loss": 0.5292,
      "step": 6738
    },
    {
      "epoch": 0.3636608925584156,
      "grad_norm": 1.125094888019104,
      "learning_rate": 7.369052120357213e-06,
      "loss": 0.549,
      "step": 6739
    },
    {
      "epoch": 0.36371485618693,
      "grad_norm": 1.1012858175380296,
      "learning_rate": 7.368358117359077e-06,
      "loss": 0.5166,
      "step": 6740
    },
    {
      "epoch": 0.3637688198154444,
      "grad_norm": 1.0385585502371224,
      "learning_rate": 7.367664060662443e-06,
      "loss": 0.483,
      "step": 6741
    },
    {
      "epoch": 0.3638227834439588,
      "grad_norm": 0.9916849295281638,
      "learning_rate": 7.366969950287255e-06,
      "loss": 0.4475,
      "step": 6742
    },
    {
      "epoch": 0.36387674707247314,
      "grad_norm": 1.0764620030564624,
      "learning_rate": 7.366275786253467e-06,
      "loss": 0.6334,
      "step": 6743
    },
    {
      "epoch": 0.36393071070098754,
      "grad_norm": 0.9802143050498165,
      "learning_rate": 7.365581568581025e-06,
      "loss": 0.4284,
      "step": 6744
    },
    {
      "epoch": 0.36398467432950193,
      "grad_norm": 0.9968729323424695,
      "learning_rate": 7.364887297289885e-06,
      "loss": 0.543,
      "step": 6745
    },
    {
      "epoch": 0.3640386379580163,
      "grad_norm": 0.8621732472834347,
      "learning_rate": 7.364192972399999e-06,
      "loss": 0.4126,
      "step": 6746
    },
    {
      "epoch": 0.3640926015865307,
      "grad_norm": 1.0447857769164788,
      "learning_rate": 7.363498593931322e-06,
      "loss": 0.4791,
      "step": 6747
    },
    {
      "epoch": 0.36414656521504507,
      "grad_norm": 1.3142185446720986,
      "learning_rate": 7.362804161903814e-06,
      "loss": 0.4826,
      "step": 6748
    },
    {
      "epoch": 0.36420052884355947,
      "grad_norm": 1.4009547425371287,
      "learning_rate": 7.362109676337431e-06,
      "loss": 0.7531,
      "step": 6749
    },
    {
      "epoch": 0.3642544924720738,
      "grad_norm": 0.852144177773267,
      "learning_rate": 7.361415137252137e-06,
      "loss": 0.3479,
      "step": 6750
    },
    {
      "epoch": 0.3643084561005882,
      "grad_norm": 1.10314631748616,
      "learning_rate": 7.3607205446678885e-06,
      "loss": 0.4608,
      "step": 6751
    },
    {
      "epoch": 0.3643624197291026,
      "grad_norm": 0.8659837677213514,
      "learning_rate": 7.360025898604652e-06,
      "loss": 0.3841,
      "step": 6752
    },
    {
      "epoch": 0.36441638335761695,
      "grad_norm": 1.0278529941264452,
      "learning_rate": 7.359331199082391e-06,
      "loss": 0.429,
      "step": 6753
    },
    {
      "epoch": 0.36447034698613134,
      "grad_norm": 0.9328477401193067,
      "learning_rate": 7.3586364461210745e-06,
      "loss": 0.396,
      "step": 6754
    },
    {
      "epoch": 0.36452431061464574,
      "grad_norm": 0.9633088038107659,
      "learning_rate": 7.357941639740668e-06,
      "loss": 0.4499,
      "step": 6755
    },
    {
      "epoch": 0.3645782742431601,
      "grad_norm": 1.2395475181597273,
      "learning_rate": 7.357246779961141e-06,
      "loss": 0.5976,
      "step": 6756
    },
    {
      "epoch": 0.3646322378716745,
      "grad_norm": 1.0431985116152698,
      "learning_rate": 7.356551866802466e-06,
      "loss": 0.5749,
      "step": 6757
    },
    {
      "epoch": 0.3646862015001889,
      "grad_norm": 1.083703758947604,
      "learning_rate": 7.3558569002846125e-06,
      "loss": 0.4558,
      "step": 6758
    },
    {
      "epoch": 0.3647401651287033,
      "grad_norm": 1.055396242941038,
      "learning_rate": 7.355161880427557e-06,
      "loss": 0.602,
      "step": 6759
    },
    {
      "epoch": 0.3647941287572176,
      "grad_norm": 0.9180587924791588,
      "learning_rate": 7.354466807251276e-06,
      "loss": 0.411,
      "step": 6760
    },
    {
      "epoch": 0.364848092385732,
      "grad_norm": 0.8500713506775616,
      "learning_rate": 7.353771680775743e-06,
      "loss": 0.3704,
      "step": 6761
    },
    {
      "epoch": 0.3649020560142464,
      "grad_norm": 0.9362406625538126,
      "learning_rate": 7.3530765010209405e-06,
      "loss": 0.4435,
      "step": 6762
    },
    {
      "epoch": 0.36495601964276075,
      "grad_norm": 1.029283665363141,
      "learning_rate": 7.352381268006848e-06,
      "loss": 0.5332,
      "step": 6763
    },
    {
      "epoch": 0.36500998327127515,
      "grad_norm": 1.0327972417398112,
      "learning_rate": 7.3516859817534436e-06,
      "loss": 0.4883,
      "step": 6764
    },
    {
      "epoch": 0.36506394689978955,
      "grad_norm": 0.8603522833494767,
      "learning_rate": 7.3509906422807145e-06,
      "loss": 0.3843,
      "step": 6765
    },
    {
      "epoch": 0.36511791052830395,
      "grad_norm": 0.9824524103640179,
      "learning_rate": 7.350295249608642e-06,
      "loss": 0.4324,
      "step": 6766
    },
    {
      "epoch": 0.3651718741568183,
      "grad_norm": 0.899226797922604,
      "learning_rate": 7.349599803757217e-06,
      "loss": 0.4087,
      "step": 6767
    },
    {
      "epoch": 0.3652258377853327,
      "grad_norm": 0.9153123616413975,
      "learning_rate": 7.348904304746424e-06,
      "loss": 0.3655,
      "step": 6768
    },
    {
      "epoch": 0.3652798014138471,
      "grad_norm": 0.8767536905973499,
      "learning_rate": 7.348208752596253e-06,
      "loss": 0.4583,
      "step": 6769
    },
    {
      "epoch": 0.3653337650423614,
      "grad_norm": 0.7768088193198115,
      "learning_rate": 7.347513147326696e-06,
      "loss": 0.3936,
      "step": 6770
    },
    {
      "epoch": 0.3653877286708758,
      "grad_norm": 0.9988745331901796,
      "learning_rate": 7.346817488957742e-06,
      "loss": 0.396,
      "step": 6771
    },
    {
      "epoch": 0.3654416922993902,
      "grad_norm": 1.1541971983895656,
      "learning_rate": 7.34612177750939e-06,
      "loss": 0.5361,
      "step": 6772
    },
    {
      "epoch": 0.3654956559279046,
      "grad_norm": 1.079823346857731,
      "learning_rate": 7.345426013001631e-06,
      "loss": 0.5324,
      "step": 6773
    },
    {
      "epoch": 0.36554961955641896,
      "grad_norm": 1.0177048358787961,
      "learning_rate": 7.344730195454463e-06,
      "loss": 0.5315,
      "step": 6774
    },
    {
      "epoch": 0.36560358318493336,
      "grad_norm": 0.919720585672862,
      "learning_rate": 7.344034324887888e-06,
      "loss": 0.412,
      "step": 6775
    },
    {
      "epoch": 0.36565754681344775,
      "grad_norm": 0.9423818155170164,
      "learning_rate": 7.3433384013219e-06,
      "loss": 0.4551,
      "step": 6776
    },
    {
      "epoch": 0.3657115104419621,
      "grad_norm": 1.0032051801446293,
      "learning_rate": 7.342642424776505e-06,
      "loss": 0.488,
      "step": 6777
    },
    {
      "epoch": 0.3657654740704765,
      "grad_norm": 1.245420045383414,
      "learning_rate": 7.341946395271705e-06,
      "loss": 0.5852,
      "step": 6778
    },
    {
      "epoch": 0.3658194376989909,
      "grad_norm": 1.0175203319425785,
      "learning_rate": 7.341250312827506e-06,
      "loss": 0.3845,
      "step": 6779
    },
    {
      "epoch": 0.3658734013275053,
      "grad_norm": 1.0922527579502934,
      "learning_rate": 7.34055417746391e-06,
      "loss": 0.5221,
      "step": 6780
    },
    {
      "epoch": 0.36592736495601963,
      "grad_norm": 1.1655791005276739,
      "learning_rate": 7.3398579892009295e-06,
      "loss": 0.6484,
      "step": 6781
    },
    {
      "epoch": 0.365981328584534,
      "grad_norm": 1.2719842826560663,
      "learning_rate": 7.3391617480585695e-06,
      "loss": 0.5424,
      "step": 6782
    },
    {
      "epoch": 0.3660352922130484,
      "grad_norm": 1.1313166981019636,
      "learning_rate": 7.338465454056843e-06,
      "loss": 0.6351,
      "step": 6783
    },
    {
      "epoch": 0.36608925584156277,
      "grad_norm": 0.8634822717469182,
      "learning_rate": 7.337769107215763e-06,
      "loss": 0.3591,
      "step": 6784
    },
    {
      "epoch": 0.36614321947007716,
      "grad_norm": 0.8793026577942592,
      "learning_rate": 7.337072707555342e-06,
      "loss": 0.3948,
      "step": 6785
    },
    {
      "epoch": 0.36619718309859156,
      "grad_norm": 0.9692756615984774,
      "learning_rate": 7.336376255095592e-06,
      "loss": 0.4096,
      "step": 6786
    },
    {
      "epoch": 0.3662511467271059,
      "grad_norm": 1.018498899543461,
      "learning_rate": 7.335679749856536e-06,
      "loss": 0.5883,
      "step": 6787
    },
    {
      "epoch": 0.3663051103556203,
      "grad_norm": 0.879902373920261,
      "learning_rate": 7.334983191858188e-06,
      "loss": 0.3519,
      "step": 6788
    },
    {
      "epoch": 0.3663590739841347,
      "grad_norm": 1.3018838435427131,
      "learning_rate": 7.33428658112057e-06,
      "loss": 0.6822,
      "step": 6789
    },
    {
      "epoch": 0.3664130376126491,
      "grad_norm": 1.1609380989156672,
      "learning_rate": 7.3335899176637004e-06,
      "loss": 0.5357,
      "step": 6790
    },
    {
      "epoch": 0.36646700124116344,
      "grad_norm": 0.9938051622691001,
      "learning_rate": 7.332893201507605e-06,
      "loss": 0.4368,
      "step": 6791
    },
    {
      "epoch": 0.36652096486967783,
      "grad_norm": 1.0330275915559484,
      "learning_rate": 7.3321964326723075e-06,
      "loss": 0.5243,
      "step": 6792
    },
    {
      "epoch": 0.36657492849819223,
      "grad_norm": 0.9081434788455025,
      "learning_rate": 7.331499611177832e-06,
      "loss": 0.3684,
      "step": 6793
    },
    {
      "epoch": 0.3666288921267066,
      "grad_norm": 1.0061672223206344,
      "learning_rate": 7.330802737044207e-06,
      "loss": 0.5064,
      "step": 6794
    },
    {
      "epoch": 0.36668285575522097,
      "grad_norm": 1.0520981249052732,
      "learning_rate": 7.330105810291462e-06,
      "loss": 0.4099,
      "step": 6795
    },
    {
      "epoch": 0.36673681938373537,
      "grad_norm": 0.9068023570325237,
      "learning_rate": 7.329408830939627e-06,
      "loss": 0.4412,
      "step": 6796
    },
    {
      "epoch": 0.36679078301224977,
      "grad_norm": 0.8533863335025363,
      "learning_rate": 7.328711799008733e-06,
      "loss": 0.3613,
      "step": 6797
    },
    {
      "epoch": 0.3668447466407641,
      "grad_norm": 1.0696801033942818,
      "learning_rate": 7.328014714518816e-06,
      "loss": 0.5622,
      "step": 6798
    },
    {
      "epoch": 0.3668987102692785,
      "grad_norm": 1.0404485813062858,
      "learning_rate": 7.327317577489908e-06,
      "loss": 0.4903,
      "step": 6799
    },
    {
      "epoch": 0.3669526738977929,
      "grad_norm": 0.919094672518673,
      "learning_rate": 7.326620387942045e-06,
      "loss": 0.4139,
      "step": 6800
    },
    {
      "epoch": 0.36700663752630724,
      "grad_norm": 1.1653548100057403,
      "learning_rate": 7.325923145895268e-06,
      "loss": 0.5486,
      "step": 6801
    },
    {
      "epoch": 0.36706060115482164,
      "grad_norm": 1.1840614639628242,
      "learning_rate": 7.325225851369614e-06,
      "loss": 0.5902,
      "step": 6802
    },
    {
      "epoch": 0.36711456478333604,
      "grad_norm": 1.0502965559448754,
      "learning_rate": 7.3245285043851255e-06,
      "loss": 0.5338,
      "step": 6803
    },
    {
      "epoch": 0.36716852841185044,
      "grad_norm": 1.223795166334336,
      "learning_rate": 7.323831104961845e-06,
      "loss": 0.6876,
      "step": 6804
    },
    {
      "epoch": 0.3672224920403648,
      "grad_norm": 0.9326514393996463,
      "learning_rate": 7.323133653119814e-06,
      "loss": 0.4537,
      "step": 6805
    },
    {
      "epoch": 0.3672764556688792,
      "grad_norm": 0.8687848273581751,
      "learning_rate": 7.322436148879079e-06,
      "loss": 0.4288,
      "step": 6806
    },
    {
      "epoch": 0.3673304192973936,
      "grad_norm": 1.106684911721268,
      "learning_rate": 7.321738592259689e-06,
      "loss": 0.5906,
      "step": 6807
    },
    {
      "epoch": 0.3673843829259079,
      "grad_norm": 1.0910248129547129,
      "learning_rate": 7.321040983281692e-06,
      "loss": 0.6057,
      "step": 6808
    },
    {
      "epoch": 0.3674383465544223,
      "grad_norm": 1.223105410979182,
      "learning_rate": 7.320343321965136e-06,
      "loss": 0.5443,
      "step": 6809
    },
    {
      "epoch": 0.3674923101829367,
      "grad_norm": 1.0586091743606716,
      "learning_rate": 7.319645608330074e-06,
      "loss": 0.519,
      "step": 6810
    },
    {
      "epoch": 0.3675462738114511,
      "grad_norm": 0.8879723847207124,
      "learning_rate": 7.318947842396558e-06,
      "loss": 0.3314,
      "step": 6811
    },
    {
      "epoch": 0.36760023743996545,
      "grad_norm": 0.9014242052605601,
      "learning_rate": 7.3182500241846446e-06,
      "loss": 0.3639,
      "step": 6812
    },
    {
      "epoch": 0.36765420106847985,
      "grad_norm": 0.9510600066553361,
      "learning_rate": 7.3175521537143855e-06,
      "loss": 0.4216,
      "step": 6813
    },
    {
      "epoch": 0.36770816469699424,
      "grad_norm": 1.1350347922085335,
      "learning_rate": 7.316854231005842e-06,
      "loss": 0.4323,
      "step": 6814
    },
    {
      "epoch": 0.3677621283255086,
      "grad_norm": 1.0833143899584625,
      "learning_rate": 7.316156256079075e-06,
      "loss": 0.5838,
      "step": 6815
    },
    {
      "epoch": 0.367816091954023,
      "grad_norm": 1.2685845869287136,
      "learning_rate": 7.31545822895414e-06,
      "loss": 0.7139,
      "step": 6816
    },
    {
      "epoch": 0.3678700555825374,
      "grad_norm": 0.8756989809879977,
      "learning_rate": 7.314760149651099e-06,
      "loss": 0.4728,
      "step": 6817
    },
    {
      "epoch": 0.3679240192110518,
      "grad_norm": 0.9481467330858468,
      "learning_rate": 7.31406201819002e-06,
      "loss": 0.4305,
      "step": 6818
    },
    {
      "epoch": 0.3679779828395661,
      "grad_norm": 0.9881342203174235,
      "learning_rate": 7.3133638345909665e-06,
      "loss": 0.4666,
      "step": 6819
    },
    {
      "epoch": 0.3680319464680805,
      "grad_norm": 0.9640593117887198,
      "learning_rate": 7.312665598874001e-06,
      "loss": 0.4426,
      "step": 6820
    },
    {
      "epoch": 0.3680859100965949,
      "grad_norm": 1.0341250681327931,
      "learning_rate": 7.311967311059197e-06,
      "loss": 0.4275,
      "step": 6821
    },
    {
      "epoch": 0.36813987372510926,
      "grad_norm": 0.9227435855685743,
      "learning_rate": 7.3112689711666215e-06,
      "loss": 0.5436,
      "step": 6822
    },
    {
      "epoch": 0.36819383735362365,
      "grad_norm": 0.8982453947467975,
      "learning_rate": 7.310570579216345e-06,
      "loss": 0.4178,
      "step": 6823
    },
    {
      "epoch": 0.36824780098213805,
      "grad_norm": 1.063232771217943,
      "learning_rate": 7.309872135228441e-06,
      "loss": 0.5482,
      "step": 6824
    },
    {
      "epoch": 0.3683017646106524,
      "grad_norm": 0.8710197797161899,
      "learning_rate": 7.309173639222983e-06,
      "loss": 0.5107,
      "step": 6825
    },
    {
      "epoch": 0.3683557282391668,
      "grad_norm": 0.9562294351354688,
      "learning_rate": 7.308475091220047e-06,
      "loss": 0.4227,
      "step": 6826
    },
    {
      "epoch": 0.3684096918676812,
      "grad_norm": 0.8647796533499764,
      "learning_rate": 7.307776491239709e-06,
      "loss": 0.4426,
      "step": 6827
    },
    {
      "epoch": 0.3684636554961956,
      "grad_norm": 1.026384558891628,
      "learning_rate": 7.307077839302048e-06,
      "loss": 0.48,
      "step": 6828
    },
    {
      "epoch": 0.3685176191247099,
      "grad_norm": 0.9938155297542473,
      "learning_rate": 7.306379135427144e-06,
      "loss": 0.4332,
      "step": 6829
    },
    {
      "epoch": 0.3685715827532243,
      "grad_norm": 0.8363350697339719,
      "learning_rate": 7.305680379635078e-06,
      "loss": 0.3891,
      "step": 6830
    },
    {
      "epoch": 0.3686255463817387,
      "grad_norm": 1.0187486961096952,
      "learning_rate": 7.304981571945933e-06,
      "loss": 0.4412,
      "step": 6831
    },
    {
      "epoch": 0.36867951001025306,
      "grad_norm": 0.9464343226330632,
      "learning_rate": 7.304282712379796e-06,
      "loss": 0.3759,
      "step": 6832
    },
    {
      "epoch": 0.36873347363876746,
      "grad_norm": 1.1792880015819223,
      "learning_rate": 7.3035838009567495e-06,
      "loss": 0.5158,
      "step": 6833
    },
    {
      "epoch": 0.36878743726728186,
      "grad_norm": 0.9018161663533822,
      "learning_rate": 7.302884837696881e-06,
      "loss": 0.4473,
      "step": 6834
    },
    {
      "epoch": 0.36884140089579626,
      "grad_norm": 0.8056861465554771,
      "learning_rate": 7.302185822620281e-06,
      "loss": 0.3965,
      "step": 6835
    },
    {
      "epoch": 0.3688953645243106,
      "grad_norm": 1.1069205291340383,
      "learning_rate": 7.301486755747041e-06,
      "loss": 0.4831,
      "step": 6836
    },
    {
      "epoch": 0.368949328152825,
      "grad_norm": 0.8899513588353588,
      "learning_rate": 7.300787637097251e-06,
      "loss": 0.3716,
      "step": 6837
    },
    {
      "epoch": 0.3690032917813394,
      "grad_norm": 0.698910719433765,
      "learning_rate": 7.3000884666910045e-06,
      "loss": 0.2599,
      "step": 6838
    },
    {
      "epoch": 0.36905725540985374,
      "grad_norm": 1.0015956398595878,
      "learning_rate": 7.299389244548396e-06,
      "loss": 0.4188,
      "step": 6839
    },
    {
      "epoch": 0.36911121903836813,
      "grad_norm": 1.1282404778416344,
      "learning_rate": 7.298689970689522e-06,
      "loss": 0.5693,
      "step": 6840
    },
    {
      "epoch": 0.36916518266688253,
      "grad_norm": 0.9199500199619473,
      "learning_rate": 7.297990645134483e-06,
      "loss": 0.4674,
      "step": 6841
    },
    {
      "epoch": 0.3692191462953969,
      "grad_norm": 1.0239895821875582,
      "learning_rate": 7.2972912679033735e-06,
      "loss": 0.4962,
      "step": 6842
    },
    {
      "epoch": 0.36927310992391127,
      "grad_norm": 1.0135742176839524,
      "learning_rate": 7.296591839016299e-06,
      "loss": 0.5337,
      "step": 6843
    },
    {
      "epoch": 0.36932707355242567,
      "grad_norm": 1.0702536271566394,
      "learning_rate": 7.295892358493358e-06,
      "loss": 0.4959,
      "step": 6844
    },
    {
      "epoch": 0.36938103718094006,
      "grad_norm": 1.1999581757319084,
      "learning_rate": 7.295192826354657e-06,
      "loss": 0.5454,
      "step": 6845
    },
    {
      "epoch": 0.3694350008094544,
      "grad_norm": 0.9430599959552601,
      "learning_rate": 7.294493242620301e-06,
      "loss": 0.3899,
      "step": 6846
    },
    {
      "epoch": 0.3694889644379688,
      "grad_norm": 0.9974049073239135,
      "learning_rate": 7.293793607310396e-06,
      "loss": 0.4955,
      "step": 6847
    },
    {
      "epoch": 0.3695429280664832,
      "grad_norm": 0.9805359863885424,
      "learning_rate": 7.293093920445048e-06,
      "loss": 0.3468,
      "step": 6848
    },
    {
      "epoch": 0.3695968916949976,
      "grad_norm": 0.6685714629164718,
      "learning_rate": 7.292394182044371e-06,
      "loss": 0.2802,
      "step": 6849
    },
    {
      "epoch": 0.36965085532351194,
      "grad_norm": 1.1984384060536521,
      "learning_rate": 7.291694392128474e-06,
      "loss": 0.596,
      "step": 6850
    },
    {
      "epoch": 0.36970481895202634,
      "grad_norm": 1.0330509015130926,
      "learning_rate": 7.290994550717468e-06,
      "loss": 0.5566,
      "step": 6851
    },
    {
      "epoch": 0.36975878258054073,
      "grad_norm": 0.9591512299014374,
      "learning_rate": 7.290294657831471e-06,
      "loss": 0.3498,
      "step": 6852
    },
    {
      "epoch": 0.3698127462090551,
      "grad_norm": 0.9684663808235012,
      "learning_rate": 7.289594713490595e-06,
      "loss": 0.4258,
      "step": 6853
    },
    {
      "epoch": 0.3698667098375695,
      "grad_norm": 1.3749001239394132,
      "learning_rate": 7.28889471771496e-06,
      "loss": 0.6131,
      "step": 6854
    },
    {
      "epoch": 0.36992067346608387,
      "grad_norm": 1.1857924658353167,
      "learning_rate": 7.288194670524682e-06,
      "loss": 0.5378,
      "step": 6855
    },
    {
      "epoch": 0.3699746370945982,
      "grad_norm": 0.8308104950786264,
      "learning_rate": 7.287494571939884e-06,
      "loss": 0.3556,
      "step": 6856
    },
    {
      "epoch": 0.3700286007231126,
      "grad_norm": 0.9826560432428025,
      "learning_rate": 7.286794421980684e-06,
      "loss": 0.4588,
      "step": 6857
    },
    {
      "epoch": 0.370082564351627,
      "grad_norm": 1.0942386209436197,
      "learning_rate": 7.286094220667208e-06,
      "loss": 0.7319,
      "step": 6858
    },
    {
      "epoch": 0.3701365279801414,
      "grad_norm": 1.0793873734521517,
      "learning_rate": 7.28539396801958e-06,
      "loss": 0.539,
      "step": 6859
    },
    {
      "epoch": 0.37019049160865575,
      "grad_norm": 1.0280388303082675,
      "learning_rate": 7.284693664057923e-06,
      "loss": 0.4153,
      "step": 6860
    },
    {
      "epoch": 0.37024445523717014,
      "grad_norm": 0.9995849166646903,
      "learning_rate": 7.28399330880237e-06,
      "loss": 0.4356,
      "step": 6861
    },
    {
      "epoch": 0.37029841886568454,
      "grad_norm": 1.0563650370088837,
      "learning_rate": 7.2832929022730445e-06,
      "loss": 0.5329,
      "step": 6862
    },
    {
      "epoch": 0.3703523824941989,
      "grad_norm": 0.926292493214605,
      "learning_rate": 7.282592444490081e-06,
      "loss": 0.3783,
      "step": 6863
    },
    {
      "epoch": 0.3704063461227133,
      "grad_norm": 0.9241841036465496,
      "learning_rate": 7.281891935473607e-06,
      "loss": 0.3038,
      "step": 6864
    },
    {
      "epoch": 0.3704603097512277,
      "grad_norm": 1.0562348090282008,
      "learning_rate": 7.281191375243759e-06,
      "loss": 0.5533,
      "step": 6865
    },
    {
      "epoch": 0.3705142733797421,
      "grad_norm": 1.1850226921764784,
      "learning_rate": 7.280490763820671e-06,
      "loss": 0.4244,
      "step": 6866
    },
    {
      "epoch": 0.3705682370082564,
      "grad_norm": 1.2534287872455863,
      "learning_rate": 7.279790101224481e-06,
      "loss": 0.4805,
      "step": 6867
    },
    {
      "epoch": 0.3706222006367708,
      "grad_norm": 0.9391626210322489,
      "learning_rate": 7.2790893874753244e-06,
      "loss": 0.3739,
      "step": 6868
    },
    {
      "epoch": 0.3706761642652852,
      "grad_norm": 1.195295820641455,
      "learning_rate": 7.27838862259334e-06,
      "loss": 0.5032,
      "step": 6869
    },
    {
      "epoch": 0.37073012789379955,
      "grad_norm": 1.1516459235652914,
      "learning_rate": 7.277687806598671e-06,
      "loss": 0.5955,
      "step": 6870
    },
    {
      "epoch": 0.37078409152231395,
      "grad_norm": 0.9614959256784787,
      "learning_rate": 7.276986939511458e-06,
      "loss": 0.478,
      "step": 6871
    },
    {
      "epoch": 0.37083805515082835,
      "grad_norm": 1.0052149746700163,
      "learning_rate": 7.276286021351844e-06,
      "loss": 0.4797,
      "step": 6872
    },
    {
      "epoch": 0.37089201877934275,
      "grad_norm": 1.1422647860935524,
      "learning_rate": 7.275585052139975e-06,
      "loss": 0.5694,
      "step": 6873
    },
    {
      "epoch": 0.3709459824078571,
      "grad_norm": 1.1802395162122075,
      "learning_rate": 7.274884031895999e-06,
      "loss": 0.5292,
      "step": 6874
    },
    {
      "epoch": 0.3709999460363715,
      "grad_norm": 1.016477199622449,
      "learning_rate": 7.27418296064006e-06,
      "loss": 0.4325,
      "step": 6875
    },
    {
      "epoch": 0.3710539096648859,
      "grad_norm": 0.8485650411447343,
      "learning_rate": 7.273481838392312e-06,
      "loss": 0.3347,
      "step": 6876
    },
    {
      "epoch": 0.3711078732934002,
      "grad_norm": 1.2880122715545421,
      "learning_rate": 7.2727806651729045e-06,
      "loss": 0.6218,
      "step": 6877
    },
    {
      "epoch": 0.3711618369219146,
      "grad_norm": 1.3238818269936965,
      "learning_rate": 7.272079441001988e-06,
      "loss": 0.5616,
      "step": 6878
    },
    {
      "epoch": 0.371215800550429,
      "grad_norm": 0.9741192111599163,
      "learning_rate": 7.271378165899718e-06,
      "loss": 0.4729,
      "step": 6879
    },
    {
      "epoch": 0.3712697641789434,
      "grad_norm": 0.9497978842002315,
      "learning_rate": 7.270676839886251e-06,
      "loss": 0.4322,
      "step": 6880
    },
    {
      "epoch": 0.37132372780745776,
      "grad_norm": 0.9180193116411299,
      "learning_rate": 7.26997546298174e-06,
      "loss": 0.4834,
      "step": 6881
    },
    {
      "epoch": 0.37137769143597216,
      "grad_norm": 0.9810016952992067,
      "learning_rate": 7.269274035206347e-06,
      "loss": 0.3654,
      "step": 6882
    },
    {
      "epoch": 0.37143165506448655,
      "grad_norm": 0.9305524799307355,
      "learning_rate": 7.268572556580231e-06,
      "loss": 0.4371,
      "step": 6883
    },
    {
      "epoch": 0.3714856186930009,
      "grad_norm": 1.083050374895501,
      "learning_rate": 7.2678710271235545e-06,
      "loss": 0.512,
      "step": 6884
    },
    {
      "epoch": 0.3715395823215153,
      "grad_norm": 0.9474965006218362,
      "learning_rate": 7.267169446856475e-06,
      "loss": 0.4707,
      "step": 6885
    },
    {
      "epoch": 0.3715935459500297,
      "grad_norm": 1.3560878492325625,
      "learning_rate": 7.26646781579916e-06,
      "loss": 0.6223,
      "step": 6886
    },
    {
      "epoch": 0.3716475095785441,
      "grad_norm": 1.2013716884832917,
      "learning_rate": 7.265766133971777e-06,
      "loss": 0.6074,
      "step": 6887
    },
    {
      "epoch": 0.37170147320705843,
      "grad_norm": 0.8049158069036503,
      "learning_rate": 7.26506440139449e-06,
      "loss": 0.3855,
      "step": 6888
    },
    {
      "epoch": 0.37175543683557283,
      "grad_norm": 1.151713368184195,
      "learning_rate": 7.2643626180874705e-06,
      "loss": 0.6237,
      "step": 6889
    },
    {
      "epoch": 0.3718094004640872,
      "grad_norm": 1.1672530263551806,
      "learning_rate": 7.263660784070884e-06,
      "loss": 0.4497,
      "step": 6890
    },
    {
      "epoch": 0.37186336409260157,
      "grad_norm": 0.9076097961775188,
      "learning_rate": 7.262958899364907e-06,
      "loss": 0.3824,
      "step": 6891
    },
    {
      "epoch": 0.37191732772111596,
      "grad_norm": 1.2590068386288806,
      "learning_rate": 7.262256963989709e-06,
      "loss": 0.6934,
      "step": 6892
    },
    {
      "epoch": 0.37197129134963036,
      "grad_norm": 1.0511018630651543,
      "learning_rate": 7.261554977965464e-06,
      "loss": 0.5735,
      "step": 6893
    },
    {
      "epoch": 0.3720252549781447,
      "grad_norm": 0.8467117673473454,
      "learning_rate": 7.260852941312351e-06,
      "loss": 0.487,
      "step": 6894
    },
    {
      "epoch": 0.3720792186066591,
      "grad_norm": 1.1022197033192875,
      "learning_rate": 7.260150854050543e-06,
      "loss": 0.5453,
      "step": 6895
    },
    {
      "epoch": 0.3721331822351735,
      "grad_norm": 1.2078930120898936,
      "learning_rate": 7.259448716200223e-06,
      "loss": 0.5233,
      "step": 6896
    },
    {
      "epoch": 0.3721871458636879,
      "grad_norm": 1.036284215848604,
      "learning_rate": 7.258746527781567e-06,
      "loss": 0.5514,
      "step": 6897
    },
    {
      "epoch": 0.37224110949220224,
      "grad_norm": 1.1762736776713212,
      "learning_rate": 7.258044288814761e-06,
      "loss": 0.569,
      "step": 6898
    },
    {
      "epoch": 0.37229507312071664,
      "grad_norm": 0.9654807400439736,
      "learning_rate": 7.257341999319985e-06,
      "loss": 0.3936,
      "step": 6899
    },
    {
      "epoch": 0.37234903674923103,
      "grad_norm": 0.872319520455724,
      "learning_rate": 7.256639659317424e-06,
      "loss": 0.3645,
      "step": 6900
    },
    {
      "epoch": 0.3724030003777454,
      "grad_norm": 0.8793321009032167,
      "learning_rate": 7.255937268827264e-06,
      "loss": 0.4179,
      "step": 6901
    },
    {
      "epoch": 0.37245696400625977,
      "grad_norm": 0.6308134596303165,
      "learning_rate": 7.255234827869693e-06,
      "loss": 0.2526,
      "step": 6902
    },
    {
      "epoch": 0.37251092763477417,
      "grad_norm": 1.1208581845341734,
      "learning_rate": 7.254532336464899e-06,
      "loss": 0.3858,
      "step": 6903
    },
    {
      "epoch": 0.37256489126328857,
      "grad_norm": 1.0487431509815637,
      "learning_rate": 7.253829794633073e-06,
      "loss": 0.4305,
      "step": 6904
    },
    {
      "epoch": 0.3726188548918029,
      "grad_norm": 0.9606937545923605,
      "learning_rate": 7.253127202394407e-06,
      "loss": 0.5142,
      "step": 6905
    },
    {
      "epoch": 0.3726728185203173,
      "grad_norm": 1.1876231003642062,
      "learning_rate": 7.252424559769092e-06,
      "loss": 0.4935,
      "step": 6906
    },
    {
      "epoch": 0.3727267821488317,
      "grad_norm": 1.1160672901766309,
      "learning_rate": 7.251721866777326e-06,
      "loss": 0.4706,
      "step": 6907
    },
    {
      "epoch": 0.37278074577734605,
      "grad_norm": 1.427491097467988,
      "learning_rate": 7.251019123439305e-06,
      "loss": 0.6918,
      "step": 6908
    },
    {
      "epoch": 0.37283470940586044,
      "grad_norm": 0.8796965688808642,
      "learning_rate": 7.250316329775223e-06,
      "loss": 0.442,
      "step": 6909
    },
    {
      "epoch": 0.37288867303437484,
      "grad_norm": 1.0571475014453204,
      "learning_rate": 7.249613485805282e-06,
      "loss": 0.5762,
      "step": 6910
    },
    {
      "epoch": 0.37294263666288924,
      "grad_norm": 0.8377632021848902,
      "learning_rate": 7.2489105915496805e-06,
      "loss": 0.3352,
      "step": 6911
    },
    {
      "epoch": 0.3729966002914036,
      "grad_norm": 1.2836702198123393,
      "learning_rate": 7.248207647028624e-06,
      "loss": 0.5446,
      "step": 6912
    },
    {
      "epoch": 0.373050563919918,
      "grad_norm": 1.048118427575878,
      "learning_rate": 7.247504652262312e-06,
      "loss": 0.4362,
      "step": 6913
    },
    {
      "epoch": 0.3731045275484324,
      "grad_norm": 1.0222376806806541,
      "learning_rate": 7.246801607270949e-06,
      "loss": 0.4374,
      "step": 6914
    },
    {
      "epoch": 0.3731584911769467,
      "grad_norm": 0.9966544161430414,
      "learning_rate": 7.246098512074745e-06,
      "loss": 0.4106,
      "step": 6915
    },
    {
      "epoch": 0.3732124548054611,
      "grad_norm": 0.9545944967871358,
      "learning_rate": 7.245395366693903e-06,
      "loss": 0.4766,
      "step": 6916
    },
    {
      "epoch": 0.3732664184339755,
      "grad_norm": 0.9999704765408531,
      "learning_rate": 7.2446921711486364e-06,
      "loss": 0.3495,
      "step": 6917
    },
    {
      "epoch": 0.3733203820624899,
      "grad_norm": 1.310731377898673,
      "learning_rate": 7.243988925459154e-06,
      "loss": 0.6372,
      "step": 6918
    },
    {
      "epoch": 0.37337434569100425,
      "grad_norm": 0.9129002855118924,
      "learning_rate": 7.243285629645667e-06,
      "loss": 0.403,
      "step": 6919
    },
    {
      "epoch": 0.37342830931951865,
      "grad_norm": 0.9680231435862053,
      "learning_rate": 7.24258228372839e-06,
      "loss": 0.4785,
      "step": 6920
    },
    {
      "epoch": 0.37348227294803304,
      "grad_norm": 1.0513572833424338,
      "learning_rate": 7.241878887727537e-06,
      "loss": 0.3996,
      "step": 6921
    },
    {
      "epoch": 0.3735362365765474,
      "grad_norm": 1.3077776428608345,
      "learning_rate": 7.2411754416633255e-06,
      "loss": 0.5792,
      "step": 6922
    },
    {
      "epoch": 0.3735902002050618,
      "grad_norm": 0.8118315962513533,
      "learning_rate": 7.240471945555971e-06,
      "loss": 0.5007,
      "step": 6923
    },
    {
      "epoch": 0.3736441638335762,
      "grad_norm": 0.9665198880486371,
      "learning_rate": 7.239768399425695e-06,
      "loss": 0.3869,
      "step": 6924
    },
    {
      "epoch": 0.3736981274620905,
      "grad_norm": 1.063033097673364,
      "learning_rate": 7.239064803292718e-06,
      "loss": 0.5949,
      "step": 6925
    },
    {
      "epoch": 0.3737520910906049,
      "grad_norm": 0.7361857673166283,
      "learning_rate": 7.238361157177261e-06,
      "loss": 0.2602,
      "step": 6926
    },
    {
      "epoch": 0.3738060547191193,
      "grad_norm": 0.9562627325551762,
      "learning_rate": 7.2376574610995475e-06,
      "loss": 0.4432,
      "step": 6927
    },
    {
      "epoch": 0.3738600183476337,
      "grad_norm": 0.9847762688950488,
      "learning_rate": 7.2369537150798035e-06,
      "loss": 0.4827,
      "step": 6928
    },
    {
      "epoch": 0.37391398197614806,
      "grad_norm": 0.8316062011741119,
      "learning_rate": 7.236249919138253e-06,
      "loss": 0.3833,
      "step": 6929
    },
    {
      "epoch": 0.37396794560466246,
      "grad_norm": 1.277086817830862,
      "learning_rate": 7.235546073295127e-06,
      "loss": 0.6237,
      "step": 6930
    },
    {
      "epoch": 0.37402190923317685,
      "grad_norm": 1.0772332081083147,
      "learning_rate": 7.234842177570653e-06,
      "loss": 0.5001,
      "step": 6931
    },
    {
      "epoch": 0.3740758728616912,
      "grad_norm": 1.0876688477877543,
      "learning_rate": 7.2341382319850606e-06,
      "loss": 0.6204,
      "step": 6932
    },
    {
      "epoch": 0.3741298364902056,
      "grad_norm": 0.9715396695637002,
      "learning_rate": 7.2334342365585845e-06,
      "loss": 0.5307,
      "step": 6933
    },
    {
      "epoch": 0.37418380011872,
      "grad_norm": 0.8099095639290982,
      "learning_rate": 7.232730191311456e-06,
      "loss": 0.4243,
      "step": 6934
    },
    {
      "epoch": 0.3742377637472344,
      "grad_norm": 1.1073060570472009,
      "learning_rate": 7.232026096263911e-06,
      "loss": 0.6736,
      "step": 6935
    },
    {
      "epoch": 0.37429172737574873,
      "grad_norm": 0.9519820450138246,
      "learning_rate": 7.2313219514361875e-06,
      "loss": 0.457,
      "step": 6936
    },
    {
      "epoch": 0.3743456910042631,
      "grad_norm": 0.8273514978522809,
      "learning_rate": 7.230617756848521e-06,
      "loss": 0.3054,
      "step": 6937
    },
    {
      "epoch": 0.3743996546327775,
      "grad_norm": 1.1390732580964629,
      "learning_rate": 7.2299135125211505e-06,
      "loss": 0.5186,
      "step": 6938
    },
    {
      "epoch": 0.37445361826129187,
      "grad_norm": 1.0343797298995598,
      "learning_rate": 7.22920921847432e-06,
      "loss": 0.5556,
      "step": 6939
    },
    {
      "epoch": 0.37450758188980626,
      "grad_norm": 1.1886789778263926,
      "learning_rate": 7.228504874728266e-06,
      "loss": 0.5518,
      "step": 6940
    },
    {
      "epoch": 0.37456154551832066,
      "grad_norm": 0.9866180615249266,
      "learning_rate": 7.227800481303238e-06,
      "loss": 0.5597,
      "step": 6941
    },
    {
      "epoch": 0.37461550914683506,
      "grad_norm": 1.223121878808218,
      "learning_rate": 7.227096038219477e-06,
      "loss": 0.5435,
      "step": 6942
    },
    {
      "epoch": 0.3746694727753494,
      "grad_norm": 1.140965152571399,
      "learning_rate": 7.226391545497232e-06,
      "loss": 0.5376,
      "step": 6943
    },
    {
      "epoch": 0.3747234364038638,
      "grad_norm": 0.852521675509372,
      "learning_rate": 7.225687003156749e-06,
      "loss": 0.3093,
      "step": 6944
    },
    {
      "epoch": 0.3747774000323782,
      "grad_norm": 0.9306056139332489,
      "learning_rate": 7.224982411218278e-06,
      "loss": 0.4905,
      "step": 6945
    },
    {
      "epoch": 0.37483136366089254,
      "grad_norm": 1.1901022228595592,
      "learning_rate": 7.22427776970207e-06,
      "loss": 0.4879,
      "step": 6946
    },
    {
      "epoch": 0.37488532728940693,
      "grad_norm": 0.8133459499009361,
      "learning_rate": 7.223573078628375e-06,
      "loss": 0.3526,
      "step": 6947
    },
    {
      "epoch": 0.37493929091792133,
      "grad_norm": 0.9415855804708764,
      "learning_rate": 7.222868338017449e-06,
      "loss": 0.4503,
      "step": 6948
    },
    {
      "epoch": 0.37499325454643573,
      "grad_norm": 0.9265492366376306,
      "learning_rate": 7.222163547889546e-06,
      "loss": 0.4509,
      "step": 6949
    },
    {
      "epoch": 0.37504721817495007,
      "grad_norm": 1.3150095320973068,
      "learning_rate": 7.221458708264923e-06,
      "loss": 0.5171,
      "step": 6950
    },
    {
      "epoch": 0.37510118180346447,
      "grad_norm": 1.2551290560767592,
      "learning_rate": 7.220753819163838e-06,
      "loss": 0.5046,
      "step": 6951
    },
    {
      "epoch": 0.37515514543197886,
      "grad_norm": 0.898760396950055,
      "learning_rate": 7.220048880606548e-06,
      "loss": 0.3871,
      "step": 6952
    },
    {
      "epoch": 0.3752091090604932,
      "grad_norm": 1.1307511394296448,
      "learning_rate": 7.219343892613316e-06,
      "loss": 0.6383,
      "step": 6953
    },
    {
      "epoch": 0.3752630726890076,
      "grad_norm": 1.1146084760259063,
      "learning_rate": 7.218638855204403e-06,
      "loss": 0.3827,
      "step": 6954
    },
    {
      "epoch": 0.375317036317522,
      "grad_norm": 0.9600914785032998,
      "learning_rate": 7.217933768400072e-06,
      "loss": 0.4991,
      "step": 6955
    },
    {
      "epoch": 0.3753709999460364,
      "grad_norm": 0.8010863772686861,
      "learning_rate": 7.21722863222059e-06,
      "loss": 0.4678,
      "step": 6956
    },
    {
      "epoch": 0.37542496357455074,
      "grad_norm": 0.9864037905308912,
      "learning_rate": 7.216523446686221e-06,
      "loss": 0.4681,
      "step": 6957
    },
    {
      "epoch": 0.37547892720306514,
      "grad_norm": 0.9902117511594252,
      "learning_rate": 7.215818211817235e-06,
      "loss": 0.4924,
      "step": 6958
    },
    {
      "epoch": 0.37553289083157954,
      "grad_norm": 0.8551210867602257,
      "learning_rate": 7.215112927633898e-06,
      "loss": 0.4244,
      "step": 6959
    },
    {
      "epoch": 0.3755868544600939,
      "grad_norm": 1.18228049747753,
      "learning_rate": 7.2144075941564835e-06,
      "loss": 0.5987,
      "step": 6960
    },
    {
      "epoch": 0.3756408180886083,
      "grad_norm": 1.1942132227087285,
      "learning_rate": 7.213702211405263e-06,
      "loss": 0.5088,
      "step": 6961
    },
    {
      "epoch": 0.37569478171712267,
      "grad_norm": 1.0980434364519334,
      "learning_rate": 7.212996779400508e-06,
      "loss": 0.5012,
      "step": 6962
    },
    {
      "epoch": 0.375748745345637,
      "grad_norm": 0.8752062709162536,
      "learning_rate": 7.212291298162497e-06,
      "loss": 0.3579,
      "step": 6963
    },
    {
      "epoch": 0.3758027089741514,
      "grad_norm": 1.0596314658242807,
      "learning_rate": 7.211585767711502e-06,
      "loss": 0.488,
      "step": 6964
    },
    {
      "epoch": 0.3758566726026658,
      "grad_norm": 0.8897832942721821,
      "learning_rate": 7.210880188067803e-06,
      "loss": 0.4717,
      "step": 6965
    },
    {
      "epoch": 0.3759106362311802,
      "grad_norm": 1.1883846321806144,
      "learning_rate": 7.210174559251677e-06,
      "loss": 0.5669,
      "step": 6966
    },
    {
      "epoch": 0.37596459985969455,
      "grad_norm": 1.2266080684763248,
      "learning_rate": 7.209468881283409e-06,
      "loss": 0.5221,
      "step": 6967
    },
    {
      "epoch": 0.37601856348820895,
      "grad_norm": 0.9804230657604749,
      "learning_rate": 7.208763154183277e-06,
      "loss": 0.472,
      "step": 6968
    },
    {
      "epoch": 0.37607252711672334,
      "grad_norm": 1.122334968323653,
      "learning_rate": 7.208057377971565e-06,
      "loss": 0.5787,
      "step": 6969
    },
    {
      "epoch": 0.3761264907452377,
      "grad_norm": 0.9331703277946313,
      "learning_rate": 7.207351552668559e-06,
      "loss": 0.4536,
      "step": 6970
    },
    {
      "epoch": 0.3761804543737521,
      "grad_norm": 0.9659017422710044,
      "learning_rate": 7.2066456782945435e-06,
      "loss": 0.504,
      "step": 6971
    },
    {
      "epoch": 0.3762344180022665,
      "grad_norm": 1.100575609088681,
      "learning_rate": 7.205939754869807e-06,
      "loss": 0.5687,
      "step": 6972
    },
    {
      "epoch": 0.3762883816307809,
      "grad_norm": 0.6354951599506269,
      "learning_rate": 7.205233782414638e-06,
      "loss": 0.2376,
      "step": 6973
    },
    {
      "epoch": 0.3763423452592952,
      "grad_norm": 1.055794778160967,
      "learning_rate": 7.204527760949328e-06,
      "loss": 0.6155,
      "step": 6974
    },
    {
      "epoch": 0.3763963088878096,
      "grad_norm": 1.0610542222257762,
      "learning_rate": 7.203821690494166e-06,
      "loss": 0.5295,
      "step": 6975
    },
    {
      "epoch": 0.376450272516324,
      "grad_norm": 0.9284585272566351,
      "learning_rate": 7.203115571069449e-06,
      "loss": 0.4823,
      "step": 6976
    },
    {
      "epoch": 0.37650423614483836,
      "grad_norm": 1.1248439335483837,
      "learning_rate": 7.20240940269547e-06,
      "loss": 0.6971,
      "step": 6977
    },
    {
      "epoch": 0.37655819977335275,
      "grad_norm": 0.8955474565088192,
      "learning_rate": 7.201703185392523e-06,
      "loss": 0.4394,
      "step": 6978
    },
    {
      "epoch": 0.37661216340186715,
      "grad_norm": 0.8675733039893668,
      "learning_rate": 7.200996919180906e-06,
      "loss": 0.364,
      "step": 6979
    },
    {
      "epoch": 0.37666612703038155,
      "grad_norm": 0.8343671115061105,
      "learning_rate": 7.200290604080919e-06,
      "loss": 0.3756,
      "step": 6980
    },
    {
      "epoch": 0.3767200906588959,
      "grad_norm": 0.8942856079959411,
      "learning_rate": 7.199584240112865e-06,
      "loss": 0.3729,
      "step": 6981
    },
    {
      "epoch": 0.3767740542874103,
      "grad_norm": 1.0885419410682318,
      "learning_rate": 7.198877827297039e-06,
      "loss": 0.5373,
      "step": 6982
    },
    {
      "epoch": 0.3768280179159247,
      "grad_norm": 0.9875712028888981,
      "learning_rate": 7.19817136565375e-06,
      "loss": 0.4567,
      "step": 6983
    },
    {
      "epoch": 0.376881981544439,
      "grad_norm": 1.2221349923454623,
      "learning_rate": 7.1974648552033e-06,
      "loss": 0.6445,
      "step": 6984
    },
    {
      "epoch": 0.3769359451729534,
      "grad_norm": 1.0718006483040892,
      "learning_rate": 7.196758295965994e-06,
      "loss": 0.494,
      "step": 6985
    },
    {
      "epoch": 0.3769899088014678,
      "grad_norm": 0.9588058399856357,
      "learning_rate": 7.1960516879621415e-06,
      "loss": 0.4885,
      "step": 6986
    },
    {
      "epoch": 0.3770438724299822,
      "grad_norm": 0.9238615292036465,
      "learning_rate": 7.195345031212048e-06,
      "loss": 0.4155,
      "step": 6987
    },
    {
      "epoch": 0.37709783605849656,
      "grad_norm": 1.1020279510813313,
      "learning_rate": 7.1946383257360264e-06,
      "loss": 0.5185,
      "step": 6988
    },
    {
      "epoch": 0.37715179968701096,
      "grad_norm": 1.1048907541370179,
      "learning_rate": 7.193931571554387e-06,
      "loss": 0.4229,
      "step": 6989
    },
    {
      "epoch": 0.37720576331552536,
      "grad_norm": 1.0209909538555413,
      "learning_rate": 7.193224768687443e-06,
      "loss": 0.5509,
      "step": 6990
    },
    {
      "epoch": 0.3772597269440397,
      "grad_norm": 1.039764491174435,
      "learning_rate": 7.1925179171555075e-06,
      "loss": 0.6809,
      "step": 6991
    },
    {
      "epoch": 0.3773136905725541,
      "grad_norm": 0.8557102767080685,
      "learning_rate": 7.191811016978896e-06,
      "loss": 0.4039,
      "step": 6992
    },
    {
      "epoch": 0.3773676542010685,
      "grad_norm": 1.0941310598784544,
      "learning_rate": 7.191104068177927e-06,
      "loss": 0.5908,
      "step": 6993
    },
    {
      "epoch": 0.37742161782958283,
      "grad_norm": 0.9379140360496497,
      "learning_rate": 7.19039707077292e-06,
      "loss": 0.4297,
      "step": 6994
    },
    {
      "epoch": 0.37747558145809723,
      "grad_norm": 0.9177692414666432,
      "learning_rate": 7.189690024784192e-06,
      "loss": 0.3998,
      "step": 6995
    },
    {
      "epoch": 0.37752954508661163,
      "grad_norm": 1.1018460862449784,
      "learning_rate": 7.188982930232066e-06,
      "loss": 0.3711,
      "step": 6996
    },
    {
      "epoch": 0.377583508715126,
      "grad_norm": 1.1596327361759011,
      "learning_rate": 7.188275787136862e-06,
      "loss": 0.5013,
      "step": 6997
    },
    {
      "epoch": 0.37763747234364037,
      "grad_norm": 1.0455865358557612,
      "learning_rate": 7.187568595518908e-06,
      "loss": 0.4654,
      "step": 6998
    },
    {
      "epoch": 0.37769143597215477,
      "grad_norm": 0.8575421319902543,
      "learning_rate": 7.186861355398526e-06,
      "loss": 0.3939,
      "step": 6999
    },
    {
      "epoch": 0.37774539960066916,
      "grad_norm": 1.1218358121342649,
      "learning_rate": 7.186154066796045e-06,
      "loss": 0.5401,
      "step": 7000
    },
    {
      "epoch": 0.37774539960066916,
      "eval_loss": 0.5605905055999756,
      "eval_runtime": 162.8616,
      "eval_samples_per_second": 21.116,
      "eval_steps_per_second": 0.884,
      "step": 7000
    },
    {
      "epoch": 0.3777993632291835,
      "grad_norm": 1.1560634531374228,
      "learning_rate": 7.1854467297317935e-06,
      "loss": 0.4905,
      "step": 7001
    },
    {
      "epoch": 0.3778533268576979,
      "grad_norm": 1.0053200433683418,
      "learning_rate": 7.184739344226098e-06,
      "loss": 0.5393,
      "step": 7002
    },
    {
      "epoch": 0.3779072904862123,
      "grad_norm": 1.1615479575939966,
      "learning_rate": 7.184031910299292e-06,
      "loss": 0.587,
      "step": 7003
    },
    {
      "epoch": 0.3779612541147267,
      "grad_norm": 1.0279025794405205,
      "learning_rate": 7.183324427971707e-06,
      "loss": 0.5261,
      "step": 7004
    },
    {
      "epoch": 0.37801521774324104,
      "grad_norm": 0.9725617125236297,
      "learning_rate": 7.182616897263678e-06,
      "loss": 0.5119,
      "step": 7005
    },
    {
      "epoch": 0.37806918137175544,
      "grad_norm": 1.1637327927669623,
      "learning_rate": 7.181909318195539e-06,
      "loss": 0.5084,
      "step": 7006
    },
    {
      "epoch": 0.37812314500026983,
      "grad_norm": 1.2187877283703035,
      "learning_rate": 7.181201690787626e-06,
      "loss": 0.5985,
      "step": 7007
    },
    {
      "epoch": 0.3781771086287842,
      "grad_norm": 1.1171912107555106,
      "learning_rate": 7.180494015060278e-06,
      "loss": 0.4276,
      "step": 7008
    },
    {
      "epoch": 0.3782310722572986,
      "grad_norm": 0.8828222769684416,
      "learning_rate": 7.179786291033834e-06,
      "loss": 0.3548,
      "step": 7009
    },
    {
      "epoch": 0.37828503588581297,
      "grad_norm": 0.8914654082081784,
      "learning_rate": 7.179078518728635e-06,
      "loss": 0.3229,
      "step": 7010
    },
    {
      "epoch": 0.37833899951432737,
      "grad_norm": 0.9884145728287924,
      "learning_rate": 7.178370698165022e-06,
      "loss": 0.485,
      "step": 7011
    },
    {
      "epoch": 0.3783929631428417,
      "grad_norm": 1.042006120200116,
      "learning_rate": 7.177662829363342e-06,
      "loss": 0.6208,
      "step": 7012
    },
    {
      "epoch": 0.3784469267713561,
      "grad_norm": 1.0842866049885342,
      "learning_rate": 7.176954912343934e-06,
      "loss": 0.5097,
      "step": 7013
    },
    {
      "epoch": 0.3785008903998705,
      "grad_norm": 1.17449811129165,
      "learning_rate": 7.176246947127148e-06,
      "loss": 0.6594,
      "step": 7014
    },
    {
      "epoch": 0.37855485402838485,
      "grad_norm": 0.9878203238988538,
      "learning_rate": 7.175538933733332e-06,
      "loss": 0.4787,
      "step": 7015
    },
    {
      "epoch": 0.37860881765689924,
      "grad_norm": 0.959912623142717,
      "learning_rate": 7.174830872182833e-06,
      "loss": 0.4025,
      "step": 7016
    },
    {
      "epoch": 0.37866278128541364,
      "grad_norm": 0.9555376274760874,
      "learning_rate": 7.174122762496003e-06,
      "loss": 0.6747,
      "step": 7017
    },
    {
      "epoch": 0.37871674491392804,
      "grad_norm": 1.0175567596831172,
      "learning_rate": 7.173414604693193e-06,
      "loss": 0.4455,
      "step": 7018
    },
    {
      "epoch": 0.3787707085424424,
      "grad_norm": 0.8651521471195998,
      "learning_rate": 7.172706398794758e-06,
      "loss": 0.4644,
      "step": 7019
    },
    {
      "epoch": 0.3788246721709568,
      "grad_norm": 0.9310049279941783,
      "learning_rate": 7.171998144821049e-06,
      "loss": 0.4601,
      "step": 7020
    },
    {
      "epoch": 0.3788786357994712,
      "grad_norm": 1.1056088395125412,
      "learning_rate": 7.1712898427924235e-06,
      "loss": 0.5281,
      "step": 7021
    },
    {
      "epoch": 0.3789325994279855,
      "grad_norm": 0.9711011152256366,
      "learning_rate": 7.170581492729242e-06,
      "loss": 0.3924,
      "step": 7022
    },
    {
      "epoch": 0.3789865630564999,
      "grad_norm": 1.0036545459654718,
      "learning_rate": 7.169873094651858e-06,
      "loss": 0.3833,
      "step": 7023
    },
    {
      "epoch": 0.3790405266850143,
      "grad_norm": 1.251563186491286,
      "learning_rate": 7.169164648580633e-06,
      "loss": 0.5144,
      "step": 7024
    },
    {
      "epoch": 0.3790944903135287,
      "grad_norm": 1.0437386377767044,
      "learning_rate": 7.168456154535931e-06,
      "loss": 0.6036,
      "step": 7025
    },
    {
      "epoch": 0.37914845394204305,
      "grad_norm": 1.1557792825783992,
      "learning_rate": 7.167747612538115e-06,
      "loss": 0.5523,
      "step": 7026
    },
    {
      "epoch": 0.37920241757055745,
      "grad_norm": 0.972454590403336,
      "learning_rate": 7.167039022607546e-06,
      "loss": 0.4995,
      "step": 7027
    },
    {
      "epoch": 0.37925638119907185,
      "grad_norm": 1.0444963863269934,
      "learning_rate": 7.166330384764591e-06,
      "loss": 0.6478,
      "step": 7028
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 1.1356458291397966,
      "learning_rate": 7.165621699029615e-06,
      "loss": 0.5202,
      "step": 7029
    },
    {
      "epoch": 0.3793643084561006,
      "grad_norm": 0.846887285174065,
      "learning_rate": 7.164912965422991e-06,
      "loss": 0.4425,
      "step": 7030
    },
    {
      "epoch": 0.379418272084615,
      "grad_norm": 0.9697369823725206,
      "learning_rate": 7.164204183965085e-06,
      "loss": 0.5323,
      "step": 7031
    },
    {
      "epoch": 0.3794722357131293,
      "grad_norm": 0.8751957801557593,
      "learning_rate": 7.163495354676269e-06,
      "loss": 0.3903,
      "step": 7032
    },
    {
      "epoch": 0.3795261993416437,
      "grad_norm": 0.9173218878604463,
      "learning_rate": 7.162786477576916e-06,
      "loss": 0.4004,
      "step": 7033
    },
    {
      "epoch": 0.3795801629701581,
      "grad_norm": 0.9336291817344491,
      "learning_rate": 7.1620775526874e-06,
      "loss": 0.4734,
      "step": 7034
    },
    {
      "epoch": 0.3796341265986725,
      "grad_norm": 1.2907097609536318,
      "learning_rate": 7.161368580028095e-06,
      "loss": 0.6509,
      "step": 7035
    },
    {
      "epoch": 0.37968809022718686,
      "grad_norm": 1.0339099622969616,
      "learning_rate": 7.160659559619379e-06,
      "loss": 0.5301,
      "step": 7036
    },
    {
      "epoch": 0.37974205385570126,
      "grad_norm": 1.0253728472001105,
      "learning_rate": 7.15995049148163e-06,
      "loss": 0.5609,
      "step": 7037
    },
    {
      "epoch": 0.37979601748421565,
      "grad_norm": 0.9944978363864054,
      "learning_rate": 7.159241375635224e-06,
      "loss": 0.5077,
      "step": 7038
    },
    {
      "epoch": 0.37984998111273,
      "grad_norm": 0.9770593445708848,
      "learning_rate": 7.158532212100548e-06,
      "loss": 0.6439,
      "step": 7039
    },
    {
      "epoch": 0.3799039447412444,
      "grad_norm": 1.1082486835219814,
      "learning_rate": 7.157823000897977e-06,
      "loss": 0.4602,
      "step": 7040
    },
    {
      "epoch": 0.3799579083697588,
      "grad_norm": 0.9180680762962835,
      "learning_rate": 7.157113742047899e-06,
      "loss": 0.3974,
      "step": 7041
    },
    {
      "epoch": 0.3800118719982732,
      "grad_norm": 1.0153429842225434,
      "learning_rate": 7.156404435570698e-06,
      "loss": 0.435,
      "step": 7042
    },
    {
      "epoch": 0.38006583562678753,
      "grad_norm": 1.1245672119684262,
      "learning_rate": 7.1556950814867595e-06,
      "loss": 0.4595,
      "step": 7043
    },
    {
      "epoch": 0.3801197992553019,
      "grad_norm": 1.0118821672640739,
      "learning_rate": 7.1549856798164715e-06,
      "loss": 0.4963,
      "step": 7044
    },
    {
      "epoch": 0.3801737628838163,
      "grad_norm": 1.0733545313084645,
      "learning_rate": 7.1542762305802225e-06,
      "loss": 0.5423,
      "step": 7045
    },
    {
      "epoch": 0.38022772651233067,
      "grad_norm": 0.9808536887973087,
      "learning_rate": 7.153566733798405e-06,
      "loss": 0.4952,
      "step": 7046
    },
    {
      "epoch": 0.38028169014084506,
      "grad_norm": 1.1204979177439232,
      "learning_rate": 7.152857189491406e-06,
      "loss": 0.603,
      "step": 7047
    },
    {
      "epoch": 0.38033565376935946,
      "grad_norm": 0.9924163898705943,
      "learning_rate": 7.152147597679623e-06,
      "loss": 0.5996,
      "step": 7048
    },
    {
      "epoch": 0.38038961739787386,
      "grad_norm": 0.7976745628240199,
      "learning_rate": 7.151437958383448e-06,
      "loss": 0.3166,
      "step": 7049
    },
    {
      "epoch": 0.3804435810263882,
      "grad_norm": 1.2239695684630665,
      "learning_rate": 7.150728271623277e-06,
      "loss": 0.5438,
      "step": 7050
    },
    {
      "epoch": 0.3804975446549026,
      "grad_norm": 0.8369295927919345,
      "learning_rate": 7.150018537419508e-06,
      "loss": 0.3885,
      "step": 7051
    },
    {
      "epoch": 0.380551508283417,
      "grad_norm": 1.0463418150811383,
      "learning_rate": 7.14930875579254e-06,
      "loss": 0.4608,
      "step": 7052
    },
    {
      "epoch": 0.38060547191193134,
      "grad_norm": 1.1286567120340327,
      "learning_rate": 7.148598926762772e-06,
      "loss": 0.594,
      "step": 7053
    },
    {
      "epoch": 0.38065943554044573,
      "grad_norm": 0.982988578476732,
      "learning_rate": 7.1478890503506035e-06,
      "loss": 0.5006,
      "step": 7054
    },
    {
      "epoch": 0.38071339916896013,
      "grad_norm": 0.9470661952728016,
      "learning_rate": 7.147179126576439e-06,
      "loss": 0.4293,
      "step": 7055
    },
    {
      "epoch": 0.38076736279747453,
      "grad_norm": 1.1500669028368993,
      "learning_rate": 7.146469155460683e-06,
      "loss": 0.5725,
      "step": 7056
    },
    {
      "epoch": 0.38082132642598887,
      "grad_norm": 1.0273649273453214,
      "learning_rate": 7.145759137023739e-06,
      "loss": 0.5557,
      "step": 7057
    },
    {
      "epoch": 0.38087529005450327,
      "grad_norm": 0.7927788227930566,
      "learning_rate": 7.145049071286013e-06,
      "loss": 0.3164,
      "step": 7058
    },
    {
      "epoch": 0.38092925368301767,
      "grad_norm": 0.9403495569723332,
      "learning_rate": 7.144338958267917e-06,
      "loss": 0.4916,
      "step": 7059
    },
    {
      "epoch": 0.380983217311532,
      "grad_norm": 0.697577001371987,
      "learning_rate": 7.143628797989858e-06,
      "loss": 0.3165,
      "step": 7060
    },
    {
      "epoch": 0.3810371809400464,
      "grad_norm": 1.039598110387755,
      "learning_rate": 7.142918590472244e-06,
      "loss": 0.4882,
      "step": 7061
    },
    {
      "epoch": 0.3810911445685608,
      "grad_norm": 1.16144188994013,
      "learning_rate": 7.142208335735491e-06,
      "loss": 0.5649,
      "step": 7062
    },
    {
      "epoch": 0.38114510819707514,
      "grad_norm": 0.9066929074648491,
      "learning_rate": 7.141498033800012e-06,
      "loss": 0.5044,
      "step": 7063
    },
    {
      "epoch": 0.38119907182558954,
      "grad_norm": 1.2474159253922128,
      "learning_rate": 7.1407876846862205e-06,
      "loss": 0.6316,
      "step": 7064
    },
    {
      "epoch": 0.38125303545410394,
      "grad_norm": 1.0056805627989949,
      "learning_rate": 7.140077288414532e-06,
      "loss": 0.4375,
      "step": 7065
    },
    {
      "epoch": 0.38130699908261834,
      "grad_norm": 1.0336610517154494,
      "learning_rate": 7.139366845005366e-06,
      "loss": 0.4706,
      "step": 7066
    },
    {
      "epoch": 0.3813609627111327,
      "grad_norm": 0.9399374537163359,
      "learning_rate": 7.13865635447914e-06,
      "loss": 0.4487,
      "step": 7067
    },
    {
      "epoch": 0.3814149263396471,
      "grad_norm": 0.9529239917125278,
      "learning_rate": 7.137945816856276e-06,
      "loss": 0.4648,
      "step": 7068
    },
    {
      "epoch": 0.3814688899681615,
      "grad_norm": 0.9913642253176614,
      "learning_rate": 7.137235232157192e-06,
      "loss": 0.4311,
      "step": 7069
    },
    {
      "epoch": 0.3815228535966758,
      "grad_norm": 1.462036318179698,
      "learning_rate": 7.136524600402315e-06,
      "loss": 0.6083,
      "step": 7070
    },
    {
      "epoch": 0.3815768172251902,
      "grad_norm": 0.9929831928580635,
      "learning_rate": 7.1358139216120655e-06,
      "loss": 0.4829,
      "step": 7071
    },
    {
      "epoch": 0.3816307808537046,
      "grad_norm": 1.060270646546809,
      "learning_rate": 7.135103195806872e-06,
      "loss": 0.4438,
      "step": 7072
    },
    {
      "epoch": 0.381684744482219,
      "grad_norm": 1.1636922748726746,
      "learning_rate": 7.134392423007162e-06,
      "loss": 0.4459,
      "step": 7073
    },
    {
      "epoch": 0.38173870811073335,
      "grad_norm": 1.0072669476342324,
      "learning_rate": 7.13368160323336e-06,
      "loss": 0.4551,
      "step": 7074
    },
    {
      "epoch": 0.38179267173924775,
      "grad_norm": 1.0058699790543133,
      "learning_rate": 7.1329707365059e-06,
      "loss": 0.5872,
      "step": 7075
    },
    {
      "epoch": 0.38184663536776214,
      "grad_norm": 1.2077065808833078,
      "learning_rate": 7.132259822845209e-06,
      "loss": 0.4921,
      "step": 7076
    },
    {
      "epoch": 0.3819005989962765,
      "grad_norm": 1.1241221299731419,
      "learning_rate": 7.131548862271722e-06,
      "loss": 0.5221,
      "step": 7077
    },
    {
      "epoch": 0.3819545626247909,
      "grad_norm": 1.2889684412514781,
      "learning_rate": 7.130837854805873e-06,
      "loss": 0.6595,
      "step": 7078
    },
    {
      "epoch": 0.3820085262533053,
      "grad_norm": 0.6497624686592273,
      "learning_rate": 7.1301268004680945e-06,
      "loss": 0.3274,
      "step": 7079
    },
    {
      "epoch": 0.3820624898818197,
      "grad_norm": 0.9274043740305404,
      "learning_rate": 7.1294156992788265e-06,
      "loss": 0.434,
      "step": 7080
    },
    {
      "epoch": 0.382116453510334,
      "grad_norm": 0.9295793633690538,
      "learning_rate": 7.128704551258505e-06,
      "loss": 0.3892,
      "step": 7081
    },
    {
      "epoch": 0.3821704171388484,
      "grad_norm": 1.1044574355017553,
      "learning_rate": 7.127993356427568e-06,
      "loss": 0.6711,
      "step": 7082
    },
    {
      "epoch": 0.3822243807673628,
      "grad_norm": 1.103033647412057,
      "learning_rate": 7.127282114806456e-06,
      "loss": 0.5789,
      "step": 7083
    },
    {
      "epoch": 0.38227834439587716,
      "grad_norm": 0.8156842862976554,
      "learning_rate": 7.126570826415614e-06,
      "loss": 0.2496,
      "step": 7084
    },
    {
      "epoch": 0.38233230802439155,
      "grad_norm": 1.2599284771480748,
      "learning_rate": 7.125859491275482e-06,
      "loss": 0.569,
      "step": 7085
    },
    {
      "epoch": 0.38238627165290595,
      "grad_norm": 0.919740635654696,
      "learning_rate": 7.125148109406504e-06,
      "loss": 0.494,
      "step": 7086
    },
    {
      "epoch": 0.38244023528142035,
      "grad_norm": 0.9637185972817239,
      "learning_rate": 7.124436680829128e-06,
      "loss": 0.3611,
      "step": 7087
    },
    {
      "epoch": 0.3824941989099347,
      "grad_norm": 0.9849685295863924,
      "learning_rate": 7.123725205563803e-06,
      "loss": 0.4652,
      "step": 7088
    },
    {
      "epoch": 0.3825481625384491,
      "grad_norm": 1.0029198575451135,
      "learning_rate": 7.123013683630972e-06,
      "loss": 0.4336,
      "step": 7089
    },
    {
      "epoch": 0.3826021261669635,
      "grad_norm": 1.1075525525727516,
      "learning_rate": 7.1223021150510885e-06,
      "loss": 0.4762,
      "step": 7090
    },
    {
      "epoch": 0.3826560897954778,
      "grad_norm": 0.7420565547532824,
      "learning_rate": 7.121590499844604e-06,
      "loss": 0.3496,
      "step": 7091
    },
    {
      "epoch": 0.3827100534239922,
      "grad_norm": 1.060954325935764,
      "learning_rate": 7.120878838031969e-06,
      "loss": 0.3414,
      "step": 7092
    },
    {
      "epoch": 0.3827640170525066,
      "grad_norm": 0.9773295965412607,
      "learning_rate": 7.120167129633638e-06,
      "loss": 0.423,
      "step": 7093
    },
    {
      "epoch": 0.382817980681021,
      "grad_norm": 0.9642921070438216,
      "learning_rate": 7.119455374670068e-06,
      "loss": 0.3988,
      "step": 7094
    },
    {
      "epoch": 0.38287194430953536,
      "grad_norm": 1.1385182171735764,
      "learning_rate": 7.1187435731617135e-06,
      "loss": 0.5359,
      "step": 7095
    },
    {
      "epoch": 0.38292590793804976,
      "grad_norm": 0.9663787184249133,
      "learning_rate": 7.118031725129033e-06,
      "loss": 0.5206,
      "step": 7096
    },
    {
      "epoch": 0.38297987156656416,
      "grad_norm": 1.062153457706783,
      "learning_rate": 7.117319830592485e-06,
      "loss": 0.4935,
      "step": 7097
    },
    {
      "epoch": 0.3830338351950785,
      "grad_norm": 1.2165884168473715,
      "learning_rate": 7.116607889572533e-06,
      "loss": 0.716,
      "step": 7098
    },
    {
      "epoch": 0.3830877988235929,
      "grad_norm": 0.8724377963198745,
      "learning_rate": 7.115895902089635e-06,
      "loss": 0.3099,
      "step": 7099
    },
    {
      "epoch": 0.3831417624521073,
      "grad_norm": 1.1120702537414002,
      "learning_rate": 7.115183868164257e-06,
      "loss": 0.5846,
      "step": 7100
    },
    {
      "epoch": 0.38319572608062163,
      "grad_norm": 1.0464143765369258,
      "learning_rate": 7.114471787816862e-06,
      "loss": 0.5569,
      "step": 7101
    },
    {
      "epoch": 0.38324968970913603,
      "grad_norm": 0.9213089670858199,
      "learning_rate": 7.113759661067917e-06,
      "loss": 0.3718,
      "step": 7102
    },
    {
      "epoch": 0.38330365333765043,
      "grad_norm": 1.2132145901734046,
      "learning_rate": 7.1130474879378886e-06,
      "loss": 0.6988,
      "step": 7103
    },
    {
      "epoch": 0.3833576169661648,
      "grad_norm": 1.034736106727486,
      "learning_rate": 7.1123352684472445e-06,
      "loss": 0.4799,
      "step": 7104
    },
    {
      "epoch": 0.38341158059467917,
      "grad_norm": 1.063297091877675,
      "learning_rate": 7.111623002616459e-06,
      "loss": 0.6107,
      "step": 7105
    },
    {
      "epoch": 0.38346554422319357,
      "grad_norm": 1.05256114320061,
      "learning_rate": 7.1109106904659964e-06,
      "loss": 0.4745,
      "step": 7106
    },
    {
      "epoch": 0.38351950785170796,
      "grad_norm": 0.9915479572382622,
      "learning_rate": 7.110198332016334e-06,
      "loss": 0.4537,
      "step": 7107
    },
    {
      "epoch": 0.3835734714802223,
      "grad_norm": 1.1606877691514548,
      "learning_rate": 7.1094859272879435e-06,
      "loss": 0.5435,
      "step": 7108
    },
    {
      "epoch": 0.3836274351087367,
      "grad_norm": 1.0426600000798874,
      "learning_rate": 7.108773476301302e-06,
      "loss": 0.4918,
      "step": 7109
    },
    {
      "epoch": 0.3836813987372511,
      "grad_norm": 0.9562706075557893,
      "learning_rate": 7.108060979076885e-06,
      "loss": 0.4266,
      "step": 7110
    },
    {
      "epoch": 0.3837353623657655,
      "grad_norm": 1.1648405529061945,
      "learning_rate": 7.1073484356351715e-06,
      "loss": 0.5566,
      "step": 7111
    },
    {
      "epoch": 0.38378932599427984,
      "grad_norm": 0.9192670509764957,
      "learning_rate": 7.10663584599664e-06,
      "loss": 0.4417,
      "step": 7112
    },
    {
      "epoch": 0.38384328962279424,
      "grad_norm": 0.9929416534731506,
      "learning_rate": 7.10592321018177e-06,
      "loss": 0.4811,
      "step": 7113
    },
    {
      "epoch": 0.38389725325130863,
      "grad_norm": 1.0627248494896002,
      "learning_rate": 7.105210528211044e-06,
      "loss": 0.5636,
      "step": 7114
    },
    {
      "epoch": 0.383951216879823,
      "grad_norm": 1.0732106681059543,
      "learning_rate": 7.104497800104947e-06,
      "loss": 0.4384,
      "step": 7115
    },
    {
      "epoch": 0.3840051805083374,
      "grad_norm": 0.8965534118919051,
      "learning_rate": 7.10378502588396e-06,
      "loss": 0.3766,
      "step": 7116
    },
    {
      "epoch": 0.38405914413685177,
      "grad_norm": 1.2682003521416387,
      "learning_rate": 7.103072205568573e-06,
      "loss": 0.6621,
      "step": 7117
    },
    {
      "epoch": 0.38411310776536617,
      "grad_norm": 0.9471164574207247,
      "learning_rate": 7.1023593391792686e-06,
      "loss": 0.4164,
      "step": 7118
    },
    {
      "epoch": 0.3841670713938805,
      "grad_norm": 1.0882104084512663,
      "learning_rate": 7.1016464267365404e-06,
      "loss": 0.6048,
      "step": 7119
    },
    {
      "epoch": 0.3842210350223949,
      "grad_norm": 0.7458542911698984,
      "learning_rate": 7.100933468260874e-06,
      "loss": 0.3401,
      "step": 7120
    },
    {
      "epoch": 0.3842749986509093,
      "grad_norm": 0.7901450629168862,
      "learning_rate": 7.100220463772762e-06,
      "loss": 0.4414,
      "step": 7121
    },
    {
      "epoch": 0.38432896227942365,
      "grad_norm": 1.023105860278067,
      "learning_rate": 7.099507413292697e-06,
      "loss": 0.4411,
      "step": 7122
    },
    {
      "epoch": 0.38438292590793804,
      "grad_norm": 0.9229392990431791,
      "learning_rate": 7.098794316841175e-06,
      "loss": 0.4575,
      "step": 7123
    },
    {
      "epoch": 0.38443688953645244,
      "grad_norm": 0.8662001339135197,
      "learning_rate": 7.0980811744386855e-06,
      "loss": 0.4488,
      "step": 7124
    },
    {
      "epoch": 0.38449085316496684,
      "grad_norm": 0.9372548480448775,
      "learning_rate": 7.09736798610573e-06,
      "loss": 0.4763,
      "step": 7125
    },
    {
      "epoch": 0.3845448167934812,
      "grad_norm": 0.9088288956530416,
      "learning_rate": 7.096654751862805e-06,
      "loss": 0.4207,
      "step": 7126
    },
    {
      "epoch": 0.3845987804219956,
      "grad_norm": 1.0762002911766866,
      "learning_rate": 7.095941471730408e-06,
      "loss": 0.5215,
      "step": 7127
    },
    {
      "epoch": 0.38465274405051,
      "grad_norm": 1.2464436997477302,
      "learning_rate": 7.095228145729041e-06,
      "loss": 0.5932,
      "step": 7128
    },
    {
      "epoch": 0.3847067076790243,
      "grad_norm": 0.8331466414265337,
      "learning_rate": 7.0945147738792055e-06,
      "loss": 0.3068,
      "step": 7129
    },
    {
      "epoch": 0.3847606713075387,
      "grad_norm": 1.3065869811930102,
      "learning_rate": 7.093801356201403e-06,
      "loss": 0.6571,
      "step": 7130
    },
    {
      "epoch": 0.3848146349360531,
      "grad_norm": 1.226712268185015,
      "learning_rate": 7.0930878927161405e-06,
      "loss": 0.4885,
      "step": 7131
    },
    {
      "epoch": 0.38486859856456745,
      "grad_norm": 1.2026442888993771,
      "learning_rate": 7.0923743834439215e-06,
      "loss": 0.6074,
      "step": 7132
    },
    {
      "epoch": 0.38492256219308185,
      "grad_norm": 1.0776890496189788,
      "learning_rate": 7.0916608284052544e-06,
      "loss": 0.561,
      "step": 7133
    },
    {
      "epoch": 0.38497652582159625,
      "grad_norm": 0.9843011732735338,
      "learning_rate": 7.090947227620646e-06,
      "loss": 0.3488,
      "step": 7134
    },
    {
      "epoch": 0.38503048945011065,
      "grad_norm": 1.186524276821606,
      "learning_rate": 7.090233581110608e-06,
      "loss": 0.5732,
      "step": 7135
    },
    {
      "epoch": 0.385084453078625,
      "grad_norm": 1.0979776879028649,
      "learning_rate": 7.089519888895648e-06,
      "loss": 0.453,
      "step": 7136
    },
    {
      "epoch": 0.3851384167071394,
      "grad_norm": 0.9995108687191485,
      "learning_rate": 7.088806150996281e-06,
      "loss": 0.5055,
      "step": 7137
    },
    {
      "epoch": 0.3851923803356538,
      "grad_norm": 0.9120729093562829,
      "learning_rate": 7.088092367433022e-06,
      "loss": 0.3815,
      "step": 7138
    },
    {
      "epoch": 0.3852463439641681,
      "grad_norm": 1.0158641614489672,
      "learning_rate": 7.087378538226381e-06,
      "loss": 0.4488,
      "step": 7139
    },
    {
      "epoch": 0.3853003075926825,
      "grad_norm": 1.0344503246881993,
      "learning_rate": 7.0866646633968795e-06,
      "loss": 0.6048,
      "step": 7140
    },
    {
      "epoch": 0.3853542712211969,
      "grad_norm": 1.1971698488834972,
      "learning_rate": 7.0859507429650296e-06,
      "loss": 0.5987,
      "step": 7141
    },
    {
      "epoch": 0.3854082348497113,
      "grad_norm": 0.9401294787234151,
      "learning_rate": 7.085236776951354e-06,
      "loss": 0.394,
      "step": 7142
    },
    {
      "epoch": 0.38546219847822566,
      "grad_norm": 0.8449491398687429,
      "learning_rate": 7.084522765376374e-06,
      "loss": 0.4778,
      "step": 7143
    },
    {
      "epoch": 0.38551616210674006,
      "grad_norm": 0.953354842794333,
      "learning_rate": 7.0838087082606054e-06,
      "loss": 0.5349,
      "step": 7144
    },
    {
      "epoch": 0.38557012573525445,
      "grad_norm": 1.0468489862211805,
      "learning_rate": 7.083094605624574e-06,
      "loss": 0.5007,
      "step": 7145
    },
    {
      "epoch": 0.3856240893637688,
      "grad_norm": 0.9471131130888596,
      "learning_rate": 7.0823804574888065e-06,
      "loss": 0.501,
      "step": 7146
    },
    {
      "epoch": 0.3856780529922832,
      "grad_norm": 0.9130830609528583,
      "learning_rate": 7.081666263873824e-06,
      "loss": 0.3816,
      "step": 7147
    },
    {
      "epoch": 0.3857320166207976,
      "grad_norm": 0.9303626655705144,
      "learning_rate": 7.0809520248001565e-06,
      "loss": 0.4286,
      "step": 7148
    },
    {
      "epoch": 0.385785980249312,
      "grad_norm": 1.034402357264734,
      "learning_rate": 7.080237740288328e-06,
      "loss": 0.4928,
      "step": 7149
    },
    {
      "epoch": 0.38583994387782633,
      "grad_norm": 0.9764894319236294,
      "learning_rate": 7.079523410358873e-06,
      "loss": 0.4552,
      "step": 7150
    },
    {
      "epoch": 0.3858939075063407,
      "grad_norm": 1.0555463097969435,
      "learning_rate": 7.078809035032317e-06,
      "loss": 0.5333,
      "step": 7151
    },
    {
      "epoch": 0.3859478711348551,
      "grad_norm": 1.098031922007203,
      "learning_rate": 7.078094614329195e-06,
      "loss": 0.5165,
      "step": 7152
    },
    {
      "epoch": 0.38600183476336947,
      "grad_norm": 0.9919541710985963,
      "learning_rate": 7.07738014827004e-06,
      "loss": 0.4473,
      "step": 7153
    },
    {
      "epoch": 0.38605579839188386,
      "grad_norm": 1.1764797714051183,
      "learning_rate": 7.076665636875384e-06,
      "loss": 0.5697,
      "step": 7154
    },
    {
      "epoch": 0.38610976202039826,
      "grad_norm": 1.023379832879976,
      "learning_rate": 7.075951080165765e-06,
      "loss": 0.5068,
      "step": 7155
    },
    {
      "epoch": 0.38616372564891266,
      "grad_norm": 0.7060380034492929,
      "learning_rate": 7.07523647816172e-06,
      "loss": 0.347,
      "step": 7156
    },
    {
      "epoch": 0.386217689277427,
      "grad_norm": 0.9274208367146264,
      "learning_rate": 7.074521830883787e-06,
      "loss": 0.4828,
      "step": 7157
    },
    {
      "epoch": 0.3862716529059414,
      "grad_norm": 0.8945485547409492,
      "learning_rate": 7.073807138352506e-06,
      "loss": 0.3628,
      "step": 7158
    },
    {
      "epoch": 0.3863256165344558,
      "grad_norm": 0.9861317239492713,
      "learning_rate": 7.0730924005884165e-06,
      "loss": 0.5182,
      "step": 7159
    },
    {
      "epoch": 0.38637958016297014,
      "grad_norm": 0.9266853032703605,
      "learning_rate": 7.072377617612062e-06,
      "loss": 0.4944,
      "step": 7160
    },
    {
      "epoch": 0.38643354379148454,
      "grad_norm": 0.7917828122443779,
      "learning_rate": 7.071662789443986e-06,
      "loss": 0.3137,
      "step": 7161
    },
    {
      "epoch": 0.38648750741999893,
      "grad_norm": 1.1751427811865813,
      "learning_rate": 7.070947916104735e-06,
      "loss": 0.4857,
      "step": 7162
    },
    {
      "epoch": 0.3865414710485133,
      "grad_norm": 1.277641393297814,
      "learning_rate": 7.070232997614852e-06,
      "loss": 0.7244,
      "step": 7163
    },
    {
      "epoch": 0.38659543467702767,
      "grad_norm": 1.0282769903913904,
      "learning_rate": 7.069518033994886e-06,
      "loss": 0.4645,
      "step": 7164
    },
    {
      "epoch": 0.38664939830554207,
      "grad_norm": 1.077618043518193,
      "learning_rate": 7.068803025265386e-06,
      "loss": 0.4832,
      "step": 7165
    },
    {
      "epoch": 0.38670336193405647,
      "grad_norm": 0.930397629419174,
      "learning_rate": 7.068087971446902e-06,
      "loss": 0.5129,
      "step": 7166
    },
    {
      "epoch": 0.3867573255625708,
      "grad_norm": 1.1015822832660194,
      "learning_rate": 7.0673728725599855e-06,
      "loss": 0.5738,
      "step": 7167
    },
    {
      "epoch": 0.3868112891910852,
      "grad_norm": 1.0767470700972241,
      "learning_rate": 7.0666577286251895e-06,
      "loss": 0.3527,
      "step": 7168
    },
    {
      "epoch": 0.3868652528195996,
      "grad_norm": 1.1710833788466661,
      "learning_rate": 7.0659425396630665e-06,
      "loss": 0.4915,
      "step": 7169
    },
    {
      "epoch": 0.38691921644811395,
      "grad_norm": 1.255486128019382,
      "learning_rate": 7.065227305694173e-06,
      "loss": 0.6422,
      "step": 7170
    },
    {
      "epoch": 0.38697318007662834,
      "grad_norm": 0.941733134473826,
      "learning_rate": 7.064512026739064e-06,
      "loss": 0.459,
      "step": 7171
    },
    {
      "epoch": 0.38702714370514274,
      "grad_norm": 1.0107485023545026,
      "learning_rate": 7.0637967028183e-06,
      "loss": 0.3339,
      "step": 7172
    },
    {
      "epoch": 0.38708110733365714,
      "grad_norm": 1.034652975828915,
      "learning_rate": 7.063081333952439e-06,
      "loss": 0.4795,
      "step": 7173
    },
    {
      "epoch": 0.3871350709621715,
      "grad_norm": 0.9516667697952352,
      "learning_rate": 7.062365920162041e-06,
      "loss": 0.4958,
      "step": 7174
    },
    {
      "epoch": 0.3871890345906859,
      "grad_norm": 1.057235885495645,
      "learning_rate": 7.061650461467666e-06,
      "loss": 0.4966,
      "step": 7175
    },
    {
      "epoch": 0.3872429982192003,
      "grad_norm": 1.132586087963951,
      "learning_rate": 7.06093495788988e-06,
      "loss": 0.4475,
      "step": 7176
    },
    {
      "epoch": 0.3872969618477146,
      "grad_norm": 1.0395971098901065,
      "learning_rate": 7.060219409449247e-06,
      "loss": 0.5493,
      "step": 7177
    },
    {
      "epoch": 0.387350925476229,
      "grad_norm": 1.0317731792643587,
      "learning_rate": 7.0595038161663295e-06,
      "loss": 0.4934,
      "step": 7178
    },
    {
      "epoch": 0.3874048891047434,
      "grad_norm": 0.9666342972898421,
      "learning_rate": 7.058788178061698e-06,
      "loss": 0.4802,
      "step": 7179
    },
    {
      "epoch": 0.3874588527332578,
      "grad_norm": 0.9085884539362401,
      "learning_rate": 7.0580724951559185e-06,
      "loss": 0.4202,
      "step": 7180
    },
    {
      "epoch": 0.38751281636177215,
      "grad_norm": 0.8730183313954777,
      "learning_rate": 7.057356767469563e-06,
      "loss": 0.4206,
      "step": 7181
    },
    {
      "epoch": 0.38756677999028655,
      "grad_norm": 0.9509450669788764,
      "learning_rate": 7.056640995023199e-06,
      "loss": 0.6737,
      "step": 7182
    },
    {
      "epoch": 0.38762074361880094,
      "grad_norm": 0.8717379001901906,
      "learning_rate": 7.0559251778373995e-06,
      "loss": 0.3701,
      "step": 7183
    },
    {
      "epoch": 0.3876747072473153,
      "grad_norm": 1.2082073343371884,
      "learning_rate": 7.055209315932738e-06,
      "loss": 0.5439,
      "step": 7184
    },
    {
      "epoch": 0.3877286708758297,
      "grad_norm": 0.785822838791441,
      "learning_rate": 7.05449340932979e-06,
      "loss": 0.3603,
      "step": 7185
    },
    {
      "epoch": 0.3877826345043441,
      "grad_norm": 0.9010915637090584,
      "learning_rate": 7.0537774580491325e-06,
      "loss": 0.3701,
      "step": 7186
    },
    {
      "epoch": 0.3878365981328585,
      "grad_norm": 1.15367957180958,
      "learning_rate": 7.053061462111338e-06,
      "loss": 0.4921,
      "step": 7187
    },
    {
      "epoch": 0.3878905617613728,
      "grad_norm": 0.9970428394282319,
      "learning_rate": 7.052345421536989e-06,
      "loss": 0.4775,
      "step": 7188
    },
    {
      "epoch": 0.3879445253898872,
      "grad_norm": 0.8959197416137857,
      "learning_rate": 7.0516293363466636e-06,
      "loss": 0.4573,
      "step": 7189
    },
    {
      "epoch": 0.3879984890184016,
      "grad_norm": 1.1117595046405957,
      "learning_rate": 7.0509132065609446e-06,
      "loss": 0.5436,
      "step": 7190
    },
    {
      "epoch": 0.38805245264691596,
      "grad_norm": 0.8900668561743117,
      "learning_rate": 7.050197032200412e-06,
      "loss": 0.3922,
      "step": 7191
    },
    {
      "epoch": 0.38810641627543035,
      "grad_norm": 1.1924314925482522,
      "learning_rate": 7.049480813285651e-06,
      "loss": 0.5425,
      "step": 7192
    },
    {
      "epoch": 0.38816037990394475,
      "grad_norm": 1.0180136829689388,
      "learning_rate": 7.048764549837244e-06,
      "loss": 0.4372,
      "step": 7193
    },
    {
      "epoch": 0.38821434353245915,
      "grad_norm": 0.9585067729255137,
      "learning_rate": 7.0480482418757805e-06,
      "loss": 0.4317,
      "step": 7194
    },
    {
      "epoch": 0.3882683071609735,
      "grad_norm": 1.0963372241770633,
      "learning_rate": 7.047331889421846e-06,
      "loss": 0.4755,
      "step": 7195
    },
    {
      "epoch": 0.3883222707894879,
      "grad_norm": 1.0365354550424546,
      "learning_rate": 7.0466154924960296e-06,
      "loss": 0.426,
      "step": 7196
    },
    {
      "epoch": 0.3883762344180023,
      "grad_norm": 0.9419839395520084,
      "learning_rate": 7.045899051118921e-06,
      "loss": 0.4087,
      "step": 7197
    },
    {
      "epoch": 0.38843019804651663,
      "grad_norm": 1.0303980386663267,
      "learning_rate": 7.045182565311112e-06,
      "loss": 0.4673,
      "step": 7198
    },
    {
      "epoch": 0.388484161675031,
      "grad_norm": 1.0721416271194737,
      "learning_rate": 7.044466035093194e-06,
      "loss": 0.5022,
      "step": 7199
    },
    {
      "epoch": 0.3885381253035454,
      "grad_norm": 1.211098016480153,
      "learning_rate": 7.043749460485763e-06,
      "loss": 0.5835,
      "step": 7200
    },
    {
      "epoch": 0.38859208893205976,
      "grad_norm": 0.9670605427881132,
      "learning_rate": 7.043032841509412e-06,
      "loss": 0.4229,
      "step": 7201
    },
    {
      "epoch": 0.38864605256057416,
      "grad_norm": 0.7833572812396475,
      "learning_rate": 7.042316178184737e-06,
      "loss": 0.3889,
      "step": 7202
    },
    {
      "epoch": 0.38870001618908856,
      "grad_norm": 0.9266040312720747,
      "learning_rate": 7.041599470532339e-06,
      "loss": 0.41,
      "step": 7203
    },
    {
      "epoch": 0.38875397981760296,
      "grad_norm": 1.1931329029825855,
      "learning_rate": 7.040882718572811e-06,
      "loss": 0.4791,
      "step": 7204
    },
    {
      "epoch": 0.3888079434461173,
      "grad_norm": 0.8907613060056407,
      "learning_rate": 7.04016592232676e-06,
      "loss": 0.381,
      "step": 7205
    },
    {
      "epoch": 0.3888619070746317,
      "grad_norm": 0.8936559938593469,
      "learning_rate": 7.039449081814782e-06,
      "loss": 0.4726,
      "step": 7206
    },
    {
      "epoch": 0.3889158707031461,
      "grad_norm": 0.9096899482588223,
      "learning_rate": 7.038732197057483e-06,
      "loss": 0.5035,
      "step": 7207
    },
    {
      "epoch": 0.38896983433166044,
      "grad_norm": 0.8564365435946649,
      "learning_rate": 7.0380152680754665e-06,
      "loss": 0.293,
      "step": 7208
    },
    {
      "epoch": 0.38902379796017483,
      "grad_norm": 1.030939401818031,
      "learning_rate": 7.037298294889337e-06,
      "loss": 0.5405,
      "step": 7209
    },
    {
      "epoch": 0.38907776158868923,
      "grad_norm": 0.8156352019225015,
      "learning_rate": 7.0365812775197e-06,
      "loss": 0.3407,
      "step": 7210
    },
    {
      "epoch": 0.38913172521720363,
      "grad_norm": 0.8959917160228557,
      "learning_rate": 7.035864215987164e-06,
      "loss": 0.416,
      "step": 7211
    },
    {
      "epoch": 0.38918568884571797,
      "grad_norm": 0.8564084387389894,
      "learning_rate": 7.035147110312338e-06,
      "loss": 0.4639,
      "step": 7212
    },
    {
      "epoch": 0.38923965247423237,
      "grad_norm": 0.8421483994153907,
      "learning_rate": 7.034429960515835e-06,
      "loss": 0.3208,
      "step": 7213
    },
    {
      "epoch": 0.38929361610274676,
      "grad_norm": 1.0619468884089582,
      "learning_rate": 7.033712766618264e-06,
      "loss": 0.3954,
      "step": 7214
    },
    {
      "epoch": 0.3893475797312611,
      "grad_norm": 1.1193664515127848,
      "learning_rate": 7.032995528640239e-06,
      "loss": 0.4765,
      "step": 7215
    },
    {
      "epoch": 0.3894015433597755,
      "grad_norm": 0.8177462030389688,
      "learning_rate": 7.032278246602373e-06,
      "loss": 0.3515,
      "step": 7216
    },
    {
      "epoch": 0.3894555069882899,
      "grad_norm": 0.9966554277408534,
      "learning_rate": 7.031560920525281e-06,
      "loss": 0.4903,
      "step": 7217
    },
    {
      "epoch": 0.3895094706168043,
      "grad_norm": 1.0337519609089925,
      "learning_rate": 7.0308435504295815e-06,
      "loss": 0.4779,
      "step": 7218
    },
    {
      "epoch": 0.38956343424531864,
      "grad_norm": 0.9800705214860455,
      "learning_rate": 7.0301261363358915e-06,
      "loss": 0.3946,
      "step": 7219
    },
    {
      "epoch": 0.38961739787383304,
      "grad_norm": 0.8525677807156149,
      "learning_rate": 7.029408678264831e-06,
      "loss": 0.3557,
      "step": 7220
    },
    {
      "epoch": 0.38967136150234744,
      "grad_norm": 0.8058616122151263,
      "learning_rate": 7.028691176237018e-06,
      "loss": 0.349,
      "step": 7221
    },
    {
      "epoch": 0.3897253251308618,
      "grad_norm": 1.1761745000742074,
      "learning_rate": 7.027973630273077e-06,
      "loss": 0.4281,
      "step": 7222
    },
    {
      "epoch": 0.3897792887593762,
      "grad_norm": 0.8373419086931093,
      "learning_rate": 7.02725604039363e-06,
      "loss": 0.4475,
      "step": 7223
    },
    {
      "epoch": 0.38983325238789057,
      "grad_norm": 1.0985215493215759,
      "learning_rate": 7.026538406619301e-06,
      "loss": 0.6293,
      "step": 7224
    },
    {
      "epoch": 0.38988721601640497,
      "grad_norm": 1.0229374001690932,
      "learning_rate": 7.0258207289707145e-06,
      "loss": 0.5391,
      "step": 7225
    },
    {
      "epoch": 0.3899411796449193,
      "grad_norm": 0.8769962120893854,
      "learning_rate": 7.025103007468502e-06,
      "loss": 0.4002,
      "step": 7226
    },
    {
      "epoch": 0.3899951432734337,
      "grad_norm": 1.0958594134710034,
      "learning_rate": 7.024385242133286e-06,
      "loss": 0.435,
      "step": 7227
    },
    {
      "epoch": 0.3900491069019481,
      "grad_norm": 0.8972533948034126,
      "learning_rate": 7.023667432985697e-06,
      "loss": 0.4173,
      "step": 7228
    },
    {
      "epoch": 0.39010307053046245,
      "grad_norm": 1.0789616426778064,
      "learning_rate": 7.022949580046367e-06,
      "loss": 0.6232,
      "step": 7229
    },
    {
      "epoch": 0.39015703415897685,
      "grad_norm": 0.9537355588163308,
      "learning_rate": 7.0222316833359275e-06,
      "loss": 0.53,
      "step": 7230
    },
    {
      "epoch": 0.39021099778749124,
      "grad_norm": 0.9543028721252254,
      "learning_rate": 7.021513742875011e-06,
      "loss": 0.4977,
      "step": 7231
    },
    {
      "epoch": 0.3902649614160056,
      "grad_norm": 0.7659776709532105,
      "learning_rate": 7.020795758684253e-06,
      "loss": 0.3067,
      "step": 7232
    },
    {
      "epoch": 0.39031892504452,
      "grad_norm": 1.0748354462136311,
      "learning_rate": 7.0200777307842875e-06,
      "loss": 0.5044,
      "step": 7233
    },
    {
      "epoch": 0.3903728886730344,
      "grad_norm": 0.9853379338503038,
      "learning_rate": 7.019359659195752e-06,
      "loss": 0.4546,
      "step": 7234
    },
    {
      "epoch": 0.3904268523015488,
      "grad_norm": 0.7333583136660703,
      "learning_rate": 7.0186415439392855e-06,
      "loss": 0.3826,
      "step": 7235
    },
    {
      "epoch": 0.3904808159300631,
      "grad_norm": 1.0292533974858298,
      "learning_rate": 7.017923385035525e-06,
      "loss": 0.473,
      "step": 7236
    },
    {
      "epoch": 0.3905347795585775,
      "grad_norm": 1.1348003935542461,
      "learning_rate": 7.0172051825051136e-06,
      "loss": 0.4709,
      "step": 7237
    },
    {
      "epoch": 0.3905887431870919,
      "grad_norm": 1.15543624233146,
      "learning_rate": 7.016486936368693e-06,
      "loss": 0.5017,
      "step": 7238
    },
    {
      "epoch": 0.39064270681560626,
      "grad_norm": 1.1295906102138595,
      "learning_rate": 7.015768646646905e-06,
      "loss": 0.5352,
      "step": 7239
    },
    {
      "epoch": 0.39069667044412065,
      "grad_norm": 1.1053646692394374,
      "learning_rate": 7.0150503133603935e-06,
      "loss": 0.4332,
      "step": 7240
    },
    {
      "epoch": 0.39075063407263505,
      "grad_norm": 0.9590893703691303,
      "learning_rate": 7.014331936529806e-06,
      "loss": 0.5376,
      "step": 7241
    },
    {
      "epoch": 0.39080459770114945,
      "grad_norm": 1.0086592089816349,
      "learning_rate": 7.013613516175788e-06,
      "loss": 0.4296,
      "step": 7242
    },
    {
      "epoch": 0.3908585613296638,
      "grad_norm": 1.03641713735584,
      "learning_rate": 7.01289505231899e-06,
      "loss": 0.575,
      "step": 7243
    },
    {
      "epoch": 0.3909125249581782,
      "grad_norm": 1.0200168185723724,
      "learning_rate": 7.012176544980057e-06,
      "loss": 0.4415,
      "step": 7244
    },
    {
      "epoch": 0.3909664885866926,
      "grad_norm": 0.9762736029474288,
      "learning_rate": 7.011457994179643e-06,
      "loss": 0.4961,
      "step": 7245
    },
    {
      "epoch": 0.3910204522152069,
      "grad_norm": 1.0959573340426039,
      "learning_rate": 7.010739399938401e-06,
      "loss": 0.5827,
      "step": 7246
    },
    {
      "epoch": 0.3910744158437213,
      "grad_norm": 1.0635511777680617,
      "learning_rate": 7.01002076227698e-06,
      "loss": 0.5779,
      "step": 7247
    },
    {
      "epoch": 0.3911283794722357,
      "grad_norm": 1.128534522223739,
      "learning_rate": 7.009302081216036e-06,
      "loss": 0.4173,
      "step": 7248
    },
    {
      "epoch": 0.3911823431007501,
      "grad_norm": 0.8696701511681613,
      "learning_rate": 7.008583356776226e-06,
      "loss": 0.3832,
      "step": 7249
    },
    {
      "epoch": 0.39123630672926446,
      "grad_norm": 0.8690374324591784,
      "learning_rate": 7.007864588978208e-06,
      "loss": 0.3103,
      "step": 7250
    },
    {
      "epoch": 0.39129027035777886,
      "grad_norm": 1.042111707988464,
      "learning_rate": 7.0071457778426366e-06,
      "loss": 0.4952,
      "step": 7251
    },
    {
      "epoch": 0.39134423398629326,
      "grad_norm": 0.9734874268073874,
      "learning_rate": 7.006426923390171e-06,
      "loss": 0.5903,
      "step": 7252
    },
    {
      "epoch": 0.3913981976148076,
      "grad_norm": 1.0030761967790662,
      "learning_rate": 7.005708025641475e-06,
      "loss": 0.4342,
      "step": 7253
    },
    {
      "epoch": 0.391452161243322,
      "grad_norm": 1.1670486393762667,
      "learning_rate": 7.00498908461721e-06,
      "loss": 0.4916,
      "step": 7254
    },
    {
      "epoch": 0.3915061248718364,
      "grad_norm": 1.2599757416224033,
      "learning_rate": 7.004270100338037e-06,
      "loss": 0.5497,
      "step": 7255
    },
    {
      "epoch": 0.3915600885003508,
      "grad_norm": 1.19803033672981,
      "learning_rate": 7.00355107282462e-06,
      "loss": 0.5243,
      "step": 7256
    },
    {
      "epoch": 0.39161405212886513,
      "grad_norm": 0.9254486315159782,
      "learning_rate": 7.002832002097628e-06,
      "loss": 0.4854,
      "step": 7257
    },
    {
      "epoch": 0.39166801575737953,
      "grad_norm": 1.1035220708748605,
      "learning_rate": 7.002112888177725e-06,
      "loss": 0.7131,
      "step": 7258
    },
    {
      "epoch": 0.3917219793858939,
      "grad_norm": 1.2591655873019791,
      "learning_rate": 7.001393731085578e-06,
      "loss": 0.6249,
      "step": 7259
    },
    {
      "epoch": 0.39177594301440827,
      "grad_norm": 0.9636921861142513,
      "learning_rate": 7.000674530841861e-06,
      "loss": 0.3811,
      "step": 7260
    },
    {
      "epoch": 0.39182990664292267,
      "grad_norm": 1.101326954642365,
      "learning_rate": 6.99995528746724e-06,
      "loss": 0.6531,
      "step": 7261
    },
    {
      "epoch": 0.39188387027143706,
      "grad_norm": 1.0332783310162637,
      "learning_rate": 6.9992360009823905e-06,
      "loss": 0.4783,
      "step": 7262
    },
    {
      "epoch": 0.39193783389995146,
      "grad_norm": 0.9562874449110266,
      "learning_rate": 6.998516671407981e-06,
      "loss": 0.5066,
      "step": 7263
    },
    {
      "epoch": 0.3919917975284658,
      "grad_norm": 0.7928153506579074,
      "learning_rate": 6.997797298764689e-06,
      "loss": 0.3062,
      "step": 7264
    },
    {
      "epoch": 0.3920457611569802,
      "grad_norm": 1.191335448375976,
      "learning_rate": 6.9970778830731885e-06,
      "loss": 0.5226,
      "step": 7265
    },
    {
      "epoch": 0.3920997247854946,
      "grad_norm": 0.9146161021733743,
      "learning_rate": 6.9963584243541585e-06,
      "loss": 0.3885,
      "step": 7266
    },
    {
      "epoch": 0.39215368841400894,
      "grad_norm": 0.8630724937085266,
      "learning_rate": 6.9956389226282765e-06,
      "loss": 0.355,
      "step": 7267
    },
    {
      "epoch": 0.39220765204252334,
      "grad_norm": 0.9138588201961969,
      "learning_rate": 6.994919377916217e-06,
      "loss": 0.4228,
      "step": 7268
    },
    {
      "epoch": 0.39226161567103773,
      "grad_norm": 0.7964711662199958,
      "learning_rate": 6.994199790238667e-06,
      "loss": 0.3266,
      "step": 7269
    },
    {
      "epoch": 0.3923155792995521,
      "grad_norm": 1.1155913337915295,
      "learning_rate": 6.993480159616305e-06,
      "loss": 0.5108,
      "step": 7270
    },
    {
      "epoch": 0.3923695429280665,
      "grad_norm": 0.9455559613478185,
      "learning_rate": 6.992760486069812e-06,
      "loss": 0.4448,
      "step": 7271
    },
    {
      "epoch": 0.39242350655658087,
      "grad_norm": 0.8345209239811572,
      "learning_rate": 6.992040769619876e-06,
      "loss": 0.3708,
      "step": 7272
    },
    {
      "epoch": 0.39247747018509527,
      "grad_norm": 0.9116268634002278,
      "learning_rate": 6.99132101028718e-06,
      "loss": 0.5158,
      "step": 7273
    },
    {
      "epoch": 0.3925314338136096,
      "grad_norm": 0.9327978182193459,
      "learning_rate": 6.990601208092413e-06,
      "loss": 0.4622,
      "step": 7274
    },
    {
      "epoch": 0.392585397442124,
      "grad_norm": 0.9193700729033507,
      "learning_rate": 6.989881363056259e-06,
      "loss": 0.4865,
      "step": 7275
    },
    {
      "epoch": 0.3926393610706384,
      "grad_norm": 1.0297730939952272,
      "learning_rate": 6.989161475199409e-06,
      "loss": 0.5207,
      "step": 7276
    },
    {
      "epoch": 0.39269332469915275,
      "grad_norm": 1.010191024112803,
      "learning_rate": 6.988441544542556e-06,
      "loss": 0.5104,
      "step": 7277
    },
    {
      "epoch": 0.39274728832766714,
      "grad_norm": 0.9285584334818006,
      "learning_rate": 6.987721571106387e-06,
      "loss": 0.3639,
      "step": 7278
    },
    {
      "epoch": 0.39280125195618154,
      "grad_norm": 1.0291656196273573,
      "learning_rate": 6.987001554911598e-06,
      "loss": 0.3606,
      "step": 7279
    },
    {
      "epoch": 0.39285521558469594,
      "grad_norm": 1.0278733303581613,
      "learning_rate": 6.98628149597888e-06,
      "loss": 0.57,
      "step": 7280
    },
    {
      "epoch": 0.3929091792132103,
      "grad_norm": 1.4485565601033363,
      "learning_rate": 6.9855613943289326e-06,
      "loss": 0.5595,
      "step": 7281
    },
    {
      "epoch": 0.3929631428417247,
      "grad_norm": 1.0134234040328098,
      "learning_rate": 6.984841249982447e-06,
      "loss": 0.4677,
      "step": 7282
    },
    {
      "epoch": 0.3930171064702391,
      "grad_norm": 1.0850167792433496,
      "learning_rate": 6.9841210629601255e-06,
      "loss": 0.5431,
      "step": 7283
    },
    {
      "epoch": 0.3930710700987534,
      "grad_norm": 1.2117763336471414,
      "learning_rate": 6.983400833282665e-06,
      "loss": 0.6201,
      "step": 7284
    },
    {
      "epoch": 0.3931250337272678,
      "grad_norm": 0.9551810417962849,
      "learning_rate": 6.982680560970766e-06,
      "loss": 0.4107,
      "step": 7285
    },
    {
      "epoch": 0.3931789973557822,
      "grad_norm": 0.9634884333069673,
      "learning_rate": 6.981960246045128e-06,
      "loss": 0.5562,
      "step": 7286
    },
    {
      "epoch": 0.3932329609842966,
      "grad_norm": 0.9664024358454562,
      "learning_rate": 6.981239888526454e-06,
      "loss": 0.4762,
      "step": 7287
    },
    {
      "epoch": 0.39328692461281095,
      "grad_norm": 0.8788152396077874,
      "learning_rate": 6.980519488435451e-06,
      "loss": 0.4106,
      "step": 7288
    },
    {
      "epoch": 0.39334088824132535,
      "grad_norm": 0.965978351735014,
      "learning_rate": 6.9797990457928215e-06,
      "loss": 0.3329,
      "step": 7289
    },
    {
      "epoch": 0.39339485186983975,
      "grad_norm": 1.0332800407573766,
      "learning_rate": 6.9790785606192725e-06,
      "loss": 0.4745,
      "step": 7290
    },
    {
      "epoch": 0.3934488154983541,
      "grad_norm": 1.1177772761216382,
      "learning_rate": 6.978358032935512e-06,
      "loss": 0.5537,
      "step": 7291
    },
    {
      "epoch": 0.3935027791268685,
      "grad_norm": 1.0574971493488603,
      "learning_rate": 6.977637462762247e-06,
      "loss": 0.4304,
      "step": 7292
    },
    {
      "epoch": 0.3935567427553829,
      "grad_norm": 1.0835481091205992,
      "learning_rate": 6.976916850120186e-06,
      "loss": 0.5322,
      "step": 7293
    },
    {
      "epoch": 0.3936107063838973,
      "grad_norm": 1.011807646620418,
      "learning_rate": 6.976196195030045e-06,
      "loss": 0.4487,
      "step": 7294
    },
    {
      "epoch": 0.3936646700124116,
      "grad_norm": 1.1481946888803534,
      "learning_rate": 6.975475497512535e-06,
      "loss": 0.4752,
      "step": 7295
    },
    {
      "epoch": 0.393718633640926,
      "grad_norm": 1.1472699112619476,
      "learning_rate": 6.974754757588366e-06,
      "loss": 0.5453,
      "step": 7296
    },
    {
      "epoch": 0.3937725972694404,
      "grad_norm": 1.01127671171724,
      "learning_rate": 6.974033975278256e-06,
      "loss": 0.429,
      "step": 7297
    },
    {
      "epoch": 0.39382656089795476,
      "grad_norm": 1.0999096657764258,
      "learning_rate": 6.973313150602919e-06,
      "loss": 0.5535,
      "step": 7298
    },
    {
      "epoch": 0.39388052452646916,
      "grad_norm": 0.9157420891503849,
      "learning_rate": 6.972592283583075e-06,
      "loss": 0.4409,
      "step": 7299
    },
    {
      "epoch": 0.39393448815498355,
      "grad_norm": 0.6961270851459377,
      "learning_rate": 6.971871374239439e-06,
      "loss": 0.2959,
      "step": 7300
    },
    {
      "epoch": 0.3939884517834979,
      "grad_norm": 0.9839009173822624,
      "learning_rate": 6.971150422592735e-06,
      "loss": 0.5245,
      "step": 7301
    },
    {
      "epoch": 0.3940424154120123,
      "grad_norm": 0.9890475129561677,
      "learning_rate": 6.97042942866368e-06,
      "loss": 0.5016,
      "step": 7302
    },
    {
      "epoch": 0.3940963790405267,
      "grad_norm": 0.8942564408350887,
      "learning_rate": 6.969708392472997e-06,
      "loss": 0.3809,
      "step": 7303
    },
    {
      "epoch": 0.3941503426690411,
      "grad_norm": 1.1199570224069297,
      "learning_rate": 6.96898731404141e-06,
      "loss": 0.6085,
      "step": 7304
    },
    {
      "epoch": 0.39420430629755543,
      "grad_norm": 0.8810069279824563,
      "learning_rate": 6.968266193389645e-06,
      "loss": 0.4505,
      "step": 7305
    },
    {
      "epoch": 0.3942582699260698,
      "grad_norm": 0.9089628114739008,
      "learning_rate": 6.967545030538424e-06,
      "loss": 0.3548,
      "step": 7306
    },
    {
      "epoch": 0.3943122335545842,
      "grad_norm": 0.9842616244967018,
      "learning_rate": 6.966823825508475e-06,
      "loss": 0.4781,
      "step": 7307
    },
    {
      "epoch": 0.39436619718309857,
      "grad_norm": 0.8971557710401814,
      "learning_rate": 6.966102578320531e-06,
      "loss": 0.2965,
      "step": 7308
    },
    {
      "epoch": 0.39442016081161296,
      "grad_norm": 0.941848702688513,
      "learning_rate": 6.965381288995315e-06,
      "loss": 0.4286,
      "step": 7309
    },
    {
      "epoch": 0.39447412444012736,
      "grad_norm": 0.9010608924667852,
      "learning_rate": 6.964659957553559e-06,
      "loss": 0.4724,
      "step": 7310
    },
    {
      "epoch": 0.39452808806864176,
      "grad_norm": 0.9989821157248231,
      "learning_rate": 6.963938584015997e-06,
      "loss": 0.7211,
      "step": 7311
    },
    {
      "epoch": 0.3945820516971561,
      "grad_norm": 1.007520951789721,
      "learning_rate": 6.9632171684033614e-06,
      "loss": 0.499,
      "step": 7312
    },
    {
      "epoch": 0.3946360153256705,
      "grad_norm": 0.9393299682377289,
      "learning_rate": 6.962495710736386e-06,
      "loss": 0.3612,
      "step": 7313
    },
    {
      "epoch": 0.3946899789541849,
      "grad_norm": 1.115652712677839,
      "learning_rate": 6.961774211035805e-06,
      "loss": 0.4956,
      "step": 7314
    },
    {
      "epoch": 0.39474394258269924,
      "grad_norm": 0.9499080566972553,
      "learning_rate": 6.9610526693223565e-06,
      "loss": 0.341,
      "step": 7315
    },
    {
      "epoch": 0.39479790621121363,
      "grad_norm": 0.9507301207502706,
      "learning_rate": 6.960331085616778e-06,
      "loss": 0.5336,
      "step": 7316
    },
    {
      "epoch": 0.39485186983972803,
      "grad_norm": 1.1490695044847667,
      "learning_rate": 6.959609459939809e-06,
      "loss": 0.6681,
      "step": 7317
    },
    {
      "epoch": 0.39490583346824243,
      "grad_norm": 1.2738683892734768,
      "learning_rate": 6.9588877923121886e-06,
      "loss": 0.5697,
      "step": 7318
    },
    {
      "epoch": 0.39495979709675677,
      "grad_norm": 1.133284752242553,
      "learning_rate": 6.958166082754661e-06,
      "loss": 0.535,
      "step": 7319
    },
    {
      "epoch": 0.39501376072527117,
      "grad_norm": 0.8796959147060349,
      "learning_rate": 6.957444331287965e-06,
      "loss": 0.3798,
      "step": 7320
    },
    {
      "epoch": 0.39506772435378557,
      "grad_norm": 1.0908859427825204,
      "learning_rate": 6.9567225379328464e-06,
      "loss": 0.5771,
      "step": 7321
    },
    {
      "epoch": 0.3951216879822999,
      "grad_norm": 1.3565909705024344,
      "learning_rate": 6.956000702710051e-06,
      "loss": 0.602,
      "step": 7322
    },
    {
      "epoch": 0.3951756516108143,
      "grad_norm": 1.0733969199820161,
      "learning_rate": 6.955278825640323e-06,
      "loss": 0.4623,
      "step": 7323
    },
    {
      "epoch": 0.3952296152393287,
      "grad_norm": 1.168757410621451,
      "learning_rate": 6.954556906744412e-06,
      "loss": 0.5599,
      "step": 7324
    },
    {
      "epoch": 0.3952835788678431,
      "grad_norm": 0.9379821457022032,
      "learning_rate": 6.953834946043065e-06,
      "loss": 0.3355,
      "step": 7325
    },
    {
      "epoch": 0.39533754249635744,
      "grad_norm": 0.8635305358219658,
      "learning_rate": 6.953112943557035e-06,
      "loss": 0.4587,
      "step": 7326
    },
    {
      "epoch": 0.39539150612487184,
      "grad_norm": 1.175969633333356,
      "learning_rate": 6.952390899307066e-06,
      "loss": 0.487,
      "step": 7327
    },
    {
      "epoch": 0.39544546975338624,
      "grad_norm": 1.0520500182100254,
      "learning_rate": 6.951668813313919e-06,
      "loss": 0.5829,
      "step": 7328
    },
    {
      "epoch": 0.3954994333819006,
      "grad_norm": 1.0003832009783513,
      "learning_rate": 6.950946685598342e-06,
      "loss": 0.4606,
      "step": 7329
    },
    {
      "epoch": 0.395553397010415,
      "grad_norm": 0.8753920467546847,
      "learning_rate": 6.950224516181092e-06,
      "loss": 0.3909,
      "step": 7330
    },
    {
      "epoch": 0.3956073606389294,
      "grad_norm": 0.8331379090004266,
      "learning_rate": 6.949502305082923e-06,
      "loss": 0.4033,
      "step": 7331
    },
    {
      "epoch": 0.39566132426744377,
      "grad_norm": 1.145611283936693,
      "learning_rate": 6.948780052324594e-06,
      "loss": 0.5329,
      "step": 7332
    },
    {
      "epoch": 0.3957152878959581,
      "grad_norm": 0.987691911883217,
      "learning_rate": 6.948057757926863e-06,
      "loss": 0.4965,
      "step": 7333
    },
    {
      "epoch": 0.3957692515244725,
      "grad_norm": 1.0621368921858534,
      "learning_rate": 6.947335421910487e-06,
      "loss": 0.4641,
      "step": 7334
    },
    {
      "epoch": 0.3958232151529869,
      "grad_norm": 0.9370148205996476,
      "learning_rate": 6.9466130442962294e-06,
      "loss": 0.4491,
      "step": 7335
    },
    {
      "epoch": 0.39587717878150125,
      "grad_norm": 0.9166660282323302,
      "learning_rate": 6.945890625104852e-06,
      "loss": 0.4576,
      "step": 7336
    },
    {
      "epoch": 0.39593114241001565,
      "grad_norm": 0.8319097175821002,
      "learning_rate": 6.945168164357117e-06,
      "loss": 0.3466,
      "step": 7337
    },
    {
      "epoch": 0.39598510603853004,
      "grad_norm": 0.8842716313565426,
      "learning_rate": 6.944445662073787e-06,
      "loss": 0.4657,
      "step": 7338
    },
    {
      "epoch": 0.3960390696670444,
      "grad_norm": 1.2257993139512444,
      "learning_rate": 6.943723118275631e-06,
      "loss": 0.5387,
      "step": 7339
    },
    {
      "epoch": 0.3960930332955588,
      "grad_norm": 1.0719152401027197,
      "learning_rate": 6.943000532983413e-06,
      "loss": 0.5147,
      "step": 7340
    },
    {
      "epoch": 0.3961469969240732,
      "grad_norm": 1.0999414301822612,
      "learning_rate": 6.9422779062179015e-06,
      "loss": 0.5486,
      "step": 7341
    },
    {
      "epoch": 0.3962009605525876,
      "grad_norm": 1.0946976938022532,
      "learning_rate": 6.941555237999866e-06,
      "loss": 0.4823,
      "step": 7342
    },
    {
      "epoch": 0.3962549241811019,
      "grad_norm": 0.8518769356416481,
      "learning_rate": 6.940832528350077e-06,
      "loss": 0.3982,
      "step": 7343
    },
    {
      "epoch": 0.3963088878096163,
      "grad_norm": 0.8772455185616743,
      "learning_rate": 6.940109777289304e-06,
      "loss": 0.434,
      "step": 7344
    },
    {
      "epoch": 0.3963628514381307,
      "grad_norm": 0.9882598322136172,
      "learning_rate": 6.9393869848383214e-06,
      "loss": 0.4677,
      "step": 7345
    },
    {
      "epoch": 0.39641681506664506,
      "grad_norm": 0.9952991142534502,
      "learning_rate": 6.938664151017903e-06,
      "loss": 0.4449,
      "step": 7346
    },
    {
      "epoch": 0.39647077869515945,
      "grad_norm": 0.9814742887229432,
      "learning_rate": 6.937941275848823e-06,
      "loss": 0.4139,
      "step": 7347
    },
    {
      "epoch": 0.39652474232367385,
      "grad_norm": 1.155573138112056,
      "learning_rate": 6.937218359351857e-06,
      "loss": 0.5893,
      "step": 7348
    },
    {
      "epoch": 0.39657870595218825,
      "grad_norm": 0.9376832875631692,
      "learning_rate": 6.936495401547784e-06,
      "loss": 0.3567,
      "step": 7349
    },
    {
      "epoch": 0.3966326695807026,
      "grad_norm": 0.9704350101745409,
      "learning_rate": 6.935772402457383e-06,
      "loss": 0.4834,
      "step": 7350
    },
    {
      "epoch": 0.396686633209217,
      "grad_norm": 1.2282985235158532,
      "learning_rate": 6.9350493621014294e-06,
      "loss": 0.698,
      "step": 7351
    },
    {
      "epoch": 0.3967405968377314,
      "grad_norm": 1.2233005019209653,
      "learning_rate": 6.9343262805007086e-06,
      "loss": 0.492,
      "step": 7352
    },
    {
      "epoch": 0.3967945604662457,
      "grad_norm": 0.9335233370039169,
      "learning_rate": 6.933603157676001e-06,
      "loss": 0.4409,
      "step": 7353
    },
    {
      "epoch": 0.3968485240947601,
      "grad_norm": 0.9817664727059547,
      "learning_rate": 6.93287999364809e-06,
      "loss": 0.5589,
      "step": 7354
    },
    {
      "epoch": 0.3969024877232745,
      "grad_norm": 1.1679786123238152,
      "learning_rate": 6.932156788437761e-06,
      "loss": 0.4845,
      "step": 7355
    },
    {
      "epoch": 0.3969564513517889,
      "grad_norm": 1.0199058549748143,
      "learning_rate": 6.9314335420657985e-06,
      "loss": 0.4011,
      "step": 7356
    },
    {
      "epoch": 0.39701041498030326,
      "grad_norm": 0.9181191607718128,
      "learning_rate": 6.930710254552989e-06,
      "loss": 0.4672,
      "step": 7357
    },
    {
      "epoch": 0.39706437860881766,
      "grad_norm": 0.78227009410512,
      "learning_rate": 6.929986925920121e-06,
      "loss": 0.2682,
      "step": 7358
    },
    {
      "epoch": 0.39711834223733206,
      "grad_norm": 0.8952285701485744,
      "learning_rate": 6.929263556187985e-06,
      "loss": 0.4715,
      "step": 7359
    },
    {
      "epoch": 0.3971723058658464,
      "grad_norm": 0.9644531706227256,
      "learning_rate": 6.92854014537737e-06,
      "loss": 0.4303,
      "step": 7360
    },
    {
      "epoch": 0.3972262694943608,
      "grad_norm": 1.0519019190237167,
      "learning_rate": 6.927816693509067e-06,
      "loss": 0.569,
      "step": 7361
    },
    {
      "epoch": 0.3972802331228752,
      "grad_norm": 0.9866100255963471,
      "learning_rate": 6.92709320060387e-06,
      "loss": 0.3967,
      "step": 7362
    },
    {
      "epoch": 0.3973341967513896,
      "grad_norm": 1.0318057617500094,
      "learning_rate": 6.926369666682573e-06,
      "loss": 0.4257,
      "step": 7363
    },
    {
      "epoch": 0.39738816037990393,
      "grad_norm": 0.9264055715588506,
      "learning_rate": 6.925646091765969e-06,
      "loss": 0.3904,
      "step": 7364
    },
    {
      "epoch": 0.39744212400841833,
      "grad_norm": 1.1007287131951302,
      "learning_rate": 6.924922475874855e-06,
      "loss": 0.4957,
      "step": 7365
    },
    {
      "epoch": 0.3974960876369327,
      "grad_norm": 0.9599885795092227,
      "learning_rate": 6.924198819030032e-06,
      "loss": 0.444,
      "step": 7366
    },
    {
      "epoch": 0.39755005126544707,
      "grad_norm": 0.7760640523358238,
      "learning_rate": 6.923475121252295e-06,
      "loss": 0.3337,
      "step": 7367
    },
    {
      "epoch": 0.39760401489396147,
      "grad_norm": 1.0094978284850458,
      "learning_rate": 6.922751382562445e-06,
      "loss": 0.437,
      "step": 7368
    },
    {
      "epoch": 0.39765797852247586,
      "grad_norm": 0.9207834787686776,
      "learning_rate": 6.9220276029812825e-06,
      "loss": 0.484,
      "step": 7369
    },
    {
      "epoch": 0.3977119421509902,
      "grad_norm": 0.8557650507547608,
      "learning_rate": 6.9213037825296095e-06,
      "loss": 0.4155,
      "step": 7370
    },
    {
      "epoch": 0.3977659057795046,
      "grad_norm": 1.101239006968196,
      "learning_rate": 6.92057992122823e-06,
      "loss": 0.6449,
      "step": 7371
    },
    {
      "epoch": 0.397819869408019,
      "grad_norm": 1.171464565375767,
      "learning_rate": 6.919856019097948e-06,
      "loss": 0.5332,
      "step": 7372
    },
    {
      "epoch": 0.3978738330365334,
      "grad_norm": 1.020880702065345,
      "learning_rate": 6.919132076159569e-06,
      "loss": 0.4323,
      "step": 7373
    },
    {
      "epoch": 0.39792779666504774,
      "grad_norm": 1.0971454769292677,
      "learning_rate": 6.918408092433902e-06,
      "loss": 0.4816,
      "step": 7374
    },
    {
      "epoch": 0.39798176029356214,
      "grad_norm": 1.2714749410406507,
      "learning_rate": 6.917684067941751e-06,
      "loss": 0.6774,
      "step": 7375
    },
    {
      "epoch": 0.39803572392207653,
      "grad_norm": 1.1708828346481746,
      "learning_rate": 6.916960002703927e-06,
      "loss": 0.5032,
      "step": 7376
    },
    {
      "epoch": 0.3980896875505909,
      "grad_norm": 1.0156931386360397,
      "learning_rate": 6.916235896741243e-06,
      "loss": 0.428,
      "step": 7377
    },
    {
      "epoch": 0.3981436511791053,
      "grad_norm": 1.1102423042048066,
      "learning_rate": 6.9155117500745065e-06,
      "loss": 0.4033,
      "step": 7378
    },
    {
      "epoch": 0.39819761480761967,
      "grad_norm": 0.8929356063250794,
      "learning_rate": 6.914787562724533e-06,
      "loss": 0.3748,
      "step": 7379
    },
    {
      "epoch": 0.39825157843613407,
      "grad_norm": 0.9073345389577914,
      "learning_rate": 6.914063334712136e-06,
      "loss": 0.4249,
      "step": 7380
    },
    {
      "epoch": 0.3983055420646484,
      "grad_norm": 0.8023287248387875,
      "learning_rate": 6.913339066058129e-06,
      "loss": 0.3931,
      "step": 7381
    },
    {
      "epoch": 0.3983595056931628,
      "grad_norm": 1.288212264311498,
      "learning_rate": 6.912614756783329e-06,
      "loss": 0.7606,
      "step": 7382
    },
    {
      "epoch": 0.3984134693216772,
      "grad_norm": 1.0888621182649214,
      "learning_rate": 6.911890406908554e-06,
      "loss": 0.5458,
      "step": 7383
    },
    {
      "epoch": 0.39846743295019155,
      "grad_norm": 1.2057243393225157,
      "learning_rate": 6.9111660164546225e-06,
      "loss": 0.7101,
      "step": 7384
    },
    {
      "epoch": 0.39852139657870594,
      "grad_norm": 1.0782923834299436,
      "learning_rate": 6.910441585442353e-06,
      "loss": 0.4761,
      "step": 7385
    },
    {
      "epoch": 0.39857536020722034,
      "grad_norm": 1.0737868329773743,
      "learning_rate": 6.909717113892567e-06,
      "loss": 0.4911,
      "step": 7386
    },
    {
      "epoch": 0.39862932383573474,
      "grad_norm": 1.087461947949343,
      "learning_rate": 6.908992601826087e-06,
      "loss": 0.4479,
      "step": 7387
    },
    {
      "epoch": 0.3986832874642491,
      "grad_norm": 1.1554375256611862,
      "learning_rate": 6.9082680492637364e-06,
      "loss": 0.5771,
      "step": 7388
    },
    {
      "epoch": 0.3987372510927635,
      "grad_norm": 0.921951538945157,
      "learning_rate": 6.907543456226339e-06,
      "loss": 0.406,
      "step": 7389
    },
    {
      "epoch": 0.3987912147212779,
      "grad_norm": 1.1882786370681226,
      "learning_rate": 6.90681882273472e-06,
      "loss": 0.6643,
      "step": 7390
    },
    {
      "epoch": 0.3988451783497922,
      "grad_norm": 1.059806028332631,
      "learning_rate": 6.906094148809708e-06,
      "loss": 0.4533,
      "step": 7391
    },
    {
      "epoch": 0.3988991419783066,
      "grad_norm": 0.9999288894521997,
      "learning_rate": 6.905369434472127e-06,
      "loss": 0.4158,
      "step": 7392
    },
    {
      "epoch": 0.398953105606821,
      "grad_norm": 1.0450696152711116,
      "learning_rate": 6.904644679742809e-06,
      "loss": 0.4818,
      "step": 7393
    },
    {
      "epoch": 0.3990070692353354,
      "grad_norm": 1.2644153026381855,
      "learning_rate": 6.9039198846425836e-06,
      "loss": 0.4476,
      "step": 7394
    },
    {
      "epoch": 0.39906103286384975,
      "grad_norm": 0.8985276129022756,
      "learning_rate": 6.903195049192285e-06,
      "loss": 0.361,
      "step": 7395
    },
    {
      "epoch": 0.39911499649236415,
      "grad_norm": 0.7512154533590325,
      "learning_rate": 6.902470173412739e-06,
      "loss": 0.3129,
      "step": 7396
    },
    {
      "epoch": 0.39916896012087855,
      "grad_norm": 0.8918173668190926,
      "learning_rate": 6.901745257324786e-06,
      "loss": 0.3586,
      "step": 7397
    },
    {
      "epoch": 0.3992229237493929,
      "grad_norm": 1.2094021443982044,
      "learning_rate": 6.901020300949257e-06,
      "loss": 0.601,
      "step": 7398
    },
    {
      "epoch": 0.3992768873779073,
      "grad_norm": 1.23906409757255,
      "learning_rate": 6.900295304306987e-06,
      "loss": 0.5486,
      "step": 7399
    },
    {
      "epoch": 0.3993308510064217,
      "grad_norm": 1.2584505419173238,
      "learning_rate": 6.899570267418818e-06,
      "loss": 0.5424,
      "step": 7400
    },
    {
      "epoch": 0.3993848146349361,
      "grad_norm": 0.9360162736217269,
      "learning_rate": 6.898845190305583e-06,
      "loss": 0.4827,
      "step": 7401
    },
    {
      "epoch": 0.3994387782634504,
      "grad_norm": 1.046184058017311,
      "learning_rate": 6.898120072988127e-06,
      "loss": 0.4719,
      "step": 7402
    },
    {
      "epoch": 0.3994927418919648,
      "grad_norm": 0.8787748917289807,
      "learning_rate": 6.897394915487284e-06,
      "loss": 0.4037,
      "step": 7403
    },
    {
      "epoch": 0.3995467055204792,
      "grad_norm": 1.0198588765783025,
      "learning_rate": 6.8966697178239e-06,
      "loss": 0.519,
      "step": 7404
    },
    {
      "epoch": 0.39960066914899356,
      "grad_norm": 1.092223303801719,
      "learning_rate": 6.895944480018819e-06,
      "loss": 0.4816,
      "step": 7405
    },
    {
      "epoch": 0.39965463277750796,
      "grad_norm": 1.1756305919461671,
      "learning_rate": 6.895219202092882e-06,
      "loss": 0.6118,
      "step": 7406
    },
    {
      "epoch": 0.39970859640602235,
      "grad_norm": 1.1759752493577824,
      "learning_rate": 6.894493884066934e-06,
      "loss": 0.4735,
      "step": 7407
    },
    {
      "epoch": 0.3997625600345367,
      "grad_norm": 0.9398615015415274,
      "learning_rate": 6.893768525961825e-06,
      "loss": 0.3725,
      "step": 7408
    },
    {
      "epoch": 0.3998165236630511,
      "grad_norm": 1.0730535084503168,
      "learning_rate": 6.893043127798398e-06,
      "loss": 0.536,
      "step": 7409
    },
    {
      "epoch": 0.3998704872915655,
      "grad_norm": 0.9915702891297549,
      "learning_rate": 6.8923176895975065e-06,
      "loss": 0.5206,
      "step": 7410
    },
    {
      "epoch": 0.3999244509200799,
      "grad_norm": 1.1174807128643185,
      "learning_rate": 6.891592211379995e-06,
      "loss": 0.5537,
      "step": 7411
    },
    {
      "epoch": 0.39997841454859423,
      "grad_norm": 0.8705396537027286,
      "learning_rate": 6.890866693166719e-06,
      "loss": 0.3489,
      "step": 7412
    },
    {
      "epoch": 0.4000323781771086,
      "grad_norm": 0.9287044392606056,
      "learning_rate": 6.890141134978529e-06,
      "loss": 0.5821,
      "step": 7413
    },
    {
      "epoch": 0.400086341805623,
      "grad_norm": 1.0479318322055813,
      "learning_rate": 6.8894155368362766e-06,
      "loss": 0.3686,
      "step": 7414
    },
    {
      "epoch": 0.40014030543413737,
      "grad_norm": 0.799069691525588,
      "learning_rate": 6.888689898760819e-06,
      "loss": 0.4302,
      "step": 7415
    },
    {
      "epoch": 0.40019426906265176,
      "grad_norm": 0.8590001903091148,
      "learning_rate": 6.88796422077301e-06,
      "loss": 0.3657,
      "step": 7416
    },
    {
      "epoch": 0.40024823269116616,
      "grad_norm": 1.1169965896271177,
      "learning_rate": 6.887238502893707e-06,
      "loss": 0.5201,
      "step": 7417
    },
    {
      "epoch": 0.40030219631968056,
      "grad_norm": 1.0343760707293472,
      "learning_rate": 6.886512745143768e-06,
      "loss": 0.4912,
      "step": 7418
    },
    {
      "epoch": 0.4003561599481949,
      "grad_norm": 1.053534393407756,
      "learning_rate": 6.8857869475440505e-06,
      "loss": 0.5298,
      "step": 7419
    },
    {
      "epoch": 0.4004101235767093,
      "grad_norm": 1.0355645691339856,
      "learning_rate": 6.885061110115418e-06,
      "loss": 0.6798,
      "step": 7420
    },
    {
      "epoch": 0.4004640872052237,
      "grad_norm": 0.9813152412824858,
      "learning_rate": 6.884335232878728e-06,
      "loss": 0.456,
      "step": 7421
    },
    {
      "epoch": 0.40051805083373804,
      "grad_norm": 1.0180771388818712,
      "learning_rate": 6.883609315854846e-06,
      "loss": 0.5082,
      "step": 7422
    },
    {
      "epoch": 0.40057201446225243,
      "grad_norm": 1.0068640739746724,
      "learning_rate": 6.882883359064634e-06,
      "loss": 0.5287,
      "step": 7423
    },
    {
      "epoch": 0.40062597809076683,
      "grad_norm": 1.1611763703645002,
      "learning_rate": 6.882157362528955e-06,
      "loss": 0.506,
      "step": 7424
    },
    {
      "epoch": 0.40067994171928123,
      "grad_norm": 1.1598748801714478,
      "learning_rate": 6.881431326268678e-06,
      "loss": 0.5628,
      "step": 7425
    },
    {
      "epoch": 0.40073390534779557,
      "grad_norm": 1.057387444705196,
      "learning_rate": 6.88070525030467e-06,
      "loss": 0.4574,
      "step": 7426
    },
    {
      "epoch": 0.40078786897630997,
      "grad_norm": 1.236511696570724,
      "learning_rate": 6.879979134657797e-06,
      "loss": 0.6323,
      "step": 7427
    },
    {
      "epoch": 0.40084183260482437,
      "grad_norm": 1.0618581146782926,
      "learning_rate": 6.8792529793489294e-06,
      "loss": 0.6071,
      "step": 7428
    },
    {
      "epoch": 0.4008957962333387,
      "grad_norm": 0.9514098954990928,
      "learning_rate": 6.878526784398938e-06,
      "loss": 0.5519,
      "step": 7429
    },
    {
      "epoch": 0.4009497598618531,
      "grad_norm": 0.9513325328957637,
      "learning_rate": 6.877800549828694e-06,
      "loss": 0.3678,
      "step": 7430
    },
    {
      "epoch": 0.4010037234903675,
      "grad_norm": 1.2687720805135903,
      "learning_rate": 6.877074275659069e-06,
      "loss": 0.5549,
      "step": 7431
    },
    {
      "epoch": 0.4010576871188819,
      "grad_norm": 0.8961639237626571,
      "learning_rate": 6.876347961910938e-06,
      "loss": 0.3899,
      "step": 7432
    },
    {
      "epoch": 0.40111165074739624,
      "grad_norm": 0.8503756505926594,
      "learning_rate": 6.8756216086051766e-06,
      "loss": 0.4103,
      "step": 7433
    },
    {
      "epoch": 0.40116561437591064,
      "grad_norm": 0.9034626615202787,
      "learning_rate": 6.87489521576266e-06,
      "loss": 0.41,
      "step": 7434
    },
    {
      "epoch": 0.40121957800442504,
      "grad_norm": 1.2056837234997013,
      "learning_rate": 6.874168783404266e-06,
      "loss": 0.564,
      "step": 7435
    },
    {
      "epoch": 0.4012735416329394,
      "grad_norm": 1.1357757377519186,
      "learning_rate": 6.873442311550873e-06,
      "loss": 0.5908,
      "step": 7436
    },
    {
      "epoch": 0.4013275052614538,
      "grad_norm": 0.9313218853284131,
      "learning_rate": 6.872715800223359e-06,
      "loss": 0.3783,
      "step": 7437
    },
    {
      "epoch": 0.4013814688899682,
      "grad_norm": 1.155563053243819,
      "learning_rate": 6.871989249442605e-06,
      "loss": 0.5926,
      "step": 7438
    },
    {
      "epoch": 0.4014354325184825,
      "grad_norm": 0.8996284899361667,
      "learning_rate": 6.871262659229496e-06,
      "loss": 0.4025,
      "step": 7439
    },
    {
      "epoch": 0.4014893961469969,
      "grad_norm": 0.8845793529901812,
      "learning_rate": 6.870536029604912e-06,
      "loss": 0.407,
      "step": 7440
    },
    {
      "epoch": 0.4015433597755113,
      "grad_norm": 0.8223896169049989,
      "learning_rate": 6.8698093605897365e-06,
      "loss": 0.4421,
      "step": 7441
    },
    {
      "epoch": 0.4015973234040257,
      "grad_norm": 0.9414673973948463,
      "learning_rate": 6.869082652204857e-06,
      "loss": 0.408,
      "step": 7442
    },
    {
      "epoch": 0.40165128703254005,
      "grad_norm": 0.9085305304075144,
      "learning_rate": 6.868355904471158e-06,
      "loss": 0.4668,
      "step": 7443
    },
    {
      "epoch": 0.40170525066105445,
      "grad_norm": 1.0312921882842883,
      "learning_rate": 6.867629117409529e-06,
      "loss": 0.4487,
      "step": 7444
    },
    {
      "epoch": 0.40175921428956884,
      "grad_norm": 1.200632082204616,
      "learning_rate": 6.866902291040856e-06,
      "loss": 0.5366,
      "step": 7445
    },
    {
      "epoch": 0.4018131779180832,
      "grad_norm": 0.9881660640315135,
      "learning_rate": 6.86617542538603e-06,
      "loss": 0.3806,
      "step": 7446
    },
    {
      "epoch": 0.4018671415465976,
      "grad_norm": 0.929625133655214,
      "learning_rate": 6.865448520465943e-06,
      "loss": 0.3999,
      "step": 7447
    },
    {
      "epoch": 0.401921105175112,
      "grad_norm": 1.1174250382244793,
      "learning_rate": 6.8647215763014845e-06,
      "loss": 0.6896,
      "step": 7448
    },
    {
      "epoch": 0.4019750688036264,
      "grad_norm": 1.0038643175666273,
      "learning_rate": 6.8639945929135485e-06,
      "loss": 0.3346,
      "step": 7449
    },
    {
      "epoch": 0.4020290324321407,
      "grad_norm": 1.1563592668375249,
      "learning_rate": 6.863267570323032e-06,
      "loss": 0.5085,
      "step": 7450
    },
    {
      "epoch": 0.4020829960606551,
      "grad_norm": 1.1116327283424083,
      "learning_rate": 6.862540508550827e-06,
      "loss": 0.564,
      "step": 7451
    },
    {
      "epoch": 0.4021369596891695,
      "grad_norm": 1.1603719437970714,
      "learning_rate": 6.86181340761783e-06,
      "loss": 0.568,
      "step": 7452
    },
    {
      "epoch": 0.40219092331768386,
      "grad_norm": 1.058065333800366,
      "learning_rate": 6.861086267544942e-06,
      "loss": 0.5885,
      "step": 7453
    },
    {
      "epoch": 0.40224488694619825,
      "grad_norm": 1.0069163159700363,
      "learning_rate": 6.860359088353057e-06,
      "loss": 0.5453,
      "step": 7454
    },
    {
      "epoch": 0.40229885057471265,
      "grad_norm": 0.9385666636293563,
      "learning_rate": 6.859631870063077e-06,
      "loss": 0.6166,
      "step": 7455
    },
    {
      "epoch": 0.40235281420322705,
      "grad_norm": 1.3012322873709101,
      "learning_rate": 6.858904612695906e-06,
      "loss": 0.5792,
      "step": 7456
    },
    {
      "epoch": 0.4024067778317414,
      "grad_norm": 0.9776935961570965,
      "learning_rate": 6.858177316272441e-06,
      "loss": 0.3797,
      "step": 7457
    },
    {
      "epoch": 0.4024607414602558,
      "grad_norm": 1.1118260821267112,
      "learning_rate": 6.857449980813589e-06,
      "loss": 0.6064,
      "step": 7458
    },
    {
      "epoch": 0.4025147050887702,
      "grad_norm": 1.3267568383972237,
      "learning_rate": 6.856722606340251e-06,
      "loss": 0.6381,
      "step": 7459
    },
    {
      "epoch": 0.40256866871728453,
      "grad_norm": 1.04924190290178,
      "learning_rate": 6.855995192873336e-06,
      "loss": 0.4451,
      "step": 7460
    },
    {
      "epoch": 0.4026226323457989,
      "grad_norm": 1.0624287744130938,
      "learning_rate": 6.855267740433748e-06,
      "loss": 0.5318,
      "step": 7461
    },
    {
      "epoch": 0.4026765959743133,
      "grad_norm": 0.9697913526782592,
      "learning_rate": 6.854540249042397e-06,
      "loss": 0.4564,
      "step": 7462
    },
    {
      "epoch": 0.4027305596028277,
      "grad_norm": 0.8758203230983929,
      "learning_rate": 6.85381271872019e-06,
      "loss": 0.4214,
      "step": 7463
    },
    {
      "epoch": 0.40278452323134206,
      "grad_norm": 1.0358890284903266,
      "learning_rate": 6.853085149488038e-06,
      "loss": 0.4522,
      "step": 7464
    },
    {
      "epoch": 0.40283848685985646,
      "grad_norm": 1.0074240561037584,
      "learning_rate": 6.852357541366851e-06,
      "loss": 0.5253,
      "step": 7465
    },
    {
      "epoch": 0.40289245048837086,
      "grad_norm": 1.0884130227220628,
      "learning_rate": 6.851629894377542e-06,
      "loss": 0.4924,
      "step": 7466
    },
    {
      "epoch": 0.4029464141168852,
      "grad_norm": 1.1759556250156717,
      "learning_rate": 6.850902208541024e-06,
      "loss": 0.4528,
      "step": 7467
    },
    {
      "epoch": 0.4030003777453996,
      "grad_norm": 0.880225177163445,
      "learning_rate": 6.850174483878211e-06,
      "loss": 0.4197,
      "step": 7468
    },
    {
      "epoch": 0.403054341373914,
      "grad_norm": 1.0382024707070134,
      "learning_rate": 6.849446720410019e-06,
      "loss": 0.4766,
      "step": 7469
    },
    {
      "epoch": 0.4031083050024284,
      "grad_norm": 0.953585543028039,
      "learning_rate": 6.8487189181573656e-06,
      "loss": 0.4752,
      "step": 7470
    },
    {
      "epoch": 0.40316226863094273,
      "grad_norm": 1.0264949522896405,
      "learning_rate": 6.847991077141168e-06,
      "loss": 0.5286,
      "step": 7471
    },
    {
      "epoch": 0.40321623225945713,
      "grad_norm": 1.0825797290921273,
      "learning_rate": 6.847263197382345e-06,
      "loss": 0.4685,
      "step": 7472
    },
    {
      "epoch": 0.4032701958879715,
      "grad_norm": 1.0800804442899046,
      "learning_rate": 6.846535278901815e-06,
      "loss": 0.4152,
      "step": 7473
    },
    {
      "epoch": 0.40332415951648587,
      "grad_norm": 0.9680618608713457,
      "learning_rate": 6.845807321720503e-06,
      "loss": 0.6097,
      "step": 7474
    },
    {
      "epoch": 0.40337812314500027,
      "grad_norm": 0.8809066812618978,
      "learning_rate": 6.845079325859326e-06,
      "loss": 0.3879,
      "step": 7475
    },
    {
      "epoch": 0.40343208677351466,
      "grad_norm": 0.9438149331246248,
      "learning_rate": 6.844351291339211e-06,
      "loss": 0.4552,
      "step": 7476
    },
    {
      "epoch": 0.403486050402029,
      "grad_norm": 1.1426728305193914,
      "learning_rate": 6.843623218181081e-06,
      "loss": 0.4203,
      "step": 7477
    },
    {
      "epoch": 0.4035400140305434,
      "grad_norm": 0.9797420152753918,
      "learning_rate": 6.8428951064058634e-06,
      "loss": 0.4744,
      "step": 7478
    },
    {
      "epoch": 0.4035939776590578,
      "grad_norm": 1.217060320735071,
      "learning_rate": 6.842166956034483e-06,
      "loss": 0.6055,
      "step": 7479
    },
    {
      "epoch": 0.4036479412875722,
      "grad_norm": 0.9245265391375899,
      "learning_rate": 6.841438767087867e-06,
      "loss": 0.4398,
      "step": 7480
    },
    {
      "epoch": 0.40370190491608654,
      "grad_norm": 0.965177937195023,
      "learning_rate": 6.840710539586948e-06,
      "loss": 0.4907,
      "step": 7481
    },
    {
      "epoch": 0.40375586854460094,
      "grad_norm": 1.2629263114110907,
      "learning_rate": 6.839982273552651e-06,
      "loss": 0.6956,
      "step": 7482
    },
    {
      "epoch": 0.40380983217311534,
      "grad_norm": 1.1470679668772006,
      "learning_rate": 6.83925396900591e-06,
      "loss": 0.4389,
      "step": 7483
    },
    {
      "epoch": 0.4038637958016297,
      "grad_norm": 0.9178001033842014,
      "learning_rate": 6.838525625967657e-06,
      "loss": 0.4685,
      "step": 7484
    },
    {
      "epoch": 0.4039177594301441,
      "grad_norm": 1.0222283180441178,
      "learning_rate": 6.837797244458826e-06,
      "loss": 0.5736,
      "step": 7485
    },
    {
      "epoch": 0.40397172305865847,
      "grad_norm": 0.7666455893795111,
      "learning_rate": 6.837068824500348e-06,
      "loss": 0.2732,
      "step": 7486
    },
    {
      "epoch": 0.40402568668717287,
      "grad_norm": 1.0843833044925837,
      "learning_rate": 6.836340366113163e-06,
      "loss": 0.435,
      "step": 7487
    },
    {
      "epoch": 0.4040796503156872,
      "grad_norm": 1.0063179762407373,
      "learning_rate": 6.835611869318206e-06,
      "loss": 0.5379,
      "step": 7488
    },
    {
      "epoch": 0.4041336139442016,
      "grad_norm": 0.9843475557023736,
      "learning_rate": 6.834883334136413e-06,
      "loss": 0.5725,
      "step": 7489
    },
    {
      "epoch": 0.404187577572716,
      "grad_norm": 1.0185877060506705,
      "learning_rate": 6.834154760588723e-06,
      "loss": 0.5264,
      "step": 7490
    },
    {
      "epoch": 0.40424154120123035,
      "grad_norm": 0.9590359741842849,
      "learning_rate": 6.83342614869608e-06,
      "loss": 0.4472,
      "step": 7491
    },
    {
      "epoch": 0.40429550482974475,
      "grad_norm": 1.183126844254113,
      "learning_rate": 6.832697498479419e-06,
      "loss": 0.7068,
      "step": 7492
    },
    {
      "epoch": 0.40434946845825914,
      "grad_norm": 1.1307991526578705,
      "learning_rate": 6.831968809959687e-06,
      "loss": 0.5922,
      "step": 7493
    },
    {
      "epoch": 0.40440343208677354,
      "grad_norm": 1.0458904175067592,
      "learning_rate": 6.831240083157827e-06,
      "loss": 0.4676,
      "step": 7494
    },
    {
      "epoch": 0.4044573957152879,
      "grad_norm": 0.9768321621729886,
      "learning_rate": 6.83051131809478e-06,
      "loss": 0.4164,
      "step": 7495
    },
    {
      "epoch": 0.4045113593438023,
      "grad_norm": 1.0137345756288594,
      "learning_rate": 6.829782514791494e-06,
      "loss": 0.4322,
      "step": 7496
    },
    {
      "epoch": 0.4045653229723167,
      "grad_norm": 1.110675615134294,
      "learning_rate": 6.829053673268914e-06,
      "loss": 0.5366,
      "step": 7497
    },
    {
      "epoch": 0.404619286600831,
      "grad_norm": 0.8597882266362329,
      "learning_rate": 6.8283247935479905e-06,
      "loss": 0.3379,
      "step": 7498
    },
    {
      "epoch": 0.4046732502293454,
      "grad_norm": 1.3741169040876424,
      "learning_rate": 6.827595875649668e-06,
      "loss": 0.5546,
      "step": 7499
    },
    {
      "epoch": 0.4047272138578598,
      "grad_norm": 1.219943334540381,
      "learning_rate": 6.8268669195949e-06,
      "loss": 0.4878,
      "step": 7500
    },
    {
      "epoch": 0.4047272138578598,
      "eval_loss": 0.5583013892173767,
      "eval_runtime": 165.7815,
      "eval_samples_per_second": 20.744,
      "eval_steps_per_second": 0.869,
      "step": 7500
    },
    {
      "epoch": 0.4047811774863742,
      "grad_norm": 1.0552853424627657,
      "learning_rate": 6.826137925404635e-06,
      "loss": 0.6024,
      "step": 7501
    },
    {
      "epoch": 0.40483514111488855,
      "grad_norm": 0.9131848846130454,
      "learning_rate": 6.825408893099828e-06,
      "loss": 0.2693,
      "step": 7502
    },
    {
      "epoch": 0.40488910474340295,
      "grad_norm": 1.140690442601447,
      "learning_rate": 6.824679822701428e-06,
      "loss": 0.5655,
      "step": 7503
    },
    {
      "epoch": 0.40494306837191735,
      "grad_norm": 0.8196183357076992,
      "learning_rate": 6.823950714230393e-06,
      "loss": 0.4239,
      "step": 7504
    },
    {
      "epoch": 0.4049970320004317,
      "grad_norm": 0.9605209575037147,
      "learning_rate": 6.823221567707677e-06,
      "loss": 0.4902,
      "step": 7505
    },
    {
      "epoch": 0.4050509956289461,
      "grad_norm": 0.9419530362151369,
      "learning_rate": 6.822492383154235e-06,
      "loss": 0.346,
      "step": 7506
    },
    {
      "epoch": 0.4051049592574605,
      "grad_norm": 0.9590137525125092,
      "learning_rate": 6.821763160591024e-06,
      "loss": 0.576,
      "step": 7507
    },
    {
      "epoch": 0.4051589228859748,
      "grad_norm": 0.8877022729819032,
      "learning_rate": 6.821033900039008e-06,
      "loss": 0.4281,
      "step": 7508
    },
    {
      "epoch": 0.4052128865144892,
      "grad_norm": 0.8905284049940813,
      "learning_rate": 6.8203046015191395e-06,
      "loss": 0.4098,
      "step": 7509
    },
    {
      "epoch": 0.4052668501430036,
      "grad_norm": 1.2081350397547608,
      "learning_rate": 6.819575265052384e-06,
      "loss": 0.6063,
      "step": 7510
    },
    {
      "epoch": 0.405320813771518,
      "grad_norm": 1.294975696563038,
      "learning_rate": 6.818845890659702e-06,
      "loss": 0.5337,
      "step": 7511
    },
    {
      "epoch": 0.40537477740003236,
      "grad_norm": 0.7946295939688212,
      "learning_rate": 6.818116478362058e-06,
      "loss": 0.2971,
      "step": 7512
    },
    {
      "epoch": 0.40542874102854676,
      "grad_norm": 1.1609785129977837,
      "learning_rate": 6.817387028180412e-06,
      "loss": 0.6851,
      "step": 7513
    },
    {
      "epoch": 0.40548270465706115,
      "grad_norm": 0.773158934580305,
      "learning_rate": 6.816657540135733e-06,
      "loss": 0.3672,
      "step": 7514
    },
    {
      "epoch": 0.4055366682855755,
      "grad_norm": 0.9595224472036586,
      "learning_rate": 6.815928014248986e-06,
      "loss": 0.5144,
      "step": 7515
    },
    {
      "epoch": 0.4055906319140899,
      "grad_norm": 1.031984536946006,
      "learning_rate": 6.8151984505411385e-06,
      "loss": 0.4525,
      "step": 7516
    },
    {
      "epoch": 0.4056445955426043,
      "grad_norm": 1.3337563226311617,
      "learning_rate": 6.814468849033158e-06,
      "loss": 0.7379,
      "step": 7517
    },
    {
      "epoch": 0.4056985591711187,
      "grad_norm": 1.1204448930587543,
      "learning_rate": 6.813739209746014e-06,
      "loss": 0.4586,
      "step": 7518
    },
    {
      "epoch": 0.40575252279963303,
      "grad_norm": 1.081771678850783,
      "learning_rate": 6.8130095327006785e-06,
      "loss": 0.4894,
      "step": 7519
    },
    {
      "epoch": 0.40580648642814743,
      "grad_norm": 0.915494290361841,
      "learning_rate": 6.812279817918122e-06,
      "loss": 0.3839,
      "step": 7520
    },
    {
      "epoch": 0.4058604500566618,
      "grad_norm": 1.094493672970002,
      "learning_rate": 6.8115500654193175e-06,
      "loss": 0.6199,
      "step": 7521
    },
    {
      "epoch": 0.40591441368517617,
      "grad_norm": 1.0194901174355704,
      "learning_rate": 6.81082027522524e-06,
      "loss": 0.491,
      "step": 7522
    },
    {
      "epoch": 0.40596837731369056,
      "grad_norm": 1.1279999869250843,
      "learning_rate": 6.810090447356863e-06,
      "loss": 0.5424,
      "step": 7523
    },
    {
      "epoch": 0.40602234094220496,
      "grad_norm": 0.9787689440138151,
      "learning_rate": 6.809360581835161e-06,
      "loss": 0.304,
      "step": 7524
    },
    {
      "epoch": 0.40607630457071936,
      "grad_norm": 0.9254681192031328,
      "learning_rate": 6.8086306786811145e-06,
      "loss": 0.3386,
      "step": 7525
    },
    {
      "epoch": 0.4061302681992337,
      "grad_norm": 1.0553603089785617,
      "learning_rate": 6.8079007379157e-06,
      "loss": 0.4456,
      "step": 7526
    },
    {
      "epoch": 0.4061842318277481,
      "grad_norm": 1.0933990115909586,
      "learning_rate": 6.807170759559896e-06,
      "loss": 0.4727,
      "step": 7527
    },
    {
      "epoch": 0.4062381954562625,
      "grad_norm": 1.172876062336393,
      "learning_rate": 6.806440743634683e-06,
      "loss": 0.5729,
      "step": 7528
    },
    {
      "epoch": 0.40629215908477684,
      "grad_norm": 1.2760309340383158,
      "learning_rate": 6.805710690161046e-06,
      "loss": 0.684,
      "step": 7529
    },
    {
      "epoch": 0.40634612271329124,
      "grad_norm": 1.1133473582942983,
      "learning_rate": 6.804980599159963e-06,
      "loss": 0.5705,
      "step": 7530
    },
    {
      "epoch": 0.40640008634180563,
      "grad_norm": 0.991127554625731,
      "learning_rate": 6.804250470652417e-06,
      "loss": 0.4539,
      "step": 7531
    },
    {
      "epoch": 0.40645404997032003,
      "grad_norm": 0.9841668022017447,
      "learning_rate": 6.803520304659396e-06,
      "loss": 0.3919,
      "step": 7532
    },
    {
      "epoch": 0.4065080135988344,
      "grad_norm": 1.065858408731654,
      "learning_rate": 6.802790101201885e-06,
      "loss": 0.5211,
      "step": 7533
    },
    {
      "epoch": 0.40656197722734877,
      "grad_norm": 1.054092506582284,
      "learning_rate": 6.802059860300869e-06,
      "loss": 0.4725,
      "step": 7534
    },
    {
      "epoch": 0.40661594085586317,
      "grad_norm": 1.39948928729766,
      "learning_rate": 6.801329581977338e-06,
      "loss": 0.5703,
      "step": 7535
    },
    {
      "epoch": 0.4066699044843775,
      "grad_norm": 1.1653015386019334,
      "learning_rate": 6.800599266252279e-06,
      "loss": 0.6272,
      "step": 7536
    },
    {
      "epoch": 0.4067238681128919,
      "grad_norm": 1.0273994479014035,
      "learning_rate": 6.799868913146683e-06,
      "loss": 0.5823,
      "step": 7537
    },
    {
      "epoch": 0.4067778317414063,
      "grad_norm": 0.9892809482026573,
      "learning_rate": 6.799138522681542e-06,
      "loss": 0.429,
      "step": 7538
    },
    {
      "epoch": 0.40683179536992065,
      "grad_norm": 1.2576271613221424,
      "learning_rate": 6.798408094877847e-06,
      "loss": 0.7948,
      "step": 7539
    },
    {
      "epoch": 0.40688575899843504,
      "grad_norm": 1.114618605925581,
      "learning_rate": 6.797677629756591e-06,
      "loss": 0.4514,
      "step": 7540
    },
    {
      "epoch": 0.40693972262694944,
      "grad_norm": 0.9140679929472066,
      "learning_rate": 6.796947127338768e-06,
      "loss": 0.4563,
      "step": 7541
    },
    {
      "epoch": 0.40699368625546384,
      "grad_norm": 1.0625122294154816,
      "learning_rate": 6.796216587645374e-06,
      "loss": 0.5374,
      "step": 7542
    },
    {
      "epoch": 0.4070476498839782,
      "grad_norm": 1.3134810360989808,
      "learning_rate": 6.7954860106974065e-06,
      "loss": 0.6455,
      "step": 7543
    },
    {
      "epoch": 0.4071016135124926,
      "grad_norm": 1.0996546381645484,
      "learning_rate": 6.794755396515862e-06,
      "loss": 0.4961,
      "step": 7544
    },
    {
      "epoch": 0.407155577141007,
      "grad_norm": 1.3112552158266912,
      "learning_rate": 6.794024745121738e-06,
      "loss": 0.5662,
      "step": 7545
    },
    {
      "epoch": 0.4072095407695213,
      "grad_norm": 0.959326163198918,
      "learning_rate": 6.7932940565360375e-06,
      "loss": 0.3728,
      "step": 7546
    },
    {
      "epoch": 0.4072635043980357,
      "grad_norm": 0.9186566530650735,
      "learning_rate": 6.792563330779757e-06,
      "loss": 0.5392,
      "step": 7547
    },
    {
      "epoch": 0.4073174680265501,
      "grad_norm": 0.992646347534432,
      "learning_rate": 6.791832567873902e-06,
      "loss": 0.4142,
      "step": 7548
    },
    {
      "epoch": 0.4073714316550645,
      "grad_norm": 0.8873433798106148,
      "learning_rate": 6.791101767839472e-06,
      "loss": 0.4271,
      "step": 7549
    },
    {
      "epoch": 0.40742539528357885,
      "grad_norm": 0.9594934973262861,
      "learning_rate": 6.7903709306974745e-06,
      "loss": 0.515,
      "step": 7550
    },
    {
      "epoch": 0.40747935891209325,
      "grad_norm": 1.043207877365533,
      "learning_rate": 6.789640056468911e-06,
      "loss": 0.3814,
      "step": 7551
    },
    {
      "epoch": 0.40753332254060765,
      "grad_norm": 0.922967190895919,
      "learning_rate": 6.7889091451747894e-06,
      "loss": 0.4577,
      "step": 7552
    },
    {
      "epoch": 0.407587286169122,
      "grad_norm": 0.8386069079612846,
      "learning_rate": 6.788178196836118e-06,
      "loss": 0.4068,
      "step": 7553
    },
    {
      "epoch": 0.4076412497976364,
      "grad_norm": 1.1453995378585662,
      "learning_rate": 6.787447211473902e-06,
      "loss": 0.4606,
      "step": 7554
    },
    {
      "epoch": 0.4076952134261508,
      "grad_norm": 1.0152585487905392,
      "learning_rate": 6.786716189109153e-06,
      "loss": 0.4023,
      "step": 7555
    },
    {
      "epoch": 0.4077491770546652,
      "grad_norm": 1.1243032267600435,
      "learning_rate": 6.785985129762881e-06,
      "loss": 0.5693,
      "step": 7556
    },
    {
      "epoch": 0.4078031406831795,
      "grad_norm": 0.6307953743109266,
      "learning_rate": 6.785254033456096e-06,
      "loss": 0.2699,
      "step": 7557
    },
    {
      "epoch": 0.4078571043116939,
      "grad_norm": 1.067236559805759,
      "learning_rate": 6.784522900209811e-06,
      "loss": 0.4552,
      "step": 7558
    },
    {
      "epoch": 0.4079110679402083,
      "grad_norm": 1.086559953574209,
      "learning_rate": 6.7837917300450414e-06,
      "loss": 0.3796,
      "step": 7559
    },
    {
      "epoch": 0.40796503156872266,
      "grad_norm": 0.9973013518950935,
      "learning_rate": 6.7830605229828005e-06,
      "loss": 0.4176,
      "step": 7560
    },
    {
      "epoch": 0.40801899519723706,
      "grad_norm": 1.1122066223938112,
      "learning_rate": 6.782329279044103e-06,
      "loss": 0.5679,
      "step": 7561
    },
    {
      "epoch": 0.40807295882575145,
      "grad_norm": 0.840760887229507,
      "learning_rate": 6.7815979982499655e-06,
      "loss": 0.3165,
      "step": 7562
    },
    {
      "epoch": 0.40812692245426585,
      "grad_norm": 0.9728957970376029,
      "learning_rate": 6.780866680621407e-06,
      "loss": 0.4235,
      "step": 7563
    },
    {
      "epoch": 0.4081808860827802,
      "grad_norm": 1.187367273392454,
      "learning_rate": 6.780135326179447e-06,
      "loss": 0.5339,
      "step": 7564
    },
    {
      "epoch": 0.4082348497112946,
      "grad_norm": 0.9677355542074414,
      "learning_rate": 6.779403934945103e-06,
      "loss": 0.4031,
      "step": 7565
    },
    {
      "epoch": 0.408288813339809,
      "grad_norm": 1.0238675966647772,
      "learning_rate": 6.778672506939398e-06,
      "loss": 0.5097,
      "step": 7566
    },
    {
      "epoch": 0.40834277696832333,
      "grad_norm": 0.912666045631566,
      "learning_rate": 6.7779410421833525e-06,
      "loss": 0.4783,
      "step": 7567
    },
    {
      "epoch": 0.4083967405968377,
      "grad_norm": 0.976084008734294,
      "learning_rate": 6.77720954069799e-06,
      "loss": 0.444,
      "step": 7568
    },
    {
      "epoch": 0.4084507042253521,
      "grad_norm": 0.9435906068948745,
      "learning_rate": 6.776478002504335e-06,
      "loss": 0.396,
      "step": 7569
    },
    {
      "epoch": 0.4085046678538665,
      "grad_norm": 1.005116291594624,
      "learning_rate": 6.775746427623413e-06,
      "loss": 0.4793,
      "step": 7570
    },
    {
      "epoch": 0.40855863148238086,
      "grad_norm": 0.8904701157419485,
      "learning_rate": 6.77501481607625e-06,
      "loss": 0.4546,
      "step": 7571
    },
    {
      "epoch": 0.40861259511089526,
      "grad_norm": 1.168635157384942,
      "learning_rate": 6.774283167883871e-06,
      "loss": 0.5257,
      "step": 7572
    },
    {
      "epoch": 0.40866655873940966,
      "grad_norm": 0.9131103124197575,
      "learning_rate": 6.773551483067308e-06,
      "loss": 0.4541,
      "step": 7573
    },
    {
      "epoch": 0.408720522367924,
      "grad_norm": 1.0470166082191037,
      "learning_rate": 6.772819761647588e-06,
      "loss": 0.425,
      "step": 7574
    },
    {
      "epoch": 0.4087744859964384,
      "grad_norm": 1.1555396530150739,
      "learning_rate": 6.772088003645741e-06,
      "loss": 0.5683,
      "step": 7575
    },
    {
      "epoch": 0.4088284496249528,
      "grad_norm": 1.0130522384780438,
      "learning_rate": 6.771356209082801e-06,
      "loss": 0.507,
      "step": 7576
    },
    {
      "epoch": 0.40888241325346714,
      "grad_norm": 1.1290576311986695,
      "learning_rate": 6.770624377979798e-06,
      "loss": 0.5615,
      "step": 7577
    },
    {
      "epoch": 0.40893637688198153,
      "grad_norm": 1.1804053488068822,
      "learning_rate": 6.7698925103577676e-06,
      "loss": 0.531,
      "step": 7578
    },
    {
      "epoch": 0.40899034051049593,
      "grad_norm": 1.1233449814464416,
      "learning_rate": 6.769160606237741e-06,
      "loss": 0.6211,
      "step": 7579
    },
    {
      "epoch": 0.40904430413901033,
      "grad_norm": 0.8064305283203516,
      "learning_rate": 6.768428665640758e-06,
      "loss": 0.3043,
      "step": 7580
    },
    {
      "epoch": 0.40909826776752467,
      "grad_norm": 0.9317865720600396,
      "learning_rate": 6.767696688587854e-06,
      "loss": 0.6461,
      "step": 7581
    },
    {
      "epoch": 0.40915223139603907,
      "grad_norm": 1.0756165890056868,
      "learning_rate": 6.766964675100065e-06,
      "loss": 0.4348,
      "step": 7582
    },
    {
      "epoch": 0.40920619502455347,
      "grad_norm": 0.9861352697882854,
      "learning_rate": 6.766232625198431e-06,
      "loss": 0.4023,
      "step": 7583
    },
    {
      "epoch": 0.4092601586530678,
      "grad_norm": 0.9577546817200113,
      "learning_rate": 6.765500538903993e-06,
      "loss": 0.3159,
      "step": 7584
    },
    {
      "epoch": 0.4093141222815822,
      "grad_norm": 0.9234481047849848,
      "learning_rate": 6.76476841623779e-06,
      "loss": 0.5196,
      "step": 7585
    },
    {
      "epoch": 0.4093680859100966,
      "grad_norm": 1.0339590924349675,
      "learning_rate": 6.764036257220865e-06,
      "loss": 0.4502,
      "step": 7586
    },
    {
      "epoch": 0.409422049538611,
      "grad_norm": 0.9744250438024435,
      "learning_rate": 6.76330406187426e-06,
      "loss": 0.4142,
      "step": 7587
    },
    {
      "epoch": 0.40947601316712534,
      "grad_norm": 0.9067440202221327,
      "learning_rate": 6.7625718302190215e-06,
      "loss": 0.423,
      "step": 7588
    },
    {
      "epoch": 0.40952997679563974,
      "grad_norm": 1.0331148539466044,
      "learning_rate": 6.7618395622761915e-06,
      "loss": 0.4945,
      "step": 7589
    },
    {
      "epoch": 0.40958394042415414,
      "grad_norm": 0.9843819692798136,
      "learning_rate": 6.761107258066818e-06,
      "loss": 0.4529,
      "step": 7590
    },
    {
      "epoch": 0.4096379040526685,
      "grad_norm": 0.9131506729743512,
      "learning_rate": 6.760374917611948e-06,
      "loss": 0.4373,
      "step": 7591
    },
    {
      "epoch": 0.4096918676811829,
      "grad_norm": 1.033680759540545,
      "learning_rate": 6.759642540932628e-06,
      "loss": 0.404,
      "step": 7592
    },
    {
      "epoch": 0.4097458313096973,
      "grad_norm": 1.1867959390970622,
      "learning_rate": 6.758910128049909e-06,
      "loss": 0.6558,
      "step": 7593
    },
    {
      "epoch": 0.40979979493821167,
      "grad_norm": 0.8973494477708069,
      "learning_rate": 6.758177678984842e-06,
      "loss": 0.3911,
      "step": 7594
    },
    {
      "epoch": 0.409853758566726,
      "grad_norm": 1.2041954804215937,
      "learning_rate": 6.757445193758476e-06,
      "loss": 0.5895,
      "step": 7595
    },
    {
      "epoch": 0.4099077221952404,
      "grad_norm": 1.0594497875463043,
      "learning_rate": 6.756712672391865e-06,
      "loss": 0.5099,
      "step": 7596
    },
    {
      "epoch": 0.4099616858237548,
      "grad_norm": 1.2560997138109604,
      "learning_rate": 6.7559801149060625e-06,
      "loss": 0.6206,
      "step": 7597
    },
    {
      "epoch": 0.41001564945226915,
      "grad_norm": 0.9994304033489262,
      "learning_rate": 6.755247521322122e-06,
      "loss": 0.546,
      "step": 7598
    },
    {
      "epoch": 0.41006961308078355,
      "grad_norm": 1.1646336927008052,
      "learning_rate": 6.754514891661099e-06,
      "loss": 0.5776,
      "step": 7599
    },
    {
      "epoch": 0.41012357670929794,
      "grad_norm": 1.1702044695655607,
      "learning_rate": 6.753782225944051e-06,
      "loss": 0.6134,
      "step": 7600
    },
    {
      "epoch": 0.41017754033781234,
      "grad_norm": 0.8865512008750267,
      "learning_rate": 6.7530495241920334e-06,
      "loss": 0.3912,
      "step": 7601
    },
    {
      "epoch": 0.4102315039663267,
      "grad_norm": 1.178952043974545,
      "learning_rate": 6.7523167864261095e-06,
      "loss": 0.5268,
      "step": 7602
    },
    {
      "epoch": 0.4102854675948411,
      "grad_norm": 1.1564745680355186,
      "learning_rate": 6.751584012667334e-06,
      "loss": 0.5357,
      "step": 7603
    },
    {
      "epoch": 0.4103394312233555,
      "grad_norm": 1.008678514949802,
      "learning_rate": 6.750851202936769e-06,
      "loss": 0.443,
      "step": 7604
    },
    {
      "epoch": 0.4103933948518698,
      "grad_norm": 1.073093196636254,
      "learning_rate": 6.750118357255477e-06,
      "loss": 0.5584,
      "step": 7605
    },
    {
      "epoch": 0.4104473584803842,
      "grad_norm": 1.2173997063428343,
      "learning_rate": 6.74938547564452e-06,
      "loss": 0.4574,
      "step": 7606
    },
    {
      "epoch": 0.4105013221088986,
      "grad_norm": 1.1077892642005789,
      "learning_rate": 6.7486525581249614e-06,
      "loss": 0.548,
      "step": 7607
    },
    {
      "epoch": 0.41055528573741296,
      "grad_norm": 1.2658895522079476,
      "learning_rate": 6.7479196047178695e-06,
      "loss": 0.6846,
      "step": 7608
    },
    {
      "epoch": 0.41060924936592735,
      "grad_norm": 1.246820772395013,
      "learning_rate": 6.747186615444304e-06,
      "loss": 0.5081,
      "step": 7609
    },
    {
      "epoch": 0.41066321299444175,
      "grad_norm": 1.096987552047168,
      "learning_rate": 6.746453590325336e-06,
      "loss": 0.5769,
      "step": 7610
    },
    {
      "epoch": 0.41071717662295615,
      "grad_norm": 1.0653859903713836,
      "learning_rate": 6.745720529382033e-06,
      "loss": 0.5094,
      "step": 7611
    },
    {
      "epoch": 0.4107711402514705,
      "grad_norm": 1.0272659153562347,
      "learning_rate": 6.744987432635463e-06,
      "loss": 0.5134,
      "step": 7612
    },
    {
      "epoch": 0.4108251038799849,
      "grad_norm": 1.1667007636446742,
      "learning_rate": 6.7442543001066954e-06,
      "loss": 0.4401,
      "step": 7613
    },
    {
      "epoch": 0.4108790675084993,
      "grad_norm": 0.971640466841577,
      "learning_rate": 6.743521131816801e-06,
      "loss": 0.3558,
      "step": 7614
    },
    {
      "epoch": 0.4109330311370136,
      "grad_norm": 1.1786801765716473,
      "learning_rate": 6.742787927786857e-06,
      "loss": 0.6721,
      "step": 7615
    },
    {
      "epoch": 0.410986994765528,
      "grad_norm": 1.103240351659209,
      "learning_rate": 6.742054688037929e-06,
      "loss": 0.4973,
      "step": 7616
    },
    {
      "epoch": 0.4110409583940424,
      "grad_norm": 1.122231384761107,
      "learning_rate": 6.741321412591094e-06,
      "loss": 0.6702,
      "step": 7617
    },
    {
      "epoch": 0.4110949220225568,
      "grad_norm": 1.1076903084176246,
      "learning_rate": 6.740588101467428e-06,
      "loss": 0.5632,
      "step": 7618
    },
    {
      "epoch": 0.41114888565107116,
      "grad_norm": 1.0619176271266038,
      "learning_rate": 6.739854754688006e-06,
      "loss": 0.567,
      "step": 7619
    },
    {
      "epoch": 0.41120284927958556,
      "grad_norm": 0.9297136655990516,
      "learning_rate": 6.7391213722739055e-06,
      "loss": 0.4369,
      "step": 7620
    },
    {
      "epoch": 0.41125681290809996,
      "grad_norm": 1.0258217704285317,
      "learning_rate": 6.738387954246206e-06,
      "loss": 0.6,
      "step": 7621
    },
    {
      "epoch": 0.4113107765366143,
      "grad_norm": 0.9677333729556463,
      "learning_rate": 6.737654500625985e-06,
      "loss": 0.5075,
      "step": 7622
    },
    {
      "epoch": 0.4113647401651287,
      "grad_norm": 1.0784996383055916,
      "learning_rate": 6.736921011434323e-06,
      "loss": 0.3545,
      "step": 7623
    },
    {
      "epoch": 0.4114187037936431,
      "grad_norm": 1.103381403928868,
      "learning_rate": 6.736187486692302e-06,
      "loss": 0.6194,
      "step": 7624
    },
    {
      "epoch": 0.4114726674221575,
      "grad_norm": 1.1527952272591555,
      "learning_rate": 6.735453926421001e-06,
      "loss": 0.5563,
      "step": 7625
    },
    {
      "epoch": 0.41152663105067183,
      "grad_norm": 0.9284164357452473,
      "learning_rate": 6.734720330641509e-06,
      "loss": 0.3699,
      "step": 7626
    },
    {
      "epoch": 0.41158059467918623,
      "grad_norm": 1.04408371727465,
      "learning_rate": 6.733986699374906e-06,
      "loss": 0.3747,
      "step": 7627
    },
    {
      "epoch": 0.4116345583077006,
      "grad_norm": 0.9516816500079481,
      "learning_rate": 6.733253032642277e-06,
      "loss": 0.3226,
      "step": 7628
    },
    {
      "epoch": 0.41168852193621497,
      "grad_norm": 1.1020687602089865,
      "learning_rate": 6.732519330464711e-06,
      "loss": 0.513,
      "step": 7629
    },
    {
      "epoch": 0.41174248556472937,
      "grad_norm": 1.1504802521293893,
      "learning_rate": 6.731785592863293e-06,
      "loss": 0.528,
      "step": 7630
    },
    {
      "epoch": 0.41179644919324376,
      "grad_norm": 0.9661424828438891,
      "learning_rate": 6.731051819859115e-06,
      "loss": 0.4744,
      "step": 7631
    },
    {
      "epoch": 0.41185041282175816,
      "grad_norm": 1.0654390011746704,
      "learning_rate": 6.730318011473261e-06,
      "loss": 0.5528,
      "step": 7632
    },
    {
      "epoch": 0.4119043764502725,
      "grad_norm": 0.7943126007116983,
      "learning_rate": 6.729584167726826e-06,
      "loss": 0.4139,
      "step": 7633
    },
    {
      "epoch": 0.4119583400787869,
      "grad_norm": 0.9154252549707663,
      "learning_rate": 6.728850288640899e-06,
      "loss": 0.4672,
      "step": 7634
    },
    {
      "epoch": 0.4120123037073013,
      "grad_norm": 0.9923976890999333,
      "learning_rate": 6.728116374236573e-06,
      "loss": 0.4917,
      "step": 7635
    },
    {
      "epoch": 0.41206626733581564,
      "grad_norm": 1.0353572604473646,
      "learning_rate": 6.727382424534942e-06,
      "loss": 0.572,
      "step": 7636
    },
    {
      "epoch": 0.41212023096433004,
      "grad_norm": 1.027880488652444,
      "learning_rate": 6.726648439557099e-06,
      "loss": 0.3535,
      "step": 7637
    },
    {
      "epoch": 0.41217419459284443,
      "grad_norm": 1.1868125580913245,
      "learning_rate": 6.725914419324141e-06,
      "loss": 0.5437,
      "step": 7638
    },
    {
      "epoch": 0.41222815822135883,
      "grad_norm": 0.8630820527547929,
      "learning_rate": 6.725180363857162e-06,
      "loss": 0.3486,
      "step": 7639
    },
    {
      "epoch": 0.4122821218498732,
      "grad_norm": 1.0683834325749735,
      "learning_rate": 6.7244462731772645e-06,
      "loss": 0.4609,
      "step": 7640
    },
    {
      "epoch": 0.41233608547838757,
      "grad_norm": 1.234195366409918,
      "learning_rate": 6.723712147305542e-06,
      "loss": 0.596,
      "step": 7641
    },
    {
      "epoch": 0.41239004910690197,
      "grad_norm": 1.0066001917032066,
      "learning_rate": 6.722977986263096e-06,
      "loss": 0.3547,
      "step": 7642
    },
    {
      "epoch": 0.4124440127354163,
      "grad_norm": 1.0046143968257766,
      "learning_rate": 6.722243790071027e-06,
      "loss": 0.5508,
      "step": 7643
    },
    {
      "epoch": 0.4124979763639307,
      "grad_norm": 0.8886237134452628,
      "learning_rate": 6.721509558750438e-06,
      "loss": 0.4647,
      "step": 7644
    },
    {
      "epoch": 0.4125519399924451,
      "grad_norm": 1.1018432310557105,
      "learning_rate": 6.720775292322429e-06,
      "loss": 0.6361,
      "step": 7645
    },
    {
      "epoch": 0.41260590362095945,
      "grad_norm": 0.9665163822249392,
      "learning_rate": 6.720040990808104e-06,
      "loss": 0.5047,
      "step": 7646
    },
    {
      "epoch": 0.41265986724947384,
      "grad_norm": 0.9567722169420492,
      "learning_rate": 6.719306654228569e-06,
      "loss": 0.4024,
      "step": 7647
    },
    {
      "epoch": 0.41271383087798824,
      "grad_norm": 0.9222497535105044,
      "learning_rate": 6.718572282604927e-06,
      "loss": 0.4185,
      "step": 7648
    },
    {
      "epoch": 0.41276779450650264,
      "grad_norm": 1.1656640278794532,
      "learning_rate": 6.717837875958287e-06,
      "loss": 0.4187,
      "step": 7649
    },
    {
      "epoch": 0.412821758135017,
      "grad_norm": 1.1641689629686858,
      "learning_rate": 6.717103434309757e-06,
      "loss": 0.483,
      "step": 7650
    },
    {
      "epoch": 0.4128757217635314,
      "grad_norm": 0.7923833910831188,
      "learning_rate": 6.716368957680445e-06,
      "loss": 0.3003,
      "step": 7651
    },
    {
      "epoch": 0.4129296853920458,
      "grad_norm": 1.1185727426654974,
      "learning_rate": 6.7156344460914604e-06,
      "loss": 0.4648,
      "step": 7652
    },
    {
      "epoch": 0.4129836490205601,
      "grad_norm": 1.0204013316384901,
      "learning_rate": 6.714899899563912e-06,
      "loss": 0.4921,
      "step": 7653
    },
    {
      "epoch": 0.4130376126490745,
      "grad_norm": 0.9869879680395924,
      "learning_rate": 6.714165318118912e-06,
      "loss": 0.3916,
      "step": 7654
    },
    {
      "epoch": 0.4130915762775889,
      "grad_norm": 1.0032480439215195,
      "learning_rate": 6.713430701777576e-06,
      "loss": 0.5511,
      "step": 7655
    },
    {
      "epoch": 0.4131455399061033,
      "grad_norm": 1.1799242645062122,
      "learning_rate": 6.712696050561014e-06,
      "loss": 0.5597,
      "step": 7656
    },
    {
      "epoch": 0.41319950353461765,
      "grad_norm": 0.981315717594818,
      "learning_rate": 6.7119613644903435e-06,
      "loss": 0.5777,
      "step": 7657
    },
    {
      "epoch": 0.41325346716313205,
      "grad_norm": 0.8924887310239625,
      "learning_rate": 6.711226643586679e-06,
      "loss": 0.3983,
      "step": 7658
    },
    {
      "epoch": 0.41330743079164645,
      "grad_norm": 1.0721082397556114,
      "learning_rate": 6.710491887871135e-06,
      "loss": 0.5701,
      "step": 7659
    },
    {
      "epoch": 0.4133613944201608,
      "grad_norm": 0.8629083679093378,
      "learning_rate": 6.7097570973648326e-06,
      "loss": 0.4454,
      "step": 7660
    },
    {
      "epoch": 0.4134153580486752,
      "grad_norm": 0.9365858958194917,
      "learning_rate": 6.7090222720888884e-06,
      "loss": 0.485,
      "step": 7661
    },
    {
      "epoch": 0.4134693216771896,
      "grad_norm": 0.9985219025884867,
      "learning_rate": 6.708287412064421e-06,
      "loss": 0.3911,
      "step": 7662
    },
    {
      "epoch": 0.413523285305704,
      "grad_norm": 0.9419055835739972,
      "learning_rate": 6.707552517312555e-06,
      "loss": 0.4184,
      "step": 7663
    },
    {
      "epoch": 0.4135772489342183,
      "grad_norm": 1.0049469833636604,
      "learning_rate": 6.706817587854408e-06,
      "loss": 0.4152,
      "step": 7664
    },
    {
      "epoch": 0.4136312125627327,
      "grad_norm": 0.8778789838192276,
      "learning_rate": 6.7060826237111056e-06,
      "loss": 0.3587,
      "step": 7665
    },
    {
      "epoch": 0.4136851761912471,
      "grad_norm": 1.3522987053697317,
      "learning_rate": 6.705347624903767e-06,
      "loss": 0.63,
      "step": 7666
    },
    {
      "epoch": 0.41373913981976146,
      "grad_norm": 0.892107317966253,
      "learning_rate": 6.704612591453523e-06,
      "loss": 0.397,
      "step": 7667
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.8372897259500505,
      "learning_rate": 6.703877523381495e-06,
      "loss": 0.3236,
      "step": 7668
    },
    {
      "epoch": 0.41384706707679025,
      "grad_norm": 1.1426882829906473,
      "learning_rate": 6.70314242070881e-06,
      "loss": 0.6227,
      "step": 7669
    },
    {
      "epoch": 0.41390103070530465,
      "grad_norm": 1.1475143961058738,
      "learning_rate": 6.7024072834565955e-06,
      "loss": 0.6112,
      "step": 7670
    },
    {
      "epoch": 0.413954994333819,
      "grad_norm": 0.9217568536702735,
      "learning_rate": 6.701672111645982e-06,
      "loss": 0.414,
      "step": 7671
    },
    {
      "epoch": 0.4140089579623334,
      "grad_norm": 1.1276839494115682,
      "learning_rate": 6.700936905298096e-06,
      "loss": 0.4869,
      "step": 7672
    },
    {
      "epoch": 0.4140629215908478,
      "grad_norm": 0.9951110757381574,
      "learning_rate": 6.700201664434072e-06,
      "loss": 0.4377,
      "step": 7673
    },
    {
      "epoch": 0.41411688521936213,
      "grad_norm": 0.862866744033978,
      "learning_rate": 6.6994663890750386e-06,
      "loss": 0.4175,
      "step": 7674
    },
    {
      "epoch": 0.4141708488478765,
      "grad_norm": 0.865574237957905,
      "learning_rate": 6.69873107924213e-06,
      "loss": 0.3683,
      "step": 7675
    },
    {
      "epoch": 0.4142248124763909,
      "grad_norm": 0.720673719841367,
      "learning_rate": 6.697995734956477e-06,
      "loss": 0.2366,
      "step": 7676
    },
    {
      "epoch": 0.41427877610490527,
      "grad_norm": 1.0836532266259111,
      "learning_rate": 6.697260356239218e-06,
      "loss": 0.5442,
      "step": 7677
    },
    {
      "epoch": 0.41433273973341966,
      "grad_norm": 1.0823555761561603,
      "learning_rate": 6.696524943111485e-06,
      "loss": 0.4443,
      "step": 7678
    },
    {
      "epoch": 0.41438670336193406,
      "grad_norm": 1.0562360158260566,
      "learning_rate": 6.695789495594417e-06,
      "loss": 0.4682,
      "step": 7679
    },
    {
      "epoch": 0.41444066699044846,
      "grad_norm": 0.9224165312237081,
      "learning_rate": 6.695054013709152e-06,
      "loss": 0.3597,
      "step": 7680
    },
    {
      "epoch": 0.4144946306189628,
      "grad_norm": 1.226348400478821,
      "learning_rate": 6.694318497476826e-06,
      "loss": 0.5449,
      "step": 7681
    },
    {
      "epoch": 0.4145485942474772,
      "grad_norm": 0.923017727645661,
      "learning_rate": 6.69358294691858e-06,
      "loss": 0.5326,
      "step": 7682
    },
    {
      "epoch": 0.4146025578759916,
      "grad_norm": 0.9118756884023664,
      "learning_rate": 6.692847362055552e-06,
      "loss": 0.3969,
      "step": 7683
    },
    {
      "epoch": 0.41465652150450594,
      "grad_norm": 1.1353698458089057,
      "learning_rate": 6.692111742908888e-06,
      "loss": 0.4939,
      "step": 7684
    },
    {
      "epoch": 0.41471048513302033,
      "grad_norm": 1.1423593138128627,
      "learning_rate": 6.691376089499727e-06,
      "loss": 0.5041,
      "step": 7685
    },
    {
      "epoch": 0.41476444876153473,
      "grad_norm": 0.9874111637696947,
      "learning_rate": 6.6906404018492135e-06,
      "loss": 0.5262,
      "step": 7686
    },
    {
      "epoch": 0.41481841239004913,
      "grad_norm": 1.0111341929282947,
      "learning_rate": 6.689904679978491e-06,
      "loss": 0.4222,
      "step": 7687
    },
    {
      "epoch": 0.41487237601856347,
      "grad_norm": 0.838523897787651,
      "learning_rate": 6.689168923908707e-06,
      "loss": 0.359,
      "step": 7688
    },
    {
      "epoch": 0.41492633964707787,
      "grad_norm": 1.1918205631180137,
      "learning_rate": 6.688433133661007e-06,
      "loss": 0.4779,
      "step": 7689
    },
    {
      "epoch": 0.41498030327559227,
      "grad_norm": 0.9968139343557337,
      "learning_rate": 6.687697309256536e-06,
      "loss": 0.434,
      "step": 7690
    },
    {
      "epoch": 0.4150342669041066,
      "grad_norm": 1.0881959282736553,
      "learning_rate": 6.686961450716446e-06,
      "loss": 0.5286,
      "step": 7691
    },
    {
      "epoch": 0.415088230532621,
      "grad_norm": 0.9653996081541671,
      "learning_rate": 6.686225558061883e-06,
      "loss": 0.39,
      "step": 7692
    },
    {
      "epoch": 0.4151421941611354,
      "grad_norm": 0.9859411725158641,
      "learning_rate": 6.685489631313999e-06,
      "loss": 0.4415,
      "step": 7693
    },
    {
      "epoch": 0.4151961577896498,
      "grad_norm": 0.7722805353414243,
      "learning_rate": 6.684753670493945e-06,
      "loss": 0.3524,
      "step": 7694
    },
    {
      "epoch": 0.41525012141816414,
      "grad_norm": 0.8425649568227274,
      "learning_rate": 6.684017675622875e-06,
      "loss": 0.3295,
      "step": 7695
    },
    {
      "epoch": 0.41530408504667854,
      "grad_norm": 1.0692116092359243,
      "learning_rate": 6.683281646721939e-06,
      "loss": 0.589,
      "step": 7696
    },
    {
      "epoch": 0.41535804867519294,
      "grad_norm": 0.8988281527593945,
      "learning_rate": 6.682545583812294e-06,
      "loss": 0.4246,
      "step": 7697
    },
    {
      "epoch": 0.4154120123037073,
      "grad_norm": 1.2523741898570613,
      "learning_rate": 6.681809486915093e-06,
      "loss": 0.5495,
      "step": 7698
    },
    {
      "epoch": 0.4154659759322217,
      "grad_norm": 1.1238742463383296,
      "learning_rate": 6.681073356051494e-06,
      "loss": 0.4113,
      "step": 7699
    },
    {
      "epoch": 0.4155199395607361,
      "grad_norm": 0.9417358033363995,
      "learning_rate": 6.680337191242653e-06,
      "loss": 0.5045,
      "step": 7700
    },
    {
      "epoch": 0.41557390318925047,
      "grad_norm": 0.9573415323956321,
      "learning_rate": 6.67960099250973e-06,
      "loss": 0.4619,
      "step": 7701
    },
    {
      "epoch": 0.4156278668177648,
      "grad_norm": 1.074991429956297,
      "learning_rate": 6.678864759873881e-06,
      "loss": 0.6141,
      "step": 7702
    },
    {
      "epoch": 0.4156818304462792,
      "grad_norm": 0.9922477652108836,
      "learning_rate": 6.6781284933562675e-06,
      "loss": 0.3896,
      "step": 7703
    },
    {
      "epoch": 0.4157357940747936,
      "grad_norm": 0.9340324758366886,
      "learning_rate": 6.677392192978051e-06,
      "loss": 0.4116,
      "step": 7704
    },
    {
      "epoch": 0.41578975770330795,
      "grad_norm": 1.0372221152312187,
      "learning_rate": 6.676655858760394e-06,
      "loss": 0.4833,
      "step": 7705
    },
    {
      "epoch": 0.41584372133182235,
      "grad_norm": 0.852785633166194,
      "learning_rate": 6.675919490724456e-06,
      "loss": 0.4208,
      "step": 7706
    },
    {
      "epoch": 0.41589768496033674,
      "grad_norm": 1.0002412466781019,
      "learning_rate": 6.675183088891406e-06,
      "loss": 0.4517,
      "step": 7707
    },
    {
      "epoch": 0.41595164858885114,
      "grad_norm": 1.2568111893381517,
      "learning_rate": 6.674446653282407e-06,
      "loss": 0.7358,
      "step": 7708
    },
    {
      "epoch": 0.4160056122173655,
      "grad_norm": 1.0051378579336638,
      "learning_rate": 6.6737101839186245e-06,
      "loss": 0.4831,
      "step": 7709
    },
    {
      "epoch": 0.4160595758458799,
      "grad_norm": 1.0475563322612476,
      "learning_rate": 6.672973680821224e-06,
      "loss": 0.4927,
      "step": 7710
    },
    {
      "epoch": 0.4161135394743943,
      "grad_norm": 0.9512274702627562,
      "learning_rate": 6.672237144011374e-06,
      "loss": 0.4255,
      "step": 7711
    },
    {
      "epoch": 0.4161675031029086,
      "grad_norm": 0.9418195881251247,
      "learning_rate": 6.6715005735102466e-06,
      "loss": 0.4199,
      "step": 7712
    },
    {
      "epoch": 0.416221466731423,
      "grad_norm": 0.926315245279255,
      "learning_rate": 6.670763969339008e-06,
      "loss": 0.5246,
      "step": 7713
    },
    {
      "epoch": 0.4162754303599374,
      "grad_norm": 0.8153015445235215,
      "learning_rate": 6.67002733151883e-06,
      "loss": 0.3984,
      "step": 7714
    },
    {
      "epoch": 0.41632939398845176,
      "grad_norm": 0.9711997703379833,
      "learning_rate": 6.669290660070884e-06,
      "loss": 0.4662,
      "step": 7715
    },
    {
      "epoch": 0.41638335761696615,
      "grad_norm": 1.1483770499637191,
      "learning_rate": 6.668553955016343e-06,
      "loss": 0.5364,
      "step": 7716
    },
    {
      "epoch": 0.41643732124548055,
      "grad_norm": 0.7748353607065362,
      "learning_rate": 6.66781721637638e-06,
      "loss": 0.2755,
      "step": 7717
    },
    {
      "epoch": 0.41649128487399495,
      "grad_norm": 1.039990759613914,
      "learning_rate": 6.66708044417217e-06,
      "loss": 0.4846,
      "step": 7718
    },
    {
      "epoch": 0.4165452485025093,
      "grad_norm": 0.9964149495377834,
      "learning_rate": 6.666343638424891e-06,
      "loss": 0.4021,
      "step": 7719
    },
    {
      "epoch": 0.4165992121310237,
      "grad_norm": 1.1052553345654368,
      "learning_rate": 6.665606799155715e-06,
      "loss": 0.5379,
      "step": 7720
    },
    {
      "epoch": 0.4166531757595381,
      "grad_norm": 1.0363708610845994,
      "learning_rate": 6.664869926385823e-06,
      "loss": 0.4762,
      "step": 7721
    },
    {
      "epoch": 0.4167071393880524,
      "grad_norm": 1.060145899787564,
      "learning_rate": 6.664133020136391e-06,
      "loss": 0.4845,
      "step": 7722
    },
    {
      "epoch": 0.4167611030165668,
      "grad_norm": 0.8813489712731463,
      "learning_rate": 6.663396080428601e-06,
      "loss": 0.496,
      "step": 7723
    },
    {
      "epoch": 0.4168150666450812,
      "grad_norm": 1.0281069388119133,
      "learning_rate": 6.662659107283631e-06,
      "loss": 0.5066,
      "step": 7724
    },
    {
      "epoch": 0.4168690302735956,
      "grad_norm": 0.8442008940412116,
      "learning_rate": 6.661922100722665e-06,
      "loss": 0.3357,
      "step": 7725
    },
    {
      "epoch": 0.41692299390210996,
      "grad_norm": 1.0519496665098675,
      "learning_rate": 6.661185060766884e-06,
      "loss": 0.5384,
      "step": 7726
    },
    {
      "epoch": 0.41697695753062436,
      "grad_norm": 0.9794761132221237,
      "learning_rate": 6.660447987437469e-06,
      "loss": 0.5539,
      "step": 7727
    },
    {
      "epoch": 0.41703092115913876,
      "grad_norm": 0.8945343286169914,
      "learning_rate": 6.659710880755608e-06,
      "loss": 0.4365,
      "step": 7728
    },
    {
      "epoch": 0.4170848847876531,
      "grad_norm": 0.8514127289364112,
      "learning_rate": 6.658973740742484e-06,
      "loss": 0.4296,
      "step": 7729
    },
    {
      "epoch": 0.4171388484161675,
      "grad_norm": 0.9341048714044063,
      "learning_rate": 6.658236567419282e-06,
      "loss": 0.3508,
      "step": 7730
    },
    {
      "epoch": 0.4171928120446819,
      "grad_norm": 0.9119270917524162,
      "learning_rate": 6.657499360807192e-06,
      "loss": 0.4233,
      "step": 7731
    },
    {
      "epoch": 0.4172467756731963,
      "grad_norm": 1.2114645856764281,
      "learning_rate": 6.656762120927402e-06,
      "loss": 0.5369,
      "step": 7732
    },
    {
      "epoch": 0.41730073930171063,
      "grad_norm": 1.1093220251875684,
      "learning_rate": 6.6560248478010995e-06,
      "loss": 0.5401,
      "step": 7733
    },
    {
      "epoch": 0.41735470293022503,
      "grad_norm": 1.0821100436005389,
      "learning_rate": 6.655287541449475e-06,
      "loss": 0.6453,
      "step": 7734
    },
    {
      "epoch": 0.4174086665587394,
      "grad_norm": 1.053260896540987,
      "learning_rate": 6.654550201893718e-06,
      "loss": 0.4305,
      "step": 7735
    },
    {
      "epoch": 0.41746263018725377,
      "grad_norm": 1.0245598962400617,
      "learning_rate": 6.653812829155022e-06,
      "loss": 0.5153,
      "step": 7736
    },
    {
      "epoch": 0.41751659381576817,
      "grad_norm": 0.9888540950043225,
      "learning_rate": 6.65307542325458e-06,
      "loss": 0.4652,
      "step": 7737
    },
    {
      "epoch": 0.41757055744428256,
      "grad_norm": 1.057157088648307,
      "learning_rate": 6.652337984213584e-06,
      "loss": 0.4525,
      "step": 7738
    },
    {
      "epoch": 0.41762452107279696,
      "grad_norm": 1.2039896405930992,
      "learning_rate": 6.651600512053231e-06,
      "loss": 0.509,
      "step": 7739
    },
    {
      "epoch": 0.4176784847013113,
      "grad_norm": 1.0975000453399402,
      "learning_rate": 6.650863006794716e-06,
      "loss": 0.5685,
      "step": 7740
    },
    {
      "epoch": 0.4177324483298257,
      "grad_norm": 0.9682522167155524,
      "learning_rate": 6.650125468459233e-06,
      "loss": 0.4628,
      "step": 7741
    },
    {
      "epoch": 0.4177864119583401,
      "grad_norm": 0.943441741037434,
      "learning_rate": 6.6493878970679836e-06,
      "loss": 0.4206,
      "step": 7742
    },
    {
      "epoch": 0.41784037558685444,
      "grad_norm": 1.0846250769761805,
      "learning_rate": 6.648650292642166e-06,
      "loss": 0.4928,
      "step": 7743
    },
    {
      "epoch": 0.41789433921536884,
      "grad_norm": 0.9727695251646855,
      "learning_rate": 6.647912655202977e-06,
      "loss": 0.4406,
      "step": 7744
    },
    {
      "epoch": 0.41794830284388323,
      "grad_norm": 1.0276312545126804,
      "learning_rate": 6.647174984771619e-06,
      "loss": 0.4769,
      "step": 7745
    },
    {
      "epoch": 0.4180022664723976,
      "grad_norm": 1.018220624298367,
      "learning_rate": 6.646437281369292e-06,
      "loss": 0.4867,
      "step": 7746
    },
    {
      "epoch": 0.418056230100912,
      "grad_norm": 1.1243014255378048,
      "learning_rate": 6.645699545017199e-06,
      "loss": 0.4464,
      "step": 7747
    },
    {
      "epoch": 0.41811019372942637,
      "grad_norm": 1.0579772006245831,
      "learning_rate": 6.644961775736543e-06,
      "loss": 0.4905,
      "step": 7748
    },
    {
      "epoch": 0.41816415735794077,
      "grad_norm": 1.055102604163941,
      "learning_rate": 6.6442239735485295e-06,
      "loss": 0.4954,
      "step": 7749
    },
    {
      "epoch": 0.4182181209864551,
      "grad_norm": 0.8708371009626861,
      "learning_rate": 6.643486138474363e-06,
      "loss": 0.3513,
      "step": 7750
    },
    {
      "epoch": 0.4182720846149695,
      "grad_norm": 1.1079858020279711,
      "learning_rate": 6.642748270535249e-06,
      "loss": 0.5938,
      "step": 7751
    },
    {
      "epoch": 0.4183260482434839,
      "grad_norm": 1.0988526306356867,
      "learning_rate": 6.6420103697523956e-06,
      "loss": 0.5741,
      "step": 7752
    },
    {
      "epoch": 0.41838001187199825,
      "grad_norm": 1.062089854354666,
      "learning_rate": 6.641272436147011e-06,
      "loss": 0.6019,
      "step": 7753
    },
    {
      "epoch": 0.41843397550051264,
      "grad_norm": 1.008804784232101,
      "learning_rate": 6.6405344697403015e-06,
      "loss": 0.5008,
      "step": 7754
    },
    {
      "epoch": 0.41848793912902704,
      "grad_norm": 0.8921464394982095,
      "learning_rate": 6.63979647055348e-06,
      "loss": 0.3818,
      "step": 7755
    },
    {
      "epoch": 0.41854190275754144,
      "grad_norm": 1.0961383607238422,
      "learning_rate": 6.639058438607755e-06,
      "loss": 0.4628,
      "step": 7756
    },
    {
      "epoch": 0.4185958663860558,
      "grad_norm": 1.147905413376192,
      "learning_rate": 6.638320373924342e-06,
      "loss": 0.4621,
      "step": 7757
    },
    {
      "epoch": 0.4186498300145702,
      "grad_norm": 0.8585187053076972,
      "learning_rate": 6.63758227652445e-06,
      "loss": 0.3995,
      "step": 7758
    },
    {
      "epoch": 0.4187037936430846,
      "grad_norm": 1.0111918491004452,
      "learning_rate": 6.6368441464292945e-06,
      "loss": 0.4617,
      "step": 7759
    },
    {
      "epoch": 0.4187577572715989,
      "grad_norm": 1.0938188146804737,
      "learning_rate": 6.636105983660089e-06,
      "loss": 0.5124,
      "step": 7760
    },
    {
      "epoch": 0.4188117209001133,
      "grad_norm": 1.000891344108365,
      "learning_rate": 6.635367788238048e-06,
      "loss": 0.4309,
      "step": 7761
    },
    {
      "epoch": 0.4188656845286277,
      "grad_norm": 1.0588034394319876,
      "learning_rate": 6.634629560184391e-06,
      "loss": 0.4612,
      "step": 7762
    },
    {
      "epoch": 0.4189196481571421,
      "grad_norm": 1.0453863050710182,
      "learning_rate": 6.633891299520334e-06,
      "loss": 0.5576,
      "step": 7763
    },
    {
      "epoch": 0.41897361178565645,
      "grad_norm": 1.2692290279377507,
      "learning_rate": 6.633153006267096e-06,
      "loss": 0.6017,
      "step": 7764
    },
    {
      "epoch": 0.41902757541417085,
      "grad_norm": 0.9496727426924458,
      "learning_rate": 6.632414680445894e-06,
      "loss": 0.3274,
      "step": 7765
    },
    {
      "epoch": 0.41908153904268525,
      "grad_norm": 1.0591229154831199,
      "learning_rate": 6.6316763220779514e-06,
      "loss": 0.4753,
      "step": 7766
    },
    {
      "epoch": 0.4191355026711996,
      "grad_norm": 1.130285433723368,
      "learning_rate": 6.630937931184489e-06,
      "loss": 0.6046,
      "step": 7767
    },
    {
      "epoch": 0.419189466299714,
      "grad_norm": 0.9900435987533374,
      "learning_rate": 6.630199507786724e-06,
      "loss": 0.3196,
      "step": 7768
    },
    {
      "epoch": 0.4192434299282284,
      "grad_norm": 0.9460307522221565,
      "learning_rate": 6.6294610519058866e-06,
      "loss": 0.4983,
      "step": 7769
    },
    {
      "epoch": 0.4192973935567428,
      "grad_norm": 1.2514171052386276,
      "learning_rate": 6.628722563563194e-06,
      "loss": 0.7269,
      "step": 7770
    },
    {
      "epoch": 0.4193513571852571,
      "grad_norm": 0.8667510067228321,
      "learning_rate": 6.627984042779877e-06,
      "loss": 0.3671,
      "step": 7771
    },
    {
      "epoch": 0.4194053208137715,
      "grad_norm": 1.1306985620698198,
      "learning_rate": 6.627245489577159e-06,
      "loss": 0.6191,
      "step": 7772
    },
    {
      "epoch": 0.4194592844422859,
      "grad_norm": 0.9324429709424953,
      "learning_rate": 6.626506903976265e-06,
      "loss": 0.3762,
      "step": 7773
    },
    {
      "epoch": 0.41951324807080026,
      "grad_norm": 0.9669956294515418,
      "learning_rate": 6.625768285998425e-06,
      "loss": 0.4218,
      "step": 7774
    },
    {
      "epoch": 0.41956721169931466,
      "grad_norm": 1.1285087648423466,
      "learning_rate": 6.625029635664867e-06,
      "loss": 0.5254,
      "step": 7775
    },
    {
      "epoch": 0.41962117532782905,
      "grad_norm": 1.1683379288462716,
      "learning_rate": 6.62429095299682e-06,
      "loss": 0.6028,
      "step": 7776
    },
    {
      "epoch": 0.41967513895634345,
      "grad_norm": 0.8868467489553826,
      "learning_rate": 6.623552238015516e-06,
      "loss": 0.3803,
      "step": 7777
    },
    {
      "epoch": 0.4197291025848578,
      "grad_norm": 1.0676556399560408,
      "learning_rate": 6.622813490742186e-06,
      "loss": 0.4608,
      "step": 7778
    },
    {
      "epoch": 0.4197830662133722,
      "grad_norm": 0.8193665734649929,
      "learning_rate": 6.62207471119806e-06,
      "loss": 0.3257,
      "step": 7779
    },
    {
      "epoch": 0.4198370298418866,
      "grad_norm": 0.9798255341882586,
      "learning_rate": 6.621335899404373e-06,
      "loss": 0.5325,
      "step": 7780
    },
    {
      "epoch": 0.41989099347040093,
      "grad_norm": 0.8458572165580065,
      "learning_rate": 6.6205970553823615e-06,
      "loss": 0.4301,
      "step": 7781
    },
    {
      "epoch": 0.41994495709891533,
      "grad_norm": 0.8991486852518086,
      "learning_rate": 6.619858179153257e-06,
      "loss": 0.4399,
      "step": 7782
    },
    {
      "epoch": 0.4199989207274297,
      "grad_norm": 1.1562903790624,
      "learning_rate": 6.619119270738296e-06,
      "loss": 0.5605,
      "step": 7783
    },
    {
      "epoch": 0.42005288435594407,
      "grad_norm": 0.9199516659641727,
      "learning_rate": 6.618380330158718e-06,
      "loss": 0.48,
      "step": 7784
    },
    {
      "epoch": 0.42010684798445846,
      "grad_norm": 1.0491509727781185,
      "learning_rate": 6.617641357435758e-06,
      "loss": 0.489,
      "step": 7785
    },
    {
      "epoch": 0.42016081161297286,
      "grad_norm": 1.0145074191748635,
      "learning_rate": 6.616902352590656e-06,
      "loss": 0.5132,
      "step": 7786
    },
    {
      "epoch": 0.42021477524148726,
      "grad_norm": 1.124281629184787,
      "learning_rate": 6.6161633156446525e-06,
      "loss": 0.4838,
      "step": 7787
    },
    {
      "epoch": 0.4202687388700016,
      "grad_norm": 1.002743952280283,
      "learning_rate": 6.615424246618987e-06,
      "loss": 0.4723,
      "step": 7788
    },
    {
      "epoch": 0.420322702498516,
      "grad_norm": 0.9995524800169185,
      "learning_rate": 6.614685145534902e-06,
      "loss": 0.5729,
      "step": 7789
    },
    {
      "epoch": 0.4203766661270304,
      "grad_norm": 1.0752440887621597,
      "learning_rate": 6.61394601241364e-06,
      "loss": 0.6043,
      "step": 7790
    },
    {
      "epoch": 0.42043062975554474,
      "grad_norm": 1.1501232180622876,
      "learning_rate": 6.613206847276444e-06,
      "loss": 0.588,
      "step": 7791
    },
    {
      "epoch": 0.42048459338405914,
      "grad_norm": 1.0263677919693812,
      "learning_rate": 6.612467650144558e-06,
      "loss": 0.5683,
      "step": 7792
    },
    {
      "epoch": 0.42053855701257353,
      "grad_norm": 0.7675123401213942,
      "learning_rate": 6.611728421039227e-06,
      "loss": 0.3715,
      "step": 7793
    },
    {
      "epoch": 0.42059252064108793,
      "grad_norm": 1.1036730561362893,
      "learning_rate": 6.6109891599816985e-06,
      "loss": 0.4553,
      "step": 7794
    },
    {
      "epoch": 0.42064648426960227,
      "grad_norm": 0.9581763454745819,
      "learning_rate": 6.61024986699322e-06,
      "loss": 0.4299,
      "step": 7795
    },
    {
      "epoch": 0.42070044789811667,
      "grad_norm": 0.8279836596270003,
      "learning_rate": 6.609510542095037e-06,
      "loss": 0.4236,
      "step": 7796
    },
    {
      "epoch": 0.42075441152663107,
      "grad_norm": 0.9513580628012458,
      "learning_rate": 6.608771185308401e-06,
      "loss": 0.5073,
      "step": 7797
    },
    {
      "epoch": 0.4208083751551454,
      "grad_norm": 1.0735693349817763,
      "learning_rate": 6.608031796654559e-06,
      "loss": 0.4844,
      "step": 7798
    },
    {
      "epoch": 0.4208623387836598,
      "grad_norm": 0.8969257117589575,
      "learning_rate": 6.607292376154766e-06,
      "loss": 0.4285,
      "step": 7799
    },
    {
      "epoch": 0.4209163024121742,
      "grad_norm": 1.143306665451789,
      "learning_rate": 6.6065529238302685e-06,
      "loss": 0.6008,
      "step": 7800
    },
    {
      "epoch": 0.4209702660406886,
      "grad_norm": 1.064741603621822,
      "learning_rate": 6.605813439702323e-06,
      "loss": 0.4753,
      "step": 7801
    },
    {
      "epoch": 0.42102422966920294,
      "grad_norm": 0.92499046748397,
      "learning_rate": 6.605073923792183e-06,
      "loss": 0.4098,
      "step": 7802
    },
    {
      "epoch": 0.42107819329771734,
      "grad_norm": 0.9423727333617983,
      "learning_rate": 6.604334376121101e-06,
      "loss": 0.4073,
      "step": 7803
    },
    {
      "epoch": 0.42113215692623174,
      "grad_norm": 0.9512374344461975,
      "learning_rate": 6.6035947967103345e-06,
      "loss": 0.5366,
      "step": 7804
    },
    {
      "epoch": 0.4211861205547461,
      "grad_norm": 1.1776820621942818,
      "learning_rate": 6.602855185581136e-06,
      "loss": 0.5083,
      "step": 7805
    },
    {
      "epoch": 0.4212400841832605,
      "grad_norm": 1.0242391241129205,
      "learning_rate": 6.602115542754768e-06,
      "loss": 0.516,
      "step": 7806
    },
    {
      "epoch": 0.4212940478117749,
      "grad_norm": 1.0217435792485094,
      "learning_rate": 6.601375868252483e-06,
      "loss": 0.519,
      "step": 7807
    },
    {
      "epoch": 0.42134801144028927,
      "grad_norm": 1.3606981229417652,
      "learning_rate": 6.600636162095544e-06,
      "loss": 0.4309,
      "step": 7808
    },
    {
      "epoch": 0.4214019750688036,
      "grad_norm": 0.8945565279044887,
      "learning_rate": 6.5998964243052095e-06,
      "loss": 0.2956,
      "step": 7809
    },
    {
      "epoch": 0.421455938697318,
      "grad_norm": 0.7762997011529004,
      "learning_rate": 6.5991566549027406e-06,
      "loss": 0.3351,
      "step": 7810
    },
    {
      "epoch": 0.4215099023258324,
      "grad_norm": 1.034762129693803,
      "learning_rate": 6.5984168539093985e-06,
      "loss": 0.5062,
      "step": 7811
    },
    {
      "epoch": 0.42156386595434675,
      "grad_norm": 1.3049134480734557,
      "learning_rate": 6.5976770213464466e-06,
      "loss": 0.5295,
      "step": 7812
    },
    {
      "epoch": 0.42161782958286115,
      "grad_norm": 1.167667406003611,
      "learning_rate": 6.596937157235147e-06,
      "loss": 0.4985,
      "step": 7813
    },
    {
      "epoch": 0.42167179321137555,
      "grad_norm": 1.1247294035236357,
      "learning_rate": 6.596197261596765e-06,
      "loss": 0.4675,
      "step": 7814
    },
    {
      "epoch": 0.4217257568398899,
      "grad_norm": 1.088702954794712,
      "learning_rate": 6.595457334452568e-06,
      "loss": 0.4403,
      "step": 7815
    },
    {
      "epoch": 0.4217797204684043,
      "grad_norm": 0.9485167101058809,
      "learning_rate": 6.594717375823819e-06,
      "loss": 0.5457,
      "step": 7816
    },
    {
      "epoch": 0.4218336840969187,
      "grad_norm": 1.049024637549353,
      "learning_rate": 6.593977385731786e-06,
      "loss": 0.3848,
      "step": 7817
    },
    {
      "epoch": 0.4218876477254331,
      "grad_norm": 1.263031342058167,
      "learning_rate": 6.5932373641977376e-06,
      "loss": 0.5934,
      "step": 7818
    },
    {
      "epoch": 0.4219416113539474,
      "grad_norm": 1.000460521712132,
      "learning_rate": 6.592497311242944e-06,
      "loss": 0.4632,
      "step": 7819
    },
    {
      "epoch": 0.4219955749824618,
      "grad_norm": 0.9092427495045363,
      "learning_rate": 6.5917572268886746e-06,
      "loss": 0.3836,
      "step": 7820
    },
    {
      "epoch": 0.4220495386109762,
      "grad_norm": 0.8142914628193029,
      "learning_rate": 6.591017111156196e-06,
      "loss": 0.4127,
      "step": 7821
    },
    {
      "epoch": 0.42210350223949056,
      "grad_norm": 1.0069381525057215,
      "learning_rate": 6.590276964066785e-06,
      "loss": 0.5575,
      "step": 7822
    },
    {
      "epoch": 0.42215746586800496,
      "grad_norm": 0.717327075830327,
      "learning_rate": 6.589536785641712e-06,
      "loss": 0.3264,
      "step": 7823
    },
    {
      "epoch": 0.42221142949651935,
      "grad_norm": 1.0302790952393663,
      "learning_rate": 6.588796575902253e-06,
      "loss": 0.4521,
      "step": 7824
    },
    {
      "epoch": 0.42226539312503375,
      "grad_norm": 1.2320076566501479,
      "learning_rate": 6.588056334869678e-06,
      "loss": 0.6709,
      "step": 7825
    },
    {
      "epoch": 0.4223193567535481,
      "grad_norm": 1.0190578352246416,
      "learning_rate": 6.587316062565265e-06,
      "loss": 0.5236,
      "step": 7826
    },
    {
      "epoch": 0.4223733203820625,
      "grad_norm": 1.034545015001414,
      "learning_rate": 6.586575759010289e-06,
      "loss": 0.5559,
      "step": 7827
    },
    {
      "epoch": 0.4224272840105769,
      "grad_norm": 1.0071350049547674,
      "learning_rate": 6.585835424226029e-06,
      "loss": 0.4282,
      "step": 7828
    },
    {
      "epoch": 0.42248124763909123,
      "grad_norm": 1.2191075359663384,
      "learning_rate": 6.585095058233761e-06,
      "loss": 0.5636,
      "step": 7829
    },
    {
      "epoch": 0.4225352112676056,
      "grad_norm": 0.9658901177075248,
      "learning_rate": 6.584354661054765e-06,
      "loss": 0.4765,
      "step": 7830
    },
    {
      "epoch": 0.42258917489612,
      "grad_norm": 0.9374580914162814,
      "learning_rate": 6.58361423271032e-06,
      "loss": 0.3498,
      "step": 7831
    },
    {
      "epoch": 0.4226431385246344,
      "grad_norm": 0.895758330164406,
      "learning_rate": 6.582873773221707e-06,
      "loss": 0.4861,
      "step": 7832
    },
    {
      "epoch": 0.42269710215314876,
      "grad_norm": 0.8944225336029444,
      "learning_rate": 6.582133282610209e-06,
      "loss": 0.4296,
      "step": 7833
    },
    {
      "epoch": 0.42275106578166316,
      "grad_norm": 1.1077189302112191,
      "learning_rate": 6.581392760897106e-06,
      "loss": 0.5182,
      "step": 7834
    },
    {
      "epoch": 0.42280502941017756,
      "grad_norm": 0.8301822107658756,
      "learning_rate": 6.5806522081036826e-06,
      "loss": 0.4132,
      "step": 7835
    },
    {
      "epoch": 0.4228589930386919,
      "grad_norm": 0.6268144217525957,
      "learning_rate": 6.5799116242512225e-06,
      "loss": 0.2906,
      "step": 7836
    },
    {
      "epoch": 0.4229129566672063,
      "grad_norm": 0.9896523217798194,
      "learning_rate": 6.579171009361012e-06,
      "loss": 0.6334,
      "step": 7837
    },
    {
      "epoch": 0.4229669202957207,
      "grad_norm": 0.8350419924873439,
      "learning_rate": 6.578430363454335e-06,
      "loss": 0.3916,
      "step": 7838
    },
    {
      "epoch": 0.4230208839242351,
      "grad_norm": 1.0517521928880973,
      "learning_rate": 6.577689686552481e-06,
      "loss": 0.5376,
      "step": 7839
    },
    {
      "epoch": 0.42307484755274943,
      "grad_norm": 1.11797029679992,
      "learning_rate": 6.576948978676736e-06,
      "loss": 0.5482,
      "step": 7840
    },
    {
      "epoch": 0.42312881118126383,
      "grad_norm": 1.2988290814149865,
      "learning_rate": 6.576208239848389e-06,
      "loss": 0.6326,
      "step": 7841
    },
    {
      "epoch": 0.42318277480977823,
      "grad_norm": 0.9337397808453255,
      "learning_rate": 6.5754674700887275e-06,
      "loss": 0.344,
      "step": 7842
    },
    {
      "epoch": 0.42323673843829257,
      "grad_norm": 0.8631833910426425,
      "learning_rate": 6.574726669419048e-06,
      "loss": 0.346,
      "step": 7843
    },
    {
      "epoch": 0.42329070206680697,
      "grad_norm": 1.192643046022319,
      "learning_rate": 6.573985837860636e-06,
      "loss": 0.6669,
      "step": 7844
    },
    {
      "epoch": 0.42334466569532136,
      "grad_norm": 0.9537288597040855,
      "learning_rate": 6.573244975434786e-06,
      "loss": 0.457,
      "step": 7845
    },
    {
      "epoch": 0.42339862932383576,
      "grad_norm": 1.1434855409991436,
      "learning_rate": 6.572504082162791e-06,
      "loss": 0.5275,
      "step": 7846
    },
    {
      "epoch": 0.4234525929523501,
      "grad_norm": 1.1836437635085957,
      "learning_rate": 6.571763158065946e-06,
      "loss": 0.6804,
      "step": 7847
    },
    {
      "epoch": 0.4235065565808645,
      "grad_norm": 0.9160717618933008,
      "learning_rate": 6.5710222031655446e-06,
      "loss": 0.5165,
      "step": 7848
    },
    {
      "epoch": 0.4235605202093789,
      "grad_norm": 0.9201978977447625,
      "learning_rate": 6.570281217482882e-06,
      "loss": 0.3791,
      "step": 7849
    },
    {
      "epoch": 0.42361448383789324,
      "grad_norm": 1.066159198059642,
      "learning_rate": 6.5695402010392566e-06,
      "loss": 0.4569,
      "step": 7850
    },
    {
      "epoch": 0.42366844746640764,
      "grad_norm": 0.7939236768458465,
      "learning_rate": 6.5687991538559635e-06,
      "loss": 0.3678,
      "step": 7851
    },
    {
      "epoch": 0.42372241109492204,
      "grad_norm": 1.2515332684550198,
      "learning_rate": 6.568058075954304e-06,
      "loss": 0.5549,
      "step": 7852
    },
    {
      "epoch": 0.4237763747234364,
      "grad_norm": 1.1243208012733275,
      "learning_rate": 6.567316967355577e-06,
      "loss": 0.4527,
      "step": 7853
    },
    {
      "epoch": 0.4238303383519508,
      "grad_norm": 0.952789979296829,
      "learning_rate": 6.566575828081081e-06,
      "loss": 0.4472,
      "step": 7854
    },
    {
      "epoch": 0.4238843019804652,
      "grad_norm": 0.8750794234103747,
      "learning_rate": 6.565834658152117e-06,
      "loss": 0.4304,
      "step": 7855
    },
    {
      "epoch": 0.42393826560897957,
      "grad_norm": 1.0518528209974933,
      "learning_rate": 6.56509345758999e-06,
      "loss": 0.471,
      "step": 7856
    },
    {
      "epoch": 0.4239922292374939,
      "grad_norm": 1.0571308562445403,
      "learning_rate": 6.564352226416001e-06,
      "loss": 0.4665,
      "step": 7857
    },
    {
      "epoch": 0.4240461928660083,
      "grad_norm": 1.0594644800431554,
      "learning_rate": 6.563610964651451e-06,
      "loss": 0.6139,
      "step": 7858
    },
    {
      "epoch": 0.4241001564945227,
      "grad_norm": 1.0454436526851603,
      "learning_rate": 6.562869672317649e-06,
      "loss": 0.5317,
      "step": 7859
    },
    {
      "epoch": 0.42415412012303705,
      "grad_norm": 1.24020282265456,
      "learning_rate": 6.562128349435899e-06,
      "loss": 0.7382,
      "step": 7860
    },
    {
      "epoch": 0.42420808375155145,
      "grad_norm": 1.0892418148459613,
      "learning_rate": 6.561386996027507e-06,
      "loss": 0.4053,
      "step": 7861
    },
    {
      "epoch": 0.42426204738006584,
      "grad_norm": 1.0220002922575258,
      "learning_rate": 6.560645612113779e-06,
      "loss": 0.5085,
      "step": 7862
    },
    {
      "epoch": 0.42431601100858024,
      "grad_norm": 1.4223807911278348,
      "learning_rate": 6.559904197716025e-06,
      "loss": 0.5826,
      "step": 7863
    },
    {
      "epoch": 0.4243699746370946,
      "grad_norm": 1.164251164994881,
      "learning_rate": 6.559162752855554e-06,
      "loss": 0.6624,
      "step": 7864
    },
    {
      "epoch": 0.424423938265609,
      "grad_norm": 1.0749444945715358,
      "learning_rate": 6.558421277553675e-06,
      "loss": 0.5624,
      "step": 7865
    },
    {
      "epoch": 0.4244779018941234,
      "grad_norm": 1.048261327446229,
      "learning_rate": 6.557679771831698e-06,
      "loss": 0.4769,
      "step": 7866
    },
    {
      "epoch": 0.4245318655226377,
      "grad_norm": 0.9442833418615849,
      "learning_rate": 6.556938235710938e-06,
      "loss": 0.3909,
      "step": 7867
    },
    {
      "epoch": 0.4245858291511521,
      "grad_norm": 1.1319720414956345,
      "learning_rate": 6.556196669212702e-06,
      "loss": 0.4852,
      "step": 7868
    },
    {
      "epoch": 0.4246397927796665,
      "grad_norm": 1.052014469079334,
      "learning_rate": 6.5554550723583075e-06,
      "loss": 0.4921,
      "step": 7869
    },
    {
      "epoch": 0.4246937564081809,
      "grad_norm": 1.1235685861392024,
      "learning_rate": 6.554713445169068e-06,
      "loss": 0.5263,
      "step": 7870
    },
    {
      "epoch": 0.42474772003669525,
      "grad_norm": 1.1386474321578377,
      "learning_rate": 6.5539717876663e-06,
      "loss": 0.4861,
      "step": 7871
    },
    {
      "epoch": 0.42480168366520965,
      "grad_norm": 0.9575977368058407,
      "learning_rate": 6.553230099871318e-06,
      "loss": 0.4009,
      "step": 7872
    },
    {
      "epoch": 0.42485564729372405,
      "grad_norm": 1.1063285148284108,
      "learning_rate": 6.552488381805436e-06,
      "loss": 0.492,
      "step": 7873
    },
    {
      "epoch": 0.4249096109222384,
      "grad_norm": 1.204554810975995,
      "learning_rate": 6.551746633489976e-06,
      "loss": 0.6741,
      "step": 7874
    },
    {
      "epoch": 0.4249635745507528,
      "grad_norm": 0.930057334471502,
      "learning_rate": 6.551004854946255e-06,
      "loss": 0.421,
      "step": 7875
    },
    {
      "epoch": 0.4250175381792672,
      "grad_norm": 1.2133036502479575,
      "learning_rate": 6.550263046195591e-06,
      "loss": 0.5385,
      "step": 7876
    },
    {
      "epoch": 0.4250715018077816,
      "grad_norm": 0.935436758230232,
      "learning_rate": 6.549521207259308e-06,
      "loss": 0.4051,
      "step": 7877
    },
    {
      "epoch": 0.4251254654362959,
      "grad_norm": 1.1831219363151877,
      "learning_rate": 6.548779338158724e-06,
      "loss": 0.4557,
      "step": 7878
    },
    {
      "epoch": 0.4251794290648103,
      "grad_norm": 1.0288225156355828,
      "learning_rate": 6.548037438915161e-06,
      "loss": 0.479,
      "step": 7879
    },
    {
      "epoch": 0.4252333926933247,
      "grad_norm": 1.0688235620197069,
      "learning_rate": 6.547295509549946e-06,
      "loss": 0.4923,
      "step": 7880
    },
    {
      "epoch": 0.42528735632183906,
      "grad_norm": 1.0801609640581717,
      "learning_rate": 6.546553550084398e-06,
      "loss": 0.5184,
      "step": 7881
    },
    {
      "epoch": 0.42534131995035346,
      "grad_norm": 0.8725134693275037,
      "learning_rate": 6.545811560539843e-06,
      "loss": 0.3938,
      "step": 7882
    },
    {
      "epoch": 0.42539528357886786,
      "grad_norm": 0.8710258728784571,
      "learning_rate": 6.545069540937607e-06,
      "loss": 0.41,
      "step": 7883
    },
    {
      "epoch": 0.4254492472073822,
      "grad_norm": 0.8592848460981608,
      "learning_rate": 6.544327491299019e-06,
      "loss": 0.347,
      "step": 7884
    },
    {
      "epoch": 0.4255032108358966,
      "grad_norm": 0.8782234756180605,
      "learning_rate": 6.543585411645401e-06,
      "loss": 0.3697,
      "step": 7885
    },
    {
      "epoch": 0.425557174464411,
      "grad_norm": 1.031162586977422,
      "learning_rate": 6.542843301998085e-06,
      "loss": 0.5044,
      "step": 7886
    },
    {
      "epoch": 0.4256111380929254,
      "grad_norm": 0.8883330611783522,
      "learning_rate": 6.542101162378397e-06,
      "loss": 0.4087,
      "step": 7887
    },
    {
      "epoch": 0.42566510172143973,
      "grad_norm": 1.2102891257776733,
      "learning_rate": 6.541358992807671e-06,
      "loss": 0.5828,
      "step": 7888
    },
    {
      "epoch": 0.42571906534995413,
      "grad_norm": 1.0374352880276512,
      "learning_rate": 6.540616793307235e-06,
      "loss": 0.5721,
      "step": 7889
    },
    {
      "epoch": 0.4257730289784685,
      "grad_norm": 1.1349447291995236,
      "learning_rate": 6.53987456389842e-06,
      "loss": 0.5534,
      "step": 7890
    },
    {
      "epoch": 0.42582699260698287,
      "grad_norm": 1.2036280693015913,
      "learning_rate": 6.53913230460256e-06,
      "loss": 0.5168,
      "step": 7891
    },
    {
      "epoch": 0.42588095623549727,
      "grad_norm": 0.9093076054244045,
      "learning_rate": 6.538390015440987e-06,
      "loss": 0.397,
      "step": 7892
    },
    {
      "epoch": 0.42593491986401166,
      "grad_norm": 1.0136557091778857,
      "learning_rate": 6.537647696435035e-06,
      "loss": 0.4915,
      "step": 7893
    },
    {
      "epoch": 0.42598888349252606,
      "grad_norm": 1.0698799942849961,
      "learning_rate": 6.53690534760604e-06,
      "loss": 0.5014,
      "step": 7894
    },
    {
      "epoch": 0.4260428471210404,
      "grad_norm": 1.0501582154162956,
      "learning_rate": 6.5361629689753395e-06,
      "loss": 0.5126,
      "step": 7895
    },
    {
      "epoch": 0.4260968107495548,
      "grad_norm": 0.9840439418147563,
      "learning_rate": 6.535420560564267e-06,
      "loss": 0.4919,
      "step": 7896
    },
    {
      "epoch": 0.4261507743780692,
      "grad_norm": 1.0671901704533855,
      "learning_rate": 6.534678122394161e-06,
      "loss": 0.5331,
      "step": 7897
    },
    {
      "epoch": 0.42620473800658354,
      "grad_norm": 0.7918471542269024,
      "learning_rate": 6.533935654486361e-06,
      "loss": 0.3214,
      "step": 7898
    },
    {
      "epoch": 0.42625870163509794,
      "grad_norm": 0.9011671399727269,
      "learning_rate": 6.533193156862206e-06,
      "loss": 0.4676,
      "step": 7899
    },
    {
      "epoch": 0.42631266526361233,
      "grad_norm": 1.1985156063571485,
      "learning_rate": 6.532450629543035e-06,
      "loss": 0.441,
      "step": 7900
    },
    {
      "epoch": 0.42636662889212673,
      "grad_norm": 0.8999357944710695,
      "learning_rate": 6.5317080725501895e-06,
      "loss": 0.5057,
      "step": 7901
    },
    {
      "epoch": 0.4264205925206411,
      "grad_norm": 1.1319670999135547,
      "learning_rate": 6.5309654859050105e-06,
      "loss": 0.556,
      "step": 7902
    },
    {
      "epoch": 0.42647455614915547,
      "grad_norm": 1.080390029645909,
      "learning_rate": 6.530222869628844e-06,
      "loss": 0.5142,
      "step": 7903
    },
    {
      "epoch": 0.42652851977766987,
      "grad_norm": 0.9117917229278997,
      "learning_rate": 6.529480223743029e-06,
      "loss": 0.4969,
      "step": 7904
    },
    {
      "epoch": 0.4265824834061842,
      "grad_norm": 0.9411410592243175,
      "learning_rate": 6.5287375482689144e-06,
      "loss": 0.3978,
      "step": 7905
    },
    {
      "epoch": 0.4266364470346986,
      "grad_norm": 1.0826400551446058,
      "learning_rate": 6.527994843227842e-06,
      "loss": 0.4793,
      "step": 7906
    },
    {
      "epoch": 0.426690410663213,
      "grad_norm": 1.03550714822426,
      "learning_rate": 6.527252108641159e-06,
      "loss": 0.4833,
      "step": 7907
    },
    {
      "epoch": 0.4267443742917274,
      "grad_norm": 1.2501866983120542,
      "learning_rate": 6.526509344530212e-06,
      "loss": 0.5302,
      "step": 7908
    },
    {
      "epoch": 0.42679833792024174,
      "grad_norm": 0.8819800977656523,
      "learning_rate": 6.525766550916352e-06,
      "loss": 0.3731,
      "step": 7909
    },
    {
      "epoch": 0.42685230154875614,
      "grad_norm": 1.1198496565464324,
      "learning_rate": 6.525023727820922e-06,
      "loss": 0.5125,
      "step": 7910
    },
    {
      "epoch": 0.42690626517727054,
      "grad_norm": 1.1310846654143047,
      "learning_rate": 6.524280875265274e-06,
      "loss": 0.5418,
      "step": 7911
    },
    {
      "epoch": 0.4269602288057849,
      "grad_norm": 0.960488343703718,
      "learning_rate": 6.523537993270761e-06,
      "loss": 0.4327,
      "step": 7912
    },
    {
      "epoch": 0.4270141924342993,
      "grad_norm": 0.784766116185314,
      "learning_rate": 6.52279508185873e-06,
      "loss": 0.354,
      "step": 7913
    },
    {
      "epoch": 0.4270681560628137,
      "grad_norm": 1.0290480890895213,
      "learning_rate": 6.522052141050536e-06,
      "loss": 0.5283,
      "step": 7914
    },
    {
      "epoch": 0.427122119691328,
      "grad_norm": 0.9935958855490318,
      "learning_rate": 6.521309170867532e-06,
      "loss": 0.3786,
      "step": 7915
    },
    {
      "epoch": 0.4271760833198424,
      "grad_norm": 0.7352870374874995,
      "learning_rate": 6.520566171331067e-06,
      "loss": 0.3569,
      "step": 7916
    },
    {
      "epoch": 0.4272300469483568,
      "grad_norm": 0.963731007082754,
      "learning_rate": 6.519823142462501e-06,
      "loss": 0.4505,
      "step": 7917
    },
    {
      "epoch": 0.4272840105768712,
      "grad_norm": 1.0030082120337285,
      "learning_rate": 6.519080084283187e-06,
      "loss": 0.3375,
      "step": 7918
    },
    {
      "epoch": 0.42733797420538555,
      "grad_norm": 1.03661784886935,
      "learning_rate": 6.518336996814483e-06,
      "loss": 0.4212,
      "step": 7919
    },
    {
      "epoch": 0.42739193783389995,
      "grad_norm": 1.1157506103807897,
      "learning_rate": 6.517593880077744e-06,
      "loss": 0.5509,
      "step": 7920
    },
    {
      "epoch": 0.42744590146241435,
      "grad_norm": 0.9911867851820126,
      "learning_rate": 6.516850734094328e-06,
      "loss": 0.3999,
      "step": 7921
    },
    {
      "epoch": 0.4274998650909287,
      "grad_norm": 0.9850298105711124,
      "learning_rate": 6.516107558885595e-06,
      "loss": 0.4765,
      "step": 7922
    },
    {
      "epoch": 0.4275538287194431,
      "grad_norm": 0.8724980641331733,
      "learning_rate": 6.515364354472904e-06,
      "loss": 0.4526,
      "step": 7923
    },
    {
      "epoch": 0.4276077923479575,
      "grad_norm": 0.9373870837600559,
      "learning_rate": 6.514621120877614e-06,
      "loss": 0.4205,
      "step": 7924
    },
    {
      "epoch": 0.4276617559764719,
      "grad_norm": 1.112776809122554,
      "learning_rate": 6.51387785812109e-06,
      "loss": 0.582,
      "step": 7925
    },
    {
      "epoch": 0.4277157196049862,
      "grad_norm": 1.0519421505793276,
      "learning_rate": 6.5131345662246905e-06,
      "loss": 0.4908,
      "step": 7926
    },
    {
      "epoch": 0.4277696832335006,
      "grad_norm": 1.2510851087084056,
      "learning_rate": 6.51239124520978e-06,
      "loss": 0.5699,
      "step": 7927
    },
    {
      "epoch": 0.427823646862015,
      "grad_norm": 1.0172699519012407,
      "learning_rate": 6.5116478950977224e-06,
      "loss": 0.5477,
      "step": 7928
    },
    {
      "epoch": 0.42787761049052936,
      "grad_norm": 0.9979069677691887,
      "learning_rate": 6.510904515909884e-06,
      "loss": 0.4683,
      "step": 7929
    },
    {
      "epoch": 0.42793157411904376,
      "grad_norm": 0.8898701623591555,
      "learning_rate": 6.510161107667627e-06,
      "loss": 0.3595,
      "step": 7930
    },
    {
      "epoch": 0.42798553774755815,
      "grad_norm": 1.0802252419158502,
      "learning_rate": 6.5094176703923175e-06,
      "loss": 0.6147,
      "step": 7931
    },
    {
      "epoch": 0.42803950137607255,
      "grad_norm": 1.0973492830116818,
      "learning_rate": 6.508674204105324e-06,
      "loss": 0.4953,
      "step": 7932
    },
    {
      "epoch": 0.4280934650045869,
      "grad_norm": 0.9303407388314952,
      "learning_rate": 6.507930708828019e-06,
      "loss": 0.4524,
      "step": 7933
    },
    {
      "epoch": 0.4281474286331013,
      "grad_norm": 0.8978471583569569,
      "learning_rate": 6.507187184581763e-06,
      "loss": 0.4538,
      "step": 7934
    },
    {
      "epoch": 0.4282013922616157,
      "grad_norm": 0.9287534694005591,
      "learning_rate": 6.5064436313879306e-06,
      "loss": 0.3743,
      "step": 7935
    },
    {
      "epoch": 0.42825535589013003,
      "grad_norm": 0.9964726067406385,
      "learning_rate": 6.505700049267893e-06,
      "loss": 0.6128,
      "step": 7936
    },
    {
      "epoch": 0.4283093195186444,
      "grad_norm": 1.104509746118136,
      "learning_rate": 6.504956438243017e-06,
      "loss": 0.5888,
      "step": 7937
    },
    {
      "epoch": 0.4283632831471588,
      "grad_norm": 0.9144669923308995,
      "learning_rate": 6.504212798334678e-06,
      "loss": 0.4597,
      "step": 7938
    },
    {
      "epoch": 0.4284172467756732,
      "grad_norm": 0.8718643045464086,
      "learning_rate": 6.503469129564249e-06,
      "loss": 0.3489,
      "step": 7939
    },
    {
      "epoch": 0.42847121040418756,
      "grad_norm": 1.0144561307939426,
      "learning_rate": 6.502725431953105e-06,
      "loss": 0.5418,
      "step": 7940
    },
    {
      "epoch": 0.42852517403270196,
      "grad_norm": 0.8393232357207013,
      "learning_rate": 6.501981705522617e-06,
      "loss": 0.3163,
      "step": 7941
    },
    {
      "epoch": 0.42857913766121636,
      "grad_norm": 1.0852202392465276,
      "learning_rate": 6.501237950294162e-06,
      "loss": 0.4852,
      "step": 7942
    },
    {
      "epoch": 0.4286331012897307,
      "grad_norm": 1.0253867678180413,
      "learning_rate": 6.500494166289119e-06,
      "loss": 0.5402,
      "step": 7943
    },
    {
      "epoch": 0.4286870649182451,
      "grad_norm": 0.906074018292121,
      "learning_rate": 6.499750353528861e-06,
      "loss": 0.2975,
      "step": 7944
    },
    {
      "epoch": 0.4287410285467595,
      "grad_norm": 1.0271366324591424,
      "learning_rate": 6.499006512034768e-06,
      "loss": 0.4251,
      "step": 7945
    },
    {
      "epoch": 0.4287949921752739,
      "grad_norm": 0.821837432737237,
      "learning_rate": 6.498262641828218e-06,
      "loss": 0.3485,
      "step": 7946
    },
    {
      "epoch": 0.42884895580378823,
      "grad_norm": 1.356700536686015,
      "learning_rate": 6.497518742930591e-06,
      "loss": 0.6673,
      "step": 7947
    },
    {
      "epoch": 0.42890291943230263,
      "grad_norm": 0.9300836948882877,
      "learning_rate": 6.496774815363268e-06,
      "loss": 0.4523,
      "step": 7948
    },
    {
      "epoch": 0.42895688306081703,
      "grad_norm": 1.2611585801026053,
      "learning_rate": 6.496030859147628e-06,
      "loss": 0.4314,
      "step": 7949
    },
    {
      "epoch": 0.42901084668933137,
      "grad_norm": 1.237487737794435,
      "learning_rate": 6.495286874305057e-06,
      "loss": 0.4343,
      "step": 7950
    },
    {
      "epoch": 0.42906481031784577,
      "grad_norm": 0.8231913275096322,
      "learning_rate": 6.494542860856935e-06,
      "loss": 0.3724,
      "step": 7951
    },
    {
      "epoch": 0.42911877394636017,
      "grad_norm": 1.0103044372241177,
      "learning_rate": 6.493798818824646e-06,
      "loss": 0.4141,
      "step": 7952
    },
    {
      "epoch": 0.4291727375748745,
      "grad_norm": 1.063909439250591,
      "learning_rate": 6.493054748229573e-06,
      "loss": 0.4624,
      "step": 7953
    },
    {
      "epoch": 0.4292267012033889,
      "grad_norm": 0.9053833963494851,
      "learning_rate": 6.4923106490931055e-06,
      "loss": 0.5142,
      "step": 7954
    },
    {
      "epoch": 0.4292806648319033,
      "grad_norm": 1.1783021034140946,
      "learning_rate": 6.491566521436627e-06,
      "loss": 0.5219,
      "step": 7955
    },
    {
      "epoch": 0.4293346284604177,
      "grad_norm": 0.9624738544314634,
      "learning_rate": 6.4908223652815236e-06,
      "loss": 0.4415,
      "step": 7956
    },
    {
      "epoch": 0.42938859208893204,
      "grad_norm": 0.9711778814964833,
      "learning_rate": 6.490078180649186e-06,
      "loss": 0.4294,
      "step": 7957
    },
    {
      "epoch": 0.42944255571744644,
      "grad_norm": 0.9972705904645048,
      "learning_rate": 6.489333967561e-06,
      "loss": 0.5048,
      "step": 7958
    },
    {
      "epoch": 0.42949651934596084,
      "grad_norm": 1.0269519863186192,
      "learning_rate": 6.4885897260383556e-06,
      "loss": 0.5705,
      "step": 7959
    },
    {
      "epoch": 0.4295504829744752,
      "grad_norm": 1.114008004273625,
      "learning_rate": 6.487845456102645e-06,
      "loss": 0.4088,
      "step": 7960
    },
    {
      "epoch": 0.4296044466029896,
      "grad_norm": 1.1968023049346563,
      "learning_rate": 6.487101157775257e-06,
      "loss": 0.674,
      "step": 7961
    },
    {
      "epoch": 0.429658410231504,
      "grad_norm": 1.061390498161719,
      "learning_rate": 6.486356831077584e-06,
      "loss": 0.5433,
      "step": 7962
    },
    {
      "epoch": 0.42971237386001837,
      "grad_norm": 1.0596716119619518,
      "learning_rate": 6.485612476031019e-06,
      "loss": 0.4144,
      "step": 7963
    },
    {
      "epoch": 0.4297663374885327,
      "grad_norm": 1.1300407892256106,
      "learning_rate": 6.4848680926569565e-06,
      "loss": 0.5082,
      "step": 7964
    },
    {
      "epoch": 0.4298203011170471,
      "grad_norm": 1.081227335739126,
      "learning_rate": 6.48412368097679e-06,
      "loss": 0.4936,
      "step": 7965
    },
    {
      "epoch": 0.4298742647455615,
      "grad_norm": 0.8463809941918291,
      "learning_rate": 6.483379241011913e-06,
      "loss": 0.4332,
      "step": 7966
    },
    {
      "epoch": 0.42992822837407585,
      "grad_norm": 1.0603295311294068,
      "learning_rate": 6.482634772783725e-06,
      "loss": 0.4259,
      "step": 7967
    },
    {
      "epoch": 0.42998219200259025,
      "grad_norm": 1.00497304319385,
      "learning_rate": 6.4818902763136195e-06,
      "loss": 0.4144,
      "step": 7968
    },
    {
      "epoch": 0.43003615563110464,
      "grad_norm": 1.090616774902251,
      "learning_rate": 6.481145751622994e-06,
      "loss": 0.4345,
      "step": 7969
    },
    {
      "epoch": 0.43009011925961904,
      "grad_norm": 1.0708404719021594,
      "learning_rate": 6.48040119873325e-06,
      "loss": 0.5187,
      "step": 7970
    },
    {
      "epoch": 0.4301440828881334,
      "grad_norm": 0.9919900045996579,
      "learning_rate": 6.479656617665785e-06,
      "loss": 0.4554,
      "step": 7971
    },
    {
      "epoch": 0.4301980465166478,
      "grad_norm": 1.2814516421469635,
      "learning_rate": 6.478912008441999e-06,
      "loss": 0.5316,
      "step": 7972
    },
    {
      "epoch": 0.4302520101451622,
      "grad_norm": 1.1140540841017712,
      "learning_rate": 6.478167371083292e-06,
      "loss": 0.4904,
      "step": 7973
    },
    {
      "epoch": 0.4303059737736765,
      "grad_norm": 1.1315510742878296,
      "learning_rate": 6.477422705611067e-06,
      "loss": 0.481,
      "step": 7974
    },
    {
      "epoch": 0.4303599374021909,
      "grad_norm": 1.0761918450490597,
      "learning_rate": 6.476678012046725e-06,
      "loss": 0.5082,
      "step": 7975
    },
    {
      "epoch": 0.4304139010307053,
      "grad_norm": 1.0460941366449146,
      "learning_rate": 6.47593329041167e-06,
      "loss": 0.5471,
      "step": 7976
    },
    {
      "epoch": 0.4304678646592197,
      "grad_norm": 1.119650556548085,
      "learning_rate": 6.475188540727306e-06,
      "loss": 0.4849,
      "step": 7977
    },
    {
      "epoch": 0.43052182828773405,
      "grad_norm": 1.0214626505058457,
      "learning_rate": 6.474443763015039e-06,
      "loss": 0.4893,
      "step": 7978
    },
    {
      "epoch": 0.43057579191624845,
      "grad_norm": 0.9433567318662003,
      "learning_rate": 6.4736989572962725e-06,
      "loss": 0.4156,
      "step": 7979
    },
    {
      "epoch": 0.43062975554476285,
      "grad_norm": 0.9496464732559519,
      "learning_rate": 6.472954123592415e-06,
      "loss": 0.4403,
      "step": 7980
    },
    {
      "epoch": 0.4306837191732772,
      "grad_norm": 1.0101994253906252,
      "learning_rate": 6.472209261924874e-06,
      "loss": 0.5364,
      "step": 7981
    },
    {
      "epoch": 0.4307376828017916,
      "grad_norm": 1.5101717681756666,
      "learning_rate": 6.471464372315055e-06,
      "loss": 0.5231,
      "step": 7982
    },
    {
      "epoch": 0.430791646430306,
      "grad_norm": 1.1165661716952662,
      "learning_rate": 6.470719454784369e-06,
      "loss": 0.5773,
      "step": 7983
    },
    {
      "epoch": 0.4308456100588203,
      "grad_norm": 1.0694760095824354,
      "learning_rate": 6.4699745093542254e-06,
      "loss": 0.453,
      "step": 7984
    },
    {
      "epoch": 0.4308995736873347,
      "grad_norm": 1.0302393261749057,
      "learning_rate": 6.469229536046033e-06,
      "loss": 0.4842,
      "step": 7985
    },
    {
      "epoch": 0.4309535373158491,
      "grad_norm": 0.9388323192944762,
      "learning_rate": 6.468484534881206e-06,
      "loss": 0.3682,
      "step": 7986
    },
    {
      "epoch": 0.4310075009443635,
      "grad_norm": 1.0427892575394033,
      "learning_rate": 6.467739505881152e-06,
      "loss": 0.4745,
      "step": 7987
    },
    {
      "epoch": 0.43106146457287786,
      "grad_norm": 1.03478702783235,
      "learning_rate": 6.466994449067289e-06,
      "loss": 0.5533,
      "step": 7988
    },
    {
      "epoch": 0.43111542820139226,
      "grad_norm": 1.0044919513893809,
      "learning_rate": 6.466249364461028e-06,
      "loss": 0.4237,
      "step": 7989
    },
    {
      "epoch": 0.43116939182990666,
      "grad_norm": 1.0943553435622573,
      "learning_rate": 6.465504252083784e-06,
      "loss": 0.6121,
      "step": 7990
    },
    {
      "epoch": 0.431223355458421,
      "grad_norm": 1.0595969140536157,
      "learning_rate": 6.464759111956973e-06,
      "loss": 0.5318,
      "step": 7991
    },
    {
      "epoch": 0.4312773190869354,
      "grad_norm": 1.1863521769814103,
      "learning_rate": 6.46401394410201e-06,
      "loss": 0.6781,
      "step": 7992
    },
    {
      "epoch": 0.4313312827154498,
      "grad_norm": 1.136189988061752,
      "learning_rate": 6.4632687485403115e-06,
      "loss": 0.4817,
      "step": 7993
    },
    {
      "epoch": 0.4313852463439642,
      "grad_norm": 0.8079749771212793,
      "learning_rate": 6.462523525293295e-06,
      "loss": 0.3373,
      "step": 7994
    },
    {
      "epoch": 0.43143920997247853,
      "grad_norm": 1.0524757804639808,
      "learning_rate": 6.461778274382382e-06,
      "loss": 0.43,
      "step": 7995
    },
    {
      "epoch": 0.43149317360099293,
      "grad_norm": 0.7703881499431162,
      "learning_rate": 6.461032995828989e-06,
      "loss": 0.3599,
      "step": 7996
    },
    {
      "epoch": 0.4315471372295073,
      "grad_norm": 1.3202555821405633,
      "learning_rate": 6.460287689654535e-06,
      "loss": 0.6113,
      "step": 7997
    },
    {
      "epoch": 0.43160110085802167,
      "grad_norm": 1.0655317222366374,
      "learning_rate": 6.459542355880445e-06,
      "loss": 0.4125,
      "step": 7998
    },
    {
      "epoch": 0.43165506448653607,
      "grad_norm": 1.0805607215880697,
      "learning_rate": 6.458796994528136e-06,
      "loss": 0.4727,
      "step": 7999
    },
    {
      "epoch": 0.43170902811505046,
      "grad_norm": 0.8582092338717855,
      "learning_rate": 6.4580516056190334e-06,
      "loss": 0.4791,
      "step": 8000
    },
    {
      "epoch": 0.43170902811505046,
      "eval_loss": 0.555820882320404,
      "eval_runtime": 166.1295,
      "eval_samples_per_second": 20.701,
      "eval_steps_per_second": 0.867,
      "step": 8000
    },
    {
      "epoch": 0.43176299174356486,
      "grad_norm": 0.9966235165546949,
      "learning_rate": 6.457306189174559e-06,
      "loss": 0.3833,
      "step": 8001
    },
    {
      "epoch": 0.4318169553720792,
      "grad_norm": 1.027525741745414,
      "learning_rate": 6.456560745216139e-06,
      "loss": 0.3956,
      "step": 8002
    },
    {
      "epoch": 0.4318709190005936,
      "grad_norm": 1.1317187601111425,
      "learning_rate": 6.455815273765195e-06,
      "loss": 0.5781,
      "step": 8003
    },
    {
      "epoch": 0.431924882629108,
      "grad_norm": 1.0803088013711326,
      "learning_rate": 6.4550697748431545e-06,
      "loss": 0.4742,
      "step": 8004
    },
    {
      "epoch": 0.43197884625762234,
      "grad_norm": 0.9001692389401343,
      "learning_rate": 6.454324248471444e-06,
      "loss": 0.5093,
      "step": 8005
    },
    {
      "epoch": 0.43203280988613674,
      "grad_norm": 0.8206242173169033,
      "learning_rate": 6.453578694671491e-06,
      "loss": 0.4099,
      "step": 8006
    },
    {
      "epoch": 0.43208677351465113,
      "grad_norm": 0.955584064034573,
      "learning_rate": 6.452833113464722e-06,
      "loss": 0.4345,
      "step": 8007
    },
    {
      "epoch": 0.43214073714316553,
      "grad_norm": 1.0185153974031833,
      "learning_rate": 6.452087504872566e-06,
      "loss": 0.525,
      "step": 8008
    },
    {
      "epoch": 0.4321947007716799,
      "grad_norm": 1.0616851725071146,
      "learning_rate": 6.451341868916454e-06,
      "loss": 0.4836,
      "step": 8009
    },
    {
      "epoch": 0.43224866440019427,
      "grad_norm": 1.1175284008490858,
      "learning_rate": 6.450596205617815e-06,
      "loss": 0.4566,
      "step": 8010
    },
    {
      "epoch": 0.43230262802870867,
      "grad_norm": 1.0602954402529055,
      "learning_rate": 6.4498505149980815e-06,
      "loss": 0.6617,
      "step": 8011
    },
    {
      "epoch": 0.432356591657223,
      "grad_norm": 1.1091641879052583,
      "learning_rate": 6.4491047970786844e-06,
      "loss": 0.5464,
      "step": 8012
    },
    {
      "epoch": 0.4324105552857374,
      "grad_norm": 0.9664216949819006,
      "learning_rate": 6.448359051881055e-06,
      "loss": 0.4639,
      "step": 8013
    },
    {
      "epoch": 0.4324645189142518,
      "grad_norm": 1.0671588085083237,
      "learning_rate": 6.447613279426629e-06,
      "loss": 0.5246,
      "step": 8014
    },
    {
      "epoch": 0.4325184825427662,
      "grad_norm": 0.9953050598485057,
      "learning_rate": 6.446867479736839e-06,
      "loss": 0.4657,
      "step": 8015
    },
    {
      "epoch": 0.43257244617128054,
      "grad_norm": 1.136955749124388,
      "learning_rate": 6.446121652833123e-06,
      "loss": 0.494,
      "step": 8016
    },
    {
      "epoch": 0.43262640979979494,
      "grad_norm": 0.9929719131533222,
      "learning_rate": 6.445375798736913e-06,
      "loss": 0.4886,
      "step": 8017
    },
    {
      "epoch": 0.43268037342830934,
      "grad_norm": 1.087271087464239,
      "learning_rate": 6.444629917469647e-06,
      "loss": 0.5384,
      "step": 8018
    },
    {
      "epoch": 0.4327343370568237,
      "grad_norm": 1.1264946557462712,
      "learning_rate": 6.443884009052764e-06,
      "loss": 0.517,
      "step": 8019
    },
    {
      "epoch": 0.4327883006853381,
      "grad_norm": 1.2124720119031085,
      "learning_rate": 6.443138073507701e-06,
      "loss": 0.4419,
      "step": 8020
    },
    {
      "epoch": 0.4328422643138525,
      "grad_norm": 0.955610731677924,
      "learning_rate": 6.442392110855896e-06,
      "loss": 0.4156,
      "step": 8021
    },
    {
      "epoch": 0.4328962279423668,
      "grad_norm": 1.1352347708537454,
      "learning_rate": 6.44164612111879e-06,
      "loss": 0.5334,
      "step": 8022
    },
    {
      "epoch": 0.4329501915708812,
      "grad_norm": 0.9740608868368066,
      "learning_rate": 6.440900104317823e-06,
      "loss": 0.5386,
      "step": 8023
    },
    {
      "epoch": 0.4330041551993956,
      "grad_norm": 1.0113133293311571,
      "learning_rate": 6.440154060474435e-06,
      "loss": 0.4102,
      "step": 8024
    },
    {
      "epoch": 0.43305811882791,
      "grad_norm": 1.0627563267923887,
      "learning_rate": 6.43940798961007e-06,
      "loss": 0.5465,
      "step": 8025
    },
    {
      "epoch": 0.43311208245642435,
      "grad_norm": 0.9189982080698558,
      "learning_rate": 6.438661891746171e-06,
      "loss": 0.3948,
      "step": 8026
    },
    {
      "epoch": 0.43316604608493875,
      "grad_norm": 0.9608110351552288,
      "learning_rate": 6.43791576690418e-06,
      "loss": 0.3871,
      "step": 8027
    },
    {
      "epoch": 0.43322000971345315,
      "grad_norm": 1.128664149700366,
      "learning_rate": 6.437169615105543e-06,
      "loss": 0.6029,
      "step": 8028
    },
    {
      "epoch": 0.4332739733419675,
      "grad_norm": 1.23838706342371,
      "learning_rate": 6.436423436371704e-06,
      "loss": 0.6062,
      "step": 8029
    },
    {
      "epoch": 0.4333279369704819,
      "grad_norm": 1.1331837253287147,
      "learning_rate": 6.435677230724108e-06,
      "loss": 0.444,
      "step": 8030
    },
    {
      "epoch": 0.4333819005989963,
      "grad_norm": 0.9055049538561607,
      "learning_rate": 6.434930998184203e-06,
      "loss": 0.4771,
      "step": 8031
    },
    {
      "epoch": 0.4334358642275107,
      "grad_norm": 1.15706699723877,
      "learning_rate": 6.434184738773438e-06,
      "loss": 0.6248,
      "step": 8032
    },
    {
      "epoch": 0.433489827856025,
      "grad_norm": 0.9927963828321534,
      "learning_rate": 6.43343845251326e-06,
      "loss": 0.4158,
      "step": 8033
    },
    {
      "epoch": 0.4335437914845394,
      "grad_norm": 1.0006950383734416,
      "learning_rate": 6.4326921394251155e-06,
      "loss": 0.4932,
      "step": 8034
    },
    {
      "epoch": 0.4335977551130538,
      "grad_norm": 1.0255084737081837,
      "learning_rate": 6.431945799530458e-06,
      "loss": 0.4654,
      "step": 8035
    },
    {
      "epoch": 0.43365171874156816,
      "grad_norm": 0.9161309082014335,
      "learning_rate": 6.431199432850738e-06,
      "loss": 0.6028,
      "step": 8036
    },
    {
      "epoch": 0.43370568237008256,
      "grad_norm": 1.0022456988498338,
      "learning_rate": 6.430453039407403e-06,
      "loss": 0.4768,
      "step": 8037
    },
    {
      "epoch": 0.43375964599859695,
      "grad_norm": 1.115672070292445,
      "learning_rate": 6.429706619221907e-06,
      "loss": 0.6394,
      "step": 8038
    },
    {
      "epoch": 0.43381360962711135,
      "grad_norm": 1.0370494415761422,
      "learning_rate": 6.428960172315705e-06,
      "loss": 0.6024,
      "step": 8039
    },
    {
      "epoch": 0.4338675732556257,
      "grad_norm": 0.9400693565634222,
      "learning_rate": 6.42821369871025e-06,
      "loss": 0.4955,
      "step": 8040
    },
    {
      "epoch": 0.4339215368841401,
      "grad_norm": 1.0786392721469908,
      "learning_rate": 6.427467198426993e-06,
      "loss": 0.5728,
      "step": 8041
    },
    {
      "epoch": 0.4339755005126545,
      "grad_norm": 1.1366709880506471,
      "learning_rate": 6.426720671487392e-06,
      "loss": 0.6078,
      "step": 8042
    },
    {
      "epoch": 0.43402946414116883,
      "grad_norm": 1.0084031768941273,
      "learning_rate": 6.425974117912904e-06,
      "loss": 0.4871,
      "step": 8043
    },
    {
      "epoch": 0.4340834277696832,
      "grad_norm": 1.1338140760266902,
      "learning_rate": 6.4252275377249824e-06,
      "loss": 0.5438,
      "step": 8044
    },
    {
      "epoch": 0.4341373913981976,
      "grad_norm": 1.0537999960133295,
      "learning_rate": 6.424480930945087e-06,
      "loss": 0.5842,
      "step": 8045
    },
    {
      "epoch": 0.434191355026712,
      "grad_norm": 1.0903483285334878,
      "learning_rate": 6.423734297594676e-06,
      "loss": 0.4026,
      "step": 8046
    },
    {
      "epoch": 0.43424531865522636,
      "grad_norm": 1.012008774465215,
      "learning_rate": 6.422987637695208e-06,
      "loss": 0.4515,
      "step": 8047
    },
    {
      "epoch": 0.43429928228374076,
      "grad_norm": 1.1304014399621816,
      "learning_rate": 6.422240951268144e-06,
      "loss": 0.4715,
      "step": 8048
    },
    {
      "epoch": 0.43435324591225516,
      "grad_norm": 1.064400887053191,
      "learning_rate": 6.42149423833494e-06,
      "loss": 0.4188,
      "step": 8049
    },
    {
      "epoch": 0.4344072095407695,
      "grad_norm": 1.149042268668801,
      "learning_rate": 6.4207474989170634e-06,
      "loss": 0.4762,
      "step": 8050
    },
    {
      "epoch": 0.4344611731692839,
      "grad_norm": 1.1475140499568488,
      "learning_rate": 6.420000733035973e-06,
      "loss": 0.4585,
      "step": 8051
    },
    {
      "epoch": 0.4345151367977983,
      "grad_norm": 1.083605079893737,
      "learning_rate": 6.41925394071313e-06,
      "loss": 0.5235,
      "step": 8052
    },
    {
      "epoch": 0.43456910042631264,
      "grad_norm": 1.033835674097321,
      "learning_rate": 6.4185071219700005e-06,
      "loss": 0.5947,
      "step": 8053
    },
    {
      "epoch": 0.43462306405482704,
      "grad_norm": 0.8582916096244454,
      "learning_rate": 6.417760276828048e-06,
      "loss": 0.393,
      "step": 8054
    },
    {
      "epoch": 0.43467702768334143,
      "grad_norm": 1.0419185696017907,
      "learning_rate": 6.417013405308738e-06,
      "loss": 0.3591,
      "step": 8055
    },
    {
      "epoch": 0.43473099131185583,
      "grad_norm": 1.0709311587178505,
      "learning_rate": 6.416266507433537e-06,
      "loss": 0.4724,
      "step": 8056
    },
    {
      "epoch": 0.43478495494037017,
      "grad_norm": 1.074971733883941,
      "learning_rate": 6.415519583223912e-06,
      "loss": 0.5165,
      "step": 8057
    },
    {
      "epoch": 0.43483891856888457,
      "grad_norm": 1.1353685124645925,
      "learning_rate": 6.414772632701326e-06,
      "loss": 0.591,
      "step": 8058
    },
    {
      "epoch": 0.43489288219739897,
      "grad_norm": 1.1320142978827112,
      "learning_rate": 6.414025655887251e-06,
      "loss": 0.6093,
      "step": 8059
    },
    {
      "epoch": 0.4349468458259133,
      "grad_norm": 0.9892705211490621,
      "learning_rate": 6.413278652803158e-06,
      "loss": 0.4913,
      "step": 8060
    },
    {
      "epoch": 0.4350008094544277,
      "grad_norm": 0.9529040242943534,
      "learning_rate": 6.412531623470512e-06,
      "loss": 0.4892,
      "step": 8061
    },
    {
      "epoch": 0.4350547730829421,
      "grad_norm": 0.8702550960721849,
      "learning_rate": 6.411784567910783e-06,
      "loss": 0.388,
      "step": 8062
    },
    {
      "epoch": 0.4351087367114565,
      "grad_norm": 0.8945978521745916,
      "learning_rate": 6.411037486145446e-06,
      "loss": 0.3707,
      "step": 8063
    },
    {
      "epoch": 0.43516270033997084,
      "grad_norm": 1.0473644185786157,
      "learning_rate": 6.410290378195971e-06,
      "loss": 0.5794,
      "step": 8064
    },
    {
      "epoch": 0.43521666396848524,
      "grad_norm": 1.288300269468444,
      "learning_rate": 6.409543244083831e-06,
      "loss": 0.6389,
      "step": 8065
    },
    {
      "epoch": 0.43527062759699964,
      "grad_norm": 1.042710238895446,
      "learning_rate": 6.4087960838304995e-06,
      "loss": 0.459,
      "step": 8066
    },
    {
      "epoch": 0.435324591225514,
      "grad_norm": 0.9528224469913216,
      "learning_rate": 6.40804889745745e-06,
      "loss": 0.3872,
      "step": 8067
    },
    {
      "epoch": 0.4353785548540284,
      "grad_norm": 1.0420034195641645,
      "learning_rate": 6.407301684986157e-06,
      "loss": 0.4581,
      "step": 8068
    },
    {
      "epoch": 0.4354325184825428,
      "grad_norm": 1.030198547140232,
      "learning_rate": 6.406554446438097e-06,
      "loss": 0.4401,
      "step": 8069
    },
    {
      "epoch": 0.43548648211105717,
      "grad_norm": 1.082601079448098,
      "learning_rate": 6.405807181834747e-06,
      "loss": 0.5042,
      "step": 8070
    },
    {
      "epoch": 0.4355404457395715,
      "grad_norm": 1.163902238198311,
      "learning_rate": 6.4050598911975825e-06,
      "loss": 0.6521,
      "step": 8071
    },
    {
      "epoch": 0.4355944093680859,
      "grad_norm": 0.916552049491407,
      "learning_rate": 6.404312574548083e-06,
      "loss": 0.4371,
      "step": 8072
    },
    {
      "epoch": 0.4356483729966003,
      "grad_norm": 1.0182016372993206,
      "learning_rate": 6.403565231907724e-06,
      "loss": 0.564,
      "step": 8073
    },
    {
      "epoch": 0.43570233662511465,
      "grad_norm": 1.12107038092545,
      "learning_rate": 6.402817863297991e-06,
      "loss": 0.4994,
      "step": 8074
    },
    {
      "epoch": 0.43575630025362905,
      "grad_norm": 1.0470176871027743,
      "learning_rate": 6.402070468740358e-06,
      "loss": 0.4609,
      "step": 8075
    },
    {
      "epoch": 0.43581026388214344,
      "grad_norm": 1.0647090637761971,
      "learning_rate": 6.4013230482563086e-06,
      "loss": 0.653,
      "step": 8076
    },
    {
      "epoch": 0.43586422751065784,
      "grad_norm": 0.8831818413182673,
      "learning_rate": 6.400575601867323e-06,
      "loss": 0.4519,
      "step": 8077
    },
    {
      "epoch": 0.4359181911391722,
      "grad_norm": 0.8472336573540319,
      "learning_rate": 6.3998281295948865e-06,
      "loss": 0.3507,
      "step": 8078
    },
    {
      "epoch": 0.4359721547676866,
      "grad_norm": 1.068479235534307,
      "learning_rate": 6.399080631460479e-06,
      "loss": 0.5811,
      "step": 8079
    },
    {
      "epoch": 0.436026118396201,
      "grad_norm": 1.0104117912516317,
      "learning_rate": 6.398333107485587e-06,
      "loss": 0.5943,
      "step": 8080
    },
    {
      "epoch": 0.4360800820247153,
      "grad_norm": 1.0608152851224375,
      "learning_rate": 6.3975855576916934e-06,
      "loss": 0.6378,
      "step": 8081
    },
    {
      "epoch": 0.4361340456532297,
      "grad_norm": 0.9965171404627531,
      "learning_rate": 6.396837982100284e-06,
      "loss": 0.483,
      "step": 8082
    },
    {
      "epoch": 0.4361880092817441,
      "grad_norm": 0.9961876180060363,
      "learning_rate": 6.3960903807328435e-06,
      "loss": 0.5455,
      "step": 8083
    },
    {
      "epoch": 0.4362419729102585,
      "grad_norm": 1.0989478057070279,
      "learning_rate": 6.395342753610861e-06,
      "loss": 0.4914,
      "step": 8084
    },
    {
      "epoch": 0.43629593653877285,
      "grad_norm": 0.8721894517490578,
      "learning_rate": 6.394595100755824e-06,
      "loss": 0.4328,
      "step": 8085
    },
    {
      "epoch": 0.43634990016728725,
      "grad_norm": 1.098108333041575,
      "learning_rate": 6.393847422189218e-06,
      "loss": 0.5682,
      "step": 8086
    },
    {
      "epoch": 0.43640386379580165,
      "grad_norm": 0.8411764767745809,
      "learning_rate": 6.3930997179325345e-06,
      "loss": 0.4709,
      "step": 8087
    },
    {
      "epoch": 0.436457827424316,
      "grad_norm": 0.9541046927701743,
      "learning_rate": 6.392351988007264e-06,
      "loss": 0.4826,
      "step": 8088
    },
    {
      "epoch": 0.4365117910528304,
      "grad_norm": 0.9456780092062623,
      "learning_rate": 6.391604232434895e-06,
      "loss": 0.4175,
      "step": 8089
    },
    {
      "epoch": 0.4365657546813448,
      "grad_norm": 1.1913724530178627,
      "learning_rate": 6.390856451236917e-06,
      "loss": 0.6237,
      "step": 8090
    },
    {
      "epoch": 0.43661971830985913,
      "grad_norm": 1.1821435120283528,
      "learning_rate": 6.390108644434828e-06,
      "loss": 0.48,
      "step": 8091
    },
    {
      "epoch": 0.4366736819383735,
      "grad_norm": 1.0206977484768158,
      "learning_rate": 6.389360812050117e-06,
      "loss": 0.4698,
      "step": 8092
    },
    {
      "epoch": 0.4367276455668879,
      "grad_norm": 1.165265685388421,
      "learning_rate": 6.388612954104276e-06,
      "loss": 0.6669,
      "step": 8093
    },
    {
      "epoch": 0.4367816091954023,
      "grad_norm": 1.3769500238470522,
      "learning_rate": 6.387865070618801e-06,
      "loss": 0.7257,
      "step": 8094
    },
    {
      "epoch": 0.43683557282391666,
      "grad_norm": 1.0852223288059764,
      "learning_rate": 6.3871171616151874e-06,
      "loss": 0.4733,
      "step": 8095
    },
    {
      "epoch": 0.43688953645243106,
      "grad_norm": 0.9689878733323666,
      "learning_rate": 6.38636922711493e-06,
      "loss": 0.3621,
      "step": 8096
    },
    {
      "epoch": 0.43694350008094546,
      "grad_norm": 1.0001688379111937,
      "learning_rate": 6.385621267139525e-06,
      "loss": 0.4352,
      "step": 8097
    },
    {
      "epoch": 0.4369974637094598,
      "grad_norm": 1.0424637729186745,
      "learning_rate": 6.3848732817104705e-06,
      "loss": 0.5924,
      "step": 8098
    },
    {
      "epoch": 0.4370514273379742,
      "grad_norm": 1.200248853860949,
      "learning_rate": 6.384125270849263e-06,
      "loss": 0.5096,
      "step": 8099
    },
    {
      "epoch": 0.4371053909664886,
      "grad_norm": 1.028276266773512,
      "learning_rate": 6.383377234577403e-06,
      "loss": 0.5292,
      "step": 8100
    },
    {
      "epoch": 0.437159354595003,
      "grad_norm": 0.9455452204419401,
      "learning_rate": 6.382629172916388e-06,
      "loss": 0.4091,
      "step": 8101
    },
    {
      "epoch": 0.43721331822351733,
      "grad_norm": 1.1147202392381956,
      "learning_rate": 6.381881085887719e-06,
      "loss": 0.6211,
      "step": 8102
    },
    {
      "epoch": 0.43726728185203173,
      "grad_norm": 1.071053832735288,
      "learning_rate": 6.381132973512896e-06,
      "loss": 0.5623,
      "step": 8103
    },
    {
      "epoch": 0.43732124548054613,
      "grad_norm": 1.0535799831939132,
      "learning_rate": 6.380384835813421e-06,
      "loss": 0.4967,
      "step": 8104
    },
    {
      "epoch": 0.43737520910906047,
      "grad_norm": 0.9339735034567102,
      "learning_rate": 6.379636672810796e-06,
      "loss": 0.457,
      "step": 8105
    },
    {
      "epoch": 0.43742917273757487,
      "grad_norm": 0.8375217093843261,
      "learning_rate": 6.378888484526525e-06,
      "loss": 0.3512,
      "step": 8106
    },
    {
      "epoch": 0.43748313636608926,
      "grad_norm": 0.9654754565033535,
      "learning_rate": 6.378140270982111e-06,
      "loss": 0.4298,
      "step": 8107
    },
    {
      "epoch": 0.43753709999460366,
      "grad_norm": 1.0205784661289765,
      "learning_rate": 6.377392032199057e-06,
      "loss": 0.3585,
      "step": 8108
    },
    {
      "epoch": 0.437591063623118,
      "grad_norm": 0.9327675162870027,
      "learning_rate": 6.376643768198871e-06,
      "loss": 0.5,
      "step": 8109
    },
    {
      "epoch": 0.4376450272516324,
      "grad_norm": 0.9970830839253926,
      "learning_rate": 6.3758954790030565e-06,
      "loss": 0.4923,
      "step": 8110
    },
    {
      "epoch": 0.4376989908801468,
      "grad_norm": 1.1855793239058616,
      "learning_rate": 6.375147164633121e-06,
      "loss": 0.6608,
      "step": 8111
    },
    {
      "epoch": 0.43775295450866114,
      "grad_norm": 0.9869786164084325,
      "learning_rate": 6.374398825110572e-06,
      "loss": 0.4502,
      "step": 8112
    },
    {
      "epoch": 0.43780691813717554,
      "grad_norm": 0.9700881786400077,
      "learning_rate": 6.373650460456918e-06,
      "loss": 0.4897,
      "step": 8113
    },
    {
      "epoch": 0.43786088176568994,
      "grad_norm": 1.0699086688969823,
      "learning_rate": 6.372902070693666e-06,
      "loss": 0.5657,
      "step": 8114
    },
    {
      "epoch": 0.43791484539420433,
      "grad_norm": 0.9806957932412118,
      "learning_rate": 6.372153655842326e-06,
      "loss": 0.4564,
      "step": 8115
    },
    {
      "epoch": 0.4379688090227187,
      "grad_norm": 1.1931456001925733,
      "learning_rate": 6.37140521592441e-06,
      "loss": 0.5545,
      "step": 8116
    },
    {
      "epoch": 0.43802277265123307,
      "grad_norm": 0.757168779368349,
      "learning_rate": 6.370656750961427e-06,
      "loss": 0.31,
      "step": 8117
    },
    {
      "epoch": 0.43807673627974747,
      "grad_norm": 0.9856305847307492,
      "learning_rate": 6.369908260974889e-06,
      "loss": 0.3844,
      "step": 8118
    },
    {
      "epoch": 0.4381306999082618,
      "grad_norm": 0.8907570899339623,
      "learning_rate": 6.369159745986309e-06,
      "loss": 0.5752,
      "step": 8119
    },
    {
      "epoch": 0.4381846635367762,
      "grad_norm": 0.9880571819847156,
      "learning_rate": 6.368411206017199e-06,
      "loss": 0.6538,
      "step": 8120
    },
    {
      "epoch": 0.4382386271652906,
      "grad_norm": 0.9933601646139172,
      "learning_rate": 6.367662641089074e-06,
      "loss": 0.4543,
      "step": 8121
    },
    {
      "epoch": 0.43829259079380495,
      "grad_norm": 1.0133913158345669,
      "learning_rate": 6.366914051223449e-06,
      "loss": 0.4424,
      "step": 8122
    },
    {
      "epoch": 0.43834655442231935,
      "grad_norm": 1.1396161296497513,
      "learning_rate": 6.366165436441836e-06,
      "loss": 0.5371,
      "step": 8123
    },
    {
      "epoch": 0.43840051805083374,
      "grad_norm": 1.0524884732792306,
      "learning_rate": 6.365416796765755e-06,
      "loss": 0.5011,
      "step": 8124
    },
    {
      "epoch": 0.43845448167934814,
      "grad_norm": 1.1626336410302738,
      "learning_rate": 6.3646681322167196e-06,
      "loss": 0.559,
      "step": 8125
    },
    {
      "epoch": 0.4385084453078625,
      "grad_norm": 1.1335010796602056,
      "learning_rate": 6.36391944281625e-06,
      "loss": 0.5815,
      "step": 8126
    },
    {
      "epoch": 0.4385624089363769,
      "grad_norm": 0.9666629066518239,
      "learning_rate": 6.3631707285858605e-06,
      "loss": 0.4598,
      "step": 8127
    },
    {
      "epoch": 0.4386163725648913,
      "grad_norm": 1.0461815096795528,
      "learning_rate": 6.362421989547073e-06,
      "loss": 0.4925,
      "step": 8128
    },
    {
      "epoch": 0.4386703361934056,
      "grad_norm": 1.0858235241843988,
      "learning_rate": 6.361673225721406e-06,
      "loss": 0.5601,
      "step": 8129
    },
    {
      "epoch": 0.43872429982192,
      "grad_norm": 0.8876433346724394,
      "learning_rate": 6.36092443713038e-06,
      "loss": 0.3838,
      "step": 8130
    },
    {
      "epoch": 0.4387782634504344,
      "grad_norm": 1.0780156616152339,
      "learning_rate": 6.360175623795516e-06,
      "loss": 0.5177,
      "step": 8131
    },
    {
      "epoch": 0.4388322270789488,
      "grad_norm": 0.9840045639439228,
      "learning_rate": 6.359426785738334e-06,
      "loss": 0.4199,
      "step": 8132
    },
    {
      "epoch": 0.43888619070746315,
      "grad_norm": 1.123332377726366,
      "learning_rate": 6.358677922980361e-06,
      "loss": 0.4749,
      "step": 8133
    },
    {
      "epoch": 0.43894015433597755,
      "grad_norm": 0.8783892431051276,
      "learning_rate": 6.357929035543116e-06,
      "loss": 0.3921,
      "step": 8134
    },
    {
      "epoch": 0.43899411796449195,
      "grad_norm": 1.034136451834459,
      "learning_rate": 6.357180123448121e-06,
      "loss": 0.4977,
      "step": 8135
    },
    {
      "epoch": 0.4390480815930063,
      "grad_norm": 0.9879846902445798,
      "learning_rate": 6.356431186716905e-06,
      "loss": 0.5931,
      "step": 8136
    },
    {
      "epoch": 0.4391020452215207,
      "grad_norm": 0.8512448211390865,
      "learning_rate": 6.355682225370991e-06,
      "loss": 0.3508,
      "step": 8137
    },
    {
      "epoch": 0.4391560088500351,
      "grad_norm": 0.8924670967537378,
      "learning_rate": 6.354933239431905e-06,
      "loss": 0.5193,
      "step": 8138
    },
    {
      "epoch": 0.4392099724785495,
      "grad_norm": 1.0896040941650025,
      "learning_rate": 6.354184228921174e-06,
      "loss": 0.524,
      "step": 8139
    },
    {
      "epoch": 0.4392639361070638,
      "grad_norm": 1.1589943101813422,
      "learning_rate": 6.353435193860326e-06,
      "loss": 0.5319,
      "step": 8140
    },
    {
      "epoch": 0.4393178997355782,
      "grad_norm": 0.8871687377890529,
      "learning_rate": 6.352686134270888e-06,
      "loss": 0.4259,
      "step": 8141
    },
    {
      "epoch": 0.4393718633640926,
      "grad_norm": 0.7771725599384475,
      "learning_rate": 6.3519370501743885e-06,
      "loss": 0.3838,
      "step": 8142
    },
    {
      "epoch": 0.43942582699260696,
      "grad_norm": 1.200714298117275,
      "learning_rate": 6.3511879415923586e-06,
      "loss": 0.64,
      "step": 8143
    },
    {
      "epoch": 0.43947979062112136,
      "grad_norm": 1.047937932962672,
      "learning_rate": 6.350438808546324e-06,
      "loss": 0.5083,
      "step": 8144
    },
    {
      "epoch": 0.43953375424963576,
      "grad_norm": 1.1978782473578398,
      "learning_rate": 6.349689651057821e-06,
      "loss": 0.3857,
      "step": 8145
    },
    {
      "epoch": 0.43958771787815015,
      "grad_norm": 1.2029055049494446,
      "learning_rate": 6.3489404691483795e-06,
      "loss": 0.6547,
      "step": 8146
    },
    {
      "epoch": 0.4396416815066645,
      "grad_norm": 0.9581872662538926,
      "learning_rate": 6.348191262839529e-06,
      "loss": 0.3754,
      "step": 8147
    },
    {
      "epoch": 0.4396956451351789,
      "grad_norm": 1.083073680084396,
      "learning_rate": 6.347442032152806e-06,
      "loss": 0.4862,
      "step": 8148
    },
    {
      "epoch": 0.4397496087636933,
      "grad_norm": 1.1286799181299354,
      "learning_rate": 6.346692777109742e-06,
      "loss": 0.4204,
      "step": 8149
    },
    {
      "epoch": 0.43980357239220763,
      "grad_norm": 0.9743331328771458,
      "learning_rate": 6.345943497731873e-06,
      "loss": 0.3359,
      "step": 8150
    },
    {
      "epoch": 0.43985753602072203,
      "grad_norm": 0.9023054104083283,
      "learning_rate": 6.345194194040733e-06,
      "loss": 0.3835,
      "step": 8151
    },
    {
      "epoch": 0.4399114996492364,
      "grad_norm": 1.1186681761070714,
      "learning_rate": 6.344444866057857e-06,
      "loss": 0.5416,
      "step": 8152
    },
    {
      "epoch": 0.4399654632777508,
      "grad_norm": 1.2603636586906632,
      "learning_rate": 6.343695513804783e-06,
      "loss": 0.5895,
      "step": 8153
    },
    {
      "epoch": 0.44001942690626517,
      "grad_norm": 0.8864033774209207,
      "learning_rate": 6.342946137303046e-06,
      "loss": 0.5217,
      "step": 8154
    },
    {
      "epoch": 0.44007339053477956,
      "grad_norm": 0.9497093675618014,
      "learning_rate": 6.342196736574188e-06,
      "loss": 0.4329,
      "step": 8155
    },
    {
      "epoch": 0.44012735416329396,
      "grad_norm": 0.9765983099653647,
      "learning_rate": 6.341447311639744e-06,
      "loss": 0.5436,
      "step": 8156
    },
    {
      "epoch": 0.4401813177918083,
      "grad_norm": 1.1042772422743248,
      "learning_rate": 6.340697862521252e-06,
      "loss": 0.4615,
      "step": 8157
    },
    {
      "epoch": 0.4402352814203227,
      "grad_norm": 0.9380930829445366,
      "learning_rate": 6.339948389240256e-06,
      "loss": 0.3542,
      "step": 8158
    },
    {
      "epoch": 0.4402892450488371,
      "grad_norm": 1.1455537076511488,
      "learning_rate": 6.339198891818294e-06,
      "loss": 0.5423,
      "step": 8159
    },
    {
      "epoch": 0.44034320867735144,
      "grad_norm": 1.1138363715485986,
      "learning_rate": 6.338449370276908e-06,
      "loss": 0.6028,
      "step": 8160
    },
    {
      "epoch": 0.44039717230586584,
      "grad_norm": 0.8219761166196592,
      "learning_rate": 6.3376998246376385e-06,
      "loss": 0.3883,
      "step": 8161
    },
    {
      "epoch": 0.44045113593438023,
      "grad_norm": 1.2886158321449712,
      "learning_rate": 6.336950254922032e-06,
      "loss": 0.6709,
      "step": 8162
    },
    {
      "epoch": 0.44050509956289463,
      "grad_norm": 1.1059109511587166,
      "learning_rate": 6.336200661151627e-06,
      "loss": 0.5769,
      "step": 8163
    },
    {
      "epoch": 0.440559063191409,
      "grad_norm": 1.0291512470794295,
      "learning_rate": 6.335451043347972e-06,
      "loss": 0.3985,
      "step": 8164
    },
    {
      "epoch": 0.44061302681992337,
      "grad_norm": 0.924734911999811,
      "learning_rate": 6.334701401532609e-06,
      "loss": 0.4196,
      "step": 8165
    },
    {
      "epoch": 0.44066699044843777,
      "grad_norm": 1.039201638852191,
      "learning_rate": 6.333951735727085e-06,
      "loss": 0.5746,
      "step": 8166
    },
    {
      "epoch": 0.4407209540769521,
      "grad_norm": 1.0363687422139771,
      "learning_rate": 6.333202045952945e-06,
      "loss": 0.4912,
      "step": 8167
    },
    {
      "epoch": 0.4407749177054665,
      "grad_norm": 1.1160911142843102,
      "learning_rate": 6.332452332231737e-06,
      "loss": 0.5686,
      "step": 8168
    },
    {
      "epoch": 0.4408288813339809,
      "grad_norm": 0.9348136448116612,
      "learning_rate": 6.3317025945850064e-06,
      "loss": 0.3763,
      "step": 8169
    },
    {
      "epoch": 0.4408828449624953,
      "grad_norm": 0.8627989511814855,
      "learning_rate": 6.330952833034304e-06,
      "loss": 0.4318,
      "step": 8170
    },
    {
      "epoch": 0.44093680859100964,
      "grad_norm": 0.9499500987728281,
      "learning_rate": 6.330203047601178e-06,
      "loss": 0.5124,
      "step": 8171
    },
    {
      "epoch": 0.44099077221952404,
      "grad_norm": 0.9239559119230113,
      "learning_rate": 6.329453238307176e-06,
      "loss": 0.3238,
      "step": 8172
    },
    {
      "epoch": 0.44104473584803844,
      "grad_norm": 0.9714223164850403,
      "learning_rate": 6.328703405173851e-06,
      "loss": 0.3856,
      "step": 8173
    },
    {
      "epoch": 0.4410986994765528,
      "grad_norm": 1.2146435155933755,
      "learning_rate": 6.327953548222753e-06,
      "loss": 0.7072,
      "step": 8174
    },
    {
      "epoch": 0.4411526631050672,
      "grad_norm": 1.0192817279028203,
      "learning_rate": 6.327203667475432e-06,
      "loss": 0.5503,
      "step": 8175
    },
    {
      "epoch": 0.4412066267335816,
      "grad_norm": 0.807412687359062,
      "learning_rate": 6.326453762953442e-06,
      "loss": 0.4799,
      "step": 8176
    },
    {
      "epoch": 0.441260590362096,
      "grad_norm": 0.9864247211217492,
      "learning_rate": 6.3257038346783374e-06,
      "loss": 0.459,
      "step": 8177
    },
    {
      "epoch": 0.4413145539906103,
      "grad_norm": 1.0599515214229398,
      "learning_rate": 6.32495388267167e-06,
      "loss": 0.4568,
      "step": 8178
    },
    {
      "epoch": 0.4413685176191247,
      "grad_norm": 1.023233013595788,
      "learning_rate": 6.324203906954993e-06,
      "loss": 0.506,
      "step": 8179
    },
    {
      "epoch": 0.4414224812476391,
      "grad_norm": 0.9538386719147528,
      "learning_rate": 6.323453907549864e-06,
      "loss": 0.3746,
      "step": 8180
    },
    {
      "epoch": 0.44147644487615345,
      "grad_norm": 0.8962383666827701,
      "learning_rate": 6.322703884477836e-06,
      "loss": 0.5594,
      "step": 8181
    },
    {
      "epoch": 0.44153040850466785,
      "grad_norm": 0.8091687386812749,
      "learning_rate": 6.3219538377604695e-06,
      "loss": 0.3542,
      "step": 8182
    },
    {
      "epoch": 0.44158437213318225,
      "grad_norm": 1.0838700672742936,
      "learning_rate": 6.321203767419317e-06,
      "loss": 0.6996,
      "step": 8183
    },
    {
      "epoch": 0.44163833576169664,
      "grad_norm": 0.9145949421530559,
      "learning_rate": 6.320453673475939e-06,
      "loss": 0.3333,
      "step": 8184
    },
    {
      "epoch": 0.441692299390211,
      "grad_norm": 0.9877080777036605,
      "learning_rate": 6.319703555951895e-06,
      "loss": 0.4811,
      "step": 8185
    },
    {
      "epoch": 0.4417462630187254,
      "grad_norm": 1.2919528204982245,
      "learning_rate": 6.3189534148687405e-06,
      "loss": 0.4917,
      "step": 8186
    },
    {
      "epoch": 0.4418002266472398,
      "grad_norm": 0.9677924561092979,
      "learning_rate": 6.318203250248038e-06,
      "loss": 0.4048,
      "step": 8187
    },
    {
      "epoch": 0.4418541902757541,
      "grad_norm": 1.1427039498956448,
      "learning_rate": 6.317453062111349e-06,
      "loss": 0.5838,
      "step": 8188
    },
    {
      "epoch": 0.4419081539042685,
      "grad_norm": 1.0162853819867161,
      "learning_rate": 6.31670285048023e-06,
      "loss": 0.4848,
      "step": 8189
    },
    {
      "epoch": 0.4419621175327829,
      "grad_norm": 1.118976014419981,
      "learning_rate": 6.3159526153762486e-06,
      "loss": 0.4782,
      "step": 8190
    },
    {
      "epoch": 0.44201608116129726,
      "grad_norm": 1.0351736930399245,
      "learning_rate": 6.3152023568209644e-06,
      "loss": 0.4111,
      "step": 8191
    },
    {
      "epoch": 0.44207004478981166,
      "grad_norm": 1.0442399579263657,
      "learning_rate": 6.314452074835939e-06,
      "loss": 0.4251,
      "step": 8192
    },
    {
      "epoch": 0.44212400841832605,
      "grad_norm": 1.213416172423367,
      "learning_rate": 6.313701769442739e-06,
      "loss": 0.5543,
      "step": 8193
    },
    {
      "epoch": 0.44217797204684045,
      "grad_norm": 1.0284956045707423,
      "learning_rate": 6.312951440662929e-06,
      "loss": 0.3544,
      "step": 8194
    },
    {
      "epoch": 0.4422319356753548,
      "grad_norm": 0.987123132657458,
      "learning_rate": 6.312201088518072e-06,
      "loss": 0.4435,
      "step": 8195
    },
    {
      "epoch": 0.4422858993038692,
      "grad_norm": 1.2694454068825594,
      "learning_rate": 6.311450713029735e-06,
      "loss": 0.587,
      "step": 8196
    },
    {
      "epoch": 0.4423398629323836,
      "grad_norm": 0.8288299916585606,
      "learning_rate": 6.3107003142194845e-06,
      "loss": 0.3729,
      "step": 8197
    },
    {
      "epoch": 0.44239382656089793,
      "grad_norm": 0.9023748236410828,
      "learning_rate": 6.309949892108888e-06,
      "loss": 0.3403,
      "step": 8198
    },
    {
      "epoch": 0.4424477901894123,
      "grad_norm": 1.0731385703159089,
      "learning_rate": 6.309199446719514e-06,
      "loss": 0.4381,
      "step": 8199
    },
    {
      "epoch": 0.4425017538179267,
      "grad_norm": 0.9221213853348443,
      "learning_rate": 6.308448978072929e-06,
      "loss": 0.373,
      "step": 8200
    },
    {
      "epoch": 0.4425557174464411,
      "grad_norm": 0.9584848058545735,
      "learning_rate": 6.307698486190703e-06,
      "loss": 0.4064,
      "step": 8201
    },
    {
      "epoch": 0.44260968107495546,
      "grad_norm": 1.0326958618881306,
      "learning_rate": 6.306947971094408e-06,
      "loss": 0.4916,
      "step": 8202
    },
    {
      "epoch": 0.44266364470346986,
      "grad_norm": 1.0361854963113555,
      "learning_rate": 6.306197432805613e-06,
      "loss": 0.6035,
      "step": 8203
    },
    {
      "epoch": 0.44271760833198426,
      "grad_norm": 1.1644927444367013,
      "learning_rate": 6.305446871345887e-06,
      "loss": 0.4753,
      "step": 8204
    },
    {
      "epoch": 0.4427715719604986,
      "grad_norm": 1.0237868915227617,
      "learning_rate": 6.304696286736806e-06,
      "loss": 0.4134,
      "step": 8205
    },
    {
      "epoch": 0.442825535589013,
      "grad_norm": 1.0892502066123266,
      "learning_rate": 6.30394567899994e-06,
      "loss": 0.6368,
      "step": 8206
    },
    {
      "epoch": 0.4428794992175274,
      "grad_norm": 0.8741782152441725,
      "learning_rate": 6.303195048156864e-06,
      "loss": 0.4734,
      "step": 8207
    },
    {
      "epoch": 0.4429334628460418,
      "grad_norm": 0.8381479574640173,
      "learning_rate": 6.30244439422915e-06,
      "loss": 0.2939,
      "step": 8208
    },
    {
      "epoch": 0.44298742647455613,
      "grad_norm": 1.09090737181508,
      "learning_rate": 6.301693717238373e-06,
      "loss": 0.4413,
      "step": 8209
    },
    {
      "epoch": 0.44304139010307053,
      "grad_norm": 1.054117729719635,
      "learning_rate": 6.300943017206108e-06,
      "loss": 0.4736,
      "step": 8210
    },
    {
      "epoch": 0.44309535373158493,
      "grad_norm": 1.1298269340883949,
      "learning_rate": 6.300192294153932e-06,
      "loss": 0.4254,
      "step": 8211
    },
    {
      "epoch": 0.44314931736009927,
      "grad_norm": 1.0717817649100072,
      "learning_rate": 6.299441548103421e-06,
      "loss": 0.5625,
      "step": 8212
    },
    {
      "epoch": 0.44320328098861367,
      "grad_norm": 0.9637728890140532,
      "learning_rate": 6.298690779076153e-06,
      "loss": 0.3913,
      "step": 8213
    },
    {
      "epoch": 0.44325724461712807,
      "grad_norm": 0.8325067726490245,
      "learning_rate": 6.297939987093703e-06,
      "loss": 0.2968,
      "step": 8214
    },
    {
      "epoch": 0.44331120824564246,
      "grad_norm": 1.154098914995121,
      "learning_rate": 6.297189172177652e-06,
      "loss": 0.5368,
      "step": 8215
    },
    {
      "epoch": 0.4433651718741568,
      "grad_norm": 1.1915841719540385,
      "learning_rate": 6.2964383343495806e-06,
      "loss": 0.4341,
      "step": 8216
    },
    {
      "epoch": 0.4434191355026712,
      "grad_norm": 0.9890774317884556,
      "learning_rate": 6.295687473631066e-06,
      "loss": 0.634,
      "step": 8217
    },
    {
      "epoch": 0.4434730991311856,
      "grad_norm": 0.9752077991895186,
      "learning_rate": 6.294936590043688e-06,
      "loss": 0.4282,
      "step": 8218
    },
    {
      "epoch": 0.44352706275969994,
      "grad_norm": 1.1067859025873192,
      "learning_rate": 6.29418568360903e-06,
      "loss": 0.4941,
      "step": 8219
    },
    {
      "epoch": 0.44358102638821434,
      "grad_norm": 1.1825300788237498,
      "learning_rate": 6.293434754348673e-06,
      "loss": 0.5656,
      "step": 8220
    },
    {
      "epoch": 0.44363499001672874,
      "grad_norm": 0.8317822048245727,
      "learning_rate": 6.292683802284199e-06,
      "loss": 0.3694,
      "step": 8221
    },
    {
      "epoch": 0.44368895364524313,
      "grad_norm": 0.8759998119374629,
      "learning_rate": 6.291932827437191e-06,
      "loss": 0.4113,
      "step": 8222
    },
    {
      "epoch": 0.4437429172737575,
      "grad_norm": 0.8129601882371275,
      "learning_rate": 6.291181829829236e-06,
      "loss": 0.3685,
      "step": 8223
    },
    {
      "epoch": 0.4437968809022719,
      "grad_norm": 0.7864725287095008,
      "learning_rate": 6.2904308094819145e-06,
      "loss": 0.2948,
      "step": 8224
    },
    {
      "epoch": 0.44385084453078627,
      "grad_norm": 0.954843239468029,
      "learning_rate": 6.289679766416812e-06,
      "loss": 0.3572,
      "step": 8225
    },
    {
      "epoch": 0.4439048081593006,
      "grad_norm": 1.1197378145195935,
      "learning_rate": 6.288928700655516e-06,
      "loss": 0.4824,
      "step": 8226
    },
    {
      "epoch": 0.443958771787815,
      "grad_norm": 1.1840832461807584,
      "learning_rate": 6.288177612219612e-06,
      "loss": 0.6109,
      "step": 8227
    },
    {
      "epoch": 0.4440127354163294,
      "grad_norm": 1.0799232856004923,
      "learning_rate": 6.287426501130686e-06,
      "loss": 0.5627,
      "step": 8228
    },
    {
      "epoch": 0.44406669904484375,
      "grad_norm": 1.4080392936876376,
      "learning_rate": 6.286675367410329e-06,
      "loss": 0.7502,
      "step": 8229
    },
    {
      "epoch": 0.44412066267335815,
      "grad_norm": 1.0806069744294307,
      "learning_rate": 6.285924211080124e-06,
      "loss": 0.4956,
      "step": 8230
    },
    {
      "epoch": 0.44417462630187254,
      "grad_norm": 0.7834499491958148,
      "learning_rate": 6.285173032161665e-06,
      "loss": 0.3141,
      "step": 8231
    },
    {
      "epoch": 0.44422858993038694,
      "grad_norm": 1.0866560282979345,
      "learning_rate": 6.284421830676538e-06,
      "loss": 0.4347,
      "step": 8232
    },
    {
      "epoch": 0.4442825535589013,
      "grad_norm": 0.9480702697277158,
      "learning_rate": 6.283670606646336e-06,
      "loss": 0.4358,
      "step": 8233
    },
    {
      "epoch": 0.4443365171874157,
      "grad_norm": 0.9220059359209927,
      "learning_rate": 6.282919360092649e-06,
      "loss": 0.4317,
      "step": 8234
    },
    {
      "epoch": 0.4443904808159301,
      "grad_norm": 1.1988366135052093,
      "learning_rate": 6.282168091037067e-06,
      "loss": 0.669,
      "step": 8235
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 1.1111110270476083,
      "learning_rate": 6.281416799501188e-06,
      "loss": 0.5348,
      "step": 8236
    },
    {
      "epoch": 0.4444984080729588,
      "grad_norm": 1.0270444262977099,
      "learning_rate": 6.280665485506596e-06,
      "loss": 0.4979,
      "step": 8237
    },
    {
      "epoch": 0.4445523717014732,
      "grad_norm": 1.0021732095779479,
      "learning_rate": 6.2799141490748915e-06,
      "loss": 0.4451,
      "step": 8238
    },
    {
      "epoch": 0.4446063353299876,
      "grad_norm": 0.9185118107048618,
      "learning_rate": 6.279162790227666e-06,
      "loss": 0.4588,
      "step": 8239
    },
    {
      "epoch": 0.44466029895850195,
      "grad_norm": 0.7817167615419632,
      "learning_rate": 6.278411408986515e-06,
      "loss": 0.3816,
      "step": 8240
    },
    {
      "epoch": 0.44471426258701635,
      "grad_norm": 1.0672024629976542,
      "learning_rate": 6.277660005373033e-06,
      "loss": 0.6394,
      "step": 8241
    },
    {
      "epoch": 0.44476822621553075,
      "grad_norm": 0.8798847536131332,
      "learning_rate": 6.276908579408817e-06,
      "loss": 0.4219,
      "step": 8242
    },
    {
      "epoch": 0.4448221898440451,
      "grad_norm": 1.3314417753540595,
      "learning_rate": 6.276157131115463e-06,
      "loss": 0.5775,
      "step": 8243
    },
    {
      "epoch": 0.4448761534725595,
      "grad_norm": 0.8686468218945674,
      "learning_rate": 6.27540566051457e-06,
      "loss": 0.4752,
      "step": 8244
    },
    {
      "epoch": 0.4449301171010739,
      "grad_norm": 0.9270347874248582,
      "learning_rate": 6.274654167627732e-06,
      "loss": 0.4318,
      "step": 8245
    },
    {
      "epoch": 0.4449840807295883,
      "grad_norm": 1.1602155861773826,
      "learning_rate": 6.2739026524765535e-06,
      "loss": 0.5141,
      "step": 8246
    },
    {
      "epoch": 0.4450380443581026,
      "grad_norm": 0.8589735913922834,
      "learning_rate": 6.27315111508263e-06,
      "loss": 0.3548,
      "step": 8247
    },
    {
      "epoch": 0.445092007986617,
      "grad_norm": 0.8935179007468842,
      "learning_rate": 6.272399555467562e-06,
      "loss": 0.5043,
      "step": 8248
    },
    {
      "epoch": 0.4451459716151314,
      "grad_norm": 1.1337664081952104,
      "learning_rate": 6.27164797365295e-06,
      "loss": 0.5161,
      "step": 8249
    },
    {
      "epoch": 0.44519993524364576,
      "grad_norm": 1.0080019638462099,
      "learning_rate": 6.270896369660396e-06,
      "loss": 0.3475,
      "step": 8250
    },
    {
      "epoch": 0.44525389887216016,
      "grad_norm": 0.8889086586750827,
      "learning_rate": 6.2701447435115e-06,
      "loss": 0.3803,
      "step": 8251
    },
    {
      "epoch": 0.44530786250067456,
      "grad_norm": 1.0485007018478847,
      "learning_rate": 6.269393095227867e-06,
      "loss": 0.4782,
      "step": 8252
    },
    {
      "epoch": 0.44536182612918895,
      "grad_norm": 0.976537839521626,
      "learning_rate": 6.268641424831099e-06,
      "loss": 0.4625,
      "step": 8253
    },
    {
      "epoch": 0.4454157897577033,
      "grad_norm": 0.9007209484689772,
      "learning_rate": 6.267889732342801e-06,
      "loss": 0.342,
      "step": 8254
    },
    {
      "epoch": 0.4454697533862177,
      "grad_norm": 0.9943723513430234,
      "learning_rate": 6.267138017784574e-06,
      "loss": 0.4716,
      "step": 8255
    },
    {
      "epoch": 0.4455237170147321,
      "grad_norm": 1.1286679947400704,
      "learning_rate": 6.266386281178027e-06,
      "loss": 0.6132,
      "step": 8256
    },
    {
      "epoch": 0.44557768064324643,
      "grad_norm": 1.207032797681148,
      "learning_rate": 6.265634522544764e-06,
      "loss": 0.4628,
      "step": 8257
    },
    {
      "epoch": 0.44563164427176083,
      "grad_norm": 0.9660067187717021,
      "learning_rate": 6.2648827419063895e-06,
      "loss": 0.5171,
      "step": 8258
    },
    {
      "epoch": 0.4456856079002752,
      "grad_norm": 1.141916691625657,
      "learning_rate": 6.264130939284512e-06,
      "loss": 0.481,
      "step": 8259
    },
    {
      "epoch": 0.44573957152878957,
      "grad_norm": 1.0683508061794467,
      "learning_rate": 6.2633791147007405e-06,
      "loss": 0.4122,
      "step": 8260
    },
    {
      "epoch": 0.44579353515730397,
      "grad_norm": 0.8414162974271109,
      "learning_rate": 6.262627268176682e-06,
      "loss": 0.4005,
      "step": 8261
    },
    {
      "epoch": 0.44584749878581836,
      "grad_norm": 0.7957324879708162,
      "learning_rate": 6.261875399733944e-06,
      "loss": 0.3489,
      "step": 8262
    },
    {
      "epoch": 0.44590146241433276,
      "grad_norm": 1.1980065814666387,
      "learning_rate": 6.261123509394138e-06,
      "loss": 0.4865,
      "step": 8263
    },
    {
      "epoch": 0.4459554260428471,
      "grad_norm": 1.0324751228374605,
      "learning_rate": 6.2603715971788735e-06,
      "loss": 0.5615,
      "step": 8264
    },
    {
      "epoch": 0.4460093896713615,
      "grad_norm": 1.0205668026300676,
      "learning_rate": 6.259619663109762e-06,
      "loss": 0.4992,
      "step": 8265
    },
    {
      "epoch": 0.4460633532998759,
      "grad_norm": 0.9961386734942749,
      "learning_rate": 6.258867707208412e-06,
      "loss": 0.5174,
      "step": 8266
    },
    {
      "epoch": 0.44611731692839024,
      "grad_norm": 0.9390337589808205,
      "learning_rate": 6.258115729496438e-06,
      "loss": 0.4027,
      "step": 8267
    },
    {
      "epoch": 0.44617128055690464,
      "grad_norm": 0.9822247636911285,
      "learning_rate": 6.257363729995452e-06,
      "loss": 0.4304,
      "step": 8268
    },
    {
      "epoch": 0.44622524418541903,
      "grad_norm": 0.9810901313736503,
      "learning_rate": 6.256611708727067e-06,
      "loss": 0.4036,
      "step": 8269
    },
    {
      "epoch": 0.44627920781393343,
      "grad_norm": 1.0819434472966634,
      "learning_rate": 6.255859665712897e-06,
      "loss": 0.5296,
      "step": 8270
    },
    {
      "epoch": 0.4463331714424478,
      "grad_norm": 0.8559395133237707,
      "learning_rate": 6.255107600974557e-06,
      "loss": 0.3429,
      "step": 8271
    },
    {
      "epoch": 0.44638713507096217,
      "grad_norm": 1.0165905916419231,
      "learning_rate": 6.254355514533661e-06,
      "loss": 0.4707,
      "step": 8272
    },
    {
      "epoch": 0.44644109869947657,
      "grad_norm": 1.1912514391587248,
      "learning_rate": 6.2536034064118245e-06,
      "loss": 0.5647,
      "step": 8273
    },
    {
      "epoch": 0.4464950623279909,
      "grad_norm": 0.8507150505266678,
      "learning_rate": 6.2528512766306674e-06,
      "loss": 0.4495,
      "step": 8274
    },
    {
      "epoch": 0.4465490259565053,
      "grad_norm": 1.1838393451015867,
      "learning_rate": 6.2520991252118005e-06,
      "loss": 0.5475,
      "step": 8275
    },
    {
      "epoch": 0.4466029895850197,
      "grad_norm": 0.8320536995197675,
      "learning_rate": 6.251346952176847e-06,
      "loss": 0.3026,
      "step": 8276
    },
    {
      "epoch": 0.4466569532135341,
      "grad_norm": 1.2358260879971017,
      "learning_rate": 6.250594757547422e-06,
      "loss": 0.3825,
      "step": 8277
    },
    {
      "epoch": 0.44671091684204844,
      "grad_norm": 1.2582339478268636,
      "learning_rate": 6.2498425413451455e-06,
      "loss": 0.6025,
      "step": 8278
    },
    {
      "epoch": 0.44676488047056284,
      "grad_norm": 1.0614539401854632,
      "learning_rate": 6.2490903035916365e-06,
      "loss": 0.4201,
      "step": 8279
    },
    {
      "epoch": 0.44681884409907724,
      "grad_norm": 1.0489981466148692,
      "learning_rate": 6.248338044308515e-06,
      "loss": 0.5237,
      "step": 8280
    },
    {
      "epoch": 0.4468728077275916,
      "grad_norm": 1.0946284703330373,
      "learning_rate": 6.247585763517402e-06,
      "loss": 0.4316,
      "step": 8281
    },
    {
      "epoch": 0.446926771356106,
      "grad_norm": 0.9949303875092664,
      "learning_rate": 6.246833461239921e-06,
      "loss": 0.4473,
      "step": 8282
    },
    {
      "epoch": 0.4469807349846204,
      "grad_norm": 1.124923389048746,
      "learning_rate": 6.246081137497689e-06,
      "loss": 0.5006,
      "step": 8283
    },
    {
      "epoch": 0.4470346986131348,
      "grad_norm": 1.1364614517118319,
      "learning_rate": 6.245328792312332e-06,
      "loss": 0.6883,
      "step": 8284
    },
    {
      "epoch": 0.4470886622416491,
      "grad_norm": 0.8968541085967444,
      "learning_rate": 6.2445764257054715e-06,
      "loss": 0.4419,
      "step": 8285
    },
    {
      "epoch": 0.4471426258701635,
      "grad_norm": 1.01887116537739,
      "learning_rate": 6.243824037698734e-06,
      "loss": 0.5444,
      "step": 8286
    },
    {
      "epoch": 0.4471965894986779,
      "grad_norm": 1.1322230658838248,
      "learning_rate": 6.243071628313741e-06,
      "loss": 0.5171,
      "step": 8287
    },
    {
      "epoch": 0.44725055312719225,
      "grad_norm": 0.9174488917296533,
      "learning_rate": 6.2423191975721195e-06,
      "loss": 0.4484,
      "step": 8288
    },
    {
      "epoch": 0.44730451675570665,
      "grad_norm": 1.0608661624047486,
      "learning_rate": 6.241566745495493e-06,
      "loss": 0.4541,
      "step": 8289
    },
    {
      "epoch": 0.44735848038422105,
      "grad_norm": 0.9951625557270805,
      "learning_rate": 6.24081427210549e-06,
      "loss": 0.4436,
      "step": 8290
    },
    {
      "epoch": 0.44741244401273544,
      "grad_norm": 1.0419475416538353,
      "learning_rate": 6.2400617774237345e-06,
      "loss": 0.5386,
      "step": 8291
    },
    {
      "epoch": 0.4474664076412498,
      "grad_norm": 1.2347215256909756,
      "learning_rate": 6.2393092614718585e-06,
      "loss": 0.6277,
      "step": 8292
    },
    {
      "epoch": 0.4475203712697642,
      "grad_norm": 0.7676799061597555,
      "learning_rate": 6.238556724271485e-06,
      "loss": 0.382,
      "step": 8293
    },
    {
      "epoch": 0.4475743348982786,
      "grad_norm": 0.9588430752199348,
      "learning_rate": 6.237804165844247e-06,
      "loss": 0.4675,
      "step": 8294
    },
    {
      "epoch": 0.4476282985267929,
      "grad_norm": 0.9119965845666558,
      "learning_rate": 6.237051586211771e-06,
      "loss": 0.4194,
      "step": 8295
    },
    {
      "epoch": 0.4476822621553073,
      "grad_norm": 1.118102031488165,
      "learning_rate": 6.236298985395687e-06,
      "loss": 0.5585,
      "step": 8296
    },
    {
      "epoch": 0.4477362257838217,
      "grad_norm": 1.272839837054946,
      "learning_rate": 6.235546363417626e-06,
      "loss": 0.7868,
      "step": 8297
    },
    {
      "epoch": 0.44779018941233606,
      "grad_norm": 1.2080708868311492,
      "learning_rate": 6.23479372029922e-06,
      "loss": 0.4729,
      "step": 8298
    },
    {
      "epoch": 0.44784415304085046,
      "grad_norm": 0.882292242508779,
      "learning_rate": 6.2340410560621e-06,
      "loss": 0.4413,
      "step": 8299
    },
    {
      "epoch": 0.44789811666936485,
      "grad_norm": 1.1725534847153798,
      "learning_rate": 6.233288370727898e-06,
      "loss": 0.7341,
      "step": 8300
    },
    {
      "epoch": 0.44795208029787925,
      "grad_norm": 0.8902496313702011,
      "learning_rate": 6.232535664318248e-06,
      "loss": 0.3883,
      "step": 8301
    },
    {
      "epoch": 0.4480060439263936,
      "grad_norm": 0.9755453942074706,
      "learning_rate": 6.231782936854783e-06,
      "loss": 0.4675,
      "step": 8302
    },
    {
      "epoch": 0.448060007554908,
      "grad_norm": 1.0832986342892634,
      "learning_rate": 6.231030188359136e-06,
      "loss": 0.469,
      "step": 8303
    },
    {
      "epoch": 0.4481139711834224,
      "grad_norm": 0.9594125867379704,
      "learning_rate": 6.230277418852942e-06,
      "loss": 0.5057,
      "step": 8304
    },
    {
      "epoch": 0.44816793481193673,
      "grad_norm": 1.0079701577881373,
      "learning_rate": 6.229524628357838e-06,
      "loss": 0.3428,
      "step": 8305
    },
    {
      "epoch": 0.4482218984404511,
      "grad_norm": 0.962475534638248,
      "learning_rate": 6.228771816895459e-06,
      "loss": 0.5037,
      "step": 8306
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 1.4026839396885045,
      "learning_rate": 6.228018984487443e-06,
      "loss": 0.6352,
      "step": 8307
    },
    {
      "epoch": 0.4483298256974799,
      "grad_norm": 1.0513368844937503,
      "learning_rate": 6.2272661311554235e-06,
      "loss": 0.4776,
      "step": 8308
    },
    {
      "epoch": 0.44838378932599426,
      "grad_norm": 0.9121626387900661,
      "learning_rate": 6.226513256921042e-06,
      "loss": 0.4673,
      "step": 8309
    },
    {
      "epoch": 0.44843775295450866,
      "grad_norm": 0.8371355145290016,
      "learning_rate": 6.2257603618059335e-06,
      "loss": 0.359,
      "step": 8310
    },
    {
      "epoch": 0.44849171658302306,
      "grad_norm": 1.151421452181055,
      "learning_rate": 6.225007445831741e-06,
      "loss": 0.5794,
      "step": 8311
    },
    {
      "epoch": 0.4485456802115374,
      "grad_norm": 0.9334231884552951,
      "learning_rate": 6.2242545090201025e-06,
      "loss": 0.4177,
      "step": 8312
    },
    {
      "epoch": 0.4485996438400518,
      "grad_norm": 1.0115269486691854,
      "learning_rate": 6.223501551392656e-06,
      "loss": 0.4407,
      "step": 8313
    },
    {
      "epoch": 0.4486536074685662,
      "grad_norm": 1.098379406537149,
      "learning_rate": 6.222748572971044e-06,
      "loss": 0.5261,
      "step": 8314
    },
    {
      "epoch": 0.4487075710970806,
      "grad_norm": 0.9984254964387551,
      "learning_rate": 6.221995573776907e-06,
      "loss": 0.4602,
      "step": 8315
    },
    {
      "epoch": 0.44876153472559493,
      "grad_norm": 1.0385443864372967,
      "learning_rate": 6.221242553831889e-06,
      "loss": 0.5939,
      "step": 8316
    },
    {
      "epoch": 0.44881549835410933,
      "grad_norm": 0.8991944262663476,
      "learning_rate": 6.2204895131576315e-06,
      "loss": 0.4789,
      "step": 8317
    },
    {
      "epoch": 0.44886946198262373,
      "grad_norm": 1.0129856418267855,
      "learning_rate": 6.219736451775777e-06,
      "loss": 0.4682,
      "step": 8318
    },
    {
      "epoch": 0.44892342561113807,
      "grad_norm": 0.9235633118556752,
      "learning_rate": 6.218983369707969e-06,
      "loss": 0.52,
      "step": 8319
    },
    {
      "epoch": 0.44897738923965247,
      "grad_norm": 0.8832354205589996,
      "learning_rate": 6.2182302669758534e-06,
      "loss": 0.4493,
      "step": 8320
    },
    {
      "epoch": 0.44903135286816687,
      "grad_norm": 1.131605401898967,
      "learning_rate": 6.217477143601073e-06,
      "loss": 0.4993,
      "step": 8321
    },
    {
      "epoch": 0.44908531649668126,
      "grad_norm": 1.0258323143659747,
      "learning_rate": 6.216723999605275e-06,
      "loss": 0.4683,
      "step": 8322
    },
    {
      "epoch": 0.4491392801251956,
      "grad_norm": 0.8246839525219615,
      "learning_rate": 6.2159708350101064e-06,
      "loss": 0.3551,
      "step": 8323
    },
    {
      "epoch": 0.44919324375371,
      "grad_norm": 1.1903603655530142,
      "learning_rate": 6.215217649837212e-06,
      "loss": 0.4971,
      "step": 8324
    },
    {
      "epoch": 0.4492472073822244,
      "grad_norm": 1.04491046368811,
      "learning_rate": 6.214464444108238e-06,
      "loss": 0.5212,
      "step": 8325
    },
    {
      "epoch": 0.44930117101073874,
      "grad_norm": 0.7817496636603269,
      "learning_rate": 6.2137112178448356e-06,
      "loss": 0.4048,
      "step": 8326
    },
    {
      "epoch": 0.44935513463925314,
      "grad_norm": 1.0367474577104214,
      "learning_rate": 6.212957971068652e-06,
      "loss": 0.5398,
      "step": 8327
    },
    {
      "epoch": 0.44940909826776754,
      "grad_norm": 0.9083244646446756,
      "learning_rate": 6.212204703801334e-06,
      "loss": 0.3629,
      "step": 8328
    },
    {
      "epoch": 0.4494630618962819,
      "grad_norm": 0.9703374855091751,
      "learning_rate": 6.211451416064536e-06,
      "loss": 0.4534,
      "step": 8329
    },
    {
      "epoch": 0.4495170255247963,
      "grad_norm": 1.0219002599005267,
      "learning_rate": 6.210698107879903e-06,
      "loss": 0.538,
      "step": 8330
    },
    {
      "epoch": 0.4495709891533107,
      "grad_norm": 0.8619922925486874,
      "learning_rate": 6.209944779269088e-06,
      "loss": 0.353,
      "step": 8331
    },
    {
      "epoch": 0.44962495278182507,
      "grad_norm": 0.9930825798100148,
      "learning_rate": 6.209191430253743e-06,
      "loss": 0.3964,
      "step": 8332
    },
    {
      "epoch": 0.4496789164103394,
      "grad_norm": 0.9691815783001185,
      "learning_rate": 6.208438060855522e-06,
      "loss": 0.3916,
      "step": 8333
    },
    {
      "epoch": 0.4497328800388538,
      "grad_norm": 1.0870076654952308,
      "learning_rate": 6.207684671096073e-06,
      "loss": 0.5516,
      "step": 8334
    },
    {
      "epoch": 0.4497868436673682,
      "grad_norm": 1.0839231078343547,
      "learning_rate": 6.206931260997052e-06,
      "loss": 0.4722,
      "step": 8335
    },
    {
      "epoch": 0.44984080729588255,
      "grad_norm": 1.1291991686581544,
      "learning_rate": 6.206177830580113e-06,
      "loss": 0.5012,
      "step": 8336
    },
    {
      "epoch": 0.44989477092439695,
      "grad_norm": 1.1284066685510572,
      "learning_rate": 6.205424379866909e-06,
      "loss": 0.5494,
      "step": 8337
    },
    {
      "epoch": 0.44994873455291134,
      "grad_norm": 0.9301465594589194,
      "learning_rate": 6.204670908879095e-06,
      "loss": 0.3552,
      "step": 8338
    },
    {
      "epoch": 0.45000269818142574,
      "grad_norm": 1.1763205306953284,
      "learning_rate": 6.203917417638327e-06,
      "loss": 0.5184,
      "step": 8339
    },
    {
      "epoch": 0.4500566618099401,
      "grad_norm": 1.06251513760377,
      "learning_rate": 6.203163906166263e-06,
      "loss": 0.5114,
      "step": 8340
    },
    {
      "epoch": 0.4501106254384545,
      "grad_norm": 1.1333362481581362,
      "learning_rate": 6.202410374484557e-06,
      "loss": 0.5058,
      "step": 8341
    },
    {
      "epoch": 0.4501645890669689,
      "grad_norm": 0.9318898124912248,
      "learning_rate": 6.201656822614865e-06,
      "loss": 0.3844,
      "step": 8342
    },
    {
      "epoch": 0.4502185526954832,
      "grad_norm": 1.0563534968644528,
      "learning_rate": 6.20090325057885e-06,
      "loss": 0.528,
      "step": 8343
    },
    {
      "epoch": 0.4502725163239976,
      "grad_norm": 0.9993405420382931,
      "learning_rate": 6.200149658398166e-06,
      "loss": 0.5493,
      "step": 8344
    },
    {
      "epoch": 0.450326479952512,
      "grad_norm": 0.9457576795836067,
      "learning_rate": 6.1993960460944736e-06,
      "loss": 0.4024,
      "step": 8345
    },
    {
      "epoch": 0.4503804435810264,
      "grad_norm": 1.1096445902737497,
      "learning_rate": 6.198642413689432e-06,
      "loss": 0.5194,
      "step": 8346
    },
    {
      "epoch": 0.45043440720954075,
      "grad_norm": 1.2304162848135645,
      "learning_rate": 6.197888761204702e-06,
      "loss": 0.4624,
      "step": 8347
    },
    {
      "epoch": 0.45048837083805515,
      "grad_norm": 1.0012605899037046,
      "learning_rate": 6.197135088661943e-06,
      "loss": 0.3959,
      "step": 8348
    },
    {
      "epoch": 0.45054233446656955,
      "grad_norm": 1.216003438390205,
      "learning_rate": 6.1963813960828165e-06,
      "loss": 0.6588,
      "step": 8349
    },
    {
      "epoch": 0.4505962980950839,
      "grad_norm": 0.994540218474625,
      "learning_rate": 6.195627683488987e-06,
      "loss": 0.5286,
      "step": 8350
    },
    {
      "epoch": 0.4506502617235983,
      "grad_norm": 1.038760729273064,
      "learning_rate": 6.194873950902114e-06,
      "loss": 0.5037,
      "step": 8351
    },
    {
      "epoch": 0.4507042253521127,
      "grad_norm": 1.120989378994448,
      "learning_rate": 6.19412019834386e-06,
      "loss": 0.673,
      "step": 8352
    },
    {
      "epoch": 0.4507581889806271,
      "grad_norm": 1.0111310770412425,
      "learning_rate": 6.193366425835893e-06,
      "loss": 0.5204,
      "step": 8353
    },
    {
      "epoch": 0.4508121526091414,
      "grad_norm": 0.9703600015052282,
      "learning_rate": 6.1926126333998734e-06,
      "loss": 0.5822,
      "step": 8354
    },
    {
      "epoch": 0.4508661162376558,
      "grad_norm": 0.9622357499434295,
      "learning_rate": 6.191858821057466e-06,
      "loss": 0.45,
      "step": 8355
    },
    {
      "epoch": 0.4509200798661702,
      "grad_norm": 1.039420922409652,
      "learning_rate": 6.191104988830338e-06,
      "loss": 0.4756,
      "step": 8356
    },
    {
      "epoch": 0.45097404349468456,
      "grad_norm": 0.9776075490361592,
      "learning_rate": 6.190351136740153e-06,
      "loss": 0.5327,
      "step": 8357
    },
    {
      "epoch": 0.45102800712319896,
      "grad_norm": 1.0125723817976613,
      "learning_rate": 6.1895972648085804e-06,
      "loss": 0.4998,
      "step": 8358
    },
    {
      "epoch": 0.45108197075171336,
      "grad_norm": 0.9999117314031796,
      "learning_rate": 6.188843373057284e-06,
      "loss": 0.5464,
      "step": 8359
    },
    {
      "epoch": 0.4511359343802277,
      "grad_norm": 1.0246430134366973,
      "learning_rate": 6.188089461507933e-06,
      "loss": 0.5534,
      "step": 8360
    },
    {
      "epoch": 0.4511898980087421,
      "grad_norm": 0.9853291286500127,
      "learning_rate": 6.187335530182197e-06,
      "loss": 0.4887,
      "step": 8361
    },
    {
      "epoch": 0.4512438616372565,
      "grad_norm": 0.8041286237980081,
      "learning_rate": 6.1865815791017435e-06,
      "loss": 0.2845,
      "step": 8362
    },
    {
      "epoch": 0.4512978252657709,
      "grad_norm": 1.012319729362179,
      "learning_rate": 6.1858276082882406e-06,
      "loss": 0.5868,
      "step": 8363
    },
    {
      "epoch": 0.45135178889428523,
      "grad_norm": 1.0324055590509715,
      "learning_rate": 6.185073617763359e-06,
      "loss": 0.5689,
      "step": 8364
    },
    {
      "epoch": 0.45140575252279963,
      "grad_norm": 0.980216297373514,
      "learning_rate": 6.184319607548771e-06,
      "loss": 0.5346,
      "step": 8365
    },
    {
      "epoch": 0.45145971615131403,
      "grad_norm": 0.9669955495748339,
      "learning_rate": 6.183565577666144e-06,
      "loss": 0.5843,
      "step": 8366
    },
    {
      "epoch": 0.45151367977982837,
      "grad_norm": 1.1596689194399683,
      "learning_rate": 6.182811528137154e-06,
      "loss": 0.6075,
      "step": 8367
    },
    {
      "epoch": 0.45156764340834277,
      "grad_norm": 0.842196522859393,
      "learning_rate": 6.182057458983469e-06,
      "loss": 0.4574,
      "step": 8368
    },
    {
      "epoch": 0.45162160703685716,
      "grad_norm": 1.0362438583159617,
      "learning_rate": 6.181303370226764e-06,
      "loss": 0.4925,
      "step": 8369
    },
    {
      "epoch": 0.45167557066537156,
      "grad_norm": 1.2152759413158918,
      "learning_rate": 6.180549261888711e-06,
      "loss": 0.5789,
      "step": 8370
    },
    {
      "epoch": 0.4517295342938859,
      "grad_norm": 0.9103548226380835,
      "learning_rate": 6.179795133990986e-06,
      "loss": 0.4999,
      "step": 8371
    },
    {
      "epoch": 0.4517834979224003,
      "grad_norm": 1.1487731602898519,
      "learning_rate": 6.179040986555261e-06,
      "loss": 0.5211,
      "step": 8372
    },
    {
      "epoch": 0.4518374615509147,
      "grad_norm": 1.0812676407560937,
      "learning_rate": 6.178286819603212e-06,
      "loss": 0.4619,
      "step": 8373
    },
    {
      "epoch": 0.45189142517942904,
      "grad_norm": 0.8267472326923944,
      "learning_rate": 6.1775326331565154e-06,
      "loss": 0.273,
      "step": 8374
    },
    {
      "epoch": 0.45194538880794344,
      "grad_norm": 0.9225753201805581,
      "learning_rate": 6.1767784272368445e-06,
      "loss": 0.441,
      "step": 8375
    },
    {
      "epoch": 0.45199935243645784,
      "grad_norm": 1.204706833911825,
      "learning_rate": 6.176024201865879e-06,
      "loss": 0.5823,
      "step": 8376
    },
    {
      "epoch": 0.45205331606497223,
      "grad_norm": 0.9391251589629964,
      "learning_rate": 6.175269957065294e-06,
      "loss": 0.3979,
      "step": 8377
    },
    {
      "epoch": 0.4521072796934866,
      "grad_norm": 0.9858315946353605,
      "learning_rate": 6.174515692856769e-06,
      "loss": 0.392,
      "step": 8378
    },
    {
      "epoch": 0.45216124332200097,
      "grad_norm": 0.9188130667507144,
      "learning_rate": 6.173761409261982e-06,
      "loss": 0.4426,
      "step": 8379
    },
    {
      "epoch": 0.45221520695051537,
      "grad_norm": 1.0188839161177743,
      "learning_rate": 6.1730071063026095e-06,
      "loss": 0.4881,
      "step": 8380
    },
    {
      "epoch": 0.4522691705790297,
      "grad_norm": 0.9640000889921666,
      "learning_rate": 6.1722527840003354e-06,
      "loss": 0.4899,
      "step": 8381
    },
    {
      "epoch": 0.4523231342075441,
      "grad_norm": 1.0662342160455445,
      "learning_rate": 6.171498442376834e-06,
      "loss": 0.4939,
      "step": 8382
    },
    {
      "epoch": 0.4523770978360585,
      "grad_norm": 1.173047501051285,
      "learning_rate": 6.170744081453791e-06,
      "loss": 0.4825,
      "step": 8383
    },
    {
      "epoch": 0.4524310614645729,
      "grad_norm": 1.0085298781216618,
      "learning_rate": 6.169989701252884e-06,
      "loss": 0.501,
      "step": 8384
    },
    {
      "epoch": 0.45248502509308725,
      "grad_norm": 0.9506675705246955,
      "learning_rate": 6.169235301795796e-06,
      "loss": 0.3623,
      "step": 8385
    },
    {
      "epoch": 0.45253898872160164,
      "grad_norm": 1.0177712703639705,
      "learning_rate": 6.16848088310421e-06,
      "loss": 0.4606,
      "step": 8386
    },
    {
      "epoch": 0.45259295235011604,
      "grad_norm": 1.1285838234095045,
      "learning_rate": 6.167726445199808e-06,
      "loss": 0.5457,
      "step": 8387
    },
    {
      "epoch": 0.4526469159786304,
      "grad_norm": 1.2700764543458856,
      "learning_rate": 6.166971988104272e-06,
      "loss": 0.6198,
      "step": 8388
    },
    {
      "epoch": 0.4527008796071448,
      "grad_norm": 1.0182056657847405,
      "learning_rate": 6.1662175118392886e-06,
      "loss": 0.4384,
      "step": 8389
    },
    {
      "epoch": 0.4527548432356592,
      "grad_norm": 1.007168745803475,
      "learning_rate": 6.1654630164265406e-06,
      "loss": 0.4493,
      "step": 8390
    },
    {
      "epoch": 0.4528088068641736,
      "grad_norm": 1.082975471430919,
      "learning_rate": 6.164708501887713e-06,
      "loss": 0.5555,
      "step": 8391
    },
    {
      "epoch": 0.4528627704926879,
      "grad_norm": 1.20443162823091,
      "learning_rate": 6.1639539682444915e-06,
      "loss": 0.6704,
      "step": 8392
    },
    {
      "epoch": 0.4529167341212023,
      "grad_norm": 1.0178248637649168,
      "learning_rate": 6.163199415518562e-06,
      "loss": 0.5266,
      "step": 8393
    },
    {
      "epoch": 0.4529706977497167,
      "grad_norm": 0.9568505557307015,
      "learning_rate": 6.162444843731611e-06,
      "loss": 0.4566,
      "step": 8394
    },
    {
      "epoch": 0.45302466137823105,
      "grad_norm": 1.1306115302604292,
      "learning_rate": 6.161690252905325e-06,
      "loss": 0.5377,
      "step": 8395
    },
    {
      "epoch": 0.45307862500674545,
      "grad_norm": 0.9686777562326195,
      "learning_rate": 6.160935643061394e-06,
      "loss": 0.3678,
      "step": 8396
    },
    {
      "epoch": 0.45313258863525985,
      "grad_norm": 1.4316369198651668,
      "learning_rate": 6.160181014221503e-06,
      "loss": 0.4487,
      "step": 8397
    },
    {
      "epoch": 0.4531865522637742,
      "grad_norm": 0.9194427818136042,
      "learning_rate": 6.159426366407345e-06,
      "loss": 0.4422,
      "step": 8398
    },
    {
      "epoch": 0.4532405158922886,
      "grad_norm": 1.0726317007948751,
      "learning_rate": 6.158671699640606e-06,
      "loss": 0.5604,
      "step": 8399
    },
    {
      "epoch": 0.453294479520803,
      "grad_norm": 1.1646538564684907,
      "learning_rate": 6.157917013942977e-06,
      "loss": 0.4353,
      "step": 8400
    },
    {
      "epoch": 0.4533484431493174,
      "grad_norm": 1.013790709910214,
      "learning_rate": 6.157162309336146e-06,
      "loss": 0.4955,
      "step": 8401
    },
    {
      "epoch": 0.4534024067778317,
      "grad_norm": 0.8692143727078429,
      "learning_rate": 6.156407585841809e-06,
      "loss": 0.3412,
      "step": 8402
    },
    {
      "epoch": 0.4534563704063461,
      "grad_norm": 1.0236675467290766,
      "learning_rate": 6.155652843481655e-06,
      "loss": 0.4563,
      "step": 8403
    },
    {
      "epoch": 0.4535103340348605,
      "grad_norm": 1.1577464304035896,
      "learning_rate": 6.154898082277374e-06,
      "loss": 0.6952,
      "step": 8404
    },
    {
      "epoch": 0.45356429766337486,
      "grad_norm": 0.9653272623430008,
      "learning_rate": 6.154143302250661e-06,
      "loss": 0.4835,
      "step": 8405
    },
    {
      "epoch": 0.45361826129188926,
      "grad_norm": 1.2152141772947251,
      "learning_rate": 6.153388503423208e-06,
      "loss": 0.6698,
      "step": 8406
    },
    {
      "epoch": 0.45367222492040366,
      "grad_norm": 0.9519933989119045,
      "learning_rate": 6.1526336858167095e-06,
      "loss": 0.4883,
      "step": 8407
    },
    {
      "epoch": 0.45372618854891805,
      "grad_norm": 1.0341868436421364,
      "learning_rate": 6.1518788494528596e-06,
      "loss": 0.4129,
      "step": 8408
    },
    {
      "epoch": 0.4537801521774324,
      "grad_norm": 1.3135231331330393,
      "learning_rate": 6.1511239943533515e-06,
      "loss": 0.6155,
      "step": 8409
    },
    {
      "epoch": 0.4538341158059468,
      "grad_norm": 1.0899223236121354,
      "learning_rate": 6.1503691205398855e-06,
      "loss": 0.578,
      "step": 8410
    },
    {
      "epoch": 0.4538880794344612,
      "grad_norm": 1.3154815500184927,
      "learning_rate": 6.1496142280341506e-06,
      "loss": 0.6484,
      "step": 8411
    },
    {
      "epoch": 0.45394204306297553,
      "grad_norm": 1.1646261336130328,
      "learning_rate": 6.148859316857847e-06,
      "loss": 0.6288,
      "step": 8412
    },
    {
      "epoch": 0.45399600669148993,
      "grad_norm": 1.0544223359081173,
      "learning_rate": 6.1481043870326705e-06,
      "loss": 0.5162,
      "step": 8413
    },
    {
      "epoch": 0.4540499703200043,
      "grad_norm": 1.0392206489482072,
      "learning_rate": 6.1473494385803205e-06,
      "loss": 0.4698,
      "step": 8414
    },
    {
      "epoch": 0.4541039339485187,
      "grad_norm": 0.8609037082670334,
      "learning_rate": 6.146594471522492e-06,
      "loss": 0.4052,
      "step": 8415
    },
    {
      "epoch": 0.45415789757703307,
      "grad_norm": 1.0197117928968813,
      "learning_rate": 6.145839485880888e-06,
      "loss": 0.6074,
      "step": 8416
    },
    {
      "epoch": 0.45421186120554746,
      "grad_norm": 1.2330149711482408,
      "learning_rate": 6.145084481677202e-06,
      "loss": 0.5834,
      "step": 8417
    },
    {
      "epoch": 0.45426582483406186,
      "grad_norm": 0.9762581434531056,
      "learning_rate": 6.144329458933136e-06,
      "loss": 0.5018,
      "step": 8418
    },
    {
      "epoch": 0.4543197884625762,
      "grad_norm": 1.0473449784789306,
      "learning_rate": 6.143574417670392e-06,
      "loss": 0.5464,
      "step": 8419
    },
    {
      "epoch": 0.4543737520910906,
      "grad_norm": 1.025767839612092,
      "learning_rate": 6.142819357910668e-06,
      "loss": 0.4865,
      "step": 8420
    },
    {
      "epoch": 0.454427715719605,
      "grad_norm": 0.8045375711693429,
      "learning_rate": 6.142064279675666e-06,
      "loss": 0.4349,
      "step": 8421
    },
    {
      "epoch": 0.4544816793481194,
      "grad_norm": 1.4000566304986641,
      "learning_rate": 6.141309182987088e-06,
      "loss": 0.7409,
      "step": 8422
    },
    {
      "epoch": 0.45453564297663374,
      "grad_norm": 1.1533345558626389,
      "learning_rate": 6.1405540678666355e-06,
      "loss": 0.5492,
      "step": 8423
    },
    {
      "epoch": 0.45458960660514813,
      "grad_norm": 1.1651246286190822,
      "learning_rate": 6.1397989343360135e-06,
      "loss": 0.5318,
      "step": 8424
    },
    {
      "epoch": 0.45464357023366253,
      "grad_norm": 0.8395877309237959,
      "learning_rate": 6.139043782416922e-06,
      "loss": 0.3905,
      "step": 8425
    },
    {
      "epoch": 0.4546975338621769,
      "grad_norm": 1.0018641529477679,
      "learning_rate": 6.138288612131068e-06,
      "loss": 0.4648,
      "step": 8426
    },
    {
      "epoch": 0.45475149749069127,
      "grad_norm": 0.9188233555625183,
      "learning_rate": 6.1375334235001525e-06,
      "loss": 0.44,
      "step": 8427
    },
    {
      "epoch": 0.45480546111920567,
      "grad_norm": 1.1617255350860531,
      "learning_rate": 6.136778216545883e-06,
      "loss": 0.4955,
      "step": 8428
    },
    {
      "epoch": 0.45485942474772,
      "grad_norm": 0.9770796015753895,
      "learning_rate": 6.136022991289965e-06,
      "loss": 0.4278,
      "step": 8429
    },
    {
      "epoch": 0.4549133883762344,
      "grad_norm": 1.1101169000782731,
      "learning_rate": 6.135267747754102e-06,
      "loss": 0.473,
      "step": 8430
    },
    {
      "epoch": 0.4549673520047488,
      "grad_norm": 0.9638366606720424,
      "learning_rate": 6.134512485960003e-06,
      "loss": 0.4238,
      "step": 8431
    },
    {
      "epoch": 0.4550213156332632,
      "grad_norm": 0.9352684946429645,
      "learning_rate": 6.133757205929375e-06,
      "loss": 0.3438,
      "step": 8432
    },
    {
      "epoch": 0.45507527926177754,
      "grad_norm": 1.0500956939638022,
      "learning_rate": 6.133001907683923e-06,
      "loss": 0.3817,
      "step": 8433
    },
    {
      "epoch": 0.45512924289029194,
      "grad_norm": 1.107319073815282,
      "learning_rate": 6.132246591245358e-06,
      "loss": 0.5657,
      "step": 8434
    },
    {
      "epoch": 0.45518320651880634,
      "grad_norm": 1.0755078350298863,
      "learning_rate": 6.131491256635386e-06,
      "loss": 0.4565,
      "step": 8435
    },
    {
      "epoch": 0.4552371701473207,
      "grad_norm": 1.021547240217654,
      "learning_rate": 6.130735903875717e-06,
      "loss": 0.4506,
      "step": 8436
    },
    {
      "epoch": 0.4552911337758351,
      "grad_norm": 0.9503372261437775,
      "learning_rate": 6.12998053298806e-06,
      "loss": 0.3925,
      "step": 8437
    },
    {
      "epoch": 0.4553450974043495,
      "grad_norm": 0.8986971076519543,
      "learning_rate": 6.129225143994128e-06,
      "loss": 0.3797,
      "step": 8438
    },
    {
      "epoch": 0.45539906103286387,
      "grad_norm": 1.0924280393162988,
      "learning_rate": 6.1284697369156276e-06,
      "loss": 0.5575,
      "step": 8439
    },
    {
      "epoch": 0.4554530246613782,
      "grad_norm": 1.0662171701577052,
      "learning_rate": 6.127714311774274e-06,
      "loss": 0.6119,
      "step": 8440
    },
    {
      "epoch": 0.4555069882898926,
      "grad_norm": 1.1499326283133573,
      "learning_rate": 6.126958868591776e-06,
      "loss": 0.442,
      "step": 8441
    },
    {
      "epoch": 0.455560951918407,
      "grad_norm": 1.0986606405606907,
      "learning_rate": 6.126203407389847e-06,
      "loss": 0.3998,
      "step": 8442
    },
    {
      "epoch": 0.45561491554692135,
      "grad_norm": 0.8962837891663541,
      "learning_rate": 6.125447928190198e-06,
      "loss": 0.5006,
      "step": 8443
    },
    {
      "epoch": 0.45566887917543575,
      "grad_norm": 0.9971383229845352,
      "learning_rate": 6.124692431014545e-06,
      "loss": 0.6899,
      "step": 8444
    },
    {
      "epoch": 0.45572284280395015,
      "grad_norm": 0.8485723512707384,
      "learning_rate": 6.1239369158846e-06,
      "loss": 0.3514,
      "step": 8445
    },
    {
      "epoch": 0.45577680643246454,
      "grad_norm": 1.0580834044856593,
      "learning_rate": 6.1231813828220765e-06,
      "loss": 0.432,
      "step": 8446
    },
    {
      "epoch": 0.4558307700609789,
      "grad_norm": 1.1477871486260398,
      "learning_rate": 6.122425831848692e-06,
      "loss": 0.5182,
      "step": 8447
    },
    {
      "epoch": 0.4558847336894933,
      "grad_norm": 0.9197826516206705,
      "learning_rate": 6.121670262986159e-06,
      "loss": 0.4773,
      "step": 8448
    },
    {
      "epoch": 0.4559386973180077,
      "grad_norm": 1.028800803676154,
      "learning_rate": 6.120914676256194e-06,
      "loss": 0.4905,
      "step": 8449
    },
    {
      "epoch": 0.455992660946522,
      "grad_norm": 1.1323639001759163,
      "learning_rate": 6.120159071680516e-06,
      "loss": 0.6318,
      "step": 8450
    },
    {
      "epoch": 0.4560466245750364,
      "grad_norm": 0.9226195557320429,
      "learning_rate": 6.119403449280838e-06,
      "loss": 0.5202,
      "step": 8451
    },
    {
      "epoch": 0.4561005882035508,
      "grad_norm": 1.0149540456819108,
      "learning_rate": 6.118647809078879e-06,
      "loss": 0.4867,
      "step": 8452
    },
    {
      "epoch": 0.4561545518320652,
      "grad_norm": 1.240496641598694,
      "learning_rate": 6.117892151096357e-06,
      "loss": 0.7369,
      "step": 8453
    },
    {
      "epoch": 0.45620851546057956,
      "grad_norm": 1.1263990257026721,
      "learning_rate": 6.117136475354992e-06,
      "loss": 0.5227,
      "step": 8454
    },
    {
      "epoch": 0.45626247908909395,
      "grad_norm": 0.9613535018947932,
      "learning_rate": 6.1163807818765e-06,
      "loss": 0.5261,
      "step": 8455
    },
    {
      "epoch": 0.45631644271760835,
      "grad_norm": 0.9330383900526112,
      "learning_rate": 6.115625070682599e-06,
      "loss": 0.4279,
      "step": 8456
    },
    {
      "epoch": 0.4563704063461227,
      "grad_norm": 1.1216426898183989,
      "learning_rate": 6.114869341795016e-06,
      "loss": 0.4996,
      "step": 8457
    },
    {
      "epoch": 0.4564243699746371,
      "grad_norm": 1.0693226349887006,
      "learning_rate": 6.114113595235463e-06,
      "loss": 0.4831,
      "step": 8458
    },
    {
      "epoch": 0.4564783336031515,
      "grad_norm": 0.9244463053224956,
      "learning_rate": 6.113357831025666e-06,
      "loss": 0.4473,
      "step": 8459
    },
    {
      "epoch": 0.4565322972316659,
      "grad_norm": 1.1140635189354953,
      "learning_rate": 6.112602049187346e-06,
      "loss": 0.5454,
      "step": 8460
    },
    {
      "epoch": 0.4565862608601802,
      "grad_norm": 0.8572697561533181,
      "learning_rate": 6.111846249742223e-06,
      "loss": 0.3565,
      "step": 8461
    },
    {
      "epoch": 0.4566402244886946,
      "grad_norm": 0.9353356161947738,
      "learning_rate": 6.11109043271202e-06,
      "loss": 0.4668,
      "step": 8462
    },
    {
      "epoch": 0.456694188117209,
      "grad_norm": 0.8861661180500667,
      "learning_rate": 6.110334598118463e-06,
      "loss": 0.4136,
      "step": 8463
    },
    {
      "epoch": 0.45674815174572336,
      "grad_norm": 0.9218541589092455,
      "learning_rate": 6.109578745983271e-06,
      "loss": 0.5015,
      "step": 8464
    },
    {
      "epoch": 0.45680211537423776,
      "grad_norm": 0.9450886038216134,
      "learning_rate": 6.10882287632817e-06,
      "loss": 0.4391,
      "step": 8465
    },
    {
      "epoch": 0.45685607900275216,
      "grad_norm": 0.9693866807681021,
      "learning_rate": 6.108066989174885e-06,
      "loss": 0.435,
      "step": 8466
    },
    {
      "epoch": 0.4569100426312665,
      "grad_norm": 1.1243327528541787,
      "learning_rate": 6.10731108454514e-06,
      "loss": 0.4661,
      "step": 8467
    },
    {
      "epoch": 0.4569640062597809,
      "grad_norm": 1.0630308142947953,
      "learning_rate": 6.1065551624606604e-06,
      "loss": 0.445,
      "step": 8468
    },
    {
      "epoch": 0.4570179698882953,
      "grad_norm": 1.1700574430663637,
      "learning_rate": 6.105799222943174e-06,
      "loss": 0.5276,
      "step": 8469
    },
    {
      "epoch": 0.4570719335168097,
      "grad_norm": 1.0265078683983655,
      "learning_rate": 6.105043266014403e-06,
      "loss": 0.4411,
      "step": 8470
    },
    {
      "epoch": 0.45712589714532403,
      "grad_norm": 1.1374559221522296,
      "learning_rate": 6.1042872916960806e-06,
      "loss": 0.5157,
      "step": 8471
    },
    {
      "epoch": 0.45717986077383843,
      "grad_norm": 0.998708861454792,
      "learning_rate": 6.1035313000099295e-06,
      "loss": 0.4489,
      "step": 8472
    },
    {
      "epoch": 0.45723382440235283,
      "grad_norm": 1.169383222178501,
      "learning_rate": 6.102775290977677e-06,
      "loss": 0.4883,
      "step": 8473
    },
    {
      "epoch": 0.45728778803086717,
      "grad_norm": 1.0020367954374028,
      "learning_rate": 6.102019264621056e-06,
      "loss": 0.5212,
      "step": 8474
    },
    {
      "epoch": 0.45734175165938157,
      "grad_norm": 1.1450746089380885,
      "learning_rate": 6.101263220961793e-06,
      "loss": 0.5074,
      "step": 8475
    },
    {
      "epoch": 0.45739571528789597,
      "grad_norm": 1.0397486987654283,
      "learning_rate": 6.100507160021616e-06,
      "loss": 0.4283,
      "step": 8476
    },
    {
      "epoch": 0.45744967891641036,
      "grad_norm": 1.0796175625232014,
      "learning_rate": 6.099751081822258e-06,
      "loss": 0.5379,
      "step": 8477
    },
    {
      "epoch": 0.4575036425449247,
      "grad_norm": 1.1806903062657337,
      "learning_rate": 6.098994986385447e-06,
      "loss": 0.5238,
      "step": 8478
    },
    {
      "epoch": 0.4575576061734391,
      "grad_norm": 1.1241456515390165,
      "learning_rate": 6.098238873732916e-06,
      "loss": 0.4644,
      "step": 8479
    },
    {
      "epoch": 0.4576115698019535,
      "grad_norm": 1.0745543983247552,
      "learning_rate": 6.097482743886394e-06,
      "loss": 0.4618,
      "step": 8480
    },
    {
      "epoch": 0.45766553343046784,
      "grad_norm": 1.027403151556168,
      "learning_rate": 6.096726596867615e-06,
      "loss": 0.4792,
      "step": 8481
    },
    {
      "epoch": 0.45771949705898224,
      "grad_norm": 1.0969413909195633,
      "learning_rate": 6.09597043269831e-06,
      "loss": 0.5457,
      "step": 8482
    },
    {
      "epoch": 0.45777346068749664,
      "grad_norm": 1.1293282512055978,
      "learning_rate": 6.0952142514002136e-06,
      "loss": 0.504,
      "step": 8483
    },
    {
      "epoch": 0.45782742431601103,
      "grad_norm": 0.9197931789113908,
      "learning_rate": 6.094458052995056e-06,
      "loss": 0.3922,
      "step": 8484
    },
    {
      "epoch": 0.4578813879445254,
      "grad_norm": 1.0631695486416,
      "learning_rate": 6.093701837504575e-06,
      "loss": 0.4285,
      "step": 8485
    },
    {
      "epoch": 0.4579353515730398,
      "grad_norm": 0.8850271693059165,
      "learning_rate": 6.092945604950504e-06,
      "loss": 0.3399,
      "step": 8486
    },
    {
      "epoch": 0.45798931520155417,
      "grad_norm": 1.0660977822856967,
      "learning_rate": 6.092189355354575e-06,
      "loss": 0.4323,
      "step": 8487
    },
    {
      "epoch": 0.4580432788300685,
      "grad_norm": 0.9442403405791846,
      "learning_rate": 6.091433088738528e-06,
      "loss": 0.6314,
      "step": 8488
    },
    {
      "epoch": 0.4580972424585829,
      "grad_norm": 0.8417808395386996,
      "learning_rate": 6.090676805124094e-06,
      "loss": 0.3953,
      "step": 8489
    },
    {
      "epoch": 0.4581512060870973,
      "grad_norm": 1.177309652042817,
      "learning_rate": 6.089920504533012e-06,
      "loss": 0.5619,
      "step": 8490
    },
    {
      "epoch": 0.4582051697156117,
      "grad_norm": 1.2190354503390892,
      "learning_rate": 6.08916418698702e-06,
      "loss": 0.4667,
      "step": 8491
    },
    {
      "epoch": 0.45825913334412605,
      "grad_norm": 1.2528474275779622,
      "learning_rate": 6.088407852507854e-06,
      "loss": 0.6317,
      "step": 8492
    },
    {
      "epoch": 0.45831309697264044,
      "grad_norm": 1.009060114392703,
      "learning_rate": 6.087651501117252e-06,
      "loss": 0.4385,
      "step": 8493
    },
    {
      "epoch": 0.45836706060115484,
      "grad_norm": 0.9796257034831455,
      "learning_rate": 6.0868951328369505e-06,
      "loss": 0.4888,
      "step": 8494
    },
    {
      "epoch": 0.4584210242296692,
      "grad_norm": 0.8566386081332966,
      "learning_rate": 6.086138747688691e-06,
      "loss": 0.3991,
      "step": 8495
    },
    {
      "epoch": 0.4584749878581836,
      "grad_norm": 0.7232924677461603,
      "learning_rate": 6.085382345694212e-06,
      "loss": 0.3592,
      "step": 8496
    },
    {
      "epoch": 0.458528951486698,
      "grad_norm": 1.0287008324676763,
      "learning_rate": 6.084625926875251e-06,
      "loss": 0.3856,
      "step": 8497
    },
    {
      "epoch": 0.4585829151152123,
      "grad_norm": 0.7519772674964819,
      "learning_rate": 6.083869491253553e-06,
      "loss": 0.2965,
      "step": 8498
    },
    {
      "epoch": 0.4586368787437267,
      "grad_norm": 0.8789876564779575,
      "learning_rate": 6.083113038850854e-06,
      "loss": 0.389,
      "step": 8499
    },
    {
      "epoch": 0.4586908423722411,
      "grad_norm": 0.9747420238012463,
      "learning_rate": 6.082356569688898e-06,
      "loss": 0.4102,
      "step": 8500
    },
    {
      "epoch": 0.4586908423722411,
      "eval_loss": 0.5534877777099609,
      "eval_runtime": 165.1603,
      "eval_samples_per_second": 20.822,
      "eval_steps_per_second": 0.872,
      "step": 8500
    },
    {
      "epoch": 0.4587448060007555,
      "grad_norm": 1.259660922576211,
      "learning_rate": 6.081600083789424e-06,
      "loss": 0.5101,
      "step": 8501
    },
    {
      "epoch": 0.45879876962926985,
      "grad_norm": 0.8878362879094391,
      "learning_rate": 6.080843581174179e-06,
      "loss": 0.428,
      "step": 8502
    },
    {
      "epoch": 0.45885273325778425,
      "grad_norm": 0.9917270658596542,
      "learning_rate": 6.0800870618649e-06,
      "loss": 0.4654,
      "step": 8503
    },
    {
      "epoch": 0.45890669688629865,
      "grad_norm": 1.027263575300223,
      "learning_rate": 6.079330525883333e-06,
      "loss": 0.4326,
      "step": 8504
    },
    {
      "epoch": 0.458960660514813,
      "grad_norm": 1.3970476087036428,
      "learning_rate": 6.0785739732512216e-06,
      "loss": 0.4995,
      "step": 8505
    },
    {
      "epoch": 0.4590146241433274,
      "grad_norm": 0.787434660045765,
      "learning_rate": 6.07781740399031e-06,
      "loss": 0.324,
      "step": 8506
    },
    {
      "epoch": 0.4590685877718418,
      "grad_norm": 0.8530666082730316,
      "learning_rate": 6.077060818122343e-06,
      "loss": 0.3912,
      "step": 8507
    },
    {
      "epoch": 0.4591225514003562,
      "grad_norm": 0.8339493756166022,
      "learning_rate": 6.076304215669063e-06,
      "loss": 0.3573,
      "step": 8508
    },
    {
      "epoch": 0.4591765150288705,
      "grad_norm": 1.2327967212482644,
      "learning_rate": 6.0755475966522196e-06,
      "loss": 0.5945,
      "step": 8509
    },
    {
      "epoch": 0.4592304786573849,
      "grad_norm": 0.9543070845205326,
      "learning_rate": 6.074790961093555e-06,
      "loss": 0.4308,
      "step": 8510
    },
    {
      "epoch": 0.4592844422858993,
      "grad_norm": 1.0061190070855432,
      "learning_rate": 6.074034309014818e-06,
      "loss": 0.3884,
      "step": 8511
    },
    {
      "epoch": 0.45933840591441366,
      "grad_norm": 1.0139016435576889,
      "learning_rate": 6.073277640437754e-06,
      "loss": 0.4559,
      "step": 8512
    },
    {
      "epoch": 0.45939236954292806,
      "grad_norm": 1.1283108096314673,
      "learning_rate": 6.072520955384112e-06,
      "loss": 0.4811,
      "step": 8513
    },
    {
      "epoch": 0.45944633317144246,
      "grad_norm": 1.357177500835346,
      "learning_rate": 6.0717642538756405e-06,
      "loss": 0.6323,
      "step": 8514
    },
    {
      "epoch": 0.45950029679995685,
      "grad_norm": 1.0322337971309237,
      "learning_rate": 6.071007535934085e-06,
      "loss": 0.4207,
      "step": 8515
    },
    {
      "epoch": 0.4595542604284712,
      "grad_norm": 0.9192357711273357,
      "learning_rate": 6.0702508015811975e-06,
      "loss": 0.4171,
      "step": 8516
    },
    {
      "epoch": 0.4596082240569856,
      "grad_norm": 1.1949038511022465,
      "learning_rate": 6.069494050838725e-06,
      "loss": 0.55,
      "step": 8517
    },
    {
      "epoch": 0.4596621876855,
      "grad_norm": 0.9064737040866253,
      "learning_rate": 6.068737283728418e-06,
      "loss": 0.4357,
      "step": 8518
    },
    {
      "epoch": 0.45971615131401433,
      "grad_norm": 1.0812144808727229,
      "learning_rate": 6.067980500272028e-06,
      "loss": 0.5141,
      "step": 8519
    },
    {
      "epoch": 0.45977011494252873,
      "grad_norm": 1.316884031013826,
      "learning_rate": 6.067223700491303e-06,
      "loss": 0.5405,
      "step": 8520
    },
    {
      "epoch": 0.4598240785710431,
      "grad_norm": 1.0552758836854492,
      "learning_rate": 6.066466884407996e-06,
      "loss": 0.3664,
      "step": 8521
    },
    {
      "epoch": 0.4598780421995575,
      "grad_norm": 1.358218242937385,
      "learning_rate": 6.0657100520438575e-06,
      "loss": 0.8129,
      "step": 8522
    },
    {
      "epoch": 0.45993200582807187,
      "grad_norm": 0.9676680314635113,
      "learning_rate": 6.064953203420643e-06,
      "loss": 0.3972,
      "step": 8523
    },
    {
      "epoch": 0.45998596945658626,
      "grad_norm": 0.9761665341879637,
      "learning_rate": 6.0641963385601e-06,
      "loss": 0.4191,
      "step": 8524
    },
    {
      "epoch": 0.46003993308510066,
      "grad_norm": 1.1934176106614867,
      "learning_rate": 6.063439457483985e-06,
      "loss": 0.6565,
      "step": 8525
    },
    {
      "epoch": 0.460093896713615,
      "grad_norm": 0.7876575221085872,
      "learning_rate": 6.062682560214053e-06,
      "loss": 0.3919,
      "step": 8526
    },
    {
      "epoch": 0.4601478603421294,
      "grad_norm": 1.0183255491891021,
      "learning_rate": 6.061925646772052e-06,
      "loss": 0.5872,
      "step": 8527
    },
    {
      "epoch": 0.4602018239706438,
      "grad_norm": 0.999810071179452,
      "learning_rate": 6.061168717179742e-06,
      "loss": 0.4011,
      "step": 8528
    },
    {
      "epoch": 0.4602557875991582,
      "grad_norm": 0.8695218284031195,
      "learning_rate": 6.060411771458875e-06,
      "loss": 0.3683,
      "step": 8529
    },
    {
      "epoch": 0.46030975122767254,
      "grad_norm": 0.8227514524752766,
      "learning_rate": 6.0596548096312094e-06,
      "loss": 0.3394,
      "step": 8530
    },
    {
      "epoch": 0.46036371485618693,
      "grad_norm": 1.0136288952084063,
      "learning_rate": 6.058897831718497e-06,
      "loss": 0.5219,
      "step": 8531
    },
    {
      "epoch": 0.46041767848470133,
      "grad_norm": 0.9860509814102855,
      "learning_rate": 6.0581408377424975e-06,
      "loss": 0.466,
      "step": 8532
    },
    {
      "epoch": 0.4604716421132157,
      "grad_norm": 0.9043191050195626,
      "learning_rate": 6.057383827724967e-06,
      "loss": 0.3647,
      "step": 8533
    },
    {
      "epoch": 0.46052560574173007,
      "grad_norm": 1.1274046832432478,
      "learning_rate": 6.056626801687659e-06,
      "loss": 0.5511,
      "step": 8534
    },
    {
      "epoch": 0.46057956937024447,
      "grad_norm": 0.9500314073784283,
      "learning_rate": 6.055869759652337e-06,
      "loss": 0.4194,
      "step": 8535
    },
    {
      "epoch": 0.4606335329987588,
      "grad_norm": 1.078723657235146,
      "learning_rate": 6.055112701640755e-06,
      "loss": 0.4527,
      "step": 8536
    },
    {
      "epoch": 0.4606874966272732,
      "grad_norm": 1.0509973134061226,
      "learning_rate": 6.054355627674675e-06,
      "loss": 0.5997,
      "step": 8537
    },
    {
      "epoch": 0.4607414602557876,
      "grad_norm": 1.0567412526116664,
      "learning_rate": 6.0535985377758526e-06,
      "loss": 0.605,
      "step": 8538
    },
    {
      "epoch": 0.460795423884302,
      "grad_norm": 1.0501482710398908,
      "learning_rate": 6.052841431966049e-06,
      "loss": 0.4637,
      "step": 8539
    },
    {
      "epoch": 0.46084938751281634,
      "grad_norm": 0.742983539175541,
      "learning_rate": 6.052084310267024e-06,
      "loss": 0.28,
      "step": 8540
    },
    {
      "epoch": 0.46090335114133074,
      "grad_norm": 0.98456093525691,
      "learning_rate": 6.051327172700538e-06,
      "loss": 0.553,
      "step": 8541
    },
    {
      "epoch": 0.46095731476984514,
      "grad_norm": 0.9686361649333369,
      "learning_rate": 6.050570019288352e-06,
      "loss": 0.4231,
      "step": 8542
    },
    {
      "epoch": 0.4610112783983595,
      "grad_norm": 0.9895918654959686,
      "learning_rate": 6.049812850052228e-06,
      "loss": 0.367,
      "step": 8543
    },
    {
      "epoch": 0.4610652420268739,
      "grad_norm": 1.1165659280192266,
      "learning_rate": 6.049055665013928e-06,
      "loss": 0.4272,
      "step": 8544
    },
    {
      "epoch": 0.4611192056553883,
      "grad_norm": 0.9276420419133237,
      "learning_rate": 6.048298464195211e-06,
      "loss": 0.423,
      "step": 8545
    },
    {
      "epoch": 0.4611731692839027,
      "grad_norm": 1.2714418871058233,
      "learning_rate": 6.047541247617842e-06,
      "loss": 0.5129,
      "step": 8546
    },
    {
      "epoch": 0.461227132912417,
      "grad_norm": 1.077623074923214,
      "learning_rate": 6.046784015303587e-06,
      "loss": 0.5662,
      "step": 8547
    },
    {
      "epoch": 0.4612810965409314,
      "grad_norm": 1.029902406998799,
      "learning_rate": 6.046026767274205e-06,
      "loss": 0.6473,
      "step": 8548
    },
    {
      "epoch": 0.4613350601694458,
      "grad_norm": 0.9062009741714702,
      "learning_rate": 6.045269503551462e-06,
      "loss": 0.3871,
      "step": 8549
    },
    {
      "epoch": 0.46138902379796015,
      "grad_norm": 1.0724087982549324,
      "learning_rate": 6.044512224157124e-06,
      "loss": 0.5333,
      "step": 8550
    },
    {
      "epoch": 0.46144298742647455,
      "grad_norm": 0.9772232457464044,
      "learning_rate": 6.0437549291129535e-06,
      "loss": 0.4184,
      "step": 8551
    },
    {
      "epoch": 0.46149695105498895,
      "grad_norm": 1.2521652587631042,
      "learning_rate": 6.042997618440717e-06,
      "loss": 0.5943,
      "step": 8552
    },
    {
      "epoch": 0.46155091468350334,
      "grad_norm": 1.0367157104738083,
      "learning_rate": 6.04224029216218e-06,
      "loss": 0.4879,
      "step": 8553
    },
    {
      "epoch": 0.4616048783120177,
      "grad_norm": 0.9224847184811149,
      "learning_rate": 6.04148295029911e-06,
      "loss": 0.3864,
      "step": 8554
    },
    {
      "epoch": 0.4616588419405321,
      "grad_norm": 1.2778128402413684,
      "learning_rate": 6.040725592873271e-06,
      "loss": 0.6752,
      "step": 8555
    },
    {
      "epoch": 0.4617128055690465,
      "grad_norm": 1.0600632123132452,
      "learning_rate": 6.0399682199064335e-06,
      "loss": 0.4502,
      "step": 8556
    },
    {
      "epoch": 0.4617667691975608,
      "grad_norm": 0.9617921555886252,
      "learning_rate": 6.039210831420364e-06,
      "loss": 0.3858,
      "step": 8557
    },
    {
      "epoch": 0.4618207328260752,
      "grad_norm": 1.0125221631731958,
      "learning_rate": 6.03845342743683e-06,
      "loss": 0.4271,
      "step": 8558
    },
    {
      "epoch": 0.4618746964545896,
      "grad_norm": 0.9749110099984198,
      "learning_rate": 6.0376960079776e-06,
      "loss": 0.4663,
      "step": 8559
    },
    {
      "epoch": 0.461928660083104,
      "grad_norm": 1.0148470171082413,
      "learning_rate": 6.036938573064444e-06,
      "loss": 0.3924,
      "step": 8560
    },
    {
      "epoch": 0.46198262371161836,
      "grad_norm": 1.0651568150976634,
      "learning_rate": 6.03618112271913e-06,
      "loss": 0.5021,
      "step": 8561
    },
    {
      "epoch": 0.46203658734013275,
      "grad_norm": 0.8561499505579699,
      "learning_rate": 6.03542365696343e-06,
      "loss": 0.425,
      "step": 8562
    },
    {
      "epoch": 0.46209055096864715,
      "grad_norm": 1.1554439844180773,
      "learning_rate": 6.034666175819112e-06,
      "loss": 0.6158,
      "step": 8563
    },
    {
      "epoch": 0.4621445145971615,
      "grad_norm": 1.042240371432287,
      "learning_rate": 6.033908679307948e-06,
      "loss": 0.4431,
      "step": 8564
    },
    {
      "epoch": 0.4621984782256759,
      "grad_norm": 0.9429062317031697,
      "learning_rate": 6.03315116745171e-06,
      "loss": 0.4238,
      "step": 8565
    },
    {
      "epoch": 0.4622524418541903,
      "grad_norm": 1.038968178225938,
      "learning_rate": 6.032393640272168e-06,
      "loss": 0.5498,
      "step": 8566
    },
    {
      "epoch": 0.46230640548270463,
      "grad_norm": 0.9996180026665001,
      "learning_rate": 6.031636097791095e-06,
      "loss": 0.4259,
      "step": 8567
    },
    {
      "epoch": 0.462360369111219,
      "grad_norm": 1.4088311085072287,
      "learning_rate": 6.030878540030263e-06,
      "loss": 0.5896,
      "step": 8568
    },
    {
      "epoch": 0.4624143327397334,
      "grad_norm": 1.2196789446862595,
      "learning_rate": 6.0301209670114455e-06,
      "loss": 0.57,
      "step": 8569
    },
    {
      "epoch": 0.4624682963682478,
      "grad_norm": 1.069297016181253,
      "learning_rate": 6.029363378756416e-06,
      "loss": 0.4972,
      "step": 8570
    },
    {
      "epoch": 0.46252225999676216,
      "grad_norm": 0.847628714537642,
      "learning_rate": 6.028605775286949e-06,
      "loss": 0.4613,
      "step": 8571
    },
    {
      "epoch": 0.46257622362527656,
      "grad_norm": 0.9614153166371943,
      "learning_rate": 6.027848156624817e-06,
      "loss": 0.4015,
      "step": 8572
    },
    {
      "epoch": 0.46263018725379096,
      "grad_norm": 1.194854146475304,
      "learning_rate": 6.027090522791795e-06,
      "loss": 0.5305,
      "step": 8573
    },
    {
      "epoch": 0.4626841508823053,
      "grad_norm": 0.9591179197863521,
      "learning_rate": 6.0263328738096605e-06,
      "loss": 0.4321,
      "step": 8574
    },
    {
      "epoch": 0.4627381145108197,
      "grad_norm": 1.1022285839550956,
      "learning_rate": 6.025575209700185e-06,
      "loss": 0.4578,
      "step": 8575
    },
    {
      "epoch": 0.4627920781393341,
      "grad_norm": 1.1976645143557847,
      "learning_rate": 6.0248175304851485e-06,
      "loss": 0.4736,
      "step": 8576
    },
    {
      "epoch": 0.4628460417678485,
      "grad_norm": 0.8064290764371849,
      "learning_rate": 6.024059836186325e-06,
      "loss": 0.3826,
      "step": 8577
    },
    {
      "epoch": 0.46290000539636283,
      "grad_norm": 1.21187762228379,
      "learning_rate": 6.0233021268254945e-06,
      "loss": 0.6037,
      "step": 8578
    },
    {
      "epoch": 0.46295396902487723,
      "grad_norm": 1.000005090807188,
      "learning_rate": 6.02254440242443e-06,
      "loss": 0.4842,
      "step": 8579
    },
    {
      "epoch": 0.46300793265339163,
      "grad_norm": 1.0915912759227546,
      "learning_rate": 6.021786663004911e-06,
      "loss": 0.4485,
      "step": 8580
    },
    {
      "epoch": 0.46306189628190597,
      "grad_norm": 1.1506182195244363,
      "learning_rate": 6.021028908588718e-06,
      "loss": 0.4641,
      "step": 8581
    },
    {
      "epoch": 0.46311585991042037,
      "grad_norm": 1.306168161121709,
      "learning_rate": 6.0202711391976266e-06,
      "loss": 0.6068,
      "step": 8582
    },
    {
      "epoch": 0.46316982353893477,
      "grad_norm": 0.8050353487310665,
      "learning_rate": 6.019513354853417e-06,
      "loss": 0.3136,
      "step": 8583
    },
    {
      "epoch": 0.46322378716744916,
      "grad_norm": 1.1093392832045035,
      "learning_rate": 6.0187555555778684e-06,
      "loss": 0.5977,
      "step": 8584
    },
    {
      "epoch": 0.4632777507959635,
      "grad_norm": 1.0378005311193506,
      "learning_rate": 6.0179977413927615e-06,
      "loss": 0.4783,
      "step": 8585
    },
    {
      "epoch": 0.4633317144244779,
      "grad_norm": 0.8708008323294189,
      "learning_rate": 6.017239912319876e-06,
      "loss": 0.3866,
      "step": 8586
    },
    {
      "epoch": 0.4633856780529923,
      "grad_norm": 0.8506874453996385,
      "learning_rate": 6.0164820683809925e-06,
      "loss": 0.4495,
      "step": 8587
    },
    {
      "epoch": 0.46343964168150664,
      "grad_norm": 1.1436534590196734,
      "learning_rate": 6.015724209597892e-06,
      "loss": 0.54,
      "step": 8588
    },
    {
      "epoch": 0.46349360531002104,
      "grad_norm": 1.0292667667921005,
      "learning_rate": 6.014966335992357e-06,
      "loss": 0.3911,
      "step": 8589
    },
    {
      "epoch": 0.46354756893853544,
      "grad_norm": 1.143158253477684,
      "learning_rate": 6.01420844758617e-06,
      "loss": 0.4598,
      "step": 8590
    },
    {
      "epoch": 0.46360153256704983,
      "grad_norm": 1.1126650464912493,
      "learning_rate": 6.013450544401112e-06,
      "loss": 0.4511,
      "step": 8591
    },
    {
      "epoch": 0.4636554961955642,
      "grad_norm": 0.9293196319675322,
      "learning_rate": 6.012692626458966e-06,
      "loss": 0.4007,
      "step": 8592
    },
    {
      "epoch": 0.4637094598240786,
      "grad_norm": 1.1344476813603983,
      "learning_rate": 6.011934693781516e-06,
      "loss": 0.6098,
      "step": 8593
    },
    {
      "epoch": 0.46376342345259297,
      "grad_norm": 1.0066490869710114,
      "learning_rate": 6.011176746390546e-06,
      "loss": 0.5155,
      "step": 8594
    },
    {
      "epoch": 0.4638173870811073,
      "grad_norm": 1.147548677211478,
      "learning_rate": 6.010418784307841e-06,
      "loss": 0.5037,
      "step": 8595
    },
    {
      "epoch": 0.4638713507096217,
      "grad_norm": 0.9827103486089183,
      "learning_rate": 6.0096608075551836e-06,
      "loss": 0.6119,
      "step": 8596
    },
    {
      "epoch": 0.4639253143381361,
      "grad_norm": 1.0262904440112586,
      "learning_rate": 6.0089028161543584e-06,
      "loss": 0.5174,
      "step": 8597
    },
    {
      "epoch": 0.4639792779666505,
      "grad_norm": 0.8985165450390238,
      "learning_rate": 6.008144810127152e-06,
      "loss": 0.4626,
      "step": 8598
    },
    {
      "epoch": 0.46403324159516485,
      "grad_norm": 0.8878817196548485,
      "learning_rate": 6.007386789495353e-06,
      "loss": 0.3629,
      "step": 8599
    },
    {
      "epoch": 0.46408720522367924,
      "grad_norm": 1.0643114841731385,
      "learning_rate": 6.006628754280743e-06,
      "loss": 0.4279,
      "step": 8600
    },
    {
      "epoch": 0.46414116885219364,
      "grad_norm": 1.1400574849213891,
      "learning_rate": 6.005870704505111e-06,
      "loss": 0.503,
      "step": 8601
    },
    {
      "epoch": 0.464195132480708,
      "grad_norm": 0.9674012203029255,
      "learning_rate": 6.005112640190245e-06,
      "loss": 0.3654,
      "step": 8602
    },
    {
      "epoch": 0.4642490961092224,
      "grad_norm": 0.8927207597936652,
      "learning_rate": 6.004354561357929e-06,
      "loss": 0.3698,
      "step": 8603
    },
    {
      "epoch": 0.4643030597377368,
      "grad_norm": 1.016120966265503,
      "learning_rate": 6.003596468029955e-06,
      "loss": 0.4365,
      "step": 8604
    },
    {
      "epoch": 0.4643570233662511,
      "grad_norm": 0.9689884486453738,
      "learning_rate": 6.002838360228109e-06,
      "loss": 0.4496,
      "step": 8605
    },
    {
      "epoch": 0.4644109869947655,
      "grad_norm": 1.0609104025785,
      "learning_rate": 6.00208023797418e-06,
      "loss": 0.5653,
      "step": 8606
    },
    {
      "epoch": 0.4644649506232799,
      "grad_norm": 1.017684071468246,
      "learning_rate": 6.001322101289958e-06,
      "loss": 0.4175,
      "step": 8607
    },
    {
      "epoch": 0.4645189142517943,
      "grad_norm": 0.9377182470974565,
      "learning_rate": 6.000563950197232e-06,
      "loss": 0.4425,
      "step": 8608
    },
    {
      "epoch": 0.46457287788030865,
      "grad_norm": 1.2566410031905575,
      "learning_rate": 5.999805784717795e-06,
      "loss": 0.6786,
      "step": 8609
    },
    {
      "epoch": 0.46462684150882305,
      "grad_norm": 1.1302888762913454,
      "learning_rate": 5.9990476048734315e-06,
      "loss": 0.4598,
      "step": 8610
    },
    {
      "epoch": 0.46468080513733745,
      "grad_norm": 0.971848437441932,
      "learning_rate": 5.998289410685936e-06,
      "loss": 0.4333,
      "step": 8611
    },
    {
      "epoch": 0.4647347687658518,
      "grad_norm": 0.8410969303393462,
      "learning_rate": 5.997531202177101e-06,
      "loss": 0.3987,
      "step": 8612
    },
    {
      "epoch": 0.4647887323943662,
      "grad_norm": 0.9409245952108177,
      "learning_rate": 5.996772979368715e-06,
      "loss": 0.5592,
      "step": 8613
    },
    {
      "epoch": 0.4648426960228806,
      "grad_norm": 0.9086144598938792,
      "learning_rate": 5.996014742282572e-06,
      "loss": 0.4397,
      "step": 8614
    },
    {
      "epoch": 0.464896659651395,
      "grad_norm": 1.0321295483225879,
      "learning_rate": 5.9952564909404644e-06,
      "loss": 0.4302,
      "step": 8615
    },
    {
      "epoch": 0.4649506232799093,
      "grad_norm": 0.9587372313047655,
      "learning_rate": 5.994498225364185e-06,
      "loss": 0.4664,
      "step": 8616
    },
    {
      "epoch": 0.4650045869084237,
      "grad_norm": 0.9677270038644902,
      "learning_rate": 5.993739945575525e-06,
      "loss": 0.487,
      "step": 8617
    },
    {
      "epoch": 0.4650585505369381,
      "grad_norm": 0.9520085588485381,
      "learning_rate": 5.992981651596281e-06,
      "loss": 0.3724,
      "step": 8618
    },
    {
      "epoch": 0.46511251416545246,
      "grad_norm": 1.1850861090941784,
      "learning_rate": 5.992223343448248e-06,
      "loss": 0.5672,
      "step": 8619
    },
    {
      "epoch": 0.46516647779396686,
      "grad_norm": 0.9174174007326882,
      "learning_rate": 5.9914650211532175e-06,
      "loss": 0.4056,
      "step": 8620
    },
    {
      "epoch": 0.46522044142248126,
      "grad_norm": 0.8492182845552501,
      "learning_rate": 5.990706684732986e-06,
      "loss": 0.3057,
      "step": 8621
    },
    {
      "epoch": 0.46527440505099565,
      "grad_norm": 0.9928685979559576,
      "learning_rate": 5.989948334209348e-06,
      "loss": 0.5669,
      "step": 8622
    },
    {
      "epoch": 0.46532836867951,
      "grad_norm": 0.8756728017447322,
      "learning_rate": 5.9891899696041e-06,
      "loss": 0.3668,
      "step": 8623
    },
    {
      "epoch": 0.4653823323080244,
      "grad_norm": 1.3894544617329678,
      "learning_rate": 5.988431590939039e-06,
      "loss": 0.6959,
      "step": 8624
    },
    {
      "epoch": 0.4654362959365388,
      "grad_norm": 1.1278402905927059,
      "learning_rate": 5.9876731982359595e-06,
      "loss": 0.6408,
      "step": 8625
    },
    {
      "epoch": 0.46549025956505313,
      "grad_norm": 0.9755040935430775,
      "learning_rate": 5.986914791516661e-06,
      "loss": 0.4474,
      "step": 8626
    },
    {
      "epoch": 0.46554422319356753,
      "grad_norm": 1.0804591375034982,
      "learning_rate": 5.986156370802937e-06,
      "loss": 0.6488,
      "step": 8627
    },
    {
      "epoch": 0.4655981868220819,
      "grad_norm": 0.7571997016695278,
      "learning_rate": 5.9853979361165895e-06,
      "loss": 0.4047,
      "step": 8628
    },
    {
      "epoch": 0.4656521504505963,
      "grad_norm": 0.959641176439372,
      "learning_rate": 5.984639487479413e-06,
      "loss": 0.4029,
      "step": 8629
    },
    {
      "epoch": 0.46570611407911067,
      "grad_norm": 0.8377408388289379,
      "learning_rate": 5.983881024913209e-06,
      "loss": 0.4099,
      "step": 8630
    },
    {
      "epoch": 0.46576007770762506,
      "grad_norm": 1.1169547614008717,
      "learning_rate": 5.983122548439776e-06,
      "loss": 0.5216,
      "step": 8631
    },
    {
      "epoch": 0.46581404133613946,
      "grad_norm": 1.1271155065417413,
      "learning_rate": 5.982364058080913e-06,
      "loss": 0.4254,
      "step": 8632
    },
    {
      "epoch": 0.4658680049646538,
      "grad_norm": 1.045926416685868,
      "learning_rate": 5.9816055538584185e-06,
      "loss": 0.4406,
      "step": 8633
    },
    {
      "epoch": 0.4659219685931682,
      "grad_norm": 1.0132631284220903,
      "learning_rate": 5.980847035794095e-06,
      "loss": 0.4225,
      "step": 8634
    },
    {
      "epoch": 0.4659759322216826,
      "grad_norm": 1.3212278499133725,
      "learning_rate": 5.98008850390974e-06,
      "loss": 0.6129,
      "step": 8635
    },
    {
      "epoch": 0.46602989585019694,
      "grad_norm": 0.9909720703523236,
      "learning_rate": 5.979329958227158e-06,
      "loss": 0.6022,
      "step": 8636
    },
    {
      "epoch": 0.46608385947871134,
      "grad_norm": 1.013249023681921,
      "learning_rate": 5.978571398768148e-06,
      "loss": 0.5471,
      "step": 8637
    },
    {
      "epoch": 0.46613782310722573,
      "grad_norm": 0.9600659820943548,
      "learning_rate": 5.977812825554512e-06,
      "loss": 0.4279,
      "step": 8638
    },
    {
      "epoch": 0.46619178673574013,
      "grad_norm": 1.0294527471016084,
      "learning_rate": 5.977054238608053e-06,
      "loss": 0.4615,
      "step": 8639
    },
    {
      "epoch": 0.4662457503642545,
      "grad_norm": 0.9970590549086533,
      "learning_rate": 5.976295637950573e-06,
      "loss": 0.4488,
      "step": 8640
    },
    {
      "epoch": 0.46629971399276887,
      "grad_norm": 1.1568941941632573,
      "learning_rate": 5.975537023603874e-06,
      "loss": 0.4906,
      "step": 8641
    },
    {
      "epoch": 0.46635367762128327,
      "grad_norm": 1.1536367244848806,
      "learning_rate": 5.974778395589762e-06,
      "loss": 0.5514,
      "step": 8642
    },
    {
      "epoch": 0.4664076412497976,
      "grad_norm": 1.106082519604652,
      "learning_rate": 5.9740197539300395e-06,
      "loss": 0.4758,
      "step": 8643
    },
    {
      "epoch": 0.466461604878312,
      "grad_norm": 0.9092684144906557,
      "learning_rate": 5.973261098646509e-06,
      "loss": 0.3812,
      "step": 8644
    },
    {
      "epoch": 0.4665155685068264,
      "grad_norm": 1.1398591100678763,
      "learning_rate": 5.972502429760977e-06,
      "loss": 0.6398,
      "step": 8645
    },
    {
      "epoch": 0.4665695321353408,
      "grad_norm": 0.91177618828031,
      "learning_rate": 5.971743747295248e-06,
      "loss": 0.4378,
      "step": 8646
    },
    {
      "epoch": 0.46662349576385515,
      "grad_norm": 0.9769211124444062,
      "learning_rate": 5.970985051271126e-06,
      "loss": 0.4691,
      "step": 8647
    },
    {
      "epoch": 0.46667745939236954,
      "grad_norm": 0.8792803592308595,
      "learning_rate": 5.970226341710419e-06,
      "loss": 0.4188,
      "step": 8648
    },
    {
      "epoch": 0.46673142302088394,
      "grad_norm": 1.2153514875671256,
      "learning_rate": 5.969467618634932e-06,
      "loss": 0.671,
      "step": 8649
    },
    {
      "epoch": 0.4667853866493983,
      "grad_norm": 1.134562666337781,
      "learning_rate": 5.968708882066471e-06,
      "loss": 0.4859,
      "step": 8650
    },
    {
      "epoch": 0.4668393502779127,
      "grad_norm": 1.0308161043834947,
      "learning_rate": 5.967950132026842e-06,
      "loss": 0.5612,
      "step": 8651
    },
    {
      "epoch": 0.4668933139064271,
      "grad_norm": 1.2681787615113551,
      "learning_rate": 5.967191368537854e-06,
      "loss": 0.5963,
      "step": 8652
    },
    {
      "epoch": 0.4669472775349415,
      "grad_norm": 0.9695833567676125,
      "learning_rate": 5.966432591621314e-06,
      "loss": 0.5549,
      "step": 8653
    },
    {
      "epoch": 0.4670012411634558,
      "grad_norm": 0.9944462198205192,
      "learning_rate": 5.965673801299031e-06,
      "loss": 0.4263,
      "step": 8654
    },
    {
      "epoch": 0.4670552047919702,
      "grad_norm": 1.211725473564839,
      "learning_rate": 5.964914997592811e-06,
      "loss": 0.8016,
      "step": 8655
    },
    {
      "epoch": 0.4671091684204846,
      "grad_norm": 1.1130634568421225,
      "learning_rate": 5.964156180524465e-06,
      "loss": 0.4792,
      "step": 8656
    },
    {
      "epoch": 0.46716313204899895,
      "grad_norm": 0.9272633091932245,
      "learning_rate": 5.9633973501158015e-06,
      "loss": 0.3858,
      "step": 8657
    },
    {
      "epoch": 0.46721709567751335,
      "grad_norm": 1.111980051623778,
      "learning_rate": 5.962638506388631e-06,
      "loss": 0.5334,
      "step": 8658
    },
    {
      "epoch": 0.46727105930602775,
      "grad_norm": 0.9959655793165458,
      "learning_rate": 5.96187964936476e-06,
      "loss": 0.3664,
      "step": 8659
    },
    {
      "epoch": 0.46732502293454214,
      "grad_norm": 0.9450880319227349,
      "learning_rate": 5.961120779066002e-06,
      "loss": 0.4359,
      "step": 8660
    },
    {
      "epoch": 0.4673789865630565,
      "grad_norm": 0.7740723716997817,
      "learning_rate": 5.960361895514168e-06,
      "loss": 0.3293,
      "step": 8661
    },
    {
      "epoch": 0.4674329501915709,
      "grad_norm": 1.1629766557514967,
      "learning_rate": 5.959602998731066e-06,
      "loss": 0.3929,
      "step": 8662
    },
    {
      "epoch": 0.4674869138200853,
      "grad_norm": 1.0484223106265693,
      "learning_rate": 5.9588440887385105e-06,
      "loss": 0.4318,
      "step": 8663
    },
    {
      "epoch": 0.4675408774485996,
      "grad_norm": 1.070297364540619,
      "learning_rate": 5.958085165558312e-06,
      "loss": 0.6086,
      "step": 8664
    },
    {
      "epoch": 0.467594841077114,
      "grad_norm": 1.0386263023486975,
      "learning_rate": 5.957326229212283e-06,
      "loss": 0.5537,
      "step": 8665
    },
    {
      "epoch": 0.4676488047056284,
      "grad_norm": 0.9901979181556742,
      "learning_rate": 5.9565672797222365e-06,
      "loss": 0.3861,
      "step": 8666
    },
    {
      "epoch": 0.4677027683341428,
      "grad_norm": 0.9169226527247082,
      "learning_rate": 5.955808317109985e-06,
      "loss": 0.3829,
      "step": 8667
    },
    {
      "epoch": 0.46775673196265716,
      "grad_norm": 0.9014195737659311,
      "learning_rate": 5.9550493413973405e-06,
      "loss": 0.4217,
      "step": 8668
    },
    {
      "epoch": 0.46781069559117155,
      "grad_norm": 1.0898136639518043,
      "learning_rate": 5.9542903526061195e-06,
      "loss": 0.5368,
      "step": 8669
    },
    {
      "epoch": 0.46786465921968595,
      "grad_norm": 1.08831008164724,
      "learning_rate": 5.9535313507581346e-06,
      "loss": 0.4792,
      "step": 8670
    },
    {
      "epoch": 0.4679186228482003,
      "grad_norm": 1.0856731585625872,
      "learning_rate": 5.952772335875201e-06,
      "loss": 0.4616,
      "step": 8671
    },
    {
      "epoch": 0.4679725864767147,
      "grad_norm": 1.0994229227965029,
      "learning_rate": 5.952013307979132e-06,
      "loss": 0.4915,
      "step": 8672
    },
    {
      "epoch": 0.4680265501052291,
      "grad_norm": 1.2464641631293885,
      "learning_rate": 5.951254267091744e-06,
      "loss": 0.5125,
      "step": 8673
    },
    {
      "epoch": 0.46808051373374343,
      "grad_norm": 0.9589187534829755,
      "learning_rate": 5.950495213234853e-06,
      "loss": 0.4439,
      "step": 8674
    },
    {
      "epoch": 0.46813447736225783,
      "grad_norm": 1.052962174195603,
      "learning_rate": 5.949736146430274e-06,
      "loss": 0.419,
      "step": 8675
    },
    {
      "epoch": 0.4681884409907722,
      "grad_norm": 0.9779969625861619,
      "learning_rate": 5.9489770666998234e-06,
      "loss": 0.452,
      "step": 8676
    },
    {
      "epoch": 0.4682424046192866,
      "grad_norm": 1.075875119603226,
      "learning_rate": 5.9482179740653175e-06,
      "loss": 0.5157,
      "step": 8677
    },
    {
      "epoch": 0.46829636824780096,
      "grad_norm": 1.0908216055182927,
      "learning_rate": 5.947458868548577e-06,
      "loss": 0.5152,
      "step": 8678
    },
    {
      "epoch": 0.46835033187631536,
      "grad_norm": 1.1153021726561794,
      "learning_rate": 5.946699750171414e-06,
      "loss": 0.5496,
      "step": 8679
    },
    {
      "epoch": 0.46840429550482976,
      "grad_norm": 1.056936103234692,
      "learning_rate": 5.945940618955649e-06,
      "loss": 0.5754,
      "step": 8680
    },
    {
      "epoch": 0.4684582591333441,
      "grad_norm": 1.0353464689814496,
      "learning_rate": 5.9451814749231005e-06,
      "loss": 0.5063,
      "step": 8681
    },
    {
      "epoch": 0.4685122227618585,
      "grad_norm": 1.1607496901134593,
      "learning_rate": 5.944422318095588e-06,
      "loss": 0.5439,
      "step": 8682
    },
    {
      "epoch": 0.4685661863903729,
      "grad_norm": 0.9532401056867751,
      "learning_rate": 5.943663148494926e-06,
      "loss": 0.4003,
      "step": 8683
    },
    {
      "epoch": 0.4686201500188873,
      "grad_norm": 1.0164448079897388,
      "learning_rate": 5.942903966142939e-06,
      "loss": 0.4375,
      "step": 8684
    },
    {
      "epoch": 0.46867411364740164,
      "grad_norm": 1.0201298576660234,
      "learning_rate": 5.942144771061445e-06,
      "loss": 0.4595,
      "step": 8685
    },
    {
      "epoch": 0.46872807727591603,
      "grad_norm": 1.10065015961312,
      "learning_rate": 5.941385563272262e-06,
      "loss": 0.3963,
      "step": 8686
    },
    {
      "epoch": 0.46878204090443043,
      "grad_norm": 1.1932535709773886,
      "learning_rate": 5.9406263427972126e-06,
      "loss": 0.6041,
      "step": 8687
    },
    {
      "epoch": 0.4688360045329448,
      "grad_norm": 0.8615032633122801,
      "learning_rate": 5.939867109658117e-06,
      "loss": 0.3318,
      "step": 8688
    },
    {
      "epoch": 0.46888996816145917,
      "grad_norm": 1.0298053379530727,
      "learning_rate": 5.939107863876795e-06,
      "loss": 0.4638,
      "step": 8689
    },
    {
      "epoch": 0.46894393178997357,
      "grad_norm": 0.9034787322488955,
      "learning_rate": 5.938348605475071e-06,
      "loss": 0.3577,
      "step": 8690
    },
    {
      "epoch": 0.46899789541848796,
      "grad_norm": 1.0053786197675298,
      "learning_rate": 5.937589334474763e-06,
      "loss": 0.4011,
      "step": 8691
    },
    {
      "epoch": 0.4690518590470023,
      "grad_norm": 1.2246981481002477,
      "learning_rate": 5.936830050897697e-06,
      "loss": 0.545,
      "step": 8692
    },
    {
      "epoch": 0.4691058226755167,
      "grad_norm": 1.0296028200638963,
      "learning_rate": 5.936070754765693e-06,
      "loss": 0.4039,
      "step": 8693
    },
    {
      "epoch": 0.4691597863040311,
      "grad_norm": 0.8887519278600245,
      "learning_rate": 5.935311446100574e-06,
      "loss": 0.3496,
      "step": 8694
    },
    {
      "epoch": 0.46921374993254544,
      "grad_norm": 0.9470570573745989,
      "learning_rate": 5.9345521249241674e-06,
      "loss": 0.4731,
      "step": 8695
    },
    {
      "epoch": 0.46926771356105984,
      "grad_norm": 0.8639015663189205,
      "learning_rate": 5.933792791258293e-06,
      "loss": 0.3973,
      "step": 8696
    },
    {
      "epoch": 0.46932167718957424,
      "grad_norm": 1.0628684888056679,
      "learning_rate": 5.933033445124773e-06,
      "loss": 0.5114,
      "step": 8697
    },
    {
      "epoch": 0.46937564081808864,
      "grad_norm": 1.0055262234465847,
      "learning_rate": 5.932274086545435e-06,
      "loss": 0.4691,
      "step": 8698
    },
    {
      "epoch": 0.469429604446603,
      "grad_norm": 0.8454228379320959,
      "learning_rate": 5.9315147155421045e-06,
      "loss": 0.318,
      "step": 8699
    },
    {
      "epoch": 0.4694835680751174,
      "grad_norm": 0.9329160293069266,
      "learning_rate": 5.930755332136604e-06,
      "loss": 0.456,
      "step": 8700
    },
    {
      "epoch": 0.46953753170363177,
      "grad_norm": 0.9703792854480903,
      "learning_rate": 5.929995936350761e-06,
      "loss": 0.4318,
      "step": 8701
    },
    {
      "epoch": 0.4695914953321461,
      "grad_norm": 0.911391456820878,
      "learning_rate": 5.929236528206399e-06,
      "loss": 0.4449,
      "step": 8702
    },
    {
      "epoch": 0.4696454589606605,
      "grad_norm": 1.2299975917880814,
      "learning_rate": 5.928477107725347e-06,
      "loss": 0.6134,
      "step": 8703
    },
    {
      "epoch": 0.4696994225891749,
      "grad_norm": 0.8713363324081187,
      "learning_rate": 5.927717674929428e-06,
      "loss": 0.3967,
      "step": 8704
    },
    {
      "epoch": 0.46975338621768925,
      "grad_norm": 1.0527699428703525,
      "learning_rate": 5.926958229840473e-06,
      "loss": 0.3955,
      "step": 8705
    },
    {
      "epoch": 0.46980734984620365,
      "grad_norm": 0.956870204832798,
      "learning_rate": 5.926198772480305e-06,
      "loss": 0.3996,
      "step": 8706
    },
    {
      "epoch": 0.46986131347471805,
      "grad_norm": 0.9865029857600905,
      "learning_rate": 5.925439302870755e-06,
      "loss": 0.4696,
      "step": 8707
    },
    {
      "epoch": 0.46991527710323244,
      "grad_norm": 1.102505368236341,
      "learning_rate": 5.924679821033649e-06,
      "loss": 0.7191,
      "step": 8708
    },
    {
      "epoch": 0.4699692407317468,
      "grad_norm": 0.8619072770788635,
      "learning_rate": 5.923920326990816e-06,
      "loss": 0.3307,
      "step": 8709
    },
    {
      "epoch": 0.4700232043602612,
      "grad_norm": 1.031311878318018,
      "learning_rate": 5.923160820764083e-06,
      "loss": 0.46,
      "step": 8710
    },
    {
      "epoch": 0.4700771679887756,
      "grad_norm": 1.033092906980219,
      "learning_rate": 5.922401302375282e-06,
      "loss": 0.5759,
      "step": 8711
    },
    {
      "epoch": 0.4701311316172899,
      "grad_norm": 0.9717009675402507,
      "learning_rate": 5.921641771846241e-06,
      "loss": 0.5654,
      "step": 8712
    },
    {
      "epoch": 0.4701850952458043,
      "grad_norm": 1.0111446128673538,
      "learning_rate": 5.92088222919879e-06,
      "loss": 0.4698,
      "step": 8713
    },
    {
      "epoch": 0.4702390588743187,
      "grad_norm": 1.0409147179508291,
      "learning_rate": 5.920122674454757e-06,
      "loss": 0.4988,
      "step": 8714
    },
    {
      "epoch": 0.4702930225028331,
      "grad_norm": 0.8807493532515743,
      "learning_rate": 5.919363107635975e-06,
      "loss": 0.4112,
      "step": 8715
    },
    {
      "epoch": 0.47034698613134746,
      "grad_norm": 0.9305971199727108,
      "learning_rate": 5.9186035287642715e-06,
      "loss": 0.4356,
      "step": 8716
    },
    {
      "epoch": 0.47040094975986185,
      "grad_norm": 1.14520039606566,
      "learning_rate": 5.9178439378614816e-06,
      "loss": 0.3361,
      "step": 8717
    },
    {
      "epoch": 0.47045491338837625,
      "grad_norm": 1.160185690780033,
      "learning_rate": 5.917084334949435e-06,
      "loss": 0.4936,
      "step": 8718
    },
    {
      "epoch": 0.4705088770168906,
      "grad_norm": 0.9328728690203343,
      "learning_rate": 5.916324720049961e-06,
      "loss": 0.4566,
      "step": 8719
    },
    {
      "epoch": 0.470562840645405,
      "grad_norm": 1.1490939620644454,
      "learning_rate": 5.9155650931848955e-06,
      "loss": 0.5615,
      "step": 8720
    },
    {
      "epoch": 0.4706168042739194,
      "grad_norm": 1.0537279684521474,
      "learning_rate": 5.914805454376069e-06,
      "loss": 0.4399,
      "step": 8721
    },
    {
      "epoch": 0.4706707679024338,
      "grad_norm": 1.0606354504826565,
      "learning_rate": 5.9140458036453144e-06,
      "loss": 0.5231,
      "step": 8722
    },
    {
      "epoch": 0.4707247315309481,
      "grad_norm": 1.182499202142192,
      "learning_rate": 5.913286141014465e-06,
      "loss": 0.7202,
      "step": 8723
    },
    {
      "epoch": 0.4707786951594625,
      "grad_norm": 1.0442918064632247,
      "learning_rate": 5.912526466505355e-06,
      "loss": 0.5068,
      "step": 8724
    },
    {
      "epoch": 0.4708326587879769,
      "grad_norm": 1.0746182252657717,
      "learning_rate": 5.911766780139817e-06,
      "loss": 0.4413,
      "step": 8725
    },
    {
      "epoch": 0.47088662241649126,
      "grad_norm": 1.2134429067828767,
      "learning_rate": 5.911007081939686e-06,
      "loss": 0.6278,
      "step": 8726
    },
    {
      "epoch": 0.47094058604500566,
      "grad_norm": 1.0318070352898485,
      "learning_rate": 5.910247371926795e-06,
      "loss": 0.4455,
      "step": 8727
    },
    {
      "epoch": 0.47099454967352006,
      "grad_norm": 1.1686103154068637,
      "learning_rate": 5.909487650122981e-06,
      "loss": 0.4437,
      "step": 8728
    },
    {
      "epoch": 0.47104851330203446,
      "grad_norm": 0.9316429964173001,
      "learning_rate": 5.9087279165500775e-06,
      "loss": 0.3522,
      "step": 8729
    },
    {
      "epoch": 0.4711024769305488,
      "grad_norm": 1.1070740778675323,
      "learning_rate": 5.907968171229923e-06,
      "loss": 0.6205,
      "step": 8730
    },
    {
      "epoch": 0.4711564405590632,
      "grad_norm": 0.8607908448060101,
      "learning_rate": 5.907208414184349e-06,
      "loss": 0.3616,
      "step": 8731
    },
    {
      "epoch": 0.4712104041875776,
      "grad_norm": 0.6657758524236569,
      "learning_rate": 5.906448645435195e-06,
      "loss": 0.3562,
      "step": 8732
    },
    {
      "epoch": 0.47126436781609193,
      "grad_norm": 1.0550156162622841,
      "learning_rate": 5.905688865004295e-06,
      "loss": 0.5256,
      "step": 8733
    },
    {
      "epoch": 0.47131833144460633,
      "grad_norm": 1.0044533736145986,
      "learning_rate": 5.9049290729134865e-06,
      "loss": 0.4409,
      "step": 8734
    },
    {
      "epoch": 0.47137229507312073,
      "grad_norm": 1.0346477928542372,
      "learning_rate": 5.904169269184609e-06,
      "loss": 0.3886,
      "step": 8735
    },
    {
      "epoch": 0.47142625870163507,
      "grad_norm": 1.2445807670001123,
      "learning_rate": 5.903409453839498e-06,
      "loss": 0.6073,
      "step": 8736
    },
    {
      "epoch": 0.47148022233014947,
      "grad_norm": 0.9606007460620334,
      "learning_rate": 5.902649626899991e-06,
      "loss": 0.4071,
      "step": 8737
    },
    {
      "epoch": 0.47153418595866387,
      "grad_norm": 1.0416415081183634,
      "learning_rate": 5.901889788387927e-06,
      "loss": 0.59,
      "step": 8738
    },
    {
      "epoch": 0.47158814958717826,
      "grad_norm": 1.090679098085388,
      "learning_rate": 5.901129938325143e-06,
      "loss": 0.6608,
      "step": 8739
    },
    {
      "epoch": 0.4716421132156926,
      "grad_norm": 0.9906550294828262,
      "learning_rate": 5.900370076733483e-06,
      "loss": 0.4756,
      "step": 8740
    },
    {
      "epoch": 0.471696076844207,
      "grad_norm": 0.8664855916541196,
      "learning_rate": 5.89961020363478e-06,
      "loss": 0.3704,
      "step": 8741
    },
    {
      "epoch": 0.4717500404727214,
      "grad_norm": 1.0384096293963592,
      "learning_rate": 5.898850319050875e-06,
      "loss": 0.4755,
      "step": 8742
    },
    {
      "epoch": 0.47180400410123574,
      "grad_norm": 0.8807543215386913,
      "learning_rate": 5.898090423003609e-06,
      "loss": 0.4151,
      "step": 8743
    },
    {
      "epoch": 0.47185796772975014,
      "grad_norm": 1.026807749795398,
      "learning_rate": 5.897330515514823e-06,
      "loss": 0.4386,
      "step": 8744
    },
    {
      "epoch": 0.47191193135826454,
      "grad_norm": 1.045354845483807,
      "learning_rate": 5.896570596606357e-06,
      "loss": 0.5102,
      "step": 8745
    },
    {
      "epoch": 0.47196589498677893,
      "grad_norm": 1.1406799114762336,
      "learning_rate": 5.895810666300051e-06,
      "loss": 0.43,
      "step": 8746
    },
    {
      "epoch": 0.4720198586152933,
      "grad_norm": 0.9189202221858719,
      "learning_rate": 5.895050724617747e-06,
      "loss": 0.4319,
      "step": 8747
    },
    {
      "epoch": 0.4720738222438077,
      "grad_norm": 0.9277152913673969,
      "learning_rate": 5.894290771581285e-06,
      "loss": 0.4071,
      "step": 8748
    },
    {
      "epoch": 0.47212778587232207,
      "grad_norm": 0.8011755125156785,
      "learning_rate": 5.893530807212508e-06,
      "loss": 0.3745,
      "step": 8749
    },
    {
      "epoch": 0.4721817495008364,
      "grad_norm": 1.1348237593190482,
      "learning_rate": 5.892770831533258e-06,
      "loss": 0.5047,
      "step": 8750
    },
    {
      "epoch": 0.4722357131293508,
      "grad_norm": 0.8834888466402445,
      "learning_rate": 5.892010844565379e-06,
      "loss": 0.4554,
      "step": 8751
    },
    {
      "epoch": 0.4722896767578652,
      "grad_norm": 1.1351987211800108,
      "learning_rate": 5.891250846330709e-06,
      "loss": 0.4732,
      "step": 8752
    },
    {
      "epoch": 0.4723436403863796,
      "grad_norm": 0.9316793659205794,
      "learning_rate": 5.890490836851096e-06,
      "loss": 0.3597,
      "step": 8753
    },
    {
      "epoch": 0.47239760401489395,
      "grad_norm": 1.049573675622062,
      "learning_rate": 5.889730816148382e-06,
      "loss": 0.3616,
      "step": 8754
    },
    {
      "epoch": 0.47245156764340834,
      "grad_norm": 1.0981922241342856,
      "learning_rate": 5.888970784244411e-06,
      "loss": 0.4813,
      "step": 8755
    },
    {
      "epoch": 0.47250553127192274,
      "grad_norm": 0.9289055139216408,
      "learning_rate": 5.888210741161027e-06,
      "loss": 0.4903,
      "step": 8756
    },
    {
      "epoch": 0.4725594949004371,
      "grad_norm": 0.9748833317215747,
      "learning_rate": 5.887450686920074e-06,
      "loss": 0.4475,
      "step": 8757
    },
    {
      "epoch": 0.4726134585289515,
      "grad_norm": 1.1077311348093375,
      "learning_rate": 5.886690621543395e-06,
      "loss": 0.4738,
      "step": 8758
    },
    {
      "epoch": 0.4726674221574659,
      "grad_norm": 1.1282389816139495,
      "learning_rate": 5.885930545052838e-06,
      "loss": 0.5652,
      "step": 8759
    },
    {
      "epoch": 0.4727213857859803,
      "grad_norm": 0.9665546034335349,
      "learning_rate": 5.885170457470246e-06,
      "loss": 0.5715,
      "step": 8760
    },
    {
      "epoch": 0.4727753494144946,
      "grad_norm": 1.1470002795291008,
      "learning_rate": 5.884410358817467e-06,
      "loss": 0.5121,
      "step": 8761
    },
    {
      "epoch": 0.472829313043009,
      "grad_norm": 1.0676241257440153,
      "learning_rate": 5.883650249116346e-06,
      "loss": 0.5323,
      "step": 8762
    },
    {
      "epoch": 0.4728832766715234,
      "grad_norm": 1.1827647929069065,
      "learning_rate": 5.882890128388729e-06,
      "loss": 0.5387,
      "step": 8763
    },
    {
      "epoch": 0.47293724030003775,
      "grad_norm": 1.1355278077546374,
      "learning_rate": 5.882129996656464e-06,
      "loss": 0.4728,
      "step": 8764
    },
    {
      "epoch": 0.47299120392855215,
      "grad_norm": 1.1107131982025897,
      "learning_rate": 5.881369853941394e-06,
      "loss": 0.5407,
      "step": 8765
    },
    {
      "epoch": 0.47304516755706655,
      "grad_norm": 1.086083125628493,
      "learning_rate": 5.8806097002653705e-06,
      "loss": 0.4038,
      "step": 8766
    },
    {
      "epoch": 0.47309913118558095,
      "grad_norm": 1.2543490257033674,
      "learning_rate": 5.879849535650239e-06,
      "loss": 0.7451,
      "step": 8767
    },
    {
      "epoch": 0.4731530948140953,
      "grad_norm": 1.0975104520520291,
      "learning_rate": 5.879089360117849e-06,
      "loss": 0.5605,
      "step": 8768
    },
    {
      "epoch": 0.4732070584426097,
      "grad_norm": 1.1561746000599853,
      "learning_rate": 5.878329173690047e-06,
      "loss": 0.5764,
      "step": 8769
    },
    {
      "epoch": 0.4732610220711241,
      "grad_norm": 1.0525536988902107,
      "learning_rate": 5.877568976388683e-06,
      "loss": 0.565,
      "step": 8770
    },
    {
      "epoch": 0.4733149856996384,
      "grad_norm": 0.8676917791022474,
      "learning_rate": 5.8768087682356054e-06,
      "loss": 0.3781,
      "step": 8771
    },
    {
      "epoch": 0.4733689493281528,
      "grad_norm": 1.0211882753078003,
      "learning_rate": 5.876048549252662e-06,
      "loss": 0.5628,
      "step": 8772
    },
    {
      "epoch": 0.4734229129566672,
      "grad_norm": 0.9702473802518192,
      "learning_rate": 5.875288319461704e-06,
      "loss": 0.3836,
      "step": 8773
    },
    {
      "epoch": 0.47347687658518156,
      "grad_norm": 1.045512879272203,
      "learning_rate": 5.87452807888458e-06,
      "loss": 0.613,
      "step": 8774
    },
    {
      "epoch": 0.47353084021369596,
      "grad_norm": 0.8589056271062326,
      "learning_rate": 5.873767827543142e-06,
      "loss": 0.3663,
      "step": 8775
    },
    {
      "epoch": 0.47358480384221036,
      "grad_norm": 1.038677143323233,
      "learning_rate": 5.873007565459238e-06,
      "loss": 0.4317,
      "step": 8776
    },
    {
      "epoch": 0.47363876747072475,
      "grad_norm": 0.9065100783460267,
      "learning_rate": 5.872247292654718e-06,
      "loss": 0.4689,
      "step": 8777
    },
    {
      "epoch": 0.4736927310992391,
      "grad_norm": 0.9402962827237474,
      "learning_rate": 5.871487009151438e-06,
      "loss": 0.365,
      "step": 8778
    },
    {
      "epoch": 0.4737466947277535,
      "grad_norm": 1.1068375045079477,
      "learning_rate": 5.870726714971243e-06,
      "loss": 0.5468,
      "step": 8779
    },
    {
      "epoch": 0.4738006583562679,
      "grad_norm": 1.1018901936084629,
      "learning_rate": 5.869966410135989e-06,
      "loss": 0.4568,
      "step": 8780
    },
    {
      "epoch": 0.47385462198478223,
      "grad_norm": 1.0685733288575598,
      "learning_rate": 5.869206094667526e-06,
      "loss": 0.3838,
      "step": 8781
    },
    {
      "epoch": 0.47390858561329663,
      "grad_norm": 0.7847559028156487,
      "learning_rate": 5.868445768587708e-06,
      "loss": 0.3136,
      "step": 8782
    },
    {
      "epoch": 0.473962549241811,
      "grad_norm": 0.8654011210424204,
      "learning_rate": 5.867685431918385e-06,
      "loss": 0.4267,
      "step": 8783
    },
    {
      "epoch": 0.4740165128703254,
      "grad_norm": 0.9764441472458033,
      "learning_rate": 5.866925084681412e-06,
      "loss": 0.4245,
      "step": 8784
    },
    {
      "epoch": 0.47407047649883977,
      "grad_norm": 1.3181882997220347,
      "learning_rate": 5.866164726898642e-06,
      "loss": 0.707,
      "step": 8785
    },
    {
      "epoch": 0.47412444012735416,
      "grad_norm": 1.0049553244972391,
      "learning_rate": 5.865404358591927e-06,
      "loss": 0.4488,
      "step": 8786
    },
    {
      "epoch": 0.47417840375586856,
      "grad_norm": 0.9071038064302532,
      "learning_rate": 5.86464397978312e-06,
      "loss": 0.4201,
      "step": 8787
    },
    {
      "epoch": 0.4742323673843829,
      "grad_norm": 0.9515277722318186,
      "learning_rate": 5.863883590494078e-06,
      "loss": 0.5063,
      "step": 8788
    },
    {
      "epoch": 0.4742863310128973,
      "grad_norm": 0.9897699471133655,
      "learning_rate": 5.863123190746655e-06,
      "loss": 0.4586,
      "step": 8789
    },
    {
      "epoch": 0.4743402946414117,
      "grad_norm": 0.9092217277370972,
      "learning_rate": 5.862362780562703e-06,
      "loss": 0.4236,
      "step": 8790
    },
    {
      "epoch": 0.4743942582699261,
      "grad_norm": 0.6363771073935938,
      "learning_rate": 5.861602359964078e-06,
      "loss": 0.2239,
      "step": 8791
    },
    {
      "epoch": 0.47444822189844044,
      "grad_norm": 1.199072511816314,
      "learning_rate": 5.8608419289726356e-06,
      "loss": 0.6411,
      "step": 8792
    },
    {
      "epoch": 0.47450218552695483,
      "grad_norm": 0.9913110526765556,
      "learning_rate": 5.860081487610234e-06,
      "loss": 0.4923,
      "step": 8793
    },
    {
      "epoch": 0.47455614915546923,
      "grad_norm": 1.114043437313138,
      "learning_rate": 5.859321035898725e-06,
      "loss": 0.4765,
      "step": 8794
    },
    {
      "epoch": 0.4746101127839836,
      "grad_norm": 1.1520958352645954,
      "learning_rate": 5.858560573859967e-06,
      "loss": 0.5756,
      "step": 8795
    },
    {
      "epoch": 0.47466407641249797,
      "grad_norm": 1.027757942422187,
      "learning_rate": 5.857800101515815e-06,
      "loss": 0.4738,
      "step": 8796
    },
    {
      "epoch": 0.47471804004101237,
      "grad_norm": 0.9624334918723162,
      "learning_rate": 5.857039618888126e-06,
      "loss": 0.3564,
      "step": 8797
    },
    {
      "epoch": 0.47477200366952677,
      "grad_norm": 0.7870307953188028,
      "learning_rate": 5.856279125998759e-06,
      "loss": 0.3258,
      "step": 8798
    },
    {
      "epoch": 0.4748259672980411,
      "grad_norm": 0.8278499175913212,
      "learning_rate": 5.85551862286957e-06,
      "loss": 0.4551,
      "step": 8799
    },
    {
      "epoch": 0.4748799309265555,
      "grad_norm": 1.0707484917174876,
      "learning_rate": 5.854758109522416e-06,
      "loss": 0.4567,
      "step": 8800
    },
    {
      "epoch": 0.4749338945550699,
      "grad_norm": 1.3592703504371253,
      "learning_rate": 5.853997585979155e-06,
      "loss": 0.5649,
      "step": 8801
    },
    {
      "epoch": 0.47498785818358424,
      "grad_norm": 1.1069057732051715,
      "learning_rate": 5.853237052261648e-06,
      "loss": 0.6157,
      "step": 8802
    },
    {
      "epoch": 0.47504182181209864,
      "grad_norm": 0.8948879096422329,
      "learning_rate": 5.852476508391748e-06,
      "loss": 0.4407,
      "step": 8803
    },
    {
      "epoch": 0.47509578544061304,
      "grad_norm": 1.2106875010054567,
      "learning_rate": 5.8517159543913185e-06,
      "loss": 0.5007,
      "step": 8804
    },
    {
      "epoch": 0.4751497490691274,
      "grad_norm": 0.8598634972787603,
      "learning_rate": 5.850955390282218e-06,
      "loss": 0.3567,
      "step": 8805
    },
    {
      "epoch": 0.4752037126976418,
      "grad_norm": 0.89823265893177,
      "learning_rate": 5.8501948160863035e-06,
      "loss": 0.3608,
      "step": 8806
    },
    {
      "epoch": 0.4752576763261562,
      "grad_norm": 0.9711783303233786,
      "learning_rate": 5.8494342318254375e-06,
      "loss": 0.3961,
      "step": 8807
    },
    {
      "epoch": 0.4753116399546706,
      "grad_norm": 0.9930348829191323,
      "learning_rate": 5.848673637521477e-06,
      "loss": 0.482,
      "step": 8808
    },
    {
      "epoch": 0.4753656035831849,
      "grad_norm": 0.9192731867855013,
      "learning_rate": 5.847913033196285e-06,
      "loss": 0.3906,
      "step": 8809
    },
    {
      "epoch": 0.4754195672116993,
      "grad_norm": 0.948206606991972,
      "learning_rate": 5.84715241887172e-06,
      "loss": 0.4848,
      "step": 8810
    },
    {
      "epoch": 0.4754735308402137,
      "grad_norm": 1.0680382514799773,
      "learning_rate": 5.846391794569643e-06,
      "loss": 0.5713,
      "step": 8811
    },
    {
      "epoch": 0.47552749446872805,
      "grad_norm": 1.1390566520511052,
      "learning_rate": 5.845631160311919e-06,
      "loss": 0.5061,
      "step": 8812
    },
    {
      "epoch": 0.47558145809724245,
      "grad_norm": 0.7545615613501151,
      "learning_rate": 5.844870516120405e-06,
      "loss": 0.3131,
      "step": 8813
    },
    {
      "epoch": 0.47563542172575685,
      "grad_norm": 0.843464459593225,
      "learning_rate": 5.844109862016962e-06,
      "loss": 0.3776,
      "step": 8814
    },
    {
      "epoch": 0.47568938535427124,
      "grad_norm": 1.0688402726424056,
      "learning_rate": 5.843349198023455e-06,
      "loss": 0.626,
      "step": 8815
    },
    {
      "epoch": 0.4757433489827856,
      "grad_norm": 0.9552895495145576,
      "learning_rate": 5.842588524161745e-06,
      "loss": 0.4727,
      "step": 8816
    },
    {
      "epoch": 0.4757973126113,
      "grad_norm": 0.9293578122797493,
      "learning_rate": 5.841827840453695e-06,
      "loss": 0.394,
      "step": 8817
    },
    {
      "epoch": 0.4758512762398144,
      "grad_norm": 1.0782008322830479,
      "learning_rate": 5.841067146921167e-06,
      "loss": 0.4722,
      "step": 8818
    },
    {
      "epoch": 0.4759052398683287,
      "grad_norm": 0.9676852045502109,
      "learning_rate": 5.840306443586024e-06,
      "loss": 0.527,
      "step": 8819
    },
    {
      "epoch": 0.4759592034968431,
      "grad_norm": 0.9034596292845017,
      "learning_rate": 5.83954573047013e-06,
      "loss": 0.3565,
      "step": 8820
    },
    {
      "epoch": 0.4760131671253575,
      "grad_norm": 1.147294163414714,
      "learning_rate": 5.83878500759535e-06,
      "loss": 0.4802,
      "step": 8821
    },
    {
      "epoch": 0.4760671307538719,
      "grad_norm": 1.0050032962143893,
      "learning_rate": 5.838024274983545e-06,
      "loss": 0.4135,
      "step": 8822
    },
    {
      "epoch": 0.47612109438238626,
      "grad_norm": 1.0925049397520792,
      "learning_rate": 5.837263532656583e-06,
      "loss": 0.6075,
      "step": 8823
    },
    {
      "epoch": 0.47617505801090065,
      "grad_norm": 0.9550104830098863,
      "learning_rate": 5.836502780636325e-06,
      "loss": 0.4455,
      "step": 8824
    },
    {
      "epoch": 0.47622902163941505,
      "grad_norm": 1.171530761690344,
      "learning_rate": 5.835742018944637e-06,
      "loss": 0.4156,
      "step": 8825
    },
    {
      "epoch": 0.4762829852679294,
      "grad_norm": 0.8162749723093464,
      "learning_rate": 5.834981247603384e-06,
      "loss": 0.3656,
      "step": 8826
    },
    {
      "epoch": 0.4763369488964438,
      "grad_norm": 0.9758331264386221,
      "learning_rate": 5.834220466634431e-06,
      "loss": 0.429,
      "step": 8827
    },
    {
      "epoch": 0.4763909125249582,
      "grad_norm": 1.0787803110694003,
      "learning_rate": 5.8334596760596455e-06,
      "loss": 0.4664,
      "step": 8828
    },
    {
      "epoch": 0.4764448761534726,
      "grad_norm": 0.9959427480544031,
      "learning_rate": 5.832698875900891e-06,
      "loss": 0.4272,
      "step": 8829
    },
    {
      "epoch": 0.4764988397819869,
      "grad_norm": 0.901218094320874,
      "learning_rate": 5.831938066180036e-06,
      "loss": 0.4958,
      "step": 8830
    },
    {
      "epoch": 0.4765528034105013,
      "grad_norm": 1.0304045555586219,
      "learning_rate": 5.831177246918944e-06,
      "loss": 0.5764,
      "step": 8831
    },
    {
      "epoch": 0.4766067670390157,
      "grad_norm": 1.234312144147616,
      "learning_rate": 5.830416418139484e-06,
      "loss": 0.7064,
      "step": 8832
    },
    {
      "epoch": 0.47666073066753006,
      "grad_norm": 0.9952506306494329,
      "learning_rate": 5.829655579863522e-06,
      "loss": 0.3757,
      "step": 8833
    },
    {
      "epoch": 0.47671469429604446,
      "grad_norm": 0.863998548222133,
      "learning_rate": 5.828894732112925e-06,
      "loss": 0.3876,
      "step": 8834
    },
    {
      "epoch": 0.47676865792455886,
      "grad_norm": 0.9270008079067744,
      "learning_rate": 5.828133874909562e-06,
      "loss": 0.4776,
      "step": 8835
    },
    {
      "epoch": 0.47682262155307326,
      "grad_norm": 0.8174685301978881,
      "learning_rate": 5.827373008275299e-06,
      "loss": 0.3228,
      "step": 8836
    },
    {
      "epoch": 0.4768765851815876,
      "grad_norm": 0.9579412923260727,
      "learning_rate": 5.826612132232007e-06,
      "loss": 0.4507,
      "step": 8837
    },
    {
      "epoch": 0.476930548810102,
      "grad_norm": 1.024534861713695,
      "learning_rate": 5.825851246801551e-06,
      "loss": 0.5037,
      "step": 8838
    },
    {
      "epoch": 0.4769845124386164,
      "grad_norm": 1.0105057738566499,
      "learning_rate": 5.8250903520058e-06,
      "loss": 0.4007,
      "step": 8839
    },
    {
      "epoch": 0.47703847606713073,
      "grad_norm": 0.9023289093199663,
      "learning_rate": 5.824329447866625e-06,
      "loss": 0.4965,
      "step": 8840
    },
    {
      "epoch": 0.47709243969564513,
      "grad_norm": 1.0848484974574562,
      "learning_rate": 5.823568534405893e-06,
      "loss": 0.5123,
      "step": 8841
    },
    {
      "epoch": 0.47714640332415953,
      "grad_norm": 0.831128821514199,
      "learning_rate": 5.822807611645475e-06,
      "loss": 0.3673,
      "step": 8842
    },
    {
      "epoch": 0.47720036695267387,
      "grad_norm": 0.9704444334724761,
      "learning_rate": 5.8220466796072395e-06,
      "loss": 0.427,
      "step": 8843
    },
    {
      "epoch": 0.47725433058118827,
      "grad_norm": 0.8256573141878355,
      "learning_rate": 5.821285738313057e-06,
      "loss": 0.3416,
      "step": 8844
    },
    {
      "epoch": 0.47730829420970267,
      "grad_norm": 0.9464244578589652,
      "learning_rate": 5.820524787784798e-06,
      "loss": 0.454,
      "step": 8845
    },
    {
      "epoch": 0.47736225783821706,
      "grad_norm": 1.0101224134531765,
      "learning_rate": 5.819763828044334e-06,
      "loss": 0.468,
      "step": 8846
    },
    {
      "epoch": 0.4774162214667314,
      "grad_norm": 0.8547167430240799,
      "learning_rate": 5.819002859113533e-06,
      "loss": 0.3831,
      "step": 8847
    },
    {
      "epoch": 0.4774701850952458,
      "grad_norm": 0.9272011612944001,
      "learning_rate": 5.818241881014267e-06,
      "loss": 0.5051,
      "step": 8848
    },
    {
      "epoch": 0.4775241487237602,
      "grad_norm": 1.1679801913986676,
      "learning_rate": 5.8174808937684095e-06,
      "loss": 0.4902,
      "step": 8849
    },
    {
      "epoch": 0.47757811235227454,
      "grad_norm": 1.1501105898356971,
      "learning_rate": 5.816719897397829e-06,
      "loss": 0.509,
      "step": 8850
    },
    {
      "epoch": 0.47763207598078894,
      "grad_norm": 1.2881002030575865,
      "learning_rate": 5.8159588919244e-06,
      "loss": 0.6213,
      "step": 8851
    },
    {
      "epoch": 0.47768603960930334,
      "grad_norm": 0.9764972360754548,
      "learning_rate": 5.815197877369991e-06,
      "loss": 0.4409,
      "step": 8852
    },
    {
      "epoch": 0.47774000323781773,
      "grad_norm": 0.9469126734973976,
      "learning_rate": 5.814436853756479e-06,
      "loss": 0.5842,
      "step": 8853
    },
    {
      "epoch": 0.4777939668663321,
      "grad_norm": 0.7807368992074802,
      "learning_rate": 5.813675821105733e-06,
      "loss": 0.3347,
      "step": 8854
    },
    {
      "epoch": 0.4778479304948465,
      "grad_norm": 1.0268379417114026,
      "learning_rate": 5.812914779439626e-06,
      "loss": 0.5405,
      "step": 8855
    },
    {
      "epoch": 0.47790189412336087,
      "grad_norm": 1.0826772298220964,
      "learning_rate": 5.812153728780034e-06,
      "loss": 0.4706,
      "step": 8856
    },
    {
      "epoch": 0.4779558577518752,
      "grad_norm": 1.0563077415437647,
      "learning_rate": 5.8113926691488274e-06,
      "loss": 0.6322,
      "step": 8857
    },
    {
      "epoch": 0.4780098213803896,
      "grad_norm": 0.783275506876491,
      "learning_rate": 5.810631600567881e-06,
      "loss": 0.3034,
      "step": 8858
    },
    {
      "epoch": 0.478063785008904,
      "grad_norm": 1.055601073925802,
      "learning_rate": 5.809870523059068e-06,
      "loss": 0.6073,
      "step": 8859
    },
    {
      "epoch": 0.4781177486374184,
      "grad_norm": 1.3724594431857093,
      "learning_rate": 5.809109436644263e-06,
      "loss": 0.6055,
      "step": 8860
    },
    {
      "epoch": 0.47817171226593275,
      "grad_norm": 0.8948912670954919,
      "learning_rate": 5.808348341345341e-06,
      "loss": 0.3601,
      "step": 8861
    },
    {
      "epoch": 0.47822567589444714,
      "grad_norm": 1.1041785679653497,
      "learning_rate": 5.807587237184176e-06,
      "loss": 0.4815,
      "step": 8862
    },
    {
      "epoch": 0.47827963952296154,
      "grad_norm": 0.9840051277294879,
      "learning_rate": 5.806826124182643e-06,
      "loss": 0.5368,
      "step": 8863
    },
    {
      "epoch": 0.4783336031514759,
      "grad_norm": 0.9680965164665438,
      "learning_rate": 5.806065002362618e-06,
      "loss": 0.382,
      "step": 8864
    },
    {
      "epoch": 0.4783875667799903,
      "grad_norm": 1.2248253460092964,
      "learning_rate": 5.8053038717459754e-06,
      "loss": 0.5901,
      "step": 8865
    },
    {
      "epoch": 0.4784415304085047,
      "grad_norm": 0.8389630193506976,
      "learning_rate": 5.80454273235459e-06,
      "loss": 0.3816,
      "step": 8866
    },
    {
      "epoch": 0.4784954940370191,
      "grad_norm": 0.7991333292661946,
      "learning_rate": 5.803781584210338e-06,
      "loss": 0.3301,
      "step": 8867
    },
    {
      "epoch": 0.4785494576655334,
      "grad_norm": 0.9073954995637065,
      "learning_rate": 5.803020427335099e-06,
      "loss": 0.479,
      "step": 8868
    },
    {
      "epoch": 0.4786034212940478,
      "grad_norm": 0.9065046598187955,
      "learning_rate": 5.802259261750745e-06,
      "loss": 0.364,
      "step": 8869
    },
    {
      "epoch": 0.4786573849225622,
      "grad_norm": 1.0245535135295143,
      "learning_rate": 5.801498087479154e-06,
      "loss": 0.445,
      "step": 8870
    },
    {
      "epoch": 0.47871134855107655,
      "grad_norm": 0.8779865305973908,
      "learning_rate": 5.800736904542205e-06,
      "loss": 0.4138,
      "step": 8871
    },
    {
      "epoch": 0.47876531217959095,
      "grad_norm": 1.0852887459607508,
      "learning_rate": 5.799975712961771e-06,
      "loss": 0.6283,
      "step": 8872
    },
    {
      "epoch": 0.47881927580810535,
      "grad_norm": 0.8733702788836811,
      "learning_rate": 5.799214512759733e-06,
      "loss": 0.4734,
      "step": 8873
    },
    {
      "epoch": 0.4788732394366197,
      "grad_norm": 1.0456845555660599,
      "learning_rate": 5.798453303957968e-06,
      "loss": 0.4103,
      "step": 8874
    },
    {
      "epoch": 0.4789272030651341,
      "grad_norm": 1.2275330415314751,
      "learning_rate": 5.797692086578354e-06,
      "loss": 0.5701,
      "step": 8875
    },
    {
      "epoch": 0.4789811666936485,
      "grad_norm": 1.1256097613380478,
      "learning_rate": 5.796930860642768e-06,
      "loss": 0.5225,
      "step": 8876
    },
    {
      "epoch": 0.4790351303221629,
      "grad_norm": 1.1349965115451373,
      "learning_rate": 5.79616962617309e-06,
      "loss": 0.4443,
      "step": 8877
    },
    {
      "epoch": 0.4790890939506772,
      "grad_norm": 0.9681892691867793,
      "learning_rate": 5.795408383191196e-06,
      "loss": 0.5399,
      "step": 8878
    },
    {
      "epoch": 0.4791430575791916,
      "grad_norm": 1.00383575244915,
      "learning_rate": 5.794647131718967e-06,
      "loss": 0.4244,
      "step": 8879
    },
    {
      "epoch": 0.479197021207706,
      "grad_norm": 0.92672660583232,
      "learning_rate": 5.793885871778282e-06,
      "loss": 0.4541,
      "step": 8880
    },
    {
      "epoch": 0.47925098483622036,
      "grad_norm": 0.9288889071343785,
      "learning_rate": 5.79312460339102e-06,
      "loss": 0.4369,
      "step": 8881
    },
    {
      "epoch": 0.47930494846473476,
      "grad_norm": 1.0633120284826474,
      "learning_rate": 5.792363326579061e-06,
      "loss": 0.5531,
      "step": 8882
    },
    {
      "epoch": 0.47935891209324916,
      "grad_norm": 0.8600137114115991,
      "learning_rate": 5.791602041364285e-06,
      "loss": 0.3956,
      "step": 8883
    },
    {
      "epoch": 0.47941287572176355,
      "grad_norm": 1.0490335759406204,
      "learning_rate": 5.79084074776857e-06,
      "loss": 0.6398,
      "step": 8884
    },
    {
      "epoch": 0.4794668393502779,
      "grad_norm": 0.9509894936400564,
      "learning_rate": 5.790079445813801e-06,
      "loss": 0.4231,
      "step": 8885
    },
    {
      "epoch": 0.4795208029787923,
      "grad_norm": 1.0457517323743697,
      "learning_rate": 5.789318135521853e-06,
      "loss": 0.5844,
      "step": 8886
    },
    {
      "epoch": 0.4795747666073067,
      "grad_norm": 0.8268696246696647,
      "learning_rate": 5.78855681691461e-06,
      "loss": 0.3996,
      "step": 8887
    },
    {
      "epoch": 0.47962873023582103,
      "grad_norm": 0.9871129221685812,
      "learning_rate": 5.787795490013954e-06,
      "loss": 0.5157,
      "step": 8888
    },
    {
      "epoch": 0.47968269386433543,
      "grad_norm": 1.129557657669863,
      "learning_rate": 5.787034154841763e-06,
      "loss": 0.4982,
      "step": 8889
    },
    {
      "epoch": 0.4797366574928498,
      "grad_norm": 0.7994494212548594,
      "learning_rate": 5.7862728114199205e-06,
      "loss": 0.3599,
      "step": 8890
    },
    {
      "epoch": 0.4797906211213642,
      "grad_norm": 0.9375638558451366,
      "learning_rate": 5.785511459770308e-06,
      "loss": 0.4736,
      "step": 8891
    },
    {
      "epoch": 0.47984458474987857,
      "grad_norm": 1.0015626751652345,
      "learning_rate": 5.784750099914809e-06,
      "loss": 0.4188,
      "step": 8892
    },
    {
      "epoch": 0.47989854837839296,
      "grad_norm": 1.102586695031359,
      "learning_rate": 5.7839887318753026e-06,
      "loss": 0.4733,
      "step": 8893
    },
    {
      "epoch": 0.47995251200690736,
      "grad_norm": 1.1645781076203017,
      "learning_rate": 5.783227355673675e-06,
      "loss": 0.5567,
      "step": 8894
    },
    {
      "epoch": 0.4800064756354217,
      "grad_norm": 0.9574314256299064,
      "learning_rate": 5.782465971331805e-06,
      "loss": 0.3687,
      "step": 8895
    },
    {
      "epoch": 0.4800604392639361,
      "grad_norm": 1.0190531593213792,
      "learning_rate": 5.7817045788715795e-06,
      "loss": 0.5341,
      "step": 8896
    },
    {
      "epoch": 0.4801144028924505,
      "grad_norm": 1.0621539401027396,
      "learning_rate": 5.780943178314878e-06,
      "loss": 0.399,
      "step": 8897
    },
    {
      "epoch": 0.4801683665209649,
      "grad_norm": 0.8507418523449214,
      "learning_rate": 5.780181769683587e-06,
      "loss": 0.3957,
      "step": 8898
    },
    {
      "epoch": 0.48022233014947924,
      "grad_norm": 0.9612091584900498,
      "learning_rate": 5.77942035299959e-06,
      "loss": 0.4325,
      "step": 8899
    },
    {
      "epoch": 0.48027629377799363,
      "grad_norm": 1.011532456107316,
      "learning_rate": 5.778658928284767e-06,
      "loss": 0.5285,
      "step": 8900
    },
    {
      "epoch": 0.48033025740650803,
      "grad_norm": 0.9365141855321122,
      "learning_rate": 5.777897495561007e-06,
      "loss": 0.4502,
      "step": 8901
    },
    {
      "epoch": 0.4803842210350224,
      "grad_norm": 1.0771903439266712,
      "learning_rate": 5.777136054850192e-06,
      "loss": 0.4592,
      "step": 8902
    },
    {
      "epoch": 0.48043818466353677,
      "grad_norm": 1.2452574762112747,
      "learning_rate": 5.776374606174204e-06,
      "loss": 0.5045,
      "step": 8903
    },
    {
      "epoch": 0.48049214829205117,
      "grad_norm": 1.0037633303829268,
      "learning_rate": 5.775613149554934e-06,
      "loss": 0.4026,
      "step": 8904
    },
    {
      "epoch": 0.48054611192056557,
      "grad_norm": 1.0867070098645655,
      "learning_rate": 5.774851685014262e-06,
      "loss": 0.5337,
      "step": 8905
    },
    {
      "epoch": 0.4806000755490799,
      "grad_norm": 0.9195884251987443,
      "learning_rate": 5.774090212574075e-06,
      "loss": 0.392,
      "step": 8906
    },
    {
      "epoch": 0.4806540391775943,
      "grad_norm": 1.1452438588087077,
      "learning_rate": 5.7733287322562595e-06,
      "loss": 0.5456,
      "step": 8907
    },
    {
      "epoch": 0.4807080028061087,
      "grad_norm": 0.9567835244027647,
      "learning_rate": 5.772567244082697e-06,
      "loss": 0.462,
      "step": 8908
    },
    {
      "epoch": 0.48076196643462304,
      "grad_norm": 1.0962937067542196,
      "learning_rate": 5.77180574807528e-06,
      "loss": 0.5603,
      "step": 8909
    },
    {
      "epoch": 0.48081593006313744,
      "grad_norm": 0.9258578759401118,
      "learning_rate": 5.771044244255889e-06,
      "loss": 0.4985,
      "step": 8910
    },
    {
      "epoch": 0.48086989369165184,
      "grad_norm": 0.8311079611091676,
      "learning_rate": 5.7702827326464124e-06,
      "loss": 0.4456,
      "step": 8911
    },
    {
      "epoch": 0.4809238573201662,
      "grad_norm": 1.2334326954154025,
      "learning_rate": 5.769521213268739e-06,
      "loss": 0.529,
      "step": 8912
    },
    {
      "epoch": 0.4809778209486806,
      "grad_norm": 1.1006202110605932,
      "learning_rate": 5.768759686144751e-06,
      "loss": 0.5085,
      "step": 8913
    },
    {
      "epoch": 0.481031784577195,
      "grad_norm": 1.094079420476509,
      "learning_rate": 5.76799815129634e-06,
      "loss": 0.5607,
      "step": 8914
    },
    {
      "epoch": 0.4810857482057094,
      "grad_norm": 1.1132249211671637,
      "learning_rate": 5.76723660874539e-06,
      "loss": 0.4689,
      "step": 8915
    },
    {
      "epoch": 0.4811397118342237,
      "grad_norm": 1.1060881852396283,
      "learning_rate": 5.766475058513792e-06,
      "loss": 0.6089,
      "step": 8916
    },
    {
      "epoch": 0.4811936754627381,
      "grad_norm": 1.0425288247847708,
      "learning_rate": 5.76571350062343e-06,
      "loss": 0.5859,
      "step": 8917
    },
    {
      "epoch": 0.4812476390912525,
      "grad_norm": 1.2397901988894628,
      "learning_rate": 5.764951935096194e-06,
      "loss": 0.578,
      "step": 8918
    },
    {
      "epoch": 0.48130160271976685,
      "grad_norm": 1.172354984819246,
      "learning_rate": 5.764190361953971e-06,
      "loss": 0.5803,
      "step": 8919
    },
    {
      "epoch": 0.48135556634828125,
      "grad_norm": 1.463508585005897,
      "learning_rate": 5.7634287812186495e-06,
      "loss": 0.7346,
      "step": 8920
    },
    {
      "epoch": 0.48140952997679565,
      "grad_norm": 1.1229682052135896,
      "learning_rate": 5.762667192912119e-06,
      "loss": 0.5595,
      "step": 8921
    },
    {
      "epoch": 0.48146349360531004,
      "grad_norm": 0.8261681550016152,
      "learning_rate": 5.761905597056269e-06,
      "loss": 0.3684,
      "step": 8922
    },
    {
      "epoch": 0.4815174572338244,
      "grad_norm": 1.3801487299326027,
      "learning_rate": 5.761143993672988e-06,
      "loss": 0.6978,
      "step": 8923
    },
    {
      "epoch": 0.4815714208623388,
      "grad_norm": 1.2878417445671688,
      "learning_rate": 5.760382382784163e-06,
      "loss": 0.5309,
      "step": 8924
    },
    {
      "epoch": 0.4816253844908532,
      "grad_norm": 1.2579304255340757,
      "learning_rate": 5.759620764411685e-06,
      "loss": 0.6228,
      "step": 8925
    },
    {
      "epoch": 0.4816793481193675,
      "grad_norm": 1.0495904391998365,
      "learning_rate": 5.758859138577445e-06,
      "loss": 0.6004,
      "step": 8926
    },
    {
      "epoch": 0.4817333117478819,
      "grad_norm": 0.9783846015441312,
      "learning_rate": 5.758097505303332e-06,
      "loss": 0.5088,
      "step": 8927
    },
    {
      "epoch": 0.4817872753763963,
      "grad_norm": 1.0104148047712003,
      "learning_rate": 5.7573358646112355e-06,
      "loss": 0.5835,
      "step": 8928
    },
    {
      "epoch": 0.4818412390049107,
      "grad_norm": 1.1334875837146274,
      "learning_rate": 5.756574216523045e-06,
      "loss": 0.4572,
      "step": 8929
    },
    {
      "epoch": 0.48189520263342506,
      "grad_norm": 1.22182243226313,
      "learning_rate": 5.755812561060655e-06,
      "loss": 0.5662,
      "step": 8930
    },
    {
      "epoch": 0.48194916626193945,
      "grad_norm": 0.8294062249920503,
      "learning_rate": 5.755050898245953e-06,
      "loss": 0.4676,
      "step": 8931
    },
    {
      "epoch": 0.48200312989045385,
      "grad_norm": 1.1460688883928374,
      "learning_rate": 5.754289228100828e-06,
      "loss": 0.4912,
      "step": 8932
    },
    {
      "epoch": 0.4820570935189682,
      "grad_norm": 1.1348808406924946,
      "learning_rate": 5.753527550647177e-06,
      "loss": 0.504,
      "step": 8933
    },
    {
      "epoch": 0.4821110571474826,
      "grad_norm": 1.1285900077698594,
      "learning_rate": 5.7527658659068855e-06,
      "loss": 0.7409,
      "step": 8934
    },
    {
      "epoch": 0.482165020775997,
      "grad_norm": 1.2746779428259416,
      "learning_rate": 5.752004173901849e-06,
      "loss": 0.6315,
      "step": 8935
    },
    {
      "epoch": 0.4822189844045114,
      "grad_norm": 1.0116447882994264,
      "learning_rate": 5.751242474653955e-06,
      "loss": 0.4994,
      "step": 8936
    },
    {
      "epoch": 0.48227294803302573,
      "grad_norm": 1.1203678939909256,
      "learning_rate": 5.750480768185103e-06,
      "loss": 0.6108,
      "step": 8937
    },
    {
      "epoch": 0.4823269116615401,
      "grad_norm": 1.0700397519916547,
      "learning_rate": 5.749719054517176e-06,
      "loss": 0.5585,
      "step": 8938
    },
    {
      "epoch": 0.4823808752900545,
      "grad_norm": 0.9071431914159915,
      "learning_rate": 5.748957333672075e-06,
      "loss": 0.4203,
      "step": 8939
    },
    {
      "epoch": 0.48243483891856886,
      "grad_norm": 0.9787891481946263,
      "learning_rate": 5.748195605671687e-06,
      "loss": 0.4756,
      "step": 8940
    },
    {
      "epoch": 0.48248880254708326,
      "grad_norm": 0.9368911078968601,
      "learning_rate": 5.747433870537906e-06,
      "loss": 0.3959,
      "step": 8941
    },
    {
      "epoch": 0.48254276617559766,
      "grad_norm": 1.0210008027774962,
      "learning_rate": 5.746672128292626e-06,
      "loss": 0.4669,
      "step": 8942
    },
    {
      "epoch": 0.482596729804112,
      "grad_norm": 0.9871705550131088,
      "learning_rate": 5.745910378957741e-06,
      "loss": 0.5155,
      "step": 8943
    },
    {
      "epoch": 0.4826506934326264,
      "grad_norm": 1.0248368452345042,
      "learning_rate": 5.745148622555143e-06,
      "loss": 0.414,
      "step": 8944
    },
    {
      "epoch": 0.4827046570611408,
      "grad_norm": 1.1171592727394923,
      "learning_rate": 5.744386859106723e-06,
      "loss": 0.6185,
      "step": 8945
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.9800103183825898,
      "learning_rate": 5.74362508863438e-06,
      "loss": 0.3584,
      "step": 8946
    },
    {
      "epoch": 0.48281258431816954,
      "grad_norm": 1.0649573867502167,
      "learning_rate": 5.7428633111600055e-06,
      "loss": 0.6278,
      "step": 8947
    },
    {
      "epoch": 0.48286654794668393,
      "grad_norm": 1.014284563745636,
      "learning_rate": 5.742101526705492e-06,
      "loss": 0.3942,
      "step": 8948
    },
    {
      "epoch": 0.48292051157519833,
      "grad_norm": 1.0984175549142965,
      "learning_rate": 5.741339735292739e-06,
      "loss": 0.5857,
      "step": 8949
    },
    {
      "epoch": 0.48297447520371267,
      "grad_norm": 1.175613186975254,
      "learning_rate": 5.740577936943639e-06,
      "loss": 0.5437,
      "step": 8950
    },
    {
      "epoch": 0.48302843883222707,
      "grad_norm": 0.986151807392606,
      "learning_rate": 5.739816131680083e-06,
      "loss": 0.3765,
      "step": 8951
    },
    {
      "epoch": 0.48308240246074147,
      "grad_norm": 1.075751531113955,
      "learning_rate": 5.73905431952397e-06,
      "loss": 0.4863,
      "step": 8952
    },
    {
      "epoch": 0.48313636608925586,
      "grad_norm": 1.1144126332986921,
      "learning_rate": 5.738292500497195e-06,
      "loss": 0.6746,
      "step": 8953
    },
    {
      "epoch": 0.4831903297177702,
      "grad_norm": 0.9633828516217373,
      "learning_rate": 5.737530674621653e-06,
      "loss": 0.4162,
      "step": 8954
    },
    {
      "epoch": 0.4832442933462846,
      "grad_norm": 0.9439306333038188,
      "learning_rate": 5.736768841919238e-06,
      "loss": 0.5008,
      "step": 8955
    },
    {
      "epoch": 0.483298256974799,
      "grad_norm": 1.2358260784603383,
      "learning_rate": 5.736007002411847e-06,
      "loss": 0.6265,
      "step": 8956
    },
    {
      "epoch": 0.48335222060331334,
      "grad_norm": 1.0836553524257568,
      "learning_rate": 5.735245156121379e-06,
      "loss": 0.6006,
      "step": 8957
    },
    {
      "epoch": 0.48340618423182774,
      "grad_norm": 1.0012580915814944,
      "learning_rate": 5.734483303069724e-06,
      "loss": 0.4641,
      "step": 8958
    },
    {
      "epoch": 0.48346014786034214,
      "grad_norm": 0.9668970168525486,
      "learning_rate": 5.733721443278785e-06,
      "loss": 0.3939,
      "step": 8959
    },
    {
      "epoch": 0.48351411148885654,
      "grad_norm": 1.1900248120154215,
      "learning_rate": 5.7329595767704534e-06,
      "loss": 0.4988,
      "step": 8960
    },
    {
      "epoch": 0.4835680751173709,
      "grad_norm": 1.0152406186454868,
      "learning_rate": 5.73219770356663e-06,
      "loss": 0.499,
      "step": 8961
    },
    {
      "epoch": 0.4836220387458853,
      "grad_norm": 1.1021822170069868,
      "learning_rate": 5.731435823689209e-06,
      "loss": 0.6884,
      "step": 8962
    },
    {
      "epoch": 0.48367600237439967,
      "grad_norm": 1.0006494810227646,
      "learning_rate": 5.730673937160088e-06,
      "loss": 0.287,
      "step": 8963
    },
    {
      "epoch": 0.483729966002914,
      "grad_norm": 1.1068327646258418,
      "learning_rate": 5.729912044001167e-06,
      "loss": 0.5092,
      "step": 8964
    },
    {
      "epoch": 0.4837839296314284,
      "grad_norm": 1.058126201211116,
      "learning_rate": 5.72915014423434e-06,
      "loss": 0.4685,
      "step": 8965
    },
    {
      "epoch": 0.4838378932599428,
      "grad_norm": 1.0360707359613686,
      "learning_rate": 5.728388237881506e-06,
      "loss": 0.5583,
      "step": 8966
    },
    {
      "epoch": 0.4838918568884572,
      "grad_norm": 1.01881566521573,
      "learning_rate": 5.727626324964563e-06,
      "loss": 0.4436,
      "step": 8967
    },
    {
      "epoch": 0.48394582051697155,
      "grad_norm": 1.1077278620378985,
      "learning_rate": 5.72686440550541e-06,
      "loss": 0.6414,
      "step": 8968
    },
    {
      "epoch": 0.48399978414548595,
      "grad_norm": 0.9909253950888102,
      "learning_rate": 5.726102479525946e-06,
      "loss": 0.5359,
      "step": 8969
    },
    {
      "epoch": 0.48405374777400034,
      "grad_norm": 0.9259317706327108,
      "learning_rate": 5.725340547048067e-06,
      "loss": 0.4111,
      "step": 8970
    },
    {
      "epoch": 0.4841077114025147,
      "grad_norm": 1.1040840041914934,
      "learning_rate": 5.724578608093671e-06,
      "loss": 0.5298,
      "step": 8971
    },
    {
      "epoch": 0.4841616750310291,
      "grad_norm": 0.8878371820939249,
      "learning_rate": 5.7238166626846625e-06,
      "loss": 0.4146,
      "step": 8972
    },
    {
      "epoch": 0.4842156386595435,
      "grad_norm": 1.0624860652347219,
      "learning_rate": 5.7230547108429356e-06,
      "loss": 0.4169,
      "step": 8973
    },
    {
      "epoch": 0.4842696022880579,
      "grad_norm": 0.9531422758317465,
      "learning_rate": 5.722292752590391e-06,
      "loss": 0.4657,
      "step": 8974
    },
    {
      "epoch": 0.4843235659165722,
      "grad_norm": 1.2090817474443107,
      "learning_rate": 5.721530787948927e-06,
      "loss": 0.5508,
      "step": 8975
    },
    {
      "epoch": 0.4843775295450866,
      "grad_norm": 1.0376787396018732,
      "learning_rate": 5.720768816940446e-06,
      "loss": 0.5389,
      "step": 8976
    },
    {
      "epoch": 0.484431493173601,
      "grad_norm": 1.068804389898577,
      "learning_rate": 5.720006839586845e-06,
      "loss": 0.5748,
      "step": 8977
    },
    {
      "epoch": 0.48448545680211536,
      "grad_norm": 1.0994788034046652,
      "learning_rate": 5.719244855910027e-06,
      "loss": 0.4193,
      "step": 8978
    },
    {
      "epoch": 0.48453942043062975,
      "grad_norm": 0.9513739707833284,
      "learning_rate": 5.7184828659318866e-06,
      "loss": 0.4356,
      "step": 8979
    },
    {
      "epoch": 0.48459338405914415,
      "grad_norm": 0.8428064281541723,
      "learning_rate": 5.71772086967433e-06,
      "loss": 0.3693,
      "step": 8980
    },
    {
      "epoch": 0.4846473476876585,
      "grad_norm": 0.963015309586769,
      "learning_rate": 5.716958867159257e-06,
      "loss": 0.4981,
      "step": 8981
    },
    {
      "epoch": 0.4847013113161729,
      "grad_norm": 0.8541293958376956,
      "learning_rate": 5.7161968584085645e-06,
      "loss": 0.3019,
      "step": 8982
    },
    {
      "epoch": 0.4847552749446873,
      "grad_norm": 0.9988183500721246,
      "learning_rate": 5.715434843444158e-06,
      "loss": 0.4503,
      "step": 8983
    },
    {
      "epoch": 0.4848092385732017,
      "grad_norm": 0.9363316495309228,
      "learning_rate": 5.714672822287934e-06,
      "loss": 0.4624,
      "step": 8984
    },
    {
      "epoch": 0.484863202201716,
      "grad_norm": 0.9293958585906826,
      "learning_rate": 5.7139107949618004e-06,
      "loss": 0.298,
      "step": 8985
    },
    {
      "epoch": 0.4849171658302304,
      "grad_norm": 1.032124933885376,
      "learning_rate": 5.713148761487649e-06,
      "loss": 0.4632,
      "step": 8986
    },
    {
      "epoch": 0.4849711294587448,
      "grad_norm": 1.0153976170584875,
      "learning_rate": 5.71238672188739e-06,
      "loss": 0.4461,
      "step": 8987
    },
    {
      "epoch": 0.48502509308725916,
      "grad_norm": 1.0499519929240382,
      "learning_rate": 5.711624676182922e-06,
      "loss": 0.5287,
      "step": 8988
    },
    {
      "epoch": 0.48507905671577356,
      "grad_norm": 1.0115417408825789,
      "learning_rate": 5.710862624396146e-06,
      "loss": 0.4504,
      "step": 8989
    },
    {
      "epoch": 0.48513302034428796,
      "grad_norm": 0.971420612604982,
      "learning_rate": 5.710100566548966e-06,
      "loss": 0.5301,
      "step": 8990
    },
    {
      "epoch": 0.48518698397280235,
      "grad_norm": 1.0257195551501725,
      "learning_rate": 5.709338502663283e-06,
      "loss": 0.4326,
      "step": 8991
    },
    {
      "epoch": 0.4852409476013167,
      "grad_norm": 0.9799627868139753,
      "learning_rate": 5.7085764327610006e-06,
      "loss": 0.5442,
      "step": 8992
    },
    {
      "epoch": 0.4852949112298311,
      "grad_norm": 1.1205979618437802,
      "learning_rate": 5.707814356864019e-06,
      "loss": 0.5538,
      "step": 8993
    },
    {
      "epoch": 0.4853488748583455,
      "grad_norm": 1.3131174789829034,
      "learning_rate": 5.707052274994243e-06,
      "loss": 0.5984,
      "step": 8994
    },
    {
      "epoch": 0.48540283848685983,
      "grad_norm": 0.7982198698311004,
      "learning_rate": 5.706290187173577e-06,
      "loss": 0.348,
      "step": 8995
    },
    {
      "epoch": 0.48545680211537423,
      "grad_norm": 1.0185883179732644,
      "learning_rate": 5.705528093423922e-06,
      "loss": 0.5653,
      "step": 8996
    },
    {
      "epoch": 0.48551076574388863,
      "grad_norm": 0.8967157836487984,
      "learning_rate": 5.704765993767182e-06,
      "loss": 0.4219,
      "step": 8997
    },
    {
      "epoch": 0.485564729372403,
      "grad_norm": 1.0141133278094963,
      "learning_rate": 5.704003888225259e-06,
      "loss": 0.4431,
      "step": 8998
    },
    {
      "epoch": 0.48561869300091737,
      "grad_norm": 1.1906450569622669,
      "learning_rate": 5.703241776820059e-06,
      "loss": 0.6254,
      "step": 8999
    },
    {
      "epoch": 0.48567265662943176,
      "grad_norm": 1.0696067044142734,
      "learning_rate": 5.702479659573485e-06,
      "loss": 0.438,
      "step": 9000
    },
    {
      "epoch": 0.48567265662943176,
      "eval_loss": 0.5507768988609314,
      "eval_runtime": 163.5998,
      "eval_samples_per_second": 21.021,
      "eval_steps_per_second": 0.88,
      "step": 9000
    },
    {
      "epoch": 0.48572662025794616,
      "grad_norm": 0.8258489387750173,
      "learning_rate": 5.701717536507441e-06,
      "loss": 0.4522,
      "step": 9001
    },
    {
      "epoch": 0.4857805838864605,
      "grad_norm": 1.0708704774637918,
      "learning_rate": 5.700955407643833e-06,
      "loss": 0.5225,
      "step": 9002
    },
    {
      "epoch": 0.4858345475149749,
      "grad_norm": 1.191189551222171,
      "learning_rate": 5.700193273004562e-06,
      "loss": 0.5774,
      "step": 9003
    },
    {
      "epoch": 0.4858885111434893,
      "grad_norm": 1.0683967910429446,
      "learning_rate": 5.699431132611532e-06,
      "loss": 0.4572,
      "step": 9004
    },
    {
      "epoch": 0.4859424747720037,
      "grad_norm": 0.7811941286369862,
      "learning_rate": 5.698668986486652e-06,
      "loss": 0.3449,
      "step": 9005
    },
    {
      "epoch": 0.48599643840051804,
      "grad_norm": 0.8997353390492202,
      "learning_rate": 5.697906834651825e-06,
      "loss": 0.3701,
      "step": 9006
    },
    {
      "epoch": 0.48605040202903244,
      "grad_norm": 1.0160993455366434,
      "learning_rate": 5.697144677128955e-06,
      "loss": 0.5007,
      "step": 9007
    },
    {
      "epoch": 0.48610436565754683,
      "grad_norm": 0.9810074999233304,
      "learning_rate": 5.696382513939947e-06,
      "loss": 0.4599,
      "step": 9008
    },
    {
      "epoch": 0.4861583292860612,
      "grad_norm": 1.175000637691287,
      "learning_rate": 5.695620345106709e-06,
      "loss": 0.5142,
      "step": 9009
    },
    {
      "epoch": 0.4862122929145756,
      "grad_norm": 0.9986410358138473,
      "learning_rate": 5.694858170651144e-06,
      "loss": 0.5835,
      "step": 9010
    },
    {
      "epoch": 0.48626625654308997,
      "grad_norm": 1.1869735562438195,
      "learning_rate": 5.694095990595158e-06,
      "loss": 0.8689,
      "step": 9011
    },
    {
      "epoch": 0.4863202201716043,
      "grad_norm": 1.1913636863835737,
      "learning_rate": 5.693333804960657e-06,
      "loss": 0.5273,
      "step": 9012
    },
    {
      "epoch": 0.4863741838001187,
      "grad_norm": 0.9477564463264486,
      "learning_rate": 5.692571613769546e-06,
      "loss": 0.4125,
      "step": 9013
    },
    {
      "epoch": 0.4864281474286331,
      "grad_norm": 1.0995654302186082,
      "learning_rate": 5.691809417043733e-06,
      "loss": 0.7489,
      "step": 9014
    },
    {
      "epoch": 0.4864821110571475,
      "grad_norm": 1.0543152392843842,
      "learning_rate": 5.691047214805123e-06,
      "loss": 0.6065,
      "step": 9015
    },
    {
      "epoch": 0.48653607468566185,
      "grad_norm": 1.0807222990521688,
      "learning_rate": 5.6902850070756255e-06,
      "loss": 0.4433,
      "step": 9016
    },
    {
      "epoch": 0.48659003831417624,
      "grad_norm": 1.0448691638347891,
      "learning_rate": 5.689522793877141e-06,
      "loss": 0.6474,
      "step": 9017
    },
    {
      "epoch": 0.48664400194269064,
      "grad_norm": 1.004580854451043,
      "learning_rate": 5.6887605752315825e-06,
      "loss": 0.3622,
      "step": 9018
    },
    {
      "epoch": 0.486697965571205,
      "grad_norm": 1.0538013577635714,
      "learning_rate": 5.687998351160855e-06,
      "loss": 0.5465,
      "step": 9019
    },
    {
      "epoch": 0.4867519291997194,
      "grad_norm": 1.0095164775349887,
      "learning_rate": 5.687236121686862e-06,
      "loss": 0.4933,
      "step": 9020
    },
    {
      "epoch": 0.4868058928282338,
      "grad_norm": 1.0856562773300655,
      "learning_rate": 5.686473886831516e-06,
      "loss": 0.4879,
      "step": 9021
    },
    {
      "epoch": 0.4868598564567482,
      "grad_norm": 0.9298221523035938,
      "learning_rate": 5.685711646616722e-06,
      "loss": 0.4063,
      "step": 9022
    },
    {
      "epoch": 0.4869138200852625,
      "grad_norm": 1.2314667157371089,
      "learning_rate": 5.684949401064385e-06,
      "loss": 0.4448,
      "step": 9023
    },
    {
      "epoch": 0.4869677837137769,
      "grad_norm": 1.2143350841957286,
      "learning_rate": 5.6841871501964175e-06,
      "loss": 0.3794,
      "step": 9024
    },
    {
      "epoch": 0.4870217473422913,
      "grad_norm": 0.8478134258752489,
      "learning_rate": 5.6834248940347245e-06,
      "loss": 0.3676,
      "step": 9025
    },
    {
      "epoch": 0.48707571097080565,
      "grad_norm": 1.0983243100547224,
      "learning_rate": 5.682662632601216e-06,
      "loss": 0.5803,
      "step": 9026
    },
    {
      "epoch": 0.48712967459932005,
      "grad_norm": 1.0237021261518908,
      "learning_rate": 5.6819003659177984e-06,
      "loss": 0.4509,
      "step": 9027
    },
    {
      "epoch": 0.48718363822783445,
      "grad_norm": 0.9966202257210777,
      "learning_rate": 5.68113809400638e-06,
      "loss": 0.368,
      "step": 9028
    },
    {
      "epoch": 0.48723760185634885,
      "grad_norm": 0.9748564726544412,
      "learning_rate": 5.680375816888869e-06,
      "loss": 0.3873,
      "step": 9029
    },
    {
      "epoch": 0.4872915654848632,
      "grad_norm": 0.8974139130744889,
      "learning_rate": 5.679613534587176e-06,
      "loss": 0.4133,
      "step": 9030
    },
    {
      "epoch": 0.4873455291133776,
      "grad_norm": 1.011780184903026,
      "learning_rate": 5.678851247123208e-06,
      "loss": 0.4575,
      "step": 9031
    },
    {
      "epoch": 0.487399492741892,
      "grad_norm": 1.1732641803251664,
      "learning_rate": 5.678088954518874e-06,
      "loss": 0.4771,
      "step": 9032
    },
    {
      "epoch": 0.4874534563704063,
      "grad_norm": 1.0343273495426395,
      "learning_rate": 5.6773266567960864e-06,
      "loss": 0.436,
      "step": 9033
    },
    {
      "epoch": 0.4875074199989207,
      "grad_norm": 1.161626929202276,
      "learning_rate": 5.676564353976748e-06,
      "loss": 0.5725,
      "step": 9034
    },
    {
      "epoch": 0.4875613836274351,
      "grad_norm": 1.0270623916816533,
      "learning_rate": 5.675802046082772e-06,
      "loss": 0.3951,
      "step": 9035
    },
    {
      "epoch": 0.4876153472559495,
      "grad_norm": 0.9967003999386687,
      "learning_rate": 5.67503973313607e-06,
      "loss": 0.4063,
      "step": 9036
    },
    {
      "epoch": 0.48766931088446386,
      "grad_norm": 1.015640738263481,
      "learning_rate": 5.674277415158548e-06,
      "loss": 0.4719,
      "step": 9037
    },
    {
      "epoch": 0.48772327451297826,
      "grad_norm": 1.0997044271434895,
      "learning_rate": 5.673515092172115e-06,
      "loss": 0.5361,
      "step": 9038
    },
    {
      "epoch": 0.48777723814149265,
      "grad_norm": 1.0144318479358967,
      "learning_rate": 5.6727527641986854e-06,
      "loss": 0.529,
      "step": 9039
    },
    {
      "epoch": 0.487831201770007,
      "grad_norm": 1.134180627063445,
      "learning_rate": 5.671990431260166e-06,
      "loss": 0.5206,
      "step": 9040
    },
    {
      "epoch": 0.4878851653985214,
      "grad_norm": 1.0445275579652313,
      "learning_rate": 5.671228093378468e-06,
      "loss": 0.5604,
      "step": 9041
    },
    {
      "epoch": 0.4879391290270358,
      "grad_norm": 1.0195743761493503,
      "learning_rate": 5.670465750575501e-06,
      "loss": 0.6556,
      "step": 9042
    },
    {
      "epoch": 0.4879930926555502,
      "grad_norm": 0.8494023680929432,
      "learning_rate": 5.669703402873177e-06,
      "loss": 0.4281,
      "step": 9043
    },
    {
      "epoch": 0.48804705628406453,
      "grad_norm": 1.0654554688769102,
      "learning_rate": 5.668941050293405e-06,
      "loss": 0.4872,
      "step": 9044
    },
    {
      "epoch": 0.4881010199125789,
      "grad_norm": 0.9595716703938717,
      "learning_rate": 5.6681786928580965e-06,
      "loss": 0.4723,
      "step": 9045
    },
    {
      "epoch": 0.4881549835410933,
      "grad_norm": 1.0024533210719815,
      "learning_rate": 5.667416330589161e-06,
      "loss": 0.4153,
      "step": 9046
    },
    {
      "epoch": 0.48820894716960767,
      "grad_norm": 0.9008841006155078,
      "learning_rate": 5.666653963508514e-06,
      "loss": 0.3582,
      "step": 9047
    },
    {
      "epoch": 0.48826291079812206,
      "grad_norm": 1.2054046148338748,
      "learning_rate": 5.665891591638063e-06,
      "loss": 0.6435,
      "step": 9048
    },
    {
      "epoch": 0.48831687442663646,
      "grad_norm": 0.9455730915596545,
      "learning_rate": 5.665129214999719e-06,
      "loss": 0.4715,
      "step": 9049
    },
    {
      "epoch": 0.4883708380551508,
      "grad_norm": 0.9921144873349975,
      "learning_rate": 5.664366833615395e-06,
      "loss": 0.4686,
      "step": 9050
    },
    {
      "epoch": 0.4884248016836652,
      "grad_norm": 1.110386277828727,
      "learning_rate": 5.663604447507001e-06,
      "loss": 0.6002,
      "step": 9051
    },
    {
      "epoch": 0.4884787653121796,
      "grad_norm": 1.060541804088775,
      "learning_rate": 5.66284205669645e-06,
      "loss": 0.5314,
      "step": 9052
    },
    {
      "epoch": 0.488532728940694,
      "grad_norm": 0.9038001244693592,
      "learning_rate": 5.662079661205655e-06,
      "loss": 0.413,
      "step": 9053
    },
    {
      "epoch": 0.48858669256920834,
      "grad_norm": 1.0155073119807618,
      "learning_rate": 5.661317261056527e-06,
      "loss": 0.5412,
      "step": 9054
    },
    {
      "epoch": 0.48864065619772273,
      "grad_norm": 1.080357609652199,
      "learning_rate": 5.660554856270978e-06,
      "loss": 0.401,
      "step": 9055
    },
    {
      "epoch": 0.48869461982623713,
      "grad_norm": 0.9733955886591145,
      "learning_rate": 5.659792446870919e-06,
      "loss": 0.4119,
      "step": 9056
    },
    {
      "epoch": 0.4887485834547515,
      "grad_norm": 0.9331906517111229,
      "learning_rate": 5.659030032878264e-06,
      "loss": 0.4778,
      "step": 9057
    },
    {
      "epoch": 0.48880254708326587,
      "grad_norm": 0.9812455375904994,
      "learning_rate": 5.658267614314926e-06,
      "loss": 0.4181,
      "step": 9058
    },
    {
      "epoch": 0.48885651071178027,
      "grad_norm": 1.0148244433348792,
      "learning_rate": 5.657505191202815e-06,
      "loss": 0.4687,
      "step": 9059
    },
    {
      "epoch": 0.48891047434029467,
      "grad_norm": 1.0032508778608455,
      "learning_rate": 5.656742763563848e-06,
      "loss": 0.4211,
      "step": 9060
    },
    {
      "epoch": 0.488964437968809,
      "grad_norm": 0.9839858023374348,
      "learning_rate": 5.655980331419933e-06,
      "loss": 0.4306,
      "step": 9061
    },
    {
      "epoch": 0.4890184015973234,
      "grad_norm": 0.9618285731339128,
      "learning_rate": 5.655217894792988e-06,
      "loss": 0.4284,
      "step": 9062
    },
    {
      "epoch": 0.4890723652258378,
      "grad_norm": 1.3204996827588074,
      "learning_rate": 5.654455453704924e-06,
      "loss": 0.5638,
      "step": 9063
    },
    {
      "epoch": 0.48912632885435214,
      "grad_norm": 0.9190775710802204,
      "learning_rate": 5.653693008177652e-06,
      "loss": 0.4546,
      "step": 9064
    },
    {
      "epoch": 0.48918029248286654,
      "grad_norm": 0.9498044763175791,
      "learning_rate": 5.65293055823309e-06,
      "loss": 0.4811,
      "step": 9065
    },
    {
      "epoch": 0.48923425611138094,
      "grad_norm": 0.9502161538461522,
      "learning_rate": 5.652168103893146e-06,
      "loss": 0.4583,
      "step": 9066
    },
    {
      "epoch": 0.48928821973989534,
      "grad_norm": 1.1162640153105996,
      "learning_rate": 5.651405645179739e-06,
      "loss": 0.591,
      "step": 9067
    },
    {
      "epoch": 0.4893421833684097,
      "grad_norm": 1.0754056215436516,
      "learning_rate": 5.650643182114782e-06,
      "loss": 0.4767,
      "step": 9068
    },
    {
      "epoch": 0.4893961469969241,
      "grad_norm": 1.1564695733771335,
      "learning_rate": 5.649880714720186e-06,
      "loss": 0.4673,
      "step": 9069
    },
    {
      "epoch": 0.4894501106254385,
      "grad_norm": 1.063908188974962,
      "learning_rate": 5.649118243017867e-06,
      "loss": 0.4176,
      "step": 9070
    },
    {
      "epoch": 0.4895040742539528,
      "grad_norm": 0.9433518356948424,
      "learning_rate": 5.6483557670297395e-06,
      "loss": 0.4994,
      "step": 9071
    },
    {
      "epoch": 0.4895580378824672,
      "grad_norm": 1.10802606014558,
      "learning_rate": 5.647593286777717e-06,
      "loss": 0.4953,
      "step": 9072
    },
    {
      "epoch": 0.4896120015109816,
      "grad_norm": 0.8317653896410824,
      "learning_rate": 5.646830802283715e-06,
      "loss": 0.3691,
      "step": 9073
    },
    {
      "epoch": 0.489665965139496,
      "grad_norm": 1.0105601461502003,
      "learning_rate": 5.6460683135696455e-06,
      "loss": 0.4486,
      "step": 9074
    },
    {
      "epoch": 0.48971992876801035,
      "grad_norm": 1.1308052178158472,
      "learning_rate": 5.645305820657427e-06,
      "loss": 0.5394,
      "step": 9075
    },
    {
      "epoch": 0.48977389239652475,
      "grad_norm": 1.2222763163571557,
      "learning_rate": 5.644543323568972e-06,
      "loss": 0.666,
      "step": 9076
    },
    {
      "epoch": 0.48982785602503914,
      "grad_norm": 0.9258783262699528,
      "learning_rate": 5.643780822326195e-06,
      "loss": 0.407,
      "step": 9077
    },
    {
      "epoch": 0.4898818196535535,
      "grad_norm": 1.079257025317513,
      "learning_rate": 5.643018316951013e-06,
      "loss": 0.5201,
      "step": 9078
    },
    {
      "epoch": 0.4899357832820679,
      "grad_norm": 0.9093672035536486,
      "learning_rate": 5.642255807465339e-06,
      "loss": 0.4614,
      "step": 9079
    },
    {
      "epoch": 0.4899897469105823,
      "grad_norm": 1.0528566429842607,
      "learning_rate": 5.641493293891091e-06,
      "loss": 0.5733,
      "step": 9080
    },
    {
      "epoch": 0.4900437105390966,
      "grad_norm": 1.148805807435015,
      "learning_rate": 5.640730776250182e-06,
      "loss": 0.5154,
      "step": 9081
    },
    {
      "epoch": 0.490097674167611,
      "grad_norm": 1.0766487726531782,
      "learning_rate": 5.639968254564528e-06,
      "loss": 0.4668,
      "step": 9082
    },
    {
      "epoch": 0.4901516377961254,
      "grad_norm": 1.1070000175776042,
      "learning_rate": 5.639205728856044e-06,
      "loss": 0.5848,
      "step": 9083
    },
    {
      "epoch": 0.4902056014246398,
      "grad_norm": 0.9774337320541031,
      "learning_rate": 5.6384431991466474e-06,
      "loss": 0.3472,
      "step": 9084
    },
    {
      "epoch": 0.49025956505315416,
      "grad_norm": 1.1373754921727341,
      "learning_rate": 5.637680665458254e-06,
      "loss": 0.5523,
      "step": 9085
    },
    {
      "epoch": 0.49031352868166855,
      "grad_norm": 1.0730813291771926,
      "learning_rate": 5.6369181278127785e-06,
      "loss": 0.5177,
      "step": 9086
    },
    {
      "epoch": 0.49036749231018295,
      "grad_norm": 1.072558548709178,
      "learning_rate": 5.636155586232138e-06,
      "loss": 0.3806,
      "step": 9087
    },
    {
      "epoch": 0.4904214559386973,
      "grad_norm": 1.073174073297064,
      "learning_rate": 5.635393040738248e-06,
      "loss": 0.4183,
      "step": 9088
    },
    {
      "epoch": 0.4904754195672117,
      "grad_norm": 1.0484761729931467,
      "learning_rate": 5.6346304913530255e-06,
      "loss": 0.4522,
      "step": 9089
    },
    {
      "epoch": 0.4905293831957261,
      "grad_norm": 0.9510815213356895,
      "learning_rate": 5.633867938098387e-06,
      "loss": 0.4634,
      "step": 9090
    },
    {
      "epoch": 0.4905833468242405,
      "grad_norm": 1.3693507439791255,
      "learning_rate": 5.633105380996248e-06,
      "loss": 0.6434,
      "step": 9091
    },
    {
      "epoch": 0.4906373104527548,
      "grad_norm": 0.983070201544626,
      "learning_rate": 5.632342820068527e-06,
      "loss": 0.4089,
      "step": 9092
    },
    {
      "epoch": 0.4906912740812692,
      "grad_norm": 0.919827321286587,
      "learning_rate": 5.631580255337139e-06,
      "loss": 0.339,
      "step": 9093
    },
    {
      "epoch": 0.4907452377097836,
      "grad_norm": 0.9665890643810782,
      "learning_rate": 5.630817686824001e-06,
      "loss": 0.4679,
      "step": 9094
    },
    {
      "epoch": 0.49079920133829796,
      "grad_norm": 0.96136881018418,
      "learning_rate": 5.630055114551031e-06,
      "loss": 0.3986,
      "step": 9095
    },
    {
      "epoch": 0.49085316496681236,
      "grad_norm": 1.004987836870924,
      "learning_rate": 5.629292538540146e-06,
      "loss": 0.4103,
      "step": 9096
    },
    {
      "epoch": 0.49090712859532676,
      "grad_norm": 0.9927249506479977,
      "learning_rate": 5.628529958813262e-06,
      "loss": 0.48,
      "step": 9097
    },
    {
      "epoch": 0.49096109222384116,
      "grad_norm": 1.1901896279463655,
      "learning_rate": 5.627767375392295e-06,
      "loss": 0.4074,
      "step": 9098
    },
    {
      "epoch": 0.4910150558523555,
      "grad_norm": 1.0203343233747315,
      "learning_rate": 5.6270047882991675e-06,
      "loss": 0.4849,
      "step": 9099
    },
    {
      "epoch": 0.4910690194808699,
      "grad_norm": 0.9277309253238674,
      "learning_rate": 5.6262421975557935e-06,
      "loss": 0.3905,
      "step": 9100
    },
    {
      "epoch": 0.4911229831093843,
      "grad_norm": 1.1629862645431597,
      "learning_rate": 5.625479603184092e-06,
      "loss": 0.5546,
      "step": 9101
    },
    {
      "epoch": 0.49117694673789863,
      "grad_norm": 1.0872630804343892,
      "learning_rate": 5.624717005205978e-06,
      "loss": 0.4649,
      "step": 9102
    },
    {
      "epoch": 0.49123091036641303,
      "grad_norm": 1.0316687539540388,
      "learning_rate": 5.623954403643372e-06,
      "loss": 0.5985,
      "step": 9103
    },
    {
      "epoch": 0.49128487399492743,
      "grad_norm": 0.9655504277761057,
      "learning_rate": 5.623191798518192e-06,
      "loss": 0.4303,
      "step": 9104
    },
    {
      "epoch": 0.4913388376234418,
      "grad_norm": 1.0202583867849238,
      "learning_rate": 5.6224291898523545e-06,
      "loss": 0.385,
      "step": 9105
    },
    {
      "epoch": 0.49139280125195617,
      "grad_norm": 1.0597243845431228,
      "learning_rate": 5.621666577667779e-06,
      "loss": 0.4107,
      "step": 9106
    },
    {
      "epoch": 0.49144676488047057,
      "grad_norm": 0.9343221986159986,
      "learning_rate": 5.620903961986383e-06,
      "loss": 0.3732,
      "step": 9107
    },
    {
      "epoch": 0.49150072850898496,
      "grad_norm": 0.9614633658009251,
      "learning_rate": 5.620141342830083e-06,
      "loss": 0.4382,
      "step": 9108
    },
    {
      "epoch": 0.4915546921374993,
      "grad_norm": 1.001043862398232,
      "learning_rate": 5.619378720220803e-06,
      "loss": 0.5076,
      "step": 9109
    },
    {
      "epoch": 0.4916086557660137,
      "grad_norm": 0.9541530184519891,
      "learning_rate": 5.618616094180455e-06,
      "loss": 0.4421,
      "step": 9110
    },
    {
      "epoch": 0.4916626193945281,
      "grad_norm": 1.182443582300601,
      "learning_rate": 5.617853464730961e-06,
      "loss": 0.4961,
      "step": 9111
    },
    {
      "epoch": 0.49171658302304244,
      "grad_norm": 1.0014452499363702,
      "learning_rate": 5.617090831894242e-06,
      "loss": 0.3892,
      "step": 9112
    },
    {
      "epoch": 0.49177054665155684,
      "grad_norm": 0.984245254941679,
      "learning_rate": 5.616328195692211e-06,
      "loss": 0.5264,
      "step": 9113
    },
    {
      "epoch": 0.49182451028007124,
      "grad_norm": 0.9097772916548543,
      "learning_rate": 5.615565556146792e-06,
      "loss": 0.445,
      "step": 9114
    },
    {
      "epoch": 0.49187847390858563,
      "grad_norm": 0.9628613731593044,
      "learning_rate": 5.614802913279901e-06,
      "loss": 0.5766,
      "step": 9115
    },
    {
      "epoch": 0.4919324375371,
      "grad_norm": 0.8477536070441106,
      "learning_rate": 5.614040267113459e-06,
      "loss": 0.3414,
      "step": 9116
    },
    {
      "epoch": 0.4919864011656144,
      "grad_norm": 0.9971225396369336,
      "learning_rate": 5.6132776176693844e-06,
      "loss": 0.4782,
      "step": 9117
    },
    {
      "epoch": 0.49204036479412877,
      "grad_norm": 0.8657982783810172,
      "learning_rate": 5.612514964969596e-06,
      "loss": 0.3855,
      "step": 9118
    },
    {
      "epoch": 0.4920943284226431,
      "grad_norm": 1.05908034585393,
      "learning_rate": 5.611752309036015e-06,
      "loss": 0.5481,
      "step": 9119
    },
    {
      "epoch": 0.4921482920511575,
      "grad_norm": 1.054190345423335,
      "learning_rate": 5.610989649890559e-06,
      "loss": 0.4964,
      "step": 9120
    },
    {
      "epoch": 0.4922022556796719,
      "grad_norm": 1.147373574075141,
      "learning_rate": 5.610226987555148e-06,
      "loss": 0.4547,
      "step": 9121
    },
    {
      "epoch": 0.4922562193081863,
      "grad_norm": 1.2670287892522556,
      "learning_rate": 5.609464322051702e-06,
      "loss": 0.5442,
      "step": 9122
    },
    {
      "epoch": 0.49231018293670065,
      "grad_norm": 0.9524589264005422,
      "learning_rate": 5.608701653402141e-06,
      "loss": 0.4617,
      "step": 9123
    },
    {
      "epoch": 0.49236414656521504,
      "grad_norm": 1.0393432499066622,
      "learning_rate": 5.607938981628385e-06,
      "loss": 0.5551,
      "step": 9124
    },
    {
      "epoch": 0.49241811019372944,
      "grad_norm": 0.9679025535074631,
      "learning_rate": 5.607176306752352e-06,
      "loss": 0.5185,
      "step": 9125
    },
    {
      "epoch": 0.4924720738222438,
      "grad_norm": 0.9778398508519132,
      "learning_rate": 5.606413628795965e-06,
      "loss": 0.5109,
      "step": 9126
    },
    {
      "epoch": 0.4925260374507582,
      "grad_norm": 0.97562800218688,
      "learning_rate": 5.605650947781143e-06,
      "loss": 0.576,
      "step": 9127
    },
    {
      "epoch": 0.4925800010792726,
      "grad_norm": 1.0387596127610503,
      "learning_rate": 5.604888263729805e-06,
      "loss": 0.4762,
      "step": 9128
    },
    {
      "epoch": 0.492633964707787,
      "grad_norm": 1.2629110189231894,
      "learning_rate": 5.604125576663872e-06,
      "loss": 0.6897,
      "step": 9129
    },
    {
      "epoch": 0.4926879283363013,
      "grad_norm": 0.9378347697326365,
      "learning_rate": 5.603362886605267e-06,
      "loss": 0.3219,
      "step": 9130
    },
    {
      "epoch": 0.4927418919648157,
      "grad_norm": 1.0130724593233944,
      "learning_rate": 5.6026001935759076e-06,
      "loss": 0.5077,
      "step": 9131
    },
    {
      "epoch": 0.4927958555933301,
      "grad_norm": 0.7874646816805916,
      "learning_rate": 5.601837497597714e-06,
      "loss": 0.3519,
      "step": 9132
    },
    {
      "epoch": 0.49284981922184445,
      "grad_norm": 1.1014547059404938,
      "learning_rate": 5.6010747986926086e-06,
      "loss": 0.7056,
      "step": 9133
    },
    {
      "epoch": 0.49290378285035885,
      "grad_norm": 1.1006761529465736,
      "learning_rate": 5.600312096882511e-06,
      "loss": 0.641,
      "step": 9134
    },
    {
      "epoch": 0.49295774647887325,
      "grad_norm": 0.9865244187156262,
      "learning_rate": 5.5995493921893415e-06,
      "loss": 0.5426,
      "step": 9135
    },
    {
      "epoch": 0.49301171010738765,
      "grad_norm": 1.1003810468793087,
      "learning_rate": 5.598786684635022e-06,
      "loss": 0.534,
      "step": 9136
    },
    {
      "epoch": 0.493065673735902,
      "grad_norm": 1.1356140947380464,
      "learning_rate": 5.598023974241475e-06,
      "loss": 0.5856,
      "step": 9137
    },
    {
      "epoch": 0.4931196373644164,
      "grad_norm": 0.9528600031966235,
      "learning_rate": 5.597261261030619e-06,
      "loss": 0.4131,
      "step": 9138
    },
    {
      "epoch": 0.4931736009929308,
      "grad_norm": 0.9006135093163227,
      "learning_rate": 5.596498545024376e-06,
      "loss": 0.4254,
      "step": 9139
    },
    {
      "epoch": 0.4932275646214451,
      "grad_norm": 0.8812376560515882,
      "learning_rate": 5.595735826244668e-06,
      "loss": 0.4722,
      "step": 9140
    },
    {
      "epoch": 0.4932815282499595,
      "grad_norm": 0.9026527212597094,
      "learning_rate": 5.594973104713415e-06,
      "loss": 0.3664,
      "step": 9141
    },
    {
      "epoch": 0.4933354918784739,
      "grad_norm": 0.9195701211827959,
      "learning_rate": 5.594210380452538e-06,
      "loss": 0.4012,
      "step": 9142
    },
    {
      "epoch": 0.4933894555069883,
      "grad_norm": 1.0571645343750486,
      "learning_rate": 5.593447653483961e-06,
      "loss": 0.4507,
      "step": 9143
    },
    {
      "epoch": 0.49344341913550266,
      "grad_norm": 0.8309722422917568,
      "learning_rate": 5.592684923829603e-06,
      "loss": 0.3504,
      "step": 9144
    },
    {
      "epoch": 0.49349738276401706,
      "grad_norm": 0.9683668234971872,
      "learning_rate": 5.591922191511387e-06,
      "loss": 0.4697,
      "step": 9145
    },
    {
      "epoch": 0.49355134639253145,
      "grad_norm": 1.1082981706169845,
      "learning_rate": 5.591159456551234e-06,
      "loss": 0.5393,
      "step": 9146
    },
    {
      "epoch": 0.4936053100210458,
      "grad_norm": 0.9223557175453133,
      "learning_rate": 5.590396718971067e-06,
      "loss": 0.4114,
      "step": 9147
    },
    {
      "epoch": 0.4936592736495602,
      "grad_norm": 0.9308925023962136,
      "learning_rate": 5.589633978792808e-06,
      "loss": 0.4053,
      "step": 9148
    },
    {
      "epoch": 0.4937132372780746,
      "grad_norm": 0.754557871028483,
      "learning_rate": 5.588871236038376e-06,
      "loss": 0.3232,
      "step": 9149
    },
    {
      "epoch": 0.49376720090658893,
      "grad_norm": 1.1737933400497087,
      "learning_rate": 5.588108490729694e-06,
      "loss": 0.5379,
      "step": 9150
    },
    {
      "epoch": 0.49382116453510333,
      "grad_norm": 1.0854744385004083,
      "learning_rate": 5.587345742888687e-06,
      "loss": 0.5697,
      "step": 9151
    },
    {
      "epoch": 0.4938751281636177,
      "grad_norm": 1.0048893362232754,
      "learning_rate": 5.586582992537273e-06,
      "loss": 0.489,
      "step": 9152
    },
    {
      "epoch": 0.4939290917921321,
      "grad_norm": 0.9457083795549985,
      "learning_rate": 5.5858202396973785e-06,
      "loss": 0.4913,
      "step": 9153
    },
    {
      "epoch": 0.49398305542064647,
      "grad_norm": 1.584170562344879,
      "learning_rate": 5.585057484390923e-06,
      "loss": 0.7078,
      "step": 9154
    },
    {
      "epoch": 0.49403701904916086,
      "grad_norm": 1.1707641613345312,
      "learning_rate": 5.584294726639828e-06,
      "loss": 0.5097,
      "step": 9155
    },
    {
      "epoch": 0.49409098267767526,
      "grad_norm": 1.3367624003119714,
      "learning_rate": 5.583531966466019e-06,
      "loss": 0.7103,
      "step": 9156
    },
    {
      "epoch": 0.4941449463061896,
      "grad_norm": 1.1309348125955154,
      "learning_rate": 5.582769203891417e-06,
      "loss": 0.438,
      "step": 9157
    },
    {
      "epoch": 0.494198909934704,
      "grad_norm": 0.9272721258130063,
      "learning_rate": 5.582006438937943e-06,
      "loss": 0.5267,
      "step": 9158
    },
    {
      "epoch": 0.4942528735632184,
      "grad_norm": 1.1768842573059317,
      "learning_rate": 5.581243671627522e-06,
      "loss": 0.5009,
      "step": 9159
    },
    {
      "epoch": 0.4943068371917328,
      "grad_norm": 1.162302490526769,
      "learning_rate": 5.580480901982076e-06,
      "loss": 0.5607,
      "step": 9160
    },
    {
      "epoch": 0.49436080082024714,
      "grad_norm": 0.8319147912767002,
      "learning_rate": 5.5797181300235285e-06,
      "loss": 0.3761,
      "step": 9161
    },
    {
      "epoch": 0.49441476444876153,
      "grad_norm": 0.7800196772310112,
      "learning_rate": 5.5789553557737996e-06,
      "loss": 0.3261,
      "step": 9162
    },
    {
      "epoch": 0.49446872807727593,
      "grad_norm": 1.1257978153167363,
      "learning_rate": 5.578192579254814e-06,
      "loss": 0.5072,
      "step": 9163
    },
    {
      "epoch": 0.4945226917057903,
      "grad_norm": 0.917675092514413,
      "learning_rate": 5.5774298004884965e-06,
      "loss": 0.4871,
      "step": 9164
    },
    {
      "epoch": 0.49457665533430467,
      "grad_norm": 0.9970625752675731,
      "learning_rate": 5.576667019496767e-06,
      "loss": 0.5463,
      "step": 9165
    },
    {
      "epoch": 0.49463061896281907,
      "grad_norm": 1.316769497799427,
      "learning_rate": 5.5759042363015505e-06,
      "loss": 0.5336,
      "step": 9166
    },
    {
      "epoch": 0.49468458259133347,
      "grad_norm": 1.0351388890162914,
      "learning_rate": 5.575141450924768e-06,
      "loss": 0.6001,
      "step": 9167
    },
    {
      "epoch": 0.4947385462198478,
      "grad_norm": 0.916664722167747,
      "learning_rate": 5.574378663388346e-06,
      "loss": 0.3953,
      "step": 9168
    },
    {
      "epoch": 0.4947925098483622,
      "grad_norm": 0.996479421177259,
      "learning_rate": 5.573615873714204e-06,
      "loss": 0.4465,
      "step": 9169
    },
    {
      "epoch": 0.4948464734768766,
      "grad_norm": 0.9919557817409324,
      "learning_rate": 5.572853081924269e-06,
      "loss": 0.4387,
      "step": 9170
    },
    {
      "epoch": 0.49490043710539094,
      "grad_norm": 1.1587231129343423,
      "learning_rate": 5.5720902880404635e-06,
      "loss": 0.3653,
      "step": 9171
    },
    {
      "epoch": 0.49495440073390534,
      "grad_norm": 0.8569568172603038,
      "learning_rate": 5.57132749208471e-06,
      "loss": 0.4106,
      "step": 9172
    },
    {
      "epoch": 0.49500836436241974,
      "grad_norm": 1.0413459163096053,
      "learning_rate": 5.5705646940789305e-06,
      "loss": 0.4568,
      "step": 9173
    },
    {
      "epoch": 0.49506232799093414,
      "grad_norm": 0.9020983332997344,
      "learning_rate": 5.569801894045052e-06,
      "loss": 0.41,
      "step": 9174
    },
    {
      "epoch": 0.4951162916194485,
      "grad_norm": 0.9899595180164241,
      "learning_rate": 5.569039092004995e-06,
      "loss": 0.5751,
      "step": 9175
    },
    {
      "epoch": 0.4951702552479629,
      "grad_norm": 0.9368445909766963,
      "learning_rate": 5.568276287980686e-06,
      "loss": 0.4496,
      "step": 9176
    },
    {
      "epoch": 0.4952242188764773,
      "grad_norm": 1.0512549977278118,
      "learning_rate": 5.567513481994047e-06,
      "loss": 0.4305,
      "step": 9177
    },
    {
      "epoch": 0.4952781825049916,
      "grad_norm": 1.1869171094548334,
      "learning_rate": 5.566750674067003e-06,
      "loss": 0.4879,
      "step": 9178
    },
    {
      "epoch": 0.495332146133506,
      "grad_norm": 0.8402338715223964,
      "learning_rate": 5.565987864221476e-06,
      "loss": 0.3184,
      "step": 9179
    },
    {
      "epoch": 0.4953861097620204,
      "grad_norm": 0.956738886567115,
      "learning_rate": 5.565225052479391e-06,
      "loss": 0.4244,
      "step": 9180
    },
    {
      "epoch": 0.49544007339053475,
      "grad_norm": 0.7991380589482813,
      "learning_rate": 5.5644622388626745e-06,
      "loss": 0.387,
      "step": 9181
    },
    {
      "epoch": 0.49549403701904915,
      "grad_norm": 0.965212856278813,
      "learning_rate": 5.563699423393248e-06,
      "loss": 0.4728,
      "step": 9182
    },
    {
      "epoch": 0.49554800064756355,
      "grad_norm": 0.8815015073861363,
      "learning_rate": 5.562936606093033e-06,
      "loss": 0.3656,
      "step": 9183
    },
    {
      "epoch": 0.49560196427607794,
      "grad_norm": 0.9814643405929492,
      "learning_rate": 5.562173786983958e-06,
      "loss": 0.3579,
      "step": 9184
    },
    {
      "epoch": 0.4956559279045923,
      "grad_norm": 0.7747149258364787,
      "learning_rate": 5.561410966087945e-06,
      "loss": 0.3506,
      "step": 9185
    },
    {
      "epoch": 0.4957098915331067,
      "grad_norm": 1.241510409918767,
      "learning_rate": 5.560648143426919e-06,
      "loss": 0.6019,
      "step": 9186
    },
    {
      "epoch": 0.4957638551616211,
      "grad_norm": 0.9719412275299124,
      "learning_rate": 5.559885319022803e-06,
      "loss": 0.4331,
      "step": 9187
    },
    {
      "epoch": 0.4958178187901354,
      "grad_norm": 0.9079525774259637,
      "learning_rate": 5.559122492897525e-06,
      "loss": 0.419,
      "step": 9188
    },
    {
      "epoch": 0.4958717824186498,
      "grad_norm": 0.9050783759407539,
      "learning_rate": 5.5583596650730055e-06,
      "loss": 0.4644,
      "step": 9189
    },
    {
      "epoch": 0.4959257460471642,
      "grad_norm": 0.7765840191259948,
      "learning_rate": 5.55759683557117e-06,
      "loss": 0.361,
      "step": 9190
    },
    {
      "epoch": 0.4959797096756786,
      "grad_norm": 0.892423073722221,
      "learning_rate": 5.5568340044139425e-06,
      "loss": 0.3408,
      "step": 9191
    },
    {
      "epoch": 0.49603367330419296,
      "grad_norm": 0.9367320150823351,
      "learning_rate": 5.55607117162325e-06,
      "loss": 0.3644,
      "step": 9192
    },
    {
      "epoch": 0.49608763693270735,
      "grad_norm": 1.0959787115678488,
      "learning_rate": 5.555308337221014e-06,
      "loss": 0.4929,
      "step": 9193
    },
    {
      "epoch": 0.49614160056122175,
      "grad_norm": 0.9750656341521331,
      "learning_rate": 5.55454550122916e-06,
      "loss": 0.4788,
      "step": 9194
    },
    {
      "epoch": 0.4961955641897361,
      "grad_norm": 1.3758883528219292,
      "learning_rate": 5.5537826636696155e-06,
      "loss": 0.6005,
      "step": 9195
    },
    {
      "epoch": 0.4962495278182505,
      "grad_norm": 1.0745855843060585,
      "learning_rate": 5.553019824564301e-06,
      "loss": 0.4725,
      "step": 9196
    },
    {
      "epoch": 0.4963034914467649,
      "grad_norm": 1.0661749098050677,
      "learning_rate": 5.552256983935144e-06,
      "loss": 0.51,
      "step": 9197
    },
    {
      "epoch": 0.4963574550752793,
      "grad_norm": 0.9633597528343006,
      "learning_rate": 5.551494141804068e-06,
      "loss": 0.4341,
      "step": 9198
    },
    {
      "epoch": 0.4964114187037936,
      "grad_norm": 1.00159650606954,
      "learning_rate": 5.550731298193e-06,
      "loss": 0.5516,
      "step": 9199
    },
    {
      "epoch": 0.496465382332308,
      "grad_norm": 1.3447225878837499,
      "learning_rate": 5.549968453123861e-06,
      "loss": 0.6996,
      "step": 9200
    },
    {
      "epoch": 0.4965193459608224,
      "grad_norm": 1.280700006308369,
      "learning_rate": 5.549205606618578e-06,
      "loss": 0.5394,
      "step": 9201
    },
    {
      "epoch": 0.49657330958933676,
      "grad_norm": 1.1108567295515825,
      "learning_rate": 5.548442758699077e-06,
      "loss": 0.5923,
      "step": 9202
    },
    {
      "epoch": 0.49662727321785116,
      "grad_norm": 1.1432750104183798,
      "learning_rate": 5.547679909387282e-06,
      "loss": 0.5436,
      "step": 9203
    },
    {
      "epoch": 0.49668123684636556,
      "grad_norm": 1.0983881682648324,
      "learning_rate": 5.546917058705119e-06,
      "loss": 0.5283,
      "step": 9204
    },
    {
      "epoch": 0.49673520047487996,
      "grad_norm": 1.2771346967703645,
      "learning_rate": 5.5461542066745114e-06,
      "loss": 0.7373,
      "step": 9205
    },
    {
      "epoch": 0.4967891641033943,
      "grad_norm": 0.971490736383745,
      "learning_rate": 5.545391353317386e-06,
      "loss": 0.5015,
      "step": 9206
    },
    {
      "epoch": 0.4968431277319087,
      "grad_norm": 0.8623511271125768,
      "learning_rate": 5.544628498655666e-06,
      "loss": 0.4362,
      "step": 9207
    },
    {
      "epoch": 0.4968970913604231,
      "grad_norm": 1.1503290824124768,
      "learning_rate": 5.543865642711279e-06,
      "loss": 0.4935,
      "step": 9208
    },
    {
      "epoch": 0.49695105498893744,
      "grad_norm": 1.0619661857935878,
      "learning_rate": 5.54310278550615e-06,
      "loss": 0.578,
      "step": 9209
    },
    {
      "epoch": 0.49700501861745183,
      "grad_norm": 0.9650471928730295,
      "learning_rate": 5.542339927062201e-06,
      "loss": 0.3765,
      "step": 9210
    },
    {
      "epoch": 0.49705898224596623,
      "grad_norm": 0.794416352227625,
      "learning_rate": 5.541577067401361e-06,
      "loss": 0.3715,
      "step": 9211
    },
    {
      "epoch": 0.4971129458744806,
      "grad_norm": 1.0229566478639398,
      "learning_rate": 5.540814206545554e-06,
      "loss": 0.4574,
      "step": 9212
    },
    {
      "epoch": 0.49716690950299497,
      "grad_norm": 0.8964337511047894,
      "learning_rate": 5.5400513445167056e-06,
      "loss": 0.3848,
      "step": 9213
    },
    {
      "epoch": 0.49722087313150937,
      "grad_norm": 1.0810765826900421,
      "learning_rate": 5.539288481336742e-06,
      "loss": 0.5469,
      "step": 9214
    },
    {
      "epoch": 0.49727483676002376,
      "grad_norm": 1.0535324683642482,
      "learning_rate": 5.538525617027586e-06,
      "loss": 0.4165,
      "step": 9215
    },
    {
      "epoch": 0.4973288003885381,
      "grad_norm": 0.9027860304592159,
      "learning_rate": 5.537762751611166e-06,
      "loss": 0.4026,
      "step": 9216
    },
    {
      "epoch": 0.4973827640170525,
      "grad_norm": 0.7965163023205577,
      "learning_rate": 5.536999885109406e-06,
      "loss": 0.339,
      "step": 9217
    },
    {
      "epoch": 0.4974367276455669,
      "grad_norm": 1.0239802585658118,
      "learning_rate": 5.536237017544232e-06,
      "loss": 0.3952,
      "step": 9218
    },
    {
      "epoch": 0.49749069127408124,
      "grad_norm": 0.9568623628908487,
      "learning_rate": 5.535474148937569e-06,
      "loss": 0.3961,
      "step": 9219
    },
    {
      "epoch": 0.49754465490259564,
      "grad_norm": 1.093409086915177,
      "learning_rate": 5.534711279311344e-06,
      "loss": 0.5768,
      "step": 9220
    },
    {
      "epoch": 0.49759861853111004,
      "grad_norm": 1.1255604276549973,
      "learning_rate": 5.5339484086874815e-06,
      "loss": 0.4515,
      "step": 9221
    },
    {
      "epoch": 0.49765258215962443,
      "grad_norm": 1.086018540774263,
      "learning_rate": 5.533185537087906e-06,
      "loss": 0.5212,
      "step": 9222
    },
    {
      "epoch": 0.4977065457881388,
      "grad_norm": 1.0183055843779207,
      "learning_rate": 5.532422664534547e-06,
      "loss": 0.5544,
      "step": 9223
    },
    {
      "epoch": 0.4977605094166532,
      "grad_norm": 1.3098950719552949,
      "learning_rate": 5.531659791049326e-06,
      "loss": 0.6574,
      "step": 9224
    },
    {
      "epoch": 0.49781447304516757,
      "grad_norm": 0.9864156168474227,
      "learning_rate": 5.530896916654171e-06,
      "loss": 0.3561,
      "step": 9225
    },
    {
      "epoch": 0.4978684366736819,
      "grad_norm": 0.9766306184664869,
      "learning_rate": 5.530134041371008e-06,
      "loss": 0.4734,
      "step": 9226
    },
    {
      "epoch": 0.4979224003021963,
      "grad_norm": 1.2778619486007325,
      "learning_rate": 5.529371165221761e-06,
      "loss": 0.6621,
      "step": 9227
    },
    {
      "epoch": 0.4979763639307107,
      "grad_norm": 1.1254282599609082,
      "learning_rate": 5.528608288228359e-06,
      "loss": 0.4042,
      "step": 9228
    },
    {
      "epoch": 0.4980303275592251,
      "grad_norm": 0.9641622553088225,
      "learning_rate": 5.527845410412725e-06,
      "loss": 0.4534,
      "step": 9229
    },
    {
      "epoch": 0.49808429118773945,
      "grad_norm": 0.9693040039312797,
      "learning_rate": 5.5270825317967855e-06,
      "loss": 0.4398,
      "step": 9230
    },
    {
      "epoch": 0.49813825481625384,
      "grad_norm": 0.9995685021550145,
      "learning_rate": 5.5263196524024665e-06,
      "loss": 0.4426,
      "step": 9231
    },
    {
      "epoch": 0.49819221844476824,
      "grad_norm": 1.0956018001855417,
      "learning_rate": 5.525556772251694e-06,
      "loss": 0.4537,
      "step": 9232
    },
    {
      "epoch": 0.4982461820732826,
      "grad_norm": 0.8623665474077875,
      "learning_rate": 5.524793891366396e-06,
      "loss": 0.3951,
      "step": 9233
    },
    {
      "epoch": 0.498300145701797,
      "grad_norm": 1.0760636575520381,
      "learning_rate": 5.524031009768494e-06,
      "loss": 0.4087,
      "step": 9234
    },
    {
      "epoch": 0.4983541093303114,
      "grad_norm": 1.1893784505041916,
      "learning_rate": 5.523268127479916e-06,
      "loss": 0.5255,
      "step": 9235
    },
    {
      "epoch": 0.4984080729588258,
      "grad_norm": 1.042270717143089,
      "learning_rate": 5.5225052445225906e-06,
      "loss": 0.4471,
      "step": 9236
    },
    {
      "epoch": 0.4984620365873401,
      "grad_norm": 1.1287401873514937,
      "learning_rate": 5.5217423609184416e-06,
      "loss": 0.5576,
      "step": 9237
    },
    {
      "epoch": 0.4985160002158545,
      "grad_norm": 1.2856432888116804,
      "learning_rate": 5.520979476689394e-06,
      "loss": 0.4732,
      "step": 9238
    },
    {
      "epoch": 0.4985699638443689,
      "grad_norm": 0.8867049112659914,
      "learning_rate": 5.520216591857376e-06,
      "loss": 0.3618,
      "step": 9239
    },
    {
      "epoch": 0.49862392747288325,
      "grad_norm": 1.0127296504223997,
      "learning_rate": 5.519453706444312e-06,
      "loss": 0.3582,
      "step": 9240
    },
    {
      "epoch": 0.49867789110139765,
      "grad_norm": 0.9843132711681426,
      "learning_rate": 5.518690820472129e-06,
      "loss": 0.4582,
      "step": 9241
    },
    {
      "epoch": 0.49873185472991205,
      "grad_norm": 0.8510992699167186,
      "learning_rate": 5.5179279339627535e-06,
      "loss": 0.3387,
      "step": 9242
    },
    {
      "epoch": 0.49878581835842645,
      "grad_norm": 0.97645120635193,
      "learning_rate": 5.517165046938111e-06,
      "loss": 0.4712,
      "step": 9243
    },
    {
      "epoch": 0.4988397819869408,
      "grad_norm": 0.9981143885678362,
      "learning_rate": 5.516402159420128e-06,
      "loss": 0.42,
      "step": 9244
    },
    {
      "epoch": 0.4988937456154552,
      "grad_norm": 1.1732276826333454,
      "learning_rate": 5.51563927143073e-06,
      "loss": 0.6728,
      "step": 9245
    },
    {
      "epoch": 0.4989477092439696,
      "grad_norm": 0.8261095128713817,
      "learning_rate": 5.514876382991841e-06,
      "loss": 0.3446,
      "step": 9246
    },
    {
      "epoch": 0.4990016728724839,
      "grad_norm": 0.925854981310481,
      "learning_rate": 5.514113494125393e-06,
      "loss": 0.4093,
      "step": 9247
    },
    {
      "epoch": 0.4990556365009983,
      "grad_norm": 1.0880842125976935,
      "learning_rate": 5.513350604853308e-06,
      "loss": 0.6809,
      "step": 9248
    },
    {
      "epoch": 0.4991096001295127,
      "grad_norm": 1.2506835701187293,
      "learning_rate": 5.512587715197514e-06,
      "loss": 0.4694,
      "step": 9249
    },
    {
      "epoch": 0.49916356375802706,
      "grad_norm": 0.9370986148548095,
      "learning_rate": 5.511824825179935e-06,
      "loss": 0.3858,
      "step": 9250
    },
    {
      "epoch": 0.49921752738654146,
      "grad_norm": 1.0524354732457204,
      "learning_rate": 5.511061934822499e-06,
      "loss": 0.4471,
      "step": 9251
    },
    {
      "epoch": 0.49927149101505586,
      "grad_norm": 0.9602750206154824,
      "learning_rate": 5.510299044147133e-06,
      "loss": 0.52,
      "step": 9252
    },
    {
      "epoch": 0.49932545464357025,
      "grad_norm": 1.177104723010309,
      "learning_rate": 5.50953615317576e-06,
      "loss": 0.523,
      "step": 9253
    },
    {
      "epoch": 0.4993794182720846,
      "grad_norm": 0.966807735667726,
      "learning_rate": 5.508773261930309e-06,
      "loss": 0.3839,
      "step": 9254
    },
    {
      "epoch": 0.499433381900599,
      "grad_norm": 1.003084264183784,
      "learning_rate": 5.508010370432705e-06,
      "loss": 0.4512,
      "step": 9255
    },
    {
      "epoch": 0.4994873455291134,
      "grad_norm": 0.9322349644105417,
      "learning_rate": 5.5072474787048754e-06,
      "loss": 0.3882,
      "step": 9256
    },
    {
      "epoch": 0.49954130915762773,
      "grad_norm": 0.9147706700142655,
      "learning_rate": 5.506484586768746e-06,
      "loss": 0.5042,
      "step": 9257
    },
    {
      "epoch": 0.49959527278614213,
      "grad_norm": 0.9575750800222728,
      "learning_rate": 5.5057216946462436e-06,
      "loss": 0.4438,
      "step": 9258
    },
    {
      "epoch": 0.49964923641465653,
      "grad_norm": 0.9482651653102838,
      "learning_rate": 5.504958802359294e-06,
      "loss": 0.418,
      "step": 9259
    },
    {
      "epoch": 0.4997032000431709,
      "grad_norm": 1.006118194211023,
      "learning_rate": 5.504195909929823e-06,
      "loss": 0.5049,
      "step": 9260
    },
    {
      "epoch": 0.49975716367168527,
      "grad_norm": 1.3057530844110719,
      "learning_rate": 5.5034330173797566e-06,
      "loss": 0.5077,
      "step": 9261
    },
    {
      "epoch": 0.49981112730019966,
      "grad_norm": 1.178316076609424,
      "learning_rate": 5.502670124731022e-06,
      "loss": 0.6718,
      "step": 9262
    },
    {
      "epoch": 0.49986509092871406,
      "grad_norm": 0.8997310399230333,
      "learning_rate": 5.501907232005547e-06,
      "loss": 0.4621,
      "step": 9263
    },
    {
      "epoch": 0.4999190545572284,
      "grad_norm": 1.02915296005842,
      "learning_rate": 5.5011443392252554e-06,
      "loss": 0.4749,
      "step": 9264
    },
    {
      "epoch": 0.4999730181857428,
      "grad_norm": 0.9485332686788208,
      "learning_rate": 5.500381446412073e-06,
      "loss": 0.389,
      "step": 9265
    },
    {
      "epoch": 0.5000269818142572,
      "grad_norm": 0.8658289734556776,
      "learning_rate": 5.499618553587929e-06,
      "loss": 0.3832,
      "step": 9266
    },
    {
      "epoch": 0.5000809454427716,
      "grad_norm": 1.0496372779751484,
      "learning_rate": 5.498855660774747e-06,
      "loss": 0.5114,
      "step": 9267
    },
    {
      "epoch": 0.500134909071286,
      "grad_norm": 1.0776992093626383,
      "learning_rate": 5.4980927679944544e-06,
      "loss": 0.5814,
      "step": 9268
    },
    {
      "epoch": 0.5001888726998003,
      "grad_norm": 1.0385389210045457,
      "learning_rate": 5.4973298752689795e-06,
      "loss": 0.4194,
      "step": 9269
    },
    {
      "epoch": 0.5002428363283147,
      "grad_norm": 1.2160638007009277,
      "learning_rate": 5.496566982620245e-06,
      "loss": 0.4903,
      "step": 9270
    },
    {
      "epoch": 0.5002967999568291,
      "grad_norm": 1.1032424548174509,
      "learning_rate": 5.495804090070179e-06,
      "loss": 0.5506,
      "step": 9271
    },
    {
      "epoch": 0.5003507635853435,
      "grad_norm": 1.0774081891951837,
      "learning_rate": 5.495041197640708e-06,
      "loss": 0.5145,
      "step": 9272
    },
    {
      "epoch": 0.5004047272138579,
      "grad_norm": 1.138146251468353,
      "learning_rate": 5.494278305353757e-06,
      "loss": 0.5634,
      "step": 9273
    },
    {
      "epoch": 0.5004586908423723,
      "grad_norm": 0.9021893819040978,
      "learning_rate": 5.493515413231255e-06,
      "loss": 0.4193,
      "step": 9274
    },
    {
      "epoch": 0.5005126544708867,
      "grad_norm": 1.1440714437176214,
      "learning_rate": 5.492752521295125e-06,
      "loss": 0.4581,
      "step": 9275
    },
    {
      "epoch": 0.500566618099401,
      "grad_norm": 1.05128732285786,
      "learning_rate": 5.4919896295672966e-06,
      "loss": 0.5209,
      "step": 9276
    },
    {
      "epoch": 0.5006205817279153,
      "grad_norm": 1.1391533285256124,
      "learning_rate": 5.491226738069694e-06,
      "loss": 0.5411,
      "step": 9277
    },
    {
      "epoch": 0.5006745453564297,
      "grad_norm": 1.112966057322348,
      "learning_rate": 5.490463846824242e-06,
      "loss": 0.5193,
      "step": 9278
    },
    {
      "epoch": 0.5007285089849441,
      "grad_norm": 1.0982054750071115,
      "learning_rate": 5.48970095585287e-06,
      "loss": 0.6495,
      "step": 9279
    },
    {
      "epoch": 0.5007824726134585,
      "grad_norm": 1.327618210267992,
      "learning_rate": 5.488938065177503e-06,
      "loss": 0.6983,
      "step": 9280
    },
    {
      "epoch": 0.5008364362419729,
      "grad_norm": 0.9109597438027638,
      "learning_rate": 5.488175174820066e-06,
      "loss": 0.4684,
      "step": 9281
    },
    {
      "epoch": 0.5008903998704873,
      "grad_norm": 0.9343854830522487,
      "learning_rate": 5.487412284802488e-06,
      "loss": 0.4773,
      "step": 9282
    },
    {
      "epoch": 0.5009443634990016,
      "grad_norm": 0.9932142469256169,
      "learning_rate": 5.486649395146694e-06,
      "loss": 0.4988,
      "step": 9283
    },
    {
      "epoch": 0.500998327127516,
      "grad_norm": 0.8842115771216675,
      "learning_rate": 5.485886505874608e-06,
      "loss": 0.42,
      "step": 9284
    },
    {
      "epoch": 0.5010522907560304,
      "grad_norm": 1.0530549179037192,
      "learning_rate": 5.485123617008159e-06,
      "loss": 0.4296,
      "step": 9285
    },
    {
      "epoch": 0.5011062543845448,
      "grad_norm": 0.9727413825082329,
      "learning_rate": 5.4843607285692724e-06,
      "loss": 0.386,
      "step": 9286
    },
    {
      "epoch": 0.5011602180130592,
      "grad_norm": 1.0484008924346282,
      "learning_rate": 5.483597840579874e-06,
      "loss": 0.493,
      "step": 9287
    },
    {
      "epoch": 0.5012141816415736,
      "grad_norm": 0.7750573658290201,
      "learning_rate": 5.48283495306189e-06,
      "loss": 0.3215,
      "step": 9288
    },
    {
      "epoch": 0.501268145270088,
      "grad_norm": 0.8958063048715763,
      "learning_rate": 5.482072066037247e-06,
      "loss": 0.4441,
      "step": 9289
    },
    {
      "epoch": 0.5013221088986023,
      "grad_norm": 1.3033664504080293,
      "learning_rate": 5.4813091795278725e-06,
      "loss": 0.4811,
      "step": 9290
    },
    {
      "epoch": 0.5013760725271167,
      "grad_norm": 1.0047672702741697,
      "learning_rate": 5.48054629355569e-06,
      "loss": 0.464,
      "step": 9291
    },
    {
      "epoch": 0.5014300361556311,
      "grad_norm": 0.975623346071451,
      "learning_rate": 5.479783408142626e-06,
      "loss": 0.5071,
      "step": 9292
    },
    {
      "epoch": 0.5014839997841455,
      "grad_norm": 0.9963806716973526,
      "learning_rate": 5.479020523310607e-06,
      "loss": 0.5277,
      "step": 9293
    },
    {
      "epoch": 0.5015379634126599,
      "grad_norm": 0.9461420698157814,
      "learning_rate": 5.47825763908156e-06,
      "loss": 0.4261,
      "step": 9294
    },
    {
      "epoch": 0.5015919270411743,
      "grad_norm": 0.8015498138152314,
      "learning_rate": 5.477494755477411e-06,
      "loss": 0.3557,
      "step": 9295
    },
    {
      "epoch": 0.5016458906696887,
      "grad_norm": 1.1572102211494004,
      "learning_rate": 5.476731872520083e-06,
      "loss": 0.6191,
      "step": 9296
    },
    {
      "epoch": 0.501699854298203,
      "grad_norm": 1.5270574786718074,
      "learning_rate": 5.4759689902315095e-06,
      "loss": 0.8715,
      "step": 9297
    },
    {
      "epoch": 0.5017538179267174,
      "grad_norm": 1.0661110022046727,
      "learning_rate": 5.475206108633608e-06,
      "loss": 0.4475,
      "step": 9298
    },
    {
      "epoch": 0.5018077815552318,
      "grad_norm": 1.0660481773833614,
      "learning_rate": 5.4744432277483086e-06,
      "loss": 0.4851,
      "step": 9299
    },
    {
      "epoch": 0.5018617451837462,
      "grad_norm": 1.2121572373473974,
      "learning_rate": 5.473680347597535e-06,
      "loss": 0.713,
      "step": 9300
    },
    {
      "epoch": 0.5019157088122606,
      "grad_norm": 0.9451347813859317,
      "learning_rate": 5.472917468203216e-06,
      "loss": 0.3828,
      "step": 9301
    },
    {
      "epoch": 0.501969672440775,
      "grad_norm": 0.9571776064547108,
      "learning_rate": 5.472154589587276e-06,
      "loss": 0.4208,
      "step": 9302
    },
    {
      "epoch": 0.5020236360692893,
      "grad_norm": 1.2981520078704187,
      "learning_rate": 5.471391711771642e-06,
      "loss": 0.5711,
      "step": 9303
    },
    {
      "epoch": 0.5020775996978036,
      "grad_norm": 0.9095358867037191,
      "learning_rate": 5.47062883477824e-06,
      "loss": 0.3791,
      "step": 9304
    },
    {
      "epoch": 0.502131563326318,
      "grad_norm": 0.946052561951035,
      "learning_rate": 5.4698659586289934e-06,
      "loss": 0.4169,
      "step": 9305
    },
    {
      "epoch": 0.5021855269548324,
      "grad_norm": 0.7803281694778149,
      "learning_rate": 5.46910308334583e-06,
      "loss": 0.3566,
      "step": 9306
    },
    {
      "epoch": 0.5022394905833468,
      "grad_norm": 1.1505084518122253,
      "learning_rate": 5.468340208950675e-06,
      "loss": 0.5978,
      "step": 9307
    },
    {
      "epoch": 0.5022934542118612,
      "grad_norm": 1.041247158163933,
      "learning_rate": 5.467577335465455e-06,
      "loss": 0.5054,
      "step": 9308
    },
    {
      "epoch": 0.5023474178403756,
      "grad_norm": 1.1785699052511733,
      "learning_rate": 5.4668144629120945e-06,
      "loss": 0.5466,
      "step": 9309
    },
    {
      "epoch": 0.50240138146889,
      "grad_norm": 1.142400932213463,
      "learning_rate": 5.46605159131252e-06,
      "loss": 0.5006,
      "step": 9310
    },
    {
      "epoch": 0.5024553450974043,
      "grad_norm": 1.0966349856264155,
      "learning_rate": 5.465288720688659e-06,
      "loss": 0.5167,
      "step": 9311
    },
    {
      "epoch": 0.5025093087259187,
      "grad_norm": 1.104963842653995,
      "learning_rate": 5.464525851062433e-06,
      "loss": 0.4846,
      "step": 9312
    },
    {
      "epoch": 0.5025632723544331,
      "grad_norm": 1.0829797624443036,
      "learning_rate": 5.4637629824557705e-06,
      "loss": 0.4616,
      "step": 9313
    },
    {
      "epoch": 0.5026172359829475,
      "grad_norm": 0.8689456465601028,
      "learning_rate": 5.463000114890595e-06,
      "loss": 0.4699,
      "step": 9314
    },
    {
      "epoch": 0.5026711996114619,
      "grad_norm": 0.7422690248079703,
      "learning_rate": 5.462237248388835e-06,
      "loss": 0.2795,
      "step": 9315
    },
    {
      "epoch": 0.5027251632399763,
      "grad_norm": 1.0830910701152139,
      "learning_rate": 5.461474382972415e-06,
      "loss": 0.4245,
      "step": 9316
    },
    {
      "epoch": 0.5027791268684907,
      "grad_norm": 1.2100651752704135,
      "learning_rate": 5.460711518663259e-06,
      "loss": 0.5455,
      "step": 9317
    },
    {
      "epoch": 0.502833090497005,
      "grad_norm": 0.9480068540747455,
      "learning_rate": 5.459948655483296e-06,
      "loss": 0.4692,
      "step": 9318
    },
    {
      "epoch": 0.5028870541255194,
      "grad_norm": 0.8907689997269927,
      "learning_rate": 5.459185793454448e-06,
      "loss": 0.4237,
      "step": 9319
    },
    {
      "epoch": 0.5029410177540338,
      "grad_norm": 1.0470853913637472,
      "learning_rate": 5.45842293259864e-06,
      "loss": 0.5164,
      "step": 9320
    },
    {
      "epoch": 0.5029949813825482,
      "grad_norm": 0.9642900590236484,
      "learning_rate": 5.4576600729378e-06,
      "loss": 0.3899,
      "step": 9321
    },
    {
      "epoch": 0.5030489450110626,
      "grad_norm": 1.1159835458701577,
      "learning_rate": 5.456897214493851e-06,
      "loss": 0.5039,
      "step": 9322
    },
    {
      "epoch": 0.503102908639577,
      "grad_norm": 1.1579526554987467,
      "learning_rate": 5.4561343572887215e-06,
      "loss": 0.4681,
      "step": 9323
    },
    {
      "epoch": 0.5031568722680914,
      "grad_norm": 1.276968390987905,
      "learning_rate": 5.455371501344334e-06,
      "loss": 0.5481,
      "step": 9324
    },
    {
      "epoch": 0.5032108358966056,
      "grad_norm": 1.0388321532843348,
      "learning_rate": 5.454608646682617e-06,
      "loss": 0.6374,
      "step": 9325
    },
    {
      "epoch": 0.50326479952512,
      "grad_norm": 1.2064306882348976,
      "learning_rate": 5.453845793325491e-06,
      "loss": 0.6834,
      "step": 9326
    },
    {
      "epoch": 0.5033187631536344,
      "grad_norm": 1.2186904519316717,
      "learning_rate": 5.453082941294883e-06,
      "loss": 0.5471,
      "step": 9327
    },
    {
      "epoch": 0.5033727267821488,
      "grad_norm": 1.2250099647692714,
      "learning_rate": 5.452320090612719e-06,
      "loss": 0.4969,
      "step": 9328
    },
    {
      "epoch": 0.5034266904106632,
      "grad_norm": 1.021723245023374,
      "learning_rate": 5.451557241300924e-06,
      "loss": 0.4369,
      "step": 9329
    },
    {
      "epoch": 0.5034806540391776,
      "grad_norm": 0.9541591774335554,
      "learning_rate": 5.450794393381423e-06,
      "loss": 0.436,
      "step": 9330
    },
    {
      "epoch": 0.503534617667692,
      "grad_norm": 0.92751245803067,
      "learning_rate": 5.4500315468761414e-06,
      "loss": 0.4154,
      "step": 9331
    },
    {
      "epoch": 0.5035885812962063,
      "grad_norm": 1.135144226257816,
      "learning_rate": 5.449268701807003e-06,
      "loss": 0.5144,
      "step": 9332
    },
    {
      "epoch": 0.5036425449247207,
      "grad_norm": 1.2550214622349016,
      "learning_rate": 5.448505858195933e-06,
      "loss": 0.5789,
      "step": 9333
    },
    {
      "epoch": 0.5036965085532351,
      "grad_norm": 0.9298814035593941,
      "learning_rate": 5.447743016064858e-06,
      "loss": 0.4026,
      "step": 9334
    },
    {
      "epoch": 0.5037504721817495,
      "grad_norm": 0.7773789389700129,
      "learning_rate": 5.4469801754357e-06,
      "loss": 0.3697,
      "step": 9335
    },
    {
      "epoch": 0.5038044358102639,
      "grad_norm": 0.9350294791123803,
      "learning_rate": 5.446217336330386e-06,
      "loss": 0.4988,
      "step": 9336
    },
    {
      "epoch": 0.5038583994387783,
      "grad_norm": 1.2738846720526575,
      "learning_rate": 5.44545449877084e-06,
      "loss": 0.555,
      "step": 9337
    },
    {
      "epoch": 0.5039123630672926,
      "grad_norm": 0.924502726983835,
      "learning_rate": 5.444691662778989e-06,
      "loss": 0.4182,
      "step": 9338
    },
    {
      "epoch": 0.503966326695807,
      "grad_norm": 1.057086603303938,
      "learning_rate": 5.443928828376753e-06,
      "loss": 0.5273,
      "step": 9339
    },
    {
      "epoch": 0.5040202903243214,
      "grad_norm": 0.9676354184286606,
      "learning_rate": 5.44316599558606e-06,
      "loss": 0.4544,
      "step": 9340
    },
    {
      "epoch": 0.5040742539528358,
      "grad_norm": 1.3511246881178542,
      "learning_rate": 5.442403164428832e-06,
      "loss": 0.6052,
      "step": 9341
    },
    {
      "epoch": 0.5041282175813502,
      "grad_norm": 1.044260220963436,
      "learning_rate": 5.441640334926997e-06,
      "loss": 0.5119,
      "step": 9342
    },
    {
      "epoch": 0.5041821812098646,
      "grad_norm": 0.993627760405897,
      "learning_rate": 5.440877507102476e-06,
      "loss": 0.395,
      "step": 9343
    },
    {
      "epoch": 0.504236144838379,
      "grad_norm": 0.9960225566513855,
      "learning_rate": 5.440114680977196e-06,
      "loss": 0.4481,
      "step": 9344
    },
    {
      "epoch": 0.5042901084668933,
      "grad_norm": 0.9026869645671672,
      "learning_rate": 5.439351856573084e-06,
      "loss": 0.5248,
      "step": 9345
    },
    {
      "epoch": 0.5043440720954077,
      "grad_norm": 1.1328770918576176,
      "learning_rate": 5.438589033912057e-06,
      "loss": 0.427,
      "step": 9346
    },
    {
      "epoch": 0.5043980357239221,
      "grad_norm": 1.180925041264739,
      "learning_rate": 5.4378262130160445e-06,
      "loss": 0.703,
      "step": 9347
    },
    {
      "epoch": 0.5044519993524365,
      "grad_norm": 1.0564949353217257,
      "learning_rate": 5.437063393906969e-06,
      "loss": 0.5086,
      "step": 9348
    },
    {
      "epoch": 0.5045059629809509,
      "grad_norm": 1.0312572608066808,
      "learning_rate": 5.436300576606755e-06,
      "loss": 0.4891,
      "step": 9349
    },
    {
      "epoch": 0.5045599266094652,
      "grad_norm": 1.1668966911616434,
      "learning_rate": 5.435537761137327e-06,
      "loss": 0.6023,
      "step": 9350
    },
    {
      "epoch": 0.5046138902379796,
      "grad_norm": 1.1934751376681627,
      "learning_rate": 5.434774947520609e-06,
      "loss": 0.6304,
      "step": 9351
    },
    {
      "epoch": 0.5046678538664939,
      "grad_norm": 1.066615198981386,
      "learning_rate": 5.4340121357785246e-06,
      "loss": 0.526,
      "step": 9352
    },
    {
      "epoch": 0.5047218174950083,
      "grad_norm": 1.1245175247197177,
      "learning_rate": 5.433249325932999e-06,
      "loss": 0.4364,
      "step": 9353
    },
    {
      "epoch": 0.5047757811235227,
      "grad_norm": 0.957982125928268,
      "learning_rate": 5.4324865180059535e-06,
      "loss": 0.4167,
      "step": 9354
    },
    {
      "epoch": 0.5048297447520371,
      "grad_norm": 0.7667702722615145,
      "learning_rate": 5.431723712019315e-06,
      "loss": 0.2979,
      "step": 9355
    },
    {
      "epoch": 0.5048837083805515,
      "grad_norm": 1.0031592646840641,
      "learning_rate": 5.430960907995006e-06,
      "loss": 0.4156,
      "step": 9356
    },
    {
      "epoch": 0.5049376720090659,
      "grad_norm": 0.8468498740891265,
      "learning_rate": 5.43019810595495e-06,
      "loss": 0.3304,
      "step": 9357
    },
    {
      "epoch": 0.5049916356375803,
      "grad_norm": 1.0078863985549582,
      "learning_rate": 5.429435305921072e-06,
      "loss": 0.3914,
      "step": 9358
    },
    {
      "epoch": 0.5050455992660946,
      "grad_norm": 0.9361936802532981,
      "learning_rate": 5.428672507915293e-06,
      "loss": 0.4904,
      "step": 9359
    },
    {
      "epoch": 0.505099562894609,
      "grad_norm": 0.9259988693189075,
      "learning_rate": 5.42790971195954e-06,
      "loss": 0.4591,
      "step": 9360
    },
    {
      "epoch": 0.5051535265231234,
      "grad_norm": 0.9686011991388224,
      "learning_rate": 5.427146918075733e-06,
      "loss": 0.4575,
      "step": 9361
    },
    {
      "epoch": 0.5052074901516378,
      "grad_norm": 1.0954276125105042,
      "learning_rate": 5.426384126285798e-06,
      "loss": 0.5353,
      "step": 9362
    },
    {
      "epoch": 0.5052614537801522,
      "grad_norm": 1.1005415685953186,
      "learning_rate": 5.425621336611656e-06,
      "loss": 0.5944,
      "step": 9363
    },
    {
      "epoch": 0.5053154174086666,
      "grad_norm": 0.9103563331726027,
      "learning_rate": 5.4248585490752324e-06,
      "loss": 0.3945,
      "step": 9364
    },
    {
      "epoch": 0.505369381037181,
      "grad_norm": 1.0351248138314637,
      "learning_rate": 5.424095763698452e-06,
      "loss": 0.4892,
      "step": 9365
    },
    {
      "epoch": 0.5054233446656953,
      "grad_norm": 1.059695637475595,
      "learning_rate": 5.423332980503237e-06,
      "loss": 0.3783,
      "step": 9366
    },
    {
      "epoch": 0.5054773082942097,
      "grad_norm": 1.0411662641081108,
      "learning_rate": 5.422570199511507e-06,
      "loss": 0.5338,
      "step": 9367
    },
    {
      "epoch": 0.5055312719227241,
      "grad_norm": 1.0413124501990882,
      "learning_rate": 5.421807420745188e-06,
      "loss": 0.4821,
      "step": 9368
    },
    {
      "epoch": 0.5055852355512385,
      "grad_norm": 1.116746175507263,
      "learning_rate": 5.421044644226202e-06,
      "loss": 0.47,
      "step": 9369
    },
    {
      "epoch": 0.5056391991797529,
      "grad_norm": 0.8935173049739824,
      "learning_rate": 5.420281869976474e-06,
      "loss": 0.3627,
      "step": 9370
    },
    {
      "epoch": 0.5056931628082673,
      "grad_norm": 1.1863786720400902,
      "learning_rate": 5.419519098017925e-06,
      "loss": 0.5352,
      "step": 9371
    },
    {
      "epoch": 0.5057471264367817,
      "grad_norm": 1.051763762279436,
      "learning_rate": 5.418756328372477e-06,
      "loss": 0.3915,
      "step": 9372
    },
    {
      "epoch": 0.5058010900652959,
      "grad_norm": 1.048466154531086,
      "learning_rate": 5.417993561062059e-06,
      "loss": 0.3858,
      "step": 9373
    },
    {
      "epoch": 0.5058550536938103,
      "grad_norm": 1.0695401828727997,
      "learning_rate": 5.417230796108586e-06,
      "loss": 0.4799,
      "step": 9374
    },
    {
      "epoch": 0.5059090173223247,
      "grad_norm": 1.0065354471231704,
      "learning_rate": 5.416468033533982e-06,
      "loss": 0.351,
      "step": 9375
    },
    {
      "epoch": 0.5059629809508391,
      "grad_norm": 1.042728182586893,
      "learning_rate": 5.415705273360173e-06,
      "loss": 0.5501,
      "step": 9376
    },
    {
      "epoch": 0.5060169445793535,
      "grad_norm": 1.010689750315454,
      "learning_rate": 5.414942515609078e-06,
      "loss": 0.3562,
      "step": 9377
    },
    {
      "epoch": 0.5060709082078679,
      "grad_norm": 0.8755094726805982,
      "learning_rate": 5.414179760302623e-06,
      "loss": 0.3633,
      "step": 9378
    },
    {
      "epoch": 0.5061248718363823,
      "grad_norm": 0.9495270741376999,
      "learning_rate": 5.413417007462728e-06,
      "loss": 0.421,
      "step": 9379
    },
    {
      "epoch": 0.5061788354648966,
      "grad_norm": 0.8593685531327266,
      "learning_rate": 5.412654257111315e-06,
      "loss": 0.4134,
      "step": 9380
    },
    {
      "epoch": 0.506232799093411,
      "grad_norm": 1.0699045811038568,
      "learning_rate": 5.411891509270307e-06,
      "loss": 0.4354,
      "step": 9381
    },
    {
      "epoch": 0.5062867627219254,
      "grad_norm": 0.8893994365592468,
      "learning_rate": 5.411128763961626e-06,
      "loss": 0.3717,
      "step": 9382
    },
    {
      "epoch": 0.5063407263504398,
      "grad_norm": 1.3139095691493066,
      "learning_rate": 5.410366021207194e-06,
      "loss": 0.6111,
      "step": 9383
    },
    {
      "epoch": 0.5063946899789542,
      "grad_norm": 1.1950607669509459,
      "learning_rate": 5.409603281028933e-06,
      "loss": 0.5038,
      "step": 9384
    },
    {
      "epoch": 0.5064486536074686,
      "grad_norm": 0.9788738478249895,
      "learning_rate": 5.408840543448766e-06,
      "loss": 0.4019,
      "step": 9385
    },
    {
      "epoch": 0.506502617235983,
      "grad_norm": 0.8288213805411832,
      "learning_rate": 5.408077808488613e-06,
      "loss": 0.3655,
      "step": 9386
    },
    {
      "epoch": 0.5065565808644973,
      "grad_norm": 1.2524665935527506,
      "learning_rate": 5.407315076170398e-06,
      "loss": 0.6362,
      "step": 9387
    },
    {
      "epoch": 0.5066105444930117,
      "grad_norm": 1.0746720028850933,
      "learning_rate": 5.406552346516042e-06,
      "loss": 0.3777,
      "step": 9388
    },
    {
      "epoch": 0.5066645081215261,
      "grad_norm": 0.9934691439107699,
      "learning_rate": 5.405789619547463e-06,
      "loss": 0.5021,
      "step": 9389
    },
    {
      "epoch": 0.5067184717500405,
      "grad_norm": 0.8203333770994206,
      "learning_rate": 5.4050268952865866e-06,
      "loss": 0.438,
      "step": 9390
    },
    {
      "epoch": 0.5067724353785549,
      "grad_norm": 0.7816133871320595,
      "learning_rate": 5.404264173755334e-06,
      "loss": 0.2878,
      "step": 9391
    },
    {
      "epoch": 0.5068263990070693,
      "grad_norm": 1.3144088759289876,
      "learning_rate": 5.403501454975625e-06,
      "loss": 0.6818,
      "step": 9392
    },
    {
      "epoch": 0.5068803626355837,
      "grad_norm": 0.8907729879494586,
      "learning_rate": 5.402738738969382e-06,
      "loss": 0.4069,
      "step": 9393
    },
    {
      "epoch": 0.506934326264098,
      "grad_norm": 0.9302364476112674,
      "learning_rate": 5.401976025758527e-06,
      "loss": 0.4518,
      "step": 9394
    },
    {
      "epoch": 0.5069882898926124,
      "grad_norm": 1.034071783507473,
      "learning_rate": 5.401213315364979e-06,
      "loss": 0.4416,
      "step": 9395
    },
    {
      "epoch": 0.5070422535211268,
      "grad_norm": 0.9394050142986388,
      "learning_rate": 5.40045060781066e-06,
      "loss": 0.4656,
      "step": 9396
    },
    {
      "epoch": 0.5070962171496411,
      "grad_norm": 1.0506621977933974,
      "learning_rate": 5.399687903117491e-06,
      "loss": 0.4468,
      "step": 9397
    },
    {
      "epoch": 0.5071501807781555,
      "grad_norm": 0.9107891719361408,
      "learning_rate": 5.398925201307394e-06,
      "loss": 0.4129,
      "step": 9398
    },
    {
      "epoch": 0.5072041444066699,
      "grad_norm": 1.1058774710196875,
      "learning_rate": 5.398162502402287e-06,
      "loss": 0.5214,
      "step": 9399
    },
    {
      "epoch": 0.5072581080351843,
      "grad_norm": 1.0996782278554278,
      "learning_rate": 5.397399806424094e-06,
      "loss": 0.5603,
      "step": 9400
    },
    {
      "epoch": 0.5073120716636986,
      "grad_norm": 1.00967021738363,
      "learning_rate": 5.396637113394735e-06,
      "loss": 0.4306,
      "step": 9401
    },
    {
      "epoch": 0.507366035292213,
      "grad_norm": 1.1243844499726794,
      "learning_rate": 5.3958744233361285e-06,
      "loss": 0.5971,
      "step": 9402
    },
    {
      "epoch": 0.5074199989207274,
      "grad_norm": 1.0569794155400694,
      "learning_rate": 5.395111736270196e-06,
      "loss": 0.4198,
      "step": 9403
    },
    {
      "epoch": 0.5074739625492418,
      "grad_norm": 1.0131338626089779,
      "learning_rate": 5.394349052218859e-06,
      "loss": 0.7423,
      "step": 9404
    },
    {
      "epoch": 0.5075279261777562,
      "grad_norm": 0.9786948776115592,
      "learning_rate": 5.393586371204036e-06,
      "loss": 0.4106,
      "step": 9405
    },
    {
      "epoch": 0.5075818898062706,
      "grad_norm": 1.1244090308719106,
      "learning_rate": 5.3928236932476484e-06,
      "loss": 0.5618,
      "step": 9406
    },
    {
      "epoch": 0.5076358534347849,
      "grad_norm": 0.893154450020052,
      "learning_rate": 5.392061018371618e-06,
      "loss": 0.3424,
      "step": 9407
    },
    {
      "epoch": 0.5076898170632993,
      "grad_norm": 1.1119116900667911,
      "learning_rate": 5.391298346597862e-06,
      "loss": 0.5376,
      "step": 9408
    },
    {
      "epoch": 0.5077437806918137,
      "grad_norm": 0.8961298384644844,
      "learning_rate": 5.390535677948301e-06,
      "loss": 0.468,
      "step": 9409
    },
    {
      "epoch": 0.5077977443203281,
      "grad_norm": 1.0652336094787611,
      "learning_rate": 5.3897730124448545e-06,
      "loss": 0.5296,
      "step": 9410
    },
    {
      "epoch": 0.5078517079488425,
      "grad_norm": 1.006605202122239,
      "learning_rate": 5.3890103501094435e-06,
      "loss": 0.4375,
      "step": 9411
    },
    {
      "epoch": 0.5079056715773569,
      "grad_norm": 1.0441137668653337,
      "learning_rate": 5.3882476909639865e-06,
      "loss": 0.5252,
      "step": 9412
    },
    {
      "epoch": 0.5079596352058713,
      "grad_norm": 1.0806736314001495,
      "learning_rate": 5.387485035030406e-06,
      "loss": 0.5954,
      "step": 9413
    },
    {
      "epoch": 0.5080135988343856,
      "grad_norm": 1.1180207899052648,
      "learning_rate": 5.3867223823306195e-06,
      "loss": 0.4105,
      "step": 9414
    },
    {
      "epoch": 0.5080675624629,
      "grad_norm": 1.2201159616781307,
      "learning_rate": 5.3859597328865435e-06,
      "loss": 0.5592,
      "step": 9415
    },
    {
      "epoch": 0.5081215260914144,
      "grad_norm": 1.0405259013909465,
      "learning_rate": 5.385197086720101e-06,
      "loss": 0.3972,
      "step": 9416
    },
    {
      "epoch": 0.5081754897199288,
      "grad_norm": 0.8797325875213612,
      "learning_rate": 5.3844344438532105e-06,
      "loss": 0.4242,
      "step": 9417
    },
    {
      "epoch": 0.5082294533484432,
      "grad_norm": 0.8144842574907457,
      "learning_rate": 5.383671804307789e-06,
      "loss": 0.3655,
      "step": 9418
    },
    {
      "epoch": 0.5082834169769576,
      "grad_norm": 1.2600851776122295,
      "learning_rate": 5.38290916810576e-06,
      "loss": 0.5017,
      "step": 9419
    },
    {
      "epoch": 0.508337380605472,
      "grad_norm": 1.376890497699021,
      "learning_rate": 5.3821465352690384e-06,
      "loss": 0.7611,
      "step": 9420
    },
    {
      "epoch": 0.5083913442339862,
      "grad_norm": 1.110166377031791,
      "learning_rate": 5.381383905819547e-06,
      "loss": 0.6109,
      "step": 9421
    },
    {
      "epoch": 0.5084453078625006,
      "grad_norm": 1.1443386370529234,
      "learning_rate": 5.3806212797792004e-06,
      "loss": 0.6569,
      "step": 9422
    },
    {
      "epoch": 0.508499271491015,
      "grad_norm": 0.8865349740549396,
      "learning_rate": 5.3798586571699176e-06,
      "loss": 0.4071,
      "step": 9423
    },
    {
      "epoch": 0.5085532351195294,
      "grad_norm": 0.9851042462143774,
      "learning_rate": 5.3790960380136196e-06,
      "loss": 0.4647,
      "step": 9424
    },
    {
      "epoch": 0.5086071987480438,
      "grad_norm": 1.0921081506411567,
      "learning_rate": 5.3783334223322226e-06,
      "loss": 0.5272,
      "step": 9425
    },
    {
      "epoch": 0.5086611623765582,
      "grad_norm": 1.028663338945143,
      "learning_rate": 5.377570810147646e-06,
      "loss": 0.4466,
      "step": 9426
    },
    {
      "epoch": 0.5087151260050726,
      "grad_norm": 0.9695302465418382,
      "learning_rate": 5.376808201481809e-06,
      "loss": 0.5154,
      "step": 9427
    },
    {
      "epoch": 0.5087690896335869,
      "grad_norm": 1.0613315868994098,
      "learning_rate": 5.376045596356629e-06,
      "loss": 0.5275,
      "step": 9428
    },
    {
      "epoch": 0.5088230532621013,
      "grad_norm": 0.9600155062573511,
      "learning_rate": 5.375282994794023e-06,
      "loss": 0.5677,
      "step": 9429
    },
    {
      "epoch": 0.5088770168906157,
      "grad_norm": 1.2369017440829464,
      "learning_rate": 5.37452039681591e-06,
      "loss": 0.7897,
      "step": 9430
    },
    {
      "epoch": 0.5089309805191301,
      "grad_norm": 0.8346290853701288,
      "learning_rate": 5.373757802444208e-06,
      "loss": 0.3288,
      "step": 9431
    },
    {
      "epoch": 0.5089849441476445,
      "grad_norm": 1.152197974099437,
      "learning_rate": 5.372995211700833e-06,
      "loss": 0.4831,
      "step": 9432
    },
    {
      "epoch": 0.5090389077761589,
      "grad_norm": 0.9976002295486909,
      "learning_rate": 5.372232624607705e-06,
      "loss": 0.5326,
      "step": 9433
    },
    {
      "epoch": 0.5090928714046733,
      "grad_norm": 1.105605891311398,
      "learning_rate": 5.3714700411867405e-06,
      "loss": 0.4563,
      "step": 9434
    },
    {
      "epoch": 0.5091468350331876,
      "grad_norm": 0.9801942386580135,
      "learning_rate": 5.3707074614598575e-06,
      "loss": 0.5085,
      "step": 9435
    },
    {
      "epoch": 0.509200798661702,
      "grad_norm": 0.8388572354012555,
      "learning_rate": 5.3699448854489725e-06,
      "loss": 0.3667,
      "step": 9436
    },
    {
      "epoch": 0.5092547622902164,
      "grad_norm": 0.9767356605098251,
      "learning_rate": 5.3691823131760015e-06,
      "loss": 0.401,
      "step": 9437
    },
    {
      "epoch": 0.5093087259187308,
      "grad_norm": 0.8999967874237264,
      "learning_rate": 5.368419744662864e-06,
      "loss": 0.3666,
      "step": 9438
    },
    {
      "epoch": 0.5093626895472452,
      "grad_norm": 1.0605515908494463,
      "learning_rate": 5.367657179931475e-06,
      "loss": 0.5408,
      "step": 9439
    },
    {
      "epoch": 0.5094166531757596,
      "grad_norm": 0.8918331066264457,
      "learning_rate": 5.366894619003753e-06,
      "loss": 0.3501,
      "step": 9440
    },
    {
      "epoch": 0.509470616804274,
      "grad_norm": 1.019115625299005,
      "learning_rate": 5.3661320619016135e-06,
      "loss": 0.5123,
      "step": 9441
    },
    {
      "epoch": 0.5095245804327883,
      "grad_norm": 1.4176605397142628,
      "learning_rate": 5.365369508646977e-06,
      "loss": 0.4697,
      "step": 9442
    },
    {
      "epoch": 0.5095785440613027,
      "grad_norm": 1.0553679959858606,
      "learning_rate": 5.364606959261754e-06,
      "loss": 0.4904,
      "step": 9443
    },
    {
      "epoch": 0.509632507689817,
      "grad_norm": 1.3222229409071806,
      "learning_rate": 5.363844413767864e-06,
      "loss": 0.4557,
      "step": 9444
    },
    {
      "epoch": 0.5096864713183314,
      "grad_norm": 1.0511353815921147,
      "learning_rate": 5.363081872187224e-06,
      "loss": 0.5261,
      "step": 9445
    },
    {
      "epoch": 0.5097404349468458,
      "grad_norm": 1.0385258689720664,
      "learning_rate": 5.362319334541748e-06,
      "loss": 0.5468,
      "step": 9446
    },
    {
      "epoch": 0.5097943985753602,
      "grad_norm": 1.0546387415950829,
      "learning_rate": 5.361556800853354e-06,
      "loss": 0.4149,
      "step": 9447
    },
    {
      "epoch": 0.5098483622038746,
      "grad_norm": 1.1728757843616993,
      "learning_rate": 5.360794271143956e-06,
      "loss": 0.5437,
      "step": 9448
    },
    {
      "epoch": 0.5099023258323889,
      "grad_norm": 0.872400591515787,
      "learning_rate": 5.3600317454354755e-06,
      "loss": 0.4188,
      "step": 9449
    },
    {
      "epoch": 0.5099562894609033,
      "grad_norm": 0.8344911400215077,
      "learning_rate": 5.359269223749821e-06,
      "loss": 0.3577,
      "step": 9450
    },
    {
      "epoch": 0.5100102530894177,
      "grad_norm": 1.095191739649592,
      "learning_rate": 5.358506706108912e-06,
      "loss": 0.5297,
      "step": 9451
    },
    {
      "epoch": 0.5100642167179321,
      "grad_norm": 1.1650270543834973,
      "learning_rate": 5.357744192534662e-06,
      "loss": 0.6104,
      "step": 9452
    },
    {
      "epoch": 0.5101181803464465,
      "grad_norm": 0.9825647577285964,
      "learning_rate": 5.356981683048989e-06,
      "loss": 0.4341,
      "step": 9453
    },
    {
      "epoch": 0.5101721439749609,
      "grad_norm": 0.9988132219776641,
      "learning_rate": 5.356219177673806e-06,
      "loss": 0.4224,
      "step": 9454
    },
    {
      "epoch": 0.5102261076034753,
      "grad_norm": 1.2549631660856984,
      "learning_rate": 5.3554566764310286e-06,
      "loss": 0.6569,
      "step": 9455
    },
    {
      "epoch": 0.5102800712319896,
      "grad_norm": 1.2811279465075511,
      "learning_rate": 5.354694179342575e-06,
      "loss": 0.5737,
      "step": 9456
    },
    {
      "epoch": 0.510334034860504,
      "grad_norm": 1.1614188992985968,
      "learning_rate": 5.353931686430356e-06,
      "loss": 0.5508,
      "step": 9457
    },
    {
      "epoch": 0.5103879984890184,
      "grad_norm": 0.9940653919749042,
      "learning_rate": 5.353169197716287e-06,
      "loss": 0.3259,
      "step": 9458
    },
    {
      "epoch": 0.5104419621175328,
      "grad_norm": 1.0397730949597248,
      "learning_rate": 5.352406713222284e-06,
      "loss": 0.4376,
      "step": 9459
    },
    {
      "epoch": 0.5104959257460472,
      "grad_norm": 1.1631554022276065,
      "learning_rate": 5.351644232970261e-06,
      "loss": 0.6159,
      "step": 9460
    },
    {
      "epoch": 0.5105498893745616,
      "grad_norm": 1.3605443280130733,
      "learning_rate": 5.350881756982134e-06,
      "loss": 0.554,
      "step": 9461
    },
    {
      "epoch": 0.510603853003076,
      "grad_norm": 0.9794408626484726,
      "learning_rate": 5.350119285279816e-06,
      "loss": 0.5235,
      "step": 9462
    },
    {
      "epoch": 0.5106578166315903,
      "grad_norm": 0.9025172651539392,
      "learning_rate": 5.349356817885221e-06,
      "loss": 0.3805,
      "step": 9463
    },
    {
      "epoch": 0.5107117802601047,
      "grad_norm": 0.8985388723300692,
      "learning_rate": 5.348594354820262e-06,
      "loss": 0.4272,
      "step": 9464
    },
    {
      "epoch": 0.5107657438886191,
      "grad_norm": 0.9332492290462433,
      "learning_rate": 5.347831896106855e-06,
      "loss": 0.3846,
      "step": 9465
    },
    {
      "epoch": 0.5108197075171335,
      "grad_norm": 1.1903522300638305,
      "learning_rate": 5.3470694417669125e-06,
      "loss": 0.7522,
      "step": 9466
    },
    {
      "epoch": 0.5108736711456479,
      "grad_norm": 1.052427708161447,
      "learning_rate": 5.34630699182235e-06,
      "loss": 0.4175,
      "step": 9467
    },
    {
      "epoch": 0.5109276347741623,
      "grad_norm": 1.22906191734881,
      "learning_rate": 5.345544546295079e-06,
      "loss": 0.5449,
      "step": 9468
    },
    {
      "epoch": 0.5109815984026767,
      "grad_norm": 1.0489256322244487,
      "learning_rate": 5.344782105207013e-06,
      "loss": 0.4993,
      "step": 9469
    },
    {
      "epoch": 0.5110355620311909,
      "grad_norm": 0.8407264555357644,
      "learning_rate": 5.344019668580069e-06,
      "loss": 0.3819,
      "step": 9470
    },
    {
      "epoch": 0.5110895256597053,
      "grad_norm": 1.2270976850278346,
      "learning_rate": 5.343257236436155e-06,
      "loss": 0.4704,
      "step": 9471
    },
    {
      "epoch": 0.5111434892882197,
      "grad_norm": 0.9334152416333615,
      "learning_rate": 5.342494808797186e-06,
      "loss": 0.5237,
      "step": 9472
    },
    {
      "epoch": 0.5111974529167341,
      "grad_norm": 1.0388990990932683,
      "learning_rate": 5.341732385685077e-06,
      "loss": 0.4912,
      "step": 9473
    },
    {
      "epoch": 0.5112514165452485,
      "grad_norm": 0.767349263070513,
      "learning_rate": 5.3409699671217375e-06,
      "loss": 0.2651,
      "step": 9474
    },
    {
      "epoch": 0.5113053801737629,
      "grad_norm": 0.9853482995029555,
      "learning_rate": 5.340207553129082e-06,
      "loss": 0.5135,
      "step": 9475
    },
    {
      "epoch": 0.5113593438022772,
      "grad_norm": 0.9075800943197744,
      "learning_rate": 5.339445143729025e-06,
      "loss": 0.4034,
      "step": 9476
    },
    {
      "epoch": 0.5114133074307916,
      "grad_norm": 0.967253491797828,
      "learning_rate": 5.338682738943476e-06,
      "loss": 0.5876,
      "step": 9477
    },
    {
      "epoch": 0.511467271059306,
      "grad_norm": 1.1017140193584496,
      "learning_rate": 5.337920338794347e-06,
      "loss": 0.521,
      "step": 9478
    },
    {
      "epoch": 0.5115212346878204,
      "grad_norm": 0.9055936794609204,
      "learning_rate": 5.33715794330355e-06,
      "loss": 0.4222,
      "step": 9479
    },
    {
      "epoch": 0.5115751983163348,
      "grad_norm": 1.0025306627528165,
      "learning_rate": 5.336395552493e-06,
      "loss": 0.4276,
      "step": 9480
    },
    {
      "epoch": 0.5116291619448492,
      "grad_norm": 1.1533059748446113,
      "learning_rate": 5.335633166384608e-06,
      "loss": 0.5143,
      "step": 9481
    },
    {
      "epoch": 0.5116831255733636,
      "grad_norm": 0.9801555103829211,
      "learning_rate": 5.334870785000283e-06,
      "loss": 0.4028,
      "step": 9482
    },
    {
      "epoch": 0.5117370892018779,
      "grad_norm": 1.1639309709593904,
      "learning_rate": 5.33410840836194e-06,
      "loss": 0.4526,
      "step": 9483
    },
    {
      "epoch": 0.5117910528303923,
      "grad_norm": 1.2616770189198094,
      "learning_rate": 5.333346036491488e-06,
      "loss": 0.554,
      "step": 9484
    },
    {
      "epoch": 0.5118450164589067,
      "grad_norm": 1.0434637248640053,
      "learning_rate": 5.332583669410839e-06,
      "loss": 0.5137,
      "step": 9485
    },
    {
      "epoch": 0.5118989800874211,
      "grad_norm": 1.1223669099574738,
      "learning_rate": 5.331821307141906e-06,
      "loss": 0.3561,
      "step": 9486
    },
    {
      "epoch": 0.5119529437159355,
      "grad_norm": 1.0142832592253106,
      "learning_rate": 5.331058949706597e-06,
      "loss": 0.5087,
      "step": 9487
    },
    {
      "epoch": 0.5120069073444499,
      "grad_norm": 0.7799546641256196,
      "learning_rate": 5.330296597126825e-06,
      "loss": 0.3253,
      "step": 9488
    },
    {
      "epoch": 0.5120608709729643,
      "grad_norm": 0.9415499244953475,
      "learning_rate": 5.3295342494244995e-06,
      "loss": 0.4006,
      "step": 9489
    },
    {
      "epoch": 0.5121148346014786,
      "grad_norm": 0.9812011633594355,
      "learning_rate": 5.328771906621535e-06,
      "loss": 0.3876,
      "step": 9490
    },
    {
      "epoch": 0.512168798229993,
      "grad_norm": 0.8517014601324987,
      "learning_rate": 5.3280095687398355e-06,
      "loss": 0.4424,
      "step": 9491
    },
    {
      "epoch": 0.5122227618585073,
      "grad_norm": 0.8686420387461525,
      "learning_rate": 5.327247235801317e-06,
      "loss": 0.3667,
      "step": 9492
    },
    {
      "epoch": 0.5122767254870217,
      "grad_norm": 1.0073453193635682,
      "learning_rate": 5.326484907827886e-06,
      "loss": 0.4401,
      "step": 9493
    },
    {
      "epoch": 0.5123306891155361,
      "grad_norm": 1.1297567902788728,
      "learning_rate": 5.325722584841454e-06,
      "loss": 0.5509,
      "step": 9494
    },
    {
      "epoch": 0.5123846527440505,
      "grad_norm": 1.3931697942261212,
      "learning_rate": 5.3249602668639324e-06,
      "loss": 0.621,
      "step": 9495
    },
    {
      "epoch": 0.5124386163725649,
      "grad_norm": 1.0443065346933265,
      "learning_rate": 5.3241979539172286e-06,
      "loss": 0.5686,
      "step": 9496
    },
    {
      "epoch": 0.5124925800010792,
      "grad_norm": 1.007221749009845,
      "learning_rate": 5.323435646023255e-06,
      "loss": 0.4122,
      "step": 9497
    },
    {
      "epoch": 0.5125465436295936,
      "grad_norm": 0.7622185133305635,
      "learning_rate": 5.322673343203917e-06,
      "loss": 0.3528,
      "step": 9498
    },
    {
      "epoch": 0.512600507258108,
      "grad_norm": 0.8833510376897722,
      "learning_rate": 5.321911045481127e-06,
      "loss": 0.3657,
      "step": 9499
    },
    {
      "epoch": 0.5126544708866224,
      "grad_norm": 0.9310437674593675,
      "learning_rate": 5.321148752876794e-06,
      "loss": 0.443,
      "step": 9500
    },
    {
      "epoch": 0.5126544708866224,
      "eval_loss": 0.5477094054222107,
      "eval_runtime": 163.9886,
      "eval_samples_per_second": 20.971,
      "eval_steps_per_second": 0.878,
      "step": 9500
    },
    {
      "epoch": 0.5127084345151368,
      "grad_norm": 1.0137443824087755,
      "learning_rate": 5.3203864654128255e-06,
      "loss": 0.521,
      "step": 9501
    },
    {
      "epoch": 0.5127623981436512,
      "grad_norm": 1.0439389148806155,
      "learning_rate": 5.319624183111132e-06,
      "loss": 0.5964,
      "step": 9502
    },
    {
      "epoch": 0.5128163617721656,
      "grad_norm": 0.9730908661302556,
      "learning_rate": 5.318861905993621e-06,
      "loss": 0.5301,
      "step": 9503
    },
    {
      "epoch": 0.5128703254006799,
      "grad_norm": 0.944760340021586,
      "learning_rate": 5.318099634082205e-06,
      "loss": 0.6459,
      "step": 9504
    },
    {
      "epoch": 0.5129242890291943,
      "grad_norm": 0.7458453262845326,
      "learning_rate": 5.317337367398785e-06,
      "loss": 0.2956,
      "step": 9505
    },
    {
      "epoch": 0.5129782526577087,
      "grad_norm": 1.1366312343871146,
      "learning_rate": 5.316575105965277e-06,
      "loss": 0.4922,
      "step": 9506
    },
    {
      "epoch": 0.5130322162862231,
      "grad_norm": 0.8555854261054769,
      "learning_rate": 5.315812849803585e-06,
      "loss": 0.3898,
      "step": 9507
    },
    {
      "epoch": 0.5130861799147375,
      "grad_norm": 0.8645922830620804,
      "learning_rate": 5.315050598935616e-06,
      "loss": 0.3572,
      "step": 9508
    },
    {
      "epoch": 0.5131401435432519,
      "grad_norm": 1.0909336148921638,
      "learning_rate": 5.314288353383281e-06,
      "loss": 0.5596,
      "step": 9509
    },
    {
      "epoch": 0.5131941071717663,
      "grad_norm": 1.0284089170641122,
      "learning_rate": 5.313526113168486e-06,
      "loss": 0.5023,
      "step": 9510
    },
    {
      "epoch": 0.5132480708002806,
      "grad_norm": 1.0809901877259969,
      "learning_rate": 5.3127638783131394e-06,
      "loss": 0.5193,
      "step": 9511
    },
    {
      "epoch": 0.513302034428795,
      "grad_norm": 0.9226007170378812,
      "learning_rate": 5.312001648839148e-06,
      "loss": 0.4076,
      "step": 9512
    },
    {
      "epoch": 0.5133559980573094,
      "grad_norm": 1.0523353943724705,
      "learning_rate": 5.31123942476842e-06,
      "loss": 0.4356,
      "step": 9513
    },
    {
      "epoch": 0.5134099616858238,
      "grad_norm": 1.171670516093928,
      "learning_rate": 5.3104772061228595e-06,
      "loss": 0.4759,
      "step": 9514
    },
    {
      "epoch": 0.5134639253143382,
      "grad_norm": 0.8514167938978103,
      "learning_rate": 5.3097149929243776e-06,
      "loss": 0.356,
      "step": 9515
    },
    {
      "epoch": 0.5135178889428526,
      "grad_norm": 1.0705113257051169,
      "learning_rate": 5.308952785194877e-06,
      "loss": 0.6298,
      "step": 9516
    },
    {
      "epoch": 0.513571852571367,
      "grad_norm": 0.9728649254817006,
      "learning_rate": 5.308190582956267e-06,
      "loss": 0.4238,
      "step": 9517
    },
    {
      "epoch": 0.5136258161998812,
      "grad_norm": 0.8395213143387513,
      "learning_rate": 5.307428386230456e-06,
      "loss": 0.3951,
      "step": 9518
    },
    {
      "epoch": 0.5136797798283956,
      "grad_norm": 1.0715402601181856,
      "learning_rate": 5.306666195039346e-06,
      "loss": 0.5053,
      "step": 9519
    },
    {
      "epoch": 0.51373374345691,
      "grad_norm": 0.7847792768446136,
      "learning_rate": 5.305904009404845e-06,
      "loss": 0.3116,
      "step": 9520
    },
    {
      "epoch": 0.5137877070854244,
      "grad_norm": 0.875714052729186,
      "learning_rate": 5.305141829348858e-06,
      "loss": 0.4605,
      "step": 9521
    },
    {
      "epoch": 0.5138416707139388,
      "grad_norm": 1.065208468147302,
      "learning_rate": 5.3043796548932925e-06,
      "loss": 0.5866,
      "step": 9522
    },
    {
      "epoch": 0.5138956343424532,
      "grad_norm": 1.2548103475711776,
      "learning_rate": 5.303617486060053e-06,
      "loss": 0.3948,
      "step": 9523
    },
    {
      "epoch": 0.5139495979709676,
      "grad_norm": 1.262662462725494,
      "learning_rate": 5.3028553228710455e-06,
      "loss": 0.4929,
      "step": 9524
    },
    {
      "epoch": 0.5140035615994819,
      "grad_norm": 1.1166389490465882,
      "learning_rate": 5.302093165348176e-06,
      "loss": 0.6332,
      "step": 9525
    },
    {
      "epoch": 0.5140575252279963,
      "grad_norm": 1.1650809976765597,
      "learning_rate": 5.301331013513349e-06,
      "loss": 0.4387,
      "step": 9526
    },
    {
      "epoch": 0.5141114888565107,
      "grad_norm": 1.0101074439112827,
      "learning_rate": 5.300568867388469e-06,
      "loss": 0.5445,
      "step": 9527
    },
    {
      "epoch": 0.5141654524850251,
      "grad_norm": 0.9128755355254814,
      "learning_rate": 5.299806726995441e-06,
      "loss": 0.4714,
      "step": 9528
    },
    {
      "epoch": 0.5142194161135395,
      "grad_norm": 1.0520169465363594,
      "learning_rate": 5.299044592356169e-06,
      "loss": 0.5564,
      "step": 9529
    },
    {
      "epoch": 0.5142733797420539,
      "grad_norm": 0.9331334447036628,
      "learning_rate": 5.298282463492559e-06,
      "loss": 0.4303,
      "step": 9530
    },
    {
      "epoch": 0.5143273433705683,
      "grad_norm": 1.1891690189439506,
      "learning_rate": 5.297520340426515e-06,
      "loss": 0.5342,
      "step": 9531
    },
    {
      "epoch": 0.5143813069990826,
      "grad_norm": 1.227969558026063,
      "learning_rate": 5.296758223179942e-06,
      "loss": 0.4861,
      "step": 9532
    },
    {
      "epoch": 0.514435270627597,
      "grad_norm": 1.0628138205456097,
      "learning_rate": 5.295996111774743e-06,
      "loss": 0.5621,
      "step": 9533
    },
    {
      "epoch": 0.5144892342561114,
      "grad_norm": 0.8539649117378358,
      "learning_rate": 5.29523400623282e-06,
      "loss": 0.3493,
      "step": 9534
    },
    {
      "epoch": 0.5145431978846258,
      "grad_norm": 0.7919215544234193,
      "learning_rate": 5.29447190657608e-06,
      "loss": 0.3505,
      "step": 9535
    },
    {
      "epoch": 0.5145971615131402,
      "grad_norm": 0.9642900684788295,
      "learning_rate": 5.293709812826424e-06,
      "loss": 0.3663,
      "step": 9536
    },
    {
      "epoch": 0.5146511251416546,
      "grad_norm": 0.8900006720983381,
      "learning_rate": 5.292947725005758e-06,
      "loss": 0.3536,
      "step": 9537
    },
    {
      "epoch": 0.514705088770169,
      "grad_norm": 1.0900926058327856,
      "learning_rate": 5.292185643135983e-06,
      "loss": 0.483,
      "step": 9538
    },
    {
      "epoch": 0.5147590523986832,
      "grad_norm": 0.8644080685866212,
      "learning_rate": 5.291423567239003e-06,
      "loss": 0.3063,
      "step": 9539
    },
    {
      "epoch": 0.5148130160271976,
      "grad_norm": 1.1578621567822183,
      "learning_rate": 5.29066149733672e-06,
      "loss": 0.676,
      "step": 9540
    },
    {
      "epoch": 0.514866979655712,
      "grad_norm": 1.1870323480816298,
      "learning_rate": 5.2898994334510365e-06,
      "loss": 0.5014,
      "step": 9541
    },
    {
      "epoch": 0.5149209432842264,
      "grad_norm": 1.169619843763151,
      "learning_rate": 5.289137375603855e-06,
      "loss": 0.5368,
      "step": 9542
    },
    {
      "epoch": 0.5149749069127408,
      "grad_norm": 1.0494188523794414,
      "learning_rate": 5.28837532381708e-06,
      "loss": 0.5156,
      "step": 9543
    },
    {
      "epoch": 0.5150288705412552,
      "grad_norm": 0.8637406124301432,
      "learning_rate": 5.2876132781126114e-06,
      "loss": 0.3875,
      "step": 9544
    },
    {
      "epoch": 0.5150828341697695,
      "grad_norm": 1.0281518088010726,
      "learning_rate": 5.286851238512353e-06,
      "loss": 0.3397,
      "step": 9545
    },
    {
      "epoch": 0.5151367977982839,
      "grad_norm": 0.8775821082840113,
      "learning_rate": 5.286089205038204e-06,
      "loss": 0.3369,
      "step": 9546
    },
    {
      "epoch": 0.5151907614267983,
      "grad_norm": 1.1620883236728918,
      "learning_rate": 5.285327177712067e-06,
      "loss": 0.4857,
      "step": 9547
    },
    {
      "epoch": 0.5152447250553127,
      "grad_norm": 1.1157724760052157,
      "learning_rate": 5.284565156555845e-06,
      "loss": 0.5505,
      "step": 9548
    },
    {
      "epoch": 0.5152986886838271,
      "grad_norm": 0.8643473621903006,
      "learning_rate": 5.283803141591437e-06,
      "loss": 0.3599,
      "step": 9549
    },
    {
      "epoch": 0.5153526523123415,
      "grad_norm": 1.099209185598772,
      "learning_rate": 5.283041132840745e-06,
      "loss": 0.4954,
      "step": 9550
    },
    {
      "epoch": 0.5154066159408559,
      "grad_norm": 1.164662847270761,
      "learning_rate": 5.28227913032567e-06,
      "loss": 0.5588,
      "step": 9551
    },
    {
      "epoch": 0.5154605795693702,
      "grad_norm": 0.8709885667218966,
      "learning_rate": 5.281517134068115e-06,
      "loss": 0.3723,
      "step": 9552
    },
    {
      "epoch": 0.5155145431978846,
      "grad_norm": 1.1699189129428964,
      "learning_rate": 5.280755144089977e-06,
      "loss": 0.661,
      "step": 9553
    },
    {
      "epoch": 0.515568506826399,
      "grad_norm": 0.9847006379240044,
      "learning_rate": 5.279993160413157e-06,
      "loss": 0.4695,
      "step": 9554
    },
    {
      "epoch": 0.5156224704549134,
      "grad_norm": 0.9497383863625719,
      "learning_rate": 5.279231183059557e-06,
      "loss": 0.4238,
      "step": 9555
    },
    {
      "epoch": 0.5156764340834278,
      "grad_norm": 0.9695758096426631,
      "learning_rate": 5.278469212051074e-06,
      "loss": 0.3756,
      "step": 9556
    },
    {
      "epoch": 0.5157303977119422,
      "grad_norm": 1.0765203072650535,
      "learning_rate": 5.277707247409612e-06,
      "loss": 0.5415,
      "step": 9557
    },
    {
      "epoch": 0.5157843613404566,
      "grad_norm": 1.0372300941708161,
      "learning_rate": 5.276945289157066e-06,
      "loss": 0.4963,
      "step": 9558
    },
    {
      "epoch": 0.5158383249689709,
      "grad_norm": 0.921820973349719,
      "learning_rate": 5.276183337315339e-06,
      "loss": 0.3731,
      "step": 9559
    },
    {
      "epoch": 0.5158922885974853,
      "grad_norm": 0.8242762583385296,
      "learning_rate": 5.275421391906329e-06,
      "loss": 0.4566,
      "step": 9560
    },
    {
      "epoch": 0.5159462522259997,
      "grad_norm": 1.201772590949609,
      "learning_rate": 5.274659452951936e-06,
      "loss": 0.4693,
      "step": 9561
    },
    {
      "epoch": 0.516000215854514,
      "grad_norm": 1.0646673926222765,
      "learning_rate": 5.273897520474056e-06,
      "loss": 0.4803,
      "step": 9562
    },
    {
      "epoch": 0.5160541794830285,
      "grad_norm": 1.1631062746456144,
      "learning_rate": 5.273135594494591e-06,
      "loss": 0.4397,
      "step": 9563
    },
    {
      "epoch": 0.5161081431115428,
      "grad_norm": 1.0313084597967017,
      "learning_rate": 5.272373675035437e-06,
      "loss": 0.5303,
      "step": 9564
    },
    {
      "epoch": 0.5161621067400572,
      "grad_norm": 1.1074489744138354,
      "learning_rate": 5.2716117621184946e-06,
      "loss": 0.7014,
      "step": 9565
    },
    {
      "epoch": 0.5162160703685715,
      "grad_norm": 0.8563501782023145,
      "learning_rate": 5.270849855765663e-06,
      "loss": 0.4292,
      "step": 9566
    },
    {
      "epoch": 0.5162700339970859,
      "grad_norm": 0.8505697005500529,
      "learning_rate": 5.270087955998836e-06,
      "loss": 0.4029,
      "step": 9567
    },
    {
      "epoch": 0.5163239976256003,
      "grad_norm": 0.9458912590907358,
      "learning_rate": 5.269326062839913e-06,
      "loss": 0.3972,
      "step": 9568
    },
    {
      "epoch": 0.5163779612541147,
      "grad_norm": 1.0422540532051232,
      "learning_rate": 5.268564176310793e-06,
      "loss": 0.4776,
      "step": 9569
    },
    {
      "epoch": 0.5164319248826291,
      "grad_norm": 0.8701788992545315,
      "learning_rate": 5.267802296433372e-06,
      "loss": 0.2903,
      "step": 9570
    },
    {
      "epoch": 0.5164858885111435,
      "grad_norm": 1.2606677083653655,
      "learning_rate": 5.267040423229547e-06,
      "loss": 0.5112,
      "step": 9571
    },
    {
      "epoch": 0.5165398521396579,
      "grad_norm": 1.1168572702493915,
      "learning_rate": 5.266278556721216e-06,
      "loss": 0.4891,
      "step": 9572
    },
    {
      "epoch": 0.5165938157681722,
      "grad_norm": 1.0315528970050492,
      "learning_rate": 5.265516696930277e-06,
      "loss": 0.5358,
      "step": 9573
    },
    {
      "epoch": 0.5166477793966866,
      "grad_norm": 1.0307100551744068,
      "learning_rate": 5.2647548438786245e-06,
      "loss": 0.559,
      "step": 9574
    },
    {
      "epoch": 0.516701743025201,
      "grad_norm": 1.1524939902816118,
      "learning_rate": 5.263992997588154e-06,
      "loss": 0.6051,
      "step": 9575
    },
    {
      "epoch": 0.5167557066537154,
      "grad_norm": 1.0103818164779266,
      "learning_rate": 5.263231158080764e-06,
      "loss": 0.4504,
      "step": 9576
    },
    {
      "epoch": 0.5168096702822298,
      "grad_norm": 1.0147891319081173,
      "learning_rate": 5.262469325378349e-06,
      "loss": 0.4511,
      "step": 9577
    },
    {
      "epoch": 0.5168636339107442,
      "grad_norm": 1.1409476991649232,
      "learning_rate": 5.2617074995028065e-06,
      "loss": 0.5462,
      "step": 9578
    },
    {
      "epoch": 0.5169175975392586,
      "grad_norm": 1.0825736282629932,
      "learning_rate": 5.26094568047603e-06,
      "loss": 0.5441,
      "step": 9579
    },
    {
      "epoch": 0.5169715611677729,
      "grad_norm": 0.899972200415747,
      "learning_rate": 5.260183868319919e-06,
      "loss": 0.4443,
      "step": 9580
    },
    {
      "epoch": 0.5170255247962873,
      "grad_norm": 0.8737871669079942,
      "learning_rate": 5.259422063056364e-06,
      "loss": 0.3562,
      "step": 9581
    },
    {
      "epoch": 0.5170794884248017,
      "grad_norm": 0.9721542723808108,
      "learning_rate": 5.258660264707262e-06,
      "loss": 0.4826,
      "step": 9582
    },
    {
      "epoch": 0.5171334520533161,
      "grad_norm": 0.8235930607982394,
      "learning_rate": 5.257898473294508e-06,
      "loss": 0.2592,
      "step": 9583
    },
    {
      "epoch": 0.5171874156818305,
      "grad_norm": 1.0629933665998887,
      "learning_rate": 5.257136688839997e-06,
      "loss": 0.5594,
      "step": 9584
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.8996526125580303,
      "learning_rate": 5.256374911365621e-06,
      "loss": 0.3901,
      "step": 9585
    },
    {
      "epoch": 0.5172953429388593,
      "grad_norm": 0.8905619906635535,
      "learning_rate": 5.255613140893278e-06,
      "loss": 0.451,
      "step": 9586
    },
    {
      "epoch": 0.5173493065673735,
      "grad_norm": 1.2150883996855757,
      "learning_rate": 5.25485137744486e-06,
      "loss": 0.521,
      "step": 9587
    },
    {
      "epoch": 0.5174032701958879,
      "grad_norm": 0.9989084333432823,
      "learning_rate": 5.2540896210422625e-06,
      "loss": 0.4543,
      "step": 9588
    },
    {
      "epoch": 0.5174572338244023,
      "grad_norm": 0.8805579091778825,
      "learning_rate": 5.253327871707375e-06,
      "loss": 0.4405,
      "step": 9589
    },
    {
      "epoch": 0.5175111974529167,
      "grad_norm": 0.8715443077854018,
      "learning_rate": 5.252566129462095e-06,
      "loss": 0.3297,
      "step": 9590
    },
    {
      "epoch": 0.5175651610814311,
      "grad_norm": 0.8712234379622795,
      "learning_rate": 5.251804394328315e-06,
      "loss": 0.4039,
      "step": 9591
    },
    {
      "epoch": 0.5176191247099455,
      "grad_norm": 1.1152068881845316,
      "learning_rate": 5.251042666327926e-06,
      "loss": 0.5832,
      "step": 9592
    },
    {
      "epoch": 0.5176730883384599,
      "grad_norm": 1.056480608975123,
      "learning_rate": 5.250280945482823e-06,
      "loss": 0.553,
      "step": 9593
    },
    {
      "epoch": 0.5177270519669742,
      "grad_norm": 0.9371457814465499,
      "learning_rate": 5.2495192318149e-06,
      "loss": 0.3972,
      "step": 9594
    },
    {
      "epoch": 0.5177810155954886,
      "grad_norm": 1.114822068390724,
      "learning_rate": 5.2487575253460464e-06,
      "loss": 0.6291,
      "step": 9595
    },
    {
      "epoch": 0.517834979224003,
      "grad_norm": 1.2113965559819555,
      "learning_rate": 5.247995826098154e-06,
      "loss": 0.7008,
      "step": 9596
    },
    {
      "epoch": 0.5178889428525174,
      "grad_norm": 0.8392142517577571,
      "learning_rate": 5.247234134093117e-06,
      "loss": 0.3539,
      "step": 9597
    },
    {
      "epoch": 0.5179429064810318,
      "grad_norm": 0.9314994870542145,
      "learning_rate": 5.246472449352825e-06,
      "loss": 0.3526,
      "step": 9598
    },
    {
      "epoch": 0.5179968701095462,
      "grad_norm": 1.0381821101085098,
      "learning_rate": 5.245710771899173e-06,
      "loss": 0.4492,
      "step": 9599
    },
    {
      "epoch": 0.5180508337380606,
      "grad_norm": 1.0942671314693733,
      "learning_rate": 5.244949101754049e-06,
      "loss": 0.5046,
      "step": 9600
    },
    {
      "epoch": 0.5181047973665749,
      "grad_norm": 0.9804402824645503,
      "learning_rate": 5.244187438939348e-06,
      "loss": 0.4116,
      "step": 9601
    },
    {
      "epoch": 0.5181587609950893,
      "grad_norm": 1.1841844825590797,
      "learning_rate": 5.2434257834769555e-06,
      "loss": 0.4875,
      "step": 9602
    },
    {
      "epoch": 0.5182127246236037,
      "grad_norm": 1.1257837833422584,
      "learning_rate": 5.242664135388767e-06,
      "loss": 0.4932,
      "step": 9603
    },
    {
      "epoch": 0.5182666882521181,
      "grad_norm": 1.136020172781561,
      "learning_rate": 5.24190249469667e-06,
      "loss": 0.6086,
      "step": 9604
    },
    {
      "epoch": 0.5183206518806325,
      "grad_norm": 0.8686481367272051,
      "learning_rate": 5.2411408614225555e-06,
      "loss": 0.4531,
      "step": 9605
    },
    {
      "epoch": 0.5183746155091469,
      "grad_norm": 0.9211521146023607,
      "learning_rate": 5.240379235588315e-06,
      "loss": 0.4132,
      "step": 9606
    },
    {
      "epoch": 0.5184285791376613,
      "grad_norm": 0.9926402972011605,
      "learning_rate": 5.239617617215839e-06,
      "loss": 0.4225,
      "step": 9607
    },
    {
      "epoch": 0.5184825427661756,
      "grad_norm": 1.00860105509254,
      "learning_rate": 5.238856006327015e-06,
      "loss": 0.4877,
      "step": 9608
    },
    {
      "epoch": 0.51853650639469,
      "grad_norm": 0.9283166513551118,
      "learning_rate": 5.238094402943732e-06,
      "loss": 0.5285,
      "step": 9609
    },
    {
      "epoch": 0.5185904700232044,
      "grad_norm": 0.9518125216308094,
      "learning_rate": 5.237332807087881e-06,
      "loss": 0.4046,
      "step": 9610
    },
    {
      "epoch": 0.5186444336517188,
      "grad_norm": 0.796793160783964,
      "learning_rate": 5.236571218781351e-06,
      "loss": 0.3363,
      "step": 9611
    },
    {
      "epoch": 0.5186983972802331,
      "grad_norm": 0.927304765102309,
      "learning_rate": 5.235809638046032e-06,
      "loss": 0.4423,
      "step": 9612
    },
    {
      "epoch": 0.5187523609087475,
      "grad_norm": 0.8501849320155819,
      "learning_rate": 5.2350480649038084e-06,
      "loss": 0.3526,
      "step": 9613
    },
    {
      "epoch": 0.5188063245372618,
      "grad_norm": 1.0280089690016438,
      "learning_rate": 5.234286499376573e-06,
      "loss": 0.4963,
      "step": 9614
    },
    {
      "epoch": 0.5188602881657762,
      "grad_norm": 1.0473035276511051,
      "learning_rate": 5.233524941486211e-06,
      "loss": 0.5226,
      "step": 9615
    },
    {
      "epoch": 0.5189142517942906,
      "grad_norm": 1.1359285623169324,
      "learning_rate": 5.2327633912546105e-06,
      "loss": 0.4455,
      "step": 9616
    },
    {
      "epoch": 0.518968215422805,
      "grad_norm": 0.9654088114948347,
      "learning_rate": 5.232001848703662e-06,
      "loss": 0.4283,
      "step": 9617
    },
    {
      "epoch": 0.5190221790513194,
      "grad_norm": 1.1705489238758828,
      "learning_rate": 5.231240313855249e-06,
      "loss": 0.6863,
      "step": 9618
    },
    {
      "epoch": 0.5190761426798338,
      "grad_norm": 0.7650585532366093,
      "learning_rate": 5.230478786731262e-06,
      "loss": 0.2856,
      "step": 9619
    },
    {
      "epoch": 0.5191301063083482,
      "grad_norm": 1.2775566433219763,
      "learning_rate": 5.229717267353588e-06,
      "loss": 0.614,
      "step": 9620
    },
    {
      "epoch": 0.5191840699368625,
      "grad_norm": 0.9309559301011927,
      "learning_rate": 5.228955755744114e-06,
      "loss": 0.4593,
      "step": 9621
    },
    {
      "epoch": 0.5192380335653769,
      "grad_norm": 0.9348982285625476,
      "learning_rate": 5.2281942519247234e-06,
      "loss": 0.482,
      "step": 9622
    },
    {
      "epoch": 0.5192919971938913,
      "grad_norm": 0.9444850887417824,
      "learning_rate": 5.227432755917304e-06,
      "loss": 0.4012,
      "step": 9623
    },
    {
      "epoch": 0.5193459608224057,
      "grad_norm": 0.9874277516399572,
      "learning_rate": 5.2266712677437436e-06,
      "loss": 0.5133,
      "step": 9624
    },
    {
      "epoch": 0.5193999244509201,
      "grad_norm": 0.8525642564249177,
      "learning_rate": 5.2259097874259265e-06,
      "loss": 0.4084,
      "step": 9625
    },
    {
      "epoch": 0.5194538880794345,
      "grad_norm": 0.9824873847583103,
      "learning_rate": 5.2251483149857396e-06,
      "loss": 0.585,
      "step": 9626
    },
    {
      "epoch": 0.5195078517079489,
      "grad_norm": 0.9556821911991316,
      "learning_rate": 5.2243868504450665e-06,
      "loss": 0.4259,
      "step": 9627
    },
    {
      "epoch": 0.5195618153364632,
      "grad_norm": 1.090584512215743,
      "learning_rate": 5.223625393825797e-06,
      "loss": 0.5117,
      "step": 9628
    },
    {
      "epoch": 0.5196157789649776,
      "grad_norm": 1.3172272869469277,
      "learning_rate": 5.222863945149811e-06,
      "loss": 0.5551,
      "step": 9629
    },
    {
      "epoch": 0.519669742593492,
      "grad_norm": 1.265358764081254,
      "learning_rate": 5.222102504438995e-06,
      "loss": 0.58,
      "step": 9630
    },
    {
      "epoch": 0.5197237062220064,
      "grad_norm": 1.0239914984229466,
      "learning_rate": 5.2213410717152345e-06,
      "loss": 0.4849,
      "step": 9631
    },
    {
      "epoch": 0.5197776698505208,
      "grad_norm": 0.9013851257388786,
      "learning_rate": 5.2205796470004125e-06,
      "loss": 0.3961,
      "step": 9632
    },
    {
      "epoch": 0.5198316334790352,
      "grad_norm": 0.9196823912790001,
      "learning_rate": 5.219818230316413e-06,
      "loss": 0.3843,
      "step": 9633
    },
    {
      "epoch": 0.5198855971075496,
      "grad_norm": 1.1576426952802408,
      "learning_rate": 5.219056821685122e-06,
      "loss": 0.544,
      "step": 9634
    },
    {
      "epoch": 0.5199395607360638,
      "grad_norm": 1.1172076297874685,
      "learning_rate": 5.218295421128422e-06,
      "loss": 0.4564,
      "step": 9635
    },
    {
      "epoch": 0.5199935243645782,
      "grad_norm": 0.9876044477715409,
      "learning_rate": 5.217534028668195e-06,
      "loss": 0.4979,
      "step": 9636
    },
    {
      "epoch": 0.5200474879930926,
      "grad_norm": 1.0034846833950475,
      "learning_rate": 5.216772644326326e-06,
      "loss": 0.4413,
      "step": 9637
    },
    {
      "epoch": 0.520101451621607,
      "grad_norm": 1.0444907992183496,
      "learning_rate": 5.216011268124698e-06,
      "loss": 0.5136,
      "step": 9638
    },
    {
      "epoch": 0.5201554152501214,
      "grad_norm": 1.1570987083669353,
      "learning_rate": 5.215249900085192e-06,
      "loss": 0.5991,
      "step": 9639
    },
    {
      "epoch": 0.5202093788786358,
      "grad_norm": 0.8909997092140797,
      "learning_rate": 5.214488540229693e-06,
      "loss": 0.3892,
      "step": 9640
    },
    {
      "epoch": 0.5202633425071502,
      "grad_norm": 0.7484242538203026,
      "learning_rate": 5.21372718858008e-06,
      "loss": 0.3142,
      "step": 9641
    },
    {
      "epoch": 0.5203173061356645,
      "grad_norm": 1.0435869672094704,
      "learning_rate": 5.21296584515824e-06,
      "loss": 0.4225,
      "step": 9642
    },
    {
      "epoch": 0.5203712697641789,
      "grad_norm": 0.9497086924896089,
      "learning_rate": 5.21220450998605e-06,
      "loss": 0.4226,
      "step": 9643
    },
    {
      "epoch": 0.5204252333926933,
      "grad_norm": 0.9809168577951985,
      "learning_rate": 5.211443183085392e-06,
      "loss": 0.5073,
      "step": 9644
    },
    {
      "epoch": 0.5204791970212077,
      "grad_norm": 0.805141581807964,
      "learning_rate": 5.210681864478149e-06,
      "loss": 0.3329,
      "step": 9645
    },
    {
      "epoch": 0.5205331606497221,
      "grad_norm": 1.0790794260263412,
      "learning_rate": 5.2099205541862e-06,
      "loss": 0.5211,
      "step": 9646
    },
    {
      "epoch": 0.5205871242782365,
      "grad_norm": 0.8919513825787239,
      "learning_rate": 5.209159252231431e-06,
      "loss": 0.3996,
      "step": 9647
    },
    {
      "epoch": 0.5206410879067509,
      "grad_norm": 0.8367033247102641,
      "learning_rate": 5.2083979586357156e-06,
      "loss": 0.3125,
      "step": 9648
    },
    {
      "epoch": 0.5206950515352652,
      "grad_norm": 1.028778199112551,
      "learning_rate": 5.207636673420942e-06,
      "loss": 0.4538,
      "step": 9649
    },
    {
      "epoch": 0.5207490151637796,
      "grad_norm": 1.186487267235031,
      "learning_rate": 5.206875396608981e-06,
      "loss": 0.5198,
      "step": 9650
    },
    {
      "epoch": 0.520802978792294,
      "grad_norm": 1.1448527637653159,
      "learning_rate": 5.20611412822172e-06,
      "loss": 0.4441,
      "step": 9651
    },
    {
      "epoch": 0.5208569424208084,
      "grad_norm": 1.1945708064507647,
      "learning_rate": 5.2053528682810335e-06,
      "loss": 0.4916,
      "step": 9652
    },
    {
      "epoch": 0.5209109060493228,
      "grad_norm": 1.3514620088073783,
      "learning_rate": 5.204591616808805e-06,
      "loss": 0.6016,
      "step": 9653
    },
    {
      "epoch": 0.5209648696778372,
      "grad_norm": 0.7900433796031581,
      "learning_rate": 5.2038303738269125e-06,
      "loss": 0.406,
      "step": 9654
    },
    {
      "epoch": 0.5210188333063516,
      "grad_norm": 0.9605800611054289,
      "learning_rate": 5.203069139357233e-06,
      "loss": 0.5516,
      "step": 9655
    },
    {
      "epoch": 0.5210727969348659,
      "grad_norm": 1.0022009859103247,
      "learning_rate": 5.202307913421649e-06,
      "loss": 0.4124,
      "step": 9656
    },
    {
      "epoch": 0.5211267605633803,
      "grad_norm": 0.8831524478882585,
      "learning_rate": 5.201546696042033e-06,
      "loss": 0.3616,
      "step": 9657
    },
    {
      "epoch": 0.5211807241918947,
      "grad_norm": 0.9199655354610686,
      "learning_rate": 5.200785487240268e-06,
      "loss": 0.4455,
      "step": 9658
    },
    {
      "epoch": 0.521234687820409,
      "grad_norm": 1.0644316527177442,
      "learning_rate": 5.2000242870382316e-06,
      "loss": 0.5296,
      "step": 9659
    },
    {
      "epoch": 0.5212886514489234,
      "grad_norm": 0.8342310291393389,
      "learning_rate": 5.1992630954577984e-06,
      "loss": 0.3433,
      "step": 9660
    },
    {
      "epoch": 0.5213426150774378,
      "grad_norm": 1.0519974042422013,
      "learning_rate": 5.198501912520848e-06,
      "loss": 0.4637,
      "step": 9661
    },
    {
      "epoch": 0.5213965787059522,
      "grad_norm": 0.9568982919293628,
      "learning_rate": 5.197740738249258e-06,
      "loss": 0.4854,
      "step": 9662
    },
    {
      "epoch": 0.5214505423344665,
      "grad_norm": 0.9744712278976847,
      "learning_rate": 5.196979572664904e-06,
      "loss": 0.4107,
      "step": 9663
    },
    {
      "epoch": 0.5215045059629809,
      "grad_norm": 1.1361613026257185,
      "learning_rate": 5.196218415789663e-06,
      "loss": 0.6605,
      "step": 9664
    },
    {
      "epoch": 0.5215584695914953,
      "grad_norm": 0.9598370269010816,
      "learning_rate": 5.195457267645412e-06,
      "loss": 0.5433,
      "step": 9665
    },
    {
      "epoch": 0.5216124332200097,
      "grad_norm": 1.2191054239814343,
      "learning_rate": 5.194696128254028e-06,
      "loss": 0.5207,
      "step": 9666
    },
    {
      "epoch": 0.5216663968485241,
      "grad_norm": 1.1836060091430065,
      "learning_rate": 5.193934997637383e-06,
      "loss": 0.4466,
      "step": 9667
    },
    {
      "epoch": 0.5217203604770385,
      "grad_norm": 1.0235342200021762,
      "learning_rate": 5.193173875817358e-06,
      "loss": 0.4278,
      "step": 9668
    },
    {
      "epoch": 0.5217743241055529,
      "grad_norm": 1.0416863034567074,
      "learning_rate": 5.192412762815824e-06,
      "loss": 0.551,
      "step": 9669
    },
    {
      "epoch": 0.5218282877340672,
      "grad_norm": 0.9843409050045359,
      "learning_rate": 5.19165165865466e-06,
      "loss": 0.4125,
      "step": 9670
    },
    {
      "epoch": 0.5218822513625816,
      "grad_norm": 0.9027373064956576,
      "learning_rate": 5.190890563355738e-06,
      "loss": 0.3863,
      "step": 9671
    },
    {
      "epoch": 0.521936214991096,
      "grad_norm": 0.9916368384077192,
      "learning_rate": 5.190129476940934e-06,
      "loss": 0.6031,
      "step": 9672
    },
    {
      "epoch": 0.5219901786196104,
      "grad_norm": 1.152862172148206,
      "learning_rate": 5.189368399432121e-06,
      "loss": 0.6138,
      "step": 9673
    },
    {
      "epoch": 0.5220441422481248,
      "grad_norm": 1.0157933030915443,
      "learning_rate": 5.188607330851174e-06,
      "loss": 0.4258,
      "step": 9674
    },
    {
      "epoch": 0.5220981058766392,
      "grad_norm": 1.0182717921002815,
      "learning_rate": 5.187846271219967e-06,
      "loss": 0.3822,
      "step": 9675
    },
    {
      "epoch": 0.5221520695051535,
      "grad_norm": 1.2205893676101491,
      "learning_rate": 5.187085220560376e-06,
      "loss": 0.6016,
      "step": 9676
    },
    {
      "epoch": 0.5222060331336679,
      "grad_norm": 1.03013158869447,
      "learning_rate": 5.1863241788942695e-06,
      "loss": 0.462,
      "step": 9677
    },
    {
      "epoch": 0.5222599967621823,
      "grad_norm": 1.0468405134984857,
      "learning_rate": 5.185563146243523e-06,
      "loss": 0.518,
      "step": 9678
    },
    {
      "epoch": 0.5223139603906967,
      "grad_norm": 1.0541604364729629,
      "learning_rate": 5.184802122630009e-06,
      "loss": 0.6002,
      "step": 9679
    },
    {
      "epoch": 0.5223679240192111,
      "grad_norm": 1.0861210682141307,
      "learning_rate": 5.184041108075602e-06,
      "loss": 0.4807,
      "step": 9680
    },
    {
      "epoch": 0.5224218876477255,
      "grad_norm": 1.0760653494054,
      "learning_rate": 5.183280102602172e-06,
      "loss": 0.6313,
      "step": 9681
    },
    {
      "epoch": 0.5224758512762399,
      "grad_norm": 1.161952978821668,
      "learning_rate": 5.182519106231592e-06,
      "loss": 0.4907,
      "step": 9682
    },
    {
      "epoch": 0.5225298149047541,
      "grad_norm": 0.9485251956637524,
      "learning_rate": 5.181758118985735e-06,
      "loss": 0.4728,
      "step": 9683
    },
    {
      "epoch": 0.5225837785332685,
      "grad_norm": 0.9870468666583867,
      "learning_rate": 5.180997140886469e-06,
      "loss": 0.4828,
      "step": 9684
    },
    {
      "epoch": 0.5226377421617829,
      "grad_norm": 1.0064886256724106,
      "learning_rate": 5.180236171955668e-06,
      "loss": 0.4354,
      "step": 9685
    },
    {
      "epoch": 0.5226917057902973,
      "grad_norm": 0.9184619198469036,
      "learning_rate": 5.1794752122152035e-06,
      "loss": 0.4824,
      "step": 9686
    },
    {
      "epoch": 0.5227456694188117,
      "grad_norm": 1.1486371908113884,
      "learning_rate": 5.178714261686945e-06,
      "loss": 0.4845,
      "step": 9687
    },
    {
      "epoch": 0.5227996330473261,
      "grad_norm": 1.17066951596827,
      "learning_rate": 5.177953320392763e-06,
      "loss": 0.5198,
      "step": 9688
    },
    {
      "epoch": 0.5228535966758405,
      "grad_norm": 0.8143927021320968,
      "learning_rate": 5.177192388354527e-06,
      "loss": 0.3615,
      "step": 9689
    },
    {
      "epoch": 0.5229075603043548,
      "grad_norm": 1.1883570313485465,
      "learning_rate": 5.176431465594109e-06,
      "loss": 0.5521,
      "step": 9690
    },
    {
      "epoch": 0.5229615239328692,
      "grad_norm": 0.8149011046811768,
      "learning_rate": 5.175670552133378e-06,
      "loss": 0.4208,
      "step": 9691
    },
    {
      "epoch": 0.5230154875613836,
      "grad_norm": 1.0687523226023858,
      "learning_rate": 5.174909647994202e-06,
      "loss": 0.5109,
      "step": 9692
    },
    {
      "epoch": 0.523069451189898,
      "grad_norm": 1.1371724134509407,
      "learning_rate": 5.174148753198452e-06,
      "loss": 0.4535,
      "step": 9693
    },
    {
      "epoch": 0.5231234148184124,
      "grad_norm": 1.1033189258389058,
      "learning_rate": 5.173387867767995e-06,
      "loss": 0.5574,
      "step": 9694
    },
    {
      "epoch": 0.5231773784469268,
      "grad_norm": 0.9905204814640702,
      "learning_rate": 5.1726269917247015e-06,
      "loss": 0.3895,
      "step": 9695
    },
    {
      "epoch": 0.5232313420754412,
      "grad_norm": 0.7891119086685983,
      "learning_rate": 5.171866125090439e-06,
      "loss": 0.329,
      "step": 9696
    },
    {
      "epoch": 0.5232853057039555,
      "grad_norm": 1.0163090467758964,
      "learning_rate": 5.171105267887077e-06,
      "loss": 0.4421,
      "step": 9697
    },
    {
      "epoch": 0.5233392693324699,
      "grad_norm": 0.9143440246336719,
      "learning_rate": 5.170344420136481e-06,
      "loss": 0.3859,
      "step": 9698
    },
    {
      "epoch": 0.5233932329609843,
      "grad_norm": 0.9692121901936291,
      "learning_rate": 5.169583581860519e-06,
      "loss": 0.6322,
      "step": 9699
    },
    {
      "epoch": 0.5234471965894987,
      "grad_norm": 0.6842032840651432,
      "learning_rate": 5.168822753081058e-06,
      "loss": 0.3103,
      "step": 9700
    },
    {
      "epoch": 0.5235011602180131,
      "grad_norm": 0.9378063048005031,
      "learning_rate": 5.168061933819966e-06,
      "loss": 0.4271,
      "step": 9701
    },
    {
      "epoch": 0.5235551238465275,
      "grad_norm": 1.0662412278005384,
      "learning_rate": 5.16730112409911e-06,
      "loss": 0.5343,
      "step": 9702
    },
    {
      "epoch": 0.5236090874750419,
      "grad_norm": 0.967490425313805,
      "learning_rate": 5.166540323940356e-06,
      "loss": 0.4082,
      "step": 9703
    },
    {
      "epoch": 0.5236630511035562,
      "grad_norm": 0.9505329831695838,
      "learning_rate": 5.16577953336557e-06,
      "loss": 0.4676,
      "step": 9704
    },
    {
      "epoch": 0.5237170147320706,
      "grad_norm": 0.9684178693793721,
      "learning_rate": 5.165018752396618e-06,
      "loss": 0.3817,
      "step": 9705
    },
    {
      "epoch": 0.523770978360585,
      "grad_norm": 1.1749223918869116,
      "learning_rate": 5.164257981055365e-06,
      "loss": 0.6451,
      "step": 9706
    },
    {
      "epoch": 0.5238249419890993,
      "grad_norm": 1.034066185665713,
      "learning_rate": 5.163497219363677e-06,
      "loss": 0.426,
      "step": 9707
    },
    {
      "epoch": 0.5238789056176137,
      "grad_norm": 0.9465132270021277,
      "learning_rate": 5.1627364673434185e-06,
      "loss": 0.4616,
      "step": 9708
    },
    {
      "epoch": 0.5239328692461281,
      "grad_norm": 1.153542668255986,
      "learning_rate": 5.161975725016455e-06,
      "loss": 0.4784,
      "step": 9709
    },
    {
      "epoch": 0.5239868328746425,
      "grad_norm": 1.156759960239962,
      "learning_rate": 5.1612149924046505e-06,
      "loss": 0.5767,
      "step": 9710
    },
    {
      "epoch": 0.5240407965031568,
      "grad_norm": 0.8827901210162064,
      "learning_rate": 5.160454269529871e-06,
      "loss": 0.3561,
      "step": 9711
    },
    {
      "epoch": 0.5240947601316712,
      "grad_norm": 0.9261287038070382,
      "learning_rate": 5.159693556413977e-06,
      "loss": 0.5562,
      "step": 9712
    },
    {
      "epoch": 0.5241487237601856,
      "grad_norm": 1.0377818799449898,
      "learning_rate": 5.158932853078835e-06,
      "loss": 0.4596,
      "step": 9713
    },
    {
      "epoch": 0.5242026873887,
      "grad_norm": 1.032974823125582,
      "learning_rate": 5.158172159546308e-06,
      "loss": 0.4947,
      "step": 9714
    },
    {
      "epoch": 0.5242566510172144,
      "grad_norm": 1.050730777489599,
      "learning_rate": 5.157411475838256e-06,
      "loss": 0.5466,
      "step": 9715
    },
    {
      "epoch": 0.5243106146457288,
      "grad_norm": 1.267960430937756,
      "learning_rate": 5.1566508019765475e-06,
      "loss": 0.5167,
      "step": 9716
    },
    {
      "epoch": 0.5243645782742432,
      "grad_norm": 0.9730603581028664,
      "learning_rate": 5.15589013798304e-06,
      "loss": 0.5152,
      "step": 9717
    },
    {
      "epoch": 0.5244185419027575,
      "grad_norm": 1.1921470184726675,
      "learning_rate": 5.155129483879599e-06,
      "loss": 0.5117,
      "step": 9718
    },
    {
      "epoch": 0.5244725055312719,
      "grad_norm": 0.9014781575035112,
      "learning_rate": 5.154368839688084e-06,
      "loss": 0.4296,
      "step": 9719
    },
    {
      "epoch": 0.5245264691597863,
      "grad_norm": 1.0229823205713489,
      "learning_rate": 5.153608205430357e-06,
      "loss": 0.4074,
      "step": 9720
    },
    {
      "epoch": 0.5245804327883007,
      "grad_norm": 0.9435346840281509,
      "learning_rate": 5.152847581128282e-06,
      "loss": 0.4303,
      "step": 9721
    },
    {
      "epoch": 0.5246343964168151,
      "grad_norm": 0.9785791051136097,
      "learning_rate": 5.152086966803716e-06,
      "loss": 0.4187,
      "step": 9722
    },
    {
      "epoch": 0.5246883600453295,
      "grad_norm": 1.1992421363377896,
      "learning_rate": 5.151326362478525e-06,
      "loss": 0.6337,
      "step": 9723
    },
    {
      "epoch": 0.5247423236738439,
      "grad_norm": 1.150152440823792,
      "learning_rate": 5.150565768174565e-06,
      "loss": 0.456,
      "step": 9724
    },
    {
      "epoch": 0.5247962873023582,
      "grad_norm": 1.137896889556343,
      "learning_rate": 5.1498051839137e-06,
      "loss": 0.4635,
      "step": 9725
    },
    {
      "epoch": 0.5248502509308726,
      "grad_norm": 0.9788570605792491,
      "learning_rate": 5.149044609717785e-06,
      "loss": 0.5053,
      "step": 9726
    },
    {
      "epoch": 0.524904214559387,
      "grad_norm": 0.9961449611766471,
      "learning_rate": 5.148284045608684e-06,
      "loss": 0.4729,
      "step": 9727
    },
    {
      "epoch": 0.5249581781879014,
      "grad_norm": 1.0248468072717873,
      "learning_rate": 5.147523491608253e-06,
      "loss": 0.4127,
      "step": 9728
    },
    {
      "epoch": 0.5250121418164158,
      "grad_norm": 0.9235939572103329,
      "learning_rate": 5.146762947738354e-06,
      "loss": 0.4414,
      "step": 9729
    },
    {
      "epoch": 0.5250661054449302,
      "grad_norm": 1.1854696067097619,
      "learning_rate": 5.1460024140208455e-06,
      "loss": 0.5815,
      "step": 9730
    },
    {
      "epoch": 0.5251200690734446,
      "grad_norm": 1.1380335324446935,
      "learning_rate": 5.145241890477585e-06,
      "loss": 0.5735,
      "step": 9731
    },
    {
      "epoch": 0.5251740327019588,
      "grad_norm": 0.96580988556463,
      "learning_rate": 5.144481377130432e-06,
      "loss": 0.4909,
      "step": 9732
    },
    {
      "epoch": 0.5252279963304732,
      "grad_norm": 0.9923326802890604,
      "learning_rate": 5.143720874001243e-06,
      "loss": 0.3871,
      "step": 9733
    },
    {
      "epoch": 0.5252819599589876,
      "grad_norm": 1.056261914166459,
      "learning_rate": 5.142960381111875e-06,
      "loss": 0.4823,
      "step": 9734
    },
    {
      "epoch": 0.525335923587502,
      "grad_norm": 0.9262506560552561,
      "learning_rate": 5.142199898484187e-06,
      "loss": 0.3555,
      "step": 9735
    },
    {
      "epoch": 0.5253898872160164,
      "grad_norm": 1.1033609192732796,
      "learning_rate": 5.141439426140035e-06,
      "loss": 0.5104,
      "step": 9736
    },
    {
      "epoch": 0.5254438508445308,
      "grad_norm": 0.8725494373337509,
      "learning_rate": 5.1406789641012764e-06,
      "loss": 0.3962,
      "step": 9737
    },
    {
      "epoch": 0.5254978144730452,
      "grad_norm": 0.8399309984908676,
      "learning_rate": 5.139918512389768e-06,
      "loss": 0.2853,
      "step": 9738
    },
    {
      "epoch": 0.5255517781015595,
      "grad_norm": 1.3021287229923415,
      "learning_rate": 5.139158071027366e-06,
      "loss": 0.585,
      "step": 9739
    },
    {
      "epoch": 0.5256057417300739,
      "grad_norm": 1.116731359758357,
      "learning_rate": 5.138397640035924e-06,
      "loss": 0.536,
      "step": 9740
    },
    {
      "epoch": 0.5256597053585883,
      "grad_norm": 0.9405679181978963,
      "learning_rate": 5.1376372194372996e-06,
      "loss": 0.5167,
      "step": 9741
    },
    {
      "epoch": 0.5257136689871027,
      "grad_norm": 1.0816966282508211,
      "learning_rate": 5.136876809253348e-06,
      "loss": 0.5286,
      "step": 9742
    },
    {
      "epoch": 0.5257676326156171,
      "grad_norm": 0.9043627126847292,
      "learning_rate": 5.136116409505924e-06,
      "loss": 0.3525,
      "step": 9743
    },
    {
      "epoch": 0.5258215962441315,
      "grad_norm": 1.0831634123468201,
      "learning_rate": 5.13535602021688e-06,
      "loss": 0.4502,
      "step": 9744
    },
    {
      "epoch": 0.5258755598726458,
      "grad_norm": 1.00175676361393,
      "learning_rate": 5.1345956414080764e-06,
      "loss": 0.4922,
      "step": 9745
    },
    {
      "epoch": 0.5259295235011602,
      "grad_norm": 1.2639944575337376,
      "learning_rate": 5.133835273101362e-06,
      "loss": 0.5872,
      "step": 9746
    },
    {
      "epoch": 0.5259834871296746,
      "grad_norm": 0.8658766603647664,
      "learning_rate": 5.13307491531859e-06,
      "loss": 0.3746,
      "step": 9747
    },
    {
      "epoch": 0.526037450758189,
      "grad_norm": 0.9334430512708828,
      "learning_rate": 5.132314568081617e-06,
      "loss": 0.3774,
      "step": 9748
    },
    {
      "epoch": 0.5260914143867034,
      "grad_norm": 1.029208643965735,
      "learning_rate": 5.131554231412293e-06,
      "loss": 0.4748,
      "step": 9749
    },
    {
      "epoch": 0.5261453780152178,
      "grad_norm": 1.2091086446753636,
      "learning_rate": 5.130793905332475e-06,
      "loss": 0.4154,
      "step": 9750
    },
    {
      "epoch": 0.5261993416437322,
      "grad_norm": 0.9722527354998728,
      "learning_rate": 5.1300335898640115e-06,
      "loss": 0.3803,
      "step": 9751
    },
    {
      "epoch": 0.5262533052722465,
      "grad_norm": 0.8565574203080233,
      "learning_rate": 5.12927328502876e-06,
      "loss": 0.4375,
      "step": 9752
    },
    {
      "epoch": 0.5263072689007608,
      "grad_norm": 1.0021909028308913,
      "learning_rate": 5.128512990848565e-06,
      "loss": 0.4623,
      "step": 9753
    },
    {
      "epoch": 0.5263612325292752,
      "grad_norm": 1.0879252995567534,
      "learning_rate": 5.127752707345284e-06,
      "loss": 0.5057,
      "step": 9754
    },
    {
      "epoch": 0.5264151961577896,
      "grad_norm": 1.0638904576507002,
      "learning_rate": 5.126992434540764e-06,
      "loss": 0.5198,
      "step": 9755
    },
    {
      "epoch": 0.526469159786304,
      "grad_norm": 1.0222525476895574,
      "learning_rate": 5.126232172456861e-06,
      "loss": 0.4152,
      "step": 9756
    },
    {
      "epoch": 0.5265231234148184,
      "grad_norm": 1.276592011051086,
      "learning_rate": 5.1254719211154205e-06,
      "loss": 0.5514,
      "step": 9757
    },
    {
      "epoch": 0.5265770870433328,
      "grad_norm": 1.0060512533536143,
      "learning_rate": 5.1247116805382965e-06,
      "loss": 0.5332,
      "step": 9758
    },
    {
      "epoch": 0.5266310506718471,
      "grad_norm": 1.2220599719406897,
      "learning_rate": 5.12395145074734e-06,
      "loss": 0.3746,
      "step": 9759
    },
    {
      "epoch": 0.5266850143003615,
      "grad_norm": 0.9574842687567705,
      "learning_rate": 5.123191231764396e-06,
      "loss": 0.4802,
      "step": 9760
    },
    {
      "epoch": 0.5267389779288759,
      "grad_norm": 1.5667954497497976,
      "learning_rate": 5.122431023611319e-06,
      "loss": 0.6564,
      "step": 9761
    },
    {
      "epoch": 0.5267929415573903,
      "grad_norm": 1.075828019871153,
      "learning_rate": 5.121670826309954e-06,
      "loss": 0.4728,
      "step": 9762
    },
    {
      "epoch": 0.5268469051859047,
      "grad_norm": 1.3283465965055488,
      "learning_rate": 5.120910639882153e-06,
      "loss": 0.5628,
      "step": 9763
    },
    {
      "epoch": 0.5269008688144191,
      "grad_norm": 1.2881249581082195,
      "learning_rate": 5.120150464349762e-06,
      "loss": 0.5282,
      "step": 9764
    },
    {
      "epoch": 0.5269548324429335,
      "grad_norm": 0.9648565593053944,
      "learning_rate": 5.119390299734631e-06,
      "loss": 0.3575,
      "step": 9765
    },
    {
      "epoch": 0.5270087960714478,
      "grad_norm": 0.8737912074238181,
      "learning_rate": 5.118630146058608e-06,
      "loss": 0.3233,
      "step": 9766
    },
    {
      "epoch": 0.5270627596999622,
      "grad_norm": 0.9581093052205283,
      "learning_rate": 5.117870003343539e-06,
      "loss": 0.4413,
      "step": 9767
    },
    {
      "epoch": 0.5271167233284766,
      "grad_norm": 1.1056816767299782,
      "learning_rate": 5.117109871611274e-06,
      "loss": 0.5592,
      "step": 9768
    },
    {
      "epoch": 0.527170686956991,
      "grad_norm": 1.219521000045024,
      "learning_rate": 5.116349750883656e-06,
      "loss": 0.5524,
      "step": 9769
    },
    {
      "epoch": 0.5272246505855054,
      "grad_norm": 1.0594574157616343,
      "learning_rate": 5.115589641182534e-06,
      "loss": 0.5359,
      "step": 9770
    },
    {
      "epoch": 0.5272786142140198,
      "grad_norm": 1.0074772940578125,
      "learning_rate": 5.114829542529754e-06,
      "loss": 0.4576,
      "step": 9771
    },
    {
      "epoch": 0.5273325778425342,
      "grad_norm": 1.0455002313685924,
      "learning_rate": 5.1140694549471635e-06,
      "loss": 0.4928,
      "step": 9772
    },
    {
      "epoch": 0.5273865414710485,
      "grad_norm": 0.7869968859913611,
      "learning_rate": 5.113309378456608e-06,
      "loss": 0.3415,
      "step": 9773
    },
    {
      "epoch": 0.5274405050995629,
      "grad_norm": 1.315459094930469,
      "learning_rate": 5.11254931307993e-06,
      "loss": 0.5753,
      "step": 9774
    },
    {
      "epoch": 0.5274944687280773,
      "grad_norm": 0.9700689783685743,
      "learning_rate": 5.111789258838975e-06,
      "loss": 0.4855,
      "step": 9775
    },
    {
      "epoch": 0.5275484323565917,
      "grad_norm": 1.1195646080759265,
      "learning_rate": 5.111029215755591e-06,
      "loss": 0.4962,
      "step": 9776
    },
    {
      "epoch": 0.527602395985106,
      "grad_norm": 0.861425438986499,
      "learning_rate": 5.1102691838516185e-06,
      "loss": 0.3528,
      "step": 9777
    },
    {
      "epoch": 0.5276563596136205,
      "grad_norm": 1.1611344618701709,
      "learning_rate": 5.109509163148904e-06,
      "loss": 0.6867,
      "step": 9778
    },
    {
      "epoch": 0.5277103232421348,
      "grad_norm": 0.9277800859284727,
      "learning_rate": 5.108749153669291e-06,
      "loss": 0.4409,
      "step": 9779
    },
    {
      "epoch": 0.5277642868706491,
      "grad_norm": 0.9421141034002796,
      "learning_rate": 5.107989155434625e-06,
      "loss": 0.3411,
      "step": 9780
    },
    {
      "epoch": 0.5278182504991635,
      "grad_norm": 0.9459179862262426,
      "learning_rate": 5.107229168466744e-06,
      "loss": 0.4397,
      "step": 9781
    },
    {
      "epoch": 0.5278722141276779,
      "grad_norm": 0.8666449002820407,
      "learning_rate": 5.106469192787494e-06,
      "loss": 0.4316,
      "step": 9782
    },
    {
      "epoch": 0.5279261777561923,
      "grad_norm": 0.9954097631324498,
      "learning_rate": 5.105709228418717e-06,
      "loss": 0.5242,
      "step": 9783
    },
    {
      "epoch": 0.5279801413847067,
      "grad_norm": 0.9910910667758349,
      "learning_rate": 5.104949275382255e-06,
      "loss": 0.4613,
      "step": 9784
    },
    {
      "epoch": 0.5280341050132211,
      "grad_norm": 0.9236007786414026,
      "learning_rate": 5.10418933369995e-06,
      "loss": 0.4041,
      "step": 9785
    },
    {
      "epoch": 0.5280880686417355,
      "grad_norm": 0.8410941077246671,
      "learning_rate": 5.1034294033936435e-06,
      "loss": 0.3278,
      "step": 9786
    },
    {
      "epoch": 0.5281420322702498,
      "grad_norm": 0.9254965677205332,
      "learning_rate": 5.102669484485177e-06,
      "loss": 0.3481,
      "step": 9787
    },
    {
      "epoch": 0.5281959958987642,
      "grad_norm": 1.1727121435124424,
      "learning_rate": 5.101909576996392e-06,
      "loss": 0.7748,
      "step": 9788
    },
    {
      "epoch": 0.5282499595272786,
      "grad_norm": 0.9330127096834071,
      "learning_rate": 5.101149680949127e-06,
      "loss": 0.4355,
      "step": 9789
    },
    {
      "epoch": 0.528303923155793,
      "grad_norm": 1.0660258403400662,
      "learning_rate": 5.100389796365223e-06,
      "loss": 0.5036,
      "step": 9790
    },
    {
      "epoch": 0.5283578867843074,
      "grad_norm": 1.2624947285709311,
      "learning_rate": 5.099629923266519e-06,
      "loss": 0.6403,
      "step": 9791
    },
    {
      "epoch": 0.5284118504128218,
      "grad_norm": 0.9868363276105038,
      "learning_rate": 5.0988700616748575e-06,
      "loss": 0.416,
      "step": 9792
    },
    {
      "epoch": 0.5284658140413362,
      "grad_norm": 0.8434235886846871,
      "learning_rate": 5.098110211612075e-06,
      "loss": 0.3119,
      "step": 9793
    },
    {
      "epoch": 0.5285197776698505,
      "grad_norm": 1.0241521899178454,
      "learning_rate": 5.097350373100012e-06,
      "loss": 0.4987,
      "step": 9794
    },
    {
      "epoch": 0.5285737412983649,
      "grad_norm": 1.1017898221716476,
      "learning_rate": 5.096590546160505e-06,
      "loss": 0.4884,
      "step": 9795
    },
    {
      "epoch": 0.5286277049268793,
      "grad_norm": 0.9052401377852753,
      "learning_rate": 5.095830730815393e-06,
      "loss": 0.4229,
      "step": 9796
    },
    {
      "epoch": 0.5286816685553937,
      "grad_norm": 1.091726520591521,
      "learning_rate": 5.095070927086515e-06,
      "loss": 0.4981,
      "step": 9797
    },
    {
      "epoch": 0.5287356321839081,
      "grad_norm": 1.129908766802422,
      "learning_rate": 5.094311134995707e-06,
      "loss": 0.4722,
      "step": 9798
    },
    {
      "epoch": 0.5287895958124225,
      "grad_norm": 1.1067903806506734,
      "learning_rate": 5.093551354564808e-06,
      "loss": 0.5548,
      "step": 9799
    },
    {
      "epoch": 0.5288435594409369,
      "grad_norm": 1.164834368183249,
      "learning_rate": 5.092791585815652e-06,
      "loss": 0.5298,
      "step": 9800
    },
    {
      "epoch": 0.5288975230694511,
      "grad_norm": 0.9318925025580266,
      "learning_rate": 5.0920318287700795e-06,
      "loss": 0.3744,
      "step": 9801
    },
    {
      "epoch": 0.5289514866979655,
      "grad_norm": 1.011214731257614,
      "learning_rate": 5.091272083449923e-06,
      "loss": 0.4661,
      "step": 9802
    },
    {
      "epoch": 0.5290054503264799,
      "grad_norm": 1.0672583452064317,
      "learning_rate": 5.09051234987702e-06,
      "loss": 0.4989,
      "step": 9803
    },
    {
      "epoch": 0.5290594139549943,
      "grad_norm": 1.1166791342204359,
      "learning_rate": 5.089752628073205e-06,
      "loss": 0.4372,
      "step": 9804
    },
    {
      "epoch": 0.5291133775835087,
      "grad_norm": 1.0902932370358502,
      "learning_rate": 5.088992918060316e-06,
      "loss": 0.4328,
      "step": 9805
    },
    {
      "epoch": 0.5291673412120231,
      "grad_norm": 1.0634458562082594,
      "learning_rate": 5.088233219860185e-06,
      "loss": 0.4843,
      "step": 9806
    },
    {
      "epoch": 0.5292213048405375,
      "grad_norm": 1.0908089063490158,
      "learning_rate": 5.087473533494646e-06,
      "loss": 0.4787,
      "step": 9807
    },
    {
      "epoch": 0.5292752684690518,
      "grad_norm": 0.8399636273841661,
      "learning_rate": 5.086713858985537e-06,
      "loss": 0.3228,
      "step": 9808
    },
    {
      "epoch": 0.5293292320975662,
      "grad_norm": 0.7762796229377825,
      "learning_rate": 5.085954196354688e-06,
      "loss": 0.435,
      "step": 9809
    },
    {
      "epoch": 0.5293831957260806,
      "grad_norm": 1.0468609322474196,
      "learning_rate": 5.085194545623933e-06,
      "loss": 0.4905,
      "step": 9810
    },
    {
      "epoch": 0.529437159354595,
      "grad_norm": 1.0231173852644253,
      "learning_rate": 5.084434906815107e-06,
      "loss": 0.5216,
      "step": 9811
    },
    {
      "epoch": 0.5294911229831094,
      "grad_norm": 0.895246879334178,
      "learning_rate": 5.083675279950041e-06,
      "loss": 0.4641,
      "step": 9812
    },
    {
      "epoch": 0.5295450866116238,
      "grad_norm": 1.1319540334632623,
      "learning_rate": 5.0829156650505675e-06,
      "loss": 0.4614,
      "step": 9813
    },
    {
      "epoch": 0.5295990502401381,
      "grad_norm": 1.103725361563143,
      "learning_rate": 5.08215606213852e-06,
      "loss": 0.401,
      "step": 9814
    },
    {
      "epoch": 0.5296530138686525,
      "grad_norm": 0.8238025325801022,
      "learning_rate": 5.08139647123573e-06,
      "loss": 0.3562,
      "step": 9815
    },
    {
      "epoch": 0.5297069774971669,
      "grad_norm": 0.8809782446819693,
      "learning_rate": 5.080636892364027e-06,
      "loss": 0.3809,
      "step": 9816
    },
    {
      "epoch": 0.5297609411256813,
      "grad_norm": 1.0699923369727389,
      "learning_rate": 5.079877325545245e-06,
      "loss": 0.4714,
      "step": 9817
    },
    {
      "epoch": 0.5298149047541957,
      "grad_norm": 1.2589858822775974,
      "learning_rate": 5.079117770801212e-06,
      "loss": 0.5853,
      "step": 9818
    },
    {
      "epoch": 0.5298688683827101,
      "grad_norm": 1.0082969473806576,
      "learning_rate": 5.078358228153759e-06,
      "loss": 0.4772,
      "step": 9819
    },
    {
      "epoch": 0.5299228320112245,
      "grad_norm": 0.7968675671658431,
      "learning_rate": 5.077598697624718e-06,
      "loss": 0.3038,
      "step": 9820
    },
    {
      "epoch": 0.5299767956397388,
      "grad_norm": 1.200072866757801,
      "learning_rate": 5.076839179235918e-06,
      "loss": 0.6611,
      "step": 9821
    },
    {
      "epoch": 0.5300307592682532,
      "grad_norm": 0.9198703527120782,
      "learning_rate": 5.076079673009187e-06,
      "loss": 0.4199,
      "step": 9822
    },
    {
      "epoch": 0.5300847228967676,
      "grad_norm": 0.8070481286233315,
      "learning_rate": 5.075320178966353e-06,
      "loss": 0.3776,
      "step": 9823
    },
    {
      "epoch": 0.530138686525282,
      "grad_norm": 1.2011079495854977,
      "learning_rate": 5.074560697129247e-06,
      "loss": 0.5155,
      "step": 9824
    },
    {
      "epoch": 0.5301926501537964,
      "grad_norm": 0.9693674232436924,
      "learning_rate": 5.073801227519697e-06,
      "loss": 0.5198,
      "step": 9825
    },
    {
      "epoch": 0.5302466137823107,
      "grad_norm": 0.9793365794881753,
      "learning_rate": 5.073041770159529e-06,
      "loss": 0.4599,
      "step": 9826
    },
    {
      "epoch": 0.5303005774108251,
      "grad_norm": 1.1465256907083443,
      "learning_rate": 5.072282325070573e-06,
      "loss": 0.4597,
      "step": 9827
    },
    {
      "epoch": 0.5303545410393394,
      "grad_norm": 1.19222344909917,
      "learning_rate": 5.0715228922746565e-06,
      "loss": 0.9406,
      "step": 9828
    },
    {
      "epoch": 0.5304085046678538,
      "grad_norm": 0.8308316256432817,
      "learning_rate": 5.070763471793603e-06,
      "loss": 0.4297,
      "step": 9829
    },
    {
      "epoch": 0.5304624682963682,
      "grad_norm": 1.032557663753453,
      "learning_rate": 5.070004063649241e-06,
      "loss": 0.4065,
      "step": 9830
    },
    {
      "epoch": 0.5305164319248826,
      "grad_norm": 0.9130205273050845,
      "learning_rate": 5.069244667863397e-06,
      "loss": 0.3661,
      "step": 9831
    },
    {
      "epoch": 0.530570395553397,
      "grad_norm": 1.2980879046068137,
      "learning_rate": 5.068485284457897e-06,
      "loss": 0.5156,
      "step": 9832
    },
    {
      "epoch": 0.5306243591819114,
      "grad_norm": 1.1081940422911116,
      "learning_rate": 5.067725913454565e-06,
      "loss": 0.5407,
      "step": 9833
    },
    {
      "epoch": 0.5306783228104258,
      "grad_norm": 0.9250727196067227,
      "learning_rate": 5.066966554875228e-06,
      "loss": 0.4882,
      "step": 9834
    },
    {
      "epoch": 0.5307322864389401,
      "grad_norm": 1.233912346123534,
      "learning_rate": 5.06620720874171e-06,
      "loss": 0.4335,
      "step": 9835
    },
    {
      "epoch": 0.5307862500674545,
      "grad_norm": 1.249312703962299,
      "learning_rate": 5.065447875075835e-06,
      "loss": 0.588,
      "step": 9836
    },
    {
      "epoch": 0.5308402136959689,
      "grad_norm": 1.0927634448580101,
      "learning_rate": 5.0646885538994264e-06,
      "loss": 0.5695,
      "step": 9837
    },
    {
      "epoch": 0.5308941773244833,
      "grad_norm": 0.990576666123403,
      "learning_rate": 5.063929245234309e-06,
      "loss": 0.5136,
      "step": 9838
    },
    {
      "epoch": 0.5309481409529977,
      "grad_norm": 0.8747830891475921,
      "learning_rate": 5.063169949102306e-06,
      "loss": 0.4227,
      "step": 9839
    },
    {
      "epoch": 0.5310021045815121,
      "grad_norm": 1.050525506360563,
      "learning_rate": 5.062410665525239e-06,
      "loss": 0.5972,
      "step": 9840
    },
    {
      "epoch": 0.5310560682100265,
      "grad_norm": 1.0301176009511661,
      "learning_rate": 5.061651394524932e-06,
      "loss": 0.4896,
      "step": 9841
    },
    {
      "epoch": 0.5311100318385408,
      "grad_norm": 1.0053303710032866,
      "learning_rate": 5.060892136123208e-06,
      "loss": 0.5311,
      "step": 9842
    },
    {
      "epoch": 0.5311639954670552,
      "grad_norm": 1.0620610276914073,
      "learning_rate": 5.060132890341887e-06,
      "loss": 0.4438,
      "step": 9843
    },
    {
      "epoch": 0.5312179590955696,
      "grad_norm": 1.1317969337548908,
      "learning_rate": 5.0593736572027905e-06,
      "loss": 0.4412,
      "step": 9844
    },
    {
      "epoch": 0.531271922724084,
      "grad_norm": 1.0404419773815328,
      "learning_rate": 5.05861443672774e-06,
      "loss": 0.5023,
      "step": 9845
    },
    {
      "epoch": 0.5313258863525984,
      "grad_norm": 1.0141537173468835,
      "learning_rate": 5.0578552289385575e-06,
      "loss": 0.4462,
      "step": 9846
    },
    {
      "epoch": 0.5313798499811128,
      "grad_norm": 0.9401250978202378,
      "learning_rate": 5.0570960338570615e-06,
      "loss": 0.463,
      "step": 9847
    },
    {
      "epoch": 0.5314338136096272,
      "grad_norm": 1.1221144048164005,
      "learning_rate": 5.056336851505075e-06,
      "loss": 0.4806,
      "step": 9848
    },
    {
      "epoch": 0.5314877772381414,
      "grad_norm": 1.1534553205309213,
      "learning_rate": 5.055577681904415e-06,
      "loss": 0.5452,
      "step": 9849
    },
    {
      "epoch": 0.5315417408666558,
      "grad_norm": 1.2663836573677207,
      "learning_rate": 5.054818525076901e-06,
      "loss": 0.4412,
      "step": 9850
    },
    {
      "epoch": 0.5315957044951702,
      "grad_norm": 1.0019763584537527,
      "learning_rate": 5.054059381044353e-06,
      "loss": 0.3483,
      "step": 9851
    },
    {
      "epoch": 0.5316496681236846,
      "grad_norm": 1.1120904249583754,
      "learning_rate": 5.053300249828588e-06,
      "loss": 0.6354,
      "step": 9852
    },
    {
      "epoch": 0.531703631752199,
      "grad_norm": 0.9386835938236144,
      "learning_rate": 5.052541131451425e-06,
      "loss": 0.4406,
      "step": 9853
    },
    {
      "epoch": 0.5317575953807134,
      "grad_norm": 0.9317790243009573,
      "learning_rate": 5.051782025934683e-06,
      "loss": 0.4363,
      "step": 9854
    },
    {
      "epoch": 0.5318115590092278,
      "grad_norm": 1.0522881030211095,
      "learning_rate": 5.051022933300178e-06,
      "loss": 0.5451,
      "step": 9855
    },
    {
      "epoch": 0.5318655226377421,
      "grad_norm": 0.8655048394895103,
      "learning_rate": 5.050263853569729e-06,
      "loss": 0.4232,
      "step": 9856
    },
    {
      "epoch": 0.5319194862662565,
      "grad_norm": 0.9568939782338598,
      "learning_rate": 5.049504786765149e-06,
      "loss": 0.3889,
      "step": 9857
    },
    {
      "epoch": 0.5319734498947709,
      "grad_norm": 1.1367502463525687,
      "learning_rate": 5.048745732908258e-06,
      "loss": 0.5957,
      "step": 9858
    },
    {
      "epoch": 0.5320274135232853,
      "grad_norm": 0.7633109806817424,
      "learning_rate": 5.047986692020869e-06,
      "loss": 0.3029,
      "step": 9859
    },
    {
      "epoch": 0.5320813771517997,
      "grad_norm": 0.8417295438660892,
      "learning_rate": 5.047227664124801e-06,
      "loss": 0.3256,
      "step": 9860
    },
    {
      "epoch": 0.5321353407803141,
      "grad_norm": 1.068478106083507,
      "learning_rate": 5.046468649241867e-06,
      "loss": 0.4265,
      "step": 9861
    },
    {
      "epoch": 0.5321893044088285,
      "grad_norm": 1.1689198391610829,
      "learning_rate": 5.045709647393882e-06,
      "loss": 0.49,
      "step": 9862
    },
    {
      "epoch": 0.5322432680373428,
      "grad_norm": 1.0050888489163547,
      "learning_rate": 5.044950658602661e-06,
      "loss": 0.4901,
      "step": 9863
    },
    {
      "epoch": 0.5322972316658572,
      "grad_norm": 0.9339358575043493,
      "learning_rate": 5.044191682890017e-06,
      "loss": 0.4808,
      "step": 9864
    },
    {
      "epoch": 0.5323511952943716,
      "grad_norm": 1.0027021901464566,
      "learning_rate": 5.043432720277766e-06,
      "loss": 0.4548,
      "step": 9865
    },
    {
      "epoch": 0.532405158922886,
      "grad_norm": 1.030074293410242,
      "learning_rate": 5.042673770787718e-06,
      "loss": 0.4445,
      "step": 9866
    },
    {
      "epoch": 0.5324591225514004,
      "grad_norm": 1.1744694775459374,
      "learning_rate": 5.04191483444169e-06,
      "loss": 0.6395,
      "step": 9867
    },
    {
      "epoch": 0.5325130861799148,
      "grad_norm": 1.0277624429924037,
      "learning_rate": 5.041155911261492e-06,
      "loss": 0.451,
      "step": 9868
    },
    {
      "epoch": 0.5325670498084292,
      "grad_norm": 1.1838013682332493,
      "learning_rate": 5.040397001268936e-06,
      "loss": 0.4901,
      "step": 9869
    },
    {
      "epoch": 0.5326210134369435,
      "grad_norm": 0.9336277616702371,
      "learning_rate": 5.039638104485835e-06,
      "loss": 0.3854,
      "step": 9870
    },
    {
      "epoch": 0.5326749770654579,
      "grad_norm": 0.9841520244418016,
      "learning_rate": 5.038879220934e-06,
      "loss": 0.4295,
      "step": 9871
    },
    {
      "epoch": 0.5327289406939723,
      "grad_norm": 1.3609108323575496,
      "learning_rate": 5.038120350635241e-06,
      "loss": 0.645,
      "step": 9872
    },
    {
      "epoch": 0.5327829043224867,
      "grad_norm": 1.1932349376146079,
      "learning_rate": 5.0373614936113725e-06,
      "loss": 0.6341,
      "step": 9873
    },
    {
      "epoch": 0.532836867951001,
      "grad_norm": 0.9936516635062216,
      "learning_rate": 5.0366026498842e-06,
      "loss": 0.5415,
      "step": 9874
    },
    {
      "epoch": 0.5328908315795154,
      "grad_norm": 1.2946549071647346,
      "learning_rate": 5.035843819475535e-06,
      "loss": 0.6859,
      "step": 9875
    },
    {
      "epoch": 0.5329447952080298,
      "grad_norm": 0.8782416467466894,
      "learning_rate": 5.03508500240719e-06,
      "loss": 0.4893,
      "step": 9876
    },
    {
      "epoch": 0.5329987588365441,
      "grad_norm": 0.8622742496677499,
      "learning_rate": 5.034326198700972e-06,
      "loss": 0.4459,
      "step": 9877
    },
    {
      "epoch": 0.5330527224650585,
      "grad_norm": 1.001155572713495,
      "learning_rate": 5.033567408378688e-06,
      "loss": 0.5214,
      "step": 9878
    },
    {
      "epoch": 0.5331066860935729,
      "grad_norm": 0.9413203695713206,
      "learning_rate": 5.032808631462147e-06,
      "loss": 0.5098,
      "step": 9879
    },
    {
      "epoch": 0.5331606497220873,
      "grad_norm": 0.9853599885948897,
      "learning_rate": 5.032049867973159e-06,
      "loss": 0.5277,
      "step": 9880
    },
    {
      "epoch": 0.5332146133506017,
      "grad_norm": 0.8919222299896387,
      "learning_rate": 5.031291117933531e-06,
      "loss": 0.3975,
      "step": 9881
    },
    {
      "epoch": 0.5332685769791161,
      "grad_norm": 0.8801126447326915,
      "learning_rate": 5.030532381365069e-06,
      "loss": 0.4566,
      "step": 9882
    },
    {
      "epoch": 0.5333225406076304,
      "grad_norm": 0.8732902618164328,
      "learning_rate": 5.029773658289583e-06,
      "loss": 0.415,
      "step": 9883
    },
    {
      "epoch": 0.5333765042361448,
      "grad_norm": 0.8632648443910056,
      "learning_rate": 5.029014948728875e-06,
      "loss": 0.3298,
      "step": 9884
    },
    {
      "epoch": 0.5334304678646592,
      "grad_norm": 1.0198642654019867,
      "learning_rate": 5.028256252704754e-06,
      "loss": 0.4438,
      "step": 9885
    },
    {
      "epoch": 0.5334844314931736,
      "grad_norm": 0.921745908386737,
      "learning_rate": 5.027497570239024e-06,
      "loss": 0.3589,
      "step": 9886
    },
    {
      "epoch": 0.533538395121688,
      "grad_norm": 1.1292652179956546,
      "learning_rate": 5.026738901353493e-06,
      "loss": 0.5313,
      "step": 9887
    },
    {
      "epoch": 0.5335923587502024,
      "grad_norm": 1.2291531847651713,
      "learning_rate": 5.025980246069963e-06,
      "loss": 0.5267,
      "step": 9888
    },
    {
      "epoch": 0.5336463223787168,
      "grad_norm": 1.2463751830352283,
      "learning_rate": 5.025221604410239e-06,
      "loss": 0.6619,
      "step": 9889
    },
    {
      "epoch": 0.5337002860072311,
      "grad_norm": 1.0508197961317378,
      "learning_rate": 5.024462976396127e-06,
      "loss": 0.4837,
      "step": 9890
    },
    {
      "epoch": 0.5337542496357455,
      "grad_norm": 1.3351591006918475,
      "learning_rate": 5.02370436204943e-06,
      "loss": 0.6931,
      "step": 9891
    },
    {
      "epoch": 0.5338082132642599,
      "grad_norm": 1.2738846483289565,
      "learning_rate": 5.022945761391949e-06,
      "loss": 0.6689,
      "step": 9892
    },
    {
      "epoch": 0.5338621768927743,
      "grad_norm": 1.003723637203252,
      "learning_rate": 5.02218717444549e-06,
      "loss": 0.385,
      "step": 9893
    },
    {
      "epoch": 0.5339161405212887,
      "grad_norm": 0.8228043027279296,
      "learning_rate": 5.021428601231854e-06,
      "loss": 0.3498,
      "step": 9894
    },
    {
      "epoch": 0.5339701041498031,
      "grad_norm": 0.9422487904435871,
      "learning_rate": 5.020670041772844e-06,
      "loss": 0.3991,
      "step": 9895
    },
    {
      "epoch": 0.5340240677783175,
      "grad_norm": 1.072372447185218,
      "learning_rate": 5.019911496090261e-06,
      "loss": 0.4389,
      "step": 9896
    },
    {
      "epoch": 0.5340780314068317,
      "grad_norm": 1.0247539251499445,
      "learning_rate": 5.019152964205909e-06,
      "loss": 0.4601,
      "step": 9897
    },
    {
      "epoch": 0.5341319950353461,
      "grad_norm": 0.9346400612150642,
      "learning_rate": 5.018394446141584e-06,
      "loss": 0.4112,
      "step": 9898
    },
    {
      "epoch": 0.5341859586638605,
      "grad_norm": 1.0157134136928683,
      "learning_rate": 5.01763594191909e-06,
      "loss": 0.4248,
      "step": 9899
    },
    {
      "epoch": 0.5342399222923749,
      "grad_norm": 1.1000510907534347,
      "learning_rate": 5.016877451560226e-06,
      "loss": 0.5227,
      "step": 9900
    },
    {
      "epoch": 0.5342938859208893,
      "grad_norm": 0.9351609773266695,
      "learning_rate": 5.016118975086792e-06,
      "loss": 0.3496,
      "step": 9901
    },
    {
      "epoch": 0.5343478495494037,
      "grad_norm": 0.9744558684718088,
      "learning_rate": 5.015360512520588e-06,
      "loss": 0.3825,
      "step": 9902
    },
    {
      "epoch": 0.5344018131779181,
      "grad_norm": 0.9809932430668968,
      "learning_rate": 5.014602063883413e-06,
      "loss": 0.4075,
      "step": 9903
    },
    {
      "epoch": 0.5344557768064324,
      "grad_norm": 1.0857427377407478,
      "learning_rate": 5.013843629197066e-06,
      "loss": 0.5079,
      "step": 9904
    },
    {
      "epoch": 0.5345097404349468,
      "grad_norm": 1.2321502834617455,
      "learning_rate": 5.0130852084833425e-06,
      "loss": 0.639,
      "step": 9905
    },
    {
      "epoch": 0.5345637040634612,
      "grad_norm": 0.9583361535841615,
      "learning_rate": 5.012326801764043e-06,
      "loss": 0.4372,
      "step": 9906
    },
    {
      "epoch": 0.5346176676919756,
      "grad_norm": 1.1510542377943596,
      "learning_rate": 5.011568409060963e-06,
      "loss": 0.5659,
      "step": 9907
    },
    {
      "epoch": 0.53467163132049,
      "grad_norm": 0.9507329476502328,
      "learning_rate": 5.0108100303959e-06,
      "loss": 0.5195,
      "step": 9908
    },
    {
      "epoch": 0.5347255949490044,
      "grad_norm": 1.1224062928378957,
      "learning_rate": 5.010051665790653e-06,
      "loss": 0.5348,
      "step": 9909
    },
    {
      "epoch": 0.5347795585775188,
      "grad_norm": 1.065135566106261,
      "learning_rate": 5.009293315267015e-06,
      "loss": 0.4981,
      "step": 9910
    },
    {
      "epoch": 0.5348335222060331,
      "grad_norm": 0.9873706183018753,
      "learning_rate": 5.008534978846785e-06,
      "loss": 0.3475,
      "step": 9911
    },
    {
      "epoch": 0.5348874858345475,
      "grad_norm": 0.8889299051670229,
      "learning_rate": 5.007776656551755e-06,
      "loss": 0.4004,
      "step": 9912
    },
    {
      "epoch": 0.5349414494630619,
      "grad_norm": 0.828539413485709,
      "learning_rate": 5.00701834840372e-06,
      "loss": 0.4221,
      "step": 9913
    },
    {
      "epoch": 0.5349954130915763,
      "grad_norm": 0.8820196211274777,
      "learning_rate": 5.006260054424477e-06,
      "loss": 0.4072,
      "step": 9914
    },
    {
      "epoch": 0.5350493767200907,
      "grad_norm": 1.036560531293208,
      "learning_rate": 5.005501774635818e-06,
      "loss": 0.4643,
      "step": 9915
    },
    {
      "epoch": 0.5351033403486051,
      "grad_norm": 0.9400599873703808,
      "learning_rate": 5.004743509059538e-06,
      "loss": 0.5429,
      "step": 9916
    },
    {
      "epoch": 0.5351573039771195,
      "grad_norm": 0.9222972212389214,
      "learning_rate": 5.003985257717429e-06,
      "loss": 0.5607,
      "step": 9917
    },
    {
      "epoch": 0.5352112676056338,
      "grad_norm": 1.01275295247848,
      "learning_rate": 5.003227020631287e-06,
      "loss": 0.4581,
      "step": 9918
    },
    {
      "epoch": 0.5352652312341482,
      "grad_norm": 0.937493077123418,
      "learning_rate": 5.002468797822901e-06,
      "loss": 0.3713,
      "step": 9919
    },
    {
      "epoch": 0.5353191948626626,
      "grad_norm": 0.9202956312894247,
      "learning_rate": 5.001710589314065e-06,
      "loss": 0.3609,
      "step": 9920
    },
    {
      "epoch": 0.535373158491177,
      "grad_norm": 0.6868329139027787,
      "learning_rate": 5.000952395126571e-06,
      "loss": 0.294,
      "step": 9921
    },
    {
      "epoch": 0.5354271221196913,
      "grad_norm": 0.9196054520244205,
      "learning_rate": 5.000194215282207e-06,
      "loss": 0.4313,
      "step": 9922
    },
    {
      "epoch": 0.5354810857482057,
      "grad_norm": 0.9474701432519063,
      "learning_rate": 4.9994360498027684e-06,
      "loss": 0.5766,
      "step": 9923
    },
    {
      "epoch": 0.5355350493767201,
      "grad_norm": 1.03804268584054,
      "learning_rate": 4.998677898710042e-06,
      "loss": 0.5034,
      "step": 9924
    },
    {
      "epoch": 0.5355890130052344,
      "grad_norm": 1.2472522864702218,
      "learning_rate": 4.997919762025821e-06,
      "loss": 0.6893,
      "step": 9925
    },
    {
      "epoch": 0.5356429766337488,
      "grad_norm": 1.050550411378059,
      "learning_rate": 4.997161639771892e-06,
      "loss": 0.4198,
      "step": 9926
    },
    {
      "epoch": 0.5356969402622632,
      "grad_norm": 0.9629876644259304,
      "learning_rate": 4.996403531970048e-06,
      "loss": 0.4744,
      "step": 9927
    },
    {
      "epoch": 0.5357509038907776,
      "grad_norm": 1.1310545570967452,
      "learning_rate": 4.995645438642071e-06,
      "loss": 0.4855,
      "step": 9928
    },
    {
      "epoch": 0.535804867519292,
      "grad_norm": 1.1131251685025116,
      "learning_rate": 4.9948873598097565e-06,
      "loss": 0.6041,
      "step": 9929
    },
    {
      "epoch": 0.5358588311478064,
      "grad_norm": 0.9593199274930416,
      "learning_rate": 4.994129295494889e-06,
      "loss": 0.3513,
      "step": 9930
    },
    {
      "epoch": 0.5359127947763208,
      "grad_norm": 0.9418600222505135,
      "learning_rate": 4.993371245719257e-06,
      "loss": 0.4123,
      "step": 9931
    },
    {
      "epoch": 0.5359667584048351,
      "grad_norm": 1.0131850971198793,
      "learning_rate": 4.99261321050465e-06,
      "loss": 0.5146,
      "step": 9932
    },
    {
      "epoch": 0.5360207220333495,
      "grad_norm": 1.042539650127272,
      "learning_rate": 4.991855189872848e-06,
      "loss": 0.5174,
      "step": 9933
    },
    {
      "epoch": 0.5360746856618639,
      "grad_norm": 1.0708511066650659,
      "learning_rate": 4.991097183845642e-06,
      "loss": 0.5208,
      "step": 9934
    },
    {
      "epoch": 0.5361286492903783,
      "grad_norm": 0.743763493488394,
      "learning_rate": 4.990339192444819e-06,
      "loss": 0.2784,
      "step": 9935
    },
    {
      "epoch": 0.5361826129188927,
      "grad_norm": 1.0882365536218006,
      "learning_rate": 4.989581215692161e-06,
      "loss": 0.4035,
      "step": 9936
    },
    {
      "epoch": 0.5362365765474071,
      "grad_norm": 1.0055235448222235,
      "learning_rate": 4.988823253609454e-06,
      "loss": 0.501,
      "step": 9937
    },
    {
      "epoch": 0.5362905401759215,
      "grad_norm": 0.9908096396198701,
      "learning_rate": 4.9880653062184835e-06,
      "loss": 0.5071,
      "step": 9938
    },
    {
      "epoch": 0.5363445038044358,
      "grad_norm": 0.6381296162085466,
      "learning_rate": 4.987307373541035e-06,
      "loss": 0.2287,
      "step": 9939
    },
    {
      "epoch": 0.5363984674329502,
      "grad_norm": 1.0868617962544562,
      "learning_rate": 4.9865494555988895e-06,
      "loss": 0.5341,
      "step": 9940
    },
    {
      "epoch": 0.5364524310614646,
      "grad_norm": 1.186078147742559,
      "learning_rate": 4.985791552413831e-06,
      "loss": 0.5795,
      "step": 9941
    },
    {
      "epoch": 0.536506394689979,
      "grad_norm": 0.8673224930155873,
      "learning_rate": 4.985033664007644e-06,
      "loss": 0.3904,
      "step": 9942
    },
    {
      "epoch": 0.5365603583184934,
      "grad_norm": 0.9759421130768049,
      "learning_rate": 4.98427579040211e-06,
      "loss": 0.4662,
      "step": 9943
    },
    {
      "epoch": 0.5366143219470078,
      "grad_norm": 0.8927853030036168,
      "learning_rate": 4.98351793161901e-06,
      "loss": 0.3303,
      "step": 9944
    },
    {
      "epoch": 0.5366682855755222,
      "grad_norm": 1.1988665229002389,
      "learning_rate": 4.982760087680125e-06,
      "loss": 0.6174,
      "step": 9945
    },
    {
      "epoch": 0.5367222492040364,
      "grad_norm": 1.0085113795051772,
      "learning_rate": 4.98200225860724e-06,
      "loss": 0.3234,
      "step": 9946
    },
    {
      "epoch": 0.5367762128325508,
      "grad_norm": 0.9738048223442025,
      "learning_rate": 4.981244444422134e-06,
      "loss": 0.5131,
      "step": 9947
    },
    {
      "epoch": 0.5368301764610652,
      "grad_norm": 0.9273330592169257,
      "learning_rate": 4.980486645146585e-06,
      "loss": 0.5312,
      "step": 9948
    },
    {
      "epoch": 0.5368841400895796,
      "grad_norm": 0.8856463690253309,
      "learning_rate": 4.979728860802375e-06,
      "loss": 0.3729,
      "step": 9949
    },
    {
      "epoch": 0.536938103718094,
      "grad_norm": 1.015102081148679,
      "learning_rate": 4.978971091411283e-06,
      "loss": 0.488,
      "step": 9950
    },
    {
      "epoch": 0.5369920673466084,
      "grad_norm": 0.9493161455619169,
      "learning_rate": 4.978213336995089e-06,
      "loss": 0.4124,
      "step": 9951
    },
    {
      "epoch": 0.5370460309751227,
      "grad_norm": 1.0560354669019085,
      "learning_rate": 4.977455597575572e-06,
      "loss": 0.4308,
      "step": 9952
    },
    {
      "epoch": 0.5370999946036371,
      "grad_norm": 1.0921978231321052,
      "learning_rate": 4.976697873174509e-06,
      "loss": 0.5585,
      "step": 9953
    },
    {
      "epoch": 0.5371539582321515,
      "grad_norm": 0.8200884871965322,
      "learning_rate": 4.975940163813676e-06,
      "loss": 0.3604,
      "step": 9954
    },
    {
      "epoch": 0.5372079218606659,
      "grad_norm": 0.9025525865995081,
      "learning_rate": 4.975182469514853e-06,
      "loss": 0.407,
      "step": 9955
    },
    {
      "epoch": 0.5372618854891803,
      "grad_norm": 1.2137695871309955,
      "learning_rate": 4.9744247902998155e-06,
      "loss": 0.5944,
      "step": 9956
    },
    {
      "epoch": 0.5373158491176947,
      "grad_norm": 1.2539473727861978,
      "learning_rate": 4.973667126190341e-06,
      "loss": 0.5053,
      "step": 9957
    },
    {
      "epoch": 0.5373698127462091,
      "grad_norm": 0.9461718165150234,
      "learning_rate": 4.972909477208206e-06,
      "loss": 0.3656,
      "step": 9958
    },
    {
      "epoch": 0.5374237763747234,
      "grad_norm": 1.0754887069747932,
      "learning_rate": 4.972151843375186e-06,
      "loss": 0.4998,
      "step": 9959
    },
    {
      "epoch": 0.5374777400032378,
      "grad_norm": 1.1649645093618626,
      "learning_rate": 4.971394224713053e-06,
      "loss": 0.5293,
      "step": 9960
    },
    {
      "epoch": 0.5375317036317522,
      "grad_norm": 0.9499067076848354,
      "learning_rate": 4.9706366212435856e-06,
      "loss": 0.413,
      "step": 9961
    },
    {
      "epoch": 0.5375856672602666,
      "grad_norm": 1.051429901273323,
      "learning_rate": 4.969879032988556e-06,
      "loss": 0.4286,
      "step": 9962
    },
    {
      "epoch": 0.537639630888781,
      "grad_norm": 1.2339737274249403,
      "learning_rate": 4.969121459969738e-06,
      "loss": 0.5373,
      "step": 9963
    },
    {
      "epoch": 0.5376935945172954,
      "grad_norm": 0.7793456059658681,
      "learning_rate": 4.968363902208906e-06,
      "loss": 0.3721,
      "step": 9964
    },
    {
      "epoch": 0.5377475581458098,
      "grad_norm": 1.3145167940958153,
      "learning_rate": 4.967606359727834e-06,
      "loss": 0.5204,
      "step": 9965
    },
    {
      "epoch": 0.537801521774324,
      "grad_norm": 1.1498640974596428,
      "learning_rate": 4.966848832548293e-06,
      "loss": 0.523,
      "step": 9966
    },
    {
      "epoch": 0.5378554854028385,
      "grad_norm": 0.9926044738266445,
      "learning_rate": 4.9660913206920535e-06,
      "loss": 0.5252,
      "step": 9967
    },
    {
      "epoch": 0.5379094490313528,
      "grad_norm": 1.046409600341498,
      "learning_rate": 4.965333824180889e-06,
      "loss": 0.4583,
      "step": 9968
    },
    {
      "epoch": 0.5379634126598672,
      "grad_norm": 0.9396402268187883,
      "learning_rate": 4.964576343036572e-06,
      "loss": 0.4133,
      "step": 9969
    },
    {
      "epoch": 0.5380173762883816,
      "grad_norm": 0.9818600047451499,
      "learning_rate": 4.963818877280871e-06,
      "loss": 0.4937,
      "step": 9970
    },
    {
      "epoch": 0.538071339916896,
      "grad_norm": 1.0704144035904015,
      "learning_rate": 4.963061426935557e-06,
      "loss": 0.3779,
      "step": 9971
    },
    {
      "epoch": 0.5381253035454104,
      "grad_norm": 0.9758671262158307,
      "learning_rate": 4.962303992022401e-06,
      "loss": 0.3514,
      "step": 9972
    },
    {
      "epoch": 0.5381792671739247,
      "grad_norm": 1.0071932936259704,
      "learning_rate": 4.9615465725631725e-06,
      "loss": 0.3695,
      "step": 9973
    },
    {
      "epoch": 0.5382332308024391,
      "grad_norm": 0.9053524776508026,
      "learning_rate": 4.960789168579639e-06,
      "loss": 0.3778,
      "step": 9974
    },
    {
      "epoch": 0.5382871944309535,
      "grad_norm": 1.0483345315799397,
      "learning_rate": 4.960031780093568e-06,
      "loss": 0.4662,
      "step": 9975
    },
    {
      "epoch": 0.5383411580594679,
      "grad_norm": 0.9474343835230155,
      "learning_rate": 4.95927440712673e-06,
      "loss": 0.4068,
      "step": 9976
    },
    {
      "epoch": 0.5383951216879823,
      "grad_norm": 1.0865828373788853,
      "learning_rate": 4.958517049700891e-06,
      "loss": 0.4529,
      "step": 9977
    },
    {
      "epoch": 0.5384490853164967,
      "grad_norm": 0.8694051795362834,
      "learning_rate": 4.957759707837821e-06,
      "loss": 0.5141,
      "step": 9978
    },
    {
      "epoch": 0.5385030489450111,
      "grad_norm": 1.0242475093677368,
      "learning_rate": 4.957002381559284e-06,
      "loss": 0.4869,
      "step": 9979
    },
    {
      "epoch": 0.5385570125735254,
      "grad_norm": 0.923632879600402,
      "learning_rate": 4.956245070887049e-06,
      "loss": 0.3909,
      "step": 9980
    },
    {
      "epoch": 0.5386109762020398,
      "grad_norm": 0.965631436444083,
      "learning_rate": 4.955487775842878e-06,
      "loss": 0.4837,
      "step": 9981
    },
    {
      "epoch": 0.5386649398305542,
      "grad_norm": 1.1582940810796427,
      "learning_rate": 4.9547304964485385e-06,
      "loss": 0.5154,
      "step": 9982
    },
    {
      "epoch": 0.5387189034590686,
      "grad_norm": 1.1526870170586134,
      "learning_rate": 4.953973232725796e-06,
      "loss": 0.5078,
      "step": 9983
    },
    {
      "epoch": 0.538772867087583,
      "grad_norm": 1.0148047372193552,
      "learning_rate": 4.953215984696414e-06,
      "loss": 0.3773,
      "step": 9984
    },
    {
      "epoch": 0.5388268307160974,
      "grad_norm": 0.897624058709063,
      "learning_rate": 4.9524587523821574e-06,
      "loss": 0.3919,
      "step": 9985
    },
    {
      "epoch": 0.5388807943446118,
      "grad_norm": 0.9733251239242692,
      "learning_rate": 4.95170153580479e-06,
      "loss": 0.403,
      "step": 9986
    },
    {
      "epoch": 0.5389347579731261,
      "grad_norm": 1.1212704429787337,
      "learning_rate": 4.9509443349860755e-06,
      "loss": 0.4445,
      "step": 9987
    },
    {
      "epoch": 0.5389887216016405,
      "grad_norm": 1.1846771442444723,
      "learning_rate": 4.9501871499477735e-06,
      "loss": 0.5658,
      "step": 9988
    },
    {
      "epoch": 0.5390426852301549,
      "grad_norm": 0.7852794876464328,
      "learning_rate": 4.9494299807116495e-06,
      "loss": 0.3895,
      "step": 9989
    },
    {
      "epoch": 0.5390966488586693,
      "grad_norm": 0.8609658882641552,
      "learning_rate": 4.948672827299463e-06,
      "loss": 0.5473,
      "step": 9990
    },
    {
      "epoch": 0.5391506124871837,
      "grad_norm": 1.0706317109875338,
      "learning_rate": 4.9479156897329776e-06,
      "loss": 0.5802,
      "step": 9991
    },
    {
      "epoch": 0.539204576115698,
      "grad_norm": 0.7599801541566236,
      "learning_rate": 4.947158568033953e-06,
      "loss": 0.3802,
      "step": 9992
    },
    {
      "epoch": 0.5392585397442125,
      "grad_norm": 1.0736956105114253,
      "learning_rate": 4.946401462224149e-06,
      "loss": 0.5108,
      "step": 9993
    },
    {
      "epoch": 0.5393125033727267,
      "grad_norm": 0.8649068876074764,
      "learning_rate": 4.945644372325328e-06,
      "loss": 0.4074,
      "step": 9994
    },
    {
      "epoch": 0.5393664670012411,
      "grad_norm": 0.8941368180658905,
      "learning_rate": 4.9448872983592455e-06,
      "loss": 0.3707,
      "step": 9995
    },
    {
      "epoch": 0.5394204306297555,
      "grad_norm": 0.8867481953064631,
      "learning_rate": 4.944130240347664e-06,
      "loss": 0.3967,
      "step": 9996
    },
    {
      "epoch": 0.5394743942582699,
      "grad_norm": 0.8797379594661722,
      "learning_rate": 4.943373198312341e-06,
      "loss": 0.3776,
      "step": 9997
    },
    {
      "epoch": 0.5395283578867843,
      "grad_norm": 1.2252260373158261,
      "learning_rate": 4.9426161722750355e-06,
      "loss": 0.5581,
      "step": 9998
    },
    {
      "epoch": 0.5395823215152987,
      "grad_norm": 1.1651144408525247,
      "learning_rate": 4.941859162257503e-06,
      "loss": 0.6326,
      "step": 9999
    },
    {
      "epoch": 0.5396362851438131,
      "grad_norm": 1.0691149570888858,
      "learning_rate": 4.941102168281503e-06,
      "loss": 0.6408,
      "step": 10000
    },
    {
      "epoch": 0.5396362851438131,
      "eval_loss": 0.5460962057113647,
      "eval_runtime": 160.341,
      "eval_samples_per_second": 21.448,
      "eval_steps_per_second": 0.898,
      "step": 10000
    },
    {
      "epoch": 0.5396902487723274,
      "grad_norm": 1.0460036031545223,
      "learning_rate": 4.940345190368794e-06,
      "loss": 0.5537,
      "step": 10001
    },
    {
      "epoch": 0.5397442124008418,
      "grad_norm": 1.3047804085059709,
      "learning_rate": 4.9395882285411255e-06,
      "loss": 0.5249,
      "step": 10002
    },
    {
      "epoch": 0.5397981760293562,
      "grad_norm": 1.113219962245879,
      "learning_rate": 4.93883128282026e-06,
      "loss": 0.5417,
      "step": 10003
    },
    {
      "epoch": 0.5398521396578706,
      "grad_norm": 0.9920627395802261,
      "learning_rate": 4.93807435322795e-06,
      "loss": 0.4565,
      "step": 10004
    },
    {
      "epoch": 0.539906103286385,
      "grad_norm": 1.090895486822463,
      "learning_rate": 4.937317439785949e-06,
      "loss": 0.4838,
      "step": 10005
    },
    {
      "epoch": 0.5399600669148994,
      "grad_norm": 0.7996832714004838,
      "learning_rate": 4.9365605425160155e-06,
      "loss": 0.4138,
      "step": 10006
    },
    {
      "epoch": 0.5400140305434138,
      "grad_norm": 0.7779324032367219,
      "learning_rate": 4.9358036614399e-06,
      "loss": 0.2693,
      "step": 10007
    },
    {
      "epoch": 0.5400679941719281,
      "grad_norm": 1.078114989951544,
      "learning_rate": 4.9350467965793605e-06,
      "loss": 0.4834,
      "step": 10008
    },
    {
      "epoch": 0.5401219578004425,
      "grad_norm": 0.9475549291097144,
      "learning_rate": 4.934289947956143e-06,
      "loss": 0.4158,
      "step": 10009
    },
    {
      "epoch": 0.5401759214289569,
      "grad_norm": 0.9848080847361335,
      "learning_rate": 4.933533115592006e-06,
      "loss": 0.369,
      "step": 10010
    },
    {
      "epoch": 0.5402298850574713,
      "grad_norm": 0.9623952623872648,
      "learning_rate": 4.932776299508699e-06,
      "loss": 0.4438,
      "step": 10011
    },
    {
      "epoch": 0.5402838486859857,
      "grad_norm": 1.033107810705618,
      "learning_rate": 4.932019499727973e-06,
      "loss": 0.4444,
      "step": 10012
    },
    {
      "epoch": 0.5403378123145001,
      "grad_norm": 0.8691818524181658,
      "learning_rate": 4.931262716271582e-06,
      "loss": 0.4057,
      "step": 10013
    },
    {
      "epoch": 0.5403917759430145,
      "grad_norm": 1.1479433592950212,
      "learning_rate": 4.930505949161275e-06,
      "loss": 0.4497,
      "step": 10014
    },
    {
      "epoch": 0.5404457395715287,
      "grad_norm": 1.135297485942887,
      "learning_rate": 4.929749198418803e-06,
      "loss": 0.4509,
      "step": 10015
    },
    {
      "epoch": 0.5404997032000431,
      "grad_norm": 0.9789666214374919,
      "learning_rate": 4.928992464065916e-06,
      "loss": 0.4278,
      "step": 10016
    },
    {
      "epoch": 0.5405536668285575,
      "grad_norm": 0.826805450607612,
      "learning_rate": 4.928235746124361e-06,
      "loss": 0.3297,
      "step": 10017
    },
    {
      "epoch": 0.5406076304570719,
      "grad_norm": 0.9670934693603411,
      "learning_rate": 4.927479044615888e-06,
      "loss": 0.401,
      "step": 10018
    },
    {
      "epoch": 0.5406615940855863,
      "grad_norm": 1.0652428773874405,
      "learning_rate": 4.926722359562247e-06,
      "loss": 0.4934,
      "step": 10019
    },
    {
      "epoch": 0.5407155577141007,
      "grad_norm": 1.0266094398929384,
      "learning_rate": 4.925965690985184e-06,
      "loss": 0.6038,
      "step": 10020
    },
    {
      "epoch": 0.540769521342615,
      "grad_norm": 0.8747862429657675,
      "learning_rate": 4.925209038906448e-06,
      "loss": 0.4803,
      "step": 10021
    },
    {
      "epoch": 0.5408234849711294,
      "grad_norm": 0.9755478536013618,
      "learning_rate": 4.9244524033477835e-06,
      "loss": 0.4547,
      "step": 10022
    },
    {
      "epoch": 0.5408774485996438,
      "grad_norm": 0.8774215344013134,
      "learning_rate": 4.923695784330938e-06,
      "loss": 0.3665,
      "step": 10023
    },
    {
      "epoch": 0.5409314122281582,
      "grad_norm": 0.8419542736608732,
      "learning_rate": 4.92293918187766e-06,
      "loss": 0.2918,
      "step": 10024
    },
    {
      "epoch": 0.5409853758566726,
      "grad_norm": 1.0856967300985787,
      "learning_rate": 4.922182596009692e-06,
      "loss": 0.4888,
      "step": 10025
    },
    {
      "epoch": 0.541039339485187,
      "grad_norm": 0.7610645358828035,
      "learning_rate": 4.921426026748779e-06,
      "loss": 0.3412,
      "step": 10026
    },
    {
      "epoch": 0.5410933031137014,
      "grad_norm": 0.8724876682098054,
      "learning_rate": 4.920669474116668e-06,
      "loss": 0.4232,
      "step": 10027
    },
    {
      "epoch": 0.5411472667422157,
      "grad_norm": 0.9894381552941685,
      "learning_rate": 4.919912938135103e-06,
      "loss": 0.453,
      "step": 10028
    },
    {
      "epoch": 0.5412012303707301,
      "grad_norm": 1.0764689276696202,
      "learning_rate": 4.9191564188258244e-06,
      "loss": 0.6472,
      "step": 10029
    },
    {
      "epoch": 0.5412551939992445,
      "grad_norm": 1.0761294432476125,
      "learning_rate": 4.918399916210577e-06,
      "loss": 0.442,
      "step": 10030
    },
    {
      "epoch": 0.5413091576277589,
      "grad_norm": 0.9752200158593429,
      "learning_rate": 4.9176434303111045e-06,
      "loss": 0.4475,
      "step": 10031
    },
    {
      "epoch": 0.5413631212562733,
      "grad_norm": 0.9590485593894853,
      "learning_rate": 4.916886961149148e-06,
      "loss": 0.4649,
      "step": 10032
    },
    {
      "epoch": 0.5414170848847877,
      "grad_norm": 0.9209448271919018,
      "learning_rate": 4.916130508746448e-06,
      "loss": 0.3363,
      "step": 10033
    },
    {
      "epoch": 0.5414710485133021,
      "grad_norm": 1.0818282371243608,
      "learning_rate": 4.915374073124749e-06,
      "loss": 0.5428,
      "step": 10034
    },
    {
      "epoch": 0.5415250121418164,
      "grad_norm": 1.0841808133412505,
      "learning_rate": 4.914617654305791e-06,
      "loss": 0.4051,
      "step": 10035
    },
    {
      "epoch": 0.5415789757703308,
      "grad_norm": 0.8157092094833543,
      "learning_rate": 4.913861252311312e-06,
      "loss": 0.3263,
      "step": 10036
    },
    {
      "epoch": 0.5416329393988452,
      "grad_norm": 0.9894636229782733,
      "learning_rate": 4.913104867163051e-06,
      "loss": 0.4179,
      "step": 10037
    },
    {
      "epoch": 0.5416869030273596,
      "grad_norm": 0.963067478343932,
      "learning_rate": 4.91234849888275e-06,
      "loss": 0.4128,
      "step": 10038
    },
    {
      "epoch": 0.541740866655874,
      "grad_norm": 0.9331472020106104,
      "learning_rate": 4.911592147492147e-06,
      "loss": 0.385,
      "step": 10039
    },
    {
      "epoch": 0.5417948302843884,
      "grad_norm": 1.1097720578583075,
      "learning_rate": 4.910835813012981e-06,
      "loss": 0.4093,
      "step": 10040
    },
    {
      "epoch": 0.5418487939129027,
      "grad_norm": 1.1768655186314665,
      "learning_rate": 4.910079495466988e-06,
      "loss": 0.5621,
      "step": 10041
    },
    {
      "epoch": 0.541902757541417,
      "grad_norm": 1.0523722635616122,
      "learning_rate": 4.909323194875908e-06,
      "loss": 0.5737,
      "step": 10042
    },
    {
      "epoch": 0.5419567211699314,
      "grad_norm": 0.9069458967634363,
      "learning_rate": 4.9085669112614755e-06,
      "loss": 0.5352,
      "step": 10043
    },
    {
      "epoch": 0.5420106847984458,
      "grad_norm": 1.0660887694605456,
      "learning_rate": 4.9078106446454255e-06,
      "loss": 0.5132,
      "step": 10044
    },
    {
      "epoch": 0.5420646484269602,
      "grad_norm": 0.9375065848868652,
      "learning_rate": 4.907054395049499e-06,
      "loss": 0.4219,
      "step": 10045
    },
    {
      "epoch": 0.5421186120554746,
      "grad_norm": 0.9866024911446263,
      "learning_rate": 4.906298162495427e-06,
      "loss": 0.3414,
      "step": 10046
    },
    {
      "epoch": 0.542172575683989,
      "grad_norm": 0.8233499271211256,
      "learning_rate": 4.905541947004945e-06,
      "loss": 0.3012,
      "step": 10047
    },
    {
      "epoch": 0.5422265393125034,
      "grad_norm": 1.1914272914686268,
      "learning_rate": 4.9047857485997895e-06,
      "loss": 0.6139,
      "step": 10048
    },
    {
      "epoch": 0.5422805029410177,
      "grad_norm": 1.012695305364322,
      "learning_rate": 4.904029567301692e-06,
      "loss": 0.4112,
      "step": 10049
    },
    {
      "epoch": 0.5423344665695321,
      "grad_norm": 1.0713953518486306,
      "learning_rate": 4.903273403132387e-06,
      "loss": 0.4641,
      "step": 10050
    },
    {
      "epoch": 0.5423884301980465,
      "grad_norm": 1.2846421812882145,
      "learning_rate": 4.902517256113608e-06,
      "loss": 0.4693,
      "step": 10051
    },
    {
      "epoch": 0.5424423938265609,
      "grad_norm": 1.0671676153528726,
      "learning_rate": 4.9017611262670865e-06,
      "loss": 0.5072,
      "step": 10052
    },
    {
      "epoch": 0.5424963574550753,
      "grad_norm": 0.9815334947591683,
      "learning_rate": 4.901005013614553e-06,
      "loss": 0.4252,
      "step": 10053
    },
    {
      "epoch": 0.5425503210835897,
      "grad_norm": 0.9159102003711225,
      "learning_rate": 4.900248918177744e-06,
      "loss": 0.4775,
      "step": 10054
    },
    {
      "epoch": 0.5426042847121041,
      "grad_norm": 0.8925917658732192,
      "learning_rate": 4.8994928399783834e-06,
      "loss": 0.3451,
      "step": 10055
    },
    {
      "epoch": 0.5426582483406184,
      "grad_norm": 1.2067211020211048,
      "learning_rate": 4.89873677903821e-06,
      "loss": 0.6311,
      "step": 10056
    },
    {
      "epoch": 0.5427122119691328,
      "grad_norm": 1.0215505870181143,
      "learning_rate": 4.897980735378945e-06,
      "loss": 0.5236,
      "step": 10057
    },
    {
      "epoch": 0.5427661755976472,
      "grad_norm": 0.9422642332253948,
      "learning_rate": 4.897224709022324e-06,
      "loss": 0.4161,
      "step": 10058
    },
    {
      "epoch": 0.5428201392261616,
      "grad_norm": 1.268678819630988,
      "learning_rate": 4.896468699990072e-06,
      "loss": 0.4688,
      "step": 10059
    },
    {
      "epoch": 0.542874102854676,
      "grad_norm": 1.1040914323827407,
      "learning_rate": 4.895712708303922e-06,
      "loss": 0.4855,
      "step": 10060
    },
    {
      "epoch": 0.5429280664831904,
      "grad_norm": 0.9762128020518631,
      "learning_rate": 4.894956733985597e-06,
      "loss": 0.3322,
      "step": 10061
    },
    {
      "epoch": 0.5429820301117048,
      "grad_norm": 0.9195992236037852,
      "learning_rate": 4.894200777056828e-06,
      "loss": 0.3587,
      "step": 10062
    },
    {
      "epoch": 0.543035993740219,
      "grad_norm": 1.1120762401909288,
      "learning_rate": 4.893444837539342e-06,
      "loss": 0.4589,
      "step": 10063
    },
    {
      "epoch": 0.5430899573687334,
      "grad_norm": 1.1186098283803123,
      "learning_rate": 4.892688915454863e-06,
      "loss": 0.4956,
      "step": 10064
    },
    {
      "epoch": 0.5431439209972478,
      "grad_norm": 1.089940696333436,
      "learning_rate": 4.8919330108251175e-06,
      "loss": 0.5896,
      "step": 10065
    },
    {
      "epoch": 0.5431978846257622,
      "grad_norm": 0.8267140401939764,
      "learning_rate": 4.891177123671832e-06,
      "loss": 0.3524,
      "step": 10066
    },
    {
      "epoch": 0.5432518482542766,
      "grad_norm": 1.2715905733356314,
      "learning_rate": 4.890421254016731e-06,
      "loss": 0.4445,
      "step": 10067
    },
    {
      "epoch": 0.543305811882791,
      "grad_norm": 0.9587949906170379,
      "learning_rate": 4.8896654018815395e-06,
      "loss": 0.4968,
      "step": 10068
    },
    {
      "epoch": 0.5433597755113054,
      "grad_norm": 0.8571205321545999,
      "learning_rate": 4.888909567287981e-06,
      "loss": 0.4886,
      "step": 10069
    },
    {
      "epoch": 0.5434137391398197,
      "grad_norm": 1.0677616369454366,
      "learning_rate": 4.8881537502577794e-06,
      "loss": 0.5033,
      "step": 10070
    },
    {
      "epoch": 0.5434677027683341,
      "grad_norm": 1.165182158297888,
      "learning_rate": 4.8873979508126565e-06,
      "loss": 0.4358,
      "step": 10071
    },
    {
      "epoch": 0.5435216663968485,
      "grad_norm": 0.8428853274469854,
      "learning_rate": 4.886642168974335e-06,
      "loss": 0.3917,
      "step": 10072
    },
    {
      "epoch": 0.5435756300253629,
      "grad_norm": 1.1473159697484987,
      "learning_rate": 4.885886404764538e-06,
      "loss": 0.6079,
      "step": 10073
    },
    {
      "epoch": 0.5436295936538773,
      "grad_norm": 1.074672074395011,
      "learning_rate": 4.885130658204986e-06,
      "loss": 0.4893,
      "step": 10074
    },
    {
      "epoch": 0.5436835572823917,
      "grad_norm": 1.0645377659775097,
      "learning_rate": 4.8843749293174e-06,
      "loss": 0.5073,
      "step": 10075
    },
    {
      "epoch": 0.5437375209109061,
      "grad_norm": 1.1284101119197298,
      "learning_rate": 4.883619218123502e-06,
      "loss": 0.5104,
      "step": 10076
    },
    {
      "epoch": 0.5437914845394204,
      "grad_norm": 1.1978952910048142,
      "learning_rate": 4.882863524645011e-06,
      "loss": 0.6018,
      "step": 10077
    },
    {
      "epoch": 0.5438454481679348,
      "grad_norm": 1.0585317063678539,
      "learning_rate": 4.882107848903644e-06,
      "loss": 0.5555,
      "step": 10078
    },
    {
      "epoch": 0.5438994117964492,
      "grad_norm": 1.078133838785871,
      "learning_rate": 4.881352190921121e-06,
      "loss": 0.4932,
      "step": 10079
    },
    {
      "epoch": 0.5439533754249636,
      "grad_norm": 0.9953243752933106,
      "learning_rate": 4.880596550719163e-06,
      "loss": 0.4866,
      "step": 10080
    },
    {
      "epoch": 0.544007339053478,
      "grad_norm": 1.1148912906319999,
      "learning_rate": 4.879840928319484e-06,
      "loss": 0.4633,
      "step": 10081
    },
    {
      "epoch": 0.5440613026819924,
      "grad_norm": 1.1878424466476991,
      "learning_rate": 4.8790853237438065e-06,
      "loss": 0.6407,
      "step": 10082
    },
    {
      "epoch": 0.5441152663105068,
      "grad_norm": 1.2347691360368283,
      "learning_rate": 4.8783297370138415e-06,
      "loss": 0.5109,
      "step": 10083
    },
    {
      "epoch": 0.5441692299390211,
      "grad_norm": 0.9403822645295723,
      "learning_rate": 4.877574168151311e-06,
      "loss": 0.4186,
      "step": 10084
    },
    {
      "epoch": 0.5442231935675355,
      "grad_norm": 1.2653640760678184,
      "learning_rate": 4.876818617177925e-06,
      "loss": 0.7212,
      "step": 10085
    },
    {
      "epoch": 0.5442771571960499,
      "grad_norm": 1.0439933556708731,
      "learning_rate": 4.8760630841154024e-06,
      "loss": 0.4429,
      "step": 10086
    },
    {
      "epoch": 0.5443311208245643,
      "grad_norm": 1.0104414914523685,
      "learning_rate": 4.875307568985457e-06,
      "loss": 0.5066,
      "step": 10087
    },
    {
      "epoch": 0.5443850844530786,
      "grad_norm": 1.0171476113574158,
      "learning_rate": 4.874552071809803e-06,
      "loss": 0.4849,
      "step": 10088
    },
    {
      "epoch": 0.544439048081593,
      "grad_norm": 1.0187759134941643,
      "learning_rate": 4.873796592610155e-06,
      "loss": 0.435,
      "step": 10089
    },
    {
      "epoch": 0.5444930117101073,
      "grad_norm": 0.8534047055566423,
      "learning_rate": 4.873041131408226e-06,
      "loss": 0.3745,
      "step": 10090
    },
    {
      "epoch": 0.5445469753386217,
      "grad_norm": 0.7743411408548803,
      "learning_rate": 4.872285688225729e-06,
      "loss": 0.3582,
      "step": 10091
    },
    {
      "epoch": 0.5446009389671361,
      "grad_norm": 0.8456901152937784,
      "learning_rate": 4.871530263084373e-06,
      "loss": 0.3922,
      "step": 10092
    },
    {
      "epoch": 0.5446549025956505,
      "grad_norm": 1.341337948499867,
      "learning_rate": 4.8707748560058745e-06,
      "loss": 0.5813,
      "step": 10093
    },
    {
      "epoch": 0.5447088662241649,
      "grad_norm": 1.0919036356716096,
      "learning_rate": 4.87001946701194e-06,
      "loss": 0.6109,
      "step": 10094
    },
    {
      "epoch": 0.5447628298526793,
      "grad_norm": 1.0407258768248533,
      "learning_rate": 4.8692640961242855e-06,
      "loss": 0.4193,
      "step": 10095
    },
    {
      "epoch": 0.5448167934811937,
      "grad_norm": 1.3374896324397492,
      "learning_rate": 4.8685087433646165e-06,
      "loss": 0.7184,
      "step": 10096
    },
    {
      "epoch": 0.544870757109708,
      "grad_norm": 0.9436021911567432,
      "learning_rate": 4.867753408754645e-06,
      "loss": 0.5342,
      "step": 10097
    },
    {
      "epoch": 0.5449247207382224,
      "grad_norm": 0.9338648097614709,
      "learning_rate": 4.866998092316078e-06,
      "loss": 0.4535,
      "step": 10098
    },
    {
      "epoch": 0.5449786843667368,
      "grad_norm": 1.1797188905490101,
      "learning_rate": 4.8662427940706266e-06,
      "loss": 0.5205,
      "step": 10099
    },
    {
      "epoch": 0.5450326479952512,
      "grad_norm": 0.904833413008134,
      "learning_rate": 4.865487514039997e-06,
      "loss": 0.4053,
      "step": 10100
    },
    {
      "epoch": 0.5450866116237656,
      "grad_norm": 0.9812252315431945,
      "learning_rate": 4.864732252245899e-06,
      "loss": 0.5477,
      "step": 10101
    },
    {
      "epoch": 0.54514057525228,
      "grad_norm": 1.061445273418434,
      "learning_rate": 4.8639770087100355e-06,
      "loss": 0.467,
      "step": 10102
    },
    {
      "epoch": 0.5451945388807944,
      "grad_norm": 1.023094589972531,
      "learning_rate": 4.8632217834541175e-06,
      "loss": 0.5108,
      "step": 10103
    },
    {
      "epoch": 0.5452485025093087,
      "grad_norm": 0.8636095479759734,
      "learning_rate": 4.8624665764998506e-06,
      "loss": 0.3861,
      "step": 10104
    },
    {
      "epoch": 0.5453024661378231,
      "grad_norm": 0.9951955658888558,
      "learning_rate": 4.861711387868935e-06,
      "loss": 0.4629,
      "step": 10105
    },
    {
      "epoch": 0.5453564297663375,
      "grad_norm": 1.1172705293650023,
      "learning_rate": 4.860956217583079e-06,
      "loss": 0.5238,
      "step": 10106
    },
    {
      "epoch": 0.5454103933948519,
      "grad_norm": 1.113965900086544,
      "learning_rate": 4.860201065663989e-06,
      "loss": 0.4558,
      "step": 10107
    },
    {
      "epoch": 0.5454643570233663,
      "grad_norm": 1.1307587822442025,
      "learning_rate": 4.859445932133365e-06,
      "loss": 0.5166,
      "step": 10108
    },
    {
      "epoch": 0.5455183206518807,
      "grad_norm": 1.025323824278754,
      "learning_rate": 4.858690817012914e-06,
      "loss": 0.4964,
      "step": 10109
    },
    {
      "epoch": 0.5455722842803951,
      "grad_norm": 1.132443870729538,
      "learning_rate": 4.8579357203243345e-06,
      "loss": 0.4828,
      "step": 10110
    },
    {
      "epoch": 0.5456262479089093,
      "grad_norm": 0.9694021887443264,
      "learning_rate": 4.857180642089334e-06,
      "loss": 0.4107,
      "step": 10111
    },
    {
      "epoch": 0.5456802115374237,
      "grad_norm": 1.0530655278990793,
      "learning_rate": 4.856425582329611e-06,
      "loss": 0.4334,
      "step": 10112
    },
    {
      "epoch": 0.5457341751659381,
      "grad_norm": 0.7849756342693849,
      "learning_rate": 4.855670541066866e-06,
      "loss": 0.3966,
      "step": 10113
    },
    {
      "epoch": 0.5457881387944525,
      "grad_norm": 1.2989024381280894,
      "learning_rate": 4.8549155183228005e-06,
      "loss": 0.5991,
      "step": 10114
    },
    {
      "epoch": 0.5458421024229669,
      "grad_norm": 0.8053688271961718,
      "learning_rate": 4.854160514119114e-06,
      "loss": 0.3575,
      "step": 10115
    },
    {
      "epoch": 0.5458960660514813,
      "grad_norm": 0.9720315078822583,
      "learning_rate": 4.853405528477509e-06,
      "loss": 0.5081,
      "step": 10116
    },
    {
      "epoch": 0.5459500296799957,
      "grad_norm": 1.0142142787212018,
      "learning_rate": 4.852650561419681e-06,
      "loss": 0.5354,
      "step": 10117
    },
    {
      "epoch": 0.54600399330851,
      "grad_norm": 0.9367326978498313,
      "learning_rate": 4.851895612967331e-06,
      "loss": 0.4371,
      "step": 10118
    },
    {
      "epoch": 0.5460579569370244,
      "grad_norm": 0.9608228274296237,
      "learning_rate": 4.8511406831421545e-06,
      "loss": 0.4488,
      "step": 10119
    },
    {
      "epoch": 0.5461119205655388,
      "grad_norm": 1.0183371551362592,
      "learning_rate": 4.850385771965851e-06,
      "loss": 0.3782,
      "step": 10120
    },
    {
      "epoch": 0.5461658841940532,
      "grad_norm": 0.8492904538225049,
      "learning_rate": 4.849630879460118e-06,
      "loss": 0.3618,
      "step": 10121
    },
    {
      "epoch": 0.5462198478225676,
      "grad_norm": 1.0613187820146008,
      "learning_rate": 4.84887600564665e-06,
      "loss": 0.5298,
      "step": 10122
    },
    {
      "epoch": 0.546273811451082,
      "grad_norm": 0.7135037531631587,
      "learning_rate": 4.848121150547143e-06,
      "loss": 0.2906,
      "step": 10123
    },
    {
      "epoch": 0.5463277750795964,
      "grad_norm": 1.1071287555377285,
      "learning_rate": 4.847366314183293e-06,
      "loss": 0.5518,
      "step": 10124
    },
    {
      "epoch": 0.5463817387081107,
      "grad_norm": 0.8476278089783764,
      "learning_rate": 4.846611496576794e-06,
      "loss": 0.409,
      "step": 10125
    },
    {
      "epoch": 0.5464357023366251,
      "grad_norm": 1.1916891455276686,
      "learning_rate": 4.845856697749342e-06,
      "loss": 0.5251,
      "step": 10126
    },
    {
      "epoch": 0.5464896659651395,
      "grad_norm": 1.2993961793037963,
      "learning_rate": 4.845101917722629e-06,
      "loss": 0.7282,
      "step": 10127
    },
    {
      "epoch": 0.5465436295936539,
      "grad_norm": 0.9134930556393765,
      "learning_rate": 4.844347156518348e-06,
      "loss": 0.5476,
      "step": 10128
    },
    {
      "epoch": 0.5465975932221683,
      "grad_norm": 0.8945965089679371,
      "learning_rate": 4.843592414158192e-06,
      "loss": 0.4168,
      "step": 10129
    },
    {
      "epoch": 0.5466515568506827,
      "grad_norm": 1.0023803035357055,
      "learning_rate": 4.842837690663854e-06,
      "loss": 0.5185,
      "step": 10130
    },
    {
      "epoch": 0.5467055204791971,
      "grad_norm": 0.9875160783261898,
      "learning_rate": 4.842082986057025e-06,
      "loss": 0.4236,
      "step": 10131
    },
    {
      "epoch": 0.5467594841077114,
      "grad_norm": 1.1311747593268053,
      "learning_rate": 4.841328300359397e-06,
      "loss": 0.75,
      "step": 10132
    },
    {
      "epoch": 0.5468134477362258,
      "grad_norm": 0.85962438670122,
      "learning_rate": 4.8405736335926565e-06,
      "loss": 0.3594,
      "step": 10133
    },
    {
      "epoch": 0.5468674113647402,
      "grad_norm": 0.9775138055976144,
      "learning_rate": 4.839818985778497e-06,
      "loss": 0.4128,
      "step": 10134
    },
    {
      "epoch": 0.5469213749932545,
      "grad_norm": 1.0053394860653033,
      "learning_rate": 4.839064356938607e-06,
      "loss": 0.3958,
      "step": 10135
    },
    {
      "epoch": 0.546975338621769,
      "grad_norm": 0.9765038287632535,
      "learning_rate": 4.838309747094676e-06,
      "loss": 0.4358,
      "step": 10136
    },
    {
      "epoch": 0.5470293022502833,
      "grad_norm": 1.207070199257001,
      "learning_rate": 4.837555156268391e-06,
      "loss": 0.507,
      "step": 10137
    },
    {
      "epoch": 0.5470832658787977,
      "grad_norm": 1.0279505109572353,
      "learning_rate": 4.8368005844814396e-06,
      "loss": 0.4798,
      "step": 10138
    },
    {
      "epoch": 0.547137229507312,
      "grad_norm": 1.0312270395773961,
      "learning_rate": 4.836046031755512e-06,
      "loss": 0.3675,
      "step": 10139
    },
    {
      "epoch": 0.5471911931358264,
      "grad_norm": 1.0706055442087286,
      "learning_rate": 4.835291498112289e-06,
      "loss": 0.6133,
      "step": 10140
    },
    {
      "epoch": 0.5472451567643408,
      "grad_norm": 1.0325200161317947,
      "learning_rate": 4.834536983573461e-06,
      "loss": 0.5421,
      "step": 10141
    },
    {
      "epoch": 0.5472991203928552,
      "grad_norm": 1.0647905565242464,
      "learning_rate": 4.833782488160712e-06,
      "loss": 0.4928,
      "step": 10142
    },
    {
      "epoch": 0.5473530840213696,
      "grad_norm": 0.9017628764961584,
      "learning_rate": 4.833028011895728e-06,
      "loss": 0.3699,
      "step": 10143
    },
    {
      "epoch": 0.547407047649884,
      "grad_norm": 1.0990059499606692,
      "learning_rate": 4.832273554800193e-06,
      "loss": 0.6716,
      "step": 10144
    },
    {
      "epoch": 0.5474610112783984,
      "grad_norm": 1.0550403869627913,
      "learning_rate": 4.831519116895791e-06,
      "loss": 0.5123,
      "step": 10145
    },
    {
      "epoch": 0.5475149749069127,
      "grad_norm": 1.1458203034689012,
      "learning_rate": 4.830764698204205e-06,
      "loss": 0.5986,
      "step": 10146
    },
    {
      "epoch": 0.5475689385354271,
      "grad_norm": 0.7503629029605383,
      "learning_rate": 4.830010298747117e-06,
      "loss": 0.3477,
      "step": 10147
    },
    {
      "epoch": 0.5476229021639415,
      "grad_norm": 1.1423138266253003,
      "learning_rate": 4.829255918546211e-06,
      "loss": 0.6131,
      "step": 10148
    },
    {
      "epoch": 0.5476768657924559,
      "grad_norm": 1.0922818105317555,
      "learning_rate": 4.8285015576231664e-06,
      "loss": 0.5069,
      "step": 10149
    },
    {
      "epoch": 0.5477308294209703,
      "grad_norm": 1.0559832041656216,
      "learning_rate": 4.827747215999666e-06,
      "loss": 0.439,
      "step": 10150
    },
    {
      "epoch": 0.5477847930494847,
      "grad_norm": 1.060495737339165,
      "learning_rate": 4.826992893697391e-06,
      "loss": 0.491,
      "step": 10151
    },
    {
      "epoch": 0.5478387566779991,
      "grad_norm": 0.908339879116432,
      "learning_rate": 4.826238590738019e-06,
      "loss": 0.3903,
      "step": 10152
    },
    {
      "epoch": 0.5478927203065134,
      "grad_norm": 1.0751515849450888,
      "learning_rate": 4.825484307143232e-06,
      "loss": 0.4713,
      "step": 10153
    },
    {
      "epoch": 0.5479466839350278,
      "grad_norm": 1.1502604159284042,
      "learning_rate": 4.824730042934707e-06,
      "loss": 0.6197,
      "step": 10154
    },
    {
      "epoch": 0.5480006475635422,
      "grad_norm": 0.7076065677351586,
      "learning_rate": 4.823975798134123e-06,
      "loss": 0.3027,
      "step": 10155
    },
    {
      "epoch": 0.5480546111920566,
      "grad_norm": 0.838870099653771,
      "learning_rate": 4.823221572763156e-06,
      "loss": 0.3826,
      "step": 10156
    },
    {
      "epoch": 0.548108574820571,
      "grad_norm": 1.031514280937807,
      "learning_rate": 4.822467366843486e-06,
      "loss": 0.4521,
      "step": 10157
    },
    {
      "epoch": 0.5481625384490854,
      "grad_norm": 0.6882318817625689,
      "learning_rate": 4.821713180396789e-06,
      "loss": 0.2865,
      "step": 10158
    },
    {
      "epoch": 0.5482165020775996,
      "grad_norm": 0.8819565570025678,
      "learning_rate": 4.820959013444742e-06,
      "loss": 0.3892,
      "step": 10159
    },
    {
      "epoch": 0.548270465706114,
      "grad_norm": 1.1650488152540242,
      "learning_rate": 4.820204866009016e-06,
      "loss": 0.5201,
      "step": 10160
    },
    {
      "epoch": 0.5483244293346284,
      "grad_norm": 0.9043989842634114,
      "learning_rate": 4.81945073811129e-06,
      "loss": 0.3352,
      "step": 10161
    },
    {
      "epoch": 0.5483783929631428,
      "grad_norm": 1.1341756484848076,
      "learning_rate": 4.818696629773238e-06,
      "loss": 0.4933,
      "step": 10162
    },
    {
      "epoch": 0.5484323565916572,
      "grad_norm": 1.048289506404977,
      "learning_rate": 4.8179425410165325e-06,
      "loss": 0.3982,
      "step": 10163
    },
    {
      "epoch": 0.5484863202201716,
      "grad_norm": 0.9961651137355694,
      "learning_rate": 4.817188471862847e-06,
      "loss": 0.4461,
      "step": 10164
    },
    {
      "epoch": 0.548540283848686,
      "grad_norm": 1.0685822700024377,
      "learning_rate": 4.816434422333855e-06,
      "loss": 0.5126,
      "step": 10165
    },
    {
      "epoch": 0.5485942474772003,
      "grad_norm": 1.2861509159406181,
      "learning_rate": 4.815680392451232e-06,
      "loss": 0.4937,
      "step": 10166
    },
    {
      "epoch": 0.5486482111057147,
      "grad_norm": 1.1959423250793557,
      "learning_rate": 4.814926382236642e-06,
      "loss": 0.5418,
      "step": 10167
    },
    {
      "epoch": 0.5487021747342291,
      "grad_norm": 1.0382135733047877,
      "learning_rate": 4.814172391711761e-06,
      "loss": 0.5095,
      "step": 10168
    },
    {
      "epoch": 0.5487561383627435,
      "grad_norm": 0.9363566732613944,
      "learning_rate": 4.813418420898258e-06,
      "loss": 0.4322,
      "step": 10169
    },
    {
      "epoch": 0.5488101019912579,
      "grad_norm": 1.084394209049667,
      "learning_rate": 4.812664469817805e-06,
      "loss": 0.5534,
      "step": 10170
    },
    {
      "epoch": 0.5488640656197723,
      "grad_norm": 0.9725164263125391,
      "learning_rate": 4.811910538492067e-06,
      "loss": 0.4819,
      "step": 10171
    },
    {
      "epoch": 0.5489180292482867,
      "grad_norm": 0.8808300563152796,
      "learning_rate": 4.811156626942717e-06,
      "loss": 0.4685,
      "step": 10172
    },
    {
      "epoch": 0.548971992876801,
      "grad_norm": 1.225782638417651,
      "learning_rate": 4.810402735191422e-06,
      "loss": 0.5322,
      "step": 10173
    },
    {
      "epoch": 0.5490259565053154,
      "grad_norm": 0.9346113816031715,
      "learning_rate": 4.809648863259848e-06,
      "loss": 0.3837,
      "step": 10174
    },
    {
      "epoch": 0.5490799201338298,
      "grad_norm": 1.1201767342807656,
      "learning_rate": 4.8088950111696645e-06,
      "loss": 0.489,
      "step": 10175
    },
    {
      "epoch": 0.5491338837623442,
      "grad_norm": 0.9654303071116841,
      "learning_rate": 4.808141178942536e-06,
      "loss": 0.533,
      "step": 10176
    },
    {
      "epoch": 0.5491878473908586,
      "grad_norm": 1.0952241542248298,
      "learning_rate": 4.807387366600128e-06,
      "loss": 0.4848,
      "step": 10177
    },
    {
      "epoch": 0.549241811019373,
      "grad_norm": 1.0182050519390895,
      "learning_rate": 4.806633574164108e-06,
      "loss": 0.3627,
      "step": 10178
    },
    {
      "epoch": 0.5492957746478874,
      "grad_norm": 1.1924516726252343,
      "learning_rate": 4.80587980165614e-06,
      "loss": 0.5689,
      "step": 10179
    },
    {
      "epoch": 0.5493497382764017,
      "grad_norm": 1.0082750116228008,
      "learning_rate": 4.805126049097889e-06,
      "loss": 0.5119,
      "step": 10180
    },
    {
      "epoch": 0.549403701904916,
      "grad_norm": 1.1573479790102197,
      "learning_rate": 4.804372316511016e-06,
      "loss": 0.5091,
      "step": 10181
    },
    {
      "epoch": 0.5494576655334305,
      "grad_norm": 0.8059179467705652,
      "learning_rate": 4.803618603917185e-06,
      "loss": 0.3048,
      "step": 10182
    },
    {
      "epoch": 0.5495116291619448,
      "grad_norm": 0.9296067892523999,
      "learning_rate": 4.802864911338059e-06,
      "loss": 0.4063,
      "step": 10183
    },
    {
      "epoch": 0.5495655927904592,
      "grad_norm": 1.3034874442404973,
      "learning_rate": 4.802111238795301e-06,
      "loss": 0.7265,
      "step": 10184
    },
    {
      "epoch": 0.5496195564189736,
      "grad_norm": 1.2165091147154479,
      "learning_rate": 4.801357586310568e-06,
      "loss": 0.6054,
      "step": 10185
    },
    {
      "epoch": 0.549673520047488,
      "grad_norm": 0.9001674583197556,
      "learning_rate": 4.800603953905527e-06,
      "loss": 0.3899,
      "step": 10186
    },
    {
      "epoch": 0.5497274836760023,
      "grad_norm": 0.9694670468378022,
      "learning_rate": 4.799850341601837e-06,
      "loss": 0.4123,
      "step": 10187
    },
    {
      "epoch": 0.5497814473045167,
      "grad_norm": 0.8711597879727461,
      "learning_rate": 4.799096749421153e-06,
      "loss": 0.3831,
      "step": 10188
    },
    {
      "epoch": 0.5498354109330311,
      "grad_norm": 1.0837823283548245,
      "learning_rate": 4.798343177385136e-06,
      "loss": 0.5189,
      "step": 10189
    },
    {
      "epoch": 0.5498893745615455,
      "grad_norm": 1.2699595327442252,
      "learning_rate": 4.797589625515445e-06,
      "loss": 0.5129,
      "step": 10190
    },
    {
      "epoch": 0.5499433381900599,
      "grad_norm": 0.8767937280574267,
      "learning_rate": 4.796836093833739e-06,
      "loss": 0.4932,
      "step": 10191
    },
    {
      "epoch": 0.5499973018185743,
      "grad_norm": 1.0644126146759445,
      "learning_rate": 4.796082582361673e-06,
      "loss": 0.5038,
      "step": 10192
    },
    {
      "epoch": 0.5500512654470887,
      "grad_norm": 1.1863097674946566,
      "learning_rate": 4.795329091120906e-06,
      "loss": 0.5607,
      "step": 10193
    },
    {
      "epoch": 0.550105229075603,
      "grad_norm": 0.9209684565015405,
      "learning_rate": 4.7945756201330925e-06,
      "loss": 0.4339,
      "step": 10194
    },
    {
      "epoch": 0.5501591927041174,
      "grad_norm": 1.0434422441652182,
      "learning_rate": 4.793822169419889e-06,
      "loss": 0.5297,
      "step": 10195
    },
    {
      "epoch": 0.5502131563326318,
      "grad_norm": 0.935853174121805,
      "learning_rate": 4.793068739002949e-06,
      "loss": 0.4205,
      "step": 10196
    },
    {
      "epoch": 0.5502671199611462,
      "grad_norm": 1.311141643473365,
      "learning_rate": 4.792315328903929e-06,
      "loss": 0.5382,
      "step": 10197
    },
    {
      "epoch": 0.5503210835896606,
      "grad_norm": 1.1209685771968527,
      "learning_rate": 4.791561939144481e-06,
      "loss": 0.558,
      "step": 10198
    },
    {
      "epoch": 0.550375047218175,
      "grad_norm": 0.9997063853539551,
      "learning_rate": 4.790808569746257e-06,
      "loss": 0.5262,
      "step": 10199
    },
    {
      "epoch": 0.5504290108466894,
      "grad_norm": 1.0263681881017197,
      "learning_rate": 4.790055220730912e-06,
      "loss": 0.4061,
      "step": 10200
    },
    {
      "epoch": 0.5504829744752037,
      "grad_norm": 0.796017157360773,
      "learning_rate": 4.7893018921201e-06,
      "loss": 0.4078,
      "step": 10201
    },
    {
      "epoch": 0.5505369381037181,
      "grad_norm": 1.1706118447440286,
      "learning_rate": 4.788548583935467e-06,
      "loss": 0.5886,
      "step": 10202
    },
    {
      "epoch": 0.5505909017322325,
      "grad_norm": 1.4474117792966905,
      "learning_rate": 4.787795296198667e-06,
      "loss": 0.5283,
      "step": 10203
    },
    {
      "epoch": 0.5506448653607469,
      "grad_norm": 1.0736548543631448,
      "learning_rate": 4.7870420289313505e-06,
      "loss": 0.5232,
      "step": 10204
    },
    {
      "epoch": 0.5506988289892613,
      "grad_norm": 1.0092395427755219,
      "learning_rate": 4.786288782155165e-06,
      "loss": 0.4274,
      "step": 10205
    },
    {
      "epoch": 0.5507527926177757,
      "grad_norm": 1.0551846437496304,
      "learning_rate": 4.785535555891763e-06,
      "loss": 0.6667,
      "step": 10206
    },
    {
      "epoch": 0.55080675624629,
      "grad_norm": 0.9530261154188043,
      "learning_rate": 4.784782350162789e-06,
      "loss": 0.3778,
      "step": 10207
    },
    {
      "epoch": 0.5508607198748043,
      "grad_norm": 1.0155369939974788,
      "learning_rate": 4.784029164989896e-06,
      "loss": 0.567,
      "step": 10208
    },
    {
      "epoch": 0.5509146835033187,
      "grad_norm": 0.9374044281201305,
      "learning_rate": 4.783276000394726e-06,
      "loss": 0.3894,
      "step": 10209
    },
    {
      "epoch": 0.5509686471318331,
      "grad_norm": 1.1416842935611224,
      "learning_rate": 4.782522856398928e-06,
      "loss": 0.5874,
      "step": 10210
    },
    {
      "epoch": 0.5510226107603475,
      "grad_norm": 0.7849281439782066,
      "learning_rate": 4.781769733024149e-06,
      "loss": 0.277,
      "step": 10211
    },
    {
      "epoch": 0.5510765743888619,
      "grad_norm": 0.9982174028871303,
      "learning_rate": 4.781016630292032e-06,
      "loss": 0.4531,
      "step": 10212
    },
    {
      "epoch": 0.5511305380173763,
      "grad_norm": 0.7533142949942875,
      "learning_rate": 4.780263548224224e-06,
      "loss": 0.395,
      "step": 10213
    },
    {
      "epoch": 0.5511845016458907,
      "grad_norm": 0.9525022370103455,
      "learning_rate": 4.779510486842369e-06,
      "loss": 0.3992,
      "step": 10214
    },
    {
      "epoch": 0.551238465274405,
      "grad_norm": 1.1182411432279644,
      "learning_rate": 4.778757446168113e-06,
      "loss": 0.5592,
      "step": 10215
    },
    {
      "epoch": 0.5512924289029194,
      "grad_norm": 1.0255616737499014,
      "learning_rate": 4.778004426223094e-06,
      "loss": 0.5167,
      "step": 10216
    },
    {
      "epoch": 0.5513463925314338,
      "grad_norm": 1.0626322283527228,
      "learning_rate": 4.7772514270289584e-06,
      "loss": 0.4521,
      "step": 10217
    },
    {
      "epoch": 0.5514003561599482,
      "grad_norm": 1.0257338349132643,
      "learning_rate": 4.7764984486073466e-06,
      "loss": 0.6385,
      "step": 10218
    },
    {
      "epoch": 0.5514543197884626,
      "grad_norm": 0.8545786876518366,
      "learning_rate": 4.7757454909799e-06,
      "loss": 0.3482,
      "step": 10219
    },
    {
      "epoch": 0.551508283416977,
      "grad_norm": 0.9154630045847885,
      "learning_rate": 4.77499255416826e-06,
      "loss": 0.4659,
      "step": 10220
    },
    {
      "epoch": 0.5515622470454914,
      "grad_norm": 1.092531444540202,
      "learning_rate": 4.774239638194066e-06,
      "loss": 0.5415,
      "step": 10221
    },
    {
      "epoch": 0.5516162106740057,
      "grad_norm": 0.9256002932522819,
      "learning_rate": 4.77348674307896e-06,
      "loss": 0.3517,
      "step": 10222
    },
    {
      "epoch": 0.5516701743025201,
      "grad_norm": 0.9338821816473993,
      "learning_rate": 4.772733868844578e-06,
      "loss": 0.4667,
      "step": 10223
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.9630922760303035,
      "learning_rate": 4.771981015512559e-06,
      "loss": 0.4452,
      "step": 10224
    },
    {
      "epoch": 0.5517781015595489,
      "grad_norm": 0.9014473940182888,
      "learning_rate": 4.771228183104541e-06,
      "loss": 0.424,
      "step": 10225
    },
    {
      "epoch": 0.5518320651880633,
      "grad_norm": 0.907884479640725,
      "learning_rate": 4.770475371642163e-06,
      "loss": 0.415,
      "step": 10226
    },
    {
      "epoch": 0.5518860288165777,
      "grad_norm": 0.9991091313920106,
      "learning_rate": 4.769722581147058e-06,
      "loss": 0.4348,
      "step": 10227
    },
    {
      "epoch": 0.551939992445092,
      "grad_norm": 0.9527270253063992,
      "learning_rate": 4.768969811640868e-06,
      "loss": 0.4083,
      "step": 10228
    },
    {
      "epoch": 0.5519939560736064,
      "grad_norm": 1.0942630076749227,
      "learning_rate": 4.7682170631452205e-06,
      "loss": 0.6173,
      "step": 10229
    },
    {
      "epoch": 0.5520479197021207,
      "grad_norm": 0.9767507647862661,
      "learning_rate": 4.767464335681754e-06,
      "loss": 0.4472,
      "step": 10230
    },
    {
      "epoch": 0.5521018833306351,
      "grad_norm": 0.9307617629281258,
      "learning_rate": 4.766711629272104e-06,
      "loss": 0.4253,
      "step": 10231
    },
    {
      "epoch": 0.5521558469591495,
      "grad_norm": 1.0363925811977672,
      "learning_rate": 4.765958943937901e-06,
      "loss": 0.577,
      "step": 10232
    },
    {
      "epoch": 0.5522098105876639,
      "grad_norm": 0.894798172195993,
      "learning_rate": 4.765206279700782e-06,
      "loss": 0.3854,
      "step": 10233
    },
    {
      "epoch": 0.5522637742161783,
      "grad_norm": 1.1636358822625577,
      "learning_rate": 4.764453636582375e-06,
      "loss": 0.6263,
      "step": 10234
    },
    {
      "epoch": 0.5523177378446926,
      "grad_norm": 0.8120591636313294,
      "learning_rate": 4.763701014604316e-06,
      "loss": 0.4309,
      "step": 10235
    },
    {
      "epoch": 0.552371701473207,
      "grad_norm": 0.7925082482907375,
      "learning_rate": 4.762948413788232e-06,
      "loss": 0.3227,
      "step": 10236
    },
    {
      "epoch": 0.5524256651017214,
      "grad_norm": 1.0425905580305348,
      "learning_rate": 4.762195834155755e-06,
      "loss": 0.4828,
      "step": 10237
    },
    {
      "epoch": 0.5524796287302358,
      "grad_norm": 1.222189188197597,
      "learning_rate": 4.7614432757285155e-06,
      "loss": 0.6606,
      "step": 10238
    },
    {
      "epoch": 0.5525335923587502,
      "grad_norm": 0.8679516864558767,
      "learning_rate": 4.760690738528143e-06,
      "loss": 0.4159,
      "step": 10239
    },
    {
      "epoch": 0.5525875559872646,
      "grad_norm": 1.002349065062371,
      "learning_rate": 4.759938222576264e-06,
      "loss": 0.4983,
      "step": 10240
    },
    {
      "epoch": 0.552641519615779,
      "grad_norm": 1.1123171888072971,
      "learning_rate": 4.7591857278945105e-06,
      "loss": 0.4693,
      "step": 10241
    },
    {
      "epoch": 0.5526954832442933,
      "grad_norm": 1.0877396510968529,
      "learning_rate": 4.7584332545045095e-06,
      "loss": 0.5048,
      "step": 10242
    },
    {
      "epoch": 0.5527494468728077,
      "grad_norm": 1.2604156914624927,
      "learning_rate": 4.757680802427883e-06,
      "loss": 0.6495,
      "step": 10243
    },
    {
      "epoch": 0.5528034105013221,
      "grad_norm": 1.1552629241205457,
      "learning_rate": 4.756928371686262e-06,
      "loss": 0.4867,
      "step": 10244
    },
    {
      "epoch": 0.5528573741298365,
      "grad_norm": 1.1070536703481821,
      "learning_rate": 4.756175962301267e-06,
      "loss": 0.4325,
      "step": 10245
    },
    {
      "epoch": 0.5529113377583509,
      "grad_norm": 1.058587355404886,
      "learning_rate": 4.755423574294529e-06,
      "loss": 0.4808,
      "step": 10246
    },
    {
      "epoch": 0.5529653013868653,
      "grad_norm": 1.0021605741817001,
      "learning_rate": 4.75467120768767e-06,
      "loss": 0.482,
      "step": 10247
    },
    {
      "epoch": 0.5530192650153797,
      "grad_norm": 1.3458160113344066,
      "learning_rate": 4.753918862502312e-06,
      "loss": 0.8167,
      "step": 10248
    },
    {
      "epoch": 0.553073228643894,
      "grad_norm": 0.8343208415239175,
      "learning_rate": 4.7531665387600825e-06,
      "loss": 0.4041,
      "step": 10249
    },
    {
      "epoch": 0.5531271922724084,
      "grad_norm": 1.052267778054434,
      "learning_rate": 4.752414236482599e-06,
      "loss": 0.546,
      "step": 10250
    },
    {
      "epoch": 0.5531811559009228,
      "grad_norm": 0.9773511789038181,
      "learning_rate": 4.7516619556914854e-06,
      "loss": 0.5469,
      "step": 10251
    },
    {
      "epoch": 0.5532351195294372,
      "grad_norm": 1.0166159145370344,
      "learning_rate": 4.750909696408365e-06,
      "loss": 0.3872,
      "step": 10252
    },
    {
      "epoch": 0.5532890831579516,
      "grad_norm": 0.9896540224926904,
      "learning_rate": 4.750157458654856e-06,
      "loss": 0.4426,
      "step": 10253
    },
    {
      "epoch": 0.553343046786466,
      "grad_norm": 1.4930105902271418,
      "learning_rate": 4.749405242452578e-06,
      "loss": 0.5779,
      "step": 10254
    },
    {
      "epoch": 0.5533970104149804,
      "grad_norm": 0.8990369041479075,
      "learning_rate": 4.748653047823154e-06,
      "loss": 0.4047,
      "step": 10255
    },
    {
      "epoch": 0.5534509740434946,
      "grad_norm": 1.1535149741319002,
      "learning_rate": 4.747900874788202e-06,
      "loss": 0.4823,
      "step": 10256
    },
    {
      "epoch": 0.553504937672009,
      "grad_norm": 0.8703006694703891,
      "learning_rate": 4.747148723369336e-06,
      "loss": 0.4143,
      "step": 10257
    },
    {
      "epoch": 0.5535589013005234,
      "grad_norm": 1.0574048226244162,
      "learning_rate": 4.746396593588176e-06,
      "loss": 0.4691,
      "step": 10258
    },
    {
      "epoch": 0.5536128649290378,
      "grad_norm": 1.1203648416963612,
      "learning_rate": 4.745644485466341e-06,
      "loss": 0.549,
      "step": 10259
    },
    {
      "epoch": 0.5536668285575522,
      "grad_norm": 1.0940808620306717,
      "learning_rate": 4.744892399025445e-06,
      "loss": 0.5694,
      "step": 10260
    },
    {
      "epoch": 0.5537207921860666,
      "grad_norm": 1.1395646913357078,
      "learning_rate": 4.744140334287104e-06,
      "loss": 0.4712,
      "step": 10261
    },
    {
      "epoch": 0.553774755814581,
      "grad_norm": 0.9147035778624767,
      "learning_rate": 4.743388291272934e-06,
      "loss": 0.4687,
      "step": 10262
    },
    {
      "epoch": 0.5538287194430953,
      "grad_norm": 0.9686000647321907,
      "learning_rate": 4.74263627000455e-06,
      "loss": 0.5505,
      "step": 10263
    },
    {
      "epoch": 0.5538826830716097,
      "grad_norm": 0.9784204950372776,
      "learning_rate": 4.741884270503564e-06,
      "loss": 0.4249,
      "step": 10264
    },
    {
      "epoch": 0.5539366467001241,
      "grad_norm": 0.8685226968130856,
      "learning_rate": 4.74113229279159e-06,
      "loss": 0.423,
      "step": 10265
    },
    {
      "epoch": 0.5539906103286385,
      "grad_norm": 0.8385328130206176,
      "learning_rate": 4.74038033689024e-06,
      "loss": 0.4223,
      "step": 10266
    },
    {
      "epoch": 0.5540445739571529,
      "grad_norm": 0.8986874191908752,
      "learning_rate": 4.739628402821127e-06,
      "loss": 0.3658,
      "step": 10267
    },
    {
      "epoch": 0.5540985375856673,
      "grad_norm": 0.9513013969724722,
      "learning_rate": 4.738876490605861e-06,
      "loss": 0.5437,
      "step": 10268
    },
    {
      "epoch": 0.5541525012141817,
      "grad_norm": 1.0610797976526958,
      "learning_rate": 4.738124600266056e-06,
      "loss": 0.5378,
      "step": 10269
    },
    {
      "epoch": 0.554206464842696,
      "grad_norm": 0.8899970332221802,
      "learning_rate": 4.7373727318233194e-06,
      "loss": 0.4063,
      "step": 10270
    },
    {
      "epoch": 0.5542604284712104,
      "grad_norm": 0.8809406767453768,
      "learning_rate": 4.73662088529926e-06,
      "loss": 0.3815,
      "step": 10271
    },
    {
      "epoch": 0.5543143920997248,
      "grad_norm": 1.0927379263485222,
      "learning_rate": 4.7358690607154886e-06,
      "loss": 0.5555,
      "step": 10272
    },
    {
      "epoch": 0.5543683557282392,
      "grad_norm": 0.9590401989227892,
      "learning_rate": 4.735117258093612e-06,
      "loss": 0.503,
      "step": 10273
    },
    {
      "epoch": 0.5544223193567536,
      "grad_norm": 1.1191268165293693,
      "learning_rate": 4.7343654774552395e-06,
      "loss": 0.6532,
      "step": 10274
    },
    {
      "epoch": 0.554476282985268,
      "grad_norm": 0.9705838694303852,
      "learning_rate": 4.733613718821976e-06,
      "loss": 0.5342,
      "step": 10275
    },
    {
      "epoch": 0.5545302466137824,
      "grad_norm": 1.0288376979707352,
      "learning_rate": 4.732861982215426e-06,
      "loss": 0.4631,
      "step": 10276
    },
    {
      "epoch": 0.5545842102422966,
      "grad_norm": 1.1563276114208567,
      "learning_rate": 4.732110267657202e-06,
      "loss": 0.4684,
      "step": 10277
    },
    {
      "epoch": 0.554638173870811,
      "grad_norm": 1.1987156128250958,
      "learning_rate": 4.731358575168903e-06,
      "loss": 0.5821,
      "step": 10278
    },
    {
      "epoch": 0.5546921374993254,
      "grad_norm": 0.7402913880802539,
      "learning_rate": 4.730606904772134e-06,
      "loss": 0.3138,
      "step": 10279
    },
    {
      "epoch": 0.5547461011278398,
      "grad_norm": 1.0177408728973936,
      "learning_rate": 4.7298552564885005e-06,
      "loss": 0.5228,
      "step": 10280
    },
    {
      "epoch": 0.5548000647563542,
      "grad_norm": 1.061933210216232,
      "learning_rate": 4.729103630339605e-06,
      "loss": 0.4396,
      "step": 10281
    },
    {
      "epoch": 0.5548540283848686,
      "grad_norm": 1.066323276000388,
      "learning_rate": 4.728352026347052e-06,
      "loss": 0.4738,
      "step": 10282
    },
    {
      "epoch": 0.554907992013383,
      "grad_norm": 1.0989357653910055,
      "learning_rate": 4.72760044453244e-06,
      "loss": 0.5498,
      "step": 10283
    },
    {
      "epoch": 0.5549619556418973,
      "grad_norm": 0.9866766493055045,
      "learning_rate": 4.726848884917372e-06,
      "loss": 0.4766,
      "step": 10284
    },
    {
      "epoch": 0.5550159192704117,
      "grad_norm": 1.0429959951674592,
      "learning_rate": 4.726097347523449e-06,
      "loss": 0.5653,
      "step": 10285
    },
    {
      "epoch": 0.5550698828989261,
      "grad_norm": 0.8772269261031262,
      "learning_rate": 4.725345832372269e-06,
      "loss": 0.4621,
      "step": 10286
    },
    {
      "epoch": 0.5551238465274405,
      "grad_norm": 1.100554652342403,
      "learning_rate": 4.724594339485432e-06,
      "loss": 0.5058,
      "step": 10287
    },
    {
      "epoch": 0.5551778101559549,
      "grad_norm": 0.9390972055032538,
      "learning_rate": 4.7238428688845375e-06,
      "loss": 0.5574,
      "step": 10288
    },
    {
      "epoch": 0.5552317737844693,
      "grad_norm": 0.9432022490620339,
      "learning_rate": 4.723091420591184e-06,
      "loss": 0.4056,
      "step": 10289
    },
    {
      "epoch": 0.5552857374129837,
      "grad_norm": 1.0956778361955692,
      "learning_rate": 4.7223399946269675e-06,
      "loss": 0.46,
      "step": 10290
    },
    {
      "epoch": 0.555339701041498,
      "grad_norm": 1.0654124151829423,
      "learning_rate": 4.721588591013487e-06,
      "loss": 0.4788,
      "step": 10291
    },
    {
      "epoch": 0.5553936646700124,
      "grad_norm": 0.9235022120454096,
      "learning_rate": 4.720837209772335e-06,
      "loss": 0.3209,
      "step": 10292
    },
    {
      "epoch": 0.5554476282985268,
      "grad_norm": 0.9046613525752183,
      "learning_rate": 4.720085850925109e-06,
      "loss": 0.4432,
      "step": 10293
    },
    {
      "epoch": 0.5555015919270412,
      "grad_norm": 0.8742799378313397,
      "learning_rate": 4.719334514493405e-06,
      "loss": 0.4184,
      "step": 10294
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.9375443184693335,
      "learning_rate": 4.718583200498814e-06,
      "loss": 0.3687,
      "step": 10295
    },
    {
      "epoch": 0.55560951918407,
      "grad_norm": 1.1688956820111163,
      "learning_rate": 4.717831908962932e-06,
      "loss": 0.6131,
      "step": 10296
    },
    {
      "epoch": 0.5556634828125843,
      "grad_norm": 0.988552118354493,
      "learning_rate": 4.717080639907352e-06,
      "loss": 0.4555,
      "step": 10297
    },
    {
      "epoch": 0.5557174464410987,
      "grad_norm": 1.233525082950925,
      "learning_rate": 4.716329393353666e-06,
      "loss": 0.7664,
      "step": 10298
    },
    {
      "epoch": 0.5557714100696131,
      "grad_norm": 0.9869199192969014,
      "learning_rate": 4.715578169323463e-06,
      "loss": 0.4248,
      "step": 10299
    },
    {
      "epoch": 0.5558253736981275,
      "grad_norm": 0.9557645692277662,
      "learning_rate": 4.7148269678383376e-06,
      "loss": 0.4374,
      "step": 10300
    },
    {
      "epoch": 0.5558793373266419,
      "grad_norm": 1.058738572841535,
      "learning_rate": 4.714075788919877e-06,
      "loss": 0.4271,
      "step": 10301
    },
    {
      "epoch": 0.5559333009551563,
      "grad_norm": 1.0677287243199753,
      "learning_rate": 4.713324632589673e-06,
      "loss": 0.5769,
      "step": 10302
    },
    {
      "epoch": 0.5559872645836706,
      "grad_norm": 0.9241402289432602,
      "learning_rate": 4.712573498869315e-06,
      "loss": 0.3105,
      "step": 10303
    },
    {
      "epoch": 0.5560412282121849,
      "grad_norm": 1.059241895708833,
      "learning_rate": 4.711822387780391e-06,
      "loss": 0.4902,
      "step": 10304
    },
    {
      "epoch": 0.5560951918406993,
      "grad_norm": 0.9678383960371283,
      "learning_rate": 4.7110712993444865e-06,
      "loss": 0.5583,
      "step": 10305
    },
    {
      "epoch": 0.5561491554692137,
      "grad_norm": 0.9706934018175043,
      "learning_rate": 4.71032023358319e-06,
      "loss": 0.3878,
      "step": 10306
    },
    {
      "epoch": 0.5562031190977281,
      "grad_norm": 0.9747668616863375,
      "learning_rate": 4.709569190518088e-06,
      "loss": 0.3958,
      "step": 10307
    },
    {
      "epoch": 0.5562570827262425,
      "grad_norm": 1.1214627847961955,
      "learning_rate": 4.708818170170765e-06,
      "loss": 0.5152,
      "step": 10308
    },
    {
      "epoch": 0.5563110463547569,
      "grad_norm": 1.066287530183705,
      "learning_rate": 4.708067172562808e-06,
      "loss": 0.5932,
      "step": 10309
    },
    {
      "epoch": 0.5563650099832713,
      "grad_norm": 0.807244235892807,
      "learning_rate": 4.707316197715801e-06,
      "loss": 0.4033,
      "step": 10310
    },
    {
      "epoch": 0.5564189736117856,
      "grad_norm": 0.8484979177009955,
      "learning_rate": 4.70656524565133e-06,
      "loss": 0.3811,
      "step": 10311
    },
    {
      "epoch": 0.5564729372403,
      "grad_norm": 0.9418938308117809,
      "learning_rate": 4.705814316390972e-06,
      "loss": 0.3566,
      "step": 10312
    },
    {
      "epoch": 0.5565269008688144,
      "grad_norm": 1.0772264892751013,
      "learning_rate": 4.705063409956314e-06,
      "loss": 0.4949,
      "step": 10313
    },
    {
      "epoch": 0.5565808644973288,
      "grad_norm": 1.0399389949659381,
      "learning_rate": 4.704312526368937e-06,
      "loss": 0.4414,
      "step": 10314
    },
    {
      "epoch": 0.5566348281258432,
      "grad_norm": 1.1515561338852636,
      "learning_rate": 4.703561665650422e-06,
      "loss": 0.4481,
      "step": 10315
    },
    {
      "epoch": 0.5566887917543576,
      "grad_norm": 0.8989238641097997,
      "learning_rate": 4.702810827822348e-06,
      "loss": 0.4189,
      "step": 10316
    },
    {
      "epoch": 0.556742755382872,
      "grad_norm": 1.3655556027095446,
      "learning_rate": 4.702060012906298e-06,
      "loss": 0.4912,
      "step": 10317
    },
    {
      "epoch": 0.5567967190113863,
      "grad_norm": 0.9973029937030303,
      "learning_rate": 4.7013092209238506e-06,
      "loss": 0.4994,
      "step": 10318
    },
    {
      "epoch": 0.5568506826399007,
      "grad_norm": 1.0107992210332357,
      "learning_rate": 4.700558451896581e-06,
      "loss": 0.6377,
      "step": 10319
    },
    {
      "epoch": 0.5569046462684151,
      "grad_norm": 1.56091262859535,
      "learning_rate": 4.69980770584607e-06,
      "loss": 0.7337,
      "step": 10320
    },
    {
      "epoch": 0.5569586098969295,
      "grad_norm": 0.9575313192000734,
      "learning_rate": 4.699056982793894e-06,
      "loss": 0.4798,
      "step": 10321
    },
    {
      "epoch": 0.5570125735254439,
      "grad_norm": 0.8975024309737544,
      "learning_rate": 4.698306282761628e-06,
      "loss": 0.4212,
      "step": 10322
    },
    {
      "epoch": 0.5570665371539583,
      "grad_norm": 0.9973815460233287,
      "learning_rate": 4.697555605770851e-06,
      "loss": 0.5311,
      "step": 10323
    },
    {
      "epoch": 0.5571205007824727,
      "grad_norm": 1.034387272682623,
      "learning_rate": 4.696804951843137e-06,
      "loss": 0.5166,
      "step": 10324
    },
    {
      "epoch": 0.557174464410987,
      "grad_norm": 0.8841292317090952,
      "learning_rate": 4.696054321000061e-06,
      "loss": 0.3655,
      "step": 10325
    },
    {
      "epoch": 0.5572284280395013,
      "grad_norm": 1.0798317920732936,
      "learning_rate": 4.695303713263195e-06,
      "loss": 0.3811,
      "step": 10326
    },
    {
      "epoch": 0.5572823916680157,
      "grad_norm": 0.8429098436538636,
      "learning_rate": 4.694553128654114e-06,
      "loss": 0.3134,
      "step": 10327
    },
    {
      "epoch": 0.5573363552965301,
      "grad_norm": 1.0595349968234722,
      "learning_rate": 4.6938025671943885e-06,
      "loss": 0.6119,
      "step": 10328
    },
    {
      "epoch": 0.5573903189250445,
      "grad_norm": 1.278004580885724,
      "learning_rate": 4.693052028905593e-06,
      "loss": 0.5509,
      "step": 10329
    },
    {
      "epoch": 0.5574442825535589,
      "grad_norm": 1.056036466539982,
      "learning_rate": 4.692301513809297e-06,
      "loss": 0.4784,
      "step": 10330
    },
    {
      "epoch": 0.5574982461820733,
      "grad_norm": 1.2453560375326012,
      "learning_rate": 4.691551021927072e-06,
      "loss": 0.7517,
      "step": 10331
    },
    {
      "epoch": 0.5575522098105876,
      "grad_norm": 0.974916877890773,
      "learning_rate": 4.690800553280489e-06,
      "loss": 0.377,
      "step": 10332
    },
    {
      "epoch": 0.557606173439102,
      "grad_norm": 0.9814487984076121,
      "learning_rate": 4.690050107891114e-06,
      "loss": 0.4509,
      "step": 10333
    },
    {
      "epoch": 0.5576601370676164,
      "grad_norm": 0.9914481128189413,
      "learning_rate": 4.689299685780517e-06,
      "loss": 0.4907,
      "step": 10334
    },
    {
      "epoch": 0.5577141006961308,
      "grad_norm": 1.030210175492681,
      "learning_rate": 4.688549286970267e-06,
      "loss": 0.407,
      "step": 10335
    },
    {
      "epoch": 0.5577680643246452,
      "grad_norm": 0.8338832597687463,
      "learning_rate": 4.68779891148193e-06,
      "loss": 0.4251,
      "step": 10336
    },
    {
      "epoch": 0.5578220279531596,
      "grad_norm": 0.933504888556169,
      "learning_rate": 4.687048559337073e-06,
      "loss": 0.3399,
      "step": 10337
    },
    {
      "epoch": 0.557875991581674,
      "grad_norm": 0.6813383150552782,
      "learning_rate": 4.686298230557261e-06,
      "loss": 0.3027,
      "step": 10338
    },
    {
      "epoch": 0.5579299552101883,
      "grad_norm": 0.91641431890194,
      "learning_rate": 4.685547925164062e-06,
      "loss": 0.4568,
      "step": 10339
    },
    {
      "epoch": 0.5579839188387027,
      "grad_norm": 1.3059834266021237,
      "learning_rate": 4.684797643179039e-06,
      "loss": 0.6158,
      "step": 10340
    },
    {
      "epoch": 0.5580378824672171,
      "grad_norm": 1.2015672687602583,
      "learning_rate": 4.684047384623752e-06,
      "loss": 0.5358,
      "step": 10341
    },
    {
      "epoch": 0.5580918460957315,
      "grad_norm": 0.9299500080692359,
      "learning_rate": 4.68329714951977e-06,
      "loss": 0.433,
      "step": 10342
    },
    {
      "epoch": 0.5581458097242459,
      "grad_norm": 1.0319829592040375,
      "learning_rate": 4.682546937888653e-06,
      "loss": 0.4536,
      "step": 10343
    },
    {
      "epoch": 0.5581997733527603,
      "grad_norm": 1.1179217687877732,
      "learning_rate": 4.6817967497519626e-06,
      "loss": 0.6583,
      "step": 10344
    },
    {
      "epoch": 0.5582537369812747,
      "grad_norm": 0.9120345116987045,
      "learning_rate": 4.681046585131259e-06,
      "loss": 0.3801,
      "step": 10345
    },
    {
      "epoch": 0.558307700609789,
      "grad_norm": 0.9452728209767552,
      "learning_rate": 4.680296444048108e-06,
      "loss": 0.3677,
      "step": 10346
    },
    {
      "epoch": 0.5583616642383034,
      "grad_norm": 0.9484777119482529,
      "learning_rate": 4.679546326524062e-06,
      "loss": 0.3287,
      "step": 10347
    },
    {
      "epoch": 0.5584156278668178,
      "grad_norm": 1.153388427550987,
      "learning_rate": 4.678796232580685e-06,
      "loss": 0.5118,
      "step": 10348
    },
    {
      "epoch": 0.5584695914953322,
      "grad_norm": 0.9389343122756456,
      "learning_rate": 4.678046162239534e-06,
      "loss": 0.4184,
      "step": 10349
    },
    {
      "epoch": 0.5585235551238465,
      "grad_norm": 0.9148941316547532,
      "learning_rate": 4.677296115522165e-06,
      "loss": 0.434,
      "step": 10350
    },
    {
      "epoch": 0.558577518752361,
      "grad_norm": 0.7269979920895444,
      "learning_rate": 4.676546092450139e-06,
      "loss": 0.3065,
      "step": 10351
    },
    {
      "epoch": 0.5586314823808753,
      "grad_norm": 1.2668829189340294,
      "learning_rate": 4.675796093045008e-06,
      "loss": 0.4927,
      "step": 10352
    },
    {
      "epoch": 0.5586854460093896,
      "grad_norm": 0.9273894968222307,
      "learning_rate": 4.675046117328333e-06,
      "loss": 0.4039,
      "step": 10353
    },
    {
      "epoch": 0.558739409637904,
      "grad_norm": 0.9703100687260712,
      "learning_rate": 4.674296165321665e-06,
      "loss": 0.4668,
      "step": 10354
    },
    {
      "epoch": 0.5587933732664184,
      "grad_norm": 1.2265326607000822,
      "learning_rate": 4.673546237046559e-06,
      "loss": 0.5099,
      "step": 10355
    },
    {
      "epoch": 0.5588473368949328,
      "grad_norm": 0.9750918512898227,
      "learning_rate": 4.672796332524569e-06,
      "loss": 0.4729,
      "step": 10356
    },
    {
      "epoch": 0.5589013005234472,
      "grad_norm": 0.8957956448216362,
      "learning_rate": 4.672046451777249e-06,
      "loss": 0.4092,
      "step": 10357
    },
    {
      "epoch": 0.5589552641519616,
      "grad_norm": 1.1747097759498604,
      "learning_rate": 4.67129659482615e-06,
      "loss": 0.5837,
      "step": 10358
    },
    {
      "epoch": 0.559009227780476,
      "grad_norm": 1.0594049763871234,
      "learning_rate": 4.670546761692825e-06,
      "loss": 0.4992,
      "step": 10359
    },
    {
      "epoch": 0.5590631914089903,
      "grad_norm": 0.9353801735714632,
      "learning_rate": 4.669796952398825e-06,
      "loss": 0.3859,
      "step": 10360
    },
    {
      "epoch": 0.5591171550375047,
      "grad_norm": 0.9586907908343285,
      "learning_rate": 4.669047166965698e-06,
      "loss": 0.3863,
      "step": 10361
    },
    {
      "epoch": 0.5591711186660191,
      "grad_norm": 1.0103313890937382,
      "learning_rate": 4.668297405414994e-06,
      "loss": 0.5386,
      "step": 10362
    },
    {
      "epoch": 0.5592250822945335,
      "grad_norm": 1.1178698162789336,
      "learning_rate": 4.667547667768265e-06,
      "loss": 0.4697,
      "step": 10363
    },
    {
      "epoch": 0.5592790459230479,
      "grad_norm": 0.9780613485700601,
      "learning_rate": 4.666797954047055e-06,
      "loss": 0.4018,
      "step": 10364
    },
    {
      "epoch": 0.5593330095515623,
      "grad_norm": 1.0507323922150107,
      "learning_rate": 4.666048264272915e-06,
      "loss": 0.3662,
      "step": 10365
    },
    {
      "epoch": 0.5593869731800766,
      "grad_norm": 1.1241988512681806,
      "learning_rate": 4.665298598467393e-06,
      "loss": 0.6178,
      "step": 10366
    },
    {
      "epoch": 0.559440936808591,
      "grad_norm": 0.9354840129688691,
      "learning_rate": 4.66454895665203e-06,
      "loss": 0.4456,
      "step": 10367
    },
    {
      "epoch": 0.5594949004371054,
      "grad_norm": 1.232332675183214,
      "learning_rate": 4.663799338848374e-06,
      "loss": 0.6272,
      "step": 10368
    },
    {
      "epoch": 0.5595488640656198,
      "grad_norm": 1.0986903744621233,
      "learning_rate": 4.66304974507797e-06,
      "loss": 0.5754,
      "step": 10369
    },
    {
      "epoch": 0.5596028276941342,
      "grad_norm": 1.2114691018425239,
      "learning_rate": 4.662300175362362e-06,
      "loss": 0.5058,
      "step": 10370
    },
    {
      "epoch": 0.5596567913226486,
      "grad_norm": 1.1960355565851375,
      "learning_rate": 4.661550629723094e-06,
      "loss": 0.4867,
      "step": 10371
    },
    {
      "epoch": 0.559710754951163,
      "grad_norm": 1.0588153431375231,
      "learning_rate": 4.660801108181707e-06,
      "loss": 0.49,
      "step": 10372
    },
    {
      "epoch": 0.5597647185796772,
      "grad_norm": 0.784890716721979,
      "learning_rate": 4.660051610759746e-06,
      "loss": 0.3851,
      "step": 10373
    },
    {
      "epoch": 0.5598186822081916,
      "grad_norm": 1.126102239523937,
      "learning_rate": 4.65930213747875e-06,
      "loss": 0.5445,
      "step": 10374
    },
    {
      "epoch": 0.559872645836706,
      "grad_norm": 0.8160524589359321,
      "learning_rate": 4.6585526883602584e-06,
      "loss": 0.3581,
      "step": 10375
    },
    {
      "epoch": 0.5599266094652204,
      "grad_norm": 0.8390079957255429,
      "learning_rate": 4.657803263425813e-06,
      "loss": 0.341,
      "step": 10376
    },
    {
      "epoch": 0.5599805730937348,
      "grad_norm": 1.1101222067601637,
      "learning_rate": 4.657053862696954e-06,
      "loss": 0.5306,
      "step": 10377
    },
    {
      "epoch": 0.5600345367222492,
      "grad_norm": 1.112411984487719,
      "learning_rate": 4.656304486195219e-06,
      "loss": 0.4873,
      "step": 10378
    },
    {
      "epoch": 0.5600885003507636,
      "grad_norm": 1.2377705301373216,
      "learning_rate": 4.655555133942144e-06,
      "loss": 0.5217,
      "step": 10379
    },
    {
      "epoch": 0.5601424639792779,
      "grad_norm": 0.9428711233239776,
      "learning_rate": 4.6548058059592696e-06,
      "loss": 0.415,
      "step": 10380
    },
    {
      "epoch": 0.5601964276077923,
      "grad_norm": 0.9031629897496262,
      "learning_rate": 4.654056502268129e-06,
      "loss": 0.3823,
      "step": 10381
    },
    {
      "epoch": 0.5602503912363067,
      "grad_norm": 0.9982631272634211,
      "learning_rate": 4.653307222890259e-06,
      "loss": 0.4761,
      "step": 10382
    },
    {
      "epoch": 0.5603043548648211,
      "grad_norm": 0.9469601002023227,
      "learning_rate": 4.652557967847195e-06,
      "loss": 0.3737,
      "step": 10383
    },
    {
      "epoch": 0.5603583184933355,
      "grad_norm": 1.0020789463514839,
      "learning_rate": 4.651808737160471e-06,
      "loss": 0.5069,
      "step": 10384
    },
    {
      "epoch": 0.5604122821218499,
      "grad_norm": 1.0781346539748349,
      "learning_rate": 4.651059530851623e-06,
      "loss": 0.4856,
      "step": 10385
    },
    {
      "epoch": 0.5604662457503643,
      "grad_norm": 0.8875768184666442,
      "learning_rate": 4.650310348942179e-06,
      "loss": 0.4225,
      "step": 10386
    },
    {
      "epoch": 0.5605202093788786,
      "grad_norm": 0.8852392179681133,
      "learning_rate": 4.6495611914536775e-06,
      "loss": 0.4509,
      "step": 10387
    },
    {
      "epoch": 0.560574173007393,
      "grad_norm": 1.037107063079202,
      "learning_rate": 4.6488120584076445e-06,
      "loss": 0.5758,
      "step": 10388
    },
    {
      "epoch": 0.5606281366359074,
      "grad_norm": 1.0655736864421468,
      "learning_rate": 4.648062949825614e-06,
      "loss": 0.4682,
      "step": 10389
    },
    {
      "epoch": 0.5606821002644218,
      "grad_norm": 1.180016693349369,
      "learning_rate": 4.647313865729114e-06,
      "loss": 0.567,
      "step": 10390
    },
    {
      "epoch": 0.5607360638929362,
      "grad_norm": 1.1068824354536526,
      "learning_rate": 4.646564806139675e-06,
      "loss": 0.4522,
      "step": 10391
    },
    {
      "epoch": 0.5607900275214506,
      "grad_norm": 1.0252659796430574,
      "learning_rate": 4.645815771078827e-06,
      "loss": 0.4729,
      "step": 10392
    },
    {
      "epoch": 0.560843991149965,
      "grad_norm": 0.9864252674172859,
      "learning_rate": 4.645066760568095e-06,
      "loss": 0.4693,
      "step": 10393
    },
    {
      "epoch": 0.5608979547784793,
      "grad_norm": 0.8943367445684924,
      "learning_rate": 4.644317774629011e-06,
      "loss": 0.4325,
      "step": 10394
    },
    {
      "epoch": 0.5609519184069937,
      "grad_norm": 0.9934470974525529,
      "learning_rate": 4.643568813283097e-06,
      "loss": 0.4272,
      "step": 10395
    },
    {
      "epoch": 0.561005882035508,
      "grad_norm": 1.1938919721042003,
      "learning_rate": 4.6428198765518804e-06,
      "loss": 0.548,
      "step": 10396
    },
    {
      "epoch": 0.5610598456640224,
      "grad_norm": 0.8847724981310143,
      "learning_rate": 4.642070964456887e-06,
      "loss": 0.373,
      "step": 10397
    },
    {
      "epoch": 0.5611138092925368,
      "grad_norm": 1.0059593041145987,
      "learning_rate": 4.641322077019641e-06,
      "loss": 0.509,
      "step": 10398
    },
    {
      "epoch": 0.5611677729210512,
      "grad_norm": 1.0306981004710065,
      "learning_rate": 4.640573214261665e-06,
      "loss": 0.4168,
      "step": 10399
    },
    {
      "epoch": 0.5612217365495656,
      "grad_norm": 0.927691135238471,
      "learning_rate": 4.639824376204485e-06,
      "loss": 0.4628,
      "step": 10400
    },
    {
      "epoch": 0.5612757001780799,
      "grad_norm": 1.1861306931197144,
      "learning_rate": 4.639075562869622e-06,
      "loss": 0.488,
      "step": 10401
    },
    {
      "epoch": 0.5613296638065943,
      "grad_norm": 1.1430739716277543,
      "learning_rate": 4.638326774278595e-06,
      "loss": 0.5462,
      "step": 10402
    },
    {
      "epoch": 0.5613836274351087,
      "grad_norm": 0.9861494154601086,
      "learning_rate": 4.637578010452928e-06,
      "loss": 0.4978,
      "step": 10403
    },
    {
      "epoch": 0.5614375910636231,
      "grad_norm": 0.9051246520748504,
      "learning_rate": 4.636829271414141e-06,
      "loss": 0.375,
      "step": 10404
    },
    {
      "epoch": 0.5614915546921375,
      "grad_norm": 0.9659828462594624,
      "learning_rate": 4.636080557183752e-06,
      "loss": 0.5012,
      "step": 10405
    },
    {
      "epoch": 0.5615455183206519,
      "grad_norm": 1.0747341826820602,
      "learning_rate": 4.63533186778328e-06,
      "loss": 0.376,
      "step": 10406
    },
    {
      "epoch": 0.5615994819491663,
      "grad_norm": 0.9398439520446701,
      "learning_rate": 4.634583203234245e-06,
      "loss": 0.4071,
      "step": 10407
    },
    {
      "epoch": 0.5616534455776806,
      "grad_norm": 1.1323527955155106,
      "learning_rate": 4.6338345635581665e-06,
      "loss": 0.4492,
      "step": 10408
    },
    {
      "epoch": 0.561707409206195,
      "grad_norm": 0.987406525424712,
      "learning_rate": 4.633085948776555e-06,
      "loss": 0.5324,
      "step": 10409
    },
    {
      "epoch": 0.5617613728347094,
      "grad_norm": 0.956725352837083,
      "learning_rate": 4.632337358910928e-06,
      "loss": 0.4319,
      "step": 10410
    },
    {
      "epoch": 0.5618153364632238,
      "grad_norm": 1.0033426839608848,
      "learning_rate": 4.631588793982802e-06,
      "loss": 0.4015,
      "step": 10411
    },
    {
      "epoch": 0.5618693000917382,
      "grad_norm": 1.0007141754290345,
      "learning_rate": 4.630840254013692e-06,
      "loss": 0.5036,
      "step": 10412
    },
    {
      "epoch": 0.5619232637202526,
      "grad_norm": 1.012303998099746,
      "learning_rate": 4.630091739025112e-06,
      "loss": 0.462,
      "step": 10413
    },
    {
      "epoch": 0.561977227348767,
      "grad_norm": 1.0860760668900513,
      "learning_rate": 4.629343249038573e-06,
      "loss": 0.6,
      "step": 10414
    },
    {
      "epoch": 0.5620311909772813,
      "grad_norm": 1.235164221337869,
      "learning_rate": 4.6285947840755915e-06,
      "loss": 0.446,
      "step": 10415
    },
    {
      "epoch": 0.5620851546057957,
      "grad_norm": 1.1982023044453771,
      "learning_rate": 4.6278463441576756e-06,
      "loss": 0.5431,
      "step": 10416
    },
    {
      "epoch": 0.5621391182343101,
      "grad_norm": 1.0865833190933252,
      "learning_rate": 4.627097929306336e-06,
      "loss": 0.5236,
      "step": 10417
    },
    {
      "epoch": 0.5621930818628245,
      "grad_norm": 1.1288206630423583,
      "learning_rate": 4.626349539543083e-06,
      "loss": 0.4695,
      "step": 10418
    },
    {
      "epoch": 0.5622470454913389,
      "grad_norm": 1.1538249430664769,
      "learning_rate": 4.625601174889429e-06,
      "loss": 0.6827,
      "step": 10419
    },
    {
      "epoch": 0.5623010091198533,
      "grad_norm": 0.8459246648399261,
      "learning_rate": 4.62485283536688e-06,
      "loss": 0.3615,
      "step": 10420
    },
    {
      "epoch": 0.5623549727483677,
      "grad_norm": 1.0960837412505011,
      "learning_rate": 4.624104520996944e-06,
      "loss": 0.4261,
      "step": 10421
    },
    {
      "epoch": 0.5624089363768819,
      "grad_norm": 1.0664625288175344,
      "learning_rate": 4.623356231801131e-06,
      "loss": 0.4545,
      "step": 10422
    },
    {
      "epoch": 0.5624629000053963,
      "grad_norm": 0.9941248482187476,
      "learning_rate": 4.6226079678009436e-06,
      "loss": 0.4084,
      "step": 10423
    },
    {
      "epoch": 0.5625168636339107,
      "grad_norm": 1.0659811941456354,
      "learning_rate": 4.6218597290178915e-06,
      "loss": 0.4793,
      "step": 10424
    },
    {
      "epoch": 0.5625708272624251,
      "grad_norm": 0.7504981626510868,
      "learning_rate": 4.621111515473476e-06,
      "loss": 0.2855,
      "step": 10425
    },
    {
      "epoch": 0.5626247908909395,
      "grad_norm": 1.0426496642181562,
      "learning_rate": 4.6203633271892055e-06,
      "loss": 0.5239,
      "step": 10426
    },
    {
      "epoch": 0.5626787545194539,
      "grad_norm": 1.066076224474066,
      "learning_rate": 4.61961516418658e-06,
      "loss": 0.4354,
      "step": 10427
    },
    {
      "epoch": 0.5627327181479682,
      "grad_norm": 1.0952431732132881,
      "learning_rate": 4.618867026487106e-06,
      "loss": 0.4694,
      "step": 10428
    },
    {
      "epoch": 0.5627866817764826,
      "grad_norm": 0.9520782178437929,
      "learning_rate": 4.618118914112284e-06,
      "loss": 0.6378,
      "step": 10429
    },
    {
      "epoch": 0.562840645404997,
      "grad_norm": 0.8067231581362336,
      "learning_rate": 4.617370827083613e-06,
      "loss": 0.3572,
      "step": 10430
    },
    {
      "epoch": 0.5628946090335114,
      "grad_norm": 0.8462480082730182,
      "learning_rate": 4.6166227654225994e-06,
      "loss": 0.4812,
      "step": 10431
    },
    {
      "epoch": 0.5629485726620258,
      "grad_norm": 0.9639543458116495,
      "learning_rate": 4.615874729150737e-06,
      "loss": 0.5463,
      "step": 10432
    },
    {
      "epoch": 0.5630025362905402,
      "grad_norm": 0.9982364353732058,
      "learning_rate": 4.615126718289531e-06,
      "loss": 0.5208,
      "step": 10433
    },
    {
      "epoch": 0.5630564999190546,
      "grad_norm": 0.8243331603684473,
      "learning_rate": 4.614378732860476e-06,
      "loss": 0.4104,
      "step": 10434
    },
    {
      "epoch": 0.5631104635475689,
      "grad_norm": 1.1125961336960704,
      "learning_rate": 4.613630772885073e-06,
      "loss": 0.6076,
      "step": 10435
    },
    {
      "epoch": 0.5631644271760833,
      "grad_norm": 1.2530837806331192,
      "learning_rate": 4.612882838384815e-06,
      "loss": 0.5736,
      "step": 10436
    },
    {
      "epoch": 0.5632183908045977,
      "grad_norm": 0.8418762342834804,
      "learning_rate": 4.6121349293812015e-06,
      "loss": 0.304,
      "step": 10437
    },
    {
      "epoch": 0.5632723544331121,
      "grad_norm": 0.7864831688443276,
      "learning_rate": 4.611387045895726e-06,
      "loss": 0.2755,
      "step": 10438
    },
    {
      "epoch": 0.5633263180616265,
      "grad_norm": 1.2358056639912849,
      "learning_rate": 4.610639187949886e-06,
      "loss": 0.4849,
      "step": 10439
    },
    {
      "epoch": 0.5633802816901409,
      "grad_norm": 1.23260109049792,
      "learning_rate": 4.609891355565172e-06,
      "loss": 0.5104,
      "step": 10440
    },
    {
      "epoch": 0.5634342453186553,
      "grad_norm": 1.2231120755727316,
      "learning_rate": 4.609143548763082e-06,
      "loss": 0.4937,
      "step": 10441
    },
    {
      "epoch": 0.5634882089471696,
      "grad_norm": 0.8460552938705836,
      "learning_rate": 4.608395767565108e-06,
      "loss": 0.3399,
      "step": 10442
    },
    {
      "epoch": 0.563542172575684,
      "grad_norm": 1.0533703303904305,
      "learning_rate": 4.607648011992738e-06,
      "loss": 0.6159,
      "step": 10443
    },
    {
      "epoch": 0.5635961362041984,
      "grad_norm": 0.8970560104302894,
      "learning_rate": 4.606900282067467e-06,
      "loss": 0.4594,
      "step": 10444
    },
    {
      "epoch": 0.5636500998327127,
      "grad_norm": 0.854141211172588,
      "learning_rate": 4.6061525778107845e-06,
      "loss": 0.4102,
      "step": 10445
    },
    {
      "epoch": 0.5637040634612271,
      "grad_norm": 1.2398508588764234,
      "learning_rate": 4.605404899244178e-06,
      "loss": 0.5239,
      "step": 10446
    },
    {
      "epoch": 0.5637580270897415,
      "grad_norm": 1.0856103298333295,
      "learning_rate": 4.604657246389141e-06,
      "loss": 0.5266,
      "step": 10447
    },
    {
      "epoch": 0.5638119907182559,
      "grad_norm": 0.7770382137721871,
      "learning_rate": 4.603909619267157e-06,
      "loss": 0.3225,
      "step": 10448
    },
    {
      "epoch": 0.5638659543467702,
      "grad_norm": 1.1715355614581755,
      "learning_rate": 4.603162017899718e-06,
      "loss": 0.5408,
      "step": 10449
    },
    {
      "epoch": 0.5639199179752846,
      "grad_norm": 0.8882758862806005,
      "learning_rate": 4.602414442308308e-06,
      "loss": 0.4413,
      "step": 10450
    },
    {
      "epoch": 0.563973881603799,
      "grad_norm": 0.9127773076845014,
      "learning_rate": 4.601666892514415e-06,
      "loss": 0.4923,
      "step": 10451
    },
    {
      "epoch": 0.5640278452323134,
      "grad_norm": 0.8631816417602077,
      "learning_rate": 4.600919368539523e-06,
      "loss": 0.4085,
      "step": 10452
    },
    {
      "epoch": 0.5640818088608278,
      "grad_norm": 0.9017419735588537,
      "learning_rate": 4.600171870405116e-06,
      "loss": 0.4682,
      "step": 10453
    },
    {
      "epoch": 0.5641357724893422,
      "grad_norm": 1.016984510076863,
      "learning_rate": 4.599424398132678e-06,
      "loss": 0.5359,
      "step": 10454
    },
    {
      "epoch": 0.5641897361178566,
      "grad_norm": 0.8852944412240826,
      "learning_rate": 4.598676951743694e-06,
      "loss": 0.3275,
      "step": 10455
    },
    {
      "epoch": 0.5642436997463709,
      "grad_norm": 1.0186523742949374,
      "learning_rate": 4.597929531259645e-06,
      "loss": 0.4358,
      "step": 10456
    },
    {
      "epoch": 0.5642976633748853,
      "grad_norm": 0.946002290808795,
      "learning_rate": 4.597182136702012e-06,
      "loss": 0.3759,
      "step": 10457
    },
    {
      "epoch": 0.5643516270033997,
      "grad_norm": 1.0167468786411347,
      "learning_rate": 4.596434768092277e-06,
      "loss": 0.493,
      "step": 10458
    },
    {
      "epoch": 0.5644055906319141,
      "grad_norm": 0.9069401324439225,
      "learning_rate": 4.595687425451919e-06,
      "loss": 0.4318,
      "step": 10459
    },
    {
      "epoch": 0.5644595542604285,
      "grad_norm": 0.8631004982072913,
      "learning_rate": 4.594940108802419e-06,
      "loss": 0.4057,
      "step": 10460
    },
    {
      "epoch": 0.5645135178889429,
      "grad_norm": 0.9299250777982595,
      "learning_rate": 4.594192818165255e-06,
      "loss": 0.3949,
      "step": 10461
    },
    {
      "epoch": 0.5645674815174573,
      "grad_norm": 1.0309033625358532,
      "learning_rate": 4.593445553561904e-06,
      "loss": 0.5333,
      "step": 10462
    },
    {
      "epoch": 0.5646214451459716,
      "grad_norm": 0.9561423030004743,
      "learning_rate": 4.592698315013845e-06,
      "loss": 0.4953,
      "step": 10463
    },
    {
      "epoch": 0.564675408774486,
      "grad_norm": 0.9671555622210293,
      "learning_rate": 4.591951102542553e-06,
      "loss": 0.4201,
      "step": 10464
    },
    {
      "epoch": 0.5647293724030004,
      "grad_norm": 0.8754228020017677,
      "learning_rate": 4.5912039161695036e-06,
      "loss": 0.3495,
      "step": 10465
    },
    {
      "epoch": 0.5647833360315148,
      "grad_norm": 1.1302772054924015,
      "learning_rate": 4.590456755916171e-06,
      "loss": 0.53,
      "step": 10466
    },
    {
      "epoch": 0.5648372996600292,
      "grad_norm": 0.8351331953368962,
      "learning_rate": 4.5897096218040294e-06,
      "loss": 0.3962,
      "step": 10467
    },
    {
      "epoch": 0.5648912632885436,
      "grad_norm": 0.9440805249199721,
      "learning_rate": 4.588962513854555e-06,
      "loss": 0.513,
      "step": 10468
    },
    {
      "epoch": 0.564945226917058,
      "grad_norm": 1.101379508163217,
      "learning_rate": 4.588215432089217e-06,
      "loss": 0.4718,
      "step": 10469
    },
    {
      "epoch": 0.5649991905455722,
      "grad_norm": 1.2364221436105636,
      "learning_rate": 4.5874683765294915e-06,
      "loss": 0.6042,
      "step": 10470
    },
    {
      "epoch": 0.5650531541740866,
      "grad_norm": 0.985800850701892,
      "learning_rate": 4.5867213471968455e-06,
      "loss": 0.3376,
      "step": 10471
    },
    {
      "epoch": 0.565107117802601,
      "grad_norm": 0.9129079325302729,
      "learning_rate": 4.585974344112749e-06,
      "loss": 0.4402,
      "step": 10472
    },
    {
      "epoch": 0.5651610814311154,
      "grad_norm": 1.0760406374597122,
      "learning_rate": 4.585227367298674e-06,
      "loss": 0.5239,
      "step": 10473
    },
    {
      "epoch": 0.5652150450596298,
      "grad_norm": 1.2340386890757566,
      "learning_rate": 4.5844804167760906e-06,
      "loss": 0.6443,
      "step": 10474
    },
    {
      "epoch": 0.5652690086881442,
      "grad_norm": 1.0604224505118627,
      "learning_rate": 4.583733492566463e-06,
      "loss": 0.4868,
      "step": 10475
    },
    {
      "epoch": 0.5653229723166586,
      "grad_norm": 0.9722103964779522,
      "learning_rate": 4.582986594691262e-06,
      "loss": 0.3633,
      "step": 10476
    },
    {
      "epoch": 0.5653769359451729,
      "grad_norm": 0.8566737740791921,
      "learning_rate": 4.582239723171953e-06,
      "loss": 0.3369,
      "step": 10477
    },
    {
      "epoch": 0.5654308995736873,
      "grad_norm": 1.0583734588262836,
      "learning_rate": 4.581492878030002e-06,
      "loss": 0.574,
      "step": 10478
    },
    {
      "epoch": 0.5654848632022017,
      "grad_norm": 1.0488481127007796,
      "learning_rate": 4.580746059286871e-06,
      "loss": 0.4544,
      "step": 10479
    },
    {
      "epoch": 0.5655388268307161,
      "grad_norm": 1.0206252778138516,
      "learning_rate": 4.579999266964029e-06,
      "loss": 0.4877,
      "step": 10480
    },
    {
      "epoch": 0.5655927904592305,
      "grad_norm": 0.6842875634433881,
      "learning_rate": 4.579252501082938e-06,
      "loss": 0.2895,
      "step": 10481
    },
    {
      "epoch": 0.5656467540877449,
      "grad_norm": 0.8091021866258207,
      "learning_rate": 4.578505761665061e-06,
      "loss": 0.3336,
      "step": 10482
    },
    {
      "epoch": 0.5657007177162593,
      "grad_norm": 1.0663593237080324,
      "learning_rate": 4.5777590487318595e-06,
      "loss": 0.4363,
      "step": 10483
    },
    {
      "epoch": 0.5657546813447736,
      "grad_norm": 0.9985080998935604,
      "learning_rate": 4.577012362304794e-06,
      "loss": 0.4481,
      "step": 10484
    },
    {
      "epoch": 0.565808644973288,
      "grad_norm": 1.0259516177533448,
      "learning_rate": 4.576265702405326e-06,
      "loss": 0.5135,
      "step": 10485
    },
    {
      "epoch": 0.5658626086018024,
      "grad_norm": 0.7960509646590144,
      "learning_rate": 4.575519069054914e-06,
      "loss": 0.3009,
      "step": 10486
    },
    {
      "epoch": 0.5659165722303168,
      "grad_norm": 1.1318196518923591,
      "learning_rate": 4.574772462275019e-06,
      "loss": 0.5344,
      "step": 10487
    },
    {
      "epoch": 0.5659705358588312,
      "grad_norm": 1.1599812341538092,
      "learning_rate": 4.574025882087097e-06,
      "loss": 0.5148,
      "step": 10488
    },
    {
      "epoch": 0.5660244994873456,
      "grad_norm": 1.104193440300905,
      "learning_rate": 4.5732793285126085e-06,
      "loss": 0.6625,
      "step": 10489
    },
    {
      "epoch": 0.56607846311586,
      "grad_norm": 0.971664785097364,
      "learning_rate": 4.572532801573008e-06,
      "loss": 0.382,
      "step": 10490
    },
    {
      "epoch": 0.5661324267443743,
      "grad_norm": 0.847045839342596,
      "learning_rate": 4.571786301289753e-06,
      "loss": 0.3419,
      "step": 10491
    },
    {
      "epoch": 0.5661863903728886,
      "grad_norm": 1.0025076018083392,
      "learning_rate": 4.571039827684297e-06,
      "loss": 0.6109,
      "step": 10492
    },
    {
      "epoch": 0.566240354001403,
      "grad_norm": 0.9719561822674162,
      "learning_rate": 4.570293380778095e-06,
      "loss": 0.4424,
      "step": 10493
    },
    {
      "epoch": 0.5662943176299174,
      "grad_norm": 0.8319105236420241,
      "learning_rate": 4.569546960592599e-06,
      "loss": 0.3909,
      "step": 10494
    },
    {
      "epoch": 0.5663482812584318,
      "grad_norm": 0.9925270799723463,
      "learning_rate": 4.568800567149265e-06,
      "loss": 0.4677,
      "step": 10495
    },
    {
      "epoch": 0.5664022448869462,
      "grad_norm": 0.9520855293244139,
      "learning_rate": 4.568054200469543e-06,
      "loss": 0.3786,
      "step": 10496
    },
    {
      "epoch": 0.5664562085154605,
      "grad_norm": 1.3646821439645345,
      "learning_rate": 4.567307860574887e-06,
      "loss": 0.6379,
      "step": 10497
    },
    {
      "epoch": 0.5665101721439749,
      "grad_norm": 1.161037060780476,
      "learning_rate": 4.566561547486743e-06,
      "loss": 0.527,
      "step": 10498
    },
    {
      "epoch": 0.5665641357724893,
      "grad_norm": 1.1139793078565807,
      "learning_rate": 4.565815261226564e-06,
      "loss": 0.6897,
      "step": 10499
    },
    {
      "epoch": 0.5666180994010037,
      "grad_norm": 0.9759634546330336,
      "learning_rate": 4.565069001815798e-06,
      "loss": 0.3925,
      "step": 10500
    },
    {
      "epoch": 0.5666180994010037,
      "eval_loss": 0.5438081622123718,
      "eval_runtime": 164.3918,
      "eval_samples_per_second": 20.92,
      "eval_steps_per_second": 0.876,
      "step": 10500
    },
    {
      "epoch": 0.5666720630295181,
      "grad_norm": 1.2132508406829081,
      "learning_rate": 4.564322769275893e-06,
      "loss": 0.6681,
      "step": 10501
    },
    {
      "epoch": 0.5667260266580325,
      "grad_norm": 0.8951656295534676,
      "learning_rate": 4.563576563628298e-06,
      "loss": 0.3908,
      "step": 10502
    },
    {
      "epoch": 0.5667799902865469,
      "grad_norm": 1.034408194809722,
      "learning_rate": 4.562830384894459e-06,
      "loss": 0.6627,
      "step": 10503
    },
    {
      "epoch": 0.5668339539150612,
      "grad_norm": 0.8452687506975001,
      "learning_rate": 4.562084233095821e-06,
      "loss": 0.3675,
      "step": 10504
    },
    {
      "epoch": 0.5668879175435756,
      "grad_norm": 1.211241568699674,
      "learning_rate": 4.561338108253831e-06,
      "loss": 0.6211,
      "step": 10505
    },
    {
      "epoch": 0.56694188117209,
      "grad_norm": 0.9143935184570449,
      "learning_rate": 4.560592010389931e-06,
      "loss": 0.4774,
      "step": 10506
    },
    {
      "epoch": 0.5669958448006044,
      "grad_norm": 1.2234019226153698,
      "learning_rate": 4.559845939525567e-06,
      "loss": 0.5969,
      "step": 10507
    },
    {
      "epoch": 0.5670498084291188,
      "grad_norm": 0.8991553793622392,
      "learning_rate": 4.559099895682179e-06,
      "loss": 0.421,
      "step": 10508
    },
    {
      "epoch": 0.5671037720576332,
      "grad_norm": 1.0017559533969251,
      "learning_rate": 4.558353878881211e-06,
      "loss": 0.4093,
      "step": 10509
    },
    {
      "epoch": 0.5671577356861476,
      "grad_norm": 1.1653015337490409,
      "learning_rate": 4.557607889144104e-06,
      "loss": 0.497,
      "step": 10510
    },
    {
      "epoch": 0.5672116993146619,
      "grad_norm": 1.255258636203648,
      "learning_rate": 4.556861926492302e-06,
      "loss": 0.5213,
      "step": 10511
    },
    {
      "epoch": 0.5672656629431763,
      "grad_norm": 1.1708837881373115,
      "learning_rate": 4.556115990947237e-06,
      "loss": 0.5367,
      "step": 10512
    },
    {
      "epoch": 0.5673196265716907,
      "grad_norm": 0.9824199753056545,
      "learning_rate": 4.555370082530354e-06,
      "loss": 0.4713,
      "step": 10513
    },
    {
      "epoch": 0.5673735902002051,
      "grad_norm": 1.0134020256880298,
      "learning_rate": 4.554624201263088e-06,
      "loss": 0.4971,
      "step": 10514
    },
    {
      "epoch": 0.5674275538287195,
      "grad_norm": 1.0959336425207347,
      "learning_rate": 4.553878347166878e-06,
      "loss": 0.4196,
      "step": 10515
    },
    {
      "epoch": 0.5674815174572339,
      "grad_norm": 1.0679688270717111,
      "learning_rate": 4.553132520263161e-06,
      "loss": 0.4829,
      "step": 10516
    },
    {
      "epoch": 0.5675354810857483,
      "grad_norm": 0.9045756208258133,
      "learning_rate": 4.552386720573372e-06,
      "loss": 0.4122,
      "step": 10517
    },
    {
      "epoch": 0.5675894447142625,
      "grad_norm": 1.0700682652821185,
      "learning_rate": 4.551640948118947e-06,
      "loss": 0.4049,
      "step": 10518
    },
    {
      "epoch": 0.5676434083427769,
      "grad_norm": 0.9938818995689936,
      "learning_rate": 4.550895202921319e-06,
      "loss": 0.4334,
      "step": 10519
    },
    {
      "epoch": 0.5676973719712913,
      "grad_norm": 1.1922130131925306,
      "learning_rate": 4.550149485001922e-06,
      "loss": 0.5675,
      "step": 10520
    },
    {
      "epoch": 0.5677513355998057,
      "grad_norm": 1.1517116584619278,
      "learning_rate": 4.5494037943821855e-06,
      "loss": 0.6336,
      "step": 10521
    },
    {
      "epoch": 0.5678052992283201,
      "grad_norm": 0.9102326012757607,
      "learning_rate": 4.548658131083546e-06,
      "loss": 0.3859,
      "step": 10522
    },
    {
      "epoch": 0.5678592628568345,
      "grad_norm": 0.9100686524729895,
      "learning_rate": 4.547912495127435e-06,
      "loss": 0.5856,
      "step": 10523
    },
    {
      "epoch": 0.5679132264853489,
      "grad_norm": 1.1612250822560735,
      "learning_rate": 4.547166886535279e-06,
      "loss": 0.4224,
      "step": 10524
    },
    {
      "epoch": 0.5679671901138632,
      "grad_norm": 1.124810719943975,
      "learning_rate": 4.5464213053285115e-06,
      "loss": 0.622,
      "step": 10525
    },
    {
      "epoch": 0.5680211537423776,
      "grad_norm": 0.8666698667896425,
      "learning_rate": 4.545675751528556e-06,
      "loss": 0.3742,
      "step": 10526
    },
    {
      "epoch": 0.568075117370892,
      "grad_norm": 1.2144187730549996,
      "learning_rate": 4.544930225156847e-06,
      "loss": 0.6003,
      "step": 10527
    },
    {
      "epoch": 0.5681290809994064,
      "grad_norm": 1.091008511310262,
      "learning_rate": 4.544184726234807e-06,
      "loss": 0.5005,
      "step": 10528
    },
    {
      "epoch": 0.5681830446279208,
      "grad_norm": 0.9474822575642348,
      "learning_rate": 4.543439254783864e-06,
      "loss": 0.3841,
      "step": 10529
    },
    {
      "epoch": 0.5682370082564352,
      "grad_norm": 0.9525027683818699,
      "learning_rate": 4.542693810825443e-06,
      "loss": 0.4796,
      "step": 10530
    },
    {
      "epoch": 0.5682909718849496,
      "grad_norm": 1.1096716407691811,
      "learning_rate": 4.541948394380968e-06,
      "loss": 0.5117,
      "step": 10531
    },
    {
      "epoch": 0.5683449355134639,
      "grad_norm": 1.1559454816488712,
      "learning_rate": 4.541203005471867e-06,
      "loss": 0.5466,
      "step": 10532
    },
    {
      "epoch": 0.5683988991419783,
      "grad_norm": 1.0178992963985396,
      "learning_rate": 4.540457644119558e-06,
      "loss": 0.5088,
      "step": 10533
    },
    {
      "epoch": 0.5684528627704927,
      "grad_norm": 1.037623615809952,
      "learning_rate": 4.5397123103454665e-06,
      "loss": 0.5513,
      "step": 10534
    },
    {
      "epoch": 0.5685068263990071,
      "grad_norm": 1.0430502586912973,
      "learning_rate": 4.538967004171013e-06,
      "loss": 0.5399,
      "step": 10535
    },
    {
      "epoch": 0.5685607900275215,
      "grad_norm": 0.9913932180355542,
      "learning_rate": 4.53822172561762e-06,
      "loss": 0.4715,
      "step": 10536
    },
    {
      "epoch": 0.5686147536560359,
      "grad_norm": 1.0173395766093782,
      "learning_rate": 4.537476474706706e-06,
      "loss": 0.5516,
      "step": 10537
    },
    {
      "epoch": 0.5686687172845503,
      "grad_norm": 1.0182770347367338,
      "learning_rate": 4.536731251459689e-06,
      "loss": 0.4234,
      "step": 10538
    },
    {
      "epoch": 0.5687226809130645,
      "grad_norm": 0.7806052958255845,
      "learning_rate": 4.535986055897993e-06,
      "loss": 0.3633,
      "step": 10539
    },
    {
      "epoch": 0.568776644541579,
      "grad_norm": 0.9409031177040648,
      "learning_rate": 4.535240888043029e-06,
      "loss": 0.4104,
      "step": 10540
    },
    {
      "epoch": 0.5688306081700933,
      "grad_norm": 1.0698967036525588,
      "learning_rate": 4.534495747916218e-06,
      "loss": 0.5668,
      "step": 10541
    },
    {
      "epoch": 0.5688845717986077,
      "grad_norm": 1.0822913122984983,
      "learning_rate": 4.533750635538974e-06,
      "loss": 0.6055,
      "step": 10542
    },
    {
      "epoch": 0.5689385354271221,
      "grad_norm": 1.134195390052243,
      "learning_rate": 4.533005550932712e-06,
      "loss": 0.6183,
      "step": 10543
    },
    {
      "epoch": 0.5689924990556365,
      "grad_norm": 1.1415572245749117,
      "learning_rate": 4.532260494118849e-06,
      "loss": 0.5873,
      "step": 10544
    },
    {
      "epoch": 0.5690464626841509,
      "grad_norm": 0.9114946320112447,
      "learning_rate": 4.531515465118797e-06,
      "loss": 0.3716,
      "step": 10545
    },
    {
      "epoch": 0.5691004263126652,
      "grad_norm": 0.8388432682843823,
      "learning_rate": 4.530770463953969e-06,
      "loss": 0.4114,
      "step": 10546
    },
    {
      "epoch": 0.5691543899411796,
      "grad_norm": 1.0848657138134543,
      "learning_rate": 4.530025490645778e-06,
      "loss": 0.529,
      "step": 10547
    },
    {
      "epoch": 0.569208353569694,
      "grad_norm": 0.9650432036842284,
      "learning_rate": 4.529280545215633e-06,
      "loss": 0.4884,
      "step": 10548
    },
    {
      "epoch": 0.5692623171982084,
      "grad_norm": 0.9424371966893221,
      "learning_rate": 4.528535627684946e-06,
      "loss": 0.4177,
      "step": 10549
    },
    {
      "epoch": 0.5693162808267228,
      "grad_norm": 1.1418985813333165,
      "learning_rate": 4.527790738075128e-06,
      "loss": 0.5627,
      "step": 10550
    },
    {
      "epoch": 0.5693702444552372,
      "grad_norm": 0.7918763054499546,
      "learning_rate": 4.5270458764075855e-06,
      "loss": 0.2944,
      "step": 10551
    },
    {
      "epoch": 0.5694242080837516,
      "grad_norm": 1.175055885712422,
      "learning_rate": 4.526301042703727e-06,
      "loss": 0.6097,
      "step": 10552
    },
    {
      "epoch": 0.5694781717122659,
      "grad_norm": 0.9414820088729619,
      "learning_rate": 4.5255562369849625e-06,
      "loss": 0.3798,
      "step": 10553
    },
    {
      "epoch": 0.5695321353407803,
      "grad_norm": 0.8678124951116019,
      "learning_rate": 4.524811459272694e-06,
      "loss": 0.382,
      "step": 10554
    },
    {
      "epoch": 0.5695860989692947,
      "grad_norm": 1.0811301802284357,
      "learning_rate": 4.524066709588332e-06,
      "loss": 0.5468,
      "step": 10555
    },
    {
      "epoch": 0.5696400625978091,
      "grad_norm": 1.0969387736888854,
      "learning_rate": 4.523321987953276e-06,
      "loss": 0.589,
      "step": 10556
    },
    {
      "epoch": 0.5696940262263235,
      "grad_norm": 1.459383894529047,
      "learning_rate": 4.522577294388934e-06,
      "loss": 0.6391,
      "step": 10557
    },
    {
      "epoch": 0.5697479898548379,
      "grad_norm": 1.078736179186585,
      "learning_rate": 4.5218326289167106e-06,
      "loss": 0.4589,
      "step": 10558
    },
    {
      "epoch": 0.5698019534833523,
      "grad_norm": 0.8551121739217271,
      "learning_rate": 4.521087991558003e-06,
      "loss": 0.4184,
      "step": 10559
    },
    {
      "epoch": 0.5698559171118666,
      "grad_norm": 0.9925483654509739,
      "learning_rate": 4.520343382334217e-06,
      "loss": 0.4486,
      "step": 10560
    },
    {
      "epoch": 0.569909880740381,
      "grad_norm": 1.0781736582206884,
      "learning_rate": 4.519598801266753e-06,
      "loss": 0.359,
      "step": 10561
    },
    {
      "epoch": 0.5699638443688954,
      "grad_norm": 0.9239417380792921,
      "learning_rate": 4.518854248377007e-06,
      "loss": 0.4309,
      "step": 10562
    },
    {
      "epoch": 0.5700178079974098,
      "grad_norm": 0.9330596678201261,
      "learning_rate": 4.518109723686383e-06,
      "loss": 0.5132,
      "step": 10563
    },
    {
      "epoch": 0.5700717716259242,
      "grad_norm": 0.9500103334051261,
      "learning_rate": 4.5173652272162775e-06,
      "loss": 0.4419,
      "step": 10564
    },
    {
      "epoch": 0.5701257352544385,
      "grad_norm": 1.0728585293580428,
      "learning_rate": 4.516620758988087e-06,
      "loss": 0.5091,
      "step": 10565
    },
    {
      "epoch": 0.5701796988829528,
      "grad_norm": 0.984952903966271,
      "learning_rate": 4.515876319023213e-06,
      "loss": 0.4436,
      "step": 10566
    },
    {
      "epoch": 0.5702336625114672,
      "grad_norm": 1.3209893927019474,
      "learning_rate": 4.515131907343046e-06,
      "loss": 0.7137,
      "step": 10567
    },
    {
      "epoch": 0.5702876261399816,
      "grad_norm": 0.8544703789693543,
      "learning_rate": 4.514387523968983e-06,
      "loss": 0.3787,
      "step": 10568
    },
    {
      "epoch": 0.570341589768496,
      "grad_norm": 1.2009008387930713,
      "learning_rate": 4.513643168922418e-06,
      "loss": 0.459,
      "step": 10569
    },
    {
      "epoch": 0.5703955533970104,
      "grad_norm": 1.0320861994311454,
      "learning_rate": 4.512898842224745e-06,
      "loss": 0.4445,
      "step": 10570
    },
    {
      "epoch": 0.5704495170255248,
      "grad_norm": 0.9264477902612831,
      "learning_rate": 4.512154543897357e-06,
      "loss": 0.3697,
      "step": 10571
    },
    {
      "epoch": 0.5705034806540392,
      "grad_norm": 0.9267906066561483,
      "learning_rate": 4.511410273961645e-06,
      "loss": 0.5279,
      "step": 10572
    },
    {
      "epoch": 0.5705574442825535,
      "grad_norm": 1.117232184546726,
      "learning_rate": 4.510666032439003e-06,
      "loss": 0.4677,
      "step": 10573
    },
    {
      "epoch": 0.5706114079110679,
      "grad_norm": 1.3201633979718521,
      "learning_rate": 4.509921819350817e-06,
      "loss": 0.6103,
      "step": 10574
    },
    {
      "epoch": 0.5706653715395823,
      "grad_norm": 0.7394646500033647,
      "learning_rate": 4.509177634718478e-06,
      "loss": 0.291,
      "step": 10575
    },
    {
      "epoch": 0.5707193351680967,
      "grad_norm": 0.8164238954965022,
      "learning_rate": 4.508433478563375e-06,
      "loss": 0.337,
      "step": 10576
    },
    {
      "epoch": 0.5707732987966111,
      "grad_norm": 0.9377936993896832,
      "learning_rate": 4.507689350906895e-06,
      "loss": 0.4177,
      "step": 10577
    },
    {
      "epoch": 0.5708272624251255,
      "grad_norm": 1.0285659302703003,
      "learning_rate": 4.506945251770427e-06,
      "loss": 0.4399,
      "step": 10578
    },
    {
      "epoch": 0.5708812260536399,
      "grad_norm": 1.2379819057240777,
      "learning_rate": 4.506201181175356e-06,
      "loss": 0.5213,
      "step": 10579
    },
    {
      "epoch": 0.5709351896821542,
      "grad_norm": 0.8786634095111288,
      "learning_rate": 4.505457139143068e-06,
      "loss": 0.3471,
      "step": 10580
    },
    {
      "epoch": 0.5709891533106686,
      "grad_norm": 1.0687462454976182,
      "learning_rate": 4.504713125694945e-06,
      "loss": 0.5849,
      "step": 10581
    },
    {
      "epoch": 0.571043116939183,
      "grad_norm": 0.9770542060711697,
      "learning_rate": 4.5039691408523725e-06,
      "loss": 0.4603,
      "step": 10582
    },
    {
      "epoch": 0.5710970805676974,
      "grad_norm": 0.8196514535962424,
      "learning_rate": 4.5032251846367335e-06,
      "loss": 0.3196,
      "step": 10583
    },
    {
      "epoch": 0.5711510441962118,
      "grad_norm": 1.0666495916125962,
      "learning_rate": 4.50248125706941e-06,
      "loss": 0.5725,
      "step": 10584
    },
    {
      "epoch": 0.5712050078247262,
      "grad_norm": 0.8880110331413505,
      "learning_rate": 4.501737358171784e-06,
      "loss": 0.4414,
      "step": 10585
    },
    {
      "epoch": 0.5712589714532406,
      "grad_norm": 1.0806083702655236,
      "learning_rate": 4.5009934879652325e-06,
      "loss": 0.5349,
      "step": 10586
    },
    {
      "epoch": 0.5713129350817548,
      "grad_norm": 0.976760398180387,
      "learning_rate": 4.500249646471142e-06,
      "loss": 0.522,
      "step": 10587
    },
    {
      "epoch": 0.5713668987102692,
      "grad_norm": 1.1099936156874142,
      "learning_rate": 4.499505833710884e-06,
      "loss": 0.4811,
      "step": 10588
    },
    {
      "epoch": 0.5714208623387836,
      "grad_norm": 1.1365399227678186,
      "learning_rate": 4.498762049705839e-06,
      "loss": 0.4581,
      "step": 10589
    },
    {
      "epoch": 0.571474825967298,
      "grad_norm": 1.0310848613716224,
      "learning_rate": 4.498018294477384e-06,
      "loss": 0.5865,
      "step": 10590
    },
    {
      "epoch": 0.5715287895958124,
      "grad_norm": 1.023656860971758,
      "learning_rate": 4.497274568046896e-06,
      "loss": 0.4429,
      "step": 10591
    },
    {
      "epoch": 0.5715827532243268,
      "grad_norm": 1.1138134135035493,
      "learning_rate": 4.4965308704357515e-06,
      "loss": 0.5776,
      "step": 10592
    },
    {
      "epoch": 0.5716367168528412,
      "grad_norm": 0.8935108769966343,
      "learning_rate": 4.495787201665321e-06,
      "loss": 0.4136,
      "step": 10593
    },
    {
      "epoch": 0.5716906804813555,
      "grad_norm": 1.2970485389581448,
      "learning_rate": 4.495043561756986e-06,
      "loss": 0.6513,
      "step": 10594
    },
    {
      "epoch": 0.5717446441098699,
      "grad_norm": 1.2812015334423443,
      "learning_rate": 4.49429995073211e-06,
      "loss": 0.5031,
      "step": 10595
    },
    {
      "epoch": 0.5717986077383843,
      "grad_norm": 1.1613189386346907,
      "learning_rate": 4.493556368612071e-06,
      "loss": 0.404,
      "step": 10596
    },
    {
      "epoch": 0.5718525713668987,
      "grad_norm": 1.0982884112344797,
      "learning_rate": 4.492812815418239e-06,
      "loss": 0.468,
      "step": 10597
    },
    {
      "epoch": 0.5719065349954131,
      "grad_norm": 1.026060798855499,
      "learning_rate": 4.492069291171984e-06,
      "loss": 0.4663,
      "step": 10598
    },
    {
      "epoch": 0.5719604986239275,
      "grad_norm": 1.0152422338255567,
      "learning_rate": 4.491325795894675e-06,
      "loss": 0.4974,
      "step": 10599
    },
    {
      "epoch": 0.5720144622524419,
      "grad_norm": 1.0062215408303345,
      "learning_rate": 4.490582329607683e-06,
      "loss": 0.4519,
      "step": 10600
    },
    {
      "epoch": 0.5720684258809562,
      "grad_norm": 1.0536011911807657,
      "learning_rate": 4.4898388923323755e-06,
      "loss": 0.4505,
      "step": 10601
    },
    {
      "epoch": 0.5721223895094706,
      "grad_norm": 0.9955333466003488,
      "learning_rate": 4.489095484090119e-06,
      "loss": 0.3786,
      "step": 10602
    },
    {
      "epoch": 0.572176353137985,
      "grad_norm": 0.8181339198428329,
      "learning_rate": 4.488352104902278e-06,
      "loss": 0.3635,
      "step": 10603
    },
    {
      "epoch": 0.5722303167664994,
      "grad_norm": 1.0718829164782495,
      "learning_rate": 4.487608754790221e-06,
      "loss": 0.5469,
      "step": 10604
    },
    {
      "epoch": 0.5722842803950138,
      "grad_norm": 1.020375693885567,
      "learning_rate": 4.48686543377531e-06,
      "loss": 0.3884,
      "step": 10605
    },
    {
      "epoch": 0.5723382440235282,
      "grad_norm": 1.158302511788771,
      "learning_rate": 4.486122141878912e-06,
      "loss": 0.6245,
      "step": 10606
    },
    {
      "epoch": 0.5723922076520426,
      "grad_norm": 0.9122582134344321,
      "learning_rate": 4.4853788791223865e-06,
      "loss": 0.4675,
      "step": 10607
    },
    {
      "epoch": 0.5724461712805569,
      "grad_norm": 1.0154385974893372,
      "learning_rate": 4.484635645527099e-06,
      "loss": 0.4066,
      "step": 10608
    },
    {
      "epoch": 0.5725001349090713,
      "grad_norm": 1.010129177438139,
      "learning_rate": 4.483892441114408e-06,
      "loss": 0.4488,
      "step": 10609
    },
    {
      "epoch": 0.5725540985375857,
      "grad_norm": 1.0495565327298668,
      "learning_rate": 4.4831492659056745e-06,
      "loss": 0.5249,
      "step": 10610
    },
    {
      "epoch": 0.5726080621661,
      "grad_norm": 0.8551594604325177,
      "learning_rate": 4.482406119922258e-06,
      "loss": 0.4357,
      "step": 10611
    },
    {
      "epoch": 0.5726620257946144,
      "grad_norm": 1.0453406455494891,
      "learning_rate": 4.481663003185519e-06,
      "loss": 0.4747,
      "step": 10612
    },
    {
      "epoch": 0.5727159894231288,
      "grad_norm": 1.0537798755337189,
      "learning_rate": 4.480919915716814e-06,
      "loss": 0.4889,
      "step": 10613
    },
    {
      "epoch": 0.5727699530516432,
      "grad_norm": 0.9098585356323315,
      "learning_rate": 4.480176857537499e-06,
      "loss": 0.3944,
      "step": 10614
    },
    {
      "epoch": 0.5728239166801575,
      "grad_norm": 0.9713307100252243,
      "learning_rate": 4.479433828668935e-06,
      "loss": 0.4,
      "step": 10615
    },
    {
      "epoch": 0.5728778803086719,
      "grad_norm": 0.7627267409792713,
      "learning_rate": 4.478690829132471e-06,
      "loss": 0.3067,
      "step": 10616
    },
    {
      "epoch": 0.5729318439371863,
      "grad_norm": 1.1906731762058105,
      "learning_rate": 4.477947858949465e-06,
      "loss": 0.5543,
      "step": 10617
    },
    {
      "epoch": 0.5729858075657007,
      "grad_norm": 0.8867902300040209,
      "learning_rate": 4.477204918141271e-06,
      "loss": 0.3774,
      "step": 10618
    },
    {
      "epoch": 0.5730397711942151,
      "grad_norm": 1.1020224007125945,
      "learning_rate": 4.476462006729241e-06,
      "loss": 0.4897,
      "step": 10619
    },
    {
      "epoch": 0.5730937348227295,
      "grad_norm": 1.1515094448383796,
      "learning_rate": 4.475719124734725e-06,
      "loss": 0.5414,
      "step": 10620
    },
    {
      "epoch": 0.5731476984512439,
      "grad_norm": 1.208875848794995,
      "learning_rate": 4.474976272179079e-06,
      "loss": 0.5603,
      "step": 10621
    },
    {
      "epoch": 0.5732016620797582,
      "grad_norm": 1.3763739111325086,
      "learning_rate": 4.474233449083651e-06,
      "loss": 0.6205,
      "step": 10622
    },
    {
      "epoch": 0.5732556257082726,
      "grad_norm": 1.2380235108393458,
      "learning_rate": 4.473490655469789e-06,
      "loss": 0.4517,
      "step": 10623
    },
    {
      "epoch": 0.573309589336787,
      "grad_norm": 1.0572127500698172,
      "learning_rate": 4.472747891358843e-06,
      "loss": 0.5109,
      "step": 10624
    },
    {
      "epoch": 0.5733635529653014,
      "grad_norm": 1.0261652073233762,
      "learning_rate": 4.47200515677216e-06,
      "loss": 0.4524,
      "step": 10625
    },
    {
      "epoch": 0.5734175165938158,
      "grad_norm": 1.1489572176933056,
      "learning_rate": 4.471262451731086e-06,
      "loss": 0.4725,
      "step": 10626
    },
    {
      "epoch": 0.5734714802223302,
      "grad_norm": 1.5247047165162657,
      "learning_rate": 4.470519776256971e-06,
      "loss": 0.6385,
      "step": 10627
    },
    {
      "epoch": 0.5735254438508446,
      "grad_norm": 0.910473574632374,
      "learning_rate": 4.469777130371158e-06,
      "loss": 0.3324,
      "step": 10628
    },
    {
      "epoch": 0.5735794074793589,
      "grad_norm": 1.02306860790074,
      "learning_rate": 4.46903451409499e-06,
      "loss": 0.5237,
      "step": 10629
    },
    {
      "epoch": 0.5736333711078733,
      "grad_norm": 1.0074265986182689,
      "learning_rate": 4.468291927449813e-06,
      "loss": 0.4344,
      "step": 10630
    },
    {
      "epoch": 0.5736873347363877,
      "grad_norm": 0.9534049489737269,
      "learning_rate": 4.467549370456966e-06,
      "loss": 0.5183,
      "step": 10631
    },
    {
      "epoch": 0.5737412983649021,
      "grad_norm": 0.9788153041361316,
      "learning_rate": 4.466806843137795e-06,
      "loss": 0.484,
      "step": 10632
    },
    {
      "epoch": 0.5737952619934165,
      "grad_norm": 1.1465662917526995,
      "learning_rate": 4.466064345513641e-06,
      "loss": 0.6373,
      "step": 10633
    },
    {
      "epoch": 0.5738492256219309,
      "grad_norm": 1.0909256431101533,
      "learning_rate": 4.46532187760584e-06,
      "loss": 0.5888,
      "step": 10634
    },
    {
      "epoch": 0.5739031892504451,
      "grad_norm": 1.0076671982183465,
      "learning_rate": 4.464579439435736e-06,
      "loss": 0.4381,
      "step": 10635
    },
    {
      "epoch": 0.5739571528789595,
      "grad_norm": 1.1519822433028821,
      "learning_rate": 4.463837031024663e-06,
      "loss": 0.6434,
      "step": 10636
    },
    {
      "epoch": 0.5740111165074739,
      "grad_norm": 0.941164973166366,
      "learning_rate": 4.463094652393961e-06,
      "loss": 0.5318,
      "step": 10637
    },
    {
      "epoch": 0.5740650801359883,
      "grad_norm": 0.7363305232511407,
      "learning_rate": 4.462352303564966e-06,
      "loss": 0.2642,
      "step": 10638
    },
    {
      "epoch": 0.5741190437645027,
      "grad_norm": 1.0489660972652277,
      "learning_rate": 4.461609984559014e-06,
      "loss": 0.4037,
      "step": 10639
    },
    {
      "epoch": 0.5741730073930171,
      "grad_norm": 0.9438426645962671,
      "learning_rate": 4.4608676953974416e-06,
      "loss": 0.3441,
      "step": 10640
    },
    {
      "epoch": 0.5742269710215315,
      "grad_norm": 0.925922080447645,
      "learning_rate": 4.4601254361015815e-06,
      "loss": 0.3727,
      "step": 10641
    },
    {
      "epoch": 0.5742809346500458,
      "grad_norm": 0.9954552695095218,
      "learning_rate": 4.459383206692768e-06,
      "loss": 0.42,
      "step": 10642
    },
    {
      "epoch": 0.5743348982785602,
      "grad_norm": 1.1461463009291533,
      "learning_rate": 4.458641007192332e-06,
      "loss": 0.5276,
      "step": 10643
    },
    {
      "epoch": 0.5743888619070746,
      "grad_norm": 0.848896681246346,
      "learning_rate": 4.457898837621604e-06,
      "loss": 0.2856,
      "step": 10644
    },
    {
      "epoch": 0.574442825535589,
      "grad_norm": 0.9696876079010146,
      "learning_rate": 4.457156698001917e-06,
      "loss": 0.5137,
      "step": 10645
    },
    {
      "epoch": 0.5744967891641034,
      "grad_norm": 0.8506717179810759,
      "learning_rate": 4.4564145883546e-06,
      "loss": 0.4366,
      "step": 10646
    },
    {
      "epoch": 0.5745507527926178,
      "grad_norm": 1.1390513692599813,
      "learning_rate": 4.4556725087009835e-06,
      "loss": 0.5321,
      "step": 10647
    },
    {
      "epoch": 0.5746047164211322,
      "grad_norm": 0.9331765094324005,
      "learning_rate": 4.454930459062393e-06,
      "loss": 0.4524,
      "step": 10648
    },
    {
      "epoch": 0.5746586800496465,
      "grad_norm": 1.0193378712012588,
      "learning_rate": 4.45418843946016e-06,
      "loss": 0.5267,
      "step": 10649
    },
    {
      "epoch": 0.5747126436781609,
      "grad_norm": 1.1351950304079719,
      "learning_rate": 4.453446449915605e-06,
      "loss": 0.5341,
      "step": 10650
    },
    {
      "epoch": 0.5747666073066753,
      "grad_norm": 0.9623760350941655,
      "learning_rate": 4.452704490450057e-06,
      "loss": 0.401,
      "step": 10651
    },
    {
      "epoch": 0.5748205709351897,
      "grad_norm": 1.1465314665200284,
      "learning_rate": 4.45196256108484e-06,
      "loss": 0.6252,
      "step": 10652
    },
    {
      "epoch": 0.5748745345637041,
      "grad_norm": 1.0205692254614633,
      "learning_rate": 4.4512206618412775e-06,
      "loss": 0.4225,
      "step": 10653
    },
    {
      "epoch": 0.5749284981922185,
      "grad_norm": 0.8575756349485733,
      "learning_rate": 4.450478792740693e-06,
      "loss": 0.3653,
      "step": 10654
    },
    {
      "epoch": 0.5749824618207329,
      "grad_norm": 0.968824270851426,
      "learning_rate": 4.44973695380441e-06,
      "loss": 0.3721,
      "step": 10655
    },
    {
      "epoch": 0.5750364254492472,
      "grad_norm": 0.8818761568495516,
      "learning_rate": 4.448995145053747e-06,
      "loss": 0.3959,
      "step": 10656
    },
    {
      "epoch": 0.5750903890777616,
      "grad_norm": 1.0353503820231238,
      "learning_rate": 4.448253366510027e-06,
      "loss": 0.5265,
      "step": 10657
    },
    {
      "epoch": 0.575144352706276,
      "grad_norm": 1.1088991750348278,
      "learning_rate": 4.447511618194566e-06,
      "loss": 0.5284,
      "step": 10658
    },
    {
      "epoch": 0.5751983163347903,
      "grad_norm": 0.9110503012042285,
      "learning_rate": 4.446769900128685e-06,
      "loss": 0.3901,
      "step": 10659
    },
    {
      "epoch": 0.5752522799633047,
      "grad_norm": 1.2965241922704283,
      "learning_rate": 4.446028212333702e-06,
      "loss": 0.5571,
      "step": 10660
    },
    {
      "epoch": 0.5753062435918191,
      "grad_norm": 0.7805924178267505,
      "learning_rate": 4.445286554830932e-06,
      "loss": 0.3155,
      "step": 10661
    },
    {
      "epoch": 0.5753602072203335,
      "grad_norm": 0.8118498411789405,
      "learning_rate": 4.444544927641692e-06,
      "loss": 0.3785,
      "step": 10662
    },
    {
      "epoch": 0.5754141708488478,
      "grad_norm": 1.0804680718586033,
      "learning_rate": 4.4438033307873e-06,
      "loss": 0.5537,
      "step": 10663
    },
    {
      "epoch": 0.5754681344773622,
      "grad_norm": 0.89237425371023,
      "learning_rate": 4.443061764289066e-06,
      "loss": 0.4197,
      "step": 10664
    },
    {
      "epoch": 0.5755220981058766,
      "grad_norm": 0.7820333766537252,
      "learning_rate": 4.442320228168303e-06,
      "loss": 0.2762,
      "step": 10665
    },
    {
      "epoch": 0.575576061734391,
      "grad_norm": 1.1846033614085174,
      "learning_rate": 4.441578722446327e-06,
      "loss": 0.5004,
      "step": 10666
    },
    {
      "epoch": 0.5756300253629054,
      "grad_norm": 1.1890490110874712,
      "learning_rate": 4.440837247144447e-06,
      "loss": 0.5309,
      "step": 10667
    },
    {
      "epoch": 0.5756839889914198,
      "grad_norm": 1.1940633524984023,
      "learning_rate": 4.440095802283976e-06,
      "loss": 0.5053,
      "step": 10668
    },
    {
      "epoch": 0.5757379526199342,
      "grad_norm": 1.0428366995388378,
      "learning_rate": 4.439354387886222e-06,
      "loss": 0.4148,
      "step": 10669
    },
    {
      "epoch": 0.5757919162484485,
      "grad_norm": 1.0682556446517424,
      "learning_rate": 4.438613003972496e-06,
      "loss": 0.5071,
      "step": 10670
    },
    {
      "epoch": 0.5758458798769629,
      "grad_norm": 0.8569192878246306,
      "learning_rate": 4.4378716505641025e-06,
      "loss": 0.3738,
      "step": 10671
    },
    {
      "epoch": 0.5758998435054773,
      "grad_norm": 1.050186583096519,
      "learning_rate": 4.437130327682353e-06,
      "loss": 0.3937,
      "step": 10672
    },
    {
      "epoch": 0.5759538071339917,
      "grad_norm": 0.9761970650772003,
      "learning_rate": 4.4363890353485495e-06,
      "loss": 0.3677,
      "step": 10673
    },
    {
      "epoch": 0.5760077707625061,
      "grad_norm": 1.0570058227268315,
      "learning_rate": 4.435647773584001e-06,
      "loss": 0.4751,
      "step": 10674
    },
    {
      "epoch": 0.5760617343910205,
      "grad_norm": 1.0875588764862072,
      "learning_rate": 4.434906542410011e-06,
      "loss": 0.5224,
      "step": 10675
    },
    {
      "epoch": 0.5761156980195349,
      "grad_norm": 1.0527043768908537,
      "learning_rate": 4.434165341847883e-06,
      "loss": 0.475,
      "step": 10676
    },
    {
      "epoch": 0.5761696616480492,
      "grad_norm": 1.2398426839470986,
      "learning_rate": 4.433424171918922e-06,
      "loss": 0.5993,
      "step": 10677
    },
    {
      "epoch": 0.5762236252765636,
      "grad_norm": 1.2808480113548504,
      "learning_rate": 4.4326830326444246e-06,
      "loss": 0.7322,
      "step": 10678
    },
    {
      "epoch": 0.576277588905078,
      "grad_norm": 0.9556415749767011,
      "learning_rate": 4.431941924045697e-06,
      "loss": 0.4069,
      "step": 10679
    },
    {
      "epoch": 0.5763315525335924,
      "grad_norm": 1.013089484999727,
      "learning_rate": 4.431200846144038e-06,
      "loss": 0.5584,
      "step": 10680
    },
    {
      "epoch": 0.5763855161621068,
      "grad_norm": 1.086361037713495,
      "learning_rate": 4.430459798960746e-06,
      "loss": 0.4947,
      "step": 10681
    },
    {
      "epoch": 0.5764394797906212,
      "grad_norm": 1.077873560168301,
      "learning_rate": 4.429718782517119e-06,
      "loss": 0.4885,
      "step": 10682
    },
    {
      "epoch": 0.5764934434191356,
      "grad_norm": 0.9366757726525804,
      "learning_rate": 4.428977796834457e-06,
      "loss": 0.4907,
      "step": 10683
    },
    {
      "epoch": 0.5765474070476498,
      "grad_norm": 1.0187354445144805,
      "learning_rate": 4.428236841934056e-06,
      "loss": 0.425,
      "step": 10684
    },
    {
      "epoch": 0.5766013706761642,
      "grad_norm": 1.089358238605998,
      "learning_rate": 4.42749591783721e-06,
      "loss": 0.4437,
      "step": 10685
    },
    {
      "epoch": 0.5766553343046786,
      "grad_norm": 0.9985471785190964,
      "learning_rate": 4.426755024565215e-06,
      "loss": 0.47,
      "step": 10686
    },
    {
      "epoch": 0.576709297933193,
      "grad_norm": 0.9170018732528163,
      "learning_rate": 4.426014162139365e-06,
      "loss": 0.4732,
      "step": 10687
    },
    {
      "epoch": 0.5767632615617074,
      "grad_norm": 1.2895337996614005,
      "learning_rate": 4.425273330580954e-06,
      "loss": 0.5087,
      "step": 10688
    },
    {
      "epoch": 0.5768172251902218,
      "grad_norm": 0.8628008861982333,
      "learning_rate": 4.424532529911271e-06,
      "loss": 0.4137,
      "step": 10689
    },
    {
      "epoch": 0.5768711888187362,
      "grad_norm": 1.0196109767323036,
      "learning_rate": 4.4237917601516135e-06,
      "loss": 0.5107,
      "step": 10690
    },
    {
      "epoch": 0.5769251524472505,
      "grad_norm": 0.9033378578818797,
      "learning_rate": 4.423051021323267e-06,
      "loss": 0.3724,
      "step": 10691
    },
    {
      "epoch": 0.5769791160757649,
      "grad_norm": 0.9289069913292367,
      "learning_rate": 4.422310313447522e-06,
      "loss": 0.3745,
      "step": 10692
    },
    {
      "epoch": 0.5770330797042793,
      "grad_norm": 0.9805429895073225,
      "learning_rate": 4.421569636545666e-06,
      "loss": 0.5291,
      "step": 10693
    },
    {
      "epoch": 0.5770870433327937,
      "grad_norm": 0.959044927461277,
      "learning_rate": 4.42082899063899e-06,
      "loss": 0.4176,
      "step": 10694
    },
    {
      "epoch": 0.5771410069613081,
      "grad_norm": 0.9193011070176517,
      "learning_rate": 4.420088375748779e-06,
      "loss": 0.5035,
      "step": 10695
    },
    {
      "epoch": 0.5771949705898225,
      "grad_norm": 0.9038313444328734,
      "learning_rate": 4.419347791896318e-06,
      "loss": 0.4653,
      "step": 10696
    },
    {
      "epoch": 0.5772489342183369,
      "grad_norm": 1.1645046616095565,
      "learning_rate": 4.418607239102894e-06,
      "loss": 0.6183,
      "step": 10697
    },
    {
      "epoch": 0.5773028978468512,
      "grad_norm": 1.068819902974211,
      "learning_rate": 4.417866717389793e-06,
      "loss": 0.5308,
      "step": 10698
    },
    {
      "epoch": 0.5773568614753656,
      "grad_norm": 0.922305986933035,
      "learning_rate": 4.417126226778293e-06,
      "loss": 0.4655,
      "step": 10699
    },
    {
      "epoch": 0.57741082510388,
      "grad_norm": 1.0278258551557862,
      "learning_rate": 4.416385767289681e-06,
      "loss": 0.5378,
      "step": 10700
    },
    {
      "epoch": 0.5774647887323944,
      "grad_norm": 0.9128904159691642,
      "learning_rate": 4.415645338945236e-06,
      "loss": 0.4132,
      "step": 10701
    },
    {
      "epoch": 0.5775187523609088,
      "grad_norm": 1.010410235458784,
      "learning_rate": 4.4149049417662394e-06,
      "loss": 0.5185,
      "step": 10702
    },
    {
      "epoch": 0.5775727159894232,
      "grad_norm": 0.925047176939598,
      "learning_rate": 4.414164575773971e-06,
      "loss": 0.4077,
      "step": 10703
    },
    {
      "epoch": 0.5776266796179375,
      "grad_norm": 1.1104317523572533,
      "learning_rate": 4.413424240989712e-06,
      "loss": 0.5038,
      "step": 10704
    },
    {
      "epoch": 0.5776806432464519,
      "grad_norm": 1.151220457906447,
      "learning_rate": 4.412683937434736e-06,
      "loss": 0.5185,
      "step": 10705
    },
    {
      "epoch": 0.5777346068749662,
      "grad_norm": 0.972394500919524,
      "learning_rate": 4.411943665130324e-06,
      "loss": 0.4348,
      "step": 10706
    },
    {
      "epoch": 0.5777885705034806,
      "grad_norm": 1.0822357706189414,
      "learning_rate": 4.41120342409775e-06,
      "loss": 0.4733,
      "step": 10707
    },
    {
      "epoch": 0.577842534131995,
      "grad_norm": 0.9794546826784469,
      "learning_rate": 4.410463214358289e-06,
      "loss": 0.4705,
      "step": 10708
    },
    {
      "epoch": 0.5778964977605094,
      "grad_norm": 0.9326873245421238,
      "learning_rate": 4.409723035933217e-06,
      "loss": 0.3413,
      "step": 10709
    },
    {
      "epoch": 0.5779504613890238,
      "grad_norm": 1.1384562096545403,
      "learning_rate": 4.408982888843804e-06,
      "loss": 0.5466,
      "step": 10710
    },
    {
      "epoch": 0.5780044250175381,
      "grad_norm": 1.0499275563929176,
      "learning_rate": 4.4082427731113285e-06,
      "loss": 0.4472,
      "step": 10711
    },
    {
      "epoch": 0.5780583886460525,
      "grad_norm": 0.9693497199020219,
      "learning_rate": 4.407502688757057e-06,
      "loss": 0.4061,
      "step": 10712
    },
    {
      "epoch": 0.5781123522745669,
      "grad_norm": 0.9586103448723158,
      "learning_rate": 4.406762635802263e-06,
      "loss": 0.4148,
      "step": 10713
    },
    {
      "epoch": 0.5781663159030813,
      "grad_norm": 1.1252187925730974,
      "learning_rate": 4.406022614268215e-06,
      "loss": 0.5635,
      "step": 10714
    },
    {
      "epoch": 0.5782202795315957,
      "grad_norm": 1.2956607810945564,
      "learning_rate": 4.4052826241761825e-06,
      "loss": 0.4976,
      "step": 10715
    },
    {
      "epoch": 0.5782742431601101,
      "grad_norm": 1.0677613788344091,
      "learning_rate": 4.404542665547433e-06,
      "loss": 0.4254,
      "step": 10716
    },
    {
      "epoch": 0.5783282067886245,
      "grad_norm": 0.8851473988704571,
      "learning_rate": 4.403802738403235e-06,
      "loss": 0.4147,
      "step": 10717
    },
    {
      "epoch": 0.5783821704171388,
      "grad_norm": 0.9811366972264037,
      "learning_rate": 4.403062842764855e-06,
      "loss": 0.5393,
      "step": 10718
    },
    {
      "epoch": 0.5784361340456532,
      "grad_norm": 1.173380936114301,
      "learning_rate": 4.402322978653556e-06,
      "loss": 0.6141,
      "step": 10719
    },
    {
      "epoch": 0.5784900976741676,
      "grad_norm": 0.9562298706410602,
      "learning_rate": 4.401583146090604e-06,
      "loss": 0.4646,
      "step": 10720
    },
    {
      "epoch": 0.578544061302682,
      "grad_norm": 1.465345874806283,
      "learning_rate": 4.400843345097261e-06,
      "loss": 0.5805,
      "step": 10721
    },
    {
      "epoch": 0.5785980249311964,
      "grad_norm": 0.8865917511141234,
      "learning_rate": 4.400103575694792e-06,
      "loss": 0.3894,
      "step": 10722
    },
    {
      "epoch": 0.5786519885597108,
      "grad_norm": 1.0434905189899335,
      "learning_rate": 4.3993638379044565e-06,
      "loss": 0.3687,
      "step": 10723
    },
    {
      "epoch": 0.5787059521882252,
      "grad_norm": 0.9350023757670994,
      "learning_rate": 4.398624131747518e-06,
      "loss": 0.5278,
      "step": 10724
    },
    {
      "epoch": 0.5787599158167395,
      "grad_norm": 0.889622592589819,
      "learning_rate": 4.397884457245235e-06,
      "loss": 0.3485,
      "step": 10725
    },
    {
      "epoch": 0.5788138794452539,
      "grad_norm": 0.9378720637746693,
      "learning_rate": 4.397144814418865e-06,
      "loss": 0.404,
      "step": 10726
    },
    {
      "epoch": 0.5788678430737683,
      "grad_norm": 1.1750232080292062,
      "learning_rate": 4.396405203289668e-06,
      "loss": 0.5155,
      "step": 10727
    },
    {
      "epoch": 0.5789218067022827,
      "grad_norm": 0.9745553733650358,
      "learning_rate": 4.3956656238789e-06,
      "loss": 0.4584,
      "step": 10728
    },
    {
      "epoch": 0.5789757703307971,
      "grad_norm": 1.2098082755827806,
      "learning_rate": 4.394926076207818e-06,
      "loss": 0.5191,
      "step": 10729
    },
    {
      "epoch": 0.5790297339593115,
      "grad_norm": 1.0230243290545324,
      "learning_rate": 4.394186560297678e-06,
      "loss": 0.4768,
      "step": 10730
    },
    {
      "epoch": 0.5790836975878259,
      "grad_norm": 1.093045095738849,
      "learning_rate": 4.393447076169731e-06,
      "loss": 0.5434,
      "step": 10731
    },
    {
      "epoch": 0.5791376612163401,
      "grad_norm": 1.0880120882430226,
      "learning_rate": 4.392707623845237e-06,
      "loss": 0.4579,
      "step": 10732
    },
    {
      "epoch": 0.5791916248448545,
      "grad_norm": 0.8478929725453419,
      "learning_rate": 4.3919682033454425e-06,
      "loss": 0.4097,
      "step": 10733
    },
    {
      "epoch": 0.5792455884733689,
      "grad_norm": 1.0737463149842836,
      "learning_rate": 4.391228814691601e-06,
      "loss": 0.5087,
      "step": 10734
    },
    {
      "epoch": 0.5792995521018833,
      "grad_norm": 1.0005002300606136,
      "learning_rate": 4.390489457904964e-06,
      "loss": 0.4141,
      "step": 10735
    },
    {
      "epoch": 0.5793535157303977,
      "grad_norm": 1.084228463245065,
      "learning_rate": 4.3897501330067815e-06,
      "loss": 0.4655,
      "step": 10736
    },
    {
      "epoch": 0.5794074793589121,
      "grad_norm": 1.0542054481846073,
      "learning_rate": 4.389010840018302e-06,
      "loss": 0.517,
      "step": 10737
    },
    {
      "epoch": 0.5794614429874265,
      "grad_norm": 0.9412062021707129,
      "learning_rate": 4.388271578960773e-06,
      "loss": 0.3553,
      "step": 10738
    },
    {
      "epoch": 0.5795154066159408,
      "grad_norm": 1.0003873180596075,
      "learning_rate": 4.387532349855444e-06,
      "loss": 0.3571,
      "step": 10739
    },
    {
      "epoch": 0.5795693702444552,
      "grad_norm": 1.0103696822250403,
      "learning_rate": 4.386793152723558e-06,
      "loss": 0.3991,
      "step": 10740
    },
    {
      "epoch": 0.5796233338729696,
      "grad_norm": 1.1160594619609254,
      "learning_rate": 4.386053987586361e-06,
      "loss": 0.528,
      "step": 10741
    },
    {
      "epoch": 0.579677297501484,
      "grad_norm": 1.0247355203711692,
      "learning_rate": 4.385314854465099e-06,
      "loss": 0.4178,
      "step": 10742
    },
    {
      "epoch": 0.5797312611299984,
      "grad_norm": 1.1503297863100375,
      "learning_rate": 4.384575753381014e-06,
      "loss": 0.5421,
      "step": 10743
    },
    {
      "epoch": 0.5797852247585128,
      "grad_norm": 1.032341492681949,
      "learning_rate": 4.383836684355348e-06,
      "loss": 0.5091,
      "step": 10744
    },
    {
      "epoch": 0.5798391883870272,
      "grad_norm": 0.9398492212170682,
      "learning_rate": 4.383097647409345e-06,
      "loss": 0.3908,
      "step": 10745
    },
    {
      "epoch": 0.5798931520155415,
      "grad_norm": 0.9183518931796758,
      "learning_rate": 4.382358642564244e-06,
      "loss": 0.4364,
      "step": 10746
    },
    {
      "epoch": 0.5799471156440559,
      "grad_norm": 0.83805866417832,
      "learning_rate": 4.381619669841285e-06,
      "loss": 0.321,
      "step": 10747
    },
    {
      "epoch": 0.5800010792725703,
      "grad_norm": 1.0194551879246887,
      "learning_rate": 4.380880729261705e-06,
      "loss": 0.3638,
      "step": 10748
    },
    {
      "epoch": 0.5800550429010847,
      "grad_norm": 0.9561142511228675,
      "learning_rate": 4.380141820846745e-06,
      "loss": 0.5029,
      "step": 10749
    },
    {
      "epoch": 0.5801090065295991,
      "grad_norm": 1.0704908048926518,
      "learning_rate": 4.37940294461764e-06,
      "loss": 0.4526,
      "step": 10750
    },
    {
      "epoch": 0.5801629701581135,
      "grad_norm": 1.287964066610987,
      "learning_rate": 4.378664100595627e-06,
      "loss": 0.4876,
      "step": 10751
    },
    {
      "epoch": 0.5802169337866279,
      "grad_norm": 0.8227485709827045,
      "learning_rate": 4.377925288801941e-06,
      "loss": 0.3847,
      "step": 10752
    },
    {
      "epoch": 0.5802708974151422,
      "grad_norm": 0.7725858120272292,
      "learning_rate": 4.377186509257817e-06,
      "loss": 0.3806,
      "step": 10753
    },
    {
      "epoch": 0.5803248610436565,
      "grad_norm": 0.9012299215742978,
      "learning_rate": 4.376447761984486e-06,
      "loss": 0.4186,
      "step": 10754
    },
    {
      "epoch": 0.580378824672171,
      "grad_norm": 1.0802492000612998,
      "learning_rate": 4.375709047003181e-06,
      "loss": 0.5985,
      "step": 10755
    },
    {
      "epoch": 0.5804327883006853,
      "grad_norm": 0.9649869887112489,
      "learning_rate": 4.374970364335135e-06,
      "loss": 0.3539,
      "step": 10756
    },
    {
      "epoch": 0.5804867519291997,
      "grad_norm": 0.9502144952947014,
      "learning_rate": 4.374231714001576e-06,
      "loss": 0.4596,
      "step": 10757
    },
    {
      "epoch": 0.5805407155577141,
      "grad_norm": 0.9863384687469474,
      "learning_rate": 4.3734930960237365e-06,
      "loss": 0.4203,
      "step": 10758
    },
    {
      "epoch": 0.5805946791862285,
      "grad_norm": 1.271304599468789,
      "learning_rate": 4.372754510422843e-06,
      "loss": 0.4717,
      "step": 10759
    },
    {
      "epoch": 0.5806486428147428,
      "grad_norm": 1.0226275209427604,
      "learning_rate": 4.372015957220125e-06,
      "loss": 0.4646,
      "step": 10760
    },
    {
      "epoch": 0.5807026064432572,
      "grad_norm": 0.8933758352341065,
      "learning_rate": 4.371277436436806e-06,
      "loss": 0.2568,
      "step": 10761
    },
    {
      "epoch": 0.5807565700717716,
      "grad_norm": 0.9956996771238652,
      "learning_rate": 4.3705389480941165e-06,
      "loss": 0.5058,
      "step": 10762
    },
    {
      "epoch": 0.580810533700286,
      "grad_norm": 1.0226492048875213,
      "learning_rate": 4.369800492213277e-06,
      "loss": 0.4423,
      "step": 10763
    },
    {
      "epoch": 0.5808644973288004,
      "grad_norm": 0.8527859636131525,
      "learning_rate": 4.369062068815514e-06,
      "loss": 0.4503,
      "step": 10764
    },
    {
      "epoch": 0.5809184609573148,
      "grad_norm": 0.9321737229116237,
      "learning_rate": 4.368323677922048e-06,
      "loss": 0.3511,
      "step": 10765
    },
    {
      "epoch": 0.5809724245858292,
      "grad_norm": 0.9270458977674326,
      "learning_rate": 4.367585319554105e-06,
      "loss": 0.4409,
      "step": 10766
    },
    {
      "epoch": 0.5810263882143435,
      "grad_norm": 1.0391385472247665,
      "learning_rate": 4.366846993732906e-06,
      "loss": 0.4824,
      "step": 10767
    },
    {
      "epoch": 0.5810803518428579,
      "grad_norm": 0.9624931505974168,
      "learning_rate": 4.366108700479667e-06,
      "loss": 0.4835,
      "step": 10768
    },
    {
      "epoch": 0.5811343154713723,
      "grad_norm": 0.7472747789017256,
      "learning_rate": 4.36537043981561e-06,
      "loss": 0.3508,
      "step": 10769
    },
    {
      "epoch": 0.5811882790998867,
      "grad_norm": 1.1021552275199522,
      "learning_rate": 4.364632211761953e-06,
      "loss": 0.5976,
      "step": 10770
    },
    {
      "epoch": 0.5812422427284011,
      "grad_norm": 1.1204508576785062,
      "learning_rate": 4.363894016339914e-06,
      "loss": 0.561,
      "step": 10771
    },
    {
      "epoch": 0.5812962063569155,
      "grad_norm": 1.0369134733842182,
      "learning_rate": 4.363155853570708e-06,
      "loss": 0.3448,
      "step": 10772
    },
    {
      "epoch": 0.5813501699854298,
      "grad_norm": 1.1810078589221615,
      "learning_rate": 4.362417723475553e-06,
      "loss": 0.666,
      "step": 10773
    },
    {
      "epoch": 0.5814041336139442,
      "grad_norm": 0.7891491105205495,
      "learning_rate": 4.361679626075661e-06,
      "loss": 0.2872,
      "step": 10774
    },
    {
      "epoch": 0.5814580972424586,
      "grad_norm": 0.9043243054565318,
      "learning_rate": 4.360941561392246e-06,
      "loss": 0.4276,
      "step": 10775
    },
    {
      "epoch": 0.581512060870973,
      "grad_norm": 1.1136636303600596,
      "learning_rate": 4.360203529446522e-06,
      "loss": 0.5374,
      "step": 10776
    },
    {
      "epoch": 0.5815660244994874,
      "grad_norm": 1.2072567896009667,
      "learning_rate": 4.3594655302597e-06,
      "loss": 0.3906,
      "step": 10777
    },
    {
      "epoch": 0.5816199881280018,
      "grad_norm": 0.8975909923707904,
      "learning_rate": 4.3587275638529906e-06,
      "loss": 0.3099,
      "step": 10778
    },
    {
      "epoch": 0.5816739517565161,
      "grad_norm": 0.8710788601996851,
      "learning_rate": 4.357989630247606e-06,
      "loss": 0.3546,
      "step": 10779
    },
    {
      "epoch": 0.5817279153850304,
      "grad_norm": 1.1459380014212872,
      "learning_rate": 4.357251729464752e-06,
      "loss": 0.4851,
      "step": 10780
    },
    {
      "epoch": 0.5817818790135448,
      "grad_norm": 0.9658110368898553,
      "learning_rate": 4.356513861525639e-06,
      "loss": 0.4495,
      "step": 10781
    },
    {
      "epoch": 0.5818358426420592,
      "grad_norm": 0.7521307472138556,
      "learning_rate": 4.355776026451473e-06,
      "loss": 0.3269,
      "step": 10782
    },
    {
      "epoch": 0.5818898062705736,
      "grad_norm": 0.967765688113511,
      "learning_rate": 4.355038224263459e-06,
      "loss": 0.4625,
      "step": 10783
    },
    {
      "epoch": 0.581943769899088,
      "grad_norm": 1.0955319101143879,
      "learning_rate": 4.354300454982803e-06,
      "loss": 0.4658,
      "step": 10784
    },
    {
      "epoch": 0.5819977335276024,
      "grad_norm": 0.9261400265988177,
      "learning_rate": 4.353562718630711e-06,
      "loss": 0.4705,
      "step": 10785
    },
    {
      "epoch": 0.5820516971561168,
      "grad_norm": 1.042367974870492,
      "learning_rate": 4.352825015228384e-06,
      "loss": 0.5001,
      "step": 10786
    },
    {
      "epoch": 0.5821056607846311,
      "grad_norm": 0.9510692325560809,
      "learning_rate": 4.352087344797026e-06,
      "loss": 0.5201,
      "step": 10787
    },
    {
      "epoch": 0.5821596244131455,
      "grad_norm": 1.3282907518336626,
      "learning_rate": 4.351349707357836e-06,
      "loss": 0.5732,
      "step": 10788
    },
    {
      "epoch": 0.5822135880416599,
      "grad_norm": 0.9341402362951141,
      "learning_rate": 4.350612102932017e-06,
      "loss": 0.5467,
      "step": 10789
    },
    {
      "epoch": 0.5822675516701743,
      "grad_norm": 0.9752629086916158,
      "learning_rate": 4.3498745315407675e-06,
      "loss": 0.4426,
      "step": 10790
    },
    {
      "epoch": 0.5823215152986887,
      "grad_norm": 0.8748259599752354,
      "learning_rate": 4.349136993205286e-06,
      "loss": 0.4168,
      "step": 10791
    },
    {
      "epoch": 0.5823754789272031,
      "grad_norm": 0.8329306424613497,
      "learning_rate": 4.34839948794677e-06,
      "loss": 0.3949,
      "step": 10792
    },
    {
      "epoch": 0.5824294425557175,
      "grad_norm": 0.9880349383847927,
      "learning_rate": 4.347662015786417e-06,
      "loss": 0.382,
      "step": 10793
    },
    {
      "epoch": 0.5824834061842318,
      "grad_norm": 0.9321874005482529,
      "learning_rate": 4.346924576745423e-06,
      "loss": 0.2706,
      "step": 10794
    },
    {
      "epoch": 0.5825373698127462,
      "grad_norm": 0.8658902059789751,
      "learning_rate": 4.3461871708449796e-06,
      "loss": 0.3523,
      "step": 10795
    },
    {
      "epoch": 0.5825913334412606,
      "grad_norm": 1.0350407139473208,
      "learning_rate": 4.3454497981062835e-06,
      "loss": 0.4838,
      "step": 10796
    },
    {
      "epoch": 0.582645297069775,
      "grad_norm": 1.2771424303257874,
      "learning_rate": 4.3447124585505275e-06,
      "loss": 0.5801,
      "step": 10797
    },
    {
      "epoch": 0.5826992606982894,
      "grad_norm": 1.3376493542889678,
      "learning_rate": 4.343975152198902e-06,
      "loss": 0.5823,
      "step": 10798
    },
    {
      "epoch": 0.5827532243268038,
      "grad_norm": 0.9344517897367045,
      "learning_rate": 4.343237879072598e-06,
      "loss": 0.3396,
      "step": 10799
    },
    {
      "epoch": 0.5828071879553182,
      "grad_norm": 1.2324296848730487,
      "learning_rate": 4.342500639192807e-06,
      "loss": 0.5686,
      "step": 10800
    },
    {
      "epoch": 0.5828611515838324,
      "grad_norm": 1.0680883324252228,
      "learning_rate": 4.341763432580719e-06,
      "loss": 0.4781,
      "step": 10801
    },
    {
      "epoch": 0.5829151152123468,
      "grad_norm": 0.799447953191825,
      "learning_rate": 4.3410262592575184e-06,
      "loss": 0.354,
      "step": 10802
    },
    {
      "epoch": 0.5829690788408612,
      "grad_norm": 1.1411180690895073,
      "learning_rate": 4.340289119244394e-06,
      "loss": 0.5158,
      "step": 10803
    },
    {
      "epoch": 0.5830230424693756,
      "grad_norm": 1.1657415667613888,
      "learning_rate": 4.339552012562532e-06,
      "loss": 0.5316,
      "step": 10804
    },
    {
      "epoch": 0.58307700609789,
      "grad_norm": 1.0247373722521023,
      "learning_rate": 4.338814939233118e-06,
      "loss": 0.4158,
      "step": 10805
    },
    {
      "epoch": 0.5831309697264044,
      "grad_norm": 0.9201124043069568,
      "learning_rate": 4.338077899277337e-06,
      "loss": 0.3458,
      "step": 10806
    },
    {
      "epoch": 0.5831849333549188,
      "grad_norm": 1.0067653679319337,
      "learning_rate": 4.337340892716369e-06,
      "loss": 0.4111,
      "step": 10807
    },
    {
      "epoch": 0.5832388969834331,
      "grad_norm": 0.8808130783274866,
      "learning_rate": 4.3366039195714e-06,
      "loss": 0.3608,
      "step": 10808
    },
    {
      "epoch": 0.5832928606119475,
      "grad_norm": 1.1524358544832196,
      "learning_rate": 4.33586697986361e-06,
      "loss": 0.4674,
      "step": 10809
    },
    {
      "epoch": 0.5833468242404619,
      "grad_norm": 0.9515742596894681,
      "learning_rate": 4.335130073614179e-06,
      "loss": 0.4628,
      "step": 10810
    },
    {
      "epoch": 0.5834007878689763,
      "grad_norm": 1.2789779005108033,
      "learning_rate": 4.334393200844287e-06,
      "loss": 0.6095,
      "step": 10811
    },
    {
      "epoch": 0.5834547514974907,
      "grad_norm": 1.0708047013285145,
      "learning_rate": 4.333656361575112e-06,
      "loss": 0.5121,
      "step": 10812
    },
    {
      "epoch": 0.5835087151260051,
      "grad_norm": 0.7750442829321201,
      "learning_rate": 4.332919555827831e-06,
      "loss": 0.3292,
      "step": 10813
    },
    {
      "epoch": 0.5835626787545195,
      "grad_norm": 1.1093352160515706,
      "learning_rate": 4.332182783623622e-06,
      "loss": 0.5163,
      "step": 10814
    },
    {
      "epoch": 0.5836166423830338,
      "grad_norm": 0.9674289541564522,
      "learning_rate": 4.33144604498366e-06,
      "loss": 0.3577,
      "step": 10815
    },
    {
      "epoch": 0.5836706060115482,
      "grad_norm": 1.0330852815972185,
      "learning_rate": 4.330709339929118e-06,
      "loss": 0.4843,
      "step": 10816
    },
    {
      "epoch": 0.5837245696400626,
      "grad_norm": 0.9367013911498115,
      "learning_rate": 4.329972668481173e-06,
      "loss": 0.368,
      "step": 10817
    },
    {
      "epoch": 0.583778533268577,
      "grad_norm": 0.6968575785944818,
      "learning_rate": 4.329236030660994e-06,
      "loss": 0.3159,
      "step": 10818
    },
    {
      "epoch": 0.5838324968970914,
      "grad_norm": 1.1251998352061752,
      "learning_rate": 4.328499426489754e-06,
      "loss": 0.5352,
      "step": 10819
    },
    {
      "epoch": 0.5838864605256058,
      "grad_norm": 0.9258307630179584,
      "learning_rate": 4.3277628559886254e-06,
      "loss": 0.4271,
      "step": 10820
    },
    {
      "epoch": 0.5839404241541202,
      "grad_norm": 1.0012220052547136,
      "learning_rate": 4.327026319178776e-06,
      "loss": 0.5471,
      "step": 10821
    },
    {
      "epoch": 0.5839943877826345,
      "grad_norm": 0.9385059050329881,
      "learning_rate": 4.3262898160813786e-06,
      "loss": 0.3767,
      "step": 10822
    },
    {
      "epoch": 0.5840483514111489,
      "grad_norm": 1.083170503772952,
      "learning_rate": 4.325553346717595e-06,
      "loss": 0.4818,
      "step": 10823
    },
    {
      "epoch": 0.5841023150396633,
      "grad_norm": 0.8806323862327279,
      "learning_rate": 4.324816911108594e-06,
      "loss": 0.3769,
      "step": 10824
    },
    {
      "epoch": 0.5841562786681777,
      "grad_norm": 1.159151686787587,
      "learning_rate": 4.324080509275544e-06,
      "loss": 0.5388,
      "step": 10825
    },
    {
      "epoch": 0.584210242296692,
      "grad_norm": 0.8232237332657785,
      "learning_rate": 4.323344141239608e-06,
      "loss": 0.3064,
      "step": 10826
    },
    {
      "epoch": 0.5842642059252064,
      "grad_norm": 0.8778969859905339,
      "learning_rate": 4.32260780702195e-06,
      "loss": 0.3977,
      "step": 10827
    },
    {
      "epoch": 0.5843181695537208,
      "grad_norm": 1.212872101587933,
      "learning_rate": 4.321871506643733e-06,
      "loss": 0.5722,
      "step": 10828
    },
    {
      "epoch": 0.5843721331822351,
      "grad_norm": 0.9882109998239202,
      "learning_rate": 4.321135240126122e-06,
      "loss": 0.5125,
      "step": 10829
    },
    {
      "epoch": 0.5844260968107495,
      "grad_norm": 1.08578912023213,
      "learning_rate": 4.320399007490273e-06,
      "loss": 0.406,
      "step": 10830
    },
    {
      "epoch": 0.5844800604392639,
      "grad_norm": 0.9167977630868087,
      "learning_rate": 4.319662808757349e-06,
      "loss": 0.3846,
      "step": 10831
    },
    {
      "epoch": 0.5845340240677783,
      "grad_norm": 0.8798173879334827,
      "learning_rate": 4.318926643948507e-06,
      "loss": 0.3324,
      "step": 10832
    },
    {
      "epoch": 0.5845879876962927,
      "grad_norm": 0.9011126364450855,
      "learning_rate": 4.318190513084908e-06,
      "loss": 0.4119,
      "step": 10833
    },
    {
      "epoch": 0.5846419513248071,
      "grad_norm": 0.9703719198288479,
      "learning_rate": 4.317454416187707e-06,
      "loss": 0.3779,
      "step": 10834
    },
    {
      "epoch": 0.5846959149533215,
      "grad_norm": 1.1866939449654437,
      "learning_rate": 4.316718353278062e-06,
      "loss": 0.4641,
      "step": 10835
    },
    {
      "epoch": 0.5847498785818358,
      "grad_norm": 0.8453886282589783,
      "learning_rate": 4.315982324377127e-06,
      "loss": 0.376,
      "step": 10836
    },
    {
      "epoch": 0.5848038422103502,
      "grad_norm": 1.124974207219576,
      "learning_rate": 4.315246329506057e-06,
      "loss": 0.5529,
      "step": 10837
    },
    {
      "epoch": 0.5848578058388646,
      "grad_norm": 1.0374886720657028,
      "learning_rate": 4.314510368686002e-06,
      "loss": 0.664,
      "step": 10838
    },
    {
      "epoch": 0.584911769467379,
      "grad_norm": 1.0229583540418845,
      "learning_rate": 4.313774441938119e-06,
      "loss": 0.5487,
      "step": 10839
    },
    {
      "epoch": 0.5849657330958934,
      "grad_norm": 1.0790251155164094,
      "learning_rate": 4.313038549283556e-06,
      "loss": 0.5783,
      "step": 10840
    },
    {
      "epoch": 0.5850196967244078,
      "grad_norm": 1.0428318145074607,
      "learning_rate": 4.312302690743465e-06,
      "loss": 0.4164,
      "step": 10841
    },
    {
      "epoch": 0.5850736603529221,
      "grad_norm": 1.0282500801496532,
      "learning_rate": 4.311566866338996e-06,
      "loss": 0.3854,
      "step": 10842
    },
    {
      "epoch": 0.5851276239814365,
      "grad_norm": 1.133535232107666,
      "learning_rate": 4.310831076091295e-06,
      "loss": 0.7857,
      "step": 10843
    },
    {
      "epoch": 0.5851815876099509,
      "grad_norm": 0.6956020773895513,
      "learning_rate": 4.31009532002151e-06,
      "loss": 0.2686,
      "step": 10844
    },
    {
      "epoch": 0.5852355512384653,
      "grad_norm": 1.0546692555437611,
      "learning_rate": 4.309359598150789e-06,
      "loss": 0.4177,
      "step": 10845
    },
    {
      "epoch": 0.5852895148669797,
      "grad_norm": 1.0277867848398559,
      "learning_rate": 4.308623910500275e-06,
      "loss": 0.4699,
      "step": 10846
    },
    {
      "epoch": 0.5853434784954941,
      "grad_norm": 1.2401841224841403,
      "learning_rate": 4.307888257091113e-06,
      "loss": 0.6137,
      "step": 10847
    },
    {
      "epoch": 0.5853974421240085,
      "grad_norm": 0.8335089252766369,
      "learning_rate": 4.3071526379444486e-06,
      "loss": 0.397,
      "step": 10848
    },
    {
      "epoch": 0.5854514057525227,
      "grad_norm": 0.9933158566030164,
      "learning_rate": 4.306417053081424e-06,
      "loss": 0.6724,
      "step": 10849
    },
    {
      "epoch": 0.5855053693810371,
      "grad_norm": 1.091579636715559,
      "learning_rate": 4.305681502523177e-06,
      "loss": 0.3932,
      "step": 10850
    },
    {
      "epoch": 0.5855593330095515,
      "grad_norm": 0.8379033096099369,
      "learning_rate": 4.304945986290851e-06,
      "loss": 0.3129,
      "step": 10851
    },
    {
      "epoch": 0.5856132966380659,
      "grad_norm": 1.038774404522165,
      "learning_rate": 4.304210504405584e-06,
      "loss": 0.5046,
      "step": 10852
    },
    {
      "epoch": 0.5856672602665803,
      "grad_norm": 1.2852988567723174,
      "learning_rate": 4.303475056888515e-06,
      "loss": 0.5627,
      "step": 10853
    },
    {
      "epoch": 0.5857212238950947,
      "grad_norm": 0.8225177280784101,
      "learning_rate": 4.302739643760782e-06,
      "loss": 0.3726,
      "step": 10854
    },
    {
      "epoch": 0.5857751875236091,
      "grad_norm": 1.045845087073345,
      "learning_rate": 4.302004265043523e-06,
      "loss": 0.4598,
      "step": 10855
    },
    {
      "epoch": 0.5858291511521234,
      "grad_norm": 1.0717066151245913,
      "learning_rate": 4.301268920757873e-06,
      "loss": 0.5606,
      "step": 10856
    },
    {
      "epoch": 0.5858831147806378,
      "grad_norm": 0.8963611134135105,
      "learning_rate": 4.300533610924963e-06,
      "loss": 0.478,
      "step": 10857
    },
    {
      "epoch": 0.5859370784091522,
      "grad_norm": 0.8709740646325408,
      "learning_rate": 4.29979833556593e-06,
      "loss": 0.383,
      "step": 10858
    },
    {
      "epoch": 0.5859910420376666,
      "grad_norm": 1.0690207644758276,
      "learning_rate": 4.299063094701905e-06,
      "loss": 0.4522,
      "step": 10859
    },
    {
      "epoch": 0.586045005666181,
      "grad_norm": 0.8937104318276762,
      "learning_rate": 4.29832788835402e-06,
      "loss": 0.3466,
      "step": 10860
    },
    {
      "epoch": 0.5860989692946954,
      "grad_norm": 0.9421285976601806,
      "learning_rate": 4.297592716543406e-06,
      "loss": 0.4696,
      "step": 10861
    },
    {
      "epoch": 0.5861529329232098,
      "grad_norm": 0.9720992628483462,
      "learning_rate": 4.296857579291191e-06,
      "loss": 0.4732,
      "step": 10862
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 1.0572632106703115,
      "learning_rate": 4.296122476618507e-06,
      "loss": 0.5633,
      "step": 10863
    },
    {
      "epoch": 0.5862608601802385,
      "grad_norm": 1.108269161130122,
      "learning_rate": 4.295387408546479e-06,
      "loss": 0.5243,
      "step": 10864
    },
    {
      "epoch": 0.5863148238087529,
      "grad_norm": 1.0534266002803678,
      "learning_rate": 4.294652375096233e-06,
      "loss": 0.5458,
      "step": 10865
    },
    {
      "epoch": 0.5863687874372673,
      "grad_norm": 1.1145857121059677,
      "learning_rate": 4.293917376288897e-06,
      "loss": 0.535,
      "step": 10866
    },
    {
      "epoch": 0.5864227510657817,
      "grad_norm": 1.151301459396368,
      "learning_rate": 4.293182412145592e-06,
      "loss": 0.527,
      "step": 10867
    },
    {
      "epoch": 0.5864767146942961,
      "grad_norm": 1.0733259557287107,
      "learning_rate": 4.292447482687446e-06,
      "loss": 0.4791,
      "step": 10868
    },
    {
      "epoch": 0.5865306783228105,
      "grad_norm": 0.8876370979203263,
      "learning_rate": 4.291712587935578e-06,
      "loss": 0.3931,
      "step": 10869
    },
    {
      "epoch": 0.5865846419513248,
      "grad_norm": 0.9630137750287598,
      "learning_rate": 4.290977727911114e-06,
      "loss": 0.3678,
      "step": 10870
    },
    {
      "epoch": 0.5866386055798392,
      "grad_norm": 1.0168923090682573,
      "learning_rate": 4.29024290263517e-06,
      "loss": 0.4729,
      "step": 10871
    },
    {
      "epoch": 0.5866925692083536,
      "grad_norm": 0.9943916396497462,
      "learning_rate": 4.289508112128866e-06,
      "loss": 0.4032,
      "step": 10872
    },
    {
      "epoch": 0.586746532836868,
      "grad_norm": 0.9730199429147925,
      "learning_rate": 4.288773356413324e-06,
      "loss": 0.4756,
      "step": 10873
    },
    {
      "epoch": 0.5868004964653823,
      "grad_norm": 1.0029776063797764,
      "learning_rate": 4.288038635509658e-06,
      "loss": 0.4288,
      "step": 10874
    },
    {
      "epoch": 0.5868544600938967,
      "grad_norm": 1.0779102457076701,
      "learning_rate": 4.287303949438987e-06,
      "loss": 0.5281,
      "step": 10875
    },
    {
      "epoch": 0.5869084237224111,
      "grad_norm": 0.8321280427550214,
      "learning_rate": 4.286569298222425e-06,
      "loss": 0.3829,
      "step": 10876
    },
    {
      "epoch": 0.5869623873509254,
      "grad_norm": 1.1089223767173197,
      "learning_rate": 4.285834681881089e-06,
      "loss": 0.5705,
      "step": 10877
    },
    {
      "epoch": 0.5870163509794398,
      "grad_norm": 1.0358090001915436,
      "learning_rate": 4.2851001004360905e-06,
      "loss": 0.4361,
      "step": 10878
    },
    {
      "epoch": 0.5870703146079542,
      "grad_norm": 0.9789316428592371,
      "learning_rate": 4.284365553908543e-06,
      "loss": 0.4622,
      "step": 10879
    },
    {
      "epoch": 0.5871242782364686,
      "grad_norm": 1.0251788989402013,
      "learning_rate": 4.2836310423195556e-06,
      "loss": 0.5573,
      "step": 10880
    },
    {
      "epoch": 0.587178241864983,
      "grad_norm": 1.2117162583343548,
      "learning_rate": 4.282896565690243e-06,
      "loss": 0.5384,
      "step": 10881
    },
    {
      "epoch": 0.5872322054934974,
      "grad_norm": 1.1053199197618562,
      "learning_rate": 4.282162124041712e-06,
      "loss": 0.5394,
      "step": 10882
    },
    {
      "epoch": 0.5872861691220118,
      "grad_norm": 0.8952581201269858,
      "learning_rate": 4.281427717395072e-06,
      "loss": 0.3734,
      "step": 10883
    },
    {
      "epoch": 0.5873401327505261,
      "grad_norm": 0.9335331053590437,
      "learning_rate": 4.280693345771433e-06,
      "loss": 0.4799,
      "step": 10884
    },
    {
      "epoch": 0.5873940963790405,
      "grad_norm": 0.9379927928947436,
      "learning_rate": 4.279959009191897e-06,
      "loss": 0.3793,
      "step": 10885
    },
    {
      "epoch": 0.5874480600075549,
      "grad_norm": 1.18284250982107,
      "learning_rate": 4.2792247076775726e-06,
      "loss": 0.5128,
      "step": 10886
    },
    {
      "epoch": 0.5875020236360693,
      "grad_norm": 1.0331819794274721,
      "learning_rate": 4.278490441249563e-06,
      "loss": 0.3905,
      "step": 10887
    },
    {
      "epoch": 0.5875559872645837,
      "grad_norm": 1.1079970568605035,
      "learning_rate": 4.277756209928974e-06,
      "loss": 0.495,
      "step": 10888
    },
    {
      "epoch": 0.5876099508930981,
      "grad_norm": 1.0994439850388666,
      "learning_rate": 4.277022013736904e-06,
      "loss": 0.5144,
      "step": 10889
    },
    {
      "epoch": 0.5876639145216125,
      "grad_norm": 1.0032966823530296,
      "learning_rate": 4.276287852694458e-06,
      "loss": 0.4645,
      "step": 10890
    },
    {
      "epoch": 0.5877178781501268,
      "grad_norm": 1.056717497916176,
      "learning_rate": 4.275553726822738e-06,
      "loss": 0.4559,
      "step": 10891
    },
    {
      "epoch": 0.5877718417786412,
      "grad_norm": 1.0291998105957025,
      "learning_rate": 4.274819636142838e-06,
      "loss": 0.4223,
      "step": 10892
    },
    {
      "epoch": 0.5878258054071556,
      "grad_norm": 0.8278834667878908,
      "learning_rate": 4.27408558067586e-06,
      "loss": 0.3389,
      "step": 10893
    },
    {
      "epoch": 0.58787976903567,
      "grad_norm": 1.0500008397392324,
      "learning_rate": 4.2733515604429034e-06,
      "loss": 0.5824,
      "step": 10894
    },
    {
      "epoch": 0.5879337326641844,
      "grad_norm": 1.034437968727046,
      "learning_rate": 4.272617575465059e-06,
      "loss": 0.5581,
      "step": 10895
    },
    {
      "epoch": 0.5879876962926988,
      "grad_norm": 0.9031477917485106,
      "learning_rate": 4.271883625763427e-06,
      "loss": 0.3773,
      "step": 10896
    },
    {
      "epoch": 0.5880416599212132,
      "grad_norm": 1.0592669512390782,
      "learning_rate": 4.271149711359101e-06,
      "loss": 0.4596,
      "step": 10897
    },
    {
      "epoch": 0.5880956235497274,
      "grad_norm": 1.0367256768854838,
      "learning_rate": 4.270415832273176e-06,
      "loss": 0.4513,
      "step": 10898
    },
    {
      "epoch": 0.5881495871782418,
      "grad_norm": 1.076352784151165,
      "learning_rate": 4.26968198852674e-06,
      "loss": 0.4788,
      "step": 10899
    },
    {
      "epoch": 0.5882035508067562,
      "grad_norm": 1.0136429941729483,
      "learning_rate": 4.268948180140888e-06,
      "loss": 0.3847,
      "step": 10900
    },
    {
      "epoch": 0.5882575144352706,
      "grad_norm": 1.2676501530475186,
      "learning_rate": 4.268214407136707e-06,
      "loss": 0.5682,
      "step": 10901
    },
    {
      "epoch": 0.588311478063785,
      "grad_norm": 1.1276807041776558,
      "learning_rate": 4.2674806695352895e-06,
      "loss": 0.5382,
      "step": 10902
    },
    {
      "epoch": 0.5883654416922994,
      "grad_norm": 0.7867281974868271,
      "learning_rate": 4.266746967357723e-06,
      "loss": 0.3327,
      "step": 10903
    },
    {
      "epoch": 0.5884194053208138,
      "grad_norm": 0.8484602686788164,
      "learning_rate": 4.266013300625095e-06,
      "loss": 0.3477,
      "step": 10904
    },
    {
      "epoch": 0.5884733689493281,
      "grad_norm": 1.0067213999635962,
      "learning_rate": 4.265279669358494e-06,
      "loss": 0.4097,
      "step": 10905
    },
    {
      "epoch": 0.5885273325778425,
      "grad_norm": 1.0334072156383585,
      "learning_rate": 4.264546073579e-06,
      "loss": 0.4632,
      "step": 10906
    },
    {
      "epoch": 0.5885812962063569,
      "grad_norm": 0.9097239938002113,
      "learning_rate": 4.2638125133077005e-06,
      "loss": 0.3261,
      "step": 10907
    },
    {
      "epoch": 0.5886352598348713,
      "grad_norm": 1.1914365165735643,
      "learning_rate": 4.2630789885656786e-06,
      "loss": 0.6019,
      "step": 10908
    },
    {
      "epoch": 0.5886892234633857,
      "grad_norm": 1.0119170170784235,
      "learning_rate": 4.2623454993740165e-06,
      "loss": 0.4213,
      "step": 10909
    },
    {
      "epoch": 0.5887431870919001,
      "grad_norm": 1.11262761958445,
      "learning_rate": 4.2616120457537945e-06,
      "loss": 0.4331,
      "step": 10910
    },
    {
      "epoch": 0.5887971507204144,
      "grad_norm": 0.8625303474073208,
      "learning_rate": 4.260878627726095e-06,
      "loss": 0.3676,
      "step": 10911
    },
    {
      "epoch": 0.5888511143489288,
      "grad_norm": 1.1086800243531476,
      "learning_rate": 4.260145245311995e-06,
      "loss": 0.4922,
      "step": 10912
    },
    {
      "epoch": 0.5889050779774432,
      "grad_norm": 0.9305531051231767,
      "learning_rate": 4.259411898532573e-06,
      "loss": 0.7115,
      "step": 10913
    },
    {
      "epoch": 0.5889590416059576,
      "grad_norm": 1.1418046148283931,
      "learning_rate": 4.258678587408908e-06,
      "loss": 0.4106,
      "step": 10914
    },
    {
      "epoch": 0.589013005234472,
      "grad_norm": 1.0324747865478328,
      "learning_rate": 4.257945311962073e-06,
      "loss": 0.628,
      "step": 10915
    },
    {
      "epoch": 0.5890669688629864,
      "grad_norm": 0.7050640644084826,
      "learning_rate": 4.257212072213145e-06,
      "loss": 0.2229,
      "step": 10916
    },
    {
      "epoch": 0.5891209324915008,
      "grad_norm": 1.0445930200132412,
      "learning_rate": 4.256478868183198e-06,
      "loss": 0.6044,
      "step": 10917
    },
    {
      "epoch": 0.5891748961200151,
      "grad_norm": 0.9012154964583723,
      "learning_rate": 4.255745699893307e-06,
      "loss": 0.389,
      "step": 10918
    },
    {
      "epoch": 0.5892288597485295,
      "grad_norm": 1.0303306259436482,
      "learning_rate": 4.2550125673645396e-06,
      "loss": 0.4991,
      "step": 10919
    },
    {
      "epoch": 0.5892828233770439,
      "grad_norm": 0.9747157673974188,
      "learning_rate": 4.254279470617969e-06,
      "loss": 0.4067,
      "step": 10920
    },
    {
      "epoch": 0.5893367870055582,
      "grad_norm": 1.1064949097690493,
      "learning_rate": 4.253546409674664e-06,
      "loss": 0.5987,
      "step": 10921
    },
    {
      "epoch": 0.5893907506340726,
      "grad_norm": 1.138372555435741,
      "learning_rate": 4.252813384555697e-06,
      "loss": 0.6081,
      "step": 10922
    },
    {
      "epoch": 0.589444714262587,
      "grad_norm": 1.0446330340382701,
      "learning_rate": 4.252080395282132e-06,
      "loss": 0.465,
      "step": 10923
    },
    {
      "epoch": 0.5894986778911014,
      "grad_norm": 1.0389905734468081,
      "learning_rate": 4.251347441875039e-06,
      "loss": 0.4168,
      "step": 10924
    },
    {
      "epoch": 0.5895526415196157,
      "grad_norm": 0.75425978295224,
      "learning_rate": 4.250614524355482e-06,
      "loss": 0.379,
      "step": 10925
    },
    {
      "epoch": 0.5896066051481301,
      "grad_norm": 1.024857856384912,
      "learning_rate": 4.249881642744526e-06,
      "loss": 0.5811,
      "step": 10926
    },
    {
      "epoch": 0.5896605687766445,
      "grad_norm": 1.5484273708574183,
      "learning_rate": 4.249148797063233e-06,
      "loss": 0.4995,
      "step": 10927
    },
    {
      "epoch": 0.5897145324051589,
      "grad_norm": 0.9159910174500486,
      "learning_rate": 4.248415987332668e-06,
      "loss": 0.4257,
      "step": 10928
    },
    {
      "epoch": 0.5897684960336733,
      "grad_norm": 1.0177357543080732,
      "learning_rate": 4.247683213573892e-06,
      "loss": 0.3942,
      "step": 10929
    },
    {
      "epoch": 0.5898224596621877,
      "grad_norm": 0.941862968318894,
      "learning_rate": 4.246950475807966e-06,
      "loss": 0.4498,
      "step": 10930
    },
    {
      "epoch": 0.5898764232907021,
      "grad_norm": 0.858240013492606,
      "learning_rate": 4.2462177740559495e-06,
      "loss": 0.4284,
      "step": 10931
    },
    {
      "epoch": 0.5899303869192164,
      "grad_norm": 0.8802814664276891,
      "learning_rate": 4.245485108338904e-06,
      "loss": 0.4374,
      "step": 10932
    },
    {
      "epoch": 0.5899843505477308,
      "grad_norm": 1.1833059533532375,
      "learning_rate": 4.244752478677881e-06,
      "loss": 0.4409,
      "step": 10933
    },
    {
      "epoch": 0.5900383141762452,
      "grad_norm": 0.944253774977486,
      "learning_rate": 4.24401988509394e-06,
      "loss": 0.3817,
      "step": 10934
    },
    {
      "epoch": 0.5900922778047596,
      "grad_norm": 0.9804645323411332,
      "learning_rate": 4.243287327608137e-06,
      "loss": 0.4321,
      "step": 10935
    },
    {
      "epoch": 0.590146241433274,
      "grad_norm": 1.125208681986375,
      "learning_rate": 4.242554806241526e-06,
      "loss": 0.5304,
      "step": 10936
    },
    {
      "epoch": 0.5902002050617884,
      "grad_norm": 0.9128708235564581,
      "learning_rate": 4.24182232101516e-06,
      "loss": 0.3897,
      "step": 10937
    },
    {
      "epoch": 0.5902541686903028,
      "grad_norm": 0.9230092535442815,
      "learning_rate": 4.241089871950092e-06,
      "loss": 0.4343,
      "step": 10938
    },
    {
      "epoch": 0.5903081323188171,
      "grad_norm": 0.9902241926987247,
      "learning_rate": 4.240357459067374e-06,
      "loss": 0.3782,
      "step": 10939
    },
    {
      "epoch": 0.5903620959473315,
      "grad_norm": 1.1040283943714986,
      "learning_rate": 4.239625082388055e-06,
      "loss": 0.5776,
      "step": 10940
    },
    {
      "epoch": 0.5904160595758459,
      "grad_norm": 1.075195945291856,
      "learning_rate": 4.238892741933184e-06,
      "loss": 0.5503,
      "step": 10941
    },
    {
      "epoch": 0.5904700232043603,
      "grad_norm": 1.0895552890798734,
      "learning_rate": 4.23816043772381e-06,
      "loss": 0.5449,
      "step": 10942
    },
    {
      "epoch": 0.5905239868328747,
      "grad_norm": 1.2302439170135944,
      "learning_rate": 4.237428169780981e-06,
      "loss": 0.5053,
      "step": 10943
    },
    {
      "epoch": 0.5905779504613891,
      "grad_norm": 1.1474995459810682,
      "learning_rate": 4.23669593812574e-06,
      "loss": 0.5107,
      "step": 10944
    },
    {
      "epoch": 0.5906319140899035,
      "grad_norm": 1.0008069167958131,
      "learning_rate": 4.235963742779135e-06,
      "loss": 0.4149,
      "step": 10945
    },
    {
      "epoch": 0.5906858777184177,
      "grad_norm": 1.2629172697375057,
      "learning_rate": 4.235231583762212e-06,
      "loss": 0.6186,
      "step": 10946
    },
    {
      "epoch": 0.5907398413469321,
      "grad_norm": 1.0195046247805273,
      "learning_rate": 4.234499461096009e-06,
      "loss": 0.4649,
      "step": 10947
    },
    {
      "epoch": 0.5907938049754465,
      "grad_norm": 0.9799907110608739,
      "learning_rate": 4.23376737480157e-06,
      "loss": 0.4574,
      "step": 10948
    },
    {
      "epoch": 0.5908477686039609,
      "grad_norm": 0.9912336471024078,
      "learning_rate": 4.233035324899937e-06,
      "loss": 0.4562,
      "step": 10949
    },
    {
      "epoch": 0.5909017322324753,
      "grad_norm": 0.9893554526041589,
      "learning_rate": 4.232303311412147e-06,
      "loss": 0.45,
      "step": 10950
    },
    {
      "epoch": 0.5909556958609897,
      "grad_norm": 1.0729675223156676,
      "learning_rate": 4.231571334359243e-06,
      "loss": 0.4365,
      "step": 10951
    },
    {
      "epoch": 0.5910096594895041,
      "grad_norm": 1.1519138825643778,
      "learning_rate": 4.230839393762258e-06,
      "loss": 0.5148,
      "step": 10952
    },
    {
      "epoch": 0.5910636231180184,
      "grad_norm": 1.0198104645819699,
      "learning_rate": 4.2301074896422355e-06,
      "loss": 0.3782,
      "step": 10953
    },
    {
      "epoch": 0.5911175867465328,
      "grad_norm": 0.9859417453629512,
      "learning_rate": 4.229375622020204e-06,
      "loss": 0.4581,
      "step": 10954
    },
    {
      "epoch": 0.5911715503750472,
      "grad_norm": 0.9317699594587162,
      "learning_rate": 4.228643790917201e-06,
      "loss": 0.396,
      "step": 10955
    },
    {
      "epoch": 0.5912255140035616,
      "grad_norm": 0.9036768758710577,
      "learning_rate": 4.22791199635426e-06,
      "loss": 0.3495,
      "step": 10956
    },
    {
      "epoch": 0.591279477632076,
      "grad_norm": 1.249592755366539,
      "learning_rate": 4.227180238352413e-06,
      "loss": 0.5615,
      "step": 10957
    },
    {
      "epoch": 0.5913334412605904,
      "grad_norm": 0.9045935649528373,
      "learning_rate": 4.226448516932693e-06,
      "loss": 0.3599,
      "step": 10958
    },
    {
      "epoch": 0.5913874048891048,
      "grad_norm": 1.0425835038176108,
      "learning_rate": 4.225716832116129e-06,
      "loss": 0.4709,
      "step": 10959
    },
    {
      "epoch": 0.5914413685176191,
      "grad_norm": 1.0455443291793665,
      "learning_rate": 4.2249851839237525e-06,
      "loss": 0.5269,
      "step": 10960
    },
    {
      "epoch": 0.5914953321461335,
      "grad_norm": 1.09307464010701,
      "learning_rate": 4.2242535723765875e-06,
      "loss": 0.4098,
      "step": 10961
    },
    {
      "epoch": 0.5915492957746479,
      "grad_norm": 1.2405492994648102,
      "learning_rate": 4.223521997495665e-06,
      "loss": 0.5689,
      "step": 10962
    },
    {
      "epoch": 0.5916032594031623,
      "grad_norm": 0.9646559549033716,
      "learning_rate": 4.222790459302012e-06,
      "loss": 0.5031,
      "step": 10963
    },
    {
      "epoch": 0.5916572230316767,
      "grad_norm": 1.1191743640952783,
      "learning_rate": 4.22205895781665e-06,
      "loss": 0.5382,
      "step": 10964
    },
    {
      "epoch": 0.5917111866601911,
      "grad_norm": 0.9495121834730811,
      "learning_rate": 4.221327493060604e-06,
      "loss": 0.3825,
      "step": 10965
    },
    {
      "epoch": 0.5917651502887055,
      "grad_norm": 1.0037463072654413,
      "learning_rate": 4.220596065054898e-06,
      "loss": 0.3966,
      "step": 10966
    },
    {
      "epoch": 0.5918191139172198,
      "grad_norm": 1.0281347819325084,
      "learning_rate": 4.219864673820555e-06,
      "loss": 0.4116,
      "step": 10967
    },
    {
      "epoch": 0.5918730775457341,
      "grad_norm": 0.8442836264011336,
      "learning_rate": 4.219133319378594e-06,
      "loss": 0.4382,
      "step": 10968
    },
    {
      "epoch": 0.5919270411742485,
      "grad_norm": 0.8964448032455127,
      "learning_rate": 4.218402001750036e-06,
      "loss": 0.3927,
      "step": 10969
    },
    {
      "epoch": 0.591981004802763,
      "grad_norm": 1.1047450201423767,
      "learning_rate": 4.217670720955899e-06,
      "loss": 0.579,
      "step": 10970
    },
    {
      "epoch": 0.5920349684312773,
      "grad_norm": 1.0464273411284535,
      "learning_rate": 4.216939477017201e-06,
      "loss": 0.4891,
      "step": 10971
    },
    {
      "epoch": 0.5920889320597917,
      "grad_norm": 0.924752206528834,
      "learning_rate": 4.216208269954959e-06,
      "loss": 0.506,
      "step": 10972
    },
    {
      "epoch": 0.5921428956883061,
      "grad_norm": 0.8799690263330859,
      "learning_rate": 4.215477099790188e-06,
      "loss": 0.4437,
      "step": 10973
    },
    {
      "epoch": 0.5921968593168204,
      "grad_norm": 1.2194662988368745,
      "learning_rate": 4.214745966543906e-06,
      "loss": 0.5223,
      "step": 10974
    },
    {
      "epoch": 0.5922508229453348,
      "grad_norm": 1.0456472439515208,
      "learning_rate": 4.214014870237122e-06,
      "loss": 0.4262,
      "step": 10975
    },
    {
      "epoch": 0.5923047865738492,
      "grad_norm": 0.9632941951523539,
      "learning_rate": 4.213283810890848e-06,
      "loss": 0.5406,
      "step": 10976
    },
    {
      "epoch": 0.5923587502023636,
      "grad_norm": 1.1258323437335487,
      "learning_rate": 4.212552788526099e-06,
      "loss": 0.4284,
      "step": 10977
    },
    {
      "epoch": 0.592412713830878,
      "grad_norm": 1.0600155517591565,
      "learning_rate": 4.2118218031638836e-06,
      "loss": 0.4368,
      "step": 10978
    },
    {
      "epoch": 0.5924666774593924,
      "grad_norm": 1.1047790665009887,
      "learning_rate": 4.211090854825212e-06,
      "loss": 0.5582,
      "step": 10979
    },
    {
      "epoch": 0.5925206410879067,
      "grad_norm": 1.1235383859113703,
      "learning_rate": 4.210359943531092e-06,
      "loss": 0.5523,
      "step": 10980
    },
    {
      "epoch": 0.5925746047164211,
      "grad_norm": 1.141720060811772,
      "learning_rate": 4.209629069302528e-06,
      "loss": 0.3788,
      "step": 10981
    },
    {
      "epoch": 0.5926285683449355,
      "grad_norm": 1.1159423796231283,
      "learning_rate": 4.20889823216053e-06,
      "loss": 0.5596,
      "step": 10982
    },
    {
      "epoch": 0.5926825319734499,
      "grad_norm": 1.1312680790317653,
      "learning_rate": 4.2081674321261e-06,
      "loss": 0.4226,
      "step": 10983
    },
    {
      "epoch": 0.5927364956019643,
      "grad_norm": 0.8156913292471246,
      "learning_rate": 4.207436669220244e-06,
      "loss": 0.3924,
      "step": 10984
    },
    {
      "epoch": 0.5927904592304787,
      "grad_norm": 1.070964873656353,
      "learning_rate": 4.206705943463965e-06,
      "loss": 0.4112,
      "step": 10985
    },
    {
      "epoch": 0.5928444228589931,
      "grad_norm": 1.0689752087315099,
      "learning_rate": 4.205975254878262e-06,
      "loss": 0.4293,
      "step": 10986
    },
    {
      "epoch": 0.5928983864875074,
      "grad_norm": 0.8555779194583981,
      "learning_rate": 4.20524460348414e-06,
      "loss": 0.3485,
      "step": 10987
    },
    {
      "epoch": 0.5929523501160218,
      "grad_norm": 1.020338374713288,
      "learning_rate": 4.204513989302596e-06,
      "loss": 0.3761,
      "step": 10988
    },
    {
      "epoch": 0.5930063137445362,
      "grad_norm": 0.9481835443397031,
      "learning_rate": 4.203783412354627e-06,
      "loss": 0.4614,
      "step": 10989
    },
    {
      "epoch": 0.5930602773730506,
      "grad_norm": 1.0045156591355506,
      "learning_rate": 4.203052872661233e-06,
      "loss": 0.5837,
      "step": 10990
    },
    {
      "epoch": 0.593114241001565,
      "grad_norm": 0.9565793372810458,
      "learning_rate": 4.202322370243411e-06,
      "loss": 0.5078,
      "step": 10991
    },
    {
      "epoch": 0.5931682046300794,
      "grad_norm": 0.8538731237522023,
      "learning_rate": 4.201591905122154e-06,
      "loss": 0.4059,
      "step": 10992
    },
    {
      "epoch": 0.5932221682585938,
      "grad_norm": 1.0011376409007693,
      "learning_rate": 4.200861477318459e-06,
      "loss": 0.4344,
      "step": 10993
    },
    {
      "epoch": 0.593276131887108,
      "grad_norm": 1.0020034555379485,
      "learning_rate": 4.200131086853319e-06,
      "loss": 0.4371,
      "step": 10994
    },
    {
      "epoch": 0.5933300955156224,
      "grad_norm": 0.8960779027843356,
      "learning_rate": 4.1994007337477226e-06,
      "loss": 0.4,
      "step": 10995
    },
    {
      "epoch": 0.5933840591441368,
      "grad_norm": 1.2059822367220299,
      "learning_rate": 4.198670418022663e-06,
      "loss": 0.5408,
      "step": 10996
    },
    {
      "epoch": 0.5934380227726512,
      "grad_norm": 0.9767187437282322,
      "learning_rate": 4.197940139699132e-06,
      "loss": 0.4182,
      "step": 10997
    },
    {
      "epoch": 0.5934919864011656,
      "grad_norm": 0.9917065346923863,
      "learning_rate": 4.197209898798117e-06,
      "loss": 0.4166,
      "step": 10998
    },
    {
      "epoch": 0.59354595002968,
      "grad_norm": 1.072529580279694,
      "learning_rate": 4.196479695340604e-06,
      "loss": 0.5371,
      "step": 10999
    },
    {
      "epoch": 0.5935999136581944,
      "grad_norm": 0.7097591923694451,
      "learning_rate": 4.195749529347583e-06,
      "loss": 0.2791,
      "step": 11000
    },
    {
      "epoch": 0.5935999136581944,
      "eval_loss": 0.5420788526535034,
      "eval_runtime": 164.0786,
      "eval_samples_per_second": 20.959,
      "eval_steps_per_second": 0.878,
      "step": 11000
    },
    {
      "epoch": 0.5936538772867087,
      "grad_norm": 1.028865403702055,
      "learning_rate": 4.195019400840041e-06,
      "loss": 0.5185,
      "step": 11001
    },
    {
      "epoch": 0.5937078409152231,
      "grad_norm": 1.1283669384031108,
      "learning_rate": 4.1942893098389555e-06,
      "loss": 0.6308,
      "step": 11002
    },
    {
      "epoch": 0.5937618045437375,
      "grad_norm": 0.9673086622878548,
      "learning_rate": 4.193559256365317e-06,
      "loss": 0.4362,
      "step": 11003
    },
    {
      "epoch": 0.5938157681722519,
      "grad_norm": 0.8802151664381347,
      "learning_rate": 4.192829240440105e-06,
      "loss": 0.3538,
      "step": 11004
    },
    {
      "epoch": 0.5938697318007663,
      "grad_norm": 0.9016033848380937,
      "learning_rate": 4.192099262084301e-06,
      "loss": 0.358,
      "step": 11005
    },
    {
      "epoch": 0.5939236954292807,
      "grad_norm": 0.7625127469033044,
      "learning_rate": 4.191369321318887e-06,
      "loss": 0.3233,
      "step": 11006
    },
    {
      "epoch": 0.5939776590577951,
      "grad_norm": 0.9507115592672235,
      "learning_rate": 4.190639418164839e-06,
      "loss": 0.4246,
      "step": 11007
    },
    {
      "epoch": 0.5940316226863094,
      "grad_norm": 0.8898432588010998,
      "learning_rate": 4.18990955264314e-06,
      "loss": 0.4136,
      "step": 11008
    },
    {
      "epoch": 0.5940855863148238,
      "grad_norm": 0.8777419717265519,
      "learning_rate": 4.189179724774763e-06,
      "loss": 0.4508,
      "step": 11009
    },
    {
      "epoch": 0.5941395499433382,
      "grad_norm": 0.8776384775585993,
      "learning_rate": 4.188449934580684e-06,
      "loss": 0.3434,
      "step": 11010
    },
    {
      "epoch": 0.5941935135718526,
      "grad_norm": 1.1437768352895774,
      "learning_rate": 4.187720182081879e-06,
      "loss": 0.525,
      "step": 11011
    },
    {
      "epoch": 0.594247477200367,
      "grad_norm": 1.0836989671138357,
      "learning_rate": 4.186990467299323e-06,
      "loss": 0.5341,
      "step": 11012
    },
    {
      "epoch": 0.5943014408288814,
      "grad_norm": 1.1362705109385673,
      "learning_rate": 4.186260790253987e-06,
      "loss": 0.4704,
      "step": 11013
    },
    {
      "epoch": 0.5943554044573958,
      "grad_norm": 0.8755400876777962,
      "learning_rate": 4.185531150966844e-06,
      "loss": 0.3317,
      "step": 11014
    },
    {
      "epoch": 0.59440936808591,
      "grad_norm": 0.8288124314186148,
      "learning_rate": 4.184801549458864e-06,
      "loss": 0.3753,
      "step": 11015
    },
    {
      "epoch": 0.5944633317144244,
      "grad_norm": 0.9970224306304769,
      "learning_rate": 4.184071985751016e-06,
      "loss": 0.3971,
      "step": 11016
    },
    {
      "epoch": 0.5945172953429388,
      "grad_norm": 1.0831541854501883,
      "learning_rate": 4.183342459864268e-06,
      "loss": 0.427,
      "step": 11017
    },
    {
      "epoch": 0.5945712589714532,
      "grad_norm": 1.037879113522104,
      "learning_rate": 4.182612971819588e-06,
      "loss": 0.396,
      "step": 11018
    },
    {
      "epoch": 0.5946252225999676,
      "grad_norm": 1.119223511566342,
      "learning_rate": 4.181883521637944e-06,
      "loss": 0.5169,
      "step": 11019
    },
    {
      "epoch": 0.594679186228482,
      "grad_norm": 0.92575984695025,
      "learning_rate": 4.181154109340298e-06,
      "loss": 0.4228,
      "step": 11020
    },
    {
      "epoch": 0.5947331498569964,
      "grad_norm": 1.0855645789952875,
      "learning_rate": 4.1804247349476155e-06,
      "loss": 0.4118,
      "step": 11021
    },
    {
      "epoch": 0.5947871134855107,
      "grad_norm": 0.9294617428348763,
      "learning_rate": 4.179695398480862e-06,
      "loss": 0.3601,
      "step": 11022
    },
    {
      "epoch": 0.5948410771140251,
      "grad_norm": 0.9255505656009985,
      "learning_rate": 4.1789660999609945e-06,
      "loss": 0.3787,
      "step": 11023
    },
    {
      "epoch": 0.5948950407425395,
      "grad_norm": 1.2488653349202081,
      "learning_rate": 4.178236839408977e-06,
      "loss": 0.5802,
      "step": 11024
    },
    {
      "epoch": 0.5949490043710539,
      "grad_norm": 1.1189188437555928,
      "learning_rate": 4.177507616845767e-06,
      "loss": 0.4753,
      "step": 11025
    },
    {
      "epoch": 0.5950029679995683,
      "grad_norm": 0.8385555694729772,
      "learning_rate": 4.1767784322923256e-06,
      "loss": 0.4826,
      "step": 11026
    },
    {
      "epoch": 0.5950569316280827,
      "grad_norm": 1.078048020443036,
      "learning_rate": 4.176049285769608e-06,
      "loss": 0.5416,
      "step": 11027
    },
    {
      "epoch": 0.5951108952565971,
      "grad_norm": 1.054186354216756,
      "learning_rate": 4.175320177298572e-06,
      "loss": 0.4772,
      "step": 11028
    },
    {
      "epoch": 0.5951648588851114,
      "grad_norm": 1.1510285179635926,
      "learning_rate": 4.174591106900175e-06,
      "loss": 0.4312,
      "step": 11029
    },
    {
      "epoch": 0.5952188225136258,
      "grad_norm": 1.0265001060574979,
      "learning_rate": 4.173862074595366e-06,
      "loss": 0.5201,
      "step": 11030
    },
    {
      "epoch": 0.5952727861421402,
      "grad_norm": 0.990120254343981,
      "learning_rate": 4.173133080405102e-06,
      "loss": 0.4517,
      "step": 11031
    },
    {
      "epoch": 0.5953267497706546,
      "grad_norm": 1.0270092135342384,
      "learning_rate": 4.172404124350333e-06,
      "loss": 0.4751,
      "step": 11032
    },
    {
      "epoch": 0.595380713399169,
      "grad_norm": 0.7784894600066139,
      "learning_rate": 4.17167520645201e-06,
      "loss": 0.2956,
      "step": 11033
    },
    {
      "epoch": 0.5954346770276834,
      "grad_norm": 0.8476745930597585,
      "learning_rate": 4.170946326731087e-06,
      "loss": 0.304,
      "step": 11034
    },
    {
      "epoch": 0.5954886406561978,
      "grad_norm": 0.8928641309592352,
      "learning_rate": 4.170217485208507e-06,
      "loss": 0.4078,
      "step": 11035
    },
    {
      "epoch": 0.5955426042847121,
      "grad_norm": 0.8916848477178771,
      "learning_rate": 4.169488681905222e-06,
      "loss": 0.368,
      "step": 11036
    },
    {
      "epoch": 0.5955965679132265,
      "grad_norm": 0.9546895677210628,
      "learning_rate": 4.168759916842174e-06,
      "loss": 0.4251,
      "step": 11037
    },
    {
      "epoch": 0.5956505315417409,
      "grad_norm": 0.9104164135397887,
      "learning_rate": 4.168031190040313e-06,
      "loss": 0.3288,
      "step": 11038
    },
    {
      "epoch": 0.5957044951702553,
      "grad_norm": 1.0103756370951251,
      "learning_rate": 4.167302501520582e-06,
      "loss": 0.4401,
      "step": 11039
    },
    {
      "epoch": 0.5957584587987697,
      "grad_norm": 1.0856257774737101,
      "learning_rate": 4.166573851303922e-06,
      "loss": 0.5565,
      "step": 11040
    },
    {
      "epoch": 0.595812422427284,
      "grad_norm": 1.145908636457165,
      "learning_rate": 4.165845239411278e-06,
      "loss": 0.4111,
      "step": 11041
    },
    {
      "epoch": 0.5958663860557984,
      "grad_norm": 1.083402425971385,
      "learning_rate": 4.165116665863589e-06,
      "loss": 0.4486,
      "step": 11042
    },
    {
      "epoch": 0.5959203496843127,
      "grad_norm": 0.8986021797672753,
      "learning_rate": 4.164388130681798e-06,
      "loss": 0.3536,
      "step": 11043
    },
    {
      "epoch": 0.5959743133128271,
      "grad_norm": 1.1192005033601518,
      "learning_rate": 4.163659633886839e-06,
      "loss": 0.425,
      "step": 11044
    },
    {
      "epoch": 0.5960282769413415,
      "grad_norm": 0.9323640530768588,
      "learning_rate": 4.1629311754996526e-06,
      "loss": 0.4896,
      "step": 11045
    },
    {
      "epoch": 0.5960822405698559,
      "grad_norm": 1.2529437056820838,
      "learning_rate": 4.162202755541177e-06,
      "loss": 0.424,
      "step": 11046
    },
    {
      "epoch": 0.5961362041983703,
      "grad_norm": 1.078663730478356,
      "learning_rate": 4.161474374032343e-06,
      "loss": 0.6525,
      "step": 11047
    },
    {
      "epoch": 0.5961901678268847,
      "grad_norm": 0.7096034667796915,
      "learning_rate": 4.160746030994091e-06,
      "loss": 0.2438,
      "step": 11048
    },
    {
      "epoch": 0.596244131455399,
      "grad_norm": 1.009539121957056,
      "learning_rate": 4.160017726447352e-06,
      "loss": 0.5328,
      "step": 11049
    },
    {
      "epoch": 0.5962980950839134,
      "grad_norm": 1.0927938745605676,
      "learning_rate": 4.159289460413055e-06,
      "loss": 0.4793,
      "step": 11050
    },
    {
      "epoch": 0.5963520587124278,
      "grad_norm": 0.944710588803073,
      "learning_rate": 4.158561232912134e-06,
      "loss": 0.5425,
      "step": 11051
    },
    {
      "epoch": 0.5964060223409422,
      "grad_norm": 1.0289983325038785,
      "learning_rate": 4.157833043965519e-06,
      "loss": 0.5512,
      "step": 11052
    },
    {
      "epoch": 0.5964599859694566,
      "grad_norm": 0.9774162841184549,
      "learning_rate": 4.157104893594138e-06,
      "loss": 0.5631,
      "step": 11053
    },
    {
      "epoch": 0.596513949597971,
      "grad_norm": 1.227700178299568,
      "learning_rate": 4.156376781818919e-06,
      "loss": 0.5685,
      "step": 11054
    },
    {
      "epoch": 0.5965679132264854,
      "grad_norm": 1.076849690762014,
      "learning_rate": 4.15564870866079e-06,
      "loss": 0.4993,
      "step": 11055
    },
    {
      "epoch": 0.5966218768549997,
      "grad_norm": 0.9867993404681888,
      "learning_rate": 4.154920674140677e-06,
      "loss": 0.369,
      "step": 11056
    },
    {
      "epoch": 0.5966758404835141,
      "grad_norm": 1.092557576359699,
      "learning_rate": 4.1541926782794996e-06,
      "loss": 0.4592,
      "step": 11057
    },
    {
      "epoch": 0.5967298041120285,
      "grad_norm": 0.7978266195198108,
      "learning_rate": 4.153464721098186e-06,
      "loss": 0.3689,
      "step": 11058
    },
    {
      "epoch": 0.5967837677405429,
      "grad_norm": 1.2542044294996602,
      "learning_rate": 4.152736802617658e-06,
      "loss": 0.6038,
      "step": 11059
    },
    {
      "epoch": 0.5968377313690573,
      "grad_norm": 0.896623635364973,
      "learning_rate": 4.152008922858833e-06,
      "loss": 0.3393,
      "step": 11060
    },
    {
      "epoch": 0.5968916949975717,
      "grad_norm": 0.9930942647356851,
      "learning_rate": 4.151281081842634e-06,
      "loss": 0.4753,
      "step": 11061
    },
    {
      "epoch": 0.5969456586260861,
      "grad_norm": 1.1102018972534509,
      "learning_rate": 4.150553279589981e-06,
      "loss": 0.502,
      "step": 11062
    },
    {
      "epoch": 0.5969996222546003,
      "grad_norm": 0.9762769152249332,
      "learning_rate": 4.14982551612179e-06,
      "loss": 0.4787,
      "step": 11063
    },
    {
      "epoch": 0.5970535858831147,
      "grad_norm": 1.0425334979906387,
      "learning_rate": 4.149097791458977e-06,
      "loss": 0.5247,
      "step": 11064
    },
    {
      "epoch": 0.5971075495116291,
      "grad_norm": 0.9631213831060839,
      "learning_rate": 4.148370105622459e-06,
      "loss": 0.4293,
      "step": 11065
    },
    {
      "epoch": 0.5971615131401435,
      "grad_norm": 1.0976698658961441,
      "learning_rate": 4.14764245863315e-06,
      "loss": 0.5572,
      "step": 11066
    },
    {
      "epoch": 0.5972154767686579,
      "grad_norm": 1.02968370095937,
      "learning_rate": 4.146914850511963e-06,
      "loss": 0.4131,
      "step": 11067
    },
    {
      "epoch": 0.5972694403971723,
      "grad_norm": 1.1296387326346575,
      "learning_rate": 4.14618728127981e-06,
      "loss": 0.6333,
      "step": 11068
    },
    {
      "epoch": 0.5973234040256867,
      "grad_norm": 1.1236492104062799,
      "learning_rate": 4.145459750957603e-06,
      "loss": 0.5888,
      "step": 11069
    },
    {
      "epoch": 0.597377367654201,
      "grad_norm": 0.964787254159277,
      "learning_rate": 4.144732259566253e-06,
      "loss": 0.5035,
      "step": 11070
    },
    {
      "epoch": 0.5974313312827154,
      "grad_norm": 1.3613712403618075,
      "learning_rate": 4.144004807126666e-06,
      "loss": 0.5237,
      "step": 11071
    },
    {
      "epoch": 0.5974852949112298,
      "grad_norm": 1.024765481324011,
      "learning_rate": 4.14327739365975e-06,
      "loss": 0.451,
      "step": 11072
    },
    {
      "epoch": 0.5975392585397442,
      "grad_norm": 1.0978986366597525,
      "learning_rate": 4.142550019186413e-06,
      "loss": 0.6871,
      "step": 11073
    },
    {
      "epoch": 0.5975932221682586,
      "grad_norm": 1.2786828735229758,
      "learning_rate": 4.14182268372756e-06,
      "loss": 0.5489,
      "step": 11074
    },
    {
      "epoch": 0.597647185796773,
      "grad_norm": 1.1012207074807878,
      "learning_rate": 4.141095387304096e-06,
      "loss": 0.4988,
      "step": 11075
    },
    {
      "epoch": 0.5977011494252874,
      "grad_norm": 0.8669211180704689,
      "learning_rate": 4.140368129936923e-06,
      "loss": 0.3829,
      "step": 11076
    },
    {
      "epoch": 0.5977551130538017,
      "grad_norm": 1.1906745704057233,
      "learning_rate": 4.139640911646945e-06,
      "loss": 0.6462,
      "step": 11077
    },
    {
      "epoch": 0.5978090766823161,
      "grad_norm": 1.0056326209611721,
      "learning_rate": 4.138913732455061e-06,
      "loss": 0.3973,
      "step": 11078
    },
    {
      "epoch": 0.5978630403108305,
      "grad_norm": 1.0589156739683308,
      "learning_rate": 4.138186592382172e-06,
      "loss": 0.4976,
      "step": 11079
    },
    {
      "epoch": 0.5979170039393449,
      "grad_norm": 1.1146954849022401,
      "learning_rate": 4.137459491449175e-06,
      "loss": 0.5627,
      "step": 11080
    },
    {
      "epoch": 0.5979709675678593,
      "grad_norm": 0.9292250575332416,
      "learning_rate": 4.136732429676969e-06,
      "loss": 0.4187,
      "step": 11081
    },
    {
      "epoch": 0.5980249311963737,
      "grad_norm": 0.8994965637779799,
      "learning_rate": 4.13600540708645e-06,
      "loss": 0.3583,
      "step": 11082
    },
    {
      "epoch": 0.5980788948248881,
      "grad_norm": 1.0064433197629927,
      "learning_rate": 4.135278423698517e-06,
      "loss": 0.4779,
      "step": 11083
    },
    {
      "epoch": 0.5981328584534024,
      "grad_norm": 0.7803995534844482,
      "learning_rate": 4.13455147953406e-06,
      "loss": 0.3904,
      "step": 11084
    },
    {
      "epoch": 0.5981868220819168,
      "grad_norm": 1.0762144639325626,
      "learning_rate": 4.1338245746139715e-06,
      "loss": 0.5632,
      "step": 11085
    },
    {
      "epoch": 0.5982407857104312,
      "grad_norm": 0.8302205804069329,
      "learning_rate": 4.133097708959146e-06,
      "loss": 0.3815,
      "step": 11086
    },
    {
      "epoch": 0.5982947493389456,
      "grad_norm": 1.1696139085767192,
      "learning_rate": 4.132370882590473e-06,
      "loss": 0.5876,
      "step": 11087
    },
    {
      "epoch": 0.59834871296746,
      "grad_norm": 0.8558123368203989,
      "learning_rate": 4.131644095528842e-06,
      "loss": 0.339,
      "step": 11088
    },
    {
      "epoch": 0.5984026765959743,
      "grad_norm": 1.0514783904370386,
      "learning_rate": 4.130917347795145e-06,
      "loss": 0.6101,
      "step": 11089
    },
    {
      "epoch": 0.5984566402244887,
      "grad_norm": 1.0520827567189084,
      "learning_rate": 4.130190639410264e-06,
      "loss": 0.4474,
      "step": 11090
    },
    {
      "epoch": 0.598510603853003,
      "grad_norm": 1.00140140364508,
      "learning_rate": 4.1294639703950904e-06,
      "loss": 0.4425,
      "step": 11091
    },
    {
      "epoch": 0.5985645674815174,
      "grad_norm": 1.4009344594298783,
      "learning_rate": 4.1287373407705055e-06,
      "loss": 0.5972,
      "step": 11092
    },
    {
      "epoch": 0.5986185311100318,
      "grad_norm": 0.8548737380540052,
      "learning_rate": 4.1280107505573955e-06,
      "loss": 0.3991,
      "step": 11093
    },
    {
      "epoch": 0.5986724947385462,
      "grad_norm": 1.0938090088163557,
      "learning_rate": 4.127284199776643e-06,
      "loss": 0.5764,
      "step": 11094
    },
    {
      "epoch": 0.5987264583670606,
      "grad_norm": 0.9655497813858812,
      "learning_rate": 4.1265576884491285e-06,
      "loss": 0.4183,
      "step": 11095
    },
    {
      "epoch": 0.598780421995575,
      "grad_norm": 1.1308355362410822,
      "learning_rate": 4.1258312165957345e-06,
      "loss": 0.5124,
      "step": 11096
    },
    {
      "epoch": 0.5988343856240894,
      "grad_norm": 1.037591215472827,
      "learning_rate": 4.1251047842373404e-06,
      "loss": 0.455,
      "step": 11097
    },
    {
      "epoch": 0.5988883492526037,
      "grad_norm": 1.0136892349564408,
      "learning_rate": 4.124378391394826e-06,
      "loss": 0.4771,
      "step": 11098
    },
    {
      "epoch": 0.5989423128811181,
      "grad_norm": 1.024007181548137,
      "learning_rate": 4.123652038089063e-06,
      "loss": 0.4521,
      "step": 11099
    },
    {
      "epoch": 0.5989962765096325,
      "grad_norm": 0.9965188050366893,
      "learning_rate": 4.122925724340933e-06,
      "loss": 0.4217,
      "step": 11100
    },
    {
      "epoch": 0.5990502401381469,
      "grad_norm": 1.1604751137164162,
      "learning_rate": 4.122199450171307e-06,
      "loss": 0.5727,
      "step": 11101
    },
    {
      "epoch": 0.5991042037666613,
      "grad_norm": 0.9532426435535281,
      "learning_rate": 4.121473215601063e-06,
      "loss": 0.5046,
      "step": 11102
    },
    {
      "epoch": 0.5991581673951757,
      "grad_norm": 1.0199831398009973,
      "learning_rate": 4.120747020651071e-06,
      "loss": 0.5105,
      "step": 11103
    },
    {
      "epoch": 0.5992121310236901,
      "grad_norm": 1.1549708449921712,
      "learning_rate": 4.120020865342204e-06,
      "loss": 0.5553,
      "step": 11104
    },
    {
      "epoch": 0.5992660946522044,
      "grad_norm": 1.0487922200195547,
      "learning_rate": 4.119294749695332e-06,
      "loss": 0.6123,
      "step": 11105
    },
    {
      "epoch": 0.5993200582807188,
      "grad_norm": 1.2835984534530167,
      "learning_rate": 4.118568673731323e-06,
      "loss": 0.5415,
      "step": 11106
    },
    {
      "epoch": 0.5993740219092332,
      "grad_norm": 1.038710219444791,
      "learning_rate": 4.117842637471045e-06,
      "loss": 0.516,
      "step": 11107
    },
    {
      "epoch": 0.5994279855377476,
      "grad_norm": 1.102123358937724,
      "learning_rate": 4.117116640935368e-06,
      "loss": 0.5308,
      "step": 11108
    },
    {
      "epoch": 0.599481949166262,
      "grad_norm": 1.1227984213177935,
      "learning_rate": 4.116390684145155e-06,
      "loss": 0.4512,
      "step": 11109
    },
    {
      "epoch": 0.5995359127947764,
      "grad_norm": 1.4802644942721959,
      "learning_rate": 4.115664767121272e-06,
      "loss": 0.559,
      "step": 11110
    },
    {
      "epoch": 0.5995898764232908,
      "grad_norm": 0.9165525815853295,
      "learning_rate": 4.114938889884583e-06,
      "loss": 0.4082,
      "step": 11111
    },
    {
      "epoch": 0.599643840051805,
      "grad_norm": 1.113086088166424,
      "learning_rate": 4.114213052455951e-06,
      "loss": 0.4559,
      "step": 11112
    },
    {
      "epoch": 0.5996978036803194,
      "grad_norm": 1.1129592296582869,
      "learning_rate": 4.1134872548562345e-06,
      "loss": 0.4443,
      "step": 11113
    },
    {
      "epoch": 0.5997517673088338,
      "grad_norm": 1.0356120912083353,
      "learning_rate": 4.112761497106294e-06,
      "loss": 0.4075,
      "step": 11114
    },
    {
      "epoch": 0.5998057309373482,
      "grad_norm": 1.1899599388708462,
      "learning_rate": 4.112035779226993e-06,
      "loss": 0.5519,
      "step": 11115
    },
    {
      "epoch": 0.5998596945658626,
      "grad_norm": 0.918012525697702,
      "learning_rate": 4.1113101012391835e-06,
      "loss": 0.4563,
      "step": 11116
    },
    {
      "epoch": 0.599913658194377,
      "grad_norm": 1.0632704671898991,
      "learning_rate": 4.110584463163725e-06,
      "loss": 0.5708,
      "step": 11117
    },
    {
      "epoch": 0.5999676218228913,
      "grad_norm": 1.2562895987150515,
      "learning_rate": 4.109858865021474e-06,
      "loss": 0.6185,
      "step": 11118
    },
    {
      "epoch": 0.6000215854514057,
      "grad_norm": 0.9052682160717231,
      "learning_rate": 4.1091333068332836e-06,
      "loss": 0.5206,
      "step": 11119
    },
    {
      "epoch": 0.6000755490799201,
      "grad_norm": 0.8562382794630141,
      "learning_rate": 4.108407788620007e-06,
      "loss": 0.363,
      "step": 11120
    },
    {
      "epoch": 0.6001295127084345,
      "grad_norm": 0.8699963359042175,
      "learning_rate": 4.1076823104024966e-06,
      "loss": 0.4149,
      "step": 11121
    },
    {
      "epoch": 0.6001834763369489,
      "grad_norm": 1.0942963162580295,
      "learning_rate": 4.106956872201603e-06,
      "loss": 0.5174,
      "step": 11122
    },
    {
      "epoch": 0.6002374399654633,
      "grad_norm": 0.9254979158562309,
      "learning_rate": 4.106231474038177e-06,
      "loss": 0.3445,
      "step": 11123
    },
    {
      "epoch": 0.6002914035939777,
      "grad_norm": 1.204049852626815,
      "learning_rate": 4.105506115933067e-06,
      "loss": 0.5187,
      "step": 11124
    },
    {
      "epoch": 0.600345367222492,
      "grad_norm": 1.0054002505406736,
      "learning_rate": 4.104780797907121e-06,
      "loss": 0.4222,
      "step": 11125
    },
    {
      "epoch": 0.6003993308510064,
      "grad_norm": 1.106748688314904,
      "learning_rate": 4.104055519981184e-06,
      "loss": 0.4265,
      "step": 11126
    },
    {
      "epoch": 0.6004532944795208,
      "grad_norm": 1.2019503518294674,
      "learning_rate": 4.103330282176101e-06,
      "loss": 0.6356,
      "step": 11127
    },
    {
      "epoch": 0.6005072581080352,
      "grad_norm": 0.8831026764362648,
      "learning_rate": 4.102605084512717e-06,
      "loss": 0.4038,
      "step": 11128
    },
    {
      "epoch": 0.6005612217365496,
      "grad_norm": 0.9192533864005417,
      "learning_rate": 4.101879927011875e-06,
      "loss": 0.4076,
      "step": 11129
    },
    {
      "epoch": 0.600615185365064,
      "grad_norm": 0.9696283404685454,
      "learning_rate": 4.101154809694417e-06,
      "loss": 0.4362,
      "step": 11130
    },
    {
      "epoch": 0.6006691489935784,
      "grad_norm": 0.9486057472646162,
      "learning_rate": 4.100429732581184e-06,
      "loss": 0.3748,
      "step": 11131
    },
    {
      "epoch": 0.6007231126220927,
      "grad_norm": 0.926499981709349,
      "learning_rate": 4.099704695693014e-06,
      "loss": 0.437,
      "step": 11132
    },
    {
      "epoch": 0.6007770762506071,
      "grad_norm": 1.0451168739242667,
      "learning_rate": 4.098979699050746e-06,
      "loss": 0.6189,
      "step": 11133
    },
    {
      "epoch": 0.6008310398791215,
      "grad_norm": 0.8783930523802455,
      "learning_rate": 4.098254742675216e-06,
      "loss": 0.4574,
      "step": 11134
    },
    {
      "epoch": 0.6008850035076359,
      "grad_norm": 0.9890280154005837,
      "learning_rate": 4.097529826587262e-06,
      "loss": 0.508,
      "step": 11135
    },
    {
      "epoch": 0.6009389671361502,
      "grad_norm": 1.2559583508585266,
      "learning_rate": 4.096804950807717e-06,
      "loss": 0.5074,
      "step": 11136
    },
    {
      "epoch": 0.6009929307646646,
      "grad_norm": 0.7828456042890977,
      "learning_rate": 4.096080115357416e-06,
      "loss": 0.2782,
      "step": 11137
    },
    {
      "epoch": 0.601046894393179,
      "grad_norm": 1.0178313358440585,
      "learning_rate": 4.095355320257192e-06,
      "loss": 0.482,
      "step": 11138
    },
    {
      "epoch": 0.6011008580216933,
      "grad_norm": 0.7592620423274989,
      "learning_rate": 4.094630565527875e-06,
      "loss": 0.295,
      "step": 11139
    },
    {
      "epoch": 0.6011548216502077,
      "grad_norm": 0.9720104620195043,
      "learning_rate": 4.093905851190295e-06,
      "loss": 0.5308,
      "step": 11140
    },
    {
      "epoch": 0.6012087852787221,
      "grad_norm": 0.8580692696298982,
      "learning_rate": 4.093181177265282e-06,
      "loss": 0.4205,
      "step": 11141
    },
    {
      "epoch": 0.6012627489072365,
      "grad_norm": 1.168052625981166,
      "learning_rate": 4.092456543773662e-06,
      "loss": 0.4426,
      "step": 11142
    },
    {
      "epoch": 0.6013167125357509,
      "grad_norm": 0.9856739201489465,
      "learning_rate": 4.091731950736267e-06,
      "loss": 0.5008,
      "step": 11143
    },
    {
      "epoch": 0.6013706761642653,
      "grad_norm": 0.8610228738080602,
      "learning_rate": 4.091007398173914e-06,
      "loss": 0.3731,
      "step": 11144
    },
    {
      "epoch": 0.6014246397927797,
      "grad_norm": 1.0039160803527343,
      "learning_rate": 4.090282886107435e-06,
      "loss": 0.447,
      "step": 11145
    },
    {
      "epoch": 0.601478603421294,
      "grad_norm": 1.0989502001758973,
      "learning_rate": 4.089558414557649e-06,
      "loss": 0.5375,
      "step": 11146
    },
    {
      "epoch": 0.6015325670498084,
      "grad_norm": 1.023059728165426,
      "learning_rate": 4.088833983545381e-06,
      "loss": 0.4267,
      "step": 11147
    },
    {
      "epoch": 0.6015865306783228,
      "grad_norm": 0.902676412882943,
      "learning_rate": 4.088109593091447e-06,
      "loss": 0.371,
      "step": 11148
    },
    {
      "epoch": 0.6016404943068372,
      "grad_norm": 1.3501117122312973,
      "learning_rate": 4.087385243216673e-06,
      "loss": 0.5264,
      "step": 11149
    },
    {
      "epoch": 0.6016944579353516,
      "grad_norm": 0.9844987660451298,
      "learning_rate": 4.0866609339418725e-06,
      "loss": 0.4929,
      "step": 11150
    },
    {
      "epoch": 0.601748421563866,
      "grad_norm": 0.867126965495711,
      "learning_rate": 4.085936665287866e-06,
      "loss": 0.2327,
      "step": 11151
    },
    {
      "epoch": 0.6018023851923804,
      "grad_norm": 0.9341213020281629,
      "learning_rate": 4.085212437275467e-06,
      "loss": 0.4569,
      "step": 11152
    },
    {
      "epoch": 0.6018563488208947,
      "grad_norm": 0.9921423548551549,
      "learning_rate": 4.084488249925496e-06,
      "loss": 0.4025,
      "step": 11153
    },
    {
      "epoch": 0.6019103124494091,
      "grad_norm": 0.8241113122216805,
      "learning_rate": 4.08376410325876e-06,
      "loss": 0.3608,
      "step": 11154
    },
    {
      "epoch": 0.6019642760779235,
      "grad_norm": 1.0623265571884324,
      "learning_rate": 4.083039997296074e-06,
      "loss": 0.4376,
      "step": 11155
    },
    {
      "epoch": 0.6020182397064379,
      "grad_norm": 1.094521984738121,
      "learning_rate": 4.082315932058251e-06,
      "loss": 0.4961,
      "step": 11156
    },
    {
      "epoch": 0.6020722033349523,
      "grad_norm": 1.2371559182834297,
      "learning_rate": 4.0815919075661005e-06,
      "loss": 0.5837,
      "step": 11157
    },
    {
      "epoch": 0.6021261669634667,
      "grad_norm": 1.0657022523809114,
      "learning_rate": 4.080867923840431e-06,
      "loss": 0.5608,
      "step": 11158
    },
    {
      "epoch": 0.602180130591981,
      "grad_norm": 0.9645873010833275,
      "learning_rate": 4.080143980902054e-06,
      "loss": 0.4554,
      "step": 11159
    },
    {
      "epoch": 0.6022340942204953,
      "grad_norm": 0.978124875082468,
      "learning_rate": 4.079420078771773e-06,
      "loss": 0.4249,
      "step": 11160
    },
    {
      "epoch": 0.6022880578490097,
      "grad_norm": 0.9928962415851953,
      "learning_rate": 4.078696217470393e-06,
      "loss": 0.5787,
      "step": 11161
    },
    {
      "epoch": 0.6023420214775241,
      "grad_norm": 0.8782434874175966,
      "learning_rate": 4.077972397018719e-06,
      "loss": 0.3386,
      "step": 11162
    },
    {
      "epoch": 0.6023959851060385,
      "grad_norm": 1.1792525495011137,
      "learning_rate": 4.077248617437557e-06,
      "loss": 0.5919,
      "step": 11163
    },
    {
      "epoch": 0.6024499487345529,
      "grad_norm": 1.0055279733841165,
      "learning_rate": 4.076524878747706e-06,
      "loss": 0.5882,
      "step": 11164
    },
    {
      "epoch": 0.6025039123630673,
      "grad_norm": 1.310291546384891,
      "learning_rate": 4.075801180969968e-06,
      "loss": 0.5572,
      "step": 11165
    },
    {
      "epoch": 0.6025578759915817,
      "grad_norm": 1.1568031256848617,
      "learning_rate": 4.0750775241251436e-06,
      "loss": 0.4854,
      "step": 11166
    },
    {
      "epoch": 0.602611839620096,
      "grad_norm": 1.0775031331520695,
      "learning_rate": 4.074353908234032e-06,
      "loss": 0.4417,
      "step": 11167
    },
    {
      "epoch": 0.6026658032486104,
      "grad_norm": 1.1198670721037167,
      "learning_rate": 4.073630333317429e-06,
      "loss": 0.5437,
      "step": 11168
    },
    {
      "epoch": 0.6027197668771248,
      "grad_norm": 0.830277299231369,
      "learning_rate": 4.072906799396131e-06,
      "loss": 0.3909,
      "step": 11169
    },
    {
      "epoch": 0.6027737305056392,
      "grad_norm": 1.1974260041234048,
      "learning_rate": 4.072183306490934e-06,
      "loss": 0.4742,
      "step": 11170
    },
    {
      "epoch": 0.6028276941341536,
      "grad_norm": 1.041167079402187,
      "learning_rate": 4.071459854622631e-06,
      "loss": 0.558,
      "step": 11171
    },
    {
      "epoch": 0.602881657762668,
      "grad_norm": 0.9705731832509925,
      "learning_rate": 4.070736443812016e-06,
      "loss": 0.4601,
      "step": 11172
    },
    {
      "epoch": 0.6029356213911824,
      "grad_norm": 0.9558034544135781,
      "learning_rate": 4.070013074079879e-06,
      "loss": 0.5728,
      "step": 11173
    },
    {
      "epoch": 0.6029895850196967,
      "grad_norm": 1.2057312226077863,
      "learning_rate": 4.069289745447014e-06,
      "loss": 0.6178,
      "step": 11174
    },
    {
      "epoch": 0.6030435486482111,
      "grad_norm": 1.1842008051411952,
      "learning_rate": 4.068566457934203e-06,
      "loss": 0.4608,
      "step": 11175
    },
    {
      "epoch": 0.6030975122767255,
      "grad_norm": 1.0712938234486276,
      "learning_rate": 4.067843211562241e-06,
      "loss": 0.4298,
      "step": 11176
    },
    {
      "epoch": 0.6031514759052399,
      "grad_norm": 0.8413818085047241,
      "learning_rate": 4.067120006351911e-06,
      "loss": 0.3308,
      "step": 11177
    },
    {
      "epoch": 0.6032054395337543,
      "grad_norm": 0.9530148131435706,
      "learning_rate": 4.066396842323999e-06,
      "loss": 0.3881,
      "step": 11178
    },
    {
      "epoch": 0.6032594031622687,
      "grad_norm": 1.0553866418160338,
      "learning_rate": 4.065673719499292e-06,
      "loss": 0.4634,
      "step": 11179
    },
    {
      "epoch": 0.6033133667907831,
      "grad_norm": 1.0936798041969351,
      "learning_rate": 4.064950637898571e-06,
      "loss": 0.4742,
      "step": 11180
    },
    {
      "epoch": 0.6033673304192974,
      "grad_norm": 1.0645548660066442,
      "learning_rate": 4.06422759754262e-06,
      "loss": 0.4355,
      "step": 11181
    },
    {
      "epoch": 0.6034212940478118,
      "grad_norm": 1.11366095245495,
      "learning_rate": 4.063504598452218e-06,
      "loss": 0.5224,
      "step": 11182
    },
    {
      "epoch": 0.6034752576763261,
      "grad_norm": 0.979854962647158,
      "learning_rate": 4.062781640648144e-06,
      "loss": 0.4482,
      "step": 11183
    },
    {
      "epoch": 0.6035292213048405,
      "grad_norm": 0.8925371881412476,
      "learning_rate": 4.062058724151179e-06,
      "loss": 0.448,
      "step": 11184
    },
    {
      "epoch": 0.6035831849333549,
      "grad_norm": 0.9980494150895586,
      "learning_rate": 4.061335848982098e-06,
      "loss": 0.4539,
      "step": 11185
    },
    {
      "epoch": 0.6036371485618693,
      "grad_norm": 1.0742368810953316,
      "learning_rate": 4.06061301516168e-06,
      "loss": 0.5282,
      "step": 11186
    },
    {
      "epoch": 0.6036911121903836,
      "grad_norm": 1.0711779630734388,
      "learning_rate": 4.059890222710698e-06,
      "loss": 0.5313,
      "step": 11187
    },
    {
      "epoch": 0.603745075818898,
      "grad_norm": 1.0236163220939785,
      "learning_rate": 4.059167471649925e-06,
      "loss": 0.619,
      "step": 11188
    },
    {
      "epoch": 0.6037990394474124,
      "grad_norm": 1.0974989171775191,
      "learning_rate": 4.058444762000136e-06,
      "loss": 0.4331,
      "step": 11189
    },
    {
      "epoch": 0.6038530030759268,
      "grad_norm": 0.8455528999996135,
      "learning_rate": 4.057722093782101e-06,
      "loss": 0.3037,
      "step": 11190
    },
    {
      "epoch": 0.6039069667044412,
      "grad_norm": 1.1080437931127518,
      "learning_rate": 4.056999467016589e-06,
      "loss": 0.5232,
      "step": 11191
    },
    {
      "epoch": 0.6039609303329556,
      "grad_norm": 0.9852243743975097,
      "learning_rate": 4.056276881724371e-06,
      "loss": 0.565,
      "step": 11192
    },
    {
      "epoch": 0.60401489396147,
      "grad_norm": 0.8364547995679051,
      "learning_rate": 4.055554337926213e-06,
      "loss": 0.3522,
      "step": 11193
    },
    {
      "epoch": 0.6040688575899843,
      "grad_norm": 1.1448954804611817,
      "learning_rate": 4.054831835642885e-06,
      "loss": 0.7172,
      "step": 11194
    },
    {
      "epoch": 0.6041228212184987,
      "grad_norm": 0.8578572920096436,
      "learning_rate": 4.05410937489515e-06,
      "loss": 0.3283,
      "step": 11195
    },
    {
      "epoch": 0.6041767848470131,
      "grad_norm": 0.9489393845096502,
      "learning_rate": 4.053386955703772e-06,
      "loss": 0.416,
      "step": 11196
    },
    {
      "epoch": 0.6042307484755275,
      "grad_norm": 0.9614522477109886,
      "learning_rate": 4.052664578089514e-06,
      "loss": 0.7413,
      "step": 11197
    },
    {
      "epoch": 0.6042847121040419,
      "grad_norm": 0.9779521516273103,
      "learning_rate": 4.051942242073138e-06,
      "loss": 0.4593,
      "step": 11198
    },
    {
      "epoch": 0.6043386757325563,
      "grad_norm": 1.0346443853426976,
      "learning_rate": 4.051219947675407e-06,
      "loss": 0.3502,
      "step": 11199
    },
    {
      "epoch": 0.6043926393610707,
      "grad_norm": 0.9582755719812408,
      "learning_rate": 4.050497694917077e-06,
      "loss": 0.4681,
      "step": 11200
    },
    {
      "epoch": 0.604446602989585,
      "grad_norm": 0.9785315924871287,
      "learning_rate": 4.04977548381891e-06,
      "loss": 0.6152,
      "step": 11201
    },
    {
      "epoch": 0.6045005666180994,
      "grad_norm": 1.0953920484232824,
      "learning_rate": 4.049053314401659e-06,
      "loss": 0.4702,
      "step": 11202
    },
    {
      "epoch": 0.6045545302466138,
      "grad_norm": 0.836031336137012,
      "learning_rate": 4.048331186686083e-06,
      "loss": 0.4372,
      "step": 11203
    },
    {
      "epoch": 0.6046084938751282,
      "grad_norm": 0.9624096658059166,
      "learning_rate": 4.047609100692934e-06,
      "loss": 0.4611,
      "step": 11204
    },
    {
      "epoch": 0.6046624575036426,
      "grad_norm": 0.8876849527379131,
      "learning_rate": 4.046887056442968e-06,
      "loss": 0.3707,
      "step": 11205
    },
    {
      "epoch": 0.604716421132157,
      "grad_norm": 1.0003480444159765,
      "learning_rate": 4.046165053956936e-06,
      "loss": 0.5174,
      "step": 11206
    },
    {
      "epoch": 0.6047703847606714,
      "grad_norm": 1.178171876768977,
      "learning_rate": 4.04544309325559e-06,
      "loss": 0.5236,
      "step": 11207
    },
    {
      "epoch": 0.6048243483891856,
      "grad_norm": 0.9582173598008707,
      "learning_rate": 4.044721174359679e-06,
      "loss": 0.3862,
      "step": 11208
    },
    {
      "epoch": 0.6048783120177,
      "grad_norm": 0.7561630217454359,
      "learning_rate": 4.043999297289952e-06,
      "loss": 0.3351,
      "step": 11209
    },
    {
      "epoch": 0.6049322756462144,
      "grad_norm": 0.9727632021534015,
      "learning_rate": 4.043277462067156e-06,
      "loss": 0.4063,
      "step": 11210
    },
    {
      "epoch": 0.6049862392747288,
      "grad_norm": 1.1143997843851183,
      "learning_rate": 4.042555668712037e-06,
      "loss": 0.5563,
      "step": 11211
    },
    {
      "epoch": 0.6050402029032432,
      "grad_norm": 1.0074117849287367,
      "learning_rate": 4.0418339172453415e-06,
      "loss": 0.3397,
      "step": 11212
    },
    {
      "epoch": 0.6050941665317576,
      "grad_norm": 1.1368392369018505,
      "learning_rate": 4.041112207687812e-06,
      "loss": 0.5316,
      "step": 11213
    },
    {
      "epoch": 0.605148130160272,
      "grad_norm": 0.8113507982873562,
      "learning_rate": 4.040390540060191e-06,
      "loss": 0.3837,
      "step": 11214
    },
    {
      "epoch": 0.6052020937887863,
      "grad_norm": 0.9625014928434787,
      "learning_rate": 4.039668914383223e-06,
      "loss": 0.3434,
      "step": 11215
    },
    {
      "epoch": 0.6052560574173007,
      "grad_norm": 0.8931921044733949,
      "learning_rate": 4.038947330677645e-06,
      "loss": 0.379,
      "step": 11216
    },
    {
      "epoch": 0.6053100210458151,
      "grad_norm": 0.7900778083025959,
      "learning_rate": 4.038225788964196e-06,
      "loss": 0.2854,
      "step": 11217
    },
    {
      "epoch": 0.6053639846743295,
      "grad_norm": 1.138471180529493,
      "learning_rate": 4.0375042892636175e-06,
      "loss": 0.5592,
      "step": 11218
    },
    {
      "epoch": 0.6054179483028439,
      "grad_norm": 1.2745133822451102,
      "learning_rate": 4.036782831596641e-06,
      "loss": 0.5865,
      "step": 11219
    },
    {
      "epoch": 0.6054719119313583,
      "grad_norm": 0.8390068216580323,
      "learning_rate": 4.036061415984004e-06,
      "loss": 0.4344,
      "step": 11220
    },
    {
      "epoch": 0.6055258755598727,
      "grad_norm": 0.9161442200485773,
      "learning_rate": 4.035340042446443e-06,
      "loss": 0.4559,
      "step": 11221
    },
    {
      "epoch": 0.605579839188387,
      "grad_norm": 0.8694574772102286,
      "learning_rate": 4.034618711004688e-06,
      "loss": 0.4036,
      "step": 11222
    },
    {
      "epoch": 0.6056338028169014,
      "grad_norm": 1.05057122255945,
      "learning_rate": 4.033897421679472e-06,
      "loss": 0.3781,
      "step": 11223
    },
    {
      "epoch": 0.6056877664454158,
      "grad_norm": 1.0697882861140928,
      "learning_rate": 4.0331761744915255e-06,
      "loss": 0.4211,
      "step": 11224
    },
    {
      "epoch": 0.6057417300739302,
      "grad_norm": 1.031551210494694,
      "learning_rate": 4.0324549694615774e-06,
      "loss": 0.4547,
      "step": 11225
    },
    {
      "epoch": 0.6057956937024446,
      "grad_norm": 1.0243567610927513,
      "learning_rate": 4.0317338066103574e-06,
      "loss": 0.5351,
      "step": 11226
    },
    {
      "epoch": 0.605849657330959,
      "grad_norm": 0.9196717714318754,
      "learning_rate": 4.031012685958591e-06,
      "loss": 0.472,
      "step": 11227
    },
    {
      "epoch": 0.6059036209594734,
      "grad_norm": 0.9704136689534313,
      "learning_rate": 4.0302916075270035e-06,
      "loss": 0.4621,
      "step": 11228
    },
    {
      "epoch": 0.6059575845879877,
      "grad_norm": 0.8834547003530765,
      "learning_rate": 4.029570571336322e-06,
      "loss": 0.5115,
      "step": 11229
    },
    {
      "epoch": 0.606011548216502,
      "grad_norm": 0.9287278887285458,
      "learning_rate": 4.028849577407267e-06,
      "loss": 0.3732,
      "step": 11230
    },
    {
      "epoch": 0.6060655118450164,
      "grad_norm": 0.8865994125435757,
      "learning_rate": 4.028128625760562e-06,
      "loss": 0.4696,
      "step": 11231
    },
    {
      "epoch": 0.6061194754735308,
      "grad_norm": 1.1979263696097264,
      "learning_rate": 4.027407716416927e-06,
      "loss": 0.6034,
      "step": 11232
    },
    {
      "epoch": 0.6061734391020452,
      "grad_norm": 1.1846333399617008,
      "learning_rate": 4.026686849397081e-06,
      "loss": 0.6374,
      "step": 11233
    },
    {
      "epoch": 0.6062274027305596,
      "grad_norm": 0.8927272867057675,
      "learning_rate": 4.025966024721745e-06,
      "loss": 0.4132,
      "step": 11234
    },
    {
      "epoch": 0.606281366359074,
      "grad_norm": 0.9280986887753325,
      "learning_rate": 4.025245242411635e-06,
      "loss": 0.4912,
      "step": 11235
    },
    {
      "epoch": 0.6063353299875883,
      "grad_norm": 1.0552384756896573,
      "learning_rate": 4.024524502487468e-06,
      "loss": 0.4228,
      "step": 11236
    },
    {
      "epoch": 0.6063892936161027,
      "grad_norm": 0.8725498666984352,
      "learning_rate": 4.023803804969955e-06,
      "loss": 0.4231,
      "step": 11237
    },
    {
      "epoch": 0.6064432572446171,
      "grad_norm": 0.9727194629391644,
      "learning_rate": 4.023083149879815e-06,
      "loss": 0.4395,
      "step": 11238
    },
    {
      "epoch": 0.6064972208731315,
      "grad_norm": 1.1081886015991267,
      "learning_rate": 4.022362537237755e-06,
      "loss": 0.4966,
      "step": 11239
    },
    {
      "epoch": 0.6065511845016459,
      "grad_norm": 1.138666482144874,
      "learning_rate": 4.021641967064491e-06,
      "loss": 0.464,
      "step": 11240
    },
    {
      "epoch": 0.6066051481301603,
      "grad_norm": 1.162865405313143,
      "learning_rate": 4.020921439380728e-06,
      "loss": 0.561,
      "step": 11241
    },
    {
      "epoch": 0.6066591117586747,
      "grad_norm": 0.8368755219871664,
      "learning_rate": 4.020200954207178e-06,
      "loss": 0.3409,
      "step": 11242
    },
    {
      "epoch": 0.606713075387189,
      "grad_norm": 1.3050285815486153,
      "learning_rate": 4.01948051156455e-06,
      "loss": 0.4797,
      "step": 11243
    },
    {
      "epoch": 0.6067670390157034,
      "grad_norm": 0.9251036156066423,
      "learning_rate": 4.018760111473547e-06,
      "loss": 0.3434,
      "step": 11244
    },
    {
      "epoch": 0.6068210026442178,
      "grad_norm": 1.1043664981818386,
      "learning_rate": 4.018039753954875e-06,
      "loss": 0.4941,
      "step": 11245
    },
    {
      "epoch": 0.6068749662727322,
      "grad_norm": 0.9070523821739639,
      "learning_rate": 4.0173194390292366e-06,
      "loss": 0.3461,
      "step": 11246
    },
    {
      "epoch": 0.6069289299012466,
      "grad_norm": 0.8274946098131355,
      "learning_rate": 4.0165991667173365e-06,
      "loss": 0.4236,
      "step": 11247
    },
    {
      "epoch": 0.606982893529761,
      "grad_norm": 1.0246469991913778,
      "learning_rate": 4.015878937039875e-06,
      "loss": 0.4298,
      "step": 11248
    },
    {
      "epoch": 0.6070368571582753,
      "grad_norm": 0.9601031049789809,
      "learning_rate": 4.0151587500175545e-06,
      "loss": 0.4337,
      "step": 11249
    },
    {
      "epoch": 0.6070908207867897,
      "grad_norm": 1.1381191133275408,
      "learning_rate": 4.0144386056710705e-06,
      "loss": 0.6221,
      "step": 11250
    },
    {
      "epoch": 0.6071447844153041,
      "grad_norm": 0.8835880995541003,
      "learning_rate": 4.013718504021121e-06,
      "loss": 0.3881,
      "step": 11251
    },
    {
      "epoch": 0.6071987480438185,
      "grad_norm": 0.945169545507159,
      "learning_rate": 4.012998445088404e-06,
      "loss": 0.3469,
      "step": 11252
    },
    {
      "epoch": 0.6072527116723329,
      "grad_norm": 1.0768625747641027,
      "learning_rate": 4.0122784288936145e-06,
      "loss": 0.4484,
      "step": 11253
    },
    {
      "epoch": 0.6073066753008473,
      "grad_norm": 0.8217469368707649,
      "learning_rate": 4.011558455457446e-06,
      "loss": 0.3629,
      "step": 11254
    },
    {
      "epoch": 0.6073606389293617,
      "grad_norm": 1.0332280458208443,
      "learning_rate": 4.010838524800591e-06,
      "loss": 0.4356,
      "step": 11255
    },
    {
      "epoch": 0.6074146025578759,
      "grad_norm": 0.950266112529411,
      "learning_rate": 4.010118636943743e-06,
      "loss": 0.4377,
      "step": 11256
    },
    {
      "epoch": 0.6074685661863903,
      "grad_norm": 0.990426265626684,
      "learning_rate": 4.00939879190759e-06,
      "loss": 0.3485,
      "step": 11257
    },
    {
      "epoch": 0.6075225298149047,
      "grad_norm": 0.878715608521457,
      "learning_rate": 4.008678989712822e-06,
      "loss": 0.4263,
      "step": 11258
    },
    {
      "epoch": 0.6075764934434191,
      "grad_norm": 1.0054317251161695,
      "learning_rate": 4.007959230380125e-06,
      "loss": 0.5952,
      "step": 11259
    },
    {
      "epoch": 0.6076304570719335,
      "grad_norm": 0.9309127020208549,
      "learning_rate": 4.007239513930189e-06,
      "loss": 0.5628,
      "step": 11260
    },
    {
      "epoch": 0.6076844207004479,
      "grad_norm": 1.0287018787504136,
      "learning_rate": 4.006519840383697e-06,
      "loss": 0.4848,
      "step": 11261
    },
    {
      "epoch": 0.6077383843289623,
      "grad_norm": 1.097060620967825,
      "learning_rate": 4.0058002097613344e-06,
      "loss": 0.4431,
      "step": 11262
    },
    {
      "epoch": 0.6077923479574766,
      "grad_norm": 0.9122367920296621,
      "learning_rate": 4.0050806220837845e-06,
      "loss": 0.4725,
      "step": 11263
    },
    {
      "epoch": 0.607846311585991,
      "grad_norm": 0.9532191921072232,
      "learning_rate": 4.004361077371727e-06,
      "loss": 0.4694,
      "step": 11264
    },
    {
      "epoch": 0.6079002752145054,
      "grad_norm": 1.2189006182858988,
      "learning_rate": 4.003641575645842e-06,
      "loss": 0.5467,
      "step": 11265
    },
    {
      "epoch": 0.6079542388430198,
      "grad_norm": 1.0129672702730645,
      "learning_rate": 4.002922116926812e-06,
      "loss": 0.5341,
      "step": 11266
    },
    {
      "epoch": 0.6080082024715342,
      "grad_norm": 0.9196009372406767,
      "learning_rate": 4.0022027012353125e-06,
      "loss": 0.4533,
      "step": 11267
    },
    {
      "epoch": 0.6080621661000486,
      "grad_norm": 0.9665268030127374,
      "learning_rate": 4.00148332859202e-06,
      "loss": 0.3507,
      "step": 11268
    },
    {
      "epoch": 0.608116129728563,
      "grad_norm": 1.2051561613292476,
      "learning_rate": 4.000763999017612e-06,
      "loss": 0.4799,
      "step": 11269
    },
    {
      "epoch": 0.6081700933570773,
      "grad_norm": 1.049088487036102,
      "learning_rate": 4.000044712532761e-06,
      "loss": 0.4496,
      "step": 11270
    },
    {
      "epoch": 0.6082240569855917,
      "grad_norm": 0.9317509571044644,
      "learning_rate": 3.999325469158141e-06,
      "loss": 0.4395,
      "step": 11271
    },
    {
      "epoch": 0.6082780206141061,
      "grad_norm": 1.128061141658857,
      "learning_rate": 3.9986062689144225e-06,
      "loss": 0.527,
      "step": 11272
    },
    {
      "epoch": 0.6083319842426205,
      "grad_norm": 1.0142018782937168,
      "learning_rate": 3.997887111822277e-06,
      "loss": 0.536,
      "step": 11273
    },
    {
      "epoch": 0.6083859478711349,
      "grad_norm": 1.108691741727035,
      "learning_rate": 3.997167997902373e-06,
      "loss": 0.5016,
      "step": 11274
    },
    {
      "epoch": 0.6084399114996493,
      "grad_norm": 0.9486114255410073,
      "learning_rate": 3.99644892717538e-06,
      "loss": 0.3954,
      "step": 11275
    },
    {
      "epoch": 0.6084938751281637,
      "grad_norm": 0.8068982327668329,
      "learning_rate": 3.9957298996619646e-06,
      "loss": 0.5372,
      "step": 11276
    },
    {
      "epoch": 0.608547838756678,
      "grad_norm": 0.8989840810385241,
      "learning_rate": 3.9950109153827934e-06,
      "loss": 0.4155,
      "step": 11277
    },
    {
      "epoch": 0.6086018023851923,
      "grad_norm": 0.9784852480309489,
      "learning_rate": 3.994291974358526e-06,
      "loss": 0.4356,
      "step": 11278
    },
    {
      "epoch": 0.6086557660137067,
      "grad_norm": 1.0910800867370731,
      "learning_rate": 3.99357307660983e-06,
      "loss": 0.5441,
      "step": 11279
    },
    {
      "epoch": 0.6087097296422211,
      "grad_norm": 1.1021726474859521,
      "learning_rate": 3.992854222157365e-06,
      "loss": 0.5106,
      "step": 11280
    },
    {
      "epoch": 0.6087636932707355,
      "grad_norm": 0.9179396649128452,
      "learning_rate": 3.992135411021793e-06,
      "loss": 0.3822,
      "step": 11281
    },
    {
      "epoch": 0.6088176568992499,
      "grad_norm": 1.0017417596112426,
      "learning_rate": 3.991416643223773e-06,
      "loss": 0.4303,
      "step": 11282
    },
    {
      "epoch": 0.6088716205277643,
      "grad_norm": 1.0908943941028675,
      "learning_rate": 3.990697918783964e-06,
      "loss": 0.3898,
      "step": 11283
    },
    {
      "epoch": 0.6089255841562786,
      "grad_norm": 0.9333723012773977,
      "learning_rate": 3.989979237723023e-06,
      "loss": 0.3628,
      "step": 11284
    },
    {
      "epoch": 0.608979547784793,
      "grad_norm": 1.1999071680927131,
      "learning_rate": 3.989260600061602e-06,
      "loss": 0.3874,
      "step": 11285
    },
    {
      "epoch": 0.6090335114133074,
      "grad_norm": 1.2242123147858932,
      "learning_rate": 3.988542005820358e-06,
      "loss": 0.5281,
      "step": 11286
    },
    {
      "epoch": 0.6090874750418218,
      "grad_norm": 0.9149691471030462,
      "learning_rate": 3.9878234550199445e-06,
      "loss": 0.3081,
      "step": 11287
    },
    {
      "epoch": 0.6091414386703362,
      "grad_norm": 1.0853354841819784,
      "learning_rate": 3.987104947681012e-06,
      "loss": 0.6947,
      "step": 11288
    },
    {
      "epoch": 0.6091954022988506,
      "grad_norm": 0.8720304927225979,
      "learning_rate": 3.986386483824212e-06,
      "loss": 0.3764,
      "step": 11289
    },
    {
      "epoch": 0.609249365927365,
      "grad_norm": 0.9241425905475186,
      "learning_rate": 3.985668063470194e-06,
      "loss": 0.4038,
      "step": 11290
    },
    {
      "epoch": 0.6093033295558793,
      "grad_norm": 0.7620496624255647,
      "learning_rate": 3.984949686639608e-06,
      "loss": 0.293,
      "step": 11291
    },
    {
      "epoch": 0.6093572931843937,
      "grad_norm": 0.9510751321047908,
      "learning_rate": 3.984231353353097e-06,
      "loss": 0.4689,
      "step": 11292
    },
    {
      "epoch": 0.6094112568129081,
      "grad_norm": 1.1450204389556082,
      "learning_rate": 3.9835130636313094e-06,
      "loss": 0.4885,
      "step": 11293
    },
    {
      "epoch": 0.6094652204414225,
      "grad_norm": 0.9832060345427881,
      "learning_rate": 3.982794817494889e-06,
      "loss": 0.4229,
      "step": 11294
    },
    {
      "epoch": 0.6095191840699369,
      "grad_norm": 0.8454917103125924,
      "learning_rate": 3.982076614964476e-06,
      "loss": 0.3147,
      "step": 11295
    },
    {
      "epoch": 0.6095731476984513,
      "grad_norm": 1.07712178224362,
      "learning_rate": 3.981358456060717e-06,
      "loss": 0.4777,
      "step": 11296
    },
    {
      "epoch": 0.6096271113269657,
      "grad_norm": 0.8990886360331498,
      "learning_rate": 3.980640340804249e-06,
      "loss": 0.3585,
      "step": 11297
    },
    {
      "epoch": 0.60968107495548,
      "grad_norm": 0.8947749828679066,
      "learning_rate": 3.979922269215715e-06,
      "loss": 0.4773,
      "step": 11298
    },
    {
      "epoch": 0.6097350385839944,
      "grad_norm": 1.022364391588578,
      "learning_rate": 3.97920424131575e-06,
      "loss": 0.3734,
      "step": 11299
    },
    {
      "epoch": 0.6097890022125088,
      "grad_norm": 1.0316290097436582,
      "learning_rate": 3.97848625712499e-06,
      "loss": 0.401,
      "step": 11300
    },
    {
      "epoch": 0.6098429658410232,
      "grad_norm": 0.9267509025853948,
      "learning_rate": 3.977768316664073e-06,
      "loss": 0.4116,
      "step": 11301
    },
    {
      "epoch": 0.6098969294695376,
      "grad_norm": 0.9505863823076416,
      "learning_rate": 3.977050419953634e-06,
      "loss": 0.3948,
      "step": 11302
    },
    {
      "epoch": 0.609950893098052,
      "grad_norm": 0.7493025113473181,
      "learning_rate": 3.9763325670143034e-06,
      "loss": 0.3892,
      "step": 11303
    },
    {
      "epoch": 0.6100048567265663,
      "grad_norm": 1.1288665355762224,
      "learning_rate": 3.975614757866716e-06,
      "loss": 0.4823,
      "step": 11304
    },
    {
      "epoch": 0.6100588203550806,
      "grad_norm": 1.1464619682545463,
      "learning_rate": 3.9748969925315e-06,
      "loss": 0.4859,
      "step": 11305
    },
    {
      "epoch": 0.610112783983595,
      "grad_norm": 0.7375602844699507,
      "learning_rate": 3.974179271029286e-06,
      "loss": 0.334,
      "step": 11306
    },
    {
      "epoch": 0.6101667476121094,
      "grad_norm": 1.3288730669615159,
      "learning_rate": 3.9734615933807e-06,
      "loss": 0.52,
      "step": 11307
    },
    {
      "epoch": 0.6102207112406238,
      "grad_norm": 0.9940409182134845,
      "learning_rate": 3.972743959606371e-06,
      "loss": 0.5769,
      "step": 11308
    },
    {
      "epoch": 0.6102746748691382,
      "grad_norm": 0.9266444884063763,
      "learning_rate": 3.972026369726924e-06,
      "loss": 0.45,
      "step": 11309
    },
    {
      "epoch": 0.6103286384976526,
      "grad_norm": 0.9482166234084654,
      "learning_rate": 3.971308823762983e-06,
      "loss": 0.5219,
      "step": 11310
    },
    {
      "epoch": 0.610382602126167,
      "grad_norm": 1.0241842221373116,
      "learning_rate": 3.97059132173517e-06,
      "loss": 0.5138,
      "step": 11311
    },
    {
      "epoch": 0.6104365657546813,
      "grad_norm": 0.9318765932640736,
      "learning_rate": 3.969873863664111e-06,
      "loss": 0.5197,
      "step": 11312
    },
    {
      "epoch": 0.6104905293831957,
      "grad_norm": 1.1533524464942064,
      "learning_rate": 3.969156449570421e-06,
      "loss": 0.5235,
      "step": 11313
    },
    {
      "epoch": 0.6105444930117101,
      "grad_norm": 1.104617232147211,
      "learning_rate": 3.9684390794747204e-06,
      "loss": 0.4893,
      "step": 11314
    },
    {
      "epoch": 0.6105984566402245,
      "grad_norm": 1.1844849853542643,
      "learning_rate": 3.96772175339763e-06,
      "loss": 0.4907,
      "step": 11315
    },
    {
      "epoch": 0.6106524202687389,
      "grad_norm": 1.2470168125160055,
      "learning_rate": 3.967004471359763e-06,
      "loss": 0.4528,
      "step": 11316
    },
    {
      "epoch": 0.6107063838972533,
      "grad_norm": 0.9134421342224568,
      "learning_rate": 3.966287233381736e-06,
      "loss": 0.5379,
      "step": 11317
    },
    {
      "epoch": 0.6107603475257676,
      "grad_norm": 1.175179464143768,
      "learning_rate": 3.965570039484166e-06,
      "loss": 0.5529,
      "step": 11318
    },
    {
      "epoch": 0.610814311154282,
      "grad_norm": 0.921827071728991,
      "learning_rate": 3.9648528896876625e-06,
      "loss": 0.4175,
      "step": 11319
    },
    {
      "epoch": 0.6108682747827964,
      "grad_norm": 1.0057821878975541,
      "learning_rate": 3.964135784012838e-06,
      "loss": 0.4846,
      "step": 11320
    },
    {
      "epoch": 0.6109222384113108,
      "grad_norm": 0.9838859749531169,
      "learning_rate": 3.963418722480302e-06,
      "loss": 0.473,
      "step": 11321
    },
    {
      "epoch": 0.6109762020398252,
      "grad_norm": 1.040033944748168,
      "learning_rate": 3.9627017051106665e-06,
      "loss": 0.5781,
      "step": 11322
    },
    {
      "epoch": 0.6110301656683396,
      "grad_norm": 0.9999801329215182,
      "learning_rate": 3.961984731924535e-06,
      "loss": 0.4678,
      "step": 11323
    },
    {
      "epoch": 0.611084129296854,
      "grad_norm": 0.8500727626235808,
      "learning_rate": 3.9612678029425185e-06,
      "loss": 0.3925,
      "step": 11324
    },
    {
      "epoch": 0.6111380929253682,
      "grad_norm": 1.1472490616729791,
      "learning_rate": 3.96055091818522e-06,
      "loss": 0.5328,
      "step": 11325
    },
    {
      "epoch": 0.6111920565538826,
      "grad_norm": 1.3022803009646637,
      "learning_rate": 3.9598340776732426e-06,
      "loss": 0.4849,
      "step": 11326
    },
    {
      "epoch": 0.611246020182397,
      "grad_norm": 0.7861098314050887,
      "learning_rate": 3.9591172814271895e-06,
      "loss": 0.2796,
      "step": 11327
    },
    {
      "epoch": 0.6112999838109114,
      "grad_norm": 1.0270197924999247,
      "learning_rate": 3.9584005294676645e-06,
      "loss": 0.5157,
      "step": 11328
    },
    {
      "epoch": 0.6113539474394258,
      "grad_norm": 1.0044299850429308,
      "learning_rate": 3.957683821815263e-06,
      "loss": 0.4847,
      "step": 11329
    },
    {
      "epoch": 0.6114079110679402,
      "grad_norm": 0.9869121887662532,
      "learning_rate": 3.95696715849059e-06,
      "loss": 0.3961,
      "step": 11330
    },
    {
      "epoch": 0.6114618746964546,
      "grad_norm": 0.9573290077054151,
      "learning_rate": 3.956250539514238e-06,
      "loss": 0.4011,
      "step": 11331
    },
    {
      "epoch": 0.6115158383249689,
      "grad_norm": 1.0723433862467202,
      "learning_rate": 3.955533964906808e-06,
      "loss": 0.4481,
      "step": 11332
    },
    {
      "epoch": 0.6115698019534833,
      "grad_norm": 0.9566390207463136,
      "learning_rate": 3.95481743468889e-06,
      "loss": 0.477,
      "step": 11333
    },
    {
      "epoch": 0.6116237655819977,
      "grad_norm": 0.9678578739383089,
      "learning_rate": 3.954100948881081e-06,
      "loss": 0.4287,
      "step": 11334
    },
    {
      "epoch": 0.6116777292105121,
      "grad_norm": 1.0341314444805665,
      "learning_rate": 3.953384507503973e-06,
      "loss": 0.3936,
      "step": 11335
    },
    {
      "epoch": 0.6117316928390265,
      "grad_norm": 1.049427669531221,
      "learning_rate": 3.952668110578155e-06,
      "loss": 0.6302,
      "step": 11336
    },
    {
      "epoch": 0.6117856564675409,
      "grad_norm": 1.102549125071285,
      "learning_rate": 3.951951758124221e-06,
      "loss": 0.5986,
      "step": 11337
    },
    {
      "epoch": 0.6118396200960553,
      "grad_norm": 0.9558056077294073,
      "learning_rate": 3.951235450162756e-06,
      "loss": 0.397,
      "step": 11338
    },
    {
      "epoch": 0.6118935837245696,
      "grad_norm": 1.1043143411560228,
      "learning_rate": 3.9505191867143525e-06,
      "loss": 0.496,
      "step": 11339
    },
    {
      "epoch": 0.611947547353084,
      "grad_norm": 0.9033914830563022,
      "learning_rate": 3.949802967799591e-06,
      "loss": 0.336,
      "step": 11340
    },
    {
      "epoch": 0.6120015109815984,
      "grad_norm": 1.1288104514135848,
      "learning_rate": 3.949086793439058e-06,
      "loss": 0.4377,
      "step": 11341
    },
    {
      "epoch": 0.6120554746101128,
      "grad_norm": 1.0305232594037712,
      "learning_rate": 3.948370663653338e-06,
      "loss": 0.4801,
      "step": 11342
    },
    {
      "epoch": 0.6121094382386272,
      "grad_norm": 1.1164455975398875,
      "learning_rate": 3.9476545784630125e-06,
      "loss": 0.5317,
      "step": 11343
    },
    {
      "epoch": 0.6121634018671416,
      "grad_norm": 1.0240971335785225,
      "learning_rate": 3.946938537888663e-06,
      "loss": 0.4814,
      "step": 11344
    },
    {
      "epoch": 0.612217365495656,
      "grad_norm": 1.1278119673777605,
      "learning_rate": 3.946222541950871e-06,
      "loss": 0.4445,
      "step": 11345
    },
    {
      "epoch": 0.6122713291241703,
      "grad_norm": 1.1860059635101987,
      "learning_rate": 3.945506590670211e-06,
      "loss": 0.4612,
      "step": 11346
    },
    {
      "epoch": 0.6123252927526847,
      "grad_norm": 0.9259185709228956,
      "learning_rate": 3.944790684067263e-06,
      "loss": 0.4434,
      "step": 11347
    },
    {
      "epoch": 0.612379256381199,
      "grad_norm": 1.2722907658905114,
      "learning_rate": 3.944074822162603e-06,
      "loss": 0.558,
      "step": 11348
    },
    {
      "epoch": 0.6124332200097135,
      "grad_norm": 1.0705340437569038,
      "learning_rate": 3.943359004976802e-06,
      "loss": 0.5351,
      "step": 11349
    },
    {
      "epoch": 0.6124871836382279,
      "grad_norm": 0.9711782636713336,
      "learning_rate": 3.942643232530439e-06,
      "loss": 0.5056,
      "step": 11350
    },
    {
      "epoch": 0.6125411472667422,
      "grad_norm": 0.94954848956124,
      "learning_rate": 3.941927504844082e-06,
      "loss": 0.4982,
      "step": 11351
    },
    {
      "epoch": 0.6125951108952566,
      "grad_norm": 0.7447161519533517,
      "learning_rate": 3.9412118219383025e-06,
      "loss": 0.2876,
      "step": 11352
    },
    {
      "epoch": 0.6126490745237709,
      "grad_norm": 0.7304063072846637,
      "learning_rate": 3.940496183833672e-06,
      "loss": 0.2848,
      "step": 11353
    },
    {
      "epoch": 0.6127030381522853,
      "grad_norm": 1.1220299937164737,
      "learning_rate": 3.939780590550756e-06,
      "loss": 0.4673,
      "step": 11354
    },
    {
      "epoch": 0.6127570017807997,
      "grad_norm": 1.2590760863479662,
      "learning_rate": 3.939065042110122e-06,
      "loss": 0.4112,
      "step": 11355
    },
    {
      "epoch": 0.6128109654093141,
      "grad_norm": 1.1391221164871697,
      "learning_rate": 3.938349538532336e-06,
      "loss": 0.7944,
      "step": 11356
    },
    {
      "epoch": 0.6128649290378285,
      "grad_norm": 0.9200274034782757,
      "learning_rate": 3.937634079837961e-06,
      "loss": 0.3994,
      "step": 11357
    },
    {
      "epoch": 0.6129188926663429,
      "grad_norm": 1.0678471602267254,
      "learning_rate": 3.936918666047562e-06,
      "loss": 0.5252,
      "step": 11358
    },
    {
      "epoch": 0.6129728562948573,
      "grad_norm": 0.7951082511264022,
      "learning_rate": 3.9362032971817e-06,
      "loss": 0.3044,
      "step": 11359
    },
    {
      "epoch": 0.6130268199233716,
      "grad_norm": 0.8574973323391991,
      "learning_rate": 3.935487973260937e-06,
      "loss": 0.348,
      "step": 11360
    },
    {
      "epoch": 0.613080783551886,
      "grad_norm": 0.8524268238533316,
      "learning_rate": 3.934772694305829e-06,
      "loss": 0.3456,
      "step": 11361
    },
    {
      "epoch": 0.6131347471804004,
      "grad_norm": 1.2041623870486018,
      "learning_rate": 3.934057460336935e-06,
      "loss": 0.5385,
      "step": 11362
    },
    {
      "epoch": 0.6131887108089148,
      "grad_norm": 1.1360069684638865,
      "learning_rate": 3.933342271374813e-06,
      "loss": 0.648,
      "step": 11363
    },
    {
      "epoch": 0.6132426744374292,
      "grad_norm": 1.1425654876444762,
      "learning_rate": 3.932627127440015e-06,
      "loss": 0.4447,
      "step": 11364
    },
    {
      "epoch": 0.6132966380659436,
      "grad_norm": 0.870169580811279,
      "learning_rate": 3.9319120285530985e-06,
      "loss": 0.3058,
      "step": 11365
    },
    {
      "epoch": 0.613350601694458,
      "grad_norm": 1.0835800513759921,
      "learning_rate": 3.931196974734615e-06,
      "loss": 0.622,
      "step": 11366
    },
    {
      "epoch": 0.6134045653229723,
      "grad_norm": 1.183276211146257,
      "learning_rate": 3.930481966005115e-06,
      "loss": 0.585,
      "step": 11367
    },
    {
      "epoch": 0.6134585289514867,
      "grad_norm": 1.1415271785443588,
      "learning_rate": 3.9297670023851505e-06,
      "loss": 0.413,
      "step": 11368
    },
    {
      "epoch": 0.6135124925800011,
      "grad_norm": 1.0587290888220424,
      "learning_rate": 3.929052083895268e-06,
      "loss": 0.4679,
      "step": 11369
    },
    {
      "epoch": 0.6135664562085155,
      "grad_norm": 1.0787767415357277,
      "learning_rate": 3.928337210556015e-06,
      "loss": 0.4086,
      "step": 11370
    },
    {
      "epoch": 0.6136204198370299,
      "grad_norm": 0.9412736754068449,
      "learning_rate": 3.92762238238794e-06,
      "loss": 0.3181,
      "step": 11371
    },
    {
      "epoch": 0.6136743834655443,
      "grad_norm": 0.9659676813436732,
      "learning_rate": 3.926907599411586e-06,
      "loss": 0.4165,
      "step": 11372
    },
    {
      "epoch": 0.6137283470940587,
      "grad_norm": 1.3078832925827035,
      "learning_rate": 3.926192861647496e-06,
      "loss": 0.6013,
      "step": 11373
    },
    {
      "epoch": 0.6137823107225729,
      "grad_norm": 0.8160400340412286,
      "learning_rate": 3.9254781691162145e-06,
      "loss": 0.3615,
      "step": 11374
    },
    {
      "epoch": 0.6138362743510873,
      "grad_norm": 0.9626993194917328,
      "learning_rate": 3.924763521838282e-06,
      "loss": 0.4029,
      "step": 11375
    },
    {
      "epoch": 0.6138902379796017,
      "grad_norm": 1.034955304457003,
      "learning_rate": 3.924048919834236e-06,
      "loss": 0.4818,
      "step": 11376
    },
    {
      "epoch": 0.6139442016081161,
      "grad_norm": 1.154633534537917,
      "learning_rate": 3.923334363124617e-06,
      "loss": 0.5516,
      "step": 11377
    },
    {
      "epoch": 0.6139981652366305,
      "grad_norm": 1.0324987596350268,
      "learning_rate": 3.922619851729962e-06,
      "loss": 0.4121,
      "step": 11378
    },
    {
      "epoch": 0.6140521288651449,
      "grad_norm": 0.8402165325146292,
      "learning_rate": 3.921905385670805e-06,
      "loss": 0.4406,
      "step": 11379
    },
    {
      "epoch": 0.6141060924936593,
      "grad_norm": 0.8553724450173478,
      "learning_rate": 3.921190964967682e-06,
      "loss": 0.2889,
      "step": 11380
    },
    {
      "epoch": 0.6141600561221736,
      "grad_norm": 0.9188794730076365,
      "learning_rate": 3.920476589641129e-06,
      "loss": 0.4676,
      "step": 11381
    },
    {
      "epoch": 0.614214019750688,
      "grad_norm": 1.2150269024533378,
      "learning_rate": 3.9197622597116726e-06,
      "loss": 0.6813,
      "step": 11382
    },
    {
      "epoch": 0.6142679833792024,
      "grad_norm": 0.9875817965702387,
      "learning_rate": 3.919047975199846e-06,
      "loss": 0.4119,
      "step": 11383
    },
    {
      "epoch": 0.6143219470077168,
      "grad_norm": 1.0636522390771102,
      "learning_rate": 3.918333736126176e-06,
      "loss": 0.5994,
      "step": 11384
    },
    {
      "epoch": 0.6143759106362312,
      "grad_norm": 0.9014549650377265,
      "learning_rate": 3.917619542511195e-06,
      "loss": 0.3782,
      "step": 11385
    },
    {
      "epoch": 0.6144298742647456,
      "grad_norm": 1.0987780576560622,
      "learning_rate": 3.916905394375426e-06,
      "loss": 0.5547,
      "step": 11386
    },
    {
      "epoch": 0.6144838378932599,
      "grad_norm": 0.9307305231399055,
      "learning_rate": 3.916191291739398e-06,
      "loss": 0.4329,
      "step": 11387
    },
    {
      "epoch": 0.6145378015217743,
      "grad_norm": 0.7540857507865159,
      "learning_rate": 3.91547723462363e-06,
      "loss": 0.2783,
      "step": 11388
    },
    {
      "epoch": 0.6145917651502887,
      "grad_norm": 1.0433548829536956,
      "learning_rate": 3.914763223048646e-06,
      "loss": 0.584,
      "step": 11389
    },
    {
      "epoch": 0.6146457287788031,
      "grad_norm": 0.8278777772606793,
      "learning_rate": 3.914049257034971e-06,
      "loss": 0.3441,
      "step": 11390
    },
    {
      "epoch": 0.6146996924073175,
      "grad_norm": 1.1716801881399788,
      "learning_rate": 3.913335336603123e-06,
      "loss": 0.4449,
      "step": 11391
    },
    {
      "epoch": 0.6147536560358319,
      "grad_norm": 1.0817451277724341,
      "learning_rate": 3.912621461773619e-06,
      "loss": 0.3719,
      "step": 11392
    },
    {
      "epoch": 0.6148076196643463,
      "grad_norm": 0.9996869358622376,
      "learning_rate": 3.911907632566979e-06,
      "loss": 0.3916,
      "step": 11393
    },
    {
      "epoch": 0.6148615832928606,
      "grad_norm": 1.2747924363523395,
      "learning_rate": 3.911193849003719e-06,
      "loss": 0.5107,
      "step": 11394
    },
    {
      "epoch": 0.614915546921375,
      "grad_norm": 1.1357981867112188,
      "learning_rate": 3.910480111104352e-06,
      "loss": 0.4713,
      "step": 11395
    },
    {
      "epoch": 0.6149695105498894,
      "grad_norm": 1.4094727035356098,
      "learning_rate": 3.909766418889393e-06,
      "loss": 0.5831,
      "step": 11396
    },
    {
      "epoch": 0.6150234741784038,
      "grad_norm": 1.1255776665657375,
      "learning_rate": 3.9090527723793545e-06,
      "loss": 0.6059,
      "step": 11397
    },
    {
      "epoch": 0.6150774378069181,
      "grad_norm": 1.353476561975284,
      "learning_rate": 3.908339171594746e-06,
      "loss": 0.5773,
      "step": 11398
    },
    {
      "epoch": 0.6151314014354325,
      "grad_norm": 0.8935846918933449,
      "learning_rate": 3.90762561655608e-06,
      "loss": 0.4178,
      "step": 11399
    },
    {
      "epoch": 0.6151853650639469,
      "grad_norm": 0.9385304526931998,
      "learning_rate": 3.906912107283861e-06,
      "loss": 0.4162,
      "step": 11400
    },
    {
      "epoch": 0.6152393286924612,
      "grad_norm": 1.0131944832670243,
      "learning_rate": 3.906198643798599e-06,
      "loss": 0.4113,
      "step": 11401
    },
    {
      "epoch": 0.6152932923209756,
      "grad_norm": 0.7997704549820639,
      "learning_rate": 3.905485226120797e-06,
      "loss": 0.3259,
      "step": 11402
    },
    {
      "epoch": 0.61534725594949,
      "grad_norm": 1.0826006397685939,
      "learning_rate": 3.904771854270961e-06,
      "loss": 0.4991,
      "step": 11403
    },
    {
      "epoch": 0.6154012195780044,
      "grad_norm": 1.2134416407939599,
      "learning_rate": 3.904058528269593e-06,
      "loss": 0.5557,
      "step": 11404
    },
    {
      "epoch": 0.6154551832065188,
      "grad_norm": 1.0225824935604995,
      "learning_rate": 3.903345248137197e-06,
      "loss": 0.4356,
      "step": 11405
    },
    {
      "epoch": 0.6155091468350332,
      "grad_norm": 1.023086088323714,
      "learning_rate": 3.902632013894271e-06,
      "loss": 0.4712,
      "step": 11406
    },
    {
      "epoch": 0.6155631104635476,
      "grad_norm": 1.3351196117211133,
      "learning_rate": 3.901918825561315e-06,
      "loss": 0.4948,
      "step": 11407
    },
    {
      "epoch": 0.6156170740920619,
      "grad_norm": 1.0496823885962128,
      "learning_rate": 3.901205683158828e-06,
      "loss": 0.5264,
      "step": 11408
    },
    {
      "epoch": 0.6156710377205763,
      "grad_norm": 0.9826439793469758,
      "learning_rate": 3.900492586707304e-06,
      "loss": 0.6002,
      "step": 11409
    },
    {
      "epoch": 0.6157250013490907,
      "grad_norm": 0.808257663058477,
      "learning_rate": 3.899779536227239e-06,
      "loss": 0.3979,
      "step": 11410
    },
    {
      "epoch": 0.6157789649776051,
      "grad_norm": 0.8216850049701667,
      "learning_rate": 3.899066531739127e-06,
      "loss": 0.4501,
      "step": 11411
    },
    {
      "epoch": 0.6158329286061195,
      "grad_norm": 0.7943792179157726,
      "learning_rate": 3.89835357326346e-06,
      "loss": 0.2852,
      "step": 11412
    },
    {
      "epoch": 0.6158868922346339,
      "grad_norm": 0.9521426765254029,
      "learning_rate": 3.89764066082073e-06,
      "loss": 0.4051,
      "step": 11413
    },
    {
      "epoch": 0.6159408558631483,
      "grad_norm": 1.0656886205552583,
      "learning_rate": 3.8969277944314275e-06,
      "loss": 0.5224,
      "step": 11414
    },
    {
      "epoch": 0.6159948194916626,
      "grad_norm": 0.9467145612543645,
      "learning_rate": 3.896214974116041e-06,
      "loss": 0.4484,
      "step": 11415
    },
    {
      "epoch": 0.616048783120177,
      "grad_norm": 1.0500070992365433,
      "learning_rate": 3.895502199895055e-06,
      "loss": 0.5246,
      "step": 11416
    },
    {
      "epoch": 0.6161027467486914,
      "grad_norm": 0.9435026002494481,
      "learning_rate": 3.894789471788957e-06,
      "loss": 0.3799,
      "step": 11417
    },
    {
      "epoch": 0.6161567103772058,
      "grad_norm": 1.0582754191481545,
      "learning_rate": 3.894076789818232e-06,
      "loss": 0.4585,
      "step": 11418
    },
    {
      "epoch": 0.6162106740057202,
      "grad_norm": 0.9713238340541142,
      "learning_rate": 3.893364154003362e-06,
      "loss": 0.5277,
      "step": 11419
    },
    {
      "epoch": 0.6162646376342346,
      "grad_norm": 0.9682246208696998,
      "learning_rate": 3.892651564364829e-06,
      "loss": 0.3068,
      "step": 11420
    },
    {
      "epoch": 0.616318601262749,
      "grad_norm": 0.9863308735562876,
      "learning_rate": 3.891939020923115e-06,
      "loss": 0.4454,
      "step": 11421
    },
    {
      "epoch": 0.6163725648912632,
      "grad_norm": 1.0710780247419451,
      "learning_rate": 3.891226523698699e-06,
      "loss": 0.5203,
      "step": 11422
    },
    {
      "epoch": 0.6164265285197776,
      "grad_norm": 0.9910546340473102,
      "learning_rate": 3.890514072712057e-06,
      "loss": 0.4533,
      "step": 11423
    },
    {
      "epoch": 0.616480492148292,
      "grad_norm": 0.9325890643103935,
      "learning_rate": 3.8898016679836685e-06,
      "loss": 0.3863,
      "step": 11424
    },
    {
      "epoch": 0.6165344557768064,
      "grad_norm": 1.1363982421534473,
      "learning_rate": 3.889089309534006e-06,
      "loss": 0.5264,
      "step": 11425
    },
    {
      "epoch": 0.6165884194053208,
      "grad_norm": 0.9989220282256581,
      "learning_rate": 3.8883769973835435e-06,
      "loss": 0.4758,
      "step": 11426
    },
    {
      "epoch": 0.6166423830338352,
      "grad_norm": 0.9133871461934632,
      "learning_rate": 3.887664731552755e-06,
      "loss": 0.4096,
      "step": 11427
    },
    {
      "epoch": 0.6166963466623496,
      "grad_norm": 1.1683678551750676,
      "learning_rate": 3.886952512062112e-06,
      "loss": 0.4642,
      "step": 11428
    },
    {
      "epoch": 0.6167503102908639,
      "grad_norm": 1.1604436041027186,
      "learning_rate": 3.886240338932086e-06,
      "loss": 0.568,
      "step": 11429
    },
    {
      "epoch": 0.6168042739193783,
      "grad_norm": 0.967325888881782,
      "learning_rate": 3.885528212183139e-06,
      "loss": 0.4379,
      "step": 11430
    },
    {
      "epoch": 0.6168582375478927,
      "grad_norm": 0.919805564126186,
      "learning_rate": 3.884816131835744e-06,
      "loss": 0.4255,
      "step": 11431
    },
    {
      "epoch": 0.6169122011764071,
      "grad_norm": 1.0602917015298179,
      "learning_rate": 3.884104097910366e-06,
      "loss": 0.4573,
      "step": 11432
    },
    {
      "epoch": 0.6169661648049215,
      "grad_norm": 0.923098332264558,
      "learning_rate": 3.883392110427468e-06,
      "loss": 0.5404,
      "step": 11433
    },
    {
      "epoch": 0.6170201284334359,
      "grad_norm": 1.0945801072774433,
      "learning_rate": 3.882680169407515e-06,
      "loss": 0.5095,
      "step": 11434
    },
    {
      "epoch": 0.6170740920619503,
      "grad_norm": 0.9266970306344515,
      "learning_rate": 3.881968274870968e-06,
      "loss": 0.4464,
      "step": 11435
    },
    {
      "epoch": 0.6171280556904646,
      "grad_norm": 1.1850975503395336,
      "learning_rate": 3.881256426838289e-06,
      "loss": 0.4836,
      "step": 11436
    },
    {
      "epoch": 0.617182019318979,
      "grad_norm": 0.968073522372537,
      "learning_rate": 3.880544625329934e-06,
      "loss": 0.3456,
      "step": 11437
    },
    {
      "epoch": 0.6172359829474934,
      "grad_norm": 1.1617521938932986,
      "learning_rate": 3.8798328703663635e-06,
      "loss": 0.5371,
      "step": 11438
    },
    {
      "epoch": 0.6172899465760078,
      "grad_norm": 0.9807574171001371,
      "learning_rate": 3.879121161968032e-06,
      "loss": 0.5165,
      "step": 11439
    },
    {
      "epoch": 0.6173439102045222,
      "grad_norm": 1.0466857776840641,
      "learning_rate": 3.878409500155398e-06,
      "loss": 0.6067,
      "step": 11440
    },
    {
      "epoch": 0.6173978738330366,
      "grad_norm": 0.9986824633950354,
      "learning_rate": 3.877697884948912e-06,
      "loss": 0.5418,
      "step": 11441
    },
    {
      "epoch": 0.617451837461551,
      "grad_norm": 1.1004020969794024,
      "learning_rate": 3.876986316369028e-06,
      "loss": 0.5621,
      "step": 11442
    },
    {
      "epoch": 0.6175058010900653,
      "grad_norm": 0.8397005873392287,
      "learning_rate": 3.8762747944362e-06,
      "loss": 0.3675,
      "step": 11443
    },
    {
      "epoch": 0.6175597647185797,
      "grad_norm": 0.9378993021068696,
      "learning_rate": 3.875563319170873e-06,
      "loss": 0.4543,
      "step": 11444
    },
    {
      "epoch": 0.617613728347094,
      "grad_norm": 1.0046204971321318,
      "learning_rate": 3.874851890593497e-06,
      "loss": 0.5213,
      "step": 11445
    },
    {
      "epoch": 0.6176676919756084,
      "grad_norm": 0.9706273419803746,
      "learning_rate": 3.87414050872452e-06,
      "loss": 0.4766,
      "step": 11446
    },
    {
      "epoch": 0.6177216556041228,
      "grad_norm": 0.9953685792457094,
      "learning_rate": 3.873429173584387e-06,
      "loss": 0.4447,
      "step": 11447
    },
    {
      "epoch": 0.6177756192326372,
      "grad_norm": 0.9604981397911373,
      "learning_rate": 3.872717885193544e-06,
      "loss": 0.5047,
      "step": 11448
    },
    {
      "epoch": 0.6178295828611516,
      "grad_norm": 0.8993117503768179,
      "learning_rate": 3.872006643572433e-06,
      "loss": 0.269,
      "step": 11449
    },
    {
      "epoch": 0.6178835464896659,
      "grad_norm": 1.1723964686671366,
      "learning_rate": 3.871295448741498e-06,
      "loss": 0.4847,
      "step": 11450
    },
    {
      "epoch": 0.6179375101181803,
      "grad_norm": 0.9408961096829096,
      "learning_rate": 3.870584300721175e-06,
      "loss": 0.5554,
      "step": 11451
    },
    {
      "epoch": 0.6179914737466947,
      "grad_norm": 0.8559072370455617,
      "learning_rate": 3.869873199531906e-06,
      "loss": 0.4573,
      "step": 11452
    },
    {
      "epoch": 0.6180454373752091,
      "grad_norm": 1.0202856679161696,
      "learning_rate": 3.8691621451941285e-06,
      "loss": 0.4923,
      "step": 11453
    },
    {
      "epoch": 0.6180994010037235,
      "grad_norm": 1.328097931154329,
      "learning_rate": 3.8684511377282785e-06,
      "loss": 0.6606,
      "step": 11454
    },
    {
      "epoch": 0.6181533646322379,
      "grad_norm": 1.0628884969936774,
      "learning_rate": 3.867740177154792e-06,
      "loss": 0.5592,
      "step": 11455
    },
    {
      "epoch": 0.6182073282607522,
      "grad_norm": 1.0387541442125758,
      "learning_rate": 3.867029263494104e-06,
      "loss": 0.4998,
      "step": 11456
    },
    {
      "epoch": 0.6182612918892666,
      "grad_norm": 0.9587580865184013,
      "learning_rate": 3.866318396766642e-06,
      "loss": 0.3968,
      "step": 11457
    },
    {
      "epoch": 0.618315255517781,
      "grad_norm": 0.7560806714294208,
      "learning_rate": 3.865607576992841e-06,
      "loss": 0.2952,
      "step": 11458
    },
    {
      "epoch": 0.6183692191462954,
      "grad_norm": 0.9997306079514394,
      "learning_rate": 3.864896804193129e-06,
      "loss": 0.3599,
      "step": 11459
    },
    {
      "epoch": 0.6184231827748098,
      "grad_norm": 1.1649511912681678,
      "learning_rate": 3.864186078387935e-06,
      "loss": 0.5497,
      "step": 11460
    },
    {
      "epoch": 0.6184771464033242,
      "grad_norm": 0.9898316178760267,
      "learning_rate": 3.863475399597687e-06,
      "loss": 0.4326,
      "step": 11461
    },
    {
      "epoch": 0.6185311100318386,
      "grad_norm": 0.906481412932685,
      "learning_rate": 3.862764767842808e-06,
      "loss": 0.3772,
      "step": 11462
    },
    {
      "epoch": 0.6185850736603529,
      "grad_norm": 1.0477973505032714,
      "learning_rate": 3.862054183143727e-06,
      "loss": 0.3018,
      "step": 11463
    },
    {
      "epoch": 0.6186390372888673,
      "grad_norm": 1.0614606528497068,
      "learning_rate": 3.861343645520862e-06,
      "loss": 0.5207,
      "step": 11464
    },
    {
      "epoch": 0.6186930009173817,
      "grad_norm": 0.9976053254100097,
      "learning_rate": 3.860633154994635e-06,
      "loss": 0.5785,
      "step": 11465
    },
    {
      "epoch": 0.6187469645458961,
      "grad_norm": 1.03776987200864,
      "learning_rate": 3.85992271158547e-06,
      "loss": 0.6092,
      "step": 11466
    },
    {
      "epoch": 0.6188009281744105,
      "grad_norm": 0.788305797193371,
      "learning_rate": 3.859212315313782e-06,
      "loss": 0.3818,
      "step": 11467
    },
    {
      "epoch": 0.6188548918029249,
      "grad_norm": 1.2241588443619074,
      "learning_rate": 3.858501966199989e-06,
      "loss": 0.5175,
      "step": 11468
    },
    {
      "epoch": 0.6189088554314393,
      "grad_norm": 1.1170619082439357,
      "learning_rate": 3.85779166426451e-06,
      "loss": 0.4356,
      "step": 11469
    },
    {
      "epoch": 0.6189628190599535,
      "grad_norm": 0.9396062236296504,
      "learning_rate": 3.857081409527758e-06,
      "loss": 0.4082,
      "step": 11470
    },
    {
      "epoch": 0.6190167826884679,
      "grad_norm": 0.9578321713087723,
      "learning_rate": 3.856371202010144e-06,
      "loss": 0.5265,
      "step": 11471
    },
    {
      "epoch": 0.6190707463169823,
      "grad_norm": 0.9918621728368521,
      "learning_rate": 3.855661041732084e-06,
      "loss": 0.4304,
      "step": 11472
    },
    {
      "epoch": 0.6191247099454967,
      "grad_norm": 1.0004506619323135,
      "learning_rate": 3.854950928713987e-06,
      "loss": 0.4864,
      "step": 11473
    },
    {
      "epoch": 0.6191786735740111,
      "grad_norm": 1.082242422461479,
      "learning_rate": 3.854240862976264e-06,
      "loss": 0.436,
      "step": 11474
    },
    {
      "epoch": 0.6192326372025255,
      "grad_norm": 1.0581453688847322,
      "learning_rate": 3.8535308445393195e-06,
      "loss": 0.4712,
      "step": 11475
    },
    {
      "epoch": 0.6192866008310399,
      "grad_norm": 0.8274408248305202,
      "learning_rate": 3.852820873423563e-06,
      "loss": 0.3285,
      "step": 11476
    },
    {
      "epoch": 0.6193405644595542,
      "grad_norm": 0.8522411039665704,
      "learning_rate": 3.852110949649399e-06,
      "loss": 0.4077,
      "step": 11477
    },
    {
      "epoch": 0.6193945280880686,
      "grad_norm": 1.064807303019347,
      "learning_rate": 3.851401073237231e-06,
      "loss": 0.5056,
      "step": 11478
    },
    {
      "epoch": 0.619448491716583,
      "grad_norm": 0.9304164998566877,
      "learning_rate": 3.8506912442074615e-06,
      "loss": 0.459,
      "step": 11479
    },
    {
      "epoch": 0.6195024553450974,
      "grad_norm": 1.0238521124548599,
      "learning_rate": 3.849981462580492e-06,
      "loss": 0.458,
      "step": 11480
    },
    {
      "epoch": 0.6195564189736118,
      "grad_norm": 0.8619783848628629,
      "learning_rate": 3.8492717283767235e-06,
      "loss": 0.3337,
      "step": 11481
    },
    {
      "epoch": 0.6196103826021262,
      "grad_norm": 0.9648695344089687,
      "learning_rate": 3.8485620416165525e-06,
      "loss": 0.4244,
      "step": 11482
    },
    {
      "epoch": 0.6196643462306406,
      "grad_norm": 1.0257681974629396,
      "learning_rate": 3.8478524023203775e-06,
      "loss": 0.5546,
      "step": 11483
    },
    {
      "epoch": 0.6197183098591549,
      "grad_norm": 0.951172681978254,
      "learning_rate": 3.847142810508596e-06,
      "loss": 0.4347,
      "step": 11484
    },
    {
      "epoch": 0.6197722734876693,
      "grad_norm": 0.9130287918399902,
      "learning_rate": 3.8464332662015975e-06,
      "loss": 0.4301,
      "step": 11485
    },
    {
      "epoch": 0.6198262371161837,
      "grad_norm": 1.1828126088566135,
      "learning_rate": 3.845723769419779e-06,
      "loss": 0.5594,
      "step": 11486
    },
    {
      "epoch": 0.6198802007446981,
      "grad_norm": 1.0826286496555992,
      "learning_rate": 3.84501432018353e-06,
      "loss": 0.5103,
      "step": 11487
    },
    {
      "epoch": 0.6199341643732125,
      "grad_norm": 1.0644337873115821,
      "learning_rate": 3.844304918513242e-06,
      "loss": 0.3847,
      "step": 11488
    },
    {
      "epoch": 0.6199881280017269,
      "grad_norm": 1.0283876173864241,
      "learning_rate": 3.843595564429303e-06,
      "loss": 0.3769,
      "step": 11489
    },
    {
      "epoch": 0.6200420916302413,
      "grad_norm": 1.0292078668534241,
      "learning_rate": 3.842886257952102e-06,
      "loss": 0.5957,
      "step": 11490
    },
    {
      "epoch": 0.6200960552587556,
      "grad_norm": 1.0296125768024416,
      "learning_rate": 3.8421769991020255e-06,
      "loss": 0.3724,
      "step": 11491
    },
    {
      "epoch": 0.62015001888727,
      "grad_norm": 1.0605237476882372,
      "learning_rate": 3.8414677878994555e-06,
      "loss": 0.3722,
      "step": 11492
    },
    {
      "epoch": 0.6202039825157843,
      "grad_norm": 1.3004397249037534,
      "learning_rate": 3.840758624364777e-06,
      "loss": 0.5276,
      "step": 11493
    },
    {
      "epoch": 0.6202579461442987,
      "grad_norm": 1.0172165784815832,
      "learning_rate": 3.840049508518373e-06,
      "loss": 0.3925,
      "step": 11494
    },
    {
      "epoch": 0.6203119097728131,
      "grad_norm": 1.1777293288787547,
      "learning_rate": 3.839340440380622e-06,
      "loss": 0.4593,
      "step": 11495
    },
    {
      "epoch": 0.6203658734013275,
      "grad_norm": 0.8538588004947577,
      "learning_rate": 3.838631419971906e-06,
      "loss": 0.3505,
      "step": 11496
    },
    {
      "epoch": 0.6204198370298419,
      "grad_norm": 1.2546813991769616,
      "learning_rate": 3.837922447312601e-06,
      "loss": 0.5106,
      "step": 11497
    },
    {
      "epoch": 0.6204738006583562,
      "grad_norm": 1.19454985496936,
      "learning_rate": 3.837213522423085e-06,
      "loss": 0.4484,
      "step": 11498
    },
    {
      "epoch": 0.6205277642868706,
      "grad_norm": 1.0578902677266486,
      "learning_rate": 3.836504645323732e-06,
      "loss": 0.4383,
      "step": 11499
    },
    {
      "epoch": 0.620581727915385,
      "grad_norm": 1.2225484671526736,
      "learning_rate": 3.835795816034917e-06,
      "loss": 0.5431,
      "step": 11500
    },
    {
      "epoch": 0.620581727915385,
      "eval_loss": 0.5394495725631714,
      "eval_runtime": 162.1491,
      "eval_samples_per_second": 21.209,
      "eval_steps_per_second": 0.888,
      "step": 11500
    },
    {
      "epoch": 0.6206356915438994,
      "grad_norm": 1.0928242331320581,
      "learning_rate": 3.835087034577011e-06,
      "loss": 0.6188,
      "step": 11501
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 1.0426165233477172,
      "learning_rate": 3.834378300970385e-06,
      "loss": 0.4243,
      "step": 11502
    },
    {
      "epoch": 0.6207436188009282,
      "grad_norm": 1.018727726745797,
      "learning_rate": 3.8336696152354114e-06,
      "loss": 0.4492,
      "step": 11503
    },
    {
      "epoch": 0.6207975824294426,
      "grad_norm": 1.062229775389288,
      "learning_rate": 3.8329609773924555e-06,
      "loss": 0.4622,
      "step": 11504
    },
    {
      "epoch": 0.6208515460579569,
      "grad_norm": 1.1672120435524331,
      "learning_rate": 3.832252387461888e-06,
      "loss": 0.6572,
      "step": 11505
    },
    {
      "epoch": 0.6209055096864713,
      "grad_norm": 0.9702766509155669,
      "learning_rate": 3.83154384546407e-06,
      "loss": 0.5688,
      "step": 11506
    },
    {
      "epoch": 0.6209594733149857,
      "grad_norm": 0.8592360899075672,
      "learning_rate": 3.8308353514193675e-06,
      "loss": 0.3563,
      "step": 11507
    },
    {
      "epoch": 0.6210134369435001,
      "grad_norm": 1.2011842094061358,
      "learning_rate": 3.830126905348144e-06,
      "loss": 0.5057,
      "step": 11508
    },
    {
      "epoch": 0.6210674005720145,
      "grad_norm": 1.098551600577112,
      "learning_rate": 3.8294185072707605e-06,
      "loss": 0.5195,
      "step": 11509
    },
    {
      "epoch": 0.6211213642005289,
      "grad_norm": 1.2878262394424043,
      "learning_rate": 3.828710157207577e-06,
      "loss": 0.4814,
      "step": 11510
    },
    {
      "epoch": 0.6211753278290433,
      "grad_norm": 0.8629109397765728,
      "learning_rate": 3.828001855178951e-06,
      "loss": 0.4156,
      "step": 11511
    },
    {
      "epoch": 0.6212292914575576,
      "grad_norm": 1.0924538212688848,
      "learning_rate": 3.827293601205245e-06,
      "loss": 0.4687,
      "step": 11512
    },
    {
      "epoch": 0.621283255086072,
      "grad_norm": 1.0299379912474689,
      "learning_rate": 3.826585395306808e-06,
      "loss": 0.4794,
      "step": 11513
    },
    {
      "epoch": 0.6213372187145864,
      "grad_norm": 0.9461871914481407,
      "learning_rate": 3.8258772375039986e-06,
      "loss": 0.5357,
      "step": 11514
    },
    {
      "epoch": 0.6213911823431008,
      "grad_norm": 1.0261442283498754,
      "learning_rate": 3.825169127817168e-06,
      "loss": 0.4804,
      "step": 11515
    },
    {
      "epoch": 0.6214451459716152,
      "grad_norm": 0.8504956219411951,
      "learning_rate": 3.824461066266669e-06,
      "loss": 0.4666,
      "step": 11516
    },
    {
      "epoch": 0.6214991096001296,
      "grad_norm": 0.8728832273600415,
      "learning_rate": 3.823753052872853e-06,
      "loss": 0.3662,
      "step": 11517
    },
    {
      "epoch": 0.621553073228644,
      "grad_norm": 1.0526580014321663,
      "learning_rate": 3.823045087656066e-06,
      "loss": 0.5461,
      "step": 11518
    },
    {
      "epoch": 0.6216070368571582,
      "grad_norm": 1.0568995136112054,
      "learning_rate": 3.822337170636662e-06,
      "loss": 0.4336,
      "step": 11519
    },
    {
      "epoch": 0.6216610004856726,
      "grad_norm": 1.042633568113139,
      "learning_rate": 3.821629301834979e-06,
      "loss": 0.4872,
      "step": 11520
    },
    {
      "epoch": 0.621714964114187,
      "grad_norm": 1.241596452194854,
      "learning_rate": 3.820921481271366e-06,
      "loss": 0.5869,
      "step": 11521
    },
    {
      "epoch": 0.6217689277427014,
      "grad_norm": 1.0091101443320372,
      "learning_rate": 3.820213708966167e-06,
      "loss": 0.4826,
      "step": 11522
    },
    {
      "epoch": 0.6218228913712158,
      "grad_norm": 1.1996616074585864,
      "learning_rate": 3.819505984939723e-06,
      "loss": 0.5847,
      "step": 11523
    },
    {
      "epoch": 0.6218768549997302,
      "grad_norm": 1.3203179278497292,
      "learning_rate": 3.818798309212375e-06,
      "loss": 0.5566,
      "step": 11524
    },
    {
      "epoch": 0.6219308186282445,
      "grad_norm": 1.0512340690060629,
      "learning_rate": 3.818090681804463e-06,
      "loss": 0.5127,
      "step": 11525
    },
    {
      "epoch": 0.6219847822567589,
      "grad_norm": 0.8009916852283765,
      "learning_rate": 3.817383102736324e-06,
      "loss": 0.3141,
      "step": 11526
    },
    {
      "epoch": 0.6220387458852733,
      "grad_norm": 1.1437517560317418,
      "learning_rate": 3.8166755720282944e-06,
      "loss": 0.4351,
      "step": 11527
    },
    {
      "epoch": 0.6220927095137877,
      "grad_norm": 1.317766066981831,
      "learning_rate": 3.815968089700709e-06,
      "loss": 0.5724,
      "step": 11528
    },
    {
      "epoch": 0.6221466731423021,
      "grad_norm": 0.9348700614452657,
      "learning_rate": 3.815260655773903e-06,
      "loss": 0.444,
      "step": 11529
    },
    {
      "epoch": 0.6222006367708165,
      "grad_norm": 0.9068526737380279,
      "learning_rate": 3.8145532702682087e-06,
      "loss": 0.4259,
      "step": 11530
    },
    {
      "epoch": 0.6222546003993309,
      "grad_norm": 1.1522108505644295,
      "learning_rate": 3.8138459332039557e-06,
      "loss": 0.5051,
      "step": 11531
    },
    {
      "epoch": 0.6223085640278452,
      "grad_norm": 0.9558540264445917,
      "learning_rate": 3.8131386446014764e-06,
      "loss": 0.5014,
      "step": 11532
    },
    {
      "epoch": 0.6223625276563596,
      "grad_norm": 1.1481603827623297,
      "learning_rate": 3.812431404481094e-06,
      "loss": 0.3994,
      "step": 11533
    },
    {
      "epoch": 0.622416491284874,
      "grad_norm": 0.7937764020181572,
      "learning_rate": 3.8117242128631393e-06,
      "loss": 0.3154,
      "step": 11534
    },
    {
      "epoch": 0.6224704549133884,
      "grad_norm": 1.0477472948637327,
      "learning_rate": 3.811017069767937e-06,
      "loss": 0.478,
      "step": 11535
    },
    {
      "epoch": 0.6225244185419028,
      "grad_norm": 1.3814974313038366,
      "learning_rate": 3.81030997521581e-06,
      "loss": 0.4965,
      "step": 11536
    },
    {
      "epoch": 0.6225783821704172,
      "grad_norm": 1.0559092380340673,
      "learning_rate": 3.809602929227081e-06,
      "loss": 0.4572,
      "step": 11537
    },
    {
      "epoch": 0.6226323457989316,
      "grad_norm": 0.9408874081125741,
      "learning_rate": 3.8088959318220732e-06,
      "loss": 0.3563,
      "step": 11538
    },
    {
      "epoch": 0.6226863094274458,
      "grad_norm": 0.9783586702548881,
      "learning_rate": 3.808188983021106e-06,
      "loss": 0.4092,
      "step": 11539
    },
    {
      "epoch": 0.6227402730559602,
      "grad_norm": 1.0623124699118882,
      "learning_rate": 3.8074820828444956e-06,
      "loss": 0.659,
      "step": 11540
    },
    {
      "epoch": 0.6227942366844746,
      "grad_norm": 0.8290981627078367,
      "learning_rate": 3.8067752313125604e-06,
      "loss": 0.4249,
      "step": 11541
    },
    {
      "epoch": 0.622848200312989,
      "grad_norm": 0.7806879860138479,
      "learning_rate": 3.8060684284456152e-06,
      "loss": 0.2828,
      "step": 11542
    },
    {
      "epoch": 0.6229021639415034,
      "grad_norm": 0.8421508172221233,
      "learning_rate": 3.8053616742639754e-06,
      "loss": 0.3875,
      "step": 11543
    },
    {
      "epoch": 0.6229561275700178,
      "grad_norm": 1.1226874465740895,
      "learning_rate": 3.804654968787953e-06,
      "loss": 0.6562,
      "step": 11544
    },
    {
      "epoch": 0.6230100911985322,
      "grad_norm": 0.980367538310028,
      "learning_rate": 3.8039483120378594e-06,
      "loss": 0.4623,
      "step": 11545
    },
    {
      "epoch": 0.6230640548270465,
      "grad_norm": 1.1271966600707952,
      "learning_rate": 3.8032417040340075e-06,
      "loss": 0.5555,
      "step": 11546
    },
    {
      "epoch": 0.6231180184555609,
      "grad_norm": 1.1825399020404714,
      "learning_rate": 3.8025351447967006e-06,
      "loss": 0.6318,
      "step": 11547
    },
    {
      "epoch": 0.6231719820840753,
      "grad_norm": 0.9766405443489511,
      "learning_rate": 3.801828634346252e-06,
      "loss": 0.4287,
      "step": 11548
    },
    {
      "epoch": 0.6232259457125897,
      "grad_norm": 1.1523084417664822,
      "learning_rate": 3.8011221727029623e-06,
      "loss": 0.5187,
      "step": 11549
    },
    {
      "epoch": 0.6232799093411041,
      "grad_norm": 1.2479132252919225,
      "learning_rate": 3.800415759887137e-06,
      "loss": 0.5145,
      "step": 11550
    },
    {
      "epoch": 0.6233338729696185,
      "grad_norm": 0.924546418782295,
      "learning_rate": 3.7997093959190813e-06,
      "loss": 0.3334,
      "step": 11551
    },
    {
      "epoch": 0.6233878365981329,
      "grad_norm": 0.9780000855073663,
      "learning_rate": 3.7990030808190945e-06,
      "loss": 0.4709,
      "step": 11552
    },
    {
      "epoch": 0.6234418002266472,
      "grad_norm": 0.9713783685808701,
      "learning_rate": 3.7982968146074804e-06,
      "loss": 0.4029,
      "step": 11553
    },
    {
      "epoch": 0.6234957638551616,
      "grad_norm": 1.2003220978670202,
      "learning_rate": 3.7975905973045337e-06,
      "loss": 0.5824,
      "step": 11554
    },
    {
      "epoch": 0.623549727483676,
      "grad_norm": 0.9366943475626447,
      "learning_rate": 3.796884428930553e-06,
      "loss": 0.4996,
      "step": 11555
    },
    {
      "epoch": 0.6236036911121904,
      "grad_norm": 0.8945406279443906,
      "learning_rate": 3.7961783095058347e-06,
      "loss": 0.3327,
      "step": 11556
    },
    {
      "epoch": 0.6236576547407048,
      "grad_norm": 0.9688078471149181,
      "learning_rate": 3.795472239050674e-06,
      "loss": 0.3942,
      "step": 11557
    },
    {
      "epoch": 0.6237116183692192,
      "grad_norm": 0.9170887510164533,
      "learning_rate": 3.7947662175853627e-06,
      "loss": 0.3205,
      "step": 11558
    },
    {
      "epoch": 0.6237655819977336,
      "grad_norm": 1.0084836674977515,
      "learning_rate": 3.7940602451301935e-06,
      "loss": 0.4654,
      "step": 11559
    },
    {
      "epoch": 0.6238195456262479,
      "grad_norm": 1.122805904996504,
      "learning_rate": 3.793354321705458e-06,
      "loss": 0.4057,
      "step": 11560
    },
    {
      "epoch": 0.6238735092547623,
      "grad_norm": 0.9506578959961349,
      "learning_rate": 3.7926484473314427e-06,
      "loss": 0.4768,
      "step": 11561
    },
    {
      "epoch": 0.6239274728832767,
      "grad_norm": 0.9091604382347543,
      "learning_rate": 3.791942622028436e-06,
      "loss": 0.4445,
      "step": 11562
    },
    {
      "epoch": 0.623981436511791,
      "grad_norm": 0.8983705326791864,
      "learning_rate": 3.7912368458167248e-06,
      "loss": 0.418,
      "step": 11563
    },
    {
      "epoch": 0.6240354001403055,
      "grad_norm": 0.8401039737429548,
      "learning_rate": 3.7905311187165915e-06,
      "loss": 0.3606,
      "step": 11564
    },
    {
      "epoch": 0.6240893637688198,
      "grad_norm": 0.9302936093242125,
      "learning_rate": 3.7898254407483228e-06,
      "loss": 0.3379,
      "step": 11565
    },
    {
      "epoch": 0.6241433273973342,
      "grad_norm": 0.9538797343565371,
      "learning_rate": 3.789119811932198e-06,
      "loss": 0.4008,
      "step": 11566
    },
    {
      "epoch": 0.6241972910258485,
      "grad_norm": 1.1463706446827544,
      "learning_rate": 3.788414232288501e-06,
      "loss": 0.462,
      "step": 11567
    },
    {
      "epoch": 0.6242512546543629,
      "grad_norm": 1.1583146393320152,
      "learning_rate": 3.787708701837507e-06,
      "loss": 0.5341,
      "step": 11568
    },
    {
      "epoch": 0.6243052182828773,
      "grad_norm": 1.0697244820346914,
      "learning_rate": 3.7870032205994932e-06,
      "loss": 0.4715,
      "step": 11569
    },
    {
      "epoch": 0.6243591819113917,
      "grad_norm": 1.0959630518959362,
      "learning_rate": 3.7862977885947383e-06,
      "loss": 0.502,
      "step": 11570
    },
    {
      "epoch": 0.6244131455399061,
      "grad_norm": 0.9497984371441222,
      "learning_rate": 3.785592405843518e-06,
      "loss": 0.4457,
      "step": 11571
    },
    {
      "epoch": 0.6244671091684205,
      "grad_norm": 1.1821280780503531,
      "learning_rate": 3.7848870723661034e-06,
      "loss": 0.5548,
      "step": 11572
    },
    {
      "epoch": 0.6245210727969349,
      "grad_norm": 1.030132976792053,
      "learning_rate": 3.784181788182767e-06,
      "loss": 0.451,
      "step": 11573
    },
    {
      "epoch": 0.6245750364254492,
      "grad_norm": 1.1111811361507382,
      "learning_rate": 3.7834765533137806e-06,
      "loss": 0.5717,
      "step": 11574
    },
    {
      "epoch": 0.6246290000539636,
      "grad_norm": 0.8147422995366819,
      "learning_rate": 3.782771367779412e-06,
      "loss": 0.3314,
      "step": 11575
    },
    {
      "epoch": 0.624682963682478,
      "grad_norm": 1.0774416585168225,
      "learning_rate": 3.782066231599929e-06,
      "loss": 0.5377,
      "step": 11576
    },
    {
      "epoch": 0.6247369273109924,
      "grad_norm": 1.1165243815721473,
      "learning_rate": 3.781361144795599e-06,
      "loss": 0.5664,
      "step": 11577
    },
    {
      "epoch": 0.6247908909395068,
      "grad_norm": 1.1599817289948917,
      "learning_rate": 3.7806561073866856e-06,
      "loss": 0.5419,
      "step": 11578
    },
    {
      "epoch": 0.6248448545680212,
      "grad_norm": 0.9204214764305291,
      "learning_rate": 3.7799511193934536e-06,
      "loss": 0.4776,
      "step": 11579
    },
    {
      "epoch": 0.6248988181965356,
      "grad_norm": 1.211367137196393,
      "learning_rate": 3.7792461808361646e-06,
      "loss": 0.5244,
      "step": 11580
    },
    {
      "epoch": 0.6249527818250499,
      "grad_norm": 1.1054190769534502,
      "learning_rate": 3.7785412917350784e-06,
      "loss": 0.5759,
      "step": 11581
    },
    {
      "epoch": 0.6250067454535643,
      "grad_norm": 1.0429684465018547,
      "learning_rate": 3.7778364521104562e-06,
      "loss": 0.4359,
      "step": 11582
    },
    {
      "epoch": 0.6250607090820787,
      "grad_norm": 1.0714344919891698,
      "learning_rate": 3.7771316619825525e-06,
      "loss": 0.4272,
      "step": 11583
    },
    {
      "epoch": 0.6251146727105931,
      "grad_norm": 1.059130732047016,
      "learning_rate": 3.7764269213716265e-06,
      "loss": 0.4472,
      "step": 11584
    },
    {
      "epoch": 0.6251686363391075,
      "grad_norm": 1.0573654642325392,
      "learning_rate": 3.7757222302979325e-06,
      "loss": 0.4428,
      "step": 11585
    },
    {
      "epoch": 0.6252225999676219,
      "grad_norm": 0.8975516685406597,
      "learning_rate": 3.775017588781723e-06,
      "loss": 0.4552,
      "step": 11586
    },
    {
      "epoch": 0.6252765635961363,
      "grad_norm": 0.9811041392208626,
      "learning_rate": 3.774312996843251e-06,
      "loss": 0.4632,
      "step": 11587
    },
    {
      "epoch": 0.6253305272246505,
      "grad_norm": 1.0903223695925204,
      "learning_rate": 3.7736084545027705e-06,
      "loss": 0.4566,
      "step": 11588
    },
    {
      "epoch": 0.6253844908531649,
      "grad_norm": 1.1923819753638967,
      "learning_rate": 3.772903961780524e-06,
      "loss": 0.6346,
      "step": 11589
    },
    {
      "epoch": 0.6254384544816793,
      "grad_norm": 0.9354916848440237,
      "learning_rate": 3.772199518696763e-06,
      "loss": 0.4485,
      "step": 11590
    },
    {
      "epoch": 0.6254924181101937,
      "grad_norm": 0.7347372659440651,
      "learning_rate": 3.7714951252717347e-06,
      "loss": 0.2314,
      "step": 11591
    },
    {
      "epoch": 0.6255463817387081,
      "grad_norm": 0.8559321772126645,
      "learning_rate": 3.7707907815256816e-06,
      "loss": 0.2799,
      "step": 11592
    },
    {
      "epoch": 0.6256003453672225,
      "grad_norm": 1.1525961488874417,
      "learning_rate": 3.77008648747885e-06,
      "loss": 0.4059,
      "step": 11593
    },
    {
      "epoch": 0.6256543089957368,
      "grad_norm": 0.9276976353189834,
      "learning_rate": 3.7693822431514816e-06,
      "loss": 0.3989,
      "step": 11594
    },
    {
      "epoch": 0.6257082726242512,
      "grad_norm": 0.9178772225317449,
      "learning_rate": 3.768678048563814e-06,
      "loss": 0.4041,
      "step": 11595
    },
    {
      "epoch": 0.6257622362527656,
      "grad_norm": 1.2488143611067255,
      "learning_rate": 3.7679739037360895e-06,
      "loss": 0.5703,
      "step": 11596
    },
    {
      "epoch": 0.62581619988128,
      "grad_norm": 1.0591032936061235,
      "learning_rate": 3.767269808688545e-06,
      "loss": 0.459,
      "step": 11597
    },
    {
      "epoch": 0.6258701635097944,
      "grad_norm": 0.7958163705623074,
      "learning_rate": 3.7665657634414177e-06,
      "loss": 0.3811,
      "step": 11598
    },
    {
      "epoch": 0.6259241271383088,
      "grad_norm": 1.2330052091864983,
      "learning_rate": 3.7658617680149413e-06,
      "loss": 0.5289,
      "step": 11599
    },
    {
      "epoch": 0.6259780907668232,
      "grad_norm": 1.0478984650886545,
      "learning_rate": 3.7651578224293496e-06,
      "loss": 0.5487,
      "step": 11600
    },
    {
      "epoch": 0.6260320543953375,
      "grad_norm": 0.7845806347612052,
      "learning_rate": 3.764453926704875e-06,
      "loss": 0.4793,
      "step": 11601
    },
    {
      "epoch": 0.6260860180238519,
      "grad_norm": 0.8435117173153428,
      "learning_rate": 3.763750080861749e-06,
      "loss": 0.3384,
      "step": 11602
    },
    {
      "epoch": 0.6261399816523663,
      "grad_norm": 0.9056833438074271,
      "learning_rate": 3.7630462849202e-06,
      "loss": 0.4874,
      "step": 11603
    },
    {
      "epoch": 0.6261939452808807,
      "grad_norm": 1.0203060482471775,
      "learning_rate": 3.7623425389004544e-06,
      "loss": 0.5253,
      "step": 11604
    },
    {
      "epoch": 0.6262479089093951,
      "grad_norm": 0.8984658379203077,
      "learning_rate": 3.7616388428227402e-06,
      "loss": 0.3284,
      "step": 11605
    },
    {
      "epoch": 0.6263018725379095,
      "grad_norm": 0.8403292830617339,
      "learning_rate": 3.7609351967072827e-06,
      "loss": 0.3541,
      "step": 11606
    },
    {
      "epoch": 0.6263558361664239,
      "grad_norm": 1.2006389251135348,
      "learning_rate": 3.7602316005743043e-06,
      "loss": 0.5912,
      "step": 11607
    },
    {
      "epoch": 0.6264097997949382,
      "grad_norm": 0.922217729377116,
      "learning_rate": 3.7595280544440304e-06,
      "loss": 0.4611,
      "step": 11608
    },
    {
      "epoch": 0.6264637634234526,
      "grad_norm": 1.0451672658523248,
      "learning_rate": 3.758824558336677e-06,
      "loss": 0.4556,
      "step": 11609
    },
    {
      "epoch": 0.626517727051967,
      "grad_norm": 1.0518079519448627,
      "learning_rate": 3.7581211122724647e-06,
      "loss": 0.4753,
      "step": 11610
    },
    {
      "epoch": 0.6265716906804814,
      "grad_norm": 1.0395275725651656,
      "learning_rate": 3.757417716271612e-06,
      "loss": 0.4022,
      "step": 11611
    },
    {
      "epoch": 0.6266256543089957,
      "grad_norm": 0.8110092596466567,
      "learning_rate": 3.756714370354334e-06,
      "loss": 0.4436,
      "step": 11612
    },
    {
      "epoch": 0.6266796179375101,
      "grad_norm": 1.0364811668645844,
      "learning_rate": 3.7560110745408473e-06,
      "loss": 0.4835,
      "step": 11613
    },
    {
      "epoch": 0.6267335815660245,
      "grad_norm": 1.007840666062246,
      "learning_rate": 3.7553078288513646e-06,
      "loss": 0.4058,
      "step": 11614
    },
    {
      "epoch": 0.6267875451945388,
      "grad_norm": 1.080112167871475,
      "learning_rate": 3.754604633306098e-06,
      "loss": 0.4381,
      "step": 11615
    },
    {
      "epoch": 0.6268415088230532,
      "grad_norm": 0.7776489531961565,
      "learning_rate": 3.753901487925258e-06,
      "loss": 0.2505,
      "step": 11616
    },
    {
      "epoch": 0.6268954724515676,
      "grad_norm": 1.0771798668283927,
      "learning_rate": 3.7531983927290527e-06,
      "loss": 0.5081,
      "step": 11617
    },
    {
      "epoch": 0.626949436080082,
      "grad_norm": 0.8841763735992666,
      "learning_rate": 3.75249534773769e-06,
      "loss": 0.4043,
      "step": 11618
    },
    {
      "epoch": 0.6270033997085964,
      "grad_norm": 0.996249346525534,
      "learning_rate": 3.7517923529713782e-06,
      "loss": 0.3962,
      "step": 11619
    },
    {
      "epoch": 0.6270573633371108,
      "grad_norm": 0.9461680106798867,
      "learning_rate": 3.7510894084503192e-06,
      "loss": 0.3866,
      "step": 11620
    },
    {
      "epoch": 0.6271113269656252,
      "grad_norm": 1.1548245341890442,
      "learning_rate": 3.7503865141947194e-06,
      "loss": 0.4428,
      "step": 11621
    },
    {
      "epoch": 0.6271652905941395,
      "grad_norm": 1.1337988179244398,
      "learning_rate": 3.7496836702247785e-06,
      "loss": 0.5397,
      "step": 11622
    },
    {
      "epoch": 0.6272192542226539,
      "grad_norm": 1.1702019146193667,
      "learning_rate": 3.7489808765606982e-06,
      "loss": 0.4561,
      "step": 11623
    },
    {
      "epoch": 0.6272732178511683,
      "grad_norm": 1.21364607321981,
      "learning_rate": 3.7482781332226747e-06,
      "loss": 0.6952,
      "step": 11624
    },
    {
      "epoch": 0.6273271814796827,
      "grad_norm": 1.0767799230851904,
      "learning_rate": 3.7475754402309094e-06,
      "loss": 0.3205,
      "step": 11625
    },
    {
      "epoch": 0.6273811451081971,
      "grad_norm": 1.138832281184855,
      "learning_rate": 3.746872797605595e-06,
      "loss": 0.523,
      "step": 11626
    },
    {
      "epoch": 0.6274351087367115,
      "grad_norm": 1.1422907069043178,
      "learning_rate": 3.746170205366929e-06,
      "loss": 0.5401,
      "step": 11627
    },
    {
      "epoch": 0.6274890723652259,
      "grad_norm": 0.9796105119088424,
      "learning_rate": 3.745467663535103e-06,
      "loss": 0.3963,
      "step": 11628
    },
    {
      "epoch": 0.6275430359937402,
      "grad_norm": 0.8546858449491889,
      "learning_rate": 3.74476517213031e-06,
      "loss": 0.4562,
      "step": 11629
    },
    {
      "epoch": 0.6275969996222546,
      "grad_norm": 0.8136013168374662,
      "learning_rate": 3.7440627311727383e-06,
      "loss": 0.3517,
      "step": 11630
    },
    {
      "epoch": 0.627650963250769,
      "grad_norm": 1.1031326340416858,
      "learning_rate": 3.743360340682578e-06,
      "loss": 0.4331,
      "step": 11631
    },
    {
      "epoch": 0.6277049268792834,
      "grad_norm": 0.9879294550014037,
      "learning_rate": 3.7426580006800174e-06,
      "loss": 0.5062,
      "step": 11632
    },
    {
      "epoch": 0.6277588905077978,
      "grad_norm": 0.9691511852109512,
      "learning_rate": 3.7419557111852413e-06,
      "loss": 0.3562,
      "step": 11633
    },
    {
      "epoch": 0.6278128541363122,
      "grad_norm": 1.1109121074475676,
      "learning_rate": 3.7412534722184336e-06,
      "loss": 0.5568,
      "step": 11634
    },
    {
      "epoch": 0.6278668177648266,
      "grad_norm": 1.227783362767662,
      "learning_rate": 3.740551283799778e-06,
      "loss": 0.6361,
      "step": 11635
    },
    {
      "epoch": 0.6279207813933408,
      "grad_norm": 1.0118662703843477,
      "learning_rate": 3.739849145949459e-06,
      "loss": 0.4964,
      "step": 11636
    },
    {
      "epoch": 0.6279747450218552,
      "grad_norm": 1.0274253880676425,
      "learning_rate": 3.739147058687652e-06,
      "loss": 0.4158,
      "step": 11637
    },
    {
      "epoch": 0.6280287086503696,
      "grad_norm": 1.0182225473257707,
      "learning_rate": 3.738445022034537e-06,
      "loss": 0.3259,
      "step": 11638
    },
    {
      "epoch": 0.628082672278884,
      "grad_norm": 1.1090592997009976,
      "learning_rate": 3.737743036010294e-06,
      "loss": 0.4758,
      "step": 11639
    },
    {
      "epoch": 0.6281366359073984,
      "grad_norm": 1.0986149292798815,
      "learning_rate": 3.7370411006350947e-06,
      "loss": 0.384,
      "step": 11640
    },
    {
      "epoch": 0.6281905995359128,
      "grad_norm": 0.972626637905103,
      "learning_rate": 3.7363392159291157e-06,
      "loss": 0.3796,
      "step": 11641
    },
    {
      "epoch": 0.6282445631644272,
      "grad_norm": 1.0386424619811678,
      "learning_rate": 3.73563738191253e-06,
      "loss": 0.4509,
      "step": 11642
    },
    {
      "epoch": 0.6282985267929415,
      "grad_norm": 1.2220456985026242,
      "learning_rate": 3.7349355986055114e-06,
      "loss": 0.4609,
      "step": 11643
    },
    {
      "epoch": 0.6283524904214559,
      "grad_norm": 1.054705295893981,
      "learning_rate": 3.7342338660282253e-06,
      "loss": 0.5099,
      "step": 11644
    },
    {
      "epoch": 0.6284064540499703,
      "grad_norm": 0.9647182094566822,
      "learning_rate": 3.733532184200841e-06,
      "loss": 0.4806,
      "step": 11645
    },
    {
      "epoch": 0.6284604176784847,
      "grad_norm": 0.8836595615388159,
      "learning_rate": 3.732830553143527e-06,
      "loss": 0.4165,
      "step": 11646
    },
    {
      "epoch": 0.6285143813069991,
      "grad_norm": 1.3702280279025734,
      "learning_rate": 3.7321289728764486e-06,
      "loss": 0.5697,
      "step": 11647
    },
    {
      "epoch": 0.6285683449355135,
      "grad_norm": 1.4315060187621809,
      "learning_rate": 3.7314274434197693e-06,
      "loss": 0.4328,
      "step": 11648
    },
    {
      "epoch": 0.6286223085640279,
      "grad_norm": 1.0472382471392156,
      "learning_rate": 3.7307259647936532e-06,
      "loss": 0.6339,
      "step": 11649
    },
    {
      "epoch": 0.6286762721925422,
      "grad_norm": 0.8660328281585147,
      "learning_rate": 3.730024537018261e-06,
      "loss": 0.3906,
      "step": 11650
    },
    {
      "epoch": 0.6287302358210566,
      "grad_norm": 1.2445898764758339,
      "learning_rate": 3.729323160113751e-06,
      "loss": 0.4481,
      "step": 11651
    },
    {
      "epoch": 0.628784199449571,
      "grad_norm": 0.933750700421194,
      "learning_rate": 3.728621834100283e-06,
      "loss": 0.3148,
      "step": 11652
    },
    {
      "epoch": 0.6288381630780854,
      "grad_norm": 1.1723498380488844,
      "learning_rate": 3.727920558998013e-06,
      "loss": 0.6755,
      "step": 11653
    },
    {
      "epoch": 0.6288921267065998,
      "grad_norm": 1.1932719201980875,
      "learning_rate": 3.727219334827098e-06,
      "loss": 0.5757,
      "step": 11654
    },
    {
      "epoch": 0.6289460903351142,
      "grad_norm": 0.9868582756941889,
      "learning_rate": 3.7265181616076885e-06,
      "loss": 0.4277,
      "step": 11655
    },
    {
      "epoch": 0.6290000539636286,
      "grad_norm": 1.0309852601601595,
      "learning_rate": 3.7258170393599404e-06,
      "loss": 0.409,
      "step": 11656
    },
    {
      "epoch": 0.6290540175921429,
      "grad_norm": 0.9173323510012511,
      "learning_rate": 3.725115968104004e-06,
      "loss": 0.3147,
      "step": 11657
    },
    {
      "epoch": 0.6291079812206573,
      "grad_norm": 1.1190774515763384,
      "learning_rate": 3.724414947860027e-06,
      "loss": 0.4382,
      "step": 11658
    },
    {
      "epoch": 0.6291619448491717,
      "grad_norm": 1.199977446228294,
      "learning_rate": 3.723713978648157e-06,
      "loss": 0.6637,
      "step": 11659
    },
    {
      "epoch": 0.629215908477686,
      "grad_norm": 0.8427856822086506,
      "learning_rate": 3.7230130604885437e-06,
      "loss": 0.4044,
      "step": 11660
    },
    {
      "epoch": 0.6292698721062004,
      "grad_norm": 0.731891875933744,
      "learning_rate": 3.7223121934013306e-06,
      "loss": 0.3277,
      "step": 11661
    },
    {
      "epoch": 0.6293238357347148,
      "grad_norm": 1.0048410123749991,
      "learning_rate": 3.72161137740666e-06,
      "loss": 0.4143,
      "step": 11662
    },
    {
      "epoch": 0.6293777993632291,
      "grad_norm": 1.0700409042219783,
      "learning_rate": 3.7209106125246787e-06,
      "loss": 0.5134,
      "step": 11663
    },
    {
      "epoch": 0.6294317629917435,
      "grad_norm": 0.970091099034213,
      "learning_rate": 3.720209898775522e-06,
      "loss": 0.4349,
      "step": 11664
    },
    {
      "epoch": 0.6294857266202579,
      "grad_norm": 1.2023324763135415,
      "learning_rate": 3.7195092361793306e-06,
      "loss": 0.6205,
      "step": 11665
    },
    {
      "epoch": 0.6295396902487723,
      "grad_norm": 1.1679626523955267,
      "learning_rate": 3.7188086247562427e-06,
      "loss": 0.624,
      "step": 11666
    },
    {
      "epoch": 0.6295936538772867,
      "grad_norm": 1.0010094433406067,
      "learning_rate": 3.7181080645263946e-06,
      "loss": 0.4779,
      "step": 11667
    },
    {
      "epoch": 0.6296476175058011,
      "grad_norm": 0.9958824760280602,
      "learning_rate": 3.717407555509922e-06,
      "loss": 0.3253,
      "step": 11668
    },
    {
      "epoch": 0.6297015811343155,
      "grad_norm": 1.0197433857095815,
      "learning_rate": 3.716707097726957e-06,
      "loss": 0.4008,
      "step": 11669
    },
    {
      "epoch": 0.6297555447628298,
      "grad_norm": 1.1055915310681876,
      "learning_rate": 3.7160066911976334e-06,
      "loss": 0.6593,
      "step": 11670
    },
    {
      "epoch": 0.6298095083913442,
      "grad_norm": 1.286844932121823,
      "learning_rate": 3.715306335942078e-06,
      "loss": 0.7143,
      "step": 11671
    },
    {
      "epoch": 0.6298634720198586,
      "grad_norm": 1.256101813301474,
      "learning_rate": 3.7146060319804224e-06,
      "loss": 0.5866,
      "step": 11672
    },
    {
      "epoch": 0.629917435648373,
      "grad_norm": 0.8722356688289632,
      "learning_rate": 3.7139057793327938e-06,
      "loss": 0.376,
      "step": 11673
    },
    {
      "epoch": 0.6299713992768874,
      "grad_norm": 0.8813420606954312,
      "learning_rate": 3.7132055780193166e-06,
      "loss": 0.3838,
      "step": 11674
    },
    {
      "epoch": 0.6300253629054018,
      "grad_norm": 0.9021867270093612,
      "learning_rate": 3.712505428060118e-06,
      "loss": 0.3751,
      "step": 11675
    },
    {
      "epoch": 0.6300793265339162,
      "grad_norm": 0.9959021176662156,
      "learning_rate": 3.711805329475319e-06,
      "loss": 0.3355,
      "step": 11676
    },
    {
      "epoch": 0.6301332901624305,
      "grad_norm": 0.9858848285381606,
      "learning_rate": 3.7111052822850425e-06,
      "loss": 0.4605,
      "step": 11677
    },
    {
      "epoch": 0.6301872537909449,
      "grad_norm": 1.076277645842063,
      "learning_rate": 3.7104052865094065e-06,
      "loss": 0.504,
      "step": 11678
    },
    {
      "epoch": 0.6302412174194593,
      "grad_norm": 0.9976856307869747,
      "learning_rate": 3.7097053421685315e-06,
      "loss": 0.4596,
      "step": 11679
    },
    {
      "epoch": 0.6302951810479737,
      "grad_norm": 0.7545665241132605,
      "learning_rate": 3.7090054492825335e-06,
      "loss": 0.2854,
      "step": 11680
    },
    {
      "epoch": 0.6303491446764881,
      "grad_norm": 1.095332499610052,
      "learning_rate": 3.7083056078715273e-06,
      "loss": 0.4593,
      "step": 11681
    },
    {
      "epoch": 0.6304031083050025,
      "grad_norm": 0.8453865941654672,
      "learning_rate": 3.7076058179556295e-06,
      "loss": 0.3531,
      "step": 11682
    },
    {
      "epoch": 0.6304570719335169,
      "grad_norm": 1.076654476342824,
      "learning_rate": 3.7069060795549516e-06,
      "loss": 0.6186,
      "step": 11683
    },
    {
      "epoch": 0.6305110355620311,
      "grad_norm": 1.128150137968342,
      "learning_rate": 3.7062063926896073e-06,
      "loss": 0.498,
      "step": 11684
    },
    {
      "epoch": 0.6305649991905455,
      "grad_norm": 1.042648985356353,
      "learning_rate": 3.705506757379701e-06,
      "loss": 0.5151,
      "step": 11685
    },
    {
      "epoch": 0.6306189628190599,
      "grad_norm": 0.8118238529726451,
      "learning_rate": 3.704807173645344e-06,
      "loss": 0.3448,
      "step": 11686
    },
    {
      "epoch": 0.6306729264475743,
      "grad_norm": 0.8986399490274769,
      "learning_rate": 3.7041076415066425e-06,
      "loss": 0.4204,
      "step": 11687
    },
    {
      "epoch": 0.6307268900760887,
      "grad_norm": 0.9099447086226061,
      "learning_rate": 3.7034081609837025e-06,
      "loss": 0.4171,
      "step": 11688
    },
    {
      "epoch": 0.6307808537046031,
      "grad_norm": 0.9891140489437573,
      "learning_rate": 3.702708732096627e-06,
      "loss": 0.5149,
      "step": 11689
    },
    {
      "epoch": 0.6308348173331175,
      "grad_norm": 0.8196665577549329,
      "learning_rate": 3.702009354865518e-06,
      "loss": 0.2636,
      "step": 11690
    },
    {
      "epoch": 0.6308887809616318,
      "grad_norm": 1.083380036235683,
      "learning_rate": 3.70131002931048e-06,
      "loss": 0.5408,
      "step": 11691
    },
    {
      "epoch": 0.6309427445901462,
      "grad_norm": 0.8782187997392663,
      "learning_rate": 3.7006107554516057e-06,
      "loss": 0.3047,
      "step": 11692
    },
    {
      "epoch": 0.6309967082186606,
      "grad_norm": 0.8269741178468035,
      "learning_rate": 3.6999115333089986e-06,
      "loss": 0.3119,
      "step": 11693
    },
    {
      "epoch": 0.631050671847175,
      "grad_norm": 0.914725171557676,
      "learning_rate": 3.699212362902751e-06,
      "loss": 0.3776,
      "step": 11694
    },
    {
      "epoch": 0.6311046354756894,
      "grad_norm": 0.9846181252164405,
      "learning_rate": 3.6985132442529605e-06,
      "loss": 0.4353,
      "step": 11695
    },
    {
      "epoch": 0.6311585991042038,
      "grad_norm": 0.9741012372124475,
      "learning_rate": 3.6978141773797194e-06,
      "loss": 0.4273,
      "step": 11696
    },
    {
      "epoch": 0.6312125627327182,
      "grad_norm": 1.2728170911635095,
      "learning_rate": 3.69711516230312e-06,
      "loss": 0.5774,
      "step": 11697
    },
    {
      "epoch": 0.6312665263612325,
      "grad_norm": 1.0066299198571715,
      "learning_rate": 3.6964161990432545e-06,
      "loss": 0.5732,
      "step": 11698
    },
    {
      "epoch": 0.6313204899897469,
      "grad_norm": 0.9843562650606072,
      "learning_rate": 3.6957172876202074e-06,
      "loss": 0.5185,
      "step": 11699
    },
    {
      "epoch": 0.6313744536182613,
      "grad_norm": 1.144806269912816,
      "learning_rate": 3.695018428054068e-06,
      "loss": 0.5157,
      "step": 11700
    },
    {
      "epoch": 0.6314284172467757,
      "grad_norm": 0.934304918588507,
      "learning_rate": 3.6943196203649246e-06,
      "loss": 0.41,
      "step": 11701
    },
    {
      "epoch": 0.6314823808752901,
      "grad_norm": 1.030752103267561,
      "learning_rate": 3.693620864572859e-06,
      "loss": 0.4689,
      "step": 11702
    },
    {
      "epoch": 0.6315363445038045,
      "grad_norm": 0.976610073811376,
      "learning_rate": 3.6929221606979536e-06,
      "loss": 0.4614,
      "step": 11703
    },
    {
      "epoch": 0.6315903081323189,
      "grad_norm": 1.114618296797929,
      "learning_rate": 3.692223508760293e-06,
      "loss": 0.4964,
      "step": 11704
    },
    {
      "epoch": 0.6316442717608332,
      "grad_norm": 0.9587586310768876,
      "learning_rate": 3.6915249087799554e-06,
      "loss": 0.5071,
      "step": 11705
    },
    {
      "epoch": 0.6316982353893476,
      "grad_norm": 1.1630814868514119,
      "learning_rate": 3.690826360777019e-06,
      "loss": 0.4578,
      "step": 11706
    },
    {
      "epoch": 0.631752199017862,
      "grad_norm": 1.3410467696147954,
      "learning_rate": 3.69012786477156e-06,
      "loss": 0.5614,
      "step": 11707
    },
    {
      "epoch": 0.6318061626463763,
      "grad_norm": 0.9651012589296448,
      "learning_rate": 3.6894294207836556e-06,
      "loss": 0.547,
      "step": 11708
    },
    {
      "epoch": 0.6318601262748907,
      "grad_norm": 0.8981317530965073,
      "learning_rate": 3.68873102883338e-06,
      "loss": 0.4149,
      "step": 11709
    },
    {
      "epoch": 0.6319140899034051,
      "grad_norm": 1.0997023135446904,
      "learning_rate": 3.6880326889408034e-06,
      "loss": 0.471,
      "step": 11710
    },
    {
      "epoch": 0.6319680535319195,
      "grad_norm": 1.266019058637315,
      "learning_rate": 3.6873344011259983e-06,
      "loss": 0.424,
      "step": 11711
    },
    {
      "epoch": 0.6320220171604338,
      "grad_norm": 0.9887344924106649,
      "learning_rate": 3.6866361654090367e-06,
      "loss": 0.4476,
      "step": 11712
    },
    {
      "epoch": 0.6320759807889482,
      "grad_norm": 1.1167670976026245,
      "learning_rate": 3.6859379818099818e-06,
      "loss": 0.7563,
      "step": 11713
    },
    {
      "epoch": 0.6321299444174626,
      "grad_norm": 1.0084734912661986,
      "learning_rate": 3.6852398503489016e-06,
      "loss": 0.4309,
      "step": 11714
    },
    {
      "epoch": 0.632183908045977,
      "grad_norm": 1.0384013179529583,
      "learning_rate": 3.684541771045862e-06,
      "loss": 0.5946,
      "step": 11715
    },
    {
      "epoch": 0.6322378716744914,
      "grad_norm": 0.7992303223349975,
      "learning_rate": 3.6838437439209274e-06,
      "loss": 0.4608,
      "step": 11716
    },
    {
      "epoch": 0.6322918353030058,
      "grad_norm": 0.9190972725837366,
      "learning_rate": 3.6831457689941574e-06,
      "loss": 0.3273,
      "step": 11717
    },
    {
      "epoch": 0.6323457989315202,
      "grad_norm": 1.082777640117104,
      "learning_rate": 3.682447846285615e-06,
      "loss": 0.4226,
      "step": 11718
    },
    {
      "epoch": 0.6323997625600345,
      "grad_norm": 1.0624136579897412,
      "learning_rate": 3.6817499758153585e-06,
      "loss": 0.48,
      "step": 11719
    },
    {
      "epoch": 0.6324537261885489,
      "grad_norm": 0.9306937391065592,
      "learning_rate": 3.6810521576034443e-06,
      "loss": 0.4004,
      "step": 11720
    },
    {
      "epoch": 0.6325076898170633,
      "grad_norm": 1.1300526621415414,
      "learning_rate": 3.680354391669928e-06,
      "loss": 0.6041,
      "step": 11721
    },
    {
      "epoch": 0.6325616534455777,
      "grad_norm": 0.9389758204929867,
      "learning_rate": 3.6796566780348654e-06,
      "loss": 0.4431,
      "step": 11722
    },
    {
      "epoch": 0.6326156170740921,
      "grad_norm": 0.9998577474096917,
      "learning_rate": 3.6789590167183096e-06,
      "loss": 0.4533,
      "step": 11723
    },
    {
      "epoch": 0.6326695807026065,
      "grad_norm": 0.9364030542920666,
      "learning_rate": 3.6782614077403113e-06,
      "loss": 0.3821,
      "step": 11724
    },
    {
      "epoch": 0.6327235443311209,
      "grad_norm": 0.9977902826417548,
      "learning_rate": 3.677563851120921e-06,
      "loss": 0.449,
      "step": 11725
    },
    {
      "epoch": 0.6327775079596352,
      "grad_norm": 1.014305020670388,
      "learning_rate": 3.676866346880188e-06,
      "loss": 0.3804,
      "step": 11726
    },
    {
      "epoch": 0.6328314715881496,
      "grad_norm": 1.3218627861607557,
      "learning_rate": 3.676168895038158e-06,
      "loss": 0.6249,
      "step": 11727
    },
    {
      "epoch": 0.632885435216664,
      "grad_norm": 1.0301926009667381,
      "learning_rate": 3.675471495614875e-06,
      "loss": 0.4775,
      "step": 11728
    },
    {
      "epoch": 0.6329393988451784,
      "grad_norm": 1.0086342543379974,
      "learning_rate": 3.674774148630388e-06,
      "loss": 0.4421,
      "step": 11729
    },
    {
      "epoch": 0.6329933624736928,
      "grad_norm": 0.8618297181325053,
      "learning_rate": 3.6740768541047333e-06,
      "loss": 0.3693,
      "step": 11730
    },
    {
      "epoch": 0.6330473261022072,
      "grad_norm": 1.0623789744551002,
      "learning_rate": 3.6733796120579557e-06,
      "loss": 0.351,
      "step": 11731
    },
    {
      "epoch": 0.6331012897307214,
      "grad_norm": 0.9965590613522162,
      "learning_rate": 3.6726824225100942e-06,
      "loss": 0.4471,
      "step": 11732
    },
    {
      "epoch": 0.6331552533592358,
      "grad_norm": 0.9457573040209599,
      "learning_rate": 3.6719852854811868e-06,
      "loss": 0.4117,
      "step": 11733
    },
    {
      "epoch": 0.6332092169877502,
      "grad_norm": 1.010615084047796,
      "learning_rate": 3.6712882009912677e-06,
      "loss": 0.3935,
      "step": 11734
    },
    {
      "epoch": 0.6332631806162646,
      "grad_norm": 0.7802987425201245,
      "learning_rate": 3.6705911690603745e-06,
      "loss": 0.2624,
      "step": 11735
    },
    {
      "epoch": 0.633317144244779,
      "grad_norm": 1.1635770869940834,
      "learning_rate": 3.669894189708538e-06,
      "loss": 0.5546,
      "step": 11736
    },
    {
      "epoch": 0.6333711078732934,
      "grad_norm": 0.9650047359301667,
      "learning_rate": 3.6691972629557936e-06,
      "loss": 0.4187,
      "step": 11737
    },
    {
      "epoch": 0.6334250715018078,
      "grad_norm": 1.0599117757331322,
      "learning_rate": 3.6685003888221694e-06,
      "loss": 0.4452,
      "step": 11738
    },
    {
      "epoch": 0.6334790351303221,
      "grad_norm": 1.0180645462683335,
      "learning_rate": 3.6678035673276956e-06,
      "loss": 0.4997,
      "step": 11739
    },
    {
      "epoch": 0.6335329987588365,
      "grad_norm": 1.1214621536844407,
      "learning_rate": 3.667106798492397e-06,
      "loss": 0.4597,
      "step": 11740
    },
    {
      "epoch": 0.6335869623873509,
      "grad_norm": 0.9338134232366744,
      "learning_rate": 3.6664100823363014e-06,
      "loss": 0.4799,
      "step": 11741
    },
    {
      "epoch": 0.6336409260158653,
      "grad_norm": 1.0793192232475415,
      "learning_rate": 3.665713418879433e-06,
      "loss": 0.5026,
      "step": 11742
    },
    {
      "epoch": 0.6336948896443797,
      "grad_norm": 1.2837248177445746,
      "learning_rate": 3.6650168081418137e-06,
      "loss": 0.675,
      "step": 11743
    },
    {
      "epoch": 0.6337488532728941,
      "grad_norm": 0.8821112382703615,
      "learning_rate": 3.664320250143465e-06,
      "loss": 0.4615,
      "step": 11744
    },
    {
      "epoch": 0.6338028169014085,
      "grad_norm": 1.0390053588204784,
      "learning_rate": 3.6636237449044077e-06,
      "loss": 0.5128,
      "step": 11745
    },
    {
      "epoch": 0.6338567805299228,
      "grad_norm": 1.1083850881743846,
      "learning_rate": 3.6629272924446622e-06,
      "loss": 0.5426,
      "step": 11746
    },
    {
      "epoch": 0.6339107441584372,
      "grad_norm": 0.9221822393442772,
      "learning_rate": 3.66223089278424e-06,
      "loss": 0.3907,
      "step": 11747
    },
    {
      "epoch": 0.6339647077869516,
      "grad_norm": 1.1716208385700353,
      "learning_rate": 3.6615345459431583e-06,
      "loss": 0.5321,
      "step": 11748
    },
    {
      "epoch": 0.634018671415466,
      "grad_norm": 1.3624881507810442,
      "learning_rate": 3.660838251941432e-06,
      "loss": 0.4521,
      "step": 11749
    },
    {
      "epoch": 0.6340726350439804,
      "grad_norm": 0.9901953072390286,
      "learning_rate": 3.6601420107990727e-06,
      "loss": 0.4687,
      "step": 11750
    },
    {
      "epoch": 0.6341265986724948,
      "grad_norm": 1.0978701902977286,
      "learning_rate": 3.6594458225360906e-06,
      "loss": 0.5425,
      "step": 11751
    },
    {
      "epoch": 0.6341805623010092,
      "grad_norm": 0.9191404888677512,
      "learning_rate": 3.658749687172496e-06,
      "loss": 0.4357,
      "step": 11752
    },
    {
      "epoch": 0.6342345259295235,
      "grad_norm": 1.0095608054477112,
      "learning_rate": 3.6580536047282955e-06,
      "loss": 0.5322,
      "step": 11753
    },
    {
      "epoch": 0.6342884895580378,
      "grad_norm": 0.9293680994779544,
      "learning_rate": 3.657357575223496e-06,
      "loss": 0.4135,
      "step": 11754
    },
    {
      "epoch": 0.6343424531865522,
      "grad_norm": 1.0221860814496269,
      "learning_rate": 3.6566615986781006e-06,
      "loss": 0.46,
      "step": 11755
    },
    {
      "epoch": 0.6343964168150666,
      "grad_norm": 1.1175126011922853,
      "learning_rate": 3.655965675112114e-06,
      "loss": 0.6329,
      "step": 11756
    },
    {
      "epoch": 0.634450380443581,
      "grad_norm": 0.9387935258083498,
      "learning_rate": 3.655269804545537e-06,
      "loss": 0.6033,
      "step": 11757
    },
    {
      "epoch": 0.6345043440720954,
      "grad_norm": 1.0754432684328101,
      "learning_rate": 3.654573986998371e-06,
      "loss": 0.4818,
      "step": 11758
    },
    {
      "epoch": 0.6345583077006098,
      "grad_norm": 0.8793464758021582,
      "learning_rate": 3.653878222490611e-06,
      "loss": 0.398,
      "step": 11759
    },
    {
      "epoch": 0.6346122713291241,
      "grad_norm": 0.9718448636916581,
      "learning_rate": 3.653182511042259e-06,
      "loss": 0.419,
      "step": 11760
    },
    {
      "epoch": 0.6346662349576385,
      "grad_norm": 1.11300667295268,
      "learning_rate": 3.6524868526733073e-06,
      "loss": 0.5988,
      "step": 11761
    },
    {
      "epoch": 0.6347201985861529,
      "grad_norm": 1.0403284241725854,
      "learning_rate": 3.6517912474037476e-06,
      "loss": 0.4666,
      "step": 11762
    },
    {
      "epoch": 0.6347741622146673,
      "grad_norm": 0.8139932392406047,
      "learning_rate": 3.651095695253577e-06,
      "loss": 0.3183,
      "step": 11763
    },
    {
      "epoch": 0.6348281258431817,
      "grad_norm": 1.2571364390478301,
      "learning_rate": 3.650400196242784e-06,
      "loss": 0.5483,
      "step": 11764
    },
    {
      "epoch": 0.6348820894716961,
      "grad_norm": 1.1759958343925172,
      "learning_rate": 3.6497047503913572e-06,
      "loss": 0.5401,
      "step": 11765
    },
    {
      "epoch": 0.6349360531002105,
      "grad_norm": 1.2313514524367877,
      "learning_rate": 3.6490093577192865e-06,
      "loss": 0.648,
      "step": 11766
    },
    {
      "epoch": 0.6349900167287248,
      "grad_norm": 1.330833572791146,
      "learning_rate": 3.6483140182465583e-06,
      "loss": 0.523,
      "step": 11767
    },
    {
      "epoch": 0.6350439803572392,
      "grad_norm": 1.0048408878482038,
      "learning_rate": 3.6476187319931545e-06,
      "loss": 0.3735,
      "step": 11768
    },
    {
      "epoch": 0.6350979439857536,
      "grad_norm": 1.11546434498582,
      "learning_rate": 3.6469234989790605e-06,
      "loss": 0.4847,
      "step": 11769
    },
    {
      "epoch": 0.635151907614268,
      "grad_norm": 1.0820873463712257,
      "learning_rate": 3.6462283192242577e-06,
      "loss": 0.4686,
      "step": 11770
    },
    {
      "epoch": 0.6352058712427824,
      "grad_norm": 1.1136450400106024,
      "learning_rate": 3.6455331927487257e-06,
      "loss": 0.4593,
      "step": 11771
    },
    {
      "epoch": 0.6352598348712968,
      "grad_norm": 1.431720286416296,
      "learning_rate": 3.6448381195724436e-06,
      "loss": 0.4203,
      "step": 11772
    },
    {
      "epoch": 0.6353137984998112,
      "grad_norm": 1.002835473352176,
      "learning_rate": 3.644143099715388e-06,
      "loss": 0.4591,
      "step": 11773
    },
    {
      "epoch": 0.6353677621283255,
      "grad_norm": 0.8744532022576739,
      "learning_rate": 3.643448133197538e-06,
      "loss": 0.4838,
      "step": 11774
    },
    {
      "epoch": 0.6354217257568399,
      "grad_norm": 0.978742622056173,
      "learning_rate": 3.6427532200388617e-06,
      "loss": 0.4442,
      "step": 11775
    },
    {
      "epoch": 0.6354756893853543,
      "grad_norm": 1.1064385219350261,
      "learning_rate": 3.642058360259334e-06,
      "loss": 0.4714,
      "step": 11776
    },
    {
      "epoch": 0.6355296530138687,
      "grad_norm": 0.9806712768613906,
      "learning_rate": 3.6413635538789274e-06,
      "loss": 0.3649,
      "step": 11777
    },
    {
      "epoch": 0.635583616642383,
      "grad_norm": 1.0775012060147213,
      "learning_rate": 3.6406688009176095e-06,
      "loss": 0.651,
      "step": 11778
    },
    {
      "epoch": 0.6356375802708975,
      "grad_norm": 1.0076089895086626,
      "learning_rate": 3.6399741013953495e-06,
      "loss": 0.4275,
      "step": 11779
    },
    {
      "epoch": 0.6356915438994118,
      "grad_norm": 0.8334387672030034,
      "learning_rate": 3.6392794553321133e-06,
      "loss": 0.2848,
      "step": 11780
    },
    {
      "epoch": 0.6357455075279261,
      "grad_norm": 0.9330924501678919,
      "learning_rate": 3.638584862747866e-06,
      "loss": 0.447,
      "step": 11781
    },
    {
      "epoch": 0.6357994711564405,
      "grad_norm": 1.0187147514537467,
      "learning_rate": 3.637890323662571e-06,
      "loss": 0.4628,
      "step": 11782
    },
    {
      "epoch": 0.6358534347849549,
      "grad_norm": 0.8587067300209367,
      "learning_rate": 3.6371958380961882e-06,
      "loss": 0.3701,
      "step": 11783
    },
    {
      "epoch": 0.6359073984134693,
      "grad_norm": 0.8742220274216445,
      "learning_rate": 3.6365014060686788e-06,
      "loss": 0.3538,
      "step": 11784
    },
    {
      "epoch": 0.6359613620419837,
      "grad_norm": 1.2212896191119018,
      "learning_rate": 3.6358070276000036e-06,
      "loss": 0.5523,
      "step": 11785
    },
    {
      "epoch": 0.6360153256704981,
      "grad_norm": 1.0571852103517925,
      "learning_rate": 3.6351127027101175e-06,
      "loss": 0.4595,
      "step": 11786
    },
    {
      "epoch": 0.6360692892990125,
      "grad_norm": 1.046079150814298,
      "learning_rate": 3.6344184314189755e-06,
      "loss": 0.4774,
      "step": 11787
    },
    {
      "epoch": 0.6361232529275268,
      "grad_norm": 1.1446625630041969,
      "learning_rate": 3.6337242137465356e-06,
      "loss": 0.6144,
      "step": 11788
    },
    {
      "epoch": 0.6361772165560412,
      "grad_norm": 0.9637619789927471,
      "learning_rate": 3.633030049712746e-06,
      "loss": 0.4927,
      "step": 11789
    },
    {
      "epoch": 0.6362311801845556,
      "grad_norm": 1.1293497807962525,
      "learning_rate": 3.632335939337559e-06,
      "loss": 0.5546,
      "step": 11790
    },
    {
      "epoch": 0.63628514381307,
      "grad_norm": 1.1177844797491676,
      "learning_rate": 3.6316418826409238e-06,
      "loss": 0.5566,
      "step": 11791
    },
    {
      "epoch": 0.6363391074415844,
      "grad_norm": 0.9074640985434141,
      "learning_rate": 3.6309478796427887e-06,
      "loss": 0.4249,
      "step": 11792
    },
    {
      "epoch": 0.6363930710700988,
      "grad_norm": 0.8334220979043656,
      "learning_rate": 3.6302539303631003e-06,
      "loss": 0.3264,
      "step": 11793
    },
    {
      "epoch": 0.6364470346986132,
      "grad_norm": 1.008592552734655,
      "learning_rate": 3.629560034821804e-06,
      "loss": 0.411,
      "step": 11794
    },
    {
      "epoch": 0.6365009983271275,
      "grad_norm": 1.0296156176220612,
      "learning_rate": 3.6288661930388434e-06,
      "loss": 0.4417,
      "step": 11795
    },
    {
      "epoch": 0.6365549619556419,
      "grad_norm": 0.9906240420321379,
      "learning_rate": 3.628172405034157e-06,
      "loss": 0.456,
      "step": 11796
    },
    {
      "epoch": 0.6366089255841563,
      "grad_norm": 1.0160956466584452,
      "learning_rate": 3.6274786708276875e-06,
      "loss": 0.4539,
      "step": 11797
    },
    {
      "epoch": 0.6366628892126707,
      "grad_norm": 0.7707062941858106,
      "learning_rate": 3.6267849904393726e-06,
      "loss": 0.3556,
      "step": 11798
    },
    {
      "epoch": 0.6367168528411851,
      "grad_norm": 1.1040790885441074,
      "learning_rate": 3.6260913638891504e-06,
      "loss": 0.4418,
      "step": 11799
    },
    {
      "epoch": 0.6367708164696995,
      "grad_norm": 0.8577942777848409,
      "learning_rate": 3.6253977911969553e-06,
      "loss": 0.3636,
      "step": 11800
    },
    {
      "epoch": 0.6368247800982137,
      "grad_norm": 1.4709207495565477,
      "learning_rate": 3.624704272382723e-06,
      "loss": 0.6872,
      "step": 11801
    },
    {
      "epoch": 0.6368787437267281,
      "grad_norm": 0.9317767469216888,
      "learning_rate": 3.624010807466385e-06,
      "loss": 0.3399,
      "step": 11802
    },
    {
      "epoch": 0.6369327073552425,
      "grad_norm": 1.1684498908522445,
      "learning_rate": 3.6233173964678693e-06,
      "loss": 0.5054,
      "step": 11803
    },
    {
      "epoch": 0.6369866709837569,
      "grad_norm": 1.1519871793443182,
      "learning_rate": 3.622624039407109e-06,
      "loss": 0.6649,
      "step": 11804
    },
    {
      "epoch": 0.6370406346122713,
      "grad_norm": 1.1529575706695985,
      "learning_rate": 3.621930736304031e-06,
      "loss": 0.5264,
      "step": 11805
    },
    {
      "epoch": 0.6370945982407857,
      "grad_norm": 0.9194174560045032,
      "learning_rate": 3.6212374871785616e-06,
      "loss": 0.4148,
      "step": 11806
    },
    {
      "epoch": 0.6371485618693001,
      "grad_norm": 0.9341801292140862,
      "learning_rate": 3.620544292050625e-06,
      "loss": 0.3989,
      "step": 11807
    },
    {
      "epoch": 0.6372025254978144,
      "grad_norm": 1.0024498158620978,
      "learning_rate": 3.6198511509401467e-06,
      "loss": 0.4114,
      "step": 11808
    },
    {
      "epoch": 0.6372564891263288,
      "grad_norm": 0.957103362197787,
      "learning_rate": 3.6191580638670433e-06,
      "loss": 0.3812,
      "step": 11809
    },
    {
      "epoch": 0.6373104527548432,
      "grad_norm": 1.0058274417991109,
      "learning_rate": 3.618465030851238e-06,
      "loss": 0.3655,
      "step": 11810
    },
    {
      "epoch": 0.6373644163833576,
      "grad_norm": 0.9730580400485673,
      "learning_rate": 3.617772051912649e-06,
      "loss": 0.4483,
      "step": 11811
    },
    {
      "epoch": 0.637418380011872,
      "grad_norm": 1.0455260279646992,
      "learning_rate": 3.6170791270711925e-06,
      "loss": 0.5493,
      "step": 11812
    },
    {
      "epoch": 0.6374723436403864,
      "grad_norm": 0.9864622350730408,
      "learning_rate": 3.6163862563467854e-06,
      "loss": 0.4234,
      "step": 11813
    },
    {
      "epoch": 0.6375263072689008,
      "grad_norm": 0.8609351932831913,
      "learning_rate": 3.61569343975934e-06,
      "loss": 0.2874,
      "step": 11814
    },
    {
      "epoch": 0.6375802708974151,
      "grad_norm": 1.203990877520913,
      "learning_rate": 3.6150006773287694e-06,
      "loss": 0.5322,
      "step": 11815
    },
    {
      "epoch": 0.6376342345259295,
      "grad_norm": 0.9629819169263105,
      "learning_rate": 3.614307969074984e-06,
      "loss": 0.5394,
      "step": 11816
    },
    {
      "epoch": 0.6376881981544439,
      "grad_norm": 0.9846492842809291,
      "learning_rate": 3.613615315017892e-06,
      "loss": 0.3247,
      "step": 11817
    },
    {
      "epoch": 0.6377421617829583,
      "grad_norm": 0.9230098423040671,
      "learning_rate": 3.6129227151774027e-06,
      "loss": 0.5149,
      "step": 11818
    },
    {
      "epoch": 0.6377961254114727,
      "grad_norm": 0.9835220657053889,
      "learning_rate": 3.612230169573423e-06,
      "loss": 0.4401,
      "step": 11819
    },
    {
      "epoch": 0.6378500890399871,
      "grad_norm": 1.083103854230946,
      "learning_rate": 3.611537678225854e-06,
      "loss": 0.4839,
      "step": 11820
    },
    {
      "epoch": 0.6379040526685015,
      "grad_norm": 1.0107211048620754,
      "learning_rate": 3.610845241154599e-06,
      "loss": 0.4062,
      "step": 11821
    },
    {
      "epoch": 0.6379580162970158,
      "grad_norm": 0.8200151887418612,
      "learning_rate": 3.610152858379563e-06,
      "loss": 0.3761,
      "step": 11822
    },
    {
      "epoch": 0.6380119799255302,
      "grad_norm": 1.049109013075734,
      "learning_rate": 3.609460529920642e-06,
      "loss": 0.483,
      "step": 11823
    },
    {
      "epoch": 0.6380659435540446,
      "grad_norm": 1.2449391376260857,
      "learning_rate": 3.6087682557977354e-06,
      "loss": 0.5463,
      "step": 11824
    },
    {
      "epoch": 0.638119907182559,
      "grad_norm": 1.0392869453789984,
      "learning_rate": 3.60807603603074e-06,
      "loss": 0.5292,
      "step": 11825
    },
    {
      "epoch": 0.6381738708110734,
      "grad_norm": 0.9623057773497181,
      "learning_rate": 3.6073838706395515e-06,
      "loss": 0.4404,
      "step": 11826
    },
    {
      "epoch": 0.6382278344395877,
      "grad_norm": 1.0243394298402049,
      "learning_rate": 3.606691759644063e-06,
      "loss": 0.4221,
      "step": 11827
    },
    {
      "epoch": 0.6382817980681021,
      "grad_norm": 0.9227254197408107,
      "learning_rate": 3.6059997030641667e-06,
      "loss": 0.4289,
      "step": 11828
    },
    {
      "epoch": 0.6383357616966164,
      "grad_norm": 0.7130346358499351,
      "learning_rate": 3.6053077009197533e-06,
      "loss": 0.3031,
      "step": 11829
    },
    {
      "epoch": 0.6383897253251308,
      "grad_norm": 1.37439822397219,
      "learning_rate": 3.6046157532307102e-06,
      "loss": 0.6102,
      "step": 11830
    },
    {
      "epoch": 0.6384436889536452,
      "grad_norm": 0.9087633221020424,
      "learning_rate": 3.6039238600169246e-06,
      "loss": 0.3659,
      "step": 11831
    },
    {
      "epoch": 0.6384976525821596,
      "grad_norm": 1.2668161293989115,
      "learning_rate": 3.603232021298284e-06,
      "loss": 0.6256,
      "step": 11832
    },
    {
      "epoch": 0.638551616210674,
      "grad_norm": 0.959193550555773,
      "learning_rate": 3.6025402370946716e-06,
      "loss": 0.4063,
      "step": 11833
    },
    {
      "epoch": 0.6386055798391884,
      "grad_norm": 1.0784065121318387,
      "learning_rate": 3.60184850742597e-06,
      "loss": 0.6447,
      "step": 11834
    },
    {
      "epoch": 0.6386595434677028,
      "grad_norm": 1.1486671849872434,
      "learning_rate": 3.6011568323120593e-06,
      "loss": 0.5143,
      "step": 11835
    },
    {
      "epoch": 0.6387135070962171,
      "grad_norm": 1.0886315358567045,
      "learning_rate": 3.6004652117728233e-06,
      "loss": 0.4846,
      "step": 11836
    },
    {
      "epoch": 0.6387674707247315,
      "grad_norm": 1.059717028293982,
      "learning_rate": 3.599773645828134e-06,
      "loss": 0.4171,
      "step": 11837
    },
    {
      "epoch": 0.6388214343532459,
      "grad_norm": 0.9765178831125351,
      "learning_rate": 3.59908213449787e-06,
      "loss": 0.4814,
      "step": 11838
    },
    {
      "epoch": 0.6388753979817603,
      "grad_norm": 1.0601128174880687,
      "learning_rate": 3.598390677801906e-06,
      "loss": 0.4595,
      "step": 11839
    },
    {
      "epoch": 0.6389293616102747,
      "grad_norm": 1.156416807405208,
      "learning_rate": 3.597699275760115e-06,
      "loss": 0.4891,
      "step": 11840
    },
    {
      "epoch": 0.6389833252387891,
      "grad_norm": 1.0141580000687147,
      "learning_rate": 3.59700792839237e-06,
      "loss": 0.4731,
      "step": 11841
    },
    {
      "epoch": 0.6390372888673035,
      "grad_norm": 0.9870610302995407,
      "learning_rate": 3.5963166357185386e-06,
      "loss": 0.3475,
      "step": 11842
    },
    {
      "epoch": 0.6390912524958178,
      "grad_norm": 0.8680721287595856,
      "learning_rate": 3.595625397758493e-06,
      "loss": 0.3664,
      "step": 11843
    },
    {
      "epoch": 0.6391452161243322,
      "grad_norm": 1.2814171187417949,
      "learning_rate": 3.5949342145320963e-06,
      "loss": 0.5359,
      "step": 11844
    },
    {
      "epoch": 0.6391991797528466,
      "grad_norm": 0.8516095364958165,
      "learning_rate": 3.5942430860592157e-06,
      "loss": 0.3809,
      "step": 11845
    },
    {
      "epoch": 0.639253143381361,
      "grad_norm": 1.155759997730833,
      "learning_rate": 3.5935520123597135e-06,
      "loss": 0.5307,
      "step": 11846
    },
    {
      "epoch": 0.6393071070098754,
      "grad_norm": 1.0822458140400364,
      "learning_rate": 3.592860993453453e-06,
      "loss": 0.444,
      "step": 11847
    },
    {
      "epoch": 0.6393610706383898,
      "grad_norm": 0.9613989606704618,
      "learning_rate": 3.5921700293602945e-06,
      "loss": 0.4564,
      "step": 11848
    },
    {
      "epoch": 0.6394150342669042,
      "grad_norm": 1.0888574610948478,
      "learning_rate": 3.5914791201000976e-06,
      "loss": 0.4911,
      "step": 11849
    },
    {
      "epoch": 0.6394689978954184,
      "grad_norm": 1.1989843537209741,
      "learning_rate": 3.5907882656927197e-06,
      "loss": 0.6406,
      "step": 11850
    },
    {
      "epoch": 0.6395229615239328,
      "grad_norm": 0.990515817940133,
      "learning_rate": 3.5900974661580158e-06,
      "loss": 0.4356,
      "step": 11851
    },
    {
      "epoch": 0.6395769251524472,
      "grad_norm": 0.7850713174218881,
      "learning_rate": 3.58940672151584e-06,
      "loss": 0.3019,
      "step": 11852
    },
    {
      "epoch": 0.6396308887809616,
      "grad_norm": 0.7590479416073787,
      "learning_rate": 3.588716031786046e-06,
      "loss": 0.4237,
      "step": 11853
    },
    {
      "epoch": 0.639684852409476,
      "grad_norm": 1.1047549655136097,
      "learning_rate": 3.5880253969884836e-06,
      "loss": 0.4799,
      "step": 11854
    },
    {
      "epoch": 0.6397388160379904,
      "grad_norm": 1.1125404702873787,
      "learning_rate": 3.5873348171430032e-06,
      "loss": 0.5958,
      "step": 11855
    },
    {
      "epoch": 0.6397927796665048,
      "grad_norm": 0.9105936432571181,
      "learning_rate": 3.5866442922694533e-06,
      "loss": 0.3617,
      "step": 11856
    },
    {
      "epoch": 0.6398467432950191,
      "grad_norm": 1.2031948758219544,
      "learning_rate": 3.585953822387681e-06,
      "loss": 0.6519,
      "step": 11857
    },
    {
      "epoch": 0.6399007069235335,
      "grad_norm": 0.8786240910467211,
      "learning_rate": 3.5852634075175296e-06,
      "loss": 0.4227,
      "step": 11858
    },
    {
      "epoch": 0.6399546705520479,
      "grad_norm": 1.076703702229077,
      "learning_rate": 3.584573047678841e-06,
      "loss": 0.446,
      "step": 11859
    },
    {
      "epoch": 0.6400086341805623,
      "grad_norm": 0.9352386752843622,
      "learning_rate": 3.583882742891458e-06,
      "loss": 0.4432,
      "step": 11860
    },
    {
      "epoch": 0.6400625978090767,
      "grad_norm": 1.1917024856374494,
      "learning_rate": 3.5831924931752228e-06,
      "loss": 0.5753,
      "step": 11861
    },
    {
      "epoch": 0.6401165614375911,
      "grad_norm": 0.8973400470615696,
      "learning_rate": 3.5825022985499714e-06,
      "loss": 0.4015,
      "step": 11862
    },
    {
      "epoch": 0.6401705250661055,
      "grad_norm": 1.1400537019160801,
      "learning_rate": 3.5818121590355406e-06,
      "loss": 0.5438,
      "step": 11863
    },
    {
      "epoch": 0.6402244886946198,
      "grad_norm": 1.0355132503485394,
      "learning_rate": 3.5811220746517695e-06,
      "loss": 0.5055,
      "step": 11864
    },
    {
      "epoch": 0.6402784523231342,
      "grad_norm": 1.036482477837173,
      "learning_rate": 3.580432045418486e-06,
      "loss": 0.5454,
      "step": 11865
    },
    {
      "epoch": 0.6403324159516486,
      "grad_norm": 1.0180056641327966,
      "learning_rate": 3.5797420713555265e-06,
      "loss": 0.4254,
      "step": 11866
    },
    {
      "epoch": 0.640386379580163,
      "grad_norm": 0.917505645711224,
      "learning_rate": 3.5790521524827192e-06,
      "loss": 0.4709,
      "step": 11867
    },
    {
      "epoch": 0.6404403432086774,
      "grad_norm": 1.1545946425835654,
      "learning_rate": 3.5783622888198943e-06,
      "loss": 0.5769,
      "step": 11868
    },
    {
      "epoch": 0.6404943068371918,
      "grad_norm": 1.0431991640835372,
      "learning_rate": 3.577672480386878e-06,
      "loss": 0.5169,
      "step": 11869
    },
    {
      "epoch": 0.6405482704657061,
      "grad_norm": 0.9701285881827786,
      "learning_rate": 3.5769827272034975e-06,
      "loss": 0.5117,
      "step": 11870
    },
    {
      "epoch": 0.6406022340942205,
      "grad_norm": 1.0616998107306277,
      "learning_rate": 3.576293029289577e-06,
      "loss": 0.4275,
      "step": 11871
    },
    {
      "epoch": 0.6406561977227349,
      "grad_norm": 0.8677877830975936,
      "learning_rate": 3.5756033866649386e-06,
      "loss": 0.4124,
      "step": 11872
    },
    {
      "epoch": 0.6407101613512493,
      "grad_norm": 0.9352147137459771,
      "learning_rate": 3.5749137993494033e-06,
      "loss": 0.3787,
      "step": 11873
    },
    {
      "epoch": 0.6407641249797636,
      "grad_norm": 1.0011680153862383,
      "learning_rate": 3.5742242673627894e-06,
      "loss": 0.5526,
      "step": 11874
    },
    {
      "epoch": 0.640818088608278,
      "grad_norm": 1.053986654617322,
      "learning_rate": 3.5735347907249174e-06,
      "loss": 0.5636,
      "step": 11875
    },
    {
      "epoch": 0.6408720522367924,
      "grad_norm": 1.0706253210945322,
      "learning_rate": 3.572845369455601e-06,
      "loss": 0.5475,
      "step": 11876
    },
    {
      "epoch": 0.6409260158653067,
      "grad_norm": 0.9202411363457638,
      "learning_rate": 3.572156003574658e-06,
      "loss": 0.3487,
      "step": 11877
    },
    {
      "epoch": 0.6409799794938211,
      "grad_norm": 0.899863689164434,
      "learning_rate": 3.5714666931018983e-06,
      "loss": 0.3778,
      "step": 11878
    },
    {
      "epoch": 0.6410339431223355,
      "grad_norm": 1.0236463650333387,
      "learning_rate": 3.5707774380571335e-06,
      "loss": 0.4522,
      "step": 11879
    },
    {
      "epoch": 0.6410879067508499,
      "grad_norm": 1.1261955438551223,
      "learning_rate": 3.5700882384601755e-06,
      "loss": 0.5915,
      "step": 11880
    },
    {
      "epoch": 0.6411418703793643,
      "grad_norm": 1.0258030500347606,
      "learning_rate": 3.56939909433083e-06,
      "loss": 0.5951,
      "step": 11881
    },
    {
      "epoch": 0.6411958340078787,
      "grad_norm": 1.2158896633769478,
      "learning_rate": 3.5687100056889067e-06,
      "loss": 0.5773,
      "step": 11882
    },
    {
      "epoch": 0.6412497976363931,
      "grad_norm": 0.8365068509555995,
      "learning_rate": 3.5680209725542092e-06,
      "loss": 0.3966,
      "step": 11883
    },
    {
      "epoch": 0.6413037612649074,
      "grad_norm": 1.2697214576393172,
      "learning_rate": 3.5673319949465423e-06,
      "loss": 0.6151,
      "step": 11884
    },
    {
      "epoch": 0.6413577248934218,
      "grad_norm": 1.0620040751671826,
      "learning_rate": 3.566643072885705e-06,
      "loss": 0.4062,
      "step": 11885
    },
    {
      "epoch": 0.6414116885219362,
      "grad_norm": 1.1544363238427984,
      "learning_rate": 3.5659542063914986e-06,
      "loss": 0.4378,
      "step": 11886
    },
    {
      "epoch": 0.6414656521504506,
      "grad_norm": 1.1052501798928747,
      "learning_rate": 3.5652653954837248e-06,
      "loss": 0.3971,
      "step": 11887
    },
    {
      "epoch": 0.641519615778965,
      "grad_norm": 1.1378833052133042,
      "learning_rate": 3.5645766401821757e-06,
      "loss": 0.5962,
      "step": 11888
    },
    {
      "epoch": 0.6415735794074794,
      "grad_norm": 1.0205537698597862,
      "learning_rate": 3.563887940506651e-06,
      "loss": 0.4951,
      "step": 11889
    },
    {
      "epoch": 0.6416275430359938,
      "grad_norm": 0.8769584623939011,
      "learning_rate": 3.5631992964769435e-06,
      "loss": 0.4205,
      "step": 11890
    },
    {
      "epoch": 0.6416815066645081,
      "grad_norm": 0.9806507707233583,
      "learning_rate": 3.5625107081128465e-06,
      "loss": 0.4013,
      "step": 11891
    },
    {
      "epoch": 0.6417354702930225,
      "grad_norm": 1.158147346907811,
      "learning_rate": 3.5618221754341476e-06,
      "loss": 0.5503,
      "step": 11892
    },
    {
      "epoch": 0.6417894339215369,
      "grad_norm": 0.8313650267337248,
      "learning_rate": 3.561133698460637e-06,
      "loss": 0.3713,
      "step": 11893
    },
    {
      "epoch": 0.6418433975500513,
      "grad_norm": 1.0546089949668522,
      "learning_rate": 3.5604452772121047e-06,
      "loss": 0.4169,
      "step": 11894
    },
    {
      "epoch": 0.6418973611785657,
      "grad_norm": 1.071391496847875,
      "learning_rate": 3.559756911708333e-06,
      "loss": 0.4536,
      "step": 11895
    },
    {
      "epoch": 0.6419513248070801,
      "grad_norm": 1.006932852231791,
      "learning_rate": 3.55906860196911e-06,
      "loss": 0.4264,
      "step": 11896
    },
    {
      "epoch": 0.6420052884355945,
      "grad_norm": 1.2268579885616375,
      "learning_rate": 3.5583803480142153e-06,
      "loss": 0.5848,
      "step": 11897
    },
    {
      "epoch": 0.6420592520641087,
      "grad_norm": 1.217043942412284,
      "learning_rate": 3.5576921498634332e-06,
      "loss": 0.5177,
      "step": 11898
    },
    {
      "epoch": 0.6421132156926231,
      "grad_norm": 1.0088739460244016,
      "learning_rate": 3.5570040075365397e-06,
      "loss": 0.4419,
      "step": 11899
    },
    {
      "epoch": 0.6421671793211375,
      "grad_norm": 1.003730399324751,
      "learning_rate": 3.556315921053315e-06,
      "loss": 0.4928,
      "step": 11900
    },
    {
      "epoch": 0.6422211429496519,
      "grad_norm": 1.0670291735385713,
      "learning_rate": 3.5556278904335327e-06,
      "loss": 0.4995,
      "step": 11901
    },
    {
      "epoch": 0.6422751065781663,
      "grad_norm": 1.1270239846743635,
      "learning_rate": 3.554939915696971e-06,
      "loss": 0.725,
      "step": 11902
    },
    {
      "epoch": 0.6423290702066807,
      "grad_norm": 1.1539306702718852,
      "learning_rate": 3.5542519968634014e-06,
      "loss": 0.5147,
      "step": 11903
    },
    {
      "epoch": 0.6423830338351951,
      "grad_norm": 1.0589081125115505,
      "learning_rate": 3.553564133952595e-06,
      "loss": 0.4141,
      "step": 11904
    },
    {
      "epoch": 0.6424369974637094,
      "grad_norm": 0.8704627889021651,
      "learning_rate": 3.552876326984325e-06,
      "loss": 0.4628,
      "step": 11905
    },
    {
      "epoch": 0.6424909610922238,
      "grad_norm": 1.2241137542495393,
      "learning_rate": 3.5521885759783538e-06,
      "loss": 0.5089,
      "step": 11906
    },
    {
      "epoch": 0.6425449247207382,
      "grad_norm": 1.0431986454290278,
      "learning_rate": 3.5515008809544516e-06,
      "loss": 0.4745,
      "step": 11907
    },
    {
      "epoch": 0.6425988883492526,
      "grad_norm": 1.0667832693960269,
      "learning_rate": 3.5508132419323825e-06,
      "loss": 0.5309,
      "step": 11908
    },
    {
      "epoch": 0.642652851977767,
      "grad_norm": 1.0231897734631454,
      "learning_rate": 3.55012565893191e-06,
      "loss": 0.4872,
      "step": 11909
    },
    {
      "epoch": 0.6427068156062814,
      "grad_norm": 1.0107114621207502,
      "learning_rate": 3.549438131972797e-06,
      "loss": 0.5285,
      "step": 11910
    },
    {
      "epoch": 0.6427607792347958,
      "grad_norm": 0.851671503612425,
      "learning_rate": 3.5487506610748035e-06,
      "loss": 0.4303,
      "step": 11911
    },
    {
      "epoch": 0.6428147428633101,
      "grad_norm": 1.0272095779233312,
      "learning_rate": 3.548063246257688e-06,
      "loss": 0.4818,
      "step": 11912
    },
    {
      "epoch": 0.6428687064918245,
      "grad_norm": 1.084939575481687,
      "learning_rate": 3.5473758875412058e-06,
      "loss": 0.5225,
      "step": 11913
    },
    {
      "epoch": 0.6429226701203389,
      "grad_norm": 0.8491112396046413,
      "learning_rate": 3.546688584945114e-06,
      "loss": 0.3734,
      "step": 11914
    },
    {
      "epoch": 0.6429766337488533,
      "grad_norm": 0.8532936301543291,
      "learning_rate": 3.5460013384891658e-06,
      "loss": 0.352,
      "step": 11915
    },
    {
      "epoch": 0.6430305973773677,
      "grad_norm": 0.849745824626063,
      "learning_rate": 3.545314148193113e-06,
      "loss": 0.4371,
      "step": 11916
    },
    {
      "epoch": 0.6430845610058821,
      "grad_norm": 1.0984460195119274,
      "learning_rate": 3.544627014076707e-06,
      "loss": 0.5781,
      "step": 11917
    },
    {
      "epoch": 0.6431385246343965,
      "grad_norm": 1.2499596474846406,
      "learning_rate": 3.543939936159697e-06,
      "loss": 0.6563,
      "step": 11918
    },
    {
      "epoch": 0.6431924882629108,
      "grad_norm": 1.1322220896063537,
      "learning_rate": 3.5432529144618287e-06,
      "loss": 0.3856,
      "step": 11919
    },
    {
      "epoch": 0.6432464518914252,
      "grad_norm": 1.217226763452663,
      "learning_rate": 3.5425659490028484e-06,
      "loss": 0.5335,
      "step": 11920
    },
    {
      "epoch": 0.6433004155199396,
      "grad_norm": 1.1192795845824188,
      "learning_rate": 3.5418790398025017e-06,
      "loss": 0.4824,
      "step": 11921
    },
    {
      "epoch": 0.643354379148454,
      "grad_norm": 0.9877301624910121,
      "learning_rate": 3.54119218688053e-06,
      "loss": 0.4844,
      "step": 11922
    },
    {
      "epoch": 0.6434083427769683,
      "grad_norm": 1.1772423062509456,
      "learning_rate": 3.540505390256673e-06,
      "loss": 0.7378,
      "step": 11923
    },
    {
      "epoch": 0.6434623064054827,
      "grad_norm": 1.140672273415446,
      "learning_rate": 3.5398186499506727e-06,
      "loss": 0.5456,
      "step": 11924
    },
    {
      "epoch": 0.6435162700339971,
      "grad_norm": 1.1576295079430625,
      "learning_rate": 3.5391319659822655e-06,
      "loss": 0.5606,
      "step": 11925
    },
    {
      "epoch": 0.6435702336625114,
      "grad_norm": 1.1158734007624782,
      "learning_rate": 3.538445338371185e-06,
      "loss": 0.441,
      "step": 11926
    },
    {
      "epoch": 0.6436241972910258,
      "grad_norm": 1.0895387961647696,
      "learning_rate": 3.5377587671371684e-06,
      "loss": 0.4575,
      "step": 11927
    },
    {
      "epoch": 0.6436781609195402,
      "grad_norm": 0.9028356284805057,
      "learning_rate": 3.5370722522999468e-06,
      "loss": 0.3728,
      "step": 11928
    },
    {
      "epoch": 0.6437321245480546,
      "grad_norm": 1.194750387660433,
      "learning_rate": 3.536385793879252e-06,
      "loss": 0.5629,
      "step": 11929
    },
    {
      "epoch": 0.643786088176569,
      "grad_norm": 1.0048290462940337,
      "learning_rate": 3.5356993918948144e-06,
      "loss": 0.4298,
      "step": 11930
    },
    {
      "epoch": 0.6438400518050834,
      "grad_norm": 0.9120050962829911,
      "learning_rate": 3.53501304636636e-06,
      "loss": 0.3681,
      "step": 11931
    },
    {
      "epoch": 0.6438940154335978,
      "grad_norm": 1.0980971405553193,
      "learning_rate": 3.5343267573136162e-06,
      "loss": 0.4277,
      "step": 11932
    },
    {
      "epoch": 0.6439479790621121,
      "grad_norm": 0.9674674229318213,
      "learning_rate": 3.533640524756311e-06,
      "loss": 0.358,
      "step": 11933
    },
    {
      "epoch": 0.6440019426906265,
      "grad_norm": 1.0497592456279705,
      "learning_rate": 3.5329543487141605e-06,
      "loss": 0.4791,
      "step": 11934
    },
    {
      "epoch": 0.6440559063191409,
      "grad_norm": 1.0397268461070002,
      "learning_rate": 3.53226822920689e-06,
      "loss": 0.4646,
      "step": 11935
    },
    {
      "epoch": 0.6441098699476553,
      "grad_norm": 1.013968402626697,
      "learning_rate": 3.5315821662542184e-06,
      "loss": 0.4242,
      "step": 11936
    },
    {
      "epoch": 0.6441638335761697,
      "grad_norm": 0.686591279484563,
      "learning_rate": 3.530896159875865e-06,
      "loss": 0.2521,
      "step": 11937
    },
    {
      "epoch": 0.6442177972046841,
      "grad_norm": 1.3280297322686236,
      "learning_rate": 3.5302102100915457e-06,
      "loss": 0.5804,
      "step": 11938
    },
    {
      "epoch": 0.6442717608331984,
      "grad_norm": 0.9911727796565426,
      "learning_rate": 3.529524316920977e-06,
      "loss": 0.4233,
      "step": 11939
    },
    {
      "epoch": 0.6443257244617128,
      "grad_norm": 1.0949596262836387,
      "learning_rate": 3.5288384803838684e-06,
      "loss": 0.4113,
      "step": 11940
    },
    {
      "epoch": 0.6443796880902272,
      "grad_norm": 0.954651637684788,
      "learning_rate": 3.5281527004999327e-06,
      "loss": 0.5168,
      "step": 11941
    },
    {
      "epoch": 0.6444336517187416,
      "grad_norm": 0.758775086446862,
      "learning_rate": 3.5274669772888823e-06,
      "loss": 0.3177,
      "step": 11942
    },
    {
      "epoch": 0.644487615347256,
      "grad_norm": 1.0077864652770185,
      "learning_rate": 3.526781310770423e-06,
      "loss": 0.5296,
      "step": 11943
    },
    {
      "epoch": 0.6445415789757704,
      "grad_norm": 0.8561574002161556,
      "learning_rate": 3.5260957009642637e-06,
      "loss": 0.3607,
      "step": 11944
    },
    {
      "epoch": 0.6445955426042848,
      "grad_norm": 1.0364908697485768,
      "learning_rate": 3.525410147890107e-06,
      "loss": 0.4743,
      "step": 11945
    },
    {
      "epoch": 0.644649506232799,
      "grad_norm": 1.0588290200723498,
      "learning_rate": 3.5247246515676604e-06,
      "loss": 0.3644,
      "step": 11946
    },
    {
      "epoch": 0.6447034698613134,
      "grad_norm": 0.9094091043401185,
      "learning_rate": 3.5240392120166213e-06,
      "loss": 0.3983,
      "step": 11947
    },
    {
      "epoch": 0.6447574334898278,
      "grad_norm": 1.0012866905639008,
      "learning_rate": 3.5233538292566926e-06,
      "loss": 0.446,
      "step": 11948
    },
    {
      "epoch": 0.6448113971183422,
      "grad_norm": 0.9925163396189625,
      "learning_rate": 3.522668503307571e-06,
      "loss": 0.5024,
      "step": 11949
    },
    {
      "epoch": 0.6448653607468566,
      "grad_norm": 1.0756836157076615,
      "learning_rate": 3.5219832341889547e-06,
      "loss": 0.5328,
      "step": 11950
    },
    {
      "epoch": 0.644919324375371,
      "grad_norm": 0.9860805759431336,
      "learning_rate": 3.5212980219205393e-06,
      "loss": 0.4539,
      "step": 11951
    },
    {
      "epoch": 0.6449732880038854,
      "grad_norm": 1.0339469568083965,
      "learning_rate": 3.5206128665220185e-06,
      "loss": 0.4384,
      "step": 11952
    },
    {
      "epoch": 0.6450272516323997,
      "grad_norm": 0.9929974677253116,
      "learning_rate": 3.519927768013085e-06,
      "loss": 0.4799,
      "step": 11953
    },
    {
      "epoch": 0.6450812152609141,
      "grad_norm": 1.1859969354093964,
      "learning_rate": 3.5192427264134265e-06,
      "loss": 0.6296,
      "step": 11954
    },
    {
      "epoch": 0.6451351788894285,
      "grad_norm": 1.072597003008902,
      "learning_rate": 3.518557741742734e-06,
      "loss": 0.5244,
      "step": 11955
    },
    {
      "epoch": 0.6451891425179429,
      "grad_norm": 0.7052607853891402,
      "learning_rate": 3.517872814020693e-06,
      "loss": 0.2658,
      "step": 11956
    },
    {
      "epoch": 0.6452431061464573,
      "grad_norm": 1.1657041904256165,
      "learning_rate": 3.5171879432669907e-06,
      "loss": 0.4379,
      "step": 11957
    },
    {
      "epoch": 0.6452970697749717,
      "grad_norm": 1.130605479739754,
      "learning_rate": 3.516503129501311e-06,
      "loss": 0.4864,
      "step": 11958
    },
    {
      "epoch": 0.6453510334034861,
      "grad_norm": 0.980993114354136,
      "learning_rate": 3.515818372743335e-06,
      "loss": 0.4699,
      "step": 11959
    },
    {
      "epoch": 0.6454049970320004,
      "grad_norm": 0.8947894708812496,
      "learning_rate": 3.5151336730127462e-06,
      "loss": 0.4314,
      "step": 11960
    },
    {
      "epoch": 0.6454589606605148,
      "grad_norm": 0.8610539387097311,
      "learning_rate": 3.514449030329218e-06,
      "loss": 0.4741,
      "step": 11961
    },
    {
      "epoch": 0.6455129242890292,
      "grad_norm": 1.1871442600655469,
      "learning_rate": 3.513764444712432e-06,
      "loss": 0.5105,
      "step": 11962
    },
    {
      "epoch": 0.6455668879175436,
      "grad_norm": 1.0590143693726168,
      "learning_rate": 3.513079916182063e-06,
      "loss": 0.4842,
      "step": 11963
    },
    {
      "epoch": 0.645620851546058,
      "grad_norm": 0.9741806404140553,
      "learning_rate": 3.5123954447577836e-06,
      "loss": 0.607,
      "step": 11964
    },
    {
      "epoch": 0.6456748151745724,
      "grad_norm": 1.374233784852015,
      "learning_rate": 3.5117110304592682e-06,
      "loss": 0.5261,
      "step": 11965
    },
    {
      "epoch": 0.6457287788030868,
      "grad_norm": 1.1013421291945056,
      "learning_rate": 3.511026673306187e-06,
      "loss": 0.4686,
      "step": 11966
    },
    {
      "epoch": 0.645782742431601,
      "grad_norm": 1.0729944528064397,
      "learning_rate": 3.5103423733182086e-06,
      "loss": 0.4908,
      "step": 11967
    },
    {
      "epoch": 0.6458367060601155,
      "grad_norm": 0.8531198874159256,
      "learning_rate": 3.5096581305150036e-06,
      "loss": 0.4415,
      "step": 11968
    },
    {
      "epoch": 0.6458906696886298,
      "grad_norm": 0.9055682881444185,
      "learning_rate": 3.508973944916232e-06,
      "loss": 0.4578,
      "step": 11969
    },
    {
      "epoch": 0.6459446333171442,
      "grad_norm": 1.0994233941725013,
      "learning_rate": 3.5082898165415614e-06,
      "loss": 0.5986,
      "step": 11970
    },
    {
      "epoch": 0.6459985969456586,
      "grad_norm": 1.0705638019984045,
      "learning_rate": 3.507605745410654e-06,
      "loss": 0.4163,
      "step": 11971
    },
    {
      "epoch": 0.646052560574173,
      "grad_norm": 1.2316473614039445,
      "learning_rate": 3.5069217315431705e-06,
      "loss": 0.7238,
      "step": 11972
    },
    {
      "epoch": 0.6461065242026874,
      "grad_norm": 0.9890461397716981,
      "learning_rate": 3.5062377749587706e-06,
      "loss": 0.4961,
      "step": 11973
    },
    {
      "epoch": 0.6461604878312017,
      "grad_norm": 1.2127660521699346,
      "learning_rate": 3.5055538756771114e-06,
      "loss": 0.3634,
      "step": 11974
    },
    {
      "epoch": 0.6462144514597161,
      "grad_norm": 1.2597035317663392,
      "learning_rate": 3.5048700337178487e-06,
      "loss": 0.6444,
      "step": 11975
    },
    {
      "epoch": 0.6462684150882305,
      "grad_norm": 1.026868310020619,
      "learning_rate": 3.5041862491006375e-06,
      "loss": 0.4596,
      "step": 11976
    },
    {
      "epoch": 0.6463223787167449,
      "grad_norm": 1.0539895145314626,
      "learning_rate": 3.5035025218451302e-06,
      "loss": 0.4728,
      "step": 11977
    },
    {
      "epoch": 0.6463763423452593,
      "grad_norm": 0.993709324151771,
      "learning_rate": 3.502818851970977e-06,
      "loss": 0.5062,
      "step": 11978
    },
    {
      "epoch": 0.6464303059737737,
      "grad_norm": 0.8855176811870283,
      "learning_rate": 3.502135239497829e-06,
      "loss": 0.3457,
      "step": 11979
    },
    {
      "epoch": 0.6464842696022881,
      "grad_norm": 0.7861643009498145,
      "learning_rate": 3.501451684445333e-06,
      "loss": 0.2857,
      "step": 11980
    },
    {
      "epoch": 0.6465382332308024,
      "grad_norm": 0.8551614489047985,
      "learning_rate": 3.5007681868331357e-06,
      "loss": 0.364,
      "step": 11981
    },
    {
      "epoch": 0.6465921968593168,
      "grad_norm": 0.983725984186243,
      "learning_rate": 3.5000847466808795e-06,
      "loss": 0.4826,
      "step": 11982
    },
    {
      "epoch": 0.6466461604878312,
      "grad_norm": 1.0616287305277774,
      "learning_rate": 3.499401364008209e-06,
      "loss": 0.5545,
      "step": 11983
    },
    {
      "epoch": 0.6467001241163456,
      "grad_norm": 1.0858162030902956,
      "learning_rate": 3.498718038834764e-06,
      "loss": 0.535,
      "step": 11984
    },
    {
      "epoch": 0.64675408774486,
      "grad_norm": 1.0123656383763082,
      "learning_rate": 3.4980347711801853e-06,
      "loss": 0.4825,
      "step": 11985
    },
    {
      "epoch": 0.6468080513733744,
      "grad_norm": 1.082969447071066,
      "learning_rate": 3.4973515610641095e-06,
      "loss": 0.5023,
      "step": 11986
    },
    {
      "epoch": 0.6468620150018888,
      "grad_norm": 0.9688310911420915,
      "learning_rate": 3.4966684085061743e-06,
      "loss": 0.4552,
      "step": 11987
    },
    {
      "epoch": 0.6469159786304031,
      "grad_norm": 0.9405301421491425,
      "learning_rate": 3.4959853135260147e-06,
      "loss": 0.4323,
      "step": 11988
    },
    {
      "epoch": 0.6469699422589175,
      "grad_norm": 1.008925601390938,
      "learning_rate": 3.49530227614326e-06,
      "loss": 0.4316,
      "step": 11989
    },
    {
      "epoch": 0.6470239058874319,
      "grad_norm": 0.9466784918981792,
      "learning_rate": 3.4946192963775426e-06,
      "loss": 0.5245,
      "step": 11990
    },
    {
      "epoch": 0.6470778695159463,
      "grad_norm": 1.0912654405501792,
      "learning_rate": 3.493936374248494e-06,
      "loss": 0.4359,
      "step": 11991
    },
    {
      "epoch": 0.6471318331444607,
      "grad_norm": 0.8532623630940516,
      "learning_rate": 3.4932535097757415e-06,
      "loss": 0.364,
      "step": 11992
    },
    {
      "epoch": 0.647185796772975,
      "grad_norm": 1.223304457568847,
      "learning_rate": 3.49257070297891e-06,
      "loss": 0.3755,
      "step": 11993
    },
    {
      "epoch": 0.6472397604014895,
      "grad_norm": 0.9421915313611954,
      "learning_rate": 3.491887953877625e-06,
      "loss": 0.4689,
      "step": 11994
    },
    {
      "epoch": 0.6472937240300037,
      "grad_norm": 1.1605546645582148,
      "learning_rate": 3.4912052624915103e-06,
      "loss": 0.5835,
      "step": 11995
    },
    {
      "epoch": 0.6473476876585181,
      "grad_norm": 1.1380754956260215,
      "learning_rate": 3.490522628840185e-06,
      "loss": 0.4987,
      "step": 11996
    },
    {
      "epoch": 0.6474016512870325,
      "grad_norm": 1.0582605546154766,
      "learning_rate": 3.4898400529432703e-06,
      "loss": 0.4886,
      "step": 11997
    },
    {
      "epoch": 0.6474556149155469,
      "grad_norm": 1.2750742778075972,
      "learning_rate": 3.489157534820383e-06,
      "loss": 0.5606,
      "step": 11998
    },
    {
      "epoch": 0.6475095785440613,
      "grad_norm": 0.7461107255434049,
      "learning_rate": 3.48847507449114e-06,
      "loss": 0.3248,
      "step": 11999
    },
    {
      "epoch": 0.6475635421725757,
      "grad_norm": 1.1380250153657816,
      "learning_rate": 3.4877926719751565e-06,
      "loss": 0.4932,
      "step": 12000
    },
    {
      "epoch": 0.6475635421725757,
      "eval_loss": 0.5388334393501282,
      "eval_runtime": 161.06,
      "eval_samples_per_second": 21.352,
      "eval_steps_per_second": 0.894,
      "step": 12000
    },
    {
      "epoch": 0.64761750580109,
      "grad_norm": 1.0115408255471097,
      "learning_rate": 3.4871103272920454e-06,
      "loss": 0.5404,
      "step": 12001
    },
    {
      "epoch": 0.6476714694296044,
      "grad_norm": 0.9856560116097753,
      "learning_rate": 3.486428040461416e-06,
      "loss": 0.6018,
      "step": 12002
    },
    {
      "epoch": 0.6477254330581188,
      "grad_norm": 1.2572996292485785,
      "learning_rate": 3.48574581150288e-06,
      "loss": 0.6961,
      "step": 12003
    },
    {
      "epoch": 0.6477793966866332,
      "grad_norm": 1.0654168950996765,
      "learning_rate": 3.485063640436044e-06,
      "loss": 0.4839,
      "step": 12004
    },
    {
      "epoch": 0.6478333603151476,
      "grad_norm": 0.8445614781514537,
      "learning_rate": 3.484381527280515e-06,
      "loss": 0.3003,
      "step": 12005
    },
    {
      "epoch": 0.647887323943662,
      "grad_norm": 1.107279703029957,
      "learning_rate": 3.483699472055897e-06,
      "loss": 0.4626,
      "step": 12006
    },
    {
      "epoch": 0.6479412875721764,
      "grad_norm": 0.7782653290260381,
      "learning_rate": 3.4830174747817945e-06,
      "loss": 0.3494,
      "step": 12007
    },
    {
      "epoch": 0.6479952512006907,
      "grad_norm": 0.8418006945509531,
      "learning_rate": 3.4823355354778095e-06,
      "loss": 0.3649,
      "step": 12008
    },
    {
      "epoch": 0.6480492148292051,
      "grad_norm": 1.0051765011949316,
      "learning_rate": 3.481653654163537e-06,
      "loss": 0.4649,
      "step": 12009
    },
    {
      "epoch": 0.6481031784577195,
      "grad_norm": 1.0418608877793682,
      "learning_rate": 3.480971830858579e-06,
      "loss": 0.5339,
      "step": 12010
    },
    {
      "epoch": 0.6481571420862339,
      "grad_norm": 0.9858402214220267,
      "learning_rate": 3.4802900655825306e-06,
      "loss": 0.395,
      "step": 12011
    },
    {
      "epoch": 0.6482111057147483,
      "grad_norm": 0.8972181112207309,
      "learning_rate": 3.4796083583549857e-06,
      "loss": 0.4511,
      "step": 12012
    },
    {
      "epoch": 0.6482650693432627,
      "grad_norm": 0.9486437575474898,
      "learning_rate": 3.4789267091955393e-06,
      "loss": 0.3876,
      "step": 12013
    },
    {
      "epoch": 0.6483190329717771,
      "grad_norm": 1.3014709305218788,
      "learning_rate": 3.47824511812378e-06,
      "loss": 0.6734,
      "step": 12014
    },
    {
      "epoch": 0.6483729966002914,
      "grad_norm": 1.19203353433188,
      "learning_rate": 3.4775635851593014e-06,
      "loss": 0.4587,
      "step": 12015
    },
    {
      "epoch": 0.6484269602288057,
      "grad_norm": 1.0150378010398968,
      "learning_rate": 3.4768821103216877e-06,
      "loss": 0.505,
      "step": 12016
    },
    {
      "epoch": 0.6484809238573201,
      "grad_norm": 1.0467699279193174,
      "learning_rate": 3.4762006936305247e-06,
      "loss": 0.3933,
      "step": 12017
    },
    {
      "epoch": 0.6485348874858345,
      "grad_norm": 1.1245701343259038,
      "learning_rate": 3.4755193351053997e-06,
      "loss": 0.559,
      "step": 12018
    },
    {
      "epoch": 0.6485888511143489,
      "grad_norm": 1.2285838020178694,
      "learning_rate": 3.474838034765895e-06,
      "loss": 0.5629,
      "step": 12019
    },
    {
      "epoch": 0.6486428147428633,
      "grad_norm": 1.0260145817164394,
      "learning_rate": 3.47415679263159e-06,
      "loss": 0.4311,
      "step": 12020
    },
    {
      "epoch": 0.6486967783713777,
      "grad_norm": 1.0992158387974078,
      "learning_rate": 3.473475608722067e-06,
      "loss": 0.4847,
      "step": 12021
    },
    {
      "epoch": 0.648750741999892,
      "grad_norm": 0.7984862352711917,
      "learning_rate": 3.472794483056903e-06,
      "loss": 0.3833,
      "step": 12022
    },
    {
      "epoch": 0.6488047056284064,
      "grad_norm": 0.9009574698005273,
      "learning_rate": 3.4721134156556728e-06,
      "loss": 0.374,
      "step": 12023
    },
    {
      "epoch": 0.6488586692569208,
      "grad_norm": 0.9332665180892579,
      "learning_rate": 3.4714324065379523e-06,
      "loss": 0.591,
      "step": 12024
    },
    {
      "epoch": 0.6489126328854352,
      "grad_norm": 1.0429458478307991,
      "learning_rate": 3.4707514557233145e-06,
      "loss": 0.453,
      "step": 12025
    },
    {
      "epoch": 0.6489665965139496,
      "grad_norm": 0.8104326481644326,
      "learning_rate": 3.47007056323133e-06,
      "loss": 0.4274,
      "step": 12026
    },
    {
      "epoch": 0.649020560142464,
      "grad_norm": 1.4430002574786829,
      "learning_rate": 3.4693897290815694e-06,
      "loss": 0.6101,
      "step": 12027
    },
    {
      "epoch": 0.6490745237709784,
      "grad_norm": 1.1476886511454112,
      "learning_rate": 3.468708953293599e-06,
      "loss": 0.5493,
      "step": 12028
    },
    {
      "epoch": 0.6491284873994927,
      "grad_norm": 1.1376286674194502,
      "learning_rate": 3.468028235886987e-06,
      "loss": 0.5055,
      "step": 12029
    },
    {
      "epoch": 0.6491824510280071,
      "grad_norm": 1.1045598809183939,
      "learning_rate": 3.467347576881296e-06,
      "loss": 0.4155,
      "step": 12030
    },
    {
      "epoch": 0.6492364146565215,
      "grad_norm": 0.9987958477232202,
      "learning_rate": 3.4666669762960887e-06,
      "loss": 0.4178,
      "step": 12031
    },
    {
      "epoch": 0.6492903782850359,
      "grad_norm": 0.9558059388619287,
      "learning_rate": 3.4659864341509274e-06,
      "loss": 0.4288,
      "step": 12032
    },
    {
      "epoch": 0.6493443419135503,
      "grad_norm": 1.1274401728056826,
      "learning_rate": 3.465305950465371e-06,
      "loss": 0.4567,
      "step": 12033
    },
    {
      "epoch": 0.6493983055420647,
      "grad_norm": 0.9761423869164767,
      "learning_rate": 3.4646255252589778e-06,
      "loss": 0.4831,
      "step": 12034
    },
    {
      "epoch": 0.6494522691705791,
      "grad_norm": 0.8673761080874788,
      "learning_rate": 3.463945158551303e-06,
      "loss": 0.4881,
      "step": 12035
    },
    {
      "epoch": 0.6495062327990934,
      "grad_norm": 0.907881271846122,
      "learning_rate": 3.4632648503619043e-06,
      "loss": 0.3465,
      "step": 12036
    },
    {
      "epoch": 0.6495601964276078,
      "grad_norm": 0.9106197536196463,
      "learning_rate": 3.462584600710329e-06,
      "loss": 0.3285,
      "step": 12037
    },
    {
      "epoch": 0.6496141600561222,
      "grad_norm": 0.9239975593297185,
      "learning_rate": 3.461904409616132e-06,
      "loss": 0.5045,
      "step": 12038
    },
    {
      "epoch": 0.6496681236846366,
      "grad_norm": 1.2736656382242142,
      "learning_rate": 3.4612242770988613e-06,
      "loss": 0.6152,
      "step": 12039
    },
    {
      "epoch": 0.649722087313151,
      "grad_norm": 0.868424591061882,
      "learning_rate": 3.4605442031780646e-06,
      "loss": 0.4453,
      "step": 12040
    },
    {
      "epoch": 0.6497760509416654,
      "grad_norm": 0.9160307972520578,
      "learning_rate": 3.4598641878732888e-06,
      "loss": 0.415,
      "step": 12041
    },
    {
      "epoch": 0.6498300145701797,
      "grad_norm": 0.9909482378209951,
      "learning_rate": 3.459184231204077e-06,
      "loss": 0.3878,
      "step": 12042
    },
    {
      "epoch": 0.649883978198694,
      "grad_norm": 0.8570057976277617,
      "learning_rate": 3.458504333189974e-06,
      "loss": 0.4288,
      "step": 12043
    },
    {
      "epoch": 0.6499379418272084,
      "grad_norm": 1.0381066731193054,
      "learning_rate": 3.457824493850518e-06,
      "loss": 0.4066,
      "step": 12044
    },
    {
      "epoch": 0.6499919054557228,
      "grad_norm": 1.061688414586502,
      "learning_rate": 3.4571447132052494e-06,
      "loss": 0.4623,
      "step": 12045
    },
    {
      "epoch": 0.6500458690842372,
      "grad_norm": 0.925140010291013,
      "learning_rate": 3.456464991273705e-06,
      "loss": 0.4449,
      "step": 12046
    },
    {
      "epoch": 0.6500998327127516,
      "grad_norm": 0.9381655655012658,
      "learning_rate": 3.4557853280754234e-06,
      "loss": 0.4348,
      "step": 12047
    },
    {
      "epoch": 0.650153796341266,
      "grad_norm": 0.9690739719938267,
      "learning_rate": 3.455105723629936e-06,
      "loss": 0.4933,
      "step": 12048
    },
    {
      "epoch": 0.6502077599697804,
      "grad_norm": 1.1263254684311492,
      "learning_rate": 3.4544261779567765e-06,
      "loss": 0.4487,
      "step": 12049
    },
    {
      "epoch": 0.6502617235982947,
      "grad_norm": 1.0967202822851743,
      "learning_rate": 3.4537466910754767e-06,
      "loss": 0.589,
      "step": 12050
    },
    {
      "epoch": 0.6503156872268091,
      "grad_norm": 1.1323356630119494,
      "learning_rate": 3.453067263005564e-06,
      "loss": 0.5127,
      "step": 12051
    },
    {
      "epoch": 0.6503696508553235,
      "grad_norm": 1.0420852224694899,
      "learning_rate": 3.4523878937665654e-06,
      "loss": 0.3772,
      "step": 12052
    },
    {
      "epoch": 0.6504236144838379,
      "grad_norm": 1.114835595117865,
      "learning_rate": 3.4517085833780086e-06,
      "loss": 0.4884,
      "step": 12053
    },
    {
      "epoch": 0.6504775781123523,
      "grad_norm": 0.9306999061265518,
      "learning_rate": 3.4510293318594156e-06,
      "loss": 0.3552,
      "step": 12054
    },
    {
      "epoch": 0.6505315417408667,
      "grad_norm": 1.0722482279023124,
      "learning_rate": 3.450350139230311e-06,
      "loss": 0.5598,
      "step": 12055
    },
    {
      "epoch": 0.6505855053693811,
      "grad_norm": 1.139652697175992,
      "learning_rate": 3.449671005510213e-06,
      "loss": 0.5758,
      "step": 12056
    },
    {
      "epoch": 0.6506394689978954,
      "grad_norm": 0.9341892524726666,
      "learning_rate": 3.4489919307186457e-06,
      "loss": 0.4022,
      "step": 12057
    },
    {
      "epoch": 0.6506934326264098,
      "grad_norm": 1.1893047058579265,
      "learning_rate": 3.44831291487512e-06,
      "loss": 0.5705,
      "step": 12058
    },
    {
      "epoch": 0.6507473962549242,
      "grad_norm": 0.8711572846257112,
      "learning_rate": 3.4476339579991534e-06,
      "loss": 0.4066,
      "step": 12059
    },
    {
      "epoch": 0.6508013598834386,
      "grad_norm": 0.7171390747338711,
      "learning_rate": 3.446955060110262e-06,
      "loss": 0.2923,
      "step": 12060
    },
    {
      "epoch": 0.650855323511953,
      "grad_norm": 0.9185297130487637,
      "learning_rate": 3.446276221227956e-06,
      "loss": 0.4733,
      "step": 12061
    },
    {
      "epoch": 0.6509092871404674,
      "grad_norm": 0.9299591708073434,
      "learning_rate": 3.4455974413717463e-06,
      "loss": 0.4266,
      "step": 12062
    },
    {
      "epoch": 0.6509632507689818,
      "grad_norm": 1.070053569526316,
      "learning_rate": 3.4449187205611424e-06,
      "loss": 0.4518,
      "step": 12063
    },
    {
      "epoch": 0.651017214397496,
      "grad_norm": 1.016086445760248,
      "learning_rate": 3.4442400588156527e-06,
      "loss": 0.5487,
      "step": 12064
    },
    {
      "epoch": 0.6510711780260104,
      "grad_norm": 1.5090633676511294,
      "learning_rate": 3.443561456154779e-06,
      "loss": 0.5515,
      "step": 12065
    },
    {
      "epoch": 0.6511251416545248,
      "grad_norm": 0.991835774058149,
      "learning_rate": 3.442882912598027e-06,
      "loss": 0.4629,
      "step": 12066
    },
    {
      "epoch": 0.6511791052830392,
      "grad_norm": 1.0720584303687162,
      "learning_rate": 3.442204428164898e-06,
      "loss": 0.5837,
      "step": 12067
    },
    {
      "epoch": 0.6512330689115536,
      "grad_norm": 0.9888823288419507,
      "learning_rate": 3.441526002874894e-06,
      "loss": 0.4492,
      "step": 12068
    },
    {
      "epoch": 0.651287032540068,
      "grad_norm": 0.9160279570221956,
      "learning_rate": 3.440847636747512e-06,
      "loss": 0.3868,
      "step": 12069
    },
    {
      "epoch": 0.6513409961685823,
      "grad_norm": 0.8870083060447506,
      "learning_rate": 3.44016932980225e-06,
      "loss": 0.4323,
      "step": 12070
    },
    {
      "epoch": 0.6513949597970967,
      "grad_norm": 0.7644712840650895,
      "learning_rate": 3.4394910820586023e-06,
      "loss": 0.3361,
      "step": 12071
    },
    {
      "epoch": 0.6514489234256111,
      "grad_norm": 1.0460522085101054,
      "learning_rate": 3.438812893536063e-06,
      "loss": 0.4521,
      "step": 12072
    },
    {
      "epoch": 0.6515028870541255,
      "grad_norm": 0.9723179160435146,
      "learning_rate": 3.438134764254124e-06,
      "loss": 0.4652,
      "step": 12073
    },
    {
      "epoch": 0.6515568506826399,
      "grad_norm": 1.034454536650743,
      "learning_rate": 3.4374566942322767e-06,
      "loss": 0.537,
      "step": 12074
    },
    {
      "epoch": 0.6516108143111543,
      "grad_norm": 1.1125324832046957,
      "learning_rate": 3.436778683490007e-06,
      "loss": 0.4935,
      "step": 12075
    },
    {
      "epoch": 0.6516647779396687,
      "grad_norm": 0.9541508221699821,
      "learning_rate": 3.436100732046802e-06,
      "loss": 0.3821,
      "step": 12076
    },
    {
      "epoch": 0.651718741568183,
      "grad_norm": 1.073856585479336,
      "learning_rate": 3.435422839922148e-06,
      "loss": 0.4752,
      "step": 12077
    },
    {
      "epoch": 0.6517727051966974,
      "grad_norm": 1.0164899017747888,
      "learning_rate": 3.434745007135527e-06,
      "loss": 0.6277,
      "step": 12078
    },
    {
      "epoch": 0.6518266688252118,
      "grad_norm": 0.8463394398607668,
      "learning_rate": 3.434067233706422e-06,
      "loss": 0.3472,
      "step": 12079
    },
    {
      "epoch": 0.6518806324537262,
      "grad_norm": 1.068269652892902,
      "learning_rate": 3.4333895196543116e-06,
      "loss": 0.5364,
      "step": 12080
    },
    {
      "epoch": 0.6519345960822406,
      "grad_norm": 1.0198114206388127,
      "learning_rate": 3.4327118649986747e-06,
      "loss": 0.4892,
      "step": 12081
    },
    {
      "epoch": 0.651988559710755,
      "grad_norm": 1.2469448596883117,
      "learning_rate": 3.432034269758987e-06,
      "loss": 0.4737,
      "step": 12082
    },
    {
      "epoch": 0.6520425233392694,
      "grad_norm": 1.0515914250814684,
      "learning_rate": 3.431356733954725e-06,
      "loss": 0.5124,
      "step": 12083
    },
    {
      "epoch": 0.6520964869677837,
      "grad_norm": 0.9210027819503481,
      "learning_rate": 3.4306792576053614e-06,
      "loss": 0.3991,
      "step": 12084
    },
    {
      "epoch": 0.6521504505962981,
      "grad_norm": 0.9785544251808219,
      "learning_rate": 3.430001840730366e-06,
      "loss": 0.5077,
      "step": 12085
    },
    {
      "epoch": 0.6522044142248125,
      "grad_norm": 1.0972261552613138,
      "learning_rate": 3.429324483349209e-06,
      "loss": 0.5601,
      "step": 12086
    },
    {
      "epoch": 0.6522583778533269,
      "grad_norm": 0.8870965489347724,
      "learning_rate": 3.4286471854813587e-06,
      "loss": 0.4261,
      "step": 12087
    },
    {
      "epoch": 0.6523123414818413,
      "grad_norm": 1.1421663772785913,
      "learning_rate": 3.427969947146281e-06,
      "loss": 0.5266,
      "step": 12088
    },
    {
      "epoch": 0.6523663051103556,
      "grad_norm": 0.8273664677084684,
      "learning_rate": 3.4272927683634405e-06,
      "loss": 0.3535,
      "step": 12089
    },
    {
      "epoch": 0.65242026873887,
      "grad_norm": 1.034966027311295,
      "learning_rate": 3.4266156491523006e-06,
      "loss": 0.4722,
      "step": 12090
    },
    {
      "epoch": 0.6524742323673843,
      "grad_norm": 1.054308112110283,
      "learning_rate": 3.425938589532323e-06,
      "loss": 0.5454,
      "step": 12091
    },
    {
      "epoch": 0.6525281959958987,
      "grad_norm": 1.1345221444458555,
      "learning_rate": 3.4252615895229644e-06,
      "loss": 0.5197,
      "step": 12092
    },
    {
      "epoch": 0.6525821596244131,
      "grad_norm": 0.9995206426003981,
      "learning_rate": 3.424584649143685e-06,
      "loss": 0.2966,
      "step": 12093
    },
    {
      "epoch": 0.6526361232529275,
      "grad_norm": 1.0702710767274346,
      "learning_rate": 3.423907768413939e-06,
      "loss": 0.527,
      "step": 12094
    },
    {
      "epoch": 0.6526900868814419,
      "grad_norm": 1.14933448866209,
      "learning_rate": 3.423230947353182e-06,
      "loss": 0.5051,
      "step": 12095
    },
    {
      "epoch": 0.6527440505099563,
      "grad_norm": 1.6334132017073408,
      "learning_rate": 3.422554185980865e-06,
      "loss": 0.5419,
      "step": 12096
    },
    {
      "epoch": 0.6527980141384707,
      "grad_norm": 0.8394070519835319,
      "learning_rate": 3.4218774843164416e-06,
      "loss": 0.4215,
      "step": 12097
    },
    {
      "epoch": 0.652851977766985,
      "grad_norm": 1.135597535306332,
      "learning_rate": 3.4212008423793596e-06,
      "loss": 0.4697,
      "step": 12098
    },
    {
      "epoch": 0.6529059413954994,
      "grad_norm": 1.1096776193644327,
      "learning_rate": 3.420524260189064e-06,
      "loss": 0.4905,
      "step": 12099
    },
    {
      "epoch": 0.6529599050240138,
      "grad_norm": 0.9544527290955739,
      "learning_rate": 3.419847737765003e-06,
      "loss": 0.4403,
      "step": 12100
    },
    {
      "epoch": 0.6530138686525282,
      "grad_norm": 1.110385338918088,
      "learning_rate": 3.4191712751266195e-06,
      "loss": 0.4987,
      "step": 12101
    },
    {
      "epoch": 0.6530678322810426,
      "grad_norm": 0.9694162372826827,
      "learning_rate": 3.418494872293356e-06,
      "loss": 0.408,
      "step": 12102
    },
    {
      "epoch": 0.653121795909557,
      "grad_norm": 0.9754280321498997,
      "learning_rate": 3.4178185292846534e-06,
      "loss": 0.3983,
      "step": 12103
    },
    {
      "epoch": 0.6531757595380714,
      "grad_norm": 1.306879773451148,
      "learning_rate": 3.417142246119949e-06,
      "loss": 0.5704,
      "step": 12104
    },
    {
      "epoch": 0.6532297231665857,
      "grad_norm": 0.9515799919138271,
      "learning_rate": 3.416466022818684e-06,
      "loss": 0.4152,
      "step": 12105
    },
    {
      "epoch": 0.6532836867951001,
      "grad_norm": 1.1135366818530072,
      "learning_rate": 3.4157898594002884e-06,
      "loss": 0.5425,
      "step": 12106
    },
    {
      "epoch": 0.6533376504236145,
      "grad_norm": 1.0089123108444913,
      "learning_rate": 3.4151137558841984e-06,
      "loss": 0.4075,
      "step": 12107
    },
    {
      "epoch": 0.6533916140521289,
      "grad_norm": 1.0120473532717573,
      "learning_rate": 3.4144377122898465e-06,
      "loss": 0.4212,
      "step": 12108
    },
    {
      "epoch": 0.6534455776806433,
      "grad_norm": 0.952685159746953,
      "learning_rate": 3.4137617286366608e-06,
      "loss": 0.3671,
      "step": 12109
    },
    {
      "epoch": 0.6534995413091577,
      "grad_norm": 0.9980305494096618,
      "learning_rate": 3.4130858049440717e-06,
      "loss": 0.3239,
      "step": 12110
    },
    {
      "epoch": 0.6535535049376721,
      "grad_norm": 1.008994558582359,
      "learning_rate": 3.412409941231505e-06,
      "loss": 0.4312,
      "step": 12111
    },
    {
      "epoch": 0.6536074685661863,
      "grad_norm": 1.2048666213483523,
      "learning_rate": 3.4117341375183875e-06,
      "loss": 0.6028,
      "step": 12112
    },
    {
      "epoch": 0.6536614321947007,
      "grad_norm": 1.3484625756463664,
      "learning_rate": 3.4110583938241393e-06,
      "loss": 0.5498,
      "step": 12113
    },
    {
      "epoch": 0.6537153958232151,
      "grad_norm": 0.9989149539752589,
      "learning_rate": 3.410382710168184e-06,
      "loss": 0.5203,
      "step": 12114
    },
    {
      "epoch": 0.6537693594517295,
      "grad_norm": 1.1562384366566094,
      "learning_rate": 3.40970708656994e-06,
      "loss": 0.5376,
      "step": 12115
    },
    {
      "epoch": 0.6538233230802439,
      "grad_norm": 1.1213423113443282,
      "learning_rate": 3.409031523048828e-06,
      "loss": 0.5057,
      "step": 12116
    },
    {
      "epoch": 0.6538772867087583,
      "grad_norm": 0.8861429552945734,
      "learning_rate": 3.4083560196242617e-06,
      "loss": 0.4784,
      "step": 12117
    },
    {
      "epoch": 0.6539312503372727,
      "grad_norm": 1.112429768513656,
      "learning_rate": 3.4076805763156573e-06,
      "loss": 0.491,
      "step": 12118
    },
    {
      "epoch": 0.653985213965787,
      "grad_norm": 1.1086948563787358,
      "learning_rate": 3.4070051931424284e-06,
      "loss": 0.4846,
      "step": 12119
    },
    {
      "epoch": 0.6540391775943014,
      "grad_norm": 1.1354289306587633,
      "learning_rate": 3.4063298701239832e-06,
      "loss": 0.395,
      "step": 12120
    },
    {
      "epoch": 0.6540931412228158,
      "grad_norm": 1.0378982492090187,
      "learning_rate": 3.4056546072797344e-06,
      "loss": 0.5287,
      "step": 12121
    },
    {
      "epoch": 0.6541471048513302,
      "grad_norm": 1.2266164045772887,
      "learning_rate": 3.404979404629087e-06,
      "loss": 0.5329,
      "step": 12122
    },
    {
      "epoch": 0.6542010684798446,
      "grad_norm": 1.0441612695699238,
      "learning_rate": 3.404304262191449e-06,
      "loss": 0.5705,
      "step": 12123
    },
    {
      "epoch": 0.654255032108359,
      "grad_norm": 0.9754754466286323,
      "learning_rate": 3.4036291799862237e-06,
      "loss": 0.3627,
      "step": 12124
    },
    {
      "epoch": 0.6543089957368734,
      "grad_norm": 1.121923839022091,
      "learning_rate": 3.402954158032814e-06,
      "loss": 0.3975,
      "step": 12125
    },
    {
      "epoch": 0.6543629593653877,
      "grad_norm": 0.9047502521131885,
      "learning_rate": 3.402279196350622e-06,
      "loss": 0.3991,
      "step": 12126
    },
    {
      "epoch": 0.6544169229939021,
      "grad_norm": 0.9075175389431006,
      "learning_rate": 3.401604294959044e-06,
      "loss": 0.5051,
      "step": 12127
    },
    {
      "epoch": 0.6544708866224165,
      "grad_norm": 0.9063448776144766,
      "learning_rate": 3.4009294538774786e-06,
      "loss": 0.3976,
      "step": 12128
    },
    {
      "epoch": 0.6545248502509309,
      "grad_norm": 0.8582429870616064,
      "learning_rate": 3.400254673125323e-06,
      "loss": 0.3812,
      "step": 12129
    },
    {
      "epoch": 0.6545788138794453,
      "grad_norm": 1.302243887851338,
      "learning_rate": 3.3995799527219687e-06,
      "loss": 0.6367,
      "step": 12130
    },
    {
      "epoch": 0.6546327775079597,
      "grad_norm": 0.9951293836612463,
      "learning_rate": 3.3989052926868094e-06,
      "loss": 0.4002,
      "step": 12131
    },
    {
      "epoch": 0.6546867411364741,
      "grad_norm": 0.8580841558076736,
      "learning_rate": 3.398230693039235e-06,
      "loss": 0.3496,
      "step": 12132
    },
    {
      "epoch": 0.6547407047649884,
      "grad_norm": 1.14243652280191,
      "learning_rate": 3.3975561537986356e-06,
      "loss": 0.6479,
      "step": 12133
    },
    {
      "epoch": 0.6547946683935028,
      "grad_norm": 0.8949416506701103,
      "learning_rate": 3.396881674984395e-06,
      "loss": 0.491,
      "step": 12134
    },
    {
      "epoch": 0.6548486320220172,
      "grad_norm": 0.85843642984806,
      "learning_rate": 3.3962072566159e-06,
      "loss": 0.4148,
      "step": 12135
    },
    {
      "epoch": 0.6549025956505315,
      "grad_norm": 1.0867720046662457,
      "learning_rate": 3.3955328987125348e-06,
      "loss": 0.4745,
      "step": 12136
    },
    {
      "epoch": 0.654956559279046,
      "grad_norm": 0.9093568248135352,
      "learning_rate": 3.394858601293679e-06,
      "loss": 0.444,
      "step": 12137
    },
    {
      "epoch": 0.6550105229075603,
      "grad_norm": 1.0751662244535751,
      "learning_rate": 3.3941843643787166e-06,
      "loss": 0.456,
      "step": 12138
    },
    {
      "epoch": 0.6550644865360746,
      "grad_norm": 1.0402423735756046,
      "learning_rate": 3.3935101879870225e-06,
      "loss": 0.459,
      "step": 12139
    },
    {
      "epoch": 0.655118450164589,
      "grad_norm": 1.0151058524208336,
      "learning_rate": 3.392836072137974e-06,
      "loss": 0.3637,
      "step": 12140
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 1.1234945874573554,
      "learning_rate": 3.392162016850945e-06,
      "loss": 0.6899,
      "step": 12141
    },
    {
      "epoch": 0.6552263774216178,
      "grad_norm": 1.0044285895397762,
      "learning_rate": 3.39148802214531e-06,
      "loss": 0.476,
      "step": 12142
    },
    {
      "epoch": 0.6552803410501322,
      "grad_norm": 1.0742968907249537,
      "learning_rate": 3.3908140880404395e-06,
      "loss": 0.6608,
      "step": 12143
    },
    {
      "epoch": 0.6553343046786466,
      "grad_norm": 1.1635184753210615,
      "learning_rate": 3.3901402145557035e-06,
      "loss": 0.5652,
      "step": 12144
    },
    {
      "epoch": 0.655388268307161,
      "grad_norm": 0.9494032212360588,
      "learning_rate": 3.3894664017104695e-06,
      "loss": 0.3351,
      "step": 12145
    },
    {
      "epoch": 0.6554422319356753,
      "grad_norm": 1.00114269483659,
      "learning_rate": 3.3887926495241054e-06,
      "loss": 0.4381,
      "step": 12146
    },
    {
      "epoch": 0.6554961955641897,
      "grad_norm": 1.042188379874871,
      "learning_rate": 3.388118958015972e-06,
      "loss": 0.4573,
      "step": 12147
    },
    {
      "epoch": 0.6555501591927041,
      "grad_norm": 1.0939901414315485,
      "learning_rate": 3.387445327205433e-06,
      "loss": 0.4892,
      "step": 12148
    },
    {
      "epoch": 0.6556041228212185,
      "grad_norm": 1.1528318508086728,
      "learning_rate": 3.386771757111851e-06,
      "loss": 0.5992,
      "step": 12149
    },
    {
      "epoch": 0.6556580864497329,
      "grad_norm": 1.0067458181955184,
      "learning_rate": 3.386098247754583e-06,
      "loss": 0.6315,
      "step": 12150
    },
    {
      "epoch": 0.6557120500782473,
      "grad_norm": 1.0938134982814334,
      "learning_rate": 3.385424799152988e-06,
      "loss": 0.5071,
      "step": 12151
    },
    {
      "epoch": 0.6557660137067617,
      "grad_norm": 1.00058059127814,
      "learning_rate": 3.384751411326421e-06,
      "loss": 0.4661,
      "step": 12152
    },
    {
      "epoch": 0.655819977335276,
      "grad_norm": 1.2303754868518313,
      "learning_rate": 3.384078084294237e-06,
      "loss": 0.4711,
      "step": 12153
    },
    {
      "epoch": 0.6558739409637904,
      "grad_norm": 1.1312480019703492,
      "learning_rate": 3.383404818075785e-06,
      "loss": 0.4892,
      "step": 12154
    },
    {
      "epoch": 0.6559279045923048,
      "grad_norm": 1.022243255742534,
      "learning_rate": 3.382731612690417e-06,
      "loss": 0.4483,
      "step": 12155
    },
    {
      "epoch": 0.6559818682208192,
      "grad_norm": 0.9637087863840178,
      "learning_rate": 3.3820584681574824e-06,
      "loss": 0.5963,
      "step": 12156
    },
    {
      "epoch": 0.6560358318493336,
      "grad_norm": 1.0522111684803257,
      "learning_rate": 3.3813853844963273e-06,
      "loss": 0.409,
      "step": 12157
    },
    {
      "epoch": 0.656089795477848,
      "grad_norm": 0.954550152796691,
      "learning_rate": 3.3807123617262972e-06,
      "loss": 0.4487,
      "step": 12158
    },
    {
      "epoch": 0.6561437591063624,
      "grad_norm": 0.9322011756917016,
      "learning_rate": 3.3800393998667346e-06,
      "loss": 0.4575,
      "step": 12159
    },
    {
      "epoch": 0.6561977227348766,
      "grad_norm": 1.022577014050822,
      "learning_rate": 3.3793664989369836e-06,
      "loss": 0.4251,
      "step": 12160
    },
    {
      "epoch": 0.656251686363391,
      "grad_norm": 1.185693590072727,
      "learning_rate": 3.3786936589563803e-06,
      "loss": 0.5736,
      "step": 12161
    },
    {
      "epoch": 0.6563056499919054,
      "grad_norm": 0.9442073239862351,
      "learning_rate": 3.378020879944265e-06,
      "loss": 0.4278,
      "step": 12162
    },
    {
      "epoch": 0.6563596136204198,
      "grad_norm": 0.7815244467994428,
      "learning_rate": 3.3773481619199743e-06,
      "loss": 0.3923,
      "step": 12163
    },
    {
      "epoch": 0.6564135772489342,
      "grad_norm": 0.9368062150422383,
      "learning_rate": 3.3766755049028417e-06,
      "loss": 0.3805,
      "step": 12164
    },
    {
      "epoch": 0.6564675408774486,
      "grad_norm": 1.1804718088695256,
      "learning_rate": 3.3760029089121993e-06,
      "loss": 0.6681,
      "step": 12165
    },
    {
      "epoch": 0.656521504505963,
      "grad_norm": 1.0801534754427602,
      "learning_rate": 3.3753303739673816e-06,
      "loss": 0.3612,
      "step": 12166
    },
    {
      "epoch": 0.6565754681344773,
      "grad_norm": 1.1841713812398418,
      "learning_rate": 3.374657900087716e-06,
      "loss": 0.6189,
      "step": 12167
    },
    {
      "epoch": 0.6566294317629917,
      "grad_norm": 0.9380881393035884,
      "learning_rate": 3.3739854872925285e-06,
      "loss": 0.4989,
      "step": 12168
    },
    {
      "epoch": 0.6566833953915061,
      "grad_norm": 0.9790180037542231,
      "learning_rate": 3.3733131356011468e-06,
      "loss": 0.5181,
      "step": 12169
    },
    {
      "epoch": 0.6567373590200205,
      "grad_norm": 0.9522124072034627,
      "learning_rate": 3.372640845032894e-06,
      "loss": 0.5104,
      "step": 12170
    },
    {
      "epoch": 0.6567913226485349,
      "grad_norm": 1.0645186726080271,
      "learning_rate": 3.371968615607093e-06,
      "loss": 0.4319,
      "step": 12171
    },
    {
      "epoch": 0.6568452862770493,
      "grad_norm": 1.037184924207552,
      "learning_rate": 3.3712964473430633e-06,
      "loss": 0.3866,
      "step": 12172
    },
    {
      "epoch": 0.6568992499055637,
      "grad_norm": 0.9373937148245929,
      "learning_rate": 3.3706243402601257e-06,
      "loss": 0.3651,
      "step": 12173
    },
    {
      "epoch": 0.656953213534078,
      "grad_norm": 1.0244755554822482,
      "learning_rate": 3.3699522943775957e-06,
      "loss": 0.5154,
      "step": 12174
    },
    {
      "epoch": 0.6570071771625924,
      "grad_norm": 1.1886502395125522,
      "learning_rate": 3.369280309714789e-06,
      "loss": 0.5904,
      "step": 12175
    },
    {
      "epoch": 0.6570611407911068,
      "grad_norm": 0.9286856637993067,
      "learning_rate": 3.3686083862910197e-06,
      "loss": 0.3797,
      "step": 12176
    },
    {
      "epoch": 0.6571151044196212,
      "grad_norm": 1.0879204370384175,
      "learning_rate": 3.3679365241255984e-06,
      "loss": 0.3942,
      "step": 12177
    },
    {
      "epoch": 0.6571690680481356,
      "grad_norm": 1.0388791497416898,
      "learning_rate": 3.3672647232378362e-06,
      "loss": 0.4734,
      "step": 12178
    },
    {
      "epoch": 0.65722303167665,
      "grad_norm": 0.8996147259832636,
      "learning_rate": 3.36659298364704e-06,
      "loss": 0.4159,
      "step": 12179
    },
    {
      "epoch": 0.6572769953051644,
      "grad_norm": 1.0553742792449772,
      "learning_rate": 3.365921305372519e-06,
      "loss": 0.4839,
      "step": 12180
    },
    {
      "epoch": 0.6573309589336787,
      "grad_norm": 0.9078863643115785,
      "learning_rate": 3.3652496884335752e-06,
      "loss": 0.3723,
      "step": 12181
    },
    {
      "epoch": 0.657384922562193,
      "grad_norm": 1.0884198503250355,
      "learning_rate": 3.3645781328495116e-06,
      "loss": 0.4603,
      "step": 12182
    },
    {
      "epoch": 0.6574388861907074,
      "grad_norm": 1.1130969762554492,
      "learning_rate": 3.36390663863963e-06,
      "loss": 0.5096,
      "step": 12183
    },
    {
      "epoch": 0.6574928498192218,
      "grad_norm": 0.9538956939587255,
      "learning_rate": 3.3632352058232304e-06,
      "loss": 0.4605,
      "step": 12184
    },
    {
      "epoch": 0.6575468134477362,
      "grad_norm": 1.048531128880802,
      "learning_rate": 3.36256383441961e-06,
      "loss": 0.5209,
      "step": 12185
    },
    {
      "epoch": 0.6576007770762506,
      "grad_norm": 1.1202353800658622,
      "learning_rate": 3.361892524448065e-06,
      "loss": 0.4167,
      "step": 12186
    },
    {
      "epoch": 0.657654740704765,
      "grad_norm": 0.8865106313273421,
      "learning_rate": 3.3612212759278876e-06,
      "loss": 0.3767,
      "step": 12187
    },
    {
      "epoch": 0.6577087043332793,
      "grad_norm": 1.0838164709522071,
      "learning_rate": 3.3605500888783756e-06,
      "loss": 0.5275,
      "step": 12188
    },
    {
      "epoch": 0.6577626679617937,
      "grad_norm": 0.948477124170272,
      "learning_rate": 3.359878963318813e-06,
      "loss": 0.5345,
      "step": 12189
    },
    {
      "epoch": 0.6578166315903081,
      "grad_norm": 1.02294667275426,
      "learning_rate": 3.3592078992684916e-06,
      "loss": 0.4662,
      "step": 12190
    },
    {
      "epoch": 0.6578705952188225,
      "grad_norm": 0.9452756299791263,
      "learning_rate": 3.3585368967466986e-06,
      "loss": 0.453,
      "step": 12191
    },
    {
      "epoch": 0.6579245588473369,
      "grad_norm": 0.8044540639299731,
      "learning_rate": 3.357865955772719e-06,
      "loss": 0.3359,
      "step": 12192
    },
    {
      "epoch": 0.6579785224758513,
      "grad_norm": 1.0882995784077858,
      "learning_rate": 3.357195076365836e-06,
      "loss": 0.5098,
      "step": 12193
    },
    {
      "epoch": 0.6580324861043657,
      "grad_norm": 0.782231330441835,
      "learning_rate": 3.3565242585453313e-06,
      "loss": 0.2558,
      "step": 12194
    },
    {
      "epoch": 0.65808644973288,
      "grad_norm": 1.2483132067618739,
      "learning_rate": 3.3558535023304873e-06,
      "loss": 0.6465,
      "step": 12195
    },
    {
      "epoch": 0.6581404133613944,
      "grad_norm": 0.9884210505857033,
      "learning_rate": 3.355182807740579e-06,
      "loss": 0.4795,
      "step": 12196
    },
    {
      "epoch": 0.6581943769899088,
      "grad_norm": 0.9737533426779393,
      "learning_rate": 3.3545121747948834e-06,
      "loss": 0.5507,
      "step": 12197
    },
    {
      "epoch": 0.6582483406184232,
      "grad_norm": 0.9812998041549411,
      "learning_rate": 3.353841603512676e-06,
      "loss": 0.4807,
      "step": 12198
    },
    {
      "epoch": 0.6583023042469376,
      "grad_norm": 0.7624453323717157,
      "learning_rate": 3.353171093913229e-06,
      "loss": 0.3031,
      "step": 12199
    },
    {
      "epoch": 0.658356267875452,
      "grad_norm": 1.0202084758022034,
      "learning_rate": 3.352500646015814e-06,
      "loss": 0.4007,
      "step": 12200
    },
    {
      "epoch": 0.6584102315039664,
      "grad_norm": 0.8602368584755781,
      "learning_rate": 3.3518302598397006e-06,
      "loss": 0.3394,
      "step": 12201
    },
    {
      "epoch": 0.6584641951324807,
      "grad_norm": 1.2541901143756347,
      "learning_rate": 3.3511599354041578e-06,
      "loss": 0.6662,
      "step": 12202
    },
    {
      "epoch": 0.6585181587609951,
      "grad_norm": 0.9702604009711312,
      "learning_rate": 3.3504896727284485e-06,
      "loss": 0.4823,
      "step": 12203
    },
    {
      "epoch": 0.6585721223895095,
      "grad_norm": 0.9427146167118085,
      "learning_rate": 3.3498194718318373e-06,
      "loss": 0.4446,
      "step": 12204
    },
    {
      "epoch": 0.6586260860180239,
      "grad_norm": 0.9716802761429558,
      "learning_rate": 3.3491493327335865e-06,
      "loss": 0.405,
      "step": 12205
    },
    {
      "epoch": 0.6586800496465383,
      "grad_norm": 1.1890106636612499,
      "learning_rate": 3.3484792554529577e-06,
      "loss": 0.573,
      "step": 12206
    },
    {
      "epoch": 0.6587340132750527,
      "grad_norm": 0.9226141506328585,
      "learning_rate": 3.3478092400092094e-06,
      "loss": 0.3758,
      "step": 12207
    },
    {
      "epoch": 0.6587879769035669,
      "grad_norm": 0.8356932889245835,
      "learning_rate": 3.3471392864216e-06,
      "loss": 0.3322,
      "step": 12208
    },
    {
      "epoch": 0.6588419405320813,
      "grad_norm": 0.9571008278539685,
      "learning_rate": 3.3464693947093806e-06,
      "loss": 0.3961,
      "step": 12209
    },
    {
      "epoch": 0.6588959041605957,
      "grad_norm": 0.9777187331600683,
      "learning_rate": 3.3457995648918074e-06,
      "loss": 0.4027,
      "step": 12210
    },
    {
      "epoch": 0.6589498677891101,
      "grad_norm": 0.9358816641031806,
      "learning_rate": 3.345129796988131e-06,
      "loss": 0.366,
      "step": 12211
    },
    {
      "epoch": 0.6590038314176245,
      "grad_norm": 0.7826089547171353,
      "learning_rate": 3.344460091017601e-06,
      "loss": 0.325,
      "step": 12212
    },
    {
      "epoch": 0.6590577950461389,
      "grad_norm": 1.1502360347298302,
      "learning_rate": 3.3437904469994666e-06,
      "loss": 0.4785,
      "step": 12213
    },
    {
      "epoch": 0.6591117586746533,
      "grad_norm": 1.173928151965994,
      "learning_rate": 3.3431208649529736e-06,
      "loss": 0.4297,
      "step": 12214
    },
    {
      "epoch": 0.6591657223031676,
      "grad_norm": 1.0587934037257585,
      "learning_rate": 3.3424513448973673e-06,
      "loss": 0.4156,
      "step": 12215
    },
    {
      "epoch": 0.659219685931682,
      "grad_norm": 1.0392485305965347,
      "learning_rate": 3.341781886851888e-06,
      "loss": 0.5544,
      "step": 12216
    },
    {
      "epoch": 0.6592736495601964,
      "grad_norm": 1.1981757940821114,
      "learning_rate": 3.3411124908357783e-06,
      "loss": 0.7293,
      "step": 12217
    },
    {
      "epoch": 0.6593276131887108,
      "grad_norm": 0.9546313280702993,
      "learning_rate": 3.340443156868277e-06,
      "loss": 0.4546,
      "step": 12218
    },
    {
      "epoch": 0.6593815768172252,
      "grad_norm": 0.949185195549188,
      "learning_rate": 3.339773884968621e-06,
      "loss": 0.3623,
      "step": 12219
    },
    {
      "epoch": 0.6594355404457396,
      "grad_norm": 1.0788401839087185,
      "learning_rate": 3.339104675156047e-06,
      "loss": 0.5909,
      "step": 12220
    },
    {
      "epoch": 0.659489504074254,
      "grad_norm": 1.1581845294250535,
      "learning_rate": 3.3384355274497875e-06,
      "loss": 0.473,
      "step": 12221
    },
    {
      "epoch": 0.6595434677027683,
      "grad_norm": 0.8208718733004026,
      "learning_rate": 3.3377664418690746e-06,
      "loss": 0.2774,
      "step": 12222
    },
    {
      "epoch": 0.6595974313312827,
      "grad_norm": 0.8504207927978109,
      "learning_rate": 3.33709741843314e-06,
      "loss": 0.4262,
      "step": 12223
    },
    {
      "epoch": 0.6596513949597971,
      "grad_norm": 1.2549731288314645,
      "learning_rate": 3.3364284571612125e-06,
      "loss": 0.4444,
      "step": 12224
    },
    {
      "epoch": 0.6597053585883115,
      "grad_norm": 1.240237445929437,
      "learning_rate": 3.335759558072515e-06,
      "loss": 0.6003,
      "step": 12225
    },
    {
      "epoch": 0.6597593222168259,
      "grad_norm": 1.188913708534342,
      "learning_rate": 3.335090721186275e-06,
      "loss": 0.4332,
      "step": 12226
    },
    {
      "epoch": 0.6598132858453403,
      "grad_norm": 1.0778273829687348,
      "learning_rate": 3.3344219465217155e-06,
      "loss": 0.4514,
      "step": 12227
    },
    {
      "epoch": 0.6598672494738547,
      "grad_norm": 0.7479582206195925,
      "learning_rate": 3.3337532340980573e-06,
      "loss": 0.3055,
      "step": 12228
    },
    {
      "epoch": 0.659921213102369,
      "grad_norm": 1.0737175098820917,
      "learning_rate": 3.3330845839345196e-06,
      "loss": 0.4033,
      "step": 12229
    },
    {
      "epoch": 0.6599751767308834,
      "grad_norm": 1.0598536169339137,
      "learning_rate": 3.3324159960503198e-06,
      "loss": 0.4562,
      "step": 12230
    },
    {
      "epoch": 0.6600291403593977,
      "grad_norm": 0.8780672378325353,
      "learning_rate": 3.3317474704646753e-06,
      "loss": 0.3296,
      "step": 12231
    },
    {
      "epoch": 0.6600831039879121,
      "grad_norm": 0.9867162385675662,
      "learning_rate": 3.331079007196799e-06,
      "loss": 0.5471,
      "step": 12232
    },
    {
      "epoch": 0.6601370676164265,
      "grad_norm": 1.0508204795855765,
      "learning_rate": 3.330410606265904e-06,
      "loss": 0.4962,
      "step": 12233
    },
    {
      "epoch": 0.6601910312449409,
      "grad_norm": 1.0612917671591466,
      "learning_rate": 3.3297422676912007e-06,
      "loss": 0.5287,
      "step": 12234
    },
    {
      "epoch": 0.6602449948734553,
      "grad_norm": 0.8008331881374267,
      "learning_rate": 3.329073991491897e-06,
      "loss": 0.3163,
      "step": 12235
    },
    {
      "epoch": 0.6602989585019696,
      "grad_norm": 1.0390195876166075,
      "learning_rate": 3.328405777687202e-06,
      "loss": 0.5613,
      "step": 12236
    },
    {
      "epoch": 0.660352922130484,
      "grad_norm": 0.799754072595658,
      "learning_rate": 3.3277376262963174e-06,
      "loss": 0.3514,
      "step": 12237
    },
    {
      "epoch": 0.6604068857589984,
      "grad_norm": 1.4848576046570632,
      "learning_rate": 3.3270695373384488e-06,
      "loss": 0.5934,
      "step": 12238
    },
    {
      "epoch": 0.6604608493875128,
      "grad_norm": 1.0513071092814519,
      "learning_rate": 3.3264015108327962e-06,
      "loss": 0.6486,
      "step": 12239
    },
    {
      "epoch": 0.6605148130160272,
      "grad_norm": 1.2070462145444176,
      "learning_rate": 3.325733546798562e-06,
      "loss": 0.4518,
      "step": 12240
    },
    {
      "epoch": 0.6605687766445416,
      "grad_norm": 0.9529697645243869,
      "learning_rate": 3.325065645254941e-06,
      "loss": 0.4636,
      "step": 12241
    },
    {
      "epoch": 0.660622740273056,
      "grad_norm": 1.0020539449320847,
      "learning_rate": 3.3243978062211325e-06,
      "loss": 0.4623,
      "step": 12242
    },
    {
      "epoch": 0.6606767039015703,
      "grad_norm": 1.1228428477912078,
      "learning_rate": 3.32373002971633e-06,
      "loss": 0.5096,
      "step": 12243
    },
    {
      "epoch": 0.6607306675300847,
      "grad_norm": 1.1918665723736026,
      "learning_rate": 3.323062315759724e-06,
      "loss": 0.6541,
      "step": 12244
    },
    {
      "epoch": 0.6607846311585991,
      "grad_norm": 1.1690607812554088,
      "learning_rate": 3.322394664370507e-06,
      "loss": 0.4254,
      "step": 12245
    },
    {
      "epoch": 0.6608385947871135,
      "grad_norm": 1.009091810995177,
      "learning_rate": 3.321727075567867e-06,
      "loss": 0.4312,
      "step": 12246
    },
    {
      "epoch": 0.6608925584156279,
      "grad_norm": 0.8977342482618287,
      "learning_rate": 3.3210595493709928e-06,
      "loss": 0.3218,
      "step": 12247
    },
    {
      "epoch": 0.6609465220441423,
      "grad_norm": 0.8755678387028514,
      "learning_rate": 3.320392085799068e-06,
      "loss": 0.3261,
      "step": 12248
    },
    {
      "epoch": 0.6610004856726567,
      "grad_norm": 0.915321156602366,
      "learning_rate": 3.319724684871277e-06,
      "loss": 0.4698,
      "step": 12249
    },
    {
      "epoch": 0.661054449301171,
      "grad_norm": 1.0091092637069579,
      "learning_rate": 3.3190573466068044e-06,
      "loss": 0.3502,
      "step": 12250
    },
    {
      "epoch": 0.6611084129296854,
      "grad_norm": 1.0408562249313136,
      "learning_rate": 3.318390071024824e-06,
      "loss": 0.5303,
      "step": 12251
    },
    {
      "epoch": 0.6611623765581998,
      "grad_norm": 1.1440654510925317,
      "learning_rate": 3.3177228581445175e-06,
      "loss": 0.4794,
      "step": 12252
    },
    {
      "epoch": 0.6612163401867142,
      "grad_norm": 1.0742478070175858,
      "learning_rate": 3.317055707985062e-06,
      "loss": 0.5111,
      "step": 12253
    },
    {
      "epoch": 0.6612703038152286,
      "grad_norm": 1.1070637108901493,
      "learning_rate": 3.3163886205656307e-06,
      "loss": 0.4847,
      "step": 12254
    },
    {
      "epoch": 0.661324267443743,
      "grad_norm": 1.0902545059334203,
      "learning_rate": 3.315721595905397e-06,
      "loss": 0.5897,
      "step": 12255
    },
    {
      "epoch": 0.6613782310722573,
      "grad_norm": 1.1273818527927026,
      "learning_rate": 3.3150546340235316e-06,
      "loss": 0.5434,
      "step": 12256
    },
    {
      "epoch": 0.6614321947007716,
      "grad_norm": 0.9067226109925057,
      "learning_rate": 3.3143877349392055e-06,
      "loss": 0.4589,
      "step": 12257
    },
    {
      "epoch": 0.661486158329286,
      "grad_norm": 1.140436368949196,
      "learning_rate": 3.3137208986715823e-06,
      "loss": 0.4286,
      "step": 12258
    },
    {
      "epoch": 0.6615401219578004,
      "grad_norm": 1.203566615149496,
      "learning_rate": 3.3130541252398298e-06,
      "loss": 0.5266,
      "step": 12259
    },
    {
      "epoch": 0.6615940855863148,
      "grad_norm": 0.9657645858743545,
      "learning_rate": 3.3123874146631117e-06,
      "loss": 0.4039,
      "step": 12260
    },
    {
      "epoch": 0.6616480492148292,
      "grad_norm": 0.9895307653846924,
      "learning_rate": 3.311720766960589e-06,
      "loss": 0.4219,
      "step": 12261
    },
    {
      "epoch": 0.6617020128433436,
      "grad_norm": 0.9627087101560815,
      "learning_rate": 3.311054182151423e-06,
      "loss": 0.5045,
      "step": 12262
    },
    {
      "epoch": 0.661755976471858,
      "grad_norm": 1.0996241050073883,
      "learning_rate": 3.310387660254772e-06,
      "loss": 0.3977,
      "step": 12263
    },
    {
      "epoch": 0.6618099401003723,
      "grad_norm": 1.0173905744812322,
      "learning_rate": 3.309721201289794e-06,
      "loss": 0.4094,
      "step": 12264
    },
    {
      "epoch": 0.6618639037288867,
      "grad_norm": 1.0867268094893792,
      "learning_rate": 3.309054805275639e-06,
      "loss": 0.4563,
      "step": 12265
    },
    {
      "epoch": 0.6619178673574011,
      "grad_norm": 0.9975244524229353,
      "learning_rate": 3.3083884722314624e-06,
      "loss": 0.4086,
      "step": 12266
    },
    {
      "epoch": 0.6619718309859155,
      "grad_norm": 1.1802889749439103,
      "learning_rate": 3.307722202176417e-06,
      "loss": 0.5342,
      "step": 12267
    },
    {
      "epoch": 0.6620257946144299,
      "grad_norm": 1.0131807846438936,
      "learning_rate": 3.30705599512965e-06,
      "loss": 0.4973,
      "step": 12268
    },
    {
      "epoch": 0.6620797582429443,
      "grad_norm": 1.0110332500984178,
      "learning_rate": 3.30638985111031e-06,
      "loss": 0.4427,
      "step": 12269
    },
    {
      "epoch": 0.6621337218714587,
      "grad_norm": 1.0122444327216822,
      "learning_rate": 3.305723770137541e-06,
      "loss": 0.5084,
      "step": 12270
    },
    {
      "epoch": 0.662187685499973,
      "grad_norm": 0.9487475464827033,
      "learning_rate": 3.3050577522304903e-06,
      "loss": 0.4468,
      "step": 12271
    },
    {
      "epoch": 0.6622416491284874,
      "grad_norm": 1.2752246457949856,
      "learning_rate": 3.304391797408296e-06,
      "loss": 0.6757,
      "step": 12272
    },
    {
      "epoch": 0.6622956127570018,
      "grad_norm": 1.1066168130289051,
      "learning_rate": 3.3037259056900994e-06,
      "loss": 0.4638,
      "step": 12273
    },
    {
      "epoch": 0.6623495763855162,
      "grad_norm": 1.197230076595792,
      "learning_rate": 3.3030600770950406e-06,
      "loss": 0.5658,
      "step": 12274
    },
    {
      "epoch": 0.6624035400140306,
      "grad_norm": 0.8226303821835945,
      "learning_rate": 3.302394311642254e-06,
      "loss": 0.3699,
      "step": 12275
    },
    {
      "epoch": 0.662457503642545,
      "grad_norm": 0.9290210444676861,
      "learning_rate": 3.301728609350876e-06,
      "loss": 0.3833,
      "step": 12276
    },
    {
      "epoch": 0.6625114672710593,
      "grad_norm": 1.0479046055117764,
      "learning_rate": 3.3010629702400387e-06,
      "loss": 0.51,
      "step": 12277
    },
    {
      "epoch": 0.6625654308995736,
      "grad_norm": 1.0536524189436811,
      "learning_rate": 3.3003973943288735e-06,
      "loss": 0.428,
      "step": 12278
    },
    {
      "epoch": 0.662619394528088,
      "grad_norm": 0.8761353433537651,
      "learning_rate": 3.29973188163651e-06,
      "loss": 0.3461,
      "step": 12279
    },
    {
      "epoch": 0.6626733581566024,
      "grad_norm": 1.13687573786356,
      "learning_rate": 3.2990664321820753e-06,
      "loss": 0.4621,
      "step": 12280
    },
    {
      "epoch": 0.6627273217851168,
      "grad_norm": 1.0900065377158878,
      "learning_rate": 3.2984010459846948e-06,
      "loss": 0.6846,
      "step": 12281
    },
    {
      "epoch": 0.6627812854136312,
      "grad_norm": 1.2131994482478845,
      "learning_rate": 3.297735723063493e-06,
      "loss": 0.4263,
      "step": 12282
    },
    {
      "epoch": 0.6628352490421456,
      "grad_norm": 1.1246783217029643,
      "learning_rate": 3.2970704634375915e-06,
      "loss": 0.504,
      "step": 12283
    },
    {
      "epoch": 0.6628892126706599,
      "grad_norm": 1.0641911186818613,
      "learning_rate": 3.296405267126114e-06,
      "loss": 0.456,
      "step": 12284
    },
    {
      "epoch": 0.6629431762991743,
      "grad_norm": 0.940960316702678,
      "learning_rate": 3.295740134148172e-06,
      "loss": 0.4507,
      "step": 12285
    },
    {
      "epoch": 0.6629971399276887,
      "grad_norm": 0.9832344728519783,
      "learning_rate": 3.2950750645228868e-06,
      "loss": 0.4535,
      "step": 12286
    },
    {
      "epoch": 0.6630511035562031,
      "grad_norm": 1.1385669144373034,
      "learning_rate": 3.2944100582693717e-06,
      "loss": 0.5047,
      "step": 12287
    },
    {
      "epoch": 0.6631050671847175,
      "grad_norm": 1.0054123034935745,
      "learning_rate": 3.2937451154067403e-06,
      "loss": 0.4942,
      "step": 12288
    },
    {
      "epoch": 0.6631590308132319,
      "grad_norm": 0.8871940820885871,
      "learning_rate": 3.293080235954104e-06,
      "loss": 0.3786,
      "step": 12289
    },
    {
      "epoch": 0.6632129944417463,
      "grad_norm": 1.2173029324688913,
      "learning_rate": 3.292415419930571e-06,
      "loss": 0.367,
      "step": 12290
    },
    {
      "epoch": 0.6632669580702606,
      "grad_norm": 1.0567106506106088,
      "learning_rate": 3.2917506673552514e-06,
      "loss": 0.4384,
      "step": 12291
    },
    {
      "epoch": 0.663320921698775,
      "grad_norm": 1.2591132056098104,
      "learning_rate": 3.291085978247247e-06,
      "loss": 0.6048,
      "step": 12292
    },
    {
      "epoch": 0.6633748853272894,
      "grad_norm": 0.9730099870654649,
      "learning_rate": 3.290421352625664e-06,
      "loss": 0.6353,
      "step": 12293
    },
    {
      "epoch": 0.6634288489558038,
      "grad_norm": 0.8050400126153652,
      "learning_rate": 3.289756790509604e-06,
      "loss": 0.3587,
      "step": 12294
    },
    {
      "epoch": 0.6634828125843182,
      "grad_norm": 1.1382932967756263,
      "learning_rate": 3.289092291918167e-06,
      "loss": 0.5188,
      "step": 12295
    },
    {
      "epoch": 0.6635367762128326,
      "grad_norm": 1.3276462869402965,
      "learning_rate": 3.288427856870451e-06,
      "loss": 0.789,
      "step": 12296
    },
    {
      "epoch": 0.663590739841347,
      "grad_norm": 1.2317524044319534,
      "learning_rate": 3.2877634853855534e-06,
      "loss": 0.524,
      "step": 12297
    },
    {
      "epoch": 0.6636447034698613,
      "grad_norm": 1.1769963930182263,
      "learning_rate": 3.28709917748257e-06,
      "loss": 0.4946,
      "step": 12298
    },
    {
      "epoch": 0.6636986670983757,
      "grad_norm": 1.0155236338182156,
      "learning_rate": 3.28643493318059e-06,
      "loss": 0.5435,
      "step": 12299
    },
    {
      "epoch": 0.6637526307268901,
      "grad_norm": 0.8761272135918134,
      "learning_rate": 3.285770752498708e-06,
      "loss": 0.3819,
      "step": 12300
    },
    {
      "epoch": 0.6638065943554045,
      "grad_norm": 0.9450125650388921,
      "learning_rate": 3.2851066354560113e-06,
      "loss": 0.4806,
      "step": 12301
    },
    {
      "epoch": 0.6638605579839189,
      "grad_norm": 0.886842047027112,
      "learning_rate": 3.2844425820715874e-06,
      "loss": 0.3645,
      "step": 12302
    },
    {
      "epoch": 0.6639145216124333,
      "grad_norm": 0.9534722180372669,
      "learning_rate": 3.283778592364523e-06,
      "loss": 0.4082,
      "step": 12303
    },
    {
      "epoch": 0.6639684852409476,
      "grad_norm": 1.0419762164358175,
      "learning_rate": 3.2831146663539005e-06,
      "loss": 0.4179,
      "step": 12304
    },
    {
      "epoch": 0.6640224488694619,
      "grad_norm": 1.0995510255803589,
      "learning_rate": 3.282450804058804e-06,
      "loss": 0.4446,
      "step": 12305
    },
    {
      "epoch": 0.6640764124979763,
      "grad_norm": 1.1564253375772022,
      "learning_rate": 3.2817870054983127e-06,
      "loss": 0.6377,
      "step": 12306
    },
    {
      "epoch": 0.6641303761264907,
      "grad_norm": 1.0337355511529431,
      "learning_rate": 3.2811232706915018e-06,
      "loss": 0.5734,
      "step": 12307
    },
    {
      "epoch": 0.6641843397550051,
      "grad_norm": 0.9636586743795145,
      "learning_rate": 3.2804595996574507e-06,
      "loss": 0.3831,
      "step": 12308
    },
    {
      "epoch": 0.6642383033835195,
      "grad_norm": 1.213459088576215,
      "learning_rate": 3.2797959924152335e-06,
      "loss": 0.614,
      "step": 12309
    },
    {
      "epoch": 0.6642922670120339,
      "grad_norm": 0.9499648162673885,
      "learning_rate": 3.279132448983923e-06,
      "loss": 0.6809,
      "step": 12310
    },
    {
      "epoch": 0.6643462306405483,
      "grad_norm": 1.0484144984404247,
      "learning_rate": 3.27846896938259e-06,
      "loss": 0.4262,
      "step": 12311
    },
    {
      "epoch": 0.6644001942690626,
      "grad_norm": 0.9221714612338353,
      "learning_rate": 3.2778055536303042e-06,
      "loss": 0.2954,
      "step": 12312
    },
    {
      "epoch": 0.664454157897577,
      "grad_norm": 0.996160777510321,
      "learning_rate": 3.277142201746132e-06,
      "loss": 0.4754,
      "step": 12313
    },
    {
      "epoch": 0.6645081215260914,
      "grad_norm": 1.0575855839198571,
      "learning_rate": 3.2764789137491375e-06,
      "loss": 0.4332,
      "step": 12314
    },
    {
      "epoch": 0.6645620851546058,
      "grad_norm": 0.9931279686967059,
      "learning_rate": 3.2758156896583865e-06,
      "loss": 0.4825,
      "step": 12315
    },
    {
      "epoch": 0.6646160487831202,
      "grad_norm": 0.9875673519489118,
      "learning_rate": 3.2751525294929398e-06,
      "loss": 0.5105,
      "step": 12316
    },
    {
      "epoch": 0.6646700124116346,
      "grad_norm": 0.8880909339402102,
      "learning_rate": 3.2744894332718576e-06,
      "loss": 0.3762,
      "step": 12317
    },
    {
      "epoch": 0.664723976040149,
      "grad_norm": 0.9615724606234294,
      "learning_rate": 3.2738264010141974e-06,
      "loss": 0.5275,
      "step": 12318
    },
    {
      "epoch": 0.6647779396686633,
      "grad_norm": 1.1590976449051928,
      "learning_rate": 3.2731634327390182e-06,
      "loss": 0.5842,
      "step": 12319
    },
    {
      "epoch": 0.6648319032971777,
      "grad_norm": 0.9555928918133184,
      "learning_rate": 3.27250052846537e-06,
      "loss": 0.366,
      "step": 12320
    },
    {
      "epoch": 0.6648858669256921,
      "grad_norm": 0.9934289291026825,
      "learning_rate": 3.271837688212307e-06,
      "loss": 0.668,
      "step": 12321
    },
    {
      "epoch": 0.6649398305542065,
      "grad_norm": 1.2568837232677808,
      "learning_rate": 3.2711749119988812e-06,
      "loss": 0.4813,
      "step": 12322
    },
    {
      "epoch": 0.6649937941827209,
      "grad_norm": 1.1192395104947572,
      "learning_rate": 3.2705121998441384e-06,
      "loss": 0.4267,
      "step": 12323
    },
    {
      "epoch": 0.6650477578112353,
      "grad_norm": 1.1633132361437082,
      "learning_rate": 3.2698495517671297e-06,
      "loss": 0.5635,
      "step": 12324
    },
    {
      "epoch": 0.6651017214397497,
      "grad_norm": 0.8846721829242467,
      "learning_rate": 3.2691869677868977e-06,
      "loss": 0.4594,
      "step": 12325
    },
    {
      "epoch": 0.665155685068264,
      "grad_norm": 0.9198175363408206,
      "learning_rate": 3.2685244479224866e-06,
      "loss": 0.411,
      "step": 12326
    },
    {
      "epoch": 0.6652096486967783,
      "grad_norm": 0.924933117782707,
      "learning_rate": 3.267861992192937e-06,
      "loss": 0.3775,
      "step": 12327
    },
    {
      "epoch": 0.6652636123252927,
      "grad_norm": 0.7929835001724763,
      "learning_rate": 3.267199600617289e-06,
      "loss": 0.3076,
      "step": 12328
    },
    {
      "epoch": 0.6653175759538071,
      "grad_norm": 1.0058604091851406,
      "learning_rate": 3.266537273214582e-06,
      "loss": 0.5648,
      "step": 12329
    },
    {
      "epoch": 0.6653715395823215,
      "grad_norm": 0.8256984226627744,
      "learning_rate": 3.2658750100038507e-06,
      "loss": 0.3371,
      "step": 12330
    },
    {
      "epoch": 0.6654255032108359,
      "grad_norm": 1.2823918901619331,
      "learning_rate": 3.2652128110041277e-06,
      "loss": 0.6623,
      "step": 12331
    },
    {
      "epoch": 0.6654794668393503,
      "grad_norm": 1.01794539808815,
      "learning_rate": 3.264550676234447e-06,
      "loss": 0.5861,
      "step": 12332
    },
    {
      "epoch": 0.6655334304678646,
      "grad_norm": 1.0043160156983775,
      "learning_rate": 3.263888605713838e-06,
      "loss": 0.4525,
      "step": 12333
    },
    {
      "epoch": 0.665587394096379,
      "grad_norm": 1.26761373447209,
      "learning_rate": 3.2632265994613305e-06,
      "loss": 0.5509,
      "step": 12334
    },
    {
      "epoch": 0.6656413577248934,
      "grad_norm": 1.1971648009554738,
      "learning_rate": 3.262564657495951e-06,
      "loss": 0.594,
      "step": 12335
    },
    {
      "epoch": 0.6656953213534078,
      "grad_norm": 1.0144595246822048,
      "learning_rate": 3.261902779836723e-06,
      "loss": 0.4115,
      "step": 12336
    },
    {
      "epoch": 0.6657492849819222,
      "grad_norm": 1.0223204975749298,
      "learning_rate": 3.2612409665026712e-06,
      "loss": 0.4523,
      "step": 12337
    },
    {
      "epoch": 0.6658032486104366,
      "grad_norm": 0.895107628570298,
      "learning_rate": 3.260579217512815e-06,
      "loss": 0.3806,
      "step": 12338
    },
    {
      "epoch": 0.665857212238951,
      "grad_norm": 1.1548502267808474,
      "learning_rate": 3.2599175328861764e-06,
      "loss": 0.644,
      "step": 12339
    },
    {
      "epoch": 0.6659111758674653,
      "grad_norm": 0.9730964702599441,
      "learning_rate": 3.2592559126417723e-06,
      "loss": 0.4599,
      "step": 12340
    },
    {
      "epoch": 0.6659651394959797,
      "grad_norm": 0.9591535512049275,
      "learning_rate": 3.2585943567986156e-06,
      "loss": 0.4469,
      "step": 12341
    },
    {
      "epoch": 0.6660191031244941,
      "grad_norm": 0.7377201917412024,
      "learning_rate": 3.257932865375722e-06,
      "loss": 0.3553,
      "step": 12342
    },
    {
      "epoch": 0.6660730667530085,
      "grad_norm": 0.8051966424979499,
      "learning_rate": 3.2572714383921035e-06,
      "loss": 0.3587,
      "step": 12343
    },
    {
      "epoch": 0.6661270303815229,
      "grad_norm": 0.7890411760952944,
      "learning_rate": 3.2566100758667698e-06,
      "loss": 0.3621,
      "step": 12344
    },
    {
      "epoch": 0.6661809940100373,
      "grad_norm": 1.1731285025545173,
      "learning_rate": 3.255948777818729e-06,
      "loss": 0.5876,
      "step": 12345
    },
    {
      "epoch": 0.6662349576385516,
      "grad_norm": 0.9205837391813163,
      "learning_rate": 3.2552875442669896e-06,
      "loss": 0.4056,
      "step": 12346
    },
    {
      "epoch": 0.666288921267066,
      "grad_norm": 1.1253506943727885,
      "learning_rate": 3.254626375230553e-06,
      "loss": 0.4969,
      "step": 12347
    },
    {
      "epoch": 0.6663428848955804,
      "grad_norm": 0.9814810509970907,
      "learning_rate": 3.253965270728423e-06,
      "loss": 0.412,
      "step": 12348
    },
    {
      "epoch": 0.6663968485240948,
      "grad_norm": 0.8854114314641077,
      "learning_rate": 3.2533042307796004e-06,
      "loss": 0.3381,
      "step": 12349
    },
    {
      "epoch": 0.6664508121526092,
      "grad_norm": 1.1745992740023736,
      "learning_rate": 3.252643255403084e-06,
      "loss": 0.5069,
      "step": 12350
    },
    {
      "epoch": 0.6665047757811235,
      "grad_norm": 1.078594956025498,
      "learning_rate": 3.251982344617872e-06,
      "loss": 0.4386,
      "step": 12351
    },
    {
      "epoch": 0.666558739409638,
      "grad_norm": 1.2407835714824633,
      "learning_rate": 3.251321498442959e-06,
      "loss": 0.6154,
      "step": 12352
    },
    {
      "epoch": 0.6666127030381522,
      "grad_norm": 1.0439835417324956,
      "learning_rate": 3.2506607168973385e-06,
      "loss": 0.5407,
      "step": 12353
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.1629096088472213,
      "learning_rate": 3.2500000000000015e-06,
      "loss": 0.623,
      "step": 12354
    },
    {
      "epoch": 0.666720630295181,
      "grad_norm": 0.936880660740031,
      "learning_rate": 3.249339347769937e-06,
      "loss": 0.5012,
      "step": 12355
    },
    {
      "epoch": 0.6667745939236954,
      "grad_norm": 0.8823209558608377,
      "learning_rate": 3.248678760226135e-06,
      "loss": 0.4077,
      "step": 12356
    },
    {
      "epoch": 0.6668285575522098,
      "grad_norm": 1.037434378890483,
      "learning_rate": 3.248018237387579e-06,
      "loss": 0.4423,
      "step": 12357
    },
    {
      "epoch": 0.6668825211807242,
      "grad_norm": 0.9469526083012364,
      "learning_rate": 3.247357779273255e-06,
      "loss": 0.4474,
      "step": 12358
    },
    {
      "epoch": 0.6669364848092386,
      "grad_norm": 0.852142360381308,
      "learning_rate": 3.2466973859021446e-06,
      "loss": 0.3878,
      "step": 12359
    },
    {
      "epoch": 0.6669904484377529,
      "grad_norm": 1.1153423285679775,
      "learning_rate": 3.2460370572932292e-06,
      "loss": 0.499,
      "step": 12360
    },
    {
      "epoch": 0.6670444120662673,
      "grad_norm": 0.9520157046969236,
      "learning_rate": 3.245376793465485e-06,
      "loss": 0.5672,
      "step": 12361
    },
    {
      "epoch": 0.6670983756947817,
      "grad_norm": 1.173193303402801,
      "learning_rate": 3.24471659443789e-06,
      "loss": 0.5541,
      "step": 12362
    },
    {
      "epoch": 0.6671523393232961,
      "grad_norm": 1.1141271844721001,
      "learning_rate": 3.2440564602294185e-06,
      "loss": 0.3981,
      "step": 12363
    },
    {
      "epoch": 0.6672063029518105,
      "grad_norm": 0.7936050093678741,
      "learning_rate": 3.243396390859044e-06,
      "loss": 0.3551,
      "step": 12364
    },
    {
      "epoch": 0.6672602665803249,
      "grad_norm": 1.003281988979681,
      "learning_rate": 3.242736386345737e-06,
      "loss": 0.4323,
      "step": 12365
    },
    {
      "epoch": 0.6673142302088393,
      "grad_norm": 0.9956373957679652,
      "learning_rate": 3.2420764467084674e-06,
      "loss": 0.318,
      "step": 12366
    },
    {
      "epoch": 0.6673681938373536,
      "grad_norm": 0.9899000592190348,
      "learning_rate": 3.241416571966204e-06,
      "loss": 0.5113,
      "step": 12367
    },
    {
      "epoch": 0.667422157465868,
      "grad_norm": 0.932198131396753,
      "learning_rate": 3.240756762137909e-06,
      "loss": 0.3277,
      "step": 12368
    },
    {
      "epoch": 0.6674761210943824,
      "grad_norm": 1.1901650500322785,
      "learning_rate": 3.2400970172425472e-06,
      "loss": 0.5311,
      "step": 12369
    },
    {
      "epoch": 0.6675300847228968,
      "grad_norm": 1.0037583520224604,
      "learning_rate": 3.2394373372990806e-06,
      "loss": 0.4447,
      "step": 12370
    },
    {
      "epoch": 0.6675840483514112,
      "grad_norm": 0.8120452855265773,
      "learning_rate": 3.238777722326468e-06,
      "loss": 0.2604,
      "step": 12371
    },
    {
      "epoch": 0.6676380119799256,
      "grad_norm": 1.1063214074375392,
      "learning_rate": 3.2381181723436707e-06,
      "loss": 0.6125,
      "step": 12372
    },
    {
      "epoch": 0.66769197560844,
      "grad_norm": 1.043449808974125,
      "learning_rate": 3.2374586873696406e-06,
      "loss": 0.5612,
      "step": 12373
    },
    {
      "epoch": 0.6677459392369542,
      "grad_norm": 0.9424853978419362,
      "learning_rate": 3.236799267423336e-06,
      "loss": 0.4018,
      "step": 12374
    },
    {
      "epoch": 0.6677999028654686,
      "grad_norm": 1.3523067849054293,
      "learning_rate": 3.2361399125237062e-06,
      "loss": 0.5797,
      "step": 12375
    },
    {
      "epoch": 0.667853866493983,
      "grad_norm": 1.116793307043734,
      "learning_rate": 3.235480622689703e-06,
      "loss": 0.5967,
      "step": 12376
    },
    {
      "epoch": 0.6679078301224974,
      "grad_norm": 1.0768017663188467,
      "learning_rate": 3.2348213979402743e-06,
      "loss": 0.4332,
      "step": 12377
    },
    {
      "epoch": 0.6679617937510118,
      "grad_norm": 1.0053062289491403,
      "learning_rate": 3.234162238294367e-06,
      "loss": 0.4378,
      "step": 12378
    },
    {
      "epoch": 0.6680157573795262,
      "grad_norm": 0.9628041666990697,
      "learning_rate": 3.2335031437709267e-06,
      "loss": 0.5465,
      "step": 12379
    },
    {
      "epoch": 0.6680697210080406,
      "grad_norm": 0.8611795754559531,
      "learning_rate": 3.2328441143888965e-06,
      "loss": 0.329,
      "step": 12380
    },
    {
      "epoch": 0.6681236846365549,
      "grad_norm": 1.0501431251896747,
      "learning_rate": 3.2321851501672173e-06,
      "loss": 0.466,
      "step": 12381
    },
    {
      "epoch": 0.6681776482650693,
      "grad_norm": 0.9775546361625078,
      "learning_rate": 3.231526251124828e-06,
      "loss": 0.5297,
      "step": 12382
    },
    {
      "epoch": 0.6682316118935837,
      "grad_norm": 1.0058615389391146,
      "learning_rate": 3.2308674172806666e-06,
      "loss": 0.4203,
      "step": 12383
    },
    {
      "epoch": 0.6682855755220981,
      "grad_norm": 1.357306495521644,
      "learning_rate": 3.230208648653668e-06,
      "loss": 0.6187,
      "step": 12384
    },
    {
      "epoch": 0.6683395391506125,
      "grad_norm": 0.904370164125031,
      "learning_rate": 3.2295499452627665e-06,
      "loss": 0.3995,
      "step": 12385
    },
    {
      "epoch": 0.6683935027791269,
      "grad_norm": 0.7658731597397775,
      "learning_rate": 3.2288913071268945e-06,
      "loss": 0.3277,
      "step": 12386
    },
    {
      "epoch": 0.6684474664076413,
      "grad_norm": 1.0038712937876297,
      "learning_rate": 3.22823273426498e-06,
      "loss": 0.4448,
      "step": 12387
    },
    {
      "epoch": 0.6685014300361556,
      "grad_norm": 1.098301133661135,
      "learning_rate": 3.227574226695954e-06,
      "loss": 0.4787,
      "step": 12388
    },
    {
      "epoch": 0.66855539366467,
      "grad_norm": 1.0980722861209757,
      "learning_rate": 3.2269157844387394e-06,
      "loss": 0.4573,
      "step": 12389
    },
    {
      "epoch": 0.6686093572931844,
      "grad_norm": 0.8850143425657481,
      "learning_rate": 3.226257407512262e-06,
      "loss": 0.3725,
      "step": 12390
    },
    {
      "epoch": 0.6686633209216988,
      "grad_norm": 0.9754567345400976,
      "learning_rate": 3.2255990959354444e-06,
      "loss": 0.6684,
      "step": 12391
    },
    {
      "epoch": 0.6687172845502132,
      "grad_norm": 0.9759492136339822,
      "learning_rate": 3.2249408497272067e-06,
      "loss": 0.3893,
      "step": 12392
    },
    {
      "epoch": 0.6687712481787276,
      "grad_norm": 1.1703930849686768,
      "learning_rate": 3.2242826689064676e-06,
      "loss": 0.5033,
      "step": 12393
    },
    {
      "epoch": 0.668825211807242,
      "grad_norm": 1.1181579206684296,
      "learning_rate": 3.223624553492144e-06,
      "loss": 0.4967,
      "step": 12394
    },
    {
      "epoch": 0.6688791754357563,
      "grad_norm": 1.0791806859611004,
      "learning_rate": 3.2229665035031533e-06,
      "loss": 0.5056,
      "step": 12395
    },
    {
      "epoch": 0.6689331390642707,
      "grad_norm": 0.9122266460404179,
      "learning_rate": 3.2223085189584036e-06,
      "loss": 0.3611,
      "step": 12396
    },
    {
      "epoch": 0.668987102692785,
      "grad_norm": 0.873889389224243,
      "learning_rate": 3.2216505998768083e-06,
      "loss": 0.2659,
      "step": 12397
    },
    {
      "epoch": 0.6690410663212994,
      "grad_norm": 0.7465301798098738,
      "learning_rate": 3.2209927462772774e-06,
      "loss": 0.3581,
      "step": 12398
    },
    {
      "epoch": 0.6690950299498138,
      "grad_norm": 1.072929820076802,
      "learning_rate": 3.2203349581787175e-06,
      "loss": 0.3616,
      "step": 12399
    },
    {
      "epoch": 0.6691489935783282,
      "grad_norm": 1.2073481417732908,
      "learning_rate": 3.219677235600034e-06,
      "loss": 0.5546,
      "step": 12400
    },
    {
      "epoch": 0.6692029572068426,
      "grad_norm": 1.0435399179487563,
      "learning_rate": 3.219019578560132e-06,
      "loss": 0.3519,
      "step": 12401
    },
    {
      "epoch": 0.6692569208353569,
      "grad_norm": 1.0598047841911067,
      "learning_rate": 3.2183619870779125e-06,
      "loss": 0.6494,
      "step": 12402
    },
    {
      "epoch": 0.6693108844638713,
      "grad_norm": 0.7997497311296136,
      "learning_rate": 3.217704461172274e-06,
      "loss": 0.3476,
      "step": 12403
    },
    {
      "epoch": 0.6693648480923857,
      "grad_norm": 1.0992737888719173,
      "learning_rate": 3.2170470008621145e-06,
      "loss": 0.5185,
      "step": 12404
    },
    {
      "epoch": 0.6694188117209001,
      "grad_norm": 0.9501856939987475,
      "learning_rate": 3.216389606166332e-06,
      "loss": 0.3849,
      "step": 12405
    },
    {
      "epoch": 0.6694727753494145,
      "grad_norm": 0.7045781394764885,
      "learning_rate": 3.215732277103819e-06,
      "loss": 0.252,
      "step": 12406
    },
    {
      "epoch": 0.6695267389779289,
      "grad_norm": 1.0382597754865721,
      "learning_rate": 3.2150750136934687e-06,
      "loss": 0.3968,
      "step": 12407
    },
    {
      "epoch": 0.6695807026064433,
      "grad_norm": 1.038354696335282,
      "learning_rate": 3.2144178159541716e-06,
      "loss": 0.5212,
      "step": 12408
    },
    {
      "epoch": 0.6696346662349576,
      "grad_norm": 1.0738892607190234,
      "learning_rate": 3.2137606839048174e-06,
      "loss": 0.5239,
      "step": 12409
    },
    {
      "epoch": 0.669688629863472,
      "grad_norm": 1.2663652441750548,
      "learning_rate": 3.2131036175642884e-06,
      "loss": 0.5935,
      "step": 12410
    },
    {
      "epoch": 0.6697425934919864,
      "grad_norm": 0.9570454757476815,
      "learning_rate": 3.2124466169514734e-06,
      "loss": 0.4188,
      "step": 12411
    },
    {
      "epoch": 0.6697965571205008,
      "grad_norm": 1.0159692430604688,
      "learning_rate": 3.2117896820852534e-06,
      "loss": 0.4211,
      "step": 12412
    },
    {
      "epoch": 0.6698505207490152,
      "grad_norm": 1.0113274848904037,
      "learning_rate": 3.2111328129845103e-06,
      "loss": 0.3937,
      "step": 12413
    },
    {
      "epoch": 0.6699044843775296,
      "grad_norm": 1.055042077771013,
      "learning_rate": 3.210476009668123e-06,
      "loss": 0.5291,
      "step": 12414
    },
    {
      "epoch": 0.6699584480060439,
      "grad_norm": 1.1283024921285176,
      "learning_rate": 3.2098192721549704e-06,
      "loss": 0.5489,
      "step": 12415
    },
    {
      "epoch": 0.6700124116345583,
      "grad_norm": 1.2015417853271846,
      "learning_rate": 3.2091626004639237e-06,
      "loss": 0.5348,
      "step": 12416
    },
    {
      "epoch": 0.6700663752630727,
      "grad_norm": 1.11953106961071,
      "learning_rate": 3.208505994613859e-06,
      "loss": 0.5668,
      "step": 12417
    },
    {
      "epoch": 0.6701203388915871,
      "grad_norm": 0.9880332969908763,
      "learning_rate": 3.207849454623647e-06,
      "loss": 0.4615,
      "step": 12418
    },
    {
      "epoch": 0.6701743025201015,
      "grad_norm": 0.8284850153408269,
      "learning_rate": 3.207192980512159e-06,
      "loss": 0.3403,
      "step": 12419
    },
    {
      "epoch": 0.6702282661486159,
      "grad_norm": 1.0439980346025834,
      "learning_rate": 3.20653657229826e-06,
      "loss": 0.5373,
      "step": 12420
    },
    {
      "epoch": 0.6702822297771303,
      "grad_norm": 1.3162178922125916,
      "learning_rate": 3.205880230000818e-06,
      "loss": 0.6363,
      "step": 12421
    },
    {
      "epoch": 0.6703361934056445,
      "grad_norm": 0.995620564904958,
      "learning_rate": 3.2052239536386974e-06,
      "loss": 0.4452,
      "step": 12422
    },
    {
      "epoch": 0.6703901570341589,
      "grad_norm": 1.0322492076761298,
      "learning_rate": 3.2045677432307582e-06,
      "loss": 0.4927,
      "step": 12423
    },
    {
      "epoch": 0.6704441206626733,
      "grad_norm": 1.3726583166210202,
      "learning_rate": 3.20391159879586e-06,
      "loss": 0.5544,
      "step": 12424
    },
    {
      "epoch": 0.6704980842911877,
      "grad_norm": 1.2022952020588167,
      "learning_rate": 3.2032555203528637e-06,
      "loss": 0.5544,
      "step": 12425
    },
    {
      "epoch": 0.6705520479197021,
      "grad_norm": 1.1149517435214429,
      "learning_rate": 3.202599507920624e-06,
      "loss": 0.4547,
      "step": 12426
    },
    {
      "epoch": 0.6706060115482165,
      "grad_norm": 1.1178131258619826,
      "learning_rate": 3.201943561517996e-06,
      "loss": 0.4186,
      "step": 12427
    },
    {
      "epoch": 0.6706599751767309,
      "grad_norm": 1.0663680378726743,
      "learning_rate": 3.2012876811638325e-06,
      "loss": 0.5685,
      "step": 12428
    },
    {
      "epoch": 0.6707139388052452,
      "grad_norm": 0.8602105913374478,
      "learning_rate": 3.2006318668769833e-06,
      "loss": 0.382,
      "step": 12429
    },
    {
      "epoch": 0.6707679024337596,
      "grad_norm": 0.9097605924457389,
      "learning_rate": 3.1999761186762968e-06,
      "loss": 0.4184,
      "step": 12430
    },
    {
      "epoch": 0.670821866062274,
      "grad_norm": 1.293219107223365,
      "learning_rate": 3.199320436580622e-06,
      "loss": 0.4675,
      "step": 12431
    },
    {
      "epoch": 0.6708758296907884,
      "grad_norm": 1.0250787071955714,
      "learning_rate": 3.198664820608801e-06,
      "loss": 0.4823,
      "step": 12432
    },
    {
      "epoch": 0.6709297933193028,
      "grad_norm": 1.4885123468534653,
      "learning_rate": 3.1980092707796795e-06,
      "loss": 0.6486,
      "step": 12433
    },
    {
      "epoch": 0.6709837569478172,
      "grad_norm": 0.9703350876185489,
      "learning_rate": 3.1973537871120975e-06,
      "loss": 0.5426,
      "step": 12434
    },
    {
      "epoch": 0.6710377205763316,
      "grad_norm": 1.660026920460805,
      "learning_rate": 3.196698369624896e-06,
      "loss": 0.5425,
      "step": 12435
    },
    {
      "epoch": 0.6710916842048459,
      "grad_norm": 1.3465265811464069,
      "learning_rate": 3.1960430183369086e-06,
      "loss": 0.7166,
      "step": 12436
    },
    {
      "epoch": 0.6711456478333603,
      "grad_norm": 1.0216537302216104,
      "learning_rate": 3.195387733266973e-06,
      "loss": 0.4944,
      "step": 12437
    },
    {
      "epoch": 0.6711996114618747,
      "grad_norm": 1.0928065064463952,
      "learning_rate": 3.1947325144339224e-06,
      "loss": 0.4976,
      "step": 12438
    },
    {
      "epoch": 0.6712535750903891,
      "grad_norm": 0.8220869152026234,
      "learning_rate": 3.194077361856589e-06,
      "loss": 0.307,
      "step": 12439
    },
    {
      "epoch": 0.6713075387189035,
      "grad_norm": 1.0947173590911028,
      "learning_rate": 3.1934222755538026e-06,
      "loss": 0.6814,
      "step": 12440
    },
    {
      "epoch": 0.6713615023474179,
      "grad_norm": 0.8437172096917992,
      "learning_rate": 3.19276725554439e-06,
      "loss": 0.3311,
      "step": 12441
    },
    {
      "epoch": 0.6714154659759323,
      "grad_norm": 0.920484881339541,
      "learning_rate": 3.1921123018471777e-06,
      "loss": 0.4069,
      "step": 12442
    },
    {
      "epoch": 0.6714694296044466,
      "grad_norm": 0.9815091229544484,
      "learning_rate": 3.1914574144809916e-06,
      "loss": 0.5763,
      "step": 12443
    },
    {
      "epoch": 0.671523393232961,
      "grad_norm": 1.2478437619276084,
      "learning_rate": 3.19080259346465e-06,
      "loss": 0.4371,
      "step": 12444
    },
    {
      "epoch": 0.6715773568614753,
      "grad_norm": 1.0432916696216878,
      "learning_rate": 3.1901478388169754e-06,
      "loss": 0.4391,
      "step": 12445
    },
    {
      "epoch": 0.6716313204899897,
      "grad_norm": 1.0707674860605227,
      "learning_rate": 3.189493150556786e-06,
      "loss": 0.4256,
      "step": 12446
    },
    {
      "epoch": 0.6716852841185041,
      "grad_norm": 1.080439289176709,
      "learning_rate": 3.1888385287028983e-06,
      "loss": 0.3821,
      "step": 12447
    },
    {
      "epoch": 0.6717392477470185,
      "grad_norm": 1.2384454951813317,
      "learning_rate": 3.188183973274126e-06,
      "loss": 0.5385,
      "step": 12448
    },
    {
      "epoch": 0.6717932113755329,
      "grad_norm": 0.9833797260594502,
      "learning_rate": 3.1875294842892823e-06,
      "loss": 0.3995,
      "step": 12449
    },
    {
      "epoch": 0.6718471750040472,
      "grad_norm": 0.866106399746397,
      "learning_rate": 3.1868750617671793e-06,
      "loss": 0.3224,
      "step": 12450
    },
    {
      "epoch": 0.6719011386325616,
      "grad_norm": 0.8393215875288713,
      "learning_rate": 3.1862207057266235e-06,
      "loss": 0.3459,
      "step": 12451
    },
    {
      "epoch": 0.671955102261076,
      "grad_norm": 0.8941907409047861,
      "learning_rate": 3.185566416186422e-06,
      "loss": 0.3054,
      "step": 12452
    },
    {
      "epoch": 0.6720090658895904,
      "grad_norm": 1.1537474588429242,
      "learning_rate": 3.1849121931653797e-06,
      "loss": 0.3669,
      "step": 12453
    },
    {
      "epoch": 0.6720630295181048,
      "grad_norm": 0.9469888563771627,
      "learning_rate": 3.1842580366823007e-06,
      "loss": 0.3626,
      "step": 12454
    },
    {
      "epoch": 0.6721169931466192,
      "grad_norm": 0.889708163693672,
      "learning_rate": 3.183603946755986e-06,
      "loss": 0.4095,
      "step": 12455
    },
    {
      "epoch": 0.6721709567751336,
      "grad_norm": 1.09812605875971,
      "learning_rate": 3.182949923405234e-06,
      "loss": 0.6064,
      "step": 12456
    },
    {
      "epoch": 0.6722249204036479,
      "grad_norm": 0.8955850993905421,
      "learning_rate": 3.182295966648844e-06,
      "loss": 0.3595,
      "step": 12457
    },
    {
      "epoch": 0.6722788840321623,
      "grad_norm": 1.0510632759448908,
      "learning_rate": 3.181642076505608e-06,
      "loss": 0.4565,
      "step": 12458
    },
    {
      "epoch": 0.6723328476606767,
      "grad_norm": 0.8492451313010254,
      "learning_rate": 3.1809882529943212e-06,
      "loss": 0.3064,
      "step": 12459
    },
    {
      "epoch": 0.6723868112891911,
      "grad_norm": 0.7597745033783163,
      "learning_rate": 3.1803344961337758e-06,
      "loss": 0.2643,
      "step": 12460
    },
    {
      "epoch": 0.6724407749177055,
      "grad_norm": 1.1788212753411735,
      "learning_rate": 3.1796808059427608e-06,
      "loss": 0.509,
      "step": 12461
    },
    {
      "epoch": 0.6724947385462199,
      "grad_norm": 1.0467330988614503,
      "learning_rate": 3.179027182440064e-06,
      "loss": 0.4565,
      "step": 12462
    },
    {
      "epoch": 0.6725487021747343,
      "grad_norm": 1.0543890039873764,
      "learning_rate": 3.1783736256444704e-06,
      "loss": 0.4176,
      "step": 12463
    },
    {
      "epoch": 0.6726026658032486,
      "grad_norm": 1.1339557311912754,
      "learning_rate": 3.1777201355747665e-06,
      "loss": 0.4962,
      "step": 12464
    },
    {
      "epoch": 0.672656629431763,
      "grad_norm": 1.1245693257956215,
      "learning_rate": 3.177066712249731e-06,
      "loss": 0.6303,
      "step": 12465
    },
    {
      "epoch": 0.6727105930602774,
      "grad_norm": 1.2754580674934812,
      "learning_rate": 3.176413355688145e-06,
      "loss": 0.6977,
      "step": 12466
    },
    {
      "epoch": 0.6727645566887918,
      "grad_norm": 0.8595240895090033,
      "learning_rate": 3.175760065908787e-06,
      "loss": 0.3781,
      "step": 12467
    },
    {
      "epoch": 0.6728185203173062,
      "grad_norm": 1.1461152308701572,
      "learning_rate": 3.1751068429304343e-06,
      "loss": 0.389,
      "step": 12468
    },
    {
      "epoch": 0.6728724839458206,
      "grad_norm": 1.163073050189621,
      "learning_rate": 3.174453686771859e-06,
      "loss": 0.4906,
      "step": 12469
    },
    {
      "epoch": 0.672926447574335,
      "grad_norm": 1.0860347525842324,
      "learning_rate": 3.173800597451835e-06,
      "loss": 0.5056,
      "step": 12470
    },
    {
      "epoch": 0.6729804112028492,
      "grad_norm": 1.078296414646916,
      "learning_rate": 3.1731475749891335e-06,
      "loss": 0.4139,
      "step": 12471
    },
    {
      "epoch": 0.6730343748313636,
      "grad_norm": 0.8107043007334677,
      "learning_rate": 3.1724946194025204e-06,
      "loss": 0.3567,
      "step": 12472
    },
    {
      "epoch": 0.673088338459878,
      "grad_norm": 0.897884304199088,
      "learning_rate": 3.1718417307107634e-06,
      "loss": 0.465,
      "step": 12473
    },
    {
      "epoch": 0.6731423020883924,
      "grad_norm": 1.0158651440998,
      "learning_rate": 3.1711889089326276e-06,
      "loss": 0.4817,
      "step": 12474
    },
    {
      "epoch": 0.6731962657169068,
      "grad_norm": 1.0549593084984992,
      "learning_rate": 3.1705361540868764e-06,
      "loss": 0.4879,
      "step": 12475
    },
    {
      "epoch": 0.6732502293454212,
      "grad_norm": 0.993195361368971,
      "learning_rate": 3.1698834661922683e-06,
      "loss": 0.4459,
      "step": 12476
    },
    {
      "epoch": 0.6733041929739356,
      "grad_norm": 0.8335979854030285,
      "learning_rate": 3.169230845267565e-06,
      "loss": 0.4411,
      "step": 12477
    },
    {
      "epoch": 0.6733581566024499,
      "grad_norm": 0.870199727372066,
      "learning_rate": 3.168578291331522e-06,
      "loss": 0.4315,
      "step": 12478
    },
    {
      "epoch": 0.6734121202309643,
      "grad_norm": 1.0512288834062855,
      "learning_rate": 3.1679258044028965e-06,
      "loss": 0.4696,
      "step": 12479
    },
    {
      "epoch": 0.6734660838594787,
      "grad_norm": 0.8923534038725774,
      "learning_rate": 3.167273384500438e-06,
      "loss": 0.3467,
      "step": 12480
    },
    {
      "epoch": 0.6735200474879931,
      "grad_norm": 1.1233162904969056,
      "learning_rate": 3.1666210316428993e-06,
      "loss": 0.6298,
      "step": 12481
    },
    {
      "epoch": 0.6735740111165075,
      "grad_norm": 0.9751493200313842,
      "learning_rate": 3.1659687458490297e-06,
      "loss": 0.5559,
      "step": 12482
    },
    {
      "epoch": 0.6736279747450219,
      "grad_norm": 1.157179883579112,
      "learning_rate": 3.1653165271375776e-06,
      "loss": 0.4943,
      "step": 12483
    },
    {
      "epoch": 0.6736819383735362,
      "grad_norm": 1.125575593575788,
      "learning_rate": 3.1646643755272865e-06,
      "loss": 0.4879,
      "step": 12484
    },
    {
      "epoch": 0.6737359020020506,
      "grad_norm": 1.0944689511411934,
      "learning_rate": 3.164012291036901e-06,
      "loss": 0.4394,
      "step": 12485
    },
    {
      "epoch": 0.673789865630565,
      "grad_norm": 0.9909343957952464,
      "learning_rate": 3.163360273685163e-06,
      "loss": 0.4952,
      "step": 12486
    },
    {
      "epoch": 0.6738438292590794,
      "grad_norm": 1.248966102323525,
      "learning_rate": 3.162708323490812e-06,
      "loss": 0.5298,
      "step": 12487
    },
    {
      "epoch": 0.6738977928875938,
      "grad_norm": 1.1199986622272946,
      "learning_rate": 3.1620564404725852e-06,
      "loss": 0.5807,
      "step": 12488
    },
    {
      "epoch": 0.6739517565161082,
      "grad_norm": 0.9444850163489719,
      "learning_rate": 3.1614046246492187e-06,
      "loss": 0.3991,
      "step": 12489
    },
    {
      "epoch": 0.6740057201446226,
      "grad_norm": 1.0321433874134422,
      "learning_rate": 3.1607528760394455e-06,
      "loss": 0.4975,
      "step": 12490
    },
    {
      "epoch": 0.6740596837731369,
      "grad_norm": 1.1727988344832683,
      "learning_rate": 3.1601011946620006e-06,
      "loss": 0.5668,
      "step": 12491
    },
    {
      "epoch": 0.6741136474016513,
      "grad_norm": 0.9172244776320095,
      "learning_rate": 3.1594495805356095e-06,
      "loss": 0.3187,
      "step": 12492
    },
    {
      "epoch": 0.6741676110301656,
      "grad_norm": 0.8357798770921838,
      "learning_rate": 3.1587980336790037e-06,
      "loss": 0.3116,
      "step": 12493
    },
    {
      "epoch": 0.67422157465868,
      "grad_norm": 1.3873139126779312,
      "learning_rate": 3.1581465541109067e-06,
      "loss": 0.5952,
      "step": 12494
    },
    {
      "epoch": 0.6742755382871944,
      "grad_norm": 0.9740829406838187,
      "learning_rate": 3.1574951418500445e-06,
      "loss": 0.3723,
      "step": 12495
    },
    {
      "epoch": 0.6743295019157088,
      "grad_norm": 0.7764021303769214,
      "learning_rate": 3.1568437969151395e-06,
      "loss": 0.3283,
      "step": 12496
    },
    {
      "epoch": 0.6743834655442232,
      "grad_norm": 1.20442334798481,
      "learning_rate": 3.1561925193249104e-06,
      "loss": 0.3889,
      "step": 12497
    },
    {
      "epoch": 0.6744374291727375,
      "grad_norm": 0.9956533322533199,
      "learning_rate": 3.1555413090980787e-06,
      "loss": 0.3711,
      "step": 12498
    },
    {
      "epoch": 0.6744913928012519,
      "grad_norm": 0.979760991540413,
      "learning_rate": 3.1548901662533558e-06,
      "loss": 0.426,
      "step": 12499
    },
    {
      "epoch": 0.6745453564297663,
      "grad_norm": 0.9452040631302806,
      "learning_rate": 3.1542390908094606e-06,
      "loss": 0.3867,
      "step": 12500
    },
    {
      "epoch": 0.6745453564297663,
      "eval_loss": 0.5368213653564453,
      "eval_runtime": 163.942,
      "eval_samples_per_second": 20.977,
      "eval_steps_per_second": 0.878,
      "step": 12500
    },
    {
      "epoch": 0.6745993200582807,
      "grad_norm": 0.9571429397203504,
      "learning_rate": 3.153588082785104e-06,
      "loss": 0.33,
      "step": 12501
    },
    {
      "epoch": 0.6746532836867951,
      "grad_norm": 1.1423131821487138,
      "learning_rate": 3.152937142198996e-06,
      "loss": 0.4871,
      "step": 12502
    },
    {
      "epoch": 0.6747072473153095,
      "grad_norm": 1.08110896798623,
      "learning_rate": 3.152286269069847e-06,
      "loss": 0.4805,
      "step": 12503
    },
    {
      "epoch": 0.6747612109438239,
      "grad_norm": 1.0894989170111733,
      "learning_rate": 3.1516354634163625e-06,
      "loss": 0.4522,
      "step": 12504
    },
    {
      "epoch": 0.6748151745723382,
      "grad_norm": 0.9350860762668982,
      "learning_rate": 3.150984725257249e-06,
      "loss": 0.3974,
      "step": 12505
    },
    {
      "epoch": 0.6748691382008526,
      "grad_norm": 0.9598369570975777,
      "learning_rate": 3.150334054611206e-06,
      "loss": 0.3548,
      "step": 12506
    },
    {
      "epoch": 0.674923101829367,
      "grad_norm": 0.8836477731598379,
      "learning_rate": 3.1496834514969368e-06,
      "loss": 0.3618,
      "step": 12507
    },
    {
      "epoch": 0.6749770654578814,
      "grad_norm": 1.225816907401292,
      "learning_rate": 3.14903291593314e-06,
      "loss": 0.479,
      "step": 12508
    },
    {
      "epoch": 0.6750310290863958,
      "grad_norm": 1.3303839739971224,
      "learning_rate": 3.1483824479385127e-06,
      "loss": 0.5311,
      "step": 12509
    },
    {
      "epoch": 0.6750849927149102,
      "grad_norm": 1.2909697991993074,
      "learning_rate": 3.1477320475317506e-06,
      "loss": 0.5832,
      "step": 12510
    },
    {
      "epoch": 0.6751389563434246,
      "grad_norm": 1.0475144942867909,
      "learning_rate": 3.1470817147315454e-06,
      "loss": 0.5019,
      "step": 12511
    },
    {
      "epoch": 0.6751929199719389,
      "grad_norm": 1.2939685275802708,
      "learning_rate": 3.1464314495565914e-06,
      "loss": 0.6392,
      "step": 12512
    },
    {
      "epoch": 0.6752468836004533,
      "grad_norm": 1.1392797448187244,
      "learning_rate": 3.145781252025574e-06,
      "loss": 0.5239,
      "step": 12513
    },
    {
      "epoch": 0.6753008472289677,
      "grad_norm": 0.96507513010418,
      "learning_rate": 3.145131122157181e-06,
      "loss": 0.4198,
      "step": 12514
    },
    {
      "epoch": 0.6753548108574821,
      "grad_norm": 0.9906796534579229,
      "learning_rate": 3.1444810599701e-06,
      "loss": 0.4341,
      "step": 12515
    },
    {
      "epoch": 0.6754087744859965,
      "grad_norm": 0.9930140483307325,
      "learning_rate": 3.143831065483014e-06,
      "loss": 0.5738,
      "step": 12516
    },
    {
      "epoch": 0.6754627381145109,
      "grad_norm": 0.8875725174695971,
      "learning_rate": 3.1431811387146025e-06,
      "loss": 0.4713,
      "step": 12517
    },
    {
      "epoch": 0.6755167017430252,
      "grad_norm": 0.8576127550501019,
      "learning_rate": 3.142531279683547e-06,
      "loss": 0.363,
      "step": 12518
    },
    {
      "epoch": 0.6755706653715395,
      "grad_norm": 0.890218055408893,
      "learning_rate": 3.141881488408527e-06,
      "loss": 0.424,
      "step": 12519
    },
    {
      "epoch": 0.6756246290000539,
      "grad_norm": 1.0451223104737808,
      "learning_rate": 3.1412317649082134e-06,
      "loss": 0.4112,
      "step": 12520
    },
    {
      "epoch": 0.6756785926285683,
      "grad_norm": 1.1477077281273154,
      "learning_rate": 3.1405821092012824e-06,
      "loss": 0.5033,
      "step": 12521
    },
    {
      "epoch": 0.6757325562570827,
      "grad_norm": 0.8520788070004383,
      "learning_rate": 3.139932521306405e-06,
      "loss": 0.3883,
      "step": 12522
    },
    {
      "epoch": 0.6757865198855971,
      "grad_norm": 0.9778952707083785,
      "learning_rate": 3.1392830012422525e-06,
      "loss": 0.4341,
      "step": 12523
    },
    {
      "epoch": 0.6758404835141115,
      "grad_norm": 1.0913159882219015,
      "learning_rate": 3.1386335490274917e-06,
      "loss": 0.472,
      "step": 12524
    },
    {
      "epoch": 0.6758944471426259,
      "grad_norm": 0.7403046814695414,
      "learning_rate": 3.1379841646807883e-06,
      "loss": 0.3477,
      "step": 12525
    },
    {
      "epoch": 0.6759484107711402,
      "grad_norm": 1.2280726501799701,
      "learning_rate": 3.137334848220809e-06,
      "loss": 0.7131,
      "step": 12526
    },
    {
      "epoch": 0.6760023743996546,
      "grad_norm": 1.01053090004462,
      "learning_rate": 3.1366855996662104e-06,
      "loss": 0.5446,
      "step": 12527
    },
    {
      "epoch": 0.676056338028169,
      "grad_norm": 0.8895239621192281,
      "learning_rate": 3.136036419035656e-06,
      "loss": 0.4047,
      "step": 12528
    },
    {
      "epoch": 0.6761103016566834,
      "grad_norm": 0.9421582641984142,
      "learning_rate": 3.135387306347805e-06,
      "loss": 0.4504,
      "step": 12529
    },
    {
      "epoch": 0.6761642652851978,
      "grad_norm": 0.9424334600471115,
      "learning_rate": 3.1347382616213114e-06,
      "loss": 0.4253,
      "step": 12530
    },
    {
      "epoch": 0.6762182289137122,
      "grad_norm": 1.0122859443487862,
      "learning_rate": 3.1340892848748295e-06,
      "loss": 0.3451,
      "step": 12531
    },
    {
      "epoch": 0.6762721925422266,
      "grad_norm": 1.081719914555442,
      "learning_rate": 3.133440376127012e-06,
      "loss": 0.4325,
      "step": 12532
    },
    {
      "epoch": 0.6763261561707409,
      "grad_norm": 1.060582630200462,
      "learning_rate": 3.13279153539651e-06,
      "loss": 0.3994,
      "step": 12533
    },
    {
      "epoch": 0.6763801197992553,
      "grad_norm": 0.7651549793746555,
      "learning_rate": 3.132142762701971e-06,
      "loss": 0.304,
      "step": 12534
    },
    {
      "epoch": 0.6764340834277697,
      "grad_norm": 0.9315051354136454,
      "learning_rate": 3.131494058062041e-06,
      "loss": 0.5469,
      "step": 12535
    },
    {
      "epoch": 0.6764880470562841,
      "grad_norm": 1.0124631383658982,
      "learning_rate": 3.1308454214953653e-06,
      "loss": 0.4943,
      "step": 12536
    },
    {
      "epoch": 0.6765420106847985,
      "grad_norm": 0.8894507696597683,
      "learning_rate": 3.1301968530205867e-06,
      "loss": 0.3459,
      "step": 12537
    },
    {
      "epoch": 0.6765959743133129,
      "grad_norm": 1.0980792733254192,
      "learning_rate": 3.1295483526563446e-06,
      "loss": 0.5039,
      "step": 12538
    },
    {
      "epoch": 0.6766499379418273,
      "grad_norm": 0.9661265480596312,
      "learning_rate": 3.1288999204212777e-06,
      "loss": 0.3887,
      "step": 12539
    },
    {
      "epoch": 0.6767039015703415,
      "grad_norm": 1.3479687321021294,
      "learning_rate": 3.1282515563340247e-06,
      "loss": 0.5613,
      "step": 12540
    },
    {
      "epoch": 0.676757865198856,
      "grad_norm": 0.9341726786516737,
      "learning_rate": 3.1276032604132173e-06,
      "loss": 0.4838,
      "step": 12541
    },
    {
      "epoch": 0.6768118288273703,
      "grad_norm": 1.0667217369329645,
      "learning_rate": 3.126955032677489e-06,
      "loss": 0.5301,
      "step": 12542
    },
    {
      "epoch": 0.6768657924558847,
      "grad_norm": 1.073598153320754,
      "learning_rate": 3.1263068731454706e-06,
      "loss": 0.4384,
      "step": 12543
    },
    {
      "epoch": 0.6769197560843991,
      "grad_norm": 0.8456914375092188,
      "learning_rate": 3.125658781835791e-06,
      "loss": 0.3173,
      "step": 12544
    },
    {
      "epoch": 0.6769737197129135,
      "grad_norm": 0.9173480500794219,
      "learning_rate": 3.1250107587670777e-06,
      "loss": 0.3714,
      "step": 12545
    },
    {
      "epoch": 0.6770276833414279,
      "grad_norm": 0.8170093633735841,
      "learning_rate": 3.1243628039579544e-06,
      "loss": 0.3311,
      "step": 12546
    },
    {
      "epoch": 0.6770816469699422,
      "grad_norm": 1.1167131910980208,
      "learning_rate": 3.1237149174270465e-06,
      "loss": 0.4908,
      "step": 12547
    },
    {
      "epoch": 0.6771356105984566,
      "grad_norm": 0.9377850786378774,
      "learning_rate": 3.123067099192971e-06,
      "loss": 0.4023,
      "step": 12548
    },
    {
      "epoch": 0.677189574226971,
      "grad_norm": 0.8707626135572417,
      "learning_rate": 3.1224193492743495e-06,
      "loss": 0.324,
      "step": 12549
    },
    {
      "epoch": 0.6772435378554854,
      "grad_norm": 0.8143062644943083,
      "learning_rate": 3.1217716676897975e-06,
      "loss": 0.3997,
      "step": 12550
    },
    {
      "epoch": 0.6772975014839998,
      "grad_norm": 1.0519723280185604,
      "learning_rate": 3.1211240544579313e-06,
      "loss": 0.5422,
      "step": 12551
    },
    {
      "epoch": 0.6773514651125142,
      "grad_norm": 1.0639368694121791,
      "learning_rate": 3.1204765095973626e-06,
      "loss": 0.3847,
      "step": 12552
    },
    {
      "epoch": 0.6774054287410285,
      "grad_norm": 1.1941453382283562,
      "learning_rate": 3.1198290331267055e-06,
      "loss": 0.576,
      "step": 12553
    },
    {
      "epoch": 0.6774593923695429,
      "grad_norm": 1.0791123478090205,
      "learning_rate": 3.119181625064566e-06,
      "loss": 0.621,
      "step": 12554
    },
    {
      "epoch": 0.6775133559980573,
      "grad_norm": 0.9976154236149422,
      "learning_rate": 3.118534285429552e-06,
      "loss": 0.5007,
      "step": 12555
    },
    {
      "epoch": 0.6775673196265717,
      "grad_norm": 1.0366372341834067,
      "learning_rate": 3.1178870142402685e-06,
      "loss": 0.4185,
      "step": 12556
    },
    {
      "epoch": 0.6776212832550861,
      "grad_norm": 1.1880258547576286,
      "learning_rate": 3.1172398115153193e-06,
      "loss": 0.4925,
      "step": 12557
    },
    {
      "epoch": 0.6776752468836005,
      "grad_norm": 1.024137049595589,
      "learning_rate": 3.1165926772733053e-06,
      "loss": 0.4286,
      "step": 12558
    },
    {
      "epoch": 0.6777292105121149,
      "grad_norm": 1.0928563951031838,
      "learning_rate": 3.1159456115328257e-06,
      "loss": 0.5829,
      "step": 12559
    },
    {
      "epoch": 0.6777831741406292,
      "grad_norm": 1.0392567848576464,
      "learning_rate": 3.115298614312481e-06,
      "loss": 0.4742,
      "step": 12560
    },
    {
      "epoch": 0.6778371377691436,
      "grad_norm": 1.0552367840127157,
      "learning_rate": 3.114651685630861e-06,
      "loss": 0.5318,
      "step": 12561
    },
    {
      "epoch": 0.677891101397658,
      "grad_norm": 0.7747160033788929,
      "learning_rate": 3.114004825506562e-06,
      "loss": 0.2809,
      "step": 12562
    },
    {
      "epoch": 0.6779450650261724,
      "grad_norm": 0.9880987788889704,
      "learning_rate": 3.113358033958175e-06,
      "loss": 0.425,
      "step": 12563
    },
    {
      "epoch": 0.6779990286546868,
      "grad_norm": 1.0970920204374452,
      "learning_rate": 3.1127113110042893e-06,
      "loss": 0.6359,
      "step": 12564
    },
    {
      "epoch": 0.6780529922832012,
      "grad_norm": 0.9371212775721022,
      "learning_rate": 3.112064656663494e-06,
      "loss": 0.4757,
      "step": 12565
    },
    {
      "epoch": 0.6781069559117155,
      "grad_norm": 0.9260124838912506,
      "learning_rate": 3.111418070954373e-06,
      "loss": 0.4819,
      "step": 12566
    },
    {
      "epoch": 0.6781609195402298,
      "grad_norm": 1.0386917603920927,
      "learning_rate": 3.1107715538955107e-06,
      "loss": 0.4319,
      "step": 12567
    },
    {
      "epoch": 0.6782148831687442,
      "grad_norm": 0.99242959717869,
      "learning_rate": 3.1101251055054877e-06,
      "loss": 0.4616,
      "step": 12568
    },
    {
      "epoch": 0.6782688467972586,
      "grad_norm": 1.248465738272871,
      "learning_rate": 3.1094787258028837e-06,
      "loss": 0.5013,
      "step": 12569
    },
    {
      "epoch": 0.678322810425773,
      "grad_norm": 1.1358692018014827,
      "learning_rate": 3.1088324148062753e-06,
      "loss": 0.4493,
      "step": 12570
    },
    {
      "epoch": 0.6783767740542874,
      "grad_norm": 1.0763794320356304,
      "learning_rate": 3.108186172534241e-06,
      "loss": 0.5234,
      "step": 12571
    },
    {
      "epoch": 0.6784307376828018,
      "grad_norm": 1.1138566523636388,
      "learning_rate": 3.1075399990053522e-06,
      "loss": 0.5141,
      "step": 12572
    },
    {
      "epoch": 0.6784847013113162,
      "grad_norm": 1.0418303039442804,
      "learning_rate": 3.106893894238181e-06,
      "loss": 0.4881,
      "step": 12573
    },
    {
      "epoch": 0.6785386649398305,
      "grad_norm": 0.9645644695915627,
      "learning_rate": 3.1062478582512988e-06,
      "loss": 0.3944,
      "step": 12574
    },
    {
      "epoch": 0.6785926285683449,
      "grad_norm": 0.9684710964307409,
      "learning_rate": 3.1056018910632705e-06,
      "loss": 0.4344,
      "step": 12575
    },
    {
      "epoch": 0.6786465921968593,
      "grad_norm": 1.0681043327414068,
      "learning_rate": 3.1049559926926633e-06,
      "loss": 0.4651,
      "step": 12576
    },
    {
      "epoch": 0.6787005558253737,
      "grad_norm": 0.8059890754065552,
      "learning_rate": 3.1043101631580413e-06,
      "loss": 0.311,
      "step": 12577
    },
    {
      "epoch": 0.6787545194538881,
      "grad_norm": 1.1422558084269914,
      "learning_rate": 3.1036644024779657e-06,
      "loss": 0.5071,
      "step": 12578
    },
    {
      "epoch": 0.6788084830824025,
      "grad_norm": 0.7248523190672969,
      "learning_rate": 3.1030187106709963e-06,
      "loss": 0.2697,
      "step": 12579
    },
    {
      "epoch": 0.6788624467109169,
      "grad_norm": 0.9508927483904276,
      "learning_rate": 3.1023730877556906e-06,
      "loss": 0.4491,
      "step": 12580
    },
    {
      "epoch": 0.6789164103394312,
      "grad_norm": 1.1222046245409567,
      "learning_rate": 3.101727533750606e-06,
      "loss": 0.61,
      "step": 12581
    },
    {
      "epoch": 0.6789703739679456,
      "grad_norm": 0.9919619243683457,
      "learning_rate": 3.1010820486742945e-06,
      "loss": 0.4105,
      "step": 12582
    },
    {
      "epoch": 0.67902433759646,
      "grad_norm": 0.9274774804854894,
      "learning_rate": 3.1004366325453095e-06,
      "loss": 0.3426,
      "step": 12583
    },
    {
      "epoch": 0.6790783012249744,
      "grad_norm": 0.8582554478788331,
      "learning_rate": 3.0997912853822e-06,
      "loss": 0.286,
      "step": 12584
    },
    {
      "epoch": 0.6791322648534888,
      "grad_norm": 1.1587799698121426,
      "learning_rate": 3.099146007203515e-06,
      "loss": 0.4292,
      "step": 12585
    },
    {
      "epoch": 0.6791862284820032,
      "grad_norm": 0.925321100601125,
      "learning_rate": 3.0985007980277994e-06,
      "loss": 0.4032,
      "step": 12586
    },
    {
      "epoch": 0.6792401921105176,
      "grad_norm": 1.2917848724780916,
      "learning_rate": 3.0978556578735963e-06,
      "loss": 0.5848,
      "step": 12587
    },
    {
      "epoch": 0.6792941557390318,
      "grad_norm": 0.9207796303396527,
      "learning_rate": 3.097210586759449e-06,
      "loss": 0.4291,
      "step": 12588
    },
    {
      "epoch": 0.6793481193675462,
      "grad_norm": 0.9492494260022888,
      "learning_rate": 3.0965655847038977e-06,
      "loss": 0.3693,
      "step": 12589
    },
    {
      "epoch": 0.6794020829960606,
      "grad_norm": 1.1477568449370277,
      "learning_rate": 3.0959206517254803e-06,
      "loss": 0.5119,
      "step": 12590
    },
    {
      "epoch": 0.679456046624575,
      "grad_norm": 1.1582127077926427,
      "learning_rate": 3.095275787842732e-06,
      "loss": 0.5233,
      "step": 12591
    },
    {
      "epoch": 0.6795100102530894,
      "grad_norm": 1.1035162940918315,
      "learning_rate": 3.0946309930741875e-06,
      "loss": 0.4516,
      "step": 12592
    },
    {
      "epoch": 0.6795639738816038,
      "grad_norm": 0.8463918172694993,
      "learning_rate": 3.0939862674383787e-06,
      "loss": 0.3777,
      "step": 12593
    },
    {
      "epoch": 0.6796179375101182,
      "grad_norm": 0.9928689531126598,
      "learning_rate": 3.0933416109538354e-06,
      "loss": 0.391,
      "step": 12594
    },
    {
      "epoch": 0.6796719011386325,
      "grad_norm": 1.0813041886848058,
      "learning_rate": 3.0926970236390876e-06,
      "loss": 0.4321,
      "step": 12595
    },
    {
      "epoch": 0.6797258647671469,
      "grad_norm": 1.116584632274721,
      "learning_rate": 3.092052505512658e-06,
      "loss": 0.5157,
      "step": 12596
    },
    {
      "epoch": 0.6797798283956613,
      "grad_norm": 1.2576092010249982,
      "learning_rate": 3.091408056593073e-06,
      "loss": 0.44,
      "step": 12597
    },
    {
      "epoch": 0.6798337920241757,
      "grad_norm": 1.1953669672963545,
      "learning_rate": 3.090763676898854e-06,
      "loss": 0.464,
      "step": 12598
    },
    {
      "epoch": 0.6798877556526901,
      "grad_norm": 0.9471987328848365,
      "learning_rate": 3.0901193664485213e-06,
      "loss": 0.3399,
      "step": 12599
    },
    {
      "epoch": 0.6799417192812045,
      "grad_norm": 1.0273246911459386,
      "learning_rate": 3.0894751252605928e-06,
      "loss": 0.481,
      "step": 12600
    },
    {
      "epoch": 0.6799956829097189,
      "grad_norm": 1.1362921745098056,
      "learning_rate": 3.0888309533535853e-06,
      "loss": 0.6258,
      "step": 12601
    },
    {
      "epoch": 0.6800496465382332,
      "grad_norm": 0.9993341642174376,
      "learning_rate": 3.088186850746014e-06,
      "loss": 0.4631,
      "step": 12602
    },
    {
      "epoch": 0.6801036101667476,
      "grad_norm": 0.9266952240555277,
      "learning_rate": 3.087542817456387e-06,
      "loss": 0.3899,
      "step": 12603
    },
    {
      "epoch": 0.680157573795262,
      "grad_norm": 0.9894636206510858,
      "learning_rate": 3.086898853503218e-06,
      "loss": 0.4669,
      "step": 12604
    },
    {
      "epoch": 0.6802115374237764,
      "grad_norm": 0.8560918570141612,
      "learning_rate": 3.086254958905014e-06,
      "loss": 0.4453,
      "step": 12605
    },
    {
      "epoch": 0.6802655010522908,
      "grad_norm": 0.9357941289161964,
      "learning_rate": 3.085611133680282e-06,
      "loss": 0.4302,
      "step": 12606
    },
    {
      "epoch": 0.6803194646808052,
      "grad_norm": 1.0849546390734923,
      "learning_rate": 3.084967377847524e-06,
      "loss": 0.4785,
      "step": 12607
    },
    {
      "epoch": 0.6803734283093196,
      "grad_norm": 1.0720492753542559,
      "learning_rate": 3.084323691425245e-06,
      "loss": 0.5944,
      "step": 12608
    },
    {
      "epoch": 0.6804273919378339,
      "grad_norm": 0.8163589899177673,
      "learning_rate": 3.0836800744319454e-06,
      "loss": 0.3869,
      "step": 12609
    },
    {
      "epoch": 0.6804813555663483,
      "grad_norm": 1.008806291373415,
      "learning_rate": 3.0830365268861202e-06,
      "loss": 0.5182,
      "step": 12610
    },
    {
      "epoch": 0.6805353191948627,
      "grad_norm": 1.1409125918027845,
      "learning_rate": 3.082393048806267e-06,
      "loss": 0.5107,
      "step": 12611
    },
    {
      "epoch": 0.680589282823377,
      "grad_norm": 1.0611748008137616,
      "learning_rate": 3.0817496402108817e-06,
      "loss": 0.4543,
      "step": 12612
    },
    {
      "epoch": 0.6806432464518914,
      "grad_norm": 0.9054356115197474,
      "learning_rate": 3.081106301118454e-06,
      "loss": 0.3781,
      "step": 12613
    },
    {
      "epoch": 0.6806972100804058,
      "grad_norm": 0.924990267829609,
      "learning_rate": 3.0804630315474757e-06,
      "loss": 0.496,
      "step": 12614
    },
    {
      "epoch": 0.6807511737089202,
      "grad_norm": 1.130146282650754,
      "learning_rate": 3.0798198315164345e-06,
      "loss": 0.5075,
      "step": 12615
    },
    {
      "epoch": 0.6808051373374345,
      "grad_norm": 1.0059843116703926,
      "learning_rate": 3.079176701043819e-06,
      "loss": 0.4208,
      "step": 12616
    },
    {
      "epoch": 0.6808591009659489,
      "grad_norm": 1.0375302024428765,
      "learning_rate": 3.0785336401481087e-06,
      "loss": 0.441,
      "step": 12617
    },
    {
      "epoch": 0.6809130645944633,
      "grad_norm": 1.1410768643435854,
      "learning_rate": 3.0778906488477882e-06,
      "loss": 0.4156,
      "step": 12618
    },
    {
      "epoch": 0.6809670282229777,
      "grad_norm": 1.093732200671502,
      "learning_rate": 3.077247727161339e-06,
      "loss": 0.4273,
      "step": 12619
    },
    {
      "epoch": 0.6810209918514921,
      "grad_norm": 1.1662700660195735,
      "learning_rate": 3.076604875107237e-06,
      "loss": 0.5387,
      "step": 12620
    },
    {
      "epoch": 0.6810749554800065,
      "grad_norm": 0.9265317908262547,
      "learning_rate": 3.07596209270396e-06,
      "loss": 0.4331,
      "step": 12621
    },
    {
      "epoch": 0.6811289191085208,
      "grad_norm": 0.9373421531566584,
      "learning_rate": 3.0753193799699833e-06,
      "loss": 0.5373,
      "step": 12622
    },
    {
      "epoch": 0.6811828827370352,
      "grad_norm": 0.8886804577128159,
      "learning_rate": 3.0746767369237757e-06,
      "loss": 0.3398,
      "step": 12623
    },
    {
      "epoch": 0.6812368463655496,
      "grad_norm": 1.0001045304277278,
      "learning_rate": 3.07403416358381e-06,
      "loss": 0.4254,
      "step": 12624
    },
    {
      "epoch": 0.681290809994064,
      "grad_norm": 0.9166374403919453,
      "learning_rate": 3.073391659968553e-06,
      "loss": 0.3886,
      "step": 12625
    },
    {
      "epoch": 0.6813447736225784,
      "grad_norm": 0.9780711736921542,
      "learning_rate": 3.072749226096472e-06,
      "loss": 0.4457,
      "step": 12626
    },
    {
      "epoch": 0.6813987372510928,
      "grad_norm": 1.1086685930467592,
      "learning_rate": 3.0721068619860306e-06,
      "loss": 0.6839,
      "step": 12627
    },
    {
      "epoch": 0.6814527008796072,
      "grad_norm": 1.0797882875135867,
      "learning_rate": 3.071464567655691e-06,
      "loss": 0.4892,
      "step": 12628
    },
    {
      "epoch": 0.6815066645081215,
      "grad_norm": 1.2974204110442549,
      "learning_rate": 3.0708223431239156e-06,
      "loss": 0.505,
      "step": 12629
    },
    {
      "epoch": 0.6815606281366359,
      "grad_norm": 1.0753471892170139,
      "learning_rate": 3.070180188409159e-06,
      "loss": 0.5295,
      "step": 12630
    },
    {
      "epoch": 0.6816145917651503,
      "grad_norm": 1.276063980898254,
      "learning_rate": 3.0695381035298787e-06,
      "loss": 0.5461,
      "step": 12631
    },
    {
      "epoch": 0.6816685553936647,
      "grad_norm": 0.9164829886378051,
      "learning_rate": 3.0688960885045294e-06,
      "loss": 0.4836,
      "step": 12632
    },
    {
      "epoch": 0.6817225190221791,
      "grad_norm": 1.0884192625062559,
      "learning_rate": 3.0682541433515624e-06,
      "loss": 0.4604,
      "step": 12633
    },
    {
      "epoch": 0.6817764826506935,
      "grad_norm": 0.9917088483247262,
      "learning_rate": 3.067612268089429e-06,
      "loss": 0.452,
      "step": 12634
    },
    {
      "epoch": 0.6818304462792079,
      "grad_norm": 0.9841445939811543,
      "learning_rate": 3.066970462736576e-06,
      "loss": 0.4974,
      "step": 12635
    },
    {
      "epoch": 0.6818844099077221,
      "grad_norm": 0.9542133163022625,
      "learning_rate": 3.066328727311451e-06,
      "loss": 0.43,
      "step": 12636
    },
    {
      "epoch": 0.6819383735362365,
      "grad_norm": 1.1015159427113865,
      "learning_rate": 3.0656870618324975e-06,
      "loss": 0.491,
      "step": 12637
    },
    {
      "epoch": 0.6819923371647509,
      "grad_norm": 1.082376686227805,
      "learning_rate": 3.065045466318157e-06,
      "loss": 0.4122,
      "step": 12638
    },
    {
      "epoch": 0.6820463007932653,
      "grad_norm": 0.8166482495697612,
      "learning_rate": 3.0644039407868705e-06,
      "loss": 0.2809,
      "step": 12639
    },
    {
      "epoch": 0.6821002644217797,
      "grad_norm": 0.8826774003998288,
      "learning_rate": 3.0637624852570757e-06,
      "loss": 0.3284,
      "step": 12640
    },
    {
      "epoch": 0.6821542280502941,
      "grad_norm": 1.0542326090732264,
      "learning_rate": 3.0631210997472084e-06,
      "loss": 0.4068,
      "step": 12641
    },
    {
      "epoch": 0.6822081916788085,
      "grad_norm": 1.0713893224789965,
      "learning_rate": 3.0624797842757027e-06,
      "loss": 0.5491,
      "step": 12642
    },
    {
      "epoch": 0.6822621553073228,
      "grad_norm": 0.8857340182055169,
      "learning_rate": 3.061838538860993e-06,
      "loss": 0.4768,
      "step": 12643
    },
    {
      "epoch": 0.6823161189358372,
      "grad_norm": 1.078015850499123,
      "learning_rate": 3.0611973635215066e-06,
      "loss": 0.3582,
      "step": 12644
    },
    {
      "epoch": 0.6823700825643516,
      "grad_norm": 1.1238528898831581,
      "learning_rate": 3.060556258275671e-06,
      "loss": 0.6395,
      "step": 12645
    },
    {
      "epoch": 0.682424046192866,
      "grad_norm": 1.0445012733833214,
      "learning_rate": 3.059915223141914e-06,
      "loss": 0.4366,
      "step": 12646
    },
    {
      "epoch": 0.6824780098213804,
      "grad_norm": 0.8407677588637472,
      "learning_rate": 3.0592742581386585e-06,
      "loss": 0.3312,
      "step": 12647
    },
    {
      "epoch": 0.6825319734498948,
      "grad_norm": 1.071469680996547,
      "learning_rate": 3.0586333632843268e-06,
      "loss": 0.5237,
      "step": 12648
    },
    {
      "epoch": 0.6825859370784092,
      "grad_norm": 0.7134973318081653,
      "learning_rate": 3.0579925385973395e-06,
      "loss": 0.2769,
      "step": 12649
    },
    {
      "epoch": 0.6826399007069235,
      "grad_norm": 1.150841107917194,
      "learning_rate": 3.0573517840961164e-06,
      "loss": 0.4797,
      "step": 12650
    },
    {
      "epoch": 0.6826938643354379,
      "grad_norm": 0.8656777621485527,
      "learning_rate": 3.056711099799069e-06,
      "loss": 0.3484,
      "step": 12651
    },
    {
      "epoch": 0.6827478279639523,
      "grad_norm": 0.7323445557044926,
      "learning_rate": 3.056070485724613e-06,
      "loss": 0.2872,
      "step": 12652
    },
    {
      "epoch": 0.6828017915924667,
      "grad_norm": 0.7583763123707657,
      "learning_rate": 3.055429941891161e-06,
      "loss": 0.2827,
      "step": 12653
    },
    {
      "epoch": 0.6828557552209811,
      "grad_norm": 1.0863264419131908,
      "learning_rate": 3.054789468317122e-06,
      "loss": 0.4883,
      "step": 12654
    },
    {
      "epoch": 0.6829097188494955,
      "grad_norm": 0.9860521528548339,
      "learning_rate": 3.054149065020905e-06,
      "loss": 0.4903,
      "step": 12655
    },
    {
      "epoch": 0.6829636824780099,
      "grad_norm": 1.0952652856259264,
      "learning_rate": 3.0535087320209145e-06,
      "loss": 0.4318,
      "step": 12656
    },
    {
      "epoch": 0.6830176461065242,
      "grad_norm": 0.8921205113801666,
      "learning_rate": 3.0528684693355575e-06,
      "loss": 0.4638,
      "step": 12657
    },
    {
      "epoch": 0.6830716097350386,
      "grad_norm": 0.9382146530197142,
      "learning_rate": 3.052228276983231e-06,
      "loss": 0.4333,
      "step": 12658
    },
    {
      "epoch": 0.683125573363553,
      "grad_norm": 0.8342004687840429,
      "learning_rate": 3.0515881549823377e-06,
      "loss": 0.3667,
      "step": 12659
    },
    {
      "epoch": 0.6831795369920673,
      "grad_norm": 1.0564749684844854,
      "learning_rate": 3.0509481033512743e-06,
      "loss": 0.4453,
      "step": 12660
    },
    {
      "epoch": 0.6832335006205817,
      "grad_norm": 0.81822993690754,
      "learning_rate": 3.050308122108438e-06,
      "loss": 0.3635,
      "step": 12661
    },
    {
      "epoch": 0.6832874642490961,
      "grad_norm": 1.0354231330552925,
      "learning_rate": 3.049668211272221e-06,
      "loss": 0.3627,
      "step": 12662
    },
    {
      "epoch": 0.6833414278776105,
      "grad_norm": 0.923002398589932,
      "learning_rate": 3.0490283708610147e-06,
      "loss": 0.444,
      "step": 12663
    },
    {
      "epoch": 0.6833953915061248,
      "grad_norm": 1.1217331551716418,
      "learning_rate": 3.048388600893212e-06,
      "loss": 0.4573,
      "step": 12664
    },
    {
      "epoch": 0.6834493551346392,
      "grad_norm": 0.9476313230085514,
      "learning_rate": 3.0477489013871964e-06,
      "loss": 0.3555,
      "step": 12665
    },
    {
      "epoch": 0.6835033187631536,
      "grad_norm": 1.2581918717337999,
      "learning_rate": 3.047109272361355e-06,
      "loss": 0.5422,
      "step": 12666
    },
    {
      "epoch": 0.683557282391668,
      "grad_norm": 1.023918367088428,
      "learning_rate": 3.046469713834071e-06,
      "loss": 0.4558,
      "step": 12667
    },
    {
      "epoch": 0.6836112460201824,
      "grad_norm": 0.8633624241321597,
      "learning_rate": 3.0458302258237272e-06,
      "loss": 0.5295,
      "step": 12668
    },
    {
      "epoch": 0.6836652096486968,
      "grad_norm": 1.0806135865754323,
      "learning_rate": 3.0451908083487025e-06,
      "loss": 0.3605,
      "step": 12669
    },
    {
      "epoch": 0.6837191732772112,
      "grad_norm": 1.0268492895278352,
      "learning_rate": 3.0445514614273743e-06,
      "loss": 0.3971,
      "step": 12670
    },
    {
      "epoch": 0.6837731369057255,
      "grad_norm": 1.149274722702989,
      "learning_rate": 3.0439121850781196e-06,
      "loss": 0.5205,
      "step": 12671
    },
    {
      "epoch": 0.6838271005342399,
      "grad_norm": 1.2922088723939693,
      "learning_rate": 3.043272979319308e-06,
      "loss": 0.5147,
      "step": 12672
    },
    {
      "epoch": 0.6838810641627543,
      "grad_norm": 0.9523610297594303,
      "learning_rate": 3.042633844169315e-06,
      "loss": 0.373,
      "step": 12673
    },
    {
      "epoch": 0.6839350277912687,
      "grad_norm": 1.0806817246647884,
      "learning_rate": 3.0419947796465073e-06,
      "loss": 0.5046,
      "step": 12674
    },
    {
      "epoch": 0.6839889914197831,
      "grad_norm": 0.9086836699982206,
      "learning_rate": 3.041355785769253e-06,
      "loss": 0.3893,
      "step": 12675
    },
    {
      "epoch": 0.6840429550482975,
      "grad_norm": 1.1259510233450283,
      "learning_rate": 3.0407168625559187e-06,
      "loss": 0.4786,
      "step": 12676
    },
    {
      "epoch": 0.6840969186768119,
      "grad_norm": 1.0550378495994337,
      "learning_rate": 3.0400780100248663e-06,
      "loss": 0.405,
      "step": 12677
    },
    {
      "epoch": 0.6841508823053262,
      "grad_norm": 1.04313224152795,
      "learning_rate": 3.0394392281944585e-06,
      "loss": 0.4801,
      "step": 12678
    },
    {
      "epoch": 0.6842048459338406,
      "grad_norm": 0.9935208336225612,
      "learning_rate": 3.0388005170830525e-06,
      "loss": 0.4316,
      "step": 12679
    },
    {
      "epoch": 0.684258809562355,
      "grad_norm": 0.8536603550933983,
      "learning_rate": 3.038161876709006e-06,
      "loss": 0.5173,
      "step": 12680
    },
    {
      "epoch": 0.6843127731908694,
      "grad_norm": 1.093904279695257,
      "learning_rate": 3.037523307090676e-06,
      "loss": 0.5125,
      "step": 12681
    },
    {
      "epoch": 0.6843667368193838,
      "grad_norm": 1.077112926265179,
      "learning_rate": 3.0368848082464125e-06,
      "loss": 0.5851,
      "step": 12682
    },
    {
      "epoch": 0.6844207004478982,
      "grad_norm": 0.9086217160384871,
      "learning_rate": 3.0362463801945698e-06,
      "loss": 0.374,
      "step": 12683
    },
    {
      "epoch": 0.6844746640764126,
      "grad_norm": 0.9028834399483868,
      "learning_rate": 3.0356080229534947e-06,
      "loss": 0.4731,
      "step": 12684
    },
    {
      "epoch": 0.6845286277049268,
      "grad_norm": 1.0732844208954762,
      "learning_rate": 3.0349697365415353e-06,
      "loss": 0.4343,
      "step": 12685
    },
    {
      "epoch": 0.6845825913334412,
      "grad_norm": 0.8803259810964136,
      "learning_rate": 3.0343315209770363e-06,
      "loss": 0.4,
      "step": 12686
    },
    {
      "epoch": 0.6846365549619556,
      "grad_norm": 0.943562997615735,
      "learning_rate": 3.033693376278341e-06,
      "loss": 0.4353,
      "step": 12687
    },
    {
      "epoch": 0.68469051859047,
      "grad_norm": 1.201821833456644,
      "learning_rate": 3.03305530246379e-06,
      "loss": 0.6316,
      "step": 12688
    },
    {
      "epoch": 0.6847444822189844,
      "grad_norm": 0.9802262613915448,
      "learning_rate": 3.032417299551722e-06,
      "loss": 0.4613,
      "step": 12689
    },
    {
      "epoch": 0.6847984458474988,
      "grad_norm": 0.9562057463996573,
      "learning_rate": 3.0317793675604746e-06,
      "loss": 0.4435,
      "step": 12690
    },
    {
      "epoch": 0.6848524094760131,
      "grad_norm": 0.9431549856319231,
      "learning_rate": 3.0311415065083838e-06,
      "loss": 0.3428,
      "step": 12691
    },
    {
      "epoch": 0.6849063731045275,
      "grad_norm": 0.8324324480400573,
      "learning_rate": 3.0305037164137785e-06,
      "loss": 0.3019,
      "step": 12692
    },
    {
      "epoch": 0.6849603367330419,
      "grad_norm": 1.2571921263147945,
      "learning_rate": 3.0298659972949917e-06,
      "loss": 0.5086,
      "step": 12693
    },
    {
      "epoch": 0.6850143003615563,
      "grad_norm": 1.1571894716790918,
      "learning_rate": 3.0292283491703527e-06,
      "loss": 0.4808,
      "step": 12694
    },
    {
      "epoch": 0.6850682639900707,
      "grad_norm": 1.0665843087650773,
      "learning_rate": 3.028590772058187e-06,
      "loss": 0.4488,
      "step": 12695
    },
    {
      "epoch": 0.6851222276185851,
      "grad_norm": 1.0327212709632838,
      "learning_rate": 3.0279532659768194e-06,
      "loss": 0.4271,
      "step": 12696
    },
    {
      "epoch": 0.6851761912470995,
      "grad_norm": 1.273466929140729,
      "learning_rate": 3.0273158309445737e-06,
      "loss": 0.5292,
      "step": 12697
    },
    {
      "epoch": 0.6852301548756138,
      "grad_norm": 0.9368044973170376,
      "learning_rate": 3.0266784669797704e-06,
      "loss": 0.4592,
      "step": 12698
    },
    {
      "epoch": 0.6852841185041282,
      "grad_norm": 0.9967815606875823,
      "learning_rate": 3.0260411741007258e-06,
      "loss": 0.4795,
      "step": 12699
    },
    {
      "epoch": 0.6853380821326426,
      "grad_norm": 0.880963010383675,
      "learning_rate": 3.0254039523257572e-06,
      "loss": 0.3831,
      "step": 12700
    },
    {
      "epoch": 0.685392045761157,
      "grad_norm": 1.3034087299681338,
      "learning_rate": 3.0247668016731803e-06,
      "loss": 0.5394,
      "step": 12701
    },
    {
      "epoch": 0.6854460093896714,
      "grad_norm": 0.70984561928335,
      "learning_rate": 3.024129722161305e-06,
      "loss": 0.2572,
      "step": 12702
    },
    {
      "epoch": 0.6854999730181858,
      "grad_norm": 1.2371971611257149,
      "learning_rate": 3.0234927138084443e-06,
      "loss": 0.5539,
      "step": 12703
    },
    {
      "epoch": 0.6855539366467002,
      "grad_norm": 1.053413426524323,
      "learning_rate": 3.0228557766329047e-06,
      "loss": 0.4626,
      "step": 12704
    },
    {
      "epoch": 0.6856079002752145,
      "grad_norm": 1.014991213952614,
      "learning_rate": 3.0222189106529953e-06,
      "loss": 0.3832,
      "step": 12705
    },
    {
      "epoch": 0.6856618639037289,
      "grad_norm": 1.3214078402524474,
      "learning_rate": 3.0215821158870163e-06,
      "loss": 0.5807,
      "step": 12706
    },
    {
      "epoch": 0.6857158275322432,
      "grad_norm": 1.0321501640300252,
      "learning_rate": 3.0209453923532704e-06,
      "loss": 0.5332,
      "step": 12707
    },
    {
      "epoch": 0.6857697911607576,
      "grad_norm": 1.0988265220574764,
      "learning_rate": 3.0203087400700603e-06,
      "loss": 0.4843,
      "step": 12708
    },
    {
      "epoch": 0.685823754789272,
      "grad_norm": 1.06169350525357,
      "learning_rate": 3.0196721590556822e-06,
      "loss": 0.4802,
      "step": 12709
    },
    {
      "epoch": 0.6858777184177864,
      "grad_norm": 0.8831078009059795,
      "learning_rate": 3.0190356493284322e-06,
      "loss": 0.4144,
      "step": 12710
    },
    {
      "epoch": 0.6859316820463008,
      "grad_norm": 1.0128377030253473,
      "learning_rate": 3.0183992109066053e-06,
      "loss": 0.5153,
      "step": 12711
    },
    {
      "epoch": 0.6859856456748151,
      "grad_norm": 1.2563031006959888,
      "learning_rate": 3.0177628438084927e-06,
      "loss": 0.5364,
      "step": 12712
    },
    {
      "epoch": 0.6860396093033295,
      "grad_norm": 1.058063619181959,
      "learning_rate": 3.0171265480523835e-06,
      "loss": 0.4769,
      "step": 12713
    },
    {
      "epoch": 0.6860935729318439,
      "grad_norm": 1.0278367782836342,
      "learning_rate": 3.0164903236565657e-06,
      "loss": 0.5072,
      "step": 12714
    },
    {
      "epoch": 0.6861475365603583,
      "grad_norm": 0.7616513751162358,
      "learning_rate": 3.015854170639325e-06,
      "loss": 0.2992,
      "step": 12715
    },
    {
      "epoch": 0.6862015001888727,
      "grad_norm": 1.049721391009651,
      "learning_rate": 3.0152180890189462e-06,
      "loss": 0.554,
      "step": 12716
    },
    {
      "epoch": 0.6862554638173871,
      "grad_norm": 0.9074011367355389,
      "learning_rate": 3.01458207881371e-06,
      "loss": 0.342,
      "step": 12717
    },
    {
      "epoch": 0.6863094274459015,
      "grad_norm": 0.9891598427279928,
      "learning_rate": 3.013946140041897e-06,
      "loss": 0.417,
      "step": 12718
    },
    {
      "epoch": 0.6863633910744158,
      "grad_norm": 0.9967838476854528,
      "learning_rate": 3.0133102727217843e-06,
      "loss": 0.4177,
      "step": 12719
    },
    {
      "epoch": 0.6864173547029302,
      "grad_norm": 1.2351139905032635,
      "learning_rate": 3.0126744768716453e-06,
      "loss": 0.4177,
      "step": 12720
    },
    {
      "epoch": 0.6864713183314446,
      "grad_norm": 1.021506460156865,
      "learning_rate": 3.0120387525097555e-06,
      "loss": 0.5684,
      "step": 12721
    },
    {
      "epoch": 0.686525281959959,
      "grad_norm": 0.9727627045952555,
      "learning_rate": 3.0114030996543863e-06,
      "loss": 0.5288,
      "step": 12722
    },
    {
      "epoch": 0.6865792455884734,
      "grad_norm": 0.8258285449158584,
      "learning_rate": 3.010767518323806e-06,
      "loss": 0.4009,
      "step": 12723
    },
    {
      "epoch": 0.6866332092169878,
      "grad_norm": 1.0101042920915568,
      "learning_rate": 3.0101320085362827e-06,
      "loss": 0.4204,
      "step": 12724
    },
    {
      "epoch": 0.6866871728455022,
      "grad_norm": 0.9144914437519924,
      "learning_rate": 3.0094965703100808e-06,
      "loss": 0.3762,
      "step": 12725
    },
    {
      "epoch": 0.6867411364740165,
      "grad_norm": 1.059062840193236,
      "learning_rate": 3.008861203663466e-06,
      "loss": 0.538,
      "step": 12726
    },
    {
      "epoch": 0.6867951001025309,
      "grad_norm": 1.1495860666088755,
      "learning_rate": 3.008225908614696e-06,
      "loss": 0.5385,
      "step": 12727
    },
    {
      "epoch": 0.6868490637310453,
      "grad_norm": 1.0272588474806184,
      "learning_rate": 3.00759068518203e-06,
      "loss": 0.4282,
      "step": 12728
    },
    {
      "epoch": 0.6869030273595597,
      "grad_norm": 0.9261986323023068,
      "learning_rate": 3.0069555333837277e-06,
      "loss": 0.4226,
      "step": 12729
    },
    {
      "epoch": 0.6869569909880741,
      "grad_norm": 0.9953545060654383,
      "learning_rate": 3.006320453238041e-06,
      "loss": 0.4551,
      "step": 12730
    },
    {
      "epoch": 0.6870109546165885,
      "grad_norm": 1.1869557342465598,
      "learning_rate": 3.0056854447632254e-06,
      "loss": 0.5484,
      "step": 12731
    },
    {
      "epoch": 0.6870649182451029,
      "grad_norm": 1.0152731760643607,
      "learning_rate": 3.0050505079775293e-06,
      "loss": 0.3455,
      "step": 12732
    },
    {
      "epoch": 0.6871188818736171,
      "grad_norm": 0.9403908082633645,
      "learning_rate": 3.004415642899203e-06,
      "loss": 0.4801,
      "step": 12733
    },
    {
      "epoch": 0.6871728455021315,
      "grad_norm": 1.131325497731077,
      "learning_rate": 3.003780849546495e-06,
      "loss": 0.475,
      "step": 12734
    },
    {
      "epoch": 0.6872268091306459,
      "grad_norm": 1.3420347490130498,
      "learning_rate": 3.003146127937645e-06,
      "loss": 0.6216,
      "step": 12735
    },
    {
      "epoch": 0.6872807727591603,
      "grad_norm": 0.934724226804627,
      "learning_rate": 3.002511478090898e-06,
      "loss": 0.4469,
      "step": 12736
    },
    {
      "epoch": 0.6873347363876747,
      "grad_norm": 0.8880204441674767,
      "learning_rate": 3.001876900024496e-06,
      "loss": 0.3355,
      "step": 12737
    },
    {
      "epoch": 0.6873887000161891,
      "grad_norm": 1.1570794064193979,
      "learning_rate": 3.001242393756675e-06,
      "loss": 0.5955,
      "step": 12738
    },
    {
      "epoch": 0.6874426636447035,
      "grad_norm": 0.9563138060989563,
      "learning_rate": 3.0006079593056725e-06,
      "loss": 0.4503,
      "step": 12739
    },
    {
      "epoch": 0.6874966272732178,
      "grad_norm": 0.9666227748296949,
      "learning_rate": 2.9999735966897233e-06,
      "loss": 0.4342,
      "step": 12740
    },
    {
      "epoch": 0.6875505909017322,
      "grad_norm": 1.0605296401292492,
      "learning_rate": 2.999339305927058e-06,
      "loss": 0.4453,
      "step": 12741
    },
    {
      "epoch": 0.6876045545302466,
      "grad_norm": 0.9008556272320122,
      "learning_rate": 2.9987050870359092e-06,
      "loss": 0.4026,
      "step": 12742
    },
    {
      "epoch": 0.687658518158761,
      "grad_norm": 1.1051085613182805,
      "learning_rate": 2.998070940034502e-06,
      "loss": 0.4353,
      "step": 12743
    },
    {
      "epoch": 0.6877124817872754,
      "grad_norm": 1.1206816398691255,
      "learning_rate": 2.9974368649410652e-06,
      "loss": 0.5611,
      "step": 12744
    },
    {
      "epoch": 0.6877664454157898,
      "grad_norm": 0.953748980346162,
      "learning_rate": 2.996802861773821e-06,
      "loss": 0.4819,
      "step": 12745
    },
    {
      "epoch": 0.6878204090443042,
      "grad_norm": 1.073510749860126,
      "learning_rate": 2.9961689305509915e-06,
      "loss": 0.5311,
      "step": 12746
    },
    {
      "epoch": 0.6878743726728185,
      "grad_norm": 1.1155665183210635,
      "learning_rate": 2.9955350712907993e-06,
      "loss": 0.4529,
      "step": 12747
    },
    {
      "epoch": 0.6879283363013329,
      "grad_norm": 0.9748291945705497,
      "learning_rate": 2.9949012840114577e-06,
      "loss": 0.3618,
      "step": 12748
    },
    {
      "epoch": 0.6879822999298473,
      "grad_norm": 0.9507670320690967,
      "learning_rate": 2.9942675687311846e-06,
      "loss": 0.4869,
      "step": 12749
    },
    {
      "epoch": 0.6880362635583617,
      "grad_norm": 0.922976834234685,
      "learning_rate": 2.9936339254681934e-06,
      "loss": 0.4583,
      "step": 12750
    },
    {
      "epoch": 0.6880902271868761,
      "grad_norm": 0.671243853143658,
      "learning_rate": 2.993000354240696e-06,
      "loss": 0.2799,
      "step": 12751
    },
    {
      "epoch": 0.6881441908153905,
      "grad_norm": 0.889854971568285,
      "learning_rate": 2.9923668550669004e-06,
      "loss": 0.4982,
      "step": 12752
    },
    {
      "epoch": 0.6881981544439048,
      "grad_norm": 1.0175401041974248,
      "learning_rate": 2.9917334279650178e-06,
      "loss": 0.5227,
      "step": 12753
    },
    {
      "epoch": 0.6882521180724192,
      "grad_norm": 0.9826513006131506,
      "learning_rate": 2.991100072953249e-06,
      "loss": 0.4221,
      "step": 12754
    },
    {
      "epoch": 0.6883060817009335,
      "grad_norm": 1.1895500903057927,
      "learning_rate": 2.990466790049799e-06,
      "loss": 0.5329,
      "step": 12755
    },
    {
      "epoch": 0.688360045329448,
      "grad_norm": 1.2865643520609955,
      "learning_rate": 2.9898335792728696e-06,
      "loss": 0.63,
      "step": 12756
    },
    {
      "epoch": 0.6884140089579623,
      "grad_norm": 1.0989712702155578,
      "learning_rate": 2.989200440640659e-06,
      "loss": 0.4825,
      "step": 12757
    },
    {
      "epoch": 0.6884679725864767,
      "grad_norm": 0.8946290690389784,
      "learning_rate": 2.988567374171365e-06,
      "loss": 0.461,
      "step": 12758
    },
    {
      "epoch": 0.6885219362149911,
      "grad_norm": 0.9533144461807532,
      "learning_rate": 2.9879343798831823e-06,
      "loss": 0.436,
      "step": 12759
    },
    {
      "epoch": 0.6885758998435054,
      "grad_norm": 0.994147465782264,
      "learning_rate": 2.987301457794305e-06,
      "loss": 0.4058,
      "step": 12760
    },
    {
      "epoch": 0.6886298634720198,
      "grad_norm": 1.0067046624809628,
      "learning_rate": 2.986668607922921e-06,
      "loss": 0.5064,
      "step": 12761
    },
    {
      "epoch": 0.6886838271005342,
      "grad_norm": 1.0557670670104513,
      "learning_rate": 2.9860358302872217e-06,
      "loss": 0.4829,
      "step": 12762
    },
    {
      "epoch": 0.6887377907290486,
      "grad_norm": 1.0763808284635679,
      "learning_rate": 2.9854031249053915e-06,
      "loss": 0.5697,
      "step": 12763
    },
    {
      "epoch": 0.688791754357563,
      "grad_norm": 0.9685494515694384,
      "learning_rate": 2.9847704917956176e-06,
      "loss": 0.4759,
      "step": 12764
    },
    {
      "epoch": 0.6888457179860774,
      "grad_norm": 0.9617306247323356,
      "learning_rate": 2.984137930976081e-06,
      "loss": 0.4081,
      "step": 12765
    },
    {
      "epoch": 0.6888996816145918,
      "grad_norm": 1.274929988234217,
      "learning_rate": 2.9835054424649617e-06,
      "loss": 0.6508,
      "step": 12766
    },
    {
      "epoch": 0.6889536452431061,
      "grad_norm": 1.0463103203746498,
      "learning_rate": 2.982873026280441e-06,
      "loss": 0.5252,
      "step": 12767
    },
    {
      "epoch": 0.6890076088716205,
      "grad_norm": 1.0279205832770368,
      "learning_rate": 2.982240682440691e-06,
      "loss": 0.4898,
      "step": 12768
    },
    {
      "epoch": 0.6890615725001349,
      "grad_norm": 1.1543333868432266,
      "learning_rate": 2.9816084109638877e-06,
      "loss": 0.5339,
      "step": 12769
    },
    {
      "epoch": 0.6891155361286493,
      "grad_norm": 1.0280381767904392,
      "learning_rate": 2.980976211868203e-06,
      "loss": 0.4243,
      "step": 12770
    },
    {
      "epoch": 0.6891694997571637,
      "grad_norm": 1.27618783712121,
      "learning_rate": 2.980344085171808e-06,
      "loss": 0.523,
      "step": 12771
    },
    {
      "epoch": 0.6892234633856781,
      "grad_norm": 1.1861103466234624,
      "learning_rate": 2.979712030892869e-06,
      "loss": 0.6508,
      "step": 12772
    },
    {
      "epoch": 0.6892774270141925,
      "grad_norm": 1.0990688249428893,
      "learning_rate": 2.9790800490495535e-06,
      "loss": 0.5152,
      "step": 12773
    },
    {
      "epoch": 0.6893313906427068,
      "grad_norm": 1.041512158676093,
      "learning_rate": 2.9784481396600262e-06,
      "loss": 0.4349,
      "step": 12774
    },
    {
      "epoch": 0.6893853542712212,
      "grad_norm": 0.9946457718109984,
      "learning_rate": 2.9778163027424457e-06,
      "loss": 0.3723,
      "step": 12775
    },
    {
      "epoch": 0.6894393178997356,
      "grad_norm": 0.9053831850530457,
      "learning_rate": 2.9771845383149723e-06,
      "loss": 0.3668,
      "step": 12776
    },
    {
      "epoch": 0.68949328152825,
      "grad_norm": 1.20188578848208,
      "learning_rate": 2.976552846395765e-06,
      "loss": 0.5144,
      "step": 12777
    },
    {
      "epoch": 0.6895472451567644,
      "grad_norm": 1.0876389452420203,
      "learning_rate": 2.975921227002979e-06,
      "loss": 0.5104,
      "step": 12778
    },
    {
      "epoch": 0.6896012087852788,
      "grad_norm": 0.6735362549239753,
      "learning_rate": 2.9752896801547664e-06,
      "loss": 0.2535,
      "step": 12779
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 1.1012285456099549,
      "learning_rate": 2.9746582058692803e-06,
      "loss": 0.4243,
      "step": 12780
    },
    {
      "epoch": 0.6897091360423074,
      "grad_norm": 0.8280515276750501,
      "learning_rate": 2.9740268041646705e-06,
      "loss": 0.3878,
      "step": 12781
    },
    {
      "epoch": 0.6897630996708218,
      "grad_norm": 1.0428676099608645,
      "learning_rate": 2.9733954750590805e-06,
      "loss": 0.4638,
      "step": 12782
    },
    {
      "epoch": 0.6898170632993362,
      "grad_norm": 1.0360979456659163,
      "learning_rate": 2.9727642185706573e-06,
      "loss": 0.4009,
      "step": 12783
    },
    {
      "epoch": 0.6898710269278506,
      "grad_norm": 0.8974776707338593,
      "learning_rate": 2.972133034717545e-06,
      "loss": 0.434,
      "step": 12784
    },
    {
      "epoch": 0.689924990556365,
      "grad_norm": 1.1047652086884894,
      "learning_rate": 2.9715019235178833e-06,
      "loss": 0.463,
      "step": 12785
    },
    {
      "epoch": 0.6899789541848794,
      "grad_norm": 1.142281241198386,
      "learning_rate": 2.9708708849898106e-06,
      "loss": 0.51,
      "step": 12786
    },
    {
      "epoch": 0.6900329178133938,
      "grad_norm": 0.9306838414242946,
      "learning_rate": 2.9702399191514648e-06,
      "loss": 0.4318,
      "step": 12787
    },
    {
      "epoch": 0.6900868814419081,
      "grad_norm": 1.133975773637936,
      "learning_rate": 2.9696090260209787e-06,
      "loss": 0.5321,
      "step": 12788
    },
    {
      "epoch": 0.6901408450704225,
      "grad_norm": 1.1201717209616342,
      "learning_rate": 2.9689782056164874e-06,
      "loss": 0.605,
      "step": 12789
    },
    {
      "epoch": 0.6901948086989369,
      "grad_norm": 0.9994785030656648,
      "learning_rate": 2.9683474579561195e-06,
      "loss": 0.4958,
      "step": 12790
    },
    {
      "epoch": 0.6902487723274513,
      "grad_norm": 0.685438639854132,
      "learning_rate": 2.9677167830580035e-06,
      "loss": 0.2199,
      "step": 12791
    },
    {
      "epoch": 0.6903027359559657,
      "grad_norm": 1.0839501941598289,
      "learning_rate": 2.967086180940266e-06,
      "loss": 0.6265,
      "step": 12792
    },
    {
      "epoch": 0.6903566995844801,
      "grad_norm": 1.2373106329913193,
      "learning_rate": 2.9664556516210306e-06,
      "loss": 0.5534,
      "step": 12793
    },
    {
      "epoch": 0.6904106632129945,
      "grad_norm": 1.0137098475843154,
      "learning_rate": 2.9658251951184204e-06,
      "loss": 0.5928,
      "step": 12794
    },
    {
      "epoch": 0.6904646268415088,
      "grad_norm": 0.9098991973635244,
      "learning_rate": 2.965194811450556e-06,
      "loss": 0.3586,
      "step": 12795
    },
    {
      "epoch": 0.6905185904700232,
      "grad_norm": 1.1520693370503128,
      "learning_rate": 2.9645645006355516e-06,
      "loss": 0.5604,
      "step": 12796
    },
    {
      "epoch": 0.6905725540985376,
      "grad_norm": 1.0659510666442944,
      "learning_rate": 2.963934262691526e-06,
      "loss": 0.5049,
      "step": 12797
    },
    {
      "epoch": 0.690626517727052,
      "grad_norm": 1.138469039816294,
      "learning_rate": 2.9633040976365923e-06,
      "loss": 0.6063,
      "step": 12798
    },
    {
      "epoch": 0.6906804813555664,
      "grad_norm": 0.764085544235665,
      "learning_rate": 2.962674005488862e-06,
      "loss": 0.3419,
      "step": 12799
    },
    {
      "epoch": 0.6907344449840808,
      "grad_norm": 0.9199802083116927,
      "learning_rate": 2.9620439862664442e-06,
      "loss": 0.4645,
      "step": 12800
    },
    {
      "epoch": 0.6907884086125952,
      "grad_norm": 1.0350934841007453,
      "learning_rate": 2.961414039987447e-06,
      "loss": 0.4662,
      "step": 12801
    },
    {
      "epoch": 0.6908423722411094,
      "grad_norm": 1.0788544293015672,
      "learning_rate": 2.9607841666699764e-06,
      "loss": 0.4953,
      "step": 12802
    },
    {
      "epoch": 0.6908963358696238,
      "grad_norm": 1.052727462178044,
      "learning_rate": 2.9601543663321342e-06,
      "loss": 0.4706,
      "step": 12803
    },
    {
      "epoch": 0.6909502994981382,
      "grad_norm": 1.0284677710985632,
      "learning_rate": 2.959524638992021e-06,
      "loss": 0.4818,
      "step": 12804
    },
    {
      "epoch": 0.6910042631266526,
      "grad_norm": 0.983814961542173,
      "learning_rate": 2.958894984667736e-06,
      "loss": 0.453,
      "step": 12805
    },
    {
      "epoch": 0.691058226755167,
      "grad_norm": 1.2077043201157578,
      "learning_rate": 2.9582654033773774e-06,
      "loss": 0.4542,
      "step": 12806
    },
    {
      "epoch": 0.6911121903836814,
      "grad_norm": 1.0514266767472853,
      "learning_rate": 2.9576358951390393e-06,
      "loss": 0.4371,
      "step": 12807
    },
    {
      "epoch": 0.6911661540121958,
      "grad_norm": 0.9613212468182926,
      "learning_rate": 2.957006459970815e-06,
      "loss": 0.3781,
      "step": 12808
    },
    {
      "epoch": 0.6912201176407101,
      "grad_norm": 0.9894138901574346,
      "learning_rate": 2.9563770978907945e-06,
      "loss": 0.4225,
      "step": 12809
    },
    {
      "epoch": 0.6912740812692245,
      "grad_norm": 1.0604461927076396,
      "learning_rate": 2.9557478089170666e-06,
      "loss": 0.3967,
      "step": 12810
    },
    {
      "epoch": 0.6913280448977389,
      "grad_norm": 1.142045932852909,
      "learning_rate": 2.9551185930677164e-06,
      "loss": 0.5271,
      "step": 12811
    },
    {
      "epoch": 0.6913820085262533,
      "grad_norm": 1.1036941784297516,
      "learning_rate": 2.9544894503608296e-06,
      "loss": 0.6816,
      "step": 12812
    },
    {
      "epoch": 0.6914359721547677,
      "grad_norm": 1.122021114108549,
      "learning_rate": 2.9538603808144883e-06,
      "loss": 0.5161,
      "step": 12813
    },
    {
      "epoch": 0.6914899357832821,
      "grad_norm": 1.213969808302271,
      "learning_rate": 2.953231384446772e-06,
      "loss": 0.4975,
      "step": 12814
    },
    {
      "epoch": 0.6915438994117965,
      "grad_norm": 0.9515384534676911,
      "learning_rate": 2.9526024612757586e-06,
      "loss": 0.4014,
      "step": 12815
    },
    {
      "epoch": 0.6915978630403108,
      "grad_norm": 0.8577377069896276,
      "learning_rate": 2.951973611319527e-06,
      "loss": 0.3528,
      "step": 12816
    },
    {
      "epoch": 0.6916518266688252,
      "grad_norm": 0.9292944792173938,
      "learning_rate": 2.951344834596146e-06,
      "loss": 0.447,
      "step": 12817
    },
    {
      "epoch": 0.6917057902973396,
      "grad_norm": 0.8346644194549784,
      "learning_rate": 2.95071613112369e-06,
      "loss": 0.3123,
      "step": 12818
    },
    {
      "epoch": 0.691759753925854,
      "grad_norm": 1.1390289959374753,
      "learning_rate": 2.950087500920228e-06,
      "loss": 0.4898,
      "step": 12819
    },
    {
      "epoch": 0.6918137175543684,
      "grad_norm": 1.0979177218669127,
      "learning_rate": 2.9494589440038278e-06,
      "loss": 0.3971,
      "step": 12820
    },
    {
      "epoch": 0.6918676811828828,
      "grad_norm": 1.1418788249689829,
      "learning_rate": 2.948830460392555e-06,
      "loss": 0.5967,
      "step": 12821
    },
    {
      "epoch": 0.6919216448113971,
      "grad_norm": 1.2444672462530921,
      "learning_rate": 2.948202050104474e-06,
      "loss": 0.551,
      "step": 12822
    },
    {
      "epoch": 0.6919756084399115,
      "grad_norm": 1.1496721954742772,
      "learning_rate": 2.947573713157643e-06,
      "loss": 0.606,
      "step": 12823
    },
    {
      "epoch": 0.6920295720684259,
      "grad_norm": 1.2679227411450105,
      "learning_rate": 2.9469454495701223e-06,
      "loss": 0.7106,
      "step": 12824
    },
    {
      "epoch": 0.6920835356969403,
      "grad_norm": 1.0431428752633156,
      "learning_rate": 2.9463172593599688e-06,
      "loss": 0.4116,
      "step": 12825
    },
    {
      "epoch": 0.6921374993254547,
      "grad_norm": 1.1172435362836337,
      "learning_rate": 2.9456891425452383e-06,
      "loss": 0.5489,
      "step": 12826
    },
    {
      "epoch": 0.692191462953969,
      "grad_norm": 0.9130826728260667,
      "learning_rate": 2.945061099143982e-06,
      "loss": 0.3221,
      "step": 12827
    },
    {
      "epoch": 0.6922454265824834,
      "grad_norm": 1.026841822949245,
      "learning_rate": 2.9444331291742524e-06,
      "loss": 0.477,
      "step": 12828
    },
    {
      "epoch": 0.6922993902109977,
      "grad_norm": 1.1550470888886188,
      "learning_rate": 2.943805232654097e-06,
      "loss": 0.3965,
      "step": 12829
    },
    {
      "epoch": 0.6923533538395121,
      "grad_norm": 0.971067926112307,
      "learning_rate": 2.9431774096015613e-06,
      "loss": 0.4463,
      "step": 12830
    },
    {
      "epoch": 0.6924073174680265,
      "grad_norm": 0.9876562016159565,
      "learning_rate": 2.94254966003469e-06,
      "loss": 0.428,
      "step": 12831
    },
    {
      "epoch": 0.6924612810965409,
      "grad_norm": 0.9138679242861155,
      "learning_rate": 2.9419219839715264e-06,
      "loss": 0.3441,
      "step": 12832
    },
    {
      "epoch": 0.6925152447250553,
      "grad_norm": 1.1251477284386113,
      "learning_rate": 2.941294381430109e-06,
      "loss": 0.4598,
      "step": 12833
    },
    {
      "epoch": 0.6925692083535697,
      "grad_norm": 1.1267706879465222,
      "learning_rate": 2.9406668524284764e-06,
      "loss": 0.5949,
      "step": 12834
    },
    {
      "epoch": 0.6926231719820841,
      "grad_norm": 1.0176819238250523,
      "learning_rate": 2.940039396984665e-06,
      "loss": 0.4134,
      "step": 12835
    },
    {
      "epoch": 0.6926771356105984,
      "grad_norm": 1.1822581646977657,
      "learning_rate": 2.939412015116707e-06,
      "loss": 0.5472,
      "step": 12836
    },
    {
      "epoch": 0.6927310992391128,
      "grad_norm": 0.861134468157677,
      "learning_rate": 2.9387847068426355e-06,
      "loss": 0.3796,
      "step": 12837
    },
    {
      "epoch": 0.6927850628676272,
      "grad_norm": 0.9688210676742628,
      "learning_rate": 2.9381574721804803e-06,
      "loss": 0.3774,
      "step": 12838
    },
    {
      "epoch": 0.6928390264961416,
      "grad_norm": 1.289014146593846,
      "learning_rate": 2.9375303111482674e-06,
      "loss": 0.6295,
      "step": 12839
    },
    {
      "epoch": 0.692892990124656,
      "grad_norm": 0.9856520429966804,
      "learning_rate": 2.936903223764024e-06,
      "loss": 0.3841,
      "step": 12840
    },
    {
      "epoch": 0.6929469537531704,
      "grad_norm": 1.2479226986039564,
      "learning_rate": 2.93627621004577e-06,
      "loss": 0.6013,
      "step": 12841
    },
    {
      "epoch": 0.6930009173816848,
      "grad_norm": 1.0624119384967545,
      "learning_rate": 2.935649270011528e-06,
      "loss": 0.4386,
      "step": 12842
    },
    {
      "epoch": 0.6930548810101991,
      "grad_norm": 1.1949556795403318,
      "learning_rate": 2.9350224036793174e-06,
      "loss": 0.4706,
      "step": 12843
    },
    {
      "epoch": 0.6931088446387135,
      "grad_norm": 0.8689130047078445,
      "learning_rate": 2.9343956110671546e-06,
      "loss": 0.4316,
      "step": 12844
    },
    {
      "epoch": 0.6931628082672279,
      "grad_norm": 1.0109999059991253,
      "learning_rate": 2.933768892193054e-06,
      "loss": 0.5558,
      "step": 12845
    },
    {
      "epoch": 0.6932167718957423,
      "grad_norm": 1.0778598863392521,
      "learning_rate": 2.9331422470750293e-06,
      "loss": 0.5348,
      "step": 12846
    },
    {
      "epoch": 0.6932707355242567,
      "grad_norm": 1.2042754343618705,
      "learning_rate": 2.932515675731089e-06,
      "loss": 0.4958,
      "step": 12847
    },
    {
      "epoch": 0.6933246991527711,
      "grad_norm": 1.1567067745555395,
      "learning_rate": 2.9318891781792434e-06,
      "loss": 0.5976,
      "step": 12848
    },
    {
      "epoch": 0.6933786627812855,
      "grad_norm": 0.8641205314285554,
      "learning_rate": 2.9312627544374967e-06,
      "loss": 0.3752,
      "step": 12849
    },
    {
      "epoch": 0.6934326264097997,
      "grad_norm": 0.9107358194982551,
      "learning_rate": 2.9306364045238563e-06,
      "loss": 0.5038,
      "step": 12850
    },
    {
      "epoch": 0.6934865900383141,
      "grad_norm": 0.905221303383921,
      "learning_rate": 2.9300101284563197e-06,
      "loss": 0.473,
      "step": 12851
    },
    {
      "epoch": 0.6935405536668285,
      "grad_norm": 1.2036915729803042,
      "learning_rate": 2.929383926252889e-06,
      "loss": 0.6014,
      "step": 12852
    },
    {
      "epoch": 0.6935945172953429,
      "grad_norm": 1.0060470117090254,
      "learning_rate": 2.9287577979315617e-06,
      "loss": 0.5597,
      "step": 12853
    },
    {
      "epoch": 0.6936484809238573,
      "grad_norm": 1.2393005424388008,
      "learning_rate": 2.928131743510334e-06,
      "loss": 0.6308,
      "step": 12854
    },
    {
      "epoch": 0.6937024445523717,
      "grad_norm": 1.1640071101929124,
      "learning_rate": 2.9275057630071974e-06,
      "loss": 0.6734,
      "step": 12855
    },
    {
      "epoch": 0.6937564081808861,
      "grad_norm": 0.9394419132303418,
      "learning_rate": 2.926879856440146e-06,
      "loss": 0.4284,
      "step": 12856
    },
    {
      "epoch": 0.6938103718094004,
      "grad_norm": 1.1560894485305147,
      "learning_rate": 2.9262540238271675e-06,
      "loss": 0.5605,
      "step": 12857
    },
    {
      "epoch": 0.6938643354379148,
      "grad_norm": 1.0748689251923167,
      "learning_rate": 2.9256282651862476e-06,
      "loss": 0.5326,
      "step": 12858
    },
    {
      "epoch": 0.6939182990664292,
      "grad_norm": 1.0911462700612236,
      "learning_rate": 2.925002580535373e-06,
      "loss": 0.4865,
      "step": 12859
    },
    {
      "epoch": 0.6939722626949436,
      "grad_norm": 1.0505641545358602,
      "learning_rate": 2.9243769698925263e-06,
      "loss": 0.4873,
      "step": 12860
    },
    {
      "epoch": 0.694026226323458,
      "grad_norm": 0.9451905969082554,
      "learning_rate": 2.923751433275688e-06,
      "loss": 0.4574,
      "step": 12861
    },
    {
      "epoch": 0.6940801899519724,
      "grad_norm": 0.8538714868563984,
      "learning_rate": 2.9231259707028357e-06,
      "loss": 0.3275,
      "step": 12862
    },
    {
      "epoch": 0.6941341535804868,
      "grad_norm": 0.9944239288036242,
      "learning_rate": 2.9225005821919476e-06,
      "loss": 0.4969,
      "step": 12863
    },
    {
      "epoch": 0.6941881172090011,
      "grad_norm": 1.0290944634440025,
      "learning_rate": 2.9218752677609986e-06,
      "loss": 0.4881,
      "step": 12864
    },
    {
      "epoch": 0.6942420808375155,
      "grad_norm": 0.8460546529779057,
      "learning_rate": 2.9212500274279574e-06,
      "loss": 0.4128,
      "step": 12865
    },
    {
      "epoch": 0.6942960444660299,
      "grad_norm": 0.7783539437856293,
      "learning_rate": 2.9206248612107968e-06,
      "loss": 0.3212,
      "step": 12866
    },
    {
      "epoch": 0.6943500080945443,
      "grad_norm": 0.9382656439671115,
      "learning_rate": 2.919999769127484e-06,
      "loss": 0.4214,
      "step": 12867
    },
    {
      "epoch": 0.6944039717230587,
      "grad_norm": 1.1143408475980336,
      "learning_rate": 2.9193747511959835e-06,
      "loss": 0.5387,
      "step": 12868
    },
    {
      "epoch": 0.6944579353515731,
      "grad_norm": 0.9477430055700514,
      "learning_rate": 2.918749807434261e-06,
      "loss": 0.3927,
      "step": 12869
    },
    {
      "epoch": 0.6945118989800875,
      "grad_norm": 0.9490471035171822,
      "learning_rate": 2.9181249378602774e-06,
      "loss": 0.4052,
      "step": 12870
    },
    {
      "epoch": 0.6945658626086018,
      "grad_norm": 1.2560340030337758,
      "learning_rate": 2.917500142491993e-06,
      "loss": 0.5212,
      "step": 12871
    },
    {
      "epoch": 0.6946198262371162,
      "grad_norm": 0.9932031827400105,
      "learning_rate": 2.9168754213473628e-06,
      "loss": 0.3797,
      "step": 12872
    },
    {
      "epoch": 0.6946737898656306,
      "grad_norm": 1.2444545396703854,
      "learning_rate": 2.9162507744443423e-06,
      "loss": 0.6569,
      "step": 12873
    },
    {
      "epoch": 0.694727753494145,
      "grad_norm": 1.2702249043633798,
      "learning_rate": 2.9156262018008863e-06,
      "loss": 0.5807,
      "step": 12874
    },
    {
      "epoch": 0.6947817171226593,
      "grad_norm": 0.9775047331302462,
      "learning_rate": 2.915001703434944e-06,
      "loss": 0.3617,
      "step": 12875
    },
    {
      "epoch": 0.6948356807511737,
      "grad_norm": 1.2580869617076678,
      "learning_rate": 2.914377279364464e-06,
      "loss": 0.6335,
      "step": 12876
    },
    {
      "epoch": 0.6948896443796881,
      "grad_norm": 0.9490033050497612,
      "learning_rate": 2.9137529296073952e-06,
      "loss": 0.4407,
      "step": 12877
    },
    {
      "epoch": 0.6949436080082024,
      "grad_norm": 0.9676004391855759,
      "learning_rate": 2.9131286541816806e-06,
      "loss": 0.4324,
      "step": 12878
    },
    {
      "epoch": 0.6949975716367168,
      "grad_norm": 0.9564224047589261,
      "learning_rate": 2.9125044531052617e-06,
      "loss": 0.4345,
      "step": 12879
    },
    {
      "epoch": 0.6950515352652312,
      "grad_norm": 0.9928597824104464,
      "learning_rate": 2.9118803263960786e-06,
      "loss": 0.3463,
      "step": 12880
    },
    {
      "epoch": 0.6951054988937456,
      "grad_norm": 0.8143170528392323,
      "learning_rate": 2.91125627407207e-06,
      "loss": 0.2503,
      "step": 12881
    },
    {
      "epoch": 0.69515946252226,
      "grad_norm": 1.0427484278282806,
      "learning_rate": 2.9106322961511725e-06,
      "loss": 0.3909,
      "step": 12882
    },
    {
      "epoch": 0.6952134261507744,
      "grad_norm": 1.0885815789910442,
      "learning_rate": 2.910008392651319e-06,
      "loss": 0.416,
      "step": 12883
    },
    {
      "epoch": 0.6952673897792888,
      "grad_norm": 1.0272594747465376,
      "learning_rate": 2.9093845635904405e-06,
      "loss": 0.4533,
      "step": 12884
    },
    {
      "epoch": 0.6953213534078031,
      "grad_norm": 1.042567206667852,
      "learning_rate": 2.9087608089864695e-06,
      "loss": 0.4142,
      "step": 12885
    },
    {
      "epoch": 0.6953753170363175,
      "grad_norm": 1.0258770208503403,
      "learning_rate": 2.9081371288573292e-06,
      "loss": 0.449,
      "step": 12886
    },
    {
      "epoch": 0.6954292806648319,
      "grad_norm": 0.9594773490159131,
      "learning_rate": 2.9075135232209466e-06,
      "loss": 0.5071,
      "step": 12887
    },
    {
      "epoch": 0.6954832442933463,
      "grad_norm": 0.8854613153592871,
      "learning_rate": 2.9068899920952452e-06,
      "loss": 0.3507,
      "step": 12888
    },
    {
      "epoch": 0.6955372079218607,
      "grad_norm": 0.9203885460880191,
      "learning_rate": 2.9062665354981454e-06,
      "loss": 0.4746,
      "step": 12889
    },
    {
      "epoch": 0.6955911715503751,
      "grad_norm": 1.1651070118938545,
      "learning_rate": 2.9056431534475666e-06,
      "loss": 0.409,
      "step": 12890
    },
    {
      "epoch": 0.6956451351788894,
      "grad_norm": 1.175098657537262,
      "learning_rate": 2.9050198459614243e-06,
      "loss": 0.6136,
      "step": 12891
    },
    {
      "epoch": 0.6956990988074038,
      "grad_norm": 1.2086980687125146,
      "learning_rate": 2.904396613057634e-06,
      "loss": 0.5182,
      "step": 12892
    },
    {
      "epoch": 0.6957530624359182,
      "grad_norm": 1.1110022549618668,
      "learning_rate": 2.903773454754108e-06,
      "loss": 0.4285,
      "step": 12893
    },
    {
      "epoch": 0.6958070260644326,
      "grad_norm": 1.090840836097646,
      "learning_rate": 2.903150371068756e-06,
      "loss": 0.4506,
      "step": 12894
    },
    {
      "epoch": 0.695860989692947,
      "grad_norm": 0.9626698731139083,
      "learning_rate": 2.902527362019486e-06,
      "loss": 0.4799,
      "step": 12895
    },
    {
      "epoch": 0.6959149533214614,
      "grad_norm": 1.0098823480966108,
      "learning_rate": 2.9019044276242046e-06,
      "loss": 0.4269,
      "step": 12896
    },
    {
      "epoch": 0.6959689169499758,
      "grad_norm": 1.049464053740126,
      "learning_rate": 2.9012815679008155e-06,
      "loss": 0.4203,
      "step": 12897
    },
    {
      "epoch": 0.69602288057849,
      "grad_norm": 0.778048613037931,
      "learning_rate": 2.9006587828672205e-06,
      "loss": 0.3043,
      "step": 12898
    },
    {
      "epoch": 0.6960768442070044,
      "grad_norm": 0.9702105756186081,
      "learning_rate": 2.900036072541317e-06,
      "loss": 0.4419,
      "step": 12899
    },
    {
      "epoch": 0.6961308078355188,
      "grad_norm": 0.9078255756930769,
      "learning_rate": 2.8994134369410042e-06,
      "loss": 0.3893,
      "step": 12900
    },
    {
      "epoch": 0.6961847714640332,
      "grad_norm": 1.1077271708482619,
      "learning_rate": 2.8987908760841777e-06,
      "loss": 0.4663,
      "step": 12901
    },
    {
      "epoch": 0.6962387350925476,
      "grad_norm": 1.0710628809059515,
      "learning_rate": 2.898168389988728e-06,
      "loss": 0.4284,
      "step": 12902
    },
    {
      "epoch": 0.696292698721062,
      "grad_norm": 1.0203672494966032,
      "learning_rate": 2.897545978672549e-06,
      "loss": 0.5479,
      "step": 12903
    },
    {
      "epoch": 0.6963466623495764,
      "grad_norm": 1.1883344481235694,
      "learning_rate": 2.8969236421535287e-06,
      "loss": 0.4992,
      "step": 12904
    },
    {
      "epoch": 0.6964006259780907,
      "grad_norm": 1.0256511557360934,
      "learning_rate": 2.896301380449553e-06,
      "loss": 0.5288,
      "step": 12905
    },
    {
      "epoch": 0.6964545896066051,
      "grad_norm": 1.0588783767200949,
      "learning_rate": 2.895679193578506e-06,
      "loss": 0.3623,
      "step": 12906
    },
    {
      "epoch": 0.6965085532351195,
      "grad_norm": 0.9726017449109245,
      "learning_rate": 2.89505708155827e-06,
      "loss": 0.4796,
      "step": 12907
    },
    {
      "epoch": 0.6965625168636339,
      "grad_norm": 1.152626570949335,
      "learning_rate": 2.894435044406726e-06,
      "loss": 0.4958,
      "step": 12908
    },
    {
      "epoch": 0.6966164804921483,
      "grad_norm": 0.8107552941293773,
      "learning_rate": 2.893813082141751e-06,
      "loss": 0.3148,
      "step": 12909
    },
    {
      "epoch": 0.6966704441206627,
      "grad_norm": 0.9742327366538469,
      "learning_rate": 2.8931911947812224e-06,
      "loss": 0.3894,
      "step": 12910
    },
    {
      "epoch": 0.6967244077491771,
      "grad_norm": 1.0581285050418603,
      "learning_rate": 2.8925693823430117e-06,
      "loss": 0.5557,
      "step": 12911
    },
    {
      "epoch": 0.6967783713776914,
      "grad_norm": 1.0413657580950326,
      "learning_rate": 2.8919476448449944e-06,
      "loss": 0.4307,
      "step": 12912
    },
    {
      "epoch": 0.6968323350062058,
      "grad_norm": 1.005331417951079,
      "learning_rate": 2.8913259823050345e-06,
      "loss": 0.3737,
      "step": 12913
    },
    {
      "epoch": 0.6968862986347202,
      "grad_norm": 1.1016199062589507,
      "learning_rate": 2.890704394741002e-06,
      "loss": 0.5772,
      "step": 12914
    },
    {
      "epoch": 0.6969402622632346,
      "grad_norm": 0.9264686509540279,
      "learning_rate": 2.8900828821707614e-06,
      "loss": 0.4083,
      "step": 12915
    },
    {
      "epoch": 0.696994225891749,
      "grad_norm": 0.8943909375505588,
      "learning_rate": 2.889461444612176e-06,
      "loss": 0.34,
      "step": 12916
    },
    {
      "epoch": 0.6970481895202634,
      "grad_norm": 0.8079244924076091,
      "learning_rate": 2.8888400820831063e-06,
      "loss": 0.3487,
      "step": 12917
    },
    {
      "epoch": 0.6971021531487778,
      "grad_norm": 0.9553139771531703,
      "learning_rate": 2.888218794601412e-06,
      "loss": 0.3316,
      "step": 12918
    },
    {
      "epoch": 0.6971561167772921,
      "grad_norm": 0.881176082034555,
      "learning_rate": 2.887597582184949e-06,
      "loss": 0.3862,
      "step": 12919
    },
    {
      "epoch": 0.6972100804058065,
      "grad_norm": 0.9429409945433771,
      "learning_rate": 2.8869764448515702e-06,
      "loss": 0.3817,
      "step": 12920
    },
    {
      "epoch": 0.6972640440343209,
      "grad_norm": 1.1578707347011084,
      "learning_rate": 2.8863553826191283e-06,
      "loss": 0.5246,
      "step": 12921
    },
    {
      "epoch": 0.6973180076628352,
      "grad_norm": 0.8967520683780582,
      "learning_rate": 2.885734395505474e-06,
      "loss": 0.457,
      "step": 12922
    },
    {
      "epoch": 0.6973719712913496,
      "grad_norm": 0.9932738210975065,
      "learning_rate": 2.8851134835284546e-06,
      "loss": 0.4541,
      "step": 12923
    },
    {
      "epoch": 0.697425934919864,
      "grad_norm": 0.9854958833620842,
      "learning_rate": 2.8844926467059154e-06,
      "loss": 0.5381,
      "step": 12924
    },
    {
      "epoch": 0.6974798985483784,
      "grad_norm": 1.0387134169824275,
      "learning_rate": 2.8838718850557e-06,
      "loss": 0.3801,
      "step": 12925
    },
    {
      "epoch": 0.6975338621768927,
      "grad_norm": 1.05149151172047,
      "learning_rate": 2.8832511985956524e-06,
      "loss": 0.5175,
      "step": 12926
    },
    {
      "epoch": 0.6975878258054071,
      "grad_norm": 1.2602958665132944,
      "learning_rate": 2.8826305873436076e-06,
      "loss": 0.4822,
      "step": 12927
    },
    {
      "epoch": 0.6976417894339215,
      "grad_norm": 1.0789958947609768,
      "learning_rate": 2.8820100513174042e-06,
      "loss": 0.6905,
      "step": 12928
    },
    {
      "epoch": 0.6976957530624359,
      "grad_norm": 0.9858086354837914,
      "learning_rate": 2.881389590534878e-06,
      "loss": 0.5798,
      "step": 12929
    },
    {
      "epoch": 0.6977497166909503,
      "grad_norm": 0.8868523476500845,
      "learning_rate": 2.8807692050138604e-06,
      "loss": 0.3894,
      "step": 12930
    },
    {
      "epoch": 0.6978036803194647,
      "grad_norm": 0.9294987412565143,
      "learning_rate": 2.8801488947721818e-06,
      "loss": 0.3437,
      "step": 12931
    },
    {
      "epoch": 0.6978576439479791,
      "grad_norm": 1.0108922310797102,
      "learning_rate": 2.879528659827672e-06,
      "loss": 0.4687,
      "step": 12932
    },
    {
      "epoch": 0.6979116075764934,
      "grad_norm": 0.9396824300570262,
      "learning_rate": 2.878908500198157e-06,
      "loss": 0.4303,
      "step": 12933
    },
    {
      "epoch": 0.6979655712050078,
      "grad_norm": 0.8925678922445839,
      "learning_rate": 2.878288415901459e-06,
      "loss": 0.4633,
      "step": 12934
    },
    {
      "epoch": 0.6980195348335222,
      "grad_norm": 0.9433177683589723,
      "learning_rate": 2.8776684069554006e-06,
      "loss": 0.3796,
      "step": 12935
    },
    {
      "epoch": 0.6980734984620366,
      "grad_norm": 1.2063324756923042,
      "learning_rate": 2.8770484733778027e-06,
      "loss": 0.5849,
      "step": 12936
    },
    {
      "epoch": 0.698127462090551,
      "grad_norm": 0.7331134858124366,
      "learning_rate": 2.8764286151864818e-06,
      "loss": 0.3961,
      "step": 12937
    },
    {
      "epoch": 0.6981814257190654,
      "grad_norm": 1.1619805067459676,
      "learning_rate": 2.8758088323992527e-06,
      "loss": 0.5107,
      "step": 12938
    },
    {
      "epoch": 0.6982353893475798,
      "grad_norm": 0.8865085221984703,
      "learning_rate": 2.87518912503393e-06,
      "loss": 0.3576,
      "step": 12939
    },
    {
      "epoch": 0.6982893529760941,
      "grad_norm": 1.069890110359377,
      "learning_rate": 2.8745694931083234e-06,
      "loss": 0.4692,
      "step": 12940
    },
    {
      "epoch": 0.6983433166046085,
      "grad_norm": 0.8036546186489738,
      "learning_rate": 2.873949936640243e-06,
      "loss": 0.3559,
      "step": 12941
    },
    {
      "epoch": 0.6983972802331229,
      "grad_norm": 1.1617986736347894,
      "learning_rate": 2.8733304556474946e-06,
      "loss": 0.4084,
      "step": 12942
    },
    {
      "epoch": 0.6984512438616373,
      "grad_norm": 0.9263656535667706,
      "learning_rate": 2.8727110501478827e-06,
      "loss": 0.3834,
      "step": 12943
    },
    {
      "epoch": 0.6985052074901517,
      "grad_norm": 0.9176318156307719,
      "learning_rate": 2.87209172015921e-06,
      "loss": 0.5834,
      "step": 12944
    },
    {
      "epoch": 0.6985591711186661,
      "grad_norm": 0.8373167529559262,
      "learning_rate": 2.8714724656992764e-06,
      "loss": 0.395,
      "step": 12945
    },
    {
      "epoch": 0.6986131347471805,
      "grad_norm": 0.9201747919475872,
      "learning_rate": 2.8708532867858823e-06,
      "loss": 0.3777,
      "step": 12946
    },
    {
      "epoch": 0.6986670983756947,
      "grad_norm": 1.1483053010574185,
      "learning_rate": 2.8702341834368196e-06,
      "loss": 0.419,
      "step": 12947
    },
    {
      "epoch": 0.6987210620042091,
      "grad_norm": 1.085991431174362,
      "learning_rate": 2.8696151556698837e-06,
      "loss": 0.5315,
      "step": 12948
    },
    {
      "epoch": 0.6987750256327235,
      "grad_norm": 1.0244724316698472,
      "learning_rate": 2.8689962035028655e-06,
      "loss": 0.5262,
      "step": 12949
    },
    {
      "epoch": 0.6988289892612379,
      "grad_norm": 1.1116852469469143,
      "learning_rate": 2.8683773269535554e-06,
      "loss": 0.5146,
      "step": 12950
    },
    {
      "epoch": 0.6988829528897523,
      "grad_norm": 0.7131104441066115,
      "learning_rate": 2.8677585260397394e-06,
      "loss": 0.2743,
      "step": 12951
    },
    {
      "epoch": 0.6989369165182667,
      "grad_norm": 1.0910564656413746,
      "learning_rate": 2.8671398007792034e-06,
      "loss": 0.4474,
      "step": 12952
    },
    {
      "epoch": 0.6989908801467811,
      "grad_norm": 0.9544232239703293,
      "learning_rate": 2.8665211511897296e-06,
      "loss": 0.3873,
      "step": 12953
    },
    {
      "epoch": 0.6990448437752954,
      "grad_norm": 1.1432511577477051,
      "learning_rate": 2.8659025772891003e-06,
      "loss": 0.4477,
      "step": 12954
    },
    {
      "epoch": 0.6990988074038098,
      "grad_norm": 0.973986112396438,
      "learning_rate": 2.865284079095092e-06,
      "loss": 0.3751,
      "step": 12955
    },
    {
      "epoch": 0.6991527710323242,
      "grad_norm": 1.2231616023515723,
      "learning_rate": 2.8646656566254803e-06,
      "loss": 0.6086,
      "step": 12956
    },
    {
      "epoch": 0.6992067346608386,
      "grad_norm": 0.9556905245050342,
      "learning_rate": 2.8640473098980404e-06,
      "loss": 0.3298,
      "step": 12957
    },
    {
      "epoch": 0.699260698289353,
      "grad_norm": 1.0391009545463676,
      "learning_rate": 2.8634290389305454e-06,
      "loss": 0.4588,
      "step": 12958
    },
    {
      "epoch": 0.6993146619178674,
      "grad_norm": 1.1271598236666098,
      "learning_rate": 2.8628108437407633e-06,
      "loss": 0.5385,
      "step": 12959
    },
    {
      "epoch": 0.6993686255463817,
      "grad_norm": 1.0909593885095143,
      "learning_rate": 2.862192724346463e-06,
      "loss": 0.5398,
      "step": 12960
    },
    {
      "epoch": 0.6994225891748961,
      "grad_norm": 0.970489551654379,
      "learning_rate": 2.861574680765409e-06,
      "loss": 0.4929,
      "step": 12961
    },
    {
      "epoch": 0.6994765528034105,
      "grad_norm": 0.969708483849196,
      "learning_rate": 2.8609567130153633e-06,
      "loss": 0.465,
      "step": 12962
    },
    {
      "epoch": 0.6995305164319249,
      "grad_norm": 0.9941266954207261,
      "learning_rate": 2.8603388211140886e-06,
      "loss": 0.5499,
      "step": 12963
    },
    {
      "epoch": 0.6995844800604393,
      "grad_norm": 0.9613418576181274,
      "learning_rate": 2.8597210050793435e-06,
      "loss": 0.4431,
      "step": 12964
    },
    {
      "epoch": 0.6996384436889537,
      "grad_norm": 1.0984638216631157,
      "learning_rate": 2.8591032649288846e-06,
      "loss": 0.5586,
      "step": 12965
    },
    {
      "epoch": 0.6996924073174681,
      "grad_norm": 1.012950214666351,
      "learning_rate": 2.858485600680466e-06,
      "loss": 0.5465,
      "step": 12966
    },
    {
      "epoch": 0.6997463709459824,
      "grad_norm": 0.9387914768136617,
      "learning_rate": 2.857868012351842e-06,
      "loss": 0.3329,
      "step": 12967
    },
    {
      "epoch": 0.6998003345744968,
      "grad_norm": 1.175156450389991,
      "learning_rate": 2.8572504999607597e-06,
      "loss": 0.4165,
      "step": 12968
    },
    {
      "epoch": 0.6998542982030111,
      "grad_norm": 0.8894618814742211,
      "learning_rate": 2.8566330635249682e-06,
      "loss": 0.3284,
      "step": 12969
    },
    {
      "epoch": 0.6999082618315255,
      "grad_norm": 1.2161289781084283,
      "learning_rate": 2.856015703062214e-06,
      "loss": 0.4336,
      "step": 12970
    },
    {
      "epoch": 0.6999622254600399,
      "grad_norm": 1.001673379308925,
      "learning_rate": 2.855398418590239e-06,
      "loss": 0.4056,
      "step": 12971
    },
    {
      "epoch": 0.7000161890885543,
      "grad_norm": 1.059691907435687,
      "learning_rate": 2.8547812101267856e-06,
      "loss": 0.4667,
      "step": 12972
    },
    {
      "epoch": 0.7000701527170687,
      "grad_norm": 1.2703593986459272,
      "learning_rate": 2.8541640776895934e-06,
      "loss": 0.5237,
      "step": 12973
    },
    {
      "epoch": 0.700124116345583,
      "grad_norm": 1.0690454179495672,
      "learning_rate": 2.853547021296401e-06,
      "loss": 0.5773,
      "step": 12974
    },
    {
      "epoch": 0.7001780799740974,
      "grad_norm": 0.9152468059966911,
      "learning_rate": 2.8529300409649395e-06,
      "loss": 0.424,
      "step": 12975
    },
    {
      "epoch": 0.7002320436026118,
      "grad_norm": 1.1022082237926278,
      "learning_rate": 2.852313136712943e-06,
      "loss": 0.4881,
      "step": 12976
    },
    {
      "epoch": 0.7002860072311262,
      "grad_norm": 0.8970499536489263,
      "learning_rate": 2.851696308558142e-06,
      "loss": 0.4297,
      "step": 12977
    },
    {
      "epoch": 0.7003399708596406,
      "grad_norm": 0.9714205401431717,
      "learning_rate": 2.8510795565182656e-06,
      "loss": 0.4061,
      "step": 12978
    },
    {
      "epoch": 0.700393934488155,
      "grad_norm": 0.8877727922961122,
      "learning_rate": 2.850462880611039e-06,
      "loss": 0.2947,
      "step": 12979
    },
    {
      "epoch": 0.7004478981166694,
      "grad_norm": 1.1022785289051134,
      "learning_rate": 2.8498462808541865e-06,
      "loss": 0.523,
      "step": 12980
    },
    {
      "epoch": 0.7005018617451837,
      "grad_norm": 0.9110190920895537,
      "learning_rate": 2.8492297572654305e-06,
      "loss": 0.3825,
      "step": 12981
    },
    {
      "epoch": 0.7005558253736981,
      "grad_norm": 1.0714977445417089,
      "learning_rate": 2.848613309862489e-06,
      "loss": 0.427,
      "step": 12982
    },
    {
      "epoch": 0.7006097890022125,
      "grad_norm": 0.7550204507933227,
      "learning_rate": 2.8479969386630796e-06,
      "loss": 0.3099,
      "step": 12983
    },
    {
      "epoch": 0.7006637526307269,
      "grad_norm": 0.8941615655717171,
      "learning_rate": 2.8473806436849183e-06,
      "loss": 0.4074,
      "step": 12984
    },
    {
      "epoch": 0.7007177162592413,
      "grad_norm": 0.9514481769718431,
      "learning_rate": 2.846764424945717e-06,
      "loss": 0.4134,
      "step": 12985
    },
    {
      "epoch": 0.7007716798877557,
      "grad_norm": 0.8784077279945546,
      "learning_rate": 2.8461482824631883e-06,
      "loss": 0.387,
      "step": 12986
    },
    {
      "epoch": 0.7008256435162701,
      "grad_norm": 1.2133829296431922,
      "learning_rate": 2.845532216255038e-06,
      "loss": 0.6764,
      "step": 12987
    },
    {
      "epoch": 0.7008796071447844,
      "grad_norm": 1.23391218345741,
      "learning_rate": 2.844916226338974e-06,
      "loss": 0.5535,
      "step": 12988
    },
    {
      "epoch": 0.7009335707732988,
      "grad_norm": 0.8220678919514699,
      "learning_rate": 2.844300312732701e-06,
      "loss": 0.3129,
      "step": 12989
    },
    {
      "epoch": 0.7009875344018132,
      "grad_norm": 0.890965656113664,
      "learning_rate": 2.8436844754539217e-06,
      "loss": 0.3615,
      "step": 12990
    },
    {
      "epoch": 0.7010414980303276,
      "grad_norm": 1.0985265313520398,
      "learning_rate": 2.843068714520333e-06,
      "loss": 0.5382,
      "step": 12991
    },
    {
      "epoch": 0.701095461658842,
      "grad_norm": 0.8440279723636199,
      "learning_rate": 2.8424530299496345e-06,
      "loss": 0.4153,
      "step": 12992
    },
    {
      "epoch": 0.7011494252873564,
      "grad_norm": 0.9791161487316945,
      "learning_rate": 2.841837421759521e-06,
      "loss": 0.4473,
      "step": 12993
    },
    {
      "epoch": 0.7012033889158708,
      "grad_norm": 1.0787858084713418,
      "learning_rate": 2.8412218899676857e-06,
      "loss": 0.5041,
      "step": 12994
    },
    {
      "epoch": 0.701257352544385,
      "grad_norm": 1.0700533089352169,
      "learning_rate": 2.84060643459182e-06,
      "loss": 0.491,
      "step": 12995
    },
    {
      "epoch": 0.7013113161728994,
      "grad_norm": 1.218940987239314,
      "learning_rate": 2.8399910556496126e-06,
      "loss": 0.6347,
      "step": 12996
    },
    {
      "epoch": 0.7013652798014138,
      "grad_norm": 0.7664033350885981,
      "learning_rate": 2.8393757531587495e-06,
      "loss": 0.3795,
      "step": 12997
    },
    {
      "epoch": 0.7014192434299282,
      "grad_norm": 1.098783766452608,
      "learning_rate": 2.8387605271369156e-06,
      "loss": 0.5272,
      "step": 12998
    },
    {
      "epoch": 0.7014732070584426,
      "grad_norm": 0.9158361199546222,
      "learning_rate": 2.838145377601793e-06,
      "loss": 0.4115,
      "step": 12999
    },
    {
      "epoch": 0.701527170686957,
      "grad_norm": 0.9986176229963939,
      "learning_rate": 2.837530304571063e-06,
      "loss": 0.6343,
      "step": 13000
    },
    {
      "epoch": 0.701527170686957,
      "eval_loss": 0.5355429649353027,
      "eval_runtime": 164.4209,
      "eval_samples_per_second": 20.916,
      "eval_steps_per_second": 0.876,
      "step": 13000
    },
    {
      "epoch": 0.7015811343154714,
      "grad_norm": 0.996595930805883,
      "learning_rate": 2.8369153080624013e-06,
      "loss": 0.3917,
      "step": 13001
    },
    {
      "epoch": 0.7016350979439857,
      "grad_norm": 1.1455880390495106,
      "learning_rate": 2.8363003880934866e-06,
      "loss": 0.5528,
      "step": 13002
    },
    {
      "epoch": 0.7016890615725001,
      "grad_norm": 0.9904917930105941,
      "learning_rate": 2.8356855446819876e-06,
      "loss": 0.4379,
      "step": 13003
    },
    {
      "epoch": 0.7017430252010145,
      "grad_norm": 1.080466738903755,
      "learning_rate": 2.8350707778455794e-06,
      "loss": 0.5141,
      "step": 13004
    },
    {
      "epoch": 0.7017969888295289,
      "grad_norm": 1.1796887113654464,
      "learning_rate": 2.8344560876019293e-06,
      "loss": 0.4489,
      "step": 13005
    },
    {
      "epoch": 0.7018509524580433,
      "grad_norm": 0.8882446696757533,
      "learning_rate": 2.833841473968705e-06,
      "loss": 0.4007,
      "step": 13006
    },
    {
      "epoch": 0.7019049160865577,
      "grad_norm": 0.891459988599545,
      "learning_rate": 2.83322693696357e-06,
      "loss": 0.3809,
      "step": 13007
    },
    {
      "epoch": 0.7019588797150721,
      "grad_norm": 0.8916230932083137,
      "learning_rate": 2.8326124766041875e-06,
      "loss": 0.4695,
      "step": 13008
    },
    {
      "epoch": 0.7020128433435864,
      "grad_norm": 1.1527854167343397,
      "learning_rate": 2.8319980929082192e-06,
      "loss": 0.5072,
      "step": 13009
    },
    {
      "epoch": 0.7020668069721008,
      "grad_norm": 1.0398119058366753,
      "learning_rate": 2.8313837858933204e-06,
      "loss": 0.4467,
      "step": 13010
    },
    {
      "epoch": 0.7021207706006152,
      "grad_norm": 1.029893027929824,
      "learning_rate": 2.8307695555771473e-06,
      "loss": 0.4338,
      "step": 13011
    },
    {
      "epoch": 0.7021747342291296,
      "grad_norm": 0.8124298157360522,
      "learning_rate": 2.830155401977355e-06,
      "loss": 0.4514,
      "step": 13012
    },
    {
      "epoch": 0.702228697857644,
      "grad_norm": 1.0851632947123264,
      "learning_rate": 2.8295413251115933e-06,
      "loss": 0.5173,
      "step": 13013
    },
    {
      "epoch": 0.7022826614861584,
      "grad_norm": 0.7589029577566443,
      "learning_rate": 2.8289273249975123e-06,
      "loss": 0.3112,
      "step": 13014
    },
    {
      "epoch": 0.7023366251146728,
      "grad_norm": 0.9477779394614375,
      "learning_rate": 2.828313401652759e-06,
      "loss": 0.4566,
      "step": 13015
    },
    {
      "epoch": 0.702390588743187,
      "grad_norm": 1.201259040975867,
      "learning_rate": 2.8276995550949806e-06,
      "loss": 0.5038,
      "step": 13016
    },
    {
      "epoch": 0.7024445523717014,
      "grad_norm": 1.0455054158281298,
      "learning_rate": 2.8270857853418155e-06,
      "loss": 0.5357,
      "step": 13017
    },
    {
      "epoch": 0.7024985160002158,
      "grad_norm": 1.1041843674439185,
      "learning_rate": 2.8264720924109045e-06,
      "loss": 0.4489,
      "step": 13018
    },
    {
      "epoch": 0.7025524796287302,
      "grad_norm": 0.9089745667709747,
      "learning_rate": 2.825858476319888e-06,
      "loss": 0.5305,
      "step": 13019
    },
    {
      "epoch": 0.7026064432572446,
      "grad_norm": 0.9939081471060669,
      "learning_rate": 2.8252449370864e-06,
      "loss": 0.3485,
      "step": 13020
    },
    {
      "epoch": 0.702660406885759,
      "grad_norm": 1.1870215038016638,
      "learning_rate": 2.824631474728076e-06,
      "loss": 0.5953,
      "step": 13021
    },
    {
      "epoch": 0.7027143705142734,
      "grad_norm": 1.1182544652513735,
      "learning_rate": 2.824018089262546e-06,
      "loss": 0.5149,
      "step": 13022
    },
    {
      "epoch": 0.7027683341427877,
      "grad_norm": 1.0058210010240727,
      "learning_rate": 2.8234047807074423e-06,
      "loss": 0.5077,
      "step": 13023
    },
    {
      "epoch": 0.7028222977713021,
      "grad_norm": 1.1264014671905709,
      "learning_rate": 2.822791549080388e-06,
      "loss": 0.4612,
      "step": 13024
    },
    {
      "epoch": 0.7028762613998165,
      "grad_norm": 0.8990954914531564,
      "learning_rate": 2.822178394399009e-06,
      "loss": 0.4495,
      "step": 13025
    },
    {
      "epoch": 0.7029302250283309,
      "grad_norm": 1.0984104188104251,
      "learning_rate": 2.821565316680929e-06,
      "loss": 0.4928,
      "step": 13026
    },
    {
      "epoch": 0.7029841886568453,
      "grad_norm": 1.1126935262792854,
      "learning_rate": 2.8209523159437684e-06,
      "loss": 0.4631,
      "step": 13027
    },
    {
      "epoch": 0.7030381522853597,
      "grad_norm": 0.9999328647167958,
      "learning_rate": 2.820339392205145e-06,
      "loss": 0.4031,
      "step": 13028
    },
    {
      "epoch": 0.703092115913874,
      "grad_norm": 1.1306384270140757,
      "learning_rate": 2.8197265454826774e-06,
      "loss": 0.5018,
      "step": 13029
    },
    {
      "epoch": 0.7031460795423884,
      "grad_norm": 0.8735255399762555,
      "learning_rate": 2.8191137757939747e-06,
      "loss": 0.3792,
      "step": 13030
    },
    {
      "epoch": 0.7032000431709028,
      "grad_norm": 1.1949672833678369,
      "learning_rate": 2.8185010831566507e-06,
      "loss": 0.5102,
      "step": 13031
    },
    {
      "epoch": 0.7032540067994172,
      "grad_norm": 1.207977652785274,
      "learning_rate": 2.8178884675883156e-06,
      "loss": 0.5498,
      "step": 13032
    },
    {
      "epoch": 0.7033079704279316,
      "grad_norm": 0.9335739754608943,
      "learning_rate": 2.817275929106576e-06,
      "loss": 0.3556,
      "step": 13033
    },
    {
      "epoch": 0.703361934056446,
      "grad_norm": 0.9663375590565338,
      "learning_rate": 2.816663467729037e-06,
      "loss": 0.4838,
      "step": 13034
    },
    {
      "epoch": 0.7034158976849604,
      "grad_norm": 1.1939507687306725,
      "learning_rate": 2.8160510834733e-06,
      "loss": 0.5332,
      "step": 13035
    },
    {
      "epoch": 0.7034698613134747,
      "grad_norm": 0.8217518074155711,
      "learning_rate": 2.81543877635697e-06,
      "loss": 0.3563,
      "step": 13036
    },
    {
      "epoch": 0.7035238249419891,
      "grad_norm": 1.0986183688642583,
      "learning_rate": 2.8148265463976397e-06,
      "loss": 0.5324,
      "step": 13037
    },
    {
      "epoch": 0.7035777885705035,
      "grad_norm": 0.9195849219557485,
      "learning_rate": 2.8142143936129072e-06,
      "loss": 0.4031,
      "step": 13038
    },
    {
      "epoch": 0.7036317521990179,
      "grad_norm": 1.0471143540401262,
      "learning_rate": 2.813602318020369e-06,
      "loss": 0.4623,
      "step": 13039
    },
    {
      "epoch": 0.7036857158275323,
      "grad_norm": 1.1639418335222713,
      "learning_rate": 2.812990319637612e-06,
      "loss": 0.5683,
      "step": 13040
    },
    {
      "epoch": 0.7037396794560467,
      "grad_norm": 1.0878062013090393,
      "learning_rate": 2.8123783984822296e-06,
      "loss": 0.5384,
      "step": 13041
    },
    {
      "epoch": 0.703793643084561,
      "grad_norm": 1.1279163830832089,
      "learning_rate": 2.8117665545718065e-06,
      "loss": 0.446,
      "step": 13042
    },
    {
      "epoch": 0.7038476067130753,
      "grad_norm": 1.0897483618394614,
      "learning_rate": 2.8111547879239303e-06,
      "loss": 0.5023,
      "step": 13043
    },
    {
      "epoch": 0.7039015703415897,
      "grad_norm": 1.100359865641062,
      "learning_rate": 2.8105430985561816e-06,
      "loss": 0.5427,
      "step": 13044
    },
    {
      "epoch": 0.7039555339701041,
      "grad_norm": 1.0625241814938284,
      "learning_rate": 2.809931486486142e-06,
      "loss": 0.5393,
      "step": 13045
    },
    {
      "epoch": 0.7040094975986185,
      "grad_norm": 0.9675337380134282,
      "learning_rate": 2.8093199517313896e-06,
      "loss": 0.4158,
      "step": 13046
    },
    {
      "epoch": 0.7040634612271329,
      "grad_norm": 0.9838849220195337,
      "learning_rate": 2.8087084943095e-06,
      "loss": 0.4313,
      "step": 13047
    },
    {
      "epoch": 0.7041174248556473,
      "grad_norm": 0.8269455632601661,
      "learning_rate": 2.8080971142380486e-06,
      "loss": 0.4301,
      "step": 13048
    },
    {
      "epoch": 0.7041713884841617,
      "grad_norm": 0.8344177225360069,
      "learning_rate": 2.8074858115346054e-06,
      "loss": 0.3272,
      "step": 13049
    },
    {
      "epoch": 0.704225352112676,
      "grad_norm": 1.0845656158364825,
      "learning_rate": 2.8068745862167423e-06,
      "loss": 0.4704,
      "step": 13050
    },
    {
      "epoch": 0.7042793157411904,
      "grad_norm": 1.2446297392616135,
      "learning_rate": 2.806263438302024e-06,
      "loss": 0.5924,
      "step": 13051
    },
    {
      "epoch": 0.7043332793697048,
      "grad_norm": 0.9584484260979105,
      "learning_rate": 2.805652367808015e-06,
      "loss": 0.3035,
      "step": 13052
    },
    {
      "epoch": 0.7043872429982192,
      "grad_norm": 0.9800382490330949,
      "learning_rate": 2.80504137475228e-06,
      "loss": 0.4242,
      "step": 13053
    },
    {
      "epoch": 0.7044412066267336,
      "grad_norm": 1.0407842585281557,
      "learning_rate": 2.8044304591523785e-06,
      "loss": 0.3801,
      "step": 13054
    },
    {
      "epoch": 0.704495170255248,
      "grad_norm": 0.9326690358173315,
      "learning_rate": 2.80381962102587e-06,
      "loss": 0.4358,
      "step": 13055
    },
    {
      "epoch": 0.7045491338837624,
      "grad_norm": 1.1958010477825447,
      "learning_rate": 2.8032088603903095e-06,
      "loss": 0.4473,
      "step": 13056
    },
    {
      "epoch": 0.7046030975122767,
      "grad_norm": 1.0487652739161568,
      "learning_rate": 2.802598177263253e-06,
      "loss": 0.5553,
      "step": 13057
    },
    {
      "epoch": 0.7046570611407911,
      "grad_norm": 0.9303126085314012,
      "learning_rate": 2.8019875716622484e-06,
      "loss": 0.3733,
      "step": 13058
    },
    {
      "epoch": 0.7047110247693055,
      "grad_norm": 1.1837198431229963,
      "learning_rate": 2.8013770436048483e-06,
      "loss": 0.5919,
      "step": 13059
    },
    {
      "epoch": 0.7047649883978199,
      "grad_norm": 0.8625004885750298,
      "learning_rate": 2.8007665931085974e-06,
      "loss": 0.3274,
      "step": 13060
    },
    {
      "epoch": 0.7048189520263343,
      "grad_norm": 1.0054512199218175,
      "learning_rate": 2.8001562201910422e-06,
      "loss": 0.5327,
      "step": 13061
    },
    {
      "epoch": 0.7048729156548487,
      "grad_norm": 0.9986648037043501,
      "learning_rate": 2.7995459248697256e-06,
      "loss": 0.463,
      "step": 13062
    },
    {
      "epoch": 0.7049268792833631,
      "grad_norm": 0.9521536341535253,
      "learning_rate": 2.7989357071621882e-06,
      "loss": 0.3776,
      "step": 13063
    },
    {
      "epoch": 0.7049808429118773,
      "grad_norm": 1.0524318374431594,
      "learning_rate": 2.7983255670859684e-06,
      "loss": 0.5458,
      "step": 13064
    },
    {
      "epoch": 0.7050348065403917,
      "grad_norm": 0.8403016335417227,
      "learning_rate": 2.7977155046586e-06,
      "loss": 0.3191,
      "step": 13065
    },
    {
      "epoch": 0.7050887701689061,
      "grad_norm": 1.0389740724256136,
      "learning_rate": 2.7971055198976195e-06,
      "loss": 0.5136,
      "step": 13066
    },
    {
      "epoch": 0.7051427337974205,
      "grad_norm": 1.1760540537392443,
      "learning_rate": 2.796495612820557e-06,
      "loss": 0.5008,
      "step": 13067
    },
    {
      "epoch": 0.7051966974259349,
      "grad_norm": 1.2807941817034658,
      "learning_rate": 2.7958857834449417e-06,
      "loss": 0.4062,
      "step": 13068
    },
    {
      "epoch": 0.7052506610544493,
      "grad_norm": 1.218844368223324,
      "learning_rate": 2.795276031788302e-06,
      "loss": 0.5431,
      "step": 13069
    },
    {
      "epoch": 0.7053046246829637,
      "grad_norm": 1.0058268001671038,
      "learning_rate": 2.794666357868161e-06,
      "loss": 0.6061,
      "step": 13070
    },
    {
      "epoch": 0.705358588311478,
      "grad_norm": 0.8122780290388153,
      "learning_rate": 2.7940567617020453e-06,
      "loss": 0.3161,
      "step": 13071
    },
    {
      "epoch": 0.7054125519399924,
      "grad_norm": 0.9530569512651462,
      "learning_rate": 2.79344724330747e-06,
      "loss": 0.327,
      "step": 13072
    },
    {
      "epoch": 0.7054665155685068,
      "grad_norm": 0.9260017557289202,
      "learning_rate": 2.7928378027019564e-06,
      "loss": 0.431,
      "step": 13073
    },
    {
      "epoch": 0.7055204791970212,
      "grad_norm": 0.9286129603474494,
      "learning_rate": 2.7922284399030198e-06,
      "loss": 0.4352,
      "step": 13074
    },
    {
      "epoch": 0.7055744428255356,
      "grad_norm": 1.0853673821383802,
      "learning_rate": 2.791619154928173e-06,
      "loss": 0.6228,
      "step": 13075
    },
    {
      "epoch": 0.70562840645405,
      "grad_norm": 0.9600055297110929,
      "learning_rate": 2.791009947794929e-06,
      "loss": 0.4342,
      "step": 13076
    },
    {
      "epoch": 0.7056823700825644,
      "grad_norm": 1.0943157555824032,
      "learning_rate": 2.7904008185207955e-06,
      "loss": 0.5069,
      "step": 13077
    },
    {
      "epoch": 0.7057363337110787,
      "grad_norm": 1.1970715513857342,
      "learning_rate": 2.7897917671232825e-06,
      "loss": 0.6761,
      "step": 13078
    },
    {
      "epoch": 0.7057902973395931,
      "grad_norm": 0.9945706429572404,
      "learning_rate": 2.7891827936198914e-06,
      "loss": 0.4484,
      "step": 13079
    },
    {
      "epoch": 0.7058442609681075,
      "grad_norm": 0.9780944096479152,
      "learning_rate": 2.788573898028125e-06,
      "loss": 0.4552,
      "step": 13080
    },
    {
      "epoch": 0.7058982245966219,
      "grad_norm": 1.132829691964871,
      "learning_rate": 2.7879650803654856e-06,
      "loss": 0.4845,
      "step": 13081
    },
    {
      "epoch": 0.7059521882251363,
      "grad_norm": 1.0797041445849078,
      "learning_rate": 2.7873563406494687e-06,
      "loss": 0.4318,
      "step": 13082
    },
    {
      "epoch": 0.7060061518536507,
      "grad_norm": 1.2193800457267148,
      "learning_rate": 2.7867476788975727e-06,
      "loss": 0.4477,
      "step": 13083
    },
    {
      "epoch": 0.7060601154821651,
      "grad_norm": 1.1652198594242897,
      "learning_rate": 2.7861390951272895e-06,
      "loss": 0.4304,
      "step": 13084
    },
    {
      "epoch": 0.7061140791106794,
      "grad_norm": 1.1546969970782421,
      "learning_rate": 2.7855305893561123e-06,
      "loss": 0.4276,
      "step": 13085
    },
    {
      "epoch": 0.7061680427391938,
      "grad_norm": 1.0083104303102652,
      "learning_rate": 2.7849221616015277e-06,
      "loss": 0.4745,
      "step": 13086
    },
    {
      "epoch": 0.7062220063677082,
      "grad_norm": 1.0719212171403862,
      "learning_rate": 2.784313811881023e-06,
      "loss": 0.4861,
      "step": 13087
    },
    {
      "epoch": 0.7062759699962226,
      "grad_norm": 1.0346907874861577,
      "learning_rate": 2.7837055402120834e-06,
      "loss": 0.5629,
      "step": 13088
    },
    {
      "epoch": 0.706329933624737,
      "grad_norm": 1.1782964167291268,
      "learning_rate": 2.7830973466121914e-06,
      "loss": 0.7236,
      "step": 13089
    },
    {
      "epoch": 0.7063838972532513,
      "grad_norm": 1.0755119645720623,
      "learning_rate": 2.782489231098826e-06,
      "loss": 0.4281,
      "step": 13090
    },
    {
      "epoch": 0.7064378608817657,
      "grad_norm": 0.7670640174505496,
      "learning_rate": 2.781881193689467e-06,
      "loss": 0.2882,
      "step": 13091
    },
    {
      "epoch": 0.70649182451028,
      "grad_norm": 1.1217362835961913,
      "learning_rate": 2.7812732344015885e-06,
      "loss": 0.4635,
      "step": 13092
    },
    {
      "epoch": 0.7065457881387944,
      "grad_norm": 0.8639041525163884,
      "learning_rate": 2.780665353252665e-06,
      "loss": 0.3782,
      "step": 13093
    },
    {
      "epoch": 0.7065997517673088,
      "grad_norm": 1.035864339908178,
      "learning_rate": 2.7800575502601664e-06,
      "loss": 0.338,
      "step": 13094
    },
    {
      "epoch": 0.7066537153958232,
      "grad_norm": 0.7941475987212196,
      "learning_rate": 2.7794498254415636e-06,
      "loss": 0.3186,
      "step": 13095
    },
    {
      "epoch": 0.7067076790243376,
      "grad_norm": 1.188104109316251,
      "learning_rate": 2.7788421788143204e-06,
      "loss": 0.5806,
      "step": 13096
    },
    {
      "epoch": 0.706761642652852,
      "grad_norm": 1.1260155836654815,
      "learning_rate": 2.778234610395902e-06,
      "loss": 0.589,
      "step": 13097
    },
    {
      "epoch": 0.7068156062813663,
      "grad_norm": 0.9264751450333825,
      "learning_rate": 2.777627120203771e-06,
      "loss": 0.403,
      "step": 13098
    },
    {
      "epoch": 0.7068695699098807,
      "grad_norm": 1.130933006404503,
      "learning_rate": 2.777019708255388e-06,
      "loss": 0.5611,
      "step": 13099
    },
    {
      "epoch": 0.7069235335383951,
      "grad_norm": 1.0464913644176592,
      "learning_rate": 2.7764123745682096e-06,
      "loss": 0.5156,
      "step": 13100
    },
    {
      "epoch": 0.7069774971669095,
      "grad_norm": 1.0263637991212162,
      "learning_rate": 2.7758051191596918e-06,
      "loss": 0.3585,
      "step": 13101
    },
    {
      "epoch": 0.7070314607954239,
      "grad_norm": 1.0153809608560864,
      "learning_rate": 2.7751979420472874e-06,
      "loss": 0.3887,
      "step": 13102
    },
    {
      "epoch": 0.7070854244239383,
      "grad_norm": 0.9950758524934253,
      "learning_rate": 2.774590843248447e-06,
      "loss": 0.4882,
      "step": 13103
    },
    {
      "epoch": 0.7071393880524527,
      "grad_norm": 1.1429176368978327,
      "learning_rate": 2.7739838227806196e-06,
      "loss": 0.447,
      "step": 13104
    },
    {
      "epoch": 0.707193351680967,
      "grad_norm": 1.038826650481438,
      "learning_rate": 2.7733768806612536e-06,
      "loss": 0.4092,
      "step": 13105
    },
    {
      "epoch": 0.7072473153094814,
      "grad_norm": 1.014656447513123,
      "learning_rate": 2.7727700169077885e-06,
      "loss": 0.4617,
      "step": 13106
    },
    {
      "epoch": 0.7073012789379958,
      "grad_norm": 0.938494170913895,
      "learning_rate": 2.7721632315376703e-06,
      "loss": 0.3603,
      "step": 13107
    },
    {
      "epoch": 0.7073552425665102,
      "grad_norm": 1.0402742957516007,
      "learning_rate": 2.7715565245683364e-06,
      "loss": 0.3637,
      "step": 13108
    },
    {
      "epoch": 0.7074092061950246,
      "grad_norm": 0.867719340477086,
      "learning_rate": 2.7709498960172254e-06,
      "loss": 0.4548,
      "step": 13109
    },
    {
      "epoch": 0.707463169823539,
      "grad_norm": 1.132188636364043,
      "learning_rate": 2.7703433459017703e-06,
      "loss": 0.624,
      "step": 13110
    },
    {
      "epoch": 0.7075171334520534,
      "grad_norm": 1.0064525796031931,
      "learning_rate": 2.769736874239407e-06,
      "loss": 0.4635,
      "step": 13111
    },
    {
      "epoch": 0.7075710970805676,
      "grad_norm": 1.290136001042839,
      "learning_rate": 2.7691304810475662e-06,
      "loss": 0.6072,
      "step": 13112
    },
    {
      "epoch": 0.707625060709082,
      "grad_norm": 1.1306427880447638,
      "learning_rate": 2.7685241663436722e-06,
      "loss": 0.5081,
      "step": 13113
    },
    {
      "epoch": 0.7076790243375964,
      "grad_norm": 1.0252672085650592,
      "learning_rate": 2.767917930145154e-06,
      "loss": 0.5017,
      "step": 13114
    },
    {
      "epoch": 0.7077329879661108,
      "grad_norm": 0.9652394481129307,
      "learning_rate": 2.767311772469436e-06,
      "loss": 0.3537,
      "step": 13115
    },
    {
      "epoch": 0.7077869515946252,
      "grad_norm": 1.0607820044583118,
      "learning_rate": 2.7667056933339376e-06,
      "loss": 0.4746,
      "step": 13116
    },
    {
      "epoch": 0.7078409152231396,
      "grad_norm": 1.0565518759386583,
      "learning_rate": 2.76609969275608e-06,
      "loss": 0.5209,
      "step": 13117
    },
    {
      "epoch": 0.707894878851654,
      "grad_norm": 1.0857899010834557,
      "learning_rate": 2.7654937707532796e-06,
      "loss": 0.4381,
      "step": 13118
    },
    {
      "epoch": 0.7079488424801683,
      "grad_norm": 1.0950508502069594,
      "learning_rate": 2.764887927342953e-06,
      "loss": 0.5571,
      "step": 13119
    },
    {
      "epoch": 0.7080028061086827,
      "grad_norm": 1.1787340086910514,
      "learning_rate": 2.7642821625425088e-06,
      "loss": 0.4198,
      "step": 13120
    },
    {
      "epoch": 0.7080567697371971,
      "grad_norm": 0.9519026302125825,
      "learning_rate": 2.7636764763693603e-06,
      "loss": 0.3771,
      "step": 13121
    },
    {
      "epoch": 0.7081107333657115,
      "grad_norm": 1.008771328759152,
      "learning_rate": 2.763070868840914e-06,
      "loss": 0.469,
      "step": 13122
    },
    {
      "epoch": 0.7081646969942259,
      "grad_norm": 1.0920876976572427,
      "learning_rate": 2.762465339974576e-06,
      "loss": 0.4483,
      "step": 13123
    },
    {
      "epoch": 0.7082186606227403,
      "grad_norm": 0.9315602994409745,
      "learning_rate": 2.7618598897877515e-06,
      "loss": 0.4799,
      "step": 13124
    },
    {
      "epoch": 0.7082726242512547,
      "grad_norm": 1.1172879915176048,
      "learning_rate": 2.76125451829784e-06,
      "loss": 0.617,
      "step": 13125
    },
    {
      "epoch": 0.708326587879769,
      "grad_norm": 1.228137830106851,
      "learning_rate": 2.7606492255222416e-06,
      "loss": 0.5113,
      "step": 13126
    },
    {
      "epoch": 0.7083805515082834,
      "grad_norm": 1.0466773014911352,
      "learning_rate": 2.760044011478352e-06,
      "loss": 0.5105,
      "step": 13127
    },
    {
      "epoch": 0.7084345151367978,
      "grad_norm": 0.9213181581164044,
      "learning_rate": 2.7594388761835654e-06,
      "loss": 0.396,
      "step": 13128
    },
    {
      "epoch": 0.7084884787653122,
      "grad_norm": 0.9402498865698745,
      "learning_rate": 2.7588338196552746e-06,
      "loss": 0.3724,
      "step": 13129
    },
    {
      "epoch": 0.7085424423938266,
      "grad_norm": 0.948868159528535,
      "learning_rate": 2.75822884191087e-06,
      "loss": 0.5209,
      "step": 13130
    },
    {
      "epoch": 0.708596406022341,
      "grad_norm": 1.1827079020074853,
      "learning_rate": 2.757623942967739e-06,
      "loss": 0.5946,
      "step": 13131
    },
    {
      "epoch": 0.7086503696508554,
      "grad_norm": 0.9596532693623243,
      "learning_rate": 2.757019122843266e-06,
      "loss": 0.3987,
      "step": 13132
    },
    {
      "epoch": 0.7087043332793697,
      "grad_norm": 1.0506910640144465,
      "learning_rate": 2.756414381554837e-06,
      "loss": 0.5308,
      "step": 13133
    },
    {
      "epoch": 0.7087582969078841,
      "grad_norm": 0.8941855824360141,
      "learning_rate": 2.755809719119829e-06,
      "loss": 0.4266,
      "step": 13134
    },
    {
      "epoch": 0.7088122605363985,
      "grad_norm": 0.908570304879144,
      "learning_rate": 2.755205135555623e-06,
      "loss": 0.4105,
      "step": 13135
    },
    {
      "epoch": 0.7088662241649129,
      "grad_norm": 1.1165462413666525,
      "learning_rate": 2.754600630879595e-06,
      "loss": 0.4527,
      "step": 13136
    },
    {
      "epoch": 0.7089201877934272,
      "grad_norm": 0.9972159802373246,
      "learning_rate": 2.7539962051091185e-06,
      "loss": 0.4058,
      "step": 13137
    },
    {
      "epoch": 0.7089741514219416,
      "grad_norm": 0.9823415793683125,
      "learning_rate": 2.7533918582615656e-06,
      "loss": 0.4663,
      "step": 13138
    },
    {
      "epoch": 0.709028115050456,
      "grad_norm": 0.9226849699131008,
      "learning_rate": 2.7527875903543056e-06,
      "loss": 0.4869,
      "step": 13139
    },
    {
      "epoch": 0.7090820786789703,
      "grad_norm": 0.8184316444486526,
      "learning_rate": 2.7521834014047084e-06,
      "loss": 0.3948,
      "step": 13140
    },
    {
      "epoch": 0.7091360423074847,
      "grad_norm": 0.939832697571284,
      "learning_rate": 2.7515792914301348e-06,
      "loss": 0.4826,
      "step": 13141
    },
    {
      "epoch": 0.7091900059359991,
      "grad_norm": 0.8606581684069843,
      "learning_rate": 2.7509752604479488e-06,
      "loss": 0.4273,
      "step": 13142
    },
    {
      "epoch": 0.7092439695645135,
      "grad_norm": 0.9211513463204489,
      "learning_rate": 2.750371308475513e-06,
      "loss": 0.3476,
      "step": 13143
    },
    {
      "epoch": 0.7092979331930279,
      "grad_norm": 1.4164502755600485,
      "learning_rate": 2.749767435530183e-06,
      "loss": 0.4202,
      "step": 13144
    },
    {
      "epoch": 0.7093518968215423,
      "grad_norm": 1.0565958109696345,
      "learning_rate": 2.7491636416293167e-06,
      "loss": 0.4839,
      "step": 13145
    },
    {
      "epoch": 0.7094058604500567,
      "grad_norm": 0.8483346139147242,
      "learning_rate": 2.748559926790265e-06,
      "loss": 0.3286,
      "step": 13146
    },
    {
      "epoch": 0.709459824078571,
      "grad_norm": 1.3584298363804665,
      "learning_rate": 2.7479562910303824e-06,
      "loss": 0.659,
      "step": 13147
    },
    {
      "epoch": 0.7095137877070854,
      "grad_norm": 1.2129510573535576,
      "learning_rate": 2.7473527343670177e-06,
      "loss": 0.5291,
      "step": 13148
    },
    {
      "epoch": 0.7095677513355998,
      "grad_norm": 1.082046236715628,
      "learning_rate": 2.7467492568175154e-06,
      "loss": 0.4789,
      "step": 13149
    },
    {
      "epoch": 0.7096217149641142,
      "grad_norm": 0.9614485312838701,
      "learning_rate": 2.746145858399222e-06,
      "loss": 0.3866,
      "step": 13150
    },
    {
      "epoch": 0.7096756785926286,
      "grad_norm": 0.7744276445603744,
      "learning_rate": 2.7455425391294788e-06,
      "loss": 0.314,
      "step": 13151
    },
    {
      "epoch": 0.709729642221143,
      "grad_norm": 0.9736422069050968,
      "learning_rate": 2.7449392990256263e-06,
      "loss": 0.4035,
      "step": 13152
    },
    {
      "epoch": 0.7097836058496574,
      "grad_norm": 1.2184246301286172,
      "learning_rate": 2.744336138105002e-06,
      "loss": 0.5237,
      "step": 13153
    },
    {
      "epoch": 0.7098375694781717,
      "grad_norm": 1.033028820375112,
      "learning_rate": 2.743733056384944e-06,
      "loss": 0.4431,
      "step": 13154
    },
    {
      "epoch": 0.7098915331066861,
      "grad_norm": 1.1337068116553453,
      "learning_rate": 2.7431300538827804e-06,
      "loss": 0.5004,
      "step": 13155
    },
    {
      "epoch": 0.7099454967352005,
      "grad_norm": 0.9275733233124511,
      "learning_rate": 2.742527130615845e-06,
      "loss": 0.5036,
      "step": 13156
    },
    {
      "epoch": 0.7099994603637149,
      "grad_norm": 1.1560844004136976,
      "learning_rate": 2.741924286601465e-06,
      "loss": 0.6425,
      "step": 13157
    },
    {
      "epoch": 0.7100534239922293,
      "grad_norm": 0.9567361019887721,
      "learning_rate": 2.74132152185697e-06,
      "loss": 0.4875,
      "step": 13158
    },
    {
      "epoch": 0.7101073876207437,
      "grad_norm": 0.8216279308959288,
      "learning_rate": 2.74071883639968e-06,
      "loss": 0.3134,
      "step": 13159
    },
    {
      "epoch": 0.710161351249258,
      "grad_norm": 0.9452424854796572,
      "learning_rate": 2.74011623024692e-06,
      "loss": 0.4339,
      "step": 13160
    },
    {
      "epoch": 0.7102153148777723,
      "grad_norm": 0.9288022523712728,
      "learning_rate": 2.7395137034160093e-06,
      "loss": 0.5035,
      "step": 13161
    },
    {
      "epoch": 0.7102692785062867,
      "grad_norm": 1.045709780918447,
      "learning_rate": 2.738911255924262e-06,
      "loss": 0.4372,
      "step": 13162
    },
    {
      "epoch": 0.7103232421348011,
      "grad_norm": 1.06511866770516,
      "learning_rate": 2.7383088877889958e-06,
      "loss": 0.53,
      "step": 13163
    },
    {
      "epoch": 0.7103772057633155,
      "grad_norm": 1.0216732842678784,
      "learning_rate": 2.737706599027522e-06,
      "loss": 0.4532,
      "step": 13164
    },
    {
      "epoch": 0.7104311693918299,
      "grad_norm": 1.0236943747870901,
      "learning_rate": 2.7371043896571516e-06,
      "loss": 0.4097,
      "step": 13165
    },
    {
      "epoch": 0.7104851330203443,
      "grad_norm": 0.9954614095793402,
      "learning_rate": 2.7365022596951936e-06,
      "loss": 0.4992,
      "step": 13166
    },
    {
      "epoch": 0.7105390966488586,
      "grad_norm": 0.8734019779904878,
      "learning_rate": 2.7359002091589535e-06,
      "loss": 0.3375,
      "step": 13167
    },
    {
      "epoch": 0.710593060277373,
      "grad_norm": 0.9031312112383583,
      "learning_rate": 2.735298238065733e-06,
      "loss": 0.4714,
      "step": 13168
    },
    {
      "epoch": 0.7106470239058874,
      "grad_norm": 1.0508096502280095,
      "learning_rate": 2.734696346432836e-06,
      "loss": 0.5285,
      "step": 13169
    },
    {
      "epoch": 0.7107009875344018,
      "grad_norm": 0.8967519229845292,
      "learning_rate": 2.734094534277559e-06,
      "loss": 0.3191,
      "step": 13170
    },
    {
      "epoch": 0.7107549511629162,
      "grad_norm": 1.1246549861554855,
      "learning_rate": 2.7334928016171997e-06,
      "loss": 0.5259,
      "step": 13171
    },
    {
      "epoch": 0.7108089147914306,
      "grad_norm": 1.2063027011437977,
      "learning_rate": 2.7328911484690527e-06,
      "loss": 0.5941,
      "step": 13172
    },
    {
      "epoch": 0.710862878419945,
      "grad_norm": 1.1729411236156362,
      "learning_rate": 2.73228957485041e-06,
      "loss": 0.5991,
      "step": 13173
    },
    {
      "epoch": 0.7109168420484593,
      "grad_norm": 0.9226914194662332,
      "learning_rate": 2.7316880807785635e-06,
      "loss": 0.3421,
      "step": 13174
    },
    {
      "epoch": 0.7109708056769737,
      "grad_norm": 1.1024220332327437,
      "learning_rate": 2.731086666270797e-06,
      "loss": 0.5276,
      "step": 13175
    },
    {
      "epoch": 0.7110247693054881,
      "grad_norm": 1.024129773770123,
      "learning_rate": 2.730485331344397e-06,
      "loss": 0.3344,
      "step": 13176
    },
    {
      "epoch": 0.7110787329340025,
      "grad_norm": 1.4343394273855226,
      "learning_rate": 2.7298840760166473e-06,
      "loss": 0.5688,
      "step": 13177
    },
    {
      "epoch": 0.7111326965625169,
      "grad_norm": 1.10963471045379,
      "learning_rate": 2.729282900304828e-06,
      "loss": 0.6018,
      "step": 13178
    },
    {
      "epoch": 0.7111866601910313,
      "grad_norm": 1.2385169566013434,
      "learning_rate": 2.7286818042262187e-06,
      "loss": 0.5205,
      "step": 13179
    },
    {
      "epoch": 0.7112406238195457,
      "grad_norm": 0.8931774110757966,
      "learning_rate": 2.7280807877980936e-06,
      "loss": 0.4131,
      "step": 13180
    },
    {
      "epoch": 0.71129458744806,
      "grad_norm": 0.9778783547179932,
      "learning_rate": 2.72747985103773e-06,
      "loss": 0.3661,
      "step": 13181
    },
    {
      "epoch": 0.7113485510765744,
      "grad_norm": 0.9010223076726723,
      "learning_rate": 2.726878993962394e-06,
      "loss": 0.4383,
      "step": 13182
    },
    {
      "epoch": 0.7114025147050888,
      "grad_norm": 0.8810799059699322,
      "learning_rate": 2.7262782165893586e-06,
      "loss": 0.3539,
      "step": 13183
    },
    {
      "epoch": 0.7114564783336031,
      "grad_norm": 0.9534954373416316,
      "learning_rate": 2.7256775189358897e-06,
      "loss": 0.3971,
      "step": 13184
    },
    {
      "epoch": 0.7115104419621175,
      "grad_norm": 1.0812430628566978,
      "learning_rate": 2.7250769010192526e-06,
      "loss": 0.4879,
      "step": 13185
    },
    {
      "epoch": 0.7115644055906319,
      "grad_norm": 1.1976230073098244,
      "learning_rate": 2.7244763628567085e-06,
      "loss": 0.5887,
      "step": 13186
    },
    {
      "epoch": 0.7116183692191463,
      "grad_norm": 1.2195072289533524,
      "learning_rate": 2.723875904465519e-06,
      "loss": 0.555,
      "step": 13187
    },
    {
      "epoch": 0.7116723328476606,
      "grad_norm": 0.8142273372093012,
      "learning_rate": 2.723275525862942e-06,
      "loss": 0.3068,
      "step": 13188
    },
    {
      "epoch": 0.711726296476175,
      "grad_norm": 1.021227290198614,
      "learning_rate": 2.7226752270662306e-06,
      "loss": 0.5165,
      "step": 13189
    },
    {
      "epoch": 0.7117802601046894,
      "grad_norm": 1.1626099098948026,
      "learning_rate": 2.7220750080926393e-06,
      "loss": 0.5572,
      "step": 13190
    },
    {
      "epoch": 0.7118342237332038,
      "grad_norm": 1.2086179796841363,
      "learning_rate": 2.7214748689594194e-06,
      "loss": 0.5259,
      "step": 13191
    },
    {
      "epoch": 0.7118881873617182,
      "grad_norm": 1.0826830374526055,
      "learning_rate": 2.7208748096838195e-06,
      "loss": 0.3927,
      "step": 13192
    },
    {
      "epoch": 0.7119421509902326,
      "grad_norm": 0.8508746325058004,
      "learning_rate": 2.720274830283085e-06,
      "loss": 0.372,
      "step": 13193
    },
    {
      "epoch": 0.711996114618747,
      "grad_norm": 1.1286694542653122,
      "learning_rate": 2.7196749307744607e-06,
      "loss": 0.4576,
      "step": 13194
    },
    {
      "epoch": 0.7120500782472613,
      "grad_norm": 0.8238344051080262,
      "learning_rate": 2.719075111175189e-06,
      "loss": 0.3493,
      "step": 13195
    },
    {
      "epoch": 0.7121040418757757,
      "grad_norm": 0.8489989396531308,
      "learning_rate": 2.7184753715025082e-06,
      "loss": 0.4733,
      "step": 13196
    },
    {
      "epoch": 0.7121580055042901,
      "grad_norm": 1.2966222862804193,
      "learning_rate": 2.7178757117736553e-06,
      "loss": 0.5777,
      "step": 13197
    },
    {
      "epoch": 0.7122119691328045,
      "grad_norm": 0.9836424656808218,
      "learning_rate": 2.717276132005866e-06,
      "loss": 0.5234,
      "step": 13198
    },
    {
      "epoch": 0.7122659327613189,
      "grad_norm": 1.217110315896518,
      "learning_rate": 2.7166766322163717e-06,
      "loss": 0.4991,
      "step": 13199
    },
    {
      "epoch": 0.7123198963898333,
      "grad_norm": 0.9566030069736929,
      "learning_rate": 2.7160772124224044e-06,
      "loss": 0.3986,
      "step": 13200
    },
    {
      "epoch": 0.7123738600183477,
      "grad_norm": 0.8557981701131225,
      "learning_rate": 2.715477872641192e-06,
      "loss": 0.3708,
      "step": 13201
    },
    {
      "epoch": 0.712427823646862,
      "grad_norm": 0.749710224553375,
      "learning_rate": 2.714878612889957e-06,
      "loss": 0.3783,
      "step": 13202
    },
    {
      "epoch": 0.7124817872753764,
      "grad_norm": 0.9413269959907998,
      "learning_rate": 2.7142794331859257e-06,
      "loss": 0.4261,
      "step": 13203
    },
    {
      "epoch": 0.7125357509038908,
      "grad_norm": 1.0033722914318448,
      "learning_rate": 2.713680333546318e-06,
      "loss": 0.4768,
      "step": 13204
    },
    {
      "epoch": 0.7125897145324052,
      "grad_norm": 0.7632303804658391,
      "learning_rate": 2.7130813139883526e-06,
      "loss": 0.3219,
      "step": 13205
    },
    {
      "epoch": 0.7126436781609196,
      "grad_norm": 1.1844773485858,
      "learning_rate": 2.712482374529247e-06,
      "loss": 0.4821,
      "step": 13206
    },
    {
      "epoch": 0.712697641789434,
      "grad_norm": 0.833220695642105,
      "learning_rate": 2.7118835151862137e-06,
      "loss": 0.3437,
      "step": 13207
    },
    {
      "epoch": 0.7127516054179484,
      "grad_norm": 1.0379164798607257,
      "learning_rate": 2.711284735976465e-06,
      "loss": 0.6064,
      "step": 13208
    },
    {
      "epoch": 0.7128055690464626,
      "grad_norm": 1.055102803815856,
      "learning_rate": 2.7106860369172133e-06,
      "loss": 0.4022,
      "step": 13209
    },
    {
      "epoch": 0.712859532674977,
      "grad_norm": 0.7933042505184966,
      "learning_rate": 2.710087418025662e-06,
      "loss": 0.3497,
      "step": 13210
    },
    {
      "epoch": 0.7129134963034914,
      "grad_norm": 1.4548406895902832,
      "learning_rate": 2.709488879319016e-06,
      "loss": 0.6629,
      "step": 13211
    },
    {
      "epoch": 0.7129674599320058,
      "grad_norm": 1.0054688760578265,
      "learning_rate": 2.7088904208144793e-06,
      "loss": 0.5057,
      "step": 13212
    },
    {
      "epoch": 0.7130214235605202,
      "grad_norm": 1.2040285557074921,
      "learning_rate": 2.7082920425292534e-06,
      "loss": 0.4262,
      "step": 13213
    },
    {
      "epoch": 0.7130753871890346,
      "grad_norm": 1.2466780856408375,
      "learning_rate": 2.707693744480534e-06,
      "loss": 0.3894,
      "step": 13214
    },
    {
      "epoch": 0.713129350817549,
      "grad_norm": 0.8416414274029633,
      "learning_rate": 2.707095526685518e-06,
      "loss": 0.3926,
      "step": 13215
    },
    {
      "epoch": 0.7131833144460633,
      "grad_norm": 1.0381739717997371,
      "learning_rate": 2.7064973891614004e-06,
      "loss": 0.4493,
      "step": 13216
    },
    {
      "epoch": 0.7132372780745777,
      "grad_norm": 0.7904088485494563,
      "learning_rate": 2.705899331925369e-06,
      "loss": 0.3001,
      "step": 13217
    },
    {
      "epoch": 0.7132912417030921,
      "grad_norm": 1.0782919848152703,
      "learning_rate": 2.705301354994614e-06,
      "loss": 0.5077,
      "step": 13218
    },
    {
      "epoch": 0.7133452053316065,
      "grad_norm": 0.8766498142978533,
      "learning_rate": 2.7047034583863217e-06,
      "loss": 0.4246,
      "step": 13219
    },
    {
      "epoch": 0.7133991689601209,
      "grad_norm": 1.0510952382999337,
      "learning_rate": 2.704105642117677e-06,
      "loss": 0.5168,
      "step": 13220
    },
    {
      "epoch": 0.7134531325886353,
      "grad_norm": 0.9763845969440835,
      "learning_rate": 2.70350790620586e-06,
      "loss": 0.4663,
      "step": 13221
    },
    {
      "epoch": 0.7135070962171497,
      "grad_norm": 1.0595809261484235,
      "learning_rate": 2.702910250668052e-06,
      "loss": 0.6219,
      "step": 13222
    },
    {
      "epoch": 0.713561059845664,
      "grad_norm": 1.282392960561827,
      "learning_rate": 2.7023126755214312e-06,
      "loss": 0.5643,
      "step": 13223
    },
    {
      "epoch": 0.7136150234741784,
      "grad_norm": 1.0074172142399567,
      "learning_rate": 2.70171518078317e-06,
      "loss": 0.5517,
      "step": 13224
    },
    {
      "epoch": 0.7136689871026928,
      "grad_norm": 0.8116301615913486,
      "learning_rate": 2.7011177664704414e-06,
      "loss": 0.3993,
      "step": 13225
    },
    {
      "epoch": 0.7137229507312072,
      "grad_norm": 1.019622224102281,
      "learning_rate": 2.7005204326004166e-06,
      "loss": 0.6558,
      "step": 13226
    },
    {
      "epoch": 0.7137769143597216,
      "grad_norm": 0.9400987742206453,
      "learning_rate": 2.6999231791902623e-06,
      "loss": 0.4032,
      "step": 13227
    },
    {
      "epoch": 0.713830877988236,
      "grad_norm": 0.945807513515949,
      "learning_rate": 2.699326006257146e-06,
      "loss": 0.4073,
      "step": 13228
    },
    {
      "epoch": 0.7138848416167504,
      "grad_norm": 0.8547156582551306,
      "learning_rate": 2.69872891381823e-06,
      "loss": 0.3704,
      "step": 13229
    },
    {
      "epoch": 0.7139388052452647,
      "grad_norm": 1.0035580184531196,
      "learning_rate": 2.6981319018906772e-06,
      "loss": 0.5091,
      "step": 13230
    },
    {
      "epoch": 0.713992768873779,
      "grad_norm": 1.0324455148475713,
      "learning_rate": 2.697534970491642e-06,
      "loss": 0.5444,
      "step": 13231
    },
    {
      "epoch": 0.7140467325022934,
      "grad_norm": 1.2333248628014588,
      "learning_rate": 2.696938119638285e-06,
      "loss": 0.5209,
      "step": 13232
    },
    {
      "epoch": 0.7141006961308078,
      "grad_norm": 0.9664834641863045,
      "learning_rate": 2.6963413493477573e-06,
      "loss": 0.4183,
      "step": 13233
    },
    {
      "epoch": 0.7141546597593222,
      "grad_norm": 1.0841694859852071,
      "learning_rate": 2.695744659637213e-06,
      "loss": 0.4471,
      "step": 13234
    },
    {
      "epoch": 0.7142086233878366,
      "grad_norm": 0.9644334911967835,
      "learning_rate": 2.6951480505238004e-06,
      "loss": 0.4466,
      "step": 13235
    },
    {
      "epoch": 0.7142625870163509,
      "grad_norm": 0.8398925691314961,
      "learning_rate": 2.6945515220246687e-06,
      "loss": 0.3561,
      "step": 13236
    },
    {
      "epoch": 0.7143165506448653,
      "grad_norm": 1.0424603529217031,
      "learning_rate": 2.6939550741569593e-06,
      "loss": 0.4136,
      "step": 13237
    },
    {
      "epoch": 0.7143705142733797,
      "grad_norm": 0.9987975578461854,
      "learning_rate": 2.6933587069378165e-06,
      "loss": 0.5871,
      "step": 13238
    },
    {
      "epoch": 0.7144244779018941,
      "grad_norm": 0.7938424329959459,
      "learning_rate": 2.6927624203843792e-06,
      "loss": 0.3353,
      "step": 13239
    },
    {
      "epoch": 0.7144784415304085,
      "grad_norm": 0.8787651742520503,
      "learning_rate": 2.6921662145137885e-06,
      "loss": 0.4455,
      "step": 13240
    },
    {
      "epoch": 0.7145324051589229,
      "grad_norm": 1.2430738006924706,
      "learning_rate": 2.6915700893431763e-06,
      "loss": 0.6142,
      "step": 13241
    },
    {
      "epoch": 0.7145863687874373,
      "grad_norm": 1.092518004786396,
      "learning_rate": 2.690974044889678e-06,
      "loss": 0.6072,
      "step": 13242
    },
    {
      "epoch": 0.7146403324159516,
      "grad_norm": 1.1952128044946557,
      "learning_rate": 2.690378081170424e-06,
      "loss": 0.5104,
      "step": 13243
    },
    {
      "epoch": 0.714694296044466,
      "grad_norm": 0.7334551820438912,
      "learning_rate": 2.689782198202543e-06,
      "loss": 0.2768,
      "step": 13244
    },
    {
      "epoch": 0.7147482596729804,
      "grad_norm": 1.119554756953071,
      "learning_rate": 2.6891863960031626e-06,
      "loss": 0.5603,
      "step": 13245
    },
    {
      "epoch": 0.7148022233014948,
      "grad_norm": 0.9232546776617528,
      "learning_rate": 2.688590674589403e-06,
      "loss": 0.3676,
      "step": 13246
    },
    {
      "epoch": 0.7148561869300092,
      "grad_norm": 0.9233913418182438,
      "learning_rate": 2.6879950339783887e-06,
      "loss": 0.3612,
      "step": 13247
    },
    {
      "epoch": 0.7149101505585236,
      "grad_norm": 1.042664294696345,
      "learning_rate": 2.687399474187239e-06,
      "loss": 0.4857,
      "step": 13248
    },
    {
      "epoch": 0.714964114187038,
      "grad_norm": 0.9424720040863117,
      "learning_rate": 2.686803995233069e-06,
      "loss": 0.4724,
      "step": 13249
    },
    {
      "epoch": 0.7150180778155523,
      "grad_norm": 1.062157845113935,
      "learning_rate": 2.6862085971329964e-06,
      "loss": 0.4559,
      "step": 13250
    },
    {
      "epoch": 0.7150720414440667,
      "grad_norm": 1.0760827156999582,
      "learning_rate": 2.6856132799041308e-06,
      "loss": 0.4295,
      "step": 13251
    },
    {
      "epoch": 0.7151260050725811,
      "grad_norm": 1.0173462309012151,
      "learning_rate": 2.6850180435635836e-06,
      "loss": 0.4043,
      "step": 13252
    },
    {
      "epoch": 0.7151799687010955,
      "grad_norm": 0.9570669349832445,
      "learning_rate": 2.6844228881284615e-06,
      "loss": 0.4028,
      "step": 13253
    },
    {
      "epoch": 0.7152339323296099,
      "grad_norm": 0.93167713720306,
      "learning_rate": 2.6838278136158713e-06,
      "loss": 0.5179,
      "step": 13254
    },
    {
      "epoch": 0.7152878959581243,
      "grad_norm": 0.9230214225488681,
      "learning_rate": 2.683232820042916e-06,
      "loss": 0.4395,
      "step": 13255
    },
    {
      "epoch": 0.7153418595866387,
      "grad_norm": 0.9740229453066266,
      "learning_rate": 2.682637907426694e-06,
      "loss": 0.4564,
      "step": 13256
    },
    {
      "epoch": 0.7153958232151529,
      "grad_norm": 0.888360043144517,
      "learning_rate": 2.6820430757843076e-06,
      "loss": 0.375,
      "step": 13257
    },
    {
      "epoch": 0.7154497868436673,
      "grad_norm": 0.9987812677875207,
      "learning_rate": 2.6814483251328488e-06,
      "loss": 0.4693,
      "step": 13258
    },
    {
      "epoch": 0.7155037504721817,
      "grad_norm": 0.9583611942579091,
      "learning_rate": 2.680853655489413e-06,
      "loss": 0.372,
      "step": 13259
    },
    {
      "epoch": 0.7155577141006961,
      "grad_norm": 1.1269935631987493,
      "learning_rate": 2.6802590668710916e-06,
      "loss": 0.5154,
      "step": 13260
    },
    {
      "epoch": 0.7156116777292105,
      "grad_norm": 1.088998456207822,
      "learning_rate": 2.6796645592949744e-06,
      "loss": 0.3569,
      "step": 13261
    },
    {
      "epoch": 0.7156656413577249,
      "grad_norm": 0.9945935809960993,
      "learning_rate": 2.6790701327781467e-06,
      "loss": 0.4329,
      "step": 13262
    },
    {
      "epoch": 0.7157196049862393,
      "grad_norm": 1.1989325584577946,
      "learning_rate": 2.6784757873376945e-06,
      "loss": 0.5132,
      "step": 13263
    },
    {
      "epoch": 0.7157735686147536,
      "grad_norm": 1.0428823232448752,
      "learning_rate": 2.6778815229906994e-06,
      "loss": 0.4258,
      "step": 13264
    },
    {
      "epoch": 0.715827532243268,
      "grad_norm": 0.9354301866831964,
      "learning_rate": 2.677287339754241e-06,
      "loss": 0.3328,
      "step": 13265
    },
    {
      "epoch": 0.7158814958717824,
      "grad_norm": 0.9172893725100263,
      "learning_rate": 2.676693237645395e-06,
      "loss": 0.402,
      "step": 13266
    },
    {
      "epoch": 0.7159354595002968,
      "grad_norm": 0.983889994770096,
      "learning_rate": 2.6760992166812396e-06,
      "loss": 0.4646,
      "step": 13267
    },
    {
      "epoch": 0.7159894231288112,
      "grad_norm": 1.1708155379853766,
      "learning_rate": 2.6755052768788447e-06,
      "loss": 0.5944,
      "step": 13268
    },
    {
      "epoch": 0.7160433867573256,
      "grad_norm": 0.8861047897532095,
      "learning_rate": 2.6749114182552827e-06,
      "loss": 0.4037,
      "step": 13269
    },
    {
      "epoch": 0.71609735038584,
      "grad_norm": 1.1690298429364787,
      "learning_rate": 2.6743176408276215e-06,
      "loss": 0.5783,
      "step": 13270
    },
    {
      "epoch": 0.7161513140143543,
      "grad_norm": 1.2228243930557834,
      "learning_rate": 2.673723944612927e-06,
      "loss": 0.5624,
      "step": 13271
    },
    {
      "epoch": 0.7162052776428687,
      "grad_norm": 0.8621574509821913,
      "learning_rate": 2.6731303296282614e-06,
      "loss": 0.355,
      "step": 13272
    },
    {
      "epoch": 0.7162592412713831,
      "grad_norm": 0.9567079843700925,
      "learning_rate": 2.672536795890687e-06,
      "loss": 0.5317,
      "step": 13273
    },
    {
      "epoch": 0.7163132048998975,
      "grad_norm": 0.9649198063741043,
      "learning_rate": 2.6719433434172603e-06,
      "loss": 0.4161,
      "step": 13274
    },
    {
      "epoch": 0.7163671685284119,
      "grad_norm": 1.0886696745002253,
      "learning_rate": 2.6713499722250403e-06,
      "loss": 0.3889,
      "step": 13275
    },
    {
      "epoch": 0.7164211321569263,
      "grad_norm": 1.113282136502745,
      "learning_rate": 2.6707566823310803e-06,
      "loss": 0.4996,
      "step": 13276
    },
    {
      "epoch": 0.7164750957854407,
      "grad_norm": 1.1070749729080978,
      "learning_rate": 2.6701634737524323e-06,
      "loss": 0.3575,
      "step": 13277
    },
    {
      "epoch": 0.716529059413955,
      "grad_norm": 1.3873544784332612,
      "learning_rate": 2.6695703465061463e-06,
      "loss": 0.5762,
      "step": 13278
    },
    {
      "epoch": 0.7165830230424693,
      "grad_norm": 1.1576047664751277,
      "learning_rate": 2.668977300609268e-06,
      "loss": 0.3888,
      "step": 13279
    },
    {
      "epoch": 0.7166369866709837,
      "grad_norm": 0.9660189638618495,
      "learning_rate": 2.6683843360788415e-06,
      "loss": 0.3229,
      "step": 13280
    },
    {
      "epoch": 0.7166909502994981,
      "grad_norm": 0.9700735634443828,
      "learning_rate": 2.6677914529319103e-06,
      "loss": 0.4643,
      "step": 13281
    },
    {
      "epoch": 0.7167449139280125,
      "grad_norm": 1.115046992482619,
      "learning_rate": 2.6671986511855154e-06,
      "loss": 0.4689,
      "step": 13282
    },
    {
      "epoch": 0.7167988775565269,
      "grad_norm": 1.0349221073257826,
      "learning_rate": 2.6666059308566926e-06,
      "loss": 0.4359,
      "step": 13283
    },
    {
      "epoch": 0.7168528411850413,
      "grad_norm": 0.8536206114152576,
      "learning_rate": 2.6660132919624793e-06,
      "loss": 0.4098,
      "step": 13284
    },
    {
      "epoch": 0.7169068048135556,
      "grad_norm": 1.0386831621262411,
      "learning_rate": 2.6654207345199084e-06,
      "loss": 0.4638,
      "step": 13285
    },
    {
      "epoch": 0.71696076844207,
      "grad_norm": 1.027545738874129,
      "learning_rate": 2.6648282585460086e-06,
      "loss": 0.4656,
      "step": 13286
    },
    {
      "epoch": 0.7170147320705844,
      "grad_norm": 0.9790351929886741,
      "learning_rate": 2.6642358640578086e-06,
      "loss": 0.3951,
      "step": 13287
    },
    {
      "epoch": 0.7170686956990988,
      "grad_norm": 1.4776573244707627,
      "learning_rate": 2.6636435510723357e-06,
      "loss": 0.5339,
      "step": 13288
    },
    {
      "epoch": 0.7171226593276132,
      "grad_norm": 1.262035834028656,
      "learning_rate": 2.6630513196066133e-06,
      "loss": 0.6814,
      "step": 13289
    },
    {
      "epoch": 0.7171766229561276,
      "grad_norm": 1.1380460741563556,
      "learning_rate": 2.6624591696776623e-06,
      "loss": 0.5252,
      "step": 13290
    },
    {
      "epoch": 0.717230586584642,
      "grad_norm": 0.9421209954909444,
      "learning_rate": 2.661867101302502e-06,
      "loss": 0.3523,
      "step": 13291
    },
    {
      "epoch": 0.7172845502131563,
      "grad_norm": 0.7931646988618413,
      "learning_rate": 2.6612751144981502e-06,
      "loss": 0.3496,
      "step": 13292
    },
    {
      "epoch": 0.7173385138416707,
      "grad_norm": 1.0116450424103758,
      "learning_rate": 2.6606832092816183e-06,
      "loss": 0.5476,
      "step": 13293
    },
    {
      "epoch": 0.7173924774701851,
      "grad_norm": 1.1694923329584692,
      "learning_rate": 2.6600913856699206e-06,
      "loss": 0.5143,
      "step": 13294
    },
    {
      "epoch": 0.7174464410986995,
      "grad_norm": 1.0439990577497027,
      "learning_rate": 2.659499643680065e-06,
      "loss": 0.503,
      "step": 13295
    },
    {
      "epoch": 0.7175004047272139,
      "grad_norm": 1.1491498238907525,
      "learning_rate": 2.6589079833290608e-06,
      "loss": 0.5123,
      "step": 13296
    },
    {
      "epoch": 0.7175543683557283,
      "grad_norm": 0.8412888951603144,
      "learning_rate": 2.6583164046339112e-06,
      "loss": 0.3628,
      "step": 13297
    },
    {
      "epoch": 0.7176083319842427,
      "grad_norm": 1.046209302027235,
      "learning_rate": 2.657724907611619e-06,
      "loss": 0.4588,
      "step": 13298
    },
    {
      "epoch": 0.717662295612757,
      "grad_norm": 1.0942489871402077,
      "learning_rate": 2.6571334922791858e-06,
      "loss": 0.4354,
      "step": 13299
    },
    {
      "epoch": 0.7177162592412714,
      "grad_norm": 1.1244782105080735,
      "learning_rate": 2.6565421586536085e-06,
      "loss": 0.4198,
      "step": 13300
    },
    {
      "epoch": 0.7177702228697858,
      "grad_norm": 0.6990059782411363,
      "learning_rate": 2.655950906751882e-06,
      "loss": 0.2644,
      "step": 13301
    },
    {
      "epoch": 0.7178241864983002,
      "grad_norm": 1.2232957091094334,
      "learning_rate": 2.6553597365910004e-06,
      "loss": 0.4704,
      "step": 13302
    },
    {
      "epoch": 0.7178781501268146,
      "grad_norm": 1.0587331184659299,
      "learning_rate": 2.6547686481879542e-06,
      "loss": 0.4956,
      "step": 13303
    },
    {
      "epoch": 0.717932113755329,
      "grad_norm": 0.9264798800524818,
      "learning_rate": 2.654177641559732e-06,
      "loss": 0.4328,
      "step": 13304
    },
    {
      "epoch": 0.7179860773838432,
      "grad_norm": 0.9983079620042615,
      "learning_rate": 2.653586716723322e-06,
      "loss": 0.5073,
      "step": 13305
    },
    {
      "epoch": 0.7180400410123576,
      "grad_norm": 0.9712213318601072,
      "learning_rate": 2.652995873695704e-06,
      "loss": 0.5775,
      "step": 13306
    },
    {
      "epoch": 0.718094004640872,
      "grad_norm": 0.9722447385673195,
      "learning_rate": 2.652405112493861e-06,
      "loss": 0.4388,
      "step": 13307
    },
    {
      "epoch": 0.7181479682693864,
      "grad_norm": 1.046009711966584,
      "learning_rate": 2.651814433134773e-06,
      "loss": 0.5073,
      "step": 13308
    },
    {
      "epoch": 0.7182019318979008,
      "grad_norm": 0.6788485216298138,
      "learning_rate": 2.6512238356354148e-06,
      "loss": 0.2764,
      "step": 13309
    },
    {
      "epoch": 0.7182558955264152,
      "grad_norm": 1.1038209080376011,
      "learning_rate": 2.6506333200127633e-06,
      "loss": 0.579,
      "step": 13310
    },
    {
      "epoch": 0.7183098591549296,
      "grad_norm": 1.0355262471832192,
      "learning_rate": 2.6500428862837878e-06,
      "loss": 0.4646,
      "step": 13311
    },
    {
      "epoch": 0.7183638227834439,
      "grad_norm": 1.2367885021703338,
      "learning_rate": 2.649452534465462e-06,
      "loss": 0.4145,
      "step": 13312
    },
    {
      "epoch": 0.7184177864119583,
      "grad_norm": 1.0099756250747256,
      "learning_rate": 2.648862264574749e-06,
      "loss": 0.4183,
      "step": 13313
    },
    {
      "epoch": 0.7184717500404727,
      "grad_norm": 0.9347681737726682,
      "learning_rate": 2.648272076628616e-06,
      "loss": 0.5466,
      "step": 13314
    },
    {
      "epoch": 0.7185257136689871,
      "grad_norm": 1.0799355554489087,
      "learning_rate": 2.6476819706440242e-06,
      "loss": 0.5158,
      "step": 13315
    },
    {
      "epoch": 0.7185796772975015,
      "grad_norm": 1.28006892222168,
      "learning_rate": 2.6470919466379346e-06,
      "loss": 0.6678,
      "step": 13316
    },
    {
      "epoch": 0.7186336409260159,
      "grad_norm": 0.9103106999199244,
      "learning_rate": 2.646502004627305e-06,
      "loss": 0.3715,
      "step": 13317
    },
    {
      "epoch": 0.7186876045545303,
      "grad_norm": 1.0809759931941167,
      "learning_rate": 2.6459121446290913e-06,
      "loss": 0.4253,
      "step": 13318
    },
    {
      "epoch": 0.7187415681830446,
      "grad_norm": 0.850990093440804,
      "learning_rate": 2.6453223666602486e-06,
      "loss": 0.3361,
      "step": 13319
    },
    {
      "epoch": 0.718795531811559,
      "grad_norm": 0.9022283777896292,
      "learning_rate": 2.644732670737723e-06,
      "loss": 0.4412,
      "step": 13320
    },
    {
      "epoch": 0.7188494954400734,
      "grad_norm": 1.0324457958351054,
      "learning_rate": 2.644143056878466e-06,
      "loss": 0.5075,
      "step": 13321
    },
    {
      "epoch": 0.7189034590685878,
      "grad_norm": 1.0622441168485444,
      "learning_rate": 2.6435535250994225e-06,
      "loss": 0.4004,
      "step": 13322
    },
    {
      "epoch": 0.7189574226971022,
      "grad_norm": 1.2345112902168511,
      "learning_rate": 2.642964075417538e-06,
      "loss": 0.6082,
      "step": 13323
    },
    {
      "epoch": 0.7190113863256166,
      "grad_norm": 0.9033369072544559,
      "learning_rate": 2.642374707849752e-06,
      "loss": 0.3467,
      "step": 13324
    },
    {
      "epoch": 0.719065349954131,
      "grad_norm": 1.110861907016974,
      "learning_rate": 2.6417854224130045e-06,
      "loss": 0.5541,
      "step": 13325
    },
    {
      "epoch": 0.7191193135826452,
      "grad_norm": 1.054662168254958,
      "learning_rate": 2.6411962191242336e-06,
      "loss": 0.4239,
      "step": 13326
    },
    {
      "epoch": 0.7191732772111596,
      "grad_norm": 0.9245187832613796,
      "learning_rate": 2.6406070980003705e-06,
      "loss": 0.3075,
      "step": 13327
    },
    {
      "epoch": 0.719227240839674,
      "grad_norm": 1.3757893422438396,
      "learning_rate": 2.640018059058348e-06,
      "loss": 0.6885,
      "step": 13328
    },
    {
      "epoch": 0.7192812044681884,
      "grad_norm": 1.059567460082278,
      "learning_rate": 2.6394291023150975e-06,
      "loss": 0.5588,
      "step": 13329
    },
    {
      "epoch": 0.7193351680967028,
      "grad_norm": 0.9095210469799303,
      "learning_rate": 2.638840227787544e-06,
      "loss": 0.332,
      "step": 13330
    },
    {
      "epoch": 0.7193891317252172,
      "grad_norm": 1.0251321780202733,
      "learning_rate": 2.6382514354926135e-06,
      "loss": 0.4278,
      "step": 13331
    },
    {
      "epoch": 0.7194430953537316,
      "grad_norm": 0.9043921216248175,
      "learning_rate": 2.6376627254472287e-06,
      "loss": 0.3895,
      "step": 13332
    },
    {
      "epoch": 0.7194970589822459,
      "grad_norm": 0.8462545838853975,
      "learning_rate": 2.6370740976683108e-06,
      "loss": 0.5038,
      "step": 13333
    },
    {
      "epoch": 0.7195510226107603,
      "grad_norm": 0.8728069422776251,
      "learning_rate": 2.6364855521727737e-06,
      "loss": 0.3258,
      "step": 13334
    },
    {
      "epoch": 0.7196049862392747,
      "grad_norm": 1.143134116069217,
      "learning_rate": 2.635897088977536e-06,
      "loss": 0.6689,
      "step": 13335
    },
    {
      "epoch": 0.7196589498677891,
      "grad_norm": 1.1262566446839823,
      "learning_rate": 2.63530870809951e-06,
      "loss": 0.494,
      "step": 13336
    },
    {
      "epoch": 0.7197129134963035,
      "grad_norm": 1.044147974697176,
      "learning_rate": 2.634720409555606e-06,
      "loss": 0.5335,
      "step": 13337
    },
    {
      "epoch": 0.7197668771248179,
      "grad_norm": 1.222035072313866,
      "learning_rate": 2.6341321933627327e-06,
      "loss": 0.6227,
      "step": 13338
    },
    {
      "epoch": 0.7198208407533323,
      "grad_norm": 1.0395755569582927,
      "learning_rate": 2.6335440595377957e-06,
      "loss": 0.5965,
      "step": 13339
    },
    {
      "epoch": 0.7198748043818466,
      "grad_norm": 0.8866248245066388,
      "learning_rate": 2.632956008097701e-06,
      "loss": 0.4333,
      "step": 13340
    },
    {
      "epoch": 0.719928768010361,
      "grad_norm": 1.1266294740935232,
      "learning_rate": 2.6323680390593454e-06,
      "loss": 0.4582,
      "step": 13341
    },
    {
      "epoch": 0.7199827316388754,
      "grad_norm": 1.0365631193763172,
      "learning_rate": 2.6317801524396304e-06,
      "loss": 0.4067,
      "step": 13342
    },
    {
      "epoch": 0.7200366952673898,
      "grad_norm": 1.2500611994750814,
      "learning_rate": 2.6311923482554513e-06,
      "loss": 0.5949,
      "step": 13343
    },
    {
      "epoch": 0.7200906588959042,
      "grad_norm": 1.1883191130209865,
      "learning_rate": 2.630604626523703e-06,
      "loss": 0.4652,
      "step": 13344
    },
    {
      "epoch": 0.7201446225244186,
      "grad_norm": 0.7908549345485597,
      "learning_rate": 2.6300169872612768e-06,
      "loss": 0.3869,
      "step": 13345
    },
    {
      "epoch": 0.720198586152933,
      "grad_norm": 0.8211297297374475,
      "learning_rate": 2.629429430485063e-06,
      "loss": 0.3035,
      "step": 13346
    },
    {
      "epoch": 0.7202525497814473,
      "grad_norm": 1.271109007667921,
      "learning_rate": 2.628841956211948e-06,
      "loss": 0.6146,
      "step": 13347
    },
    {
      "epoch": 0.7203065134099617,
      "grad_norm": 1.0856460892834938,
      "learning_rate": 2.628254564458815e-06,
      "loss": 0.4812,
      "step": 13348
    },
    {
      "epoch": 0.720360477038476,
      "grad_norm": 0.9051195223617481,
      "learning_rate": 2.6276672552425485e-06,
      "loss": 0.3836,
      "step": 13349
    },
    {
      "epoch": 0.7204144406669905,
      "grad_norm": 0.9196845608309803,
      "learning_rate": 2.6270800285800268e-06,
      "loss": 0.4353,
      "step": 13350
    },
    {
      "epoch": 0.7204684042955048,
      "grad_norm": 1.0186971290573599,
      "learning_rate": 2.62649288448813e-06,
      "loss": 0.4908,
      "step": 13351
    },
    {
      "epoch": 0.7205223679240192,
      "grad_norm": 1.0945365479637779,
      "learning_rate": 2.62590582298373e-06,
      "loss": 0.4769,
      "step": 13352
    },
    {
      "epoch": 0.7205763315525336,
      "grad_norm": 1.1863620372870605,
      "learning_rate": 2.6253188440837e-06,
      "loss": 0.5071,
      "step": 13353
    },
    {
      "epoch": 0.7206302951810479,
      "grad_norm": 1.0597391744809816,
      "learning_rate": 2.624731947804911e-06,
      "loss": 0.4467,
      "step": 13354
    },
    {
      "epoch": 0.7206842588095623,
      "grad_norm": 0.8131773599336587,
      "learning_rate": 2.6241451341642317e-06,
      "loss": 0.333,
      "step": 13355
    },
    {
      "epoch": 0.7207382224380767,
      "grad_norm": 1.1090944625871009,
      "learning_rate": 2.6235584031785268e-06,
      "loss": 0.5467,
      "step": 13356
    },
    {
      "epoch": 0.7207921860665911,
      "grad_norm": 1.0252940983541994,
      "learning_rate": 2.6229717548646592e-06,
      "loss": 0.4827,
      "step": 13357
    },
    {
      "epoch": 0.7208461496951055,
      "grad_norm": 0.7778224259606199,
      "learning_rate": 2.6223851892394914e-06,
      "loss": 0.3416,
      "step": 13358
    },
    {
      "epoch": 0.7209001133236199,
      "grad_norm": 1.3489244281892114,
      "learning_rate": 2.62179870631988e-06,
      "loss": 0.6663,
      "step": 13359
    },
    {
      "epoch": 0.7209540769521343,
      "grad_norm": 1.0365589194355815,
      "learning_rate": 2.6212123061226825e-06,
      "loss": 0.5552,
      "step": 13360
    },
    {
      "epoch": 0.7210080405806486,
      "grad_norm": 0.8988408443706264,
      "learning_rate": 2.6206259886647533e-06,
      "loss": 0.3817,
      "step": 13361
    },
    {
      "epoch": 0.721062004209163,
      "grad_norm": 1.067960249452255,
      "learning_rate": 2.6200397539629418e-06,
      "loss": 0.4182,
      "step": 13362
    },
    {
      "epoch": 0.7211159678376774,
      "grad_norm": 1.2370291401363136,
      "learning_rate": 2.619453602034097e-06,
      "loss": 0.7044,
      "step": 13363
    },
    {
      "epoch": 0.7211699314661918,
      "grad_norm": 1.025091635631566,
      "learning_rate": 2.6188675328950665e-06,
      "loss": 0.4706,
      "step": 13364
    },
    {
      "epoch": 0.7212238950947062,
      "grad_norm": 0.9636420401400061,
      "learning_rate": 2.6182815465626943e-06,
      "loss": 0.5533,
      "step": 13365
    },
    {
      "epoch": 0.7212778587232206,
      "grad_norm": 0.6893176321208639,
      "learning_rate": 2.6176956430538225e-06,
      "loss": 0.3112,
      "step": 13366
    },
    {
      "epoch": 0.721331822351735,
      "grad_norm": 1.0299483787681833,
      "learning_rate": 2.6171098223852896e-06,
      "loss": 0.4938,
      "step": 13367
    },
    {
      "epoch": 0.7213857859802493,
      "grad_norm": 0.9390610185474342,
      "learning_rate": 2.616524084573936e-06,
      "loss": 0.4909,
      "step": 13368
    },
    {
      "epoch": 0.7214397496087637,
      "grad_norm": 1.1209523498051224,
      "learning_rate": 2.6159384296365908e-06,
      "loss": 0.4912,
      "step": 13369
    },
    {
      "epoch": 0.7214937132372781,
      "grad_norm": 0.952356167557773,
      "learning_rate": 2.61535285759009e-06,
      "loss": 0.3578,
      "step": 13370
    },
    {
      "epoch": 0.7215476768657925,
      "grad_norm": 1.0612729591430898,
      "learning_rate": 2.6147673684512625e-06,
      "loss": 0.4151,
      "step": 13371
    },
    {
      "epoch": 0.7216016404943069,
      "grad_norm": 0.8696471916677225,
      "learning_rate": 2.614181962236937e-06,
      "loss": 0.3292,
      "step": 13372
    },
    {
      "epoch": 0.7216556041228213,
      "grad_norm": 0.9750245167594038,
      "learning_rate": 2.6135966389639368e-06,
      "loss": 0.4625,
      "step": 13373
    },
    {
      "epoch": 0.7217095677513355,
      "grad_norm": 1.4664200522554673,
      "learning_rate": 2.6130113986490876e-06,
      "loss": 0.5737,
      "step": 13374
    },
    {
      "epoch": 0.7217635313798499,
      "grad_norm": 1.0433685219271545,
      "learning_rate": 2.612426241309207e-06,
      "loss": 0.5445,
      "step": 13375
    },
    {
      "epoch": 0.7218174950083643,
      "grad_norm": 1.243240134938577,
      "learning_rate": 2.611841166961112e-06,
      "loss": 0.6417,
      "step": 13376
    },
    {
      "epoch": 0.7218714586368787,
      "grad_norm": 0.9297417461358365,
      "learning_rate": 2.611256175621622e-06,
      "loss": 0.4172,
      "step": 13377
    },
    {
      "epoch": 0.7219254222653931,
      "grad_norm": 0.9800100035186387,
      "learning_rate": 2.610671267307548e-06,
      "loss": 0.3942,
      "step": 13378
    },
    {
      "epoch": 0.7219793858939075,
      "grad_norm": 1.3858848249656541,
      "learning_rate": 2.6100864420357007e-06,
      "loss": 0.5304,
      "step": 13379
    },
    {
      "epoch": 0.7220333495224219,
      "grad_norm": 0.8999764526867546,
      "learning_rate": 2.6095016998228896e-06,
      "loss": 0.4209,
      "step": 13380
    },
    {
      "epoch": 0.7220873131509362,
      "grad_norm": 0.8346293751623973,
      "learning_rate": 2.6089170406859223e-06,
      "loss": 0.3763,
      "step": 13381
    },
    {
      "epoch": 0.7221412767794506,
      "grad_norm": 0.9039919118207718,
      "learning_rate": 2.6083324646415975e-06,
      "loss": 0.3612,
      "step": 13382
    },
    {
      "epoch": 0.722195240407965,
      "grad_norm": 1.1546840769463755,
      "learning_rate": 2.6077479717067208e-06,
      "loss": 0.5268,
      "step": 13383
    },
    {
      "epoch": 0.7222492040364794,
      "grad_norm": 1.077537886080401,
      "learning_rate": 2.6071635618980888e-06,
      "loss": 0.4941,
      "step": 13384
    },
    {
      "epoch": 0.7223031676649938,
      "grad_norm": 0.9066313291098543,
      "learning_rate": 2.606579235232499e-06,
      "loss": 0.476,
      "step": 13385
    },
    {
      "epoch": 0.7223571312935082,
      "grad_norm": 1.0968952459798218,
      "learning_rate": 2.605994991726747e-06,
      "loss": 0.3986,
      "step": 13386
    },
    {
      "epoch": 0.7224110949220226,
      "grad_norm": 1.2058814298839027,
      "learning_rate": 2.605410831397622e-06,
      "loss": 0.5272,
      "step": 13387
    },
    {
      "epoch": 0.7224650585505369,
      "grad_norm": 1.2445512966540968,
      "learning_rate": 2.604826754261916e-06,
      "loss": 0.4773,
      "step": 13388
    },
    {
      "epoch": 0.7225190221790513,
      "grad_norm": 1.2496243249704435,
      "learning_rate": 2.6042427603364128e-06,
      "loss": 0.544,
      "step": 13389
    },
    {
      "epoch": 0.7225729858075657,
      "grad_norm": 1.237062632110237,
      "learning_rate": 2.603658849637899e-06,
      "loss": 0.4594,
      "step": 13390
    },
    {
      "epoch": 0.7226269494360801,
      "grad_norm": 1.1233287926250792,
      "learning_rate": 2.603075022183157e-06,
      "loss": 0.5654,
      "step": 13391
    },
    {
      "epoch": 0.7226809130645945,
      "grad_norm": 1.1428239001013867,
      "learning_rate": 2.6024912779889644e-06,
      "loss": 0.3955,
      "step": 13392
    },
    {
      "epoch": 0.7227348766931089,
      "grad_norm": 1.0847401353029984,
      "learning_rate": 2.6019076170721016e-06,
      "loss": 0.3706,
      "step": 13393
    },
    {
      "epoch": 0.7227888403216233,
      "grad_norm": 1.007130351347291,
      "learning_rate": 2.601324039449342e-06,
      "loss": 0.4198,
      "step": 13394
    },
    {
      "epoch": 0.7228428039501376,
      "grad_norm": 1.0234229307057372,
      "learning_rate": 2.600740545137459e-06,
      "loss": 0.4032,
      "step": 13395
    },
    {
      "epoch": 0.722896767578652,
      "grad_norm": 1.1563523393555486,
      "learning_rate": 2.6001571341532217e-06,
      "loss": 0.4983,
      "step": 13396
    },
    {
      "epoch": 0.7229507312071664,
      "grad_norm": 0.864507303855469,
      "learning_rate": 2.599573806513398e-06,
      "loss": 0.3891,
      "step": 13397
    },
    {
      "epoch": 0.7230046948356808,
      "grad_norm": 0.9588198198261277,
      "learning_rate": 2.5989905622347538e-06,
      "loss": 0.4734,
      "step": 13398
    },
    {
      "epoch": 0.7230586584641951,
      "grad_norm": 0.8226653585812616,
      "learning_rate": 2.598407401334052e-06,
      "loss": 0.2411,
      "step": 13399
    },
    {
      "epoch": 0.7231126220927095,
      "grad_norm": 1.0402009078820305,
      "learning_rate": 2.5978243238280532e-06,
      "loss": 0.5403,
      "step": 13400
    },
    {
      "epoch": 0.7231665857212239,
      "grad_norm": 0.9485042295359996,
      "learning_rate": 2.5972413297335163e-06,
      "loss": 0.4309,
      "step": 13401
    },
    {
      "epoch": 0.7232205493497382,
      "grad_norm": 0.9447265521248864,
      "learning_rate": 2.596658419067196e-06,
      "loss": 0.3598,
      "step": 13402
    },
    {
      "epoch": 0.7232745129782526,
      "grad_norm": 1.181815959235528,
      "learning_rate": 2.5960755918458468e-06,
      "loss": 0.438,
      "step": 13403
    },
    {
      "epoch": 0.723328476606767,
      "grad_norm": 1.1464676224211592,
      "learning_rate": 2.5954928480862198e-06,
      "loss": 0.5186,
      "step": 13404
    },
    {
      "epoch": 0.7233824402352814,
      "grad_norm": 0.8878367893950712,
      "learning_rate": 2.594910187805062e-06,
      "loss": 0.3149,
      "step": 13405
    },
    {
      "epoch": 0.7234364038637958,
      "grad_norm": 1.0687067475905487,
      "learning_rate": 2.594327611019121e-06,
      "loss": 0.5901,
      "step": 13406
    },
    {
      "epoch": 0.7234903674923102,
      "grad_norm": 1.0076934363885313,
      "learning_rate": 2.593745117745141e-06,
      "loss": 0.4113,
      "step": 13407
    },
    {
      "epoch": 0.7235443311208246,
      "grad_norm": 0.8418182694276831,
      "learning_rate": 2.593162707999862e-06,
      "loss": 0.2749,
      "step": 13408
    },
    {
      "epoch": 0.7235982947493389,
      "grad_norm": 0.9521206149220434,
      "learning_rate": 2.5925803818000267e-06,
      "loss": 0.349,
      "step": 13409
    },
    {
      "epoch": 0.7236522583778533,
      "grad_norm": 1.0032223209487299,
      "learning_rate": 2.5919981391623665e-06,
      "loss": 0.4346,
      "step": 13410
    },
    {
      "epoch": 0.7237062220063677,
      "grad_norm": 1.2622900497425407,
      "learning_rate": 2.5914159801036186e-06,
      "loss": 0.5471,
      "step": 13411
    },
    {
      "epoch": 0.7237601856348821,
      "grad_norm": 1.0024412227741937,
      "learning_rate": 2.590833904640514e-06,
      "loss": 0.4045,
      "step": 13412
    },
    {
      "epoch": 0.7238141492633965,
      "grad_norm": 1.0685179347140787,
      "learning_rate": 2.590251912789783e-06,
      "loss": 0.6033,
      "step": 13413
    },
    {
      "epoch": 0.7238681128919109,
      "grad_norm": 0.9030563653256031,
      "learning_rate": 2.589670004568151e-06,
      "loss": 0.3373,
      "step": 13414
    },
    {
      "epoch": 0.7239220765204253,
      "grad_norm": 1.2139581660842473,
      "learning_rate": 2.589088179992345e-06,
      "loss": 0.5355,
      "step": 13415
    },
    {
      "epoch": 0.7239760401489396,
      "grad_norm": 1.108809755600344,
      "learning_rate": 2.588506439079087e-06,
      "loss": 0.4288,
      "step": 13416
    },
    {
      "epoch": 0.724030003777454,
      "grad_norm": 1.086472818029685,
      "learning_rate": 2.5879247818450935e-06,
      "loss": 0.5728,
      "step": 13417
    },
    {
      "epoch": 0.7240839674059684,
      "grad_norm": 1.0433047513856017,
      "learning_rate": 2.587343208307085e-06,
      "loss": 0.3749,
      "step": 13418
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 1.103855316603011,
      "learning_rate": 2.586761718481776e-06,
      "loss": 0.665,
      "step": 13419
    },
    {
      "epoch": 0.7241918946629972,
      "grad_norm": 1.150991632653524,
      "learning_rate": 2.586180312385878e-06,
      "loss": 0.5238,
      "step": 13420
    },
    {
      "epoch": 0.7242458582915116,
      "grad_norm": 0.8619040356298572,
      "learning_rate": 2.585598990036103e-06,
      "loss": 0.4174,
      "step": 13421
    },
    {
      "epoch": 0.724299821920026,
      "grad_norm": 1.0253954306758601,
      "learning_rate": 2.585017751449157e-06,
      "loss": 0.4326,
      "step": 13422
    },
    {
      "epoch": 0.7243537855485402,
      "grad_norm": 1.0438291757955294,
      "learning_rate": 2.5844365966417485e-06,
      "loss": 0.4519,
      "step": 13423
    },
    {
      "epoch": 0.7244077491770546,
      "grad_norm": 0.9144167466397051,
      "learning_rate": 2.5838555256305757e-06,
      "loss": 0.3725,
      "step": 13424
    },
    {
      "epoch": 0.724461712805569,
      "grad_norm": 1.0208303272009667,
      "learning_rate": 2.583274538432342e-06,
      "loss": 0.6227,
      "step": 13425
    },
    {
      "epoch": 0.7245156764340834,
      "grad_norm": 1.0057592403421007,
      "learning_rate": 2.5826936350637453e-06,
      "loss": 0.3794,
      "step": 13426
    },
    {
      "epoch": 0.7245696400625978,
      "grad_norm": 0.9345804932280786,
      "learning_rate": 2.582112815541481e-06,
      "loss": 0.4171,
      "step": 13427
    },
    {
      "epoch": 0.7246236036911122,
      "grad_norm": 1.2513109999716336,
      "learning_rate": 2.5815320798822435e-06,
      "loss": 0.5249,
      "step": 13428
    },
    {
      "epoch": 0.7246775673196266,
      "grad_norm": 0.8875130433898615,
      "learning_rate": 2.5809514281027226e-06,
      "loss": 0.5018,
      "step": 13429
    },
    {
      "epoch": 0.7247315309481409,
      "grad_norm": 0.7784890537793363,
      "learning_rate": 2.5803708602196078e-06,
      "loss": 0.4323,
      "step": 13430
    },
    {
      "epoch": 0.7247854945766553,
      "grad_norm": 1.0115762304953475,
      "learning_rate": 2.579790376249584e-06,
      "loss": 0.4737,
      "step": 13431
    },
    {
      "epoch": 0.7248394582051697,
      "grad_norm": 1.1113212079010701,
      "learning_rate": 2.579209976209335e-06,
      "loss": 0.5808,
      "step": 13432
    },
    {
      "epoch": 0.7248934218336841,
      "grad_norm": 0.9544041116536818,
      "learning_rate": 2.5786296601155427e-06,
      "loss": 0.3852,
      "step": 13433
    },
    {
      "epoch": 0.7249473854621985,
      "grad_norm": 1.0139030283395896,
      "learning_rate": 2.5780494279848854e-06,
      "loss": 0.4283,
      "step": 13434
    },
    {
      "epoch": 0.7250013490907129,
      "grad_norm": 0.9006901988034783,
      "learning_rate": 2.5774692798340407e-06,
      "loss": 0.447,
      "step": 13435
    },
    {
      "epoch": 0.7250553127192273,
      "grad_norm": 1.0434744400956788,
      "learning_rate": 2.5768892156796815e-06,
      "loss": 0.4215,
      "step": 13436
    },
    {
      "epoch": 0.7251092763477416,
      "grad_norm": 1.071165558705846,
      "learning_rate": 2.5763092355384812e-06,
      "loss": 0.5398,
      "step": 13437
    },
    {
      "epoch": 0.725163239976256,
      "grad_norm": 1.0606191089530117,
      "learning_rate": 2.575729339427106e-06,
      "loss": 0.4677,
      "step": 13438
    },
    {
      "epoch": 0.7252172036047704,
      "grad_norm": 1.1135351453426188,
      "learning_rate": 2.575149527362225e-06,
      "loss": 0.496,
      "step": 13439
    },
    {
      "epoch": 0.7252711672332848,
      "grad_norm": 1.065480867671898,
      "learning_rate": 2.574569799360501e-06,
      "loss": 0.4069,
      "step": 13440
    },
    {
      "epoch": 0.7253251308617992,
      "grad_norm": 1.1215321995261585,
      "learning_rate": 2.5739901554385977e-06,
      "loss": 0.5248,
      "step": 13441
    },
    {
      "epoch": 0.7253790944903136,
      "grad_norm": 1.142664167430887,
      "learning_rate": 2.5734105956131737e-06,
      "loss": 0.503,
      "step": 13442
    },
    {
      "epoch": 0.7254330581188279,
      "grad_norm": 0.8838402510335802,
      "learning_rate": 2.572831119900888e-06,
      "loss": 0.3703,
      "step": 13443
    },
    {
      "epoch": 0.7254870217473423,
      "grad_norm": 0.7440157655215867,
      "learning_rate": 2.5722517283183914e-06,
      "loss": 0.2576,
      "step": 13444
    },
    {
      "epoch": 0.7255409853758567,
      "grad_norm": 1.106035769709015,
      "learning_rate": 2.5716724208823383e-06,
      "loss": 0.4654,
      "step": 13445
    },
    {
      "epoch": 0.725594949004371,
      "grad_norm": 1.1140617198996015,
      "learning_rate": 2.5710931976093797e-06,
      "loss": 0.4646,
      "step": 13446
    },
    {
      "epoch": 0.7256489126328854,
      "grad_norm": 0.9160680846340389,
      "learning_rate": 2.5705140585161613e-06,
      "loss": 0.3731,
      "step": 13447
    },
    {
      "epoch": 0.7257028762613998,
      "grad_norm": 1.0429912849896734,
      "learning_rate": 2.5699350036193287e-06,
      "loss": 0.5195,
      "step": 13448
    },
    {
      "epoch": 0.7257568398899142,
      "grad_norm": 1.0244798290902768,
      "learning_rate": 2.569356032935525e-06,
      "loss": 0.3341,
      "step": 13449
    },
    {
      "epoch": 0.7258108035184285,
      "grad_norm": 0.9959707542058127,
      "learning_rate": 2.5687771464813906e-06,
      "loss": 0.4311,
      "step": 13450
    },
    {
      "epoch": 0.7258647671469429,
      "grad_norm": 1.1487090686999863,
      "learning_rate": 2.5681983442735626e-06,
      "loss": 0.4549,
      "step": 13451
    },
    {
      "epoch": 0.7259187307754573,
      "grad_norm": 0.9134372755459692,
      "learning_rate": 2.5676196263286755e-06,
      "loss": 0.3628,
      "step": 13452
    },
    {
      "epoch": 0.7259726944039717,
      "grad_norm": 1.0126358520911054,
      "learning_rate": 2.5670409926633642e-06,
      "loss": 0.439,
      "step": 13453
    },
    {
      "epoch": 0.7260266580324861,
      "grad_norm": 1.249538990912904,
      "learning_rate": 2.5664624432942585e-06,
      "loss": 0.5967,
      "step": 13454
    },
    {
      "epoch": 0.7260806216610005,
      "grad_norm": 1.0292338761775375,
      "learning_rate": 2.565883978237986e-06,
      "loss": 0.5033,
      "step": 13455
    },
    {
      "epoch": 0.7261345852895149,
      "grad_norm": 1.1682916736302917,
      "learning_rate": 2.565305597511173e-06,
      "loss": 0.5272,
      "step": 13456
    },
    {
      "epoch": 0.7261885489180292,
      "grad_norm": 0.770031822145691,
      "learning_rate": 2.5647273011304435e-06,
      "loss": 0.2736,
      "step": 13457
    },
    {
      "epoch": 0.7262425125465436,
      "grad_norm": 0.8966492766576014,
      "learning_rate": 2.5641490891124168e-06,
      "loss": 0.4594,
      "step": 13458
    },
    {
      "epoch": 0.726296476175058,
      "grad_norm": 1.0921978061560218,
      "learning_rate": 2.5635709614737103e-06,
      "loss": 0.5292,
      "step": 13459
    },
    {
      "epoch": 0.7263504398035724,
      "grad_norm": 1.0503412928697025,
      "learning_rate": 2.562992918230942e-06,
      "loss": 0.4296,
      "step": 13460
    },
    {
      "epoch": 0.7264044034320868,
      "grad_norm": 0.8969878652004784,
      "learning_rate": 2.562414959400725e-06,
      "loss": 0.3714,
      "step": 13461
    },
    {
      "epoch": 0.7264583670606012,
      "grad_norm": 1.1431141316852604,
      "learning_rate": 2.56183708499967e-06,
      "loss": 0.508,
      "step": 13462
    },
    {
      "epoch": 0.7265123306891156,
      "grad_norm": 1.106903686496617,
      "learning_rate": 2.561259295044386e-06,
      "loss": 0.6066,
      "step": 13463
    },
    {
      "epoch": 0.7265662943176299,
      "grad_norm": 0.8486312353684848,
      "learning_rate": 2.56068158955148e-06,
      "loss": 0.3854,
      "step": 13464
    },
    {
      "epoch": 0.7266202579461443,
      "grad_norm": 1.0310339183529158,
      "learning_rate": 2.5601039685375545e-06,
      "loss": 0.5855,
      "step": 13465
    },
    {
      "epoch": 0.7266742215746587,
      "grad_norm": 0.9602753127440643,
      "learning_rate": 2.5595264320192115e-06,
      "loss": 0.4226,
      "step": 13466
    },
    {
      "epoch": 0.7267281852031731,
      "grad_norm": 1.0829597269121054,
      "learning_rate": 2.5589489800130502e-06,
      "loss": 0.4118,
      "step": 13467
    },
    {
      "epoch": 0.7267821488316875,
      "grad_norm": 0.9558687924618091,
      "learning_rate": 2.558371612535666e-06,
      "loss": 0.442,
      "step": 13468
    },
    {
      "epoch": 0.7268361124602019,
      "grad_norm": 1.1711716406608983,
      "learning_rate": 2.557794329603655e-06,
      "loss": 0.5272,
      "step": 13469
    },
    {
      "epoch": 0.7268900760887163,
      "grad_norm": 1.000493603825545,
      "learning_rate": 2.557217131233607e-06,
      "loss": 0.514,
      "step": 13470
    },
    {
      "epoch": 0.7269440397172305,
      "grad_norm": 1.0261834783316424,
      "learning_rate": 2.556640017442114e-06,
      "loss": 0.4245,
      "step": 13471
    },
    {
      "epoch": 0.7269980033457449,
      "grad_norm": 0.9762760330617286,
      "learning_rate": 2.55606298824576e-06,
      "loss": 0.4514,
      "step": 13472
    },
    {
      "epoch": 0.7270519669742593,
      "grad_norm": 1.0071282819837983,
      "learning_rate": 2.5554860436611293e-06,
      "loss": 0.4028,
      "step": 13473
    },
    {
      "epoch": 0.7271059306027737,
      "grad_norm": 1.0434496217640248,
      "learning_rate": 2.554909183704806e-06,
      "loss": 0.4258,
      "step": 13474
    },
    {
      "epoch": 0.7271598942312881,
      "grad_norm": 1.1341355863625404,
      "learning_rate": 2.5543324083933673e-06,
      "loss": 0.486,
      "step": 13475
    },
    {
      "epoch": 0.7272138578598025,
      "grad_norm": 0.9697054685378246,
      "learning_rate": 2.5537557177433923e-06,
      "loss": 0.414,
      "step": 13476
    },
    {
      "epoch": 0.7272678214883169,
      "grad_norm": 0.9446106254973385,
      "learning_rate": 2.553179111771455e-06,
      "loss": 0.4844,
      "step": 13477
    },
    {
      "epoch": 0.7273217851168312,
      "grad_norm": 1.0125567262800435,
      "learning_rate": 2.5526025904941287e-06,
      "loss": 0.4557,
      "step": 13478
    },
    {
      "epoch": 0.7273757487453456,
      "grad_norm": 1.1275198041556171,
      "learning_rate": 2.552026153927981e-06,
      "loss": 0.5271,
      "step": 13479
    },
    {
      "epoch": 0.72742971237386,
      "grad_norm": 1.3798050295767914,
      "learning_rate": 2.551449802089581e-06,
      "loss": 0.6914,
      "step": 13480
    },
    {
      "epoch": 0.7274836760023744,
      "grad_norm": 0.8693669713270226,
      "learning_rate": 2.5508735349954915e-06,
      "loss": 0.3983,
      "step": 13481
    },
    {
      "epoch": 0.7275376396308888,
      "grad_norm": 0.8918954738618847,
      "learning_rate": 2.5502973526622765e-06,
      "loss": 0.4955,
      "step": 13482
    },
    {
      "epoch": 0.7275916032594032,
      "grad_norm": 1.0883456788876844,
      "learning_rate": 2.5497212551064975e-06,
      "loss": 0.5499,
      "step": 13483
    },
    {
      "epoch": 0.7276455668879176,
      "grad_norm": 1.1435157567624432,
      "learning_rate": 2.5491452423447092e-06,
      "loss": 0.5367,
      "step": 13484
    },
    {
      "epoch": 0.7276995305164319,
      "grad_norm": 1.021574063750877,
      "learning_rate": 2.5485693143934704e-06,
      "loss": 0.4892,
      "step": 13485
    },
    {
      "epoch": 0.7277534941449463,
      "grad_norm": 1.1201109276828871,
      "learning_rate": 2.5479934712693304e-06,
      "loss": 0.3436,
      "step": 13486
    },
    {
      "epoch": 0.7278074577734607,
      "grad_norm": 0.859690391785482,
      "learning_rate": 2.5474177129888405e-06,
      "loss": 0.4377,
      "step": 13487
    },
    {
      "epoch": 0.7278614214019751,
      "grad_norm": 1.014996765384532,
      "learning_rate": 2.546842039568549e-06,
      "loss": 0.3723,
      "step": 13488
    },
    {
      "epoch": 0.7279153850304895,
      "grad_norm": 1.0336977387838089,
      "learning_rate": 2.546266451025001e-06,
      "loss": 0.3774,
      "step": 13489
    },
    {
      "epoch": 0.7279693486590039,
      "grad_norm": 1.0266008437310108,
      "learning_rate": 2.5456909473747404e-06,
      "loss": 0.4434,
      "step": 13490
    },
    {
      "epoch": 0.7280233122875183,
      "grad_norm": 1.2584558355916085,
      "learning_rate": 2.5451155286343065e-06,
      "loss": 0.6913,
      "step": 13491
    },
    {
      "epoch": 0.7280772759160326,
      "grad_norm": 0.954501971458651,
      "learning_rate": 2.54454019482024e-06,
      "loss": 0.379,
      "step": 13492
    },
    {
      "epoch": 0.728131239544547,
      "grad_norm": 1.0240603495366793,
      "learning_rate": 2.5439649459490724e-06,
      "loss": 0.3903,
      "step": 13493
    },
    {
      "epoch": 0.7281852031730613,
      "grad_norm": 1.1967391184068452,
      "learning_rate": 2.54338978203734e-06,
      "loss": 0.5204,
      "step": 13494
    },
    {
      "epoch": 0.7282391668015757,
      "grad_norm": 0.7955382193684571,
      "learning_rate": 2.5428147031015716e-06,
      "loss": 0.3488,
      "step": 13495
    },
    {
      "epoch": 0.7282931304300901,
      "grad_norm": 0.8121678564221708,
      "learning_rate": 2.5422397091582973e-06,
      "loss": 0.3459,
      "step": 13496
    },
    {
      "epoch": 0.7283470940586045,
      "grad_norm": 1.0608005243740393,
      "learning_rate": 2.541664800224042e-06,
      "loss": 0.3905,
      "step": 13497
    },
    {
      "epoch": 0.7284010576871189,
      "grad_norm": 0.8965358985749823,
      "learning_rate": 2.54108997631533e-06,
      "loss": 0.3919,
      "step": 13498
    },
    {
      "epoch": 0.7284550213156332,
      "grad_norm": 1.0057197680911782,
      "learning_rate": 2.540515237448681e-06,
      "loss": 0.4765,
      "step": 13499
    },
    {
      "epoch": 0.7285089849441476,
      "grad_norm": 0.7639110577649296,
      "learning_rate": 2.5399405836406165e-06,
      "loss": 0.3225,
      "step": 13500
    },
    {
      "epoch": 0.7285089849441476,
      "eval_loss": 0.5345942378044128,
      "eval_runtime": 166.1813,
      "eval_samples_per_second": 20.694,
      "eval_steps_per_second": 0.867,
      "step": 13500
    },
    {
      "epoch": 0.728562948572662,
      "grad_norm": 0.9832073302542869,
      "learning_rate": 2.5393660149076495e-06,
      "loss": 0.4193,
      "step": 13501
    },
    {
      "epoch": 0.7286169122011764,
      "grad_norm": 1.001608204073589,
      "learning_rate": 2.538791531266294e-06,
      "loss": 0.378,
      "step": 13502
    },
    {
      "epoch": 0.7286708758296908,
      "grad_norm": 1.1774002168067876,
      "learning_rate": 2.538217132733063e-06,
      "loss": 0.465,
      "step": 13503
    },
    {
      "epoch": 0.7287248394582052,
      "grad_norm": 1.1741578246835056,
      "learning_rate": 2.537642819324463e-06,
      "loss": 0.4887,
      "step": 13504
    },
    {
      "epoch": 0.7287788030867196,
      "grad_norm": 0.9866896679432774,
      "learning_rate": 2.5370685910570027e-06,
      "loss": 0.479,
      "step": 13505
    },
    {
      "epoch": 0.7288327667152339,
      "grad_norm": 1.143813844002504,
      "learning_rate": 2.5364944479471836e-06,
      "loss": 0.5776,
      "step": 13506
    },
    {
      "epoch": 0.7288867303437483,
      "grad_norm": 1.1072261202458875,
      "learning_rate": 2.5359203900115093e-06,
      "loss": 0.5148,
      "step": 13507
    },
    {
      "epoch": 0.7289406939722627,
      "grad_norm": 1.0148745440345333,
      "learning_rate": 2.5353464172664777e-06,
      "loss": 0.4708,
      "step": 13508
    },
    {
      "epoch": 0.7289946576007771,
      "grad_norm": 1.1720563683039176,
      "learning_rate": 2.5347725297285864e-06,
      "loss": 0.4011,
      "step": 13509
    },
    {
      "epoch": 0.7290486212292915,
      "grad_norm": 0.9104143572540124,
      "learning_rate": 2.5341987274143274e-06,
      "loss": 0.4954,
      "step": 13510
    },
    {
      "epoch": 0.7291025848578059,
      "grad_norm": 0.9017142091992483,
      "learning_rate": 2.5336250103401948e-06,
      "loss": 0.4026,
      "step": 13511
    },
    {
      "epoch": 0.7291565484863202,
      "grad_norm": 1.0736617218350282,
      "learning_rate": 2.5330513785226786e-06,
      "loss": 0.5487,
      "step": 13512
    },
    {
      "epoch": 0.7292105121148346,
      "grad_norm": 1.0644121129066744,
      "learning_rate": 2.5324778319782617e-06,
      "loss": 0.4671,
      "step": 13513
    },
    {
      "epoch": 0.729264475743349,
      "grad_norm": 1.1947133589908543,
      "learning_rate": 2.531904370723429e-06,
      "loss": 0.5197,
      "step": 13514
    },
    {
      "epoch": 0.7293184393718634,
      "grad_norm": 1.01309768018476,
      "learning_rate": 2.5313309947746656e-06,
      "loss": 0.5126,
      "step": 13515
    },
    {
      "epoch": 0.7293724030003778,
      "grad_norm": 0.8288609496775764,
      "learning_rate": 2.530757704148448e-06,
      "loss": 0.3009,
      "step": 13516
    },
    {
      "epoch": 0.7294263666288922,
      "grad_norm": 1.0208044554015725,
      "learning_rate": 2.5301844988612544e-06,
      "loss": 0.3674,
      "step": 13517
    },
    {
      "epoch": 0.7294803302574066,
      "grad_norm": 0.901744983702518,
      "learning_rate": 2.5296113789295597e-06,
      "loss": 0.4087,
      "step": 13518
    },
    {
      "epoch": 0.7295342938859208,
      "grad_norm": 0.9915973537370705,
      "learning_rate": 2.529038344369836e-06,
      "loss": 0.4455,
      "step": 13519
    },
    {
      "epoch": 0.7295882575144352,
      "grad_norm": 1.1429009277003839,
      "learning_rate": 2.5284653951985513e-06,
      "loss": 0.5089,
      "step": 13520
    },
    {
      "epoch": 0.7296422211429496,
      "grad_norm": 1.202684908676763,
      "learning_rate": 2.527892531432173e-06,
      "loss": 0.64,
      "step": 13521
    },
    {
      "epoch": 0.729696184771464,
      "grad_norm": 1.1542696897555655,
      "learning_rate": 2.5273197530871664e-06,
      "loss": 0.516,
      "step": 13522
    },
    {
      "epoch": 0.7297501483999784,
      "grad_norm": 0.9751391069634686,
      "learning_rate": 2.526747060179995e-06,
      "loss": 0.4185,
      "step": 13523
    },
    {
      "epoch": 0.7298041120284928,
      "grad_norm": 1.0488335385185037,
      "learning_rate": 2.526174452727116e-06,
      "loss": 0.4039,
      "step": 13524
    },
    {
      "epoch": 0.7298580756570072,
      "grad_norm": 1.059894313143924,
      "learning_rate": 2.525601930744989e-06,
      "loss": 0.5263,
      "step": 13525
    },
    {
      "epoch": 0.7299120392855215,
      "grad_norm": 1.0019448435580318,
      "learning_rate": 2.5250294942500687e-06,
      "loss": 0.525,
      "step": 13526
    },
    {
      "epoch": 0.7299660029140359,
      "grad_norm": 0.9827383612756054,
      "learning_rate": 2.5244571432588048e-06,
      "loss": 0.3808,
      "step": 13527
    },
    {
      "epoch": 0.7300199665425503,
      "grad_norm": 1.0578705980188305,
      "learning_rate": 2.52388487778765e-06,
      "loss": 0.4301,
      "step": 13528
    },
    {
      "epoch": 0.7300739301710647,
      "grad_norm": 1.0158402282433983,
      "learning_rate": 2.523312697853051e-06,
      "loss": 0.4724,
      "step": 13529
    },
    {
      "epoch": 0.7301278937995791,
      "grad_norm": 1.1174990974858967,
      "learning_rate": 2.522740603471453e-06,
      "loss": 0.6041,
      "step": 13530
    },
    {
      "epoch": 0.7301818574280935,
      "grad_norm": 1.049559431345731,
      "learning_rate": 2.522168594659298e-06,
      "loss": 0.3712,
      "step": 13531
    },
    {
      "epoch": 0.7302358210566079,
      "grad_norm": 1.0261827957359164,
      "learning_rate": 2.5215966714330266e-06,
      "loss": 0.5127,
      "step": 13532
    },
    {
      "epoch": 0.7302897846851222,
      "grad_norm": 1.0254581952971413,
      "learning_rate": 2.521024833809077e-06,
      "loss": 0.4189,
      "step": 13533
    },
    {
      "epoch": 0.7303437483136366,
      "grad_norm": 0.9211364903514767,
      "learning_rate": 2.520453081803883e-06,
      "loss": 0.4084,
      "step": 13534
    },
    {
      "epoch": 0.730397711942151,
      "grad_norm": 0.991218799677247,
      "learning_rate": 2.519881415433878e-06,
      "loss": 0.4296,
      "step": 13535
    },
    {
      "epoch": 0.7304516755706654,
      "grad_norm": 0.8895299094397353,
      "learning_rate": 2.5193098347154926e-06,
      "loss": 0.3414,
      "step": 13536
    },
    {
      "epoch": 0.7305056391991798,
      "grad_norm": 1.0574901772512477,
      "learning_rate": 2.5187383396651538e-06,
      "loss": 0.5157,
      "step": 13537
    },
    {
      "epoch": 0.7305596028276942,
      "grad_norm": 1.24006147316401,
      "learning_rate": 2.5181669302992872e-06,
      "loss": 0.5984,
      "step": 13538
    },
    {
      "epoch": 0.7306135664562086,
      "grad_norm": 1.1038539916501577,
      "learning_rate": 2.5175956066343165e-06,
      "loss": 0.6318,
      "step": 13539
    },
    {
      "epoch": 0.7306675300847228,
      "grad_norm": 1.0898484799760983,
      "learning_rate": 2.517024368686663e-06,
      "loss": 0.5657,
      "step": 13540
    },
    {
      "epoch": 0.7307214937132372,
      "grad_norm": 1.5924613782108885,
      "learning_rate": 2.516453216472742e-06,
      "loss": 0.7013,
      "step": 13541
    },
    {
      "epoch": 0.7307754573417516,
      "grad_norm": 0.8855286969647098,
      "learning_rate": 2.515882150008969e-06,
      "loss": 0.3234,
      "step": 13542
    },
    {
      "epoch": 0.730829420970266,
      "grad_norm": 0.8072671750465032,
      "learning_rate": 2.51531116931176e-06,
      "loss": 0.386,
      "step": 13543
    },
    {
      "epoch": 0.7308833845987804,
      "grad_norm": 0.9840547638066884,
      "learning_rate": 2.5147402743975223e-06,
      "loss": 0.524,
      "step": 13544
    },
    {
      "epoch": 0.7309373482272948,
      "grad_norm": 0.7673897870816924,
      "learning_rate": 2.5141694652826664e-06,
      "loss": 0.2922,
      "step": 13545
    },
    {
      "epoch": 0.7309913118558092,
      "grad_norm": 0.9949154018146212,
      "learning_rate": 2.5135987419835966e-06,
      "loss": 0.4517,
      "step": 13546
    },
    {
      "epoch": 0.7310452754843235,
      "grad_norm": 1.2938239723651193,
      "learning_rate": 2.5130281045167183e-06,
      "loss": 0.4854,
      "step": 13547
    },
    {
      "epoch": 0.7310992391128379,
      "grad_norm": 0.9376110424973763,
      "learning_rate": 2.5124575528984286e-06,
      "loss": 0.3981,
      "step": 13548
    },
    {
      "epoch": 0.7311532027413523,
      "grad_norm": 1.0644833033560988,
      "learning_rate": 2.511887087145127e-06,
      "loss": 0.4767,
      "step": 13549
    },
    {
      "epoch": 0.7312071663698667,
      "grad_norm": 1.0740469115062115,
      "learning_rate": 2.511316707273211e-06,
      "loss": 0.4535,
      "step": 13550
    },
    {
      "epoch": 0.7312611299983811,
      "grad_norm": 0.8780480252466446,
      "learning_rate": 2.510746413299072e-06,
      "loss": 0.3376,
      "step": 13551
    },
    {
      "epoch": 0.7313150936268955,
      "grad_norm": 1.1841828827984417,
      "learning_rate": 2.5101762052391023e-06,
      "loss": 0.6054,
      "step": 13552
    },
    {
      "epoch": 0.7313690572554099,
      "grad_norm": 1.0696982444712746,
      "learning_rate": 2.509606083109688e-06,
      "loss": 0.4148,
      "step": 13553
    },
    {
      "epoch": 0.7314230208839242,
      "grad_norm": 1.6449616865206025,
      "learning_rate": 2.509036046927218e-06,
      "loss": 0.6017,
      "step": 13554
    },
    {
      "epoch": 0.7314769845124386,
      "grad_norm": 0.9176156445957037,
      "learning_rate": 2.508466096708073e-06,
      "loss": 0.4544,
      "step": 13555
    },
    {
      "epoch": 0.731530948140953,
      "grad_norm": 1.0569769097494108,
      "learning_rate": 2.507896232468635e-06,
      "loss": 0.6096,
      "step": 13556
    },
    {
      "epoch": 0.7315849117694674,
      "grad_norm": 1.0097783821285016,
      "learning_rate": 2.507326454225283e-06,
      "loss": 0.4378,
      "step": 13557
    },
    {
      "epoch": 0.7316388753979818,
      "grad_norm": 1.0055814594832413,
      "learning_rate": 2.5067567619943934e-06,
      "loss": 0.6154,
      "step": 13558
    },
    {
      "epoch": 0.7316928390264962,
      "grad_norm": 1.0235893226596933,
      "learning_rate": 2.506187155792338e-06,
      "loss": 0.5591,
      "step": 13559
    },
    {
      "epoch": 0.7317468026550106,
      "grad_norm": 1.1688391134108418,
      "learning_rate": 2.505617635635489e-06,
      "loss": 0.6448,
      "step": 13560
    },
    {
      "epoch": 0.7318007662835249,
      "grad_norm": 0.8983101615377616,
      "learning_rate": 2.505048201540216e-06,
      "loss": 0.3132,
      "step": 13561
    },
    {
      "epoch": 0.7318547299120393,
      "grad_norm": 0.8376998328789775,
      "learning_rate": 2.504478853522882e-06,
      "loss": 0.4014,
      "step": 13562
    },
    {
      "epoch": 0.7319086935405537,
      "grad_norm": 1.2400460269268394,
      "learning_rate": 2.5039095915998535e-06,
      "loss": 0.5787,
      "step": 13563
    },
    {
      "epoch": 0.731962657169068,
      "grad_norm": 1.05612058390643,
      "learning_rate": 2.5033404157874903e-06,
      "loss": 0.6273,
      "step": 13564
    },
    {
      "epoch": 0.7320166207975825,
      "grad_norm": 0.9434358015120581,
      "learning_rate": 2.5027713261021513e-06,
      "loss": 0.3476,
      "step": 13565
    },
    {
      "epoch": 0.7320705844260968,
      "grad_norm": 1.0927515238425687,
      "learning_rate": 2.5022023225601926e-06,
      "loss": 0.424,
      "step": 13566
    },
    {
      "epoch": 0.7321245480546112,
      "grad_norm": 1.091197315691455,
      "learning_rate": 2.501633405177969e-06,
      "loss": 0.5136,
      "step": 13567
    },
    {
      "epoch": 0.7321785116831255,
      "grad_norm": 0.8696997598052195,
      "learning_rate": 2.501064573971832e-06,
      "loss": 0.4966,
      "step": 13568
    },
    {
      "epoch": 0.7322324753116399,
      "grad_norm": 1.0215244184162315,
      "learning_rate": 2.500495828958127e-06,
      "loss": 0.379,
      "step": 13569
    },
    {
      "epoch": 0.7322864389401543,
      "grad_norm": 1.0416543829776885,
      "learning_rate": 2.499927170153204e-06,
      "loss": 0.5633,
      "step": 13570
    },
    {
      "epoch": 0.7323404025686687,
      "grad_norm": 0.9436452647503643,
      "learning_rate": 2.499358597573405e-06,
      "loss": 0.4487,
      "step": 13571
    },
    {
      "epoch": 0.7323943661971831,
      "grad_norm": 1.0279375124071728,
      "learning_rate": 2.498790111235072e-06,
      "loss": 0.4935,
      "step": 13572
    },
    {
      "epoch": 0.7324483298256975,
      "grad_norm": 1.0926967528621099,
      "learning_rate": 2.4982217111545434e-06,
      "loss": 0.6538,
      "step": 13573
    },
    {
      "epoch": 0.7325022934542118,
      "grad_norm": 1.0496287782618587,
      "learning_rate": 2.4976533973481573e-06,
      "loss": 0.4201,
      "step": 13574
    },
    {
      "epoch": 0.7325562570827262,
      "grad_norm": 0.9466200707442465,
      "learning_rate": 2.497085169832245e-06,
      "loss": 0.3804,
      "step": 13575
    },
    {
      "epoch": 0.7326102207112406,
      "grad_norm": 1.0367277401205253,
      "learning_rate": 2.496517028623139e-06,
      "loss": 0.3724,
      "step": 13576
    },
    {
      "epoch": 0.732664184339755,
      "grad_norm": 0.9542886878599554,
      "learning_rate": 2.495948973737168e-06,
      "loss": 0.4414,
      "step": 13577
    },
    {
      "epoch": 0.7327181479682694,
      "grad_norm": 1.0080974744991127,
      "learning_rate": 2.49538100519066e-06,
      "loss": 0.4808,
      "step": 13578
    },
    {
      "epoch": 0.7327721115967838,
      "grad_norm": 1.0180316867987034,
      "learning_rate": 2.494813122999937e-06,
      "loss": 0.433,
      "step": 13579
    },
    {
      "epoch": 0.7328260752252982,
      "grad_norm": 1.1500089721177336,
      "learning_rate": 2.4942453271813214e-06,
      "loss": 0.5298,
      "step": 13580
    },
    {
      "epoch": 0.7328800388538125,
      "grad_norm": 0.9379713348556743,
      "learning_rate": 2.493677617751134e-06,
      "loss": 0.3808,
      "step": 13581
    },
    {
      "epoch": 0.7329340024823269,
      "grad_norm": 0.9380738813480819,
      "learning_rate": 2.493109994725688e-06,
      "loss": 0.5119,
      "step": 13582
    },
    {
      "epoch": 0.7329879661108413,
      "grad_norm": 1.0860798537628213,
      "learning_rate": 2.492542458121299e-06,
      "loss": 0.3805,
      "step": 13583
    },
    {
      "epoch": 0.7330419297393557,
      "grad_norm": 1.1372175037614276,
      "learning_rate": 2.491975007954278e-06,
      "loss": 0.4384,
      "step": 13584
    },
    {
      "epoch": 0.7330958933678701,
      "grad_norm": 1.0981269523554062,
      "learning_rate": 2.491407644240935e-06,
      "loss": 0.6351,
      "step": 13585
    },
    {
      "epoch": 0.7331498569963845,
      "grad_norm": 1.0645259090570434,
      "learning_rate": 2.490840366997576e-06,
      "loss": 0.5334,
      "step": 13586
    },
    {
      "epoch": 0.7332038206248989,
      "grad_norm": 1.0598964425689201,
      "learning_rate": 2.4902731762405063e-06,
      "loss": 0.5264,
      "step": 13587
    },
    {
      "epoch": 0.7332577842534131,
      "grad_norm": 0.9550600736919211,
      "learning_rate": 2.4897060719860274e-06,
      "loss": 0.4049,
      "step": 13588
    },
    {
      "epoch": 0.7333117478819275,
      "grad_norm": 0.9954754278515476,
      "learning_rate": 2.4891390542504363e-06,
      "loss": 0.4307,
      "step": 13589
    },
    {
      "epoch": 0.7333657115104419,
      "grad_norm": 0.9733028427572028,
      "learning_rate": 2.488572123050031e-06,
      "loss": 0.4574,
      "step": 13590
    },
    {
      "epoch": 0.7334196751389563,
      "grad_norm": 0.9176073096006121,
      "learning_rate": 2.488005278401106e-06,
      "loss": 0.4576,
      "step": 13591
    },
    {
      "epoch": 0.7334736387674707,
      "grad_norm": 1.0802478905146666,
      "learning_rate": 2.4874385203199524e-06,
      "loss": 0.5042,
      "step": 13592
    },
    {
      "epoch": 0.7335276023959851,
      "grad_norm": 1.0421548996171521,
      "learning_rate": 2.4868718488228597e-06,
      "loss": 0.4073,
      "step": 13593
    },
    {
      "epoch": 0.7335815660244995,
      "grad_norm": 1.0629989707680672,
      "learning_rate": 2.4863052639261144e-06,
      "loss": 0.498,
      "step": 13594
    },
    {
      "epoch": 0.7336355296530138,
      "grad_norm": 1.0240286857011502,
      "learning_rate": 2.485738765646003e-06,
      "loss": 0.4064,
      "step": 13595
    },
    {
      "epoch": 0.7336894932815282,
      "grad_norm": 0.9520207376926747,
      "learning_rate": 2.4851723539988034e-06,
      "loss": 0.3592,
      "step": 13596
    },
    {
      "epoch": 0.7337434569100426,
      "grad_norm": 1.04106787167802,
      "learning_rate": 2.4846060290007967e-06,
      "loss": 0.5149,
      "step": 13597
    },
    {
      "epoch": 0.733797420538557,
      "grad_norm": 1.0146410992623738,
      "learning_rate": 2.484039790668259e-06,
      "loss": 0.4253,
      "step": 13598
    },
    {
      "epoch": 0.7338513841670714,
      "grad_norm": 1.1754319416356693,
      "learning_rate": 2.4834736390174656e-06,
      "loss": 0.5845,
      "step": 13599
    },
    {
      "epoch": 0.7339053477955858,
      "grad_norm": 0.9747670393679951,
      "learning_rate": 2.4829075740646876e-06,
      "loss": 0.5589,
      "step": 13600
    },
    {
      "epoch": 0.7339593114241002,
      "grad_norm": 1.1321315685064057,
      "learning_rate": 2.482341595826195e-06,
      "loss": 0.5668,
      "step": 13601
    },
    {
      "epoch": 0.7340132750526145,
      "grad_norm": 1.101005115242968,
      "learning_rate": 2.4817757043182536e-06,
      "loss": 0.5146,
      "step": 13602
    },
    {
      "epoch": 0.7340672386811289,
      "grad_norm": 1.1648727827301775,
      "learning_rate": 2.4812098995571284e-06,
      "loss": 0.733,
      "step": 13603
    },
    {
      "epoch": 0.7341212023096433,
      "grad_norm": 1.0850837100201067,
      "learning_rate": 2.480644181559081e-06,
      "loss": 0.5119,
      "step": 13604
    },
    {
      "epoch": 0.7341751659381577,
      "grad_norm": 0.8897772348148945,
      "learning_rate": 2.48007855034037e-06,
      "loss": 0.3671,
      "step": 13605
    },
    {
      "epoch": 0.7342291295666721,
      "grad_norm": 0.87418832366154,
      "learning_rate": 2.4795130059172547e-06,
      "loss": 0.3947,
      "step": 13606
    },
    {
      "epoch": 0.7342830931951865,
      "grad_norm": 0.9755676194728852,
      "learning_rate": 2.478947548305986e-06,
      "loss": 0.3876,
      "step": 13607
    },
    {
      "epoch": 0.7343370568237009,
      "grad_norm": 0.977952299064298,
      "learning_rate": 2.478382177522818e-06,
      "loss": 0.4769,
      "step": 13608
    },
    {
      "epoch": 0.7343910204522152,
      "grad_norm": 1.0739838906969983,
      "learning_rate": 2.4778168935839987e-06,
      "loss": 0.4822,
      "step": 13609
    },
    {
      "epoch": 0.7344449840807296,
      "grad_norm": 0.9898485374483439,
      "learning_rate": 2.4772516965057757e-06,
      "loss": 0.4725,
      "step": 13610
    },
    {
      "epoch": 0.734498947709244,
      "grad_norm": 1.0428449620106803,
      "learning_rate": 2.476686586304393e-06,
      "loss": 0.4219,
      "step": 13611
    },
    {
      "epoch": 0.7345529113377584,
      "grad_norm": 0.9856419712172332,
      "learning_rate": 2.476121562996093e-06,
      "loss": 0.4196,
      "step": 13612
    },
    {
      "epoch": 0.7346068749662727,
      "grad_norm": 1.1462230080751992,
      "learning_rate": 2.4755566265971147e-06,
      "loss": 0.5165,
      "step": 13613
    },
    {
      "epoch": 0.7346608385947871,
      "grad_norm": 1.0550535251954816,
      "learning_rate": 2.474991777123695e-06,
      "loss": 0.3833,
      "step": 13614
    },
    {
      "epoch": 0.7347148022233015,
      "grad_norm": 1.0074439880866302,
      "learning_rate": 2.4744270145920675e-06,
      "loss": 0.4751,
      "step": 13615
    },
    {
      "epoch": 0.7347687658518158,
      "grad_norm": 0.9468332717744762,
      "learning_rate": 2.4738623390184666e-06,
      "loss": 0.3485,
      "step": 13616
    },
    {
      "epoch": 0.7348227294803302,
      "grad_norm": 0.9635066569822315,
      "learning_rate": 2.4732977504191185e-06,
      "loss": 0.4575,
      "step": 13617
    },
    {
      "epoch": 0.7348766931088446,
      "grad_norm": 1.17262902412489,
      "learning_rate": 2.472733248810251e-06,
      "loss": 0.5926,
      "step": 13618
    },
    {
      "epoch": 0.734930656737359,
      "grad_norm": 1.2751822765755176,
      "learning_rate": 2.472168834208089e-06,
      "loss": 0.5341,
      "step": 13619
    },
    {
      "epoch": 0.7349846203658734,
      "grad_norm": 0.9303027016025562,
      "learning_rate": 2.471604506628854e-06,
      "loss": 0.4653,
      "step": 13620
    },
    {
      "epoch": 0.7350385839943878,
      "grad_norm": 1.3230079238450747,
      "learning_rate": 2.4710402660887655e-06,
      "loss": 0.5251,
      "step": 13621
    },
    {
      "epoch": 0.7350925476229022,
      "grad_norm": 0.9670266134905476,
      "learning_rate": 2.4704761126040404e-06,
      "loss": 0.4706,
      "step": 13622
    },
    {
      "epoch": 0.7351465112514165,
      "grad_norm": 1.1396093191413559,
      "learning_rate": 2.469912046190894e-06,
      "loss": 0.4757,
      "step": 13623
    },
    {
      "epoch": 0.7352004748799309,
      "grad_norm": 0.9620759976206323,
      "learning_rate": 2.469348066865536e-06,
      "loss": 0.4038,
      "step": 13624
    },
    {
      "epoch": 0.7352544385084453,
      "grad_norm": 0.9298080851115904,
      "learning_rate": 2.4687841746441766e-06,
      "loss": 0.3902,
      "step": 13625
    },
    {
      "epoch": 0.7353084021369597,
      "grad_norm": 1.0178579390376412,
      "learning_rate": 2.468220369543023e-06,
      "loss": 0.4395,
      "step": 13626
    },
    {
      "epoch": 0.7353623657654741,
      "grad_norm": 0.9984670481600476,
      "learning_rate": 2.4676566515782792e-06,
      "loss": 0.4079,
      "step": 13627
    },
    {
      "epoch": 0.7354163293939885,
      "grad_norm": 1.0716192236181983,
      "learning_rate": 2.4670930207661474e-06,
      "loss": 0.459,
      "step": 13628
    },
    {
      "epoch": 0.7354702930225029,
      "grad_norm": 0.8730602459956811,
      "learning_rate": 2.4665294771228267e-06,
      "loss": 0.4428,
      "step": 13629
    },
    {
      "epoch": 0.7355242566510172,
      "grad_norm": 1.080405267970498,
      "learning_rate": 2.4659660206645153e-06,
      "loss": 0.4513,
      "step": 13630
    },
    {
      "epoch": 0.7355782202795316,
      "grad_norm": 1.1140804468789152,
      "learning_rate": 2.4654026514074043e-06,
      "loss": 0.5027,
      "step": 13631
    },
    {
      "epoch": 0.735632183908046,
      "grad_norm": 1.0587115137618324,
      "learning_rate": 2.464839369367688e-06,
      "loss": 0.429,
      "step": 13632
    },
    {
      "epoch": 0.7356861475365604,
      "grad_norm": 1.015505211807759,
      "learning_rate": 2.464276174561555e-06,
      "loss": 0.4797,
      "step": 13633
    },
    {
      "epoch": 0.7357401111650748,
      "grad_norm": 1.0489429798040013,
      "learning_rate": 2.4637130670051913e-06,
      "loss": 0.4782,
      "step": 13634
    },
    {
      "epoch": 0.7357940747935892,
      "grad_norm": 1.3163777288203045,
      "learning_rate": 2.463150046714782e-06,
      "loss": 0.603,
      "step": 13635
    },
    {
      "epoch": 0.7358480384221036,
      "grad_norm": 1.1186934562030635,
      "learning_rate": 2.46258711370651e-06,
      "loss": 0.5213,
      "step": 13636
    },
    {
      "epoch": 0.7359020020506178,
      "grad_norm": 1.078850872804275,
      "learning_rate": 2.462024267996554e-06,
      "loss": 0.5037,
      "step": 13637
    },
    {
      "epoch": 0.7359559656791322,
      "grad_norm": 1.0887400252370762,
      "learning_rate": 2.4614615096010894e-06,
      "loss": 0.4041,
      "step": 13638
    },
    {
      "epoch": 0.7360099293076466,
      "grad_norm": 0.8505007585673173,
      "learning_rate": 2.4608988385362907e-06,
      "loss": 0.4162,
      "step": 13639
    },
    {
      "epoch": 0.736063892936161,
      "grad_norm": 1.0884182926644816,
      "learning_rate": 2.460336254818331e-06,
      "loss": 0.6185,
      "step": 13640
    },
    {
      "epoch": 0.7361178565646754,
      "grad_norm": 1.0374396732770828,
      "learning_rate": 2.459773758463377e-06,
      "loss": 0.4999,
      "step": 13641
    },
    {
      "epoch": 0.7361718201931898,
      "grad_norm": 0.9805132244145264,
      "learning_rate": 2.459211349487599e-06,
      "loss": 0.5234,
      "step": 13642
    },
    {
      "epoch": 0.7362257838217041,
      "grad_norm": 1.0678965749028502,
      "learning_rate": 2.4586490279071597e-06,
      "loss": 0.4883,
      "step": 13643
    },
    {
      "epoch": 0.7362797474502185,
      "grad_norm": 0.9254355073156404,
      "learning_rate": 2.458086793738219e-06,
      "loss": 0.435,
      "step": 13644
    },
    {
      "epoch": 0.7363337110787329,
      "grad_norm": 0.9497452178753402,
      "learning_rate": 2.457524646996938e-06,
      "loss": 0.4346,
      "step": 13645
    },
    {
      "epoch": 0.7363876747072473,
      "grad_norm": 1.1111253448502625,
      "learning_rate": 2.456962587699472e-06,
      "loss": 0.4149,
      "step": 13646
    },
    {
      "epoch": 0.7364416383357617,
      "grad_norm": 1.11908850019378,
      "learning_rate": 2.4564006158619764e-06,
      "loss": 0.6062,
      "step": 13647
    },
    {
      "epoch": 0.7364956019642761,
      "grad_norm": 1.1899083835755424,
      "learning_rate": 2.4558387315006022e-06,
      "loss": 0.6,
      "step": 13648
    },
    {
      "epoch": 0.7365495655927905,
      "grad_norm": 0.983051466800003,
      "learning_rate": 2.455276934631499e-06,
      "loss": 0.402,
      "step": 13649
    },
    {
      "epoch": 0.7366035292213048,
      "grad_norm": 0.957602831588132,
      "learning_rate": 2.454715225270815e-06,
      "loss": 0.4131,
      "step": 13650
    },
    {
      "epoch": 0.7366574928498192,
      "grad_norm": 0.9957465363494457,
      "learning_rate": 2.4541536034346906e-06,
      "loss": 0.4844,
      "step": 13651
    },
    {
      "epoch": 0.7367114564783336,
      "grad_norm": 0.9593326257838737,
      "learning_rate": 2.4535920691392696e-06,
      "loss": 0.472,
      "step": 13652
    },
    {
      "epoch": 0.736765420106848,
      "grad_norm": 1.155382600831078,
      "learning_rate": 2.45303062240069e-06,
      "loss": 0.4199,
      "step": 13653
    },
    {
      "epoch": 0.7368193837353624,
      "grad_norm": 0.9215367143850469,
      "learning_rate": 2.4524692632350896e-06,
      "loss": 0.4403,
      "step": 13654
    },
    {
      "epoch": 0.7368733473638768,
      "grad_norm": 0.8759481814771279,
      "learning_rate": 2.451907991658602e-06,
      "loss": 0.3967,
      "step": 13655
    },
    {
      "epoch": 0.7369273109923912,
      "grad_norm": 1.02003606941435,
      "learning_rate": 2.4513468076873586e-06,
      "loss": 0.5502,
      "step": 13656
    },
    {
      "epoch": 0.7369812746209055,
      "grad_norm": 0.9020539164559835,
      "learning_rate": 2.450785711337488e-06,
      "loss": 0.3377,
      "step": 13657
    },
    {
      "epoch": 0.7370352382494199,
      "grad_norm": 1.2789257119461575,
      "learning_rate": 2.450224702625117e-06,
      "loss": 0.6103,
      "step": 13658
    },
    {
      "epoch": 0.7370892018779343,
      "grad_norm": 0.9467371461208075,
      "learning_rate": 2.4496637815663697e-06,
      "loss": 0.3928,
      "step": 13659
    },
    {
      "epoch": 0.7371431655064486,
      "grad_norm": 1.282642003345757,
      "learning_rate": 2.4491029481773675e-06,
      "loss": 0.6181,
      "step": 13660
    },
    {
      "epoch": 0.737197129134963,
      "grad_norm": 1.0226351416055555,
      "learning_rate": 2.448542202474229e-06,
      "loss": 0.4441,
      "step": 13661
    },
    {
      "epoch": 0.7372510927634774,
      "grad_norm": 1.0255037843302663,
      "learning_rate": 2.4479815444730714e-06,
      "loss": 0.49,
      "step": 13662
    },
    {
      "epoch": 0.7373050563919918,
      "grad_norm": 1.0827441597276284,
      "learning_rate": 2.447420974190008e-06,
      "loss": 0.5063,
      "step": 13663
    },
    {
      "epoch": 0.7373590200205061,
      "grad_norm": 1.1497380483289108,
      "learning_rate": 2.4468604916411516e-06,
      "loss": 0.5123,
      "step": 13664
    },
    {
      "epoch": 0.7374129836490205,
      "grad_norm": 0.9270781308245263,
      "learning_rate": 2.4463000968426077e-06,
      "loss": 0.4217,
      "step": 13665
    },
    {
      "epoch": 0.7374669472775349,
      "grad_norm": 1.1850070057491957,
      "learning_rate": 2.4457397898104856e-06,
      "loss": 0.4693,
      "step": 13666
    },
    {
      "epoch": 0.7375209109060493,
      "grad_norm": 0.9889364870837063,
      "learning_rate": 2.445179570560887e-06,
      "loss": 0.4101,
      "step": 13667
    },
    {
      "epoch": 0.7375748745345637,
      "grad_norm": 1.0182072520515841,
      "learning_rate": 2.4446194391099146e-06,
      "loss": 0.437,
      "step": 13668
    },
    {
      "epoch": 0.7376288381630781,
      "grad_norm": 1.3250955693155826,
      "learning_rate": 2.4440593954736666e-06,
      "loss": 0.4124,
      "step": 13669
    },
    {
      "epoch": 0.7376828017915925,
      "grad_norm": 1.0018573914160396,
      "learning_rate": 2.4434994396682397e-06,
      "loss": 0.6161,
      "step": 13670
    },
    {
      "epoch": 0.7377367654201068,
      "grad_norm": 0.8994548574169627,
      "learning_rate": 2.442939571709729e-06,
      "loss": 0.3547,
      "step": 13671
    },
    {
      "epoch": 0.7377907290486212,
      "grad_norm": 1.0749108245670753,
      "learning_rate": 2.442379791614222e-06,
      "loss": 0.505,
      "step": 13672
    },
    {
      "epoch": 0.7378446926771356,
      "grad_norm": 0.7342420213281813,
      "learning_rate": 2.441820099397809e-06,
      "loss": 0.3142,
      "step": 13673
    },
    {
      "epoch": 0.73789865630565,
      "grad_norm": 1.0702027751905223,
      "learning_rate": 2.441260495076577e-06,
      "loss": 0.5122,
      "step": 13674
    },
    {
      "epoch": 0.7379526199341644,
      "grad_norm": 1.1150972268539592,
      "learning_rate": 2.4407009786666083e-06,
      "loss": 0.3647,
      "step": 13675
    },
    {
      "epoch": 0.7380065835626788,
      "grad_norm": 0.9424367369836135,
      "learning_rate": 2.4401415501839856e-06,
      "loss": 0.494,
      "step": 13676
    },
    {
      "epoch": 0.7380605471911932,
      "grad_norm": 1.1507105030112639,
      "learning_rate": 2.439582209644786e-06,
      "loss": 0.4721,
      "step": 13677
    },
    {
      "epoch": 0.7381145108197075,
      "grad_norm": 0.9715291468544067,
      "learning_rate": 2.4390229570650877e-06,
      "loss": 0.4561,
      "step": 13678
    },
    {
      "epoch": 0.7381684744482219,
      "grad_norm": 1.315805504367494,
      "learning_rate": 2.438463792460961e-06,
      "loss": 0.6337,
      "step": 13679
    },
    {
      "epoch": 0.7382224380767363,
      "grad_norm": 1.0276749016827882,
      "learning_rate": 2.4379047158484782e-06,
      "loss": 0.5278,
      "step": 13680
    },
    {
      "epoch": 0.7382764017052507,
      "grad_norm": 1.1476555799581736,
      "learning_rate": 2.4373457272437074e-06,
      "loss": 0.5078,
      "step": 13681
    },
    {
      "epoch": 0.7383303653337651,
      "grad_norm": 1.0758844904611882,
      "learning_rate": 2.4367868266627156e-06,
      "loss": 0.4978,
      "step": 13682
    },
    {
      "epoch": 0.7383843289622795,
      "grad_norm": 1.1040785181034605,
      "learning_rate": 2.4362280141215656e-06,
      "loss": 0.4797,
      "step": 13683
    },
    {
      "epoch": 0.7384382925907939,
      "grad_norm": 0.7164325344886534,
      "learning_rate": 2.435669289636318e-06,
      "loss": 0.2639,
      "step": 13684
    },
    {
      "epoch": 0.7384922562193081,
      "grad_norm": 1.150942715004878,
      "learning_rate": 2.4351106532230324e-06,
      "loss": 0.6395,
      "step": 13685
    },
    {
      "epoch": 0.7385462198478225,
      "grad_norm": 1.0671006193424986,
      "learning_rate": 2.434552104897763e-06,
      "loss": 0.5917,
      "step": 13686
    },
    {
      "epoch": 0.7386001834763369,
      "grad_norm": 1.122983313886147,
      "learning_rate": 2.4339936446765626e-06,
      "loss": 0.5119,
      "step": 13687
    },
    {
      "epoch": 0.7386541471048513,
      "grad_norm": 1.149519177163638,
      "learning_rate": 2.433435272575483e-06,
      "loss": 0.5393,
      "step": 13688
    },
    {
      "epoch": 0.7387081107333657,
      "grad_norm": 0.9418991323786952,
      "learning_rate": 2.4328769886105727e-06,
      "loss": 0.46,
      "step": 13689
    },
    {
      "epoch": 0.7387620743618801,
      "grad_norm": 1.0207255114982146,
      "learning_rate": 2.432318792797876e-06,
      "loss": 0.6218,
      "step": 13690
    },
    {
      "epoch": 0.7388160379903945,
      "grad_norm": 1.0603094091727008,
      "learning_rate": 2.431760685153438e-06,
      "loss": 0.4732,
      "step": 13691
    },
    {
      "epoch": 0.7388700016189088,
      "grad_norm": 1.0561993270447907,
      "learning_rate": 2.431202665693299e-06,
      "loss": 0.65,
      "step": 13692
    },
    {
      "epoch": 0.7389239652474232,
      "grad_norm": 1.234141060046457,
      "learning_rate": 2.4306447344334944e-06,
      "loss": 0.5043,
      "step": 13693
    },
    {
      "epoch": 0.7389779288759376,
      "grad_norm": 0.9253077232396937,
      "learning_rate": 2.4300868913900614e-06,
      "loss": 0.3343,
      "step": 13694
    },
    {
      "epoch": 0.739031892504452,
      "grad_norm": 0.9937966782348889,
      "learning_rate": 2.429529136579034e-06,
      "loss": 0.4951,
      "step": 13695
    },
    {
      "epoch": 0.7390858561329664,
      "grad_norm": 0.8713841586850773,
      "learning_rate": 2.428971470016441e-06,
      "loss": 0.3861,
      "step": 13696
    },
    {
      "epoch": 0.7391398197614808,
      "grad_norm": 0.9315616895734296,
      "learning_rate": 2.428413891718311e-06,
      "loss": 0.3671,
      "step": 13697
    },
    {
      "epoch": 0.7391937833899952,
      "grad_norm": 0.9724083157325145,
      "learning_rate": 2.4278564017006697e-06,
      "loss": 0.4183,
      "step": 13698
    },
    {
      "epoch": 0.7392477470185095,
      "grad_norm": 0.8592093201675339,
      "learning_rate": 2.427298999979541e-06,
      "loss": 0.4937,
      "step": 13699
    },
    {
      "epoch": 0.7393017106470239,
      "grad_norm": 0.8841585858935,
      "learning_rate": 2.426741686570942e-06,
      "loss": 0.5018,
      "step": 13700
    },
    {
      "epoch": 0.7393556742755383,
      "grad_norm": 1.2010347959847076,
      "learning_rate": 2.426184461490892e-06,
      "loss": 0.5115,
      "step": 13701
    },
    {
      "epoch": 0.7394096379040527,
      "grad_norm": 0.9595410183125147,
      "learning_rate": 2.425627324755407e-06,
      "loss": 0.3275,
      "step": 13702
    },
    {
      "epoch": 0.7394636015325671,
      "grad_norm": 1.0116580052975386,
      "learning_rate": 2.425070276380499e-06,
      "loss": 0.4731,
      "step": 13703
    },
    {
      "epoch": 0.7395175651610815,
      "grad_norm": 1.1331323914910418,
      "learning_rate": 2.4245133163821786e-06,
      "loss": 0.5286,
      "step": 13704
    },
    {
      "epoch": 0.7395715287895959,
      "grad_norm": 0.965721998091864,
      "learning_rate": 2.4239564447764526e-06,
      "loss": 0.4567,
      "step": 13705
    },
    {
      "epoch": 0.7396254924181102,
      "grad_norm": 1.0871088715804735,
      "learning_rate": 2.4233996615793264e-06,
      "loss": 0.4945,
      "step": 13706
    },
    {
      "epoch": 0.7396794560466246,
      "grad_norm": 0.8891263507732957,
      "learning_rate": 2.422842966806803e-06,
      "loss": 0.3681,
      "step": 13707
    },
    {
      "epoch": 0.739733419675139,
      "grad_norm": 1.0631152902708518,
      "learning_rate": 2.422286360474881e-06,
      "loss": 0.4676,
      "step": 13708
    },
    {
      "epoch": 0.7397873833036533,
      "grad_norm": 1.0740771947266015,
      "learning_rate": 2.42172984259956e-06,
      "loss": 0.4305,
      "step": 13709
    },
    {
      "epoch": 0.7398413469321677,
      "grad_norm": 1.366298892919677,
      "learning_rate": 2.4211734131968333e-06,
      "loss": 0.4217,
      "step": 13710
    },
    {
      "epoch": 0.7398953105606821,
      "grad_norm": 0.8213648237364805,
      "learning_rate": 2.4206170722826937e-06,
      "loss": 0.4172,
      "step": 13711
    },
    {
      "epoch": 0.7399492741891964,
      "grad_norm": 0.8316913464395386,
      "learning_rate": 2.420060819873132e-06,
      "loss": 0.4531,
      "step": 13712
    },
    {
      "epoch": 0.7400032378177108,
      "grad_norm": 1.3046650194159426,
      "learning_rate": 2.4195046559841328e-06,
      "loss": 0.5476,
      "step": 13713
    },
    {
      "epoch": 0.7400572014462252,
      "grad_norm": 0.9939936224287365,
      "learning_rate": 2.418948580631683e-06,
      "loss": 0.453,
      "step": 13714
    },
    {
      "epoch": 0.7401111650747396,
      "grad_norm": 1.0473066879246253,
      "learning_rate": 2.4183925938317636e-06,
      "loss": 0.4768,
      "step": 13715
    },
    {
      "epoch": 0.740165128703254,
      "grad_norm": 1.101800801714389,
      "learning_rate": 2.4178366956003555e-06,
      "loss": 0.5632,
      "step": 13716
    },
    {
      "epoch": 0.7402190923317684,
      "grad_norm": 1.151336232101916,
      "learning_rate": 2.417280885953435e-06,
      "loss": 0.5397,
      "step": 13717
    },
    {
      "epoch": 0.7402730559602828,
      "grad_norm": 1.1003132418092016,
      "learning_rate": 2.416725164906976e-06,
      "loss": 0.5815,
      "step": 13718
    },
    {
      "epoch": 0.7403270195887971,
      "grad_norm": 1.1196128294876557,
      "learning_rate": 2.416169532476953e-06,
      "loss": 0.4559,
      "step": 13719
    },
    {
      "epoch": 0.7403809832173115,
      "grad_norm": 1.2847161137848364,
      "learning_rate": 2.415613988679332e-06,
      "loss": 0.5314,
      "step": 13720
    },
    {
      "epoch": 0.7404349468458259,
      "grad_norm": 0.905725827509395,
      "learning_rate": 2.4150585335300823e-06,
      "loss": 0.3688,
      "step": 13721
    },
    {
      "epoch": 0.7404889104743403,
      "grad_norm": 0.9331445897672912,
      "learning_rate": 2.4145031670451664e-06,
      "loss": 0.418,
      "step": 13722
    },
    {
      "epoch": 0.7405428741028547,
      "grad_norm": 0.9678438992702885,
      "learning_rate": 2.413947889240548e-06,
      "loss": 0.3283,
      "step": 13723
    },
    {
      "epoch": 0.7405968377313691,
      "grad_norm": 0.9668414890025637,
      "learning_rate": 2.4133927001321844e-06,
      "loss": 0.3424,
      "step": 13724
    },
    {
      "epoch": 0.7406508013598835,
      "grad_norm": 1.1793042044889026,
      "learning_rate": 2.4128375997360345e-06,
      "loss": 0.6142,
      "step": 13725
    },
    {
      "epoch": 0.7407047649883978,
      "grad_norm": 1.0491357141544038,
      "learning_rate": 2.4122825880680524e-06,
      "loss": 0.6195,
      "step": 13726
    },
    {
      "epoch": 0.7407587286169122,
      "grad_norm": 1.0182503156057756,
      "learning_rate": 2.411727665144187e-06,
      "loss": 0.4819,
      "step": 13727
    },
    {
      "epoch": 0.7408126922454266,
      "grad_norm": 0.8484597813744122,
      "learning_rate": 2.411172830980389e-06,
      "loss": 0.3027,
      "step": 13728
    },
    {
      "epoch": 0.740866655873941,
      "grad_norm": 0.8811491422541216,
      "learning_rate": 2.4106180855926047e-06,
      "loss": 0.5434,
      "step": 13729
    },
    {
      "epoch": 0.7409206195024554,
      "grad_norm": 1.1984682472476476,
      "learning_rate": 2.410063428996779e-06,
      "loss": 0.4817,
      "step": 13730
    },
    {
      "epoch": 0.7409745831309698,
      "grad_norm": 0.9503037022430559,
      "learning_rate": 2.409508861208852e-06,
      "loss": 0.4125,
      "step": 13731
    },
    {
      "epoch": 0.7410285467594842,
      "grad_norm": 1.0514323190152437,
      "learning_rate": 2.4089543822447625e-06,
      "loss": 0.5191,
      "step": 13732
    },
    {
      "epoch": 0.7410825103879984,
      "grad_norm": 0.9450518949259739,
      "learning_rate": 2.4083999921204494e-06,
      "loss": 0.3329,
      "step": 13733
    },
    {
      "epoch": 0.7411364740165128,
      "grad_norm": 1.1163883584995937,
      "learning_rate": 2.407845690851843e-06,
      "loss": 0.3758,
      "step": 13734
    },
    {
      "epoch": 0.7411904376450272,
      "grad_norm": 1.1852532065061756,
      "learning_rate": 2.4072914784548758e-06,
      "loss": 0.5194,
      "step": 13735
    },
    {
      "epoch": 0.7412444012735416,
      "grad_norm": 1.008302472986563,
      "learning_rate": 2.4067373549454763e-06,
      "loss": 0.4232,
      "step": 13736
    },
    {
      "epoch": 0.741298364902056,
      "grad_norm": 0.9919007032138214,
      "learning_rate": 2.406183320339571e-06,
      "loss": 0.5281,
      "step": 13737
    },
    {
      "epoch": 0.7413523285305704,
      "grad_norm": 1.0848160318087539,
      "learning_rate": 2.4056293746530827e-06,
      "loss": 0.4149,
      "step": 13738
    },
    {
      "epoch": 0.7414062921590848,
      "grad_norm": 1.2642470827812446,
      "learning_rate": 2.405075517901933e-06,
      "loss": 0.6762,
      "step": 13739
    },
    {
      "epoch": 0.7414602557875991,
      "grad_norm": 0.8860661897964235,
      "learning_rate": 2.404521750102041e-06,
      "loss": 0.375,
      "step": 13740
    },
    {
      "epoch": 0.7415142194161135,
      "grad_norm": 1.071061933988195,
      "learning_rate": 2.403968071269321e-06,
      "loss": 0.5575,
      "step": 13741
    },
    {
      "epoch": 0.7415681830446279,
      "grad_norm": 0.8810004739765295,
      "learning_rate": 2.403414481419687e-06,
      "loss": 0.4363,
      "step": 13742
    },
    {
      "epoch": 0.7416221466731423,
      "grad_norm": 0.8208026749574069,
      "learning_rate": 2.4028609805690497e-06,
      "loss": 0.4042,
      "step": 13743
    },
    {
      "epoch": 0.7416761103016567,
      "grad_norm": 1.0613778563980258,
      "learning_rate": 2.4023075687333164e-06,
      "loss": 0.5204,
      "step": 13744
    },
    {
      "epoch": 0.7417300739301711,
      "grad_norm": 1.103517846305272,
      "learning_rate": 2.4017542459283942e-06,
      "loss": 0.5989,
      "step": 13745
    },
    {
      "epoch": 0.7417840375586855,
      "grad_norm": 1.0642533134106034,
      "learning_rate": 2.4012010121701853e-06,
      "loss": 0.4206,
      "step": 13746
    },
    {
      "epoch": 0.7418380011871998,
      "grad_norm": 1.0412196996176735,
      "learning_rate": 2.400647867474592e-06,
      "loss": 0.4719,
      "step": 13747
    },
    {
      "epoch": 0.7418919648157142,
      "grad_norm": 1.032660266180819,
      "learning_rate": 2.4000948118575095e-06,
      "loss": 0.4689,
      "step": 13748
    },
    {
      "epoch": 0.7419459284442286,
      "grad_norm": 0.912428024473744,
      "learning_rate": 2.399541845334834e-06,
      "loss": 0.3993,
      "step": 13749
    },
    {
      "epoch": 0.741999892072743,
      "grad_norm": 1.1320335935144874,
      "learning_rate": 2.398988967922459e-06,
      "loss": 0.4421,
      "step": 13750
    },
    {
      "epoch": 0.7420538557012574,
      "grad_norm": 1.0017586660319573,
      "learning_rate": 2.3984361796362736e-06,
      "loss": 0.3701,
      "step": 13751
    },
    {
      "epoch": 0.7421078193297718,
      "grad_norm": 1.080069741561656,
      "learning_rate": 2.397883480492167e-06,
      "loss": 0.4945,
      "step": 13752
    },
    {
      "epoch": 0.7421617829582862,
      "grad_norm": 1.1308331905670144,
      "learning_rate": 2.397330870506023e-06,
      "loss": 0.3845,
      "step": 13753
    },
    {
      "epoch": 0.7422157465868005,
      "grad_norm": 1.2991499534930335,
      "learning_rate": 2.3967783496937257e-06,
      "loss": 0.5128,
      "step": 13754
    },
    {
      "epoch": 0.7422697102153148,
      "grad_norm": 1.0622439811957038,
      "learning_rate": 2.3962259180711535e-06,
      "loss": 0.4933,
      "step": 13755
    },
    {
      "epoch": 0.7423236738438292,
      "grad_norm": 0.9279697015938064,
      "learning_rate": 2.395673575654186e-06,
      "loss": 0.3394,
      "step": 13756
    },
    {
      "epoch": 0.7423776374723436,
      "grad_norm": 1.152018746229512,
      "learning_rate": 2.3951213224586952e-06,
      "loss": 0.5376,
      "step": 13757
    },
    {
      "epoch": 0.742431601100858,
      "grad_norm": 1.109857964289796,
      "learning_rate": 2.394569158500556e-06,
      "loss": 0.4527,
      "step": 13758
    },
    {
      "epoch": 0.7424855647293724,
      "grad_norm": 1.0246520040970153,
      "learning_rate": 2.3940170837956355e-06,
      "loss": 0.5165,
      "step": 13759
    },
    {
      "epoch": 0.7425395283578868,
      "grad_norm": 0.9894776757505382,
      "learning_rate": 2.393465098359803e-06,
      "loss": 0.4319,
      "step": 13760
    },
    {
      "epoch": 0.7425934919864011,
      "grad_norm": 0.9762314606003716,
      "learning_rate": 2.3929132022089226e-06,
      "loss": 0.3804,
      "step": 13761
    },
    {
      "epoch": 0.7426474556149155,
      "grad_norm": 1.012694306685135,
      "learning_rate": 2.3923613953588557e-06,
      "loss": 0.5295,
      "step": 13762
    },
    {
      "epoch": 0.7427014192434299,
      "grad_norm": 0.9671423106023049,
      "learning_rate": 2.391809677825463e-06,
      "loss": 0.4376,
      "step": 13763
    },
    {
      "epoch": 0.7427553828719443,
      "grad_norm": 0.9361471538045009,
      "learning_rate": 2.3912580496246005e-06,
      "loss": 0.3615,
      "step": 13764
    },
    {
      "epoch": 0.7428093465004587,
      "grad_norm": 0.9060281298727937,
      "learning_rate": 2.390706510772122e-06,
      "loss": 0.3711,
      "step": 13765
    },
    {
      "epoch": 0.7428633101289731,
      "grad_norm": 0.9913451955992864,
      "learning_rate": 2.3901550612838813e-06,
      "loss": 0.4483,
      "step": 13766
    },
    {
      "epoch": 0.7429172737574875,
      "grad_norm": 1.055559688541436,
      "learning_rate": 2.389603701175726e-06,
      "loss": 0.5486,
      "step": 13767
    },
    {
      "epoch": 0.7429712373860018,
      "grad_norm": 0.9993980484579671,
      "learning_rate": 2.3890524304635046e-06,
      "loss": 0.5141,
      "step": 13768
    },
    {
      "epoch": 0.7430252010145162,
      "grad_norm": 1.1904213081696895,
      "learning_rate": 2.3885012491630584e-06,
      "loss": 0.3744,
      "step": 13769
    },
    {
      "epoch": 0.7430791646430306,
      "grad_norm": 0.9682592631167551,
      "learning_rate": 2.3879501572902297e-06,
      "loss": 0.4878,
      "step": 13770
    },
    {
      "epoch": 0.743133128271545,
      "grad_norm": 1.015666374152435,
      "learning_rate": 2.3873991548608585e-06,
      "loss": 0.4673,
      "step": 13771
    },
    {
      "epoch": 0.7431870919000594,
      "grad_norm": 1.036437490523783,
      "learning_rate": 2.3868482418907802e-06,
      "loss": 0.4024,
      "step": 13772
    },
    {
      "epoch": 0.7432410555285738,
      "grad_norm": 0.9935265710147462,
      "learning_rate": 2.3862974183958296e-06,
      "loss": 0.4879,
      "step": 13773
    },
    {
      "epoch": 0.7432950191570882,
      "grad_norm": 1.1326060695396332,
      "learning_rate": 2.3857466843918373e-06,
      "loss": 0.4733,
      "step": 13774
    },
    {
      "epoch": 0.7433489827856025,
      "grad_norm": 1.310265642735977,
      "learning_rate": 2.385196039894633e-06,
      "loss": 0.628,
      "step": 13775
    },
    {
      "epoch": 0.7434029464141169,
      "grad_norm": 0.8669568076758674,
      "learning_rate": 2.3846454849200407e-06,
      "loss": 0.4615,
      "step": 13776
    },
    {
      "epoch": 0.7434569100426313,
      "grad_norm": 1.0805997239011107,
      "learning_rate": 2.384095019483885e-06,
      "loss": 0.5001,
      "step": 13777
    },
    {
      "epoch": 0.7435108736711457,
      "grad_norm": 1.096603561482585,
      "learning_rate": 2.3835446436019867e-06,
      "loss": 0.5918,
      "step": 13778
    },
    {
      "epoch": 0.74356483729966,
      "grad_norm": 1.2216267516814752,
      "learning_rate": 2.3829943572901647e-06,
      "loss": 0.5178,
      "step": 13779
    },
    {
      "epoch": 0.7436188009281745,
      "grad_norm": 1.1274079123242335,
      "learning_rate": 2.3824441605642336e-06,
      "loss": 0.4917,
      "step": 13780
    },
    {
      "epoch": 0.7436727645566887,
      "grad_norm": 1.0894475553323304,
      "learning_rate": 2.3818940534400094e-06,
      "loss": 0.4137,
      "step": 13781
    },
    {
      "epoch": 0.7437267281852031,
      "grad_norm": 1.5699155312577897,
      "learning_rate": 2.381344035933299e-06,
      "loss": 0.6076,
      "step": 13782
    },
    {
      "epoch": 0.7437806918137175,
      "grad_norm": 0.8814909667958198,
      "learning_rate": 2.3807941080599127e-06,
      "loss": 0.4378,
      "step": 13783
    },
    {
      "epoch": 0.7438346554422319,
      "grad_norm": 1.2751995379648908,
      "learning_rate": 2.380244269835655e-06,
      "loss": 0.5956,
      "step": 13784
    },
    {
      "epoch": 0.7438886190707463,
      "grad_norm": 1.036647707405406,
      "learning_rate": 2.379694521276329e-06,
      "loss": 0.5286,
      "step": 13785
    },
    {
      "epoch": 0.7439425826992607,
      "grad_norm": 0.9365790629900254,
      "learning_rate": 2.3791448623977355e-06,
      "loss": 0.5007,
      "step": 13786
    },
    {
      "epoch": 0.7439965463277751,
      "grad_norm": 0.894772996811788,
      "learning_rate": 2.378595293215672e-06,
      "loss": 0.3147,
      "step": 13787
    },
    {
      "epoch": 0.7440505099562894,
      "grad_norm": 1.053407350631881,
      "learning_rate": 2.3780458137459357e-06,
      "loss": 0.531,
      "step": 13788
    },
    {
      "epoch": 0.7441044735848038,
      "grad_norm": 1.0304153325450238,
      "learning_rate": 2.377496424004315e-06,
      "loss": 0.4198,
      "step": 13789
    },
    {
      "epoch": 0.7441584372133182,
      "grad_norm": 0.9101456239384091,
      "learning_rate": 2.376947124006603e-06,
      "loss": 0.3002,
      "step": 13790
    },
    {
      "epoch": 0.7442124008418326,
      "grad_norm": 1.1174036635410245,
      "learning_rate": 2.376397913768586e-06,
      "loss": 0.4323,
      "step": 13791
    },
    {
      "epoch": 0.744266364470347,
      "grad_norm": 1.1049241632795346,
      "learning_rate": 2.3758487933060487e-06,
      "loss": 0.4443,
      "step": 13792
    },
    {
      "epoch": 0.7443203280988614,
      "grad_norm": 1.2015833386334946,
      "learning_rate": 2.375299762634774e-06,
      "loss": 0.4555,
      "step": 13793
    },
    {
      "epoch": 0.7443742917273758,
      "grad_norm": 0.8594089736446202,
      "learning_rate": 2.3747508217705416e-06,
      "loss": 0.4406,
      "step": 13794
    },
    {
      "epoch": 0.7444282553558901,
      "grad_norm": 0.8742014033470046,
      "learning_rate": 2.3742019707291296e-06,
      "loss": 0.3099,
      "step": 13795
    },
    {
      "epoch": 0.7444822189844045,
      "grad_norm": 1.060963690080123,
      "learning_rate": 2.3736532095263105e-06,
      "loss": 0.4565,
      "step": 13796
    },
    {
      "epoch": 0.7445361826129189,
      "grad_norm": 0.8320523048708396,
      "learning_rate": 2.3731045381778557e-06,
      "loss": 0.3359,
      "step": 13797
    },
    {
      "epoch": 0.7445901462414333,
      "grad_norm": 0.8573467769601333,
      "learning_rate": 2.3725559566995373e-06,
      "loss": 0.4066,
      "step": 13798
    },
    {
      "epoch": 0.7446441098699477,
      "grad_norm": 1.0673434379914584,
      "learning_rate": 2.3720074651071202e-06,
      "loss": 0.37,
      "step": 13799
    },
    {
      "epoch": 0.7446980734984621,
      "grad_norm": 1.2937021763798593,
      "learning_rate": 2.3714590634163696e-06,
      "loss": 0.5532,
      "step": 13800
    },
    {
      "epoch": 0.7447520371269765,
      "grad_norm": 1.103289798143979,
      "learning_rate": 2.3709107516430464e-06,
      "loss": 0.4403,
      "step": 13801
    },
    {
      "epoch": 0.7448060007554907,
      "grad_norm": 1.1704785439797945,
      "learning_rate": 2.370362529802911e-06,
      "loss": 0.503,
      "step": 13802
    },
    {
      "epoch": 0.7448599643840051,
      "grad_norm": 1.038288956407734,
      "learning_rate": 2.3698143979117177e-06,
      "loss": 0.5855,
      "step": 13803
    },
    {
      "epoch": 0.7449139280125195,
      "grad_norm": 0.8857637613251428,
      "learning_rate": 2.3692663559852212e-06,
      "loss": 0.3766,
      "step": 13804
    },
    {
      "epoch": 0.7449678916410339,
      "grad_norm": 1.1413475335896897,
      "learning_rate": 2.3687184040391737e-06,
      "loss": 0.5588,
      "step": 13805
    },
    {
      "epoch": 0.7450218552695483,
      "grad_norm": 1.04276318580816,
      "learning_rate": 2.3681705420893227e-06,
      "loss": 0.4103,
      "step": 13806
    },
    {
      "epoch": 0.7450758188980627,
      "grad_norm": 1.2485561003632042,
      "learning_rate": 2.367622770151415e-06,
      "loss": 0.5486,
      "step": 13807
    },
    {
      "epoch": 0.7451297825265771,
      "grad_norm": 1.1290160104078468,
      "learning_rate": 2.3670750882411935e-06,
      "loss": 0.4975,
      "step": 13808
    },
    {
      "epoch": 0.7451837461550914,
      "grad_norm": 0.8360162742896027,
      "learning_rate": 2.3665274963743995e-06,
      "loss": 0.277,
      "step": 13809
    },
    {
      "epoch": 0.7452377097836058,
      "grad_norm": 0.8628724662214751,
      "learning_rate": 2.3659799945667717e-06,
      "loss": 0.3399,
      "step": 13810
    },
    {
      "epoch": 0.7452916734121202,
      "grad_norm": 0.8793707646838839,
      "learning_rate": 2.365432582834045e-06,
      "loss": 0.357,
      "step": 13811
    },
    {
      "epoch": 0.7453456370406346,
      "grad_norm": 0.9861971324201783,
      "learning_rate": 2.3648852611919544e-06,
      "loss": 0.4494,
      "step": 13812
    },
    {
      "epoch": 0.745399600669149,
      "grad_norm": 1.2487522244358646,
      "learning_rate": 2.3643380296562284e-06,
      "loss": 0.7373,
      "step": 13813
    },
    {
      "epoch": 0.7454535642976634,
      "grad_norm": 0.8530605760902752,
      "learning_rate": 2.3637908882425954e-06,
      "loss": 0.3561,
      "step": 13814
    },
    {
      "epoch": 0.7455075279261778,
      "grad_norm": 0.8584480902636806,
      "learning_rate": 2.3632438369667817e-06,
      "loss": 0.3312,
      "step": 13815
    },
    {
      "epoch": 0.7455614915546921,
      "grad_norm": 1.022913835148295,
      "learning_rate": 2.362696875844511e-06,
      "loss": 0.4936,
      "step": 13816
    },
    {
      "epoch": 0.7456154551832065,
      "grad_norm": 0.91513934472465,
      "learning_rate": 2.3621500048915004e-06,
      "loss": 0.4475,
      "step": 13817
    },
    {
      "epoch": 0.7456694188117209,
      "grad_norm": 1.073321145457848,
      "learning_rate": 2.3616032241234697e-06,
      "loss": 0.5151,
      "step": 13818
    },
    {
      "epoch": 0.7457233824402353,
      "grad_norm": 1.067086769307415,
      "learning_rate": 2.361056533556133e-06,
      "loss": 0.5296,
      "step": 13819
    },
    {
      "epoch": 0.7457773460687497,
      "grad_norm": 1.0799638784072596,
      "learning_rate": 2.360509933205203e-06,
      "loss": 0.4912,
      "step": 13820
    },
    {
      "epoch": 0.7458313096972641,
      "grad_norm": 1.0849559030965712,
      "learning_rate": 2.359963423086391e-06,
      "loss": 0.4463,
      "step": 13821
    },
    {
      "epoch": 0.7458852733257785,
      "grad_norm": 1.0863929223532487,
      "learning_rate": 2.359417003215402e-06,
      "loss": 0.4017,
      "step": 13822
    },
    {
      "epoch": 0.7459392369542928,
      "grad_norm": 0.9950650456064172,
      "learning_rate": 2.3588706736079424e-06,
      "loss": 0.4828,
      "step": 13823
    },
    {
      "epoch": 0.7459932005828072,
      "grad_norm": 0.7921325208881091,
      "learning_rate": 2.358324434279713e-06,
      "loss": 0.3795,
      "step": 13824
    },
    {
      "epoch": 0.7460471642113216,
      "grad_norm": 0.9676722532986726,
      "learning_rate": 2.357778285246414e-06,
      "loss": 0.4836,
      "step": 13825
    },
    {
      "epoch": 0.746101127839836,
      "grad_norm": 0.9558544113196051,
      "learning_rate": 2.357232226523742e-06,
      "loss": 0.4746,
      "step": 13826
    },
    {
      "epoch": 0.7461550914683504,
      "grad_norm": 1.0816936781493736,
      "learning_rate": 2.3566862581273905e-06,
      "loss": 0.5095,
      "step": 13827
    },
    {
      "epoch": 0.7462090550968647,
      "grad_norm": 1.0018724730405548,
      "learning_rate": 2.356140380073053e-06,
      "loss": 0.4796,
      "step": 13828
    },
    {
      "epoch": 0.7462630187253791,
      "grad_norm": 0.9495529639720932,
      "learning_rate": 2.3555945923764166e-06,
      "loss": 0.4459,
      "step": 13829
    },
    {
      "epoch": 0.7463169823538934,
      "grad_norm": 1.0916657440131894,
      "learning_rate": 2.355048895053171e-06,
      "loss": 0.4417,
      "step": 13830
    },
    {
      "epoch": 0.7463709459824078,
      "grad_norm": 0.9756424395715568,
      "learning_rate": 2.354503288118997e-06,
      "loss": 0.5197,
      "step": 13831
    },
    {
      "epoch": 0.7464249096109222,
      "grad_norm": 0.946673421669155,
      "learning_rate": 2.353957771589576e-06,
      "loss": 0.3948,
      "step": 13832
    },
    {
      "epoch": 0.7464788732394366,
      "grad_norm": 1.1069637139217243,
      "learning_rate": 2.353412345480587e-06,
      "loss": 0.5882,
      "step": 13833
    },
    {
      "epoch": 0.746532836867951,
      "grad_norm": 1.0743254544185008,
      "learning_rate": 2.3528670098077074e-06,
      "loss": 0.517,
      "step": 13834
    },
    {
      "epoch": 0.7465868004964654,
      "grad_norm": 1.0178847687459385,
      "learning_rate": 2.3523217645866094e-06,
      "loss": 0.4745,
      "step": 13835
    },
    {
      "epoch": 0.7466407641249798,
      "grad_norm": 0.854359428661137,
      "learning_rate": 2.351776609832965e-06,
      "loss": 0.3678,
      "step": 13836
    },
    {
      "epoch": 0.7466947277534941,
      "grad_norm": 0.9761637169520345,
      "learning_rate": 2.3512315455624426e-06,
      "loss": 0.532,
      "step": 13837
    },
    {
      "epoch": 0.7467486913820085,
      "grad_norm": 0.954597902077252,
      "learning_rate": 2.3506865717907064e-06,
      "loss": 0.4454,
      "step": 13838
    },
    {
      "epoch": 0.7468026550105229,
      "grad_norm": 1.122086390046103,
      "learning_rate": 2.35014168853342e-06,
      "loss": 0.5337,
      "step": 13839
    },
    {
      "epoch": 0.7468566186390373,
      "grad_norm": 1.091415959701526,
      "learning_rate": 2.349596895806244e-06,
      "loss": 0.4682,
      "step": 13840
    },
    {
      "epoch": 0.7469105822675517,
      "grad_norm": 0.9600569014067727,
      "learning_rate": 2.3490521936248365e-06,
      "loss": 0.4434,
      "step": 13841
    },
    {
      "epoch": 0.7469645458960661,
      "grad_norm": 0.9319223279167853,
      "learning_rate": 2.348507582004853e-06,
      "loss": 0.4092,
      "step": 13842
    },
    {
      "epoch": 0.7470185095245805,
      "grad_norm": 1.1145067796071018,
      "learning_rate": 2.347963060961946e-06,
      "loss": 0.4092,
      "step": 13843
    },
    {
      "epoch": 0.7470724731530948,
      "grad_norm": 1.3630301926926642,
      "learning_rate": 2.347418630511767e-06,
      "loss": 0.5381,
      "step": 13844
    },
    {
      "epoch": 0.7471264367816092,
      "grad_norm": 0.8660056737015015,
      "learning_rate": 2.34687429066996e-06,
      "loss": 0.4007,
      "step": 13845
    },
    {
      "epoch": 0.7471804004101236,
      "grad_norm": 0.9374074367344646,
      "learning_rate": 2.3463300414521723e-06,
      "loss": 0.4047,
      "step": 13846
    },
    {
      "epoch": 0.747234364038638,
      "grad_norm": 1.2693323402838448,
      "learning_rate": 2.3457858828740456e-06,
      "loss": 0.5171,
      "step": 13847
    },
    {
      "epoch": 0.7472883276671524,
      "grad_norm": 1.0372571018608676,
      "learning_rate": 2.34524181495122e-06,
      "loss": 0.4351,
      "step": 13848
    },
    {
      "epoch": 0.7473422912956668,
      "grad_norm": 0.9907971465472057,
      "learning_rate": 2.344697837699332e-06,
      "loss": 0.4426,
      "step": 13849
    },
    {
      "epoch": 0.747396254924181,
      "grad_norm": 1.0399103208492853,
      "learning_rate": 2.3441539511340183e-06,
      "loss": 0.4998,
      "step": 13850
    },
    {
      "epoch": 0.7474502185526954,
      "grad_norm": 0.8835383897600424,
      "learning_rate": 2.3436101552709065e-06,
      "loss": 0.3218,
      "step": 13851
    },
    {
      "epoch": 0.7475041821812098,
      "grad_norm": 1.0836387793865145,
      "learning_rate": 2.3430664501256287e-06,
      "loss": 0.5031,
      "step": 13852
    },
    {
      "epoch": 0.7475581458097242,
      "grad_norm": 1.166993759618592,
      "learning_rate": 2.3425228357138107e-06,
      "loss": 0.3687,
      "step": 13853
    },
    {
      "epoch": 0.7476121094382386,
      "grad_norm": 1.1643009362728767,
      "learning_rate": 2.341979312051077e-06,
      "loss": 0.5458,
      "step": 13854
    },
    {
      "epoch": 0.747666073066753,
      "grad_norm": 0.9661020292141071,
      "learning_rate": 2.3414358791530487e-06,
      "loss": 0.3801,
      "step": 13855
    },
    {
      "epoch": 0.7477200366952674,
      "grad_norm": 0.7767369554945415,
      "learning_rate": 2.340892537035345e-06,
      "loss": 0.3052,
      "step": 13856
    },
    {
      "epoch": 0.7477740003237817,
      "grad_norm": 1.0478652922912999,
      "learning_rate": 2.340349285713582e-06,
      "loss": 0.4699,
      "step": 13857
    },
    {
      "epoch": 0.7478279639522961,
      "grad_norm": 0.8708834059660753,
      "learning_rate": 2.3398061252033723e-06,
      "loss": 0.3708,
      "step": 13858
    },
    {
      "epoch": 0.7478819275808105,
      "grad_norm": 1.1082664917936471,
      "learning_rate": 2.3392630555203282e-06,
      "loss": 0.5985,
      "step": 13859
    },
    {
      "epoch": 0.7479358912093249,
      "grad_norm": 0.9559940790709079,
      "learning_rate": 2.3387200766800575e-06,
      "loss": 0.5273,
      "step": 13860
    },
    {
      "epoch": 0.7479898548378393,
      "grad_norm": 0.8248772637152212,
      "learning_rate": 2.338177188698168e-06,
      "loss": 0.5071,
      "step": 13861
    },
    {
      "epoch": 0.7480438184663537,
      "grad_norm": 1.035049290168375,
      "learning_rate": 2.337634391590258e-06,
      "loss": 0.6254,
      "step": 13862
    },
    {
      "epoch": 0.7480977820948681,
      "grad_norm": 0.877827136765907,
      "learning_rate": 2.3370916853719324e-06,
      "loss": 0.3536,
      "step": 13863
    },
    {
      "epoch": 0.7481517457233824,
      "grad_norm": 0.9934921072485104,
      "learning_rate": 2.3365490700587874e-06,
      "loss": 0.5967,
      "step": 13864
    },
    {
      "epoch": 0.7482057093518968,
      "grad_norm": 1.0411642703379984,
      "learning_rate": 2.336006545666418e-06,
      "loss": 0.518,
      "step": 13865
    },
    {
      "epoch": 0.7482596729804112,
      "grad_norm": 0.9130278927698046,
      "learning_rate": 2.335464112210418e-06,
      "loss": 0.3416,
      "step": 13866
    },
    {
      "epoch": 0.7483136366089256,
      "grad_norm": 0.9027868798275885,
      "learning_rate": 2.3349217697063767e-06,
      "loss": 0.3194,
      "step": 13867
    },
    {
      "epoch": 0.74836760023744,
      "grad_norm": 1.077666019839338,
      "learning_rate": 2.334379518169882e-06,
      "loss": 0.5004,
      "step": 13868
    },
    {
      "epoch": 0.7484215638659544,
      "grad_norm": 1.0968449613076456,
      "learning_rate": 2.333837357616518e-06,
      "loss": 0.6516,
      "step": 13869
    },
    {
      "epoch": 0.7484755274944688,
      "grad_norm": 1.1021636979794451,
      "learning_rate": 2.3332952880618677e-06,
      "loss": 0.7228,
      "step": 13870
    },
    {
      "epoch": 0.7485294911229831,
      "grad_norm": 0.9240080669251083,
      "learning_rate": 2.332753309521512e-06,
      "loss": 0.4977,
      "step": 13871
    },
    {
      "epoch": 0.7485834547514975,
      "grad_norm": 1.0128425219380541,
      "learning_rate": 2.332211422011026e-06,
      "loss": 0.3663,
      "step": 13872
    },
    {
      "epoch": 0.7486374183800119,
      "grad_norm": 1.0126395479367398,
      "learning_rate": 2.3316696255459838e-06,
      "loss": 0.5,
      "step": 13873
    },
    {
      "epoch": 0.7486913820085263,
      "grad_norm": 1.1642802066840596,
      "learning_rate": 2.3311279201419577e-06,
      "loss": 0.6003,
      "step": 13874
    },
    {
      "epoch": 0.7487453456370406,
      "grad_norm": 1.0172385822036754,
      "learning_rate": 2.3305863058145174e-06,
      "loss": 0.3729,
      "step": 13875
    },
    {
      "epoch": 0.748799309265555,
      "grad_norm": 0.7739083816895926,
      "learning_rate": 2.330044782579229e-06,
      "loss": 0.2766,
      "step": 13876
    },
    {
      "epoch": 0.7488532728940694,
      "grad_norm": 0.9493346939129942,
      "learning_rate": 2.329503350451656e-06,
      "loss": 0.3901,
      "step": 13877
    },
    {
      "epoch": 0.7489072365225837,
      "grad_norm": 0.9022201748842198,
      "learning_rate": 2.3289620094473624e-06,
      "loss": 0.3594,
      "step": 13878
    },
    {
      "epoch": 0.7489612001510981,
      "grad_norm": 0.9434494011742915,
      "learning_rate": 2.3284207595819028e-06,
      "loss": 0.3685,
      "step": 13879
    },
    {
      "epoch": 0.7490151637796125,
      "grad_norm": 0.9806707446549446,
      "learning_rate": 2.3278796008708357e-06,
      "loss": 0.3912,
      "step": 13880
    },
    {
      "epoch": 0.7490691274081269,
      "grad_norm": 0.7843899307359191,
      "learning_rate": 2.327338533329714e-06,
      "loss": 0.2971,
      "step": 13881
    },
    {
      "epoch": 0.7491230910366413,
      "grad_norm": 1.2347059656236767,
      "learning_rate": 2.326797556974089e-06,
      "loss": 0.4525,
      "step": 13882
    },
    {
      "epoch": 0.7491770546651557,
      "grad_norm": 1.03413736900149,
      "learning_rate": 2.3262566718195075e-06,
      "loss": 0.5572,
      "step": 13883
    },
    {
      "epoch": 0.7492310182936701,
      "grad_norm": 0.8866682133445495,
      "learning_rate": 2.325715877881517e-06,
      "loss": 0.444,
      "step": 13884
    },
    {
      "epoch": 0.7492849819221844,
      "grad_norm": 1.0034744194851308,
      "learning_rate": 2.3251751751756603e-06,
      "loss": 0.5889,
      "step": 13885
    },
    {
      "epoch": 0.7493389455506988,
      "grad_norm": 1.014114210736665,
      "learning_rate": 2.324634563717476e-06,
      "loss": 0.4455,
      "step": 13886
    },
    {
      "epoch": 0.7493929091792132,
      "grad_norm": 1.2251327276293535,
      "learning_rate": 2.3240940435225024e-06,
      "loss": 0.6765,
      "step": 13887
    },
    {
      "epoch": 0.7494468728077276,
      "grad_norm": 1.0085081107874403,
      "learning_rate": 2.3235536146062754e-06,
      "loss": 0.4723,
      "step": 13888
    },
    {
      "epoch": 0.749500836436242,
      "grad_norm": 0.9809147178811293,
      "learning_rate": 2.3230132769843267e-06,
      "loss": 0.4686,
      "step": 13889
    },
    {
      "epoch": 0.7495548000647564,
      "grad_norm": 1.0375204586339257,
      "learning_rate": 2.322473030672187e-06,
      "loss": 0.4763,
      "step": 13890
    },
    {
      "epoch": 0.7496087636932708,
      "grad_norm": 0.9002632225183947,
      "learning_rate": 2.321932875685383e-06,
      "loss": 0.4673,
      "step": 13891
    },
    {
      "epoch": 0.7496627273217851,
      "grad_norm": 0.7864064003814552,
      "learning_rate": 2.321392812039441e-06,
      "loss": 0.3506,
      "step": 13892
    },
    {
      "epoch": 0.7497166909502995,
      "grad_norm": 1.063057447801405,
      "learning_rate": 2.3208528397498793e-06,
      "loss": 0.4692,
      "step": 13893
    },
    {
      "epoch": 0.7497706545788139,
      "grad_norm": 1.0867849989947265,
      "learning_rate": 2.3203129588322194e-06,
      "loss": 0.5268,
      "step": 13894
    },
    {
      "epoch": 0.7498246182073283,
      "grad_norm": 0.7389927052721725,
      "learning_rate": 2.319773169301978e-06,
      "loss": 0.3311,
      "step": 13895
    },
    {
      "epoch": 0.7498785818358427,
      "grad_norm": 1.0303906003743488,
      "learning_rate": 2.319233471174669e-06,
      "loss": 0.4148,
      "step": 13896
    },
    {
      "epoch": 0.7499325454643571,
      "grad_norm": 0.7769908444070815,
      "learning_rate": 2.3186938644658043e-06,
      "loss": 0.3539,
      "step": 13897
    },
    {
      "epoch": 0.7499865090928715,
      "grad_norm": 1.3711606945478032,
      "learning_rate": 2.318154349190892e-06,
      "loss": 0.6757,
      "step": 13898
    },
    {
      "epoch": 0.7500404727213857,
      "grad_norm": 1.2609983664713675,
      "learning_rate": 2.317614925365441e-06,
      "loss": 0.4842,
      "step": 13899
    },
    {
      "epoch": 0.7500944363499001,
      "grad_norm": 0.941628279472951,
      "learning_rate": 2.3170755930049497e-06,
      "loss": 0.4943,
      "step": 13900
    },
    {
      "epoch": 0.7501483999784145,
      "grad_norm": 0.8392853439945963,
      "learning_rate": 2.3165363521249233e-06,
      "loss": 0.3131,
      "step": 13901
    },
    {
      "epoch": 0.7502023636069289,
      "grad_norm": 1.059485293529084,
      "learning_rate": 2.3159972027408584e-06,
      "loss": 0.5012,
      "step": 13902
    },
    {
      "epoch": 0.7502563272354433,
      "grad_norm": 0.9626318778083127,
      "learning_rate": 2.315458144868251e-06,
      "loss": 0.5125,
      "step": 13903
    },
    {
      "epoch": 0.7503102908639577,
      "grad_norm": 0.9336132967557704,
      "learning_rate": 2.314919178522594e-06,
      "loss": 0.4478,
      "step": 13904
    },
    {
      "epoch": 0.7503642544924721,
      "grad_norm": 1.1196736041258337,
      "learning_rate": 2.3143803037193783e-06,
      "loss": 0.5109,
      "step": 13905
    },
    {
      "epoch": 0.7504182181209864,
      "grad_norm": 1.3531374178677031,
      "learning_rate": 2.313841520474093e-06,
      "loss": 0.4355,
      "step": 13906
    },
    {
      "epoch": 0.7504721817495008,
      "grad_norm": 1.060131367256263,
      "learning_rate": 2.3133028288022203e-06,
      "loss": 0.4916,
      "step": 13907
    },
    {
      "epoch": 0.7505261453780152,
      "grad_norm": 0.9513431821987536,
      "learning_rate": 2.3127642287192444e-06,
      "loss": 0.4265,
      "step": 13908
    },
    {
      "epoch": 0.7505801090065296,
      "grad_norm": 1.1389279373359416,
      "learning_rate": 2.3122257202406444e-06,
      "loss": 0.4818,
      "step": 13909
    },
    {
      "epoch": 0.750634072635044,
      "grad_norm": 0.9989919462580521,
      "learning_rate": 2.311687303381899e-06,
      "loss": 0.383,
      "step": 13910
    },
    {
      "epoch": 0.7506880362635584,
      "grad_norm": 1.079402584020198,
      "learning_rate": 2.311148978158482e-06,
      "loss": 0.4942,
      "step": 13911
    },
    {
      "epoch": 0.7507419998920728,
      "grad_norm": 1.1041052828361313,
      "learning_rate": 2.3106107445858647e-06,
      "loss": 0.449,
      "step": 13912
    },
    {
      "epoch": 0.7507959635205871,
      "grad_norm": 0.9842580119143084,
      "learning_rate": 2.3100726026795183e-06,
      "loss": 0.5134,
      "step": 13913
    },
    {
      "epoch": 0.7508499271491015,
      "grad_norm": 0.9450788564796143,
      "learning_rate": 2.3095345524549076e-06,
      "loss": 0.3966,
      "step": 13914
    },
    {
      "epoch": 0.7509038907776159,
      "grad_norm": 0.8458323762100993,
      "learning_rate": 2.308996593927498e-06,
      "loss": 0.2922,
      "step": 13915
    },
    {
      "epoch": 0.7509578544061303,
      "grad_norm": 0.9558156328933171,
      "learning_rate": 2.3084587271127506e-06,
      "loss": 0.409,
      "step": 13916
    },
    {
      "epoch": 0.7510118180346447,
      "grad_norm": 1.0482624850213207,
      "learning_rate": 2.307920952026124e-06,
      "loss": 0.5502,
      "step": 13917
    },
    {
      "epoch": 0.7510657816631591,
      "grad_norm": 1.0425250159021064,
      "learning_rate": 2.3073832686830745e-06,
      "loss": 0.392,
      "step": 13918
    },
    {
      "epoch": 0.7511197452916734,
      "grad_norm": 0.9988277249060221,
      "learning_rate": 2.3068456770990576e-06,
      "loss": 0.4381,
      "step": 13919
    },
    {
      "epoch": 0.7511737089201878,
      "grad_norm": 1.153643635431112,
      "learning_rate": 2.3063081772895203e-06,
      "loss": 0.4842,
      "step": 13920
    },
    {
      "epoch": 0.7512276725487022,
      "grad_norm": 0.9670721308238392,
      "learning_rate": 2.305770769269914e-06,
      "loss": 0.3651,
      "step": 13921
    },
    {
      "epoch": 0.7512816361772165,
      "grad_norm": 1.0155191184004237,
      "learning_rate": 2.3052334530556823e-06,
      "loss": 0.4204,
      "step": 13922
    },
    {
      "epoch": 0.751335599805731,
      "grad_norm": 1.053197898745572,
      "learning_rate": 2.3046962286622697e-06,
      "loss": 0.4481,
      "step": 13923
    },
    {
      "epoch": 0.7513895634342453,
      "grad_norm": 1.059930182915908,
      "learning_rate": 2.3041590961051165e-06,
      "loss": 0.4725,
      "step": 13924
    },
    {
      "epoch": 0.7514435270627597,
      "grad_norm": 1.0728715593452358,
      "learning_rate": 2.3036220553996593e-06,
      "loss": 0.6513,
      "step": 13925
    },
    {
      "epoch": 0.751497490691274,
      "grad_norm": 1.0641860683521378,
      "learning_rate": 2.303085106561336e-06,
      "loss": 0.472,
      "step": 13926
    },
    {
      "epoch": 0.7515514543197884,
      "grad_norm": 1.0284422588963145,
      "learning_rate": 2.302548249605575e-06,
      "loss": 0.4558,
      "step": 13927
    },
    {
      "epoch": 0.7516054179483028,
      "grad_norm": 0.9302938241294749,
      "learning_rate": 2.3020114845478087e-06,
      "loss": 0.4778,
      "step": 13928
    },
    {
      "epoch": 0.7516593815768172,
      "grad_norm": 0.8223284883197628,
      "learning_rate": 2.301474811403463e-06,
      "loss": 0.4057,
      "step": 13929
    },
    {
      "epoch": 0.7517133452053316,
      "grad_norm": 0.9859580312894238,
      "learning_rate": 2.3009382301879636e-06,
      "loss": 0.5278,
      "step": 13930
    },
    {
      "epoch": 0.751767308833846,
      "grad_norm": 0.9645528171438593,
      "learning_rate": 2.3004017409167315e-06,
      "loss": 0.4694,
      "step": 13931
    },
    {
      "epoch": 0.7518212724623604,
      "grad_norm": 0.9750089893922079,
      "learning_rate": 2.2998653436051864e-06,
      "loss": 0.4163,
      "step": 13932
    },
    {
      "epoch": 0.7518752360908747,
      "grad_norm": 1.0411375184947833,
      "learning_rate": 2.299329038268746e-06,
      "loss": 0.5314,
      "step": 13933
    },
    {
      "epoch": 0.7519291997193891,
      "grad_norm": 1.1383386471466754,
      "learning_rate": 2.2987928249228226e-06,
      "loss": 0.6169,
      "step": 13934
    },
    {
      "epoch": 0.7519831633479035,
      "grad_norm": 0.9546003616574823,
      "learning_rate": 2.298256703582827e-06,
      "loss": 0.4001,
      "step": 13935
    },
    {
      "epoch": 0.7520371269764179,
      "grad_norm": 0.8354379060924966,
      "learning_rate": 2.297720674264169e-06,
      "loss": 0.3944,
      "step": 13936
    },
    {
      "epoch": 0.7520910906049323,
      "grad_norm": 0.8688536682157534,
      "learning_rate": 2.297184736982255e-06,
      "loss": 0.4041,
      "step": 13937
    },
    {
      "epoch": 0.7521450542334467,
      "grad_norm": 0.9673040290306656,
      "learning_rate": 2.2966488917524877e-06,
      "loss": 0.4497,
      "step": 13938
    },
    {
      "epoch": 0.7521990178619611,
      "grad_norm": 1.046302118952138,
      "learning_rate": 2.296113138590268e-06,
      "loss": 0.4914,
      "step": 13939
    },
    {
      "epoch": 0.7522529814904754,
      "grad_norm": 0.8064320777360616,
      "learning_rate": 2.2955774775109955e-06,
      "loss": 0.3208,
      "step": 13940
    },
    {
      "epoch": 0.7523069451189898,
      "grad_norm": 1.049437797134903,
      "learning_rate": 2.2950419085300627e-06,
      "loss": 0.4554,
      "step": 13941
    },
    {
      "epoch": 0.7523609087475042,
      "grad_norm": 0.963044979443334,
      "learning_rate": 2.2945064316628634e-06,
      "loss": 0.4605,
      "step": 13942
    },
    {
      "epoch": 0.7524148723760186,
      "grad_norm": 0.9277369030402229,
      "learning_rate": 2.2939710469247886e-06,
      "loss": 0.3891,
      "step": 13943
    },
    {
      "epoch": 0.752468836004533,
      "grad_norm": 1.038255899529012,
      "learning_rate": 2.2934357543312253e-06,
      "loss": 0.4647,
      "step": 13944
    },
    {
      "epoch": 0.7525227996330474,
      "grad_norm": 0.8308779826793878,
      "learning_rate": 2.292900553897558e-06,
      "loss": 0.2524,
      "step": 13945
    },
    {
      "epoch": 0.7525767632615618,
      "grad_norm": 1.1901001166279568,
      "learning_rate": 2.2923654456391696e-06,
      "loss": 0.5329,
      "step": 13946
    },
    {
      "epoch": 0.752630726890076,
      "grad_norm": 0.9153584119867314,
      "learning_rate": 2.291830429571441e-06,
      "loss": 0.391,
      "step": 13947
    },
    {
      "epoch": 0.7526846905185904,
      "grad_norm": 1.0398886500903242,
      "learning_rate": 2.291295505709746e-06,
      "loss": 0.3989,
      "step": 13948
    },
    {
      "epoch": 0.7527386541471048,
      "grad_norm": 0.8813885605314811,
      "learning_rate": 2.29076067406946e-06,
      "loss": 0.3499,
      "step": 13949
    },
    {
      "epoch": 0.7527926177756192,
      "grad_norm": 1.0567013872588795,
      "learning_rate": 2.2902259346659555e-06,
      "loss": 0.3763,
      "step": 13950
    },
    {
      "epoch": 0.7528465814041336,
      "grad_norm": 1.0239319776396805,
      "learning_rate": 2.2896912875145997e-06,
      "loss": 0.4325,
      "step": 13951
    },
    {
      "epoch": 0.752900545032648,
      "grad_norm": 1.043169568219525,
      "learning_rate": 2.2891567326307615e-06,
      "loss": 0.6055,
      "step": 13952
    },
    {
      "epoch": 0.7529545086611624,
      "grad_norm": 1.315158268320146,
      "learning_rate": 2.288622270029802e-06,
      "loss": 0.6208,
      "step": 13953
    },
    {
      "epoch": 0.7530084722896767,
      "grad_norm": 0.8877865196588159,
      "learning_rate": 2.2880878997270855e-06,
      "loss": 0.356,
      "step": 13954
    },
    {
      "epoch": 0.7530624359181911,
      "grad_norm": 0.9634296170178792,
      "learning_rate": 2.2875536217379666e-06,
      "loss": 0.4345,
      "step": 13955
    },
    {
      "epoch": 0.7531163995467055,
      "grad_norm": 0.8824796326431276,
      "learning_rate": 2.2870194360778026e-06,
      "loss": 0.3775,
      "step": 13956
    },
    {
      "epoch": 0.7531703631752199,
      "grad_norm": 1.1490704949596644,
      "learning_rate": 2.2864853427619464e-06,
      "loss": 0.4787,
      "step": 13957
    },
    {
      "epoch": 0.7532243268037343,
      "grad_norm": 1.0472843810317451,
      "learning_rate": 2.285951341805748e-06,
      "loss": 0.4538,
      "step": 13958
    },
    {
      "epoch": 0.7532782904322487,
      "grad_norm": 1.1603455904862483,
      "learning_rate": 2.2854174332245563e-06,
      "loss": 0.5279,
      "step": 13959
    },
    {
      "epoch": 0.7533322540607631,
      "grad_norm": 0.9812901696258612,
      "learning_rate": 2.2848836170337154e-06,
      "loss": 0.5151,
      "step": 13960
    },
    {
      "epoch": 0.7533862176892774,
      "grad_norm": 1.0601332914699295,
      "learning_rate": 2.2843498932485677e-06,
      "loss": 0.473,
      "step": 13961
    },
    {
      "epoch": 0.7534401813177918,
      "grad_norm": 0.8731307163576292,
      "learning_rate": 2.2838162618844532e-06,
      "loss": 0.3811,
      "step": 13962
    },
    {
      "epoch": 0.7534941449463062,
      "grad_norm": 0.8598279330023306,
      "learning_rate": 2.2832827229567098e-06,
      "loss": 0.322,
      "step": 13963
    },
    {
      "epoch": 0.7535481085748206,
      "grad_norm": 0.9160434513497467,
      "learning_rate": 2.2827492764806707e-06,
      "loss": 0.417,
      "step": 13964
    },
    {
      "epoch": 0.753602072203335,
      "grad_norm": 1.2753003756815688,
      "learning_rate": 2.282215922471668e-06,
      "loss": 0.4938,
      "step": 13965
    },
    {
      "epoch": 0.7536560358318494,
      "grad_norm": 1.0413033390431994,
      "learning_rate": 2.281682660945031e-06,
      "loss": 0.4525,
      "step": 13966
    },
    {
      "epoch": 0.7537099994603638,
      "grad_norm": 1.055866639858768,
      "learning_rate": 2.2811494919160883e-06,
      "loss": 0.4635,
      "step": 13967
    },
    {
      "epoch": 0.753763963088878,
      "grad_norm": 1.0670306556065037,
      "learning_rate": 2.2806164154001597e-06,
      "loss": 0.4695,
      "step": 13968
    },
    {
      "epoch": 0.7538179267173925,
      "grad_norm": 0.8832360134891668,
      "learning_rate": 2.2800834314125687e-06,
      "loss": 0.404,
      "step": 13969
    },
    {
      "epoch": 0.7538718903459068,
      "grad_norm": 1.0154232844028097,
      "learning_rate": 2.279550539968633e-06,
      "loss": 0.5647,
      "step": 13970
    },
    {
      "epoch": 0.7539258539744212,
      "grad_norm": 1.06918318546569,
      "learning_rate": 2.2790177410836686e-06,
      "loss": 0.5077,
      "step": 13971
    },
    {
      "epoch": 0.7539798176029356,
      "grad_norm": 1.0394868545589164,
      "learning_rate": 2.278485034772989e-06,
      "loss": 0.4387,
      "step": 13972
    },
    {
      "epoch": 0.75403378123145,
      "grad_norm": 1.0190498775526138,
      "learning_rate": 2.2779524210519046e-06,
      "loss": 0.4276,
      "step": 13973
    },
    {
      "epoch": 0.7540877448599644,
      "grad_norm": 1.0477537681727538,
      "learning_rate": 2.277419899935724e-06,
      "loss": 0.4977,
      "step": 13974
    },
    {
      "epoch": 0.7541417084884787,
      "grad_norm": 1.038687709598905,
      "learning_rate": 2.276887471439752e-06,
      "loss": 0.4205,
      "step": 13975
    },
    {
      "epoch": 0.7541956721169931,
      "grad_norm": 0.8550754999882894,
      "learning_rate": 2.27635513557929e-06,
      "loss": 0.3732,
      "step": 13976
    },
    {
      "epoch": 0.7542496357455075,
      "grad_norm": 1.002579757208517,
      "learning_rate": 2.2758228923696383e-06,
      "loss": 0.3779,
      "step": 13977
    },
    {
      "epoch": 0.7543035993740219,
      "grad_norm": 0.8757684609903668,
      "learning_rate": 2.2752907418260945e-06,
      "loss": 0.3862,
      "step": 13978
    },
    {
      "epoch": 0.7543575630025363,
      "grad_norm": 1.039766692050184,
      "learning_rate": 2.2747586839639535e-06,
      "loss": 0.455,
      "step": 13979
    },
    {
      "epoch": 0.7544115266310507,
      "grad_norm": 1.269311961145948,
      "learning_rate": 2.2742267187985063e-06,
      "loss": 0.4942,
      "step": 13980
    },
    {
      "epoch": 0.7544654902595651,
      "grad_norm": 0.9116306124318664,
      "learning_rate": 2.2736948463450428e-06,
      "loss": 0.4422,
      "step": 13981
    },
    {
      "epoch": 0.7545194538880794,
      "grad_norm": 1.4209363205709982,
      "learning_rate": 2.2731630666188513e-06,
      "loss": 0.5185,
      "step": 13982
    },
    {
      "epoch": 0.7545734175165938,
      "grad_norm": 1.091474460441169,
      "learning_rate": 2.2726313796352113e-06,
      "loss": 0.3717,
      "step": 13983
    },
    {
      "epoch": 0.7546273811451082,
      "grad_norm": 0.9502580375185803,
      "learning_rate": 2.2720997854094072e-06,
      "loss": 0.5635,
      "step": 13984
    },
    {
      "epoch": 0.7546813447736226,
      "grad_norm": 1.097727131566793,
      "learning_rate": 2.2715682839567167e-06,
      "loss": 0.5103,
      "step": 13985
    },
    {
      "epoch": 0.754735308402137,
      "grad_norm": 0.87476582051389,
      "learning_rate": 2.2710368752924157e-06,
      "loss": 0.3137,
      "step": 13986
    },
    {
      "epoch": 0.7547892720306514,
      "grad_norm": 1.0824908030959848,
      "learning_rate": 2.2705055594317776e-06,
      "loss": 0.507,
      "step": 13987
    },
    {
      "epoch": 0.7548432356591657,
      "grad_norm": 1.2412557048661557,
      "learning_rate": 2.2699743363900743e-06,
      "loss": 0.62,
      "step": 13988
    },
    {
      "epoch": 0.7548971992876801,
      "grad_norm": 0.7619064252134992,
      "learning_rate": 2.269443206182571e-06,
      "loss": 0.2743,
      "step": 13989
    },
    {
      "epoch": 0.7549511629161945,
      "grad_norm": 1.0291794920616641,
      "learning_rate": 2.2689121688245337e-06,
      "loss": 0.4238,
      "step": 13990
    },
    {
      "epoch": 0.7550051265447089,
      "grad_norm": 0.7634420652607231,
      "learning_rate": 2.2683812243312257e-06,
      "loss": 0.2759,
      "step": 13991
    },
    {
      "epoch": 0.7550590901732233,
      "grad_norm": 0.888660662533418,
      "learning_rate": 2.267850372717907e-06,
      "loss": 0.3637,
      "step": 13992
    },
    {
      "epoch": 0.7551130538017377,
      "grad_norm": 1.0938418505343273,
      "learning_rate": 2.267319613999834e-06,
      "loss": 0.5777,
      "step": 13993
    },
    {
      "epoch": 0.755167017430252,
      "grad_norm": 1.281393687038122,
      "learning_rate": 2.266788948192262e-06,
      "loss": 0.5981,
      "step": 13994
    },
    {
      "epoch": 0.7552209810587663,
      "grad_norm": 0.9156048909418717,
      "learning_rate": 2.2662583753104436e-06,
      "loss": 0.397,
      "step": 13995
    },
    {
      "epoch": 0.7552749446872807,
      "grad_norm": 1.3000762897722957,
      "learning_rate": 2.265727895369626e-06,
      "loss": 0.4909,
      "step": 13996
    },
    {
      "epoch": 0.7553289083157951,
      "grad_norm": 0.8873700642481425,
      "learning_rate": 2.265197508385056e-06,
      "loss": 0.4425,
      "step": 13997
    },
    {
      "epoch": 0.7553828719443095,
      "grad_norm": 1.2342881044884315,
      "learning_rate": 2.2646672143719787e-06,
      "loss": 0.5763,
      "step": 13998
    },
    {
      "epoch": 0.7554368355728239,
      "grad_norm": 1.404536929865371,
      "learning_rate": 2.2641370133456347e-06,
      "loss": 0.6666,
      "step": 13999
    },
    {
      "epoch": 0.7554907992013383,
      "grad_norm": 1.068235239302593,
      "learning_rate": 2.2636069053212624e-06,
      "loss": 0.4564,
      "step": 14000
    },
    {
      "epoch": 0.7554907992013383,
      "eval_loss": 0.533542811870575,
      "eval_runtime": 164.5859,
      "eval_samples_per_second": 20.895,
      "eval_steps_per_second": 0.875,
      "step": 14000
    },
    {
      "epoch": 0.7555447628298527,
      "grad_norm": 0.9354936789278996,
      "learning_rate": 2.263076890314098e-06,
      "loss": 0.4387,
      "step": 14001
    },
    {
      "epoch": 0.755598726458367,
      "grad_norm": 0.9931809275498044,
      "learning_rate": 2.262546968339375e-06,
      "loss": 0.4557,
      "step": 14002
    },
    {
      "epoch": 0.7556526900868814,
      "grad_norm": 1.0022497036178668,
      "learning_rate": 2.262017139412323e-06,
      "loss": 0.4709,
      "step": 14003
    },
    {
      "epoch": 0.7557066537153958,
      "grad_norm": 1.0504671486400556,
      "learning_rate": 2.2614874035481697e-06,
      "loss": 0.5235,
      "step": 14004
    },
    {
      "epoch": 0.7557606173439102,
      "grad_norm": 0.903311764010699,
      "learning_rate": 2.2609577607621405e-06,
      "loss": 0.3745,
      "step": 14005
    },
    {
      "epoch": 0.7558145809724246,
      "grad_norm": 1.0634134743913592,
      "learning_rate": 2.2604282110694588e-06,
      "loss": 0.5042,
      "step": 14006
    },
    {
      "epoch": 0.755868544600939,
      "grad_norm": 1.09724279283852,
      "learning_rate": 2.2598987544853428e-06,
      "loss": 0.6339,
      "step": 14007
    },
    {
      "epoch": 0.7559225082294534,
      "grad_norm": 0.9340576035854213,
      "learning_rate": 2.2593693910250107e-06,
      "loss": 0.414,
      "step": 14008
    },
    {
      "epoch": 0.7559764718579677,
      "grad_norm": 1.1319018665907776,
      "learning_rate": 2.258840120703677e-06,
      "loss": 0.5375,
      "step": 14009
    },
    {
      "epoch": 0.7560304354864821,
      "grad_norm": 0.9371574397537727,
      "learning_rate": 2.2583109435365535e-06,
      "loss": 0.4329,
      "step": 14010
    },
    {
      "epoch": 0.7560843991149965,
      "grad_norm": 0.8465797703266068,
      "learning_rate": 2.257781859538849e-06,
      "loss": 0.3663,
      "step": 14011
    },
    {
      "epoch": 0.7561383627435109,
      "grad_norm": 0.9799546732909562,
      "learning_rate": 2.2572528687257697e-06,
      "loss": 0.3774,
      "step": 14012
    },
    {
      "epoch": 0.7561923263720253,
      "grad_norm": 1.0731783137962008,
      "learning_rate": 2.256723971112519e-06,
      "loss": 0.4569,
      "step": 14013
    },
    {
      "epoch": 0.7562462900005397,
      "grad_norm": 0.8695403505905999,
      "learning_rate": 2.256195166714299e-06,
      "loss": 0.3188,
      "step": 14014
    },
    {
      "epoch": 0.7563002536290541,
      "grad_norm": 0.943825112091173,
      "learning_rate": 2.255666455546307e-06,
      "loss": 0.4951,
      "step": 14015
    },
    {
      "epoch": 0.7563542172575684,
      "grad_norm": 1.2331045022486962,
      "learning_rate": 2.2551378376237395e-06,
      "loss": 0.6782,
      "step": 14016
    },
    {
      "epoch": 0.7564081808860827,
      "grad_norm": 1.4798032267275172,
      "learning_rate": 2.2546093129617893e-06,
      "loss": 0.5434,
      "step": 14017
    },
    {
      "epoch": 0.7564621445145971,
      "grad_norm": 0.9938028535199084,
      "learning_rate": 2.2540808815756467e-06,
      "loss": 0.3908,
      "step": 14018
    },
    {
      "epoch": 0.7565161081431115,
      "grad_norm": 1.0779670982988565,
      "learning_rate": 2.2535525434804987e-06,
      "loss": 0.4575,
      "step": 14019
    },
    {
      "epoch": 0.7565700717716259,
      "grad_norm": 0.9610341649051867,
      "learning_rate": 2.2530242986915314e-06,
      "loss": 0.4521,
      "step": 14020
    },
    {
      "epoch": 0.7566240354001403,
      "grad_norm": 1.0548144970262483,
      "learning_rate": 2.2524961472239264e-06,
      "loss": 0.5371,
      "step": 14021
    },
    {
      "epoch": 0.7566779990286547,
      "grad_norm": 1.0359540052835028,
      "learning_rate": 2.251968089092864e-06,
      "loss": 0.4842,
      "step": 14022
    },
    {
      "epoch": 0.756731962657169,
      "grad_norm": 1.1252138429991847,
      "learning_rate": 2.2514401243135216e-06,
      "loss": 0.5575,
      "step": 14023
    },
    {
      "epoch": 0.7567859262856834,
      "grad_norm": 1.0030169820521018,
      "learning_rate": 2.250912252901071e-06,
      "loss": 0.4537,
      "step": 14024
    },
    {
      "epoch": 0.7568398899141978,
      "grad_norm": 1.1314355028232312,
      "learning_rate": 2.2503844748706854e-06,
      "loss": 0.4044,
      "step": 14025
    },
    {
      "epoch": 0.7568938535427122,
      "grad_norm": 1.1767965295077552,
      "learning_rate": 2.2498567902375335e-06,
      "loss": 0.4921,
      "step": 14026
    },
    {
      "epoch": 0.7569478171712266,
      "grad_norm": 0.8726495729706913,
      "learning_rate": 2.2493291990167814e-06,
      "loss": 0.355,
      "step": 14027
    },
    {
      "epoch": 0.757001780799741,
      "grad_norm": 1.051174877906105,
      "learning_rate": 2.2488017012235922e-06,
      "loss": 0.5024,
      "step": 14028
    },
    {
      "epoch": 0.7570557444282554,
      "grad_norm": 1.2255619159247264,
      "learning_rate": 2.2482742968731276e-06,
      "loss": 0.6038,
      "step": 14029
    },
    {
      "epoch": 0.7571097080567697,
      "grad_norm": 1.0387737497554268,
      "learning_rate": 2.247746985980547e-06,
      "loss": 0.5265,
      "step": 14030
    },
    {
      "epoch": 0.7571636716852841,
      "grad_norm": 1.1051019382834246,
      "learning_rate": 2.2472197685610026e-06,
      "loss": 0.5472,
      "step": 14031
    },
    {
      "epoch": 0.7572176353137985,
      "grad_norm": 1.1298131798795752,
      "learning_rate": 2.2466926446296487e-06,
      "loss": 0.578,
      "step": 14032
    },
    {
      "epoch": 0.7572715989423129,
      "grad_norm": 0.8247354359424252,
      "learning_rate": 2.246165614201636e-06,
      "loss": 0.3487,
      "step": 14033
    },
    {
      "epoch": 0.7573255625708273,
      "grad_norm": 0.9147531285338606,
      "learning_rate": 2.245638677292111e-06,
      "loss": 0.4712,
      "step": 14034
    },
    {
      "epoch": 0.7573795261993417,
      "grad_norm": 0.9962392298538044,
      "learning_rate": 2.245111833916218e-06,
      "loss": 0.482,
      "step": 14035
    },
    {
      "epoch": 0.7574334898278561,
      "grad_norm": 1.2219595437358248,
      "learning_rate": 2.2445850840891e-06,
      "loss": 0.4479,
      "step": 14036
    },
    {
      "epoch": 0.7574874534563704,
      "grad_norm": 0.9829072754572074,
      "learning_rate": 2.244058427825898e-06,
      "loss": 0.5882,
      "step": 14037
    },
    {
      "epoch": 0.7575414170848848,
      "grad_norm": 1.0715133363125136,
      "learning_rate": 2.2435318651417443e-06,
      "loss": 0.3446,
      "step": 14038
    },
    {
      "epoch": 0.7575953807133992,
      "grad_norm": 1.0532794651427677,
      "learning_rate": 2.243005396051776e-06,
      "loss": 0.3683,
      "step": 14039
    },
    {
      "epoch": 0.7576493443419136,
      "grad_norm": 1.2657654250446775,
      "learning_rate": 2.2424790205711232e-06,
      "loss": 0.6074,
      "step": 14040
    },
    {
      "epoch": 0.757703307970428,
      "grad_norm": 0.8303571842159004,
      "learning_rate": 2.2419527387149155e-06,
      "loss": 0.3366,
      "step": 14041
    },
    {
      "epoch": 0.7577572715989424,
      "grad_norm": 1.1109347240107132,
      "learning_rate": 2.2414265504982775e-06,
      "loss": 0.4992,
      "step": 14042
    },
    {
      "epoch": 0.7578112352274567,
      "grad_norm": 0.8947093340476518,
      "learning_rate": 2.2409004559363336e-06,
      "loss": 0.3795,
      "step": 14043
    },
    {
      "epoch": 0.757865198855971,
      "grad_norm": 1.2874307539552958,
      "learning_rate": 2.240374455044204e-06,
      "loss": 0.5297,
      "step": 14044
    },
    {
      "epoch": 0.7579191624844854,
      "grad_norm": 1.1151955526239983,
      "learning_rate": 2.239848547837005e-06,
      "loss": 0.4746,
      "step": 14045
    },
    {
      "epoch": 0.7579731261129998,
      "grad_norm": 0.9757719585476565,
      "learning_rate": 2.239322734329854e-06,
      "loss": 0.4732,
      "step": 14046
    },
    {
      "epoch": 0.7580270897415142,
      "grad_norm": 0.960906462502796,
      "learning_rate": 2.2387970145378613e-06,
      "loss": 0.4091,
      "step": 14047
    },
    {
      "epoch": 0.7580810533700286,
      "grad_norm": 0.8644604168307908,
      "learning_rate": 2.238271388476138e-06,
      "loss": 0.3521,
      "step": 14048
    },
    {
      "epoch": 0.758135016998543,
      "grad_norm": 0.8754311658601478,
      "learning_rate": 2.2377458561597903e-06,
      "loss": 0.3968,
      "step": 14049
    },
    {
      "epoch": 0.7581889806270574,
      "grad_norm": 1.022753319384135,
      "learning_rate": 2.2372204176039237e-06,
      "loss": 0.4365,
      "step": 14050
    },
    {
      "epoch": 0.7582429442555717,
      "grad_norm": 0.7037696112676267,
      "learning_rate": 2.2366950728236396e-06,
      "loss": 0.2856,
      "step": 14051
    },
    {
      "epoch": 0.7582969078840861,
      "grad_norm": 1.1603634650448553,
      "learning_rate": 2.2361698218340356e-06,
      "loss": 0.6064,
      "step": 14052
    },
    {
      "epoch": 0.7583508715126005,
      "grad_norm": 0.8946276753020095,
      "learning_rate": 2.235644664650209e-06,
      "loss": 0.4482,
      "step": 14053
    },
    {
      "epoch": 0.7584048351411149,
      "grad_norm": 1.0900952725093074,
      "learning_rate": 2.2351196012872524e-06,
      "loss": 0.4911,
      "step": 14054
    },
    {
      "epoch": 0.7584587987696293,
      "grad_norm": 1.0434579186318522,
      "learning_rate": 2.2345946317602577e-06,
      "loss": 0.5979,
      "step": 14055
    },
    {
      "epoch": 0.7585127623981437,
      "grad_norm": 1.2318179300551424,
      "learning_rate": 2.2340697560843127e-06,
      "loss": 0.5244,
      "step": 14056
    },
    {
      "epoch": 0.758566726026658,
      "grad_norm": 0.9363785292792274,
      "learning_rate": 2.233544974274504e-06,
      "loss": 0.432,
      "step": 14057
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 1.0351200631672817,
      "learning_rate": 2.2330202863459123e-06,
      "loss": 0.4116,
      "step": 14058
    },
    {
      "epoch": 0.7586746532836868,
      "grad_norm": 1.0772944519371075,
      "learning_rate": 2.232495692313619e-06,
      "loss": 0.4812,
      "step": 14059
    },
    {
      "epoch": 0.7587286169122012,
      "grad_norm": 0.9345690297230624,
      "learning_rate": 2.2319711921927005e-06,
      "loss": 0.3545,
      "step": 14060
    },
    {
      "epoch": 0.7587825805407156,
      "grad_norm": 1.0437092098766032,
      "learning_rate": 2.231446785998232e-06,
      "loss": 0.6237,
      "step": 14061
    },
    {
      "epoch": 0.75883654416923,
      "grad_norm": 1.058723526718693,
      "learning_rate": 2.230922473745286e-06,
      "loss": 0.4297,
      "step": 14062
    },
    {
      "epoch": 0.7588905077977444,
      "grad_norm": 1.0184821947065592,
      "learning_rate": 2.230398255448931e-06,
      "loss": 0.5125,
      "step": 14063
    },
    {
      "epoch": 0.7589444714262586,
      "grad_norm": 1.1870944654756497,
      "learning_rate": 2.2298741311242334e-06,
      "loss": 0.5936,
      "step": 14064
    },
    {
      "epoch": 0.758998435054773,
      "grad_norm": 1.0288548404742364,
      "learning_rate": 2.2293501007862584e-06,
      "loss": 0.459,
      "step": 14065
    },
    {
      "epoch": 0.7590523986832874,
      "grad_norm": 1.0000619082019335,
      "learning_rate": 2.2288261644500652e-06,
      "loss": 0.4297,
      "step": 14066
    },
    {
      "epoch": 0.7591063623118018,
      "grad_norm": 0.8352114368200882,
      "learning_rate": 2.2283023221307138e-06,
      "loss": 0.3187,
      "step": 14067
    },
    {
      "epoch": 0.7591603259403162,
      "grad_norm": 1.0119472095432958,
      "learning_rate": 2.2277785738432596e-06,
      "loss": 0.4523,
      "step": 14068
    },
    {
      "epoch": 0.7592142895688306,
      "grad_norm": 0.9683223852011174,
      "learning_rate": 2.2272549196027553e-06,
      "loss": 0.4315,
      "step": 14069
    },
    {
      "epoch": 0.759268253197345,
      "grad_norm": 0.9026347161140674,
      "learning_rate": 2.226731359424252e-06,
      "loss": 0.3789,
      "step": 14070
    },
    {
      "epoch": 0.7593222168258593,
      "grad_norm": 1.008241487806419,
      "learning_rate": 2.2262078933227976e-06,
      "loss": 0.4114,
      "step": 14071
    },
    {
      "epoch": 0.7593761804543737,
      "grad_norm": 0.9643104761945684,
      "learning_rate": 2.225684521313435e-06,
      "loss": 0.3934,
      "step": 14072
    },
    {
      "epoch": 0.7594301440828881,
      "grad_norm": 0.9044016698067384,
      "learning_rate": 2.225161243411208e-06,
      "loss": 0.4271,
      "step": 14073
    },
    {
      "epoch": 0.7594841077114025,
      "grad_norm": 1.1884586731218172,
      "learning_rate": 2.2246380596311555e-06,
      "loss": 0.598,
      "step": 14074
    },
    {
      "epoch": 0.7595380713399169,
      "grad_norm": 0.9864132670642436,
      "learning_rate": 2.224114969988315e-06,
      "loss": 0.463,
      "step": 14075
    },
    {
      "epoch": 0.7595920349684313,
      "grad_norm": 1.3309934198088655,
      "learning_rate": 2.22359197449772e-06,
      "loss": 0.6471,
      "step": 14076
    },
    {
      "epoch": 0.7596459985969457,
      "grad_norm": 0.7669020242829246,
      "learning_rate": 2.2230690731744027e-06,
      "loss": 0.3698,
      "step": 14077
    },
    {
      "epoch": 0.75969996222546,
      "grad_norm": 0.9011193220538629,
      "learning_rate": 2.222546266033392e-06,
      "loss": 0.4986,
      "step": 14078
    },
    {
      "epoch": 0.7597539258539744,
      "grad_norm": 0.8253816512140858,
      "learning_rate": 2.222023553089712e-06,
      "loss": 0.3449,
      "step": 14079
    },
    {
      "epoch": 0.7598078894824888,
      "grad_norm": 0.9593549382720731,
      "learning_rate": 2.2215009343583877e-06,
      "loss": 0.6023,
      "step": 14080
    },
    {
      "epoch": 0.7598618531110032,
      "grad_norm": 0.9206240189425108,
      "learning_rate": 2.220978409854439e-06,
      "loss": 0.4004,
      "step": 14081
    },
    {
      "epoch": 0.7599158167395176,
      "grad_norm": 1.034067088604144,
      "learning_rate": 2.220455979592884e-06,
      "loss": 0.5054,
      "step": 14082
    },
    {
      "epoch": 0.759969780368032,
      "grad_norm": 1.0525801116777458,
      "learning_rate": 2.219933643588738e-06,
      "loss": 0.4722,
      "step": 14083
    },
    {
      "epoch": 0.7600237439965464,
      "grad_norm": 1.0905118359563968,
      "learning_rate": 2.2194114018570136e-06,
      "loss": 0.4881,
      "step": 14084
    },
    {
      "epoch": 0.7600777076250607,
      "grad_norm": 0.9572745724599465,
      "learning_rate": 2.2188892544127213e-06,
      "loss": 0.3695,
      "step": 14085
    },
    {
      "epoch": 0.7601316712535751,
      "grad_norm": 1.1559502840868627,
      "learning_rate": 2.2183672012708656e-06,
      "loss": 0.4817,
      "step": 14086
    },
    {
      "epoch": 0.7601856348820895,
      "grad_norm": 1.0847997119539958,
      "learning_rate": 2.2178452424464526e-06,
      "loss": 0.5115,
      "step": 14087
    },
    {
      "epoch": 0.7602395985106039,
      "grad_norm": 0.9553736328831977,
      "learning_rate": 2.2173233779544832e-06,
      "loss": 0.5437,
      "step": 14088
    },
    {
      "epoch": 0.7602935621391183,
      "grad_norm": 1.2411575101739827,
      "learning_rate": 2.2168016078099576e-06,
      "loss": 0.6462,
      "step": 14089
    },
    {
      "epoch": 0.7603475257676326,
      "grad_norm": 1.1008434186157212,
      "learning_rate": 2.2162799320278707e-06,
      "loss": 0.4988,
      "step": 14090
    },
    {
      "epoch": 0.760401489396147,
      "grad_norm": 0.9544268175890872,
      "learning_rate": 2.2157583506232167e-06,
      "loss": 0.3743,
      "step": 14091
    },
    {
      "epoch": 0.7604554530246613,
      "grad_norm": 1.0623833826343994,
      "learning_rate": 2.2152368636109877e-06,
      "loss": 0.5235,
      "step": 14092
    },
    {
      "epoch": 0.7605094166531757,
      "grad_norm": 1.095349650283702,
      "learning_rate": 2.2147154710061686e-06,
      "loss": 0.4706,
      "step": 14093
    },
    {
      "epoch": 0.7605633802816901,
      "grad_norm": 1.079017033617189,
      "learning_rate": 2.2141941728237467e-06,
      "loss": 0.4716,
      "step": 14094
    },
    {
      "epoch": 0.7606173439102045,
      "grad_norm": 1.0994110985912298,
      "learning_rate": 2.213672969078704e-06,
      "loss": 0.4631,
      "step": 14095
    },
    {
      "epoch": 0.7606713075387189,
      "grad_norm": 0.9312769563651165,
      "learning_rate": 2.2131518597860214e-06,
      "loss": 0.3883,
      "step": 14096
    },
    {
      "epoch": 0.7607252711672333,
      "grad_norm": 1.050316798886754,
      "learning_rate": 2.212630844960675e-06,
      "loss": 0.4083,
      "step": 14097
    },
    {
      "epoch": 0.7607792347957477,
      "grad_norm": 1.1134487024014956,
      "learning_rate": 2.21210992461764e-06,
      "loss": 0.4578,
      "step": 14098
    },
    {
      "epoch": 0.760833198424262,
      "grad_norm": 1.0960197421337934,
      "learning_rate": 2.211589098771889e-06,
      "loss": 0.5087,
      "step": 14099
    },
    {
      "epoch": 0.7608871620527764,
      "grad_norm": 0.8576646686766642,
      "learning_rate": 2.2110683674383887e-06,
      "loss": 0.3286,
      "step": 14100
    },
    {
      "epoch": 0.7609411256812908,
      "grad_norm": 0.8975381305868854,
      "learning_rate": 2.2105477306321068e-06,
      "loss": 0.4361,
      "step": 14101
    },
    {
      "epoch": 0.7609950893098052,
      "grad_norm": 0.9608841246615539,
      "learning_rate": 2.210027188368006e-06,
      "loss": 0.3888,
      "step": 14102
    },
    {
      "epoch": 0.7610490529383196,
      "grad_norm": 1.0148271247897842,
      "learning_rate": 2.2095067406610496e-06,
      "loss": 0.5752,
      "step": 14103
    },
    {
      "epoch": 0.761103016566834,
      "grad_norm": 0.7432483846045046,
      "learning_rate": 2.2089863875261936e-06,
      "loss": 0.266,
      "step": 14104
    },
    {
      "epoch": 0.7611569801953484,
      "grad_norm": 1.3547244580104505,
      "learning_rate": 2.2084661289783943e-06,
      "loss": 0.4121,
      "step": 14105
    },
    {
      "epoch": 0.7612109438238627,
      "grad_norm": 0.9899404515292248,
      "learning_rate": 2.2079459650326055e-06,
      "loss": 0.5122,
      "step": 14106
    },
    {
      "epoch": 0.7612649074523771,
      "grad_norm": 1.0719810476322928,
      "learning_rate": 2.207425895703774e-06,
      "loss": 0.5035,
      "step": 14107
    },
    {
      "epoch": 0.7613188710808915,
      "grad_norm": 1.2057477895575381,
      "learning_rate": 2.2069059210068506e-06,
      "loss": 0.5813,
      "step": 14108
    },
    {
      "epoch": 0.7613728347094059,
      "grad_norm": 1.3077407342789313,
      "learning_rate": 2.206386040956778e-06,
      "loss": 0.5667,
      "step": 14109
    },
    {
      "epoch": 0.7614267983379203,
      "grad_norm": 1.0891011486178523,
      "learning_rate": 2.2058662555684976e-06,
      "loss": 0.4929,
      "step": 14110
    },
    {
      "epoch": 0.7614807619664347,
      "grad_norm": 0.876300571343487,
      "learning_rate": 2.20534656485695e-06,
      "loss": 0.4158,
      "step": 14111
    },
    {
      "epoch": 0.7615347255949491,
      "grad_norm": 1.2609000277855016,
      "learning_rate": 2.2048269688370715e-06,
      "loss": 0.5348,
      "step": 14112
    },
    {
      "epoch": 0.7615886892234633,
      "grad_norm": 1.255577216504524,
      "learning_rate": 2.2043074675237953e-06,
      "loss": 0.5882,
      "step": 14113
    },
    {
      "epoch": 0.7616426528519777,
      "grad_norm": 0.9810583477098861,
      "learning_rate": 2.203788060932053e-06,
      "loss": 0.4572,
      "step": 14114
    },
    {
      "epoch": 0.7616966164804921,
      "grad_norm": 1.015088715850209,
      "learning_rate": 2.2032687490767714e-06,
      "loss": 0.5748,
      "step": 14115
    },
    {
      "epoch": 0.7617505801090065,
      "grad_norm": 1.1294025485613624,
      "learning_rate": 2.2027495319728774e-06,
      "loss": 0.4734,
      "step": 14116
    },
    {
      "epoch": 0.7618045437375209,
      "grad_norm": 0.9008141026401497,
      "learning_rate": 2.202230409635295e-06,
      "loss": 0.398,
      "step": 14117
    },
    {
      "epoch": 0.7618585073660353,
      "grad_norm": 0.9815570918188072,
      "learning_rate": 2.2017113820789414e-06,
      "loss": 0.5603,
      "step": 14118
    },
    {
      "epoch": 0.7619124709945497,
      "grad_norm": 1.0641470744699095,
      "learning_rate": 2.2011924493187353e-06,
      "loss": 0.5121,
      "step": 14119
    },
    {
      "epoch": 0.761966434623064,
      "grad_norm": 0.8850332303789288,
      "learning_rate": 2.2006736113695914e-06,
      "loss": 0.4148,
      "step": 14120
    },
    {
      "epoch": 0.7620203982515784,
      "grad_norm": 1.0720982611199925,
      "learning_rate": 2.2001548682464213e-06,
      "loss": 0.5362,
      "step": 14121
    },
    {
      "epoch": 0.7620743618800928,
      "grad_norm": 1.1485372621498815,
      "learning_rate": 2.199636219964135e-06,
      "loss": 0.5404,
      "step": 14122
    },
    {
      "epoch": 0.7621283255086072,
      "grad_norm": 0.9201928139685264,
      "learning_rate": 2.199117666537639e-06,
      "loss": 0.3676,
      "step": 14123
    },
    {
      "epoch": 0.7621822891371216,
      "grad_norm": 0.7476943928426587,
      "learning_rate": 2.198599207981836e-06,
      "loss": 0.3376,
      "step": 14124
    },
    {
      "epoch": 0.762236252765636,
      "grad_norm": 0.9774180712911992,
      "learning_rate": 2.1980808443116274e-06,
      "loss": 0.452,
      "step": 14125
    },
    {
      "epoch": 0.7622902163941503,
      "grad_norm": 1.0079382991202668,
      "learning_rate": 2.1975625755419133e-06,
      "loss": 0.4358,
      "step": 14126
    },
    {
      "epoch": 0.7623441800226647,
      "grad_norm": 0.9349152087273793,
      "learning_rate": 2.1970444016875857e-06,
      "loss": 0.4344,
      "step": 14127
    },
    {
      "epoch": 0.7623981436511791,
      "grad_norm": 1.169278198242889,
      "learning_rate": 2.19652632276354e-06,
      "loss": 0.6267,
      "step": 14128
    },
    {
      "epoch": 0.7624521072796935,
      "grad_norm": 1.1551902957713174,
      "learning_rate": 2.196008338784666e-06,
      "loss": 0.6579,
      "step": 14129
    },
    {
      "epoch": 0.7625060709082079,
      "grad_norm": 1.2287926249260928,
      "learning_rate": 2.19549044976585e-06,
      "loss": 0.6022,
      "step": 14130
    },
    {
      "epoch": 0.7625600345367223,
      "grad_norm": 1.1457493789188624,
      "learning_rate": 2.194972655721978e-06,
      "loss": 0.6283,
      "step": 14131
    },
    {
      "epoch": 0.7626139981652367,
      "grad_norm": 1.226954303229493,
      "learning_rate": 2.194454956667931e-06,
      "loss": 0.5536,
      "step": 14132
    },
    {
      "epoch": 0.762667961793751,
      "grad_norm": 0.9980505997504556,
      "learning_rate": 2.19393735261859e-06,
      "loss": 0.3807,
      "step": 14133
    },
    {
      "epoch": 0.7627219254222654,
      "grad_norm": 1.1613050201561097,
      "learning_rate": 2.1934198435888286e-06,
      "loss": 0.635,
      "step": 14134
    },
    {
      "epoch": 0.7627758890507798,
      "grad_norm": 0.719146258817388,
      "learning_rate": 2.1929024295935224e-06,
      "loss": 0.3052,
      "step": 14135
    },
    {
      "epoch": 0.7628298526792942,
      "grad_norm": 1.1269172232683708,
      "learning_rate": 2.192385110647542e-06,
      "loss": 0.5619,
      "step": 14136
    },
    {
      "epoch": 0.7628838163078085,
      "grad_norm": 1.358585481082049,
      "learning_rate": 2.191867886765755e-06,
      "loss": 0.6047,
      "step": 14137
    },
    {
      "epoch": 0.762937779936323,
      "grad_norm": 0.9031446240978861,
      "learning_rate": 2.1913507579630285e-06,
      "loss": 0.3885,
      "step": 14138
    },
    {
      "epoch": 0.7629917435648373,
      "grad_norm": 1.35876267265125,
      "learning_rate": 2.1908337242542227e-06,
      "loss": 0.6019,
      "step": 14139
    },
    {
      "epoch": 0.7630457071933516,
      "grad_norm": 1.139517603283508,
      "learning_rate": 2.190316785654202e-06,
      "loss": 0.5293,
      "step": 14140
    },
    {
      "epoch": 0.763099670821866,
      "grad_norm": 0.8268150960788346,
      "learning_rate": 2.1897999421778195e-06,
      "loss": 0.3497,
      "step": 14141
    },
    {
      "epoch": 0.7631536344503804,
      "grad_norm": 1.1243207609895531,
      "learning_rate": 2.1892831938399313e-06,
      "loss": 0.5638,
      "step": 14142
    },
    {
      "epoch": 0.7632075980788948,
      "grad_norm": 1.0337416102456323,
      "learning_rate": 2.1887665406553895e-06,
      "loss": 0.5032,
      "step": 14143
    },
    {
      "epoch": 0.7632615617074092,
      "grad_norm": 1.1711393254163078,
      "learning_rate": 2.188249982639043e-06,
      "loss": 0.5308,
      "step": 14144
    },
    {
      "epoch": 0.7633155253359236,
      "grad_norm": 1.0842393658176575,
      "learning_rate": 2.187733519805738e-06,
      "loss": 0.4788,
      "step": 14145
    },
    {
      "epoch": 0.763369488964438,
      "grad_norm": 1.1263597740909923,
      "learning_rate": 2.187217152170319e-06,
      "loss": 0.4335,
      "step": 14146
    },
    {
      "epoch": 0.7634234525929523,
      "grad_norm": 1.1549469661226268,
      "learning_rate": 2.1867008797476265e-06,
      "loss": 0.4775,
      "step": 14147
    },
    {
      "epoch": 0.7634774162214667,
      "grad_norm": 1.3166143354409474,
      "learning_rate": 2.1861847025524985e-06,
      "loss": 0.6455,
      "step": 14148
    },
    {
      "epoch": 0.7635313798499811,
      "grad_norm": 1.0538583863379278,
      "learning_rate": 2.1856686205997702e-06,
      "loss": 0.4138,
      "step": 14149
    },
    {
      "epoch": 0.7635853434784955,
      "grad_norm": 1.0515666686156713,
      "learning_rate": 2.1851526339042745e-06,
      "loss": 0.5078,
      "step": 14150
    },
    {
      "epoch": 0.7636393071070099,
      "grad_norm": 0.9308575701809098,
      "learning_rate": 2.1846367424808414e-06,
      "loss": 0.4475,
      "step": 14151
    },
    {
      "epoch": 0.7636932707355243,
      "grad_norm": 0.8918375461148187,
      "learning_rate": 2.1841209463442985e-06,
      "loss": 0.2741,
      "step": 14152
    },
    {
      "epoch": 0.7637472343640387,
      "grad_norm": 1.3268228984779902,
      "learning_rate": 2.1836052455094696e-06,
      "loss": 0.6183,
      "step": 14153
    },
    {
      "epoch": 0.763801197992553,
      "grad_norm": 1.0515211657722692,
      "learning_rate": 2.183089639991178e-06,
      "loss": 0.426,
      "step": 14154
    },
    {
      "epoch": 0.7638551616210674,
      "grad_norm": 0.88292273492877,
      "learning_rate": 2.1825741298042414e-06,
      "loss": 0.4358,
      "step": 14155
    },
    {
      "epoch": 0.7639091252495818,
      "grad_norm": 1.0911168238694267,
      "learning_rate": 2.182058714963476e-06,
      "loss": 0.4283,
      "step": 14156
    },
    {
      "epoch": 0.7639630888780962,
      "grad_norm": 0.8911752224151559,
      "learning_rate": 2.181543395483695e-06,
      "loss": 0.399,
      "step": 14157
    },
    {
      "epoch": 0.7640170525066106,
      "grad_norm": 0.7167491549474045,
      "learning_rate": 2.1810281713797107e-06,
      "loss": 0.2991,
      "step": 14158
    },
    {
      "epoch": 0.764071016135125,
      "grad_norm": 1.0635658019508292,
      "learning_rate": 2.1805130426663297e-06,
      "loss": 0.4326,
      "step": 14159
    },
    {
      "epoch": 0.7641249797636394,
      "grad_norm": 1.0288199523801118,
      "learning_rate": 2.1799980093583582e-06,
      "loss": 0.4989,
      "step": 14160
    },
    {
      "epoch": 0.7641789433921536,
      "grad_norm": 0.9558932197269022,
      "learning_rate": 2.1794830714706e-06,
      "loss": 0.4825,
      "step": 14161
    },
    {
      "epoch": 0.764232907020668,
      "grad_norm": 0.9454661459190418,
      "learning_rate": 2.178968229017852e-06,
      "loss": 0.426,
      "step": 14162
    },
    {
      "epoch": 0.7642868706491824,
      "grad_norm": 1.2708702160228194,
      "learning_rate": 2.178453482014913e-06,
      "loss": 0.5117,
      "step": 14163
    },
    {
      "epoch": 0.7643408342776968,
      "grad_norm": 0.9654540568971489,
      "learning_rate": 2.177938830476577e-06,
      "loss": 0.4161,
      "step": 14164
    },
    {
      "epoch": 0.7643947979062112,
      "grad_norm": 1.1887865943075255,
      "learning_rate": 2.1774242744176354e-06,
      "loss": 0.527,
      "step": 14165
    },
    {
      "epoch": 0.7644487615347256,
      "grad_norm": 0.8727706057012192,
      "learning_rate": 2.1769098138528772e-06,
      "loss": 0.3616,
      "step": 14166
    },
    {
      "epoch": 0.76450272516324,
      "grad_norm": 0.8895798914935643,
      "learning_rate": 2.1763954487970894e-06,
      "loss": 0.3446,
      "step": 14167
    },
    {
      "epoch": 0.7645566887917543,
      "grad_norm": 0.989852874801464,
      "learning_rate": 2.175881179265054e-06,
      "loss": 0.4102,
      "step": 14168
    },
    {
      "epoch": 0.7646106524202687,
      "grad_norm": 1.064101463955328,
      "learning_rate": 2.175367005271553e-06,
      "loss": 0.6475,
      "step": 14169
    },
    {
      "epoch": 0.7646646160487831,
      "grad_norm": 0.9220361048747693,
      "learning_rate": 2.1748529268313627e-06,
      "loss": 0.4591,
      "step": 14170
    },
    {
      "epoch": 0.7647185796772975,
      "grad_norm": 1.2082277028305115,
      "learning_rate": 2.1743389439592593e-06,
      "loss": 0.5257,
      "step": 14171
    },
    {
      "epoch": 0.7647725433058119,
      "grad_norm": 0.9730970180234755,
      "learning_rate": 2.1738250566700154e-06,
      "loss": 0.5218,
      "step": 14172
    },
    {
      "epoch": 0.7648265069343263,
      "grad_norm": 1.1175320014981787,
      "learning_rate": 2.1733112649784e-06,
      "loss": 0.5523,
      "step": 14173
    },
    {
      "epoch": 0.7648804705628407,
      "grad_norm": 1.141492548903176,
      "learning_rate": 2.1727975688991804e-06,
      "loss": 0.5721,
      "step": 14174
    },
    {
      "epoch": 0.764934434191355,
      "grad_norm": 1.061004409578363,
      "learning_rate": 2.172283968447122e-06,
      "loss": 0.4685,
      "step": 14175
    },
    {
      "epoch": 0.7649883978198694,
      "grad_norm": 1.0558149381490756,
      "learning_rate": 2.1717704636369826e-06,
      "loss": 0.5687,
      "step": 14176
    },
    {
      "epoch": 0.7650423614483838,
      "grad_norm": 0.9978846676182903,
      "learning_rate": 2.171257054483524e-06,
      "loss": 0.3818,
      "step": 14177
    },
    {
      "epoch": 0.7650963250768982,
      "grad_norm": 1.075598428910838,
      "learning_rate": 2.170743741001501e-06,
      "loss": 0.4331,
      "step": 14178
    },
    {
      "epoch": 0.7651502887054126,
      "grad_norm": 1.0637113607872113,
      "learning_rate": 2.1702305232056662e-06,
      "loss": 0.4469,
      "step": 14179
    },
    {
      "epoch": 0.765204252333927,
      "grad_norm": 1.0536867574106419,
      "learning_rate": 2.1697174011107712e-06,
      "loss": 0.4384,
      "step": 14180
    },
    {
      "epoch": 0.7652582159624414,
      "grad_norm": 1.0466022976721183,
      "learning_rate": 2.1692043747315628e-06,
      "loss": 0.4576,
      "step": 14181
    },
    {
      "epoch": 0.7653121795909557,
      "grad_norm": 0.975279154385196,
      "learning_rate": 2.1686914440827877e-06,
      "loss": 0.3916,
      "step": 14182
    },
    {
      "epoch": 0.76536614321947,
      "grad_norm": 1.050325752292318,
      "learning_rate": 2.168178609179185e-06,
      "loss": 0.4562,
      "step": 14183
    },
    {
      "epoch": 0.7654201068479844,
      "grad_norm": 1.3308253921321456,
      "learning_rate": 2.167665870035496e-06,
      "loss": 0.5308,
      "step": 14184
    },
    {
      "epoch": 0.7654740704764988,
      "grad_norm": 1.1650281575672303,
      "learning_rate": 2.1671532266664562e-06,
      "loss": 0.6941,
      "step": 14185
    },
    {
      "epoch": 0.7655280341050132,
      "grad_norm": 1.0050826948381852,
      "learning_rate": 2.166640679086801e-06,
      "loss": 0.3894,
      "step": 14186
    },
    {
      "epoch": 0.7655819977335276,
      "grad_norm": 1.0356948220090099,
      "learning_rate": 2.1661282273112604e-06,
      "loss": 0.4736,
      "step": 14187
    },
    {
      "epoch": 0.765635961362042,
      "grad_norm": 0.9600368350789804,
      "learning_rate": 2.1656158713545635e-06,
      "loss": 0.401,
      "step": 14188
    },
    {
      "epoch": 0.7656899249905563,
      "grad_norm": 1.0799072749872123,
      "learning_rate": 2.165103611231437e-06,
      "loss": 0.4584,
      "step": 14189
    },
    {
      "epoch": 0.7657438886190707,
      "grad_norm": 1.3311947456817206,
      "learning_rate": 2.1645914469566006e-06,
      "loss": 0.5596,
      "step": 14190
    },
    {
      "epoch": 0.7657978522475851,
      "grad_norm": 0.9389451810837417,
      "learning_rate": 2.1640793785447766e-06,
      "loss": 0.4406,
      "step": 14191
    },
    {
      "epoch": 0.7658518158760995,
      "grad_norm": 0.887318733688726,
      "learning_rate": 2.1635674060106816e-06,
      "loss": 0.3229,
      "step": 14192
    },
    {
      "epoch": 0.7659057795046139,
      "grad_norm": 1.2113104262460652,
      "learning_rate": 2.1630555293690307e-06,
      "loss": 0.7512,
      "step": 14193
    },
    {
      "epoch": 0.7659597431331283,
      "grad_norm": 0.8703942727321965,
      "learning_rate": 2.1625437486345356e-06,
      "loss": 0.399,
      "step": 14194
    },
    {
      "epoch": 0.7660137067616426,
      "grad_norm": 1.2510142177818817,
      "learning_rate": 2.162032063821907e-06,
      "loss": 0.5029,
      "step": 14195
    },
    {
      "epoch": 0.766067670390157,
      "grad_norm": 1.1203145092525542,
      "learning_rate": 2.1615204749458486e-06,
      "loss": 0.4734,
      "step": 14196
    },
    {
      "epoch": 0.7661216340186714,
      "grad_norm": 1.197828515284275,
      "learning_rate": 2.1610089820210643e-06,
      "loss": 0.5636,
      "step": 14197
    },
    {
      "epoch": 0.7661755976471858,
      "grad_norm": 1.096667298324891,
      "learning_rate": 2.160497585062256e-06,
      "loss": 0.4598,
      "step": 14198
    },
    {
      "epoch": 0.7662295612757002,
      "grad_norm": 1.0165928435783191,
      "learning_rate": 2.1599862840841213e-06,
      "loss": 0.4308,
      "step": 14199
    },
    {
      "epoch": 0.7662835249042146,
      "grad_norm": 0.9981656631074851,
      "learning_rate": 2.1594750791013557e-06,
      "loss": 0.5132,
      "step": 14200
    },
    {
      "epoch": 0.766337488532729,
      "grad_norm": 1.1228662329921573,
      "learning_rate": 2.158963970128652e-06,
      "loss": 0.4867,
      "step": 14201
    },
    {
      "epoch": 0.7663914521612433,
      "grad_norm": 0.9518765215134315,
      "learning_rate": 2.1584529571807006e-06,
      "loss": 0.5246,
      "step": 14202
    },
    {
      "epoch": 0.7664454157897577,
      "grad_norm": 0.8835602761469125,
      "learning_rate": 2.157942040272186e-06,
      "loss": 0.3645,
      "step": 14203
    },
    {
      "epoch": 0.7664993794182721,
      "grad_norm": 0.7841990020267345,
      "learning_rate": 2.1574312194177945e-06,
      "loss": 0.3968,
      "step": 14204
    },
    {
      "epoch": 0.7665533430467865,
      "grad_norm": 1.0091223711030786,
      "learning_rate": 2.1569204946322072e-06,
      "loss": 0.5128,
      "step": 14205
    },
    {
      "epoch": 0.7666073066753009,
      "grad_norm": 0.8440343387684025,
      "learning_rate": 2.156409865930103e-06,
      "loss": 0.2964,
      "step": 14206
    },
    {
      "epoch": 0.7666612703038153,
      "grad_norm": 1.1542418756099928,
      "learning_rate": 2.1558993333261572e-06,
      "loss": 0.5922,
      "step": 14207
    },
    {
      "epoch": 0.7667152339323297,
      "grad_norm": 0.9867860427489499,
      "learning_rate": 2.1553888968350437e-06,
      "loss": 0.4621,
      "step": 14208
    },
    {
      "epoch": 0.7667691975608439,
      "grad_norm": 0.9787632574096162,
      "learning_rate": 2.1548785564714344e-06,
      "loss": 0.4274,
      "step": 14209
    },
    {
      "epoch": 0.7668231611893583,
      "grad_norm": 0.9783912784166661,
      "learning_rate": 2.154368312249994e-06,
      "loss": 0.5039,
      "step": 14210
    },
    {
      "epoch": 0.7668771248178727,
      "grad_norm": 1.0302440891054205,
      "learning_rate": 2.1538581641853885e-06,
      "loss": 0.4108,
      "step": 14211
    },
    {
      "epoch": 0.7669310884463871,
      "grad_norm": 1.3470618513672132,
      "learning_rate": 2.1533481122922804e-06,
      "loss": 0.595,
      "step": 14212
    },
    {
      "epoch": 0.7669850520749015,
      "grad_norm": 0.9578399120610472,
      "learning_rate": 2.1528381565853297e-06,
      "loss": 0.4495,
      "step": 14213
    },
    {
      "epoch": 0.7670390157034159,
      "grad_norm": 0.8629556371905172,
      "learning_rate": 2.1523282970791914e-06,
      "loss": 0.3305,
      "step": 14214
    },
    {
      "epoch": 0.7670929793319303,
      "grad_norm": 1.0093581037235846,
      "learning_rate": 2.1518185337885213e-06,
      "loss": 0.4366,
      "step": 14215
    },
    {
      "epoch": 0.7671469429604446,
      "grad_norm": 0.8955786521217854,
      "learning_rate": 2.1513088667279696e-06,
      "loss": 0.3984,
      "step": 14216
    },
    {
      "epoch": 0.767200906588959,
      "grad_norm": 0.9767114827414054,
      "learning_rate": 2.150799295912184e-06,
      "loss": 0.5098,
      "step": 14217
    },
    {
      "epoch": 0.7672548702174734,
      "grad_norm": 1.1021459671390357,
      "learning_rate": 2.1502898213558117e-06,
      "loss": 0.4459,
      "step": 14218
    },
    {
      "epoch": 0.7673088338459878,
      "grad_norm": 1.1719967600501304,
      "learning_rate": 2.1497804430734937e-06,
      "loss": 0.6049,
      "step": 14219
    },
    {
      "epoch": 0.7673627974745022,
      "grad_norm": 1.0085223728680084,
      "learning_rate": 2.149271161079871e-06,
      "loss": 0.4297,
      "step": 14220
    },
    {
      "epoch": 0.7674167611030166,
      "grad_norm": 0.8556113453301768,
      "learning_rate": 2.1487619753895817e-06,
      "loss": 0.4203,
      "step": 14221
    },
    {
      "epoch": 0.767470724731531,
      "grad_norm": 1.030769886544714,
      "learning_rate": 2.1482528860172585e-06,
      "loss": 0.4516,
      "step": 14222
    },
    {
      "epoch": 0.7675246883600453,
      "grad_norm": 1.080861895240368,
      "learning_rate": 2.147743892977536e-06,
      "loss": 0.356,
      "step": 14223
    },
    {
      "epoch": 0.7675786519885597,
      "grad_norm": 1.0128758835186016,
      "learning_rate": 2.147234996285039e-06,
      "loss": 0.4541,
      "step": 14224
    },
    {
      "epoch": 0.7676326156170741,
      "grad_norm": 0.929413556309162,
      "learning_rate": 2.146726195954397e-06,
      "loss": 0.3513,
      "step": 14225
    },
    {
      "epoch": 0.7676865792455885,
      "grad_norm": 0.9619793196371735,
      "learning_rate": 2.1462174920002315e-06,
      "loss": 0.4065,
      "step": 14226
    },
    {
      "epoch": 0.7677405428741029,
      "grad_norm": 1.0191993280076568,
      "learning_rate": 2.1457088844371644e-06,
      "loss": 0.6887,
      "step": 14227
    },
    {
      "epoch": 0.7677945065026173,
      "grad_norm": 0.9106749451473213,
      "learning_rate": 2.1452003732798133e-06,
      "loss": 0.3279,
      "step": 14228
    },
    {
      "epoch": 0.7678484701311317,
      "grad_norm": 1.0253190116067716,
      "learning_rate": 2.144691958542793e-06,
      "loss": 0.4967,
      "step": 14229
    },
    {
      "epoch": 0.767902433759646,
      "grad_norm": 1.1024949043569654,
      "learning_rate": 2.1441836402407175e-06,
      "loss": 0.4994,
      "step": 14230
    },
    {
      "epoch": 0.7679563973881604,
      "grad_norm": 1.121303841238638,
      "learning_rate": 2.1436754183881934e-06,
      "loss": 0.4662,
      "step": 14231
    },
    {
      "epoch": 0.7680103610166747,
      "grad_norm": 1.0286104271549372,
      "learning_rate": 2.1431672929998297e-06,
      "loss": 0.4208,
      "step": 14232
    },
    {
      "epoch": 0.7680643246451891,
      "grad_norm": 1.0944492165295847,
      "learning_rate": 2.142659264090229e-06,
      "loss": 0.4691,
      "step": 14233
    },
    {
      "epoch": 0.7681182882737035,
      "grad_norm": 1.1564859558490694,
      "learning_rate": 2.1421513316739945e-06,
      "loss": 0.4118,
      "step": 14234
    },
    {
      "epoch": 0.7681722519022179,
      "grad_norm": 1.0483422017477237,
      "learning_rate": 2.1416434957657227e-06,
      "loss": 0.4086,
      "step": 14235
    },
    {
      "epoch": 0.7682262155307323,
      "grad_norm": 0.9536361341713236,
      "learning_rate": 2.1411357563800106e-06,
      "loss": 0.4085,
      "step": 14236
    },
    {
      "epoch": 0.7682801791592466,
      "grad_norm": 0.9689108930763332,
      "learning_rate": 2.1406281135314523e-06,
      "loss": 0.3889,
      "step": 14237
    },
    {
      "epoch": 0.768334142787761,
      "grad_norm": 1.0006025609323963,
      "learning_rate": 2.1401205672346347e-06,
      "loss": 0.3848,
      "step": 14238
    },
    {
      "epoch": 0.7683881064162754,
      "grad_norm": 1.306741109114828,
      "learning_rate": 2.1396131175041467e-06,
      "loss": 0.671,
      "step": 14239
    },
    {
      "epoch": 0.7684420700447898,
      "grad_norm": 1.0370096571887635,
      "learning_rate": 2.1391057643545736e-06,
      "loss": 0.5284,
      "step": 14240
    },
    {
      "epoch": 0.7684960336733042,
      "grad_norm": 0.9081107053821491,
      "learning_rate": 2.1385985078004965e-06,
      "loss": 0.4168,
      "step": 14241
    },
    {
      "epoch": 0.7685499973018186,
      "grad_norm": 1.0336358240987997,
      "learning_rate": 2.138091347856495e-06,
      "loss": 0.3939,
      "step": 14242
    },
    {
      "epoch": 0.768603960930333,
      "grad_norm": 0.9168266007324878,
      "learning_rate": 2.137584284537145e-06,
      "loss": 0.4293,
      "step": 14243
    },
    {
      "epoch": 0.7686579245588473,
      "grad_norm": 0.7869110937545207,
      "learning_rate": 2.1370773178570216e-06,
      "loss": 0.3932,
      "step": 14244
    },
    {
      "epoch": 0.7687118881873617,
      "grad_norm": 1.1559919219411525,
      "learning_rate": 2.136570447830692e-06,
      "loss": 0.5533,
      "step": 14245
    },
    {
      "epoch": 0.7687658518158761,
      "grad_norm": 0.9963454711940187,
      "learning_rate": 2.136063674472727e-06,
      "loss": 0.4934,
      "step": 14246
    },
    {
      "epoch": 0.7688198154443905,
      "grad_norm": 0.8039381481472839,
      "learning_rate": 2.135556997797691e-06,
      "loss": 0.3326,
      "step": 14247
    },
    {
      "epoch": 0.7688737790729049,
      "grad_norm": 0.896315153703183,
      "learning_rate": 2.135050417820146e-06,
      "loss": 0.3246,
      "step": 14248
    },
    {
      "epoch": 0.7689277427014193,
      "grad_norm": 0.9294680108089669,
      "learning_rate": 2.1345439345546524e-06,
      "loss": 0.4219,
      "step": 14249
    },
    {
      "epoch": 0.7689817063299337,
      "grad_norm": 1.040677759763582,
      "learning_rate": 2.1340375480157667e-06,
      "loss": 0.4349,
      "step": 14250
    },
    {
      "epoch": 0.769035669958448,
      "grad_norm": 1.2527667249799772,
      "learning_rate": 2.1335312582180433e-06,
      "loss": 0.7291,
      "step": 14251
    },
    {
      "epoch": 0.7690896335869624,
      "grad_norm": 1.38159144242021,
      "learning_rate": 2.1330250651760333e-06,
      "loss": 0.5498,
      "step": 14252
    },
    {
      "epoch": 0.7691435972154768,
      "grad_norm": 0.9098150124832253,
      "learning_rate": 2.1325189689042836e-06,
      "loss": 0.3961,
      "step": 14253
    },
    {
      "epoch": 0.7691975608439912,
      "grad_norm": 0.9004359139429029,
      "learning_rate": 2.132012969417342e-06,
      "loss": 0.4398,
      "step": 14254
    },
    {
      "epoch": 0.7692515244725056,
      "grad_norm": 0.9673873151897715,
      "learning_rate": 2.131507066729751e-06,
      "loss": 0.4435,
      "step": 14255
    },
    {
      "epoch": 0.76930548810102,
      "grad_norm": 0.9783466807608096,
      "learning_rate": 2.13100126085605e-06,
      "loss": 0.3887,
      "step": 14256
    },
    {
      "epoch": 0.7693594517295343,
      "grad_norm": 1.2669337579433886,
      "learning_rate": 2.1304955518107772e-06,
      "loss": 0.5495,
      "step": 14257
    },
    {
      "epoch": 0.7694134153580486,
      "grad_norm": 0.9665587734335788,
      "learning_rate": 2.129989939608468e-06,
      "loss": 0.4791,
      "step": 14258
    },
    {
      "epoch": 0.769467378986563,
      "grad_norm": 0.9343944605128095,
      "learning_rate": 2.129484424263652e-06,
      "loss": 0.4635,
      "step": 14259
    },
    {
      "epoch": 0.7695213426150774,
      "grad_norm": 1.102448937766551,
      "learning_rate": 2.1289790057908592e-06,
      "loss": 0.4225,
      "step": 14260
    },
    {
      "epoch": 0.7695753062435918,
      "grad_norm": 1.1233919853113272,
      "learning_rate": 2.128473684204616e-06,
      "loss": 0.5577,
      "step": 14261
    },
    {
      "epoch": 0.7696292698721062,
      "grad_norm": 1.073993698625925,
      "learning_rate": 2.1279684595194453e-06,
      "loss": 0.4473,
      "step": 14262
    },
    {
      "epoch": 0.7696832335006206,
      "grad_norm": 1.1432879563612997,
      "learning_rate": 2.127463331749869e-06,
      "loss": 0.4986,
      "step": 14263
    },
    {
      "epoch": 0.7697371971291349,
      "grad_norm": 0.9694714796635561,
      "learning_rate": 2.1269583009104038e-06,
      "loss": 0.4473,
      "step": 14264
    },
    {
      "epoch": 0.7697911607576493,
      "grad_norm": 1.0396245757403264,
      "learning_rate": 2.1264533670155657e-06,
      "loss": 0.507,
      "step": 14265
    },
    {
      "epoch": 0.7698451243861637,
      "grad_norm": 1.0396770391116028,
      "learning_rate": 2.1259485300798675e-06,
      "loss": 0.4667,
      "step": 14266
    },
    {
      "epoch": 0.7698990880146781,
      "grad_norm": 0.8489437717516952,
      "learning_rate": 2.125443790117816e-06,
      "loss": 0.3607,
      "step": 14267
    },
    {
      "epoch": 0.7699530516431925,
      "grad_norm": 0.691182175608516,
      "learning_rate": 2.1249391471439206e-06,
      "loss": 0.2762,
      "step": 14268
    },
    {
      "epoch": 0.7700070152717069,
      "grad_norm": 1.079948133310876,
      "learning_rate": 2.124434601172684e-06,
      "loss": 0.5946,
      "step": 14269
    },
    {
      "epoch": 0.7700609789002213,
      "grad_norm": 1.115839397569839,
      "learning_rate": 2.123930152218607e-06,
      "loss": 0.4393,
      "step": 14270
    },
    {
      "epoch": 0.7701149425287356,
      "grad_norm": 0.9530309149603373,
      "learning_rate": 2.12342580029619e-06,
      "loss": 0.51,
      "step": 14271
    },
    {
      "epoch": 0.77016890615725,
      "grad_norm": 0.844352729393124,
      "learning_rate": 2.1229215454199265e-06,
      "loss": 0.3533,
      "step": 14272
    },
    {
      "epoch": 0.7702228697857644,
      "grad_norm": 0.9673428916122069,
      "learning_rate": 2.1224173876043104e-06,
      "loss": 0.4245,
      "step": 14273
    },
    {
      "epoch": 0.7702768334142788,
      "grad_norm": 1.0104721342411072,
      "learning_rate": 2.1219133268638307e-06,
      "loss": 0.4977,
      "step": 14274
    },
    {
      "epoch": 0.7703307970427932,
      "grad_norm": 1.2013452055612124,
      "learning_rate": 2.121409363212976e-06,
      "loss": 0.6165,
      "step": 14275
    },
    {
      "epoch": 0.7703847606713076,
      "grad_norm": 0.971873228303481,
      "learning_rate": 2.1209054966662303e-06,
      "loss": 0.4357,
      "step": 14276
    },
    {
      "epoch": 0.770438724299822,
      "grad_norm": 1.0110038025001873,
      "learning_rate": 2.1204017272380744e-06,
      "loss": 0.439,
      "step": 14277
    },
    {
      "epoch": 0.7704926879283363,
      "grad_norm": 1.066893556705141,
      "learning_rate": 2.119898054942989e-06,
      "loss": 0.4634,
      "step": 14278
    },
    {
      "epoch": 0.7705466515568506,
      "grad_norm": 0.8869175953016482,
      "learning_rate": 2.1193944797954473e-06,
      "loss": 0.4061,
      "step": 14279
    },
    {
      "epoch": 0.770600615185365,
      "grad_norm": 0.9208767090581513,
      "learning_rate": 2.1188910018099245e-06,
      "loss": 0.3859,
      "step": 14280
    },
    {
      "epoch": 0.7706545788138794,
      "grad_norm": 0.9867387224768924,
      "learning_rate": 2.118387621000891e-06,
      "loss": 0.4175,
      "step": 14281
    },
    {
      "epoch": 0.7707085424423938,
      "grad_norm": 0.9658982801667303,
      "learning_rate": 2.1178843373828135e-06,
      "loss": 0.4617,
      "step": 14282
    },
    {
      "epoch": 0.7707625060709082,
      "grad_norm": 1.0681469412889457,
      "learning_rate": 2.1173811509701575e-06,
      "loss": 0.4099,
      "step": 14283
    },
    {
      "epoch": 0.7708164696994226,
      "grad_norm": 1.1178953577981978,
      "learning_rate": 2.1168780617773856e-06,
      "loss": 0.4394,
      "step": 14284
    },
    {
      "epoch": 0.7708704333279369,
      "grad_norm": 0.8958802081916195,
      "learning_rate": 2.116375069818957e-06,
      "loss": 0.4143,
      "step": 14285
    },
    {
      "epoch": 0.7709243969564513,
      "grad_norm": 1.1924630492388575,
      "learning_rate": 2.1158721751093277e-06,
      "loss": 0.5219,
      "step": 14286
    },
    {
      "epoch": 0.7709783605849657,
      "grad_norm": 1.1741886962966726,
      "learning_rate": 2.115369377662951e-06,
      "loss": 0.5439,
      "step": 14287
    },
    {
      "epoch": 0.7710323242134801,
      "grad_norm": 1.0216582429965113,
      "learning_rate": 2.1148666774942785e-06,
      "loss": 0.3303,
      "step": 14288
    },
    {
      "epoch": 0.7710862878419945,
      "grad_norm": 1.236522005946748,
      "learning_rate": 2.1143640746177576e-06,
      "loss": 0.5081,
      "step": 14289
    },
    {
      "epoch": 0.7711402514705089,
      "grad_norm": 0.7961106974665461,
      "learning_rate": 2.113861569047834e-06,
      "loss": 0.3337,
      "step": 14290
    },
    {
      "epoch": 0.7711942150990233,
      "grad_norm": 1.0452259180502994,
      "learning_rate": 2.113359160798951e-06,
      "loss": 0.615,
      "step": 14291
    },
    {
      "epoch": 0.7712481787275376,
      "grad_norm": 0.9544171447792474,
      "learning_rate": 2.112856849885549e-06,
      "loss": 0.3842,
      "step": 14292
    },
    {
      "epoch": 0.771302142356052,
      "grad_norm": 0.9542611324505222,
      "learning_rate": 2.112354636322062e-06,
      "loss": 0.5281,
      "step": 14293
    },
    {
      "epoch": 0.7713561059845664,
      "grad_norm": 1.19495098040227,
      "learning_rate": 2.1118525201229254e-06,
      "loss": 0.7226,
      "step": 14294
    },
    {
      "epoch": 0.7714100696130808,
      "grad_norm": 0.9837780138903477,
      "learning_rate": 2.1113505013025713e-06,
      "loss": 0.4383,
      "step": 14295
    },
    {
      "epoch": 0.7714640332415952,
      "grad_norm": 1.0560853040044587,
      "learning_rate": 2.1108485798754275e-06,
      "loss": 0.5338,
      "step": 14296
    },
    {
      "epoch": 0.7715179968701096,
      "grad_norm": 1.1480644460917355,
      "learning_rate": 2.1103467558559203e-06,
      "loss": 0.6061,
      "step": 14297
    },
    {
      "epoch": 0.771571960498624,
      "grad_norm": 1.0340674054706762,
      "learning_rate": 2.109845029258472e-06,
      "loss": 0.4852,
      "step": 14298
    },
    {
      "epoch": 0.7716259241271383,
      "grad_norm": 0.9653277987224296,
      "learning_rate": 2.1093434000975047e-06,
      "loss": 0.3485,
      "step": 14299
    },
    {
      "epoch": 0.7716798877556527,
      "grad_norm": 1.0325467132311257,
      "learning_rate": 2.1088418683874327e-06,
      "loss": 0.4124,
      "step": 14300
    },
    {
      "epoch": 0.7717338513841671,
      "grad_norm": 1.0821064158381222,
      "learning_rate": 2.108340434142672e-06,
      "loss": 0.4119,
      "step": 14301
    },
    {
      "epoch": 0.7717878150126815,
      "grad_norm": 1.0884120212675803,
      "learning_rate": 2.1078390973776338e-06,
      "loss": 0.4339,
      "step": 14302
    },
    {
      "epoch": 0.7718417786411959,
      "grad_norm": 1.0849884438670336,
      "learning_rate": 2.107337858106728e-06,
      "loss": 0.4853,
      "step": 14303
    },
    {
      "epoch": 0.7718957422697102,
      "grad_norm": 1.1094765971279958,
      "learning_rate": 2.1068367163443596e-06,
      "loss": 0.5364,
      "step": 14304
    },
    {
      "epoch": 0.7719497058982246,
      "grad_norm": 1.1314439579594944,
      "learning_rate": 2.1063356721049333e-06,
      "loss": 0.5238,
      "step": 14305
    },
    {
      "epoch": 0.7720036695267389,
      "grad_norm": 0.8591360438066771,
      "learning_rate": 2.105834725402849e-06,
      "loss": 0.4131,
      "step": 14306
    },
    {
      "epoch": 0.7720576331552533,
      "grad_norm": 1.2300666986408573,
      "learning_rate": 2.1053338762525044e-06,
      "loss": 0.4682,
      "step": 14307
    },
    {
      "epoch": 0.7721115967837677,
      "grad_norm": 1.1913549131930028,
      "learning_rate": 2.104833124668293e-06,
      "loss": 0.5309,
      "step": 14308
    },
    {
      "epoch": 0.7721655604122821,
      "grad_norm": 0.9755711569933762,
      "learning_rate": 2.104332470664609e-06,
      "loss": 0.4183,
      "step": 14309
    },
    {
      "epoch": 0.7722195240407965,
      "grad_norm": 0.7273154489952732,
      "learning_rate": 2.1038319142558404e-06,
      "loss": 0.3599,
      "step": 14310
    },
    {
      "epoch": 0.7722734876693109,
      "grad_norm": 0.934213678997547,
      "learning_rate": 2.1033314554563743e-06,
      "loss": 0.3613,
      "step": 14311
    },
    {
      "epoch": 0.7723274512978253,
      "grad_norm": 0.9869528944148045,
      "learning_rate": 2.1028310942805944e-06,
      "loss": 0.4968,
      "step": 14312
    },
    {
      "epoch": 0.7723814149263396,
      "grad_norm": 1.1493784909207156,
      "learning_rate": 2.1023308307428826e-06,
      "loss": 0.5266,
      "step": 14313
    },
    {
      "epoch": 0.772435378554854,
      "grad_norm": 0.9387360642326492,
      "learning_rate": 2.1018306648576143e-06,
      "loss": 0.3935,
      "step": 14314
    },
    {
      "epoch": 0.7724893421833684,
      "grad_norm": 1.062510820733369,
      "learning_rate": 2.1013305966391665e-06,
      "loss": 0.4395,
      "step": 14315
    },
    {
      "epoch": 0.7725433058118828,
      "grad_norm": 0.8262267227732525,
      "learning_rate": 2.100830626101912e-06,
      "loss": 0.3324,
      "step": 14316
    },
    {
      "epoch": 0.7725972694403972,
      "grad_norm": 1.2435848776817182,
      "learning_rate": 2.100330753260219e-06,
      "loss": 0.5007,
      "step": 14317
    },
    {
      "epoch": 0.7726512330689116,
      "grad_norm": 0.9338430535118151,
      "learning_rate": 2.0998309781284552e-06,
      "loss": 0.3908,
      "step": 14318
    },
    {
      "epoch": 0.772705196697426,
      "grad_norm": 1.1663077577933865,
      "learning_rate": 2.0993313007209855e-06,
      "loss": 0.5217,
      "step": 14319
    },
    {
      "epoch": 0.7727591603259403,
      "grad_norm": 1.3257649393710185,
      "learning_rate": 2.09883172105217e-06,
      "loss": 0.5353,
      "step": 14320
    },
    {
      "epoch": 0.7728131239544547,
      "grad_norm": 1.0987590163461898,
      "learning_rate": 2.0983322391363677e-06,
      "loss": 0.5294,
      "step": 14321
    },
    {
      "epoch": 0.7728670875829691,
      "grad_norm": 1.2388604233043665,
      "learning_rate": 2.0978328549879335e-06,
      "loss": 0.5069,
      "step": 14322
    },
    {
      "epoch": 0.7729210512114835,
      "grad_norm": 0.9706719423805242,
      "learning_rate": 2.0973335686212208e-06,
      "loss": 0.4075,
      "step": 14323
    },
    {
      "epoch": 0.7729750148399979,
      "grad_norm": 1.058463769787896,
      "learning_rate": 2.09683438005058e-06,
      "loss": 0.4506,
      "step": 14324
    },
    {
      "epoch": 0.7730289784685123,
      "grad_norm": 1.0994357133679753,
      "learning_rate": 2.096335289290357e-06,
      "loss": 0.5042,
      "step": 14325
    },
    {
      "epoch": 0.7730829420970265,
      "grad_norm": 1.0013080357516266,
      "learning_rate": 2.095836296354899e-06,
      "loss": 0.5048,
      "step": 14326
    },
    {
      "epoch": 0.773136905725541,
      "grad_norm": 0.987420812848804,
      "learning_rate": 2.095337401258544e-06,
      "loss": 0.3857,
      "step": 14327
    },
    {
      "epoch": 0.7731908693540553,
      "grad_norm": 0.836910579882509,
      "learning_rate": 2.0948386040156317e-06,
      "loss": 0.3548,
      "step": 14328
    },
    {
      "epoch": 0.7732448329825697,
      "grad_norm": 0.8372340179199624,
      "learning_rate": 2.0943399046404987e-06,
      "loss": 0.4051,
      "step": 14329
    },
    {
      "epoch": 0.7732987966110841,
      "grad_norm": 1.1063531799983626,
      "learning_rate": 2.093841303147479e-06,
      "loss": 0.5324,
      "step": 14330
    },
    {
      "epoch": 0.7733527602395985,
      "grad_norm": 1.1825221131395094,
      "learning_rate": 2.0933427995509008e-06,
      "loss": 0.5282,
      "step": 14331
    },
    {
      "epoch": 0.7734067238681129,
      "grad_norm": 0.8861686125751678,
      "learning_rate": 2.0928443938650926e-06,
      "loss": 0.3157,
      "step": 14332
    },
    {
      "epoch": 0.7734606874966272,
      "grad_norm": 1.079579262303941,
      "learning_rate": 2.092346086104381e-06,
      "loss": 0.5429,
      "step": 14333
    },
    {
      "epoch": 0.7735146511251416,
      "grad_norm": 1.0186039028542024,
      "learning_rate": 2.0918478762830852e-06,
      "loss": 0.4188,
      "step": 14334
    },
    {
      "epoch": 0.773568614753656,
      "grad_norm": 1.0202251089662961,
      "learning_rate": 2.091349764415524e-06,
      "loss": 0.3858,
      "step": 14335
    },
    {
      "epoch": 0.7736225783821704,
      "grad_norm": 1.0658978603771039,
      "learning_rate": 2.0908517505160157e-06,
      "loss": 0.5141,
      "step": 14336
    },
    {
      "epoch": 0.7736765420106848,
      "grad_norm": 1.1478683205385043,
      "learning_rate": 2.0903538345988726e-06,
      "loss": 0.4743,
      "step": 14337
    },
    {
      "epoch": 0.7737305056391992,
      "grad_norm": 0.9234933914379745,
      "learning_rate": 2.0898560166784054e-06,
      "loss": 0.4121,
      "step": 14338
    },
    {
      "epoch": 0.7737844692677136,
      "grad_norm": 0.7353720006650354,
      "learning_rate": 2.089358296768922e-06,
      "loss": 0.3007,
      "step": 14339
    },
    {
      "epoch": 0.7738384328962279,
      "grad_norm": 1.160203013691209,
      "learning_rate": 2.088860674884729e-06,
      "loss": 0.4961,
      "step": 14340
    },
    {
      "epoch": 0.7738923965247423,
      "grad_norm": 1.1901919205743785,
      "learning_rate": 2.088363151040125e-06,
      "loss": 0.5558,
      "step": 14341
    },
    {
      "epoch": 0.7739463601532567,
      "grad_norm": 1.0757500728856886,
      "learning_rate": 2.087865725249411e-06,
      "loss": 0.5013,
      "step": 14342
    },
    {
      "epoch": 0.7740003237817711,
      "grad_norm": 1.2253694724614346,
      "learning_rate": 2.0873683975268847e-06,
      "loss": 0.3725,
      "step": 14343
    },
    {
      "epoch": 0.7740542874102855,
      "grad_norm": 1.028363710697949,
      "learning_rate": 2.086871167886839e-06,
      "loss": 0.3879,
      "step": 14344
    },
    {
      "epoch": 0.7741082510387999,
      "grad_norm": 1.142734041857299,
      "learning_rate": 2.0863740363435632e-06,
      "loss": 0.4855,
      "step": 14345
    },
    {
      "epoch": 0.7741622146673143,
      "grad_norm": 1.2561496969639887,
      "learning_rate": 2.0858770029113478e-06,
      "loss": 0.4657,
      "step": 14346
    },
    {
      "epoch": 0.7742161782958286,
      "grad_norm": 0.9144290021262295,
      "learning_rate": 2.0853800676044785e-06,
      "loss": 0.3505,
      "step": 14347
    },
    {
      "epoch": 0.774270141924343,
      "grad_norm": 1.0579784536567547,
      "learning_rate": 2.0848832304372347e-06,
      "loss": 0.5187,
      "step": 14348
    },
    {
      "epoch": 0.7743241055528574,
      "grad_norm": 1.054784929562731,
      "learning_rate": 2.0843864914238978e-06,
      "loss": 0.5176,
      "step": 14349
    },
    {
      "epoch": 0.7743780691813718,
      "grad_norm": 0.9262335345117498,
      "learning_rate": 2.083889850578744e-06,
      "loss": 0.4082,
      "step": 14350
    },
    {
      "epoch": 0.7744320328098862,
      "grad_norm": 0.9470698706197943,
      "learning_rate": 2.0833933079160473e-06,
      "loss": 0.3342,
      "step": 14351
    },
    {
      "epoch": 0.7744859964384005,
      "grad_norm": 1.1861652098478488,
      "learning_rate": 2.08289686345008e-06,
      "loss": 0.438,
      "step": 14352
    },
    {
      "epoch": 0.774539960066915,
      "grad_norm": 1.1219133485829607,
      "learning_rate": 2.082400517195109e-06,
      "loss": 0.4311,
      "step": 14353
    },
    {
      "epoch": 0.7745939236954292,
      "grad_norm": 1.0150602469393966,
      "learning_rate": 2.0819042691654023e-06,
      "loss": 0.4004,
      "step": 14354
    },
    {
      "epoch": 0.7746478873239436,
      "grad_norm": 1.3054476800728159,
      "learning_rate": 2.081408119375219e-06,
      "loss": 0.3654,
      "step": 14355
    },
    {
      "epoch": 0.774701850952458,
      "grad_norm": 0.9796761624209479,
      "learning_rate": 2.0809120678388202e-06,
      "loss": 0.555,
      "step": 14356
    },
    {
      "epoch": 0.7747558145809724,
      "grad_norm": 1.0953217420802197,
      "learning_rate": 2.080416114570464e-06,
      "loss": 0.4251,
      "step": 14357
    },
    {
      "epoch": 0.7748097782094868,
      "grad_norm": 1.0333740031127727,
      "learning_rate": 2.0799202595844033e-06,
      "loss": 0.3472,
      "step": 14358
    },
    {
      "epoch": 0.7748637418380012,
      "grad_norm": 1.0158728755176933,
      "learning_rate": 2.0794245028948904e-06,
      "loss": 0.355,
      "step": 14359
    },
    {
      "epoch": 0.7749177054665156,
      "grad_norm": 0.8097227111131159,
      "learning_rate": 2.0789288445161736e-06,
      "loss": 0.3196,
      "step": 14360
    },
    {
      "epoch": 0.7749716690950299,
      "grad_norm": 1.020862447306454,
      "learning_rate": 2.0784332844624994e-06,
      "loss": 0.396,
      "step": 14361
    },
    {
      "epoch": 0.7750256327235443,
      "grad_norm": 1.0765236816529666,
      "learning_rate": 2.077937822748109e-06,
      "loss": 0.4822,
      "step": 14362
    },
    {
      "epoch": 0.7750795963520587,
      "grad_norm": 1.1305948413338578,
      "learning_rate": 2.077442459387243e-06,
      "loss": 0.5108,
      "step": 14363
    },
    {
      "epoch": 0.7751335599805731,
      "grad_norm": 1.0551246459773433,
      "learning_rate": 2.076947194394139e-06,
      "loss": 0.5606,
      "step": 14364
    },
    {
      "epoch": 0.7751875236090875,
      "grad_norm": 1.1311752063643037,
      "learning_rate": 2.0764520277830312e-06,
      "loss": 0.6118,
      "step": 14365
    },
    {
      "epoch": 0.7752414872376019,
      "grad_norm": 1.0499145864352535,
      "learning_rate": 2.075956959568152e-06,
      "loss": 0.4381,
      "step": 14366
    },
    {
      "epoch": 0.7752954508661163,
      "grad_norm": 0.9496719552405049,
      "learning_rate": 2.075461989763729e-06,
      "loss": 0.3996,
      "step": 14367
    },
    {
      "epoch": 0.7753494144946306,
      "grad_norm": 1.276578970790503,
      "learning_rate": 2.074967118383989e-06,
      "loss": 0.5822,
      "step": 14368
    },
    {
      "epoch": 0.775403378123145,
      "grad_norm": 0.8687332110688493,
      "learning_rate": 2.0744723454431547e-06,
      "loss": 0.362,
      "step": 14369
    },
    {
      "epoch": 0.7754573417516594,
      "grad_norm": 1.0007149470769077,
      "learning_rate": 2.073977670955446e-06,
      "loss": 0.4037,
      "step": 14370
    },
    {
      "epoch": 0.7755113053801738,
      "grad_norm": 1.0360504846959904,
      "learning_rate": 2.073483094935081e-06,
      "loss": 0.3986,
      "step": 14371
    },
    {
      "epoch": 0.7755652690086882,
      "grad_norm": 0.9598494425256857,
      "learning_rate": 2.072988617396276e-06,
      "loss": 0.4312,
      "step": 14372
    },
    {
      "epoch": 0.7756192326372026,
      "grad_norm": 1.0283033312013563,
      "learning_rate": 2.0724942383532387e-06,
      "loss": 0.4248,
      "step": 14373
    },
    {
      "epoch": 0.775673196265717,
      "grad_norm": 1.0725744464930982,
      "learning_rate": 2.0719999578201803e-06,
      "loss": 0.5608,
      "step": 14374
    },
    {
      "epoch": 0.7757271598942312,
      "grad_norm": 0.8964152460758551,
      "learning_rate": 2.071505775811308e-06,
      "loss": 0.313,
      "step": 14375
    },
    {
      "epoch": 0.7757811235227456,
      "grad_norm": 0.8387030913175174,
      "learning_rate": 2.0710116923408226e-06,
      "loss": 0.2842,
      "step": 14376
    },
    {
      "epoch": 0.77583508715126,
      "grad_norm": 1.0972072867608762,
      "learning_rate": 2.0705177074229264e-06,
      "loss": 0.5419,
      "step": 14377
    },
    {
      "epoch": 0.7758890507797744,
      "grad_norm": 0.8053923197315072,
      "learning_rate": 2.0700238210718165e-06,
      "loss": 0.4089,
      "step": 14378
    },
    {
      "epoch": 0.7759430144082888,
      "grad_norm": 1.1671393645957278,
      "learning_rate": 2.069530033301688e-06,
      "loss": 0.4847,
      "step": 14379
    },
    {
      "epoch": 0.7759969780368032,
      "grad_norm": 0.9685300367127977,
      "learning_rate": 2.0690363441267324e-06,
      "loss": 0.4278,
      "step": 14380
    },
    {
      "epoch": 0.7760509416653176,
      "grad_norm": 1.05425181815556,
      "learning_rate": 2.0685427535611387e-06,
      "loss": 0.4749,
      "step": 14381
    },
    {
      "epoch": 0.7761049052938319,
      "grad_norm": 1.0113027903349576,
      "learning_rate": 2.068049261619095e-06,
      "loss": 0.5165,
      "step": 14382
    },
    {
      "epoch": 0.7761588689223463,
      "grad_norm": 1.1259911360388273,
      "learning_rate": 2.0675558683147823e-06,
      "loss": 0.5087,
      "step": 14383
    },
    {
      "epoch": 0.7762128325508607,
      "grad_norm": 0.7211926499314123,
      "learning_rate": 2.0670625736623816e-06,
      "loss": 0.2655,
      "step": 14384
    },
    {
      "epoch": 0.7762667961793751,
      "grad_norm": 0.9835995575578486,
      "learning_rate": 2.0665693776760717e-06,
      "loss": 0.3718,
      "step": 14385
    },
    {
      "epoch": 0.7763207598078895,
      "grad_norm": 0.8223119645906154,
      "learning_rate": 2.0660762803700267e-06,
      "loss": 0.3696,
      "step": 14386
    },
    {
      "epoch": 0.7763747234364039,
      "grad_norm": 1.0760840311062059,
      "learning_rate": 2.0655832817584193e-06,
      "loss": 0.5231,
      "step": 14387
    },
    {
      "epoch": 0.7764286870649183,
      "grad_norm": 1.0057243395012716,
      "learning_rate": 2.0650903818554187e-06,
      "loss": 0.4305,
      "step": 14388
    },
    {
      "epoch": 0.7764826506934326,
      "grad_norm": 1.0719124799698891,
      "learning_rate": 2.0645975806751926e-06,
      "loss": 0.4625,
      "step": 14389
    },
    {
      "epoch": 0.776536614321947,
      "grad_norm": 1.0455960047342394,
      "learning_rate": 2.064104878231902e-06,
      "loss": 0.4216,
      "step": 14390
    },
    {
      "epoch": 0.7765905779504614,
      "grad_norm": 0.9128095241420763,
      "learning_rate": 2.063612274539709e-06,
      "loss": 0.4446,
      "step": 14391
    },
    {
      "epoch": 0.7766445415789758,
      "grad_norm": 0.7802035930995922,
      "learning_rate": 2.063119769612771e-06,
      "loss": 0.3446,
      "step": 14392
    },
    {
      "epoch": 0.7766985052074902,
      "grad_norm": 1.161108802392436,
      "learning_rate": 2.0626273634652437e-06,
      "loss": 0.5247,
      "step": 14393
    },
    {
      "epoch": 0.7767524688360046,
      "grad_norm": 0.9272204876071786,
      "learning_rate": 2.062135056111279e-06,
      "loss": 0.3808,
      "step": 14394
    },
    {
      "epoch": 0.7768064324645189,
      "grad_norm": 1.2133573149954733,
      "learning_rate": 2.061642847565028e-06,
      "loss": 0.4893,
      "step": 14395
    },
    {
      "epoch": 0.7768603960930333,
      "grad_norm": 1.2200092919907302,
      "learning_rate": 2.0611507378406346e-06,
      "loss": 0.5801,
      "step": 14396
    },
    {
      "epoch": 0.7769143597215477,
      "grad_norm": 1.0109356966178489,
      "learning_rate": 2.0606587269522438e-06,
      "loss": 0.466,
      "step": 14397
    },
    {
      "epoch": 0.776968323350062,
      "grad_norm": 1.113021699656678,
      "learning_rate": 2.0601668149139963e-06,
      "loss": 0.5077,
      "step": 14398
    },
    {
      "epoch": 0.7770222869785764,
      "grad_norm": 0.9414631942392963,
      "learning_rate": 2.05967500174003e-06,
      "loss": 0.4067,
      "step": 14399
    },
    {
      "epoch": 0.7770762506070908,
      "grad_norm": 1.1966292087506343,
      "learning_rate": 2.0591832874444803e-06,
      "loss": 0.3936,
      "step": 14400
    },
    {
      "epoch": 0.7771302142356052,
      "grad_norm": 1.0229313821570796,
      "learning_rate": 2.05869167204148e-06,
      "loss": 0.4955,
      "step": 14401
    },
    {
      "epoch": 0.7771841778641195,
      "grad_norm": 1.0884067483616644,
      "learning_rate": 2.0582001555451596e-06,
      "loss": 0.4968,
      "step": 14402
    },
    {
      "epoch": 0.7772381414926339,
      "grad_norm": 1.1496113895405562,
      "learning_rate": 2.0577087379696427e-06,
      "loss": 0.4997,
      "step": 14403
    },
    {
      "epoch": 0.7772921051211483,
      "grad_norm": 1.0316091585506433,
      "learning_rate": 2.057217419329055e-06,
      "loss": 0.4616,
      "step": 14404
    },
    {
      "epoch": 0.7773460687496627,
      "grad_norm": 1.0188820428988987,
      "learning_rate": 2.0567261996375177e-06,
      "loss": 0.4651,
      "step": 14405
    },
    {
      "epoch": 0.7774000323781771,
      "grad_norm": 1.0805457812576844,
      "learning_rate": 2.0562350789091474e-06,
      "loss": 0.5871,
      "step": 14406
    },
    {
      "epoch": 0.7774539960066915,
      "grad_norm": 1.119655554710967,
      "learning_rate": 2.0557440571580615e-06,
      "loss": 0.4714,
      "step": 14407
    },
    {
      "epoch": 0.7775079596352059,
      "grad_norm": 1.1749782463613263,
      "learning_rate": 2.0552531343983715e-06,
      "loss": 0.5361,
      "step": 14408
    },
    {
      "epoch": 0.7775619232637202,
      "grad_norm": 0.9616241526148085,
      "learning_rate": 2.054762310644189e-06,
      "loss": 0.3761,
      "step": 14409
    },
    {
      "epoch": 0.7776158868922346,
      "grad_norm": 1.176193877243342,
      "learning_rate": 2.054271585909617e-06,
      "loss": 0.5797,
      "step": 14410
    },
    {
      "epoch": 0.777669850520749,
      "grad_norm": 1.119055597181627,
      "learning_rate": 2.0537809602087607e-06,
      "loss": 0.6086,
      "step": 14411
    },
    {
      "epoch": 0.7777238141492634,
      "grad_norm": 0.9609628723473139,
      "learning_rate": 2.0532904335557225e-06,
      "loss": 0.4962,
      "step": 14412
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 0.965922113810176,
      "learning_rate": 2.0528000059646e-06,
      "loss": 0.4794,
      "step": 14413
    },
    {
      "epoch": 0.7778317414062922,
      "grad_norm": 1.033505263402845,
      "learning_rate": 2.052309677449488e-06,
      "loss": 0.5029,
      "step": 14414
    },
    {
      "epoch": 0.7778857050348066,
      "grad_norm": 0.927512485467369,
      "learning_rate": 2.0518194480244795e-06,
      "loss": 0.4097,
      "step": 14415
    },
    {
      "epoch": 0.7779396686633209,
      "grad_norm": 0.8795300774054674,
      "learning_rate": 2.051329317703665e-06,
      "loss": 0.3341,
      "step": 14416
    },
    {
      "epoch": 0.7779936322918353,
      "grad_norm": 0.9623697370881142,
      "learning_rate": 2.0508392865011304e-06,
      "loss": 0.3764,
      "step": 14417
    },
    {
      "epoch": 0.7780475959203497,
      "grad_norm": 0.9902206413756315,
      "learning_rate": 2.050349354430959e-06,
      "loss": 0.4537,
      "step": 14418
    },
    {
      "epoch": 0.7781015595488641,
      "grad_norm": 0.7042370685625253,
      "learning_rate": 2.049859521507233e-06,
      "loss": 0.3146,
      "step": 14419
    },
    {
      "epoch": 0.7781555231773785,
      "grad_norm": 1.1992730750468843,
      "learning_rate": 2.049369787744031e-06,
      "loss": 0.5277,
      "step": 14420
    },
    {
      "epoch": 0.7782094868058929,
      "grad_norm": 1.2285729890173887,
      "learning_rate": 2.0488801531554275e-06,
      "loss": 0.6661,
      "step": 14421
    },
    {
      "epoch": 0.7782634504344073,
      "grad_norm": 0.8964973857904519,
      "learning_rate": 2.0483906177554953e-06,
      "loss": 0.3293,
      "step": 14422
    },
    {
      "epoch": 0.7783174140629215,
      "grad_norm": 0.8173687043236592,
      "learning_rate": 2.0479011815583043e-06,
      "loss": 0.3657,
      "step": 14423
    },
    {
      "epoch": 0.7783713776914359,
      "grad_norm": 1.1450500831117785,
      "learning_rate": 2.047411844577922e-06,
      "loss": 0.5061,
      "step": 14424
    },
    {
      "epoch": 0.7784253413199503,
      "grad_norm": 1.1443693699499344,
      "learning_rate": 2.0469226068284117e-06,
      "loss": 0.4878,
      "step": 14425
    },
    {
      "epoch": 0.7784793049484647,
      "grad_norm": 1.228838390917046,
      "learning_rate": 2.046433468323834e-06,
      "loss": 0.5364,
      "step": 14426
    },
    {
      "epoch": 0.7785332685769791,
      "grad_norm": 1.1320499738938357,
      "learning_rate": 2.0459444290782488e-06,
      "loss": 0.615,
      "step": 14427
    },
    {
      "epoch": 0.7785872322054935,
      "grad_norm": 0.7859517898059896,
      "learning_rate": 2.04545548910571e-06,
      "loss": 0.335,
      "step": 14428
    },
    {
      "epoch": 0.7786411958340079,
      "grad_norm": 0.9314475316010032,
      "learning_rate": 2.044966648420272e-06,
      "loss": 0.5007,
      "step": 14429
    },
    {
      "epoch": 0.7786951594625222,
      "grad_norm": 1.022743549693376,
      "learning_rate": 2.044477907035984e-06,
      "loss": 0.3695,
      "step": 14430
    },
    {
      "epoch": 0.7787491230910366,
      "grad_norm": 0.9557945164575815,
      "learning_rate": 2.0439892649668918e-06,
      "loss": 0.4189,
      "step": 14431
    },
    {
      "epoch": 0.778803086719551,
      "grad_norm": 0.9278745865160652,
      "learning_rate": 2.04350072222704e-06,
      "loss": 0.3587,
      "step": 14432
    },
    {
      "epoch": 0.7788570503480654,
      "grad_norm": 1.140461045450529,
      "learning_rate": 2.04301227883047e-06,
      "loss": 0.4414,
      "step": 14433
    },
    {
      "epoch": 0.7789110139765798,
      "grad_norm": 1.090949650275269,
      "learning_rate": 2.04252393479122e-06,
      "loss": 0.6113,
      "step": 14434
    },
    {
      "epoch": 0.7789649776050942,
      "grad_norm": 0.934314375940731,
      "learning_rate": 2.0420356901233256e-06,
      "loss": 0.4277,
      "step": 14435
    },
    {
      "epoch": 0.7790189412336086,
      "grad_norm": 0.9272732000360229,
      "learning_rate": 2.0415475448408195e-06,
      "loss": 0.3367,
      "step": 14436
    },
    {
      "epoch": 0.7790729048621229,
      "grad_norm": 0.9427236644770652,
      "learning_rate": 2.041059498957733e-06,
      "loss": 0.4637,
      "step": 14437
    },
    {
      "epoch": 0.7791268684906373,
      "grad_norm": 1.0290089913095755,
      "learning_rate": 2.04057155248809e-06,
      "loss": 0.433,
      "step": 14438
    },
    {
      "epoch": 0.7791808321191517,
      "grad_norm": 1.0104285536623612,
      "learning_rate": 2.0400837054459165e-06,
      "loss": 0.5414,
      "step": 14439
    },
    {
      "epoch": 0.7792347957476661,
      "grad_norm": 1.2058858741274943,
      "learning_rate": 2.039595957845233e-06,
      "loss": 0.5443,
      "step": 14440
    },
    {
      "epoch": 0.7792887593761805,
      "grad_norm": 0.8988980148026456,
      "learning_rate": 2.039108309700058e-06,
      "loss": 0.2988,
      "step": 14441
    },
    {
      "epoch": 0.7793427230046949,
      "grad_norm": 1.2661650798940882,
      "learning_rate": 2.0386207610244073e-06,
      "loss": 0.6166,
      "step": 14442
    },
    {
      "epoch": 0.7793966866332093,
      "grad_norm": 1.0909141755802485,
      "learning_rate": 2.038133311832294e-06,
      "loss": 0.4535,
      "step": 14443
    },
    {
      "epoch": 0.7794506502617236,
      "grad_norm": 1.0209353091442916,
      "learning_rate": 2.0376459621377274e-06,
      "loss": 0.5598,
      "step": 14444
    },
    {
      "epoch": 0.779504613890238,
      "grad_norm": 0.8932556600163734,
      "learning_rate": 2.0371587119547133e-06,
      "loss": 0.3668,
      "step": 14445
    },
    {
      "epoch": 0.7795585775187523,
      "grad_norm": 1.2737370946830107,
      "learning_rate": 2.036671561297257e-06,
      "loss": 0.414,
      "step": 14446
    },
    {
      "epoch": 0.7796125411472667,
      "grad_norm": 1.037955347250833,
      "learning_rate": 2.0361845101793597e-06,
      "loss": 0.4399,
      "step": 14447
    },
    {
      "epoch": 0.7796665047757811,
      "grad_norm": 1.2044028187848255,
      "learning_rate": 2.035697558615019e-06,
      "loss": 0.5112,
      "step": 14448
    },
    {
      "epoch": 0.7797204684042955,
      "grad_norm": 1.2188368998085295,
      "learning_rate": 2.0352107066182312e-06,
      "loss": 0.4468,
      "step": 14449
    },
    {
      "epoch": 0.7797744320328099,
      "grad_norm": 1.0109406777428043,
      "learning_rate": 2.034723954202989e-06,
      "loss": 0.4905,
      "step": 14450
    },
    {
      "epoch": 0.7798283956613242,
      "grad_norm": 0.9923325755110624,
      "learning_rate": 2.034237301383283e-06,
      "loss": 0.3956,
      "step": 14451
    },
    {
      "epoch": 0.7798823592898386,
      "grad_norm": 1.0201644367358738,
      "learning_rate": 2.0337507481730962e-06,
      "loss": 0.4434,
      "step": 14452
    },
    {
      "epoch": 0.779936322918353,
      "grad_norm": 1.2982836971463227,
      "learning_rate": 2.0332642945864165e-06,
      "loss": 0.6354,
      "step": 14453
    },
    {
      "epoch": 0.7799902865468674,
      "grad_norm": 1.0749164710417765,
      "learning_rate": 2.032777940637224e-06,
      "loss": 0.4394,
      "step": 14454
    },
    {
      "epoch": 0.7800442501753818,
      "grad_norm": 0.9809456651532976,
      "learning_rate": 2.0322916863394964e-06,
      "loss": 0.4937,
      "step": 14455
    },
    {
      "epoch": 0.7800982138038962,
      "grad_norm": 1.0395330957157807,
      "learning_rate": 2.0318055317072094e-06,
      "loss": 0.4315,
      "step": 14456
    },
    {
      "epoch": 0.7801521774324106,
      "grad_norm": 0.9248108032665019,
      "learning_rate": 2.031319476754336e-06,
      "loss": 0.3659,
      "step": 14457
    },
    {
      "epoch": 0.7802061410609249,
      "grad_norm": 1.0012895896883862,
      "learning_rate": 2.0308335214948477e-06,
      "loss": 0.4294,
      "step": 14458
    },
    {
      "epoch": 0.7802601046894393,
      "grad_norm": 1.098053149539852,
      "learning_rate": 2.0303476659427073e-06,
      "loss": 0.5854,
      "step": 14459
    },
    {
      "epoch": 0.7803140683179537,
      "grad_norm": 0.9780917670674242,
      "learning_rate": 2.0298619101118807e-06,
      "loss": 0.4355,
      "step": 14460
    },
    {
      "epoch": 0.7803680319464681,
      "grad_norm": 0.8427677745087346,
      "learning_rate": 2.029376254016329e-06,
      "loss": 0.384,
      "step": 14461
    },
    {
      "epoch": 0.7804219955749825,
      "grad_norm": 1.1406418680394432,
      "learning_rate": 2.0288906976700113e-06,
      "loss": 0.5202,
      "step": 14462
    },
    {
      "epoch": 0.7804759592034969,
      "grad_norm": 1.254188639345077,
      "learning_rate": 2.028405241086882e-06,
      "loss": 0.4134,
      "step": 14463
    },
    {
      "epoch": 0.7805299228320112,
      "grad_norm": 1.0492251735850429,
      "learning_rate": 2.0279198842808946e-06,
      "loss": 0.4175,
      "step": 14464
    },
    {
      "epoch": 0.7805838864605256,
      "grad_norm": 1.0579110904317792,
      "learning_rate": 2.0274346272659973e-06,
      "loss": 0.5302,
      "step": 14465
    },
    {
      "epoch": 0.78063785008904,
      "grad_norm": 1.1737391589162267,
      "learning_rate": 2.0269494700561372e-06,
      "loss": 0.5993,
      "step": 14466
    },
    {
      "epoch": 0.7806918137175544,
      "grad_norm": 1.2485248934360795,
      "learning_rate": 2.026464412665259e-06,
      "loss": 0.6083,
      "step": 14467
    },
    {
      "epoch": 0.7807457773460688,
      "grad_norm": 1.1304342612532223,
      "learning_rate": 2.0259794551073033e-06,
      "loss": 0.544,
      "step": 14468
    },
    {
      "epoch": 0.7807997409745832,
      "grad_norm": 0.8648598273865524,
      "learning_rate": 2.0254945973962074e-06,
      "loss": 0.4178,
      "step": 14469
    },
    {
      "epoch": 0.7808537046030976,
      "grad_norm": 1.2575831402439106,
      "learning_rate": 2.025009839545909e-06,
      "loss": 0.5421,
      "step": 14470
    },
    {
      "epoch": 0.7809076682316118,
      "grad_norm": 0.8857441769585577,
      "learning_rate": 2.024525181570338e-06,
      "loss": 0.3173,
      "step": 14471
    },
    {
      "epoch": 0.7809616318601262,
      "grad_norm": 1.0066849779756508,
      "learning_rate": 2.0240406234834255e-06,
      "loss": 0.4779,
      "step": 14472
    },
    {
      "epoch": 0.7810155954886406,
      "grad_norm": 1.2275309832958314,
      "learning_rate": 2.023556165299097e-06,
      "loss": 0.5706,
      "step": 14473
    },
    {
      "epoch": 0.781069559117155,
      "grad_norm": 0.9900239791453298,
      "learning_rate": 2.0230718070312776e-06,
      "loss": 0.4283,
      "step": 14474
    },
    {
      "epoch": 0.7811235227456694,
      "grad_norm": 0.9687414298778189,
      "learning_rate": 2.0225875486938873e-06,
      "loss": 0.4308,
      "step": 14475
    },
    {
      "epoch": 0.7811774863741838,
      "grad_norm": 1.0549694628033899,
      "learning_rate": 2.022103390300845e-06,
      "loss": 0.5082,
      "step": 14476
    },
    {
      "epoch": 0.7812314500026982,
      "grad_norm": 1.1071721553353007,
      "learning_rate": 2.0216193318660647e-06,
      "loss": 0.5297,
      "step": 14477
    },
    {
      "epoch": 0.7812854136312125,
      "grad_norm": 1.1786026297534808,
      "learning_rate": 2.0211353734034616e-06,
      "loss": 0.5893,
      "step": 14478
    },
    {
      "epoch": 0.7813393772597269,
      "grad_norm": 0.9907743011426239,
      "learning_rate": 2.020651514926941e-06,
      "loss": 0.4203,
      "step": 14479
    },
    {
      "epoch": 0.7813933408882413,
      "grad_norm": 0.9411050437879911,
      "learning_rate": 2.020167756450412e-06,
      "loss": 0.4515,
      "step": 14480
    },
    {
      "epoch": 0.7814473045167557,
      "grad_norm": 0.8430810236880563,
      "learning_rate": 2.0196840979877773e-06,
      "loss": 0.3761,
      "step": 14481
    },
    {
      "epoch": 0.7815012681452701,
      "grad_norm": 1.1588162596435063,
      "learning_rate": 2.0192005395529385e-06,
      "loss": 0.6014,
      "step": 14482
    },
    {
      "epoch": 0.7815552317737845,
      "grad_norm": 0.9977850186804282,
      "learning_rate": 2.0187170811597926e-06,
      "loss": 0.4551,
      "step": 14483
    },
    {
      "epoch": 0.7816091954022989,
      "grad_norm": 1.008223282880729,
      "learning_rate": 2.0182337228222366e-06,
      "loss": 0.4626,
      "step": 14484
    },
    {
      "epoch": 0.7816631590308132,
      "grad_norm": 1.1051713262692409,
      "learning_rate": 2.0177504645541623e-06,
      "loss": 0.4952,
      "step": 14485
    },
    {
      "epoch": 0.7817171226593276,
      "grad_norm": 1.1868618035459073,
      "learning_rate": 2.0172673063694566e-06,
      "loss": 0.5616,
      "step": 14486
    },
    {
      "epoch": 0.781771086287842,
      "grad_norm": 1.1805049111733419,
      "learning_rate": 2.0167842482820077e-06,
      "loss": 0.4974,
      "step": 14487
    },
    {
      "epoch": 0.7818250499163564,
      "grad_norm": 1.1702871891226332,
      "learning_rate": 2.0163012903056987e-06,
      "loss": 0.5857,
      "step": 14488
    },
    {
      "epoch": 0.7818790135448708,
      "grad_norm": 1.2479803201837485,
      "learning_rate": 2.0158184324544118e-06,
      "loss": 0.6273,
      "step": 14489
    },
    {
      "epoch": 0.7819329771733852,
      "grad_norm": 1.02857235519036,
      "learning_rate": 2.0153356747420225e-06,
      "loss": 0.498,
      "step": 14490
    },
    {
      "epoch": 0.7819869408018996,
      "grad_norm": 1.1579185071793177,
      "learning_rate": 2.014853017182408e-06,
      "loss": 0.4698,
      "step": 14491
    },
    {
      "epoch": 0.7820409044304139,
      "grad_norm": 0.871069709967407,
      "learning_rate": 2.0143704597894388e-06,
      "loss": 0.4174,
      "step": 14492
    },
    {
      "epoch": 0.7820948680589282,
      "grad_norm": 1.3203712565476016,
      "learning_rate": 2.0138880025769847e-06,
      "loss": 0.5711,
      "step": 14493
    },
    {
      "epoch": 0.7821488316874426,
      "grad_norm": 0.86232776291382,
      "learning_rate": 2.0134056455589114e-06,
      "loss": 0.3138,
      "step": 14494
    },
    {
      "epoch": 0.782202795315957,
      "grad_norm": 0.9088384766482491,
      "learning_rate": 2.012923388749083e-06,
      "loss": 0.3711,
      "step": 14495
    },
    {
      "epoch": 0.7822567589444714,
      "grad_norm": 0.9011836768532737,
      "learning_rate": 2.0124412321613594e-06,
      "loss": 0.3593,
      "step": 14496
    },
    {
      "epoch": 0.7823107225729858,
      "grad_norm": 0.9488262373048938,
      "learning_rate": 2.011959175809599e-06,
      "loss": 0.4802,
      "step": 14497
    },
    {
      "epoch": 0.7823646862015002,
      "grad_norm": 1.0321900076323258,
      "learning_rate": 2.0114772197076564e-06,
      "loss": 0.4377,
      "step": 14498
    },
    {
      "epoch": 0.7824186498300145,
      "grad_norm": 1.1728024797736873,
      "learning_rate": 2.010995363869385e-06,
      "loss": 0.4675,
      "step": 14499
    },
    {
      "epoch": 0.7824726134585289,
      "grad_norm": 0.945161989143589,
      "learning_rate": 2.0105136083086303e-06,
      "loss": 0.4364,
      "step": 14500
    },
    {
      "epoch": 0.7824726134585289,
      "eval_loss": 0.5324375629425049,
      "eval_runtime": 164.3441,
      "eval_samples_per_second": 20.926,
      "eval_steps_per_second": 0.876,
      "step": 14500
    },
    {
      "epoch": 0.7825265770870433,
      "grad_norm": 1.0353846129642752,
      "learning_rate": 2.010031953039241e-06,
      "loss": 0.4253,
      "step": 14501
    },
    {
      "epoch": 0.7825805407155577,
      "grad_norm": 0.9617501953433107,
      "learning_rate": 2.0095503980750586e-06,
      "loss": 0.4071,
      "step": 14502
    },
    {
      "epoch": 0.7826345043440721,
      "grad_norm": 1.1316488692146285,
      "learning_rate": 2.0090689434299256e-06,
      "loss": 0.4627,
      "step": 14503
    },
    {
      "epoch": 0.7826884679725865,
      "grad_norm": 1.1922138238286948,
      "learning_rate": 2.008587589117678e-06,
      "loss": 0.5853,
      "step": 14504
    },
    {
      "epoch": 0.7827424316011009,
      "grad_norm": 1.1350677370168967,
      "learning_rate": 2.0081063351521518e-06,
      "loss": 0.5289,
      "step": 14505
    },
    {
      "epoch": 0.7827963952296152,
      "grad_norm": 0.9102313093006867,
      "learning_rate": 2.007625181547178e-06,
      "loss": 0.3898,
      "step": 14506
    },
    {
      "epoch": 0.7828503588581296,
      "grad_norm": 0.8304204735228492,
      "learning_rate": 2.0071441283165847e-06,
      "loss": 0.3726,
      "step": 14507
    },
    {
      "epoch": 0.782904322486644,
      "grad_norm": 0.9821913928769301,
      "learning_rate": 2.0066631754741976e-06,
      "loss": 0.45,
      "step": 14508
    },
    {
      "epoch": 0.7829582861151584,
      "grad_norm": 0.9426820772123301,
      "learning_rate": 2.0061823230338414e-06,
      "loss": 0.4777,
      "step": 14509
    },
    {
      "epoch": 0.7830122497436728,
      "grad_norm": 1.0131095534742025,
      "learning_rate": 2.005701571009335e-06,
      "loss": 0.4378,
      "step": 14510
    },
    {
      "epoch": 0.7830662133721872,
      "grad_norm": 1.1223231170420371,
      "learning_rate": 2.005220919414497e-06,
      "loss": 0.4285,
      "step": 14511
    },
    {
      "epoch": 0.7831201770007016,
      "grad_norm": 0.9923094342799185,
      "learning_rate": 2.0047403682631405e-06,
      "loss": 0.3785,
      "step": 14512
    },
    {
      "epoch": 0.7831741406292159,
      "grad_norm": 1.1368355777241224,
      "learning_rate": 2.0042599175690784e-06,
      "loss": 0.3905,
      "step": 14513
    },
    {
      "epoch": 0.7832281042577303,
      "grad_norm": 0.876358998378452,
      "learning_rate": 2.003779567346118e-06,
      "loss": 0.3619,
      "step": 14514
    },
    {
      "epoch": 0.7832820678862447,
      "grad_norm": 0.8944103544261888,
      "learning_rate": 2.0032993176080645e-06,
      "loss": 0.3212,
      "step": 14515
    },
    {
      "epoch": 0.7833360315147591,
      "grad_norm": 0.9263869719107248,
      "learning_rate": 2.002819168368722e-06,
      "loss": 0.4233,
      "step": 14516
    },
    {
      "epoch": 0.7833899951432735,
      "grad_norm": 1.0650469051883267,
      "learning_rate": 2.0023391196418905e-06,
      "loss": 0.4235,
      "step": 14517
    },
    {
      "epoch": 0.7834439587717879,
      "grad_norm": 1.0336447862200027,
      "learning_rate": 2.001859171441367e-06,
      "loss": 0.5651,
      "step": 14518
    },
    {
      "epoch": 0.7834979224003022,
      "grad_norm": 1.2057059050511176,
      "learning_rate": 2.0013793237809454e-06,
      "loss": 0.4926,
      "step": 14519
    },
    {
      "epoch": 0.7835518860288165,
      "grad_norm": 0.9365287480066077,
      "learning_rate": 2.0008995766744173e-06,
      "loss": 0.3103,
      "step": 14520
    },
    {
      "epoch": 0.7836058496573309,
      "grad_norm": 1.0475193906281244,
      "learning_rate": 2.000419930135572e-06,
      "loss": 0.4171,
      "step": 14521
    },
    {
      "epoch": 0.7836598132858453,
      "grad_norm": 1.013503977196661,
      "learning_rate": 1.9999403841781925e-06,
      "loss": 0.494,
      "step": 14522
    },
    {
      "epoch": 0.7837137769143597,
      "grad_norm": 0.8250730115690996,
      "learning_rate": 1.999460938816063e-06,
      "loss": 0.3494,
      "step": 14523
    },
    {
      "epoch": 0.7837677405428741,
      "grad_norm": 1.0981165292556543,
      "learning_rate": 1.9989815940629635e-06,
      "loss": 0.4657,
      "step": 14524
    },
    {
      "epoch": 0.7838217041713885,
      "grad_norm": 0.9767573672555151,
      "learning_rate": 1.9985023499326705e-06,
      "loss": 0.434,
      "step": 14525
    },
    {
      "epoch": 0.7838756677999029,
      "grad_norm": 1.0417134922571631,
      "learning_rate": 1.9980232064389578e-06,
      "loss": 0.4772,
      "step": 14526
    },
    {
      "epoch": 0.7839296314284172,
      "grad_norm": 0.9349967649175002,
      "learning_rate": 1.9975441635955965e-06,
      "loss": 0.3744,
      "step": 14527
    },
    {
      "epoch": 0.7839835950569316,
      "grad_norm": 0.9041636491919524,
      "learning_rate": 1.9970652214163547e-06,
      "loss": 0.392,
      "step": 14528
    },
    {
      "epoch": 0.784037558685446,
      "grad_norm": 0.8995478858751499,
      "learning_rate": 1.9965863799149988e-06,
      "loss": 0.3891,
      "step": 14529
    },
    {
      "epoch": 0.7840915223139604,
      "grad_norm": 1.0090148456458585,
      "learning_rate": 1.9961076391052893e-06,
      "loss": 0.4327,
      "step": 14530
    },
    {
      "epoch": 0.7841454859424748,
      "grad_norm": 1.0011835593994496,
      "learning_rate": 1.995628999000987e-06,
      "loss": 0.4694,
      "step": 14531
    },
    {
      "epoch": 0.7841994495709892,
      "grad_norm": 1.0241981781107876,
      "learning_rate": 1.9951504596158484e-06,
      "loss": 0.4115,
      "step": 14532
    },
    {
      "epoch": 0.7842534131995035,
      "grad_norm": 0.8217138392426513,
      "learning_rate": 1.994672020963628e-06,
      "loss": 0.3208,
      "step": 14533
    },
    {
      "epoch": 0.7843073768280179,
      "grad_norm": 1.1553323282329495,
      "learning_rate": 1.9941936830580747e-06,
      "loss": 0.5156,
      "step": 14534
    },
    {
      "epoch": 0.7843613404565323,
      "grad_norm": 0.8293158794750655,
      "learning_rate": 1.9937154459129364e-06,
      "loss": 0.3394,
      "step": 14535
    },
    {
      "epoch": 0.7844153040850467,
      "grad_norm": 0.8342557247820733,
      "learning_rate": 1.99323730954196e-06,
      "loss": 0.292,
      "step": 14536
    },
    {
      "epoch": 0.7844692677135611,
      "grad_norm": 0.9156244096674335,
      "learning_rate": 1.9927592739588864e-06,
      "loss": 0.3896,
      "step": 14537
    },
    {
      "epoch": 0.7845232313420755,
      "grad_norm": 0.9287009469797889,
      "learning_rate": 1.992281339177455e-06,
      "loss": 0.436,
      "step": 14538
    },
    {
      "epoch": 0.7845771949705899,
      "grad_norm": 0.9713892905296607,
      "learning_rate": 1.991803505211402e-06,
      "loss": 0.511,
      "step": 14539
    },
    {
      "epoch": 0.7846311585991042,
      "grad_norm": 0.8738529260398595,
      "learning_rate": 1.991325772074463e-06,
      "loss": 0.4812,
      "step": 14540
    },
    {
      "epoch": 0.7846851222276185,
      "grad_norm": 1.069978642742901,
      "learning_rate": 1.990848139780364e-06,
      "loss": 0.7341,
      "step": 14541
    },
    {
      "epoch": 0.784739085856133,
      "grad_norm": 0.9230346084709934,
      "learning_rate": 1.9903706083428363e-06,
      "loss": 0.409,
      "step": 14542
    },
    {
      "epoch": 0.7847930494846473,
      "grad_norm": 0.8421839116379571,
      "learning_rate": 1.989893177775603e-06,
      "loss": 0.3569,
      "step": 14543
    },
    {
      "epoch": 0.7848470131131617,
      "grad_norm": 1.0410140526870304,
      "learning_rate": 1.989415848092387e-06,
      "loss": 0.4455,
      "step": 14544
    },
    {
      "epoch": 0.7849009767416761,
      "grad_norm": 0.8727437842261191,
      "learning_rate": 1.9889386193069064e-06,
      "loss": 0.4315,
      "step": 14545
    },
    {
      "epoch": 0.7849549403701905,
      "grad_norm": 0.9326318032478996,
      "learning_rate": 1.988461491432877e-06,
      "loss": 0.3895,
      "step": 14546
    },
    {
      "epoch": 0.7850089039987048,
      "grad_norm": 1.0320452812984966,
      "learning_rate": 1.9879844644840148e-06,
      "loss": 0.4316,
      "step": 14547
    },
    {
      "epoch": 0.7850628676272192,
      "grad_norm": 1.1123391946361239,
      "learning_rate": 1.9875075384740255e-06,
      "loss": 0.5044,
      "step": 14548
    },
    {
      "epoch": 0.7851168312557336,
      "grad_norm": 1.0418362989425352,
      "learning_rate": 1.9870307134166196e-06,
      "loss": 0.4159,
      "step": 14549
    },
    {
      "epoch": 0.785170794884248,
      "grad_norm": 0.9244888386789647,
      "learning_rate": 1.9865539893255e-06,
      "loss": 0.4579,
      "step": 14550
    },
    {
      "epoch": 0.7852247585127624,
      "grad_norm": 1.1458244696798083,
      "learning_rate": 1.986077366214369e-06,
      "loss": 0.4527,
      "step": 14551
    },
    {
      "epoch": 0.7852787221412768,
      "grad_norm": 0.9912633749272167,
      "learning_rate": 1.9856008440969253e-06,
      "loss": 0.3148,
      "step": 14552
    },
    {
      "epoch": 0.7853326857697912,
      "grad_norm": 0.9247364275283552,
      "learning_rate": 1.985124422986864e-06,
      "loss": 0.3358,
      "step": 14553
    },
    {
      "epoch": 0.7853866493983055,
      "grad_norm": 0.9927487479966528,
      "learning_rate": 1.9846481028978793e-06,
      "loss": 0.4986,
      "step": 14554
    },
    {
      "epoch": 0.7854406130268199,
      "grad_norm": 0.9813022766760525,
      "learning_rate": 1.98417188384366e-06,
      "loss": 0.4218,
      "step": 14555
    },
    {
      "epoch": 0.7854945766553343,
      "grad_norm": 0.9118419843640746,
      "learning_rate": 1.9836957658378917e-06,
      "loss": 0.3339,
      "step": 14556
    },
    {
      "epoch": 0.7855485402838487,
      "grad_norm": 0.9595747141682858,
      "learning_rate": 1.9832197488942614e-06,
      "loss": 0.4361,
      "step": 14557
    },
    {
      "epoch": 0.7856025039123631,
      "grad_norm": 1.0400634196970147,
      "learning_rate": 1.9827438330264475e-06,
      "loss": 0.4812,
      "step": 14558
    },
    {
      "epoch": 0.7856564675408775,
      "grad_norm": 0.9053376960437989,
      "learning_rate": 1.9822680182481306e-06,
      "loss": 0.3813,
      "step": 14559
    },
    {
      "epoch": 0.7857104311693919,
      "grad_norm": 0.9242120045080967,
      "learning_rate": 1.981792304572986e-06,
      "loss": 0.4672,
      "step": 14560
    },
    {
      "epoch": 0.7857643947979062,
      "grad_norm": 1.0026425104287817,
      "learning_rate": 1.9813166920146854e-06,
      "loss": 0.3997,
      "step": 14561
    },
    {
      "epoch": 0.7858183584264206,
      "grad_norm": 1.0164268815817612,
      "learning_rate": 1.980841180586898e-06,
      "loss": 0.5048,
      "step": 14562
    },
    {
      "epoch": 0.785872322054935,
      "grad_norm": 0.8824254190975734,
      "learning_rate": 1.980365770303291e-06,
      "loss": 0.3829,
      "step": 14563
    },
    {
      "epoch": 0.7859262856834494,
      "grad_norm": 1.1090674621178995,
      "learning_rate": 1.979890461177528e-06,
      "loss": 0.5719,
      "step": 14564
    },
    {
      "epoch": 0.7859802493119638,
      "grad_norm": 1.0277712093655211,
      "learning_rate": 1.9794152532232696e-06,
      "loss": 0.4221,
      "step": 14565
    },
    {
      "epoch": 0.7860342129404781,
      "grad_norm": 1.109636482810379,
      "learning_rate": 1.978940146454174e-06,
      "loss": 0.4797,
      "step": 14566
    },
    {
      "epoch": 0.7860881765689925,
      "grad_norm": 0.8719448094763242,
      "learning_rate": 1.9784651408838966e-06,
      "loss": 0.3701,
      "step": 14567
    },
    {
      "epoch": 0.7861421401975068,
      "grad_norm": 0.6917187417779371,
      "learning_rate": 1.9779902365260906e-06,
      "loss": 0.29,
      "step": 14568
    },
    {
      "epoch": 0.7861961038260212,
      "grad_norm": 1.1010768429165132,
      "learning_rate": 1.9775154333944024e-06,
      "loss": 0.4108,
      "step": 14569
    },
    {
      "epoch": 0.7862500674545356,
      "grad_norm": 1.177161036320019,
      "learning_rate": 1.977040731502481e-06,
      "loss": 0.5081,
      "step": 14570
    },
    {
      "epoch": 0.78630403108305,
      "grad_norm": 1.2740437649420884,
      "learning_rate": 1.9765661308639677e-06,
      "loss": 0.5229,
      "step": 14571
    },
    {
      "epoch": 0.7863579947115644,
      "grad_norm": 1.0715639090120188,
      "learning_rate": 1.9760916314925038e-06,
      "loss": 0.4769,
      "step": 14572
    },
    {
      "epoch": 0.7864119583400788,
      "grad_norm": 1.105726045225631,
      "learning_rate": 1.9756172334017273e-06,
      "loss": 0.5088,
      "step": 14573
    },
    {
      "epoch": 0.7864659219685932,
      "grad_norm": 1.0647358363333272,
      "learning_rate": 1.9751429366052726e-06,
      "loss": 0.389,
      "step": 14574
    },
    {
      "epoch": 0.7865198855971075,
      "grad_norm": 1.1353298651037742,
      "learning_rate": 1.9746687411167715e-06,
      "loss": 0.5701,
      "step": 14575
    },
    {
      "epoch": 0.7865738492256219,
      "grad_norm": 1.0274057492681332,
      "learning_rate": 1.974194646949853e-06,
      "loss": 0.4412,
      "step": 14576
    },
    {
      "epoch": 0.7866278128541363,
      "grad_norm": 1.0453580205384694,
      "learning_rate": 1.973720654118143e-06,
      "loss": 0.5148,
      "step": 14577
    },
    {
      "epoch": 0.7866817764826507,
      "grad_norm": 0.8991452673025677,
      "learning_rate": 1.9732467626352646e-06,
      "loss": 0.3408,
      "step": 14578
    },
    {
      "epoch": 0.7867357401111651,
      "grad_norm": 1.1501329993109375,
      "learning_rate": 1.972772972514837e-06,
      "loss": 0.427,
      "step": 14579
    },
    {
      "epoch": 0.7867897037396795,
      "grad_norm": 0.9872379786277704,
      "learning_rate": 1.972299283770479e-06,
      "loss": 0.4763,
      "step": 14580
    },
    {
      "epoch": 0.7868436673681939,
      "grad_norm": 1.1753246199241705,
      "learning_rate": 1.9718256964158033e-06,
      "loss": 0.5492,
      "step": 14581
    },
    {
      "epoch": 0.7868976309967082,
      "grad_norm": 0.9336711373710453,
      "learning_rate": 1.971352210464424e-06,
      "loss": 0.3532,
      "step": 14582
    },
    {
      "epoch": 0.7869515946252226,
      "grad_norm": 1.1399169765568866,
      "learning_rate": 1.9708788259299463e-06,
      "loss": 0.4481,
      "step": 14583
    },
    {
      "epoch": 0.787005558253737,
      "grad_norm": 1.0486413266759491,
      "learning_rate": 1.9704055428259763e-06,
      "loss": 0.437,
      "step": 14584
    },
    {
      "epoch": 0.7870595218822514,
      "grad_norm": 0.9555552968870438,
      "learning_rate": 1.969932361166118e-06,
      "loss": 0.4435,
      "step": 14585
    },
    {
      "epoch": 0.7871134855107658,
      "grad_norm": 1.001808400411889,
      "learning_rate": 1.9694592809639705e-06,
      "loss": 0.3734,
      "step": 14586
    },
    {
      "epoch": 0.7871674491392802,
      "grad_norm": 0.7375081586672189,
      "learning_rate": 1.9689863022331308e-06,
      "loss": 0.4185,
      "step": 14587
    },
    {
      "epoch": 0.7872214127677946,
      "grad_norm": 0.9213188442921666,
      "learning_rate": 1.968513424987192e-06,
      "loss": 0.4506,
      "step": 14588
    },
    {
      "epoch": 0.7872753763963088,
      "grad_norm": 1.0559839526868657,
      "learning_rate": 1.9680406492397476e-06,
      "loss": 0.505,
      "step": 14589
    },
    {
      "epoch": 0.7873293400248232,
      "grad_norm": 1.0419308833432812,
      "learning_rate": 1.9675679750043817e-06,
      "loss": 0.4701,
      "step": 14590
    },
    {
      "epoch": 0.7873833036533376,
      "grad_norm": 0.9616046469989242,
      "learning_rate": 1.9670954022946826e-06,
      "loss": 0.3328,
      "step": 14591
    },
    {
      "epoch": 0.787437267281852,
      "grad_norm": 0.9040819571329246,
      "learning_rate": 1.9666229311242303e-06,
      "loss": 0.3662,
      "step": 14592
    },
    {
      "epoch": 0.7874912309103664,
      "grad_norm": 1.1345561409696694,
      "learning_rate": 1.9661505615066058e-06,
      "loss": 0.5443,
      "step": 14593
    },
    {
      "epoch": 0.7875451945388808,
      "grad_norm": 1.1657201297365287,
      "learning_rate": 1.965678293455385e-06,
      "loss": 0.6008,
      "step": 14594
    },
    {
      "epoch": 0.7875991581673952,
      "grad_norm": 1.1236248936800637,
      "learning_rate": 1.96520612698414e-06,
      "loss": 0.4303,
      "step": 14595
    },
    {
      "epoch": 0.7876531217959095,
      "grad_norm": 1.1887382871394314,
      "learning_rate": 1.9647340621064455e-06,
      "loss": 0.5023,
      "step": 14596
    },
    {
      "epoch": 0.7877070854244239,
      "grad_norm": 1.1167323224158185,
      "learning_rate": 1.9642620988358635e-06,
      "loss": 0.51,
      "step": 14597
    },
    {
      "epoch": 0.7877610490529383,
      "grad_norm": 0.9725598658298157,
      "learning_rate": 1.963790237185962e-06,
      "loss": 0.4917,
      "step": 14598
    },
    {
      "epoch": 0.7878150126814527,
      "grad_norm": 1.1766063516941194,
      "learning_rate": 1.9633184771703022e-06,
      "loss": 0.4994,
      "step": 14599
    },
    {
      "epoch": 0.7878689763099671,
      "grad_norm": 0.9588054871739451,
      "learning_rate": 1.962846818802443e-06,
      "loss": 0.463,
      "step": 14600
    },
    {
      "epoch": 0.7879229399384815,
      "grad_norm": 1.053901011648955,
      "learning_rate": 1.96237526209594e-06,
      "loss": 0.5106,
      "step": 14601
    },
    {
      "epoch": 0.7879769035669958,
      "grad_norm": 1.1504303191946863,
      "learning_rate": 1.9619038070643476e-06,
      "loss": 0.4879,
      "step": 14602
    },
    {
      "epoch": 0.7880308671955102,
      "grad_norm": 1.0669827823038642,
      "learning_rate": 1.961432453721214e-06,
      "loss": 0.5072,
      "step": 14603
    },
    {
      "epoch": 0.7880848308240246,
      "grad_norm": 0.9782645001759421,
      "learning_rate": 1.9609612020800866e-06,
      "loss": 0.4439,
      "step": 14604
    },
    {
      "epoch": 0.788138794452539,
      "grad_norm": 1.359828050116991,
      "learning_rate": 1.9604900521545107e-06,
      "loss": 0.6595,
      "step": 14605
    },
    {
      "epoch": 0.7881927580810534,
      "grad_norm": 1.22670663652383,
      "learning_rate": 1.9600190039580267e-06,
      "loss": 0.5579,
      "step": 14606
    },
    {
      "epoch": 0.7882467217095678,
      "grad_norm": 0.9832801412819628,
      "learning_rate": 1.9595480575041737e-06,
      "loss": 0.4679,
      "step": 14607
    },
    {
      "epoch": 0.7883006853380822,
      "grad_norm": 1.063455423744853,
      "learning_rate": 1.959077212806487e-06,
      "loss": 0.521,
      "step": 14608
    },
    {
      "epoch": 0.7883546489665965,
      "grad_norm": 1.0184793325192882,
      "learning_rate": 1.9586064698785e-06,
      "loss": 0.3785,
      "step": 14609
    },
    {
      "epoch": 0.7884086125951109,
      "grad_norm": 0.9612397042067697,
      "learning_rate": 1.9581358287337405e-06,
      "loss": 0.5349,
      "step": 14610
    },
    {
      "epoch": 0.7884625762236253,
      "grad_norm": 1.0876314603954356,
      "learning_rate": 1.9576652893857355e-06,
      "loss": 0.6497,
      "step": 14611
    },
    {
      "epoch": 0.7885165398521397,
      "grad_norm": 0.8442428382866128,
      "learning_rate": 1.95719485184801e-06,
      "loss": 0.4138,
      "step": 14612
    },
    {
      "epoch": 0.788570503480654,
      "grad_norm": 1.0370312246105564,
      "learning_rate": 1.9567245161340834e-06,
      "loss": 0.4297,
      "step": 14613
    },
    {
      "epoch": 0.7886244671091684,
      "grad_norm": 0.9831280330808988,
      "learning_rate": 1.956254282257475e-06,
      "loss": 0.4283,
      "step": 14614
    },
    {
      "epoch": 0.7886784307376828,
      "grad_norm": 1.0810431846276434,
      "learning_rate": 1.955784150231699e-06,
      "loss": 0.5893,
      "step": 14615
    },
    {
      "epoch": 0.7887323943661971,
      "grad_norm": 1.1802345429070025,
      "learning_rate": 1.955314120070269e-06,
      "loss": 0.547,
      "step": 14616
    },
    {
      "epoch": 0.7887863579947115,
      "grad_norm": 1.069473762120721,
      "learning_rate": 1.954844191786692e-06,
      "loss": 0.3753,
      "step": 14617
    },
    {
      "epoch": 0.7888403216232259,
      "grad_norm": 1.0239668779978723,
      "learning_rate": 1.954374365394474e-06,
      "loss": 0.3447,
      "step": 14618
    },
    {
      "epoch": 0.7888942852517403,
      "grad_norm": 1.0831535143537088,
      "learning_rate": 1.9539046409071206e-06,
      "loss": 0.4251,
      "step": 14619
    },
    {
      "epoch": 0.7889482488802547,
      "grad_norm": 1.1492900822974768,
      "learning_rate": 1.95343501833813e-06,
      "loss": 0.5577,
      "step": 14620
    },
    {
      "epoch": 0.7890022125087691,
      "grad_norm": 0.7583297713124629,
      "learning_rate": 1.9529654977010005e-06,
      "loss": 0.323,
      "step": 14621
    },
    {
      "epoch": 0.7890561761372835,
      "grad_norm": 0.9096020058222228,
      "learning_rate": 1.9524960790092274e-06,
      "loss": 0.3826,
      "step": 14622
    },
    {
      "epoch": 0.7891101397657978,
      "grad_norm": 0.9778620177521072,
      "learning_rate": 1.9520267622763013e-06,
      "loss": 0.4296,
      "step": 14623
    },
    {
      "epoch": 0.7891641033943122,
      "grad_norm": 1.0741544767826348,
      "learning_rate": 1.9515575475157104e-06,
      "loss": 0.4401,
      "step": 14624
    },
    {
      "epoch": 0.7892180670228266,
      "grad_norm": 0.9765940157495852,
      "learning_rate": 1.9510884347409412e-06,
      "loss": 0.5735,
      "step": 14625
    },
    {
      "epoch": 0.789272030651341,
      "grad_norm": 1.2053500147056275,
      "learning_rate": 1.9506194239654764e-06,
      "loss": 0.4886,
      "step": 14626
    },
    {
      "epoch": 0.7893259942798554,
      "grad_norm": 1.057664835411804,
      "learning_rate": 1.950150515202797e-06,
      "loss": 0.5093,
      "step": 14627
    },
    {
      "epoch": 0.7893799579083698,
      "grad_norm": 0.9053075275254887,
      "learning_rate": 1.949681708466378e-06,
      "loss": 0.4229,
      "step": 14628
    },
    {
      "epoch": 0.7894339215368842,
      "grad_norm": 1.4660527909542647,
      "learning_rate": 1.949213003769693e-06,
      "loss": 0.5433,
      "step": 14629
    },
    {
      "epoch": 0.7894878851653985,
      "grad_norm": 0.8426924354439126,
      "learning_rate": 1.9487444011262148e-06,
      "loss": 0.3952,
      "step": 14630
    },
    {
      "epoch": 0.7895418487939129,
      "grad_norm": 0.8901930112317316,
      "learning_rate": 1.9482759005494104e-06,
      "loss": 0.3724,
      "step": 14631
    },
    {
      "epoch": 0.7895958124224273,
      "grad_norm": 1.015796746177723,
      "learning_rate": 1.9478075020527453e-06,
      "loss": 0.3919,
      "step": 14632
    },
    {
      "epoch": 0.7896497760509417,
      "grad_norm": 1.0053810531367111,
      "learning_rate": 1.9473392056496825e-06,
      "loss": 0.5391,
      "step": 14633
    },
    {
      "epoch": 0.7897037396794561,
      "grad_norm": 0.879850645077835,
      "learning_rate": 1.9468710113536802e-06,
      "loss": 0.3311,
      "step": 14634
    },
    {
      "epoch": 0.7897577033079705,
      "grad_norm": 1.1052677512858058,
      "learning_rate": 1.9464029191781948e-06,
      "loss": 0.446,
      "step": 14635
    },
    {
      "epoch": 0.7898116669364849,
      "grad_norm": 1.0155677497942495,
      "learning_rate": 1.9459349291366805e-06,
      "loss": 0.6197,
      "step": 14636
    },
    {
      "epoch": 0.7898656305649991,
      "grad_norm": 1.2018586687388273,
      "learning_rate": 1.945467041242589e-06,
      "loss": 0.4817,
      "step": 14637
    },
    {
      "epoch": 0.7899195941935135,
      "grad_norm": 1.0869570579438435,
      "learning_rate": 1.9449992555093645e-06,
      "loss": 0.5304,
      "step": 14638
    },
    {
      "epoch": 0.7899735578220279,
      "grad_norm": 0.9497907662467563,
      "learning_rate": 1.9445315719504536e-06,
      "loss": 0.4894,
      "step": 14639
    },
    {
      "epoch": 0.7900275214505423,
      "grad_norm": 1.0053172531095134,
      "learning_rate": 1.9440639905792983e-06,
      "loss": 0.4993,
      "step": 14640
    },
    {
      "epoch": 0.7900814850790567,
      "grad_norm": 1.0519826044040437,
      "learning_rate": 1.9435965114093364e-06,
      "loss": 0.4603,
      "step": 14641
    },
    {
      "epoch": 0.7901354487075711,
      "grad_norm": 1.121994285303685,
      "learning_rate": 1.9431291344540047e-06,
      "loss": 0.4758,
      "step": 14642
    },
    {
      "epoch": 0.7901894123360855,
      "grad_norm": 0.8946920116199533,
      "learning_rate": 1.9426618597267355e-06,
      "loss": 0.2618,
      "step": 14643
    },
    {
      "epoch": 0.7902433759645998,
      "grad_norm": 1.113220607357293,
      "learning_rate": 1.94219468724096e-06,
      "loss": 0.4654,
      "step": 14644
    },
    {
      "epoch": 0.7902973395931142,
      "grad_norm": 1.0580874674336034,
      "learning_rate": 1.941727617010103e-06,
      "loss": 0.4891,
      "step": 14645
    },
    {
      "epoch": 0.7903513032216286,
      "grad_norm": 1.0364647805909184,
      "learning_rate": 1.94126064904759e-06,
      "loss": 0.45,
      "step": 14646
    },
    {
      "epoch": 0.790405266850143,
      "grad_norm": 1.0490587399810778,
      "learning_rate": 1.9407937833668408e-06,
      "loss": 0.4759,
      "step": 14647
    },
    {
      "epoch": 0.7904592304786574,
      "grad_norm": 0.9432544641532278,
      "learning_rate": 1.940327019981275e-06,
      "loss": 0.3557,
      "step": 14648
    },
    {
      "epoch": 0.7905131941071718,
      "grad_norm": 1.1337849867019563,
      "learning_rate": 1.9398603589043074e-06,
      "loss": 0.4166,
      "step": 14649
    },
    {
      "epoch": 0.7905671577356862,
      "grad_norm": 0.9628411686784104,
      "learning_rate": 1.9393938001493514e-06,
      "loss": 0.4034,
      "step": 14650
    },
    {
      "epoch": 0.7906211213642005,
      "grad_norm": 1.111355875500459,
      "learning_rate": 1.938927343729815e-06,
      "loss": 0.4596,
      "step": 14651
    },
    {
      "epoch": 0.7906750849927149,
      "grad_norm": 0.9017676529933735,
      "learning_rate": 1.9384609896591043e-06,
      "loss": 0.4594,
      "step": 14652
    },
    {
      "epoch": 0.7907290486212293,
      "grad_norm": 0.8261720474104377,
      "learning_rate": 1.937994737950624e-06,
      "loss": 0.3802,
      "step": 14653
    },
    {
      "epoch": 0.7907830122497437,
      "grad_norm": 1.072229210797679,
      "learning_rate": 1.9375285886177738e-06,
      "loss": 0.52,
      "step": 14654
    },
    {
      "epoch": 0.7908369758782581,
      "grad_norm": 1.0910878308133074,
      "learning_rate": 1.9370625416739507e-06,
      "loss": 0.4756,
      "step": 14655
    },
    {
      "epoch": 0.7908909395067725,
      "grad_norm": 0.8586268695833523,
      "learning_rate": 1.9365965971325514e-06,
      "loss": 0.3749,
      "step": 14656
    },
    {
      "epoch": 0.7909449031352869,
      "grad_norm": 1.0057634185191764,
      "learning_rate": 1.936130755006966e-06,
      "loss": 0.4992,
      "step": 14657
    },
    {
      "epoch": 0.7909988667638012,
      "grad_norm": 1.6150398588663673,
      "learning_rate": 1.9356650153105847e-06,
      "loss": 0.4619,
      "step": 14658
    },
    {
      "epoch": 0.7910528303923156,
      "grad_norm": 1.0251777035348544,
      "learning_rate": 1.9351993780567912e-06,
      "loss": 0.4286,
      "step": 14659
    },
    {
      "epoch": 0.79110679402083,
      "grad_norm": 1.0231706273543215,
      "learning_rate": 1.9347338432589697e-06,
      "loss": 0.3845,
      "step": 14660
    },
    {
      "epoch": 0.7911607576493443,
      "grad_norm": 0.9315099261473602,
      "learning_rate": 1.9342684109305e-06,
      "loss": 0.4123,
      "step": 14661
    },
    {
      "epoch": 0.7912147212778587,
      "grad_norm": 1.0407359457055043,
      "learning_rate": 1.9338030810847593e-06,
      "loss": 0.6077,
      "step": 14662
    },
    {
      "epoch": 0.7912686849063731,
      "grad_norm": 0.9916866429737331,
      "learning_rate": 1.9333378537351206e-06,
      "loss": 0.3936,
      "step": 14663
    },
    {
      "epoch": 0.7913226485348875,
      "grad_norm": 0.888361047527817,
      "learning_rate": 1.9328727288949567e-06,
      "loss": 0.4563,
      "step": 14664
    },
    {
      "epoch": 0.7913766121634018,
      "grad_norm": 1.0274301388597396,
      "learning_rate": 1.9324077065776355e-06,
      "loss": 0.5094,
      "step": 14665
    },
    {
      "epoch": 0.7914305757919162,
      "grad_norm": 0.921856580276973,
      "learning_rate": 1.931942786796521e-06,
      "loss": 0.3809,
      "step": 14666
    },
    {
      "epoch": 0.7914845394204306,
      "grad_norm": 1.034561716415534,
      "learning_rate": 1.931477969564976e-06,
      "loss": 0.4305,
      "step": 14667
    },
    {
      "epoch": 0.791538503048945,
      "grad_norm": 0.9502745333496354,
      "learning_rate": 1.9310132548963593e-06,
      "loss": 0.3065,
      "step": 14668
    },
    {
      "epoch": 0.7915924666774594,
      "grad_norm": 1.1840242400956,
      "learning_rate": 1.9305486428040283e-06,
      "loss": 0.4666,
      "step": 14669
    },
    {
      "epoch": 0.7916464303059738,
      "grad_norm": 1.1169356083965438,
      "learning_rate": 1.930084133301336e-06,
      "loss": 0.427,
      "step": 14670
    },
    {
      "epoch": 0.7917003939344881,
      "grad_norm": 1.023123042642945,
      "learning_rate": 1.9296197264016337e-06,
      "loss": 0.3979,
      "step": 14671
    },
    {
      "epoch": 0.7917543575630025,
      "grad_norm": 1.0554826563828439,
      "learning_rate": 1.929155422118267e-06,
      "loss": 0.5007,
      "step": 14672
    },
    {
      "epoch": 0.7918083211915169,
      "grad_norm": 0.9737572176380674,
      "learning_rate": 1.9286912204645812e-06,
      "loss": 0.3973,
      "step": 14673
    },
    {
      "epoch": 0.7918622848200313,
      "grad_norm": 1.1623836894968198,
      "learning_rate": 1.928227121453919e-06,
      "loss": 0.4656,
      "step": 14674
    },
    {
      "epoch": 0.7919162484485457,
      "grad_norm": 0.9376393575868741,
      "learning_rate": 1.9277631250996182e-06,
      "loss": 0.3689,
      "step": 14675
    },
    {
      "epoch": 0.7919702120770601,
      "grad_norm": 1.04390223492826,
      "learning_rate": 1.9272992314150144e-06,
      "loss": 0.402,
      "step": 14676
    },
    {
      "epoch": 0.7920241757055745,
      "grad_norm": 1.0322256347024008,
      "learning_rate": 1.9268354404134407e-06,
      "loss": 0.4102,
      "step": 14677
    },
    {
      "epoch": 0.7920781393340888,
      "grad_norm": 1.0449109987877059,
      "learning_rate": 1.926371752108227e-06,
      "loss": 0.4433,
      "step": 14678
    },
    {
      "epoch": 0.7921321029626032,
      "grad_norm": 1.0368468534175435,
      "learning_rate": 1.9259081665127e-06,
      "loss": 0.5125,
      "step": 14679
    },
    {
      "epoch": 0.7921860665911176,
      "grad_norm": 1.0797972209112312,
      "learning_rate": 1.9254446836401835e-06,
      "loss": 0.3963,
      "step": 14680
    },
    {
      "epoch": 0.792240030219632,
      "grad_norm": 1.0909159040676193,
      "learning_rate": 1.9249813035039984e-06,
      "loss": 0.5694,
      "step": 14681
    },
    {
      "epoch": 0.7922939938481464,
      "grad_norm": 1.1645583006570448,
      "learning_rate": 1.9245180261174635e-06,
      "loss": 0.5176,
      "step": 14682
    },
    {
      "epoch": 0.7923479574766608,
      "grad_norm": 1.2057847053649595,
      "learning_rate": 1.9240548514938928e-06,
      "loss": 0.6152,
      "step": 14683
    },
    {
      "epoch": 0.7924019211051752,
      "grad_norm": 0.9080371031529645,
      "learning_rate": 1.923591779646599e-06,
      "loss": 0.3399,
      "step": 14684
    },
    {
      "epoch": 0.7924558847336894,
      "grad_norm": 1.1587000176992215,
      "learning_rate": 1.923128810588893e-06,
      "loss": 0.4928,
      "step": 14685
    },
    {
      "epoch": 0.7925098483622038,
      "grad_norm": 1.1940806224629577,
      "learning_rate": 1.9226659443340774e-06,
      "loss": 0.5321,
      "step": 14686
    },
    {
      "epoch": 0.7925638119907182,
      "grad_norm": 0.9433538820088282,
      "learning_rate": 1.922203180895457e-06,
      "loss": 0.5364,
      "step": 14687
    },
    {
      "epoch": 0.7926177756192326,
      "grad_norm": 0.9562633786115002,
      "learning_rate": 1.9217405202863325e-06,
      "loss": 0.3673,
      "step": 14688
    },
    {
      "epoch": 0.792671739247747,
      "grad_norm": 0.9207957651230827,
      "learning_rate": 1.9212779625200008e-06,
      "loss": 0.3171,
      "step": 14689
    },
    {
      "epoch": 0.7927257028762614,
      "grad_norm": 1.0041870101883708,
      "learning_rate": 1.920815507609757e-06,
      "loss": 0.4341,
      "step": 14690
    },
    {
      "epoch": 0.7927796665047758,
      "grad_norm": 1.0255983190137221,
      "learning_rate": 1.9203531555688913e-06,
      "loss": 0.5445,
      "step": 14691
    },
    {
      "epoch": 0.7928336301332901,
      "grad_norm": 1.1145455360977843,
      "learning_rate": 1.919890906410695e-06,
      "loss": 0.5018,
      "step": 14692
    },
    {
      "epoch": 0.7928875937618045,
      "grad_norm": 1.053575793297646,
      "learning_rate": 1.9194287601484496e-06,
      "loss": 0.6112,
      "step": 14693
    },
    {
      "epoch": 0.7929415573903189,
      "grad_norm": 1.129890802658404,
      "learning_rate": 1.9189667167954395e-06,
      "loss": 0.4163,
      "step": 14694
    },
    {
      "epoch": 0.7929955210188333,
      "grad_norm": 0.9971143942702119,
      "learning_rate": 1.918504776364945e-06,
      "loss": 0.4133,
      "step": 14695
    },
    {
      "epoch": 0.7930494846473477,
      "grad_norm": 0.828246794843553,
      "learning_rate": 1.9180429388702414e-06,
      "loss": 0.3685,
      "step": 14696
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 0.8479513596844016,
      "learning_rate": 1.9175812043246034e-06,
      "loss": 0.3271,
      "step": 14697
    },
    {
      "epoch": 0.7931574119043765,
      "grad_norm": 1.1243099086835642,
      "learning_rate": 1.917119572741301e-06,
      "loss": 0.5523,
      "step": 14698
    },
    {
      "epoch": 0.7932113755328908,
      "grad_norm": 0.9875972402353539,
      "learning_rate": 1.9166580441336043e-06,
      "loss": 0.5131,
      "step": 14699
    },
    {
      "epoch": 0.7932653391614052,
      "grad_norm": 1.0048136692105383,
      "learning_rate": 1.916196618514774e-06,
      "loss": 0.3896,
      "step": 14700
    },
    {
      "epoch": 0.7933193027899196,
      "grad_norm": 0.9156991604281818,
      "learning_rate": 1.9157352958980743e-06,
      "loss": 0.3503,
      "step": 14701
    },
    {
      "epoch": 0.793373266418434,
      "grad_norm": 0.9290362434969981,
      "learning_rate": 1.9152740762967636e-06,
      "loss": 0.4249,
      "step": 14702
    },
    {
      "epoch": 0.7934272300469484,
      "grad_norm": 0.8883630918949069,
      "learning_rate": 1.9148129597240984e-06,
      "loss": 0.3974,
      "step": 14703
    },
    {
      "epoch": 0.7934811936754628,
      "grad_norm": 1.0265765300504754,
      "learning_rate": 1.9143519461933313e-06,
      "loss": 0.5065,
      "step": 14704
    },
    {
      "epoch": 0.7935351573039772,
      "grad_norm": 0.9646740803721207,
      "learning_rate": 1.913891035717712e-06,
      "loss": 0.4677,
      "step": 14705
    },
    {
      "epoch": 0.7935891209324915,
      "grad_norm": 1.1173312613779238,
      "learning_rate": 1.9134302283104894e-06,
      "loss": 0.5185,
      "step": 14706
    },
    {
      "epoch": 0.7936430845610059,
      "grad_norm": 1.0480630540340863,
      "learning_rate": 1.9129695239849047e-06,
      "loss": 0.4234,
      "step": 14707
    },
    {
      "epoch": 0.7936970481895202,
      "grad_norm": 1.3204896184984523,
      "learning_rate": 1.9125089227542005e-06,
      "loss": 0.5089,
      "step": 14708
    },
    {
      "epoch": 0.7937510118180346,
      "grad_norm": 0.9922520433747511,
      "learning_rate": 1.9120484246316147e-06,
      "loss": 0.4104,
      "step": 14709
    },
    {
      "epoch": 0.793804975446549,
      "grad_norm": 1.1014923012566726,
      "learning_rate": 1.911588029630382e-06,
      "loss": 0.4784,
      "step": 14710
    },
    {
      "epoch": 0.7938589390750634,
      "grad_norm": 1.0634956849871278,
      "learning_rate": 1.9111277377637357e-06,
      "loss": 0.5374,
      "step": 14711
    },
    {
      "epoch": 0.7939129027035778,
      "grad_norm": 1.048762337801115,
      "learning_rate": 1.9106675490449043e-06,
      "loss": 0.3949,
      "step": 14712
    },
    {
      "epoch": 0.7939668663320921,
      "grad_norm": 1.3857732550325788,
      "learning_rate": 1.910207463487116e-06,
      "loss": 0.8371,
      "step": 14713
    },
    {
      "epoch": 0.7940208299606065,
      "grad_norm": 1.0843348717776715,
      "learning_rate": 1.9097474811035903e-06,
      "loss": 0.5509,
      "step": 14714
    },
    {
      "epoch": 0.7940747935891209,
      "grad_norm": 0.9699804361405676,
      "learning_rate": 1.90928760190755e-06,
      "loss": 0.4629,
      "step": 14715
    },
    {
      "epoch": 0.7941287572176353,
      "grad_norm": 1.409965799386098,
      "learning_rate": 1.908827825912213e-06,
      "loss": 0.482,
      "step": 14716
    },
    {
      "epoch": 0.7941827208461497,
      "grad_norm": 0.947238591271362,
      "learning_rate": 1.908368153130792e-06,
      "loss": 0.353,
      "step": 14717
    },
    {
      "epoch": 0.7942366844746641,
      "grad_norm": 1.36389969610373,
      "learning_rate": 1.9079085835764994e-06,
      "loss": 0.6026,
      "step": 14718
    },
    {
      "epoch": 0.7942906481031785,
      "grad_norm": 1.1519277611316663,
      "learning_rate": 1.9074491172625437e-06,
      "loss": 0.5681,
      "step": 14719
    },
    {
      "epoch": 0.7943446117316928,
      "grad_norm": 1.1076512005633572,
      "learning_rate": 1.9069897542021317e-06,
      "loss": 0.4524,
      "step": 14720
    },
    {
      "epoch": 0.7943985753602072,
      "grad_norm": 1.017848421325392,
      "learning_rate": 1.9065304944084634e-06,
      "loss": 0.5098,
      "step": 14721
    },
    {
      "epoch": 0.7944525389887216,
      "grad_norm": 0.937765737512193,
      "learning_rate": 1.90607133789474e-06,
      "loss": 0.3411,
      "step": 14722
    },
    {
      "epoch": 0.794506502617236,
      "grad_norm": 1.073202808900377,
      "learning_rate": 1.9056122846741577e-06,
      "loss": 0.5777,
      "step": 14723
    },
    {
      "epoch": 0.7945604662457504,
      "grad_norm": 1.1296825422326133,
      "learning_rate": 1.9051533347599094e-06,
      "loss": 0.402,
      "step": 14724
    },
    {
      "epoch": 0.7946144298742648,
      "grad_norm": 0.8173768052235465,
      "learning_rate": 1.9046944881651874e-06,
      "loss": 0.2975,
      "step": 14725
    },
    {
      "epoch": 0.7946683935027792,
      "grad_norm": 1.1794373399455296,
      "learning_rate": 1.9042357449031785e-06,
      "loss": 0.6981,
      "step": 14726
    },
    {
      "epoch": 0.7947223571312935,
      "grad_norm": 1.1123641890988822,
      "learning_rate": 1.9037771049870668e-06,
      "loss": 0.5596,
      "step": 14727
    },
    {
      "epoch": 0.7947763207598079,
      "grad_norm": 1.2357123113849133,
      "learning_rate": 1.903318568430036e-06,
      "loss": 0.4851,
      "step": 14728
    },
    {
      "epoch": 0.7948302843883223,
      "grad_norm": 1.1441593422059406,
      "learning_rate": 1.9028601352452628e-06,
      "loss": 0.5436,
      "step": 14729
    },
    {
      "epoch": 0.7948842480168367,
      "grad_norm": 0.9781521793624168,
      "learning_rate": 1.9024018054459246e-06,
      "loss": 0.4038,
      "step": 14730
    },
    {
      "epoch": 0.7949382116453511,
      "grad_norm": 1.2311397131124744,
      "learning_rate": 1.901943579045193e-06,
      "loss": 0.6452,
      "step": 14731
    },
    {
      "epoch": 0.7949921752738655,
      "grad_norm": 0.9010140984652273,
      "learning_rate": 1.901485456056239e-06,
      "loss": 0.391,
      "step": 14732
    },
    {
      "epoch": 0.7950461389023799,
      "grad_norm": 0.9463505220214544,
      "learning_rate": 1.9010274364922298e-06,
      "loss": 0.4638,
      "step": 14733
    },
    {
      "epoch": 0.7951001025308941,
      "grad_norm": 0.9936048123130422,
      "learning_rate": 1.9005695203663276e-06,
      "loss": 0.3912,
      "step": 14734
    },
    {
      "epoch": 0.7951540661594085,
      "grad_norm": 1.33041783630449,
      "learning_rate": 1.9001117076916941e-06,
      "loss": 0.6522,
      "step": 14735
    },
    {
      "epoch": 0.7952080297879229,
      "grad_norm": 1.0508493125225786,
      "learning_rate": 1.8996539984814882e-06,
      "loss": 0.4,
      "step": 14736
    },
    {
      "epoch": 0.7952619934164373,
      "grad_norm": 1.0494029222890209,
      "learning_rate": 1.8991963927488635e-06,
      "loss": 0.4771,
      "step": 14737
    },
    {
      "epoch": 0.7953159570449517,
      "grad_norm": 1.0545694244403985,
      "learning_rate": 1.8987388905069731e-06,
      "loss": 0.5372,
      "step": 14738
    },
    {
      "epoch": 0.7953699206734661,
      "grad_norm": 0.9671201359390122,
      "learning_rate": 1.8982814917689664e-06,
      "loss": 0.4741,
      "step": 14739
    },
    {
      "epoch": 0.7954238843019804,
      "grad_norm": 0.8776667955522179,
      "learning_rate": 1.8978241965479894e-06,
      "loss": 0.374,
      "step": 14740
    },
    {
      "epoch": 0.7954778479304948,
      "grad_norm": 1.1490784023313267,
      "learning_rate": 1.8973670048571836e-06,
      "loss": 0.4851,
      "step": 14741
    },
    {
      "epoch": 0.7955318115590092,
      "grad_norm": 0.954857012397209,
      "learning_rate": 1.89690991670969e-06,
      "loss": 0.4751,
      "step": 14742
    },
    {
      "epoch": 0.7955857751875236,
      "grad_norm": 0.9630626942063284,
      "learning_rate": 1.8964529321186464e-06,
      "loss": 0.3477,
      "step": 14743
    },
    {
      "epoch": 0.795639738816038,
      "grad_norm": 1.0151186509726797,
      "learning_rate": 1.895996051097187e-06,
      "loss": 0.4953,
      "step": 14744
    },
    {
      "epoch": 0.7956937024445524,
      "grad_norm": 0.8379192698790362,
      "learning_rate": 1.8955392736584425e-06,
      "loss": 0.3317,
      "step": 14745
    },
    {
      "epoch": 0.7957476660730668,
      "grad_norm": 1.0851161359368682,
      "learning_rate": 1.895082599815541e-06,
      "loss": 0.5159,
      "step": 14746
    },
    {
      "epoch": 0.7958016297015811,
      "grad_norm": 1.2478104662804392,
      "learning_rate": 1.8946260295816094e-06,
      "loss": 0.4625,
      "step": 14747
    },
    {
      "epoch": 0.7958555933300955,
      "grad_norm": 0.7742887885056116,
      "learning_rate": 1.8941695629697673e-06,
      "loss": 0.3163,
      "step": 14748
    },
    {
      "epoch": 0.7959095569586099,
      "grad_norm": 1.1622273708228705,
      "learning_rate": 1.8937131999931356e-06,
      "loss": 0.5126,
      "step": 14749
    },
    {
      "epoch": 0.7959635205871243,
      "grad_norm": 0.9005113732268086,
      "learning_rate": 1.8932569406648305e-06,
      "loss": 0.4271,
      "step": 14750
    },
    {
      "epoch": 0.7960174842156387,
      "grad_norm": 0.9804885453601415,
      "learning_rate": 1.8928007849979655e-06,
      "loss": 0.4355,
      "step": 14751
    },
    {
      "epoch": 0.7960714478441531,
      "grad_norm": 1.0436313548469127,
      "learning_rate": 1.8923447330056502e-06,
      "loss": 0.4373,
      "step": 14752
    },
    {
      "epoch": 0.7961254114726675,
      "grad_norm": 0.9109588063388007,
      "learning_rate": 1.891888784700993e-06,
      "loss": 0.426,
      "step": 14753
    },
    {
      "epoch": 0.7961793751011818,
      "grad_norm": 1.0762700531991114,
      "learning_rate": 1.8914329400970988e-06,
      "loss": 0.4699,
      "step": 14754
    },
    {
      "epoch": 0.7962333387296961,
      "grad_norm": 0.8861200813844414,
      "learning_rate": 1.890977199207067e-06,
      "loss": 0.3702,
      "step": 14755
    },
    {
      "epoch": 0.7962873023582105,
      "grad_norm": 0.9524228528688526,
      "learning_rate": 1.8905215620439966e-06,
      "loss": 0.4173,
      "step": 14756
    },
    {
      "epoch": 0.7963412659867249,
      "grad_norm": 1.003601682473795,
      "learning_rate": 1.8900660286209843e-06,
      "loss": 0.3874,
      "step": 14757
    },
    {
      "epoch": 0.7963952296152393,
      "grad_norm": 0.9770116603968483,
      "learning_rate": 1.8896105989511214e-06,
      "loss": 0.3487,
      "step": 14758
    },
    {
      "epoch": 0.7964491932437537,
      "grad_norm": 1.0921888857640392,
      "learning_rate": 1.8891552730474986e-06,
      "loss": 0.5435,
      "step": 14759
    },
    {
      "epoch": 0.7965031568722681,
      "grad_norm": 0.8985530295698759,
      "learning_rate": 1.888700050923201e-06,
      "loss": 0.3803,
      "step": 14760
    },
    {
      "epoch": 0.7965571205007824,
      "grad_norm": 1.0272766593325036,
      "learning_rate": 1.8882449325913149e-06,
      "loss": 0.4374,
      "step": 14761
    },
    {
      "epoch": 0.7966110841292968,
      "grad_norm": 1.196360621861719,
      "learning_rate": 1.8877899180649168e-06,
      "loss": 0.6302,
      "step": 14762
    },
    {
      "epoch": 0.7966650477578112,
      "grad_norm": 0.9966199547257808,
      "learning_rate": 1.8873350073570868e-06,
      "loss": 0.3572,
      "step": 14763
    },
    {
      "epoch": 0.7967190113863256,
      "grad_norm": 1.1335156988378492,
      "learning_rate": 1.886880200480899e-06,
      "loss": 0.4459,
      "step": 14764
    },
    {
      "epoch": 0.79677297501484,
      "grad_norm": 1.1490114188716736,
      "learning_rate": 1.886425497449425e-06,
      "loss": 0.4226,
      "step": 14765
    },
    {
      "epoch": 0.7968269386433544,
      "grad_norm": 1.0059514956825795,
      "learning_rate": 1.8859708982757335e-06,
      "loss": 0.4157,
      "step": 14766
    },
    {
      "epoch": 0.7968809022718688,
      "grad_norm": 0.7956385282690904,
      "learning_rate": 1.8855164029728902e-06,
      "loss": 0.4154,
      "step": 14767
    },
    {
      "epoch": 0.7969348659003831,
      "grad_norm": 1.027229258560275,
      "learning_rate": 1.8850620115539586e-06,
      "loss": 0.4882,
      "step": 14768
    },
    {
      "epoch": 0.7969888295288975,
      "grad_norm": 0.9200074034127403,
      "learning_rate": 1.884607724031996e-06,
      "loss": 0.3864,
      "step": 14769
    },
    {
      "epoch": 0.7970427931574119,
      "grad_norm": 0.887075731931406,
      "learning_rate": 1.8841535404200605e-06,
      "loss": 0.3955,
      "step": 14770
    },
    {
      "epoch": 0.7970967567859263,
      "grad_norm": 1.2438792107003147,
      "learning_rate": 1.8836994607312065e-06,
      "loss": 0.5332,
      "step": 14771
    },
    {
      "epoch": 0.7971507204144407,
      "grad_norm": 1.0307681574932173,
      "learning_rate": 1.8832454849784836e-06,
      "loss": 0.4218,
      "step": 14772
    },
    {
      "epoch": 0.7972046840429551,
      "grad_norm": 0.9959415972932533,
      "learning_rate": 1.8827916131749396e-06,
      "loss": 0.4544,
      "step": 14773
    },
    {
      "epoch": 0.7972586476714695,
      "grad_norm": 1.1420035912272635,
      "learning_rate": 1.88233784533362e-06,
      "loss": 0.5161,
      "step": 14774
    },
    {
      "epoch": 0.7973126112999838,
      "grad_norm": 0.9139977786354526,
      "learning_rate": 1.8818841814675658e-06,
      "loss": 0.384,
      "step": 14775
    },
    {
      "epoch": 0.7973665749284982,
      "grad_norm": 1.1253187578119446,
      "learning_rate": 1.8814306215898165e-06,
      "loss": 0.4422,
      "step": 14776
    },
    {
      "epoch": 0.7974205385570126,
      "grad_norm": 0.8713162778267493,
      "learning_rate": 1.8809771657134086e-06,
      "loss": 0.5021,
      "step": 14777
    },
    {
      "epoch": 0.797474502185527,
      "grad_norm": 1.1005253140300735,
      "learning_rate": 1.880523813851372e-06,
      "loss": 0.5531,
      "step": 14778
    },
    {
      "epoch": 0.7975284658140414,
      "grad_norm": 1.1386980644003926,
      "learning_rate": 1.8800705660167384e-06,
      "loss": 0.5642,
      "step": 14779
    },
    {
      "epoch": 0.7975824294425558,
      "grad_norm": 1.1828925739550857,
      "learning_rate": 1.8796174222225345e-06,
      "loss": 0.5738,
      "step": 14780
    },
    {
      "epoch": 0.7976363930710701,
      "grad_norm": 0.9292486094603751,
      "learning_rate": 1.879164382481784e-06,
      "loss": 0.3698,
      "step": 14781
    },
    {
      "epoch": 0.7976903566995844,
      "grad_norm": 1.0984978322483399,
      "learning_rate": 1.878711446807508e-06,
      "loss": 0.4803,
      "step": 14782
    },
    {
      "epoch": 0.7977443203280988,
      "grad_norm": 1.0808667381546153,
      "learning_rate": 1.8782586152127236e-06,
      "loss": 0.5648,
      "step": 14783
    },
    {
      "epoch": 0.7977982839566132,
      "grad_norm": 1.0103599267923162,
      "learning_rate": 1.877805887710446e-06,
      "loss": 0.4267,
      "step": 14784
    },
    {
      "epoch": 0.7978522475851276,
      "grad_norm": 1.127591519733452,
      "learning_rate": 1.8773532643136877e-06,
      "loss": 0.5792,
      "step": 14785
    },
    {
      "epoch": 0.797906211213642,
      "grad_norm": 0.9309712047612532,
      "learning_rate": 1.876900745035457e-06,
      "loss": 0.5485,
      "step": 14786
    },
    {
      "epoch": 0.7979601748421564,
      "grad_norm": 1.128324876273577,
      "learning_rate": 1.8764483298887593e-06,
      "loss": 0.5225,
      "step": 14787
    },
    {
      "epoch": 0.7980141384706708,
      "grad_norm": 0.884910801788974,
      "learning_rate": 1.8759960188865984e-06,
      "loss": 0.4142,
      "step": 14788
    },
    {
      "epoch": 0.7980681020991851,
      "grad_norm": 0.9471320499228437,
      "learning_rate": 1.8755438120419745e-06,
      "loss": 0.4434,
      "step": 14789
    },
    {
      "epoch": 0.7981220657276995,
      "grad_norm": 0.8134948222090379,
      "learning_rate": 1.8750917093678824e-06,
      "loss": 0.3054,
      "step": 14790
    },
    {
      "epoch": 0.7981760293562139,
      "grad_norm": 0.9682750170772254,
      "learning_rate": 1.8746397108773178e-06,
      "loss": 0.4417,
      "step": 14791
    },
    {
      "epoch": 0.7982299929847283,
      "grad_norm": 1.036577158250916,
      "learning_rate": 1.8741878165832712e-06,
      "loss": 0.5362,
      "step": 14792
    },
    {
      "epoch": 0.7982839566132427,
      "grad_norm": 1.163347762033495,
      "learning_rate": 1.8737360264987292e-06,
      "loss": 0.5545,
      "step": 14793
    },
    {
      "epoch": 0.7983379202417571,
      "grad_norm": 1.0995744107310832,
      "learning_rate": 1.8732843406366792e-06,
      "loss": 0.4869,
      "step": 14794
    },
    {
      "epoch": 0.7983918838702715,
      "grad_norm": 1.0229166415700088,
      "learning_rate": 1.8728327590101009e-06,
      "loss": 0.4744,
      "step": 14795
    },
    {
      "epoch": 0.7984458474987858,
      "grad_norm": 0.9607900755609097,
      "learning_rate": 1.8723812816319753e-06,
      "loss": 0.3749,
      "step": 14796
    },
    {
      "epoch": 0.7984998111273002,
      "grad_norm": 1.1734748183851609,
      "learning_rate": 1.8719299085152767e-06,
      "loss": 0.4903,
      "step": 14797
    },
    {
      "epoch": 0.7985537747558146,
      "grad_norm": 1.02817619512855,
      "learning_rate": 1.8714786396729775e-06,
      "loss": 0.3206,
      "step": 14798
    },
    {
      "epoch": 0.798607738384329,
      "grad_norm": 1.0437421904178203,
      "learning_rate": 1.871027475118049e-06,
      "loss": 0.4057,
      "step": 14799
    },
    {
      "epoch": 0.7986617020128434,
      "grad_norm": 0.9650162500973755,
      "learning_rate": 1.8705764148634581e-06,
      "loss": 0.3601,
      "step": 14800
    },
    {
      "epoch": 0.7987156656413578,
      "grad_norm": 0.8539200665972264,
      "learning_rate": 1.8701254589221674e-06,
      "loss": 0.3216,
      "step": 14801
    },
    {
      "epoch": 0.7987696292698722,
      "grad_norm": 0.9478621133524248,
      "learning_rate": 1.8696746073071394e-06,
      "loss": 0.4205,
      "step": 14802
    },
    {
      "epoch": 0.7988235928983864,
      "grad_norm": 1.2313558330083039,
      "learning_rate": 1.8692238600313324e-06,
      "loss": 0.4798,
      "step": 14803
    },
    {
      "epoch": 0.7988775565269008,
      "grad_norm": 0.9485377490832243,
      "learning_rate": 1.868773217107699e-06,
      "loss": 0.4027,
      "step": 14804
    },
    {
      "epoch": 0.7989315201554152,
      "grad_norm": 1.022802011819007,
      "learning_rate": 1.8683226785491923e-06,
      "loss": 0.5217,
      "step": 14805
    },
    {
      "epoch": 0.7989854837839296,
      "grad_norm": 1.1219222287987456,
      "learning_rate": 1.867872244368762e-06,
      "loss": 0.4711,
      "step": 14806
    },
    {
      "epoch": 0.799039447412444,
      "grad_norm": 0.7372547253334664,
      "learning_rate": 1.8674219145793525e-06,
      "loss": 0.3157,
      "step": 14807
    },
    {
      "epoch": 0.7990934110409584,
      "grad_norm": 1.0320274121862387,
      "learning_rate": 1.8669716891939083e-06,
      "loss": 0.4678,
      "step": 14808
    },
    {
      "epoch": 0.7991473746694727,
      "grad_norm": 1.2294591615519743,
      "learning_rate": 1.8665215682253696e-06,
      "loss": 0.5302,
      "step": 14809
    },
    {
      "epoch": 0.7992013382979871,
      "grad_norm": 0.9773987588766254,
      "learning_rate": 1.8660715516866706e-06,
      "loss": 0.613,
      "step": 14810
    },
    {
      "epoch": 0.7992553019265015,
      "grad_norm": 1.2147389764750902,
      "learning_rate": 1.865621639590748e-06,
      "loss": 0.5757,
      "step": 14811
    },
    {
      "epoch": 0.7993092655550159,
      "grad_norm": 0.9401454724725238,
      "learning_rate": 1.8651718319505312e-06,
      "loss": 0.4048,
      "step": 14812
    },
    {
      "epoch": 0.7993632291835303,
      "grad_norm": 0.9517836672760154,
      "learning_rate": 1.8647221287789488e-06,
      "loss": 0.5221,
      "step": 14813
    },
    {
      "epoch": 0.7994171928120447,
      "grad_norm": 1.110762375303299,
      "learning_rate": 1.864272530088926e-06,
      "loss": 0.6151,
      "step": 14814
    },
    {
      "epoch": 0.7994711564405591,
      "grad_norm": 1.0193248049048989,
      "learning_rate": 1.863823035893384e-06,
      "loss": 0.6308,
      "step": 14815
    },
    {
      "epoch": 0.7995251200690734,
      "grad_norm": 1.05604216479492,
      "learning_rate": 1.8633736462052432e-06,
      "loss": 0.4483,
      "step": 14816
    },
    {
      "epoch": 0.7995790836975878,
      "grad_norm": 1.074631013728031,
      "learning_rate": 1.862924361037417e-06,
      "loss": 0.4908,
      "step": 14817
    },
    {
      "epoch": 0.7996330473261022,
      "grad_norm": 1.1984826118264553,
      "learning_rate": 1.8624751804028196e-06,
      "loss": 0.4117,
      "step": 14818
    },
    {
      "epoch": 0.7996870109546166,
      "grad_norm": 0.9798695348898983,
      "learning_rate": 1.8620261043143616e-06,
      "loss": 0.4278,
      "step": 14819
    },
    {
      "epoch": 0.799740974583131,
      "grad_norm": 1.0197562633385497,
      "learning_rate": 1.8615771327849486e-06,
      "loss": 0.4843,
      "step": 14820
    },
    {
      "epoch": 0.7997949382116454,
      "grad_norm": 1.360094125839444,
      "learning_rate": 1.8611282658274853e-06,
      "loss": 0.7185,
      "step": 14821
    },
    {
      "epoch": 0.7998489018401598,
      "grad_norm": 1.1174267692521802,
      "learning_rate": 1.8606795034548727e-06,
      "loss": 0.5623,
      "step": 14822
    },
    {
      "epoch": 0.7999028654686741,
      "grad_norm": 1.0881951461208044,
      "learning_rate": 1.8602308456800095e-06,
      "loss": 0.542,
      "step": 14823
    },
    {
      "epoch": 0.7999568290971885,
      "grad_norm": 0.9855754390166537,
      "learning_rate": 1.8597822925157883e-06,
      "loss": 0.555,
      "step": 14824
    },
    {
      "epoch": 0.8000107927257029,
      "grad_norm": 1.1022811016941738,
      "learning_rate": 1.8593338439751022e-06,
      "loss": 0.4787,
      "step": 14825
    },
    {
      "epoch": 0.8000647563542173,
      "grad_norm": 1.245202377260087,
      "learning_rate": 1.8588855000708397e-06,
      "loss": 0.505,
      "step": 14826
    },
    {
      "epoch": 0.8001187199827317,
      "grad_norm": 1.0709447889774986,
      "learning_rate": 1.8584372608158875e-06,
      "loss": 0.4763,
      "step": 14827
    },
    {
      "epoch": 0.800172683611246,
      "grad_norm": 1.1800619612185912,
      "learning_rate": 1.8579891262231275e-06,
      "loss": 0.5308,
      "step": 14828
    },
    {
      "epoch": 0.8002266472397604,
      "grad_norm": 0.9204724358070794,
      "learning_rate": 1.8575410963054407e-06,
      "loss": 0.2741,
      "step": 14829
    },
    {
      "epoch": 0.8002806108682747,
      "grad_norm": 0.994022041402739,
      "learning_rate": 1.8570931710757028e-06,
      "loss": 0.518,
      "step": 14830
    },
    {
      "epoch": 0.8003345744967891,
      "grad_norm": 0.9422201697703397,
      "learning_rate": 1.856645350546788e-06,
      "loss": 0.4082,
      "step": 14831
    },
    {
      "epoch": 0.8003885381253035,
      "grad_norm": 0.982440631638376,
      "learning_rate": 1.8561976347315675e-06,
      "loss": 0.429,
      "step": 14832
    },
    {
      "epoch": 0.8004425017538179,
      "grad_norm": 1.0557796318425101,
      "learning_rate": 1.8557500236429086e-06,
      "loss": 0.4192,
      "step": 14833
    },
    {
      "epoch": 0.8004964653823323,
      "grad_norm": 0.9481502418370241,
      "learning_rate": 1.8553025172936762e-06,
      "loss": 0.3895,
      "step": 14834
    },
    {
      "epoch": 0.8005504290108467,
      "grad_norm": 1.0170468570929492,
      "learning_rate": 1.8548551156967318e-06,
      "loss": 0.4005,
      "step": 14835
    },
    {
      "epoch": 0.8006043926393611,
      "grad_norm": 1.1047325988975143,
      "learning_rate": 1.854407818864935e-06,
      "loss": 0.5188,
      "step": 14836
    },
    {
      "epoch": 0.8006583562678754,
      "grad_norm": 1.031524217695675,
      "learning_rate": 1.853960626811142e-06,
      "loss": 0.485,
      "step": 14837
    },
    {
      "epoch": 0.8007123198963898,
      "grad_norm": 1.162140224190474,
      "learning_rate": 1.8535135395482038e-06,
      "loss": 0.4989,
      "step": 14838
    },
    {
      "epoch": 0.8007662835249042,
      "grad_norm": 1.2349089712846228,
      "learning_rate": 1.8530665570889704e-06,
      "loss": 0.4955,
      "step": 14839
    },
    {
      "epoch": 0.8008202471534186,
      "grad_norm": 1.0767333385297704,
      "learning_rate": 1.8526196794462898e-06,
      "loss": 0.5216,
      "step": 14840
    },
    {
      "epoch": 0.800874210781933,
      "grad_norm": 1.0481536421062496,
      "learning_rate": 1.8521729066330047e-06,
      "loss": 0.4008,
      "step": 14841
    },
    {
      "epoch": 0.8009281744104474,
      "grad_norm": 0.9619177211874119,
      "learning_rate": 1.8517262386619561e-06,
      "loss": 0.395,
      "step": 14842
    },
    {
      "epoch": 0.8009821380389618,
      "grad_norm": 1.1224281469708115,
      "learning_rate": 1.8512796755459816e-06,
      "loss": 0.4083,
      "step": 14843
    },
    {
      "epoch": 0.8010361016674761,
      "grad_norm": 0.9164417573962617,
      "learning_rate": 1.8508332172979172e-06,
      "loss": 0.4138,
      "step": 14844
    },
    {
      "epoch": 0.8010900652959905,
      "grad_norm": 1.1474931795160652,
      "learning_rate": 1.850386863930592e-06,
      "loss": 0.3693,
      "step": 14845
    },
    {
      "epoch": 0.8011440289245049,
      "grad_norm": 1.083078399936675,
      "learning_rate": 1.8499406154568361e-06,
      "loss": 0.5743,
      "step": 14846
    },
    {
      "epoch": 0.8011979925530193,
      "grad_norm": 1.0840862623104552,
      "learning_rate": 1.8494944718894756e-06,
      "loss": 0.4105,
      "step": 14847
    },
    {
      "epoch": 0.8012519561815337,
      "grad_norm": 0.6091667664591182,
      "learning_rate": 1.8490484332413322e-06,
      "loss": 0.1826,
      "step": 14848
    },
    {
      "epoch": 0.8013059198100481,
      "grad_norm": 1.0085393137402148,
      "learning_rate": 1.848602499525226e-06,
      "loss": 0.5533,
      "step": 14849
    },
    {
      "epoch": 0.8013598834385625,
      "grad_norm": 1.2522589257185284,
      "learning_rate": 1.8481566707539727e-06,
      "loss": 0.6378,
      "step": 14850
    },
    {
      "epoch": 0.8014138470670767,
      "grad_norm": 1.0408550182451914,
      "learning_rate": 1.847710946940388e-06,
      "loss": 0.3927,
      "step": 14851
    },
    {
      "epoch": 0.8014678106955911,
      "grad_norm": 0.984302905834938,
      "learning_rate": 1.8472653280972808e-06,
      "loss": 0.4355,
      "step": 14852
    },
    {
      "epoch": 0.8015217743241055,
      "grad_norm": 1.0100942959360883,
      "learning_rate": 1.846819814237458e-06,
      "loss": 0.443,
      "step": 14853
    },
    {
      "epoch": 0.8015757379526199,
      "grad_norm": 0.946854540615051,
      "learning_rate": 1.8463744053737254e-06,
      "loss": 0.387,
      "step": 14854
    },
    {
      "epoch": 0.8016297015811343,
      "grad_norm": 0.8768283991857462,
      "learning_rate": 1.8459291015188841e-06,
      "loss": 0.4245,
      "step": 14855
    },
    {
      "epoch": 0.8016836652096487,
      "grad_norm": 0.8946967481900465,
      "learning_rate": 1.845483902685732e-06,
      "loss": 0.3883,
      "step": 14856
    },
    {
      "epoch": 0.8017376288381631,
      "grad_norm": 0.940535041229304,
      "learning_rate": 1.8450388088870657e-06,
      "loss": 0.3987,
      "step": 14857
    },
    {
      "epoch": 0.8017915924666774,
      "grad_norm": 1.221502934085716,
      "learning_rate": 1.8445938201356783e-06,
      "loss": 0.4223,
      "step": 14858
    },
    {
      "epoch": 0.8018455560951918,
      "grad_norm": 1.2224243487987394,
      "learning_rate": 1.8441489364443568e-06,
      "loss": 0.7295,
      "step": 14859
    },
    {
      "epoch": 0.8018995197237062,
      "grad_norm": 1.0806682871089455,
      "learning_rate": 1.843704157825889e-06,
      "loss": 0.4475,
      "step": 14860
    },
    {
      "epoch": 0.8019534833522206,
      "grad_norm": 1.0541236266539329,
      "learning_rate": 1.843259484293058e-06,
      "loss": 0.4467,
      "step": 14861
    },
    {
      "epoch": 0.802007446980735,
      "grad_norm": 1.6573794761381317,
      "learning_rate": 1.8428149158586438e-06,
      "loss": 0.4755,
      "step": 14862
    },
    {
      "epoch": 0.8020614106092494,
      "grad_norm": 0.964174586843505,
      "learning_rate": 1.8423704525354247e-06,
      "loss": 0.3688,
      "step": 14863
    },
    {
      "epoch": 0.8021153742377638,
      "grad_norm": 1.1134043635819921,
      "learning_rate": 1.8419260943361743e-06,
      "loss": 0.5964,
      "step": 14864
    },
    {
      "epoch": 0.8021693378662781,
      "grad_norm": 1.1345024966063415,
      "learning_rate": 1.8414818412736657e-06,
      "loss": 0.5014,
      "step": 14865
    },
    {
      "epoch": 0.8022233014947925,
      "grad_norm": 1.136672695267758,
      "learning_rate": 1.8410376933606644e-06,
      "loss": 0.6105,
      "step": 14866
    },
    {
      "epoch": 0.8022772651233069,
      "grad_norm": 1.0081075047634147,
      "learning_rate": 1.840593650609937e-06,
      "loss": 0.4465,
      "step": 14867
    },
    {
      "epoch": 0.8023312287518213,
      "grad_norm": 1.0258975073860266,
      "learning_rate": 1.840149713034245e-06,
      "loss": 0.5233,
      "step": 14868
    },
    {
      "epoch": 0.8023851923803357,
      "grad_norm": 1.0874742521148508,
      "learning_rate": 1.8397058806463491e-06,
      "loss": 0.4121,
      "step": 14869
    },
    {
      "epoch": 0.8024391560088501,
      "grad_norm": 0.8342244516417907,
      "learning_rate": 1.839262153459004e-06,
      "loss": 0.3514,
      "step": 14870
    },
    {
      "epoch": 0.8024931196373645,
      "grad_norm": 0.9905578183520143,
      "learning_rate": 1.838818531484964e-06,
      "loss": 0.4699,
      "step": 14871
    },
    {
      "epoch": 0.8025470832658788,
      "grad_norm": 1.3025108718167793,
      "learning_rate": 1.8383750147369795e-06,
      "loss": 0.5032,
      "step": 14872
    },
    {
      "epoch": 0.8026010468943932,
      "grad_norm": 0.9442464478431325,
      "learning_rate": 1.8379316032277955e-06,
      "loss": 0.3419,
      "step": 14873
    },
    {
      "epoch": 0.8026550105229076,
      "grad_norm": 0.8532057014131399,
      "learning_rate": 1.837488296970158e-06,
      "loss": 0.3165,
      "step": 14874
    },
    {
      "epoch": 0.802708974151422,
      "grad_norm": 1.087983332882838,
      "learning_rate": 1.8370450959768077e-06,
      "loss": 0.5392,
      "step": 14875
    },
    {
      "epoch": 0.8027629377799363,
      "grad_norm": 1.0358739416827556,
      "learning_rate": 1.8366020002604823e-06,
      "loss": 0.4967,
      "step": 14876
    },
    {
      "epoch": 0.8028169014084507,
      "grad_norm": 1.0539420297924937,
      "learning_rate": 1.8361590098339168e-06,
      "loss": 0.448,
      "step": 14877
    },
    {
      "epoch": 0.802870865036965,
      "grad_norm": 0.9705497081071237,
      "learning_rate": 1.8357161247098431e-06,
      "loss": 0.3624,
      "step": 14878
    },
    {
      "epoch": 0.8029248286654794,
      "grad_norm": 1.073161273398357,
      "learning_rate": 1.8352733449009914e-06,
      "loss": 0.6208,
      "step": 14879
    },
    {
      "epoch": 0.8029787922939938,
      "grad_norm": 1.055156864659904,
      "learning_rate": 1.8348306704200859e-06,
      "loss": 0.4476,
      "step": 14880
    },
    {
      "epoch": 0.8030327559225082,
      "grad_norm": 0.7822144268295095,
      "learning_rate": 1.834388101279851e-06,
      "loss": 0.2875,
      "step": 14881
    },
    {
      "epoch": 0.8030867195510226,
      "grad_norm": 1.1108834775176897,
      "learning_rate": 1.8339456374930053e-06,
      "loss": 0.5091,
      "step": 14882
    },
    {
      "epoch": 0.803140683179537,
      "grad_norm": 1.0132494450251692,
      "learning_rate": 1.8335032790722674e-06,
      "loss": 0.4036,
      "step": 14883
    },
    {
      "epoch": 0.8031946468080514,
      "grad_norm": 1.0494940086681137,
      "learning_rate": 1.833061026030349e-06,
      "loss": 0.4833,
      "step": 14884
    },
    {
      "epoch": 0.8032486104365657,
      "grad_norm": 0.9409484376686271,
      "learning_rate": 1.832618878379962e-06,
      "loss": 0.3543,
      "step": 14885
    },
    {
      "epoch": 0.8033025740650801,
      "grad_norm": 0.9287032919470498,
      "learning_rate": 1.8321768361338139e-06,
      "loss": 0.4069,
      "step": 14886
    },
    {
      "epoch": 0.8033565376935945,
      "grad_norm": 0.8634983491242433,
      "learning_rate": 1.8317348993046094e-06,
      "loss": 0.3427,
      "step": 14887
    },
    {
      "epoch": 0.8034105013221089,
      "grad_norm": 0.7291706127039025,
      "learning_rate": 1.8312930679050512e-06,
      "loss": 0.3481,
      "step": 14888
    },
    {
      "epoch": 0.8034644649506233,
      "grad_norm": 0.990692591551886,
      "learning_rate": 1.8308513419478363e-06,
      "loss": 0.3881,
      "step": 14889
    },
    {
      "epoch": 0.8035184285791377,
      "grad_norm": 1.0030693153396586,
      "learning_rate": 1.8304097214456618e-06,
      "loss": 0.4915,
      "step": 14890
    },
    {
      "epoch": 0.8035723922076521,
      "grad_norm": 0.7767291989627656,
      "learning_rate": 1.8299682064112196e-06,
      "loss": 0.2906,
      "step": 14891
    },
    {
      "epoch": 0.8036263558361664,
      "grad_norm": 1.094021916396228,
      "learning_rate": 1.8295267968572005e-06,
      "loss": 0.5207,
      "step": 14892
    },
    {
      "epoch": 0.8036803194646808,
      "grad_norm": 1.1749131854970998,
      "learning_rate": 1.8290854927962886e-06,
      "loss": 0.4895,
      "step": 14893
    },
    {
      "epoch": 0.8037342830931952,
      "grad_norm": 1.0607811994489298,
      "learning_rate": 1.8286442942411692e-06,
      "loss": 0.3938,
      "step": 14894
    },
    {
      "epoch": 0.8037882467217096,
      "grad_norm": 0.9117354168175209,
      "learning_rate": 1.8282032012045222e-06,
      "loss": 0.4272,
      "step": 14895
    },
    {
      "epoch": 0.803842210350224,
      "grad_norm": 0.9857441065327186,
      "learning_rate": 1.8277622136990253e-06,
      "loss": 0.4238,
      "step": 14896
    },
    {
      "epoch": 0.8038961739787384,
      "grad_norm": 1.014995242077085,
      "learning_rate": 1.8273213317373534e-06,
      "loss": 0.4147,
      "step": 14897
    },
    {
      "epoch": 0.8039501376072528,
      "grad_norm": 0.9367949136322224,
      "learning_rate": 1.8268805553321766e-06,
      "loss": 0.3837,
      "step": 14898
    },
    {
      "epoch": 0.804004101235767,
      "grad_norm": 0.9906727856071421,
      "learning_rate": 1.8264398844961657e-06,
      "loss": 0.4906,
      "step": 14899
    },
    {
      "epoch": 0.8040580648642814,
      "grad_norm": 0.8062355015692951,
      "learning_rate": 1.8259993192419833e-06,
      "loss": 0.3752,
      "step": 14900
    },
    {
      "epoch": 0.8041120284927958,
      "grad_norm": 1.1894756184352309,
      "learning_rate": 1.8255588595822927e-06,
      "loss": 0.5302,
      "step": 14901
    },
    {
      "epoch": 0.8041659921213102,
      "grad_norm": 0.9758674926882615,
      "learning_rate": 1.8251185055297533e-06,
      "loss": 0.3437,
      "step": 14902
    },
    {
      "epoch": 0.8042199557498246,
      "grad_norm": 0.9876948716458978,
      "learning_rate": 1.8246782570970207e-06,
      "loss": 0.4206,
      "step": 14903
    },
    {
      "epoch": 0.804273919378339,
      "grad_norm": 0.9577555278383736,
      "learning_rate": 1.8242381142967491e-06,
      "loss": 0.4024,
      "step": 14904
    },
    {
      "epoch": 0.8043278830068534,
      "grad_norm": 0.903248763719223,
      "learning_rate": 1.823798077141588e-06,
      "loss": 0.349,
      "step": 14905
    },
    {
      "epoch": 0.8043818466353677,
      "grad_norm": 1.1046407512277232,
      "learning_rate": 1.8233581456441862e-06,
      "loss": 0.3762,
      "step": 14906
    },
    {
      "epoch": 0.8044358102638821,
      "grad_norm": 0.6580234371148095,
      "learning_rate": 1.8229183198171846e-06,
      "loss": 0.2546,
      "step": 14907
    },
    {
      "epoch": 0.8044897738923965,
      "grad_norm": 1.0806330916845162,
      "learning_rate": 1.8224785996732263e-06,
      "loss": 0.5294,
      "step": 14908
    },
    {
      "epoch": 0.8045437375209109,
      "grad_norm": 0.9263223811457929,
      "learning_rate": 1.8220389852249486e-06,
      "loss": 0.3744,
      "step": 14909
    },
    {
      "epoch": 0.8045977011494253,
      "grad_norm": 0.8455803283602408,
      "learning_rate": 1.821599476484987e-06,
      "loss": 0.3316,
      "step": 14910
    },
    {
      "epoch": 0.8046516647779397,
      "grad_norm": 1.0439169384952034,
      "learning_rate": 1.8211600734659735e-06,
      "loss": 0.4248,
      "step": 14911
    },
    {
      "epoch": 0.8047056284064541,
      "grad_norm": 0.9087322470945891,
      "learning_rate": 1.8207207761805361e-06,
      "loss": 0.4269,
      "step": 14912
    },
    {
      "epoch": 0.8047595920349684,
      "grad_norm": 0.7926332198596833,
      "learning_rate": 1.8202815846413025e-06,
      "loss": 0.3373,
      "step": 14913
    },
    {
      "epoch": 0.8048135556634828,
      "grad_norm": 1.1330138133778485,
      "learning_rate": 1.819842498860893e-06,
      "loss": 0.4328,
      "step": 14914
    },
    {
      "epoch": 0.8048675192919972,
      "grad_norm": 0.9237203115154883,
      "learning_rate": 1.8194035188519294e-06,
      "loss": 0.3359,
      "step": 14915
    },
    {
      "epoch": 0.8049214829205116,
      "grad_norm": 0.8752440739375261,
      "learning_rate": 1.8189646446270265e-06,
      "loss": 0.3275,
      "step": 14916
    },
    {
      "epoch": 0.804975446549026,
      "grad_norm": 1.1421747025087434,
      "learning_rate": 1.8185258761988e-06,
      "loss": 0.5606,
      "step": 14917
    },
    {
      "epoch": 0.8050294101775404,
      "grad_norm": 1.0333628821566332,
      "learning_rate": 1.8180872135798594e-06,
      "loss": 0.4535,
      "step": 14918
    },
    {
      "epoch": 0.8050833738060548,
      "grad_norm": 1.0470871013490217,
      "learning_rate": 1.8176486567828127e-06,
      "loss": 0.5128,
      "step": 14919
    },
    {
      "epoch": 0.8051373374345691,
      "grad_norm": 0.8426683055029965,
      "learning_rate": 1.817210205820265e-06,
      "loss": 0.4155,
      "step": 14920
    },
    {
      "epoch": 0.8051913010630835,
      "grad_norm": 0.9124348915522433,
      "learning_rate": 1.8167718607048164e-06,
      "loss": 0.4121,
      "step": 14921
    },
    {
      "epoch": 0.8052452646915979,
      "grad_norm": 1.155530921667954,
      "learning_rate": 1.8163336214490668e-06,
      "loss": 0.5767,
      "step": 14922
    },
    {
      "epoch": 0.8052992283201122,
      "grad_norm": 0.971090482523904,
      "learning_rate": 1.8158954880656099e-06,
      "loss": 0.4347,
      "step": 14923
    },
    {
      "epoch": 0.8053531919486266,
      "grad_norm": 0.8543214083354823,
      "learning_rate": 1.81545746056704e-06,
      "loss": 0.3602,
      "step": 14924
    },
    {
      "epoch": 0.805407155577141,
      "grad_norm": 1.2350446268928017,
      "learning_rate": 1.8150195389659446e-06,
      "loss": 0.4601,
      "step": 14925
    },
    {
      "epoch": 0.8054611192056554,
      "grad_norm": 1.1650666900499977,
      "learning_rate": 1.8145817232749118e-06,
      "loss": 0.6227,
      "step": 14926
    },
    {
      "epoch": 0.8055150828341697,
      "grad_norm": 1.1028149537478762,
      "learning_rate": 1.8141440135065253e-06,
      "loss": 0.5892,
      "step": 14927
    },
    {
      "epoch": 0.8055690464626841,
      "grad_norm": 1.127625915231473,
      "learning_rate": 1.8137064096733628e-06,
      "loss": 0.5826,
      "step": 14928
    },
    {
      "epoch": 0.8056230100911985,
      "grad_norm": 0.9528063076572508,
      "learning_rate": 1.8132689117880029e-06,
      "loss": 0.4848,
      "step": 14929
    },
    {
      "epoch": 0.8056769737197129,
      "grad_norm": 0.9479943547418739,
      "learning_rate": 1.8128315198630195e-06,
      "loss": 0.4662,
      "step": 14930
    },
    {
      "epoch": 0.8057309373482273,
      "grad_norm": 1.1839034240923088,
      "learning_rate": 1.812394233910984e-06,
      "loss": 0.5281,
      "step": 14931
    },
    {
      "epoch": 0.8057849009767417,
      "grad_norm": 1.1277863995063973,
      "learning_rate": 1.8119570539444641e-06,
      "loss": 0.5561,
      "step": 14932
    },
    {
      "epoch": 0.8058388646052561,
      "grad_norm": 0.8745497663551298,
      "learning_rate": 1.8115199799760251e-06,
      "loss": 0.4066,
      "step": 14933
    },
    {
      "epoch": 0.8058928282337704,
      "grad_norm": 1.1514325675933397,
      "learning_rate": 1.8110830120182287e-06,
      "loss": 0.5129,
      "step": 14934
    },
    {
      "epoch": 0.8059467918622848,
      "grad_norm": 1.0778910881744626,
      "learning_rate": 1.8106461500836345e-06,
      "loss": 0.4657,
      "step": 14935
    },
    {
      "epoch": 0.8060007554907992,
      "grad_norm": 1.1042558576558912,
      "learning_rate": 1.810209394184797e-06,
      "loss": 0.5772,
      "step": 14936
    },
    {
      "epoch": 0.8060547191193136,
      "grad_norm": 1.1568424155069672,
      "learning_rate": 1.8097727443342702e-06,
      "loss": 0.5108,
      "step": 14937
    },
    {
      "epoch": 0.806108682747828,
      "grad_norm": 0.837572558348581,
      "learning_rate": 1.8093362005446032e-06,
      "loss": 0.2649,
      "step": 14938
    },
    {
      "epoch": 0.8061626463763424,
      "grad_norm": 0.9650766217331049,
      "learning_rate": 1.8088997628283429e-06,
      "loss": 0.4303,
      "step": 14939
    },
    {
      "epoch": 0.8062166100048568,
      "grad_norm": 1.0683719638628948,
      "learning_rate": 1.808463431198033e-06,
      "loss": 0.4666,
      "step": 14940
    },
    {
      "epoch": 0.8062705736333711,
      "grad_norm": 1.0256846334417231,
      "learning_rate": 1.8080272056662158e-06,
      "loss": 0.5534,
      "step": 14941
    },
    {
      "epoch": 0.8063245372618855,
      "grad_norm": 1.1823339246448672,
      "learning_rate": 1.8075910862454254e-06,
      "loss": 0.442,
      "step": 14942
    },
    {
      "epoch": 0.8063785008903999,
      "grad_norm": 0.8986031410947927,
      "learning_rate": 1.8071550729481984e-06,
      "loss": 0.3387,
      "step": 14943
    },
    {
      "epoch": 0.8064324645189143,
      "grad_norm": 0.9340110763787425,
      "learning_rate": 1.806719165787066e-06,
      "loss": 0.4021,
      "step": 14944
    },
    {
      "epoch": 0.8064864281474287,
      "grad_norm": 1.085852510115472,
      "learning_rate": 1.806283364774556e-06,
      "loss": 0.3275,
      "step": 14945
    },
    {
      "epoch": 0.806540391775943,
      "grad_norm": 1.1364319764806416,
      "learning_rate": 1.8058476699231947e-06,
      "loss": 0.5651,
      "step": 14946
    },
    {
      "epoch": 0.8065943554044573,
      "grad_norm": 1.2764468224019285,
      "learning_rate": 1.8054120812455053e-06,
      "loss": 0.525,
      "step": 14947
    },
    {
      "epoch": 0.8066483190329717,
      "grad_norm": 1.159503606183495,
      "learning_rate": 1.8049765987540047e-06,
      "loss": 0.4545,
      "step": 14948
    },
    {
      "epoch": 0.8067022826614861,
      "grad_norm": 1.1395060064264508,
      "learning_rate": 1.80454122246121e-06,
      "loss": 0.5564,
      "step": 14949
    },
    {
      "epoch": 0.8067562462900005,
      "grad_norm": 0.8336616875565236,
      "learning_rate": 1.8041059523796345e-06,
      "loss": 0.3829,
      "step": 14950
    },
    {
      "epoch": 0.8068102099185149,
      "grad_norm": 1.0539589910938114,
      "learning_rate": 1.803670788521789e-06,
      "loss": 0.5007,
      "step": 14951
    },
    {
      "epoch": 0.8068641735470293,
      "grad_norm": 0.914580639616152,
      "learning_rate": 1.8032357309001791e-06,
      "loss": 0.3904,
      "step": 14952
    },
    {
      "epoch": 0.8069181371755437,
      "grad_norm": 1.009529856483741,
      "learning_rate": 1.80280077952731e-06,
      "loss": 0.398,
      "step": 14953
    },
    {
      "epoch": 0.806972100804058,
      "grad_norm": 1.104746471422327,
      "learning_rate": 1.8023659344156833e-06,
      "loss": 0.4922,
      "step": 14954
    },
    {
      "epoch": 0.8070260644325724,
      "grad_norm": 1.267548476021106,
      "learning_rate": 1.8019311955777942e-06,
      "loss": 0.5947,
      "step": 14955
    },
    {
      "epoch": 0.8070800280610868,
      "grad_norm": 1.0439939128513471,
      "learning_rate": 1.8014965630261396e-06,
      "loss": 0.4267,
      "step": 14956
    },
    {
      "epoch": 0.8071339916896012,
      "grad_norm": 0.8372970192184802,
      "learning_rate": 1.801062036773211e-06,
      "loss": 0.3953,
      "step": 14957
    },
    {
      "epoch": 0.8071879553181156,
      "grad_norm": 0.9190879973606361,
      "learning_rate": 1.8006276168314969e-06,
      "loss": 0.4926,
      "step": 14958
    },
    {
      "epoch": 0.80724191894663,
      "grad_norm": 1.0292497193473313,
      "learning_rate": 1.8001933032134827e-06,
      "loss": 0.385,
      "step": 14959
    },
    {
      "epoch": 0.8072958825751444,
      "grad_norm": 0.9801563690338797,
      "learning_rate": 1.7997590959316516e-06,
      "loss": 0.47,
      "step": 14960
    },
    {
      "epoch": 0.8073498462036587,
      "grad_norm": 0.9321822240731279,
      "learning_rate": 1.7993249949984842e-06,
      "loss": 0.3759,
      "step": 14961
    },
    {
      "epoch": 0.8074038098321731,
      "grad_norm": 0.9563053320242623,
      "learning_rate": 1.7988910004264542e-06,
      "loss": 0.3856,
      "step": 14962
    },
    {
      "epoch": 0.8074577734606875,
      "grad_norm": 0.8571348686757044,
      "learning_rate": 1.7984571122280372e-06,
      "loss": 0.4418,
      "step": 14963
    },
    {
      "epoch": 0.8075117370892019,
      "grad_norm": 1.070332704067573,
      "learning_rate": 1.7980233304157025e-06,
      "loss": 0.4729,
      "step": 14964
    },
    {
      "epoch": 0.8075657007177163,
      "grad_norm": 0.9184247681591252,
      "learning_rate": 1.7975896550019173e-06,
      "loss": 0.457,
      "step": 14965
    },
    {
      "epoch": 0.8076196643462307,
      "grad_norm": 1.1165326065132974,
      "learning_rate": 1.7971560859991471e-06,
      "loss": 0.5484,
      "step": 14966
    },
    {
      "epoch": 0.8076736279747451,
      "grad_norm": 0.9863118032775459,
      "learning_rate": 1.7967226234198525e-06,
      "loss": 0.4191,
      "step": 14967
    },
    {
      "epoch": 0.8077275916032594,
      "grad_norm": 1.1860739047909694,
      "learning_rate": 1.7962892672764925e-06,
      "loss": 0.4851,
      "step": 14968
    },
    {
      "epoch": 0.8077815552317738,
      "grad_norm": 0.7894256296856095,
      "learning_rate": 1.7958560175815204e-06,
      "loss": 0.3412,
      "step": 14969
    },
    {
      "epoch": 0.8078355188602881,
      "grad_norm": 1.0568125051268067,
      "learning_rate": 1.795422874347389e-06,
      "loss": 0.3956,
      "step": 14970
    },
    {
      "epoch": 0.8078894824888025,
      "grad_norm": 0.9599838317133688,
      "learning_rate": 1.7949898375865473e-06,
      "loss": 0.4401,
      "step": 14971
    },
    {
      "epoch": 0.8079434461173169,
      "grad_norm": 0.8130173749173386,
      "learning_rate": 1.794556907311441e-06,
      "loss": 0.2779,
      "step": 14972
    },
    {
      "epoch": 0.8079974097458313,
      "grad_norm": 0.9078919781123607,
      "learning_rate": 1.794124083534514e-06,
      "loss": 0.3555,
      "step": 14973
    },
    {
      "epoch": 0.8080513733743457,
      "grad_norm": 0.9084122922611785,
      "learning_rate": 1.7936913662682046e-06,
      "loss": 0.3522,
      "step": 14974
    },
    {
      "epoch": 0.80810533700286,
      "grad_norm": 0.887351414631382,
      "learning_rate": 1.7932587555249517e-06,
      "loss": 0.3329,
      "step": 14975
    },
    {
      "epoch": 0.8081593006313744,
      "grad_norm": 1.0560674036483217,
      "learning_rate": 1.7928262513171868e-06,
      "loss": 0.4639,
      "step": 14976
    },
    {
      "epoch": 0.8082132642598888,
      "grad_norm": 1.0109169675477818,
      "learning_rate": 1.7923938536573408e-06,
      "loss": 0.4551,
      "step": 14977
    },
    {
      "epoch": 0.8082672278884032,
      "grad_norm": 1.1793991819063474,
      "learning_rate": 1.7919615625578418e-06,
      "loss": 0.5174,
      "step": 14978
    },
    {
      "epoch": 0.8083211915169176,
      "grad_norm": 1.0014114535798888,
      "learning_rate": 1.7915293780311145e-06,
      "loss": 0.4813,
      "step": 14979
    },
    {
      "epoch": 0.808375155145432,
      "grad_norm": 1.0124623551860468,
      "learning_rate": 1.7910973000895793e-06,
      "loss": 0.4352,
      "step": 14980
    },
    {
      "epoch": 0.8084291187739464,
      "grad_norm": 1.0090925628776648,
      "learning_rate": 1.790665328745656e-06,
      "loss": 0.5218,
      "step": 14981
    },
    {
      "epoch": 0.8084830824024607,
      "grad_norm": 1.2327965910475613,
      "learning_rate": 1.7902334640117586e-06,
      "loss": 0.5195,
      "step": 14982
    },
    {
      "epoch": 0.8085370460309751,
      "grad_norm": 1.1547113245905833,
      "learning_rate": 1.7898017059003002e-06,
      "loss": 0.4874,
      "step": 14983
    },
    {
      "epoch": 0.8085910096594895,
      "grad_norm": 0.985503494722898,
      "learning_rate": 1.7893700544236897e-06,
      "loss": 0.4217,
      "step": 14984
    },
    {
      "epoch": 0.8086449732880039,
      "grad_norm": 1.2340858512506225,
      "learning_rate": 1.788938509594333e-06,
      "loss": 0.4677,
      "step": 14985
    },
    {
      "epoch": 0.8086989369165183,
      "grad_norm": 1.1056450129954292,
      "learning_rate": 1.7885070714246333e-06,
      "loss": 0.577,
      "step": 14986
    },
    {
      "epoch": 0.8087529005450327,
      "grad_norm": 1.0647169751318186,
      "learning_rate": 1.788075739926991e-06,
      "loss": 0.5555,
      "step": 14987
    },
    {
      "epoch": 0.8088068641735471,
      "grad_norm": 1.117389251365727,
      "learning_rate": 1.7876445151138034e-06,
      "loss": 0.7261,
      "step": 14988
    },
    {
      "epoch": 0.8088608278020614,
      "grad_norm": 1.1345824698241858,
      "learning_rate": 1.787213396997462e-06,
      "loss": 0.4991,
      "step": 14989
    },
    {
      "epoch": 0.8089147914305758,
      "grad_norm": 1.1242228676654655,
      "learning_rate": 1.7867823855903592e-06,
      "loss": 0.4845,
      "step": 14990
    },
    {
      "epoch": 0.8089687550590902,
      "grad_norm": 0.9204631404080857,
      "learning_rate": 1.7863514809048827e-06,
      "loss": 0.4009,
      "step": 14991
    },
    {
      "epoch": 0.8090227186876046,
      "grad_norm": 1.2561591540542034,
      "learning_rate": 1.7859206829534175e-06,
      "loss": 0.5299,
      "step": 14992
    },
    {
      "epoch": 0.809076682316119,
      "grad_norm": 1.1683708825450252,
      "learning_rate": 1.7854899917483438e-06,
      "loss": 0.4639,
      "step": 14993
    },
    {
      "epoch": 0.8091306459446334,
      "grad_norm": 1.0376969752541425,
      "learning_rate": 1.785059407302041e-06,
      "loss": 0.4499,
      "step": 14994
    },
    {
      "epoch": 0.8091846095731478,
      "grad_norm": 1.0871378881220306,
      "learning_rate": 1.7846289296268853e-06,
      "loss": 0.4923,
      "step": 14995
    },
    {
      "epoch": 0.809238573201662,
      "grad_norm": 0.976759349907928,
      "learning_rate": 1.7841985587352492e-06,
      "loss": 0.4123,
      "step": 14996
    },
    {
      "epoch": 0.8092925368301764,
      "grad_norm": 1.0629330365357155,
      "learning_rate": 1.7837682946394997e-06,
      "loss": 0.5319,
      "step": 14997
    },
    {
      "epoch": 0.8093465004586908,
      "grad_norm": 0.9205182449534607,
      "learning_rate": 1.7833381373520041e-06,
      "loss": 0.3385,
      "step": 14998
    },
    {
      "epoch": 0.8094004640872052,
      "grad_norm": 1.2154714236721322,
      "learning_rate": 1.7829080868851262e-06,
      "loss": 0.5878,
      "step": 14999
    },
    {
      "epoch": 0.8094544277157196,
      "grad_norm": 1.2976292256622846,
      "learning_rate": 1.7824781432512256e-06,
      "loss": 0.6482,
      "step": 15000
    },
    {
      "epoch": 0.8094544277157196,
      "eval_loss": 0.5320729613304138,
      "eval_runtime": 163.8564,
      "eval_samples_per_second": 20.988,
      "eval_steps_per_second": 0.879,
      "step": 15000
    },
    {
      "epoch": 0.809508391344234,
      "grad_norm": 1.1863441200170959,
      "learning_rate": 1.7820483064626597e-06,
      "loss": 0.4728,
      "step": 15001
    },
    {
      "epoch": 0.8095623549727484,
      "grad_norm": 1.0080234605331135,
      "learning_rate": 1.7816185765317822e-06,
      "loss": 0.4781,
      "step": 15002
    },
    {
      "epoch": 0.8096163186012627,
      "grad_norm": 0.8850413627677761,
      "learning_rate": 1.7811889534709449e-06,
      "loss": 0.4378,
      "step": 15003
    },
    {
      "epoch": 0.8096702822297771,
      "grad_norm": 0.8345954079871095,
      "learning_rate": 1.780759437292493e-06,
      "loss": 0.295,
      "step": 15004
    },
    {
      "epoch": 0.8097242458582915,
      "grad_norm": 1.1139046844990743,
      "learning_rate": 1.7803300280087738e-06,
      "loss": 0.5285,
      "step": 15005
    },
    {
      "epoch": 0.8097782094868059,
      "grad_norm": 1.0593706783806836,
      "learning_rate": 1.7799007256321276e-06,
      "loss": 0.4337,
      "step": 15006
    },
    {
      "epoch": 0.8098321731153203,
      "grad_norm": 0.9804437252946457,
      "learning_rate": 1.7794715301748936e-06,
      "loss": 0.4022,
      "step": 15007
    },
    {
      "epoch": 0.8098861367438347,
      "grad_norm": 1.1375057083100792,
      "learning_rate": 1.7790424416494074e-06,
      "loss": 0.5933,
      "step": 15008
    },
    {
      "epoch": 0.8099401003723491,
      "grad_norm": 1.1636874970256401,
      "learning_rate": 1.7786134600680001e-06,
      "loss": 0.5059,
      "step": 15009
    },
    {
      "epoch": 0.8099940640008634,
      "grad_norm": 1.1135987378545398,
      "learning_rate": 1.778184585443004e-06,
      "loss": 0.5402,
      "step": 15010
    },
    {
      "epoch": 0.8100480276293778,
      "grad_norm": 1.0000052746056893,
      "learning_rate": 1.7777558177867427e-06,
      "loss": 0.4939,
      "step": 15011
    },
    {
      "epoch": 0.8101019912578922,
      "grad_norm": 0.991326883153044,
      "learning_rate": 1.7773271571115403e-06,
      "loss": 0.4456,
      "step": 15012
    },
    {
      "epoch": 0.8101559548864066,
      "grad_norm": 1.05906384632891,
      "learning_rate": 1.776898603429716e-06,
      "loss": 0.4838,
      "step": 15013
    },
    {
      "epoch": 0.810209918514921,
      "grad_norm": 1.002298963206294,
      "learning_rate": 1.7764701567535885e-06,
      "loss": 0.3233,
      "step": 15014
    },
    {
      "epoch": 0.8102638821434354,
      "grad_norm": 1.3738403079161232,
      "learning_rate": 1.7760418170954708e-06,
      "loss": 0.6579,
      "step": 15015
    },
    {
      "epoch": 0.8103178457719497,
      "grad_norm": 1.6598733687139255,
      "learning_rate": 1.7756135844676752e-06,
      "loss": 0.422,
      "step": 15016
    },
    {
      "epoch": 0.810371809400464,
      "grad_norm": 1.1336587353622936,
      "learning_rate": 1.7751854588825075e-06,
      "loss": 0.5138,
      "step": 15017
    },
    {
      "epoch": 0.8104257730289784,
      "grad_norm": 1.0175921252993692,
      "learning_rate": 1.7747574403522733e-06,
      "loss": 0.4678,
      "step": 15018
    },
    {
      "epoch": 0.8104797366574928,
      "grad_norm": 1.1661461083938378,
      "learning_rate": 1.7743295288892743e-06,
      "loss": 0.5124,
      "step": 15019
    },
    {
      "epoch": 0.8105337002860072,
      "grad_norm": 0.860716818273619,
      "learning_rate": 1.7739017245058089e-06,
      "loss": 0.3894,
      "step": 15020
    },
    {
      "epoch": 0.8105876639145216,
      "grad_norm": 0.8168400528370259,
      "learning_rate": 1.773474027214173e-06,
      "loss": 0.3716,
      "step": 15021
    },
    {
      "epoch": 0.810641627543036,
      "grad_norm": 1.1685792593057673,
      "learning_rate": 1.7730464370266587e-06,
      "loss": 0.563,
      "step": 15022
    },
    {
      "epoch": 0.8106955911715503,
      "grad_norm": 1.3472048182129714,
      "learning_rate": 1.7726189539555572e-06,
      "loss": 0.6815,
      "step": 15023
    },
    {
      "epoch": 0.8107495548000647,
      "grad_norm": 1.0736375926752653,
      "learning_rate": 1.7721915780131522e-06,
      "loss": 0.5605,
      "step": 15024
    },
    {
      "epoch": 0.8108035184285791,
      "grad_norm": 1.0211496883191755,
      "learning_rate": 1.771764309211727e-06,
      "loss": 0.3775,
      "step": 15025
    },
    {
      "epoch": 0.8108574820570935,
      "grad_norm": 0.9090755112037462,
      "learning_rate": 1.7713371475635637e-06,
      "loss": 0.4407,
      "step": 15026
    },
    {
      "epoch": 0.8109114456856079,
      "grad_norm": 1.050819920008424,
      "learning_rate": 1.7709100930809382e-06,
      "loss": 0.4721,
      "step": 15027
    },
    {
      "epoch": 0.8109654093141223,
      "grad_norm": 1.0183332833978376,
      "learning_rate": 1.770483145776124e-06,
      "loss": 0.5717,
      "step": 15028
    },
    {
      "epoch": 0.8110193729426367,
      "grad_norm": 1.2949718287047092,
      "learning_rate": 1.7700563056613928e-06,
      "loss": 0.5345,
      "step": 15029
    },
    {
      "epoch": 0.811073336571151,
      "grad_norm": 0.9807674701731001,
      "learning_rate": 1.7696295727490127e-06,
      "loss": 0.465,
      "step": 15030
    },
    {
      "epoch": 0.8111273001996654,
      "grad_norm": 0.7975593296569758,
      "learning_rate": 1.7692029470512473e-06,
      "loss": 0.4181,
      "step": 15031
    },
    {
      "epoch": 0.8111812638281798,
      "grad_norm": 0.9554080022033713,
      "learning_rate": 1.7687764285803602e-06,
      "loss": 0.3528,
      "step": 15032
    },
    {
      "epoch": 0.8112352274566942,
      "grad_norm": 1.0494824058899817,
      "learning_rate": 1.768350017348608e-06,
      "loss": 0.5595,
      "step": 15033
    },
    {
      "epoch": 0.8112891910852086,
      "grad_norm": 1.031737000930814,
      "learning_rate": 1.7679237133682465e-06,
      "loss": 0.458,
      "step": 15034
    },
    {
      "epoch": 0.811343154713723,
      "grad_norm": 0.8947440835027869,
      "learning_rate": 1.767497516651528e-06,
      "loss": 0.3653,
      "step": 15035
    },
    {
      "epoch": 0.8113971183422374,
      "grad_norm": 1.4364257981975208,
      "learning_rate": 1.767071427210703e-06,
      "loss": 0.5575,
      "step": 15036
    },
    {
      "epoch": 0.8114510819707517,
      "grad_norm": 1.11092716619213,
      "learning_rate": 1.7666454450580166e-06,
      "loss": 0.5065,
      "step": 15037
    },
    {
      "epoch": 0.8115050455992661,
      "grad_norm": 1.0225252214628613,
      "learning_rate": 1.7662195702057127e-06,
      "loss": 0.4453,
      "step": 15038
    },
    {
      "epoch": 0.8115590092277805,
      "grad_norm": 0.842351008972849,
      "learning_rate": 1.7657938026660308e-06,
      "loss": 0.4553,
      "step": 15039
    },
    {
      "epoch": 0.8116129728562949,
      "grad_norm": 0.8657925350717867,
      "learning_rate": 1.7653681424512086e-06,
      "loss": 0.3252,
      "step": 15040
    },
    {
      "epoch": 0.8116669364848093,
      "grad_norm": 1.1338157187208147,
      "learning_rate": 1.764942589573479e-06,
      "loss": 0.5433,
      "step": 15041
    },
    {
      "epoch": 0.8117209001133237,
      "grad_norm": 0.9501993780123866,
      "learning_rate": 1.7645171440450737e-06,
      "loss": 0.4986,
      "step": 15042
    },
    {
      "epoch": 0.811774863741838,
      "grad_norm": 0.9946240155934433,
      "learning_rate": 1.7640918058782196e-06,
      "loss": 0.4144,
      "step": 15043
    },
    {
      "epoch": 0.8118288273703523,
      "grad_norm": 1.0747504347797308,
      "learning_rate": 1.7636665750851435e-06,
      "loss": 0.4966,
      "step": 15044
    },
    {
      "epoch": 0.8118827909988667,
      "grad_norm": 1.0718500224286507,
      "learning_rate": 1.763241451678064e-06,
      "loss": 0.5328,
      "step": 15045
    },
    {
      "epoch": 0.8119367546273811,
      "grad_norm": 0.9745615218813367,
      "learning_rate": 1.7628164356692012e-06,
      "loss": 0.5111,
      "step": 15046
    },
    {
      "epoch": 0.8119907182558955,
      "grad_norm": 1.0201795589662783,
      "learning_rate": 1.7623915270707704e-06,
      "loss": 0.3848,
      "step": 15047
    },
    {
      "epoch": 0.8120446818844099,
      "grad_norm": 0.9797688631806711,
      "learning_rate": 1.761966725894983e-06,
      "loss": 0.4517,
      "step": 15048
    },
    {
      "epoch": 0.8120986455129243,
      "grad_norm": 1.220546859588374,
      "learning_rate": 1.7615420321540495e-06,
      "loss": 0.6075,
      "step": 15049
    },
    {
      "epoch": 0.8121526091414387,
      "grad_norm": 0.9286313305979726,
      "learning_rate": 1.761117445860175e-06,
      "loss": 0.3389,
      "step": 15050
    },
    {
      "epoch": 0.812206572769953,
      "grad_norm": 1.0906033186195883,
      "learning_rate": 1.760692967025564e-06,
      "loss": 0.4641,
      "step": 15051
    },
    {
      "epoch": 0.8122605363984674,
      "grad_norm": 0.9281618600735212,
      "learning_rate": 1.7602685956624143e-06,
      "loss": 0.4128,
      "step": 15052
    },
    {
      "epoch": 0.8123145000269818,
      "grad_norm": 0.9228913279007626,
      "learning_rate": 1.7598443317829238e-06,
      "loss": 0.3918,
      "step": 15053
    },
    {
      "epoch": 0.8123684636554962,
      "grad_norm": 0.9726166968848565,
      "learning_rate": 1.7594201753992867e-06,
      "loss": 0.467,
      "step": 15054
    },
    {
      "epoch": 0.8124224272840106,
      "grad_norm": 0.9505287269078845,
      "learning_rate": 1.7589961265236933e-06,
      "loss": 0.4632,
      "step": 15055
    },
    {
      "epoch": 0.812476390912525,
      "grad_norm": 1.017786292321823,
      "learning_rate": 1.7585721851683315e-06,
      "loss": 0.4069,
      "step": 15056
    },
    {
      "epoch": 0.8125303545410394,
      "grad_norm": 1.161272788386706,
      "learning_rate": 1.758148351345385e-06,
      "loss": 0.504,
      "step": 15057
    },
    {
      "epoch": 0.8125843181695537,
      "grad_norm": 1.0816243987592278,
      "learning_rate": 1.7577246250670371e-06,
      "loss": 0.3928,
      "step": 15058
    },
    {
      "epoch": 0.8126382817980681,
      "grad_norm": 1.0700417249672256,
      "learning_rate": 1.7573010063454638e-06,
      "loss": 0.5,
      "step": 15059
    },
    {
      "epoch": 0.8126922454265825,
      "grad_norm": 0.9613288741890027,
      "learning_rate": 1.7568774951928409e-06,
      "loss": 0.5336,
      "step": 15060
    },
    {
      "epoch": 0.8127462090550969,
      "grad_norm": 0.8590534834836154,
      "learning_rate": 1.7564540916213415e-06,
      "loss": 0.2727,
      "step": 15061
    },
    {
      "epoch": 0.8128001726836113,
      "grad_norm": 1.1449316477622389,
      "learning_rate": 1.756030795643134e-06,
      "loss": 0.5823,
      "step": 15062
    },
    {
      "epoch": 0.8128541363121257,
      "grad_norm": 0.8688042572141161,
      "learning_rate": 1.7556076072703836e-06,
      "loss": 0.3848,
      "step": 15063
    },
    {
      "epoch": 0.8129080999406401,
      "grad_norm": 1.1398900047288822,
      "learning_rate": 1.7551845265152551e-06,
      "loss": 0.5134,
      "step": 15064
    },
    {
      "epoch": 0.8129620635691543,
      "grad_norm": 1.2288621092842575,
      "learning_rate": 1.7547615533899076e-06,
      "loss": 0.4351,
      "step": 15065
    },
    {
      "epoch": 0.8130160271976687,
      "grad_norm": 1.1285164601993736,
      "learning_rate": 1.7543386879064967e-06,
      "loss": 0.4712,
      "step": 15066
    },
    {
      "epoch": 0.8130699908261831,
      "grad_norm": 0.8491767051728157,
      "learning_rate": 1.753915930077176e-06,
      "loss": 0.3256,
      "step": 15067
    },
    {
      "epoch": 0.8131239544546975,
      "grad_norm": 0.867821730795633,
      "learning_rate": 1.7534932799140975e-06,
      "loss": 0.3285,
      "step": 15068
    },
    {
      "epoch": 0.8131779180832119,
      "grad_norm": 0.8942739222155119,
      "learning_rate": 1.7530707374294068e-06,
      "loss": 0.5627,
      "step": 15069
    },
    {
      "epoch": 0.8132318817117263,
      "grad_norm": 1.142258686153707,
      "learning_rate": 1.7526483026352498e-06,
      "loss": 0.5152,
      "step": 15070
    },
    {
      "epoch": 0.8132858453402407,
      "grad_norm": 1.2945810026361777,
      "learning_rate": 1.7522259755437666e-06,
      "loss": 0.5458,
      "step": 15071
    },
    {
      "epoch": 0.813339808968755,
      "grad_norm": 1.177462380169229,
      "learning_rate": 1.7518037561670967e-06,
      "loss": 0.4323,
      "step": 15072
    },
    {
      "epoch": 0.8133937725972694,
      "grad_norm": 1.032273033352269,
      "learning_rate": 1.7513816445173734e-06,
      "loss": 0.5061,
      "step": 15073
    },
    {
      "epoch": 0.8134477362257838,
      "grad_norm": 1.1759257729250066,
      "learning_rate": 1.7509596406067286e-06,
      "loss": 0.4707,
      "step": 15074
    },
    {
      "epoch": 0.8135016998542982,
      "grad_norm": 1.0687775208740045,
      "learning_rate": 1.7505377444472926e-06,
      "loss": 0.427,
      "step": 15075
    },
    {
      "epoch": 0.8135556634828126,
      "grad_norm": 0.9950248208213496,
      "learning_rate": 1.7501159560511905e-06,
      "loss": 0.5003,
      "step": 15076
    },
    {
      "epoch": 0.813609627111327,
      "grad_norm": 1.0381643494009056,
      "learning_rate": 1.7496942754305441e-06,
      "loss": 0.5011,
      "step": 15077
    },
    {
      "epoch": 0.8136635907398413,
      "grad_norm": 0.9622982546521305,
      "learning_rate": 1.749272702597475e-06,
      "loss": 0.4457,
      "step": 15078
    },
    {
      "epoch": 0.8137175543683557,
      "grad_norm": 0.9257103328895527,
      "learning_rate": 1.7488512375640976e-06,
      "loss": 0.3438,
      "step": 15079
    },
    {
      "epoch": 0.8137715179968701,
      "grad_norm": 1.0628779641462953,
      "learning_rate": 1.7484298803425253e-06,
      "loss": 0.4762,
      "step": 15080
    },
    {
      "epoch": 0.8138254816253845,
      "grad_norm": 1.0903607370443784,
      "learning_rate": 1.748008630944869e-06,
      "loss": 0.5062,
      "step": 15081
    },
    {
      "epoch": 0.8138794452538989,
      "grad_norm": 1.0372253890873118,
      "learning_rate": 1.7475874893832363e-06,
      "loss": 0.4499,
      "step": 15082
    },
    {
      "epoch": 0.8139334088824133,
      "grad_norm": 1.0732225787618093,
      "learning_rate": 1.7471664556697305e-06,
      "loss": 0.3982,
      "step": 15083
    },
    {
      "epoch": 0.8139873725109277,
      "grad_norm": 0.9649086873186808,
      "learning_rate": 1.7467455298164525e-06,
      "loss": 0.451,
      "step": 15084
    },
    {
      "epoch": 0.814041336139442,
      "grad_norm": 1.0065573090965891,
      "learning_rate": 1.7463247118355009e-06,
      "loss": 0.4062,
      "step": 15085
    },
    {
      "epoch": 0.8140952997679564,
      "grad_norm": 0.9683149991183063,
      "learning_rate": 1.745904001738969e-06,
      "loss": 0.4351,
      "step": 15086
    },
    {
      "epoch": 0.8141492633964708,
      "grad_norm": 0.9664827107852813,
      "learning_rate": 1.7454833995389498e-06,
      "loss": 0.358,
      "step": 15087
    },
    {
      "epoch": 0.8142032270249852,
      "grad_norm": 1.0322280720681658,
      "learning_rate": 1.7450629052475312e-06,
      "loss": 0.5193,
      "step": 15088
    },
    {
      "epoch": 0.8142571906534996,
      "grad_norm": 0.905768647720355,
      "learning_rate": 1.744642518876799e-06,
      "loss": 0.3446,
      "step": 15089
    },
    {
      "epoch": 0.814311154282014,
      "grad_norm": 1.0122542823513263,
      "learning_rate": 1.744222240438835e-06,
      "loss": 0.3496,
      "step": 15090
    },
    {
      "epoch": 0.8143651179105283,
      "grad_norm": 0.9526966313163798,
      "learning_rate": 1.7438020699457192e-06,
      "loss": 0.4415,
      "step": 15091
    },
    {
      "epoch": 0.8144190815390426,
      "grad_norm": 1.1722786517777881,
      "learning_rate": 1.7433820074095281e-06,
      "loss": 0.5076,
      "step": 15092
    },
    {
      "epoch": 0.814473045167557,
      "grad_norm": 0.8690401099128786,
      "learning_rate": 1.742962052842333e-06,
      "loss": 0.252,
      "step": 15093
    },
    {
      "epoch": 0.8145270087960714,
      "grad_norm": 0.9669428287104388,
      "learning_rate": 1.7425422062562042e-06,
      "loss": 0.48,
      "step": 15094
    },
    {
      "epoch": 0.8145809724245858,
      "grad_norm": 1.3425039882915293,
      "learning_rate": 1.7421224676632097e-06,
      "loss": 0.4968,
      "step": 15095
    },
    {
      "epoch": 0.8146349360531002,
      "grad_norm": 1.051796953321871,
      "learning_rate": 1.7417028370754124e-06,
      "loss": 0.5064,
      "step": 15096
    },
    {
      "epoch": 0.8146888996816146,
      "grad_norm": 1.111424443600791,
      "learning_rate": 1.7412833145048725e-06,
      "loss": 0.4134,
      "step": 15097
    },
    {
      "epoch": 0.814742863310129,
      "grad_norm": 1.4429364988121711,
      "learning_rate": 1.7408638999636481e-06,
      "loss": 0.5257,
      "step": 15098
    },
    {
      "epoch": 0.8147968269386433,
      "grad_norm": 0.8955639375008229,
      "learning_rate": 1.7404445934637954e-06,
      "loss": 0.3548,
      "step": 15099
    },
    {
      "epoch": 0.8148507905671577,
      "grad_norm": 1.1712014707507226,
      "learning_rate": 1.7400253950173618e-06,
      "loss": 0.5068,
      "step": 15100
    },
    {
      "epoch": 0.8149047541956721,
      "grad_norm": 1.110871267303176,
      "learning_rate": 1.7396063046363981e-06,
      "loss": 0.4628,
      "step": 15101
    },
    {
      "epoch": 0.8149587178241865,
      "grad_norm": 1.1708196856809225,
      "learning_rate": 1.7391873223329491e-06,
      "loss": 0.3986,
      "step": 15102
    },
    {
      "epoch": 0.8150126814527009,
      "grad_norm": 1.1820238998553476,
      "learning_rate": 1.7387684481190559e-06,
      "loss": 0.6195,
      "step": 15103
    },
    {
      "epoch": 0.8150666450812153,
      "grad_norm": 1.0028754147914025,
      "learning_rate": 1.738349682006758e-06,
      "loss": 0.386,
      "step": 15104
    },
    {
      "epoch": 0.8151206087097297,
      "grad_norm": 0.9976206099272383,
      "learning_rate": 1.7379310240080913e-06,
      "loss": 0.4719,
      "step": 15105
    },
    {
      "epoch": 0.815174572338244,
      "grad_norm": 0.8360013072997218,
      "learning_rate": 1.7375124741350897e-06,
      "loss": 0.3464,
      "step": 15106
    },
    {
      "epoch": 0.8152285359667584,
      "grad_norm": 1.0523357493428456,
      "learning_rate": 1.7370940323997794e-06,
      "loss": 0.4834,
      "step": 15107
    },
    {
      "epoch": 0.8152824995952728,
      "grad_norm": 0.8712314369171777,
      "learning_rate": 1.7366756988141892e-06,
      "loss": 0.3176,
      "step": 15108
    },
    {
      "epoch": 0.8153364632237872,
      "grad_norm": 1.1199600100351657,
      "learning_rate": 1.736257473390342e-06,
      "loss": 0.4858,
      "step": 15109
    },
    {
      "epoch": 0.8153904268523016,
      "grad_norm": 1.0674403693648564,
      "learning_rate": 1.735839356140258e-06,
      "loss": 0.4886,
      "step": 15110
    },
    {
      "epoch": 0.815444390480816,
      "grad_norm": 1.0901726533090381,
      "learning_rate": 1.7354213470759543e-06,
      "loss": 0.5467,
      "step": 15111
    },
    {
      "epoch": 0.8154983541093304,
      "grad_norm": 1.0339069268329082,
      "learning_rate": 1.7350034462094451e-06,
      "loss": 0.4451,
      "step": 15112
    },
    {
      "epoch": 0.8155523177378446,
      "grad_norm": 1.0057614378727346,
      "learning_rate": 1.7345856535527413e-06,
      "loss": 0.4198,
      "step": 15113
    },
    {
      "epoch": 0.815606281366359,
      "grad_norm": 1.0159768008715466,
      "learning_rate": 1.7341679691178502e-06,
      "loss": 0.5239,
      "step": 15114
    },
    {
      "epoch": 0.8156602449948734,
      "grad_norm": 1.0455589995478431,
      "learning_rate": 1.7337503929167764e-06,
      "loss": 0.4742,
      "step": 15115
    },
    {
      "epoch": 0.8157142086233878,
      "grad_norm": 1.0614386793007953,
      "learning_rate": 1.7333329249615228e-06,
      "loss": 0.4726,
      "step": 15116
    },
    {
      "epoch": 0.8157681722519022,
      "grad_norm": 1.0767425261300112,
      "learning_rate": 1.732915565264086e-06,
      "loss": 0.4747,
      "step": 15117
    },
    {
      "epoch": 0.8158221358804166,
      "grad_norm": 1.1241505074808311,
      "learning_rate": 1.732498313836462e-06,
      "loss": 0.6552,
      "step": 15118
    },
    {
      "epoch": 0.815876099508931,
      "grad_norm": 0.9334109126824673,
      "learning_rate": 1.732081170690644e-06,
      "loss": 0.3467,
      "step": 15119
    },
    {
      "epoch": 0.8159300631374453,
      "grad_norm": 1.244613518243582,
      "learning_rate": 1.731664135838621e-06,
      "loss": 0.6001,
      "step": 15120
    },
    {
      "epoch": 0.8159840267659597,
      "grad_norm": 1.2344020015749921,
      "learning_rate": 1.731247209292377e-06,
      "loss": 0.6271,
      "step": 15121
    },
    {
      "epoch": 0.8160379903944741,
      "grad_norm": 1.1818382982229965,
      "learning_rate": 1.7308303910638968e-06,
      "loss": 0.4584,
      "step": 15122
    },
    {
      "epoch": 0.8160919540229885,
      "grad_norm": 0.921933653289056,
      "learning_rate": 1.7304136811651595e-06,
      "loss": 0.4343,
      "step": 15123
    },
    {
      "epoch": 0.8161459176515029,
      "grad_norm": 1.1762653204705922,
      "learning_rate": 1.7299970796081418e-06,
      "loss": 0.5568,
      "step": 15124
    },
    {
      "epoch": 0.8161998812800173,
      "grad_norm": 0.9152830710848486,
      "learning_rate": 1.7295805864048172e-06,
      "loss": 0.3551,
      "step": 15125
    },
    {
      "epoch": 0.8162538449085317,
      "grad_norm": 0.8106504220108,
      "learning_rate": 1.729164201567156e-06,
      "loss": 0.2962,
      "step": 15126
    },
    {
      "epoch": 0.816307808537046,
      "grad_norm": 0.9651932021039273,
      "learning_rate": 1.7287479251071276e-06,
      "loss": 0.3988,
      "step": 15127
    },
    {
      "epoch": 0.8163617721655604,
      "grad_norm": 1.0344336122073787,
      "learning_rate": 1.7283317570366932e-06,
      "loss": 0.5532,
      "step": 15128
    },
    {
      "epoch": 0.8164157357940748,
      "grad_norm": 0.9841066214194347,
      "learning_rate": 1.7279156973678146e-06,
      "loss": 0.3836,
      "step": 15129
    },
    {
      "epoch": 0.8164696994225892,
      "grad_norm": 1.1979796659556876,
      "learning_rate": 1.7274997461124505e-06,
      "loss": 0.5245,
      "step": 15130
    },
    {
      "epoch": 0.8165236630511036,
      "grad_norm": 1.2160194836749223,
      "learning_rate": 1.727083903282556e-06,
      "loss": 0.5113,
      "step": 15131
    },
    {
      "epoch": 0.816577626679618,
      "grad_norm": 1.0834107492549452,
      "learning_rate": 1.7266681688900818e-06,
      "loss": 0.4522,
      "step": 15132
    },
    {
      "epoch": 0.8166315903081324,
      "grad_norm": 1.2704657925601976,
      "learning_rate": 1.7262525429469772e-06,
      "loss": 0.6654,
      "step": 15133
    },
    {
      "epoch": 0.8166855539366467,
      "grad_norm": 0.8886308246887907,
      "learning_rate": 1.7258370254651878e-06,
      "loss": 0.346,
      "step": 15134
    },
    {
      "epoch": 0.816739517565161,
      "grad_norm": 1.2057503006807524,
      "learning_rate": 1.7254216164566557e-06,
      "loss": 0.546,
      "step": 15135
    },
    {
      "epoch": 0.8167934811936755,
      "grad_norm": 0.9872574723305986,
      "learning_rate": 1.72500631593332e-06,
      "loss": 0.4473,
      "step": 15136
    },
    {
      "epoch": 0.8168474448221898,
      "grad_norm": 0.9550430535742108,
      "learning_rate": 1.7245911239071173e-06,
      "loss": 0.4153,
      "step": 15137
    },
    {
      "epoch": 0.8169014084507042,
      "grad_norm": 1.0637829910401888,
      "learning_rate": 1.724176040389982e-06,
      "loss": 0.4444,
      "step": 15138
    },
    {
      "epoch": 0.8169553720792186,
      "grad_norm": 0.9615300470904677,
      "learning_rate": 1.723761065393841e-06,
      "loss": 0.4073,
      "step": 15139
    },
    {
      "epoch": 0.817009335707733,
      "grad_norm": 0.8760107586559983,
      "learning_rate": 1.7233461989306228e-06,
      "loss": 0.3442,
      "step": 15140
    },
    {
      "epoch": 0.8170632993362473,
      "grad_norm": 1.0898336738932737,
      "learning_rate": 1.7229314410122505e-06,
      "loss": 0.3841,
      "step": 15141
    },
    {
      "epoch": 0.8171172629647617,
      "grad_norm": 0.9843084497830475,
      "learning_rate": 1.7225167916506452e-06,
      "loss": 0.501,
      "step": 15142
    },
    {
      "epoch": 0.8171712265932761,
      "grad_norm": 1.2199929395560487,
      "learning_rate": 1.7221022508577245e-06,
      "loss": 0.4557,
      "step": 15143
    },
    {
      "epoch": 0.8172251902217905,
      "grad_norm": 1.0547251240118944,
      "learning_rate": 1.7216878186454019e-06,
      "loss": 0.446,
      "step": 15144
    },
    {
      "epoch": 0.8172791538503049,
      "grad_norm": 0.9961280188763848,
      "learning_rate": 1.7212734950255891e-06,
      "loss": 0.4058,
      "step": 15145
    },
    {
      "epoch": 0.8173331174788193,
      "grad_norm": 1.1509943920364634,
      "learning_rate": 1.7208592800101947e-06,
      "loss": 0.5363,
      "step": 15146
    },
    {
      "epoch": 0.8173870811073336,
      "grad_norm": 1.0177544506846612,
      "learning_rate": 1.7204451736111238e-06,
      "loss": 0.3991,
      "step": 15147
    },
    {
      "epoch": 0.817441044735848,
      "grad_norm": 0.844767114487663,
      "learning_rate": 1.7200311758402765e-06,
      "loss": 0.401,
      "step": 15148
    },
    {
      "epoch": 0.8174950083643624,
      "grad_norm": 0.9407685433048384,
      "learning_rate": 1.7196172867095528e-06,
      "loss": 0.4434,
      "step": 15149
    },
    {
      "epoch": 0.8175489719928768,
      "grad_norm": 1.065244587996238,
      "learning_rate": 1.7192035062308476e-06,
      "loss": 0.4303,
      "step": 15150
    },
    {
      "epoch": 0.8176029356213912,
      "grad_norm": 0.9311825906199078,
      "learning_rate": 1.7187898344160547e-06,
      "loss": 0.374,
      "step": 15151
    },
    {
      "epoch": 0.8176568992499056,
      "grad_norm": 1.149769920024768,
      "learning_rate": 1.7183762712770619e-06,
      "loss": 0.484,
      "step": 15152
    },
    {
      "epoch": 0.81771086287842,
      "grad_norm": 1.1495573386857607,
      "learning_rate": 1.7179628168257567e-06,
      "loss": 0.3943,
      "step": 15153
    },
    {
      "epoch": 0.8177648265069343,
      "grad_norm": 1.0451360011956505,
      "learning_rate": 1.7175494710740223e-06,
      "loss": 0.4849,
      "step": 15154
    },
    {
      "epoch": 0.8178187901354487,
      "grad_norm": 1.057213380359949,
      "learning_rate": 1.7171362340337367e-06,
      "loss": 0.6289,
      "step": 15155
    },
    {
      "epoch": 0.8178727537639631,
      "grad_norm": 0.9364574570580809,
      "learning_rate": 1.7167231057167782e-06,
      "loss": 0.4733,
      "step": 15156
    },
    {
      "epoch": 0.8179267173924775,
      "grad_norm": 1.2795891799871753,
      "learning_rate": 1.7163100861350208e-06,
      "loss": 0.5759,
      "step": 15157
    },
    {
      "epoch": 0.8179806810209919,
      "grad_norm": 1.1747363932518176,
      "learning_rate": 1.715897175300334e-06,
      "loss": 0.5677,
      "step": 15158
    },
    {
      "epoch": 0.8180346446495063,
      "grad_norm": 0.8931459010240307,
      "learning_rate": 1.7154843732245862e-06,
      "loss": 0.4422,
      "step": 15159
    },
    {
      "epoch": 0.8180886082780207,
      "grad_norm": 1.0055573861401688,
      "learning_rate": 1.7150716799196414e-06,
      "loss": 0.5087,
      "step": 15160
    },
    {
      "epoch": 0.8181425719065349,
      "grad_norm": 1.2377504555005403,
      "learning_rate": 1.7146590953973625e-06,
      "loss": 0.5498,
      "step": 15161
    },
    {
      "epoch": 0.8181965355350493,
      "grad_norm": 1.361495944776977,
      "learning_rate": 1.7142466196696044e-06,
      "loss": 0.651,
      "step": 15162
    },
    {
      "epoch": 0.8182504991635637,
      "grad_norm": 1.1331645703832274,
      "learning_rate": 1.7138342527482238e-06,
      "loss": 0.4738,
      "step": 15163
    },
    {
      "epoch": 0.8183044627920781,
      "grad_norm": 1.0351599573079375,
      "learning_rate": 1.7134219946450725e-06,
      "loss": 0.3814,
      "step": 15164
    },
    {
      "epoch": 0.8183584264205925,
      "grad_norm": 1.360444761976536,
      "learning_rate": 1.713009845371999e-06,
      "loss": 0.5723,
      "step": 15165
    },
    {
      "epoch": 0.8184123900491069,
      "grad_norm": 1.102710551027252,
      "learning_rate": 1.7125978049408486e-06,
      "loss": 0.4076,
      "step": 15166
    },
    {
      "epoch": 0.8184663536776213,
      "grad_norm": 0.9755193654078085,
      "learning_rate": 1.712185873363465e-06,
      "loss": 0.4893,
      "step": 15167
    },
    {
      "epoch": 0.8185203173061356,
      "grad_norm": 1.0520917216630519,
      "learning_rate": 1.7117740506516874e-06,
      "loss": 0.4777,
      "step": 15168
    },
    {
      "epoch": 0.81857428093465,
      "grad_norm": 0.997824178128833,
      "learning_rate": 1.7113623368173504e-06,
      "loss": 0.3743,
      "step": 15169
    },
    {
      "epoch": 0.8186282445631644,
      "grad_norm": 0.9263484601998064,
      "learning_rate": 1.710950731872288e-06,
      "loss": 0.4078,
      "step": 15170
    },
    {
      "epoch": 0.8186822081916788,
      "grad_norm": 1.120297231608471,
      "learning_rate": 1.71053923582833e-06,
      "loss": 0.5405,
      "step": 15171
    },
    {
      "epoch": 0.8187361718201932,
      "grad_norm": 1.0129089694326077,
      "learning_rate": 1.7101278486973033e-06,
      "loss": 0.4254,
      "step": 15172
    },
    {
      "epoch": 0.8187901354487076,
      "grad_norm": 0.9514323351833918,
      "learning_rate": 1.7097165704910319e-06,
      "loss": 0.3767,
      "step": 15173
    },
    {
      "epoch": 0.818844099077222,
      "grad_norm": 0.9661659430917452,
      "learning_rate": 1.7093054012213355e-06,
      "loss": 0.5089,
      "step": 15174
    },
    {
      "epoch": 0.8188980627057363,
      "grad_norm": 0.9268128258056492,
      "learning_rate": 1.7088943409000336e-06,
      "loss": 0.3478,
      "step": 15175
    },
    {
      "epoch": 0.8189520263342507,
      "grad_norm": 1.0937691238798186,
      "learning_rate": 1.7084833895389374e-06,
      "loss": 0.5081,
      "step": 15176
    },
    {
      "epoch": 0.8190059899627651,
      "grad_norm": 1.0125312335778345,
      "learning_rate": 1.7080725471498603e-06,
      "loss": 0.3877,
      "step": 15177
    },
    {
      "epoch": 0.8190599535912795,
      "grad_norm": 0.8845595572174,
      "learning_rate": 1.707661813744609e-06,
      "loss": 0.4497,
      "step": 15178
    },
    {
      "epoch": 0.8191139172197939,
      "grad_norm": 1.0211368833174785,
      "learning_rate": 1.707251189334989e-06,
      "loss": 0.468,
      "step": 15179
    },
    {
      "epoch": 0.8191678808483083,
      "grad_norm": 0.9534858323104747,
      "learning_rate": 1.706840673932803e-06,
      "loss": 0.3681,
      "step": 15180
    },
    {
      "epoch": 0.8192218444768227,
      "grad_norm": 0.9160056799040834,
      "learning_rate": 1.7064302675498481e-06,
      "loss": 0.4413,
      "step": 15181
    },
    {
      "epoch": 0.819275808105337,
      "grad_norm": 1.0341806283417736,
      "learning_rate": 1.7060199701979218e-06,
      "loss": 0.5375,
      "step": 15182
    },
    {
      "epoch": 0.8193297717338514,
      "grad_norm": 1.0673690645939127,
      "learning_rate": 1.705609781888814e-06,
      "loss": 0.4299,
      "step": 15183
    },
    {
      "epoch": 0.8193837353623658,
      "grad_norm": 1.1681129524617253,
      "learning_rate": 1.7051997026343148e-06,
      "loss": 0.5788,
      "step": 15184
    },
    {
      "epoch": 0.8194376989908801,
      "grad_norm": 0.8467781199960405,
      "learning_rate": 1.7047897324462104e-06,
      "loss": 0.3111,
      "step": 15185
    },
    {
      "epoch": 0.8194916626193945,
      "grad_norm": 1.0322381942757666,
      "learning_rate": 1.7043798713362842e-06,
      "loss": 0.5266,
      "step": 15186
    },
    {
      "epoch": 0.8195456262479089,
      "grad_norm": 1.33939831830161,
      "learning_rate": 1.7039701193163156e-06,
      "loss": 0.5874,
      "step": 15187
    },
    {
      "epoch": 0.8195995898764233,
      "grad_norm": 1.1074320809595735,
      "learning_rate": 1.7035604763980812e-06,
      "loss": 0.5073,
      "step": 15188
    },
    {
      "epoch": 0.8196535535049376,
      "grad_norm": 1.1102030340292126,
      "learning_rate": 1.7031509425933546e-06,
      "loss": 0.457,
      "step": 15189
    },
    {
      "epoch": 0.819707517133452,
      "grad_norm": 0.9555393779954622,
      "learning_rate": 1.7027415179139068e-06,
      "loss": 0.4502,
      "step": 15190
    },
    {
      "epoch": 0.8197614807619664,
      "grad_norm": 1.1164453235548275,
      "learning_rate": 1.7023322023715048e-06,
      "loss": 0.4891,
      "step": 15191
    },
    {
      "epoch": 0.8198154443904808,
      "grad_norm": 1.072119526866778,
      "learning_rate": 1.701922995977912e-06,
      "loss": 0.4004,
      "step": 15192
    },
    {
      "epoch": 0.8198694080189952,
      "grad_norm": 0.96430374649677,
      "learning_rate": 1.70151389874489e-06,
      "loss": 0.4469,
      "step": 15193
    },
    {
      "epoch": 0.8199233716475096,
      "grad_norm": 0.9580832223783925,
      "learning_rate": 1.701104910684197e-06,
      "loss": 0.6075,
      "step": 15194
    },
    {
      "epoch": 0.819977335276024,
      "grad_norm": 1.0872272543927284,
      "learning_rate": 1.7006960318075872e-06,
      "loss": 0.4832,
      "step": 15195
    },
    {
      "epoch": 0.8200312989045383,
      "grad_norm": 1.1923898264720125,
      "learning_rate": 1.7002872621268135e-06,
      "loss": 0.3911,
      "step": 15196
    },
    {
      "epoch": 0.8200852625330527,
      "grad_norm": 1.0283255162914302,
      "learning_rate": 1.6998786016536223e-06,
      "loss": 0.3966,
      "step": 15197
    },
    {
      "epoch": 0.8201392261615671,
      "grad_norm": 1.1185873154251138,
      "learning_rate": 1.699470050399759e-06,
      "loss": 0.497,
      "step": 15198
    },
    {
      "epoch": 0.8201931897900815,
      "grad_norm": 1.030085881861573,
      "learning_rate": 1.6990616083769678e-06,
      "loss": 0.4303,
      "step": 15199
    },
    {
      "epoch": 0.8202471534185959,
      "grad_norm": 0.8852793080058973,
      "learning_rate": 1.6986532755969854e-06,
      "loss": 0.3803,
      "step": 15200
    },
    {
      "epoch": 0.8203011170471103,
      "grad_norm": 0.9893078256397938,
      "learning_rate": 1.69824505207155e-06,
      "loss": 0.3654,
      "step": 15201
    },
    {
      "epoch": 0.8203550806756247,
      "grad_norm": 0.9727138521177023,
      "learning_rate": 1.6978369378123922e-06,
      "loss": 0.3476,
      "step": 15202
    },
    {
      "epoch": 0.820409044304139,
      "grad_norm": 0.7612902500712936,
      "learning_rate": 1.697428932831244e-06,
      "loss": 0.3039,
      "step": 15203
    },
    {
      "epoch": 0.8204630079326534,
      "grad_norm": 0.9597916191809853,
      "learning_rate": 1.6970210371398299e-06,
      "loss": 0.4491,
      "step": 15204
    },
    {
      "epoch": 0.8205169715611678,
      "grad_norm": 0.9444731911654156,
      "learning_rate": 1.6966132507498734e-06,
      "loss": 0.395,
      "step": 15205
    },
    {
      "epoch": 0.8205709351896822,
      "grad_norm": 0.7937933132303412,
      "learning_rate": 1.6962055736730949e-06,
      "loss": 0.3291,
      "step": 15206
    },
    {
      "epoch": 0.8206248988181966,
      "grad_norm": 1.0115446334117435,
      "learning_rate": 1.6957980059212114e-06,
      "loss": 0.6179,
      "step": 15207
    },
    {
      "epoch": 0.820678862446711,
      "grad_norm": 1.0002018211489654,
      "learning_rate": 1.6953905475059379e-06,
      "loss": 0.4277,
      "step": 15208
    },
    {
      "epoch": 0.8207328260752254,
      "grad_norm": 1.1708166516554661,
      "learning_rate": 1.6949831984389843e-06,
      "loss": 0.5028,
      "step": 15209
    },
    {
      "epoch": 0.8207867897037396,
      "grad_norm": 1.0714092071168877,
      "learning_rate": 1.6945759587320587e-06,
      "loss": 0.5435,
      "step": 15210
    },
    {
      "epoch": 0.820840753332254,
      "grad_norm": 0.9961441919936915,
      "learning_rate": 1.6941688283968644e-06,
      "loss": 0.3604,
      "step": 15211
    },
    {
      "epoch": 0.8208947169607684,
      "grad_norm": 1.0255774522573817,
      "learning_rate": 1.6937618074451034e-06,
      "loss": 0.5432,
      "step": 15212
    },
    {
      "epoch": 0.8209486805892828,
      "grad_norm": 1.0831078464151398,
      "learning_rate": 1.6933548958884744e-06,
      "loss": 0.3171,
      "step": 15213
    },
    {
      "epoch": 0.8210026442177972,
      "grad_norm": 1.0068235602847235,
      "learning_rate": 1.6929480937386716e-06,
      "loss": 0.4286,
      "step": 15214
    },
    {
      "epoch": 0.8210566078463116,
      "grad_norm": 1.0575133589707755,
      "learning_rate": 1.6925414010073876e-06,
      "loss": 0.4495,
      "step": 15215
    },
    {
      "epoch": 0.8211105714748259,
      "grad_norm": 0.7346566803368823,
      "learning_rate": 1.6921348177063116e-06,
      "loss": 0.2661,
      "step": 15216
    },
    {
      "epoch": 0.8211645351033403,
      "grad_norm": 1.2708810179683256,
      "learning_rate": 1.6917283438471276e-06,
      "loss": 0.5395,
      "step": 15217
    },
    {
      "epoch": 0.8212184987318547,
      "grad_norm": 1.0041273748397244,
      "learning_rate": 1.6913219794415191e-06,
      "loss": 0.4293,
      "step": 15218
    },
    {
      "epoch": 0.8212724623603691,
      "grad_norm": 1.1481636173389942,
      "learning_rate": 1.6909157245011653e-06,
      "loss": 0.5606,
      "step": 15219
    },
    {
      "epoch": 0.8213264259888835,
      "grad_norm": 1.0486392581911248,
      "learning_rate": 1.6905095790377424e-06,
      "loss": 0.5641,
      "step": 15220
    },
    {
      "epoch": 0.8213803896173979,
      "grad_norm": 0.9368245880370383,
      "learning_rate": 1.6901035430629229e-06,
      "loss": 0.41,
      "step": 15221
    },
    {
      "epoch": 0.8214343532459123,
      "grad_norm": 1.1266425989379656,
      "learning_rate": 1.6896976165883778e-06,
      "loss": 0.5366,
      "step": 15222
    },
    {
      "epoch": 0.8214883168744266,
      "grad_norm": 0.9727424893779304,
      "learning_rate": 1.689291799625774e-06,
      "loss": 0.3264,
      "step": 15223
    },
    {
      "epoch": 0.821542280502941,
      "grad_norm": 1.040179566534727,
      "learning_rate": 1.6888860921867733e-06,
      "loss": 0.4803,
      "step": 15224
    },
    {
      "epoch": 0.8215962441314554,
      "grad_norm": 1.1701016359833818,
      "learning_rate": 1.6884804942830373e-06,
      "loss": 0.4633,
      "step": 15225
    },
    {
      "epoch": 0.8216502077599698,
      "grad_norm": 0.7011096013236991,
      "learning_rate": 1.688075005926223e-06,
      "loss": 0.2525,
      "step": 15226
    },
    {
      "epoch": 0.8217041713884842,
      "grad_norm": 0.9905039077018664,
      "learning_rate": 1.6876696271279846e-06,
      "loss": 0.5482,
      "step": 15227
    },
    {
      "epoch": 0.8217581350169986,
      "grad_norm": 1.0803946995998157,
      "learning_rate": 1.6872643578999737e-06,
      "loss": 0.4277,
      "step": 15228
    },
    {
      "epoch": 0.821812098645513,
      "grad_norm": 1.0400495994480834,
      "learning_rate": 1.686859198253837e-06,
      "loss": 0.5455,
      "step": 15229
    },
    {
      "epoch": 0.8218660622740273,
      "grad_norm": 1.010128671237372,
      "learning_rate": 1.6864541482012215e-06,
      "loss": 0.4466,
      "step": 15230
    },
    {
      "epoch": 0.8219200259025417,
      "grad_norm": 0.9575118245129701,
      "learning_rate": 1.6860492077537654e-06,
      "loss": 0.363,
      "step": 15231
    },
    {
      "epoch": 0.821973989531056,
      "grad_norm": 1.0234621837132485,
      "learning_rate": 1.685644376923109e-06,
      "loss": 0.4032,
      "step": 15232
    },
    {
      "epoch": 0.8220279531595704,
      "grad_norm": 1.2004785287443844,
      "learning_rate": 1.6852396557208879e-06,
      "loss": 0.5484,
      "step": 15233
    },
    {
      "epoch": 0.8220819167880848,
      "grad_norm": 1.078887260609283,
      "learning_rate": 1.684835044158733e-06,
      "loss": 0.4227,
      "step": 15234
    },
    {
      "epoch": 0.8221358804165992,
      "grad_norm": 1.4590680628441808,
      "learning_rate": 1.6844305422482738e-06,
      "loss": 0.691,
      "step": 15235
    },
    {
      "epoch": 0.8221898440451136,
      "grad_norm": 1.2480920491415,
      "learning_rate": 1.6840261500011362e-06,
      "loss": 0.4859,
      "step": 15236
    },
    {
      "epoch": 0.8222438076736279,
      "grad_norm": 1.0346299656002698,
      "learning_rate": 1.6836218674289435e-06,
      "loss": 0.4868,
      "step": 15237
    },
    {
      "epoch": 0.8222977713021423,
      "grad_norm": 1.0065199045088573,
      "learning_rate": 1.6832176945433137e-06,
      "loss": 0.3191,
      "step": 15238
    },
    {
      "epoch": 0.8223517349306567,
      "grad_norm": 0.9489358030933921,
      "learning_rate": 1.6828136313558646e-06,
      "loss": 0.4306,
      "step": 15239
    },
    {
      "epoch": 0.8224056985591711,
      "grad_norm": 0.792785621453142,
      "learning_rate": 1.6824096778782082e-06,
      "loss": 0.2892,
      "step": 15240
    },
    {
      "epoch": 0.8224596621876855,
      "grad_norm": 0.9661409010971205,
      "learning_rate": 1.6820058341219553e-06,
      "loss": 0.4272,
      "step": 15241
    },
    {
      "epoch": 0.8225136258161999,
      "grad_norm": 0.8564701058663273,
      "learning_rate": 1.681602100098712e-06,
      "loss": 0.4342,
      "step": 15242
    },
    {
      "epoch": 0.8225675894447143,
      "grad_norm": 1.022451030645694,
      "learning_rate": 1.6811984758200833e-06,
      "loss": 0.4564,
      "step": 15243
    },
    {
      "epoch": 0.8226215530732286,
      "grad_norm": 1.049153988759497,
      "learning_rate": 1.68079496129767e-06,
      "loss": 0.4529,
      "step": 15244
    },
    {
      "epoch": 0.822675516701743,
      "grad_norm": 1.1292566559492871,
      "learning_rate": 1.6803915565430672e-06,
      "loss": 0.5754,
      "step": 15245
    },
    {
      "epoch": 0.8227294803302574,
      "grad_norm": 0.9079465948219066,
      "learning_rate": 1.6799882615678703e-06,
      "loss": 0.4485,
      "step": 15246
    },
    {
      "epoch": 0.8227834439587718,
      "grad_norm": 1.0155934404026505,
      "learning_rate": 1.6795850763836708e-06,
      "loss": 0.5628,
      "step": 15247
    },
    {
      "epoch": 0.8228374075872862,
      "grad_norm": 1.1764315046248373,
      "learning_rate": 1.6791820010020568e-06,
      "loss": 0.5022,
      "step": 15248
    },
    {
      "epoch": 0.8228913712158006,
      "grad_norm": 1.100738548215173,
      "learning_rate": 1.678779035434612e-06,
      "loss": 0.5808,
      "step": 15249
    },
    {
      "epoch": 0.822945334844315,
      "grad_norm": 0.9849390506516633,
      "learning_rate": 1.6783761796929193e-06,
      "loss": 0.4148,
      "step": 15250
    },
    {
      "epoch": 0.8229992984728293,
      "grad_norm": 0.8745954548936484,
      "learning_rate": 1.6779734337885578e-06,
      "loss": 0.4044,
      "step": 15251
    },
    {
      "epoch": 0.8230532621013437,
      "grad_norm": 0.736396353759655,
      "learning_rate": 1.6775707977331008e-06,
      "loss": 0.2608,
      "step": 15252
    },
    {
      "epoch": 0.8231072257298581,
      "grad_norm": 0.920702597703036,
      "learning_rate": 1.677168271538121e-06,
      "loss": 0.4324,
      "step": 15253
    },
    {
      "epoch": 0.8231611893583725,
      "grad_norm": 1.0488014199500135,
      "learning_rate": 1.6767658552151878e-06,
      "loss": 0.4942,
      "step": 15254
    },
    {
      "epoch": 0.8232151529868869,
      "grad_norm": 0.9593326466663169,
      "learning_rate": 1.6763635487758668e-06,
      "loss": 0.4176,
      "step": 15255
    },
    {
      "epoch": 0.8232691166154013,
      "grad_norm": 1.0052775996662797,
      "learning_rate": 1.6759613522317214e-06,
      "loss": 0.3579,
      "step": 15256
    },
    {
      "epoch": 0.8233230802439157,
      "grad_norm": 0.8962863776418115,
      "learning_rate": 1.6755592655943103e-06,
      "loss": 0.3879,
      "step": 15257
    },
    {
      "epoch": 0.8233770438724299,
      "grad_norm": 1.158062090981049,
      "learning_rate": 1.675157288875191e-06,
      "loss": 0.5486,
      "step": 15258
    },
    {
      "epoch": 0.8234310075009443,
      "grad_norm": 0.8762405266793526,
      "learning_rate": 1.6747554220859153e-06,
      "loss": 0.3529,
      "step": 15259
    },
    {
      "epoch": 0.8234849711294587,
      "grad_norm": 1.1217545633979755,
      "learning_rate": 1.6743536652380344e-06,
      "loss": 0.4672,
      "step": 15260
    },
    {
      "epoch": 0.8235389347579731,
      "grad_norm": 0.9562218278969031,
      "learning_rate": 1.6739520183430934e-06,
      "loss": 0.4804,
      "step": 15261
    },
    {
      "epoch": 0.8235928983864875,
      "grad_norm": 1.1282345637610869,
      "learning_rate": 1.6735504814126385e-06,
      "loss": 0.482,
      "step": 15262
    },
    {
      "epoch": 0.8236468620150019,
      "grad_norm": 1.0658878244195737,
      "learning_rate": 1.6731490544582086e-06,
      "loss": 0.4978,
      "step": 15263
    },
    {
      "epoch": 0.8237008256435163,
      "grad_norm": 1.0209844104943446,
      "learning_rate": 1.6727477374913415e-06,
      "loss": 0.6033,
      "step": 15264
    },
    {
      "epoch": 0.8237547892720306,
      "grad_norm": 1.0675909764184786,
      "learning_rate": 1.6723465305235727e-06,
      "loss": 0.4,
      "step": 15265
    },
    {
      "epoch": 0.823808752900545,
      "grad_norm": 1.0741853297900583,
      "learning_rate": 1.6719454335664315e-06,
      "loss": 0.4067,
      "step": 15266
    },
    {
      "epoch": 0.8238627165290594,
      "grad_norm": 0.9782178042780708,
      "learning_rate": 1.6715444466314462e-06,
      "loss": 0.4295,
      "step": 15267
    },
    {
      "epoch": 0.8239166801575738,
      "grad_norm": 1.0612408994113318,
      "learning_rate": 1.671143569730142e-06,
      "loss": 0.3816,
      "step": 15268
    },
    {
      "epoch": 0.8239706437860882,
      "grad_norm": 1.1400180863833498,
      "learning_rate": 1.6707428028740402e-06,
      "loss": 0.64,
      "step": 15269
    },
    {
      "epoch": 0.8240246074146026,
      "grad_norm": 0.919419911835329,
      "learning_rate": 1.67034214607466e-06,
      "loss": 0.3435,
      "step": 15270
    },
    {
      "epoch": 0.824078571043117,
      "grad_norm": 1.054872539270685,
      "learning_rate": 1.6699415993435154e-06,
      "loss": 0.4764,
      "step": 15271
    },
    {
      "epoch": 0.8241325346716313,
      "grad_norm": 1.0658727204629213,
      "learning_rate": 1.6695411626921205e-06,
      "loss": 0.3804,
      "step": 15272
    },
    {
      "epoch": 0.8241864983001457,
      "grad_norm": 0.9353168687161953,
      "learning_rate": 1.6691408361319823e-06,
      "loss": 0.3439,
      "step": 15273
    },
    {
      "epoch": 0.8242404619286601,
      "grad_norm": 1.3723189725860312,
      "learning_rate": 1.6687406196746063e-06,
      "loss": 0.5516,
      "step": 15274
    },
    {
      "epoch": 0.8242944255571745,
      "grad_norm": 0.9791749248950546,
      "learning_rate": 1.668340513331497e-06,
      "loss": 0.4388,
      "step": 15275
    },
    {
      "epoch": 0.8243483891856889,
      "grad_norm": 1.134389714060165,
      "learning_rate": 1.6679405171141528e-06,
      "loss": 0.6094,
      "step": 15276
    },
    {
      "epoch": 0.8244023528142033,
      "grad_norm": 1.2409679470973485,
      "learning_rate": 1.6675406310340698e-06,
      "loss": 0.4468,
      "step": 15277
    },
    {
      "epoch": 0.8244563164427177,
      "grad_norm": 0.877506017831225,
      "learning_rate": 1.6671408551027419e-06,
      "loss": 0.396,
      "step": 15278
    },
    {
      "epoch": 0.824510280071232,
      "grad_norm": 0.8884766588142108,
      "learning_rate": 1.6667411893316592e-06,
      "loss": 0.4363,
      "step": 15279
    },
    {
      "epoch": 0.8245642436997463,
      "grad_norm": 1.1552380995314846,
      "learning_rate": 1.6663416337323074e-06,
      "loss": 0.5106,
      "step": 15280
    },
    {
      "epoch": 0.8246182073282607,
      "grad_norm": 0.945257535211169,
      "learning_rate": 1.6659421883161709e-06,
      "loss": 0.3614,
      "step": 15281
    },
    {
      "epoch": 0.8246721709567751,
      "grad_norm": 0.9657387074278247,
      "learning_rate": 1.6655428530947293e-06,
      "loss": 0.3992,
      "step": 15282
    },
    {
      "epoch": 0.8247261345852895,
      "grad_norm": 1.1451314236179333,
      "learning_rate": 1.6651436280794606e-06,
      "loss": 0.5964,
      "step": 15283
    },
    {
      "epoch": 0.8247800982138039,
      "grad_norm": 0.9482422709385145,
      "learning_rate": 1.6647445132818393e-06,
      "loss": 0.4665,
      "step": 15284
    },
    {
      "epoch": 0.8248340618423182,
      "grad_norm": 0.8469977240633774,
      "learning_rate": 1.6643455087133354e-06,
      "loss": 0.3531,
      "step": 15285
    },
    {
      "epoch": 0.8248880254708326,
      "grad_norm": 0.8494720351294899,
      "learning_rate": 1.6639466143854177e-06,
      "loss": 0.344,
      "step": 15286
    },
    {
      "epoch": 0.824941989099347,
      "grad_norm": 1.0006468357966578,
      "learning_rate": 1.6635478303095508e-06,
      "loss": 0.3963,
      "step": 15287
    },
    {
      "epoch": 0.8249959527278614,
      "grad_norm": 1.1825062703229994,
      "learning_rate": 1.6631491564971954e-06,
      "loss": 0.5821,
      "step": 15288
    },
    {
      "epoch": 0.8250499163563758,
      "grad_norm": 0.9703370374855572,
      "learning_rate": 1.6627505929598098e-06,
      "loss": 0.4698,
      "step": 15289
    },
    {
      "epoch": 0.8251038799848902,
      "grad_norm": 0.7704826166576337,
      "learning_rate": 1.6623521397088494e-06,
      "loss": 0.309,
      "step": 15290
    },
    {
      "epoch": 0.8251578436134046,
      "grad_norm": 0.7475413397779399,
      "learning_rate": 1.6619537967557658e-06,
      "loss": 0.3164,
      "step": 15291
    },
    {
      "epoch": 0.8252118072419189,
      "grad_norm": 0.9815406048652813,
      "learning_rate": 1.6615555641120087e-06,
      "loss": 0.4722,
      "step": 15292
    },
    {
      "epoch": 0.8252657708704333,
      "grad_norm": 1.1830777588943997,
      "learning_rate": 1.661157441789023e-06,
      "loss": 0.4208,
      "step": 15293
    },
    {
      "epoch": 0.8253197344989477,
      "grad_norm": 1.002791344793153,
      "learning_rate": 1.660759429798251e-06,
      "loss": 0.4328,
      "step": 15294
    },
    {
      "epoch": 0.8253736981274621,
      "grad_norm": 0.9963884093504602,
      "learning_rate": 1.6603615281511327e-06,
      "loss": 0.3678,
      "step": 15295
    },
    {
      "epoch": 0.8254276617559765,
      "grad_norm": 1.0621158113636961,
      "learning_rate": 1.6599637368591031e-06,
      "loss": 0.4961,
      "step": 15296
    },
    {
      "epoch": 0.8254816253844909,
      "grad_norm": 1.1883653775029228,
      "learning_rate": 1.659566055933597e-06,
      "loss": 0.5059,
      "step": 15297
    },
    {
      "epoch": 0.8255355890130053,
      "grad_norm": 0.9616357470863179,
      "learning_rate": 1.6591684853860419e-06,
      "loss": 0.4451,
      "step": 15298
    },
    {
      "epoch": 0.8255895526415196,
      "grad_norm": 0.954410842348908,
      "learning_rate": 1.6587710252278666e-06,
      "loss": 0.3717,
      "step": 15299
    },
    {
      "epoch": 0.825643516270034,
      "grad_norm": 1.0226124432135435,
      "learning_rate": 1.658373675470492e-06,
      "loss": 0.5448,
      "step": 15300
    },
    {
      "epoch": 0.8256974798985484,
      "grad_norm": 1.051332942469408,
      "learning_rate": 1.6579764361253402e-06,
      "loss": 0.4587,
      "step": 15301
    },
    {
      "epoch": 0.8257514435270628,
      "grad_norm": 1.0064378918233974,
      "learning_rate": 1.6575793072038277e-06,
      "loss": 0.5064,
      "step": 15302
    },
    {
      "epoch": 0.8258054071555772,
      "grad_norm": 0.8280355004178112,
      "learning_rate": 1.6571822887173682e-06,
      "loss": 0.3153,
      "step": 15303
    },
    {
      "epoch": 0.8258593707840916,
      "grad_norm": 0.9433685809385743,
      "learning_rate": 1.6567853806773725e-06,
      "loss": 0.4151,
      "step": 15304
    },
    {
      "epoch": 0.825913334412606,
      "grad_norm": 1.0278959797644276,
      "learning_rate": 1.6563885830952481e-06,
      "loss": 0.4695,
      "step": 15305
    },
    {
      "epoch": 0.8259672980411202,
      "grad_norm": 1.0951052825092307,
      "learning_rate": 1.6559918959824007e-06,
      "loss": 0.6042,
      "step": 15306
    },
    {
      "epoch": 0.8260212616696346,
      "grad_norm": 1.1560765268794617,
      "learning_rate": 1.6555953193502294e-06,
      "loss": 0.4197,
      "step": 15307
    },
    {
      "epoch": 0.826075225298149,
      "grad_norm": 0.9341647133161034,
      "learning_rate": 1.6551988532101334e-06,
      "loss": 0.358,
      "step": 15308
    },
    {
      "epoch": 0.8261291889266634,
      "grad_norm": 0.9596059882092927,
      "learning_rate": 1.6548024975735068e-06,
      "loss": 0.4001,
      "step": 15309
    },
    {
      "epoch": 0.8261831525551778,
      "grad_norm": 0.9496388432571761,
      "learning_rate": 1.6544062524517417e-06,
      "loss": 0.4992,
      "step": 15310
    },
    {
      "epoch": 0.8262371161836922,
      "grad_norm": 1.0679020293879193,
      "learning_rate": 1.6540101178562268e-06,
      "loss": 0.4602,
      "step": 15311
    },
    {
      "epoch": 0.8262910798122066,
      "grad_norm": 1.0117689512059747,
      "learning_rate": 1.6536140937983469e-06,
      "loss": 0.5547,
      "step": 15312
    },
    {
      "epoch": 0.8263450434407209,
      "grad_norm": 0.9387888661809785,
      "learning_rate": 1.6532181802894858e-06,
      "loss": 0.3941,
      "step": 15313
    },
    {
      "epoch": 0.8263990070692353,
      "grad_norm": 1.038077720108746,
      "learning_rate": 1.6528223773410202e-06,
      "loss": 0.4535,
      "step": 15314
    },
    {
      "epoch": 0.8264529706977497,
      "grad_norm": 0.8492798883185232,
      "learning_rate": 1.6524266849643263e-06,
      "loss": 0.4195,
      "step": 15315
    },
    {
      "epoch": 0.8265069343262641,
      "grad_norm": 0.984466346648735,
      "learning_rate": 1.6520311031707776e-06,
      "loss": 0.4196,
      "step": 15316
    },
    {
      "epoch": 0.8265608979547785,
      "grad_norm": 0.9525329023144323,
      "learning_rate": 1.6516356319717424e-06,
      "loss": 0.3348,
      "step": 15317
    },
    {
      "epoch": 0.8266148615832929,
      "grad_norm": 1.0401583310194717,
      "learning_rate": 1.651240271378588e-06,
      "loss": 0.3512,
      "step": 15318
    },
    {
      "epoch": 0.8266688252118073,
      "grad_norm": 1.121124164012634,
      "learning_rate": 1.6508450214026779e-06,
      "loss": 0.643,
      "step": 15319
    },
    {
      "epoch": 0.8267227888403216,
      "grad_norm": 1.3295100307235266,
      "learning_rate": 1.6504498820553708e-06,
      "loss": 0.5374,
      "step": 15320
    },
    {
      "epoch": 0.826776752468836,
      "grad_norm": 1.022600458689407,
      "learning_rate": 1.6500548533480237e-06,
      "loss": 0.3431,
      "step": 15321
    },
    {
      "epoch": 0.8268307160973504,
      "grad_norm": 0.952025877056846,
      "learning_rate": 1.6496599352919895e-06,
      "loss": 0.3692,
      "step": 15322
    },
    {
      "epoch": 0.8268846797258648,
      "grad_norm": 0.9850382153017488,
      "learning_rate": 1.6492651278986199e-06,
      "loss": 0.4874,
      "step": 15323
    },
    {
      "epoch": 0.8269386433543792,
      "grad_norm": 0.9754257827735962,
      "learning_rate": 1.6488704311792614e-06,
      "loss": 0.4086,
      "step": 15324
    },
    {
      "epoch": 0.8269926069828936,
      "grad_norm": 1.0875429196686848,
      "learning_rate": 1.6484758451452576e-06,
      "loss": 0.5221,
      "step": 15325
    },
    {
      "epoch": 0.827046570611408,
      "grad_norm": 0.9844649431257165,
      "learning_rate": 1.64808136980795e-06,
      "loss": 0.3816,
      "step": 15326
    },
    {
      "epoch": 0.8271005342399222,
      "grad_norm": 0.9131930073443923,
      "learning_rate": 1.647687005178677e-06,
      "loss": 0.4081,
      "step": 15327
    },
    {
      "epoch": 0.8271544978684366,
      "grad_norm": 0.9712443854969908,
      "learning_rate": 1.6472927512687703e-06,
      "loss": 0.4141,
      "step": 15328
    },
    {
      "epoch": 0.827208461496951,
      "grad_norm": 1.0355973072206333,
      "learning_rate": 1.6468986080895632e-06,
      "loss": 0.3999,
      "step": 15329
    },
    {
      "epoch": 0.8272624251254654,
      "grad_norm": 0.9727168487786289,
      "learning_rate": 1.6465045756523835e-06,
      "loss": 0.4082,
      "step": 15330
    },
    {
      "epoch": 0.8273163887539798,
      "grad_norm": 1.0137916292130777,
      "learning_rate": 1.6461106539685557e-06,
      "loss": 0.4452,
      "step": 15331
    },
    {
      "epoch": 0.8273703523824942,
      "grad_norm": 0.9848647473256945,
      "learning_rate": 1.645716843049402e-06,
      "loss": 0.3723,
      "step": 15332
    },
    {
      "epoch": 0.8274243160110086,
      "grad_norm": 1.0315841005455213,
      "learning_rate": 1.6453231429062406e-06,
      "loss": 0.5252,
      "step": 15333
    },
    {
      "epoch": 0.8274782796395229,
      "grad_norm": 1.115264253986196,
      "learning_rate": 1.6449295535503879e-06,
      "loss": 0.5367,
      "step": 15334
    },
    {
      "epoch": 0.8275322432680373,
      "grad_norm": 1.1932735938232533,
      "learning_rate": 1.644536074993154e-06,
      "loss": 0.5656,
      "step": 15335
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 1.1770821675523802,
      "learning_rate": 1.6441427072458493e-06,
      "loss": 0.5771,
      "step": 15336
    },
    {
      "epoch": 0.8276401705250661,
      "grad_norm": 1.1470761731590955,
      "learning_rate": 1.6437494503197793e-06,
      "loss": 0.4973,
      "step": 15337
    },
    {
      "epoch": 0.8276941341535805,
      "grad_norm": 1.112590310917873,
      "learning_rate": 1.6433563042262462e-06,
      "loss": 0.4466,
      "step": 15338
    },
    {
      "epoch": 0.8277480977820949,
      "grad_norm": 1.179742942604993,
      "learning_rate": 1.6429632689765495e-06,
      "loss": 0.7439,
      "step": 15339
    },
    {
      "epoch": 0.8278020614106093,
      "grad_norm": 0.9418536515824265,
      "learning_rate": 1.6425703445819868e-06,
      "loss": 0.5878,
      "step": 15340
    },
    {
      "epoch": 0.8278560250391236,
      "grad_norm": 1.1606521531608371,
      "learning_rate": 1.6421775310538491e-06,
      "loss": 0.5657,
      "step": 15341
    },
    {
      "epoch": 0.827909988667638,
      "grad_norm": 0.8111427005460786,
      "learning_rate": 1.6417848284034277e-06,
      "loss": 0.3825,
      "step": 15342
    },
    {
      "epoch": 0.8279639522961524,
      "grad_norm": 0.82409662291311,
      "learning_rate": 1.6413922366420083e-06,
      "loss": 0.2925,
      "step": 15343
    },
    {
      "epoch": 0.8280179159246668,
      "grad_norm": 1.0134838940954902,
      "learning_rate": 1.640999755780875e-06,
      "loss": 0.4218,
      "step": 15344
    },
    {
      "epoch": 0.8280718795531812,
      "grad_norm": 1.359957018304399,
      "learning_rate": 1.6406073858313082e-06,
      "loss": 0.492,
      "step": 15345
    },
    {
      "epoch": 0.8281258431816956,
      "grad_norm": 0.9848664247455619,
      "learning_rate": 1.6402151268045845e-06,
      "loss": 0.46,
      "step": 15346
    },
    {
      "epoch": 0.82817980681021,
      "grad_norm": 0.9882096098383427,
      "learning_rate": 1.6398229787119783e-06,
      "loss": 0.5366,
      "step": 15347
    },
    {
      "epoch": 0.8282337704387243,
      "grad_norm": 1.1467199926370104,
      "learning_rate": 1.6394309415647607e-06,
      "loss": 0.6507,
      "step": 15348
    },
    {
      "epoch": 0.8282877340672387,
      "grad_norm": 1.1089960433376567,
      "learning_rate": 1.639039015374198e-06,
      "loss": 0.405,
      "step": 15349
    },
    {
      "epoch": 0.828341697695753,
      "grad_norm": 0.9034260353095049,
      "learning_rate": 1.6386472001515551e-06,
      "loss": 0.4843,
      "step": 15350
    },
    {
      "epoch": 0.8283956613242675,
      "grad_norm": 0.9372743728434907,
      "learning_rate": 1.6382554959080935e-06,
      "loss": 0.326,
      "step": 15351
    },
    {
      "epoch": 0.8284496249527818,
      "grad_norm": 0.9203627778104434,
      "learning_rate": 1.6378639026550709e-06,
      "loss": 0.476,
      "step": 15352
    },
    {
      "epoch": 0.8285035885812962,
      "grad_norm": 0.9919834063939926,
      "learning_rate": 1.6374724204037424e-06,
      "loss": 0.3753,
      "step": 15353
    },
    {
      "epoch": 0.8285575522098105,
      "grad_norm": 1.0934076519936318,
      "learning_rate": 1.6370810491653598e-06,
      "loss": 0.5406,
      "step": 15354
    },
    {
      "epoch": 0.8286115158383249,
      "grad_norm": 1.0013347753243231,
      "learning_rate": 1.63668978895117e-06,
      "loss": 0.3907,
      "step": 15355
    },
    {
      "epoch": 0.8286654794668393,
      "grad_norm": 1.1236995563846215,
      "learning_rate": 1.6362986397724195e-06,
      "loss": 0.4294,
      "step": 15356
    },
    {
      "epoch": 0.8287194430953537,
      "grad_norm": 0.993775266249401,
      "learning_rate": 1.6359076016403502e-06,
      "loss": 0.3901,
      "step": 15357
    },
    {
      "epoch": 0.8287734067238681,
      "grad_norm": 1.2536948145051494,
      "learning_rate": 1.6355166745662004e-06,
      "loss": 0.5583,
      "step": 15358
    },
    {
      "epoch": 0.8288273703523825,
      "grad_norm": 0.9206707188964893,
      "learning_rate": 1.6351258585612066e-06,
      "loss": 0.3841,
      "step": 15359
    },
    {
      "epoch": 0.8288813339808969,
      "grad_norm": 0.9173560508696751,
      "learning_rate": 1.6347351536366007e-06,
      "loss": 0.4087,
      "step": 15360
    },
    {
      "epoch": 0.8289352976094112,
      "grad_norm": 1.1409267254422841,
      "learning_rate": 1.634344559803613e-06,
      "loss": 0.4314,
      "step": 15361
    },
    {
      "epoch": 0.8289892612379256,
      "grad_norm": 1.4403809437476502,
      "learning_rate": 1.6339540770734675e-06,
      "loss": 0.588,
      "step": 15362
    },
    {
      "epoch": 0.82904322486644,
      "grad_norm": 1.1553253568377821,
      "learning_rate": 1.6335637054573885e-06,
      "loss": 0.5501,
      "step": 15363
    },
    {
      "epoch": 0.8290971884949544,
      "grad_norm": 1.279456055629867,
      "learning_rate": 1.6331734449665951e-06,
      "loss": 0.657,
      "step": 15364
    },
    {
      "epoch": 0.8291511521234688,
      "grad_norm": 1.0347977504619774,
      "learning_rate": 1.6327832956123042e-06,
      "loss": 0.4935,
      "step": 15365
    },
    {
      "epoch": 0.8292051157519832,
      "grad_norm": 1.1320971733316116,
      "learning_rate": 1.6323932574057286e-06,
      "loss": 0.4609,
      "step": 15366
    },
    {
      "epoch": 0.8292590793804976,
      "grad_norm": 1.1737440614733097,
      "learning_rate": 1.6320033303580787e-06,
      "loss": 0.5883,
      "step": 15367
    },
    {
      "epoch": 0.8293130430090119,
      "grad_norm": 1.0241309139919301,
      "learning_rate": 1.6316135144805627e-06,
      "loss": 0.4134,
      "step": 15368
    },
    {
      "epoch": 0.8293670066375263,
      "grad_norm": 1.052990435201778,
      "learning_rate": 1.631223809784382e-06,
      "loss": 0.5443,
      "step": 15369
    },
    {
      "epoch": 0.8294209702660407,
      "grad_norm": 0.9953715784078616,
      "learning_rate": 1.6308342162807378e-06,
      "loss": 0.5802,
      "step": 15370
    },
    {
      "epoch": 0.8294749338945551,
      "grad_norm": 1.1573935995493407,
      "learning_rate": 1.630444733980828e-06,
      "loss": 0.5258,
      "step": 15371
    },
    {
      "epoch": 0.8295288975230695,
      "grad_norm": 0.9760629770655556,
      "learning_rate": 1.6300553628958465e-06,
      "loss": 0.4057,
      "step": 15372
    },
    {
      "epoch": 0.8295828611515839,
      "grad_norm": 0.9405748451846002,
      "learning_rate": 1.6296661030369842e-06,
      "loss": 0.4053,
      "step": 15373
    },
    {
      "epoch": 0.8296368247800983,
      "grad_norm": 1.1409267966358656,
      "learning_rate": 1.6292769544154286e-06,
      "loss": 0.4725,
      "step": 15374
    },
    {
      "epoch": 0.8296907884086125,
      "grad_norm": 1.3473920011749605,
      "learning_rate": 1.6288879170423655e-06,
      "loss": 0.3913,
      "step": 15375
    },
    {
      "epoch": 0.8297447520371269,
      "grad_norm": 1.0210178350199894,
      "learning_rate": 1.6284989909289742e-06,
      "loss": 0.5578,
      "step": 15376
    },
    {
      "epoch": 0.8297987156656413,
      "grad_norm": 1.3074107239874468,
      "learning_rate": 1.6281101760864331e-06,
      "loss": 0.5884,
      "step": 15377
    },
    {
      "epoch": 0.8298526792941557,
      "grad_norm": 1.2864957659619,
      "learning_rate": 1.6277214725259186e-06,
      "loss": 0.5754,
      "step": 15378
    },
    {
      "epoch": 0.8299066429226701,
      "grad_norm": 1.0087017551091262,
      "learning_rate": 1.6273328802586014e-06,
      "loss": 0.44,
      "step": 15379
    },
    {
      "epoch": 0.8299606065511845,
      "grad_norm": 1.058579090990101,
      "learning_rate": 1.6269443992956506e-06,
      "loss": 0.4641,
      "step": 15380
    },
    {
      "epoch": 0.8300145701796989,
      "grad_norm": 1.0959609816684277,
      "learning_rate": 1.6265560296482308e-06,
      "loss": 0.6074,
      "step": 15381
    },
    {
      "epoch": 0.8300685338082132,
      "grad_norm": 0.9473560366849721,
      "learning_rate": 1.6261677713275056e-06,
      "loss": 0.3645,
      "step": 15382
    },
    {
      "epoch": 0.8301224974367276,
      "grad_norm": 0.9885028322314426,
      "learning_rate": 1.6257796243446316e-06,
      "loss": 0.4401,
      "step": 15383
    },
    {
      "epoch": 0.830176461065242,
      "grad_norm": 0.9124421877386184,
      "learning_rate": 1.6253915887107662e-06,
      "loss": 0.4123,
      "step": 15384
    },
    {
      "epoch": 0.8302304246937564,
      "grad_norm": 0.974782845582678,
      "learning_rate": 1.6250036644370614e-06,
      "loss": 0.4333,
      "step": 15385
    },
    {
      "epoch": 0.8302843883222708,
      "grad_norm": 0.8827787107749782,
      "learning_rate": 1.6246158515346666e-06,
      "loss": 0.4255,
      "step": 15386
    },
    {
      "epoch": 0.8303383519507852,
      "grad_norm": 1.1217396031088451,
      "learning_rate": 1.6242281500147283e-06,
      "loss": 0.5426,
      "step": 15387
    },
    {
      "epoch": 0.8303923155792996,
      "grad_norm": 0.942461516956828,
      "learning_rate": 1.6238405598883892e-06,
      "loss": 0.3866,
      "step": 15388
    },
    {
      "epoch": 0.8304462792078139,
      "grad_norm": 1.0525126029461804,
      "learning_rate": 1.6234530811667888e-06,
      "loss": 0.4491,
      "step": 15389
    },
    {
      "epoch": 0.8305002428363283,
      "grad_norm": 1.007521273389118,
      "learning_rate": 1.6230657138610636e-06,
      "loss": 0.4761,
      "step": 15390
    },
    {
      "epoch": 0.8305542064648427,
      "grad_norm": 1.0782207584626167,
      "learning_rate": 1.6226784579823473e-06,
      "loss": 0.576,
      "step": 15391
    },
    {
      "epoch": 0.8306081700933571,
      "grad_norm": 1.2713924843577549,
      "learning_rate": 1.62229131354177e-06,
      "loss": 0.4287,
      "step": 15392
    },
    {
      "epoch": 0.8306621337218715,
      "grad_norm": 1.3201762229068055,
      "learning_rate": 1.6219042805504591e-06,
      "loss": 0.4662,
      "step": 15393
    },
    {
      "epoch": 0.8307160973503859,
      "grad_norm": 1.0997816194577656,
      "learning_rate": 1.6215173590195375e-06,
      "loss": 0.7125,
      "step": 15394
    },
    {
      "epoch": 0.8307700609789003,
      "grad_norm": 1.4264876999674316,
      "learning_rate": 1.621130548960125e-06,
      "loss": 0.5558,
      "step": 15395
    },
    {
      "epoch": 0.8308240246074146,
      "grad_norm": 0.9044083506550118,
      "learning_rate": 1.6207438503833403e-06,
      "loss": 0.3904,
      "step": 15396
    },
    {
      "epoch": 0.830877988235929,
      "grad_norm": 0.9593840809195331,
      "learning_rate": 1.6203572633002967e-06,
      "loss": 0.448,
      "step": 15397
    },
    {
      "epoch": 0.8309319518644434,
      "grad_norm": 0.8372278402376571,
      "learning_rate": 1.619970787722106e-06,
      "loss": 0.3658,
      "step": 15398
    },
    {
      "epoch": 0.8309859154929577,
      "grad_norm": 1.2818194586584315,
      "learning_rate": 1.619584423659875e-06,
      "loss": 0.5185,
      "step": 15399
    },
    {
      "epoch": 0.8310398791214721,
      "grad_norm": 0.9128565011562656,
      "learning_rate": 1.6191981711247088e-06,
      "loss": 0.3617,
      "step": 15400
    },
    {
      "epoch": 0.8310938427499865,
      "grad_norm": 1.111651066693923,
      "learning_rate": 1.6188120301277077e-06,
      "loss": 0.5754,
      "step": 15401
    },
    {
      "epoch": 0.8311478063785009,
      "grad_norm": 0.8489233010440381,
      "learning_rate": 1.6184260006799713e-06,
      "loss": 0.3246,
      "step": 15402
    },
    {
      "epoch": 0.8312017700070152,
      "grad_norm": 0.9727107313473871,
      "learning_rate": 1.6180400827925947e-06,
      "loss": 0.4512,
      "step": 15403
    },
    {
      "epoch": 0.8312557336355296,
      "grad_norm": 0.8534793087467182,
      "learning_rate": 1.617654276476667e-06,
      "loss": 0.4352,
      "step": 15404
    },
    {
      "epoch": 0.831309697264044,
      "grad_norm": 1.193872356677087,
      "learning_rate": 1.6172685817432786e-06,
      "loss": 0.6557,
      "step": 15405
    },
    {
      "epoch": 0.8313636608925584,
      "grad_norm": 1.0002804448382285,
      "learning_rate": 1.6168829986035142e-06,
      "loss": 0.413,
      "step": 15406
    },
    {
      "epoch": 0.8314176245210728,
      "grad_norm": 1.1382512632168367,
      "learning_rate": 1.6164975270684563e-06,
      "loss": 0.4019,
      "step": 15407
    },
    {
      "epoch": 0.8314715881495872,
      "grad_norm": 1.5034149998405386,
      "learning_rate": 1.6161121671491833e-06,
      "loss": 0.4728,
      "step": 15408
    },
    {
      "epoch": 0.8315255517781016,
      "grad_norm": 0.8020719572717433,
      "learning_rate": 1.6157269188567707e-06,
      "loss": 0.2878,
      "step": 15409
    },
    {
      "epoch": 0.8315795154066159,
      "grad_norm": 1.1982361298988031,
      "learning_rate": 1.6153417822022924e-06,
      "loss": 0.5711,
      "step": 15410
    },
    {
      "epoch": 0.8316334790351303,
      "grad_norm": 0.8731803604544135,
      "learning_rate": 1.6149567571968155e-06,
      "loss": 0.4187,
      "step": 15411
    },
    {
      "epoch": 0.8316874426636447,
      "grad_norm": 0.8721333860861324,
      "learning_rate": 1.6145718438514072e-06,
      "loss": 0.3787,
      "step": 15412
    },
    {
      "epoch": 0.8317414062921591,
      "grad_norm": 1.2286064509049628,
      "learning_rate": 1.61418704217713e-06,
      "loss": 0.495,
      "step": 15413
    },
    {
      "epoch": 0.8317953699206735,
      "grad_norm": 1.0517934024118158,
      "learning_rate": 1.6138023521850432e-06,
      "loss": 0.4482,
      "step": 15414
    },
    {
      "epoch": 0.8318493335491879,
      "grad_norm": 1.2375432854548964,
      "learning_rate": 1.6134177738862039e-06,
      "loss": 0.738,
      "step": 15415
    },
    {
      "epoch": 0.8319032971777023,
      "grad_norm": 1.1225499681073126,
      "learning_rate": 1.6130333072916646e-06,
      "loss": 0.4777,
      "step": 15416
    },
    {
      "epoch": 0.8319572608062166,
      "grad_norm": 0.8425856452675937,
      "learning_rate": 1.6126489524124764e-06,
      "loss": 0.3074,
      "step": 15417
    },
    {
      "epoch": 0.832011224434731,
      "grad_norm": 1.2060189078267347,
      "learning_rate": 1.612264709259684e-06,
      "loss": 0.5855,
      "step": 15418
    },
    {
      "epoch": 0.8320651880632454,
      "grad_norm": 0.8777255808701291,
      "learning_rate": 1.6118805778443327e-06,
      "loss": 0.3978,
      "step": 15419
    },
    {
      "epoch": 0.8321191516917598,
      "grad_norm": 0.8415611672750207,
      "learning_rate": 1.6114965581774618e-06,
      "loss": 0.378,
      "step": 15420
    },
    {
      "epoch": 0.8321731153202742,
      "grad_norm": 0.8998017275692519,
      "learning_rate": 1.6111126502701093e-06,
      "loss": 0.4069,
      "step": 15421
    },
    {
      "epoch": 0.8322270789487886,
      "grad_norm": 1.0829321595191217,
      "learning_rate": 1.6107288541333087e-06,
      "loss": 0.4262,
      "step": 15422
    },
    {
      "epoch": 0.8322810425773028,
      "grad_norm": 1.0371350334094551,
      "learning_rate": 1.6103451697780914e-06,
      "loss": 0.5284,
      "step": 15423
    },
    {
      "epoch": 0.8323350062058172,
      "grad_norm": 1.0613850534229425,
      "learning_rate": 1.6099615972154832e-06,
      "loss": 0.5963,
      "step": 15424
    },
    {
      "epoch": 0.8323889698343316,
      "grad_norm": 1.2120816671785168,
      "learning_rate": 1.6095781364565094e-06,
      "loss": 0.6594,
      "step": 15425
    },
    {
      "epoch": 0.832442933462846,
      "grad_norm": 1.3435391936281014,
      "learning_rate": 1.6091947875121905e-06,
      "loss": 0.5859,
      "step": 15426
    },
    {
      "epoch": 0.8324968970913604,
      "grad_norm": 1.0089868674008657,
      "learning_rate": 1.6088115503935454e-06,
      "loss": 0.5091,
      "step": 15427
    },
    {
      "epoch": 0.8325508607198748,
      "grad_norm": 1.1006106154910875,
      "learning_rate": 1.6084284251115877e-06,
      "loss": 0.4729,
      "step": 15428
    },
    {
      "epoch": 0.8326048243483892,
      "grad_norm": 1.251822668941307,
      "learning_rate": 1.6080454116773292e-06,
      "loss": 0.5016,
      "step": 15429
    },
    {
      "epoch": 0.8326587879769035,
      "grad_norm": 1.093031540117436,
      "learning_rate": 1.6076625101017793e-06,
      "loss": 0.4558,
      "step": 15430
    },
    {
      "epoch": 0.8327127516054179,
      "grad_norm": 0.8084222569184688,
      "learning_rate": 1.6072797203959402e-06,
      "loss": 0.28,
      "step": 15431
    },
    {
      "epoch": 0.8327667152339323,
      "grad_norm": 1.0632321543272978,
      "learning_rate": 1.6068970425708164e-06,
      "loss": 0.4485,
      "step": 15432
    },
    {
      "epoch": 0.8328206788624467,
      "grad_norm": 0.9949950780979813,
      "learning_rate": 1.6065144766374041e-06,
      "loss": 0.4578,
      "step": 15433
    },
    {
      "epoch": 0.8328746424909611,
      "grad_norm": 1.0680871728637895,
      "learning_rate": 1.6061320226067003e-06,
      "loss": 0.5031,
      "step": 15434
    },
    {
      "epoch": 0.8329286061194755,
      "grad_norm": 1.0682046669520722,
      "learning_rate": 1.6057496804896972e-06,
      "loss": 0.4953,
      "step": 15435
    },
    {
      "epoch": 0.8329825697479899,
      "grad_norm": 1.0756701641920334,
      "learning_rate": 1.6053674502973823e-06,
      "loss": 0.6223,
      "step": 15436
    },
    {
      "epoch": 0.8330365333765042,
      "grad_norm": 0.9629766957992251,
      "learning_rate": 1.6049853320407432e-06,
      "loss": 0.3974,
      "step": 15437
    },
    {
      "epoch": 0.8330904970050186,
      "grad_norm": 1.06396352588827,
      "learning_rate": 1.6046033257307605e-06,
      "loss": 0.3937,
      "step": 15438
    },
    {
      "epoch": 0.833144460633533,
      "grad_norm": 0.8147411131856522,
      "learning_rate": 1.6042214313784147e-06,
      "loss": 0.3568,
      "step": 15439
    },
    {
      "epoch": 0.8331984242620474,
      "grad_norm": 0.9533864023835867,
      "learning_rate": 1.6038396489946806e-06,
      "loss": 0.3532,
      "step": 15440
    },
    {
      "epoch": 0.8332523878905618,
      "grad_norm": 1.0302196846677762,
      "learning_rate": 1.603457978590532e-06,
      "loss": 0.402,
      "step": 15441
    },
    {
      "epoch": 0.8333063515190762,
      "grad_norm": 0.9758146464841674,
      "learning_rate": 1.6030764201769384e-06,
      "loss": 0.549,
      "step": 15442
    },
    {
      "epoch": 0.8333603151475906,
      "grad_norm": 0.8020670586479,
      "learning_rate": 1.6026949737648657e-06,
      "loss": 0.5788,
      "step": 15443
    },
    {
      "epoch": 0.8334142787761049,
      "grad_norm": 0.9608324653599793,
      "learning_rate": 1.6023136393652778e-06,
      "loss": 0.4775,
      "step": 15444
    },
    {
      "epoch": 0.8334682424046193,
      "grad_norm": 1.0236748245535792,
      "learning_rate": 1.601932416989134e-06,
      "loss": 0.4332,
      "step": 15445
    },
    {
      "epoch": 0.8335222060331337,
      "grad_norm": 1.0793062751377,
      "learning_rate": 1.6015513066473915e-06,
      "loss": 0.5708,
      "step": 15446
    },
    {
      "epoch": 0.833576169661648,
      "grad_norm": 1.2646977882591044,
      "learning_rate": 1.6011703083510036e-06,
      "loss": 0.492,
      "step": 15447
    },
    {
      "epoch": 0.8336301332901624,
      "grad_norm": 1.0332386509432254,
      "learning_rate": 1.6007894221109202e-06,
      "loss": 0.4787,
      "step": 15448
    },
    {
      "epoch": 0.8336840969186768,
      "grad_norm": 0.9728259660034616,
      "learning_rate": 1.6004086479380884e-06,
      "loss": 0.4202,
      "step": 15449
    },
    {
      "epoch": 0.8337380605471912,
      "grad_norm": 1.1584952033457256,
      "learning_rate": 1.6000279858434529e-06,
      "loss": 0.4996,
      "step": 15450
    },
    {
      "epoch": 0.8337920241757055,
      "grad_norm": 1.2649563185585226,
      "learning_rate": 1.5996474358379546e-06,
      "loss": 0.667,
      "step": 15451
    },
    {
      "epoch": 0.8338459878042199,
      "grad_norm": 0.9069482119480039,
      "learning_rate": 1.5992669979325288e-06,
      "loss": 0.4403,
      "step": 15452
    },
    {
      "epoch": 0.8338999514327343,
      "grad_norm": 0.9398816949568085,
      "learning_rate": 1.5988866721381108e-06,
      "loss": 0.3711,
      "step": 15453
    },
    {
      "epoch": 0.8339539150612487,
      "grad_norm": 0.9286961469888042,
      "learning_rate": 1.5985064584656322e-06,
      "loss": 0.3879,
      "step": 15454
    },
    {
      "epoch": 0.8340078786897631,
      "grad_norm": 1.0675522780680875,
      "learning_rate": 1.5981263569260195e-06,
      "loss": 0.4197,
      "step": 15455
    },
    {
      "epoch": 0.8340618423182775,
      "grad_norm": 1.102339013243039,
      "learning_rate": 1.5977463675301979e-06,
      "loss": 0.4749,
      "step": 15456
    },
    {
      "epoch": 0.8341158059467919,
      "grad_norm": 0.9948520773399109,
      "learning_rate": 1.5973664902890889e-06,
      "loss": 0.4192,
      "step": 15457
    },
    {
      "epoch": 0.8341697695753062,
      "grad_norm": 1.0397151584757678,
      "learning_rate": 1.5969867252136113e-06,
      "loss": 0.3953,
      "step": 15458
    },
    {
      "epoch": 0.8342237332038206,
      "grad_norm": 1.0861944609042549,
      "learning_rate": 1.5966070723146782e-06,
      "loss": 0.3692,
      "step": 15459
    },
    {
      "epoch": 0.834277696832335,
      "grad_norm": 1.1243146014513827,
      "learning_rate": 1.5962275316032015e-06,
      "loss": 0.4827,
      "step": 15460
    },
    {
      "epoch": 0.8343316604608494,
      "grad_norm": 0.9593071977186706,
      "learning_rate": 1.5958481030900901e-06,
      "loss": 0.5513,
      "step": 15461
    },
    {
      "epoch": 0.8343856240893638,
      "grad_norm": 0.9268842176998455,
      "learning_rate": 1.5954687867862489e-06,
      "loss": 0.4535,
      "step": 15462
    },
    {
      "epoch": 0.8344395877178782,
      "grad_norm": 1.0673988750977188,
      "learning_rate": 1.5950895827025803e-06,
      "loss": 0.4033,
      "step": 15463
    },
    {
      "epoch": 0.8344935513463926,
      "grad_norm": 0.8715056957113748,
      "learning_rate": 1.5947104908499824e-06,
      "loss": 0.4577,
      "step": 15464
    },
    {
      "epoch": 0.8345475149749069,
      "grad_norm": 0.8987602501928993,
      "learning_rate": 1.5943315112393526e-06,
      "loss": 0.4029,
      "step": 15465
    },
    {
      "epoch": 0.8346014786034213,
      "grad_norm": 0.8773751497525236,
      "learning_rate": 1.59395264388158e-06,
      "loss": 0.3383,
      "step": 15466
    },
    {
      "epoch": 0.8346554422319357,
      "grad_norm": 1.0843570612023368,
      "learning_rate": 1.5935738887875554e-06,
      "loss": 0.5083,
      "step": 15467
    },
    {
      "epoch": 0.8347094058604501,
      "grad_norm": 1.134885321277727,
      "learning_rate": 1.5931952459681642e-06,
      "loss": 0.4528,
      "step": 15468
    },
    {
      "epoch": 0.8347633694889645,
      "grad_norm": 1.1872079318245,
      "learning_rate": 1.5928167154342897e-06,
      "loss": 0.4426,
      "step": 15469
    },
    {
      "epoch": 0.8348173331174789,
      "grad_norm": 1.0479680987545812,
      "learning_rate": 1.59243829719681e-06,
      "loss": 0.4619,
      "step": 15470
    },
    {
      "epoch": 0.8348712967459933,
      "grad_norm": 1.0781794782543832,
      "learning_rate": 1.5920599912666023e-06,
      "loss": 0.554,
      "step": 15471
    },
    {
      "epoch": 0.8349252603745075,
      "grad_norm": 0.8385822015470974,
      "learning_rate": 1.5916817976545401e-06,
      "loss": 0.313,
      "step": 15472
    },
    {
      "epoch": 0.8349792240030219,
      "grad_norm": 1.0637144790626056,
      "learning_rate": 1.5913037163714918e-06,
      "loss": 0.4704,
      "step": 15473
    },
    {
      "epoch": 0.8350331876315363,
      "grad_norm": 1.0384855461031282,
      "learning_rate": 1.590925747428323e-06,
      "loss": 0.4625,
      "step": 15474
    },
    {
      "epoch": 0.8350871512600507,
      "grad_norm": 1.0344086942224813,
      "learning_rate": 1.5905478908358986e-06,
      "loss": 0.3915,
      "step": 15475
    },
    {
      "epoch": 0.8351411148885651,
      "grad_norm": 0.9382600541660834,
      "learning_rate": 1.590170146605078e-06,
      "loss": 0.4082,
      "step": 15476
    },
    {
      "epoch": 0.8351950785170795,
      "grad_norm": 0.9435627897058326,
      "learning_rate": 1.589792514746718e-06,
      "loss": 0.3756,
      "step": 15477
    },
    {
      "epoch": 0.8352490421455939,
      "grad_norm": 0.815861921340752,
      "learning_rate": 1.589414995271672e-06,
      "loss": 0.3624,
      "step": 15478
    },
    {
      "epoch": 0.8353030057741082,
      "grad_norm": 1.0895511276126646,
      "learning_rate": 1.5890375881907913e-06,
      "loss": 0.537,
      "step": 15479
    },
    {
      "epoch": 0.8353569694026226,
      "grad_norm": 0.9628337526826546,
      "learning_rate": 1.5886602935149213e-06,
      "loss": 0.4138,
      "step": 15480
    },
    {
      "epoch": 0.835410933031137,
      "grad_norm": 1.1839720253833979,
      "learning_rate": 1.5882831112549062e-06,
      "loss": 0.4764,
      "step": 15481
    },
    {
      "epoch": 0.8354648966596514,
      "grad_norm": 0.9446969237386693,
      "learning_rate": 1.5879060414215872e-06,
      "loss": 0.4724,
      "step": 15482
    },
    {
      "epoch": 0.8355188602881658,
      "grad_norm": 1.0523577150402004,
      "learning_rate": 1.5875290840258018e-06,
      "loss": 0.4539,
      "step": 15483
    },
    {
      "epoch": 0.8355728239166802,
      "grad_norm": 1.1426080841347372,
      "learning_rate": 1.5871522390783837e-06,
      "loss": 0.4716,
      "step": 15484
    },
    {
      "epoch": 0.8356267875451946,
      "grad_norm": 1.0680833212343255,
      "learning_rate": 1.5867755065901633e-06,
      "loss": 0.514,
      "step": 15485
    },
    {
      "epoch": 0.8356807511737089,
      "grad_norm": 1.0718875817183453,
      "learning_rate": 1.5863988865719702e-06,
      "loss": 0.5649,
      "step": 15486
    },
    {
      "epoch": 0.8357347148022233,
      "grad_norm": 1.1958154353870627,
      "learning_rate": 1.5860223790346268e-06,
      "loss": 0.5051,
      "step": 15487
    },
    {
      "epoch": 0.8357886784307377,
      "grad_norm": 1.0346655311914728,
      "learning_rate": 1.585645983988955e-06,
      "loss": 0.4995,
      "step": 15488
    },
    {
      "epoch": 0.8358426420592521,
      "grad_norm": 0.9746658368748649,
      "learning_rate": 1.585269701445772e-06,
      "loss": 0.4792,
      "step": 15489
    },
    {
      "epoch": 0.8358966056877665,
      "grad_norm": 0.9860016438021469,
      "learning_rate": 1.5848935314158942e-06,
      "loss": 0.3884,
      "step": 15490
    },
    {
      "epoch": 0.8359505693162809,
      "grad_norm": 1.0554058493433476,
      "learning_rate": 1.5845174739101318e-06,
      "loss": 0.4706,
      "step": 15491
    },
    {
      "epoch": 0.8360045329447952,
      "grad_norm": 1.091638406478731,
      "learning_rate": 1.5841415289392936e-06,
      "loss": 0.5771,
      "step": 15492
    },
    {
      "epoch": 0.8360584965733096,
      "grad_norm": 1.0463120367282213,
      "learning_rate": 1.5837656965141846e-06,
      "loss": 0.4753,
      "step": 15493
    },
    {
      "epoch": 0.836112460201824,
      "grad_norm": 1.078644891273764,
      "learning_rate": 1.5833899766456068e-06,
      "loss": 0.4695,
      "step": 15494
    },
    {
      "epoch": 0.8361664238303383,
      "grad_norm": 1.1321392919911106,
      "learning_rate": 1.5830143693443581e-06,
      "loss": 0.6149,
      "step": 15495
    },
    {
      "epoch": 0.8362203874588527,
      "grad_norm": 1.0958851221375978,
      "learning_rate": 1.582638874621234e-06,
      "loss": 0.4649,
      "step": 15496
    },
    {
      "epoch": 0.8362743510873671,
      "grad_norm": 1.0387402235829108,
      "learning_rate": 1.5822634924870272e-06,
      "loss": 0.425,
      "step": 15497
    },
    {
      "epoch": 0.8363283147158815,
      "grad_norm": 1.14659320874304,
      "learning_rate": 1.5818882229525268e-06,
      "loss": 0.4939,
      "step": 15498
    },
    {
      "epoch": 0.8363822783443958,
      "grad_norm": 0.96507869949211,
      "learning_rate": 1.581513066028518e-06,
      "loss": 0.6267,
      "step": 15499
    },
    {
      "epoch": 0.8364362419729102,
      "grad_norm": 1.2469349457589842,
      "learning_rate": 1.5811380217257825e-06,
      "loss": 0.5586,
      "step": 15500
    },
    {
      "epoch": 0.8364362419729102,
      "eval_loss": 0.5311984419822693,
      "eval_runtime": 164.6487,
      "eval_samples_per_second": 20.887,
      "eval_steps_per_second": 0.875,
      "step": 15500
    },
    {
      "epoch": 0.8364902056014246,
      "grad_norm": 1.0208297743669947,
      "learning_rate": 1.5807630900550996e-06,
      "loss": 0.4772,
      "step": 15501
    },
    {
      "epoch": 0.836544169229939,
      "grad_norm": 1.0485464961239297,
      "learning_rate": 1.5803882710272455e-06,
      "loss": 0.4318,
      "step": 15502
    },
    {
      "epoch": 0.8365981328584534,
      "grad_norm": 1.0674857903449697,
      "learning_rate": 1.5800135646529931e-06,
      "loss": 0.4156,
      "step": 15503
    },
    {
      "epoch": 0.8366520964869678,
      "grad_norm": 1.302290883065623,
      "learning_rate": 1.5796389709431123e-06,
      "loss": 0.636,
      "step": 15504
    },
    {
      "epoch": 0.8367060601154822,
      "grad_norm": 1.0727640844906052,
      "learning_rate": 1.5792644899083682e-06,
      "loss": 0.5362,
      "step": 15505
    },
    {
      "epoch": 0.8367600237439965,
      "grad_norm": 0.8858143947268418,
      "learning_rate": 1.578890121559525e-06,
      "loss": 0.3987,
      "step": 15506
    },
    {
      "epoch": 0.8368139873725109,
      "grad_norm": 0.9847345018805407,
      "learning_rate": 1.5785158659073411e-06,
      "loss": 0.4161,
      "step": 15507
    },
    {
      "epoch": 0.8368679510010253,
      "grad_norm": 0.7122367733409533,
      "learning_rate": 1.5781417229625739e-06,
      "loss": 0.2901,
      "step": 15508
    },
    {
      "epoch": 0.8369219146295397,
      "grad_norm": 1.0069260467648573,
      "learning_rate": 1.5777676927359756e-06,
      "loss": 0.5402,
      "step": 15509
    },
    {
      "epoch": 0.8369758782580541,
      "grad_norm": 0.9837379156901922,
      "learning_rate": 1.5773937752382974e-06,
      "loss": 0.5058,
      "step": 15510
    },
    {
      "epoch": 0.8370298418865685,
      "grad_norm": 1.3462639398938983,
      "learning_rate": 1.5770199704802858e-06,
      "loss": 0.5068,
      "step": 15511
    },
    {
      "epoch": 0.8370838055150829,
      "grad_norm": 1.1479964795810278,
      "learning_rate": 1.5766462784726838e-06,
      "loss": 0.5063,
      "step": 15512
    },
    {
      "epoch": 0.8371377691435972,
      "grad_norm": 1.0948436488742808,
      "learning_rate": 1.5762726992262326e-06,
      "loss": 0.5249,
      "step": 15513
    },
    {
      "epoch": 0.8371917327721116,
      "grad_norm": 1.331084626382694,
      "learning_rate": 1.575899232751668e-06,
      "loss": 0.5525,
      "step": 15514
    },
    {
      "epoch": 0.837245696400626,
      "grad_norm": 0.9861529514681407,
      "learning_rate": 1.5755258790597247e-06,
      "loss": 0.4615,
      "step": 15515
    },
    {
      "epoch": 0.8372996600291404,
      "grad_norm": 0.9108610317477531,
      "learning_rate": 1.575152638161133e-06,
      "loss": 0.3566,
      "step": 15516
    },
    {
      "epoch": 0.8373536236576548,
      "grad_norm": 1.0563884223890743,
      "learning_rate": 1.57477951006662e-06,
      "loss": 0.5625,
      "step": 15517
    },
    {
      "epoch": 0.8374075872861692,
      "grad_norm": 1.1190767850282246,
      "learning_rate": 1.5744064947869105e-06,
      "loss": 0.4973,
      "step": 15518
    },
    {
      "epoch": 0.8374615509146836,
      "grad_norm": 1.0483703009262524,
      "learning_rate": 1.574033592332724e-06,
      "loss": 0.5296,
      "step": 15519
    },
    {
      "epoch": 0.8375155145431978,
      "grad_norm": 0.7750788972529037,
      "learning_rate": 1.5736608027147804e-06,
      "loss": 0.3216,
      "step": 15520
    },
    {
      "epoch": 0.8375694781717122,
      "grad_norm": 0.9981656318019938,
      "learning_rate": 1.5732881259437921e-06,
      "loss": 0.5326,
      "step": 15521
    },
    {
      "epoch": 0.8376234418002266,
      "grad_norm": 1.0788586415213026,
      "learning_rate": 1.57291556203047e-06,
      "loss": 0.4626,
      "step": 15522
    },
    {
      "epoch": 0.837677405428741,
      "grad_norm": 1.1277969752216248,
      "learning_rate": 1.5725431109855233e-06,
      "loss": 0.6338,
      "step": 15523
    },
    {
      "epoch": 0.8377313690572554,
      "grad_norm": 0.8806320956964613,
      "learning_rate": 1.5721707728196554e-06,
      "loss": 0.4228,
      "step": 15524
    },
    {
      "epoch": 0.8377853326857698,
      "grad_norm": 0.9024850767853645,
      "learning_rate": 1.5717985475435684e-06,
      "loss": 0.5006,
      "step": 15525
    },
    {
      "epoch": 0.8378392963142842,
      "grad_norm": 0.9576024766338505,
      "learning_rate": 1.5714264351679607e-06,
      "loss": 0.4977,
      "step": 15526
    },
    {
      "epoch": 0.8378932599427985,
      "grad_norm": 1.0040510660789583,
      "learning_rate": 1.5710544357035279e-06,
      "loss": 0.4607,
      "step": 15527
    },
    {
      "epoch": 0.8379472235713129,
      "grad_norm": 1.121756012829598,
      "learning_rate": 1.5706825491609587e-06,
      "loss": 0.4487,
      "step": 15528
    },
    {
      "epoch": 0.8380011871998273,
      "grad_norm": 1.0666597297888356,
      "learning_rate": 1.570310775550944e-06,
      "loss": 0.4233,
      "step": 15529
    },
    {
      "epoch": 0.8380551508283417,
      "grad_norm": 1.4548882030090533,
      "learning_rate": 1.5699391148841677e-06,
      "loss": 0.7751,
      "step": 15530
    },
    {
      "epoch": 0.8381091144568561,
      "grad_norm": 1.1710269364680927,
      "learning_rate": 1.5695675671713132e-06,
      "loss": 0.6118,
      "step": 15531
    },
    {
      "epoch": 0.8381630780853705,
      "grad_norm": 0.9908887257628076,
      "learning_rate": 1.5691961324230575e-06,
      "loss": 0.4487,
      "step": 15532
    },
    {
      "epoch": 0.8382170417138849,
      "grad_norm": 1.1444201620574257,
      "learning_rate": 1.568824810650077e-06,
      "loss": 0.4457,
      "step": 15533
    },
    {
      "epoch": 0.8382710053423992,
      "grad_norm": 1.1356439758538068,
      "learning_rate": 1.5684536018630442e-06,
      "loss": 0.6472,
      "step": 15534
    },
    {
      "epoch": 0.8383249689709136,
      "grad_norm": 1.0396377427167394,
      "learning_rate": 1.5680825060726269e-06,
      "loss": 0.4528,
      "step": 15535
    },
    {
      "epoch": 0.838378932599428,
      "grad_norm": 1.0178773801284349,
      "learning_rate": 1.567711523289491e-06,
      "loss": 0.4244,
      "step": 15536
    },
    {
      "epoch": 0.8384328962279424,
      "grad_norm": 1.1168994781879635,
      "learning_rate": 1.5673406535242992e-06,
      "loss": 0.4732,
      "step": 15537
    },
    {
      "epoch": 0.8384868598564568,
      "grad_norm": 1.3991284872716359,
      "learning_rate": 1.5669698967877108e-06,
      "loss": 0.5227,
      "step": 15538
    },
    {
      "epoch": 0.8385408234849712,
      "grad_norm": 1.125999648831656,
      "learning_rate": 1.5665992530903821e-06,
      "loss": 0.521,
      "step": 15539
    },
    {
      "epoch": 0.8385947871134856,
      "grad_norm": 0.7754519326792063,
      "learning_rate": 1.5662287224429645e-06,
      "loss": 0.3976,
      "step": 15540
    },
    {
      "epoch": 0.8386487507419998,
      "grad_norm": 0.9693054231220607,
      "learning_rate": 1.5658583048561085e-06,
      "loss": 0.5432,
      "step": 15541
    },
    {
      "epoch": 0.8387027143705142,
      "grad_norm": 1.1667820808394238,
      "learning_rate": 1.56548800034046e-06,
      "loss": 0.5246,
      "step": 15542
    },
    {
      "epoch": 0.8387566779990286,
      "grad_norm": 0.9784687575834556,
      "learning_rate": 1.5651178089066627e-06,
      "loss": 0.3657,
      "step": 15543
    },
    {
      "epoch": 0.838810641627543,
      "grad_norm": 1.1177386700991623,
      "learning_rate": 1.5647477305653546e-06,
      "loss": 0.5693,
      "step": 15544
    },
    {
      "epoch": 0.8388646052560574,
      "grad_norm": 0.9100597312971741,
      "learning_rate": 1.5643777653271734e-06,
      "loss": 0.4045,
      "step": 15545
    },
    {
      "epoch": 0.8389185688845718,
      "grad_norm": 0.8614887604229023,
      "learning_rate": 1.564007913202751e-06,
      "loss": 0.3508,
      "step": 15546
    },
    {
      "epoch": 0.8389725325130862,
      "grad_norm": 1.1168961458309108,
      "learning_rate": 1.5636381742027188e-06,
      "loss": 0.5634,
      "step": 15547
    },
    {
      "epoch": 0.8390264961416005,
      "grad_norm": 1.1020084523379883,
      "learning_rate": 1.5632685483377031e-06,
      "loss": 0.4267,
      "step": 15548
    },
    {
      "epoch": 0.8390804597701149,
      "grad_norm": 1.1048948092208573,
      "learning_rate": 1.5628990356183267e-06,
      "loss": 0.5287,
      "step": 15549
    },
    {
      "epoch": 0.8391344233986293,
      "grad_norm": 1.1453961644181934,
      "learning_rate": 1.56252963605521e-06,
      "loss": 0.4221,
      "step": 15550
    },
    {
      "epoch": 0.8391883870271437,
      "grad_norm": 1.3050703741966283,
      "learning_rate": 1.5621603496589708e-06,
      "loss": 0.4957,
      "step": 15551
    },
    {
      "epoch": 0.8392423506556581,
      "grad_norm": 1.0183456901240044,
      "learning_rate": 1.5617911764402212e-06,
      "loss": 0.5258,
      "step": 15552
    },
    {
      "epoch": 0.8392963142841725,
      "grad_norm": 1.0992568929611217,
      "learning_rate": 1.5614221164095732e-06,
      "loss": 0.5173,
      "step": 15553
    },
    {
      "epoch": 0.8393502779126869,
      "grad_norm": 1.0238101910879376,
      "learning_rate": 1.5610531695776327e-06,
      "loss": 0.3312,
      "step": 15554
    },
    {
      "epoch": 0.8394042415412012,
      "grad_norm": 1.16937732421084,
      "learning_rate": 1.5606843359550053e-06,
      "loss": 0.4706,
      "step": 15555
    },
    {
      "epoch": 0.8394582051697156,
      "grad_norm": 0.939287629188896,
      "learning_rate": 1.5603156155522895e-06,
      "loss": 0.4166,
      "step": 15556
    },
    {
      "epoch": 0.83951216879823,
      "grad_norm": 0.9834324787031573,
      "learning_rate": 1.5599470083800837e-06,
      "loss": 0.3255,
      "step": 15557
    },
    {
      "epoch": 0.8395661324267444,
      "grad_norm": 0.9443240760881272,
      "learning_rate": 1.559578514448982e-06,
      "loss": 0.3834,
      "step": 15558
    },
    {
      "epoch": 0.8396200960552588,
      "grad_norm": 1.231388934292451,
      "learning_rate": 1.559210133769575e-06,
      "loss": 0.3705,
      "step": 15559
    },
    {
      "epoch": 0.8396740596837732,
      "grad_norm": 0.9077461754332709,
      "learning_rate": 1.5588418663524513e-06,
      "loss": 0.4103,
      "step": 15560
    },
    {
      "epoch": 0.8397280233122875,
      "grad_norm": 0.8792249230134237,
      "learning_rate": 1.558473712208195e-06,
      "loss": 0.3633,
      "step": 15561
    },
    {
      "epoch": 0.8397819869408019,
      "grad_norm": 1.087648120561072,
      "learning_rate": 1.558105671347386e-06,
      "loss": 0.4301,
      "step": 15562
    },
    {
      "epoch": 0.8398359505693163,
      "grad_norm": 1.2110174302942933,
      "learning_rate": 1.5577377437806033e-06,
      "loss": 0.5583,
      "step": 15563
    },
    {
      "epoch": 0.8398899141978307,
      "grad_norm": 0.8581004118813542,
      "learning_rate": 1.5573699295184206e-06,
      "loss": 0.4359,
      "step": 15564
    },
    {
      "epoch": 0.839943877826345,
      "grad_norm": 1.067440705344865,
      "learning_rate": 1.55700222857141e-06,
      "loss": 0.4869,
      "step": 15565
    },
    {
      "epoch": 0.8399978414548595,
      "grad_norm": 1.03293553982358,
      "learning_rate": 1.5566346409501399e-06,
      "loss": 0.4807,
      "step": 15566
    },
    {
      "epoch": 0.8400518050833738,
      "grad_norm": 1.0428222011764312,
      "learning_rate": 1.5562671666651747e-06,
      "loss": 0.5519,
      "step": 15567
    },
    {
      "epoch": 0.8401057687118881,
      "grad_norm": 1.0986706673808653,
      "learning_rate": 1.5558998057270766e-06,
      "loss": 0.4343,
      "step": 15568
    },
    {
      "epoch": 0.8401597323404025,
      "grad_norm": 1.0237842225787237,
      "learning_rate": 1.5555325581464024e-06,
      "loss": 0.5529,
      "step": 15569
    },
    {
      "epoch": 0.8402136959689169,
      "grad_norm": 1.0644490306371286,
      "learning_rate": 1.5551654239337083e-06,
      "loss": 0.3911,
      "step": 15570
    },
    {
      "epoch": 0.8402676595974313,
      "grad_norm": 0.9083267249104359,
      "learning_rate": 1.5547984030995452e-06,
      "loss": 0.3876,
      "step": 15571
    },
    {
      "epoch": 0.8403216232259457,
      "grad_norm": 0.9575762132747923,
      "learning_rate": 1.554431495654463e-06,
      "loss": 0.5125,
      "step": 15572
    },
    {
      "epoch": 0.8403755868544601,
      "grad_norm": 0.8523796270181165,
      "learning_rate": 1.5540647016090066e-06,
      "loss": 0.3148,
      "step": 15573
    },
    {
      "epoch": 0.8404295504829745,
      "grad_norm": 0.9978047047219434,
      "learning_rate": 1.5536980209737172e-06,
      "loss": 0.3877,
      "step": 15574
    },
    {
      "epoch": 0.8404835141114888,
      "grad_norm": 1.0426434581066748,
      "learning_rate": 1.5533314537591352e-06,
      "loss": 0.4041,
      "step": 15575
    },
    {
      "epoch": 0.8405374777400032,
      "grad_norm": 1.1240320302166988,
      "learning_rate": 1.552964999975794e-06,
      "loss": 0.4652,
      "step": 15576
    },
    {
      "epoch": 0.8405914413685176,
      "grad_norm": 1.0664260973418827,
      "learning_rate": 1.5525986596342268e-06,
      "loss": 0.5393,
      "step": 15577
    },
    {
      "epoch": 0.840645404997032,
      "grad_norm": 0.7787820743046387,
      "learning_rate": 1.5522324327449628e-06,
      "loss": 0.3669,
      "step": 15578
    },
    {
      "epoch": 0.8406993686255464,
      "grad_norm": 0.9804192928563061,
      "learning_rate": 1.551866319318528e-06,
      "loss": 0.4494,
      "step": 15579
    },
    {
      "epoch": 0.8407533322540608,
      "grad_norm": 1.0194823777864181,
      "learning_rate": 1.5515003193654438e-06,
      "loss": 0.5125,
      "step": 15580
    },
    {
      "epoch": 0.8408072958825752,
      "grad_norm": 0.9308877682468607,
      "learning_rate": 1.5511344328962301e-06,
      "loss": 0.4123,
      "step": 15581
    },
    {
      "epoch": 0.8408612595110895,
      "grad_norm": 1.0647955140508276,
      "learning_rate": 1.5507686599214047e-06,
      "loss": 0.32,
      "step": 15582
    },
    {
      "epoch": 0.8409152231396039,
      "grad_norm": 0.9359130234146371,
      "learning_rate": 1.5504030004514767e-06,
      "loss": 0.5042,
      "step": 15583
    },
    {
      "epoch": 0.8409691867681183,
      "grad_norm": 1.1931017853893984,
      "learning_rate": 1.550037454496957e-06,
      "loss": 0.6229,
      "step": 15584
    },
    {
      "epoch": 0.8410231503966327,
      "grad_norm": 1.0537813193171577,
      "learning_rate": 1.549672022068353e-06,
      "loss": 0.5197,
      "step": 15585
    },
    {
      "epoch": 0.8410771140251471,
      "grad_norm": 1.1506779838849222,
      "learning_rate": 1.5493067031761657e-06,
      "loss": 0.5619,
      "step": 15586
    },
    {
      "epoch": 0.8411310776536615,
      "grad_norm": 1.1455182848332406,
      "learning_rate": 1.548941497830896e-06,
      "loss": 0.4871,
      "step": 15587
    },
    {
      "epoch": 0.8411850412821759,
      "grad_norm": 0.9543181418540896,
      "learning_rate": 1.54857640604304e-06,
      "loss": 0.3547,
      "step": 15588
    },
    {
      "epoch": 0.8412390049106901,
      "grad_norm": 1.0538407370586427,
      "learning_rate": 1.5482114278230915e-06,
      "loss": 0.5522,
      "step": 15589
    },
    {
      "epoch": 0.8412929685392045,
      "grad_norm": 1.0582181067276633,
      "learning_rate": 1.5478465631815387e-06,
      "loss": 0.676,
      "step": 15590
    },
    {
      "epoch": 0.8413469321677189,
      "grad_norm": 0.747398100069921,
      "learning_rate": 1.5474818121288687e-06,
      "loss": 0.2231,
      "step": 15591
    },
    {
      "epoch": 0.8414008957962333,
      "grad_norm": 0.778502224727566,
      "learning_rate": 1.5471171746755653e-06,
      "loss": 0.2815,
      "step": 15592
    },
    {
      "epoch": 0.8414548594247477,
      "grad_norm": 0.793909066386534,
      "learning_rate": 1.5467526508321088e-06,
      "loss": 0.3181,
      "step": 15593
    },
    {
      "epoch": 0.8415088230532621,
      "grad_norm": 0.8397800854947285,
      "learning_rate": 1.5463882406089752e-06,
      "loss": 0.4084,
      "step": 15594
    },
    {
      "epoch": 0.8415627866817765,
      "grad_norm": 0.9170500887745147,
      "learning_rate": 1.5460239440166386e-06,
      "loss": 0.3721,
      "step": 15595
    },
    {
      "epoch": 0.8416167503102908,
      "grad_norm": 1.0216364230396604,
      "learning_rate": 1.5456597610655692e-06,
      "loss": 0.4339,
      "step": 15596
    },
    {
      "epoch": 0.8416707139388052,
      "grad_norm": 1.258023987744892,
      "learning_rate": 1.5452956917662332e-06,
      "loss": 0.4564,
      "step": 15597
    },
    {
      "epoch": 0.8417246775673196,
      "grad_norm": 1.3739793071366757,
      "learning_rate": 1.5449317361290956e-06,
      "loss": 0.6747,
      "step": 15598
    },
    {
      "epoch": 0.841778641195834,
      "grad_norm": 1.0992868643478324,
      "learning_rate": 1.5445678941646156e-06,
      "loss": 0.5635,
      "step": 15599
    },
    {
      "epoch": 0.8418326048243484,
      "grad_norm": 1.0542431107578454,
      "learning_rate": 1.5442041658832516e-06,
      "loss": 0.4258,
      "step": 15600
    },
    {
      "epoch": 0.8418865684528628,
      "grad_norm": 0.9675619511544512,
      "learning_rate": 1.5438405512954565e-06,
      "loss": 0.3489,
      "step": 15601
    },
    {
      "epoch": 0.8419405320813772,
      "grad_norm": 1.052713208143923,
      "learning_rate": 1.5434770504116814e-06,
      "loss": 0.5039,
      "step": 15602
    },
    {
      "epoch": 0.8419944957098915,
      "grad_norm": 1.1854999968485487,
      "learning_rate": 1.5431136632423749e-06,
      "loss": 0.3994,
      "step": 15603
    },
    {
      "epoch": 0.8420484593384059,
      "grad_norm": 1.2654173408310079,
      "learning_rate": 1.5427503897979787e-06,
      "loss": 0.6131,
      "step": 15604
    },
    {
      "epoch": 0.8421024229669203,
      "grad_norm": 1.090355424192914,
      "learning_rate": 1.5423872300889347e-06,
      "loss": 0.4192,
      "step": 15605
    },
    {
      "epoch": 0.8421563865954347,
      "grad_norm": 1.1128558673473303,
      "learning_rate": 1.542024184125681e-06,
      "loss": 0.4157,
      "step": 15606
    },
    {
      "epoch": 0.8422103502239491,
      "grad_norm": 1.0147074341448536,
      "learning_rate": 1.5416612519186508e-06,
      "loss": 0.4434,
      "step": 15607
    },
    {
      "epoch": 0.8422643138524635,
      "grad_norm": 0.7364606036601585,
      "learning_rate": 1.5412984334782762e-06,
      "loss": 0.2912,
      "step": 15608
    },
    {
      "epoch": 0.8423182774809779,
      "grad_norm": 1.0401434329791017,
      "learning_rate": 1.5409357288149846e-06,
      "loss": 0.4572,
      "step": 15609
    },
    {
      "epoch": 0.8423722411094922,
      "grad_norm": 1.2336517186647886,
      "learning_rate": 1.540573137939202e-06,
      "loss": 0.5057,
      "step": 15610
    },
    {
      "epoch": 0.8424262047380066,
      "grad_norm": 0.954521521727013,
      "learning_rate": 1.5402106608613463e-06,
      "loss": 0.3518,
      "step": 15611
    },
    {
      "epoch": 0.842480168366521,
      "grad_norm": 1.1020829093871387,
      "learning_rate": 1.5398482975918378e-06,
      "loss": 0.5578,
      "step": 15612
    },
    {
      "epoch": 0.8425341319950354,
      "grad_norm": 1.0749108977759225,
      "learning_rate": 1.5394860481410903e-06,
      "loss": 0.4575,
      "step": 15613
    },
    {
      "epoch": 0.8425880956235497,
      "grad_norm": 1.2015204421810688,
      "learning_rate": 1.5391239125195157e-06,
      "loss": 0.5051,
      "step": 15614
    },
    {
      "epoch": 0.8426420592520641,
      "grad_norm": 0.9140882933900291,
      "learning_rate": 1.538761890737522e-06,
      "loss": 0.3763,
      "step": 15615
    },
    {
      "epoch": 0.8426960228805785,
      "grad_norm": 0.8307259138850285,
      "learning_rate": 1.5383999828055142e-06,
      "loss": 0.2464,
      "step": 15616
    },
    {
      "epoch": 0.8427499865090928,
      "grad_norm": 0.6738548208830848,
      "learning_rate": 1.5380381887338951e-06,
      "loss": 0.2718,
      "step": 15617
    },
    {
      "epoch": 0.8428039501376072,
      "grad_norm": 1.0215386898135,
      "learning_rate": 1.5376765085330603e-06,
      "loss": 0.4021,
      "step": 15618
    },
    {
      "epoch": 0.8428579137661216,
      "grad_norm": 0.9302651269309149,
      "learning_rate": 1.5373149422134064e-06,
      "loss": 0.43,
      "step": 15619
    },
    {
      "epoch": 0.842911877394636,
      "grad_norm": 1.0094367776411532,
      "learning_rate": 1.5369534897853256e-06,
      "loss": 0.4959,
      "step": 15620
    },
    {
      "epoch": 0.8429658410231504,
      "grad_norm": 0.8003082844027756,
      "learning_rate": 1.5365921512592053e-06,
      "loss": 0.2892,
      "step": 15621
    },
    {
      "epoch": 0.8430198046516648,
      "grad_norm": 0.7779880128013374,
      "learning_rate": 1.5362309266454308e-06,
      "loss": 0.2617,
      "step": 15622
    },
    {
      "epoch": 0.8430737682801792,
      "grad_norm": 0.8994193036741248,
      "learning_rate": 1.5358698159543853e-06,
      "loss": 0.3638,
      "step": 15623
    },
    {
      "epoch": 0.8431277319086935,
      "grad_norm": 1.233392741986116,
      "learning_rate": 1.5355088191964473e-06,
      "loss": 0.5541,
      "step": 15624
    },
    {
      "epoch": 0.8431816955372079,
      "grad_norm": 0.8539974526849857,
      "learning_rate": 1.535147936381991e-06,
      "loss": 0.3277,
      "step": 15625
    },
    {
      "epoch": 0.8432356591657223,
      "grad_norm": 1.0076607624096026,
      "learning_rate": 1.5347871675213893e-06,
      "loss": 0.4668,
      "step": 15626
    },
    {
      "epoch": 0.8432896227942367,
      "grad_norm": 1.0347599262852203,
      "learning_rate": 1.5344265126250107e-06,
      "loss": 0.4307,
      "step": 15627
    },
    {
      "epoch": 0.8433435864227511,
      "grad_norm": 1.1866362871254612,
      "learning_rate": 1.5340659717032211e-06,
      "loss": 0.5849,
      "step": 15628
    },
    {
      "epoch": 0.8433975500512655,
      "grad_norm": 0.9744844127451343,
      "learning_rate": 1.533705544766383e-06,
      "loss": 0.51,
      "step": 15629
    },
    {
      "epoch": 0.8434515136797798,
      "grad_norm": 0.8072184838957053,
      "learning_rate": 1.533345231824856e-06,
      "loss": 0.3813,
      "step": 15630
    },
    {
      "epoch": 0.8435054773082942,
      "grad_norm": 1.011812807273064,
      "learning_rate": 1.5329850328889936e-06,
      "loss": 0.4298,
      "step": 15631
    },
    {
      "epoch": 0.8435594409368086,
      "grad_norm": 1.0198565800579311,
      "learning_rate": 1.5326249479691502e-06,
      "loss": 0.51,
      "step": 15632
    },
    {
      "epoch": 0.843613404565323,
      "grad_norm": 0.972540294264195,
      "learning_rate": 1.5322649770756746e-06,
      "loss": 0.3957,
      "step": 15633
    },
    {
      "epoch": 0.8436673681938374,
      "grad_norm": 1.0752082361849231,
      "learning_rate": 1.5319051202189125e-06,
      "loss": 0.4533,
      "step": 15634
    },
    {
      "epoch": 0.8437213318223518,
      "grad_norm": 1.0356576727901199,
      "learning_rate": 1.5315453774092065e-06,
      "loss": 0.5183,
      "step": 15635
    },
    {
      "epoch": 0.8437752954508662,
      "grad_norm": 1.1379719120174503,
      "learning_rate": 1.5311857486568967e-06,
      "loss": 0.4075,
      "step": 15636
    },
    {
      "epoch": 0.8438292590793804,
      "grad_norm": 1.0051130526428098,
      "learning_rate": 1.5308262339723193e-06,
      "loss": 0.4177,
      "step": 15637
    },
    {
      "epoch": 0.8438832227078948,
      "grad_norm": 1.1734405175079952,
      "learning_rate": 1.5304668333658056e-06,
      "loss": 0.5285,
      "step": 15638
    },
    {
      "epoch": 0.8439371863364092,
      "grad_norm": 0.9116643477268707,
      "learning_rate": 1.5301075468476862e-06,
      "loss": 0.3737,
      "step": 15639
    },
    {
      "epoch": 0.8439911499649236,
      "grad_norm": 1.109970703551109,
      "learning_rate": 1.5297483744282873e-06,
      "loss": 0.4999,
      "step": 15640
    },
    {
      "epoch": 0.844045113593438,
      "grad_norm": 1.004180089464532,
      "learning_rate": 1.5293893161179313e-06,
      "loss": 0.3884,
      "step": 15641
    },
    {
      "epoch": 0.8440990772219524,
      "grad_norm": 1.1681651758296154,
      "learning_rate": 1.5290303719269383e-06,
      "loss": 0.5603,
      "step": 15642
    },
    {
      "epoch": 0.8441530408504668,
      "grad_norm": 1.0991786872799718,
      "learning_rate": 1.528671541865625e-06,
      "loss": 0.5582,
      "step": 15643
    },
    {
      "epoch": 0.8442070044789811,
      "grad_norm": 1.1946474945385264,
      "learning_rate": 1.5283128259443048e-06,
      "loss": 0.4571,
      "step": 15644
    },
    {
      "epoch": 0.8442609681074955,
      "grad_norm": 1.002998397826846,
      "learning_rate": 1.5279542241732864e-06,
      "loss": 0.3947,
      "step": 15645
    },
    {
      "epoch": 0.8443149317360099,
      "grad_norm": 0.955698653651919,
      "learning_rate": 1.5275957365628773e-06,
      "loss": 0.43,
      "step": 15646
    },
    {
      "epoch": 0.8443688953645243,
      "grad_norm": 1.0831566663886352,
      "learning_rate": 1.5272373631233806e-06,
      "loss": 0.6253,
      "step": 15647
    },
    {
      "epoch": 0.8444228589930387,
      "grad_norm": 1.151361359194399,
      "learning_rate": 1.5268791038650975e-06,
      "loss": 0.5058,
      "step": 15648
    },
    {
      "epoch": 0.8444768226215531,
      "grad_norm": 0.8796930606746819,
      "learning_rate": 1.5265209587983221e-06,
      "loss": 0.3226,
      "step": 15649
    },
    {
      "epoch": 0.8445307862500675,
      "grad_norm": 1.172956063308295,
      "learning_rate": 1.5261629279333492e-06,
      "loss": 0.6813,
      "step": 15650
    },
    {
      "epoch": 0.8445847498785818,
      "grad_norm": 1.206714564346388,
      "learning_rate": 1.5258050112804693e-06,
      "loss": 0.5728,
      "step": 15651
    },
    {
      "epoch": 0.8446387135070962,
      "grad_norm": 0.9416206385126298,
      "learning_rate": 1.5254472088499685e-06,
      "loss": 0.3645,
      "step": 15652
    },
    {
      "epoch": 0.8446926771356106,
      "grad_norm": 1.0452365084521218,
      "learning_rate": 1.5250895206521316e-06,
      "loss": 0.4241,
      "step": 15653
    },
    {
      "epoch": 0.844746640764125,
      "grad_norm": 0.8368733906809237,
      "learning_rate": 1.524731946697238e-06,
      "loss": 0.3788,
      "step": 15654
    },
    {
      "epoch": 0.8448006043926394,
      "grad_norm": 1.130356665915359,
      "learning_rate": 1.5243744869955648e-06,
      "loss": 0.4596,
      "step": 15655
    },
    {
      "epoch": 0.8448545680211538,
      "grad_norm": 1.0542698490689562,
      "learning_rate": 1.5240171415573862e-06,
      "loss": 0.5717,
      "step": 15656
    },
    {
      "epoch": 0.8449085316496682,
      "grad_norm": 1.4176149069543755,
      "learning_rate": 1.5236599103929722e-06,
      "loss": 0.7412,
      "step": 15657
    },
    {
      "epoch": 0.8449624952781825,
      "grad_norm": 1.0819664043252837,
      "learning_rate": 1.523302793512591e-06,
      "loss": 0.57,
      "step": 15658
    },
    {
      "epoch": 0.8450164589066969,
      "grad_norm": 1.2290749183877465,
      "learning_rate": 1.5229457909265053e-06,
      "loss": 0.6637,
      "step": 15659
    },
    {
      "epoch": 0.8450704225352113,
      "grad_norm": 1.0489037396459566,
      "learning_rate": 1.5225889026449754e-06,
      "loss": 0.4026,
      "step": 15660
    },
    {
      "epoch": 0.8451243861637256,
      "grad_norm": 1.0948028629880235,
      "learning_rate": 1.5222321286782594e-06,
      "loss": 0.4641,
      "step": 15661
    },
    {
      "epoch": 0.84517834979224,
      "grad_norm": 0.8551910326132586,
      "learning_rate": 1.5218754690366114e-06,
      "loss": 0.4152,
      "step": 15662
    },
    {
      "epoch": 0.8452323134207544,
      "grad_norm": 1.0170543971328168,
      "learning_rate": 1.521518923730282e-06,
      "loss": 0.504,
      "step": 15663
    },
    {
      "epoch": 0.8452862770492688,
      "grad_norm": 1.135491872139825,
      "learning_rate": 1.521162492769519e-06,
      "loss": 0.4609,
      "step": 15664
    },
    {
      "epoch": 0.8453402406777831,
      "grad_norm": 1.169891679399081,
      "learning_rate": 1.5208061761645671e-06,
      "loss": 0.5186,
      "step": 15665
    },
    {
      "epoch": 0.8453942043062975,
      "grad_norm": 1.1450771696472846,
      "learning_rate": 1.5204499739256653e-06,
      "loss": 0.4862,
      "step": 15666
    },
    {
      "epoch": 0.8454481679348119,
      "grad_norm": 1.0244272046879965,
      "learning_rate": 1.520093886063052e-06,
      "loss": 0.3807,
      "step": 15667
    },
    {
      "epoch": 0.8455021315633263,
      "grad_norm": 1.0959603534024962,
      "learning_rate": 1.5197379125869624e-06,
      "loss": 0.518,
      "step": 15668
    },
    {
      "epoch": 0.8455560951918407,
      "grad_norm": 1.0867438939061882,
      "learning_rate": 1.5193820535076264e-06,
      "loss": 0.5516,
      "step": 15669
    },
    {
      "epoch": 0.8456100588203551,
      "grad_norm": 1.102048936742892,
      "learning_rate": 1.5190263088352727e-06,
      "loss": 0.5641,
      "step": 15670
    },
    {
      "epoch": 0.8456640224488695,
      "grad_norm": 1.0895360331259327,
      "learning_rate": 1.5186706785801254e-06,
      "loss": 0.5802,
      "step": 15671
    },
    {
      "epoch": 0.8457179860773838,
      "grad_norm": 0.9451019320585996,
      "learning_rate": 1.5183151627524065e-06,
      "loss": 0.4569,
      "step": 15672
    },
    {
      "epoch": 0.8457719497058982,
      "grad_norm": 0.9978216294560042,
      "learning_rate": 1.517959761362332e-06,
      "loss": 0.4188,
      "step": 15673
    },
    {
      "epoch": 0.8458259133344126,
      "grad_norm": 1.1198097111779806,
      "learning_rate": 1.5176044744201173e-06,
      "loss": 0.4682,
      "step": 15674
    },
    {
      "epoch": 0.845879876962927,
      "grad_norm": 1.255605327038385,
      "learning_rate": 1.517249301935974e-06,
      "loss": 0.4846,
      "step": 15675
    },
    {
      "epoch": 0.8459338405914414,
      "grad_norm": 1.0436531186155327,
      "learning_rate": 1.5168942439201107e-06,
      "loss": 0.5202,
      "step": 15676
    },
    {
      "epoch": 0.8459878042199558,
      "grad_norm": 0.9232969061135506,
      "learning_rate": 1.5165393003827308e-06,
      "loss": 0.4021,
      "step": 15677
    },
    {
      "epoch": 0.8460417678484702,
      "grad_norm": 0.8147570932043685,
      "learning_rate": 1.5161844713340364e-06,
      "loss": 0.3094,
      "step": 15678
    },
    {
      "epoch": 0.8460957314769845,
      "grad_norm": 1.2627248711227055,
      "learning_rate": 1.5158297567842266e-06,
      "loss": 0.5915,
      "step": 15679
    },
    {
      "epoch": 0.8461496951054989,
      "grad_norm": 1.1886283842150887,
      "learning_rate": 1.5154751567434944e-06,
      "loss": 0.532,
      "step": 15680
    },
    {
      "epoch": 0.8462036587340133,
      "grad_norm": 1.0283885610952799,
      "learning_rate": 1.5151206712220326e-06,
      "loss": 0.4946,
      "step": 15681
    },
    {
      "epoch": 0.8462576223625277,
      "grad_norm": 1.0107088324610436,
      "learning_rate": 1.514766300230029e-06,
      "loss": 0.3893,
      "step": 15682
    },
    {
      "epoch": 0.8463115859910421,
      "grad_norm": 1.301034999949981,
      "learning_rate": 1.5144120437776686e-06,
      "loss": 0.5841,
      "step": 15683
    },
    {
      "epoch": 0.8463655496195565,
      "grad_norm": 1.072419932641718,
      "learning_rate": 1.5140579018751327e-06,
      "loss": 0.4495,
      "step": 15684
    },
    {
      "epoch": 0.8464195132480709,
      "grad_norm": 1.0478313501813261,
      "learning_rate": 1.513703874532601e-06,
      "loss": 0.4316,
      "step": 15685
    },
    {
      "epoch": 0.8464734768765851,
      "grad_norm": 0.9397128301918715,
      "learning_rate": 1.5133499617602488e-06,
      "loss": 0.4092,
      "step": 15686
    },
    {
      "epoch": 0.8465274405050995,
      "grad_norm": 1.02324865240044,
      "learning_rate": 1.5129961635682455e-06,
      "loss": 0.4742,
      "step": 15687
    },
    {
      "epoch": 0.8465814041336139,
      "grad_norm": 0.7205241479806055,
      "learning_rate": 1.5126424799667608e-06,
      "loss": 0.2645,
      "step": 15688
    },
    {
      "epoch": 0.8466353677621283,
      "grad_norm": 1.3436038829895383,
      "learning_rate": 1.5122889109659605e-06,
      "loss": 0.4738,
      "step": 15689
    },
    {
      "epoch": 0.8466893313906427,
      "grad_norm": 1.1203097929474553,
      "learning_rate": 1.511935456576006e-06,
      "loss": 0.4884,
      "step": 15690
    },
    {
      "epoch": 0.8467432950191571,
      "grad_norm": 0.849693838477334,
      "learning_rate": 1.5115821168070566e-06,
      "loss": 0.3563,
      "step": 15691
    },
    {
      "epoch": 0.8467972586476715,
      "grad_norm": 0.9942231877881641,
      "learning_rate": 1.511228891669267e-06,
      "loss": 0.4779,
      "step": 15692
    },
    {
      "epoch": 0.8468512222761858,
      "grad_norm": 0.9646936269492785,
      "learning_rate": 1.5108757811727904e-06,
      "loss": 0.3296,
      "step": 15693
    },
    {
      "epoch": 0.8469051859047002,
      "grad_norm": 0.9863326215000598,
      "learning_rate": 1.510522785327773e-06,
      "loss": 0.3582,
      "step": 15694
    },
    {
      "epoch": 0.8469591495332146,
      "grad_norm": 1.2541564055945007,
      "learning_rate": 1.5101699041443626e-06,
      "loss": 0.5427,
      "step": 15695
    },
    {
      "epoch": 0.847013113161729,
      "grad_norm": 1.067874565456146,
      "learning_rate": 1.5098171376327e-06,
      "loss": 0.434,
      "step": 15696
    },
    {
      "epoch": 0.8470670767902434,
      "grad_norm": 0.8676080622440123,
      "learning_rate": 1.5094644858029251e-06,
      "loss": 0.311,
      "step": 15697
    },
    {
      "epoch": 0.8471210404187578,
      "grad_norm": 1.1110851080837238,
      "learning_rate": 1.5091119486651729e-06,
      "loss": 0.4165,
      "step": 15698
    },
    {
      "epoch": 0.8471750040472721,
      "grad_norm": 1.082143923214254,
      "learning_rate": 1.508759526229576e-06,
      "loss": 0.4826,
      "step": 15699
    },
    {
      "epoch": 0.8472289676757865,
      "grad_norm": 1.2370947067573443,
      "learning_rate": 1.5084072185062627e-06,
      "loss": 0.5745,
      "step": 15700
    },
    {
      "epoch": 0.8472829313043009,
      "grad_norm": 0.9899907439625864,
      "learning_rate": 1.5080550255053594e-06,
      "loss": 0.4032,
      "step": 15701
    },
    {
      "epoch": 0.8473368949328153,
      "grad_norm": 0.8673386970844211,
      "learning_rate": 1.5077029472369884e-06,
      "loss": 0.3717,
      "step": 15702
    },
    {
      "epoch": 0.8473908585613297,
      "grad_norm": 0.9120663057235976,
      "learning_rate": 1.507350983711268e-06,
      "loss": 0.5253,
      "step": 15703
    },
    {
      "epoch": 0.8474448221898441,
      "grad_norm": 1.0678633393656072,
      "learning_rate": 1.5069991349383151e-06,
      "loss": 0.4357,
      "step": 15704
    },
    {
      "epoch": 0.8474987858183585,
      "grad_norm": 0.9813157967987356,
      "learning_rate": 1.5066474009282417e-06,
      "loss": 0.4855,
      "step": 15705
    },
    {
      "epoch": 0.8475527494468728,
      "grad_norm": 1.0290988232134095,
      "learning_rate": 1.506295781691158e-06,
      "loss": 0.5004,
      "step": 15706
    },
    {
      "epoch": 0.8476067130753872,
      "grad_norm": 0.8550078135398885,
      "learning_rate": 1.505944277237168e-06,
      "loss": 0.3387,
      "step": 15707
    },
    {
      "epoch": 0.8476606767039015,
      "grad_norm": 0.9119805372373545,
      "learning_rate": 1.5055928875763745e-06,
      "loss": 0.3789,
      "step": 15708
    },
    {
      "epoch": 0.847714640332416,
      "grad_norm": 1.1350851933650363,
      "learning_rate": 1.5052416127188777e-06,
      "loss": 0.488,
      "step": 15709
    },
    {
      "epoch": 0.8477686039609303,
      "grad_norm": 1.1240290847970673,
      "learning_rate": 1.5048904526747738e-06,
      "loss": 0.4973,
      "step": 15710
    },
    {
      "epoch": 0.8478225675894447,
      "grad_norm": 1.0365015847659047,
      "learning_rate": 1.5045394074541548e-06,
      "loss": 0.401,
      "step": 15711
    },
    {
      "epoch": 0.8478765312179591,
      "grad_norm": 1.0373086829482396,
      "learning_rate": 1.5041884770671106e-06,
      "loss": 0.5639,
      "step": 15712
    },
    {
      "epoch": 0.8479304948464734,
      "grad_norm": 1.0822040456419593,
      "learning_rate": 1.5038376615237275e-06,
      "loss": 0.4455,
      "step": 15713
    },
    {
      "epoch": 0.8479844584749878,
      "grad_norm": 1.215595558886416,
      "learning_rate": 1.5034869608340873e-06,
      "loss": 0.5225,
      "step": 15714
    },
    {
      "epoch": 0.8480384221035022,
      "grad_norm": 0.9455240806719459,
      "learning_rate": 1.5031363750082698e-06,
      "loss": 0.4571,
      "step": 15715
    },
    {
      "epoch": 0.8480923857320166,
      "grad_norm": 1.0189232957429506,
      "learning_rate": 1.5027859040563513e-06,
      "loss": 0.4905,
      "step": 15716
    },
    {
      "epoch": 0.848146349360531,
      "grad_norm": 1.0045969371783228,
      "learning_rate": 1.5024355479884052e-06,
      "loss": 0.4549,
      "step": 15717
    },
    {
      "epoch": 0.8482003129890454,
      "grad_norm": 1.0064237149291162,
      "learning_rate": 1.5020853068145004e-06,
      "loss": 0.4408,
      "step": 15718
    },
    {
      "epoch": 0.8482542766175598,
      "grad_norm": 0.9655046603832823,
      "learning_rate": 1.5017351805447033e-06,
      "loss": 0.4198,
      "step": 15719
    },
    {
      "epoch": 0.8483082402460741,
      "grad_norm": 1.1765629565948794,
      "learning_rate": 1.5013851691890787e-06,
      "loss": 0.56,
      "step": 15720
    },
    {
      "epoch": 0.8483622038745885,
      "grad_norm": 0.9662417779794124,
      "learning_rate": 1.5010352727576832e-06,
      "loss": 0.4166,
      "step": 15721
    },
    {
      "epoch": 0.8484161675031029,
      "grad_norm": 1.2365817854568666,
      "learning_rate": 1.5006854912605745e-06,
      "loss": 0.4678,
      "step": 15722
    },
    {
      "epoch": 0.8484701311316173,
      "grad_norm": 0.9288138813109479,
      "learning_rate": 1.5003358247078061e-06,
      "loss": 0.4048,
      "step": 15723
    },
    {
      "epoch": 0.8485240947601317,
      "grad_norm": 1.1818950142331557,
      "learning_rate": 1.4999862731094272e-06,
      "loss": 0.3427,
      "step": 15724
    },
    {
      "epoch": 0.8485780583886461,
      "grad_norm": 1.012070299554473,
      "learning_rate": 1.499636836475485e-06,
      "loss": 0.438,
      "step": 15725
    },
    {
      "epoch": 0.8486320220171605,
      "grad_norm": 1.022620678299921,
      "learning_rate": 1.4992875148160216e-06,
      "loss": 0.3855,
      "step": 15726
    },
    {
      "epoch": 0.8486859856456748,
      "grad_norm": 0.9592909300998097,
      "learning_rate": 1.498938308141079e-06,
      "loss": 0.5367,
      "step": 15727
    },
    {
      "epoch": 0.8487399492741892,
      "grad_norm": 1.0212417101416968,
      "learning_rate": 1.4985892164606908e-06,
      "loss": 0.6372,
      "step": 15728
    },
    {
      "epoch": 0.8487939129027036,
      "grad_norm": 0.9225495333572487,
      "learning_rate": 1.498240239784892e-06,
      "loss": 0.3465,
      "step": 15729
    },
    {
      "epoch": 0.848847876531218,
      "grad_norm": 1.0116591351076443,
      "learning_rate": 1.4978913781237116e-06,
      "loss": 0.3832,
      "step": 15730
    },
    {
      "epoch": 0.8489018401597324,
      "grad_norm": 0.895027797929562,
      "learning_rate": 1.4975426314871774e-06,
      "loss": 0.4381,
      "step": 15731
    },
    {
      "epoch": 0.8489558037882468,
      "grad_norm": 0.9383857222527567,
      "learning_rate": 1.4971939998853118e-06,
      "loss": 0.4793,
      "step": 15732
    },
    {
      "epoch": 0.8490097674167612,
      "grad_norm": 1.0135467489670569,
      "learning_rate": 1.4968454833281354e-06,
      "loss": 0.422,
      "step": 15733
    },
    {
      "epoch": 0.8490637310452754,
      "grad_norm": 0.9789793481090988,
      "learning_rate": 1.4964970818256658e-06,
      "loss": 0.432,
      "step": 15734
    },
    {
      "epoch": 0.8491176946737898,
      "grad_norm": 0.9623969228342123,
      "learning_rate": 1.4961487953879142e-06,
      "loss": 0.4182,
      "step": 15735
    },
    {
      "epoch": 0.8491716583023042,
      "grad_norm": 0.9402201300082402,
      "learning_rate": 1.4958006240248918e-06,
      "loss": 0.3837,
      "step": 15736
    },
    {
      "epoch": 0.8492256219308186,
      "grad_norm": 1.2675933983595005,
      "learning_rate": 1.495452567746605e-06,
      "loss": 0.4639,
      "step": 15737
    },
    {
      "epoch": 0.849279585559333,
      "grad_norm": 1.20562386002614,
      "learning_rate": 1.4951046265630583e-06,
      "loss": 0.7426,
      "step": 15738
    },
    {
      "epoch": 0.8493335491878474,
      "grad_norm": 1.3534321249055234,
      "learning_rate": 1.4947568004842511e-06,
      "loss": 0.475,
      "step": 15739
    },
    {
      "epoch": 0.8493875128163618,
      "grad_norm": 0.9662891615368887,
      "learning_rate": 1.4944090895201802e-06,
      "loss": 0.5155,
      "step": 15740
    },
    {
      "epoch": 0.8494414764448761,
      "grad_norm": 1.0366847106062607,
      "learning_rate": 1.4940614936808406e-06,
      "loss": 0.3993,
      "step": 15741
    },
    {
      "epoch": 0.8494954400733905,
      "grad_norm": 0.9707253952123226,
      "learning_rate": 1.4937140129762207e-06,
      "loss": 0.538,
      "step": 15742
    },
    {
      "epoch": 0.8495494037019049,
      "grad_norm": 1.202387720064826,
      "learning_rate": 1.4933666474163078e-06,
      "loss": 0.5635,
      "step": 15743
    },
    {
      "epoch": 0.8496033673304193,
      "grad_norm": 1.0800994648502338,
      "learning_rate": 1.4930193970110859e-06,
      "loss": 0.4975,
      "step": 15744
    },
    {
      "epoch": 0.8496573309589337,
      "grad_norm": 0.8932254080114047,
      "learning_rate": 1.4926722617705353e-06,
      "loss": 0.3613,
      "step": 15745
    },
    {
      "epoch": 0.8497112945874481,
      "grad_norm": 1.319133754716954,
      "learning_rate": 1.492325241704633e-06,
      "loss": 0.4482,
      "step": 15746
    },
    {
      "epoch": 0.8497652582159625,
      "grad_norm": 1.299527876943168,
      "learning_rate": 1.4919783368233525e-06,
      "loss": 0.418,
      "step": 15747
    },
    {
      "epoch": 0.8498192218444768,
      "grad_norm": 1.0885212037106737,
      "learning_rate": 1.491631547136665e-06,
      "loss": 0.5217,
      "step": 15748
    },
    {
      "epoch": 0.8498731854729912,
      "grad_norm": 1.3232458407812677,
      "learning_rate": 1.4912848726545364e-06,
      "loss": 0.6393,
      "step": 15749
    },
    {
      "epoch": 0.8499271491015056,
      "grad_norm": 1.0639296034871653,
      "learning_rate": 1.4909383133869312e-06,
      "loss": 0.3847,
      "step": 15750
    },
    {
      "epoch": 0.84998111273002,
      "grad_norm": 0.9865725281776033,
      "learning_rate": 1.4905918693438098e-06,
      "loss": 0.5118,
      "step": 15751
    },
    {
      "epoch": 0.8500350763585344,
      "grad_norm": 1.0107564927626724,
      "learning_rate": 1.4902455405351295e-06,
      "loss": 0.4243,
      "step": 15752
    },
    {
      "epoch": 0.8500890399870488,
      "grad_norm": 0.9784188129070368,
      "learning_rate": 1.489899326970844e-06,
      "loss": 0.4338,
      "step": 15753
    },
    {
      "epoch": 0.8501430036155632,
      "grad_norm": 0.9414754023800073,
      "learning_rate": 1.489553228660904e-06,
      "loss": 0.41,
      "step": 15754
    },
    {
      "epoch": 0.8501969672440775,
      "grad_norm": 0.8985769242794818,
      "learning_rate": 1.489207245615256e-06,
      "loss": 0.4013,
      "step": 15755
    },
    {
      "epoch": 0.8502509308725918,
      "grad_norm": 0.8765765661657081,
      "learning_rate": 1.4888613778438443e-06,
      "loss": 0.3748,
      "step": 15756
    },
    {
      "epoch": 0.8503048945011062,
      "grad_norm": 0.9125343515198978,
      "learning_rate": 1.4885156253566096e-06,
      "loss": 0.3874,
      "step": 15757
    },
    {
      "epoch": 0.8503588581296206,
      "grad_norm": 0.9756591840293476,
      "learning_rate": 1.4881699881634892e-06,
      "loss": 0.514,
      "step": 15758
    },
    {
      "epoch": 0.850412821758135,
      "grad_norm": 1.1546873648066518,
      "learning_rate": 1.4878244662744169e-06,
      "loss": 0.4839,
      "step": 15759
    },
    {
      "epoch": 0.8504667853866494,
      "grad_norm": 1.023417048054535,
      "learning_rate": 1.4874790596993236e-06,
      "loss": 0.4756,
      "step": 15760
    },
    {
      "epoch": 0.8505207490151638,
      "grad_norm": 0.9578463887343279,
      "learning_rate": 1.4871337684481366e-06,
      "loss": 0.4548,
      "step": 15761
    },
    {
      "epoch": 0.8505747126436781,
      "grad_norm": 1.0469285446925227,
      "learning_rate": 1.48678859253078e-06,
      "loss": 0.4421,
      "step": 15762
    },
    {
      "epoch": 0.8506286762721925,
      "grad_norm": 0.9946958864502419,
      "learning_rate": 1.4864435319571742e-06,
      "loss": 0.4689,
      "step": 15763
    },
    {
      "epoch": 0.8506826399007069,
      "grad_norm": 1.1488126702329948,
      "learning_rate": 1.4860985867372365e-06,
      "loss": 0.4742,
      "step": 15764
    },
    {
      "epoch": 0.8507366035292213,
      "grad_norm": 0.8039318657227673,
      "learning_rate": 1.4857537568808808e-06,
      "loss": 0.3507,
      "step": 15765
    },
    {
      "epoch": 0.8507905671577357,
      "grad_norm": 0.9224411999445356,
      "learning_rate": 1.4854090423980186e-06,
      "loss": 0.4048,
      "step": 15766
    },
    {
      "epoch": 0.8508445307862501,
      "grad_norm": 0.9134814432106474,
      "learning_rate": 1.4850644432985567e-06,
      "loss": 0.4494,
      "step": 15767
    },
    {
      "epoch": 0.8508984944147644,
      "grad_norm": 1.121736983361893,
      "learning_rate": 1.4847199595924005e-06,
      "loss": 0.5818,
      "step": 15768
    },
    {
      "epoch": 0.8509524580432788,
      "grad_norm": 1.1620112856097646,
      "learning_rate": 1.484375591289449e-06,
      "loss": 0.5362,
      "step": 15769
    },
    {
      "epoch": 0.8510064216717932,
      "grad_norm": 1.1494533341107662,
      "learning_rate": 1.4840313383996001e-06,
      "loss": 0.5495,
      "step": 15770
    },
    {
      "epoch": 0.8510603853003076,
      "grad_norm": 1.0686185199518534,
      "learning_rate": 1.4836872009327491e-06,
      "loss": 0.4825,
      "step": 15771
    },
    {
      "epoch": 0.851114348928822,
      "grad_norm": 1.0419026466347046,
      "learning_rate": 1.4833431788987854e-06,
      "loss": 0.382,
      "step": 15772
    },
    {
      "epoch": 0.8511683125573364,
      "grad_norm": 1.0381524929024746,
      "learning_rate": 1.4829992723075976e-06,
      "loss": 0.4721,
      "step": 15773
    },
    {
      "epoch": 0.8512222761858508,
      "grad_norm": 1.0249330002717119,
      "learning_rate": 1.4826554811690697e-06,
      "loss": 0.3721,
      "step": 15774
    },
    {
      "epoch": 0.8512762398143651,
      "grad_norm": 0.8526058899312333,
      "learning_rate": 1.4823118054930834e-06,
      "loss": 0.498,
      "step": 15775
    },
    {
      "epoch": 0.8513302034428795,
      "grad_norm": 1.2195580877410226,
      "learning_rate": 1.4819682452895142e-06,
      "loss": 0.5318,
      "step": 15776
    },
    {
      "epoch": 0.8513841670713939,
      "grad_norm": 1.006026111709622,
      "learning_rate": 1.4816248005682379e-06,
      "loss": 0.4567,
      "step": 15777
    },
    {
      "epoch": 0.8514381306999083,
      "grad_norm": 1.133734495278117,
      "learning_rate": 1.4812814713391252e-06,
      "loss": 0.4928,
      "step": 15778
    },
    {
      "epoch": 0.8514920943284227,
      "grad_norm": 1.0579972772339776,
      "learning_rate": 1.480938257612043e-06,
      "loss": 0.4726,
      "step": 15779
    },
    {
      "epoch": 0.851546057956937,
      "grad_norm": 0.9204732811600153,
      "learning_rate": 1.4805951593968571e-06,
      "loss": 0.4825,
      "step": 15780
    },
    {
      "epoch": 0.8516000215854514,
      "grad_norm": 1.2094388911244773,
      "learning_rate": 1.480252176703427e-06,
      "loss": 0.6022,
      "step": 15781
    },
    {
      "epoch": 0.8516539852139657,
      "grad_norm": 0.8147993491552107,
      "learning_rate": 1.4799093095416117e-06,
      "loss": 0.24,
      "step": 15782
    },
    {
      "epoch": 0.8517079488424801,
      "grad_norm": 1.2079275173860349,
      "learning_rate": 1.4795665579212645e-06,
      "loss": 0.5638,
      "step": 15783
    },
    {
      "epoch": 0.8517619124709945,
      "grad_norm": 1.3134812046221398,
      "learning_rate": 1.4792239218522367e-06,
      "loss": 0.4358,
      "step": 15784
    },
    {
      "epoch": 0.8518158760995089,
      "grad_norm": 0.9413218946320672,
      "learning_rate": 1.4788814013443758e-06,
      "loss": 0.4035,
      "step": 15785
    },
    {
      "epoch": 0.8518698397280233,
      "grad_norm": 1.0440111736259523,
      "learning_rate": 1.4785389964075267e-06,
      "loss": 0.4921,
      "step": 15786
    },
    {
      "epoch": 0.8519238033565377,
      "grad_norm": 0.8433954335303583,
      "learning_rate": 1.4781967070515302e-06,
      "loss": 0.3445,
      "step": 15787
    },
    {
      "epoch": 0.8519777669850521,
      "grad_norm": 0.8984090171328069,
      "learning_rate": 1.477854533286224e-06,
      "loss": 0.3537,
      "step": 15788
    },
    {
      "epoch": 0.8520317306135664,
      "grad_norm": 0.7747507281965911,
      "learning_rate": 1.4775124751214436e-06,
      "loss": 0.4312,
      "step": 15789
    },
    {
      "epoch": 0.8520856942420808,
      "grad_norm": 1.080136267408936,
      "learning_rate": 1.4771705325670183e-06,
      "loss": 0.5117,
      "step": 15790
    },
    {
      "epoch": 0.8521396578705952,
      "grad_norm": 1.0170026921178028,
      "learning_rate": 1.4768287056327762e-06,
      "loss": 0.4868,
      "step": 15791
    },
    {
      "epoch": 0.8521936214991096,
      "grad_norm": 0.9500726582741089,
      "learning_rate": 1.4764869943285425e-06,
      "loss": 0.5448,
      "step": 15792
    },
    {
      "epoch": 0.852247585127624,
      "grad_norm": 1.0665257048718844,
      "learning_rate": 1.4761453986641383e-06,
      "loss": 0.4186,
      "step": 15793
    },
    {
      "epoch": 0.8523015487561384,
      "grad_norm": 1.1803345585722,
      "learning_rate": 1.475803918649381e-06,
      "loss": 0.5873,
      "step": 15794
    },
    {
      "epoch": 0.8523555123846528,
      "grad_norm": 1.0021283205644853,
      "learning_rate": 1.4754625542940848e-06,
      "loss": 0.405,
      "step": 15795
    },
    {
      "epoch": 0.8524094760131671,
      "grad_norm": 0.8136987213904351,
      "learning_rate": 1.475121305608062e-06,
      "loss": 0.2955,
      "step": 15796
    },
    {
      "epoch": 0.8524634396416815,
      "grad_norm": 1.0079844761549812,
      "learning_rate": 1.474780172601119e-06,
      "loss": 0.4548,
      "step": 15797
    },
    {
      "epoch": 0.8525174032701959,
      "grad_norm": 0.8728727798334589,
      "learning_rate": 1.4744391552830626e-06,
      "loss": 0.3893,
      "step": 15798
    },
    {
      "epoch": 0.8525713668987103,
      "grad_norm": 1.0942580555514418,
      "learning_rate": 1.474098253663691e-06,
      "loss": 0.5079,
      "step": 15799
    },
    {
      "epoch": 0.8526253305272247,
      "grad_norm": 0.960035837727071,
      "learning_rate": 1.4737574677528034e-06,
      "loss": 0.4878,
      "step": 15800
    },
    {
      "epoch": 0.8526792941557391,
      "grad_norm": 0.9338648183653308,
      "learning_rate": 1.4734167975601948e-06,
      "loss": 0.3326,
      "step": 15801
    },
    {
      "epoch": 0.8527332577842535,
      "grad_norm": 0.8973901080261126,
      "learning_rate": 1.4730762430956557e-06,
      "loss": 0.4335,
      "step": 15802
    },
    {
      "epoch": 0.8527872214127677,
      "grad_norm": 0.9691510010864917,
      "learning_rate": 1.4727358043689747e-06,
      "loss": 0.4065,
      "step": 15803
    },
    {
      "epoch": 0.8528411850412821,
      "grad_norm": 1.1289848961700548,
      "learning_rate": 1.4723954813899357e-06,
      "loss": 0.5244,
      "step": 15804
    },
    {
      "epoch": 0.8528951486697965,
      "grad_norm": 1.020518667326881,
      "learning_rate": 1.4720552741683203e-06,
      "loss": 0.4372,
      "step": 15805
    },
    {
      "epoch": 0.8529491122983109,
      "grad_norm": 0.8680844314276434,
      "learning_rate": 1.4717151827139059e-06,
      "loss": 0.3244,
      "step": 15806
    },
    {
      "epoch": 0.8530030759268253,
      "grad_norm": 1.0001988550738208,
      "learning_rate": 1.471375207036468e-06,
      "loss": 0.441,
      "step": 15807
    },
    {
      "epoch": 0.8530570395553397,
      "grad_norm": 1.0503053979060033,
      "learning_rate": 1.4710353471457767e-06,
      "loss": 0.3961,
      "step": 15808
    },
    {
      "epoch": 0.8531110031838541,
      "grad_norm": 1.0903334805389977,
      "learning_rate": 1.4706956030516004e-06,
      "loss": 0.4714,
      "step": 15809
    },
    {
      "epoch": 0.8531649668123684,
      "grad_norm": 1.037032592063396,
      "learning_rate": 1.4703559747637053e-06,
      "loss": 0.4099,
      "step": 15810
    },
    {
      "epoch": 0.8532189304408828,
      "grad_norm": 0.8901861589672014,
      "learning_rate": 1.47001646229185e-06,
      "loss": 0.3164,
      "step": 15811
    },
    {
      "epoch": 0.8532728940693972,
      "grad_norm": 1.1260546210390316,
      "learning_rate": 1.4696770656457934e-06,
      "loss": 0.4738,
      "step": 15812
    },
    {
      "epoch": 0.8533268576979116,
      "grad_norm": 1.1083632891910054,
      "learning_rate": 1.4693377848352902e-06,
      "loss": 0.6003,
      "step": 15813
    },
    {
      "epoch": 0.853380821326426,
      "grad_norm": 1.011466981322459,
      "learning_rate": 1.4689986198700924e-06,
      "loss": 0.5891,
      "step": 15814
    },
    {
      "epoch": 0.8534347849549404,
      "grad_norm": 0.9080254174991468,
      "learning_rate": 1.4686595707599466e-06,
      "loss": 0.3369,
      "step": 15815
    },
    {
      "epoch": 0.8534887485834548,
      "grad_norm": 1.156074163260614,
      "learning_rate": 1.4683206375145983e-06,
      "loss": 0.4933,
      "step": 15816
    },
    {
      "epoch": 0.8535427122119691,
      "grad_norm": 1.1892608864246066,
      "learning_rate": 1.4679818201437894e-06,
      "loss": 0.4603,
      "step": 15817
    },
    {
      "epoch": 0.8535966758404835,
      "grad_norm": 0.918170454751779,
      "learning_rate": 1.4676431186572561e-06,
      "loss": 0.4888,
      "step": 15818
    },
    {
      "epoch": 0.8536506394689979,
      "grad_norm": 1.1629390981585814,
      "learning_rate": 1.4673045330647344e-06,
      "loss": 0.5214,
      "step": 15819
    },
    {
      "epoch": 0.8537046030975123,
      "grad_norm": 1.0721371460333542,
      "learning_rate": 1.4669660633759553e-06,
      "loss": 0.4362,
      "step": 15820
    },
    {
      "epoch": 0.8537585667260267,
      "grad_norm": 0.9601621311010247,
      "learning_rate": 1.466627709600646e-06,
      "loss": 0.3606,
      "step": 15821
    },
    {
      "epoch": 0.8538125303545411,
      "grad_norm": 1.0065214315108326,
      "learning_rate": 1.4662894717485318e-06,
      "loss": 0.4377,
      "step": 15822
    },
    {
      "epoch": 0.8538664939830555,
      "grad_norm": 0.8390647145513496,
      "learning_rate": 1.465951349829334e-06,
      "loss": 0.332,
      "step": 15823
    },
    {
      "epoch": 0.8539204576115698,
      "grad_norm": 0.837788779839645,
      "learning_rate": 1.4656133438527716e-06,
      "loss": 0.323,
      "step": 15824
    },
    {
      "epoch": 0.8539744212400842,
      "grad_norm": 1.1952772635498945,
      "learning_rate": 1.4652754538285574e-06,
      "loss": 0.5159,
      "step": 15825
    },
    {
      "epoch": 0.8540283848685986,
      "grad_norm": 1.0857620036651507,
      "learning_rate": 1.4649376797664034e-06,
      "loss": 0.4396,
      "step": 15826
    },
    {
      "epoch": 0.854082348497113,
      "grad_norm": 1.1796186870998873,
      "learning_rate": 1.4646000216760173e-06,
      "loss": 0.6277,
      "step": 15827
    },
    {
      "epoch": 0.8541363121256274,
      "grad_norm": 0.9863713433548841,
      "learning_rate": 1.4642624795671043e-06,
      "loss": 0.5365,
      "step": 15828
    },
    {
      "epoch": 0.8541902757541417,
      "grad_norm": 1.153378483911638,
      "learning_rate": 1.4639250534493654e-06,
      "loss": 0.4688,
      "step": 15829
    },
    {
      "epoch": 0.854244239382656,
      "grad_norm": 0.9370545292126401,
      "learning_rate": 1.4635877433324994e-06,
      "loss": 0.4454,
      "step": 15830
    },
    {
      "epoch": 0.8542982030111704,
      "grad_norm": 0.9733987150656904,
      "learning_rate": 1.463250549226199e-06,
      "loss": 0.5954,
      "step": 15831
    },
    {
      "epoch": 0.8543521666396848,
      "grad_norm": 0.9480551003777602,
      "learning_rate": 1.4629134711401571e-06,
      "loss": 0.4911,
      "step": 15832
    },
    {
      "epoch": 0.8544061302681992,
      "grad_norm": 0.8342078927967626,
      "learning_rate": 1.4625765090840612e-06,
      "loss": 0.3804,
      "step": 15833
    },
    {
      "epoch": 0.8544600938967136,
      "grad_norm": 1.0866599174190996,
      "learning_rate": 1.462239663067596e-06,
      "loss": 0.4724,
      "step": 15834
    },
    {
      "epoch": 0.854514057525228,
      "grad_norm": 0.9327010145077209,
      "learning_rate": 1.4619029331004423e-06,
      "loss": 0.4429,
      "step": 15835
    },
    {
      "epoch": 0.8545680211537424,
      "grad_norm": 1.2182815759290246,
      "learning_rate": 1.4615663191922788e-06,
      "loss": 0.4346,
      "step": 15836
    },
    {
      "epoch": 0.8546219847822567,
      "grad_norm": 1.1245294249652618,
      "learning_rate": 1.4612298213527807e-06,
      "loss": 0.4539,
      "step": 15837
    },
    {
      "epoch": 0.8546759484107711,
      "grad_norm": 0.8170143521439835,
      "learning_rate": 1.4608934395916174e-06,
      "loss": 0.339,
      "step": 15838
    },
    {
      "epoch": 0.8547299120392855,
      "grad_norm": 1.0323241354808055,
      "learning_rate": 1.4605571739184582e-06,
      "loss": 0.6143,
      "step": 15839
    },
    {
      "epoch": 0.8547838756677999,
      "grad_norm": 0.9540524375503299,
      "learning_rate": 1.460221024342967e-06,
      "loss": 0.3567,
      "step": 15840
    },
    {
      "epoch": 0.8548378392963143,
      "grad_norm": 1.2224515217428633,
      "learning_rate": 1.4598849908748059e-06,
      "loss": 0.5034,
      "step": 15841
    },
    {
      "epoch": 0.8548918029248287,
      "grad_norm": 1.1314576200969684,
      "learning_rate": 1.459549073523632e-06,
      "loss": 0.3654,
      "step": 15842
    },
    {
      "epoch": 0.8549457665533431,
      "grad_norm": 0.9541611977051758,
      "learning_rate": 1.4592132722991007e-06,
      "loss": 0.3485,
      "step": 15843
    },
    {
      "epoch": 0.8549997301818574,
      "grad_norm": 1.0771320473297965,
      "learning_rate": 1.4588775872108634e-06,
      "loss": 0.4654,
      "step": 15844
    },
    {
      "epoch": 0.8550536938103718,
      "grad_norm": 1.1030949693992358,
      "learning_rate": 1.4585420182685669e-06,
      "loss": 0.4748,
      "step": 15845
    },
    {
      "epoch": 0.8551076574388862,
      "grad_norm": 1.4379408497840662,
      "learning_rate": 1.4582065654818561e-06,
      "loss": 0.6368,
      "step": 15846
    },
    {
      "epoch": 0.8551616210674006,
      "grad_norm": 1.255951601812409,
      "learning_rate": 1.4578712288603727e-06,
      "loss": 0.4439,
      "step": 15847
    },
    {
      "epoch": 0.855215584695915,
      "grad_norm": 1.081231307300269,
      "learning_rate": 1.4575360084137542e-06,
      "loss": 0.6327,
      "step": 15848
    },
    {
      "epoch": 0.8552695483244294,
      "grad_norm": 0.8629403049634136,
      "learning_rate": 1.4572009041516357e-06,
      "loss": 0.3976,
      "step": 15849
    },
    {
      "epoch": 0.8553235119529438,
      "grad_norm": 0.9276008678244068,
      "learning_rate": 1.456865916083648e-06,
      "loss": 0.3878,
      "step": 15850
    },
    {
      "epoch": 0.855377475581458,
      "grad_norm": 1.047688622786715,
      "learning_rate": 1.4565310442194193e-06,
      "loss": 0.52,
      "step": 15851
    },
    {
      "epoch": 0.8554314392099724,
      "grad_norm": 0.9749084213510643,
      "learning_rate": 1.456196288568574e-06,
      "loss": 0.4235,
      "step": 15852
    },
    {
      "epoch": 0.8554854028384868,
      "grad_norm": 0.8992123251891002,
      "learning_rate": 1.4558616491407335e-06,
      "loss": 0.3438,
      "step": 15853
    },
    {
      "epoch": 0.8555393664670012,
      "grad_norm": 1.0667766303489499,
      "learning_rate": 1.455527125945515e-06,
      "loss": 0.3558,
      "step": 15854
    },
    {
      "epoch": 0.8555933300955156,
      "grad_norm": 0.854030670602688,
      "learning_rate": 1.455192718992534e-06,
      "loss": 0.3772,
      "step": 15855
    },
    {
      "epoch": 0.85564729372403,
      "grad_norm": 0.9802884654391336,
      "learning_rate": 1.4548584282914014e-06,
      "loss": 0.5382,
      "step": 15856
    },
    {
      "epoch": 0.8557012573525444,
      "grad_norm": 0.8981881561280495,
      "learning_rate": 1.454524253851724e-06,
      "loss": 0.5086,
      "step": 15857
    },
    {
      "epoch": 0.8557552209810587,
      "grad_norm": 0.9353584485033251,
      "learning_rate": 1.4541901956831089e-06,
      "loss": 0.3782,
      "step": 15858
    },
    {
      "epoch": 0.8558091846095731,
      "grad_norm": 1.1259992995224728,
      "learning_rate": 1.4538562537951545e-06,
      "loss": 0.4391,
      "step": 15859
    },
    {
      "epoch": 0.8558631482380875,
      "grad_norm": 1.181240710942633,
      "learning_rate": 1.4535224281974597e-06,
      "loss": 0.5302,
      "step": 15860
    },
    {
      "epoch": 0.8559171118666019,
      "grad_norm": 1.0176253549943677,
      "learning_rate": 1.4531887188996187e-06,
      "loss": 0.3315,
      "step": 15861
    },
    {
      "epoch": 0.8559710754951163,
      "grad_norm": 1.08460360291916,
      "learning_rate": 1.4528551259112233e-06,
      "loss": 0.4559,
      "step": 15862
    },
    {
      "epoch": 0.8560250391236307,
      "grad_norm": 0.9684425954219358,
      "learning_rate": 1.4525216492418607e-06,
      "loss": 0.3626,
      "step": 15863
    },
    {
      "epoch": 0.8560790027521451,
      "grad_norm": 0.7642836153530456,
      "learning_rate": 1.4521882889011157e-06,
      "loss": 0.3131,
      "step": 15864
    },
    {
      "epoch": 0.8561329663806594,
      "grad_norm": 0.829008815355002,
      "learning_rate": 1.4518550448985702e-06,
      "loss": 0.3249,
      "step": 15865
    },
    {
      "epoch": 0.8561869300091738,
      "grad_norm": 1.0349547029256545,
      "learning_rate": 1.4515219172438002e-06,
      "loss": 0.4868,
      "step": 15866
    },
    {
      "epoch": 0.8562408936376882,
      "grad_norm": 1.0132757867236892,
      "learning_rate": 1.4511889059463807e-06,
      "loss": 0.5031,
      "step": 15867
    },
    {
      "epoch": 0.8562948572662026,
      "grad_norm": 1.0921997922649795,
      "learning_rate": 1.450856011015883e-06,
      "loss": 0.5425,
      "step": 15868
    },
    {
      "epoch": 0.856348820894717,
      "grad_norm": 1.050629999246534,
      "learning_rate": 1.4505232324618758e-06,
      "loss": 0.4621,
      "step": 15869
    },
    {
      "epoch": 0.8564027845232314,
      "grad_norm": 1.0932315655026048,
      "learning_rate": 1.4501905702939215e-06,
      "loss": 0.5528,
      "step": 15870
    },
    {
      "epoch": 0.8564567481517458,
      "grad_norm": 1.0359276172153218,
      "learning_rate": 1.4498580245215828e-06,
      "loss": 0.4534,
      "step": 15871
    },
    {
      "epoch": 0.8565107117802601,
      "grad_norm": 1.0163036056099466,
      "learning_rate": 1.4495255951544178e-06,
      "loss": 0.3357,
      "step": 15872
    },
    {
      "epoch": 0.8565646754087745,
      "grad_norm": 1.0669617782809329,
      "learning_rate": 1.4491932822019786e-06,
      "loss": 0.4171,
      "step": 15873
    },
    {
      "epoch": 0.8566186390372889,
      "grad_norm": 0.8592681192155163,
      "learning_rate": 1.448861085673818e-06,
      "loss": 0.3333,
      "step": 15874
    },
    {
      "epoch": 0.8566726026658033,
      "grad_norm": 1.1252787145913525,
      "learning_rate": 1.4485290055794832e-06,
      "loss": 0.5327,
      "step": 15875
    },
    {
      "epoch": 0.8567265662943176,
      "grad_norm": 0.8725678219792498,
      "learning_rate": 1.4481970419285179e-06,
      "loss": 0.4063,
      "step": 15876
    },
    {
      "epoch": 0.856780529922832,
      "grad_norm": 1.109102577345782,
      "learning_rate": 1.447865194730464e-06,
      "loss": 0.5195,
      "step": 15877
    },
    {
      "epoch": 0.8568344935513464,
      "grad_norm": 1.0604682812137793,
      "learning_rate": 1.4475334639948587e-06,
      "loss": 0.4866,
      "step": 15878
    },
    {
      "epoch": 0.8568884571798607,
      "grad_norm": 1.1563847223817985,
      "learning_rate": 1.4472018497312373e-06,
      "loss": 0.4441,
      "step": 15879
    },
    {
      "epoch": 0.8569424208083751,
      "grad_norm": 1.0909792206969944,
      "learning_rate": 1.446870351949129e-06,
      "loss": 0.5326,
      "step": 15880
    },
    {
      "epoch": 0.8569963844368895,
      "grad_norm": 1.2026644556031447,
      "learning_rate": 1.4465389706580624e-06,
      "loss": 0.5225,
      "step": 15881
    },
    {
      "epoch": 0.8570503480654039,
      "grad_norm": 0.997671093408529,
      "learning_rate": 1.446207705867561e-06,
      "loss": 0.3887,
      "step": 15882
    },
    {
      "epoch": 0.8571043116939183,
      "grad_norm": 0.9384232153161337,
      "learning_rate": 1.4458765575871465e-06,
      "loss": 0.5017,
      "step": 15883
    },
    {
      "epoch": 0.8571582753224327,
      "grad_norm": 0.9342080544699762,
      "learning_rate": 1.445545525826336e-06,
      "loss": 0.418,
      "step": 15884
    },
    {
      "epoch": 0.8572122389509471,
      "grad_norm": 1.0689499785096086,
      "learning_rate": 1.4452146105946438e-06,
      "loss": 0.4425,
      "step": 15885
    },
    {
      "epoch": 0.8572662025794614,
      "grad_norm": 1.1703820126289595,
      "learning_rate": 1.4448838119015814e-06,
      "loss": 0.5747,
      "step": 15886
    },
    {
      "epoch": 0.8573201662079758,
      "grad_norm": 1.024961206266867,
      "learning_rate": 1.444553129756655e-06,
      "loss": 0.4594,
      "step": 15887
    },
    {
      "epoch": 0.8573741298364902,
      "grad_norm": 1.0893436851946077,
      "learning_rate": 1.4442225641693697e-06,
      "loss": 0.4369,
      "step": 15888
    },
    {
      "epoch": 0.8574280934650046,
      "grad_norm": 1.0949898289595388,
      "learning_rate": 1.4438921151492258e-06,
      "loss": 0.3972,
      "step": 15889
    },
    {
      "epoch": 0.857482057093519,
      "grad_norm": 0.8469801304065431,
      "learning_rate": 1.4435617827057204e-06,
      "loss": 0.3808,
      "step": 15890
    },
    {
      "epoch": 0.8575360207220334,
      "grad_norm": 1.0605278101920335,
      "learning_rate": 1.4432315668483488e-06,
      "loss": 0.4227,
      "step": 15891
    },
    {
      "epoch": 0.8575899843505478,
      "grad_norm": 1.2457192297941235,
      "learning_rate": 1.4429014675866007e-06,
      "loss": 0.6217,
      "step": 15892
    },
    {
      "epoch": 0.8576439479790621,
      "grad_norm": 1.025495169455793,
      "learning_rate": 1.442571484929965e-06,
      "loss": 0.3927,
      "step": 15893
    },
    {
      "epoch": 0.8576979116075765,
      "grad_norm": 0.955404595589251,
      "learning_rate": 1.442241618887923e-06,
      "loss": 0.3838,
      "step": 15894
    },
    {
      "epoch": 0.8577518752360909,
      "grad_norm": 1.1734594831744913,
      "learning_rate": 1.4419118694699575e-06,
      "loss": 0.4931,
      "step": 15895
    },
    {
      "epoch": 0.8578058388646053,
      "grad_norm": 1.1938011189629854,
      "learning_rate": 1.4415822366855453e-06,
      "loss": 0.6036,
      "step": 15896
    },
    {
      "epoch": 0.8578598024931197,
      "grad_norm": 1.152551952517387,
      "learning_rate": 1.44125272054416e-06,
      "loss": 0.5432,
      "step": 15897
    },
    {
      "epoch": 0.8579137661216341,
      "grad_norm": 1.06589876494821,
      "learning_rate": 1.4409233210552728e-06,
      "loss": 0.4612,
      "step": 15898
    },
    {
      "epoch": 0.8579677297501483,
      "grad_norm": 0.9825946619655737,
      "learning_rate": 1.4405940382283505e-06,
      "loss": 0.4093,
      "step": 15899
    },
    {
      "epoch": 0.8580216933786627,
      "grad_norm": 1.097304330679007,
      "learning_rate": 1.4402648720728574e-06,
      "loss": 0.5552,
      "step": 15900
    },
    {
      "epoch": 0.8580756570071771,
      "grad_norm": 0.9962777717849847,
      "learning_rate": 1.439935822598254e-06,
      "loss": 0.3911,
      "step": 15901
    },
    {
      "epoch": 0.8581296206356915,
      "grad_norm": 0.9406688914542373,
      "learning_rate": 1.4396068898139979e-06,
      "loss": 0.3552,
      "step": 15902
    },
    {
      "epoch": 0.8581835842642059,
      "grad_norm": 1.2208059315184703,
      "learning_rate": 1.4392780737295418e-06,
      "loss": 0.5402,
      "step": 15903
    },
    {
      "epoch": 0.8582375478927203,
      "grad_norm": 1.1359782079116498,
      "learning_rate": 1.4389493743543379e-06,
      "loss": 0.4671,
      "step": 15904
    },
    {
      "epoch": 0.8582915115212347,
      "grad_norm": 1.009861463843404,
      "learning_rate": 1.4386207916978317e-06,
      "loss": 0.4471,
      "step": 15905
    },
    {
      "epoch": 0.858345475149749,
      "grad_norm": 0.9056339857402622,
      "learning_rate": 1.4382923257694679e-06,
      "loss": 0.39,
      "step": 15906
    },
    {
      "epoch": 0.8583994387782634,
      "grad_norm": 0.9386697005872333,
      "learning_rate": 1.4379639765786866e-06,
      "loss": 0.3542,
      "step": 15907
    },
    {
      "epoch": 0.8584534024067778,
      "grad_norm": 0.994111870969634,
      "learning_rate": 1.4376357441349248e-06,
      "loss": 0.3975,
      "step": 15908
    },
    {
      "epoch": 0.8585073660352922,
      "grad_norm": 1.0369749692462495,
      "learning_rate": 1.4373076284476164e-06,
      "loss": 0.3246,
      "step": 15909
    },
    {
      "epoch": 0.8585613296638066,
      "grad_norm": 0.984069845666785,
      "learning_rate": 1.4369796295261918e-06,
      "loss": 0.4683,
      "step": 15910
    },
    {
      "epoch": 0.858615293292321,
      "grad_norm": 0.9208427184139549,
      "learning_rate": 1.4366517473800782e-06,
      "loss": 0.3794,
      "step": 15911
    },
    {
      "epoch": 0.8586692569208354,
      "grad_norm": 0.9870178500037841,
      "learning_rate": 1.4363239820186985e-06,
      "loss": 0.3571,
      "step": 15912
    },
    {
      "epoch": 0.8587232205493497,
      "grad_norm": 1.1506313222705482,
      "learning_rate": 1.4359963334514754e-06,
      "loss": 0.4619,
      "step": 15913
    },
    {
      "epoch": 0.8587771841778641,
      "grad_norm": 0.9169992119429607,
      "learning_rate": 1.4356688016878226e-06,
      "loss": 0.3198,
      "step": 15914
    },
    {
      "epoch": 0.8588311478063785,
      "grad_norm": 1.1369237681324578,
      "learning_rate": 1.4353413867371557e-06,
      "loss": 0.5378,
      "step": 15915
    },
    {
      "epoch": 0.8588851114348929,
      "grad_norm": 0.9847893228974146,
      "learning_rate": 1.4350140886088836e-06,
      "loss": 0.4116,
      "step": 15916
    },
    {
      "epoch": 0.8589390750634073,
      "grad_norm": 1.1364652917023306,
      "learning_rate": 1.4346869073124148e-06,
      "loss": 0.5708,
      "step": 15917
    },
    {
      "epoch": 0.8589930386919217,
      "grad_norm": 0.8638220438489722,
      "learning_rate": 1.4343598428571514e-06,
      "loss": 0.3193,
      "step": 15918
    },
    {
      "epoch": 0.8590470023204361,
      "grad_norm": 0.9476155888701241,
      "learning_rate": 1.434032895252494e-06,
      "loss": 0.4945,
      "step": 15919
    },
    {
      "epoch": 0.8591009659489504,
      "grad_norm": 1.1576573645855195,
      "learning_rate": 1.433706064507841e-06,
      "loss": 0.4854,
      "step": 15920
    },
    {
      "epoch": 0.8591549295774648,
      "grad_norm": 0.9565109131368867,
      "learning_rate": 1.4333793506325832e-06,
      "loss": 0.3604,
      "step": 15921
    },
    {
      "epoch": 0.8592088932059792,
      "grad_norm": 0.9728394462776149,
      "learning_rate": 1.4330527536361121e-06,
      "loss": 0.4729,
      "step": 15922
    },
    {
      "epoch": 0.8592628568344935,
      "grad_norm": 1.0614510301418998,
      "learning_rate": 1.4327262735278142e-06,
      "loss": 0.5868,
      "step": 15923
    },
    {
      "epoch": 0.859316820463008,
      "grad_norm": 1.0824139467197467,
      "learning_rate": 1.432399910317072e-06,
      "loss": 0.3664,
      "step": 15924
    },
    {
      "epoch": 0.8593707840915223,
      "grad_norm": 1.0083672502427832,
      "learning_rate": 1.4320736640132673e-06,
      "loss": 0.3831,
      "step": 15925
    },
    {
      "epoch": 0.8594247477200367,
      "grad_norm": 0.9864891861566125,
      "learning_rate": 1.4317475346257755e-06,
      "loss": 0.4938,
      "step": 15926
    },
    {
      "epoch": 0.859478711348551,
      "grad_norm": 1.447390308701108,
      "learning_rate": 1.4314215221639715e-06,
      "loss": 0.5771,
      "step": 15927
    },
    {
      "epoch": 0.8595326749770654,
      "grad_norm": 0.9279244200635454,
      "learning_rate": 1.4310956266372228e-06,
      "loss": 0.347,
      "step": 15928
    },
    {
      "epoch": 0.8595866386055798,
      "grad_norm": 1.081968419002428,
      "learning_rate": 1.430769848054897e-06,
      "loss": 0.42,
      "step": 15929
    },
    {
      "epoch": 0.8596406022340942,
      "grad_norm": 0.76770282858641,
      "learning_rate": 1.430444186426358e-06,
      "loss": 0.3927,
      "step": 15930
    },
    {
      "epoch": 0.8596945658626086,
      "grad_norm": 0.9855381777597689,
      "learning_rate": 1.4301186417609645e-06,
      "loss": 0.4205,
      "step": 15931
    },
    {
      "epoch": 0.859748529491123,
      "grad_norm": 1.0318940910107028,
      "learning_rate": 1.429793214068073e-06,
      "loss": 0.4803,
      "step": 15932
    },
    {
      "epoch": 0.8598024931196374,
      "grad_norm": 0.9699503935014682,
      "learning_rate": 1.4294679033570383e-06,
      "loss": 0.4068,
      "step": 15933
    },
    {
      "epoch": 0.8598564567481517,
      "grad_norm": 1.2192412971609141,
      "learning_rate": 1.4291427096372096e-06,
      "loss": 0.5045,
      "step": 15934
    },
    {
      "epoch": 0.8599104203766661,
      "grad_norm": 1.0672920101549561,
      "learning_rate": 1.4288176329179315e-06,
      "loss": 0.4336,
      "step": 15935
    },
    {
      "epoch": 0.8599643840051805,
      "grad_norm": 1.0343209846855994,
      "learning_rate": 1.4284926732085486e-06,
      "loss": 0.46,
      "step": 15936
    },
    {
      "epoch": 0.8600183476336949,
      "grad_norm": 1.058777102408671,
      "learning_rate": 1.4281678305184002e-06,
      "loss": 0.4421,
      "step": 15937
    },
    {
      "epoch": 0.8600723112622093,
      "grad_norm": 0.9647120422009714,
      "learning_rate": 1.4278431048568225e-06,
      "loss": 0.4214,
      "step": 15938
    },
    {
      "epoch": 0.8601262748907237,
      "grad_norm": 1.4331258408884076,
      "learning_rate": 1.4275184962331486e-06,
      "loss": 0.6528,
      "step": 15939
    },
    {
      "epoch": 0.8601802385192381,
      "grad_norm": 1.121417448114517,
      "learning_rate": 1.427194004656708e-06,
      "loss": 0.4845,
      "step": 15940
    },
    {
      "epoch": 0.8602342021477524,
      "grad_norm": 0.9651501684627006,
      "learning_rate": 1.4268696301368278e-06,
      "loss": 0.5314,
      "step": 15941
    },
    {
      "epoch": 0.8602881657762668,
      "grad_norm": 1.145268699570462,
      "learning_rate": 1.4265453726828293e-06,
      "loss": 0.4915,
      "step": 15942
    },
    {
      "epoch": 0.8603421294047812,
      "grad_norm": 0.9791759234874867,
      "learning_rate": 1.4262212323040325e-06,
      "loss": 0.3752,
      "step": 15943
    },
    {
      "epoch": 0.8603960930332956,
      "grad_norm": 0.9051315390867263,
      "learning_rate": 1.4258972090097541e-06,
      "loss": 0.3417,
      "step": 15944
    },
    {
      "epoch": 0.86045005666181,
      "grad_norm": 1.0471410376825796,
      "learning_rate": 1.4255733028093066e-06,
      "loss": 0.4429,
      "step": 15945
    },
    {
      "epoch": 0.8605040202903244,
      "grad_norm": 1.1850548336388953,
      "learning_rate": 1.425249513711999e-06,
      "loss": 0.4425,
      "step": 15946
    },
    {
      "epoch": 0.8605579839188388,
      "grad_norm": 1.074197136046372,
      "learning_rate": 1.4249258417271373e-06,
      "loss": 0.5196,
      "step": 15947
    },
    {
      "epoch": 0.860611947547353,
      "grad_norm": 1.0162868048902367,
      "learning_rate": 1.4246022868640258e-06,
      "loss": 0.4103,
      "step": 15948
    },
    {
      "epoch": 0.8606659111758674,
      "grad_norm": 0.9348489496610543,
      "learning_rate": 1.4242788491319615e-06,
      "loss": 0.4252,
      "step": 15949
    },
    {
      "epoch": 0.8607198748043818,
      "grad_norm": 1.2017771273442426,
      "learning_rate": 1.4239555285402409e-06,
      "loss": 0.5225,
      "step": 15950
    },
    {
      "epoch": 0.8607738384328962,
      "grad_norm": 1.0507939921817726,
      "learning_rate": 1.4236323250981576e-06,
      "loss": 0.4735,
      "step": 15951
    },
    {
      "epoch": 0.8608278020614106,
      "grad_norm": 0.984977330016026,
      "learning_rate": 1.4233092388149997e-06,
      "loss": 0.4511,
      "step": 15952
    },
    {
      "epoch": 0.860881765689925,
      "grad_norm": 1.2144885822169362,
      "learning_rate": 1.4229862697000535e-06,
      "loss": 0.4916,
      "step": 15953
    },
    {
      "epoch": 0.8609357293184394,
      "grad_norm": 1.0438250539544476,
      "learning_rate": 1.4226634177626017e-06,
      "loss": 0.3664,
      "step": 15954
    },
    {
      "epoch": 0.8609896929469537,
      "grad_norm": 1.0744890804873033,
      "learning_rate": 1.4223406830119226e-06,
      "loss": 0.4929,
      "step": 15955
    },
    {
      "epoch": 0.8610436565754681,
      "grad_norm": 0.9204804239178237,
      "learning_rate": 1.4220180654572933e-06,
      "loss": 0.3916,
      "step": 15956
    },
    {
      "epoch": 0.8610976202039825,
      "grad_norm": 1.2693343315533419,
      "learning_rate": 1.4216955651079844e-06,
      "loss": 0.4356,
      "step": 15957
    },
    {
      "epoch": 0.8611515838324969,
      "grad_norm": 1.033125623539115,
      "learning_rate": 1.4213731819732661e-06,
      "loss": 0.5017,
      "step": 15958
    },
    {
      "epoch": 0.8612055474610113,
      "grad_norm": 1.129121797181992,
      "learning_rate": 1.4210509160624042e-06,
      "loss": 0.5605,
      "step": 15959
    },
    {
      "epoch": 0.8612595110895257,
      "grad_norm": 1.2563254114625597,
      "learning_rate": 1.4207287673846598e-06,
      "loss": 0.6091,
      "step": 15960
    },
    {
      "epoch": 0.8613134747180401,
      "grad_norm": 1.2560325420972729,
      "learning_rate": 1.4204067359492929e-06,
      "loss": 0.6504,
      "step": 15961
    },
    {
      "epoch": 0.8613674383465544,
      "grad_norm": 1.0869344249827126,
      "learning_rate": 1.4200848217655595e-06,
      "loss": 0.5154,
      "step": 15962
    },
    {
      "epoch": 0.8614214019750688,
      "grad_norm": 1.0487511700831937,
      "learning_rate": 1.4197630248427091e-06,
      "loss": 0.5328,
      "step": 15963
    },
    {
      "epoch": 0.8614753656035832,
      "grad_norm": 1.1312728803623147,
      "learning_rate": 1.4194413451899927e-06,
      "loss": 0.4699,
      "step": 15964
    },
    {
      "epoch": 0.8615293292320976,
      "grad_norm": 1.045843582259091,
      "learning_rate": 1.4191197828166548e-06,
      "loss": 0.4534,
      "step": 15965
    },
    {
      "epoch": 0.861583292860612,
      "grad_norm": 0.9374846363756233,
      "learning_rate": 1.4187983377319384e-06,
      "loss": 0.4319,
      "step": 15966
    },
    {
      "epoch": 0.8616372564891264,
      "grad_norm": 0.9224466198078187,
      "learning_rate": 1.4184770099450812e-06,
      "loss": 0.3596,
      "step": 15967
    },
    {
      "epoch": 0.8616912201176407,
      "grad_norm": 0.7949052251616,
      "learning_rate": 1.4181557994653191e-06,
      "loss": 0.3517,
      "step": 15968
    },
    {
      "epoch": 0.861745183746155,
      "grad_norm": 0.8976535165002293,
      "learning_rate": 1.4178347063018837e-06,
      "loss": 0.3541,
      "step": 15969
    },
    {
      "epoch": 0.8617991473746694,
      "grad_norm": 1.059221410318111,
      "learning_rate": 1.4175137304640032e-06,
      "loss": 0.4538,
      "step": 15970
    },
    {
      "epoch": 0.8618531110031838,
      "grad_norm": 1.1811230115637241,
      "learning_rate": 1.4171928719609034e-06,
      "loss": 0.599,
      "step": 15971
    },
    {
      "epoch": 0.8619070746316982,
      "grad_norm": 0.8752475460601022,
      "learning_rate": 1.4168721308018055e-06,
      "loss": 0.3394,
      "step": 15972
    },
    {
      "epoch": 0.8619610382602126,
      "grad_norm": 1.2339807609986468,
      "learning_rate": 1.4165515069959284e-06,
      "loss": 0.6023,
      "step": 15973
    },
    {
      "epoch": 0.862015001888727,
      "grad_norm": 1.2078786419694487,
      "learning_rate": 1.416231000552487e-06,
      "loss": 0.5201,
      "step": 15974
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 1.0615114035869715,
      "learning_rate": 1.4159106114806943e-06,
      "loss": 0.4179,
      "step": 15975
    },
    {
      "epoch": 0.8621229291457557,
      "grad_norm": 1.0768319746046544,
      "learning_rate": 1.4155903397897558e-06,
      "loss": 0.3841,
      "step": 15976
    },
    {
      "epoch": 0.8621768927742701,
      "grad_norm": 0.9069604757752442,
      "learning_rate": 1.4152701854888782e-06,
      "loss": 0.3718,
      "step": 15977
    },
    {
      "epoch": 0.8622308564027845,
      "grad_norm": 1.2947096656716894,
      "learning_rate": 1.4149501485872628e-06,
      "loss": 0.5341,
      "step": 15978
    },
    {
      "epoch": 0.8622848200312989,
      "grad_norm": 1.0154531319501023,
      "learning_rate": 1.4146302290941075e-06,
      "loss": 0.4093,
      "step": 15979
    },
    {
      "epoch": 0.8623387836598133,
      "grad_norm": 0.846573145614561,
      "learning_rate": 1.4143104270186076e-06,
      "loss": 0.4138,
      "step": 15980
    },
    {
      "epoch": 0.8623927472883277,
      "grad_norm": 0.961360587893776,
      "learning_rate": 1.413990742369954e-06,
      "loss": 0.4016,
      "step": 15981
    },
    {
      "epoch": 0.862446710916842,
      "grad_norm": 1.123183681422894,
      "learning_rate": 1.4136711751573362e-06,
      "loss": 0.4834,
      "step": 15982
    },
    {
      "epoch": 0.8625006745453564,
      "grad_norm": 1.0260146284690133,
      "learning_rate": 1.4133517253899373e-06,
      "loss": 0.377,
      "step": 15983
    },
    {
      "epoch": 0.8625546381738708,
      "grad_norm": 0.8427135377080797,
      "learning_rate": 1.4130323930769383e-06,
      "loss": 0.3091,
      "step": 15984
    },
    {
      "epoch": 0.8626086018023852,
      "grad_norm": 1.1817841602821046,
      "learning_rate": 1.4127131782275188e-06,
      "loss": 0.444,
      "step": 15985
    },
    {
      "epoch": 0.8626625654308996,
      "grad_norm": 1.247391221336531,
      "learning_rate": 1.4123940808508517e-06,
      "loss": 0.6012,
      "step": 15986
    },
    {
      "epoch": 0.862716529059414,
      "grad_norm": 0.9923787974961044,
      "learning_rate": 1.4120751009561095e-06,
      "loss": 0.4112,
      "step": 15987
    },
    {
      "epoch": 0.8627704926879284,
      "grad_norm": 1.066839641643829,
      "learning_rate": 1.4117562385524589e-06,
      "loss": 0.513,
      "step": 15988
    },
    {
      "epoch": 0.8628244563164427,
      "grad_norm": 0.9755266380696296,
      "learning_rate": 1.411437493649066e-06,
      "loss": 0.4549,
      "step": 15989
    },
    {
      "epoch": 0.8628784199449571,
      "grad_norm": 0.9984450423966413,
      "learning_rate": 1.4111188662550899e-06,
      "loss": 0.5047,
      "step": 15990
    },
    {
      "epoch": 0.8629323835734715,
      "grad_norm": 0.731918722567138,
      "learning_rate": 1.4108003563796888e-06,
      "loss": 0.2404,
      "step": 15991
    },
    {
      "epoch": 0.8629863472019859,
      "grad_norm": 1.018612702173676,
      "learning_rate": 1.4104819640320175e-06,
      "loss": 0.421,
      "step": 15992
    },
    {
      "epoch": 0.8630403108305003,
      "grad_norm": 0.8102219016189661,
      "learning_rate": 1.410163689221227e-06,
      "loss": 0.2974,
      "step": 15993
    },
    {
      "epoch": 0.8630942744590147,
      "grad_norm": 0.9865223773135915,
      "learning_rate": 1.409845531956464e-06,
      "loss": 0.4049,
      "step": 15994
    },
    {
      "epoch": 0.863148238087529,
      "grad_norm": 1.1540373982572925,
      "learning_rate": 1.4095274922468735e-06,
      "loss": 0.6225,
      "step": 15995
    },
    {
      "epoch": 0.8632022017160433,
      "grad_norm": 1.131291556117137,
      "learning_rate": 1.4092095701015967e-06,
      "loss": 0.4926,
      "step": 15996
    },
    {
      "epoch": 0.8632561653445577,
      "grad_norm": 0.9767646726005362,
      "learning_rate": 1.4088917655297698e-06,
      "loss": 0.5098,
      "step": 15997
    },
    {
      "epoch": 0.8633101289730721,
      "grad_norm": 1.2586348700649965,
      "learning_rate": 1.4085740785405272e-06,
      "loss": 0.4799,
      "step": 15998
    },
    {
      "epoch": 0.8633640926015865,
      "grad_norm": 1.203479650304201,
      "learning_rate": 1.4082565091429995e-06,
      "loss": 0.6744,
      "step": 15999
    },
    {
      "epoch": 0.8634180562301009,
      "grad_norm": 0.8379793734539012,
      "learning_rate": 1.407939057346314e-06,
      "loss": 0.33,
      "step": 16000
    },
    {
      "epoch": 0.8634180562301009,
      "eval_loss": 0.530173659324646,
      "eval_runtime": 159.5405,
      "eval_samples_per_second": 21.556,
      "eval_steps_per_second": 0.903,
      "step": 16000
    },
    {
      "epoch": 0.8634720198586153,
      "grad_norm": 0.887453276793935,
      "learning_rate": 1.4076217231595952e-06,
      "loss": 0.3317,
      "step": 16001
    },
    {
      "epoch": 0.8635259834871297,
      "grad_norm": 1.0613328518457446,
      "learning_rate": 1.4073045065919626e-06,
      "loss": 0.4116,
      "step": 16002
    },
    {
      "epoch": 0.863579947115644,
      "grad_norm": 0.94971006500543,
      "learning_rate": 1.4069874076525345e-06,
      "loss": 0.474,
      "step": 16003
    },
    {
      "epoch": 0.8636339107441584,
      "grad_norm": 1.0642828222655067,
      "learning_rate": 1.4066704263504236e-06,
      "loss": 0.5063,
      "step": 16004
    },
    {
      "epoch": 0.8636878743726728,
      "grad_norm": 0.968726643612459,
      "learning_rate": 1.4063535626947405e-06,
      "loss": 0.487,
      "step": 16005
    },
    {
      "epoch": 0.8637418380011872,
      "grad_norm": 0.8758106860713164,
      "learning_rate": 1.4060368166945925e-06,
      "loss": 0.5536,
      "step": 16006
    },
    {
      "epoch": 0.8637958016297016,
      "grad_norm": 1.0617495382858972,
      "learning_rate": 1.4057201883590832e-06,
      "loss": 0.4976,
      "step": 16007
    },
    {
      "epoch": 0.863849765258216,
      "grad_norm": 1.040591413242043,
      "learning_rate": 1.4054036776973123e-06,
      "loss": 0.5468,
      "step": 16008
    },
    {
      "epoch": 0.8639037288867304,
      "grad_norm": 1.084507076859897,
      "learning_rate": 1.4050872847183773e-06,
      "loss": 0.4374,
      "step": 16009
    },
    {
      "epoch": 0.8639576925152447,
      "grad_norm": 1.0125501252786353,
      "learning_rate": 1.4047710094313721e-06,
      "loss": 0.4792,
      "step": 16010
    },
    {
      "epoch": 0.8640116561437591,
      "grad_norm": 1.1550284623050406,
      "learning_rate": 1.404454851845385e-06,
      "loss": 0.5422,
      "step": 16011
    },
    {
      "epoch": 0.8640656197722735,
      "grad_norm": 1.0653542174368869,
      "learning_rate": 1.4041388119695034e-06,
      "loss": 0.4778,
      "step": 16012
    },
    {
      "epoch": 0.8641195834007879,
      "grad_norm": 1.0171155384734847,
      "learning_rate": 1.4038228898128115e-06,
      "loss": 0.4549,
      "step": 16013
    },
    {
      "epoch": 0.8641735470293023,
      "grad_norm": 0.9157559560410936,
      "learning_rate": 1.4035070853843886e-06,
      "loss": 0.342,
      "step": 16014
    },
    {
      "epoch": 0.8642275106578167,
      "grad_norm": 0.9719171588895529,
      "learning_rate": 1.4031913986933112e-06,
      "loss": 0.4759,
      "step": 16015
    },
    {
      "epoch": 0.8642814742863311,
      "grad_norm": 0.771018276778794,
      "learning_rate": 1.4028758297486524e-06,
      "loss": 0.2712,
      "step": 16016
    },
    {
      "epoch": 0.8643354379148454,
      "grad_norm": 1.0295528936070981,
      "learning_rate": 1.4025603785594826e-06,
      "loss": 0.4352,
      "step": 16017
    },
    {
      "epoch": 0.8643894015433597,
      "grad_norm": 0.819469310851545,
      "learning_rate": 1.4022450451348679e-06,
      "loss": 0.3663,
      "step": 16018
    },
    {
      "epoch": 0.8644433651718741,
      "grad_norm": 0.92170869806711,
      "learning_rate": 1.40192982948387e-06,
      "loss": 0.4241,
      "step": 16019
    },
    {
      "epoch": 0.8644973288003885,
      "grad_norm": 1.007724800725122,
      "learning_rate": 1.40161473161555e-06,
      "loss": 0.4388,
      "step": 16020
    },
    {
      "epoch": 0.8645512924289029,
      "grad_norm": 1.086398251259194,
      "learning_rate": 1.401299751538964e-06,
      "loss": 0.6251,
      "step": 16021
    },
    {
      "epoch": 0.8646052560574173,
      "grad_norm": 0.9887508220337071,
      "learning_rate": 1.400984889263164e-06,
      "loss": 0.4632,
      "step": 16022
    },
    {
      "epoch": 0.8646592196859317,
      "grad_norm": 1.1253776939820697,
      "learning_rate": 1.4006701447972004e-06,
      "loss": 0.4686,
      "step": 16023
    },
    {
      "epoch": 0.864713183314446,
      "grad_norm": 0.8844839621485916,
      "learning_rate": 1.40035551815012e-06,
      "loss": 0.3742,
      "step": 16024
    },
    {
      "epoch": 0.8647671469429604,
      "grad_norm": 1.0639229825836636,
      "learning_rate": 1.4000410093309632e-06,
      "loss": 0.4578,
      "step": 16025
    },
    {
      "epoch": 0.8648211105714748,
      "grad_norm": 1.0013419126193146,
      "learning_rate": 1.3997266183487704e-06,
      "loss": 0.3754,
      "step": 16026
    },
    {
      "epoch": 0.8648750741999892,
      "grad_norm": 1.290661101417018,
      "learning_rate": 1.399412345212578e-06,
      "loss": 0.567,
      "step": 16027
    },
    {
      "epoch": 0.8649290378285036,
      "grad_norm": 0.9474516112739214,
      "learning_rate": 1.399098189931418e-06,
      "loss": 0.4514,
      "step": 16028
    },
    {
      "epoch": 0.864983001457018,
      "grad_norm": 0.955052883263688,
      "learning_rate": 1.3987841525143192e-06,
      "loss": 0.4875,
      "step": 16029
    },
    {
      "epoch": 0.8650369650855324,
      "grad_norm": 1.079703992629002,
      "learning_rate": 1.398470232970308e-06,
      "loss": 0.6401,
      "step": 16030
    },
    {
      "epoch": 0.8650909287140467,
      "grad_norm": 0.993149832037558,
      "learning_rate": 1.398156431308408e-06,
      "loss": 0.4345,
      "step": 16031
    },
    {
      "epoch": 0.8651448923425611,
      "grad_norm": 1.1694787954590027,
      "learning_rate": 1.3978427475376355e-06,
      "loss": 0.5396,
      "step": 16032
    },
    {
      "epoch": 0.8651988559710755,
      "grad_norm": 0.8604624256722553,
      "learning_rate": 1.3975291816670078e-06,
      "loss": 0.3088,
      "step": 16033
    },
    {
      "epoch": 0.8652528195995899,
      "grad_norm": 1.0829544121259367,
      "learning_rate": 1.3972157337055365e-06,
      "loss": 0.6836,
      "step": 16034
    },
    {
      "epoch": 0.8653067832281043,
      "grad_norm": 1.1240740643668043,
      "learning_rate": 1.3969024036622303e-06,
      "loss": 0.4721,
      "step": 16035
    },
    {
      "epoch": 0.8653607468566187,
      "grad_norm": 1.0640674801608045,
      "learning_rate": 1.3965891915460946e-06,
      "loss": 0.5535,
      "step": 16036
    },
    {
      "epoch": 0.865414710485133,
      "grad_norm": 0.9907022767114316,
      "learning_rate": 1.3962760973661333e-06,
      "loss": 0.4608,
      "step": 16037
    },
    {
      "epoch": 0.8654686741136474,
      "grad_norm": 1.2247088910889385,
      "learning_rate": 1.395963121131342e-06,
      "loss": 0.4413,
      "step": 16038
    },
    {
      "epoch": 0.8655226377421618,
      "grad_norm": 1.009051020017518,
      "learning_rate": 1.395650262850718e-06,
      "loss": 0.4979,
      "step": 16039
    },
    {
      "epoch": 0.8655766013706762,
      "grad_norm": 1.1772209284283381,
      "learning_rate": 1.3953375225332519e-06,
      "loss": 0.5238,
      "step": 16040
    },
    {
      "epoch": 0.8656305649991906,
      "grad_norm": 1.1297710078642407,
      "learning_rate": 1.3950249001879334e-06,
      "loss": 0.505,
      "step": 16041
    },
    {
      "epoch": 0.865684528627705,
      "grad_norm": 1.1385790350210376,
      "learning_rate": 1.3947123958237469e-06,
      "loss": 0.5596,
      "step": 16042
    },
    {
      "epoch": 0.8657384922562193,
      "grad_norm": 0.9890371350841766,
      "learning_rate": 1.394400009449674e-06,
      "loss": 0.4273,
      "step": 16043
    },
    {
      "epoch": 0.8657924558847336,
      "grad_norm": 0.9737856023223564,
      "learning_rate": 1.3940877410746948e-06,
      "loss": 0.4287,
      "step": 16044
    },
    {
      "epoch": 0.865846419513248,
      "grad_norm": 0.9781804679410638,
      "learning_rate": 1.393775590707781e-06,
      "loss": 0.4705,
      "step": 16045
    },
    {
      "epoch": 0.8659003831417624,
      "grad_norm": 1.0263800575123965,
      "learning_rate": 1.3934635583579062e-06,
      "loss": 0.4649,
      "step": 16046
    },
    {
      "epoch": 0.8659543467702768,
      "grad_norm": 1.0996046726559146,
      "learning_rate": 1.3931516440340382e-06,
      "loss": 0.4406,
      "step": 16047
    },
    {
      "epoch": 0.8660083103987912,
      "grad_norm": 1.1046343841556652,
      "learning_rate": 1.392839847745141e-06,
      "loss": 0.4062,
      "step": 16048
    },
    {
      "epoch": 0.8660622740273056,
      "grad_norm": 0.8652012530827887,
      "learning_rate": 1.3925281695001772e-06,
      "loss": 0.3848,
      "step": 16049
    },
    {
      "epoch": 0.86611623765582,
      "grad_norm": 0.9605319806712386,
      "learning_rate": 1.392216609308104e-06,
      "loss": 0.4347,
      "step": 16050
    },
    {
      "epoch": 0.8661702012843343,
      "grad_norm": 1.1102586264780712,
      "learning_rate": 1.3919051671778758e-06,
      "loss": 0.6059,
      "step": 16051
    },
    {
      "epoch": 0.8662241649128487,
      "grad_norm": 0.8940590130358088,
      "learning_rate": 1.3915938431184442e-06,
      "loss": 0.4109,
      "step": 16052
    },
    {
      "epoch": 0.8662781285413631,
      "grad_norm": 1.1791815894272986,
      "learning_rate": 1.3912826371387575e-06,
      "loss": 0.5774,
      "step": 16053
    },
    {
      "epoch": 0.8663320921698775,
      "grad_norm": 1.0620188208551782,
      "learning_rate": 1.3909715492477588e-06,
      "loss": 0.4613,
      "step": 16054
    },
    {
      "epoch": 0.8663860557983919,
      "grad_norm": 1.1043702610923958,
      "learning_rate": 1.390660579454389e-06,
      "loss": 0.4283,
      "step": 16055
    },
    {
      "epoch": 0.8664400194269063,
      "grad_norm": 1.0162210616030949,
      "learning_rate": 1.390349727767587e-06,
      "loss": 0.6004,
      "step": 16056
    },
    {
      "epoch": 0.8664939830554207,
      "grad_norm": 0.997451131207885,
      "learning_rate": 1.3900389941962861e-06,
      "loss": 0.4671,
      "step": 16057
    },
    {
      "epoch": 0.866547946683935,
      "grad_norm": 1.0554052076574567,
      "learning_rate": 1.389728378749417e-06,
      "loss": 0.4872,
      "step": 16058
    },
    {
      "epoch": 0.8666019103124494,
      "grad_norm": 1.0251757289552526,
      "learning_rate": 1.3894178814359077e-06,
      "loss": 0.4916,
      "step": 16059
    },
    {
      "epoch": 0.8666558739409638,
      "grad_norm": 1.046077585448734,
      "learning_rate": 1.3891075022646822e-06,
      "loss": 0.431,
      "step": 16060
    },
    {
      "epoch": 0.8667098375694782,
      "grad_norm": 1.31230653298192,
      "learning_rate": 1.3887972412446608e-06,
      "loss": 0.6478,
      "step": 16061
    },
    {
      "epoch": 0.8667638011979926,
      "grad_norm": 1.0632964279311767,
      "learning_rate": 1.3884870983847604e-06,
      "loss": 0.4656,
      "step": 16062
    },
    {
      "epoch": 0.866817764826507,
      "grad_norm": 0.8854448388034084,
      "learning_rate": 1.3881770736938954e-06,
      "loss": 0.419,
      "step": 16063
    },
    {
      "epoch": 0.8668717284550214,
      "grad_norm": 1.0009328867494571,
      "learning_rate": 1.3878671671809759e-06,
      "loss": 0.436,
      "step": 16064
    },
    {
      "epoch": 0.8669256920835356,
      "grad_norm": 0.9921178278969827,
      "learning_rate": 1.3875573788549095e-06,
      "loss": 0.4074,
      "step": 16065
    },
    {
      "epoch": 0.86697965571205,
      "grad_norm": 1.156789898969116,
      "learning_rate": 1.387247708724599e-06,
      "loss": 0.4307,
      "step": 16066
    },
    {
      "epoch": 0.8670336193405644,
      "grad_norm": 0.860507172211298,
      "learning_rate": 1.386938156798945e-06,
      "loss": 0.4075,
      "step": 16067
    },
    {
      "epoch": 0.8670875829690788,
      "grad_norm": 1.1254649436321356,
      "learning_rate": 1.3866287230868445e-06,
      "loss": 0.5214,
      "step": 16068
    },
    {
      "epoch": 0.8671415465975932,
      "grad_norm": 1.118931643320334,
      "learning_rate": 1.3863194075971904e-06,
      "loss": 0.3571,
      "step": 16069
    },
    {
      "epoch": 0.8671955102261076,
      "grad_norm": 0.8595429709281893,
      "learning_rate": 1.3860102103388728e-06,
      "loss": 0.3689,
      "step": 16070
    },
    {
      "epoch": 0.867249473854622,
      "grad_norm": 0.9526729450411437,
      "learning_rate": 1.3857011313207791e-06,
      "loss": 0.3684,
      "step": 16071
    },
    {
      "epoch": 0.8673034374831363,
      "grad_norm": 1.0246077664936666,
      "learning_rate": 1.385392170551793e-06,
      "loss": 0.4082,
      "step": 16072
    },
    {
      "epoch": 0.8673574011116507,
      "grad_norm": 1.0617800257989989,
      "learning_rate": 1.3850833280407924e-06,
      "loss": 0.574,
      "step": 16073
    },
    {
      "epoch": 0.8674113647401651,
      "grad_norm": 0.9404674090415638,
      "learning_rate": 1.3847746037966541e-06,
      "loss": 0.4172,
      "step": 16074
    },
    {
      "epoch": 0.8674653283686795,
      "grad_norm": 1.124740125721108,
      "learning_rate": 1.384465997828253e-06,
      "loss": 0.5465,
      "step": 16075
    },
    {
      "epoch": 0.8675192919971939,
      "grad_norm": 1.136932845887837,
      "learning_rate": 1.384157510144457e-06,
      "loss": 0.5971,
      "step": 16076
    },
    {
      "epoch": 0.8675732556257083,
      "grad_norm": 0.9855516888648608,
      "learning_rate": 1.3838491407541329e-06,
      "loss": 0.4069,
      "step": 16077
    },
    {
      "epoch": 0.8676272192542227,
      "grad_norm": 0.9351424469033709,
      "learning_rate": 1.3835408896661437e-06,
      "loss": 0.321,
      "step": 16078
    },
    {
      "epoch": 0.867681182882737,
      "grad_norm": 1.0285498158706627,
      "learning_rate": 1.3832327568893496e-06,
      "loss": 0.4238,
      "step": 16079
    },
    {
      "epoch": 0.8677351465112514,
      "grad_norm": 1.0379522360000892,
      "learning_rate": 1.3829247424326044e-06,
      "loss": 0.4773,
      "step": 16080
    },
    {
      "epoch": 0.8677891101397658,
      "grad_norm": 0.939613253615824,
      "learning_rate": 1.382616846304763e-06,
      "loss": 0.388,
      "step": 16081
    },
    {
      "epoch": 0.8678430737682802,
      "grad_norm": 0.9539592372622561,
      "learning_rate": 1.3823090685146728e-06,
      "loss": 0.3699,
      "step": 16082
    },
    {
      "epoch": 0.8678970373967946,
      "grad_norm": 1.2394409833549738,
      "learning_rate": 1.3820014090711808e-06,
      "loss": 0.5191,
      "step": 16083
    },
    {
      "epoch": 0.867951001025309,
      "grad_norm": 0.9562219117111755,
      "learning_rate": 1.3816938679831299e-06,
      "loss": 0.4508,
      "step": 16084
    },
    {
      "epoch": 0.8680049646538234,
      "grad_norm": 0.9186425441389877,
      "learning_rate": 1.3813864452593578e-06,
      "loss": 0.4134,
      "step": 16085
    },
    {
      "epoch": 0.8680589282823377,
      "grad_norm": 1.065583652826002,
      "learning_rate": 1.3810791409087022e-06,
      "loss": 0.4189,
      "step": 16086
    },
    {
      "epoch": 0.8681128919108521,
      "grad_norm": 0.9330314109104152,
      "learning_rate": 1.3807719549399926e-06,
      "loss": 0.4813,
      "step": 16087
    },
    {
      "epoch": 0.8681668555393665,
      "grad_norm": 1.0810382849207258,
      "learning_rate": 1.3804648873620597e-06,
      "loss": 0.5205,
      "step": 16088
    },
    {
      "epoch": 0.8682208191678809,
      "grad_norm": 0.9933615974959051,
      "learning_rate": 1.380157938183728e-06,
      "loss": 0.4406,
      "step": 16089
    },
    {
      "epoch": 0.8682747827963953,
      "grad_norm": 1.0261553517428226,
      "learning_rate": 1.3798511074138204e-06,
      "loss": 0.4614,
      "step": 16090
    },
    {
      "epoch": 0.8683287464249096,
      "grad_norm": 1.236979341473934,
      "learning_rate": 1.3795443950611548e-06,
      "loss": 0.4813,
      "step": 16091
    },
    {
      "epoch": 0.868382710053424,
      "grad_norm": 0.9875497459273016,
      "learning_rate": 1.3792378011345464e-06,
      "loss": 0.3791,
      "step": 16092
    },
    {
      "epoch": 0.8684366736819383,
      "grad_norm": 0.98501396240504,
      "learning_rate": 1.3789313256428083e-06,
      "loss": 0.3687,
      "step": 16093
    },
    {
      "epoch": 0.8684906373104527,
      "grad_norm": 1.1451115166223478,
      "learning_rate": 1.3786249685947473e-06,
      "loss": 0.4908,
      "step": 16094
    },
    {
      "epoch": 0.8685446009389671,
      "grad_norm": 0.9149378750060622,
      "learning_rate": 1.378318729999169e-06,
      "loss": 0.3245,
      "step": 16095
    },
    {
      "epoch": 0.8685985645674815,
      "grad_norm": 1.0195229136598205,
      "learning_rate": 1.378012609864875e-06,
      "loss": 0.3878,
      "step": 16096
    },
    {
      "epoch": 0.8686525281959959,
      "grad_norm": 0.8680100036010613,
      "learning_rate": 1.3777066082006638e-06,
      "loss": 0.3232,
      "step": 16097
    },
    {
      "epoch": 0.8687064918245103,
      "grad_norm": 0.9386649720105785,
      "learning_rate": 1.3774007250153293e-06,
      "loss": 0.5095,
      "step": 16098
    },
    {
      "epoch": 0.8687604554530247,
      "grad_norm": 0.9952900314270993,
      "learning_rate": 1.3770949603176643e-06,
      "loss": 0.5587,
      "step": 16099
    },
    {
      "epoch": 0.868814419081539,
      "grad_norm": 0.9876989212897406,
      "learning_rate": 1.3767893141164567e-06,
      "loss": 0.4281,
      "step": 16100
    },
    {
      "epoch": 0.8688683827100534,
      "grad_norm": 1.2985161861331027,
      "learning_rate": 1.376483786420489e-06,
      "loss": 0.8057,
      "step": 16101
    },
    {
      "epoch": 0.8689223463385678,
      "grad_norm": 1.1783287776719449,
      "learning_rate": 1.3761783772385445e-06,
      "loss": 0.4225,
      "step": 16102
    },
    {
      "epoch": 0.8689763099670822,
      "grad_norm": 1.1411742002104799,
      "learning_rate": 1.3758730865794002e-06,
      "loss": 0.4815,
      "step": 16103
    },
    {
      "epoch": 0.8690302735955966,
      "grad_norm": 1.0747917860418272,
      "learning_rate": 1.37556791445183e-06,
      "loss": 0.5573,
      "step": 16104
    },
    {
      "epoch": 0.869084237224111,
      "grad_norm": 0.9664008391731783,
      "learning_rate": 1.3752628608646056e-06,
      "loss": 0.5101,
      "step": 16105
    },
    {
      "epoch": 0.8691382008526253,
      "grad_norm": 0.8700667458863904,
      "learning_rate": 1.3749579258264946e-06,
      "loss": 0.4118,
      "step": 16106
    },
    {
      "epoch": 0.8691921644811397,
      "grad_norm": 1.0715963139976257,
      "learning_rate": 1.3746531093462607e-06,
      "loss": 0.5085,
      "step": 16107
    },
    {
      "epoch": 0.8692461281096541,
      "grad_norm": 0.9753353159673824,
      "learning_rate": 1.3743484114326643e-06,
      "loss": 0.4375,
      "step": 16108
    },
    {
      "epoch": 0.8693000917381685,
      "grad_norm": 0.9021106636994022,
      "learning_rate": 1.3740438320944637e-06,
      "loss": 0.3694,
      "step": 16109
    },
    {
      "epoch": 0.8693540553666829,
      "grad_norm": 1.105276085436314,
      "learning_rate": 1.3737393713404127e-06,
      "loss": 0.438,
      "step": 16110
    },
    {
      "epoch": 0.8694080189951973,
      "grad_norm": 1.1391061817507306,
      "learning_rate": 1.3734350291792611e-06,
      "loss": 0.6086,
      "step": 16111
    },
    {
      "epoch": 0.8694619826237117,
      "grad_norm": 0.9773434045830147,
      "learning_rate": 1.3731308056197559e-06,
      "loss": 0.4328,
      "step": 16112
    },
    {
      "epoch": 0.869515946252226,
      "grad_norm": 1.1565451872062975,
      "learning_rate": 1.372826700670643e-06,
      "loss": 0.6308,
      "step": 16113
    },
    {
      "epoch": 0.8695699098807403,
      "grad_norm": 0.8256899847413528,
      "learning_rate": 1.3725227143406595e-06,
      "loss": 0.3108,
      "step": 16114
    },
    {
      "epoch": 0.8696238735092547,
      "grad_norm": 0.9562243797391092,
      "learning_rate": 1.372218846638544e-06,
      "loss": 0.4535,
      "step": 16115
    },
    {
      "epoch": 0.8696778371377691,
      "grad_norm": 1.0428583380074818,
      "learning_rate": 1.3719150975730294e-06,
      "loss": 0.5863,
      "step": 16116
    },
    {
      "epoch": 0.8697318007662835,
      "grad_norm": 1.531613523366415,
      "learning_rate": 1.3716114671528457e-06,
      "loss": 0.6786,
      "step": 16117
    },
    {
      "epoch": 0.8697857643947979,
      "grad_norm": 0.8599125551370869,
      "learning_rate": 1.3713079553867204e-06,
      "loss": 0.3009,
      "step": 16118
    },
    {
      "epoch": 0.8698397280233123,
      "grad_norm": 1.2156802473526944,
      "learning_rate": 1.3710045622833763e-06,
      "loss": 0.4883,
      "step": 16119
    },
    {
      "epoch": 0.8698936916518266,
      "grad_norm": 1.082416925571627,
      "learning_rate": 1.3707012878515336e-06,
      "loss": 0.522,
      "step": 16120
    },
    {
      "epoch": 0.869947655280341,
      "grad_norm": 0.9109854946275128,
      "learning_rate": 1.3703981320999072e-06,
      "loss": 0.3727,
      "step": 16121
    },
    {
      "epoch": 0.8700016189088554,
      "grad_norm": 1.1797095393090407,
      "learning_rate": 1.3700950950372113e-06,
      "loss": 0.4824,
      "step": 16122
    },
    {
      "epoch": 0.8700555825373698,
      "grad_norm": 0.8815374069717248,
      "learning_rate": 1.3697921766721556e-06,
      "loss": 0.3721,
      "step": 16123
    },
    {
      "epoch": 0.8701095461658842,
      "grad_norm": 1.003291726951096,
      "learning_rate": 1.3694893770134459e-06,
      "loss": 0.4137,
      "step": 16124
    },
    {
      "epoch": 0.8701635097943986,
      "grad_norm": 1.1663913342693344,
      "learning_rate": 1.3691866960697853e-06,
      "loss": 0.4814,
      "step": 16125
    },
    {
      "epoch": 0.870217473422913,
      "grad_norm": 1.1046820591635083,
      "learning_rate": 1.3688841338498727e-06,
      "loss": 0.4396,
      "step": 16126
    },
    {
      "epoch": 0.8702714370514273,
      "grad_norm": 0.9521805725281192,
      "learning_rate": 1.368581690362405e-06,
      "loss": 0.5982,
      "step": 16127
    },
    {
      "epoch": 0.8703254006799417,
      "grad_norm": 1.267862301747905,
      "learning_rate": 1.3682793656160734e-06,
      "loss": 0.541,
      "step": 16128
    },
    {
      "epoch": 0.8703793643084561,
      "grad_norm": 1.0201033485188375,
      "learning_rate": 1.3679771596195672e-06,
      "loss": 0.4457,
      "step": 16129
    },
    {
      "epoch": 0.8704333279369705,
      "grad_norm": 1.043246794389693,
      "learning_rate": 1.367675072381573e-06,
      "loss": 0.4474,
      "step": 16130
    },
    {
      "epoch": 0.8704872915654849,
      "grad_norm": 1.0799382703429783,
      "learning_rate": 1.3673731039107726e-06,
      "loss": 0.5305,
      "step": 16131
    },
    {
      "epoch": 0.8705412551939993,
      "grad_norm": 1.1577623643432198,
      "learning_rate": 1.367071254215845e-06,
      "loss": 0.4998,
      "step": 16132
    },
    {
      "epoch": 0.8705952188225137,
      "grad_norm": 0.990625915361084,
      "learning_rate": 1.3667695233054651e-06,
      "loss": 0.3781,
      "step": 16133
    },
    {
      "epoch": 0.870649182451028,
      "grad_norm": 0.9940018695270442,
      "learning_rate": 1.3664679111883068e-06,
      "loss": 0.3528,
      "step": 16134
    },
    {
      "epoch": 0.8707031460795424,
      "grad_norm": 1.0279633049485066,
      "learning_rate": 1.366166417873036e-06,
      "loss": 0.4739,
      "step": 16135
    },
    {
      "epoch": 0.8707571097080568,
      "grad_norm": 1.0916213917274067,
      "learning_rate": 1.3658650433683198e-06,
      "loss": 0.5409,
      "step": 16136
    },
    {
      "epoch": 0.8708110733365712,
      "grad_norm": 1.1360261833068148,
      "learning_rate": 1.3655637876828192e-06,
      "loss": 0.4516,
      "step": 16137
    },
    {
      "epoch": 0.8708650369650855,
      "grad_norm": 1.111808575701815,
      "learning_rate": 1.365262650825193e-06,
      "loss": 0.5902,
      "step": 16138
    },
    {
      "epoch": 0.8709190005936,
      "grad_norm": 1.2311601385314508,
      "learning_rate": 1.364961632804096e-06,
      "loss": 0.4378,
      "step": 16139
    },
    {
      "epoch": 0.8709729642221143,
      "grad_norm": 0.8985519086167008,
      "learning_rate": 1.3646607336281798e-06,
      "loss": 0.4333,
      "step": 16140
    },
    {
      "epoch": 0.8710269278506286,
      "grad_norm": 0.7758171060322469,
      "learning_rate": 1.3643599533060933e-06,
      "loss": 0.4019,
      "step": 16141
    },
    {
      "epoch": 0.871080891479143,
      "grad_norm": 0.9402180790839404,
      "learning_rate": 1.3640592918464799e-06,
      "loss": 0.3713,
      "step": 16142
    },
    {
      "epoch": 0.8711348551076574,
      "grad_norm": 1.0241980903106382,
      "learning_rate": 1.3637587492579812e-06,
      "loss": 0.4631,
      "step": 16143
    },
    {
      "epoch": 0.8711888187361718,
      "grad_norm": 0.8692352588090932,
      "learning_rate": 1.3634583255492357e-06,
      "loss": 0.3084,
      "step": 16144
    },
    {
      "epoch": 0.8712427823646862,
      "grad_norm": 0.9208504216967156,
      "learning_rate": 1.3631580207288777e-06,
      "loss": 0.4352,
      "step": 16145
    },
    {
      "epoch": 0.8712967459932006,
      "grad_norm": 1.1170767535255977,
      "learning_rate": 1.3628578348055377e-06,
      "loss": 0.5356,
      "step": 16146
    },
    {
      "epoch": 0.871350709621715,
      "grad_norm": 1.0298898966486347,
      "learning_rate": 1.3625577677878443e-06,
      "loss": 0.531,
      "step": 16147
    },
    {
      "epoch": 0.8714046732502293,
      "grad_norm": 0.9957206196275015,
      "learning_rate": 1.3622578196844215e-06,
      "loss": 0.4559,
      "step": 16148
    },
    {
      "epoch": 0.8714586368787437,
      "grad_norm": 1.0992135603182238,
      "learning_rate": 1.3619579905038892e-06,
      "loss": 0.4101,
      "step": 16149
    },
    {
      "epoch": 0.8715126005072581,
      "grad_norm": 1.0894893584160243,
      "learning_rate": 1.3616582802548662e-06,
      "loss": 0.3852,
      "step": 16150
    },
    {
      "epoch": 0.8715665641357725,
      "grad_norm": 0.9914274292165268,
      "learning_rate": 1.3613586889459646e-06,
      "loss": 0.4456,
      "step": 16151
    },
    {
      "epoch": 0.8716205277642869,
      "grad_norm": 0.9345336537946539,
      "learning_rate": 1.361059216585797e-06,
      "loss": 0.4005,
      "step": 16152
    },
    {
      "epoch": 0.8716744913928013,
      "grad_norm": 1.1244778765597754,
      "learning_rate": 1.3607598631829688e-06,
      "loss": 0.4472,
      "step": 16153
    },
    {
      "epoch": 0.8717284550213157,
      "grad_norm": 0.9721936954906572,
      "learning_rate": 1.3604606287460853e-06,
      "loss": 0.4495,
      "step": 16154
    },
    {
      "epoch": 0.87178241864983,
      "grad_norm": 0.9135220477467266,
      "learning_rate": 1.3601615132837454e-06,
      "loss": 0.3482,
      "step": 16155
    },
    {
      "epoch": 0.8718363822783444,
      "grad_norm": 1.200500867553237,
      "learning_rate": 1.3598625168045467e-06,
      "loss": 0.6387,
      "step": 16156
    },
    {
      "epoch": 0.8718903459068588,
      "grad_norm": 1.086063182182612,
      "learning_rate": 1.3595636393170828e-06,
      "loss": 0.4782,
      "step": 16157
    },
    {
      "epoch": 0.8719443095353732,
      "grad_norm": 0.8378772527067964,
      "learning_rate": 1.3592648808299433e-06,
      "loss": 0.3562,
      "step": 16158
    },
    {
      "epoch": 0.8719982731638876,
      "grad_norm": 0.9614197833527857,
      "learning_rate": 1.3589662413517163e-06,
      "loss": 0.42,
      "step": 16159
    },
    {
      "epoch": 0.872052236792402,
      "grad_norm": 0.9465480557806304,
      "learning_rate": 1.3586677208909825e-06,
      "loss": 0.4747,
      "step": 16160
    },
    {
      "epoch": 0.8721062004209164,
      "grad_norm": 1.3492633965183105,
      "learning_rate": 1.358369319456323e-06,
      "loss": 0.5267,
      "step": 16161
    },
    {
      "epoch": 0.8721601640494306,
      "grad_norm": 1.1737112770384424,
      "learning_rate": 1.3580710370563142e-06,
      "loss": 0.4838,
      "step": 16162
    },
    {
      "epoch": 0.872214127677945,
      "grad_norm": 0.9885937936727277,
      "learning_rate": 1.3577728736995288e-06,
      "loss": 0.4485,
      "step": 16163
    },
    {
      "epoch": 0.8722680913064594,
      "grad_norm": 0.9841119724552323,
      "learning_rate": 1.357474829394536e-06,
      "loss": 0.4902,
      "step": 16164
    },
    {
      "epoch": 0.8723220549349738,
      "grad_norm": 0.8847402583752003,
      "learning_rate": 1.3571769041499034e-06,
      "loss": 0.3662,
      "step": 16165
    },
    {
      "epoch": 0.8723760185634882,
      "grad_norm": 1.1051869415893845,
      "learning_rate": 1.3568790979741914e-06,
      "loss": 0.4421,
      "step": 16166
    },
    {
      "epoch": 0.8724299821920026,
      "grad_norm": 1.0235224503654698,
      "learning_rate": 1.3565814108759612e-06,
      "loss": 0.4538,
      "step": 16167
    },
    {
      "epoch": 0.872483945820517,
      "grad_norm": 0.8767621757428213,
      "learning_rate": 1.3562838428637679e-06,
      "loss": 0.4266,
      "step": 16168
    },
    {
      "epoch": 0.8725379094490313,
      "grad_norm": 0.9631959669288327,
      "learning_rate": 1.3559863939461643e-06,
      "loss": 0.4322,
      "step": 16169
    },
    {
      "epoch": 0.8725918730775457,
      "grad_norm": 1.019213285185678,
      "learning_rate": 1.3556890641316983e-06,
      "loss": 0.6214,
      "step": 16170
    },
    {
      "epoch": 0.8726458367060601,
      "grad_norm": 1.3264550709130312,
      "learning_rate": 1.3553918534289163e-06,
      "loss": 0.539,
      "step": 16171
    },
    {
      "epoch": 0.8726998003345745,
      "grad_norm": 1.127972437255977,
      "learning_rate": 1.35509476184636e-06,
      "loss": 0.4962,
      "step": 16172
    },
    {
      "epoch": 0.8727537639630889,
      "grad_norm": 0.9157499315450833,
      "learning_rate": 1.3547977893925685e-06,
      "loss": 0.5323,
      "step": 16173
    },
    {
      "epoch": 0.8728077275916033,
      "grad_norm": 0.9339540478830759,
      "learning_rate": 1.3545009360760766e-06,
      "loss": 0.3551,
      "step": 16174
    },
    {
      "epoch": 0.8728616912201176,
      "grad_norm": 1.003224520860702,
      "learning_rate": 1.3542042019054174e-06,
      "loss": 0.3907,
      "step": 16175
    },
    {
      "epoch": 0.872915654848632,
      "grad_norm": 0.9016734753049422,
      "learning_rate": 1.353907586889118e-06,
      "loss": 0.5232,
      "step": 16176
    },
    {
      "epoch": 0.8729696184771464,
      "grad_norm": 0.9589432607051979,
      "learning_rate": 1.3536110910357039e-06,
      "loss": 0.3985,
      "step": 16177
    },
    {
      "epoch": 0.8730235821056608,
      "grad_norm": 0.9519870917865427,
      "learning_rate": 1.3533147143536964e-06,
      "loss": 0.4161,
      "step": 16178
    },
    {
      "epoch": 0.8730775457341752,
      "grad_norm": 1.1021329574623013,
      "learning_rate": 1.3530184568516142e-06,
      "loss": 0.3881,
      "step": 16179
    },
    {
      "epoch": 0.8731315093626896,
      "grad_norm": 1.115403530729816,
      "learning_rate": 1.3527223185379714e-06,
      "loss": 0.4815,
      "step": 16180
    },
    {
      "epoch": 0.873185472991204,
      "grad_norm": 0.9078095454632906,
      "learning_rate": 1.35242629942128e-06,
      "loss": 0.3796,
      "step": 16181
    },
    {
      "epoch": 0.8732394366197183,
      "grad_norm": 1.0095614864348466,
      "learning_rate": 1.3521303995100479e-06,
      "loss": 0.3688,
      "step": 16182
    },
    {
      "epoch": 0.8732934002482327,
      "grad_norm": 1.0237823280447695,
      "learning_rate": 1.3518346188127786e-06,
      "loss": 0.3799,
      "step": 16183
    },
    {
      "epoch": 0.873347363876747,
      "grad_norm": 1.25877722748606,
      "learning_rate": 1.3515389573379745e-06,
      "loss": 0.6,
      "step": 16184
    },
    {
      "epoch": 0.8734013275052614,
      "grad_norm": 0.9682231216834499,
      "learning_rate": 1.3512434150941317e-06,
      "loss": 0.5315,
      "step": 16185
    },
    {
      "epoch": 0.8734552911337758,
      "grad_norm": 0.9429742196446539,
      "learning_rate": 1.3509479920897456e-06,
      "loss": 0.4265,
      "step": 16186
    },
    {
      "epoch": 0.8735092547622902,
      "grad_norm": 0.9554849076351533,
      "learning_rate": 1.3506526883333064e-06,
      "loss": 0.3468,
      "step": 16187
    },
    {
      "epoch": 0.8735632183908046,
      "grad_norm": 0.9366176276202473,
      "learning_rate": 1.3503575038333012e-06,
      "loss": 0.3759,
      "step": 16188
    },
    {
      "epoch": 0.8736171820193189,
      "grad_norm": 0.9202095223270981,
      "learning_rate": 1.3500624385982152e-06,
      "loss": 0.3038,
      "step": 16189
    },
    {
      "epoch": 0.8736711456478333,
      "grad_norm": 1.0565379545044808,
      "learning_rate": 1.349767492636527e-06,
      "loss": 0.403,
      "step": 16190
    },
    {
      "epoch": 0.8737251092763477,
      "grad_norm": 0.9597680428527434,
      "learning_rate": 1.3494726659567148e-06,
      "loss": 0.3364,
      "step": 16191
    },
    {
      "epoch": 0.8737790729048621,
      "grad_norm": 1.0424276035307118,
      "learning_rate": 1.349177958567252e-06,
      "loss": 0.3094,
      "step": 16192
    },
    {
      "epoch": 0.8738330365333765,
      "grad_norm": 1.184409482562362,
      "learning_rate": 1.3488833704766089e-06,
      "loss": 0.728,
      "step": 16193
    },
    {
      "epoch": 0.8738870001618909,
      "grad_norm": 0.9233311877398541,
      "learning_rate": 1.3485889016932522e-06,
      "loss": 0.3386,
      "step": 16194
    },
    {
      "epoch": 0.8739409637904053,
      "grad_norm": 1.145729624079814,
      "learning_rate": 1.3482945522256447e-06,
      "loss": 0.5515,
      "step": 16195
    },
    {
      "epoch": 0.8739949274189196,
      "grad_norm": 1.0944963804644126,
      "learning_rate": 1.3480003220822477e-06,
      "loss": 0.5841,
      "step": 16196
    },
    {
      "epoch": 0.874048891047434,
      "grad_norm": 1.0295699846959496,
      "learning_rate": 1.3477062112715162e-06,
      "loss": 0.4498,
      "step": 16197
    },
    {
      "epoch": 0.8741028546759484,
      "grad_norm": 0.9087576121256383,
      "learning_rate": 1.3474122198019038e-06,
      "loss": 0.462,
      "step": 16198
    },
    {
      "epoch": 0.8741568183044628,
      "grad_norm": 1.230293339728579,
      "learning_rate": 1.3471183476818598e-06,
      "loss": 0.558,
      "step": 16199
    },
    {
      "epoch": 0.8742107819329772,
      "grad_norm": 1.2086249145093795,
      "learning_rate": 1.3468245949198308e-06,
      "loss": 0.5747,
      "step": 16200
    },
    {
      "epoch": 0.8742647455614916,
      "grad_norm": 1.0783169231626186,
      "learning_rate": 1.3465309615242589e-06,
      "loss": 0.4034,
      "step": 16201
    },
    {
      "epoch": 0.874318709190006,
      "grad_norm": 1.0019962311657518,
      "learning_rate": 1.3462374475035845e-06,
      "loss": 0.3985,
      "step": 16202
    },
    {
      "epoch": 0.8743726728185203,
      "grad_norm": 1.093208681886094,
      "learning_rate": 1.3459440528662441e-06,
      "loss": 0.4938,
      "step": 16203
    },
    {
      "epoch": 0.8744266364470347,
      "grad_norm": 1.2193558521799472,
      "learning_rate": 1.3456507776206674e-06,
      "loss": 0.4969,
      "step": 16204
    },
    {
      "epoch": 0.8744806000755491,
      "grad_norm": 1.071824311393677,
      "learning_rate": 1.3453576217752854e-06,
      "loss": 0.493,
      "step": 16205
    },
    {
      "epoch": 0.8745345637040635,
      "grad_norm": 1.090456470598965,
      "learning_rate": 1.3450645853385235e-06,
      "loss": 0.5096,
      "step": 16206
    },
    {
      "epoch": 0.8745885273325779,
      "grad_norm": 1.1684221708857356,
      "learning_rate": 1.3447716683188032e-06,
      "loss": 0.5251,
      "step": 16207
    },
    {
      "epoch": 0.8746424909610923,
      "grad_norm": 0.976421832261836,
      "learning_rate": 1.344478870724544e-06,
      "loss": 0.4241,
      "step": 16208
    },
    {
      "epoch": 0.8746964545896067,
      "grad_norm": 0.8095598871256602,
      "learning_rate": 1.3441861925641609e-06,
      "loss": 0.2806,
      "step": 16209
    },
    {
      "epoch": 0.8747504182181209,
      "grad_norm": 0.828570874100992,
      "learning_rate": 1.3438936338460653e-06,
      "loss": 0.3215,
      "step": 16210
    },
    {
      "epoch": 0.8748043818466353,
      "grad_norm": 1.0357945673540385,
      "learning_rate": 1.3436011945786665e-06,
      "loss": 0.492,
      "step": 16211
    },
    {
      "epoch": 0.8748583454751497,
      "grad_norm": 0.971975235569279,
      "learning_rate": 1.343308874770369e-06,
      "loss": 0.4122,
      "step": 16212
    },
    {
      "epoch": 0.8749123091036641,
      "grad_norm": 0.8686189719093521,
      "learning_rate": 1.3430166744295753e-06,
      "loss": 0.2868,
      "step": 16213
    },
    {
      "epoch": 0.8749662727321785,
      "grad_norm": 0.9193858394409322,
      "learning_rate": 1.3427245935646813e-06,
      "loss": 0.5665,
      "step": 16214
    },
    {
      "epoch": 0.8750202363606929,
      "grad_norm": 1.0279637718014387,
      "learning_rate": 1.3424326321840842e-06,
      "loss": 0.4225,
      "step": 16215
    },
    {
      "epoch": 0.8750741999892073,
      "grad_norm": 1.2095369099495306,
      "learning_rate": 1.3421407902961733e-06,
      "loss": 0.631,
      "step": 16216
    },
    {
      "epoch": 0.8751281636177216,
      "grad_norm": 1.1051721292706589,
      "learning_rate": 1.3418490679093388e-06,
      "loss": 0.6027,
      "step": 16217
    },
    {
      "epoch": 0.875182127246236,
      "grad_norm": 1.3806581313165702,
      "learning_rate": 1.3415574650319628e-06,
      "loss": 0.5689,
      "step": 16218
    },
    {
      "epoch": 0.8752360908747504,
      "grad_norm": 1.018399059119866,
      "learning_rate": 1.3412659816724266e-06,
      "loss": 0.4072,
      "step": 16219
    },
    {
      "epoch": 0.8752900545032648,
      "grad_norm": 1.1273643154385393,
      "learning_rate": 1.3409746178391084e-06,
      "loss": 0.4681,
      "step": 16220
    },
    {
      "epoch": 0.8753440181317792,
      "grad_norm": 1.1111287148987132,
      "learning_rate": 1.3406833735403819e-06,
      "loss": 0.503,
      "step": 16221
    },
    {
      "epoch": 0.8753979817602936,
      "grad_norm": 1.1810402487094827,
      "learning_rate": 1.340392248784618e-06,
      "loss": 0.2904,
      "step": 16222
    },
    {
      "epoch": 0.875451945388808,
      "grad_norm": 0.825549075306673,
      "learning_rate": 1.3401012435801837e-06,
      "loss": 0.43,
      "step": 16223
    },
    {
      "epoch": 0.8755059090173223,
      "grad_norm": 0.989321325589098,
      "learning_rate": 1.3398103579354444e-06,
      "loss": 0.412,
      "step": 16224
    },
    {
      "epoch": 0.8755598726458367,
      "grad_norm": 0.9825084829744649,
      "learning_rate": 1.3395195918587574e-06,
      "loss": 0.4834,
      "step": 16225
    },
    {
      "epoch": 0.8756138362743511,
      "grad_norm": 1.1180643560626067,
      "learning_rate": 1.3392289453584816e-06,
      "loss": 0.5126,
      "step": 16226
    },
    {
      "epoch": 0.8756677999028655,
      "grad_norm": 1.0700545515820141,
      "learning_rate": 1.33893841844297e-06,
      "loss": 0.5359,
      "step": 16227
    },
    {
      "epoch": 0.8757217635313799,
      "grad_norm": 0.8962622859253799,
      "learning_rate": 1.338648011120573e-06,
      "loss": 0.3817,
      "step": 16228
    },
    {
      "epoch": 0.8757757271598943,
      "grad_norm": 1.0494403783418564,
      "learning_rate": 1.3383577233996365e-06,
      "loss": 0.4043,
      "step": 16229
    },
    {
      "epoch": 0.8758296907884087,
      "grad_norm": 0.9641242982344483,
      "learning_rate": 1.338067555288504e-06,
      "loss": 0.4663,
      "step": 16230
    },
    {
      "epoch": 0.875883654416923,
      "grad_norm": 1.0591959959224793,
      "learning_rate": 1.3377775067955158e-06,
      "loss": 0.4235,
      "step": 16231
    },
    {
      "epoch": 0.8759376180454373,
      "grad_norm": 0.8492782775488025,
      "learning_rate": 1.3374875779290076e-06,
      "loss": 0.3581,
      "step": 16232
    },
    {
      "epoch": 0.8759915816739517,
      "grad_norm": 0.8455419438000532,
      "learning_rate": 1.337197768697312e-06,
      "loss": 0.3399,
      "step": 16233
    },
    {
      "epoch": 0.8760455453024661,
      "grad_norm": 0.8886929873726982,
      "learning_rate": 1.336908079108759e-06,
      "loss": 0.4403,
      "step": 16234
    },
    {
      "epoch": 0.8760995089309805,
      "grad_norm": 0.9418267707651176,
      "learning_rate": 1.3366185091716738e-06,
      "loss": 0.372,
      "step": 16235
    },
    {
      "epoch": 0.8761534725594949,
      "grad_norm": 1.0365418339388308,
      "learning_rate": 1.33632905889438e-06,
      "loss": 0.4273,
      "step": 16236
    },
    {
      "epoch": 0.8762074361880093,
      "grad_norm": 1.1473859308375958,
      "learning_rate": 1.336039728285196e-06,
      "loss": 0.3898,
      "step": 16237
    },
    {
      "epoch": 0.8762613998165236,
      "grad_norm": 1.1139524906931684,
      "learning_rate": 1.3357505173524379e-06,
      "loss": 0.552,
      "step": 16238
    },
    {
      "epoch": 0.876315363445038,
      "grad_norm": 1.0690433025245032,
      "learning_rate": 1.3354614261044171e-06,
      "loss": 0.4938,
      "step": 16239
    },
    {
      "epoch": 0.8763693270735524,
      "grad_norm": 1.0361623322611004,
      "learning_rate": 1.335172454549443e-06,
      "loss": 0.3914,
      "step": 16240
    },
    {
      "epoch": 0.8764232907020668,
      "grad_norm": 1.324977538579406,
      "learning_rate": 1.3348836026958206e-06,
      "loss": 0.5913,
      "step": 16241
    },
    {
      "epoch": 0.8764772543305812,
      "grad_norm": 1.139345839799654,
      "learning_rate": 1.3345948705518518e-06,
      "loss": 0.5611,
      "step": 16242
    },
    {
      "epoch": 0.8765312179590956,
      "grad_norm": 0.8520748806038332,
      "learning_rate": 1.3343062581258356e-06,
      "loss": 0.3411,
      "step": 16243
    },
    {
      "epoch": 0.8765851815876099,
      "grad_norm": 1.0429905459336324,
      "learning_rate": 1.3340177654260671e-06,
      "loss": 0.4028,
      "step": 16244
    },
    {
      "epoch": 0.8766391452161243,
      "grad_norm": 1.0024855965241013,
      "learning_rate": 1.3337293924608375e-06,
      "loss": 0.4284,
      "step": 16245
    },
    {
      "epoch": 0.8766931088446387,
      "grad_norm": 0.954267327529957,
      "learning_rate": 1.333441139238434e-06,
      "loss": 0.4437,
      "step": 16246
    },
    {
      "epoch": 0.8767470724731531,
      "grad_norm": 0.9909663586897963,
      "learning_rate": 1.3331530057671426e-06,
      "loss": 0.4253,
      "step": 16247
    },
    {
      "epoch": 0.8768010361016675,
      "grad_norm": 1.056526542700705,
      "learning_rate": 1.3328649920552442e-06,
      "loss": 0.4754,
      "step": 16248
    },
    {
      "epoch": 0.8768549997301819,
      "grad_norm": 1.228122553137251,
      "learning_rate": 1.332577098111016e-06,
      "loss": 0.5173,
      "step": 16249
    },
    {
      "epoch": 0.8769089633586963,
      "grad_norm": 1.1959084745564956,
      "learning_rate": 1.3322893239427331e-06,
      "loss": 0.4654,
      "step": 16250
    },
    {
      "epoch": 0.8769629269872106,
      "grad_norm": 0.8081133166010039,
      "learning_rate": 1.3320016695586673e-06,
      "loss": 0.2631,
      "step": 16251
    },
    {
      "epoch": 0.877016890615725,
      "grad_norm": 1.2128518778678408,
      "learning_rate": 1.3317141349670837e-06,
      "loss": 0.4838,
      "step": 16252
    },
    {
      "epoch": 0.8770708542442394,
      "grad_norm": 1.0326010209938243,
      "learning_rate": 1.331426720176248e-06,
      "loss": 0.4467,
      "step": 16253
    },
    {
      "epoch": 0.8771248178727538,
      "grad_norm": 0.9669595460962193,
      "learning_rate": 1.3311394251944201e-06,
      "loss": 0.4061,
      "step": 16254
    },
    {
      "epoch": 0.8771787815012682,
      "grad_norm": 1.151576126519565,
      "learning_rate": 1.330852250029858e-06,
      "loss": 0.5281,
      "step": 16255
    },
    {
      "epoch": 0.8772327451297826,
      "grad_norm": 1.4582218738483015,
      "learning_rate": 1.330565194690814e-06,
      "loss": 0.4123,
      "step": 16256
    },
    {
      "epoch": 0.877286708758297,
      "grad_norm": 0.7612206730464174,
      "learning_rate": 1.3302782591855396e-06,
      "loss": 0.2508,
      "step": 16257
    },
    {
      "epoch": 0.8773406723868112,
      "grad_norm": 1.1365948333274234,
      "learning_rate": 1.3299914435222815e-06,
      "loss": 0.5228,
      "step": 16258
    },
    {
      "epoch": 0.8773946360153256,
      "grad_norm": 1.0988728260710177,
      "learning_rate": 1.3297047477092826e-06,
      "loss": 0.4832,
      "step": 16259
    },
    {
      "epoch": 0.87744859964384,
      "grad_norm": 0.9161579948645353,
      "learning_rate": 1.3294181717547827e-06,
      "loss": 0.3357,
      "step": 16260
    },
    {
      "epoch": 0.8775025632723544,
      "grad_norm": 1.044317377943981,
      "learning_rate": 1.3291317156670192e-06,
      "loss": 0.6142,
      "step": 16261
    },
    {
      "epoch": 0.8775565269008688,
      "grad_norm": 1.0141679052160875,
      "learning_rate": 1.3288453794542244e-06,
      "loss": 0.3848,
      "step": 16262
    },
    {
      "epoch": 0.8776104905293832,
      "grad_norm": 1.1473738622462137,
      "learning_rate": 1.3285591631246278e-06,
      "loss": 0.4666,
      "step": 16263
    },
    {
      "epoch": 0.8776644541578976,
      "grad_norm": 0.8929392251063433,
      "learning_rate": 1.3282730666864558e-06,
      "loss": 0.4781,
      "step": 16264
    },
    {
      "epoch": 0.8777184177864119,
      "grad_norm": 1.0910401444815376,
      "learning_rate": 1.327987090147932e-06,
      "loss": 0.5229,
      "step": 16265
    },
    {
      "epoch": 0.8777723814149263,
      "grad_norm": 1.1430890082677725,
      "learning_rate": 1.327701233517274e-06,
      "loss": 0.5055,
      "step": 16266
    },
    {
      "epoch": 0.8778263450434407,
      "grad_norm": 0.8423276584927167,
      "learning_rate": 1.3274154968026978e-06,
      "loss": 0.3907,
      "step": 16267
    },
    {
      "epoch": 0.8778803086719551,
      "grad_norm": 0.8797936187027491,
      "learning_rate": 1.3271298800124172e-06,
      "loss": 0.4005,
      "step": 16268
    },
    {
      "epoch": 0.8779342723004695,
      "grad_norm": 0.9952257154881801,
      "learning_rate": 1.32684438315464e-06,
      "loss": 0.49,
      "step": 16269
    },
    {
      "epoch": 0.8779882359289839,
      "grad_norm": 1.1519047074517272,
      "learning_rate": 1.3265590062375712e-06,
      "loss": 0.4759,
      "step": 16270
    },
    {
      "epoch": 0.8780421995574983,
      "grad_norm": 0.9495959789456008,
      "learning_rate": 1.3262737492694144e-06,
      "loss": 0.4568,
      "step": 16271
    },
    {
      "epoch": 0.8780961631860126,
      "grad_norm": 1.2718010313498125,
      "learning_rate": 1.3259886122583674e-06,
      "loss": 0.6703,
      "step": 16272
    },
    {
      "epoch": 0.878150126814527,
      "grad_norm": 1.1281112484178857,
      "learning_rate": 1.3257035952126246e-06,
      "loss": 0.4407,
      "step": 16273
    },
    {
      "epoch": 0.8782040904430414,
      "grad_norm": 1.163730690891375,
      "learning_rate": 1.3254186981403789e-06,
      "loss": 0.5563,
      "step": 16274
    },
    {
      "epoch": 0.8782580540715558,
      "grad_norm": 0.8861846476819842,
      "learning_rate": 1.3251339210498171e-06,
      "loss": 0.3662,
      "step": 16275
    },
    {
      "epoch": 0.8783120177000702,
      "grad_norm": 1.1940297873957384,
      "learning_rate": 1.3248492639491249e-06,
      "loss": 0.5453,
      "step": 16276
    },
    {
      "epoch": 0.8783659813285846,
      "grad_norm": 1.0393743275512835,
      "learning_rate": 1.324564726846484e-06,
      "loss": 0.4336,
      "step": 16277
    },
    {
      "epoch": 0.878419944957099,
      "grad_norm": 0.9736588115283616,
      "learning_rate": 1.3242803097500714e-06,
      "loss": 0.5693,
      "step": 16278
    },
    {
      "epoch": 0.8784739085856133,
      "grad_norm": 1.054150887677765,
      "learning_rate": 1.3239960126680628e-06,
      "loss": 0.4937,
      "step": 16279
    },
    {
      "epoch": 0.8785278722141276,
      "grad_norm": 1.0605386710162192,
      "learning_rate": 1.3237118356086277e-06,
      "loss": 0.5145,
      "step": 16280
    },
    {
      "epoch": 0.878581835842642,
      "grad_norm": 1.084493336948569,
      "learning_rate": 1.3234277785799343e-06,
      "loss": 0.458,
      "step": 16281
    },
    {
      "epoch": 0.8786357994711564,
      "grad_norm": 1.1238115004553337,
      "learning_rate": 1.323143841590147e-06,
      "loss": 0.4468,
      "step": 16282
    },
    {
      "epoch": 0.8786897630996708,
      "grad_norm": 1.005988046506106,
      "learning_rate": 1.3228600246474255e-06,
      "loss": 0.4485,
      "step": 16283
    },
    {
      "epoch": 0.8787437267281852,
      "grad_norm": 0.9884198148694481,
      "learning_rate": 1.3225763277599277e-06,
      "loss": 0.402,
      "step": 16284
    },
    {
      "epoch": 0.8787976903566996,
      "grad_norm": 1.2182579825488973,
      "learning_rate": 1.3222927509358075e-06,
      "loss": 0.5584,
      "step": 16285
    },
    {
      "epoch": 0.8788516539852139,
      "grad_norm": 0.9642691254251994,
      "learning_rate": 1.3220092941832154e-06,
      "loss": 0.4548,
      "step": 16286
    },
    {
      "epoch": 0.8789056176137283,
      "grad_norm": 1.0118242796238452,
      "learning_rate": 1.321725957510297e-06,
      "loss": 0.5285,
      "step": 16287
    },
    {
      "epoch": 0.8789595812422427,
      "grad_norm": 0.9921622604962324,
      "learning_rate": 1.3214427409251965e-06,
      "loss": 0.4938,
      "step": 16288
    },
    {
      "epoch": 0.8790135448707571,
      "grad_norm": 1.162160560333159,
      "learning_rate": 1.3211596444360535e-06,
      "loss": 0.5148,
      "step": 16289
    },
    {
      "epoch": 0.8790675084992715,
      "grad_norm": 1.0728172272259302,
      "learning_rate": 1.320876668051005e-06,
      "loss": 0.4141,
      "step": 16290
    },
    {
      "epoch": 0.8791214721277859,
      "grad_norm": 1.002430518001186,
      "learning_rate": 1.3205938117781836e-06,
      "loss": 0.431,
      "step": 16291
    },
    {
      "epoch": 0.8791754357563003,
      "grad_norm": 1.020604290787138,
      "learning_rate": 1.3203110756257193e-06,
      "loss": 0.4525,
      "step": 16292
    },
    {
      "epoch": 0.8792293993848146,
      "grad_norm": 1.2584930814628903,
      "learning_rate": 1.3200284596017384e-06,
      "loss": 0.3627,
      "step": 16293
    },
    {
      "epoch": 0.879283363013329,
      "grad_norm": 0.8189246848053718,
      "learning_rate": 1.3197459637143622e-06,
      "loss": 0.3195,
      "step": 16294
    },
    {
      "epoch": 0.8793373266418434,
      "grad_norm": 0.8684090660133488,
      "learning_rate": 1.3194635879717115e-06,
      "loss": 0.3873,
      "step": 16295
    },
    {
      "epoch": 0.8793912902703578,
      "grad_norm": 0.9943214435637469,
      "learning_rate": 1.3191813323819012e-06,
      "loss": 0.4067,
      "step": 16296
    },
    {
      "epoch": 0.8794452538988722,
      "grad_norm": 1.0003310440870399,
      "learning_rate": 1.3188991969530434e-06,
      "loss": 0.3536,
      "step": 16297
    },
    {
      "epoch": 0.8794992175273866,
      "grad_norm": 1.1062223503083537,
      "learning_rate": 1.3186171816932477e-06,
      "loss": 0.4716,
      "step": 16298
    },
    {
      "epoch": 0.879553181155901,
      "grad_norm": 1.051498998880181,
      "learning_rate": 1.318335286610619e-06,
      "loss": 0.4848,
      "step": 16299
    },
    {
      "epoch": 0.8796071447844153,
      "grad_norm": 1.1142429870573873,
      "learning_rate": 1.3180535117132606e-06,
      "loss": 0.4206,
      "step": 16300
    },
    {
      "epoch": 0.8796611084129297,
      "grad_norm": 1.092564932281193,
      "learning_rate": 1.3177718570092684e-06,
      "loss": 0.4857,
      "step": 16301
    },
    {
      "epoch": 0.8797150720414441,
      "grad_norm": 1.0662420595925992,
      "learning_rate": 1.3174903225067395e-06,
      "loss": 0.4736,
      "step": 16302
    },
    {
      "epoch": 0.8797690356699585,
      "grad_norm": 1.0900640418566727,
      "learning_rate": 1.3172089082137646e-06,
      "loss": 0.3477,
      "step": 16303
    },
    {
      "epoch": 0.8798229992984729,
      "grad_norm": 1.0214498841325603,
      "learning_rate": 1.3169276141384323e-06,
      "loss": 0.4155,
      "step": 16304
    },
    {
      "epoch": 0.8798769629269872,
      "grad_norm": 1.0322377284065911,
      "learning_rate": 1.3166464402888265e-06,
      "loss": 0.4335,
      "step": 16305
    },
    {
      "epoch": 0.8799309265555016,
      "grad_norm": 0.8937511083639723,
      "learning_rate": 1.3163653866730295e-06,
      "loss": 0.3172,
      "step": 16306
    },
    {
      "epoch": 0.8799848901840159,
      "grad_norm": 0.9090745380658347,
      "learning_rate": 1.3160844532991185e-06,
      "loss": 0.3075,
      "step": 16307
    },
    {
      "epoch": 0.8800388538125303,
      "grad_norm": 0.9560315617941141,
      "learning_rate": 1.3158036401751675e-06,
      "loss": 0.425,
      "step": 16308
    },
    {
      "epoch": 0.8800928174410447,
      "grad_norm": 1.0274506953536002,
      "learning_rate": 1.3155229473092484e-06,
      "loss": 0.3914,
      "step": 16309
    },
    {
      "epoch": 0.8801467810695591,
      "grad_norm": 0.9764322133841534,
      "learning_rate": 1.3152423747094275e-06,
      "loss": 0.4817,
      "step": 16310
    },
    {
      "epoch": 0.8802007446980735,
      "grad_norm": 1.2006463471050075,
      "learning_rate": 1.3149619223837689e-06,
      "loss": 0.5441,
      "step": 16311
    },
    {
      "epoch": 0.8802547083265879,
      "grad_norm": 0.8727541896273092,
      "learning_rate": 1.3146815903403336e-06,
      "loss": 0.4283,
      "step": 16312
    },
    {
      "epoch": 0.8803086719551022,
      "grad_norm": 1.1975838672454306,
      "learning_rate": 1.3144013785871785e-06,
      "loss": 0.4389,
      "step": 16313
    },
    {
      "epoch": 0.8803626355836166,
      "grad_norm": 1.1387391331738848,
      "learning_rate": 1.3141212871323564e-06,
      "loss": 0.53,
      "step": 16314
    },
    {
      "epoch": 0.880416599212131,
      "grad_norm": 1.2963482341722576,
      "learning_rate": 1.3138413159839186e-06,
      "loss": 0.5771,
      "step": 16315
    },
    {
      "epoch": 0.8804705628406454,
      "grad_norm": 1.2321489481052343,
      "learning_rate": 1.313561465149911e-06,
      "loss": 0.5682,
      "step": 16316
    },
    {
      "epoch": 0.8805245264691598,
      "grad_norm": 1.0115514473672362,
      "learning_rate": 1.3132817346383768e-06,
      "loss": 0.4291,
      "step": 16317
    },
    {
      "epoch": 0.8805784900976742,
      "grad_norm": 0.9695907867074682,
      "learning_rate": 1.313002124457356e-06,
      "loss": 0.4329,
      "step": 16318
    },
    {
      "epoch": 0.8806324537261886,
      "grad_norm": 0.9451894949924334,
      "learning_rate": 1.3127226346148846e-06,
      "loss": 0.4107,
      "step": 16319
    },
    {
      "epoch": 0.8806864173547029,
      "grad_norm": 0.8122714306373786,
      "learning_rate": 1.312443265118997e-06,
      "loss": 0.3706,
      "step": 16320
    },
    {
      "epoch": 0.8807403809832173,
      "grad_norm": 0.9802070734311316,
      "learning_rate": 1.3121640159777197e-06,
      "loss": 0.5194,
      "step": 16321
    },
    {
      "epoch": 0.8807943446117317,
      "grad_norm": 1.0021207485052832,
      "learning_rate": 1.3118848871990805e-06,
      "loss": 0.4204,
      "step": 16322
    },
    {
      "epoch": 0.8808483082402461,
      "grad_norm": 0.7861167473018752,
      "learning_rate": 1.311605878791101e-06,
      "loss": 0.2756,
      "step": 16323
    },
    {
      "epoch": 0.8809022718687605,
      "grad_norm": 1.1705889369574873,
      "learning_rate": 1.3113269907618012e-06,
      "loss": 0.4848,
      "step": 16324
    },
    {
      "epoch": 0.8809562354972749,
      "grad_norm": 1.0345147934271375,
      "learning_rate": 1.3110482231191957e-06,
      "loss": 0.4206,
      "step": 16325
    },
    {
      "epoch": 0.8810101991257893,
      "grad_norm": 1.2129802832634726,
      "learning_rate": 1.3107695758712971e-06,
      "loss": 0.57,
      "step": 16326
    },
    {
      "epoch": 0.8810641627543035,
      "grad_norm": 1.038242858607382,
      "learning_rate": 1.3104910490261145e-06,
      "loss": 0.412,
      "step": 16327
    },
    {
      "epoch": 0.881118126382818,
      "grad_norm": 1.0011833385531665,
      "learning_rate": 1.3102126425916513e-06,
      "loss": 0.4001,
      "step": 16328
    },
    {
      "epoch": 0.8811720900113323,
      "grad_norm": 1.0154417372411588,
      "learning_rate": 1.30993435657591e-06,
      "loss": 0.5307,
      "step": 16329
    },
    {
      "epoch": 0.8812260536398467,
      "grad_norm": 1.118090838711728,
      "learning_rate": 1.30965619098689e-06,
      "loss": 0.5883,
      "step": 16330
    },
    {
      "epoch": 0.8812800172683611,
      "grad_norm": 1.0056359936343076,
      "learning_rate": 1.3093781458325844e-06,
      "loss": 0.4588,
      "step": 16331
    },
    {
      "epoch": 0.8813339808968755,
      "grad_norm": 0.9452585500044672,
      "learning_rate": 1.3091002211209853e-06,
      "loss": 0.4124,
      "step": 16332
    },
    {
      "epoch": 0.8813879445253899,
      "grad_norm": 1.0415532879888756,
      "learning_rate": 1.3088224168600805e-06,
      "loss": 0.4028,
      "step": 16333
    },
    {
      "epoch": 0.8814419081539042,
      "grad_norm": 0.9465635993855074,
      "learning_rate": 1.3085447330578551e-06,
      "loss": 0.3763,
      "step": 16334
    },
    {
      "epoch": 0.8814958717824186,
      "grad_norm": 0.8494108898611828,
      "learning_rate": 1.3082671697222888e-06,
      "loss": 0.4193,
      "step": 16335
    },
    {
      "epoch": 0.881549835410933,
      "grad_norm": 1.0805863944042722,
      "learning_rate": 1.3079897268613586e-06,
      "loss": 0.5026,
      "step": 16336
    },
    {
      "epoch": 0.8816037990394474,
      "grad_norm": 1.410091983415008,
      "learning_rate": 1.3077124044830403e-06,
      "loss": 0.6464,
      "step": 16337
    },
    {
      "epoch": 0.8816577626679618,
      "grad_norm": 1.1653730334290522,
      "learning_rate": 1.307435202595303e-06,
      "loss": 0.6247,
      "step": 16338
    },
    {
      "epoch": 0.8817117262964762,
      "grad_norm": 1.2598140051114024,
      "learning_rate": 1.307158121206114e-06,
      "loss": 0.5201,
      "step": 16339
    },
    {
      "epoch": 0.8817656899249906,
      "grad_norm": 0.7430563731218666,
      "learning_rate": 1.306881160323438e-06,
      "loss": 0.3296,
      "step": 16340
    },
    {
      "epoch": 0.8818196535535049,
      "grad_norm": 0.9912882769534633,
      "learning_rate": 1.3066043199552344e-06,
      "loss": 0.4126,
      "step": 16341
    },
    {
      "epoch": 0.8818736171820193,
      "grad_norm": 1.1267548079484635,
      "learning_rate": 1.3063276001094594e-06,
      "loss": 0.423,
      "step": 16342
    },
    {
      "epoch": 0.8819275808105337,
      "grad_norm": 1.2567577484779262,
      "learning_rate": 1.3060510007940659e-06,
      "loss": 0.5822,
      "step": 16343
    },
    {
      "epoch": 0.8819815444390481,
      "grad_norm": 0.8549632879371133,
      "learning_rate": 1.3057745220170048e-06,
      "loss": 0.3933,
      "step": 16344
    },
    {
      "epoch": 0.8820355080675625,
      "grad_norm": 1.1284381277882378,
      "learning_rate": 1.3054981637862215e-06,
      "loss": 0.4522,
      "step": 16345
    },
    {
      "epoch": 0.8820894716960769,
      "grad_norm": 1.134526738087843,
      "learning_rate": 1.3052219261096602e-06,
      "loss": 0.4045,
      "step": 16346
    },
    {
      "epoch": 0.8821434353245913,
      "grad_norm": 0.8303747636044022,
      "learning_rate": 1.3049458089952583e-06,
      "loss": 0.4033,
      "step": 16347
    },
    {
      "epoch": 0.8821973989531056,
      "grad_norm": 1.0653920681971525,
      "learning_rate": 1.3046698124509535e-06,
      "loss": 0.4707,
      "step": 16348
    },
    {
      "epoch": 0.88225136258162,
      "grad_norm": 0.9851862929910917,
      "learning_rate": 1.3043939364846773e-06,
      "loss": 0.4299,
      "step": 16349
    },
    {
      "epoch": 0.8823053262101344,
      "grad_norm": 1.0319375814748237,
      "learning_rate": 1.304118181104358e-06,
      "loss": 0.4195,
      "step": 16350
    },
    {
      "epoch": 0.8823592898386488,
      "grad_norm": 0.9504788332112963,
      "learning_rate": 1.3038425463179218e-06,
      "loss": 0.443,
      "step": 16351
    },
    {
      "epoch": 0.8824132534671632,
      "grad_norm": 1.0747054378420613,
      "learning_rate": 1.303567032133291e-06,
      "loss": 0.5255,
      "step": 16352
    },
    {
      "epoch": 0.8824672170956775,
      "grad_norm": 0.9986198894569697,
      "learning_rate": 1.3032916385583836e-06,
      "loss": 0.4467,
      "step": 16353
    },
    {
      "epoch": 0.882521180724192,
      "grad_norm": 0.9617094428782382,
      "learning_rate": 1.3030163656011156e-06,
      "loss": 0.3755,
      "step": 16354
    },
    {
      "epoch": 0.8825751443527062,
      "grad_norm": 1.0550446644592442,
      "learning_rate": 1.3027412132693981e-06,
      "loss": 0.4058,
      "step": 16355
    },
    {
      "epoch": 0.8826291079812206,
      "grad_norm": 1.2047229261507832,
      "learning_rate": 1.3024661815711387e-06,
      "loss": 0.3166,
      "step": 16356
    },
    {
      "epoch": 0.882683071609735,
      "grad_norm": 0.9911468637815285,
      "learning_rate": 1.3021912705142423e-06,
      "loss": 0.3818,
      "step": 16357
    },
    {
      "epoch": 0.8827370352382494,
      "grad_norm": 1.0625514569716301,
      "learning_rate": 1.3019164801066108e-06,
      "loss": 0.6496,
      "step": 16358
    },
    {
      "epoch": 0.8827909988667638,
      "grad_norm": 1.178329684062857,
      "learning_rate": 1.3016418103561411e-06,
      "loss": 0.4474,
      "step": 16359
    },
    {
      "epoch": 0.8828449624952782,
      "grad_norm": 1.1620133142654654,
      "learning_rate": 1.3013672612707281e-06,
      "loss": 0.6303,
      "step": 16360
    },
    {
      "epoch": 0.8828989261237926,
      "grad_norm": 0.8902551163234473,
      "learning_rate": 1.3010928328582629e-06,
      "loss": 0.4184,
      "step": 16361
    },
    {
      "epoch": 0.8829528897523069,
      "grad_norm": 0.9698666511251441,
      "learning_rate": 1.3008185251266315e-06,
      "loss": 0.5415,
      "step": 16362
    },
    {
      "epoch": 0.8830068533808213,
      "grad_norm": 0.9136388490165804,
      "learning_rate": 1.3005443380837188e-06,
      "loss": 0.3938,
      "step": 16363
    },
    {
      "epoch": 0.8830608170093357,
      "grad_norm": 1.0263546395930037,
      "learning_rate": 1.3002702717374054e-06,
      "loss": 0.5587,
      "step": 16364
    },
    {
      "epoch": 0.8831147806378501,
      "grad_norm": 0.8966050535143485,
      "learning_rate": 1.2999963260955678e-06,
      "loss": 0.3742,
      "step": 16365
    },
    {
      "epoch": 0.8831687442663645,
      "grad_norm": 0.8946996138187174,
      "learning_rate": 1.2997225011660795e-06,
      "loss": 0.3398,
      "step": 16366
    },
    {
      "epoch": 0.8832227078948789,
      "grad_norm": 1.112348334987563,
      "learning_rate": 1.299448796956811e-06,
      "loss": 0.4108,
      "step": 16367
    },
    {
      "epoch": 0.8832766715233933,
      "grad_norm": 1.2343550785587607,
      "learning_rate": 1.2991752134756278e-06,
      "loss": 0.5867,
      "step": 16368
    },
    {
      "epoch": 0.8833306351519076,
      "grad_norm": 0.9195066623619099,
      "learning_rate": 1.2989017507303944e-06,
      "loss": 0.3703,
      "step": 16369
    },
    {
      "epoch": 0.883384598780422,
      "grad_norm": 1.1365859584556606,
      "learning_rate": 1.2986284087289691e-06,
      "loss": 0.6064,
      "step": 16370
    },
    {
      "epoch": 0.8834385624089364,
      "grad_norm": 1.1635733551956753,
      "learning_rate": 1.2983551874792086e-06,
      "loss": 0.555,
      "step": 16371
    },
    {
      "epoch": 0.8834925260374508,
      "grad_norm": 1.0521671185156922,
      "learning_rate": 1.2980820869889651e-06,
      "loss": 0.3635,
      "step": 16372
    },
    {
      "epoch": 0.8835464896659652,
      "grad_norm": 0.9348341114302332,
      "learning_rate": 1.2978091072660884e-06,
      "loss": 0.417,
      "step": 16373
    },
    {
      "epoch": 0.8836004532944796,
      "grad_norm": 0.8995182468958178,
      "learning_rate": 1.2975362483184239e-06,
      "loss": 0.4144,
      "step": 16374
    },
    {
      "epoch": 0.883654416922994,
      "grad_norm": 1.1337785856229525,
      "learning_rate": 1.2972635101538142e-06,
      "loss": 0.4758,
      "step": 16375
    },
    {
      "epoch": 0.8837083805515082,
      "grad_norm": 0.8951134363350084,
      "learning_rate": 1.296990892780098e-06,
      "loss": 0.4646,
      "step": 16376
    },
    {
      "epoch": 0.8837623441800226,
      "grad_norm": 0.9840971984922269,
      "learning_rate": 1.2967183962051097e-06,
      "loss": 0.5684,
      "step": 16377
    },
    {
      "epoch": 0.883816307808537,
      "grad_norm": 0.8809038442374075,
      "learning_rate": 1.296446020436682e-06,
      "loss": 0.3909,
      "step": 16378
    },
    {
      "epoch": 0.8838702714370514,
      "grad_norm": 1.088991111222965,
      "learning_rate": 1.2961737654826432e-06,
      "loss": 0.3985,
      "step": 16379
    },
    {
      "epoch": 0.8839242350655658,
      "grad_norm": 0.8516374219043824,
      "learning_rate": 1.2959016313508178e-06,
      "loss": 0.4043,
      "step": 16380
    },
    {
      "epoch": 0.8839781986940802,
      "grad_norm": 0.7360247716777377,
      "learning_rate": 1.2956296180490278e-06,
      "loss": 0.2913,
      "step": 16381
    },
    {
      "epoch": 0.8840321623225945,
      "grad_norm": 1.0017739607996576,
      "learning_rate": 1.2953577255850907e-06,
      "loss": 0.4822,
      "step": 16382
    },
    {
      "epoch": 0.8840861259511089,
      "grad_norm": 0.831955594760269,
      "learning_rate": 1.2950859539668208e-06,
      "loss": 0.2627,
      "step": 16383
    },
    {
      "epoch": 0.8841400895796233,
      "grad_norm": 1.4616939205791322,
      "learning_rate": 1.2948143032020296e-06,
      "loss": 0.6098,
      "step": 16384
    },
    {
      "epoch": 0.8841940532081377,
      "grad_norm": 1.0555503936414146,
      "learning_rate": 1.2945427732985242e-06,
      "loss": 0.4732,
      "step": 16385
    },
    {
      "epoch": 0.8842480168366521,
      "grad_norm": 1.0962561779684534,
      "learning_rate": 1.2942713642641086e-06,
      "loss": 0.4363,
      "step": 16386
    },
    {
      "epoch": 0.8843019804651665,
      "grad_norm": 1.0287623960233445,
      "learning_rate": 1.2940000761065837e-06,
      "loss": 0.3768,
      "step": 16387
    },
    {
      "epoch": 0.8843559440936809,
      "grad_norm": 0.8100250947921226,
      "learning_rate": 1.2937289088337464e-06,
      "loss": 0.2968,
      "step": 16388
    },
    {
      "epoch": 0.8844099077221952,
      "grad_norm": 0.9334458354411618,
      "learning_rate": 1.2934578624533913e-06,
      "loss": 0.4371,
      "step": 16389
    },
    {
      "epoch": 0.8844638713507096,
      "grad_norm": 1.038967261868459,
      "learning_rate": 1.2931869369733065e-06,
      "loss": 0.5252,
      "step": 16390
    },
    {
      "epoch": 0.884517834979224,
      "grad_norm": 1.079875836918772,
      "learning_rate": 1.2929161324012804e-06,
      "loss": 0.4523,
      "step": 16391
    },
    {
      "epoch": 0.8845717986077384,
      "grad_norm": 1.0253542262590924,
      "learning_rate": 1.2926454487450953e-06,
      "loss": 0.3709,
      "step": 16392
    },
    {
      "epoch": 0.8846257622362528,
      "grad_norm": 0.9320960900673885,
      "learning_rate": 1.2923748860125313e-06,
      "loss": 0.3165,
      "step": 16393
    },
    {
      "epoch": 0.8846797258647672,
      "grad_norm": 1.2605177479763339,
      "learning_rate": 1.2921044442113647e-06,
      "loss": 0.5414,
      "step": 16394
    },
    {
      "epoch": 0.8847336894932816,
      "grad_norm": 1.081987345831397,
      "learning_rate": 1.2918341233493675e-06,
      "loss": 0.4982,
      "step": 16395
    },
    {
      "epoch": 0.8847876531217959,
      "grad_norm": 0.9654210339201181,
      "learning_rate": 1.291563923434311e-06,
      "loss": 0.3622,
      "step": 16396
    },
    {
      "epoch": 0.8848416167503103,
      "grad_norm": 1.1419374959863091,
      "learning_rate": 1.2912938444739587e-06,
      "loss": 0.6342,
      "step": 16397
    },
    {
      "epoch": 0.8848955803788247,
      "grad_norm": 0.8093493129835441,
      "learning_rate": 1.291023886476074e-06,
      "loss": 0.2975,
      "step": 16398
    },
    {
      "epoch": 0.884949544007339,
      "grad_norm": 1.20709070793282,
      "learning_rate": 1.2907540494484155e-06,
      "loss": 0.5039,
      "step": 16399
    },
    {
      "epoch": 0.8850035076358534,
      "grad_norm": 1.166165220879286,
      "learning_rate": 1.290484333398739e-06,
      "loss": 0.4084,
      "step": 16400
    },
    {
      "epoch": 0.8850574712643678,
      "grad_norm": 0.8835260819370534,
      "learning_rate": 1.290214738334796e-06,
      "loss": 0.4651,
      "step": 16401
    },
    {
      "epoch": 0.8851114348928822,
      "grad_norm": 0.7861476195240932,
      "learning_rate": 1.289945264264335e-06,
      "loss": 0.3078,
      "step": 16402
    },
    {
      "epoch": 0.8851653985213965,
      "grad_norm": 1.0378189422021185,
      "learning_rate": 1.2896759111951018e-06,
      "loss": 0.4216,
      "step": 16403
    },
    {
      "epoch": 0.8852193621499109,
      "grad_norm": 1.0287793796246105,
      "learning_rate": 1.2894066791348365e-06,
      "loss": 0.4273,
      "step": 16404
    },
    {
      "epoch": 0.8852733257784253,
      "grad_norm": 0.8964218092818278,
      "learning_rate": 1.2891375680912773e-06,
      "loss": 0.3087,
      "step": 16405
    },
    {
      "epoch": 0.8853272894069397,
      "grad_norm": 1.1856203571411077,
      "learning_rate": 1.2888685780721602e-06,
      "loss": 0.4933,
      "step": 16406
    },
    {
      "epoch": 0.8853812530354541,
      "grad_norm": 1.1615505767755219,
      "learning_rate": 1.2885997090852146e-06,
      "loss": 0.5066,
      "step": 16407
    },
    {
      "epoch": 0.8854352166639685,
      "grad_norm": 1.073095644857882,
      "learning_rate": 1.2883309611381686e-06,
      "loss": 0.3942,
      "step": 16408
    },
    {
      "epoch": 0.8854891802924829,
      "grad_norm": 0.92725053438267,
      "learning_rate": 1.2880623342387461e-06,
      "loss": 0.3924,
      "step": 16409
    },
    {
      "epoch": 0.8855431439209972,
      "grad_norm": 1.1362806845785176,
      "learning_rate": 1.2877938283946685e-06,
      "loss": 0.5555,
      "step": 16410
    },
    {
      "epoch": 0.8855971075495116,
      "grad_norm": 1.0100703540242997,
      "learning_rate": 1.2875254436136523e-06,
      "loss": 0.4284,
      "step": 16411
    },
    {
      "epoch": 0.885651071178026,
      "grad_norm": 1.107278794918129,
      "learning_rate": 1.287257179903411e-06,
      "loss": 0.4864,
      "step": 16412
    },
    {
      "epoch": 0.8857050348065404,
      "grad_norm": 1.2299812254296707,
      "learning_rate": 1.2869890372716554e-06,
      "loss": 0.5976,
      "step": 16413
    },
    {
      "epoch": 0.8857589984350548,
      "grad_norm": 1.2670854198039954,
      "learning_rate": 1.2867210157260917e-06,
      "loss": 0.5084,
      "step": 16414
    },
    {
      "epoch": 0.8858129620635692,
      "grad_norm": 0.7505876168584866,
      "learning_rate": 1.2864531152744232e-06,
      "loss": 0.2983,
      "step": 16415
    },
    {
      "epoch": 0.8858669256920836,
      "grad_norm": 0.9863252537314471,
      "learning_rate": 1.2861853359243492e-06,
      "loss": 0.3659,
      "step": 16416
    },
    {
      "epoch": 0.8859208893205979,
      "grad_norm": 0.9810064654805469,
      "learning_rate": 1.2859176776835671e-06,
      "loss": 0.4477,
      "step": 16417
    },
    {
      "epoch": 0.8859748529491123,
      "grad_norm": 1.041505178014188,
      "learning_rate": 1.2856501405597683e-06,
      "loss": 0.5491,
      "step": 16418
    },
    {
      "epoch": 0.8860288165776267,
      "grad_norm": 0.9557228680251226,
      "learning_rate": 1.2853827245606435e-06,
      "loss": 0.5254,
      "step": 16419
    },
    {
      "epoch": 0.8860827802061411,
      "grad_norm": 1.2908854821482454,
      "learning_rate": 1.285115429693877e-06,
      "loss": 0.5596,
      "step": 16420
    },
    {
      "epoch": 0.8861367438346555,
      "grad_norm": 1.2914466202607622,
      "learning_rate": 1.2848482559671522e-06,
      "loss": 0.6656,
      "step": 16421
    },
    {
      "epoch": 0.8861907074631699,
      "grad_norm": 1.104565974705371,
      "learning_rate": 1.2845812033881478e-06,
      "loss": 0.3877,
      "step": 16422
    },
    {
      "epoch": 0.8862446710916843,
      "grad_norm": 1.0328896805146894,
      "learning_rate": 1.2843142719645381e-06,
      "loss": 0.4138,
      "step": 16423
    },
    {
      "epoch": 0.8862986347201985,
      "grad_norm": 0.8151965530529152,
      "learning_rate": 1.2840474617039977e-06,
      "loss": 0.3329,
      "step": 16424
    },
    {
      "epoch": 0.8863525983487129,
      "grad_norm": 1.0709161453888374,
      "learning_rate": 1.2837807726141919e-06,
      "loss": 0.4615,
      "step": 16425
    },
    {
      "epoch": 0.8864065619772273,
      "grad_norm": 1.060571317968483,
      "learning_rate": 1.2835142047027868e-06,
      "loss": 0.5468,
      "step": 16426
    },
    {
      "epoch": 0.8864605256057417,
      "grad_norm": 1.0674326013600595,
      "learning_rate": 1.283247757977444e-06,
      "loss": 0.525,
      "step": 16427
    },
    {
      "epoch": 0.8865144892342561,
      "grad_norm": 0.9698898416388564,
      "learning_rate": 1.2829814324458214e-06,
      "loss": 0.4013,
      "step": 16428
    },
    {
      "epoch": 0.8865684528627705,
      "grad_norm": 1.0769489003217798,
      "learning_rate": 1.2827152281155736e-06,
      "loss": 0.3654,
      "step": 16429
    },
    {
      "epoch": 0.8866224164912849,
      "grad_norm": 1.0490606327606042,
      "learning_rate": 1.2824491449943512e-06,
      "loss": 0.5048,
      "step": 16430
    },
    {
      "epoch": 0.8866763801197992,
      "grad_norm": 1.0899988682707076,
      "learning_rate": 1.2821831830898023e-06,
      "loss": 0.4952,
      "step": 16431
    },
    {
      "epoch": 0.8867303437483136,
      "grad_norm": 0.8887075321321516,
      "learning_rate": 1.2819173424095705e-06,
      "loss": 0.4443,
      "step": 16432
    },
    {
      "epoch": 0.886784307376828,
      "grad_norm": 1.2079843910976504,
      "learning_rate": 1.281651622961296e-06,
      "loss": 0.6927,
      "step": 16433
    },
    {
      "epoch": 0.8868382710053424,
      "grad_norm": 0.8972898506224976,
      "learning_rate": 1.2813860247526163e-06,
      "loss": 0.5287,
      "step": 16434
    },
    {
      "epoch": 0.8868922346338568,
      "grad_norm": 1.0506842897137345,
      "learning_rate": 1.2811205477911648e-06,
      "loss": 0.4441,
      "step": 16435
    },
    {
      "epoch": 0.8869461982623712,
      "grad_norm": 1.002325114802535,
      "learning_rate": 1.2808551920845713e-06,
      "loss": 0.4567,
      "step": 16436
    },
    {
      "epoch": 0.8870001618908856,
      "grad_norm": 0.9860623277686188,
      "learning_rate": 1.280589957640463e-06,
      "loss": 0.4985,
      "step": 16437
    },
    {
      "epoch": 0.8870541255193999,
      "grad_norm": 1.0565656584899619,
      "learning_rate": 1.2803248444664631e-06,
      "loss": 0.4944,
      "step": 16438
    },
    {
      "epoch": 0.8871080891479143,
      "grad_norm": 1.0295451897352184,
      "learning_rate": 1.2800598525701905e-06,
      "loss": 0.407,
      "step": 16439
    },
    {
      "epoch": 0.8871620527764287,
      "grad_norm": 1.004366283270853,
      "learning_rate": 1.2797949819592618e-06,
      "loss": 0.4054,
      "step": 16440
    },
    {
      "epoch": 0.8872160164049431,
      "grad_norm": 1.0213623081127725,
      "learning_rate": 1.2795302326412892e-06,
      "loss": 0.4755,
      "step": 16441
    },
    {
      "epoch": 0.8872699800334575,
      "grad_norm": 1.1541480890476314,
      "learning_rate": 1.2792656046238823e-06,
      "loss": 0.5604,
      "step": 16442
    },
    {
      "epoch": 0.8873239436619719,
      "grad_norm": 0.8456243326759948,
      "learning_rate": 1.2790010979146467e-06,
      "loss": 0.3824,
      "step": 16443
    },
    {
      "epoch": 0.8873779072904863,
      "grad_norm": 1.240206881450869,
      "learning_rate": 1.2787367125211844e-06,
      "loss": 0.4309,
      "step": 16444
    },
    {
      "epoch": 0.8874318709190006,
      "grad_norm": 0.8075517665117974,
      "learning_rate": 1.2784724484510952e-06,
      "loss": 0.333,
      "step": 16445
    },
    {
      "epoch": 0.887485834547515,
      "grad_norm": 0.9864276806200334,
      "learning_rate": 1.2782083057119723e-06,
      "loss": 0.4129,
      "step": 16446
    },
    {
      "epoch": 0.8875397981760293,
      "grad_norm": 0.9795681691921394,
      "learning_rate": 1.2779442843114087e-06,
      "loss": 0.4754,
      "step": 16447
    },
    {
      "epoch": 0.8875937618045437,
      "grad_norm": 1.290433594121106,
      "learning_rate": 1.2776803842569924e-06,
      "loss": 0.5115,
      "step": 16448
    },
    {
      "epoch": 0.8876477254330581,
      "grad_norm": 0.9854478626227076,
      "learning_rate": 1.2774166055563084e-06,
      "loss": 0.537,
      "step": 16449
    },
    {
      "epoch": 0.8877016890615725,
      "grad_norm": 0.7989539072069418,
      "learning_rate": 1.2771529482169379e-06,
      "loss": 0.3242,
      "step": 16450
    },
    {
      "epoch": 0.8877556526900868,
      "grad_norm": 0.9835767919956414,
      "learning_rate": 1.2768894122464588e-06,
      "loss": 0.4498,
      "step": 16451
    },
    {
      "epoch": 0.8878096163186012,
      "grad_norm": 0.8525160141914656,
      "learning_rate": 1.2766259976524444e-06,
      "loss": 0.3453,
      "step": 16452
    },
    {
      "epoch": 0.8878635799471156,
      "grad_norm": 0.8189167659545582,
      "learning_rate": 1.2763627044424665e-06,
      "loss": 0.3584,
      "step": 16453
    },
    {
      "epoch": 0.88791754357563,
      "grad_norm": 0.9919440461034471,
      "learning_rate": 1.2760995326240923e-06,
      "loss": 0.457,
      "step": 16454
    },
    {
      "epoch": 0.8879715072041444,
      "grad_norm": 0.9519144645747675,
      "learning_rate": 1.2758364822048855e-06,
      "loss": 0.416,
      "step": 16455
    },
    {
      "epoch": 0.8880254708326588,
      "grad_norm": 1.0663090938568633,
      "learning_rate": 1.2755735531924063e-06,
      "loss": 0.4505,
      "step": 16456
    },
    {
      "epoch": 0.8880794344611732,
      "grad_norm": 1.051992675941245,
      "learning_rate": 1.275310745594211e-06,
      "loss": 0.5079,
      "step": 16457
    },
    {
      "epoch": 0.8881333980896875,
      "grad_norm": 0.9160025602100328,
      "learning_rate": 1.275048059417855e-06,
      "loss": 0.4837,
      "step": 16458
    },
    {
      "epoch": 0.8881873617182019,
      "grad_norm": 1.0354868373294255,
      "learning_rate": 1.274785494670886e-06,
      "loss": 0.5487,
      "step": 16459
    },
    {
      "epoch": 0.8882413253467163,
      "grad_norm": 0.8962256058130614,
      "learning_rate": 1.2745230513608512e-06,
      "loss": 0.3515,
      "step": 16460
    },
    {
      "epoch": 0.8882952889752307,
      "grad_norm": 1.1315919440814448,
      "learning_rate": 1.2742607294952936e-06,
      "loss": 0.5272,
      "step": 16461
    },
    {
      "epoch": 0.8883492526037451,
      "grad_norm": 0.9438513699188169,
      "learning_rate": 1.2739985290817519e-06,
      "loss": 0.3895,
      "step": 16462
    },
    {
      "epoch": 0.8884032162322595,
      "grad_norm": 1.1674968190840456,
      "learning_rate": 1.2737364501277627e-06,
      "loss": 0.623,
      "step": 16463
    },
    {
      "epoch": 0.8884571798607739,
      "grad_norm": 1.0396420498317855,
      "learning_rate": 1.2734744926408585e-06,
      "loss": 0.5119,
      "step": 16464
    },
    {
      "epoch": 0.8885111434892882,
      "grad_norm": 1.1078827166579142,
      "learning_rate": 1.2732126566285676e-06,
      "loss": 0.4782,
      "step": 16465
    },
    {
      "epoch": 0.8885651071178026,
      "grad_norm": 1.0856388166542545,
      "learning_rate": 1.272950942098416e-06,
      "loss": 0.5306,
      "step": 16466
    },
    {
      "epoch": 0.888619070746317,
      "grad_norm": 0.8935620009613143,
      "learning_rate": 1.2726893490579253e-06,
      "loss": 0.3611,
      "step": 16467
    },
    {
      "epoch": 0.8886730343748314,
      "grad_norm": 0.8170034191561761,
      "learning_rate": 1.2724278775146143e-06,
      "loss": 0.2925,
      "step": 16468
    },
    {
      "epoch": 0.8887269980033458,
      "grad_norm": 0.8609069150892673,
      "learning_rate": 1.2721665274759979e-06,
      "loss": 0.3795,
      "step": 16469
    },
    {
      "epoch": 0.8887809616318602,
      "grad_norm": 1.009447510559249,
      "learning_rate": 1.271905298949587e-06,
      "loss": 0.4837,
      "step": 16470
    },
    {
      "epoch": 0.8888349252603746,
      "grad_norm": 0.7341122017535853,
      "learning_rate": 1.2716441919428903e-06,
      "loss": 0.3017,
      "step": 16471
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 1.1207995725947584,
      "learning_rate": 1.2713832064634127e-06,
      "loss": 0.4981,
      "step": 16472
    },
    {
      "epoch": 0.8889428525174032,
      "grad_norm": 1.0568906055695508,
      "learning_rate": 1.2711223425186539e-06,
      "loss": 0.4212,
      "step": 16473
    },
    {
      "epoch": 0.8889968161459176,
      "grad_norm": 0.7845839562116749,
      "learning_rate": 1.270861600116111e-06,
      "loss": 0.3198,
      "step": 16474
    },
    {
      "epoch": 0.889050779774432,
      "grad_norm": 0.7979085036851493,
      "learning_rate": 1.2706009792632799e-06,
      "loss": 0.3025,
      "step": 16475
    },
    {
      "epoch": 0.8891047434029464,
      "grad_norm": 1.0221419998354604,
      "learning_rate": 1.2703404799676497e-06,
      "loss": 0.4756,
      "step": 16476
    },
    {
      "epoch": 0.8891587070314608,
      "grad_norm": 1.2026524847805997,
      "learning_rate": 1.270080102236708e-06,
      "loss": 0.5554,
      "step": 16477
    },
    {
      "epoch": 0.8892126706599752,
      "grad_norm": 0.9589015940299753,
      "learning_rate": 1.2698198460779388e-06,
      "loss": 0.4472,
      "step": 16478
    },
    {
      "epoch": 0.8892666342884895,
      "grad_norm": 0.9712733035366702,
      "learning_rate": 1.2695597114988213e-06,
      "loss": 0.4188,
      "step": 16479
    },
    {
      "epoch": 0.8893205979170039,
      "grad_norm": 1.0225060043070002,
      "learning_rate": 1.2692996985068317e-06,
      "loss": 0.4887,
      "step": 16480
    },
    {
      "epoch": 0.8893745615455183,
      "grad_norm": 1.0844642834987794,
      "learning_rate": 1.269039807109444e-06,
      "loss": 0.4168,
      "step": 16481
    },
    {
      "epoch": 0.8894285251740327,
      "grad_norm": 0.7601127380542876,
      "learning_rate": 1.2687800373141272e-06,
      "loss": 0.2665,
      "step": 16482
    },
    {
      "epoch": 0.8894824888025471,
      "grad_norm": 1.2674125123268303,
      "learning_rate": 1.2685203891283473e-06,
      "loss": 0.5738,
      "step": 16483
    },
    {
      "epoch": 0.8895364524310615,
      "grad_norm": 0.8785650203446898,
      "learning_rate": 1.2682608625595668e-06,
      "loss": 0.316,
      "step": 16484
    },
    {
      "epoch": 0.8895904160595759,
      "grad_norm": 0.7980527998176535,
      "learning_rate": 1.2680014576152452e-06,
      "loss": 0.2798,
      "step": 16485
    },
    {
      "epoch": 0.8896443796880902,
      "grad_norm": 1.011726549358425,
      "learning_rate": 1.2677421743028385e-06,
      "loss": 0.5324,
      "step": 16486
    },
    {
      "epoch": 0.8896983433166046,
      "grad_norm": 1.1770644150639689,
      "learning_rate": 1.2674830126297973e-06,
      "loss": 0.6624,
      "step": 16487
    },
    {
      "epoch": 0.889752306945119,
      "grad_norm": 0.9616074781754076,
      "learning_rate": 1.2672239726035712e-06,
      "loss": 0.3578,
      "step": 16488
    },
    {
      "epoch": 0.8898062705736334,
      "grad_norm": 1.0117915401309545,
      "learning_rate": 1.2669650542316048e-06,
      "loss": 0.4478,
      "step": 16489
    },
    {
      "epoch": 0.8898602342021478,
      "grad_norm": 1.1999455862455493,
      "learning_rate": 1.2667062575213403e-06,
      "loss": 0.4966,
      "step": 16490
    },
    {
      "epoch": 0.8899141978306622,
      "grad_norm": 1.0242588006543893,
      "learning_rate": 1.2664475824802152e-06,
      "loss": 0.3755,
      "step": 16491
    },
    {
      "epoch": 0.8899681614591766,
      "grad_norm": 1.105176615139589,
      "learning_rate": 1.266189029115664e-06,
      "loss": 0.5054,
      "step": 16492
    },
    {
      "epoch": 0.8900221250876909,
      "grad_norm": 0.8907792860475334,
      "learning_rate": 1.265930597435119e-06,
      "loss": 0.3603,
      "step": 16493
    },
    {
      "epoch": 0.8900760887162052,
      "grad_norm": 1.0198933135094712,
      "learning_rate": 1.2656722874460062e-06,
      "loss": 0.497,
      "step": 16494
    },
    {
      "epoch": 0.8901300523447196,
      "grad_norm": 1.044316266920108,
      "learning_rate": 1.26541409915575e-06,
      "loss": 0.4672,
      "step": 16495
    },
    {
      "epoch": 0.890184015973234,
      "grad_norm": 0.8473874034473863,
      "learning_rate": 1.2651560325717717e-06,
      "loss": 0.2748,
      "step": 16496
    },
    {
      "epoch": 0.8902379796017484,
      "grad_norm": 0.9419617667005832,
      "learning_rate": 1.264898087701488e-06,
      "loss": 0.3286,
      "step": 16497
    },
    {
      "epoch": 0.8902919432302628,
      "grad_norm": 1.308104805888339,
      "learning_rate": 1.2646402645523124e-06,
      "loss": 0.5435,
      "step": 16498
    },
    {
      "epoch": 0.8903459068587772,
      "grad_norm": 0.9358306876718675,
      "learning_rate": 1.264382563131655e-06,
      "loss": 0.4782,
      "step": 16499
    },
    {
      "epoch": 0.8903998704872915,
      "grad_norm": 1.084478990800832,
      "learning_rate": 1.2641249834469236e-06,
      "loss": 0.5039,
      "step": 16500
    },
    {
      "epoch": 0.8903998704872915,
      "eval_loss": 0.5293381810188293,
      "eval_runtime": 163.501,
      "eval_samples_per_second": 21.034,
      "eval_steps_per_second": 0.881,
      "step": 16500
    },
    {
      "epoch": 0.8904538341158059,
      "grad_norm": 1.0217530888648618,
      "learning_rate": 1.263867525505519e-06,
      "loss": 0.3723,
      "step": 16501
    },
    {
      "epoch": 0.8905077977443203,
      "grad_norm": 0.8472186345214957,
      "learning_rate": 1.2636101893148422e-06,
      "loss": 0.3893,
      "step": 16502
    },
    {
      "epoch": 0.8905617613728347,
      "grad_norm": 1.2526475131335866,
      "learning_rate": 1.2633529748822894e-06,
      "loss": 0.6571,
      "step": 16503
    },
    {
      "epoch": 0.8906157250013491,
      "grad_norm": 1.302715439291715,
      "learning_rate": 1.2630958822152526e-06,
      "loss": 0.5258,
      "step": 16504
    },
    {
      "epoch": 0.8906696886298635,
      "grad_norm": 0.9318817368330088,
      "learning_rate": 1.2628389113211214e-06,
      "loss": 0.4626,
      "step": 16505
    },
    {
      "epoch": 0.8907236522583779,
      "grad_norm": 0.974352124059338,
      "learning_rate": 1.2625820622072808e-06,
      "loss": 0.3899,
      "step": 16506
    },
    {
      "epoch": 0.8907776158868922,
      "grad_norm": 1.0275919221901608,
      "learning_rate": 1.2623253348811143e-06,
      "loss": 0.5347,
      "step": 16507
    },
    {
      "epoch": 0.8908315795154066,
      "grad_norm": 0.9639482680189808,
      "learning_rate": 1.2620687293499992e-06,
      "loss": 0.4032,
      "step": 16508
    },
    {
      "epoch": 0.890885543143921,
      "grad_norm": 1.1985002830238212,
      "learning_rate": 1.2618122456213105e-06,
      "loss": 0.5555,
      "step": 16509
    },
    {
      "epoch": 0.8909395067724354,
      "grad_norm": 0.9135884283006279,
      "learning_rate": 1.2615558837024204e-06,
      "loss": 0.4077,
      "step": 16510
    },
    {
      "epoch": 0.8909934704009498,
      "grad_norm": 1.0172487447063896,
      "learning_rate": 1.2612996436006969e-06,
      "loss": 0.4663,
      "step": 16511
    },
    {
      "epoch": 0.8910474340294642,
      "grad_norm": 1.0748884954276077,
      "learning_rate": 1.2610435253235042e-06,
      "loss": 0.4069,
      "step": 16512
    },
    {
      "epoch": 0.8911013976579786,
      "grad_norm": 1.1684402971671777,
      "learning_rate": 1.2607875288782043e-06,
      "loss": 0.5033,
      "step": 16513
    },
    {
      "epoch": 0.8911553612864929,
      "grad_norm": 0.966950269708176,
      "learning_rate": 1.2605316542721541e-06,
      "loss": 0.3671,
      "step": 16514
    },
    {
      "epoch": 0.8912093249150073,
      "grad_norm": 1.3241290628388032,
      "learning_rate": 1.2602759015127075e-06,
      "loss": 0.5862,
      "step": 16515
    },
    {
      "epoch": 0.8912632885435217,
      "grad_norm": 1.1703286419328354,
      "learning_rate": 1.2600202706072157e-06,
      "loss": 0.6544,
      "step": 16516
    },
    {
      "epoch": 0.8913172521720361,
      "grad_norm": 1.138413146711598,
      "learning_rate": 1.259764761563025e-06,
      "loss": 0.4253,
      "step": 16517
    },
    {
      "epoch": 0.8913712158005505,
      "grad_norm": 1.0627597472587622,
      "learning_rate": 1.2595093743874804e-06,
      "loss": 0.398,
      "step": 16518
    },
    {
      "epoch": 0.8914251794290649,
      "grad_norm": 0.9108117835236497,
      "learning_rate": 1.2592541090879204e-06,
      "loss": 0.4787,
      "step": 16519
    },
    {
      "epoch": 0.8914791430575791,
      "grad_norm": 1.200334765692928,
      "learning_rate": 1.2589989656716828e-06,
      "loss": 0.6281,
      "step": 16520
    },
    {
      "epoch": 0.8915331066860935,
      "grad_norm": 0.994082617632326,
      "learning_rate": 1.2587439441460997e-06,
      "loss": 0.421,
      "step": 16521
    },
    {
      "epoch": 0.8915870703146079,
      "grad_norm": 1.1961546795934537,
      "learning_rate": 1.2584890445185017e-06,
      "loss": 0.6503,
      "step": 16522
    },
    {
      "epoch": 0.8916410339431223,
      "grad_norm": 1.0001549706647135,
      "learning_rate": 1.2582342667962136e-06,
      "loss": 0.4951,
      "step": 16523
    },
    {
      "epoch": 0.8916949975716367,
      "grad_norm": 1.0615005433632017,
      "learning_rate": 1.2579796109865586e-06,
      "loss": 0.4716,
      "step": 16524
    },
    {
      "epoch": 0.8917489612001511,
      "grad_norm": 0.8826479559599854,
      "learning_rate": 1.2577250770968564e-06,
      "loss": 0.3994,
      "step": 16525
    },
    {
      "epoch": 0.8918029248286655,
      "grad_norm": 1.215901497263955,
      "learning_rate": 1.2574706651344218e-06,
      "loss": 0.5102,
      "step": 16526
    },
    {
      "epoch": 0.8918568884571798,
      "grad_norm": 0.9888895200018744,
      "learning_rate": 1.2572163751065671e-06,
      "loss": 0.4634,
      "step": 16527
    },
    {
      "epoch": 0.8919108520856942,
      "grad_norm": 1.118666577379255,
      "learning_rate": 1.2569622070206005e-06,
      "loss": 0.4976,
      "step": 16528
    },
    {
      "epoch": 0.8919648157142086,
      "grad_norm": 1.1091448515902338,
      "learning_rate": 1.256708160883828e-06,
      "loss": 0.449,
      "step": 16529
    },
    {
      "epoch": 0.892018779342723,
      "grad_norm": 1.108640379548791,
      "learning_rate": 1.2564542367035502e-06,
      "loss": 0.4259,
      "step": 16530
    },
    {
      "epoch": 0.8920727429712374,
      "grad_norm": 1.1091838449648266,
      "learning_rate": 1.256200434487065e-06,
      "loss": 0.3998,
      "step": 16531
    },
    {
      "epoch": 0.8921267065997518,
      "grad_norm": 1.0176849206478824,
      "learning_rate": 1.2559467542416673e-06,
      "loss": 0.5137,
      "step": 16532
    },
    {
      "epoch": 0.8921806702282662,
      "grad_norm": 1.203340161308398,
      "learning_rate": 1.2556931959746485e-06,
      "loss": 0.4936,
      "step": 16533
    },
    {
      "epoch": 0.8922346338567805,
      "grad_norm": 0.8327366562798605,
      "learning_rate": 1.2554397596932963e-06,
      "loss": 0.3888,
      "step": 16534
    },
    {
      "epoch": 0.8922885974852949,
      "grad_norm": 4.393572328472059,
      "learning_rate": 1.2551864454048935e-06,
      "loss": 0.3703,
      "step": 16535
    },
    {
      "epoch": 0.8923425611138093,
      "grad_norm": 0.8959613009695632,
      "learning_rate": 1.2549332531167216e-06,
      "loss": 0.409,
      "step": 16536
    },
    {
      "epoch": 0.8923965247423237,
      "grad_norm": 0.9148845356521775,
      "learning_rate": 1.2546801828360572e-06,
      "loss": 0.4549,
      "step": 16537
    },
    {
      "epoch": 0.8924504883708381,
      "grad_norm": 1.0401684755893603,
      "learning_rate": 1.2544272345701742e-06,
      "loss": 0.5105,
      "step": 16538
    },
    {
      "epoch": 0.8925044519993525,
      "grad_norm": 1.055691724688839,
      "learning_rate": 1.2541744083263422e-06,
      "loss": 0.5184,
      "step": 16539
    },
    {
      "epoch": 0.8925584156278669,
      "grad_norm": 1.108073633361213,
      "learning_rate": 1.2539217041118275e-06,
      "loss": 0.6506,
      "step": 16540
    },
    {
      "epoch": 0.8926123792563811,
      "grad_norm": 1.1510604963061541,
      "learning_rate": 1.253669121933894e-06,
      "loss": 0.5981,
      "step": 16541
    },
    {
      "epoch": 0.8926663428848955,
      "grad_norm": 1.1748632438433584,
      "learning_rate": 1.2534166617998004e-06,
      "loss": 0.4788,
      "step": 16542
    },
    {
      "epoch": 0.89272030651341,
      "grad_norm": 0.9486444859560507,
      "learning_rate": 1.253164323716802e-06,
      "loss": 0.4692,
      "step": 16543
    },
    {
      "epoch": 0.8927742701419243,
      "grad_norm": 1.060085343321271,
      "learning_rate": 1.2529121076921526e-06,
      "loss": 0.565,
      "step": 16544
    },
    {
      "epoch": 0.8928282337704387,
      "grad_norm": 1.107667743922369,
      "learning_rate": 1.2526600137331007e-06,
      "loss": 0.5324,
      "step": 16545
    },
    {
      "epoch": 0.8928821973989531,
      "grad_norm": 1.0208831219249432,
      "learning_rate": 1.2524080418468914e-06,
      "loss": 0.4214,
      "step": 16546
    },
    {
      "epoch": 0.8929361610274675,
      "grad_norm": 1.0174501093969073,
      "learning_rate": 1.252156192040767e-06,
      "loss": 0.5105,
      "step": 16547
    },
    {
      "epoch": 0.8929901246559818,
      "grad_norm": 0.9869706570967811,
      "learning_rate": 1.2519044643219662e-06,
      "loss": 0.4753,
      "step": 16548
    },
    {
      "epoch": 0.8930440882844962,
      "grad_norm": 0.8215668508517365,
      "learning_rate": 1.2516528586977232e-06,
      "loss": 0.3511,
      "step": 16549
    },
    {
      "epoch": 0.8930980519130106,
      "grad_norm": 1.3135463816799575,
      "learning_rate": 1.2514013751752695e-06,
      "loss": 0.6137,
      "step": 16550
    },
    {
      "epoch": 0.893152015541525,
      "grad_norm": 0.9959825424413834,
      "learning_rate": 1.2511500137618329e-06,
      "loss": 0.4249,
      "step": 16551
    },
    {
      "epoch": 0.8932059791700394,
      "grad_norm": 1.0272823321431281,
      "learning_rate": 1.2508987744646384e-06,
      "loss": 0.4823,
      "step": 16552
    },
    {
      "epoch": 0.8932599427985538,
      "grad_norm": 0.9258018896104172,
      "learning_rate": 1.2506476572909062e-06,
      "loss": 0.3954,
      "step": 16553
    },
    {
      "epoch": 0.8933139064270682,
      "grad_norm": 0.7578424444688344,
      "learning_rate": 1.2503966622478542e-06,
      "loss": 0.322,
      "step": 16554
    },
    {
      "epoch": 0.8933678700555825,
      "grad_norm": 1.0080123328317325,
      "learning_rate": 1.250145789342697e-06,
      "loss": 0.4534,
      "step": 16555
    },
    {
      "epoch": 0.8934218336840969,
      "grad_norm": 0.9650576438984382,
      "learning_rate": 1.2498950385826427e-06,
      "loss": 0.4054,
      "step": 16556
    },
    {
      "epoch": 0.8934757973126113,
      "grad_norm": 0.8483603364968832,
      "learning_rate": 1.2496444099748997e-06,
      "loss": 0.3958,
      "step": 16557
    },
    {
      "epoch": 0.8935297609411257,
      "grad_norm": 0.8181061791071963,
      "learning_rate": 1.2493939035266712e-06,
      "loss": 0.3266,
      "step": 16558
    },
    {
      "epoch": 0.8935837245696401,
      "grad_norm": 1.0075346426934482,
      "learning_rate": 1.249143519245157e-06,
      "loss": 0.6376,
      "step": 16559
    },
    {
      "epoch": 0.8936376881981545,
      "grad_norm": 1.036891000265129,
      "learning_rate": 1.2488932571375527e-06,
      "loss": 0.6937,
      "step": 16560
    },
    {
      "epoch": 0.8936916518266689,
      "grad_norm": 0.9519098701554323,
      "learning_rate": 1.2486431172110516e-06,
      "loss": 0.4522,
      "step": 16561
    },
    {
      "epoch": 0.8937456154551832,
      "grad_norm": 1.170704134399399,
      "learning_rate": 1.248393099472843e-06,
      "loss": 0.5588,
      "step": 16562
    },
    {
      "epoch": 0.8937995790836976,
      "grad_norm": 0.9444481580915027,
      "learning_rate": 1.2481432039301134e-06,
      "loss": 0.3761,
      "step": 16563
    },
    {
      "epoch": 0.893853542712212,
      "grad_norm": 1.2190771469882187,
      "learning_rate": 1.2478934305900445e-06,
      "loss": 0.5295,
      "step": 16564
    },
    {
      "epoch": 0.8939075063407264,
      "grad_norm": 1.179284402695383,
      "learning_rate": 1.247643779459814e-06,
      "loss": 0.35,
      "step": 16565
    },
    {
      "epoch": 0.8939614699692408,
      "grad_norm": 0.9252540307693538,
      "learning_rate": 1.2473942505465978e-06,
      "loss": 0.4644,
      "step": 16566
    },
    {
      "epoch": 0.8940154335977551,
      "grad_norm": 1.0768920642088355,
      "learning_rate": 1.2471448438575682e-06,
      "loss": 0.4941,
      "step": 16567
    },
    {
      "epoch": 0.8940693972262695,
      "grad_norm": 1.0970095163635307,
      "learning_rate": 1.246895559399893e-06,
      "loss": 0.4218,
      "step": 16568
    },
    {
      "epoch": 0.8941233608547838,
      "grad_norm": 1.0444501113468312,
      "learning_rate": 1.2466463971807368e-06,
      "loss": 0.4468,
      "step": 16569
    },
    {
      "epoch": 0.8941773244832982,
      "grad_norm": 1.2698907470415834,
      "learning_rate": 1.2463973572072604e-06,
      "loss": 0.5248,
      "step": 16570
    },
    {
      "epoch": 0.8942312881118126,
      "grad_norm": 1.1076601204940808,
      "learning_rate": 1.2461484394866224e-06,
      "loss": 0.5472,
      "step": 16571
    },
    {
      "epoch": 0.894285251740327,
      "grad_norm": 1.0560689343137104,
      "learning_rate": 1.2458996440259766e-06,
      "loss": 0.4427,
      "step": 16572
    },
    {
      "epoch": 0.8943392153688414,
      "grad_norm": 1.0054769613027383,
      "learning_rate": 1.2456509708324733e-06,
      "loss": 0.3454,
      "step": 16573
    },
    {
      "epoch": 0.8943931789973558,
      "grad_norm": 1.359604787302321,
      "learning_rate": 1.2454024199132594e-06,
      "loss": 0.4874,
      "step": 16574
    },
    {
      "epoch": 0.8944471426258702,
      "grad_norm": 0.834195320309321,
      "learning_rate": 1.2451539912754795e-06,
      "loss": 0.282,
      "step": 16575
    },
    {
      "epoch": 0.8945011062543845,
      "grad_norm": 1.0379056513969505,
      "learning_rate": 1.2449056849262735e-06,
      "loss": 0.4906,
      "step": 16576
    },
    {
      "epoch": 0.8945550698828989,
      "grad_norm": 1.0311886052535417,
      "learning_rate": 1.2446575008727765e-06,
      "loss": 0.4416,
      "step": 16577
    },
    {
      "epoch": 0.8946090335114133,
      "grad_norm": 0.9215855354907386,
      "learning_rate": 1.2444094391221235e-06,
      "loss": 0.3505,
      "step": 16578
    },
    {
      "epoch": 0.8946629971399277,
      "grad_norm": 1.17542271939718,
      "learning_rate": 1.2441614996814425e-06,
      "loss": 0.5333,
      "step": 16579
    },
    {
      "epoch": 0.8947169607684421,
      "grad_norm": 0.8544448824884102,
      "learning_rate": 1.2439136825578605e-06,
      "loss": 0.3386,
      "step": 16580
    },
    {
      "epoch": 0.8947709243969565,
      "grad_norm": 1.1827145759065303,
      "learning_rate": 1.2436659877584998e-06,
      "loss": 0.4428,
      "step": 16581
    },
    {
      "epoch": 0.8948248880254709,
      "grad_norm": 0.8889807992430249,
      "learning_rate": 1.2434184152904792e-06,
      "loss": 0.3553,
      "step": 16582
    },
    {
      "epoch": 0.8948788516539852,
      "grad_norm": 0.910967540231939,
      "learning_rate": 1.2431709651609147e-06,
      "loss": 0.3701,
      "step": 16583
    },
    {
      "epoch": 0.8949328152824996,
      "grad_norm": 1.2303874609595469,
      "learning_rate": 1.2429236373769175e-06,
      "loss": 0.5456,
      "step": 16584
    },
    {
      "epoch": 0.894986778911014,
      "grad_norm": 0.9799532161018254,
      "learning_rate": 1.2426764319455964e-06,
      "loss": 0.4361,
      "step": 16585
    },
    {
      "epoch": 0.8950407425395284,
      "grad_norm": 1.1075065324683198,
      "learning_rate": 1.2424293488740564e-06,
      "loss": 0.4961,
      "step": 16586
    },
    {
      "epoch": 0.8950947061680428,
      "grad_norm": 1.0057056562968245,
      "learning_rate": 1.2421823881693988e-06,
      "loss": 0.4948,
      "step": 16587
    },
    {
      "epoch": 0.8951486697965572,
      "grad_norm": 1.0257909555267728,
      "learning_rate": 1.2419355498387215e-06,
      "loss": 0.5134,
      "step": 16588
    },
    {
      "epoch": 0.8952026334250714,
      "grad_norm": 1.110669033244554,
      "learning_rate": 1.2416888338891198e-06,
      "loss": 0.518,
      "step": 16589
    },
    {
      "epoch": 0.8952565970535858,
      "grad_norm": 1.0501869037450313,
      "learning_rate": 1.241442240327683e-06,
      "loss": 0.4213,
      "step": 16590
    },
    {
      "epoch": 0.8953105606821002,
      "grad_norm": 0.9824561878084013,
      "learning_rate": 1.241195769161499e-06,
      "loss": 0.4979,
      "step": 16591
    },
    {
      "epoch": 0.8953645243106146,
      "grad_norm": 1.2188297950057785,
      "learning_rate": 1.2409494203976523e-06,
      "loss": 0.5692,
      "step": 16592
    },
    {
      "epoch": 0.895418487939129,
      "grad_norm": 1.2422629455908751,
      "learning_rate": 1.2407031940432222e-06,
      "loss": 0.5979,
      "step": 16593
    },
    {
      "epoch": 0.8954724515676434,
      "grad_norm": 1.0726866733247764,
      "learning_rate": 1.240457090105286e-06,
      "loss": 0.4447,
      "step": 16594
    },
    {
      "epoch": 0.8955264151961578,
      "grad_norm": 0.9095784964048158,
      "learning_rate": 1.2402111085909171e-06,
      "loss": 0.3662,
      "step": 16595
    },
    {
      "epoch": 0.8955803788246721,
      "grad_norm": 1.193822793723948,
      "learning_rate": 1.239965249507186e-06,
      "loss": 0.5798,
      "step": 16596
    },
    {
      "epoch": 0.8956343424531865,
      "grad_norm": 0.8376405544396237,
      "learning_rate": 1.2397195128611577e-06,
      "loss": 0.4392,
      "step": 16597
    },
    {
      "epoch": 0.8956883060817009,
      "grad_norm": 0.8780492847457859,
      "learning_rate": 1.2394738986598947e-06,
      "loss": 0.3965,
      "step": 16598
    },
    {
      "epoch": 0.8957422697102153,
      "grad_norm": 1.1744904278538202,
      "learning_rate": 1.2392284069104567e-06,
      "loss": 0.5452,
      "step": 16599
    },
    {
      "epoch": 0.8957962333387297,
      "grad_norm": 1.0342346237573024,
      "learning_rate": 1.2389830376199001e-06,
      "loss": 0.3535,
      "step": 16600
    },
    {
      "epoch": 0.8958501969672441,
      "grad_norm": 1.079316604871361,
      "learning_rate": 1.2387377907952758e-06,
      "loss": 0.3567,
      "step": 16601
    },
    {
      "epoch": 0.8959041605957585,
      "grad_norm": 1.060413161250887,
      "learning_rate": 1.2384926664436337e-06,
      "loss": 0.3661,
      "step": 16602
    },
    {
      "epoch": 0.8959581242242728,
      "grad_norm": 0.9125578785050198,
      "learning_rate": 1.2382476645720182e-06,
      "loss": 0.3678,
      "step": 16603
    },
    {
      "epoch": 0.8960120878527872,
      "grad_norm": 1.0493319708316957,
      "learning_rate": 1.2380027851874711e-06,
      "loss": 0.5027,
      "step": 16604
    },
    {
      "epoch": 0.8960660514813016,
      "grad_norm": 0.9335126376802196,
      "learning_rate": 1.2377580282970301e-06,
      "loss": 0.3236,
      "step": 16605
    },
    {
      "epoch": 0.896120015109816,
      "grad_norm": 0.8194309305720052,
      "learning_rate": 1.2375133939077303e-06,
      "loss": 0.2602,
      "step": 16606
    },
    {
      "epoch": 0.8961739787383304,
      "grad_norm": 1.246286250743248,
      "learning_rate": 1.237268882026602e-06,
      "loss": 0.4087,
      "step": 16607
    },
    {
      "epoch": 0.8962279423668448,
      "grad_norm": 1.1922214335501442,
      "learning_rate": 1.237024492660673e-06,
      "loss": 0.4764,
      "step": 16608
    },
    {
      "epoch": 0.8962819059953592,
      "grad_norm": 0.9287961995002572,
      "learning_rate": 1.2367802258169683e-06,
      "loss": 0.3628,
      "step": 16609
    },
    {
      "epoch": 0.8963358696238735,
      "grad_norm": 1.4348745743586708,
      "learning_rate": 1.236536081502508e-06,
      "loss": 0.554,
      "step": 16610
    },
    {
      "epoch": 0.8963898332523879,
      "grad_norm": 0.9921380084129724,
      "learning_rate": 1.2362920597243074e-06,
      "loss": 0.4572,
      "step": 16611
    },
    {
      "epoch": 0.8964437968809023,
      "grad_norm": 1.0439618257091634,
      "learning_rate": 1.236048160489382e-06,
      "loss": 0.5197,
      "step": 16612
    },
    {
      "epoch": 0.8964977605094167,
      "grad_norm": 0.8907244918413592,
      "learning_rate": 1.2358043838047404e-06,
      "loss": 0.3505,
      "step": 16613
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 1.0465177807830133,
      "learning_rate": 1.2355607296773896e-06,
      "loss": 0.535,
      "step": 16614
    },
    {
      "epoch": 0.8966056877664454,
      "grad_norm": 1.1278392209618169,
      "learning_rate": 1.235317198114332e-06,
      "loss": 0.4382,
      "step": 16615
    },
    {
      "epoch": 0.8966596513949598,
      "grad_norm": 1.0936893214142978,
      "learning_rate": 1.2350737891225675e-06,
      "loss": 0.4891,
      "step": 16616
    },
    {
      "epoch": 0.8967136150234741,
      "grad_norm": 0.9670248450091921,
      "learning_rate": 1.2348305027090923e-06,
      "loss": 0.4992,
      "step": 16617
    },
    {
      "epoch": 0.8967675786519885,
      "grad_norm": 0.8009880066409537,
      "learning_rate": 1.2345873388808971e-06,
      "loss": 0.2583,
      "step": 16618
    },
    {
      "epoch": 0.8968215422805029,
      "grad_norm": 1.0752712582113488,
      "learning_rate": 1.2343442976449724e-06,
      "loss": 0.4609,
      "step": 16619
    },
    {
      "epoch": 0.8968755059090173,
      "grad_norm": 1.056995790472711,
      "learning_rate": 1.234101379008302e-06,
      "loss": 0.4896,
      "step": 16620
    },
    {
      "epoch": 0.8969294695375317,
      "grad_norm": 1.1023990704721909,
      "learning_rate": 1.2338585829778687e-06,
      "loss": 0.5226,
      "step": 16621
    },
    {
      "epoch": 0.8969834331660461,
      "grad_norm": 1.025448883346239,
      "learning_rate": 1.2336159095606506e-06,
      "loss": 0.464,
      "step": 16622
    },
    {
      "epoch": 0.8970373967945605,
      "grad_norm": 1.3397719380793285,
      "learning_rate": 1.2333733587636218e-06,
      "loss": 0.723,
      "step": 16623
    },
    {
      "epoch": 0.8970913604230748,
      "grad_norm": 1.1292982508486942,
      "learning_rate": 1.2331309305937547e-06,
      "loss": 0.5367,
      "step": 16624
    },
    {
      "epoch": 0.8971453240515892,
      "grad_norm": 1.074915879893168,
      "learning_rate": 1.2328886250580154e-06,
      "loss": 0.5696,
      "step": 16625
    },
    {
      "epoch": 0.8971992876801036,
      "grad_norm": 0.8564645218629254,
      "learning_rate": 1.2326464421633683e-06,
      "loss": 0.3229,
      "step": 16626
    },
    {
      "epoch": 0.897253251308618,
      "grad_norm": 0.9520230570401239,
      "learning_rate": 1.2324043819167748e-06,
      "loss": 0.3929,
      "step": 16627
    },
    {
      "epoch": 0.8973072149371324,
      "grad_norm": 1.0420041235147786,
      "learning_rate": 1.232162444325191e-06,
      "loss": 0.4885,
      "step": 16628
    },
    {
      "epoch": 0.8973611785656468,
      "grad_norm": 0.7140381366790686,
      "learning_rate": 1.231920629395571e-06,
      "loss": 0.3258,
      "step": 16629
    },
    {
      "epoch": 0.8974151421941612,
      "grad_norm": 1.0047477850701425,
      "learning_rate": 1.2316789371348647e-06,
      "loss": 0.4146,
      "step": 16630
    },
    {
      "epoch": 0.8974691058226755,
      "grad_norm": 1.008172866486972,
      "learning_rate": 1.231437367550019e-06,
      "loss": 0.4609,
      "step": 16631
    },
    {
      "epoch": 0.8975230694511899,
      "grad_norm": 0.9191653075451771,
      "learning_rate": 1.2311959206479757e-06,
      "loss": 0.4168,
      "step": 16632
    },
    {
      "epoch": 0.8975770330797043,
      "grad_norm": 1.1092036903888478,
      "learning_rate": 1.2309545964356754e-06,
      "loss": 0.4981,
      "step": 16633
    },
    {
      "epoch": 0.8976309967082187,
      "grad_norm": 1.052883919466102,
      "learning_rate": 1.2307133949200533e-06,
      "loss": 0.4477,
      "step": 16634
    },
    {
      "epoch": 0.8976849603367331,
      "grad_norm": 1.020870035153527,
      "learning_rate": 1.230472316108042e-06,
      "loss": 0.5646,
      "step": 16635
    },
    {
      "epoch": 0.8977389239652475,
      "grad_norm": 1.277904283487938,
      "learning_rate": 1.230231360006571e-06,
      "loss": 0.6889,
      "step": 16636
    },
    {
      "epoch": 0.8977928875937619,
      "grad_norm": 1.0136179356200992,
      "learning_rate": 1.2299905266225642e-06,
      "loss": 0.4926,
      "step": 16637
    },
    {
      "epoch": 0.8978468512222761,
      "grad_norm": 0.9633269651853454,
      "learning_rate": 1.2297498159629454e-06,
      "loss": 0.4382,
      "step": 16638
    },
    {
      "epoch": 0.8979008148507905,
      "grad_norm": 0.9344173242507018,
      "learning_rate": 1.2295092280346304e-06,
      "loss": 0.3608,
      "step": 16639
    },
    {
      "epoch": 0.8979547784793049,
      "grad_norm": 0.893180567705878,
      "learning_rate": 1.2292687628445357e-06,
      "loss": 0.3469,
      "step": 16640
    },
    {
      "epoch": 0.8980087421078193,
      "grad_norm": 1.1845004225780589,
      "learning_rate": 1.2290284203995719e-06,
      "loss": 0.4748,
      "step": 16641
    },
    {
      "epoch": 0.8980627057363337,
      "grad_norm": 1.0018513288286304,
      "learning_rate": 1.228788200706647e-06,
      "loss": 0.5446,
      "step": 16642
    },
    {
      "epoch": 0.8981166693648481,
      "grad_norm": 0.9201558595620329,
      "learning_rate": 1.2285481037726644e-06,
      "loss": 0.3716,
      "step": 16643
    },
    {
      "epoch": 0.8981706329933625,
      "grad_norm": 1.1679579687150983,
      "learning_rate": 1.2283081296045256e-06,
      "loss": 0.5931,
      "step": 16644
    },
    {
      "epoch": 0.8982245966218768,
      "grad_norm": 1.0080828075109014,
      "learning_rate": 1.228068278209128e-06,
      "loss": 0.5802,
      "step": 16645
    },
    {
      "epoch": 0.8982785602503912,
      "grad_norm": 1.1866440329471075,
      "learning_rate": 1.2278285495933642e-06,
      "loss": 0.4731,
      "step": 16646
    },
    {
      "epoch": 0.8983325238789056,
      "grad_norm": 1.2356251508686347,
      "learning_rate": 1.227588943764124e-06,
      "loss": 0.5503,
      "step": 16647
    },
    {
      "epoch": 0.89838648750742,
      "grad_norm": 1.0169657124305187,
      "learning_rate": 1.227349460728295e-06,
      "loss": 0.4662,
      "step": 16648
    },
    {
      "epoch": 0.8984404511359344,
      "grad_norm": 1.1524975939463453,
      "learning_rate": 1.2271101004927594e-06,
      "loss": 0.4256,
      "step": 16649
    },
    {
      "epoch": 0.8984944147644488,
      "grad_norm": 1.036860463792441,
      "learning_rate": 1.226870863064397e-06,
      "loss": 0.4402,
      "step": 16650
    },
    {
      "epoch": 0.8985483783929631,
      "grad_norm": 0.9283860236441164,
      "learning_rate": 1.2266317484500847e-06,
      "loss": 0.4741,
      "step": 16651
    },
    {
      "epoch": 0.8986023420214775,
      "grad_norm": 1.037705740790301,
      "learning_rate": 1.2263927566566928e-06,
      "loss": 0.603,
      "step": 16652
    },
    {
      "epoch": 0.8986563056499919,
      "grad_norm": 1.3095954646334673,
      "learning_rate": 1.2261538876910913e-06,
      "loss": 0.5896,
      "step": 16653
    },
    {
      "epoch": 0.8987102692785063,
      "grad_norm": 1.0579842250108549,
      "learning_rate": 1.2259151415601454e-06,
      "loss": 0.5089,
      "step": 16654
    },
    {
      "epoch": 0.8987642329070207,
      "grad_norm": 1.153059921399229,
      "learning_rate": 1.2256765182707174e-06,
      "loss": 0.4582,
      "step": 16655
    },
    {
      "epoch": 0.8988181965355351,
      "grad_norm": 0.8833540060816297,
      "learning_rate": 1.2254380178296649e-06,
      "loss": 0.4287,
      "step": 16656
    },
    {
      "epoch": 0.8988721601640495,
      "grad_norm": 0.988157051319922,
      "learning_rate": 1.2251996402438427e-06,
      "loss": 0.4747,
      "step": 16657
    },
    {
      "epoch": 0.8989261237925638,
      "grad_norm": 0.9082953327580486,
      "learning_rate": 1.2249613855201033e-06,
      "loss": 0.3843,
      "step": 16658
    },
    {
      "epoch": 0.8989800874210782,
      "grad_norm": 0.9647390001788155,
      "learning_rate": 1.2247232536652924e-06,
      "loss": 0.4294,
      "step": 16659
    },
    {
      "epoch": 0.8990340510495926,
      "grad_norm": 1.1255155547550646,
      "learning_rate": 1.2244852446862548e-06,
      "loss": 0.4375,
      "step": 16660
    },
    {
      "epoch": 0.899088014678107,
      "grad_norm": 1.0764517322127314,
      "learning_rate": 1.2242473585898317e-06,
      "loss": 0.4759,
      "step": 16661
    },
    {
      "epoch": 0.8991419783066213,
      "grad_norm": 1.1629322726063878,
      "learning_rate": 1.2240095953828596e-06,
      "loss": 0.5314,
      "step": 16662
    },
    {
      "epoch": 0.8991959419351357,
      "grad_norm": 0.9282711611115064,
      "learning_rate": 1.2237719550721723e-06,
      "loss": 0.4152,
      "step": 16663
    },
    {
      "epoch": 0.8992499055636501,
      "grad_norm": 1.197274846773252,
      "learning_rate": 1.2235344376646006e-06,
      "loss": 0.5891,
      "step": 16664
    },
    {
      "epoch": 0.8993038691921644,
      "grad_norm": 1.1396567600240226,
      "learning_rate": 1.2232970431669692e-06,
      "loss": 0.5419,
      "step": 16665
    },
    {
      "epoch": 0.8993578328206788,
      "grad_norm": 1.0602268543486988,
      "learning_rate": 1.2230597715861022e-06,
      "loss": 0.4613,
      "step": 16666
    },
    {
      "epoch": 0.8994117964491932,
      "grad_norm": 0.9809160172553654,
      "learning_rate": 1.2228226229288198e-06,
      "loss": 0.4152,
      "step": 16667
    },
    {
      "epoch": 0.8994657600777076,
      "grad_norm": 1.1156612719428147,
      "learning_rate": 1.222585597201936e-06,
      "loss": 0.5685,
      "step": 16668
    },
    {
      "epoch": 0.899519723706222,
      "grad_norm": 1.334358576441543,
      "learning_rate": 1.2223486944122644e-06,
      "loss": 0.5226,
      "step": 16669
    },
    {
      "epoch": 0.8995736873347364,
      "grad_norm": 1.0222748285677334,
      "learning_rate": 1.2221119145666142e-06,
      "loss": 0.3418,
      "step": 16670
    },
    {
      "epoch": 0.8996276509632508,
      "grad_norm": 1.2810471455307793,
      "learning_rate": 1.2218752576717897e-06,
      "loss": 0.5283,
      "step": 16671
    },
    {
      "epoch": 0.8996816145917651,
      "grad_norm": 1.0409551710609215,
      "learning_rate": 1.221638723734593e-06,
      "loss": 0.4881,
      "step": 16672
    },
    {
      "epoch": 0.8997355782202795,
      "grad_norm": 1.2198758224019,
      "learning_rate": 1.2214023127618223e-06,
      "loss": 0.4445,
      "step": 16673
    },
    {
      "epoch": 0.8997895418487939,
      "grad_norm": 0.9752782855696729,
      "learning_rate": 1.2211660247602718e-06,
      "loss": 0.4459,
      "step": 16674
    },
    {
      "epoch": 0.8998435054773083,
      "grad_norm": 1.2029567147971472,
      "learning_rate": 1.2209298597367336e-06,
      "loss": 0.551,
      "step": 16675
    },
    {
      "epoch": 0.8998974691058227,
      "grad_norm": 0.9236222277284928,
      "learning_rate": 1.2206938176979953e-06,
      "loss": 0.4224,
      "step": 16676
    },
    {
      "epoch": 0.8999514327343371,
      "grad_norm": 1.043343136207993,
      "learning_rate": 1.2204578986508404e-06,
      "loss": 0.4164,
      "step": 16677
    },
    {
      "epoch": 0.9000053963628515,
      "grad_norm": 0.9348803678505607,
      "learning_rate": 1.2202221026020495e-06,
      "loss": 0.3914,
      "step": 16678
    },
    {
      "epoch": 0.9000593599913658,
      "grad_norm": 0.8943326144959638,
      "learning_rate": 1.2199864295584e-06,
      "loss": 0.4186,
      "step": 16679
    },
    {
      "epoch": 0.9001133236198802,
      "grad_norm": 0.9481628039620029,
      "learning_rate": 1.219750879526665e-06,
      "loss": 0.4207,
      "step": 16680
    },
    {
      "epoch": 0.9001672872483946,
      "grad_norm": 0.9327812003408481,
      "learning_rate": 1.2195154525136146e-06,
      "loss": 0.3761,
      "step": 16681
    },
    {
      "epoch": 0.900221250876909,
      "grad_norm": 0.937116890290192,
      "learning_rate": 1.2192801485260152e-06,
      "loss": 0.3913,
      "step": 16682
    },
    {
      "epoch": 0.9002752145054234,
      "grad_norm": 0.9716758981647424,
      "learning_rate": 1.2190449675706296e-06,
      "loss": 0.5011,
      "step": 16683
    },
    {
      "epoch": 0.9003291781339378,
      "grad_norm": 0.9829481874556457,
      "learning_rate": 1.2188099096542175e-06,
      "loss": 0.3641,
      "step": 16684
    },
    {
      "epoch": 0.9003831417624522,
      "grad_norm": 1.029847414136919,
      "learning_rate": 1.2185749747835338e-06,
      "loss": 0.3964,
      "step": 16685
    },
    {
      "epoch": 0.9004371053909664,
      "grad_norm": 1.2573203983310157,
      "learning_rate": 1.2183401629653324e-06,
      "loss": 0.4872,
      "step": 16686
    },
    {
      "epoch": 0.9004910690194808,
      "grad_norm": 1.1200858806749747,
      "learning_rate": 1.2181054742063603e-06,
      "loss": 0.6089,
      "step": 16687
    },
    {
      "epoch": 0.9005450326479952,
      "grad_norm": 1.087561655776066,
      "learning_rate": 1.2178709085133635e-06,
      "loss": 0.5319,
      "step": 16688
    },
    {
      "epoch": 0.9005989962765096,
      "grad_norm": 0.9477195107024905,
      "learning_rate": 1.2176364658930836e-06,
      "loss": 0.338,
      "step": 16689
    },
    {
      "epoch": 0.900652959905024,
      "grad_norm": 1.0095256887997281,
      "learning_rate": 1.217402146352259e-06,
      "loss": 0.4237,
      "step": 16690
    },
    {
      "epoch": 0.9007069235335384,
      "grad_norm": 1.0804802081211164,
      "learning_rate": 1.2171679498976231e-06,
      "loss": 0.5543,
      "step": 16691
    },
    {
      "epoch": 0.9007608871620528,
      "grad_norm": 1.1574384358259182,
      "learning_rate": 1.2169338765359087e-06,
      "loss": 0.515,
      "step": 16692
    },
    {
      "epoch": 0.9008148507905671,
      "grad_norm": 0.8851573826354238,
      "learning_rate": 1.2166999262738422e-06,
      "loss": 0.3283,
      "step": 16693
    },
    {
      "epoch": 0.9008688144190815,
      "grad_norm": 1.148451787502722,
      "learning_rate": 1.2164660991181478e-06,
      "loss": 0.402,
      "step": 16694
    },
    {
      "epoch": 0.9009227780475959,
      "grad_norm": 1.474635778767568,
      "learning_rate": 1.2162323950755458e-06,
      "loss": 0.6205,
      "step": 16695
    },
    {
      "epoch": 0.9009767416761103,
      "grad_norm": 1.3058334604106872,
      "learning_rate": 1.2159988141527529e-06,
      "loss": 0.5471,
      "step": 16696
    },
    {
      "epoch": 0.9010307053046247,
      "grad_norm": 0.9146713661810555,
      "learning_rate": 1.2157653563564831e-06,
      "loss": 0.4147,
      "step": 16697
    },
    {
      "epoch": 0.9010846689331391,
      "grad_norm": 1.1984015135542683,
      "learning_rate": 1.2155320216934454e-06,
      "loss": 0.5521,
      "step": 16698
    },
    {
      "epoch": 0.9011386325616535,
      "grad_norm": 0.9906711173075657,
      "learning_rate": 1.215298810170347e-06,
      "loss": 0.4888,
      "step": 16699
    },
    {
      "epoch": 0.9011925961901678,
      "grad_norm": 0.9786313126934569,
      "learning_rate": 1.2150657217938908e-06,
      "loss": 0.5349,
      "step": 16700
    },
    {
      "epoch": 0.9012465598186822,
      "grad_norm": 1.240289660254431,
      "learning_rate": 1.2148327565707742e-06,
      "loss": 0.5487,
      "step": 16701
    },
    {
      "epoch": 0.9013005234471966,
      "grad_norm": 1.1775805606210468,
      "learning_rate": 1.2145999145076945e-06,
      "loss": 0.5929,
      "step": 16702
    },
    {
      "epoch": 0.901354487075711,
      "grad_norm": 1.0084077787543544,
      "learning_rate": 1.2143671956113434e-06,
      "loss": 0.5914,
      "step": 16703
    },
    {
      "epoch": 0.9014084507042254,
      "grad_norm": 0.8310169807351301,
      "learning_rate": 1.2141345998884092e-06,
      "loss": 0.2939,
      "step": 16704
    },
    {
      "epoch": 0.9014624143327398,
      "grad_norm": 1.2796960252357983,
      "learning_rate": 1.2139021273455773e-06,
      "loss": 0.597,
      "step": 16705
    },
    {
      "epoch": 0.9015163779612542,
      "grad_norm": 0.9635007877309409,
      "learning_rate": 1.213669777989529e-06,
      "loss": 0.5354,
      "step": 16706
    },
    {
      "epoch": 0.9015703415897685,
      "grad_norm": 1.052478715631544,
      "learning_rate": 1.2134375518269429e-06,
      "loss": 0.4678,
      "step": 16707
    },
    {
      "epoch": 0.9016243052182829,
      "grad_norm": 1.0066414619194994,
      "learning_rate": 1.213205448864492e-06,
      "loss": 0.5239,
      "step": 16708
    },
    {
      "epoch": 0.9016782688467972,
      "grad_norm": 1.1865257244174428,
      "learning_rate": 1.2129734691088483e-06,
      "loss": 0.6521,
      "step": 16709
    },
    {
      "epoch": 0.9017322324753116,
      "grad_norm": 0.9675514312992051,
      "learning_rate": 1.2127416125666786e-06,
      "loss": 0.3917,
      "step": 16710
    },
    {
      "epoch": 0.901786196103826,
      "grad_norm": 1.247112586659996,
      "learning_rate": 1.2125098792446468e-06,
      "loss": 0.702,
      "step": 16711
    },
    {
      "epoch": 0.9018401597323404,
      "grad_norm": 1.000971853570305,
      "learning_rate": 1.2122782691494137e-06,
      "loss": 0.4259,
      "step": 16712
    },
    {
      "epoch": 0.9018941233608548,
      "grad_norm": 1.0897955521408966,
      "learning_rate": 1.2120467822876353e-06,
      "loss": 0.5454,
      "step": 16713
    },
    {
      "epoch": 0.9019480869893691,
      "grad_norm": 0.9607833471185613,
      "learning_rate": 1.2118154186659653e-06,
      "loss": 0.3562,
      "step": 16714
    },
    {
      "epoch": 0.9020020506178835,
      "grad_norm": 1.1886920083458952,
      "learning_rate": 1.211584178291053e-06,
      "loss": 0.4838,
      "step": 16715
    },
    {
      "epoch": 0.9020560142463979,
      "grad_norm": 0.8666227219347404,
      "learning_rate": 1.2113530611695443e-06,
      "loss": 0.4211,
      "step": 16716
    },
    {
      "epoch": 0.9021099778749123,
      "grad_norm": 1.0676656350972396,
      "learning_rate": 1.211122067308082e-06,
      "loss": 0.4639,
      "step": 16717
    },
    {
      "epoch": 0.9021639415034267,
      "grad_norm": 1.0420219671031778,
      "learning_rate": 1.2108911967133052e-06,
      "loss": 0.4714,
      "step": 16718
    },
    {
      "epoch": 0.9022179051319411,
      "grad_norm": 0.9793702142375112,
      "learning_rate": 1.210660449391849e-06,
      "loss": 0.4423,
      "step": 16719
    },
    {
      "epoch": 0.9022718687604554,
      "grad_norm": 0.9971024162901069,
      "learning_rate": 1.2104298253503461e-06,
      "loss": 0.3833,
      "step": 16720
    },
    {
      "epoch": 0.9023258323889698,
      "grad_norm": 1.100232205796732,
      "learning_rate": 1.2101993245954235e-06,
      "loss": 0.5271,
      "step": 16721
    },
    {
      "epoch": 0.9023797960174842,
      "grad_norm": 1.0717112133471145,
      "learning_rate": 1.2099689471337077e-06,
      "loss": 0.4019,
      "step": 16722
    },
    {
      "epoch": 0.9024337596459986,
      "grad_norm": 1.0975203890941791,
      "learning_rate": 1.2097386929718185e-06,
      "loss": 0.5274,
      "step": 16723
    },
    {
      "epoch": 0.902487723274513,
      "grad_norm": 1.1832403109950573,
      "learning_rate": 1.2095085621163747e-06,
      "loss": 0.6526,
      "step": 16724
    },
    {
      "epoch": 0.9025416869030274,
      "grad_norm": 0.8650952522058243,
      "learning_rate": 1.2092785545739893e-06,
      "loss": 0.3577,
      "step": 16725
    },
    {
      "epoch": 0.9025956505315418,
      "grad_norm": 0.9758424292926259,
      "learning_rate": 1.209048670351274e-06,
      "loss": 0.392,
      "step": 16726
    },
    {
      "epoch": 0.9026496141600561,
      "grad_norm": 0.8790984968824037,
      "learning_rate": 1.2088189094548366e-06,
      "loss": 0.3699,
      "step": 16727
    },
    {
      "epoch": 0.9027035777885705,
      "grad_norm": 1.1532803703071761,
      "learning_rate": 1.2085892718912789e-06,
      "loss": 0.4486,
      "step": 16728
    },
    {
      "epoch": 0.9027575414170849,
      "grad_norm": 1.1126263273588575,
      "learning_rate": 1.2083597576672019e-06,
      "loss": 0.5473,
      "step": 16729
    },
    {
      "epoch": 0.9028115050455993,
      "grad_norm": 1.0358348190920725,
      "learning_rate": 1.2081303667892013e-06,
      "loss": 0.4508,
      "step": 16730
    },
    {
      "epoch": 0.9028654686741137,
      "grad_norm": 1.1368768306323236,
      "learning_rate": 1.2079010992638713e-06,
      "loss": 0.5042,
      "step": 16731
    },
    {
      "epoch": 0.9029194323026281,
      "grad_norm": 1.1796211875784193,
      "learning_rate": 1.2076719550977998e-06,
      "loss": 0.406,
      "step": 16732
    },
    {
      "epoch": 0.9029733959311425,
      "grad_norm": 1.0256604864613124,
      "learning_rate": 1.2074429342975737e-06,
      "loss": 0.4966,
      "step": 16733
    },
    {
      "epoch": 0.9030273595596567,
      "grad_norm": 1.0112261493181969,
      "learning_rate": 1.2072140368697757e-06,
      "loss": 0.509,
      "step": 16734
    },
    {
      "epoch": 0.9030813231881711,
      "grad_norm": 1.0453950935088143,
      "learning_rate": 1.2069852628209832e-06,
      "loss": 0.4298,
      "step": 16735
    },
    {
      "epoch": 0.9031352868166855,
      "grad_norm": 0.9053106866789684,
      "learning_rate": 1.2067566121577715e-06,
      "loss": 0.3341,
      "step": 16736
    },
    {
      "epoch": 0.9031892504451999,
      "grad_norm": 1.0596651281974154,
      "learning_rate": 1.2065280848867133e-06,
      "loss": 0.5446,
      "step": 16737
    },
    {
      "epoch": 0.9032432140737143,
      "grad_norm": 0.9527883659564679,
      "learning_rate": 1.2062996810143763e-06,
      "loss": 0.4473,
      "step": 16738
    },
    {
      "epoch": 0.9032971777022287,
      "grad_norm": 1.1710613258362486,
      "learning_rate": 1.2060714005473247e-06,
      "loss": 0.4834,
      "step": 16739
    },
    {
      "epoch": 0.9033511413307431,
      "grad_norm": 0.9219973479180453,
      "learning_rate": 1.2058432434921196e-06,
      "loss": 0.3612,
      "step": 16740
    },
    {
      "epoch": 0.9034051049592574,
      "grad_norm": 0.8073586153969352,
      "learning_rate": 1.2056152098553194e-06,
      "loss": 0.3163,
      "step": 16741
    },
    {
      "epoch": 0.9034590685877718,
      "grad_norm": 1.0377857267467192,
      "learning_rate": 1.2053872996434769e-06,
      "loss": 0.3817,
      "step": 16742
    },
    {
      "epoch": 0.9035130322162862,
      "grad_norm": 0.859846617526315,
      "learning_rate": 1.2051595128631425e-06,
      "loss": 0.4129,
      "step": 16743
    },
    {
      "epoch": 0.9035669958448006,
      "grad_norm": 0.9955365089079901,
      "learning_rate": 1.2049318495208633e-06,
      "loss": 0.4412,
      "step": 16744
    },
    {
      "epoch": 0.903620959473315,
      "grad_norm": 1.0889744374520482,
      "learning_rate": 1.2047043096231826e-06,
      "loss": 0.4781,
      "step": 16745
    },
    {
      "epoch": 0.9036749231018294,
      "grad_norm": 1.1119264924734953,
      "learning_rate": 1.2044768931766396e-06,
      "loss": 0.451,
      "step": 16746
    },
    {
      "epoch": 0.9037288867303438,
      "grad_norm": 1.0173120815631487,
      "learning_rate": 1.204249600187772e-06,
      "loss": 0.55,
      "step": 16747
    },
    {
      "epoch": 0.9037828503588581,
      "grad_norm": 1.10126709178847,
      "learning_rate": 1.2040224306631115e-06,
      "loss": 0.3977,
      "step": 16748
    },
    {
      "epoch": 0.9038368139873725,
      "grad_norm": 0.9398889817332698,
      "learning_rate": 1.2037953846091863e-06,
      "loss": 0.377,
      "step": 16749
    },
    {
      "epoch": 0.9038907776158869,
      "grad_norm": 1.0322759814475428,
      "learning_rate": 1.2035684620325235e-06,
      "loss": 0.3974,
      "step": 16750
    },
    {
      "epoch": 0.9039447412444013,
      "grad_norm": 1.212783386839925,
      "learning_rate": 1.2033416629396438e-06,
      "loss": 0.3724,
      "step": 16751
    },
    {
      "epoch": 0.9039987048729157,
      "grad_norm": 1.0616762878345487,
      "learning_rate": 1.2031149873370662e-06,
      "loss": 0.4393,
      "step": 16752
    },
    {
      "epoch": 0.9040526685014301,
      "grad_norm": 1.1173407057526572,
      "learning_rate": 1.2028884352313052e-06,
      "loss": 0.4926,
      "step": 16753
    },
    {
      "epoch": 0.9041066321299445,
      "grad_norm": 1.0621731804999577,
      "learning_rate": 1.202662006628873e-06,
      "loss": 0.4506,
      "step": 16754
    },
    {
      "epoch": 0.9041605957584588,
      "grad_norm": 1.0907736599534423,
      "learning_rate": 1.2024357015362775e-06,
      "loss": 0.4617,
      "step": 16755
    },
    {
      "epoch": 0.9042145593869731,
      "grad_norm": 1.0969531438374895,
      "learning_rate": 1.2022095199600212e-06,
      "loss": 0.5341,
      "step": 16756
    },
    {
      "epoch": 0.9042685230154875,
      "grad_norm": 1.0061623871393737,
      "learning_rate": 1.2019834619066064e-06,
      "loss": 0.4644,
      "step": 16757
    },
    {
      "epoch": 0.9043224866440019,
      "grad_norm": 1.275793234882058,
      "learning_rate": 1.2017575273825297e-06,
      "loss": 0.5038,
      "step": 16758
    },
    {
      "epoch": 0.9043764502725163,
      "grad_norm": 1.1151777687604314,
      "learning_rate": 1.2015317163942842e-06,
      "loss": 0.6202,
      "step": 16759
    },
    {
      "epoch": 0.9044304139010307,
      "grad_norm": 0.9917883550645258,
      "learning_rate": 1.201306028948361e-06,
      "loss": 0.4203,
      "step": 16760
    },
    {
      "epoch": 0.9044843775295451,
      "grad_norm": 0.898235738604466,
      "learning_rate": 1.201080465051246e-06,
      "loss": 0.3784,
      "step": 16761
    },
    {
      "epoch": 0.9045383411580594,
      "grad_norm": 0.8981332052897137,
      "learning_rate": 1.2008550247094224e-06,
      "loss": 0.3393,
      "step": 16762
    },
    {
      "epoch": 0.9045923047865738,
      "grad_norm": 0.9579351308154923,
      "learning_rate": 1.2006297079293695e-06,
      "loss": 0.4044,
      "step": 16763
    },
    {
      "epoch": 0.9046462684150882,
      "grad_norm": 0.9976494898412256,
      "learning_rate": 1.2004045147175623e-06,
      "loss": 0.4348,
      "step": 16764
    },
    {
      "epoch": 0.9047002320436026,
      "grad_norm": 0.9964891272790907,
      "learning_rate": 1.2001794450804739e-06,
      "loss": 0.5737,
      "step": 16765
    },
    {
      "epoch": 0.904754195672117,
      "grad_norm": 1.0379170980123198,
      "learning_rate": 1.199954499024573e-06,
      "loss": 0.5533,
      "step": 16766
    },
    {
      "epoch": 0.9048081593006314,
      "grad_norm": 1.2664982818860355,
      "learning_rate": 1.1997296765563243e-06,
      "loss": 0.6408,
      "step": 16767
    },
    {
      "epoch": 0.9048621229291458,
      "grad_norm": 1.1331017074317575,
      "learning_rate": 1.1995049776821907e-06,
      "loss": 0.4041,
      "step": 16768
    },
    {
      "epoch": 0.9049160865576601,
      "grad_norm": 0.9183550394211663,
      "learning_rate": 1.1992804024086288e-06,
      "loss": 0.4577,
      "step": 16769
    },
    {
      "epoch": 0.9049700501861745,
      "grad_norm": 1.108108244799261,
      "learning_rate": 1.1990559507420934e-06,
      "loss": 0.4312,
      "step": 16770
    },
    {
      "epoch": 0.9050240138146889,
      "grad_norm": 0.7854628099704745,
      "learning_rate": 1.198831622689036e-06,
      "loss": 0.3901,
      "step": 16771
    },
    {
      "epoch": 0.9050779774432033,
      "grad_norm": 0.9512897334411671,
      "learning_rate": 1.1986074182559038e-06,
      "loss": 0.4343,
      "step": 16772
    },
    {
      "epoch": 0.9051319410717177,
      "grad_norm": 1.1171915817025302,
      "learning_rate": 1.198383337449141e-06,
      "loss": 0.4723,
      "step": 16773
    },
    {
      "epoch": 0.9051859047002321,
      "grad_norm": 0.8471324655079376,
      "learning_rate": 1.1981593802751869e-06,
      "loss": 0.3513,
      "step": 16774
    },
    {
      "epoch": 0.9052398683287465,
      "grad_norm": 1.1764630277106005,
      "learning_rate": 1.1979355467404794e-06,
      "loss": 0.3686,
      "step": 16775
    },
    {
      "epoch": 0.9052938319572608,
      "grad_norm": 0.8863528402103076,
      "learning_rate": 1.1977118368514519e-06,
      "loss": 0.36,
      "step": 16776
    },
    {
      "epoch": 0.9053477955857752,
      "grad_norm": 1.0642932794287812,
      "learning_rate": 1.1974882506145325e-06,
      "loss": 0.4999,
      "step": 16777
    },
    {
      "epoch": 0.9054017592142896,
      "grad_norm": 0.9346699779231484,
      "learning_rate": 1.197264788036148e-06,
      "loss": 0.3671,
      "step": 16778
    },
    {
      "epoch": 0.905455722842804,
      "grad_norm": 1.059855853203582,
      "learning_rate": 1.1970414491227214e-06,
      "loss": 0.4463,
      "step": 16779
    },
    {
      "epoch": 0.9055096864713184,
      "grad_norm": 1.342374004052335,
      "learning_rate": 1.1968182338806713e-06,
      "loss": 0.4931,
      "step": 16780
    },
    {
      "epoch": 0.9055636500998328,
      "grad_norm": 1.1131122888168414,
      "learning_rate": 1.1965951423164128e-06,
      "loss": 0.444,
      "step": 16781
    },
    {
      "epoch": 0.9056176137283471,
      "grad_norm": 1.0104393039296322,
      "learning_rate": 1.1963721744363587e-06,
      "loss": 0.4099,
      "step": 16782
    },
    {
      "epoch": 0.9056715773568614,
      "grad_norm": 0.9736517005836665,
      "learning_rate": 1.1961493302469172e-06,
      "loss": 0.4052,
      "step": 16783
    },
    {
      "epoch": 0.9057255409853758,
      "grad_norm": 0.9619747454262176,
      "learning_rate": 1.195926609754492e-06,
      "loss": 0.4523,
      "step": 16784
    },
    {
      "epoch": 0.9057795046138902,
      "grad_norm": 1.0270582073863601,
      "learning_rate": 1.1957040129654852e-06,
      "loss": 0.5369,
      "step": 16785
    },
    {
      "epoch": 0.9058334682424046,
      "grad_norm": 1.0933902127379427,
      "learning_rate": 1.1954815398862936e-06,
      "loss": 0.4323,
      "step": 16786
    },
    {
      "epoch": 0.905887431870919,
      "grad_norm": 0.8653221044850875,
      "learning_rate": 1.1952591905233126e-06,
      "loss": 0.3149,
      "step": 16787
    },
    {
      "epoch": 0.9059413954994334,
      "grad_norm": 1.1813095963737361,
      "learning_rate": 1.1950369648829324e-06,
      "loss": 0.4578,
      "step": 16788
    },
    {
      "epoch": 0.9059953591279477,
      "grad_norm": 0.9529122289285663,
      "learning_rate": 1.1948148629715393e-06,
      "loss": 0.3778,
      "step": 16789
    },
    {
      "epoch": 0.9060493227564621,
      "grad_norm": 1.0443828436900413,
      "learning_rate": 1.1945928847955174e-06,
      "loss": 0.5184,
      "step": 16790
    },
    {
      "epoch": 0.9061032863849765,
      "grad_norm": 1.0880913420430582,
      "learning_rate": 1.1943710303612459e-06,
      "loss": 0.5902,
      "step": 16791
    },
    {
      "epoch": 0.9061572500134909,
      "grad_norm": 1.085663411744397,
      "learning_rate": 1.1941492996751015e-06,
      "loss": 0.3788,
      "step": 16792
    },
    {
      "epoch": 0.9062112136420053,
      "grad_norm": 0.8516716658693528,
      "learning_rate": 1.1939276927434567e-06,
      "loss": 0.3165,
      "step": 16793
    },
    {
      "epoch": 0.9062651772705197,
      "grad_norm": 1.2044128902721896,
      "learning_rate": 1.1937062095726818e-06,
      "loss": 0.5981,
      "step": 16794
    },
    {
      "epoch": 0.9063191408990341,
      "grad_norm": 1.0289571109223823,
      "learning_rate": 1.193484850169141e-06,
      "loss": 0.4018,
      "step": 16795
    },
    {
      "epoch": 0.9063731045275484,
      "grad_norm": 0.9033119282200087,
      "learning_rate": 1.1932636145391977e-06,
      "loss": 0.4905,
      "step": 16796
    },
    {
      "epoch": 0.9064270681560628,
      "grad_norm": 0.9417746886274232,
      "learning_rate": 1.1930425026892092e-06,
      "loss": 0.3652,
      "step": 16797
    },
    {
      "epoch": 0.9064810317845772,
      "grad_norm": 0.9618644943975591,
      "learning_rate": 1.192821514625531e-06,
      "loss": 0.5013,
      "step": 16798
    },
    {
      "epoch": 0.9065349954130916,
      "grad_norm": 0.8289480810942114,
      "learning_rate": 1.1926006503545147e-06,
      "loss": 0.3175,
      "step": 16799
    },
    {
      "epoch": 0.906588959041606,
      "grad_norm": 1.0260893544658707,
      "learning_rate": 1.1923799098825076e-06,
      "loss": 0.523,
      "step": 16800
    },
    {
      "epoch": 0.9066429226701204,
      "grad_norm": 1.0530980186913874,
      "learning_rate": 1.192159293215855e-06,
      "loss": 0.4332,
      "step": 16801
    },
    {
      "epoch": 0.9066968862986348,
      "grad_norm": 1.1206235478683007,
      "learning_rate": 1.1919388003608965e-06,
      "loss": 0.5926,
      "step": 16802
    },
    {
      "epoch": 0.906750849927149,
      "grad_norm": 1.0761510979215079,
      "learning_rate": 1.1917184313239706e-06,
      "loss": 0.4463,
      "step": 16803
    },
    {
      "epoch": 0.9068048135556634,
      "grad_norm": 1.4373580615063613,
      "learning_rate": 1.19149818611141e-06,
      "loss": 0.6423,
      "step": 16804
    },
    {
      "epoch": 0.9068587771841778,
      "grad_norm": 0.8710218384759507,
      "learning_rate": 1.1912780647295444e-06,
      "loss": 0.3943,
      "step": 16805
    },
    {
      "epoch": 0.9069127408126922,
      "grad_norm": 0.8941716909841638,
      "learning_rate": 1.1910580671847014e-06,
      "loss": 0.3958,
      "step": 16806
    },
    {
      "epoch": 0.9069667044412066,
      "grad_norm": 1.0866830078220886,
      "learning_rate": 1.1908381934832031e-06,
      "loss": 0.5705,
      "step": 16807
    },
    {
      "epoch": 0.907020668069721,
      "grad_norm": 1.1046454934114835,
      "learning_rate": 1.1906184436313694e-06,
      "loss": 0.4645,
      "step": 16808
    },
    {
      "epoch": 0.9070746316982354,
      "grad_norm": 1.0912570530602865,
      "learning_rate": 1.1903988176355156e-06,
      "loss": 0.4093,
      "step": 16809
    },
    {
      "epoch": 0.9071285953267497,
      "grad_norm": 0.9762157369650749,
      "learning_rate": 1.1901793155019554e-06,
      "loss": 0.3816,
      "step": 16810
    },
    {
      "epoch": 0.9071825589552641,
      "grad_norm": 0.8887553835459238,
      "learning_rate": 1.1899599372369952e-06,
      "loss": 0.3764,
      "step": 16811
    },
    {
      "epoch": 0.9072365225837785,
      "grad_norm": 0.8205506671709637,
      "learning_rate": 1.1897406828469417e-06,
      "loss": 0.3849,
      "step": 16812
    },
    {
      "epoch": 0.9072904862122929,
      "grad_norm": 0.7259338815642064,
      "learning_rate": 1.1895215523380962e-06,
      "loss": 0.2773,
      "step": 16813
    },
    {
      "epoch": 0.9073444498408073,
      "grad_norm": 0.9871838753333931,
      "learning_rate": 1.1893025457167567e-06,
      "loss": 0.4494,
      "step": 16814
    },
    {
      "epoch": 0.9073984134693217,
      "grad_norm": 1.0315884644247137,
      "learning_rate": 1.1890836629892178e-06,
      "loss": 0.5172,
      "step": 16815
    },
    {
      "epoch": 0.9074523770978361,
      "grad_norm": 0.9084785595970151,
      "learning_rate": 1.18886490416177e-06,
      "loss": 0.3586,
      "step": 16816
    },
    {
      "epoch": 0.9075063407263504,
      "grad_norm": 0.9975832576333824,
      "learning_rate": 1.1886462692407015e-06,
      "loss": 0.5283,
      "step": 16817
    },
    {
      "epoch": 0.9075603043548648,
      "grad_norm": 1.2246729654002966,
      "learning_rate": 1.188427758232295e-06,
      "loss": 0.6909,
      "step": 16818
    },
    {
      "epoch": 0.9076142679833792,
      "grad_norm": 0.9187406787658606,
      "learning_rate": 1.1882093711428317e-06,
      "loss": 0.3021,
      "step": 16819
    },
    {
      "epoch": 0.9076682316118936,
      "grad_norm": 0.9050591895090271,
      "learning_rate": 1.187991107978588e-06,
      "loss": 0.4473,
      "step": 16820
    },
    {
      "epoch": 0.907722195240408,
      "grad_norm": 0.874098640939496,
      "learning_rate": 1.1877729687458365e-06,
      "loss": 0.3799,
      "step": 16821
    },
    {
      "epoch": 0.9077761588689224,
      "grad_norm": 0.9597996847441804,
      "learning_rate": 1.1875549534508469e-06,
      "loss": 0.4021,
      "step": 16822
    },
    {
      "epoch": 0.9078301224974368,
      "grad_norm": 1.1547451714294281,
      "learning_rate": 1.1873370620998853e-06,
      "loss": 0.5515,
      "step": 16823
    },
    {
      "epoch": 0.9078840861259511,
      "grad_norm": 1.083522615765843,
      "learning_rate": 1.1871192946992142e-06,
      "loss": 0.4589,
      "step": 16824
    },
    {
      "epoch": 0.9079380497544655,
      "grad_norm": 1.1088767500850647,
      "learning_rate": 1.1869016512550922e-06,
      "loss": 0.511,
      "step": 16825
    },
    {
      "epoch": 0.9079920133829799,
      "grad_norm": 1.0033789473205046,
      "learning_rate": 1.1866841317737753e-06,
      "loss": 0.5229,
      "step": 16826
    },
    {
      "epoch": 0.9080459770114943,
      "grad_norm": 1.1136461941909117,
      "learning_rate": 1.1864667362615143e-06,
      "loss": 0.4669,
      "step": 16827
    },
    {
      "epoch": 0.9080999406400087,
      "grad_norm": 0.990556255757785,
      "learning_rate": 1.186249464724558e-06,
      "loss": 0.4876,
      "step": 16828
    },
    {
      "epoch": 0.908153904268523,
      "grad_norm": 0.9729339178517875,
      "learning_rate": 1.186032317169151e-06,
      "loss": 0.5033,
      "step": 16829
    },
    {
      "epoch": 0.9082078678970374,
      "grad_norm": 1.1091273458290871,
      "learning_rate": 1.1858152936015336e-06,
      "loss": 0.4183,
      "step": 16830
    },
    {
      "epoch": 0.9082618315255517,
      "grad_norm": 1.403796230703734,
      "learning_rate": 1.1855983940279447e-06,
      "loss": 0.5139,
      "step": 16831
    },
    {
      "epoch": 0.9083157951540661,
      "grad_norm": 1.1266710870467602,
      "learning_rate": 1.185381618454617e-06,
      "loss": 0.5037,
      "step": 16832
    },
    {
      "epoch": 0.9083697587825805,
      "grad_norm": 1.0638481160418614,
      "learning_rate": 1.1851649668877812e-06,
      "loss": 0.4482,
      "step": 16833
    },
    {
      "epoch": 0.9084237224110949,
      "grad_norm": 0.8032821170037179,
      "learning_rate": 1.1849484393336636e-06,
      "loss": 0.3429,
      "step": 16834
    },
    {
      "epoch": 0.9084776860396093,
      "grad_norm": 1.1617278888622353,
      "learning_rate": 1.1847320357984881e-06,
      "loss": 0.4578,
      "step": 16835
    },
    {
      "epoch": 0.9085316496681237,
      "grad_norm": 1.2370079234726632,
      "learning_rate": 1.1845157562884743e-06,
      "loss": 0.4756,
      "step": 16836
    },
    {
      "epoch": 0.9085856132966381,
      "grad_norm": 1.0364067968823278,
      "learning_rate": 1.184299600809838e-06,
      "loss": 0.2732,
      "step": 16837
    },
    {
      "epoch": 0.9086395769251524,
      "grad_norm": 0.8736544656034154,
      "learning_rate": 1.1840835693687927e-06,
      "loss": 0.3894,
      "step": 16838
    },
    {
      "epoch": 0.9086935405536668,
      "grad_norm": 1.0567370029865544,
      "learning_rate": 1.1838676619715458e-06,
      "loss": 0.4961,
      "step": 16839
    },
    {
      "epoch": 0.9087475041821812,
      "grad_norm": 1.050688817949281,
      "learning_rate": 1.1836518786243034e-06,
      "loss": 0.3375,
      "step": 16840
    },
    {
      "epoch": 0.9088014678106956,
      "grad_norm": 1.1075523722116583,
      "learning_rate": 1.1834362193332676e-06,
      "loss": 0.5446,
      "step": 16841
    },
    {
      "epoch": 0.90885543143921,
      "grad_norm": 1.3545956915177442,
      "learning_rate": 1.183220684104636e-06,
      "loss": 0.4044,
      "step": 16842
    },
    {
      "epoch": 0.9089093950677244,
      "grad_norm": 1.1250805211705208,
      "learning_rate": 1.183005272944604e-06,
      "loss": 0.4315,
      "step": 16843
    },
    {
      "epoch": 0.9089633586962388,
      "grad_norm": 0.8979991980791777,
      "learning_rate": 1.1827899858593625e-06,
      "loss": 0.3624,
      "step": 16844
    },
    {
      "epoch": 0.9090173223247531,
      "grad_norm": 0.9611009031380302,
      "learning_rate": 1.1825748228550995e-06,
      "loss": 0.3663,
      "step": 16845
    },
    {
      "epoch": 0.9090712859532675,
      "grad_norm": 0.9806338837117654,
      "learning_rate": 1.182359783937998e-06,
      "loss": 0.4423,
      "step": 16846
    },
    {
      "epoch": 0.9091252495817819,
      "grad_norm": 1.195829841147055,
      "learning_rate": 1.1821448691142388e-06,
      "loss": 0.6147,
      "step": 16847
    },
    {
      "epoch": 0.9091792132102963,
      "grad_norm": 0.9858585830501105,
      "learning_rate": 1.1819300783899989e-06,
      "loss": 0.4594,
      "step": 16848
    },
    {
      "epoch": 0.9092331768388107,
      "grad_norm": 1.3454615683381237,
      "learning_rate": 1.181715411771452e-06,
      "loss": 0.5912,
      "step": 16849
    },
    {
      "epoch": 0.9092871404673251,
      "grad_norm": 1.0998985644713724,
      "learning_rate": 1.1815008692647672e-06,
      "loss": 0.4943,
      "step": 16850
    },
    {
      "epoch": 0.9093411040958395,
      "grad_norm": 0.9919698531777323,
      "learning_rate": 1.181286450876111e-06,
      "loss": 0.366,
      "step": 16851
    },
    {
      "epoch": 0.9093950677243537,
      "grad_norm": 1.1263026359526167,
      "learning_rate": 1.1810721566116465e-06,
      "loss": 0.6189,
      "step": 16852
    },
    {
      "epoch": 0.9094490313528681,
      "grad_norm": 1.1300459543833647,
      "learning_rate": 1.1808579864775318e-06,
      "loss": 0.5346,
      "step": 16853
    },
    {
      "epoch": 0.9095029949813825,
      "grad_norm": 0.9666786823740185,
      "learning_rate": 1.1806439404799224e-06,
      "loss": 0.4578,
      "step": 16854
    },
    {
      "epoch": 0.9095569586098969,
      "grad_norm": 1.0499863825496538,
      "learning_rate": 1.1804300186249706e-06,
      "loss": 0.4679,
      "step": 16855
    },
    {
      "epoch": 0.9096109222384113,
      "grad_norm": 1.1824737839625958,
      "learning_rate": 1.1802162209188248e-06,
      "loss": 0.5694,
      "step": 16856
    },
    {
      "epoch": 0.9096648858669257,
      "grad_norm": 1.134878762070186,
      "learning_rate": 1.1800025473676296e-06,
      "loss": 0.4552,
      "step": 16857
    },
    {
      "epoch": 0.90971884949544,
      "grad_norm": 0.9043992787089347,
      "learning_rate": 1.179788997977527e-06,
      "loss": 0.4184,
      "step": 16858
    },
    {
      "epoch": 0.9097728131239544,
      "grad_norm": 0.7583470576267999,
      "learning_rate": 1.1795755727546528e-06,
      "loss": 0.3305,
      "step": 16859
    },
    {
      "epoch": 0.9098267767524688,
      "grad_norm": 1.0433820775080942,
      "learning_rate": 1.1793622717051426e-06,
      "loss": 0.4983,
      "step": 16860
    },
    {
      "epoch": 0.9098807403809832,
      "grad_norm": 1.0227055987259839,
      "learning_rate": 1.179149094835126e-06,
      "loss": 0.3968,
      "step": 16861
    },
    {
      "epoch": 0.9099347040094976,
      "grad_norm": 1.080768203351878,
      "learning_rate": 1.1789360421507305e-06,
      "loss": 0.407,
      "step": 16862
    },
    {
      "epoch": 0.909988667638012,
      "grad_norm": 1.0179116995892183,
      "learning_rate": 1.1787231136580793e-06,
      "loss": 0.4314,
      "step": 16863
    },
    {
      "epoch": 0.9100426312665264,
      "grad_norm": 1.051107876174947,
      "learning_rate": 1.178510309363292e-06,
      "loss": 0.4054,
      "step": 16864
    },
    {
      "epoch": 0.9100965948950407,
      "grad_norm": 0.928022803155768,
      "learning_rate": 1.1782976292724858e-06,
      "loss": 0.464,
      "step": 16865
    },
    {
      "epoch": 0.9101505585235551,
      "grad_norm": 1.108490965968445,
      "learning_rate": 1.178085073391772e-06,
      "loss": 0.5364,
      "step": 16866
    },
    {
      "epoch": 0.9102045221520695,
      "grad_norm": 1.3798449481006136,
      "learning_rate": 1.1778726417272599e-06,
      "loss": 0.4796,
      "step": 16867
    },
    {
      "epoch": 0.9102584857805839,
      "grad_norm": 0.9918125094895963,
      "learning_rate": 1.177660334285055e-06,
      "loss": 0.3887,
      "step": 16868
    },
    {
      "epoch": 0.9103124494090983,
      "grad_norm": 0.951484698639378,
      "learning_rate": 1.17744815107126e-06,
      "loss": 0.3107,
      "step": 16869
    },
    {
      "epoch": 0.9103664130376127,
      "grad_norm": 1.1719765456659375,
      "learning_rate": 1.1772360920919725e-06,
      "loss": 0.494,
      "step": 16870
    },
    {
      "epoch": 0.9104203766661271,
      "grad_norm": 1.427314154120136,
      "learning_rate": 1.1770241573532875e-06,
      "loss": 0.6771,
      "step": 16871
    },
    {
      "epoch": 0.9104743402946414,
      "grad_norm": 1.059229018404341,
      "learning_rate": 1.1768123468612968e-06,
      "loss": 0.4515,
      "step": 16872
    },
    {
      "epoch": 0.9105283039231558,
      "grad_norm": 1.0925142997059742,
      "learning_rate": 1.1766006606220866e-06,
      "loss": 0.4737,
      "step": 16873
    },
    {
      "epoch": 0.9105822675516702,
      "grad_norm": 1.2346115217102818,
      "learning_rate": 1.1763890986417426e-06,
      "loss": 0.6227,
      "step": 16874
    },
    {
      "epoch": 0.9106362311801846,
      "grad_norm": 1.267881158392703,
      "learning_rate": 1.1761776609263446e-06,
      "loss": 0.6349,
      "step": 16875
    },
    {
      "epoch": 0.910690194808699,
      "grad_norm": 1.0405024478839842,
      "learning_rate": 1.1759663474819688e-06,
      "loss": 0.4499,
      "step": 16876
    },
    {
      "epoch": 0.9107441584372133,
      "grad_norm": 0.992217987948118,
      "learning_rate": 1.1757551583146897e-06,
      "loss": 0.4335,
      "step": 16877
    },
    {
      "epoch": 0.9107981220657277,
      "grad_norm": 1.001423784072231,
      "learning_rate": 1.175544093430577e-06,
      "loss": 0.5019,
      "step": 16878
    },
    {
      "epoch": 0.910852085694242,
      "grad_norm": 1.0363973222331415,
      "learning_rate": 1.175333152835697e-06,
      "loss": 0.4376,
      "step": 16879
    },
    {
      "epoch": 0.9109060493227564,
      "grad_norm": 0.8688669417645254,
      "learning_rate": 1.1751223365361114e-06,
      "loss": 0.4468,
      "step": 16880
    },
    {
      "epoch": 0.9109600129512708,
      "grad_norm": 0.8773068439552222,
      "learning_rate": 1.1749116445378798e-06,
      "loss": 0.3512,
      "step": 16881
    },
    {
      "epoch": 0.9110139765797852,
      "grad_norm": 1.0555960470178207,
      "learning_rate": 1.1747010768470576e-06,
      "loss": 0.4125,
      "step": 16882
    },
    {
      "epoch": 0.9110679402082996,
      "grad_norm": 1.0025470799065912,
      "learning_rate": 1.1744906334696969e-06,
      "loss": 0.4808,
      "step": 16883
    },
    {
      "epoch": 0.911121903836814,
      "grad_norm": 0.824373075472008,
      "learning_rate": 1.174280314411846e-06,
      "loss": 0.3676,
      "step": 16884
    },
    {
      "epoch": 0.9111758674653284,
      "grad_norm": 1.1058322650940604,
      "learning_rate": 1.1740701196795496e-06,
      "loss": 0.4872,
      "step": 16885
    },
    {
      "epoch": 0.9112298310938427,
      "grad_norm": 1.1404207331336609,
      "learning_rate": 1.1738600492788495e-06,
      "loss": 0.3994,
      "step": 16886
    },
    {
      "epoch": 0.9112837947223571,
      "grad_norm": 1.0734189328576162,
      "learning_rate": 1.1736501032157827e-06,
      "loss": 0.4858,
      "step": 16887
    },
    {
      "epoch": 0.9113377583508715,
      "grad_norm": 1.0183101278272144,
      "learning_rate": 1.173440281496383e-06,
      "loss": 0.5195,
      "step": 16888
    },
    {
      "epoch": 0.9113917219793859,
      "grad_norm": 1.0626484909225495,
      "learning_rate": 1.1732305841266814e-06,
      "loss": 0.4371,
      "step": 16889
    },
    {
      "epoch": 0.9114456856079003,
      "grad_norm": 0.9478137928140227,
      "learning_rate": 1.1730210111127046e-06,
      "loss": 0.3613,
      "step": 16890
    },
    {
      "epoch": 0.9114996492364147,
      "grad_norm": 1.2298279653042206,
      "learning_rate": 1.1728115624604762e-06,
      "loss": 0.5203,
      "step": 16891
    },
    {
      "epoch": 0.9115536128649291,
      "grad_norm": 1.183539691896239,
      "learning_rate": 1.1726022381760159e-06,
      "loss": 0.4561,
      "step": 16892
    },
    {
      "epoch": 0.9116075764934434,
      "grad_norm": 0.9791766270297672,
      "learning_rate": 1.1723930382653398e-06,
      "loss": 0.3574,
      "step": 16893
    },
    {
      "epoch": 0.9116615401219578,
      "grad_norm": 0.9051945698672417,
      "learning_rate": 1.1721839627344606e-06,
      "loss": 0.547,
      "step": 16894
    },
    {
      "epoch": 0.9117155037504722,
      "grad_norm": 0.9626661709113797,
      "learning_rate": 1.1719750115893865e-06,
      "loss": 0.4983,
      "step": 16895
    },
    {
      "epoch": 0.9117694673789866,
      "grad_norm": 1.105431017892257,
      "learning_rate": 1.1717661848361245e-06,
      "loss": 0.5486,
      "step": 16896
    },
    {
      "epoch": 0.911823431007501,
      "grad_norm": 0.9936592463666849,
      "learning_rate": 1.1715574824806756e-06,
      "loss": 0.4417,
      "step": 16897
    },
    {
      "epoch": 0.9118773946360154,
      "grad_norm": 1.1674816246072603,
      "learning_rate": 1.1713489045290378e-06,
      "loss": 0.5095,
      "step": 16898
    },
    {
      "epoch": 0.9119313582645298,
      "grad_norm": 0.8689335628480996,
      "learning_rate": 1.1711404509872062e-06,
      "loss": 0.4287,
      "step": 16899
    },
    {
      "epoch": 0.911985321893044,
      "grad_norm": 0.9941597777952049,
      "learning_rate": 1.1709321218611731e-06,
      "loss": 0.4762,
      "step": 16900
    },
    {
      "epoch": 0.9120392855215584,
      "grad_norm": 0.9988806628324017,
      "learning_rate": 1.1707239171569242e-06,
      "loss": 0.3618,
      "step": 16901
    },
    {
      "epoch": 0.9120932491500728,
      "grad_norm": 1.0780730264934164,
      "learning_rate": 1.1705158368804445e-06,
      "loss": 0.5722,
      "step": 16902
    },
    {
      "epoch": 0.9121472127785872,
      "grad_norm": 1.0872730174832148,
      "learning_rate": 1.170307881037714e-06,
      "loss": 0.3882,
      "step": 16903
    },
    {
      "epoch": 0.9122011764071016,
      "grad_norm": 1.0922398138690594,
      "learning_rate": 1.1701000496347098e-06,
      "loss": 0.4929,
      "step": 16904
    },
    {
      "epoch": 0.912255140035616,
      "grad_norm": 0.9070075270983222,
      "learning_rate": 1.1698923426774057e-06,
      "loss": 0.3486,
      "step": 16905
    },
    {
      "epoch": 0.9123091036641304,
      "grad_norm": 1.0097538847201213,
      "learning_rate": 1.1696847601717704e-06,
      "loss": 0.454,
      "step": 16906
    },
    {
      "epoch": 0.9123630672926447,
      "grad_norm": 0.9577960226751314,
      "learning_rate": 1.1694773021237714e-06,
      "loss": 0.4537,
      "step": 16907
    },
    {
      "epoch": 0.9124170309211591,
      "grad_norm": 1.0530423808167795,
      "learning_rate": 1.16926996853937e-06,
      "loss": 0.4209,
      "step": 16908
    },
    {
      "epoch": 0.9124709945496735,
      "grad_norm": 1.2101800871826536,
      "learning_rate": 1.169062759424525e-06,
      "loss": 0.4484,
      "step": 16909
    },
    {
      "epoch": 0.9125249581781879,
      "grad_norm": 0.8898463287825683,
      "learning_rate": 1.168855674785193e-06,
      "loss": 0.4085,
      "step": 16910
    },
    {
      "epoch": 0.9125789218067023,
      "grad_norm": 0.98996862264736,
      "learning_rate": 1.1686487146273251e-06,
      "loss": 0.3847,
      "step": 16911
    },
    {
      "epoch": 0.9126328854352167,
      "grad_norm": 1.1031345405815276,
      "learning_rate": 1.1684418789568692e-06,
      "loss": 0.5635,
      "step": 16912
    },
    {
      "epoch": 0.9126868490637311,
      "grad_norm": 1.0128103338518741,
      "learning_rate": 1.1682351677797708e-06,
      "loss": 0.4614,
      "step": 16913
    },
    {
      "epoch": 0.9127408126922454,
      "grad_norm": 0.9485855556713889,
      "learning_rate": 1.1680285811019712e-06,
      "loss": 0.4142,
      "step": 16914
    },
    {
      "epoch": 0.9127947763207598,
      "grad_norm": 0.7421410704731143,
      "learning_rate": 1.1678221189294065e-06,
      "loss": 0.3003,
      "step": 16915
    },
    {
      "epoch": 0.9128487399492742,
      "grad_norm": 1.132683247003801,
      "learning_rate": 1.167615781268012e-06,
      "loss": 0.5233,
      "step": 16916
    },
    {
      "epoch": 0.9129027035777886,
      "grad_norm": 1.0033156577917535,
      "learning_rate": 1.167409568123717e-06,
      "loss": 0.4335,
      "step": 16917
    },
    {
      "epoch": 0.912956667206303,
      "grad_norm": 0.8197734071933604,
      "learning_rate": 1.1672034795024486e-06,
      "loss": 0.3627,
      "step": 16918
    },
    {
      "epoch": 0.9130106308348174,
      "grad_norm": 1.0228482283353368,
      "learning_rate": 1.1669975154101306e-06,
      "loss": 0.4696,
      "step": 16919
    },
    {
      "epoch": 0.9130645944633318,
      "grad_norm": 1.1195546609838904,
      "learning_rate": 1.1667916758526818e-06,
      "loss": 0.4858,
      "step": 16920
    },
    {
      "epoch": 0.913118558091846,
      "grad_norm": 1.0239018892206044,
      "learning_rate": 1.166585960836019e-06,
      "loss": 0.5144,
      "step": 16921
    },
    {
      "epoch": 0.9131725217203605,
      "grad_norm": 1.014630920873624,
      "learning_rate": 1.1663803703660543e-06,
      "loss": 0.4369,
      "step": 16922
    },
    {
      "epoch": 0.9132264853488749,
      "grad_norm": 0.8826738320412871,
      "learning_rate": 1.1661749044486962e-06,
      "loss": 0.3426,
      "step": 16923
    },
    {
      "epoch": 0.9132804489773892,
      "grad_norm": 1.0057867419034845,
      "learning_rate": 1.1659695630898507e-06,
      "loss": 0.4401,
      "step": 16924
    },
    {
      "epoch": 0.9133344126059036,
      "grad_norm": 1.0506411927342532,
      "learning_rate": 1.1657643462954194e-06,
      "loss": 0.5208,
      "step": 16925
    },
    {
      "epoch": 0.913388376234418,
      "grad_norm": 0.8986404685200945,
      "learning_rate": 1.1655592540712997e-06,
      "loss": 0.3333,
      "step": 16926
    },
    {
      "epoch": 0.9134423398629323,
      "grad_norm": 0.8947295748724384,
      "learning_rate": 1.165354286423387e-06,
      "loss": 0.3834,
      "step": 16927
    },
    {
      "epoch": 0.9134963034914467,
      "grad_norm": 1.0086139544807173,
      "learning_rate": 1.1651494433575717e-06,
      "loss": 0.5561,
      "step": 16928
    },
    {
      "epoch": 0.9135502671199611,
      "grad_norm": 1.0077866524111596,
      "learning_rate": 1.1649447248797416e-06,
      "loss": 0.3764,
      "step": 16929
    },
    {
      "epoch": 0.9136042307484755,
      "grad_norm": 1.0155951022679206,
      "learning_rate": 1.1647401309957805e-06,
      "loss": 0.3978,
      "step": 16930
    },
    {
      "epoch": 0.9136581943769899,
      "grad_norm": 1.1438370125493493,
      "learning_rate": 1.164535661711568e-06,
      "loss": 0.5294,
      "step": 16931
    },
    {
      "epoch": 0.9137121580055043,
      "grad_norm": 1.1129915099421646,
      "learning_rate": 1.1643313170329818e-06,
      "loss": 0.442,
      "step": 16932
    },
    {
      "epoch": 0.9137661216340187,
      "grad_norm": 1.3597479594018367,
      "learning_rate": 1.1641270969658942e-06,
      "loss": 0.5389,
      "step": 16933
    },
    {
      "epoch": 0.913820085262533,
      "grad_norm": 1.069723178584762,
      "learning_rate": 1.163923001516175e-06,
      "loss": 0.5008,
      "step": 16934
    },
    {
      "epoch": 0.9138740488910474,
      "grad_norm": 1.2416942829925761,
      "learning_rate": 1.16371903068969e-06,
      "loss": 0.6111,
      "step": 16935
    },
    {
      "epoch": 0.9139280125195618,
      "grad_norm": 0.8305094500166785,
      "learning_rate": 1.1635151844923017e-06,
      "loss": 0.3738,
      "step": 16936
    },
    {
      "epoch": 0.9139819761480762,
      "grad_norm": 1.0306578846344712,
      "learning_rate": 1.1633114629298681e-06,
      "loss": 0.408,
      "step": 16937
    },
    {
      "epoch": 0.9140359397765906,
      "grad_norm": 0.9951584248706308,
      "learning_rate": 1.1631078660082453e-06,
      "loss": 0.3763,
      "step": 16938
    },
    {
      "epoch": 0.914089903405105,
      "grad_norm": 1.1548323295505134,
      "learning_rate": 1.1629043937332843e-06,
      "loss": 0.6563,
      "step": 16939
    },
    {
      "epoch": 0.9141438670336194,
      "grad_norm": 1.3088069572689804,
      "learning_rate": 1.162701046110834e-06,
      "loss": 0.6121,
      "step": 16940
    },
    {
      "epoch": 0.9141978306621337,
      "grad_norm": 0.8539940645028278,
      "learning_rate": 1.162497823146738e-06,
      "loss": 0.3781,
      "step": 16941
    },
    {
      "epoch": 0.9142517942906481,
      "grad_norm": 1.284262987567295,
      "learning_rate": 1.1622947248468369e-06,
      "loss": 0.6883,
      "step": 16942
    },
    {
      "epoch": 0.9143057579191625,
      "grad_norm": 1.0759572582854429,
      "learning_rate": 1.1620917512169689e-06,
      "loss": 0.4421,
      "step": 16943
    },
    {
      "epoch": 0.9143597215476769,
      "grad_norm": 1.111741988068475,
      "learning_rate": 1.1618889022629663e-06,
      "loss": 0.4417,
      "step": 16944
    },
    {
      "epoch": 0.9144136851761913,
      "grad_norm": 0.9582464901250636,
      "learning_rate": 1.1616861779906603e-06,
      "loss": 0.3493,
      "step": 16945
    },
    {
      "epoch": 0.9144676488047057,
      "grad_norm": 0.8084475950723385,
      "learning_rate": 1.1614835784058776e-06,
      "loss": 0.2846,
      "step": 16946
    },
    {
      "epoch": 0.91452161243322,
      "grad_norm": 1.0554539092608264,
      "learning_rate": 1.16128110351444e-06,
      "loss": 0.4522,
      "step": 16947
    },
    {
      "epoch": 0.9145755760617343,
      "grad_norm": 1.011038529604387,
      "learning_rate": 1.1610787533221678e-06,
      "loss": 0.4876,
      "step": 16948
    },
    {
      "epoch": 0.9146295396902487,
      "grad_norm": 1.007069694024192,
      "learning_rate": 1.1608765278348766e-06,
      "loss": 0.4317,
      "step": 16949
    },
    {
      "epoch": 0.9146835033187631,
      "grad_norm": 1.136726347428227,
      "learning_rate": 1.1606744270583776e-06,
      "loss": 0.5333,
      "step": 16950
    },
    {
      "epoch": 0.9147374669472775,
      "grad_norm": 1.3134431504355117,
      "learning_rate": 1.1604724509984805e-06,
      "loss": 0.5045,
      "step": 16951
    },
    {
      "epoch": 0.9147914305757919,
      "grad_norm": 0.9654851806094548,
      "learning_rate": 1.1602705996609897e-06,
      "loss": 0.3679,
      "step": 16952
    },
    {
      "epoch": 0.9148453942043063,
      "grad_norm": 1.042880331932488,
      "learning_rate": 1.1600688730517075e-06,
      "loss": 0.4848,
      "step": 16953
    },
    {
      "epoch": 0.9148993578328207,
      "grad_norm": 0.8517091179157101,
      "learning_rate": 1.1598672711764303e-06,
      "loss": 0.3885,
      "step": 16954
    },
    {
      "epoch": 0.914953321461335,
      "grad_norm": 1.1279226438269596,
      "learning_rate": 1.159665794040954e-06,
      "loss": 0.4635,
      "step": 16955
    },
    {
      "epoch": 0.9150072850898494,
      "grad_norm": 0.7872390429726648,
      "learning_rate": 1.1594644416510676e-06,
      "loss": 0.3994,
      "step": 16956
    },
    {
      "epoch": 0.9150612487183638,
      "grad_norm": 1.028743709888183,
      "learning_rate": 1.1592632140125594e-06,
      "loss": 0.4185,
      "step": 16957
    },
    {
      "epoch": 0.9151152123468782,
      "grad_norm": 1.0723119635976255,
      "learning_rate": 1.159062111131212e-06,
      "loss": 0.4565,
      "step": 16958
    },
    {
      "epoch": 0.9151691759753926,
      "grad_norm": 1.073772788781618,
      "learning_rate": 1.1588611330128058e-06,
      "loss": 0.459,
      "step": 16959
    },
    {
      "epoch": 0.915223139603907,
      "grad_norm": 0.9126417226068954,
      "learning_rate": 1.158660279663117e-06,
      "loss": 0.3961,
      "step": 16960
    },
    {
      "epoch": 0.9152771032324214,
      "grad_norm": 1.046765441052405,
      "learning_rate": 1.1584595510879186e-06,
      "loss": 0.474,
      "step": 16961
    },
    {
      "epoch": 0.9153310668609357,
      "grad_norm": 0.8221019719035789,
      "learning_rate": 1.1582589472929798e-06,
      "loss": 0.3279,
      "step": 16962
    },
    {
      "epoch": 0.9153850304894501,
      "grad_norm": 1.1057198289963421,
      "learning_rate": 1.1580584682840653e-06,
      "loss": 0.5864,
      "step": 16963
    },
    {
      "epoch": 0.9154389941179645,
      "grad_norm": 0.9057140436914779,
      "learning_rate": 1.157858114066938e-06,
      "loss": 0.4203,
      "step": 16964
    },
    {
      "epoch": 0.9154929577464789,
      "grad_norm": 0.9928056462712203,
      "learning_rate": 1.1576578846473558e-06,
      "loss": 0.4263,
      "step": 16965
    },
    {
      "epoch": 0.9155469213749933,
      "grad_norm": 0.8288688400917101,
      "learning_rate": 1.1574577800310733e-06,
      "loss": 0.3815,
      "step": 16966
    },
    {
      "epoch": 0.9156008850035077,
      "grad_norm": 1.0859338307608108,
      "learning_rate": 1.1572578002238424e-06,
      "loss": 0.5694,
      "step": 16967
    },
    {
      "epoch": 0.9156548486320221,
      "grad_norm": 0.900635605981182,
      "learning_rate": 1.1570579452314102e-06,
      "loss": 0.3727,
      "step": 16968
    },
    {
      "epoch": 0.9157088122605364,
      "grad_norm": 1.212689092470864,
      "learning_rate": 1.1568582150595213e-06,
      "loss": 0.5387,
      "step": 16969
    },
    {
      "epoch": 0.9157627758890508,
      "grad_norm": 0.7491935600034412,
      "learning_rate": 1.156658609713915e-06,
      "loss": 0.2649,
      "step": 16970
    },
    {
      "epoch": 0.9158167395175651,
      "grad_norm": 1.1541771907842062,
      "learning_rate": 1.1564591292003296e-06,
      "loss": 0.5503,
      "step": 16971
    },
    {
      "epoch": 0.9158707031460795,
      "grad_norm": 1.1319563736164848,
      "learning_rate": 1.156259773524497e-06,
      "loss": 0.4652,
      "step": 16972
    },
    {
      "epoch": 0.9159246667745939,
      "grad_norm": 1.0150001952676329,
      "learning_rate": 1.1560605426921475e-06,
      "loss": 0.4932,
      "step": 16973
    },
    {
      "epoch": 0.9159786304031083,
      "grad_norm": 0.9571521568337467,
      "learning_rate": 1.1558614367090078e-06,
      "loss": 0.4205,
      "step": 16974
    },
    {
      "epoch": 0.9160325940316227,
      "grad_norm": 1.1493457450394349,
      "learning_rate": 1.1556624555807993e-06,
      "loss": 0.5325,
      "step": 16975
    },
    {
      "epoch": 0.916086557660137,
      "grad_norm": 0.9845167299502717,
      "learning_rate": 1.1554635993132419e-06,
      "loss": 0.4315,
      "step": 16976
    },
    {
      "epoch": 0.9161405212886514,
      "grad_norm": 1.2323099922116445,
      "learning_rate": 1.15526486791205e-06,
      "loss": 0.4658,
      "step": 16977
    },
    {
      "epoch": 0.9161944849171658,
      "grad_norm": 1.1113009236894305,
      "learning_rate": 1.1550662613829365e-06,
      "loss": 0.4177,
      "step": 16978
    },
    {
      "epoch": 0.9162484485456802,
      "grad_norm": 0.9418218512729671,
      "learning_rate": 1.1548677797316085e-06,
      "loss": 0.5034,
      "step": 16979
    },
    {
      "epoch": 0.9163024121741946,
      "grad_norm": 1.0920404228604519,
      "learning_rate": 1.1546694229637704e-06,
      "loss": 0.4276,
      "step": 16980
    },
    {
      "epoch": 0.916356375802709,
      "grad_norm": 0.9558595666841655,
      "learning_rate": 1.1544711910851244e-06,
      "loss": 0.4113,
      "step": 16981
    },
    {
      "epoch": 0.9164103394312234,
      "grad_norm": 1.0236719175302933,
      "learning_rate": 1.1542730841013669e-06,
      "loss": 0.3792,
      "step": 16982
    },
    {
      "epoch": 0.9164643030597377,
      "grad_norm": 1.0805072352798952,
      "learning_rate": 1.1540751020181928e-06,
      "loss": 0.4178,
      "step": 16983
    },
    {
      "epoch": 0.9165182666882521,
      "grad_norm": 1.0251806620167006,
      "learning_rate": 1.1538772448412908e-06,
      "loss": 0.3497,
      "step": 16984
    },
    {
      "epoch": 0.9165722303167665,
      "grad_norm": 1.2894571927182308,
      "learning_rate": 1.1536795125763484e-06,
      "loss": 0.5691,
      "step": 16985
    },
    {
      "epoch": 0.9166261939452809,
      "grad_norm": 0.7723369556549816,
      "learning_rate": 1.1534819052290485e-06,
      "loss": 0.3295,
      "step": 16986
    },
    {
      "epoch": 0.9166801575737953,
      "grad_norm": 0.975575689707718,
      "learning_rate": 1.1532844228050704e-06,
      "loss": 0.4602,
      "step": 16987
    },
    {
      "epoch": 0.9167341212023097,
      "grad_norm": 1.1892590155214395,
      "learning_rate": 1.1530870653100899e-06,
      "loss": 0.4315,
      "step": 16988
    },
    {
      "epoch": 0.9167880848308241,
      "grad_norm": 1.073504029566406,
      "learning_rate": 1.152889832749779e-06,
      "loss": 0.4778,
      "step": 16989
    },
    {
      "epoch": 0.9168420484593384,
      "grad_norm": 1.0081862369836172,
      "learning_rate": 1.1526927251298076e-06,
      "loss": 0.3521,
      "step": 16990
    },
    {
      "epoch": 0.9168960120878528,
      "grad_norm": 0.9333604770467135,
      "learning_rate": 1.1524957424558397e-06,
      "loss": 0.3443,
      "step": 16991
    },
    {
      "epoch": 0.9169499757163672,
      "grad_norm": 0.9450674538494087,
      "learning_rate": 1.1522988847335364e-06,
      "loss": 0.5058,
      "step": 16992
    },
    {
      "epoch": 0.9170039393448816,
      "grad_norm": 1.106172691271883,
      "learning_rate": 1.1521021519685568e-06,
      "loss": 0.5654,
      "step": 16993
    },
    {
      "epoch": 0.917057902973396,
      "grad_norm": 0.8745514117081873,
      "learning_rate": 1.1519055441665543e-06,
      "loss": 0.3819,
      "step": 16994
    },
    {
      "epoch": 0.9171118666019104,
      "grad_norm": 1.0609949368110423,
      "learning_rate": 1.15170906133318e-06,
      "loss": 0.5089,
      "step": 16995
    },
    {
      "epoch": 0.9171658302304246,
      "grad_norm": 1.2138329950584097,
      "learning_rate": 1.151512703474081e-06,
      "loss": 0.4594,
      "step": 16996
    },
    {
      "epoch": 0.917219793858939,
      "grad_norm": 1.083636295949006,
      "learning_rate": 1.1513164705949006e-06,
      "loss": 0.5259,
      "step": 16997
    },
    {
      "epoch": 0.9172737574874534,
      "grad_norm": 0.9425070332663033,
      "learning_rate": 1.151120362701279e-06,
      "loss": 0.4298,
      "step": 16998
    },
    {
      "epoch": 0.9173277211159678,
      "grad_norm": 0.9038492433593869,
      "learning_rate": 1.1509243797988523e-06,
      "loss": 0.3843,
      "step": 16999
    },
    {
      "epoch": 0.9173816847444822,
      "grad_norm": 1.1866900096261452,
      "learning_rate": 1.1507285218932529e-06,
      "loss": 0.4999,
      "step": 17000
    },
    {
      "epoch": 0.9173816847444822,
      "eval_loss": 0.5285828709602356,
      "eval_runtime": 164.3598,
      "eval_samples_per_second": 20.924,
      "eval_steps_per_second": 0.876,
      "step": 17000
    },
    {
      "epoch": 0.9174356483729966,
      "grad_norm": 1.1155783507408839,
      "learning_rate": 1.150532788990111e-06,
      "loss": 0.4426,
      "step": 17001
    },
    {
      "epoch": 0.917489612001511,
      "grad_norm": 1.088237415312259,
      "learning_rate": 1.1503371810950514e-06,
      "loss": 0.5544,
      "step": 17002
    },
    {
      "epoch": 0.9175435756300253,
      "grad_norm": 0.8358327920475628,
      "learning_rate": 1.1501416982136967e-06,
      "loss": 0.3636,
      "step": 17003
    },
    {
      "epoch": 0.9175975392585397,
      "grad_norm": 0.9647551009148343,
      "learning_rate": 1.149946340351664e-06,
      "loss": 0.4291,
      "step": 17004
    },
    {
      "epoch": 0.9176515028870541,
      "grad_norm": 1.1061946708540777,
      "learning_rate": 1.1497511075145689e-06,
      "loss": 0.4437,
      "step": 17005
    },
    {
      "epoch": 0.9177054665155685,
      "grad_norm": 1.1745516579705224,
      "learning_rate": 1.1495559997080231e-06,
      "loss": 0.4748,
      "step": 17006
    },
    {
      "epoch": 0.9177594301440829,
      "grad_norm": 1.098433588168236,
      "learning_rate": 1.149361016937633e-06,
      "loss": 0.4418,
      "step": 17007
    },
    {
      "epoch": 0.9178133937725973,
      "grad_norm": 1.164364168280332,
      "learning_rate": 1.149166159209004e-06,
      "loss": 0.4921,
      "step": 17008
    },
    {
      "epoch": 0.9178673574011117,
      "grad_norm": 0.8363004858697384,
      "learning_rate": 1.1489714265277352e-06,
      "loss": 0.4179,
      "step": 17009
    },
    {
      "epoch": 0.917921321029626,
      "grad_norm": 1.0548350388803647,
      "learning_rate": 1.1487768188994248e-06,
      "loss": 0.4554,
      "step": 17010
    },
    {
      "epoch": 0.9179752846581404,
      "grad_norm": 1.116210290767277,
      "learning_rate": 1.1485823363296646e-06,
      "loss": 0.5774,
      "step": 17011
    },
    {
      "epoch": 0.9180292482866548,
      "grad_norm": 0.8716425216474286,
      "learning_rate": 1.148387978824045e-06,
      "loss": 0.4093,
      "step": 17012
    },
    {
      "epoch": 0.9180832119151692,
      "grad_norm": 1.1240306117838947,
      "learning_rate": 1.1481937463881514e-06,
      "loss": 0.4718,
      "step": 17013
    },
    {
      "epoch": 0.9181371755436836,
      "grad_norm": 0.990497087661862,
      "learning_rate": 1.147999639027567e-06,
      "loss": 0.4847,
      "step": 17014
    },
    {
      "epoch": 0.918191139172198,
      "grad_norm": 0.9867570265340146,
      "learning_rate": 1.1478056567478708e-06,
      "loss": 0.454,
      "step": 17015
    },
    {
      "epoch": 0.9182451028007124,
      "grad_norm": 0.9120879093433375,
      "learning_rate": 1.1476117995546371e-06,
      "loss": 0.3977,
      "step": 17016
    },
    {
      "epoch": 0.9182990664292267,
      "grad_norm": 1.1273761401346343,
      "learning_rate": 1.1474180674534384e-06,
      "loss": 0.4822,
      "step": 17017
    },
    {
      "epoch": 0.918353030057741,
      "grad_norm": 1.0352979098880144,
      "learning_rate": 1.1472244604498425e-06,
      "loss": 0.5565,
      "step": 17018
    },
    {
      "epoch": 0.9184069936862554,
      "grad_norm": 1.151319556675141,
      "learning_rate": 1.147030978549413e-06,
      "loss": 0.4297,
      "step": 17019
    },
    {
      "epoch": 0.9184609573147698,
      "grad_norm": 1.1265850355796767,
      "learning_rate": 1.1468376217577124e-06,
      "loss": 0.5055,
      "step": 17020
    },
    {
      "epoch": 0.9185149209432842,
      "grad_norm": 0.9721762345578367,
      "learning_rate": 1.1466443900802964e-06,
      "loss": 0.3163,
      "step": 17021
    },
    {
      "epoch": 0.9185688845717986,
      "grad_norm": 0.900237386840704,
      "learning_rate": 1.1464512835227194e-06,
      "loss": 0.4876,
      "step": 17022
    },
    {
      "epoch": 0.918622848200313,
      "grad_norm": 1.362122660503254,
      "learning_rate": 1.146258302090532e-06,
      "loss": 0.4905,
      "step": 17023
    },
    {
      "epoch": 0.9186768118288273,
      "grad_norm": 0.9162471402076313,
      "learning_rate": 1.1460654457892797e-06,
      "loss": 0.4241,
      "step": 17024
    },
    {
      "epoch": 0.9187307754573417,
      "grad_norm": 1.1224590258572404,
      "learning_rate": 1.1458727146245058e-06,
      "loss": 0.473,
      "step": 17025
    },
    {
      "epoch": 0.9187847390858561,
      "grad_norm": 0.8916376445884127,
      "learning_rate": 1.14568010860175e-06,
      "loss": 0.4442,
      "step": 17026
    },
    {
      "epoch": 0.9188387027143705,
      "grad_norm": 1.0374183930381475,
      "learning_rate": 1.1454876277265474e-06,
      "loss": 0.5648,
      "step": 17027
    },
    {
      "epoch": 0.9188926663428849,
      "grad_norm": 0.9886527990337742,
      "learning_rate": 1.1452952720044303e-06,
      "loss": 0.3942,
      "step": 17028
    },
    {
      "epoch": 0.9189466299713993,
      "grad_norm": 1.1326640341159453,
      "learning_rate": 1.1451030414409266e-06,
      "loss": 0.5084,
      "step": 17029
    },
    {
      "epoch": 0.9190005935999137,
      "grad_norm": 0.8828127212879949,
      "learning_rate": 1.1449109360415625e-06,
      "loss": 0.3348,
      "step": 17030
    },
    {
      "epoch": 0.919054557228428,
      "grad_norm": 1.1157326869438646,
      "learning_rate": 1.1447189558118591e-06,
      "loss": 0.3628,
      "step": 17031
    },
    {
      "epoch": 0.9191085208569424,
      "grad_norm": 1.1238648269189224,
      "learning_rate": 1.1445271007573329e-06,
      "loss": 0.536,
      "step": 17032
    },
    {
      "epoch": 0.9191624844854568,
      "grad_norm": 1.028174013329178,
      "learning_rate": 1.1443353708834983e-06,
      "loss": 0.5468,
      "step": 17033
    },
    {
      "epoch": 0.9192164481139712,
      "grad_norm": 1.1044161399820003,
      "learning_rate": 1.1441437661958668e-06,
      "loss": 0.5068,
      "step": 17034
    },
    {
      "epoch": 0.9192704117424856,
      "grad_norm": 1.0219623345979358,
      "learning_rate": 1.1439522866999443e-06,
      "loss": 0.4524,
      "step": 17035
    },
    {
      "epoch": 0.919324375371,
      "grad_norm": 1.0541482944309646,
      "learning_rate": 1.143760932401235e-06,
      "loss": 0.4213,
      "step": 17036
    },
    {
      "epoch": 0.9193783389995144,
      "grad_norm": 1.1580814771599297,
      "learning_rate": 1.1435697033052381e-06,
      "loss": 0.5353,
      "step": 17037
    },
    {
      "epoch": 0.9194323026280287,
      "grad_norm": 1.1243527196951757,
      "learning_rate": 1.1433785994174501e-06,
      "loss": 0.5016,
      "step": 17038
    },
    {
      "epoch": 0.9194862662565431,
      "grad_norm": 1.0893624226320795,
      "learning_rate": 1.1431876207433628e-06,
      "loss": 0.5734,
      "step": 17039
    },
    {
      "epoch": 0.9195402298850575,
      "grad_norm": 1.0862059095158552,
      "learning_rate": 1.1429967672884653e-06,
      "loss": 0.4973,
      "step": 17040
    },
    {
      "epoch": 0.9195941935135719,
      "grad_norm": 0.886309627725469,
      "learning_rate": 1.1428060390582433e-06,
      "loss": 0.3487,
      "step": 17041
    },
    {
      "epoch": 0.9196481571420863,
      "grad_norm": 1.0585815538239522,
      "learning_rate": 1.1426154360581782e-06,
      "loss": 0.393,
      "step": 17042
    },
    {
      "epoch": 0.9197021207706007,
      "grad_norm": 1.1782601818750436,
      "learning_rate": 1.1424249582937485e-06,
      "loss": 0.4978,
      "step": 17043
    },
    {
      "epoch": 0.919756084399115,
      "grad_norm": 1.055006352915145,
      "learning_rate": 1.1422346057704287e-06,
      "loss": 0.3948,
      "step": 17044
    },
    {
      "epoch": 0.9198100480276293,
      "grad_norm": 1.2846705883695575,
      "learning_rate": 1.1420443784936894e-06,
      "loss": 0.6511,
      "step": 17045
    },
    {
      "epoch": 0.9198640116561437,
      "grad_norm": 1.0814506287307002,
      "learning_rate": 1.141854276468998e-06,
      "loss": 0.6203,
      "step": 17046
    },
    {
      "epoch": 0.9199179752846581,
      "grad_norm": 1.0968791816291783,
      "learning_rate": 1.1416642997018185e-06,
      "loss": 0.4518,
      "step": 17047
    },
    {
      "epoch": 0.9199719389131725,
      "grad_norm": 1.0274509489929953,
      "learning_rate": 1.1414744481976105e-06,
      "loss": 0.4897,
      "step": 17048
    },
    {
      "epoch": 0.9200259025416869,
      "grad_norm": 1.1295071116784903,
      "learning_rate": 1.1412847219618304e-06,
      "loss": 0.5577,
      "step": 17049
    },
    {
      "epoch": 0.9200798661702013,
      "grad_norm": 0.9170240695275826,
      "learning_rate": 1.1410951209999322e-06,
      "loss": 0.4143,
      "step": 17050
    },
    {
      "epoch": 0.9201338297987157,
      "grad_norm": 1.0957323082543307,
      "learning_rate": 1.1409056453173644e-06,
      "loss": 0.4437,
      "step": 17051
    },
    {
      "epoch": 0.92018779342723,
      "grad_norm": 1.210417880116114,
      "learning_rate": 1.1407162949195732e-06,
      "loss": 0.4704,
      "step": 17052
    },
    {
      "epoch": 0.9202417570557444,
      "grad_norm": 1.1573002554251823,
      "learning_rate": 1.1405270698120003e-06,
      "loss": 0.5054,
      "step": 17053
    },
    {
      "epoch": 0.9202957206842588,
      "grad_norm": 1.0398608465687487,
      "learning_rate": 1.140337970000084e-06,
      "loss": 0.4437,
      "step": 17054
    },
    {
      "epoch": 0.9203496843127732,
      "grad_norm": 0.8815851983788495,
      "learning_rate": 1.14014899548926e-06,
      "loss": 0.39,
      "step": 17055
    },
    {
      "epoch": 0.9204036479412876,
      "grad_norm": 1.0235498842215274,
      "learning_rate": 1.1399601462849583e-06,
      "loss": 0.4799,
      "step": 17056
    },
    {
      "epoch": 0.920457611569802,
      "grad_norm": 1.031980673044282,
      "learning_rate": 1.1397714223926082e-06,
      "loss": 0.4805,
      "step": 17057
    },
    {
      "epoch": 0.9205115751983164,
      "grad_norm": 1.0818361430716121,
      "learning_rate": 1.139582823817633e-06,
      "loss": 0.4017,
      "step": 17058
    },
    {
      "epoch": 0.9205655388268307,
      "grad_norm": 1.2076673289197333,
      "learning_rate": 1.1393943505654539e-06,
      "loss": 0.5072,
      "step": 17059
    },
    {
      "epoch": 0.9206195024553451,
      "grad_norm": 1.04528249771486,
      "learning_rate": 1.1392060026414866e-06,
      "loss": 0.388,
      "step": 17060
    },
    {
      "epoch": 0.9206734660838595,
      "grad_norm": 1.1233164887601286,
      "learning_rate": 1.1390177800511448e-06,
      "loss": 0.4477,
      "step": 17061
    },
    {
      "epoch": 0.9207274297123739,
      "grad_norm": 1.1271513990039128,
      "learning_rate": 1.1388296827998387e-06,
      "loss": 0.4982,
      "step": 17062
    },
    {
      "epoch": 0.9207813933408883,
      "grad_norm": 0.8524433233965182,
      "learning_rate": 1.1386417108929744e-06,
      "loss": 0.2946,
      "step": 17063
    },
    {
      "epoch": 0.9208353569694027,
      "grad_norm": 1.0545686855009162,
      "learning_rate": 1.1384538643359541e-06,
      "loss": 0.5361,
      "step": 17064
    },
    {
      "epoch": 0.920889320597917,
      "grad_norm": 1.1580989174403744,
      "learning_rate": 1.1382661431341769e-06,
      "loss": 0.4756,
      "step": 17065
    },
    {
      "epoch": 0.9209432842264313,
      "grad_norm": 1.1106304842621835,
      "learning_rate": 1.1380785472930378e-06,
      "loss": 0.5145,
      "step": 17066
    },
    {
      "epoch": 0.9209972478549457,
      "grad_norm": 1.1791707125633641,
      "learning_rate": 1.1378910768179284e-06,
      "loss": 0.4882,
      "step": 17067
    },
    {
      "epoch": 0.9210512114834601,
      "grad_norm": 0.9534236996505927,
      "learning_rate": 1.137703731714237e-06,
      "loss": 0.4203,
      "step": 17068
    },
    {
      "epoch": 0.9211051751119745,
      "grad_norm": 1.0999607342994862,
      "learning_rate": 1.1375165119873482e-06,
      "loss": 0.406,
      "step": 17069
    },
    {
      "epoch": 0.9211591387404889,
      "grad_norm": 1.0135776779418948,
      "learning_rate": 1.1373294176426432e-06,
      "loss": 0.5796,
      "step": 17070
    },
    {
      "epoch": 0.9212131023690033,
      "grad_norm": 1.070523103915182,
      "learning_rate": 1.1371424486854982e-06,
      "loss": 0.4645,
      "step": 17071
    },
    {
      "epoch": 0.9212670659975176,
      "grad_norm": 0.9756657946035469,
      "learning_rate": 1.1369556051212882e-06,
      "loss": 0.404,
      "step": 17072
    },
    {
      "epoch": 0.921321029626032,
      "grad_norm": 1.040922304839918,
      "learning_rate": 1.1367688869553823e-06,
      "loss": 0.4798,
      "step": 17073
    },
    {
      "epoch": 0.9213749932545464,
      "grad_norm": 0.7890646642457021,
      "learning_rate": 1.136582294193148e-06,
      "loss": 0.2575,
      "step": 17074
    },
    {
      "epoch": 0.9214289568830608,
      "grad_norm": 1.075064477087498,
      "learning_rate": 1.136395826839947e-06,
      "loss": 0.4548,
      "step": 17075
    },
    {
      "epoch": 0.9214829205115752,
      "grad_norm": 1.1018460358345108,
      "learning_rate": 1.1362094849011386e-06,
      "loss": 0.4652,
      "step": 17076
    },
    {
      "epoch": 0.9215368841400896,
      "grad_norm": 1.0956946113548025,
      "learning_rate": 1.1360232683820795e-06,
      "loss": 0.5099,
      "step": 17077
    },
    {
      "epoch": 0.921590847768604,
      "grad_norm": 1.1813705198965359,
      "learning_rate": 1.1358371772881204e-06,
      "loss": 0.4539,
      "step": 17078
    },
    {
      "epoch": 0.9216448113971183,
      "grad_norm": 0.953211387826852,
      "learning_rate": 1.1356512116246111e-06,
      "loss": 0.4959,
      "step": 17079
    },
    {
      "epoch": 0.9216987750256327,
      "grad_norm": 1.0781003571580279,
      "learning_rate": 1.1354653713968955e-06,
      "loss": 0.3908,
      "step": 17080
    },
    {
      "epoch": 0.9217527386541471,
      "grad_norm": 1.1235241210100717,
      "learning_rate": 1.1352796566103155e-06,
      "loss": 0.5921,
      "step": 17081
    },
    {
      "epoch": 0.9218067022826615,
      "grad_norm": 0.7780656021408747,
      "learning_rate": 1.1350940672702085e-06,
      "loss": 0.2707,
      "step": 17082
    },
    {
      "epoch": 0.9218606659111759,
      "grad_norm": 0.9503794878089104,
      "learning_rate": 1.134908603381908e-06,
      "loss": 0.4029,
      "step": 17083
    },
    {
      "epoch": 0.9219146295396903,
      "grad_norm": 1.0417896858242564,
      "learning_rate": 1.1347232649507452e-06,
      "loss": 0.4403,
      "step": 17084
    },
    {
      "epoch": 0.9219685931682047,
      "grad_norm": 1.1441889718527973,
      "learning_rate": 1.1345380519820464e-06,
      "loss": 0.5491,
      "step": 17085
    },
    {
      "epoch": 0.922022556796719,
      "grad_norm": 1.1955788790678965,
      "learning_rate": 1.134352964481135e-06,
      "loss": 0.5337,
      "step": 17086
    },
    {
      "epoch": 0.9220765204252334,
      "grad_norm": 0.9766053204211788,
      "learning_rate": 1.1341680024533308e-06,
      "loss": 0.3935,
      "step": 17087
    },
    {
      "epoch": 0.9221304840537478,
      "grad_norm": 0.9936432766759872,
      "learning_rate": 1.133983165903949e-06,
      "loss": 0.5345,
      "step": 17088
    },
    {
      "epoch": 0.9221844476822622,
      "grad_norm": 1.303900751500316,
      "learning_rate": 1.1337984548383027e-06,
      "loss": 0.4372,
      "step": 17089
    },
    {
      "epoch": 0.9222384113107766,
      "grad_norm": 0.8973116089762633,
      "learning_rate": 1.1336138692617008e-06,
      "loss": 0.4089,
      "step": 17090
    },
    {
      "epoch": 0.922292374939291,
      "grad_norm": 1.1169234211151209,
      "learning_rate": 1.133429409179448e-06,
      "loss": 0.4716,
      "step": 17091
    },
    {
      "epoch": 0.9223463385678053,
      "grad_norm": 1.3783416456332866,
      "learning_rate": 1.133245074596846e-06,
      "loss": 0.4538,
      "step": 17092
    },
    {
      "epoch": 0.9224003021963196,
      "grad_norm": 1.1102316359296656,
      "learning_rate": 1.1330608655191933e-06,
      "loss": 0.4622,
      "step": 17093
    },
    {
      "epoch": 0.922454265824834,
      "grad_norm": 1.1289029400290056,
      "learning_rate": 1.1328767819517835e-06,
      "loss": 0.3707,
      "step": 17094
    },
    {
      "epoch": 0.9225082294533484,
      "grad_norm": 0.7536033001466376,
      "learning_rate": 1.1326928238999076e-06,
      "loss": 0.3129,
      "step": 17095
    },
    {
      "epoch": 0.9225621930818628,
      "grad_norm": 1.0422039236377312,
      "learning_rate": 1.1325089913688528e-06,
      "loss": 0.4999,
      "step": 17096
    },
    {
      "epoch": 0.9226161567103772,
      "grad_norm": 1.109389978565776,
      "learning_rate": 1.1323252843639027e-06,
      "loss": 0.434,
      "step": 17097
    },
    {
      "epoch": 0.9226701203388916,
      "grad_norm": 1.1391214952837603,
      "learning_rate": 1.1321417028903373e-06,
      "loss": 0.4822,
      "step": 17098
    },
    {
      "epoch": 0.922724083967406,
      "grad_norm": 0.9926647855838342,
      "learning_rate": 1.1319582469534324e-06,
      "loss": 0.3999,
      "step": 17099
    },
    {
      "epoch": 0.9227780475959203,
      "grad_norm": 1.092255994618334,
      "learning_rate": 1.1317749165584615e-06,
      "loss": 0.5098,
      "step": 17100
    },
    {
      "epoch": 0.9228320112244347,
      "grad_norm": 0.9890223491671929,
      "learning_rate": 1.1315917117106926e-06,
      "loss": 0.4589,
      "step": 17101
    },
    {
      "epoch": 0.9228859748529491,
      "grad_norm": 0.9204321965445273,
      "learning_rate": 1.1314086324153922e-06,
      "loss": 0.3734,
      "step": 17102
    },
    {
      "epoch": 0.9229399384814635,
      "grad_norm": 0.9501815354081277,
      "learning_rate": 1.1312256786778219e-06,
      "loss": 0.3095,
      "step": 17103
    },
    {
      "epoch": 0.9229939021099779,
      "grad_norm": 0.8888529357759046,
      "learning_rate": 1.1310428505032397e-06,
      "loss": 0.3913,
      "step": 17104
    },
    {
      "epoch": 0.9230478657384923,
      "grad_norm": 1.0211170407469718,
      "learning_rate": 1.1308601478969007e-06,
      "loss": 0.4369,
      "step": 17105
    },
    {
      "epoch": 0.9231018293670067,
      "grad_norm": 1.0756297481705168,
      "learning_rate": 1.130677570864056e-06,
      "loss": 0.5124,
      "step": 17106
    },
    {
      "epoch": 0.923155792995521,
      "grad_norm": 1.0432492250141208,
      "learning_rate": 1.1304951194099527e-06,
      "loss": 0.4023,
      "step": 17107
    },
    {
      "epoch": 0.9232097566240354,
      "grad_norm": 1.0995340398068094,
      "learning_rate": 1.1303127935398346e-06,
      "loss": 0.4011,
      "step": 17108
    },
    {
      "epoch": 0.9232637202525498,
      "grad_norm": 0.819933435158619,
      "learning_rate": 1.1301305932589419e-06,
      "loss": 0.4285,
      "step": 17109
    },
    {
      "epoch": 0.9233176838810642,
      "grad_norm": 1.1604469709711045,
      "learning_rate": 1.1299485185725115e-06,
      "loss": 0.43,
      "step": 17110
    },
    {
      "epoch": 0.9233716475095786,
      "grad_norm": 1.1537034770087642,
      "learning_rate": 1.1297665694857763e-06,
      "loss": 0.6285,
      "step": 17111
    },
    {
      "epoch": 0.923425611138093,
      "grad_norm": 0.8693944776839003,
      "learning_rate": 1.1295847460039655e-06,
      "loss": 0.4008,
      "step": 17112
    },
    {
      "epoch": 0.9234795747666074,
      "grad_norm": 1.2508552647148314,
      "learning_rate": 1.129403048132305e-06,
      "loss": 0.6882,
      "step": 17113
    },
    {
      "epoch": 0.9235335383951216,
      "grad_norm": 1.0056142815107392,
      "learning_rate": 1.1292214758760178e-06,
      "loss": 0.4088,
      "step": 17114
    },
    {
      "epoch": 0.923587502023636,
      "grad_norm": 1.1335008756648373,
      "learning_rate": 1.1290400292403214e-06,
      "loss": 0.3617,
      "step": 17115
    },
    {
      "epoch": 0.9236414656521504,
      "grad_norm": 1.048711645470727,
      "learning_rate": 1.1288587082304307e-06,
      "loss": 0.4579,
      "step": 17116
    },
    {
      "epoch": 0.9236954292806648,
      "grad_norm": 1.1200453210847177,
      "learning_rate": 1.1286775128515576e-06,
      "loss": 0.4916,
      "step": 17117
    },
    {
      "epoch": 0.9237493929091792,
      "grad_norm": 1.3141451218483213,
      "learning_rate": 1.12849644310891e-06,
      "loss": 0.4754,
      "step": 17118
    },
    {
      "epoch": 0.9238033565376936,
      "grad_norm": 1.0849959869446433,
      "learning_rate": 1.1283154990076911e-06,
      "loss": 0.458,
      "step": 17119
    },
    {
      "epoch": 0.923857320166208,
      "grad_norm": 1.1316401129916418,
      "learning_rate": 1.1281346805531024e-06,
      "loss": 0.417,
      "step": 17120
    },
    {
      "epoch": 0.9239112837947223,
      "grad_norm": 1.0808390937688435,
      "learning_rate": 1.1279539877503408e-06,
      "loss": 0.4629,
      "step": 17121
    },
    {
      "epoch": 0.9239652474232367,
      "grad_norm": 1.1527773689519094,
      "learning_rate": 1.127773420604599e-06,
      "loss": 0.5622,
      "step": 17122
    },
    {
      "epoch": 0.9240192110517511,
      "grad_norm": 1.1893621151238225,
      "learning_rate": 1.1275929791210665e-06,
      "loss": 0.5139,
      "step": 17123
    },
    {
      "epoch": 0.9240731746802655,
      "grad_norm": 0.9200976762749007,
      "learning_rate": 1.12741266330493e-06,
      "loss": 0.4983,
      "step": 17124
    },
    {
      "epoch": 0.9241271383087799,
      "grad_norm": 1.164780516796796,
      "learning_rate": 1.1272324731613716e-06,
      "loss": 0.4464,
      "step": 17125
    },
    {
      "epoch": 0.9241811019372943,
      "grad_norm": 1.1784610414245527,
      "learning_rate": 1.1270524086955707e-06,
      "loss": 0.3972,
      "step": 17126
    },
    {
      "epoch": 0.9242350655658087,
      "grad_norm": 0.9127131635431531,
      "learning_rate": 1.1268724699127019e-06,
      "loss": 0.4663,
      "step": 17127
    },
    {
      "epoch": 0.924289029194323,
      "grad_norm": 1.0582830518339936,
      "learning_rate": 1.1266926568179365e-06,
      "loss": 0.3845,
      "step": 17128
    },
    {
      "epoch": 0.9243429928228374,
      "grad_norm": 1.02019230198679,
      "learning_rate": 1.1265129694164438e-06,
      "loss": 0.3848,
      "step": 17129
    },
    {
      "epoch": 0.9243969564513518,
      "grad_norm": 1.027008304472066,
      "learning_rate": 1.126333407713387e-06,
      "loss": 0.4289,
      "step": 17130
    },
    {
      "epoch": 0.9244509200798662,
      "grad_norm": 1.1693731058055012,
      "learning_rate": 1.1261539717139276e-06,
      "loss": 0.4516,
      "step": 17131
    },
    {
      "epoch": 0.9245048837083806,
      "grad_norm": 1.1814510619410472,
      "learning_rate": 1.125974661423222e-06,
      "loss": 0.5784,
      "step": 17132
    },
    {
      "epoch": 0.924558847336895,
      "grad_norm": 1.0294309539058584,
      "learning_rate": 1.1257954768464248e-06,
      "loss": 0.4895,
      "step": 17133
    },
    {
      "epoch": 0.9246128109654093,
      "grad_norm": 0.9731056968971341,
      "learning_rate": 1.1256164179886857e-06,
      "loss": 0.3443,
      "step": 17134
    },
    {
      "epoch": 0.9246667745939237,
      "grad_norm": 0.9876508021307268,
      "learning_rate": 1.1254374848551504e-06,
      "loss": 0.474,
      "step": 17135
    },
    {
      "epoch": 0.924720738222438,
      "grad_norm": 0.8075787376420062,
      "learning_rate": 1.1252586774509616e-06,
      "loss": 0.3051,
      "step": 17136
    },
    {
      "epoch": 0.9247747018509525,
      "grad_norm": 1.1927649976299126,
      "learning_rate": 1.125079995781259e-06,
      "loss": 0.5053,
      "step": 17137
    },
    {
      "epoch": 0.9248286654794668,
      "grad_norm": 1.2369831410687326,
      "learning_rate": 1.124901439851178e-06,
      "loss": 0.4847,
      "step": 17138
    },
    {
      "epoch": 0.9248826291079812,
      "grad_norm": 0.953134832801891,
      "learning_rate": 1.1247230096658498e-06,
      "loss": 0.4206,
      "step": 17139
    },
    {
      "epoch": 0.9249365927364956,
      "grad_norm": 1.0282754225114121,
      "learning_rate": 1.124544705230404e-06,
      "loss": 0.5549,
      "step": 17140
    },
    {
      "epoch": 0.9249905563650099,
      "grad_norm": 1.488968903156188,
      "learning_rate": 1.1243665265499646e-06,
      "loss": 0.6487,
      "step": 17141
    },
    {
      "epoch": 0.9250445199935243,
      "grad_norm": 0.9810361591947774,
      "learning_rate": 1.1241884736296519e-06,
      "loss": 0.5267,
      "step": 17142
    },
    {
      "epoch": 0.9250984836220387,
      "grad_norm": 0.865661985282122,
      "learning_rate": 1.1240105464745838e-06,
      "loss": 0.3648,
      "step": 17143
    },
    {
      "epoch": 0.9251524472505531,
      "grad_norm": 1.0552733968127936,
      "learning_rate": 1.123832745089874e-06,
      "loss": 0.4474,
      "step": 17144
    },
    {
      "epoch": 0.9252064108790675,
      "grad_norm": 1.21209292144695,
      "learning_rate": 1.123655069480633e-06,
      "loss": 0.5531,
      "step": 17145
    },
    {
      "epoch": 0.9252603745075819,
      "grad_norm": 1.1514588514187536,
      "learning_rate": 1.1234775196519678e-06,
      "loss": 0.5145,
      "step": 17146
    },
    {
      "epoch": 0.9253143381360963,
      "grad_norm": 1.0010429097115787,
      "learning_rate": 1.1233000956089802e-06,
      "loss": 0.4279,
      "step": 17147
    },
    {
      "epoch": 0.9253683017646106,
      "grad_norm": 0.873765198824895,
      "learning_rate": 1.1231227973567712e-06,
      "loss": 0.4622,
      "step": 17148
    },
    {
      "epoch": 0.925422265393125,
      "grad_norm": 1.2663105465946654,
      "learning_rate": 1.1229456249004352e-06,
      "loss": 0.5332,
      "step": 17149
    },
    {
      "epoch": 0.9254762290216394,
      "grad_norm": 1.0499196363951542,
      "learning_rate": 1.1227685782450638e-06,
      "loss": 0.2898,
      "step": 17150
    },
    {
      "epoch": 0.9255301926501538,
      "grad_norm": 0.9006006524404785,
      "learning_rate": 1.122591657395747e-06,
      "loss": 0.418,
      "step": 17151
    },
    {
      "epoch": 0.9255841562786682,
      "grad_norm": 0.9961550392107617,
      "learning_rate": 1.1224148623575688e-06,
      "loss": 0.4381,
      "step": 17152
    },
    {
      "epoch": 0.9256381199071826,
      "grad_norm": 0.9373773081017521,
      "learning_rate": 1.1222381931356106e-06,
      "loss": 0.4319,
      "step": 17153
    },
    {
      "epoch": 0.925692083535697,
      "grad_norm": 0.9093651047061327,
      "learning_rate": 1.12206164973495e-06,
      "loss": 0.3954,
      "step": 17154
    },
    {
      "epoch": 0.9257460471642113,
      "grad_norm": 1.0945158871032434,
      "learning_rate": 1.121885232160662e-06,
      "loss": 0.5049,
      "step": 17155
    },
    {
      "epoch": 0.9258000107927257,
      "grad_norm": 0.9699883734010841,
      "learning_rate": 1.1217089404178158e-06,
      "loss": 0.394,
      "step": 17156
    },
    {
      "epoch": 0.9258539744212401,
      "grad_norm": 1.0137439667609587,
      "learning_rate": 1.1215327745114784e-06,
      "loss": 0.5073,
      "step": 17157
    },
    {
      "epoch": 0.9259079380497545,
      "grad_norm": 1.0170979649907703,
      "learning_rate": 1.121356734446713e-06,
      "loss": 0.3884,
      "step": 17158
    },
    {
      "epoch": 0.9259619016782689,
      "grad_norm": 1.1075305414261314,
      "learning_rate": 1.1211808202285797e-06,
      "loss": 0.4916,
      "step": 17159
    },
    {
      "epoch": 0.9260158653067833,
      "grad_norm": 1.1193242084217927,
      "learning_rate": 1.121005031862134e-06,
      "loss": 0.5069,
      "step": 17160
    },
    {
      "epoch": 0.9260698289352977,
      "grad_norm": 0.9520306816081967,
      "learning_rate": 1.1208293693524282e-06,
      "loss": 0.3347,
      "step": 17161
    },
    {
      "epoch": 0.9261237925638119,
      "grad_norm": 1.1324635660416942,
      "learning_rate": 1.1206538327045116e-06,
      "loss": 0.5558,
      "step": 17162
    },
    {
      "epoch": 0.9261777561923263,
      "grad_norm": 0.8891962068242001,
      "learning_rate": 1.1204784219234282e-06,
      "loss": 0.4262,
      "step": 17163
    },
    {
      "epoch": 0.9262317198208407,
      "grad_norm": 0.8889060953592636,
      "learning_rate": 1.1203031370142204e-06,
      "loss": 0.3264,
      "step": 17164
    },
    {
      "epoch": 0.9262856834493551,
      "grad_norm": 1.1808920056980299,
      "learning_rate": 1.1201279779819257e-06,
      "loss": 0.5799,
      "step": 17165
    },
    {
      "epoch": 0.9263396470778695,
      "grad_norm": 1.0079423902035387,
      "learning_rate": 1.1199529448315787e-06,
      "loss": 0.3691,
      "step": 17166
    },
    {
      "epoch": 0.9263936107063839,
      "grad_norm": 1.0049527634231838,
      "learning_rate": 1.1197780375682093e-06,
      "loss": 0.493,
      "step": 17167
    },
    {
      "epoch": 0.9264475743348983,
      "grad_norm": 1.0249945730504453,
      "learning_rate": 1.1196032561968456e-06,
      "loss": 0.4408,
      "step": 17168
    },
    {
      "epoch": 0.9265015379634126,
      "grad_norm": 1.1837771105176669,
      "learning_rate": 1.1194286007225102e-06,
      "loss": 0.5013,
      "step": 17169
    },
    {
      "epoch": 0.926555501591927,
      "grad_norm": 0.9991441585249675,
      "learning_rate": 1.1192540711502234e-06,
      "loss": 0.4524,
      "step": 17170
    },
    {
      "epoch": 0.9266094652204414,
      "grad_norm": 0.8473605159145541,
      "learning_rate": 1.1190796674850003e-06,
      "loss": 0.3729,
      "step": 17171
    },
    {
      "epoch": 0.9266634288489558,
      "grad_norm": 1.0261411970709997,
      "learning_rate": 1.1189053897318546e-06,
      "loss": 0.3716,
      "step": 17172
    },
    {
      "epoch": 0.9267173924774702,
      "grad_norm": 1.0400818934533709,
      "learning_rate": 1.1187312378957944e-06,
      "loss": 0.3879,
      "step": 17173
    },
    {
      "epoch": 0.9267713561059846,
      "grad_norm": 1.1701124127562916,
      "learning_rate": 1.1185572119818256e-06,
      "loss": 0.4509,
      "step": 17174
    },
    {
      "epoch": 0.926825319734499,
      "grad_norm": 0.9201409078863747,
      "learning_rate": 1.1183833119949495e-06,
      "loss": 0.4702,
      "step": 17175
    },
    {
      "epoch": 0.9268792833630133,
      "grad_norm": 1.1238238754939784,
      "learning_rate": 1.1182095379401648e-06,
      "loss": 0.4871,
      "step": 17176
    },
    {
      "epoch": 0.9269332469915277,
      "grad_norm": 1.1297655963330888,
      "learning_rate": 1.1180358898224647e-06,
      "loss": 0.5699,
      "step": 17177
    },
    {
      "epoch": 0.9269872106200421,
      "grad_norm": 1.0475347489763582,
      "learning_rate": 1.1178623676468415e-06,
      "loss": 0.5164,
      "step": 17178
    },
    {
      "epoch": 0.9270411742485565,
      "grad_norm": 0.7877484435589853,
      "learning_rate": 1.1176889714182814e-06,
      "loss": 0.3532,
      "step": 17179
    },
    {
      "epoch": 0.9270951378770709,
      "grad_norm": 1.1040034175356985,
      "learning_rate": 1.1175157011417683e-06,
      "loss": 0.5007,
      "step": 17180
    },
    {
      "epoch": 0.9271491015055853,
      "grad_norm": 0.8964377489021044,
      "learning_rate": 1.1173425568222823e-06,
      "loss": 0.4924,
      "step": 17181
    },
    {
      "epoch": 0.9272030651340997,
      "grad_norm": 1.0617332127500285,
      "learning_rate": 1.1171695384647993e-06,
      "loss": 0.4225,
      "step": 17182
    },
    {
      "epoch": 0.927257028762614,
      "grad_norm": 1.278284611133064,
      "learning_rate": 1.116996646074292e-06,
      "loss": 0.4289,
      "step": 17183
    },
    {
      "epoch": 0.9273109923911284,
      "grad_norm": 1.0113130488371063,
      "learning_rate": 1.1168238796557303e-06,
      "loss": 0.4381,
      "step": 17184
    },
    {
      "epoch": 0.9273649560196427,
      "grad_norm": 1.1715697308905562,
      "learning_rate": 1.1166512392140788e-06,
      "loss": 0.4783,
      "step": 17185
    },
    {
      "epoch": 0.9274189196481571,
      "grad_norm": 0.8541843181844087,
      "learning_rate": 1.1164787247543003e-06,
      "loss": 0.3183,
      "step": 17186
    },
    {
      "epoch": 0.9274728832766715,
      "grad_norm": 1.2893666194743563,
      "learning_rate": 1.1163063362813516e-06,
      "loss": 0.4971,
      "step": 17187
    },
    {
      "epoch": 0.9275268469051859,
      "grad_norm": 0.9762654116425453,
      "learning_rate": 1.1161340738001887e-06,
      "loss": 0.4342,
      "step": 17188
    },
    {
      "epoch": 0.9275808105337003,
      "grad_norm": 1.148166195753637,
      "learning_rate": 1.1159619373157623e-06,
      "loss": 0.5335,
      "step": 17189
    },
    {
      "epoch": 0.9276347741622146,
      "grad_norm": 1.1248471771807793,
      "learning_rate": 1.11578992683302e-06,
      "loss": 0.5505,
      "step": 17190
    },
    {
      "epoch": 0.927688737790729,
      "grad_norm": 0.9591595305276363,
      "learning_rate": 1.115618042356904e-06,
      "loss": 0.3476,
      "step": 17191
    },
    {
      "epoch": 0.9277427014192434,
      "grad_norm": 1.2464198608963695,
      "learning_rate": 1.115446283892356e-06,
      "loss": 0.5416,
      "step": 17192
    },
    {
      "epoch": 0.9277966650477578,
      "grad_norm": 1.5266093741284787,
      "learning_rate": 1.1152746514443122e-06,
      "loss": 0.4077,
      "step": 17193
    },
    {
      "epoch": 0.9278506286762722,
      "grad_norm": 1.10609014632624,
      "learning_rate": 1.1151031450177055e-06,
      "loss": 0.4715,
      "step": 17194
    },
    {
      "epoch": 0.9279045923047866,
      "grad_norm": 0.9409276284781632,
      "learning_rate": 1.114931764617465e-06,
      "loss": 0.4105,
      "step": 17195
    },
    {
      "epoch": 0.927958555933301,
      "grad_norm": 0.8944432616540952,
      "learning_rate": 1.1147605102485155e-06,
      "loss": 0.372,
      "step": 17196
    },
    {
      "epoch": 0.9280125195618153,
      "grad_norm": 1.0588476469608041,
      "learning_rate": 1.1145893819157811e-06,
      "loss": 0.4149,
      "step": 17197
    },
    {
      "epoch": 0.9280664831903297,
      "grad_norm": 1.1755485684601388,
      "learning_rate": 1.1144183796241783e-06,
      "loss": 0.4254,
      "step": 17198
    },
    {
      "epoch": 0.9281204468188441,
      "grad_norm": 0.9801181404880135,
      "learning_rate": 1.114247503378623e-06,
      "loss": 0.4816,
      "step": 17199
    },
    {
      "epoch": 0.9281744104473585,
      "grad_norm": 1.1849025667434465,
      "learning_rate": 1.1140767531840254e-06,
      "loss": 0.4675,
      "step": 17200
    },
    {
      "epoch": 0.9282283740758729,
      "grad_norm": 0.9715956901709254,
      "learning_rate": 1.1139061290452942e-06,
      "loss": 0.5519,
      "step": 17201
    },
    {
      "epoch": 0.9282823377043873,
      "grad_norm": 0.950570690411513,
      "learning_rate": 1.1137356309673323e-06,
      "loss": 0.3672,
      "step": 17202
    },
    {
      "epoch": 0.9283363013329016,
      "grad_norm": 0.8179084170153493,
      "learning_rate": 1.1135652589550406e-06,
      "loss": 0.3813,
      "step": 17203
    },
    {
      "epoch": 0.928390264961416,
      "grad_norm": 1.0349329214833236,
      "learning_rate": 1.113395013013315e-06,
      "loss": 0.4487,
      "step": 17204
    },
    {
      "epoch": 0.9284442285899304,
      "grad_norm": 1.038789373933383,
      "learning_rate": 1.1132248931470496e-06,
      "loss": 0.5069,
      "step": 17205
    },
    {
      "epoch": 0.9284981922184448,
      "grad_norm": 1.0655276848745483,
      "learning_rate": 1.113054899361133e-06,
      "loss": 0.4858,
      "step": 17206
    },
    {
      "epoch": 0.9285521558469592,
      "grad_norm": 1.199973074897376,
      "learning_rate": 1.1128850316604515e-06,
      "loss": 0.6595,
      "step": 17207
    },
    {
      "epoch": 0.9286061194754736,
      "grad_norm": 0.9801633869212708,
      "learning_rate": 1.1127152900498868e-06,
      "loss": 0.4323,
      "step": 17208
    },
    {
      "epoch": 0.928660083103988,
      "grad_norm": 1.0758294277047462,
      "learning_rate": 1.1125456745343175e-06,
      "loss": 0.5692,
      "step": 17209
    },
    {
      "epoch": 0.9287140467325022,
      "grad_norm": 1.1626022374927458,
      "learning_rate": 1.1123761851186196e-06,
      "loss": 0.5068,
      "step": 17210
    },
    {
      "epoch": 0.9287680103610166,
      "grad_norm": 0.8895867803767645,
      "learning_rate": 1.1122068218076626e-06,
      "loss": 0.4129,
      "step": 17211
    },
    {
      "epoch": 0.928821973989531,
      "grad_norm": 1.0401002039998533,
      "learning_rate": 1.1120375846063158e-06,
      "loss": 0.4531,
      "step": 17212
    },
    {
      "epoch": 0.9288759376180454,
      "grad_norm": 0.9702461234855705,
      "learning_rate": 1.1118684735194417e-06,
      "loss": 0.3607,
      "step": 17213
    },
    {
      "epoch": 0.9289299012465598,
      "grad_norm": 0.92075711915728,
      "learning_rate": 1.111699488551902e-06,
      "loss": 0.379,
      "step": 17214
    },
    {
      "epoch": 0.9289838648750742,
      "grad_norm": 1.0518613529888734,
      "learning_rate": 1.1115306297085531e-06,
      "loss": 0.5308,
      "step": 17215
    },
    {
      "epoch": 0.9290378285035886,
      "grad_norm": 0.8659809856294727,
      "learning_rate": 1.111361896994248e-06,
      "loss": 0.3412,
      "step": 17216
    },
    {
      "epoch": 0.9290917921321029,
      "grad_norm": 1.2886770368604843,
      "learning_rate": 1.1111932904138366e-06,
      "loss": 0.5891,
      "step": 17217
    },
    {
      "epoch": 0.9291457557606173,
      "grad_norm": 0.9892016191962122,
      "learning_rate": 1.1110248099721646e-06,
      "loss": 0.4154,
      "step": 17218
    },
    {
      "epoch": 0.9291997193891317,
      "grad_norm": 0.9051110145093129,
      "learning_rate": 1.1108564556740738e-06,
      "loss": 0.3513,
      "step": 17219
    },
    {
      "epoch": 0.9292536830176461,
      "grad_norm": 1.1760371053648504,
      "learning_rate": 1.1106882275244036e-06,
      "loss": 0.422,
      "step": 17220
    },
    {
      "epoch": 0.9293076466461605,
      "grad_norm": 0.9312030733706808,
      "learning_rate": 1.1105201255279894e-06,
      "loss": 0.4449,
      "step": 17221
    },
    {
      "epoch": 0.9293616102746749,
      "grad_norm": 1.014785417066404,
      "learning_rate": 1.110352149689661e-06,
      "loss": 0.5429,
      "step": 17222
    },
    {
      "epoch": 0.9294155739031893,
      "grad_norm": 0.9143504590684891,
      "learning_rate": 1.1101843000142477e-06,
      "loss": 0.4212,
      "step": 17223
    },
    {
      "epoch": 0.9294695375317036,
      "grad_norm": 1.3478028252809429,
      "learning_rate": 1.1100165765065737e-06,
      "loss": 0.5319,
      "step": 17224
    },
    {
      "epoch": 0.929523501160218,
      "grad_norm": 1.1185421020861572,
      "learning_rate": 1.1098489791714586e-06,
      "loss": 0.5507,
      "step": 17225
    },
    {
      "epoch": 0.9295774647887324,
      "grad_norm": 1.1111966993969193,
      "learning_rate": 1.1096815080137196e-06,
      "loss": 0.4859,
      "step": 17226
    },
    {
      "epoch": 0.9296314284172468,
      "grad_norm": 1.1063374142569322,
      "learning_rate": 1.1095141630381709e-06,
      "loss": 0.5138,
      "step": 17227
    },
    {
      "epoch": 0.9296853920457612,
      "grad_norm": 1.0927822915494732,
      "learning_rate": 1.1093469442496203e-06,
      "loss": 0.4989,
      "step": 17228
    },
    {
      "epoch": 0.9297393556742756,
      "grad_norm": 1.0416881321912916,
      "learning_rate": 1.109179851652876e-06,
      "loss": 0.5016,
      "step": 17229
    },
    {
      "epoch": 0.92979331930279,
      "grad_norm": 1.0677881337606276,
      "learning_rate": 1.1090128852527387e-06,
      "loss": 0.4817,
      "step": 17230
    },
    {
      "epoch": 0.9298472829313043,
      "grad_norm": 0.9101789279210969,
      "learning_rate": 1.1088460450540076e-06,
      "loss": 0.3948,
      "step": 17231
    },
    {
      "epoch": 0.9299012465598187,
      "grad_norm": 0.8861957591524564,
      "learning_rate": 1.1086793310614787e-06,
      "loss": 0.3342,
      "step": 17232
    },
    {
      "epoch": 0.929955210188333,
      "grad_norm": 1.2229258747245766,
      "learning_rate": 1.1085127432799427e-06,
      "loss": 0.423,
      "step": 17233
    },
    {
      "epoch": 0.9300091738168474,
      "grad_norm": 0.9137827059361836,
      "learning_rate": 1.1083462817141878e-06,
      "loss": 0.3629,
      "step": 17234
    },
    {
      "epoch": 0.9300631374453618,
      "grad_norm": 0.9656139237967916,
      "learning_rate": 1.1081799463689982e-06,
      "loss": 0.427,
      "step": 17235
    },
    {
      "epoch": 0.9301171010738762,
      "grad_norm": 1.1771460207850526,
      "learning_rate": 1.1080137372491545e-06,
      "loss": 0.4559,
      "step": 17236
    },
    {
      "epoch": 0.9301710647023906,
      "grad_norm": 0.9388769313970659,
      "learning_rate": 1.1078476543594336e-06,
      "loss": 0.4278,
      "step": 17237
    },
    {
      "epoch": 0.9302250283309049,
      "grad_norm": 0.9166087878782561,
      "learning_rate": 1.1076816977046096e-06,
      "loss": 0.4373,
      "step": 17238
    },
    {
      "epoch": 0.9302789919594193,
      "grad_norm": 1.0703586235736726,
      "learning_rate": 1.1075158672894513e-06,
      "loss": 0.4228,
      "step": 17239
    },
    {
      "epoch": 0.9303329555879337,
      "grad_norm": 0.8864505086993925,
      "learning_rate": 1.1073501631187256e-06,
      "loss": 0.3836,
      "step": 17240
    },
    {
      "epoch": 0.9303869192164481,
      "grad_norm": 0.8553034091142702,
      "learning_rate": 1.1071845851971944e-06,
      "loss": 0.325,
      "step": 17241
    },
    {
      "epoch": 0.9304408828449625,
      "grad_norm": 1.0495765790130687,
      "learning_rate": 1.1070191335296169e-06,
      "loss": 0.6224,
      "step": 17242
    },
    {
      "epoch": 0.9304948464734769,
      "grad_norm": 1.1961312798202317,
      "learning_rate": 1.1068538081207478e-06,
      "loss": 0.5098,
      "step": 17243
    },
    {
      "epoch": 0.9305488101019913,
      "grad_norm": 0.883967226041481,
      "learning_rate": 1.1066886089753397e-06,
      "loss": 0.341,
      "step": 17244
    },
    {
      "epoch": 0.9306027737305056,
      "grad_norm": 0.9600200395820533,
      "learning_rate": 1.1065235360981404e-06,
      "loss": 0.586,
      "step": 17245
    },
    {
      "epoch": 0.93065673735902,
      "grad_norm": 0.7652902755684388,
      "learning_rate": 1.1063585894938936e-06,
      "loss": 0.3368,
      "step": 17246
    },
    {
      "epoch": 0.9307107009875344,
      "grad_norm": 1.2378360102053838,
      "learning_rate": 1.1061937691673407e-06,
      "loss": 0.5139,
      "step": 17247
    },
    {
      "epoch": 0.9307646646160488,
      "grad_norm": 1.2221357051678798,
      "learning_rate": 1.1060290751232177e-06,
      "loss": 0.6267,
      "step": 17248
    },
    {
      "epoch": 0.9308186282445632,
      "grad_norm": 0.8719454965965641,
      "learning_rate": 1.1058645073662593e-06,
      "loss": 0.4509,
      "step": 17249
    },
    {
      "epoch": 0.9308725918730776,
      "grad_norm": 0.8855958700325937,
      "learning_rate": 1.1057000659011952e-06,
      "loss": 0.3575,
      "step": 17250
    },
    {
      "epoch": 0.930926555501592,
      "grad_norm": 0.9822031219976025,
      "learning_rate": 1.105535750732751e-06,
      "loss": 0.4387,
      "step": 17251
    },
    {
      "epoch": 0.9309805191301063,
      "grad_norm": 0.987807586379645,
      "learning_rate": 1.1053715618656498e-06,
      "loss": 0.4537,
      "step": 17252
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 1.0068578910740493,
      "learning_rate": 1.1052074993046102e-06,
      "loss": 0.4794,
      "step": 17253
    },
    {
      "epoch": 0.9310884463871351,
      "grad_norm": 1.0121036778970915,
      "learning_rate": 1.105043563054348e-06,
      "loss": 0.3514,
      "step": 17254
    },
    {
      "epoch": 0.9311424100156495,
      "grad_norm": 1.0679524111160856,
      "learning_rate": 1.104879753119574e-06,
      "loss": 0.4076,
      "step": 17255
    },
    {
      "epoch": 0.9311963736441639,
      "grad_norm": 1.1864636830274964,
      "learning_rate": 1.1047160695049974e-06,
      "loss": 0.6043,
      "step": 17256
    },
    {
      "epoch": 0.9312503372726783,
      "grad_norm": 0.8092753965597969,
      "learning_rate": 1.1045525122153214e-06,
      "loss": 0.2931,
      "step": 17257
    },
    {
      "epoch": 0.9313043009011926,
      "grad_norm": 1.3347551052221265,
      "learning_rate": 1.1043890812552477e-06,
      "loss": 0.472,
      "step": 17258
    },
    {
      "epoch": 0.9313582645297069,
      "grad_norm": 1.1094679335527144,
      "learning_rate": 1.1042257766294743e-06,
      "loss": 0.4596,
      "step": 17259
    },
    {
      "epoch": 0.9314122281582213,
      "grad_norm": 0.9046028373275663,
      "learning_rate": 1.1040625983426924e-06,
      "loss": 0.3391,
      "step": 17260
    },
    {
      "epoch": 0.9314661917867357,
      "grad_norm": 1.1355250577646598,
      "learning_rate": 1.1038995463995935e-06,
      "loss": 0.5188,
      "step": 17261
    },
    {
      "epoch": 0.9315201554152501,
      "grad_norm": 1.290877329588031,
      "learning_rate": 1.1037366208048636e-06,
      "loss": 0.6492,
      "step": 17262
    },
    {
      "epoch": 0.9315741190437645,
      "grad_norm": 1.0693484767055255,
      "learning_rate": 1.1035738215631856e-06,
      "loss": 0.4421,
      "step": 17263
    },
    {
      "epoch": 0.9316280826722789,
      "grad_norm": 1.0979597898597038,
      "learning_rate": 1.103411148679238e-06,
      "loss": 0.4442,
      "step": 17264
    },
    {
      "epoch": 0.9316820463007933,
      "grad_norm": 1.0761266028692755,
      "learning_rate": 1.1032486021576961e-06,
      "loss": 0.4086,
      "step": 17265
    },
    {
      "epoch": 0.9317360099293076,
      "grad_norm": 0.9574249927458519,
      "learning_rate": 1.1030861820032328e-06,
      "loss": 0.4927,
      "step": 17266
    },
    {
      "epoch": 0.931789973557822,
      "grad_norm": 1.0212729419615736,
      "learning_rate": 1.102923888220515e-06,
      "loss": 0.3841,
      "step": 17267
    },
    {
      "epoch": 0.9318439371863364,
      "grad_norm": 1.13896800575113,
      "learning_rate": 1.1027617208142076e-06,
      "loss": 0.5671,
      "step": 17268
    },
    {
      "epoch": 0.9318979008148508,
      "grad_norm": 0.9932309306353466,
      "learning_rate": 1.1025996797889712e-06,
      "loss": 0.3995,
      "step": 17269
    },
    {
      "epoch": 0.9319518644433652,
      "grad_norm": 0.9138083109924741,
      "learning_rate": 1.102437765149463e-06,
      "loss": 0.3358,
      "step": 17270
    },
    {
      "epoch": 0.9320058280718796,
      "grad_norm": 1.1906258384324042,
      "learning_rate": 1.1022759769003373e-06,
      "loss": 0.499,
      "step": 17271
    },
    {
      "epoch": 0.9320597917003939,
      "grad_norm": 0.9155554400136157,
      "learning_rate": 1.102114315046244e-06,
      "loss": 0.4681,
      "step": 17272
    },
    {
      "epoch": 0.9321137553289083,
      "grad_norm": 1.0279245675060757,
      "learning_rate": 1.1019527795918284e-06,
      "loss": 0.5581,
      "step": 17273
    },
    {
      "epoch": 0.9321677189574227,
      "grad_norm": 1.0497827393025896,
      "learning_rate": 1.1017913705417337e-06,
      "loss": 0.4492,
      "step": 17274
    },
    {
      "epoch": 0.9322216825859371,
      "grad_norm": 1.1367575641384589,
      "learning_rate": 1.1016300879005993e-06,
      "loss": 0.5066,
      "step": 17275
    },
    {
      "epoch": 0.9322756462144515,
      "grad_norm": 1.2069170255904667,
      "learning_rate": 1.1014689316730604e-06,
      "loss": 0.5807,
      "step": 17276
    },
    {
      "epoch": 0.9323296098429659,
      "grad_norm": 1.042358674046549,
      "learning_rate": 1.1013079018637486e-06,
      "loss": 0.3813,
      "step": 17277
    },
    {
      "epoch": 0.9323835734714803,
      "grad_norm": 0.8998528440625521,
      "learning_rate": 1.1011469984772924e-06,
      "loss": 0.3756,
      "step": 17278
    },
    {
      "epoch": 0.9324375370999946,
      "grad_norm": 1.2383650627914182,
      "learning_rate": 1.1009862215183162e-06,
      "loss": 0.4473,
      "step": 17279
    },
    {
      "epoch": 0.932491500728509,
      "grad_norm": 1.2485503192537872,
      "learning_rate": 1.1008255709914414e-06,
      "loss": 0.5904,
      "step": 17280
    },
    {
      "epoch": 0.9325454643570233,
      "grad_norm": 1.0999215789066727,
      "learning_rate": 1.100665046901284e-06,
      "loss": 0.3905,
      "step": 17281
    },
    {
      "epoch": 0.9325994279855377,
      "grad_norm": 1.0830867755534237,
      "learning_rate": 1.1005046492524587e-06,
      "loss": 0.5057,
      "step": 17282
    },
    {
      "epoch": 0.9326533916140521,
      "grad_norm": 0.8455875049785918,
      "learning_rate": 1.1003443780495748e-06,
      "loss": 0.4082,
      "step": 17283
    },
    {
      "epoch": 0.9327073552425665,
      "grad_norm": 1.004147302318887,
      "learning_rate": 1.1001842332972393e-06,
      "loss": 0.4817,
      "step": 17284
    },
    {
      "epoch": 0.9327613188710809,
      "grad_norm": 1.0621554192860336,
      "learning_rate": 1.1000242150000543e-06,
      "loss": 0.4361,
      "step": 17285
    },
    {
      "epoch": 0.9328152824995952,
      "grad_norm": 1.0840429396551567,
      "learning_rate": 1.0998643231626199e-06,
      "loss": 0.5123,
      "step": 17286
    },
    {
      "epoch": 0.9328692461281096,
      "grad_norm": 0.9924910142864513,
      "learning_rate": 1.0997045577895306e-06,
      "loss": 0.4585,
      "step": 17287
    },
    {
      "epoch": 0.932923209756624,
      "grad_norm": 1.1427535912544042,
      "learning_rate": 1.0995449188853782e-06,
      "loss": 0.5234,
      "step": 17288
    },
    {
      "epoch": 0.9329771733851384,
      "grad_norm": 0.9100148467988263,
      "learning_rate": 1.099385406454751e-06,
      "loss": 0.6417,
      "step": 17289
    },
    {
      "epoch": 0.9330311370136528,
      "grad_norm": 1.0354326402288565,
      "learning_rate": 1.0992260205022344e-06,
      "loss": 0.5027,
      "step": 17290
    },
    {
      "epoch": 0.9330851006421672,
      "grad_norm": 0.9710623637983505,
      "learning_rate": 1.0990667610324081e-06,
      "loss": 0.4556,
      "step": 17291
    },
    {
      "epoch": 0.9331390642706816,
      "grad_norm": 0.7993467081693396,
      "learning_rate": 1.0989076280498497e-06,
      "loss": 0.3276,
      "step": 17292
    },
    {
      "epoch": 0.9331930278991959,
      "grad_norm": 0.8287510056045735,
      "learning_rate": 1.0987486215591341e-06,
      "loss": 0.3535,
      "step": 17293
    },
    {
      "epoch": 0.9332469915277103,
      "grad_norm": 1.257693218501587,
      "learning_rate": 1.0985897415648292e-06,
      "loss": 0.4823,
      "step": 17294
    },
    {
      "epoch": 0.9333009551562247,
      "grad_norm": 1.121306031977901,
      "learning_rate": 1.0984309880715032e-06,
      "loss": 0.4786,
      "step": 17295
    },
    {
      "epoch": 0.9333549187847391,
      "grad_norm": 0.9252553705902091,
      "learning_rate": 1.0982723610837175e-06,
      "loss": 0.3902,
      "step": 17296
    },
    {
      "epoch": 0.9334088824132535,
      "grad_norm": 0.839664696630705,
      "learning_rate": 1.0981138606060317e-06,
      "loss": 0.347,
      "step": 17297
    },
    {
      "epoch": 0.9334628460417679,
      "grad_norm": 0.8604812978540517,
      "learning_rate": 1.0979554866430013e-06,
      "loss": 0.3627,
      "step": 17298
    },
    {
      "epoch": 0.9335168096702823,
      "grad_norm": 0.9975137402880859,
      "learning_rate": 1.0977972391991785e-06,
      "loss": 0.4892,
      "step": 17299
    },
    {
      "epoch": 0.9335707732987966,
      "grad_norm": 1.0943248339858238,
      "learning_rate": 1.0976391182791117e-06,
      "loss": 0.4674,
      "step": 17300
    },
    {
      "epoch": 0.933624736927311,
      "grad_norm": 0.8991630330645857,
      "learning_rate": 1.097481123887344e-06,
      "loss": 0.2761,
      "step": 17301
    },
    {
      "epoch": 0.9336787005558254,
      "grad_norm": 1.0757337931288524,
      "learning_rate": 1.0973232560284177e-06,
      "loss": 0.4579,
      "step": 17302
    },
    {
      "epoch": 0.9337326641843398,
      "grad_norm": 1.0692363710818704,
      "learning_rate": 1.0971655147068688e-06,
      "loss": 0.4828,
      "step": 17303
    },
    {
      "epoch": 0.9337866278128542,
      "grad_norm": 0.9320224961587416,
      "learning_rate": 1.0970078999272326e-06,
      "loss": 0.4052,
      "step": 17304
    },
    {
      "epoch": 0.9338405914413686,
      "grad_norm": 1.1708782271518505,
      "learning_rate": 1.096850411694038e-06,
      "loss": 0.5298,
      "step": 17305
    },
    {
      "epoch": 0.933894555069883,
      "grad_norm": 1.0385046496227677,
      "learning_rate": 1.0966930500118112e-06,
      "loss": 0.4281,
      "step": 17306
    },
    {
      "epoch": 0.9339485186983972,
      "grad_norm": 1.023965315750552,
      "learning_rate": 1.0965358148850765e-06,
      "loss": 0.5053,
      "step": 17307
    },
    {
      "epoch": 0.9340024823269116,
      "grad_norm": 0.8956742176122396,
      "learning_rate": 1.0963787063183508e-06,
      "loss": 0.3806,
      "step": 17308
    },
    {
      "epoch": 0.934056445955426,
      "grad_norm": 1.028431903032215,
      "learning_rate": 1.0962217243161508e-06,
      "loss": 0.5742,
      "step": 17309
    },
    {
      "epoch": 0.9341104095839404,
      "grad_norm": 0.9880966949317165,
      "learning_rate": 1.0960648688829882e-06,
      "loss": 0.4137,
      "step": 17310
    },
    {
      "epoch": 0.9341643732124548,
      "grad_norm": 1.318045546494377,
      "learning_rate": 1.0959081400233712e-06,
      "loss": 0.2976,
      "step": 17311
    },
    {
      "epoch": 0.9342183368409692,
      "grad_norm": 1.09621462969618,
      "learning_rate": 1.0957515377418044e-06,
      "loss": 0.4035,
      "step": 17312
    },
    {
      "epoch": 0.9342723004694836,
      "grad_norm": 1.2078827649447292,
      "learning_rate": 1.0955950620427883e-06,
      "loss": 0.5537,
      "step": 17313
    },
    {
      "epoch": 0.9343262640979979,
      "grad_norm": 1.0515224366861584,
      "learning_rate": 1.0954387129308208e-06,
      "loss": 0.4372,
      "step": 17314
    },
    {
      "epoch": 0.9343802277265123,
      "grad_norm": 1.0967027208612905,
      "learning_rate": 1.0952824904103948e-06,
      "loss": 0.5038,
      "step": 17315
    },
    {
      "epoch": 0.9344341913550267,
      "grad_norm": 0.8598139248525903,
      "learning_rate": 1.095126394486001e-06,
      "loss": 0.3313,
      "step": 17316
    },
    {
      "epoch": 0.9344881549835411,
      "grad_norm": 1.005043325931441,
      "learning_rate": 1.0949704251621254e-06,
      "loss": 0.3897,
      "step": 17317
    },
    {
      "epoch": 0.9345421186120555,
      "grad_norm": 0.9646680502668209,
      "learning_rate": 1.0948145824432503e-06,
      "loss": 0.4251,
      "step": 17318
    },
    {
      "epoch": 0.9345960822405699,
      "grad_norm": 1.276188620379973,
      "learning_rate": 1.0946588663338552e-06,
      "loss": 0.564,
      "step": 17319
    },
    {
      "epoch": 0.9346500458690843,
      "grad_norm": 1.0213304326100074,
      "learning_rate": 1.094503276838416e-06,
      "loss": 0.5081,
      "step": 17320
    },
    {
      "epoch": 0.9347040094975986,
      "grad_norm": 1.0102188045359872,
      "learning_rate": 1.0943478139614042e-06,
      "loss": 0.5079,
      "step": 17321
    },
    {
      "epoch": 0.934757973126113,
      "grad_norm": 1.032358349859664,
      "learning_rate": 1.0941924777072875e-06,
      "loss": 0.3739,
      "step": 17322
    },
    {
      "epoch": 0.9348119367546274,
      "grad_norm": 0.9352808195075534,
      "learning_rate": 1.0940372680805305e-06,
      "loss": 0.4339,
      "step": 17323
    },
    {
      "epoch": 0.9348659003831418,
      "grad_norm": 0.9761123854245445,
      "learning_rate": 1.0938821850855947e-06,
      "loss": 0.4471,
      "step": 17324
    },
    {
      "epoch": 0.9349198640116562,
      "grad_norm": 0.8462229878271017,
      "learning_rate": 1.0937272287269367e-06,
      "loss": 0.3372,
      "step": 17325
    },
    {
      "epoch": 0.9349738276401706,
      "grad_norm": 1.0407690450066263,
      "learning_rate": 1.0935723990090105e-06,
      "loss": 0.4216,
      "step": 17326
    },
    {
      "epoch": 0.935027791268685,
      "grad_norm": 1.2065963866030907,
      "learning_rate": 1.0934176959362659e-06,
      "loss": 0.5568,
      "step": 17327
    },
    {
      "epoch": 0.9350817548971992,
      "grad_norm": 1.1235905854120753,
      "learning_rate": 1.093263119513149e-06,
      "loss": 0.501,
      "step": 17328
    },
    {
      "epoch": 0.9351357185257136,
      "grad_norm": 1.0296468876434302,
      "learning_rate": 1.0931086697441032e-06,
      "loss": 0.4273,
      "step": 17329
    },
    {
      "epoch": 0.935189682154228,
      "grad_norm": 1.1959373754682798,
      "learning_rate": 1.092954346633567e-06,
      "loss": 0.3959,
      "step": 17330
    },
    {
      "epoch": 0.9352436457827424,
      "grad_norm": 0.9781533465024763,
      "learning_rate": 1.0928001501859757e-06,
      "loss": 0.5428,
      "step": 17331
    },
    {
      "epoch": 0.9352976094112568,
      "grad_norm": 0.8024007871169889,
      "learning_rate": 1.0926460804057614e-06,
      "loss": 0.4676,
      "step": 17332
    },
    {
      "epoch": 0.9353515730397712,
      "grad_norm": 1.083358847962274,
      "learning_rate": 1.0924921372973516e-06,
      "loss": 0.3748,
      "step": 17333
    },
    {
      "epoch": 0.9354055366682856,
      "grad_norm": 0.8859466537156724,
      "learning_rate": 1.0923383208651713e-06,
      "loss": 0.3174,
      "step": 17334
    },
    {
      "epoch": 0.9354595002967999,
      "grad_norm": 0.967545067686146,
      "learning_rate": 1.0921846311136417e-06,
      "loss": 0.3952,
      "step": 17335
    },
    {
      "epoch": 0.9355134639253143,
      "grad_norm": 1.326186446056577,
      "learning_rate": 1.0920310680471792e-06,
      "loss": 0.5043,
      "step": 17336
    },
    {
      "epoch": 0.9355674275538287,
      "grad_norm": 1.0353649065096666,
      "learning_rate": 1.0918776316701978e-06,
      "loss": 0.3937,
      "step": 17337
    },
    {
      "epoch": 0.9356213911823431,
      "grad_norm": 0.8816947652763565,
      "learning_rate": 1.0917243219871077e-06,
      "loss": 0.418,
      "step": 17338
    },
    {
      "epoch": 0.9356753548108575,
      "grad_norm": 1.0487251307851215,
      "learning_rate": 1.0915711390023146e-06,
      "loss": 0.5768,
      "step": 17339
    },
    {
      "epoch": 0.9357293184393719,
      "grad_norm": 1.1638287634432303,
      "learning_rate": 1.091418082720221e-06,
      "loss": 0.3479,
      "step": 17340
    },
    {
      "epoch": 0.9357832820678862,
      "grad_norm": 1.206250616631468,
      "learning_rate": 1.091265153145227e-06,
      "loss": 0.5249,
      "step": 17341
    },
    {
      "epoch": 0.9358372456964006,
      "grad_norm": 1.1562606649507132,
      "learning_rate": 1.0911123502817267e-06,
      "loss": 0.3847,
      "step": 17342
    },
    {
      "epoch": 0.935891209324915,
      "grad_norm": 0.9991782959530328,
      "learning_rate": 1.0909596741341127e-06,
      "loss": 0.3968,
      "step": 17343
    },
    {
      "epoch": 0.9359451729534294,
      "grad_norm": 1.0016780230208167,
      "learning_rate": 1.090807124706772e-06,
      "loss": 0.3587,
      "step": 17344
    },
    {
      "epoch": 0.9359991365819438,
      "grad_norm": 0.9058851754954083,
      "learning_rate": 1.0906547020040901e-06,
      "loss": 0.4171,
      "step": 17345
    },
    {
      "epoch": 0.9360531002104582,
      "grad_norm": 0.9369029717872934,
      "learning_rate": 1.0905024060304475e-06,
      "loss": 0.365,
      "step": 17346
    },
    {
      "epoch": 0.9361070638389726,
      "grad_norm": 0.8084239191607253,
      "learning_rate": 1.0903502367902208e-06,
      "loss": 0.2306,
      "step": 17347
    },
    {
      "epoch": 0.9361610274674869,
      "grad_norm": 1.2338940830837064,
      "learning_rate": 1.090198194287785e-06,
      "loss": 0.5438,
      "step": 17348
    },
    {
      "epoch": 0.9362149910960013,
      "grad_norm": 0.9805079648038706,
      "learning_rate": 1.0900462785275079e-06,
      "loss": 0.3722,
      "step": 17349
    },
    {
      "epoch": 0.9362689547245157,
      "grad_norm": 1.1464342544552608,
      "learning_rate": 1.0898944895137568e-06,
      "loss": 0.5976,
      "step": 17350
    },
    {
      "epoch": 0.93632291835303,
      "grad_norm": 1.1088231383231355,
      "learning_rate": 1.0897428272508939e-06,
      "loss": 0.5493,
      "step": 17351
    },
    {
      "epoch": 0.9363768819815445,
      "grad_norm": 1.035419427688752,
      "learning_rate": 1.0895912917432788e-06,
      "loss": 0.4553,
      "step": 17352
    },
    {
      "epoch": 0.9364308456100588,
      "grad_norm": 1.1094929735141852,
      "learning_rate": 1.0894398829952662e-06,
      "loss": 0.4774,
      "step": 17353
    },
    {
      "epoch": 0.9364848092385732,
      "grad_norm": 0.9466578367412182,
      "learning_rate": 1.0892886010112077e-06,
      "loss": 0.4297,
      "step": 17354
    },
    {
      "epoch": 0.9365387728670875,
      "grad_norm": 1.733687342556291,
      "learning_rate": 1.089137445795452e-06,
      "loss": 0.4569,
      "step": 17355
    },
    {
      "epoch": 0.9365927364956019,
      "grad_norm": 0.8837718947727119,
      "learning_rate": 1.088986417352343e-06,
      "loss": 0.3582,
      "step": 17356
    },
    {
      "epoch": 0.9366467001241163,
      "grad_norm": 1.0545065604719959,
      "learning_rate": 1.0888355156862212e-06,
      "loss": 0.4401,
      "step": 17357
    },
    {
      "epoch": 0.9367006637526307,
      "grad_norm": 1.1683909058766477,
      "learning_rate": 1.0886847408014236e-06,
      "loss": 0.4023,
      "step": 17358
    },
    {
      "epoch": 0.9367546273811451,
      "grad_norm": 0.9822943197379215,
      "learning_rate": 1.0885340927022842e-06,
      "loss": 0.4913,
      "step": 17359
    },
    {
      "epoch": 0.9368085910096595,
      "grad_norm": 0.9519505516386172,
      "learning_rate": 1.0883835713931318e-06,
      "loss": 0.4369,
      "step": 17360
    },
    {
      "epoch": 0.9368625546381739,
      "grad_norm": 1.031649618768374,
      "learning_rate": 1.088233176878294e-06,
      "loss": 0.498,
      "step": 17361
    },
    {
      "epoch": 0.9369165182666882,
      "grad_norm": 1.0280464851993039,
      "learning_rate": 1.0880829091620922e-06,
      "loss": 0.359,
      "step": 17362
    },
    {
      "epoch": 0.9369704818952026,
      "grad_norm": 0.9537800848247804,
      "learning_rate": 1.0879327682488452e-06,
      "loss": 0.3983,
      "step": 17363
    },
    {
      "epoch": 0.937024445523717,
      "grad_norm": 0.9550657718994277,
      "learning_rate": 1.087782754142869e-06,
      "loss": 0.3932,
      "step": 17364
    },
    {
      "epoch": 0.9370784091522314,
      "grad_norm": 1.0088667958978885,
      "learning_rate": 1.0876328668484741e-06,
      "loss": 0.5102,
      "step": 17365
    },
    {
      "epoch": 0.9371323727807458,
      "grad_norm": 0.9771555942378689,
      "learning_rate": 1.0874831063699695e-06,
      "loss": 0.4715,
      "step": 17366
    },
    {
      "epoch": 0.9371863364092602,
      "grad_norm": 1.2294456029194378,
      "learning_rate": 1.0873334727116585e-06,
      "loss": 0.4784,
      "step": 17367
    },
    {
      "epoch": 0.9372403000377746,
      "grad_norm": 1.1435225492865144,
      "learning_rate": 1.0871839658778424e-06,
      "loss": 0.5601,
      "step": 17368
    },
    {
      "epoch": 0.9372942636662889,
      "grad_norm": 0.9851466094571982,
      "learning_rate": 1.0870345858728182e-06,
      "loss": 0.4334,
      "step": 17369
    },
    {
      "epoch": 0.9373482272948033,
      "grad_norm": 1.0388403914319213,
      "learning_rate": 1.0868853327008788e-06,
      "loss": 0.4107,
      "step": 17370
    },
    {
      "epoch": 0.9374021909233177,
      "grad_norm": 0.7694108463261368,
      "learning_rate": 1.0867362063663137e-06,
      "loss": 0.2978,
      "step": 17371
    },
    {
      "epoch": 0.9374561545518321,
      "grad_norm": 0.9117317975729406,
      "learning_rate": 1.0865872068734092e-06,
      "loss": 0.4295,
      "step": 17372
    },
    {
      "epoch": 0.9375101181803465,
      "grad_norm": 1.2237130483140148,
      "learning_rate": 1.0864383342264484e-06,
      "loss": 0.5778,
      "step": 17373
    },
    {
      "epoch": 0.9375640818088609,
      "grad_norm": 1.163291274840639,
      "learning_rate": 1.0862895884297091e-06,
      "loss": 0.4932,
      "step": 17374
    },
    {
      "epoch": 0.9376180454373753,
      "grad_norm": 1.0003931965906736,
      "learning_rate": 1.0861409694874668e-06,
      "loss": 0.3709,
      "step": 17375
    },
    {
      "epoch": 0.9376720090658895,
      "grad_norm": 0.9892953070017084,
      "learning_rate": 1.0859924774039934e-06,
      "loss": 0.4821,
      "step": 17376
    },
    {
      "epoch": 0.9377259726944039,
      "grad_norm": 0.9181756315618775,
      "learning_rate": 1.0858441121835556e-06,
      "loss": 0.3834,
      "step": 17377
    },
    {
      "epoch": 0.9377799363229183,
      "grad_norm": 0.946810878306497,
      "learning_rate": 1.0856958738304186e-06,
      "loss": 0.3552,
      "step": 17378
    },
    {
      "epoch": 0.9378338999514327,
      "grad_norm": 0.9464000715673325,
      "learning_rate": 1.0855477623488423e-06,
      "loss": 0.3773,
      "step": 17379
    },
    {
      "epoch": 0.9378878635799471,
      "grad_norm": 1.227134518858664,
      "learning_rate": 1.0853997777430837e-06,
      "loss": 0.4586,
      "step": 17380
    },
    {
      "epoch": 0.9379418272084615,
      "grad_norm": 0.9330748839347619,
      "learning_rate": 1.085251920017396e-06,
      "loss": 0.3408,
      "step": 17381
    },
    {
      "epoch": 0.9379957908369759,
      "grad_norm": 1.0366177623876756,
      "learning_rate": 1.0851041891760292e-06,
      "loss": 0.3988,
      "step": 17382
    },
    {
      "epoch": 0.9380497544654902,
      "grad_norm": 1.148629041603677,
      "learning_rate": 1.084956585223229e-06,
      "loss": 0.4445,
      "step": 17383
    },
    {
      "epoch": 0.9381037180940046,
      "grad_norm": 1.0574689889780948,
      "learning_rate": 1.0848091081632372e-06,
      "loss": 0.4593,
      "step": 17384
    },
    {
      "epoch": 0.938157681722519,
      "grad_norm": 1.0753221454234463,
      "learning_rate": 1.084661758000293e-06,
      "loss": 0.5109,
      "step": 17385
    },
    {
      "epoch": 0.9382116453510334,
      "grad_norm": 1.1332287640763128,
      "learning_rate": 1.0845145347386313e-06,
      "loss": 0.5417,
      "step": 17386
    },
    {
      "epoch": 0.9382656089795478,
      "grad_norm": 1.0825640839128843,
      "learning_rate": 1.0843674383824836e-06,
      "loss": 0.3649,
      "step": 17387
    },
    {
      "epoch": 0.9383195726080622,
      "grad_norm": 1.1759936865884797,
      "learning_rate": 1.0842204689360773e-06,
      "loss": 0.6493,
      "step": 17388
    },
    {
      "epoch": 0.9383735362365766,
      "grad_norm": 1.053742650757147,
      "learning_rate": 1.0840736264036362e-06,
      "loss": 0.4967,
      "step": 17389
    },
    {
      "epoch": 0.9384274998650909,
      "grad_norm": 1.113705928508162,
      "learning_rate": 1.0839269107893815e-06,
      "loss": 0.4884,
      "step": 17390
    },
    {
      "epoch": 0.9384814634936053,
      "grad_norm": 0.906716693258534,
      "learning_rate": 1.0837803220975296e-06,
      "loss": 0.4423,
      "step": 17391
    },
    {
      "epoch": 0.9385354271221197,
      "grad_norm": 1.062830331852171,
      "learning_rate": 1.0836338603322932e-06,
      "loss": 0.3925,
      "step": 17392
    },
    {
      "epoch": 0.9385893907506341,
      "grad_norm": 1.0624860195009338,
      "learning_rate": 1.0834875254978822e-06,
      "loss": 0.4163,
      "step": 17393
    },
    {
      "epoch": 0.9386433543791485,
      "grad_norm": 1.2574376112007684,
      "learning_rate": 1.083341317598502e-06,
      "loss": 0.5203,
      "step": 17394
    },
    {
      "epoch": 0.9386973180076629,
      "grad_norm": 1.1541638713473246,
      "learning_rate": 1.0831952366383549e-06,
      "loss": 0.4568,
      "step": 17395
    },
    {
      "epoch": 0.9387512816361773,
      "grad_norm": 0.997749482003211,
      "learning_rate": 1.0830492826216405e-06,
      "loss": 0.4047,
      "step": 17396
    },
    {
      "epoch": 0.9388052452646916,
      "grad_norm": 1.1039128185265104,
      "learning_rate": 1.0829034555525522e-06,
      "loss": 0.5303,
      "step": 17397
    },
    {
      "epoch": 0.938859208893206,
      "grad_norm": 1.0535255162626413,
      "learning_rate": 1.0827577554352818e-06,
      "loss": 0.5072,
      "step": 17398
    },
    {
      "epoch": 0.9389131725217204,
      "grad_norm": 1.1991173584109622,
      "learning_rate": 1.0826121822740162e-06,
      "loss": 0.5938,
      "step": 17399
    },
    {
      "epoch": 0.9389671361502347,
      "grad_norm": 1.085107974641101,
      "learning_rate": 1.0824667360729408e-06,
      "loss": 0.5229,
      "step": 17400
    },
    {
      "epoch": 0.9390210997787491,
      "grad_norm": 1.1486466398284558,
      "learning_rate": 1.0823214168362343e-06,
      "loss": 0.5757,
      "step": 17401
    },
    {
      "epoch": 0.9390750634072635,
      "grad_norm": 1.0984857625046873,
      "learning_rate": 1.0821762245680744e-06,
      "loss": 0.6785,
      "step": 17402
    },
    {
      "epoch": 0.9391290270357778,
      "grad_norm": 0.9585735502272028,
      "learning_rate": 1.0820311592726339e-06,
      "loss": 0.3766,
      "step": 17403
    },
    {
      "epoch": 0.9391829906642922,
      "grad_norm": 1.1284454027913027,
      "learning_rate": 1.0818862209540813e-06,
      "loss": 0.5033,
      "step": 17404
    },
    {
      "epoch": 0.9392369542928066,
      "grad_norm": 1.2766011041950835,
      "learning_rate": 1.0817414096165832e-06,
      "loss": 0.5345,
      "step": 17405
    },
    {
      "epoch": 0.939290917921321,
      "grad_norm": 0.8707958888255534,
      "learning_rate": 1.081596725264301e-06,
      "loss": 0.3901,
      "step": 17406
    },
    {
      "epoch": 0.9393448815498354,
      "grad_norm": 0.9647120504237828,
      "learning_rate": 1.081452167901394e-06,
      "loss": 0.358,
      "step": 17407
    },
    {
      "epoch": 0.9393988451783498,
      "grad_norm": 0.8844395353060923,
      "learning_rate": 1.0813077375320158e-06,
      "loss": 0.4195,
      "step": 17408
    },
    {
      "epoch": 0.9394528088068642,
      "grad_norm": 0.9293176741023896,
      "learning_rate": 1.0811634341603184e-06,
      "loss": 0.394,
      "step": 17409
    },
    {
      "epoch": 0.9395067724353785,
      "grad_norm": 1.3134197692893612,
      "learning_rate": 1.0810192577904486e-06,
      "loss": 0.5357,
      "step": 17410
    },
    {
      "epoch": 0.9395607360638929,
      "grad_norm": 0.9076212295919399,
      "learning_rate": 1.0808752084265505e-06,
      "loss": 0.4133,
      "step": 17411
    },
    {
      "epoch": 0.9396146996924073,
      "grad_norm": 1.2650666090173655,
      "learning_rate": 1.0807312860727636e-06,
      "loss": 0.6318,
      "step": 17412
    },
    {
      "epoch": 0.9396686633209217,
      "grad_norm": 1.1461707646831374,
      "learning_rate": 1.0805874907332256e-06,
      "loss": 0.5255,
      "step": 17413
    },
    {
      "epoch": 0.9397226269494361,
      "grad_norm": 0.9052485505109944,
      "learning_rate": 1.0804438224120684e-06,
      "loss": 0.3503,
      "step": 17414
    },
    {
      "epoch": 0.9397765905779505,
      "grad_norm": 0.8831609976170246,
      "learning_rate": 1.080300281113421e-06,
      "loss": 0.3446,
      "step": 17415
    },
    {
      "epoch": 0.9398305542064649,
      "grad_norm": 1.2358757085573848,
      "learning_rate": 1.0801568668414096e-06,
      "loss": 0.4093,
      "step": 17416
    },
    {
      "epoch": 0.9398845178349792,
      "grad_norm": 0.9145291171311667,
      "learning_rate": 1.080013579600156e-06,
      "loss": 0.3794,
      "step": 17417
    },
    {
      "epoch": 0.9399384814634936,
      "grad_norm": 1.464378316901581,
      "learning_rate": 1.079870419393778e-06,
      "loss": 0.6542,
      "step": 17418
    },
    {
      "epoch": 0.939992445092008,
      "grad_norm": 0.7797195896780417,
      "learning_rate": 1.0797273862263903e-06,
      "loss": 0.4027,
      "step": 17419
    },
    {
      "epoch": 0.9400464087205224,
      "grad_norm": 0.9222089823528181,
      "learning_rate": 1.0795844801021037e-06,
      "loss": 0.3671,
      "step": 17420
    },
    {
      "epoch": 0.9401003723490368,
      "grad_norm": 1.0458920747579148,
      "learning_rate": 1.0794417010250258e-06,
      "loss": 0.4104,
      "step": 17421
    },
    {
      "epoch": 0.9401543359775512,
      "grad_norm": 0.9876519761980403,
      "learning_rate": 1.07929904899926e-06,
      "loss": 0.3686,
      "step": 17422
    },
    {
      "epoch": 0.9402082996060656,
      "grad_norm": 1.1939810701902476,
      "learning_rate": 1.0791565240289063e-06,
      "loss": 0.5895,
      "step": 17423
    },
    {
      "epoch": 0.9402622632345798,
      "grad_norm": 0.8325523521679353,
      "learning_rate": 1.0790141261180615e-06,
      "loss": 0.3962,
      "step": 17424
    },
    {
      "epoch": 0.9403162268630942,
      "grad_norm": 1.052476896210643,
      "learning_rate": 1.0788718552708172e-06,
      "loss": 0.4419,
      "step": 17425
    },
    {
      "epoch": 0.9403701904916086,
      "grad_norm": 1.1959243148688423,
      "learning_rate": 1.0787297114912635e-06,
      "loss": 0.583,
      "step": 17426
    },
    {
      "epoch": 0.940424154120123,
      "grad_norm": 0.9013798225756159,
      "learning_rate": 1.0785876947834847e-06,
      "loss": 0.3217,
      "step": 17427
    },
    {
      "epoch": 0.9404781177486374,
      "grad_norm": 0.9509687207712175,
      "learning_rate": 1.0784458051515634e-06,
      "loss": 0.4444,
      "step": 17428
    },
    {
      "epoch": 0.9405320813771518,
      "grad_norm": 1.3817594736621224,
      "learning_rate": 1.078304042599577e-06,
      "loss": 0.5556,
      "step": 17429
    },
    {
      "epoch": 0.9405860450056662,
      "grad_norm": 1.1637515427403333,
      "learning_rate": 1.0781624071316006e-06,
      "loss": 0.4102,
      "step": 17430
    },
    {
      "epoch": 0.9406400086341805,
      "grad_norm": 1.0019392158731615,
      "learning_rate": 1.0780208987517041e-06,
      "loss": 0.4848,
      "step": 17431
    },
    {
      "epoch": 0.9406939722626949,
      "grad_norm": 1.176316262226128,
      "learning_rate": 1.0778795174639555e-06,
      "loss": 0.5086,
      "step": 17432
    },
    {
      "epoch": 0.9407479358912093,
      "grad_norm": 1.0184269740411365,
      "learning_rate": 1.0777382632724181e-06,
      "loss": 0.418,
      "step": 17433
    },
    {
      "epoch": 0.9408018995197237,
      "grad_norm": 1.1241779303072872,
      "learning_rate": 1.077597136181151e-06,
      "loss": 0.3524,
      "step": 17434
    },
    {
      "epoch": 0.9408558631482381,
      "grad_norm": 1.100817884753176,
      "learning_rate": 1.0774561361942102e-06,
      "loss": 0.5107,
      "step": 17435
    },
    {
      "epoch": 0.9409098267767525,
      "grad_norm": 1.1181688777828696,
      "learning_rate": 1.0773152633156498e-06,
      "loss": 0.4124,
      "step": 17436
    },
    {
      "epoch": 0.9409637904052669,
      "grad_norm": 0.917681286226624,
      "learning_rate": 1.0771745175495167e-06,
      "loss": 0.492,
      "step": 17437
    },
    {
      "epoch": 0.9410177540337812,
      "grad_norm": 1.2540739393980092,
      "learning_rate": 1.077033898899857e-06,
      "loss": 0.4314,
      "step": 17438
    },
    {
      "epoch": 0.9410717176622956,
      "grad_norm": 1.2570786254911834,
      "learning_rate": 1.0768934073707124e-06,
      "loss": 0.4354,
      "step": 17439
    },
    {
      "epoch": 0.94112568129081,
      "grad_norm": 1.4124252878057995,
      "learning_rate": 1.0767530429661204e-06,
      "loss": 0.6175,
      "step": 17440
    },
    {
      "epoch": 0.9411796449193244,
      "grad_norm": 0.8793254917124304,
      "learning_rate": 1.0766128056901152e-06,
      "loss": 0.2811,
      "step": 17441
    },
    {
      "epoch": 0.9412336085478388,
      "grad_norm": 0.9981789171755551,
      "learning_rate": 1.0764726955467273e-06,
      "loss": 0.3626,
      "step": 17442
    },
    {
      "epoch": 0.9412875721763532,
      "grad_norm": 0.970804315546047,
      "learning_rate": 1.0763327125399842e-06,
      "loss": 0.4832,
      "step": 17443
    },
    {
      "epoch": 0.9413415358048676,
      "grad_norm": 1.006026357254184,
      "learning_rate": 1.0761928566739082e-06,
      "loss": 0.3932,
      "step": 17444
    },
    {
      "epoch": 0.9413954994333819,
      "grad_norm": 1.2446846373686593,
      "learning_rate": 1.0760531279525202e-06,
      "loss": 0.5419,
      "step": 17445
    },
    {
      "epoch": 0.9414494630618963,
      "grad_norm": 0.8979209940790261,
      "learning_rate": 1.0759135263798349e-06,
      "loss": 0.3624,
      "step": 17446
    },
    {
      "epoch": 0.9415034266904106,
      "grad_norm": 1.14922461168997,
      "learning_rate": 1.075774051959865e-06,
      "loss": 0.5505,
      "step": 17447
    },
    {
      "epoch": 0.941557390318925,
      "grad_norm": 1.2232568171282814,
      "learning_rate": 1.0756347046966192e-06,
      "loss": 0.5831,
      "step": 17448
    },
    {
      "epoch": 0.9416113539474394,
      "grad_norm": 1.0817623423195535,
      "learning_rate": 1.0754954845941026e-06,
      "loss": 0.4528,
      "step": 17449
    },
    {
      "epoch": 0.9416653175759538,
      "grad_norm": 0.9853619957234634,
      "learning_rate": 1.0753563916563165e-06,
      "loss": 0.2971,
      "step": 17450
    },
    {
      "epoch": 0.9417192812044682,
      "grad_norm": 1.206265962956966,
      "learning_rate": 1.0752174258872582e-06,
      "loss": 0.4867,
      "step": 17451
    },
    {
      "epoch": 0.9417732448329825,
      "grad_norm": 0.9770687171476103,
      "learning_rate": 1.0750785872909226e-06,
      "loss": 0.4861,
      "step": 17452
    },
    {
      "epoch": 0.9418272084614969,
      "grad_norm": 0.9213204478559972,
      "learning_rate": 1.0749398758712987e-06,
      "loss": 0.3959,
      "step": 17453
    },
    {
      "epoch": 0.9418811720900113,
      "grad_norm": 1.261965679402492,
      "learning_rate": 1.0748012916323747e-06,
      "loss": 0.4515,
      "step": 17454
    },
    {
      "epoch": 0.9419351357185257,
      "grad_norm": 1.233065448617164,
      "learning_rate": 1.0746628345781323e-06,
      "loss": 0.5819,
      "step": 17455
    },
    {
      "epoch": 0.9419890993470401,
      "grad_norm": 1.3106406699152944,
      "learning_rate": 1.0745245047125516e-06,
      "loss": 0.5164,
      "step": 17456
    },
    {
      "epoch": 0.9420430629755545,
      "grad_norm": 0.8455112173572287,
      "learning_rate": 1.0743863020396084e-06,
      "loss": 0.2847,
      "step": 17457
    },
    {
      "epoch": 0.9420970266040689,
      "grad_norm": 1.0244272919907735,
      "learning_rate": 1.0742482265632744e-06,
      "loss": 0.4901,
      "step": 17458
    },
    {
      "epoch": 0.9421509902325832,
      "grad_norm": 1.136795616190668,
      "learning_rate": 1.074110278287519e-06,
      "loss": 0.5852,
      "step": 17459
    },
    {
      "epoch": 0.9422049538610976,
      "grad_norm": 1.1420590831941488,
      "learning_rate": 1.0739724572163057e-06,
      "loss": 0.5316,
      "step": 17460
    },
    {
      "epoch": 0.942258917489612,
      "grad_norm": 0.9874084782198072,
      "learning_rate": 1.0738347633535958e-06,
      "loss": 0.4379,
      "step": 17461
    },
    {
      "epoch": 0.9423128811181264,
      "grad_norm": 0.9298717739239738,
      "learning_rate": 1.0736971967033476e-06,
      "loss": 0.3578,
      "step": 17462
    },
    {
      "epoch": 0.9423668447466408,
      "grad_norm": 0.9786934391686091,
      "learning_rate": 1.0735597572695142e-06,
      "loss": 0.507,
      "step": 17463
    },
    {
      "epoch": 0.9424208083751552,
      "grad_norm": 0.9922910674994573,
      "learning_rate": 1.073422445056046e-06,
      "loss": 0.4539,
      "step": 17464
    },
    {
      "epoch": 0.9424747720036696,
      "grad_norm": 1.2788390673274184,
      "learning_rate": 1.0732852600668896e-06,
      "loss": 0.6293,
      "step": 17465
    },
    {
      "epoch": 0.9425287356321839,
      "grad_norm": 1.2944126625171304,
      "learning_rate": 1.073148202305988e-06,
      "loss": 0.4715,
      "step": 17466
    },
    {
      "epoch": 0.9425826992606983,
      "grad_norm": 1.2385237715333102,
      "learning_rate": 1.0730112717772798e-06,
      "loss": 0.6687,
      "step": 17467
    },
    {
      "epoch": 0.9426366628892127,
      "grad_norm": 1.165117535254885,
      "learning_rate": 1.0728744684847007e-06,
      "loss": 0.4253,
      "step": 17468
    },
    {
      "epoch": 0.9426906265177271,
      "grad_norm": 1.1222126617819985,
      "learning_rate": 1.0727377924321825e-06,
      "loss": 0.4945,
      "step": 17469
    },
    {
      "epoch": 0.9427445901462415,
      "grad_norm": 0.9810341291433643,
      "learning_rate": 1.0726012436236537e-06,
      "loss": 0.4917,
      "step": 17470
    },
    {
      "epoch": 0.9427985537747559,
      "grad_norm": 0.9984896125891163,
      "learning_rate": 1.0724648220630387e-06,
      "loss": 0.5047,
      "step": 17471
    },
    {
      "epoch": 0.9428525174032701,
      "grad_norm": 1.0607805272165745,
      "learning_rate": 1.0723285277542586e-06,
      "loss": 0.485,
      "step": 17472
    },
    {
      "epoch": 0.9429064810317845,
      "grad_norm": 1.102093979612336,
      "learning_rate": 1.0721923607012299e-06,
      "loss": 0.4888,
      "step": 17473
    },
    {
      "epoch": 0.9429604446602989,
      "grad_norm": 1.1346747988337693,
      "learning_rate": 1.072056320907867e-06,
      "loss": 0.5951,
      "step": 17474
    },
    {
      "epoch": 0.9430144082888133,
      "grad_norm": 1.1635474516759199,
      "learning_rate": 1.07192040837808e-06,
      "loss": 0.4826,
      "step": 17475
    },
    {
      "epoch": 0.9430683719173277,
      "grad_norm": 1.1072672605258773,
      "learning_rate": 1.0717846231157744e-06,
      "loss": 0.4378,
      "step": 17476
    },
    {
      "epoch": 0.9431223355458421,
      "grad_norm": 0.9348939813565327,
      "learning_rate": 1.0716489651248528e-06,
      "loss": 0.3991,
      "step": 17477
    },
    {
      "epoch": 0.9431762991743565,
      "grad_norm": 1.046904466672987,
      "learning_rate": 1.071513434409215e-06,
      "loss": 0.445,
      "step": 17478
    },
    {
      "epoch": 0.9432302628028708,
      "grad_norm": 1.0959023916400938,
      "learning_rate": 1.0713780309727559e-06,
      "loss": 0.4541,
      "step": 17479
    },
    {
      "epoch": 0.9432842264313852,
      "grad_norm": 0.8030607571122733,
      "learning_rate": 1.0712427548193662e-06,
      "loss": 0.2875,
      "step": 17480
    },
    {
      "epoch": 0.9433381900598996,
      "grad_norm": 0.895934003857164,
      "learning_rate": 1.0711076059529353e-06,
      "loss": 0.317,
      "step": 17481
    },
    {
      "epoch": 0.943392153688414,
      "grad_norm": 0.8558935227871727,
      "learning_rate": 1.0709725843773467e-06,
      "loss": 0.3083,
      "step": 17482
    },
    {
      "epoch": 0.9434461173169284,
      "grad_norm": 0.988726291921667,
      "learning_rate": 1.0708376900964812e-06,
      "loss": 0.4394,
      "step": 17483
    },
    {
      "epoch": 0.9435000809454428,
      "grad_norm": 1.0163670888177232,
      "learning_rate": 1.0707029231142153e-06,
      "loss": 0.5626,
      "step": 17484
    },
    {
      "epoch": 0.9435540445739572,
      "grad_norm": 1.2212543412395898,
      "learning_rate": 1.0705682834344235e-06,
      "loss": 0.5302,
      "step": 17485
    },
    {
      "epoch": 0.9436080082024715,
      "grad_norm": 1.179022527255604,
      "learning_rate": 1.0704337710609745e-06,
      "loss": 0.4327,
      "step": 17486
    },
    {
      "epoch": 0.9436619718309859,
      "grad_norm": 1.352685876292664,
      "learning_rate": 1.070299385997735e-06,
      "loss": 0.5746,
      "step": 17487
    },
    {
      "epoch": 0.9437159354595003,
      "grad_norm": 0.8445142232159811,
      "learning_rate": 1.070165128248567e-06,
      "loss": 0.3344,
      "step": 17488
    },
    {
      "epoch": 0.9437698990880147,
      "grad_norm": 1.3041571846027975,
      "learning_rate": 1.070030997817329e-06,
      "loss": 0.3908,
      "step": 17489
    },
    {
      "epoch": 0.9438238627165291,
      "grad_norm": 1.163952804179614,
      "learning_rate": 1.0698969947078763e-06,
      "loss": 0.5228,
      "step": 17490
    },
    {
      "epoch": 0.9438778263450435,
      "grad_norm": 0.7919936556816627,
      "learning_rate": 1.0697631189240606e-06,
      "loss": 0.2409,
      "step": 17491
    },
    {
      "epoch": 0.9439317899735579,
      "grad_norm": 0.9680736874425601,
      "learning_rate": 1.0696293704697288e-06,
      "loss": 0.4938,
      "step": 17492
    },
    {
      "epoch": 0.9439857536020722,
      "grad_norm": 0.8938878562788537,
      "learning_rate": 1.0694957493487258e-06,
      "loss": 0.3746,
      "step": 17493
    },
    {
      "epoch": 0.9440397172305866,
      "grad_norm": 1.1889658585241398,
      "learning_rate": 1.0693622555648918e-06,
      "loss": 0.4858,
      "step": 17494
    },
    {
      "epoch": 0.944093680859101,
      "grad_norm": 1.1671254435836316,
      "learning_rate": 1.0692288891220629e-06,
      "loss": 0.6597,
      "step": 17495
    },
    {
      "epoch": 0.9441476444876153,
      "grad_norm": 0.8851486100435733,
      "learning_rate": 1.0690956500240733e-06,
      "loss": 0.3159,
      "step": 17496
    },
    {
      "epoch": 0.9442016081161297,
      "grad_norm": 1.0063725542891657,
      "learning_rate": 1.0689625382747511e-06,
      "loss": 0.414,
      "step": 17497
    },
    {
      "epoch": 0.9442555717446441,
      "grad_norm": 0.9328836909399898,
      "learning_rate": 1.0688295538779236e-06,
      "loss": 0.3431,
      "step": 17498
    },
    {
      "epoch": 0.9443095353731585,
      "grad_norm": 1.0453397583896495,
      "learning_rate": 1.0686966968374116e-06,
      "loss": 0.6067,
      "step": 17499
    },
    {
      "epoch": 0.9443634990016728,
      "grad_norm": 1.1653692126429538,
      "learning_rate": 1.0685639671570345e-06,
      "loss": 0.5373,
      "step": 17500
    },
    {
      "epoch": 0.9443634990016728,
      "eval_loss": 0.5278103947639465,
      "eval_runtime": 165.3254,
      "eval_samples_per_second": 20.801,
      "eval_steps_per_second": 0.871,
      "step": 17500
    },
    {
      "epoch": 0.9444174626301872,
      "grad_norm": 0.919340152874765,
      "learning_rate": 1.0684313648406057e-06,
      "loss": 0.5487,
      "step": 17501
    },
    {
      "epoch": 0.9444714262587016,
      "grad_norm": 0.9150036836390414,
      "learning_rate": 1.068298889891938e-06,
      "loss": 0.4556,
      "step": 17502
    },
    {
      "epoch": 0.944525389887216,
      "grad_norm": 1.155778563432956,
      "learning_rate": 1.0681665423148379e-06,
      "loss": 0.5486,
      "step": 17503
    },
    {
      "epoch": 0.9445793535157304,
      "grad_norm": 1.0435667597772245,
      "learning_rate": 1.0680343221131096e-06,
      "loss": 0.5244,
      "step": 17504
    },
    {
      "epoch": 0.9446333171442448,
      "grad_norm": 0.9736620268372393,
      "learning_rate": 1.0679022292905525e-06,
      "loss": 0.4842,
      "step": 17505
    },
    {
      "epoch": 0.9446872807727592,
      "grad_norm": 1.0361002247771984,
      "learning_rate": 1.067770263850964e-06,
      "loss": 0.4402,
      "step": 17506
    },
    {
      "epoch": 0.9447412444012735,
      "grad_norm": 1.028144567229061,
      "learning_rate": 1.0676384257981372e-06,
      "loss": 0.4014,
      "step": 17507
    },
    {
      "epoch": 0.9447952080297879,
      "grad_norm": 1.1525582286743679,
      "learning_rate": 1.0675067151358601e-06,
      "loss": 0.4911,
      "step": 17508
    },
    {
      "epoch": 0.9448491716583023,
      "grad_norm": 1.0147048166925388,
      "learning_rate": 1.0673751318679185e-06,
      "loss": 0.4119,
      "step": 17509
    },
    {
      "epoch": 0.9449031352868167,
      "grad_norm": 0.8644611896719882,
      "learning_rate": 1.0672436759980946e-06,
      "loss": 0.3423,
      "step": 17510
    },
    {
      "epoch": 0.9449570989153311,
      "grad_norm": 1.0256899491736409,
      "learning_rate": 1.0671123475301659e-06,
      "loss": 0.5553,
      "step": 17511
    },
    {
      "epoch": 0.9450110625438455,
      "grad_norm": 1.1390558002464215,
      "learning_rate": 1.066981146467908e-06,
      "loss": 0.543,
      "step": 17512
    },
    {
      "epoch": 0.9450650261723599,
      "grad_norm": 0.9660885293403647,
      "learning_rate": 1.0668500728150914e-06,
      "loss": 0.3982,
      "step": 17513
    },
    {
      "epoch": 0.9451189898008742,
      "grad_norm": 0.7934473405406189,
      "learning_rate": 1.0667191265754829e-06,
      "loss": 0.3691,
      "step": 17514
    },
    {
      "epoch": 0.9451729534293886,
      "grad_norm": 0.9574450685398503,
      "learning_rate": 1.0665883077528458e-06,
      "loss": 0.4128,
      "step": 17515
    },
    {
      "epoch": 0.945226917057903,
      "grad_norm": 0.8788458067952307,
      "learning_rate": 1.0664576163509409e-06,
      "loss": 0.3994,
      "step": 17516
    },
    {
      "epoch": 0.9452808806864174,
      "grad_norm": 0.9432924275034762,
      "learning_rate": 1.0663270523735237e-06,
      "loss": 0.4187,
      "step": 17517
    },
    {
      "epoch": 0.9453348443149318,
      "grad_norm": 1.1377993465063772,
      "learning_rate": 1.0661966158243464e-06,
      "loss": 0.4816,
      "step": 17518
    },
    {
      "epoch": 0.9453888079434462,
      "grad_norm": 0.933365555393075,
      "learning_rate": 1.0660663067071592e-06,
      "loss": 0.4845,
      "step": 17519
    },
    {
      "epoch": 0.9454427715719605,
      "grad_norm": 0.9528193697523573,
      "learning_rate": 1.065936125025706e-06,
      "loss": 0.3789,
      "step": 17520
    },
    {
      "epoch": 0.9454967352004748,
      "grad_norm": 0.9847312305441132,
      "learning_rate": 1.0658060707837291e-06,
      "loss": 0.4155,
      "step": 17521
    },
    {
      "epoch": 0.9455506988289892,
      "grad_norm": 0.9716974289218435,
      "learning_rate": 1.0656761439849665e-06,
      "loss": 0.2857,
      "step": 17522
    },
    {
      "epoch": 0.9456046624575036,
      "grad_norm": 0.9876711267398114,
      "learning_rate": 1.0655463446331512e-06,
      "loss": 0.4961,
      "step": 17523
    },
    {
      "epoch": 0.945658626086018,
      "grad_norm": 0.9008783420411685,
      "learning_rate": 1.0654166727320155e-06,
      "loss": 0.393,
      "step": 17524
    },
    {
      "epoch": 0.9457125897145324,
      "grad_norm": 1.0488967404853808,
      "learning_rate": 1.0652871282852848e-06,
      "loss": 0.4367,
      "step": 17525
    },
    {
      "epoch": 0.9457665533430468,
      "grad_norm": 1.0189066070846666,
      "learning_rate": 1.065157711296683e-06,
      "loss": 0.4212,
      "step": 17526
    },
    {
      "epoch": 0.9458205169715612,
      "grad_norm": 0.6336356854142714,
      "learning_rate": 1.06502842176993e-06,
      "loss": 0.2625,
      "step": 17527
    },
    {
      "epoch": 0.9458744806000755,
      "grad_norm": 0.983526769121989,
      "learning_rate": 1.0648992597087414e-06,
      "loss": 0.4035,
      "step": 17528
    },
    {
      "epoch": 0.9459284442285899,
      "grad_norm": 0.9823036754445538,
      "learning_rate": 1.0647702251168294e-06,
      "loss": 0.4695,
      "step": 17529
    },
    {
      "epoch": 0.9459824078571043,
      "grad_norm": 0.9601582275869454,
      "learning_rate": 1.0646413179979024e-06,
      "loss": 0.4189,
      "step": 17530
    },
    {
      "epoch": 0.9460363714856187,
      "grad_norm": 1.162472408342118,
      "learning_rate": 1.0645125383556658e-06,
      "loss": 0.6454,
      "step": 17531
    },
    {
      "epoch": 0.9460903351141331,
      "grad_norm": 1.0278878371578082,
      "learning_rate": 1.06438388619382e-06,
      "loss": 0.5125,
      "step": 17532
    },
    {
      "epoch": 0.9461442987426475,
      "grad_norm": 1.1352965342578698,
      "learning_rate": 1.0642553615160634e-06,
      "loss": 0.45,
      "step": 17533
    },
    {
      "epoch": 0.9461982623711619,
      "grad_norm": 1.0802149010826647,
      "learning_rate": 1.0641269643260898e-06,
      "loss": 0.5747,
      "step": 17534
    },
    {
      "epoch": 0.9462522259996762,
      "grad_norm": 1.144835587920776,
      "learning_rate": 1.0639986946275896e-06,
      "loss": 0.3966,
      "step": 17535
    },
    {
      "epoch": 0.9463061896281906,
      "grad_norm": 0.863545944178848,
      "learning_rate": 1.0638705524242487e-06,
      "loss": 0.3088,
      "step": 17536
    },
    {
      "epoch": 0.946360153256705,
      "grad_norm": 1.0342548441944406,
      "learning_rate": 1.063742537719751e-06,
      "loss": 0.4181,
      "step": 17537
    },
    {
      "epoch": 0.9464141168852194,
      "grad_norm": 1.161319569984265,
      "learning_rate": 1.063614650517775e-06,
      "loss": 0.4243,
      "step": 17538
    },
    {
      "epoch": 0.9464680805137338,
      "grad_norm": 0.947729188106141,
      "learning_rate": 1.063486890821997e-06,
      "loss": 0.3993,
      "step": 17539
    },
    {
      "epoch": 0.9465220441422482,
      "grad_norm": 0.9318621259595681,
      "learning_rate": 1.0633592586360877e-06,
      "loss": 0.3539,
      "step": 17540
    },
    {
      "epoch": 0.9465760077707625,
      "grad_norm": 1.0512478474523645,
      "learning_rate": 1.063231753963717e-06,
      "loss": 0.4573,
      "step": 17541
    },
    {
      "epoch": 0.9466299713992768,
      "grad_norm": 1.2036629277734847,
      "learning_rate": 1.0631043768085485e-06,
      "loss": 0.6013,
      "step": 17542
    },
    {
      "epoch": 0.9466839350277912,
      "grad_norm": 0.9416817520148757,
      "learning_rate": 1.0629771271742433e-06,
      "loss": 0.3835,
      "step": 17543
    },
    {
      "epoch": 0.9467378986563056,
      "grad_norm": 1.0986199190501218,
      "learning_rate": 1.0628500050644586e-06,
      "loss": 0.47,
      "step": 17544
    },
    {
      "epoch": 0.94679186228482,
      "grad_norm": 0.9808269933649015,
      "learning_rate": 1.0627230104828484e-06,
      "loss": 0.4299,
      "step": 17545
    },
    {
      "epoch": 0.9468458259133344,
      "grad_norm": 1.0006281936131571,
      "learning_rate": 1.0625961434330622e-06,
      "loss": 0.4489,
      "step": 17546
    },
    {
      "epoch": 0.9468997895418488,
      "grad_norm": 1.099572255639449,
      "learning_rate": 1.0624694039187464e-06,
      "loss": 0.4566,
      "step": 17547
    },
    {
      "epoch": 0.9469537531703631,
      "grad_norm": 0.9223443952339713,
      "learning_rate": 1.0623427919435442e-06,
      "loss": 0.4004,
      "step": 17548
    },
    {
      "epoch": 0.9470077167988775,
      "grad_norm": 0.86231144845256,
      "learning_rate": 1.062216307511094e-06,
      "loss": 0.3584,
      "step": 17549
    },
    {
      "epoch": 0.9470616804273919,
      "grad_norm": 1.093046457625342,
      "learning_rate": 1.0620899506250306e-06,
      "loss": 0.4756,
      "step": 17550
    },
    {
      "epoch": 0.9471156440559063,
      "grad_norm": 1.1680058494490955,
      "learning_rate": 1.0619637212889868e-06,
      "loss": 0.4151,
      "step": 17551
    },
    {
      "epoch": 0.9471696076844207,
      "grad_norm": 0.8517988101754712,
      "learning_rate": 1.0618376195065896e-06,
      "loss": 0.3505,
      "step": 17552
    },
    {
      "epoch": 0.9472235713129351,
      "grad_norm": 1.0393381640380404,
      "learning_rate": 1.0617116452814639e-06,
      "loss": 0.4632,
      "step": 17553
    },
    {
      "epoch": 0.9472775349414495,
      "grad_norm": 1.064498866760536,
      "learning_rate": 1.0615857986172298e-06,
      "loss": 0.4316,
      "step": 17554
    },
    {
      "epoch": 0.9473314985699638,
      "grad_norm": 1.1094018618912822,
      "learning_rate": 1.0614600795175052e-06,
      "loss": 0.4617,
      "step": 17555
    },
    {
      "epoch": 0.9473854621984782,
      "grad_norm": 0.9247202620547346,
      "learning_rate": 1.0613344879859023e-06,
      "loss": 0.3885,
      "step": 17556
    },
    {
      "epoch": 0.9474394258269926,
      "grad_norm": 0.9865073565869363,
      "learning_rate": 1.0612090240260308e-06,
      "loss": 0.4632,
      "step": 17557
    },
    {
      "epoch": 0.947493389455507,
      "grad_norm": 1.0499795225951762,
      "learning_rate": 1.0610836876414973e-06,
      "loss": 0.4581,
      "step": 17558
    },
    {
      "epoch": 0.9475473530840214,
      "grad_norm": 1.1342490858315624,
      "learning_rate": 1.060958478835904e-06,
      "loss": 0.4816,
      "step": 17559
    },
    {
      "epoch": 0.9476013167125358,
      "grad_norm": 1.117976623299202,
      "learning_rate": 1.060833397612849e-06,
      "loss": 0.5111,
      "step": 17560
    },
    {
      "epoch": 0.9476552803410502,
      "grad_norm": 0.8930671083412475,
      "learning_rate": 1.0607084439759282e-06,
      "loss": 0.3523,
      "step": 17561
    },
    {
      "epoch": 0.9477092439695645,
      "grad_norm": 0.972350348590994,
      "learning_rate": 1.060583617928732e-06,
      "loss": 0.4514,
      "step": 17562
    },
    {
      "epoch": 0.9477632075980789,
      "grad_norm": 0.8689446205790831,
      "learning_rate": 1.0604589194748482e-06,
      "loss": 0.4961,
      "step": 17563
    },
    {
      "epoch": 0.9478171712265933,
      "grad_norm": 1.1688850461171667,
      "learning_rate": 1.060334348617861e-06,
      "loss": 0.4595,
      "step": 17564
    },
    {
      "epoch": 0.9478711348551077,
      "grad_norm": 1.0500546007362237,
      "learning_rate": 1.06020990536135e-06,
      "loss": 0.3544,
      "step": 17565
    },
    {
      "epoch": 0.947925098483622,
      "grad_norm": 1.093478386678579,
      "learning_rate": 1.0600855897088933e-06,
      "loss": 0.7618,
      "step": 17566
    },
    {
      "epoch": 0.9479790621121365,
      "grad_norm": 1.0472726264363874,
      "learning_rate": 1.059961401664062e-06,
      "loss": 0.5343,
      "step": 17567
    },
    {
      "epoch": 0.9480330257406508,
      "grad_norm": 1.0174380401161265,
      "learning_rate": 1.0598373412304269e-06,
      "loss": 0.3745,
      "step": 17568
    },
    {
      "epoch": 0.9480869893691651,
      "grad_norm": 0.9701479959311213,
      "learning_rate": 1.0597134084115535e-06,
      "loss": 0.3435,
      "step": 17569
    },
    {
      "epoch": 0.9481409529976795,
      "grad_norm": 1.0758037337031652,
      "learning_rate": 1.0595896032110028e-06,
      "loss": 0.4351,
      "step": 17570
    },
    {
      "epoch": 0.9481949166261939,
      "grad_norm": 1.3877638457066415,
      "learning_rate": 1.0594659256323337e-06,
      "loss": 0.5476,
      "step": 17571
    },
    {
      "epoch": 0.9482488802547083,
      "grad_norm": 0.9599622371717922,
      "learning_rate": 1.0593423756791007e-06,
      "loss": 0.4404,
      "step": 17572
    },
    {
      "epoch": 0.9483028438832227,
      "grad_norm": 0.971526332635944,
      "learning_rate": 1.059218953354855e-06,
      "loss": 0.4025,
      "step": 17573
    },
    {
      "epoch": 0.9483568075117371,
      "grad_norm": 1.193566931817505,
      "learning_rate": 1.0590956586631432e-06,
      "loss": 0.4367,
      "step": 17574
    },
    {
      "epoch": 0.9484107711402515,
      "grad_norm": 1.1450846100184533,
      "learning_rate": 1.0589724916075096e-06,
      "loss": 0.6168,
      "step": 17575
    },
    {
      "epoch": 0.9484647347687658,
      "grad_norm": 0.9571375163285176,
      "learning_rate": 1.0588494521914944e-06,
      "loss": 0.3829,
      "step": 17576
    },
    {
      "epoch": 0.9485186983972802,
      "grad_norm": 1.011596998126651,
      "learning_rate": 1.0587265404186327e-06,
      "loss": 0.4433,
      "step": 17577
    },
    {
      "epoch": 0.9485726620257946,
      "grad_norm": 1.4301156770818677,
      "learning_rate": 1.058603756292458e-06,
      "loss": 0.6002,
      "step": 17578
    },
    {
      "epoch": 0.948626625654309,
      "grad_norm": 1.0528021604406168,
      "learning_rate": 1.0584810998164994e-06,
      "loss": 0.4737,
      "step": 17579
    },
    {
      "epoch": 0.9486805892828234,
      "grad_norm": 1.1104296407052574,
      "learning_rate": 1.0583585709942815e-06,
      "loss": 0.524,
      "step": 17580
    },
    {
      "epoch": 0.9487345529113378,
      "grad_norm": 0.8814833968636292,
      "learning_rate": 1.0582361698293258e-06,
      "loss": 0.3353,
      "step": 17581
    },
    {
      "epoch": 0.9487885165398522,
      "grad_norm": 1.1161010211198492,
      "learning_rate": 1.058113896325151e-06,
      "loss": 0.5981,
      "step": 17582
    },
    {
      "epoch": 0.9488424801683665,
      "grad_norm": 1.2077814905368793,
      "learning_rate": 1.0579917504852715e-06,
      "loss": 0.5307,
      "step": 17583
    },
    {
      "epoch": 0.9488964437968809,
      "grad_norm": 1.2096194317661646,
      "learning_rate": 1.057869732313197e-06,
      "loss": 0.5494,
      "step": 17584
    },
    {
      "epoch": 0.9489504074253953,
      "grad_norm": 0.9647337152071195,
      "learning_rate": 1.057747841812435e-06,
      "loss": 0.3488,
      "step": 17585
    },
    {
      "epoch": 0.9490043710539097,
      "grad_norm": 1.1277414698847856,
      "learning_rate": 1.0576260789864887e-06,
      "loss": 0.5059,
      "step": 17586
    },
    {
      "epoch": 0.9490583346824241,
      "grad_norm": 1.2171344550142191,
      "learning_rate": 1.0575044438388571e-06,
      "loss": 0.5606,
      "step": 17587
    },
    {
      "epoch": 0.9491122983109385,
      "grad_norm": 1.1839853163796918,
      "learning_rate": 1.0573829363730367e-06,
      "loss": 0.6986,
      "step": 17588
    },
    {
      "epoch": 0.9491662619394529,
      "grad_norm": 1.1057264440704324,
      "learning_rate": 1.05726155659252e-06,
      "loss": 0.4344,
      "step": 17589
    },
    {
      "epoch": 0.9492202255679671,
      "grad_norm": 1.0053517225601263,
      "learning_rate": 1.0571403045007948e-06,
      "loss": 0.4158,
      "step": 17590
    },
    {
      "epoch": 0.9492741891964815,
      "grad_norm": 0.9632841273201004,
      "learning_rate": 1.0570191801013466e-06,
      "loss": 0.5073,
      "step": 17591
    },
    {
      "epoch": 0.9493281528249959,
      "grad_norm": 0.8528989255233609,
      "learning_rate": 1.0568981833976566e-06,
      "loss": 0.3174,
      "step": 17592
    },
    {
      "epoch": 0.9493821164535103,
      "grad_norm": 1.0992333611789769,
      "learning_rate": 1.0567773143932019e-06,
      "loss": 0.5261,
      "step": 17593
    },
    {
      "epoch": 0.9494360800820247,
      "grad_norm": 0.9180169721633795,
      "learning_rate": 1.0566565730914572e-06,
      "loss": 0.3734,
      "step": 17594
    },
    {
      "epoch": 0.9494900437105391,
      "grad_norm": 1.05186056575915,
      "learning_rate": 1.0565359594958921e-06,
      "loss": 0.4177,
      "step": 17595
    },
    {
      "epoch": 0.9495440073390535,
      "grad_norm": 1.0898070464731402,
      "learning_rate": 1.0564154736099733e-06,
      "loss": 0.4927,
      "step": 17596
    },
    {
      "epoch": 0.9495979709675678,
      "grad_norm": 0.8980480215430352,
      "learning_rate": 1.0562951154371636e-06,
      "loss": 0.3933,
      "step": 17597
    },
    {
      "epoch": 0.9496519345960822,
      "grad_norm": 1.0357125731852825,
      "learning_rate": 1.0561748849809227e-06,
      "loss": 0.4825,
      "step": 17598
    },
    {
      "epoch": 0.9497058982245966,
      "grad_norm": 1.3026461077253615,
      "learning_rate": 1.0560547822447053e-06,
      "loss": 0.6274,
      "step": 17599
    },
    {
      "epoch": 0.949759861853111,
      "grad_norm": 1.060659815126623,
      "learning_rate": 1.0559348072319639e-06,
      "loss": 0.4234,
      "step": 17600
    },
    {
      "epoch": 0.9498138254816254,
      "grad_norm": 0.9206066763629969,
      "learning_rate": 1.0558149599461462e-06,
      "loss": 0.4403,
      "step": 17601
    },
    {
      "epoch": 0.9498677891101398,
      "grad_norm": 0.9553472412905634,
      "learning_rate": 1.0556952403906978e-06,
      "loss": 0.3699,
      "step": 17602
    },
    {
      "epoch": 0.9499217527386542,
      "grad_norm": 1.011732583421941,
      "learning_rate": 1.055575648569058e-06,
      "loss": 0.3778,
      "step": 17603
    },
    {
      "epoch": 0.9499757163671685,
      "grad_norm": 0.9875932311424257,
      "learning_rate": 1.0554561844846657e-06,
      "loss": 0.3969,
      "step": 17604
    },
    {
      "epoch": 0.9500296799956829,
      "grad_norm": 1.03024449031879,
      "learning_rate": 1.0553368481409528e-06,
      "loss": 0.3203,
      "step": 17605
    },
    {
      "epoch": 0.9500836436241973,
      "grad_norm": 1.1260197738062632,
      "learning_rate": 1.0552176395413502e-06,
      "loss": 0.4982,
      "step": 17606
    },
    {
      "epoch": 0.9501376072527117,
      "grad_norm": 1.0769001657135813,
      "learning_rate": 1.0550985586892838e-06,
      "loss": 0.5221,
      "step": 17607
    },
    {
      "epoch": 0.9501915708812261,
      "grad_norm": 1.0292782108295273,
      "learning_rate": 1.054979605588176e-06,
      "loss": 0.3786,
      "step": 17608
    },
    {
      "epoch": 0.9502455345097405,
      "grad_norm": 1.0263272246545248,
      "learning_rate": 1.0548607802414458e-06,
      "loss": 0.4202,
      "step": 17609
    },
    {
      "epoch": 0.9502994981382548,
      "grad_norm": 1.0199168901497793,
      "learning_rate": 1.0547420826525085e-06,
      "loss": 0.4217,
      "step": 17610
    },
    {
      "epoch": 0.9503534617667692,
      "grad_norm": 1.0903832026451736,
      "learning_rate": 1.0546235128247748e-06,
      "loss": 0.4922,
      "step": 17611
    },
    {
      "epoch": 0.9504074253952836,
      "grad_norm": 1.250600295550576,
      "learning_rate": 1.054505070761653e-06,
      "loss": 0.568,
      "step": 17612
    },
    {
      "epoch": 0.950461389023798,
      "grad_norm": 0.7999368929545015,
      "learning_rate": 1.054386756466548e-06,
      "loss": 0.2575,
      "step": 17613
    },
    {
      "epoch": 0.9505153526523124,
      "grad_norm": 1.0003526289890181,
      "learning_rate": 1.0542685699428593e-06,
      "loss": 0.3298,
      "step": 17614
    },
    {
      "epoch": 0.9505693162808267,
      "grad_norm": 1.2882564006627486,
      "learning_rate": 1.0541505111939832e-06,
      "loss": 0.6094,
      "step": 17615
    },
    {
      "epoch": 0.9506232799093411,
      "grad_norm": 0.8942514990971155,
      "learning_rate": 1.0540325802233143e-06,
      "loss": 0.3076,
      "step": 17616
    },
    {
      "epoch": 0.9506772435378554,
      "grad_norm": 0.8664887793453793,
      "learning_rate": 1.0539147770342416e-06,
      "loss": 0.4171,
      "step": 17617
    },
    {
      "epoch": 0.9507312071663698,
      "grad_norm": 1.0380235114048717,
      "learning_rate": 1.0537971016301507e-06,
      "loss": 0.3839,
      "step": 17618
    },
    {
      "epoch": 0.9507851707948842,
      "grad_norm": 1.1934669675639038,
      "learning_rate": 1.0536795540144232e-06,
      "loss": 0.6089,
      "step": 17619
    },
    {
      "epoch": 0.9508391344233986,
      "grad_norm": 1.119083941449759,
      "learning_rate": 1.053562134190438e-06,
      "loss": 0.5283,
      "step": 17620
    },
    {
      "epoch": 0.950893098051913,
      "grad_norm": 1.0166359927264608,
      "learning_rate": 1.0534448421615702e-06,
      "loss": 0.4177,
      "step": 17621
    },
    {
      "epoch": 0.9509470616804274,
      "grad_norm": 1.121585250181695,
      "learning_rate": 1.0533276779311906e-06,
      "loss": 0.4544,
      "step": 17622
    },
    {
      "epoch": 0.9510010253089418,
      "grad_norm": 1.1036883059044593,
      "learning_rate": 1.053210641502666e-06,
      "loss": 0.5292,
      "step": 17623
    },
    {
      "epoch": 0.9510549889374561,
      "grad_norm": 0.9160297024000854,
      "learning_rate": 1.0530937328793615e-06,
      "loss": 0.3979,
      "step": 17624
    },
    {
      "epoch": 0.9511089525659705,
      "grad_norm": 1.1817338968566953,
      "learning_rate": 1.052976952064636e-06,
      "loss": 0.5171,
      "step": 17625
    },
    {
      "epoch": 0.9511629161944849,
      "grad_norm": 1.008804934363091,
      "learning_rate": 1.0528602990618459e-06,
      "loss": 0.5387,
      "step": 17626
    },
    {
      "epoch": 0.9512168798229993,
      "grad_norm": 1.0010144745255987,
      "learning_rate": 1.0527437738743444e-06,
      "loss": 0.4229,
      "step": 17627
    },
    {
      "epoch": 0.9512708434515137,
      "grad_norm": 0.9309813156841535,
      "learning_rate": 1.052627376505481e-06,
      "loss": 0.3715,
      "step": 17628
    },
    {
      "epoch": 0.9513248070800281,
      "grad_norm": 1.0531797435876542,
      "learning_rate": 1.0525111069586002e-06,
      "loss": 0.6202,
      "step": 17629
    },
    {
      "epoch": 0.9513787707085425,
      "grad_norm": 0.889345664946669,
      "learning_rate": 1.052394965237044e-06,
      "loss": 0.3591,
      "step": 17630
    },
    {
      "epoch": 0.9514327343370568,
      "grad_norm": 1.1066690737170983,
      "learning_rate": 1.0522789513441508e-06,
      "loss": 0.3865,
      "step": 17631
    },
    {
      "epoch": 0.9514866979655712,
      "grad_norm": 1.107699108375548,
      "learning_rate": 1.0521630652832542e-06,
      "loss": 0.554,
      "step": 17632
    },
    {
      "epoch": 0.9515406615940856,
      "grad_norm": 1.0309008815380227,
      "learning_rate": 1.0520473070576857e-06,
      "loss": 0.4469,
      "step": 17633
    },
    {
      "epoch": 0.9515946252226,
      "grad_norm": 1.1362166527863702,
      "learning_rate": 1.051931676670772e-06,
      "loss": 0.5731,
      "step": 17634
    },
    {
      "epoch": 0.9516485888511144,
      "grad_norm": 1.2212576989212656,
      "learning_rate": 1.051816174125836e-06,
      "loss": 0.5999,
      "step": 17635
    },
    {
      "epoch": 0.9517025524796288,
      "grad_norm": 1.1601667896209962,
      "learning_rate": 1.0517007994261974e-06,
      "loss": 0.5036,
      "step": 17636
    },
    {
      "epoch": 0.9517565161081432,
      "grad_norm": 0.8389435394866919,
      "learning_rate": 1.051585552575173e-06,
      "loss": 0.3647,
      "step": 17637
    },
    {
      "epoch": 0.9518104797366574,
      "grad_norm": 1.1036808014179826,
      "learning_rate": 1.0514704335760745e-06,
      "loss": 0.5584,
      "step": 17638
    },
    {
      "epoch": 0.9518644433651718,
      "grad_norm": 0.9016261404037019,
      "learning_rate": 1.0513554424322108e-06,
      "loss": 0.3668,
      "step": 17639
    },
    {
      "epoch": 0.9519184069936862,
      "grad_norm": 0.8962507866122025,
      "learning_rate": 1.0512405791468866e-06,
      "loss": 0.3721,
      "step": 17640
    },
    {
      "epoch": 0.9519723706222006,
      "grad_norm": 1.0335624806950634,
      "learning_rate": 1.051125843723403e-06,
      "loss": 0.4562,
      "step": 17641
    },
    {
      "epoch": 0.952026334250715,
      "grad_norm": 0.843583024126452,
      "learning_rate": 1.0510112361650584e-06,
      "loss": 0.3341,
      "step": 17642
    },
    {
      "epoch": 0.9520802978792294,
      "grad_norm": 1.0770652278479815,
      "learning_rate": 1.0508967564751457e-06,
      "loss": 0.3958,
      "step": 17643
    },
    {
      "epoch": 0.9521342615077438,
      "grad_norm": 0.9708368305725501,
      "learning_rate": 1.0507824046569563e-06,
      "loss": 0.3391,
      "step": 17644
    },
    {
      "epoch": 0.9521882251362581,
      "grad_norm": 0.9851122262730378,
      "learning_rate": 1.0506681807137762e-06,
      "loss": 0.4206,
      "step": 17645
    },
    {
      "epoch": 0.9522421887647725,
      "grad_norm": 0.9297507241168641,
      "learning_rate": 1.0505540846488883e-06,
      "loss": 0.4296,
      "step": 17646
    },
    {
      "epoch": 0.9522961523932869,
      "grad_norm": 1.0457631285836326,
      "learning_rate": 1.050440116465571e-06,
      "loss": 0.4153,
      "step": 17647
    },
    {
      "epoch": 0.9523501160218013,
      "grad_norm": 1.0753969675962822,
      "learning_rate": 1.0503262761671015e-06,
      "loss": 0.3985,
      "step": 17648
    },
    {
      "epoch": 0.9524040796503157,
      "grad_norm": 1.3357804711696608,
      "learning_rate": 1.0502125637567507e-06,
      "loss": 0.4159,
      "step": 17649
    },
    {
      "epoch": 0.9524580432788301,
      "grad_norm": 0.9555765359466383,
      "learning_rate": 1.0500989792377868e-06,
      "loss": 0.3315,
      "step": 17650
    },
    {
      "epoch": 0.9525120069073445,
      "grad_norm": 1.010547543904692,
      "learning_rate": 1.049985522613475e-06,
      "loss": 0.4438,
      "step": 17651
    },
    {
      "epoch": 0.9525659705358588,
      "grad_norm": 1.2242796264990534,
      "learning_rate": 1.0498721938870757e-06,
      "loss": 0.4773,
      "step": 17652
    },
    {
      "epoch": 0.9526199341643732,
      "grad_norm": 0.8741964857863869,
      "learning_rate": 1.0497589930618456e-06,
      "loss": 0.3859,
      "step": 17653
    },
    {
      "epoch": 0.9526738977928876,
      "grad_norm": 1.0402375372208164,
      "learning_rate": 1.049645920141039e-06,
      "loss": 0.4422,
      "step": 17654
    },
    {
      "epoch": 0.952727861421402,
      "grad_norm": 1.1431143254444935,
      "learning_rate": 1.0495329751279049e-06,
      "loss": 0.4153,
      "step": 17655
    },
    {
      "epoch": 0.9527818250499164,
      "grad_norm": 0.9032345684703901,
      "learning_rate": 1.0494201580256909e-06,
      "loss": 0.3631,
      "step": 17656
    },
    {
      "epoch": 0.9528357886784308,
      "grad_norm": 1.1045862970631781,
      "learning_rate": 1.0493074688376377e-06,
      "loss": 0.5038,
      "step": 17657
    },
    {
      "epoch": 0.9528897523069452,
      "grad_norm": 0.9820575398578418,
      "learning_rate": 1.049194907566985e-06,
      "loss": 0.5346,
      "step": 17658
    },
    {
      "epoch": 0.9529437159354595,
      "grad_norm": 1.3080387519678802,
      "learning_rate": 1.0490824742169681e-06,
      "loss": 0.6324,
      "step": 17659
    },
    {
      "epoch": 0.9529976795639739,
      "grad_norm": 1.165268554844112,
      "learning_rate": 1.0489701687908186e-06,
      "loss": 0.5677,
      "step": 17660
    },
    {
      "epoch": 0.9530516431924883,
      "grad_norm": 1.0662527186166062,
      "learning_rate": 1.0488579912917633e-06,
      "loss": 0.4551,
      "step": 17661
    },
    {
      "epoch": 0.9531056068210026,
      "grad_norm": 1.1187684012649728,
      "learning_rate": 1.0487459417230272e-06,
      "loss": 0.5619,
      "step": 17662
    },
    {
      "epoch": 0.953159570449517,
      "grad_norm": 1.2088529727759703,
      "learning_rate": 1.04863402008783e-06,
      "loss": 0.4408,
      "step": 17663
    },
    {
      "epoch": 0.9532135340780314,
      "grad_norm": 1.177533437145635,
      "learning_rate": 1.048522226389389e-06,
      "loss": 0.4022,
      "step": 17664
    },
    {
      "epoch": 0.9532674977065458,
      "grad_norm": 1.234571508720051,
      "learning_rate": 1.0484105606309173e-06,
      "loss": 0.5027,
      "step": 17665
    },
    {
      "epoch": 0.9533214613350601,
      "grad_norm": 1.091445740940015,
      "learning_rate": 1.0482990228156244e-06,
      "loss": 0.4347,
      "step": 17666
    },
    {
      "epoch": 0.9533754249635745,
      "grad_norm": 1.2553942688133366,
      "learning_rate": 1.0481876129467155e-06,
      "loss": 0.5557,
      "step": 17667
    },
    {
      "epoch": 0.9534293885920889,
      "grad_norm": 1.3443417619220825,
      "learning_rate": 1.0480763310273926e-06,
      "loss": 0.6185,
      "step": 17668
    },
    {
      "epoch": 0.9534833522206033,
      "grad_norm": 0.9731637448298938,
      "learning_rate": 1.0479651770608542e-06,
      "loss": 0.4857,
      "step": 17669
    },
    {
      "epoch": 0.9535373158491177,
      "grad_norm": 1.0115749730655552,
      "learning_rate": 1.0478541510502949e-06,
      "loss": 0.54,
      "step": 17670
    },
    {
      "epoch": 0.9535912794776321,
      "grad_norm": 1.0083989685724766,
      "learning_rate": 1.0477432529989063e-06,
      "loss": 0.3684,
      "step": 17671
    },
    {
      "epoch": 0.9536452431061465,
      "grad_norm": 0.990603171907023,
      "learning_rate": 1.0476324829098752e-06,
      "loss": 0.5231,
      "step": 17672
    },
    {
      "epoch": 0.9536992067346608,
      "grad_norm": 1.085726614527059,
      "learning_rate": 1.0475218407863856e-06,
      "loss": 0.43,
      "step": 17673
    },
    {
      "epoch": 0.9537531703631752,
      "grad_norm": 0.9610723731221705,
      "learning_rate": 1.0474113266316165e-06,
      "loss": 0.4765,
      "step": 17674
    },
    {
      "epoch": 0.9538071339916896,
      "grad_norm": 1.1290952107149033,
      "learning_rate": 1.0473009404487457e-06,
      "loss": 0.4295,
      "step": 17675
    },
    {
      "epoch": 0.953861097620204,
      "grad_norm": 0.9215067393297959,
      "learning_rate": 1.0471906822409442e-06,
      "loss": 0.4157,
      "step": 17676
    },
    {
      "epoch": 0.9539150612487184,
      "grad_norm": 1.0470756764388338,
      "learning_rate": 1.0470805520113819e-06,
      "loss": 0.4045,
      "step": 17677
    },
    {
      "epoch": 0.9539690248772328,
      "grad_norm": 1.0307174850827003,
      "learning_rate": 1.046970549763224e-06,
      "loss": 0.4437,
      "step": 17678
    },
    {
      "epoch": 0.9540229885057471,
      "grad_norm": 1.11061759362116,
      "learning_rate": 1.0468606754996326e-06,
      "loss": 0.4742,
      "step": 17679
    },
    {
      "epoch": 0.9540769521342615,
      "grad_norm": 1.114854173150465,
      "learning_rate": 1.046750929223764e-06,
      "loss": 0.6119,
      "step": 17680
    },
    {
      "epoch": 0.9541309157627759,
      "grad_norm": 0.9847414804928707,
      "learning_rate": 1.046641310938774e-06,
      "loss": 0.4232,
      "step": 17681
    },
    {
      "epoch": 0.9541848793912903,
      "grad_norm": 1.1109607203669796,
      "learning_rate": 1.0465318206478124e-06,
      "loss": 0.4372,
      "step": 17682
    },
    {
      "epoch": 0.9542388430198047,
      "grad_norm": 0.9582056349965231,
      "learning_rate": 1.0464224583540259e-06,
      "loss": 0.3866,
      "step": 17683
    },
    {
      "epoch": 0.9542928066483191,
      "grad_norm": 1.1201153569349884,
      "learning_rate": 1.0463132240605577e-06,
      "loss": 0.5326,
      "step": 17684
    },
    {
      "epoch": 0.9543467702768335,
      "grad_norm": 1.2324118633377332,
      "learning_rate": 1.0462041177705482e-06,
      "loss": 0.6902,
      "step": 17685
    },
    {
      "epoch": 0.9544007339053477,
      "grad_norm": 1.0798541694354742,
      "learning_rate": 1.0460951394871321e-06,
      "loss": 0.3508,
      "step": 17686
    },
    {
      "epoch": 0.9544546975338621,
      "grad_norm": 0.9359774790878951,
      "learning_rate": 1.0459862892134425e-06,
      "loss": 0.4451,
      "step": 17687
    },
    {
      "epoch": 0.9545086611623765,
      "grad_norm": 0.718418275040109,
      "learning_rate": 1.0458775669526073e-06,
      "loss": 0.2675,
      "step": 17688
    },
    {
      "epoch": 0.9545626247908909,
      "grad_norm": 0.9827792135680532,
      "learning_rate": 1.045768972707751e-06,
      "loss": 0.4183,
      "step": 17689
    },
    {
      "epoch": 0.9546165884194053,
      "grad_norm": 0.9902427001287208,
      "learning_rate": 1.0456605064819953e-06,
      "loss": 0.425,
      "step": 17690
    },
    {
      "epoch": 0.9546705520479197,
      "grad_norm": 1.0348500837253471,
      "learning_rate": 1.045552168278458e-06,
      "loss": 0.4558,
      "step": 17691
    },
    {
      "epoch": 0.9547245156764341,
      "grad_norm": 0.8575155310673214,
      "learning_rate": 1.0454439581002516e-06,
      "loss": 0.397,
      "step": 17692
    },
    {
      "epoch": 0.9547784793049484,
      "grad_norm": 1.4360496251926085,
      "learning_rate": 1.045335875950487e-06,
      "loss": 0.5174,
      "step": 17693
    },
    {
      "epoch": 0.9548324429334628,
      "grad_norm": 1.024975834218003,
      "learning_rate": 1.0452279218322704e-06,
      "loss": 0.4369,
      "step": 17694
    },
    {
      "epoch": 0.9548864065619772,
      "grad_norm": 1.092512806303529,
      "learning_rate": 1.0451200957487044e-06,
      "loss": 0.4264,
      "step": 17695
    },
    {
      "epoch": 0.9549403701904916,
      "grad_norm": 1.0611428554073434,
      "learning_rate": 1.0450123977028886e-06,
      "loss": 0.5135,
      "step": 17696
    },
    {
      "epoch": 0.954994333819006,
      "grad_norm": 0.9980648977585351,
      "learning_rate": 1.0449048276979174e-06,
      "loss": 0.3864,
      "step": 17697
    },
    {
      "epoch": 0.9550482974475204,
      "grad_norm": 0.9461468803910928,
      "learning_rate": 1.0447973857368834e-06,
      "loss": 0.3863,
      "step": 17698
    },
    {
      "epoch": 0.9551022610760348,
      "grad_norm": 1.0978084771973087,
      "learning_rate": 1.0446900718228743e-06,
      "loss": 0.483,
      "step": 17699
    },
    {
      "epoch": 0.9551562247045491,
      "grad_norm": 0.8978731375611855,
      "learning_rate": 1.0445828859589743e-06,
      "loss": 0.5017,
      "step": 17700
    },
    {
      "epoch": 0.9552101883330635,
      "grad_norm": 1.0979394112677254,
      "learning_rate": 1.0444758281482637e-06,
      "loss": 0.5262,
      "step": 17701
    },
    {
      "epoch": 0.9552641519615779,
      "grad_norm": 1.0616813898680568,
      "learning_rate": 1.0443688983938199e-06,
      "loss": 0.4212,
      "step": 17702
    },
    {
      "epoch": 0.9553181155900923,
      "grad_norm": 0.9891963158647209,
      "learning_rate": 1.044262096698716e-06,
      "loss": 0.508,
      "step": 17703
    },
    {
      "epoch": 0.9553720792186067,
      "grad_norm": 1.0020101319230459,
      "learning_rate": 1.0441554230660214e-06,
      "loss": 0.3942,
      "step": 17704
    },
    {
      "epoch": 0.9554260428471211,
      "grad_norm": 1.1869993088750936,
      "learning_rate": 1.0440488774988025e-06,
      "loss": 0.5137,
      "step": 17705
    },
    {
      "epoch": 0.9554800064756355,
      "grad_norm": 1.0442410427488011,
      "learning_rate": 1.0439424600001213e-06,
      "loss": 0.4338,
      "step": 17706
    },
    {
      "epoch": 0.9555339701041498,
      "grad_norm": 1.3410162127961305,
      "learning_rate": 1.0438361705730366e-06,
      "loss": 0.6366,
      "step": 17707
    },
    {
      "epoch": 0.9555879337326642,
      "grad_norm": 1.1380491493760103,
      "learning_rate": 1.0437300092206027e-06,
      "loss": 0.5098,
      "step": 17708
    },
    {
      "epoch": 0.9556418973611785,
      "grad_norm": 1.2099188914345962,
      "learning_rate": 1.0436239759458706e-06,
      "loss": 0.5391,
      "step": 17709
    },
    {
      "epoch": 0.955695860989693,
      "grad_norm": 1.030921659759116,
      "learning_rate": 1.0435180707518893e-06,
      "loss": 0.3893,
      "step": 17710
    },
    {
      "epoch": 0.9557498246182073,
      "grad_norm": 1.028493883766659,
      "learning_rate": 1.0434122936417008e-06,
      "loss": 0.4023,
      "step": 17711
    },
    {
      "epoch": 0.9558037882467217,
      "grad_norm": 1.139975423688594,
      "learning_rate": 1.0433066446183461e-06,
      "loss": 0.4774,
      "step": 17712
    },
    {
      "epoch": 0.9558577518752361,
      "grad_norm": 0.9959768846473165,
      "learning_rate": 1.0432011236848617e-06,
      "loss": 0.4092,
      "step": 17713
    },
    {
      "epoch": 0.9559117155037504,
      "grad_norm": 1.086792776682233,
      "learning_rate": 1.0430957308442803e-06,
      "loss": 0.413,
      "step": 17714
    },
    {
      "epoch": 0.9559656791322648,
      "grad_norm": 0.9659619943331393,
      "learning_rate": 1.0429904660996313e-06,
      "loss": 0.5156,
      "step": 17715
    },
    {
      "epoch": 0.9560196427607792,
      "grad_norm": 1.1205599090875018,
      "learning_rate": 1.0428853294539394e-06,
      "loss": 0.6102,
      "step": 17716
    },
    {
      "epoch": 0.9560736063892936,
      "grad_norm": 1.1462236769240362,
      "learning_rate": 1.0427803209102264e-06,
      "loss": 0.5216,
      "step": 17717
    },
    {
      "epoch": 0.956127570017808,
      "grad_norm": 0.9592046640094596,
      "learning_rate": 1.042675440471511e-06,
      "loss": 0.4165,
      "step": 17718
    },
    {
      "epoch": 0.9561815336463224,
      "grad_norm": 0.9212851544249544,
      "learning_rate": 1.0425706881408077e-06,
      "loss": 0.4389,
      "step": 17719
    },
    {
      "epoch": 0.9562354972748368,
      "grad_norm": 1.3534285352857323,
      "learning_rate": 1.042466063921126e-06,
      "loss": 0.5107,
      "step": 17720
    },
    {
      "epoch": 0.9562894609033511,
      "grad_norm": 0.8444942617126312,
      "learning_rate": 1.0423615678154742e-06,
      "loss": 0.4232,
      "step": 17721
    },
    {
      "epoch": 0.9563434245318655,
      "grad_norm": 0.978843321585457,
      "learning_rate": 1.042257199826855e-06,
      "loss": 0.4703,
      "step": 17722
    },
    {
      "epoch": 0.9563973881603799,
      "grad_norm": 1.0406275146846216,
      "learning_rate": 1.0421529599582675e-06,
      "loss": 0.433,
      "step": 17723
    },
    {
      "epoch": 0.9564513517888943,
      "grad_norm": 1.077547345639178,
      "learning_rate": 1.0420488482127087e-06,
      "loss": 0.4628,
      "step": 17724
    },
    {
      "epoch": 0.9565053154174087,
      "grad_norm": 1.0770319302788138,
      "learning_rate": 1.0419448645931704e-06,
      "loss": 0.4472,
      "step": 17725
    },
    {
      "epoch": 0.9565592790459231,
      "grad_norm": 1.2328458454538282,
      "learning_rate": 1.0418410091026415e-06,
      "loss": 0.5473,
      "step": 17726
    },
    {
      "epoch": 0.9566132426744375,
      "grad_norm": 0.8348238701452746,
      "learning_rate": 1.0417372817441059e-06,
      "loss": 0.3618,
      "step": 17727
    },
    {
      "epoch": 0.9566672063029518,
      "grad_norm": 1.6358276578772126,
      "learning_rate": 1.0416336825205465e-06,
      "loss": 0.4475,
      "step": 17728
    },
    {
      "epoch": 0.9567211699314662,
      "grad_norm": 0.8473469149249803,
      "learning_rate": 1.0415302114349395e-06,
      "loss": 0.4059,
      "step": 17729
    },
    {
      "epoch": 0.9567751335599806,
      "grad_norm": 1.0164194582845456,
      "learning_rate": 1.0414268684902588e-06,
      "loss": 0.4518,
      "step": 17730
    },
    {
      "epoch": 0.956829097188495,
      "grad_norm": 1.015227693201472,
      "learning_rate": 1.0413236536894755e-06,
      "loss": 0.3668,
      "step": 17731
    },
    {
      "epoch": 0.9568830608170094,
      "grad_norm": 0.8883798306481594,
      "learning_rate": 1.0412205670355555e-06,
      "loss": 0.4853,
      "step": 17732
    },
    {
      "epoch": 0.9569370244455238,
      "grad_norm": 0.929913013414162,
      "learning_rate": 1.041117608531461e-06,
      "loss": 0.3863,
      "step": 17733
    },
    {
      "epoch": 0.9569909880740382,
      "grad_norm": 0.9269784530881331,
      "learning_rate": 1.0410147781801526e-06,
      "loss": 0.3679,
      "step": 17734
    },
    {
      "epoch": 0.9570449517025524,
      "grad_norm": 1.0222152220900902,
      "learning_rate": 1.0409120759845847e-06,
      "loss": 0.5078,
      "step": 17735
    },
    {
      "epoch": 0.9570989153310668,
      "grad_norm": 0.896020226354129,
      "learning_rate": 1.0408095019477093e-06,
      "loss": 0.405,
      "step": 17736
    },
    {
      "epoch": 0.9571528789595812,
      "grad_norm": 1.1147115781505392,
      "learning_rate": 1.0407070560724743e-06,
      "loss": 0.5496,
      "step": 17737
    },
    {
      "epoch": 0.9572068425880956,
      "grad_norm": 1.249905312818438,
      "learning_rate": 1.0406047383618248e-06,
      "loss": 0.4694,
      "step": 17738
    },
    {
      "epoch": 0.95726080621661,
      "grad_norm": 1.2587793953450306,
      "learning_rate": 1.0405025488187003e-06,
      "loss": 0.4567,
      "step": 17739
    },
    {
      "epoch": 0.9573147698451244,
      "grad_norm": 0.95219212579897,
      "learning_rate": 1.040400487446039e-06,
      "loss": 0.3957,
      "step": 17740
    },
    {
      "epoch": 0.9573687334736388,
      "grad_norm": 1.0338858317263628,
      "learning_rate": 1.0402985542467738e-06,
      "loss": 0.4931,
      "step": 17741
    },
    {
      "epoch": 0.9574226971021531,
      "grad_norm": 0.975737406563452,
      "learning_rate": 1.040196749223834e-06,
      "loss": 0.5264,
      "step": 17742
    },
    {
      "epoch": 0.9574766607306675,
      "grad_norm": 0.9314578463781079,
      "learning_rate": 1.0400950723801464e-06,
      "loss": 0.3643,
      "step": 17743
    },
    {
      "epoch": 0.9575306243591819,
      "grad_norm": 1.2187869295547722,
      "learning_rate": 1.0399935237186322e-06,
      "loss": 0.7999,
      "step": 17744
    },
    {
      "epoch": 0.9575845879876963,
      "grad_norm": 0.970237359790173,
      "learning_rate": 1.039892103242211e-06,
      "loss": 0.3741,
      "step": 17745
    },
    {
      "epoch": 0.9576385516162107,
      "grad_norm": 1.0040419361998687,
      "learning_rate": 1.0397908109537977e-06,
      "loss": 0.4514,
      "step": 17746
    },
    {
      "epoch": 0.9576925152447251,
      "grad_norm": 1.072397400898371,
      "learning_rate": 1.0396896468563026e-06,
      "loss": 0.4617,
      "step": 17747
    },
    {
      "epoch": 0.9577464788732394,
      "grad_norm": 1.1114026267859385,
      "learning_rate": 1.0395886109526346e-06,
      "loss": 0.4701,
      "step": 17748
    },
    {
      "epoch": 0.9578004425017538,
      "grad_norm": 0.86983322205349,
      "learning_rate": 1.0394877032456962e-06,
      "loss": 0.3734,
      "step": 17749
    },
    {
      "epoch": 0.9578544061302682,
      "grad_norm": 0.9360601557914175,
      "learning_rate": 1.0393869237383885e-06,
      "loss": 0.4767,
      "step": 17750
    },
    {
      "epoch": 0.9579083697587826,
      "grad_norm": 1.104363051885875,
      "learning_rate": 1.0392862724336077e-06,
      "loss": 0.5442,
      "step": 17751
    },
    {
      "epoch": 0.957962333387297,
      "grad_norm": 1.2150470593995297,
      "learning_rate": 1.0391857493342466e-06,
      "loss": 0.5096,
      "step": 17752
    },
    {
      "epoch": 0.9580162970158114,
      "grad_norm": 1.2235229531774363,
      "learning_rate": 1.0390853544431943e-06,
      "loss": 0.5891,
      "step": 17753
    },
    {
      "epoch": 0.9580702606443258,
      "grad_norm": 1.0840329855731063,
      "learning_rate": 1.0389850877633363e-06,
      "loss": 0.4794,
      "step": 17754
    },
    {
      "epoch": 0.95812422427284,
      "grad_norm": 0.9712401167125208,
      "learning_rate": 1.0388849492975549e-06,
      "loss": 0.5553,
      "step": 17755
    },
    {
      "epoch": 0.9581781879013545,
      "grad_norm": 0.8535542663321446,
      "learning_rate": 1.0387849390487275e-06,
      "loss": 0.333,
      "step": 17756
    },
    {
      "epoch": 0.9582321515298688,
      "grad_norm": 1.1196008375013198,
      "learning_rate": 1.0386850570197283e-06,
      "loss": 0.3848,
      "step": 17757
    },
    {
      "epoch": 0.9582861151583832,
      "grad_norm": 1.0148381817806675,
      "learning_rate": 1.0385853032134286e-06,
      "loss": 0.5283,
      "step": 17758
    },
    {
      "epoch": 0.9583400787868976,
      "grad_norm": 1.4057983999056316,
      "learning_rate": 1.0384856776326954e-06,
      "loss": 0.3568,
      "step": 17759
    },
    {
      "epoch": 0.958394042415412,
      "grad_norm": 1.0817293548021443,
      "learning_rate": 1.0383861802803913e-06,
      "loss": 0.408,
      "step": 17760
    },
    {
      "epoch": 0.9584480060439264,
      "grad_norm": 1.0885837367144853,
      "learning_rate": 1.0382868111593773e-06,
      "loss": 0.5009,
      "step": 17761
    },
    {
      "epoch": 0.9585019696724407,
      "grad_norm": 0.9030553297122709,
      "learning_rate": 1.0381875702725084e-06,
      "loss": 0.3447,
      "step": 17762
    },
    {
      "epoch": 0.9585559333009551,
      "grad_norm": 1.0400000936401956,
      "learning_rate": 1.0380884576226366e-06,
      "loss": 0.5068,
      "step": 17763
    },
    {
      "epoch": 0.9586098969294695,
      "grad_norm": 0.8792210321934569,
      "learning_rate": 1.037989473212611e-06,
      "loss": 0.3478,
      "step": 17764
    },
    {
      "epoch": 0.9586638605579839,
      "grad_norm": 1.2571053756366484,
      "learning_rate": 1.0378906170452769e-06,
      "loss": 0.5545,
      "step": 17765
    },
    {
      "epoch": 0.9587178241864983,
      "grad_norm": 0.9786951632352334,
      "learning_rate": 1.0377918891234745e-06,
      "loss": 0.4642,
      "step": 17766
    },
    {
      "epoch": 0.9587717878150127,
      "grad_norm": 0.9267528113855232,
      "learning_rate": 1.0376932894500425e-06,
      "loss": 0.3882,
      "step": 17767
    },
    {
      "epoch": 0.9588257514435271,
      "grad_norm": 1.0571608441531606,
      "learning_rate": 1.0375948180278139e-06,
      "loss": 0.4726,
      "step": 17768
    },
    {
      "epoch": 0.9588797150720414,
      "grad_norm": 1.0253784562121409,
      "learning_rate": 1.0374964748596194e-06,
      "loss": 0.4982,
      "step": 17769
    },
    {
      "epoch": 0.9589336787005558,
      "grad_norm": 1.138370982693256,
      "learning_rate": 1.037398259948285e-06,
      "loss": 0.4369,
      "step": 17770
    },
    {
      "epoch": 0.9589876423290702,
      "grad_norm": 1.2362092049183677,
      "learning_rate": 1.0373001732966336e-06,
      "loss": 0.5716,
      "step": 17771
    },
    {
      "epoch": 0.9590416059575846,
      "grad_norm": 0.9131615057677653,
      "learning_rate": 1.037202214907485e-06,
      "loss": 0.3085,
      "step": 17772
    },
    {
      "epoch": 0.959095569586099,
      "grad_norm": 1.2377854372747885,
      "learning_rate": 1.0371043847836535e-06,
      "loss": 0.4293,
      "step": 17773
    },
    {
      "epoch": 0.9591495332146134,
      "grad_norm": 0.8502394748359379,
      "learning_rate": 1.0370066829279518e-06,
      "loss": 0.4171,
      "step": 17774
    },
    {
      "epoch": 0.9592034968431278,
      "grad_norm": 1.1054012230251269,
      "learning_rate": 1.0369091093431873e-06,
      "loss": 0.5485,
      "step": 17775
    },
    {
      "epoch": 0.9592574604716421,
      "grad_norm": 1.0173493943618388,
      "learning_rate": 1.036811664032165e-06,
      "loss": 0.4111,
      "step": 17776
    },
    {
      "epoch": 0.9593114241001565,
      "grad_norm": 0.9556756036768218,
      "learning_rate": 1.036714346997685e-06,
      "loss": 0.5103,
      "step": 17777
    },
    {
      "epoch": 0.9593653877286709,
      "grad_norm": 1.1293090632549063,
      "learning_rate": 1.0366171582425444e-06,
      "loss": 0.5426,
      "step": 17778
    },
    {
      "epoch": 0.9594193513571853,
      "grad_norm": 0.9558760919447268,
      "learning_rate": 1.0365200977695365e-06,
      "loss": 0.3854,
      "step": 17779
    },
    {
      "epoch": 0.9594733149856997,
      "grad_norm": 1.3165901407414835,
      "learning_rate": 1.0364231655814513e-06,
      "loss": 0.5533,
      "step": 17780
    },
    {
      "epoch": 0.959527278614214,
      "grad_norm": 1.0784371606843117,
      "learning_rate": 1.0363263616810742e-06,
      "loss": 0.3897,
      "step": 17781
    },
    {
      "epoch": 0.9595812422427284,
      "grad_norm": 0.9843913026337026,
      "learning_rate": 1.0362296860711879e-06,
      "loss": 0.3492,
      "step": 17782
    },
    {
      "epoch": 0.9596352058712427,
      "grad_norm": 0.961559767168133,
      "learning_rate": 1.0361331387545707e-06,
      "loss": 0.3924,
      "step": 17783
    },
    {
      "epoch": 0.9596891694997571,
      "grad_norm": 1.1283605172314912,
      "learning_rate": 1.036036719733997e-06,
      "loss": 0.3911,
      "step": 17784
    },
    {
      "epoch": 0.9597431331282715,
      "grad_norm": 1.1232262417522745,
      "learning_rate": 1.0359404290122391e-06,
      "loss": 0.6024,
      "step": 17785
    },
    {
      "epoch": 0.9597970967567859,
      "grad_norm": 0.936757241373038,
      "learning_rate": 1.0358442665920637e-06,
      "loss": 0.3737,
      "step": 17786
    },
    {
      "epoch": 0.9598510603853003,
      "grad_norm": 0.9288426980696525,
      "learning_rate": 1.0357482324762347e-06,
      "loss": 0.4834,
      "step": 17787
    },
    {
      "epoch": 0.9599050240138147,
      "grad_norm": 0.9108370126880697,
      "learning_rate": 1.0356523266675119e-06,
      "loss": 0.4213,
      "step": 17788
    },
    {
      "epoch": 0.9599589876423291,
      "grad_norm": 0.9650317172733208,
      "learning_rate": 1.0355565491686526e-06,
      "loss": 0.4192,
      "step": 17789
    },
    {
      "epoch": 0.9600129512708434,
      "grad_norm": 1.2219971105843666,
      "learning_rate": 1.0354608999824092e-06,
      "loss": 0.6119,
      "step": 17790
    },
    {
      "epoch": 0.9600669148993578,
      "grad_norm": 1.0789889265466652,
      "learning_rate": 1.0353653791115303e-06,
      "loss": 0.5469,
      "step": 17791
    },
    {
      "epoch": 0.9601208785278722,
      "grad_norm": 0.8035186573370074,
      "learning_rate": 1.0352699865587618e-06,
      "loss": 0.4639,
      "step": 17792
    },
    {
      "epoch": 0.9601748421563866,
      "grad_norm": 0.9584016640655257,
      "learning_rate": 1.0351747223268449e-06,
      "loss": 0.4382,
      "step": 17793
    },
    {
      "epoch": 0.960228805784901,
      "grad_norm": 1.0085559776137847,
      "learning_rate": 1.035079586418518e-06,
      "loss": 0.3836,
      "step": 17794
    },
    {
      "epoch": 0.9602827694134154,
      "grad_norm": 1.13162926891348,
      "learning_rate": 1.0349845788365153e-06,
      "loss": 0.439,
      "step": 17795
    },
    {
      "epoch": 0.9603367330419298,
      "grad_norm": 1.0277144240689935,
      "learning_rate": 1.0348896995835676e-06,
      "loss": 0.3953,
      "step": 17796
    },
    {
      "epoch": 0.9603906966704441,
      "grad_norm": 0.9725216861478407,
      "learning_rate": 1.0347949486624017e-06,
      "loss": 0.5384,
      "step": 17797
    },
    {
      "epoch": 0.9604446602989585,
      "grad_norm": 1.1687313052463928,
      "learning_rate": 1.03470032607574e-06,
      "loss": 0.5086,
      "step": 17798
    },
    {
      "epoch": 0.9604986239274729,
      "grad_norm": 0.9307316422153171,
      "learning_rate": 1.0346058318263034e-06,
      "loss": 0.4179,
      "step": 17799
    },
    {
      "epoch": 0.9605525875559873,
      "grad_norm": 0.8951515470042894,
      "learning_rate": 1.0345114659168068e-06,
      "loss": 0.3346,
      "step": 17800
    },
    {
      "epoch": 0.9606065511845017,
      "grad_norm": 0.8456784465567858,
      "learning_rate": 1.034417228349963e-06,
      "loss": 0.4632,
      "step": 17801
    },
    {
      "epoch": 0.9606605148130161,
      "grad_norm": 1.0938200510937397,
      "learning_rate": 1.0343231191284798e-06,
      "loss": 0.4015,
      "step": 17802
    },
    {
      "epoch": 0.9607144784415305,
      "grad_norm": 1.1273739224444321,
      "learning_rate": 1.0342291382550626e-06,
      "loss": 0.5294,
      "step": 17803
    },
    {
      "epoch": 0.9607684420700447,
      "grad_norm": 0.9337108271471135,
      "learning_rate": 1.0341352857324128e-06,
      "loss": 0.3409,
      "step": 17804
    },
    {
      "epoch": 0.9608224056985591,
      "grad_norm": 1.1945625392753083,
      "learning_rate": 1.0340415615632266e-06,
      "loss": 0.4398,
      "step": 17805
    },
    {
      "epoch": 0.9608763693270735,
      "grad_norm": 1.0412908117891129,
      "learning_rate": 1.0339479657501987e-06,
      "loss": 0.4362,
      "step": 17806
    },
    {
      "epoch": 0.9609303329555879,
      "grad_norm": 1.2821620271787078,
      "learning_rate": 1.0338544982960188e-06,
      "loss": 0.4057,
      "step": 17807
    },
    {
      "epoch": 0.9609842965841023,
      "grad_norm": 1.1469856592848473,
      "learning_rate": 1.0337611592033732e-06,
      "loss": 0.5388,
      "step": 17808
    },
    {
      "epoch": 0.9610382602126167,
      "grad_norm": 0.9771851589078671,
      "learning_rate": 1.033667948474945e-06,
      "loss": 0.4011,
      "step": 17809
    },
    {
      "epoch": 0.9610922238411311,
      "grad_norm": 1.1093452589481345,
      "learning_rate": 1.0335748661134129e-06,
      "loss": 0.4837,
      "step": 17810
    },
    {
      "epoch": 0.9611461874696454,
      "grad_norm": 1.1499146791759003,
      "learning_rate": 1.0334819121214524e-06,
      "loss": 0.4578,
      "step": 17811
    },
    {
      "epoch": 0.9612001510981598,
      "grad_norm": 0.9174664721278442,
      "learning_rate": 1.033389086501734e-06,
      "loss": 0.3225,
      "step": 17812
    },
    {
      "epoch": 0.9612541147266742,
      "grad_norm": 1.148450469967518,
      "learning_rate": 1.033296389256927e-06,
      "loss": 0.5432,
      "step": 17813
    },
    {
      "epoch": 0.9613080783551886,
      "grad_norm": 1.1224472497300408,
      "learning_rate": 1.0332038203896949e-06,
      "loss": 0.4559,
      "step": 17814
    },
    {
      "epoch": 0.961362041983703,
      "grad_norm": 1.032105816573149,
      "learning_rate": 1.0331113799026982e-06,
      "loss": 0.4058,
      "step": 17815
    },
    {
      "epoch": 0.9614160056122174,
      "grad_norm": 1.17900329558276,
      "learning_rate": 1.033019067798594e-06,
      "loss": 0.5411,
      "step": 17816
    },
    {
      "epoch": 0.9614699692407317,
      "grad_norm": 1.462501033840754,
      "learning_rate": 1.0329268840800361e-06,
      "loss": 0.5191,
      "step": 17817
    },
    {
      "epoch": 0.9615239328692461,
      "grad_norm": 1.0432674034473248,
      "learning_rate": 1.032834828749672e-06,
      "loss": 0.6402,
      "step": 17818
    },
    {
      "epoch": 0.9615778964977605,
      "grad_norm": 1.0687312478938558,
      "learning_rate": 1.0327429018101495e-06,
      "loss": 0.4453,
      "step": 17819
    },
    {
      "epoch": 0.9616318601262749,
      "grad_norm": 0.8943684571218878,
      "learning_rate": 1.0326511032641094e-06,
      "loss": 0.4535,
      "step": 17820
    },
    {
      "epoch": 0.9616858237547893,
      "grad_norm": 0.8181611872106016,
      "learning_rate": 1.0325594331141902e-06,
      "loss": 0.3733,
      "step": 17821
    },
    {
      "epoch": 0.9617397873833037,
      "grad_norm": 1.0347529754775624,
      "learning_rate": 1.0324678913630275e-06,
      "loss": 0.2954,
      "step": 17822
    },
    {
      "epoch": 0.9617937510118181,
      "grad_norm": 0.8041383947890971,
      "learning_rate": 1.0323764780132517e-06,
      "loss": 0.3668,
      "step": 17823
    },
    {
      "epoch": 0.9618477146403324,
      "grad_norm": 1.2354958407611445,
      "learning_rate": 1.0322851930674903e-06,
      "loss": 0.6399,
      "step": 17824
    },
    {
      "epoch": 0.9619016782688468,
      "grad_norm": 0.9428310295754792,
      "learning_rate": 1.0321940365283662e-06,
      "loss": 0.4399,
      "step": 17825
    },
    {
      "epoch": 0.9619556418973612,
      "grad_norm": 0.9852785550982763,
      "learning_rate": 1.0321030083985004e-06,
      "loss": 0.4506,
      "step": 17826
    },
    {
      "epoch": 0.9620096055258756,
      "grad_norm": 1.018632599472547,
      "learning_rate": 1.0320121086805083e-06,
      "loss": 0.4259,
      "step": 17827
    },
    {
      "epoch": 0.96206356915439,
      "grad_norm": 0.8843087021489069,
      "learning_rate": 1.0319213373770025e-06,
      "loss": 0.3149,
      "step": 17828
    },
    {
      "epoch": 0.9621175327829043,
      "grad_norm": 1.0436087781426167,
      "learning_rate": 1.0318306944905923e-06,
      "loss": 0.5354,
      "step": 17829
    },
    {
      "epoch": 0.9621714964114187,
      "grad_norm": 1.2437877744021688,
      "learning_rate": 1.031740180023883e-06,
      "loss": 0.5098,
      "step": 17830
    },
    {
      "epoch": 0.962225460039933,
      "grad_norm": 1.1825161284636074,
      "learning_rate": 1.031649793979476e-06,
      "loss": 0.4955,
      "step": 17831
    },
    {
      "epoch": 0.9622794236684474,
      "grad_norm": 0.9987501100771807,
      "learning_rate": 1.0315595363599682e-06,
      "loss": 0.4037,
      "step": 17832
    },
    {
      "epoch": 0.9623333872969618,
      "grad_norm": 1.1868750833476218,
      "learning_rate": 1.0314694071679546e-06,
      "loss": 0.525,
      "step": 17833
    },
    {
      "epoch": 0.9623873509254762,
      "grad_norm": 0.9129871513173932,
      "learning_rate": 1.0313794064060253e-06,
      "loss": 0.3429,
      "step": 17834
    },
    {
      "epoch": 0.9624413145539906,
      "grad_norm": 0.8597746158332128,
      "learning_rate": 1.0312895340767674e-06,
      "loss": 0.398,
      "step": 17835
    },
    {
      "epoch": 0.962495278182505,
      "grad_norm": 1.190475073824763,
      "learning_rate": 1.031199790182763e-06,
      "loss": 0.4711,
      "step": 17836
    },
    {
      "epoch": 0.9625492418110194,
      "grad_norm": 0.9272317329151514,
      "learning_rate": 1.0311101747265927e-06,
      "loss": 0.3592,
      "step": 17837
    },
    {
      "epoch": 0.9626032054395337,
      "grad_norm": 0.8753608076286427,
      "learning_rate": 1.0310206877108308e-06,
      "loss": 0.3308,
      "step": 17838
    },
    {
      "epoch": 0.9626571690680481,
      "grad_norm": 0.9922325850964324,
      "learning_rate": 1.0309313291380507e-06,
      "loss": 0.4727,
      "step": 17839
    },
    {
      "epoch": 0.9627111326965625,
      "grad_norm": 1.1794026089140381,
      "learning_rate": 1.0308420990108194e-06,
      "loss": 0.6268,
      "step": 17840
    },
    {
      "epoch": 0.9627650963250769,
      "grad_norm": 1.0426036482956351,
      "learning_rate": 1.0307529973317021e-06,
      "loss": 0.3908,
      "step": 17841
    },
    {
      "epoch": 0.9628190599535913,
      "grad_norm": 0.9439085166138109,
      "learning_rate": 1.0306640241032591e-06,
      "loss": 0.4145,
      "step": 17842
    },
    {
      "epoch": 0.9628730235821057,
      "grad_norm": 1.002191734151182,
      "learning_rate": 1.030575179328048e-06,
      "loss": 0.5489,
      "step": 17843
    },
    {
      "epoch": 0.9629269872106201,
      "grad_norm": 0.8320335958375719,
      "learning_rate": 1.0304864630086228e-06,
      "loss": 0.2349,
      "step": 17844
    },
    {
      "epoch": 0.9629809508391344,
      "grad_norm": 0.956815211118935,
      "learning_rate": 1.0303978751475323e-06,
      "loss": 0.4539,
      "step": 17845
    },
    {
      "epoch": 0.9630349144676488,
      "grad_norm": 1.0717500178554453,
      "learning_rate": 1.030309415747323e-06,
      "loss": 0.3612,
      "step": 17846
    },
    {
      "epoch": 0.9630888780961632,
      "grad_norm": 0.8863490072070994,
      "learning_rate": 1.0302210848105376e-06,
      "loss": 0.2537,
      "step": 17847
    },
    {
      "epoch": 0.9631428417246776,
      "grad_norm": 1.0623019564729872,
      "learning_rate": 1.0301328823397146e-06,
      "loss": 0.7749,
      "step": 17848
    },
    {
      "epoch": 0.963196805353192,
      "grad_norm": 0.8674116118255991,
      "learning_rate": 1.0300448083373893e-06,
      "loss": 0.3717,
      "step": 17849
    },
    {
      "epoch": 0.9632507689817064,
      "grad_norm": 0.7822389675961525,
      "learning_rate": 1.0299568628060923e-06,
      "loss": 0.3227,
      "step": 17850
    },
    {
      "epoch": 0.9633047326102208,
      "grad_norm": 1.1865948109257531,
      "learning_rate": 1.0298690457483518e-06,
      "loss": 0.4439,
      "step": 17851
    },
    {
      "epoch": 0.963358696238735,
      "grad_norm": 0.9856340931394383,
      "learning_rate": 1.029781357166692e-06,
      "loss": 0.4523,
      "step": 17852
    },
    {
      "epoch": 0.9634126598672494,
      "grad_norm": 0.8895006942207564,
      "learning_rate": 1.0296937970636328e-06,
      "loss": 0.2738,
      "step": 17853
    },
    {
      "epoch": 0.9634666234957638,
      "grad_norm": 1.0641206274440027,
      "learning_rate": 1.0296063654416904e-06,
      "loss": 0.4292,
      "step": 17854
    },
    {
      "epoch": 0.9635205871242782,
      "grad_norm": 0.9297197545576139,
      "learning_rate": 1.0295190623033784e-06,
      "loss": 0.4333,
      "step": 17855
    },
    {
      "epoch": 0.9635745507527926,
      "grad_norm": 1.085500276263674,
      "learning_rate": 1.0294318876512055e-06,
      "loss": 0.5555,
      "step": 17856
    },
    {
      "epoch": 0.963628514381307,
      "grad_norm": 1.1678778031458226,
      "learning_rate": 1.0293448414876773e-06,
      "loss": 0.4371,
      "step": 17857
    },
    {
      "epoch": 0.9636824780098214,
      "grad_norm": 1.0212188194120606,
      "learning_rate": 1.0292579238152952e-06,
      "loss": 0.3865,
      "step": 17858
    },
    {
      "epoch": 0.9637364416383357,
      "grad_norm": 1.0924321840487399,
      "learning_rate": 1.0291711346365585e-06,
      "loss": 0.3798,
      "step": 17859
    },
    {
      "epoch": 0.9637904052668501,
      "grad_norm": 1.164620020855037,
      "learning_rate": 1.0290844739539606e-06,
      "loss": 0.4551,
      "step": 17860
    },
    {
      "epoch": 0.9638443688953645,
      "grad_norm": 1.0276925261270873,
      "learning_rate": 1.0289979417699924e-06,
      "loss": 0.3858,
      "step": 17861
    },
    {
      "epoch": 0.9638983325238789,
      "grad_norm": 0.9567818566884306,
      "learning_rate": 1.028911538087141e-06,
      "loss": 0.4068,
      "step": 17862
    },
    {
      "epoch": 0.9639522961523933,
      "grad_norm": 1.038829230738043,
      "learning_rate": 1.0288252629078897e-06,
      "loss": 0.3733,
      "step": 17863
    },
    {
      "epoch": 0.9640062597809077,
      "grad_norm": 1.0274048190455871,
      "learning_rate": 1.0287391162347177e-06,
      "loss": 0.4476,
      "step": 17864
    },
    {
      "epoch": 0.9640602234094221,
      "grad_norm": 1.1132675243969998,
      "learning_rate": 1.0286530980701018e-06,
      "loss": 0.5052,
      "step": 17865
    },
    {
      "epoch": 0.9641141870379364,
      "grad_norm": 1.0015813628045236,
      "learning_rate": 1.028567208416514e-06,
      "loss": 0.3708,
      "step": 17866
    },
    {
      "epoch": 0.9641681506664508,
      "grad_norm": 1.1448434152323455,
      "learning_rate": 1.0284814472764225e-06,
      "loss": 0.4733,
      "step": 17867
    },
    {
      "epoch": 0.9642221142949652,
      "grad_norm": 0.8336423099155317,
      "learning_rate": 1.0283958146522925e-06,
      "loss": 0.3679,
      "step": 17868
    },
    {
      "epoch": 0.9642760779234796,
      "grad_norm": 0.9625165106633883,
      "learning_rate": 1.0283103105465847e-06,
      "loss": 0.4152,
      "step": 17869
    },
    {
      "epoch": 0.964330041551994,
      "grad_norm": 1.0651891545157988,
      "learning_rate": 1.028224934961757e-06,
      "loss": 0.5732,
      "step": 17870
    },
    {
      "epoch": 0.9643840051805084,
      "grad_norm": 0.9171351378608484,
      "learning_rate": 1.028139687900263e-06,
      "loss": 0.4724,
      "step": 17871
    },
    {
      "epoch": 0.9644379688090228,
      "grad_norm": 1.0078628063634751,
      "learning_rate": 1.0280545693645528e-06,
      "loss": 0.5491,
      "step": 17872
    },
    {
      "epoch": 0.9644919324375371,
      "grad_norm": 0.8899626469964094,
      "learning_rate": 1.0279695793570731e-06,
      "loss": 0.297,
      "step": 17873
    },
    {
      "epoch": 0.9645458960660515,
      "grad_norm": 1.2555265010507757,
      "learning_rate": 1.0278847178802664e-06,
      "loss": 0.4736,
      "step": 17874
    },
    {
      "epoch": 0.9645998596945659,
      "grad_norm": 1.0857085744545776,
      "learning_rate": 1.0277999849365714e-06,
      "loss": 0.3695,
      "step": 17875
    },
    {
      "epoch": 0.9646538233230803,
      "grad_norm": 0.929586095032255,
      "learning_rate": 1.0277153805284236e-06,
      "loss": 0.3793,
      "step": 17876
    },
    {
      "epoch": 0.9647077869515946,
      "grad_norm": 0.8241015551057623,
      "learning_rate": 1.0276309046582545e-06,
      "loss": 0.375,
      "step": 17877
    },
    {
      "epoch": 0.964761750580109,
      "grad_norm": 0.7523499404912642,
      "learning_rate": 1.0275465573284928e-06,
      "loss": 0.2889,
      "step": 17878
    },
    {
      "epoch": 0.9648157142086234,
      "grad_norm": 0.9765269084361601,
      "learning_rate": 1.0274623385415616e-06,
      "loss": 0.4251,
      "step": 17879
    },
    {
      "epoch": 0.9648696778371377,
      "grad_norm": 1.0172661525891928,
      "learning_rate": 1.0273782482998827e-06,
      "loss": 0.4331,
      "step": 17880
    },
    {
      "epoch": 0.9649236414656521,
      "grad_norm": 1.0108747618632523,
      "learning_rate": 1.0272942866058714e-06,
      "loss": 0.4178,
      "step": 17881
    },
    {
      "epoch": 0.9649776050941665,
      "grad_norm": 0.9897210263063176,
      "learning_rate": 1.0272104534619421e-06,
      "loss": 0.3518,
      "step": 17882
    },
    {
      "epoch": 0.9650315687226809,
      "grad_norm": 1.2570710169809765,
      "learning_rate": 1.0271267488705033e-06,
      "loss": 0.6787,
      "step": 17883
    },
    {
      "epoch": 0.9650855323511953,
      "grad_norm": 0.996014547114282,
      "learning_rate": 1.0270431728339615e-06,
      "loss": 0.4036,
      "step": 17884
    },
    {
      "epoch": 0.9651394959797097,
      "grad_norm": 1.0025569354056054,
      "learning_rate": 1.026959725354719e-06,
      "loss": 0.3969,
      "step": 17885
    },
    {
      "epoch": 0.965193459608224,
      "grad_norm": 1.0031363834946845,
      "learning_rate": 1.0268764064351732e-06,
      "loss": 0.4562,
      "step": 17886
    },
    {
      "epoch": 0.9652474232367384,
      "grad_norm": 0.7549133103591573,
      "learning_rate": 1.0267932160777195e-06,
      "loss": 0.343,
      "step": 17887
    },
    {
      "epoch": 0.9653013868652528,
      "grad_norm": 1.125111672093527,
      "learning_rate": 1.0267101542847482e-06,
      "loss": 0.4846,
      "step": 17888
    },
    {
      "epoch": 0.9653553504937672,
      "grad_norm": 1.295898205205464,
      "learning_rate": 1.0266272210586473e-06,
      "loss": 0.655,
      "step": 17889
    },
    {
      "epoch": 0.9654093141222816,
      "grad_norm": 0.9620376330144821,
      "learning_rate": 1.0265444164017998e-06,
      "loss": 0.4134,
      "step": 17890
    },
    {
      "epoch": 0.965463277750796,
      "grad_norm": 1.0820406277605394,
      "learning_rate": 1.026461740316586e-06,
      "loss": 0.7075,
      "step": 17891
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 1.116645817768908,
      "learning_rate": 1.026379192805382e-06,
      "loss": 0.5177,
      "step": 17892
    },
    {
      "epoch": 0.9655712050078247,
      "grad_norm": 0.9204012021783984,
      "learning_rate": 1.02629677387056e-06,
      "loss": 0.4117,
      "step": 17893
    },
    {
      "epoch": 0.9656251686363391,
      "grad_norm": 1.0818601627594364,
      "learning_rate": 1.0262144835144897e-06,
      "loss": 0.5074,
      "step": 17894
    },
    {
      "epoch": 0.9656791322648535,
      "grad_norm": 0.9690285914804339,
      "learning_rate": 1.0261323217395352e-06,
      "loss": 0.4375,
      "step": 17895
    },
    {
      "epoch": 0.9657330958933679,
      "grad_norm": 0.9430185168624731,
      "learning_rate": 1.026050288548058e-06,
      "loss": 0.4065,
      "step": 17896
    },
    {
      "epoch": 0.9657870595218823,
      "grad_norm": 1.122293553442814,
      "learning_rate": 1.0259683839424167e-06,
      "loss": 0.6528,
      "step": 17897
    },
    {
      "epoch": 0.9658410231503967,
      "grad_norm": 0.855262542982004,
      "learning_rate": 1.025886607924964e-06,
      "loss": 0.3905,
      "step": 17898
    },
    {
      "epoch": 0.9658949867789111,
      "grad_norm": 0.8098971092793059,
      "learning_rate": 1.0258049604980515e-06,
      "loss": 0.3398,
      "step": 17899
    },
    {
      "epoch": 0.9659489504074253,
      "grad_norm": 0.9911239813641783,
      "learning_rate": 1.0257234416640255e-06,
      "loss": 0.3477,
      "step": 17900
    },
    {
      "epoch": 0.9660029140359397,
      "grad_norm": 1.2693430116626974,
      "learning_rate": 1.0256420514252285e-06,
      "loss": 0.4855,
      "step": 17901
    },
    {
      "epoch": 0.9660568776644541,
      "grad_norm": 0.9471655070882822,
      "learning_rate": 1.0255607897839994e-06,
      "loss": 0.3178,
      "step": 17902
    },
    {
      "epoch": 0.9661108412929685,
      "grad_norm": 1.0471143560756995,
      "learning_rate": 1.0254796567426745e-06,
      "loss": 0.4504,
      "step": 17903
    },
    {
      "epoch": 0.9661648049214829,
      "grad_norm": 1.2698459158725,
      "learning_rate": 1.0253986523035857e-06,
      "loss": 0.4673,
      "step": 17904
    },
    {
      "epoch": 0.9662187685499973,
      "grad_norm": 0.9068773711643514,
      "learning_rate": 1.0253177764690613e-06,
      "loss": 0.3652,
      "step": 17905
    },
    {
      "epoch": 0.9662727321785117,
      "grad_norm": 1.0738579287715162,
      "learning_rate": 1.0252370292414244e-06,
      "loss": 0.4006,
      "step": 17906
    },
    {
      "epoch": 0.966326695807026,
      "grad_norm": 0.9741741110029672,
      "learning_rate": 1.0251564106229974e-06,
      "loss": 0.4628,
      "step": 17907
    },
    {
      "epoch": 0.9663806594355404,
      "grad_norm": 1.2059434844354031,
      "learning_rate": 1.0250759206160964e-06,
      "loss": 0.6102,
      "step": 17908
    },
    {
      "epoch": 0.9664346230640548,
      "grad_norm": 1.082691631896521,
      "learning_rate": 1.0249955592230354e-06,
      "loss": 0.462,
      "step": 17909
    },
    {
      "epoch": 0.9664885866925692,
      "grad_norm": 1.1360780866970472,
      "learning_rate": 1.0249153264461232e-06,
      "loss": 0.5068,
      "step": 17910
    },
    {
      "epoch": 0.9665425503210836,
      "grad_norm": 0.9105956607980784,
      "learning_rate": 1.0248352222876661e-06,
      "loss": 0.3663,
      "step": 17911
    },
    {
      "epoch": 0.966596513949598,
      "grad_norm": 1.00702825364352,
      "learning_rate": 1.024755246749967e-06,
      "loss": 0.3843,
      "step": 17912
    },
    {
      "epoch": 0.9666504775781124,
      "grad_norm": 1.0104475076986612,
      "learning_rate": 1.0246753998353237e-06,
      "loss": 0.3216,
      "step": 17913
    },
    {
      "epoch": 0.9667044412066267,
      "grad_norm": 1.071184757683129,
      "learning_rate": 1.024595681546032e-06,
      "loss": 0.4045,
      "step": 17914
    },
    {
      "epoch": 0.9667584048351411,
      "grad_norm": 0.9994208332500781,
      "learning_rate": 1.0245160918843817e-06,
      "loss": 0.5676,
      "step": 17915
    },
    {
      "epoch": 0.9668123684636555,
      "grad_norm": 0.916376068978545,
      "learning_rate": 1.0244366308526619e-06,
      "loss": 0.4003,
      "step": 17916
    },
    {
      "epoch": 0.9668663320921699,
      "grad_norm": 1.0373828504978186,
      "learning_rate": 1.0243572984531548e-06,
      "loss": 0.4601,
      "step": 17917
    },
    {
      "epoch": 0.9669202957206843,
      "grad_norm": 1.130847378079657,
      "learning_rate": 1.0242780946881417e-06,
      "loss": 0.6079,
      "step": 17918
    },
    {
      "epoch": 0.9669742593491987,
      "grad_norm": 1.0822500759583193,
      "learning_rate": 1.0241990195598981e-06,
      "loss": 0.5366,
      "step": 17919
    },
    {
      "epoch": 0.9670282229777131,
      "grad_norm": 0.9734413231881124,
      "learning_rate": 1.0241200730706973e-06,
      "loss": 0.3896,
      "step": 17920
    },
    {
      "epoch": 0.9670821866062274,
      "grad_norm": 0.9734939348230095,
      "learning_rate": 1.0240412552228085e-06,
      "loss": 0.433,
      "step": 17921
    },
    {
      "epoch": 0.9671361502347418,
      "grad_norm": 0.8697652291976674,
      "learning_rate": 1.0239625660184965e-06,
      "loss": 0.277,
      "step": 17922
    },
    {
      "epoch": 0.9671901138632562,
      "grad_norm": 1.0339336759384465,
      "learning_rate": 1.023884005460023e-06,
      "loss": 0.4507,
      "step": 17923
    },
    {
      "epoch": 0.9672440774917705,
      "grad_norm": 0.8494610379040256,
      "learning_rate": 1.0238055735496454e-06,
      "loss": 0.3624,
      "step": 17924
    },
    {
      "epoch": 0.967298041120285,
      "grad_norm": 0.9096978146570516,
      "learning_rate": 1.0237272702896188e-06,
      "loss": 0.4068,
      "step": 17925
    },
    {
      "epoch": 0.9673520047487993,
      "grad_norm": 0.8960002173067981,
      "learning_rate": 1.023649095682194e-06,
      "loss": 0.3817,
      "step": 17926
    },
    {
      "epoch": 0.9674059683773137,
      "grad_norm": 1.1087403570936085,
      "learning_rate": 1.023571049729617e-06,
      "loss": 0.5568,
      "step": 17927
    },
    {
      "epoch": 0.967459932005828,
      "grad_norm": 1.1425657631016721,
      "learning_rate": 1.0234931324341307e-06,
      "loss": 0.59,
      "step": 17928
    },
    {
      "epoch": 0.9675138956343424,
      "grad_norm": 1.2249175710668825,
      "learning_rate": 1.0234153437979752e-06,
      "loss": 0.5461,
      "step": 17929
    },
    {
      "epoch": 0.9675678592628568,
      "grad_norm": 1.1516156467956786,
      "learning_rate": 1.0233376838233861e-06,
      "loss": 0.4433,
      "step": 17930
    },
    {
      "epoch": 0.9676218228913712,
      "grad_norm": 1.0688404454392078,
      "learning_rate": 1.0232601525125952e-06,
      "loss": 0.4762,
      "step": 17931
    },
    {
      "epoch": 0.9676757865198856,
      "grad_norm": 0.8945699694856074,
      "learning_rate": 1.0231827498678306e-06,
      "loss": 0.3463,
      "step": 17932
    },
    {
      "epoch": 0.9677297501484,
      "grad_norm": 0.9381772251488613,
      "learning_rate": 1.0231054758913176e-06,
      "loss": 0.3875,
      "step": 17933
    },
    {
      "epoch": 0.9677837137769144,
      "grad_norm": 1.153159367677306,
      "learning_rate": 1.0230283305852767e-06,
      "loss": 0.4666,
      "step": 17934
    },
    {
      "epoch": 0.9678376774054287,
      "grad_norm": 0.9790019482812711,
      "learning_rate": 1.0229513139519254e-06,
      "loss": 0.3567,
      "step": 17935
    },
    {
      "epoch": 0.9678916410339431,
      "grad_norm": 1.097073553141336,
      "learning_rate": 1.0228744259934772e-06,
      "loss": 0.5562,
      "step": 17936
    },
    {
      "epoch": 0.9679456046624575,
      "grad_norm": 1.1877005452087197,
      "learning_rate": 1.0227976667121417e-06,
      "loss": 0.4755,
      "step": 17937
    },
    {
      "epoch": 0.9679995682909719,
      "grad_norm": 0.983369509608898,
      "learning_rate": 1.0227210361101247e-06,
      "loss": 0.5646,
      "step": 17938
    },
    {
      "epoch": 0.9680535319194863,
      "grad_norm": 1.127318474731123,
      "learning_rate": 1.0226445341896294e-06,
      "loss": 0.4433,
      "step": 17939
    },
    {
      "epoch": 0.9681074955480007,
      "grad_norm": 1.003917119121436,
      "learning_rate": 1.022568160952854e-06,
      "loss": 0.3939,
      "step": 17940
    },
    {
      "epoch": 0.9681614591765151,
      "grad_norm": 1.0721522673342416,
      "learning_rate": 1.0224919164019943e-06,
      "loss": 0.432,
      "step": 17941
    },
    {
      "epoch": 0.9682154228050294,
      "grad_norm": 1.0532234847367967,
      "learning_rate": 1.0224158005392405e-06,
      "loss": 0.4268,
      "step": 17942
    },
    {
      "epoch": 0.9682693864335438,
      "grad_norm": 1.0337799555045897,
      "learning_rate": 1.0223398133667812e-06,
      "loss": 0.4303,
      "step": 17943
    },
    {
      "epoch": 0.9683233500620582,
      "grad_norm": 1.0335658553614406,
      "learning_rate": 1.0222639548868001e-06,
      "loss": 0.3635,
      "step": 17944
    },
    {
      "epoch": 0.9683773136905726,
      "grad_norm": 1.0770778407190655,
      "learning_rate": 1.022188225101477e-06,
      "loss": 0.4892,
      "step": 17945
    },
    {
      "epoch": 0.968431277319087,
      "grad_norm": 1.1709460009929626,
      "learning_rate": 1.022112624012989e-06,
      "loss": 0.7669,
      "step": 17946
    },
    {
      "epoch": 0.9684852409476014,
      "grad_norm": 1.0091662442482074,
      "learning_rate": 1.0220371516235089e-06,
      "loss": 0.465,
      "step": 17947
    },
    {
      "epoch": 0.9685392045761158,
      "grad_norm": 1.0740569857368223,
      "learning_rate": 1.0219618079352057e-06,
      "loss": 0.5232,
      "step": 17948
    },
    {
      "epoch": 0.96859316820463,
      "grad_norm": 1.3056445853555478,
      "learning_rate": 1.021886592950245e-06,
      "loss": 0.4805,
      "step": 17949
    },
    {
      "epoch": 0.9686471318331444,
      "grad_norm": 0.7248177777846264,
      "learning_rate": 1.0218115066707878e-06,
      "loss": 0.2545,
      "step": 17950
    },
    {
      "epoch": 0.9687010954616588,
      "grad_norm": 0.9658544320808473,
      "learning_rate": 1.0217365490989932e-06,
      "loss": 0.3123,
      "step": 17951
    },
    {
      "epoch": 0.9687550590901732,
      "grad_norm": 1.0362310464996607,
      "learning_rate": 1.021661720237015e-06,
      "loss": 0.3614,
      "step": 17952
    },
    {
      "epoch": 0.9688090227186876,
      "grad_norm": 1.108763890757111,
      "learning_rate": 1.0215870200870043e-06,
      "loss": 0.4466,
      "step": 17953
    },
    {
      "epoch": 0.968862986347202,
      "grad_norm": 1.0859257408716558,
      "learning_rate": 1.0215124486511076e-06,
      "loss": 0.5687,
      "step": 17954
    },
    {
      "epoch": 0.9689169499757163,
      "grad_norm": 0.9638020329819141,
      "learning_rate": 1.0214380059314684e-06,
      "loss": 0.3274,
      "step": 17955
    },
    {
      "epoch": 0.9689709136042307,
      "grad_norm": 0.9648403456141517,
      "learning_rate": 1.0213636919302257e-06,
      "loss": 0.3536,
      "step": 17956
    },
    {
      "epoch": 0.9690248772327451,
      "grad_norm": 1.1370965429815654,
      "learning_rate": 1.0212895066495162e-06,
      "loss": 0.5114,
      "step": 17957
    },
    {
      "epoch": 0.9690788408612595,
      "grad_norm": 0.9892726633311988,
      "learning_rate": 1.021215450091472e-06,
      "loss": 0.4854,
      "step": 17958
    },
    {
      "epoch": 0.9691328044897739,
      "grad_norm": 0.9537775241528009,
      "learning_rate": 1.0211415222582202e-06,
      "loss": 0.446,
      "step": 17959
    },
    {
      "epoch": 0.9691867681182883,
      "grad_norm": 1.3388648731217332,
      "learning_rate": 1.0210677231518878e-06,
      "loss": 0.6358,
      "step": 17960
    },
    {
      "epoch": 0.9692407317468027,
      "grad_norm": 1.321525863893946,
      "learning_rate": 1.0209940527745937e-06,
      "loss": 0.4884,
      "step": 17961
    },
    {
      "epoch": 0.969294695375317,
      "grad_norm": 1.3675563393319181,
      "learning_rate": 1.0209205111284568e-06,
      "loss": 0.5611,
      "step": 17962
    },
    {
      "epoch": 0.9693486590038314,
      "grad_norm": 1.1503612726313361,
      "learning_rate": 1.02084709821559e-06,
      "loss": 0.614,
      "step": 17963
    },
    {
      "epoch": 0.9694026226323458,
      "grad_norm": 1.0008190163529656,
      "learning_rate": 1.0207738140381037e-06,
      "loss": 0.3521,
      "step": 17964
    },
    {
      "epoch": 0.9694565862608602,
      "grad_norm": 0.8840560622903401,
      "learning_rate": 1.0207006585981038e-06,
      "loss": 0.3702,
      "step": 17965
    },
    {
      "epoch": 0.9695105498893746,
      "grad_norm": 1.0086970738993442,
      "learning_rate": 1.0206276318976927e-06,
      "loss": 0.4817,
      "step": 17966
    },
    {
      "epoch": 0.969564513517889,
      "grad_norm": 0.9723861008800753,
      "learning_rate": 1.0205547339389697e-06,
      "loss": 0.4418,
      "step": 17967
    },
    {
      "epoch": 0.9696184771464034,
      "grad_norm": 0.9263613503185468,
      "learning_rate": 1.02048196472403e-06,
      "loss": 0.3829,
      "step": 17968
    },
    {
      "epoch": 0.9696724407749177,
      "grad_norm": 0.9502685549357336,
      "learning_rate": 1.0204093242549645e-06,
      "loss": 0.3532,
      "step": 17969
    },
    {
      "epoch": 0.969726404403432,
      "grad_norm": 0.9661400662419387,
      "learning_rate": 1.0203368125338615e-06,
      "loss": 0.4812,
      "step": 17970
    },
    {
      "epoch": 0.9697803680319464,
      "grad_norm": 0.8962112155940124,
      "learning_rate": 1.020264429562805e-06,
      "loss": 0.4371,
      "step": 17971
    },
    {
      "epoch": 0.9698343316604608,
      "grad_norm": 0.9795120444014361,
      "learning_rate": 1.0201921753438751e-06,
      "loss": 0.4647,
      "step": 17972
    },
    {
      "epoch": 0.9698882952889752,
      "grad_norm": 0.9320153696267194,
      "learning_rate": 1.0201200498791488e-06,
      "loss": 0.3535,
      "step": 17973
    },
    {
      "epoch": 0.9699422589174896,
      "grad_norm": 1.074744220607773,
      "learning_rate": 1.0200480531706982e-06,
      "loss": 0.4855,
      "step": 17974
    },
    {
      "epoch": 0.969996222546004,
      "grad_norm": 1.0572480481681181,
      "learning_rate": 1.0199761852205937e-06,
      "loss": 0.3882,
      "step": 17975
    },
    {
      "epoch": 0.9700501861745183,
      "grad_norm": 1.2906687263699386,
      "learning_rate": 1.0199044460309006e-06,
      "loss": 0.4311,
      "step": 17976
    },
    {
      "epoch": 0.9701041498030327,
      "grad_norm": 1.111316494394137,
      "learning_rate": 1.0198328356036805e-06,
      "loss": 0.4808,
      "step": 17977
    },
    {
      "epoch": 0.9701581134315471,
      "grad_norm": 1.0641438294794052,
      "learning_rate": 1.019761353940991e-06,
      "loss": 0.4324,
      "step": 17978
    },
    {
      "epoch": 0.9702120770600615,
      "grad_norm": 1.198288584808613,
      "learning_rate": 1.0196900010448879e-06,
      "loss": 0.6303,
      "step": 17979
    },
    {
      "epoch": 0.9702660406885759,
      "grad_norm": 1.260344827559928,
      "learning_rate": 1.0196187769174206e-06,
      "loss": 0.4536,
      "step": 17980
    },
    {
      "epoch": 0.9703200043170903,
      "grad_norm": 1.0202265780362707,
      "learning_rate": 1.019547681560637e-06,
      "loss": 0.4242,
      "step": 17981
    },
    {
      "epoch": 0.9703739679456047,
      "grad_norm": 1.0393088977398341,
      "learning_rate": 1.01947671497658e-06,
      "loss": 0.5378,
      "step": 17982
    },
    {
      "epoch": 0.970427931574119,
      "grad_norm": 0.9038681792198786,
      "learning_rate": 1.0194058771672896e-06,
      "loss": 0.3819,
      "step": 17983
    },
    {
      "epoch": 0.9704818952026334,
      "grad_norm": 0.9490614708162891,
      "learning_rate": 1.019335168134802e-06,
      "loss": 0.3325,
      "step": 17984
    },
    {
      "epoch": 0.9705358588311478,
      "grad_norm": 1.175677515510177,
      "learning_rate": 1.0192645878811484e-06,
      "loss": 0.621,
      "step": 17985
    },
    {
      "epoch": 0.9705898224596622,
      "grad_norm": 0.9742705421471248,
      "learning_rate": 1.0191941364083582e-06,
      "loss": 0.4162,
      "step": 17986
    },
    {
      "epoch": 0.9706437860881766,
      "grad_norm": 1.158591201346969,
      "learning_rate": 1.019123813718456e-06,
      "loss": 0.4924,
      "step": 17987
    },
    {
      "epoch": 0.970697749716691,
      "grad_norm": 0.9398136893635369,
      "learning_rate": 1.019053619813463e-06,
      "loss": 0.385,
      "step": 17988
    },
    {
      "epoch": 0.9707517133452054,
      "grad_norm": 0.969448654866024,
      "learning_rate": 1.0189835546953966e-06,
      "loss": 0.5117,
      "step": 17989
    },
    {
      "epoch": 0.9708056769737197,
      "grad_norm": 1.0354562893644046,
      "learning_rate": 1.0189136183662708e-06,
      "loss": 0.4802,
      "step": 17990
    },
    {
      "epoch": 0.9708596406022341,
      "grad_norm": 0.9628997418989812,
      "learning_rate": 1.0188438108280952e-06,
      "loss": 0.3718,
      "step": 17991
    },
    {
      "epoch": 0.9709136042307485,
      "grad_norm": 1.0655703729613528,
      "learning_rate": 1.0187741320828765e-06,
      "loss": 0.4613,
      "step": 17992
    },
    {
      "epoch": 0.9709675678592629,
      "grad_norm": 0.6971511371982585,
      "learning_rate": 1.018704582132617e-06,
      "loss": 0.3221,
      "step": 17993
    },
    {
      "epoch": 0.9710215314877773,
      "grad_norm": 0.9287856752858177,
      "learning_rate": 1.0186351609793161e-06,
      "loss": 0.3847,
      "step": 17994
    },
    {
      "epoch": 0.9710754951162917,
      "grad_norm": 1.097259970205671,
      "learning_rate": 1.0185658686249688e-06,
      "loss": 0.4001,
      "step": 17995
    },
    {
      "epoch": 0.971129458744806,
      "grad_norm": 1.2089053302622532,
      "learning_rate": 1.0184967050715663e-06,
      "loss": 0.5217,
      "step": 17996
    },
    {
      "epoch": 0.9711834223733203,
      "grad_norm": 0.9899671896659042,
      "learning_rate": 1.0184276703210967e-06,
      "loss": 0.3994,
      "step": 17997
    },
    {
      "epoch": 0.9712373860018347,
      "grad_norm": 1.1178972957769615,
      "learning_rate": 1.0183587643755443e-06,
      "loss": 0.5384,
      "step": 17998
    },
    {
      "epoch": 0.9712913496303491,
      "grad_norm": 1.0975141207731514,
      "learning_rate": 1.0182899872368895e-06,
      "loss": 0.4725,
      "step": 17999
    },
    {
      "epoch": 0.9713453132588635,
      "grad_norm": 0.9627333429157267,
      "learning_rate": 1.018221338907109e-06,
      "loss": 0.3884,
      "step": 18000
    },
    {
      "epoch": 0.9713453132588635,
      "eval_loss": 0.5274714231491089,
      "eval_runtime": 161.4197,
      "eval_samples_per_second": 21.305,
      "eval_steps_per_second": 0.892,
      "step": 18000
    },
    {
      "epoch": 0.9713992768873779,
      "grad_norm": 1.1328763531926356,
      "learning_rate": 1.0181528193881752e-06,
      "loss": 0.429,
      "step": 18001
    },
    {
      "epoch": 0.9714532405158923,
      "grad_norm": 1.0306978235519206,
      "learning_rate": 1.0180844286820581e-06,
      "loss": 0.6274,
      "step": 18002
    },
    {
      "epoch": 0.9715072041444067,
      "grad_norm": 1.002171745650088,
      "learning_rate": 1.0180161667907233e-06,
      "loss": 0.4979,
      "step": 18003
    },
    {
      "epoch": 0.971561167772921,
      "grad_norm": 1.0997233807343583,
      "learning_rate": 1.017948033716133e-06,
      "loss": 0.4954,
      "step": 18004
    },
    {
      "epoch": 0.9716151314014354,
      "grad_norm": 1.0212534442859866,
      "learning_rate": 1.0178800294602446e-06,
      "loss": 0.4144,
      "step": 18005
    },
    {
      "epoch": 0.9716690950299498,
      "grad_norm": 1.171328013613981,
      "learning_rate": 1.0178121540250127e-06,
      "loss": 0.4489,
      "step": 18006
    },
    {
      "epoch": 0.9717230586584642,
      "grad_norm": 1.0803249901108014,
      "learning_rate": 1.017744407412389e-06,
      "loss": 0.502,
      "step": 18007
    },
    {
      "epoch": 0.9717770222869786,
      "grad_norm": 0.944312005227451,
      "learning_rate": 1.0176767896243198e-06,
      "loss": 0.4296,
      "step": 18008
    },
    {
      "epoch": 0.971830985915493,
      "grad_norm": 1.023631639050111,
      "learning_rate": 1.0176093006627485e-06,
      "loss": 0.4585,
      "step": 18009
    },
    {
      "epoch": 0.9718849495440074,
      "grad_norm": 0.8655145733764561,
      "learning_rate": 1.017541940529615e-06,
      "loss": 0.3885,
      "step": 18010
    },
    {
      "epoch": 0.9719389131725217,
      "grad_norm": 1.0188007506358656,
      "learning_rate": 1.0174747092268556e-06,
      "loss": 0.4793,
      "step": 18011
    },
    {
      "epoch": 0.9719928768010361,
      "grad_norm": 0.8262735025143435,
      "learning_rate": 1.0174076067564024e-06,
      "loss": 0.3157,
      "step": 18012
    },
    {
      "epoch": 0.9720468404295505,
      "grad_norm": 0.9385289914734662,
      "learning_rate": 1.0173406331201837e-06,
      "loss": 0.3985,
      "step": 18013
    },
    {
      "epoch": 0.9721008040580649,
      "grad_norm": 1.0448583862785652,
      "learning_rate": 1.0172737883201244e-06,
      "loss": 0.449,
      "step": 18014
    },
    {
      "epoch": 0.9721547676865793,
      "grad_norm": 0.8104506879596775,
      "learning_rate": 1.017207072358146e-06,
      "loss": 0.3265,
      "step": 18015
    },
    {
      "epoch": 0.9722087313150937,
      "grad_norm": 1.0848780167041987,
      "learning_rate": 1.0171404852361663e-06,
      "loss": 0.4303,
      "step": 18016
    },
    {
      "epoch": 0.9722626949436081,
      "grad_norm": 0.9300776913010911,
      "learning_rate": 1.0170740269560982e-06,
      "loss": 0.5227,
      "step": 18017
    },
    {
      "epoch": 0.9723166585721223,
      "grad_norm": 0.9759178786310342,
      "learning_rate": 1.0170076975198525e-06,
      "loss": 0.3182,
      "step": 18018
    },
    {
      "epoch": 0.9723706222006367,
      "grad_norm": 0.9517121234564407,
      "learning_rate": 1.0169414969293352e-06,
      "loss": 0.4019,
      "step": 18019
    },
    {
      "epoch": 0.9724245858291511,
      "grad_norm": 0.9223500733878252,
      "learning_rate": 1.0168754251864487e-06,
      "loss": 0.4654,
      "step": 18020
    },
    {
      "epoch": 0.9724785494576655,
      "grad_norm": 1.248658007734825,
      "learning_rate": 1.016809482293093e-06,
      "loss": 0.5129,
      "step": 18021
    },
    {
      "epoch": 0.9725325130861799,
      "grad_norm": 0.762163682543234,
      "learning_rate": 1.0167436682511622e-06,
      "loss": 0.2941,
      "step": 18022
    },
    {
      "epoch": 0.9725864767146943,
      "grad_norm": 1.1364166920600156,
      "learning_rate": 1.016677983062549e-06,
      "loss": 0.4145,
      "step": 18023
    },
    {
      "epoch": 0.9726404403432086,
      "grad_norm": 0.9499605204605104,
      "learning_rate": 1.0166124267291403e-06,
      "loss": 0.4228,
      "step": 18024
    },
    {
      "epoch": 0.972694403971723,
      "grad_norm": 0.993330907392355,
      "learning_rate": 1.0165469992528203e-06,
      "loss": 0.3718,
      "step": 18025
    },
    {
      "epoch": 0.9727483676002374,
      "grad_norm": 1.10620652430404,
      "learning_rate": 1.0164817006354702e-06,
      "loss": 0.4252,
      "step": 18026
    },
    {
      "epoch": 0.9728023312287518,
      "grad_norm": 1.02636794987454,
      "learning_rate": 1.0164165308789658e-06,
      "loss": 0.4001,
      "step": 18027
    },
    {
      "epoch": 0.9728562948572662,
      "grad_norm": 1.0145050539979432,
      "learning_rate": 1.0163514899851807e-06,
      "loss": 0.4238,
      "step": 18028
    },
    {
      "epoch": 0.9729102584857806,
      "grad_norm": 1.2074874907431528,
      "learning_rate": 1.0162865779559843e-06,
      "loss": 0.4629,
      "step": 18029
    },
    {
      "epoch": 0.972964222114295,
      "grad_norm": 1.0052872753058684,
      "learning_rate": 1.0162217947932425e-06,
      "loss": 0.467,
      "step": 18030
    },
    {
      "epoch": 0.9730181857428093,
      "grad_norm": 1.0006345896175435,
      "learning_rate": 1.0161571404988163e-06,
      "loss": 0.3675,
      "step": 18031
    },
    {
      "epoch": 0.9730721493713237,
      "grad_norm": 0.8213445728258764,
      "learning_rate": 1.016092615074565e-06,
      "loss": 0.3485,
      "step": 18032
    },
    {
      "epoch": 0.9731261129998381,
      "grad_norm": 1.0402457711805533,
      "learning_rate": 1.0160282185223419e-06,
      "loss": 0.4131,
      "step": 18033
    },
    {
      "epoch": 0.9731800766283525,
      "grad_norm": 0.9428716444633622,
      "learning_rate": 1.0159639508439993e-06,
      "loss": 0.4261,
      "step": 18034
    },
    {
      "epoch": 0.9732340402568669,
      "grad_norm": 1.1118724429153695,
      "learning_rate": 1.0158998120413827e-06,
      "loss": 0.394,
      "step": 18035
    },
    {
      "epoch": 0.9732880038853813,
      "grad_norm": 1.273067003670178,
      "learning_rate": 1.0158358021163367e-06,
      "loss": 0.5248,
      "step": 18036
    },
    {
      "epoch": 0.9733419675138957,
      "grad_norm": 1.1554296508776936,
      "learning_rate": 1.0157719210707012e-06,
      "loss": 0.4668,
      "step": 18037
    },
    {
      "epoch": 0.97339593114241,
      "grad_norm": 1.0345303950012612,
      "learning_rate": 1.015708168906311e-06,
      "loss": 0.3635,
      "step": 18038
    },
    {
      "epoch": 0.9734498947709244,
      "grad_norm": 1.0598998491873557,
      "learning_rate": 1.0156445456249995e-06,
      "loss": 0.4431,
      "step": 18039
    },
    {
      "epoch": 0.9735038583994388,
      "grad_norm": 1.0576053209848826,
      "learning_rate": 1.0155810512285945e-06,
      "loss": 0.477,
      "step": 18040
    },
    {
      "epoch": 0.9735578220279532,
      "grad_norm": 0.9469843734794028,
      "learning_rate": 1.0155176857189217e-06,
      "loss": 0.4268,
      "step": 18041
    },
    {
      "epoch": 0.9736117856564676,
      "grad_norm": 0.790758614085866,
      "learning_rate": 1.015454449097801e-06,
      "loss": 0.3501,
      "step": 18042
    },
    {
      "epoch": 0.973665749284982,
      "grad_norm": 1.2214685586476977,
      "learning_rate": 1.0153913413670514e-06,
      "loss": 0.4719,
      "step": 18043
    },
    {
      "epoch": 0.9737197129134963,
      "grad_norm": 1.0245133507821669,
      "learning_rate": 1.0153283625284863e-06,
      "loss": 0.362,
      "step": 18044
    },
    {
      "epoch": 0.9737736765420106,
      "grad_norm": 1.1514262785520362,
      "learning_rate": 1.015265512583915e-06,
      "loss": 0.5175,
      "step": 18045
    },
    {
      "epoch": 0.973827640170525,
      "grad_norm": 0.845297602103749,
      "learning_rate": 1.0152027915351443e-06,
      "loss": 0.285,
      "step": 18046
    },
    {
      "epoch": 0.9738816037990394,
      "grad_norm": 0.9991348460688204,
      "learning_rate": 1.0151401993839772e-06,
      "loss": 0.4199,
      "step": 18047
    },
    {
      "epoch": 0.9739355674275538,
      "grad_norm": 1.2644439135386834,
      "learning_rate": 1.0150777361322123e-06,
      "loss": 0.417,
      "step": 18048
    },
    {
      "epoch": 0.9739895310560682,
      "grad_norm": 0.9687601733782427,
      "learning_rate": 1.015015401781645e-06,
      "loss": 0.4322,
      "step": 18049
    },
    {
      "epoch": 0.9740434946845826,
      "grad_norm": 0.9745143892758288,
      "learning_rate": 1.0149531963340665e-06,
      "loss": 0.4595,
      "step": 18050
    },
    {
      "epoch": 0.974097458313097,
      "grad_norm": 0.9088534825989094,
      "learning_rate": 1.0148911197912654e-06,
      "loss": 0.3922,
      "step": 18051
    },
    {
      "epoch": 0.9741514219416113,
      "grad_norm": 1.0698219269637987,
      "learning_rate": 1.0148291721550254e-06,
      "loss": 0.4231,
      "step": 18052
    },
    {
      "epoch": 0.9742053855701257,
      "grad_norm": 0.8973189162744587,
      "learning_rate": 1.0147673534271266e-06,
      "loss": 0.4007,
      "step": 18053
    },
    {
      "epoch": 0.9742593491986401,
      "grad_norm": 0.919528507832268,
      "learning_rate": 1.014705663609346e-06,
      "loss": 0.3579,
      "step": 18054
    },
    {
      "epoch": 0.9743133128271545,
      "grad_norm": 1.059930800017334,
      "learning_rate": 1.014644102703457e-06,
      "loss": 0.5876,
      "step": 18055
    },
    {
      "epoch": 0.9743672764556689,
      "grad_norm": 1.0878077362574954,
      "learning_rate": 1.0145826707112284e-06,
      "loss": 0.4552,
      "step": 18056
    },
    {
      "epoch": 0.9744212400841833,
      "grad_norm": 0.9780916765598362,
      "learning_rate": 1.014521367634426e-06,
      "loss": 0.3991,
      "step": 18057
    },
    {
      "epoch": 0.9744752037126977,
      "grad_norm": 1.0687809565633837,
      "learning_rate": 1.0144601934748123e-06,
      "loss": 0.4888,
      "step": 18058
    },
    {
      "epoch": 0.974529167341212,
      "grad_norm": 0.8776835530088282,
      "learning_rate": 1.0143991482341441e-06,
      "loss": 0.3376,
      "step": 18059
    },
    {
      "epoch": 0.9745831309697264,
      "grad_norm": 1.2798073666925707,
      "learning_rate": 1.0143382319141775e-06,
      "loss": 0.6058,
      "step": 18060
    },
    {
      "epoch": 0.9746370945982408,
      "grad_norm": 0.8937949449702395,
      "learning_rate": 1.0142774445166618e-06,
      "loss": 0.347,
      "step": 18061
    },
    {
      "epoch": 0.9746910582267552,
      "grad_norm": 0.8820387598596883,
      "learning_rate": 1.0142167860433455e-06,
      "loss": 0.3725,
      "step": 18062
    },
    {
      "epoch": 0.9747450218552696,
      "grad_norm": 1.0144998420830902,
      "learning_rate": 1.0141562564959707e-06,
      "loss": 0.3642,
      "step": 18063
    },
    {
      "epoch": 0.974798985483784,
      "grad_norm": 1.0271247759111297,
      "learning_rate": 1.0140958558762783e-06,
      "loss": 0.5319,
      "step": 18064
    },
    {
      "epoch": 0.9748529491122984,
      "grad_norm": 1.1095181607857743,
      "learning_rate": 1.0140355841860032e-06,
      "loss": 0.483,
      "step": 18065
    },
    {
      "epoch": 0.9749069127408126,
      "grad_norm": 1.2714455606123745,
      "learning_rate": 1.0139754414268787e-06,
      "loss": 0.5139,
      "step": 18066
    },
    {
      "epoch": 0.974960876369327,
      "grad_norm": 0.9277221124613251,
      "learning_rate": 1.0139154276006322e-06,
      "loss": 0.4348,
      "step": 18067
    },
    {
      "epoch": 0.9750148399978414,
      "grad_norm": 1.061710294978301,
      "learning_rate": 1.0138555427089897e-06,
      "loss": 0.4173,
      "step": 18068
    },
    {
      "epoch": 0.9750688036263558,
      "grad_norm": 0.9949897615254458,
      "learning_rate": 1.0137957867536714e-06,
      "loss": 0.3916,
      "step": 18069
    },
    {
      "epoch": 0.9751227672548702,
      "grad_norm": 0.8878688591420227,
      "learning_rate": 1.013736159736395e-06,
      "loss": 0.3277,
      "step": 18070
    },
    {
      "epoch": 0.9751767308833846,
      "grad_norm": 1.0522244172311517,
      "learning_rate": 1.0136766616588748e-06,
      "loss": 0.445,
      "step": 18071
    },
    {
      "epoch": 0.975230694511899,
      "grad_norm": 1.0713606697345743,
      "learning_rate": 1.0136172925228205e-06,
      "loss": 0.5239,
      "step": 18072
    },
    {
      "epoch": 0.9752846581404133,
      "grad_norm": 0.9948468703010666,
      "learning_rate": 1.0135580523299384e-06,
      "loss": 0.3708,
      "step": 18073
    },
    {
      "epoch": 0.9753386217689277,
      "grad_norm": 0.9076397304546973,
      "learning_rate": 1.013498941081931e-06,
      "loss": 0.4023,
      "step": 18074
    },
    {
      "epoch": 0.9753925853974421,
      "grad_norm": 1.076820001182323,
      "learning_rate": 1.0134399587804975e-06,
      "loss": 0.4614,
      "step": 18075
    },
    {
      "epoch": 0.9754465490259565,
      "grad_norm": 0.9155163002381717,
      "learning_rate": 1.0133811054273324e-06,
      "loss": 0.397,
      "step": 18076
    },
    {
      "epoch": 0.9755005126544709,
      "grad_norm": 1.1775540998061875,
      "learning_rate": 1.0133223810241278e-06,
      "loss": 0.5366,
      "step": 18077
    },
    {
      "epoch": 0.9755544762829853,
      "grad_norm": 0.8871661975453764,
      "learning_rate": 1.013263785572572e-06,
      "loss": 0.2946,
      "step": 18078
    },
    {
      "epoch": 0.9756084399114997,
      "grad_norm": 1.0631971254903316,
      "learning_rate": 1.0132053190743481e-06,
      "loss": 0.4143,
      "step": 18079
    },
    {
      "epoch": 0.975662403540014,
      "grad_norm": 1.1122853943000228,
      "learning_rate": 1.013146981531137e-06,
      "loss": 0.484,
      "step": 18080
    },
    {
      "epoch": 0.9757163671685284,
      "grad_norm": 1.1166792031664092,
      "learning_rate": 1.0130887729446152e-06,
      "loss": 0.5109,
      "step": 18081
    },
    {
      "epoch": 0.9757703307970428,
      "grad_norm": 0.8902107840888837,
      "learning_rate": 1.0130306933164557e-06,
      "loss": 0.3799,
      "step": 18082
    },
    {
      "epoch": 0.9758242944255572,
      "grad_norm": 1.021386503212034,
      "learning_rate": 1.012972742648328e-06,
      "loss": 0.3747,
      "step": 18083
    },
    {
      "epoch": 0.9758782580540716,
      "grad_norm": 1.1694817312060117,
      "learning_rate": 1.0129149209418972e-06,
      "loss": 0.3697,
      "step": 18084
    },
    {
      "epoch": 0.975932221682586,
      "grad_norm": 1.128239919081581,
      "learning_rate": 1.0128572281988259e-06,
      "loss": 0.4284,
      "step": 18085
    },
    {
      "epoch": 0.9759861853111004,
      "grad_norm": 0.8845508203752075,
      "learning_rate": 1.0127996644207711e-06,
      "loss": 0.49,
      "step": 18086
    },
    {
      "epoch": 0.9760401489396147,
      "grad_norm": 1.1007899491637163,
      "learning_rate": 1.0127422296093888e-06,
      "loss": 0.5224,
      "step": 18087
    },
    {
      "epoch": 0.9760941125681291,
      "grad_norm": 1.030130818003272,
      "learning_rate": 1.0126849237663284e-06,
      "loss": 0.4841,
      "step": 18088
    },
    {
      "epoch": 0.9761480761966435,
      "grad_norm": 1.1261481412751173,
      "learning_rate": 1.0126277468932372e-06,
      "loss": 0.4721,
      "step": 18089
    },
    {
      "epoch": 0.9762020398251579,
      "grad_norm": 0.9570717482173775,
      "learning_rate": 1.012570698991759e-06,
      "loss": 0.352,
      "step": 18090
    },
    {
      "epoch": 0.9762560034536722,
      "grad_norm": 1.2416516377528521,
      "learning_rate": 1.0125137800635332e-06,
      "loss": 0.5088,
      "step": 18091
    },
    {
      "epoch": 0.9763099670821866,
      "grad_norm": 1.365636903424236,
      "learning_rate": 1.0124569901101955e-06,
      "loss": 0.4724,
      "step": 18092
    },
    {
      "epoch": 0.9763639307107009,
      "grad_norm": 1.047732205441396,
      "learning_rate": 1.012400329133378e-06,
      "loss": 0.4506,
      "step": 18093
    },
    {
      "epoch": 0.9764178943392153,
      "grad_norm": 1.0665553929723155,
      "learning_rate": 1.0123437971347098e-06,
      "loss": 0.4633,
      "step": 18094
    },
    {
      "epoch": 0.9764718579677297,
      "grad_norm": 0.9598739415894162,
      "learning_rate": 1.0122873941158153e-06,
      "loss": 0.463,
      "step": 18095
    },
    {
      "epoch": 0.9765258215962441,
      "grad_norm": 1.220803164805238,
      "learning_rate": 1.0122311200783154e-06,
      "loss": 0.6544,
      "step": 18096
    },
    {
      "epoch": 0.9765797852247585,
      "grad_norm": 1.042986223605916,
      "learning_rate": 1.0121749750238279e-06,
      "loss": 0.4811,
      "step": 18097
    },
    {
      "epoch": 0.9766337488532729,
      "grad_norm": 0.9266844776555484,
      "learning_rate": 1.0121189589539662e-06,
      "loss": 0.4392,
      "step": 18098
    },
    {
      "epoch": 0.9766877124817873,
      "grad_norm": 0.9544992510563778,
      "learning_rate": 1.0120630718703397e-06,
      "loss": 0.3253,
      "step": 18099
    },
    {
      "epoch": 0.9767416761103016,
      "grad_norm": 1.3082673723914955,
      "learning_rate": 1.0120073137745557e-06,
      "loss": 0.6495,
      "step": 18100
    },
    {
      "epoch": 0.976795639738816,
      "grad_norm": 0.8977290577675325,
      "learning_rate": 1.0119516846682165e-06,
      "loss": 0.3739,
      "step": 18101
    },
    {
      "epoch": 0.9768496033673304,
      "grad_norm": 0.932588049512579,
      "learning_rate": 1.0118961845529204e-06,
      "loss": 0.4511,
      "step": 18102
    },
    {
      "epoch": 0.9769035669958448,
      "grad_norm": 1.2099227209616008,
      "learning_rate": 1.011840813430263e-06,
      "loss": 0.5139,
      "step": 18103
    },
    {
      "epoch": 0.9769575306243592,
      "grad_norm": 0.9918164606383799,
      "learning_rate": 1.0117855713018356e-06,
      "loss": 0.3853,
      "step": 18104
    },
    {
      "epoch": 0.9770114942528736,
      "grad_norm": 1.13099420030521,
      "learning_rate": 1.0117304581692261e-06,
      "loss": 0.5049,
      "step": 18105
    },
    {
      "epoch": 0.977065457881388,
      "grad_norm": 0.7893086738490819,
      "learning_rate": 1.0116754740340176e-06,
      "loss": 0.3166,
      "step": 18106
    },
    {
      "epoch": 0.9771194215099023,
      "grad_norm": 0.9897265959083399,
      "learning_rate": 1.0116206188977918e-06,
      "loss": 0.4042,
      "step": 18107
    },
    {
      "epoch": 0.9771733851384167,
      "grad_norm": 1.109337742688425,
      "learning_rate": 1.0115658927621242e-06,
      "loss": 0.5344,
      "step": 18108
    },
    {
      "epoch": 0.9772273487669311,
      "grad_norm": 1.395479482543543,
      "learning_rate": 1.0115112956285886e-06,
      "loss": 0.6934,
      "step": 18109
    },
    {
      "epoch": 0.9772813123954455,
      "grad_norm": 1.0004340051323202,
      "learning_rate": 1.011456827498753e-06,
      "loss": 0.4164,
      "step": 18110
    },
    {
      "epoch": 0.9773352760239599,
      "grad_norm": 0.9757732982036992,
      "learning_rate": 1.0114024883741835e-06,
      "loss": 0.5241,
      "step": 18111
    },
    {
      "epoch": 0.9773892396524743,
      "grad_norm": 1.077248122988553,
      "learning_rate": 1.0113482782564424e-06,
      "loss": 0.4202,
      "step": 18112
    },
    {
      "epoch": 0.9774432032809887,
      "grad_norm": 1.10720238150688,
      "learning_rate": 1.011294197147087e-06,
      "loss": 0.4188,
      "step": 18113
    },
    {
      "epoch": 0.977497166909503,
      "grad_norm": 0.8616868165464051,
      "learning_rate": 1.011240245047672e-06,
      "loss": 0.3447,
      "step": 18114
    },
    {
      "epoch": 0.9775511305380173,
      "grad_norm": 1.0005921722472457,
      "learning_rate": 1.011186421959748e-06,
      "loss": 0.5084,
      "step": 18115
    },
    {
      "epoch": 0.9776050941665317,
      "grad_norm": 0.9450088196074476,
      "learning_rate": 1.0111327278848616e-06,
      "loss": 0.4369,
      "step": 18116
    },
    {
      "epoch": 0.9776590577950461,
      "grad_norm": 1.0212645045982813,
      "learning_rate": 1.0110791628245562e-06,
      "loss": 0.3825,
      "step": 18117
    },
    {
      "epoch": 0.9777130214235605,
      "grad_norm": 1.0038556670448877,
      "learning_rate": 1.0110257267803713e-06,
      "loss": 0.4267,
      "step": 18118
    },
    {
      "epoch": 0.9777669850520749,
      "grad_norm": 0.8301868517497739,
      "learning_rate": 1.010972419753843e-06,
      "loss": 0.3331,
      "step": 18119
    },
    {
      "epoch": 0.9778209486805893,
      "grad_norm": 0.862951565202753,
      "learning_rate": 1.0109192417465032e-06,
      "loss": 0.3455,
      "step": 18120
    },
    {
      "epoch": 0.9778749123091036,
      "grad_norm": 1.1781202287271402,
      "learning_rate": 1.0108661927598801e-06,
      "loss": 0.5539,
      "step": 18121
    },
    {
      "epoch": 0.977928875937618,
      "grad_norm": 0.9337509440004241,
      "learning_rate": 1.0108132727954992e-06,
      "loss": 0.4734,
      "step": 18122
    },
    {
      "epoch": 0.9779828395661324,
      "grad_norm": 1.0152066625072462,
      "learning_rate": 1.0107604818548803e-06,
      "loss": 0.4454,
      "step": 18123
    },
    {
      "epoch": 0.9780368031946468,
      "grad_norm": 1.097816332292557,
      "learning_rate": 1.0107078199395412e-06,
      "loss": 0.4234,
      "step": 18124
    },
    {
      "epoch": 0.9780907668231612,
      "grad_norm": 1.2156524692412445,
      "learning_rate": 1.0106552870509954e-06,
      "loss": 0.4834,
      "step": 18125
    },
    {
      "epoch": 0.9781447304516756,
      "grad_norm": 0.8971273857009229,
      "learning_rate": 1.0106028831907528e-06,
      "loss": 0.4471,
      "step": 18126
    },
    {
      "epoch": 0.97819869408019,
      "grad_norm": 1.2180385526090078,
      "learning_rate": 1.0105506083603194e-06,
      "loss": 0.5007,
      "step": 18127
    },
    {
      "epoch": 0.9782526577087043,
      "grad_norm": 0.980561212891799,
      "learning_rate": 1.0104984625611983e-06,
      "loss": 0.3225,
      "step": 18128
    },
    {
      "epoch": 0.9783066213372187,
      "grad_norm": 1.1222499466707594,
      "learning_rate": 1.0104464457948874e-06,
      "loss": 0.5624,
      "step": 18129
    },
    {
      "epoch": 0.9783605849657331,
      "grad_norm": 0.827916510612417,
      "learning_rate": 1.0103945580628819e-06,
      "loss": 0.3301,
      "step": 18130
    },
    {
      "epoch": 0.9784145485942475,
      "grad_norm": 0.9870376600276756,
      "learning_rate": 1.0103427993666734e-06,
      "loss": 0.3735,
      "step": 18131
    },
    {
      "epoch": 0.9784685122227619,
      "grad_norm": 1.0546495304772863,
      "learning_rate": 1.0102911697077493e-06,
      "loss": 0.4328,
      "step": 18132
    },
    {
      "epoch": 0.9785224758512763,
      "grad_norm": 0.8936157840489036,
      "learning_rate": 1.0102396690875933e-06,
      "loss": 0.396,
      "step": 18133
    },
    {
      "epoch": 0.9785764394797907,
      "grad_norm": 0.9965196863527791,
      "learning_rate": 1.0101882975076858e-06,
      "loss": 0.5175,
      "step": 18134
    },
    {
      "epoch": 0.978630403108305,
      "grad_norm": 0.7855980035152029,
      "learning_rate": 1.0101370549695036e-06,
      "loss": 0.3472,
      "step": 18135
    },
    {
      "epoch": 0.9786843667368194,
      "grad_norm": 0.9879028949683455,
      "learning_rate": 1.010085941474519e-06,
      "loss": 0.4519,
      "step": 18136
    },
    {
      "epoch": 0.9787383303653338,
      "grad_norm": 1.1303004022196244,
      "learning_rate": 1.010034957024201e-06,
      "loss": 0.4345,
      "step": 18137
    },
    {
      "epoch": 0.9787922939938482,
      "grad_norm": 1.1326582539831522,
      "learning_rate": 1.009984101620015e-06,
      "loss": 0.4768,
      "step": 18138
    },
    {
      "epoch": 0.9788462576223625,
      "grad_norm": 1.1087613203064681,
      "learning_rate": 1.009933375263423e-06,
      "loss": 0.484,
      "step": 18139
    },
    {
      "epoch": 0.978900221250877,
      "grad_norm": 1.049135708799523,
      "learning_rate": 1.0098827779558828e-06,
      "loss": 0.4053,
      "step": 18140
    },
    {
      "epoch": 0.9789541848793913,
      "grad_norm": 1.2797077019584613,
      "learning_rate": 1.0098323096988482e-06,
      "loss": 0.5732,
      "step": 18141
    },
    {
      "epoch": 0.9790081485079056,
      "grad_norm": 1.013127326389871,
      "learning_rate": 1.00978197049377e-06,
      "loss": 0.3665,
      "step": 18142
    },
    {
      "epoch": 0.97906211213642,
      "grad_norm": 1.2609954697528132,
      "learning_rate": 1.009731760342095e-06,
      "loss": 0.6353,
      "step": 18143
    },
    {
      "epoch": 0.9791160757649344,
      "grad_norm": 0.809306903059424,
      "learning_rate": 1.0096816792452664e-06,
      "loss": 0.4416,
      "step": 18144
    },
    {
      "epoch": 0.9791700393934488,
      "grad_norm": 0.9907922030743701,
      "learning_rate": 1.0096317272047235e-06,
      "loss": 0.4811,
      "step": 18145
    },
    {
      "epoch": 0.9792240030219632,
      "grad_norm": 1.163130820144943,
      "learning_rate": 1.0095819042219021e-06,
      "loss": 0.4712,
      "step": 18146
    },
    {
      "epoch": 0.9792779666504776,
      "grad_norm": 0.8367245820436914,
      "learning_rate": 1.0095322102982335e-06,
      "loss": 0.3716,
      "step": 18147
    },
    {
      "epoch": 0.979331930278992,
      "grad_norm": 1.0872510078203723,
      "learning_rate": 1.0094826454351467e-06,
      "loss": 0.5669,
      "step": 18148
    },
    {
      "epoch": 0.9793858939075063,
      "grad_norm": 1.0113968467370829,
      "learning_rate": 1.0094332096340665e-06,
      "loss": 0.3827,
      "step": 18149
    },
    {
      "epoch": 0.9794398575360207,
      "grad_norm": 0.9757532364728224,
      "learning_rate": 1.0093839028964127e-06,
      "loss": 0.4252,
      "step": 18150
    },
    {
      "epoch": 0.9794938211645351,
      "grad_norm": 1.017550642637091,
      "learning_rate": 1.0093347252236032e-06,
      "loss": 0.5442,
      "step": 18151
    },
    {
      "epoch": 0.9795477847930495,
      "grad_norm": 1.1081291747849564,
      "learning_rate": 1.009285676617051e-06,
      "loss": 0.5742,
      "step": 18152
    },
    {
      "epoch": 0.9796017484215639,
      "grad_norm": 1.016243254478458,
      "learning_rate": 1.0092367570781662e-06,
      "loss": 0.4543,
      "step": 18153
    },
    {
      "epoch": 0.9796557120500783,
      "grad_norm": 1.042707370596398,
      "learning_rate": 1.0091879666083543e-06,
      "loss": 0.4029,
      "step": 18154
    },
    {
      "epoch": 0.9797096756785926,
      "grad_norm": 1.0885245434916806,
      "learning_rate": 1.0091393052090178e-06,
      "loss": 0.4568,
      "step": 18155
    },
    {
      "epoch": 0.979763639307107,
      "grad_norm": 1.096945357734605,
      "learning_rate": 1.009090772881556e-06,
      "loss": 0.4954,
      "step": 18156
    },
    {
      "epoch": 0.9798176029356214,
      "grad_norm": 1.4061241342917765,
      "learning_rate": 1.0090423696273622e-06,
      "loss": 0.675,
      "step": 18157
    },
    {
      "epoch": 0.9798715665641358,
      "grad_norm": 1.6067755764102338,
      "learning_rate": 1.0089940954478294e-06,
      "loss": 0.5401,
      "step": 18158
    },
    {
      "epoch": 0.9799255301926502,
      "grad_norm": 0.9744164225233323,
      "learning_rate": 1.0089459503443435e-06,
      "loss": 0.434,
      "step": 18159
    },
    {
      "epoch": 0.9799794938211646,
      "grad_norm": 1.2286964899415982,
      "learning_rate": 1.0088979343182893e-06,
      "loss": 0.5718,
      "step": 18160
    },
    {
      "epoch": 0.980033457449679,
      "grad_norm": 1.0881118896369042,
      "learning_rate": 1.008850047371046e-06,
      "loss": 0.446,
      "step": 18161
    },
    {
      "epoch": 0.9800874210781932,
      "grad_norm": 1.0466462775271386,
      "learning_rate": 1.0088022895039907e-06,
      "loss": 0.5322,
      "step": 18162
    },
    {
      "epoch": 0.9801413847067076,
      "grad_norm": 0.8850562603511756,
      "learning_rate": 1.0087546607184951e-06,
      "loss": 0.3247,
      "step": 18163
    },
    {
      "epoch": 0.980195348335222,
      "grad_norm": 1.1048349582346668,
      "learning_rate": 1.0087071610159289e-06,
      "loss": 0.5944,
      "step": 18164
    },
    {
      "epoch": 0.9802493119637364,
      "grad_norm": 1.0553682201993737,
      "learning_rate": 1.0086597903976574e-06,
      "loss": 0.4352,
      "step": 18165
    },
    {
      "epoch": 0.9803032755922508,
      "grad_norm": 1.0109869448420752,
      "learning_rate": 1.008612548865041e-06,
      "loss": 0.319,
      "step": 18166
    },
    {
      "epoch": 0.9803572392207652,
      "grad_norm": 1.2179284930988414,
      "learning_rate": 1.0085654364194384e-06,
      "loss": 0.5053,
      "step": 18167
    },
    {
      "epoch": 0.9804112028492796,
      "grad_norm": 1.124397495186284,
      "learning_rate": 1.0085184530622036e-06,
      "loss": 0.4701,
      "step": 18168
    },
    {
      "epoch": 0.9804651664777939,
      "grad_norm": 0.9691286676573289,
      "learning_rate": 1.0084715987946868e-06,
      "loss": 0.3604,
      "step": 18169
    },
    {
      "epoch": 0.9805191301063083,
      "grad_norm": 1.148808657024611,
      "learning_rate": 1.0084248736182348e-06,
      "loss": 0.4632,
      "step": 18170
    },
    {
      "epoch": 0.9805730937348227,
      "grad_norm": 1.1482145516739721,
      "learning_rate": 1.0083782775341896e-06,
      "loss": 0.5492,
      "step": 18171
    },
    {
      "epoch": 0.9806270573633371,
      "grad_norm": 1.04684841000243,
      "learning_rate": 1.0083318105438916e-06,
      "loss": 0.6575,
      "step": 18172
    },
    {
      "epoch": 0.9806810209918515,
      "grad_norm": 0.8746654764198996,
      "learning_rate": 1.0082854726486762e-06,
      "loss": 0.3835,
      "step": 18173
    },
    {
      "epoch": 0.9807349846203659,
      "grad_norm": 0.9826383417690083,
      "learning_rate": 1.0082392638498743e-06,
      "loss": 0.479,
      "step": 18174
    },
    {
      "epoch": 0.9807889482488803,
      "grad_norm": 1.1799744077112881,
      "learning_rate": 1.008193184148815e-06,
      "loss": 0.6455,
      "step": 18175
    },
    {
      "epoch": 0.9808429118773946,
      "grad_norm": 1.183458635674134,
      "learning_rate": 1.008147233546822e-06,
      "loss": 0.5509,
      "step": 18176
    },
    {
      "epoch": 0.980896875505909,
      "grad_norm": 1.1632075306345713,
      "learning_rate": 1.0081014120452159e-06,
      "loss": 0.4979,
      "step": 18177
    },
    {
      "epoch": 0.9809508391344234,
      "grad_norm": 1.1836925945347285,
      "learning_rate": 1.0080557196453143e-06,
      "loss": 0.4678,
      "step": 18178
    },
    {
      "epoch": 0.9810048027629378,
      "grad_norm": 1.0398366517841071,
      "learning_rate": 1.0080101563484299e-06,
      "loss": 0.4987,
      "step": 18179
    },
    {
      "epoch": 0.9810587663914522,
      "grad_norm": 0.9729466164429071,
      "learning_rate": 1.0079647221558725e-06,
      "loss": 0.4997,
      "step": 18180
    },
    {
      "epoch": 0.9811127300199666,
      "grad_norm": 0.9685842662128966,
      "learning_rate": 1.0079194170689482e-06,
      "loss": 0.3667,
      "step": 18181
    },
    {
      "epoch": 0.981166693648481,
      "grad_norm": 0.966241600667927,
      "learning_rate": 1.0078742410889583e-06,
      "loss": 0.4646,
      "step": 18182
    },
    {
      "epoch": 0.9812206572769953,
      "grad_norm": 1.14052648188633,
      "learning_rate": 1.0078291942172018e-06,
      "loss": 0.4961,
      "step": 18183
    },
    {
      "epoch": 0.9812746209055097,
      "grad_norm": 0.950189912568297,
      "learning_rate": 1.0077842764549738e-06,
      "loss": 0.4096,
      "step": 18184
    },
    {
      "epoch": 0.981328584534024,
      "grad_norm": 0.8953338701019746,
      "learning_rate": 1.0077394878035642e-06,
      "loss": 0.4281,
      "step": 18185
    },
    {
      "epoch": 0.9813825481625384,
      "grad_norm": 1.0997180637953392,
      "learning_rate": 1.007694828264261e-06,
      "loss": 0.5758,
      "step": 18186
    },
    {
      "epoch": 0.9814365117910528,
      "grad_norm": 1.0227271502629796,
      "learning_rate": 1.0076502978383474e-06,
      "loss": 0.418,
      "step": 18187
    },
    {
      "epoch": 0.9814904754195672,
      "grad_norm": 0.9192749042914306,
      "learning_rate": 1.007605896527103e-06,
      "loss": 0.4067,
      "step": 18188
    },
    {
      "epoch": 0.9815444390480816,
      "grad_norm": 0.7734473257320773,
      "learning_rate": 1.0075616243318053e-06,
      "loss": 0.376,
      "step": 18189
    },
    {
      "epoch": 0.9815984026765959,
      "grad_norm": 0.9018667239248379,
      "learning_rate": 1.0075174812537254e-06,
      "loss": 0.3639,
      "step": 18190
    },
    {
      "epoch": 0.9816523663051103,
      "grad_norm": 1.2452280963652373,
      "learning_rate": 1.0074734672941325e-06,
      "loss": 0.6006,
      "step": 18191
    },
    {
      "epoch": 0.9817063299336247,
      "grad_norm": 0.8761715411188457,
      "learning_rate": 1.0074295824542913e-06,
      "loss": 0.449,
      "step": 18192
    },
    {
      "epoch": 0.9817602935621391,
      "grad_norm": 1.1489912186007136,
      "learning_rate": 1.0073858267354632e-06,
      "loss": 0.4977,
      "step": 18193
    },
    {
      "epoch": 0.9818142571906535,
      "grad_norm": 1.0717418091020081,
      "learning_rate": 1.0073422001389063e-06,
      "loss": 0.4049,
      "step": 18194
    },
    {
      "epoch": 0.9818682208191679,
      "grad_norm": 1.1470621841439548,
      "learning_rate": 1.0072987026658738e-06,
      "loss": 0.5002,
      "step": 18195
    },
    {
      "epoch": 0.9819221844476823,
      "grad_norm": 0.9233279024462852,
      "learning_rate": 1.0072553343176158e-06,
      "loss": 0.4683,
      "step": 18196
    },
    {
      "epoch": 0.9819761480761966,
      "grad_norm": 1.2870965510194652,
      "learning_rate": 1.0072120950953795e-06,
      "loss": 0.6111,
      "step": 18197
    },
    {
      "epoch": 0.982030111704711,
      "grad_norm": 1.1447051199033667,
      "learning_rate": 1.0071689850004073e-06,
      "loss": 0.5625,
      "step": 18198
    },
    {
      "epoch": 0.9820840753332254,
      "grad_norm": 1.1349408350999848,
      "learning_rate": 1.0071260040339378e-06,
      "loss": 0.3781,
      "step": 18199
    },
    {
      "epoch": 0.9821380389617398,
      "grad_norm": 1.0449355200693438,
      "learning_rate": 1.0070831521972072e-06,
      "loss": 0.5821,
      "step": 18200
    },
    {
      "epoch": 0.9821920025902542,
      "grad_norm": 1.329689120131221,
      "learning_rate": 1.0070404294914463e-06,
      "loss": 0.5435,
      "step": 18201
    },
    {
      "epoch": 0.9822459662187686,
      "grad_norm": 1.0965281736248673,
      "learning_rate": 1.0069978359178832e-06,
      "loss": 0.4836,
      "step": 18202
    },
    {
      "epoch": 0.982299929847283,
      "grad_norm": 1.0176554257984893,
      "learning_rate": 1.0069553714777423e-06,
      "loss": 0.4513,
      "step": 18203
    },
    {
      "epoch": 0.9823538934757973,
      "grad_norm": 1.040582656031757,
      "learning_rate": 1.0069130361722436e-06,
      "loss": 0.5312,
      "step": 18204
    },
    {
      "epoch": 0.9824078571043117,
      "grad_norm": 1.0010686329789062,
      "learning_rate": 1.0068708300026043e-06,
      "loss": 0.4399,
      "step": 18205
    },
    {
      "epoch": 0.9824618207328261,
      "grad_norm": 0.9041465211844968,
      "learning_rate": 1.0068287529700377e-06,
      "loss": 0.425,
      "step": 18206
    },
    {
      "epoch": 0.9825157843613405,
      "grad_norm": 0.9932138029374277,
      "learning_rate": 1.006786805075752e-06,
      "loss": 0.45,
      "step": 18207
    },
    {
      "epoch": 0.9825697479898549,
      "grad_norm": 1.2245529047345556,
      "learning_rate": 1.006744986320954e-06,
      "loss": 0.6005,
      "step": 18208
    },
    {
      "epoch": 0.9826237116183693,
      "grad_norm": 0.9026483170159882,
      "learning_rate": 1.0067032967068455e-06,
      "loss": 0.3546,
      "step": 18209
    },
    {
      "epoch": 0.9826776752468837,
      "grad_norm": 1.1038519195075662,
      "learning_rate": 1.0066617362346238e-06,
      "loss": 0.4307,
      "step": 18210
    },
    {
      "epoch": 0.9827316388753979,
      "grad_norm": 0.9008630562636736,
      "learning_rate": 1.0066203049054843e-06,
      "loss": 0.4698,
      "step": 18211
    },
    {
      "epoch": 0.9827856025039123,
      "grad_norm": 1.0777629658208496,
      "learning_rate": 1.0065790027206176e-06,
      "loss": 0.4676,
      "step": 18212
    },
    {
      "epoch": 0.9828395661324267,
      "grad_norm": 1.1308371782448063,
      "learning_rate": 1.006537829681211e-06,
      "loss": 0.4926,
      "step": 18213
    },
    {
      "epoch": 0.9828935297609411,
      "grad_norm": 1.0564011279118206,
      "learning_rate": 1.0064967857884463e-06,
      "loss": 0.4305,
      "step": 18214
    },
    {
      "epoch": 0.9829474933894555,
      "grad_norm": 1.0803578712037736,
      "learning_rate": 1.0064558710435052e-06,
      "loss": 0.4842,
      "step": 18215
    },
    {
      "epoch": 0.9830014570179699,
      "grad_norm": 0.8734324500360199,
      "learning_rate": 1.0064150854475627e-06,
      "loss": 0.2905,
      "step": 18216
    },
    {
      "epoch": 0.9830554206464843,
      "grad_norm": 1.0700042510862018,
      "learning_rate": 1.0063744290017907e-06,
      "loss": 0.4753,
      "step": 18217
    },
    {
      "epoch": 0.9831093842749986,
      "grad_norm": 1.0044028723119616,
      "learning_rate": 1.0063339017073588e-06,
      "loss": 0.3678,
      "step": 18218
    },
    {
      "epoch": 0.983163347903513,
      "grad_norm": 0.8365898435909265,
      "learning_rate": 1.0062935035654305e-06,
      "loss": 0.3748,
      "step": 18219
    },
    {
      "epoch": 0.9832173115320274,
      "grad_norm": 1.1914293477419051,
      "learning_rate": 1.0062532345771678e-06,
      "loss": 0.4886,
      "step": 18220
    },
    {
      "epoch": 0.9832712751605418,
      "grad_norm": 1.0647671882819592,
      "learning_rate": 1.0062130947437272e-06,
      "loss": 0.4334,
      "step": 18221
    },
    {
      "epoch": 0.9833252387890562,
      "grad_norm": 1.034577191696908,
      "learning_rate": 1.0061730840662634e-06,
      "loss": 0.4641,
      "step": 18222
    },
    {
      "epoch": 0.9833792024175706,
      "grad_norm": 1.2306640220021285,
      "learning_rate": 1.0061332025459254e-06,
      "loss": 0.4458,
      "step": 18223
    },
    {
      "epoch": 0.9834331660460849,
      "grad_norm": 1.112177325256744,
      "learning_rate": 1.00609345018386e-06,
      "loss": 0.5474,
      "step": 18224
    },
    {
      "epoch": 0.9834871296745993,
      "grad_norm": 1.0437642462191972,
      "learning_rate": 1.0060538269812102e-06,
      "loss": 0.378,
      "step": 18225
    },
    {
      "epoch": 0.9835410933031137,
      "grad_norm": 1.2021338265143033,
      "learning_rate": 1.0060143329391136e-06,
      "loss": 0.466,
      "step": 18226
    },
    {
      "epoch": 0.9835950569316281,
      "grad_norm": 1.0398276269324418,
      "learning_rate": 1.0059749680587062e-06,
      "loss": 0.4721,
      "step": 18227
    },
    {
      "epoch": 0.9836490205601425,
      "grad_norm": 0.9692963770283574,
      "learning_rate": 1.0059357323411188e-06,
      "loss": 0.4055,
      "step": 18228
    },
    {
      "epoch": 0.9837029841886569,
      "grad_norm": 1.2000271330654384,
      "learning_rate": 1.0058966257874795e-06,
      "loss": 0.3847,
      "step": 18229
    },
    {
      "epoch": 0.9837569478171713,
      "grad_norm": 1.0512583845130368,
      "learning_rate": 1.0058576483989122e-06,
      "loss": 0.3482,
      "step": 18230
    },
    {
      "epoch": 0.9838109114456856,
      "grad_norm": 1.1243446046343173,
      "learning_rate": 1.005818800176537e-06,
      "loss": 0.4107,
      "step": 18231
    },
    {
      "epoch": 0.9838648750742,
      "grad_norm": 1.1610840379514549,
      "learning_rate": 1.0057800811214707e-06,
      "loss": 0.4495,
      "step": 18232
    },
    {
      "epoch": 0.9839188387027143,
      "grad_norm": 0.9316836874987512,
      "learning_rate": 1.005741491234826e-06,
      "loss": 0.4049,
      "step": 18233
    },
    {
      "epoch": 0.9839728023312287,
      "grad_norm": 1.1134203056649714,
      "learning_rate": 1.0057030305177118e-06,
      "loss": 0.4329,
      "step": 18234
    },
    {
      "epoch": 0.9840267659597431,
      "grad_norm": 1.1743277057931736,
      "learning_rate": 1.0056646989712334e-06,
      "loss": 0.5086,
      "step": 18235
    },
    {
      "epoch": 0.9840807295882575,
      "grad_norm": 0.9216401843672322,
      "learning_rate": 1.0056264965964932e-06,
      "loss": 0.3802,
      "step": 18236
    },
    {
      "epoch": 0.9841346932167719,
      "grad_norm": 0.8592541274796697,
      "learning_rate": 1.0055884233945886e-06,
      "loss": 0.3381,
      "step": 18237
    },
    {
      "epoch": 0.9841886568452862,
      "grad_norm": 0.8829869270783117,
      "learning_rate": 1.0055504793666139e-06,
      "loss": 0.3898,
      "step": 18238
    },
    {
      "epoch": 0.9842426204738006,
      "grad_norm": 0.8629232867351901,
      "learning_rate": 1.0055126645136596e-06,
      "loss": 0.3301,
      "step": 18239
    },
    {
      "epoch": 0.984296584102315,
      "grad_norm": 0.9050872495089389,
      "learning_rate": 1.0054749788368129e-06,
      "loss": 0.3133,
      "step": 18240
    },
    {
      "epoch": 0.9843505477308294,
      "grad_norm": 0.814666072193666,
      "learning_rate": 1.0054374223371565e-06,
      "loss": 0.3693,
      "step": 18241
    },
    {
      "epoch": 0.9844045113593438,
      "grad_norm": 0.9396071443650622,
      "learning_rate": 1.00539999501577e-06,
      "loss": 0.3867,
      "step": 18242
    },
    {
      "epoch": 0.9844584749878582,
      "grad_norm": 1.0118339830823415,
      "learning_rate": 1.0053626968737292e-06,
      "loss": 0.4123,
      "step": 18243
    },
    {
      "epoch": 0.9845124386163726,
      "grad_norm": 1.0145481664941927,
      "learning_rate": 1.005325527912106e-06,
      "loss": 0.5768,
      "step": 18244
    },
    {
      "epoch": 0.9845664022448869,
      "grad_norm": 0.8696590983676783,
      "learning_rate": 1.005288488131969e-06,
      "loss": 0.3358,
      "step": 18245
    },
    {
      "epoch": 0.9846203658734013,
      "grad_norm": 0.8980830216114041,
      "learning_rate": 1.0052515775343822e-06,
      "loss": 0.3976,
      "step": 18246
    },
    {
      "epoch": 0.9846743295019157,
      "grad_norm": 1.0478407376333694,
      "learning_rate": 1.0052147961204062e-06,
      "loss": 0.349,
      "step": 18247
    },
    {
      "epoch": 0.9847282931304301,
      "grad_norm": 1.1066506661839204,
      "learning_rate": 1.005178143891099e-06,
      "loss": 0.5298,
      "step": 18248
    },
    {
      "epoch": 0.9847822567589445,
      "grad_norm": 1.093684639579365,
      "learning_rate": 1.0051416208475135e-06,
      "loss": 0.5162,
      "step": 18249
    },
    {
      "epoch": 0.9848362203874589,
      "grad_norm": 1.2256587780017763,
      "learning_rate": 1.0051052269906997e-06,
      "loss": 0.6069,
      "step": 18250
    },
    {
      "epoch": 0.9848901840159733,
      "grad_norm": 0.9593915962190848,
      "learning_rate": 1.0050689623217035e-06,
      "loss": 0.4657,
      "step": 18251
    },
    {
      "epoch": 0.9849441476444876,
      "grad_norm": 1.0962049463225507,
      "learning_rate": 1.005032826841567e-06,
      "loss": 0.5599,
      "step": 18252
    },
    {
      "epoch": 0.984998111273002,
      "grad_norm": 1.0583406537142928,
      "learning_rate": 1.004996820551329e-06,
      "loss": 0.5445,
      "step": 18253
    },
    {
      "epoch": 0.9850520749015164,
      "grad_norm": 1.0345727799297262,
      "learning_rate": 1.004960943452024e-06,
      "loss": 0.3963,
      "step": 18254
    },
    {
      "epoch": 0.9851060385300308,
      "grad_norm": 1.308932196241908,
      "learning_rate": 1.0049251955446834e-06,
      "loss": 0.6512,
      "step": 18255
    },
    {
      "epoch": 0.9851600021585452,
      "grad_norm": 1.1085152381252514,
      "learning_rate": 1.004889576830335e-06,
      "loss": 0.5703,
      "step": 18256
    },
    {
      "epoch": 0.9852139657870596,
      "grad_norm": 1.2414541017410305,
      "learning_rate": 1.0048540873100019e-06,
      "loss": 0.6488,
      "step": 18257
    },
    {
      "epoch": 0.985267929415574,
      "grad_norm": 1.1522782318195677,
      "learning_rate": 1.004818726984704e-06,
      "loss": 0.5326,
      "step": 18258
    },
    {
      "epoch": 0.9853218930440882,
      "grad_norm": 1.164448385879329,
      "learning_rate": 1.0047834958554586e-06,
      "loss": 0.4495,
      "step": 18259
    },
    {
      "epoch": 0.9853758566726026,
      "grad_norm": 1.0209892377338907,
      "learning_rate": 1.0047483939232771e-06,
      "loss": 0.3622,
      "step": 18260
    },
    {
      "epoch": 0.985429820301117,
      "grad_norm": 1.230433249042254,
      "learning_rate": 1.0047134211891691e-06,
      "loss": 0.5905,
      "step": 18261
    },
    {
      "epoch": 0.9854837839296314,
      "grad_norm": 0.9850723713908215,
      "learning_rate": 1.0046785776541392e-06,
      "loss": 0.5285,
      "step": 18262
    },
    {
      "epoch": 0.9855377475581458,
      "grad_norm": 1.1918676048872154,
      "learning_rate": 1.0046438633191896e-06,
      "loss": 0.4449,
      "step": 18263
    },
    {
      "epoch": 0.9855917111866602,
      "grad_norm": 1.0949680205912922,
      "learning_rate": 1.0046092781853173e-06,
      "loss": 0.4855,
      "step": 18264
    },
    {
      "epoch": 0.9856456748151746,
      "grad_norm": 1.1016646318917303,
      "learning_rate": 1.004574822253517e-06,
      "loss": 0.6209,
      "step": 18265
    },
    {
      "epoch": 0.9856996384436889,
      "grad_norm": 1.0480242424525588,
      "learning_rate": 1.0045404955247787e-06,
      "loss": 0.4905,
      "step": 18266
    },
    {
      "epoch": 0.9857536020722033,
      "grad_norm": 0.9649795113407106,
      "learning_rate": 1.004506298000088e-06,
      "loss": 0.4235,
      "step": 18267
    },
    {
      "epoch": 0.9858075657007177,
      "grad_norm": 1.0961290326494726,
      "learning_rate": 1.0044722296804299e-06,
      "loss": 0.509,
      "step": 18268
    },
    {
      "epoch": 0.9858615293292321,
      "grad_norm": 0.9931244635726947,
      "learning_rate": 1.0044382905667814e-06,
      "loss": 0.4623,
      "step": 18269
    },
    {
      "epoch": 0.9859154929577465,
      "grad_norm": 0.8983103031769314,
      "learning_rate": 1.0044044806601188e-06,
      "loss": 0.3958,
      "step": 18270
    },
    {
      "epoch": 0.9859694565862609,
      "grad_norm": 0.9824877709717669,
      "learning_rate": 1.0043707999614145e-06,
      "loss": 0.3334,
      "step": 18271
    },
    {
      "epoch": 0.9860234202147753,
      "grad_norm": 1.1206280975716212,
      "learning_rate": 1.0043372484716354e-06,
      "loss": 0.5302,
      "step": 18272
    },
    {
      "epoch": 0.9860773838432896,
      "grad_norm": 0.850834284070151,
      "learning_rate": 1.0043038261917468e-06,
      "loss": 0.3346,
      "step": 18273
    },
    {
      "epoch": 0.986131347471804,
      "grad_norm": 0.9432469932889882,
      "learning_rate": 1.0042705331227082e-06,
      "loss": 0.4346,
      "step": 18274
    },
    {
      "epoch": 0.9861853111003184,
      "grad_norm": 1.1029279909154361,
      "learning_rate": 1.0042373692654773e-06,
      "loss": 0.4031,
      "step": 18275
    },
    {
      "epoch": 0.9862392747288328,
      "grad_norm": 0.9813679784605401,
      "learning_rate": 1.0042043346210074e-06,
      "loss": 0.4045,
      "step": 18276
    },
    {
      "epoch": 0.9862932383573472,
      "grad_norm": 1.1138866055146197,
      "learning_rate": 1.004171429190247e-06,
      "loss": 0.5222,
      "step": 18277
    },
    {
      "epoch": 0.9863472019858616,
      "grad_norm": 0.9178131726245456,
      "learning_rate": 1.0041386529741428e-06,
      "loss": 0.2554,
      "step": 18278
    },
    {
      "epoch": 0.986401165614376,
      "grad_norm": 1.2169431313211292,
      "learning_rate": 1.0041060059736363e-06,
      "loss": 0.5749,
      "step": 18279
    },
    {
      "epoch": 0.9864551292428902,
      "grad_norm": 0.9078935693463321,
      "learning_rate": 1.0040734881896664e-06,
      "loss": 0.3543,
      "step": 18280
    },
    {
      "epoch": 0.9865090928714046,
      "grad_norm": 1.1620783499335867,
      "learning_rate": 1.0040410996231672e-06,
      "loss": 0.4035,
      "step": 18281
    },
    {
      "epoch": 0.986563056499919,
      "grad_norm": 1.014865994007913,
      "learning_rate": 1.004008840275069e-06,
      "loss": 0.4407,
      "step": 18282
    },
    {
      "epoch": 0.9866170201284334,
      "grad_norm": 0.9784484360173267,
      "learning_rate": 1.0039767101462998e-06,
      "loss": 0.3863,
      "step": 18283
    },
    {
      "epoch": 0.9866709837569478,
      "grad_norm": 1.06214586436164,
      "learning_rate": 1.0039447092377832e-06,
      "loss": 0.4745,
      "step": 18284
    },
    {
      "epoch": 0.9867249473854622,
      "grad_norm": 1.103956061930172,
      "learning_rate": 1.0039128375504383e-06,
      "loss": 0.4326,
      "step": 18285
    },
    {
      "epoch": 0.9867789110139766,
      "grad_norm": 1.1560465020155366,
      "learning_rate": 1.0038810950851816e-06,
      "loss": 0.5157,
      "step": 18286
    },
    {
      "epoch": 0.9868328746424909,
      "grad_norm": 0.8174886292650656,
      "learning_rate": 1.003849481842925e-06,
      "loss": 0.3161,
      "step": 18287
    },
    {
      "epoch": 0.9868868382710053,
      "grad_norm": 0.9765321252278635,
      "learning_rate": 1.0038179978245777e-06,
      "loss": 0.446,
      "step": 18288
    },
    {
      "epoch": 0.9869408018995197,
      "grad_norm": 1.0605564076666492,
      "learning_rate": 1.0037866430310437e-06,
      "loss": 0.5177,
      "step": 18289
    },
    {
      "epoch": 0.9869947655280341,
      "grad_norm": 1.0395547023376426,
      "learning_rate": 1.0037554174632251e-06,
      "loss": 0.4179,
      "step": 18290
    },
    {
      "epoch": 0.9870487291565485,
      "grad_norm": 1.0823578565099994,
      "learning_rate": 1.0037243211220187e-06,
      "loss": 0.4319,
      "step": 18291
    },
    {
      "epoch": 0.9871026927850629,
      "grad_norm": 0.9009497451485586,
      "learning_rate": 1.0036933540083183e-06,
      "loss": 0.3799,
      "step": 18292
    },
    {
      "epoch": 0.9871566564135772,
      "grad_norm": 1.1257290424317652,
      "learning_rate": 1.0036625161230146e-06,
      "loss": 0.5548,
      "step": 18293
    },
    {
      "epoch": 0.9872106200420916,
      "grad_norm": 1.0618977617744,
      "learning_rate": 1.003631807466993e-06,
      "loss": 0.4447,
      "step": 18294
    },
    {
      "epoch": 0.987264583670606,
      "grad_norm": 1.0919200086610794,
      "learning_rate": 1.0036012280411366e-06,
      "loss": 0.4259,
      "step": 18295
    },
    {
      "epoch": 0.9873185472991204,
      "grad_norm": 1.0753906018115966,
      "learning_rate": 1.0035707778463244e-06,
      "loss": 0.4556,
      "step": 18296
    },
    {
      "epoch": 0.9873725109276348,
      "grad_norm": 1.267899617075028,
      "learning_rate": 1.0035404568834308e-06,
      "loss": 0.4891,
      "step": 18297
    },
    {
      "epoch": 0.9874264745561492,
      "grad_norm": 0.8983139915447558,
      "learning_rate": 1.0035102651533285e-06,
      "loss": 0.4394,
      "step": 18298
    },
    {
      "epoch": 0.9874804381846636,
      "grad_norm": 1.3557085263618613,
      "learning_rate": 1.0034802026568842e-06,
      "loss": 0.5345,
      "step": 18299
    },
    {
      "epoch": 0.9875344018131779,
      "grad_norm": 1.199900587713111,
      "learning_rate": 1.003450269394962e-06,
      "loss": 0.5055,
      "step": 18300
    },
    {
      "epoch": 0.9875883654416923,
      "grad_norm": 1.0543647407175287,
      "learning_rate": 1.003420465368423e-06,
      "loss": 0.4389,
      "step": 18301
    },
    {
      "epoch": 0.9876423290702067,
      "grad_norm": 0.9726110282002045,
      "learning_rate": 1.003390790578123e-06,
      "loss": 0.3642,
      "step": 18302
    },
    {
      "epoch": 0.9876962926987211,
      "grad_norm": 1.1788532230638067,
      "learning_rate": 1.0033612450249152e-06,
      "loss": 0.4348,
      "step": 18303
    },
    {
      "epoch": 0.9877502563272355,
      "grad_norm": 1.026343981519351,
      "learning_rate": 1.0033318287096487e-06,
      "loss": 0.4196,
      "step": 18304
    },
    {
      "epoch": 0.9878042199557499,
      "grad_norm": 0.9041360507158261,
      "learning_rate": 1.0033025416331692e-06,
      "loss": 0.4252,
      "step": 18305
    },
    {
      "epoch": 0.9878581835842642,
      "grad_norm": 0.7729073542744118,
      "learning_rate": 1.0032733837963182e-06,
      "loss": 0.3015,
      "step": 18306
    },
    {
      "epoch": 0.9879121472127785,
      "grad_norm": 0.9851343341420472,
      "learning_rate": 1.0032443551999335e-06,
      "loss": 0.4195,
      "step": 18307
    },
    {
      "epoch": 0.9879661108412929,
      "grad_norm": 0.9849982239090818,
      "learning_rate": 1.0032154558448502e-06,
      "loss": 0.4183,
      "step": 18308
    },
    {
      "epoch": 0.9880200744698073,
      "grad_norm": 1.1344501365313695,
      "learning_rate": 1.0031866857318975e-06,
      "loss": 0.5231,
      "step": 18309
    },
    {
      "epoch": 0.9880740380983217,
      "grad_norm": 0.9729465678746738,
      "learning_rate": 1.003158044861904e-06,
      "loss": 0.3778,
      "step": 18310
    },
    {
      "epoch": 0.9881280017268361,
      "grad_norm": 1.1666556995258823,
      "learning_rate": 1.0031295332356914e-06,
      "loss": 0.4016,
      "step": 18311
    },
    {
      "epoch": 0.9881819653553505,
      "grad_norm": 1.0216438191854136,
      "learning_rate": 1.00310115085408e-06,
      "loss": 0.4335,
      "step": 18312
    },
    {
      "epoch": 0.9882359289838649,
      "grad_norm": 1.2932886889274218,
      "learning_rate": 1.0030728977178854e-06,
      "loss": 0.7647,
      "step": 18313
    },
    {
      "epoch": 0.9882898926123792,
      "grad_norm": 0.9643036938347801,
      "learning_rate": 1.0030447738279195e-06,
      "loss": 0.5105,
      "step": 18314
    },
    {
      "epoch": 0.9883438562408936,
      "grad_norm": 0.8920087782880494,
      "learning_rate": 1.0030167791849907e-06,
      "loss": 0.3923,
      "step": 18315
    },
    {
      "epoch": 0.988397819869408,
      "grad_norm": 1.3145218527508202,
      "learning_rate": 1.0029889137899034e-06,
      "loss": 0.5274,
      "step": 18316
    },
    {
      "epoch": 0.9884517834979224,
      "grad_norm": 1.0744809346904274,
      "learning_rate": 1.0029611776434588e-06,
      "loss": 0.4969,
      "step": 18317
    },
    {
      "epoch": 0.9885057471264368,
      "grad_norm": 0.9080886806558113,
      "learning_rate": 1.002933570746454e-06,
      "loss": 0.3739,
      "step": 18318
    },
    {
      "epoch": 0.9885597107549512,
      "grad_norm": 1.0392965399430079,
      "learning_rate": 1.002906093099682e-06,
      "loss": 0.4986,
      "step": 18319
    },
    {
      "epoch": 0.9886136743834656,
      "grad_norm": 1.2253051490616929,
      "learning_rate": 1.0028787447039328e-06,
      "loss": 0.5424,
      "step": 18320
    },
    {
      "epoch": 0.9886676380119799,
      "grad_norm": 1.0781878166388663,
      "learning_rate": 1.0028515255599927e-06,
      "loss": 0.5953,
      "step": 18321
    },
    {
      "epoch": 0.9887216016404943,
      "grad_norm": 0.9586334071939676,
      "learning_rate": 1.0028244356686435e-06,
      "loss": 0.3556,
      "step": 18322
    },
    {
      "epoch": 0.9887755652690087,
      "grad_norm": 1.143543690527548,
      "learning_rate": 1.0027974750306643e-06,
      "loss": 0.5412,
      "step": 18323
    },
    {
      "epoch": 0.9888295288975231,
      "grad_norm": 1.2168848366629923,
      "learning_rate": 1.0027706436468302e-06,
      "loss": 0.3527,
      "step": 18324
    },
    {
      "epoch": 0.9888834925260375,
      "grad_norm": 0.8038001885478543,
      "learning_rate": 1.002743941517911e-06,
      "loss": 0.3098,
      "step": 18325
    },
    {
      "epoch": 0.9889374561545519,
      "grad_norm": 1.1501976199462125,
      "learning_rate": 1.002717368644676e-06,
      "loss": 0.5583,
      "step": 18326
    },
    {
      "epoch": 0.9889914197830663,
      "grad_norm": 1.0458674390584841,
      "learning_rate": 1.0026909250278875e-06,
      "loss": 0.3512,
      "step": 18327
    },
    {
      "epoch": 0.9890453834115805,
      "grad_norm": 0.9215082740320958,
      "learning_rate": 1.0026646106683061e-06,
      "loss": 0.5135,
      "step": 18328
    },
    {
      "epoch": 0.989099347040095,
      "grad_norm": 0.9944481128050141,
      "learning_rate": 1.0026384255666876e-06,
      "loss": 0.3901,
      "step": 18329
    },
    {
      "epoch": 0.9891533106686093,
      "grad_norm": 1.0945692553158084,
      "learning_rate": 1.0026123697237854e-06,
      "loss": 0.4885,
      "step": 18330
    },
    {
      "epoch": 0.9892072742971237,
      "grad_norm": 1.0166013518014263,
      "learning_rate": 1.0025864431403483e-06,
      "loss": 0.5264,
      "step": 18331
    },
    {
      "epoch": 0.9892612379256381,
      "grad_norm": 1.184214513389542,
      "learning_rate": 1.0025606458171206e-06,
      "loss": 0.5963,
      "step": 18332
    },
    {
      "epoch": 0.9893152015541525,
      "grad_norm": 1.139553714869688,
      "learning_rate": 1.0025349777548445e-06,
      "loss": 0.515,
      "step": 18333
    },
    {
      "epoch": 0.9893691651826669,
      "grad_norm": 1.0193654634684444,
      "learning_rate": 1.0025094389542579e-06,
      "loss": 0.5054,
      "step": 18334
    },
    {
      "epoch": 0.9894231288111812,
      "grad_norm": 1.1947542964153786,
      "learning_rate": 1.0024840294160936e-06,
      "loss": 0.5128,
      "step": 18335
    },
    {
      "epoch": 0.9894770924396956,
      "grad_norm": 0.8995876039764699,
      "learning_rate": 1.0024587491410834e-06,
      "loss": 0.3897,
      "step": 18336
    },
    {
      "epoch": 0.98953105606821,
      "grad_norm": 0.8578067122970012,
      "learning_rate": 1.0024335981299525e-06,
      "loss": 0.4387,
      "step": 18337
    },
    {
      "epoch": 0.9895850196967244,
      "grad_norm": 1.1837956999771444,
      "learning_rate": 1.0024085763834248e-06,
      "loss": 0.6932,
      "step": 18338
    },
    {
      "epoch": 0.9896389833252388,
      "grad_norm": 1.1535021712522362,
      "learning_rate": 1.002383683902219e-06,
      "loss": 0.5652,
      "step": 18339
    },
    {
      "epoch": 0.9896929469537532,
      "grad_norm": 1.2514453156549379,
      "learning_rate": 1.002358920687051e-06,
      "loss": 0.6481,
      "step": 18340
    },
    {
      "epoch": 0.9897469105822676,
      "grad_norm": 1.2411429186322565,
      "learning_rate": 1.002334286738632e-06,
      "loss": 0.525,
      "step": 18341
    },
    {
      "epoch": 0.9898008742107819,
      "grad_norm": 1.0740097857084852,
      "learning_rate": 1.00230978205767e-06,
      "loss": 0.5156,
      "step": 18342
    },
    {
      "epoch": 0.9898548378392963,
      "grad_norm": 1.103589238652413,
      "learning_rate": 1.0022854066448694e-06,
      "loss": 0.5069,
      "step": 18343
    },
    {
      "epoch": 0.9899088014678107,
      "grad_norm": 0.9663477784777028,
      "learning_rate": 1.0022611605009309e-06,
      "loss": 0.3375,
      "step": 18344
    },
    {
      "epoch": 0.9899627650963251,
      "grad_norm": 0.8746129084834625,
      "learning_rate": 1.0022370436265513e-06,
      "loss": 0.2935,
      "step": 18345
    },
    {
      "epoch": 0.9900167287248395,
      "grad_norm": 0.9755198597388265,
      "learning_rate": 1.0022130560224238e-06,
      "loss": 0.4055,
      "step": 18346
    },
    {
      "epoch": 0.9900706923533539,
      "grad_norm": 0.8641563753235636,
      "learning_rate": 1.0021891976892377e-06,
      "loss": 0.3442,
      "step": 18347
    },
    {
      "epoch": 0.9901246559818683,
      "grad_norm": 1.1223832691252933,
      "learning_rate": 1.0021654686276785e-06,
      "loss": 0.4355,
      "step": 18348
    },
    {
      "epoch": 0.9901786196103826,
      "grad_norm": 1.0096865343119064,
      "learning_rate": 1.0021418688384287e-06,
      "loss": 0.3916,
      "step": 18349
    },
    {
      "epoch": 0.990232583238897,
      "grad_norm": 0.9938323241284259,
      "learning_rate": 1.0021183983221662e-06,
      "loss": 0.4249,
      "step": 18350
    },
    {
      "epoch": 0.9902865468674114,
      "grad_norm": 0.9770875401195024,
      "learning_rate": 1.002095057079566e-06,
      "loss": 0.3578,
      "step": 18351
    },
    {
      "epoch": 0.9903405104959258,
      "grad_norm": 1.0376236197147999,
      "learning_rate": 1.0020718451112984e-06,
      "loss": 0.4475,
      "step": 18352
    },
    {
      "epoch": 0.9903944741244401,
      "grad_norm": 0.901504593003279,
      "learning_rate": 1.002048762418031e-06,
      "loss": 0.4348,
      "step": 18353
    },
    {
      "epoch": 0.9904484377529545,
      "grad_norm": 1.1533133356232634,
      "learning_rate": 1.002025809000427e-06,
      "loss": 0.5228,
      "step": 18354
    },
    {
      "epoch": 0.9905024013814689,
      "grad_norm": 1.2358395099757273,
      "learning_rate": 1.002002984859146e-06,
      "loss": 0.6421,
      "step": 18355
    },
    {
      "epoch": 0.9905563650099832,
      "grad_norm": 1.0244787176520822,
      "learning_rate": 1.0019802899948443e-06,
      "loss": 0.502,
      "step": 18356
    },
    {
      "epoch": 0.9906103286384976,
      "grad_norm": 1.23294297739529,
      "learning_rate": 1.0019577244081736e-06,
      "loss": 0.6117,
      "step": 18357
    },
    {
      "epoch": 0.990664292267012,
      "grad_norm": 0.8909486683820619,
      "learning_rate": 1.0019352880997832e-06,
      "loss": 0.3499,
      "step": 18358
    },
    {
      "epoch": 0.9907182558955264,
      "grad_norm": 1.0525581346518027,
      "learning_rate": 1.0019129810703171e-06,
      "loss": 0.4516,
      "step": 18359
    },
    {
      "epoch": 0.9907722195240408,
      "grad_norm": 0.9040450482318629,
      "learning_rate": 1.0018908033204178e-06,
      "loss": 0.3149,
      "step": 18360
    },
    {
      "epoch": 0.9908261831525552,
      "grad_norm": 1.023266638012293,
      "learning_rate": 1.0018687548507212e-06,
      "loss": 0.4593,
      "step": 18361
    },
    {
      "epoch": 0.9908801467810695,
      "grad_norm": 0.9783588268619413,
      "learning_rate": 1.0018468356618616e-06,
      "loss": 0.4835,
      "step": 18362
    },
    {
      "epoch": 0.9909341104095839,
      "grad_norm": 1.0783952239946148,
      "learning_rate": 1.0018250457544692e-06,
      "loss": 0.4296,
      "step": 18363
    },
    {
      "epoch": 0.9909880740380983,
      "grad_norm": 0.7904294734284498,
      "learning_rate": 1.00180338512917e-06,
      "loss": 0.3907,
      "step": 18364
    },
    {
      "epoch": 0.9910420376666127,
      "grad_norm": 0.9480792552743722,
      "learning_rate": 1.0017818537865866e-06,
      "loss": 0.3578,
      "step": 18365
    },
    {
      "epoch": 0.9910960012951271,
      "grad_norm": 1.1259146636400907,
      "learning_rate": 1.0017604517273378e-06,
      "loss": 0.5216,
      "step": 18366
    },
    {
      "epoch": 0.9911499649236415,
      "grad_norm": 1.127263832397505,
      "learning_rate": 1.0017391789520385e-06,
      "loss": 0.4962,
      "step": 18367
    },
    {
      "epoch": 0.9912039285521559,
      "grad_norm": 1.0390230515977825,
      "learning_rate": 1.0017180354613007e-06,
      "loss": 0.483,
      "step": 18368
    },
    {
      "epoch": 0.9912578921806702,
      "grad_norm": 0.8836978686143901,
      "learning_rate": 1.0016970212557315e-06,
      "loss": 0.3925,
      "step": 18369
    },
    {
      "epoch": 0.9913118558091846,
      "grad_norm": 1.1148368559451598,
      "learning_rate": 1.0016761363359355e-06,
      "loss": 0.5086,
      "step": 18370
    },
    {
      "epoch": 0.991365819437699,
      "grad_norm": 0.8930625394979353,
      "learning_rate": 1.0016553807025126e-06,
      "loss": 0.3637,
      "step": 18371
    },
    {
      "epoch": 0.9914197830662134,
      "grad_norm": 0.95530829572851,
      "learning_rate": 1.0016347543560587e-06,
      "loss": 0.4168,
      "step": 18372
    },
    {
      "epoch": 0.9914737466947278,
      "grad_norm": 0.9848685547687274,
      "learning_rate": 1.0016142572971676e-06,
      "loss": 0.3664,
      "step": 18373
    },
    {
      "epoch": 0.9915277103232422,
      "grad_norm": 1.1034275370258138,
      "learning_rate": 1.001593889526428e-06,
      "loss": 0.4282,
      "step": 18374
    },
    {
      "epoch": 0.9915816739517566,
      "grad_norm": 1.0762335724900403,
      "learning_rate": 1.001573651044425e-06,
      "loss": 0.5589,
      "step": 18375
    },
    {
      "epoch": 0.9916356375802708,
      "grad_norm": 1.0832882694564596,
      "learning_rate": 1.0015535418517412e-06,
      "loss": 0.4344,
      "step": 18376
    },
    {
      "epoch": 0.9916896012087852,
      "grad_norm": 0.9741162958993312,
      "learning_rate": 1.001533561948954e-06,
      "loss": 0.4243,
      "step": 18377
    },
    {
      "epoch": 0.9917435648372996,
      "grad_norm": 1.0538435157374961,
      "learning_rate": 1.001513711336637e-06,
      "loss": 0.3754,
      "step": 18378
    },
    {
      "epoch": 0.991797528465814,
      "grad_norm": 0.9393132399666979,
      "learning_rate": 1.0014939900153614e-06,
      "loss": 0.4748,
      "step": 18379
    },
    {
      "epoch": 0.9918514920943284,
      "grad_norm": 1.0165722228879244,
      "learning_rate": 1.0014743979856944e-06,
      "loss": 0.4148,
      "step": 18380
    },
    {
      "epoch": 0.9919054557228428,
      "grad_norm": 0.8212015058230271,
      "learning_rate": 1.0014549352481982e-06,
      "loss": 0.3125,
      "step": 18381
    },
    {
      "epoch": 0.9919594193513572,
      "grad_norm": 0.9577020112030686,
      "learning_rate": 1.0014356018034331e-06,
      "loss": 0.3572,
      "step": 18382
    },
    {
      "epoch": 0.9920133829798715,
      "grad_norm": 1.1998388303556253,
      "learning_rate": 1.001416397651954e-06,
      "loss": 0.5905,
      "step": 18383
    },
    {
      "epoch": 0.9920673466083859,
      "grad_norm": 0.7629371555310218,
      "learning_rate": 1.001397322794313e-06,
      "loss": 0.3084,
      "step": 18384
    },
    {
      "epoch": 0.9921213102369003,
      "grad_norm": 1.1516287777766407,
      "learning_rate": 1.0013783772310584e-06,
      "loss": 0.5435,
      "step": 18385
    },
    {
      "epoch": 0.9921752738654147,
      "grad_norm": 0.9510711013953445,
      "learning_rate": 1.001359560962735e-06,
      "loss": 0.337,
      "step": 18386
    },
    {
      "epoch": 0.9922292374939291,
      "grad_norm": 0.9517043076354024,
      "learning_rate": 1.0013408739898833e-06,
      "loss": 0.379,
      "step": 18387
    },
    {
      "epoch": 0.9922832011224435,
      "grad_norm": 1.2029351126352912,
      "learning_rate": 1.0013223163130402e-06,
      "loss": 0.6458,
      "step": 18388
    },
    {
      "epoch": 0.9923371647509579,
      "grad_norm": 1.0955888460614083,
      "learning_rate": 1.0013038879327394e-06,
      "loss": 0.4326,
      "step": 18389
    },
    {
      "epoch": 0.9923911283794722,
      "grad_norm": 1.0751470288464444,
      "learning_rate": 1.0012855888495108e-06,
      "loss": 0.4891,
      "step": 18390
    },
    {
      "epoch": 0.9924450920079866,
      "grad_norm": 0.9190426947517256,
      "learning_rate": 1.0012674190638795e-06,
      "loss": 0.3628,
      "step": 18391
    },
    {
      "epoch": 0.992499055636501,
      "grad_norm": 0.95073847634167,
      "learning_rate": 1.0012493785763689e-06,
      "loss": 0.5119,
      "step": 18392
    },
    {
      "epoch": 0.9925530192650154,
      "grad_norm": 0.930747869874686,
      "learning_rate": 1.0012314673874961e-06,
      "loss": 0.3933,
      "step": 18393
    },
    {
      "epoch": 0.9926069828935298,
      "grad_norm": 0.9499963741104693,
      "learning_rate": 1.0012136854977772e-06,
      "loss": 0.3633,
      "step": 18394
    },
    {
      "epoch": 0.9926609465220442,
      "grad_norm": 1.0858446387932625,
      "learning_rate": 1.0011960329077223e-06,
      "loss": 0.5875,
      "step": 18395
    },
    {
      "epoch": 0.9927149101505586,
      "grad_norm": 1.1518466083744512,
      "learning_rate": 1.0011785096178393e-06,
      "loss": 0.5438,
      "step": 18396
    },
    {
      "epoch": 0.9927688737790729,
      "grad_norm": 1.0811681321936708,
      "learning_rate": 1.0011611156286311e-06,
      "loss": 0.4842,
      "step": 18397
    },
    {
      "epoch": 0.9928228374075873,
      "grad_norm": 0.8843053358299039,
      "learning_rate": 1.001143850940599e-06,
      "loss": 0.4117,
      "step": 18398
    },
    {
      "epoch": 0.9928768010361017,
      "grad_norm": 1.2024688789809779,
      "learning_rate": 1.0011267155542377e-06,
      "loss": 0.4346,
      "step": 18399
    },
    {
      "epoch": 0.992930764664616,
      "grad_norm": 1.2967011364989678,
      "learning_rate": 1.0011097094700408e-06,
      "loss": 0.5817,
      "step": 18400
    },
    {
      "epoch": 0.9929847282931304,
      "grad_norm": 0.7955485266642397,
      "learning_rate": 1.0010928326884966e-06,
      "loss": 0.3022,
      "step": 18401
    },
    {
      "epoch": 0.9930386919216448,
      "grad_norm": 0.923764783683336,
      "learning_rate": 1.00107608521009e-06,
      "loss": 0.3615,
      "step": 18402
    },
    {
      "epoch": 0.9930926555501592,
      "grad_norm": 0.9749643380748793,
      "learning_rate": 1.0010594670353027e-06,
      "loss": 0.4051,
      "step": 18403
    },
    {
      "epoch": 0.9931466191786735,
      "grad_norm": 1.035658140594875,
      "learning_rate": 1.001042978164612e-06,
      "loss": 0.5249,
      "step": 18404
    },
    {
      "epoch": 0.9932005828071879,
      "grad_norm": 1.0568138016088358,
      "learning_rate": 1.0010266185984921e-06,
      "loss": 0.4302,
      "step": 18405
    },
    {
      "epoch": 0.9932545464357023,
      "grad_norm": 1.0740356356009348,
      "learning_rate": 1.001010388337413e-06,
      "loss": 0.527,
      "step": 18406
    },
    {
      "epoch": 0.9933085100642167,
      "grad_norm": 0.9737360580959202,
      "learning_rate": 1.0009942873818416e-06,
      "loss": 0.6706,
      "step": 18407
    },
    {
      "epoch": 0.9933624736927311,
      "grad_norm": 0.6775708065987449,
      "learning_rate": 1.0009783157322397e-06,
      "loss": 0.3566,
      "step": 18408
    },
    {
      "epoch": 0.9934164373212455,
      "grad_norm": 0.9959302803493967,
      "learning_rate": 1.0009624733890674e-06,
      "loss": 0.3142,
      "step": 18409
    },
    {
      "epoch": 0.9934704009497599,
      "grad_norm": 1.0393653993614935,
      "learning_rate": 1.0009467603527793e-06,
      "loss": 0.4812,
      "step": 18410
    },
    {
      "epoch": 0.9935243645782742,
      "grad_norm": 1.1407391525944253,
      "learning_rate": 1.0009311766238277e-06,
      "loss": 0.4533,
      "step": 18411
    },
    {
      "epoch": 0.9935783282067886,
      "grad_norm": 1.1092502471514785,
      "learning_rate": 1.00091572220266e-06,
      "loss": 0.5114,
      "step": 18412
    },
    {
      "epoch": 0.993632291835303,
      "grad_norm": 0.7989078186076607,
      "learning_rate": 1.0009003970897197e-06,
      "loss": 0.3002,
      "step": 18413
    },
    {
      "epoch": 0.9936862554638174,
      "grad_norm": 1.167664944283381,
      "learning_rate": 1.0008852012854482e-06,
      "loss": 0.605,
      "step": 18414
    },
    {
      "epoch": 0.9937402190923318,
      "grad_norm": 1.1305384630048814,
      "learning_rate": 1.0008701347902827e-06,
      "loss": 0.5035,
      "step": 18415
    },
    {
      "epoch": 0.9937941827208462,
      "grad_norm": 0.8674929221047492,
      "learning_rate": 1.000855197604655e-06,
      "loss": 0.4271,
      "step": 18416
    },
    {
      "epoch": 0.9938481463493606,
      "grad_norm": 1.1583306752921834,
      "learning_rate": 1.0008403897289952e-06,
      "loss": 0.5807,
      "step": 18417
    },
    {
      "epoch": 0.9939021099778749,
      "grad_norm": 1.230555165269435,
      "learning_rate": 1.0008257111637285e-06,
      "loss": 0.5731,
      "step": 18418
    },
    {
      "epoch": 0.9939560736063893,
      "grad_norm": 1.0581796193135984,
      "learning_rate": 1.0008111619092773e-06,
      "loss": 0.459,
      "step": 18419
    },
    {
      "epoch": 0.9940100372349037,
      "grad_norm": 0.7622152453154231,
      "learning_rate": 1.0007967419660592e-06,
      "loss": 0.3016,
      "step": 18420
    },
    {
      "epoch": 0.9940640008634181,
      "grad_norm": 1.0353011308919744,
      "learning_rate": 1.0007824513344886e-06,
      "loss": 0.3731,
      "step": 18421
    },
    {
      "epoch": 0.9941179644919325,
      "grad_norm": 1.3480856936603853,
      "learning_rate": 1.0007682900149766e-06,
      "loss": 0.5996,
      "step": 18422
    },
    {
      "epoch": 0.9941719281204469,
      "grad_norm": 1.1174011637319903,
      "learning_rate": 1.0007542580079302e-06,
      "loss": 0.5645,
      "step": 18423
    },
    {
      "epoch": 0.9942258917489613,
      "grad_norm": 0.8257801964247194,
      "learning_rate": 1.0007403553137524e-06,
      "loss": 0.2721,
      "step": 18424
    },
    {
      "epoch": 0.9942798553774755,
      "grad_norm": 1.0408760875791672,
      "learning_rate": 1.0007265819328437e-06,
      "loss": 0.3688,
      "step": 18425
    },
    {
      "epoch": 0.9943338190059899,
      "grad_norm": 1.1270854376399546,
      "learning_rate": 1.000712937865598e-06,
      "loss": 0.4614,
      "step": 18426
    },
    {
      "epoch": 0.9943877826345043,
      "grad_norm": 1.0175077814531839,
      "learning_rate": 1.0006994231124095e-06,
      "loss": 0.5991,
      "step": 18427
    },
    {
      "epoch": 0.9944417462630187,
      "grad_norm": 1.0840588258282462,
      "learning_rate": 1.0006860376736655e-06,
      "loss": 0.5761,
      "step": 18428
    },
    {
      "epoch": 0.9944957098915331,
      "grad_norm": 1.081991988769724,
      "learning_rate": 1.0006727815497511e-06,
      "loss": 0.4691,
      "step": 18429
    },
    {
      "epoch": 0.9945496735200475,
      "grad_norm": 0.9791711874572566,
      "learning_rate": 1.0006596547410475e-06,
      "loss": 0.5432,
      "step": 18430
    },
    {
      "epoch": 0.9946036371485618,
      "grad_norm": 0.934016518513832,
      "learning_rate": 1.0006466572479311e-06,
      "loss": 0.4284,
      "step": 18431
    },
    {
      "epoch": 0.9946576007770762,
      "grad_norm": 1.0471205434015671,
      "learning_rate": 1.0006337890707764e-06,
      "loss": 0.4074,
      "step": 18432
    },
    {
      "epoch": 0.9947115644055906,
      "grad_norm": 0.9709949281442777,
      "learning_rate": 1.0006210502099527e-06,
      "loss": 0.439,
      "step": 18433
    },
    {
      "epoch": 0.994765528034105,
      "grad_norm": 1.0319603788403233,
      "learning_rate": 1.0006084406658263e-06,
      "loss": 0.3731,
      "step": 18434
    },
    {
      "epoch": 0.9948194916626194,
      "grad_norm": 1.1243795166210273,
      "learning_rate": 1.00059596043876e-06,
      "loss": 0.3781,
      "step": 18435
    },
    {
      "epoch": 0.9948734552911338,
      "grad_norm": 0.9512007943250874,
      "learning_rate": 1.0005836095291116e-06,
      "loss": 0.3594,
      "step": 18436
    },
    {
      "epoch": 0.9949274189196482,
      "grad_norm": 1.0383897528421595,
      "learning_rate": 1.000571387937237e-06,
      "loss": 0.3928,
      "step": 18437
    },
    {
      "epoch": 0.9949813825481625,
      "grad_norm": 1.0940583949837202,
      "learning_rate": 1.0005592956634867e-06,
      "loss": 0.4174,
      "step": 18438
    },
    {
      "epoch": 0.9950353461766769,
      "grad_norm": 1.0568149430203366,
      "learning_rate": 1.0005473327082089e-06,
      "loss": 0.5858,
      "step": 18439
    },
    {
      "epoch": 0.9950893098051913,
      "grad_norm": 0.8174669010508476,
      "learning_rate": 1.0005354990717472e-06,
      "loss": 0.3311,
      "step": 18440
    },
    {
      "epoch": 0.9951432734337057,
      "grad_norm": 1.2467594726548978,
      "learning_rate": 1.0005237947544411e-06,
      "loss": 0.4504,
      "step": 18441
    },
    {
      "epoch": 0.9951972370622201,
      "grad_norm": 1.0650601044891532,
      "learning_rate": 1.000512219756628e-06,
      "loss": 0.4366,
      "step": 18442
    },
    {
      "epoch": 0.9952512006907345,
      "grad_norm": 1.2101391844157494,
      "learning_rate": 1.0005007740786396e-06,
      "loss": 0.4282,
      "step": 18443
    },
    {
      "epoch": 0.9953051643192489,
      "grad_norm": 1.0443617491936916,
      "learning_rate": 1.0004894577208057e-06,
      "loss": 0.4815,
      "step": 18444
    },
    {
      "epoch": 0.9953591279477632,
      "grad_norm": 1.1946815732524485,
      "learning_rate": 1.0004782706834513e-06,
      "loss": 0.5652,
      "step": 18445
    },
    {
      "epoch": 0.9954130915762776,
      "grad_norm": 1.1880164362836878,
      "learning_rate": 1.000467212966898e-06,
      "loss": 0.4606,
      "step": 18446
    },
    {
      "epoch": 0.995467055204792,
      "grad_norm": 1.1558755508925032,
      "learning_rate": 1.0004562845714629e-06,
      "loss": 0.458,
      "step": 18447
    },
    {
      "epoch": 0.9955210188333063,
      "grad_norm": 1.0736900782497427,
      "learning_rate": 1.0004454854974607e-06,
      "loss": 0.4606,
      "step": 18448
    },
    {
      "epoch": 0.9955749824618207,
      "grad_norm": 1.2182351023462827,
      "learning_rate": 1.000434815745202e-06,
      "loss": 0.3983,
      "step": 18449
    },
    {
      "epoch": 0.9956289460903351,
      "grad_norm": 1.0677978256505927,
      "learning_rate": 1.0004242753149934e-06,
      "loss": 0.4469,
      "step": 18450
    },
    {
      "epoch": 0.9956829097188495,
      "grad_norm": 0.6395682212083621,
      "learning_rate": 1.0004138642071371e-06,
      "loss": 0.2482,
      "step": 18451
    },
    {
      "epoch": 0.9957368733473638,
      "grad_norm": 0.8296826586991661,
      "learning_rate": 1.0004035824219334e-06,
      "loss": 0.2612,
      "step": 18452
    },
    {
      "epoch": 0.9957908369758782,
      "grad_norm": 1.0048259610996664,
      "learning_rate": 1.0003934299596769e-06,
      "loss": 0.4729,
      "step": 18453
    },
    {
      "epoch": 0.9958448006043926,
      "grad_norm": 0.9542158817059453,
      "learning_rate": 1.0003834068206597e-06,
      "loss": 0.4195,
      "step": 18454
    },
    {
      "epoch": 0.995898764232907,
      "grad_norm": 1.2992264810648746,
      "learning_rate": 1.0003735130051698e-06,
      "loss": 0.5161,
      "step": 18455
    },
    {
      "epoch": 0.9959527278614214,
      "grad_norm": 0.949937494940201,
      "learning_rate": 1.0003637485134921e-06,
      "loss": 0.5017,
      "step": 18456
    },
    {
      "epoch": 0.9960066914899358,
      "grad_norm": 1.0234515795574237,
      "learning_rate": 1.0003541133459063e-06,
      "loss": 0.4782,
      "step": 18457
    },
    {
      "epoch": 0.9960606551184502,
      "grad_norm": 1.0725541641268932,
      "learning_rate": 1.0003446075026902e-06,
      "loss": 0.3464,
      "step": 18458
    },
    {
      "epoch": 0.9961146187469645,
      "grad_norm": 1.2342650459286182,
      "learning_rate": 1.0003352309841166e-06,
      "loss": 0.4981,
      "step": 18459
    },
    {
      "epoch": 0.9961685823754789,
      "grad_norm": 0.8611438178691448,
      "learning_rate": 1.0003259837904555e-06,
      "loss": 0.3856,
      "step": 18460
    },
    {
      "epoch": 0.9962225460039933,
      "grad_norm": 1.0097774627651561,
      "learning_rate": 1.0003168659219717e-06,
      "loss": 0.3344,
      "step": 18461
    },
    {
      "epoch": 0.9962765096325077,
      "grad_norm": 0.95818852402117,
      "learning_rate": 1.0003078773789279e-06,
      "loss": 0.446,
      "step": 18462
    },
    {
      "epoch": 0.9963304732610221,
      "grad_norm": 0.9284324794492886,
      "learning_rate": 1.0002990181615821e-06,
      "loss": 0.4524,
      "step": 18463
    },
    {
      "epoch": 0.9963844368895365,
      "grad_norm": 1.05683498049531,
      "learning_rate": 1.0002902882701888e-06,
      "loss": 0.3809,
      "step": 18464
    },
    {
      "epoch": 0.9964384005180509,
      "grad_norm": 1.0848425931906238,
      "learning_rate": 1.0002816877049999e-06,
      "loss": 0.4483,
      "step": 18465
    },
    {
      "epoch": 0.9964923641465652,
      "grad_norm": 1.2403069683654322,
      "learning_rate": 1.0002732164662618e-06,
      "loss": 0.5565,
      "step": 18466
    },
    {
      "epoch": 0.9965463277750796,
      "grad_norm": 0.9236353907334149,
      "learning_rate": 1.0002648745542181e-06,
      "loss": 0.3768,
      "step": 18467
    },
    {
      "epoch": 0.996600291403594,
      "grad_norm": 0.8173897229028756,
      "learning_rate": 1.0002566619691082e-06,
      "loss": 0.3635,
      "step": 18468
    },
    {
      "epoch": 0.9966542550321084,
      "grad_norm": 1.153126541301487,
      "learning_rate": 1.0002485787111687e-06,
      "loss": 0.3295,
      "step": 18469
    },
    {
      "epoch": 0.9967082186606228,
      "grad_norm": 1.1644339498191476,
      "learning_rate": 1.0002406247806319e-06,
      "loss": 0.5345,
      "step": 18470
    },
    {
      "epoch": 0.9967621822891372,
      "grad_norm": 1.0862242848214658,
      "learning_rate": 1.0002328001777263e-06,
      "loss": 0.4913,
      "step": 18471
    },
    {
      "epoch": 0.9968161459176516,
      "grad_norm": 1.1026482639617785,
      "learning_rate": 1.0002251049026762e-06,
      "loss": 0.445,
      "step": 18472
    },
    {
      "epoch": 0.9968701095461658,
      "grad_norm": 1.0018764522824024,
      "learning_rate": 1.0002175389557037e-06,
      "loss": 0.4788,
      "step": 18473
    },
    {
      "epoch": 0.9969240731746802,
      "grad_norm": 1.0102257144856674,
      "learning_rate": 1.000210102337026e-06,
      "loss": 0.4552,
      "step": 18474
    },
    {
      "epoch": 0.9969780368031946,
      "grad_norm": 1.3933708877797142,
      "learning_rate": 1.0002027950468564e-06,
      "loss": 0.6766,
      "step": 18475
    },
    {
      "epoch": 0.997032000431709,
      "grad_norm": 0.9302169750457506,
      "learning_rate": 1.0001956170854051e-06,
      "loss": 0.3355,
      "step": 18476
    },
    {
      "epoch": 0.9970859640602234,
      "grad_norm": 1.1552161736911235,
      "learning_rate": 1.0001885684528785e-06,
      "loss": 0.4796,
      "step": 18477
    },
    {
      "epoch": 0.9971399276887378,
      "grad_norm": 0.9029034204146128,
      "learning_rate": 1.0001816491494793e-06,
      "loss": 0.511,
      "step": 18478
    },
    {
      "epoch": 0.9971938913172522,
      "grad_norm": 0.8282888823759261,
      "learning_rate": 1.0001748591754058e-06,
      "loss": 0.3374,
      "step": 18479
    },
    {
      "epoch": 0.9972478549457665,
      "grad_norm": 1.2440616319322944,
      "learning_rate": 1.000168198530854e-06,
      "loss": 0.701,
      "step": 18480
    },
    {
      "epoch": 0.9973018185742809,
      "grad_norm": 1.218837811911342,
      "learning_rate": 1.000161667216015e-06,
      "loss": 0.4887,
      "step": 18481
    },
    {
      "epoch": 0.9973557822027953,
      "grad_norm": 0.9913560782834503,
      "learning_rate": 1.0001552652310763e-06,
      "loss": 0.3704,
      "step": 18482
    },
    {
      "epoch": 0.9974097458313097,
      "grad_norm": 1.083338640168778,
      "learning_rate": 1.0001489925762223e-06,
      "loss": 0.4631,
      "step": 18483
    },
    {
      "epoch": 0.9974637094598241,
      "grad_norm": 1.205246622145469,
      "learning_rate": 1.0001428492516326e-06,
      "loss": 0.6625,
      "step": 18484
    },
    {
      "epoch": 0.9975176730883385,
      "grad_norm": 0.8565959092290395,
      "learning_rate": 1.0001368352574845e-06,
      "loss": 0.3393,
      "step": 18485
    },
    {
      "epoch": 0.9975716367168529,
      "grad_norm": 1.0300081874188178,
      "learning_rate": 1.00013095059395e-06,
      "loss": 0.5637,
      "step": 18486
    },
    {
      "epoch": 0.9976256003453672,
      "grad_norm": 0.9834691041930705,
      "learning_rate": 1.0001251952611991e-06,
      "loss": 0.414,
      "step": 18487
    },
    {
      "epoch": 0.9976795639738816,
      "grad_norm": 1.1629181261064778,
      "learning_rate": 1.0001195692593972e-06,
      "loss": 0.4638,
      "step": 18488
    },
    {
      "epoch": 0.997733527602396,
      "grad_norm": 0.9788066965044775,
      "learning_rate": 1.0001140725887052e-06,
      "loss": 0.4518,
      "step": 18489
    },
    {
      "epoch": 0.9977874912309104,
      "grad_norm": 1.2282858037990427,
      "learning_rate": 1.0001087052492817e-06,
      "loss": 0.6476,
      "step": 18490
    },
    {
      "epoch": 0.9978414548594248,
      "grad_norm": 1.16931709141275,
      "learning_rate": 1.0001034672412809e-06,
      "loss": 0.511,
      "step": 18491
    },
    {
      "epoch": 0.9978954184879392,
      "grad_norm": 1.0003444301489588,
      "learning_rate": 1.0000983585648536e-06,
      "loss": 0.3795,
      "step": 18492
    },
    {
      "epoch": 0.9979493821164536,
      "grad_norm": 0.9238154243232823,
      "learning_rate": 1.000093379220146e-06,
      "loss": 0.497,
      "step": 18493
    },
    {
      "epoch": 0.9980033457449679,
      "grad_norm": 1.0165961505919756,
      "learning_rate": 1.0000885292073011e-06,
      "loss": 0.42,
      "step": 18494
    },
    {
      "epoch": 0.9980573093734822,
      "grad_norm": 1.2355485625727491,
      "learning_rate": 1.0000838085264594e-06,
      "loss": 0.4648,
      "step": 18495
    },
    {
      "epoch": 0.9981112730019966,
      "grad_norm": 1.074809869867822,
      "learning_rate": 1.0000792171777552e-06,
      "loss": 0.536,
      "step": 18496
    },
    {
      "epoch": 0.998165236630511,
      "grad_norm": 1.026334321632097,
      "learning_rate": 1.0000747551613218e-06,
      "loss": 0.406,
      "step": 18497
    },
    {
      "epoch": 0.9982192002590254,
      "grad_norm": 1.1312012954783295,
      "learning_rate": 1.0000704224772864e-06,
      "loss": 0.5423,
      "step": 18498
    },
    {
      "epoch": 0.9982731638875398,
      "grad_norm": 0.9747373496655249,
      "learning_rate": 1.000066219125774e-06,
      "loss": 0.4517,
      "step": 18499
    },
    {
      "epoch": 0.9983271275160541,
      "grad_norm": 1.1569161390039195,
      "learning_rate": 1.000062145106905e-06,
      "loss": 0.4424,
      "step": 18500
    },
    {
      "epoch": 0.9983271275160541,
      "eval_loss": 0.526906430721283,
      "eval_runtime": 163.1155,
      "eval_samples_per_second": 21.083,
      "eval_steps_per_second": 0.883,
      "step": 18500
    },
    {
      "epoch": 0.9983810911445685,
      "grad_norm": 1.163719449613361,
      "learning_rate": 1.0000582004207968e-06,
      "loss": 0.4895,
      "step": 18501
    },
    {
      "epoch": 0.9984350547730829,
      "grad_norm": 1.149728121926013,
      "learning_rate": 1.0000543850675636e-06,
      "loss": 0.5625,
      "step": 18502
    },
    {
      "epoch": 0.9984890184015973,
      "grad_norm": 1.0144566652765594,
      "learning_rate": 1.0000506990473135e-06,
      "loss": 0.4175,
      "step": 18503
    },
    {
      "epoch": 0.9985429820301117,
      "grad_norm": 1.1109847492181069,
      "learning_rate": 1.0000471423601532e-06,
      "loss": 0.445,
      "step": 18504
    },
    {
      "epoch": 0.9985969456586261,
      "grad_norm": 0.9688401126550942,
      "learning_rate": 1.000043715006185e-06,
      "loss": 0.4086,
      "step": 18505
    },
    {
      "epoch": 0.9986509092871405,
      "grad_norm": 0.9361366216141924,
      "learning_rate": 1.0000404169855076e-06,
      "loss": 0.4652,
      "step": 18506
    },
    {
      "epoch": 0.9987048729156548,
      "grad_norm": 1.1167180563340167,
      "learning_rate": 1.000037248298215e-06,
      "loss": 0.4973,
      "step": 18507
    },
    {
      "epoch": 0.9987588365441692,
      "grad_norm": 1.0932232512383238,
      "learning_rate": 1.0000342089443993e-06,
      "loss": 0.4951,
      "step": 18508
    },
    {
      "epoch": 0.9988128001726836,
      "grad_norm": 1.2186056130783893,
      "learning_rate": 1.000031298924147e-06,
      "loss": 0.3838,
      "step": 18509
    },
    {
      "epoch": 0.998866763801198,
      "grad_norm": 1.1255029218921468,
      "learning_rate": 1.0000285182375422e-06,
      "loss": 0.5443,
      "step": 18510
    },
    {
      "epoch": 0.9989207274297124,
      "grad_norm": 0.974934430479258,
      "learning_rate": 1.0000258668846643e-06,
      "loss": 0.3418,
      "step": 18511
    },
    {
      "epoch": 0.9989746910582268,
      "grad_norm": 1.0549676270814177,
      "learning_rate": 1.0000233448655903e-06,
      "loss": 0.3935,
      "step": 18512
    },
    {
      "epoch": 0.9990286546867412,
      "grad_norm": 1.0916380697085497,
      "learning_rate": 1.0000209521803917e-06,
      "loss": 0.4953,
      "step": 18513
    },
    {
      "epoch": 0.9990826183152555,
      "grad_norm": 1.1054779283956757,
      "learning_rate": 1.0000186888291385e-06,
      "loss": 0.4542,
      "step": 18514
    },
    {
      "epoch": 0.9991365819437699,
      "grad_norm": 0.8938320323817286,
      "learning_rate": 1.000016554811895e-06,
      "loss": 0.3199,
      "step": 18515
    },
    {
      "epoch": 0.9991905455722843,
      "grad_norm": 1.0392140901034475,
      "learning_rate": 1.0000145501287225e-06,
      "loss": 0.4311,
      "step": 18516
    },
    {
      "epoch": 0.9992445092007987,
      "grad_norm": 1.0965112771103944,
      "learning_rate": 1.0000126747796783e-06,
      "loss": 0.4336,
      "step": 18517
    },
    {
      "epoch": 0.9992984728293131,
      "grad_norm": 1.0130596924882802,
      "learning_rate": 1.0000109287648174e-06,
      "loss": 0.4628,
      "step": 18518
    },
    {
      "epoch": 0.9993524364578275,
      "grad_norm": 0.9469850989468779,
      "learning_rate": 1.0000093120841888e-06,
      "loss": 0.5356,
      "step": 18519
    },
    {
      "epoch": 0.9994064000863419,
      "grad_norm": 0.7398953533968945,
      "learning_rate": 1.0000078247378398e-06,
      "loss": 0.3257,
      "step": 18520
    },
    {
      "epoch": 0.9994603637148561,
      "grad_norm": 0.9716800188101506,
      "learning_rate": 1.0000064667258125e-06,
      "loss": 0.3607,
      "step": 18521
    },
    {
      "epoch": 0.9995143273433705,
      "grad_norm": 0.8158076879826773,
      "learning_rate": 1.000005238048147e-06,
      "loss": 0.2466,
      "step": 18522
    },
    {
      "epoch": 0.9995682909718849,
      "grad_norm": 0.9051159237310575,
      "learning_rate": 1.000004138704877e-06,
      "loss": 0.3083,
      "step": 18523
    },
    {
      "epoch": 0.9996222546003993,
      "grad_norm": 1.0624751290394137,
      "learning_rate": 1.0000031686960351e-06,
      "loss": 0.4858,
      "step": 18524
    },
    {
      "epoch": 0.9996762182289137,
      "grad_norm": 1.0178736989462844,
      "learning_rate": 1.0000023280216497e-06,
      "loss": 0.4919,
      "step": 18525
    },
    {
      "epoch": 0.9997301818574281,
      "grad_norm": 0.8231678679540493,
      "learning_rate": 1.0000016166817437e-06,
      "loss": 0.3226,
      "step": 18526
    },
    {
      "epoch": 0.9997841454859425,
      "grad_norm": 1.2681969720075554,
      "learning_rate": 1.0000010346763381e-06,
      "loss": 0.7385,
      "step": 18527
    },
    {
      "epoch": 0.9998381091144568,
      "grad_norm": 1.1563989415460567,
      "learning_rate": 1.00000058200545e-06,
      "loss": 0.5028,
      "step": 18528
    },
    {
      "epoch": 0.9998920727429712,
      "grad_norm": 1.0234185229700692,
      "learning_rate": 1.0000002586690923e-06,
      "loss": 0.4663,
      "step": 18529
    },
    {
      "epoch": 0.9999460363714856,
      "grad_norm": 0.9519100534808891,
      "learning_rate": 1.0000000646672737e-06,
      "loss": 0.392,
      "step": 18530
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8605762284142406,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.294,
      "step": 18531
    }
  ],
  "logging_steps": 1,
  "max_steps": 18531,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3167434831822848.0,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}