{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999992215899803,
  "eval_steps": 500,
  "global_step": 64233,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 7.784100196937735e-05,
      "grad_norm": 18.677021026611328,
      "learning_rate": 1.2453300124533002e-09,
      "loss": 1.0606,
      "step": 5
    },
    {
      "epoch": 0.0001556820039387547,
      "grad_norm": 8.49953842163086,
      "learning_rate": 2.801992528019925e-09,
      "loss": 1.0495,
      "step": 10
    },
    {
      "epoch": 0.00023352300590813205,
      "grad_norm": 4.764372825622559,
      "learning_rate": 4.358655043586551e-09,
      "loss": 1.0324,
      "step": 15
    },
    {
      "epoch": 0.0003113640078775094,
      "grad_norm": 6.350065231323242,
      "learning_rate": 5.915317559153175e-09,
      "loss": 1.0236,
      "step": 20
    },
    {
      "epoch": 0.00038920500984688676,
      "grad_norm": 21.48935317993164,
      "learning_rate": 7.471980074719801e-09,
      "loss": 1.0156,
      "step": 25
    },
    {
      "epoch": 0.0004670460118162641,
      "grad_norm": 11.964753150939941,
      "learning_rate": 9.028642590286426e-09,
      "loss": 1.1045,
      "step": 30
    },
    {
      "epoch": 0.0005448870137856414,
      "grad_norm": 16.780696868896484,
      "learning_rate": 1.0585305105853052e-08,
      "loss": 0.9926,
      "step": 35
    },
    {
      "epoch": 0.0006227280157550188,
      "grad_norm": 9.311758041381836,
      "learning_rate": 1.2141967621419675e-08,
      "loss": 1.0672,
      "step": 40
    },
    {
      "epoch": 0.0007005690177243961,
      "grad_norm": 10.671490669250488,
      "learning_rate": 1.36986301369863e-08,
      "loss": 1.1432,
      "step": 45
    },
    {
      "epoch": 0.0007784100196937735,
      "grad_norm": 6.056899547576904,
      "learning_rate": 1.5255292652552926e-08,
      "loss": 1.0207,
      "step": 50
    },
    {
      "epoch": 0.0008562510216631508,
      "grad_norm": 12.727471351623535,
      "learning_rate": 1.6811955168119553e-08,
      "loss": 1.0626,
      "step": 55
    },
    {
      "epoch": 0.0009340920236325282,
      "grad_norm": 3.2297894954681396,
      "learning_rate": 1.8368617683686178e-08,
      "loss": 1.0551,
      "step": 60
    },
    {
      "epoch": 0.0010119330256019056,
      "grad_norm": 5.09151554107666,
      "learning_rate": 1.9925280199252803e-08,
      "loss": 1.0051,
      "step": 65
    },
    {
      "epoch": 0.0010897740275712829,
      "grad_norm": 9.829240798950195,
      "learning_rate": 2.1481942714819424e-08,
      "loss": 1.0936,
      "step": 70
    },
    {
      "epoch": 0.0011676150295406602,
      "grad_norm": 4.1169023513793945,
      "learning_rate": 2.3038605230386048e-08,
      "loss": 0.9367,
      "step": 75
    },
    {
      "epoch": 0.0012454560315100377,
      "grad_norm": 22.784198760986328,
      "learning_rate": 2.4595267745952676e-08,
      "loss": 1.1475,
      "step": 80
    },
    {
      "epoch": 0.001323297033479415,
      "grad_norm": 6.216701507568359,
      "learning_rate": 2.61519302615193e-08,
      "loss": 1.1738,
      "step": 85
    },
    {
      "epoch": 0.0014011380354487922,
      "grad_norm": 8.767633438110352,
      "learning_rate": 2.7708592777085925e-08,
      "loss": 1.0995,
      "step": 90
    },
    {
      "epoch": 0.0014789790374181697,
      "grad_norm": 6.444150447845459,
      "learning_rate": 2.926525529265255e-08,
      "loss": 1.1674,
      "step": 95
    },
    {
      "epoch": 0.001556820039387547,
      "grad_norm": 21.939842224121094,
      "learning_rate": 3.082191780821918e-08,
      "loss": 0.9658,
      "step": 100
    },
    {
      "epoch": 0.0016346610413569243,
      "grad_norm": 6.07455587387085,
      "learning_rate": 3.23785803237858e-08,
      "loss": 1.3969,
      "step": 105
    },
    {
      "epoch": 0.0017125020433263016,
      "grad_norm": 7.66893196105957,
      "learning_rate": 3.3935242839352427e-08,
      "loss": 1.0274,
      "step": 110
    },
    {
      "epoch": 0.0017903430452956791,
      "grad_norm": 6.411283016204834,
      "learning_rate": 3.549190535491906e-08,
      "loss": 1.0922,
      "step": 115
    },
    {
      "epoch": 0.0018681840472650564,
      "grad_norm": 15.535103797912598,
      "learning_rate": 3.704856787048568e-08,
      "loss": 1.058,
      "step": 120
    },
    {
      "epoch": 0.0019460250492344337,
      "grad_norm": 13.108068466186523,
      "learning_rate": 3.860523038605231e-08,
      "loss": 1.1347,
      "step": 125
    },
    {
      "epoch": 0.002023866051203811,
      "grad_norm": 5.452599048614502,
      "learning_rate": 4.016189290161893e-08,
      "loss": 1.07,
      "step": 130
    },
    {
      "epoch": 0.0021017070531731885,
      "grad_norm": 13.57522964477539,
      "learning_rate": 4.1718555417185556e-08,
      "loss": 1.0556,
      "step": 135
    },
    {
      "epoch": 0.0021795480551425658,
      "grad_norm": 4.844541072845459,
      "learning_rate": 4.3275217932752174e-08,
      "loss": 1.0538,
      "step": 140
    },
    {
      "epoch": 0.002257389057111943,
      "grad_norm": 7.6000800132751465,
      "learning_rate": 4.48318804483188e-08,
      "loss": 1.0884,
      "step": 145
    },
    {
      "epoch": 0.0023352300590813203,
      "grad_norm": 6.445258617401123,
      "learning_rate": 4.638854296388542e-08,
      "loss": 1.0247,
      "step": 150
    },
    {
      "epoch": 0.002413071061050698,
      "grad_norm": 4.861091136932373,
      "learning_rate": 4.794520547945205e-08,
      "loss": 1.0553,
      "step": 155
    },
    {
      "epoch": 0.0024909120630200753,
      "grad_norm": 6.040435314178467,
      "learning_rate": 4.950186799501867e-08,
      "loss": 1.0682,
      "step": 160
    },
    {
      "epoch": 0.0025687530649894526,
      "grad_norm": 10.561899185180664,
      "learning_rate": 5.10585305105853e-08,
      "loss": 0.9904,
      "step": 165
    },
    {
      "epoch": 0.00264659406695883,
      "grad_norm": 8.238300323486328,
      "learning_rate": 5.261519302615193e-08,
      "loss": 1.0576,
      "step": 170
    },
    {
      "epoch": 0.002724435068928207,
      "grad_norm": 10.821751594543457,
      "learning_rate": 5.417185554171855e-08,
      "loss": 1.0507,
      "step": 175
    },
    {
      "epoch": 0.0028022760708975845,
      "grad_norm": 20.215164184570312,
      "learning_rate": 5.5728518057285177e-08,
      "loss": 1.2059,
      "step": 180
    },
    {
      "epoch": 0.0028801170728669618,
      "grad_norm": 15.447042465209961,
      "learning_rate": 5.72851805728518e-08,
      "loss": 1.1047,
      "step": 185
    },
    {
      "epoch": 0.0029579580748363395,
      "grad_norm": 13.472341537475586,
      "learning_rate": 5.8841843088418426e-08,
      "loss": 1.0508,
      "step": 190
    },
    {
      "epoch": 0.003035799076805717,
      "grad_norm": 16.09784507751465,
      "learning_rate": 6.039850560398505e-08,
      "loss": 1.1144,
      "step": 195
    },
    {
      "epoch": 0.003113640078775094,
      "grad_norm": 5.519948959350586,
      "learning_rate": 6.195516811955167e-08,
      "loss": 1.1127,
      "step": 200
    },
    {
      "epoch": 0.0031914810807444714,
      "grad_norm": 9.467545509338379,
      "learning_rate": 6.351183063511831e-08,
      "loss": 1.0882,
      "step": 205
    },
    {
      "epoch": 0.0032693220827138486,
      "grad_norm": 8.895452499389648,
      "learning_rate": 6.506849315068492e-08,
      "loss": 1.0845,
      "step": 210
    },
    {
      "epoch": 0.003347163084683226,
      "grad_norm": 10.007709503173828,
      "learning_rate": 6.662515566625156e-08,
      "loss": 1.1585,
      "step": 215
    },
    {
      "epoch": 0.0034250040866526032,
      "grad_norm": 10.499605178833008,
      "learning_rate": 6.818181818181817e-08,
      "loss": 1.103,
      "step": 220
    },
    {
      "epoch": 0.003502845088621981,
      "grad_norm": 5.367983818054199,
      "learning_rate": 6.973848069738481e-08,
      "loss": 1.1586,
      "step": 225
    },
    {
      "epoch": 0.0035806860905913582,
      "grad_norm": 19.27895164489746,
      "learning_rate": 7.129514321295142e-08,
      "loss": 1.0898,
      "step": 230
    },
    {
      "epoch": 0.0036585270925607355,
      "grad_norm": 3.7263176441192627,
      "learning_rate": 7.285180572851806e-08,
      "loss": 1.0541,
      "step": 235
    },
    {
      "epoch": 0.003736368094530113,
      "grad_norm": 21.48790740966797,
      "learning_rate": 7.440846824408468e-08,
      "loss": 1.2173,
      "step": 240
    },
    {
      "epoch": 0.00381420909649949,
      "grad_norm": 5.5661702156066895,
      "learning_rate": 7.596513075965131e-08,
      "loss": 1.1816,
      "step": 245
    },
    {
      "epoch": 0.0038920500984688674,
      "grad_norm": 13.601526260375977,
      "learning_rate": 7.752179327521793e-08,
      "loss": 1.0989,
      "step": 250
    },
    {
      "epoch": 0.003969891100438245,
      "grad_norm": 9.873005867004395,
      "learning_rate": 7.907845579078456e-08,
      "loss": 1.1859,
      "step": 255
    },
    {
      "epoch": 0.004047732102407622,
      "grad_norm": 4.8417277336120605,
      "learning_rate": 8.063511830635118e-08,
      "loss": 0.9859,
      "step": 260
    },
    {
      "epoch": 0.004125573104377,
      "grad_norm": 3.8291945457458496,
      "learning_rate": 8.21917808219178e-08,
      "loss": 0.9709,
      "step": 265
    },
    {
      "epoch": 0.004203414106346377,
      "grad_norm": 5.504295349121094,
      "learning_rate": 8.374844333748443e-08,
      "loss": 1.1271,
      "step": 270
    },
    {
      "epoch": 0.004281255108315754,
      "grad_norm": 10.665711402893066,
      "learning_rate": 8.530510585305104e-08,
      "loss": 1.1773,
      "step": 275
    },
    {
      "epoch": 0.0043590961102851315,
      "grad_norm": 8.259835243225098,
      "learning_rate": 8.686176836861768e-08,
      "loss": 1.1428,
      "step": 280
    },
    {
      "epoch": 0.004436937112254509,
      "grad_norm": 15.531925201416016,
      "learning_rate": 8.84184308841843e-08,
      "loss": 1.1423,
      "step": 285
    },
    {
      "epoch": 0.004514778114223886,
      "grad_norm": 17.920616149902344,
      "learning_rate": 8.997509339975093e-08,
      "loss": 1.3731,
      "step": 290
    },
    {
      "epoch": 0.004592619116193263,
      "grad_norm": 5.740132808685303,
      "learning_rate": 9.153175591531755e-08,
      "loss": 0.9373,
      "step": 295
    },
    {
      "epoch": 0.004670460118162641,
      "grad_norm": 6.698586463928223,
      "learning_rate": 9.308841843088418e-08,
      "loss": 1.0603,
      "step": 300
    },
    {
      "epoch": 0.004748301120132018,
      "grad_norm": 4.851785182952881,
      "learning_rate": 9.46450809464508e-08,
      "loss": 1.004,
      "step": 305
    },
    {
      "epoch": 0.004826142122101396,
      "grad_norm": 7.876951217651367,
      "learning_rate": 9.620174346201743e-08,
      "loss": 1.1607,
      "step": 310
    },
    {
      "epoch": 0.004903983124070773,
      "grad_norm": 9.093779563903809,
      "learning_rate": 9.775840597758405e-08,
      "loss": 1.1045,
      "step": 315
    },
    {
      "epoch": 0.004981824126040151,
      "grad_norm": 16.582103729248047,
      "learning_rate": 9.931506849315068e-08,
      "loss": 1.1154,
      "step": 320
    },
    {
      "epoch": 0.005059665128009528,
      "grad_norm": 13.140198707580566,
      "learning_rate": 1.008717310087173e-07,
      "loss": 1.17,
      "step": 325
    },
    {
      "epoch": 0.005137506129978905,
      "grad_norm": 3.4895646572113037,
      "learning_rate": 1.0242839352428394e-07,
      "loss": 0.9535,
      "step": 330
    },
    {
      "epoch": 0.0052153471319482825,
      "grad_norm": 6.645687103271484,
      "learning_rate": 1.0398505603985055e-07,
      "loss": 1.05,
      "step": 335
    },
    {
      "epoch": 0.00529318813391766,
      "grad_norm": 7.615957736968994,
      "learning_rate": 1.0554171855541719e-07,
      "loss": 0.9174,
      "step": 340
    },
    {
      "epoch": 0.005371029135887037,
      "grad_norm": 8.536812782287598,
      "learning_rate": 1.070983810709838e-07,
      "loss": 1.1338,
      "step": 345
    },
    {
      "epoch": 0.005448870137856414,
      "grad_norm": 4.573184967041016,
      "learning_rate": 1.0865504358655044e-07,
      "loss": 1.2018,
      "step": 350
    },
    {
      "epoch": 0.005526711139825792,
      "grad_norm": 11.614198684692383,
      "learning_rate": 1.1021170610211705e-07,
      "loss": 1.1853,
      "step": 355
    },
    {
      "epoch": 0.005604552141795169,
      "grad_norm": 12.930988311767578,
      "learning_rate": 1.1176836861768369e-07,
      "loss": 1.1772,
      "step": 360
    },
    {
      "epoch": 0.005682393143764546,
      "grad_norm": 5.334465980529785,
      "learning_rate": 1.133250311332503e-07,
      "loss": 1.1402,
      "step": 365
    },
    {
      "epoch": 0.0057602341457339236,
      "grad_norm": 19.55135726928711,
      "learning_rate": 1.1488169364881693e-07,
      "loss": 0.9569,
      "step": 370
    },
    {
      "epoch": 0.005838075147703301,
      "grad_norm": 14.209831237792969,
      "learning_rate": 1.1643835616438355e-07,
      "loss": 1.1239,
      "step": 375
    },
    {
      "epoch": 0.005915916149672679,
      "grad_norm": 5.5656352043151855,
      "learning_rate": 1.1799501867995018e-07,
      "loss": 1.1074,
      "step": 380
    },
    {
      "epoch": 0.005993757151642056,
      "grad_norm": 10.571775436401367,
      "learning_rate": 1.1955168119551682e-07,
      "loss": 1.1663,
      "step": 385
    },
    {
      "epoch": 0.006071598153611434,
      "grad_norm": 5.807967662811279,
      "learning_rate": 1.2110834371108342e-07,
      "loss": 1.1868,
      "step": 390
    },
    {
      "epoch": 0.006149439155580811,
      "grad_norm": 7.003355503082275,
      "learning_rate": 1.2266500622665007e-07,
      "loss": 1.0249,
      "step": 395
    },
    {
      "epoch": 0.006227280157550188,
      "grad_norm": 14.337294578552246,
      "learning_rate": 1.2422166874221667e-07,
      "loss": 1.0405,
      "step": 400
    },
    {
      "epoch": 0.006305121159519565,
      "grad_norm": 12.388212203979492,
      "learning_rate": 1.2577833125778332e-07,
      "loss": 1.1801,
      "step": 405
    },
    {
      "epoch": 0.006382962161488943,
      "grad_norm": 11.25795841217041,
      "learning_rate": 1.2733499377334994e-07,
      "loss": 1.1672,
      "step": 410
    },
    {
      "epoch": 0.00646080316345832,
      "grad_norm": 15.970906257629395,
      "learning_rate": 1.2889165628891654e-07,
      "loss": 1.0815,
      "step": 415
    },
    {
      "epoch": 0.006538644165427697,
      "grad_norm": 16.4951114654541,
      "learning_rate": 1.3044831880448317e-07,
      "loss": 1.039,
      "step": 420
    },
    {
      "epoch": 0.006616485167397075,
      "grad_norm": 16.199981689453125,
      "learning_rate": 1.3200498132004982e-07,
      "loss": 1.1636,
      "step": 425
    },
    {
      "epoch": 0.006694326169366452,
      "grad_norm": 7.787930965423584,
      "learning_rate": 1.3356164383561644e-07,
      "loss": 1.0949,
      "step": 430
    },
    {
      "epoch": 0.006772167171335829,
      "grad_norm": 4.226932525634766,
      "learning_rate": 1.3511830635118307e-07,
      "loss": 1.0409,
      "step": 435
    },
    {
      "epoch": 0.0068500081733052064,
      "grad_norm": 19.068387985229492,
      "learning_rate": 1.3667496886674967e-07,
      "loss": 0.9881,
      "step": 440
    },
    {
      "epoch": 0.006927849175274584,
      "grad_norm": 3.8829450607299805,
      "learning_rate": 1.3823163138231632e-07,
      "loss": 0.9989,
      "step": 445
    },
    {
      "epoch": 0.007005690177243962,
      "grad_norm": 5.948785305023193,
      "learning_rate": 1.3978829389788294e-07,
      "loss": 1.007,
      "step": 450
    },
    {
      "epoch": 0.007083531179213339,
      "grad_norm": 5.125,
      "learning_rate": 1.4134495641344957e-07,
      "loss": 0.8301,
      "step": 455
    },
    {
      "epoch": 0.0071613721811827164,
      "grad_norm": 12.499361038208008,
      "learning_rate": 1.4290161892901616e-07,
      "loss": 1.0923,
      "step": 460
    },
    {
      "epoch": 0.007239213183152094,
      "grad_norm": 6.266834735870361,
      "learning_rate": 1.4445828144458281e-07,
      "loss": 0.88,
      "step": 465
    },
    {
      "epoch": 0.007317054185121471,
      "grad_norm": 9.417441368103027,
      "learning_rate": 1.4601494396014944e-07,
      "loss": 0.9669,
      "step": 470
    },
    {
      "epoch": 0.007394895187090848,
      "grad_norm": 9.376644134521484,
      "learning_rate": 1.4757160647571606e-07,
      "loss": 1.0241,
      "step": 475
    },
    {
      "epoch": 0.007472736189060226,
      "grad_norm": 10.515301704406738,
      "learning_rate": 1.491282689912827e-07,
      "loss": 1.0468,
      "step": 480
    },
    {
      "epoch": 0.007550577191029603,
      "grad_norm": 8.439921379089355,
      "learning_rate": 1.506849315068493e-07,
      "loss": 0.9356,
      "step": 485
    },
    {
      "epoch": 0.00762841819299898,
      "grad_norm": 8.198512077331543,
      "learning_rate": 1.5224159402241594e-07,
      "loss": 1.1591,
      "step": 490
    },
    {
      "epoch": 0.0077062591949683575,
      "grad_norm": 6.289046287536621,
      "learning_rate": 1.5379825653798256e-07,
      "loss": 1.0259,
      "step": 495
    },
    {
      "epoch": 0.007784100196937735,
      "grad_norm": 18.078012466430664,
      "learning_rate": 1.5535491905354919e-07,
      "loss": 1.1059,
      "step": 500
    },
    {
      "epoch": 0.007861941198907112,
      "grad_norm": 8.508400917053223,
      "learning_rate": 1.569115815691158e-07,
      "loss": 1.001,
      "step": 505
    },
    {
      "epoch": 0.00793978220087649,
      "grad_norm": 6.552981853485107,
      "learning_rate": 1.5846824408468243e-07,
      "loss": 1.0819,
      "step": 510
    },
    {
      "epoch": 0.008017623202845867,
      "grad_norm": 5.941412925720215,
      "learning_rate": 1.6002490660024906e-07,
      "loss": 1.0143,
      "step": 515
    },
    {
      "epoch": 0.008095464204815245,
      "grad_norm": 10.764496803283691,
      "learning_rate": 1.6158156911581568e-07,
      "loss": 1.0228,
      "step": 520
    },
    {
      "epoch": 0.008173305206784621,
      "grad_norm": 5.186371326446533,
      "learning_rate": 1.6313823163138233e-07,
      "loss": 0.9794,
      "step": 525
    },
    {
      "epoch": 0.008251146208754,
      "grad_norm": 11.401899337768555,
      "learning_rate": 1.6469489414694893e-07,
      "loss": 1.0773,
      "step": 530
    },
    {
      "epoch": 0.008328987210723376,
      "grad_norm": 5.4313788414001465,
      "learning_rate": 1.6625155666251556e-07,
      "loss": 0.9984,
      "step": 535
    },
    {
      "epoch": 0.008406828212692754,
      "grad_norm": 7.18859338760376,
      "learning_rate": 1.6780821917808218e-07,
      "loss": 1.0894,
      "step": 540
    },
    {
      "epoch": 0.00848466921466213,
      "grad_norm": 5.814337253570557,
      "learning_rate": 1.6936488169364883e-07,
      "loss": 0.983,
      "step": 545
    },
    {
      "epoch": 0.008562510216631508,
      "grad_norm": 11.842198371887207,
      "learning_rate": 1.7092154420921543e-07,
      "loss": 0.994,
      "step": 550
    },
    {
      "epoch": 0.008640351218600887,
      "grad_norm": 10.12619400024414,
      "learning_rate": 1.7247820672478206e-07,
      "loss": 1.015,
      "step": 555
    },
    {
      "epoch": 0.008718192220570263,
      "grad_norm": 7.895757675170898,
      "learning_rate": 1.7403486924034868e-07,
      "loss": 1.1194,
      "step": 560
    },
    {
      "epoch": 0.008796033222539641,
      "grad_norm": 5.340054512023926,
      "learning_rate": 1.755915317559153e-07,
      "loss": 1.0283,
      "step": 565
    },
    {
      "epoch": 0.008873874224509018,
      "grad_norm": 13.950590133666992,
      "learning_rate": 1.7714819427148193e-07,
      "loss": 1.155,
      "step": 570
    },
    {
      "epoch": 0.008951715226478396,
      "grad_norm": 10.90434741973877,
      "learning_rate": 1.7870485678704855e-07,
      "loss": 1.0135,
      "step": 575
    },
    {
      "epoch": 0.009029556228447772,
      "grad_norm": 4.94070291519165,
      "learning_rate": 1.8026151930261518e-07,
      "loss": 0.9843,
      "step": 580
    },
    {
      "epoch": 0.00910739723041715,
      "grad_norm": 9.50981616973877,
      "learning_rate": 1.818181818181818e-07,
      "loss": 1.0777,
      "step": 585
    },
    {
      "epoch": 0.009185238232386527,
      "grad_norm": 9.218316078186035,
      "learning_rate": 1.8337484433374845e-07,
      "loss": 1.1192,
      "step": 590
    },
    {
      "epoch": 0.009263079234355905,
      "grad_norm": 17.782791137695312,
      "learning_rate": 1.8493150684931505e-07,
      "loss": 1.0513,
      "step": 595
    },
    {
      "epoch": 0.009340920236325281,
      "grad_norm": 5.774691581726074,
      "learning_rate": 1.8648816936488168e-07,
      "loss": 0.872,
      "step": 600
    },
    {
      "epoch": 0.00941876123829466,
      "grad_norm": 6.310098171234131,
      "learning_rate": 1.880448318804483e-07,
      "loss": 0.9395,
      "step": 605
    },
    {
      "epoch": 0.009496602240264036,
      "grad_norm": 6.68503999710083,
      "learning_rate": 1.8960149439601495e-07,
      "loss": 1.0394,
      "step": 610
    },
    {
      "epoch": 0.009574443242233414,
      "grad_norm": 6.972198486328125,
      "learning_rate": 1.9115815691158155e-07,
      "loss": 1.066,
      "step": 615
    },
    {
      "epoch": 0.009652284244202792,
      "grad_norm": 6.581061363220215,
      "learning_rate": 1.9271481942714817e-07,
      "loss": 0.9428,
      "step": 620
    },
    {
      "epoch": 0.009730125246172169,
      "grad_norm": 10.010781288146973,
      "learning_rate": 1.942714819427148e-07,
      "loss": 1.097,
      "step": 625
    },
    {
      "epoch": 0.009807966248141547,
      "grad_norm": 10.270834922790527,
      "learning_rate": 1.9582814445828145e-07,
      "loss": 1.0373,
      "step": 630
    },
    {
      "epoch": 0.009885807250110923,
      "grad_norm": 7.189127445220947,
      "learning_rate": 1.9738480697384807e-07,
      "loss": 1.2356,
      "step": 635
    },
    {
      "epoch": 0.009963648252080301,
      "grad_norm": 11.2526216506958,
      "learning_rate": 1.9894146948941467e-07,
      "loss": 0.8422,
      "step": 640
    },
    {
      "epoch": 0.010041489254049678,
      "grad_norm": 5.1716203689575195,
      "learning_rate": 2.004981320049813e-07,
      "loss": 1.0132,
      "step": 645
    },
    {
      "epoch": 0.010119330256019056,
      "grad_norm": 4.592648983001709,
      "learning_rate": 2.0205479452054795e-07,
      "loss": 1.0677,
      "step": 650
    },
    {
      "epoch": 0.010197171257988432,
      "grad_norm": 4.74710750579834,
      "learning_rate": 2.0361145703611457e-07,
      "loss": 0.9573,
      "step": 655
    },
    {
      "epoch": 0.01027501225995781,
      "grad_norm": 5.075165748596191,
      "learning_rate": 2.0516811955168117e-07,
      "loss": 0.919,
      "step": 660
    },
    {
      "epoch": 0.010352853261927187,
      "grad_norm": 5.705000400543213,
      "learning_rate": 2.067247820672478e-07,
      "loss": 0.9722,
      "step": 665
    },
    {
      "epoch": 0.010430694263896565,
      "grad_norm": 8.337606430053711,
      "learning_rate": 2.0828144458281445e-07,
      "loss": 1.0511,
      "step": 670
    },
    {
      "epoch": 0.010508535265865942,
      "grad_norm": 7.93868350982666,
      "learning_rate": 2.0983810709838107e-07,
      "loss": 1.0511,
      "step": 675
    },
    {
      "epoch": 0.01058637626783532,
      "grad_norm": 7.5352325439453125,
      "learning_rate": 2.113947696139477e-07,
      "loss": 1.0529,
      "step": 680
    },
    {
      "epoch": 0.010664217269804696,
      "grad_norm": 12.067502975463867,
      "learning_rate": 2.129514321295143e-07,
      "loss": 0.9818,
      "step": 685
    },
    {
      "epoch": 0.010742058271774074,
      "grad_norm": 4.793339729309082,
      "learning_rate": 2.1450809464508094e-07,
      "loss": 0.9757,
      "step": 690
    },
    {
      "epoch": 0.010819899273743452,
      "grad_norm": 5.648492336273193,
      "learning_rate": 2.1606475716064757e-07,
      "loss": 0.9838,
      "step": 695
    },
    {
      "epoch": 0.010897740275712829,
      "grad_norm": 14.50791072845459,
      "learning_rate": 2.176214196762142e-07,
      "loss": 1.025,
      "step": 700
    },
    {
      "epoch": 0.010975581277682207,
      "grad_norm": 6.976552486419678,
      "learning_rate": 2.191780821917808e-07,
      "loss": 1.0636,
      "step": 705
    },
    {
      "epoch": 0.011053422279651583,
      "grad_norm": 8.440703392028809,
      "learning_rate": 2.2073474470734744e-07,
      "loss": 0.8979,
      "step": 710
    },
    {
      "epoch": 0.011131263281620962,
      "grad_norm": 17.822824478149414,
      "learning_rate": 2.2229140722291407e-07,
      "loss": 1.126,
      "step": 715
    },
    {
      "epoch": 0.011209104283590338,
      "grad_norm": 3.6384825706481934,
      "learning_rate": 2.238480697384807e-07,
      "loss": 0.9756,
      "step": 720
    },
    {
      "epoch": 0.011286945285559716,
      "grad_norm": 9.758706092834473,
      "learning_rate": 2.2540473225404732e-07,
      "loss": 1.0776,
      "step": 725
    },
    {
      "epoch": 0.011364786287529093,
      "grad_norm": 6.821314334869385,
      "learning_rate": 2.2696139476961394e-07,
      "loss": 1.0809,
      "step": 730
    },
    {
      "epoch": 0.01144262728949847,
      "grad_norm": 5.796785831451416,
      "learning_rate": 2.2851805728518056e-07,
      "loss": 1.0388,
      "step": 735
    },
    {
      "epoch": 0.011520468291467847,
      "grad_norm": 14.487456321716309,
      "learning_rate": 2.300747198007472e-07,
      "loss": 0.9906,
      "step": 740
    },
    {
      "epoch": 0.011598309293437225,
      "grad_norm": 5.587100505828857,
      "learning_rate": 2.3163138231631381e-07,
      "loss": 0.955,
      "step": 745
    },
    {
      "epoch": 0.011676150295406602,
      "grad_norm": 5.029387474060059,
      "learning_rate": 2.3318804483188044e-07,
      "loss": 0.9509,
      "step": 750
    },
    {
      "epoch": 0.01175399129737598,
      "grad_norm": 16.782621383666992,
      "learning_rate": 2.3474470734744706e-07,
      "loss": 1.0278,
      "step": 755
    },
    {
      "epoch": 0.011831832299345358,
      "grad_norm": 8.211995124816895,
      "learning_rate": 2.363013698630137e-07,
      "loss": 1.1128,
      "step": 760
    },
    {
      "epoch": 0.011909673301314734,
      "grad_norm": 8.179312705993652,
      "learning_rate": 2.378580323785803e-07,
      "loss": 0.9175,
      "step": 765
    },
    {
      "epoch": 0.011987514303284113,
      "grad_norm": 3.8183233737945557,
      "learning_rate": 2.3941469489414696e-07,
      "loss": 1.0116,
      "step": 770
    },
    {
      "epoch": 0.012065355305253489,
      "grad_norm": 11.87375545501709,
      "learning_rate": 2.4097135740971356e-07,
      "loss": 0.975,
      "step": 775
    },
    {
      "epoch": 0.012143196307222867,
      "grad_norm": 20.000045776367188,
      "learning_rate": 2.425280199252802e-07,
      "loss": 1.0561,
      "step": 780
    },
    {
      "epoch": 0.012221037309192244,
      "grad_norm": 4.025638103485107,
      "learning_rate": 2.440846824408468e-07,
      "loss": 0.977,
      "step": 785
    },
    {
      "epoch": 0.012298878311161622,
      "grad_norm": 8.214958190917969,
      "learning_rate": 2.4564134495641346e-07,
      "loss": 0.9728,
      "step": 790
    },
    {
      "epoch": 0.012376719313130998,
      "grad_norm": 13.562061309814453,
      "learning_rate": 2.4719800747198006e-07,
      "loss": 0.8595,
      "step": 795
    },
    {
      "epoch": 0.012454560315100376,
      "grad_norm": 4.473455905914307,
      "learning_rate": 2.4875466998754666e-07,
      "loss": 0.8972,
      "step": 800
    },
    {
      "epoch": 0.012532401317069753,
      "grad_norm": 5.311202049255371,
      "learning_rate": 2.503113325031133e-07,
      "loss": 0.8578,
      "step": 805
    },
    {
      "epoch": 0.01261024231903913,
      "grad_norm": 11.063155174255371,
      "learning_rate": 2.5186799501867996e-07,
      "loss": 0.9938,
      "step": 810
    },
    {
      "epoch": 0.012688083321008507,
      "grad_norm": 7.260047435760498,
      "learning_rate": 2.5342465753424656e-07,
      "loss": 0.9745,
      "step": 815
    },
    {
      "epoch": 0.012765924322977885,
      "grad_norm": 5.3101067543029785,
      "learning_rate": 2.549813200498132e-07,
      "loss": 1.0001,
      "step": 820
    },
    {
      "epoch": 0.012843765324947264,
      "grad_norm": 4.430516719818115,
      "learning_rate": 2.5653798256537986e-07,
      "loss": 0.9283,
      "step": 825
    },
    {
      "epoch": 0.01292160632691664,
      "grad_norm": 4.081624507904053,
      "learning_rate": 2.580946450809464e-07,
      "loss": 0.9694,
      "step": 830
    },
    {
      "epoch": 0.012999447328886018,
      "grad_norm": 6.339404106140137,
      "learning_rate": 2.5965130759651306e-07,
      "loss": 1.0485,
      "step": 835
    },
    {
      "epoch": 0.013077288330855395,
      "grad_norm": 8.773398399353027,
      "learning_rate": 2.6120797011207965e-07,
      "loss": 0.8494,
      "step": 840
    },
    {
      "epoch": 0.013155129332824773,
      "grad_norm": 9.235841751098633,
      "learning_rate": 2.627646326276463e-07,
      "loss": 0.9223,
      "step": 845
    },
    {
      "epoch": 0.01323297033479415,
      "grad_norm": 5.350943565368652,
      "learning_rate": 2.6432129514321296e-07,
      "loss": 1.0852,
      "step": 850
    },
    {
      "epoch": 0.013310811336763527,
      "grad_norm": 5.6170268058776855,
      "learning_rate": 2.6587795765877955e-07,
      "loss": 0.9529,
      "step": 855
    },
    {
      "epoch": 0.013388652338732904,
      "grad_norm": 6.033858776092529,
      "learning_rate": 2.674346201743462e-07,
      "loss": 0.9264,
      "step": 860
    },
    {
      "epoch": 0.013466493340702282,
      "grad_norm": 10.408087730407715,
      "learning_rate": 2.6899128268991286e-07,
      "loss": 0.9435,
      "step": 865
    },
    {
      "epoch": 0.013544334342671658,
      "grad_norm": 3.902411460876465,
      "learning_rate": 2.7054794520547945e-07,
      "loss": 1.0717,
      "step": 870
    },
    {
      "epoch": 0.013622175344641036,
      "grad_norm": 6.315438270568848,
      "learning_rate": 2.7210460772104605e-07,
      "loss": 1.0222,
      "step": 875
    },
    {
      "epoch": 0.013700016346610413,
      "grad_norm": 9.6283540725708,
      "learning_rate": 2.7366127023661265e-07,
      "loss": 0.9607,
      "step": 880
    },
    {
      "epoch": 0.013777857348579791,
      "grad_norm": 8.017468452453613,
      "learning_rate": 2.752179327521793e-07,
      "loss": 0.908,
      "step": 885
    },
    {
      "epoch": 0.013855698350549167,
      "grad_norm": 4.06109094619751,
      "learning_rate": 2.7677459526774595e-07,
      "loss": 0.9779,
      "step": 890
    },
    {
      "epoch": 0.013933539352518546,
      "grad_norm": 4.540249347686768,
      "learning_rate": 2.7833125778331255e-07,
      "loss": 1.0685,
      "step": 895
    },
    {
      "epoch": 0.014011380354487924,
      "grad_norm": 5.971028804779053,
      "learning_rate": 2.798879202988792e-07,
      "loss": 1.0487,
      "step": 900
    },
    {
      "epoch": 0.0140892213564573,
      "grad_norm": 7.365455150604248,
      "learning_rate": 2.8144458281444585e-07,
      "loss": 1.0255,
      "step": 905
    },
    {
      "epoch": 0.014167062358426678,
      "grad_norm": 5.49646520614624,
      "learning_rate": 2.8300124533001245e-07,
      "loss": 1.0066,
      "step": 910
    },
    {
      "epoch": 0.014244903360396055,
      "grad_norm": 5.0211615562438965,
      "learning_rate": 2.845579078455791e-07,
      "loss": 0.9018,
      "step": 915
    },
    {
      "epoch": 0.014322744362365433,
      "grad_norm": 3.7670419216156006,
      "learning_rate": 2.8611457036114565e-07,
      "loss": 1.0793,
      "step": 920
    },
    {
      "epoch": 0.01440058536433481,
      "grad_norm": 10.098974227905273,
      "learning_rate": 2.876712328767123e-07,
      "loss": 1.0537,
      "step": 925
    },
    {
      "epoch": 0.014478426366304187,
      "grad_norm": 8.83332633972168,
      "learning_rate": 2.8922789539227895e-07,
      "loss": 0.9274,
      "step": 930
    },
    {
      "epoch": 0.014556267368273564,
      "grad_norm": 13.259550094604492,
      "learning_rate": 2.9078455790784555e-07,
      "loss": 0.9684,
      "step": 935
    },
    {
      "epoch": 0.014634108370242942,
      "grad_norm": 9.241827964782715,
      "learning_rate": 2.923412204234122e-07,
      "loss": 1.031,
      "step": 940
    },
    {
      "epoch": 0.014711949372212318,
      "grad_norm": 7.292890548706055,
      "learning_rate": 2.9389788293897885e-07,
      "loss": 0.9006,
      "step": 945
    },
    {
      "epoch": 0.014789790374181697,
      "grad_norm": 4.794684886932373,
      "learning_rate": 2.9545454545454545e-07,
      "loss": 1.0308,
      "step": 950
    },
    {
      "epoch": 0.014867631376151073,
      "grad_norm": 4.3201518058776855,
      "learning_rate": 2.970112079701121e-07,
      "loss": 0.9706,
      "step": 955
    },
    {
      "epoch": 0.014945472378120451,
      "grad_norm": 3.5388669967651367,
      "learning_rate": 2.985678704856787e-07,
      "loss": 0.9782,
      "step": 960
    },
    {
      "epoch": 0.01502331338008983,
      "grad_norm": 10.980652809143066,
      "learning_rate": 3.001245330012453e-07,
      "loss": 1.1015,
      "step": 965
    },
    {
      "epoch": 0.015101154382059206,
      "grad_norm": 7.639592170715332,
      "learning_rate": 3.0168119551681194e-07,
      "loss": 0.9046,
      "step": 970
    },
    {
      "epoch": 0.015178995384028584,
      "grad_norm": 5.50681734085083,
      "learning_rate": 3.0323785803237854e-07,
      "loss": 1.0192,
      "step": 975
    },
    {
      "epoch": 0.01525683638599796,
      "grad_norm": 4.924655437469482,
      "learning_rate": 3.047945205479452e-07,
      "loss": 1.0545,
      "step": 980
    },
    {
      "epoch": 0.015334677387967338,
      "grad_norm": 6.294414043426514,
      "learning_rate": 3.0635118306351184e-07,
      "loss": 0.9807,
      "step": 985
    },
    {
      "epoch": 0.015412518389936715,
      "grad_norm": 4.609034538269043,
      "learning_rate": 3.0790784557907844e-07,
      "loss": 1.0205,
      "step": 990
    },
    {
      "epoch": 0.015490359391906093,
      "grad_norm": 3.4544599056243896,
      "learning_rate": 3.094645080946451e-07,
      "loss": 0.9761,
      "step": 995
    },
    {
      "epoch": 0.01556820039387547,
      "grad_norm": 5.186591148376465,
      "learning_rate": 3.110211706102117e-07,
      "loss": 0.9933,
      "step": 1000
    },
    {
      "epoch": 0.015646041395844846,
      "grad_norm": 4.516424179077148,
      "learning_rate": 3.125778331257783e-07,
      "loss": 0.9057,
      "step": 1005
    },
    {
      "epoch": 0.015723882397814224,
      "grad_norm": 4.458924293518066,
      "learning_rate": 3.1413449564134494e-07,
      "loss": 0.9526,
      "step": 1010
    },
    {
      "epoch": 0.015801723399783602,
      "grad_norm": 5.840490341186523,
      "learning_rate": 3.1569115815691154e-07,
      "loss": 1.0032,
      "step": 1015
    },
    {
      "epoch": 0.01587956440175298,
      "grad_norm": 13.803277015686035,
      "learning_rate": 3.172478206724782e-07,
      "loss": 1.0162,
      "step": 1020
    },
    {
      "epoch": 0.015957405403722355,
      "grad_norm": 3.742831230163574,
      "learning_rate": 3.1880448318804484e-07,
      "loss": 0.9714,
      "step": 1025
    },
    {
      "epoch": 0.016035246405691733,
      "grad_norm": 5.748800277709961,
      "learning_rate": 3.2036114570361144e-07,
      "loss": 0.9545,
      "step": 1030
    },
    {
      "epoch": 0.01611308740766111,
      "grad_norm": 4.5021491050720215,
      "learning_rate": 3.219178082191781e-07,
      "loss": 0.9503,
      "step": 1035
    },
    {
      "epoch": 0.01619092840963049,
      "grad_norm": 6.095613956451416,
      "learning_rate": 3.234744707347447e-07,
      "loss": 0.991,
      "step": 1040
    },
    {
      "epoch": 0.016268769411599868,
      "grad_norm": 4.993571758270264,
      "learning_rate": 3.2503113325031134e-07,
      "loss": 0.9221,
      "step": 1045
    },
    {
      "epoch": 0.016346610413569242,
      "grad_norm": 5.949316501617432,
      "learning_rate": 3.2658779576587794e-07,
      "loss": 0.9897,
      "step": 1050
    },
    {
      "epoch": 0.01642445141553862,
      "grad_norm": 5.225283622741699,
      "learning_rate": 3.2814445828144453e-07,
      "loss": 0.9719,
      "step": 1055
    },
    {
      "epoch": 0.016502292417508,
      "grad_norm": 15.378800392150879,
      "learning_rate": 3.297011207970112e-07,
      "loss": 0.8633,
      "step": 1060
    },
    {
      "epoch": 0.016580133419477377,
      "grad_norm": 4.347599506378174,
      "learning_rate": 3.312577833125778e-07,
      "loss": 0.9054,
      "step": 1065
    },
    {
      "epoch": 0.01665797442144675,
      "grad_norm": 5.208911895751953,
      "learning_rate": 3.3281444582814443e-07,
      "loss": 0.899,
      "step": 1070
    },
    {
      "epoch": 0.01673581542341613,
      "grad_norm": 6.316863059997559,
      "learning_rate": 3.343711083437111e-07,
      "loss": 1.0165,
      "step": 1075
    },
    {
      "epoch": 0.016813656425385508,
      "grad_norm": 5.477814197540283,
      "learning_rate": 3.359277708592777e-07,
      "loss": 0.96,
      "step": 1080
    },
    {
      "epoch": 0.016891497427354886,
      "grad_norm": 4.848371505737305,
      "learning_rate": 3.3748443337484433e-07,
      "loss": 0.9703,
      "step": 1085
    },
    {
      "epoch": 0.01696933842932426,
      "grad_norm": 9.025872230529785,
      "learning_rate": 3.39041095890411e-07,
      "loss": 0.9239,
      "step": 1090
    },
    {
      "epoch": 0.01704717943129364,
      "grad_norm": 3.3916220664978027,
      "learning_rate": 3.4059775840597753e-07,
      "loss": 1.0136,
      "step": 1095
    },
    {
      "epoch": 0.017125020433263017,
      "grad_norm": 9.25607967376709,
      "learning_rate": 3.421544209215442e-07,
      "loss": 0.9626,
      "step": 1100
    },
    {
      "epoch": 0.017202861435232395,
      "grad_norm": 7.245452880859375,
      "learning_rate": 3.437110834371108e-07,
      "loss": 1.0026,
      "step": 1105
    },
    {
      "epoch": 0.017280702437201773,
      "grad_norm": 3.3463306427001953,
      "learning_rate": 3.4526774595267743e-07,
      "loss": 0.9243,
      "step": 1110
    },
    {
      "epoch": 0.017358543439171148,
      "grad_norm": 5.334697723388672,
      "learning_rate": 3.468244084682441e-07,
      "loss": 1.1001,
      "step": 1115
    },
    {
      "epoch": 0.017436384441140526,
      "grad_norm": 4.7469305992126465,
      "learning_rate": 3.483810709838107e-07,
      "loss": 1.0421,
      "step": 1120
    },
    {
      "epoch": 0.017514225443109904,
      "grad_norm": 4.398116111755371,
      "learning_rate": 3.4993773349937733e-07,
      "loss": 0.9502,
      "step": 1125
    },
    {
      "epoch": 0.017592066445079282,
      "grad_norm": 3.972031831741333,
      "learning_rate": 3.51494396014944e-07,
      "loss": 0.8974,
      "step": 1130
    },
    {
      "epoch": 0.017669907447048657,
      "grad_norm": 5.13526725769043,
      "learning_rate": 3.530510585305106e-07,
      "loss": 0.9668,
      "step": 1135
    },
    {
      "epoch": 0.017747748449018035,
      "grad_norm": 3.752171754837036,
      "learning_rate": 3.546077210460772e-07,
      "loss": 1.0589,
      "step": 1140
    },
    {
      "epoch": 0.017825589450987413,
      "grad_norm": 6.005197048187256,
      "learning_rate": 3.561643835616438e-07,
      "loss": 0.9786,
      "step": 1145
    },
    {
      "epoch": 0.01790343045295679,
      "grad_norm": 5.12382173538208,
      "learning_rate": 3.5772104607721043e-07,
      "loss": 0.936,
      "step": 1150
    },
    {
      "epoch": 0.017981271454926166,
      "grad_norm": 7.456275939941406,
      "learning_rate": 3.592777085927771e-07,
      "loss": 0.9261,
      "step": 1155
    },
    {
      "epoch": 0.018059112456895544,
      "grad_norm": 3.7287797927856445,
      "learning_rate": 3.608343711083437e-07,
      "loss": 0.9291,
      "step": 1160
    },
    {
      "epoch": 0.018136953458864923,
      "grad_norm": 3.916651725769043,
      "learning_rate": 3.6239103362391033e-07,
      "loss": 0.9193,
      "step": 1165
    },
    {
      "epoch": 0.0182147944608343,
      "grad_norm": 4.2813720703125,
      "learning_rate": 3.63947696139477e-07,
      "loss": 1.0079,
      "step": 1170
    },
    {
      "epoch": 0.01829263546280368,
      "grad_norm": 8.352608680725098,
      "learning_rate": 3.655043586550436e-07,
      "loss": 0.9901,
      "step": 1175
    },
    {
      "epoch": 0.018370476464773054,
      "grad_norm": 5.297429084777832,
      "learning_rate": 3.6706102117061023e-07,
      "loss": 0.9049,
      "step": 1180
    },
    {
      "epoch": 0.01844831746674243,
      "grad_norm": 4.064713478088379,
      "learning_rate": 3.6861768368617677e-07,
      "loss": 1.1307,
      "step": 1185
    },
    {
      "epoch": 0.01852615846871181,
      "grad_norm": 6.08450174331665,
      "learning_rate": 3.701743462017434e-07,
      "loss": 1.0267,
      "step": 1190
    },
    {
      "epoch": 0.018603999470681188,
      "grad_norm": 4.351869106292725,
      "learning_rate": 3.717310087173101e-07,
      "loss": 1.0315,
      "step": 1195
    },
    {
      "epoch": 0.018681840472650563,
      "grad_norm": 7.120603084564209,
      "learning_rate": 3.7328767123287667e-07,
      "loss": 1.0092,
      "step": 1200
    },
    {
      "epoch": 0.01875968147461994,
      "grad_norm": 4.8134660720825195,
      "learning_rate": 3.748443337484433e-07,
      "loss": 0.9491,
      "step": 1205
    },
    {
      "epoch": 0.01883752247658932,
      "grad_norm": 5.852837085723877,
      "learning_rate": 3.7640099626401e-07,
      "loss": 1.0029,
      "step": 1210
    },
    {
      "epoch": 0.018915363478558697,
      "grad_norm": 5.291375160217285,
      "learning_rate": 3.7795765877957657e-07,
      "loss": 0.8198,
      "step": 1215
    },
    {
      "epoch": 0.018993204480528072,
      "grad_norm": 3.2667717933654785,
      "learning_rate": 3.795143212951432e-07,
      "loss": 0.9155,
      "step": 1220
    },
    {
      "epoch": 0.01907104548249745,
      "grad_norm": 4.952467918395996,
      "learning_rate": 3.810709838107098e-07,
      "loss": 0.9144,
      "step": 1225
    },
    {
      "epoch": 0.019148886484466828,
      "grad_norm": 4.495504379272461,
      "learning_rate": 3.826276463262764e-07,
      "loss": 0.9236,
      "step": 1230
    },
    {
      "epoch": 0.019226727486436206,
      "grad_norm": 5.554149627685547,
      "learning_rate": 3.8418430884184307e-07,
      "loss": 0.7856,
      "step": 1235
    },
    {
      "epoch": 0.019304568488405584,
      "grad_norm": 6.092937469482422,
      "learning_rate": 3.8574097135740967e-07,
      "loss": 0.8681,
      "step": 1240
    },
    {
      "epoch": 0.01938240949037496,
      "grad_norm": 3.9643170833587646,
      "learning_rate": 3.872976338729763e-07,
      "loss": 0.836,
      "step": 1245
    },
    {
      "epoch": 0.019460250492344337,
      "grad_norm": 3.9617724418640137,
      "learning_rate": 3.8885429638854297e-07,
      "loss": 1.015,
      "step": 1250
    },
    {
      "epoch": 0.019538091494313715,
      "grad_norm": 8.572834014892578,
      "learning_rate": 3.9041095890410957e-07,
      "loss": 0.9308,
      "step": 1255
    },
    {
      "epoch": 0.019615932496283094,
      "grad_norm": 6.380552291870117,
      "learning_rate": 3.919676214196762e-07,
      "loss": 0.9056,
      "step": 1260
    },
    {
      "epoch": 0.01969377349825247,
      "grad_norm": 5.703736782073975,
      "learning_rate": 3.935242839352428e-07,
      "loss": 0.9864,
      "step": 1265
    },
    {
      "epoch": 0.019771614500221846,
      "grad_norm": 4.2661824226379395,
      "learning_rate": 3.9508094645080947e-07,
      "loss": 0.8801,
      "step": 1270
    },
    {
      "epoch": 0.019849455502191225,
      "grad_norm": 3.4654171466827393,
      "learning_rate": 3.9663760896637607e-07,
      "loss": 0.8954,
      "step": 1275
    },
    {
      "epoch": 0.019927296504160603,
      "grad_norm": 5.910457611083984,
      "learning_rate": 3.9819427148194266e-07,
      "loss": 0.8456,
      "step": 1280
    },
    {
      "epoch": 0.020005137506129977,
      "grad_norm": 6.345880031585693,
      "learning_rate": 3.997509339975093e-07,
      "loss": 0.9825,
      "step": 1285
    },
    {
      "epoch": 0.020082978508099356,
      "grad_norm": 11.178544044494629,
      "learning_rate": 4.0130759651307597e-07,
      "loss": 0.9716,
      "step": 1290
    },
    {
      "epoch": 0.020160819510068734,
      "grad_norm": 3.9438936710357666,
      "learning_rate": 4.0286425902864256e-07,
      "loss": 0.9629,
      "step": 1295
    },
    {
      "epoch": 0.020238660512038112,
      "grad_norm": 7.4510273933410645,
      "learning_rate": 4.044209215442092e-07,
      "loss": 0.9815,
      "step": 1300
    },
    {
      "epoch": 0.020316501514007487,
      "grad_norm": 6.15594482421875,
      "learning_rate": 4.059775840597758e-07,
      "loss": 0.9805,
      "step": 1305
    },
    {
      "epoch": 0.020394342515976865,
      "grad_norm": 5.105663776397705,
      "learning_rate": 4.0753424657534246e-07,
      "loss": 1.0047,
      "step": 1310
    },
    {
      "epoch": 0.020472183517946243,
      "grad_norm": 4.2579779624938965,
      "learning_rate": 4.090909090909091e-07,
      "loss": 0.797,
      "step": 1315
    },
    {
      "epoch": 0.02055002451991562,
      "grad_norm": 3.6263747215270996,
      "learning_rate": 4.1064757160647566e-07,
      "loss": 0.9526,
      "step": 1320
    },
    {
      "epoch": 0.020627865521885,
      "grad_norm": 4.003891944885254,
      "learning_rate": 4.122042341220423e-07,
      "loss": 0.862,
      "step": 1325
    },
    {
      "epoch": 0.020705706523854374,
      "grad_norm": 4.833682060241699,
      "learning_rate": 4.137608966376089e-07,
      "loss": 1.0438,
      "step": 1330
    },
    {
      "epoch": 0.020783547525823752,
      "grad_norm": 8.875425338745117,
      "learning_rate": 4.1531755915317556e-07,
      "loss": 1.0013,
      "step": 1335
    },
    {
      "epoch": 0.02086138852779313,
      "grad_norm": 5.356649398803711,
      "learning_rate": 4.168742216687422e-07,
      "loss": 0.8956,
      "step": 1340
    },
    {
      "epoch": 0.02093922952976251,
      "grad_norm": 5.640366554260254,
      "learning_rate": 4.184308841843088e-07,
      "loss": 0.8346,
      "step": 1345
    },
    {
      "epoch": 0.021017070531731883,
      "grad_norm": 3.717663288116455,
      "learning_rate": 4.1998754669987546e-07,
      "loss": 0.8512,
      "step": 1350
    },
    {
      "epoch": 0.02109491153370126,
      "grad_norm": 3.557542324066162,
      "learning_rate": 4.215442092154421e-07,
      "loss": 0.9313,
      "step": 1355
    },
    {
      "epoch": 0.02117275253567064,
      "grad_norm": 5.178566932678223,
      "learning_rate": 4.231008717310087e-07,
      "loss": 0.9086,
      "step": 1360
    },
    {
      "epoch": 0.021250593537640017,
      "grad_norm": 5.773383140563965,
      "learning_rate": 4.246575342465753e-07,
      "loss": 0.9678,
      "step": 1365
    },
    {
      "epoch": 0.021328434539609392,
      "grad_norm": 4.725634574890137,
      "learning_rate": 4.262141967621419e-07,
      "loss": 0.9356,
      "step": 1370
    },
    {
      "epoch": 0.02140627554157877,
      "grad_norm": 3.0198757648468018,
      "learning_rate": 4.2777085927770856e-07,
      "loss": 0.9342,
      "step": 1375
    },
    {
      "epoch": 0.02148411654354815,
      "grad_norm": 5.704006195068359,
      "learning_rate": 4.293275217932752e-07,
      "loss": 1.0469,
      "step": 1380
    },
    {
      "epoch": 0.021561957545517527,
      "grad_norm": 4.559571743011475,
      "learning_rate": 4.308841843088418e-07,
      "loss": 0.845,
      "step": 1385
    },
    {
      "epoch": 0.021639798547486905,
      "grad_norm": 9.018213272094727,
      "learning_rate": 4.3244084682440846e-07,
      "loss": 0.9221,
      "step": 1390
    },
    {
      "epoch": 0.02171763954945628,
      "grad_norm": 6.414641380310059,
      "learning_rate": 4.339975093399751e-07,
      "loss": 1.0877,
      "step": 1395
    },
    {
      "epoch": 0.021795480551425658,
      "grad_norm": 4.217600345611572,
      "learning_rate": 4.355541718555417e-07,
      "loss": 0.8761,
      "step": 1400
    },
    {
      "epoch": 0.021873321553395036,
      "grad_norm": 5.274855136871338,
      "learning_rate": 4.3711083437110836e-07,
      "loss": 0.9046,
      "step": 1405
    },
    {
      "epoch": 0.021951162555364414,
      "grad_norm": 11.607494354248047,
      "learning_rate": 4.386674968866749e-07,
      "loss": 0.981,
      "step": 1410
    },
    {
      "epoch": 0.02202900355733379,
      "grad_norm": 5.442785263061523,
      "learning_rate": 4.4022415940224155e-07,
      "loss": 0.9357,
      "step": 1415
    },
    {
      "epoch": 0.022106844559303167,
      "grad_norm": 4.934208869934082,
      "learning_rate": 4.417808219178082e-07,
      "loss": 0.9124,
      "step": 1420
    },
    {
      "epoch": 0.022184685561272545,
      "grad_norm": 5.2812933921813965,
      "learning_rate": 4.433374844333748e-07,
      "loss": 0.9655,
      "step": 1425
    },
    {
      "epoch": 0.022262526563241923,
      "grad_norm": 3.1578216552734375,
      "learning_rate": 4.4489414694894145e-07,
      "loss": 0.9452,
      "step": 1430
    },
    {
      "epoch": 0.022340367565211298,
      "grad_norm": 10.148691177368164,
      "learning_rate": 4.464508094645081e-07,
      "loss": 0.9015,
      "step": 1435
    },
    {
      "epoch": 0.022418208567180676,
      "grad_norm": 3.150479793548584,
      "learning_rate": 4.480074719800747e-07,
      "loss": 0.8744,
      "step": 1440
    },
    {
      "epoch": 0.022496049569150054,
      "grad_norm": 5.963056564331055,
      "learning_rate": 4.4956413449564135e-07,
      "loss": 1.0312,
      "step": 1445
    },
    {
      "epoch": 0.022573890571119432,
      "grad_norm": 5.098721981048584,
      "learning_rate": 4.5112079701120795e-07,
      "loss": 0.8031,
      "step": 1450
    },
    {
      "epoch": 0.02265173157308881,
      "grad_norm": 5.2625017166137695,
      "learning_rate": 4.5267745952677455e-07,
      "loss": 0.8132,
      "step": 1455
    },
    {
      "epoch": 0.022729572575058185,
      "grad_norm": 8.537793159484863,
      "learning_rate": 4.542341220423412e-07,
      "loss": 0.8296,
      "step": 1460
    },
    {
      "epoch": 0.022807413577027563,
      "grad_norm": 6.819812774658203,
      "learning_rate": 4.557907845579078e-07,
      "loss": 0.9476,
      "step": 1465
    },
    {
      "epoch": 0.02288525457899694,
      "grad_norm": 4.941056251525879,
      "learning_rate": 4.5734744707347445e-07,
      "loss": 0.8785,
      "step": 1470
    },
    {
      "epoch": 0.02296309558096632,
      "grad_norm": 5.378219127655029,
      "learning_rate": 4.589041095890411e-07,
      "loss": 0.8987,
      "step": 1475
    },
    {
      "epoch": 0.023040936582935694,
      "grad_norm": 4.793314456939697,
      "learning_rate": 4.604607721046077e-07,
      "loss": 0.8309,
      "step": 1480
    },
    {
      "epoch": 0.023118777584905072,
      "grad_norm": 7.7251434326171875,
      "learning_rate": 4.6201743462017435e-07,
      "loss": 0.959,
      "step": 1485
    },
    {
      "epoch": 0.02319661858687445,
      "grad_norm": 3.7208149433135986,
      "learning_rate": 4.6357409713574095e-07,
      "loss": 0.9126,
      "step": 1490
    },
    {
      "epoch": 0.02327445958884383,
      "grad_norm": 4.322316646575928,
      "learning_rate": 4.651307596513076e-07,
      "loss": 0.9567,
      "step": 1495
    },
    {
      "epoch": 0.023352300590813203,
      "grad_norm": 5.451142311096191,
      "learning_rate": 4.666874221668742e-07,
      "loss": 0.9943,
      "step": 1500
    },
    {
      "epoch": 0.02343014159278258,
      "grad_norm": 6.478999614715576,
      "learning_rate": 4.682440846824408e-07,
      "loss": 0.8577,
      "step": 1505
    },
    {
      "epoch": 0.02350798259475196,
      "grad_norm": 5.626023292541504,
      "learning_rate": 4.6980074719800745e-07,
      "loss": 0.9176,
      "step": 1510
    },
    {
      "epoch": 0.023585823596721338,
      "grad_norm": 9.153360366821289,
      "learning_rate": 4.713574097135741e-07,
      "loss": 1.0269,
      "step": 1515
    },
    {
      "epoch": 0.023663664598690716,
      "grad_norm": 11.129598617553711,
      "learning_rate": 4.729140722291407e-07,
      "loss": 0.9428,
      "step": 1520
    },
    {
      "epoch": 0.02374150560066009,
      "grad_norm": 5.8177313804626465,
      "learning_rate": 4.7447073474470735e-07,
      "loss": 1.0107,
      "step": 1525
    },
    {
      "epoch": 0.02381934660262947,
      "grad_norm": 6.537820816040039,
      "learning_rate": 4.7602739726027394e-07,
      "loss": 0.8104,
      "step": 1530
    },
    {
      "epoch": 0.023897187604598847,
      "grad_norm": 4.420594692230225,
      "learning_rate": 4.775840597758406e-07,
      "loss": 0.8484,
      "step": 1535
    },
    {
      "epoch": 0.023975028606568225,
      "grad_norm": 6.306564807891846,
      "learning_rate": 4.791407222914072e-07,
      "loss": 0.796,
      "step": 1540
    },
    {
      "epoch": 0.0240528696085376,
      "grad_norm": 11.836288452148438,
      "learning_rate": 4.806973848069738e-07,
      "loss": 0.8949,
      "step": 1545
    },
    {
      "epoch": 0.024130710610506978,
      "grad_norm": 4.565202713012695,
      "learning_rate": 4.822540473225404e-07,
      "loss": 0.8881,
      "step": 1550
    },
    {
      "epoch": 0.024208551612476356,
      "grad_norm": 4.610184669494629,
      "learning_rate": 4.83810709838107e-07,
      "loss": 1.0267,
      "step": 1555
    },
    {
      "epoch": 0.024286392614445734,
      "grad_norm": 4.136282444000244,
      "learning_rate": 4.853673723536737e-07,
      "loss": 0.9593,
      "step": 1560
    },
    {
      "epoch": 0.02436423361641511,
      "grad_norm": 4.203325271606445,
      "learning_rate": 4.869240348692403e-07,
      "loss": 0.9377,
      "step": 1565
    },
    {
      "epoch": 0.024442074618384487,
      "grad_norm": 8.468722343444824,
      "learning_rate": 4.88480697384807e-07,
      "loss": 0.9949,
      "step": 1570
    },
    {
      "epoch": 0.024519915620353865,
      "grad_norm": 7.116949558258057,
      "learning_rate": 4.900373599003736e-07,
      "loss": 0.92,
      "step": 1575
    },
    {
      "epoch": 0.024597756622323243,
      "grad_norm": 4.857876777648926,
      "learning_rate": 4.915940224159402e-07,
      "loss": 0.8945,
      "step": 1580
    },
    {
      "epoch": 0.02467559762429262,
      "grad_norm": 7.421228408813477,
      "learning_rate": 4.931506849315068e-07,
      "loss": 0.8807,
      "step": 1585
    },
    {
      "epoch": 0.024753438626261996,
      "grad_norm": 7.203330993652344,
      "learning_rate": 4.947073474470734e-07,
      "loss": 0.8998,
      "step": 1590
    },
    {
      "epoch": 0.024831279628231374,
      "grad_norm": 12.598939895629883,
      "learning_rate": 4.9626400996264e-07,
      "loss": 0.8843,
      "step": 1595
    },
    {
      "epoch": 0.024909120630200753,
      "grad_norm": 6.573790073394775,
      "learning_rate": 4.978206724782067e-07,
      "loss": 0.8954,
      "step": 1600
    },
    {
      "epoch": 0.02498696163217013,
      "grad_norm": 5.063882350921631,
      "learning_rate": 4.993773349937733e-07,
      "loss": 0.8539,
      "step": 1605
    },
    {
      "epoch": 0.025064802634139505,
      "grad_norm": 5.859914779663086,
      "learning_rate": 5.0093399750934e-07,
      "loss": 0.9102,
      "step": 1610
    },
    {
      "epoch": 0.025142643636108884,
      "grad_norm": 4.542943954467773,
      "learning_rate": 5.024906600249066e-07,
      "loss": 0.876,
      "step": 1615
    },
    {
      "epoch": 0.02522048463807826,
      "grad_norm": 6.943472862243652,
      "learning_rate": 5.040473225404732e-07,
      "loss": 0.8886,
      "step": 1620
    },
    {
      "epoch": 0.02529832564004764,
      "grad_norm": 5.794211387634277,
      "learning_rate": 5.056039850560398e-07,
      "loss": 0.9136,
      "step": 1625
    },
    {
      "epoch": 0.025376166642017015,
      "grad_norm": 3.58612322807312,
      "learning_rate": 5.071606475716065e-07,
      "loss": 0.8483,
      "step": 1630
    },
    {
      "epoch": 0.025454007643986393,
      "grad_norm": 8.513461112976074,
      "learning_rate": 5.087173100871731e-07,
      "loss": 0.9484,
      "step": 1635
    },
    {
      "epoch": 0.02553184864595577,
      "grad_norm": 3.152209997177124,
      "learning_rate": 5.102739726027398e-07,
      "loss": 0.8564,
      "step": 1640
    },
    {
      "epoch": 0.02560968964792515,
      "grad_norm": 11.711703300476074,
      "learning_rate": 5.118306351183063e-07,
      "loss": 0.7842,
      "step": 1645
    },
    {
      "epoch": 0.025687530649894527,
      "grad_norm": 4.101468086242676,
      "learning_rate": 5.13387297633873e-07,
      "loss": 0.8646,
      "step": 1650
    },
    {
      "epoch": 0.025765371651863902,
      "grad_norm": 3.844512462615967,
      "learning_rate": 5.149439601494395e-07,
      "loss": 0.8094,
      "step": 1655
    },
    {
      "epoch": 0.02584321265383328,
      "grad_norm": 3.546029567718506,
      "learning_rate": 5.165006226650062e-07,
      "loss": 0.87,
      "step": 1660
    },
    {
      "epoch": 0.025921053655802658,
      "grad_norm": 3.3729195594787598,
      "learning_rate": 5.180572851805728e-07,
      "loss": 0.9467,
      "step": 1665
    },
    {
      "epoch": 0.025998894657772036,
      "grad_norm": 3.984131336212158,
      "learning_rate": 5.196139476961394e-07,
      "loss": 0.8571,
      "step": 1670
    },
    {
      "epoch": 0.02607673565974141,
      "grad_norm": 5.9442291259765625,
      "learning_rate": 5.21170610211706e-07,
      "loss": 0.892,
      "step": 1675
    },
    {
      "epoch": 0.02615457666171079,
      "grad_norm": 6.404414653778076,
      "learning_rate": 5.227272727272727e-07,
      "loss": 0.8978,
      "step": 1680
    },
    {
      "epoch": 0.026232417663680167,
      "grad_norm": 8.53201961517334,
      "learning_rate": 5.242839352428393e-07,
      "loss": 0.8403,
      "step": 1685
    },
    {
      "epoch": 0.026310258665649545,
      "grad_norm": 7.944653511047363,
      "learning_rate": 5.25840597758406e-07,
      "loss": 0.7752,
      "step": 1690
    },
    {
      "epoch": 0.02638809966761892,
      "grad_norm": 4.13915491104126,
      "learning_rate": 5.273972602739725e-07,
      "loss": 1.053,
      "step": 1695
    },
    {
      "epoch": 0.0264659406695883,
      "grad_norm": 9.199925422668457,
      "learning_rate": 5.289539227895392e-07,
      "loss": 1.0808,
      "step": 1700
    },
    {
      "epoch": 0.026543781671557676,
      "grad_norm": 4.507978439331055,
      "learning_rate": 5.305105853051058e-07,
      "loss": 0.9654,
      "step": 1705
    },
    {
      "epoch": 0.026621622673527055,
      "grad_norm": 5.004615783691406,
      "learning_rate": 5.320672478206725e-07,
      "loss": 0.9003,
      "step": 1710
    },
    {
      "epoch": 0.02669946367549643,
      "grad_norm": 9.572540283203125,
      "learning_rate": 5.336239103362391e-07,
      "loss": 1.0496,
      "step": 1715
    },
    {
      "epoch": 0.026777304677465807,
      "grad_norm": 6.494607925415039,
      "learning_rate": 5.351805728518058e-07,
      "loss": 0.9283,
      "step": 1720
    },
    {
      "epoch": 0.026855145679435186,
      "grad_norm": 6.419877529144287,
      "learning_rate": 5.367372353673723e-07,
      "loss": 0.9118,
      "step": 1725
    },
    {
      "epoch": 0.026932986681404564,
      "grad_norm": 8.065162658691406,
      "learning_rate": 5.38293897882939e-07,
      "loss": 0.9376,
      "step": 1730
    },
    {
      "epoch": 0.027010827683373942,
      "grad_norm": 11.1658935546875,
      "learning_rate": 5.398505603985056e-07,
      "loss": 0.9734,
      "step": 1735
    },
    {
      "epoch": 0.027088668685343317,
      "grad_norm": 8.80482006072998,
      "learning_rate": 5.414072229140723e-07,
      "loss": 0.9357,
      "step": 1740
    },
    {
      "epoch": 0.027166509687312695,
      "grad_norm": 9.545907974243164,
      "learning_rate": 5.429638854296388e-07,
      "loss": 1.0168,
      "step": 1745
    },
    {
      "epoch": 0.027244350689282073,
      "grad_norm": 3.2502315044403076,
      "learning_rate": 5.445205479452054e-07,
      "loss": 0.8814,
      "step": 1750
    },
    {
      "epoch": 0.02732219169125145,
      "grad_norm": 7.160440921783447,
      "learning_rate": 5.46077210460772e-07,
      "loss": 0.8058,
      "step": 1755
    },
    {
      "epoch": 0.027400032693220826,
      "grad_norm": 4.625821113586426,
      "learning_rate": 5.476338729763387e-07,
      "loss": 0.8834,
      "step": 1760
    },
    {
      "epoch": 0.027477873695190204,
      "grad_norm": 6.714595317840576,
      "learning_rate": 5.491905354919053e-07,
      "loss": 0.9709,
      "step": 1765
    },
    {
      "epoch": 0.027555714697159582,
      "grad_norm": 5.669415473937988,
      "learning_rate": 5.50747198007472e-07,
      "loss": 0.9646,
      "step": 1770
    },
    {
      "epoch": 0.02763355569912896,
      "grad_norm": 6.046622276306152,
      "learning_rate": 5.523038605230385e-07,
      "loss": 0.8469,
      "step": 1775
    },
    {
      "epoch": 0.027711396701098335,
      "grad_norm": 16.526947021484375,
      "learning_rate": 5.538605230386052e-07,
      "loss": 0.877,
      "step": 1780
    },
    {
      "epoch": 0.027789237703067713,
      "grad_norm": 4.415500164031982,
      "learning_rate": 5.554171855541718e-07,
      "loss": 0.8356,
      "step": 1785
    },
    {
      "epoch": 0.02786707870503709,
      "grad_norm": 4.823260307312012,
      "learning_rate": 5.569738480697385e-07,
      "loss": 0.9628,
      "step": 1790
    },
    {
      "epoch": 0.02794491970700647,
      "grad_norm": 8.501585006713867,
      "learning_rate": 5.585305105853051e-07,
      "loss": 0.7859,
      "step": 1795
    },
    {
      "epoch": 0.028022760708975848,
      "grad_norm": 10.616768836975098,
      "learning_rate": 5.600871731008718e-07,
      "loss": 0.9145,
      "step": 1800
    },
    {
      "epoch": 0.028100601710945222,
      "grad_norm": 6.610407829284668,
      "learning_rate": 5.616438356164383e-07,
      "loss": 0.8058,
      "step": 1805
    },
    {
      "epoch": 0.0281784427129146,
      "grad_norm": 4.978299617767334,
      "learning_rate": 5.63200498132005e-07,
      "loss": 0.8,
      "step": 1810
    },
    {
      "epoch": 0.02825628371488398,
      "grad_norm": 4.922807693481445,
      "learning_rate": 5.647571606475716e-07,
      "loss": 1.0388,
      "step": 1815
    },
    {
      "epoch": 0.028334124716853357,
      "grad_norm": 6.71333122253418,
      "learning_rate": 5.663138231631383e-07,
      "loss": 0.9221,
      "step": 1820
    },
    {
      "epoch": 0.02841196571882273,
      "grad_norm": 4.787428379058838,
      "learning_rate": 5.678704856787049e-07,
      "loss": 0.8824,
      "step": 1825
    },
    {
      "epoch": 0.02848980672079211,
      "grad_norm": 8.047598838806152,
      "learning_rate": 5.694271481942715e-07,
      "loss": 0.8187,
      "step": 1830
    },
    {
      "epoch": 0.028567647722761488,
      "grad_norm": 6.064495086669922,
      "learning_rate": 5.70983810709838e-07,
      "loss": 0.7595,
      "step": 1835
    },
    {
      "epoch": 0.028645488724730866,
      "grad_norm": 4.46295690536499,
      "learning_rate": 5.725404732254047e-07,
      "loss": 0.9397,
      "step": 1840
    },
    {
      "epoch": 0.02872332972670024,
      "grad_norm": 7.761974334716797,
      "learning_rate": 5.740971357409713e-07,
      "loss": 0.915,
      "step": 1845
    },
    {
      "epoch": 0.02880117072866962,
      "grad_norm": 5.134248733520508,
      "learning_rate": 5.75653798256538e-07,
      "loss": 0.862,
      "step": 1850
    },
    {
      "epoch": 0.028879011730638997,
      "grad_norm": 5.424485206604004,
      "learning_rate": 5.772104607721045e-07,
      "loss": 0.8359,
      "step": 1855
    },
    {
      "epoch": 0.028956852732608375,
      "grad_norm": 2.9714298248291016,
      "learning_rate": 5.787671232876712e-07,
      "loss": 0.7999,
      "step": 1860
    },
    {
      "epoch": 0.029034693734577753,
      "grad_norm": 6.131465911865234,
      "learning_rate": 5.803237858032378e-07,
      "loss": 0.8619,
      "step": 1865
    },
    {
      "epoch": 0.029112534736547128,
      "grad_norm": 7.894665241241455,
      "learning_rate": 5.818804483188045e-07,
      "loss": 0.771,
      "step": 1870
    },
    {
      "epoch": 0.029190375738516506,
      "grad_norm": 3.163548469543457,
      "learning_rate": 5.834371108343711e-07,
      "loss": 0.7482,
      "step": 1875
    },
    {
      "epoch": 0.029268216740485884,
      "grad_norm": 5.383469581604004,
      "learning_rate": 5.849937733499378e-07,
      "loss": 0.895,
      "step": 1880
    },
    {
      "epoch": 0.029346057742455262,
      "grad_norm": 6.841033935546875,
      "learning_rate": 5.865504358655043e-07,
      "loss": 0.8822,
      "step": 1885
    },
    {
      "epoch": 0.029423898744424637,
      "grad_norm": 9.069436073303223,
      "learning_rate": 5.88107098381071e-07,
      "loss": 0.8947,
      "step": 1890
    },
    {
      "epoch": 0.029501739746394015,
      "grad_norm": 5.3066725730896,
      "learning_rate": 5.896637608966376e-07,
      "loss": 0.8046,
      "step": 1895
    },
    {
      "epoch": 0.029579580748363393,
      "grad_norm": 5.761783599853516,
      "learning_rate": 5.912204234122043e-07,
      "loss": 0.9087,
      "step": 1900
    },
    {
      "epoch": 0.02965742175033277,
      "grad_norm": 3.4487996101379395,
      "learning_rate": 5.927770859277709e-07,
      "loss": 0.9291,
      "step": 1905
    },
    {
      "epoch": 0.029735262752302146,
      "grad_norm": 5.8793816566467285,
      "learning_rate": 5.943337484433375e-07,
      "loss": 0.8244,
      "step": 1910
    },
    {
      "epoch": 0.029813103754271524,
      "grad_norm": 6.812746047973633,
      "learning_rate": 5.958904109589041e-07,
      "loss": 0.8169,
      "step": 1915
    },
    {
      "epoch": 0.029890944756240902,
      "grad_norm": 5.695523738861084,
      "learning_rate": 5.974470734744707e-07,
      "loss": 0.9072,
      "step": 1920
    },
    {
      "epoch": 0.02996878575821028,
      "grad_norm": 3.443061590194702,
      "learning_rate": 5.990037359900373e-07,
      "loss": 0.8709,
      "step": 1925
    },
    {
      "epoch": 0.03004662676017966,
      "grad_norm": 6.014828681945801,
      "learning_rate": 6.00560398505604e-07,
      "loss": 0.917,
      "step": 1930
    },
    {
      "epoch": 0.030124467762149033,
      "grad_norm": 4.14946985244751,
      "learning_rate": 6.021170610211705e-07,
      "loss": 0.8827,
      "step": 1935
    },
    {
      "epoch": 0.03020230876411841,
      "grad_norm": 4.128273963928223,
      "learning_rate": 6.036737235367372e-07,
      "loss": 0.8062,
      "step": 1940
    },
    {
      "epoch": 0.03028014976608779,
      "grad_norm": 5.5036115646362305,
      "learning_rate": 6.052303860523038e-07,
      "loss": 0.9832,
      "step": 1945
    },
    {
      "epoch": 0.030357990768057168,
      "grad_norm": 5.694386005401611,
      "learning_rate": 6.067870485678705e-07,
      "loss": 0.7735,
      "step": 1950
    },
    {
      "epoch": 0.030435831770026543,
      "grad_norm": 3.861293315887451,
      "learning_rate": 6.083437110834371e-07,
      "loss": 0.8782,
      "step": 1955
    },
    {
      "epoch": 0.03051367277199592,
      "grad_norm": 5.179184436798096,
      "learning_rate": 6.099003735990037e-07,
      "loss": 0.908,
      "step": 1960
    },
    {
      "epoch": 0.0305915137739653,
      "grad_norm": 4.929222106933594,
      "learning_rate": 6.114570361145703e-07,
      "loss": 0.8967,
      "step": 1965
    },
    {
      "epoch": 0.030669354775934677,
      "grad_norm": 3.300053596496582,
      "learning_rate": 6.13013698630137e-07,
      "loss": 0.9517,
      "step": 1970
    },
    {
      "epoch": 0.03074719577790405,
      "grad_norm": 4.976810932159424,
      "learning_rate": 6.145703611457036e-07,
      "loss": 0.8676,
      "step": 1975
    },
    {
      "epoch": 0.03082503677987343,
      "grad_norm": 3.866328477859497,
      "learning_rate": 6.161270236612703e-07,
      "loss": 0.9735,
      "step": 1980
    },
    {
      "epoch": 0.030902877781842808,
      "grad_norm": 4.272680759429932,
      "learning_rate": 6.176836861768369e-07,
      "loss": 0.9716,
      "step": 1985
    },
    {
      "epoch": 0.030980718783812186,
      "grad_norm": 6.74641752243042,
      "learning_rate": 6.192403486924035e-07,
      "loss": 0.815,
      "step": 1990
    },
    {
      "epoch": 0.031058559785781564,
      "grad_norm": 3.4278452396392822,
      "learning_rate": 6.207970112079701e-07,
      "loss": 0.9165,
      "step": 1995
    },
    {
      "epoch": 0.03113640078775094,
      "grad_norm": 7.538846492767334,
      "learning_rate": 6.223536737235368e-07,
      "loss": 0.9513,
      "step": 2000
    },
    {
      "epoch": 0.031214241789720317,
      "grad_norm": 4.969770431518555,
      "learning_rate": 6.239103362391034e-07,
      "loss": 0.816,
      "step": 2005
    },
    {
      "epoch": 0.03129208279168969,
      "grad_norm": 9.244134902954102,
      "learning_rate": 6.2546699875467e-07,
      "loss": 0.9293,
      "step": 2010
    },
    {
      "epoch": 0.03136992379365907,
      "grad_norm": 5.617055416107178,
      "learning_rate": 6.270236612702365e-07,
      "loss": 0.8553,
      "step": 2015
    },
    {
      "epoch": 0.03144776479562845,
      "grad_norm": 4.888432502746582,
      "learning_rate": 6.285803237858031e-07,
      "loss": 0.8679,
      "step": 2020
    },
    {
      "epoch": 0.031525605797597826,
      "grad_norm": 4.528554916381836,
      "learning_rate": 6.301369863013698e-07,
      "loss": 0.9086,
      "step": 2025
    },
    {
      "epoch": 0.031603446799567204,
      "grad_norm": 6.504762172698975,
      "learning_rate": 6.316936488169364e-07,
      "loss": 0.9501,
      "step": 2030
    },
    {
      "epoch": 0.03168128780153658,
      "grad_norm": 3.974257230758667,
      "learning_rate": 6.332503113325031e-07,
      "loss": 0.888,
      "step": 2035
    },
    {
      "epoch": 0.03175912880350596,
      "grad_norm": 8.628198623657227,
      "learning_rate": 6.348069738480696e-07,
      "loss": 0.8664,
      "step": 2040
    },
    {
      "epoch": 0.03183696980547534,
      "grad_norm": 4.1892805099487305,
      "learning_rate": 6.363636363636363e-07,
      "loss": 0.9896,
      "step": 2045
    },
    {
      "epoch": 0.03191481080744471,
      "grad_norm": 5.350588321685791,
      "learning_rate": 6.37920298879203e-07,
      "loss": 0.8815,
      "step": 2050
    },
    {
      "epoch": 0.03199265180941409,
      "grad_norm": 5.569740295410156,
      "learning_rate": 6.394769613947696e-07,
      "loss": 0.8785,
      "step": 2055
    },
    {
      "epoch": 0.032070492811383466,
      "grad_norm": 7.358509063720703,
      "learning_rate": 6.410336239103362e-07,
      "loss": 0.8415,
      "step": 2060
    },
    {
      "epoch": 0.032148333813352845,
      "grad_norm": 5.384446144104004,
      "learning_rate": 6.425902864259029e-07,
      "loss": 0.9108,
      "step": 2065
    },
    {
      "epoch": 0.03222617481532222,
      "grad_norm": 4.48892068862915,
      "learning_rate": 6.441469489414694e-07,
      "loss": 0.9423,
      "step": 2070
    },
    {
      "epoch": 0.0323040158172916,
      "grad_norm": 4.302936553955078,
      "learning_rate": 6.457036114570361e-07,
      "loss": 0.8849,
      "step": 2075
    },
    {
      "epoch": 0.03238185681926098,
      "grad_norm": 5.185121536254883,
      "learning_rate": 6.472602739726027e-07,
      "loss": 0.8177,
      "step": 2080
    },
    {
      "epoch": 0.03245969782123036,
      "grad_norm": 3.2999234199523926,
      "learning_rate": 6.488169364881694e-07,
      "loss": 0.9199,
      "step": 2085
    },
    {
      "epoch": 0.032537538823199735,
      "grad_norm": 19.133163452148438,
      "learning_rate": 6.50373599003736e-07,
      "loss": 0.9283,
      "step": 2090
    },
    {
      "epoch": 0.03261537982516911,
      "grad_norm": 3.4535083770751953,
      "learning_rate": 6.519302615193026e-07,
      "loss": 0.9707,
      "step": 2095
    },
    {
      "epoch": 0.032693220827138485,
      "grad_norm": 11.507316589355469,
      "learning_rate": 6.534869240348691e-07,
      "loss": 0.8878,
      "step": 2100
    },
    {
      "epoch": 0.03277106182910786,
      "grad_norm": 21.363101959228516,
      "learning_rate": 6.550435865504358e-07,
      "loss": 0.9761,
      "step": 2105
    },
    {
      "epoch": 0.03284890283107724,
      "grad_norm": 4.29213285446167,
      "learning_rate": 6.566002490660024e-07,
      "loss": 0.7789,
      "step": 2110
    },
    {
      "epoch": 0.03292674383304662,
      "grad_norm": 7.540319442749023,
      "learning_rate": 6.581569115815691e-07,
      "loss": 0.7991,
      "step": 2115
    },
    {
      "epoch": 0.033004584835016,
      "grad_norm": 3.658780097961426,
      "learning_rate": 6.597135740971356e-07,
      "loss": 0.9211,
      "step": 2120
    },
    {
      "epoch": 0.033082425836985375,
      "grad_norm": 8.205567359924316,
      "learning_rate": 6.612702366127023e-07,
      "loss": 0.8083,
      "step": 2125
    },
    {
      "epoch": 0.033160266838954754,
      "grad_norm": 6.272342681884766,
      "learning_rate": 6.628268991282689e-07,
      "loss": 0.9724,
      "step": 2130
    },
    {
      "epoch": 0.03323810784092413,
      "grad_norm": 7.037917137145996,
      "learning_rate": 6.643835616438356e-07,
      "loss": 0.8821,
      "step": 2135
    },
    {
      "epoch": 0.0333159488428935,
      "grad_norm": 6.3946027755737305,
      "learning_rate": 6.659402241594022e-07,
      "loss": 0.9065,
      "step": 2140
    },
    {
      "epoch": 0.03339378984486288,
      "grad_norm": 7.079307556152344,
      "learning_rate": 6.674968866749689e-07,
      "loss": 0.993,
      "step": 2145
    },
    {
      "epoch": 0.03347163084683226,
      "grad_norm": 6.372123718261719,
      "learning_rate": 6.690535491905354e-07,
      "loss": 0.9372,
      "step": 2150
    },
    {
      "epoch": 0.03354947184880164,
      "grad_norm": 2.9949862957000732,
      "learning_rate": 6.706102117061021e-07,
      "loss": 0.7785,
      "step": 2155
    },
    {
      "epoch": 0.033627312850771016,
      "grad_norm": 5.278440475463867,
      "learning_rate": 6.721668742216687e-07,
      "loss": 0.8646,
      "step": 2160
    },
    {
      "epoch": 0.033705153852740394,
      "grad_norm": 3.972559928894043,
      "learning_rate": 6.737235367372354e-07,
      "loss": 0.8816,
      "step": 2165
    },
    {
      "epoch": 0.03378299485470977,
      "grad_norm": 7.038811683654785,
      "learning_rate": 6.75280199252802e-07,
      "loss": 0.8586,
      "step": 2170
    },
    {
      "epoch": 0.03386083585667915,
      "grad_norm": 4.659327507019043,
      "learning_rate": 6.768368617683686e-07,
      "loss": 0.8671,
      "step": 2175
    },
    {
      "epoch": 0.03393867685864852,
      "grad_norm": 3.272244453430176,
      "learning_rate": 6.783935242839352e-07,
      "loss": 0.8553,
      "step": 2180
    },
    {
      "epoch": 0.0340165178606179,
      "grad_norm": 4.486519813537598,
      "learning_rate": 6.799501867995019e-07,
      "loss": 0.9569,
      "step": 2185
    },
    {
      "epoch": 0.03409435886258728,
      "grad_norm": 3.5172436237335205,
      "learning_rate": 6.815068493150684e-07,
      "loss": 0.9172,
      "step": 2190
    },
    {
      "epoch": 0.034172199864556656,
      "grad_norm": 8.919556617736816,
      "learning_rate": 6.830635118306351e-07,
      "loss": 0.9005,
      "step": 2195
    },
    {
      "epoch": 0.034250040866526034,
      "grad_norm": 3.1688411235809326,
      "learning_rate": 6.846201743462016e-07,
      "loss": 0.9052,
      "step": 2200
    },
    {
      "epoch": 0.03432788186849541,
      "grad_norm": 8.181324005126953,
      "learning_rate": 6.861768368617683e-07,
      "loss": 0.9344,
      "step": 2205
    },
    {
      "epoch": 0.03440572287046479,
      "grad_norm": 14.188647270202637,
      "learning_rate": 6.877334993773349e-07,
      "loss": 0.8221,
      "step": 2210
    },
    {
      "epoch": 0.03448356387243417,
      "grad_norm": 2.8779571056365967,
      "learning_rate": 6.892901618929016e-07,
      "loss": 0.8213,
      "step": 2215
    },
    {
      "epoch": 0.034561404874403547,
      "grad_norm": 4.762483596801758,
      "learning_rate": 6.908468244084682e-07,
      "loss": 0.9539,
      "step": 2220
    },
    {
      "epoch": 0.03463924587637292,
      "grad_norm": 5.372674942016602,
      "learning_rate": 6.924034869240348e-07,
      "loss": 0.9323,
      "step": 2225
    },
    {
      "epoch": 0.034717086878342296,
      "grad_norm": 4.73727560043335,
      "learning_rate": 6.939601494396014e-07,
      "loss": 0.9555,
      "step": 2230
    },
    {
      "epoch": 0.034794927880311674,
      "grad_norm": 2.479062557220459,
      "learning_rate": 6.955168119551681e-07,
      "loss": 0.8026,
      "step": 2235
    },
    {
      "epoch": 0.03487276888228105,
      "grad_norm": 4.98023796081543,
      "learning_rate": 6.970734744707347e-07,
      "loss": 0.9514,
      "step": 2240
    },
    {
      "epoch": 0.03495060988425043,
      "grad_norm": 4.072389125823975,
      "learning_rate": 6.986301369863014e-07,
      "loss": 0.9739,
      "step": 2245
    },
    {
      "epoch": 0.03502845088621981,
      "grad_norm": 3.26598858833313,
      "learning_rate": 7.00186799501868e-07,
      "loss": 0.8112,
      "step": 2250
    },
    {
      "epoch": 0.03510629188818919,
      "grad_norm": 10.324394226074219,
      "learning_rate": 7.017434620174346e-07,
      "loss": 0.8578,
      "step": 2255
    },
    {
      "epoch": 0.035184132890158565,
      "grad_norm": 7.579793453216553,
      "learning_rate": 7.033001245330012e-07,
      "loss": 0.8586,
      "step": 2260
    },
    {
      "epoch": 0.035261973892127936,
      "grad_norm": 3.6266613006591797,
      "learning_rate": 7.048567870485679e-07,
      "loss": 0.8904,
      "step": 2265
    },
    {
      "epoch": 0.035339814894097314,
      "grad_norm": 4.336295127868652,
      "learning_rate": 7.064134495641345e-07,
      "loss": 0.84,
      "step": 2270
    },
    {
      "epoch": 0.03541765589606669,
      "grad_norm": 3.5872817039489746,
      "learning_rate": 7.079701120797012e-07,
      "loss": 0.7951,
      "step": 2275
    },
    {
      "epoch": 0.03549549689803607,
      "grad_norm": 4.598228454589844,
      "learning_rate": 7.095267745952676e-07,
      "loss": 0.7467,
      "step": 2280
    },
    {
      "epoch": 0.03557333790000545,
      "grad_norm": 3.560222625732422,
      "learning_rate": 7.110834371108343e-07,
      "loss": 0.9047,
      "step": 2285
    },
    {
      "epoch": 0.03565117890197483,
      "grad_norm": 2.8487563133239746,
      "learning_rate": 7.126400996264009e-07,
      "loss": 0.9243,
      "step": 2290
    },
    {
      "epoch": 0.035729019903944205,
      "grad_norm": 5.525490760803223,
      "learning_rate": 7.141967621419676e-07,
      "loss": 0.8549,
      "step": 2295
    },
    {
      "epoch": 0.03580686090591358,
      "grad_norm": 3.5428950786590576,
      "learning_rate": 7.157534246575342e-07,
      "loss": 0.9336,
      "step": 2300
    },
    {
      "epoch": 0.03588470190788296,
      "grad_norm": 8.396724700927734,
      "learning_rate": 7.173100871731008e-07,
      "loss": 0.9101,
      "step": 2305
    },
    {
      "epoch": 0.03596254290985233,
      "grad_norm": 6.355068206787109,
      "learning_rate": 7.188667496886674e-07,
      "loss": 0.8673,
      "step": 2310
    },
    {
      "epoch": 0.03604038391182171,
      "grad_norm": 8.388739585876465,
      "learning_rate": 7.204234122042341e-07,
      "loss": 0.9225,
      "step": 2315
    },
    {
      "epoch": 0.03611822491379109,
      "grad_norm": 4.088027477264404,
      "learning_rate": 7.219800747198007e-07,
      "loss": 0.8003,
      "step": 2320
    },
    {
      "epoch": 0.03619606591576047,
      "grad_norm": 3.6764137744903564,
      "learning_rate": 7.235367372353674e-07,
      "loss": 0.782,
      "step": 2325
    },
    {
      "epoch": 0.036273906917729845,
      "grad_norm": 3.6554110050201416,
      "learning_rate": 7.25093399750934e-07,
      "loss": 0.9257,
      "step": 2330
    },
    {
      "epoch": 0.03635174791969922,
      "grad_norm": 6.99379301071167,
      "learning_rate": 7.266500622665006e-07,
      "loss": 0.9329,
      "step": 2335
    },
    {
      "epoch": 0.0364295889216686,
      "grad_norm": 3.984800100326538,
      "learning_rate": 7.282067247820672e-07,
      "loss": 0.9596,
      "step": 2340
    },
    {
      "epoch": 0.03650742992363798,
      "grad_norm": 7.992112159729004,
      "learning_rate": 7.297633872976339e-07,
      "loss": 0.8945,
      "step": 2345
    },
    {
      "epoch": 0.03658527092560736,
      "grad_norm": 3.314192295074463,
      "learning_rate": 7.313200498132005e-07,
      "loss": 0.817,
      "step": 2350
    },
    {
      "epoch": 0.03666311192757673,
      "grad_norm": 5.738452434539795,
      "learning_rate": 7.328767123287672e-07,
      "loss": 0.8133,
      "step": 2355
    },
    {
      "epoch": 0.03674095292954611,
      "grad_norm": 4.364063739776611,
      "learning_rate": 7.344333748443337e-07,
      "loss": 0.7555,
      "step": 2360
    },
    {
      "epoch": 0.036818793931515485,
      "grad_norm": 6.397834777832031,
      "learning_rate": 7.359900373599004e-07,
      "loss": 0.9289,
      "step": 2365
    },
    {
      "epoch": 0.03689663493348486,
      "grad_norm": 4.602386951446533,
      "learning_rate": 7.375466998754669e-07,
      "loss": 0.8466,
      "step": 2370
    },
    {
      "epoch": 0.03697447593545424,
      "grad_norm": 4.438021659851074,
      "learning_rate": 7.391033623910336e-07,
      "loss": 0.8155,
      "step": 2375
    },
    {
      "epoch": 0.03705231693742362,
      "grad_norm": 5.829861164093018,
      "learning_rate": 7.406600249066002e-07,
      "loss": 0.9119,
      "step": 2380
    },
    {
      "epoch": 0.037130157939393,
      "grad_norm": 3.999397039413452,
      "learning_rate": 7.422166874221668e-07,
      "loss": 0.9544,
      "step": 2385
    },
    {
      "epoch": 0.037207998941362376,
      "grad_norm": 7.094069480895996,
      "learning_rate": 7.437733499377334e-07,
      "loss": 0.8562,
      "step": 2390
    },
    {
      "epoch": 0.03728583994333175,
      "grad_norm": 7.502668857574463,
      "learning_rate": 7.453300124533001e-07,
      "loss": 0.7626,
      "step": 2395
    },
    {
      "epoch": 0.037363680945301125,
      "grad_norm": 4.224865913391113,
      "learning_rate": 7.468866749688667e-07,
      "loss": 0.8287,
      "step": 2400
    },
    {
      "epoch": 0.037441521947270504,
      "grad_norm": 4.2678046226501465,
      "learning_rate": 7.484433374844334e-07,
      "loss": 0.9631,
      "step": 2405
    },
    {
      "epoch": 0.03751936294923988,
      "grad_norm": 4.143566608428955,
      "learning_rate": 7.5e-07,
      "loss": 0.9013,
      "step": 2410
    },
    {
      "epoch": 0.03759720395120926,
      "grad_norm": 3.8706650733947754,
      "learning_rate": 7.515566625155666e-07,
      "loss": 0.8463,
      "step": 2415
    },
    {
      "epoch": 0.03767504495317864,
      "grad_norm": 6.372035503387451,
      "learning_rate": 7.531133250311332e-07,
      "loss": 0.8966,
      "step": 2420
    },
    {
      "epoch": 0.037752885955148016,
      "grad_norm": 4.3398613929748535,
      "learning_rate": 7.546699875466999e-07,
      "loss": 0.8045,
      "step": 2425
    },
    {
      "epoch": 0.037830726957117394,
      "grad_norm": 2.7824904918670654,
      "learning_rate": 7.562266500622665e-07,
      "loss": 0.8311,
      "step": 2430
    },
    {
      "epoch": 0.03790856795908677,
      "grad_norm": 3.9570069313049316,
      "learning_rate": 7.577833125778332e-07,
      "loss": 0.9548,
      "step": 2435
    },
    {
      "epoch": 0.037986408961056144,
      "grad_norm": 4.316530227661133,
      "learning_rate": 7.593399750933997e-07,
      "loss": 0.7945,
      "step": 2440
    },
    {
      "epoch": 0.03806424996302552,
      "grad_norm": 4.4045844078063965,
      "learning_rate": 7.608966376089664e-07,
      "loss": 0.9145,
      "step": 2445
    },
    {
      "epoch": 0.0381420909649949,
      "grad_norm": 3.736820697784424,
      "learning_rate": 7.62453300124533e-07,
      "loss": 0.9292,
      "step": 2450
    },
    {
      "epoch": 0.03821993196696428,
      "grad_norm": 3.8448410034179688,
      "learning_rate": 7.640099626400996e-07,
      "loss": 0.8863,
      "step": 2455
    },
    {
      "epoch": 0.038297772968933656,
      "grad_norm": 7.468678951263428,
      "learning_rate": 7.655666251556662e-07,
      "loss": 0.8776,
      "step": 2460
    },
    {
      "epoch": 0.038375613970903034,
      "grad_norm": 4.066128253936768,
      "learning_rate": 7.671232876712328e-07,
      "loss": 0.906,
      "step": 2465
    },
    {
      "epoch": 0.03845345497287241,
      "grad_norm": 8.009504318237305,
      "learning_rate": 7.686799501867994e-07,
      "loss": 0.9295,
      "step": 2470
    },
    {
      "epoch": 0.03853129597484179,
      "grad_norm": 3.9662601947784424,
      "learning_rate": 7.702366127023661e-07,
      "loss": 0.927,
      "step": 2475
    },
    {
      "epoch": 0.03860913697681117,
      "grad_norm": 3.94587779045105,
      "learning_rate": 7.717932752179327e-07,
      "loss": 0.9679,
      "step": 2480
    },
    {
      "epoch": 0.03868697797878054,
      "grad_norm": 3.856196641921997,
      "learning_rate": 7.733499377334994e-07,
      "loss": 0.981,
      "step": 2485
    },
    {
      "epoch": 0.03876481898074992,
      "grad_norm": 12.542234420776367,
      "learning_rate": 7.749066002490659e-07,
      "loss": 0.8065,
      "step": 2490
    },
    {
      "epoch": 0.038842659982719296,
      "grad_norm": 5.717936038970947,
      "learning_rate": 7.764632627646326e-07,
      "loss": 0.8912,
      "step": 2495
    },
    {
      "epoch": 0.038920500984688675,
      "grad_norm": 9.94604206085205,
      "learning_rate": 7.780199252801992e-07,
      "loss": 0.9442,
      "step": 2500
    },
    {
      "epoch": 0.03899834198665805,
      "grad_norm": 5.26216983795166,
      "learning_rate": 7.795765877957659e-07,
      "loss": 0.9552,
      "step": 2505
    },
    {
      "epoch": 0.03907618298862743,
      "grad_norm": 6.468954563140869,
      "learning_rate": 7.811332503113325e-07,
      "loss": 0.9372,
      "step": 2510
    },
    {
      "epoch": 0.03915402399059681,
      "grad_norm": 2.9301857948303223,
      "learning_rate": 7.826899128268992e-07,
      "loss": 0.8479,
      "step": 2515
    },
    {
      "epoch": 0.03923186499256619,
      "grad_norm": 6.389108657836914,
      "learning_rate": 7.842465753424657e-07,
      "loss": 0.9255,
      "step": 2520
    },
    {
      "epoch": 0.03930970599453556,
      "grad_norm": 4.842959880828857,
      "learning_rate": 7.858032378580324e-07,
      "loss": 0.8478,
      "step": 2525
    },
    {
      "epoch": 0.03938754699650494,
      "grad_norm": 3.118706464767456,
      "learning_rate": 7.87359900373599e-07,
      "loss": 0.8668,
      "step": 2530
    },
    {
      "epoch": 0.039465387998474315,
      "grad_norm": 6.257364273071289,
      "learning_rate": 7.889165628891657e-07,
      "loss": 0.873,
      "step": 2535
    },
    {
      "epoch": 0.03954322900044369,
      "grad_norm": 4.405180931091309,
      "learning_rate": 7.904732254047323e-07,
      "loss": 0.841,
      "step": 2540
    },
    {
      "epoch": 0.03962107000241307,
      "grad_norm": 9.870434761047363,
      "learning_rate": 7.920298879202987e-07,
      "loss": 0.937,
      "step": 2545
    },
    {
      "epoch": 0.03969891100438245,
      "grad_norm": 3.4615135192871094,
      "learning_rate": 7.935865504358654e-07,
      "loss": 0.9465,
      "step": 2550
    },
    {
      "epoch": 0.03977675200635183,
      "grad_norm": 4.9833760261535645,
      "learning_rate": 7.95143212951432e-07,
      "loss": 0.8261,
      "step": 2555
    },
    {
      "epoch": 0.039854593008321205,
      "grad_norm": 4.042236804962158,
      "learning_rate": 7.966998754669987e-07,
      "loss": 0.9602,
      "step": 2560
    },
    {
      "epoch": 0.039932434010290584,
      "grad_norm": 4.549630641937256,
      "learning_rate": 7.982565379825654e-07,
      "loss": 0.8674,
      "step": 2565
    },
    {
      "epoch": 0.040010275012259955,
      "grad_norm": 3.67543363571167,
      "learning_rate": 7.998132004981319e-07,
      "loss": 0.9073,
      "step": 2570
    },
    {
      "epoch": 0.04008811601422933,
      "grad_norm": 6.078221321105957,
      "learning_rate": 8.013698630136985e-07,
      "loss": 0.9359,
      "step": 2575
    },
    {
      "epoch": 0.04016595701619871,
      "grad_norm": 5.599534034729004,
      "learning_rate": 8.029265255292652e-07,
      "loss": 0.8628,
      "step": 2580
    },
    {
      "epoch": 0.04024379801816809,
      "grad_norm": 5.098958492279053,
      "learning_rate": 8.044831880448319e-07,
      "loss": 0.9412,
      "step": 2585
    },
    {
      "epoch": 0.04032163902013747,
      "grad_norm": 7.108897686004639,
      "learning_rate": 8.060398505603985e-07,
      "loss": 0.9517,
      "step": 2590
    },
    {
      "epoch": 0.040399480022106846,
      "grad_norm": 4.495419979095459,
      "learning_rate": 8.075965130759652e-07,
      "loss": 0.8605,
      "step": 2595
    },
    {
      "epoch": 0.040477321024076224,
      "grad_norm": 4.583033084869385,
      "learning_rate": 8.091531755915317e-07,
      "loss": 0.8956,
      "step": 2600
    },
    {
      "epoch": 0.0405551620260456,
      "grad_norm": 5.067065238952637,
      "learning_rate": 8.107098381070983e-07,
      "loss": 0.8306,
      "step": 2605
    },
    {
      "epoch": 0.04063300302801497,
      "grad_norm": 7.724658012390137,
      "learning_rate": 8.12266500622665e-07,
      "loss": 0.8716,
      "step": 2610
    },
    {
      "epoch": 0.04071084402998435,
      "grad_norm": 2.8972911834716797,
      "learning_rate": 8.138231631382317e-07,
      "loss": 0.8987,
      "step": 2615
    },
    {
      "epoch": 0.04078868503195373,
      "grad_norm": 7.840747833251953,
      "learning_rate": 8.153798256537983e-07,
      "loss": 0.8418,
      "step": 2620
    },
    {
      "epoch": 0.04086652603392311,
      "grad_norm": 7.727685928344727,
      "learning_rate": 8.169364881693648e-07,
      "loss": 0.9107,
      "step": 2625
    },
    {
      "epoch": 0.040944367035892486,
      "grad_norm": 3.801807165145874,
      "learning_rate": 8.184931506849315e-07,
      "loss": 0.8083,
      "step": 2630
    },
    {
      "epoch": 0.041022208037861864,
      "grad_norm": 12.985006332397461,
      "learning_rate": 8.20049813200498e-07,
      "loss": 0.9866,
      "step": 2635
    },
    {
      "epoch": 0.04110004903983124,
      "grad_norm": 3.2062785625457764,
      "learning_rate": 8.216064757160647e-07,
      "loss": 0.8945,
      "step": 2640
    },
    {
      "epoch": 0.04117789004180062,
      "grad_norm": 4.6915459632873535,
      "learning_rate": 8.231631382316313e-07,
      "loss": 0.8236,
      "step": 2645
    },
    {
      "epoch": 0.04125573104377,
      "grad_norm": 3.5803701877593994,
      "learning_rate": 8.247198007471979e-07,
      "loss": 0.8776,
      "step": 2650
    },
    {
      "epoch": 0.04133357204573937,
      "grad_norm": 12.053580284118652,
      "learning_rate": 8.262764632627645e-07,
      "loss": 0.9105,
      "step": 2655
    },
    {
      "epoch": 0.04141141304770875,
      "grad_norm": 6.285280227661133,
      "learning_rate": 8.278331257783312e-07,
      "loss": 0.9066,
      "step": 2660
    },
    {
      "epoch": 0.041489254049678126,
      "grad_norm": 5.232326984405518,
      "learning_rate": 8.293897882938978e-07,
      "loss": 0.8508,
      "step": 2665
    },
    {
      "epoch": 0.041567095051647504,
      "grad_norm": 3.038318395614624,
      "learning_rate": 8.309464508094645e-07,
      "loss": 0.8356,
      "step": 2670
    },
    {
      "epoch": 0.04164493605361688,
      "grad_norm": 7.6262335777282715,
      "learning_rate": 8.32503113325031e-07,
      "loss": 0.9568,
      "step": 2675
    },
    {
      "epoch": 0.04172277705558626,
      "grad_norm": 3.0321080684661865,
      "learning_rate": 8.340597758405977e-07,
      "loss": 0.7881,
      "step": 2680
    },
    {
      "epoch": 0.04180061805755564,
      "grad_norm": 9.739387512207031,
      "learning_rate": 8.356164383561643e-07,
      "loss": 0.74,
      "step": 2685
    },
    {
      "epoch": 0.04187845905952502,
      "grad_norm": 8.000276565551758,
      "learning_rate": 8.37173100871731e-07,
      "loss": 0.9413,
      "step": 2690
    },
    {
      "epoch": 0.041956300061494395,
      "grad_norm": 6.706925868988037,
      "learning_rate": 8.387297633872976e-07,
      "loss": 0.7788,
      "step": 2695
    },
    {
      "epoch": 0.042034141063463766,
      "grad_norm": 6.568419933319092,
      "learning_rate": 8.402864259028643e-07,
      "loss": 0.8554,
      "step": 2700
    },
    {
      "epoch": 0.042111982065433144,
      "grad_norm": 3.8879165649414062,
      "learning_rate": 8.418430884184308e-07,
      "loss": 0.9076,
      "step": 2705
    },
    {
      "epoch": 0.04218982306740252,
      "grad_norm": 5.89036226272583,
      "learning_rate": 8.433997509339975e-07,
      "loss": 0.91,
      "step": 2710
    },
    {
      "epoch": 0.0422676640693719,
      "grad_norm": 5.522625923156738,
      "learning_rate": 8.449564134495641e-07,
      "loss": 0.9102,
      "step": 2715
    },
    {
      "epoch": 0.04234550507134128,
      "grad_norm": 4.862393379211426,
      "learning_rate": 8.465130759651308e-07,
      "loss": 0.8503,
      "step": 2720
    },
    {
      "epoch": 0.04242334607331066,
      "grad_norm": 8.545342445373535,
      "learning_rate": 8.480697384806973e-07,
      "loss": 0.7882,
      "step": 2725
    },
    {
      "epoch": 0.042501187075280035,
      "grad_norm": 3.1325466632843018,
      "learning_rate": 8.496264009962639e-07,
      "loss": 0.7993,
      "step": 2730
    },
    {
      "epoch": 0.04257902807724941,
      "grad_norm": 3.6244635581970215,
      "learning_rate": 8.511830635118305e-07,
      "loss": 0.7498,
      "step": 2735
    },
    {
      "epoch": 0.042656869079218784,
      "grad_norm": 7.154248237609863,
      "learning_rate": 8.527397260273972e-07,
      "loss": 0.856,
      "step": 2740
    },
    {
      "epoch": 0.04273471008118816,
      "grad_norm": 4.3253960609436035,
      "learning_rate": 8.542963885429638e-07,
      "loss": 1.0214,
      "step": 2745
    },
    {
      "epoch": 0.04281255108315754,
      "grad_norm": 4.56231164932251,
      "learning_rate": 8.558530510585305e-07,
      "loss": 0.8501,
      "step": 2750
    },
    {
      "epoch": 0.04289039208512692,
      "grad_norm": 3.396204710006714,
      "learning_rate": 8.57409713574097e-07,
      "loss": 0.9002,
      "step": 2755
    },
    {
      "epoch": 0.0429682330870963,
      "grad_norm": 5.2896952629089355,
      "learning_rate": 8.589663760896637e-07,
      "loss": 0.8824,
      "step": 2760
    },
    {
      "epoch": 0.043046074089065675,
      "grad_norm": 2.9441330432891846,
      "learning_rate": 8.605230386052303e-07,
      "loss": 0.915,
      "step": 2765
    },
    {
      "epoch": 0.04312391509103505,
      "grad_norm": 3.7935092449188232,
      "learning_rate": 8.62079701120797e-07,
      "loss": 0.9273,
      "step": 2770
    },
    {
      "epoch": 0.04320175609300443,
      "grad_norm": 2.87821102142334,
      "learning_rate": 8.636363636363636e-07,
      "loss": 0.7991,
      "step": 2775
    },
    {
      "epoch": 0.04327959709497381,
      "grad_norm": 6.359185218811035,
      "learning_rate": 8.651930261519303e-07,
      "loss": 0.8009,
      "step": 2780
    },
    {
      "epoch": 0.04335743809694318,
      "grad_norm": 4.339592456817627,
      "learning_rate": 8.667496886674968e-07,
      "loss": 0.9357,
      "step": 2785
    },
    {
      "epoch": 0.04343527909891256,
      "grad_norm": 5.373045921325684,
      "learning_rate": 8.683063511830635e-07,
      "loss": 0.9278,
      "step": 2790
    },
    {
      "epoch": 0.04351312010088194,
      "grad_norm": 4.687058448791504,
      "learning_rate": 8.698630136986301e-07,
      "loss": 0.8102,
      "step": 2795
    },
    {
      "epoch": 0.043590961102851315,
      "grad_norm": 3.0270180702209473,
      "learning_rate": 8.714196762141968e-07,
      "loss": 0.7115,
      "step": 2800
    },
    {
      "epoch": 0.04366880210482069,
      "grad_norm": 4.379403114318848,
      "learning_rate": 8.729763387297634e-07,
      "loss": 0.9258,
      "step": 2805
    },
    {
      "epoch": 0.04374664310679007,
      "grad_norm": 5.339996814727783,
      "learning_rate": 8.7453300124533e-07,
      "loss": 0.8254,
      "step": 2810
    },
    {
      "epoch": 0.04382448410875945,
      "grad_norm": 8.269269943237305,
      "learning_rate": 8.760896637608965e-07,
      "loss": 0.8743,
      "step": 2815
    },
    {
      "epoch": 0.04390232511072883,
      "grad_norm": 3.314060688018799,
      "learning_rate": 8.776463262764632e-07,
      "loss": 0.8462,
      "step": 2820
    },
    {
      "epoch": 0.043980166112698206,
      "grad_norm": 5.628077507019043,
      "learning_rate": 8.792029887920298e-07,
      "loss": 0.8659,
      "step": 2825
    },
    {
      "epoch": 0.04405800711466758,
      "grad_norm": 5.050654888153076,
      "learning_rate": 8.807596513075965e-07,
      "loss": 0.9523,
      "step": 2830
    },
    {
      "epoch": 0.044135848116636955,
      "grad_norm": 3.8180229663848877,
      "learning_rate": 8.82316313823163e-07,
      "loss": 0.8504,
      "step": 2835
    },
    {
      "epoch": 0.044213689118606334,
      "grad_norm": 3.3269238471984863,
      "learning_rate": 8.838729763387297e-07,
      "loss": 0.8544,
      "step": 2840
    },
    {
      "epoch": 0.04429153012057571,
      "grad_norm": 5.524733066558838,
      "learning_rate": 8.854296388542963e-07,
      "loss": 0.8963,
      "step": 2845
    },
    {
      "epoch": 0.04436937112254509,
      "grad_norm": 4.741829872131348,
      "learning_rate": 8.86986301369863e-07,
      "loss": 0.8262,
      "step": 2850
    },
    {
      "epoch": 0.04444721212451447,
      "grad_norm": 4.213449001312256,
      "learning_rate": 8.885429638854296e-07,
      "loss": 0.8722,
      "step": 2855
    },
    {
      "epoch": 0.044525053126483846,
      "grad_norm": 3.3883323669433594,
      "learning_rate": 8.900996264009963e-07,
      "loss": 0.8378,
      "step": 2860
    },
    {
      "epoch": 0.044602894128453224,
      "grad_norm": 5.192069053649902,
      "learning_rate": 8.916562889165628e-07,
      "loss": 0.8204,
      "step": 2865
    },
    {
      "epoch": 0.044680735130422596,
      "grad_norm": 3.5852484703063965,
      "learning_rate": 8.932129514321295e-07,
      "loss": 0.9008,
      "step": 2870
    },
    {
      "epoch": 0.044758576132391974,
      "grad_norm": 5.334090709686279,
      "learning_rate": 8.947696139476961e-07,
      "loss": 0.7885,
      "step": 2875
    },
    {
      "epoch": 0.04483641713436135,
      "grad_norm": 5.502117156982422,
      "learning_rate": 8.963262764632628e-07,
      "loss": 0.8715,
      "step": 2880
    },
    {
      "epoch": 0.04491425813633073,
      "grad_norm": 3.577226400375366,
      "learning_rate": 8.978829389788294e-07,
      "loss": 0.8076,
      "step": 2885
    },
    {
      "epoch": 0.04499209913830011,
      "grad_norm": 2.6925950050354004,
      "learning_rate": 8.99439601494396e-07,
      "loss": 0.8624,
      "step": 2890
    },
    {
      "epoch": 0.045069940140269486,
      "grad_norm": 15.992047309875488,
      "learning_rate": 9.009962640099626e-07,
      "loss": 0.8628,
      "step": 2895
    },
    {
      "epoch": 0.045147781142238864,
      "grad_norm": 4.682984352111816,
      "learning_rate": 9.025529265255293e-07,
      "loss": 0.876,
      "step": 2900
    },
    {
      "epoch": 0.04522562214420824,
      "grad_norm": 3.694166421890259,
      "learning_rate": 9.041095890410958e-07,
      "loss": 0.7707,
      "step": 2905
    },
    {
      "epoch": 0.04530346314617762,
      "grad_norm": 6.382852077484131,
      "learning_rate": 9.056662515566625e-07,
      "loss": 0.8372,
      "step": 2910
    },
    {
      "epoch": 0.04538130414814699,
      "grad_norm": 21.609174728393555,
      "learning_rate": 9.07222914072229e-07,
      "loss": 0.9099,
      "step": 2915
    },
    {
      "epoch": 0.04545914515011637,
      "grad_norm": 2.4359121322631836,
      "learning_rate": 9.087795765877957e-07,
      "loss": 0.9096,
      "step": 2920
    },
    {
      "epoch": 0.04553698615208575,
      "grad_norm": 3.6651458740234375,
      "learning_rate": 9.103362391033623e-07,
      "loss": 0.9408,
      "step": 2925
    },
    {
      "epoch": 0.045614827154055126,
      "grad_norm": 5.385332107543945,
      "learning_rate": 9.11892901618929e-07,
      "loss": 0.9312,
      "step": 2930
    },
    {
      "epoch": 0.045692668156024505,
      "grad_norm": 3.9548020362854004,
      "learning_rate": 9.134495641344956e-07,
      "loss": 0.8158,
      "step": 2935
    },
    {
      "epoch": 0.04577050915799388,
      "grad_norm": 2.7402524948120117,
      "learning_rate": 9.150062266500622e-07,
      "loss": 0.7451,
      "step": 2940
    },
    {
      "epoch": 0.04584835015996326,
      "grad_norm": 6.259926795959473,
      "learning_rate": 9.165628891656288e-07,
      "loss": 0.7399,
      "step": 2945
    },
    {
      "epoch": 0.04592619116193264,
      "grad_norm": 2.590571403503418,
      "learning_rate": 9.181195516811955e-07,
      "loss": 0.9088,
      "step": 2950
    },
    {
      "epoch": 0.04600403216390201,
      "grad_norm": 4.331900596618652,
      "learning_rate": 9.196762141967621e-07,
      "loss": 0.8636,
      "step": 2955
    },
    {
      "epoch": 0.04608187316587139,
      "grad_norm": 5.567667007446289,
      "learning_rate": 9.212328767123288e-07,
      "loss": 0.9059,
      "step": 2960
    },
    {
      "epoch": 0.04615971416784077,
      "grad_norm": 4.826610565185547,
      "learning_rate": 9.227895392278954e-07,
      "loss": 0.7942,
      "step": 2965
    },
    {
      "epoch": 0.046237555169810145,
      "grad_norm": 7.561775207519531,
      "learning_rate": 9.24346201743462e-07,
      "loss": 0.8461,
      "step": 2970
    },
    {
      "epoch": 0.04631539617177952,
      "grad_norm": 4.251841068267822,
      "learning_rate": 9.259028642590286e-07,
      "loss": 0.8541,
      "step": 2975
    },
    {
      "epoch": 0.0463932371737489,
      "grad_norm": 5.157746315002441,
      "learning_rate": 9.274595267745953e-07,
      "loss": 0.8863,
      "step": 2980
    },
    {
      "epoch": 0.04647107817571828,
      "grad_norm": 4.906675815582275,
      "learning_rate": 9.290161892901619e-07,
      "loss": 0.8432,
      "step": 2985
    },
    {
      "epoch": 0.04654891917768766,
      "grad_norm": 4.339780807495117,
      "learning_rate": 9.305728518057285e-07,
      "loss": 0.8448,
      "step": 2990
    },
    {
      "epoch": 0.046626760179657036,
      "grad_norm": 7.889379024505615,
      "learning_rate": 9.32129514321295e-07,
      "loss": 0.9431,
      "step": 2995
    },
    {
      "epoch": 0.04670460118162641,
      "grad_norm": 3.7697620391845703,
      "learning_rate": 9.336861768368617e-07,
      "loss": 0.8569,
      "step": 3000
    },
    {
      "epoch": 0.046782442183595785,
      "grad_norm": 8.277153968811035,
      "learning_rate": 9.352428393524283e-07,
      "loss": 0.7687,
      "step": 3005
    },
    {
      "epoch": 0.04686028318556516,
      "grad_norm": 6.6820149421691895,
      "learning_rate": 9.36799501867995e-07,
      "loss": 0.8283,
      "step": 3010
    },
    {
      "epoch": 0.04693812418753454,
      "grad_norm": 6.457581996917725,
      "learning_rate": 9.383561643835616e-07,
      "loss": 0.8447,
      "step": 3015
    },
    {
      "epoch": 0.04701596518950392,
      "grad_norm": 8.55042552947998,
      "learning_rate": 9.399128268991282e-07,
      "loss": 0.8986,
      "step": 3020
    },
    {
      "epoch": 0.0470938061914733,
      "grad_norm": 8.297921180725098,
      "learning_rate": 9.414694894146948e-07,
      "loss": 0.9012,
      "step": 3025
    },
    {
      "epoch": 0.047171647193442676,
      "grad_norm": 7.09883975982666,
      "learning_rate": 9.430261519302615e-07,
      "loss": 0.7454,
      "step": 3030
    },
    {
      "epoch": 0.047249488195412054,
      "grad_norm": 8.166378021240234,
      "learning_rate": 9.445828144458281e-07,
      "loss": 0.8405,
      "step": 3035
    },
    {
      "epoch": 0.04732732919738143,
      "grad_norm": 4.509795665740967,
      "learning_rate": 9.461394769613948e-07,
      "loss": 0.8076,
      "step": 3040
    },
    {
      "epoch": 0.0474051701993508,
      "grad_norm": 8.811022758483887,
      "learning_rate": 9.476961394769614e-07,
      "loss": 0.8387,
      "step": 3045
    },
    {
      "epoch": 0.04748301120132018,
      "grad_norm": 3.122080087661743,
      "learning_rate": 9.49252801992528e-07,
      "loss": 0.7069,
      "step": 3050
    },
    {
      "epoch": 0.04756085220328956,
      "grad_norm": 6.84942626953125,
      "learning_rate": 9.508094645080946e-07,
      "loss": 0.7412,
      "step": 3055
    },
    {
      "epoch": 0.04763869320525894,
      "grad_norm": 7.2728424072265625,
      "learning_rate": 9.523661270236613e-07,
      "loss": 0.8156,
      "step": 3060
    },
    {
      "epoch": 0.047716534207228316,
      "grad_norm": 4.008538246154785,
      "learning_rate": 9.539227895392278e-07,
      "loss": 0.8618,
      "step": 3065
    },
    {
      "epoch": 0.047794375209197694,
      "grad_norm": 3.9167230129241943,
      "learning_rate": 9.554794520547946e-07,
      "loss": 0.8026,
      "step": 3070
    },
    {
      "epoch": 0.04787221621116707,
      "grad_norm": 3.683629274368286,
      "learning_rate": 9.570361145703611e-07,
      "loss": 0.9918,
      "step": 3075
    },
    {
      "epoch": 0.04795005721313645,
      "grad_norm": 4.745260238647461,
      "learning_rate": 9.585927770859277e-07,
      "loss": 0.8148,
      "step": 3080
    },
    {
      "epoch": 0.04802789821510582,
      "grad_norm": 2.839996814727783,
      "learning_rate": 9.601494396014944e-07,
      "loss": 0.8176,
      "step": 3085
    },
    {
      "epoch": 0.0481057392170752,
      "grad_norm": 5.715896129608154,
      "learning_rate": 9.61706102117061e-07,
      "loss": 0.9266,
      "step": 3090
    },
    {
      "epoch": 0.04818358021904458,
      "grad_norm": 3.44376540184021,
      "learning_rate": 9.632627646326275e-07,
      "loss": 0.7851,
      "step": 3095
    },
    {
      "epoch": 0.048261421221013956,
      "grad_norm": 4.266874313354492,
      "learning_rate": 9.648194271481943e-07,
      "loss": 0.8701,
      "step": 3100
    },
    {
      "epoch": 0.048339262222983334,
      "grad_norm": 3.6180949211120605,
      "learning_rate": 9.663760896637608e-07,
      "loss": 0.9206,
      "step": 3105
    },
    {
      "epoch": 0.04841710322495271,
      "grad_norm": 4.103425979614258,
      "learning_rate": 9.679327521793276e-07,
      "loss": 0.9025,
      "step": 3110
    },
    {
      "epoch": 0.04849494422692209,
      "grad_norm": 3.339601516723633,
      "learning_rate": 9.69489414694894e-07,
      "loss": 0.7176,
      "step": 3115
    },
    {
      "epoch": 0.04857278522889147,
      "grad_norm": 5.723580360412598,
      "learning_rate": 9.710460772104606e-07,
      "loss": 0.7712,
      "step": 3120
    },
    {
      "epoch": 0.04865062623086085,
      "grad_norm": 9.84174919128418,
      "learning_rate": 9.726027397260274e-07,
      "loss": 0.8301,
      "step": 3125
    },
    {
      "epoch": 0.04872846723283022,
      "grad_norm": 4.004911422729492,
      "learning_rate": 9.74159402241594e-07,
      "loss": 0.8232,
      "step": 3130
    },
    {
      "epoch": 0.048806308234799596,
      "grad_norm": 4.22821044921875,
      "learning_rate": 9.757160647571607e-07,
      "loss": 0.8408,
      "step": 3135
    },
    {
      "epoch": 0.048884149236768974,
      "grad_norm": 3.268477439880371,
      "learning_rate": 9.772727272727273e-07,
      "loss": 0.854,
      "step": 3140
    },
    {
      "epoch": 0.04896199023873835,
      "grad_norm": 3.3312723636627197,
      "learning_rate": 9.788293897882938e-07,
      "loss": 0.7584,
      "step": 3145
    },
    {
      "epoch": 0.04903983124070773,
      "grad_norm": 2.6721420288085938,
      "learning_rate": 9.803860523038606e-07,
      "loss": 0.7101,
      "step": 3150
    },
    {
      "epoch": 0.04911767224267711,
      "grad_norm": 3.1036221981048584,
      "learning_rate": 9.81942714819427e-07,
      "loss": 0.8943,
      "step": 3155
    },
    {
      "epoch": 0.04919551324464649,
      "grad_norm": 4.581750869750977,
      "learning_rate": 9.834993773349939e-07,
      "loss": 0.956,
      "step": 3160
    },
    {
      "epoch": 0.049273354246615865,
      "grad_norm": 5.273120880126953,
      "learning_rate": 9.850560398505604e-07,
      "loss": 0.8077,
      "step": 3165
    },
    {
      "epoch": 0.04935119524858524,
      "grad_norm": 7.310013771057129,
      "learning_rate": 9.86612702366127e-07,
      "loss": 0.8506,
      "step": 3170
    },
    {
      "epoch": 0.049429036250554614,
      "grad_norm": 5.949068069458008,
      "learning_rate": 9.881693648816935e-07,
      "loss": 0.8132,
      "step": 3175
    },
    {
      "epoch": 0.04950687725252399,
      "grad_norm": 5.228186130523682,
      "learning_rate": 9.897260273972602e-07,
      "loss": 0.9303,
      "step": 3180
    },
    {
      "epoch": 0.04958471825449337,
      "grad_norm": 4.407035827636719,
      "learning_rate": 9.912826899128268e-07,
      "loss": 0.8184,
      "step": 3185
    },
    {
      "epoch": 0.04966255925646275,
      "grad_norm": 4.605864524841309,
      "learning_rate": 9.928393524283936e-07,
      "loss": 0.9336,
      "step": 3190
    },
    {
      "epoch": 0.04974040025843213,
      "grad_norm": 3.0708847045898438,
      "learning_rate": 9.9439601494396e-07,
      "loss": 0.8725,
      "step": 3195
    },
    {
      "epoch": 0.049818241260401505,
      "grad_norm": 3.3742926120758057,
      "learning_rate": 9.959526774595266e-07,
      "loss": 0.8121,
      "step": 3200
    },
    {
      "epoch": 0.04989608226237088,
      "grad_norm": 2.685382843017578,
      "learning_rate": 9.975093399750934e-07,
      "loss": 0.7798,
      "step": 3205
    },
    {
      "epoch": 0.04997392326434026,
      "grad_norm": 4.932633876800537,
      "learning_rate": 9.9906600249066e-07,
      "loss": 0.8492,
      "step": 3210
    },
    {
      "epoch": 0.05005176426630963,
      "grad_norm": 8.489307403564453,
      "learning_rate": 9.999672243981579e-07,
      "loss": 0.8355,
      "step": 3215
    },
    {
      "epoch": 0.05012960526827901,
      "grad_norm": 4.679005146026611,
      "learning_rate": 9.99885285393553e-07,
      "loss": 0.9012,
      "step": 3220
    },
    {
      "epoch": 0.05020744627024839,
      "grad_norm": 6.65717887878418,
      "learning_rate": 9.99803346388948e-07,
      "loss": 1.0277,
      "step": 3225
    },
    {
      "epoch": 0.05028528727221777,
      "grad_norm": 5.373363494873047,
      "learning_rate": 9.99721407384343e-07,
      "loss": 1.0007,
      "step": 3230
    },
    {
      "epoch": 0.050363128274187145,
      "grad_norm": 3.9103312492370605,
      "learning_rate": 9.996394683797382e-07,
      "loss": 0.8015,
      "step": 3235
    },
    {
      "epoch": 0.05044096927615652,
      "grad_norm": 6.019688606262207,
      "learning_rate": 9.995575293751332e-07,
      "loss": 0.8575,
      "step": 3240
    },
    {
      "epoch": 0.0505188102781259,
      "grad_norm": 17.253416061401367,
      "learning_rate": 9.99475590370528e-07,
      "loss": 0.7818,
      "step": 3245
    },
    {
      "epoch": 0.05059665128009528,
      "grad_norm": 9.291438102722168,
      "learning_rate": 9.993936513659232e-07,
      "loss": 0.9093,
      "step": 3250
    },
    {
      "epoch": 0.05067449228206466,
      "grad_norm": 4.7031121253967285,
      "learning_rate": 9.993117123613182e-07,
      "loss": 0.792,
      "step": 3255
    },
    {
      "epoch": 0.05075233328403403,
      "grad_norm": 3.9141600131988525,
      "learning_rate": 9.992297733567131e-07,
      "loss": 0.8803,
      "step": 3260
    },
    {
      "epoch": 0.05083017428600341,
      "grad_norm": 5.731180191040039,
      "learning_rate": 9.991478343521082e-07,
      "loss": 0.9267,
      "step": 3265
    },
    {
      "epoch": 0.050908015287972785,
      "grad_norm": 4.963929653167725,
      "learning_rate": 9.990658953475033e-07,
      "loss": 0.8762,
      "step": 3270
    },
    {
      "epoch": 0.050985856289942164,
      "grad_norm": 5.126701831817627,
      "learning_rate": 9.989839563428983e-07,
      "loss": 0.9762,
      "step": 3275
    },
    {
      "epoch": 0.05106369729191154,
      "grad_norm": 5.1071953773498535,
      "learning_rate": 9.989020173382934e-07,
      "loss": 0.9504,
      "step": 3280
    },
    {
      "epoch": 0.05114153829388092,
      "grad_norm": 4.061114311218262,
      "learning_rate": 9.988200783336883e-07,
      "loss": 0.821,
      "step": 3285
    },
    {
      "epoch": 0.0512193792958503,
      "grad_norm": 3.604483127593994,
      "learning_rate": 9.987381393290833e-07,
      "loss": 0.9565,
      "step": 3290
    },
    {
      "epoch": 0.051297220297819676,
      "grad_norm": 4.070693016052246,
      "learning_rate": 9.986562003244784e-07,
      "loss": 0.7469,
      "step": 3295
    },
    {
      "epoch": 0.051375061299789054,
      "grad_norm": 3.4125092029571533,
      "learning_rate": 9.985742613198735e-07,
      "loss": 0.7926,
      "step": 3300
    },
    {
      "epoch": 0.051452902301758426,
      "grad_norm": 7.950231075286865,
      "learning_rate": 9.984923223152686e-07,
      "loss": 0.8422,
      "step": 3305
    },
    {
      "epoch": 0.051530743303727804,
      "grad_norm": 3.185955762863159,
      "learning_rate": 9.984103833106634e-07,
      "loss": 0.7959,
      "step": 3310
    },
    {
      "epoch": 0.05160858430569718,
      "grad_norm": 4.626750946044922,
      "learning_rate": 9.983284443060585e-07,
      "loss": 0.9932,
      "step": 3315
    },
    {
      "epoch": 0.05168642530766656,
      "grad_norm": 2.5758249759674072,
      "learning_rate": 9.982465053014536e-07,
      "loss": 0.7739,
      "step": 3320
    },
    {
      "epoch": 0.05176426630963594,
      "grad_norm": 3.6274349689483643,
      "learning_rate": 9.981645662968484e-07,
      "loss": 0.8351,
      "step": 3325
    },
    {
      "epoch": 0.051842107311605316,
      "grad_norm": 3.520857572555542,
      "learning_rate": 9.980826272922435e-07,
      "loss": 0.8815,
      "step": 3330
    },
    {
      "epoch": 0.051919948313574694,
      "grad_norm": 4.665640354156494,
      "learning_rate": 9.980006882876386e-07,
      "loss": 0.8575,
      "step": 3335
    },
    {
      "epoch": 0.05199778931554407,
      "grad_norm": 5.597052574157715,
      "learning_rate": 9.979187492830337e-07,
      "loss": 0.8373,
      "step": 3340
    },
    {
      "epoch": 0.052075630317513444,
      "grad_norm": 5.660586357116699,
      "learning_rate": 9.978368102784287e-07,
      "loss": 0.9164,
      "step": 3345
    },
    {
      "epoch": 0.05215347131948282,
      "grad_norm": 11.376925468444824,
      "learning_rate": 9.977548712738238e-07,
      "loss": 0.8779,
      "step": 3350
    },
    {
      "epoch": 0.0522313123214522,
      "grad_norm": 3.930678606033325,
      "learning_rate": 9.976729322692187e-07,
      "loss": 0.8638,
      "step": 3355
    },
    {
      "epoch": 0.05230915332342158,
      "grad_norm": 4.059145450592041,
      "learning_rate": 9.975909932646138e-07,
      "loss": 0.7965,
      "step": 3360
    },
    {
      "epoch": 0.052386994325390956,
      "grad_norm": 4.585720539093018,
      "learning_rate": 9.975090542600088e-07,
      "loss": 0.8034,
      "step": 3365
    },
    {
      "epoch": 0.052464835327360335,
      "grad_norm": 5.015563488006592,
      "learning_rate": 9.974271152554037e-07,
      "loss": 0.8109,
      "step": 3370
    },
    {
      "epoch": 0.05254267632932971,
      "grad_norm": 3.2969090938568115,
      "learning_rate": 9.973451762507988e-07,
      "loss": 0.9502,
      "step": 3375
    },
    {
      "epoch": 0.05262051733129909,
      "grad_norm": 3.2702388763427734,
      "learning_rate": 9.972632372461938e-07,
      "loss": 0.8148,
      "step": 3380
    },
    {
      "epoch": 0.05269835833326847,
      "grad_norm": 2.95889949798584,
      "learning_rate": 9.97181298241589e-07,
      "loss": 0.8935,
      "step": 3385
    },
    {
      "epoch": 0.05277619933523784,
      "grad_norm": 5.157326698303223,
      "learning_rate": 9.97099359236984e-07,
      "loss": 0.9001,
      "step": 3390
    },
    {
      "epoch": 0.05285404033720722,
      "grad_norm": 3.6577107906341553,
      "learning_rate": 9.97017420232379e-07,
      "loss": 0.7983,
      "step": 3395
    },
    {
      "epoch": 0.0529318813391766,
      "grad_norm": 2.539867401123047,
      "learning_rate": 9.969354812277741e-07,
      "loss": 0.732,
      "step": 3400
    },
    {
      "epoch": 0.053009722341145975,
      "grad_norm": 6.6847076416015625,
      "learning_rate": 9.96853542223169e-07,
      "loss": 0.8909,
      "step": 3405
    },
    {
      "epoch": 0.05308756334311535,
      "grad_norm": 3.6293387413024902,
      "learning_rate": 9.96771603218564e-07,
      "loss": 0.7757,
      "step": 3410
    },
    {
      "epoch": 0.05316540434508473,
      "grad_norm": 9.500846862792969,
      "learning_rate": 9.966896642139592e-07,
      "loss": 0.8709,
      "step": 3415
    },
    {
      "epoch": 0.05324324534705411,
      "grad_norm": 8.317655563354492,
      "learning_rate": 9.96607725209354e-07,
      "loss": 0.8833,
      "step": 3420
    },
    {
      "epoch": 0.05332108634902349,
      "grad_norm": 6.386698246002197,
      "learning_rate": 9.96525786204749e-07,
      "loss": 0.9136,
      "step": 3425
    },
    {
      "epoch": 0.05339892735099286,
      "grad_norm": 3.567600965499878,
      "learning_rate": 9.964438472001442e-07,
      "loss": 0.8465,
      "step": 3430
    },
    {
      "epoch": 0.05347676835296224,
      "grad_norm": 7.062701225280762,
      "learning_rate": 9.963619081955392e-07,
      "loss": 0.8179,
      "step": 3435
    },
    {
      "epoch": 0.053554609354931615,
      "grad_norm": 3.983492851257324,
      "learning_rate": 9.962799691909343e-07,
      "loss": 0.899,
      "step": 3440
    },
    {
      "epoch": 0.05363245035690099,
      "grad_norm": 7.150521278381348,
      "learning_rate": 9.961980301863292e-07,
      "loss": 0.7949,
      "step": 3445
    },
    {
      "epoch": 0.05371029135887037,
      "grad_norm": 5.3643107414245605,
      "learning_rate": 9.961160911817243e-07,
      "loss": 0.893,
      "step": 3450
    },
    {
      "epoch": 0.05378813236083975,
      "grad_norm": 7.8569440841674805,
      "learning_rate": 9.960341521771193e-07,
      "loss": 0.7597,
      "step": 3455
    },
    {
      "epoch": 0.05386597336280913,
      "grad_norm": 2.990384817123413,
      "learning_rate": 9.959522131725144e-07,
      "loss": 0.8968,
      "step": 3460
    },
    {
      "epoch": 0.053943814364778506,
      "grad_norm": 11.023333549499512,
      "learning_rate": 9.958702741679093e-07,
      "loss": 0.8577,
      "step": 3465
    },
    {
      "epoch": 0.054021655366747884,
      "grad_norm": 3.8599610328674316,
      "learning_rate": 9.957883351633043e-07,
      "loss": 0.8187,
      "step": 3470
    },
    {
      "epoch": 0.054099496368717255,
      "grad_norm": 4.514223575592041,
      "learning_rate": 9.957063961586994e-07,
      "loss": 0.8948,
      "step": 3475
    },
    {
      "epoch": 0.05417733737068663,
      "grad_norm": 5.561735153198242,
      "learning_rate": 9.956244571540945e-07,
      "loss": 0.7144,
      "step": 3480
    },
    {
      "epoch": 0.05425517837265601,
      "grad_norm": 2.5921874046325684,
      "learning_rate": 9.955425181494896e-07,
      "loss": 0.8599,
      "step": 3485
    },
    {
      "epoch": 0.05433301937462539,
      "grad_norm": 4.871161937713623,
      "learning_rate": 9.954605791448844e-07,
      "loss": 0.9644,
      "step": 3490
    },
    {
      "epoch": 0.05441086037659477,
      "grad_norm": 6.471960544586182,
      "learning_rate": 9.953786401402795e-07,
      "loss": 0.764,
      "step": 3495
    },
    {
      "epoch": 0.054488701378564146,
      "grad_norm": 5.133829593658447,
      "learning_rate": 9.952967011356746e-07,
      "loss": 0.8484,
      "step": 3500
    },
    {
      "epoch": 0.054566542380533524,
      "grad_norm": 15.294747352600098,
      "learning_rate": 9.952147621310697e-07,
      "loss": 0.9278,
      "step": 3505
    },
    {
      "epoch": 0.0546443833825029,
      "grad_norm": 4.0458526611328125,
      "learning_rate": 9.951328231264645e-07,
      "loss": 0.8015,
      "step": 3510
    },
    {
      "epoch": 0.05472222438447228,
      "grad_norm": 3.96840238571167,
      "learning_rate": 9.950508841218596e-07,
      "loss": 0.9182,
      "step": 3515
    },
    {
      "epoch": 0.05480006538644165,
      "grad_norm": 3.493230104446411,
      "learning_rate": 9.949689451172547e-07,
      "loss": 0.7351,
      "step": 3520
    },
    {
      "epoch": 0.05487790638841103,
      "grad_norm": 6.453081130981445,
      "learning_rate": 9.948870061126497e-07,
      "loss": 0.6706,
      "step": 3525
    },
    {
      "epoch": 0.05495574739038041,
      "grad_norm": 4.883228302001953,
      "learning_rate": 9.948050671080446e-07,
      "loss": 0.8926,
      "step": 3530
    },
    {
      "epoch": 0.055033588392349786,
      "grad_norm": 8.88487434387207,
      "learning_rate": 9.947231281034397e-07,
      "loss": 0.7815,
      "step": 3535
    },
    {
      "epoch": 0.055111429394319164,
      "grad_norm": 3.5414915084838867,
      "learning_rate": 9.946411890988348e-07,
      "loss": 0.8066,
      "step": 3540
    },
    {
      "epoch": 0.05518927039628854,
      "grad_norm": 3.3924477100372314,
      "learning_rate": 9.945592500942298e-07,
      "loss": 0.8942,
      "step": 3545
    },
    {
      "epoch": 0.05526711139825792,
      "grad_norm": 8.606155395507812,
      "learning_rate": 9.94477311089625e-07,
      "loss": 0.8062,
      "step": 3550
    },
    {
      "epoch": 0.0553449524002273,
      "grad_norm": 3.5798611640930176,
      "learning_rate": 9.9439537208502e-07,
      "loss": 0.8145,
      "step": 3555
    },
    {
      "epoch": 0.05542279340219667,
      "grad_norm": 4.816424369812012,
      "learning_rate": 9.943134330804148e-07,
      "loss": 0.9767,
      "step": 3560
    },
    {
      "epoch": 0.05550063440416605,
      "grad_norm": 3.161212682723999,
      "learning_rate": 9.9423149407581e-07,
      "loss": 0.7526,
      "step": 3565
    },
    {
      "epoch": 0.055578475406135426,
      "grad_norm": 5.3241143226623535,
      "learning_rate": 9.94149555071205e-07,
      "loss": 0.8756,
      "step": 3570
    },
    {
      "epoch": 0.055656316408104804,
      "grad_norm": 4.702089786529541,
      "learning_rate": 9.940676160665999e-07,
      "loss": 0.7844,
      "step": 3575
    },
    {
      "epoch": 0.05573415741007418,
      "grad_norm": 3.6324615478515625,
      "learning_rate": 9.93985677061995e-07,
      "loss": 0.9757,
      "step": 3580
    },
    {
      "epoch": 0.05581199841204356,
      "grad_norm": 5.574779510498047,
      "learning_rate": 9.9390373805739e-07,
      "loss": 0.8368,
      "step": 3585
    },
    {
      "epoch": 0.05588983941401294,
      "grad_norm": 3.3760433197021484,
      "learning_rate": 9.93821799052785e-07,
      "loss": 0.9753,
      "step": 3590
    },
    {
      "epoch": 0.05596768041598232,
      "grad_norm": 3.6447086334228516,
      "learning_rate": 9.937398600481802e-07,
      "loss": 0.8654,
      "step": 3595
    },
    {
      "epoch": 0.056045521417951695,
      "grad_norm": 10.935750007629395,
      "learning_rate": 9.936579210435752e-07,
      "loss": 0.8504,
      "step": 3600
    },
    {
      "epoch": 0.056123362419921066,
      "grad_norm": 5.356347560882568,
      "learning_rate": 9.9357598203897e-07,
      "loss": 0.8439,
      "step": 3605
    },
    {
      "epoch": 0.056201203421890444,
      "grad_norm": 7.737555027008057,
      "learning_rate": 9.934940430343652e-07,
      "loss": 0.8997,
      "step": 3610
    },
    {
      "epoch": 0.05627904442385982,
      "grad_norm": 4.059571266174316,
      "learning_rate": 9.934121040297602e-07,
      "loss": 0.736,
      "step": 3615
    },
    {
      "epoch": 0.0563568854258292,
      "grad_norm": 10.28212833404541,
      "learning_rate": 9.933301650251551e-07,
      "loss": 0.8219,
      "step": 3620
    },
    {
      "epoch": 0.05643472642779858,
      "grad_norm": 7.522468090057373,
      "learning_rate": 9.932482260205502e-07,
      "loss": 0.7058,
      "step": 3625
    },
    {
      "epoch": 0.05651256742976796,
      "grad_norm": 4.0811872482299805,
      "learning_rate": 9.931662870159453e-07,
      "loss": 0.8334,
      "step": 3630
    },
    {
      "epoch": 0.056590408431737335,
      "grad_norm": 2.533539295196533,
      "learning_rate": 9.930843480113403e-07,
      "loss": 0.8185,
      "step": 3635
    },
    {
      "epoch": 0.05666824943370671,
      "grad_norm": 2.272587776184082,
      "learning_rate": 9.930024090067354e-07,
      "loss": 0.8294,
      "step": 3640
    },
    {
      "epoch": 0.05674609043567609,
      "grad_norm": 4.402963638305664,
      "learning_rate": 9.929204700021305e-07,
      "loss": 0.8253,
      "step": 3645
    },
    {
      "epoch": 0.05682393143764546,
      "grad_norm": 4.450977802276611,
      "learning_rate": 9.928385309975253e-07,
      "loss": 0.7287,
      "step": 3650
    },
    {
      "epoch": 0.05690177243961484,
      "grad_norm": 4.995216369628906,
      "learning_rate": 9.927565919929204e-07,
      "loss": 0.7744,
      "step": 3655
    },
    {
      "epoch": 0.05697961344158422,
      "grad_norm": 4.42352294921875,
      "learning_rate": 9.926746529883155e-07,
      "loss": 0.8216,
      "step": 3660
    },
    {
      "epoch": 0.0570574544435536,
      "grad_norm": 5.005922317504883,
      "learning_rate": 9.925927139837106e-07,
      "loss": 0.88,
      "step": 3665
    },
    {
      "epoch": 0.057135295445522975,
      "grad_norm": 4.319427013397217,
      "learning_rate": 9.925107749791054e-07,
      "loss": 0.9386,
      "step": 3670
    },
    {
      "epoch": 0.05721313644749235,
      "grad_norm": 4.61904239654541,
      "learning_rate": 9.924288359745005e-07,
      "loss": 0.8248,
      "step": 3675
    },
    {
      "epoch": 0.05729097744946173,
      "grad_norm": 3.656996250152588,
      "learning_rate": 9.923468969698956e-07,
      "loss": 0.8898,
      "step": 3680
    },
    {
      "epoch": 0.05736881845143111,
      "grad_norm": 10.73847484588623,
      "learning_rate": 9.922649579652907e-07,
      "loss": 0.7295,
      "step": 3685
    },
    {
      "epoch": 0.05744665945340048,
      "grad_norm": 3.2956910133361816,
      "learning_rate": 9.921830189606855e-07,
      "loss": 0.7937,
      "step": 3690
    },
    {
      "epoch": 0.05752450045536986,
      "grad_norm": 3.310476541519165,
      "learning_rate": 9.921010799560806e-07,
      "loss": 0.7597,
      "step": 3695
    },
    {
      "epoch": 0.05760234145733924,
      "grad_norm": 6.073892116546631,
      "learning_rate": 9.920191409514757e-07,
      "loss": 0.775,
      "step": 3700
    },
    {
      "epoch": 0.057680182459308615,
      "grad_norm": 4.651096820831299,
      "learning_rate": 9.919372019468707e-07,
      "loss": 0.9085,
      "step": 3705
    },
    {
      "epoch": 0.057758023461277994,
      "grad_norm": 5.112009048461914,
      "learning_rate": 9.918552629422658e-07,
      "loss": 0.854,
      "step": 3710
    },
    {
      "epoch": 0.05783586446324737,
      "grad_norm": 3.9226460456848145,
      "learning_rate": 9.917733239376607e-07,
      "loss": 0.8815,
      "step": 3715
    },
    {
      "epoch": 0.05791370546521675,
      "grad_norm": 5.9531707763671875,
      "learning_rate": 9.916913849330558e-07,
      "loss": 0.8794,
      "step": 3720
    },
    {
      "epoch": 0.05799154646718613,
      "grad_norm": 7.749881744384766,
      "learning_rate": 9.916094459284508e-07,
      "loss": 0.8347,
      "step": 3725
    },
    {
      "epoch": 0.058069387469155506,
      "grad_norm": 3.2161874771118164,
      "learning_rate": 9.915275069238457e-07,
      "loss": 0.8297,
      "step": 3730
    },
    {
      "epoch": 0.05814722847112488,
      "grad_norm": 3.4381978511810303,
      "learning_rate": 9.914455679192408e-07,
      "loss": 0.8016,
      "step": 3735
    },
    {
      "epoch": 0.058225069473094256,
      "grad_norm": 6.175289630889893,
      "learning_rate": 9.913636289146358e-07,
      "loss": 0.7378,
      "step": 3740
    },
    {
      "epoch": 0.058302910475063634,
      "grad_norm": 7.808245658874512,
      "learning_rate": 9.91281689910031e-07,
      "loss": 0.8631,
      "step": 3745
    },
    {
      "epoch": 0.05838075147703301,
      "grad_norm": 8.13048267364502,
      "learning_rate": 9.91199750905426e-07,
      "loss": 0.7241,
      "step": 3750
    },
    {
      "epoch": 0.05845859247900239,
      "grad_norm": 14.47769546508789,
      "learning_rate": 9.91117811900821e-07,
      "loss": 0.7604,
      "step": 3755
    },
    {
      "epoch": 0.05853643348097177,
      "grad_norm": 13.544578552246094,
      "learning_rate": 9.91035872896216e-07,
      "loss": 0.9168,
      "step": 3760
    },
    {
      "epoch": 0.058614274482941146,
      "grad_norm": 3.012338638305664,
      "learning_rate": 9.90953933891611e-07,
      "loss": 0.8438,
      "step": 3765
    },
    {
      "epoch": 0.058692115484910524,
      "grad_norm": 8.543879508972168,
      "learning_rate": 9.90871994887006e-07,
      "loss": 0.8027,
      "step": 3770
    },
    {
      "epoch": 0.058769956486879896,
      "grad_norm": 3.5552265644073486,
      "learning_rate": 9.907900558824012e-07,
      "loss": 0.9394,
      "step": 3775
    },
    {
      "epoch": 0.058847797488849274,
      "grad_norm": 2.7634129524230957,
      "learning_rate": 9.90708116877796e-07,
      "loss": 0.8544,
      "step": 3780
    },
    {
      "epoch": 0.05892563849081865,
      "grad_norm": 4.050414085388184,
      "learning_rate": 9.90626177873191e-07,
      "loss": 0.8405,
      "step": 3785
    },
    {
      "epoch": 0.05900347949278803,
      "grad_norm": 3.3038461208343506,
      "learning_rate": 9.905442388685862e-07,
      "loss": 0.763,
      "step": 3790
    },
    {
      "epoch": 0.05908132049475741,
      "grad_norm": 5.79196834564209,
      "learning_rate": 9.904622998639812e-07,
      "loss": 0.8174,
      "step": 3795
    },
    {
      "epoch": 0.059159161496726786,
      "grad_norm": 4.359936714172363,
      "learning_rate": 9.903803608593763e-07,
      "loss": 0.8229,
      "step": 3800
    },
    {
      "epoch": 0.059237002498696165,
      "grad_norm": 6.546017169952393,
      "learning_rate": 9.902984218547714e-07,
      "loss": 0.835,
      "step": 3805
    },
    {
      "epoch": 0.05931484350066554,
      "grad_norm": 6.203246593475342,
      "learning_rate": 9.902164828501663e-07,
      "loss": 0.9859,
      "step": 3810
    },
    {
      "epoch": 0.05939268450263492,
      "grad_norm": 3.92028546333313,
      "learning_rate": 9.901345438455613e-07,
      "loss": 0.84,
      "step": 3815
    },
    {
      "epoch": 0.05947052550460429,
      "grad_norm": 4.098803520202637,
      "learning_rate": 9.900526048409564e-07,
      "loss": 0.8088,
      "step": 3820
    },
    {
      "epoch": 0.05954836650657367,
      "grad_norm": 4.060965061187744,
      "learning_rate": 9.899706658363513e-07,
      "loss": 0.8048,
      "step": 3825
    },
    {
      "epoch": 0.05962620750854305,
      "grad_norm": 7.130313873291016,
      "learning_rate": 9.898887268317463e-07,
      "loss": 0.991,
      "step": 3830
    },
    {
      "epoch": 0.05970404851051243,
      "grad_norm": 4.355027198791504,
      "learning_rate": 9.898067878271414e-07,
      "loss": 0.9168,
      "step": 3835
    },
    {
      "epoch": 0.059781889512481805,
      "grad_norm": 4.409844398498535,
      "learning_rate": 9.897248488225365e-07,
      "loss": 0.7811,
      "step": 3840
    },
    {
      "epoch": 0.05985973051445118,
      "grad_norm": 4.593713283538818,
      "learning_rate": 9.896429098179316e-07,
      "loss": 0.9282,
      "step": 3845
    },
    {
      "epoch": 0.05993757151642056,
      "grad_norm": 3.813417911529541,
      "learning_rate": 9.895609708133266e-07,
      "loss": 0.8671,
      "step": 3850
    },
    {
      "epoch": 0.06001541251838994,
      "grad_norm": 9.554966926574707,
      "learning_rate": 9.894790318087215e-07,
      "loss": 0.8516,
      "step": 3855
    },
    {
      "epoch": 0.06009325352035932,
      "grad_norm": 3.616415500640869,
      "learning_rate": 9.893970928041166e-07,
      "loss": 0.8382,
      "step": 3860
    },
    {
      "epoch": 0.06017109452232869,
      "grad_norm": 3.379333019256592,
      "learning_rate": 9.893151537995117e-07,
      "loss": 0.9661,
      "step": 3865
    },
    {
      "epoch": 0.06024893552429807,
      "grad_norm": 2.6693906784057617,
      "learning_rate": 9.892332147949065e-07,
      "loss": 0.8133,
      "step": 3870
    },
    {
      "epoch": 0.060326776526267445,
      "grad_norm": 4.557685375213623,
      "learning_rate": 9.891512757903016e-07,
      "loss": 0.8617,
      "step": 3875
    },
    {
      "epoch": 0.06040461752823682,
      "grad_norm": 2.69423770904541,
      "learning_rate": 9.890693367856967e-07,
      "loss": 0.7904,
      "step": 3880
    },
    {
      "epoch": 0.0604824585302062,
      "grad_norm": 3.213026762008667,
      "learning_rate": 9.889873977810917e-07,
      "loss": 0.7852,
      "step": 3885
    },
    {
      "epoch": 0.06056029953217558,
      "grad_norm": 3.25534725189209,
      "learning_rate": 9.889054587764868e-07,
      "loss": 0.8165,
      "step": 3890
    },
    {
      "epoch": 0.06063814053414496,
      "grad_norm": 5.834784984588623,
      "learning_rate": 9.888235197718817e-07,
      "loss": 0.9304,
      "step": 3895
    },
    {
      "epoch": 0.060715981536114336,
      "grad_norm": 3.369537353515625,
      "learning_rate": 9.887415807672768e-07,
      "loss": 0.7562,
      "step": 3900
    },
    {
      "epoch": 0.06079382253808371,
      "grad_norm": 5.367571830749512,
      "learning_rate": 9.886596417626718e-07,
      "loss": 0.8158,
      "step": 3905
    },
    {
      "epoch": 0.060871663540053085,
      "grad_norm": 4.397671222686768,
      "learning_rate": 9.88577702758067e-07,
      "loss": 0.8699,
      "step": 3910
    },
    {
      "epoch": 0.06094950454202246,
      "grad_norm": 3.270768404006958,
      "learning_rate": 9.88495763753462e-07,
      "loss": 0.9022,
      "step": 3915
    },
    {
      "epoch": 0.06102734554399184,
      "grad_norm": 4.194687366485596,
      "learning_rate": 9.884138247488568e-07,
      "loss": 0.937,
      "step": 3920
    },
    {
      "epoch": 0.06110518654596122,
      "grad_norm": 3.5028905868530273,
      "learning_rate": 9.88331885744252e-07,
      "loss": 0.9853,
      "step": 3925
    },
    {
      "epoch": 0.0611830275479306,
      "grad_norm": 9.81811237335205,
      "learning_rate": 9.88249946739647e-07,
      "loss": 0.9332,
      "step": 3930
    },
    {
      "epoch": 0.061260868549899976,
      "grad_norm": 9.531314849853516,
      "learning_rate": 9.881680077350419e-07,
      "loss": 0.8402,
      "step": 3935
    },
    {
      "epoch": 0.061338709551869354,
      "grad_norm": 6.465907096862793,
      "learning_rate": 9.88086068730437e-07,
      "loss": 0.9443,
      "step": 3940
    },
    {
      "epoch": 0.06141655055383873,
      "grad_norm": 9.462715148925781,
      "learning_rate": 9.88004129725832e-07,
      "loss": 0.7544,
      "step": 3945
    },
    {
      "epoch": 0.0614943915558081,
      "grad_norm": 4.005988121032715,
      "learning_rate": 9.87922190721227e-07,
      "loss": 0.8655,
      "step": 3950
    },
    {
      "epoch": 0.06157223255777748,
      "grad_norm": 6.533730983734131,
      "learning_rate": 9.878402517166222e-07,
      "loss": 0.9202,
      "step": 3955
    },
    {
      "epoch": 0.06165007355974686,
      "grad_norm": 4.695230484008789,
      "learning_rate": 9.877583127120172e-07,
      "loss": 0.7832,
      "step": 3960
    },
    {
      "epoch": 0.06172791456171624,
      "grad_norm": 4.281477451324463,
      "learning_rate": 9.87676373707412e-07,
      "loss": 0.8885,
      "step": 3965
    },
    {
      "epoch": 0.061805755563685616,
      "grad_norm": 4.162761688232422,
      "learning_rate": 9.875944347028072e-07,
      "loss": 0.8782,
      "step": 3970
    },
    {
      "epoch": 0.061883596565654994,
      "grad_norm": 3.2788217067718506,
      "learning_rate": 9.875124956982022e-07,
      "loss": 0.901,
      "step": 3975
    },
    {
      "epoch": 0.06196143756762437,
      "grad_norm": 3.823699951171875,
      "learning_rate": 9.874305566935971e-07,
      "loss": 0.8811,
      "step": 3980
    },
    {
      "epoch": 0.06203927856959375,
      "grad_norm": 5.366037368774414,
      "learning_rate": 9.873486176889922e-07,
      "loss": 0.9297,
      "step": 3985
    },
    {
      "epoch": 0.06211711957156313,
      "grad_norm": 3.4064414501190186,
      "learning_rate": 9.872666786843873e-07,
      "loss": 0.8329,
      "step": 3990
    },
    {
      "epoch": 0.0621949605735325,
      "grad_norm": 6.189504146575928,
      "learning_rate": 9.871847396797823e-07,
      "loss": 0.8325,
      "step": 3995
    },
    {
      "epoch": 0.06227280157550188,
      "grad_norm": 2.825984001159668,
      "learning_rate": 9.871028006751774e-07,
      "loss": 0.7901,
      "step": 4000
    },
    {
      "epoch": 0.062350642577471256,
      "grad_norm": 3.610321521759033,
      "learning_rate": 9.870208616705725e-07,
      "loss": 0.7974,
      "step": 4005
    },
    {
      "epoch": 0.062428483579440634,
      "grad_norm": 4.4487128257751465,
      "learning_rate": 9.869389226659676e-07,
      "loss": 0.9165,
      "step": 4010
    },
    {
      "epoch": 0.06250632458141,
      "grad_norm": 3.201486110687256,
      "learning_rate": 9.868569836613624e-07,
      "loss": 0.9165,
      "step": 4015
    },
    {
      "epoch": 0.06258416558337938,
      "grad_norm": 6.013232231140137,
      "learning_rate": 9.867750446567575e-07,
      "loss": 0.7316,
      "step": 4020
    },
    {
      "epoch": 0.06266200658534876,
      "grad_norm": 4.562684535980225,
      "learning_rate": 9.866931056521526e-07,
      "loss": 0.8648,
      "step": 4025
    },
    {
      "epoch": 0.06273984758731814,
      "grad_norm": 3.915780544281006,
      "learning_rate": 9.866111666475474e-07,
      "loss": 0.7684,
      "step": 4030
    },
    {
      "epoch": 0.06281768858928752,
      "grad_norm": 13.098698616027832,
      "learning_rate": 9.865292276429425e-07,
      "loss": 0.8222,
      "step": 4035
    },
    {
      "epoch": 0.0628955295912569,
      "grad_norm": 5.85524320602417,
      "learning_rate": 9.864472886383376e-07,
      "loss": 0.8593,
      "step": 4040
    },
    {
      "epoch": 0.06297337059322627,
      "grad_norm": 12.446966171264648,
      "learning_rate": 9.863653496337327e-07,
      "loss": 0.6881,
      "step": 4045
    },
    {
      "epoch": 0.06305121159519565,
      "grad_norm": 3.663348436355591,
      "learning_rate": 9.862834106291277e-07,
      "loss": 0.6791,
      "step": 4050
    },
    {
      "epoch": 0.06312905259716503,
      "grad_norm": 5.9468159675598145,
      "learning_rate": 9.862014716245226e-07,
      "loss": 0.883,
      "step": 4055
    },
    {
      "epoch": 0.06320689359913441,
      "grad_norm": 4.544028282165527,
      "learning_rate": 9.861195326199177e-07,
      "loss": 0.6979,
      "step": 4060
    },
    {
      "epoch": 0.06328473460110379,
      "grad_norm": 4.25548791885376,
      "learning_rate": 9.860375936153127e-07,
      "loss": 0.757,
      "step": 4065
    },
    {
      "epoch": 0.06336257560307317,
      "grad_norm": 4.892475128173828,
      "learning_rate": 9.859556546107078e-07,
      "loss": 0.8346,
      "step": 4070
    },
    {
      "epoch": 0.06344041660504254,
      "grad_norm": 3.967132091522217,
      "learning_rate": 9.858737156061027e-07,
      "loss": 0.7614,
      "step": 4075
    },
    {
      "epoch": 0.06351825760701192,
      "grad_norm": 9.065237998962402,
      "learning_rate": 9.857917766014978e-07,
      "loss": 0.8471,
      "step": 4080
    },
    {
      "epoch": 0.0635960986089813,
      "grad_norm": 5.109429359436035,
      "learning_rate": 9.857098375968928e-07,
      "loss": 0.7441,
      "step": 4085
    },
    {
      "epoch": 0.06367393961095068,
      "grad_norm": 13.242950439453125,
      "learning_rate": 9.85627898592288e-07,
      "loss": 0.7784,
      "step": 4090
    },
    {
      "epoch": 0.06375178061292006,
      "grad_norm": 7.870430946350098,
      "learning_rate": 9.855459595876828e-07,
      "loss": 0.9225,
      "step": 4095
    },
    {
      "epoch": 0.06382962161488942,
      "grad_norm": 6.2109761238098145,
      "learning_rate": 9.854640205830778e-07,
      "loss": 0.8741,
      "step": 4100
    },
    {
      "epoch": 0.0639074626168588,
      "grad_norm": 4.566768169403076,
      "learning_rate": 9.85382081578473e-07,
      "loss": 0.7312,
      "step": 4105
    },
    {
      "epoch": 0.06398530361882818,
      "grad_norm": 4.343275547027588,
      "learning_rate": 9.85300142573868e-07,
      "loss": 0.8077,
      "step": 4110
    },
    {
      "epoch": 0.06406314462079755,
      "grad_norm": 3.710590124130249,
      "learning_rate": 9.85218203569263e-07,
      "loss": 0.8512,
      "step": 4115
    },
    {
      "epoch": 0.06414098562276693,
      "grad_norm": 5.875495433807373,
      "learning_rate": 9.85136264564658e-07,
      "loss": 0.8588,
      "step": 4120
    },
    {
      "epoch": 0.06421882662473631,
      "grad_norm": 5.609859943389893,
      "learning_rate": 9.85054325560053e-07,
      "loss": 0.9756,
      "step": 4125
    },
    {
      "epoch": 0.06429666762670569,
      "grad_norm": 3.695260763168335,
      "learning_rate": 9.84972386555448e-07,
      "loss": 0.8677,
      "step": 4130
    },
    {
      "epoch": 0.06437450862867507,
      "grad_norm": 4.265758991241455,
      "learning_rate": 9.848904475508432e-07,
      "loss": 0.868,
      "step": 4135
    },
    {
      "epoch": 0.06445234963064445,
      "grad_norm": 5.0540361404418945,
      "learning_rate": 9.84808508546238e-07,
      "loss": 0.7448,
      "step": 4140
    },
    {
      "epoch": 0.06453019063261382,
      "grad_norm": 3.1422078609466553,
      "learning_rate": 9.84726569541633e-07,
      "loss": 0.8085,
      "step": 4145
    },
    {
      "epoch": 0.0646080316345832,
      "grad_norm": 3.257333755493164,
      "learning_rate": 9.846446305370282e-07,
      "loss": 0.904,
      "step": 4150
    },
    {
      "epoch": 0.06468587263655258,
      "grad_norm": 6.303824424743652,
      "learning_rate": 9.845626915324232e-07,
      "loss": 0.6844,
      "step": 4155
    },
    {
      "epoch": 0.06476371363852196,
      "grad_norm": 7.541611194610596,
      "learning_rate": 9.844807525278183e-07,
      "loss": 0.9385,
      "step": 4160
    },
    {
      "epoch": 0.06484155464049134,
      "grad_norm": 3.217496633529663,
      "learning_rate": 9.843988135232134e-07,
      "loss": 0.8201,
      "step": 4165
    },
    {
      "epoch": 0.06491939564246071,
      "grad_norm": 4.375589370727539,
      "learning_rate": 9.843168745186083e-07,
      "loss": 0.8042,
      "step": 4170
    },
    {
      "epoch": 0.06499723664443009,
      "grad_norm": 6.62051248550415,
      "learning_rate": 9.842349355140033e-07,
      "loss": 0.7035,
      "step": 4175
    },
    {
      "epoch": 0.06507507764639947,
      "grad_norm": 4.503577709197998,
      "learning_rate": 9.841529965093984e-07,
      "loss": 0.8564,
      "step": 4180
    },
    {
      "epoch": 0.06515291864836883,
      "grad_norm": 3.583695411682129,
      "learning_rate": 9.840710575047933e-07,
      "loss": 0.9069,
      "step": 4185
    },
    {
      "epoch": 0.06523075965033821,
      "grad_norm": 4.029445648193359,
      "learning_rate": 9.839891185001883e-07,
      "loss": 0.8835,
      "step": 4190
    },
    {
      "epoch": 0.06530860065230759,
      "grad_norm": 3.6656410694122314,
      "learning_rate": 9.839071794955834e-07,
      "loss": 0.7814,
      "step": 4195
    },
    {
      "epoch": 0.06538644165427697,
      "grad_norm": 3.0505213737487793,
      "learning_rate": 9.838252404909785e-07,
      "loss": 0.7942,
      "step": 4200
    },
    {
      "epoch": 0.06546428265624635,
      "grad_norm": 4.775297164916992,
      "learning_rate": 9.837433014863736e-07,
      "loss": 0.8875,
      "step": 4205
    },
    {
      "epoch": 0.06554212365821573,
      "grad_norm": 5.490566253662109,
      "learning_rate": 9.836613624817686e-07,
      "loss": 0.7635,
      "step": 4210
    },
    {
      "epoch": 0.0656199646601851,
      "grad_norm": 3.202033519744873,
      "learning_rate": 9.835794234771635e-07,
      "loss": 0.7558,
      "step": 4215
    },
    {
      "epoch": 0.06569780566215448,
      "grad_norm": 5.484325408935547,
      "learning_rate": 9.834974844725586e-07,
      "loss": 0.8066,
      "step": 4220
    },
    {
      "epoch": 0.06577564666412386,
      "grad_norm": 2.903610944747925,
      "learning_rate": 9.834155454679537e-07,
      "loss": 0.7833,
      "step": 4225
    },
    {
      "epoch": 0.06585348766609324,
      "grad_norm": 3.188546895980835,
      "learning_rate": 9.833336064633485e-07,
      "loss": 0.7774,
      "step": 4230
    },
    {
      "epoch": 0.06593132866806262,
      "grad_norm": 3.055574655532837,
      "learning_rate": 9.832516674587436e-07,
      "loss": 0.6551,
      "step": 4235
    },
    {
      "epoch": 0.066009169670032,
      "grad_norm": 4.439972877502441,
      "learning_rate": 9.831697284541387e-07,
      "loss": 0.7456,
      "step": 4240
    },
    {
      "epoch": 0.06608701067200137,
      "grad_norm": 2.4513139724731445,
      "learning_rate": 9.830877894495337e-07,
      "loss": 0.7752,
      "step": 4245
    },
    {
      "epoch": 0.06616485167397075,
      "grad_norm": 4.66846227645874,
      "learning_rate": 9.830058504449288e-07,
      "loss": 0.9322,
      "step": 4250
    },
    {
      "epoch": 0.06624269267594013,
      "grad_norm": 4.819527626037598,
      "learning_rate": 9.82923911440324e-07,
      "loss": 0.9361,
      "step": 4255
    },
    {
      "epoch": 0.06632053367790951,
      "grad_norm": 8.028414726257324,
      "learning_rate": 9.828419724357188e-07,
      "loss": 0.8214,
      "step": 4260
    },
    {
      "epoch": 0.06639837467987889,
      "grad_norm": 3.565459728240967,
      "learning_rate": 9.827600334311138e-07,
      "loss": 0.7668,
      "step": 4265
    },
    {
      "epoch": 0.06647621568184826,
      "grad_norm": 2.9492602348327637,
      "learning_rate": 9.82678094426509e-07,
      "loss": 0.7513,
      "step": 4270
    },
    {
      "epoch": 0.06655405668381763,
      "grad_norm": 4.8683576583862305,
      "learning_rate": 9.82596155421904e-07,
      "loss": 0.8725,
      "step": 4275
    },
    {
      "epoch": 0.066631897685787,
      "grad_norm": 4.162265300750732,
      "learning_rate": 9.825142164172989e-07,
      "loss": 0.8275,
      "step": 4280
    },
    {
      "epoch": 0.06670973868775638,
      "grad_norm": 3.6537702083587646,
      "learning_rate": 9.82432277412694e-07,
      "loss": 0.7264,
      "step": 4285
    },
    {
      "epoch": 0.06678757968972576,
      "grad_norm": 3.9282073974609375,
      "learning_rate": 9.82350338408089e-07,
      "loss": 0.977,
      "step": 4290
    },
    {
      "epoch": 0.06686542069169514,
      "grad_norm": 5.129037857055664,
      "learning_rate": 9.82268399403484e-07,
      "loss": 0.8609,
      "step": 4295
    },
    {
      "epoch": 0.06694326169366452,
      "grad_norm": 4.563994884490967,
      "learning_rate": 9.82186460398879e-07,
      "loss": 0.8303,
      "step": 4300
    },
    {
      "epoch": 0.0670211026956339,
      "grad_norm": 3.177889585494995,
      "learning_rate": 9.82104521394274e-07,
      "loss": 0.9134,
      "step": 4305
    },
    {
      "epoch": 0.06709894369760327,
      "grad_norm": 4.675817966461182,
      "learning_rate": 9.82022582389669e-07,
      "loss": 0.7188,
      "step": 4310
    },
    {
      "epoch": 0.06717678469957265,
      "grad_norm": 6.9661173820495605,
      "learning_rate": 9.819406433850642e-07,
      "loss": 0.7871,
      "step": 4315
    },
    {
      "epoch": 0.06725462570154203,
      "grad_norm": 6.177728176116943,
      "learning_rate": 9.818587043804592e-07,
      "loss": 0.7438,
      "step": 4320
    },
    {
      "epoch": 0.06733246670351141,
      "grad_norm": 3.9021103382110596,
      "learning_rate": 9.81776765375854e-07,
      "loss": 0.8456,
      "step": 4325
    },
    {
      "epoch": 0.06741030770548079,
      "grad_norm": 6.576573371887207,
      "learning_rate": 9.816948263712492e-07,
      "loss": 0.8173,
      "step": 4330
    },
    {
      "epoch": 0.06748814870745017,
      "grad_norm": 3.117799997329712,
      "learning_rate": 9.816128873666442e-07,
      "loss": 0.8552,
      "step": 4335
    },
    {
      "epoch": 0.06756598970941954,
      "grad_norm": 5.52931022644043,
      "learning_rate": 9.815309483620391e-07,
      "loss": 0.7353,
      "step": 4340
    },
    {
      "epoch": 0.06764383071138892,
      "grad_norm": 3.3571298122406006,
      "learning_rate": 9.814490093574342e-07,
      "loss": 0.7253,
      "step": 4345
    },
    {
      "epoch": 0.0677216717133583,
      "grad_norm": 4.7125468254089355,
      "learning_rate": 9.813670703528293e-07,
      "loss": 0.8708,
      "step": 4350
    },
    {
      "epoch": 0.06779951271532768,
      "grad_norm": 3.7811620235443115,
      "learning_rate": 9.812851313482243e-07,
      "loss": 0.744,
      "step": 4355
    },
    {
      "epoch": 0.06787735371729704,
      "grad_norm": 4.079869270324707,
      "learning_rate": 9.812031923436194e-07,
      "loss": 0.8291,
      "step": 4360
    },
    {
      "epoch": 0.06795519471926642,
      "grad_norm": 2.9714179039001465,
      "learning_rate": 9.811212533390145e-07,
      "loss": 0.879,
      "step": 4365
    },
    {
      "epoch": 0.0680330357212358,
      "grad_norm": 4.301975250244141,
      "learning_rate": 9.810393143344094e-07,
      "loss": 0.7528,
      "step": 4370
    },
    {
      "epoch": 0.06811087672320518,
      "grad_norm": 4.707742214202881,
      "learning_rate": 9.809573753298044e-07,
      "loss": 0.7686,
      "step": 4375
    },
    {
      "epoch": 0.06818871772517456,
      "grad_norm": 2.911092758178711,
      "learning_rate": 9.808754363251995e-07,
      "loss": 0.8224,
      "step": 4380
    },
    {
      "epoch": 0.06826655872714393,
      "grad_norm": 3.809354543685913,
      "learning_rate": 9.807934973205944e-07,
      "loss": 0.9337,
      "step": 4385
    },
    {
      "epoch": 0.06834439972911331,
      "grad_norm": 3.0105934143066406,
      "learning_rate": 9.807115583159894e-07,
      "loss": 0.7952,
      "step": 4390
    },
    {
      "epoch": 0.06842224073108269,
      "grad_norm": 4.267519474029541,
      "learning_rate": 9.806296193113845e-07,
      "loss": 0.9312,
      "step": 4395
    },
    {
      "epoch": 0.06850008173305207,
      "grad_norm": 13.714824676513672,
      "learning_rate": 9.805476803067796e-07,
      "loss": 0.7266,
      "step": 4400
    },
    {
      "epoch": 0.06857792273502145,
      "grad_norm": 5.861302852630615,
      "learning_rate": 9.804657413021747e-07,
      "loss": 0.8267,
      "step": 4405
    },
    {
      "epoch": 0.06865576373699082,
      "grad_norm": 4.226170539855957,
      "learning_rate": 9.803838022975697e-07,
      "loss": 0.6108,
      "step": 4410
    },
    {
      "epoch": 0.0687336047389602,
      "grad_norm": 4.260887145996094,
      "learning_rate": 9.803018632929648e-07,
      "loss": 0.7917,
      "step": 4415
    },
    {
      "epoch": 0.06881144574092958,
      "grad_norm": 2.1800050735473633,
      "learning_rate": 9.802199242883597e-07,
      "loss": 0.7279,
      "step": 4420
    },
    {
      "epoch": 0.06888928674289896,
      "grad_norm": 4.386568069458008,
      "learning_rate": 9.801379852837548e-07,
      "loss": 0.7997,
      "step": 4425
    },
    {
      "epoch": 0.06896712774486834,
      "grad_norm": 7.1831135749816895,
      "learning_rate": 9.800560462791498e-07,
      "loss": 0.9703,
      "step": 4430
    },
    {
      "epoch": 0.06904496874683771,
      "grad_norm": 7.631860733032227,
      "learning_rate": 9.799741072745447e-07,
      "loss": 0.7836,
      "step": 4435
    },
    {
      "epoch": 0.06912280974880709,
      "grad_norm": 3.6150078773498535,
      "learning_rate": 9.798921682699398e-07,
      "loss": 0.6982,
      "step": 4440
    },
    {
      "epoch": 0.06920065075077646,
      "grad_norm": 5.267273902893066,
      "learning_rate": 9.798102292653348e-07,
      "loss": 0.9292,
      "step": 4445
    },
    {
      "epoch": 0.06927849175274584,
      "grad_norm": 6.139009952545166,
      "learning_rate": 9.7972829026073e-07,
      "loss": 0.776,
      "step": 4450
    },
    {
      "epoch": 0.06935633275471521,
      "grad_norm": 6.20229959487915,
      "learning_rate": 9.79646351256125e-07,
      "loss": 0.8239,
      "step": 4455
    },
    {
      "epoch": 0.06943417375668459,
      "grad_norm": 3.204371929168701,
      "learning_rate": 9.7956441225152e-07,
      "loss": 0.8123,
      "step": 4460
    },
    {
      "epoch": 0.06951201475865397,
      "grad_norm": 4.521599769592285,
      "learning_rate": 9.79482473246915e-07,
      "loss": 0.7049,
      "step": 4465
    },
    {
      "epoch": 0.06958985576062335,
      "grad_norm": 5.0935750007629395,
      "learning_rate": 9.7940053424231e-07,
      "loss": 0.8673,
      "step": 4470
    },
    {
      "epoch": 0.06966769676259273,
      "grad_norm": 7.926290512084961,
      "learning_rate": 9.79318595237705e-07,
      "loss": 0.8195,
      "step": 4475
    },
    {
      "epoch": 0.0697455377645621,
      "grad_norm": 4.315165042877197,
      "learning_rate": 9.792366562331e-07,
      "loss": 0.8674,
      "step": 4480
    },
    {
      "epoch": 0.06982337876653148,
      "grad_norm": 3.775836706161499,
      "learning_rate": 9.79154717228495e-07,
      "loss": 0.8334,
      "step": 4485
    },
    {
      "epoch": 0.06990121976850086,
      "grad_norm": 2.560904026031494,
      "learning_rate": 9.7907277822389e-07,
      "loss": 0.8271,
      "step": 4490
    },
    {
      "epoch": 0.06997906077047024,
      "grad_norm": 11.29925537109375,
      "learning_rate": 9.789908392192852e-07,
      "loss": 0.9633,
      "step": 4495
    },
    {
      "epoch": 0.07005690177243962,
      "grad_norm": 4.101975917816162,
      "learning_rate": 9.789089002146802e-07,
      "loss": 0.8858,
      "step": 4500
    },
    {
      "epoch": 0.070134742774409,
      "grad_norm": 2.970782518386841,
      "learning_rate": 9.78826961210075e-07,
      "loss": 0.8608,
      "step": 4505
    },
    {
      "epoch": 0.07021258377637837,
      "grad_norm": 7.289088726043701,
      "learning_rate": 9.787450222054702e-07,
      "loss": 0.8347,
      "step": 4510
    },
    {
      "epoch": 0.07029042477834775,
      "grad_norm": 7.107760429382324,
      "learning_rate": 9.786630832008653e-07,
      "loss": 0.771,
      "step": 4515
    },
    {
      "epoch": 0.07036826578031713,
      "grad_norm": 3.630275249481201,
      "learning_rate": 9.785811441962603e-07,
      "loss": 0.7113,
      "step": 4520
    },
    {
      "epoch": 0.07044610678228651,
      "grad_norm": 4.681270122528076,
      "learning_rate": 9.784992051916554e-07,
      "loss": 0.8278,
      "step": 4525
    },
    {
      "epoch": 0.07052394778425587,
      "grad_norm": 3.6923000812530518,
      "learning_rate": 9.784172661870503e-07,
      "loss": 0.8067,
      "step": 4530
    },
    {
      "epoch": 0.07060178878622525,
      "grad_norm": 3.538496255874634,
      "learning_rate": 9.783353271824453e-07,
      "loss": 0.7577,
      "step": 4535
    },
    {
      "epoch": 0.07067962978819463,
      "grad_norm": 3.3996520042419434,
      "learning_rate": 9.782533881778404e-07,
      "loss": 0.9051,
      "step": 4540
    },
    {
      "epoch": 0.070757470790164,
      "grad_norm": 4.107473850250244,
      "learning_rate": 9.781714491732353e-07,
      "loss": 0.8982,
      "step": 4545
    },
    {
      "epoch": 0.07083531179213338,
      "grad_norm": 2.9986937046051025,
      "learning_rate": 9.780895101686304e-07,
      "loss": 0.9025,
      "step": 4550
    },
    {
      "epoch": 0.07091315279410276,
      "grad_norm": 3.413224697113037,
      "learning_rate": 9.780075711640254e-07,
      "loss": 0.7699,
      "step": 4555
    },
    {
      "epoch": 0.07099099379607214,
      "grad_norm": 3.332380771636963,
      "learning_rate": 9.779256321594205e-07,
      "loss": 0.6913,
      "step": 4560
    },
    {
      "epoch": 0.07106883479804152,
      "grad_norm": 3.161701202392578,
      "learning_rate": 9.778436931548156e-07,
      "loss": 0.7502,
      "step": 4565
    },
    {
      "epoch": 0.0711466758000109,
      "grad_norm": 3.6863913536071777,
      "learning_rate": 9.777617541502107e-07,
      "loss": 0.7959,
      "step": 4570
    },
    {
      "epoch": 0.07122451680198028,
      "grad_norm": 4.537403583526611,
      "learning_rate": 9.776798151456055e-07,
      "loss": 0.8646,
      "step": 4575
    },
    {
      "epoch": 0.07130235780394965,
      "grad_norm": 4.111873149871826,
      "learning_rate": 9.775978761410006e-07,
      "loss": 0.8503,
      "step": 4580
    },
    {
      "epoch": 0.07138019880591903,
      "grad_norm": 8.788448333740234,
      "learning_rate": 9.775159371363957e-07,
      "loss": 0.8046,
      "step": 4585
    },
    {
      "epoch": 0.07145803980788841,
      "grad_norm": 5.538233757019043,
      "learning_rate": 9.774339981317905e-07,
      "loss": 0.8018,
      "step": 4590
    },
    {
      "epoch": 0.07153588080985779,
      "grad_norm": 6.06341028213501,
      "learning_rate": 9.773520591271856e-07,
      "loss": 0.8073,
      "step": 4595
    },
    {
      "epoch": 0.07161372181182717,
      "grad_norm": 3.6553616523742676,
      "learning_rate": 9.772701201225807e-07,
      "loss": 0.8142,
      "step": 4600
    },
    {
      "epoch": 0.07169156281379654,
      "grad_norm": 4.252196311950684,
      "learning_rate": 9.771881811179758e-07,
      "loss": 0.7505,
      "step": 4605
    },
    {
      "epoch": 0.07176940381576592,
      "grad_norm": 3.3813109397888184,
      "learning_rate": 9.771062421133708e-07,
      "loss": 0.7076,
      "step": 4610
    },
    {
      "epoch": 0.0718472448177353,
      "grad_norm": 9.012163162231445,
      "learning_rate": 9.77024303108766e-07,
      "loss": 0.8812,
      "step": 4615
    },
    {
      "epoch": 0.07192508581970466,
      "grad_norm": 6.12354040145874,
      "learning_rate": 9.769423641041608e-07,
      "loss": 0.8218,
      "step": 4620
    },
    {
      "epoch": 0.07200292682167404,
      "grad_norm": 3.364898681640625,
      "learning_rate": 9.768604250995558e-07,
      "loss": 0.8775,
      "step": 4625
    },
    {
      "epoch": 0.07208076782364342,
      "grad_norm": 13.047234535217285,
      "learning_rate": 9.76778486094951e-07,
      "loss": 0.7578,
      "step": 4630
    },
    {
      "epoch": 0.0721586088256128,
      "grad_norm": 6.722197532653809,
      "learning_rate": 9.766965470903458e-07,
      "loss": 0.8446,
      "step": 4635
    },
    {
      "epoch": 0.07223644982758218,
      "grad_norm": 4.028960227966309,
      "learning_rate": 9.766146080857409e-07,
      "loss": 0.8284,
      "step": 4640
    },
    {
      "epoch": 0.07231429082955156,
      "grad_norm": 3.668736219406128,
      "learning_rate": 9.76532669081136e-07,
      "loss": 0.8298,
      "step": 4645
    },
    {
      "epoch": 0.07239213183152093,
      "grad_norm": 3.391463041305542,
      "learning_rate": 9.76450730076531e-07,
      "loss": 0.8038,
      "step": 4650
    },
    {
      "epoch": 0.07246997283349031,
      "grad_norm": 2.8080356121063232,
      "learning_rate": 9.76368791071926e-07,
      "loss": 0.8614,
      "step": 4655
    },
    {
      "epoch": 0.07254781383545969,
      "grad_norm": 3.9080796241760254,
      "learning_rate": 9.762868520673212e-07,
      "loss": 0.8239,
      "step": 4660
    },
    {
      "epoch": 0.07262565483742907,
      "grad_norm": 3.0968992710113525,
      "learning_rate": 9.76204913062716e-07,
      "loss": 0.8751,
      "step": 4665
    },
    {
      "epoch": 0.07270349583939845,
      "grad_norm": 6.975797176361084,
      "learning_rate": 9.76122974058111e-07,
      "loss": 0.7877,
      "step": 4670
    },
    {
      "epoch": 0.07278133684136782,
      "grad_norm": 5.175839424133301,
      "learning_rate": 9.760410350535062e-07,
      "loss": 0.7242,
      "step": 4675
    },
    {
      "epoch": 0.0728591778433372,
      "grad_norm": 3.86811900138855,
      "learning_rate": 9.759590960489012e-07,
      "loss": 0.8628,
      "step": 4680
    },
    {
      "epoch": 0.07293701884530658,
      "grad_norm": 4.670974254608154,
      "learning_rate": 9.75877157044296e-07,
      "loss": 0.7741,
      "step": 4685
    },
    {
      "epoch": 0.07301485984727596,
      "grad_norm": 3.4863369464874268,
      "learning_rate": 9.757952180396912e-07,
      "loss": 0.9401,
      "step": 4690
    },
    {
      "epoch": 0.07309270084924534,
      "grad_norm": 4.012441158294678,
      "learning_rate": 9.757132790350863e-07,
      "loss": 0.8949,
      "step": 4695
    },
    {
      "epoch": 0.07317054185121472,
      "grad_norm": 3.7120773792266846,
      "learning_rate": 9.756313400304813e-07,
      "loss": 0.912,
      "step": 4700
    },
    {
      "epoch": 0.07324838285318408,
      "grad_norm": 4.149153232574463,
      "learning_rate": 9.755494010258762e-07,
      "loss": 0.7284,
      "step": 4705
    },
    {
      "epoch": 0.07332622385515346,
      "grad_norm": 3.724862813949585,
      "learning_rate": 9.754674620212713e-07,
      "loss": 0.8486,
      "step": 4710
    },
    {
      "epoch": 0.07340406485712284,
      "grad_norm": 5.275464057922363,
      "learning_rate": 9.753855230166663e-07,
      "loss": 0.7661,
      "step": 4715
    },
    {
      "epoch": 0.07348190585909221,
      "grad_norm": 8.389967918395996,
      "learning_rate": 9.753035840120614e-07,
      "loss": 0.8646,
      "step": 4720
    },
    {
      "epoch": 0.07355974686106159,
      "grad_norm": 4.1537017822265625,
      "learning_rate": 9.752216450074565e-07,
      "loss": 0.8596,
      "step": 4725
    },
    {
      "epoch": 0.07363758786303097,
      "grad_norm": 3.4971349239349365,
      "learning_rate": 9.751397060028514e-07,
      "loss": 0.7888,
      "step": 4730
    },
    {
      "epoch": 0.07371542886500035,
      "grad_norm": 2.9952375888824463,
      "learning_rate": 9.750577669982464e-07,
      "loss": 0.8127,
      "step": 4735
    },
    {
      "epoch": 0.07379326986696973,
      "grad_norm": 3.0156424045562744,
      "learning_rate": 9.749758279936415e-07,
      "loss": 0.6775,
      "step": 4740
    },
    {
      "epoch": 0.0738711108689391,
      "grad_norm": 4.386186122894287,
      "learning_rate": 9.748938889890366e-07,
      "loss": 0.8813,
      "step": 4745
    },
    {
      "epoch": 0.07394895187090848,
      "grad_norm": 8.352777481079102,
      "learning_rate": 9.748119499844314e-07,
      "loss": 0.841,
      "step": 4750
    },
    {
      "epoch": 0.07402679287287786,
      "grad_norm": 3.9071156978607178,
      "learning_rate": 9.747300109798265e-07,
      "loss": 0.8124,
      "step": 4755
    },
    {
      "epoch": 0.07410463387484724,
      "grad_norm": 6.337040901184082,
      "learning_rate": 9.746480719752216e-07,
      "loss": 0.7601,
      "step": 4760
    },
    {
      "epoch": 0.07418247487681662,
      "grad_norm": 4.740725040435791,
      "learning_rate": 9.745661329706167e-07,
      "loss": 0.8621,
      "step": 4765
    },
    {
      "epoch": 0.074260315878786,
      "grad_norm": 3.6366703510284424,
      "learning_rate": 9.744841939660117e-07,
      "loss": 0.7521,
      "step": 4770
    },
    {
      "epoch": 0.07433815688075537,
      "grad_norm": 5.869968891143799,
      "learning_rate": 9.744022549614068e-07,
      "loss": 0.8219,
      "step": 4775
    },
    {
      "epoch": 0.07441599788272475,
      "grad_norm": 4.8249006271362305,
      "learning_rate": 9.743203159568017e-07,
      "loss": 0.8733,
      "step": 4780
    },
    {
      "epoch": 0.07449383888469413,
      "grad_norm": 3.9930624961853027,
      "learning_rate": 9.742383769521968e-07,
      "loss": 0.7968,
      "step": 4785
    },
    {
      "epoch": 0.0745716798866635,
      "grad_norm": 5.8335418701171875,
      "learning_rate": 9.741564379475918e-07,
      "loss": 0.8224,
      "step": 4790
    },
    {
      "epoch": 0.07464952088863287,
      "grad_norm": 5.657021522521973,
      "learning_rate": 9.740744989429867e-07,
      "loss": 0.7896,
      "step": 4795
    },
    {
      "epoch": 0.07472736189060225,
      "grad_norm": 6.225119590759277,
      "learning_rate": 9.739925599383818e-07,
      "loss": 0.8297,
      "step": 4800
    },
    {
      "epoch": 0.07480520289257163,
      "grad_norm": 3.373596429824829,
      "learning_rate": 9.739106209337768e-07,
      "loss": 0.8255,
      "step": 4805
    },
    {
      "epoch": 0.07488304389454101,
      "grad_norm": 2.2436752319335938,
      "learning_rate": 9.73828681929172e-07,
      "loss": 0.7097,
      "step": 4810
    },
    {
      "epoch": 0.07496088489651039,
      "grad_norm": 3.6879262924194336,
      "learning_rate": 9.73746742924567e-07,
      "loss": 0.6912,
      "step": 4815
    },
    {
      "epoch": 0.07503872589847976,
      "grad_norm": 3.399632692337036,
      "learning_rate": 9.73664803919962e-07,
      "loss": 0.8371,
      "step": 4820
    },
    {
      "epoch": 0.07511656690044914,
      "grad_norm": 9.186985969543457,
      "learning_rate": 9.73582864915357e-07,
      "loss": 0.8157,
      "step": 4825
    },
    {
      "epoch": 0.07519440790241852,
      "grad_norm": 3.6295411586761475,
      "learning_rate": 9.73500925910752e-07,
      "loss": 0.7785,
      "step": 4830
    },
    {
      "epoch": 0.0752722489043879,
      "grad_norm": 3.534175395965576,
      "learning_rate": 9.73418986906147e-07,
      "loss": 0.878,
      "step": 4835
    },
    {
      "epoch": 0.07535008990635728,
      "grad_norm": 3.0165436267852783,
      "learning_rate": 9.73337047901542e-07,
      "loss": 0.7593,
      "step": 4840
    },
    {
      "epoch": 0.07542793090832665,
      "grad_norm": 4.980969429016113,
      "learning_rate": 9.73255108896937e-07,
      "loss": 0.7592,
      "step": 4845
    },
    {
      "epoch": 0.07550577191029603,
      "grad_norm": 3.377429723739624,
      "learning_rate": 9.73173169892332e-07,
      "loss": 0.7849,
      "step": 4850
    },
    {
      "epoch": 0.07558361291226541,
      "grad_norm": 5.916225910186768,
      "learning_rate": 9.730912308877272e-07,
      "loss": 0.8283,
      "step": 4855
    },
    {
      "epoch": 0.07566145391423479,
      "grad_norm": 6.396664619445801,
      "learning_rate": 9.730092918831222e-07,
      "loss": 0.8801,
      "step": 4860
    },
    {
      "epoch": 0.07573929491620417,
      "grad_norm": 3.2279000282287598,
      "learning_rate": 9.729273528785173e-07,
      "loss": 0.7746,
      "step": 4865
    },
    {
      "epoch": 0.07581713591817354,
      "grad_norm": 3.3522236347198486,
      "learning_rate": 9.728454138739122e-07,
      "loss": 1.0139,
      "step": 4870
    },
    {
      "epoch": 0.07589497692014291,
      "grad_norm": 7.16496467590332,
      "learning_rate": 9.727634748693073e-07,
      "loss": 0.8203,
      "step": 4875
    },
    {
      "epoch": 0.07597281792211229,
      "grad_norm": 3.7520346641540527,
      "learning_rate": 9.726815358647023e-07,
      "loss": 0.7859,
      "step": 4880
    },
    {
      "epoch": 0.07605065892408167,
      "grad_norm": 5.511653900146484,
      "learning_rate": 9.725995968600972e-07,
      "loss": 0.8594,
      "step": 4885
    },
    {
      "epoch": 0.07612849992605104,
      "grad_norm": 5.89841365814209,
      "learning_rate": 9.725176578554923e-07,
      "loss": 0.6535,
      "step": 4890
    },
    {
      "epoch": 0.07620634092802042,
      "grad_norm": 4.694098949432373,
      "learning_rate": 9.724357188508873e-07,
      "loss": 0.8221,
      "step": 4895
    },
    {
      "epoch": 0.0762841819299898,
      "grad_norm": 4.192508220672607,
      "learning_rate": 9.723537798462824e-07,
      "loss": 0.8594,
      "step": 4900
    },
    {
      "epoch": 0.07636202293195918,
      "grad_norm": 4.4052534103393555,
      "learning_rate": 9.722718408416775e-07,
      "loss": 0.7326,
      "step": 4905
    },
    {
      "epoch": 0.07643986393392856,
      "grad_norm": 4.1674299240112305,
      "learning_rate": 9.721899018370724e-07,
      "loss": 0.7525,
      "step": 4910
    },
    {
      "epoch": 0.07651770493589793,
      "grad_norm": 4.367162227630615,
      "learning_rate": 9.721079628324674e-07,
      "loss": 0.9038,
      "step": 4915
    },
    {
      "epoch": 0.07659554593786731,
      "grad_norm": 3.0957272052764893,
      "learning_rate": 9.720260238278625e-07,
      "loss": 0.8247,
      "step": 4920
    },
    {
      "epoch": 0.07667338693983669,
      "grad_norm": 6.5740532875061035,
      "learning_rate": 9.719440848232576e-07,
      "loss": 0.7707,
      "step": 4925
    },
    {
      "epoch": 0.07675122794180607,
      "grad_norm": 5.08697509765625,
      "learning_rate": 9.718621458186527e-07,
      "loss": 0.7561,
      "step": 4930
    },
    {
      "epoch": 0.07682906894377545,
      "grad_norm": 6.9134063720703125,
      "learning_rate": 9.717802068140475e-07,
      "loss": 0.7645,
      "step": 4935
    },
    {
      "epoch": 0.07690690994574483,
      "grad_norm": 3.2047367095947266,
      "learning_rate": 9.716982678094426e-07,
      "loss": 0.854,
      "step": 4940
    },
    {
      "epoch": 0.0769847509477142,
      "grad_norm": 3.4643442630767822,
      "learning_rate": 9.716163288048377e-07,
      "loss": 0.7598,
      "step": 4945
    },
    {
      "epoch": 0.07706259194968358,
      "grad_norm": 5.208106517791748,
      "learning_rate": 9.715343898002325e-07,
      "loss": 0.7699,
      "step": 4950
    },
    {
      "epoch": 0.07714043295165296,
      "grad_norm": 4.935080051422119,
      "learning_rate": 9.714524507956276e-07,
      "loss": 0.8839,
      "step": 4955
    },
    {
      "epoch": 0.07721827395362234,
      "grad_norm": 4.052170753479004,
      "learning_rate": 9.713705117910227e-07,
      "loss": 0.841,
      "step": 4960
    },
    {
      "epoch": 0.0772961149555917,
      "grad_norm": 3.409742593765259,
      "learning_rate": 9.712885727864178e-07,
      "loss": 0.8768,
      "step": 4965
    },
    {
      "epoch": 0.07737395595756108,
      "grad_norm": 5.577835559844971,
      "learning_rate": 9.712066337818128e-07,
      "loss": 0.7609,
      "step": 4970
    },
    {
      "epoch": 0.07745179695953046,
      "grad_norm": 4.966437816619873,
      "learning_rate": 9.71124694777208e-07,
      "loss": 0.9478,
      "step": 4975
    },
    {
      "epoch": 0.07752963796149984,
      "grad_norm": 5.092791557312012,
      "learning_rate": 9.710427557726028e-07,
      "loss": 0.8237,
      "step": 4980
    },
    {
      "epoch": 0.07760747896346921,
      "grad_norm": 2.992233991622925,
      "learning_rate": 9.709608167679978e-07,
      "loss": 0.8486,
      "step": 4985
    },
    {
      "epoch": 0.07768531996543859,
      "grad_norm": 3.947547197341919,
      "learning_rate": 9.70878877763393e-07,
      "loss": 1.0368,
      "step": 4990
    },
    {
      "epoch": 0.07776316096740797,
      "grad_norm": 4.660312652587891,
      "learning_rate": 9.707969387587878e-07,
      "loss": 0.7254,
      "step": 4995
    },
    {
      "epoch": 0.07784100196937735,
      "grad_norm": 5.542099475860596,
      "learning_rate": 9.707149997541829e-07,
      "loss": 0.7871,
      "step": 5000
    },
    {
      "epoch": 0.07791884297134673,
      "grad_norm": 2.8513717651367188,
      "learning_rate": 9.70633060749578e-07,
      "loss": 0.7405,
      "step": 5005
    },
    {
      "epoch": 0.0779966839733161,
      "grad_norm": 3.067697286605835,
      "learning_rate": 9.70551121744973e-07,
      "loss": 0.8251,
      "step": 5010
    },
    {
      "epoch": 0.07807452497528548,
      "grad_norm": 4.706809043884277,
      "learning_rate": 9.70469182740368e-07,
      "loss": 0.894,
      "step": 5015
    },
    {
      "epoch": 0.07815236597725486,
      "grad_norm": 3.1183722019195557,
      "learning_rate": 9.703872437357632e-07,
      "loss": 0.7985,
      "step": 5020
    },
    {
      "epoch": 0.07823020697922424,
      "grad_norm": 3.95314884185791,
      "learning_rate": 9.703053047311582e-07,
      "loss": 0.8673,
      "step": 5025
    },
    {
      "epoch": 0.07830804798119362,
      "grad_norm": 5.186405658721924,
      "learning_rate": 9.70223365726553e-07,
      "loss": 0.8834,
      "step": 5030
    },
    {
      "epoch": 0.078385888983163,
      "grad_norm": 6.825442790985107,
      "learning_rate": 9.701414267219482e-07,
      "loss": 0.8638,
      "step": 5035
    },
    {
      "epoch": 0.07846372998513237,
      "grad_norm": 4.547275543212891,
      "learning_rate": 9.700594877173432e-07,
      "loss": 0.7591,
      "step": 5040
    },
    {
      "epoch": 0.07854157098710175,
      "grad_norm": 3.055347204208374,
      "learning_rate": 9.69977548712738e-07,
      "loss": 0.7955,
      "step": 5045
    },
    {
      "epoch": 0.07861941198907112,
      "grad_norm": 6.120547294616699,
      "learning_rate": 9.698956097081332e-07,
      "loss": 0.8606,
      "step": 5050
    },
    {
      "epoch": 0.0786972529910405,
      "grad_norm": 8.125998497009277,
      "learning_rate": 9.698136707035283e-07,
      "loss": 0.7676,
      "step": 5055
    },
    {
      "epoch": 0.07877509399300987,
      "grad_norm": 3.794414758682251,
      "learning_rate": 9.697317316989233e-07,
      "loss": 0.8594,
      "step": 5060
    },
    {
      "epoch": 0.07885293499497925,
      "grad_norm": 4.892978191375732,
      "learning_rate": 9.696497926943184e-07,
      "loss": 0.774,
      "step": 5065
    },
    {
      "epoch": 0.07893077599694863,
      "grad_norm": 4.139584064483643,
      "learning_rate": 9.695678536897135e-07,
      "loss": 0.83,
      "step": 5070
    },
    {
      "epoch": 0.07900861699891801,
      "grad_norm": 7.144068241119385,
      "learning_rate": 9.694859146851083e-07,
      "loss": 0.8541,
      "step": 5075
    },
    {
      "epoch": 0.07908645800088739,
      "grad_norm": 10.283439636230469,
      "learning_rate": 9.694039756805034e-07,
      "loss": 0.9355,
      "step": 5080
    },
    {
      "epoch": 0.07916429900285676,
      "grad_norm": 3.0185656547546387,
      "learning_rate": 9.693220366758985e-07,
      "loss": 0.8761,
      "step": 5085
    },
    {
      "epoch": 0.07924214000482614,
      "grad_norm": 3.299808979034424,
      "learning_rate": 9.692400976712934e-07,
      "loss": 0.715,
      "step": 5090
    },
    {
      "epoch": 0.07931998100679552,
      "grad_norm": 7.163717746734619,
      "learning_rate": 9.691581586666884e-07,
      "loss": 0.6974,
      "step": 5095
    },
    {
      "epoch": 0.0793978220087649,
      "grad_norm": 3.0995216369628906,
      "learning_rate": 9.690762196620835e-07,
      "loss": 0.8159,
      "step": 5100
    },
    {
      "epoch": 0.07947566301073428,
      "grad_norm": 2.7312302589416504,
      "learning_rate": 9.689942806574786e-07,
      "loss": 0.8009,
      "step": 5105
    },
    {
      "epoch": 0.07955350401270365,
      "grad_norm": 4.884325981140137,
      "learning_rate": 9.689123416528737e-07,
      "loss": 0.8839,
      "step": 5110
    },
    {
      "epoch": 0.07963134501467303,
      "grad_norm": 3.1511213779449463,
      "learning_rate": 9.688304026482685e-07,
      "loss": 0.6633,
      "step": 5115
    },
    {
      "epoch": 0.07970918601664241,
      "grad_norm": 3.034996271133423,
      "learning_rate": 9.687484636436636e-07,
      "loss": 0.8636,
      "step": 5120
    },
    {
      "epoch": 0.07978702701861179,
      "grad_norm": 6.756342887878418,
      "learning_rate": 9.686665246390587e-07,
      "loss": 0.844,
      "step": 5125
    },
    {
      "epoch": 0.07986486802058117,
      "grad_norm": 4.012609958648682,
      "learning_rate": 9.685845856344537e-07,
      "loss": 1.0347,
      "step": 5130
    },
    {
      "epoch": 0.07994270902255053,
      "grad_norm": 4.039714336395264,
      "learning_rate": 9.685026466298486e-07,
      "loss": 0.8549,
      "step": 5135
    },
    {
      "epoch": 0.08002055002451991,
      "grad_norm": 4.654749393463135,
      "learning_rate": 9.684207076252437e-07,
      "loss": 0.7294,
      "step": 5140
    },
    {
      "epoch": 0.08009839102648929,
      "grad_norm": 5.652122497558594,
      "learning_rate": 9.683387686206388e-07,
      "loss": 0.8808,
      "step": 5145
    },
    {
      "epoch": 0.08017623202845867,
      "grad_norm": 5.13718318939209,
      "learning_rate": 9.682568296160338e-07,
      "loss": 0.8012,
      "step": 5150
    },
    {
      "epoch": 0.08025407303042804,
      "grad_norm": 4.274785995483398,
      "learning_rate": 9.681748906114287e-07,
      "loss": 0.9497,
      "step": 5155
    },
    {
      "epoch": 0.08033191403239742,
      "grad_norm": 3.5715765953063965,
      "learning_rate": 9.680929516068238e-07,
      "loss": 0.6932,
      "step": 5160
    },
    {
      "epoch": 0.0804097550343668,
      "grad_norm": 3.721369504928589,
      "learning_rate": 9.680110126022188e-07,
      "loss": 0.7604,
      "step": 5165
    },
    {
      "epoch": 0.08048759603633618,
      "grad_norm": 4.815948486328125,
      "learning_rate": 9.67929073597614e-07,
      "loss": 0.8056,
      "step": 5170
    },
    {
      "epoch": 0.08056543703830556,
      "grad_norm": 3.9973649978637695,
      "learning_rate": 9.67847134593009e-07,
      "loss": 0.8245,
      "step": 5175
    },
    {
      "epoch": 0.08064327804027493,
      "grad_norm": 6.30864143371582,
      "learning_rate": 9.67765195588404e-07,
      "loss": 0.8158,
      "step": 5180
    },
    {
      "epoch": 0.08072111904224431,
      "grad_norm": 3.627049207687378,
      "learning_rate": 9.67683256583799e-07,
      "loss": 0.6924,
      "step": 5185
    },
    {
      "epoch": 0.08079896004421369,
      "grad_norm": 3.445680618286133,
      "learning_rate": 9.67601317579194e-07,
      "loss": 0.7619,
      "step": 5190
    },
    {
      "epoch": 0.08087680104618307,
      "grad_norm": 5.6612868309021,
      "learning_rate": 9.67519378574589e-07,
      "loss": 0.8712,
      "step": 5195
    },
    {
      "epoch": 0.08095464204815245,
      "grad_norm": 8.172099113464355,
      "learning_rate": 9.67437439569984e-07,
      "loss": 0.875,
      "step": 5200
    },
    {
      "epoch": 0.08103248305012183,
      "grad_norm": 3.6549482345581055,
      "learning_rate": 9.67355500565379e-07,
      "loss": 0.7416,
      "step": 5205
    },
    {
      "epoch": 0.0811103240520912,
      "grad_norm": 4.237252712249756,
      "learning_rate": 9.67273561560774e-07,
      "loss": 0.7864,
      "step": 5210
    },
    {
      "epoch": 0.08118816505406058,
      "grad_norm": 3.6416895389556885,
      "learning_rate": 9.671916225561692e-07,
      "loss": 0.8346,
      "step": 5215
    },
    {
      "epoch": 0.08126600605602995,
      "grad_norm": 7.055088996887207,
      "learning_rate": 9.671096835515642e-07,
      "loss": 0.8257,
      "step": 5220
    },
    {
      "epoch": 0.08134384705799932,
      "grad_norm": 4.3031511306762695,
      "learning_rate": 9.670277445469593e-07,
      "loss": 0.9243,
      "step": 5225
    },
    {
      "epoch": 0.0814216880599687,
      "grad_norm": 12.051529884338379,
      "learning_rate": 9.669458055423542e-07,
      "loss": 0.8012,
      "step": 5230
    },
    {
      "epoch": 0.08149952906193808,
      "grad_norm": 3.5274226665496826,
      "learning_rate": 9.668638665377493e-07,
      "loss": 0.8752,
      "step": 5235
    },
    {
      "epoch": 0.08157737006390746,
      "grad_norm": 3.1642568111419678,
      "learning_rate": 9.667819275331443e-07,
      "loss": 0.7385,
      "step": 5240
    },
    {
      "epoch": 0.08165521106587684,
      "grad_norm": 3.645951271057129,
      "learning_rate": 9.666999885285392e-07,
      "loss": 0.7538,
      "step": 5245
    },
    {
      "epoch": 0.08173305206784622,
      "grad_norm": 5.045301914215088,
      "learning_rate": 9.666180495239343e-07,
      "loss": 0.8496,
      "step": 5250
    },
    {
      "epoch": 0.0818108930698156,
      "grad_norm": 3.8335864543914795,
      "learning_rate": 9.665361105193293e-07,
      "loss": 0.8149,
      "step": 5255
    },
    {
      "epoch": 0.08188873407178497,
      "grad_norm": 5.525310516357422,
      "learning_rate": 9.664541715147244e-07,
      "loss": 0.8061,
      "step": 5260
    },
    {
      "epoch": 0.08196657507375435,
      "grad_norm": 3.721007823944092,
      "learning_rate": 9.663722325101195e-07,
      "loss": 0.7835,
      "step": 5265
    },
    {
      "epoch": 0.08204441607572373,
      "grad_norm": 4.0820393562316895,
      "learning_rate": 9.662902935055146e-07,
      "loss": 0.8629,
      "step": 5270
    },
    {
      "epoch": 0.0821222570776931,
      "grad_norm": 2.5007712841033936,
      "learning_rate": 9.662083545009094e-07,
      "loss": 0.8987,
      "step": 5275
    },
    {
      "epoch": 0.08220009807966248,
      "grad_norm": 5.49976110458374,
      "learning_rate": 9.661264154963045e-07,
      "loss": 0.8664,
      "step": 5280
    },
    {
      "epoch": 0.08227793908163186,
      "grad_norm": 3.953249931335449,
      "learning_rate": 9.660444764916996e-07,
      "loss": 0.8364,
      "step": 5285
    },
    {
      "epoch": 0.08235578008360124,
      "grad_norm": 5.422050476074219,
      "learning_rate": 9.659625374870947e-07,
      "loss": 0.825,
      "step": 5290
    },
    {
      "epoch": 0.08243362108557062,
      "grad_norm": 6.019737720489502,
      "learning_rate": 9.658805984824895e-07,
      "loss": 0.7429,
      "step": 5295
    },
    {
      "epoch": 0.08251146208754,
      "grad_norm": 4.360890865325928,
      "learning_rate": 9.657986594778846e-07,
      "loss": 0.8384,
      "step": 5300
    },
    {
      "epoch": 0.08258930308950937,
      "grad_norm": 2.676135540008545,
      "learning_rate": 9.657167204732797e-07,
      "loss": 0.7928,
      "step": 5305
    },
    {
      "epoch": 0.08266714409147874,
      "grad_norm": 2.602173328399658,
      "learning_rate": 9.656347814686747e-07,
      "loss": 0.8301,
      "step": 5310
    },
    {
      "epoch": 0.08274498509344812,
      "grad_norm": 3.2521512508392334,
      "learning_rate": 9.655528424640696e-07,
      "loss": 0.8491,
      "step": 5315
    },
    {
      "epoch": 0.0828228260954175,
      "grad_norm": 3.9603660106658936,
      "learning_rate": 9.654709034594647e-07,
      "loss": 0.9349,
      "step": 5320
    },
    {
      "epoch": 0.08290066709738687,
      "grad_norm": 3.615999698638916,
      "learning_rate": 9.653889644548598e-07,
      "loss": 0.7493,
      "step": 5325
    },
    {
      "epoch": 0.08297850809935625,
      "grad_norm": 4.19753360748291,
      "learning_rate": 9.653070254502548e-07,
      "loss": 0.8548,
      "step": 5330
    },
    {
      "epoch": 0.08305634910132563,
      "grad_norm": 3.5472726821899414,
      "learning_rate": 9.6522508644565e-07,
      "loss": 0.6982,
      "step": 5335
    },
    {
      "epoch": 0.08313419010329501,
      "grad_norm": 8.160552024841309,
      "learning_rate": 9.651431474410448e-07,
      "loss": 0.801,
      "step": 5340
    },
    {
      "epoch": 0.08321203110526439,
      "grad_norm": 5.538876056671143,
      "learning_rate": 9.650612084364398e-07,
      "loss": 0.7735,
      "step": 5345
    },
    {
      "epoch": 0.08328987210723376,
      "grad_norm": 5.047536849975586,
      "learning_rate": 9.64979269431835e-07,
      "loss": 0.9173,
      "step": 5350
    },
    {
      "epoch": 0.08336771310920314,
      "grad_norm": 3.526073932647705,
      "learning_rate": 9.648973304272298e-07,
      "loss": 0.9703,
      "step": 5355
    },
    {
      "epoch": 0.08344555411117252,
      "grad_norm": 12.305222511291504,
      "learning_rate": 9.648153914226249e-07,
      "loss": 0.8025,
      "step": 5360
    },
    {
      "epoch": 0.0835233951131419,
      "grad_norm": 4.003148078918457,
      "learning_rate": 9.6473345241802e-07,
      "loss": 0.8206,
      "step": 5365
    },
    {
      "epoch": 0.08360123611511128,
      "grad_norm": 3.3531124591827393,
      "learning_rate": 9.64651513413415e-07,
      "loss": 0.7922,
      "step": 5370
    },
    {
      "epoch": 0.08367907711708066,
      "grad_norm": 3.3483853340148926,
      "learning_rate": 9.6456957440881e-07,
      "loss": 0.805,
      "step": 5375
    },
    {
      "epoch": 0.08375691811905003,
      "grad_norm": 3.580211639404297,
      "learning_rate": 9.644876354042052e-07,
      "loss": 0.8038,
      "step": 5380
    },
    {
      "epoch": 0.08383475912101941,
      "grad_norm": 4.441928863525391,
      "learning_rate": 9.644056963996e-07,
      "loss": 0.8188,
      "step": 5385
    },
    {
      "epoch": 0.08391260012298879,
      "grad_norm": 4.342660903930664,
      "learning_rate": 9.64323757394995e-07,
      "loss": 0.9376,
      "step": 5390
    },
    {
      "epoch": 0.08399044112495815,
      "grad_norm": 3.5513997077941895,
      "learning_rate": 9.642418183903902e-07,
      "loss": 0.8407,
      "step": 5395
    },
    {
      "epoch": 0.08406828212692753,
      "grad_norm": 3.7131507396698,
      "learning_rate": 9.64159879385785e-07,
      "loss": 0.8832,
      "step": 5400
    },
    {
      "epoch": 0.08414612312889691,
      "grad_norm": 4.675576686859131,
      "learning_rate": 9.6407794038118e-07,
      "loss": 0.8137,
      "step": 5405
    },
    {
      "epoch": 0.08422396413086629,
      "grad_norm": 5.775442600250244,
      "learning_rate": 9.639960013765752e-07,
      "loss": 0.8705,
      "step": 5410
    },
    {
      "epoch": 0.08430180513283567,
      "grad_norm": 3.2232508659362793,
      "learning_rate": 9.639140623719703e-07,
      "loss": 0.8266,
      "step": 5415
    },
    {
      "epoch": 0.08437964613480504,
      "grad_norm": 3.5636298656463623,
      "learning_rate": 9.638321233673653e-07,
      "loss": 0.739,
      "step": 5420
    },
    {
      "epoch": 0.08445748713677442,
      "grad_norm": 6.0133442878723145,
      "learning_rate": 9.637501843627604e-07,
      "loss": 0.7714,
      "step": 5425
    },
    {
      "epoch": 0.0845353281387438,
      "grad_norm": 3.2928476333618164,
      "learning_rate": 9.636682453581555e-07,
      "loss": 0.7455,
      "step": 5430
    },
    {
      "epoch": 0.08461316914071318,
      "grad_norm": 3.734174966812134,
      "learning_rate": 9.635863063535503e-07,
      "loss": 0.8668,
      "step": 5435
    },
    {
      "epoch": 0.08469101014268256,
      "grad_norm": 3.125318765640259,
      "learning_rate": 9.635043673489454e-07,
      "loss": 0.8308,
      "step": 5440
    },
    {
      "epoch": 0.08476885114465194,
      "grad_norm": 4.821923732757568,
      "learning_rate": 9.634224283443405e-07,
      "loss": 0.7993,
      "step": 5445
    },
    {
      "epoch": 0.08484669214662131,
      "grad_norm": 3.52372407913208,
      "learning_rate": 9.633404893397354e-07,
      "loss": 0.9816,
      "step": 5450
    },
    {
      "epoch": 0.08492453314859069,
      "grad_norm": 4.727661609649658,
      "learning_rate": 9.632585503351304e-07,
      "loss": 0.8819,
      "step": 5455
    },
    {
      "epoch": 0.08500237415056007,
      "grad_norm": 6.401661396026611,
      "learning_rate": 9.631766113305255e-07,
      "loss": 0.879,
      "step": 5460
    },
    {
      "epoch": 0.08508021515252945,
      "grad_norm": 3.203312873840332,
      "learning_rate": 9.630946723259206e-07,
      "loss": 0.7982,
      "step": 5465
    },
    {
      "epoch": 0.08515805615449883,
      "grad_norm": 4.19862174987793,
      "learning_rate": 9.630127333213157e-07,
      "loss": 0.8441,
      "step": 5470
    },
    {
      "epoch": 0.0852358971564682,
      "grad_norm": 3.9910812377929688,
      "learning_rate": 9.629307943167107e-07,
      "loss": 0.9595,
      "step": 5475
    },
    {
      "epoch": 0.08531373815843757,
      "grad_norm": 3.739917755126953,
      "learning_rate": 9.628488553121056e-07,
      "loss": 0.9027,
      "step": 5480
    },
    {
      "epoch": 0.08539157916040695,
      "grad_norm": 3.8963537216186523,
      "learning_rate": 9.627669163075007e-07,
      "loss": 0.7635,
      "step": 5485
    },
    {
      "epoch": 0.08546942016237633,
      "grad_norm": 11.066873550415039,
      "learning_rate": 9.626849773028957e-07,
      "loss": 0.828,
      "step": 5490
    },
    {
      "epoch": 0.0855472611643457,
      "grad_norm": 5.069997310638428,
      "learning_rate": 9.626030382982906e-07,
      "loss": 0.892,
      "step": 5495
    },
    {
      "epoch": 0.08562510216631508,
      "grad_norm": 2.1011128425598145,
      "learning_rate": 9.625210992936857e-07,
      "loss": 0.7157,
      "step": 5500
    },
    {
      "epoch": 0.08570294316828446,
      "grad_norm": 5.490849494934082,
      "learning_rate": 9.624391602890808e-07,
      "loss": 0.7912,
      "step": 5505
    },
    {
      "epoch": 0.08578078417025384,
      "grad_norm": 5.189328670501709,
      "learning_rate": 9.623572212844758e-07,
      "loss": 0.6751,
      "step": 5510
    },
    {
      "epoch": 0.08585862517222322,
      "grad_norm": 3.257615089416504,
      "learning_rate": 9.62275282279871e-07,
      "loss": 0.7616,
      "step": 5515
    },
    {
      "epoch": 0.0859364661741926,
      "grad_norm": 3.8244619369506836,
      "learning_rate": 9.621933432752658e-07,
      "loss": 0.7356,
      "step": 5520
    },
    {
      "epoch": 0.08601430717616197,
      "grad_norm": 4.616507530212402,
      "learning_rate": 9.621114042706608e-07,
      "loss": 0.8844,
      "step": 5525
    },
    {
      "epoch": 0.08609214817813135,
      "grad_norm": 8.950932502746582,
      "learning_rate": 9.62029465266056e-07,
      "loss": 0.8493,
      "step": 5530
    },
    {
      "epoch": 0.08616998918010073,
      "grad_norm": 3.257582187652588,
      "learning_rate": 9.61947526261451e-07,
      "loss": 0.7523,
      "step": 5535
    },
    {
      "epoch": 0.0862478301820701,
      "grad_norm": 9.792999267578125,
      "learning_rate": 9.61865587256846e-07,
      "loss": 0.8037,
      "step": 5540
    },
    {
      "epoch": 0.08632567118403948,
      "grad_norm": 3.294633626937866,
      "learning_rate": 9.61783648252241e-07,
      "loss": 0.7864,
      "step": 5545
    },
    {
      "epoch": 0.08640351218600886,
      "grad_norm": 3.527974843978882,
      "learning_rate": 9.61701709247636e-07,
      "loss": 0.7226,
      "step": 5550
    },
    {
      "epoch": 0.08648135318797824,
      "grad_norm": 10.905069351196289,
      "learning_rate": 9.61619770243031e-07,
      "loss": 0.8916,
      "step": 5555
    },
    {
      "epoch": 0.08655919418994762,
      "grad_norm": 5.191342830657959,
      "learning_rate": 9.61537831238426e-07,
      "loss": 0.7855,
      "step": 5560
    },
    {
      "epoch": 0.08663703519191698,
      "grad_norm": 4.45928430557251,
      "learning_rate": 9.61455892233821e-07,
      "loss": 0.6989,
      "step": 5565
    },
    {
      "epoch": 0.08671487619388636,
      "grad_norm": 3.573596954345703,
      "learning_rate": 9.61373953229216e-07,
      "loss": 0.718,
      "step": 5570
    },
    {
      "epoch": 0.08679271719585574,
      "grad_norm": 3.490968942642212,
      "learning_rate": 9.612920142246112e-07,
      "loss": 0.9475,
      "step": 5575
    },
    {
      "epoch": 0.08687055819782512,
      "grad_norm": 5.315331935882568,
      "learning_rate": 9.612100752200062e-07,
      "loss": 0.7643,
      "step": 5580
    },
    {
      "epoch": 0.0869483991997945,
      "grad_norm": 5.576305389404297,
      "learning_rate": 9.611281362154013e-07,
      "loss": 0.8273,
      "step": 5585
    },
    {
      "epoch": 0.08702624020176387,
      "grad_norm": 3.3249528408050537,
      "learning_rate": 9.610461972107962e-07,
      "loss": 0.7327,
      "step": 5590
    },
    {
      "epoch": 0.08710408120373325,
      "grad_norm": 5.021561622619629,
      "learning_rate": 9.609642582061913e-07,
      "loss": 0.7693,
      "step": 5595
    },
    {
      "epoch": 0.08718192220570263,
      "grad_norm": 4.7560834884643555,
      "learning_rate": 9.608823192015863e-07,
      "loss": 0.7056,
      "step": 5600
    },
    {
      "epoch": 0.08725976320767201,
      "grad_norm": 4.182785987854004,
      "learning_rate": 9.608003801969812e-07,
      "loss": 0.7807,
      "step": 5605
    },
    {
      "epoch": 0.08733760420964139,
      "grad_norm": 3.104510545730591,
      "learning_rate": 9.607184411923763e-07,
      "loss": 0.771,
      "step": 5610
    },
    {
      "epoch": 0.08741544521161076,
      "grad_norm": 4.316323280334473,
      "learning_rate": 9.606365021877713e-07,
      "loss": 0.8881,
      "step": 5615
    },
    {
      "epoch": 0.08749328621358014,
      "grad_norm": 4.004445552825928,
      "learning_rate": 9.605545631831664e-07,
      "loss": 0.8125,
      "step": 5620
    },
    {
      "epoch": 0.08757112721554952,
      "grad_norm": 5.998608112335205,
      "learning_rate": 9.604726241785615e-07,
      "loss": 0.8896,
      "step": 5625
    },
    {
      "epoch": 0.0876489682175189,
      "grad_norm": 6.251708507537842,
      "learning_rate": 9.603906851739566e-07,
      "loss": 0.97,
      "step": 5630
    },
    {
      "epoch": 0.08772680921948828,
      "grad_norm": 4.202377796173096,
      "learning_rate": 9.603087461693514e-07,
      "loss": 0.9106,
      "step": 5635
    },
    {
      "epoch": 0.08780465022145766,
      "grad_norm": 7.748138427734375,
      "learning_rate": 9.602268071647465e-07,
      "loss": 0.8634,
      "step": 5640
    },
    {
      "epoch": 0.08788249122342703,
      "grad_norm": 5.498707294464111,
      "learning_rate": 9.601448681601416e-07,
      "loss": 0.7508,
      "step": 5645
    },
    {
      "epoch": 0.08796033222539641,
      "grad_norm": 3.504171133041382,
      "learning_rate": 9.600629291555364e-07,
      "loss": 0.8638,
      "step": 5650
    },
    {
      "epoch": 0.08803817322736578,
      "grad_norm": 4.243772983551025,
      "learning_rate": 9.599809901509315e-07,
      "loss": 0.8651,
      "step": 5655
    },
    {
      "epoch": 0.08811601422933515,
      "grad_norm": 2.332878589630127,
      "learning_rate": 9.598990511463266e-07,
      "loss": 0.6884,
      "step": 5660
    },
    {
      "epoch": 0.08819385523130453,
      "grad_norm": 5.470850944519043,
      "learning_rate": 9.598171121417217e-07,
      "loss": 0.9159,
      "step": 5665
    },
    {
      "epoch": 0.08827169623327391,
      "grad_norm": 3.167588710784912,
      "learning_rate": 9.597351731371167e-07,
      "loss": 0.7925,
      "step": 5670
    },
    {
      "epoch": 0.08834953723524329,
      "grad_norm": 8.463876724243164,
      "learning_rate": 9.596532341325118e-07,
      "loss": 0.773,
      "step": 5675
    },
    {
      "epoch": 0.08842737823721267,
      "grad_norm": 5.318755626678467,
      "learning_rate": 9.595712951279067e-07,
      "loss": 0.8523,
      "step": 5680
    },
    {
      "epoch": 0.08850521923918205,
      "grad_norm": 8.276546478271484,
      "learning_rate": 9.594893561233018e-07,
      "loss": 0.7573,
      "step": 5685
    },
    {
      "epoch": 0.08858306024115142,
      "grad_norm": 3.6410884857177734,
      "learning_rate": 9.594074171186968e-07,
      "loss": 1.0501,
      "step": 5690
    },
    {
      "epoch": 0.0886609012431208,
      "grad_norm": 4.722231388092041,
      "learning_rate": 9.59325478114092e-07,
      "loss": 0.8005,
      "step": 5695
    },
    {
      "epoch": 0.08873874224509018,
      "grad_norm": 4.808355808258057,
      "learning_rate": 9.592435391094868e-07,
      "loss": 0.8858,
      "step": 5700
    },
    {
      "epoch": 0.08881658324705956,
      "grad_norm": 3.3222663402557373,
      "learning_rate": 9.591616001048818e-07,
      "loss": 0.8279,
      "step": 5705
    },
    {
      "epoch": 0.08889442424902894,
      "grad_norm": 6.019637584686279,
      "learning_rate": 9.59079661100277e-07,
      "loss": 0.8425,
      "step": 5710
    },
    {
      "epoch": 0.08897226525099831,
      "grad_norm": 3.4430840015411377,
      "learning_rate": 9.58997722095672e-07,
      "loss": 0.8277,
      "step": 5715
    },
    {
      "epoch": 0.08905010625296769,
      "grad_norm": 2.4599595069885254,
      "learning_rate": 9.58915783091067e-07,
      "loss": 0.8047,
      "step": 5720
    },
    {
      "epoch": 0.08912794725493707,
      "grad_norm": 5.123390197753906,
      "learning_rate": 9.58833844086462e-07,
      "loss": 0.7134,
      "step": 5725
    },
    {
      "epoch": 0.08920578825690645,
      "grad_norm": 5.264007568359375,
      "learning_rate": 9.58751905081857e-07,
      "loss": 0.914,
      "step": 5730
    },
    {
      "epoch": 0.08928362925887583,
      "grad_norm": 2.6615512371063232,
      "learning_rate": 9.58669966077252e-07,
      "loss": 0.8613,
      "step": 5735
    },
    {
      "epoch": 0.08936147026084519,
      "grad_norm": 2.7306411266326904,
      "learning_rate": 9.585880270726472e-07,
      "loss": 0.6556,
      "step": 5740
    },
    {
      "epoch": 0.08943931126281457,
      "grad_norm": 3.124546766281128,
      "learning_rate": 9.58506088068042e-07,
      "loss": 0.7402,
      "step": 5745
    },
    {
      "epoch": 0.08951715226478395,
      "grad_norm": 3.2253921031951904,
      "learning_rate": 9.58424149063437e-07,
      "loss": 0.8324,
      "step": 5750
    },
    {
      "epoch": 0.08959499326675333,
      "grad_norm": 4.765871524810791,
      "learning_rate": 9.583422100588322e-07,
      "loss": 0.8187,
      "step": 5755
    },
    {
      "epoch": 0.0896728342687227,
      "grad_norm": 5.348093509674072,
      "learning_rate": 9.582602710542272e-07,
      "loss": 0.8607,
      "step": 5760
    },
    {
      "epoch": 0.08975067527069208,
      "grad_norm": 3.306044340133667,
      "learning_rate": 9.581783320496221e-07,
      "loss": 0.8503,
      "step": 5765
    },
    {
      "epoch": 0.08982851627266146,
      "grad_norm": 5.045707702636719,
      "learning_rate": 9.580963930450172e-07,
      "loss": 0.8423,
      "step": 5770
    },
    {
      "epoch": 0.08990635727463084,
      "grad_norm": 6.576409816741943,
      "learning_rate": 9.580144540404123e-07,
      "loss": 0.9193,
      "step": 5775
    },
    {
      "epoch": 0.08998419827660022,
      "grad_norm": 7.775379180908203,
      "learning_rate": 9.579325150358073e-07,
      "loss": 0.8257,
      "step": 5780
    },
    {
      "epoch": 0.0900620392785696,
      "grad_norm": 8.183690071105957,
      "learning_rate": 9.578505760312024e-07,
      "loss": 0.7978,
      "step": 5785
    },
    {
      "epoch": 0.09013988028053897,
      "grad_norm": 5.8509087562561035,
      "learning_rate": 9.577686370265975e-07,
      "loss": 0.7486,
      "step": 5790
    },
    {
      "epoch": 0.09021772128250835,
      "grad_norm": 7.354578971862793,
      "learning_rate": 9.576866980219923e-07,
      "loss": 0.8432,
      "step": 5795
    },
    {
      "epoch": 0.09029556228447773,
      "grad_norm": 3.6449766159057617,
      "learning_rate": 9.576047590173874e-07,
      "loss": 0.8296,
      "step": 5800
    },
    {
      "epoch": 0.09037340328644711,
      "grad_norm": 9.557231903076172,
      "learning_rate": 9.575228200127825e-07,
      "loss": 0.7974,
      "step": 5805
    },
    {
      "epoch": 0.09045124428841649,
      "grad_norm": 4.868302345275879,
      "learning_rate": 9.574408810081774e-07,
      "loss": 0.9218,
      "step": 5810
    },
    {
      "epoch": 0.09052908529038586,
      "grad_norm": 4.260608196258545,
      "learning_rate": 9.573589420035724e-07,
      "loss": 0.7428,
      "step": 5815
    },
    {
      "epoch": 0.09060692629235524,
      "grad_norm": 3.023204803466797,
      "learning_rate": 9.572770029989675e-07,
      "loss": 0.8903,
      "step": 5820
    },
    {
      "epoch": 0.0906847672943246,
      "grad_norm": 3.036348819732666,
      "learning_rate": 9.571950639943626e-07,
      "loss": 0.89,
      "step": 5825
    },
    {
      "epoch": 0.09076260829629398,
      "grad_norm": 4.273719310760498,
      "learning_rate": 9.571131249897577e-07,
      "loss": 0.8611,
      "step": 5830
    },
    {
      "epoch": 0.09084044929826336,
      "grad_norm": 3.51576828956604,
      "learning_rate": 9.570311859851527e-07,
      "loss": 0.829,
      "step": 5835
    },
    {
      "epoch": 0.09091829030023274,
      "grad_norm": 3.904651641845703,
      "learning_rate": 9.569492469805476e-07,
      "loss": 0.9588,
      "step": 5840
    },
    {
      "epoch": 0.09099613130220212,
      "grad_norm": 3.824842691421509,
      "learning_rate": 9.568673079759427e-07,
      "loss": 0.7191,
      "step": 5845
    },
    {
      "epoch": 0.0910739723041715,
      "grad_norm": 5.210089683532715,
      "learning_rate": 9.567853689713377e-07,
      "loss": 0.8695,
      "step": 5850
    },
    {
      "epoch": 0.09115181330614087,
      "grad_norm": 2.847330093383789,
      "learning_rate": 9.567034299667326e-07,
      "loss": 0.8151,
      "step": 5855
    },
    {
      "epoch": 0.09122965430811025,
      "grad_norm": 4.297481060028076,
      "learning_rate": 9.566214909621277e-07,
      "loss": 0.91,
      "step": 5860
    },
    {
      "epoch": 0.09130749531007963,
      "grad_norm": 5.124939918518066,
      "learning_rate": 9.565395519575228e-07,
      "loss": 0.7646,
      "step": 5865
    },
    {
      "epoch": 0.09138533631204901,
      "grad_norm": 4.084904193878174,
      "learning_rate": 9.564576129529178e-07,
      "loss": 0.8639,
      "step": 5870
    },
    {
      "epoch": 0.09146317731401839,
      "grad_norm": 7.2979254722595215,
      "learning_rate": 9.56375673948313e-07,
      "loss": 0.909,
      "step": 5875
    },
    {
      "epoch": 0.09154101831598777,
      "grad_norm": 3.893127202987671,
      "learning_rate": 9.56293734943708e-07,
      "loss": 0.7701,
      "step": 5880
    },
    {
      "epoch": 0.09161885931795714,
      "grad_norm": 3.9665653705596924,
      "learning_rate": 9.562117959391028e-07,
      "loss": 0.8257,
      "step": 5885
    },
    {
      "epoch": 0.09169670031992652,
      "grad_norm": 3.298375129699707,
      "learning_rate": 9.56129856934498e-07,
      "loss": 0.8005,
      "step": 5890
    },
    {
      "epoch": 0.0917745413218959,
      "grad_norm": 3.643336057662964,
      "learning_rate": 9.56047917929893e-07,
      "loss": 0.8725,
      "step": 5895
    },
    {
      "epoch": 0.09185238232386528,
      "grad_norm": 11.060576438903809,
      "learning_rate": 9.559659789252879e-07,
      "loss": 0.8681,
      "step": 5900
    },
    {
      "epoch": 0.09193022332583466,
      "grad_norm": 4.433795928955078,
      "learning_rate": 9.55884039920683e-07,
      "loss": 0.8007,
      "step": 5905
    },
    {
      "epoch": 0.09200806432780402,
      "grad_norm": 6.115171909332275,
      "learning_rate": 9.55802100916078e-07,
      "loss": 0.73,
      "step": 5910
    },
    {
      "epoch": 0.0920859053297734,
      "grad_norm": 4.329387187957764,
      "learning_rate": 9.55720161911473e-07,
      "loss": 0.7698,
      "step": 5915
    },
    {
      "epoch": 0.09216374633174278,
      "grad_norm": 4.206638813018799,
      "learning_rate": 9.556382229068682e-07,
      "loss": 0.7973,
      "step": 5920
    },
    {
      "epoch": 0.09224158733371216,
      "grad_norm": 3.0813913345336914,
      "learning_rate": 9.55556283902263e-07,
      "loss": 0.7804,
      "step": 5925
    },
    {
      "epoch": 0.09231942833568153,
      "grad_norm": 6.411551475524902,
      "learning_rate": 9.55474344897658e-07,
      "loss": 0.8881,
      "step": 5930
    },
    {
      "epoch": 0.09239726933765091,
      "grad_norm": 2.5208792686462402,
      "learning_rate": 9.553924058930532e-07,
      "loss": 0.841,
      "step": 5935
    },
    {
      "epoch": 0.09247511033962029,
      "grad_norm": 2.8447041511535645,
      "learning_rate": 9.553104668884482e-07,
      "loss": 0.791,
      "step": 5940
    },
    {
      "epoch": 0.09255295134158967,
      "grad_norm": 4.374822616577148,
      "learning_rate": 9.552285278838433e-07,
      "loss": 0.8153,
      "step": 5945
    },
    {
      "epoch": 0.09263079234355905,
      "grad_norm": 6.252150058746338,
      "learning_rate": 9.551465888792382e-07,
      "loss": 0.8223,
      "step": 5950
    },
    {
      "epoch": 0.09270863334552842,
      "grad_norm": 3.3018994331359863,
      "learning_rate": 9.550646498746333e-07,
      "loss": 0.9067,
      "step": 5955
    },
    {
      "epoch": 0.0927864743474978,
      "grad_norm": 4.026679515838623,
      "learning_rate": 9.549827108700283e-07,
      "loss": 0.7962,
      "step": 5960
    },
    {
      "epoch": 0.09286431534946718,
      "grad_norm": 3.2476413249969482,
      "learning_rate": 9.549007718654232e-07,
      "loss": 0.8074,
      "step": 5965
    },
    {
      "epoch": 0.09294215635143656,
      "grad_norm": 2.852954149246216,
      "learning_rate": 9.548188328608183e-07,
      "loss": 0.8071,
      "step": 5970
    },
    {
      "epoch": 0.09301999735340594,
      "grad_norm": 3.4490416049957275,
      "learning_rate": 9.547368938562133e-07,
      "loss": 0.9519,
      "step": 5975
    },
    {
      "epoch": 0.09309783835537531,
      "grad_norm": 2.473008155822754,
      "learning_rate": 9.546549548516084e-07,
      "loss": 0.8446,
      "step": 5980
    },
    {
      "epoch": 0.09317567935734469,
      "grad_norm": 7.381313800811768,
      "learning_rate": 9.545730158470035e-07,
      "loss": 0.802,
      "step": 5985
    },
    {
      "epoch": 0.09325352035931407,
      "grad_norm": 4.133596897125244,
      "learning_rate": 9.544910768423986e-07,
      "loss": 0.7167,
      "step": 5990
    },
    {
      "epoch": 0.09333136136128345,
      "grad_norm": 4.466127872467041,
      "learning_rate": 9.544091378377934e-07,
      "loss": 0.9329,
      "step": 5995
    },
    {
      "epoch": 0.09340920236325281,
      "grad_norm": 4.252684593200684,
      "learning_rate": 9.543271988331885e-07,
      "loss": 0.856,
      "step": 6000
    },
    {
      "epoch": 0.09348704336522219,
      "grad_norm": 3.630127429962158,
      "learning_rate": 9.542452598285836e-07,
      "loss": 0.8738,
      "step": 6005
    },
    {
      "epoch": 0.09356488436719157,
      "grad_norm": 8.133733749389648,
      "learning_rate": 9.541633208239784e-07,
      "loss": 0.8339,
      "step": 6010
    },
    {
      "epoch": 0.09364272536916095,
      "grad_norm": 11.18271541595459,
      "learning_rate": 9.540813818193735e-07,
      "loss": 0.7978,
      "step": 6015
    },
    {
      "epoch": 0.09372056637113033,
      "grad_norm": 6.3515214920043945,
      "learning_rate": 9.539994428147686e-07,
      "loss": 0.7742,
      "step": 6020
    },
    {
      "epoch": 0.0937984073730997,
      "grad_norm": 3.030446767807007,
      "learning_rate": 9.539175038101637e-07,
      "loss": 0.6648,
      "step": 6025
    },
    {
      "epoch": 0.09387624837506908,
      "grad_norm": 5.10403299331665,
      "learning_rate": 9.538355648055587e-07,
      "loss": 0.9393,
      "step": 6030
    },
    {
      "epoch": 0.09395408937703846,
      "grad_norm": 2.5483992099761963,
      "learning_rate": 9.537536258009538e-07,
      "loss": 0.8155,
      "step": 6035
    },
    {
      "epoch": 0.09403193037900784,
      "grad_norm": 6.4216814041137695,
      "learning_rate": 9.536716867963488e-07,
      "loss": 0.9805,
      "step": 6040
    },
    {
      "epoch": 0.09410977138097722,
      "grad_norm": 8.188408851623535,
      "learning_rate": 9.535897477917439e-07,
      "loss": 0.8092,
      "step": 6045
    },
    {
      "epoch": 0.0941876123829466,
      "grad_norm": 4.022781848907471,
      "learning_rate": 9.535078087871387e-07,
      "loss": 0.8009,
      "step": 6050
    },
    {
      "epoch": 0.09426545338491597,
      "grad_norm": 3.303135871887207,
      "learning_rate": 9.534258697825338e-07,
      "loss": 0.7763,
      "step": 6055
    },
    {
      "epoch": 0.09434329438688535,
      "grad_norm": 5.49419641494751,
      "learning_rate": 9.533439307779289e-07,
      "loss": 0.8246,
      "step": 6060
    },
    {
      "epoch": 0.09442113538885473,
      "grad_norm": 4.208410739898682,
      "learning_rate": 9.532619917733238e-07,
      "loss": 0.8875,
      "step": 6065
    },
    {
      "epoch": 0.09449897639082411,
      "grad_norm": 2.5194616317749023,
      "learning_rate": 9.531800527687189e-07,
      "loss": 0.8782,
      "step": 6070
    },
    {
      "epoch": 0.09457681739279349,
      "grad_norm": 3.4753055572509766,
      "learning_rate": 9.53098113764114e-07,
      "loss": 0.7719,
      "step": 6075
    },
    {
      "epoch": 0.09465465839476286,
      "grad_norm": 4.319244861602783,
      "learning_rate": 9.53016174759509e-07,
      "loss": 0.9141,
      "step": 6080
    },
    {
      "epoch": 0.09473249939673223,
      "grad_norm": 2.9613096714019775,
      "learning_rate": 9.52934235754904e-07,
      "loss": 0.7196,
      "step": 6085
    },
    {
      "epoch": 0.0948103403987016,
      "grad_norm": 6.506518840789795,
      "learning_rate": 9.52852296750299e-07,
      "loss": 0.8589,
      "step": 6090
    },
    {
      "epoch": 0.09488818140067098,
      "grad_norm": 7.105751037597656,
      "learning_rate": 9.527703577456941e-07,
      "loss": 0.8392,
      "step": 6095
    },
    {
      "epoch": 0.09496602240264036,
      "grad_norm": 3.0105667114257812,
      "learning_rate": 9.52688418741089e-07,
      "loss": 0.6962,
      "step": 6100
    },
    {
      "epoch": 0.09504386340460974,
      "grad_norm": 7.148667812347412,
      "learning_rate": 9.526064797364841e-07,
      "loss": 0.7958,
      "step": 6105
    },
    {
      "epoch": 0.09512170440657912,
      "grad_norm": 11.00757122039795,
      "learning_rate": 9.525245407318792e-07,
      "loss": 0.8516,
      "step": 6110
    },
    {
      "epoch": 0.0951995454085485,
      "grad_norm": 5.520313739776611,
      "learning_rate": 9.524426017272742e-07,
      "loss": 0.9142,
      "step": 6115
    },
    {
      "epoch": 0.09527738641051788,
      "grad_norm": 3.418109893798828,
      "learning_rate": 9.523606627226692e-07,
      "loss": 0.7326,
      "step": 6120
    },
    {
      "epoch": 0.09535522741248725,
      "grad_norm": 3.9674932956695557,
      "learning_rate": 9.522787237180643e-07,
      "loss": 0.8561,
      "step": 6125
    },
    {
      "epoch": 0.09543306841445663,
      "grad_norm": 3.5758800506591797,
      "learning_rate": 9.521967847134592e-07,
      "loss": 0.8119,
      "step": 6130
    },
    {
      "epoch": 0.09551090941642601,
      "grad_norm": 4.39679479598999,
      "learning_rate": 9.521148457088543e-07,
      "loss": 0.8427,
      "step": 6135
    },
    {
      "epoch": 0.09558875041839539,
      "grad_norm": 5.498544216156006,
      "learning_rate": 9.520329067042493e-07,
      "loss": 0.8786,
      "step": 6140
    },
    {
      "epoch": 0.09566659142036477,
      "grad_norm": 3.5914194583892822,
      "learning_rate": 9.519509676996443e-07,
      "loss": 0.8465,
      "step": 6145
    },
    {
      "epoch": 0.09574443242233414,
      "grad_norm": 5.639887809753418,
      "learning_rate": 9.518690286950394e-07,
      "loss": 0.764,
      "step": 6150
    },
    {
      "epoch": 0.09582227342430352,
      "grad_norm": 7.638282299041748,
      "learning_rate": 9.517870896904345e-07,
      "loss": 0.858,
      "step": 6155
    },
    {
      "epoch": 0.0959001144262729,
      "grad_norm": 3.7768096923828125,
      "learning_rate": 9.517051506858294e-07,
      "loss": 0.8739,
      "step": 6160
    },
    {
      "epoch": 0.09597795542824228,
      "grad_norm": 3.388122081756592,
      "learning_rate": 9.516232116812245e-07,
      "loss": 0.7646,
      "step": 6165
    },
    {
      "epoch": 0.09605579643021164,
      "grad_norm": 3.0460891723632812,
      "learning_rate": 9.515412726766195e-07,
      "loss": 0.9288,
      "step": 6170
    },
    {
      "epoch": 0.09613363743218102,
      "grad_norm": 3.9041805267333984,
      "learning_rate": 9.514593336720144e-07,
      "loss": 0.735,
      "step": 6175
    },
    {
      "epoch": 0.0962114784341504,
      "grad_norm": 3.894850254058838,
      "learning_rate": 9.513773946674095e-07,
      "loss": 0.7104,
      "step": 6180
    },
    {
      "epoch": 0.09628931943611978,
      "grad_norm": 3.6872172355651855,
      "learning_rate": 9.512954556628046e-07,
      "loss": 0.7743,
      "step": 6185
    },
    {
      "epoch": 0.09636716043808916,
      "grad_norm": 2.9574503898620605,
      "learning_rate": 9.512135166581996e-07,
      "loss": 0.8923,
      "step": 6190
    },
    {
      "epoch": 0.09644500144005853,
      "grad_norm": 3.6874301433563232,
      "learning_rate": 9.511315776535946e-07,
      "loss": 0.7798,
      "step": 6195
    },
    {
      "epoch": 0.09652284244202791,
      "grad_norm": 5.110114574432373,
      "learning_rate": 9.510496386489897e-07,
      "loss": 0.8202,
      "step": 6200
    },
    {
      "epoch": 0.09660068344399729,
      "grad_norm": 4.243130683898926,
      "learning_rate": 9.509676996443848e-07,
      "loss": 0.9956,
      "step": 6205
    },
    {
      "epoch": 0.09667852444596667,
      "grad_norm": 9.388118743896484,
      "learning_rate": 9.508857606397796e-07,
      "loss": 0.8103,
      "step": 6210
    },
    {
      "epoch": 0.09675636544793605,
      "grad_norm": 4.399020671844482,
      "learning_rate": 9.508038216351747e-07,
      "loss": 0.9899,
      "step": 6215
    },
    {
      "epoch": 0.09683420644990542,
      "grad_norm": 5.260294437408447,
      "learning_rate": 9.507218826305698e-07,
      "loss": 0.7776,
      "step": 6220
    },
    {
      "epoch": 0.0969120474518748,
      "grad_norm": 2.903243064880371,
      "learning_rate": 9.506399436259648e-07,
      "loss": 0.8065,
      "step": 6225
    },
    {
      "epoch": 0.09698988845384418,
      "grad_norm": 7.704418182373047,
      "learning_rate": 9.505580046213598e-07,
      "loss": 0.6625,
      "step": 6230
    },
    {
      "epoch": 0.09706772945581356,
      "grad_norm": 3.7152814865112305,
      "learning_rate": 9.504760656167549e-07,
      "loss": 0.8702,
      "step": 6235
    },
    {
      "epoch": 0.09714557045778294,
      "grad_norm": 6.636418342590332,
      "learning_rate": 9.503941266121499e-07,
      "loss": 0.7592,
      "step": 6240
    },
    {
      "epoch": 0.09722341145975232,
      "grad_norm": 5.01901388168335,
      "learning_rate": 9.50312187607545e-07,
      "loss": 0.8377,
      "step": 6245
    },
    {
      "epoch": 0.0973012524617217,
      "grad_norm": 6.149816989898682,
      "learning_rate": 9.502302486029399e-07,
      "loss": 0.7698,
      "step": 6250
    },
    {
      "epoch": 0.09737909346369107,
      "grad_norm": 4.017423152923584,
      "learning_rate": 9.501483095983349e-07,
      "loss": 0.7522,
      "step": 6255
    },
    {
      "epoch": 0.09745693446566044,
      "grad_norm": 3.395038366317749,
      "learning_rate": 9.5006637059373e-07,
      "loss": 0.7814,
      "step": 6260
    },
    {
      "epoch": 0.09753477546762981,
      "grad_norm": 8.359529495239258,
      "learning_rate": 9.49984431589125e-07,
      "loss": 0.8817,
      "step": 6265
    },
    {
      "epoch": 0.09761261646959919,
      "grad_norm": 5.801593780517578,
      "learning_rate": 9.4990249258452e-07,
      "loss": 0.8346,
      "step": 6270
    },
    {
      "epoch": 0.09769045747156857,
      "grad_norm": 3.440136432647705,
      "learning_rate": 9.498205535799151e-07,
      "loss": 0.7793,
      "step": 6275
    },
    {
      "epoch": 0.09776829847353795,
      "grad_norm": 3.3891918659210205,
      "learning_rate": 9.497386145753102e-07,
      "loss": 0.7846,
      "step": 6280
    },
    {
      "epoch": 0.09784613947550733,
      "grad_norm": 3.6862120628356934,
      "learning_rate": 9.496566755707051e-07,
      "loss": 0.8316,
      "step": 6285
    },
    {
      "epoch": 0.0979239804774767,
      "grad_norm": 4.2608642578125,
      "learning_rate": 9.495747365661001e-07,
      "loss": 0.8244,
      "step": 6290
    },
    {
      "epoch": 0.09800182147944608,
      "grad_norm": 4.8404459953308105,
      "learning_rate": 9.494927975614952e-07,
      "loss": 0.7919,
      "step": 6295
    },
    {
      "epoch": 0.09807966248141546,
      "grad_norm": 2.3203227519989014,
      "learning_rate": 9.494108585568901e-07,
      "loss": 0.7593,
      "step": 6300
    },
    {
      "epoch": 0.09815750348338484,
      "grad_norm": 2.870492935180664,
      "learning_rate": 9.493289195522852e-07,
      "loss": 0.7388,
      "step": 6305
    },
    {
      "epoch": 0.09823534448535422,
      "grad_norm": 3.6634552478790283,
      "learning_rate": 9.492469805476803e-07,
      "loss": 0.6632,
      "step": 6310
    },
    {
      "epoch": 0.0983131854873236,
      "grad_norm": 4.569755554199219,
      "learning_rate": 9.491650415430753e-07,
      "loss": 0.866,
      "step": 6315
    },
    {
      "epoch": 0.09839102648929297,
      "grad_norm": 3.319843053817749,
      "learning_rate": 9.490831025384703e-07,
      "loss": 0.8812,
      "step": 6320
    },
    {
      "epoch": 0.09846886749126235,
      "grad_norm": 5.647189140319824,
      "learning_rate": 9.490011635338654e-07,
      "loss": 0.893,
      "step": 6325
    },
    {
      "epoch": 0.09854670849323173,
      "grad_norm": 4.285895347595215,
      "learning_rate": 9.489192245292603e-07,
      "loss": 0.9623,
      "step": 6330
    },
    {
      "epoch": 0.09862454949520111,
      "grad_norm": 4.257463455200195,
      "learning_rate": 9.488372855246553e-07,
      "loss": 0.7959,
      "step": 6335
    },
    {
      "epoch": 0.09870239049717049,
      "grad_norm": 4.747158050537109,
      "learning_rate": 9.487553465200504e-07,
      "loss": 0.6869,
      "step": 6340
    },
    {
      "epoch": 0.09878023149913985,
      "grad_norm": 4.191068172454834,
      "learning_rate": 9.486734075154455e-07,
      "loss": 0.8163,
      "step": 6345
    },
    {
      "epoch": 0.09885807250110923,
      "grad_norm": 4.583565711975098,
      "learning_rate": 9.485914685108405e-07,
      "loss": 0.7348,
      "step": 6350
    },
    {
      "epoch": 0.0989359135030786,
      "grad_norm": 3.9108059406280518,
      "learning_rate": 9.485095295062355e-07,
      "loss": 0.7968,
      "step": 6355
    },
    {
      "epoch": 0.09901375450504799,
      "grad_norm": 5.722688674926758,
      "learning_rate": 9.484275905016306e-07,
      "loss": 0.8138,
      "step": 6360
    },
    {
      "epoch": 0.09909159550701736,
      "grad_norm": 2.972755193710327,
      "learning_rate": 9.483456514970256e-07,
      "loss": 0.795,
      "step": 6365
    },
    {
      "epoch": 0.09916943650898674,
      "grad_norm": 8.901226997375488,
      "learning_rate": 9.482637124924207e-07,
      "loss": 0.7491,
      "step": 6370
    },
    {
      "epoch": 0.09924727751095612,
      "grad_norm": 7.961559772491455,
      "learning_rate": 9.481817734878156e-07,
      "loss": 1.0205,
      "step": 6375
    },
    {
      "epoch": 0.0993251185129255,
      "grad_norm": 6.042298316955566,
      "learning_rate": 9.480998344832106e-07,
      "loss": 0.7984,
      "step": 6380
    },
    {
      "epoch": 0.09940295951489488,
      "grad_norm": 10.556697845458984,
      "learning_rate": 9.480178954786057e-07,
      "loss": 0.8639,
      "step": 6385
    },
    {
      "epoch": 0.09948080051686425,
      "grad_norm": 2.7401647567749023,
      "learning_rate": 9.479359564740007e-07,
      "loss": 0.799,
      "step": 6390
    },
    {
      "epoch": 0.09955864151883363,
      "grad_norm": 6.288196086883545,
      "learning_rate": 9.478540174693957e-07,
      "loss": 0.9338,
      "step": 6395
    },
    {
      "epoch": 0.09963648252080301,
      "grad_norm": 4.34282112121582,
      "learning_rate": 9.477720784647908e-07,
      "loss": 0.8355,
      "step": 6400
    },
    {
      "epoch": 0.09971432352277239,
      "grad_norm": 3.5038483142852783,
      "learning_rate": 9.476901394601859e-07,
      "loss": 0.7796,
      "step": 6405
    },
    {
      "epoch": 0.09979216452474177,
      "grad_norm": 4.715381622314453,
      "learning_rate": 9.476082004555808e-07,
      "loss": 0.815,
      "step": 6410
    },
    {
      "epoch": 0.09987000552671114,
      "grad_norm": 3.7928483486175537,
      "learning_rate": 9.475262614509758e-07,
      "loss": 0.8346,
      "step": 6415
    },
    {
      "epoch": 0.09994784652868052,
      "grad_norm": 5.622752666473389,
      "learning_rate": 9.474443224463709e-07,
      "loss": 0.8558,
      "step": 6420
    },
    {
      "epoch": 0.1000256875306499,
      "grad_norm": 5.325289726257324,
      "learning_rate": 9.473623834417658e-07,
      "loss": 0.8078,
      "step": 6425
    },
    {
      "epoch": 0.10010352853261927,
      "grad_norm": 3.5389554500579834,
      "learning_rate": 9.472804444371609e-07,
      "loss": 0.7461,
      "step": 6430
    },
    {
      "epoch": 0.10018136953458864,
      "grad_norm": 8.74923038482666,
      "learning_rate": 9.47198505432556e-07,
      "loss": 0.7609,
      "step": 6435
    },
    {
      "epoch": 0.10025921053655802,
      "grad_norm": 4.170187473297119,
      "learning_rate": 9.47116566427951e-07,
      "loss": 0.8243,
      "step": 6440
    },
    {
      "epoch": 0.1003370515385274,
      "grad_norm": 12.243910789489746,
      "learning_rate": 9.47034627423346e-07,
      "loss": 0.7365,
      "step": 6445
    },
    {
      "epoch": 0.10041489254049678,
      "grad_norm": 6.56355094909668,
      "learning_rate": 9.469526884187411e-07,
      "loss": 0.8693,
      "step": 6450
    },
    {
      "epoch": 0.10049273354246616,
      "grad_norm": 4.917191982269287,
      "learning_rate": 9.46870749414136e-07,
      "loss": 0.8323,
      "step": 6455
    },
    {
      "epoch": 0.10057057454443553,
      "grad_norm": 4.455476760864258,
      "learning_rate": 9.467888104095311e-07,
      "loss": 0.7539,
      "step": 6460
    },
    {
      "epoch": 0.10064841554640491,
      "grad_norm": 4.135006904602051,
      "learning_rate": 9.467068714049261e-07,
      "loss": 0.772,
      "step": 6465
    },
    {
      "epoch": 0.10072625654837429,
      "grad_norm": 5.814565658569336,
      "learning_rate": 9.466249324003212e-07,
      "loss": 0.8389,
      "step": 6470
    },
    {
      "epoch": 0.10080409755034367,
      "grad_norm": 3.4807469844818115,
      "learning_rate": 9.465429933957162e-07,
      "loss": 0.8365,
      "step": 6475
    },
    {
      "epoch": 0.10088193855231305,
      "grad_norm": 5.241673946380615,
      "learning_rate": 9.464610543911112e-07,
      "loss": 0.6689,
      "step": 6480
    },
    {
      "epoch": 0.10095977955428243,
      "grad_norm": 6.0900678634643555,
      "learning_rate": 9.463791153865063e-07,
      "loss": 0.813,
      "step": 6485
    },
    {
      "epoch": 0.1010376205562518,
      "grad_norm": 4.659064769744873,
      "learning_rate": 9.462971763819013e-07,
      "loss": 0.8624,
      "step": 6490
    },
    {
      "epoch": 0.10111546155822118,
      "grad_norm": 7.9358320236206055,
      "learning_rate": 9.462152373772963e-07,
      "loss": 0.7209,
      "step": 6495
    },
    {
      "epoch": 0.10119330256019056,
      "grad_norm": 3.9600491523742676,
      "learning_rate": 9.461332983726913e-07,
      "loss": 0.7995,
      "step": 6500
    },
    {
      "epoch": 0.10127114356215994,
      "grad_norm": 4.832655906677246,
      "learning_rate": 9.460513593680863e-07,
      "loss": 0.8935,
      "step": 6505
    },
    {
      "epoch": 0.10134898456412932,
      "grad_norm": 4.184332370758057,
      "learning_rate": 9.459694203634814e-07,
      "loss": 0.7468,
      "step": 6510
    },
    {
      "epoch": 0.10142682556609868,
      "grad_norm": 3.2403645515441895,
      "learning_rate": 9.458874813588765e-07,
      "loss": 0.7502,
      "step": 6515
    },
    {
      "epoch": 0.10150466656806806,
      "grad_norm": 6.45439338684082,
      "learning_rate": 9.458055423542714e-07,
      "loss": 0.8359,
      "step": 6520
    },
    {
      "epoch": 0.10158250757003744,
      "grad_norm": 3.9225785732269287,
      "learning_rate": 9.457236033496665e-07,
      "loss": 0.822,
      "step": 6525
    },
    {
      "epoch": 0.10166034857200681,
      "grad_norm": 6.211043834686279,
      "learning_rate": 9.456416643450616e-07,
      "loss": 0.7675,
      "step": 6530
    },
    {
      "epoch": 0.10173818957397619,
      "grad_norm": 5.109851360321045,
      "learning_rate": 9.455597253404564e-07,
      "loss": 0.8539,
      "step": 6535
    },
    {
      "epoch": 0.10181603057594557,
      "grad_norm": 3.1654608249664307,
      "learning_rate": 9.454777863358515e-07,
      "loss": 0.851,
      "step": 6540
    },
    {
      "epoch": 0.10189387157791495,
      "grad_norm": 6.30355167388916,
      "learning_rate": 9.453958473312466e-07,
      "loss": 0.8668,
      "step": 6545
    },
    {
      "epoch": 0.10197171257988433,
      "grad_norm": 2.9073293209075928,
      "learning_rate": 9.453139083266416e-07,
      "loss": 0.8354,
      "step": 6550
    },
    {
      "epoch": 0.1020495535818537,
      "grad_norm": 4.239645481109619,
      "learning_rate": 9.452319693220366e-07,
      "loss": 0.8394,
      "step": 6555
    },
    {
      "epoch": 0.10212739458382308,
      "grad_norm": 4.341432094573975,
      "learning_rate": 9.451500303174317e-07,
      "loss": 0.8248,
      "step": 6560
    },
    {
      "epoch": 0.10220523558579246,
      "grad_norm": 5.958523273468018,
      "learning_rate": 9.450680913128267e-07,
      "loss": 0.7696,
      "step": 6565
    },
    {
      "epoch": 0.10228307658776184,
      "grad_norm": 2.9546141624450684,
      "learning_rate": 9.449861523082217e-07,
      "loss": 0.8008,
      "step": 6570
    },
    {
      "epoch": 0.10236091758973122,
      "grad_norm": 3.216296672821045,
      "learning_rate": 9.449042133036167e-07,
      "loss": 0.8286,
      "step": 6575
    },
    {
      "epoch": 0.1024387585917006,
      "grad_norm": 5.784662246704102,
      "learning_rate": 9.448222742990117e-07,
      "loss": 0.8118,
      "step": 6580
    },
    {
      "epoch": 0.10251659959366997,
      "grad_norm": 18.038049697875977,
      "learning_rate": 9.447403352944068e-07,
      "loss": 0.7844,
      "step": 6585
    },
    {
      "epoch": 0.10259444059563935,
      "grad_norm": 3.434221029281616,
      "learning_rate": 9.446583962898018e-07,
      "loss": 0.7745,
      "step": 6590
    },
    {
      "epoch": 0.10267228159760873,
      "grad_norm": 3.0332863330841064,
      "learning_rate": 9.445764572851969e-07,
      "loss": 0.9026,
      "step": 6595
    },
    {
      "epoch": 0.10275012259957811,
      "grad_norm": 4.459526538848877,
      "learning_rate": 9.444945182805919e-07,
      "loss": 0.8231,
      "step": 6600
    },
    {
      "epoch": 0.10282796360154747,
      "grad_norm": 11.914100646972656,
      "learning_rate": 9.44412579275987e-07,
      "loss": 0.7191,
      "step": 6605
    },
    {
      "epoch": 0.10290580460351685,
      "grad_norm": 9.491118431091309,
      "learning_rate": 9.44330640271382e-07,
      "loss": 0.8438,
      "step": 6610
    },
    {
      "epoch": 0.10298364560548623,
      "grad_norm": 3.312546968460083,
      "learning_rate": 9.442487012667769e-07,
      "loss": 0.7754,
      "step": 6615
    },
    {
      "epoch": 0.10306148660745561,
      "grad_norm": 2.4198150634765625,
      "learning_rate": 9.44166762262172e-07,
      "loss": 0.7373,
      "step": 6620
    },
    {
      "epoch": 0.10313932760942499,
      "grad_norm": 3.8953001499176025,
      "learning_rate": 9.44084823257567e-07,
      "loss": 0.6972,
      "step": 6625
    },
    {
      "epoch": 0.10321716861139436,
      "grad_norm": 3.971245050430298,
      "learning_rate": 9.44002884252962e-07,
      "loss": 0.7959,
      "step": 6630
    },
    {
      "epoch": 0.10329500961336374,
      "grad_norm": 3.9119505882263184,
      "learning_rate": 9.439209452483571e-07,
      "loss": 0.8072,
      "step": 6635
    },
    {
      "epoch": 0.10337285061533312,
      "grad_norm": 3.6322784423828125,
      "learning_rate": 9.438390062437522e-07,
      "loss": 0.8389,
      "step": 6640
    },
    {
      "epoch": 0.1034506916173025,
      "grad_norm": 3.221548557281494,
      "learning_rate": 9.437570672391471e-07,
      "loss": 0.8113,
      "step": 6645
    },
    {
      "epoch": 0.10352853261927188,
      "grad_norm": 3.768453598022461,
      "learning_rate": 9.436751282345422e-07,
      "loss": 0.8859,
      "step": 6650
    },
    {
      "epoch": 0.10360637362124125,
      "grad_norm": 3.436704635620117,
      "learning_rate": 9.435931892299372e-07,
      "loss": 0.9308,
      "step": 6655
    },
    {
      "epoch": 0.10368421462321063,
      "grad_norm": 3.5479848384857178,
      "learning_rate": 9.435112502253321e-07,
      "loss": 0.8624,
      "step": 6660
    },
    {
      "epoch": 0.10376205562518001,
      "grad_norm": 5.35614538192749,
      "learning_rate": 9.434293112207272e-07,
      "loss": 0.8095,
      "step": 6665
    },
    {
      "epoch": 0.10383989662714939,
      "grad_norm": 3.0815038681030273,
      "learning_rate": 9.433473722161223e-07,
      "loss": 0.6846,
      "step": 6670
    },
    {
      "epoch": 0.10391773762911877,
      "grad_norm": 5.047412872314453,
      "learning_rate": 9.432654332115173e-07,
      "loss": 0.7785,
      "step": 6675
    },
    {
      "epoch": 0.10399557863108815,
      "grad_norm": 4.322173595428467,
      "learning_rate": 9.431834942069123e-07,
      "loss": 0.8362,
      "step": 6680
    },
    {
      "epoch": 0.10407341963305752,
      "grad_norm": 4.15039587020874,
      "learning_rate": 9.431015552023074e-07,
      "loss": 0.7475,
      "step": 6685
    },
    {
      "epoch": 0.10415126063502689,
      "grad_norm": 3.9758059978485107,
      "learning_rate": 9.430196161977024e-07,
      "loss": 0.8058,
      "step": 6690
    },
    {
      "epoch": 0.10422910163699627,
      "grad_norm": 3.778308629989624,
      "learning_rate": 9.429376771930975e-07,
      "loss": 0.9209,
      "step": 6695
    },
    {
      "epoch": 0.10430694263896564,
      "grad_norm": 7.6523566246032715,
      "learning_rate": 9.428557381884924e-07,
      "loss": 0.765,
      "step": 6700
    },
    {
      "epoch": 0.10438478364093502,
      "grad_norm": 4.295438289642334,
      "learning_rate": 9.427737991838874e-07,
      "loss": 0.8556,
      "step": 6705
    },
    {
      "epoch": 0.1044626246429044,
      "grad_norm": 9.01634407043457,
      "learning_rate": 9.426918601792825e-07,
      "loss": 0.7078,
      "step": 6710
    },
    {
      "epoch": 0.10454046564487378,
      "grad_norm": 4.1538987159729,
      "learning_rate": 9.426099211746775e-07,
      "loss": 1.0486,
      "step": 6715
    },
    {
      "epoch": 0.10461830664684316,
      "grad_norm": 5.460824489593506,
      "learning_rate": 9.425279821700726e-07,
      "loss": 0.7296,
      "step": 6720
    },
    {
      "epoch": 0.10469614764881253,
      "grad_norm": 8.995347023010254,
      "learning_rate": 9.424460431654676e-07,
      "loss": 0.7228,
      "step": 6725
    },
    {
      "epoch": 0.10477398865078191,
      "grad_norm": 3.254420042037964,
      "learning_rate": 9.423641041608627e-07,
      "loss": 0.8683,
      "step": 6730
    },
    {
      "epoch": 0.10485182965275129,
      "grad_norm": 3.987894058227539,
      "learning_rate": 9.422821651562577e-07,
      "loss": 0.7663,
      "step": 6735
    },
    {
      "epoch": 0.10492967065472067,
      "grad_norm": 3.244363307952881,
      "learning_rate": 9.422002261516526e-07,
      "loss": 0.8443,
      "step": 6740
    },
    {
      "epoch": 0.10500751165669005,
      "grad_norm": 4.715000152587891,
      "learning_rate": 9.421182871470477e-07,
      "loss": 0.8059,
      "step": 6745
    },
    {
      "epoch": 0.10508535265865943,
      "grad_norm": 5.014405727386475,
      "learning_rate": 9.420363481424427e-07,
      "loss": 0.8381,
      "step": 6750
    },
    {
      "epoch": 0.1051631936606288,
      "grad_norm": 4.085587978363037,
      "learning_rate": 9.419544091378377e-07,
      "loss": 0.8104,
      "step": 6755
    },
    {
      "epoch": 0.10524103466259818,
      "grad_norm": 12.998879432678223,
      "learning_rate": 9.418724701332328e-07,
      "loss": 0.8814,
      "step": 6760
    },
    {
      "epoch": 0.10531887566456756,
      "grad_norm": 5.057702541351318,
      "learning_rate": 9.417905311286279e-07,
      "loss": 0.7355,
      "step": 6765
    },
    {
      "epoch": 0.10539671666653694,
      "grad_norm": 4.116156578063965,
      "learning_rate": 9.417085921240228e-07,
      "loss": 0.907,
      "step": 6770
    },
    {
      "epoch": 0.1054745576685063,
      "grad_norm": 3.0526468753814697,
      "learning_rate": 9.416266531194179e-07,
      "loss": 0.832,
      "step": 6775
    },
    {
      "epoch": 0.10555239867047568,
      "grad_norm": 5.218168258666992,
      "learning_rate": 9.415447141148129e-07,
      "loss": 0.7835,
      "step": 6780
    },
    {
      "epoch": 0.10563023967244506,
      "grad_norm": 7.534468650817871,
      "learning_rate": 9.414627751102078e-07,
      "loss": 0.8966,
      "step": 6785
    },
    {
      "epoch": 0.10570808067441444,
      "grad_norm": 14.327566146850586,
      "learning_rate": 9.413808361056029e-07,
      "loss": 0.789,
      "step": 6790
    },
    {
      "epoch": 0.10578592167638382,
      "grad_norm": 4.58953332901001,
      "learning_rate": 9.41298897100998e-07,
      "loss": 0.8134,
      "step": 6795
    },
    {
      "epoch": 0.1058637626783532,
      "grad_norm": 3.4943652153015137,
      "learning_rate": 9.41216958096393e-07,
      "loss": 0.9615,
      "step": 6800
    },
    {
      "epoch": 0.10594160368032257,
      "grad_norm": 5.815013408660889,
      "learning_rate": 9.41135019091788e-07,
      "loss": 0.7671,
      "step": 6805
    },
    {
      "epoch": 0.10601944468229195,
      "grad_norm": 4.9490580558776855,
      "learning_rate": 9.410530800871831e-07,
      "loss": 0.7993,
      "step": 6810
    },
    {
      "epoch": 0.10609728568426133,
      "grad_norm": 3.030304193496704,
      "learning_rate": 9.409711410825781e-07,
      "loss": 0.7573,
      "step": 6815
    },
    {
      "epoch": 0.1061751266862307,
      "grad_norm": 3.3977646827697754,
      "learning_rate": 9.408892020779731e-07,
      "loss": 0.7966,
      "step": 6820
    },
    {
      "epoch": 0.10625296768820008,
      "grad_norm": 9.117260932922363,
      "learning_rate": 9.408072630733681e-07,
      "loss": 0.8123,
      "step": 6825
    },
    {
      "epoch": 0.10633080869016946,
      "grad_norm": 3.8861453533172607,
      "learning_rate": 9.407253240687631e-07,
      "loss": 0.8245,
      "step": 6830
    },
    {
      "epoch": 0.10640864969213884,
      "grad_norm": 3.4242775440216064,
      "learning_rate": 9.406433850641582e-07,
      "loss": 0.7318,
      "step": 6835
    },
    {
      "epoch": 0.10648649069410822,
      "grad_norm": 4.729854106903076,
      "learning_rate": 9.405614460595532e-07,
      "loss": 0.7959,
      "step": 6840
    },
    {
      "epoch": 0.1065643316960776,
      "grad_norm": 3.1164026260375977,
      "learning_rate": 9.404795070549483e-07,
      "loss": 0.8327,
      "step": 6845
    },
    {
      "epoch": 0.10664217269804697,
      "grad_norm": 4.031877040863037,
      "learning_rate": 9.403975680503433e-07,
      "loss": 0.807,
      "step": 6850
    },
    {
      "epoch": 0.10672001370001635,
      "grad_norm": 3.205714702606201,
      "learning_rate": 9.403156290457384e-07,
      "loss": 0.7962,
      "step": 6855
    },
    {
      "epoch": 0.10679785470198572,
      "grad_norm": 3.2358205318450928,
      "learning_rate": 9.402336900411333e-07,
      "loss": 0.8694,
      "step": 6860
    },
    {
      "epoch": 0.1068756957039551,
      "grad_norm": 2.7498748302459717,
      "learning_rate": 9.401517510365283e-07,
      "loss": 0.7554,
      "step": 6865
    },
    {
      "epoch": 0.10695353670592447,
      "grad_norm": 7.2536420822143555,
      "learning_rate": 9.400698120319234e-07,
      "loss": 0.9135,
      "step": 6870
    },
    {
      "epoch": 0.10703137770789385,
      "grad_norm": 5.090606689453125,
      "learning_rate": 9.399878730273185e-07,
      "loss": 0.8358,
      "step": 6875
    },
    {
      "epoch": 0.10710921870986323,
      "grad_norm": 3.6972696781158447,
      "learning_rate": 9.399059340227134e-07,
      "loss": 0.9106,
      "step": 6880
    },
    {
      "epoch": 0.10718705971183261,
      "grad_norm": 3.833972692489624,
      "learning_rate": 9.398239950181085e-07,
      "loss": 0.9021,
      "step": 6885
    },
    {
      "epoch": 0.10726490071380199,
      "grad_norm": 6.692166805267334,
      "learning_rate": 9.397420560135036e-07,
      "loss": 0.9014,
      "step": 6890
    },
    {
      "epoch": 0.10734274171577136,
      "grad_norm": 3.5323872566223145,
      "learning_rate": 9.396601170088985e-07,
      "loss": 0.8122,
      "step": 6895
    },
    {
      "epoch": 0.10742058271774074,
      "grad_norm": 5.148552894592285,
      "learning_rate": 9.395781780042935e-07,
      "loss": 0.7712,
      "step": 6900
    },
    {
      "epoch": 0.10749842371971012,
      "grad_norm": 4.791245460510254,
      "learning_rate": 9.394962389996886e-07,
      "loss": 0.8229,
      "step": 6905
    },
    {
      "epoch": 0.1075762647216795,
      "grad_norm": 7.922582149505615,
      "learning_rate": 9.394142999950836e-07,
      "loss": 0.8641,
      "step": 6910
    },
    {
      "epoch": 0.10765410572364888,
      "grad_norm": 4.787046432495117,
      "learning_rate": 9.393323609904786e-07,
      "loss": 0.8567,
      "step": 6915
    },
    {
      "epoch": 0.10773194672561826,
      "grad_norm": 7.581035137176514,
      "learning_rate": 9.392504219858737e-07,
      "loss": 0.9418,
      "step": 6920
    },
    {
      "epoch": 0.10780978772758763,
      "grad_norm": 3.7408881187438965,
      "learning_rate": 9.391684829812687e-07,
      "loss": 0.7569,
      "step": 6925
    },
    {
      "epoch": 0.10788762872955701,
      "grad_norm": 4.957324981689453,
      "learning_rate": 9.390865439766637e-07,
      "loss": 0.9865,
      "step": 6930
    },
    {
      "epoch": 0.10796546973152639,
      "grad_norm": 4.249368190765381,
      "learning_rate": 9.390046049720588e-07,
      "loss": 0.7513,
      "step": 6935
    },
    {
      "epoch": 0.10804331073349577,
      "grad_norm": 4.029480934143066,
      "learning_rate": 9.389226659674537e-07,
      "loss": 0.8097,
      "step": 6940
    },
    {
      "epoch": 0.10812115173546515,
      "grad_norm": 3.9717726707458496,
      "learning_rate": 9.388407269628488e-07,
      "loss": 0.8199,
      "step": 6945
    },
    {
      "epoch": 0.10819899273743451,
      "grad_norm": 4.825889587402344,
      "learning_rate": 9.387587879582438e-07,
      "loss": 0.8086,
      "step": 6950
    },
    {
      "epoch": 0.10827683373940389,
      "grad_norm": 6.288622856140137,
      "learning_rate": 9.386768489536388e-07,
      "loss": 0.764,
      "step": 6955
    },
    {
      "epoch": 0.10835467474137327,
      "grad_norm": 4.316305637359619,
      "learning_rate": 9.385949099490339e-07,
      "loss": 0.8855,
      "step": 6960
    },
    {
      "epoch": 0.10843251574334264,
      "grad_norm": 2.9733645915985107,
      "learning_rate": 9.38512970944429e-07,
      "loss": 0.8997,
      "step": 6965
    },
    {
      "epoch": 0.10851035674531202,
      "grad_norm": 7.586787700653076,
      "learning_rate": 9.38431031939824e-07,
      "loss": 0.7612,
      "step": 6970
    },
    {
      "epoch": 0.1085881977472814,
      "grad_norm": 6.496944904327393,
      "learning_rate": 9.38349092935219e-07,
      "loss": 0.8064,
      "step": 6975
    },
    {
      "epoch": 0.10866603874925078,
      "grad_norm": 10.352307319641113,
      "learning_rate": 9.382671539306141e-07,
      "loss": 0.8054,
      "step": 6980
    },
    {
      "epoch": 0.10874387975122016,
      "grad_norm": 3.3039493560791016,
      "learning_rate": 9.38185214926009e-07,
      "loss": 0.7777,
      "step": 6985
    },
    {
      "epoch": 0.10882172075318954,
      "grad_norm": 2.823133945465088,
      "learning_rate": 9.38103275921404e-07,
      "loss": 0.8755,
      "step": 6990
    },
    {
      "epoch": 0.10889956175515891,
      "grad_norm": 5.029725074768066,
      "learning_rate": 9.380213369167991e-07,
      "loss": 0.7765,
      "step": 6995
    },
    {
      "epoch": 0.10897740275712829,
      "grad_norm": 5.5392889976501465,
      "learning_rate": 9.379393979121942e-07,
      "loss": 0.8166,
      "step": 7000
    },
    {
      "epoch": 0.10905524375909767,
      "grad_norm": 5.657524585723877,
      "learning_rate": 9.378574589075891e-07,
      "loss": 0.7781,
      "step": 7005
    },
    {
      "epoch": 0.10913308476106705,
      "grad_norm": 5.030917167663574,
      "learning_rate": 9.377755199029842e-07,
      "loss": 0.8928,
      "step": 7010
    },
    {
      "epoch": 0.10921092576303643,
      "grad_norm": 4.488454341888428,
      "learning_rate": 9.376935808983793e-07,
      "loss": 0.835,
      "step": 7015
    },
    {
      "epoch": 0.1092887667650058,
      "grad_norm": 3.592827081680298,
      "learning_rate": 9.376116418937743e-07,
      "loss": 0.7534,
      "step": 7020
    },
    {
      "epoch": 0.10936660776697518,
      "grad_norm": 6.4170331954956055,
      "learning_rate": 9.375297028891692e-07,
      "loss": 0.7978,
      "step": 7025
    },
    {
      "epoch": 0.10944444876894456,
      "grad_norm": 3.166126251220703,
      "learning_rate": 9.374477638845643e-07,
      "loss": 0.8483,
      "step": 7030
    },
    {
      "epoch": 0.10952228977091392,
      "grad_norm": 2.965501070022583,
      "learning_rate": 9.373658248799593e-07,
      "loss": 0.7824,
      "step": 7035
    },
    {
      "epoch": 0.1096001307728833,
      "grad_norm": 4.2378058433532715,
      "learning_rate": 9.372838858753543e-07,
      "loss": 0.9217,
      "step": 7040
    },
    {
      "epoch": 0.10967797177485268,
      "grad_norm": 5.209421634674072,
      "learning_rate": 9.372019468707494e-07,
      "loss": 0.835,
      "step": 7045
    },
    {
      "epoch": 0.10975581277682206,
      "grad_norm": 4.27461576461792,
      "learning_rate": 9.371200078661444e-07,
      "loss": 0.9258,
      "step": 7050
    },
    {
      "epoch": 0.10983365377879144,
      "grad_norm": 2.5676474571228027,
      "learning_rate": 9.370380688615395e-07,
      "loss": 0.8691,
      "step": 7055
    },
    {
      "epoch": 0.10991149478076082,
      "grad_norm": 3.422879934310913,
      "learning_rate": 9.369561298569345e-07,
      "loss": 0.7527,
      "step": 7060
    },
    {
      "epoch": 0.1099893357827302,
      "grad_norm": 4.083531379699707,
      "learning_rate": 9.368741908523294e-07,
      "loss": 0.7531,
      "step": 7065
    },
    {
      "epoch": 0.11006717678469957,
      "grad_norm": 4.684252738952637,
      "learning_rate": 9.367922518477245e-07,
      "loss": 0.7772,
      "step": 7070
    },
    {
      "epoch": 0.11014501778666895,
      "grad_norm": 3.0496606826782227,
      "learning_rate": 9.367103128431195e-07,
      "loss": 0.7993,
      "step": 7075
    },
    {
      "epoch": 0.11022285878863833,
      "grad_norm": 3.641996145248413,
      "learning_rate": 9.366283738385145e-07,
      "loss": 0.7938,
      "step": 7080
    },
    {
      "epoch": 0.1103006997906077,
      "grad_norm": 3.0637736320495605,
      "learning_rate": 9.365464348339096e-07,
      "loss": 0.8944,
      "step": 7085
    },
    {
      "epoch": 0.11037854079257708,
      "grad_norm": 4.9412455558776855,
      "learning_rate": 9.364644958293047e-07,
      "loss": 0.7653,
      "step": 7090
    },
    {
      "epoch": 0.11045638179454646,
      "grad_norm": 4.0071516036987305,
      "learning_rate": 9.363825568246997e-07,
      "loss": 0.8582,
      "step": 7095
    },
    {
      "epoch": 0.11053422279651584,
      "grad_norm": 3.297551155090332,
      "learning_rate": 9.363006178200947e-07,
      "loss": 0.6726,
      "step": 7100
    },
    {
      "epoch": 0.11061206379848522,
      "grad_norm": 6.013480186462402,
      "learning_rate": 9.362186788154897e-07,
      "loss": 0.777,
      "step": 7105
    },
    {
      "epoch": 0.1106899048004546,
      "grad_norm": 4.557566165924072,
      "learning_rate": 9.361367398108848e-07,
      "loss": 0.9073,
      "step": 7110
    },
    {
      "epoch": 0.11076774580242398,
      "grad_norm": 3.922395706176758,
      "learning_rate": 9.360548008062797e-07,
      "loss": 0.7608,
      "step": 7115
    },
    {
      "epoch": 0.11084558680439334,
      "grad_norm": 4.4782867431640625,
      "learning_rate": 9.359728618016748e-07,
      "loss": 0.8138,
      "step": 7120
    },
    {
      "epoch": 0.11092342780636272,
      "grad_norm": 3.320688486099243,
      "learning_rate": 9.358909227970699e-07,
      "loss": 0.7954,
      "step": 7125
    },
    {
      "epoch": 0.1110012688083321,
      "grad_norm": 14.582179069519043,
      "learning_rate": 9.358089837924648e-07,
      "loss": 0.8295,
      "step": 7130
    },
    {
      "epoch": 0.11107910981030147,
      "grad_norm": 5.56791877746582,
      "learning_rate": 9.357270447878599e-07,
      "loss": 0.7778,
      "step": 7135
    },
    {
      "epoch": 0.11115695081227085,
      "grad_norm": 4.387538909912109,
      "learning_rate": 9.35645105783255e-07,
      "loss": 0.7354,
      "step": 7140
    },
    {
      "epoch": 0.11123479181424023,
      "grad_norm": 3.590179443359375,
      "learning_rate": 9.355631667786499e-07,
      "loss": 0.7492,
      "step": 7145
    },
    {
      "epoch": 0.11131263281620961,
      "grad_norm": 2.9361941814422607,
      "learning_rate": 9.354812277740449e-07,
      "loss": 0.9192,
      "step": 7150
    },
    {
      "epoch": 0.11139047381817899,
      "grad_norm": 4.104539394378662,
      "learning_rate": 9.3539928876944e-07,
      "loss": 0.7818,
      "step": 7155
    },
    {
      "epoch": 0.11146831482014836,
      "grad_norm": 3.3516862392425537,
      "learning_rate": 9.35317349764835e-07,
      "loss": 0.7937,
      "step": 7160
    },
    {
      "epoch": 0.11154615582211774,
      "grad_norm": 3.5534565448760986,
      "learning_rate": 9.3523541076023e-07,
      "loss": 0.7168,
      "step": 7165
    },
    {
      "epoch": 0.11162399682408712,
      "grad_norm": 3.8620402812957764,
      "learning_rate": 9.351534717556251e-07,
      "loss": 0.7666,
      "step": 7170
    },
    {
      "epoch": 0.1117018378260565,
      "grad_norm": 5.330255031585693,
      "learning_rate": 9.350715327510201e-07,
      "loss": 0.7302,
      "step": 7175
    },
    {
      "epoch": 0.11177967882802588,
      "grad_norm": 3.8225488662719727,
      "learning_rate": 9.349895937464152e-07,
      "loss": 0.7087,
      "step": 7180
    },
    {
      "epoch": 0.11185751982999526,
      "grad_norm": 4.536187648773193,
      "learning_rate": 9.349076547418101e-07,
      "loss": 0.6757,
      "step": 7185
    },
    {
      "epoch": 0.11193536083196463,
      "grad_norm": 3.3316333293914795,
      "learning_rate": 9.348257157372051e-07,
      "loss": 0.8506,
      "step": 7190
    },
    {
      "epoch": 0.11201320183393401,
      "grad_norm": 4.451030731201172,
      "learning_rate": 9.347437767326002e-07,
      "loss": 0.8501,
      "step": 7195
    },
    {
      "epoch": 0.11209104283590339,
      "grad_norm": 6.453036308288574,
      "learning_rate": 9.346618377279953e-07,
      "loss": 0.7665,
      "step": 7200
    },
    {
      "epoch": 0.11216888383787275,
      "grad_norm": 3.7804341316223145,
      "learning_rate": 9.345798987233902e-07,
      "loss": 0.8482,
      "step": 7205
    },
    {
      "epoch": 0.11224672483984213,
      "grad_norm": 4.703028678894043,
      "learning_rate": 9.344979597187853e-07,
      "loss": 0.8001,
      "step": 7210
    },
    {
      "epoch": 0.11232456584181151,
      "grad_norm": 3.5996668338775635,
      "learning_rate": 9.344160207141804e-07,
      "loss": 0.8508,
      "step": 7215
    },
    {
      "epoch": 0.11240240684378089,
      "grad_norm": 3.470485210418701,
      "learning_rate": 9.343340817095754e-07,
      "loss": 0.8929,
      "step": 7220
    },
    {
      "epoch": 0.11248024784575027,
      "grad_norm": 9.36178970336914,
      "learning_rate": 9.342521427049703e-07,
      "loss": 0.7707,
      "step": 7225
    },
    {
      "epoch": 0.11255808884771965,
      "grad_norm": 7.091135025024414,
      "learning_rate": 9.341702037003654e-07,
      "loss": 0.9335,
      "step": 7230
    },
    {
      "epoch": 0.11263592984968902,
      "grad_norm": 4.552675247192383,
      "learning_rate": 9.340882646957605e-07,
      "loss": 0.9234,
      "step": 7235
    },
    {
      "epoch": 0.1127137708516584,
      "grad_norm": 2.9877877235412598,
      "learning_rate": 9.340063256911554e-07,
      "loss": 0.8213,
      "step": 7240
    },
    {
      "epoch": 0.11279161185362778,
      "grad_norm": 3.49109148979187,
      "learning_rate": 9.339243866865505e-07,
      "loss": 0.7713,
      "step": 7245
    },
    {
      "epoch": 0.11286945285559716,
      "grad_norm": 3.662997245788574,
      "learning_rate": 9.338424476819456e-07,
      "loss": 0.8424,
      "step": 7250
    },
    {
      "epoch": 0.11294729385756654,
      "grad_norm": 3.4681499004364014,
      "learning_rate": 9.337605086773405e-07,
      "loss": 0.9026,
      "step": 7255
    },
    {
      "epoch": 0.11302513485953591,
      "grad_norm": 3.360700845718384,
      "learning_rate": 9.336785696727356e-07,
      "loss": 0.9131,
      "step": 7260
    },
    {
      "epoch": 0.11310297586150529,
      "grad_norm": 3.470808982849121,
      "learning_rate": 9.335966306681306e-07,
      "loss": 0.8477,
      "step": 7265
    },
    {
      "epoch": 0.11318081686347467,
      "grad_norm": 4.002136707305908,
      "learning_rate": 9.335146916635256e-07,
      "loss": 0.8753,
      "step": 7270
    },
    {
      "epoch": 0.11325865786544405,
      "grad_norm": 3.301177978515625,
      "learning_rate": 9.334327526589206e-07,
      "loss": 0.8266,
      "step": 7275
    },
    {
      "epoch": 0.11333649886741343,
      "grad_norm": 3.6634960174560547,
      "learning_rate": 9.333508136543157e-07,
      "loss": 0.6648,
      "step": 7280
    },
    {
      "epoch": 0.1134143398693828,
      "grad_norm": 4.124035835266113,
      "learning_rate": 9.332688746497107e-07,
      "loss": 0.8151,
      "step": 7285
    },
    {
      "epoch": 0.11349218087135218,
      "grad_norm": 5.273459434509277,
      "learning_rate": 9.331869356451058e-07,
      "loss": 0.7747,
      "step": 7290
    },
    {
      "epoch": 0.11357002187332155,
      "grad_norm": 3.6614978313446045,
      "learning_rate": 9.331049966405008e-07,
      "loss": 0.8157,
      "step": 7295
    },
    {
      "epoch": 0.11364786287529093,
      "grad_norm": 5.9759368896484375,
      "learning_rate": 9.330230576358958e-07,
      "loss": 0.7475,
      "step": 7300
    },
    {
      "epoch": 0.1137257038772603,
      "grad_norm": 3.271934747695923,
      "learning_rate": 9.329411186312909e-07,
      "loss": 0.8176,
      "step": 7305
    },
    {
      "epoch": 0.11380354487922968,
      "grad_norm": 6.224942207336426,
      "learning_rate": 9.328591796266858e-07,
      "loss": 0.7874,
      "step": 7310
    },
    {
      "epoch": 0.11388138588119906,
      "grad_norm": 4.060842990875244,
      "learning_rate": 9.327772406220808e-07,
      "loss": 0.7104,
      "step": 7315
    },
    {
      "epoch": 0.11395922688316844,
      "grad_norm": 3.273303985595703,
      "learning_rate": 9.326953016174759e-07,
      "loss": 0.9288,
      "step": 7320
    },
    {
      "epoch": 0.11403706788513782,
      "grad_norm": 5.721134662628174,
      "learning_rate": 9.32613362612871e-07,
      "loss": 0.9447,
      "step": 7325
    },
    {
      "epoch": 0.1141149088871072,
      "grad_norm": 3.942401885986328,
      "learning_rate": 9.325314236082659e-07,
      "loss": 0.6934,
      "step": 7330
    },
    {
      "epoch": 0.11419274988907657,
      "grad_norm": 8.272555351257324,
      "learning_rate": 9.32449484603661e-07,
      "loss": 0.7917,
      "step": 7335
    },
    {
      "epoch": 0.11427059089104595,
      "grad_norm": 9.704336166381836,
      "learning_rate": 9.323675455990561e-07,
      "loss": 0.816,
      "step": 7340
    },
    {
      "epoch": 0.11434843189301533,
      "grad_norm": 3.2745420932769775,
      "learning_rate": 9.322856065944512e-07,
      "loss": 0.8432,
      "step": 7345
    },
    {
      "epoch": 0.1144262728949847,
      "grad_norm": 4.300100803375244,
      "learning_rate": 9.32203667589846e-07,
      "loss": 0.8954,
      "step": 7350
    },
    {
      "epoch": 0.11450411389695409,
      "grad_norm": 3.520085334777832,
      "learning_rate": 9.321217285852411e-07,
      "loss": 0.6849,
      "step": 7355
    },
    {
      "epoch": 0.11458195489892346,
      "grad_norm": 3.1472392082214355,
      "learning_rate": 9.320397895806362e-07,
      "loss": 0.9081,
      "step": 7360
    },
    {
      "epoch": 0.11465979590089284,
      "grad_norm": 5.217727184295654,
      "learning_rate": 9.319578505760311e-07,
      "loss": 0.9009,
      "step": 7365
    },
    {
      "epoch": 0.11473763690286222,
      "grad_norm": 2.461811065673828,
      "learning_rate": 9.318759115714262e-07,
      "loss": 0.8018,
      "step": 7370
    },
    {
      "epoch": 0.1148154779048316,
      "grad_norm": 4.464178562164307,
      "learning_rate": 9.317939725668213e-07,
      "loss": 0.7952,
      "step": 7375
    },
    {
      "epoch": 0.11489331890680096,
      "grad_norm": 8.113191604614258,
      "learning_rate": 9.317120335622163e-07,
      "loss": 0.8338,
      "step": 7380
    },
    {
      "epoch": 0.11497115990877034,
      "grad_norm": 3.7389204502105713,
      "learning_rate": 9.316300945576113e-07,
      "loss": 0.7343,
      "step": 7385
    },
    {
      "epoch": 0.11504900091073972,
      "grad_norm": 3.955479145050049,
      "learning_rate": 9.315481555530063e-07,
      "loss": 0.8868,
      "step": 7390
    },
    {
      "epoch": 0.1151268419127091,
      "grad_norm": 3.5472121238708496,
      "learning_rate": 9.314662165484013e-07,
      "loss": 0.7299,
      "step": 7395
    },
    {
      "epoch": 0.11520468291467847,
      "grad_norm": 3.3682563304901123,
      "learning_rate": 9.313842775437963e-07,
      "loss": 0.8079,
      "step": 7400
    },
    {
      "epoch": 0.11528252391664785,
      "grad_norm": 4.840713977813721,
      "learning_rate": 9.313023385391914e-07,
      "loss": 0.8041,
      "step": 7405
    },
    {
      "epoch": 0.11536036491861723,
      "grad_norm": 6.368528366088867,
      "learning_rate": 9.312203995345864e-07,
      "loss": 0.8727,
      "step": 7410
    },
    {
      "epoch": 0.11543820592058661,
      "grad_norm": 8.412065505981445,
      "learning_rate": 9.311384605299815e-07,
      "loss": 0.7993,
      "step": 7415
    },
    {
      "epoch": 0.11551604692255599,
      "grad_norm": 4.714696884155273,
      "learning_rate": 9.310565215253765e-07,
      "loss": 0.7354,
      "step": 7420
    },
    {
      "epoch": 0.11559388792452537,
      "grad_norm": 5.104187488555908,
      "learning_rate": 9.309745825207715e-07,
      "loss": 0.8301,
      "step": 7425
    },
    {
      "epoch": 0.11567172892649474,
      "grad_norm": 2.960247039794922,
      "learning_rate": 9.308926435161665e-07,
      "loss": 0.7949,
      "step": 7430
    },
    {
      "epoch": 0.11574956992846412,
      "grad_norm": 5.484702110290527,
      "learning_rate": 9.308107045115615e-07,
      "loss": 0.7747,
      "step": 7435
    },
    {
      "epoch": 0.1158274109304335,
      "grad_norm": 3.192422866821289,
      "learning_rate": 9.307287655069565e-07,
      "loss": 0.8138,
      "step": 7440
    },
    {
      "epoch": 0.11590525193240288,
      "grad_norm": 4.892508029937744,
      "learning_rate": 9.306468265023516e-07,
      "loss": 0.8059,
      "step": 7445
    },
    {
      "epoch": 0.11598309293437226,
      "grad_norm": 3.488111972808838,
      "learning_rate": 9.305648874977467e-07,
      "loss": 0.7975,
      "step": 7450
    },
    {
      "epoch": 0.11606093393634163,
      "grad_norm": 4.7030029296875,
      "learning_rate": 9.304829484931416e-07,
      "loss": 0.752,
      "step": 7455
    },
    {
      "epoch": 0.11613877493831101,
      "grad_norm": 5.344095706939697,
      "learning_rate": 9.304010094885367e-07,
      "loss": 0.774,
      "step": 7460
    },
    {
      "epoch": 0.11621661594028038,
      "grad_norm": 2.944584846496582,
      "learning_rate": 9.303190704839318e-07,
      "loss": 0.7948,
      "step": 7465
    },
    {
      "epoch": 0.11629445694224975,
      "grad_norm": 2.8411998748779297,
      "learning_rate": 9.302371314793266e-07,
      "loss": 0.7698,
      "step": 7470
    },
    {
      "epoch": 0.11637229794421913,
      "grad_norm": 6.780023097991943,
      "learning_rate": 9.301551924747217e-07,
      "loss": 0.8843,
      "step": 7475
    },
    {
      "epoch": 0.11645013894618851,
      "grad_norm": 4.495726108551025,
      "learning_rate": 9.300732534701168e-07,
      "loss": 0.9019,
      "step": 7480
    },
    {
      "epoch": 0.11652797994815789,
      "grad_norm": 4.61745023727417,
      "learning_rate": 9.299913144655119e-07,
      "loss": 0.8717,
      "step": 7485
    },
    {
      "epoch": 0.11660582095012727,
      "grad_norm": 3.0337278842926025,
      "learning_rate": 9.299093754609068e-07,
      "loss": 0.791,
      "step": 7490
    },
    {
      "epoch": 0.11668366195209665,
      "grad_norm": 6.76452112197876,
      "learning_rate": 9.298274364563019e-07,
      "loss": 0.7373,
      "step": 7495
    },
    {
      "epoch": 0.11676150295406602,
      "grad_norm": 2.7412028312683105,
      "learning_rate": 9.29745497451697e-07,
      "loss": 0.7638,
      "step": 7500
    },
    {
      "epoch": 0.1168393439560354,
      "grad_norm": 4.742910861968994,
      "learning_rate": 9.29663558447092e-07,
      "loss": 0.8951,
      "step": 7505
    },
    {
      "epoch": 0.11691718495800478,
      "grad_norm": 2.368957042694092,
      "learning_rate": 9.295816194424869e-07,
      "loss": 0.7237,
      "step": 7510
    },
    {
      "epoch": 0.11699502595997416,
      "grad_norm": 4.270255088806152,
      "learning_rate": 9.29499680437882e-07,
      "loss": 0.789,
      "step": 7515
    },
    {
      "epoch": 0.11707286696194354,
      "grad_norm": 4.186617851257324,
      "learning_rate": 9.29417741433277e-07,
      "loss": 0.8655,
      "step": 7520
    },
    {
      "epoch": 0.11715070796391291,
      "grad_norm": 2.4891741275787354,
      "learning_rate": 9.29335802428672e-07,
      "loss": 0.8008,
      "step": 7525
    },
    {
      "epoch": 0.11722854896588229,
      "grad_norm": 3.5113582611083984,
      "learning_rate": 9.292538634240671e-07,
      "loss": 0.8465,
      "step": 7530
    },
    {
      "epoch": 0.11730638996785167,
      "grad_norm": 4.365012168884277,
      "learning_rate": 9.291719244194621e-07,
      "loss": 0.8378,
      "step": 7535
    },
    {
      "epoch": 0.11738423096982105,
      "grad_norm": 4.17802619934082,
      "learning_rate": 9.290899854148572e-07,
      "loss": 0.7605,
      "step": 7540
    },
    {
      "epoch": 0.11746207197179043,
      "grad_norm": 3.046833038330078,
      "learning_rate": 9.290080464102522e-07,
      "loss": 0.6796,
      "step": 7545
    },
    {
      "epoch": 0.11753991297375979,
      "grad_norm": 2.9827167987823486,
      "learning_rate": 9.289261074056471e-07,
      "loss": 0.7666,
      "step": 7550
    },
    {
      "epoch": 0.11761775397572917,
      "grad_norm": 2.8555498123168945,
      "learning_rate": 9.288441684010422e-07,
      "loss": 0.8337,
      "step": 7555
    },
    {
      "epoch": 0.11769559497769855,
      "grad_norm": 4.013402462005615,
      "learning_rate": 9.287622293964373e-07,
      "loss": 0.9353,
      "step": 7560
    },
    {
      "epoch": 0.11777343597966793,
      "grad_norm": 6.579110622406006,
      "learning_rate": 9.286802903918322e-07,
      "loss": 0.7464,
      "step": 7565
    },
    {
      "epoch": 0.1178512769816373,
      "grad_norm": 6.643786907196045,
      "learning_rate": 9.285983513872273e-07,
      "loss": 0.8852,
      "step": 7570
    },
    {
      "epoch": 0.11792911798360668,
      "grad_norm": 3.837496042251587,
      "learning_rate": 9.285164123826224e-07,
      "loss": 0.7909,
      "step": 7575
    },
    {
      "epoch": 0.11800695898557606,
      "grad_norm": 3.4954562187194824,
      "learning_rate": 9.284344733780173e-07,
      "loss": 0.8559,
      "step": 7580
    },
    {
      "epoch": 0.11808479998754544,
      "grad_norm": 3.962890148162842,
      "learning_rate": 9.283525343734124e-07,
      "loss": 0.833,
      "step": 7585
    },
    {
      "epoch": 0.11816264098951482,
      "grad_norm": 7.724937438964844,
      "learning_rate": 9.282705953688074e-07,
      "loss": 0.9339,
      "step": 7590
    },
    {
      "epoch": 0.1182404819914842,
      "grad_norm": 5.779001235961914,
      "learning_rate": 9.281886563642024e-07,
      "loss": 0.6968,
      "step": 7595
    },
    {
      "epoch": 0.11831832299345357,
      "grad_norm": 6.85791540145874,
      "learning_rate": 9.281067173595974e-07,
      "loss": 0.8279,
      "step": 7600
    },
    {
      "epoch": 0.11839616399542295,
      "grad_norm": 3.720306158065796,
      "learning_rate": 9.280247783549925e-07,
      "loss": 0.7751,
      "step": 7605
    },
    {
      "epoch": 0.11847400499739233,
      "grad_norm": 4.524914741516113,
      "learning_rate": 9.279428393503876e-07,
      "loss": 0.7227,
      "step": 7610
    },
    {
      "epoch": 0.11855184599936171,
      "grad_norm": 3.6757309436798096,
      "learning_rate": 9.278609003457825e-07,
      "loss": 0.8816,
      "step": 7615
    },
    {
      "epoch": 0.11862968700133109,
      "grad_norm": 3.6419453620910645,
      "learning_rate": 9.277789613411776e-07,
      "loss": 0.7741,
      "step": 7620
    },
    {
      "epoch": 0.11870752800330046,
      "grad_norm": 3.0829155445098877,
      "learning_rate": 9.276970223365727e-07,
      "loss": 0.7712,
      "step": 7625
    },
    {
      "epoch": 0.11878536900526984,
      "grad_norm": 6.524753570556641,
      "learning_rate": 9.276150833319677e-07,
      "loss": 0.8025,
      "step": 7630
    },
    {
      "epoch": 0.11886321000723922,
      "grad_norm": 2.959907054901123,
      "learning_rate": 9.275331443273626e-07,
      "loss": 0.932,
      "step": 7635
    },
    {
      "epoch": 0.11894105100920858,
      "grad_norm": 3.6245338916778564,
      "learning_rate": 9.274512053227577e-07,
      "loss": 0.6997,
      "step": 7640
    },
    {
      "epoch": 0.11901889201117796,
      "grad_norm": 4.328200340270996,
      "learning_rate": 9.273692663181527e-07,
      "loss": 0.8786,
      "step": 7645
    },
    {
      "epoch": 0.11909673301314734,
      "grad_norm": 4.048555374145508,
      "learning_rate": 9.272873273135478e-07,
      "loss": 0.6936,
      "step": 7650
    },
    {
      "epoch": 0.11917457401511672,
      "grad_norm": 3.2797539234161377,
      "learning_rate": 9.272053883089428e-07,
      "loss": 0.7915,
      "step": 7655
    },
    {
      "epoch": 0.1192524150170861,
      "grad_norm": 9.63089656829834,
      "learning_rate": 9.271234493043378e-07,
      "loss": 0.8417,
      "step": 7660
    },
    {
      "epoch": 0.11933025601905548,
      "grad_norm": 4.521198272705078,
      "learning_rate": 9.270415102997329e-07,
      "loss": 0.8831,
      "step": 7665
    },
    {
      "epoch": 0.11940809702102485,
      "grad_norm": 16.66404151916504,
      "learning_rate": 9.269595712951279e-07,
      "loss": 0.8582,
      "step": 7670
    },
    {
      "epoch": 0.11948593802299423,
      "grad_norm": 8.25177001953125,
      "learning_rate": 9.268776322905228e-07,
      "loss": 0.818,
      "step": 7675
    },
    {
      "epoch": 0.11956377902496361,
      "grad_norm": 3.3690059185028076,
      "learning_rate": 9.267956932859179e-07,
      "loss": 0.689,
      "step": 7680
    },
    {
      "epoch": 0.11964162002693299,
      "grad_norm": 4.987194061279297,
      "learning_rate": 9.26713754281313e-07,
      "loss": 0.8771,
      "step": 7685
    },
    {
      "epoch": 0.11971946102890237,
      "grad_norm": 5.207099914550781,
      "learning_rate": 9.266318152767079e-07,
      "loss": 0.898,
      "step": 7690
    },
    {
      "epoch": 0.11979730203087174,
      "grad_norm": 4.396019458770752,
      "learning_rate": 9.26549876272103e-07,
      "loss": 0.7181,
      "step": 7695
    },
    {
      "epoch": 0.11987514303284112,
      "grad_norm": 3.1832330226898193,
      "learning_rate": 9.264679372674981e-07,
      "loss": 0.8264,
      "step": 7700
    },
    {
      "epoch": 0.1199529840348105,
      "grad_norm": 3.0821919441223145,
      "learning_rate": 9.26385998262893e-07,
      "loss": 0.8412,
      "step": 7705
    },
    {
      "epoch": 0.12003082503677988,
      "grad_norm": 4.4039626121521,
      "learning_rate": 9.263040592582881e-07,
      "loss": 0.7891,
      "step": 7710
    },
    {
      "epoch": 0.12010866603874926,
      "grad_norm": 3.0394608974456787,
      "learning_rate": 9.262221202536831e-07,
      "loss": 0.7594,
      "step": 7715
    },
    {
      "epoch": 0.12018650704071863,
      "grad_norm": 3.430525779724121,
      "learning_rate": 9.261401812490781e-07,
      "loss": 0.9487,
      "step": 7720
    },
    {
      "epoch": 0.120264348042688,
      "grad_norm": 3.6181464195251465,
      "learning_rate": 9.260582422444731e-07,
      "loss": 0.881,
      "step": 7725
    },
    {
      "epoch": 0.12034218904465738,
      "grad_norm": 9.917842864990234,
      "learning_rate": 9.259763032398682e-07,
      "loss": 0.7738,
      "step": 7730
    },
    {
      "epoch": 0.12042003004662676,
      "grad_norm": 5.7242255210876465,
      "learning_rate": 9.258943642352633e-07,
      "loss": 0.9353,
      "step": 7735
    },
    {
      "epoch": 0.12049787104859613,
      "grad_norm": 5.354982852935791,
      "learning_rate": 9.258124252306583e-07,
      "loss": 0.7719,
      "step": 7740
    },
    {
      "epoch": 0.12057571205056551,
      "grad_norm": 5.967920780181885,
      "learning_rate": 9.257304862260533e-07,
      "loss": 0.7304,
      "step": 7745
    },
    {
      "epoch": 0.12065355305253489,
      "grad_norm": 3.245041847229004,
      "learning_rate": 9.256485472214484e-07,
      "loss": 0.6752,
      "step": 7750
    },
    {
      "epoch": 0.12073139405450427,
      "grad_norm": 6.362682342529297,
      "learning_rate": 9.255666082168433e-07,
      "loss": 0.8159,
      "step": 7755
    },
    {
      "epoch": 0.12080923505647365,
      "grad_norm": 5.759777069091797,
      "learning_rate": 9.254846692122383e-07,
      "loss": 0.6947,
      "step": 7760
    },
    {
      "epoch": 0.12088707605844302,
      "grad_norm": 3.7885186672210693,
      "learning_rate": 9.254027302076334e-07,
      "loss": 0.8279,
      "step": 7765
    },
    {
      "epoch": 0.1209649170604124,
      "grad_norm": 2.970975399017334,
      "learning_rate": 9.253207912030284e-07,
      "loss": 0.8366,
      "step": 7770
    },
    {
      "epoch": 0.12104275806238178,
      "grad_norm": 3.9208946228027344,
      "learning_rate": 9.252388521984235e-07,
      "loss": 0.8122,
      "step": 7775
    },
    {
      "epoch": 0.12112059906435116,
      "grad_norm": 3.411137580871582,
      "learning_rate": 9.251569131938185e-07,
      "loss": 0.736,
      "step": 7780
    },
    {
      "epoch": 0.12119844006632054,
      "grad_norm": 10.479853630065918,
      "learning_rate": 9.250749741892135e-07,
      "loss": 0.9075,
      "step": 7785
    },
    {
      "epoch": 0.12127628106828992,
      "grad_norm": 4.920512676239014,
      "learning_rate": 9.249930351846086e-07,
      "loss": 0.8346,
      "step": 7790
    },
    {
      "epoch": 0.1213541220702593,
      "grad_norm": 5.350609302520752,
      "learning_rate": 9.249110961800035e-07,
      "loss": 0.8034,
      "step": 7795
    },
    {
      "epoch": 0.12143196307222867,
      "grad_norm": 6.461511611938477,
      "learning_rate": 9.248291571753985e-07,
      "loss": 0.6603,
      "step": 7800
    },
    {
      "epoch": 0.12150980407419805,
      "grad_norm": 4.6280837059021,
      "learning_rate": 9.247472181707936e-07,
      "loss": 0.801,
      "step": 7805
    },
    {
      "epoch": 0.12158764507616741,
      "grad_norm": 8.22148323059082,
      "learning_rate": 9.246652791661887e-07,
      "loss": 0.8335,
      "step": 7810
    },
    {
      "epoch": 0.12166548607813679,
      "grad_norm": 7.433074951171875,
      "learning_rate": 9.245833401615836e-07,
      "loss": 0.7171,
      "step": 7815
    },
    {
      "epoch": 0.12174332708010617,
      "grad_norm": 5.692368507385254,
      "learning_rate": 9.245014011569787e-07,
      "loss": 0.8649,
      "step": 7820
    },
    {
      "epoch": 0.12182116808207555,
      "grad_norm": 3.0104875564575195,
      "learning_rate": 9.244194621523738e-07,
      "loss": 0.6021,
      "step": 7825
    },
    {
      "epoch": 0.12189900908404493,
      "grad_norm": 6.160313606262207,
      "learning_rate": 9.243375231477688e-07,
      "loss": 0.8652,
      "step": 7830
    },
    {
      "epoch": 0.1219768500860143,
      "grad_norm": 4.373056411743164,
      "learning_rate": 9.242555841431637e-07,
      "loss": 0.7689,
      "step": 7835
    },
    {
      "epoch": 0.12205469108798368,
      "grad_norm": 5.760363578796387,
      "learning_rate": 9.241736451385588e-07,
      "loss": 0.8888,
      "step": 7840
    },
    {
      "epoch": 0.12213253208995306,
      "grad_norm": 4.07029914855957,
      "learning_rate": 9.240917061339538e-07,
      "loss": 0.7693,
      "step": 7845
    },
    {
      "epoch": 0.12221037309192244,
      "grad_norm": 3.7920992374420166,
      "learning_rate": 9.240097671293488e-07,
      "loss": 0.8221,
      "step": 7850
    },
    {
      "epoch": 0.12228821409389182,
      "grad_norm": 7.572727203369141,
      "learning_rate": 9.239278281247439e-07,
      "loss": 0.8255,
      "step": 7855
    },
    {
      "epoch": 0.1223660550958612,
      "grad_norm": 5.047962188720703,
      "learning_rate": 9.23845889120139e-07,
      "loss": 0.8189,
      "step": 7860
    },
    {
      "epoch": 0.12244389609783057,
      "grad_norm": 6.445250988006592,
      "learning_rate": 9.23763950115534e-07,
      "loss": 0.9137,
      "step": 7865
    },
    {
      "epoch": 0.12252173709979995,
      "grad_norm": 2.9722390174865723,
      "learning_rate": 9.23682011110929e-07,
      "loss": 0.895,
      "step": 7870
    },
    {
      "epoch": 0.12259957810176933,
      "grad_norm": 3.0519466400146484,
      "learning_rate": 9.23600072106324e-07,
      "loss": 0.8048,
      "step": 7875
    },
    {
      "epoch": 0.12267741910373871,
      "grad_norm": 5.44154167175293,
      "learning_rate": 9.23518133101719e-07,
      "loss": 0.6992,
      "step": 7880
    },
    {
      "epoch": 0.12275526010570809,
      "grad_norm": 11.307611465454102,
      "learning_rate": 9.23436194097114e-07,
      "loss": 0.8595,
      "step": 7885
    },
    {
      "epoch": 0.12283310110767746,
      "grad_norm": 6.504110336303711,
      "learning_rate": 9.233542550925091e-07,
      "loss": 0.7882,
      "step": 7890
    },
    {
      "epoch": 0.12291094210964683,
      "grad_norm": 2.5726773738861084,
      "learning_rate": 9.232723160879041e-07,
      "loss": 0.8535,
      "step": 7895
    },
    {
      "epoch": 0.1229887831116162,
      "grad_norm": 3.1990294456481934,
      "learning_rate": 9.231903770832992e-07,
      "loss": 0.7941,
      "step": 7900
    },
    {
      "epoch": 0.12306662411358558,
      "grad_norm": 3.614875555038452,
      "learning_rate": 9.231084380786942e-07,
      "loss": 0.6803,
      "step": 7905
    },
    {
      "epoch": 0.12314446511555496,
      "grad_norm": 5.973575592041016,
      "learning_rate": 9.230264990740892e-07,
      "loss": 0.7994,
      "step": 7910
    },
    {
      "epoch": 0.12322230611752434,
      "grad_norm": 3.558375358581543,
      "learning_rate": 9.229445600694842e-07,
      "loss": 0.8292,
      "step": 7915
    },
    {
      "epoch": 0.12330014711949372,
      "grad_norm": 4.37818717956543,
      "learning_rate": 9.228626210648793e-07,
      "loss": 0.8512,
      "step": 7920
    },
    {
      "epoch": 0.1233779881214631,
      "grad_norm": 7.052612781524658,
      "learning_rate": 9.227806820602742e-07,
      "loss": 0.7453,
      "step": 7925
    },
    {
      "epoch": 0.12345582912343248,
      "grad_norm": 9.181270599365234,
      "learning_rate": 9.226987430556693e-07,
      "loss": 0.6847,
      "step": 7930
    },
    {
      "epoch": 0.12353367012540185,
      "grad_norm": 4.499727249145508,
      "learning_rate": 9.226168040510644e-07,
      "loss": 0.7718,
      "step": 7935
    },
    {
      "epoch": 0.12361151112737123,
      "grad_norm": 3.2490360736846924,
      "learning_rate": 9.225348650464593e-07,
      "loss": 0.7631,
      "step": 7940
    },
    {
      "epoch": 0.12368935212934061,
      "grad_norm": 3.26198410987854,
      "learning_rate": 9.224529260418544e-07,
      "loss": 0.8257,
      "step": 7945
    },
    {
      "epoch": 0.12376719313130999,
      "grad_norm": 7.192509174346924,
      "learning_rate": 9.223709870372495e-07,
      "loss": 0.809,
      "step": 7950
    },
    {
      "epoch": 0.12384503413327937,
      "grad_norm": 3.1407008171081543,
      "learning_rate": 9.222890480326445e-07,
      "loss": 0.8444,
      "step": 7955
    },
    {
      "epoch": 0.12392287513524874,
      "grad_norm": 4.12625789642334,
      "learning_rate": 9.222071090280394e-07,
      "loss": 0.8275,
      "step": 7960
    },
    {
      "epoch": 0.12400071613721812,
      "grad_norm": 5.525479316711426,
      "learning_rate": 9.221251700234345e-07,
      "loss": 0.8999,
      "step": 7965
    },
    {
      "epoch": 0.1240785571391875,
      "grad_norm": 6.302455902099609,
      "learning_rate": 9.220432310188295e-07,
      "loss": 0.7642,
      "step": 7970
    },
    {
      "epoch": 0.12415639814115688,
      "grad_norm": 4.649979114532471,
      "learning_rate": 9.219612920142245e-07,
      "loss": 0.7002,
      "step": 7975
    },
    {
      "epoch": 0.12423423914312626,
      "grad_norm": 5.463395595550537,
      "learning_rate": 9.218793530096196e-07,
      "loss": 0.8473,
      "step": 7980
    },
    {
      "epoch": 0.12431208014509562,
      "grad_norm": 6.704756736755371,
      "learning_rate": 9.217974140050147e-07,
      "loss": 0.6967,
      "step": 7985
    },
    {
      "epoch": 0.124389921147065,
      "grad_norm": 4.5808539390563965,
      "learning_rate": 9.217154750004097e-07,
      "loss": 0.7898,
      "step": 7990
    },
    {
      "epoch": 0.12446776214903438,
      "grad_norm": 2.9757680892944336,
      "learning_rate": 9.216335359958047e-07,
      "loss": 0.698,
      "step": 7995
    },
    {
      "epoch": 0.12454560315100376,
      "grad_norm": 5.5388617515563965,
      "learning_rate": 9.215515969911997e-07,
      "loss": 0.8798,
      "step": 8000
    },
    {
      "epoch": 0.12462344415297313,
      "grad_norm": 5.424650192260742,
      "learning_rate": 9.214696579865947e-07,
      "loss": 0.814,
      "step": 8005
    },
    {
      "epoch": 0.12470128515494251,
      "grad_norm": 4.0730109214782715,
      "learning_rate": 9.213877189819898e-07,
      "loss": 0.6547,
      "step": 8010
    },
    {
      "epoch": 0.12477912615691189,
      "grad_norm": 5.436715126037598,
      "learning_rate": 9.213057799773848e-07,
      "loss": 0.776,
      "step": 8015
    },
    {
      "epoch": 0.12485696715888127,
      "grad_norm": 3.6354191303253174,
      "learning_rate": 9.212238409727798e-07,
      "loss": 0.7196,
      "step": 8020
    },
    {
      "epoch": 0.12493480816085065,
      "grad_norm": 3.5990068912506104,
      "learning_rate": 9.211419019681749e-07,
      "loss": 0.8079,
      "step": 8025
    },
    {
      "epoch": 0.12501264916282,
      "grad_norm": 4.007763385772705,
      "learning_rate": 9.2105996296357e-07,
      "loss": 0.7711,
      "step": 8030
    },
    {
      "epoch": 0.1250904901647894,
      "grad_norm": 4.223349571228027,
      "learning_rate": 9.209780239589649e-07,
      "loss": 0.8481,
      "step": 8035
    },
    {
      "epoch": 0.12516833116675877,
      "grad_norm": 4.40108060836792,
      "learning_rate": 9.208960849543599e-07,
      "loss": 0.7981,
      "step": 8040
    },
    {
      "epoch": 0.12524617216872816,
      "grad_norm": 2.7487142086029053,
      "learning_rate": 9.20814145949755e-07,
      "loss": 0.6693,
      "step": 8045
    },
    {
      "epoch": 0.12532401317069752,
      "grad_norm": 3.568763017654419,
      "learning_rate": 9.207322069451499e-07,
      "loss": 0.7716,
      "step": 8050
    },
    {
      "epoch": 0.12540185417266692,
      "grad_norm": 3.7271010875701904,
      "learning_rate": 9.20650267940545e-07,
      "loss": 0.7969,
      "step": 8055
    },
    {
      "epoch": 0.12547969517463628,
      "grad_norm": 4.352176189422607,
      "learning_rate": 9.205683289359401e-07,
      "loss": 0.8513,
      "step": 8060
    },
    {
      "epoch": 0.12555753617660567,
      "grad_norm": 3.4279236793518066,
      "learning_rate": 9.20486389931335e-07,
      "loss": 0.8315,
      "step": 8065
    },
    {
      "epoch": 0.12563537717857504,
      "grad_norm": 5.192807197570801,
      "learning_rate": 9.204044509267301e-07,
      "loss": 0.6887,
      "step": 8070
    },
    {
      "epoch": 0.12571321818054443,
      "grad_norm": 7.44572114944458,
      "learning_rate": 9.203225119221252e-07,
      "loss": 0.8179,
      "step": 8075
    },
    {
      "epoch": 0.1257910591825138,
      "grad_norm": 2.792656183242798,
      "learning_rate": 9.202405729175201e-07,
      "loss": 0.8136,
      "step": 8080
    },
    {
      "epoch": 0.12586890018448318,
      "grad_norm": 3.840090036392212,
      "learning_rate": 9.201586339129151e-07,
      "loss": 0.7216,
      "step": 8085
    },
    {
      "epoch": 0.12594674118645255,
      "grad_norm": 6.609809398651123,
      "learning_rate": 9.200766949083102e-07,
      "loss": 0.8109,
      "step": 8090
    },
    {
      "epoch": 0.12602458218842194,
      "grad_norm": 3.8064773082733154,
      "learning_rate": 9.199947559037052e-07,
      "loss": 0.7874,
      "step": 8095
    },
    {
      "epoch": 0.1261024231903913,
      "grad_norm": 4.612837791442871,
      "learning_rate": 9.199128168991003e-07,
      "loss": 0.8989,
      "step": 8100
    },
    {
      "epoch": 0.1261802641923607,
      "grad_norm": 3.837954044342041,
      "learning_rate": 9.198308778944953e-07,
      "loss": 0.7964,
      "step": 8105
    },
    {
      "epoch": 0.12625810519433006,
      "grad_norm": 5.080657482147217,
      "learning_rate": 9.197489388898904e-07,
      "loss": 0.7835,
      "step": 8110
    },
    {
      "epoch": 0.12633594619629943,
      "grad_norm": 5.589513301849365,
      "learning_rate": 9.196669998852854e-07,
      "loss": 0.8027,
      "step": 8115
    },
    {
      "epoch": 0.12641378719826882,
      "grad_norm": 5.392527103424072,
      "learning_rate": 9.195850608806803e-07,
      "loss": 0.7721,
      "step": 8120
    },
    {
      "epoch": 0.12649162820023818,
      "grad_norm": 8.648225784301758,
      "learning_rate": 9.195031218760754e-07,
      "loss": 0.8396,
      "step": 8125
    },
    {
      "epoch": 0.12656946920220757,
      "grad_norm": 3.0926201343536377,
      "learning_rate": 9.194211828714704e-07,
      "loss": 0.8282,
      "step": 8130
    },
    {
      "epoch": 0.12664731020417694,
      "grad_norm": 4.515932083129883,
      "learning_rate": 9.193392438668655e-07,
      "loss": 0.8238,
      "step": 8135
    },
    {
      "epoch": 0.12672515120614633,
      "grad_norm": 6.131096839904785,
      "learning_rate": 9.192573048622605e-07,
      "loss": 0.8638,
      "step": 8140
    },
    {
      "epoch": 0.1268029922081157,
      "grad_norm": 3.634267807006836,
      "learning_rate": 9.191753658576555e-07,
      "loss": 0.716,
      "step": 8145
    },
    {
      "epoch": 0.1268808332100851,
      "grad_norm": 4.676586627960205,
      "learning_rate": 9.190934268530506e-07,
      "loss": 0.8273,
      "step": 8150
    },
    {
      "epoch": 0.12695867421205445,
      "grad_norm": 3.584019422531128,
      "learning_rate": 9.190114878484457e-07,
      "loss": 0.805,
      "step": 8155
    },
    {
      "epoch": 0.12703651521402384,
      "grad_norm": 3.7464358806610107,
      "learning_rate": 9.189295488438405e-07,
      "loss": 0.7724,
      "step": 8160
    },
    {
      "epoch": 0.1271143562159932,
      "grad_norm": 3.974726438522339,
      "learning_rate": 9.188476098392356e-07,
      "loss": 0.807,
      "step": 8165
    },
    {
      "epoch": 0.1271921972179626,
      "grad_norm": 5.144652843475342,
      "learning_rate": 9.187656708346307e-07,
      "loss": 0.8721,
      "step": 8170
    },
    {
      "epoch": 0.12727003821993196,
      "grad_norm": 7.679945945739746,
      "learning_rate": 9.186837318300256e-07,
      "loss": 0.7945,
      "step": 8175
    },
    {
      "epoch": 0.12734787922190136,
      "grad_norm": 5.530436992645264,
      "learning_rate": 9.186017928254207e-07,
      "loss": 0.9354,
      "step": 8180
    },
    {
      "epoch": 0.12742572022387072,
      "grad_norm": 3.981515884399414,
      "learning_rate": 9.185198538208158e-07,
      "loss": 0.6024,
      "step": 8185
    },
    {
      "epoch": 0.1275035612258401,
      "grad_norm": 3.5425384044647217,
      "learning_rate": 9.184379148162108e-07,
      "loss": 0.8061,
      "step": 8190
    },
    {
      "epoch": 0.12758140222780948,
      "grad_norm": 3.2570059299468994,
      "learning_rate": 9.183559758116058e-07,
      "loss": 0.7803,
      "step": 8195
    },
    {
      "epoch": 0.12765924322977884,
      "grad_norm": 3.267265558242798,
      "learning_rate": 9.182740368070008e-07,
      "loss": 0.7927,
      "step": 8200
    },
    {
      "epoch": 0.12773708423174823,
      "grad_norm": 3.4302942752838135,
      "learning_rate": 9.181920978023958e-07,
      "loss": 0.8104,
      "step": 8205
    },
    {
      "epoch": 0.1278149252337176,
      "grad_norm": 4.691220760345459,
      "learning_rate": 9.181101587977908e-07,
      "loss": 0.7826,
      "step": 8210
    },
    {
      "epoch": 0.127892766235687,
      "grad_norm": 4.321291446685791,
      "learning_rate": 9.180282197931859e-07,
      "loss": 0.8503,
      "step": 8215
    },
    {
      "epoch": 0.12797060723765635,
      "grad_norm": 6.356113433837891,
      "learning_rate": 9.179462807885809e-07,
      "loss": 0.8026,
      "step": 8220
    },
    {
      "epoch": 0.12804844823962575,
      "grad_norm": 6.146854400634766,
      "learning_rate": 9.17864341783976e-07,
      "loss": 0.7779,
      "step": 8225
    },
    {
      "epoch": 0.1281262892415951,
      "grad_norm": 4.845507621765137,
      "learning_rate": 9.17782402779371e-07,
      "loss": 0.855,
      "step": 8230
    },
    {
      "epoch": 0.1282041302435645,
      "grad_norm": 4.020603179931641,
      "learning_rate": 9.177004637747661e-07,
      "loss": 0.8117,
      "step": 8235
    },
    {
      "epoch": 0.12828197124553387,
      "grad_norm": 3.1681554317474365,
      "learning_rate": 9.17618524770161e-07,
      "loss": 1.0093,
      "step": 8240
    },
    {
      "epoch": 0.12835981224750326,
      "grad_norm": 12.957433700561523,
      "learning_rate": 9.17536585765556e-07,
      "loss": 0.8047,
      "step": 8245
    },
    {
      "epoch": 0.12843765324947262,
      "grad_norm": 3.9429116249084473,
      "learning_rate": 9.174546467609511e-07,
      "loss": 0.7415,
      "step": 8250
    },
    {
      "epoch": 0.12851549425144201,
      "grad_norm": 5.232780933380127,
      "learning_rate": 9.173727077563461e-07,
      "loss": 0.9999,
      "step": 8255
    },
    {
      "epoch": 0.12859333525341138,
      "grad_norm": 5.215351104736328,
      "learning_rate": 9.172907687517412e-07,
      "loss": 0.7994,
      "step": 8260
    },
    {
      "epoch": 0.12867117625538077,
      "grad_norm": 3.917405366897583,
      "learning_rate": 9.172088297471362e-07,
      "loss": 0.7882,
      "step": 8265
    },
    {
      "epoch": 0.12874901725735013,
      "grad_norm": 5.005404949188232,
      "learning_rate": 9.171268907425312e-07,
      "loss": 0.8112,
      "step": 8270
    },
    {
      "epoch": 0.12882685825931953,
      "grad_norm": 3.008211851119995,
      "learning_rate": 9.170449517379263e-07,
      "loss": 0.8249,
      "step": 8275
    },
    {
      "epoch": 0.1289046992612889,
      "grad_norm": 9.127318382263184,
      "learning_rate": 9.169630127333214e-07,
      "loss": 0.8224,
      "step": 8280
    },
    {
      "epoch": 0.12898254026325826,
      "grad_norm": 4.283410549163818,
      "learning_rate": 9.168810737287162e-07,
      "loss": 0.8319,
      "step": 8285
    },
    {
      "epoch": 0.12906038126522765,
      "grad_norm": 9.957722663879395,
      "learning_rate": 9.167991347241113e-07,
      "loss": 0.7951,
      "step": 8290
    },
    {
      "epoch": 0.129138222267197,
      "grad_norm": 3.6089370250701904,
      "learning_rate": 9.167171957195064e-07,
      "loss": 0.7409,
      "step": 8295
    },
    {
      "epoch": 0.1292160632691664,
      "grad_norm": 6.038301467895508,
      "learning_rate": 9.166352567149013e-07,
      "loss": 0.802,
      "step": 8300
    },
    {
      "epoch": 0.12929390427113577,
      "grad_norm": 5.123950481414795,
      "learning_rate": 9.165533177102964e-07,
      "loss": 0.723,
      "step": 8305
    },
    {
      "epoch": 0.12937174527310516,
      "grad_norm": 3.86824369430542,
      "learning_rate": 9.164713787056915e-07,
      "loss": 0.811,
      "step": 8310
    },
    {
      "epoch": 0.12944958627507452,
      "grad_norm": 4.312297821044922,
      "learning_rate": 9.163894397010865e-07,
      "loss": 0.7,
      "step": 8315
    },
    {
      "epoch": 0.12952742727704392,
      "grad_norm": 2.920485258102417,
      "learning_rate": 9.163075006964815e-07,
      "loss": 0.8012,
      "step": 8320
    },
    {
      "epoch": 0.12960526827901328,
      "grad_norm": 3.8033828735351562,
      "learning_rate": 9.162255616918765e-07,
      "loss": 0.8236,
      "step": 8325
    },
    {
      "epoch": 0.12968310928098267,
      "grad_norm": 7.963630199432373,
      "learning_rate": 9.161436226872715e-07,
      "loss": 0.8041,
      "step": 8330
    },
    {
      "epoch": 0.12976095028295204,
      "grad_norm": 3.964761734008789,
      "learning_rate": 9.160616836826666e-07,
      "loss": 0.8897,
      "step": 8335
    },
    {
      "epoch": 0.12983879128492143,
      "grad_norm": 3.360156297683716,
      "learning_rate": 9.159797446780616e-07,
      "loss": 0.7087,
      "step": 8340
    },
    {
      "epoch": 0.1299166322868908,
      "grad_norm": 4.731776237487793,
      "learning_rate": 9.158978056734567e-07,
      "loss": 0.7926,
      "step": 8345
    },
    {
      "epoch": 0.12999447328886019,
      "grad_norm": 3.076554775238037,
      "learning_rate": 9.158158666688517e-07,
      "loss": 0.7662,
      "step": 8350
    },
    {
      "epoch": 0.13007231429082955,
      "grad_norm": 3.2537529468536377,
      "learning_rate": 9.157339276642467e-07,
      "loss": 0.8568,
      "step": 8355
    },
    {
      "epoch": 0.13015015529279894,
      "grad_norm": 4.160289287567139,
      "learning_rate": 9.156519886596418e-07,
      "loss": 0.7748,
      "step": 8360
    },
    {
      "epoch": 0.1302279962947683,
      "grad_norm": 3.388763666152954,
      "learning_rate": 9.155700496550367e-07,
      "loss": 0.7411,
      "step": 8365
    },
    {
      "epoch": 0.13030583729673767,
      "grad_norm": 3.7193074226379395,
      "learning_rate": 9.154881106504318e-07,
      "loss": 0.7553,
      "step": 8370
    },
    {
      "epoch": 0.13038367829870706,
      "grad_norm": 4.124868392944336,
      "learning_rate": 9.154061716458268e-07,
      "loss": 0.8061,
      "step": 8375
    },
    {
      "epoch": 0.13046151930067643,
      "grad_norm": 3.1243176460266113,
      "learning_rate": 9.153242326412218e-07,
      "loss": 0.7673,
      "step": 8380
    },
    {
      "epoch": 0.13053936030264582,
      "grad_norm": 3.175187587738037,
      "learning_rate": 9.152422936366169e-07,
      "loss": 0.636,
      "step": 8385
    },
    {
      "epoch": 0.13061720130461518,
      "grad_norm": 3.486941337585449,
      "learning_rate": 9.15160354632012e-07,
      "loss": 0.768,
      "step": 8390
    },
    {
      "epoch": 0.13069504230658457,
      "grad_norm": 3.408848762512207,
      "learning_rate": 9.150784156274069e-07,
      "loss": 0.8326,
      "step": 8395
    },
    {
      "epoch": 0.13077288330855394,
      "grad_norm": 5.37129545211792,
      "learning_rate": 9.14996476622802e-07,
      "loss": 0.7712,
      "step": 8400
    },
    {
      "epoch": 0.13085072431052333,
      "grad_norm": 2.642165422439575,
      "learning_rate": 9.14914537618197e-07,
      "loss": 0.8611,
      "step": 8405
    },
    {
      "epoch": 0.1309285653124927,
      "grad_norm": 3.8213489055633545,
      "learning_rate": 9.148325986135919e-07,
      "loss": 0.7938,
      "step": 8410
    },
    {
      "epoch": 0.1310064063144621,
      "grad_norm": 3.893542528152466,
      "learning_rate": 9.14750659608987e-07,
      "loss": 0.7761,
      "step": 8415
    },
    {
      "epoch": 0.13108424731643145,
      "grad_norm": 3.6185567378997803,
      "learning_rate": 9.146687206043821e-07,
      "loss": 0.776,
      "step": 8420
    },
    {
      "epoch": 0.13116208831840084,
      "grad_norm": 7.76255989074707,
      "learning_rate": 9.14586781599777e-07,
      "loss": 0.8723,
      "step": 8425
    },
    {
      "epoch": 0.1312399293203702,
      "grad_norm": 3.2361936569213867,
      "learning_rate": 9.145048425951721e-07,
      "loss": 0.7151,
      "step": 8430
    },
    {
      "epoch": 0.1313177703223396,
      "grad_norm": 7.259923458099365,
      "learning_rate": 9.144229035905672e-07,
      "loss": 0.7714,
      "step": 8435
    },
    {
      "epoch": 0.13139561132430896,
      "grad_norm": 8.949355125427246,
      "learning_rate": 9.143409645859622e-07,
      "loss": 0.8829,
      "step": 8440
    },
    {
      "epoch": 0.13147345232627836,
      "grad_norm": 5.755862712860107,
      "learning_rate": 9.142590255813571e-07,
      "loss": 0.8672,
      "step": 8445
    },
    {
      "epoch": 0.13155129332824772,
      "grad_norm": 4.543202877044678,
      "learning_rate": 9.141770865767522e-07,
      "loss": 0.8165,
      "step": 8450
    },
    {
      "epoch": 0.1316291343302171,
      "grad_norm": 3.6541123390197754,
      "learning_rate": 9.140951475721472e-07,
      "loss": 0.8017,
      "step": 8455
    },
    {
      "epoch": 0.13170697533218648,
      "grad_norm": 3.5702321529388428,
      "learning_rate": 9.140132085675423e-07,
      "loss": 0.8063,
      "step": 8460
    },
    {
      "epoch": 0.13178481633415584,
      "grad_norm": 3.831411361694336,
      "learning_rate": 9.139312695629373e-07,
      "loss": 0.832,
      "step": 8465
    },
    {
      "epoch": 0.13186265733612523,
      "grad_norm": 3.0359880924224854,
      "learning_rate": 9.138493305583324e-07,
      "loss": 0.6776,
      "step": 8470
    },
    {
      "epoch": 0.1319404983380946,
      "grad_norm": 4.264082908630371,
      "learning_rate": 9.137673915537274e-07,
      "loss": 0.6582,
      "step": 8475
    },
    {
      "epoch": 0.132018339340064,
      "grad_norm": 4.2727508544921875,
      "learning_rate": 9.136854525491225e-07,
      "loss": 0.7432,
      "step": 8480
    },
    {
      "epoch": 0.13209618034203335,
      "grad_norm": 7.4130539894104,
      "learning_rate": 9.136035135445174e-07,
      "loss": 0.7692,
      "step": 8485
    },
    {
      "epoch": 0.13217402134400275,
      "grad_norm": 5.631756782531738,
      "learning_rate": 9.135215745399124e-07,
      "loss": 0.7297,
      "step": 8490
    },
    {
      "epoch": 0.1322518623459721,
      "grad_norm": 4.975935459136963,
      "learning_rate": 9.134396355353075e-07,
      "loss": 0.8036,
      "step": 8495
    },
    {
      "epoch": 0.1323297033479415,
      "grad_norm": 3.5903608798980713,
      "learning_rate": 9.133576965307025e-07,
      "loss": 0.8833,
      "step": 8500
    },
    {
      "epoch": 0.13240754434991087,
      "grad_norm": 4.563084602355957,
      "learning_rate": 9.132757575260975e-07,
      "loss": 0.7386,
      "step": 8505
    },
    {
      "epoch": 0.13248538535188026,
      "grad_norm": 5.170467853546143,
      "learning_rate": 9.131938185214926e-07,
      "loss": 0.8154,
      "step": 8510
    },
    {
      "epoch": 0.13256322635384962,
      "grad_norm": 6.0783233642578125,
      "learning_rate": 9.131118795168877e-07,
      "loss": 0.8525,
      "step": 8515
    },
    {
      "epoch": 0.13264106735581901,
      "grad_norm": 3.2167932987213135,
      "learning_rate": 9.130299405122826e-07,
      "loss": 0.7605,
      "step": 8520
    },
    {
      "epoch": 0.13271890835778838,
      "grad_norm": 4.35626745223999,
      "learning_rate": 9.129480015076776e-07,
      "loss": 0.7577,
      "step": 8525
    },
    {
      "epoch": 0.13279674935975777,
      "grad_norm": 6.491021633148193,
      "learning_rate": 9.128660625030727e-07,
      "loss": 0.8463,
      "step": 8530
    },
    {
      "epoch": 0.13287459036172714,
      "grad_norm": 3.174940347671509,
      "learning_rate": 9.127841234984676e-07,
      "loss": 0.7382,
      "step": 8535
    },
    {
      "epoch": 0.13295243136369653,
      "grad_norm": 4.856359004974365,
      "learning_rate": 9.127021844938627e-07,
      "loss": 0.7025,
      "step": 8540
    },
    {
      "epoch": 0.1330302723656659,
      "grad_norm": 3.6859755516052246,
      "learning_rate": 9.126202454892578e-07,
      "loss": 0.642,
      "step": 8545
    },
    {
      "epoch": 0.13310811336763526,
      "grad_norm": 3.6811821460723877,
      "learning_rate": 9.125383064846528e-07,
      "loss": 0.828,
      "step": 8550
    },
    {
      "epoch": 0.13318595436960465,
      "grad_norm": 7.908188819885254,
      "learning_rate": 9.124563674800478e-07,
      "loss": 0.8043,
      "step": 8555
    },
    {
      "epoch": 0.133263795371574,
      "grad_norm": 3.508521795272827,
      "learning_rate": 9.123744284754429e-07,
      "loss": 0.8052,
      "step": 8560
    },
    {
      "epoch": 0.1333416363735434,
      "grad_norm": 4.004594326019287,
      "learning_rate": 9.122924894708378e-07,
      "loss": 0.8126,
      "step": 8565
    },
    {
      "epoch": 0.13341947737551277,
      "grad_norm": 4.229026794433594,
      "learning_rate": 9.122105504662328e-07,
      "loss": 0.7793,
      "step": 8570
    },
    {
      "epoch": 0.13349731837748216,
      "grad_norm": 5.319640159606934,
      "learning_rate": 9.121286114616279e-07,
      "loss": 0.765,
      "step": 8575
    },
    {
      "epoch": 0.13357515937945152,
      "grad_norm": 3.8783299922943115,
      "learning_rate": 9.120466724570229e-07,
      "loss": 0.7036,
      "step": 8580
    },
    {
      "epoch": 0.13365300038142092,
      "grad_norm": 5.243676662445068,
      "learning_rate": 9.11964733452418e-07,
      "loss": 0.8525,
      "step": 8585
    },
    {
      "epoch": 0.13373084138339028,
      "grad_norm": 6.118826866149902,
      "learning_rate": 9.11882794447813e-07,
      "loss": 0.8391,
      "step": 8590
    },
    {
      "epoch": 0.13380868238535967,
      "grad_norm": 2.1617751121520996,
      "learning_rate": 9.118008554432081e-07,
      "loss": 0.7353,
      "step": 8595
    },
    {
      "epoch": 0.13388652338732904,
      "grad_norm": 7.590507507324219,
      "learning_rate": 9.117189164386031e-07,
      "loss": 0.8919,
      "step": 8600
    },
    {
      "epoch": 0.13396436438929843,
      "grad_norm": 3.345130443572998,
      "learning_rate": 9.116369774339982e-07,
      "loss": 0.7615,
      "step": 8605
    },
    {
      "epoch": 0.1340422053912678,
      "grad_norm": 3.144171953201294,
      "learning_rate": 9.115550384293931e-07,
      "loss": 0.7714,
      "step": 8610
    },
    {
      "epoch": 0.13412004639323719,
      "grad_norm": 5.711132049560547,
      "learning_rate": 9.114730994247881e-07,
      "loss": 0.9119,
      "step": 8615
    },
    {
      "epoch": 0.13419788739520655,
      "grad_norm": 3.997664451599121,
      "learning_rate": 9.113911604201832e-07,
      "loss": 0.7821,
      "step": 8620
    },
    {
      "epoch": 0.13427572839717594,
      "grad_norm": 5.540621757507324,
      "learning_rate": 9.113092214155782e-07,
      "loss": 0.8796,
      "step": 8625
    },
    {
      "epoch": 0.1343535693991453,
      "grad_norm": 4.296466827392578,
      "learning_rate": 9.112272824109732e-07,
      "loss": 0.839,
      "step": 8630
    },
    {
      "epoch": 0.13443141040111467,
      "grad_norm": 5.527231693267822,
      "learning_rate": 9.111453434063683e-07,
      "loss": 0.7966,
      "step": 8635
    },
    {
      "epoch": 0.13450925140308406,
      "grad_norm": 4.798453330993652,
      "learning_rate": 9.110634044017634e-07,
      "loss": 0.8808,
      "step": 8640
    },
    {
      "epoch": 0.13458709240505343,
      "grad_norm": 4.989645957946777,
      "learning_rate": 9.109814653971583e-07,
      "loss": 0.7845,
      "step": 8645
    },
    {
      "epoch": 0.13466493340702282,
      "grad_norm": 5.687048435211182,
      "learning_rate": 9.108995263925533e-07,
      "loss": 0.8627,
      "step": 8650
    },
    {
      "epoch": 0.13474277440899218,
      "grad_norm": 3.1098756790161133,
      "learning_rate": 9.108175873879484e-07,
      "loss": 0.7594,
      "step": 8655
    },
    {
      "epoch": 0.13482061541096158,
      "grad_norm": 8.40995979309082,
      "learning_rate": 9.107356483833433e-07,
      "loss": 0.7599,
      "step": 8660
    },
    {
      "epoch": 0.13489845641293094,
      "grad_norm": 6.420483589172363,
      "learning_rate": 9.106537093787384e-07,
      "loss": 0.8199,
      "step": 8665
    },
    {
      "epoch": 0.13497629741490033,
      "grad_norm": 5.545563697814941,
      "learning_rate": 9.105717703741335e-07,
      "loss": 0.7403,
      "step": 8670
    },
    {
      "epoch": 0.1350541384168697,
      "grad_norm": 8.34080696105957,
      "learning_rate": 9.104898313695285e-07,
      "loss": 0.7979,
      "step": 8675
    },
    {
      "epoch": 0.1351319794188391,
      "grad_norm": 6.4696550369262695,
      "learning_rate": 9.104078923649235e-07,
      "loss": 0.6935,
      "step": 8680
    },
    {
      "epoch": 0.13520982042080845,
      "grad_norm": 7.9759521484375,
      "learning_rate": 9.103259533603186e-07,
      "loss": 0.7174,
      "step": 8685
    },
    {
      "epoch": 0.13528766142277784,
      "grad_norm": 4.094228267669678,
      "learning_rate": 9.102440143557135e-07,
      "loss": 0.9191,
      "step": 8690
    },
    {
      "epoch": 0.1353655024247472,
      "grad_norm": 3.2784035205841064,
      "learning_rate": 9.101620753511086e-07,
      "loss": 0.8157,
      "step": 8695
    },
    {
      "epoch": 0.1354433434267166,
      "grad_norm": 3.680067300796509,
      "learning_rate": 9.100801363465036e-07,
      "loss": 0.8478,
      "step": 8700
    },
    {
      "epoch": 0.13552118442868596,
      "grad_norm": 3.991107225418091,
      "learning_rate": 9.099981973418986e-07,
      "loss": 0.9003,
      "step": 8705
    },
    {
      "epoch": 0.13559902543065536,
      "grad_norm": 2.9558584690093994,
      "learning_rate": 9.099162583372937e-07,
      "loss": 0.7571,
      "step": 8710
    },
    {
      "epoch": 0.13567686643262472,
      "grad_norm": 2.952221632003784,
      "learning_rate": 9.098343193326887e-07,
      "loss": 0.8597,
      "step": 8715
    },
    {
      "epoch": 0.13575470743459409,
      "grad_norm": 8.548612594604492,
      "learning_rate": 9.097523803280838e-07,
      "loss": 0.7225,
      "step": 8720
    },
    {
      "epoch": 0.13583254843656348,
      "grad_norm": 9.161630630493164,
      "learning_rate": 9.096704413234788e-07,
      "loss": 0.6646,
      "step": 8725
    },
    {
      "epoch": 0.13591038943853284,
      "grad_norm": 4.948508262634277,
      "learning_rate": 9.095885023188738e-07,
      "loss": 0.6339,
      "step": 8730
    },
    {
      "epoch": 0.13598823044050223,
      "grad_norm": 6.353402614593506,
      "learning_rate": 9.095065633142688e-07,
      "loss": 0.7525,
      "step": 8735
    },
    {
      "epoch": 0.1360660714424716,
      "grad_norm": 2.8166439533233643,
      "learning_rate": 9.094246243096638e-07,
      "loss": 0.7942,
      "step": 8740
    },
    {
      "epoch": 0.136143912444441,
      "grad_norm": 3.2891948223114014,
      "learning_rate": 9.093426853050589e-07,
      "loss": 0.7832,
      "step": 8745
    },
    {
      "epoch": 0.13622175344641035,
      "grad_norm": 6.029998779296875,
      "learning_rate": 9.09260746300454e-07,
      "loss": 0.9606,
      "step": 8750
    },
    {
      "epoch": 0.13629959444837975,
      "grad_norm": 3.9812381267547607,
      "learning_rate": 9.091788072958489e-07,
      "loss": 0.8119,
      "step": 8755
    },
    {
      "epoch": 0.1363774354503491,
      "grad_norm": 2.964101791381836,
      "learning_rate": 9.09096868291244e-07,
      "loss": 0.7792,
      "step": 8760
    },
    {
      "epoch": 0.1364552764523185,
      "grad_norm": 5.025110244750977,
      "learning_rate": 9.090149292866391e-07,
      "loss": 0.9316,
      "step": 8765
    },
    {
      "epoch": 0.13653311745428787,
      "grad_norm": 8.131609916687012,
      "learning_rate": 9.089329902820339e-07,
      "loss": 0.8358,
      "step": 8770
    },
    {
      "epoch": 0.13661095845625726,
      "grad_norm": 6.380354881286621,
      "learning_rate": 9.08851051277429e-07,
      "loss": 0.8002,
      "step": 8775
    },
    {
      "epoch": 0.13668879945822662,
      "grad_norm": 3.922022581100464,
      "learning_rate": 9.087691122728241e-07,
      "loss": 0.7736,
      "step": 8780
    },
    {
      "epoch": 0.13676664046019602,
      "grad_norm": 7.275602340698242,
      "learning_rate": 9.08687173268219e-07,
      "loss": 0.8083,
      "step": 8785
    },
    {
      "epoch": 0.13684448146216538,
      "grad_norm": 4.583987236022949,
      "learning_rate": 9.086052342636141e-07,
      "loss": 0.7587,
      "step": 8790
    },
    {
      "epoch": 0.13692232246413477,
      "grad_norm": 2.945908784866333,
      "learning_rate": 9.085232952590092e-07,
      "loss": 0.7746,
      "step": 8795
    },
    {
      "epoch": 0.13700016346610414,
      "grad_norm": 3.0301320552825928,
      "learning_rate": 9.084413562544042e-07,
      "loss": 0.7701,
      "step": 8800
    },
    {
      "epoch": 0.1370780044680735,
      "grad_norm": 2.6927056312561035,
      "learning_rate": 9.083594172497992e-07,
      "loss": 0.6471,
      "step": 8805
    },
    {
      "epoch": 0.1371558454700429,
      "grad_norm": 3.8082404136657715,
      "learning_rate": 9.082774782451942e-07,
      "loss": 0.8069,
      "step": 8810
    },
    {
      "epoch": 0.13723368647201226,
      "grad_norm": 3.635481595993042,
      "learning_rate": 9.081955392405892e-07,
      "loss": 0.7559,
      "step": 8815
    },
    {
      "epoch": 0.13731152747398165,
      "grad_norm": 3.12910795211792,
      "learning_rate": 9.081136002359843e-07,
      "loss": 0.7618,
      "step": 8820
    },
    {
      "epoch": 0.137389368475951,
      "grad_norm": 4.329128742218018,
      "learning_rate": 9.080316612313793e-07,
      "loss": 0.7491,
      "step": 8825
    },
    {
      "epoch": 0.1374672094779204,
      "grad_norm": 3.8912899494171143,
      "learning_rate": 9.079497222267743e-07,
      "loss": 0.8315,
      "step": 8830
    },
    {
      "epoch": 0.13754505047988977,
      "grad_norm": 5.182728290557861,
      "learning_rate": 9.078677832221694e-07,
      "loss": 0.8504,
      "step": 8835
    },
    {
      "epoch": 0.13762289148185916,
      "grad_norm": 2.6342267990112305,
      "learning_rate": 9.077858442175645e-07,
      "loss": 0.8111,
      "step": 8840
    },
    {
      "epoch": 0.13770073248382853,
      "grad_norm": 5.059619903564453,
      "learning_rate": 9.077039052129595e-07,
      "loss": 0.9168,
      "step": 8845
    },
    {
      "epoch": 0.13777857348579792,
      "grad_norm": 3.4851198196411133,
      "learning_rate": 9.076219662083544e-07,
      "loss": 0.8299,
      "step": 8850
    },
    {
      "epoch": 0.13785641448776728,
      "grad_norm": 8.638031959533691,
      "learning_rate": 9.075400272037495e-07,
      "loss": 0.6855,
      "step": 8855
    },
    {
      "epoch": 0.13793425548973667,
      "grad_norm": 4.0812811851501465,
      "learning_rate": 9.074580881991445e-07,
      "loss": 0.7596,
      "step": 8860
    },
    {
      "epoch": 0.13801209649170604,
      "grad_norm": 3.0092952251434326,
      "learning_rate": 9.073761491945395e-07,
      "loss": 0.7839,
      "step": 8865
    },
    {
      "epoch": 0.13808993749367543,
      "grad_norm": 3.5875368118286133,
      "learning_rate": 9.072942101899346e-07,
      "loss": 0.8173,
      "step": 8870
    },
    {
      "epoch": 0.1381677784956448,
      "grad_norm": 6.807621955871582,
      "learning_rate": 9.072122711853297e-07,
      "loss": 0.8204,
      "step": 8875
    },
    {
      "epoch": 0.13824561949761419,
      "grad_norm": 5.541608810424805,
      "learning_rate": 9.071303321807246e-07,
      "loss": 0.7516,
      "step": 8880
    },
    {
      "epoch": 0.13832346049958355,
      "grad_norm": 4.05478048324585,
      "learning_rate": 9.070483931761197e-07,
      "loss": 0.841,
      "step": 8885
    },
    {
      "epoch": 0.13840130150155291,
      "grad_norm": 3.128432512283325,
      "learning_rate": 9.069664541715148e-07,
      "loss": 0.694,
      "step": 8890
    },
    {
      "epoch": 0.1384791425035223,
      "grad_norm": 4.354421138763428,
      "learning_rate": 9.068845151669096e-07,
      "loss": 0.885,
      "step": 8895
    },
    {
      "epoch": 0.13855698350549167,
      "grad_norm": 4.6781134605407715,
      "learning_rate": 9.068025761623047e-07,
      "loss": 0.9041,
      "step": 8900
    },
    {
      "epoch": 0.13863482450746106,
      "grad_norm": 12.059392929077148,
      "learning_rate": 9.067206371576998e-07,
      "loss": 0.7997,
      "step": 8905
    },
    {
      "epoch": 0.13871266550943043,
      "grad_norm": 2.994907855987549,
      "learning_rate": 9.066386981530948e-07,
      "loss": 0.7988,
      "step": 8910
    },
    {
      "epoch": 0.13879050651139982,
      "grad_norm": 4.420156478881836,
      "learning_rate": 9.065567591484898e-07,
      "loss": 0.8653,
      "step": 8915
    },
    {
      "epoch": 0.13886834751336918,
      "grad_norm": 8.454998016357422,
      "learning_rate": 9.064748201438849e-07,
      "loss": 0.8991,
      "step": 8920
    },
    {
      "epoch": 0.13894618851533858,
      "grad_norm": 3.3839731216430664,
      "learning_rate": 9.063928811392799e-07,
      "loss": 0.7173,
      "step": 8925
    },
    {
      "epoch": 0.13902402951730794,
      "grad_norm": 5.453253746032715,
      "learning_rate": 9.06310942134675e-07,
      "loss": 0.7141,
      "step": 8930
    },
    {
      "epoch": 0.13910187051927733,
      "grad_norm": 9.229926109313965,
      "learning_rate": 9.062290031300699e-07,
      "loss": 0.7292,
      "step": 8935
    },
    {
      "epoch": 0.1391797115212467,
      "grad_norm": 5.860846042633057,
      "learning_rate": 9.061470641254649e-07,
      "loss": 0.8563,
      "step": 8940
    },
    {
      "epoch": 0.1392575525232161,
      "grad_norm": 4.182551383972168,
      "learning_rate": 9.0606512512086e-07,
      "loss": 0.7552,
      "step": 8945
    },
    {
      "epoch": 0.13933539352518545,
      "grad_norm": 3.165614604949951,
      "learning_rate": 9.05983186116255e-07,
      "loss": 0.7887,
      "step": 8950
    },
    {
      "epoch": 0.13941323452715484,
      "grad_norm": 3.805906057357788,
      "learning_rate": 9.0590124711165e-07,
      "loss": 0.8226,
      "step": 8955
    },
    {
      "epoch": 0.1394910755291242,
      "grad_norm": 4.190842151641846,
      "learning_rate": 9.058193081070451e-07,
      "loss": 0.8951,
      "step": 8960
    },
    {
      "epoch": 0.1395689165310936,
      "grad_norm": 3.0468575954437256,
      "learning_rate": 9.057373691024402e-07,
      "loss": 0.7346,
      "step": 8965
    },
    {
      "epoch": 0.13964675753306297,
      "grad_norm": 4.917840003967285,
      "learning_rate": 9.056554300978352e-07,
      "loss": 0.8478,
      "step": 8970
    },
    {
      "epoch": 0.13972459853503233,
      "grad_norm": 3.5989246368408203,
      "learning_rate": 9.055734910932301e-07,
      "loss": 0.702,
      "step": 8975
    },
    {
      "epoch": 0.13980243953700172,
      "grad_norm": 3.706799030303955,
      "learning_rate": 9.054915520886252e-07,
      "loss": 0.6731,
      "step": 8980
    },
    {
      "epoch": 0.13988028053897109,
      "grad_norm": 2.194471836090088,
      "learning_rate": 9.054096130840202e-07,
      "loss": 0.7718,
      "step": 8985
    },
    {
      "epoch": 0.13995812154094048,
      "grad_norm": 4.610592842102051,
      "learning_rate": 9.053276740794152e-07,
      "loss": 0.657,
      "step": 8990
    },
    {
      "epoch": 0.14003596254290984,
      "grad_norm": 9.641939163208008,
      "learning_rate": 9.052457350748103e-07,
      "loss": 0.7753,
      "step": 8995
    },
    {
      "epoch": 0.14011380354487923,
      "grad_norm": 4.634001731872559,
      "learning_rate": 9.051637960702054e-07,
      "loss": 0.8575,
      "step": 9000
    },
    {
      "epoch": 0.1401916445468486,
      "grad_norm": 4.204237461090088,
      "learning_rate": 9.050818570656003e-07,
      "loss": 0.8538,
      "step": 9005
    },
    {
      "epoch": 0.140269485548818,
      "grad_norm": 4.357415199279785,
      "learning_rate": 9.049999180609954e-07,
      "loss": 0.7938,
      "step": 9010
    },
    {
      "epoch": 0.14034732655078735,
      "grad_norm": 6.758501052856445,
      "learning_rate": 9.049179790563904e-07,
      "loss": 0.867,
      "step": 9015
    },
    {
      "epoch": 0.14042516755275675,
      "grad_norm": 4.51594877243042,
      "learning_rate": 9.048360400517853e-07,
      "loss": 0.7706,
      "step": 9020
    },
    {
      "epoch": 0.1405030085547261,
      "grad_norm": 3.3741414546966553,
      "learning_rate": 9.047541010471804e-07,
      "loss": 0.8655,
      "step": 9025
    },
    {
      "epoch": 0.1405808495566955,
      "grad_norm": 5.02528190612793,
      "learning_rate": 9.046721620425755e-07,
      "loss": 0.7714,
      "step": 9030
    },
    {
      "epoch": 0.14065869055866487,
      "grad_norm": 3.8374040126800537,
      "learning_rate": 9.045902230379705e-07,
      "loss": 0.8834,
      "step": 9035
    },
    {
      "epoch": 0.14073653156063426,
      "grad_norm": 3.829531192779541,
      "learning_rate": 9.045082840333655e-07,
      "loss": 0.8477,
      "step": 9040
    },
    {
      "epoch": 0.14081437256260362,
      "grad_norm": 7.790329456329346,
      "learning_rate": 9.044263450287606e-07,
      "loss": 0.8359,
      "step": 9045
    },
    {
      "epoch": 0.14089221356457302,
      "grad_norm": 4.2702460289001465,
      "learning_rate": 9.043444060241556e-07,
      "loss": 0.8458,
      "step": 9050
    },
    {
      "epoch": 0.14097005456654238,
      "grad_norm": 5.226367950439453,
      "learning_rate": 9.042624670195506e-07,
      "loss": 0.9577,
      "step": 9055
    },
    {
      "epoch": 0.14104789556851174,
      "grad_norm": 4.627621650695801,
      "learning_rate": 9.041805280149456e-07,
      "loss": 0.8082,
      "step": 9060
    },
    {
      "epoch": 0.14112573657048114,
      "grad_norm": 4.152894973754883,
      "learning_rate": 9.040985890103406e-07,
      "loss": 0.6948,
      "step": 9065
    },
    {
      "epoch": 0.1412035775724505,
      "grad_norm": 6.219531059265137,
      "learning_rate": 9.040166500057357e-07,
      "loss": 0.8201,
      "step": 9070
    },
    {
      "epoch": 0.1412814185744199,
      "grad_norm": 4.419485569000244,
      "learning_rate": 9.039347110011307e-07,
      "loss": 0.8803,
      "step": 9075
    },
    {
      "epoch": 0.14135925957638926,
      "grad_norm": 4.359714508056641,
      "learning_rate": 9.038527719965257e-07,
      "loss": 0.7647,
      "step": 9080
    },
    {
      "epoch": 0.14143710057835865,
      "grad_norm": 6.366148948669434,
      "learning_rate": 9.037708329919208e-07,
      "loss": 0.8826,
      "step": 9085
    },
    {
      "epoch": 0.141514941580328,
      "grad_norm": 3.5685646533966064,
      "learning_rate": 9.036888939873159e-07,
      "loss": 0.7997,
      "step": 9090
    },
    {
      "epoch": 0.1415927825822974,
      "grad_norm": 3.0451033115386963,
      "learning_rate": 9.036069549827107e-07,
      "loss": 0.7774,
      "step": 9095
    },
    {
      "epoch": 0.14167062358426677,
      "grad_norm": 4.7669291496276855,
      "learning_rate": 9.035250159781058e-07,
      "loss": 0.8111,
      "step": 9100
    },
    {
      "epoch": 0.14174846458623616,
      "grad_norm": 3.8222289085388184,
      "learning_rate": 9.034430769735009e-07,
      "loss": 0.8369,
      "step": 9105
    },
    {
      "epoch": 0.14182630558820553,
      "grad_norm": 9.232769966125488,
      "learning_rate": 9.03361137968896e-07,
      "loss": 0.7547,
      "step": 9110
    },
    {
      "epoch": 0.14190414659017492,
      "grad_norm": 2.8610751628875732,
      "learning_rate": 9.032791989642909e-07,
      "loss": 0.6919,
      "step": 9115
    },
    {
      "epoch": 0.14198198759214428,
      "grad_norm": 5.900112152099609,
      "learning_rate": 9.03197259959686e-07,
      "loss": 0.8343,
      "step": 9120
    },
    {
      "epoch": 0.14205982859411367,
      "grad_norm": 4.022780418395996,
      "learning_rate": 9.031153209550811e-07,
      "loss": 0.867,
      "step": 9125
    },
    {
      "epoch": 0.14213766959608304,
      "grad_norm": 3.068723201751709,
      "learning_rate": 9.03033381950476e-07,
      "loss": 0.7645,
      "step": 9130
    },
    {
      "epoch": 0.14221551059805243,
      "grad_norm": 3.156970977783203,
      "learning_rate": 9.02951442945871e-07,
      "loss": 0.6631,
      "step": 9135
    },
    {
      "epoch": 0.1422933516000218,
      "grad_norm": 3.7697770595550537,
      "learning_rate": 9.028695039412661e-07,
      "loss": 0.7945,
      "step": 9140
    },
    {
      "epoch": 0.1423711926019912,
      "grad_norm": 9.904714584350586,
      "learning_rate": 9.027875649366611e-07,
      "loss": 0.8339,
      "step": 9145
    },
    {
      "epoch": 0.14244903360396055,
      "grad_norm": 3.7701642513275146,
      "learning_rate": 9.027056259320561e-07,
      "loss": 0.7597,
      "step": 9150
    },
    {
      "epoch": 0.14252687460592992,
      "grad_norm": 4.5266828536987305,
      "learning_rate": 9.026236869274512e-07,
      "loss": 0.7383,
      "step": 9155
    },
    {
      "epoch": 0.1426047156078993,
      "grad_norm": 5.614555358886719,
      "learning_rate": 9.025417479228462e-07,
      "loss": 0.7889,
      "step": 9160
    },
    {
      "epoch": 0.14268255660986867,
      "grad_norm": 7.812586784362793,
      "learning_rate": 9.024598089182412e-07,
      "loss": 0.8535,
      "step": 9165
    },
    {
      "epoch": 0.14276039761183806,
      "grad_norm": 5.218164443969727,
      "learning_rate": 9.023778699136363e-07,
      "loss": 0.814,
      "step": 9170
    },
    {
      "epoch": 0.14283823861380743,
      "grad_norm": 2.824345588684082,
      "learning_rate": 9.022959309090312e-07,
      "loss": 0.8305,
      "step": 9175
    },
    {
      "epoch": 0.14291607961577682,
      "grad_norm": 3.954688787460327,
      "learning_rate": 9.022139919044263e-07,
      "loss": 0.8707,
      "step": 9180
    },
    {
      "epoch": 0.14299392061774618,
      "grad_norm": 4.215836048126221,
      "learning_rate": 9.021320528998213e-07,
      "loss": 0.9382,
      "step": 9185
    },
    {
      "epoch": 0.14307176161971558,
      "grad_norm": 9.17077350616455,
      "learning_rate": 9.020501138952163e-07,
      "loss": 0.8275,
      "step": 9190
    },
    {
      "epoch": 0.14314960262168494,
      "grad_norm": 6.527652740478516,
      "learning_rate": 9.019681748906114e-07,
      "loss": 0.7504,
      "step": 9195
    },
    {
      "epoch": 0.14322744362365433,
      "grad_norm": 3.313584089279175,
      "learning_rate": 9.018862358860065e-07,
      "loss": 0.8984,
      "step": 9200
    },
    {
      "epoch": 0.1433052846256237,
      "grad_norm": 4.872673988342285,
      "learning_rate": 9.018042968814014e-07,
      "loss": 0.9078,
      "step": 9205
    },
    {
      "epoch": 0.1433831256275931,
      "grad_norm": 4.736050128936768,
      "learning_rate": 9.017223578767965e-07,
      "loss": 0.7167,
      "step": 9210
    },
    {
      "epoch": 0.14346096662956245,
      "grad_norm": 3.0113682746887207,
      "learning_rate": 9.016404188721916e-07,
      "loss": 0.901,
      "step": 9215
    },
    {
      "epoch": 0.14353880763153185,
      "grad_norm": 4.565042495727539,
      "learning_rate": 9.015584798675864e-07,
      "loss": 0.7495,
      "step": 9220
    },
    {
      "epoch": 0.1436166486335012,
      "grad_norm": 3.0236637592315674,
      "learning_rate": 9.014765408629815e-07,
      "loss": 0.7655,
      "step": 9225
    },
    {
      "epoch": 0.1436944896354706,
      "grad_norm": 5.903986930847168,
      "learning_rate": 9.013946018583766e-07,
      "loss": 0.8993,
      "step": 9230
    },
    {
      "epoch": 0.14377233063743997,
      "grad_norm": 4.84224796295166,
      "learning_rate": 9.013126628537717e-07,
      "loss": 0.7633,
      "step": 9235
    },
    {
      "epoch": 0.14385017163940933,
      "grad_norm": 3.5455453395843506,
      "learning_rate": 9.012307238491666e-07,
      "loss": 0.7984,
      "step": 9240
    },
    {
      "epoch": 0.14392801264137872,
      "grad_norm": 9.098531723022461,
      "learning_rate": 9.011487848445617e-07,
      "loss": 0.8721,
      "step": 9245
    },
    {
      "epoch": 0.1440058536433481,
      "grad_norm": 3.4469218254089355,
      "learning_rate": 9.010668458399568e-07,
      "loss": 0.7364,
      "step": 9250
    },
    {
      "epoch": 0.14408369464531748,
      "grad_norm": 3.0461971759796143,
      "learning_rate": 9.009849068353517e-07,
      "loss": 0.8158,
      "step": 9255
    },
    {
      "epoch": 0.14416153564728684,
      "grad_norm": 5.753521919250488,
      "learning_rate": 9.009029678307467e-07,
      "loss": 0.7637,
      "step": 9260
    },
    {
      "epoch": 0.14423937664925623,
      "grad_norm": 10.035242080688477,
      "learning_rate": 9.008210288261418e-07,
      "loss": 0.8554,
      "step": 9265
    },
    {
      "epoch": 0.1443172176512256,
      "grad_norm": 3.796072483062744,
      "learning_rate": 9.007390898215368e-07,
      "loss": 0.8241,
      "step": 9270
    },
    {
      "epoch": 0.144395058653195,
      "grad_norm": 4.330554962158203,
      "learning_rate": 9.006571508169318e-07,
      "loss": 0.7456,
      "step": 9275
    },
    {
      "epoch": 0.14447289965516436,
      "grad_norm": 3.3564655780792236,
      "learning_rate": 9.005752118123269e-07,
      "loss": 0.8016,
      "step": 9280
    },
    {
      "epoch": 0.14455074065713375,
      "grad_norm": 2.9985568523406982,
      "learning_rate": 9.004932728077219e-07,
      "loss": 0.8099,
      "step": 9285
    },
    {
      "epoch": 0.1446285816591031,
      "grad_norm": 2.705263376235962,
      "learning_rate": 9.00411333803117e-07,
      "loss": 0.7769,
      "step": 9290
    },
    {
      "epoch": 0.1447064226610725,
      "grad_norm": 3.700831890106201,
      "learning_rate": 9.00329394798512e-07,
      "loss": 0.7327,
      "step": 9295
    },
    {
      "epoch": 0.14478426366304187,
      "grad_norm": 6.56169319152832,
      "learning_rate": 9.002474557939069e-07,
      "loss": 0.7674,
      "step": 9300
    },
    {
      "epoch": 0.14486210466501126,
      "grad_norm": 4.469850063323975,
      "learning_rate": 9.00165516789302e-07,
      "loss": 0.8658,
      "step": 9305
    },
    {
      "epoch": 0.14493994566698062,
      "grad_norm": 8.878783226013184,
      "learning_rate": 9.00083577784697e-07,
      "loss": 0.7462,
      "step": 9310
    },
    {
      "epoch": 0.14501778666895002,
      "grad_norm": 2.435040235519409,
      "learning_rate": 9.00001638780092e-07,
      "loss": 0.8363,
      "step": 9315
    },
    {
      "epoch": 0.14509562767091938,
      "grad_norm": 3.0248606204986572,
      "learning_rate": 8.999196997754871e-07,
      "loss": 0.8078,
      "step": 9320
    },
    {
      "epoch": 0.14517346867288874,
      "grad_norm": 3.466975450515747,
      "learning_rate": 8.998377607708822e-07,
      "loss": 0.7242,
      "step": 9325
    },
    {
      "epoch": 0.14525130967485814,
      "grad_norm": 3.6273536682128906,
      "learning_rate": 8.997558217662771e-07,
      "loss": 0.8677,
      "step": 9330
    },
    {
      "epoch": 0.1453291506768275,
      "grad_norm": 6.970489025115967,
      "learning_rate": 8.996738827616722e-07,
      "loss": 0.8854,
      "step": 9335
    },
    {
      "epoch": 0.1454069916787969,
      "grad_norm": 3.735153913497925,
      "learning_rate": 8.995919437570672e-07,
      "loss": 0.7732,
      "step": 9340
    },
    {
      "epoch": 0.14548483268076626,
      "grad_norm": 8.700521469116211,
      "learning_rate": 8.995100047524621e-07,
      "loss": 0.9042,
      "step": 9345
    },
    {
      "epoch": 0.14556267368273565,
      "grad_norm": 7.9652276039123535,
      "learning_rate": 8.994280657478572e-07,
      "loss": 0.867,
      "step": 9350
    },
    {
      "epoch": 0.145640514684705,
      "grad_norm": 3.8336143493652344,
      "learning_rate": 8.993461267432523e-07,
      "loss": 0.8283,
      "step": 9355
    },
    {
      "epoch": 0.1457183556866744,
      "grad_norm": 7.037674903869629,
      "learning_rate": 8.992641877386474e-07,
      "loss": 0.7827,
      "step": 9360
    },
    {
      "epoch": 0.14579619668864377,
      "grad_norm": 6.455174922943115,
      "learning_rate": 8.991822487340423e-07,
      "loss": 0.7666,
      "step": 9365
    },
    {
      "epoch": 0.14587403769061316,
      "grad_norm": 3.610822916030884,
      "learning_rate": 8.991003097294374e-07,
      "loss": 0.8234,
      "step": 9370
    },
    {
      "epoch": 0.14595187869258253,
      "grad_norm": 4.823882102966309,
      "learning_rate": 8.990183707248325e-07,
      "loss": 0.7568,
      "step": 9375
    },
    {
      "epoch": 0.14602971969455192,
      "grad_norm": 5.625290393829346,
      "learning_rate": 8.989364317202273e-07,
      "loss": 0.7957,
      "step": 9380
    },
    {
      "epoch": 0.14610756069652128,
      "grad_norm": 5.454258441925049,
      "learning_rate": 8.988544927156224e-07,
      "loss": 0.6412,
      "step": 9385
    },
    {
      "epoch": 0.14618540169849067,
      "grad_norm": 5.896919250488281,
      "learning_rate": 8.987725537110175e-07,
      "loss": 0.7889,
      "step": 9390
    },
    {
      "epoch": 0.14626324270046004,
      "grad_norm": 3.3774161338806152,
      "learning_rate": 8.986906147064125e-07,
      "loss": 0.9058,
      "step": 9395
    },
    {
      "epoch": 0.14634108370242943,
      "grad_norm": 3.2311689853668213,
      "learning_rate": 8.986086757018075e-07,
      "loss": 0.7764,
      "step": 9400
    },
    {
      "epoch": 0.1464189247043988,
      "grad_norm": 5.298305511474609,
      "learning_rate": 8.985267366972026e-07,
      "loss": 0.7435,
      "step": 9405
    },
    {
      "epoch": 0.14649676570636816,
      "grad_norm": 4.45401668548584,
      "learning_rate": 8.984447976925976e-07,
      "loss": 0.7631,
      "step": 9410
    },
    {
      "epoch": 0.14657460670833755,
      "grad_norm": 3.3566505908966064,
      "learning_rate": 8.983628586879927e-07,
      "loss": 0.8497,
      "step": 9415
    },
    {
      "epoch": 0.14665244771030692,
      "grad_norm": 3.440115213394165,
      "learning_rate": 8.982809196833876e-07,
      "loss": 0.7614,
      "step": 9420
    },
    {
      "epoch": 0.1467302887122763,
      "grad_norm": 3.5758979320526123,
      "learning_rate": 8.981989806787826e-07,
      "loss": 0.651,
      "step": 9425
    },
    {
      "epoch": 0.14680812971424567,
      "grad_norm": 4.706757545471191,
      "learning_rate": 8.981170416741777e-07,
      "loss": 0.9389,
      "step": 9430
    },
    {
      "epoch": 0.14688597071621506,
      "grad_norm": 4.142934799194336,
      "learning_rate": 8.980351026695727e-07,
      "loss": 0.9398,
      "step": 9435
    },
    {
      "epoch": 0.14696381171818443,
      "grad_norm": 3.549654483795166,
      "learning_rate": 8.979531636649677e-07,
      "loss": 0.7552,
      "step": 9440
    },
    {
      "epoch": 0.14704165272015382,
      "grad_norm": 3.9092509746551514,
      "learning_rate": 8.978712246603628e-07,
      "loss": 0.8207,
      "step": 9445
    },
    {
      "epoch": 0.14711949372212318,
      "grad_norm": 4.208457946777344,
      "learning_rate": 8.977892856557579e-07,
      "loss": 0.7046,
      "step": 9450
    },
    {
      "epoch": 0.14719733472409258,
      "grad_norm": 8.339073181152344,
      "learning_rate": 8.977073466511528e-07,
      "loss": 0.7736,
      "step": 9455
    },
    {
      "epoch": 0.14727517572606194,
      "grad_norm": 2.834137439727783,
      "learning_rate": 8.976254076465478e-07,
      "loss": 0.893,
      "step": 9460
    },
    {
      "epoch": 0.14735301672803133,
      "grad_norm": 4.870460033416748,
      "learning_rate": 8.975434686419429e-07,
      "loss": 0.819,
      "step": 9465
    },
    {
      "epoch": 0.1474308577300007,
      "grad_norm": 4.1774582862854,
      "learning_rate": 8.974615296373378e-07,
      "loss": 0.7164,
      "step": 9470
    },
    {
      "epoch": 0.1475086987319701,
      "grad_norm": 3.573882818222046,
      "learning_rate": 8.973795906327329e-07,
      "loss": 0.7982,
      "step": 9475
    },
    {
      "epoch": 0.14758653973393945,
      "grad_norm": 3.5654754638671875,
      "learning_rate": 8.97297651628128e-07,
      "loss": 0.8293,
      "step": 9480
    },
    {
      "epoch": 0.14766438073590885,
      "grad_norm": 3.387242078781128,
      "learning_rate": 8.972157126235231e-07,
      "loss": 0.7351,
      "step": 9485
    },
    {
      "epoch": 0.1477422217378782,
      "grad_norm": 7.289836883544922,
      "learning_rate": 8.97133773618918e-07,
      "loss": 0.8013,
      "step": 9490
    },
    {
      "epoch": 0.14782006273984757,
      "grad_norm": 4.657222270965576,
      "learning_rate": 8.970518346143131e-07,
      "loss": 0.8237,
      "step": 9495
    },
    {
      "epoch": 0.14789790374181697,
      "grad_norm": 4.462035179138184,
      "learning_rate": 8.969698956097081e-07,
      "loss": 0.7025,
      "step": 9500
    },
    {
      "epoch": 0.14797574474378633,
      "grad_norm": 4.082338809967041,
      "learning_rate": 8.968879566051031e-07,
      "loss": 0.7665,
      "step": 9505
    },
    {
      "epoch": 0.14805358574575572,
      "grad_norm": 4.678539276123047,
      "learning_rate": 8.968060176004981e-07,
      "loss": 0.8325,
      "step": 9510
    },
    {
      "epoch": 0.1481314267477251,
      "grad_norm": 4.395145893096924,
      "learning_rate": 8.967240785958932e-07,
      "loss": 0.7172,
      "step": 9515
    },
    {
      "epoch": 0.14820926774969448,
      "grad_norm": 3.5010783672332764,
      "learning_rate": 8.966421395912882e-07,
      "loss": 0.7459,
      "step": 9520
    },
    {
      "epoch": 0.14828710875166384,
      "grad_norm": 5.013630390167236,
      "learning_rate": 8.965602005866832e-07,
      "loss": 0.7536,
      "step": 9525
    },
    {
      "epoch": 0.14836494975363324,
      "grad_norm": 3.0716755390167236,
      "learning_rate": 8.964782615820783e-07,
      "loss": 0.8559,
      "step": 9530
    },
    {
      "epoch": 0.1484427907556026,
      "grad_norm": 8.94082260131836,
      "learning_rate": 8.963963225774733e-07,
      "loss": 0.8765,
      "step": 9535
    },
    {
      "epoch": 0.148520631757572,
      "grad_norm": 5.629791736602783,
      "learning_rate": 8.963143835728684e-07,
      "loss": 0.9076,
      "step": 9540
    },
    {
      "epoch": 0.14859847275954136,
      "grad_norm": 4.679615020751953,
      "learning_rate": 8.962324445682633e-07,
      "loss": 0.7733,
      "step": 9545
    },
    {
      "epoch": 0.14867631376151075,
      "grad_norm": 2.7258129119873047,
      "learning_rate": 8.961505055636583e-07,
      "loss": 0.7082,
      "step": 9550
    },
    {
      "epoch": 0.1487541547634801,
      "grad_norm": 4.701254844665527,
      "learning_rate": 8.960685665590534e-07,
      "loss": 0.6848,
      "step": 9555
    },
    {
      "epoch": 0.1488319957654495,
      "grad_norm": 3.3973538875579834,
      "learning_rate": 8.959866275544485e-07,
      "loss": 0.87,
      "step": 9560
    },
    {
      "epoch": 0.14890983676741887,
      "grad_norm": 5.04672384262085,
      "learning_rate": 8.959046885498434e-07,
      "loss": 0.7941,
      "step": 9565
    },
    {
      "epoch": 0.14898767776938826,
      "grad_norm": 5.399540901184082,
      "learning_rate": 8.958227495452385e-07,
      "loss": 0.7677,
      "step": 9570
    },
    {
      "epoch": 0.14906551877135762,
      "grad_norm": 3.8206582069396973,
      "learning_rate": 8.957408105406336e-07,
      "loss": 0.7326,
      "step": 9575
    },
    {
      "epoch": 0.149143359773327,
      "grad_norm": 2.07293963432312,
      "learning_rate": 8.956588715360285e-07,
      "loss": 0.7492,
      "step": 9580
    },
    {
      "epoch": 0.14922120077529638,
      "grad_norm": 8.431818008422852,
      "learning_rate": 8.955769325314235e-07,
      "loss": 0.7455,
      "step": 9585
    },
    {
      "epoch": 0.14929904177726575,
      "grad_norm": 7.549500465393066,
      "learning_rate": 8.954949935268186e-07,
      "loss": 0.7746,
      "step": 9590
    },
    {
      "epoch": 0.14937688277923514,
      "grad_norm": 5.316057205200195,
      "learning_rate": 8.954130545222136e-07,
      "loss": 0.7876,
      "step": 9595
    },
    {
      "epoch": 0.1494547237812045,
      "grad_norm": 7.522194862365723,
      "learning_rate": 8.953311155176086e-07,
      "loss": 0.802,
      "step": 9600
    },
    {
      "epoch": 0.1495325647831739,
      "grad_norm": 3.6271915435791016,
      "learning_rate": 8.952491765130037e-07,
      "loss": 0.7795,
      "step": 9605
    },
    {
      "epoch": 0.14961040578514326,
      "grad_norm": 5.297993183135986,
      "learning_rate": 8.951672375083988e-07,
      "loss": 0.7834,
      "step": 9610
    },
    {
      "epoch": 0.14968824678711265,
      "grad_norm": 3.3175265789031982,
      "learning_rate": 8.950852985037937e-07,
      "loss": 0.6759,
      "step": 9615
    },
    {
      "epoch": 0.14976608778908201,
      "grad_norm": 3.857908010482788,
      "learning_rate": 8.950033594991888e-07,
      "loss": 0.7961,
      "step": 9620
    },
    {
      "epoch": 0.1498439287910514,
      "grad_norm": 6.590905666351318,
      "learning_rate": 8.949214204945838e-07,
      "loss": 0.8008,
      "step": 9625
    },
    {
      "epoch": 0.14992176979302077,
      "grad_norm": 3.9845802783966064,
      "learning_rate": 8.948394814899788e-07,
      "loss": 0.8439,
      "step": 9630
    },
    {
      "epoch": 0.14999961079499016,
      "grad_norm": 4.234025001525879,
      "learning_rate": 8.947575424853738e-07,
      "loss": 0.829,
      "step": 9635
    },
    {
      "epoch": 0.15007745179695953,
      "grad_norm": 4.937478542327881,
      "learning_rate": 8.946756034807689e-07,
      "loss": 0.8498,
      "step": 9640
    },
    {
      "epoch": 0.15015529279892892,
      "grad_norm": 5.999597072601318,
      "learning_rate": 8.945936644761639e-07,
      "loss": 0.7593,
      "step": 9645
    },
    {
      "epoch": 0.15023313380089828,
      "grad_norm": 5.461806774139404,
      "learning_rate": 8.94511725471559e-07,
      "loss": 0.8598,
      "step": 9650
    },
    {
      "epoch": 0.15031097480286768,
      "grad_norm": 5.157901763916016,
      "learning_rate": 8.94429786466954e-07,
      "loss": 0.8738,
      "step": 9655
    },
    {
      "epoch": 0.15038881580483704,
      "grad_norm": 3.0504953861236572,
      "learning_rate": 8.94347847462349e-07,
      "loss": 0.7486,
      "step": 9660
    },
    {
      "epoch": 0.1504666568068064,
      "grad_norm": 10.305487632751465,
      "learning_rate": 8.94265908457744e-07,
      "loss": 0.7844,
      "step": 9665
    },
    {
      "epoch": 0.1505444978087758,
      "grad_norm": 3.5192887783050537,
      "learning_rate": 8.94183969453139e-07,
      "loss": 0.8386,
      "step": 9670
    },
    {
      "epoch": 0.15062233881074516,
      "grad_norm": 4.425221920013428,
      "learning_rate": 8.94102030448534e-07,
      "loss": 0.7428,
      "step": 9675
    },
    {
      "epoch": 0.15070017981271455,
      "grad_norm": 5.125747203826904,
      "learning_rate": 8.940200914439291e-07,
      "loss": 0.7855,
      "step": 9680
    },
    {
      "epoch": 0.15077802081468392,
      "grad_norm": 5.673059463500977,
      "learning_rate": 8.939381524393242e-07,
      "loss": 0.9193,
      "step": 9685
    },
    {
      "epoch": 0.1508558618166533,
      "grad_norm": 2.934014320373535,
      "learning_rate": 8.938562134347191e-07,
      "loss": 0.6473,
      "step": 9690
    },
    {
      "epoch": 0.15093370281862267,
      "grad_norm": 6.39523983001709,
      "learning_rate": 8.937742744301142e-07,
      "loss": 0.7309,
      "step": 9695
    },
    {
      "epoch": 0.15101154382059206,
      "grad_norm": 3.2489750385284424,
      "learning_rate": 8.936923354255093e-07,
      "loss": 0.8391,
      "step": 9700
    },
    {
      "epoch": 0.15108938482256143,
      "grad_norm": 3.2904443740844727,
      "learning_rate": 8.936103964209041e-07,
      "loss": 0.6639,
      "step": 9705
    },
    {
      "epoch": 0.15116722582453082,
      "grad_norm": 5.773887634277344,
      "learning_rate": 8.935284574162992e-07,
      "loss": 0.7859,
      "step": 9710
    },
    {
      "epoch": 0.15124506682650019,
      "grad_norm": 4.7919697761535645,
      "learning_rate": 8.934465184116943e-07,
      "loss": 0.7519,
      "step": 9715
    },
    {
      "epoch": 0.15132290782846958,
      "grad_norm": 4.2467942237854,
      "learning_rate": 8.933645794070893e-07,
      "loss": 0.8234,
      "step": 9720
    },
    {
      "epoch": 0.15140074883043894,
      "grad_norm": 3.2644877433776855,
      "learning_rate": 8.932826404024843e-07,
      "loss": 0.6703,
      "step": 9725
    },
    {
      "epoch": 0.15147858983240833,
      "grad_norm": 4.7553534507751465,
      "learning_rate": 8.932007013978794e-07,
      "loss": 0.7273,
      "step": 9730
    },
    {
      "epoch": 0.1515564308343777,
      "grad_norm": 3.009917974472046,
      "learning_rate": 8.931187623932745e-07,
      "loss": 0.8056,
      "step": 9735
    },
    {
      "epoch": 0.1516342718363471,
      "grad_norm": 2.296459436416626,
      "learning_rate": 8.930368233886695e-07,
      "loss": 0.7561,
      "step": 9740
    },
    {
      "epoch": 0.15171211283831645,
      "grad_norm": 9.049659729003906,
      "learning_rate": 8.929548843840644e-07,
      "loss": 0.7144,
      "step": 9745
    },
    {
      "epoch": 0.15178995384028582,
      "grad_norm": 6.223394393920898,
      "learning_rate": 8.928729453794595e-07,
      "loss": 0.737,
      "step": 9750
    },
    {
      "epoch": 0.1518677948422552,
      "grad_norm": 7.632771968841553,
      "learning_rate": 8.927910063748545e-07,
      "loss": 0.8498,
      "step": 9755
    },
    {
      "epoch": 0.15194563584422457,
      "grad_norm": 6.848038196563721,
      "learning_rate": 8.927090673702495e-07,
      "loss": 0.806,
      "step": 9760
    },
    {
      "epoch": 0.15202347684619397,
      "grad_norm": 4.97011137008667,
      "learning_rate": 8.926271283656446e-07,
      "loss": 0.7157,
      "step": 9765
    },
    {
      "epoch": 0.15210131784816333,
      "grad_norm": 3.5741093158721924,
      "learning_rate": 8.925451893610396e-07,
      "loss": 0.8259,
      "step": 9770
    },
    {
      "epoch": 0.15217915885013272,
      "grad_norm": 3.7775120735168457,
      "learning_rate": 8.924632503564347e-07,
      "loss": 0.7166,
      "step": 9775
    },
    {
      "epoch": 0.1522569998521021,
      "grad_norm": 3.5487003326416016,
      "learning_rate": 8.923813113518297e-07,
      "loss": 0.7971,
      "step": 9780
    },
    {
      "epoch": 0.15233484085407148,
      "grad_norm": 4.203275203704834,
      "learning_rate": 8.922993723472246e-07,
      "loss": 0.8493,
      "step": 9785
    },
    {
      "epoch": 0.15241268185604084,
      "grad_norm": 3.4415955543518066,
      "learning_rate": 8.922174333426197e-07,
      "loss": 0.7878,
      "step": 9790
    },
    {
      "epoch": 0.15249052285801024,
      "grad_norm": 3.142867088317871,
      "learning_rate": 8.921354943380148e-07,
      "loss": 0.7424,
      "step": 9795
    },
    {
      "epoch": 0.1525683638599796,
      "grad_norm": 4.120011806488037,
      "learning_rate": 8.920535553334097e-07,
      "loss": 0.8535,
      "step": 9800
    },
    {
      "epoch": 0.152646204861949,
      "grad_norm": 3.5447614192962646,
      "learning_rate": 8.919716163288048e-07,
      "loss": 0.7767,
      "step": 9805
    },
    {
      "epoch": 0.15272404586391836,
      "grad_norm": 3.2715671062469482,
      "learning_rate": 8.918896773241999e-07,
      "loss": 0.8062,
      "step": 9810
    },
    {
      "epoch": 0.15280188686588775,
      "grad_norm": 2.4431989192962646,
      "learning_rate": 8.918077383195948e-07,
      "loss": 0.7578,
      "step": 9815
    },
    {
      "epoch": 0.1528797278678571,
      "grad_norm": 3.4583024978637695,
      "learning_rate": 8.917257993149899e-07,
      "loss": 0.7531,
      "step": 9820
    },
    {
      "epoch": 0.1529575688698265,
      "grad_norm": 2.993116617202759,
      "learning_rate": 8.916438603103849e-07,
      "loss": 0.7672,
      "step": 9825
    },
    {
      "epoch": 0.15303540987179587,
      "grad_norm": 3.1958775520324707,
      "learning_rate": 8.915619213057799e-07,
      "loss": 0.8598,
      "step": 9830
    },
    {
      "epoch": 0.15311325087376526,
      "grad_norm": 6.617530345916748,
      "learning_rate": 8.914799823011749e-07,
      "loss": 0.6993,
      "step": 9835
    },
    {
      "epoch": 0.15319109187573463,
      "grad_norm": 3.9187653064727783,
      "learning_rate": 8.9139804329657e-07,
      "loss": 0.7635,
      "step": 9840
    },
    {
      "epoch": 0.153268932877704,
      "grad_norm": 3.482074499130249,
      "learning_rate": 8.91316104291965e-07,
      "loss": 0.7737,
      "step": 9845
    },
    {
      "epoch": 0.15334677387967338,
      "grad_norm": 9.398933410644531,
      "learning_rate": 8.9123416528736e-07,
      "loss": 0.8737,
      "step": 9850
    },
    {
      "epoch": 0.15342461488164275,
      "grad_norm": 3.637568473815918,
      "learning_rate": 8.911522262827551e-07,
      "loss": 0.7667,
      "step": 9855
    },
    {
      "epoch": 0.15350245588361214,
      "grad_norm": 3.3821427822113037,
      "learning_rate": 8.910702872781502e-07,
      "loss": 0.7836,
      "step": 9860
    },
    {
      "epoch": 0.1535802968855815,
      "grad_norm": 2.9914300441741943,
      "learning_rate": 8.909883482735452e-07,
      "loss": 0.8564,
      "step": 9865
    },
    {
      "epoch": 0.1536581378875509,
      "grad_norm": 6.171716690063477,
      "learning_rate": 8.909064092689401e-07,
      "loss": 0.8064,
      "step": 9870
    },
    {
      "epoch": 0.15373597888952026,
      "grad_norm": 3.4968278408050537,
      "learning_rate": 8.908244702643352e-07,
      "loss": 0.7524,
      "step": 9875
    },
    {
      "epoch": 0.15381381989148965,
      "grad_norm": 5.08829402923584,
      "learning_rate": 8.907425312597302e-07,
      "loss": 0.8207,
      "step": 9880
    },
    {
      "epoch": 0.15389166089345901,
      "grad_norm": 3.5643150806427,
      "learning_rate": 8.906605922551253e-07,
      "loss": 0.7776,
      "step": 9885
    },
    {
      "epoch": 0.1539695018954284,
      "grad_norm": 3.4858274459838867,
      "learning_rate": 8.905786532505203e-07,
      "loss": 0.8979,
      "step": 9890
    },
    {
      "epoch": 0.15404734289739777,
      "grad_norm": 3.9792239665985107,
      "learning_rate": 8.904967142459153e-07,
      "loss": 0.797,
      "step": 9895
    },
    {
      "epoch": 0.15412518389936716,
      "grad_norm": 9.95738410949707,
      "learning_rate": 8.904147752413104e-07,
      "loss": 0.7492,
      "step": 9900
    },
    {
      "epoch": 0.15420302490133653,
      "grad_norm": 4.636844158172607,
      "learning_rate": 8.903328362367054e-07,
      "loss": 0.7867,
      "step": 9905
    },
    {
      "epoch": 0.15428086590330592,
      "grad_norm": 7.289409637451172,
      "learning_rate": 8.902508972321003e-07,
      "loss": 0.7982,
      "step": 9910
    },
    {
      "epoch": 0.15435870690527528,
      "grad_norm": 3.7324790954589844,
      "learning_rate": 8.901689582274954e-07,
      "loss": 0.7001,
      "step": 9915
    },
    {
      "epoch": 0.15443654790724468,
      "grad_norm": 4.735339641571045,
      "learning_rate": 8.900870192228905e-07,
      "loss": 0.7155,
      "step": 9920
    },
    {
      "epoch": 0.15451438890921404,
      "grad_norm": 8.172463417053223,
      "learning_rate": 8.900050802182854e-07,
      "loss": 0.8489,
      "step": 9925
    },
    {
      "epoch": 0.1545922299111834,
      "grad_norm": 3.3313515186309814,
      "learning_rate": 8.899231412136805e-07,
      "loss": 0.8064,
      "step": 9930
    },
    {
      "epoch": 0.1546700709131528,
      "grad_norm": 9.073691368103027,
      "learning_rate": 8.898412022090756e-07,
      "loss": 0.8577,
      "step": 9935
    },
    {
      "epoch": 0.15474791191512216,
      "grad_norm": 7.126258373260498,
      "learning_rate": 8.897592632044705e-07,
      "loss": 0.7679,
      "step": 9940
    },
    {
      "epoch": 0.15482575291709155,
      "grad_norm": 11.254960060119629,
      "learning_rate": 8.896773241998656e-07,
      "loss": 0.7725,
      "step": 9945
    },
    {
      "epoch": 0.15490359391906092,
      "grad_norm": 3.706859827041626,
      "learning_rate": 8.895953851952606e-07,
      "loss": 0.9131,
      "step": 9950
    },
    {
      "epoch": 0.1549814349210303,
      "grad_norm": 3.0554311275482178,
      "learning_rate": 8.895134461906556e-07,
      "loss": 0.7683,
      "step": 9955
    },
    {
      "epoch": 0.15505927592299967,
      "grad_norm": 3.64799165725708,
      "learning_rate": 8.894315071860506e-07,
      "loss": 0.7275,
      "step": 9960
    },
    {
      "epoch": 0.15513711692496907,
      "grad_norm": 3.383768320083618,
      "learning_rate": 8.893495681814457e-07,
      "loss": 0.8155,
      "step": 9965
    },
    {
      "epoch": 0.15521495792693843,
      "grad_norm": 3.561455726623535,
      "learning_rate": 8.892676291768407e-07,
      "loss": 0.7256,
      "step": 9970
    },
    {
      "epoch": 0.15529279892890782,
      "grad_norm": 3.7967913150787354,
      "learning_rate": 8.891856901722358e-07,
      "loss": 0.8299,
      "step": 9975
    },
    {
      "epoch": 0.15537063993087719,
      "grad_norm": 3.6479313373565674,
      "learning_rate": 8.891037511676308e-07,
      "loss": 0.7851,
      "step": 9980
    },
    {
      "epoch": 0.15544848093284658,
      "grad_norm": 4.59113883972168,
      "learning_rate": 8.890218121630259e-07,
      "loss": 0.8115,
      "step": 9985
    },
    {
      "epoch": 0.15552632193481594,
      "grad_norm": 5.832945823669434,
      "learning_rate": 8.889398731584208e-07,
      "loss": 0.9049,
      "step": 9990
    },
    {
      "epoch": 0.15560416293678533,
      "grad_norm": 3.0644237995147705,
      "learning_rate": 8.888579341538158e-07,
      "loss": 0.6787,
      "step": 9995
    },
    {
      "epoch": 0.1556820039387547,
      "grad_norm": 4.14565372467041,
      "learning_rate": 8.887759951492109e-07,
      "loss": 0.8273,
      "step": 10000
    },
    {
      "epoch": 0.1557598449407241,
      "grad_norm": 10.58462142944336,
      "learning_rate": 8.886940561446059e-07,
      "loss": 0.7526,
      "step": 10005
    },
    {
      "epoch": 0.15583768594269345,
      "grad_norm": 4.249096870422363,
      "learning_rate": 8.88612117140001e-07,
      "loss": 0.7715,
      "step": 10010
    },
    {
      "epoch": 0.15591552694466282,
      "grad_norm": 6.341519355773926,
      "learning_rate": 8.88530178135396e-07,
      "loss": 0.7869,
      "step": 10015
    },
    {
      "epoch": 0.1559933679466322,
      "grad_norm": 2.831510543823242,
      "learning_rate": 8.88448239130791e-07,
      "loss": 0.807,
      "step": 10020
    },
    {
      "epoch": 0.15607120894860158,
      "grad_norm": 3.297983169555664,
      "learning_rate": 8.883663001261861e-07,
      "loss": 0.7433,
      "step": 10025
    },
    {
      "epoch": 0.15614904995057097,
      "grad_norm": 3.2775771617889404,
      "learning_rate": 8.88284361121581e-07,
      "loss": 0.7378,
      "step": 10030
    },
    {
      "epoch": 0.15622689095254033,
      "grad_norm": 4.8905768394470215,
      "learning_rate": 8.88202422116976e-07,
      "loss": 0.9689,
      "step": 10035
    },
    {
      "epoch": 0.15630473195450972,
      "grad_norm": 5.260064601898193,
      "learning_rate": 8.881204831123711e-07,
      "loss": 0.8044,
      "step": 10040
    },
    {
      "epoch": 0.1563825729564791,
      "grad_norm": 3.0201990604400635,
      "learning_rate": 8.880385441077662e-07,
      "loss": 0.7943,
      "step": 10045
    },
    {
      "epoch": 0.15646041395844848,
      "grad_norm": 3.7755088806152344,
      "learning_rate": 8.879566051031611e-07,
      "loss": 0.7369,
      "step": 10050
    },
    {
      "epoch": 0.15653825496041784,
      "grad_norm": 4.074087619781494,
      "learning_rate": 8.878746660985562e-07,
      "loss": 0.8229,
      "step": 10055
    },
    {
      "epoch": 0.15661609596238724,
      "grad_norm": 4.723263263702393,
      "learning_rate": 8.877927270939513e-07,
      "loss": 0.6489,
      "step": 10060
    },
    {
      "epoch": 0.1566939369643566,
      "grad_norm": 4.477192401885986,
      "learning_rate": 8.877107880893463e-07,
      "loss": 0.8427,
      "step": 10065
    },
    {
      "epoch": 0.156771777966326,
      "grad_norm": 2.5688400268554688,
      "learning_rate": 8.876288490847412e-07,
      "loss": 0.8412,
      "step": 10070
    },
    {
      "epoch": 0.15684961896829536,
      "grad_norm": 4.53289794921875,
      "learning_rate": 8.875469100801363e-07,
      "loss": 0.8295,
      "step": 10075
    },
    {
      "epoch": 0.15692745997026475,
      "grad_norm": 7.105226993560791,
      "learning_rate": 8.874649710755313e-07,
      "loss": 0.8648,
      "step": 10080
    },
    {
      "epoch": 0.1570053009722341,
      "grad_norm": 2.8844571113586426,
      "learning_rate": 8.873830320709263e-07,
      "loss": 0.7657,
      "step": 10085
    },
    {
      "epoch": 0.1570831419742035,
      "grad_norm": 4.569636344909668,
      "learning_rate": 8.873010930663214e-07,
      "loss": 0.6868,
      "step": 10090
    },
    {
      "epoch": 0.15716098297617287,
      "grad_norm": 3.5974533557891846,
      "learning_rate": 8.872191540617164e-07,
      "loss": 0.9193,
      "step": 10095
    },
    {
      "epoch": 0.15723882397814223,
      "grad_norm": 4.02683687210083,
      "learning_rate": 8.871372150571115e-07,
      "loss": 0.8309,
      "step": 10100
    },
    {
      "epoch": 0.15731666498011163,
      "grad_norm": 3.6681370735168457,
      "learning_rate": 8.870552760525065e-07,
      "loss": 0.7995,
      "step": 10105
    },
    {
      "epoch": 0.157394505982081,
      "grad_norm": 3.227896213531494,
      "learning_rate": 8.869733370479014e-07,
      "loss": 0.7712,
      "step": 10110
    },
    {
      "epoch": 0.15747234698405038,
      "grad_norm": 7.877114295959473,
      "learning_rate": 8.868913980432965e-07,
      "loss": 0.7099,
      "step": 10115
    },
    {
      "epoch": 0.15755018798601975,
      "grad_norm": 4.573225021362305,
      "learning_rate": 8.868094590386915e-07,
      "loss": 0.6932,
      "step": 10120
    },
    {
      "epoch": 0.15762802898798914,
      "grad_norm": 4.980703353881836,
      "learning_rate": 8.867275200340866e-07,
      "loss": 0.9058,
      "step": 10125
    },
    {
      "epoch": 0.1577058699899585,
      "grad_norm": 4.040239334106445,
      "learning_rate": 8.866455810294816e-07,
      "loss": 0.813,
      "step": 10130
    },
    {
      "epoch": 0.1577837109919279,
      "grad_norm": 3.489124059677124,
      "learning_rate": 8.865636420248767e-07,
      "loss": 0.7237,
      "step": 10135
    },
    {
      "epoch": 0.15786155199389726,
      "grad_norm": 2.5940945148468018,
      "learning_rate": 8.864817030202717e-07,
      "loss": 0.7494,
      "step": 10140
    },
    {
      "epoch": 0.15793939299586665,
      "grad_norm": 3.312206745147705,
      "learning_rate": 8.863997640156667e-07,
      "loss": 0.8447,
      "step": 10145
    },
    {
      "epoch": 0.15801723399783602,
      "grad_norm": 3.279050350189209,
      "learning_rate": 8.863178250110617e-07,
      "loss": 0.8192,
      "step": 10150
    },
    {
      "epoch": 0.1580950749998054,
      "grad_norm": 3.221031904220581,
      "learning_rate": 8.862358860064568e-07,
      "loss": 0.8593,
      "step": 10155
    },
    {
      "epoch": 0.15817291600177477,
      "grad_norm": 5.654365062713623,
      "learning_rate": 8.861539470018517e-07,
      "loss": 0.8396,
      "step": 10160
    },
    {
      "epoch": 0.15825075700374416,
      "grad_norm": 3.0688130855560303,
      "learning_rate": 8.860720079972468e-07,
      "loss": 0.6908,
      "step": 10165
    },
    {
      "epoch": 0.15832859800571353,
      "grad_norm": 4.515695571899414,
      "learning_rate": 8.859900689926419e-07,
      "loss": 0.8116,
      "step": 10170
    },
    {
      "epoch": 0.15840643900768292,
      "grad_norm": 5.167048454284668,
      "learning_rate": 8.859081299880368e-07,
      "loss": 0.8373,
      "step": 10175
    },
    {
      "epoch": 0.15848428000965228,
      "grad_norm": 8.9721097946167,
      "learning_rate": 8.858261909834319e-07,
      "loss": 0.9478,
      "step": 10180
    },
    {
      "epoch": 0.15856212101162165,
      "grad_norm": 3.8844072818756104,
      "learning_rate": 8.85744251978827e-07,
      "loss": 0.8713,
      "step": 10185
    },
    {
      "epoch": 0.15863996201359104,
      "grad_norm": 3.705676794052124,
      "learning_rate": 8.85662312974222e-07,
      "loss": 0.817,
      "step": 10190
    },
    {
      "epoch": 0.1587178030155604,
      "grad_norm": 4.741096496582031,
      "learning_rate": 8.855803739696169e-07,
      "loss": 0.8058,
      "step": 10195
    },
    {
      "epoch": 0.1587956440175298,
      "grad_norm": 6.481576442718506,
      "learning_rate": 8.85498434965012e-07,
      "loss": 0.9077,
      "step": 10200
    },
    {
      "epoch": 0.15887348501949916,
      "grad_norm": 5.847426414489746,
      "learning_rate": 8.85416495960407e-07,
      "loss": 0.7257,
      "step": 10205
    },
    {
      "epoch": 0.15895132602146855,
      "grad_norm": 3.6452407836914062,
      "learning_rate": 8.85334556955802e-07,
      "loss": 0.7282,
      "step": 10210
    },
    {
      "epoch": 0.15902916702343792,
      "grad_norm": 3.5716402530670166,
      "learning_rate": 8.852526179511971e-07,
      "loss": 0.7504,
      "step": 10215
    },
    {
      "epoch": 0.1591070080254073,
      "grad_norm": 4.702507495880127,
      "learning_rate": 8.851706789465921e-07,
      "loss": 0.8345,
      "step": 10220
    },
    {
      "epoch": 0.15918484902737667,
      "grad_norm": 3.6276495456695557,
      "learning_rate": 8.850887399419872e-07,
      "loss": 0.8267,
      "step": 10225
    },
    {
      "epoch": 0.15926269002934607,
      "grad_norm": 3.432413101196289,
      "learning_rate": 8.850068009373822e-07,
      "loss": 0.9476,
      "step": 10230
    },
    {
      "epoch": 0.15934053103131543,
      "grad_norm": 4.081861972808838,
      "learning_rate": 8.849248619327771e-07,
      "loss": 0.7688,
      "step": 10235
    },
    {
      "epoch": 0.15941837203328482,
      "grad_norm": 6.799472332000732,
      "learning_rate": 8.848429229281722e-07,
      "loss": 0.7084,
      "step": 10240
    },
    {
      "epoch": 0.1594962130352542,
      "grad_norm": 2.736294984817505,
      "learning_rate": 8.847609839235673e-07,
      "loss": 0.7835,
      "step": 10245
    },
    {
      "epoch": 0.15957405403722358,
      "grad_norm": 4.227167129516602,
      "learning_rate": 8.846790449189623e-07,
      "loss": 0.7833,
      "step": 10250
    },
    {
      "epoch": 0.15965189503919294,
      "grad_norm": 3.3678765296936035,
      "learning_rate": 8.845971059143573e-07,
      "loss": 0.9005,
      "step": 10255
    },
    {
      "epoch": 0.15972973604116233,
      "grad_norm": 3.6415300369262695,
      "learning_rate": 8.845151669097524e-07,
      "loss": 0.8214,
      "step": 10260
    },
    {
      "epoch": 0.1598075770431317,
      "grad_norm": 4.28493595123291,
      "learning_rate": 8.844332279051474e-07,
      "loss": 0.7597,
      "step": 10265
    },
    {
      "epoch": 0.15988541804510106,
      "grad_norm": 3.2147064208984375,
      "learning_rate": 8.843512889005424e-07,
      "loss": 0.7652,
      "step": 10270
    },
    {
      "epoch": 0.15996325904707046,
      "grad_norm": 7.0327959060668945,
      "learning_rate": 8.842693498959374e-07,
      "loss": 0.7703,
      "step": 10275
    },
    {
      "epoch": 0.16004110004903982,
      "grad_norm": 3.714036464691162,
      "learning_rate": 8.841874108913325e-07,
      "loss": 0.8864,
      "step": 10280
    },
    {
      "epoch": 0.1601189410510092,
      "grad_norm": 3.1920130252838135,
      "learning_rate": 8.841054718867274e-07,
      "loss": 0.8203,
      "step": 10285
    },
    {
      "epoch": 0.16019678205297858,
      "grad_norm": 4.356620788574219,
      "learning_rate": 8.840235328821225e-07,
      "loss": 0.8215,
      "step": 10290
    },
    {
      "epoch": 0.16027462305494797,
      "grad_norm": 3.031799793243408,
      "learning_rate": 8.839415938775176e-07,
      "loss": 0.6909,
      "step": 10295
    },
    {
      "epoch": 0.16035246405691733,
      "grad_norm": 3.817898750305176,
      "learning_rate": 8.838596548729125e-07,
      "loss": 0.7237,
      "step": 10300
    },
    {
      "epoch": 0.16043030505888672,
      "grad_norm": 4.679711818695068,
      "learning_rate": 8.837777158683076e-07,
      "loss": 0.7296,
      "step": 10305
    },
    {
      "epoch": 0.1605081460608561,
      "grad_norm": 3.7384095191955566,
      "learning_rate": 8.836957768637027e-07,
      "loss": 0.9011,
      "step": 10310
    },
    {
      "epoch": 0.16058598706282548,
      "grad_norm": 4.7621541023254395,
      "learning_rate": 8.836138378590976e-07,
      "loss": 0.9026,
      "step": 10315
    },
    {
      "epoch": 0.16066382806479484,
      "grad_norm": 6.402016639709473,
      "learning_rate": 8.835318988544926e-07,
      "loss": 0.8081,
      "step": 10320
    },
    {
      "epoch": 0.16074166906676424,
      "grad_norm": 4.472290515899658,
      "learning_rate": 8.834499598498877e-07,
      "loss": 0.7606,
      "step": 10325
    },
    {
      "epoch": 0.1608195100687336,
      "grad_norm": 5.197412967681885,
      "learning_rate": 8.833680208452827e-07,
      "loss": 0.7824,
      "step": 10330
    },
    {
      "epoch": 0.160897351070703,
      "grad_norm": 3.40364408493042,
      "learning_rate": 8.832860818406778e-07,
      "loss": 0.7884,
      "step": 10335
    },
    {
      "epoch": 0.16097519207267236,
      "grad_norm": 6.786139965057373,
      "learning_rate": 8.832041428360728e-07,
      "loss": 0.8743,
      "step": 10340
    },
    {
      "epoch": 0.16105303307464175,
      "grad_norm": 4.048831462860107,
      "learning_rate": 8.831222038314678e-07,
      "loss": 0.8958,
      "step": 10345
    },
    {
      "epoch": 0.1611308740766111,
      "grad_norm": 4.102180480957031,
      "learning_rate": 8.830402648268629e-07,
      "loss": 0.7835,
      "step": 10350
    },
    {
      "epoch": 0.16120871507858048,
      "grad_norm": 3.135587453842163,
      "learning_rate": 8.829583258222578e-07,
      "loss": 0.6362,
      "step": 10355
    },
    {
      "epoch": 0.16128655608054987,
      "grad_norm": 3.9217071533203125,
      "learning_rate": 8.828763868176528e-07,
      "loss": 0.7069,
      "step": 10360
    },
    {
      "epoch": 0.16136439708251923,
      "grad_norm": 3.276561737060547,
      "learning_rate": 8.827944478130479e-07,
      "loss": 0.8588,
      "step": 10365
    },
    {
      "epoch": 0.16144223808448863,
      "grad_norm": 3.9652113914489746,
      "learning_rate": 8.82712508808443e-07,
      "loss": 0.8276,
      "step": 10370
    },
    {
      "epoch": 0.161520079086458,
      "grad_norm": 5.678191184997559,
      "learning_rate": 8.82630569803838e-07,
      "loss": 0.7526,
      "step": 10375
    },
    {
      "epoch": 0.16159792008842738,
      "grad_norm": 5.475320816040039,
      "learning_rate": 8.82548630799233e-07,
      "loss": 0.8913,
      "step": 10380
    },
    {
      "epoch": 0.16167576109039675,
      "grad_norm": 15.008077621459961,
      "learning_rate": 8.824666917946281e-07,
      "loss": 0.7912,
      "step": 10385
    },
    {
      "epoch": 0.16175360209236614,
      "grad_norm": 3.8481225967407227,
      "learning_rate": 8.823847527900232e-07,
      "loss": 0.7949,
      "step": 10390
    },
    {
      "epoch": 0.1618314430943355,
      "grad_norm": 2.759436845779419,
      "learning_rate": 8.82302813785418e-07,
      "loss": 0.7449,
      "step": 10395
    },
    {
      "epoch": 0.1619092840963049,
      "grad_norm": 8.001230239868164,
      "learning_rate": 8.822208747808131e-07,
      "loss": 0.853,
      "step": 10400
    },
    {
      "epoch": 0.16198712509827426,
      "grad_norm": 5.410604953765869,
      "learning_rate": 8.821389357762082e-07,
      "loss": 0.8032,
      "step": 10405
    },
    {
      "epoch": 0.16206496610024365,
      "grad_norm": 17.351768493652344,
      "learning_rate": 8.820569967716031e-07,
      "loss": 0.7418,
      "step": 10410
    },
    {
      "epoch": 0.16214280710221302,
      "grad_norm": 3.534604549407959,
      "learning_rate": 8.819750577669982e-07,
      "loss": 0.7653,
      "step": 10415
    },
    {
      "epoch": 0.1622206481041824,
      "grad_norm": 8.916220664978027,
      "learning_rate": 8.818931187623933e-07,
      "loss": 0.837,
      "step": 10420
    },
    {
      "epoch": 0.16229848910615177,
      "grad_norm": 4.538114547729492,
      "learning_rate": 8.818111797577883e-07,
      "loss": 0.7677,
      "step": 10425
    },
    {
      "epoch": 0.16237633010812116,
      "grad_norm": 3.6164751052856445,
      "learning_rate": 8.817292407531833e-07,
      "loss": 0.7675,
      "step": 10430
    },
    {
      "epoch": 0.16245417111009053,
      "grad_norm": 4.763157844543457,
      "learning_rate": 8.816473017485783e-07,
      "loss": 0.7237,
      "step": 10435
    },
    {
      "epoch": 0.1625320121120599,
      "grad_norm": 4.728906631469727,
      "learning_rate": 8.815653627439733e-07,
      "loss": 0.8593,
      "step": 10440
    },
    {
      "epoch": 0.16260985311402928,
      "grad_norm": 3.878848075866699,
      "learning_rate": 8.814834237393683e-07,
      "loss": 0.8701,
      "step": 10445
    },
    {
      "epoch": 0.16268769411599865,
      "grad_norm": 5.231772422790527,
      "learning_rate": 8.814014847347634e-07,
      "loss": 0.8108,
      "step": 10450
    },
    {
      "epoch": 0.16276553511796804,
      "grad_norm": 9.139507293701172,
      "learning_rate": 8.813195457301584e-07,
      "loss": 0.7747,
      "step": 10455
    },
    {
      "epoch": 0.1628433761199374,
      "grad_norm": 4.208261013031006,
      "learning_rate": 8.812376067255535e-07,
      "loss": 0.7136,
      "step": 10460
    },
    {
      "epoch": 0.1629212171219068,
      "grad_norm": 4.353900909423828,
      "learning_rate": 8.811556677209485e-07,
      "loss": 0.7753,
      "step": 10465
    },
    {
      "epoch": 0.16299905812387616,
      "grad_norm": 2.729976177215576,
      "learning_rate": 8.810737287163435e-07,
      "loss": 0.7926,
      "step": 10470
    },
    {
      "epoch": 0.16307689912584555,
      "grad_norm": 2.928906202316284,
      "learning_rate": 8.809917897117385e-07,
      "loss": 0.7747,
      "step": 10475
    },
    {
      "epoch": 0.16315474012781492,
      "grad_norm": 4.206002712249756,
      "learning_rate": 8.809098507071335e-07,
      "loss": 0.8576,
      "step": 10480
    },
    {
      "epoch": 0.1632325811297843,
      "grad_norm": 7.724112510681152,
      "learning_rate": 8.808279117025286e-07,
      "loss": 0.8393,
      "step": 10485
    },
    {
      "epoch": 0.16331042213175367,
      "grad_norm": 3.295132637023926,
      "learning_rate": 8.807459726979236e-07,
      "loss": 0.778,
      "step": 10490
    },
    {
      "epoch": 0.16338826313372307,
      "grad_norm": 3.3869988918304443,
      "learning_rate": 8.806640336933187e-07,
      "loss": 0.7569,
      "step": 10495
    },
    {
      "epoch": 0.16346610413569243,
      "grad_norm": 3.718599319458008,
      "learning_rate": 8.805820946887137e-07,
      "loss": 0.8747,
      "step": 10500
    },
    {
      "epoch": 0.16354394513766182,
      "grad_norm": 3.8476598262786865,
      "learning_rate": 8.805001556841087e-07,
      "loss": 0.6924,
      "step": 10505
    },
    {
      "epoch": 0.1636217861396312,
      "grad_norm": 3.9861888885498047,
      "learning_rate": 8.804182166795038e-07,
      "loss": 0.8145,
      "step": 10510
    },
    {
      "epoch": 0.16369962714160058,
      "grad_norm": 3.1306610107421875,
      "learning_rate": 8.803362776748989e-07,
      "loss": 0.6579,
      "step": 10515
    },
    {
      "epoch": 0.16377746814356994,
      "grad_norm": 4.904189586639404,
      "learning_rate": 8.802543386702937e-07,
      "loss": 0.8639,
      "step": 10520
    },
    {
      "epoch": 0.16385530914553934,
      "grad_norm": 5.17719841003418,
      "learning_rate": 8.801723996656888e-07,
      "loss": 0.8722,
      "step": 10525
    },
    {
      "epoch": 0.1639331501475087,
      "grad_norm": 2.9521892070770264,
      "learning_rate": 8.800904606610839e-07,
      "loss": 0.7665,
      "step": 10530
    },
    {
      "epoch": 0.16401099114947806,
      "grad_norm": 3.255908966064453,
      "learning_rate": 8.800085216564788e-07,
      "loss": 0.7702,
      "step": 10535
    },
    {
      "epoch": 0.16408883215144746,
      "grad_norm": 6.015521049499512,
      "learning_rate": 8.799265826518739e-07,
      "loss": 0.7845,
      "step": 10540
    },
    {
      "epoch": 0.16416667315341682,
      "grad_norm": 4.091128826141357,
      "learning_rate": 8.79844643647269e-07,
      "loss": 0.9369,
      "step": 10545
    },
    {
      "epoch": 0.1642445141553862,
      "grad_norm": 4.375763416290283,
      "learning_rate": 8.79762704642664e-07,
      "loss": 0.8318,
      "step": 10550
    },
    {
      "epoch": 0.16432235515735558,
      "grad_norm": 6.054563522338867,
      "learning_rate": 8.79680765638059e-07,
      "loss": 0.8422,
      "step": 10555
    },
    {
      "epoch": 0.16440019615932497,
      "grad_norm": 4.5666303634643555,
      "learning_rate": 8.79598826633454e-07,
      "loss": 0.8438,
      "step": 10560
    },
    {
      "epoch": 0.16447803716129433,
      "grad_norm": 3.462005615234375,
      "learning_rate": 8.79516887628849e-07,
      "loss": 0.7772,
      "step": 10565
    },
    {
      "epoch": 0.16455587816326372,
      "grad_norm": 3.66213321685791,
      "learning_rate": 8.79434948624244e-07,
      "loss": 0.9158,
      "step": 10570
    },
    {
      "epoch": 0.1646337191652331,
      "grad_norm": 3.454305648803711,
      "learning_rate": 8.793530096196391e-07,
      "loss": 0.7452,
      "step": 10575
    },
    {
      "epoch": 0.16471156016720248,
      "grad_norm": 4.568371295928955,
      "learning_rate": 8.792710706150341e-07,
      "loss": 0.855,
      "step": 10580
    },
    {
      "epoch": 0.16478940116917185,
      "grad_norm": 5.541522979736328,
      "learning_rate": 8.791891316104292e-07,
      "loss": 0.9156,
      "step": 10585
    },
    {
      "epoch": 0.16486724217114124,
      "grad_norm": 4.69566011428833,
      "learning_rate": 8.791071926058242e-07,
      "loss": 0.7321,
      "step": 10590
    },
    {
      "epoch": 0.1649450831731106,
      "grad_norm": 7.592962741851807,
      "learning_rate": 8.790252536012192e-07,
      "loss": 0.7888,
      "step": 10595
    },
    {
      "epoch": 0.16502292417508,
      "grad_norm": 5.3040080070495605,
      "learning_rate": 8.789433145966142e-07,
      "loss": 0.7643,
      "step": 10600
    },
    {
      "epoch": 0.16510076517704936,
      "grad_norm": 2.616908073425293,
      "learning_rate": 8.788613755920093e-07,
      "loss": 0.7614,
      "step": 10605
    },
    {
      "epoch": 0.16517860617901875,
      "grad_norm": 3.683166980743408,
      "learning_rate": 8.787794365874043e-07,
      "loss": 0.715,
      "step": 10610
    },
    {
      "epoch": 0.16525644718098811,
      "grad_norm": 4.2956132888793945,
      "learning_rate": 8.786974975827993e-07,
      "loss": 0.9153,
      "step": 10615
    },
    {
      "epoch": 0.16533428818295748,
      "grad_norm": 3.76724910736084,
      "learning_rate": 8.786155585781944e-07,
      "loss": 0.86,
      "step": 10620
    },
    {
      "epoch": 0.16541212918492687,
      "grad_norm": 3.4380204677581787,
      "learning_rate": 8.785336195735894e-07,
      "loss": 0.8662,
      "step": 10625
    },
    {
      "epoch": 0.16548997018689623,
      "grad_norm": 8.281620025634766,
      "learning_rate": 8.784516805689844e-07,
      "loss": 0.8607,
      "step": 10630
    },
    {
      "epoch": 0.16556781118886563,
      "grad_norm": 7.735804557800293,
      "learning_rate": 8.783697415643795e-07,
      "loss": 0.7513,
      "step": 10635
    },
    {
      "epoch": 0.165645652190835,
      "grad_norm": 6.491322040557861,
      "learning_rate": 8.782878025597745e-07,
      "loss": 0.824,
      "step": 10640
    },
    {
      "epoch": 0.16572349319280438,
      "grad_norm": 2.7615108489990234,
      "learning_rate": 8.782058635551694e-07,
      "loss": 0.9231,
      "step": 10645
    },
    {
      "epoch": 0.16580133419477375,
      "grad_norm": 5.219547748565674,
      "learning_rate": 8.781239245505645e-07,
      "loss": 0.7178,
      "step": 10650
    },
    {
      "epoch": 0.16587917519674314,
      "grad_norm": 8.346015930175781,
      "learning_rate": 8.780419855459596e-07,
      "loss": 0.8779,
      "step": 10655
    },
    {
      "epoch": 0.1659570161987125,
      "grad_norm": 9.989750862121582,
      "learning_rate": 8.779600465413545e-07,
      "loss": 0.7609,
      "step": 10660
    },
    {
      "epoch": 0.1660348572006819,
      "grad_norm": 6.761223793029785,
      "learning_rate": 8.778781075367496e-07,
      "loss": 0.9394,
      "step": 10665
    },
    {
      "epoch": 0.16611269820265126,
      "grad_norm": 3.055026054382324,
      "learning_rate": 8.777961685321447e-07,
      "loss": 0.8548,
      "step": 10670
    },
    {
      "epoch": 0.16619053920462065,
      "grad_norm": 7.2378950119018555,
      "learning_rate": 8.777142295275397e-07,
      "loss": 0.8538,
      "step": 10675
    },
    {
      "epoch": 0.16626838020659002,
      "grad_norm": 11.686674118041992,
      "learning_rate": 8.776322905229346e-07,
      "loss": 0.7245,
      "step": 10680
    },
    {
      "epoch": 0.1663462212085594,
      "grad_norm": 3.3093395233154297,
      "learning_rate": 8.775503515183297e-07,
      "loss": 0.7876,
      "step": 10685
    },
    {
      "epoch": 0.16642406221052877,
      "grad_norm": 5.579384803771973,
      "learning_rate": 8.774684125137247e-07,
      "loss": 0.8294,
      "step": 10690
    },
    {
      "epoch": 0.16650190321249816,
      "grad_norm": 3.092195749282837,
      "learning_rate": 8.773864735091198e-07,
      "loss": 0.8116,
      "step": 10695
    },
    {
      "epoch": 0.16657974421446753,
      "grad_norm": 5.8653130531311035,
      "learning_rate": 8.773045345045148e-07,
      "loss": 0.7827,
      "step": 10700
    },
    {
      "epoch": 0.1666575852164369,
      "grad_norm": 8.496586799621582,
      "learning_rate": 8.772225954999098e-07,
      "loss": 0.871,
      "step": 10705
    },
    {
      "epoch": 0.16673542621840629,
      "grad_norm": 3.554830312728882,
      "learning_rate": 8.771406564953049e-07,
      "loss": 0.7876,
      "step": 10710
    },
    {
      "epoch": 0.16681326722037565,
      "grad_norm": 4.988160610198975,
      "learning_rate": 8.770587174907e-07,
      "loss": 0.8173,
      "step": 10715
    },
    {
      "epoch": 0.16689110822234504,
      "grad_norm": 4.899982929229736,
      "learning_rate": 8.769767784860948e-07,
      "loss": 0.7582,
      "step": 10720
    },
    {
      "epoch": 0.1669689492243144,
      "grad_norm": 3.1257996559143066,
      "learning_rate": 8.768948394814899e-07,
      "loss": 0.7396,
      "step": 10725
    },
    {
      "epoch": 0.1670467902262838,
      "grad_norm": 7.962161064147949,
      "learning_rate": 8.76812900476885e-07,
      "loss": 0.8768,
      "step": 10730
    },
    {
      "epoch": 0.16712463122825316,
      "grad_norm": 9.329161643981934,
      "learning_rate": 8.7673096147228e-07,
      "loss": 0.7469,
      "step": 10735
    },
    {
      "epoch": 0.16720247223022255,
      "grad_norm": 2.5543742179870605,
      "learning_rate": 8.76649022467675e-07,
      "loss": 0.8198,
      "step": 10740
    },
    {
      "epoch": 0.16728031323219192,
      "grad_norm": 3.192782402038574,
      "learning_rate": 8.765670834630701e-07,
      "loss": 0.7747,
      "step": 10745
    },
    {
      "epoch": 0.1673581542341613,
      "grad_norm": 3.3908746242523193,
      "learning_rate": 8.764851444584652e-07,
      "loss": 0.8149,
      "step": 10750
    },
    {
      "epoch": 0.16743599523613067,
      "grad_norm": 3.8879611492156982,
      "learning_rate": 8.764032054538601e-07,
      "loss": 0.774,
      "step": 10755
    },
    {
      "epoch": 0.16751383623810007,
      "grad_norm": 5.076687335968018,
      "learning_rate": 8.763212664492551e-07,
      "loss": 0.799,
      "step": 10760
    },
    {
      "epoch": 0.16759167724006943,
      "grad_norm": 5.339931964874268,
      "learning_rate": 8.762393274446502e-07,
      "loss": 0.756,
      "step": 10765
    },
    {
      "epoch": 0.16766951824203882,
      "grad_norm": 4.076517581939697,
      "learning_rate": 8.761573884400451e-07,
      "loss": 0.8323,
      "step": 10770
    },
    {
      "epoch": 0.1677473592440082,
      "grad_norm": 4.2539777755737305,
      "learning_rate": 8.760754494354402e-07,
      "loss": 0.6913,
      "step": 10775
    },
    {
      "epoch": 0.16782520024597758,
      "grad_norm": 3.196856737136841,
      "learning_rate": 8.759935104308353e-07,
      "loss": 0.7324,
      "step": 10780
    },
    {
      "epoch": 0.16790304124794694,
      "grad_norm": 13.658886909484863,
      "learning_rate": 8.759115714262303e-07,
      "loss": 0.7811,
      "step": 10785
    },
    {
      "epoch": 0.1679808822499163,
      "grad_norm": 3.423370838165283,
      "learning_rate": 8.758296324216253e-07,
      "loss": 0.7902,
      "step": 10790
    },
    {
      "epoch": 0.1680587232518857,
      "grad_norm": 4.14307165145874,
      "learning_rate": 8.757476934170204e-07,
      "loss": 0.7751,
      "step": 10795
    },
    {
      "epoch": 0.16813656425385506,
      "grad_norm": 3.6597414016723633,
      "learning_rate": 8.756657544124153e-07,
      "loss": 0.7519,
      "step": 10800
    },
    {
      "epoch": 0.16821440525582446,
      "grad_norm": 3.839320182800293,
      "learning_rate": 8.755838154078103e-07,
      "loss": 0.6788,
      "step": 10805
    },
    {
      "epoch": 0.16829224625779382,
      "grad_norm": 5.377815246582031,
      "learning_rate": 8.755018764032054e-07,
      "loss": 0.8277,
      "step": 10810
    },
    {
      "epoch": 0.1683700872597632,
      "grad_norm": 2.5293519496917725,
      "learning_rate": 8.754199373986004e-07,
      "loss": 0.7754,
      "step": 10815
    },
    {
      "epoch": 0.16844792826173258,
      "grad_norm": 3.5276260375976562,
      "learning_rate": 8.753379983939955e-07,
      "loss": 0.6878,
      "step": 10820
    },
    {
      "epoch": 0.16852576926370197,
      "grad_norm": 4.435384273529053,
      "learning_rate": 8.752560593893905e-07,
      "loss": 0.7848,
      "step": 10825
    },
    {
      "epoch": 0.16860361026567133,
      "grad_norm": 4.338689804077148,
      "learning_rate": 8.751741203847855e-07,
      "loss": 0.7934,
      "step": 10830
    },
    {
      "epoch": 0.16868145126764073,
      "grad_norm": 5.802286148071289,
      "learning_rate": 8.750921813801806e-07,
      "loss": 0.6343,
      "step": 10835
    },
    {
      "epoch": 0.1687592922696101,
      "grad_norm": 5.789932727813721,
      "learning_rate": 8.750102423755757e-07,
      "loss": 0.8002,
      "step": 10840
    },
    {
      "epoch": 0.16883713327157948,
      "grad_norm": 3.8039424419403076,
      "learning_rate": 8.749283033709705e-07,
      "loss": 0.7334,
      "step": 10845
    },
    {
      "epoch": 0.16891497427354885,
      "grad_norm": 3.930701971054077,
      "learning_rate": 8.748463643663656e-07,
      "loss": 0.8077,
      "step": 10850
    },
    {
      "epoch": 0.16899281527551824,
      "grad_norm": 15.397842407226562,
      "learning_rate": 8.747644253617607e-07,
      "loss": 0.9553,
      "step": 10855
    },
    {
      "epoch": 0.1690706562774876,
      "grad_norm": 7.080071926116943,
      "learning_rate": 8.746824863571557e-07,
      "loss": 0.7472,
      "step": 10860
    },
    {
      "epoch": 0.169148497279457,
      "grad_norm": 3.849839687347412,
      "learning_rate": 8.746005473525507e-07,
      "loss": 0.7191,
      "step": 10865
    },
    {
      "epoch": 0.16922633828142636,
      "grad_norm": 3.95355486869812,
      "learning_rate": 8.745186083479458e-07,
      "loss": 0.661,
      "step": 10870
    },
    {
      "epoch": 0.16930417928339572,
      "grad_norm": 2.4966964721679688,
      "learning_rate": 8.744366693433409e-07,
      "loss": 0.8823,
      "step": 10875
    },
    {
      "epoch": 0.16938202028536511,
      "grad_norm": 6.468384742736816,
      "learning_rate": 8.743547303387358e-07,
      "loss": 0.7943,
      "step": 10880
    },
    {
      "epoch": 0.16945986128733448,
      "grad_norm": 4.023189544677734,
      "learning_rate": 8.742727913341308e-07,
      "loss": 0.7107,
      "step": 10885
    },
    {
      "epoch": 0.16953770228930387,
      "grad_norm": 3.180260419845581,
      "learning_rate": 8.741908523295259e-07,
      "loss": 0.7373,
      "step": 10890
    },
    {
      "epoch": 0.16961554329127324,
      "grad_norm": 4.347747802734375,
      "learning_rate": 8.741089133249208e-07,
      "loss": 0.7309,
      "step": 10895
    },
    {
      "epoch": 0.16969338429324263,
      "grad_norm": 6.062042713165283,
      "learning_rate": 8.740269743203159e-07,
      "loss": 0.7572,
      "step": 10900
    },
    {
      "epoch": 0.169771225295212,
      "grad_norm": 2.9714925289154053,
      "learning_rate": 8.73945035315711e-07,
      "loss": 0.7924,
      "step": 10905
    },
    {
      "epoch": 0.16984906629718138,
      "grad_norm": 3.827422618865967,
      "learning_rate": 8.73863096311106e-07,
      "loss": 0.912,
      "step": 10910
    },
    {
      "epoch": 0.16992690729915075,
      "grad_norm": 5.133169174194336,
      "learning_rate": 8.73781157306501e-07,
      "loss": 0.7078,
      "step": 10915
    },
    {
      "epoch": 0.17000474830112014,
      "grad_norm": 2.856977939605713,
      "learning_rate": 8.736992183018961e-07,
      "loss": 0.8076,
      "step": 10920
    },
    {
      "epoch": 0.1700825893030895,
      "grad_norm": 4.684539318084717,
      "learning_rate": 8.73617279297291e-07,
      "loss": 0.7045,
      "step": 10925
    },
    {
      "epoch": 0.1701604303050589,
      "grad_norm": 4.420496940612793,
      "learning_rate": 8.73535340292686e-07,
      "loss": 0.8248,
      "step": 10930
    },
    {
      "epoch": 0.17023827130702826,
      "grad_norm": 3.429518461227417,
      "learning_rate": 8.734534012880811e-07,
      "loss": 0.7643,
      "step": 10935
    },
    {
      "epoch": 0.17031611230899765,
      "grad_norm": 3.1888604164123535,
      "learning_rate": 8.733714622834761e-07,
      "loss": 0.7827,
      "step": 10940
    },
    {
      "epoch": 0.17039395331096702,
      "grad_norm": 3.491792917251587,
      "learning_rate": 8.732895232788712e-07,
      "loss": 0.8594,
      "step": 10945
    },
    {
      "epoch": 0.1704717943129364,
      "grad_norm": 4.644289493560791,
      "learning_rate": 8.732075842742662e-07,
      "loss": 0.7508,
      "step": 10950
    },
    {
      "epoch": 0.17054963531490577,
      "grad_norm": 3.222562313079834,
      "learning_rate": 8.731256452696612e-07,
      "loss": 0.7755,
      "step": 10955
    },
    {
      "epoch": 0.17062747631687514,
      "grad_norm": 7.415604591369629,
      "learning_rate": 8.730437062650563e-07,
      "loss": 0.677,
      "step": 10960
    },
    {
      "epoch": 0.17070531731884453,
      "grad_norm": 3.6360597610473633,
      "learning_rate": 8.729617672604513e-07,
      "loss": 0.8338,
      "step": 10965
    },
    {
      "epoch": 0.1707831583208139,
      "grad_norm": 9.364253044128418,
      "learning_rate": 8.728798282558462e-07,
      "loss": 0.7644,
      "step": 10970
    },
    {
      "epoch": 0.17086099932278329,
      "grad_norm": 3.551168441772461,
      "learning_rate": 8.727978892512413e-07,
      "loss": 0.7808,
      "step": 10975
    },
    {
      "epoch": 0.17093884032475265,
      "grad_norm": 6.72664213180542,
      "learning_rate": 8.727159502466364e-07,
      "loss": 0.7789,
      "step": 10980
    },
    {
      "epoch": 0.17101668132672204,
      "grad_norm": 3.8760182857513428,
      "learning_rate": 8.726340112420314e-07,
      "loss": 0.7943,
      "step": 10985
    },
    {
      "epoch": 0.1710945223286914,
      "grad_norm": 4.414963722229004,
      "learning_rate": 8.725520722374264e-07,
      "loss": 0.753,
      "step": 10990
    },
    {
      "epoch": 0.1711723633306608,
      "grad_norm": 8.448708534240723,
      "learning_rate": 8.724701332328215e-07,
      "loss": 0.704,
      "step": 10995
    },
    {
      "epoch": 0.17125020433263016,
      "grad_norm": 3.2329261302948,
      "learning_rate": 8.723881942282166e-07,
      "loss": 0.7565,
      "step": 11000
    },
    {
      "epoch": 0.17132804533459955,
      "grad_norm": 7.1403303146362305,
      "learning_rate": 8.723062552236114e-07,
      "loss": 0.7924,
      "step": 11005
    },
    {
      "epoch": 0.17140588633656892,
      "grad_norm": 2.50464129447937,
      "learning_rate": 8.722243162190065e-07,
      "loss": 0.8038,
      "step": 11010
    },
    {
      "epoch": 0.1714837273385383,
      "grad_norm": 3.236157178878784,
      "learning_rate": 8.721423772144016e-07,
      "loss": 0.7118,
      "step": 11015
    },
    {
      "epoch": 0.17156156834050768,
      "grad_norm": 4.607430934906006,
      "learning_rate": 8.720604382097966e-07,
      "loss": 0.8654,
      "step": 11020
    },
    {
      "epoch": 0.17163940934247707,
      "grad_norm": 3.002549171447754,
      "learning_rate": 8.719784992051916e-07,
      "loss": 0.7738,
      "step": 11025
    },
    {
      "epoch": 0.17171725034444643,
      "grad_norm": 3.545802116394043,
      "learning_rate": 8.718965602005867e-07,
      "loss": 0.7659,
      "step": 11030
    },
    {
      "epoch": 0.17179509134641582,
      "grad_norm": 4.228000640869141,
      "learning_rate": 8.718146211959817e-07,
      "loss": 0.89,
      "step": 11035
    },
    {
      "epoch": 0.1718729323483852,
      "grad_norm": 10.417171478271484,
      "learning_rate": 8.717326821913767e-07,
      "loss": 0.7852,
      "step": 11040
    },
    {
      "epoch": 0.17195077335035455,
      "grad_norm": 5.213564872741699,
      "learning_rate": 8.716507431867717e-07,
      "loss": 0.8346,
      "step": 11045
    },
    {
      "epoch": 0.17202861435232394,
      "grad_norm": 3.869044780731201,
      "learning_rate": 8.715688041821667e-07,
      "loss": 0.7275,
      "step": 11050
    },
    {
      "epoch": 0.1721064553542933,
      "grad_norm": 3.7918996810913086,
      "learning_rate": 8.714868651775618e-07,
      "loss": 0.9288,
      "step": 11055
    },
    {
      "epoch": 0.1721842963562627,
      "grad_norm": 3.620591640472412,
      "learning_rate": 8.714049261729568e-07,
      "loss": 0.8667,
      "step": 11060
    },
    {
      "epoch": 0.17226213735823206,
      "grad_norm": 3.3272628784179688,
      "learning_rate": 8.713229871683518e-07,
      "loss": 0.7839,
      "step": 11065
    },
    {
      "epoch": 0.17233997836020146,
      "grad_norm": 3.3981659412384033,
      "learning_rate": 8.712410481637469e-07,
      "loss": 0.8703,
      "step": 11070
    },
    {
      "epoch": 0.17241781936217082,
      "grad_norm": 2.7353670597076416,
      "learning_rate": 8.71159109159142e-07,
      "loss": 0.8323,
      "step": 11075
    },
    {
      "epoch": 0.1724956603641402,
      "grad_norm": 3.0876946449279785,
      "learning_rate": 8.710771701545369e-07,
      "loss": 0.7806,
      "step": 11080
    },
    {
      "epoch": 0.17257350136610958,
      "grad_norm": 4.163149833679199,
      "learning_rate": 8.709952311499319e-07,
      "loss": 0.738,
      "step": 11085
    },
    {
      "epoch": 0.17265134236807897,
      "grad_norm": 3.454596519470215,
      "learning_rate": 8.70913292145327e-07,
      "loss": 0.7692,
      "step": 11090
    },
    {
      "epoch": 0.17272918337004833,
      "grad_norm": 5.42886209487915,
      "learning_rate": 8.708313531407219e-07,
      "loss": 0.8061,
      "step": 11095
    },
    {
      "epoch": 0.17280702437201773,
      "grad_norm": 6.216919898986816,
      "learning_rate": 8.70749414136117e-07,
      "loss": 0.7982,
      "step": 11100
    },
    {
      "epoch": 0.1728848653739871,
      "grad_norm": 3.19822359085083,
      "learning_rate": 8.706674751315121e-07,
      "loss": 0.7365,
      "step": 11105
    },
    {
      "epoch": 0.17296270637595648,
      "grad_norm": 6.890682220458984,
      "learning_rate": 8.705855361269072e-07,
      "loss": 0.8121,
      "step": 11110
    },
    {
      "epoch": 0.17304054737792585,
      "grad_norm": 3.8538219928741455,
      "learning_rate": 8.705035971223021e-07,
      "loss": 0.8005,
      "step": 11115
    },
    {
      "epoch": 0.17311838837989524,
      "grad_norm": 4.351493835449219,
      "learning_rate": 8.704216581176972e-07,
      "loss": 0.7921,
      "step": 11120
    },
    {
      "epoch": 0.1731962293818646,
      "grad_norm": 4.173168182373047,
      "learning_rate": 8.703397191130923e-07,
      "loss": 0.7759,
      "step": 11125
    },
    {
      "epoch": 0.17327407038383397,
      "grad_norm": 3.0455987453460693,
      "learning_rate": 8.702577801084871e-07,
      "loss": 0.7061,
      "step": 11130
    },
    {
      "epoch": 0.17335191138580336,
      "grad_norm": 8.838937759399414,
      "learning_rate": 8.701758411038822e-07,
      "loss": 0.6888,
      "step": 11135
    },
    {
      "epoch": 0.17342975238777272,
      "grad_norm": 10.514293670654297,
      "learning_rate": 8.700939020992773e-07,
      "loss": 0.8403,
      "step": 11140
    },
    {
      "epoch": 0.17350759338974212,
      "grad_norm": 2.9291927814483643,
      "learning_rate": 8.700119630946723e-07,
      "loss": 0.8609,
      "step": 11145
    },
    {
      "epoch": 0.17358543439171148,
      "grad_norm": 4.258464336395264,
      "learning_rate": 8.699300240900673e-07,
      "loss": 0.8436,
      "step": 11150
    },
    {
      "epoch": 0.17366327539368087,
      "grad_norm": 4.499458312988281,
      "learning_rate": 8.698480850854624e-07,
      "loss": 0.883,
      "step": 11155
    },
    {
      "epoch": 0.17374111639565024,
      "grad_norm": 4.2294745445251465,
      "learning_rate": 8.697661460808574e-07,
      "loss": 0.711,
      "step": 11160
    },
    {
      "epoch": 0.17381895739761963,
      "grad_norm": 2.956054925918579,
      "learning_rate": 8.696842070762525e-07,
      "loss": 0.8658,
      "step": 11165
    },
    {
      "epoch": 0.173896798399589,
      "grad_norm": 6.596224308013916,
      "learning_rate": 8.696022680716474e-07,
      "loss": 0.8289,
      "step": 11170
    },
    {
      "epoch": 0.17397463940155838,
      "grad_norm": 4.582150936126709,
      "learning_rate": 8.695203290670424e-07,
      "loss": 0.7524,
      "step": 11175
    },
    {
      "epoch": 0.17405248040352775,
      "grad_norm": 2.815945625305176,
      "learning_rate": 8.694383900624375e-07,
      "loss": 0.8105,
      "step": 11180
    },
    {
      "epoch": 0.17413032140549714,
      "grad_norm": 7.399906635284424,
      "learning_rate": 8.693564510578325e-07,
      "loss": 0.628,
      "step": 11185
    },
    {
      "epoch": 0.1742081624074665,
      "grad_norm": 2.3162524700164795,
      "learning_rate": 8.692745120532275e-07,
      "loss": 0.6802,
      "step": 11190
    },
    {
      "epoch": 0.1742860034094359,
      "grad_norm": 4.691956043243408,
      "learning_rate": 8.691925730486226e-07,
      "loss": 0.9559,
      "step": 11195
    },
    {
      "epoch": 0.17436384441140526,
      "grad_norm": 5.132546901702881,
      "learning_rate": 8.691106340440177e-07,
      "loss": 0.8418,
      "step": 11200
    },
    {
      "epoch": 0.17444168541337465,
      "grad_norm": 6.407838344573975,
      "learning_rate": 8.690286950394126e-07,
      "loss": 0.9149,
      "step": 11205
    },
    {
      "epoch": 0.17451952641534402,
      "grad_norm": 2.8634774684906006,
      "learning_rate": 8.689467560348076e-07,
      "loss": 0.8361,
      "step": 11210
    },
    {
      "epoch": 0.1745973674173134,
      "grad_norm": 5.304955959320068,
      "learning_rate": 8.688648170302027e-07,
      "loss": 0.7911,
      "step": 11215
    },
    {
      "epoch": 0.17467520841928277,
      "grad_norm": 6.025475978851318,
      "learning_rate": 8.687828780255976e-07,
      "loss": 0.7571,
      "step": 11220
    },
    {
      "epoch": 0.17475304942125214,
      "grad_norm": 4.901416301727295,
      "learning_rate": 8.687009390209927e-07,
      "loss": 0.7321,
      "step": 11225
    },
    {
      "epoch": 0.17483089042322153,
      "grad_norm": 2.951046943664551,
      "learning_rate": 8.686190000163878e-07,
      "loss": 0.756,
      "step": 11230
    },
    {
      "epoch": 0.1749087314251909,
      "grad_norm": 3.0569169521331787,
      "learning_rate": 8.685370610117829e-07,
      "loss": 0.7937,
      "step": 11235
    },
    {
      "epoch": 0.1749865724271603,
      "grad_norm": 3.96439528465271,
      "learning_rate": 8.684551220071778e-07,
      "loss": 0.7065,
      "step": 11240
    },
    {
      "epoch": 0.17506441342912965,
      "grad_norm": 5.546309947967529,
      "learning_rate": 8.683731830025729e-07,
      "loss": 0.7787,
      "step": 11245
    },
    {
      "epoch": 0.17514225443109904,
      "grad_norm": 4.913859844207764,
      "learning_rate": 8.682912439979679e-07,
      "loss": 0.7689,
      "step": 11250
    },
    {
      "epoch": 0.1752200954330684,
      "grad_norm": 3.5462117195129395,
      "learning_rate": 8.682093049933628e-07,
      "loss": 0.7766,
      "step": 11255
    },
    {
      "epoch": 0.1752979364350378,
      "grad_norm": 5.14613676071167,
      "learning_rate": 8.681273659887579e-07,
      "loss": 0.8672,
      "step": 11260
    },
    {
      "epoch": 0.17537577743700716,
      "grad_norm": 3.5507776737213135,
      "learning_rate": 8.68045426984153e-07,
      "loss": 0.8519,
      "step": 11265
    },
    {
      "epoch": 0.17545361843897656,
      "grad_norm": 5.96008825302124,
      "learning_rate": 8.67963487979548e-07,
      "loss": 0.8072,
      "step": 11270
    },
    {
      "epoch": 0.17553145944094592,
      "grad_norm": 3.9602410793304443,
      "learning_rate": 8.67881548974943e-07,
      "loss": 0.7046,
      "step": 11275
    },
    {
      "epoch": 0.1756093004429153,
      "grad_norm": 9.932755470275879,
      "learning_rate": 8.677996099703381e-07,
      "loss": 0.6901,
      "step": 11280
    },
    {
      "epoch": 0.17568714144488468,
      "grad_norm": 3.0417749881744385,
      "learning_rate": 8.677176709657331e-07,
      "loss": 0.8085,
      "step": 11285
    },
    {
      "epoch": 0.17576498244685407,
      "grad_norm": 8.406500816345215,
      "learning_rate": 8.67635731961128e-07,
      "loss": 0.8751,
      "step": 11290
    },
    {
      "epoch": 0.17584282344882343,
      "grad_norm": 3.1526453495025635,
      "learning_rate": 8.675537929565231e-07,
      "loss": 0.779,
      "step": 11295
    },
    {
      "epoch": 0.17592066445079282,
      "grad_norm": 8.264491081237793,
      "learning_rate": 8.674718539519181e-07,
      "loss": 0.6696,
      "step": 11300
    },
    {
      "epoch": 0.1759985054527622,
      "grad_norm": 4.0117011070251465,
      "learning_rate": 8.673899149473132e-07,
      "loss": 0.7622,
      "step": 11305
    },
    {
      "epoch": 0.17607634645473155,
      "grad_norm": 5.8008527755737305,
      "learning_rate": 8.673079759427082e-07,
      "loss": 0.9383,
      "step": 11310
    },
    {
      "epoch": 0.17615418745670094,
      "grad_norm": 3.362180709838867,
      "learning_rate": 8.672260369381032e-07,
      "loss": 0.7315,
      "step": 11315
    },
    {
      "epoch": 0.1762320284586703,
      "grad_norm": 3.2175207138061523,
      "learning_rate": 8.671440979334983e-07,
      "loss": 0.7064,
      "step": 11320
    },
    {
      "epoch": 0.1763098694606397,
      "grad_norm": 3.180907964706421,
      "learning_rate": 8.670621589288934e-07,
      "loss": 0.7622,
      "step": 11325
    },
    {
      "epoch": 0.17638771046260907,
      "grad_norm": 4.250461578369141,
      "learning_rate": 8.669802199242882e-07,
      "loss": 0.9496,
      "step": 11330
    },
    {
      "epoch": 0.17646555146457846,
      "grad_norm": 3.719573497772217,
      "learning_rate": 8.668982809196833e-07,
      "loss": 0.7536,
      "step": 11335
    },
    {
      "epoch": 0.17654339246654782,
      "grad_norm": 4.155417442321777,
      "learning_rate": 8.668163419150784e-07,
      "loss": 0.7122,
      "step": 11340
    },
    {
      "epoch": 0.1766212334685172,
      "grad_norm": 5.191723346710205,
      "learning_rate": 8.667344029104733e-07,
      "loss": 0.6908,
      "step": 11345
    },
    {
      "epoch": 0.17669907447048658,
      "grad_norm": 6.202426910400391,
      "learning_rate": 8.666524639058684e-07,
      "loss": 0.8704,
      "step": 11350
    },
    {
      "epoch": 0.17677691547245597,
      "grad_norm": 3.2510268688201904,
      "learning_rate": 8.665705249012635e-07,
      "loss": 0.702,
      "step": 11355
    },
    {
      "epoch": 0.17685475647442533,
      "grad_norm": 7.769083023071289,
      "learning_rate": 8.664885858966586e-07,
      "loss": 0.8048,
      "step": 11360
    },
    {
      "epoch": 0.17693259747639473,
      "grad_norm": 3.8902037143707275,
      "learning_rate": 8.664066468920535e-07,
      "loss": 0.7719,
      "step": 11365
    },
    {
      "epoch": 0.1770104384783641,
      "grad_norm": 4.6013407707214355,
      "learning_rate": 8.663247078874485e-07,
      "loss": 0.7691,
      "step": 11370
    },
    {
      "epoch": 0.17708827948033348,
      "grad_norm": 7.063052654266357,
      "learning_rate": 8.662427688828436e-07,
      "loss": 0.8091,
      "step": 11375
    },
    {
      "epoch": 0.17716612048230285,
      "grad_norm": 4.544634819030762,
      "learning_rate": 8.661608298782386e-07,
      "loss": 0.9371,
      "step": 11380
    },
    {
      "epoch": 0.17724396148427224,
      "grad_norm": 4.195473670959473,
      "learning_rate": 8.660788908736336e-07,
      "loss": 0.7634,
      "step": 11385
    },
    {
      "epoch": 0.1773218024862416,
      "grad_norm": 18.608367919921875,
      "learning_rate": 8.659969518690287e-07,
      "loss": 0.7902,
      "step": 11390
    },
    {
      "epoch": 0.17739964348821097,
      "grad_norm": 3.536041498184204,
      "learning_rate": 8.659150128644237e-07,
      "loss": 0.7095,
      "step": 11395
    },
    {
      "epoch": 0.17747748449018036,
      "grad_norm": 3.1353678703308105,
      "learning_rate": 8.658330738598187e-07,
      "loss": 0.688,
      "step": 11400
    },
    {
      "epoch": 0.17755532549214972,
      "grad_norm": 2.603710889816284,
      "learning_rate": 8.657511348552138e-07,
      "loss": 0.8218,
      "step": 11405
    },
    {
      "epoch": 0.17763316649411912,
      "grad_norm": 3.3999760150909424,
      "learning_rate": 8.656691958506087e-07,
      "loss": 0.7279,
      "step": 11410
    },
    {
      "epoch": 0.17771100749608848,
      "grad_norm": 3.0341594219207764,
      "learning_rate": 8.655872568460038e-07,
      "loss": 0.754,
      "step": 11415
    },
    {
      "epoch": 0.17778884849805787,
      "grad_norm": 5.157776355743408,
      "learning_rate": 8.655053178413988e-07,
      "loss": 0.8347,
      "step": 11420
    },
    {
      "epoch": 0.17786668950002724,
      "grad_norm": 5.881651878356934,
      "learning_rate": 8.654233788367938e-07,
      "loss": 0.868,
      "step": 11425
    },
    {
      "epoch": 0.17794453050199663,
      "grad_norm": 7.980086326599121,
      "learning_rate": 8.653414398321889e-07,
      "loss": 0.8467,
      "step": 11430
    },
    {
      "epoch": 0.178022371503966,
      "grad_norm": 3.4080753326416016,
      "learning_rate": 8.65259500827584e-07,
      "loss": 0.641,
      "step": 11435
    },
    {
      "epoch": 0.17810021250593538,
      "grad_norm": 8.840136528015137,
      "learning_rate": 8.651775618229789e-07,
      "loss": 0.8391,
      "step": 11440
    },
    {
      "epoch": 0.17817805350790475,
      "grad_norm": 6.058139324188232,
      "learning_rate": 8.65095622818374e-07,
      "loss": 0.6896,
      "step": 11445
    },
    {
      "epoch": 0.17825589450987414,
      "grad_norm": 3.771533250808716,
      "learning_rate": 8.650136838137691e-07,
      "loss": 0.7656,
      "step": 11450
    },
    {
      "epoch": 0.1783337355118435,
      "grad_norm": 5.53726053237915,
      "learning_rate": 8.649317448091639e-07,
      "loss": 0.7161,
      "step": 11455
    },
    {
      "epoch": 0.1784115765138129,
      "grad_norm": 4.151732444763184,
      "learning_rate": 8.64849805804559e-07,
      "loss": 0.7258,
      "step": 11460
    },
    {
      "epoch": 0.17848941751578226,
      "grad_norm": 5.9917731285095215,
      "learning_rate": 8.647678667999541e-07,
      "loss": 0.8906,
      "step": 11465
    },
    {
      "epoch": 0.17856725851775165,
      "grad_norm": 3.5444159507751465,
      "learning_rate": 8.64685927795349e-07,
      "loss": 0.8006,
      "step": 11470
    },
    {
      "epoch": 0.17864509951972102,
      "grad_norm": 7.527516841888428,
      "learning_rate": 8.646039887907441e-07,
      "loss": 0.837,
      "step": 11475
    },
    {
      "epoch": 0.17872294052169038,
      "grad_norm": 3.505826711654663,
      "learning_rate": 8.645220497861392e-07,
      "loss": 0.7962,
      "step": 11480
    },
    {
      "epoch": 0.17880078152365977,
      "grad_norm": 4.634178161621094,
      "learning_rate": 8.644401107815343e-07,
      "loss": 0.7389,
      "step": 11485
    },
    {
      "epoch": 0.17887862252562914,
      "grad_norm": 2.834869623184204,
      "learning_rate": 8.643581717769292e-07,
      "loss": 0.837,
      "step": 11490
    },
    {
      "epoch": 0.17895646352759853,
      "grad_norm": 4.649651527404785,
      "learning_rate": 8.642762327723242e-07,
      "loss": 0.8676,
      "step": 11495
    },
    {
      "epoch": 0.1790343045295679,
      "grad_norm": 4.7943925857543945,
      "learning_rate": 8.641942937677193e-07,
      "loss": 0.7031,
      "step": 11500
    },
    {
      "epoch": 0.1791121455315373,
      "grad_norm": 3.4490435123443604,
      "learning_rate": 8.641123547631143e-07,
      "loss": 0.8332,
      "step": 11505
    },
    {
      "epoch": 0.17918998653350665,
      "grad_norm": 2.6124467849731445,
      "learning_rate": 8.640304157585093e-07,
      "loss": 0.8277,
      "step": 11510
    },
    {
      "epoch": 0.17926782753547604,
      "grad_norm": 3.379868507385254,
      "learning_rate": 8.639484767539044e-07,
      "loss": 0.8617,
      "step": 11515
    },
    {
      "epoch": 0.1793456685374454,
      "grad_norm": 4.773552417755127,
      "learning_rate": 8.638665377492994e-07,
      "loss": 0.8771,
      "step": 11520
    },
    {
      "epoch": 0.1794235095394148,
      "grad_norm": 8.424163818359375,
      "learning_rate": 8.637845987446945e-07,
      "loss": 0.8067,
      "step": 11525
    },
    {
      "epoch": 0.17950135054138416,
      "grad_norm": 4.11102294921875,
      "learning_rate": 8.637026597400895e-07,
      "loss": 0.7825,
      "step": 11530
    },
    {
      "epoch": 0.17957919154335356,
      "grad_norm": 3.056248188018799,
      "learning_rate": 8.636207207354844e-07,
      "loss": 0.805,
      "step": 11535
    },
    {
      "epoch": 0.17965703254532292,
      "grad_norm": 6.354325294494629,
      "learning_rate": 8.635387817308795e-07,
      "loss": 0.697,
      "step": 11540
    },
    {
      "epoch": 0.1797348735472923,
      "grad_norm": 4.208707809448242,
      "learning_rate": 8.634568427262745e-07,
      "loss": 0.7442,
      "step": 11545
    },
    {
      "epoch": 0.17981271454926168,
      "grad_norm": 3.256444215774536,
      "learning_rate": 8.633749037216695e-07,
      "loss": 0.7955,
      "step": 11550
    },
    {
      "epoch": 0.17989055555123107,
      "grad_norm": 2.9025378227233887,
      "learning_rate": 8.632929647170646e-07,
      "loss": 0.8062,
      "step": 11555
    },
    {
      "epoch": 0.17996839655320043,
      "grad_norm": 5.104341506958008,
      "learning_rate": 8.632110257124597e-07,
      "loss": 0.9282,
      "step": 11560
    },
    {
      "epoch": 0.1800462375551698,
      "grad_norm": 3.7267444133758545,
      "learning_rate": 8.631290867078546e-07,
      "loss": 0.7416,
      "step": 11565
    },
    {
      "epoch": 0.1801240785571392,
      "grad_norm": 2.9586052894592285,
      "learning_rate": 8.630471477032497e-07,
      "loss": 0.7961,
      "step": 11570
    },
    {
      "epoch": 0.18020191955910855,
      "grad_norm": 4.650422096252441,
      "learning_rate": 8.629652086986447e-07,
      "loss": 0.7148,
      "step": 11575
    },
    {
      "epoch": 0.18027976056107795,
      "grad_norm": 3.344991445541382,
      "learning_rate": 8.628832696940396e-07,
      "loss": 0.7804,
      "step": 11580
    },
    {
      "epoch": 0.1803576015630473,
      "grad_norm": 3.1043858528137207,
      "learning_rate": 8.628013306894347e-07,
      "loss": 0.7004,
      "step": 11585
    },
    {
      "epoch": 0.1804354425650167,
      "grad_norm": 4.275179862976074,
      "learning_rate": 8.627193916848298e-07,
      "loss": 0.702,
      "step": 11590
    },
    {
      "epoch": 0.18051328356698607,
      "grad_norm": 5.884884357452393,
      "learning_rate": 8.626374526802248e-07,
      "loss": 0.7196,
      "step": 11595
    },
    {
      "epoch": 0.18059112456895546,
      "grad_norm": 5.935474395751953,
      "learning_rate": 8.625555136756198e-07,
      "loss": 0.8548,
      "step": 11600
    },
    {
      "epoch": 0.18066896557092482,
      "grad_norm": 3.8844761848449707,
      "learning_rate": 8.624735746710149e-07,
      "loss": 0.7467,
      "step": 11605
    },
    {
      "epoch": 0.18074680657289421,
      "grad_norm": 3.4248223304748535,
      "learning_rate": 8.6239163566641e-07,
      "loss": 0.8664,
      "step": 11610
    },
    {
      "epoch": 0.18082464757486358,
      "grad_norm": 4.0504679679870605,
      "learning_rate": 8.623096966618048e-07,
      "loss": 0.8649,
      "step": 11615
    },
    {
      "epoch": 0.18090248857683297,
      "grad_norm": 3.0257060527801514,
      "learning_rate": 8.622277576571999e-07,
      "loss": 0.7091,
      "step": 11620
    },
    {
      "epoch": 0.18098032957880233,
      "grad_norm": 4.030515193939209,
      "learning_rate": 8.62145818652595e-07,
      "loss": 0.8485,
      "step": 11625
    },
    {
      "epoch": 0.18105817058077173,
      "grad_norm": 3.375437021255493,
      "learning_rate": 8.6206387964799e-07,
      "loss": 0.6832,
      "step": 11630
    },
    {
      "epoch": 0.1811360115827411,
      "grad_norm": 4.179788112640381,
      "learning_rate": 8.61981940643385e-07,
      "loss": 0.8009,
      "step": 11635
    },
    {
      "epoch": 0.18121385258471048,
      "grad_norm": 5.095760822296143,
      "learning_rate": 8.619000016387801e-07,
      "loss": 0.7984,
      "step": 11640
    },
    {
      "epoch": 0.18129169358667985,
      "grad_norm": 3.096256732940674,
      "learning_rate": 8.618180626341751e-07,
      "loss": 0.8468,
      "step": 11645
    },
    {
      "epoch": 0.1813695345886492,
      "grad_norm": 3.6533854007720947,
      "learning_rate": 8.617361236295702e-07,
      "loss": 0.7336,
      "step": 11650
    },
    {
      "epoch": 0.1814473755906186,
      "grad_norm": 3.505079984664917,
      "learning_rate": 8.616541846249651e-07,
      "loss": 0.7025,
      "step": 11655
    },
    {
      "epoch": 0.18152521659258797,
      "grad_norm": 3.6542341709136963,
      "learning_rate": 8.615722456203601e-07,
      "loss": 0.8489,
      "step": 11660
    },
    {
      "epoch": 0.18160305759455736,
      "grad_norm": 6.4186811447143555,
      "learning_rate": 8.614903066157552e-07,
      "loss": 0.7712,
      "step": 11665
    },
    {
      "epoch": 0.18168089859652672,
      "grad_norm": 5.470929145812988,
      "learning_rate": 8.614083676111502e-07,
      "loss": 0.819,
      "step": 11670
    },
    {
      "epoch": 0.18175873959849612,
      "grad_norm": 3.3178632259368896,
      "learning_rate": 8.613264286065452e-07,
      "loss": 0.813,
      "step": 11675
    },
    {
      "epoch": 0.18183658060046548,
      "grad_norm": 3.1316630840301514,
      "learning_rate": 8.612444896019403e-07,
      "loss": 0.7939,
      "step": 11680
    },
    {
      "epoch": 0.18191442160243487,
      "grad_norm": 5.289381980895996,
      "learning_rate": 8.611625505973354e-07,
      "loss": 0.7354,
      "step": 11685
    },
    {
      "epoch": 0.18199226260440424,
      "grad_norm": 2.2420859336853027,
      "learning_rate": 8.610806115927303e-07,
      "loss": 0.7458,
      "step": 11690
    },
    {
      "epoch": 0.18207010360637363,
      "grad_norm": 5.712198257446289,
      "learning_rate": 8.609986725881253e-07,
      "loss": 0.7175,
      "step": 11695
    },
    {
      "epoch": 0.182147944608343,
      "grad_norm": 4.2332353591918945,
      "learning_rate": 8.609167335835204e-07,
      "loss": 0.7712,
      "step": 11700
    },
    {
      "epoch": 0.18222578561031239,
      "grad_norm": 3.4181714057922363,
      "learning_rate": 8.608347945789153e-07,
      "loss": 0.7699,
      "step": 11705
    },
    {
      "epoch": 0.18230362661228175,
      "grad_norm": 4.554285049438477,
      "learning_rate": 8.607528555743104e-07,
      "loss": 0.7718,
      "step": 11710
    },
    {
      "epoch": 0.18238146761425114,
      "grad_norm": 4.693836212158203,
      "learning_rate": 8.606709165697055e-07,
      "loss": 0.8118,
      "step": 11715
    },
    {
      "epoch": 0.1824593086162205,
      "grad_norm": 5.117660999298096,
      "learning_rate": 8.605889775651005e-07,
      "loss": 0.7083,
      "step": 11720
    },
    {
      "epoch": 0.1825371496181899,
      "grad_norm": 7.209866046905518,
      "learning_rate": 8.605070385604955e-07,
      "loss": 0.832,
      "step": 11725
    },
    {
      "epoch": 0.18261499062015926,
      "grad_norm": 3.6614935398101807,
      "learning_rate": 8.604250995558906e-07,
      "loss": 0.7817,
      "step": 11730
    },
    {
      "epoch": 0.18269283162212863,
      "grad_norm": 3.863678216934204,
      "learning_rate": 8.603431605512855e-07,
      "loss": 0.8006,
      "step": 11735
    },
    {
      "epoch": 0.18277067262409802,
      "grad_norm": 3.742063522338867,
      "learning_rate": 8.602612215466806e-07,
      "loss": 0.7947,
      "step": 11740
    },
    {
      "epoch": 0.18284851362606738,
      "grad_norm": 2.856513261795044,
      "learning_rate": 8.601792825420756e-07,
      "loss": 0.745,
      "step": 11745
    },
    {
      "epoch": 0.18292635462803677,
      "grad_norm": 2.845435619354248,
      "learning_rate": 8.600973435374707e-07,
      "loss": 0.794,
      "step": 11750
    },
    {
      "epoch": 0.18300419563000614,
      "grad_norm": 8.055949211120605,
      "learning_rate": 8.600154045328657e-07,
      "loss": 0.7383,
      "step": 11755
    },
    {
      "epoch": 0.18308203663197553,
      "grad_norm": 4.0927510261535645,
      "learning_rate": 8.599334655282607e-07,
      "loss": 0.6759,
      "step": 11760
    },
    {
      "epoch": 0.1831598776339449,
      "grad_norm": 3.1355292797088623,
      "learning_rate": 8.598515265236558e-07,
      "loss": 0.771,
      "step": 11765
    },
    {
      "epoch": 0.1832377186359143,
      "grad_norm": 9.864462852478027,
      "learning_rate": 8.597695875190508e-07,
      "loss": 0.709,
      "step": 11770
    },
    {
      "epoch": 0.18331555963788365,
      "grad_norm": 5.377257823944092,
      "learning_rate": 8.596876485144459e-07,
      "loss": 0.903,
      "step": 11775
    },
    {
      "epoch": 0.18339340063985304,
      "grad_norm": 3.909209728240967,
      "learning_rate": 8.596057095098408e-07,
      "loss": 0.7645,
      "step": 11780
    },
    {
      "epoch": 0.1834712416418224,
      "grad_norm": 4.229231834411621,
      "learning_rate": 8.595237705052358e-07,
      "loss": 0.8469,
      "step": 11785
    },
    {
      "epoch": 0.1835490826437918,
      "grad_norm": 9.634142875671387,
      "learning_rate": 8.594418315006309e-07,
      "loss": 0.88,
      "step": 11790
    },
    {
      "epoch": 0.18362692364576116,
      "grad_norm": 2.9738218784332275,
      "learning_rate": 8.59359892496026e-07,
      "loss": 0.689,
      "step": 11795
    },
    {
      "epoch": 0.18370476464773056,
      "grad_norm": 3.0469038486480713,
      "learning_rate": 8.592779534914209e-07,
      "loss": 0.8383,
      "step": 11800
    },
    {
      "epoch": 0.18378260564969992,
      "grad_norm": 4.129268646240234,
      "learning_rate": 8.59196014486816e-07,
      "loss": 0.6902,
      "step": 11805
    },
    {
      "epoch": 0.1838604466516693,
      "grad_norm": 4.603461265563965,
      "learning_rate": 8.591140754822111e-07,
      "loss": 0.813,
      "step": 11810
    },
    {
      "epoch": 0.18393828765363868,
      "grad_norm": 4.814962387084961,
      "learning_rate": 8.59032136477606e-07,
      "loss": 0.8819,
      "step": 11815
    },
    {
      "epoch": 0.18401612865560804,
      "grad_norm": 7.185861110687256,
      "learning_rate": 8.58950197473001e-07,
      "loss": 0.8366,
      "step": 11820
    },
    {
      "epoch": 0.18409396965757743,
      "grad_norm": 7.243460655212402,
      "learning_rate": 8.588682584683961e-07,
      "loss": 0.9382,
      "step": 11825
    },
    {
      "epoch": 0.1841718106595468,
      "grad_norm": 6.482030391693115,
      "learning_rate": 8.587863194637911e-07,
      "loss": 0.8316,
      "step": 11830
    },
    {
      "epoch": 0.1842496516615162,
      "grad_norm": 12.124528884887695,
      "learning_rate": 8.587043804591861e-07,
      "loss": 0.8205,
      "step": 11835
    },
    {
      "epoch": 0.18432749266348555,
      "grad_norm": 4.70367431640625,
      "learning_rate": 8.586224414545812e-07,
      "loss": 0.6944,
      "step": 11840
    },
    {
      "epoch": 0.18440533366545495,
      "grad_norm": 6.1089558601379395,
      "learning_rate": 8.585405024499762e-07,
      "loss": 0.8094,
      "step": 11845
    },
    {
      "epoch": 0.1844831746674243,
      "grad_norm": 3.807187557220459,
      "learning_rate": 8.584585634453712e-07,
      "loss": 0.8715,
      "step": 11850
    },
    {
      "epoch": 0.1845610156693937,
      "grad_norm": 4.469877243041992,
      "learning_rate": 8.583766244407663e-07,
      "loss": 0.7794,
      "step": 11855
    },
    {
      "epoch": 0.18463885667136307,
      "grad_norm": 5.07852029800415,
      "learning_rate": 8.582946854361612e-07,
      "loss": 0.7603,
      "step": 11860
    },
    {
      "epoch": 0.18471669767333246,
      "grad_norm": 3.7635741233825684,
      "learning_rate": 8.582127464315563e-07,
      "loss": 0.817,
      "step": 11865
    },
    {
      "epoch": 0.18479453867530182,
      "grad_norm": 5.198869228363037,
      "learning_rate": 8.581308074269513e-07,
      "loss": 0.8442,
      "step": 11870
    },
    {
      "epoch": 0.18487237967727121,
      "grad_norm": 3.176208019256592,
      "learning_rate": 8.580488684223464e-07,
      "loss": 0.8135,
      "step": 11875
    },
    {
      "epoch": 0.18495022067924058,
      "grad_norm": 5.479611396789551,
      "learning_rate": 8.579669294177414e-07,
      "loss": 0.686,
      "step": 11880
    },
    {
      "epoch": 0.18502806168120997,
      "grad_norm": 5.823884010314941,
      "learning_rate": 8.578849904131365e-07,
      "loss": 0.7851,
      "step": 11885
    },
    {
      "epoch": 0.18510590268317934,
      "grad_norm": 4.050318717956543,
      "learning_rate": 8.578030514085315e-07,
      "loss": 0.7321,
      "step": 11890
    },
    {
      "epoch": 0.18518374368514873,
      "grad_norm": 2.703254461288452,
      "learning_rate": 8.577211124039265e-07,
      "loss": 0.7569,
      "step": 11895
    },
    {
      "epoch": 0.1852615846871181,
      "grad_norm": 6.201961517333984,
      "learning_rate": 8.576391733993215e-07,
      "loss": 0.8172,
      "step": 11900
    },
    {
      "epoch": 0.18533942568908748,
      "grad_norm": 4.213263511657715,
      "learning_rate": 8.575572343947165e-07,
      "loss": 0.8344,
      "step": 11905
    },
    {
      "epoch": 0.18541726669105685,
      "grad_norm": 6.157925128936768,
      "learning_rate": 8.574752953901115e-07,
      "loss": 0.7343,
      "step": 11910
    },
    {
      "epoch": 0.1854951076930262,
      "grad_norm": 4.5453691482543945,
      "learning_rate": 8.573933563855066e-07,
      "loss": 0.7038,
      "step": 11915
    },
    {
      "epoch": 0.1855729486949956,
      "grad_norm": 4.831943035125732,
      "learning_rate": 8.573114173809017e-07,
      "loss": 0.8805,
      "step": 11920
    },
    {
      "epoch": 0.18565078969696497,
      "grad_norm": 2.977743625640869,
      "learning_rate": 8.572294783762966e-07,
      "loss": 0.784,
      "step": 11925
    },
    {
      "epoch": 0.18572863069893436,
      "grad_norm": 5.694007396697998,
      "learning_rate": 8.571475393716917e-07,
      "loss": 0.6915,
      "step": 11930
    },
    {
      "epoch": 0.18580647170090372,
      "grad_norm": 4.1263017654418945,
      "learning_rate": 8.570656003670868e-07,
      "loss": 0.7476,
      "step": 11935
    },
    {
      "epoch": 0.18588431270287312,
      "grad_norm": 5.350509166717529,
      "learning_rate": 8.569836613624816e-07,
      "loss": 0.6784,
      "step": 11940
    },
    {
      "epoch": 0.18596215370484248,
      "grad_norm": 3.7994115352630615,
      "learning_rate": 8.569017223578767e-07,
      "loss": 0.7191,
      "step": 11945
    },
    {
      "epoch": 0.18603999470681187,
      "grad_norm": 5.086226940155029,
      "learning_rate": 8.568197833532718e-07,
      "loss": 0.7841,
      "step": 11950
    },
    {
      "epoch": 0.18611783570878124,
      "grad_norm": 4.248946189880371,
      "learning_rate": 8.567378443486668e-07,
      "loss": 0.8489,
      "step": 11955
    },
    {
      "epoch": 0.18619567671075063,
      "grad_norm": 2.877885580062866,
      "learning_rate": 8.566559053440618e-07,
      "loss": 0.66,
      "step": 11960
    },
    {
      "epoch": 0.18627351771272,
      "grad_norm": 6.250997066497803,
      "learning_rate": 8.565739663394569e-07,
      "loss": 0.7974,
      "step": 11965
    },
    {
      "epoch": 0.18635135871468939,
      "grad_norm": 5.646812915802002,
      "learning_rate": 8.564920273348519e-07,
      "loss": 0.87,
      "step": 11970
    },
    {
      "epoch": 0.18642919971665875,
      "grad_norm": 3.944369316101074,
      "learning_rate": 8.56410088330247e-07,
      "loss": 0.8484,
      "step": 11975
    },
    {
      "epoch": 0.18650704071862814,
      "grad_norm": 3.154167413711548,
      "learning_rate": 8.563281493256419e-07,
      "loss": 0.728,
      "step": 11980
    },
    {
      "epoch": 0.1865848817205975,
      "grad_norm": 5.012053489685059,
      "learning_rate": 8.562462103210369e-07,
      "loss": 0.7125,
      "step": 11985
    },
    {
      "epoch": 0.1866627227225669,
      "grad_norm": 5.746982097625732,
      "learning_rate": 8.56164271316432e-07,
      "loss": 0.7059,
      "step": 11990
    },
    {
      "epoch": 0.18674056372453626,
      "grad_norm": 3.114208698272705,
      "learning_rate": 8.56082332311827e-07,
      "loss": 0.76,
      "step": 11995
    },
    {
      "epoch": 0.18681840472650563,
      "grad_norm": 3.1285858154296875,
      "learning_rate": 8.560003933072221e-07,
      "loss": 0.8049,
      "step": 12000
    },
    {
      "epoch": 0.18689624572847502,
      "grad_norm": 3.38901686668396,
      "learning_rate": 8.559184543026171e-07,
      "loss": 0.6621,
      "step": 12005
    },
    {
      "epoch": 0.18697408673044438,
      "grad_norm": 3.768120050430298,
      "learning_rate": 8.558365152980122e-07,
      "loss": 0.7425,
      "step": 12010
    },
    {
      "epoch": 0.18705192773241378,
      "grad_norm": 5.235153675079346,
      "learning_rate": 8.557545762934072e-07,
      "loss": 0.731,
      "step": 12015
    },
    {
      "epoch": 0.18712976873438314,
      "grad_norm": 5.779262065887451,
      "learning_rate": 8.556726372888021e-07,
      "loss": 0.6507,
      "step": 12020
    },
    {
      "epoch": 0.18720760973635253,
      "grad_norm": 6.876397609710693,
      "learning_rate": 8.555906982841972e-07,
      "loss": 0.9053,
      "step": 12025
    },
    {
      "epoch": 0.1872854507383219,
      "grad_norm": 3.9089648723602295,
      "learning_rate": 8.555087592795922e-07,
      "loss": 0.887,
      "step": 12030
    },
    {
      "epoch": 0.1873632917402913,
      "grad_norm": 4.342790603637695,
      "learning_rate": 8.554268202749872e-07,
      "loss": 0.8031,
      "step": 12035
    },
    {
      "epoch": 0.18744113274226065,
      "grad_norm": 4.621131420135498,
      "learning_rate": 8.553448812703823e-07,
      "loss": 0.7683,
      "step": 12040
    },
    {
      "epoch": 0.18751897374423004,
      "grad_norm": 4.215606212615967,
      "learning_rate": 8.552629422657774e-07,
      "loss": 0.7317,
      "step": 12045
    },
    {
      "epoch": 0.1875968147461994,
      "grad_norm": 4.2532172203063965,
      "learning_rate": 8.551810032611723e-07,
      "loss": 0.8652,
      "step": 12050
    },
    {
      "epoch": 0.1876746557481688,
      "grad_norm": 3.5596415996551514,
      "learning_rate": 8.550990642565674e-07,
      "loss": 0.8786,
      "step": 12055
    },
    {
      "epoch": 0.18775249675013816,
      "grad_norm": 4.018403053283691,
      "learning_rate": 8.550171252519624e-07,
      "loss": 0.8089,
      "step": 12060
    },
    {
      "epoch": 0.18783033775210756,
      "grad_norm": 3.3754570484161377,
      "learning_rate": 8.549351862473573e-07,
      "loss": 0.8194,
      "step": 12065
    },
    {
      "epoch": 0.18790817875407692,
      "grad_norm": 3.737074375152588,
      "learning_rate": 8.548532472427524e-07,
      "loss": 0.881,
      "step": 12070
    },
    {
      "epoch": 0.1879860197560463,
      "grad_norm": 2.991834878921509,
      "learning_rate": 8.547713082381475e-07,
      "loss": 0.7424,
      "step": 12075
    },
    {
      "epoch": 0.18806386075801568,
      "grad_norm": 12.81692886352539,
      "learning_rate": 8.546893692335425e-07,
      "loss": 0.747,
      "step": 12080
    },
    {
      "epoch": 0.18814170175998504,
      "grad_norm": 5.266377925872803,
      "learning_rate": 8.546074302289375e-07,
      "loss": 0.799,
      "step": 12085
    },
    {
      "epoch": 0.18821954276195443,
      "grad_norm": 4.115131378173828,
      "learning_rate": 8.545254912243326e-07,
      "loss": 0.8458,
      "step": 12090
    },
    {
      "epoch": 0.1882973837639238,
      "grad_norm": 2.350405216217041,
      "learning_rate": 8.544435522197276e-07,
      "loss": 0.8161,
      "step": 12095
    },
    {
      "epoch": 0.1883752247658932,
      "grad_norm": 6.091660499572754,
      "learning_rate": 8.543616132151227e-07,
      "loss": 0.7929,
      "step": 12100
    },
    {
      "epoch": 0.18845306576786255,
      "grad_norm": 5.162372589111328,
      "learning_rate": 8.542796742105176e-07,
      "loss": 0.7737,
      "step": 12105
    },
    {
      "epoch": 0.18853090676983195,
      "grad_norm": 3.9324533939361572,
      "learning_rate": 8.541977352059126e-07,
      "loss": 0.7208,
      "step": 12110
    },
    {
      "epoch": 0.1886087477718013,
      "grad_norm": 5.336818218231201,
      "learning_rate": 8.541157962013077e-07,
      "loss": 0.8538,
      "step": 12115
    },
    {
      "epoch": 0.1886865887737707,
      "grad_norm": 4.819896221160889,
      "learning_rate": 8.540338571967027e-07,
      "loss": 0.8404,
      "step": 12120
    },
    {
      "epoch": 0.18876442977574007,
      "grad_norm": 4.829216480255127,
      "learning_rate": 8.539519181920978e-07,
      "loss": 0.8186,
      "step": 12125
    },
    {
      "epoch": 0.18884227077770946,
      "grad_norm": 3.225750684738159,
      "learning_rate": 8.538699791874928e-07,
      "loss": 0.605,
      "step": 12130
    },
    {
      "epoch": 0.18892011177967882,
      "grad_norm": 5.357349872589111,
      "learning_rate": 8.537880401828879e-07,
      "loss": 0.8215,
      "step": 12135
    },
    {
      "epoch": 0.18899795278164822,
      "grad_norm": 5.640933990478516,
      "learning_rate": 8.537061011782829e-07,
      "loss": 0.844,
      "step": 12140
    },
    {
      "epoch": 0.18907579378361758,
      "grad_norm": 6.281761169433594,
      "learning_rate": 8.536241621736778e-07,
      "loss": 0.7619,
      "step": 12145
    },
    {
      "epoch": 0.18915363478558697,
      "grad_norm": 4.4332661628723145,
      "learning_rate": 8.535422231690729e-07,
      "loss": 0.7782,
      "step": 12150
    },
    {
      "epoch": 0.18923147578755634,
      "grad_norm": 6.337418556213379,
      "learning_rate": 8.53460284164468e-07,
      "loss": 0.8855,
      "step": 12155
    },
    {
      "epoch": 0.18930931678952573,
      "grad_norm": 3.157975912094116,
      "learning_rate": 8.533783451598629e-07,
      "loss": 0.6462,
      "step": 12160
    },
    {
      "epoch": 0.1893871577914951,
      "grad_norm": 3.5442986488342285,
      "learning_rate": 8.53296406155258e-07,
      "loss": 0.7451,
      "step": 12165
    },
    {
      "epoch": 0.18946499879346446,
      "grad_norm": 3.5435712337493896,
      "learning_rate": 8.532144671506531e-07,
      "loss": 0.741,
      "step": 12170
    },
    {
      "epoch": 0.18954283979543385,
      "grad_norm": 2.5443406105041504,
      "learning_rate": 8.53132528146048e-07,
      "loss": 0.8358,
      "step": 12175
    },
    {
      "epoch": 0.1896206807974032,
      "grad_norm": 3.2555923461914062,
      "learning_rate": 8.530505891414431e-07,
      "loss": 0.7672,
      "step": 12180
    },
    {
      "epoch": 0.1896985217993726,
      "grad_norm": 4.258267879486084,
      "learning_rate": 8.529686501368381e-07,
      "loss": 0.6727,
      "step": 12185
    },
    {
      "epoch": 0.18977636280134197,
      "grad_norm": 5.2711920738220215,
      "learning_rate": 8.528867111322331e-07,
      "loss": 0.6505,
      "step": 12190
    },
    {
      "epoch": 0.18985420380331136,
      "grad_norm": 3.2238636016845703,
      "learning_rate": 8.528047721276281e-07,
      "loss": 0.8701,
      "step": 12195
    },
    {
      "epoch": 0.18993204480528073,
      "grad_norm": 3.0218734741210938,
      "learning_rate": 8.527228331230232e-07,
      "loss": 0.7897,
      "step": 12200
    },
    {
      "epoch": 0.19000988580725012,
      "grad_norm": 3.167506694793701,
      "learning_rate": 8.526408941184182e-07,
      "loss": 0.7731,
      "step": 12205
    },
    {
      "epoch": 0.19008772680921948,
      "grad_norm": 2.4751815795898438,
      "learning_rate": 8.525589551138132e-07,
      "loss": 0.6739,
      "step": 12210
    },
    {
      "epoch": 0.19016556781118887,
      "grad_norm": 6.196854114532471,
      "learning_rate": 8.524770161092083e-07,
      "loss": 0.7885,
      "step": 12215
    },
    {
      "epoch": 0.19024340881315824,
      "grad_norm": 3.2802228927612305,
      "learning_rate": 8.523950771046033e-07,
      "loss": 0.8174,
      "step": 12220
    },
    {
      "epoch": 0.19032124981512763,
      "grad_norm": 2.8430838584899902,
      "learning_rate": 8.523131380999983e-07,
      "loss": 0.7345,
      "step": 12225
    },
    {
      "epoch": 0.190399090817097,
      "grad_norm": 3.185624837875366,
      "learning_rate": 8.522311990953933e-07,
      "loss": 0.7387,
      "step": 12230
    },
    {
      "epoch": 0.1904769318190664,
      "grad_norm": 4.954017639160156,
      "learning_rate": 8.521492600907883e-07,
      "loss": 0.7631,
      "step": 12235
    },
    {
      "epoch": 0.19055477282103575,
      "grad_norm": 3.4082043170928955,
      "learning_rate": 8.520673210861834e-07,
      "loss": 0.6641,
      "step": 12240
    },
    {
      "epoch": 0.19063261382300514,
      "grad_norm": 4.00759744644165,
      "learning_rate": 8.519853820815785e-07,
      "loss": 0.6956,
      "step": 12245
    },
    {
      "epoch": 0.1907104548249745,
      "grad_norm": 3.7741944789886475,
      "learning_rate": 8.519034430769735e-07,
      "loss": 0.8373,
      "step": 12250
    },
    {
      "epoch": 0.19078829582694387,
      "grad_norm": 8.064949035644531,
      "learning_rate": 8.518215040723685e-07,
      "loss": 0.8908,
      "step": 12255
    },
    {
      "epoch": 0.19086613682891326,
      "grad_norm": 3.7881693840026855,
      "learning_rate": 8.517395650677636e-07,
      "loss": 0.7637,
      "step": 12260
    },
    {
      "epoch": 0.19094397783088263,
      "grad_norm": 3.755758762359619,
      "learning_rate": 8.516576260631585e-07,
      "loss": 0.7973,
      "step": 12265
    },
    {
      "epoch": 0.19102181883285202,
      "grad_norm": 5.2573370933532715,
      "learning_rate": 8.515756870585535e-07,
      "loss": 0.6995,
      "step": 12270
    },
    {
      "epoch": 0.19109965983482138,
      "grad_norm": 3.6603024005889893,
      "learning_rate": 8.514937480539486e-07,
      "loss": 0.7853,
      "step": 12275
    },
    {
      "epoch": 0.19117750083679078,
      "grad_norm": 5.118966102600098,
      "learning_rate": 8.514118090493437e-07,
      "loss": 0.884,
      "step": 12280
    },
    {
      "epoch": 0.19125534183876014,
      "grad_norm": 3.1857612133026123,
      "learning_rate": 8.513298700447386e-07,
      "loss": 0.804,
      "step": 12285
    },
    {
      "epoch": 0.19133318284072953,
      "grad_norm": 5.507519245147705,
      "learning_rate": 8.512479310401337e-07,
      "loss": 0.7884,
      "step": 12290
    },
    {
      "epoch": 0.1914110238426989,
      "grad_norm": 8.014042854309082,
      "learning_rate": 8.511659920355288e-07,
      "loss": 0.9034,
      "step": 12295
    },
    {
      "epoch": 0.1914888648446683,
      "grad_norm": 3.642604112625122,
      "learning_rate": 8.510840530309238e-07,
      "loss": 0.7831,
      "step": 12300
    },
    {
      "epoch": 0.19156670584663765,
      "grad_norm": 3.5193252563476562,
      "learning_rate": 8.510021140263187e-07,
      "loss": 0.8202,
      "step": 12305
    },
    {
      "epoch": 0.19164454684860704,
      "grad_norm": 4.22996187210083,
      "learning_rate": 8.509201750217138e-07,
      "loss": 0.6753,
      "step": 12310
    },
    {
      "epoch": 0.1917223878505764,
      "grad_norm": 3.661057472229004,
      "learning_rate": 8.508382360171088e-07,
      "loss": 0.785,
      "step": 12315
    },
    {
      "epoch": 0.1918002288525458,
      "grad_norm": 3.3151731491088867,
      "learning_rate": 8.507562970125038e-07,
      "loss": 0.7575,
      "step": 12320
    },
    {
      "epoch": 0.19187806985451517,
      "grad_norm": 2.6864590644836426,
      "learning_rate": 8.506743580078989e-07,
      "loss": 0.8164,
      "step": 12325
    },
    {
      "epoch": 0.19195591085648456,
      "grad_norm": 5.895023345947266,
      "learning_rate": 8.505924190032939e-07,
      "loss": 0.8764,
      "step": 12330
    },
    {
      "epoch": 0.19203375185845392,
      "grad_norm": 5.192370414733887,
      "learning_rate": 8.50510479998689e-07,
      "loss": 0.6947,
      "step": 12335
    },
    {
      "epoch": 0.19211159286042329,
      "grad_norm": 2.818570375442505,
      "learning_rate": 8.50428540994084e-07,
      "loss": 0.8609,
      "step": 12340
    },
    {
      "epoch": 0.19218943386239268,
      "grad_norm": 4.249600887298584,
      "learning_rate": 8.503466019894789e-07,
      "loss": 0.7599,
      "step": 12345
    },
    {
      "epoch": 0.19226727486436204,
      "grad_norm": 3.1460306644439697,
      "learning_rate": 8.50264662984874e-07,
      "loss": 0.6936,
      "step": 12350
    },
    {
      "epoch": 0.19234511586633143,
      "grad_norm": 9.153656959533691,
      "learning_rate": 8.50182723980269e-07,
      "loss": 0.8209,
      "step": 12355
    },
    {
      "epoch": 0.1924229568683008,
      "grad_norm": 4.739767074584961,
      "learning_rate": 8.50100784975664e-07,
      "loss": 0.8558,
      "step": 12360
    },
    {
      "epoch": 0.1925007978702702,
      "grad_norm": 2.902334451675415,
      "learning_rate": 8.500188459710591e-07,
      "loss": 0.6182,
      "step": 12365
    },
    {
      "epoch": 0.19257863887223955,
      "grad_norm": 5.967688083648682,
      "learning_rate": 8.499369069664542e-07,
      "loss": 0.843,
      "step": 12370
    },
    {
      "epoch": 0.19265647987420895,
      "grad_norm": 3.8589203357696533,
      "learning_rate": 8.498549679618492e-07,
      "loss": 0.7795,
      "step": 12375
    },
    {
      "epoch": 0.1927343208761783,
      "grad_norm": 3.3635146617889404,
      "learning_rate": 8.497730289572442e-07,
      "loss": 0.7461,
      "step": 12380
    },
    {
      "epoch": 0.1928121618781477,
      "grad_norm": 3.772067070007324,
      "learning_rate": 8.496910899526392e-07,
      "loss": 0.8692,
      "step": 12385
    },
    {
      "epoch": 0.19289000288011707,
      "grad_norm": 3.3439524173736572,
      "learning_rate": 8.496091509480343e-07,
      "loss": 0.6822,
      "step": 12390
    },
    {
      "epoch": 0.19296784388208646,
      "grad_norm": 6.4128875732421875,
      "learning_rate": 8.495272119434292e-07,
      "loss": 0.8279,
      "step": 12395
    },
    {
      "epoch": 0.19304568488405582,
      "grad_norm": 3.5066323280334473,
      "learning_rate": 8.494452729388243e-07,
      "loss": 0.8597,
      "step": 12400
    },
    {
      "epoch": 0.19312352588602522,
      "grad_norm": 4.82898473739624,
      "learning_rate": 8.493633339342194e-07,
      "loss": 0.8872,
      "step": 12405
    },
    {
      "epoch": 0.19320136688799458,
      "grad_norm": 7.195667266845703,
      "learning_rate": 8.492813949296143e-07,
      "loss": 0.7993,
      "step": 12410
    },
    {
      "epoch": 0.19327920788996397,
      "grad_norm": 3.0416855812072754,
      "learning_rate": 8.491994559250094e-07,
      "loss": 0.8976,
      "step": 12415
    },
    {
      "epoch": 0.19335704889193334,
      "grad_norm": 4.605866432189941,
      "learning_rate": 8.491175169204045e-07,
      "loss": 0.6899,
      "step": 12420
    },
    {
      "epoch": 0.1934348898939027,
      "grad_norm": 2.6966631412506104,
      "learning_rate": 8.490355779157995e-07,
      "loss": 0.7145,
      "step": 12425
    },
    {
      "epoch": 0.1935127308958721,
      "grad_norm": 4.953130722045898,
      "learning_rate": 8.489536389111944e-07,
      "loss": 0.789,
      "step": 12430
    },
    {
      "epoch": 0.19359057189784146,
      "grad_norm": 2.913949489593506,
      "learning_rate": 8.488716999065895e-07,
      "loss": 0.7734,
      "step": 12435
    },
    {
      "epoch": 0.19366841289981085,
      "grad_norm": 2.37611722946167,
      "learning_rate": 8.487897609019845e-07,
      "loss": 0.663,
      "step": 12440
    },
    {
      "epoch": 0.1937462539017802,
      "grad_norm": 6.736103534698486,
      "learning_rate": 8.487078218973795e-07,
      "loss": 0.5834,
      "step": 12445
    },
    {
      "epoch": 0.1938240949037496,
      "grad_norm": 7.212032318115234,
      "learning_rate": 8.486258828927746e-07,
      "loss": 0.7853,
      "step": 12450
    },
    {
      "epoch": 0.19390193590571897,
      "grad_norm": 3.8290274143218994,
      "learning_rate": 8.485439438881696e-07,
      "loss": 0.7891,
      "step": 12455
    },
    {
      "epoch": 0.19397977690768836,
      "grad_norm": 2.9193921089172363,
      "learning_rate": 8.484620048835647e-07,
      "loss": 0.8621,
      "step": 12460
    },
    {
      "epoch": 0.19405761790965773,
      "grad_norm": 3.4873313903808594,
      "learning_rate": 8.483800658789597e-07,
      "loss": 0.6691,
      "step": 12465
    },
    {
      "epoch": 0.19413545891162712,
      "grad_norm": 4.49635124206543,
      "learning_rate": 8.482981268743546e-07,
      "loss": 0.7417,
      "step": 12470
    },
    {
      "epoch": 0.19421329991359648,
      "grad_norm": 3.002316951751709,
      "learning_rate": 8.482161878697497e-07,
      "loss": 0.8712,
      "step": 12475
    },
    {
      "epoch": 0.19429114091556587,
      "grad_norm": 2.9635400772094727,
      "learning_rate": 8.481342488651448e-07,
      "loss": 0.7171,
      "step": 12480
    },
    {
      "epoch": 0.19436898191753524,
      "grad_norm": 3.025629997253418,
      "learning_rate": 8.480523098605397e-07,
      "loss": 0.7841,
      "step": 12485
    },
    {
      "epoch": 0.19444682291950463,
      "grad_norm": 5.012829303741455,
      "learning_rate": 8.479703708559348e-07,
      "loss": 0.7381,
      "step": 12490
    },
    {
      "epoch": 0.194524663921474,
      "grad_norm": 4.070782661437988,
      "learning_rate": 8.478884318513299e-07,
      "loss": 0.7512,
      "step": 12495
    },
    {
      "epoch": 0.1946025049234434,
      "grad_norm": 3.400742769241333,
      "learning_rate": 8.478064928467249e-07,
      "loss": 0.7998,
      "step": 12500
    },
    {
      "epoch": 0.19468034592541275,
      "grad_norm": 6.554990768432617,
      "learning_rate": 8.477245538421199e-07,
      "loss": 0.8623,
      "step": 12505
    },
    {
      "epoch": 0.19475818692738214,
      "grad_norm": 3.837843179702759,
      "learning_rate": 8.476426148375149e-07,
      "loss": 0.7377,
      "step": 12510
    },
    {
      "epoch": 0.1948360279293515,
      "grad_norm": 3.646177291870117,
      "learning_rate": 8.4756067583291e-07,
      "loss": 0.8649,
      "step": 12515
    },
    {
      "epoch": 0.19491386893132087,
      "grad_norm": 5.887502193450928,
      "learning_rate": 8.474787368283049e-07,
      "loss": 0.7227,
      "step": 12520
    },
    {
      "epoch": 0.19499170993329026,
      "grad_norm": 3.3450663089752197,
      "learning_rate": 8.473967978237e-07,
      "loss": 0.7593,
      "step": 12525
    },
    {
      "epoch": 0.19506955093525963,
      "grad_norm": 3.6260218620300293,
      "learning_rate": 8.473148588190951e-07,
      "loss": 0.6893,
      "step": 12530
    },
    {
      "epoch": 0.19514739193722902,
      "grad_norm": 3.210761308670044,
      "learning_rate": 8.4723291981449e-07,
      "loss": 0.64,
      "step": 12535
    },
    {
      "epoch": 0.19522523293919838,
      "grad_norm": 7.629252910614014,
      "learning_rate": 8.471509808098851e-07,
      "loss": 0.7134,
      "step": 12540
    },
    {
      "epoch": 0.19530307394116778,
      "grad_norm": 3.362891435623169,
      "learning_rate": 8.470690418052802e-07,
      "loss": 0.6621,
      "step": 12545
    },
    {
      "epoch": 0.19538091494313714,
      "grad_norm": 3.2264938354492188,
      "learning_rate": 8.469871028006751e-07,
      "loss": 0.7544,
      "step": 12550
    },
    {
      "epoch": 0.19545875594510653,
      "grad_norm": 4.121139049530029,
      "learning_rate": 8.469051637960701e-07,
      "loss": 0.8005,
      "step": 12555
    },
    {
      "epoch": 0.1955365969470759,
      "grad_norm": 4.077367305755615,
      "learning_rate": 8.468232247914652e-07,
      "loss": 0.7671,
      "step": 12560
    },
    {
      "epoch": 0.1956144379490453,
      "grad_norm": 4.537203311920166,
      "learning_rate": 8.467412857868602e-07,
      "loss": 0.7573,
      "step": 12565
    },
    {
      "epoch": 0.19569227895101465,
      "grad_norm": 4.151341438293457,
      "learning_rate": 8.466593467822553e-07,
      "loss": 0.7534,
      "step": 12570
    },
    {
      "epoch": 0.19577011995298405,
      "grad_norm": 4.7857584953308105,
      "learning_rate": 8.465774077776503e-07,
      "loss": 0.7586,
      "step": 12575
    },
    {
      "epoch": 0.1958479609549534,
      "grad_norm": 3.5333428382873535,
      "learning_rate": 8.464954687730453e-07,
      "loss": 0.7703,
      "step": 12580
    },
    {
      "epoch": 0.1959258019569228,
      "grad_norm": 3.486469030380249,
      "learning_rate": 8.464135297684404e-07,
      "loss": 0.7808,
      "step": 12585
    },
    {
      "epoch": 0.19600364295889217,
      "grad_norm": 5.123147487640381,
      "learning_rate": 8.463315907638353e-07,
      "loss": 0.7586,
      "step": 12590
    },
    {
      "epoch": 0.19608148396086156,
      "grad_norm": 5.875349521636963,
      "learning_rate": 8.462496517592303e-07,
      "loss": 0.8283,
      "step": 12595
    },
    {
      "epoch": 0.19615932496283092,
      "grad_norm": 3.429596424102783,
      "learning_rate": 8.461677127546254e-07,
      "loss": 0.9282,
      "step": 12600
    },
    {
      "epoch": 0.1962371659648003,
      "grad_norm": 4.255092620849609,
      "learning_rate": 8.460857737500205e-07,
      "loss": 0.7489,
      "step": 12605
    },
    {
      "epoch": 0.19631500696676968,
      "grad_norm": 5.298252582550049,
      "learning_rate": 8.460038347454154e-07,
      "loss": 0.8015,
      "step": 12610
    },
    {
      "epoch": 0.19639284796873904,
      "grad_norm": 3.1947810649871826,
      "learning_rate": 8.459218957408105e-07,
      "loss": 0.8742,
      "step": 12615
    },
    {
      "epoch": 0.19647068897070843,
      "grad_norm": 2.557737112045288,
      "learning_rate": 8.458399567362056e-07,
      "loss": 0.7221,
      "step": 12620
    },
    {
      "epoch": 0.1965485299726778,
      "grad_norm": 3.7738475799560547,
      "learning_rate": 8.457580177316007e-07,
      "loss": 0.8054,
      "step": 12625
    },
    {
      "epoch": 0.1966263709746472,
      "grad_norm": 3.820087432861328,
      "learning_rate": 8.456760787269955e-07,
      "loss": 0.8202,
      "step": 12630
    },
    {
      "epoch": 0.19670421197661656,
      "grad_norm": 9.728828430175781,
      "learning_rate": 8.455941397223906e-07,
      "loss": 0.9442,
      "step": 12635
    },
    {
      "epoch": 0.19678205297858595,
      "grad_norm": 2.848874807357788,
      "learning_rate": 8.455122007177857e-07,
      "loss": 0.8407,
      "step": 12640
    },
    {
      "epoch": 0.1968598939805553,
      "grad_norm": 2.538989305496216,
      "learning_rate": 8.454302617131806e-07,
      "loss": 0.6267,
      "step": 12645
    },
    {
      "epoch": 0.1969377349825247,
      "grad_norm": 4.164561748504639,
      "learning_rate": 8.453483227085757e-07,
      "loss": 0.8619,
      "step": 12650
    },
    {
      "epoch": 0.19701557598449407,
      "grad_norm": 2.70489239692688,
      "learning_rate": 8.452663837039708e-07,
      "loss": 0.6901,
      "step": 12655
    },
    {
      "epoch": 0.19709341698646346,
      "grad_norm": 5.329331398010254,
      "learning_rate": 8.451844446993658e-07,
      "loss": 0.8083,
      "step": 12660
    },
    {
      "epoch": 0.19717125798843282,
      "grad_norm": 3.1759660243988037,
      "learning_rate": 8.451025056947608e-07,
      "loss": 0.6953,
      "step": 12665
    },
    {
      "epoch": 0.19724909899040222,
      "grad_norm": 3.253971576690674,
      "learning_rate": 8.450205666901558e-07,
      "loss": 0.7275,
      "step": 12670
    },
    {
      "epoch": 0.19732693999237158,
      "grad_norm": 4.8069634437561035,
      "learning_rate": 8.449386276855508e-07,
      "loss": 0.874,
      "step": 12675
    },
    {
      "epoch": 0.19740478099434097,
      "grad_norm": 5.485657215118408,
      "learning_rate": 8.448566886809458e-07,
      "loss": 0.7752,
      "step": 12680
    },
    {
      "epoch": 0.19748262199631034,
      "grad_norm": 3.5033183097839355,
      "learning_rate": 8.447747496763409e-07,
      "loss": 0.8751,
      "step": 12685
    },
    {
      "epoch": 0.1975604629982797,
      "grad_norm": 3.121022939682007,
      "learning_rate": 8.446928106717359e-07,
      "loss": 0.7357,
      "step": 12690
    },
    {
      "epoch": 0.1976383040002491,
      "grad_norm": 10.636984825134277,
      "learning_rate": 8.44610871667131e-07,
      "loss": 0.8506,
      "step": 12695
    },
    {
      "epoch": 0.19771614500221846,
      "grad_norm": 5.662878513336182,
      "learning_rate": 8.44528932662526e-07,
      "loss": 0.8426,
      "step": 12700
    },
    {
      "epoch": 0.19779398600418785,
      "grad_norm": 3.862569808959961,
      "learning_rate": 8.44446993657921e-07,
      "loss": 0.8391,
      "step": 12705
    },
    {
      "epoch": 0.1978718270061572,
      "grad_norm": 5.017756938934326,
      "learning_rate": 8.44365054653316e-07,
      "loss": 0.8664,
      "step": 12710
    },
    {
      "epoch": 0.1979496680081266,
      "grad_norm": 3.679381847381592,
      "learning_rate": 8.44283115648711e-07,
      "loss": 0.95,
      "step": 12715
    },
    {
      "epoch": 0.19802750901009597,
      "grad_norm": 3.268360137939453,
      "learning_rate": 8.44201176644106e-07,
      "loss": 0.7672,
      "step": 12720
    },
    {
      "epoch": 0.19810535001206536,
      "grad_norm": 5.429895401000977,
      "learning_rate": 8.441192376395011e-07,
      "loss": 0.7644,
      "step": 12725
    },
    {
      "epoch": 0.19818319101403473,
      "grad_norm": 5.481542587280273,
      "learning_rate": 8.440372986348962e-07,
      "loss": 0.8911,
      "step": 12730
    },
    {
      "epoch": 0.19826103201600412,
      "grad_norm": 3.77721905708313,
      "learning_rate": 8.439553596302911e-07,
      "loss": 0.7636,
      "step": 12735
    },
    {
      "epoch": 0.19833887301797348,
      "grad_norm": 3.9851577281951904,
      "learning_rate": 8.438734206256862e-07,
      "loss": 0.9312,
      "step": 12740
    },
    {
      "epoch": 0.19841671401994287,
      "grad_norm": 3.65083909034729,
      "learning_rate": 8.437914816210813e-07,
      "loss": 0.7529,
      "step": 12745
    },
    {
      "epoch": 0.19849455502191224,
      "grad_norm": 3.0100839138031006,
      "learning_rate": 8.437095426164764e-07,
      "loss": 0.7147,
      "step": 12750
    },
    {
      "epoch": 0.19857239602388163,
      "grad_norm": 3.7276885509490967,
      "learning_rate": 8.436276036118712e-07,
      "loss": 0.7436,
      "step": 12755
    },
    {
      "epoch": 0.198650237025851,
      "grad_norm": 3.0928821563720703,
      "learning_rate": 8.435456646072663e-07,
      "loss": 0.7429,
      "step": 12760
    },
    {
      "epoch": 0.1987280780278204,
      "grad_norm": 3.2721216678619385,
      "learning_rate": 8.434637256026614e-07,
      "loss": 0.7477,
      "step": 12765
    },
    {
      "epoch": 0.19880591902978975,
      "grad_norm": 4.112125396728516,
      "learning_rate": 8.433817865980563e-07,
      "loss": 0.7392,
      "step": 12770
    },
    {
      "epoch": 0.19888376003175912,
      "grad_norm": 3.3520290851593018,
      "learning_rate": 8.432998475934514e-07,
      "loss": 0.8243,
      "step": 12775
    },
    {
      "epoch": 0.1989616010337285,
      "grad_norm": 2.7560410499572754,
      "learning_rate": 8.432179085888465e-07,
      "loss": 0.8496,
      "step": 12780
    },
    {
      "epoch": 0.19903944203569787,
      "grad_norm": 5.738790988922119,
      "learning_rate": 8.431359695842415e-07,
      "loss": 0.8122,
      "step": 12785
    },
    {
      "epoch": 0.19911728303766726,
      "grad_norm": 3.3104143142700195,
      "learning_rate": 8.430540305796365e-07,
      "loss": 0.8499,
      "step": 12790
    },
    {
      "epoch": 0.19919512403963663,
      "grad_norm": 3.488959312438965,
      "learning_rate": 8.429720915750315e-07,
      "loss": 0.7974,
      "step": 12795
    },
    {
      "epoch": 0.19927296504160602,
      "grad_norm": 5.885444164276123,
      "learning_rate": 8.428901525704265e-07,
      "loss": 0.8403,
      "step": 12800
    },
    {
      "epoch": 0.19935080604357538,
      "grad_norm": 7.075294017791748,
      "learning_rate": 8.428082135658215e-07,
      "loss": 0.9154,
      "step": 12805
    },
    {
      "epoch": 0.19942864704554478,
      "grad_norm": 7.824799060821533,
      "learning_rate": 8.427262745612166e-07,
      "loss": 0.8342,
      "step": 12810
    },
    {
      "epoch": 0.19950648804751414,
      "grad_norm": 3.2951488494873047,
      "learning_rate": 8.426443355566116e-07,
      "loss": 0.8083,
      "step": 12815
    },
    {
      "epoch": 0.19958432904948353,
      "grad_norm": 4.159008026123047,
      "learning_rate": 8.425623965520067e-07,
      "loss": 0.9696,
      "step": 12820
    },
    {
      "epoch": 0.1996621700514529,
      "grad_norm": 2.9382009506225586,
      "learning_rate": 8.424804575474017e-07,
      "loss": 0.7681,
      "step": 12825
    },
    {
      "epoch": 0.1997400110534223,
      "grad_norm": 4.053783893585205,
      "learning_rate": 8.423985185427967e-07,
      "loss": 0.8329,
      "step": 12830
    },
    {
      "epoch": 0.19981785205539165,
      "grad_norm": 5.080565452575684,
      "learning_rate": 8.423165795381917e-07,
      "loss": 0.777,
      "step": 12835
    },
    {
      "epoch": 0.19989569305736105,
      "grad_norm": 2.973886251449585,
      "learning_rate": 8.422346405335868e-07,
      "loss": 0.8252,
      "step": 12840
    },
    {
      "epoch": 0.1999735340593304,
      "grad_norm": 6.690211296081543,
      "learning_rate": 8.421527015289817e-07,
      "loss": 0.7814,
      "step": 12845
    },
    {
      "epoch": 0.2000513750612998,
      "grad_norm": 3.9737579822540283,
      "learning_rate": 8.420707625243768e-07,
      "loss": 0.8165,
      "step": 12850
    },
    {
      "epoch": 0.20012921606326917,
      "grad_norm": 4.720776081085205,
      "learning_rate": 8.419888235197719e-07,
      "loss": 0.8016,
      "step": 12855
    },
    {
      "epoch": 0.20020705706523853,
      "grad_norm": 3.9128286838531494,
      "learning_rate": 8.419068845151668e-07,
      "loss": 0.8222,
      "step": 12860
    },
    {
      "epoch": 0.20028489806720792,
      "grad_norm": 2.903087854385376,
      "learning_rate": 8.418249455105619e-07,
      "loss": 0.8552,
      "step": 12865
    },
    {
      "epoch": 0.2003627390691773,
      "grad_norm": 3.1490824222564697,
      "learning_rate": 8.41743006505957e-07,
      "loss": 0.7545,
      "step": 12870
    },
    {
      "epoch": 0.20044058007114668,
      "grad_norm": 7.775855541229248,
      "learning_rate": 8.41661067501352e-07,
      "loss": 0.8454,
      "step": 12875
    },
    {
      "epoch": 0.20051842107311604,
      "grad_norm": 4.137736797332764,
      "learning_rate": 8.415791284967469e-07,
      "loss": 0.8048,
      "step": 12880
    },
    {
      "epoch": 0.20059626207508544,
      "grad_norm": 2.8976640701293945,
      "learning_rate": 8.41497189492142e-07,
      "loss": 0.709,
      "step": 12885
    },
    {
      "epoch": 0.2006741030770548,
      "grad_norm": 3.4031612873077393,
      "learning_rate": 8.414152504875371e-07,
      "loss": 0.7075,
      "step": 12890
    },
    {
      "epoch": 0.2007519440790242,
      "grad_norm": 4.480353832244873,
      "learning_rate": 8.41333311482932e-07,
      "loss": 0.7674,
      "step": 12895
    },
    {
      "epoch": 0.20082978508099356,
      "grad_norm": 3.8333520889282227,
      "learning_rate": 8.412513724783271e-07,
      "loss": 0.7645,
      "step": 12900
    },
    {
      "epoch": 0.20090762608296295,
      "grad_norm": 3.1221156120300293,
      "learning_rate": 8.411694334737222e-07,
      "loss": 0.7816,
      "step": 12905
    },
    {
      "epoch": 0.2009854670849323,
      "grad_norm": 22.836925506591797,
      "learning_rate": 8.410874944691172e-07,
      "loss": 0.8316,
      "step": 12910
    },
    {
      "epoch": 0.2010633080869017,
      "grad_norm": 4.997188568115234,
      "learning_rate": 8.410055554645121e-07,
      "loss": 0.7576,
      "step": 12915
    },
    {
      "epoch": 0.20114114908887107,
      "grad_norm": 4.20601749420166,
      "learning_rate": 8.409236164599072e-07,
      "loss": 0.7546,
      "step": 12920
    },
    {
      "epoch": 0.20121899009084046,
      "grad_norm": 4.1795783042907715,
      "learning_rate": 8.408416774553022e-07,
      "loss": 0.845,
      "step": 12925
    },
    {
      "epoch": 0.20129683109280982,
      "grad_norm": 3.2515761852264404,
      "learning_rate": 8.407597384506973e-07,
      "loss": 0.7512,
      "step": 12930
    },
    {
      "epoch": 0.20137467209477922,
      "grad_norm": 5.808708190917969,
      "learning_rate": 8.406777994460923e-07,
      "loss": 0.7683,
      "step": 12935
    },
    {
      "epoch": 0.20145251309674858,
      "grad_norm": 2.7343006134033203,
      "learning_rate": 8.405958604414873e-07,
      "loss": 0.8085,
      "step": 12940
    },
    {
      "epoch": 0.20153035409871795,
      "grad_norm": 4.954883098602295,
      "learning_rate": 8.405139214368824e-07,
      "loss": 0.7211,
      "step": 12945
    },
    {
      "epoch": 0.20160819510068734,
      "grad_norm": 8.119803428649902,
      "learning_rate": 8.404319824322774e-07,
      "loss": 0.8099,
      "step": 12950
    },
    {
      "epoch": 0.2016860361026567,
      "grad_norm": 3.069868803024292,
      "learning_rate": 8.403500434276723e-07,
      "loss": 0.7915,
      "step": 12955
    },
    {
      "epoch": 0.2017638771046261,
      "grad_norm": 2.7691965103149414,
      "learning_rate": 8.402681044230674e-07,
      "loss": 0.8081,
      "step": 12960
    },
    {
      "epoch": 0.20184171810659546,
      "grad_norm": 2.716383457183838,
      "learning_rate": 8.401861654184625e-07,
      "loss": 0.8375,
      "step": 12965
    },
    {
      "epoch": 0.20191955910856485,
      "grad_norm": 4.081620693206787,
      "learning_rate": 8.401042264138574e-07,
      "loss": 0.8029,
      "step": 12970
    },
    {
      "epoch": 0.20199740011053421,
      "grad_norm": 4.608429908752441,
      "learning_rate": 8.400222874092525e-07,
      "loss": 0.9028,
      "step": 12975
    },
    {
      "epoch": 0.2020752411125036,
      "grad_norm": 2.8588662147521973,
      "learning_rate": 8.399403484046476e-07,
      "loss": 0.7334,
      "step": 12980
    },
    {
      "epoch": 0.20215308211447297,
      "grad_norm": 5.044338703155518,
      "learning_rate": 8.398584094000425e-07,
      "loss": 0.7045,
      "step": 12985
    },
    {
      "epoch": 0.20223092311644236,
      "grad_norm": 3.933218479156494,
      "learning_rate": 8.397764703954376e-07,
      "loss": 0.8562,
      "step": 12990
    },
    {
      "epoch": 0.20230876411841173,
      "grad_norm": 4.926372528076172,
      "learning_rate": 8.396945313908326e-07,
      "loss": 0.737,
      "step": 12995
    },
    {
      "epoch": 0.20238660512038112,
      "grad_norm": 3.583951711654663,
      "learning_rate": 8.396125923862277e-07,
      "loss": 0.7308,
      "step": 13000
    },
    {
      "epoch": 0.20246444612235048,
      "grad_norm": 3.6071455478668213,
      "learning_rate": 8.395306533816226e-07,
      "loss": 0.7894,
      "step": 13005
    },
    {
      "epoch": 0.20254228712431988,
      "grad_norm": 4.990063667297363,
      "learning_rate": 8.394487143770177e-07,
      "loss": 0.8604,
      "step": 13010
    },
    {
      "epoch": 0.20262012812628924,
      "grad_norm": 5.4818549156188965,
      "learning_rate": 8.393667753724128e-07,
      "loss": 0.6678,
      "step": 13015
    },
    {
      "epoch": 0.20269796912825863,
      "grad_norm": 7.279552936553955,
      "learning_rate": 8.392848363678078e-07,
      "loss": 0.8089,
      "step": 13020
    },
    {
      "epoch": 0.202775810130228,
      "grad_norm": 5.273457050323486,
      "learning_rate": 8.392028973632028e-07,
      "loss": 0.6855,
      "step": 13025
    },
    {
      "epoch": 0.20285365113219736,
      "grad_norm": 5.578176021575928,
      "learning_rate": 8.391209583585979e-07,
      "loss": 0.8303,
      "step": 13030
    },
    {
      "epoch": 0.20293149213416675,
      "grad_norm": 4.444499492645264,
      "learning_rate": 8.390390193539929e-07,
      "loss": 0.6954,
      "step": 13035
    },
    {
      "epoch": 0.20300933313613612,
      "grad_norm": 4.023461818695068,
      "learning_rate": 8.389570803493878e-07,
      "loss": 0.678,
      "step": 13040
    },
    {
      "epoch": 0.2030871741381055,
      "grad_norm": 2.834993839263916,
      "learning_rate": 8.388751413447829e-07,
      "loss": 0.7115,
      "step": 13045
    },
    {
      "epoch": 0.20316501514007487,
      "grad_norm": 4.737156391143799,
      "learning_rate": 8.387932023401779e-07,
      "loss": 0.7809,
      "step": 13050
    },
    {
      "epoch": 0.20324285614204426,
      "grad_norm": 12.347189903259277,
      "learning_rate": 8.38711263335573e-07,
      "loss": 0.7702,
      "step": 13055
    },
    {
      "epoch": 0.20332069714401363,
      "grad_norm": 4.668651103973389,
      "learning_rate": 8.38629324330968e-07,
      "loss": 0.7655,
      "step": 13060
    },
    {
      "epoch": 0.20339853814598302,
      "grad_norm": 3.7084391117095947,
      "learning_rate": 8.38547385326363e-07,
      "loss": 0.714,
      "step": 13065
    },
    {
      "epoch": 0.20347637914795239,
      "grad_norm": 2.783339262008667,
      "learning_rate": 8.384654463217581e-07,
      "loss": 0.6384,
      "step": 13070
    },
    {
      "epoch": 0.20355422014992178,
      "grad_norm": 3.5190024375915527,
      "learning_rate": 8.383835073171532e-07,
      "loss": 0.85,
      "step": 13075
    },
    {
      "epoch": 0.20363206115189114,
      "grad_norm": 3.285179615020752,
      "learning_rate": 8.38301568312548e-07,
      "loss": 0.7181,
      "step": 13080
    },
    {
      "epoch": 0.20370990215386053,
      "grad_norm": 5.458109378814697,
      "learning_rate": 8.382196293079431e-07,
      "loss": 0.9302,
      "step": 13085
    },
    {
      "epoch": 0.2037877431558299,
      "grad_norm": 4.708980083465576,
      "learning_rate": 8.381376903033382e-07,
      "loss": 0.8137,
      "step": 13090
    },
    {
      "epoch": 0.2038655841577993,
      "grad_norm": 3.8359594345092773,
      "learning_rate": 8.380557512987331e-07,
      "loss": 0.6734,
      "step": 13095
    },
    {
      "epoch": 0.20394342515976865,
      "grad_norm": 3.476069688796997,
      "learning_rate": 8.379738122941282e-07,
      "loss": 0.6435,
      "step": 13100
    },
    {
      "epoch": 0.20402126616173805,
      "grad_norm": 4.068753242492676,
      "learning_rate": 8.378918732895233e-07,
      "loss": 0.7332,
      "step": 13105
    },
    {
      "epoch": 0.2040991071637074,
      "grad_norm": 5.591004371643066,
      "learning_rate": 8.378099342849184e-07,
      "loss": 0.7835,
      "step": 13110
    },
    {
      "epoch": 0.20417694816567677,
      "grad_norm": 4.07530403137207,
      "learning_rate": 8.377279952803133e-07,
      "loss": 0.7763,
      "step": 13115
    },
    {
      "epoch": 0.20425478916764617,
      "grad_norm": 6.400569915771484,
      "learning_rate": 8.376460562757083e-07,
      "loss": 0.7816,
      "step": 13120
    },
    {
      "epoch": 0.20433263016961553,
      "grad_norm": 4.189541339874268,
      "learning_rate": 8.375641172711034e-07,
      "loss": 0.8091,
      "step": 13125
    },
    {
      "epoch": 0.20441047117158492,
      "grad_norm": 4.928549289703369,
      "learning_rate": 8.374821782664983e-07,
      "loss": 0.7569,
      "step": 13130
    },
    {
      "epoch": 0.2044883121735543,
      "grad_norm": 3.725872039794922,
      "learning_rate": 8.374002392618934e-07,
      "loss": 0.8412,
      "step": 13135
    },
    {
      "epoch": 0.20456615317552368,
      "grad_norm": 4.936779975891113,
      "learning_rate": 8.373183002572885e-07,
      "loss": 0.7803,
      "step": 13140
    },
    {
      "epoch": 0.20464399417749304,
      "grad_norm": 3.692831039428711,
      "learning_rate": 8.372363612526835e-07,
      "loss": 0.7979,
      "step": 13145
    },
    {
      "epoch": 0.20472183517946244,
      "grad_norm": 5.788731098175049,
      "learning_rate": 8.371544222480785e-07,
      "loss": 0.7817,
      "step": 13150
    },
    {
      "epoch": 0.2047996761814318,
      "grad_norm": 5.3713154792785645,
      "learning_rate": 8.370724832434736e-07,
      "loss": 0.7444,
      "step": 13155
    },
    {
      "epoch": 0.2048775171834012,
      "grad_norm": 3.327475070953369,
      "learning_rate": 8.369905442388685e-07,
      "loss": 0.853,
      "step": 13160
    },
    {
      "epoch": 0.20495535818537056,
      "grad_norm": 4.377721309661865,
      "learning_rate": 8.369086052342635e-07,
      "loss": 0.7105,
      "step": 13165
    },
    {
      "epoch": 0.20503319918733995,
      "grad_norm": 2.7629923820495605,
      "learning_rate": 8.368266662296586e-07,
      "loss": 0.7283,
      "step": 13170
    },
    {
      "epoch": 0.2051110401893093,
      "grad_norm": 4.508754730224609,
      "learning_rate": 8.367447272250536e-07,
      "loss": 0.7238,
      "step": 13175
    },
    {
      "epoch": 0.2051888811912787,
      "grad_norm": 3.4627845287323,
      "learning_rate": 8.366627882204487e-07,
      "loss": 0.7426,
      "step": 13180
    },
    {
      "epoch": 0.20526672219324807,
      "grad_norm": 4.33838415145874,
      "learning_rate": 8.365808492158437e-07,
      "loss": 0.7616,
      "step": 13185
    },
    {
      "epoch": 0.20534456319521746,
      "grad_norm": 4.036031723022461,
      "learning_rate": 8.364989102112387e-07,
      "loss": 0.6372,
      "step": 13190
    },
    {
      "epoch": 0.20542240419718683,
      "grad_norm": 9.265192985534668,
      "learning_rate": 8.364169712066338e-07,
      "loss": 0.6713,
      "step": 13195
    },
    {
      "epoch": 0.20550024519915622,
      "grad_norm": 5.069167613983154,
      "learning_rate": 8.363350322020288e-07,
      "loss": 0.8708,
      "step": 13200
    },
    {
      "epoch": 0.20557808620112558,
      "grad_norm": 3.5143802165985107,
      "learning_rate": 8.362530931974237e-07,
      "loss": 0.8575,
      "step": 13205
    },
    {
      "epoch": 0.20565592720309495,
      "grad_norm": 4.60297966003418,
      "learning_rate": 8.361711541928188e-07,
      "loss": 0.8428,
      "step": 13210
    },
    {
      "epoch": 0.20573376820506434,
      "grad_norm": 4.301421642303467,
      "learning_rate": 8.360892151882139e-07,
      "loss": 0.7471,
      "step": 13215
    },
    {
      "epoch": 0.2058116092070337,
      "grad_norm": 7.8147125244140625,
      "learning_rate": 8.360072761836088e-07,
      "loss": 0.6705,
      "step": 13220
    },
    {
      "epoch": 0.2058894502090031,
      "grad_norm": 4.782034873962402,
      "learning_rate": 8.359253371790039e-07,
      "loss": 0.695,
      "step": 13225
    },
    {
      "epoch": 0.20596729121097246,
      "grad_norm": 4.737283229827881,
      "learning_rate": 8.35843398174399e-07,
      "loss": 0.8739,
      "step": 13230
    },
    {
      "epoch": 0.20604513221294185,
      "grad_norm": 4.9094014167785645,
      "learning_rate": 8.357614591697941e-07,
      "loss": 0.7827,
      "step": 13235
    },
    {
      "epoch": 0.20612297321491121,
      "grad_norm": 4.747406959533691,
      "learning_rate": 8.356795201651889e-07,
      "loss": 0.7785,
      "step": 13240
    },
    {
      "epoch": 0.2062008142168806,
      "grad_norm": 3.4002225399017334,
      "learning_rate": 8.35597581160584e-07,
      "loss": 0.8564,
      "step": 13245
    },
    {
      "epoch": 0.20627865521884997,
      "grad_norm": 5.391127109527588,
      "learning_rate": 8.355156421559791e-07,
      "loss": 0.7454,
      "step": 13250
    },
    {
      "epoch": 0.20635649622081936,
      "grad_norm": 4.001280784606934,
      "learning_rate": 8.35433703151374e-07,
      "loss": 0.7198,
      "step": 13255
    },
    {
      "epoch": 0.20643433722278873,
      "grad_norm": 3.865567207336426,
      "learning_rate": 8.353517641467691e-07,
      "loss": 0.8462,
      "step": 13260
    },
    {
      "epoch": 0.20651217822475812,
      "grad_norm": 5.007477283477783,
      "learning_rate": 8.352698251421642e-07,
      "loss": 0.9695,
      "step": 13265
    },
    {
      "epoch": 0.20659001922672748,
      "grad_norm": 4.396978378295898,
      "learning_rate": 8.351878861375592e-07,
      "loss": 0.8082,
      "step": 13270
    },
    {
      "epoch": 0.20666786022869688,
      "grad_norm": 3.5921192169189453,
      "learning_rate": 8.351059471329542e-07,
      "loss": 0.7191,
      "step": 13275
    },
    {
      "epoch": 0.20674570123066624,
      "grad_norm": 3.7623448371887207,
      "learning_rate": 8.350240081283492e-07,
      "loss": 0.7479,
      "step": 13280
    },
    {
      "epoch": 0.20682354223263563,
      "grad_norm": 6.096179008483887,
      "learning_rate": 8.349420691237442e-07,
      "loss": 0.7466,
      "step": 13285
    },
    {
      "epoch": 0.206901383234605,
      "grad_norm": 5.423510551452637,
      "learning_rate": 8.348601301191393e-07,
      "loss": 0.7874,
      "step": 13290
    },
    {
      "epoch": 0.20697922423657436,
      "grad_norm": 4.719470500946045,
      "learning_rate": 8.347781911145343e-07,
      "loss": 0.7438,
      "step": 13295
    },
    {
      "epoch": 0.20705706523854375,
      "grad_norm": 3.9418418407440186,
      "learning_rate": 8.346962521099293e-07,
      "loss": 0.7436,
      "step": 13300
    },
    {
      "epoch": 0.20713490624051312,
      "grad_norm": 3.420037269592285,
      "learning_rate": 8.346143131053244e-07,
      "loss": 0.8426,
      "step": 13305
    },
    {
      "epoch": 0.2072127472424825,
      "grad_norm": 3.9989964962005615,
      "learning_rate": 8.345323741007194e-07,
      "loss": 0.8473,
      "step": 13310
    },
    {
      "epoch": 0.20729058824445187,
      "grad_norm": 12.799483299255371,
      "learning_rate": 8.344504350961144e-07,
      "loss": 0.719,
      "step": 13315
    },
    {
      "epoch": 0.20736842924642127,
      "grad_norm": 3.141099691390991,
      "learning_rate": 8.343684960915094e-07,
      "loss": 0.8719,
      "step": 13320
    },
    {
      "epoch": 0.20744627024839063,
      "grad_norm": 4.788917541503906,
      "learning_rate": 8.342865570869045e-07,
      "loss": 0.785,
      "step": 13325
    },
    {
      "epoch": 0.20752411125036002,
      "grad_norm": 3.2936322689056396,
      "learning_rate": 8.342046180822994e-07,
      "loss": 0.8711,
      "step": 13330
    },
    {
      "epoch": 0.20760195225232939,
      "grad_norm": 5.100861072540283,
      "learning_rate": 8.341226790776945e-07,
      "loss": 0.79,
      "step": 13335
    },
    {
      "epoch": 0.20767979325429878,
      "grad_norm": 2.7465577125549316,
      "learning_rate": 8.340407400730896e-07,
      "loss": 0.6895,
      "step": 13340
    },
    {
      "epoch": 0.20775763425626814,
      "grad_norm": 5.790363788604736,
      "learning_rate": 8.339588010684845e-07,
      "loss": 0.7492,
      "step": 13345
    },
    {
      "epoch": 0.20783547525823753,
      "grad_norm": 3.6077206134796143,
      "learning_rate": 8.338768620638796e-07,
      "loss": 0.6855,
      "step": 13350
    },
    {
      "epoch": 0.2079133162602069,
      "grad_norm": 3.6149632930755615,
      "learning_rate": 8.337949230592747e-07,
      "loss": 0.7914,
      "step": 13355
    },
    {
      "epoch": 0.2079911572621763,
      "grad_norm": 4.330005168914795,
      "learning_rate": 8.337129840546698e-07,
      "loss": 0.7572,
      "step": 13360
    },
    {
      "epoch": 0.20806899826414565,
      "grad_norm": 6.381075859069824,
      "learning_rate": 8.336310450500646e-07,
      "loss": 0.8347,
      "step": 13365
    },
    {
      "epoch": 0.20814683926611505,
      "grad_norm": 3.4584124088287354,
      "learning_rate": 8.335491060454597e-07,
      "loss": 0.7925,
      "step": 13370
    },
    {
      "epoch": 0.2082246802680844,
      "grad_norm": 5.580089092254639,
      "learning_rate": 8.334671670408548e-07,
      "loss": 0.7925,
      "step": 13375
    },
    {
      "epoch": 0.20830252127005378,
      "grad_norm": 3.6964595317840576,
      "learning_rate": 8.333852280362498e-07,
      "loss": 0.7562,
      "step": 13380
    },
    {
      "epoch": 0.20838036227202317,
      "grad_norm": 3.2514872550964355,
      "learning_rate": 8.333032890316448e-07,
      "loss": 0.7701,
      "step": 13385
    },
    {
      "epoch": 0.20845820327399253,
      "grad_norm": 4.424211502075195,
      "learning_rate": 8.332213500270399e-07,
      "loss": 0.7148,
      "step": 13390
    },
    {
      "epoch": 0.20853604427596192,
      "grad_norm": 5.740350723266602,
      "learning_rate": 8.331394110224349e-07,
      "loss": 0.8141,
      "step": 13395
    },
    {
      "epoch": 0.2086138852779313,
      "grad_norm": 3.7232847213745117,
      "learning_rate": 8.3305747201783e-07,
      "loss": 0.8316,
      "step": 13400
    },
    {
      "epoch": 0.20869172627990068,
      "grad_norm": 4.571991443634033,
      "learning_rate": 8.329755330132249e-07,
      "loss": 0.6605,
      "step": 13405
    },
    {
      "epoch": 0.20876956728187004,
      "grad_norm": 3.058192729949951,
      "learning_rate": 8.328935940086199e-07,
      "loss": 0.8583,
      "step": 13410
    },
    {
      "epoch": 0.20884740828383944,
      "grad_norm": 3.834956645965576,
      "learning_rate": 8.32811655004015e-07,
      "loss": 0.7717,
      "step": 13415
    },
    {
      "epoch": 0.2089252492858088,
      "grad_norm": 2.910698652267456,
      "learning_rate": 8.3272971599941e-07,
      "loss": 0.8361,
      "step": 13420
    },
    {
      "epoch": 0.2090030902877782,
      "grad_norm": 4.002933025360107,
      "learning_rate": 8.32647776994805e-07,
      "loss": 0.9264,
      "step": 13425
    },
    {
      "epoch": 0.20908093128974756,
      "grad_norm": 4.509486198425293,
      "learning_rate": 8.325658379902001e-07,
      "loss": 0.7,
      "step": 13430
    },
    {
      "epoch": 0.20915877229171695,
      "grad_norm": 3.453099250793457,
      "learning_rate": 8.324838989855952e-07,
      "loss": 0.8373,
      "step": 13435
    },
    {
      "epoch": 0.2092366132936863,
      "grad_norm": 5.16492223739624,
      "learning_rate": 8.324019599809901e-07,
      "loss": 0.8035,
      "step": 13440
    },
    {
      "epoch": 0.2093144542956557,
      "grad_norm": 5.247521877288818,
      "learning_rate": 8.323200209763851e-07,
      "loss": 0.7963,
      "step": 13445
    },
    {
      "epoch": 0.20939229529762507,
      "grad_norm": 2.5298125743865967,
      "learning_rate": 8.322380819717802e-07,
      "loss": 0.8136,
      "step": 13450
    },
    {
      "epoch": 0.20947013629959446,
      "grad_norm": 4.630289077758789,
      "learning_rate": 8.321561429671751e-07,
      "loss": 0.8526,
      "step": 13455
    },
    {
      "epoch": 0.20954797730156383,
      "grad_norm": 7.223544597625732,
      "learning_rate": 8.320742039625702e-07,
      "loss": 0.8431,
      "step": 13460
    },
    {
      "epoch": 0.2096258183035332,
      "grad_norm": 4.768827438354492,
      "learning_rate": 8.319922649579653e-07,
      "loss": 0.9167,
      "step": 13465
    },
    {
      "epoch": 0.20970365930550258,
      "grad_norm": 4.539663314819336,
      "learning_rate": 8.319103259533603e-07,
      "loss": 0.8486,
      "step": 13470
    },
    {
      "epoch": 0.20978150030747195,
      "grad_norm": 4.0133795738220215,
      "learning_rate": 8.318283869487553e-07,
      "loss": 0.8473,
      "step": 13475
    },
    {
      "epoch": 0.20985934130944134,
      "grad_norm": 3.0814623832702637,
      "learning_rate": 8.317464479441504e-07,
      "loss": 0.704,
      "step": 13480
    },
    {
      "epoch": 0.2099371823114107,
      "grad_norm": 4.812824726104736,
      "learning_rate": 8.316645089395453e-07,
      "loss": 0.7252,
      "step": 13485
    },
    {
      "epoch": 0.2100150233133801,
      "grad_norm": 11.381315231323242,
      "learning_rate": 8.315825699349403e-07,
      "loss": 0.7837,
      "step": 13490
    },
    {
      "epoch": 0.21009286431534946,
      "grad_norm": 4.615209579467773,
      "learning_rate": 8.315006309303354e-07,
      "loss": 0.8026,
      "step": 13495
    },
    {
      "epoch": 0.21017070531731885,
      "grad_norm": 4.024731159210205,
      "learning_rate": 8.314186919257305e-07,
      "loss": 0.8345,
      "step": 13500
    },
    {
      "epoch": 0.21024854631928822,
      "grad_norm": 5.9903764724731445,
      "learning_rate": 8.313367529211255e-07,
      "loss": 0.7786,
      "step": 13505
    },
    {
      "epoch": 0.2103263873212576,
      "grad_norm": 5.442336082458496,
      "learning_rate": 8.312548139165205e-07,
      "loss": 0.7564,
      "step": 13510
    },
    {
      "epoch": 0.21040422832322697,
      "grad_norm": 3.8722071647644043,
      "learning_rate": 8.311728749119156e-07,
      "loss": 0.8054,
      "step": 13515
    },
    {
      "epoch": 0.21048206932519636,
      "grad_norm": 6.410277843475342,
      "learning_rate": 8.310909359073106e-07,
      "loss": 0.7405,
      "step": 13520
    },
    {
      "epoch": 0.21055991032716573,
      "grad_norm": 4.545468807220459,
      "learning_rate": 8.310089969027055e-07,
      "loss": 0.6794,
      "step": 13525
    },
    {
      "epoch": 0.21063775132913512,
      "grad_norm": 5.3065667152404785,
      "learning_rate": 8.309270578981006e-07,
      "loss": 0.8487,
      "step": 13530
    },
    {
      "epoch": 0.21071559233110448,
      "grad_norm": 3.9609336853027344,
      "learning_rate": 8.308451188934956e-07,
      "loss": 0.7859,
      "step": 13535
    },
    {
      "epoch": 0.21079343333307388,
      "grad_norm": 8.753162384033203,
      "learning_rate": 8.307631798888907e-07,
      "loss": 0.7646,
      "step": 13540
    },
    {
      "epoch": 0.21087127433504324,
      "grad_norm": 5.301315784454346,
      "learning_rate": 8.306812408842857e-07,
      "loss": 0.6595,
      "step": 13545
    },
    {
      "epoch": 0.2109491153370126,
      "grad_norm": 3.3421101570129395,
      "learning_rate": 8.305993018796807e-07,
      "loss": 0.7312,
      "step": 13550
    },
    {
      "epoch": 0.211026956338982,
      "grad_norm": 6.578088283538818,
      "learning_rate": 8.305173628750758e-07,
      "loss": 0.8821,
      "step": 13555
    },
    {
      "epoch": 0.21110479734095136,
      "grad_norm": 4.213285446166992,
      "learning_rate": 8.304354238704709e-07,
      "loss": 0.8494,
      "step": 13560
    },
    {
      "epoch": 0.21118263834292075,
      "grad_norm": 3.34028959274292,
      "learning_rate": 8.303534848658657e-07,
      "loss": 0.802,
      "step": 13565
    },
    {
      "epoch": 0.21126047934489012,
      "grad_norm": 6.317659378051758,
      "learning_rate": 8.302715458612608e-07,
      "loss": 0.7412,
      "step": 13570
    },
    {
      "epoch": 0.2113383203468595,
      "grad_norm": 3.612773895263672,
      "learning_rate": 8.301896068566559e-07,
      "loss": 0.772,
      "step": 13575
    },
    {
      "epoch": 0.21141616134882887,
      "grad_norm": 6.666327476501465,
      "learning_rate": 8.301076678520508e-07,
      "loss": 0.7994,
      "step": 13580
    },
    {
      "epoch": 0.21149400235079827,
      "grad_norm": 3.3673384189605713,
      "learning_rate": 8.300257288474459e-07,
      "loss": 0.8642,
      "step": 13585
    },
    {
      "epoch": 0.21157184335276763,
      "grad_norm": 4.446502685546875,
      "learning_rate": 8.29943789842841e-07,
      "loss": 0.8119,
      "step": 13590
    },
    {
      "epoch": 0.21164968435473702,
      "grad_norm": 2.946279525756836,
      "learning_rate": 8.29861850838236e-07,
      "loss": 0.7246,
      "step": 13595
    },
    {
      "epoch": 0.2117275253567064,
      "grad_norm": 3.9437286853790283,
      "learning_rate": 8.29779911833631e-07,
      "loss": 0.7658,
      "step": 13600
    },
    {
      "epoch": 0.21180536635867578,
      "grad_norm": 5.029818534851074,
      "learning_rate": 8.29697972829026e-07,
      "loss": 0.8914,
      "step": 13605
    },
    {
      "epoch": 0.21188320736064514,
      "grad_norm": 3.175124406814575,
      "learning_rate": 8.29616033824421e-07,
      "loss": 0.7365,
      "step": 13610
    },
    {
      "epoch": 0.21196104836261453,
      "grad_norm": 4.092532634735107,
      "learning_rate": 8.29534094819816e-07,
      "loss": 0.8208,
      "step": 13615
    },
    {
      "epoch": 0.2120388893645839,
      "grad_norm": 4.064992427825928,
      "learning_rate": 8.294521558152111e-07,
      "loss": 0.7227,
      "step": 13620
    },
    {
      "epoch": 0.2121167303665533,
      "grad_norm": 4.1546244621276855,
      "learning_rate": 8.293702168106062e-07,
      "loss": 0.8453,
      "step": 13625
    },
    {
      "epoch": 0.21219457136852266,
      "grad_norm": 2.7323529720306396,
      "learning_rate": 8.292882778060012e-07,
      "loss": 0.7896,
      "step": 13630
    },
    {
      "epoch": 0.21227241237049202,
      "grad_norm": 3.339479446411133,
      "learning_rate": 8.292063388013962e-07,
      "loss": 0.7644,
      "step": 13635
    },
    {
      "epoch": 0.2123502533724614,
      "grad_norm": 5.718183994293213,
      "learning_rate": 8.291243997967913e-07,
      "loss": 0.6686,
      "step": 13640
    },
    {
      "epoch": 0.21242809437443078,
      "grad_norm": 3.4797685146331787,
      "learning_rate": 8.290424607921862e-07,
      "loss": 0.6722,
      "step": 13645
    },
    {
      "epoch": 0.21250593537640017,
      "grad_norm": 3.628688335418701,
      "learning_rate": 8.289605217875813e-07,
      "loss": 0.8625,
      "step": 13650
    },
    {
      "epoch": 0.21258377637836953,
      "grad_norm": 5.237287998199463,
      "learning_rate": 8.288785827829763e-07,
      "loss": 0.92,
      "step": 13655
    },
    {
      "epoch": 0.21266161738033892,
      "grad_norm": 3.847942590713501,
      "learning_rate": 8.287966437783713e-07,
      "loss": 0.7552,
      "step": 13660
    },
    {
      "epoch": 0.2127394583823083,
      "grad_norm": 3.620185613632202,
      "learning_rate": 8.287147047737664e-07,
      "loss": 0.7319,
      "step": 13665
    },
    {
      "epoch": 0.21281729938427768,
      "grad_norm": 2.827895402908325,
      "learning_rate": 8.286327657691615e-07,
      "loss": 0.8578,
      "step": 13670
    },
    {
      "epoch": 0.21289514038624704,
      "grad_norm": 4.379232406616211,
      "learning_rate": 8.285508267645564e-07,
      "loss": 0.7956,
      "step": 13675
    },
    {
      "epoch": 0.21297298138821644,
      "grad_norm": 5.4567036628723145,
      "learning_rate": 8.284688877599515e-07,
      "loss": 0.7348,
      "step": 13680
    },
    {
      "epoch": 0.2130508223901858,
      "grad_norm": 3.210332155227661,
      "learning_rate": 8.283869487553466e-07,
      "loss": 0.7783,
      "step": 13685
    },
    {
      "epoch": 0.2131286633921552,
      "grad_norm": 3.489346981048584,
      "learning_rate": 8.283050097507414e-07,
      "loss": 0.6617,
      "step": 13690
    },
    {
      "epoch": 0.21320650439412456,
      "grad_norm": 5.191878318786621,
      "learning_rate": 8.282230707461365e-07,
      "loss": 0.7675,
      "step": 13695
    },
    {
      "epoch": 0.21328434539609395,
      "grad_norm": 3.5417792797088623,
      "learning_rate": 8.281411317415316e-07,
      "loss": 0.8721,
      "step": 13700
    },
    {
      "epoch": 0.2133621863980633,
      "grad_norm": 3.657003164291382,
      "learning_rate": 8.280591927369266e-07,
      "loss": 0.8504,
      "step": 13705
    },
    {
      "epoch": 0.2134400274000327,
      "grad_norm": 3.983947515487671,
      "learning_rate": 8.279772537323216e-07,
      "loss": 0.6321,
      "step": 13710
    },
    {
      "epoch": 0.21351786840200207,
      "grad_norm": 5.221202373504639,
      "learning_rate": 8.278953147277167e-07,
      "loss": 0.7542,
      "step": 13715
    },
    {
      "epoch": 0.21359570940397143,
      "grad_norm": 3.3143723011016846,
      "learning_rate": 8.278133757231117e-07,
      "loss": 0.7038,
      "step": 13720
    },
    {
      "epoch": 0.21367355040594083,
      "grad_norm": 3.3426454067230225,
      "learning_rate": 8.277314367185067e-07,
      "loss": 0.8252,
      "step": 13725
    },
    {
      "epoch": 0.2137513914079102,
      "grad_norm": 4.690375328063965,
      "learning_rate": 8.276494977139017e-07,
      "loss": 0.9626,
      "step": 13730
    },
    {
      "epoch": 0.21382923240987958,
      "grad_norm": 5.993745803833008,
      "learning_rate": 8.275675587092967e-07,
      "loss": 0.81,
      "step": 13735
    },
    {
      "epoch": 0.21390707341184895,
      "grad_norm": 5.37796688079834,
      "learning_rate": 8.274856197046918e-07,
      "loss": 0.8553,
      "step": 13740
    },
    {
      "epoch": 0.21398491441381834,
      "grad_norm": 3.129485607147217,
      "learning_rate": 8.274036807000868e-07,
      "loss": 0.6703,
      "step": 13745
    },
    {
      "epoch": 0.2140627554157877,
      "grad_norm": 3.359469413757324,
      "learning_rate": 8.273217416954819e-07,
      "loss": 0.7105,
      "step": 13750
    },
    {
      "epoch": 0.2141405964177571,
      "grad_norm": 5.8597869873046875,
      "learning_rate": 8.272398026908769e-07,
      "loss": 0.9019,
      "step": 13755
    },
    {
      "epoch": 0.21421843741972646,
      "grad_norm": 2.56128191947937,
      "learning_rate": 8.27157863686272e-07,
      "loss": 0.7213,
      "step": 13760
    },
    {
      "epoch": 0.21429627842169585,
      "grad_norm": 3.979757070541382,
      "learning_rate": 8.27075924681667e-07,
      "loss": 0.8644,
      "step": 13765
    },
    {
      "epoch": 0.21437411942366522,
      "grad_norm": 4.351536273956299,
      "learning_rate": 8.269939856770619e-07,
      "loss": 0.8617,
      "step": 13770
    },
    {
      "epoch": 0.2144519604256346,
      "grad_norm": 4.314666748046875,
      "learning_rate": 8.26912046672457e-07,
      "loss": 0.7729,
      "step": 13775
    },
    {
      "epoch": 0.21452980142760397,
      "grad_norm": 4.594080448150635,
      "learning_rate": 8.26830107667852e-07,
      "loss": 0.7507,
      "step": 13780
    },
    {
      "epoch": 0.21460764242957336,
      "grad_norm": 3.6102042198181152,
      "learning_rate": 8.26748168663247e-07,
      "loss": 0.8858,
      "step": 13785
    },
    {
      "epoch": 0.21468548343154273,
      "grad_norm": 5.565709114074707,
      "learning_rate": 8.266662296586421e-07,
      "loss": 0.7261,
      "step": 13790
    },
    {
      "epoch": 0.21476332443351212,
      "grad_norm": 4.304652690887451,
      "learning_rate": 8.265842906540372e-07,
      "loss": 0.7998,
      "step": 13795
    },
    {
      "epoch": 0.21484116543548148,
      "grad_norm": 4.141101360321045,
      "learning_rate": 8.265023516494321e-07,
      "loss": 0.7796,
      "step": 13800
    },
    {
      "epoch": 0.21491900643745085,
      "grad_norm": 3.0917515754699707,
      "learning_rate": 8.264204126448272e-07,
      "loss": 0.7534,
      "step": 13805
    },
    {
      "epoch": 0.21499684743942024,
      "grad_norm": 3.7405171394348145,
      "learning_rate": 8.263384736402222e-07,
      "loss": 0.8425,
      "step": 13810
    },
    {
      "epoch": 0.2150746884413896,
      "grad_norm": 8.8234224319458,
      "learning_rate": 8.262565346356171e-07,
      "loss": 0.7727,
      "step": 13815
    },
    {
      "epoch": 0.215152529443359,
      "grad_norm": 3.842259645462036,
      "learning_rate": 8.261745956310122e-07,
      "loss": 0.7739,
      "step": 13820
    },
    {
      "epoch": 0.21523037044532836,
      "grad_norm": 6.93010950088501,
      "learning_rate": 8.260926566264073e-07,
      "loss": 0.9363,
      "step": 13825
    },
    {
      "epoch": 0.21530821144729775,
      "grad_norm": 4.047563552856445,
      "learning_rate": 8.260107176218023e-07,
      "loss": 0.8669,
      "step": 13830
    },
    {
      "epoch": 0.21538605244926712,
      "grad_norm": 3.978311777114868,
      "learning_rate": 8.259287786171973e-07,
      "loss": 0.9023,
      "step": 13835
    },
    {
      "epoch": 0.2154638934512365,
      "grad_norm": 5.298313140869141,
      "learning_rate": 8.258468396125924e-07,
      "loss": 0.7525,
      "step": 13840
    },
    {
      "epoch": 0.21554173445320587,
      "grad_norm": 3.4469170570373535,
      "learning_rate": 8.257649006079874e-07,
      "loss": 0.793,
      "step": 13845
    },
    {
      "epoch": 0.21561957545517527,
      "grad_norm": 5.635530948638916,
      "learning_rate": 8.256829616033823e-07,
      "loss": 0.7775,
      "step": 13850
    },
    {
      "epoch": 0.21569741645714463,
      "grad_norm": 5.665985584259033,
      "learning_rate": 8.256010225987774e-07,
      "loss": 0.7704,
      "step": 13855
    },
    {
      "epoch": 0.21577525745911402,
      "grad_norm": 6.2332024574279785,
      "learning_rate": 8.255190835941724e-07,
      "loss": 0.8562,
      "step": 13860
    },
    {
      "epoch": 0.2158530984610834,
      "grad_norm": 3.517777681350708,
      "learning_rate": 8.254371445895675e-07,
      "loss": 0.7324,
      "step": 13865
    },
    {
      "epoch": 0.21593093946305278,
      "grad_norm": 6.235748767852783,
      "learning_rate": 8.253552055849625e-07,
      "loss": 0.7729,
      "step": 13870
    },
    {
      "epoch": 0.21600878046502214,
      "grad_norm": 5.898641109466553,
      "learning_rate": 8.252732665803576e-07,
      "loss": 0.7918,
      "step": 13875
    },
    {
      "epoch": 0.21608662146699154,
      "grad_norm": 3.069035530090332,
      "learning_rate": 8.251913275757526e-07,
      "loss": 0.6672,
      "step": 13880
    },
    {
      "epoch": 0.2161644624689609,
      "grad_norm": 5.234879493713379,
      "learning_rate": 8.251093885711477e-07,
      "loss": 0.8629,
      "step": 13885
    },
    {
      "epoch": 0.2162423034709303,
      "grad_norm": 7.040566921234131,
      "learning_rate": 8.250274495665426e-07,
      "loss": 0.7679,
      "step": 13890
    },
    {
      "epoch": 0.21632014447289966,
      "grad_norm": 2.981306552886963,
      "learning_rate": 8.249455105619376e-07,
      "loss": 0.8243,
      "step": 13895
    },
    {
      "epoch": 0.21639798547486902,
      "grad_norm": 9.416054725646973,
      "learning_rate": 8.248635715573327e-07,
      "loss": 0.7234,
      "step": 13900
    },
    {
      "epoch": 0.2164758264768384,
      "grad_norm": 4.333786964416504,
      "learning_rate": 8.247816325527277e-07,
      "loss": 0.7538,
      "step": 13905
    },
    {
      "epoch": 0.21655366747880778,
      "grad_norm": 5.313504695892334,
      "learning_rate": 8.246996935481227e-07,
      "loss": 0.9174,
      "step": 13910
    },
    {
      "epoch": 0.21663150848077717,
      "grad_norm": 2.9732582569122314,
      "learning_rate": 8.246177545435178e-07,
      "loss": 0.7441,
      "step": 13915
    },
    {
      "epoch": 0.21670934948274653,
      "grad_norm": 2.9341235160827637,
      "learning_rate": 8.245358155389129e-07,
      "loss": 0.8753,
      "step": 13920
    },
    {
      "epoch": 0.21678719048471592,
      "grad_norm": 12.563216209411621,
      "learning_rate": 8.244538765343078e-07,
      "loss": 0.8747,
      "step": 13925
    },
    {
      "epoch": 0.2168650314866853,
      "grad_norm": 3.163398504257202,
      "learning_rate": 8.243719375297028e-07,
      "loss": 0.8394,
      "step": 13930
    },
    {
      "epoch": 0.21694287248865468,
      "grad_norm": 2.861363410949707,
      "learning_rate": 8.242899985250979e-07,
      "loss": 0.7728,
      "step": 13935
    },
    {
      "epoch": 0.21702071349062405,
      "grad_norm": 4.490225791931152,
      "learning_rate": 8.242080595204928e-07,
      "loss": 0.8157,
      "step": 13940
    },
    {
      "epoch": 0.21709855449259344,
      "grad_norm": 4.911387920379639,
      "learning_rate": 8.241261205158879e-07,
      "loss": 0.8017,
      "step": 13945
    },
    {
      "epoch": 0.2171763954945628,
      "grad_norm": 4.898251533508301,
      "learning_rate": 8.24044181511283e-07,
      "loss": 0.7364,
      "step": 13950
    },
    {
      "epoch": 0.2172542364965322,
      "grad_norm": 5.873484134674072,
      "learning_rate": 8.23962242506678e-07,
      "loss": 0.7311,
      "step": 13955
    },
    {
      "epoch": 0.21733207749850156,
      "grad_norm": 3.597964286804199,
      "learning_rate": 8.23880303502073e-07,
      "loss": 0.834,
      "step": 13960
    },
    {
      "epoch": 0.21740991850047095,
      "grad_norm": 2.7035915851593018,
      "learning_rate": 8.237983644974681e-07,
      "loss": 0.6948,
      "step": 13965
    },
    {
      "epoch": 0.21748775950244031,
      "grad_norm": 5.127129554748535,
      "learning_rate": 8.23716425492863e-07,
      "loss": 0.6872,
      "step": 13970
    },
    {
      "epoch": 0.2175656005044097,
      "grad_norm": 6.334181785583496,
      "learning_rate": 8.23634486488258e-07,
      "loss": 0.6572,
      "step": 13975
    },
    {
      "epoch": 0.21764344150637907,
      "grad_norm": 4.3845744132995605,
      "learning_rate": 8.235525474836531e-07,
      "loss": 0.8049,
      "step": 13980
    },
    {
      "epoch": 0.21772128250834843,
      "grad_norm": 3.1326639652252197,
      "learning_rate": 8.234706084790481e-07,
      "loss": 0.7638,
      "step": 13985
    },
    {
      "epoch": 0.21779912351031783,
      "grad_norm": 2.6377274990081787,
      "learning_rate": 8.233886694744432e-07,
      "loss": 0.6976,
      "step": 13990
    },
    {
      "epoch": 0.2178769645122872,
      "grad_norm": 3.378631353378296,
      "learning_rate": 8.233067304698382e-07,
      "loss": 0.7974,
      "step": 13995
    },
    {
      "epoch": 0.21795480551425658,
      "grad_norm": 3.709468364715576,
      "learning_rate": 8.232247914652333e-07,
      "loss": 0.7535,
      "step": 14000
    },
    {
      "epoch": 0.21803264651622595,
      "grad_norm": 5.473926067352295,
      "learning_rate": 8.231428524606283e-07,
      "loss": 0.8026,
      "step": 14005
    },
    {
      "epoch": 0.21811048751819534,
      "grad_norm": 3.7275307178497314,
      "learning_rate": 8.230609134560234e-07,
      "loss": 0.7762,
      "step": 14010
    },
    {
      "epoch": 0.2181883285201647,
      "grad_norm": 8.003002166748047,
      "learning_rate": 8.229789744514183e-07,
      "loss": 0.8362,
      "step": 14015
    },
    {
      "epoch": 0.2182661695221341,
      "grad_norm": 2.683030605316162,
      "learning_rate": 8.228970354468133e-07,
      "loss": 0.7537,
      "step": 14020
    },
    {
      "epoch": 0.21834401052410346,
      "grad_norm": 23.4306583404541,
      "learning_rate": 8.228150964422084e-07,
      "loss": 0.8595,
      "step": 14025
    },
    {
      "epoch": 0.21842185152607285,
      "grad_norm": 4.96809720993042,
      "learning_rate": 8.227331574376035e-07,
      "loss": 0.7552,
      "step": 14030
    },
    {
      "epoch": 0.21849969252804222,
      "grad_norm": 5.7253098487854,
      "learning_rate": 8.226512184329984e-07,
      "loss": 0.7839,
      "step": 14035
    },
    {
      "epoch": 0.2185775335300116,
      "grad_norm": 4.622411251068115,
      "learning_rate": 8.225692794283935e-07,
      "loss": 0.8006,
      "step": 14040
    },
    {
      "epoch": 0.21865537453198097,
      "grad_norm": 4.409444332122803,
      "learning_rate": 8.224873404237886e-07,
      "loss": 0.8242,
      "step": 14045
    },
    {
      "epoch": 0.21873321553395036,
      "grad_norm": 3.5355610847473145,
      "learning_rate": 8.224054014191835e-07,
      "loss": 0.8879,
      "step": 14050
    },
    {
      "epoch": 0.21881105653591973,
      "grad_norm": 7.076291084289551,
      "learning_rate": 8.223234624145785e-07,
      "loss": 0.8007,
      "step": 14055
    },
    {
      "epoch": 0.21888889753788912,
      "grad_norm": 2.4964358806610107,
      "learning_rate": 8.222415234099736e-07,
      "loss": 0.7088,
      "step": 14060
    },
    {
      "epoch": 0.21896673853985849,
      "grad_norm": 3.8193085193634033,
      "learning_rate": 8.221595844053686e-07,
      "loss": 0.779,
      "step": 14065
    },
    {
      "epoch": 0.21904457954182785,
      "grad_norm": 4.243593692779541,
      "learning_rate": 8.220776454007636e-07,
      "loss": 0.7483,
      "step": 14070
    },
    {
      "epoch": 0.21912242054379724,
      "grad_norm": 3.811239719390869,
      "learning_rate": 8.219957063961587e-07,
      "loss": 0.8161,
      "step": 14075
    },
    {
      "epoch": 0.2192002615457666,
      "grad_norm": 7.058951377868652,
      "learning_rate": 8.219137673915537e-07,
      "loss": 0.856,
      "step": 14080
    },
    {
      "epoch": 0.219278102547736,
      "grad_norm": 4.006752014160156,
      "learning_rate": 8.218318283869487e-07,
      "loss": 0.8221,
      "step": 14085
    },
    {
      "epoch": 0.21935594354970536,
      "grad_norm": 4.000454425811768,
      "learning_rate": 8.217498893823438e-07,
      "loss": 0.7499,
      "step": 14090
    },
    {
      "epoch": 0.21943378455167475,
      "grad_norm": 4.066524982452393,
      "learning_rate": 8.216679503777387e-07,
      "loss": 0.7449,
      "step": 14095
    },
    {
      "epoch": 0.21951162555364412,
      "grad_norm": 8.29305362701416,
      "learning_rate": 8.215860113731338e-07,
      "loss": 0.7864,
      "step": 14100
    },
    {
      "epoch": 0.2195894665556135,
      "grad_norm": 5.086592674255371,
      "learning_rate": 8.215040723685288e-07,
      "loss": 0.904,
      "step": 14105
    },
    {
      "epoch": 0.21966730755758287,
      "grad_norm": 5.60768985748291,
      "learning_rate": 8.214221333639238e-07,
      "loss": 0.6991,
      "step": 14110
    },
    {
      "epoch": 0.21974514855955227,
      "grad_norm": 6.108041286468506,
      "learning_rate": 8.213401943593189e-07,
      "loss": 0.8331,
      "step": 14115
    },
    {
      "epoch": 0.21982298956152163,
      "grad_norm": 3.089353561401367,
      "learning_rate": 8.21258255354714e-07,
      "loss": 0.8687,
      "step": 14120
    },
    {
      "epoch": 0.21990083056349102,
      "grad_norm": 4.127028465270996,
      "learning_rate": 8.21176316350109e-07,
      "loss": 0.7391,
      "step": 14125
    },
    {
      "epoch": 0.2199786715654604,
      "grad_norm": 7.309013843536377,
      "learning_rate": 8.21094377345504e-07,
      "loss": 0.8205,
      "step": 14130
    },
    {
      "epoch": 0.22005651256742978,
      "grad_norm": 5.600733757019043,
      "learning_rate": 8.21012438340899e-07,
      "loss": 0.7669,
      "step": 14135
    },
    {
      "epoch": 0.22013435356939914,
      "grad_norm": 5.287996768951416,
      "learning_rate": 8.20930499336294e-07,
      "loss": 0.7142,
      "step": 14140
    },
    {
      "epoch": 0.22021219457136854,
      "grad_norm": 3.8828279972076416,
      "learning_rate": 8.20848560331689e-07,
      "loss": 0.8409,
      "step": 14145
    },
    {
      "epoch": 0.2202900355733379,
      "grad_norm": 4.1061224937438965,
      "learning_rate": 8.207666213270841e-07,
      "loss": 0.6962,
      "step": 14150
    },
    {
      "epoch": 0.22036787657530726,
      "grad_norm": 2.734174966812134,
      "learning_rate": 8.206846823224792e-07,
      "loss": 0.7433,
      "step": 14155
    },
    {
      "epoch": 0.22044571757727666,
      "grad_norm": 2.9642739295959473,
      "learning_rate": 8.206027433178741e-07,
      "loss": 0.8107,
      "step": 14160
    },
    {
      "epoch": 0.22052355857924602,
      "grad_norm": 5.745486736297607,
      "learning_rate": 8.205208043132692e-07,
      "loss": 0.8115,
      "step": 14165
    },
    {
      "epoch": 0.2206013995812154,
      "grad_norm": 4.336205005645752,
      "learning_rate": 8.204388653086643e-07,
      "loss": 0.8461,
      "step": 14170
    },
    {
      "epoch": 0.22067924058318478,
      "grad_norm": 2.4944536685943604,
      "learning_rate": 8.203569263040591e-07,
      "loss": 0.6971,
      "step": 14175
    },
    {
      "epoch": 0.22075708158515417,
      "grad_norm": 4.578847408294678,
      "learning_rate": 8.202749872994542e-07,
      "loss": 0.6928,
      "step": 14180
    },
    {
      "epoch": 0.22083492258712353,
      "grad_norm": 2.760460138320923,
      "learning_rate": 8.201930482948493e-07,
      "loss": 0.7466,
      "step": 14185
    },
    {
      "epoch": 0.22091276358909293,
      "grad_norm": 3.805088520050049,
      "learning_rate": 8.201111092902443e-07,
      "loss": 0.8223,
      "step": 14190
    },
    {
      "epoch": 0.2209906045910623,
      "grad_norm": 3.4560000896453857,
      "learning_rate": 8.200291702856393e-07,
      "loss": 0.8931,
      "step": 14195
    },
    {
      "epoch": 0.22106844559303168,
      "grad_norm": 7.692502498626709,
      "learning_rate": 8.199472312810344e-07,
      "loss": 0.677,
      "step": 14200
    },
    {
      "epoch": 0.22114628659500105,
      "grad_norm": 2.7867279052734375,
      "learning_rate": 8.198652922764294e-07,
      "loss": 0.7933,
      "step": 14205
    },
    {
      "epoch": 0.22122412759697044,
      "grad_norm": 3.4029629230499268,
      "learning_rate": 8.197833532718245e-07,
      "loss": 0.7878,
      "step": 14210
    },
    {
      "epoch": 0.2213019685989398,
      "grad_norm": 6.6466546058654785,
      "learning_rate": 8.197014142672194e-07,
      "loss": 0.6714,
      "step": 14215
    },
    {
      "epoch": 0.2213798096009092,
      "grad_norm": 3.9622604846954346,
      "learning_rate": 8.196194752626144e-07,
      "loss": 0.6719,
      "step": 14220
    },
    {
      "epoch": 0.22145765060287856,
      "grad_norm": 3.521365165710449,
      "learning_rate": 8.195375362580095e-07,
      "loss": 0.8028,
      "step": 14225
    },
    {
      "epoch": 0.22153549160484795,
      "grad_norm": 3.510684013366699,
      "learning_rate": 8.194555972534045e-07,
      "loss": 0.7864,
      "step": 14230
    },
    {
      "epoch": 0.22161333260681731,
      "grad_norm": 5.596205711364746,
      "learning_rate": 8.193736582487995e-07,
      "loss": 0.8561,
      "step": 14235
    },
    {
      "epoch": 0.22169117360878668,
      "grad_norm": 3.478048086166382,
      "learning_rate": 8.192917192441946e-07,
      "loss": 0.7405,
      "step": 14240
    },
    {
      "epoch": 0.22176901461075607,
      "grad_norm": 2.7496485710144043,
      "learning_rate": 8.192097802395897e-07,
      "loss": 0.7695,
      "step": 14245
    },
    {
      "epoch": 0.22184685561272544,
      "grad_norm": 5.8107781410217285,
      "learning_rate": 8.191278412349847e-07,
      "loss": 0.8052,
      "step": 14250
    },
    {
      "epoch": 0.22192469661469483,
      "grad_norm": 3.761655330657959,
      "learning_rate": 8.190459022303796e-07,
      "loss": 0.7027,
      "step": 14255
    },
    {
      "epoch": 0.2220025376166642,
      "grad_norm": 3.4367716312408447,
      "learning_rate": 8.189639632257747e-07,
      "loss": 0.801,
      "step": 14260
    },
    {
      "epoch": 0.22208037861863358,
      "grad_norm": 3.767960548400879,
      "learning_rate": 8.188820242211697e-07,
      "loss": 0.7308,
      "step": 14265
    },
    {
      "epoch": 0.22215821962060295,
      "grad_norm": 4.997518062591553,
      "learning_rate": 8.188000852165647e-07,
      "loss": 0.8017,
      "step": 14270
    },
    {
      "epoch": 0.22223606062257234,
      "grad_norm": 2.965698480606079,
      "learning_rate": 8.187181462119598e-07,
      "loss": 0.8322,
      "step": 14275
    },
    {
      "epoch": 0.2223139016245417,
      "grad_norm": 4.3722662925720215,
      "learning_rate": 8.186362072073549e-07,
      "loss": 0.8415,
      "step": 14280
    },
    {
      "epoch": 0.2223917426265111,
      "grad_norm": 3.201127767562866,
      "learning_rate": 8.185542682027498e-07,
      "loss": 0.8765,
      "step": 14285
    },
    {
      "epoch": 0.22246958362848046,
      "grad_norm": 7.47006368637085,
      "learning_rate": 8.184723291981449e-07,
      "loss": 0.7547,
      "step": 14290
    },
    {
      "epoch": 0.22254742463044985,
      "grad_norm": 4.898451805114746,
      "learning_rate": 8.183903901935399e-07,
      "loss": 0.7302,
      "step": 14295
    },
    {
      "epoch": 0.22262526563241922,
      "grad_norm": 4.226808547973633,
      "learning_rate": 8.183084511889348e-07,
      "loss": 0.7815,
      "step": 14300
    },
    {
      "epoch": 0.2227031066343886,
      "grad_norm": 3.9173691272735596,
      "learning_rate": 8.182265121843299e-07,
      "loss": 0.8184,
      "step": 14305
    },
    {
      "epoch": 0.22278094763635797,
      "grad_norm": 4.693234443664551,
      "learning_rate": 8.18144573179725e-07,
      "loss": 0.7193,
      "step": 14310
    },
    {
      "epoch": 0.22285878863832737,
      "grad_norm": 3.06361985206604,
      "learning_rate": 8.1806263417512e-07,
      "loss": 0.727,
      "step": 14315
    },
    {
      "epoch": 0.22293662964029673,
      "grad_norm": 3.4421865940093994,
      "learning_rate": 8.17980695170515e-07,
      "loss": 0.6528,
      "step": 14320
    },
    {
      "epoch": 0.2230144706422661,
      "grad_norm": 3.3549883365631104,
      "learning_rate": 8.178987561659101e-07,
      "loss": 0.8015,
      "step": 14325
    },
    {
      "epoch": 0.22309231164423549,
      "grad_norm": 5.654025077819824,
      "learning_rate": 8.178168171613051e-07,
      "loss": 0.7657,
      "step": 14330
    },
    {
      "epoch": 0.22317015264620485,
      "grad_norm": 5.917110919952393,
      "learning_rate": 8.177348781567002e-07,
      "loss": 0.7654,
      "step": 14335
    },
    {
      "epoch": 0.22324799364817424,
      "grad_norm": 6.301812171936035,
      "learning_rate": 8.176529391520951e-07,
      "loss": 0.7356,
      "step": 14340
    },
    {
      "epoch": 0.2233258346501436,
      "grad_norm": 7.683788776397705,
      "learning_rate": 8.175710001474901e-07,
      "loss": 0.7445,
      "step": 14345
    },
    {
      "epoch": 0.223403675652113,
      "grad_norm": 3.3383662700653076,
      "learning_rate": 8.174890611428852e-07,
      "loss": 0.8202,
      "step": 14350
    },
    {
      "epoch": 0.22348151665408236,
      "grad_norm": 4.314518928527832,
      "learning_rate": 8.174071221382802e-07,
      "loss": 0.8356,
      "step": 14355
    },
    {
      "epoch": 0.22355935765605175,
      "grad_norm": 2.8232765197753906,
      "learning_rate": 8.173251831336752e-07,
      "loss": 0.7911,
      "step": 14360
    },
    {
      "epoch": 0.22363719865802112,
      "grad_norm": 6.986234188079834,
      "learning_rate": 8.172432441290703e-07,
      "loss": 0.7589,
      "step": 14365
    },
    {
      "epoch": 0.2237150396599905,
      "grad_norm": 3.733344316482544,
      "learning_rate": 8.171613051244654e-07,
      "loss": 0.7412,
      "step": 14370
    },
    {
      "epoch": 0.22379288066195988,
      "grad_norm": 3.7331125736236572,
      "learning_rate": 8.170793661198604e-07,
      "loss": 0.8827,
      "step": 14375
    },
    {
      "epoch": 0.22387072166392927,
      "grad_norm": 2.895965099334717,
      "learning_rate": 8.169974271152553e-07,
      "loss": 0.8259,
      "step": 14380
    },
    {
      "epoch": 0.22394856266589863,
      "grad_norm": 3.6977014541625977,
      "learning_rate": 8.169154881106504e-07,
      "loss": 0.8569,
      "step": 14385
    },
    {
      "epoch": 0.22402640366786802,
      "grad_norm": 6.653420925140381,
      "learning_rate": 8.168335491060455e-07,
      "loss": 0.841,
      "step": 14390
    },
    {
      "epoch": 0.2241042446698374,
      "grad_norm": 3.430891275405884,
      "learning_rate": 8.167516101014404e-07,
      "loss": 0.7575,
      "step": 14395
    },
    {
      "epoch": 0.22418208567180678,
      "grad_norm": 3.0250182151794434,
      "learning_rate": 8.166696710968355e-07,
      "loss": 0.8719,
      "step": 14400
    },
    {
      "epoch": 0.22425992667377614,
      "grad_norm": 2.6156387329101562,
      "learning_rate": 8.165877320922306e-07,
      "loss": 0.6721,
      "step": 14405
    },
    {
      "epoch": 0.2243377676757455,
      "grad_norm": 5.948540210723877,
      "learning_rate": 8.165057930876255e-07,
      "loss": 0.8922,
      "step": 14410
    },
    {
      "epoch": 0.2244156086777149,
      "grad_norm": 3.9800314903259277,
      "learning_rate": 8.164238540830206e-07,
      "loss": 0.8119,
      "step": 14415
    },
    {
      "epoch": 0.22449344967968426,
      "grad_norm": 4.236239433288574,
      "learning_rate": 8.163419150784156e-07,
      "loss": 0.8216,
      "step": 14420
    },
    {
      "epoch": 0.22457129068165366,
      "grad_norm": 3.8411571979522705,
      "learning_rate": 8.162599760738106e-07,
      "loss": 0.7578,
      "step": 14425
    },
    {
      "epoch": 0.22464913168362302,
      "grad_norm": 4.027297496795654,
      "learning_rate": 8.161780370692056e-07,
      "loss": 0.7417,
      "step": 14430
    },
    {
      "epoch": 0.2247269726855924,
      "grad_norm": 2.982266902923584,
      "learning_rate": 8.160960980646007e-07,
      "loss": 0.6673,
      "step": 14435
    },
    {
      "epoch": 0.22480481368756178,
      "grad_norm": 7.821840763092041,
      "learning_rate": 8.160141590599957e-07,
      "loss": 0.7697,
      "step": 14440
    },
    {
      "epoch": 0.22488265468953117,
      "grad_norm": 4.410890102386475,
      "learning_rate": 8.159322200553907e-07,
      "loss": 0.7914,
      "step": 14445
    },
    {
      "epoch": 0.22496049569150053,
      "grad_norm": 5.73233699798584,
      "learning_rate": 8.158502810507858e-07,
      "loss": 0.6921,
      "step": 14450
    },
    {
      "epoch": 0.22503833669346993,
      "grad_norm": 3.975536346435547,
      "learning_rate": 8.157683420461808e-07,
      "loss": 0.7207,
      "step": 14455
    },
    {
      "epoch": 0.2251161776954393,
      "grad_norm": 3.909505605697632,
      "learning_rate": 8.156864030415758e-07,
      "loss": 0.7957,
      "step": 14460
    },
    {
      "epoch": 0.22519401869740868,
      "grad_norm": 3.489048480987549,
      "learning_rate": 8.156044640369708e-07,
      "loss": 0.8026,
      "step": 14465
    },
    {
      "epoch": 0.22527185969937805,
      "grad_norm": 3.770564317703247,
      "learning_rate": 8.155225250323658e-07,
      "loss": 0.787,
      "step": 14470
    },
    {
      "epoch": 0.22534970070134744,
      "grad_norm": 3.319371223449707,
      "learning_rate": 8.154405860277609e-07,
      "loss": 0.7437,
      "step": 14475
    },
    {
      "epoch": 0.2254275417033168,
      "grad_norm": 3.6019351482391357,
      "learning_rate": 8.15358647023156e-07,
      "loss": 0.6795,
      "step": 14480
    },
    {
      "epoch": 0.2255053827052862,
      "grad_norm": 3.6138195991516113,
      "learning_rate": 8.152767080185509e-07,
      "loss": 0.9009,
      "step": 14485
    },
    {
      "epoch": 0.22558322370725556,
      "grad_norm": 2.5489044189453125,
      "learning_rate": 8.15194769013946e-07,
      "loss": 0.8335,
      "step": 14490
    },
    {
      "epoch": 0.22566106470922492,
      "grad_norm": 3.9875316619873047,
      "learning_rate": 8.151128300093411e-07,
      "loss": 0.8167,
      "step": 14495
    },
    {
      "epoch": 0.22573890571119432,
      "grad_norm": 3.159693956375122,
      "learning_rate": 8.150308910047359e-07,
      "loss": 0.7998,
      "step": 14500
    },
    {
      "epoch": 0.22581674671316368,
      "grad_norm": 6.574869632720947,
      "learning_rate": 8.14948952000131e-07,
      "loss": 0.9038,
      "step": 14505
    },
    {
      "epoch": 0.22589458771513307,
      "grad_norm": 5.537595748901367,
      "learning_rate": 8.148670129955261e-07,
      "loss": 0.7544,
      "step": 14510
    },
    {
      "epoch": 0.22597242871710244,
      "grad_norm": 7.5179057121276855,
      "learning_rate": 8.147850739909212e-07,
      "loss": 0.7809,
      "step": 14515
    },
    {
      "epoch": 0.22605026971907183,
      "grad_norm": 7.547605037689209,
      "learning_rate": 8.147031349863161e-07,
      "loss": 0.7712,
      "step": 14520
    },
    {
      "epoch": 0.2261281107210412,
      "grad_norm": 3.386322498321533,
      "learning_rate": 8.146211959817112e-07,
      "loss": 0.7648,
      "step": 14525
    },
    {
      "epoch": 0.22620595172301058,
      "grad_norm": 3.413363218307495,
      "learning_rate": 8.145392569771063e-07,
      "loss": 0.7465,
      "step": 14530
    },
    {
      "epoch": 0.22628379272497995,
      "grad_norm": 4.657474040985107,
      "learning_rate": 8.144573179725012e-07,
      "loss": 0.755,
      "step": 14535
    },
    {
      "epoch": 0.22636163372694934,
      "grad_norm": 3.3240861892700195,
      "learning_rate": 8.143753789678962e-07,
      "loss": 0.7667,
      "step": 14540
    },
    {
      "epoch": 0.2264394747289187,
      "grad_norm": 6.471132755279541,
      "learning_rate": 8.142934399632913e-07,
      "loss": 0.7321,
      "step": 14545
    },
    {
      "epoch": 0.2265173157308881,
      "grad_norm": 6.002511024475098,
      "learning_rate": 8.142115009586863e-07,
      "loss": 0.8867,
      "step": 14550
    },
    {
      "epoch": 0.22659515673285746,
      "grad_norm": 4.303719997406006,
      "learning_rate": 8.141295619540813e-07,
      "loss": 0.6478,
      "step": 14555
    },
    {
      "epoch": 0.22667299773482685,
      "grad_norm": 4.221592426300049,
      "learning_rate": 8.140476229494764e-07,
      "loss": 0.7632,
      "step": 14560
    },
    {
      "epoch": 0.22675083873679622,
      "grad_norm": 6.414244174957275,
      "learning_rate": 8.139656839448714e-07,
      "loss": 0.6771,
      "step": 14565
    },
    {
      "epoch": 0.2268286797387656,
      "grad_norm": 3.563746690750122,
      "learning_rate": 8.138837449402665e-07,
      "loss": 0.826,
      "step": 14570
    },
    {
      "epoch": 0.22690652074073497,
      "grad_norm": 3.3471519947052,
      "learning_rate": 8.138018059356615e-07,
      "loss": 0.8127,
      "step": 14575
    },
    {
      "epoch": 0.22698436174270437,
      "grad_norm": 3.464322328567505,
      "learning_rate": 8.137198669310564e-07,
      "loss": 0.851,
      "step": 14580
    },
    {
      "epoch": 0.22706220274467373,
      "grad_norm": 2.899655342102051,
      "learning_rate": 8.136379279264515e-07,
      "loss": 0.8432,
      "step": 14585
    },
    {
      "epoch": 0.2271400437466431,
      "grad_norm": 3.654853343963623,
      "learning_rate": 8.135559889218465e-07,
      "loss": 0.8792,
      "step": 14590
    },
    {
      "epoch": 0.2272178847486125,
      "grad_norm": 3.7926645278930664,
      "learning_rate": 8.134740499172415e-07,
      "loss": 0.8283,
      "step": 14595
    },
    {
      "epoch": 0.22729572575058185,
      "grad_norm": 2.3791191577911377,
      "learning_rate": 8.133921109126366e-07,
      "loss": 0.5844,
      "step": 14600
    },
    {
      "epoch": 0.22737356675255124,
      "grad_norm": 4.844578266143799,
      "learning_rate": 8.133101719080317e-07,
      "loss": 0.746,
      "step": 14605
    },
    {
      "epoch": 0.2274514077545206,
      "grad_norm": 5.1527934074401855,
      "learning_rate": 8.132282329034266e-07,
      "loss": 0.7452,
      "step": 14610
    },
    {
      "epoch": 0.22752924875649,
      "grad_norm": 2.648463010787964,
      "learning_rate": 8.131462938988217e-07,
      "loss": 0.6665,
      "step": 14615
    },
    {
      "epoch": 0.22760708975845936,
      "grad_norm": 4.308215618133545,
      "learning_rate": 8.130643548942167e-07,
      "loss": 0.814,
      "step": 14620
    },
    {
      "epoch": 0.22768493076042876,
      "grad_norm": 4.505756378173828,
      "learning_rate": 8.129824158896116e-07,
      "loss": 0.7675,
      "step": 14625
    },
    {
      "epoch": 0.22776277176239812,
      "grad_norm": 4.09812593460083,
      "learning_rate": 8.129004768850067e-07,
      "loss": 0.8119,
      "step": 14630
    },
    {
      "epoch": 0.2278406127643675,
      "grad_norm": 5.634159088134766,
      "learning_rate": 8.128185378804018e-07,
      "loss": 0.7209,
      "step": 14635
    },
    {
      "epoch": 0.22791845376633688,
      "grad_norm": 5.862931728363037,
      "learning_rate": 8.127365988757969e-07,
      "loss": 0.8805,
      "step": 14640
    },
    {
      "epoch": 0.22799629476830627,
      "grad_norm": 3.8466310501098633,
      "learning_rate": 8.126546598711918e-07,
      "loss": 0.7862,
      "step": 14645
    },
    {
      "epoch": 0.22807413577027563,
      "grad_norm": 2.5441253185272217,
      "learning_rate": 8.125727208665869e-07,
      "loss": 0.7283,
      "step": 14650
    },
    {
      "epoch": 0.22815197677224502,
      "grad_norm": 2.762737274169922,
      "learning_rate": 8.12490781861982e-07,
      "loss": 0.7527,
      "step": 14655
    },
    {
      "epoch": 0.2282298177742144,
      "grad_norm": 3.7195687294006348,
      "learning_rate": 8.12408842857377e-07,
      "loss": 0.8323,
      "step": 14660
    },
    {
      "epoch": 0.22830765877618378,
      "grad_norm": 4.145010471343994,
      "learning_rate": 8.123269038527719e-07,
      "loss": 0.936,
      "step": 14665
    },
    {
      "epoch": 0.22838549977815314,
      "grad_norm": 3.6851210594177246,
      "learning_rate": 8.12244964848167e-07,
      "loss": 0.8665,
      "step": 14670
    },
    {
      "epoch": 0.2284633407801225,
      "grad_norm": 3.9991884231567383,
      "learning_rate": 8.12163025843562e-07,
      "loss": 0.7744,
      "step": 14675
    },
    {
      "epoch": 0.2285411817820919,
      "grad_norm": 4.037894248962402,
      "learning_rate": 8.12081086838957e-07,
      "loss": 0.8248,
      "step": 14680
    },
    {
      "epoch": 0.22861902278406127,
      "grad_norm": 3.2940948009490967,
      "learning_rate": 8.119991478343521e-07,
      "loss": 0.7561,
      "step": 14685
    },
    {
      "epoch": 0.22869686378603066,
      "grad_norm": 5.218342304229736,
      "learning_rate": 8.119172088297471e-07,
      "loss": 0.838,
      "step": 14690
    },
    {
      "epoch": 0.22877470478800002,
      "grad_norm": 4.6300578117370605,
      "learning_rate": 8.118352698251422e-07,
      "loss": 0.7774,
      "step": 14695
    },
    {
      "epoch": 0.2288525457899694,
      "grad_norm": 6.564635753631592,
      "learning_rate": 8.117533308205372e-07,
      "loss": 0.7365,
      "step": 14700
    },
    {
      "epoch": 0.22893038679193878,
      "grad_norm": 7.371841907501221,
      "learning_rate": 8.116713918159321e-07,
      "loss": 0.7821,
      "step": 14705
    },
    {
      "epoch": 0.22900822779390817,
      "grad_norm": 4.6451096534729,
      "learning_rate": 8.115894528113272e-07,
      "loss": 0.7552,
      "step": 14710
    },
    {
      "epoch": 0.22908606879587753,
      "grad_norm": 3.867570161819458,
      "learning_rate": 8.115075138067222e-07,
      "loss": 0.6779,
      "step": 14715
    },
    {
      "epoch": 0.22916390979784693,
      "grad_norm": 7.25125789642334,
      "learning_rate": 8.114255748021172e-07,
      "loss": 0.7304,
      "step": 14720
    },
    {
      "epoch": 0.2292417507998163,
      "grad_norm": 3.8599355220794678,
      "learning_rate": 8.113436357975123e-07,
      "loss": 0.7558,
      "step": 14725
    },
    {
      "epoch": 0.22931959180178568,
      "grad_norm": 6.007401466369629,
      "learning_rate": 8.112616967929074e-07,
      "loss": 0.7334,
      "step": 14730
    },
    {
      "epoch": 0.22939743280375505,
      "grad_norm": 6.439403057098389,
      "learning_rate": 8.111797577883023e-07,
      "loss": 0.8409,
      "step": 14735
    },
    {
      "epoch": 0.22947527380572444,
      "grad_norm": 5.950405597686768,
      "learning_rate": 8.110978187836974e-07,
      "loss": 0.7371,
      "step": 14740
    },
    {
      "epoch": 0.2295531148076938,
      "grad_norm": 4.082195281982422,
      "learning_rate": 8.110158797790924e-07,
      "loss": 0.877,
      "step": 14745
    },
    {
      "epoch": 0.2296309558096632,
      "grad_norm": 6.829986095428467,
      "learning_rate": 8.109339407744873e-07,
      "loss": 0.6617,
      "step": 14750
    },
    {
      "epoch": 0.22970879681163256,
      "grad_norm": 4.060400485992432,
      "learning_rate": 8.108520017698824e-07,
      "loss": 0.7444,
      "step": 14755
    },
    {
      "epoch": 0.22978663781360192,
      "grad_norm": 3.6605048179626465,
      "learning_rate": 8.107700627652775e-07,
      "loss": 0.8301,
      "step": 14760
    },
    {
      "epoch": 0.22986447881557132,
      "grad_norm": 5.560822010040283,
      "learning_rate": 8.106881237606726e-07,
      "loss": 0.7338,
      "step": 14765
    },
    {
      "epoch": 0.22994231981754068,
      "grad_norm": 3.1847989559173584,
      "learning_rate": 8.106061847560675e-07,
      "loss": 0.6357,
      "step": 14770
    },
    {
      "epoch": 0.23002016081951007,
      "grad_norm": 4.955717086791992,
      "learning_rate": 8.105242457514626e-07,
      "loss": 0.8697,
      "step": 14775
    },
    {
      "epoch": 0.23009800182147944,
      "grad_norm": 7.307408332824707,
      "learning_rate": 8.104423067468577e-07,
      "loss": 0.7504,
      "step": 14780
    },
    {
      "epoch": 0.23017584282344883,
      "grad_norm": 12.575167655944824,
      "learning_rate": 8.103603677422526e-07,
      "loss": 0.7651,
      "step": 14785
    },
    {
      "epoch": 0.2302536838254182,
      "grad_norm": 3.67960524559021,
      "learning_rate": 8.102784287376476e-07,
      "loss": 0.7182,
      "step": 14790
    },
    {
      "epoch": 0.23033152482738758,
      "grad_norm": 6.872060298919678,
      "learning_rate": 8.101964897330427e-07,
      "loss": 0.8097,
      "step": 14795
    },
    {
      "epoch": 0.23040936582935695,
      "grad_norm": 4.601255416870117,
      "learning_rate": 8.101145507284377e-07,
      "loss": 0.8425,
      "step": 14800
    },
    {
      "epoch": 0.23048720683132634,
      "grad_norm": 4.442498683929443,
      "learning_rate": 8.100326117238327e-07,
      "loss": 0.8298,
      "step": 14805
    },
    {
      "epoch": 0.2305650478332957,
      "grad_norm": 5.014193058013916,
      "learning_rate": 8.099506727192278e-07,
      "loss": 0.7692,
      "step": 14810
    },
    {
      "epoch": 0.2306428888352651,
      "grad_norm": 5.516265392303467,
      "learning_rate": 8.098687337146228e-07,
      "loss": 0.7929,
      "step": 14815
    },
    {
      "epoch": 0.23072072983723446,
      "grad_norm": 2.710554361343384,
      "learning_rate": 8.097867947100179e-07,
      "loss": 0.6834,
      "step": 14820
    },
    {
      "epoch": 0.23079857083920385,
      "grad_norm": 3.622087240219116,
      "learning_rate": 8.097048557054128e-07,
      "loss": 0.9235,
      "step": 14825
    },
    {
      "epoch": 0.23087641184117322,
      "grad_norm": 3.3825948238372803,
      "learning_rate": 8.096229167008078e-07,
      "loss": 0.8111,
      "step": 14830
    },
    {
      "epoch": 0.2309542528431426,
      "grad_norm": 3.4113945960998535,
      "learning_rate": 8.095409776962029e-07,
      "loss": 0.8036,
      "step": 14835
    },
    {
      "epoch": 0.23103209384511197,
      "grad_norm": 5.018237590789795,
      "learning_rate": 8.09459038691598e-07,
      "loss": 0.7452,
      "step": 14840
    },
    {
      "epoch": 0.23110993484708134,
      "grad_norm": 2.8325562477111816,
      "learning_rate": 8.093770996869929e-07,
      "loss": 0.7413,
      "step": 14845
    },
    {
      "epoch": 0.23118777584905073,
      "grad_norm": 7.24515962600708,
      "learning_rate": 8.09295160682388e-07,
      "loss": 0.7887,
      "step": 14850
    },
    {
      "epoch": 0.2312656168510201,
      "grad_norm": 2.9999563694000244,
      "learning_rate": 8.092132216777831e-07,
      "loss": 0.8002,
      "step": 14855
    },
    {
      "epoch": 0.2313434578529895,
      "grad_norm": 2.6894278526306152,
      "learning_rate": 8.09131282673178e-07,
      "loss": 0.9197,
      "step": 14860
    },
    {
      "epoch": 0.23142129885495885,
      "grad_norm": 4.730764865875244,
      "learning_rate": 8.09049343668573e-07,
      "loss": 0.7456,
      "step": 14865
    },
    {
      "epoch": 0.23149913985692824,
      "grad_norm": 2.9281725883483887,
      "learning_rate": 8.089674046639681e-07,
      "loss": 0.782,
      "step": 14870
    },
    {
      "epoch": 0.2315769808588976,
      "grad_norm": 3.086618185043335,
      "learning_rate": 8.088854656593631e-07,
      "loss": 0.7647,
      "step": 14875
    },
    {
      "epoch": 0.231654821860867,
      "grad_norm": 3.007323980331421,
      "learning_rate": 8.088035266547581e-07,
      "loss": 0.7483,
      "step": 14880
    },
    {
      "epoch": 0.23173266286283636,
      "grad_norm": 3.2050905227661133,
      "learning_rate": 8.087215876501532e-07,
      "loss": 0.7355,
      "step": 14885
    },
    {
      "epoch": 0.23181050386480576,
      "grad_norm": 4.09403657913208,
      "learning_rate": 8.086396486455483e-07,
      "loss": 0.7159,
      "step": 14890
    },
    {
      "epoch": 0.23188834486677512,
      "grad_norm": 3.1509900093078613,
      "learning_rate": 8.085577096409432e-07,
      "loss": 0.8726,
      "step": 14895
    },
    {
      "epoch": 0.2319661858687445,
      "grad_norm": 3.883364200592041,
      "learning_rate": 8.084757706363383e-07,
      "loss": 0.8345,
      "step": 14900
    },
    {
      "epoch": 0.23204402687071388,
      "grad_norm": 2.933501720428467,
      "learning_rate": 8.083938316317333e-07,
      "loss": 0.8286,
      "step": 14905
    },
    {
      "epoch": 0.23212186787268327,
      "grad_norm": 3.5567376613616943,
      "learning_rate": 8.083118926271283e-07,
      "loss": 0.863,
      "step": 14910
    },
    {
      "epoch": 0.23219970887465263,
      "grad_norm": 4.056962490081787,
      "learning_rate": 8.082299536225233e-07,
      "loss": 0.853,
      "step": 14915
    },
    {
      "epoch": 0.23227754987662202,
      "grad_norm": 3.4202661514282227,
      "learning_rate": 8.081480146179184e-07,
      "loss": 0.7843,
      "step": 14920
    },
    {
      "epoch": 0.2323553908785914,
      "grad_norm": 3.929119348526001,
      "learning_rate": 8.080660756133134e-07,
      "loss": 0.7301,
      "step": 14925
    },
    {
      "epoch": 0.23243323188056075,
      "grad_norm": 5.521277904510498,
      "learning_rate": 8.079841366087085e-07,
      "loss": 0.7049,
      "step": 14930
    },
    {
      "epoch": 0.23251107288253015,
      "grad_norm": 3.254838466644287,
      "learning_rate": 8.079021976041035e-07,
      "loss": 0.7977,
      "step": 14935
    },
    {
      "epoch": 0.2325889138844995,
      "grad_norm": 4.433169364929199,
      "learning_rate": 8.078202585994985e-07,
      "loss": 0.8519,
      "step": 14940
    },
    {
      "epoch": 0.2326667548864689,
      "grad_norm": 3.3750112056732178,
      "learning_rate": 8.077383195948935e-07,
      "loss": 0.804,
      "step": 14945
    },
    {
      "epoch": 0.23274459588843827,
      "grad_norm": 3.0588154792785645,
      "learning_rate": 8.076563805902885e-07,
      "loss": 0.7398,
      "step": 14950
    },
    {
      "epoch": 0.23282243689040766,
      "grad_norm": 2.8940608501434326,
      "learning_rate": 8.075744415856835e-07,
      "loss": 0.7417,
      "step": 14955
    },
    {
      "epoch": 0.23290027789237702,
      "grad_norm": 6.039758205413818,
      "learning_rate": 8.074925025810786e-07,
      "loss": 0.8312,
      "step": 14960
    },
    {
      "epoch": 0.23297811889434641,
      "grad_norm": 3.923211097717285,
      "learning_rate": 8.074105635764737e-07,
      "loss": 0.7279,
      "step": 14965
    },
    {
      "epoch": 0.23305595989631578,
      "grad_norm": 2.865115165710449,
      "learning_rate": 8.073286245718686e-07,
      "loss": 0.8598,
      "step": 14970
    },
    {
      "epoch": 0.23313380089828517,
      "grad_norm": 3.202042818069458,
      "learning_rate": 8.072466855672637e-07,
      "loss": 0.8332,
      "step": 14975
    },
    {
      "epoch": 0.23321164190025453,
      "grad_norm": 4.588990211486816,
      "learning_rate": 8.071647465626588e-07,
      "loss": 0.6646,
      "step": 14980
    },
    {
      "epoch": 0.23328948290222393,
      "grad_norm": 3.9389398097991943,
      "learning_rate": 8.070828075580538e-07,
      "loss": 0.8169,
      "step": 14985
    },
    {
      "epoch": 0.2333673239041933,
      "grad_norm": 5.094730854034424,
      "learning_rate": 8.070008685534487e-07,
      "loss": 0.7144,
      "step": 14990
    },
    {
      "epoch": 0.23344516490616268,
      "grad_norm": 4.285375595092773,
      "learning_rate": 8.069189295488438e-07,
      "loss": 0.7464,
      "step": 14995
    },
    {
      "epoch": 0.23352300590813205,
      "grad_norm": 3.9989163875579834,
      "learning_rate": 8.068369905442388e-07,
      "loss": 0.7893,
      "step": 15000
    },
    {
      "epoch": 0.23360084691010144,
      "grad_norm": 3.326927661895752,
      "learning_rate": 8.067550515396338e-07,
      "loss": 0.7341,
      "step": 15005
    },
    {
      "epoch": 0.2336786879120708,
      "grad_norm": 3.371975898742676,
      "learning_rate": 8.066731125350289e-07,
      "loss": 0.7949,
      "step": 15010
    },
    {
      "epoch": 0.23375652891404017,
      "grad_norm": 8.783902168273926,
      "learning_rate": 8.06591173530424e-07,
      "loss": 0.7686,
      "step": 15015
    },
    {
      "epoch": 0.23383436991600956,
      "grad_norm": 5.643832206726074,
      "learning_rate": 8.06509234525819e-07,
      "loss": 0.6986,
      "step": 15020
    },
    {
      "epoch": 0.23391221091797892,
      "grad_norm": 2.909189462661743,
      "learning_rate": 8.06427295521214e-07,
      "loss": 0.832,
      "step": 15025
    },
    {
      "epoch": 0.23399005191994832,
      "grad_norm": 3.0144805908203125,
      "learning_rate": 8.06345356516609e-07,
      "loss": 0.7332,
      "step": 15030
    },
    {
      "epoch": 0.23406789292191768,
      "grad_norm": 4.5122246742248535,
      "learning_rate": 8.06263417512004e-07,
      "loss": 0.7674,
      "step": 15035
    },
    {
      "epoch": 0.23414573392388707,
      "grad_norm": 3.568117141723633,
      "learning_rate": 8.06181478507399e-07,
      "loss": 0.7287,
      "step": 15040
    },
    {
      "epoch": 0.23422357492585644,
      "grad_norm": 4.4060845375061035,
      "learning_rate": 8.060995395027941e-07,
      "loss": 0.8157,
      "step": 15045
    },
    {
      "epoch": 0.23430141592782583,
      "grad_norm": 4.511560440063477,
      "learning_rate": 8.060176004981891e-07,
      "loss": 0.7997,
      "step": 15050
    },
    {
      "epoch": 0.2343792569297952,
      "grad_norm": 4.127213001251221,
      "learning_rate": 8.059356614935842e-07,
      "loss": 0.6869,
      "step": 15055
    },
    {
      "epoch": 0.23445709793176459,
      "grad_norm": 3.2110824584960938,
      "learning_rate": 8.058537224889792e-07,
      "loss": 0.8188,
      "step": 15060
    },
    {
      "epoch": 0.23453493893373395,
      "grad_norm": 6.899344444274902,
      "learning_rate": 8.057717834843742e-07,
      "loss": 0.7663,
      "step": 15065
    },
    {
      "epoch": 0.23461277993570334,
      "grad_norm": 6.0621538162231445,
      "learning_rate": 8.056898444797692e-07,
      "loss": 0.8239,
      "step": 15070
    },
    {
      "epoch": 0.2346906209376727,
      "grad_norm": 5.787536144256592,
      "learning_rate": 8.056079054751643e-07,
      "loss": 0.78,
      "step": 15075
    },
    {
      "epoch": 0.2347684619396421,
      "grad_norm": 3.954349994659424,
      "learning_rate": 8.055259664705592e-07,
      "loss": 0.8584,
      "step": 15080
    },
    {
      "epoch": 0.23484630294161146,
      "grad_norm": 4.174968719482422,
      "learning_rate": 8.054440274659543e-07,
      "loss": 0.7563,
      "step": 15085
    },
    {
      "epoch": 0.23492414394358085,
      "grad_norm": 4.891374111175537,
      "learning_rate": 8.053620884613494e-07,
      "loss": 0.859,
      "step": 15090
    },
    {
      "epoch": 0.23500198494555022,
      "grad_norm": 3.874782085418701,
      "learning_rate": 8.052801494567443e-07,
      "loss": 0.825,
      "step": 15095
    },
    {
      "epoch": 0.23507982594751958,
      "grad_norm": 6.848973274230957,
      "learning_rate": 8.051982104521394e-07,
      "loss": 0.7221,
      "step": 15100
    },
    {
      "epoch": 0.23515766694948897,
      "grad_norm": 8.270047187805176,
      "learning_rate": 8.051162714475345e-07,
      "loss": 0.7624,
      "step": 15105
    },
    {
      "epoch": 0.23523550795145834,
      "grad_norm": 4.251575946807861,
      "learning_rate": 8.050343324429294e-07,
      "loss": 0.7126,
      "step": 15110
    },
    {
      "epoch": 0.23531334895342773,
      "grad_norm": 3.7854721546173096,
      "learning_rate": 8.049523934383244e-07,
      "loss": 0.8051,
      "step": 15115
    },
    {
      "epoch": 0.2353911899553971,
      "grad_norm": 3.765505790710449,
      "learning_rate": 8.048704544337195e-07,
      "loss": 0.7562,
      "step": 15120
    },
    {
      "epoch": 0.2354690309573665,
      "grad_norm": 7.419190406799316,
      "learning_rate": 8.047885154291146e-07,
      "loss": 0.7952,
      "step": 15125
    },
    {
      "epoch": 0.23554687195933585,
      "grad_norm": 3.807638168334961,
      "learning_rate": 8.047065764245095e-07,
      "loss": 0.88,
      "step": 15130
    },
    {
      "epoch": 0.23562471296130524,
      "grad_norm": 3.85225248336792,
      "learning_rate": 8.046246374199046e-07,
      "loss": 0.8683,
      "step": 15135
    },
    {
      "epoch": 0.2357025539632746,
      "grad_norm": 3.4142606258392334,
      "learning_rate": 8.045426984152997e-07,
      "loss": 0.636,
      "step": 15140
    },
    {
      "epoch": 0.235780394965244,
      "grad_norm": 4.804329872131348,
      "learning_rate": 8.044607594106947e-07,
      "loss": 0.7109,
      "step": 15145
    },
    {
      "epoch": 0.23585823596721336,
      "grad_norm": 5.378768444061279,
      "learning_rate": 8.043788204060896e-07,
      "loss": 0.8313,
      "step": 15150
    },
    {
      "epoch": 0.23593607696918276,
      "grad_norm": 5.029935359954834,
      "learning_rate": 8.042968814014847e-07,
      "loss": 0.7756,
      "step": 15155
    },
    {
      "epoch": 0.23601391797115212,
      "grad_norm": 3.552234649658203,
      "learning_rate": 8.042149423968797e-07,
      "loss": 0.7499,
      "step": 15160
    },
    {
      "epoch": 0.2360917589731215,
      "grad_norm": 5.424056053161621,
      "learning_rate": 8.041330033922748e-07,
      "loss": 0.7542,
      "step": 15165
    },
    {
      "epoch": 0.23616959997509088,
      "grad_norm": 2.835927724838257,
      "learning_rate": 8.040510643876698e-07,
      "loss": 0.7637,
      "step": 15170
    },
    {
      "epoch": 0.23624744097706027,
      "grad_norm": 3.721679210662842,
      "learning_rate": 8.039691253830648e-07,
      "loss": 0.6655,
      "step": 15175
    },
    {
      "epoch": 0.23632528197902963,
      "grad_norm": 3.6744985580444336,
      "learning_rate": 8.038871863784599e-07,
      "loss": 0.7801,
      "step": 15180
    },
    {
      "epoch": 0.236403122980999,
      "grad_norm": 4.117432117462158,
      "learning_rate": 8.038052473738549e-07,
      "loss": 0.7342,
      "step": 15185
    },
    {
      "epoch": 0.2364809639829684,
      "grad_norm": 3.8154187202453613,
      "learning_rate": 8.037233083692498e-07,
      "loss": 0.8315,
      "step": 15190
    },
    {
      "epoch": 0.23655880498493775,
      "grad_norm": 2.46156907081604,
      "learning_rate": 8.036413693646449e-07,
      "loss": 0.7739,
      "step": 15195
    },
    {
      "epoch": 0.23663664598690715,
      "grad_norm": 2.249382257461548,
      "learning_rate": 8.0355943036004e-07,
      "loss": 0.6846,
      "step": 15200
    },
    {
      "epoch": 0.2367144869888765,
      "grad_norm": 3.7491369247436523,
      "learning_rate": 8.034774913554349e-07,
      "loss": 0.6684,
      "step": 15205
    },
    {
      "epoch": 0.2367923279908459,
      "grad_norm": 5.3089599609375,
      "learning_rate": 8.0339555235083e-07,
      "loss": 0.7049,
      "step": 15210
    },
    {
      "epoch": 0.23687016899281527,
      "grad_norm": 9.281523704528809,
      "learning_rate": 8.033136133462251e-07,
      "loss": 0.8498,
      "step": 15215
    },
    {
      "epoch": 0.23694800999478466,
      "grad_norm": 7.066847324371338,
      "learning_rate": 8.0323167434162e-07,
      "loss": 0.7983,
      "step": 15220
    },
    {
      "epoch": 0.23702585099675402,
      "grad_norm": 4.059422016143799,
      "learning_rate": 8.031497353370151e-07,
      "loss": 0.8096,
      "step": 15225
    },
    {
      "epoch": 0.23710369199872341,
      "grad_norm": 3.0671002864837646,
      "learning_rate": 8.030677963324101e-07,
      "loss": 0.7808,
      "step": 15230
    },
    {
      "epoch": 0.23718153300069278,
      "grad_norm": 4.942725658416748,
      "learning_rate": 8.029858573278051e-07,
      "loss": 0.6983,
      "step": 15235
    },
    {
      "epoch": 0.23725937400266217,
      "grad_norm": 6.2389235496521,
      "learning_rate": 8.029039183232001e-07,
      "loss": 0.8416,
      "step": 15240
    },
    {
      "epoch": 0.23733721500463154,
      "grad_norm": 5.170253753662109,
      "learning_rate": 8.028219793185952e-07,
      "loss": 0.698,
      "step": 15245
    },
    {
      "epoch": 0.23741505600660093,
      "grad_norm": 3.286102294921875,
      "learning_rate": 8.027400403139903e-07,
      "loss": 0.6214,
      "step": 15250
    },
    {
      "epoch": 0.2374928970085703,
      "grad_norm": 3.854853868484497,
      "learning_rate": 8.026581013093853e-07,
      "loss": 0.7179,
      "step": 15255
    },
    {
      "epoch": 0.23757073801053968,
      "grad_norm": 3.141479015350342,
      "learning_rate": 8.025761623047803e-07,
      "loss": 0.8504,
      "step": 15260
    },
    {
      "epoch": 0.23764857901250905,
      "grad_norm": 4.667540073394775,
      "learning_rate": 8.024942233001754e-07,
      "loss": 0.7591,
      "step": 15265
    },
    {
      "epoch": 0.23772642001447844,
      "grad_norm": 4.2037882804870605,
      "learning_rate": 8.024122842955704e-07,
      "loss": 0.7391,
      "step": 15270
    },
    {
      "epoch": 0.2378042610164478,
      "grad_norm": 4.822389125823975,
      "learning_rate": 8.023303452909653e-07,
      "loss": 0.7682,
      "step": 15275
    },
    {
      "epoch": 0.23788210201841717,
      "grad_norm": 3.8853273391723633,
      "learning_rate": 8.022484062863604e-07,
      "loss": 0.7997,
      "step": 15280
    },
    {
      "epoch": 0.23795994302038656,
      "grad_norm": 3.6971592903137207,
      "learning_rate": 8.021664672817554e-07,
      "loss": 0.7844,
      "step": 15285
    },
    {
      "epoch": 0.23803778402235592,
      "grad_norm": 4.4559712409973145,
      "learning_rate": 8.020845282771505e-07,
      "loss": 0.7987,
      "step": 15290
    },
    {
      "epoch": 0.23811562502432532,
      "grad_norm": 4.48579740524292,
      "learning_rate": 8.020025892725455e-07,
      "loss": 0.692,
      "step": 15295
    },
    {
      "epoch": 0.23819346602629468,
      "grad_norm": 3.990757465362549,
      "learning_rate": 8.019206502679405e-07,
      "loss": 0.7665,
      "step": 15300
    },
    {
      "epoch": 0.23827130702826407,
      "grad_norm": 4.188387870788574,
      "learning_rate": 8.018387112633356e-07,
      "loss": 0.7027,
      "step": 15305
    },
    {
      "epoch": 0.23834914803023344,
      "grad_norm": 3.3271737098693848,
      "learning_rate": 8.017567722587307e-07,
      "loss": 0.7322,
      "step": 15310
    },
    {
      "epoch": 0.23842698903220283,
      "grad_norm": 4.802268028259277,
      "learning_rate": 8.016748332541255e-07,
      "loss": 0.8131,
      "step": 15315
    },
    {
      "epoch": 0.2385048300341722,
      "grad_norm": 3.292767286300659,
      "learning_rate": 8.015928942495206e-07,
      "loss": 0.837,
      "step": 15320
    },
    {
      "epoch": 0.23858267103614159,
      "grad_norm": 8.88932991027832,
      "learning_rate": 8.015109552449157e-07,
      "loss": 0.7772,
      "step": 15325
    },
    {
      "epoch": 0.23866051203811095,
      "grad_norm": 3.7067458629608154,
      "learning_rate": 8.014290162403106e-07,
      "loss": 0.7729,
      "step": 15330
    },
    {
      "epoch": 0.23873835304008034,
      "grad_norm": 4.151866912841797,
      "learning_rate": 8.013470772357057e-07,
      "loss": 0.8282,
      "step": 15335
    },
    {
      "epoch": 0.2388161940420497,
      "grad_norm": 5.464903831481934,
      "learning_rate": 8.012651382311008e-07,
      "loss": 0.6652,
      "step": 15340
    },
    {
      "epoch": 0.2388940350440191,
      "grad_norm": 3.9546003341674805,
      "learning_rate": 8.011831992264958e-07,
      "loss": 0.7602,
      "step": 15345
    },
    {
      "epoch": 0.23897187604598846,
      "grad_norm": 3.7713892459869385,
      "learning_rate": 8.011012602218908e-07,
      "loss": 0.823,
      "step": 15350
    },
    {
      "epoch": 0.23904971704795785,
      "grad_norm": 3.8322677612304688,
      "learning_rate": 8.010193212172858e-07,
      "loss": 0.8645,
      "step": 15355
    },
    {
      "epoch": 0.23912755804992722,
      "grad_norm": 3.5602362155914307,
      "learning_rate": 8.009373822126808e-07,
      "loss": 0.7581,
      "step": 15360
    },
    {
      "epoch": 0.23920539905189658,
      "grad_norm": 3.7427737712860107,
      "learning_rate": 8.008554432080758e-07,
      "loss": 0.9271,
      "step": 15365
    },
    {
      "epoch": 0.23928324005386598,
      "grad_norm": 3.3515186309814453,
      "learning_rate": 8.007735042034709e-07,
      "loss": 0.5673,
      "step": 15370
    },
    {
      "epoch": 0.23936108105583534,
      "grad_norm": 5.299621105194092,
      "learning_rate": 8.00691565198866e-07,
      "loss": 0.809,
      "step": 15375
    },
    {
      "epoch": 0.23943892205780473,
      "grad_norm": 3.726717233657837,
      "learning_rate": 8.00609626194261e-07,
      "loss": 0.8541,
      "step": 15380
    },
    {
      "epoch": 0.2395167630597741,
      "grad_norm": 4.9893083572387695,
      "learning_rate": 8.00527687189656e-07,
      "loss": 0.6938,
      "step": 15385
    },
    {
      "epoch": 0.2395946040617435,
      "grad_norm": 3.4375534057617188,
      "learning_rate": 8.004457481850511e-07,
      "loss": 0.6838,
      "step": 15390
    },
    {
      "epoch": 0.23967244506371285,
      "grad_norm": 10.630546569824219,
      "learning_rate": 8.00363809180446e-07,
      "loss": 0.9148,
      "step": 15395
    },
    {
      "epoch": 0.23975028606568224,
      "grad_norm": 4.449067115783691,
      "learning_rate": 8.00281870175841e-07,
      "loss": 0.7932,
      "step": 15400
    },
    {
      "epoch": 0.2398281270676516,
      "grad_norm": 3.6890759468078613,
      "learning_rate": 8.001999311712361e-07,
      "loss": 0.7808,
      "step": 15405
    },
    {
      "epoch": 0.239905968069621,
      "grad_norm": 5.842904567718506,
      "learning_rate": 8.001179921666311e-07,
      "loss": 0.784,
      "step": 15410
    },
    {
      "epoch": 0.23998380907159036,
      "grad_norm": 3.3193156719207764,
      "learning_rate": 8.000360531620262e-07,
      "loss": 0.7356,
      "step": 15415
    },
    {
      "epoch": 0.24006165007355976,
      "grad_norm": 3.566133737564087,
      "learning_rate": 7.999541141574212e-07,
      "loss": 0.77,
      "step": 15420
    },
    {
      "epoch": 0.24013949107552912,
      "grad_norm": 3.7197113037109375,
      "learning_rate": 7.998721751528162e-07,
      "loss": 0.7782,
      "step": 15425
    },
    {
      "epoch": 0.2402173320774985,
      "grad_norm": 4.48840856552124,
      "learning_rate": 7.997902361482113e-07,
      "loss": 0.7564,
      "step": 15430
    },
    {
      "epoch": 0.24029517307946788,
      "grad_norm": 3.8312463760375977,
      "learning_rate": 7.997082971436063e-07,
      "loss": 0.7527,
      "step": 15435
    },
    {
      "epoch": 0.24037301408143727,
      "grad_norm": 3.0325686931610107,
      "learning_rate": 7.996263581390012e-07,
      "loss": 0.8293,
      "step": 15440
    },
    {
      "epoch": 0.24045085508340663,
      "grad_norm": 5.339194297790527,
      "learning_rate": 7.995444191343963e-07,
      "loss": 0.7972,
      "step": 15445
    },
    {
      "epoch": 0.240528696085376,
      "grad_norm": 3.5099070072174072,
      "learning_rate": 7.994624801297914e-07,
      "loss": 0.6688,
      "step": 15450
    },
    {
      "epoch": 0.2406065370873454,
      "grad_norm": 4.041937828063965,
      "learning_rate": 7.993805411251863e-07,
      "loss": 0.7599,
      "step": 15455
    },
    {
      "epoch": 0.24068437808931475,
      "grad_norm": 2.9792280197143555,
      "learning_rate": 7.992986021205814e-07,
      "loss": 0.763,
      "step": 15460
    },
    {
      "epoch": 0.24076221909128415,
      "grad_norm": 3.3873233795166016,
      "learning_rate": 7.992166631159765e-07,
      "loss": 0.7828,
      "step": 15465
    },
    {
      "epoch": 0.2408400600932535,
      "grad_norm": 7.740002155303955,
      "learning_rate": 7.991347241113715e-07,
      "loss": 0.8105,
      "step": 15470
    },
    {
      "epoch": 0.2409179010952229,
      "grad_norm": 3.067972183227539,
      "learning_rate": 7.990527851067664e-07,
      "loss": 0.7807,
      "step": 15475
    },
    {
      "epoch": 0.24099574209719227,
      "grad_norm": 5.455771446228027,
      "learning_rate": 7.989708461021615e-07,
      "loss": 0.8325,
      "step": 15480
    },
    {
      "epoch": 0.24107358309916166,
      "grad_norm": 5.027036190032959,
      "learning_rate": 7.988889070975565e-07,
      "loss": 0.8411,
      "step": 15485
    },
    {
      "epoch": 0.24115142410113102,
      "grad_norm": 3.955979347229004,
      "learning_rate": 7.988069680929515e-07,
      "loss": 0.7524,
      "step": 15490
    },
    {
      "epoch": 0.24122926510310042,
      "grad_norm": 8.366657257080078,
      "learning_rate": 7.987250290883466e-07,
      "loss": 0.6843,
      "step": 15495
    },
    {
      "epoch": 0.24130710610506978,
      "grad_norm": 3.6064229011535645,
      "learning_rate": 7.986430900837417e-07,
      "loss": 0.6553,
      "step": 15500
    },
    {
      "epoch": 0.24138494710703917,
      "grad_norm": 4.347894668579102,
      "learning_rate": 7.985611510791367e-07,
      "loss": 0.8618,
      "step": 15505
    },
    {
      "epoch": 0.24146278810900854,
      "grad_norm": 3.172020435333252,
      "learning_rate": 7.984792120745317e-07,
      "loss": 0.7782,
      "step": 15510
    },
    {
      "epoch": 0.24154062911097793,
      "grad_norm": 2.4569530487060547,
      "learning_rate": 7.983972730699267e-07,
      "loss": 0.7837,
      "step": 15515
    },
    {
      "epoch": 0.2416184701129473,
      "grad_norm": 3.2561228275299072,
      "learning_rate": 7.983153340653217e-07,
      "loss": 0.7358,
      "step": 15520
    },
    {
      "epoch": 0.24169631111491668,
      "grad_norm": 3.373246431350708,
      "learning_rate": 7.982333950607168e-07,
      "loss": 0.698,
      "step": 15525
    },
    {
      "epoch": 0.24177415211688605,
      "grad_norm": 3.8543834686279297,
      "learning_rate": 7.981514560561118e-07,
      "loss": 0.8319,
      "step": 15530
    },
    {
      "epoch": 0.2418519931188554,
      "grad_norm": 5.210475921630859,
      "learning_rate": 7.980695170515068e-07,
      "loss": 0.7735,
      "step": 15535
    },
    {
      "epoch": 0.2419298341208248,
      "grad_norm": 5.1110358238220215,
      "learning_rate": 7.979875780469019e-07,
      "loss": 0.7848,
      "step": 15540
    },
    {
      "epoch": 0.24200767512279417,
      "grad_norm": 4.9611663818359375,
      "learning_rate": 7.979056390422969e-07,
      "loss": 0.8454,
      "step": 15545
    },
    {
      "epoch": 0.24208551612476356,
      "grad_norm": 2.9595394134521484,
      "learning_rate": 7.978237000376919e-07,
      "loss": 0.7968,
      "step": 15550
    },
    {
      "epoch": 0.24216335712673293,
      "grad_norm": 5.836178302764893,
      "learning_rate": 7.977417610330869e-07,
      "loss": 0.8707,
      "step": 15555
    },
    {
      "epoch": 0.24224119812870232,
      "grad_norm": 3.4866228103637695,
      "learning_rate": 7.97659822028482e-07,
      "loss": 0.7657,
      "step": 15560
    },
    {
      "epoch": 0.24231903913067168,
      "grad_norm": 3.4374146461486816,
      "learning_rate": 7.975778830238769e-07,
      "loss": 0.7517,
      "step": 15565
    },
    {
      "epoch": 0.24239688013264107,
      "grad_norm": 3.317115068435669,
      "learning_rate": 7.97495944019272e-07,
      "loss": 0.734,
      "step": 15570
    },
    {
      "epoch": 0.24247472113461044,
      "grad_norm": 3.37998104095459,
      "learning_rate": 7.974140050146671e-07,
      "loss": 0.7899,
      "step": 15575
    },
    {
      "epoch": 0.24255256213657983,
      "grad_norm": 4.553988933563232,
      "learning_rate": 7.97332066010062e-07,
      "loss": 0.8539,
      "step": 15580
    },
    {
      "epoch": 0.2426304031385492,
      "grad_norm": 2.3986995220184326,
      "learning_rate": 7.972501270054571e-07,
      "loss": 0.6941,
      "step": 15585
    },
    {
      "epoch": 0.2427082441405186,
      "grad_norm": 5.917320728302002,
      "learning_rate": 7.971681880008522e-07,
      "loss": 0.7735,
      "step": 15590
    },
    {
      "epoch": 0.24278608514248795,
      "grad_norm": 3.8159122467041016,
      "learning_rate": 7.970862489962472e-07,
      "loss": 0.7445,
      "step": 15595
    },
    {
      "epoch": 0.24286392614445734,
      "grad_norm": 4.68142032623291,
      "learning_rate": 7.970043099916421e-07,
      "loss": 0.7696,
      "step": 15600
    },
    {
      "epoch": 0.2429417671464267,
      "grad_norm": 3.4699482917785645,
      "learning_rate": 7.969223709870372e-07,
      "loss": 0.8056,
      "step": 15605
    },
    {
      "epoch": 0.2430196081483961,
      "grad_norm": 2.8645105361938477,
      "learning_rate": 7.968404319824322e-07,
      "loss": 0.8157,
      "step": 15610
    },
    {
      "epoch": 0.24309744915036546,
      "grad_norm": 5.023718357086182,
      "learning_rate": 7.967584929778273e-07,
      "loss": 0.7347,
      "step": 15615
    },
    {
      "epoch": 0.24317529015233483,
      "grad_norm": 5.544175148010254,
      "learning_rate": 7.966765539732223e-07,
      "loss": 0.865,
      "step": 15620
    },
    {
      "epoch": 0.24325313115430422,
      "grad_norm": 3.5016930103302,
      "learning_rate": 7.965946149686174e-07,
      "loss": 0.8305,
      "step": 15625
    },
    {
      "epoch": 0.24333097215627358,
      "grad_norm": 3.4511895179748535,
      "learning_rate": 7.965126759640124e-07,
      "loss": 0.7034,
      "step": 15630
    },
    {
      "epoch": 0.24340881315824298,
      "grad_norm": 7.369199275970459,
      "learning_rate": 7.964307369594074e-07,
      "loss": 0.7782,
      "step": 15635
    },
    {
      "epoch": 0.24348665416021234,
      "grad_norm": 3.0479769706726074,
      "learning_rate": 7.963487979548024e-07,
      "loss": 0.7787,
      "step": 15640
    },
    {
      "epoch": 0.24356449516218173,
      "grad_norm": 3.6762473583221436,
      "learning_rate": 7.962668589501974e-07,
      "loss": 0.7846,
      "step": 15645
    },
    {
      "epoch": 0.2436423361641511,
      "grad_norm": 2.366706371307373,
      "learning_rate": 7.961849199455925e-07,
      "loss": 0.8315,
      "step": 15650
    },
    {
      "epoch": 0.2437201771661205,
      "grad_norm": 3.059541940689087,
      "learning_rate": 7.961029809409875e-07,
      "loss": 0.9155,
      "step": 15655
    },
    {
      "epoch": 0.24379801816808985,
      "grad_norm": 3.0604653358459473,
      "learning_rate": 7.960210419363825e-07,
      "loss": 0.7558,
      "step": 15660
    },
    {
      "epoch": 0.24387585917005924,
      "grad_norm": 3.0823192596435547,
      "learning_rate": 7.959391029317776e-07,
      "loss": 0.7389,
      "step": 15665
    },
    {
      "epoch": 0.2439537001720286,
      "grad_norm": 4.451327800750732,
      "learning_rate": 7.958571639271727e-07,
      "loss": 0.7755,
      "step": 15670
    },
    {
      "epoch": 0.244031541173998,
      "grad_norm": 2.8147003650665283,
      "learning_rate": 7.957752249225676e-07,
      "loss": 0.7032,
      "step": 15675
    },
    {
      "epoch": 0.24410938217596737,
      "grad_norm": 3.849137544631958,
      "learning_rate": 7.956932859179626e-07,
      "loss": 0.6895,
      "step": 15680
    },
    {
      "epoch": 0.24418722317793676,
      "grad_norm": 6.4774627685546875,
      "learning_rate": 7.956113469133577e-07,
      "loss": 0.7932,
      "step": 15685
    },
    {
      "epoch": 0.24426506417990612,
      "grad_norm": 6.876297473907471,
      "learning_rate": 7.955294079087526e-07,
      "loss": 0.8019,
      "step": 15690
    },
    {
      "epoch": 0.2443429051818755,
      "grad_norm": 4.014187812805176,
      "learning_rate": 7.954474689041477e-07,
      "loss": 0.8425,
      "step": 15695
    },
    {
      "epoch": 0.24442074618384488,
      "grad_norm": 4.625513553619385,
      "learning_rate": 7.953655298995428e-07,
      "loss": 0.8028,
      "step": 15700
    },
    {
      "epoch": 0.24449858718581424,
      "grad_norm": 7.612455368041992,
      "learning_rate": 7.952835908949378e-07,
      "loss": 0.9008,
      "step": 15705
    },
    {
      "epoch": 0.24457642818778363,
      "grad_norm": 6.782050609588623,
      "learning_rate": 7.952016518903328e-07,
      "loss": 0.9255,
      "step": 15710
    },
    {
      "epoch": 0.244654269189753,
      "grad_norm": 4.165322780609131,
      "learning_rate": 7.951197128857279e-07,
      "loss": 0.8253,
      "step": 15715
    },
    {
      "epoch": 0.2447321101917224,
      "grad_norm": 6.379940032958984,
      "learning_rate": 7.950377738811228e-07,
      "loss": 0.8267,
      "step": 15720
    },
    {
      "epoch": 0.24480995119369175,
      "grad_norm": 2.715947389602661,
      "learning_rate": 7.949558348765178e-07,
      "loss": 0.7533,
      "step": 15725
    },
    {
      "epoch": 0.24488779219566115,
      "grad_norm": 5.163042068481445,
      "learning_rate": 7.948738958719129e-07,
      "loss": 0.8279,
      "step": 15730
    },
    {
      "epoch": 0.2449656331976305,
      "grad_norm": 3.102045774459839,
      "learning_rate": 7.947919568673079e-07,
      "loss": 0.8667,
      "step": 15735
    },
    {
      "epoch": 0.2450434741995999,
      "grad_norm": 4.849552631378174,
      "learning_rate": 7.94710017862703e-07,
      "loss": 0.7192,
      "step": 15740
    },
    {
      "epoch": 0.24512131520156927,
      "grad_norm": 3.216055393218994,
      "learning_rate": 7.94628078858098e-07,
      "loss": 0.8751,
      "step": 15745
    },
    {
      "epoch": 0.24519915620353866,
      "grad_norm": 7.7601237297058105,
      "learning_rate": 7.945461398534931e-07,
      "loss": 0.7621,
      "step": 15750
    },
    {
      "epoch": 0.24527699720550802,
      "grad_norm": 3.9958834648132324,
      "learning_rate": 7.944642008488881e-07,
      "loss": 0.706,
      "step": 15755
    },
    {
      "epoch": 0.24535483820747742,
      "grad_norm": 2.99743390083313,
      "learning_rate": 7.94382261844283e-07,
      "loss": 0.7574,
      "step": 15760
    },
    {
      "epoch": 0.24543267920944678,
      "grad_norm": 2.5857388973236084,
      "learning_rate": 7.943003228396781e-07,
      "loss": 0.8547,
      "step": 15765
    },
    {
      "epoch": 0.24551052021141617,
      "grad_norm": 4.666770935058594,
      "learning_rate": 7.942183838350731e-07,
      "loss": 0.7977,
      "step": 15770
    },
    {
      "epoch": 0.24558836121338554,
      "grad_norm": 7.485809803009033,
      "learning_rate": 7.941364448304682e-07,
      "loss": 0.8136,
      "step": 15775
    },
    {
      "epoch": 0.24566620221535493,
      "grad_norm": 3.478633165359497,
      "learning_rate": 7.940545058258632e-07,
      "loss": 0.7625,
      "step": 15780
    },
    {
      "epoch": 0.2457440432173243,
      "grad_norm": 4.919050216674805,
      "learning_rate": 7.939725668212582e-07,
      "loss": 0.7858,
      "step": 15785
    },
    {
      "epoch": 0.24582188421929366,
      "grad_norm": 4.7012505531311035,
      "learning_rate": 7.938906278166533e-07,
      "loss": 0.7982,
      "step": 15790
    },
    {
      "epoch": 0.24589972522126305,
      "grad_norm": 3.941943883895874,
      "learning_rate": 7.938086888120484e-07,
      "loss": 0.7948,
      "step": 15795
    },
    {
      "epoch": 0.2459775662232324,
      "grad_norm": 2.9950075149536133,
      "learning_rate": 7.937267498074432e-07,
      "loss": 0.8232,
      "step": 15800
    },
    {
      "epoch": 0.2460554072252018,
      "grad_norm": 3.231367588043213,
      "learning_rate": 7.936448108028383e-07,
      "loss": 0.6793,
      "step": 15805
    },
    {
      "epoch": 0.24613324822717117,
      "grad_norm": 6.495508670806885,
      "learning_rate": 7.935628717982334e-07,
      "loss": 0.7306,
      "step": 15810
    },
    {
      "epoch": 0.24621108922914056,
      "grad_norm": 3.594294786453247,
      "learning_rate": 7.934809327936283e-07,
      "loss": 0.8135,
      "step": 15815
    },
    {
      "epoch": 0.24628893023110993,
      "grad_norm": 4.430262088775635,
      "learning_rate": 7.933989937890234e-07,
      "loss": 0.7158,
      "step": 15820
    },
    {
      "epoch": 0.24636677123307932,
      "grad_norm": 5.1624836921691895,
      "learning_rate": 7.933170547844185e-07,
      "loss": 0.7402,
      "step": 15825
    },
    {
      "epoch": 0.24644461223504868,
      "grad_norm": 2.362281322479248,
      "learning_rate": 7.932351157798135e-07,
      "loss": 0.6743,
      "step": 15830
    },
    {
      "epoch": 0.24652245323701807,
      "grad_norm": 2.9471206665039062,
      "learning_rate": 7.931531767752085e-07,
      "loss": 0.6006,
      "step": 15835
    },
    {
      "epoch": 0.24660029423898744,
      "grad_norm": 3.280003786087036,
      "learning_rate": 7.930712377706035e-07,
      "loss": 0.7592,
      "step": 15840
    },
    {
      "epoch": 0.24667813524095683,
      "grad_norm": 5.314359188079834,
      "learning_rate": 7.929892987659985e-07,
      "loss": 0.8627,
      "step": 15845
    },
    {
      "epoch": 0.2467559762429262,
      "grad_norm": 6.30027437210083,
      "learning_rate": 7.929073597613935e-07,
      "loss": 0.8176,
      "step": 15850
    },
    {
      "epoch": 0.2468338172448956,
      "grad_norm": 8.492682456970215,
      "learning_rate": 7.928254207567886e-07,
      "loss": 0.8596,
      "step": 15855
    },
    {
      "epoch": 0.24691165824686495,
      "grad_norm": 2.7843966484069824,
      "learning_rate": 7.927434817521836e-07,
      "loss": 0.7205,
      "step": 15860
    },
    {
      "epoch": 0.24698949924883434,
      "grad_norm": 3.647057294845581,
      "learning_rate": 7.926615427475787e-07,
      "loss": 0.7801,
      "step": 15865
    },
    {
      "epoch": 0.2470673402508037,
      "grad_norm": 5.774316310882568,
      "learning_rate": 7.925796037429737e-07,
      "loss": 0.7323,
      "step": 15870
    },
    {
      "epoch": 0.2471451812527731,
      "grad_norm": 2.779609441757202,
      "learning_rate": 7.924976647383688e-07,
      "loss": 0.7907,
      "step": 15875
    },
    {
      "epoch": 0.24722302225474246,
      "grad_norm": 5.235540390014648,
      "learning_rate": 7.924157257337637e-07,
      "loss": 0.8759,
      "step": 15880
    },
    {
      "epoch": 0.24730086325671183,
      "grad_norm": 3.683673620223999,
      "learning_rate": 7.923337867291588e-07,
      "loss": 0.8407,
      "step": 15885
    },
    {
      "epoch": 0.24737870425868122,
      "grad_norm": 4.882210731506348,
      "learning_rate": 7.922518477245538e-07,
      "loss": 0.654,
      "step": 15890
    },
    {
      "epoch": 0.24745654526065058,
      "grad_norm": 6.311478137969971,
      "learning_rate": 7.921699087199488e-07,
      "loss": 0.7533,
      "step": 15895
    },
    {
      "epoch": 0.24753438626261998,
      "grad_norm": 3.136744260787964,
      "learning_rate": 7.920879697153439e-07,
      "loss": 0.7993,
      "step": 15900
    },
    {
      "epoch": 0.24761222726458934,
      "grad_norm": 3.370462656021118,
      "learning_rate": 7.92006030710739e-07,
      "loss": 0.7827,
      "step": 15905
    },
    {
      "epoch": 0.24769006826655873,
      "grad_norm": 6.1946892738342285,
      "learning_rate": 7.919240917061339e-07,
      "loss": 0.6617,
      "step": 15910
    },
    {
      "epoch": 0.2477679092685281,
      "grad_norm": 4.19883918762207,
      "learning_rate": 7.91842152701529e-07,
      "loss": 0.6765,
      "step": 15915
    },
    {
      "epoch": 0.2478457502704975,
      "grad_norm": 7.124974250793457,
      "learning_rate": 7.917602136969241e-07,
      "loss": 0.877,
      "step": 15920
    },
    {
      "epoch": 0.24792359127246685,
      "grad_norm": 3.6568546295166016,
      "learning_rate": 7.916782746923189e-07,
      "loss": 0.7851,
      "step": 15925
    },
    {
      "epoch": 0.24800143227443625,
      "grad_norm": 3.275956392288208,
      "learning_rate": 7.91596335687714e-07,
      "loss": 0.8097,
      "step": 15930
    },
    {
      "epoch": 0.2480792732764056,
      "grad_norm": 3.5456111431121826,
      "learning_rate": 7.915143966831091e-07,
      "loss": 0.825,
      "step": 15935
    },
    {
      "epoch": 0.248157114278375,
      "grad_norm": 5.092121601104736,
      "learning_rate": 7.91432457678504e-07,
      "loss": 0.7222,
      "step": 15940
    },
    {
      "epoch": 0.24823495528034437,
      "grad_norm": 3.306962251663208,
      "learning_rate": 7.913505186738991e-07,
      "loss": 0.8208,
      "step": 15945
    },
    {
      "epoch": 0.24831279628231376,
      "grad_norm": 3.8366684913635254,
      "learning_rate": 7.912685796692942e-07,
      "loss": 0.71,
      "step": 15950
    },
    {
      "epoch": 0.24839063728428312,
      "grad_norm": 7.828287124633789,
      "learning_rate": 7.911866406646892e-07,
      "loss": 0.7674,
      "step": 15955
    },
    {
      "epoch": 0.24846847828625251,
      "grad_norm": 6.6405510902404785,
      "learning_rate": 7.911047016600842e-07,
      "loss": 0.7211,
      "step": 15960
    },
    {
      "epoch": 0.24854631928822188,
      "grad_norm": 5.808908462524414,
      "learning_rate": 7.910227626554792e-07,
      "loss": 0.6894,
      "step": 15965
    },
    {
      "epoch": 0.24862416029019124,
      "grad_norm": 3.5409669876098633,
      "learning_rate": 7.909408236508742e-07,
      "loss": 0.953,
      "step": 15970
    },
    {
      "epoch": 0.24870200129216063,
      "grad_norm": 4.125702381134033,
      "learning_rate": 7.908588846462693e-07,
      "loss": 0.8751,
      "step": 15975
    },
    {
      "epoch": 0.24877984229413,
      "grad_norm": 5.823986053466797,
      "learning_rate": 7.907769456416643e-07,
      "loss": 0.7372,
      "step": 15980
    },
    {
      "epoch": 0.2488576832960994,
      "grad_norm": 2.8857569694519043,
      "learning_rate": 7.906950066370593e-07,
      "loss": 0.7657,
      "step": 15985
    },
    {
      "epoch": 0.24893552429806876,
      "grad_norm": 3.998352289199829,
      "learning_rate": 7.906130676324544e-07,
      "loss": 0.7888,
      "step": 15990
    },
    {
      "epoch": 0.24901336530003815,
      "grad_norm": 3.0391101837158203,
      "learning_rate": 7.905311286278494e-07,
      "loss": 0.8028,
      "step": 15995
    },
    {
      "epoch": 0.2490912063020075,
      "grad_norm": 5.301479816436768,
      "learning_rate": 7.904491896232445e-07,
      "loss": 0.8295,
      "step": 16000
    },
    {
      "epoch": 0.2491690473039769,
      "grad_norm": 3.8986215591430664,
      "learning_rate": 7.903672506186394e-07,
      "loss": 0.7564,
      "step": 16005
    },
    {
      "epoch": 0.24924688830594627,
      "grad_norm": 2.7448720932006836,
      "learning_rate": 7.902853116140345e-07,
      "loss": 0.6304,
      "step": 16010
    },
    {
      "epoch": 0.24932472930791566,
      "grad_norm": 9.283259391784668,
      "learning_rate": 7.902033726094295e-07,
      "loss": 0.77,
      "step": 16015
    },
    {
      "epoch": 0.24940257030988502,
      "grad_norm": 3.1082334518432617,
      "learning_rate": 7.901214336048245e-07,
      "loss": 0.7327,
      "step": 16020
    },
    {
      "epoch": 0.24948041131185442,
      "grad_norm": 3.517364025115967,
      "learning_rate": 7.900394946002196e-07,
      "loss": 0.7992,
      "step": 16025
    },
    {
      "epoch": 0.24955825231382378,
      "grad_norm": 3.1072371006011963,
      "learning_rate": 7.899575555956147e-07,
      "loss": 0.7692,
      "step": 16030
    },
    {
      "epoch": 0.24963609331579317,
      "grad_norm": 4.7211527824401855,
      "learning_rate": 7.898756165910096e-07,
      "loss": 0.6787,
      "step": 16035
    },
    {
      "epoch": 0.24971393431776254,
      "grad_norm": 5.028372764587402,
      "learning_rate": 7.897936775864047e-07,
      "loss": 0.6971,
      "step": 16040
    },
    {
      "epoch": 0.24979177531973193,
      "grad_norm": 3.3576149940490723,
      "learning_rate": 7.897117385817997e-07,
      "loss": 0.7766,
      "step": 16045
    },
    {
      "epoch": 0.2498696163217013,
      "grad_norm": 3.800384998321533,
      "learning_rate": 7.896297995771946e-07,
      "loss": 0.8661,
      "step": 16050
    },
    {
      "epoch": 0.24994745732367066,
      "grad_norm": 2.6306138038635254,
      "learning_rate": 7.895478605725897e-07,
      "loss": 0.7385,
      "step": 16055
    },
    {
      "epoch": 0.25002529832564,
      "grad_norm": 3.010930061340332,
      "learning_rate": 7.894659215679848e-07,
      "loss": 0.6919,
      "step": 16060
    },
    {
      "epoch": 0.2501031393276094,
      "grad_norm": 5.057335376739502,
      "learning_rate": 7.893839825633798e-07,
      "loss": 0.8734,
      "step": 16065
    },
    {
      "epoch": 0.2501809803295788,
      "grad_norm": 2.867112398147583,
      "learning_rate": 7.893020435587748e-07,
      "loss": 0.8205,
      "step": 16070
    },
    {
      "epoch": 0.2502588213315482,
      "grad_norm": 3.206636905670166,
      "learning_rate": 7.892201045541699e-07,
      "loss": 0.7503,
      "step": 16075
    },
    {
      "epoch": 0.25033666233351753,
      "grad_norm": 2.602874755859375,
      "learning_rate": 7.891381655495649e-07,
      "loss": 0.632,
      "step": 16080
    },
    {
      "epoch": 0.2504145033354869,
      "grad_norm": 12.468692779541016,
      "learning_rate": 7.890562265449598e-07,
      "loss": 0.8202,
      "step": 16085
    },
    {
      "epoch": 0.2504923443374563,
      "grad_norm": 5.095680236816406,
      "learning_rate": 7.889742875403549e-07,
      "loss": 0.6518,
      "step": 16090
    },
    {
      "epoch": 0.2505701853394257,
      "grad_norm": 3.303833246231079,
      "learning_rate": 7.888923485357499e-07,
      "loss": 0.6671,
      "step": 16095
    },
    {
      "epoch": 0.25064802634139505,
      "grad_norm": 4.023092746734619,
      "learning_rate": 7.88810409531145e-07,
      "loss": 0.784,
      "step": 16100
    },
    {
      "epoch": 0.25072586734336444,
      "grad_norm": 4.499414443969727,
      "learning_rate": 7.8872847052654e-07,
      "loss": 0.8782,
      "step": 16105
    },
    {
      "epoch": 0.25080370834533383,
      "grad_norm": 2.9642868041992188,
      "learning_rate": 7.88646531521935e-07,
      "loss": 0.7308,
      "step": 16110
    },
    {
      "epoch": 0.2508815493473032,
      "grad_norm": 3.3162338733673096,
      "learning_rate": 7.885645925173301e-07,
      "loss": 0.6679,
      "step": 16115
    },
    {
      "epoch": 0.25095939034927256,
      "grad_norm": 3.08477783203125,
      "learning_rate": 7.884826535127252e-07,
      "loss": 0.7544,
      "step": 16120
    },
    {
      "epoch": 0.25103723135124195,
      "grad_norm": 3.925215482711792,
      "learning_rate": 7.8840071450812e-07,
      "loss": 0.7136,
      "step": 16125
    },
    {
      "epoch": 0.25111507235321134,
      "grad_norm": 2.451855182647705,
      "learning_rate": 7.883187755035151e-07,
      "loss": 0.7571,
      "step": 16130
    },
    {
      "epoch": 0.2511929133551807,
      "grad_norm": 4.570413112640381,
      "learning_rate": 7.882368364989102e-07,
      "loss": 0.7323,
      "step": 16135
    },
    {
      "epoch": 0.2512707543571501,
      "grad_norm": 4.142566204071045,
      "learning_rate": 7.881548974943052e-07,
      "loss": 0.7532,
      "step": 16140
    },
    {
      "epoch": 0.25134859535911946,
      "grad_norm": 3.8024239540100098,
      "learning_rate": 7.880729584897002e-07,
      "loss": 0.7589,
      "step": 16145
    },
    {
      "epoch": 0.25142643636108886,
      "grad_norm": 7.670729637145996,
      "learning_rate": 7.879910194850953e-07,
      "loss": 0.7944,
      "step": 16150
    },
    {
      "epoch": 0.2515042773630582,
      "grad_norm": 4.074754238128662,
      "learning_rate": 7.879090804804904e-07,
      "loss": 0.7572,
      "step": 16155
    },
    {
      "epoch": 0.2515821183650276,
      "grad_norm": 4.261497974395752,
      "learning_rate": 7.878271414758853e-07,
      "loss": 0.7098,
      "step": 16160
    },
    {
      "epoch": 0.251659959366997,
      "grad_norm": 3.9497246742248535,
      "learning_rate": 7.877452024712803e-07,
      "loss": 0.7485,
      "step": 16165
    },
    {
      "epoch": 0.25173780036896637,
      "grad_norm": 3.1572659015655518,
      "learning_rate": 7.876632634666754e-07,
      "loss": 0.8074,
      "step": 16170
    },
    {
      "epoch": 0.2518156413709357,
      "grad_norm": 5.557485580444336,
      "learning_rate": 7.875813244620703e-07,
      "loss": 0.8997,
      "step": 16175
    },
    {
      "epoch": 0.2518934823729051,
      "grad_norm": 3.017874240875244,
      "learning_rate": 7.874993854574654e-07,
      "loss": 0.7874,
      "step": 16180
    },
    {
      "epoch": 0.2519713233748745,
      "grad_norm": 4.374793529510498,
      "learning_rate": 7.874174464528605e-07,
      "loss": 0.762,
      "step": 16185
    },
    {
      "epoch": 0.2520491643768439,
      "grad_norm": 2.7853641510009766,
      "learning_rate": 7.873355074482555e-07,
      "loss": 0.8455,
      "step": 16190
    },
    {
      "epoch": 0.2521270053788132,
      "grad_norm": 3.7544867992401123,
      "learning_rate": 7.872535684436505e-07,
      "loss": 0.699,
      "step": 16195
    },
    {
      "epoch": 0.2522048463807826,
      "grad_norm": 7.739635944366455,
      "learning_rate": 7.871716294390456e-07,
      "loss": 0.782,
      "step": 16200
    },
    {
      "epoch": 0.252282687382752,
      "grad_norm": 3.971346855163574,
      "learning_rate": 7.870896904344405e-07,
      "loss": 0.6151,
      "step": 16205
    },
    {
      "epoch": 0.2523605283847214,
      "grad_norm": 4.750772953033447,
      "learning_rate": 7.870077514298355e-07,
      "loss": 0.8195,
      "step": 16210
    },
    {
      "epoch": 0.25243836938669073,
      "grad_norm": 3.0286149978637695,
      "learning_rate": 7.869258124252306e-07,
      "loss": 0.7301,
      "step": 16215
    },
    {
      "epoch": 0.2525162103886601,
      "grad_norm": 4.7284932136535645,
      "learning_rate": 7.868438734206256e-07,
      "loss": 0.8878,
      "step": 16220
    },
    {
      "epoch": 0.2525940513906295,
      "grad_norm": 7.5153398513793945,
      "learning_rate": 7.867619344160207e-07,
      "loss": 0.7866,
      "step": 16225
    },
    {
      "epoch": 0.25267189239259885,
      "grad_norm": 5.353423595428467,
      "learning_rate": 7.866799954114157e-07,
      "loss": 0.7017,
      "step": 16230
    },
    {
      "epoch": 0.25274973339456824,
      "grad_norm": 3.3869919776916504,
      "learning_rate": 7.865980564068107e-07,
      "loss": 0.8093,
      "step": 16235
    },
    {
      "epoch": 0.25282757439653764,
      "grad_norm": 4.509708881378174,
      "learning_rate": 7.865161174022058e-07,
      "loss": 0.6865,
      "step": 16240
    },
    {
      "epoch": 0.252905415398507,
      "grad_norm": 4.100122928619385,
      "learning_rate": 7.864341783976009e-07,
      "loss": 0.8192,
      "step": 16245
    },
    {
      "epoch": 0.25298325640047636,
      "grad_norm": 6.802055358886719,
      "learning_rate": 7.863522393929957e-07,
      "loss": 0.775,
      "step": 16250
    },
    {
      "epoch": 0.25306109740244576,
      "grad_norm": 5.319190502166748,
      "learning_rate": 7.862703003883908e-07,
      "loss": 0.7606,
      "step": 16255
    },
    {
      "epoch": 0.25313893840441515,
      "grad_norm": 4.581395626068115,
      "learning_rate": 7.861883613837859e-07,
      "loss": 0.7658,
      "step": 16260
    },
    {
      "epoch": 0.25321677940638454,
      "grad_norm": 3.787923812866211,
      "learning_rate": 7.86106422379181e-07,
      "loss": 0.8342,
      "step": 16265
    },
    {
      "epoch": 0.2532946204083539,
      "grad_norm": 5.891742706298828,
      "learning_rate": 7.860244833745759e-07,
      "loss": 0.7354,
      "step": 16270
    },
    {
      "epoch": 0.25337246141032327,
      "grad_norm": 3.339656114578247,
      "learning_rate": 7.85942544369971e-07,
      "loss": 0.7455,
      "step": 16275
    },
    {
      "epoch": 0.25345030241229266,
      "grad_norm": 4.084258556365967,
      "learning_rate": 7.858606053653661e-07,
      "loss": 0.7613,
      "step": 16280
    },
    {
      "epoch": 0.25352814341426205,
      "grad_norm": 8.612784385681152,
      "learning_rate": 7.85778666360761e-07,
      "loss": 0.755,
      "step": 16285
    },
    {
      "epoch": 0.2536059844162314,
      "grad_norm": 5.70842981338501,
      "learning_rate": 7.85696727356156e-07,
      "loss": 0.6829,
      "step": 16290
    },
    {
      "epoch": 0.2536838254182008,
      "grad_norm": 4.7522501945495605,
      "learning_rate": 7.856147883515511e-07,
      "loss": 0.7412,
      "step": 16295
    },
    {
      "epoch": 0.2537616664201702,
      "grad_norm": 3.4117345809936523,
      "learning_rate": 7.85532849346946e-07,
      "loss": 0.6579,
      "step": 16300
    },
    {
      "epoch": 0.25383950742213957,
      "grad_norm": 4.271410942077637,
      "learning_rate": 7.854509103423411e-07,
      "loss": 0.8531,
      "step": 16305
    },
    {
      "epoch": 0.2539173484241089,
      "grad_norm": 4.646153450012207,
      "learning_rate": 7.853689713377362e-07,
      "loss": 0.8344,
      "step": 16310
    },
    {
      "epoch": 0.2539951894260783,
      "grad_norm": 3.57694411277771,
      "learning_rate": 7.852870323331312e-07,
      "loss": 0.7752,
      "step": 16315
    },
    {
      "epoch": 0.2540730304280477,
      "grad_norm": 5.518935680389404,
      "learning_rate": 7.852050933285262e-07,
      "loss": 0.7524,
      "step": 16320
    },
    {
      "epoch": 0.254150871430017,
      "grad_norm": 4.475656509399414,
      "learning_rate": 7.851231543239213e-07,
      "loss": 0.8602,
      "step": 16325
    },
    {
      "epoch": 0.2542287124319864,
      "grad_norm": 5.29443883895874,
      "learning_rate": 7.850412153193162e-07,
      "loss": 0.6992,
      "step": 16330
    },
    {
      "epoch": 0.2543065534339558,
      "grad_norm": 4.594540119171143,
      "learning_rate": 7.849592763147113e-07,
      "loss": 0.823,
      "step": 16335
    },
    {
      "epoch": 0.2543843944359252,
      "grad_norm": 4.070684909820557,
      "learning_rate": 7.848773373101063e-07,
      "loss": 0.8951,
      "step": 16340
    },
    {
      "epoch": 0.25446223543789454,
      "grad_norm": 4.30888032913208,
      "learning_rate": 7.847953983055013e-07,
      "loss": 0.9167,
      "step": 16345
    },
    {
      "epoch": 0.2545400764398639,
      "grad_norm": 2.9648544788360596,
      "learning_rate": 7.847134593008964e-07,
      "loss": 0.7235,
      "step": 16350
    },
    {
      "epoch": 0.2546179174418333,
      "grad_norm": 7.27984094619751,
      "learning_rate": 7.846315202962915e-07,
      "loss": 0.7889,
      "step": 16355
    },
    {
      "epoch": 0.2546957584438027,
      "grad_norm": 4.1877827644348145,
      "learning_rate": 7.845495812916864e-07,
      "loss": 0.737,
      "step": 16360
    },
    {
      "epoch": 0.25477359944577205,
      "grad_norm": 3.393552780151367,
      "learning_rate": 7.844676422870815e-07,
      "loss": 0.7156,
      "step": 16365
    },
    {
      "epoch": 0.25485144044774144,
      "grad_norm": 10.564290046691895,
      "learning_rate": 7.843857032824765e-07,
      "loss": 0.7583,
      "step": 16370
    },
    {
      "epoch": 0.25492928144971083,
      "grad_norm": 5.3602800369262695,
      "learning_rate": 7.843037642778714e-07,
      "loss": 0.7714,
      "step": 16375
    },
    {
      "epoch": 0.2550071224516802,
      "grad_norm": 5.475079536437988,
      "learning_rate": 7.842218252732665e-07,
      "loss": 0.6836,
      "step": 16380
    },
    {
      "epoch": 0.25508496345364956,
      "grad_norm": 3.7718818187713623,
      "learning_rate": 7.841398862686616e-07,
      "loss": 0.6086,
      "step": 16385
    },
    {
      "epoch": 0.25516280445561895,
      "grad_norm": 2.7791218757629395,
      "learning_rate": 7.840579472640567e-07,
      "loss": 0.7479,
      "step": 16390
    },
    {
      "epoch": 0.25524064545758834,
      "grad_norm": 5.199838161468506,
      "learning_rate": 7.839760082594516e-07,
      "loss": 0.8453,
      "step": 16395
    },
    {
      "epoch": 0.2553184864595577,
      "grad_norm": 5.045164585113525,
      "learning_rate": 7.838940692548467e-07,
      "loss": 0.7707,
      "step": 16400
    },
    {
      "epoch": 0.2553963274615271,
      "grad_norm": 3.9481828212738037,
      "learning_rate": 7.838121302502418e-07,
      "loss": 0.8319,
      "step": 16405
    },
    {
      "epoch": 0.25547416846349646,
      "grad_norm": 3.1789422035217285,
      "learning_rate": 7.837301912456366e-07,
      "loss": 0.891,
      "step": 16410
    },
    {
      "epoch": 0.25555200946546586,
      "grad_norm": 3.5486695766448975,
      "learning_rate": 7.836482522410317e-07,
      "loss": 0.8076,
      "step": 16415
    },
    {
      "epoch": 0.2556298504674352,
      "grad_norm": 2.756218910217285,
      "learning_rate": 7.835663132364268e-07,
      "loss": 0.8272,
      "step": 16420
    },
    {
      "epoch": 0.2557076914694046,
      "grad_norm": 3.782132148742676,
      "learning_rate": 7.834843742318218e-07,
      "loss": 0.6728,
      "step": 16425
    },
    {
      "epoch": 0.255785532471374,
      "grad_norm": 3.6226656436920166,
      "learning_rate": 7.834024352272168e-07,
      "loss": 0.7395,
      "step": 16430
    },
    {
      "epoch": 0.25586337347334337,
      "grad_norm": 3.3157193660736084,
      "learning_rate": 7.833204962226119e-07,
      "loss": 0.8317,
      "step": 16435
    },
    {
      "epoch": 0.2559412144753127,
      "grad_norm": 3.4248037338256836,
      "learning_rate": 7.832385572180069e-07,
      "loss": 0.7668,
      "step": 16440
    },
    {
      "epoch": 0.2560190554772821,
      "grad_norm": 7.231293201446533,
      "learning_rate": 7.83156618213402e-07,
      "loss": 0.6963,
      "step": 16445
    },
    {
      "epoch": 0.2560968964792515,
      "grad_norm": 10.40032958984375,
      "learning_rate": 7.830746792087969e-07,
      "loss": 0.6711,
      "step": 16450
    },
    {
      "epoch": 0.2561747374812209,
      "grad_norm": 6.381346702575684,
      "learning_rate": 7.829927402041919e-07,
      "loss": 0.7014,
      "step": 16455
    },
    {
      "epoch": 0.2562525784831902,
      "grad_norm": 2.866187572479248,
      "learning_rate": 7.82910801199587e-07,
      "loss": 0.841,
      "step": 16460
    },
    {
      "epoch": 0.2563304194851596,
      "grad_norm": 5.38809061050415,
      "learning_rate": 7.82828862194982e-07,
      "loss": 0.751,
      "step": 16465
    },
    {
      "epoch": 0.256408260487129,
      "grad_norm": 3.7504756450653076,
      "learning_rate": 7.82746923190377e-07,
      "loss": 0.7829,
      "step": 16470
    },
    {
      "epoch": 0.2564861014890984,
      "grad_norm": 3.9122817516326904,
      "learning_rate": 7.826649841857721e-07,
      "loss": 0.7979,
      "step": 16475
    },
    {
      "epoch": 0.25656394249106773,
      "grad_norm": 3.9444596767425537,
      "learning_rate": 7.825830451811672e-07,
      "loss": 0.8804,
      "step": 16480
    },
    {
      "epoch": 0.2566417834930371,
      "grad_norm": 4.131224155426025,
      "learning_rate": 7.825011061765621e-07,
      "loss": 0.7349,
      "step": 16485
    },
    {
      "epoch": 0.2567196244950065,
      "grad_norm": 4.119733810424805,
      "learning_rate": 7.824191671719571e-07,
      "loss": 0.612,
      "step": 16490
    },
    {
      "epoch": 0.25679746549697585,
      "grad_norm": 5.177633762359619,
      "learning_rate": 7.823372281673522e-07,
      "loss": 0.7003,
      "step": 16495
    },
    {
      "epoch": 0.25687530649894524,
      "grad_norm": 5.1837992668151855,
      "learning_rate": 7.822552891627471e-07,
      "loss": 0.7929,
      "step": 16500
    },
    {
      "epoch": 0.25695314750091464,
      "grad_norm": 6.409270763397217,
      "learning_rate": 7.821733501581422e-07,
      "loss": 0.7425,
      "step": 16505
    },
    {
      "epoch": 0.25703098850288403,
      "grad_norm": 3.130068063735962,
      "learning_rate": 7.820914111535373e-07,
      "loss": 0.6838,
      "step": 16510
    },
    {
      "epoch": 0.25710882950485336,
      "grad_norm": 2.76039719581604,
      "learning_rate": 7.820094721489324e-07,
      "loss": 0.7023,
      "step": 16515
    },
    {
      "epoch": 0.25718667050682276,
      "grad_norm": 4.199296474456787,
      "learning_rate": 7.819275331443273e-07,
      "loss": 0.777,
      "step": 16520
    },
    {
      "epoch": 0.25726451150879215,
      "grad_norm": 2.0903480052948,
      "learning_rate": 7.818455941397224e-07,
      "loss": 0.6523,
      "step": 16525
    },
    {
      "epoch": 0.25734235251076154,
      "grad_norm": 3.675006866455078,
      "learning_rate": 7.817636551351174e-07,
      "loss": 0.7007,
      "step": 16530
    },
    {
      "epoch": 0.2574201935127309,
      "grad_norm": 5.696536540985107,
      "learning_rate": 7.816817161305123e-07,
      "loss": 0.822,
      "step": 16535
    },
    {
      "epoch": 0.25749803451470027,
      "grad_norm": 3.226651668548584,
      "learning_rate": 7.815997771259074e-07,
      "loss": 0.7139,
      "step": 16540
    },
    {
      "epoch": 0.25757587551666966,
      "grad_norm": 4.20841121673584,
      "learning_rate": 7.815178381213025e-07,
      "loss": 0.795,
      "step": 16545
    },
    {
      "epoch": 0.25765371651863905,
      "grad_norm": 7.1250834465026855,
      "learning_rate": 7.814358991166975e-07,
      "loss": 0.7926,
      "step": 16550
    },
    {
      "epoch": 0.2577315575206084,
      "grad_norm": 3.3494808673858643,
      "learning_rate": 7.813539601120925e-07,
      "loss": 0.8161,
      "step": 16555
    },
    {
      "epoch": 0.2578093985225778,
      "grad_norm": 3.7074177265167236,
      "learning_rate": 7.812720211074876e-07,
      "loss": 0.8033,
      "step": 16560
    },
    {
      "epoch": 0.2578872395245472,
      "grad_norm": 3.2563693523406982,
      "learning_rate": 7.811900821028826e-07,
      "loss": 0.9385,
      "step": 16565
    },
    {
      "epoch": 0.2579650805265165,
      "grad_norm": 4.425312042236328,
      "learning_rate": 7.811081430982777e-07,
      "loss": 0.7872,
      "step": 16570
    },
    {
      "epoch": 0.2580429215284859,
      "grad_norm": 4.398519515991211,
      "learning_rate": 7.810262040936726e-07,
      "loss": 0.6697,
      "step": 16575
    },
    {
      "epoch": 0.2581207625304553,
      "grad_norm": 2.759260892868042,
      "learning_rate": 7.809442650890676e-07,
      "loss": 0.7034,
      "step": 16580
    },
    {
      "epoch": 0.2581986035324247,
      "grad_norm": 6.658419609069824,
      "learning_rate": 7.808623260844627e-07,
      "loss": 0.8203,
      "step": 16585
    },
    {
      "epoch": 0.258276444534394,
      "grad_norm": 2.738607406616211,
      "learning_rate": 7.807803870798577e-07,
      "loss": 0.7464,
      "step": 16590
    },
    {
      "epoch": 0.2583542855363634,
      "grad_norm": 3.0051686763763428,
      "learning_rate": 7.806984480752527e-07,
      "loss": 0.7817,
      "step": 16595
    },
    {
      "epoch": 0.2584321265383328,
      "grad_norm": 3.117178440093994,
      "learning_rate": 7.806165090706478e-07,
      "loss": 0.79,
      "step": 16600
    },
    {
      "epoch": 0.2585099675403022,
      "grad_norm": 4.8895158767700195,
      "learning_rate": 7.805345700660429e-07,
      "loss": 0.7484,
      "step": 16605
    },
    {
      "epoch": 0.25858780854227154,
      "grad_norm": 3.8371946811676025,
      "learning_rate": 7.804526310614378e-07,
      "loss": 0.6784,
      "step": 16610
    },
    {
      "epoch": 0.2586656495442409,
      "grad_norm": 3.57291579246521,
      "learning_rate": 7.803706920568328e-07,
      "loss": 0.7957,
      "step": 16615
    },
    {
      "epoch": 0.2587434905462103,
      "grad_norm": 3.827907085418701,
      "learning_rate": 7.802887530522279e-07,
      "loss": 0.7731,
      "step": 16620
    },
    {
      "epoch": 0.2588213315481797,
      "grad_norm": 3.4150476455688477,
      "learning_rate": 7.802068140476228e-07,
      "loss": 0.9146,
      "step": 16625
    },
    {
      "epoch": 0.25889917255014905,
      "grad_norm": 5.377663612365723,
      "learning_rate": 7.801248750430179e-07,
      "loss": 0.7602,
      "step": 16630
    },
    {
      "epoch": 0.25897701355211844,
      "grad_norm": 4.645751476287842,
      "learning_rate": 7.80042936038413e-07,
      "loss": 0.8021,
      "step": 16635
    },
    {
      "epoch": 0.25905485455408783,
      "grad_norm": 6.805863380432129,
      "learning_rate": 7.799609970338081e-07,
      "loss": 0.7908,
      "step": 16640
    },
    {
      "epoch": 0.2591326955560572,
      "grad_norm": 2.4049344062805176,
      "learning_rate": 7.79879058029203e-07,
      "loss": 0.7619,
      "step": 16645
    },
    {
      "epoch": 0.25921053655802656,
      "grad_norm": 3.3766093254089355,
      "learning_rate": 7.797971190245981e-07,
      "loss": 0.7428,
      "step": 16650
    },
    {
      "epoch": 0.25928837755999595,
      "grad_norm": 4.741754531860352,
      "learning_rate": 7.797151800199931e-07,
      "loss": 0.8635,
      "step": 16655
    },
    {
      "epoch": 0.25936621856196534,
      "grad_norm": 2.6133670806884766,
      "learning_rate": 7.79633241015388e-07,
      "loss": 0.6963,
      "step": 16660
    },
    {
      "epoch": 0.2594440595639347,
      "grad_norm": 4.859498977661133,
      "learning_rate": 7.795513020107831e-07,
      "loss": 0.7334,
      "step": 16665
    },
    {
      "epoch": 0.2595219005659041,
      "grad_norm": 3.187589406967163,
      "learning_rate": 7.794693630061782e-07,
      "loss": 0.7591,
      "step": 16670
    },
    {
      "epoch": 0.25959974156787347,
      "grad_norm": 6.532161235809326,
      "learning_rate": 7.793874240015732e-07,
      "loss": 0.7892,
      "step": 16675
    },
    {
      "epoch": 0.25967758256984286,
      "grad_norm": 4.1265177726745605,
      "learning_rate": 7.793054849969682e-07,
      "loss": 0.7637,
      "step": 16680
    },
    {
      "epoch": 0.2597554235718122,
      "grad_norm": 5.0222649574279785,
      "learning_rate": 7.792235459923633e-07,
      "loss": 0.7339,
      "step": 16685
    },
    {
      "epoch": 0.2598332645737816,
      "grad_norm": 3.554102659225464,
      "learning_rate": 7.791416069877583e-07,
      "loss": 0.821,
      "step": 16690
    },
    {
      "epoch": 0.259911105575751,
      "grad_norm": 3.7950658798217773,
      "learning_rate": 7.790596679831533e-07,
      "loss": 0.8341,
      "step": 16695
    },
    {
      "epoch": 0.25998894657772037,
      "grad_norm": 3.3983871936798096,
      "learning_rate": 7.789777289785483e-07,
      "loss": 0.7982,
      "step": 16700
    },
    {
      "epoch": 0.2600667875796897,
      "grad_norm": 3.4833576679229736,
      "learning_rate": 7.788957899739433e-07,
      "loss": 0.8516,
      "step": 16705
    },
    {
      "epoch": 0.2601446285816591,
      "grad_norm": 2.6888585090637207,
      "learning_rate": 7.788138509693384e-07,
      "loss": 0.6446,
      "step": 16710
    },
    {
      "epoch": 0.2602224695836285,
      "grad_norm": 3.0143544673919678,
      "learning_rate": 7.787319119647335e-07,
      "loss": 0.8344,
      "step": 16715
    },
    {
      "epoch": 0.2603003105855979,
      "grad_norm": 9.82540512084961,
      "learning_rate": 7.786499729601284e-07,
      "loss": 0.7017,
      "step": 16720
    },
    {
      "epoch": 0.2603781515875672,
      "grad_norm": 3.5165607929229736,
      "learning_rate": 7.785680339555235e-07,
      "loss": 0.6927,
      "step": 16725
    },
    {
      "epoch": 0.2604559925895366,
      "grad_norm": 3.0352158546447754,
      "learning_rate": 7.784860949509186e-07,
      "loss": 0.8134,
      "step": 16730
    },
    {
      "epoch": 0.260533833591506,
      "grad_norm": 2.934072494506836,
      "learning_rate": 7.784041559463134e-07,
      "loss": 0.7767,
      "step": 16735
    },
    {
      "epoch": 0.26061167459347534,
      "grad_norm": 5.560964107513428,
      "learning_rate": 7.783222169417085e-07,
      "loss": 0.835,
      "step": 16740
    },
    {
      "epoch": 0.26068951559544473,
      "grad_norm": 6.389545917510986,
      "learning_rate": 7.782402779371036e-07,
      "loss": 0.8353,
      "step": 16745
    },
    {
      "epoch": 0.2607673565974141,
      "grad_norm": 2.7560980319976807,
      "learning_rate": 7.781583389324986e-07,
      "loss": 0.7777,
      "step": 16750
    },
    {
      "epoch": 0.2608451975993835,
      "grad_norm": 5.903055667877197,
      "learning_rate": 7.780763999278936e-07,
      "loss": 0.8077,
      "step": 16755
    },
    {
      "epoch": 0.26092303860135285,
      "grad_norm": 3.2455592155456543,
      "learning_rate": 7.779944609232887e-07,
      "loss": 0.8842,
      "step": 16760
    },
    {
      "epoch": 0.26100087960332224,
      "grad_norm": 3.2503061294555664,
      "learning_rate": 7.779125219186838e-07,
      "loss": 0.7527,
      "step": 16765
    },
    {
      "epoch": 0.26107872060529164,
      "grad_norm": 13.79643440246582,
      "learning_rate": 7.778305829140787e-07,
      "loss": 0.7926,
      "step": 16770
    },
    {
      "epoch": 0.26115656160726103,
      "grad_norm": 4.413746356964111,
      "learning_rate": 7.777486439094737e-07,
      "loss": 0.6204,
      "step": 16775
    },
    {
      "epoch": 0.26123440260923037,
      "grad_norm": 4.079379081726074,
      "learning_rate": 7.776667049048688e-07,
      "loss": 0.8961,
      "step": 16780
    },
    {
      "epoch": 0.26131224361119976,
      "grad_norm": 3.8018133640289307,
      "learning_rate": 7.775847659002638e-07,
      "loss": 0.7947,
      "step": 16785
    },
    {
      "epoch": 0.26139008461316915,
      "grad_norm": 3.7187345027923584,
      "learning_rate": 7.775028268956588e-07,
      "loss": 0.7183,
      "step": 16790
    },
    {
      "epoch": 0.26146792561513854,
      "grad_norm": 3.7661221027374268,
      "learning_rate": 7.774208878910539e-07,
      "loss": 0.7923,
      "step": 16795
    },
    {
      "epoch": 0.2615457666171079,
      "grad_norm": 4.487337112426758,
      "learning_rate": 7.773389488864489e-07,
      "loss": 0.8413,
      "step": 16800
    },
    {
      "epoch": 0.26162360761907727,
      "grad_norm": 3.0049211978912354,
      "learning_rate": 7.77257009881844e-07,
      "loss": 0.7203,
      "step": 16805
    },
    {
      "epoch": 0.26170144862104666,
      "grad_norm": 9.142400741577148,
      "learning_rate": 7.77175070877239e-07,
      "loss": 0.7582,
      "step": 16810
    },
    {
      "epoch": 0.26177928962301605,
      "grad_norm": 6.436301231384277,
      "learning_rate": 7.770931318726339e-07,
      "loss": 0.651,
      "step": 16815
    },
    {
      "epoch": 0.2618571306249854,
      "grad_norm": 11.26479434967041,
      "learning_rate": 7.77011192868029e-07,
      "loss": 0.7751,
      "step": 16820
    },
    {
      "epoch": 0.2619349716269548,
      "grad_norm": 4.235225200653076,
      "learning_rate": 7.76929253863424e-07,
      "loss": 0.7435,
      "step": 16825
    },
    {
      "epoch": 0.2620128126289242,
      "grad_norm": 3.807072162628174,
      "learning_rate": 7.76847314858819e-07,
      "loss": 0.7725,
      "step": 16830
    },
    {
      "epoch": 0.2620906536308935,
      "grad_norm": 3.0418660640716553,
      "learning_rate": 7.767653758542141e-07,
      "loss": 0.7339,
      "step": 16835
    },
    {
      "epoch": 0.2621684946328629,
      "grad_norm": 4.268150329589844,
      "learning_rate": 7.766834368496092e-07,
      "loss": 0.7511,
      "step": 16840
    },
    {
      "epoch": 0.2622463356348323,
      "grad_norm": 3.158903121948242,
      "learning_rate": 7.766014978450041e-07,
      "loss": 0.8017,
      "step": 16845
    },
    {
      "epoch": 0.2623241766368017,
      "grad_norm": 2.870638132095337,
      "learning_rate": 7.765195588403992e-07,
      "loss": 0.8768,
      "step": 16850
    },
    {
      "epoch": 0.262402017638771,
      "grad_norm": 5.223047733306885,
      "learning_rate": 7.764376198357942e-07,
      "loss": 0.7877,
      "step": 16855
    },
    {
      "epoch": 0.2624798586407404,
      "grad_norm": 4.719412326812744,
      "learning_rate": 7.763556808311891e-07,
      "loss": 0.7911,
      "step": 16860
    },
    {
      "epoch": 0.2625576996427098,
      "grad_norm": 4.325451850891113,
      "learning_rate": 7.762737418265842e-07,
      "loss": 0.8931,
      "step": 16865
    },
    {
      "epoch": 0.2626355406446792,
      "grad_norm": 7.159557342529297,
      "learning_rate": 7.761918028219793e-07,
      "loss": 0.718,
      "step": 16870
    },
    {
      "epoch": 0.26271338164664854,
      "grad_norm": 3.957003116607666,
      "learning_rate": 7.761098638173743e-07,
      "loss": 0.8571,
      "step": 16875
    },
    {
      "epoch": 0.26279122264861793,
      "grad_norm": 3.6029467582702637,
      "learning_rate": 7.760279248127693e-07,
      "loss": 0.8155,
      "step": 16880
    },
    {
      "epoch": 0.2628690636505873,
      "grad_norm": 3.1437227725982666,
      "learning_rate": 7.759459858081644e-07,
      "loss": 0.7031,
      "step": 16885
    },
    {
      "epoch": 0.2629469046525567,
      "grad_norm": 8.502701759338379,
      "learning_rate": 7.758640468035595e-07,
      "loss": 0.7562,
      "step": 16890
    },
    {
      "epoch": 0.26302474565452605,
      "grad_norm": 4.806173324584961,
      "learning_rate": 7.757821077989545e-07,
      "loss": 0.8258,
      "step": 16895
    },
    {
      "epoch": 0.26310258665649544,
      "grad_norm": 15.015897750854492,
      "learning_rate": 7.757001687943494e-07,
      "loss": 0.7796,
      "step": 16900
    },
    {
      "epoch": 0.26318042765846483,
      "grad_norm": 4.395715236663818,
      "learning_rate": 7.756182297897445e-07,
      "loss": 0.6802,
      "step": 16905
    },
    {
      "epoch": 0.2632582686604342,
      "grad_norm": 4.44493293762207,
      "learning_rate": 7.755362907851395e-07,
      "loss": 0.6893,
      "step": 16910
    },
    {
      "epoch": 0.26333610966240356,
      "grad_norm": 4.132498264312744,
      "learning_rate": 7.754543517805345e-07,
      "loss": 0.7528,
      "step": 16915
    },
    {
      "epoch": 0.26341395066437295,
      "grad_norm": 4.151634693145752,
      "learning_rate": 7.753724127759296e-07,
      "loss": 0.7714,
      "step": 16920
    },
    {
      "epoch": 0.26349179166634235,
      "grad_norm": 3.9134562015533447,
      "learning_rate": 7.752904737713246e-07,
      "loss": 0.7701,
      "step": 16925
    },
    {
      "epoch": 0.2635696326683117,
      "grad_norm": 7.147616386413574,
      "learning_rate": 7.752085347667197e-07,
      "loss": 0.8402,
      "step": 16930
    },
    {
      "epoch": 0.2636474736702811,
      "grad_norm": 2.874391555786133,
      "learning_rate": 7.751265957621147e-07,
      "loss": 0.8452,
      "step": 16935
    },
    {
      "epoch": 0.26372531467225047,
      "grad_norm": 3.2869579792022705,
      "learning_rate": 7.750446567575096e-07,
      "loss": 0.7295,
      "step": 16940
    },
    {
      "epoch": 0.26380315567421986,
      "grad_norm": 4.201857566833496,
      "learning_rate": 7.749627177529047e-07,
      "loss": 0.6901,
      "step": 16945
    },
    {
      "epoch": 0.2638809966761892,
      "grad_norm": 6.836452960968018,
      "learning_rate": 7.748807787482997e-07,
      "loss": 0.8212,
      "step": 16950
    },
    {
      "epoch": 0.2639588376781586,
      "grad_norm": 3.0617053508758545,
      "learning_rate": 7.747988397436947e-07,
      "loss": 0.6808,
      "step": 16955
    },
    {
      "epoch": 0.264036678680128,
      "grad_norm": 3.325904607772827,
      "learning_rate": 7.747169007390898e-07,
      "loss": 0.7128,
      "step": 16960
    },
    {
      "epoch": 0.26411451968209737,
      "grad_norm": 10.036579132080078,
      "learning_rate": 7.746349617344849e-07,
      "loss": 0.7599,
      "step": 16965
    },
    {
      "epoch": 0.2641923606840667,
      "grad_norm": 3.4642529487609863,
      "learning_rate": 7.745530227298798e-07,
      "loss": 0.7073,
      "step": 16970
    },
    {
      "epoch": 0.2642702016860361,
      "grad_norm": 4.849530220031738,
      "learning_rate": 7.744710837252749e-07,
      "loss": 0.7978,
      "step": 16975
    },
    {
      "epoch": 0.2643480426880055,
      "grad_norm": 4.592496871948242,
      "learning_rate": 7.743891447206699e-07,
      "loss": 0.8337,
      "step": 16980
    },
    {
      "epoch": 0.2644258836899749,
      "grad_norm": 7.480959892272949,
      "learning_rate": 7.743072057160648e-07,
      "loss": 0.6988,
      "step": 16985
    },
    {
      "epoch": 0.2645037246919442,
      "grad_norm": 3.0004184246063232,
      "learning_rate": 7.742252667114599e-07,
      "loss": 0.706,
      "step": 16990
    },
    {
      "epoch": 0.2645815656939136,
      "grad_norm": 5.580563545227051,
      "learning_rate": 7.74143327706855e-07,
      "loss": 0.7985,
      "step": 16995
    },
    {
      "epoch": 0.264659406695883,
      "grad_norm": 3.1316614151000977,
      "learning_rate": 7.7406138870225e-07,
      "loss": 0.7035,
      "step": 17000
    },
    {
      "epoch": 0.26473724769785234,
      "grad_norm": 3.28928804397583,
      "learning_rate": 7.73979449697645e-07,
      "loss": 0.8902,
      "step": 17005
    },
    {
      "epoch": 0.26481508869982173,
      "grad_norm": 3.257534980773926,
      "learning_rate": 7.738975106930401e-07,
      "loss": 0.666,
      "step": 17010
    },
    {
      "epoch": 0.2648929297017911,
      "grad_norm": 3.9892382621765137,
      "learning_rate": 7.738155716884352e-07,
      "loss": 0.7318,
      "step": 17015
    },
    {
      "epoch": 0.2649707707037605,
      "grad_norm": 4.157562732696533,
      "learning_rate": 7.737336326838301e-07,
      "loss": 0.7503,
      "step": 17020
    },
    {
      "epoch": 0.26504861170572985,
      "grad_norm": 3.515887498855591,
      "learning_rate": 7.736516936792251e-07,
      "loss": 0.7764,
      "step": 17025
    },
    {
      "epoch": 0.26512645270769924,
      "grad_norm": 6.801537990570068,
      "learning_rate": 7.735697546746202e-07,
      "loss": 0.7394,
      "step": 17030
    },
    {
      "epoch": 0.26520429370966864,
      "grad_norm": 5.358715534210205,
      "learning_rate": 7.734878156700152e-07,
      "loss": 0.7025,
      "step": 17035
    },
    {
      "epoch": 0.26528213471163803,
      "grad_norm": 3.9038212299346924,
      "learning_rate": 7.734058766654102e-07,
      "loss": 0.7226,
      "step": 17040
    },
    {
      "epoch": 0.26535997571360737,
      "grad_norm": 2.7967400550842285,
      "learning_rate": 7.733239376608053e-07,
      "loss": 0.7166,
      "step": 17045
    },
    {
      "epoch": 0.26543781671557676,
      "grad_norm": 4.307606220245361,
      "learning_rate": 7.732419986562003e-07,
      "loss": 0.8099,
      "step": 17050
    },
    {
      "epoch": 0.26551565771754615,
      "grad_norm": 3.752150058746338,
      "learning_rate": 7.731600596515954e-07,
      "loss": 0.7633,
      "step": 17055
    },
    {
      "epoch": 0.26559349871951554,
      "grad_norm": 4.0929412841796875,
      "learning_rate": 7.730781206469903e-07,
      "loss": 0.8023,
      "step": 17060
    },
    {
      "epoch": 0.2656713397214849,
      "grad_norm": 3.6266283988952637,
      "learning_rate": 7.729961816423853e-07,
      "loss": 0.8214,
      "step": 17065
    },
    {
      "epoch": 0.26574918072345427,
      "grad_norm": 2.470479726791382,
      "learning_rate": 7.729142426377804e-07,
      "loss": 0.7045,
      "step": 17070
    },
    {
      "epoch": 0.26582702172542366,
      "grad_norm": 3.274996519088745,
      "learning_rate": 7.728323036331755e-07,
      "loss": 0.7351,
      "step": 17075
    },
    {
      "epoch": 0.26590486272739305,
      "grad_norm": 3.2078683376312256,
      "learning_rate": 7.727503646285704e-07,
      "loss": 0.7526,
      "step": 17080
    },
    {
      "epoch": 0.2659827037293624,
      "grad_norm": 6.345730304718018,
      "learning_rate": 7.726684256239655e-07,
      "loss": 0.6919,
      "step": 17085
    },
    {
      "epoch": 0.2660605447313318,
      "grad_norm": 3.1593596935272217,
      "learning_rate": 7.725864866193606e-07,
      "loss": 0.7087,
      "step": 17090
    },
    {
      "epoch": 0.2661383857333012,
      "grad_norm": 4.418591022491455,
      "learning_rate": 7.725045476147555e-07,
      "loss": 0.8193,
      "step": 17095
    },
    {
      "epoch": 0.2662162267352705,
      "grad_norm": 3.5372109413146973,
      "learning_rate": 7.724226086101505e-07,
      "loss": 0.7493,
      "step": 17100
    },
    {
      "epoch": 0.2662940677372399,
      "grad_norm": 3.8144540786743164,
      "learning_rate": 7.723406696055456e-07,
      "loss": 0.9164,
      "step": 17105
    },
    {
      "epoch": 0.2663719087392093,
      "grad_norm": 3.1666834354400635,
      "learning_rate": 7.722587306009406e-07,
      "loss": 0.8379,
      "step": 17110
    },
    {
      "epoch": 0.2664497497411787,
      "grad_norm": 4.0041375160217285,
      "learning_rate": 7.721767915963356e-07,
      "loss": 0.7178,
      "step": 17115
    },
    {
      "epoch": 0.266527590743148,
      "grad_norm": 8.216415405273438,
      "learning_rate": 7.720948525917307e-07,
      "loss": 0.692,
      "step": 17120
    },
    {
      "epoch": 0.2666054317451174,
      "grad_norm": 7.888249397277832,
      "learning_rate": 7.720129135871257e-07,
      "loss": 0.8232,
      "step": 17125
    },
    {
      "epoch": 0.2666832727470868,
      "grad_norm": 7.171913146972656,
      "learning_rate": 7.719309745825207e-07,
      "loss": 0.8611,
      "step": 17130
    },
    {
      "epoch": 0.2667611137490562,
      "grad_norm": 3.7759389877319336,
      "learning_rate": 7.718490355779158e-07,
      "loss": 0.8469,
      "step": 17135
    },
    {
      "epoch": 0.26683895475102554,
      "grad_norm": 4.734642028808594,
      "learning_rate": 7.717670965733107e-07,
      "loss": 0.7777,
      "step": 17140
    },
    {
      "epoch": 0.26691679575299493,
      "grad_norm": 4.96793270111084,
      "learning_rate": 7.716851575687058e-07,
      "loss": 0.843,
      "step": 17145
    },
    {
      "epoch": 0.2669946367549643,
      "grad_norm": 6.541609764099121,
      "learning_rate": 7.716032185641008e-07,
      "loss": 0.8751,
      "step": 17150
    },
    {
      "epoch": 0.2670724777569337,
      "grad_norm": 4.108456134796143,
      "learning_rate": 7.715212795594959e-07,
      "loss": 0.8128,
      "step": 17155
    },
    {
      "epoch": 0.26715031875890305,
      "grad_norm": 3.626671314239502,
      "learning_rate": 7.714393405548909e-07,
      "loss": 0.7681,
      "step": 17160
    },
    {
      "epoch": 0.26722815976087244,
      "grad_norm": 4.041867733001709,
      "learning_rate": 7.71357401550286e-07,
      "loss": 0.7371,
      "step": 17165
    },
    {
      "epoch": 0.26730600076284183,
      "grad_norm": 3.3305304050445557,
      "learning_rate": 7.71275462545681e-07,
      "loss": 0.8206,
      "step": 17170
    },
    {
      "epoch": 0.26738384176481117,
      "grad_norm": 6.281455039978027,
      "learning_rate": 7.71193523541076e-07,
      "loss": 0.6028,
      "step": 17175
    },
    {
      "epoch": 0.26746168276678056,
      "grad_norm": 2.861084222793579,
      "learning_rate": 7.711115845364711e-07,
      "loss": 0.7768,
      "step": 17180
    },
    {
      "epoch": 0.26753952376874995,
      "grad_norm": 4.843693733215332,
      "learning_rate": 7.71029645531866e-07,
      "loss": 0.8696,
      "step": 17185
    },
    {
      "epoch": 0.26761736477071935,
      "grad_norm": 2.4090218544006348,
      "learning_rate": 7.70947706527261e-07,
      "loss": 0.729,
      "step": 17190
    },
    {
      "epoch": 0.2676952057726887,
      "grad_norm": 7.0648651123046875,
      "learning_rate": 7.708657675226561e-07,
      "loss": 0.7541,
      "step": 17195
    },
    {
      "epoch": 0.2677730467746581,
      "grad_norm": 5.795253276824951,
      "learning_rate": 7.707838285180512e-07,
      "loss": 0.7521,
      "step": 17200
    },
    {
      "epoch": 0.26785088777662747,
      "grad_norm": 2.898508310317993,
      "learning_rate": 7.707018895134461e-07,
      "loss": 0.6649,
      "step": 17205
    },
    {
      "epoch": 0.26792872877859686,
      "grad_norm": 4.25831937789917,
      "learning_rate": 7.706199505088412e-07,
      "loss": 0.7483,
      "step": 17210
    },
    {
      "epoch": 0.2680065697805662,
      "grad_norm": 3.3910627365112305,
      "learning_rate": 7.705380115042363e-07,
      "loss": 0.7649,
      "step": 17215
    },
    {
      "epoch": 0.2680844107825356,
      "grad_norm": 3.8831562995910645,
      "learning_rate": 7.704560724996312e-07,
      "loss": 0.9277,
      "step": 17220
    },
    {
      "epoch": 0.268162251784505,
      "grad_norm": 3.593287229537964,
      "learning_rate": 7.703741334950262e-07,
      "loss": 0.6956,
      "step": 17225
    },
    {
      "epoch": 0.26824009278647437,
      "grad_norm": 4.377844333648682,
      "learning_rate": 7.702921944904213e-07,
      "loss": 0.7818,
      "step": 17230
    },
    {
      "epoch": 0.2683179337884437,
      "grad_norm": 3.796743392944336,
      "learning_rate": 7.702102554858163e-07,
      "loss": 0.9001,
      "step": 17235
    },
    {
      "epoch": 0.2683957747904131,
      "grad_norm": 3.4302916526794434,
      "learning_rate": 7.701283164812113e-07,
      "loss": 0.7686,
      "step": 17240
    },
    {
      "epoch": 0.2684736157923825,
      "grad_norm": 3.4582180976867676,
      "learning_rate": 7.700463774766064e-07,
      "loss": 0.778,
      "step": 17245
    },
    {
      "epoch": 0.2685514567943519,
      "grad_norm": 3.298583507537842,
      "learning_rate": 7.699644384720014e-07,
      "loss": 0.6798,
      "step": 17250
    },
    {
      "epoch": 0.2686292977963212,
      "grad_norm": 3.020228624343872,
      "learning_rate": 7.698824994673965e-07,
      "loss": 0.7602,
      "step": 17255
    },
    {
      "epoch": 0.2687071387982906,
      "grad_norm": 3.2028045654296875,
      "learning_rate": 7.698005604627915e-07,
      "loss": 0.8621,
      "step": 17260
    },
    {
      "epoch": 0.26878497980026,
      "grad_norm": 5.402839660644531,
      "learning_rate": 7.697186214581865e-07,
      "loss": 0.784,
      "step": 17265
    },
    {
      "epoch": 0.26886282080222934,
      "grad_norm": 6.431519508361816,
      "learning_rate": 7.696366824535815e-07,
      "loss": 0.7682,
      "step": 17270
    },
    {
      "epoch": 0.26894066180419873,
      "grad_norm": 3.140960931777954,
      "learning_rate": 7.695547434489765e-07,
      "loss": 0.6265,
      "step": 17275
    },
    {
      "epoch": 0.2690185028061681,
      "grad_norm": 2.3430335521698,
      "learning_rate": 7.694728044443716e-07,
      "loss": 0.7718,
      "step": 17280
    },
    {
      "epoch": 0.2690963438081375,
      "grad_norm": 3.5215811729431152,
      "learning_rate": 7.693908654397666e-07,
      "loss": 0.802,
      "step": 17285
    },
    {
      "epoch": 0.26917418481010685,
      "grad_norm": 3.7115468978881836,
      "learning_rate": 7.693089264351617e-07,
      "loss": 0.7033,
      "step": 17290
    },
    {
      "epoch": 0.26925202581207625,
      "grad_norm": 4.185376167297363,
      "learning_rate": 7.692269874305567e-07,
      "loss": 0.7204,
      "step": 17295
    },
    {
      "epoch": 0.26932986681404564,
      "grad_norm": 6.7096123695373535,
      "learning_rate": 7.691450484259517e-07,
      "loss": 0.7501,
      "step": 17300
    },
    {
      "epoch": 0.26940770781601503,
      "grad_norm": 3.2766761779785156,
      "learning_rate": 7.690631094213467e-07,
      "loss": 0.7548,
      "step": 17305
    },
    {
      "epoch": 0.26948554881798437,
      "grad_norm": 2.9786577224731445,
      "learning_rate": 7.689811704167417e-07,
      "loss": 0.944,
      "step": 17310
    },
    {
      "epoch": 0.26956338981995376,
      "grad_norm": 4.937072277069092,
      "learning_rate": 7.688992314121367e-07,
      "loss": 0.728,
      "step": 17315
    },
    {
      "epoch": 0.26964123082192315,
      "grad_norm": 3.2550084590911865,
      "learning_rate": 7.688172924075318e-07,
      "loss": 0.8478,
      "step": 17320
    },
    {
      "epoch": 0.26971907182389254,
      "grad_norm": 11.952690124511719,
      "learning_rate": 7.687353534029269e-07,
      "loss": 0.7935,
      "step": 17325
    },
    {
      "epoch": 0.2697969128258619,
      "grad_norm": 8.415304183959961,
      "learning_rate": 7.686534143983218e-07,
      "loss": 0.8136,
      "step": 17330
    },
    {
      "epoch": 0.26987475382783127,
      "grad_norm": 3.8239519596099854,
      "learning_rate": 7.685714753937169e-07,
      "loss": 0.9237,
      "step": 17335
    },
    {
      "epoch": 0.26995259482980066,
      "grad_norm": 6.529261112213135,
      "learning_rate": 7.68489536389112e-07,
      "loss": 0.8137,
      "step": 17340
    },
    {
      "epoch": 0.27003043583177,
      "grad_norm": 4.317914009094238,
      "learning_rate": 7.684075973845068e-07,
      "loss": 0.9643,
      "step": 17345
    },
    {
      "epoch": 0.2701082768337394,
      "grad_norm": 6.890023708343506,
      "learning_rate": 7.683256583799019e-07,
      "loss": 0.6974,
      "step": 17350
    },
    {
      "epoch": 0.2701861178357088,
      "grad_norm": 3.332381248474121,
      "learning_rate": 7.68243719375297e-07,
      "loss": 0.8654,
      "step": 17355
    },
    {
      "epoch": 0.2702639588376782,
      "grad_norm": 4.557866096496582,
      "learning_rate": 7.68161780370692e-07,
      "loss": 0.6106,
      "step": 17360
    },
    {
      "epoch": 0.2703417998396475,
      "grad_norm": 4.092104434967041,
      "learning_rate": 7.68079841366087e-07,
      "loss": 0.7621,
      "step": 17365
    },
    {
      "epoch": 0.2704196408416169,
      "grad_norm": 3.59782075881958,
      "learning_rate": 7.679979023614821e-07,
      "loss": 0.8053,
      "step": 17370
    },
    {
      "epoch": 0.2704974818435863,
      "grad_norm": 4.809222221374512,
      "learning_rate": 7.679159633568771e-07,
      "loss": 0.871,
      "step": 17375
    },
    {
      "epoch": 0.2705753228455557,
      "grad_norm": 3.0703799724578857,
      "learning_rate": 7.678340243522722e-07,
      "loss": 0.8575,
      "step": 17380
    },
    {
      "epoch": 0.270653163847525,
      "grad_norm": 3.1767992973327637,
      "learning_rate": 7.677520853476671e-07,
      "loss": 0.8037,
      "step": 17385
    },
    {
      "epoch": 0.2707310048494944,
      "grad_norm": 10.665190696716309,
      "learning_rate": 7.676701463430622e-07,
      "loss": 0.7498,
      "step": 17390
    },
    {
      "epoch": 0.2708088458514638,
      "grad_norm": 2.9556710720062256,
      "learning_rate": 7.675882073384572e-07,
      "loss": 0.7574,
      "step": 17395
    },
    {
      "epoch": 0.2708866868534332,
      "grad_norm": 5.384457588195801,
      "learning_rate": 7.675062683338522e-07,
      "loss": 0.7484,
      "step": 17400
    },
    {
      "epoch": 0.27096452785540254,
      "grad_norm": 3.542102575302124,
      "learning_rate": 7.674243293292473e-07,
      "loss": 0.7361,
      "step": 17405
    },
    {
      "epoch": 0.27104236885737193,
      "grad_norm": 3.0414071083068848,
      "learning_rate": 7.673423903246423e-07,
      "loss": 0.7546,
      "step": 17410
    },
    {
      "epoch": 0.2711202098593413,
      "grad_norm": 3.4305264949798584,
      "learning_rate": 7.672604513200374e-07,
      "loss": 0.8032,
      "step": 17415
    },
    {
      "epoch": 0.2711980508613107,
      "grad_norm": 3.261608123779297,
      "learning_rate": 7.671785123154324e-07,
      "loss": 0.7438,
      "step": 17420
    },
    {
      "epoch": 0.27127589186328005,
      "grad_norm": 3.1833226680755615,
      "learning_rate": 7.670965733108273e-07,
      "loss": 0.7677,
      "step": 17425
    },
    {
      "epoch": 0.27135373286524944,
      "grad_norm": 5.976661205291748,
      "learning_rate": 7.670146343062224e-07,
      "loss": 0.7372,
      "step": 17430
    },
    {
      "epoch": 0.27143157386721883,
      "grad_norm": 16.816925048828125,
      "learning_rate": 7.669326953016175e-07,
      "loss": 0.7339,
      "step": 17435
    },
    {
      "epoch": 0.27150941486918817,
      "grad_norm": 4.602113723754883,
      "learning_rate": 7.668507562970124e-07,
      "loss": 0.7989,
      "step": 17440
    },
    {
      "epoch": 0.27158725587115756,
      "grad_norm": 5.182931423187256,
      "learning_rate": 7.667688172924075e-07,
      "loss": 0.7642,
      "step": 17445
    },
    {
      "epoch": 0.27166509687312695,
      "grad_norm": 5.678241729736328,
      "learning_rate": 7.666868782878026e-07,
      "loss": 0.7479,
      "step": 17450
    },
    {
      "epoch": 0.27174293787509635,
      "grad_norm": 3.3905954360961914,
      "learning_rate": 7.666049392831975e-07,
      "loss": 0.7594,
      "step": 17455
    },
    {
      "epoch": 0.2718207788770657,
      "grad_norm": 8.445892333984375,
      "learning_rate": 7.665230002785926e-07,
      "loss": 0.8912,
      "step": 17460
    },
    {
      "epoch": 0.2718986198790351,
      "grad_norm": 2.7368478775024414,
      "learning_rate": 7.664410612739876e-07,
      "loss": 0.7081,
      "step": 17465
    },
    {
      "epoch": 0.27197646088100447,
      "grad_norm": 3.5980234146118164,
      "learning_rate": 7.663591222693826e-07,
      "loss": 0.7346,
      "step": 17470
    },
    {
      "epoch": 0.27205430188297386,
      "grad_norm": 6.195345878601074,
      "learning_rate": 7.662771832647776e-07,
      "loss": 0.7162,
      "step": 17475
    },
    {
      "epoch": 0.2721321428849432,
      "grad_norm": 8.303228378295898,
      "learning_rate": 7.661952442601727e-07,
      "loss": 0.8932,
      "step": 17480
    },
    {
      "epoch": 0.2722099838869126,
      "grad_norm": 3.866856813430786,
      "learning_rate": 7.661133052555677e-07,
      "loss": 0.7422,
      "step": 17485
    },
    {
      "epoch": 0.272287824888882,
      "grad_norm": 3.056715965270996,
      "learning_rate": 7.660313662509627e-07,
      "loss": 0.7901,
      "step": 17490
    },
    {
      "epoch": 0.27236566589085137,
      "grad_norm": 4.668761730194092,
      "learning_rate": 7.659494272463578e-07,
      "loss": 0.7008,
      "step": 17495
    },
    {
      "epoch": 0.2724435068928207,
      "grad_norm": 6.530035495758057,
      "learning_rate": 7.658674882417528e-07,
      "loss": 0.8165,
      "step": 17500
    },
    {
      "epoch": 0.2725213478947901,
      "grad_norm": 2.7709784507751465,
      "learning_rate": 7.657855492371479e-07,
      "loss": 0.7451,
      "step": 17505
    },
    {
      "epoch": 0.2725991888967595,
      "grad_norm": 2.8162500858306885,
      "learning_rate": 7.657036102325428e-07,
      "loss": 0.6419,
      "step": 17510
    },
    {
      "epoch": 0.27267702989872883,
      "grad_norm": 5.088579177856445,
      "learning_rate": 7.656216712279379e-07,
      "loss": 0.7556,
      "step": 17515
    },
    {
      "epoch": 0.2727548709006982,
      "grad_norm": 6.667660713195801,
      "learning_rate": 7.655397322233329e-07,
      "loss": 0.811,
      "step": 17520
    },
    {
      "epoch": 0.2728327119026676,
      "grad_norm": 3.972935199737549,
      "learning_rate": 7.65457793218728e-07,
      "loss": 0.7962,
      "step": 17525
    },
    {
      "epoch": 0.272910552904637,
      "grad_norm": 5.444567680358887,
      "learning_rate": 7.65375854214123e-07,
      "loss": 0.764,
      "step": 17530
    },
    {
      "epoch": 0.27298839390660634,
      "grad_norm": 5.834167957305908,
      "learning_rate": 7.65293915209518e-07,
      "loss": 0.7616,
      "step": 17535
    },
    {
      "epoch": 0.27306623490857573,
      "grad_norm": 3.4438259601593018,
      "learning_rate": 7.652119762049131e-07,
      "loss": 0.7918,
      "step": 17540
    },
    {
      "epoch": 0.2731440759105451,
      "grad_norm": 13.500309944152832,
      "learning_rate": 7.651300372003081e-07,
      "loss": 0.6592,
      "step": 17545
    },
    {
      "epoch": 0.2732219169125145,
      "grad_norm": 6.822749137878418,
      "learning_rate": 7.65048098195703e-07,
      "loss": 0.7099,
      "step": 17550
    },
    {
      "epoch": 0.27329975791448385,
      "grad_norm": 3.325993537902832,
      "learning_rate": 7.649661591910981e-07,
      "loss": 0.702,
      "step": 17555
    },
    {
      "epoch": 0.27337759891645325,
      "grad_norm": 3.109536647796631,
      "learning_rate": 7.648842201864932e-07,
      "loss": 0.6603,
      "step": 17560
    },
    {
      "epoch": 0.27345543991842264,
      "grad_norm": 4.584033012390137,
      "learning_rate": 7.648022811818881e-07,
      "loss": 0.7757,
      "step": 17565
    },
    {
      "epoch": 0.27353328092039203,
      "grad_norm": 4.5637383460998535,
      "learning_rate": 7.647203421772832e-07,
      "loss": 0.7424,
      "step": 17570
    },
    {
      "epoch": 0.27361112192236137,
      "grad_norm": 3.220829486846924,
      "learning_rate": 7.646384031726783e-07,
      "loss": 0.7742,
      "step": 17575
    },
    {
      "epoch": 0.27368896292433076,
      "grad_norm": 2.887474775314331,
      "learning_rate": 7.645564641680733e-07,
      "loss": 0.6854,
      "step": 17580
    },
    {
      "epoch": 0.27376680392630015,
      "grad_norm": 3.1210684776306152,
      "learning_rate": 7.644745251634683e-07,
      "loss": 0.7429,
      "step": 17585
    },
    {
      "epoch": 0.27384464492826954,
      "grad_norm": 3.1878061294555664,
      "learning_rate": 7.643925861588633e-07,
      "loss": 0.7759,
      "step": 17590
    },
    {
      "epoch": 0.2739224859302389,
      "grad_norm": 3.132434368133545,
      "learning_rate": 7.643106471542583e-07,
      "loss": 0.8279,
      "step": 17595
    },
    {
      "epoch": 0.27400032693220827,
      "grad_norm": 7.325407981872559,
      "learning_rate": 7.642287081496533e-07,
      "loss": 0.8546,
      "step": 17600
    },
    {
      "epoch": 0.27407816793417766,
      "grad_norm": 2.864562749862671,
      "learning_rate": 7.641467691450484e-07,
      "loss": 0.7245,
      "step": 17605
    },
    {
      "epoch": 0.274156008936147,
      "grad_norm": 4.206888675689697,
      "learning_rate": 7.640648301404434e-07,
      "loss": 0.8363,
      "step": 17610
    },
    {
      "epoch": 0.2742338499381164,
      "grad_norm": 3.8253650665283203,
      "learning_rate": 7.639828911358385e-07,
      "loss": 0.8165,
      "step": 17615
    },
    {
      "epoch": 0.2743116909400858,
      "grad_norm": 4.832096099853516,
      "learning_rate": 7.639009521312335e-07,
      "loss": 0.799,
      "step": 17620
    },
    {
      "epoch": 0.2743895319420552,
      "grad_norm": 3.5117223262786865,
      "learning_rate": 7.638190131266285e-07,
      "loss": 0.7837,
      "step": 17625
    },
    {
      "epoch": 0.2744673729440245,
      "grad_norm": 6.5699286460876465,
      "learning_rate": 7.637370741220235e-07,
      "loss": 0.6992,
      "step": 17630
    },
    {
      "epoch": 0.2745452139459939,
      "grad_norm": 4.263813018798828,
      "learning_rate": 7.636551351174185e-07,
      "loss": 0.8177,
      "step": 17635
    },
    {
      "epoch": 0.2746230549479633,
      "grad_norm": 3.31231427192688,
      "learning_rate": 7.635731961128136e-07,
      "loss": 0.7273,
      "step": 17640
    },
    {
      "epoch": 0.2747008959499327,
      "grad_norm": 3.007326602935791,
      "learning_rate": 7.634912571082086e-07,
      "loss": 0.7857,
      "step": 17645
    },
    {
      "epoch": 0.274778736951902,
      "grad_norm": 3.9053094387054443,
      "learning_rate": 7.634093181036037e-07,
      "loss": 0.6829,
      "step": 17650
    },
    {
      "epoch": 0.2748565779538714,
      "grad_norm": 4.185614585876465,
      "learning_rate": 7.633273790989987e-07,
      "loss": 0.6869,
      "step": 17655
    },
    {
      "epoch": 0.2749344189558408,
      "grad_norm": 2.9816734790802,
      "learning_rate": 7.632454400943937e-07,
      "loss": 0.7999,
      "step": 17660
    },
    {
      "epoch": 0.2750122599578102,
      "grad_norm": 2.800800085067749,
      "learning_rate": 7.631635010897888e-07,
      "loss": 0.7201,
      "step": 17665
    },
    {
      "epoch": 0.27509010095977954,
      "grad_norm": 4.0509033203125,
      "learning_rate": 7.630815620851838e-07,
      "loss": 0.7324,
      "step": 17670
    },
    {
      "epoch": 0.27516794196174893,
      "grad_norm": 4.819879531860352,
      "learning_rate": 7.629996230805787e-07,
      "loss": 0.7498,
      "step": 17675
    },
    {
      "epoch": 0.2752457829637183,
      "grad_norm": 4.175679683685303,
      "learning_rate": 7.629176840759738e-07,
      "loss": 0.7992,
      "step": 17680
    },
    {
      "epoch": 0.2753236239656877,
      "grad_norm": 5.96135139465332,
      "learning_rate": 7.628357450713689e-07,
      "loss": 0.608,
      "step": 17685
    },
    {
      "epoch": 0.27540146496765705,
      "grad_norm": 4.20062255859375,
      "learning_rate": 7.627538060667638e-07,
      "loss": 0.8923,
      "step": 17690
    },
    {
      "epoch": 0.27547930596962644,
      "grad_norm": 7.188648223876953,
      "learning_rate": 7.626718670621589e-07,
      "loss": 0.877,
      "step": 17695
    },
    {
      "epoch": 0.27555714697159583,
      "grad_norm": 6.603199481964111,
      "learning_rate": 7.62589928057554e-07,
      "loss": 0.733,
      "step": 17700
    },
    {
      "epoch": 0.27563498797356517,
      "grad_norm": 4.94620418548584,
      "learning_rate": 7.62507989052949e-07,
      "loss": 0.8015,
      "step": 17705
    },
    {
      "epoch": 0.27571282897553456,
      "grad_norm": 3.5170164108276367,
      "learning_rate": 7.624260500483439e-07,
      "loss": 0.6238,
      "step": 17710
    },
    {
      "epoch": 0.27579066997750395,
      "grad_norm": 3.5444560050964355,
      "learning_rate": 7.62344111043739e-07,
      "loss": 0.8081,
      "step": 17715
    },
    {
      "epoch": 0.27586851097947335,
      "grad_norm": 3.9345004558563232,
      "learning_rate": 7.62262172039134e-07,
      "loss": 0.67,
      "step": 17720
    },
    {
      "epoch": 0.2759463519814427,
      "grad_norm": 3.7450127601623535,
      "learning_rate": 7.62180233034529e-07,
      "loss": 0.8191,
      "step": 17725
    },
    {
      "epoch": 0.2760241929834121,
      "grad_norm": 5.089782238006592,
      "learning_rate": 7.620982940299241e-07,
      "loss": 0.7944,
      "step": 17730
    },
    {
      "epoch": 0.27610203398538147,
      "grad_norm": 3.7452938556671143,
      "learning_rate": 7.620163550253191e-07,
      "loss": 0.8261,
      "step": 17735
    },
    {
      "epoch": 0.27617987498735086,
      "grad_norm": 2.857208013534546,
      "learning_rate": 7.619344160207142e-07,
      "loss": 0.7092,
      "step": 17740
    },
    {
      "epoch": 0.2762577159893202,
      "grad_norm": 3.1661722660064697,
      "learning_rate": 7.618524770161092e-07,
      "loss": 0.9596,
      "step": 17745
    },
    {
      "epoch": 0.2763355569912896,
      "grad_norm": 4.01740837097168,
      "learning_rate": 7.617705380115041e-07,
      "loss": 0.7322,
      "step": 17750
    },
    {
      "epoch": 0.276413397993259,
      "grad_norm": 3.3057632446289062,
      "learning_rate": 7.616885990068992e-07,
      "loss": 0.7321,
      "step": 17755
    },
    {
      "epoch": 0.27649123899522837,
      "grad_norm": 4.020196437835693,
      "learning_rate": 7.616066600022943e-07,
      "loss": 0.6533,
      "step": 17760
    },
    {
      "epoch": 0.2765690799971977,
      "grad_norm": 3.5413827896118164,
      "learning_rate": 7.615247209976893e-07,
      "loss": 0.7581,
      "step": 17765
    },
    {
      "epoch": 0.2766469209991671,
      "grad_norm": 4.459136486053467,
      "learning_rate": 7.614427819930843e-07,
      "loss": 0.7323,
      "step": 17770
    },
    {
      "epoch": 0.2767247620011365,
      "grad_norm": 3.8716177940368652,
      "learning_rate": 7.613608429884794e-07,
      "loss": 0.8683,
      "step": 17775
    },
    {
      "epoch": 0.27680260300310583,
      "grad_norm": 2.5693435668945312,
      "learning_rate": 7.612789039838744e-07,
      "loss": 0.7723,
      "step": 17780
    },
    {
      "epoch": 0.2768804440050752,
      "grad_norm": 6.711716175079346,
      "learning_rate": 7.611969649792694e-07,
      "loss": 0.7423,
      "step": 17785
    },
    {
      "epoch": 0.2769582850070446,
      "grad_norm": 7.211732387542725,
      "learning_rate": 7.611150259746644e-07,
      "loss": 0.7655,
      "step": 17790
    },
    {
      "epoch": 0.277036126009014,
      "grad_norm": 3.0362424850463867,
      "learning_rate": 7.610330869700595e-07,
      "loss": 0.7852,
      "step": 17795
    },
    {
      "epoch": 0.27711396701098334,
      "grad_norm": 7.21616268157959,
      "learning_rate": 7.609511479654544e-07,
      "loss": 0.6979,
      "step": 17800
    },
    {
      "epoch": 0.27719180801295273,
      "grad_norm": 2.9571242332458496,
      "learning_rate": 7.608692089608495e-07,
      "loss": 0.736,
      "step": 17805
    },
    {
      "epoch": 0.2772696490149221,
      "grad_norm": 3.2685022354125977,
      "learning_rate": 7.607872699562446e-07,
      "loss": 0.7699,
      "step": 17810
    },
    {
      "epoch": 0.2773474900168915,
      "grad_norm": 3.817214012145996,
      "learning_rate": 7.607053309516395e-07,
      "loss": 0.7695,
      "step": 17815
    },
    {
      "epoch": 0.27742533101886085,
      "grad_norm": 3.7128727436065674,
      "learning_rate": 7.606233919470346e-07,
      "loss": 0.8229,
      "step": 17820
    },
    {
      "epoch": 0.27750317202083025,
      "grad_norm": 4.211459636688232,
      "learning_rate": 7.605414529424297e-07,
      "loss": 0.7276,
      "step": 17825
    },
    {
      "epoch": 0.27758101302279964,
      "grad_norm": 3.2515077590942383,
      "learning_rate": 7.604595139378247e-07,
      "loss": 0.7933,
      "step": 17830
    },
    {
      "epoch": 0.27765885402476903,
      "grad_norm": 3.9488120079040527,
      "learning_rate": 7.603775749332196e-07,
      "loss": 0.6916,
      "step": 17835
    },
    {
      "epoch": 0.27773669502673837,
      "grad_norm": 4.689972877502441,
      "learning_rate": 7.602956359286147e-07,
      "loss": 0.783,
      "step": 17840
    },
    {
      "epoch": 0.27781453602870776,
      "grad_norm": 3.274508476257324,
      "learning_rate": 7.602136969240097e-07,
      "loss": 0.6912,
      "step": 17845
    },
    {
      "epoch": 0.27789237703067715,
      "grad_norm": 3.4465394020080566,
      "learning_rate": 7.601317579194048e-07,
      "loss": 0.7393,
      "step": 17850
    },
    {
      "epoch": 0.27797021803264654,
      "grad_norm": 2.9954683780670166,
      "learning_rate": 7.600498189147998e-07,
      "loss": 0.6276,
      "step": 17855
    },
    {
      "epoch": 0.2780480590346159,
      "grad_norm": 6.088066101074219,
      "learning_rate": 7.599678799101948e-07,
      "loss": 0.7727,
      "step": 17860
    },
    {
      "epoch": 0.27812590003658527,
      "grad_norm": 2.8900184631347656,
      "learning_rate": 7.598859409055899e-07,
      "loss": 0.7322,
      "step": 17865
    },
    {
      "epoch": 0.27820374103855466,
      "grad_norm": 5.254889965057373,
      "learning_rate": 7.598040019009849e-07,
      "loss": 0.8718,
      "step": 17870
    },
    {
      "epoch": 0.278281582040524,
      "grad_norm": 3.176635265350342,
      "learning_rate": 7.597220628963798e-07,
      "loss": 0.6775,
      "step": 17875
    },
    {
      "epoch": 0.2783594230424934,
      "grad_norm": 4.2815260887146,
      "learning_rate": 7.596401238917749e-07,
      "loss": 0.7797,
      "step": 17880
    },
    {
      "epoch": 0.2784372640444628,
      "grad_norm": 2.851503372192383,
      "learning_rate": 7.5955818488717e-07,
      "loss": 0.6992,
      "step": 17885
    },
    {
      "epoch": 0.2785151050464322,
      "grad_norm": 4.479244232177734,
      "learning_rate": 7.59476245882565e-07,
      "loss": 0.8804,
      "step": 17890
    },
    {
      "epoch": 0.2785929460484015,
      "grad_norm": 3.6790082454681396,
      "learning_rate": 7.5939430687796e-07,
      "loss": 0.7211,
      "step": 17895
    },
    {
      "epoch": 0.2786707870503709,
      "grad_norm": 4.7184600830078125,
      "learning_rate": 7.593123678733551e-07,
      "loss": 0.862,
      "step": 17900
    },
    {
      "epoch": 0.2787486280523403,
      "grad_norm": 3.2201459407806396,
      "learning_rate": 7.592304288687502e-07,
      "loss": 0.8146,
      "step": 17905
    },
    {
      "epoch": 0.2788264690543097,
      "grad_norm": 3.8004696369171143,
      "learning_rate": 7.591484898641451e-07,
      "loss": 0.7318,
      "step": 17910
    },
    {
      "epoch": 0.278904310056279,
      "grad_norm": 5.019532203674316,
      "learning_rate": 7.590665508595401e-07,
      "loss": 0.8417,
      "step": 17915
    },
    {
      "epoch": 0.2789821510582484,
      "grad_norm": 3.6091842651367188,
      "learning_rate": 7.589846118549352e-07,
      "loss": 0.7375,
      "step": 17920
    },
    {
      "epoch": 0.2790599920602178,
      "grad_norm": 4.4064226150512695,
      "learning_rate": 7.589026728503301e-07,
      "loss": 0.7558,
      "step": 17925
    },
    {
      "epoch": 0.2791378330621872,
      "grad_norm": 2.371734619140625,
      "learning_rate": 7.588207338457252e-07,
      "loss": 0.6613,
      "step": 17930
    },
    {
      "epoch": 0.27921567406415654,
      "grad_norm": 10.808391571044922,
      "learning_rate": 7.587387948411203e-07,
      "loss": 0.879,
      "step": 17935
    },
    {
      "epoch": 0.27929351506612593,
      "grad_norm": 5.366029262542725,
      "learning_rate": 7.586568558365153e-07,
      "loss": 0.7484,
      "step": 17940
    },
    {
      "epoch": 0.2793713560680953,
      "grad_norm": 5.2202911376953125,
      "learning_rate": 7.585749168319103e-07,
      "loss": 0.6601,
      "step": 17945
    },
    {
      "epoch": 0.27944919707006466,
      "grad_norm": 5.043264865875244,
      "learning_rate": 7.584929778273054e-07,
      "loss": 0.8133,
      "step": 17950
    },
    {
      "epoch": 0.27952703807203405,
      "grad_norm": 6.051628112792969,
      "learning_rate": 7.584110388227003e-07,
      "loss": 0.8647,
      "step": 17955
    },
    {
      "epoch": 0.27960487907400344,
      "grad_norm": 3.5219998359680176,
      "learning_rate": 7.583290998180953e-07,
      "loss": 0.7895,
      "step": 17960
    },
    {
      "epoch": 0.27968272007597283,
      "grad_norm": 4.5519232749938965,
      "learning_rate": 7.582471608134904e-07,
      "loss": 0.8633,
      "step": 17965
    },
    {
      "epoch": 0.27976056107794217,
      "grad_norm": 3.4150590896606445,
      "learning_rate": 7.581652218088854e-07,
      "loss": 0.7193,
      "step": 17970
    },
    {
      "epoch": 0.27983840207991156,
      "grad_norm": 4.484484672546387,
      "learning_rate": 7.580832828042805e-07,
      "loss": 0.7381,
      "step": 17975
    },
    {
      "epoch": 0.27991624308188096,
      "grad_norm": 4.187557220458984,
      "learning_rate": 7.580013437996755e-07,
      "loss": 0.7605,
      "step": 17980
    },
    {
      "epoch": 0.27999408408385035,
      "grad_norm": 3.3712828159332275,
      "learning_rate": 7.579194047950705e-07,
      "loss": 0.718,
      "step": 17985
    },
    {
      "epoch": 0.2800719250858197,
      "grad_norm": 4.786652565002441,
      "learning_rate": 7.578374657904656e-07,
      "loss": 0.771,
      "step": 17990
    },
    {
      "epoch": 0.2801497660877891,
      "grad_norm": 3.188917875289917,
      "learning_rate": 7.577555267858605e-07,
      "loss": 0.8198,
      "step": 17995
    },
    {
      "epoch": 0.28022760708975847,
      "grad_norm": 3.1383514404296875,
      "learning_rate": 7.576735877812555e-07,
      "loss": 0.8076,
      "step": 18000
    },
    {
      "epoch": 0.28030544809172786,
      "grad_norm": 3.447678327560425,
      "learning_rate": 7.575916487766506e-07,
      "loss": 0.7277,
      "step": 18005
    },
    {
      "epoch": 0.2803832890936972,
      "grad_norm": 3.8167226314544678,
      "learning_rate": 7.575097097720457e-07,
      "loss": 0.6305,
      "step": 18010
    },
    {
      "epoch": 0.2804611300956666,
      "grad_norm": 5.272345066070557,
      "learning_rate": 7.574277707674407e-07,
      "loss": 0.8302,
      "step": 18015
    },
    {
      "epoch": 0.280538971097636,
      "grad_norm": 3.4920878410339355,
      "learning_rate": 7.573458317628357e-07,
      "loss": 0.8639,
      "step": 18020
    },
    {
      "epoch": 0.2806168120996054,
      "grad_norm": 3.8000481128692627,
      "learning_rate": 7.572638927582308e-07,
      "loss": 0.6455,
      "step": 18025
    },
    {
      "epoch": 0.2806946531015747,
      "grad_norm": 5.670722007751465,
      "learning_rate": 7.571819537536259e-07,
      "loss": 0.7034,
      "step": 18030
    },
    {
      "epoch": 0.2807724941035441,
      "grad_norm": 6.655512809753418,
      "learning_rate": 7.571000147490207e-07,
      "loss": 0.8541,
      "step": 18035
    },
    {
      "epoch": 0.2808503351055135,
      "grad_norm": 4.9885573387146,
      "learning_rate": 7.570180757444158e-07,
      "loss": 0.7457,
      "step": 18040
    },
    {
      "epoch": 0.28092817610748283,
      "grad_norm": 3.309530019760132,
      "learning_rate": 7.569361367398109e-07,
      "loss": 0.7874,
      "step": 18045
    },
    {
      "epoch": 0.2810060171094522,
      "grad_norm": 2.9954628944396973,
      "learning_rate": 7.568541977352058e-07,
      "loss": 0.8728,
      "step": 18050
    },
    {
      "epoch": 0.2810838581114216,
      "grad_norm": 9.354890823364258,
      "learning_rate": 7.567722587306009e-07,
      "loss": 0.7902,
      "step": 18055
    },
    {
      "epoch": 0.281161699113391,
      "grad_norm": 6.805416107177734,
      "learning_rate": 7.56690319725996e-07,
      "loss": 0.8763,
      "step": 18060
    },
    {
      "epoch": 0.28123954011536034,
      "grad_norm": 11.965883255004883,
      "learning_rate": 7.56608380721391e-07,
      "loss": 0.8265,
      "step": 18065
    },
    {
      "epoch": 0.28131738111732973,
      "grad_norm": 3.0321974754333496,
      "learning_rate": 7.56526441716786e-07,
      "loss": 0.7889,
      "step": 18070
    },
    {
      "epoch": 0.2813952221192991,
      "grad_norm": 3.4996132850646973,
      "learning_rate": 7.56444502712181e-07,
      "loss": 0.7354,
      "step": 18075
    },
    {
      "epoch": 0.2814730631212685,
      "grad_norm": 8.6669921875,
      "learning_rate": 7.56362563707576e-07,
      "loss": 0.7775,
      "step": 18080
    },
    {
      "epoch": 0.28155090412323786,
      "grad_norm": 6.530038356781006,
      "learning_rate": 7.56280624702971e-07,
      "loss": 0.7778,
      "step": 18085
    },
    {
      "epoch": 0.28162874512520725,
      "grad_norm": 2.507265329360962,
      "learning_rate": 7.561986856983661e-07,
      "loss": 0.7254,
      "step": 18090
    },
    {
      "epoch": 0.28170658612717664,
      "grad_norm": 3.001185655593872,
      "learning_rate": 7.561167466937611e-07,
      "loss": 0.7498,
      "step": 18095
    },
    {
      "epoch": 0.28178442712914603,
      "grad_norm": 4.192236423492432,
      "learning_rate": 7.560348076891562e-07,
      "loss": 0.7007,
      "step": 18100
    },
    {
      "epoch": 0.28186226813111537,
      "grad_norm": 6.101157188415527,
      "learning_rate": 7.559528686845512e-07,
      "loss": 0.7048,
      "step": 18105
    },
    {
      "epoch": 0.28194010913308476,
      "grad_norm": 4.111788749694824,
      "learning_rate": 7.558709296799462e-07,
      "loss": 0.7734,
      "step": 18110
    },
    {
      "epoch": 0.28201795013505415,
      "grad_norm": 3.0076723098754883,
      "learning_rate": 7.557889906753412e-07,
      "loss": 0.7345,
      "step": 18115
    },
    {
      "epoch": 0.2820957911370235,
      "grad_norm": 2.860314130783081,
      "learning_rate": 7.557070516707363e-07,
      "loss": 0.7199,
      "step": 18120
    },
    {
      "epoch": 0.2821736321389929,
      "grad_norm": 3.355159044265747,
      "learning_rate": 7.556251126661312e-07,
      "loss": 0.7313,
      "step": 18125
    },
    {
      "epoch": 0.2822514731409623,
      "grad_norm": 5.150790691375732,
      "learning_rate": 7.555431736615263e-07,
      "loss": 0.643,
      "step": 18130
    },
    {
      "epoch": 0.28232931414293166,
      "grad_norm": 3.4499988555908203,
      "learning_rate": 7.554612346569214e-07,
      "loss": 0.8814,
      "step": 18135
    },
    {
      "epoch": 0.282407155144901,
      "grad_norm": 4.242376804351807,
      "learning_rate": 7.553792956523164e-07,
      "loss": 0.7596,
      "step": 18140
    },
    {
      "epoch": 0.2824849961468704,
      "grad_norm": 3.851548433303833,
      "learning_rate": 7.552973566477114e-07,
      "loss": 0.736,
      "step": 18145
    },
    {
      "epoch": 0.2825628371488398,
      "grad_norm": 6.2824602127075195,
      "learning_rate": 7.552154176431065e-07,
      "loss": 0.759,
      "step": 18150
    },
    {
      "epoch": 0.2826406781508092,
      "grad_norm": 5.529384613037109,
      "learning_rate": 7.551334786385016e-07,
      "loss": 0.6954,
      "step": 18155
    },
    {
      "epoch": 0.2827185191527785,
      "grad_norm": 4.210619926452637,
      "learning_rate": 7.550515396338964e-07,
      "loss": 0.8275,
      "step": 18160
    },
    {
      "epoch": 0.2827963601547479,
      "grad_norm": 4.350375175476074,
      "learning_rate": 7.549696006292915e-07,
      "loss": 0.7411,
      "step": 18165
    },
    {
      "epoch": 0.2828742011567173,
      "grad_norm": 5.59158992767334,
      "learning_rate": 7.548876616246866e-07,
      "loss": 0.8171,
      "step": 18170
    },
    {
      "epoch": 0.2829520421586867,
      "grad_norm": 5.024247169494629,
      "learning_rate": 7.548057226200815e-07,
      "loss": 0.7903,
      "step": 18175
    },
    {
      "epoch": 0.283029883160656,
      "grad_norm": 2.6078712940216064,
      "learning_rate": 7.547237836154766e-07,
      "loss": 0.6803,
      "step": 18180
    },
    {
      "epoch": 0.2831077241626254,
      "grad_norm": 4.355838775634766,
      "learning_rate": 7.546418446108717e-07,
      "loss": 0.7313,
      "step": 18185
    },
    {
      "epoch": 0.2831855651645948,
      "grad_norm": 3.1983373165130615,
      "learning_rate": 7.545599056062667e-07,
      "loss": 0.7433,
      "step": 18190
    },
    {
      "epoch": 0.2832634061665642,
      "grad_norm": 2.885439395904541,
      "learning_rate": 7.544779666016617e-07,
      "loss": 0.6819,
      "step": 18195
    },
    {
      "epoch": 0.28334124716853354,
      "grad_norm": 9.0044527053833,
      "learning_rate": 7.543960275970567e-07,
      "loss": 0.8062,
      "step": 18200
    },
    {
      "epoch": 0.28341908817050293,
      "grad_norm": 4.429920673370361,
      "learning_rate": 7.543140885924517e-07,
      "loss": 0.7271,
      "step": 18205
    },
    {
      "epoch": 0.2834969291724723,
      "grad_norm": 3.505737543106079,
      "learning_rate": 7.542321495878468e-07,
      "loss": 0.7338,
      "step": 18210
    },
    {
      "epoch": 0.28357477017444166,
      "grad_norm": 3.533663272857666,
      "learning_rate": 7.541502105832418e-07,
      "loss": 0.8234,
      "step": 18215
    },
    {
      "epoch": 0.28365261117641105,
      "grad_norm": 3.877680540084839,
      "learning_rate": 7.540682715786368e-07,
      "loss": 0.7557,
      "step": 18220
    },
    {
      "epoch": 0.28373045217838044,
      "grad_norm": 2.824127435684204,
      "learning_rate": 7.539863325740319e-07,
      "loss": 0.6762,
      "step": 18225
    },
    {
      "epoch": 0.28380829318034984,
      "grad_norm": 5.132575035095215,
      "learning_rate": 7.539043935694269e-07,
      "loss": 0.8278,
      "step": 18230
    },
    {
      "epoch": 0.28388613418231917,
      "grad_norm": 3.5914008617401123,
      "learning_rate": 7.538224545648219e-07,
      "loss": 0.7782,
      "step": 18235
    },
    {
      "epoch": 0.28396397518428856,
      "grad_norm": 3.1311373710632324,
      "learning_rate": 7.537405155602169e-07,
      "loss": 0.9563,
      "step": 18240
    },
    {
      "epoch": 0.28404181618625796,
      "grad_norm": 4.662816524505615,
      "learning_rate": 7.53658576555612e-07,
      "loss": 0.8451,
      "step": 18245
    },
    {
      "epoch": 0.28411965718822735,
      "grad_norm": 4.352396011352539,
      "learning_rate": 7.535766375510069e-07,
      "loss": 0.7883,
      "step": 18250
    },
    {
      "epoch": 0.2841974981901967,
      "grad_norm": 7.063146114349365,
      "learning_rate": 7.53494698546402e-07,
      "loss": 0.795,
      "step": 18255
    },
    {
      "epoch": 0.2842753391921661,
      "grad_norm": 6.286665439605713,
      "learning_rate": 7.534127595417971e-07,
      "loss": 0.7575,
      "step": 18260
    },
    {
      "epoch": 0.28435318019413547,
      "grad_norm": 3.584050178527832,
      "learning_rate": 7.533308205371922e-07,
      "loss": 0.7749,
      "step": 18265
    },
    {
      "epoch": 0.28443102119610486,
      "grad_norm": 3.3104610443115234,
      "learning_rate": 7.532488815325871e-07,
      "loss": 0.8768,
      "step": 18270
    },
    {
      "epoch": 0.2845088621980742,
      "grad_norm": 3.0027613639831543,
      "learning_rate": 7.531669425279822e-07,
      "loss": 0.7214,
      "step": 18275
    },
    {
      "epoch": 0.2845867032000436,
      "grad_norm": 7.029557228088379,
      "learning_rate": 7.530850035233772e-07,
      "loss": 0.8627,
      "step": 18280
    },
    {
      "epoch": 0.284664544202013,
      "grad_norm": 3.829381227493286,
      "learning_rate": 7.530030645187721e-07,
      "loss": 0.7763,
      "step": 18285
    },
    {
      "epoch": 0.2847423852039824,
      "grad_norm": 3.7503976821899414,
      "learning_rate": 7.529211255141672e-07,
      "loss": 0.9111,
      "step": 18290
    },
    {
      "epoch": 0.2848202262059517,
      "grad_norm": 5.3771586418151855,
      "learning_rate": 7.528391865095623e-07,
      "loss": 0.8367,
      "step": 18295
    },
    {
      "epoch": 0.2848980672079211,
      "grad_norm": 2.6693806648254395,
      "learning_rate": 7.527572475049573e-07,
      "loss": 0.7148,
      "step": 18300
    },
    {
      "epoch": 0.2849759082098905,
      "grad_norm": 8.563738822937012,
      "learning_rate": 7.526753085003523e-07,
      "loss": 0.8822,
      "step": 18305
    },
    {
      "epoch": 0.28505374921185983,
      "grad_norm": 3.3797926902770996,
      "learning_rate": 7.525933694957474e-07,
      "loss": 0.7949,
      "step": 18310
    },
    {
      "epoch": 0.2851315902138292,
      "grad_norm": 2.9566423892974854,
      "learning_rate": 7.525114304911424e-07,
      "loss": 0.7643,
      "step": 18315
    },
    {
      "epoch": 0.2852094312157986,
      "grad_norm": 2.9523532390594482,
      "learning_rate": 7.524294914865373e-07,
      "loss": 0.6975,
      "step": 18320
    },
    {
      "epoch": 0.285287272217768,
      "grad_norm": 7.08513069152832,
      "learning_rate": 7.523475524819324e-07,
      "loss": 0.772,
      "step": 18325
    },
    {
      "epoch": 0.28536511321973734,
      "grad_norm": 3.7659263610839844,
      "learning_rate": 7.522656134773274e-07,
      "loss": 0.8242,
      "step": 18330
    },
    {
      "epoch": 0.28544295422170674,
      "grad_norm": 3.876816511154175,
      "learning_rate": 7.521836744727225e-07,
      "loss": 0.8292,
      "step": 18335
    },
    {
      "epoch": 0.2855207952236761,
      "grad_norm": 3.302157163619995,
      "learning_rate": 7.521017354681175e-07,
      "loss": 0.7672,
      "step": 18340
    },
    {
      "epoch": 0.2855986362256455,
      "grad_norm": 3.461796760559082,
      "learning_rate": 7.520197964635125e-07,
      "loss": 0.7899,
      "step": 18345
    },
    {
      "epoch": 0.28567647722761486,
      "grad_norm": 4.767899990081787,
      "learning_rate": 7.519378574589076e-07,
      "loss": 0.7807,
      "step": 18350
    },
    {
      "epoch": 0.28575431822958425,
      "grad_norm": 2.773707389831543,
      "learning_rate": 7.518559184543027e-07,
      "loss": 0.6448,
      "step": 18355
    },
    {
      "epoch": 0.28583215923155364,
      "grad_norm": 3.4016494750976562,
      "learning_rate": 7.517739794496975e-07,
      "loss": 0.697,
      "step": 18360
    },
    {
      "epoch": 0.28591000023352303,
      "grad_norm": 3.069875955581665,
      "learning_rate": 7.516920404450926e-07,
      "loss": 0.7969,
      "step": 18365
    },
    {
      "epoch": 0.28598784123549237,
      "grad_norm": 4.186999797821045,
      "learning_rate": 7.516101014404877e-07,
      "loss": 0.7641,
      "step": 18370
    },
    {
      "epoch": 0.28606568223746176,
      "grad_norm": 5.006409168243408,
      "learning_rate": 7.515281624358826e-07,
      "loss": 0.6768,
      "step": 18375
    },
    {
      "epoch": 0.28614352323943115,
      "grad_norm": 8.393912315368652,
      "learning_rate": 7.514462234312777e-07,
      "loss": 0.843,
      "step": 18380
    },
    {
      "epoch": 0.2862213642414005,
      "grad_norm": 4.208598613739014,
      "learning_rate": 7.513642844266728e-07,
      "loss": 0.87,
      "step": 18385
    },
    {
      "epoch": 0.2862992052433699,
      "grad_norm": 2.9314475059509277,
      "learning_rate": 7.512823454220679e-07,
      "loss": 0.8289,
      "step": 18390
    },
    {
      "epoch": 0.2863770462453393,
      "grad_norm": 5.165114879608154,
      "learning_rate": 7.512004064174628e-07,
      "loss": 0.7525,
      "step": 18395
    },
    {
      "epoch": 0.28645488724730866,
      "grad_norm": 5.7505669593811035,
      "learning_rate": 7.511184674128578e-07,
      "loss": 0.6644,
      "step": 18400
    },
    {
      "epoch": 0.286532728249278,
      "grad_norm": 3.4606173038482666,
      "learning_rate": 7.510365284082529e-07,
      "loss": 0.7591,
      "step": 18405
    },
    {
      "epoch": 0.2866105692512474,
      "grad_norm": 4.732656478881836,
      "learning_rate": 7.509545894036478e-07,
      "loss": 0.9361,
      "step": 18410
    },
    {
      "epoch": 0.2866884102532168,
      "grad_norm": 5.098369121551514,
      "learning_rate": 7.508726503990429e-07,
      "loss": 0.832,
      "step": 18415
    },
    {
      "epoch": 0.2867662512551862,
      "grad_norm": 4.499561786651611,
      "learning_rate": 7.50790711394438e-07,
      "loss": 0.6976,
      "step": 18420
    },
    {
      "epoch": 0.2868440922571555,
      "grad_norm": 7.913333892822266,
      "learning_rate": 7.50708772389833e-07,
      "loss": 0.9198,
      "step": 18425
    },
    {
      "epoch": 0.2869219332591249,
      "grad_norm": 2.9314780235290527,
      "learning_rate": 7.50626833385228e-07,
      "loss": 0.7318,
      "step": 18430
    },
    {
      "epoch": 0.2869997742610943,
      "grad_norm": 3.969515323638916,
      "learning_rate": 7.505448943806231e-07,
      "loss": 0.9026,
      "step": 18435
    },
    {
      "epoch": 0.2870776152630637,
      "grad_norm": 5.8380351066589355,
      "learning_rate": 7.50462955376018e-07,
      "loss": 0.7391,
      "step": 18440
    },
    {
      "epoch": 0.287155456265033,
      "grad_norm": 3.473393201828003,
      "learning_rate": 7.50381016371413e-07,
      "loss": 0.863,
      "step": 18445
    },
    {
      "epoch": 0.2872332972670024,
      "grad_norm": 5.020837783813477,
      "learning_rate": 7.502990773668081e-07,
      "loss": 0.864,
      "step": 18450
    },
    {
      "epoch": 0.2873111382689718,
      "grad_norm": 3.1780622005462646,
      "learning_rate": 7.502171383622031e-07,
      "loss": 0.7748,
      "step": 18455
    },
    {
      "epoch": 0.2873889792709412,
      "grad_norm": 3.086195230484009,
      "learning_rate": 7.501351993575982e-07,
      "loss": 0.8153,
      "step": 18460
    },
    {
      "epoch": 0.28746682027291054,
      "grad_norm": 3.143611192703247,
      "learning_rate": 7.500532603529932e-07,
      "loss": 0.7826,
      "step": 18465
    },
    {
      "epoch": 0.28754466127487993,
      "grad_norm": 3.706808090209961,
      "learning_rate": 7.499713213483882e-07,
      "loss": 0.6314,
      "step": 18470
    },
    {
      "epoch": 0.2876225022768493,
      "grad_norm": 3.3525521755218506,
      "learning_rate": 7.498893823437833e-07,
      "loss": 0.7845,
      "step": 18475
    },
    {
      "epoch": 0.28770034327881866,
      "grad_norm": 3.0572566986083984,
      "learning_rate": 7.498074433391784e-07,
      "loss": 0.7662,
      "step": 18480
    },
    {
      "epoch": 0.28777818428078805,
      "grad_norm": 3.026519298553467,
      "learning_rate": 7.497255043345732e-07,
      "loss": 0.743,
      "step": 18485
    },
    {
      "epoch": 0.28785602528275744,
      "grad_norm": 6.692337512969971,
      "learning_rate": 7.496435653299683e-07,
      "loss": 0.8563,
      "step": 18490
    },
    {
      "epoch": 0.28793386628472684,
      "grad_norm": 4.200711727142334,
      "learning_rate": 7.495616263253634e-07,
      "loss": 0.7121,
      "step": 18495
    },
    {
      "epoch": 0.2880117072866962,
      "grad_norm": 3.3459670543670654,
      "learning_rate": 7.494796873207583e-07,
      "loss": 0.8345,
      "step": 18500
    },
    {
      "epoch": 0.28808954828866556,
      "grad_norm": 3.4595234394073486,
      "learning_rate": 7.493977483161534e-07,
      "loss": 0.8437,
      "step": 18505
    },
    {
      "epoch": 0.28816738929063496,
      "grad_norm": 3.5343360900878906,
      "learning_rate": 7.493158093115485e-07,
      "loss": 0.8151,
      "step": 18510
    },
    {
      "epoch": 0.28824523029260435,
      "grad_norm": 4.573742866516113,
      "learning_rate": 7.492338703069436e-07,
      "loss": 0.7683,
      "step": 18515
    },
    {
      "epoch": 0.2883230712945737,
      "grad_norm": 6.558521747589111,
      "learning_rate": 7.491519313023385e-07,
      "loss": 0.8551,
      "step": 18520
    },
    {
      "epoch": 0.2884009122965431,
      "grad_norm": 8.159137725830078,
      "learning_rate": 7.490699922977335e-07,
      "loss": 0.8556,
      "step": 18525
    },
    {
      "epoch": 0.28847875329851247,
      "grad_norm": 4.157485008239746,
      "learning_rate": 7.489880532931286e-07,
      "loss": 0.7474,
      "step": 18530
    },
    {
      "epoch": 0.28855659430048186,
      "grad_norm": 2.989870071411133,
      "learning_rate": 7.489061142885235e-07,
      "loss": 0.7662,
      "step": 18535
    },
    {
      "epoch": 0.2886344353024512,
      "grad_norm": 3.084144115447998,
      "learning_rate": 7.488241752839186e-07,
      "loss": 0.7597,
      "step": 18540
    },
    {
      "epoch": 0.2887122763044206,
      "grad_norm": 3.9873554706573486,
      "learning_rate": 7.487422362793137e-07,
      "loss": 0.6703,
      "step": 18545
    },
    {
      "epoch": 0.28879011730639,
      "grad_norm": 2.9624643325805664,
      "learning_rate": 7.486602972747087e-07,
      "loss": 0.7915,
      "step": 18550
    },
    {
      "epoch": 0.2888679583083593,
      "grad_norm": 3.346914529800415,
      "learning_rate": 7.485783582701037e-07,
      "loss": 0.6576,
      "step": 18555
    },
    {
      "epoch": 0.2889457993103287,
      "grad_norm": 3.5888848304748535,
      "learning_rate": 7.484964192654988e-07,
      "loss": 0.7523,
      "step": 18560
    },
    {
      "epoch": 0.2890236403122981,
      "grad_norm": 4.135461807250977,
      "learning_rate": 7.484144802608937e-07,
      "loss": 0.7744,
      "step": 18565
    },
    {
      "epoch": 0.2891014813142675,
      "grad_norm": 3.826291799545288,
      "learning_rate": 7.483325412562888e-07,
      "loss": 0.8145,
      "step": 18570
    },
    {
      "epoch": 0.28917932231623683,
      "grad_norm": 4.202657699584961,
      "learning_rate": 7.482506022516838e-07,
      "loss": 0.7333,
      "step": 18575
    },
    {
      "epoch": 0.2892571633182062,
      "grad_norm": 3.219723701477051,
      "learning_rate": 7.481686632470788e-07,
      "loss": 0.7905,
      "step": 18580
    },
    {
      "epoch": 0.2893350043201756,
      "grad_norm": 3.597161293029785,
      "learning_rate": 7.480867242424739e-07,
      "loss": 0.7624,
      "step": 18585
    },
    {
      "epoch": 0.289412845322145,
      "grad_norm": 6.197475433349609,
      "learning_rate": 7.48004785237869e-07,
      "loss": 0.9616,
      "step": 18590
    },
    {
      "epoch": 0.28949068632411434,
      "grad_norm": 3.9106247425079346,
      "learning_rate": 7.479228462332639e-07,
      "loss": 0.7329,
      "step": 18595
    },
    {
      "epoch": 0.28956852732608374,
      "grad_norm": 5.9468255043029785,
      "learning_rate": 7.47840907228659e-07,
      "loss": 0.8381,
      "step": 18600
    },
    {
      "epoch": 0.2896463683280531,
      "grad_norm": 5.991184234619141,
      "learning_rate": 7.47758968224054e-07,
      "loss": 0.7476,
      "step": 18605
    },
    {
      "epoch": 0.2897242093300225,
      "grad_norm": 4.296688556671143,
      "learning_rate": 7.476770292194489e-07,
      "loss": 0.7739,
      "step": 18610
    },
    {
      "epoch": 0.28980205033199186,
      "grad_norm": 4.5617146492004395,
      "learning_rate": 7.47595090214844e-07,
      "loss": 0.7349,
      "step": 18615
    },
    {
      "epoch": 0.28987989133396125,
      "grad_norm": 6.767970085144043,
      "learning_rate": 7.475131512102391e-07,
      "loss": 0.7144,
      "step": 18620
    },
    {
      "epoch": 0.28995773233593064,
      "grad_norm": 3.624408721923828,
      "learning_rate": 7.47431212205634e-07,
      "loss": 0.7996,
      "step": 18625
    },
    {
      "epoch": 0.29003557333790003,
      "grad_norm": 2.7725491523742676,
      "learning_rate": 7.473492732010291e-07,
      "loss": 0.743,
      "step": 18630
    },
    {
      "epoch": 0.29011341433986937,
      "grad_norm": 7.392967224121094,
      "learning_rate": 7.472673341964242e-07,
      "loss": 0.896,
      "step": 18635
    },
    {
      "epoch": 0.29019125534183876,
      "grad_norm": 6.188554286956787,
      "learning_rate": 7.471853951918193e-07,
      "loss": 0.6967,
      "step": 18640
    },
    {
      "epoch": 0.29026909634380815,
      "grad_norm": 6.979849338531494,
      "learning_rate": 7.471034561872141e-07,
      "loss": 0.8246,
      "step": 18645
    },
    {
      "epoch": 0.2903469373457775,
      "grad_norm": 3.2482519149780273,
      "learning_rate": 7.470215171826092e-07,
      "loss": 0.8414,
      "step": 18650
    },
    {
      "epoch": 0.2904247783477469,
      "grad_norm": 2.916731357574463,
      "learning_rate": 7.469395781780043e-07,
      "loss": 0.7341,
      "step": 18655
    },
    {
      "epoch": 0.2905026193497163,
      "grad_norm": 2.8692774772644043,
      "learning_rate": 7.468576391733993e-07,
      "loss": 0.8345,
      "step": 18660
    },
    {
      "epoch": 0.29058046035168567,
      "grad_norm": 4.653710842132568,
      "learning_rate": 7.467757001687943e-07,
      "loss": 0.7714,
      "step": 18665
    },
    {
      "epoch": 0.290658301353655,
      "grad_norm": 3.6073105335235596,
      "learning_rate": 7.466937611641894e-07,
      "loss": 0.807,
      "step": 18670
    },
    {
      "epoch": 0.2907361423556244,
      "grad_norm": 2.567059278488159,
      "learning_rate": 7.466118221595844e-07,
      "loss": 0.6275,
      "step": 18675
    },
    {
      "epoch": 0.2908139833575938,
      "grad_norm": 4.383679389953613,
      "learning_rate": 7.465298831549794e-07,
      "loss": 0.8669,
      "step": 18680
    },
    {
      "epoch": 0.2908918243595632,
      "grad_norm": 3.321564197540283,
      "learning_rate": 7.464479441503744e-07,
      "loss": 0.6724,
      "step": 18685
    },
    {
      "epoch": 0.2909696653615325,
      "grad_norm": 6.479825973510742,
      "learning_rate": 7.463660051457694e-07,
      "loss": 0.7196,
      "step": 18690
    },
    {
      "epoch": 0.2910475063635019,
      "grad_norm": 7.732433795928955,
      "learning_rate": 7.462840661411645e-07,
      "loss": 0.8695,
      "step": 18695
    },
    {
      "epoch": 0.2911253473654713,
      "grad_norm": 2.9067394733428955,
      "learning_rate": 7.462021271365595e-07,
      "loss": 0.6753,
      "step": 18700
    },
    {
      "epoch": 0.2912031883674407,
      "grad_norm": 3.960364580154419,
      "learning_rate": 7.461201881319545e-07,
      "loss": 0.9328,
      "step": 18705
    },
    {
      "epoch": 0.29128102936941,
      "grad_norm": 2.495321750640869,
      "learning_rate": 7.460382491273496e-07,
      "loss": 0.8255,
      "step": 18710
    },
    {
      "epoch": 0.2913588703713794,
      "grad_norm": 2.928755044937134,
      "learning_rate": 7.459563101227447e-07,
      "loss": 0.7353,
      "step": 18715
    },
    {
      "epoch": 0.2914367113733488,
      "grad_norm": 2.7744498252868652,
      "learning_rate": 7.458743711181396e-07,
      "loss": 0.8438,
      "step": 18720
    },
    {
      "epoch": 0.29151455237531815,
      "grad_norm": 5.325984001159668,
      "learning_rate": 7.457924321135346e-07,
      "loss": 0.6956,
      "step": 18725
    },
    {
      "epoch": 0.29159239337728754,
      "grad_norm": 3.782843589782715,
      "learning_rate": 7.457104931089297e-07,
      "loss": 0.7555,
      "step": 18730
    },
    {
      "epoch": 0.29167023437925693,
      "grad_norm": 4.154267311096191,
      "learning_rate": 7.456285541043246e-07,
      "loss": 0.7718,
      "step": 18735
    },
    {
      "epoch": 0.2917480753812263,
      "grad_norm": 4.614317417144775,
      "learning_rate": 7.455466150997197e-07,
      "loss": 0.8251,
      "step": 18740
    },
    {
      "epoch": 0.29182591638319566,
      "grad_norm": 3.0895204544067383,
      "learning_rate": 7.454646760951148e-07,
      "loss": 0.7649,
      "step": 18745
    },
    {
      "epoch": 0.29190375738516505,
      "grad_norm": 4.956776142120361,
      "learning_rate": 7.453827370905098e-07,
      "loss": 0.7662,
      "step": 18750
    },
    {
      "epoch": 0.29198159838713444,
      "grad_norm": 4.089311599731445,
      "learning_rate": 7.453007980859048e-07,
      "loss": 0.6745,
      "step": 18755
    },
    {
      "epoch": 0.29205943938910384,
      "grad_norm": 4.280338764190674,
      "learning_rate": 7.452188590812999e-07,
      "loss": 0.8082,
      "step": 18760
    },
    {
      "epoch": 0.2921372803910732,
      "grad_norm": 4.679727554321289,
      "learning_rate": 7.451369200766948e-07,
      "loss": 0.7907,
      "step": 18765
    },
    {
      "epoch": 0.29221512139304257,
      "grad_norm": 7.030491828918457,
      "learning_rate": 7.450549810720898e-07,
      "loss": 0.822,
      "step": 18770
    },
    {
      "epoch": 0.29229296239501196,
      "grad_norm": 3.752502202987671,
      "learning_rate": 7.449730420674849e-07,
      "loss": 0.6647,
      "step": 18775
    },
    {
      "epoch": 0.29237080339698135,
      "grad_norm": 3.4395251274108887,
      "learning_rate": 7.4489110306288e-07,
      "loss": 0.9502,
      "step": 18780
    },
    {
      "epoch": 0.2924486443989507,
      "grad_norm": 4.704014778137207,
      "learning_rate": 7.44809164058275e-07,
      "loss": 0.8132,
      "step": 18785
    },
    {
      "epoch": 0.2925264854009201,
      "grad_norm": 6.6849446296691895,
      "learning_rate": 7.4472722505367e-07,
      "loss": 0.7998,
      "step": 18790
    },
    {
      "epoch": 0.29260432640288947,
      "grad_norm": 5.722506046295166,
      "learning_rate": 7.446452860490651e-07,
      "loss": 0.7584,
      "step": 18795
    },
    {
      "epoch": 0.29268216740485886,
      "grad_norm": 4.313940048217773,
      "learning_rate": 7.445633470444601e-07,
      "loss": 0.8576,
      "step": 18800
    },
    {
      "epoch": 0.2927600084068282,
      "grad_norm": 3.7770705223083496,
      "learning_rate": 7.444814080398552e-07,
      "loss": 0.8339,
      "step": 18805
    },
    {
      "epoch": 0.2928378494087976,
      "grad_norm": 7.073665618896484,
      "learning_rate": 7.443994690352501e-07,
      "loss": 0.8197,
      "step": 18810
    },
    {
      "epoch": 0.292915690410767,
      "grad_norm": 4.039837837219238,
      "learning_rate": 7.443175300306451e-07,
      "loss": 0.7743,
      "step": 18815
    },
    {
      "epoch": 0.2929935314127363,
      "grad_norm": 3.668461561203003,
      "learning_rate": 7.442355910260402e-07,
      "loss": 0.7049,
      "step": 18820
    },
    {
      "epoch": 0.2930713724147057,
      "grad_norm": 4.572951793670654,
      "learning_rate": 7.441536520214352e-07,
      "loss": 0.7926,
      "step": 18825
    },
    {
      "epoch": 0.2931492134166751,
      "grad_norm": 2.8897600173950195,
      "learning_rate": 7.440717130168302e-07,
      "loss": 0.885,
      "step": 18830
    },
    {
      "epoch": 0.2932270544186445,
      "grad_norm": 6.474730014801025,
      "learning_rate": 7.439897740122253e-07,
      "loss": 0.5874,
      "step": 18835
    },
    {
      "epoch": 0.29330489542061383,
      "grad_norm": 3.490199327468872,
      "learning_rate": 7.439078350076204e-07,
      "loss": 0.8314,
      "step": 18840
    },
    {
      "epoch": 0.2933827364225832,
      "grad_norm": 3.0774970054626465,
      "learning_rate": 7.438258960030153e-07,
      "loss": 0.7794,
      "step": 18845
    },
    {
      "epoch": 0.2934605774245526,
      "grad_norm": 4.177643299102783,
      "learning_rate": 7.437439569984103e-07,
      "loss": 0.8409,
      "step": 18850
    },
    {
      "epoch": 0.293538418426522,
      "grad_norm": 4.719580173492432,
      "learning_rate": 7.436620179938054e-07,
      "loss": 0.8033,
      "step": 18855
    },
    {
      "epoch": 0.29361625942849134,
      "grad_norm": 3.9040963649749756,
      "learning_rate": 7.435800789892003e-07,
      "loss": 0.7557,
      "step": 18860
    },
    {
      "epoch": 0.29369410043046074,
      "grad_norm": 4.212202548980713,
      "learning_rate": 7.434981399845954e-07,
      "loss": 0.789,
      "step": 18865
    },
    {
      "epoch": 0.29377194143243013,
      "grad_norm": 6.8639373779296875,
      "learning_rate": 7.434162009799905e-07,
      "loss": 0.885,
      "step": 18870
    },
    {
      "epoch": 0.2938497824343995,
      "grad_norm": 3.239004611968994,
      "learning_rate": 7.433342619753855e-07,
      "loss": 0.7409,
      "step": 18875
    },
    {
      "epoch": 0.29392762343636886,
      "grad_norm": 4.313357830047607,
      "learning_rate": 7.432523229707805e-07,
      "loss": 0.778,
      "step": 18880
    },
    {
      "epoch": 0.29400546443833825,
      "grad_norm": 4.123286724090576,
      "learning_rate": 7.431703839661756e-07,
      "loss": 0.7855,
      "step": 18885
    },
    {
      "epoch": 0.29408330544030764,
      "grad_norm": 3.8900656700134277,
      "learning_rate": 7.430884449615705e-07,
      "loss": 0.6672,
      "step": 18890
    },
    {
      "epoch": 0.294161146442277,
      "grad_norm": 5.243304252624512,
      "learning_rate": 7.430065059569656e-07,
      "loss": 0.8295,
      "step": 18895
    },
    {
      "epoch": 0.29423898744424637,
      "grad_norm": 7.33563756942749,
      "learning_rate": 7.429245669523606e-07,
      "loss": 0.6894,
      "step": 18900
    },
    {
      "epoch": 0.29431682844621576,
      "grad_norm": 3.1834230422973633,
      "learning_rate": 7.428426279477557e-07,
      "loss": 0.8793,
      "step": 18905
    },
    {
      "epoch": 0.29439466944818515,
      "grad_norm": 3.197418212890625,
      "learning_rate": 7.427606889431507e-07,
      "loss": 0.755,
      "step": 18910
    },
    {
      "epoch": 0.2944725104501545,
      "grad_norm": 3.6597862243652344,
      "learning_rate": 7.426787499385457e-07,
      "loss": 0.8454,
      "step": 18915
    },
    {
      "epoch": 0.2945503514521239,
      "grad_norm": 3.0385353565216064,
      "learning_rate": 7.425968109339408e-07,
      "loss": 0.7821,
      "step": 18920
    },
    {
      "epoch": 0.2946281924540933,
      "grad_norm": 5.031830310821533,
      "learning_rate": 7.425148719293358e-07,
      "loss": 0.7794,
      "step": 18925
    },
    {
      "epoch": 0.29470603345606267,
      "grad_norm": 4.892368793487549,
      "learning_rate": 7.424329329247308e-07,
      "loss": 0.6679,
      "step": 18930
    },
    {
      "epoch": 0.294783874458032,
      "grad_norm": 8.175716400146484,
      "learning_rate": 7.423509939201258e-07,
      "loss": 0.7837,
      "step": 18935
    },
    {
      "epoch": 0.2948617154600014,
      "grad_norm": 2.7154431343078613,
      "learning_rate": 7.422690549155208e-07,
      "loss": 0.7933,
      "step": 18940
    },
    {
      "epoch": 0.2949395564619708,
      "grad_norm": 3.6209683418273926,
      "learning_rate": 7.421871159109159e-07,
      "loss": 0.8131,
      "step": 18945
    },
    {
      "epoch": 0.2950173974639402,
      "grad_norm": 4.062808513641357,
      "learning_rate": 7.42105176906311e-07,
      "loss": 0.851,
      "step": 18950
    },
    {
      "epoch": 0.2950952384659095,
      "grad_norm": 4.193578720092773,
      "learning_rate": 7.420232379017059e-07,
      "loss": 0.6818,
      "step": 18955
    },
    {
      "epoch": 0.2951730794678789,
      "grad_norm": 7.759963512420654,
      "learning_rate": 7.41941298897101e-07,
      "loss": 0.7228,
      "step": 18960
    },
    {
      "epoch": 0.2952509204698483,
      "grad_norm": 4.498473644256592,
      "learning_rate": 7.418593598924961e-07,
      "loss": 0.7439,
      "step": 18965
    },
    {
      "epoch": 0.2953287614718177,
      "grad_norm": 5.182002067565918,
      "learning_rate": 7.417774208878909e-07,
      "loss": 0.8098,
      "step": 18970
    },
    {
      "epoch": 0.295406602473787,
      "grad_norm": 4.381763458251953,
      "learning_rate": 7.41695481883286e-07,
      "loss": 0.6748,
      "step": 18975
    },
    {
      "epoch": 0.2954844434757564,
      "grad_norm": 3.979992628097534,
      "learning_rate": 7.416135428786811e-07,
      "loss": 0.7706,
      "step": 18980
    },
    {
      "epoch": 0.2955622844777258,
      "grad_norm": 2.989089012145996,
      "learning_rate": 7.41531603874076e-07,
      "loss": 0.747,
      "step": 18985
    },
    {
      "epoch": 0.29564012547969515,
      "grad_norm": 5.946344375610352,
      "learning_rate": 7.414496648694711e-07,
      "loss": 0.6734,
      "step": 18990
    },
    {
      "epoch": 0.29571796648166454,
      "grad_norm": 2.8725976943969727,
      "learning_rate": 7.413677258648662e-07,
      "loss": 0.7419,
      "step": 18995
    },
    {
      "epoch": 0.29579580748363393,
      "grad_norm": 3.7277004718780518,
      "learning_rate": 7.412857868602612e-07,
      "loss": 0.7738,
      "step": 19000
    },
    {
      "epoch": 0.2958736484856033,
      "grad_norm": 6.9846367835998535,
      "learning_rate": 7.412038478556562e-07,
      "loss": 0.6773,
      "step": 19005
    },
    {
      "epoch": 0.29595148948757266,
      "grad_norm": 3.8853328227996826,
      "learning_rate": 7.411219088510512e-07,
      "loss": 0.6712,
      "step": 19010
    },
    {
      "epoch": 0.29602933048954205,
      "grad_norm": 6.792137145996094,
      "learning_rate": 7.410399698464462e-07,
      "loss": 0.836,
      "step": 19015
    },
    {
      "epoch": 0.29610717149151145,
      "grad_norm": 5.322309494018555,
      "learning_rate": 7.409580308418413e-07,
      "loss": 0.7975,
      "step": 19020
    },
    {
      "epoch": 0.29618501249348084,
      "grad_norm": 3.9240760803222656,
      "learning_rate": 7.408760918372363e-07,
      "loss": 0.8532,
      "step": 19025
    },
    {
      "epoch": 0.2962628534954502,
      "grad_norm": 2.847827196121216,
      "learning_rate": 7.407941528326314e-07,
      "loss": 0.7627,
      "step": 19030
    },
    {
      "epoch": 0.29634069449741957,
      "grad_norm": 4.343638896942139,
      "learning_rate": 7.407122138280264e-07,
      "loss": 0.9215,
      "step": 19035
    },
    {
      "epoch": 0.29641853549938896,
      "grad_norm": 2.5756824016571045,
      "learning_rate": 7.406302748234215e-07,
      "loss": 0.7291,
      "step": 19040
    },
    {
      "epoch": 0.29649637650135835,
      "grad_norm": 3.4114644527435303,
      "learning_rate": 7.405483358188165e-07,
      "loss": 0.8098,
      "step": 19045
    },
    {
      "epoch": 0.2965742175033277,
      "grad_norm": 3.304774522781372,
      "learning_rate": 7.404663968142114e-07,
      "loss": 0.7402,
      "step": 19050
    },
    {
      "epoch": 0.2966520585052971,
      "grad_norm": 4.325243949890137,
      "learning_rate": 7.403844578096065e-07,
      "loss": 1.008,
      "step": 19055
    },
    {
      "epoch": 0.29672989950726647,
      "grad_norm": 3.628368616104126,
      "learning_rate": 7.403025188050015e-07,
      "loss": 0.5888,
      "step": 19060
    },
    {
      "epoch": 0.29680774050923586,
      "grad_norm": 3.809199333190918,
      "learning_rate": 7.402205798003965e-07,
      "loss": 0.8118,
      "step": 19065
    },
    {
      "epoch": 0.2968855815112052,
      "grad_norm": 3.897184133529663,
      "learning_rate": 7.401386407957916e-07,
      "loss": 0.7218,
      "step": 19070
    },
    {
      "epoch": 0.2969634225131746,
      "grad_norm": 4.060847759246826,
      "learning_rate": 7.400567017911867e-07,
      "loss": 0.803,
      "step": 19075
    },
    {
      "epoch": 0.297041263515144,
      "grad_norm": 4.0685906410217285,
      "learning_rate": 7.399747627865816e-07,
      "loss": 0.7336,
      "step": 19080
    },
    {
      "epoch": 0.2971191045171133,
      "grad_norm": 3.2365095615386963,
      "learning_rate": 7.398928237819767e-07,
      "loss": 0.9283,
      "step": 19085
    },
    {
      "epoch": 0.2971969455190827,
      "grad_norm": 3.6066620349884033,
      "learning_rate": 7.398108847773718e-07,
      "loss": 0.7151,
      "step": 19090
    },
    {
      "epoch": 0.2972747865210521,
      "grad_norm": 3.55867600440979,
      "learning_rate": 7.397289457727666e-07,
      "loss": 0.7524,
      "step": 19095
    },
    {
      "epoch": 0.2973526275230215,
      "grad_norm": 2.7427852153778076,
      "learning_rate": 7.396470067681617e-07,
      "loss": 0.8824,
      "step": 19100
    },
    {
      "epoch": 0.29743046852499083,
      "grad_norm": 10.828716278076172,
      "learning_rate": 7.395650677635568e-07,
      "loss": 0.6942,
      "step": 19105
    },
    {
      "epoch": 0.2975083095269602,
      "grad_norm": 4.044482231140137,
      "learning_rate": 7.394831287589518e-07,
      "loss": 0.7965,
      "step": 19110
    },
    {
      "epoch": 0.2975861505289296,
      "grad_norm": 7.023869514465332,
      "learning_rate": 7.394011897543468e-07,
      "loss": 0.7369,
      "step": 19115
    },
    {
      "epoch": 0.297663991530899,
      "grad_norm": 6.251187801361084,
      "learning_rate": 7.393192507497419e-07,
      "loss": 0.7276,
      "step": 19120
    },
    {
      "epoch": 0.29774183253286834,
      "grad_norm": 4.2345194816589355,
      "learning_rate": 7.392373117451369e-07,
      "loss": 0.8351,
      "step": 19125
    },
    {
      "epoch": 0.29781967353483774,
      "grad_norm": 5.484807014465332,
      "learning_rate": 7.39155372740532e-07,
      "loss": 0.8511,
      "step": 19130
    },
    {
      "epoch": 0.29789751453680713,
      "grad_norm": 4.656536102294922,
      "learning_rate": 7.390734337359269e-07,
      "loss": 0.7701,
      "step": 19135
    },
    {
      "epoch": 0.2979753555387765,
      "grad_norm": 4.175867557525635,
      "learning_rate": 7.389914947313219e-07,
      "loss": 0.8358,
      "step": 19140
    },
    {
      "epoch": 0.29805319654074586,
      "grad_norm": 5.8004279136657715,
      "learning_rate": 7.38909555726717e-07,
      "loss": 0.8207,
      "step": 19145
    },
    {
      "epoch": 0.29813103754271525,
      "grad_norm": 3.534777879714966,
      "learning_rate": 7.38827616722112e-07,
      "loss": 0.7163,
      "step": 19150
    },
    {
      "epoch": 0.29820887854468464,
      "grad_norm": 2.747576951980591,
      "learning_rate": 7.387456777175071e-07,
      "loss": 0.7338,
      "step": 19155
    },
    {
      "epoch": 0.298286719546654,
      "grad_norm": 3.858243465423584,
      "learning_rate": 7.386637387129021e-07,
      "loss": 0.8268,
      "step": 19160
    },
    {
      "epoch": 0.29836456054862337,
      "grad_norm": 3.667602300643921,
      "learning_rate": 7.385817997082972e-07,
      "loss": 0.6898,
      "step": 19165
    },
    {
      "epoch": 0.29844240155059276,
      "grad_norm": 5.483829021453857,
      "learning_rate": 7.384998607036922e-07,
      "loss": 0.7615,
      "step": 19170
    },
    {
      "epoch": 0.29852024255256215,
      "grad_norm": 3.279327630996704,
      "learning_rate": 7.384179216990871e-07,
      "loss": 0.7374,
      "step": 19175
    },
    {
      "epoch": 0.2985980835545315,
      "grad_norm": 4.456769943237305,
      "learning_rate": 7.383359826944822e-07,
      "loss": 0.7688,
      "step": 19180
    },
    {
      "epoch": 0.2986759245565009,
      "grad_norm": 5.216207981109619,
      "learning_rate": 7.382540436898772e-07,
      "loss": 0.7678,
      "step": 19185
    },
    {
      "epoch": 0.2987537655584703,
      "grad_norm": 5.643771648406982,
      "learning_rate": 7.381721046852722e-07,
      "loss": 0.7999,
      "step": 19190
    },
    {
      "epoch": 0.29883160656043967,
      "grad_norm": 7.622976779937744,
      "learning_rate": 7.380901656806673e-07,
      "loss": 0.6539,
      "step": 19195
    },
    {
      "epoch": 0.298909447562409,
      "grad_norm": 3.570014476776123,
      "learning_rate": 7.380082266760624e-07,
      "loss": 0.8342,
      "step": 19200
    },
    {
      "epoch": 0.2989872885643784,
      "grad_norm": 3.948180913925171,
      "learning_rate": 7.379262876714573e-07,
      "loss": 0.656,
      "step": 19205
    },
    {
      "epoch": 0.2990651295663478,
      "grad_norm": 4.40153169631958,
      "learning_rate": 7.378443486668524e-07,
      "loss": 0.812,
      "step": 19210
    },
    {
      "epoch": 0.2991429705683172,
      "grad_norm": 4.24730110168457,
      "learning_rate": 7.377624096622474e-07,
      "loss": 0.6723,
      "step": 19215
    },
    {
      "epoch": 0.2992208115702865,
      "grad_norm": 2.9068751335144043,
      "learning_rate": 7.376804706576423e-07,
      "loss": 0.7985,
      "step": 19220
    },
    {
      "epoch": 0.2992986525722559,
      "grad_norm": 3.062253475189209,
      "learning_rate": 7.375985316530374e-07,
      "loss": 0.8082,
      "step": 19225
    },
    {
      "epoch": 0.2993764935742253,
      "grad_norm": 2.8117597103118896,
      "learning_rate": 7.375165926484325e-07,
      "loss": 0.7478,
      "step": 19230
    },
    {
      "epoch": 0.2994543345761947,
      "grad_norm": 3.2682876586914062,
      "learning_rate": 7.374346536438275e-07,
      "loss": 0.7893,
      "step": 19235
    },
    {
      "epoch": 0.29953217557816403,
      "grad_norm": 5.783231258392334,
      "learning_rate": 7.373527146392225e-07,
      "loss": 0.7702,
      "step": 19240
    },
    {
      "epoch": 0.2996100165801334,
      "grad_norm": 4.528698444366455,
      "learning_rate": 7.372707756346176e-07,
      "loss": 0.7292,
      "step": 19245
    },
    {
      "epoch": 0.2996878575821028,
      "grad_norm": 3.2343766689300537,
      "learning_rate": 7.371888366300126e-07,
      "loss": 0.7593,
      "step": 19250
    },
    {
      "epoch": 0.29976569858407215,
      "grad_norm": 3.411930799484253,
      "learning_rate": 7.371068976254076e-07,
      "loss": 0.8373,
      "step": 19255
    },
    {
      "epoch": 0.29984353958604154,
      "grad_norm": 4.680840969085693,
      "learning_rate": 7.370249586208026e-07,
      "loss": 0.8613,
      "step": 19260
    },
    {
      "epoch": 0.29992138058801093,
      "grad_norm": 2.9823901653289795,
      "learning_rate": 7.369430196161976e-07,
      "loss": 0.7563,
      "step": 19265
    },
    {
      "epoch": 0.2999992215899803,
      "grad_norm": 9.004485130310059,
      "learning_rate": 7.368610806115927e-07,
      "loss": 0.7912,
      "step": 19270
    },
    {
      "epoch": 0.30007706259194966,
      "grad_norm": 4.095830917358398,
      "learning_rate": 7.367791416069877e-07,
      "loss": 0.7193,
      "step": 19275
    },
    {
      "epoch": 0.30015490359391905,
      "grad_norm": 15.25143814086914,
      "learning_rate": 7.366972026023828e-07,
      "loss": 0.6691,
      "step": 19280
    },
    {
      "epoch": 0.30023274459588845,
      "grad_norm": 2.6462841033935547,
      "learning_rate": 7.366152635977778e-07,
      "loss": 0.8597,
      "step": 19285
    },
    {
      "epoch": 0.30031058559785784,
      "grad_norm": 3.802515983581543,
      "learning_rate": 7.365333245931729e-07,
      "loss": 0.9042,
      "step": 19290
    },
    {
      "epoch": 0.3003884265998272,
      "grad_norm": 2.9349098205566406,
      "learning_rate": 7.364513855885678e-07,
      "loss": 0.7542,
      "step": 19295
    },
    {
      "epoch": 0.30046626760179657,
      "grad_norm": 2.784654378890991,
      "learning_rate": 7.363694465839628e-07,
      "loss": 0.7769,
      "step": 19300
    },
    {
      "epoch": 0.30054410860376596,
      "grad_norm": 5.91815710067749,
      "learning_rate": 7.362875075793579e-07,
      "loss": 0.7659,
      "step": 19305
    },
    {
      "epoch": 0.30062194960573535,
      "grad_norm": 6.340529441833496,
      "learning_rate": 7.36205568574753e-07,
      "loss": 0.7498,
      "step": 19310
    },
    {
      "epoch": 0.3006997906077047,
      "grad_norm": 3.2137279510498047,
      "learning_rate": 7.361236295701479e-07,
      "loss": 0.7722,
      "step": 19315
    },
    {
      "epoch": 0.3007776316096741,
      "grad_norm": 6.5689473152160645,
      "learning_rate": 7.36041690565543e-07,
      "loss": 0.7377,
      "step": 19320
    },
    {
      "epoch": 0.30085547261164347,
      "grad_norm": 2.988903760910034,
      "learning_rate": 7.359597515609381e-07,
      "loss": 0.7134,
      "step": 19325
    },
    {
      "epoch": 0.3009333136136128,
      "grad_norm": 4.176788330078125,
      "learning_rate": 7.35877812556333e-07,
      "loss": 0.8399,
      "step": 19330
    },
    {
      "epoch": 0.3010111546155822,
      "grad_norm": 3.5837290287017822,
      "learning_rate": 7.35795873551728e-07,
      "loss": 0.7243,
      "step": 19335
    },
    {
      "epoch": 0.3010889956175516,
      "grad_norm": 6.051422119140625,
      "learning_rate": 7.357139345471231e-07,
      "loss": 0.7335,
      "step": 19340
    },
    {
      "epoch": 0.301166836619521,
      "grad_norm": 3.454967498779297,
      "learning_rate": 7.35631995542518e-07,
      "loss": 0.7165,
      "step": 19345
    },
    {
      "epoch": 0.3012446776214903,
      "grad_norm": 3.151862621307373,
      "learning_rate": 7.355500565379131e-07,
      "loss": 0.8532,
      "step": 19350
    },
    {
      "epoch": 0.3013225186234597,
      "grad_norm": 6.889316558837891,
      "learning_rate": 7.354681175333082e-07,
      "loss": 0.6624,
      "step": 19355
    },
    {
      "epoch": 0.3014003596254291,
      "grad_norm": 6.973013877868652,
      "learning_rate": 7.353861785287032e-07,
      "loss": 0.7738,
      "step": 19360
    },
    {
      "epoch": 0.3014782006273985,
      "grad_norm": 3.1961417198181152,
      "learning_rate": 7.353042395240982e-07,
      "loss": 0.8734,
      "step": 19365
    },
    {
      "epoch": 0.30155604162936783,
      "grad_norm": 4.515864372253418,
      "learning_rate": 7.352223005194933e-07,
      "loss": 0.8756,
      "step": 19370
    },
    {
      "epoch": 0.3016338826313372,
      "grad_norm": 6.735131740570068,
      "learning_rate": 7.351403615148882e-07,
      "loss": 0.6706,
      "step": 19375
    },
    {
      "epoch": 0.3017117236333066,
      "grad_norm": 5.002194881439209,
      "learning_rate": 7.350584225102833e-07,
      "loss": 0.7529,
      "step": 19380
    },
    {
      "epoch": 0.301789564635276,
      "grad_norm": 6.737076759338379,
      "learning_rate": 7.349764835056783e-07,
      "loss": 0.6168,
      "step": 19385
    },
    {
      "epoch": 0.30186740563724535,
      "grad_norm": 4.550383567810059,
      "learning_rate": 7.348945445010733e-07,
      "loss": 0.7475,
      "step": 19390
    },
    {
      "epoch": 0.30194524663921474,
      "grad_norm": 6.845092296600342,
      "learning_rate": 7.348126054964684e-07,
      "loss": 0.7623,
      "step": 19395
    },
    {
      "epoch": 0.30202308764118413,
      "grad_norm": 2.750274896621704,
      "learning_rate": 7.347306664918635e-07,
      "loss": 0.8408,
      "step": 19400
    },
    {
      "epoch": 0.3021009286431535,
      "grad_norm": 3.3687939643859863,
      "learning_rate": 7.346487274872585e-07,
      "loss": 0.7362,
      "step": 19405
    },
    {
      "epoch": 0.30217876964512286,
      "grad_norm": 4.238241195678711,
      "learning_rate": 7.345667884826535e-07,
      "loss": 0.7712,
      "step": 19410
    },
    {
      "epoch": 0.30225661064709225,
      "grad_norm": 3.3636481761932373,
      "learning_rate": 7.344848494780486e-07,
      "loss": 0.8209,
      "step": 19415
    },
    {
      "epoch": 0.30233445164906164,
      "grad_norm": 3.0912439823150635,
      "learning_rate": 7.344029104734435e-07,
      "loss": 0.7543,
      "step": 19420
    },
    {
      "epoch": 0.302412292651031,
      "grad_norm": 4.076699733734131,
      "learning_rate": 7.343209714688385e-07,
      "loss": 0.7493,
      "step": 19425
    },
    {
      "epoch": 0.30249013365300037,
      "grad_norm": 3.4594085216522217,
      "learning_rate": 7.342390324642336e-07,
      "loss": 0.7467,
      "step": 19430
    },
    {
      "epoch": 0.30256797465496976,
      "grad_norm": 4.686057090759277,
      "learning_rate": 7.341570934596287e-07,
      "loss": 0.8548,
      "step": 19435
    },
    {
      "epoch": 0.30264581565693915,
      "grad_norm": 2.9473624229431152,
      "learning_rate": 7.340751544550236e-07,
      "loss": 0.7667,
      "step": 19440
    },
    {
      "epoch": 0.3027236566589085,
      "grad_norm": 2.9822275638580322,
      "learning_rate": 7.339932154504187e-07,
      "loss": 0.806,
      "step": 19445
    },
    {
      "epoch": 0.3028014976608779,
      "grad_norm": 3.60502552986145,
      "learning_rate": 7.339112764458138e-07,
      "loss": 0.6964,
      "step": 19450
    },
    {
      "epoch": 0.3028793386628473,
      "grad_norm": 4.586121082305908,
      "learning_rate": 7.338293374412087e-07,
      "loss": 0.7756,
      "step": 19455
    },
    {
      "epoch": 0.30295717966481667,
      "grad_norm": 4.052720069885254,
      "learning_rate": 7.337473984366037e-07,
      "loss": 0.7412,
      "step": 19460
    },
    {
      "epoch": 0.303035020666786,
      "grad_norm": 3.488071918487549,
      "learning_rate": 7.336654594319988e-07,
      "loss": 0.7772,
      "step": 19465
    },
    {
      "epoch": 0.3031128616687554,
      "grad_norm": 3.216257333755493,
      "learning_rate": 7.335835204273938e-07,
      "loss": 0.8047,
      "step": 19470
    },
    {
      "epoch": 0.3031907026707248,
      "grad_norm": 2.9577155113220215,
      "learning_rate": 7.335015814227888e-07,
      "loss": 0.77,
      "step": 19475
    },
    {
      "epoch": 0.3032685436726942,
      "grad_norm": 3.1843841075897217,
      "learning_rate": 7.334196424181839e-07,
      "loss": 0.8028,
      "step": 19480
    },
    {
      "epoch": 0.3033463846746635,
      "grad_norm": 5.264181137084961,
      "learning_rate": 7.333377034135789e-07,
      "loss": 0.7464,
      "step": 19485
    },
    {
      "epoch": 0.3034242256766329,
      "grad_norm": 6.033514976501465,
      "learning_rate": 7.33255764408974e-07,
      "loss": 0.9238,
      "step": 19490
    },
    {
      "epoch": 0.3035020666786023,
      "grad_norm": 3.4538733959198,
      "learning_rate": 7.33173825404369e-07,
      "loss": 0.8628,
      "step": 19495
    },
    {
      "epoch": 0.30357990768057164,
      "grad_norm": 6.510023593902588,
      "learning_rate": 7.330918863997639e-07,
      "loss": 0.6936,
      "step": 19500
    },
    {
      "epoch": 0.30365774868254103,
      "grad_norm": 6.218939304351807,
      "learning_rate": 7.33009947395159e-07,
      "loss": 0.7854,
      "step": 19505
    },
    {
      "epoch": 0.3037355896845104,
      "grad_norm": 3.971550464630127,
      "learning_rate": 7.32928008390554e-07,
      "loss": 0.9312,
      "step": 19510
    },
    {
      "epoch": 0.3038134306864798,
      "grad_norm": 3.35803484916687,
      "learning_rate": 7.32846069385949e-07,
      "loss": 0.7496,
      "step": 19515
    },
    {
      "epoch": 0.30389127168844915,
      "grad_norm": 4.602055549621582,
      "learning_rate": 7.327641303813441e-07,
      "loss": 0.7642,
      "step": 19520
    },
    {
      "epoch": 0.30396911269041854,
      "grad_norm": 3.0348055362701416,
      "learning_rate": 7.326821913767392e-07,
      "loss": 0.7273,
      "step": 19525
    },
    {
      "epoch": 0.30404695369238793,
      "grad_norm": 4.904761791229248,
      "learning_rate": 7.326002523721342e-07,
      "loss": 0.715,
      "step": 19530
    },
    {
      "epoch": 0.3041247946943573,
      "grad_norm": 7.011703968048096,
      "learning_rate": 7.325183133675292e-07,
      "loss": 0.7044,
      "step": 19535
    },
    {
      "epoch": 0.30420263569632666,
      "grad_norm": 7.0719895362854,
      "learning_rate": 7.324363743629242e-07,
      "loss": 0.7905,
      "step": 19540
    },
    {
      "epoch": 0.30428047669829605,
      "grad_norm": 5.640370845794678,
      "learning_rate": 7.323544353583192e-07,
      "loss": 0.8111,
      "step": 19545
    },
    {
      "epoch": 0.30435831770026545,
      "grad_norm": 3.0712015628814697,
      "learning_rate": 7.322724963537142e-07,
      "loss": 0.7576,
      "step": 19550
    },
    {
      "epoch": 0.30443615870223484,
      "grad_norm": 3.424154043197632,
      "learning_rate": 7.321905573491093e-07,
      "loss": 0.8453,
      "step": 19555
    },
    {
      "epoch": 0.3045139997042042,
      "grad_norm": 3.3583383560180664,
      "learning_rate": 7.321086183445044e-07,
      "loss": 0.7752,
      "step": 19560
    },
    {
      "epoch": 0.30459184070617357,
      "grad_norm": 4.696355819702148,
      "learning_rate": 7.320266793398993e-07,
      "loss": 0.7201,
      "step": 19565
    },
    {
      "epoch": 0.30466968170814296,
      "grad_norm": 7.2936859130859375,
      "learning_rate": 7.319447403352944e-07,
      "loss": 0.8347,
      "step": 19570
    },
    {
      "epoch": 0.30474752271011235,
      "grad_norm": 4.277551651000977,
      "learning_rate": 7.318628013306895e-07,
      "loss": 0.718,
      "step": 19575
    },
    {
      "epoch": 0.3048253637120817,
      "grad_norm": 5.286831855773926,
      "learning_rate": 7.317808623260843e-07,
      "loss": 0.747,
      "step": 19580
    },
    {
      "epoch": 0.3049032047140511,
      "grad_norm": 4.06176233291626,
      "learning_rate": 7.316989233214794e-07,
      "loss": 0.7597,
      "step": 19585
    },
    {
      "epoch": 0.30498104571602047,
      "grad_norm": 5.223543643951416,
      "learning_rate": 7.316169843168745e-07,
      "loss": 0.7336,
      "step": 19590
    },
    {
      "epoch": 0.3050588867179898,
      "grad_norm": 3.522918939590454,
      "learning_rate": 7.315350453122695e-07,
      "loss": 0.7236,
      "step": 19595
    },
    {
      "epoch": 0.3051367277199592,
      "grad_norm": 2.9740731716156006,
      "learning_rate": 7.314531063076645e-07,
      "loss": 0.7833,
      "step": 19600
    },
    {
      "epoch": 0.3052145687219286,
      "grad_norm": 3.410181760787964,
      "learning_rate": 7.313711673030596e-07,
      "loss": 0.8845,
      "step": 19605
    },
    {
      "epoch": 0.305292409723898,
      "grad_norm": 4.243725299835205,
      "learning_rate": 7.312892282984546e-07,
      "loss": 0.7387,
      "step": 19610
    },
    {
      "epoch": 0.3053702507258673,
      "grad_norm": 3.667628526687622,
      "learning_rate": 7.312072892938497e-07,
      "loss": 0.8079,
      "step": 19615
    },
    {
      "epoch": 0.3054480917278367,
      "grad_norm": 3.6076266765594482,
      "learning_rate": 7.311253502892446e-07,
      "loss": 0.7033,
      "step": 19620
    },
    {
      "epoch": 0.3055259327298061,
      "grad_norm": 5.2530999183654785,
      "learning_rate": 7.310434112846396e-07,
      "loss": 0.6888,
      "step": 19625
    },
    {
      "epoch": 0.3056037737317755,
      "grad_norm": 6.23611307144165,
      "learning_rate": 7.309614722800347e-07,
      "loss": 0.8028,
      "step": 19630
    },
    {
      "epoch": 0.30568161473374483,
      "grad_norm": 3.4382402896881104,
      "learning_rate": 7.308795332754297e-07,
      "loss": 0.7078,
      "step": 19635
    },
    {
      "epoch": 0.3057594557357142,
      "grad_norm": 5.48118782043457,
      "learning_rate": 7.307975942708247e-07,
      "loss": 0.7961,
      "step": 19640
    },
    {
      "epoch": 0.3058372967376836,
      "grad_norm": 6.199230670928955,
      "learning_rate": 7.307156552662198e-07,
      "loss": 0.7278,
      "step": 19645
    },
    {
      "epoch": 0.305915137739653,
      "grad_norm": 14.622966766357422,
      "learning_rate": 7.306337162616149e-07,
      "loss": 0.7,
      "step": 19650
    },
    {
      "epoch": 0.30599297874162235,
      "grad_norm": 6.439300060272217,
      "learning_rate": 7.305517772570099e-07,
      "loss": 0.8521,
      "step": 19655
    },
    {
      "epoch": 0.30607081974359174,
      "grad_norm": 4.3287577629089355,
      "learning_rate": 7.304698382524048e-07,
      "loss": 0.6632,
      "step": 19660
    },
    {
      "epoch": 0.30614866074556113,
      "grad_norm": 5.56475305557251,
      "learning_rate": 7.303878992477999e-07,
      "loss": 0.7035,
      "step": 19665
    },
    {
      "epoch": 0.3062265017475305,
      "grad_norm": 2.9900405406951904,
      "learning_rate": 7.30305960243195e-07,
      "loss": 0.5952,
      "step": 19670
    },
    {
      "epoch": 0.30630434274949986,
      "grad_norm": 3.227675676345825,
      "learning_rate": 7.302240212385899e-07,
      "loss": 0.8055,
      "step": 19675
    },
    {
      "epoch": 0.30638218375146925,
      "grad_norm": 2.7153420448303223,
      "learning_rate": 7.30142082233985e-07,
      "loss": 0.7083,
      "step": 19680
    },
    {
      "epoch": 0.30646002475343864,
      "grad_norm": 2.3255703449249268,
      "learning_rate": 7.300601432293801e-07,
      "loss": 0.7237,
      "step": 19685
    },
    {
      "epoch": 0.306537865755408,
      "grad_norm": 3.549863576889038,
      "learning_rate": 7.29978204224775e-07,
      "loss": 0.69,
      "step": 19690
    },
    {
      "epoch": 0.30661570675737737,
      "grad_norm": 3.5451266765594482,
      "learning_rate": 7.298962652201701e-07,
      "loss": 0.8929,
      "step": 19695
    },
    {
      "epoch": 0.30669354775934676,
      "grad_norm": 6.672890663146973,
      "learning_rate": 7.298143262155651e-07,
      "loss": 0.7899,
      "step": 19700
    },
    {
      "epoch": 0.30677138876131615,
      "grad_norm": 5.6110429763793945,
      "learning_rate": 7.297323872109601e-07,
      "loss": 0.7105,
      "step": 19705
    },
    {
      "epoch": 0.3068492297632855,
      "grad_norm": 3.9284863471984863,
      "learning_rate": 7.296504482063551e-07,
      "loss": 0.6883,
      "step": 19710
    },
    {
      "epoch": 0.3069270707652549,
      "grad_norm": 3.8036837577819824,
      "learning_rate": 7.295685092017502e-07,
      "loss": 0.7643,
      "step": 19715
    },
    {
      "epoch": 0.3070049117672243,
      "grad_norm": 5.060457706451416,
      "learning_rate": 7.294865701971452e-07,
      "loss": 0.7906,
      "step": 19720
    },
    {
      "epoch": 0.30708275276919367,
      "grad_norm": 3.4861080646514893,
      "learning_rate": 7.294046311925402e-07,
      "loss": 0.7877,
      "step": 19725
    },
    {
      "epoch": 0.307160593771163,
      "grad_norm": 5.678596496582031,
      "learning_rate": 7.293226921879353e-07,
      "loss": 0.7542,
      "step": 19730
    },
    {
      "epoch": 0.3072384347731324,
      "grad_norm": 3.183572769165039,
      "learning_rate": 7.292407531833303e-07,
      "loss": 0.8266,
      "step": 19735
    },
    {
      "epoch": 0.3073162757751018,
      "grad_norm": 3.0133581161499023,
      "learning_rate": 7.291588141787254e-07,
      "loss": 0.8507,
      "step": 19740
    },
    {
      "epoch": 0.3073941167770712,
      "grad_norm": 5.432462215423584,
      "learning_rate": 7.290768751741203e-07,
      "loss": 0.8463,
      "step": 19745
    },
    {
      "epoch": 0.3074719577790405,
      "grad_norm": 4.709221363067627,
      "learning_rate": 7.289949361695153e-07,
      "loss": 0.7677,
      "step": 19750
    },
    {
      "epoch": 0.3075497987810099,
      "grad_norm": 6.288868427276611,
      "learning_rate": 7.289129971649104e-07,
      "loss": 0.8172,
      "step": 19755
    },
    {
      "epoch": 0.3076276397829793,
      "grad_norm": 3.6717214584350586,
      "learning_rate": 7.288310581603055e-07,
      "loss": 0.7405,
      "step": 19760
    },
    {
      "epoch": 0.30770548078494864,
      "grad_norm": 3.573765277862549,
      "learning_rate": 7.287491191557004e-07,
      "loss": 0.7495,
      "step": 19765
    },
    {
      "epoch": 0.30778332178691803,
      "grad_norm": 3.3548150062561035,
      "learning_rate": 7.286671801510955e-07,
      "loss": 0.7931,
      "step": 19770
    },
    {
      "epoch": 0.3078611627888874,
      "grad_norm": 4.807186126708984,
      "learning_rate": 7.285852411464906e-07,
      "loss": 0.6837,
      "step": 19775
    },
    {
      "epoch": 0.3079390037908568,
      "grad_norm": 4.181824684143066,
      "learning_rate": 7.285033021418856e-07,
      "loss": 0.8501,
      "step": 19780
    },
    {
      "epoch": 0.30801684479282615,
      "grad_norm": 3.6679091453552246,
      "learning_rate": 7.284213631372805e-07,
      "loss": 0.7887,
      "step": 19785
    },
    {
      "epoch": 0.30809468579479554,
      "grad_norm": 3.1990134716033936,
      "learning_rate": 7.283394241326756e-07,
      "loss": 0.7741,
      "step": 19790
    },
    {
      "epoch": 0.30817252679676493,
      "grad_norm": 2.814786195755005,
      "learning_rate": 7.282574851280707e-07,
      "loss": 0.7523,
      "step": 19795
    },
    {
      "epoch": 0.3082503677987343,
      "grad_norm": 6.019982814788818,
      "learning_rate": 7.281755461234656e-07,
      "loss": 0.8319,
      "step": 19800
    },
    {
      "epoch": 0.30832820880070366,
      "grad_norm": 9.461308479309082,
      "learning_rate": 7.280936071188607e-07,
      "loss": 0.7141,
      "step": 19805
    },
    {
      "epoch": 0.30840604980267305,
      "grad_norm": 5.778023719787598,
      "learning_rate": 7.280116681142558e-07,
      "loss": 0.6953,
      "step": 19810
    },
    {
      "epoch": 0.30848389080464245,
      "grad_norm": 4.680640697479248,
      "learning_rate": 7.279297291096507e-07,
      "loss": 0.9028,
      "step": 19815
    },
    {
      "epoch": 0.30856173180661184,
      "grad_norm": 5.0716400146484375,
      "learning_rate": 7.278477901050458e-07,
      "loss": 0.6978,
      "step": 19820
    },
    {
      "epoch": 0.3086395728085812,
      "grad_norm": 5.249613285064697,
      "learning_rate": 7.277658511004408e-07,
      "loss": 0.8705,
      "step": 19825
    },
    {
      "epoch": 0.30871741381055057,
      "grad_norm": 3.1706202030181885,
      "learning_rate": 7.276839120958358e-07,
      "loss": 0.7233,
      "step": 19830
    },
    {
      "epoch": 0.30879525481251996,
      "grad_norm": 2.3790881633758545,
      "learning_rate": 7.276019730912308e-07,
      "loss": 0.7208,
      "step": 19835
    },
    {
      "epoch": 0.30887309581448935,
      "grad_norm": 5.737145900726318,
      "learning_rate": 7.275200340866259e-07,
      "loss": 0.6923,
      "step": 19840
    },
    {
      "epoch": 0.3089509368164587,
      "grad_norm": 3.3187997341156006,
      "learning_rate": 7.274380950820209e-07,
      "loss": 0.7003,
      "step": 19845
    },
    {
      "epoch": 0.3090287778184281,
      "grad_norm": 3.5388314723968506,
      "learning_rate": 7.27356156077416e-07,
      "loss": 0.7766,
      "step": 19850
    },
    {
      "epoch": 0.30910661882039747,
      "grad_norm": 6.672298908233643,
      "learning_rate": 7.27274217072811e-07,
      "loss": 0.8126,
      "step": 19855
    },
    {
      "epoch": 0.3091844598223668,
      "grad_norm": 4.5849080085754395,
      "learning_rate": 7.27192278068206e-07,
      "loss": 0.8502,
      "step": 19860
    },
    {
      "epoch": 0.3092623008243362,
      "grad_norm": 8.410574913024902,
      "learning_rate": 7.27110339063601e-07,
      "loss": 0.7512,
      "step": 19865
    },
    {
      "epoch": 0.3093401418263056,
      "grad_norm": 8.312150001525879,
      "learning_rate": 7.27028400058996e-07,
      "loss": 0.6793,
      "step": 19870
    },
    {
      "epoch": 0.309417982828275,
      "grad_norm": 5.622318267822266,
      "learning_rate": 7.26946461054391e-07,
      "loss": 0.7669,
      "step": 19875
    },
    {
      "epoch": 0.3094958238302443,
      "grad_norm": 3.0122766494750977,
      "learning_rate": 7.268645220497861e-07,
      "loss": 0.7512,
      "step": 19880
    },
    {
      "epoch": 0.3095736648322137,
      "grad_norm": 4.408302307128906,
      "learning_rate": 7.267825830451812e-07,
      "loss": 0.7914,
      "step": 19885
    },
    {
      "epoch": 0.3096515058341831,
      "grad_norm": 12.037958145141602,
      "learning_rate": 7.267006440405762e-07,
      "loss": 0.8234,
      "step": 19890
    },
    {
      "epoch": 0.3097293468361525,
      "grad_norm": 3.986621618270874,
      "learning_rate": 7.266187050359712e-07,
      "loss": 0.8296,
      "step": 19895
    },
    {
      "epoch": 0.30980718783812183,
      "grad_norm": 6.8099212646484375,
      "learning_rate": 7.265367660313663e-07,
      "loss": 0.6782,
      "step": 19900
    },
    {
      "epoch": 0.3098850288400912,
      "grad_norm": 5.6810479164123535,
      "learning_rate": 7.264548270267612e-07,
      "loss": 0.6785,
      "step": 19905
    },
    {
      "epoch": 0.3099628698420606,
      "grad_norm": 3.6507179737091064,
      "learning_rate": 7.263728880221562e-07,
      "loss": 0.8,
      "step": 19910
    },
    {
      "epoch": 0.31004071084403,
      "grad_norm": 3.3082025051116943,
      "learning_rate": 7.262909490175513e-07,
      "loss": 0.742,
      "step": 19915
    },
    {
      "epoch": 0.31011855184599935,
      "grad_norm": 4.687744140625,
      "learning_rate": 7.262090100129464e-07,
      "loss": 0.722,
      "step": 19920
    },
    {
      "epoch": 0.31019639284796874,
      "grad_norm": 4.227119445800781,
      "learning_rate": 7.261270710083413e-07,
      "loss": 0.8008,
      "step": 19925
    },
    {
      "epoch": 0.31027423384993813,
      "grad_norm": 5.139937877655029,
      "learning_rate": 7.260451320037364e-07,
      "loss": 0.8128,
      "step": 19930
    },
    {
      "epoch": 0.31035207485190747,
      "grad_norm": 4.09891414642334,
      "learning_rate": 7.259631929991315e-07,
      "loss": 0.7716,
      "step": 19935
    },
    {
      "epoch": 0.31042991585387686,
      "grad_norm": 3.9586713314056396,
      "learning_rate": 7.258812539945265e-07,
      "loss": 0.7492,
      "step": 19940
    },
    {
      "epoch": 0.31050775685584625,
      "grad_norm": 3.3138248920440674,
      "learning_rate": 7.257993149899214e-07,
      "loss": 0.8005,
      "step": 19945
    },
    {
      "epoch": 0.31058559785781564,
      "grad_norm": 3.6728568077087402,
      "learning_rate": 7.257173759853165e-07,
      "loss": 0.7757,
      "step": 19950
    },
    {
      "epoch": 0.310663438859785,
      "grad_norm": 3.5933239459991455,
      "learning_rate": 7.256354369807115e-07,
      "loss": 0.8814,
      "step": 19955
    },
    {
      "epoch": 0.31074127986175437,
      "grad_norm": 3.2425715923309326,
      "learning_rate": 7.255534979761065e-07,
      "loss": 0.7549,
      "step": 19960
    },
    {
      "epoch": 0.31081912086372376,
      "grad_norm": 7.56531286239624,
      "learning_rate": 7.254715589715016e-07,
      "loss": 0.7529,
      "step": 19965
    },
    {
      "epoch": 0.31089696186569316,
      "grad_norm": 3.339479923248291,
      "learning_rate": 7.253896199668966e-07,
      "loss": 0.8662,
      "step": 19970
    },
    {
      "epoch": 0.3109748028676625,
      "grad_norm": 3.970248222351074,
      "learning_rate": 7.253076809622917e-07,
      "loss": 0.8382,
      "step": 19975
    },
    {
      "epoch": 0.3110526438696319,
      "grad_norm": 6.586206912994385,
      "learning_rate": 7.252257419576867e-07,
      "loss": 0.656,
      "step": 19980
    },
    {
      "epoch": 0.3111304848716013,
      "grad_norm": 4.172905445098877,
      "learning_rate": 7.251438029530816e-07,
      "loss": 0.8625,
      "step": 19985
    },
    {
      "epoch": 0.31120832587357067,
      "grad_norm": 3.187283754348755,
      "learning_rate": 7.250618639484767e-07,
      "loss": 0.7319,
      "step": 19990
    },
    {
      "epoch": 0.31128616687554,
      "grad_norm": 3.5772223472595215,
      "learning_rate": 7.249799249438717e-07,
      "loss": 0.8214,
      "step": 19995
    },
    {
      "epoch": 0.3113640078775094,
      "grad_norm": 3.162219762802124,
      "learning_rate": 7.248979859392667e-07,
      "loss": 0.7177,
      "step": 20000
    },
    {
      "epoch": 0.3114418488794788,
      "grad_norm": 4.707995891571045,
      "learning_rate": 7.248160469346618e-07,
      "loss": 0.7666,
      "step": 20005
    },
    {
      "epoch": 0.3115196898814482,
      "grad_norm": 4.667968273162842,
      "learning_rate": 7.247341079300569e-07,
      "loss": 0.877,
      "step": 20010
    },
    {
      "epoch": 0.3115975308834175,
      "grad_norm": 8.888596534729004,
      "learning_rate": 7.246521689254519e-07,
      "loss": 0.7889,
      "step": 20015
    },
    {
      "epoch": 0.3116753718853869,
      "grad_norm": 4.097687721252441,
      "learning_rate": 7.245702299208469e-07,
      "loss": 0.6785,
      "step": 20020
    },
    {
      "epoch": 0.3117532128873563,
      "grad_norm": 5.855955600738525,
      "learning_rate": 7.244882909162419e-07,
      "loss": 0.7627,
      "step": 20025
    },
    {
      "epoch": 0.31183105388932564,
      "grad_norm": 2.3836493492126465,
      "learning_rate": 7.24406351911637e-07,
      "loss": 0.6925,
      "step": 20030
    },
    {
      "epoch": 0.31190889489129503,
      "grad_norm": 4.272909641265869,
      "learning_rate": 7.243244129070319e-07,
      "loss": 0.7573,
      "step": 20035
    },
    {
      "epoch": 0.3119867358932644,
      "grad_norm": 8.327914237976074,
      "learning_rate": 7.24242473902427e-07,
      "loss": 0.8195,
      "step": 20040
    },
    {
      "epoch": 0.3120645768952338,
      "grad_norm": 5.052828788757324,
      "learning_rate": 7.241605348978221e-07,
      "loss": 0.9444,
      "step": 20045
    },
    {
      "epoch": 0.31214241789720315,
      "grad_norm": 6.556353569030762,
      "learning_rate": 7.24078595893217e-07,
      "loss": 0.6907,
      "step": 20050
    },
    {
      "epoch": 0.31222025889917254,
      "grad_norm": 4.166321754455566,
      "learning_rate": 7.239966568886121e-07,
      "loss": 0.7356,
      "step": 20055
    },
    {
      "epoch": 0.31229809990114193,
      "grad_norm": 3.270192861557007,
      "learning_rate": 7.239147178840072e-07,
      "loss": 0.7642,
      "step": 20060
    },
    {
      "epoch": 0.3123759409031113,
      "grad_norm": 21.332584381103516,
      "learning_rate": 7.238327788794022e-07,
      "loss": 0.7392,
      "step": 20065
    },
    {
      "epoch": 0.31245378190508066,
      "grad_norm": 3.567343235015869,
      "learning_rate": 7.237508398747971e-07,
      "loss": 0.7774,
      "step": 20070
    },
    {
      "epoch": 0.31253162290705006,
      "grad_norm": 4.1884307861328125,
      "learning_rate": 7.236689008701922e-07,
      "loss": 0.8854,
      "step": 20075
    },
    {
      "epoch": 0.31260946390901945,
      "grad_norm": 3.5007705688476562,
      "learning_rate": 7.235869618655872e-07,
      "loss": 0.6666,
      "step": 20080
    },
    {
      "epoch": 0.31268730491098884,
      "grad_norm": 4.026138782501221,
      "learning_rate": 7.235050228609822e-07,
      "loss": 0.8585,
      "step": 20085
    },
    {
      "epoch": 0.3127651459129582,
      "grad_norm": 3.0517191886901855,
      "learning_rate": 7.234230838563773e-07,
      "loss": 0.8152,
      "step": 20090
    },
    {
      "epoch": 0.31284298691492757,
      "grad_norm": 8.352131843566895,
      "learning_rate": 7.233411448517723e-07,
      "loss": 0.7127,
      "step": 20095
    },
    {
      "epoch": 0.31292082791689696,
      "grad_norm": 2.481409788131714,
      "learning_rate": 7.232592058471674e-07,
      "loss": 0.713,
      "step": 20100
    },
    {
      "epoch": 0.3129986689188663,
      "grad_norm": 9.404962539672852,
      "learning_rate": 7.231772668425624e-07,
      "loss": 0.8564,
      "step": 20105
    },
    {
      "epoch": 0.3130765099208357,
      "grad_norm": 3.086458444595337,
      "learning_rate": 7.230953278379573e-07,
      "loss": 0.704,
      "step": 20110
    },
    {
      "epoch": 0.3131543509228051,
      "grad_norm": 4.2751240730285645,
      "learning_rate": 7.230133888333524e-07,
      "loss": 0.7979,
      "step": 20115
    },
    {
      "epoch": 0.3132321919247745,
      "grad_norm": 3.2433555126190186,
      "learning_rate": 7.229314498287475e-07,
      "loss": 0.7739,
      "step": 20120
    },
    {
      "epoch": 0.3133100329267438,
      "grad_norm": 6.711822032928467,
      "learning_rate": 7.228495108241424e-07,
      "loss": 0.8249,
      "step": 20125
    },
    {
      "epoch": 0.3133878739287132,
      "grad_norm": 5.886772632598877,
      "learning_rate": 7.227675718195375e-07,
      "loss": 0.8063,
      "step": 20130
    },
    {
      "epoch": 0.3134657149306826,
      "grad_norm": 5.3925018310546875,
      "learning_rate": 7.226856328149326e-07,
      "loss": 0.8441,
      "step": 20135
    },
    {
      "epoch": 0.313543555932652,
      "grad_norm": 3.970362663269043,
      "learning_rate": 7.226036938103276e-07,
      "loss": 0.7704,
      "step": 20140
    },
    {
      "epoch": 0.3136213969346213,
      "grad_norm": 7.354496479034424,
      "learning_rate": 7.225217548057226e-07,
      "loss": 0.7853,
      "step": 20145
    },
    {
      "epoch": 0.3136992379365907,
      "grad_norm": 4.8241190910339355,
      "learning_rate": 7.224398158011176e-07,
      "loss": 0.8676,
      "step": 20150
    },
    {
      "epoch": 0.3137770789385601,
      "grad_norm": 5.4294867515563965,
      "learning_rate": 7.223578767965127e-07,
      "loss": 0.7949,
      "step": 20155
    },
    {
      "epoch": 0.3138549199405295,
      "grad_norm": 6.615969657897949,
      "learning_rate": 7.222759377919076e-07,
      "loss": 0.7947,
      "step": 20160
    },
    {
      "epoch": 0.31393276094249883,
      "grad_norm": 4.777083396911621,
      "learning_rate": 7.221939987873027e-07,
      "loss": 0.8299,
      "step": 20165
    },
    {
      "epoch": 0.3140106019444682,
      "grad_norm": 4.006468296051025,
      "learning_rate": 7.221120597826978e-07,
      "loss": 0.6731,
      "step": 20170
    },
    {
      "epoch": 0.3140884429464376,
      "grad_norm": 2.7182235717773438,
      "learning_rate": 7.220301207780927e-07,
      "loss": 0.682,
      "step": 20175
    },
    {
      "epoch": 0.314166283948407,
      "grad_norm": 4.555752277374268,
      "learning_rate": 7.219481817734878e-07,
      "loss": 0.7247,
      "step": 20180
    },
    {
      "epoch": 0.31424412495037635,
      "grad_norm": 3.7917845249176025,
      "learning_rate": 7.218662427688829e-07,
      "loss": 0.745,
      "step": 20185
    },
    {
      "epoch": 0.31432196595234574,
      "grad_norm": 4.585085391998291,
      "learning_rate": 7.217843037642778e-07,
      "loss": 0.7629,
      "step": 20190
    },
    {
      "epoch": 0.31439980695431513,
      "grad_norm": 4.027462005615234,
      "learning_rate": 7.217023647596728e-07,
      "loss": 0.7306,
      "step": 20195
    },
    {
      "epoch": 0.31447764795628447,
      "grad_norm": 3.540985107421875,
      "learning_rate": 7.216204257550679e-07,
      "loss": 0.6836,
      "step": 20200
    },
    {
      "epoch": 0.31455548895825386,
      "grad_norm": 4.648544788360596,
      "learning_rate": 7.215384867504629e-07,
      "loss": 0.7504,
      "step": 20205
    },
    {
      "epoch": 0.31463332996022325,
      "grad_norm": 3.58345103263855,
      "learning_rate": 7.21456547745858e-07,
      "loss": 0.8156,
      "step": 20210
    },
    {
      "epoch": 0.31471117096219264,
      "grad_norm": 6.841537952423096,
      "learning_rate": 7.21374608741253e-07,
      "loss": 0.7757,
      "step": 20215
    },
    {
      "epoch": 0.314789011964162,
      "grad_norm": 3.252333879470825,
      "learning_rate": 7.21292669736648e-07,
      "loss": 0.7893,
      "step": 20220
    },
    {
      "epoch": 0.31486685296613137,
      "grad_norm": 3.391663074493408,
      "learning_rate": 7.212107307320431e-07,
      "loss": 0.8041,
      "step": 20225
    },
    {
      "epoch": 0.31494469396810076,
      "grad_norm": 3.951467990875244,
      "learning_rate": 7.21128791727438e-07,
      "loss": 0.8238,
      "step": 20230
    },
    {
      "epoch": 0.31502253497007016,
      "grad_norm": 2.974583864212036,
      "learning_rate": 7.21046852722833e-07,
      "loss": 0.7669,
      "step": 20235
    },
    {
      "epoch": 0.3151003759720395,
      "grad_norm": 7.49880313873291,
      "learning_rate": 7.209649137182281e-07,
      "loss": 0.768,
      "step": 20240
    },
    {
      "epoch": 0.3151782169740089,
      "grad_norm": 3.172010660171509,
      "learning_rate": 7.208829747136232e-07,
      "loss": 0.7582,
      "step": 20245
    },
    {
      "epoch": 0.3152560579759783,
      "grad_norm": 4.700904846191406,
      "learning_rate": 7.208010357090181e-07,
      "loss": 0.6729,
      "step": 20250
    },
    {
      "epoch": 0.31533389897794767,
      "grad_norm": 6.512081146240234,
      "learning_rate": 7.207190967044132e-07,
      "loss": 0.7653,
      "step": 20255
    },
    {
      "epoch": 0.315411739979917,
      "grad_norm": 3.5270817279815674,
      "learning_rate": 7.206371576998083e-07,
      "loss": 0.6887,
      "step": 20260
    },
    {
      "epoch": 0.3154895809818864,
      "grad_norm": 7.409488201141357,
      "learning_rate": 7.205552186952034e-07,
      "loss": 0.8739,
      "step": 20265
    },
    {
      "epoch": 0.3155674219838558,
      "grad_norm": 4.6055121421813965,
      "learning_rate": 7.204732796905982e-07,
      "loss": 0.722,
      "step": 20270
    },
    {
      "epoch": 0.3156452629858252,
      "grad_norm": 3.539865016937256,
      "learning_rate": 7.203913406859933e-07,
      "loss": 0.7651,
      "step": 20275
    },
    {
      "epoch": 0.3157231039877945,
      "grad_norm": 12.878244400024414,
      "learning_rate": 7.203094016813884e-07,
      "loss": 0.8434,
      "step": 20280
    },
    {
      "epoch": 0.3158009449897639,
      "grad_norm": 3.4090824127197266,
      "learning_rate": 7.202274626767833e-07,
      "loss": 0.7503,
      "step": 20285
    },
    {
      "epoch": 0.3158787859917333,
      "grad_norm": 3.430004119873047,
      "learning_rate": 7.201455236721784e-07,
      "loss": 0.7558,
      "step": 20290
    },
    {
      "epoch": 0.31595662699370264,
      "grad_norm": 4.504459857940674,
      "learning_rate": 7.200635846675735e-07,
      "loss": 0.7472,
      "step": 20295
    },
    {
      "epoch": 0.31603446799567203,
      "grad_norm": 7.216919422149658,
      "learning_rate": 7.199816456629685e-07,
      "loss": 0.6889,
      "step": 20300
    },
    {
      "epoch": 0.3161123089976414,
      "grad_norm": 3.8278470039367676,
      "learning_rate": 7.198997066583635e-07,
      "loss": 0.7783,
      "step": 20305
    },
    {
      "epoch": 0.3161901499996108,
      "grad_norm": 4.990563869476318,
      "learning_rate": 7.198177676537585e-07,
      "loss": 0.7194,
      "step": 20310
    },
    {
      "epoch": 0.31626799100158015,
      "grad_norm": 7.858270168304443,
      "learning_rate": 7.197358286491535e-07,
      "loss": 0.7615,
      "step": 20315
    },
    {
      "epoch": 0.31634583200354954,
      "grad_norm": 3.449202299118042,
      "learning_rate": 7.196538896445485e-07,
      "loss": 0.7311,
      "step": 20320
    },
    {
      "epoch": 0.31642367300551894,
      "grad_norm": 3.601962089538574,
      "learning_rate": 7.195719506399436e-07,
      "loss": 0.8149,
      "step": 20325
    },
    {
      "epoch": 0.3165015140074883,
      "grad_norm": 4.08453893661499,
      "learning_rate": 7.194900116353386e-07,
      "loss": 0.7266,
      "step": 20330
    },
    {
      "epoch": 0.31657935500945766,
      "grad_norm": 3.3417105674743652,
      "learning_rate": 7.194080726307337e-07,
      "loss": 0.7956,
      "step": 20335
    },
    {
      "epoch": 0.31665719601142706,
      "grad_norm": 3.159672975540161,
      "learning_rate": 7.193261336261287e-07,
      "loss": 0.7221,
      "step": 20340
    },
    {
      "epoch": 0.31673503701339645,
      "grad_norm": 5.266695022583008,
      "learning_rate": 7.192441946215237e-07,
      "loss": 0.8936,
      "step": 20345
    },
    {
      "epoch": 0.31681287801536584,
      "grad_norm": 2.914886474609375,
      "learning_rate": 7.191622556169187e-07,
      "loss": 0.9082,
      "step": 20350
    },
    {
      "epoch": 0.3168907190173352,
      "grad_norm": 3.562744379043579,
      "learning_rate": 7.190803166123138e-07,
      "loss": 0.7991,
      "step": 20355
    },
    {
      "epoch": 0.31696856001930457,
      "grad_norm": 8.86327838897705,
      "learning_rate": 7.189983776077087e-07,
      "loss": 0.7872,
      "step": 20360
    },
    {
      "epoch": 0.31704640102127396,
      "grad_norm": 3.006682872772217,
      "learning_rate": 7.189164386031038e-07,
      "loss": 0.7675,
      "step": 20365
    },
    {
      "epoch": 0.3171242420232433,
      "grad_norm": 6.521688461303711,
      "learning_rate": 7.188344995984989e-07,
      "loss": 0.7401,
      "step": 20370
    },
    {
      "epoch": 0.3172020830252127,
      "grad_norm": 4.019016742706299,
      "learning_rate": 7.187525605938938e-07,
      "loss": 0.7771,
      "step": 20375
    },
    {
      "epoch": 0.3172799240271821,
      "grad_norm": 3.9266228675842285,
      "learning_rate": 7.186706215892889e-07,
      "loss": 0.6257,
      "step": 20380
    },
    {
      "epoch": 0.3173577650291515,
      "grad_norm": 4.3516621589660645,
      "learning_rate": 7.18588682584684e-07,
      "loss": 0.7592,
      "step": 20385
    },
    {
      "epoch": 0.3174356060311208,
      "grad_norm": 4.869146823883057,
      "learning_rate": 7.185067435800791e-07,
      "loss": 0.7773,
      "step": 20390
    },
    {
      "epoch": 0.3175134470330902,
      "grad_norm": 4.757040500640869,
      "learning_rate": 7.184248045754739e-07,
      "loss": 0.8382,
      "step": 20395
    },
    {
      "epoch": 0.3175912880350596,
      "grad_norm": 4.194018363952637,
      "learning_rate": 7.18342865570869e-07,
      "loss": 0.6359,
      "step": 20400
    },
    {
      "epoch": 0.317669129037029,
      "grad_norm": 3.9095518589019775,
      "learning_rate": 7.182609265662641e-07,
      "loss": 0.755,
      "step": 20405
    },
    {
      "epoch": 0.3177469700389983,
      "grad_norm": 3.327188730239868,
      "learning_rate": 7.18178987561659e-07,
      "loss": 0.7284,
      "step": 20410
    },
    {
      "epoch": 0.3178248110409677,
      "grad_norm": 3.2544198036193848,
      "learning_rate": 7.180970485570541e-07,
      "loss": 0.7143,
      "step": 20415
    },
    {
      "epoch": 0.3179026520429371,
      "grad_norm": 5.302325248718262,
      "learning_rate": 7.180151095524492e-07,
      "loss": 0.7323,
      "step": 20420
    },
    {
      "epoch": 0.3179804930449065,
      "grad_norm": 3.3627989292144775,
      "learning_rate": 7.179331705478442e-07,
      "loss": 0.7139,
      "step": 20425
    },
    {
      "epoch": 0.31805833404687583,
      "grad_norm": 2.9275786876678467,
      "learning_rate": 7.178512315432392e-07,
      "loss": 0.8896,
      "step": 20430
    },
    {
      "epoch": 0.3181361750488452,
      "grad_norm": 2.586864709854126,
      "learning_rate": 7.177692925386342e-07,
      "loss": 0.8718,
      "step": 20435
    },
    {
      "epoch": 0.3182140160508146,
      "grad_norm": 3.3972079753875732,
      "learning_rate": 7.176873535340292e-07,
      "loss": 0.7542,
      "step": 20440
    },
    {
      "epoch": 0.318291857052784,
      "grad_norm": 3.256925344467163,
      "learning_rate": 7.176054145294243e-07,
      "loss": 0.815,
      "step": 20445
    },
    {
      "epoch": 0.31836969805475335,
      "grad_norm": 3.8902106285095215,
      "learning_rate": 7.175234755248193e-07,
      "loss": 0.8117,
      "step": 20450
    },
    {
      "epoch": 0.31844753905672274,
      "grad_norm": 4.046694755554199,
      "learning_rate": 7.174415365202143e-07,
      "loss": 0.7104,
      "step": 20455
    },
    {
      "epoch": 0.31852538005869213,
      "grad_norm": 7.7123918533325195,
      "learning_rate": 7.173595975156094e-07,
      "loss": 0.6584,
      "step": 20460
    },
    {
      "epoch": 0.31860322106066147,
      "grad_norm": 4.0524516105651855,
      "learning_rate": 7.172776585110044e-07,
      "loss": 0.7106,
      "step": 20465
    },
    {
      "epoch": 0.31868106206263086,
      "grad_norm": 2.6010935306549072,
      "learning_rate": 7.171957195063994e-07,
      "loss": 0.8609,
      "step": 20470
    },
    {
      "epoch": 0.31875890306460025,
      "grad_norm": 3.557629346847534,
      "learning_rate": 7.171137805017944e-07,
      "loss": 0.7728,
      "step": 20475
    },
    {
      "epoch": 0.31883674406656964,
      "grad_norm": 6.395180702209473,
      "learning_rate": 7.170318414971895e-07,
      "loss": 0.8742,
      "step": 20480
    },
    {
      "epoch": 0.318914585068539,
      "grad_norm": 2.464134454727173,
      "learning_rate": 7.169499024925844e-07,
      "loss": 0.6831,
      "step": 20485
    },
    {
      "epoch": 0.3189924260705084,
      "grad_norm": 3.003485679626465,
      "learning_rate": 7.168679634879795e-07,
      "loss": 0.7702,
      "step": 20490
    },
    {
      "epoch": 0.31907026707247776,
      "grad_norm": 3.070361614227295,
      "learning_rate": 7.167860244833746e-07,
      "loss": 0.7564,
      "step": 20495
    },
    {
      "epoch": 0.31914810807444716,
      "grad_norm": 2.900399684906006,
      "learning_rate": 7.167040854787695e-07,
      "loss": 0.8135,
      "step": 20500
    },
    {
      "epoch": 0.3192259490764165,
      "grad_norm": 2.8308839797973633,
      "learning_rate": 7.166221464741646e-07,
      "loss": 0.6554,
      "step": 20505
    },
    {
      "epoch": 0.3193037900783859,
      "grad_norm": 3.5729901790618896,
      "learning_rate": 7.165402074695597e-07,
      "loss": 0.7582,
      "step": 20510
    },
    {
      "epoch": 0.3193816310803553,
      "grad_norm": 8.747566223144531,
      "learning_rate": 7.164582684649546e-07,
      "loss": 0.8207,
      "step": 20515
    },
    {
      "epoch": 0.31945947208232467,
      "grad_norm": 3.617783784866333,
      "learning_rate": 7.163763294603496e-07,
      "loss": 0.808,
      "step": 20520
    },
    {
      "epoch": 0.319537313084294,
      "grad_norm": 4.175931453704834,
      "learning_rate": 7.162943904557447e-07,
      "loss": 0.7721,
      "step": 20525
    },
    {
      "epoch": 0.3196151540862634,
      "grad_norm": 3.4712469577789307,
      "learning_rate": 7.162124514511398e-07,
      "loss": 0.7462,
      "step": 20530
    },
    {
      "epoch": 0.3196929950882328,
      "grad_norm": 2.555041790008545,
      "learning_rate": 7.161305124465348e-07,
      "loss": 0.7408,
      "step": 20535
    },
    {
      "epoch": 0.3197708360902021,
      "grad_norm": 3.7030394077301025,
      "learning_rate": 7.160485734419298e-07,
      "loss": 0.8261,
      "step": 20540
    },
    {
      "epoch": 0.3198486770921715,
      "grad_norm": 4.538588047027588,
      "learning_rate": 7.159666344373249e-07,
      "loss": 0.8366,
      "step": 20545
    },
    {
      "epoch": 0.3199265180941409,
      "grad_norm": 4.802341938018799,
      "learning_rate": 7.158846954327199e-07,
      "loss": 0.92,
      "step": 20550
    },
    {
      "epoch": 0.3200043590961103,
      "grad_norm": 2.9012575149536133,
      "learning_rate": 7.158027564281148e-07,
      "loss": 0.7722,
      "step": 20555
    },
    {
      "epoch": 0.32008220009807964,
      "grad_norm": 2.780606269836426,
      "learning_rate": 7.157208174235099e-07,
      "loss": 0.6411,
      "step": 20560
    },
    {
      "epoch": 0.32016004110004903,
      "grad_norm": 3.9199488162994385,
      "learning_rate": 7.156388784189049e-07,
      "loss": 0.6957,
      "step": 20565
    },
    {
      "epoch": 0.3202378821020184,
      "grad_norm": 4.829502105712891,
      "learning_rate": 7.155569394143e-07,
      "loss": 0.778,
      "step": 20570
    },
    {
      "epoch": 0.3203157231039878,
      "grad_norm": 3.4407572746276855,
      "learning_rate": 7.15475000409695e-07,
      "loss": 0.6633,
      "step": 20575
    },
    {
      "epoch": 0.32039356410595715,
      "grad_norm": 3.0549888610839844,
      "learning_rate": 7.1539306140509e-07,
      "loss": 0.6254,
      "step": 20580
    },
    {
      "epoch": 0.32047140510792654,
      "grad_norm": 2.840653896331787,
      "learning_rate": 7.153111224004851e-07,
      "loss": 0.7218,
      "step": 20585
    },
    {
      "epoch": 0.32054924610989594,
      "grad_norm": 4.250532150268555,
      "learning_rate": 7.152291833958802e-07,
      "loss": 0.8882,
      "step": 20590
    },
    {
      "epoch": 0.3206270871118653,
      "grad_norm": 3.789776086807251,
      "learning_rate": 7.15147244391275e-07,
      "loss": 0.8756,
      "step": 20595
    },
    {
      "epoch": 0.32070492811383466,
      "grad_norm": 3.888643503189087,
      "learning_rate": 7.150653053866701e-07,
      "loss": 0.7486,
      "step": 20600
    },
    {
      "epoch": 0.32078276911580406,
      "grad_norm": 2.7952585220336914,
      "learning_rate": 7.149833663820652e-07,
      "loss": 0.7398,
      "step": 20605
    },
    {
      "epoch": 0.32086061011777345,
      "grad_norm": 3.1019365787506104,
      "learning_rate": 7.149014273774601e-07,
      "loss": 0.6767,
      "step": 20610
    },
    {
      "epoch": 0.32093845111974284,
      "grad_norm": 3.4878134727478027,
      "learning_rate": 7.148194883728552e-07,
      "loss": 0.8461,
      "step": 20615
    },
    {
      "epoch": 0.3210162921217122,
      "grad_norm": 3.560169219970703,
      "learning_rate": 7.147375493682503e-07,
      "loss": 0.7525,
      "step": 20620
    },
    {
      "epoch": 0.32109413312368157,
      "grad_norm": 6.664193153381348,
      "learning_rate": 7.146556103636453e-07,
      "loss": 0.8274,
      "step": 20625
    },
    {
      "epoch": 0.32117197412565096,
      "grad_norm": 7.751073360443115,
      "learning_rate": 7.145736713590403e-07,
      "loss": 0.7782,
      "step": 20630
    },
    {
      "epoch": 0.3212498151276203,
      "grad_norm": 3.1961686611175537,
      "learning_rate": 7.144917323544353e-07,
      "loss": 0.7371,
      "step": 20635
    },
    {
      "epoch": 0.3213276561295897,
      "grad_norm": 3.616342782974243,
      "learning_rate": 7.144097933498303e-07,
      "loss": 0.7793,
      "step": 20640
    },
    {
      "epoch": 0.3214054971315591,
      "grad_norm": 14.634657859802246,
      "learning_rate": 7.143278543452253e-07,
      "loss": 0.7341,
      "step": 20645
    },
    {
      "epoch": 0.3214833381335285,
      "grad_norm": 7.2038726806640625,
      "learning_rate": 7.142459153406204e-07,
      "loss": 0.7319,
      "step": 20650
    },
    {
      "epoch": 0.3215611791354978,
      "grad_norm": 3.5353848934173584,
      "learning_rate": 7.141639763360155e-07,
      "loss": 0.8196,
      "step": 20655
    },
    {
      "epoch": 0.3216390201374672,
      "grad_norm": 2.9616966247558594,
      "learning_rate": 7.140820373314105e-07,
      "loss": 0.7542,
      "step": 20660
    },
    {
      "epoch": 0.3217168611394366,
      "grad_norm": 6.182138442993164,
      "learning_rate": 7.140000983268055e-07,
      "loss": 0.7098,
      "step": 20665
    },
    {
      "epoch": 0.321794702141406,
      "grad_norm": 3.2851719856262207,
      "learning_rate": 7.139181593222006e-07,
      "loss": 0.73,
      "step": 20670
    },
    {
      "epoch": 0.3218725431433753,
      "grad_norm": 3.859102249145508,
      "learning_rate": 7.138362203175955e-07,
      "loss": 0.737,
      "step": 20675
    },
    {
      "epoch": 0.3219503841453447,
      "grad_norm": 3.811932325363159,
      "learning_rate": 7.137542813129905e-07,
      "loss": 0.7437,
      "step": 20680
    },
    {
      "epoch": 0.3220282251473141,
      "grad_norm": 3.6481974124908447,
      "learning_rate": 7.136723423083856e-07,
      "loss": 0.8089,
      "step": 20685
    },
    {
      "epoch": 0.3221060661492835,
      "grad_norm": 7.045378684997559,
      "learning_rate": 7.135904033037806e-07,
      "loss": 0.787,
      "step": 20690
    },
    {
      "epoch": 0.32218390715125284,
      "grad_norm": 5.550210952758789,
      "learning_rate": 7.135084642991757e-07,
      "loss": 0.8496,
      "step": 20695
    },
    {
      "epoch": 0.3222617481532222,
      "grad_norm": 6.449124813079834,
      "learning_rate": 7.134265252945707e-07,
      "loss": 0.6899,
      "step": 20700
    },
    {
      "epoch": 0.3223395891551916,
      "grad_norm": 3.796647310256958,
      "learning_rate": 7.133445862899657e-07,
      "loss": 0.7465,
      "step": 20705
    },
    {
      "epoch": 0.32241743015716096,
      "grad_norm": 3.268113136291504,
      "learning_rate": 7.132626472853608e-07,
      "loss": 0.7982,
      "step": 20710
    },
    {
      "epoch": 0.32249527115913035,
      "grad_norm": 15.425836563110352,
      "learning_rate": 7.131807082807559e-07,
      "loss": 0.7995,
      "step": 20715
    },
    {
      "epoch": 0.32257311216109974,
      "grad_norm": 5.5737152099609375,
      "learning_rate": 7.130987692761507e-07,
      "loss": 0.8068,
      "step": 20720
    },
    {
      "epoch": 0.32265095316306913,
      "grad_norm": 6.799961566925049,
      "learning_rate": 7.130168302715458e-07,
      "loss": 0.7693,
      "step": 20725
    },
    {
      "epoch": 0.32272879416503847,
      "grad_norm": 6.524772644042969,
      "learning_rate": 7.129348912669409e-07,
      "loss": 0.8043,
      "step": 20730
    },
    {
      "epoch": 0.32280663516700786,
      "grad_norm": 4.884058475494385,
      "learning_rate": 7.128529522623358e-07,
      "loss": 0.7259,
      "step": 20735
    },
    {
      "epoch": 0.32288447616897725,
      "grad_norm": 4.115115165710449,
      "learning_rate": 7.127710132577309e-07,
      "loss": 0.7945,
      "step": 20740
    },
    {
      "epoch": 0.32296231717094664,
      "grad_norm": 3.3195178508758545,
      "learning_rate": 7.12689074253126e-07,
      "loss": 0.8378,
      "step": 20745
    },
    {
      "epoch": 0.323040158172916,
      "grad_norm": 2.867581367492676,
      "learning_rate": 7.12607135248521e-07,
      "loss": 0.7582,
      "step": 20750
    },
    {
      "epoch": 0.3231179991748854,
      "grad_norm": 5.850396633148193,
      "learning_rate": 7.12525196243916e-07,
      "loss": 0.7765,
      "step": 20755
    },
    {
      "epoch": 0.32319584017685477,
      "grad_norm": 7.244020938873291,
      "learning_rate": 7.12443257239311e-07,
      "loss": 0.8619,
      "step": 20760
    },
    {
      "epoch": 0.32327368117882416,
      "grad_norm": 4.443892478942871,
      "learning_rate": 7.12361318234706e-07,
      "loss": 0.707,
      "step": 20765
    },
    {
      "epoch": 0.3233515221807935,
      "grad_norm": 4.754065990447998,
      "learning_rate": 7.12279379230101e-07,
      "loss": 0.7362,
      "step": 20770
    },
    {
      "epoch": 0.3234293631827629,
      "grad_norm": 4.68637228012085,
      "learning_rate": 7.121974402254961e-07,
      "loss": 0.8056,
      "step": 20775
    },
    {
      "epoch": 0.3235072041847323,
      "grad_norm": 3.501113176345825,
      "learning_rate": 7.121155012208912e-07,
      "loss": 0.8143,
      "step": 20780
    },
    {
      "epoch": 0.32358504518670167,
      "grad_norm": 4.2049031257629395,
      "learning_rate": 7.120335622162862e-07,
      "loss": 0.7402,
      "step": 20785
    },
    {
      "epoch": 0.323662886188671,
      "grad_norm": 5.971714973449707,
      "learning_rate": 7.119516232116812e-07,
      "loss": 0.7949,
      "step": 20790
    },
    {
      "epoch": 0.3237407271906404,
      "grad_norm": 5.241507053375244,
      "learning_rate": 7.118696842070763e-07,
      "loss": 0.8283,
      "step": 20795
    },
    {
      "epoch": 0.3238185681926098,
      "grad_norm": 3.72343111038208,
      "learning_rate": 7.117877452024712e-07,
      "loss": 0.8301,
      "step": 20800
    },
    {
      "epoch": 0.3238964091945791,
      "grad_norm": 6.797969818115234,
      "learning_rate": 7.117058061978663e-07,
      "loss": 0.7162,
      "step": 20805
    },
    {
      "epoch": 0.3239742501965485,
      "grad_norm": 7.002163887023926,
      "learning_rate": 7.116238671932613e-07,
      "loss": 0.741,
      "step": 20810
    },
    {
      "epoch": 0.3240520911985179,
      "grad_norm": 3.0822055339813232,
      "learning_rate": 7.115419281886563e-07,
      "loss": 0.7807,
      "step": 20815
    },
    {
      "epoch": 0.3241299322004873,
      "grad_norm": 4.6375508308410645,
      "learning_rate": 7.114599891840514e-07,
      "loss": 0.8142,
      "step": 20820
    },
    {
      "epoch": 0.32420777320245664,
      "grad_norm": 2.537044048309326,
      "learning_rate": 7.113780501794464e-07,
      "loss": 0.7271,
      "step": 20825
    },
    {
      "epoch": 0.32428561420442603,
      "grad_norm": 4.382402420043945,
      "learning_rate": 7.112961111748414e-07,
      "loss": 0.7769,
      "step": 20830
    },
    {
      "epoch": 0.3243634552063954,
      "grad_norm": 4.217374801635742,
      "learning_rate": 7.112141721702365e-07,
      "loss": 0.7414,
      "step": 20835
    },
    {
      "epoch": 0.3244412962083648,
      "grad_norm": 5.162209510803223,
      "learning_rate": 7.111322331656315e-07,
      "loss": 0.7683,
      "step": 20840
    },
    {
      "epoch": 0.32451913721033415,
      "grad_norm": 2.4148507118225098,
      "learning_rate": 7.110502941610264e-07,
      "loss": 0.7117,
      "step": 20845
    },
    {
      "epoch": 0.32459697821230354,
      "grad_norm": 4.48330545425415,
      "learning_rate": 7.109683551564215e-07,
      "loss": 0.8636,
      "step": 20850
    },
    {
      "epoch": 0.32467481921427294,
      "grad_norm": 3.14823842048645,
      "learning_rate": 7.108864161518166e-07,
      "loss": 0.7688,
      "step": 20855
    },
    {
      "epoch": 0.32475266021624233,
      "grad_norm": 3.530456781387329,
      "learning_rate": 7.108044771472115e-07,
      "loss": 0.8039,
      "step": 20860
    },
    {
      "epoch": 0.32483050121821166,
      "grad_norm": 3.5431201457977295,
      "learning_rate": 7.107225381426066e-07,
      "loss": 0.742,
      "step": 20865
    },
    {
      "epoch": 0.32490834222018106,
      "grad_norm": 9.657413482666016,
      "learning_rate": 7.106405991380017e-07,
      "loss": 0.7895,
      "step": 20870
    },
    {
      "epoch": 0.32498618322215045,
      "grad_norm": 6.042131423950195,
      "learning_rate": 7.105586601333967e-07,
      "loss": 0.6917,
      "step": 20875
    },
    {
      "epoch": 0.3250640242241198,
      "grad_norm": 5.177880764007568,
      "learning_rate": 7.104767211287916e-07,
      "loss": 0.7696,
      "step": 20880
    },
    {
      "epoch": 0.3251418652260892,
      "grad_norm": 2.8649775981903076,
      "learning_rate": 7.103947821241867e-07,
      "loss": 0.7532,
      "step": 20885
    },
    {
      "epoch": 0.32521970622805857,
      "grad_norm": 3.3477964401245117,
      "learning_rate": 7.103128431195817e-07,
      "loss": 0.7381,
      "step": 20890
    },
    {
      "epoch": 0.32529754723002796,
      "grad_norm": 3.959502935409546,
      "learning_rate": 7.102309041149768e-07,
      "loss": 0.8313,
      "step": 20895
    },
    {
      "epoch": 0.3253753882319973,
      "grad_norm": 3.5778965950012207,
      "learning_rate": 7.101489651103718e-07,
      "loss": 0.8392,
      "step": 20900
    },
    {
      "epoch": 0.3254532292339667,
      "grad_norm": 4.421009063720703,
      "learning_rate": 7.100670261057669e-07,
      "loss": 0.9063,
      "step": 20905
    },
    {
      "epoch": 0.3255310702359361,
      "grad_norm": 4.305064678192139,
      "learning_rate": 7.099850871011619e-07,
      "loss": 0.8522,
      "step": 20910
    },
    {
      "epoch": 0.3256089112379055,
      "grad_norm": 25.120641708374023,
      "learning_rate": 7.099031480965569e-07,
      "loss": 0.7622,
      "step": 20915
    },
    {
      "epoch": 0.3256867522398748,
      "grad_norm": 4.1203227043151855,
      "learning_rate": 7.098212090919519e-07,
      "loss": 0.7762,
      "step": 20920
    },
    {
      "epoch": 0.3257645932418442,
      "grad_norm": 3.8534672260284424,
      "learning_rate": 7.097392700873469e-07,
      "loss": 0.833,
      "step": 20925
    },
    {
      "epoch": 0.3258424342438136,
      "grad_norm": 4.1801652908325195,
      "learning_rate": 7.09657331082742e-07,
      "loss": 0.7922,
      "step": 20930
    },
    {
      "epoch": 0.325920275245783,
      "grad_norm": 4.395228862762451,
      "learning_rate": 7.09575392078137e-07,
      "loss": 0.7613,
      "step": 20935
    },
    {
      "epoch": 0.3259981162477523,
      "grad_norm": 4.56093168258667,
      "learning_rate": 7.09493453073532e-07,
      "loss": 0.675,
      "step": 20940
    },
    {
      "epoch": 0.3260759572497217,
      "grad_norm": 7.025084018707275,
      "learning_rate": 7.094115140689271e-07,
      "loss": 0.759,
      "step": 20945
    },
    {
      "epoch": 0.3261537982516911,
      "grad_norm": 4.575479984283447,
      "learning_rate": 7.093295750643222e-07,
      "loss": 0.8409,
      "step": 20950
    },
    {
      "epoch": 0.3262316392536605,
      "grad_norm": 5.8464837074279785,
      "learning_rate": 7.092476360597171e-07,
      "loss": 0.813,
      "step": 20955
    },
    {
      "epoch": 0.32630948025562984,
      "grad_norm": 2.9095611572265625,
      "learning_rate": 7.091656970551121e-07,
      "loss": 0.7483,
      "step": 20960
    },
    {
      "epoch": 0.3263873212575992,
      "grad_norm": 3.054791212081909,
      "learning_rate": 7.090837580505072e-07,
      "loss": 0.8178,
      "step": 20965
    },
    {
      "epoch": 0.3264651622595686,
      "grad_norm": 7.010560512542725,
      "learning_rate": 7.090018190459021e-07,
      "loss": 0.7066,
      "step": 20970
    },
    {
      "epoch": 0.32654300326153796,
      "grad_norm": 6.855307102203369,
      "learning_rate": 7.089198800412972e-07,
      "loss": 0.8175,
      "step": 20975
    },
    {
      "epoch": 0.32662084426350735,
      "grad_norm": 3.9125053882598877,
      "learning_rate": 7.088379410366923e-07,
      "loss": 0.7047,
      "step": 20980
    },
    {
      "epoch": 0.32669868526547674,
      "grad_norm": 2.5680830478668213,
      "learning_rate": 7.087560020320873e-07,
      "loss": 0.5779,
      "step": 20985
    },
    {
      "epoch": 0.32677652626744613,
      "grad_norm": 4.910250663757324,
      "learning_rate": 7.086740630274823e-07,
      "loss": 0.8485,
      "step": 20990
    },
    {
      "epoch": 0.32685436726941547,
      "grad_norm": 4.934157371520996,
      "learning_rate": 7.085921240228774e-07,
      "loss": 0.83,
      "step": 20995
    },
    {
      "epoch": 0.32693220827138486,
      "grad_norm": 2.8414039611816406,
      "learning_rate": 7.085101850182723e-07,
      "loss": 0.8265,
      "step": 21000
    },
    {
      "epoch": 0.32701004927335425,
      "grad_norm": 3.3391027450561523,
      "learning_rate": 7.084282460136673e-07,
      "loss": 0.7878,
      "step": 21005
    },
    {
      "epoch": 0.32708789027532365,
      "grad_norm": 3.4965968132019043,
      "learning_rate": 7.083463070090624e-07,
      "loss": 0.7286,
      "step": 21010
    },
    {
      "epoch": 0.327165731277293,
      "grad_norm": 4.536909580230713,
      "learning_rate": 7.082643680044574e-07,
      "loss": 0.8608,
      "step": 21015
    },
    {
      "epoch": 0.3272435722792624,
      "grad_norm": 7.158962249755859,
      "learning_rate": 7.081824289998525e-07,
      "loss": 0.8251,
      "step": 21020
    },
    {
      "epoch": 0.32732141328123177,
      "grad_norm": 2.9187631607055664,
      "learning_rate": 7.081004899952475e-07,
      "loss": 0.8394,
      "step": 21025
    },
    {
      "epoch": 0.32739925428320116,
      "grad_norm": 3.9261114597320557,
      "learning_rate": 7.080185509906426e-07,
      "loss": 0.8444,
      "step": 21030
    },
    {
      "epoch": 0.3274770952851705,
      "grad_norm": 2.9356627464294434,
      "learning_rate": 7.079366119860376e-07,
      "loss": 0.7229,
      "step": 21035
    },
    {
      "epoch": 0.3275549362871399,
      "grad_norm": 3.0354654788970947,
      "learning_rate": 7.078546729814327e-07,
      "loss": 0.7199,
      "step": 21040
    },
    {
      "epoch": 0.3276327772891093,
      "grad_norm": 4.891841888427734,
      "learning_rate": 7.077727339768276e-07,
      "loss": 0.7688,
      "step": 21045
    },
    {
      "epoch": 0.32771061829107867,
      "grad_norm": 3.367002487182617,
      "learning_rate": 7.076907949722226e-07,
      "loss": 0.725,
      "step": 21050
    },
    {
      "epoch": 0.327788459293048,
      "grad_norm": 4.214025020599365,
      "learning_rate": 7.076088559676177e-07,
      "loss": 0.8281,
      "step": 21055
    },
    {
      "epoch": 0.3278663002950174,
      "grad_norm": 3.413587808609009,
      "learning_rate": 7.075269169630127e-07,
      "loss": 0.8031,
      "step": 21060
    },
    {
      "epoch": 0.3279441412969868,
      "grad_norm": 3.970029592514038,
      "learning_rate": 7.074449779584077e-07,
      "loss": 0.7813,
      "step": 21065
    },
    {
      "epoch": 0.3280219822989561,
      "grad_norm": 4.808574199676514,
      "learning_rate": 7.073630389538028e-07,
      "loss": 0.8128,
      "step": 21070
    },
    {
      "epoch": 0.3280998233009255,
      "grad_norm": 3.092742919921875,
      "learning_rate": 7.072810999491979e-07,
      "loss": 0.8593,
      "step": 21075
    },
    {
      "epoch": 0.3281776643028949,
      "grad_norm": 4.983108997344971,
      "learning_rate": 7.071991609445928e-07,
      "loss": 0.6809,
      "step": 21080
    },
    {
      "epoch": 0.3282555053048643,
      "grad_norm": 3.6495978832244873,
      "learning_rate": 7.071172219399878e-07,
      "loss": 0.6302,
      "step": 21085
    },
    {
      "epoch": 0.32833334630683364,
      "grad_norm": 6.250349998474121,
      "learning_rate": 7.070352829353829e-07,
      "loss": 0.7432,
      "step": 21090
    },
    {
      "epoch": 0.32841118730880303,
      "grad_norm": 5.218113422393799,
      "learning_rate": 7.069533439307778e-07,
      "loss": 0.7807,
      "step": 21095
    },
    {
      "epoch": 0.3284890283107724,
      "grad_norm": 4.160154819488525,
      "learning_rate": 7.068714049261729e-07,
      "loss": 0.7347,
      "step": 21100
    },
    {
      "epoch": 0.3285668693127418,
      "grad_norm": 3.9513416290283203,
      "learning_rate": 7.06789465921568e-07,
      "loss": 0.7936,
      "step": 21105
    },
    {
      "epoch": 0.32864471031471115,
      "grad_norm": 5.0178375244140625,
      "learning_rate": 7.06707526916963e-07,
      "loss": 0.7616,
      "step": 21110
    },
    {
      "epoch": 0.32872255131668054,
      "grad_norm": 3.6662726402282715,
      "learning_rate": 7.06625587912358e-07,
      "loss": 0.6692,
      "step": 21115
    },
    {
      "epoch": 0.32880039231864994,
      "grad_norm": 3.0202407836914062,
      "learning_rate": 7.065436489077531e-07,
      "loss": 0.7599,
      "step": 21120
    },
    {
      "epoch": 0.32887823332061933,
      "grad_norm": 3.204550266265869,
      "learning_rate": 7.06461709903148e-07,
      "loss": 0.7346,
      "step": 21125
    },
    {
      "epoch": 0.32895607432258867,
      "grad_norm": 7.8837690353393555,
      "learning_rate": 7.06379770898543e-07,
      "loss": 0.7442,
      "step": 21130
    },
    {
      "epoch": 0.32903391532455806,
      "grad_norm": 4.227644443511963,
      "learning_rate": 7.062978318939381e-07,
      "loss": 0.7455,
      "step": 21135
    },
    {
      "epoch": 0.32911175632652745,
      "grad_norm": 2.4810287952423096,
      "learning_rate": 7.062158928893331e-07,
      "loss": 0.7375,
      "step": 21140
    },
    {
      "epoch": 0.3291895973284968,
      "grad_norm": 5.757534503936768,
      "learning_rate": 7.061339538847282e-07,
      "loss": 0.7165,
      "step": 21145
    },
    {
      "epoch": 0.3292674383304662,
      "grad_norm": 6.480144500732422,
      "learning_rate": 7.060520148801232e-07,
      "loss": 0.8146,
      "step": 21150
    },
    {
      "epoch": 0.32934527933243557,
      "grad_norm": 4.645882606506348,
      "learning_rate": 7.059700758755183e-07,
      "loss": 0.8694,
      "step": 21155
    },
    {
      "epoch": 0.32942312033440496,
      "grad_norm": 7.792613983154297,
      "learning_rate": 7.058881368709133e-07,
      "loss": 0.8158,
      "step": 21160
    },
    {
      "epoch": 0.3295009613363743,
      "grad_norm": 3.325713872909546,
      "learning_rate": 7.058061978663083e-07,
      "loss": 0.7242,
      "step": 21165
    },
    {
      "epoch": 0.3295788023383437,
      "grad_norm": 5.18583345413208,
      "learning_rate": 7.057242588617033e-07,
      "loss": 0.8161,
      "step": 21170
    },
    {
      "epoch": 0.3296566433403131,
      "grad_norm": 5.270481586456299,
      "learning_rate": 7.056423198570983e-07,
      "loss": 0.7999,
      "step": 21175
    },
    {
      "epoch": 0.3297344843422825,
      "grad_norm": 6.009892463684082,
      "learning_rate": 7.055603808524934e-07,
      "loss": 0.7995,
      "step": 21180
    },
    {
      "epoch": 0.3298123253442518,
      "grad_norm": 6.691226482391357,
      "learning_rate": 7.054784418478884e-07,
      "loss": 0.8364,
      "step": 21185
    },
    {
      "epoch": 0.3298901663462212,
      "grad_norm": 2.960845947265625,
      "learning_rate": 7.053965028432834e-07,
      "loss": 0.6677,
      "step": 21190
    },
    {
      "epoch": 0.3299680073481906,
      "grad_norm": 4.097884178161621,
      "learning_rate": 7.053145638386785e-07,
      "loss": 0.8006,
      "step": 21195
    },
    {
      "epoch": 0.33004584835016,
      "grad_norm": 4.129356384277344,
      "learning_rate": 7.052326248340736e-07,
      "loss": 0.7322,
      "step": 21200
    },
    {
      "epoch": 0.3301236893521293,
      "grad_norm": 11.891890525817871,
      "learning_rate": 7.051506858294684e-07,
      "loss": 0.8696,
      "step": 21205
    },
    {
      "epoch": 0.3302015303540987,
      "grad_norm": 4.318405628204346,
      "learning_rate": 7.050687468248635e-07,
      "loss": 0.8034,
      "step": 21210
    },
    {
      "epoch": 0.3302793713560681,
      "grad_norm": 3.004293441772461,
      "learning_rate": 7.049868078202586e-07,
      "loss": 0.9435,
      "step": 21215
    },
    {
      "epoch": 0.3303572123580375,
      "grad_norm": 3.503347873687744,
      "learning_rate": 7.049048688156535e-07,
      "loss": 0.8751,
      "step": 21220
    },
    {
      "epoch": 0.33043505336000684,
      "grad_norm": 4.205401420593262,
      "learning_rate": 7.048229298110486e-07,
      "loss": 0.7662,
      "step": 21225
    },
    {
      "epoch": 0.33051289436197623,
      "grad_norm": 3.049184560775757,
      "learning_rate": 7.047409908064437e-07,
      "loss": 0.7602,
      "step": 21230
    },
    {
      "epoch": 0.3305907353639456,
      "grad_norm": 2.9818828105926514,
      "learning_rate": 7.046590518018387e-07,
      "loss": 0.7905,
      "step": 21235
    },
    {
      "epoch": 0.33066857636591496,
      "grad_norm": 5.376505374908447,
      "learning_rate": 7.045771127972337e-07,
      "loss": 0.7248,
      "step": 21240
    },
    {
      "epoch": 0.33074641736788435,
      "grad_norm": 5.111346244812012,
      "learning_rate": 7.044951737926287e-07,
      "loss": 0.7423,
      "step": 21245
    },
    {
      "epoch": 0.33082425836985374,
      "grad_norm": 5.915484428405762,
      "learning_rate": 7.044132347880237e-07,
      "loss": 0.8131,
      "step": 21250
    },
    {
      "epoch": 0.33090209937182313,
      "grad_norm": 3.484762191772461,
      "learning_rate": 7.043312957834188e-07,
      "loss": 0.7364,
      "step": 21255
    },
    {
      "epoch": 0.33097994037379247,
      "grad_norm": 8.695598602294922,
      "learning_rate": 7.042493567788138e-07,
      "loss": 0.7335,
      "step": 21260
    },
    {
      "epoch": 0.33105778137576186,
      "grad_norm": 5.287198066711426,
      "learning_rate": 7.041674177742088e-07,
      "loss": 0.8465,
      "step": 21265
    },
    {
      "epoch": 0.33113562237773125,
      "grad_norm": 5.074387550354004,
      "learning_rate": 7.040854787696039e-07,
      "loss": 0.736,
      "step": 21270
    },
    {
      "epoch": 0.33121346337970065,
      "grad_norm": 4.307338714599609,
      "learning_rate": 7.04003539764999e-07,
      "loss": 0.8083,
      "step": 21275
    },
    {
      "epoch": 0.33129130438167,
      "grad_norm": 4.100557804107666,
      "learning_rate": 7.03921600760394e-07,
      "loss": 0.7717,
      "step": 21280
    },
    {
      "epoch": 0.3313691453836394,
      "grad_norm": 10.841692924499512,
      "learning_rate": 7.038396617557889e-07,
      "loss": 0.6512,
      "step": 21285
    },
    {
      "epoch": 0.33144698638560877,
      "grad_norm": 4.395034313201904,
      "learning_rate": 7.03757722751184e-07,
      "loss": 0.8602,
      "step": 21290
    },
    {
      "epoch": 0.33152482738757816,
      "grad_norm": 4.024008750915527,
      "learning_rate": 7.03675783746579e-07,
      "loss": 0.8101,
      "step": 21295
    },
    {
      "epoch": 0.3316026683895475,
      "grad_norm": 2.2783641815185547,
      "learning_rate": 7.03593844741974e-07,
      "loss": 0.7214,
      "step": 21300
    },
    {
      "epoch": 0.3316805093915169,
      "grad_norm": 2.686100721359253,
      "learning_rate": 7.035119057373691e-07,
      "loss": 0.6999,
      "step": 21305
    },
    {
      "epoch": 0.3317583503934863,
      "grad_norm": 5.044751167297363,
      "learning_rate": 7.034299667327642e-07,
      "loss": 0.7782,
      "step": 21310
    },
    {
      "epoch": 0.3318361913954556,
      "grad_norm": 4.074419021606445,
      "learning_rate": 7.033480277281591e-07,
      "loss": 0.8148,
      "step": 21315
    },
    {
      "epoch": 0.331914032397425,
      "grad_norm": 4.122618198394775,
      "learning_rate": 7.032660887235542e-07,
      "loss": 0.853,
      "step": 21320
    },
    {
      "epoch": 0.3319918733993944,
      "grad_norm": 3.0273494720458984,
      "learning_rate": 7.031841497189493e-07,
      "loss": 0.8329,
      "step": 21325
    },
    {
      "epoch": 0.3320697144013638,
      "grad_norm": 3.330517053604126,
      "learning_rate": 7.031022107143441e-07,
      "loss": 0.7399,
      "step": 21330
    },
    {
      "epoch": 0.33214755540333313,
      "grad_norm": 8.765055656433105,
      "learning_rate": 7.030202717097392e-07,
      "loss": 0.7669,
      "step": 21335
    },
    {
      "epoch": 0.3322253964053025,
      "grad_norm": 2.513746500015259,
      "learning_rate": 7.029383327051343e-07,
      "loss": 0.7557,
      "step": 21340
    },
    {
      "epoch": 0.3323032374072719,
      "grad_norm": 2.607100009918213,
      "learning_rate": 7.028563937005293e-07,
      "loss": 0.8703,
      "step": 21345
    },
    {
      "epoch": 0.3323810784092413,
      "grad_norm": 2.982377290725708,
      "learning_rate": 7.027744546959243e-07,
      "loss": 0.7169,
      "step": 21350
    },
    {
      "epoch": 0.33245891941121064,
      "grad_norm": 3.663597822189331,
      "learning_rate": 7.026925156913194e-07,
      "loss": 0.7615,
      "step": 21355
    },
    {
      "epoch": 0.33253676041318003,
      "grad_norm": 3.046760320663452,
      "learning_rate": 7.026105766867144e-07,
      "loss": 0.7096,
      "step": 21360
    },
    {
      "epoch": 0.3326146014151494,
      "grad_norm": 3.6248621940612793,
      "learning_rate": 7.025286376821094e-07,
      "loss": 0.7434,
      "step": 21365
    },
    {
      "epoch": 0.3326924424171188,
      "grad_norm": 4.510558605194092,
      "learning_rate": 7.024466986775044e-07,
      "loss": 0.8041,
      "step": 21370
    },
    {
      "epoch": 0.33277028341908815,
      "grad_norm": 4.688333511352539,
      "learning_rate": 7.023647596728994e-07,
      "loss": 0.727,
      "step": 21375
    },
    {
      "epoch": 0.33284812442105755,
      "grad_norm": 5.505545616149902,
      "learning_rate": 7.022828206682945e-07,
      "loss": 0.8143,
      "step": 21380
    },
    {
      "epoch": 0.33292596542302694,
      "grad_norm": 3.2059834003448486,
      "learning_rate": 7.022008816636895e-07,
      "loss": 0.6771,
      "step": 21385
    },
    {
      "epoch": 0.33300380642499633,
      "grad_norm": 3.3329732418060303,
      "learning_rate": 7.021189426590845e-07,
      "loss": 0.7855,
      "step": 21390
    },
    {
      "epoch": 0.33308164742696567,
      "grad_norm": 4.149011611938477,
      "learning_rate": 7.020370036544796e-07,
      "loss": 0.8071,
      "step": 21395
    },
    {
      "epoch": 0.33315948842893506,
      "grad_norm": 4.561814308166504,
      "learning_rate": 7.019550646498747e-07,
      "loss": 0.6988,
      "step": 21400
    },
    {
      "epoch": 0.33323732943090445,
      "grad_norm": 3.655268669128418,
      "learning_rate": 7.018731256452697e-07,
      "loss": 0.7151,
      "step": 21405
    },
    {
      "epoch": 0.3333151704328738,
      "grad_norm": 4.033401012420654,
      "learning_rate": 7.017911866406646e-07,
      "loss": 0.7491,
      "step": 21410
    },
    {
      "epoch": 0.3333930114348432,
      "grad_norm": 3.786790609359741,
      "learning_rate": 7.017092476360597e-07,
      "loss": 0.7232,
      "step": 21415
    },
    {
      "epoch": 0.33347085243681257,
      "grad_norm": 3.5973622798919678,
      "learning_rate": 7.016273086314547e-07,
      "loss": 0.7981,
      "step": 21420
    },
    {
      "epoch": 0.33354869343878196,
      "grad_norm": 4.2606682777404785,
      "learning_rate": 7.015453696268497e-07,
      "loss": 0.791,
      "step": 21425
    },
    {
      "epoch": 0.3336265344407513,
      "grad_norm": 3.1162331104278564,
      "learning_rate": 7.014634306222448e-07,
      "loss": 0.7806,
      "step": 21430
    },
    {
      "epoch": 0.3337043754427207,
      "grad_norm": 2.6775388717651367,
      "learning_rate": 7.013814916176399e-07,
      "loss": 0.7989,
      "step": 21435
    },
    {
      "epoch": 0.3337822164446901,
      "grad_norm": 3.595301628112793,
      "learning_rate": 7.012995526130348e-07,
      "loss": 0.7593,
      "step": 21440
    },
    {
      "epoch": 0.3338600574466595,
      "grad_norm": 4.060016632080078,
      "learning_rate": 7.012176136084299e-07,
      "loss": 0.6843,
      "step": 21445
    },
    {
      "epoch": 0.3339378984486288,
      "grad_norm": 3.22746205329895,
      "learning_rate": 7.011356746038249e-07,
      "loss": 0.6879,
      "step": 21450
    },
    {
      "epoch": 0.3340157394505982,
      "grad_norm": 2.997220993041992,
      "learning_rate": 7.010537355992198e-07,
      "loss": 0.7744,
      "step": 21455
    },
    {
      "epoch": 0.3340935804525676,
      "grad_norm": 3.810737371444702,
      "learning_rate": 7.009717965946149e-07,
      "loss": 0.7834,
      "step": 21460
    },
    {
      "epoch": 0.334171421454537,
      "grad_norm": 4.1732988357543945,
      "learning_rate": 7.0088985759001e-07,
      "loss": 0.7195,
      "step": 21465
    },
    {
      "epoch": 0.3342492624565063,
      "grad_norm": 2.8913424015045166,
      "learning_rate": 7.00807918585405e-07,
      "loss": 0.6206,
      "step": 21470
    },
    {
      "epoch": 0.3343271034584757,
      "grad_norm": 3.0899922847747803,
      "learning_rate": 7.007259795808e-07,
      "loss": 0.7757,
      "step": 21475
    },
    {
      "epoch": 0.3344049444604451,
      "grad_norm": 5.782576084136963,
      "learning_rate": 7.006440405761951e-07,
      "loss": 0.6863,
      "step": 21480
    },
    {
      "epoch": 0.33448278546241444,
      "grad_norm": 8.251847267150879,
      "learning_rate": 7.005621015715901e-07,
      "loss": 0.6977,
      "step": 21485
    },
    {
      "epoch": 0.33456062646438384,
      "grad_norm": 3.7665064334869385,
      "learning_rate": 7.00480162566985e-07,
      "loss": 0.7871,
      "step": 21490
    },
    {
      "epoch": 0.33463846746635323,
      "grad_norm": 4.690029144287109,
      "learning_rate": 7.003982235623801e-07,
      "loss": 0.7801,
      "step": 21495
    },
    {
      "epoch": 0.3347163084683226,
      "grad_norm": 2.9794564247131348,
      "learning_rate": 7.003162845577751e-07,
      "loss": 0.7675,
      "step": 21500
    },
    {
      "epoch": 0.33479414947029196,
      "grad_norm": 4.976625442504883,
      "learning_rate": 7.002343455531702e-07,
      "loss": 0.8075,
      "step": 21505
    },
    {
      "epoch": 0.33487199047226135,
      "grad_norm": 2.9241256713867188,
      "learning_rate": 7.001524065485652e-07,
      "loss": 0.7282,
      "step": 21510
    },
    {
      "epoch": 0.33494983147423074,
      "grad_norm": 4.694438457489014,
      "learning_rate": 7.000704675439602e-07,
      "loss": 0.8939,
      "step": 21515
    },
    {
      "epoch": 0.33502767247620013,
      "grad_norm": 4.293147563934326,
      "learning_rate": 6.999885285393553e-07,
      "loss": 0.7264,
      "step": 21520
    },
    {
      "epoch": 0.33510551347816947,
      "grad_norm": 10.977949142456055,
      "learning_rate": 6.999065895347504e-07,
      "loss": 0.7685,
      "step": 21525
    },
    {
      "epoch": 0.33518335448013886,
      "grad_norm": 4.644759178161621,
      "learning_rate": 6.998246505301452e-07,
      "loss": 0.7754,
      "step": 21530
    },
    {
      "epoch": 0.33526119548210825,
      "grad_norm": 5.421177864074707,
      "learning_rate": 6.997427115255403e-07,
      "loss": 0.7976,
      "step": 21535
    },
    {
      "epoch": 0.33533903648407765,
      "grad_norm": 3.4674479961395264,
      "learning_rate": 6.996607725209354e-07,
      "loss": 0.6951,
      "step": 21540
    },
    {
      "epoch": 0.335416877486047,
      "grad_norm": 4.617788314819336,
      "learning_rate": 6.995788335163304e-07,
      "loss": 0.7562,
      "step": 21545
    },
    {
      "epoch": 0.3354947184880164,
      "grad_norm": 2.9097959995269775,
      "learning_rate": 6.994968945117254e-07,
      "loss": 0.6994,
      "step": 21550
    },
    {
      "epoch": 0.33557255948998577,
      "grad_norm": 4.879605770111084,
      "learning_rate": 6.994149555071205e-07,
      "loss": 0.7559,
      "step": 21555
    },
    {
      "epoch": 0.33565040049195516,
      "grad_norm": 2.8688557147979736,
      "learning_rate": 6.993330165025156e-07,
      "loss": 0.7733,
      "step": 21560
    },
    {
      "epoch": 0.3357282414939245,
      "grad_norm": 3.4021968841552734,
      "learning_rate": 6.992510774979105e-07,
      "loss": 0.8755,
      "step": 21565
    },
    {
      "epoch": 0.3358060824958939,
      "grad_norm": 6.432661056518555,
      "learning_rate": 6.991691384933055e-07,
      "loss": 0.6555,
      "step": 21570
    },
    {
      "epoch": 0.3358839234978633,
      "grad_norm": 3.397725820541382,
      "learning_rate": 6.990871994887006e-07,
      "loss": 0.8091,
      "step": 21575
    },
    {
      "epoch": 0.3359617644998326,
      "grad_norm": 2.418670177459717,
      "learning_rate": 6.990052604840956e-07,
      "loss": 0.7853,
      "step": 21580
    },
    {
      "epoch": 0.336039605501802,
      "grad_norm": 3.4952597618103027,
      "learning_rate": 6.989233214794906e-07,
      "loss": 0.769,
      "step": 21585
    },
    {
      "epoch": 0.3361174465037714,
      "grad_norm": 4.628364086151123,
      "learning_rate": 6.988413824748857e-07,
      "loss": 0.7542,
      "step": 21590
    },
    {
      "epoch": 0.3361952875057408,
      "grad_norm": 4.102391242980957,
      "learning_rate": 6.987594434702807e-07,
      "loss": 0.7571,
      "step": 21595
    },
    {
      "epoch": 0.33627312850771013,
      "grad_norm": 3.2335331439971924,
      "learning_rate": 6.986775044656757e-07,
      "loss": 0.6834,
      "step": 21600
    },
    {
      "epoch": 0.3363509695096795,
      "grad_norm": 7.069674968719482,
      "learning_rate": 6.985955654610708e-07,
      "loss": 0.8251,
      "step": 21605
    },
    {
      "epoch": 0.3364288105116489,
      "grad_norm": 4.058067798614502,
      "learning_rate": 6.985136264564657e-07,
      "loss": 0.8077,
      "step": 21610
    },
    {
      "epoch": 0.3365066515136183,
      "grad_norm": 3.856201171875,
      "learning_rate": 6.984316874518608e-07,
      "loss": 0.8431,
      "step": 21615
    },
    {
      "epoch": 0.33658449251558764,
      "grad_norm": 5.469402313232422,
      "learning_rate": 6.983497484472558e-07,
      "loss": 0.7463,
      "step": 21620
    },
    {
      "epoch": 0.33666233351755703,
      "grad_norm": 4.18521785736084,
      "learning_rate": 6.982678094426508e-07,
      "loss": 0.9012,
      "step": 21625
    },
    {
      "epoch": 0.3367401745195264,
      "grad_norm": 3.856855869293213,
      "learning_rate": 6.981858704380459e-07,
      "loss": 0.7926,
      "step": 21630
    },
    {
      "epoch": 0.3368180155214958,
      "grad_norm": 3.008106231689453,
      "learning_rate": 6.98103931433441e-07,
      "loss": 0.7308,
      "step": 21635
    },
    {
      "epoch": 0.33689585652346515,
      "grad_norm": 5.669660568237305,
      "learning_rate": 6.980219924288359e-07,
      "loss": 0.7562,
      "step": 21640
    },
    {
      "epoch": 0.33697369752543455,
      "grad_norm": 4.197543144226074,
      "learning_rate": 6.97940053424231e-07,
      "loss": 0.8628,
      "step": 21645
    },
    {
      "epoch": 0.33705153852740394,
      "grad_norm": 2.6384623050689697,
      "learning_rate": 6.978581144196261e-07,
      "loss": 0.6264,
      "step": 21650
    },
    {
      "epoch": 0.33712937952937333,
      "grad_norm": 4.709079265594482,
      "learning_rate": 6.977761754150209e-07,
      "loss": 0.7781,
      "step": 21655
    },
    {
      "epoch": 0.33720722053134267,
      "grad_norm": 5.100217342376709,
      "learning_rate": 6.97694236410416e-07,
      "loss": 0.7359,
      "step": 21660
    },
    {
      "epoch": 0.33728506153331206,
      "grad_norm": 4.674426555633545,
      "learning_rate": 6.976122974058111e-07,
      "loss": 0.7898,
      "step": 21665
    },
    {
      "epoch": 0.33736290253528145,
      "grad_norm": 5.750451564788818,
      "learning_rate": 6.975303584012062e-07,
      "loss": 0.7547,
      "step": 21670
    },
    {
      "epoch": 0.3374407435372508,
      "grad_norm": 3.1156442165374756,
      "learning_rate": 6.974484193966011e-07,
      "loss": 0.7708,
      "step": 21675
    },
    {
      "epoch": 0.3375185845392202,
      "grad_norm": 3.386111259460449,
      "learning_rate": 6.973664803919962e-07,
      "loss": 0.7096,
      "step": 21680
    },
    {
      "epoch": 0.33759642554118957,
      "grad_norm": 3.170990228652954,
      "learning_rate": 6.972845413873913e-07,
      "loss": 0.8063,
      "step": 21685
    },
    {
      "epoch": 0.33767426654315896,
      "grad_norm": 2.8356587886810303,
      "learning_rate": 6.972026023827862e-07,
      "loss": 0.7984,
      "step": 21690
    },
    {
      "epoch": 0.3377521075451283,
      "grad_norm": 3.965768814086914,
      "learning_rate": 6.971206633781812e-07,
      "loss": 0.8705,
      "step": 21695
    },
    {
      "epoch": 0.3378299485470977,
      "grad_norm": 4.400123119354248,
      "learning_rate": 6.970387243735763e-07,
      "loss": 0.7655,
      "step": 21700
    },
    {
      "epoch": 0.3379077895490671,
      "grad_norm": 3.1862401962280273,
      "learning_rate": 6.969567853689713e-07,
      "loss": 0.8166,
      "step": 21705
    },
    {
      "epoch": 0.3379856305510365,
      "grad_norm": 3.2615442276000977,
      "learning_rate": 6.968748463643663e-07,
      "loss": 0.7554,
      "step": 21710
    },
    {
      "epoch": 0.3380634715530058,
      "grad_norm": 6.5277419090271,
      "learning_rate": 6.967929073597614e-07,
      "loss": 0.7581,
      "step": 21715
    },
    {
      "epoch": 0.3381413125549752,
      "grad_norm": 6.898044109344482,
      "learning_rate": 6.967109683551564e-07,
      "loss": 0.826,
      "step": 21720
    },
    {
      "epoch": 0.3382191535569446,
      "grad_norm": 3.163782835006714,
      "learning_rate": 6.966290293505515e-07,
      "loss": 0.7342,
      "step": 21725
    },
    {
      "epoch": 0.338296994558914,
      "grad_norm": 4.352728843688965,
      "learning_rate": 6.965470903459465e-07,
      "loss": 0.7732,
      "step": 21730
    },
    {
      "epoch": 0.3383748355608833,
      "grad_norm": 5.486903667449951,
      "learning_rate": 6.964651513413414e-07,
      "loss": 0.845,
      "step": 21735
    },
    {
      "epoch": 0.3384526765628527,
      "grad_norm": 6.931983470916748,
      "learning_rate": 6.963832123367365e-07,
      "loss": 0.9539,
      "step": 21740
    },
    {
      "epoch": 0.3385305175648221,
      "grad_norm": 3.2124457359313965,
      "learning_rate": 6.963012733321315e-07,
      "loss": 0.7011,
      "step": 21745
    },
    {
      "epoch": 0.33860835856679145,
      "grad_norm": 4.073243618011475,
      "learning_rate": 6.962193343275265e-07,
      "loss": 0.8017,
      "step": 21750
    },
    {
      "epoch": 0.33868619956876084,
      "grad_norm": 2.7231483459472656,
      "learning_rate": 6.961373953229216e-07,
      "loss": 0.7914,
      "step": 21755
    },
    {
      "epoch": 0.33876404057073023,
      "grad_norm": 4.636868953704834,
      "learning_rate": 6.960554563183167e-07,
      "loss": 0.6745,
      "step": 21760
    },
    {
      "epoch": 0.3388418815726996,
      "grad_norm": 3.260817289352417,
      "learning_rate": 6.959735173137116e-07,
      "loss": 0.7471,
      "step": 21765
    },
    {
      "epoch": 0.33891972257466896,
      "grad_norm": 3.095573663711548,
      "learning_rate": 6.958915783091067e-07,
      "loss": 0.7908,
      "step": 21770
    },
    {
      "epoch": 0.33899756357663835,
      "grad_norm": 4.436625003814697,
      "learning_rate": 6.958096393045017e-07,
      "loss": 0.8482,
      "step": 21775
    },
    {
      "epoch": 0.33907540457860774,
      "grad_norm": 3.3770432472229004,
      "learning_rate": 6.957277002998966e-07,
      "loss": 0.6528,
      "step": 21780
    },
    {
      "epoch": 0.33915324558057713,
      "grad_norm": 4.698873996734619,
      "learning_rate": 6.956457612952917e-07,
      "loss": 0.7155,
      "step": 21785
    },
    {
      "epoch": 0.33923108658254647,
      "grad_norm": 5.815252304077148,
      "learning_rate": 6.955638222906868e-07,
      "loss": 0.7678,
      "step": 21790
    },
    {
      "epoch": 0.33930892758451586,
      "grad_norm": 5.941638946533203,
      "learning_rate": 6.954818832860819e-07,
      "loss": 0.7872,
      "step": 21795
    },
    {
      "epoch": 0.33938676858648525,
      "grad_norm": 4.159576416015625,
      "learning_rate": 6.953999442814768e-07,
      "loss": 0.8849,
      "step": 21800
    },
    {
      "epoch": 0.33946460958845465,
      "grad_norm": 2.9337544441223145,
      "learning_rate": 6.953180052768719e-07,
      "loss": 0.705,
      "step": 21805
    },
    {
      "epoch": 0.339542450590424,
      "grad_norm": 2.750241994857788,
      "learning_rate": 6.95236066272267e-07,
      "loss": 0.7139,
      "step": 21810
    },
    {
      "epoch": 0.3396202915923934,
      "grad_norm": 5.717836856842041,
      "learning_rate": 6.951541272676618e-07,
      "loss": 0.884,
      "step": 21815
    },
    {
      "epoch": 0.33969813259436277,
      "grad_norm": 3.4549524784088135,
      "learning_rate": 6.950721882630569e-07,
      "loss": 0.7177,
      "step": 21820
    },
    {
      "epoch": 0.33977597359633216,
      "grad_norm": 9.49217700958252,
      "learning_rate": 6.94990249258452e-07,
      "loss": 0.7746,
      "step": 21825
    },
    {
      "epoch": 0.3398538145983015,
      "grad_norm": 6.191872596740723,
      "learning_rate": 6.94908310253847e-07,
      "loss": 0.7398,
      "step": 21830
    },
    {
      "epoch": 0.3399316556002709,
      "grad_norm": 5.124302387237549,
      "learning_rate": 6.94826371249242e-07,
      "loss": 0.8453,
      "step": 21835
    },
    {
      "epoch": 0.3400094966022403,
      "grad_norm": 3.679274082183838,
      "learning_rate": 6.947444322446371e-07,
      "loss": 0.8208,
      "step": 21840
    },
    {
      "epoch": 0.3400873376042096,
      "grad_norm": 4.2432541847229,
      "learning_rate": 6.946624932400321e-07,
      "loss": 0.7919,
      "step": 21845
    },
    {
      "epoch": 0.340165178606179,
      "grad_norm": 4.548547267913818,
      "learning_rate": 6.945805542354272e-07,
      "loss": 0.7669,
      "step": 21850
    },
    {
      "epoch": 0.3402430196081484,
      "grad_norm": 9.537131309509277,
      "learning_rate": 6.944986152308221e-07,
      "loss": 0.7988,
      "step": 21855
    },
    {
      "epoch": 0.3403208606101178,
      "grad_norm": 3.267117500305176,
      "learning_rate": 6.944166762262171e-07,
      "loss": 0.6798,
      "step": 21860
    },
    {
      "epoch": 0.34039870161208713,
      "grad_norm": 4.591448783874512,
      "learning_rate": 6.943347372216122e-07,
      "loss": 0.7488,
      "step": 21865
    },
    {
      "epoch": 0.3404765426140565,
      "grad_norm": 2.8945159912109375,
      "learning_rate": 6.942527982170072e-07,
      "loss": 0.8226,
      "step": 21870
    },
    {
      "epoch": 0.3405543836160259,
      "grad_norm": 3.5019824504852295,
      "learning_rate": 6.941708592124022e-07,
      "loss": 0.7698,
      "step": 21875
    },
    {
      "epoch": 0.3406322246179953,
      "grad_norm": 3.800676107406616,
      "learning_rate": 6.940889202077973e-07,
      "loss": 0.7195,
      "step": 21880
    },
    {
      "epoch": 0.34071006561996464,
      "grad_norm": 6.936948776245117,
      "learning_rate": 6.940069812031924e-07,
      "loss": 0.7831,
      "step": 21885
    },
    {
      "epoch": 0.34078790662193403,
      "grad_norm": 2.993300676345825,
      "learning_rate": 6.939250421985873e-07,
      "loss": 0.7205,
      "step": 21890
    },
    {
      "epoch": 0.3408657476239034,
      "grad_norm": 4.39565372467041,
      "learning_rate": 6.938431031939823e-07,
      "loss": 0.809,
      "step": 21895
    },
    {
      "epoch": 0.3409435886258728,
      "grad_norm": 3.0325002670288086,
      "learning_rate": 6.937611641893774e-07,
      "loss": 0.6348,
      "step": 21900
    },
    {
      "epoch": 0.34102142962784215,
      "grad_norm": 2.1405999660491943,
      "learning_rate": 6.936792251847723e-07,
      "loss": 0.7687,
      "step": 21905
    },
    {
      "epoch": 0.34109927062981155,
      "grad_norm": 3.2362265586853027,
      "learning_rate": 6.935972861801674e-07,
      "loss": 0.9177,
      "step": 21910
    },
    {
      "epoch": 0.34117711163178094,
      "grad_norm": 7.863826751708984,
      "learning_rate": 6.935153471755625e-07,
      "loss": 0.8616,
      "step": 21915
    },
    {
      "epoch": 0.3412549526337503,
      "grad_norm": 3.197587490081787,
      "learning_rate": 6.934334081709576e-07,
      "loss": 0.6349,
      "step": 21920
    },
    {
      "epoch": 0.34133279363571967,
      "grad_norm": 3.8425405025482178,
      "learning_rate": 6.933514691663525e-07,
      "loss": 0.7869,
      "step": 21925
    },
    {
      "epoch": 0.34141063463768906,
      "grad_norm": 3.0469181537628174,
      "learning_rate": 6.932695301617476e-07,
      "loss": 0.8481,
      "step": 21930
    },
    {
      "epoch": 0.34148847563965845,
      "grad_norm": 2.5239310264587402,
      "learning_rate": 6.931875911571426e-07,
      "loss": 0.7255,
      "step": 21935
    },
    {
      "epoch": 0.3415663166416278,
      "grad_norm": 5.252364635467529,
      "learning_rate": 6.931056521525376e-07,
      "loss": 0.7062,
      "step": 21940
    },
    {
      "epoch": 0.3416441576435972,
      "grad_norm": 3.408431053161621,
      "learning_rate": 6.930237131479326e-07,
      "loss": 0.7958,
      "step": 21945
    },
    {
      "epoch": 0.34172199864556657,
      "grad_norm": 3.285446882247925,
      "learning_rate": 6.929417741433277e-07,
      "loss": 0.7819,
      "step": 21950
    },
    {
      "epoch": 0.34179983964753596,
      "grad_norm": 4.0770368576049805,
      "learning_rate": 6.928598351387227e-07,
      "loss": 0.8164,
      "step": 21955
    },
    {
      "epoch": 0.3418776806495053,
      "grad_norm": 3.2159502506256104,
      "learning_rate": 6.927778961341177e-07,
      "loss": 0.7799,
      "step": 21960
    },
    {
      "epoch": 0.3419555216514747,
      "grad_norm": 3.2708003520965576,
      "learning_rate": 6.926959571295128e-07,
      "loss": 0.8103,
      "step": 21965
    },
    {
      "epoch": 0.3420333626534441,
      "grad_norm": 5.362832069396973,
      "learning_rate": 6.926140181249078e-07,
      "loss": 0.7317,
      "step": 21970
    },
    {
      "epoch": 0.3421112036554135,
      "grad_norm": 3.9707159996032715,
      "learning_rate": 6.925320791203029e-07,
      "loss": 0.7181,
      "step": 21975
    },
    {
      "epoch": 0.3421890446573828,
      "grad_norm": 4.492808818817139,
      "learning_rate": 6.924501401156978e-07,
      "loss": 0.8526,
      "step": 21980
    },
    {
      "epoch": 0.3422668856593522,
      "grad_norm": 2.531547784805298,
      "learning_rate": 6.923682011110928e-07,
      "loss": 0.6497,
      "step": 21985
    },
    {
      "epoch": 0.3423447266613216,
      "grad_norm": 4.880797863006592,
      "learning_rate": 6.922862621064879e-07,
      "loss": 0.7759,
      "step": 21990
    },
    {
      "epoch": 0.342422567663291,
      "grad_norm": 5.803737640380859,
      "learning_rate": 6.92204323101883e-07,
      "loss": 0.8622,
      "step": 21995
    },
    {
      "epoch": 0.3425004086652603,
      "grad_norm": 4.203178882598877,
      "learning_rate": 6.921223840972779e-07,
      "loss": 0.8575,
      "step": 22000
    },
    {
      "epoch": 0.3425782496672297,
      "grad_norm": 6.29517936706543,
      "learning_rate": 6.92040445092673e-07,
      "loss": 0.814,
      "step": 22005
    },
    {
      "epoch": 0.3426560906691991,
      "grad_norm": 3.598208427429199,
      "learning_rate": 6.919585060880681e-07,
      "loss": 0.7939,
      "step": 22010
    },
    {
      "epoch": 0.34273393167116845,
      "grad_norm": 4.675456523895264,
      "learning_rate": 6.91876567083463e-07,
      "loss": 0.8032,
      "step": 22015
    },
    {
      "epoch": 0.34281177267313784,
      "grad_norm": 3.1422770023345947,
      "learning_rate": 6.91794628078858e-07,
      "loss": 0.8217,
      "step": 22020
    },
    {
      "epoch": 0.34288961367510723,
      "grad_norm": 4.944550037384033,
      "learning_rate": 6.917126890742531e-07,
      "loss": 0.6776,
      "step": 22025
    },
    {
      "epoch": 0.3429674546770766,
      "grad_norm": 5.1036057472229,
      "learning_rate": 6.916307500696482e-07,
      "loss": 0.6798,
      "step": 22030
    },
    {
      "epoch": 0.34304529567904596,
      "grad_norm": 4.726799488067627,
      "learning_rate": 6.915488110650431e-07,
      "loss": 0.7783,
      "step": 22035
    },
    {
      "epoch": 0.34312313668101535,
      "grad_norm": 3.355400323867798,
      "learning_rate": 6.914668720604382e-07,
      "loss": 0.7481,
      "step": 22040
    },
    {
      "epoch": 0.34320097768298474,
      "grad_norm": 4.240068435668945,
      "learning_rate": 6.913849330558333e-07,
      "loss": 0.7477,
      "step": 22045
    },
    {
      "epoch": 0.34327881868495413,
      "grad_norm": 5.435959815979004,
      "learning_rate": 6.913029940512282e-07,
      "loss": 0.8529,
      "step": 22050
    },
    {
      "epoch": 0.34335665968692347,
      "grad_norm": 4.2254438400268555,
      "learning_rate": 6.912210550466233e-07,
      "loss": 0.6756,
      "step": 22055
    },
    {
      "epoch": 0.34343450068889286,
      "grad_norm": 5.156310558319092,
      "learning_rate": 6.911391160420183e-07,
      "loss": 0.7669,
      "step": 22060
    },
    {
      "epoch": 0.34351234169086226,
      "grad_norm": 2.5324745178222656,
      "learning_rate": 6.910571770374133e-07,
      "loss": 0.7165,
      "step": 22065
    },
    {
      "epoch": 0.34359018269283165,
      "grad_norm": 5.870659828186035,
      "learning_rate": 6.909752380328083e-07,
      "loss": 0.7372,
      "step": 22070
    },
    {
      "epoch": 0.343668023694801,
      "grad_norm": 3.9795782566070557,
      "learning_rate": 6.908932990282034e-07,
      "loss": 0.7804,
      "step": 22075
    },
    {
      "epoch": 0.3437458646967704,
      "grad_norm": 3.463372230529785,
      "learning_rate": 6.908113600235984e-07,
      "loss": 0.6794,
      "step": 22080
    },
    {
      "epoch": 0.34382370569873977,
      "grad_norm": 2.5854239463806152,
      "learning_rate": 6.907294210189935e-07,
      "loss": 0.8242,
      "step": 22085
    },
    {
      "epoch": 0.3439015467007091,
      "grad_norm": 5.109464168548584,
      "learning_rate": 6.906474820143885e-07,
      "loss": 0.7327,
      "step": 22090
    },
    {
      "epoch": 0.3439793877026785,
      "grad_norm": 3.044029474258423,
      "learning_rate": 6.905655430097835e-07,
      "loss": 0.7222,
      "step": 22095
    },
    {
      "epoch": 0.3440572287046479,
      "grad_norm": 5.811017036437988,
      "learning_rate": 6.904836040051785e-07,
      "loss": 0.812,
      "step": 22100
    },
    {
      "epoch": 0.3441350697066173,
      "grad_norm": 3.0378737449645996,
      "learning_rate": 6.904016650005735e-07,
      "loss": 0.6783,
      "step": 22105
    },
    {
      "epoch": 0.3442129107085866,
      "grad_norm": 3.8113794326782227,
      "learning_rate": 6.903197259959685e-07,
      "loss": 0.7307,
      "step": 22110
    },
    {
      "epoch": 0.344290751710556,
      "grad_norm": 11.150230407714844,
      "learning_rate": 6.902377869913636e-07,
      "loss": 0.7725,
      "step": 22115
    },
    {
      "epoch": 0.3443685927125254,
      "grad_norm": 4.847995758056641,
      "learning_rate": 6.901558479867587e-07,
      "loss": 0.6859,
      "step": 22120
    },
    {
      "epoch": 0.3444464337144948,
      "grad_norm": 3.970857620239258,
      "learning_rate": 6.900739089821536e-07,
      "loss": 0.859,
      "step": 22125
    },
    {
      "epoch": 0.34452427471646413,
      "grad_norm": 4.519296646118164,
      "learning_rate": 6.899919699775487e-07,
      "loss": 0.664,
      "step": 22130
    },
    {
      "epoch": 0.3446021157184335,
      "grad_norm": 6.204890251159668,
      "learning_rate": 6.899100309729438e-07,
      "loss": 0.6662,
      "step": 22135
    },
    {
      "epoch": 0.3446799567204029,
      "grad_norm": 5.375654697418213,
      "learning_rate": 6.898280919683386e-07,
      "loss": 0.7193,
      "step": 22140
    },
    {
      "epoch": 0.3447577977223723,
      "grad_norm": 2.971890449523926,
      "learning_rate": 6.897461529637337e-07,
      "loss": 0.7933,
      "step": 22145
    },
    {
      "epoch": 0.34483563872434164,
      "grad_norm": 2.947749614715576,
      "learning_rate": 6.896642139591288e-07,
      "loss": 0.7621,
      "step": 22150
    },
    {
      "epoch": 0.34491347972631103,
      "grad_norm": 9.173848152160645,
      "learning_rate": 6.895822749545239e-07,
      "loss": 0.8844,
      "step": 22155
    },
    {
      "epoch": 0.3449913207282804,
      "grad_norm": 3.6220645904541016,
      "learning_rate": 6.895003359499188e-07,
      "loss": 0.8889,
      "step": 22160
    },
    {
      "epoch": 0.3450691617302498,
      "grad_norm": 3.2141635417938232,
      "learning_rate": 6.894183969453139e-07,
      "loss": 0.752,
      "step": 22165
    },
    {
      "epoch": 0.34514700273221915,
      "grad_norm": 3.2809619903564453,
      "learning_rate": 6.89336457940709e-07,
      "loss": 0.7753,
      "step": 22170
    },
    {
      "epoch": 0.34522484373418855,
      "grad_norm": 5.106159687042236,
      "learning_rate": 6.89254518936104e-07,
      "loss": 0.6969,
      "step": 22175
    },
    {
      "epoch": 0.34530268473615794,
      "grad_norm": 3.3226418495178223,
      "learning_rate": 6.891725799314989e-07,
      "loss": 0.7708,
      "step": 22180
    },
    {
      "epoch": 0.3453805257381273,
      "grad_norm": 2.795375347137451,
      "learning_rate": 6.89090640926894e-07,
      "loss": 0.8047,
      "step": 22185
    },
    {
      "epoch": 0.34545836674009667,
      "grad_norm": 3.6871438026428223,
      "learning_rate": 6.89008701922289e-07,
      "loss": 0.7531,
      "step": 22190
    },
    {
      "epoch": 0.34553620774206606,
      "grad_norm": 2.721118211746216,
      "learning_rate": 6.88926762917684e-07,
      "loss": 0.7146,
      "step": 22195
    },
    {
      "epoch": 0.34561404874403545,
      "grad_norm": 3.5846962928771973,
      "learning_rate": 6.888448239130791e-07,
      "loss": 0.8803,
      "step": 22200
    },
    {
      "epoch": 0.3456918897460048,
      "grad_norm": 3.11128568649292,
      "learning_rate": 6.887628849084741e-07,
      "loss": 0.803,
      "step": 22205
    },
    {
      "epoch": 0.3457697307479742,
      "grad_norm": 6.103161334991455,
      "learning_rate": 6.886809459038692e-07,
      "loss": 0.6508,
      "step": 22210
    },
    {
      "epoch": 0.34584757174994357,
      "grad_norm": 4.672749996185303,
      "learning_rate": 6.885990068992642e-07,
      "loss": 0.6122,
      "step": 22215
    },
    {
      "epoch": 0.34592541275191296,
      "grad_norm": 6.16982364654541,
      "learning_rate": 6.885170678946591e-07,
      "loss": 0.6678,
      "step": 22220
    },
    {
      "epoch": 0.3460032537538823,
      "grad_norm": 3.5904715061187744,
      "learning_rate": 6.884351288900542e-07,
      "loss": 0.7361,
      "step": 22225
    },
    {
      "epoch": 0.3460810947558517,
      "grad_norm": 3.5172922611236572,
      "learning_rate": 6.883531898854492e-07,
      "loss": 0.8611,
      "step": 22230
    },
    {
      "epoch": 0.3461589357578211,
      "grad_norm": 4.074573993682861,
      "learning_rate": 6.882712508808442e-07,
      "loss": 0.8436,
      "step": 22235
    },
    {
      "epoch": 0.3462367767597905,
      "grad_norm": 3.7133467197418213,
      "learning_rate": 6.881893118762393e-07,
      "loss": 0.8602,
      "step": 22240
    },
    {
      "epoch": 0.3463146177617598,
      "grad_norm": 3.753530263900757,
      "learning_rate": 6.881073728716344e-07,
      "loss": 0.6924,
      "step": 22245
    },
    {
      "epoch": 0.3463924587637292,
      "grad_norm": 3.2277767658233643,
      "learning_rate": 6.880254338670293e-07,
      "loss": 0.7164,
      "step": 22250
    },
    {
      "epoch": 0.3464702997656986,
      "grad_norm": 4.366305828094482,
      "learning_rate": 6.879434948624244e-07,
      "loss": 0.7426,
      "step": 22255
    },
    {
      "epoch": 0.34654814076766793,
      "grad_norm": 3.119943618774414,
      "learning_rate": 6.878615558578194e-07,
      "loss": 0.7346,
      "step": 22260
    },
    {
      "epoch": 0.3466259817696373,
      "grad_norm": 6.399360179901123,
      "learning_rate": 6.877796168532143e-07,
      "loss": 0.8213,
      "step": 22265
    },
    {
      "epoch": 0.3467038227716067,
      "grad_norm": 2.9667718410491943,
      "learning_rate": 6.876976778486094e-07,
      "loss": 0.778,
      "step": 22270
    },
    {
      "epoch": 0.3467816637735761,
      "grad_norm": 12.54854679107666,
      "learning_rate": 6.876157388440045e-07,
      "loss": 0.71,
      "step": 22275
    },
    {
      "epoch": 0.34685950477554545,
      "grad_norm": 3.2707033157348633,
      "learning_rate": 6.875337998393996e-07,
      "loss": 0.7795,
      "step": 22280
    },
    {
      "epoch": 0.34693734577751484,
      "grad_norm": 3.859928607940674,
      "learning_rate": 6.874518608347945e-07,
      "loss": 0.7869,
      "step": 22285
    },
    {
      "epoch": 0.34701518677948423,
      "grad_norm": 6.35730504989624,
      "learning_rate": 6.873699218301896e-07,
      "loss": 0.8143,
      "step": 22290
    },
    {
      "epoch": 0.3470930277814536,
      "grad_norm": 6.358939170837402,
      "learning_rate": 6.872879828255847e-07,
      "loss": 0.8757,
      "step": 22295
    },
    {
      "epoch": 0.34717086878342296,
      "grad_norm": 3.4988765716552734,
      "learning_rate": 6.872060438209797e-07,
      "loss": 0.6424,
      "step": 22300
    },
    {
      "epoch": 0.34724870978539235,
      "grad_norm": 3.9090397357940674,
      "learning_rate": 6.871241048163746e-07,
      "loss": 0.8496,
      "step": 22305
    },
    {
      "epoch": 0.34732655078736174,
      "grad_norm": 5.115474700927734,
      "learning_rate": 6.870421658117697e-07,
      "loss": 0.7153,
      "step": 22310
    },
    {
      "epoch": 0.34740439178933114,
      "grad_norm": 3.7270002365112305,
      "learning_rate": 6.869602268071647e-07,
      "loss": 0.8032,
      "step": 22315
    },
    {
      "epoch": 0.34748223279130047,
      "grad_norm": 3.265317678451538,
      "learning_rate": 6.868782878025597e-07,
      "loss": 0.7228,
      "step": 22320
    },
    {
      "epoch": 0.34756007379326986,
      "grad_norm": 2.9664700031280518,
      "learning_rate": 6.867963487979548e-07,
      "loss": 0.6512,
      "step": 22325
    },
    {
      "epoch": 0.34763791479523926,
      "grad_norm": 6.434524059295654,
      "learning_rate": 6.867144097933498e-07,
      "loss": 0.7779,
      "step": 22330
    },
    {
      "epoch": 0.34771575579720865,
      "grad_norm": 7.022162437438965,
      "learning_rate": 6.866324707887449e-07,
      "loss": 0.8811,
      "step": 22335
    },
    {
      "epoch": 0.347793596799178,
      "grad_norm": 4.4307427406311035,
      "learning_rate": 6.865505317841399e-07,
      "loss": 0.7314,
      "step": 22340
    },
    {
      "epoch": 0.3478714378011474,
      "grad_norm": 5.711907386779785,
      "learning_rate": 6.864685927795348e-07,
      "loss": 0.7039,
      "step": 22345
    },
    {
      "epoch": 0.34794927880311677,
      "grad_norm": 2.9050698280334473,
      "learning_rate": 6.863866537749299e-07,
      "loss": 0.816,
      "step": 22350
    },
    {
      "epoch": 0.3480271198050861,
      "grad_norm": 2.579324960708618,
      "learning_rate": 6.86304714770325e-07,
      "loss": 0.7305,
      "step": 22355
    },
    {
      "epoch": 0.3481049608070555,
      "grad_norm": 3.0604426860809326,
      "learning_rate": 6.862227757657199e-07,
      "loss": 0.735,
      "step": 22360
    },
    {
      "epoch": 0.3481828018090249,
      "grad_norm": 3.0506060123443604,
      "learning_rate": 6.86140836761115e-07,
      "loss": 0.7676,
      "step": 22365
    },
    {
      "epoch": 0.3482606428109943,
      "grad_norm": 3.921900510787964,
      "learning_rate": 6.860588977565101e-07,
      "loss": 0.7507,
      "step": 22370
    },
    {
      "epoch": 0.3483384838129636,
      "grad_norm": 5.635601043701172,
      "learning_rate": 6.85976958751905e-07,
      "loss": 0.7118,
      "step": 22375
    },
    {
      "epoch": 0.348416324814933,
      "grad_norm": 7.334948539733887,
      "learning_rate": 6.858950197473001e-07,
      "loss": 0.6346,
      "step": 22380
    },
    {
      "epoch": 0.3484941658169024,
      "grad_norm": 5.019484996795654,
      "learning_rate": 6.858130807426951e-07,
      "loss": 0.7292,
      "step": 22385
    },
    {
      "epoch": 0.3485720068188718,
      "grad_norm": 2.8227858543395996,
      "learning_rate": 6.857311417380901e-07,
      "loss": 0.8263,
      "step": 22390
    },
    {
      "epoch": 0.34864984782084113,
      "grad_norm": 3.0834336280822754,
      "learning_rate": 6.856492027334851e-07,
      "loss": 0.8384,
      "step": 22395
    },
    {
      "epoch": 0.3487276888228105,
      "grad_norm": 2.915942430496216,
      "learning_rate": 6.855672637288802e-07,
      "loss": 0.6825,
      "step": 22400
    },
    {
      "epoch": 0.3488055298247799,
      "grad_norm": 10.376582145690918,
      "learning_rate": 6.854853247242753e-07,
      "loss": 0.7687,
      "step": 22405
    },
    {
      "epoch": 0.3488833708267493,
      "grad_norm": 4.40883731842041,
      "learning_rate": 6.854033857196702e-07,
      "loss": 0.8783,
      "step": 22410
    },
    {
      "epoch": 0.34896121182871864,
      "grad_norm": 3.5486271381378174,
      "learning_rate": 6.853214467150653e-07,
      "loss": 0.6884,
      "step": 22415
    },
    {
      "epoch": 0.34903905283068803,
      "grad_norm": 3.900449514389038,
      "learning_rate": 6.852395077104604e-07,
      "loss": 0.7433,
      "step": 22420
    },
    {
      "epoch": 0.3491168938326574,
      "grad_norm": 4.826340675354004,
      "learning_rate": 6.851575687058553e-07,
      "loss": 0.6692,
      "step": 22425
    },
    {
      "epoch": 0.3491947348346268,
      "grad_norm": 4.2448930740356445,
      "learning_rate": 6.850756297012503e-07,
      "loss": 0.8648,
      "step": 22430
    },
    {
      "epoch": 0.34927257583659616,
      "grad_norm": 2.990662097930908,
      "learning_rate": 6.849936906966454e-07,
      "loss": 0.7769,
      "step": 22435
    },
    {
      "epoch": 0.34935041683856555,
      "grad_norm": 4.034456729888916,
      "learning_rate": 6.849117516920404e-07,
      "loss": 0.8343,
      "step": 22440
    },
    {
      "epoch": 0.34942825784053494,
      "grad_norm": 6.867827415466309,
      "learning_rate": 6.848298126874355e-07,
      "loss": 0.8551,
      "step": 22445
    },
    {
      "epoch": 0.3495060988425043,
      "grad_norm": 3.230297327041626,
      "learning_rate": 6.847478736828305e-07,
      "loss": 0.7914,
      "step": 22450
    },
    {
      "epoch": 0.34958393984447367,
      "grad_norm": 4.498978137969971,
      "learning_rate": 6.846659346782255e-07,
      "loss": 0.7035,
      "step": 22455
    },
    {
      "epoch": 0.34966178084644306,
      "grad_norm": 4.367350101470947,
      "learning_rate": 6.845839956736206e-07,
      "loss": 0.8114,
      "step": 22460
    },
    {
      "epoch": 0.34973962184841245,
      "grad_norm": 8.373358726501465,
      "learning_rate": 6.845020566690155e-07,
      "loss": 0.7263,
      "step": 22465
    },
    {
      "epoch": 0.3498174628503818,
      "grad_norm": 2.916916608810425,
      "learning_rate": 6.844201176644105e-07,
      "loss": 0.7741,
      "step": 22470
    },
    {
      "epoch": 0.3498953038523512,
      "grad_norm": 3.4376893043518066,
      "learning_rate": 6.843381786598056e-07,
      "loss": 0.7689,
      "step": 22475
    },
    {
      "epoch": 0.3499731448543206,
      "grad_norm": 4.603682994842529,
      "learning_rate": 6.842562396552007e-07,
      "loss": 0.7048,
      "step": 22480
    },
    {
      "epoch": 0.35005098585628996,
      "grad_norm": 6.618813991546631,
      "learning_rate": 6.841743006505956e-07,
      "loss": 0.7252,
      "step": 22485
    },
    {
      "epoch": 0.3501288268582593,
      "grad_norm": 2.3469929695129395,
      "learning_rate": 6.840923616459907e-07,
      "loss": 0.621,
      "step": 22490
    },
    {
      "epoch": 0.3502066678602287,
      "grad_norm": 4.470626354217529,
      "learning_rate": 6.840104226413858e-07,
      "loss": 0.7745,
      "step": 22495
    },
    {
      "epoch": 0.3502845088621981,
      "grad_norm": 3.6722328662872314,
      "learning_rate": 6.839284836367807e-07,
      "loss": 0.7833,
      "step": 22500
    },
    {
      "epoch": 0.3503623498641675,
      "grad_norm": 3.7017297744750977,
      "learning_rate": 6.838465446321757e-07,
      "loss": 0.7217,
      "step": 22505
    },
    {
      "epoch": 0.3504401908661368,
      "grad_norm": 3.2131004333496094,
      "learning_rate": 6.837646056275708e-07,
      "loss": 0.7248,
      "step": 22510
    },
    {
      "epoch": 0.3505180318681062,
      "grad_norm": 3.843090772628784,
      "learning_rate": 6.836826666229658e-07,
      "loss": 0.7478,
      "step": 22515
    },
    {
      "epoch": 0.3505958728700756,
      "grad_norm": 6.186388969421387,
      "learning_rate": 6.836007276183608e-07,
      "loss": 0.6838,
      "step": 22520
    },
    {
      "epoch": 0.35067371387204493,
      "grad_norm": 14.013809204101562,
      "learning_rate": 6.835187886137559e-07,
      "loss": 0.7547,
      "step": 22525
    },
    {
      "epoch": 0.3507515548740143,
      "grad_norm": 4.710532188415527,
      "learning_rate": 6.83436849609151e-07,
      "loss": 0.8102,
      "step": 22530
    },
    {
      "epoch": 0.3508293958759837,
      "grad_norm": 3.910647392272949,
      "learning_rate": 6.83354910604546e-07,
      "loss": 0.6972,
      "step": 22535
    },
    {
      "epoch": 0.3509072368779531,
      "grad_norm": 4.449835300445557,
      "learning_rate": 6.83272971599941e-07,
      "loss": 0.7189,
      "step": 22540
    },
    {
      "epoch": 0.35098507787992245,
      "grad_norm": 5.269973278045654,
      "learning_rate": 6.83191032595336e-07,
      "loss": 0.7083,
      "step": 22545
    },
    {
      "epoch": 0.35106291888189184,
      "grad_norm": 7.316697120666504,
      "learning_rate": 6.83109093590731e-07,
      "loss": 0.7048,
      "step": 22550
    },
    {
      "epoch": 0.35114075988386123,
      "grad_norm": 3.925175189971924,
      "learning_rate": 6.83027154586126e-07,
      "loss": 0.7584,
      "step": 22555
    },
    {
      "epoch": 0.3512186008858306,
      "grad_norm": 5.711633682250977,
      "learning_rate": 6.829452155815211e-07,
      "loss": 0.7734,
      "step": 22560
    },
    {
      "epoch": 0.35129644188779996,
      "grad_norm": 3.5585262775421143,
      "learning_rate": 6.828632765769161e-07,
      "loss": 0.7737,
      "step": 22565
    },
    {
      "epoch": 0.35137428288976935,
      "grad_norm": 2.6927120685577393,
      "learning_rate": 6.827813375723112e-07,
      "loss": 0.7866,
      "step": 22570
    },
    {
      "epoch": 0.35145212389173874,
      "grad_norm": 3.7263312339782715,
      "learning_rate": 6.826993985677062e-07,
      "loss": 0.8001,
      "step": 22575
    },
    {
      "epoch": 0.35152996489370814,
      "grad_norm": 5.524526119232178,
      "learning_rate": 6.826174595631012e-07,
      "loss": 0.7609,
      "step": 22580
    },
    {
      "epoch": 0.35160780589567747,
      "grad_norm": 6.121705055236816,
      "learning_rate": 6.825355205584962e-07,
      "loss": 0.8267,
      "step": 22585
    },
    {
      "epoch": 0.35168564689764686,
      "grad_norm": 10.191218376159668,
      "learning_rate": 6.824535815538912e-07,
      "loss": 0.8606,
      "step": 22590
    },
    {
      "epoch": 0.35176348789961626,
      "grad_norm": 3.0002195835113525,
      "learning_rate": 6.823716425492862e-07,
      "loss": 0.644,
      "step": 22595
    },
    {
      "epoch": 0.35184132890158565,
      "grad_norm": 9.535001754760742,
      "learning_rate": 6.822897035446813e-07,
      "loss": 0.7069,
      "step": 22600
    },
    {
      "epoch": 0.351919169903555,
      "grad_norm": 2.9684154987335205,
      "learning_rate": 6.822077645400764e-07,
      "loss": 0.8066,
      "step": 22605
    },
    {
      "epoch": 0.3519970109055244,
      "grad_norm": 3.142479419708252,
      "learning_rate": 6.821258255354713e-07,
      "loss": 0.8111,
      "step": 22610
    },
    {
      "epoch": 0.35207485190749377,
      "grad_norm": 4.711599349975586,
      "learning_rate": 6.820438865308664e-07,
      "loss": 0.7644,
      "step": 22615
    },
    {
      "epoch": 0.3521526929094631,
      "grad_norm": 5.315770149230957,
      "learning_rate": 6.819619475262615e-07,
      "loss": 0.6956,
      "step": 22620
    },
    {
      "epoch": 0.3522305339114325,
      "grad_norm": 3.60571026802063,
      "learning_rate": 6.818800085216565e-07,
      "loss": 0.7191,
      "step": 22625
    },
    {
      "epoch": 0.3523083749134019,
      "grad_norm": 3.6298365592956543,
      "learning_rate": 6.817980695170514e-07,
      "loss": 0.6508,
      "step": 22630
    },
    {
      "epoch": 0.3523862159153713,
      "grad_norm": 4.676616191864014,
      "learning_rate": 6.817161305124465e-07,
      "loss": 0.8316,
      "step": 22635
    },
    {
      "epoch": 0.3524640569173406,
      "grad_norm": 2.8332366943359375,
      "learning_rate": 6.816341915078415e-07,
      "loss": 0.7621,
      "step": 22640
    },
    {
      "epoch": 0.35254189791931,
      "grad_norm": 4.0415167808532715,
      "learning_rate": 6.815522525032365e-07,
      "loss": 0.6998,
      "step": 22645
    },
    {
      "epoch": 0.3526197389212794,
      "grad_norm": 3.8834304809570312,
      "learning_rate": 6.814703134986316e-07,
      "loss": 0.7337,
      "step": 22650
    },
    {
      "epoch": 0.3526975799232488,
      "grad_norm": 3.6226377487182617,
      "learning_rate": 6.813883744940267e-07,
      "loss": 0.8377,
      "step": 22655
    },
    {
      "epoch": 0.35277542092521813,
      "grad_norm": 6.299158573150635,
      "learning_rate": 6.813064354894217e-07,
      "loss": 0.8646,
      "step": 22660
    },
    {
      "epoch": 0.3528532619271875,
      "grad_norm": 7.230133533477783,
      "learning_rate": 6.812244964848167e-07,
      "loss": 0.8336,
      "step": 22665
    },
    {
      "epoch": 0.3529311029291569,
      "grad_norm": 6.897984981536865,
      "learning_rate": 6.811425574802117e-07,
      "loss": 0.8237,
      "step": 22670
    },
    {
      "epoch": 0.3530089439311263,
      "grad_norm": 5.134638786315918,
      "learning_rate": 6.810606184756067e-07,
      "loss": 0.8048,
      "step": 22675
    },
    {
      "epoch": 0.35308678493309564,
      "grad_norm": 2.6234192848205566,
      "learning_rate": 6.809786794710017e-07,
      "loss": 0.7218,
      "step": 22680
    },
    {
      "epoch": 0.35316462593506504,
      "grad_norm": 4.5663909912109375,
      "learning_rate": 6.808967404663968e-07,
      "loss": 0.7262,
      "step": 22685
    },
    {
      "epoch": 0.3532424669370344,
      "grad_norm": 3.0789310932159424,
      "learning_rate": 6.808148014617918e-07,
      "loss": 0.8084,
      "step": 22690
    },
    {
      "epoch": 0.35332030793900376,
      "grad_norm": 4.5040411949157715,
      "learning_rate": 6.807328624571869e-07,
      "loss": 0.7338,
      "step": 22695
    },
    {
      "epoch": 0.35339814894097316,
      "grad_norm": 3.41920804977417,
      "learning_rate": 6.806509234525819e-07,
      "loss": 0.8064,
      "step": 22700
    },
    {
      "epoch": 0.35347598994294255,
      "grad_norm": 9.487191200256348,
      "learning_rate": 6.805689844479769e-07,
      "loss": 0.7528,
      "step": 22705
    },
    {
      "epoch": 0.35355383094491194,
      "grad_norm": 4.45808744430542,
      "learning_rate": 6.804870454433719e-07,
      "loss": 0.7882,
      "step": 22710
    },
    {
      "epoch": 0.3536316719468813,
      "grad_norm": 4.73945426940918,
      "learning_rate": 6.80405106438767e-07,
      "loss": 0.8084,
      "step": 22715
    },
    {
      "epoch": 0.35370951294885067,
      "grad_norm": 4.235296726226807,
      "learning_rate": 6.803231674341619e-07,
      "loss": 0.8334,
      "step": 22720
    },
    {
      "epoch": 0.35378735395082006,
      "grad_norm": 4.562685489654541,
      "learning_rate": 6.80241228429557e-07,
      "loss": 0.6529,
      "step": 22725
    },
    {
      "epoch": 0.35386519495278945,
      "grad_norm": 3.7877378463745117,
      "learning_rate": 6.801592894249521e-07,
      "loss": 0.8522,
      "step": 22730
    },
    {
      "epoch": 0.3539430359547588,
      "grad_norm": 7.546112537384033,
      "learning_rate": 6.80077350420347e-07,
      "loss": 0.6945,
      "step": 22735
    },
    {
      "epoch": 0.3540208769567282,
      "grad_norm": 3.382904052734375,
      "learning_rate": 6.799954114157421e-07,
      "loss": 0.8918,
      "step": 22740
    },
    {
      "epoch": 0.3540987179586976,
      "grad_norm": 4.363940238952637,
      "learning_rate": 6.799134724111372e-07,
      "loss": 0.8784,
      "step": 22745
    },
    {
      "epoch": 0.35417655896066697,
      "grad_norm": 4.633853435516357,
      "learning_rate": 6.798315334065321e-07,
      "loss": 0.8348,
      "step": 22750
    },
    {
      "epoch": 0.3542543999626363,
      "grad_norm": 3.2815988063812256,
      "learning_rate": 6.797495944019271e-07,
      "loss": 0.7236,
      "step": 22755
    },
    {
      "epoch": 0.3543322409646057,
      "grad_norm": 3.208714723587036,
      "learning_rate": 6.796676553973222e-07,
      "loss": 0.7547,
      "step": 22760
    },
    {
      "epoch": 0.3544100819665751,
      "grad_norm": 3.2043545246124268,
      "learning_rate": 6.795857163927172e-07,
      "loss": 0.7042,
      "step": 22765
    },
    {
      "epoch": 0.3544879229685445,
      "grad_norm": 3.376326322555542,
      "learning_rate": 6.795037773881122e-07,
      "loss": 0.7247,
      "step": 22770
    },
    {
      "epoch": 0.3545657639705138,
      "grad_norm": 2.1214239597320557,
      "learning_rate": 6.794218383835073e-07,
      "loss": 0.7341,
      "step": 22775
    },
    {
      "epoch": 0.3546436049724832,
      "grad_norm": 3.4116671085357666,
      "learning_rate": 6.793398993789024e-07,
      "loss": 0.8146,
      "step": 22780
    },
    {
      "epoch": 0.3547214459744526,
      "grad_norm": 4.884197235107422,
      "learning_rate": 6.792579603742974e-07,
      "loss": 0.6964,
      "step": 22785
    },
    {
      "epoch": 0.35479928697642193,
      "grad_norm": 2.820373773574829,
      "learning_rate": 6.791760213696923e-07,
      "loss": 0.6723,
      "step": 22790
    },
    {
      "epoch": 0.3548771279783913,
      "grad_norm": 3.8949737548828125,
      "learning_rate": 6.790940823650874e-07,
      "loss": 0.8173,
      "step": 22795
    },
    {
      "epoch": 0.3549549689803607,
      "grad_norm": 4.030924320220947,
      "learning_rate": 6.790121433604824e-07,
      "loss": 0.6769,
      "step": 22800
    },
    {
      "epoch": 0.3550328099823301,
      "grad_norm": 4.274011135101318,
      "learning_rate": 6.789302043558775e-07,
      "loss": 0.6461,
      "step": 22805
    },
    {
      "epoch": 0.35511065098429945,
      "grad_norm": 2.990159749984741,
      "learning_rate": 6.788482653512725e-07,
      "loss": 0.8349,
      "step": 22810
    },
    {
      "epoch": 0.35518849198626884,
      "grad_norm": 2.8439786434173584,
      "learning_rate": 6.787663263466675e-07,
      "loss": 0.7096,
      "step": 22815
    },
    {
      "epoch": 0.35526633298823823,
      "grad_norm": 4.419504642486572,
      "learning_rate": 6.786843873420626e-07,
      "loss": 0.788,
      "step": 22820
    },
    {
      "epoch": 0.3553441739902076,
      "grad_norm": 9.070401191711426,
      "learning_rate": 6.786024483374576e-07,
      "loss": 0.7732,
      "step": 22825
    },
    {
      "epoch": 0.35542201499217696,
      "grad_norm": 4.542810440063477,
      "learning_rate": 6.785205093328525e-07,
      "loss": 0.7192,
      "step": 22830
    },
    {
      "epoch": 0.35549985599414635,
      "grad_norm": 5.248653411865234,
      "learning_rate": 6.784385703282476e-07,
      "loss": 0.7031,
      "step": 22835
    },
    {
      "epoch": 0.35557769699611574,
      "grad_norm": 4.995849132537842,
      "learning_rate": 6.783566313236427e-07,
      "loss": 0.8282,
      "step": 22840
    },
    {
      "epoch": 0.35565553799808514,
      "grad_norm": 4.0547637939453125,
      "learning_rate": 6.782746923190376e-07,
      "loss": 0.7117,
      "step": 22845
    },
    {
      "epoch": 0.3557333790000545,
      "grad_norm": 2.6417508125305176,
      "learning_rate": 6.781927533144327e-07,
      "loss": 0.6833,
      "step": 22850
    },
    {
      "epoch": 0.35581122000202386,
      "grad_norm": 3.902892827987671,
      "learning_rate": 6.781108143098278e-07,
      "loss": 0.7101,
      "step": 22855
    },
    {
      "epoch": 0.35588906100399326,
      "grad_norm": 6.848975658416748,
      "learning_rate": 6.780288753052227e-07,
      "loss": 0.7738,
      "step": 22860
    },
    {
      "epoch": 0.3559669020059626,
      "grad_norm": 5.913900375366211,
      "learning_rate": 6.779469363006178e-07,
      "loss": 0.9128,
      "step": 22865
    },
    {
      "epoch": 0.356044743007932,
      "grad_norm": 4.356508255004883,
      "learning_rate": 6.778649972960128e-07,
      "loss": 0.8422,
      "step": 22870
    },
    {
      "epoch": 0.3561225840099014,
      "grad_norm": 3.5027060508728027,
      "learning_rate": 6.777830582914078e-07,
      "loss": 0.8309,
      "step": 22875
    },
    {
      "epoch": 0.35620042501187077,
      "grad_norm": 3.5927176475524902,
      "learning_rate": 6.777011192868028e-07,
      "loss": 0.7523,
      "step": 22880
    },
    {
      "epoch": 0.3562782660138401,
      "grad_norm": 4.901709079742432,
      "learning_rate": 6.776191802821979e-07,
      "loss": 0.6756,
      "step": 22885
    },
    {
      "epoch": 0.3563561070158095,
      "grad_norm": 3.4579625129699707,
      "learning_rate": 6.775372412775929e-07,
      "loss": 0.7965,
      "step": 22890
    },
    {
      "epoch": 0.3564339480177789,
      "grad_norm": 3.0406250953674316,
      "learning_rate": 6.77455302272988e-07,
      "loss": 0.642,
      "step": 22895
    },
    {
      "epoch": 0.3565117890197483,
      "grad_norm": 4.66744327545166,
      "learning_rate": 6.77373363268383e-07,
      "loss": 0.7916,
      "step": 22900
    },
    {
      "epoch": 0.3565896300217176,
      "grad_norm": 3.9514176845550537,
      "learning_rate": 6.772914242637781e-07,
      "loss": 0.7903,
      "step": 22905
    },
    {
      "epoch": 0.356667471023687,
      "grad_norm": 3.19490385055542,
      "learning_rate": 6.77209485259173e-07,
      "loss": 0.8109,
      "step": 22910
    },
    {
      "epoch": 0.3567453120256564,
      "grad_norm": 3.2852671146392822,
      "learning_rate": 6.77127546254568e-07,
      "loss": 0.7254,
      "step": 22915
    },
    {
      "epoch": 0.3568231530276258,
      "grad_norm": 2.8636245727539062,
      "learning_rate": 6.770456072499631e-07,
      "loss": 0.8545,
      "step": 22920
    },
    {
      "epoch": 0.35690099402959513,
      "grad_norm": 3.2930498123168945,
      "learning_rate": 6.769636682453581e-07,
      "loss": 0.6733,
      "step": 22925
    },
    {
      "epoch": 0.3569788350315645,
      "grad_norm": 4.192903995513916,
      "learning_rate": 6.768817292407532e-07,
      "loss": 0.7281,
      "step": 22930
    },
    {
      "epoch": 0.3570566760335339,
      "grad_norm": 3.524688482284546,
      "learning_rate": 6.767997902361482e-07,
      "loss": 0.773,
      "step": 22935
    },
    {
      "epoch": 0.3571345170355033,
      "grad_norm": 3.3027379512786865,
      "learning_rate": 6.767178512315432e-07,
      "loss": 0.7367,
      "step": 22940
    },
    {
      "epoch": 0.35721235803747264,
      "grad_norm": 5.94819450378418,
      "learning_rate": 6.766359122269383e-07,
      "loss": 0.7967,
      "step": 22945
    },
    {
      "epoch": 0.35729019903944204,
      "grad_norm": 3.1738839149475098,
      "learning_rate": 6.765539732223334e-07,
      "loss": 0.8166,
      "step": 22950
    },
    {
      "epoch": 0.35736804004141143,
      "grad_norm": 5.154567241668701,
      "learning_rate": 6.764720342177282e-07,
      "loss": 0.8211,
      "step": 22955
    },
    {
      "epoch": 0.35744588104338076,
      "grad_norm": 3.465256452560425,
      "learning_rate": 6.763900952131233e-07,
      "loss": 0.7854,
      "step": 22960
    },
    {
      "epoch": 0.35752372204535016,
      "grad_norm": 4.536584854125977,
      "learning_rate": 6.763081562085184e-07,
      "loss": 0.7004,
      "step": 22965
    },
    {
      "epoch": 0.35760156304731955,
      "grad_norm": 8.660713195800781,
      "learning_rate": 6.762262172039133e-07,
      "loss": 0.7063,
      "step": 22970
    },
    {
      "epoch": 0.35767940404928894,
      "grad_norm": 3.680833339691162,
      "learning_rate": 6.761442781993084e-07,
      "loss": 0.691,
      "step": 22975
    },
    {
      "epoch": 0.3577572450512583,
      "grad_norm": 2.5678701400756836,
      "learning_rate": 6.760623391947035e-07,
      "loss": 0.7232,
      "step": 22980
    },
    {
      "epoch": 0.35783508605322767,
      "grad_norm": 5.785712718963623,
      "learning_rate": 6.759804001900985e-07,
      "loss": 0.7438,
      "step": 22985
    },
    {
      "epoch": 0.35791292705519706,
      "grad_norm": 4.117922782897949,
      "learning_rate": 6.758984611854935e-07,
      "loss": 0.7896,
      "step": 22990
    },
    {
      "epoch": 0.35799076805716645,
      "grad_norm": 3.6091392040252686,
      "learning_rate": 6.758165221808885e-07,
      "loss": 0.8481,
      "step": 22995
    },
    {
      "epoch": 0.3580686090591358,
      "grad_norm": 5.465922832489014,
      "learning_rate": 6.757345831762835e-07,
      "loss": 0.7218,
      "step": 23000
    },
    {
      "epoch": 0.3581464500611052,
      "grad_norm": 5.9281206130981445,
      "learning_rate": 6.756526441716785e-07,
      "loss": 0.6711,
      "step": 23005
    },
    {
      "epoch": 0.3582242910630746,
      "grad_norm": 2.9041519165039062,
      "learning_rate": 6.755707051670736e-07,
      "loss": 0.8283,
      "step": 23010
    },
    {
      "epoch": 0.35830213206504397,
      "grad_norm": 4.295798301696777,
      "learning_rate": 6.754887661624686e-07,
      "loss": 0.8098,
      "step": 23015
    },
    {
      "epoch": 0.3583799730670133,
      "grad_norm": 7.160951614379883,
      "learning_rate": 6.754068271578637e-07,
      "loss": 0.7757,
      "step": 23020
    },
    {
      "epoch": 0.3584578140689827,
      "grad_norm": 2.566189765930176,
      "learning_rate": 6.753248881532587e-07,
      "loss": 0.7379,
      "step": 23025
    },
    {
      "epoch": 0.3585356550709521,
      "grad_norm": 4.007600784301758,
      "learning_rate": 6.752429491486538e-07,
      "loss": 0.803,
      "step": 23030
    },
    {
      "epoch": 0.3586134960729215,
      "grad_norm": 3.9632365703582764,
      "learning_rate": 6.751610101440487e-07,
      "loss": 0.7326,
      "step": 23035
    },
    {
      "epoch": 0.3586913370748908,
      "grad_norm": 4.623491287231445,
      "learning_rate": 6.750790711394438e-07,
      "loss": 0.746,
      "step": 23040
    },
    {
      "epoch": 0.3587691780768602,
      "grad_norm": 5.454947471618652,
      "learning_rate": 6.749971321348388e-07,
      "loss": 0.8059,
      "step": 23045
    },
    {
      "epoch": 0.3588470190788296,
      "grad_norm": 2.595407247543335,
      "learning_rate": 6.749151931302338e-07,
      "loss": 0.7712,
      "step": 23050
    },
    {
      "epoch": 0.35892486008079894,
      "grad_norm": 4.200428485870361,
      "learning_rate": 6.748332541256289e-07,
      "loss": 0.6896,
      "step": 23055
    },
    {
      "epoch": 0.3590027010827683,
      "grad_norm": 3.9522597789764404,
      "learning_rate": 6.747513151210239e-07,
      "loss": 0.9086,
      "step": 23060
    },
    {
      "epoch": 0.3590805420847377,
      "grad_norm": 8.130274772644043,
      "learning_rate": 6.746693761164189e-07,
      "loss": 0.7731,
      "step": 23065
    },
    {
      "epoch": 0.3591583830867071,
      "grad_norm": 5.477586269378662,
      "learning_rate": 6.74587437111814e-07,
      "loss": 0.7276,
      "step": 23070
    },
    {
      "epoch": 0.35923622408867645,
      "grad_norm": 5.232213973999023,
      "learning_rate": 6.74505498107209e-07,
      "loss": 0.728,
      "step": 23075
    },
    {
      "epoch": 0.35931406509064584,
      "grad_norm": 4.288384437561035,
      "learning_rate": 6.744235591026039e-07,
      "loss": 0.7833,
      "step": 23080
    },
    {
      "epoch": 0.35939190609261523,
      "grad_norm": 10.313909530639648,
      "learning_rate": 6.74341620097999e-07,
      "loss": 0.8917,
      "step": 23085
    },
    {
      "epoch": 0.3594697470945846,
      "grad_norm": 3.7162394523620605,
      "learning_rate": 6.742596810933941e-07,
      "loss": 0.8603,
      "step": 23090
    },
    {
      "epoch": 0.35954758809655396,
      "grad_norm": 4.621378421783447,
      "learning_rate": 6.74177742088789e-07,
      "loss": 0.6715,
      "step": 23095
    },
    {
      "epoch": 0.35962542909852335,
      "grad_norm": 3.898655652999878,
      "learning_rate": 6.740958030841841e-07,
      "loss": 0.7817,
      "step": 23100
    },
    {
      "epoch": 0.35970327010049274,
      "grad_norm": 3.6938929557800293,
      "learning_rate": 6.740138640795792e-07,
      "loss": 0.7954,
      "step": 23105
    },
    {
      "epoch": 0.35978111110246214,
      "grad_norm": 3.4234812259674072,
      "learning_rate": 6.739319250749742e-07,
      "loss": 0.7543,
      "step": 23110
    },
    {
      "epoch": 0.3598589521044315,
      "grad_norm": 8.834489822387695,
      "learning_rate": 6.738499860703691e-07,
      "loss": 0.7172,
      "step": 23115
    },
    {
      "epoch": 0.35993679310640087,
      "grad_norm": 4.670121192932129,
      "learning_rate": 6.737680470657642e-07,
      "loss": 0.6233,
      "step": 23120
    },
    {
      "epoch": 0.36001463410837026,
      "grad_norm": 5.580198287963867,
      "learning_rate": 6.736861080611592e-07,
      "loss": 0.8335,
      "step": 23125
    },
    {
      "epoch": 0.3600924751103396,
      "grad_norm": 3.1863162517547607,
      "learning_rate": 6.736041690565543e-07,
      "loss": 0.7478,
      "step": 23130
    },
    {
      "epoch": 0.360170316112309,
      "grad_norm": 3.62100887298584,
      "learning_rate": 6.735222300519493e-07,
      "loss": 0.7283,
      "step": 23135
    },
    {
      "epoch": 0.3602481571142784,
      "grad_norm": 3.6080639362335205,
      "learning_rate": 6.734402910473443e-07,
      "loss": 0.7048,
      "step": 23140
    },
    {
      "epoch": 0.36032599811624777,
      "grad_norm": 4.390148639678955,
      "learning_rate": 6.733583520427394e-07,
      "loss": 0.7789,
      "step": 23145
    },
    {
      "epoch": 0.3604038391182171,
      "grad_norm": 3.059814214706421,
      "learning_rate": 6.732764130381344e-07,
      "loss": 0.7741,
      "step": 23150
    },
    {
      "epoch": 0.3604816801201865,
      "grad_norm": 8.699779510498047,
      "learning_rate": 6.731944740335293e-07,
      "loss": 0.8658,
      "step": 23155
    },
    {
      "epoch": 0.3605595211221559,
      "grad_norm": 2.8748719692230225,
      "learning_rate": 6.731125350289244e-07,
      "loss": 0.7413,
      "step": 23160
    },
    {
      "epoch": 0.3606373621241253,
      "grad_norm": 10.56540584564209,
      "learning_rate": 6.730305960243195e-07,
      "loss": 0.7527,
      "step": 23165
    },
    {
      "epoch": 0.3607152031260946,
      "grad_norm": 4.239411354064941,
      "learning_rate": 6.729486570197145e-07,
      "loss": 0.7167,
      "step": 23170
    },
    {
      "epoch": 0.360793044128064,
      "grad_norm": 4.87721061706543,
      "learning_rate": 6.728667180151095e-07,
      "loss": 0.6902,
      "step": 23175
    },
    {
      "epoch": 0.3608708851300334,
      "grad_norm": 4.051860809326172,
      "learning_rate": 6.727847790105046e-07,
      "loss": 0.7506,
      "step": 23180
    },
    {
      "epoch": 0.3609487261320028,
      "grad_norm": 6.635641574859619,
      "learning_rate": 6.727028400058997e-07,
      "loss": 0.7678,
      "step": 23185
    },
    {
      "epoch": 0.36102656713397213,
      "grad_norm": 4.147698879241943,
      "learning_rate": 6.726209010012946e-07,
      "loss": 0.7075,
      "step": 23190
    },
    {
      "epoch": 0.3611044081359415,
      "grad_norm": 3.2259066104888916,
      "learning_rate": 6.725389619966896e-07,
      "loss": 0.6925,
      "step": 23195
    },
    {
      "epoch": 0.3611822491379109,
      "grad_norm": 4.023670673370361,
      "learning_rate": 6.724570229920847e-07,
      "loss": 0.8152,
      "step": 23200
    },
    {
      "epoch": 0.3612600901398803,
      "grad_norm": 4.489516258239746,
      "learning_rate": 6.723750839874796e-07,
      "loss": 0.8063,
      "step": 23205
    },
    {
      "epoch": 0.36133793114184964,
      "grad_norm": 14.589898109436035,
      "learning_rate": 6.722931449828747e-07,
      "loss": 0.7012,
      "step": 23210
    },
    {
      "epoch": 0.36141577214381904,
      "grad_norm": 7.213819980621338,
      "learning_rate": 6.722112059782698e-07,
      "loss": 0.7749,
      "step": 23215
    },
    {
      "epoch": 0.36149361314578843,
      "grad_norm": 3.634456157684326,
      "learning_rate": 6.721292669736648e-07,
      "loss": 0.9399,
      "step": 23220
    },
    {
      "epoch": 0.36157145414775776,
      "grad_norm": 5.684159278869629,
      "learning_rate": 6.720473279690598e-07,
      "loss": 0.8365,
      "step": 23225
    },
    {
      "epoch": 0.36164929514972716,
      "grad_norm": 5.689432144165039,
      "learning_rate": 6.719653889644549e-07,
      "loss": 0.6303,
      "step": 23230
    },
    {
      "epoch": 0.36172713615169655,
      "grad_norm": 4.839646816253662,
      "learning_rate": 6.718834499598499e-07,
      "loss": 0.8117,
      "step": 23235
    },
    {
      "epoch": 0.36180497715366594,
      "grad_norm": 3.3537235260009766,
      "learning_rate": 6.718015109552448e-07,
      "loss": 0.7472,
      "step": 23240
    },
    {
      "epoch": 0.3618828181556353,
      "grad_norm": 3.9428837299346924,
      "learning_rate": 6.717195719506399e-07,
      "loss": 0.8015,
      "step": 23245
    },
    {
      "epoch": 0.36196065915760467,
      "grad_norm": 4.171321868896484,
      "learning_rate": 6.716376329460349e-07,
      "loss": 0.6913,
      "step": 23250
    },
    {
      "epoch": 0.36203850015957406,
      "grad_norm": 3.294297933578491,
      "learning_rate": 6.7155569394143e-07,
      "loss": 0.7597,
      "step": 23255
    },
    {
      "epoch": 0.36211634116154345,
      "grad_norm": 3.333055257797241,
      "learning_rate": 6.71473754936825e-07,
      "loss": 0.6821,
      "step": 23260
    },
    {
      "epoch": 0.3621941821635128,
      "grad_norm": 3.4170455932617188,
      "learning_rate": 6.7139181593222e-07,
      "loss": 0.7262,
      "step": 23265
    },
    {
      "epoch": 0.3622720231654822,
      "grad_norm": 10.967249870300293,
      "learning_rate": 6.713098769276151e-07,
      "loss": 0.6685,
      "step": 23270
    },
    {
      "epoch": 0.3623498641674516,
      "grad_norm": 4.1759467124938965,
      "learning_rate": 6.712279379230102e-07,
      "loss": 0.712,
      "step": 23275
    },
    {
      "epoch": 0.36242770516942097,
      "grad_norm": 3.056581735610962,
      "learning_rate": 6.71145998918405e-07,
      "loss": 0.7911,
      "step": 23280
    },
    {
      "epoch": 0.3625055461713903,
      "grad_norm": 4.291056156158447,
      "learning_rate": 6.710640599138001e-07,
      "loss": 0.7522,
      "step": 23285
    },
    {
      "epoch": 0.3625833871733597,
      "grad_norm": 4.416627407073975,
      "learning_rate": 6.709821209091952e-07,
      "loss": 0.739,
      "step": 23290
    },
    {
      "epoch": 0.3626612281753291,
      "grad_norm": 3.039506435394287,
      "learning_rate": 6.709001819045902e-07,
      "loss": 0.6868,
      "step": 23295
    },
    {
      "epoch": 0.3627390691772984,
      "grad_norm": 3.890329360961914,
      "learning_rate": 6.708182428999852e-07,
      "loss": 0.6522,
      "step": 23300
    },
    {
      "epoch": 0.3628169101792678,
      "grad_norm": 3.143517017364502,
      "learning_rate": 6.707363038953803e-07,
      "loss": 0.7248,
      "step": 23305
    },
    {
      "epoch": 0.3628947511812372,
      "grad_norm": 2.5914478302001953,
      "learning_rate": 6.706543648907754e-07,
      "loss": 0.7247,
      "step": 23310
    },
    {
      "epoch": 0.3629725921832066,
      "grad_norm": 3.5781683921813965,
      "learning_rate": 6.705724258861703e-07,
      "loss": 0.6967,
      "step": 23315
    },
    {
      "epoch": 0.36305043318517594,
      "grad_norm": 7.8637566566467285,
      "learning_rate": 6.704904868815653e-07,
      "loss": 0.5988,
      "step": 23320
    },
    {
      "epoch": 0.36312827418714533,
      "grad_norm": 14.063533782958984,
      "learning_rate": 6.704085478769604e-07,
      "loss": 0.7561,
      "step": 23325
    },
    {
      "epoch": 0.3632061151891147,
      "grad_norm": 4.953052043914795,
      "learning_rate": 6.703266088723553e-07,
      "loss": 0.8064,
      "step": 23330
    },
    {
      "epoch": 0.3632839561910841,
      "grad_norm": 4.630489826202393,
      "learning_rate": 6.702446698677504e-07,
      "loss": 0.8702,
      "step": 23335
    },
    {
      "epoch": 0.36336179719305345,
      "grad_norm": 3.3488337993621826,
      "learning_rate": 6.701627308631455e-07,
      "loss": 0.7062,
      "step": 23340
    },
    {
      "epoch": 0.36343963819502284,
      "grad_norm": 4.161873817443848,
      "learning_rate": 6.700807918585405e-07,
      "loss": 0.8029,
      "step": 23345
    },
    {
      "epoch": 0.36351747919699223,
      "grad_norm": 6.1759467124938965,
      "learning_rate": 6.699988528539355e-07,
      "loss": 0.6983,
      "step": 23350
    },
    {
      "epoch": 0.3635953201989616,
      "grad_norm": 4.258143901824951,
      "learning_rate": 6.699169138493306e-07,
      "loss": 0.7348,
      "step": 23355
    },
    {
      "epoch": 0.36367316120093096,
      "grad_norm": 3.5058975219726562,
      "learning_rate": 6.698349748447255e-07,
      "loss": 0.7475,
      "step": 23360
    },
    {
      "epoch": 0.36375100220290035,
      "grad_norm": 5.003944396972656,
      "learning_rate": 6.697530358401205e-07,
      "loss": 0.7452,
      "step": 23365
    },
    {
      "epoch": 0.36382884320486975,
      "grad_norm": 3.4567155838012695,
      "learning_rate": 6.696710968355156e-07,
      "loss": 0.6901,
      "step": 23370
    },
    {
      "epoch": 0.36390668420683914,
      "grad_norm": 11.741866111755371,
      "learning_rate": 6.695891578309106e-07,
      "loss": 0.7837,
      "step": 23375
    },
    {
      "epoch": 0.3639845252088085,
      "grad_norm": 3.1956512928009033,
      "learning_rate": 6.695072188263057e-07,
      "loss": 0.7313,
      "step": 23380
    },
    {
      "epoch": 0.36406236621077787,
      "grad_norm": 3.3680739402770996,
      "learning_rate": 6.694252798217007e-07,
      "loss": 0.8233,
      "step": 23385
    },
    {
      "epoch": 0.36414020721274726,
      "grad_norm": 4.3190202713012695,
      "learning_rate": 6.693433408170957e-07,
      "loss": 0.7765,
      "step": 23390
    },
    {
      "epoch": 0.3642180482147166,
      "grad_norm": 3.9722366333007812,
      "learning_rate": 6.692614018124908e-07,
      "loss": 0.8161,
      "step": 23395
    },
    {
      "epoch": 0.364295889216686,
      "grad_norm": 7.148501873016357,
      "learning_rate": 6.691794628078858e-07,
      "loss": 0.8317,
      "step": 23400
    },
    {
      "epoch": 0.3643737302186554,
      "grad_norm": 3.410125970840454,
      "learning_rate": 6.690975238032807e-07,
      "loss": 0.8454,
      "step": 23405
    },
    {
      "epoch": 0.36445157122062477,
      "grad_norm": 4.123642921447754,
      "learning_rate": 6.690155847986758e-07,
      "loss": 0.6667,
      "step": 23410
    },
    {
      "epoch": 0.3645294122225941,
      "grad_norm": 2.6929523944854736,
      "learning_rate": 6.689336457940709e-07,
      "loss": 0.6559,
      "step": 23415
    },
    {
      "epoch": 0.3646072532245635,
      "grad_norm": 6.187636375427246,
      "learning_rate": 6.688517067894659e-07,
      "loss": 0.631,
      "step": 23420
    },
    {
      "epoch": 0.3646850942265329,
      "grad_norm": 3.486956834793091,
      "learning_rate": 6.687697677848609e-07,
      "loss": 0.7464,
      "step": 23425
    },
    {
      "epoch": 0.3647629352285023,
      "grad_norm": 3.228217601776123,
      "learning_rate": 6.68687828780256e-07,
      "loss": 0.692,
      "step": 23430
    },
    {
      "epoch": 0.3648407762304716,
      "grad_norm": 3.462691307067871,
      "learning_rate": 6.686058897756511e-07,
      "loss": 0.661,
      "step": 23435
    },
    {
      "epoch": 0.364918617232441,
      "grad_norm": 3.6201229095458984,
      "learning_rate": 6.685239507710459e-07,
      "loss": 0.6984,
      "step": 23440
    },
    {
      "epoch": 0.3649964582344104,
      "grad_norm": 3.9280552864074707,
      "learning_rate": 6.68442011766441e-07,
      "loss": 0.7372,
      "step": 23445
    },
    {
      "epoch": 0.3650742992363798,
      "grad_norm": 3.9565505981445312,
      "learning_rate": 6.683600727618361e-07,
      "loss": 0.7406,
      "step": 23450
    },
    {
      "epoch": 0.36515214023834913,
      "grad_norm": 3.299711227416992,
      "learning_rate": 6.68278133757231e-07,
      "loss": 0.8143,
      "step": 23455
    },
    {
      "epoch": 0.3652299812403185,
      "grad_norm": 8.119706153869629,
      "learning_rate": 6.681961947526261e-07,
      "loss": 0.6385,
      "step": 23460
    },
    {
      "epoch": 0.3653078222422879,
      "grad_norm": 4.427123546600342,
      "learning_rate": 6.681142557480212e-07,
      "loss": 0.7992,
      "step": 23465
    },
    {
      "epoch": 0.36538566324425725,
      "grad_norm": 6.14912223815918,
      "learning_rate": 6.680323167434162e-07,
      "loss": 0.6641,
      "step": 23470
    },
    {
      "epoch": 0.36546350424622664,
      "grad_norm": 3.034275770187378,
      "learning_rate": 6.679503777388112e-07,
      "loss": 0.8102,
      "step": 23475
    },
    {
      "epoch": 0.36554134524819604,
      "grad_norm": 4.197099685668945,
      "learning_rate": 6.678684387342062e-07,
      "loss": 0.7425,
      "step": 23480
    },
    {
      "epoch": 0.36561918625016543,
      "grad_norm": 3.902829647064209,
      "learning_rate": 6.677864997296012e-07,
      "loss": 0.7829,
      "step": 23485
    },
    {
      "epoch": 0.36569702725213477,
      "grad_norm": 2.425935983657837,
      "learning_rate": 6.677045607249963e-07,
      "loss": 0.7347,
      "step": 23490
    },
    {
      "epoch": 0.36577486825410416,
      "grad_norm": 5.808422565460205,
      "learning_rate": 6.676226217203913e-07,
      "loss": 0.7398,
      "step": 23495
    },
    {
      "epoch": 0.36585270925607355,
      "grad_norm": 3.304903984069824,
      "learning_rate": 6.675406827157863e-07,
      "loss": 0.7069,
      "step": 23500
    },
    {
      "epoch": 0.36593055025804294,
      "grad_norm": 3.726069927215576,
      "learning_rate": 6.674587437111814e-07,
      "loss": 0.7961,
      "step": 23505
    },
    {
      "epoch": 0.3660083912600123,
      "grad_norm": 3.518789052963257,
      "learning_rate": 6.673768047065764e-07,
      "loss": 0.8082,
      "step": 23510
    },
    {
      "epoch": 0.36608623226198167,
      "grad_norm": 5.096744537353516,
      "learning_rate": 6.672948657019714e-07,
      "loss": 0.6841,
      "step": 23515
    },
    {
      "epoch": 0.36616407326395106,
      "grad_norm": 4.866468906402588,
      "learning_rate": 6.672129266973664e-07,
      "loss": 0.8747,
      "step": 23520
    },
    {
      "epoch": 0.36624191426592045,
      "grad_norm": 3.5222420692443848,
      "learning_rate": 6.671309876927615e-07,
      "loss": 0.7526,
      "step": 23525
    },
    {
      "epoch": 0.3663197552678898,
      "grad_norm": 4.000988960266113,
      "learning_rate": 6.670490486881564e-07,
      "loss": 0.7283,
      "step": 23530
    },
    {
      "epoch": 0.3663975962698592,
      "grad_norm": 2.9310030937194824,
      "learning_rate": 6.669671096835515e-07,
      "loss": 0.7335,
      "step": 23535
    },
    {
      "epoch": 0.3664754372718286,
      "grad_norm": 2.890139579772949,
      "learning_rate": 6.668851706789466e-07,
      "loss": 0.82,
      "step": 23540
    },
    {
      "epoch": 0.36655327827379797,
      "grad_norm": 4.196646690368652,
      "learning_rate": 6.668032316743417e-07,
      "loss": 0.6881,
      "step": 23545
    },
    {
      "epoch": 0.3666311192757673,
      "grad_norm": 3.2677407264709473,
      "learning_rate": 6.667212926697366e-07,
      "loss": 0.7409,
      "step": 23550
    },
    {
      "epoch": 0.3667089602777367,
      "grad_norm": 3.126652717590332,
      "learning_rate": 6.666393536651317e-07,
      "loss": 0.7847,
      "step": 23555
    },
    {
      "epoch": 0.3667868012797061,
      "grad_norm": 3.171036720275879,
      "learning_rate": 6.665574146605268e-07,
      "loss": 0.7718,
      "step": 23560
    },
    {
      "epoch": 0.3668646422816754,
      "grad_norm": 6.803696155548096,
      "learning_rate": 6.664754756559216e-07,
      "loss": 0.8001,
      "step": 23565
    },
    {
      "epoch": 0.3669424832836448,
      "grad_norm": 2.7900009155273438,
      "learning_rate": 6.663935366513167e-07,
      "loss": 0.7105,
      "step": 23570
    },
    {
      "epoch": 0.3670203242856142,
      "grad_norm": 7.508723258972168,
      "learning_rate": 6.663115976467118e-07,
      "loss": 0.7273,
      "step": 23575
    },
    {
      "epoch": 0.3670981652875836,
      "grad_norm": 6.162428379058838,
      "learning_rate": 6.662296586421068e-07,
      "loss": 0.7806,
      "step": 23580
    },
    {
      "epoch": 0.36717600628955294,
      "grad_norm": 3.379800319671631,
      "learning_rate": 6.661477196375018e-07,
      "loss": 0.6322,
      "step": 23585
    },
    {
      "epoch": 0.36725384729152233,
      "grad_norm": 4.739343643188477,
      "learning_rate": 6.660657806328969e-07,
      "loss": 0.6487,
      "step": 23590
    },
    {
      "epoch": 0.3673316882934917,
      "grad_norm": 3.727065086364746,
      "learning_rate": 6.659838416282919e-07,
      "loss": 0.7775,
      "step": 23595
    },
    {
      "epoch": 0.3674095292954611,
      "grad_norm": 12.969565391540527,
      "learning_rate": 6.659019026236869e-07,
      "loss": 0.8681,
      "step": 23600
    },
    {
      "epoch": 0.36748737029743045,
      "grad_norm": 3.2617616653442383,
      "learning_rate": 6.658199636190819e-07,
      "loss": 0.7413,
      "step": 23605
    },
    {
      "epoch": 0.36756521129939984,
      "grad_norm": 3.86673641204834,
      "learning_rate": 6.657380246144769e-07,
      "loss": 0.7602,
      "step": 23610
    },
    {
      "epoch": 0.36764305230136923,
      "grad_norm": 3.8551628589630127,
      "learning_rate": 6.65656085609872e-07,
      "loss": 0.7935,
      "step": 23615
    },
    {
      "epoch": 0.3677208933033386,
      "grad_norm": 5.538820266723633,
      "learning_rate": 6.65574146605267e-07,
      "loss": 0.6844,
      "step": 23620
    },
    {
      "epoch": 0.36779873430530796,
      "grad_norm": 3.6687228679656982,
      "learning_rate": 6.65492207600662e-07,
      "loss": 0.667,
      "step": 23625
    },
    {
      "epoch": 0.36787657530727735,
      "grad_norm": 3.558749198913574,
      "learning_rate": 6.654102685960571e-07,
      "loss": 0.7229,
      "step": 23630
    },
    {
      "epoch": 0.36795441630924675,
      "grad_norm": 4.3059234619140625,
      "learning_rate": 6.653283295914522e-07,
      "loss": 0.8282,
      "step": 23635
    },
    {
      "epoch": 0.3680322573112161,
      "grad_norm": 6.534751892089844,
      "learning_rate": 6.652463905868471e-07,
      "loss": 0.6982,
      "step": 23640
    },
    {
      "epoch": 0.3681100983131855,
      "grad_norm": 2.8018741607666016,
      "learning_rate": 6.651644515822421e-07,
      "loss": 0.7222,
      "step": 23645
    },
    {
      "epoch": 0.36818793931515487,
      "grad_norm": 6.206122398376465,
      "learning_rate": 6.650825125776372e-07,
      "loss": 0.7321,
      "step": 23650
    },
    {
      "epoch": 0.36826578031712426,
      "grad_norm": 2.767120122909546,
      "learning_rate": 6.650005735730321e-07,
      "loss": 0.6645,
      "step": 23655
    },
    {
      "epoch": 0.3683436213190936,
      "grad_norm": 3.2537684440612793,
      "learning_rate": 6.649186345684272e-07,
      "loss": 0.7628,
      "step": 23660
    },
    {
      "epoch": 0.368421462321063,
      "grad_norm": 5.646450996398926,
      "learning_rate": 6.648366955638223e-07,
      "loss": 0.7477,
      "step": 23665
    },
    {
      "epoch": 0.3684993033230324,
      "grad_norm": 3.675793409347534,
      "learning_rate": 6.647547565592174e-07,
      "loss": 0.6524,
      "step": 23670
    },
    {
      "epoch": 0.36857714432500177,
      "grad_norm": 3.5903780460357666,
      "learning_rate": 6.646728175546123e-07,
      "loss": 0.7101,
      "step": 23675
    },
    {
      "epoch": 0.3686549853269711,
      "grad_norm": 6.434808254241943,
      "learning_rate": 6.645908785500074e-07,
      "loss": 0.7431,
      "step": 23680
    },
    {
      "epoch": 0.3687328263289405,
      "grad_norm": 3.1390936374664307,
      "learning_rate": 6.645089395454024e-07,
      "loss": 0.7688,
      "step": 23685
    },
    {
      "epoch": 0.3688106673309099,
      "grad_norm": 3.689784526824951,
      "learning_rate": 6.644270005407973e-07,
      "loss": 0.7055,
      "step": 23690
    },
    {
      "epoch": 0.3688885083328793,
      "grad_norm": 3.707280158996582,
      "learning_rate": 6.643450615361924e-07,
      "loss": 0.7503,
      "step": 23695
    },
    {
      "epoch": 0.3689663493348486,
      "grad_norm": 4.782886981964111,
      "learning_rate": 6.642631225315875e-07,
      "loss": 0.7002,
      "step": 23700
    },
    {
      "epoch": 0.369044190336818,
      "grad_norm": 3.4875693321228027,
      "learning_rate": 6.641811835269825e-07,
      "loss": 0.8609,
      "step": 23705
    },
    {
      "epoch": 0.3691220313387874,
      "grad_norm": 4.89082670211792,
      "learning_rate": 6.640992445223775e-07,
      "loss": 0.7487,
      "step": 23710
    },
    {
      "epoch": 0.3691998723407568,
      "grad_norm": 4.395251750946045,
      "learning_rate": 6.640173055177726e-07,
      "loss": 0.7893,
      "step": 23715
    },
    {
      "epoch": 0.36927771334272613,
      "grad_norm": 3.1876797676086426,
      "learning_rate": 6.639353665131676e-07,
      "loss": 0.7706,
      "step": 23720
    },
    {
      "epoch": 0.3693555543446955,
      "grad_norm": 4.417449951171875,
      "learning_rate": 6.638534275085625e-07,
      "loss": 0.7106,
      "step": 23725
    },
    {
      "epoch": 0.3694333953466649,
      "grad_norm": 3.5318500995635986,
      "learning_rate": 6.637714885039576e-07,
      "loss": 0.6906,
      "step": 23730
    },
    {
      "epoch": 0.36951123634863425,
      "grad_norm": 6.2017822265625,
      "learning_rate": 6.636895494993526e-07,
      "loss": 0.8055,
      "step": 23735
    },
    {
      "epoch": 0.36958907735060365,
      "grad_norm": 3.228733539581299,
      "learning_rate": 6.636076104947477e-07,
      "loss": 0.8764,
      "step": 23740
    },
    {
      "epoch": 0.36966691835257304,
      "grad_norm": 8.708043098449707,
      "learning_rate": 6.635256714901427e-07,
      "loss": 0.8125,
      "step": 23745
    },
    {
      "epoch": 0.36974475935454243,
      "grad_norm": 3.2916955947875977,
      "learning_rate": 6.634437324855377e-07,
      "loss": 0.9001,
      "step": 23750
    },
    {
      "epoch": 0.36982260035651177,
      "grad_norm": 4.881289958953857,
      "learning_rate": 6.633617934809328e-07,
      "loss": 0.6748,
      "step": 23755
    },
    {
      "epoch": 0.36990044135848116,
      "grad_norm": 8.595666885375977,
      "learning_rate": 6.632798544763279e-07,
      "loss": 0.7027,
      "step": 23760
    },
    {
      "epoch": 0.36997828236045055,
      "grad_norm": 3.132411479949951,
      "learning_rate": 6.631979154717227e-07,
      "loss": 0.8831,
      "step": 23765
    },
    {
      "epoch": 0.37005612336241994,
      "grad_norm": 3.2658257484436035,
      "learning_rate": 6.631159764671178e-07,
      "loss": 0.8026,
      "step": 23770
    },
    {
      "epoch": 0.3701339643643893,
      "grad_norm": 6.149971961975098,
      "learning_rate": 6.630340374625129e-07,
      "loss": 0.7649,
      "step": 23775
    },
    {
      "epoch": 0.37021180536635867,
      "grad_norm": 4.137527942657471,
      "learning_rate": 6.629520984579078e-07,
      "loss": 0.8157,
      "step": 23780
    },
    {
      "epoch": 0.37028964636832806,
      "grad_norm": 2.82962703704834,
      "learning_rate": 6.628701594533029e-07,
      "loss": 0.8197,
      "step": 23785
    },
    {
      "epoch": 0.37036748737029745,
      "grad_norm": 6.3977370262146,
      "learning_rate": 6.62788220448698e-07,
      "loss": 0.6795,
      "step": 23790
    },
    {
      "epoch": 0.3704453283722668,
      "grad_norm": 6.537234306335449,
      "learning_rate": 6.627062814440931e-07,
      "loss": 0.6928,
      "step": 23795
    },
    {
      "epoch": 0.3705231693742362,
      "grad_norm": 3.5378894805908203,
      "learning_rate": 6.62624342439488e-07,
      "loss": 0.8536,
      "step": 23800
    },
    {
      "epoch": 0.3706010103762056,
      "grad_norm": 8.844374656677246,
      "learning_rate": 6.62542403434883e-07,
      "loss": 0.8496,
      "step": 23805
    },
    {
      "epoch": 0.37067885137817497,
      "grad_norm": 5.335997104644775,
      "learning_rate": 6.624604644302781e-07,
      "loss": 0.8001,
      "step": 23810
    },
    {
      "epoch": 0.3707566923801443,
      "grad_norm": 3.984768867492676,
      "learning_rate": 6.62378525425673e-07,
      "loss": 0.7219,
      "step": 23815
    },
    {
      "epoch": 0.3708345333821137,
      "grad_norm": 2.9257357120513916,
      "learning_rate": 6.622965864210681e-07,
      "loss": 0.8097,
      "step": 23820
    },
    {
      "epoch": 0.3709123743840831,
      "grad_norm": 3.574244737625122,
      "learning_rate": 6.622146474164632e-07,
      "loss": 0.8586,
      "step": 23825
    },
    {
      "epoch": 0.3709902153860524,
      "grad_norm": 2.2178125381469727,
      "learning_rate": 6.621327084118582e-07,
      "loss": 0.7355,
      "step": 23830
    },
    {
      "epoch": 0.3710680563880218,
      "grad_norm": 3.1225383281707764,
      "learning_rate": 6.620507694072532e-07,
      "loss": 0.665,
      "step": 23835
    },
    {
      "epoch": 0.3711458973899912,
      "grad_norm": 3.577746868133545,
      "learning_rate": 6.619688304026483e-07,
      "loss": 0.779,
      "step": 23840
    },
    {
      "epoch": 0.3712237383919606,
      "grad_norm": 6.259574890136719,
      "learning_rate": 6.618868913980432e-07,
      "loss": 0.7787,
      "step": 23845
    },
    {
      "epoch": 0.37130157939392994,
      "grad_norm": 3.1449718475341797,
      "learning_rate": 6.618049523934383e-07,
      "loss": 0.7209,
      "step": 23850
    },
    {
      "epoch": 0.37137942039589933,
      "grad_norm": 3.897055149078369,
      "learning_rate": 6.617230133888333e-07,
      "loss": 0.6962,
      "step": 23855
    },
    {
      "epoch": 0.3714572613978687,
      "grad_norm": 3.856555938720703,
      "learning_rate": 6.616410743842283e-07,
      "loss": 0.7708,
      "step": 23860
    },
    {
      "epoch": 0.3715351023998381,
      "grad_norm": 5.206638813018799,
      "learning_rate": 6.615591353796234e-07,
      "loss": 0.7102,
      "step": 23865
    },
    {
      "epoch": 0.37161294340180745,
      "grad_norm": 4.0182414054870605,
      "learning_rate": 6.614771963750184e-07,
      "loss": 0.7327,
      "step": 23870
    },
    {
      "epoch": 0.37169078440377684,
      "grad_norm": 3.4601380825042725,
      "learning_rate": 6.613952573704134e-07,
      "loss": 0.7188,
      "step": 23875
    },
    {
      "epoch": 0.37176862540574623,
      "grad_norm": 4.397197723388672,
      "learning_rate": 6.613133183658085e-07,
      "loss": 0.824,
      "step": 23880
    },
    {
      "epoch": 0.3718464664077156,
      "grad_norm": 3.7970356941223145,
      "learning_rate": 6.612313793612036e-07,
      "loss": 0.7597,
      "step": 23885
    },
    {
      "epoch": 0.37192430740968496,
      "grad_norm": 4.713395118713379,
      "learning_rate": 6.611494403565984e-07,
      "loss": 0.6744,
      "step": 23890
    },
    {
      "epoch": 0.37200214841165435,
      "grad_norm": 10.343477249145508,
      "learning_rate": 6.610675013519935e-07,
      "loss": 0.7873,
      "step": 23895
    },
    {
      "epoch": 0.37207998941362375,
      "grad_norm": 5.199718952178955,
      "learning_rate": 6.609855623473886e-07,
      "loss": 0.6932,
      "step": 23900
    },
    {
      "epoch": 0.3721578304155931,
      "grad_norm": 8.903950691223145,
      "learning_rate": 6.609036233427835e-07,
      "loss": 0.7658,
      "step": 23905
    },
    {
      "epoch": 0.3722356714175625,
      "grad_norm": 4.624990940093994,
      "learning_rate": 6.608216843381786e-07,
      "loss": 0.7684,
      "step": 23910
    },
    {
      "epoch": 0.37231351241953187,
      "grad_norm": 5.178587913513184,
      "learning_rate": 6.607397453335737e-07,
      "loss": 0.767,
      "step": 23915
    },
    {
      "epoch": 0.37239135342150126,
      "grad_norm": 4.226781368255615,
      "learning_rate": 6.606578063289688e-07,
      "loss": 0.728,
      "step": 23920
    },
    {
      "epoch": 0.3724691944234706,
      "grad_norm": 6.461855411529541,
      "learning_rate": 6.605758673243637e-07,
      "loss": 0.7692,
      "step": 23925
    },
    {
      "epoch": 0.37254703542544,
      "grad_norm": 3.171079397201538,
      "learning_rate": 6.604939283197587e-07,
      "loss": 0.6856,
      "step": 23930
    },
    {
      "epoch": 0.3726248764274094,
      "grad_norm": 4.6631693840026855,
      "learning_rate": 6.604119893151538e-07,
      "loss": 0.8407,
      "step": 23935
    },
    {
      "epoch": 0.37270271742937877,
      "grad_norm": 6.995047569274902,
      "learning_rate": 6.603300503105488e-07,
      "loss": 0.7624,
      "step": 23940
    },
    {
      "epoch": 0.3727805584313481,
      "grad_norm": 3.8059585094451904,
      "learning_rate": 6.602481113059438e-07,
      "loss": 0.7307,
      "step": 23945
    },
    {
      "epoch": 0.3728583994333175,
      "grad_norm": 5.160345554351807,
      "learning_rate": 6.601661723013389e-07,
      "loss": 0.7879,
      "step": 23950
    },
    {
      "epoch": 0.3729362404352869,
      "grad_norm": 7.343621730804443,
      "learning_rate": 6.600842332967339e-07,
      "loss": 0.7122,
      "step": 23955
    },
    {
      "epoch": 0.3730140814372563,
      "grad_norm": 5.80457067489624,
      "learning_rate": 6.60002294292129e-07,
      "loss": 0.8733,
      "step": 23960
    },
    {
      "epoch": 0.3730919224392256,
      "grad_norm": 4.190765857696533,
      "learning_rate": 6.59920355287524e-07,
      "loss": 0.7712,
      "step": 23965
    },
    {
      "epoch": 0.373169763441195,
      "grad_norm": 2.601067304611206,
      "learning_rate": 6.598384162829189e-07,
      "loss": 0.632,
      "step": 23970
    },
    {
      "epoch": 0.3732476044431644,
      "grad_norm": 4.3673834800720215,
      "learning_rate": 6.59756477278314e-07,
      "loss": 0.8203,
      "step": 23975
    },
    {
      "epoch": 0.3733254454451338,
      "grad_norm": 2.9783923625946045,
      "learning_rate": 6.59674538273709e-07,
      "loss": 0.7736,
      "step": 23980
    },
    {
      "epoch": 0.37340328644710313,
      "grad_norm": 6.698786735534668,
      "learning_rate": 6.59592599269104e-07,
      "loss": 0.8546,
      "step": 23985
    },
    {
      "epoch": 0.3734811274490725,
      "grad_norm": 3.005664348602295,
      "learning_rate": 6.595106602644991e-07,
      "loss": 0.7989,
      "step": 23990
    },
    {
      "epoch": 0.3735589684510419,
      "grad_norm": 3.8986668586730957,
      "learning_rate": 6.594287212598942e-07,
      "loss": 0.6315,
      "step": 23995
    },
    {
      "epoch": 0.37363680945301125,
      "grad_norm": 4.2483062744140625,
      "learning_rate": 6.593467822552891e-07,
      "loss": 0.8664,
      "step": 24000
    },
    {
      "epoch": 0.37371465045498065,
      "grad_norm": 2.8898262977600098,
      "learning_rate": 6.592648432506842e-07,
      "loss": 0.6167,
      "step": 24005
    },
    {
      "epoch": 0.37379249145695004,
      "grad_norm": 2.7583794593811035,
      "learning_rate": 6.591829042460792e-07,
      "loss": 0.8595,
      "step": 24010
    },
    {
      "epoch": 0.37387033245891943,
      "grad_norm": 3.2141683101654053,
      "learning_rate": 6.591009652414741e-07,
      "loss": 0.8204,
      "step": 24015
    },
    {
      "epoch": 0.37394817346088877,
      "grad_norm": 4.640387535095215,
      "learning_rate": 6.590190262368692e-07,
      "loss": 0.8476,
      "step": 24020
    },
    {
      "epoch": 0.37402601446285816,
      "grad_norm": 8.061158180236816,
      "learning_rate": 6.589370872322643e-07,
      "loss": 0.6599,
      "step": 24025
    },
    {
      "epoch": 0.37410385546482755,
      "grad_norm": 4.347311973571777,
      "learning_rate": 6.588551482276593e-07,
      "loss": 0.7904,
      "step": 24030
    },
    {
      "epoch": 0.37418169646679694,
      "grad_norm": 4.115234375,
      "learning_rate": 6.587732092230543e-07,
      "loss": 0.8568,
      "step": 24035
    },
    {
      "epoch": 0.3742595374687663,
      "grad_norm": 5.665265083312988,
      "learning_rate": 6.586912702184494e-07,
      "loss": 0.795,
      "step": 24040
    },
    {
      "epoch": 0.37433737847073567,
      "grad_norm": 6.1077423095703125,
      "learning_rate": 6.586093312138445e-07,
      "loss": 0.7835,
      "step": 24045
    },
    {
      "epoch": 0.37441521947270506,
      "grad_norm": 6.675379276275635,
      "learning_rate": 6.585273922092393e-07,
      "loss": 0.7311,
      "step": 24050
    },
    {
      "epoch": 0.37449306047467446,
      "grad_norm": 3.762505054473877,
      "learning_rate": 6.584454532046344e-07,
      "loss": 0.6828,
      "step": 24055
    },
    {
      "epoch": 0.3745709014766438,
      "grad_norm": 7.7305684089660645,
      "learning_rate": 6.583635142000295e-07,
      "loss": 0.729,
      "step": 24060
    },
    {
      "epoch": 0.3746487424786132,
      "grad_norm": 4.176987171173096,
      "learning_rate": 6.582815751954245e-07,
      "loss": 0.7009,
      "step": 24065
    },
    {
      "epoch": 0.3747265834805826,
      "grad_norm": 3.813732385635376,
      "learning_rate": 6.581996361908195e-07,
      "loss": 0.8001,
      "step": 24070
    },
    {
      "epoch": 0.3748044244825519,
      "grad_norm": 5.6425604820251465,
      "learning_rate": 6.581176971862146e-07,
      "loss": 0.7634,
      "step": 24075
    },
    {
      "epoch": 0.3748822654845213,
      "grad_norm": 4.143874168395996,
      "learning_rate": 6.580357581816096e-07,
      "loss": 0.7091,
      "step": 24080
    },
    {
      "epoch": 0.3749601064864907,
      "grad_norm": 2.3728559017181396,
      "learning_rate": 6.579538191770047e-07,
      "loss": 0.7337,
      "step": 24085
    },
    {
      "epoch": 0.3750379474884601,
      "grad_norm": 3.5923044681549072,
      "learning_rate": 6.578718801723996e-07,
      "loss": 0.6422,
      "step": 24090
    },
    {
      "epoch": 0.3751157884904294,
      "grad_norm": 3.443315029144287,
      "learning_rate": 6.577899411677946e-07,
      "loss": 0.6181,
      "step": 24095
    },
    {
      "epoch": 0.3751936294923988,
      "grad_norm": 5.803593635559082,
      "learning_rate": 6.577080021631897e-07,
      "loss": 0.6687,
      "step": 24100
    },
    {
      "epoch": 0.3752714704943682,
      "grad_norm": 3.978525400161743,
      "learning_rate": 6.576260631585847e-07,
      "loss": 0.6865,
      "step": 24105
    },
    {
      "epoch": 0.3753493114963376,
      "grad_norm": 3.4662880897521973,
      "learning_rate": 6.575441241539797e-07,
      "loss": 0.7028,
      "step": 24110
    },
    {
      "epoch": 0.37542715249830694,
      "grad_norm": 3.669604539871216,
      "learning_rate": 6.574621851493748e-07,
      "loss": 0.7207,
      "step": 24115
    },
    {
      "epoch": 0.37550499350027633,
      "grad_norm": 10.295849800109863,
      "learning_rate": 6.573802461447699e-07,
      "loss": 0.7618,
      "step": 24120
    },
    {
      "epoch": 0.3755828345022457,
      "grad_norm": 3.143021821975708,
      "learning_rate": 6.572983071401648e-07,
      "loss": 0.6281,
      "step": 24125
    },
    {
      "epoch": 0.3756606755042151,
      "grad_norm": 5.273399353027344,
      "learning_rate": 6.572163681355598e-07,
      "loss": 0.7412,
      "step": 24130
    },
    {
      "epoch": 0.37573851650618445,
      "grad_norm": 5.699292182922363,
      "learning_rate": 6.571344291309549e-07,
      "loss": 0.6383,
      "step": 24135
    },
    {
      "epoch": 0.37581635750815384,
      "grad_norm": 3.716749429702759,
      "learning_rate": 6.570524901263498e-07,
      "loss": 0.8457,
      "step": 24140
    },
    {
      "epoch": 0.37589419851012323,
      "grad_norm": 4.004367351531982,
      "learning_rate": 6.569705511217449e-07,
      "loss": 0.7126,
      "step": 24145
    },
    {
      "epoch": 0.3759720395120926,
      "grad_norm": 3.647512435913086,
      "learning_rate": 6.5688861211714e-07,
      "loss": 0.8438,
      "step": 24150
    },
    {
      "epoch": 0.37604988051406196,
      "grad_norm": 2.8271801471710205,
      "learning_rate": 6.56806673112535e-07,
      "loss": 0.7476,
      "step": 24155
    },
    {
      "epoch": 0.37612772151603135,
      "grad_norm": 5.783123016357422,
      "learning_rate": 6.5672473410793e-07,
      "loss": 0.8312,
      "step": 24160
    },
    {
      "epoch": 0.37620556251800075,
      "grad_norm": 3.1801908016204834,
      "learning_rate": 6.566427951033251e-07,
      "loss": 0.8276,
      "step": 24165
    },
    {
      "epoch": 0.3762834035199701,
      "grad_norm": 3.8137319087982178,
      "learning_rate": 6.565608560987201e-07,
      "loss": 0.6949,
      "step": 24170
    },
    {
      "epoch": 0.3763612445219395,
      "grad_norm": 6.042738437652588,
      "learning_rate": 6.56478917094115e-07,
      "loss": 0.799,
      "step": 24175
    },
    {
      "epoch": 0.37643908552390887,
      "grad_norm": 2.819098472595215,
      "learning_rate": 6.563969780895101e-07,
      "loss": 0.7652,
      "step": 24180
    },
    {
      "epoch": 0.37651692652587826,
      "grad_norm": 6.101853370666504,
      "learning_rate": 6.563150390849052e-07,
      "loss": 0.7183,
      "step": 24185
    },
    {
      "epoch": 0.3765947675278476,
      "grad_norm": 3.3164820671081543,
      "learning_rate": 6.562331000803002e-07,
      "loss": 0.787,
      "step": 24190
    },
    {
      "epoch": 0.376672608529817,
      "grad_norm": 3.9163126945495605,
      "learning_rate": 6.561511610756952e-07,
      "loss": 0.7803,
      "step": 24195
    },
    {
      "epoch": 0.3767504495317864,
      "grad_norm": 3.7802116870880127,
      "learning_rate": 6.560692220710903e-07,
      "loss": 0.718,
      "step": 24200
    },
    {
      "epoch": 0.37682829053375577,
      "grad_norm": 3.106133460998535,
      "learning_rate": 6.559872830664853e-07,
      "loss": 0.7347,
      "step": 24205
    },
    {
      "epoch": 0.3769061315357251,
      "grad_norm": 7.895877361297607,
      "learning_rate": 6.559053440618804e-07,
      "loss": 0.7739,
      "step": 24210
    },
    {
      "epoch": 0.3769839725376945,
      "grad_norm": 3.9047114849090576,
      "learning_rate": 6.558234050572753e-07,
      "loss": 0.8125,
      "step": 24215
    },
    {
      "epoch": 0.3770618135396639,
      "grad_norm": 3.2611491680145264,
      "learning_rate": 6.557414660526703e-07,
      "loss": 0.7705,
      "step": 24220
    },
    {
      "epoch": 0.3771396545416333,
      "grad_norm": 3.606032609939575,
      "learning_rate": 6.556595270480654e-07,
      "loss": 0.7115,
      "step": 24225
    },
    {
      "epoch": 0.3772174955436026,
      "grad_norm": 3.7341561317443848,
      "learning_rate": 6.555775880434604e-07,
      "loss": 0.7128,
      "step": 24230
    },
    {
      "epoch": 0.377295336545572,
      "grad_norm": 3.9536352157592773,
      "learning_rate": 6.554956490388554e-07,
      "loss": 0.7125,
      "step": 24235
    },
    {
      "epoch": 0.3773731775475414,
      "grad_norm": 3.7920241355895996,
      "learning_rate": 6.554137100342505e-07,
      "loss": 0.783,
      "step": 24240
    },
    {
      "epoch": 0.37745101854951074,
      "grad_norm": 4.251092433929443,
      "learning_rate": 6.553317710296456e-07,
      "loss": 0.7633,
      "step": 24245
    },
    {
      "epoch": 0.37752885955148013,
      "grad_norm": 3.8011977672576904,
      "learning_rate": 6.552498320250405e-07,
      "loss": 0.7667,
      "step": 24250
    },
    {
      "epoch": 0.3776067005534495,
      "grad_norm": 2.624079704284668,
      "learning_rate": 6.551678930204355e-07,
      "loss": 0.6509,
      "step": 24255
    },
    {
      "epoch": 0.3776845415554189,
      "grad_norm": 4.23248291015625,
      "learning_rate": 6.550859540158306e-07,
      "loss": 0.7741,
      "step": 24260
    },
    {
      "epoch": 0.37776238255738825,
      "grad_norm": 2.9144062995910645,
      "learning_rate": 6.550040150112256e-07,
      "loss": 0.8159,
      "step": 24265
    },
    {
      "epoch": 0.37784022355935765,
      "grad_norm": 2.4228739738464355,
      "learning_rate": 6.549220760066206e-07,
      "loss": 0.7285,
      "step": 24270
    },
    {
      "epoch": 0.37791806456132704,
      "grad_norm": 2.9209840297698975,
      "learning_rate": 6.548401370020157e-07,
      "loss": 0.6249,
      "step": 24275
    },
    {
      "epoch": 0.37799590556329643,
      "grad_norm": 3.9086556434631348,
      "learning_rate": 6.547581979974107e-07,
      "loss": 0.69,
      "step": 24280
    },
    {
      "epoch": 0.37807374656526577,
      "grad_norm": 3.2286627292633057,
      "learning_rate": 6.546762589928057e-07,
      "loss": 0.8162,
      "step": 24285
    },
    {
      "epoch": 0.37815158756723516,
      "grad_norm": 3.5103039741516113,
      "learning_rate": 6.545943199882008e-07,
      "loss": 0.6934,
      "step": 24290
    },
    {
      "epoch": 0.37822942856920455,
      "grad_norm": 4.203784465789795,
      "learning_rate": 6.545123809835958e-07,
      "loss": 0.7594,
      "step": 24295
    },
    {
      "epoch": 0.37830726957117394,
      "grad_norm": 3.7477505207061768,
      "learning_rate": 6.544304419789908e-07,
      "loss": 0.7079,
      "step": 24300
    },
    {
      "epoch": 0.3783851105731433,
      "grad_norm": 3.812622308731079,
      "learning_rate": 6.543485029743858e-07,
      "loss": 0.7349,
      "step": 24305
    },
    {
      "epoch": 0.37846295157511267,
      "grad_norm": 3.558681011199951,
      "learning_rate": 6.542665639697809e-07,
      "loss": 0.7898,
      "step": 24310
    },
    {
      "epoch": 0.37854079257708206,
      "grad_norm": 6.393659591674805,
      "learning_rate": 6.541846249651759e-07,
      "loss": 0.6253,
      "step": 24315
    },
    {
      "epoch": 0.37861863357905146,
      "grad_norm": 2.723544120788574,
      "learning_rate": 6.54102685960571e-07,
      "loss": 0.7685,
      "step": 24320
    },
    {
      "epoch": 0.3786964745810208,
      "grad_norm": 6.085930347442627,
      "learning_rate": 6.54020746955966e-07,
      "loss": 0.8936,
      "step": 24325
    },
    {
      "epoch": 0.3787743155829902,
      "grad_norm": 3.1237025260925293,
      "learning_rate": 6.53938807951361e-07,
      "loss": 0.679,
      "step": 24330
    },
    {
      "epoch": 0.3788521565849596,
      "grad_norm": 5.606638431549072,
      "learning_rate": 6.53856868946756e-07,
      "loss": 0.8498,
      "step": 24335
    },
    {
      "epoch": 0.3789299975869289,
      "grad_norm": 4.310702323913574,
      "learning_rate": 6.53774929942151e-07,
      "loss": 0.836,
      "step": 24340
    },
    {
      "epoch": 0.3790078385888983,
      "grad_norm": 4.304116725921631,
      "learning_rate": 6.53692990937546e-07,
      "loss": 0.8469,
      "step": 24345
    },
    {
      "epoch": 0.3790856795908677,
      "grad_norm": 3.8278419971466064,
      "learning_rate": 6.536110519329411e-07,
      "loss": 0.8109,
      "step": 24350
    },
    {
      "epoch": 0.3791635205928371,
      "grad_norm": 4.090198516845703,
      "learning_rate": 6.535291129283362e-07,
      "loss": 0.8282,
      "step": 24355
    },
    {
      "epoch": 0.3792413615948064,
      "grad_norm": 4.38947057723999,
      "learning_rate": 6.534471739237311e-07,
      "loss": 0.8153,
      "step": 24360
    },
    {
      "epoch": 0.3793192025967758,
      "grad_norm": 2.918142795562744,
      "learning_rate": 6.533652349191262e-07,
      "loss": 0.744,
      "step": 24365
    },
    {
      "epoch": 0.3793970435987452,
      "grad_norm": 3.9180519580841064,
      "learning_rate": 6.532832959145213e-07,
      "loss": 0.8869,
      "step": 24370
    },
    {
      "epoch": 0.3794748846007146,
      "grad_norm": 2.603149890899658,
      "learning_rate": 6.532013569099161e-07,
      "loss": 0.8022,
      "step": 24375
    },
    {
      "epoch": 0.37955272560268394,
      "grad_norm": 4.62812614440918,
      "learning_rate": 6.531194179053112e-07,
      "loss": 0.7867,
      "step": 24380
    },
    {
      "epoch": 0.37963056660465333,
      "grad_norm": 3.68599009513855,
      "learning_rate": 6.530374789007063e-07,
      "loss": 0.8338,
      "step": 24385
    },
    {
      "epoch": 0.3797084076066227,
      "grad_norm": 5.189111709594727,
      "learning_rate": 6.529555398961013e-07,
      "loss": 0.8457,
      "step": 24390
    },
    {
      "epoch": 0.3797862486085921,
      "grad_norm": 3.806553840637207,
      "learning_rate": 6.528736008914963e-07,
      "loss": 0.7669,
      "step": 24395
    },
    {
      "epoch": 0.37986408961056145,
      "grad_norm": 5.229100227355957,
      "learning_rate": 6.527916618868914e-07,
      "loss": 0.7026,
      "step": 24400
    },
    {
      "epoch": 0.37994193061253084,
      "grad_norm": 3.422163248062134,
      "learning_rate": 6.527097228822864e-07,
      "loss": 0.7962,
      "step": 24405
    },
    {
      "epoch": 0.38001977161450023,
      "grad_norm": 5.487661361694336,
      "learning_rate": 6.526277838776815e-07,
      "loss": 0.8384,
      "step": 24410
    },
    {
      "epoch": 0.3800976126164696,
      "grad_norm": 3.975858449935913,
      "learning_rate": 6.525458448730764e-07,
      "loss": 0.7887,
      "step": 24415
    },
    {
      "epoch": 0.38017545361843896,
      "grad_norm": 4.209732532501221,
      "learning_rate": 6.524639058684715e-07,
      "loss": 0.874,
      "step": 24420
    },
    {
      "epoch": 0.38025329462040836,
      "grad_norm": 3.549058198928833,
      "learning_rate": 6.523819668638665e-07,
      "loss": 0.6626,
      "step": 24425
    },
    {
      "epoch": 0.38033113562237775,
      "grad_norm": 3.021886110305786,
      "learning_rate": 6.523000278592615e-07,
      "loss": 0.7263,
      "step": 24430
    },
    {
      "epoch": 0.3804089766243471,
      "grad_norm": 2.7419252395629883,
      "learning_rate": 6.522180888546566e-07,
      "loss": 0.8412,
      "step": 24435
    },
    {
      "epoch": 0.3804868176263165,
      "grad_norm": 3.5946810245513916,
      "learning_rate": 6.521361498500516e-07,
      "loss": 0.7614,
      "step": 24440
    },
    {
      "epoch": 0.38056465862828587,
      "grad_norm": 4.112155914306641,
      "learning_rate": 6.520542108454467e-07,
      "loss": 0.6456,
      "step": 24445
    },
    {
      "epoch": 0.38064249963025526,
      "grad_norm": 2.8762810230255127,
      "learning_rate": 6.519722718408417e-07,
      "loss": 0.7133,
      "step": 24450
    },
    {
      "epoch": 0.3807203406322246,
      "grad_norm": 3.534217119216919,
      "learning_rate": 6.518903328362366e-07,
      "loss": 0.8656,
      "step": 24455
    },
    {
      "epoch": 0.380798181634194,
      "grad_norm": 3.1278042793273926,
      "learning_rate": 6.518083938316317e-07,
      "loss": 0.7489,
      "step": 24460
    },
    {
      "epoch": 0.3808760226361634,
      "grad_norm": 3.2034077644348145,
      "learning_rate": 6.517264548270267e-07,
      "loss": 0.8122,
      "step": 24465
    },
    {
      "epoch": 0.3809538636381328,
      "grad_norm": 4.331535339355469,
      "learning_rate": 6.516445158224217e-07,
      "loss": 0.6917,
      "step": 24470
    },
    {
      "epoch": 0.3810317046401021,
      "grad_norm": 3.5739593505859375,
      "learning_rate": 6.515625768178168e-07,
      "loss": 0.7332,
      "step": 24475
    },
    {
      "epoch": 0.3811095456420715,
      "grad_norm": 5.469711780548096,
      "learning_rate": 6.514806378132119e-07,
      "loss": 0.7239,
      "step": 24480
    },
    {
      "epoch": 0.3811873866440409,
      "grad_norm": 4.376361846923828,
      "learning_rate": 6.513986988086068e-07,
      "loss": 0.8151,
      "step": 24485
    },
    {
      "epoch": 0.3812652276460103,
      "grad_norm": 13.4918794631958,
      "learning_rate": 6.513167598040019e-07,
      "loss": 0.7069,
      "step": 24490
    },
    {
      "epoch": 0.3813430686479796,
      "grad_norm": 3.0612945556640625,
      "learning_rate": 6.512348207993969e-07,
      "loss": 0.7912,
      "step": 24495
    },
    {
      "epoch": 0.381420909649949,
      "grad_norm": 3.844869613647461,
      "learning_rate": 6.511528817947918e-07,
      "loss": 0.7319,
      "step": 24500
    },
    {
      "epoch": 0.3814987506519184,
      "grad_norm": 3.146636724472046,
      "learning_rate": 6.510709427901869e-07,
      "loss": 0.7234,
      "step": 24505
    },
    {
      "epoch": 0.38157659165388774,
      "grad_norm": 3.0788705348968506,
      "learning_rate": 6.50989003785582e-07,
      "loss": 0.9002,
      "step": 24510
    },
    {
      "epoch": 0.38165443265585713,
      "grad_norm": 5.646673679351807,
      "learning_rate": 6.50907064780977e-07,
      "loss": 0.7844,
      "step": 24515
    },
    {
      "epoch": 0.3817322736578265,
      "grad_norm": 2.48629093170166,
      "learning_rate": 6.50825125776372e-07,
      "loss": 0.7244,
      "step": 24520
    },
    {
      "epoch": 0.3818101146597959,
      "grad_norm": 4.52423620223999,
      "learning_rate": 6.507431867717671e-07,
      "loss": 0.7823,
      "step": 24525
    },
    {
      "epoch": 0.38188795566176525,
      "grad_norm": 3.8551182746887207,
      "learning_rate": 6.506612477671622e-07,
      "loss": 0.7537,
      "step": 24530
    },
    {
      "epoch": 0.38196579666373465,
      "grad_norm": 5.3387770652771,
      "learning_rate": 6.505793087625572e-07,
      "loss": 0.6455,
      "step": 24535
    },
    {
      "epoch": 0.38204363766570404,
      "grad_norm": 3.0548503398895264,
      "learning_rate": 6.504973697579521e-07,
      "loss": 0.7108,
      "step": 24540
    },
    {
      "epoch": 0.38212147866767343,
      "grad_norm": 4.870777130126953,
      "learning_rate": 6.504154307533472e-07,
      "loss": 0.7893,
      "step": 24545
    },
    {
      "epoch": 0.38219931966964277,
      "grad_norm": 3.0125315189361572,
      "learning_rate": 6.503334917487422e-07,
      "loss": 0.7098,
      "step": 24550
    },
    {
      "epoch": 0.38227716067161216,
      "grad_norm": 3.0246524810791016,
      "learning_rate": 6.502515527441372e-07,
      "loss": 0.7185,
      "step": 24555
    },
    {
      "epoch": 0.38235500167358155,
      "grad_norm": 2.8637657165527344,
      "learning_rate": 6.501696137395323e-07,
      "loss": 0.7185,
      "step": 24560
    },
    {
      "epoch": 0.38243284267555094,
      "grad_norm": 2.6888625621795654,
      "learning_rate": 6.500876747349273e-07,
      "loss": 0.8275,
      "step": 24565
    },
    {
      "epoch": 0.3825106836775203,
      "grad_norm": 3.1520016193389893,
      "learning_rate": 6.500057357303224e-07,
      "loss": 0.9101,
      "step": 24570
    },
    {
      "epoch": 0.38258852467948967,
      "grad_norm": 6.681828498840332,
      "learning_rate": 6.499237967257174e-07,
      "loss": 0.8141,
      "step": 24575
    },
    {
      "epoch": 0.38266636568145906,
      "grad_norm": 3.728142738342285,
      "learning_rate": 6.498418577211123e-07,
      "loss": 0.7488,
      "step": 24580
    },
    {
      "epoch": 0.38274420668342846,
      "grad_norm": 3.855109214782715,
      "learning_rate": 6.497599187165074e-07,
      "loss": 0.6411,
      "step": 24585
    },
    {
      "epoch": 0.3828220476853978,
      "grad_norm": 3.837766170501709,
      "learning_rate": 6.496779797119025e-07,
      "loss": 0.7441,
      "step": 24590
    },
    {
      "epoch": 0.3828998886873672,
      "grad_norm": 3.5934677124023438,
      "learning_rate": 6.495960407072974e-07,
      "loss": 0.6883,
      "step": 24595
    },
    {
      "epoch": 0.3829777296893366,
      "grad_norm": 4.515072345733643,
      "learning_rate": 6.495141017026925e-07,
      "loss": 0.7892,
      "step": 24600
    },
    {
      "epoch": 0.3830555706913059,
      "grad_norm": 2.74763560295105,
      "learning_rate": 6.494321626980876e-07,
      "loss": 0.7184,
      "step": 24605
    },
    {
      "epoch": 0.3831334116932753,
      "grad_norm": 5.450028896331787,
      "learning_rate": 6.493502236934825e-07,
      "loss": 0.9551,
      "step": 24610
    },
    {
      "epoch": 0.3832112526952447,
      "grad_norm": 3.213296413421631,
      "learning_rate": 6.492682846888776e-07,
      "loss": 0.8129,
      "step": 24615
    },
    {
      "epoch": 0.3832890936972141,
      "grad_norm": 4.963546276092529,
      "learning_rate": 6.491863456842726e-07,
      "loss": 0.6857,
      "step": 24620
    },
    {
      "epoch": 0.3833669346991834,
      "grad_norm": 3.2613205909729004,
      "learning_rate": 6.491044066796676e-07,
      "loss": 0.6915,
      "step": 24625
    },
    {
      "epoch": 0.3834447757011528,
      "grad_norm": 6.735360622406006,
      "learning_rate": 6.490224676750626e-07,
      "loss": 0.7666,
      "step": 24630
    },
    {
      "epoch": 0.3835226167031222,
      "grad_norm": 2.654822587966919,
      "learning_rate": 6.489405286704577e-07,
      "loss": 0.6861,
      "step": 24635
    },
    {
      "epoch": 0.3836004577050916,
      "grad_norm": 4.433813095092773,
      "learning_rate": 6.488585896658527e-07,
      "loss": 0.7513,
      "step": 24640
    },
    {
      "epoch": 0.38367829870706094,
      "grad_norm": 5.000208377838135,
      "learning_rate": 6.487766506612477e-07,
      "loss": 0.9181,
      "step": 24645
    },
    {
      "epoch": 0.38375613970903033,
      "grad_norm": 3.075925827026367,
      "learning_rate": 6.486947116566428e-07,
      "loss": 0.7587,
      "step": 24650
    },
    {
      "epoch": 0.3838339807109997,
      "grad_norm": 3.4058053493499756,
      "learning_rate": 6.486127726520379e-07,
      "loss": 0.8517,
      "step": 24655
    },
    {
      "epoch": 0.3839118217129691,
      "grad_norm": 3.323262929916382,
      "learning_rate": 6.485308336474328e-07,
      "loss": 0.7488,
      "step": 24660
    },
    {
      "epoch": 0.38398966271493845,
      "grad_norm": 3.4578330516815186,
      "learning_rate": 6.484488946428278e-07,
      "loss": 0.8329,
      "step": 24665
    },
    {
      "epoch": 0.38406750371690784,
      "grad_norm": 4.2953643798828125,
      "learning_rate": 6.483669556382229e-07,
      "loss": 0.8123,
      "step": 24670
    },
    {
      "epoch": 0.38414534471887724,
      "grad_norm": 3.698613166809082,
      "learning_rate": 6.482850166336179e-07,
      "loss": 0.7279,
      "step": 24675
    },
    {
      "epoch": 0.38422318572084657,
      "grad_norm": 6.446969985961914,
      "learning_rate": 6.48203077629013e-07,
      "loss": 0.741,
      "step": 24680
    },
    {
      "epoch": 0.38430102672281596,
      "grad_norm": 3.7218122482299805,
      "learning_rate": 6.48121138624408e-07,
      "loss": 0.6806,
      "step": 24685
    },
    {
      "epoch": 0.38437886772478536,
      "grad_norm": 3.6571853160858154,
      "learning_rate": 6.48039199619803e-07,
      "loss": 0.7325,
      "step": 24690
    },
    {
      "epoch": 0.38445670872675475,
      "grad_norm": 3.122586727142334,
      "learning_rate": 6.479572606151981e-07,
      "loss": 0.7157,
      "step": 24695
    },
    {
      "epoch": 0.3845345497287241,
      "grad_norm": 3.1738834381103516,
      "learning_rate": 6.47875321610593e-07,
      "loss": 0.7394,
      "step": 24700
    },
    {
      "epoch": 0.3846123907306935,
      "grad_norm": 3.8704867362976074,
      "learning_rate": 6.47793382605988e-07,
      "loss": 0.7922,
      "step": 24705
    },
    {
      "epoch": 0.38469023173266287,
      "grad_norm": 4.960922718048096,
      "learning_rate": 6.477114436013831e-07,
      "loss": 0.7774,
      "step": 24710
    },
    {
      "epoch": 0.38476807273463226,
      "grad_norm": 9.464720726013184,
      "learning_rate": 6.476295045967782e-07,
      "loss": 0.7547,
      "step": 24715
    },
    {
      "epoch": 0.3848459137366016,
      "grad_norm": 3.610100269317627,
      "learning_rate": 6.475475655921731e-07,
      "loss": 0.6677,
      "step": 24720
    },
    {
      "epoch": 0.384923754738571,
      "grad_norm": 3.670114278793335,
      "learning_rate": 6.474656265875682e-07,
      "loss": 0.7666,
      "step": 24725
    },
    {
      "epoch": 0.3850015957405404,
      "grad_norm": 4.544345855712891,
      "learning_rate": 6.473836875829633e-07,
      "loss": 0.6886,
      "step": 24730
    },
    {
      "epoch": 0.3850794367425098,
      "grad_norm": 2.7942495346069336,
      "learning_rate": 6.473017485783582e-07,
      "loss": 0.6796,
      "step": 24735
    },
    {
      "epoch": 0.3851572777444791,
      "grad_norm": 3.126971960067749,
      "learning_rate": 6.472198095737532e-07,
      "loss": 0.7763,
      "step": 24740
    },
    {
      "epoch": 0.3852351187464485,
      "grad_norm": 5.752474308013916,
      "learning_rate": 6.471378705691483e-07,
      "loss": 0.8314,
      "step": 24745
    },
    {
      "epoch": 0.3853129597484179,
      "grad_norm": 3.342573642730713,
      "learning_rate": 6.470559315645433e-07,
      "loss": 0.7669,
      "step": 24750
    },
    {
      "epoch": 0.3853908007503873,
      "grad_norm": 2.660649061203003,
      "learning_rate": 6.469739925599383e-07,
      "loss": 0.7613,
      "step": 24755
    },
    {
      "epoch": 0.3854686417523566,
      "grad_norm": 3.439711332321167,
      "learning_rate": 6.468920535553334e-07,
      "loss": 0.8655,
      "step": 24760
    },
    {
      "epoch": 0.385546482754326,
      "grad_norm": 2.6334118843078613,
      "learning_rate": 6.468101145507284e-07,
      "loss": 0.7644,
      "step": 24765
    },
    {
      "epoch": 0.3856243237562954,
      "grad_norm": 4.374823093414307,
      "learning_rate": 6.467281755461235e-07,
      "loss": 0.7128,
      "step": 24770
    },
    {
      "epoch": 0.38570216475826474,
      "grad_norm": 5.410842418670654,
      "learning_rate": 6.466462365415185e-07,
      "loss": 0.8169,
      "step": 24775
    },
    {
      "epoch": 0.38578000576023413,
      "grad_norm": 3.2832460403442383,
      "learning_rate": 6.465642975369134e-07,
      "loss": 0.9047,
      "step": 24780
    },
    {
      "epoch": 0.3858578467622035,
      "grad_norm": 3.3888192176818848,
      "learning_rate": 6.464823585323085e-07,
      "loss": 0.8511,
      "step": 24785
    },
    {
      "epoch": 0.3859356877641729,
      "grad_norm": 3.015672206878662,
      "learning_rate": 6.464004195277035e-07,
      "loss": 0.7058,
      "step": 24790
    },
    {
      "epoch": 0.38601352876614226,
      "grad_norm": 2.8976247310638428,
      "learning_rate": 6.463184805230986e-07,
      "loss": 0.5977,
      "step": 24795
    },
    {
      "epoch": 0.38609136976811165,
      "grad_norm": 4.574088096618652,
      "learning_rate": 6.462365415184936e-07,
      "loss": 0.7404,
      "step": 24800
    },
    {
      "epoch": 0.38616921077008104,
      "grad_norm": 3.3250582218170166,
      "learning_rate": 6.461546025138887e-07,
      "loss": 0.6752,
      "step": 24805
    },
    {
      "epoch": 0.38624705177205043,
      "grad_norm": 3.7079975605010986,
      "learning_rate": 6.460726635092837e-07,
      "loss": 0.7038,
      "step": 24810
    },
    {
      "epoch": 0.38632489277401977,
      "grad_norm": 4.7266340255737305,
      "learning_rate": 6.459907245046787e-07,
      "loss": 0.6684,
      "step": 24815
    },
    {
      "epoch": 0.38640273377598916,
      "grad_norm": 6.4636335372924805,
      "learning_rate": 6.459087855000737e-07,
      "loss": 0.7269,
      "step": 24820
    },
    {
      "epoch": 0.38648057477795855,
      "grad_norm": 3.2442867755889893,
      "learning_rate": 6.458268464954687e-07,
      "loss": 0.8114,
      "step": 24825
    },
    {
      "epoch": 0.38655841577992794,
      "grad_norm": 3.611947774887085,
      "learning_rate": 6.457449074908637e-07,
      "loss": 0.8077,
      "step": 24830
    },
    {
      "epoch": 0.3866362567818973,
      "grad_norm": 3.2939252853393555,
      "learning_rate": 6.456629684862588e-07,
      "loss": 0.7871,
      "step": 24835
    },
    {
      "epoch": 0.3867140977838667,
      "grad_norm": 6.327769756317139,
      "learning_rate": 6.455810294816539e-07,
      "loss": 0.7539,
      "step": 24840
    },
    {
      "epoch": 0.38679193878583606,
      "grad_norm": 3.1518681049346924,
      "learning_rate": 6.454990904770488e-07,
      "loss": 0.8088,
      "step": 24845
    },
    {
      "epoch": 0.3868697797878054,
      "grad_norm": 3.270970106124878,
      "learning_rate": 6.454171514724439e-07,
      "loss": 0.7782,
      "step": 24850
    },
    {
      "epoch": 0.3869476207897748,
      "grad_norm": 2.967019557952881,
      "learning_rate": 6.45335212467839e-07,
      "loss": 0.8034,
      "step": 24855
    },
    {
      "epoch": 0.3870254617917442,
      "grad_norm": 3.985086679458618,
      "learning_rate": 6.45253273463234e-07,
      "loss": 0.6594,
      "step": 24860
    },
    {
      "epoch": 0.3871033027937136,
      "grad_norm": 4.853755950927734,
      "learning_rate": 6.451713344586289e-07,
      "loss": 0.752,
      "step": 24865
    },
    {
      "epoch": 0.3871811437956829,
      "grad_norm": 4.004329681396484,
      "learning_rate": 6.45089395454024e-07,
      "loss": 0.8231,
      "step": 24870
    },
    {
      "epoch": 0.3872589847976523,
      "grad_norm": 5.640361309051514,
      "learning_rate": 6.45007456449419e-07,
      "loss": 0.7791,
      "step": 24875
    },
    {
      "epoch": 0.3873368257996217,
      "grad_norm": 4.084892749786377,
      "learning_rate": 6.44925517444814e-07,
      "loss": 0.7546,
      "step": 24880
    },
    {
      "epoch": 0.3874146668015911,
      "grad_norm": 4.119089603424072,
      "learning_rate": 6.448435784402091e-07,
      "loss": 0.6331,
      "step": 24885
    },
    {
      "epoch": 0.3874925078035604,
      "grad_norm": 2.9183003902435303,
      "learning_rate": 6.447616394356041e-07,
      "loss": 0.8319,
      "step": 24890
    },
    {
      "epoch": 0.3875703488055298,
      "grad_norm": 6.5224432945251465,
      "learning_rate": 6.446797004309992e-07,
      "loss": 0.8136,
      "step": 24895
    },
    {
      "epoch": 0.3876481898074992,
      "grad_norm": 3.6733174324035645,
      "learning_rate": 6.445977614263942e-07,
      "loss": 0.8367,
      "step": 24900
    },
    {
      "epoch": 0.3877260308094686,
      "grad_norm": 3.2794108390808105,
      "learning_rate": 6.445158224217891e-07,
      "loss": 0.8072,
      "step": 24905
    },
    {
      "epoch": 0.38780387181143794,
      "grad_norm": 2.9346797466278076,
      "learning_rate": 6.444338834171842e-07,
      "loss": 0.7321,
      "step": 24910
    },
    {
      "epoch": 0.38788171281340733,
      "grad_norm": 6.637734889984131,
      "learning_rate": 6.443519444125792e-07,
      "loss": 0.7524,
      "step": 24915
    },
    {
      "epoch": 0.3879595538153767,
      "grad_norm": 4.200438499450684,
      "learning_rate": 6.442700054079743e-07,
      "loss": 0.6879,
      "step": 24920
    },
    {
      "epoch": 0.3880373948173461,
      "grad_norm": 4.099344730377197,
      "learning_rate": 6.441880664033693e-07,
      "loss": 0.8778,
      "step": 24925
    },
    {
      "epoch": 0.38811523581931545,
      "grad_norm": 4.251168251037598,
      "learning_rate": 6.441061273987644e-07,
      "loss": 0.8255,
      "step": 24930
    },
    {
      "epoch": 0.38819307682128484,
      "grad_norm": 3.3429462909698486,
      "learning_rate": 6.440241883941594e-07,
      "loss": 0.8582,
      "step": 24935
    },
    {
      "epoch": 0.38827091782325424,
      "grad_norm": 2.959158182144165,
      "learning_rate": 6.439422493895544e-07,
      "loss": 0.8398,
      "step": 24940
    },
    {
      "epoch": 0.3883487588252236,
      "grad_norm": 3.3628015518188477,
      "learning_rate": 6.438603103849494e-07,
      "loss": 0.7873,
      "step": 24945
    },
    {
      "epoch": 0.38842659982719296,
      "grad_norm": 5.4761176109313965,
      "learning_rate": 6.437783713803445e-07,
      "loss": 0.716,
      "step": 24950
    },
    {
      "epoch": 0.38850444082916236,
      "grad_norm": 3.494793176651001,
      "learning_rate": 6.436964323757394e-07,
      "loss": 0.7331,
      "step": 24955
    },
    {
      "epoch": 0.38858228183113175,
      "grad_norm": 2.821584701538086,
      "learning_rate": 6.436144933711345e-07,
      "loss": 0.6953,
      "step": 24960
    },
    {
      "epoch": 0.3886601228331011,
      "grad_norm": 3.9607622623443604,
      "learning_rate": 6.435325543665296e-07,
      "loss": 0.8024,
      "step": 24965
    },
    {
      "epoch": 0.3887379638350705,
      "grad_norm": 4.245300769805908,
      "learning_rate": 6.434506153619245e-07,
      "loss": 0.6974,
      "step": 24970
    },
    {
      "epoch": 0.38881580483703987,
      "grad_norm": 4.189142227172852,
      "learning_rate": 6.433686763573196e-07,
      "loss": 0.6792,
      "step": 24975
    },
    {
      "epoch": 0.38889364583900926,
      "grad_norm": 5.561572074890137,
      "learning_rate": 6.432867373527147e-07,
      "loss": 0.7351,
      "step": 24980
    },
    {
      "epoch": 0.3889714868409786,
      "grad_norm": 4.93323278427124,
      "learning_rate": 6.432047983481096e-07,
      "loss": 0.8008,
      "step": 24985
    },
    {
      "epoch": 0.389049327842948,
      "grad_norm": 3.7661235332489014,
      "learning_rate": 6.431228593435046e-07,
      "loss": 0.768,
      "step": 24990
    },
    {
      "epoch": 0.3891271688449174,
      "grad_norm": 9.159685134887695,
      "learning_rate": 6.430409203388997e-07,
      "loss": 0.8869,
      "step": 24995
    },
    {
      "epoch": 0.3892050098468868,
      "grad_norm": 4.195098876953125,
      "learning_rate": 6.429589813342947e-07,
      "loss": 0.8319,
      "step": 25000
    },
    {
      "epoch": 0.3892828508488561,
      "grad_norm": 3.203718662261963,
      "learning_rate": 6.428770423296897e-07,
      "loss": 0.6736,
      "step": 25005
    },
    {
      "epoch": 0.3893606918508255,
      "grad_norm": 3.4601032733917236,
      "learning_rate": 6.427951033250848e-07,
      "loss": 0.7713,
      "step": 25010
    },
    {
      "epoch": 0.3894385328527949,
      "grad_norm": 3.0744853019714355,
      "learning_rate": 6.427131643204798e-07,
      "loss": 0.8134,
      "step": 25015
    },
    {
      "epoch": 0.3895163738547643,
      "grad_norm": 3.6189422607421875,
      "learning_rate": 6.426312253158749e-07,
      "loss": 0.7555,
      "step": 25020
    },
    {
      "epoch": 0.3895942148567336,
      "grad_norm": 7.418415069580078,
      "learning_rate": 6.425492863112698e-07,
      "loss": 0.7212,
      "step": 25025
    },
    {
      "epoch": 0.389672055858703,
      "grad_norm": 5.042339324951172,
      "learning_rate": 6.424673473066648e-07,
      "loss": 0.7072,
      "step": 25030
    },
    {
      "epoch": 0.3897498968606724,
      "grad_norm": 3.4799351692199707,
      "learning_rate": 6.423854083020599e-07,
      "loss": 0.6974,
      "step": 25035
    },
    {
      "epoch": 0.38982773786264174,
      "grad_norm": 3.5405311584472656,
      "learning_rate": 6.42303469297455e-07,
      "loss": 0.7653,
      "step": 25040
    },
    {
      "epoch": 0.38990557886461114,
      "grad_norm": 8.769624710083008,
      "learning_rate": 6.4222153029285e-07,
      "loss": 0.7843,
      "step": 25045
    },
    {
      "epoch": 0.3899834198665805,
      "grad_norm": 4.537402629852295,
      "learning_rate": 6.42139591288245e-07,
      "loss": 0.7039,
      "step": 25050
    },
    {
      "epoch": 0.3900612608685499,
      "grad_norm": 9.237510681152344,
      "learning_rate": 6.420576522836401e-07,
      "loss": 0.7074,
      "step": 25055
    },
    {
      "epoch": 0.39013910187051926,
      "grad_norm": 5.712589740753174,
      "learning_rate": 6.419757132790351e-07,
      "loss": 0.7478,
      "step": 25060
    },
    {
      "epoch": 0.39021694287248865,
      "grad_norm": 2.4703423976898193,
      "learning_rate": 6.4189377427443e-07,
      "loss": 0.7171,
      "step": 25065
    },
    {
      "epoch": 0.39029478387445804,
      "grad_norm": 4.031161308288574,
      "learning_rate": 6.418118352698251e-07,
      "loss": 0.8543,
      "step": 25070
    },
    {
      "epoch": 0.39037262487642743,
      "grad_norm": 3.8306655883789062,
      "learning_rate": 6.417298962652202e-07,
      "loss": 0.6768,
      "step": 25075
    },
    {
      "epoch": 0.39045046587839677,
      "grad_norm": 4.797236442565918,
      "learning_rate": 6.416479572606151e-07,
      "loss": 0.8301,
      "step": 25080
    },
    {
      "epoch": 0.39052830688036616,
      "grad_norm": 5.457010746002197,
      "learning_rate": 6.415660182560102e-07,
      "loss": 0.8202,
      "step": 25085
    },
    {
      "epoch": 0.39060614788233555,
      "grad_norm": 4.702629089355469,
      "learning_rate": 6.414840792514053e-07,
      "loss": 0.6733,
      "step": 25090
    },
    {
      "epoch": 0.39068398888430494,
      "grad_norm": 8.158437728881836,
      "learning_rate": 6.414021402468002e-07,
      "loss": 0.8074,
      "step": 25095
    },
    {
      "epoch": 0.3907618298862743,
      "grad_norm": 8.769953727722168,
      "learning_rate": 6.413202012421953e-07,
      "loss": 0.7251,
      "step": 25100
    },
    {
      "epoch": 0.3908396708882437,
      "grad_norm": 3.4470908641815186,
      "learning_rate": 6.412382622375903e-07,
      "loss": 0.8341,
      "step": 25105
    },
    {
      "epoch": 0.39091751189021307,
      "grad_norm": 8.288146018981934,
      "learning_rate": 6.411563232329853e-07,
      "loss": 0.8349,
      "step": 25110
    },
    {
      "epoch": 0.3909953528921824,
      "grad_norm": 3.165147066116333,
      "learning_rate": 6.410743842283803e-07,
      "loss": 0.7797,
      "step": 25115
    },
    {
      "epoch": 0.3910731938941518,
      "grad_norm": 3.657697916030884,
      "learning_rate": 6.409924452237754e-07,
      "loss": 0.763,
      "step": 25120
    },
    {
      "epoch": 0.3911510348961212,
      "grad_norm": 5.660918712615967,
      "learning_rate": 6.409105062191704e-07,
      "loss": 0.7126,
      "step": 25125
    },
    {
      "epoch": 0.3912288758980906,
      "grad_norm": 5.250837326049805,
      "learning_rate": 6.408285672145655e-07,
      "loss": 0.6867,
      "step": 25130
    },
    {
      "epoch": 0.3913067169000599,
      "grad_norm": 5.486688613891602,
      "learning_rate": 6.407466282099605e-07,
      "loss": 0.7719,
      "step": 25135
    },
    {
      "epoch": 0.3913845579020293,
      "grad_norm": 3.74428129196167,
      "learning_rate": 6.406646892053555e-07,
      "loss": 0.6571,
      "step": 25140
    },
    {
      "epoch": 0.3914623989039987,
      "grad_norm": 4.203434944152832,
      "learning_rate": 6.405827502007505e-07,
      "loss": 0.8256,
      "step": 25145
    },
    {
      "epoch": 0.3915402399059681,
      "grad_norm": 3.337033987045288,
      "learning_rate": 6.405008111961455e-07,
      "loss": 0.8205,
      "step": 25150
    },
    {
      "epoch": 0.3916180809079374,
      "grad_norm": 5.5420355796813965,
      "learning_rate": 6.404188721915405e-07,
      "loss": 0.7959,
      "step": 25155
    },
    {
      "epoch": 0.3916959219099068,
      "grad_norm": 3.8033721446990967,
      "learning_rate": 6.403369331869356e-07,
      "loss": 0.6854,
      "step": 25160
    },
    {
      "epoch": 0.3917737629118762,
      "grad_norm": 3.5326404571533203,
      "learning_rate": 6.402549941823307e-07,
      "loss": 0.8763,
      "step": 25165
    },
    {
      "epoch": 0.3918516039138456,
      "grad_norm": 2.995070219039917,
      "learning_rate": 6.401730551777257e-07,
      "loss": 0.8321,
      "step": 25170
    },
    {
      "epoch": 0.39192944491581494,
      "grad_norm": 5.510183334350586,
      "learning_rate": 6.400911161731207e-07,
      "loss": 0.6439,
      "step": 25175
    },
    {
      "epoch": 0.39200728591778433,
      "grad_norm": 3.0924971103668213,
      "learning_rate": 6.400091771685158e-07,
      "loss": 0.801,
      "step": 25180
    },
    {
      "epoch": 0.3920851269197537,
      "grad_norm": 3.860745668411255,
      "learning_rate": 6.399272381639109e-07,
      "loss": 0.7218,
      "step": 25185
    },
    {
      "epoch": 0.3921629679217231,
      "grad_norm": 3.2652904987335205,
      "learning_rate": 6.398452991593057e-07,
      "loss": 0.836,
      "step": 25190
    },
    {
      "epoch": 0.39224080892369245,
      "grad_norm": 7.08906364440918,
      "learning_rate": 6.397633601547008e-07,
      "loss": 0.8404,
      "step": 25195
    },
    {
      "epoch": 0.39231864992566184,
      "grad_norm": 2.7343451976776123,
      "learning_rate": 6.396814211500959e-07,
      "loss": 0.8146,
      "step": 25200
    },
    {
      "epoch": 0.39239649092763124,
      "grad_norm": 10.399474143981934,
      "learning_rate": 6.395994821454908e-07,
      "loss": 0.7137,
      "step": 25205
    },
    {
      "epoch": 0.3924743319296006,
      "grad_norm": 5.2434210777282715,
      "learning_rate": 6.395175431408859e-07,
      "loss": 0.7348,
      "step": 25210
    },
    {
      "epoch": 0.39255217293156996,
      "grad_norm": 3.0255560874938965,
      "learning_rate": 6.39435604136281e-07,
      "loss": 0.6458,
      "step": 25215
    },
    {
      "epoch": 0.39263001393353936,
      "grad_norm": 2.3744378089904785,
      "learning_rate": 6.39353665131676e-07,
      "loss": 0.7827,
      "step": 25220
    },
    {
      "epoch": 0.39270785493550875,
      "grad_norm": 3.563133955001831,
      "learning_rate": 6.39271726127071e-07,
      "loss": 0.7984,
      "step": 25225
    },
    {
      "epoch": 0.3927856959374781,
      "grad_norm": 4.089686393737793,
      "learning_rate": 6.39189787122466e-07,
      "loss": 0.7627,
      "step": 25230
    },
    {
      "epoch": 0.3928635369394475,
      "grad_norm": 4.701676368713379,
      "learning_rate": 6.39107848117861e-07,
      "loss": 0.8174,
      "step": 25235
    },
    {
      "epoch": 0.39294137794141687,
      "grad_norm": 4.46544075012207,
      "learning_rate": 6.39025909113256e-07,
      "loss": 0.7566,
      "step": 25240
    },
    {
      "epoch": 0.39301921894338626,
      "grad_norm": 18.95890998840332,
      "learning_rate": 6.389439701086511e-07,
      "loss": 0.7883,
      "step": 25245
    },
    {
      "epoch": 0.3930970599453556,
      "grad_norm": 7.087912559509277,
      "learning_rate": 6.388620311040461e-07,
      "loss": 0.7431,
      "step": 25250
    },
    {
      "epoch": 0.393174900947325,
      "grad_norm": 11.888138771057129,
      "learning_rate": 6.387800920994412e-07,
      "loss": 0.7625,
      "step": 25255
    },
    {
      "epoch": 0.3932527419492944,
      "grad_norm": 7.013394832611084,
      "learning_rate": 6.386981530948362e-07,
      "loss": 0.7365,
      "step": 25260
    },
    {
      "epoch": 0.3933305829512638,
      "grad_norm": 3.1967785358428955,
      "learning_rate": 6.386162140902312e-07,
      "loss": 0.7336,
      "step": 25265
    },
    {
      "epoch": 0.3934084239532331,
      "grad_norm": 3.2710204124450684,
      "learning_rate": 6.385342750856262e-07,
      "loss": 0.7243,
      "step": 25270
    },
    {
      "epoch": 0.3934862649552025,
      "grad_norm": 5.661640644073486,
      "learning_rate": 6.384523360810212e-07,
      "loss": 0.7948,
      "step": 25275
    },
    {
      "epoch": 0.3935641059571719,
      "grad_norm": 4.448530197143555,
      "learning_rate": 6.383703970764162e-07,
      "loss": 0.7979,
      "step": 25280
    },
    {
      "epoch": 0.39364194695914123,
      "grad_norm": 3.535083770751953,
      "learning_rate": 6.382884580718113e-07,
      "loss": 0.7797,
      "step": 25285
    },
    {
      "epoch": 0.3937197879611106,
      "grad_norm": 3.0342564582824707,
      "learning_rate": 6.382065190672064e-07,
      "loss": 0.6707,
      "step": 25290
    },
    {
      "epoch": 0.39379762896308,
      "grad_norm": 3.252420663833618,
      "learning_rate": 6.381245800626014e-07,
      "loss": 0.6687,
      "step": 25295
    },
    {
      "epoch": 0.3938754699650494,
      "grad_norm": 5.64980936050415,
      "learning_rate": 6.380426410579964e-07,
      "loss": 0.7643,
      "step": 25300
    },
    {
      "epoch": 0.39395331096701874,
      "grad_norm": 2.8796017169952393,
      "learning_rate": 6.379607020533915e-07,
      "loss": 0.8468,
      "step": 25305
    },
    {
      "epoch": 0.39403115196898814,
      "grad_norm": 4.910164833068848,
      "learning_rate": 6.378787630487865e-07,
      "loss": 0.8456,
      "step": 25310
    },
    {
      "epoch": 0.39410899297095753,
      "grad_norm": 3.392526865005493,
      "learning_rate": 6.377968240441814e-07,
      "loss": 0.7253,
      "step": 25315
    },
    {
      "epoch": 0.3941868339729269,
      "grad_norm": 3.7609364986419678,
      "learning_rate": 6.377148850395765e-07,
      "loss": 0.6837,
      "step": 25320
    },
    {
      "epoch": 0.39426467497489626,
      "grad_norm": 3.4429965019226074,
      "learning_rate": 6.376329460349716e-07,
      "loss": 0.7907,
      "step": 25325
    },
    {
      "epoch": 0.39434251597686565,
      "grad_norm": 3.964439630508423,
      "learning_rate": 6.375510070303665e-07,
      "loss": 0.6576,
      "step": 25330
    },
    {
      "epoch": 0.39442035697883504,
      "grad_norm": 9.316553115844727,
      "learning_rate": 6.374690680257616e-07,
      "loss": 0.7046,
      "step": 25335
    },
    {
      "epoch": 0.39449819798080443,
      "grad_norm": 3.9549331665039062,
      "learning_rate": 6.373871290211567e-07,
      "loss": 0.752,
      "step": 25340
    },
    {
      "epoch": 0.39457603898277377,
      "grad_norm": 6.220914840698242,
      "learning_rate": 6.373051900165517e-07,
      "loss": 0.8579,
      "step": 25345
    },
    {
      "epoch": 0.39465387998474316,
      "grad_norm": 5.23582124710083,
      "learning_rate": 6.372232510119466e-07,
      "loss": 0.757,
      "step": 25350
    },
    {
      "epoch": 0.39473172098671255,
      "grad_norm": 6.521960735321045,
      "learning_rate": 6.371413120073417e-07,
      "loss": 0.8323,
      "step": 25355
    },
    {
      "epoch": 0.39480956198868195,
      "grad_norm": 3.584139347076416,
      "learning_rate": 6.370593730027367e-07,
      "loss": 0.7943,
      "step": 25360
    },
    {
      "epoch": 0.3948874029906513,
      "grad_norm": 13.306678771972656,
      "learning_rate": 6.369774339981317e-07,
      "loss": 0.73,
      "step": 25365
    },
    {
      "epoch": 0.3949652439926207,
      "grad_norm": 4.9485039710998535,
      "learning_rate": 6.368954949935268e-07,
      "loss": 0.7932,
      "step": 25370
    },
    {
      "epoch": 0.39504308499459007,
      "grad_norm": 3.292325735092163,
      "learning_rate": 6.368135559889218e-07,
      "loss": 0.8218,
      "step": 25375
    },
    {
      "epoch": 0.3951209259965594,
      "grad_norm": 3.3515844345092773,
      "learning_rate": 6.367316169843169e-07,
      "loss": 0.7421,
      "step": 25380
    },
    {
      "epoch": 0.3951987669985288,
      "grad_norm": 5.699435710906982,
      "learning_rate": 6.366496779797119e-07,
      "loss": 0.7649,
      "step": 25385
    },
    {
      "epoch": 0.3952766080004982,
      "grad_norm": 8.88771915435791,
      "learning_rate": 6.365677389751068e-07,
      "loss": 0.7564,
      "step": 25390
    },
    {
      "epoch": 0.3953544490024676,
      "grad_norm": 3.6982576847076416,
      "learning_rate": 6.364857999705019e-07,
      "loss": 0.7699,
      "step": 25395
    },
    {
      "epoch": 0.3954322900044369,
      "grad_norm": 4.210885524749756,
      "learning_rate": 6.36403860965897e-07,
      "loss": 0.7438,
      "step": 25400
    },
    {
      "epoch": 0.3955101310064063,
      "grad_norm": 3.2768607139587402,
      "learning_rate": 6.363219219612919e-07,
      "loss": 0.7445,
      "step": 25405
    },
    {
      "epoch": 0.3955879720083757,
      "grad_norm": 4.871786117553711,
      "learning_rate": 6.36239982956687e-07,
      "loss": 0.7446,
      "step": 25410
    },
    {
      "epoch": 0.3956658130103451,
      "grad_norm": 5.095002174377441,
      "learning_rate": 6.361580439520821e-07,
      "loss": 0.7323,
      "step": 25415
    },
    {
      "epoch": 0.3957436540123144,
      "grad_norm": 6.775503158569336,
      "learning_rate": 6.360761049474771e-07,
      "loss": 0.7104,
      "step": 25420
    },
    {
      "epoch": 0.3958214950142838,
      "grad_norm": 3.6449718475341797,
      "learning_rate": 6.359941659428721e-07,
      "loss": 0.8384,
      "step": 25425
    },
    {
      "epoch": 0.3958993360162532,
      "grad_norm": 3.4597651958465576,
      "learning_rate": 6.359122269382671e-07,
      "loss": 0.7858,
      "step": 25430
    },
    {
      "epoch": 0.3959771770182226,
      "grad_norm": 3.478415012359619,
      "learning_rate": 6.358302879336622e-07,
      "loss": 0.8736,
      "step": 25435
    },
    {
      "epoch": 0.39605501802019194,
      "grad_norm": 4.179505348205566,
      "learning_rate": 6.357483489290571e-07,
      "loss": 0.7169,
      "step": 25440
    },
    {
      "epoch": 0.39613285902216133,
      "grad_norm": 3.7945547103881836,
      "learning_rate": 6.356664099244522e-07,
      "loss": 0.7743,
      "step": 25445
    },
    {
      "epoch": 0.3962107000241307,
      "grad_norm": 3.468790054321289,
      "learning_rate": 6.355844709198473e-07,
      "loss": 0.7189,
      "step": 25450
    },
    {
      "epoch": 0.39628854102610006,
      "grad_norm": 5.45301628112793,
      "learning_rate": 6.355025319152422e-07,
      "loss": 0.8203,
      "step": 25455
    },
    {
      "epoch": 0.39636638202806945,
      "grad_norm": 9.016706466674805,
      "learning_rate": 6.354205929106373e-07,
      "loss": 0.9262,
      "step": 25460
    },
    {
      "epoch": 0.39644422303003884,
      "grad_norm": 3.1717607975006104,
      "learning_rate": 6.353386539060324e-07,
      "loss": 0.7329,
      "step": 25465
    },
    {
      "epoch": 0.39652206403200824,
      "grad_norm": 2.699633836746216,
      "learning_rate": 6.352567149014274e-07,
      "loss": 0.6813,
      "step": 25470
    },
    {
      "epoch": 0.3965999050339776,
      "grad_norm": 4.039689064025879,
      "learning_rate": 6.351747758968223e-07,
      "loss": 0.7226,
      "step": 25475
    },
    {
      "epoch": 0.39667774603594697,
      "grad_norm": 4.107393741607666,
      "learning_rate": 6.350928368922174e-07,
      "loss": 0.824,
      "step": 25480
    },
    {
      "epoch": 0.39675558703791636,
      "grad_norm": 4.444277763366699,
      "learning_rate": 6.350108978876124e-07,
      "loss": 0.8854,
      "step": 25485
    },
    {
      "epoch": 0.39683342803988575,
      "grad_norm": 6.592438220977783,
      "learning_rate": 6.349289588830075e-07,
      "loss": 0.651,
      "step": 25490
    },
    {
      "epoch": 0.3969112690418551,
      "grad_norm": 6.207258224487305,
      "learning_rate": 6.348470198784025e-07,
      "loss": 0.6449,
      "step": 25495
    },
    {
      "epoch": 0.3969891100438245,
      "grad_norm": 4.503571033477783,
      "learning_rate": 6.347650808737975e-07,
      "loss": 0.807,
      "step": 25500
    },
    {
      "epoch": 0.39706695104579387,
      "grad_norm": 3.707956075668335,
      "learning_rate": 6.346831418691926e-07,
      "loss": 0.7628,
      "step": 25505
    },
    {
      "epoch": 0.39714479204776326,
      "grad_norm": 2.8668758869171143,
      "learning_rate": 6.346012028645876e-07,
      "loss": 0.6728,
      "step": 25510
    },
    {
      "epoch": 0.3972226330497326,
      "grad_norm": 5.505023002624512,
      "learning_rate": 6.345192638599825e-07,
      "loss": 0.7744,
      "step": 25515
    },
    {
      "epoch": 0.397300474051702,
      "grad_norm": 2.5895702838897705,
      "learning_rate": 6.344373248553776e-07,
      "loss": 0.7283,
      "step": 25520
    },
    {
      "epoch": 0.3973783150536714,
      "grad_norm": 3.2556867599487305,
      "learning_rate": 6.343553858507727e-07,
      "loss": 0.6701,
      "step": 25525
    },
    {
      "epoch": 0.3974561560556408,
      "grad_norm": 3.534254550933838,
      "learning_rate": 6.342734468461676e-07,
      "loss": 0.8053,
      "step": 25530
    },
    {
      "epoch": 0.3975339970576101,
      "grad_norm": 4.892989158630371,
      "learning_rate": 6.341915078415627e-07,
      "loss": 0.6051,
      "step": 25535
    },
    {
      "epoch": 0.3976118380595795,
      "grad_norm": 3.6802456378936768,
      "learning_rate": 6.341095688369578e-07,
      "loss": 0.7358,
      "step": 25540
    },
    {
      "epoch": 0.3976896790615489,
      "grad_norm": 3.2066876888275146,
      "learning_rate": 6.340276298323529e-07,
      "loss": 0.7943,
      "step": 25545
    },
    {
      "epoch": 0.39776752006351823,
      "grad_norm": 5.907846450805664,
      "learning_rate": 6.339456908277478e-07,
      "loss": 0.7533,
      "step": 25550
    },
    {
      "epoch": 0.3978453610654876,
      "grad_norm": 4.704529762268066,
      "learning_rate": 6.338637518231428e-07,
      "loss": 0.6999,
      "step": 25555
    },
    {
      "epoch": 0.397923202067457,
      "grad_norm": 3.5781517028808594,
      "learning_rate": 6.337818128185379e-07,
      "loss": 0.7683,
      "step": 25560
    },
    {
      "epoch": 0.3980010430694264,
      "grad_norm": 3.7871382236480713,
      "learning_rate": 6.336998738139328e-07,
      "loss": 0.903,
      "step": 25565
    },
    {
      "epoch": 0.39807888407139574,
      "grad_norm": 6.048968315124512,
      "learning_rate": 6.336179348093279e-07,
      "loss": 0.8356,
      "step": 25570
    },
    {
      "epoch": 0.39815672507336514,
      "grad_norm": 4.461136341094971,
      "learning_rate": 6.33535995804723e-07,
      "loss": 0.819,
      "step": 25575
    },
    {
      "epoch": 0.39823456607533453,
      "grad_norm": 3.7887139320373535,
      "learning_rate": 6.33454056800118e-07,
      "loss": 0.7609,
      "step": 25580
    },
    {
      "epoch": 0.3983124070773039,
      "grad_norm": 3.172938823699951,
      "learning_rate": 6.33372117795513e-07,
      "loss": 0.76,
      "step": 25585
    },
    {
      "epoch": 0.39839024807927326,
      "grad_norm": 4.127932548522949,
      "learning_rate": 6.332901787909081e-07,
      "loss": 0.808,
      "step": 25590
    },
    {
      "epoch": 0.39846808908124265,
      "grad_norm": 3.771493911743164,
      "learning_rate": 6.33208239786303e-07,
      "loss": 0.8112,
      "step": 25595
    },
    {
      "epoch": 0.39854593008321204,
      "grad_norm": 4.411073207855225,
      "learning_rate": 6.33126300781698e-07,
      "loss": 0.894,
      "step": 25600
    },
    {
      "epoch": 0.39862377108518143,
      "grad_norm": 4.447315216064453,
      "learning_rate": 6.330443617770931e-07,
      "loss": 0.7567,
      "step": 25605
    },
    {
      "epoch": 0.39870161208715077,
      "grad_norm": 6.834893703460693,
      "learning_rate": 6.329624227724881e-07,
      "loss": 0.9129,
      "step": 25610
    },
    {
      "epoch": 0.39877945308912016,
      "grad_norm": 4.036401748657227,
      "learning_rate": 6.328804837678832e-07,
      "loss": 0.8468,
      "step": 25615
    },
    {
      "epoch": 0.39885729409108955,
      "grad_norm": 3.791520833969116,
      "learning_rate": 6.327985447632782e-07,
      "loss": 0.7759,
      "step": 25620
    },
    {
      "epoch": 0.3989351350930589,
      "grad_norm": 3.5723390579223633,
      "learning_rate": 6.327166057586732e-07,
      "loss": 0.6774,
      "step": 25625
    },
    {
      "epoch": 0.3990129760950283,
      "grad_norm": 10.609052658081055,
      "learning_rate": 6.326346667540683e-07,
      "loss": 0.8008,
      "step": 25630
    },
    {
      "epoch": 0.3990908170969977,
      "grad_norm": 3.5287327766418457,
      "learning_rate": 6.325527277494633e-07,
      "loss": 0.6881,
      "step": 25635
    },
    {
      "epoch": 0.39916865809896707,
      "grad_norm": 3.306892156600952,
      "learning_rate": 6.324707887448582e-07,
      "loss": 0.7992,
      "step": 25640
    },
    {
      "epoch": 0.3992464991009364,
      "grad_norm": 3.6389737129211426,
      "learning_rate": 6.323888497402533e-07,
      "loss": 0.5907,
      "step": 25645
    },
    {
      "epoch": 0.3993243401029058,
      "grad_norm": 3.8080899715423584,
      "learning_rate": 6.323069107356484e-07,
      "loss": 0.7649,
      "step": 25650
    },
    {
      "epoch": 0.3994021811048752,
      "grad_norm": 3.856809139251709,
      "learning_rate": 6.322249717310433e-07,
      "loss": 0.783,
      "step": 25655
    },
    {
      "epoch": 0.3994800221068446,
      "grad_norm": 2.6995177268981934,
      "learning_rate": 6.321430327264384e-07,
      "loss": 0.7815,
      "step": 25660
    },
    {
      "epoch": 0.3995578631088139,
      "grad_norm": 3.1869149208068848,
      "learning_rate": 6.320610937218335e-07,
      "loss": 0.7579,
      "step": 25665
    },
    {
      "epoch": 0.3996357041107833,
      "grad_norm": 3.3936657905578613,
      "learning_rate": 6.319791547172286e-07,
      "loss": 0.6582,
      "step": 25670
    },
    {
      "epoch": 0.3997135451127527,
      "grad_norm": 3.886843204498291,
      "learning_rate": 6.318972157126234e-07,
      "loss": 0.7062,
      "step": 25675
    },
    {
      "epoch": 0.3997913861147221,
      "grad_norm": 5.519811153411865,
      "learning_rate": 6.318152767080185e-07,
      "loss": 0.7387,
      "step": 25680
    },
    {
      "epoch": 0.39986922711669143,
      "grad_norm": 4.253294467926025,
      "learning_rate": 6.317333377034136e-07,
      "loss": 0.6748,
      "step": 25685
    },
    {
      "epoch": 0.3999470681186608,
      "grad_norm": 2.7694480419158936,
      "learning_rate": 6.316513986988085e-07,
      "loss": 0.8194,
      "step": 25690
    },
    {
      "epoch": 0.4000249091206302,
      "grad_norm": 3.7576897144317627,
      "learning_rate": 6.315694596942036e-07,
      "loss": 0.6366,
      "step": 25695
    },
    {
      "epoch": 0.4001027501225996,
      "grad_norm": 5.782569408416748,
      "learning_rate": 6.314875206895987e-07,
      "loss": 0.8115,
      "step": 25700
    },
    {
      "epoch": 0.40018059112456894,
      "grad_norm": 3.806763172149658,
      "learning_rate": 6.314055816849937e-07,
      "loss": 0.791,
      "step": 25705
    },
    {
      "epoch": 0.40025843212653833,
      "grad_norm": 7.054147720336914,
      "learning_rate": 6.313236426803887e-07,
      "loss": 0.7312,
      "step": 25710
    },
    {
      "epoch": 0.4003362731285077,
      "grad_norm": 3.344073534011841,
      "learning_rate": 6.312417036757837e-07,
      "loss": 0.7061,
      "step": 25715
    },
    {
      "epoch": 0.40041411413047706,
      "grad_norm": 3.3587987422943115,
      "learning_rate": 6.311597646711787e-07,
      "loss": 0.7641,
      "step": 25720
    },
    {
      "epoch": 0.40049195513244645,
      "grad_norm": 2.2525744438171387,
      "learning_rate": 6.310778256665738e-07,
      "loss": 0.8433,
      "step": 25725
    },
    {
      "epoch": 0.40056979613441585,
      "grad_norm": 5.507537841796875,
      "learning_rate": 6.309958866619688e-07,
      "loss": 0.8923,
      "step": 25730
    },
    {
      "epoch": 0.40064763713638524,
      "grad_norm": 5.3320512771606445,
      "learning_rate": 6.309139476573638e-07,
      "loss": 0.7727,
      "step": 25735
    },
    {
      "epoch": 0.4007254781383546,
      "grad_norm": 3.6437861919403076,
      "learning_rate": 6.308320086527589e-07,
      "loss": 0.7592,
      "step": 25740
    },
    {
      "epoch": 0.40080331914032397,
      "grad_norm": 5.713900566101074,
      "learning_rate": 6.307500696481539e-07,
      "loss": 0.7082,
      "step": 25745
    },
    {
      "epoch": 0.40088116014229336,
      "grad_norm": 6.304920673370361,
      "learning_rate": 6.306681306435489e-07,
      "loss": 0.7573,
      "step": 25750
    },
    {
      "epoch": 0.40095900114426275,
      "grad_norm": 3.1046628952026367,
      "learning_rate": 6.305861916389439e-07,
      "loss": 0.8484,
      "step": 25755
    },
    {
      "epoch": 0.4010368421462321,
      "grad_norm": 2.9262259006500244,
      "learning_rate": 6.30504252634339e-07,
      "loss": 0.7134,
      "step": 25760
    },
    {
      "epoch": 0.4011146831482015,
      "grad_norm": 5.784107208251953,
      "learning_rate": 6.304223136297339e-07,
      "loss": 0.8202,
      "step": 25765
    },
    {
      "epoch": 0.40119252415017087,
      "grad_norm": 3.4631705284118652,
      "learning_rate": 6.30340374625129e-07,
      "loss": 0.9114,
      "step": 25770
    },
    {
      "epoch": 0.40127036515214026,
      "grad_norm": 3.4223835468292236,
      "learning_rate": 6.302584356205241e-07,
      "loss": 0.7365,
      "step": 25775
    },
    {
      "epoch": 0.4013482061541096,
      "grad_norm": 8.433497428894043,
      "learning_rate": 6.30176496615919e-07,
      "loss": 0.6492,
      "step": 25780
    },
    {
      "epoch": 0.401426047156079,
      "grad_norm": 4.8615827560424805,
      "learning_rate": 6.300945576113141e-07,
      "loss": 0.6824,
      "step": 25785
    },
    {
      "epoch": 0.4015038881580484,
      "grad_norm": 4.456131935119629,
      "learning_rate": 6.300126186067092e-07,
      "loss": 0.7575,
      "step": 25790
    },
    {
      "epoch": 0.4015817291600178,
      "grad_norm": 3.5618176460266113,
      "learning_rate": 6.299306796021043e-07,
      "loss": 0.7174,
      "step": 25795
    },
    {
      "epoch": 0.4016595701619871,
      "grad_norm": 3.5955419540405273,
      "learning_rate": 6.298487405974991e-07,
      "loss": 0.7488,
      "step": 25800
    },
    {
      "epoch": 0.4017374111639565,
      "grad_norm": 3.817229986190796,
      "learning_rate": 6.297668015928942e-07,
      "loss": 0.7705,
      "step": 25805
    },
    {
      "epoch": 0.4018152521659259,
      "grad_norm": 2.675896167755127,
      "learning_rate": 6.296848625882893e-07,
      "loss": 0.7554,
      "step": 25810
    },
    {
      "epoch": 0.40189309316789523,
      "grad_norm": 5.006580829620361,
      "learning_rate": 6.296029235836843e-07,
      "loss": 0.7452,
      "step": 25815
    },
    {
      "epoch": 0.4019709341698646,
      "grad_norm": 7.0662946701049805,
      "learning_rate": 6.295209845790793e-07,
      "loss": 0.7118,
      "step": 25820
    },
    {
      "epoch": 0.402048775171834,
      "grad_norm": 4.054603099822998,
      "learning_rate": 6.294390455744744e-07,
      "loss": 0.7652,
      "step": 25825
    },
    {
      "epoch": 0.4021266161738034,
      "grad_norm": 6.0550408363342285,
      "learning_rate": 6.293571065698694e-07,
      "loss": 0.6895,
      "step": 25830
    },
    {
      "epoch": 0.40220445717577274,
      "grad_norm": 4.203376293182373,
      "learning_rate": 6.292751675652644e-07,
      "loss": 0.751,
      "step": 25835
    },
    {
      "epoch": 0.40228229817774214,
      "grad_norm": 3.9787485599517822,
      "learning_rate": 6.291932285606594e-07,
      "loss": 0.6545,
      "step": 25840
    },
    {
      "epoch": 0.40236013917971153,
      "grad_norm": 3.0336601734161377,
      "learning_rate": 6.291112895560544e-07,
      "loss": 0.8396,
      "step": 25845
    },
    {
      "epoch": 0.4024379801816809,
      "grad_norm": 4.524356365203857,
      "learning_rate": 6.290293505514495e-07,
      "loss": 0.7489,
      "step": 25850
    },
    {
      "epoch": 0.40251582118365026,
      "grad_norm": 2.6053860187530518,
      "learning_rate": 6.289474115468445e-07,
      "loss": 0.7061,
      "step": 25855
    },
    {
      "epoch": 0.40259366218561965,
      "grad_norm": 3.6728532314300537,
      "learning_rate": 6.288654725422395e-07,
      "loss": 0.7726,
      "step": 25860
    },
    {
      "epoch": 0.40267150318758904,
      "grad_norm": 10.254779815673828,
      "learning_rate": 6.287835335376346e-07,
      "loss": 0.7134,
      "step": 25865
    },
    {
      "epoch": 0.40274934418955843,
      "grad_norm": 4.043647766113281,
      "learning_rate": 6.287015945330297e-07,
      "loss": 0.7779,
      "step": 25870
    },
    {
      "epoch": 0.40282718519152777,
      "grad_norm": 3.0604496002197266,
      "learning_rate": 6.286196555284246e-07,
      "loss": 0.8593,
      "step": 25875
    },
    {
      "epoch": 0.40290502619349716,
      "grad_norm": 6.564319610595703,
      "learning_rate": 6.285377165238196e-07,
      "loss": 0.6981,
      "step": 25880
    },
    {
      "epoch": 0.40298286719546655,
      "grad_norm": 4.262159824371338,
      "learning_rate": 6.284557775192147e-07,
      "loss": 0.7919,
      "step": 25885
    },
    {
      "epoch": 0.4030607081974359,
      "grad_norm": 5.1394853591918945,
      "learning_rate": 6.283738385146096e-07,
      "loss": 0.8422,
      "step": 25890
    },
    {
      "epoch": 0.4031385491994053,
      "grad_norm": 3.8123087882995605,
      "learning_rate": 6.282918995100047e-07,
      "loss": 0.7922,
      "step": 25895
    },
    {
      "epoch": 0.4032163902013747,
      "grad_norm": 6.320923805236816,
      "learning_rate": 6.282099605053998e-07,
      "loss": 0.7794,
      "step": 25900
    },
    {
      "epoch": 0.40329423120334407,
      "grad_norm": 3.729196310043335,
      "learning_rate": 6.281280215007948e-07,
      "loss": 0.6516,
      "step": 25905
    },
    {
      "epoch": 0.4033720722053134,
      "grad_norm": 4.0743255615234375,
      "learning_rate": 6.280460824961898e-07,
      "loss": 0.8156,
      "step": 25910
    },
    {
      "epoch": 0.4034499132072828,
      "grad_norm": 2.5866003036499023,
      "learning_rate": 6.279641434915849e-07,
      "loss": 0.7341,
      "step": 25915
    },
    {
      "epoch": 0.4035277542092522,
      "grad_norm": 3.3593242168426514,
      "learning_rate": 6.278822044869798e-07,
      "loss": 0.7359,
      "step": 25920
    },
    {
      "epoch": 0.4036055952112216,
      "grad_norm": 3.165477991104126,
      "learning_rate": 6.278002654823748e-07,
      "loss": 0.6969,
      "step": 25925
    },
    {
      "epoch": 0.4036834362131909,
      "grad_norm": 4.103423595428467,
      "learning_rate": 6.277183264777699e-07,
      "loss": 0.7591,
      "step": 25930
    },
    {
      "epoch": 0.4037612772151603,
      "grad_norm": 6.460220813751221,
      "learning_rate": 6.27636387473165e-07,
      "loss": 0.8588,
      "step": 25935
    },
    {
      "epoch": 0.4038391182171297,
      "grad_norm": 4.974887847900391,
      "learning_rate": 6.2755444846856e-07,
      "loss": 0.8204,
      "step": 25940
    },
    {
      "epoch": 0.4039169592190991,
      "grad_norm": 3.6187150478363037,
      "learning_rate": 6.27472509463955e-07,
      "loss": 0.8373,
      "step": 25945
    },
    {
      "epoch": 0.40399480022106843,
      "grad_norm": 4.209701061248779,
      "learning_rate": 6.273905704593501e-07,
      "loss": 0.7623,
      "step": 25950
    },
    {
      "epoch": 0.4040726412230378,
      "grad_norm": 3.2338383197784424,
      "learning_rate": 6.273086314547451e-07,
      "loss": 0.8593,
      "step": 25955
    },
    {
      "epoch": 0.4041504822250072,
      "grad_norm": 4.725485324859619,
      "learning_rate": 6.2722669245014e-07,
      "loss": 0.7365,
      "step": 25960
    },
    {
      "epoch": 0.4042283232269766,
      "grad_norm": 7.591894149780273,
      "learning_rate": 6.271447534455351e-07,
      "loss": 0.6815,
      "step": 25965
    },
    {
      "epoch": 0.40430616422894594,
      "grad_norm": 6.150928497314453,
      "learning_rate": 6.270628144409301e-07,
      "loss": 0.7711,
      "step": 25970
    },
    {
      "epoch": 0.40438400523091533,
      "grad_norm": 5.162087440490723,
      "learning_rate": 6.269808754363252e-07,
      "loss": 0.778,
      "step": 25975
    },
    {
      "epoch": 0.4044618462328847,
      "grad_norm": 11.32683277130127,
      "learning_rate": 6.268989364317202e-07,
      "loss": 0.8482,
      "step": 25980
    },
    {
      "epoch": 0.40453968723485406,
      "grad_norm": 3.6983065605163574,
      "learning_rate": 6.268169974271152e-07,
      "loss": 0.7804,
      "step": 25985
    },
    {
      "epoch": 0.40461752823682345,
      "grad_norm": 3.600092887878418,
      "learning_rate": 6.267350584225103e-07,
      "loss": 0.76,
      "step": 25990
    },
    {
      "epoch": 0.40469536923879285,
      "grad_norm": 4.2920050621032715,
      "learning_rate": 6.266531194179054e-07,
      "loss": 0.6658,
      "step": 25995
    },
    {
      "epoch": 0.40477321024076224,
      "grad_norm": 3.719242572784424,
      "learning_rate": 6.265711804133002e-07,
      "loss": 0.7362,
      "step": 26000
    },
    {
      "epoch": 0.4048510512427316,
      "grad_norm": 4.887945175170898,
      "learning_rate": 6.264892414086953e-07,
      "loss": 0.8455,
      "step": 26005
    },
    {
      "epoch": 0.40492889224470097,
      "grad_norm": 3.4599506855010986,
      "learning_rate": 6.264073024040904e-07,
      "loss": 0.8767,
      "step": 26010
    },
    {
      "epoch": 0.40500673324667036,
      "grad_norm": 3.034026861190796,
      "learning_rate": 6.263253633994853e-07,
      "loss": 0.7008,
      "step": 26015
    },
    {
      "epoch": 0.40508457424863975,
      "grad_norm": 4.146843433380127,
      "learning_rate": 6.262434243948804e-07,
      "loss": 0.7626,
      "step": 26020
    },
    {
      "epoch": 0.4051624152506091,
      "grad_norm": 6.154612064361572,
      "learning_rate": 6.261614853902755e-07,
      "loss": 0.8537,
      "step": 26025
    },
    {
      "epoch": 0.4052402562525785,
      "grad_norm": 2.850789785385132,
      "learning_rate": 6.260795463856705e-07,
      "loss": 0.7535,
      "step": 26030
    },
    {
      "epoch": 0.40531809725454787,
      "grad_norm": 3.5527889728546143,
      "learning_rate": 6.259976073810655e-07,
      "loss": 0.8057,
      "step": 26035
    },
    {
      "epoch": 0.40539593825651726,
      "grad_norm": 3.5174005031585693,
      "learning_rate": 6.259156683764605e-07,
      "loss": 0.7248,
      "step": 26040
    },
    {
      "epoch": 0.4054737792584866,
      "grad_norm": 3.333939790725708,
      "learning_rate": 6.258337293718555e-07,
      "loss": 0.8365,
      "step": 26045
    },
    {
      "epoch": 0.405551620260456,
      "grad_norm": 3.6032981872558594,
      "learning_rate": 6.257517903672505e-07,
      "loss": 0.7541,
      "step": 26050
    },
    {
      "epoch": 0.4056294612624254,
      "grad_norm": 2.9027950763702393,
      "learning_rate": 6.256698513626456e-07,
      "loss": 0.7457,
      "step": 26055
    },
    {
      "epoch": 0.4057073022643947,
      "grad_norm": 3.3258626461029053,
      "learning_rate": 6.255879123580407e-07,
      "loss": 0.8462,
      "step": 26060
    },
    {
      "epoch": 0.4057851432663641,
      "grad_norm": 7.839389801025391,
      "learning_rate": 6.255059733534357e-07,
      "loss": 0.6776,
      "step": 26065
    },
    {
      "epoch": 0.4058629842683335,
      "grad_norm": 4.488097190856934,
      "learning_rate": 6.254240343488307e-07,
      "loss": 0.7482,
      "step": 26070
    },
    {
      "epoch": 0.4059408252703029,
      "grad_norm": 5.980696678161621,
      "learning_rate": 6.253420953442258e-07,
      "loss": 0.6656,
      "step": 26075
    },
    {
      "epoch": 0.40601866627227223,
      "grad_norm": 2.6422781944274902,
      "learning_rate": 6.252601563396207e-07,
      "loss": 0.7025,
      "step": 26080
    },
    {
      "epoch": 0.4060965072742416,
      "grad_norm": 6.426515102386475,
      "learning_rate": 6.251782173350158e-07,
      "loss": 0.8232,
      "step": 26085
    },
    {
      "epoch": 0.406174348276211,
      "grad_norm": 8.42750358581543,
      "learning_rate": 6.250962783304108e-07,
      "loss": 0.7168,
      "step": 26090
    },
    {
      "epoch": 0.4062521892781804,
      "grad_norm": 3.8758537769317627,
      "learning_rate": 6.250143393258058e-07,
      "loss": 0.8671,
      "step": 26095
    },
    {
      "epoch": 0.40633003028014975,
      "grad_norm": 3.026585817337036,
      "learning_rate": 6.249324003212009e-07,
      "loss": 0.8579,
      "step": 26100
    },
    {
      "epoch": 0.40640787128211914,
      "grad_norm": 4.004497051239014,
      "learning_rate": 6.248504613165959e-07,
      "loss": 0.7519,
      "step": 26105
    },
    {
      "epoch": 0.40648571228408853,
      "grad_norm": 3.6319313049316406,
      "learning_rate": 6.247685223119909e-07,
      "loss": 0.8311,
      "step": 26110
    },
    {
      "epoch": 0.4065635532860579,
      "grad_norm": 5.742655277252197,
      "learning_rate": 6.24686583307386e-07,
      "loss": 0.8036,
      "step": 26115
    },
    {
      "epoch": 0.40664139428802726,
      "grad_norm": 8.020777702331543,
      "learning_rate": 6.246046443027811e-07,
      "loss": 0.8244,
      "step": 26120
    },
    {
      "epoch": 0.40671923528999665,
      "grad_norm": 4.790924072265625,
      "learning_rate": 6.245227052981759e-07,
      "loss": 0.6761,
      "step": 26125
    },
    {
      "epoch": 0.40679707629196604,
      "grad_norm": 4.601445198059082,
      "learning_rate": 6.24440766293571e-07,
      "loss": 0.7301,
      "step": 26130
    },
    {
      "epoch": 0.40687491729393543,
      "grad_norm": 4.280304431915283,
      "learning_rate": 6.243588272889661e-07,
      "loss": 0.7748,
      "step": 26135
    },
    {
      "epoch": 0.40695275829590477,
      "grad_norm": 6.949324607849121,
      "learning_rate": 6.24276888284361e-07,
      "loss": 0.7076,
      "step": 26140
    },
    {
      "epoch": 0.40703059929787416,
      "grad_norm": 2.876680850982666,
      "learning_rate": 6.241949492797561e-07,
      "loss": 0.6675,
      "step": 26145
    },
    {
      "epoch": 0.40710844029984355,
      "grad_norm": 3.646688461303711,
      "learning_rate": 6.241130102751512e-07,
      "loss": 0.7767,
      "step": 26150
    },
    {
      "epoch": 0.4071862813018129,
      "grad_norm": 3.557560920715332,
      "learning_rate": 6.240310712705462e-07,
      "loss": 0.7684,
      "step": 26155
    },
    {
      "epoch": 0.4072641223037823,
      "grad_norm": 6.114151954650879,
      "learning_rate": 6.239491322659412e-07,
      "loss": 0.7599,
      "step": 26160
    },
    {
      "epoch": 0.4073419633057517,
      "grad_norm": 5.147528171539307,
      "learning_rate": 6.238671932613362e-07,
      "loss": 0.7908,
      "step": 26165
    },
    {
      "epoch": 0.40741980430772107,
      "grad_norm": 3.184091091156006,
      "learning_rate": 6.237852542567312e-07,
      "loss": 0.6857,
      "step": 26170
    },
    {
      "epoch": 0.4074976453096904,
      "grad_norm": 6.086164474487305,
      "learning_rate": 6.237033152521263e-07,
      "loss": 0.8387,
      "step": 26175
    },
    {
      "epoch": 0.4075754863116598,
      "grad_norm": 2.5259017944335938,
      "learning_rate": 6.236213762475213e-07,
      "loss": 0.7188,
      "step": 26180
    },
    {
      "epoch": 0.4076533273136292,
      "grad_norm": 5.400747776031494,
      "learning_rate": 6.235394372429164e-07,
      "loss": 0.7216,
      "step": 26185
    },
    {
      "epoch": 0.4077311683155986,
      "grad_norm": 4.082566738128662,
      "learning_rate": 6.234574982383114e-07,
      "loss": 0.7062,
      "step": 26190
    },
    {
      "epoch": 0.4078090093175679,
      "grad_norm": 4.59686279296875,
      "learning_rate": 6.233755592337064e-07,
      "loss": 0.8353,
      "step": 26195
    },
    {
      "epoch": 0.4078868503195373,
      "grad_norm": 3.698218822479248,
      "learning_rate": 6.232936202291015e-07,
      "loss": 0.7123,
      "step": 26200
    },
    {
      "epoch": 0.4079646913215067,
      "grad_norm": 3.4450857639312744,
      "learning_rate": 6.232116812244964e-07,
      "loss": 0.7248,
      "step": 26205
    },
    {
      "epoch": 0.4080425323234761,
      "grad_norm": 4.156335353851318,
      "learning_rate": 6.231297422198915e-07,
      "loss": 0.7683,
      "step": 26210
    },
    {
      "epoch": 0.40812037332544543,
      "grad_norm": 7.560205459594727,
      "learning_rate": 6.230478032152865e-07,
      "loss": 0.8649,
      "step": 26215
    },
    {
      "epoch": 0.4081982143274148,
      "grad_norm": 5.544355869293213,
      "learning_rate": 6.229658642106815e-07,
      "loss": 0.6824,
      "step": 26220
    },
    {
      "epoch": 0.4082760553293842,
      "grad_norm": 2.8157401084899902,
      "learning_rate": 6.228839252060766e-07,
      "loss": 0.8322,
      "step": 26225
    },
    {
      "epoch": 0.40835389633135355,
      "grad_norm": 3.6721606254577637,
      "learning_rate": 6.228019862014717e-07,
      "loss": 0.7834,
      "step": 26230
    },
    {
      "epoch": 0.40843173733332294,
      "grad_norm": 2.750293731689453,
      "learning_rate": 6.227200471968666e-07,
      "loss": 0.7128,
      "step": 26235
    },
    {
      "epoch": 0.40850957833529233,
      "grad_norm": 6.421826362609863,
      "learning_rate": 6.226381081922617e-07,
      "loss": 0.7421,
      "step": 26240
    },
    {
      "epoch": 0.4085874193372617,
      "grad_norm": 3.734286069869995,
      "learning_rate": 6.225561691876567e-07,
      "loss": 0.7068,
      "step": 26245
    },
    {
      "epoch": 0.40866526033923106,
      "grad_norm": 3.162724494934082,
      "learning_rate": 6.224742301830516e-07,
      "loss": 0.7293,
      "step": 26250
    },
    {
      "epoch": 0.40874310134120045,
      "grad_norm": 6.575493335723877,
      "learning_rate": 6.223922911784467e-07,
      "loss": 0.6914,
      "step": 26255
    },
    {
      "epoch": 0.40882094234316985,
      "grad_norm": 5.713708877563477,
      "learning_rate": 6.223103521738418e-07,
      "loss": 0.7782,
      "step": 26260
    },
    {
      "epoch": 0.40889878334513924,
      "grad_norm": 3.8541054725646973,
      "learning_rate": 6.222284131692368e-07,
      "loss": 0.8174,
      "step": 26265
    },
    {
      "epoch": 0.4089766243471086,
      "grad_norm": 2.532059907913208,
      "learning_rate": 6.221464741646318e-07,
      "loss": 0.6114,
      "step": 26270
    },
    {
      "epoch": 0.40905446534907797,
      "grad_norm": 9.393692970275879,
      "learning_rate": 6.220645351600269e-07,
      "loss": 0.7823,
      "step": 26275
    },
    {
      "epoch": 0.40913230635104736,
      "grad_norm": 3.4701974391937256,
      "learning_rate": 6.219825961554219e-07,
      "loss": 0.7245,
      "step": 26280
    },
    {
      "epoch": 0.40921014735301675,
      "grad_norm": 9.034595489501953,
      "learning_rate": 6.219006571508168e-07,
      "loss": 0.8293,
      "step": 26285
    },
    {
      "epoch": 0.4092879883549861,
      "grad_norm": 6.221508026123047,
      "learning_rate": 6.218187181462119e-07,
      "loss": 0.7924,
      "step": 26290
    },
    {
      "epoch": 0.4093658293569555,
      "grad_norm": 3.99814772605896,
      "learning_rate": 6.217367791416069e-07,
      "loss": 0.7358,
      "step": 26295
    },
    {
      "epoch": 0.40944367035892487,
      "grad_norm": 2.8011677265167236,
      "learning_rate": 6.21654840137002e-07,
      "loss": 0.771,
      "step": 26300
    },
    {
      "epoch": 0.40952151136089426,
      "grad_norm": 4.2195515632629395,
      "learning_rate": 6.21572901132397e-07,
      "loss": 0.6896,
      "step": 26305
    },
    {
      "epoch": 0.4095993523628636,
      "grad_norm": 4.860198974609375,
      "learning_rate": 6.214909621277921e-07,
      "loss": 0.774,
      "step": 26310
    },
    {
      "epoch": 0.409677193364833,
      "grad_norm": 2.483884334564209,
      "learning_rate": 6.214090231231871e-07,
      "loss": 0.8224,
      "step": 26315
    },
    {
      "epoch": 0.4097550343668024,
      "grad_norm": 3.0667426586151123,
      "learning_rate": 6.213270841185822e-07,
      "loss": 0.7239,
      "step": 26320
    },
    {
      "epoch": 0.4098328753687717,
      "grad_norm": 6.193851470947266,
      "learning_rate": 6.212451451139771e-07,
      "loss": 0.7518,
      "step": 26325
    },
    {
      "epoch": 0.4099107163707411,
      "grad_norm": 3.35817551612854,
      "learning_rate": 6.211632061093721e-07,
      "loss": 0.8608,
      "step": 26330
    },
    {
      "epoch": 0.4099885573727105,
      "grad_norm": 4.190231800079346,
      "learning_rate": 6.210812671047672e-07,
      "loss": 0.8772,
      "step": 26335
    },
    {
      "epoch": 0.4100663983746799,
      "grad_norm": 3.7101292610168457,
      "learning_rate": 6.209993281001622e-07,
      "loss": 0.8175,
      "step": 26340
    },
    {
      "epoch": 0.41014423937664923,
      "grad_norm": 3.1132471561431885,
      "learning_rate": 6.209173890955572e-07,
      "loss": 0.7291,
      "step": 26345
    },
    {
      "epoch": 0.4102220803786186,
      "grad_norm": 4.467872142791748,
      "learning_rate": 6.208354500909523e-07,
      "loss": 0.8373,
      "step": 26350
    },
    {
      "epoch": 0.410299921380588,
      "grad_norm": 3.7462499141693115,
      "learning_rate": 6.207535110863474e-07,
      "loss": 0.7824,
      "step": 26355
    },
    {
      "epoch": 0.4103777623825574,
      "grad_norm": 3.6446940898895264,
      "learning_rate": 6.206715720817423e-07,
      "loss": 0.858,
      "step": 26360
    },
    {
      "epoch": 0.41045560338452675,
      "grad_norm": 4.626848220825195,
      "learning_rate": 6.205896330771373e-07,
      "loss": 0.8112,
      "step": 26365
    },
    {
      "epoch": 0.41053344438649614,
      "grad_norm": 12.126510620117188,
      "learning_rate": 6.205076940725324e-07,
      "loss": 0.6869,
      "step": 26370
    },
    {
      "epoch": 0.41061128538846553,
      "grad_norm": 2.845858335494995,
      "learning_rate": 6.204257550679273e-07,
      "loss": 0.6595,
      "step": 26375
    },
    {
      "epoch": 0.4106891263904349,
      "grad_norm": 6.0855712890625,
      "learning_rate": 6.203438160633224e-07,
      "loss": 0.8124,
      "step": 26380
    },
    {
      "epoch": 0.41076696739240426,
      "grad_norm": 3.1703717708587646,
      "learning_rate": 6.202618770587175e-07,
      "loss": 0.7564,
      "step": 26385
    },
    {
      "epoch": 0.41084480839437365,
      "grad_norm": 2.412677049636841,
      "learning_rate": 6.201799380541125e-07,
      "loss": 0.8151,
      "step": 26390
    },
    {
      "epoch": 0.41092264939634304,
      "grad_norm": 4.135179042816162,
      "learning_rate": 6.200979990495075e-07,
      "loss": 0.7567,
      "step": 26395
    },
    {
      "epoch": 0.41100049039831243,
      "grad_norm": 3.759422779083252,
      "learning_rate": 6.200160600449026e-07,
      "loss": 0.666,
      "step": 26400
    },
    {
      "epoch": 0.41107833140028177,
      "grad_norm": 4.042479038238525,
      "learning_rate": 6.199341210402975e-07,
      "loss": 0.7447,
      "step": 26405
    },
    {
      "epoch": 0.41115617240225116,
      "grad_norm": 3.1750473976135254,
      "learning_rate": 6.198521820356925e-07,
      "loss": 0.7239,
      "step": 26410
    },
    {
      "epoch": 0.41123401340422056,
      "grad_norm": 3.660675525665283,
      "learning_rate": 6.197702430310876e-07,
      "loss": 0.7659,
      "step": 26415
    },
    {
      "epoch": 0.4113118544061899,
      "grad_norm": 5.285322666168213,
      "learning_rate": 6.196883040264826e-07,
      "loss": 0.7527,
      "step": 26420
    },
    {
      "epoch": 0.4113896954081593,
      "grad_norm": 4.534871578216553,
      "learning_rate": 6.196063650218777e-07,
      "loss": 0.7233,
      "step": 26425
    },
    {
      "epoch": 0.4114675364101287,
      "grad_norm": 4.243113994598389,
      "learning_rate": 6.195244260172727e-07,
      "loss": 0.8813,
      "step": 26430
    },
    {
      "epoch": 0.41154537741209807,
      "grad_norm": 8.809892654418945,
      "learning_rate": 6.194424870126678e-07,
      "loss": 0.7059,
      "step": 26435
    },
    {
      "epoch": 0.4116232184140674,
      "grad_norm": 7.20442008972168,
      "learning_rate": 6.193605480080628e-07,
      "loss": 0.767,
      "step": 26440
    },
    {
      "epoch": 0.4117010594160368,
      "grad_norm": 6.816311836242676,
      "learning_rate": 6.192786090034579e-07,
      "loss": 0.727,
      "step": 26445
    },
    {
      "epoch": 0.4117789004180062,
      "grad_norm": 3.0180819034576416,
      "learning_rate": 6.191966699988528e-07,
      "loss": 0.6395,
      "step": 26450
    },
    {
      "epoch": 0.4118567414199756,
      "grad_norm": 8.097386360168457,
      "learning_rate": 6.191147309942478e-07,
      "loss": 0.8232,
      "step": 26455
    },
    {
      "epoch": 0.4119345824219449,
      "grad_norm": 2.4666662216186523,
      "learning_rate": 6.190327919896429e-07,
      "loss": 0.793,
      "step": 26460
    },
    {
      "epoch": 0.4120124234239143,
      "grad_norm": 2.839620351791382,
      "learning_rate": 6.18950852985038e-07,
      "loss": 0.7188,
      "step": 26465
    },
    {
      "epoch": 0.4120902644258837,
      "grad_norm": 4.023433685302734,
      "learning_rate": 6.188689139804329e-07,
      "loss": 0.719,
      "step": 26470
    },
    {
      "epoch": 0.4121681054278531,
      "grad_norm": 6.379566192626953,
      "learning_rate": 6.18786974975828e-07,
      "loss": 0.8544,
      "step": 26475
    },
    {
      "epoch": 0.41224594642982243,
      "grad_norm": 3.7102549076080322,
      "learning_rate": 6.187050359712231e-07,
      "loss": 0.7426,
      "step": 26480
    },
    {
      "epoch": 0.4123237874317918,
      "grad_norm": 4.392475128173828,
      "learning_rate": 6.18623096966618e-07,
      "loss": 0.8274,
      "step": 26485
    },
    {
      "epoch": 0.4124016284337612,
      "grad_norm": 3.0677952766418457,
      "learning_rate": 6.18541157962013e-07,
      "loss": 0.7238,
      "step": 26490
    },
    {
      "epoch": 0.41247946943573055,
      "grad_norm": 4.5266289710998535,
      "learning_rate": 6.184592189574081e-07,
      "loss": 0.7116,
      "step": 26495
    },
    {
      "epoch": 0.41255731043769994,
      "grad_norm": 7.1209025382995605,
      "learning_rate": 6.18377279952803e-07,
      "loss": 0.7334,
      "step": 26500
    },
    {
      "epoch": 0.41263515143966933,
      "grad_norm": 2.6258678436279297,
      "learning_rate": 6.182953409481981e-07,
      "loss": 0.6962,
      "step": 26505
    },
    {
      "epoch": 0.4127129924416387,
      "grad_norm": 2.9764554500579834,
      "learning_rate": 6.182134019435932e-07,
      "loss": 0.6064,
      "step": 26510
    },
    {
      "epoch": 0.41279083344360806,
      "grad_norm": 3.3296802043914795,
      "learning_rate": 6.181314629389882e-07,
      "loss": 0.7621,
      "step": 26515
    },
    {
      "epoch": 0.41286867444557745,
      "grad_norm": 4.500085830688477,
      "learning_rate": 6.180495239343832e-07,
      "loss": 0.5435,
      "step": 26520
    },
    {
      "epoch": 0.41294651544754685,
      "grad_norm": 3.5746519565582275,
      "learning_rate": 6.179675849297783e-07,
      "loss": 0.7156,
      "step": 26525
    },
    {
      "epoch": 0.41302435644951624,
      "grad_norm": 2.8794310092926025,
      "learning_rate": 6.178856459251732e-07,
      "loss": 0.7878,
      "step": 26530
    },
    {
      "epoch": 0.4131021974514856,
      "grad_norm": 3.6298744678497314,
      "learning_rate": 6.178037069205683e-07,
      "loss": 0.8602,
      "step": 26535
    },
    {
      "epoch": 0.41318003845345497,
      "grad_norm": 3.2164409160614014,
      "learning_rate": 6.177217679159633e-07,
      "loss": 0.756,
      "step": 26540
    },
    {
      "epoch": 0.41325787945542436,
      "grad_norm": 3.0077872276306152,
      "learning_rate": 6.176398289113583e-07,
      "loss": 0.87,
      "step": 26545
    },
    {
      "epoch": 0.41333572045739375,
      "grad_norm": 3.9495902061462402,
      "learning_rate": 6.175578899067534e-07,
      "loss": 0.7959,
      "step": 26550
    },
    {
      "epoch": 0.4134135614593631,
      "grad_norm": 4.911811351776123,
      "learning_rate": 6.174759509021484e-07,
      "loss": 0.7468,
      "step": 26555
    },
    {
      "epoch": 0.4134914024613325,
      "grad_norm": 4.713522434234619,
      "learning_rate": 6.173940118975435e-07,
      "loss": 0.7409,
      "step": 26560
    },
    {
      "epoch": 0.41356924346330187,
      "grad_norm": 3.8988125324249268,
      "learning_rate": 6.173120728929385e-07,
      "loss": 0.7385,
      "step": 26565
    },
    {
      "epoch": 0.41364708446527126,
      "grad_norm": 2.3970417976379395,
      "learning_rate": 6.172301338883335e-07,
      "loss": 0.8039,
      "step": 26570
    },
    {
      "epoch": 0.4137249254672406,
      "grad_norm": 2.4468183517456055,
      "learning_rate": 6.171481948837285e-07,
      "loss": 0.7449,
      "step": 26575
    },
    {
      "epoch": 0.41380276646921,
      "grad_norm": 3.3654592037200928,
      "learning_rate": 6.170662558791235e-07,
      "loss": 0.7903,
      "step": 26580
    },
    {
      "epoch": 0.4138806074711794,
      "grad_norm": 3.423818349838257,
      "learning_rate": 6.169843168745186e-07,
      "loss": 0.7963,
      "step": 26585
    },
    {
      "epoch": 0.4139584484731487,
      "grad_norm": 2.784715175628662,
      "learning_rate": 6.169023778699137e-07,
      "loss": 0.7938,
      "step": 26590
    },
    {
      "epoch": 0.4140362894751181,
      "grad_norm": 2.766166925430298,
      "learning_rate": 6.168204388653086e-07,
      "loss": 0.7048,
      "step": 26595
    },
    {
      "epoch": 0.4141141304770875,
      "grad_norm": 3.4769532680511475,
      "learning_rate": 6.167384998607037e-07,
      "loss": 0.7529,
      "step": 26600
    },
    {
      "epoch": 0.4141919714790569,
      "grad_norm": 3.1103856563568115,
      "learning_rate": 6.166565608560988e-07,
      "loss": 0.715,
      "step": 26605
    },
    {
      "epoch": 0.41426981248102623,
      "grad_norm": 3.8510234355926514,
      "learning_rate": 6.165746218514936e-07,
      "loss": 0.7761,
      "step": 26610
    },
    {
      "epoch": 0.4143476534829956,
      "grad_norm": 6.741697788238525,
      "learning_rate": 6.164926828468887e-07,
      "loss": 0.7113,
      "step": 26615
    },
    {
      "epoch": 0.414425494484965,
      "grad_norm": 2.8889408111572266,
      "learning_rate": 6.164107438422838e-07,
      "loss": 0.6265,
      "step": 26620
    },
    {
      "epoch": 0.4145033354869344,
      "grad_norm": 9.615153312683105,
      "learning_rate": 6.163288048376788e-07,
      "loss": 0.8047,
      "step": 26625
    },
    {
      "epoch": 0.41458117648890375,
      "grad_norm": 7.800364971160889,
      "learning_rate": 6.162468658330738e-07,
      "loss": 0.728,
      "step": 26630
    },
    {
      "epoch": 0.41465901749087314,
      "grad_norm": 2.9394280910491943,
      "learning_rate": 6.161649268284689e-07,
      "loss": 0.7267,
      "step": 26635
    },
    {
      "epoch": 0.41473685849284253,
      "grad_norm": 4.299229145050049,
      "learning_rate": 6.160829878238639e-07,
      "loss": 0.9,
      "step": 26640
    },
    {
      "epoch": 0.4148146994948119,
      "grad_norm": 4.3845672607421875,
      "learning_rate": 6.16001048819259e-07,
      "loss": 0.6087,
      "step": 26645
    },
    {
      "epoch": 0.41489254049678126,
      "grad_norm": 3.6776092052459717,
      "learning_rate": 6.159191098146539e-07,
      "loss": 0.6703,
      "step": 26650
    },
    {
      "epoch": 0.41497038149875065,
      "grad_norm": 5.197491645812988,
      "learning_rate": 6.158371708100489e-07,
      "loss": 0.7332,
      "step": 26655
    },
    {
      "epoch": 0.41504822250072004,
      "grad_norm": 3.1170854568481445,
      "learning_rate": 6.15755231805444e-07,
      "loss": 0.6487,
      "step": 26660
    },
    {
      "epoch": 0.4151260635026894,
      "grad_norm": 4.167092323303223,
      "learning_rate": 6.15673292800839e-07,
      "loss": 0.8017,
      "step": 26665
    },
    {
      "epoch": 0.41520390450465877,
      "grad_norm": 4.455737113952637,
      "learning_rate": 6.155913537962341e-07,
      "loss": 0.767,
      "step": 26670
    },
    {
      "epoch": 0.41528174550662816,
      "grad_norm": 5.93499755859375,
      "learning_rate": 6.155094147916291e-07,
      "loss": 0.8527,
      "step": 26675
    },
    {
      "epoch": 0.41535958650859756,
      "grad_norm": 3.7191712856292725,
      "learning_rate": 6.154274757870242e-07,
      "loss": 0.8184,
      "step": 26680
    },
    {
      "epoch": 0.4154374275105669,
      "grad_norm": 6.877071380615234,
      "learning_rate": 6.153455367824192e-07,
      "loss": 0.7743,
      "step": 26685
    },
    {
      "epoch": 0.4155152685125363,
      "grad_norm": 7.615885257720947,
      "learning_rate": 6.152635977778141e-07,
      "loss": 0.9191,
      "step": 26690
    },
    {
      "epoch": 0.4155931095145057,
      "grad_norm": 3.3617007732391357,
      "learning_rate": 6.151816587732092e-07,
      "loss": 0.7714,
      "step": 26695
    },
    {
      "epoch": 0.41567095051647507,
      "grad_norm": 2.519935369491577,
      "learning_rate": 6.150997197686042e-07,
      "loss": 0.765,
      "step": 26700
    },
    {
      "epoch": 0.4157487915184444,
      "grad_norm": 3.2422094345092773,
      "learning_rate": 6.150177807639992e-07,
      "loss": 0.7896,
      "step": 26705
    },
    {
      "epoch": 0.4158266325204138,
      "grad_norm": 2.624389410018921,
      "learning_rate": 6.149358417593943e-07,
      "loss": 0.6363,
      "step": 26710
    },
    {
      "epoch": 0.4159044735223832,
      "grad_norm": 3.49373197555542,
      "learning_rate": 6.148539027547894e-07,
      "loss": 0.8335,
      "step": 26715
    },
    {
      "epoch": 0.4159823145243526,
      "grad_norm": 4.526536464691162,
      "learning_rate": 6.147719637501843e-07,
      "loss": 0.7632,
      "step": 26720
    },
    {
      "epoch": 0.4160601555263219,
      "grad_norm": 7.346248149871826,
      "learning_rate": 6.146900247455794e-07,
      "loss": 0.6511,
      "step": 26725
    },
    {
      "epoch": 0.4161379965282913,
      "grad_norm": 6.548068523406982,
      "learning_rate": 6.146080857409744e-07,
      "loss": 0.7563,
      "step": 26730
    },
    {
      "epoch": 0.4162158375302607,
      "grad_norm": 3.753570079803467,
      "learning_rate": 6.145261467363693e-07,
      "loss": 0.7684,
      "step": 26735
    },
    {
      "epoch": 0.4162936785322301,
      "grad_norm": 8.335871696472168,
      "learning_rate": 6.144442077317644e-07,
      "loss": 0.8658,
      "step": 26740
    },
    {
      "epoch": 0.41637151953419943,
      "grad_norm": 4.141759395599365,
      "learning_rate": 6.143622687271595e-07,
      "loss": 0.8941,
      "step": 26745
    },
    {
      "epoch": 0.4164493605361688,
      "grad_norm": 5.216050148010254,
      "learning_rate": 6.142803297225545e-07,
      "loss": 0.7769,
      "step": 26750
    },
    {
      "epoch": 0.4165272015381382,
      "grad_norm": 3.003042697906494,
      "learning_rate": 6.141983907179495e-07,
      "loss": 0.6379,
      "step": 26755
    },
    {
      "epoch": 0.41660504254010755,
      "grad_norm": 5.297163963317871,
      "learning_rate": 6.141164517133446e-07,
      "loss": 0.8346,
      "step": 26760
    },
    {
      "epoch": 0.41668288354207694,
      "grad_norm": 6.788291931152344,
      "learning_rate": 6.140345127087396e-07,
      "loss": 0.8368,
      "step": 26765
    },
    {
      "epoch": 0.41676072454404633,
      "grad_norm": 13.022028923034668,
      "learning_rate": 6.139525737041347e-07,
      "loss": 0.8212,
      "step": 26770
    },
    {
      "epoch": 0.4168385655460157,
      "grad_norm": 3.262367010116577,
      "learning_rate": 6.138706346995296e-07,
      "loss": 0.6826,
      "step": 26775
    },
    {
      "epoch": 0.41691640654798506,
      "grad_norm": 3.736050844192505,
      "learning_rate": 6.137886956949246e-07,
      "loss": 0.7972,
      "step": 26780
    },
    {
      "epoch": 0.41699424754995446,
      "grad_norm": 8.119766235351562,
      "learning_rate": 6.137067566903197e-07,
      "loss": 0.739,
      "step": 26785
    },
    {
      "epoch": 0.41707208855192385,
      "grad_norm": 4.275023460388184,
      "learning_rate": 6.136248176857147e-07,
      "loss": 0.8204,
      "step": 26790
    },
    {
      "epoch": 0.41714992955389324,
      "grad_norm": 3.132134437561035,
      "learning_rate": 6.135428786811098e-07,
      "loss": 0.7618,
      "step": 26795
    },
    {
      "epoch": 0.4172277705558626,
      "grad_norm": 3.638437271118164,
      "learning_rate": 6.134609396765048e-07,
      "loss": 0.7926,
      "step": 26800
    },
    {
      "epoch": 0.41730561155783197,
      "grad_norm": 3.03761625289917,
      "learning_rate": 6.133790006718999e-07,
      "loss": 0.7097,
      "step": 26805
    },
    {
      "epoch": 0.41738345255980136,
      "grad_norm": 5.049458026885986,
      "learning_rate": 6.132970616672949e-07,
      "loss": 0.7085,
      "step": 26810
    },
    {
      "epoch": 0.41746129356177075,
      "grad_norm": 3.4833250045776367,
      "learning_rate": 6.132151226626898e-07,
      "loss": 0.7813,
      "step": 26815
    },
    {
      "epoch": 0.4175391345637401,
      "grad_norm": 3.814908027648926,
      "learning_rate": 6.131331836580849e-07,
      "loss": 0.7807,
      "step": 26820
    },
    {
      "epoch": 0.4176169755657095,
      "grad_norm": 3.30277419090271,
      "learning_rate": 6.1305124465348e-07,
      "loss": 0.7181,
      "step": 26825
    },
    {
      "epoch": 0.4176948165676789,
      "grad_norm": 3.3809800148010254,
      "learning_rate": 6.129693056488749e-07,
      "loss": 0.7821,
      "step": 26830
    },
    {
      "epoch": 0.4177726575696482,
      "grad_norm": 7.558841705322266,
      "learning_rate": 6.1288736664427e-07,
      "loss": 0.6552,
      "step": 26835
    },
    {
      "epoch": 0.4178504985716176,
      "grad_norm": 5.2530975341796875,
      "learning_rate": 6.128054276396651e-07,
      "loss": 0.7189,
      "step": 26840
    },
    {
      "epoch": 0.417928339573587,
      "grad_norm": 6.695279121398926,
      "learning_rate": 6.1272348863506e-07,
      "loss": 0.6808,
      "step": 26845
    },
    {
      "epoch": 0.4180061805755564,
      "grad_norm": 4.834892749786377,
      "learning_rate": 6.126415496304551e-07,
      "loss": 0.7211,
      "step": 26850
    },
    {
      "epoch": 0.4180840215775257,
      "grad_norm": 4.291613578796387,
      "learning_rate": 6.125596106258501e-07,
      "loss": 0.7151,
      "step": 26855
    },
    {
      "epoch": 0.4181618625794951,
      "grad_norm": 6.921629428863525,
      "learning_rate": 6.12477671621245e-07,
      "loss": 0.811,
      "step": 26860
    },
    {
      "epoch": 0.4182397035814645,
      "grad_norm": 3.956561326980591,
      "learning_rate": 6.123957326166401e-07,
      "loss": 0.782,
      "step": 26865
    },
    {
      "epoch": 0.4183175445834339,
      "grad_norm": 2.7356879711151123,
      "learning_rate": 6.123137936120352e-07,
      "loss": 0.6748,
      "step": 26870
    },
    {
      "epoch": 0.41839538558540323,
      "grad_norm": 3.5307228565216064,
      "learning_rate": 6.122318546074302e-07,
      "loss": 0.7285,
      "step": 26875
    },
    {
      "epoch": 0.4184732265873726,
      "grad_norm": 3.408752679824829,
      "learning_rate": 6.121499156028252e-07,
      "loss": 0.6309,
      "step": 26880
    },
    {
      "epoch": 0.418551067589342,
      "grad_norm": 3.754222869873047,
      "learning_rate": 6.120679765982203e-07,
      "loss": 0.721,
      "step": 26885
    },
    {
      "epoch": 0.4186289085913114,
      "grad_norm": 3.735696315765381,
      "learning_rate": 6.119860375936153e-07,
      "loss": 0.7713,
      "step": 26890
    },
    {
      "epoch": 0.41870674959328075,
      "grad_norm": 4.097355365753174,
      "learning_rate": 6.119040985890103e-07,
      "loss": 0.7273,
      "step": 26895
    },
    {
      "epoch": 0.41878459059525014,
      "grad_norm": 9.404135704040527,
      "learning_rate": 6.118221595844053e-07,
      "loss": 0.8026,
      "step": 26900
    },
    {
      "epoch": 0.41886243159721953,
      "grad_norm": 4.844088077545166,
      "learning_rate": 6.117402205798003e-07,
      "loss": 0.8528,
      "step": 26905
    },
    {
      "epoch": 0.4189402725991889,
      "grad_norm": 3.7302608489990234,
      "learning_rate": 6.116582815751954e-07,
      "loss": 0.8019,
      "step": 26910
    },
    {
      "epoch": 0.41901811360115826,
      "grad_norm": 5.881189823150635,
      "learning_rate": 6.115763425705905e-07,
      "loss": 0.6222,
      "step": 26915
    },
    {
      "epoch": 0.41909595460312765,
      "grad_norm": 2.738311529159546,
      "learning_rate": 6.114944035659855e-07,
      "loss": 0.6529,
      "step": 26920
    },
    {
      "epoch": 0.41917379560509704,
      "grad_norm": 3.3108832836151123,
      "learning_rate": 6.114124645613805e-07,
      "loss": 0.7377,
      "step": 26925
    },
    {
      "epoch": 0.4192516366070664,
      "grad_norm": 3.1821160316467285,
      "learning_rate": 6.113305255567756e-07,
      "loss": 0.7788,
      "step": 26930
    },
    {
      "epoch": 0.4193294776090358,
      "grad_norm": 2.7442448139190674,
      "learning_rate": 6.112485865521705e-07,
      "loss": 0.7053,
      "step": 26935
    },
    {
      "epoch": 0.41940731861100516,
      "grad_norm": 2.8890326023101807,
      "learning_rate": 6.111666475475655e-07,
      "loss": 0.7762,
      "step": 26940
    },
    {
      "epoch": 0.41948515961297456,
      "grad_norm": 4.088584899902344,
      "learning_rate": 6.110847085429606e-07,
      "loss": 0.8523,
      "step": 26945
    },
    {
      "epoch": 0.4195630006149439,
      "grad_norm": 6.8917341232299805,
      "learning_rate": 6.110027695383557e-07,
      "loss": 0.8087,
      "step": 26950
    },
    {
      "epoch": 0.4196408416169133,
      "grad_norm": 5.052600383758545,
      "learning_rate": 6.109208305337506e-07,
      "loss": 0.7649,
      "step": 26955
    },
    {
      "epoch": 0.4197186826188827,
      "grad_norm": 6.6367621421813965,
      "learning_rate": 6.108388915291457e-07,
      "loss": 0.7778,
      "step": 26960
    },
    {
      "epoch": 0.41979652362085207,
      "grad_norm": 7.393866539001465,
      "learning_rate": 6.107569525245408e-07,
      "loss": 0.6788,
      "step": 26965
    },
    {
      "epoch": 0.4198743646228214,
      "grad_norm": 11.517690658569336,
      "learning_rate": 6.106750135199357e-07,
      "loss": 0.7605,
      "step": 26970
    },
    {
      "epoch": 0.4199522056247908,
      "grad_norm": 5.074392795562744,
      "learning_rate": 6.105930745153307e-07,
      "loss": 0.6848,
      "step": 26975
    },
    {
      "epoch": 0.4200300466267602,
      "grad_norm": 6.310015678405762,
      "learning_rate": 6.105111355107258e-07,
      "loss": 0.8061,
      "step": 26980
    },
    {
      "epoch": 0.4201078876287296,
      "grad_norm": 3.8176333904266357,
      "learning_rate": 6.104291965061208e-07,
      "loss": 0.8268,
      "step": 26985
    },
    {
      "epoch": 0.4201857286306989,
      "grad_norm": 3.844820737838745,
      "learning_rate": 6.103472575015158e-07,
      "loss": 0.6873,
      "step": 26990
    },
    {
      "epoch": 0.4202635696326683,
      "grad_norm": 5.099147796630859,
      "learning_rate": 6.102653184969109e-07,
      "loss": 0.6911,
      "step": 26995
    },
    {
      "epoch": 0.4203414106346377,
      "grad_norm": 4.644072532653809,
      "learning_rate": 6.101833794923059e-07,
      "loss": 0.7186,
      "step": 27000
    },
    {
      "epoch": 0.42041925163660704,
      "grad_norm": 5.660889625549316,
      "learning_rate": 6.10101440487701e-07,
      "loss": 0.8471,
      "step": 27005
    },
    {
      "epoch": 0.42049709263857643,
      "grad_norm": 3.6035196781158447,
      "learning_rate": 6.10019501483096e-07,
      "loss": 0.7778,
      "step": 27010
    },
    {
      "epoch": 0.4205749336405458,
      "grad_norm": 3.680065393447876,
      "learning_rate": 6.099375624784909e-07,
      "loss": 0.8525,
      "step": 27015
    },
    {
      "epoch": 0.4206527746425152,
      "grad_norm": 4.620081901550293,
      "learning_rate": 6.09855623473886e-07,
      "loss": 0.7906,
      "step": 27020
    },
    {
      "epoch": 0.42073061564448455,
      "grad_norm": 2.889920473098755,
      "learning_rate": 6.09773684469281e-07,
      "loss": 0.7267,
      "step": 27025
    },
    {
      "epoch": 0.42080845664645394,
      "grad_norm": 5.654252052307129,
      "learning_rate": 6.09691745464676e-07,
      "loss": 0.7148,
      "step": 27030
    },
    {
      "epoch": 0.42088629764842334,
      "grad_norm": 6.712372303009033,
      "learning_rate": 6.096098064600711e-07,
      "loss": 0.7575,
      "step": 27035
    },
    {
      "epoch": 0.4209641386503927,
      "grad_norm": 2.887517213821411,
      "learning_rate": 6.095278674554662e-07,
      "loss": 0.7287,
      "step": 27040
    },
    {
      "epoch": 0.42104197965236206,
      "grad_norm": 3.611146926879883,
      "learning_rate": 6.094459284508612e-07,
      "loss": 0.8218,
      "step": 27045
    },
    {
      "epoch": 0.42111982065433146,
      "grad_norm": 3.5884053707122803,
      "learning_rate": 6.093639894462562e-07,
      "loss": 0.7259,
      "step": 27050
    },
    {
      "epoch": 0.42119766165630085,
      "grad_norm": 6.537292003631592,
      "learning_rate": 6.092820504416512e-07,
      "loss": 0.7751,
      "step": 27055
    },
    {
      "epoch": 0.42127550265827024,
      "grad_norm": 3.2337849140167236,
      "learning_rate": 6.092001114370462e-07,
      "loss": 0.722,
      "step": 27060
    },
    {
      "epoch": 0.4213533436602396,
      "grad_norm": 3.9660308361053467,
      "learning_rate": 6.091181724324412e-07,
      "loss": 0.7699,
      "step": 27065
    },
    {
      "epoch": 0.42143118466220897,
      "grad_norm": 3.0891098976135254,
      "learning_rate": 6.090362334278363e-07,
      "loss": 0.8119,
      "step": 27070
    },
    {
      "epoch": 0.42150902566417836,
      "grad_norm": 4.106495380401611,
      "learning_rate": 6.089542944232314e-07,
      "loss": 0.7763,
      "step": 27075
    },
    {
      "epoch": 0.42158686666614775,
      "grad_norm": 3.0975704193115234,
      "learning_rate": 6.088723554186263e-07,
      "loss": 0.819,
      "step": 27080
    },
    {
      "epoch": 0.4216647076681171,
      "grad_norm": 3.7811708450317383,
      "learning_rate": 6.087904164140214e-07,
      "loss": 0.8506,
      "step": 27085
    },
    {
      "epoch": 0.4217425486700865,
      "grad_norm": 4.70298957824707,
      "learning_rate": 6.087084774094165e-07,
      "loss": 0.8322,
      "step": 27090
    },
    {
      "epoch": 0.4218203896720559,
      "grad_norm": 2.7160446643829346,
      "learning_rate": 6.086265384048115e-07,
      "loss": 0.7179,
      "step": 27095
    },
    {
      "epoch": 0.4218982306740252,
      "grad_norm": 5.256832122802734,
      "learning_rate": 6.085445994002064e-07,
      "loss": 0.7015,
      "step": 27100
    },
    {
      "epoch": 0.4219760716759946,
      "grad_norm": 6.849483013153076,
      "learning_rate": 6.084626603956015e-07,
      "loss": 0.7913,
      "step": 27105
    },
    {
      "epoch": 0.422053912677964,
      "grad_norm": 2.9917688369750977,
      "learning_rate": 6.083807213909965e-07,
      "loss": 0.7566,
      "step": 27110
    },
    {
      "epoch": 0.4221317536799334,
      "grad_norm": 3.3257033824920654,
      "learning_rate": 6.082987823863915e-07,
      "loss": 0.6861,
      "step": 27115
    },
    {
      "epoch": 0.4222095946819027,
      "grad_norm": 3.04402232170105,
      "learning_rate": 6.082168433817866e-07,
      "loss": 0.779,
      "step": 27120
    },
    {
      "epoch": 0.4222874356838721,
      "grad_norm": 2.7107558250427246,
      "learning_rate": 6.081349043771816e-07,
      "loss": 0.7857,
      "step": 27125
    },
    {
      "epoch": 0.4223652766858415,
      "grad_norm": 8.305509567260742,
      "learning_rate": 6.080529653725767e-07,
      "loss": 0.8263,
      "step": 27130
    },
    {
      "epoch": 0.4224431176878109,
      "grad_norm": 3.3207385540008545,
      "learning_rate": 6.079710263679717e-07,
      "loss": 0.7337,
      "step": 27135
    },
    {
      "epoch": 0.42252095868978023,
      "grad_norm": 5.155167102813721,
      "learning_rate": 6.078890873633666e-07,
      "loss": 0.8061,
      "step": 27140
    },
    {
      "epoch": 0.4225987996917496,
      "grad_norm": 7.670403003692627,
      "learning_rate": 6.078071483587617e-07,
      "loss": 0.7951,
      "step": 27145
    },
    {
      "epoch": 0.422676640693719,
      "grad_norm": 3.5299322605133057,
      "learning_rate": 6.077252093541567e-07,
      "loss": 0.7142,
      "step": 27150
    },
    {
      "epoch": 0.4227544816956884,
      "grad_norm": 2.901541233062744,
      "learning_rate": 6.076432703495517e-07,
      "loss": 0.7646,
      "step": 27155
    },
    {
      "epoch": 0.42283232269765775,
      "grad_norm": 2.260049343109131,
      "learning_rate": 6.075613313449468e-07,
      "loss": 0.7741,
      "step": 27160
    },
    {
      "epoch": 0.42291016369962714,
      "grad_norm": 2.8874802589416504,
      "learning_rate": 6.074793923403419e-07,
      "loss": 0.6199,
      "step": 27165
    },
    {
      "epoch": 0.42298800470159653,
      "grad_norm": 5.7339301109313965,
      "learning_rate": 6.073974533357369e-07,
      "loss": 0.7064,
      "step": 27170
    },
    {
      "epoch": 0.4230658457035659,
      "grad_norm": 3.6913132667541504,
      "learning_rate": 6.073155143311319e-07,
      "loss": 0.8068,
      "step": 27175
    },
    {
      "epoch": 0.42314368670553526,
      "grad_norm": 3.6943602561950684,
      "learning_rate": 6.072335753265269e-07,
      "loss": 0.804,
      "step": 27180
    },
    {
      "epoch": 0.42322152770750465,
      "grad_norm": 3.9005420207977295,
      "learning_rate": 6.07151636321922e-07,
      "loss": 0.6257,
      "step": 27185
    },
    {
      "epoch": 0.42329936870947404,
      "grad_norm": 5.157369613647461,
      "learning_rate": 6.070696973173169e-07,
      "loss": 0.7783,
      "step": 27190
    },
    {
      "epoch": 0.4233772097114434,
      "grad_norm": 7.655777931213379,
      "learning_rate": 6.06987758312712e-07,
      "loss": 0.7452,
      "step": 27195
    },
    {
      "epoch": 0.4234550507134128,
      "grad_norm": 5.108643054962158,
      "learning_rate": 6.069058193081071e-07,
      "loss": 0.6303,
      "step": 27200
    },
    {
      "epoch": 0.42353289171538216,
      "grad_norm": 3.364788293838501,
      "learning_rate": 6.06823880303502e-07,
      "loss": 0.7605,
      "step": 27205
    },
    {
      "epoch": 0.42361073271735156,
      "grad_norm": 7.872039794921875,
      "learning_rate": 6.067419412988971e-07,
      "loss": 0.7943,
      "step": 27210
    },
    {
      "epoch": 0.4236885737193209,
      "grad_norm": 3.3262274265289307,
      "learning_rate": 6.066600022942922e-07,
      "loss": 0.7413,
      "step": 27215
    },
    {
      "epoch": 0.4237664147212903,
      "grad_norm": 6.564119815826416,
      "learning_rate": 6.06578063289687e-07,
      "loss": 0.751,
      "step": 27220
    },
    {
      "epoch": 0.4238442557232597,
      "grad_norm": 3.2822859287261963,
      "learning_rate": 6.064961242850821e-07,
      "loss": 0.7151,
      "step": 27225
    },
    {
      "epoch": 0.42392209672522907,
      "grad_norm": 3.66558575630188,
      "learning_rate": 6.064141852804772e-07,
      "loss": 0.7304,
      "step": 27230
    },
    {
      "epoch": 0.4239999377271984,
      "grad_norm": 4.87346887588501,
      "learning_rate": 6.063322462758722e-07,
      "loss": 0.6916,
      "step": 27235
    },
    {
      "epoch": 0.4240777787291678,
      "grad_norm": 6.911497592926025,
      "learning_rate": 6.062503072712672e-07,
      "loss": 0.8004,
      "step": 27240
    },
    {
      "epoch": 0.4241556197311372,
      "grad_norm": 6.917334079742432,
      "learning_rate": 6.061683682666623e-07,
      "loss": 0.8152,
      "step": 27245
    },
    {
      "epoch": 0.4242334607331066,
      "grad_norm": 4.561256408691406,
      "learning_rate": 6.060864292620573e-07,
      "loss": 0.6921,
      "step": 27250
    },
    {
      "epoch": 0.4243113017350759,
      "grad_norm": 3.4769482612609863,
      "learning_rate": 6.060044902574524e-07,
      "loss": 0.6964,
      "step": 27255
    },
    {
      "epoch": 0.4243891427370453,
      "grad_norm": 6.042318344116211,
      "learning_rate": 6.059225512528473e-07,
      "loss": 0.6912,
      "step": 27260
    },
    {
      "epoch": 0.4244669837390147,
      "grad_norm": 10.03291130065918,
      "learning_rate": 6.058406122482423e-07,
      "loss": 0.7412,
      "step": 27265
    },
    {
      "epoch": 0.42454482474098404,
      "grad_norm": 4.805675983428955,
      "learning_rate": 6.057586732436374e-07,
      "loss": 0.6488,
      "step": 27270
    },
    {
      "epoch": 0.42462266574295343,
      "grad_norm": 3.6138381958007812,
      "learning_rate": 6.056767342390325e-07,
      "loss": 0.6861,
      "step": 27275
    },
    {
      "epoch": 0.4247005067449228,
      "grad_norm": 3.0130345821380615,
      "learning_rate": 6.055947952344274e-07,
      "loss": 0.7606,
      "step": 27280
    },
    {
      "epoch": 0.4247783477468922,
      "grad_norm": 7.257308006286621,
      "learning_rate": 6.055128562298225e-07,
      "loss": 0.724,
      "step": 27285
    },
    {
      "epoch": 0.42485618874886155,
      "grad_norm": 3.798450469970703,
      "learning_rate": 6.054309172252176e-07,
      "loss": 0.8094,
      "step": 27290
    },
    {
      "epoch": 0.42493402975083094,
      "grad_norm": 4.286552429199219,
      "learning_rate": 6.053489782206126e-07,
      "loss": 0.7206,
      "step": 27295
    },
    {
      "epoch": 0.42501187075280034,
      "grad_norm": 7.024591445922852,
      "learning_rate": 6.052670392160075e-07,
      "loss": 0.693,
      "step": 27300
    },
    {
      "epoch": 0.42508971175476973,
      "grad_norm": 2.7912750244140625,
      "learning_rate": 6.051851002114026e-07,
      "loss": 0.7163,
      "step": 27305
    },
    {
      "epoch": 0.42516755275673906,
      "grad_norm": 4.904073715209961,
      "learning_rate": 6.051031612067977e-07,
      "loss": 0.7379,
      "step": 27310
    },
    {
      "epoch": 0.42524539375870846,
      "grad_norm": 4.198714256286621,
      "learning_rate": 6.050212222021926e-07,
      "loss": 0.6968,
      "step": 27315
    },
    {
      "epoch": 0.42532323476067785,
      "grad_norm": 8.80378532409668,
      "learning_rate": 6.049392831975877e-07,
      "loss": 0.7086,
      "step": 27320
    },
    {
      "epoch": 0.42540107576264724,
      "grad_norm": 2.636580467224121,
      "learning_rate": 6.048573441929828e-07,
      "loss": 0.6348,
      "step": 27325
    },
    {
      "epoch": 0.4254789167646166,
      "grad_norm": 2.9006733894348145,
      "learning_rate": 6.047754051883777e-07,
      "loss": 0.7363,
      "step": 27330
    },
    {
      "epoch": 0.42555675776658597,
      "grad_norm": 10.586029052734375,
      "learning_rate": 6.046934661837728e-07,
      "loss": 0.8535,
      "step": 27335
    },
    {
      "epoch": 0.42563459876855536,
      "grad_norm": 7.683661937713623,
      "learning_rate": 6.046115271791678e-07,
      "loss": 0.6484,
      "step": 27340
    },
    {
      "epoch": 0.42571243977052475,
      "grad_norm": 4.323159217834473,
      "learning_rate": 6.045295881745628e-07,
      "loss": 0.8111,
      "step": 27345
    },
    {
      "epoch": 0.4257902807724941,
      "grad_norm": 3.4367847442626953,
      "learning_rate": 6.044476491699578e-07,
      "loss": 0.7728,
      "step": 27350
    },
    {
      "epoch": 0.4258681217744635,
      "grad_norm": 4.658853530883789,
      "learning_rate": 6.043657101653529e-07,
      "loss": 0.8023,
      "step": 27355
    },
    {
      "epoch": 0.4259459627764329,
      "grad_norm": 4.2418036460876465,
      "learning_rate": 6.042837711607479e-07,
      "loss": 0.8127,
      "step": 27360
    },
    {
      "epoch": 0.4260238037784022,
      "grad_norm": 5.615492343902588,
      "learning_rate": 6.04201832156143e-07,
      "loss": 0.7324,
      "step": 27365
    },
    {
      "epoch": 0.4261016447803716,
      "grad_norm": 5.688205718994141,
      "learning_rate": 6.04119893151538e-07,
      "loss": 0.7359,
      "step": 27370
    },
    {
      "epoch": 0.426179485782341,
      "grad_norm": 2.9322752952575684,
      "learning_rate": 6.04037954146933e-07,
      "loss": 0.7647,
      "step": 27375
    },
    {
      "epoch": 0.4262573267843104,
      "grad_norm": 4.038652420043945,
      "learning_rate": 6.039560151423281e-07,
      "loss": 0.7031,
      "step": 27380
    },
    {
      "epoch": 0.4263351677862797,
      "grad_norm": 3.9757728576660156,
      "learning_rate": 6.03874076137723e-07,
      "loss": 0.7481,
      "step": 27385
    },
    {
      "epoch": 0.4264130087882491,
      "grad_norm": 3.1250228881835938,
      "learning_rate": 6.03792137133118e-07,
      "loss": 0.8038,
      "step": 27390
    },
    {
      "epoch": 0.4264908497902185,
      "grad_norm": 3.7392306327819824,
      "learning_rate": 6.037101981285131e-07,
      "loss": 0.7122,
      "step": 27395
    },
    {
      "epoch": 0.4265686907921879,
      "grad_norm": 3.21966814994812,
      "learning_rate": 6.036282591239082e-07,
      "loss": 0.7205,
      "step": 27400
    },
    {
      "epoch": 0.42664653179415724,
      "grad_norm": 3.8706374168395996,
      "learning_rate": 6.035463201193031e-07,
      "loss": 0.8379,
      "step": 27405
    },
    {
      "epoch": 0.4267243727961266,
      "grad_norm": 3.5687901973724365,
      "learning_rate": 6.034643811146982e-07,
      "loss": 0.7195,
      "step": 27410
    },
    {
      "epoch": 0.426802213798096,
      "grad_norm": 3.3177990913391113,
      "learning_rate": 6.033824421100933e-07,
      "loss": 0.7446,
      "step": 27415
    },
    {
      "epoch": 0.4268800548000654,
      "grad_norm": 3.827878475189209,
      "learning_rate": 6.033005031054884e-07,
      "loss": 0.8047,
      "step": 27420
    },
    {
      "epoch": 0.42695789580203475,
      "grad_norm": 5.776741027832031,
      "learning_rate": 6.032185641008832e-07,
      "loss": 0.7703,
      "step": 27425
    },
    {
      "epoch": 0.42703573680400414,
      "grad_norm": 2.8187882900238037,
      "learning_rate": 6.031366250962783e-07,
      "loss": 0.6875,
      "step": 27430
    },
    {
      "epoch": 0.42711357780597353,
      "grad_norm": 3.621317148208618,
      "learning_rate": 6.030546860916734e-07,
      "loss": 0.8284,
      "step": 27435
    },
    {
      "epoch": 0.42719141880794287,
      "grad_norm": 8.660701751708984,
      "learning_rate": 6.029727470870683e-07,
      "loss": 0.877,
      "step": 27440
    },
    {
      "epoch": 0.42726925980991226,
      "grad_norm": 4.218385696411133,
      "learning_rate": 6.028908080824634e-07,
      "loss": 0.7182,
      "step": 27445
    },
    {
      "epoch": 0.42734710081188165,
      "grad_norm": 3.3534364700317383,
      "learning_rate": 6.028088690778585e-07,
      "loss": 0.7942,
      "step": 27450
    },
    {
      "epoch": 0.42742494181385104,
      "grad_norm": 7.34171724319458,
      "learning_rate": 6.027269300732535e-07,
      "loss": 0.7839,
      "step": 27455
    },
    {
      "epoch": 0.4275027828158204,
      "grad_norm": 4.382108211517334,
      "learning_rate": 6.026449910686485e-07,
      "loss": 0.7193,
      "step": 27460
    },
    {
      "epoch": 0.4275806238177898,
      "grad_norm": 9.386519432067871,
      "learning_rate": 6.025630520640435e-07,
      "loss": 0.7651,
      "step": 27465
    },
    {
      "epoch": 0.42765846481975917,
      "grad_norm": 5.13021183013916,
      "learning_rate": 6.024811130594385e-07,
      "loss": 0.7346,
      "step": 27470
    },
    {
      "epoch": 0.42773630582172856,
      "grad_norm": 3.906785249710083,
      "learning_rate": 6.023991740548335e-07,
      "loss": 0.9051,
      "step": 27475
    },
    {
      "epoch": 0.4278141468236979,
      "grad_norm": 3.4074766635894775,
      "learning_rate": 6.023172350502286e-07,
      "loss": 0.633,
      "step": 27480
    },
    {
      "epoch": 0.4278919878256673,
      "grad_norm": 3.513617753982544,
      "learning_rate": 6.022352960456236e-07,
      "loss": 0.7182,
      "step": 27485
    },
    {
      "epoch": 0.4279698288276367,
      "grad_norm": 3.953442335128784,
      "learning_rate": 6.021533570410187e-07,
      "loss": 0.8214,
      "step": 27490
    },
    {
      "epoch": 0.42804766982960607,
      "grad_norm": 2.662166118621826,
      "learning_rate": 6.020714180364137e-07,
      "loss": 0.809,
      "step": 27495
    },
    {
      "epoch": 0.4281255108315754,
      "grad_norm": 3.2333011627197266,
      "learning_rate": 6.019894790318087e-07,
      "loss": 0.8559,
      "step": 27500
    },
    {
      "epoch": 0.4282033518335448,
      "grad_norm": 4.552109718322754,
      "learning_rate": 6.019075400272037e-07,
      "loss": 0.8152,
      "step": 27505
    },
    {
      "epoch": 0.4282811928355142,
      "grad_norm": 3.784883737564087,
      "learning_rate": 6.018256010225987e-07,
      "loss": 0.6774,
      "step": 27510
    },
    {
      "epoch": 0.4283590338374836,
      "grad_norm": 3.5555851459503174,
      "learning_rate": 6.017436620179937e-07,
      "loss": 0.7077,
      "step": 27515
    },
    {
      "epoch": 0.4284368748394529,
      "grad_norm": 3.111800193786621,
      "learning_rate": 6.016617230133888e-07,
      "loss": 0.742,
      "step": 27520
    },
    {
      "epoch": 0.4285147158414223,
      "grad_norm": 4.403416633605957,
      "learning_rate": 6.015797840087839e-07,
      "loss": 0.793,
      "step": 27525
    },
    {
      "epoch": 0.4285925568433917,
      "grad_norm": 3.6969947814941406,
      "learning_rate": 6.014978450041788e-07,
      "loss": 0.7078,
      "step": 27530
    },
    {
      "epoch": 0.42867039784536104,
      "grad_norm": 7.605321407318115,
      "learning_rate": 6.014159059995739e-07,
      "loss": 0.8086,
      "step": 27535
    },
    {
      "epoch": 0.42874823884733043,
      "grad_norm": 3.8635454177856445,
      "learning_rate": 6.01333966994969e-07,
      "loss": 0.7304,
      "step": 27540
    },
    {
      "epoch": 0.4288260798492998,
      "grad_norm": 4.390786647796631,
      "learning_rate": 6.012520279903638e-07,
      "loss": 0.8371,
      "step": 27545
    },
    {
      "epoch": 0.4289039208512692,
      "grad_norm": 3.810314655303955,
      "learning_rate": 6.011700889857589e-07,
      "loss": 0.7858,
      "step": 27550
    },
    {
      "epoch": 0.42898176185323855,
      "grad_norm": 3.335005283355713,
      "learning_rate": 6.01088149981154e-07,
      "loss": 0.7657,
      "step": 27555
    },
    {
      "epoch": 0.42905960285520794,
      "grad_norm": 3.9059786796569824,
      "learning_rate": 6.010062109765491e-07,
      "loss": 0.6121,
      "step": 27560
    },
    {
      "epoch": 0.42913744385717734,
      "grad_norm": 7.90656042098999,
      "learning_rate": 6.00924271971944e-07,
      "loss": 0.7579,
      "step": 27565
    },
    {
      "epoch": 0.42921528485914673,
      "grad_norm": 2.571089506149292,
      "learning_rate": 6.008423329673391e-07,
      "loss": 0.7464,
      "step": 27570
    },
    {
      "epoch": 0.42929312586111606,
      "grad_norm": 3.8170342445373535,
      "learning_rate": 6.007603939627342e-07,
      "loss": 0.8353,
      "step": 27575
    },
    {
      "epoch": 0.42937096686308546,
      "grad_norm": 3.4820406436920166,
      "learning_rate": 6.006784549581292e-07,
      "loss": 0.7535,
      "step": 27580
    },
    {
      "epoch": 0.42944880786505485,
      "grad_norm": 2.870199203491211,
      "learning_rate": 6.005965159535241e-07,
      "loss": 0.8004,
      "step": 27585
    },
    {
      "epoch": 0.42952664886702424,
      "grad_norm": 3.029043674468994,
      "learning_rate": 6.005145769489192e-07,
      "loss": 0.7063,
      "step": 27590
    },
    {
      "epoch": 0.4296044898689936,
      "grad_norm": 5.453882694244385,
      "learning_rate": 6.004326379443142e-07,
      "loss": 0.814,
      "step": 27595
    },
    {
      "epoch": 0.42968233087096297,
      "grad_norm": 3.331883430480957,
      "learning_rate": 6.003506989397092e-07,
      "loss": 0.787,
      "step": 27600
    },
    {
      "epoch": 0.42976017187293236,
      "grad_norm": 2.395042657852173,
      "learning_rate": 6.002687599351043e-07,
      "loss": 0.6732,
      "step": 27605
    },
    {
      "epoch": 0.4298380128749017,
      "grad_norm": 7.470876216888428,
      "learning_rate": 6.001868209304993e-07,
      "loss": 0.6835,
      "step": 27610
    },
    {
      "epoch": 0.4299158538768711,
      "grad_norm": 3.4728715419769287,
      "learning_rate": 6.001048819258944e-07,
      "loss": 0.6952,
      "step": 27615
    },
    {
      "epoch": 0.4299936948788405,
      "grad_norm": 2.973357677459717,
      "learning_rate": 6.000229429212894e-07,
      "loss": 0.6941,
      "step": 27620
    },
    {
      "epoch": 0.4300715358808099,
      "grad_norm": 2.8671586513519287,
      "learning_rate": 5.999410039166843e-07,
      "loss": 0.6964,
      "step": 27625
    },
    {
      "epoch": 0.4301493768827792,
      "grad_norm": 3.881767749786377,
      "learning_rate": 5.998590649120794e-07,
      "loss": 0.6912,
      "step": 27630
    },
    {
      "epoch": 0.4302272178847486,
      "grad_norm": 3.46494460105896,
      "learning_rate": 5.997771259074745e-07,
      "loss": 0.6453,
      "step": 27635
    },
    {
      "epoch": 0.430305058886718,
      "grad_norm": 6.151224613189697,
      "learning_rate": 5.996951869028694e-07,
      "loss": 0.8322,
      "step": 27640
    },
    {
      "epoch": 0.4303828998886874,
      "grad_norm": 4.667808532714844,
      "learning_rate": 5.996132478982645e-07,
      "loss": 0.8206,
      "step": 27645
    },
    {
      "epoch": 0.4304607408906567,
      "grad_norm": 5.188647747039795,
      "learning_rate": 5.995313088936596e-07,
      "loss": 0.663,
      "step": 27650
    },
    {
      "epoch": 0.4305385818926261,
      "grad_norm": 9.459535598754883,
      "learning_rate": 5.994493698890545e-07,
      "loss": 0.665,
      "step": 27655
    },
    {
      "epoch": 0.4306164228945955,
      "grad_norm": 2.7770302295684814,
      "learning_rate": 5.993674308844496e-07,
      "loss": 0.696,
      "step": 27660
    },
    {
      "epoch": 0.4306942638965649,
      "grad_norm": 3.644118070602417,
      "learning_rate": 5.992854918798446e-07,
      "loss": 0.5877,
      "step": 27665
    },
    {
      "epoch": 0.43077210489853424,
      "grad_norm": 3.66776967048645,
      "learning_rate": 5.992035528752396e-07,
      "loss": 0.8004,
      "step": 27670
    },
    {
      "epoch": 0.43084994590050363,
      "grad_norm": 3.4178576469421387,
      "learning_rate": 5.991216138706346e-07,
      "loss": 0.8062,
      "step": 27675
    },
    {
      "epoch": 0.430927786902473,
      "grad_norm": 7.568190574645996,
      "learning_rate": 5.990396748660297e-07,
      "loss": 0.7219,
      "step": 27680
    },
    {
      "epoch": 0.4310056279044424,
      "grad_norm": 3.1899938583374023,
      "learning_rate": 5.989577358614248e-07,
      "loss": 0.6335,
      "step": 27685
    },
    {
      "epoch": 0.43108346890641175,
      "grad_norm": 4.04885721206665,
      "learning_rate": 5.988757968568197e-07,
      "loss": 0.8045,
      "step": 27690
    },
    {
      "epoch": 0.43116130990838114,
      "grad_norm": 3.311727285385132,
      "learning_rate": 5.987938578522148e-07,
      "loss": 0.8141,
      "step": 27695
    },
    {
      "epoch": 0.43123915091035053,
      "grad_norm": 2.6867873668670654,
      "learning_rate": 5.987119188476099e-07,
      "loss": 0.7318,
      "step": 27700
    },
    {
      "epoch": 0.43131699191231987,
      "grad_norm": 3.6827845573425293,
      "learning_rate": 5.986299798430049e-07,
      "loss": 0.7296,
      "step": 27705
    },
    {
      "epoch": 0.43139483291428926,
      "grad_norm": 4.631564140319824,
      "learning_rate": 5.985480408383998e-07,
      "loss": 0.714,
      "step": 27710
    },
    {
      "epoch": 0.43147267391625865,
      "grad_norm": 4.947175979614258,
      "learning_rate": 5.984661018337949e-07,
      "loss": 0.892,
      "step": 27715
    },
    {
      "epoch": 0.43155051491822805,
      "grad_norm": 2.701998233795166,
      "learning_rate": 5.983841628291899e-07,
      "loss": 0.7073,
      "step": 27720
    },
    {
      "epoch": 0.4316283559201974,
      "grad_norm": 4.04004430770874,
      "learning_rate": 5.98302223824585e-07,
      "loss": 0.8043,
      "step": 27725
    },
    {
      "epoch": 0.4317061969221668,
      "grad_norm": 9.032465934753418,
      "learning_rate": 5.9822028481998e-07,
      "loss": 0.8812,
      "step": 27730
    },
    {
      "epoch": 0.43178403792413617,
      "grad_norm": 4.29990816116333,
      "learning_rate": 5.98138345815375e-07,
      "loss": 0.671,
      "step": 27735
    },
    {
      "epoch": 0.43186187892610556,
      "grad_norm": 5.574892520904541,
      "learning_rate": 5.980564068107701e-07,
      "loss": 0.7746,
      "step": 27740
    },
    {
      "epoch": 0.4319397199280749,
      "grad_norm": 4.6538214683532715,
      "learning_rate": 5.979744678061651e-07,
      "loss": 0.9419,
      "step": 27745
    },
    {
      "epoch": 0.4320175609300443,
      "grad_norm": 3.485159158706665,
      "learning_rate": 5.9789252880156e-07,
      "loss": 0.8016,
      "step": 27750
    },
    {
      "epoch": 0.4320954019320137,
      "grad_norm": 4.120754718780518,
      "learning_rate": 5.978105897969551e-07,
      "loss": 0.8587,
      "step": 27755
    },
    {
      "epoch": 0.43217324293398307,
      "grad_norm": 4.850719451904297,
      "learning_rate": 5.977286507923502e-07,
      "loss": 0.8688,
      "step": 27760
    },
    {
      "epoch": 0.4322510839359524,
      "grad_norm": 5.697882175445557,
      "learning_rate": 5.976467117877451e-07,
      "loss": 0.7384,
      "step": 27765
    },
    {
      "epoch": 0.4323289249379218,
      "grad_norm": 3.8523309230804443,
      "learning_rate": 5.975647727831402e-07,
      "loss": 0.8453,
      "step": 27770
    },
    {
      "epoch": 0.4324067659398912,
      "grad_norm": 4.005265235900879,
      "learning_rate": 5.974828337785353e-07,
      "loss": 0.7361,
      "step": 27775
    },
    {
      "epoch": 0.4324846069418606,
      "grad_norm": 4.518993377685547,
      "learning_rate": 5.974008947739302e-07,
      "loss": 0.7459,
      "step": 27780
    },
    {
      "epoch": 0.4325624479438299,
      "grad_norm": 3.953925371170044,
      "learning_rate": 5.973189557693253e-07,
      "loss": 0.7403,
      "step": 27785
    },
    {
      "epoch": 0.4326402889457993,
      "grad_norm": 4.339183807373047,
      "learning_rate": 5.972370167647203e-07,
      "loss": 0.7239,
      "step": 27790
    },
    {
      "epoch": 0.4327181299477687,
      "grad_norm": 6.377171039581299,
      "learning_rate": 5.971550777601153e-07,
      "loss": 0.7259,
      "step": 27795
    },
    {
      "epoch": 0.43279597094973804,
      "grad_norm": 4.8265180587768555,
      "learning_rate": 5.970731387555103e-07,
      "loss": 0.8928,
      "step": 27800
    },
    {
      "epoch": 0.43287381195170743,
      "grad_norm": 3.585028648376465,
      "learning_rate": 5.969911997509054e-07,
      "loss": 0.727,
      "step": 27805
    },
    {
      "epoch": 0.4329516529536768,
      "grad_norm": 4.21317195892334,
      "learning_rate": 5.969092607463005e-07,
      "loss": 0.761,
      "step": 27810
    },
    {
      "epoch": 0.4330294939556462,
      "grad_norm": 3.281339645385742,
      "learning_rate": 5.968273217416955e-07,
      "loss": 0.8078,
      "step": 27815
    },
    {
      "epoch": 0.43310733495761555,
      "grad_norm": 4.054388523101807,
      "learning_rate": 5.967453827370905e-07,
      "loss": 0.8311,
      "step": 27820
    },
    {
      "epoch": 0.43318517595958494,
      "grad_norm": 3.831932544708252,
      "learning_rate": 5.966634437324856e-07,
      "loss": 0.7771,
      "step": 27825
    },
    {
      "epoch": 0.43326301696155434,
      "grad_norm": 6.436086654663086,
      "learning_rate": 5.965815047278805e-07,
      "loss": 0.8272,
      "step": 27830
    },
    {
      "epoch": 0.43334085796352373,
      "grad_norm": 3.4208126068115234,
      "learning_rate": 5.964995657232755e-07,
      "loss": 0.6237,
      "step": 27835
    },
    {
      "epoch": 0.43341869896549307,
      "grad_norm": 2.5811595916748047,
      "learning_rate": 5.964176267186706e-07,
      "loss": 0.7275,
      "step": 27840
    },
    {
      "epoch": 0.43349653996746246,
      "grad_norm": 4.617489814758301,
      "learning_rate": 5.963356877140656e-07,
      "loss": 0.7164,
      "step": 27845
    },
    {
      "epoch": 0.43357438096943185,
      "grad_norm": 3.223447561264038,
      "learning_rate": 5.962537487094607e-07,
      "loss": 0.7278,
      "step": 27850
    },
    {
      "epoch": 0.43365222197140124,
      "grad_norm": 3.8689448833465576,
      "learning_rate": 5.961718097048557e-07,
      "loss": 0.8391,
      "step": 27855
    },
    {
      "epoch": 0.4337300629733706,
      "grad_norm": 3.374972343444824,
      "learning_rate": 5.960898707002507e-07,
      "loss": 0.7914,
      "step": 27860
    },
    {
      "epoch": 0.43380790397533997,
      "grad_norm": 6.62541389465332,
      "learning_rate": 5.960079316956458e-07,
      "loss": 0.7388,
      "step": 27865
    },
    {
      "epoch": 0.43388574497730936,
      "grad_norm": 3.3307340145111084,
      "learning_rate": 5.959259926910407e-07,
      "loss": 0.7181,
      "step": 27870
    },
    {
      "epoch": 0.4339635859792787,
      "grad_norm": 3.8283004760742188,
      "learning_rate": 5.958440536864357e-07,
      "loss": 0.6944,
      "step": 27875
    },
    {
      "epoch": 0.4340414269812481,
      "grad_norm": 3.793822765350342,
      "learning_rate": 5.957621146818308e-07,
      "loss": 0.6795,
      "step": 27880
    },
    {
      "epoch": 0.4341192679832175,
      "grad_norm": 3.196544647216797,
      "learning_rate": 5.956801756772259e-07,
      "loss": 0.8672,
      "step": 27885
    },
    {
      "epoch": 0.4341971089851869,
      "grad_norm": 4.767712116241455,
      "learning_rate": 5.955982366726208e-07,
      "loss": 0.8061,
      "step": 27890
    },
    {
      "epoch": 0.4342749499871562,
      "grad_norm": 2.1641054153442383,
      "learning_rate": 5.955162976680159e-07,
      "loss": 0.7623,
      "step": 27895
    },
    {
      "epoch": 0.4343527909891256,
      "grad_norm": 3.6267597675323486,
      "learning_rate": 5.95434358663411e-07,
      "loss": 0.6784,
      "step": 27900
    },
    {
      "epoch": 0.434430631991095,
      "grad_norm": 4.674212455749512,
      "learning_rate": 5.95352419658806e-07,
      "loss": 0.759,
      "step": 27905
    },
    {
      "epoch": 0.4345084729930644,
      "grad_norm": 6.812900543212891,
      "learning_rate": 5.952704806542009e-07,
      "loss": 0.7098,
      "step": 27910
    },
    {
      "epoch": 0.4345863139950337,
      "grad_norm": 3.000892162322998,
      "learning_rate": 5.95188541649596e-07,
      "loss": 0.6528,
      "step": 27915
    },
    {
      "epoch": 0.4346641549970031,
      "grad_norm": 13.540499687194824,
      "learning_rate": 5.95106602644991e-07,
      "loss": 0.7468,
      "step": 27920
    },
    {
      "epoch": 0.4347419959989725,
      "grad_norm": 4.585391998291016,
      "learning_rate": 5.95024663640386e-07,
      "loss": 0.6364,
      "step": 27925
    },
    {
      "epoch": 0.4348198370009419,
      "grad_norm": 5.876636505126953,
      "learning_rate": 5.949427246357811e-07,
      "loss": 0.763,
      "step": 27930
    },
    {
      "epoch": 0.43489767800291124,
      "grad_norm": 4.930020809173584,
      "learning_rate": 5.948607856311762e-07,
      "loss": 0.5975,
      "step": 27935
    },
    {
      "epoch": 0.43497551900488063,
      "grad_norm": 4.648181438446045,
      "learning_rate": 5.947788466265712e-07,
      "loss": 0.7269,
      "step": 27940
    },
    {
      "epoch": 0.43505336000685,
      "grad_norm": 4.776635646820068,
      "learning_rate": 5.946969076219662e-07,
      "loss": 0.6465,
      "step": 27945
    },
    {
      "epoch": 0.4351312010088194,
      "grad_norm": 5.510016918182373,
      "learning_rate": 5.946149686173612e-07,
      "loss": 0.7738,
      "step": 27950
    },
    {
      "epoch": 0.43520904201078875,
      "grad_norm": 2.281200647354126,
      "learning_rate": 5.945330296127562e-07,
      "loss": 0.7802,
      "step": 27955
    },
    {
      "epoch": 0.43528688301275814,
      "grad_norm": 3.1887929439544678,
      "learning_rate": 5.944510906081512e-07,
      "loss": 0.7501,
      "step": 27960
    },
    {
      "epoch": 0.43536472401472753,
      "grad_norm": 3.804816961288452,
      "learning_rate": 5.943691516035463e-07,
      "loss": 0.6792,
      "step": 27965
    },
    {
      "epoch": 0.43544256501669687,
      "grad_norm": 3.4015350341796875,
      "learning_rate": 5.942872125989413e-07,
      "loss": 0.8475,
      "step": 27970
    },
    {
      "epoch": 0.43552040601866626,
      "grad_norm": 4.119654178619385,
      "learning_rate": 5.942052735943364e-07,
      "loss": 0.8177,
      "step": 27975
    },
    {
      "epoch": 0.43559824702063565,
      "grad_norm": 8.41972827911377,
      "learning_rate": 5.941233345897314e-07,
      "loss": 0.7545,
      "step": 27980
    },
    {
      "epoch": 0.43567608802260505,
      "grad_norm": 5.477794647216797,
      "learning_rate": 5.940413955851264e-07,
      "loss": 0.7307,
      "step": 27985
    },
    {
      "epoch": 0.4357539290245744,
      "grad_norm": 7.2264838218688965,
      "learning_rate": 5.939594565805214e-07,
      "loss": 0.8613,
      "step": 27990
    },
    {
      "epoch": 0.4358317700265438,
      "grad_norm": 3.4156153202056885,
      "learning_rate": 5.938775175759165e-07,
      "loss": 0.7259,
      "step": 27995
    },
    {
      "epoch": 0.43590961102851317,
      "grad_norm": 2.4434916973114014,
      "learning_rate": 5.937955785713114e-07,
      "loss": 0.7627,
      "step": 28000
    },
    {
      "epoch": 0.43598745203048256,
      "grad_norm": 3.167893171310425,
      "learning_rate": 5.937136395667065e-07,
      "loss": 0.5847,
      "step": 28005
    },
    {
      "epoch": 0.4360652930324519,
      "grad_norm": 16.382583618164062,
      "learning_rate": 5.936317005621016e-07,
      "loss": 0.7994,
      "step": 28010
    },
    {
      "epoch": 0.4361431340344213,
      "grad_norm": 3.138211250305176,
      "learning_rate": 5.935497615574965e-07,
      "loss": 0.7655,
      "step": 28015
    },
    {
      "epoch": 0.4362209750363907,
      "grad_norm": 3.0444304943084717,
      "learning_rate": 5.934678225528916e-07,
      "loss": 0.7488,
      "step": 28020
    },
    {
      "epoch": 0.43629881603836007,
      "grad_norm": 3.296393394470215,
      "learning_rate": 5.933858835482867e-07,
      "loss": 0.685,
      "step": 28025
    },
    {
      "epoch": 0.4363766570403294,
      "grad_norm": 3.1975717544555664,
      "learning_rate": 5.933039445436817e-07,
      "loss": 0.7817,
      "step": 28030
    },
    {
      "epoch": 0.4364544980422988,
      "grad_norm": 2.862736701965332,
      "learning_rate": 5.932220055390766e-07,
      "loss": 0.8753,
      "step": 28035
    },
    {
      "epoch": 0.4365323390442682,
      "grad_norm": 10.180088996887207,
      "learning_rate": 5.931400665344717e-07,
      "loss": 0.6301,
      "step": 28040
    },
    {
      "epoch": 0.43661018004623753,
      "grad_norm": 3.7742133140563965,
      "learning_rate": 5.930581275298667e-07,
      "loss": 0.6582,
      "step": 28045
    },
    {
      "epoch": 0.4366880210482069,
      "grad_norm": 3.1581547260284424,
      "learning_rate": 5.929761885252617e-07,
      "loss": 0.7242,
      "step": 28050
    },
    {
      "epoch": 0.4367658620501763,
      "grad_norm": 10.063048362731934,
      "learning_rate": 5.928942495206568e-07,
      "loss": 1.0004,
      "step": 28055
    },
    {
      "epoch": 0.4368437030521457,
      "grad_norm": 3.4235334396362305,
      "learning_rate": 5.928123105160519e-07,
      "loss": 0.5652,
      "step": 28060
    },
    {
      "epoch": 0.43692154405411504,
      "grad_norm": 3.312119722366333,
      "learning_rate": 5.927303715114469e-07,
      "loss": 0.7674,
      "step": 28065
    },
    {
      "epoch": 0.43699938505608443,
      "grad_norm": 3.313004970550537,
      "learning_rate": 5.926484325068419e-07,
      "loss": 0.7633,
      "step": 28070
    },
    {
      "epoch": 0.4370772260580538,
      "grad_norm": 4.131656169891357,
      "learning_rate": 5.925664935022369e-07,
      "loss": 0.8007,
      "step": 28075
    },
    {
      "epoch": 0.4371550670600232,
      "grad_norm": 3.5195024013519287,
      "learning_rate": 5.924845544976319e-07,
      "loss": 0.7885,
      "step": 28080
    },
    {
      "epoch": 0.43723290806199255,
      "grad_norm": 3.162827253341675,
      "learning_rate": 5.92402615493027e-07,
      "loss": 0.7527,
      "step": 28085
    },
    {
      "epoch": 0.43731074906396195,
      "grad_norm": 4.721434593200684,
      "learning_rate": 5.92320676488422e-07,
      "loss": 0.8578,
      "step": 28090
    },
    {
      "epoch": 0.43738859006593134,
      "grad_norm": 3.5338213443756104,
      "learning_rate": 5.92238737483817e-07,
      "loss": 0.804,
      "step": 28095
    },
    {
      "epoch": 0.43746643106790073,
      "grad_norm": 3.930555582046509,
      "learning_rate": 5.921567984792121e-07,
      "loss": 0.7481,
      "step": 28100
    },
    {
      "epoch": 0.43754427206987007,
      "grad_norm": 5.562685966491699,
      "learning_rate": 5.920748594746071e-07,
      "loss": 0.8776,
      "step": 28105
    },
    {
      "epoch": 0.43762211307183946,
      "grad_norm": 3.6246979236602783,
      "learning_rate": 5.919929204700021e-07,
      "loss": 0.7283,
      "step": 28110
    },
    {
      "epoch": 0.43769995407380885,
      "grad_norm": 5.065835475921631,
      "learning_rate": 5.919109814653971e-07,
      "loss": 0.7177,
      "step": 28115
    },
    {
      "epoch": 0.43777779507577824,
      "grad_norm": 3.75182843208313,
      "learning_rate": 5.918290424607922e-07,
      "loss": 0.7741,
      "step": 28120
    },
    {
      "epoch": 0.4378556360777476,
      "grad_norm": 3.845771312713623,
      "learning_rate": 5.917471034561871e-07,
      "loss": 0.8227,
      "step": 28125
    },
    {
      "epoch": 0.43793347707971697,
      "grad_norm": 2.205892562866211,
      "learning_rate": 5.916651644515822e-07,
      "loss": 0.6918,
      "step": 28130
    },
    {
      "epoch": 0.43801131808168636,
      "grad_norm": 3.447526693344116,
      "learning_rate": 5.915832254469773e-07,
      "loss": 0.8242,
      "step": 28135
    },
    {
      "epoch": 0.4380891590836557,
      "grad_norm": 4.775432586669922,
      "learning_rate": 5.915012864423722e-07,
      "loss": 0.7072,
      "step": 28140
    },
    {
      "epoch": 0.4381670000856251,
      "grad_norm": 4.1686787605285645,
      "learning_rate": 5.914193474377673e-07,
      "loss": 0.7002,
      "step": 28145
    },
    {
      "epoch": 0.4382448410875945,
      "grad_norm": 4.378955841064453,
      "learning_rate": 5.913374084331624e-07,
      "loss": 0.794,
      "step": 28150
    },
    {
      "epoch": 0.4383226820895639,
      "grad_norm": 3.673280715942383,
      "learning_rate": 5.912554694285573e-07,
      "loss": 0.7652,
      "step": 28155
    },
    {
      "epoch": 0.4384005230915332,
      "grad_norm": 4.4424848556518555,
      "learning_rate": 5.911735304239523e-07,
      "loss": 0.6848,
      "step": 28160
    },
    {
      "epoch": 0.4384783640935026,
      "grad_norm": 3.245546579360962,
      "learning_rate": 5.910915914193474e-07,
      "loss": 0.741,
      "step": 28165
    },
    {
      "epoch": 0.438556205095472,
      "grad_norm": 2.628819227218628,
      "learning_rate": 5.910096524147424e-07,
      "loss": 0.6993,
      "step": 28170
    },
    {
      "epoch": 0.4386340460974414,
      "grad_norm": 4.761402130126953,
      "learning_rate": 5.909277134101375e-07,
      "loss": 0.732,
      "step": 28175
    },
    {
      "epoch": 0.4387118870994107,
      "grad_norm": 3.5356898307800293,
      "learning_rate": 5.908457744055325e-07,
      "loss": 0.7505,
      "step": 28180
    },
    {
      "epoch": 0.4387897281013801,
      "grad_norm": 2.67647647857666,
      "learning_rate": 5.907638354009276e-07,
      "loss": 0.7687,
      "step": 28185
    },
    {
      "epoch": 0.4388675691033495,
      "grad_norm": 9.661590576171875,
      "learning_rate": 5.906818963963226e-07,
      "loss": 0.7266,
      "step": 28190
    },
    {
      "epoch": 0.4389454101053189,
      "grad_norm": 3.347217321395874,
      "learning_rate": 5.905999573917175e-07,
      "loss": 0.7665,
      "step": 28195
    },
    {
      "epoch": 0.43902325110728824,
      "grad_norm": 2.7007672786712646,
      "learning_rate": 5.905180183871126e-07,
      "loss": 0.7923,
      "step": 28200
    },
    {
      "epoch": 0.43910109210925763,
      "grad_norm": 4.161789894104004,
      "learning_rate": 5.904360793825076e-07,
      "loss": 0.7244,
      "step": 28205
    },
    {
      "epoch": 0.439178933111227,
      "grad_norm": 2.561704158782959,
      "learning_rate": 5.903541403779027e-07,
      "loss": 0.6945,
      "step": 28210
    },
    {
      "epoch": 0.43925677411319636,
      "grad_norm": 4.047752380371094,
      "learning_rate": 5.902722013732977e-07,
      "loss": 0.6831,
      "step": 28215
    },
    {
      "epoch": 0.43933461511516575,
      "grad_norm": 5.509623050689697,
      "learning_rate": 5.901902623686927e-07,
      "loss": 0.8383,
      "step": 28220
    },
    {
      "epoch": 0.43941245611713514,
      "grad_norm": 2.814044237136841,
      "learning_rate": 5.901083233640878e-07,
      "loss": 0.711,
      "step": 28225
    },
    {
      "epoch": 0.43949029711910453,
      "grad_norm": 6.586339473724365,
      "learning_rate": 5.900263843594829e-07,
      "loss": 0.8168,
      "step": 28230
    },
    {
      "epoch": 0.43956813812107387,
      "grad_norm": 3.233229637145996,
      "learning_rate": 5.899444453548777e-07,
      "loss": 0.663,
      "step": 28235
    },
    {
      "epoch": 0.43964597912304326,
      "grad_norm": 3.1166365146636963,
      "learning_rate": 5.898625063502728e-07,
      "loss": 0.7402,
      "step": 28240
    },
    {
      "epoch": 0.43972382012501265,
      "grad_norm": 3.2115893363952637,
      "learning_rate": 5.897805673456679e-07,
      "loss": 0.8284,
      "step": 28245
    },
    {
      "epoch": 0.43980166112698205,
      "grad_norm": 4.563340187072754,
      "learning_rate": 5.896986283410628e-07,
      "loss": 0.6664,
      "step": 28250
    },
    {
      "epoch": 0.4398795021289514,
      "grad_norm": 2.2660348415374756,
      "learning_rate": 5.896166893364579e-07,
      "loss": 0.7314,
      "step": 28255
    },
    {
      "epoch": 0.4399573431309208,
      "grad_norm": 4.088837623596191,
      "learning_rate": 5.89534750331853e-07,
      "loss": 0.7606,
      "step": 28260
    },
    {
      "epoch": 0.44003518413289017,
      "grad_norm": 6.516998767852783,
      "learning_rate": 5.89452811327248e-07,
      "loss": 0.7674,
      "step": 28265
    },
    {
      "epoch": 0.44011302513485956,
      "grad_norm": 4.032287120819092,
      "learning_rate": 5.89370872322643e-07,
      "loss": 0.7683,
      "step": 28270
    },
    {
      "epoch": 0.4401908661368289,
      "grad_norm": 4.658278465270996,
      "learning_rate": 5.89288933318038e-07,
      "loss": 0.8606,
      "step": 28275
    },
    {
      "epoch": 0.4402687071387983,
      "grad_norm": 7.811373710632324,
      "learning_rate": 5.89206994313433e-07,
      "loss": 0.715,
      "step": 28280
    },
    {
      "epoch": 0.4403465481407677,
      "grad_norm": 4.1481547355651855,
      "learning_rate": 5.89125055308828e-07,
      "loss": 0.6848,
      "step": 28285
    },
    {
      "epoch": 0.44042438914273707,
      "grad_norm": 10.9120512008667,
      "learning_rate": 5.890431163042231e-07,
      "loss": 0.7103,
      "step": 28290
    },
    {
      "epoch": 0.4405022301447064,
      "grad_norm": 4.8491129875183105,
      "learning_rate": 5.889611772996181e-07,
      "loss": 0.6921,
      "step": 28295
    },
    {
      "epoch": 0.4405800711466758,
      "grad_norm": 3.8429486751556396,
      "learning_rate": 5.888792382950132e-07,
      "loss": 0.7103,
      "step": 28300
    },
    {
      "epoch": 0.4406579121486452,
      "grad_norm": 3.482661008834839,
      "learning_rate": 5.887972992904082e-07,
      "loss": 0.8208,
      "step": 28305
    },
    {
      "epoch": 0.44073575315061453,
      "grad_norm": 3.3518240451812744,
      "learning_rate": 5.887153602858033e-07,
      "loss": 0.7582,
      "step": 28310
    },
    {
      "epoch": 0.4408135941525839,
      "grad_norm": 3.1931521892547607,
      "learning_rate": 5.886334212811982e-07,
      "loss": 0.6514,
      "step": 28315
    },
    {
      "epoch": 0.4408914351545533,
      "grad_norm": 3.3006231784820557,
      "learning_rate": 5.885514822765933e-07,
      "loss": 0.7889,
      "step": 28320
    },
    {
      "epoch": 0.4409692761565227,
      "grad_norm": 3.569957733154297,
      "learning_rate": 5.884695432719883e-07,
      "loss": 0.6939,
      "step": 28325
    },
    {
      "epoch": 0.44104711715849204,
      "grad_norm": 4.447323799133301,
      "learning_rate": 5.883876042673833e-07,
      "loss": 0.6871,
      "step": 28330
    },
    {
      "epoch": 0.44112495816046143,
      "grad_norm": 4.069333076477051,
      "learning_rate": 5.883056652627784e-07,
      "loss": 0.7319,
      "step": 28335
    },
    {
      "epoch": 0.4412027991624308,
      "grad_norm": 2.6958563327789307,
      "learning_rate": 5.882237262581734e-07,
      "loss": 0.6825,
      "step": 28340
    },
    {
      "epoch": 0.4412806401644002,
      "grad_norm": 5.831766605377197,
      "learning_rate": 5.881417872535684e-07,
      "loss": 0.6912,
      "step": 28345
    },
    {
      "epoch": 0.44135848116636955,
      "grad_norm": 4.509929656982422,
      "learning_rate": 5.880598482489635e-07,
      "loss": 0.7606,
      "step": 28350
    },
    {
      "epoch": 0.44143632216833895,
      "grad_norm": 4.06305456161499,
      "learning_rate": 5.879779092443586e-07,
      "loss": 0.716,
      "step": 28355
    },
    {
      "epoch": 0.44151416317030834,
      "grad_norm": 5.638819217681885,
      "learning_rate": 5.878959702397534e-07,
      "loss": 0.7627,
      "step": 28360
    },
    {
      "epoch": 0.44159200417227773,
      "grad_norm": 5.382050514221191,
      "learning_rate": 5.878140312351485e-07,
      "loss": 0.859,
      "step": 28365
    },
    {
      "epoch": 0.44166984517424707,
      "grad_norm": 3.79328989982605,
      "learning_rate": 5.877320922305436e-07,
      "loss": 0.7537,
      "step": 28370
    },
    {
      "epoch": 0.44174768617621646,
      "grad_norm": 5.255930423736572,
      "learning_rate": 5.876501532259385e-07,
      "loss": 0.8518,
      "step": 28375
    },
    {
      "epoch": 0.44182552717818585,
      "grad_norm": 3.785076141357422,
      "learning_rate": 5.875682142213336e-07,
      "loss": 0.6591,
      "step": 28380
    },
    {
      "epoch": 0.44190336818015524,
      "grad_norm": 9.082311630249023,
      "learning_rate": 5.874862752167287e-07,
      "loss": 0.6978,
      "step": 28385
    },
    {
      "epoch": 0.4419812091821246,
      "grad_norm": 4.221014976501465,
      "learning_rate": 5.874043362121237e-07,
      "loss": 0.6371,
      "step": 28390
    },
    {
      "epoch": 0.44205905018409397,
      "grad_norm": 4.4617767333984375,
      "learning_rate": 5.873223972075187e-07,
      "loss": 0.7363,
      "step": 28395
    },
    {
      "epoch": 0.44213689118606336,
      "grad_norm": 4.976316928863525,
      "learning_rate": 5.872404582029137e-07,
      "loss": 0.758,
      "step": 28400
    },
    {
      "epoch": 0.4422147321880327,
      "grad_norm": 4.9800310134887695,
      "learning_rate": 5.871585191983087e-07,
      "loss": 0.8536,
      "step": 28405
    },
    {
      "epoch": 0.4422925731900021,
      "grad_norm": 3.337171792984009,
      "learning_rate": 5.870765801937038e-07,
      "loss": 0.7016,
      "step": 28410
    },
    {
      "epoch": 0.4423704141919715,
      "grad_norm": 3.1711952686309814,
      "learning_rate": 5.869946411890988e-07,
      "loss": 0.6803,
      "step": 28415
    },
    {
      "epoch": 0.4424482551939409,
      "grad_norm": 4.335486888885498,
      "learning_rate": 5.869127021844938e-07,
      "loss": 0.8334,
      "step": 28420
    },
    {
      "epoch": 0.4425260961959102,
      "grad_norm": 8.560162544250488,
      "learning_rate": 5.868307631798889e-07,
      "loss": 0.7931,
      "step": 28425
    },
    {
      "epoch": 0.4426039371978796,
      "grad_norm": 5.151147842407227,
      "learning_rate": 5.867488241752839e-07,
      "loss": 0.7387,
      "step": 28430
    },
    {
      "epoch": 0.442681778199849,
      "grad_norm": 6.47930908203125,
      "learning_rate": 5.86666885170679e-07,
      "loss": 0.6809,
      "step": 28435
    },
    {
      "epoch": 0.4427596192018184,
      "grad_norm": 3.0402133464813232,
      "learning_rate": 5.865849461660739e-07,
      "loss": 0.6932,
      "step": 28440
    },
    {
      "epoch": 0.4428374602037877,
      "grad_norm": 5.68237829208374,
      "learning_rate": 5.86503007161469e-07,
      "loss": 0.7512,
      "step": 28445
    },
    {
      "epoch": 0.4429153012057571,
      "grad_norm": 7.695286273956299,
      "learning_rate": 5.86421068156864e-07,
      "loss": 0.6204,
      "step": 28450
    },
    {
      "epoch": 0.4429931422077265,
      "grad_norm": 3.5092580318450928,
      "learning_rate": 5.86339129152259e-07,
      "loss": 0.7814,
      "step": 28455
    },
    {
      "epoch": 0.4430709832096959,
      "grad_norm": 2.973787784576416,
      "learning_rate": 5.862571901476541e-07,
      "loss": 0.7079,
      "step": 28460
    },
    {
      "epoch": 0.44314882421166524,
      "grad_norm": 2.6469204425811768,
      "learning_rate": 5.861752511430492e-07,
      "loss": 0.701,
      "step": 28465
    },
    {
      "epoch": 0.44322666521363463,
      "grad_norm": 3.336364507675171,
      "learning_rate": 5.860933121384441e-07,
      "loss": 0.6272,
      "step": 28470
    },
    {
      "epoch": 0.443304506215604,
      "grad_norm": 3.0087203979492188,
      "learning_rate": 5.860113731338392e-07,
      "loss": 0.6707,
      "step": 28475
    },
    {
      "epoch": 0.44338234721757336,
      "grad_norm": 2.546515941619873,
      "learning_rate": 5.859294341292342e-07,
      "loss": 0.7941,
      "step": 28480
    },
    {
      "epoch": 0.44346018821954275,
      "grad_norm": 7.917011260986328,
      "learning_rate": 5.858474951246291e-07,
      "loss": 0.8553,
      "step": 28485
    },
    {
      "epoch": 0.44353802922151214,
      "grad_norm": 7.305311679840088,
      "learning_rate": 5.857655561200242e-07,
      "loss": 0.695,
      "step": 28490
    },
    {
      "epoch": 0.44361587022348153,
      "grad_norm": 5.976718425750732,
      "learning_rate": 5.856836171154193e-07,
      "loss": 0.6468,
      "step": 28495
    },
    {
      "epoch": 0.44369371122545087,
      "grad_norm": 3.3048276901245117,
      "learning_rate": 5.856016781108143e-07,
      "loss": 0.7979,
      "step": 28500
    },
    {
      "epoch": 0.44377155222742026,
      "grad_norm": 3.263644218444824,
      "learning_rate": 5.855197391062093e-07,
      "loss": 0.8195,
      "step": 28505
    },
    {
      "epoch": 0.44384939322938965,
      "grad_norm": 2.6208372116088867,
      "learning_rate": 5.854378001016044e-07,
      "loss": 0.7456,
      "step": 28510
    },
    {
      "epoch": 0.44392723423135905,
      "grad_norm": 3.8031177520751953,
      "learning_rate": 5.853558610969994e-07,
      "loss": 0.7559,
      "step": 28515
    },
    {
      "epoch": 0.4440050752333284,
      "grad_norm": 3.0899171829223633,
      "learning_rate": 5.852739220923943e-07,
      "loss": 0.6635,
      "step": 28520
    },
    {
      "epoch": 0.4440829162352978,
      "grad_norm": 6.144171237945557,
      "learning_rate": 5.851919830877894e-07,
      "loss": 0.7759,
      "step": 28525
    },
    {
      "epoch": 0.44416075723726717,
      "grad_norm": 4.998770236968994,
      "learning_rate": 5.851100440831844e-07,
      "loss": 0.8274,
      "step": 28530
    },
    {
      "epoch": 0.44423859823923656,
      "grad_norm": 3.759521007537842,
      "learning_rate": 5.850281050785795e-07,
      "loss": 0.7592,
      "step": 28535
    },
    {
      "epoch": 0.4443164392412059,
      "grad_norm": 4.322113990783691,
      "learning_rate": 5.849461660739745e-07,
      "loss": 0.8804,
      "step": 28540
    },
    {
      "epoch": 0.4443942802431753,
      "grad_norm": 10.286017417907715,
      "learning_rate": 5.848642270693695e-07,
      "loss": 0.7539,
      "step": 28545
    },
    {
      "epoch": 0.4444721212451447,
      "grad_norm": 4.129453659057617,
      "learning_rate": 5.847822880647646e-07,
      "loss": 0.8594,
      "step": 28550
    },
    {
      "epoch": 0.44454996224711407,
      "grad_norm": 3.813709020614624,
      "learning_rate": 5.847003490601597e-07,
      "loss": 0.8825,
      "step": 28555
    },
    {
      "epoch": 0.4446278032490834,
      "grad_norm": 4.043671607971191,
      "learning_rate": 5.846184100555545e-07,
      "loss": 0.7422,
      "step": 28560
    },
    {
      "epoch": 0.4447056442510528,
      "grad_norm": 5.5891008377075195,
      "learning_rate": 5.845364710509496e-07,
      "loss": 0.8352,
      "step": 28565
    },
    {
      "epoch": 0.4447834852530222,
      "grad_norm": 3.9800727367401123,
      "learning_rate": 5.844545320463447e-07,
      "loss": 0.8648,
      "step": 28570
    },
    {
      "epoch": 0.44486132625499153,
      "grad_norm": 4.688074588775635,
      "learning_rate": 5.843725930417397e-07,
      "loss": 0.6971,
      "step": 28575
    },
    {
      "epoch": 0.4449391672569609,
      "grad_norm": 3.784926176071167,
      "learning_rate": 5.842906540371347e-07,
      "loss": 0.7281,
      "step": 28580
    },
    {
      "epoch": 0.4450170082589303,
      "grad_norm": 2.675313711166382,
      "learning_rate": 5.842087150325298e-07,
      "loss": 0.7733,
      "step": 28585
    },
    {
      "epoch": 0.4450948492608997,
      "grad_norm": 2.694478988647461,
      "learning_rate": 5.841267760279249e-07,
      "loss": 0.7713,
      "step": 28590
    },
    {
      "epoch": 0.44517269026286904,
      "grad_norm": 5.597519397735596,
      "learning_rate": 5.840448370233198e-07,
      "loss": 0.615,
      "step": 28595
    },
    {
      "epoch": 0.44525053126483843,
      "grad_norm": 3.3392319679260254,
      "learning_rate": 5.839628980187148e-07,
      "loss": 0.7699,
      "step": 28600
    },
    {
      "epoch": 0.4453283722668078,
      "grad_norm": 4.946972846984863,
      "learning_rate": 5.838809590141099e-07,
      "loss": 0.7339,
      "step": 28605
    },
    {
      "epoch": 0.4454062132687772,
      "grad_norm": 3.593247413635254,
      "learning_rate": 5.837990200095048e-07,
      "loss": 0.7862,
      "step": 28610
    },
    {
      "epoch": 0.44548405427074655,
      "grad_norm": 5.393089771270752,
      "learning_rate": 5.837170810048999e-07,
      "loss": 0.8712,
      "step": 28615
    },
    {
      "epoch": 0.44556189527271595,
      "grad_norm": 2.7674152851104736,
      "learning_rate": 5.83635142000295e-07,
      "loss": 0.7376,
      "step": 28620
    },
    {
      "epoch": 0.44563973627468534,
      "grad_norm": 3.860445499420166,
      "learning_rate": 5.8355320299569e-07,
      "loss": 0.8108,
      "step": 28625
    },
    {
      "epoch": 0.44571757727665473,
      "grad_norm": 3.221179485321045,
      "learning_rate": 5.83471263991085e-07,
      "loss": 0.7394,
      "step": 28630
    },
    {
      "epoch": 0.44579541827862407,
      "grad_norm": 3.5511248111724854,
      "learning_rate": 5.833893249864801e-07,
      "loss": 0.7899,
      "step": 28635
    },
    {
      "epoch": 0.44587325928059346,
      "grad_norm": 2.731159210205078,
      "learning_rate": 5.83307385981875e-07,
      "loss": 0.6319,
      "step": 28640
    },
    {
      "epoch": 0.44595110028256285,
      "grad_norm": 4.476842403411865,
      "learning_rate": 5.8322544697727e-07,
      "loss": 0.8062,
      "step": 28645
    },
    {
      "epoch": 0.4460289412845322,
      "grad_norm": 3.592449426651001,
      "learning_rate": 5.831435079726651e-07,
      "loss": 0.7144,
      "step": 28650
    },
    {
      "epoch": 0.4461067822865016,
      "grad_norm": 4.755378723144531,
      "learning_rate": 5.830615689680601e-07,
      "loss": 0.757,
      "step": 28655
    },
    {
      "epoch": 0.44618462328847097,
      "grad_norm": 2.9378182888031006,
      "learning_rate": 5.829796299634552e-07,
      "loss": 0.754,
      "step": 28660
    },
    {
      "epoch": 0.44626246429044036,
      "grad_norm": 4.925267696380615,
      "learning_rate": 5.828976909588502e-07,
      "loss": 0.8127,
      "step": 28665
    },
    {
      "epoch": 0.4463403052924097,
      "grad_norm": 4.041872978210449,
      "learning_rate": 5.828157519542452e-07,
      "loss": 0.6518,
      "step": 28670
    },
    {
      "epoch": 0.4464181462943791,
      "grad_norm": 6.165174961090088,
      "learning_rate": 5.827338129496403e-07,
      "loss": 0.7984,
      "step": 28675
    },
    {
      "epoch": 0.4464959872963485,
      "grad_norm": 4.192864418029785,
      "learning_rate": 5.826518739450354e-07,
      "loss": 0.7825,
      "step": 28680
    },
    {
      "epoch": 0.4465738282983179,
      "grad_norm": 3.121920108795166,
      "learning_rate": 5.825699349404302e-07,
      "loss": 0.6702,
      "step": 28685
    },
    {
      "epoch": 0.4466516693002872,
      "grad_norm": 7.205260276794434,
      "learning_rate": 5.824879959358253e-07,
      "loss": 0.8273,
      "step": 28690
    },
    {
      "epoch": 0.4467295103022566,
      "grad_norm": 3.040034532546997,
      "learning_rate": 5.824060569312204e-07,
      "loss": 0.822,
      "step": 28695
    },
    {
      "epoch": 0.446807351304226,
      "grad_norm": 2.335698127746582,
      "learning_rate": 5.823241179266154e-07,
      "loss": 0.6392,
      "step": 28700
    },
    {
      "epoch": 0.4468851923061954,
      "grad_norm": 4.273251533508301,
      "learning_rate": 5.822421789220104e-07,
      "loss": 0.6925,
      "step": 28705
    },
    {
      "epoch": 0.4469630333081647,
      "grad_norm": 3.9456238746643066,
      "learning_rate": 5.821602399174055e-07,
      "loss": 0.7481,
      "step": 28710
    },
    {
      "epoch": 0.4470408743101341,
      "grad_norm": 4.079767227172852,
      "learning_rate": 5.820783009128006e-07,
      "loss": 0.7397,
      "step": 28715
    },
    {
      "epoch": 0.4471187153121035,
      "grad_norm": 4.003576755523682,
      "learning_rate": 5.819963619081955e-07,
      "loss": 0.8125,
      "step": 28720
    },
    {
      "epoch": 0.4471965563140729,
      "grad_norm": 4.537726402282715,
      "learning_rate": 5.819144229035905e-07,
      "loss": 0.7417,
      "step": 28725
    },
    {
      "epoch": 0.44727439731604224,
      "grad_norm": 7.388315200805664,
      "learning_rate": 5.818324838989856e-07,
      "loss": 0.8641,
      "step": 28730
    },
    {
      "epoch": 0.44735223831801163,
      "grad_norm": 3.0487756729125977,
      "learning_rate": 5.817505448943805e-07,
      "loss": 0.8928,
      "step": 28735
    },
    {
      "epoch": 0.447430079319981,
      "grad_norm": 3.1655802726745605,
      "learning_rate": 5.816686058897756e-07,
      "loss": 0.707,
      "step": 28740
    },
    {
      "epoch": 0.44750792032195036,
      "grad_norm": 4.178553104400635,
      "learning_rate": 5.815866668851707e-07,
      "loss": 0.7312,
      "step": 28745
    },
    {
      "epoch": 0.44758576132391975,
      "grad_norm": 7.626601696014404,
      "learning_rate": 5.815047278805657e-07,
      "loss": 0.7456,
      "step": 28750
    },
    {
      "epoch": 0.44766360232588914,
      "grad_norm": 4.492088794708252,
      "learning_rate": 5.814227888759607e-07,
      "loss": 0.7463,
      "step": 28755
    },
    {
      "epoch": 0.44774144332785853,
      "grad_norm": 3.4210524559020996,
      "learning_rate": 5.813408498713558e-07,
      "loss": 0.7552,
      "step": 28760
    },
    {
      "epoch": 0.44781928432982787,
      "grad_norm": 4.002391338348389,
      "learning_rate": 5.812589108667507e-07,
      "loss": 0.7915,
      "step": 28765
    },
    {
      "epoch": 0.44789712533179726,
      "grad_norm": 2.579582691192627,
      "learning_rate": 5.811769718621458e-07,
      "loss": 0.845,
      "step": 28770
    },
    {
      "epoch": 0.44797496633376666,
      "grad_norm": 5.933312892913818,
      "learning_rate": 5.810950328575408e-07,
      "loss": 0.7263,
      "step": 28775
    },
    {
      "epoch": 0.44805280733573605,
      "grad_norm": 6.678994655609131,
      "learning_rate": 5.810130938529358e-07,
      "loss": 0.9789,
      "step": 28780
    },
    {
      "epoch": 0.4481306483377054,
      "grad_norm": 2.9170658588409424,
      "learning_rate": 5.809311548483309e-07,
      "loss": 0.7674,
      "step": 28785
    },
    {
      "epoch": 0.4482084893396748,
      "grad_norm": 2.6036124229431152,
      "learning_rate": 5.808492158437259e-07,
      "loss": 0.5481,
      "step": 28790
    },
    {
      "epoch": 0.44828633034164417,
      "grad_norm": 4.275371551513672,
      "learning_rate": 5.807672768391209e-07,
      "loss": 0.7133,
      "step": 28795
    },
    {
      "epoch": 0.44836417134361356,
      "grad_norm": 4.063039302825928,
      "learning_rate": 5.80685337834516e-07,
      "loss": 0.7645,
      "step": 28800
    },
    {
      "epoch": 0.4484420123455829,
      "grad_norm": 4.390212059020996,
      "learning_rate": 5.80603398829911e-07,
      "loss": 0.7717,
      "step": 28805
    },
    {
      "epoch": 0.4485198533475523,
      "grad_norm": 2.9731879234313965,
      "learning_rate": 5.80521459825306e-07,
      "loss": 0.8827,
      "step": 28810
    },
    {
      "epoch": 0.4485976943495217,
      "grad_norm": 5.1143035888671875,
      "learning_rate": 5.80439520820701e-07,
      "loss": 0.6741,
      "step": 28815
    },
    {
      "epoch": 0.448675535351491,
      "grad_norm": 4.85829496383667,
      "learning_rate": 5.803575818160961e-07,
      "loss": 0.7148,
      "step": 28820
    },
    {
      "epoch": 0.4487533763534604,
      "grad_norm": 5.24547004699707,
      "learning_rate": 5.802756428114912e-07,
      "loss": 0.7149,
      "step": 28825
    },
    {
      "epoch": 0.4488312173554298,
      "grad_norm": 4.583830833435059,
      "learning_rate": 5.801937038068861e-07,
      "loss": 0.7881,
      "step": 28830
    },
    {
      "epoch": 0.4489090583573992,
      "grad_norm": 3.0546112060546875,
      "learning_rate": 5.801117648022812e-07,
      "loss": 0.7414,
      "step": 28835
    },
    {
      "epoch": 0.44898689935936853,
      "grad_norm": 4.399287700653076,
      "learning_rate": 5.800298257976763e-07,
      "loss": 0.7823,
      "step": 28840
    },
    {
      "epoch": 0.4490647403613379,
      "grad_norm": 3.3814289569854736,
      "learning_rate": 5.799478867930711e-07,
      "loss": 0.6798,
      "step": 28845
    },
    {
      "epoch": 0.4491425813633073,
      "grad_norm": 4.2093963623046875,
      "learning_rate": 5.798659477884662e-07,
      "loss": 0.7128,
      "step": 28850
    },
    {
      "epoch": 0.4492204223652767,
      "grad_norm": 3.488926887512207,
      "learning_rate": 5.797840087838613e-07,
      "loss": 0.7998,
      "step": 28855
    },
    {
      "epoch": 0.44929826336724604,
      "grad_norm": 3.5785272121429443,
      "learning_rate": 5.797020697792563e-07,
      "loss": 0.7752,
      "step": 28860
    },
    {
      "epoch": 0.44937610436921543,
      "grad_norm": 4.4823150634765625,
      "learning_rate": 5.796201307746513e-07,
      "loss": 0.7829,
      "step": 28865
    },
    {
      "epoch": 0.4494539453711848,
      "grad_norm": 4.716001033782959,
      "learning_rate": 5.795381917700464e-07,
      "loss": 0.8278,
      "step": 28870
    },
    {
      "epoch": 0.4495317863731542,
      "grad_norm": 5.834087371826172,
      "learning_rate": 5.794562527654414e-07,
      "loss": 0.8443,
      "step": 28875
    },
    {
      "epoch": 0.44960962737512356,
      "grad_norm": 3.227731227874756,
      "learning_rate": 5.793743137608364e-07,
      "loss": 0.7663,
      "step": 28880
    },
    {
      "epoch": 0.44968746837709295,
      "grad_norm": 4.241919994354248,
      "learning_rate": 5.792923747562314e-07,
      "loss": 0.7926,
      "step": 28885
    },
    {
      "epoch": 0.44976530937906234,
      "grad_norm": 3.0052309036254883,
      "learning_rate": 5.792104357516264e-07,
      "loss": 0.7092,
      "step": 28890
    },
    {
      "epoch": 0.44984315038103173,
      "grad_norm": 4.822492599487305,
      "learning_rate": 5.791284967470215e-07,
      "loss": 0.781,
      "step": 28895
    },
    {
      "epoch": 0.44992099138300107,
      "grad_norm": 7.8753180503845215,
      "learning_rate": 5.790465577424165e-07,
      "loss": 0.7651,
      "step": 28900
    },
    {
      "epoch": 0.44999883238497046,
      "grad_norm": 3.306781053543091,
      "learning_rate": 5.789646187378115e-07,
      "loss": 0.7367,
      "step": 28905
    },
    {
      "epoch": 0.45007667338693985,
      "grad_norm": 3.1504251956939697,
      "learning_rate": 5.788826797332066e-07,
      "loss": 0.706,
      "step": 28910
    },
    {
      "epoch": 0.4501545143889092,
      "grad_norm": 4.191858291625977,
      "learning_rate": 5.788007407286017e-07,
      "loss": 0.7441,
      "step": 28915
    },
    {
      "epoch": 0.4502323553908786,
      "grad_norm": 3.7259771823883057,
      "learning_rate": 5.787188017239966e-07,
      "loss": 0.8353,
      "step": 28920
    },
    {
      "epoch": 0.450310196392848,
      "grad_norm": 4.479262351989746,
      "learning_rate": 5.786368627193916e-07,
      "loss": 0.7023,
      "step": 28925
    },
    {
      "epoch": 0.45038803739481736,
      "grad_norm": 4.838881015777588,
      "learning_rate": 5.785549237147867e-07,
      "loss": 0.7328,
      "step": 28930
    },
    {
      "epoch": 0.4504658783967867,
      "grad_norm": 3.202648401260376,
      "learning_rate": 5.784729847101817e-07,
      "loss": 0.8272,
      "step": 28935
    },
    {
      "epoch": 0.4505437193987561,
      "grad_norm": 3.269976854324341,
      "learning_rate": 5.783910457055767e-07,
      "loss": 0.7599,
      "step": 28940
    },
    {
      "epoch": 0.4506215604007255,
      "grad_norm": 3.516141653060913,
      "learning_rate": 5.783091067009718e-07,
      "loss": 0.7602,
      "step": 28945
    },
    {
      "epoch": 0.4506994014026949,
      "grad_norm": 4.920629501342773,
      "learning_rate": 5.782271676963669e-07,
      "loss": 0.7543,
      "step": 28950
    },
    {
      "epoch": 0.4507772424046642,
      "grad_norm": 5.533834457397461,
      "learning_rate": 5.781452286917618e-07,
      "loss": 0.7442,
      "step": 28955
    },
    {
      "epoch": 0.4508550834066336,
      "grad_norm": 5.30636739730835,
      "learning_rate": 5.780632896871569e-07,
      "loss": 0.676,
      "step": 28960
    },
    {
      "epoch": 0.450932924408603,
      "grad_norm": 10.619142532348633,
      "learning_rate": 5.779813506825519e-07,
      "loss": 0.8129,
      "step": 28965
    },
    {
      "epoch": 0.4510107654105724,
      "grad_norm": 6.480765342712402,
      "learning_rate": 5.778994116779468e-07,
      "loss": 0.7446,
      "step": 28970
    },
    {
      "epoch": 0.4510886064125417,
      "grad_norm": 5.204502582550049,
      "learning_rate": 5.778174726733419e-07,
      "loss": 0.7069,
      "step": 28975
    },
    {
      "epoch": 0.4511664474145111,
      "grad_norm": 4.343836784362793,
      "learning_rate": 5.77735533668737e-07,
      "loss": 0.7275,
      "step": 28980
    },
    {
      "epoch": 0.4512442884164805,
      "grad_norm": 3.211695671081543,
      "learning_rate": 5.77653594664132e-07,
      "loss": 0.7882,
      "step": 28985
    },
    {
      "epoch": 0.45132212941844985,
      "grad_norm": 4.3351569175720215,
      "learning_rate": 5.77571655659527e-07,
      "loss": 0.8266,
      "step": 28990
    },
    {
      "epoch": 0.45139997042041924,
      "grad_norm": 2.9921987056732178,
      "learning_rate": 5.774897166549221e-07,
      "loss": 0.7562,
      "step": 28995
    },
    {
      "epoch": 0.45147781142238863,
      "grad_norm": 4.759602069854736,
      "learning_rate": 5.774077776503171e-07,
      "loss": 0.8041,
      "step": 29000
    },
    {
      "epoch": 0.451555652424358,
      "grad_norm": 3.083439826965332,
      "learning_rate": 5.773258386457122e-07,
      "loss": 0.7023,
      "step": 29005
    },
    {
      "epoch": 0.45163349342632736,
      "grad_norm": 8.841672897338867,
      "learning_rate": 5.772438996411071e-07,
      "loss": 0.729,
      "step": 29010
    },
    {
      "epoch": 0.45171133442829675,
      "grad_norm": 3.8009612560272217,
      "learning_rate": 5.771619606365021e-07,
      "loss": 0.8011,
      "step": 29015
    },
    {
      "epoch": 0.45178917543026614,
      "grad_norm": 16.207178115844727,
      "learning_rate": 5.770800216318972e-07,
      "loss": 0.8641,
      "step": 29020
    },
    {
      "epoch": 0.45186701643223554,
      "grad_norm": 12.364839553833008,
      "learning_rate": 5.769980826272922e-07,
      "loss": 0.772,
      "step": 29025
    },
    {
      "epoch": 0.45194485743420487,
      "grad_norm": 8.666084289550781,
      "learning_rate": 5.769161436226872e-07,
      "loss": 0.8252,
      "step": 29030
    },
    {
      "epoch": 0.45202269843617426,
      "grad_norm": 4.036910057067871,
      "learning_rate": 5.768342046180823e-07,
      "loss": 0.6861,
      "step": 29035
    },
    {
      "epoch": 0.45210053943814366,
      "grad_norm": 4.029730796813965,
      "learning_rate": 5.767522656134774e-07,
      "loss": 0.6195,
      "step": 29040
    },
    {
      "epoch": 0.45217838044011305,
      "grad_norm": 5.024090766906738,
      "learning_rate": 5.766703266088723e-07,
      "loss": 0.6818,
      "step": 29045
    },
    {
      "epoch": 0.4522562214420824,
      "grad_norm": 4.02849817276001,
      "learning_rate": 5.765883876042673e-07,
      "loss": 0.8053,
      "step": 29050
    },
    {
      "epoch": 0.4523340624440518,
      "grad_norm": 4.624151229858398,
      "learning_rate": 5.765064485996624e-07,
      "loss": 0.8042,
      "step": 29055
    },
    {
      "epoch": 0.45241190344602117,
      "grad_norm": 24.174957275390625,
      "learning_rate": 5.764245095950574e-07,
      "loss": 0.7939,
      "step": 29060
    },
    {
      "epoch": 0.45248974444799056,
      "grad_norm": 4.263962745666504,
      "learning_rate": 5.763425705904524e-07,
      "loss": 0.6899,
      "step": 29065
    },
    {
      "epoch": 0.4525675854499599,
      "grad_norm": 2.8683125972747803,
      "learning_rate": 5.762606315858475e-07,
      "loss": 0.7562,
      "step": 29070
    },
    {
      "epoch": 0.4526454264519293,
      "grad_norm": 5.541211128234863,
      "learning_rate": 5.761786925812426e-07,
      "loss": 0.7891,
      "step": 29075
    },
    {
      "epoch": 0.4527232674538987,
      "grad_norm": 6.852053642272949,
      "learning_rate": 5.760967535766375e-07,
      "loss": 0.7556,
      "step": 29080
    },
    {
      "epoch": 0.452801108455868,
      "grad_norm": 3.5538642406463623,
      "learning_rate": 5.760148145720326e-07,
      "loss": 0.8761,
      "step": 29085
    },
    {
      "epoch": 0.4528789494578374,
      "grad_norm": 4.408565998077393,
      "learning_rate": 5.759328755674276e-07,
      "loss": 0.712,
      "step": 29090
    },
    {
      "epoch": 0.4529567904598068,
      "grad_norm": 5.765478610992432,
      "learning_rate": 5.758509365628225e-07,
      "loss": 0.7246,
      "step": 29095
    },
    {
      "epoch": 0.4530346314617762,
      "grad_norm": 4.650556564331055,
      "learning_rate": 5.757689975582176e-07,
      "loss": 0.8426,
      "step": 29100
    },
    {
      "epoch": 0.45311247246374553,
      "grad_norm": 4.318330764770508,
      "learning_rate": 5.756870585536127e-07,
      "loss": 0.7968,
      "step": 29105
    },
    {
      "epoch": 0.4531903134657149,
      "grad_norm": 4.4654388427734375,
      "learning_rate": 5.756051195490077e-07,
      "loss": 0.7472,
      "step": 29110
    },
    {
      "epoch": 0.4532681544676843,
      "grad_norm": 6.117023468017578,
      "learning_rate": 5.755231805444027e-07,
      "loss": 0.7138,
      "step": 29115
    },
    {
      "epoch": 0.4533459954696537,
      "grad_norm": 7.660935401916504,
      "learning_rate": 5.754412415397978e-07,
      "loss": 0.808,
      "step": 29120
    },
    {
      "epoch": 0.45342383647162304,
      "grad_norm": 5.994586944580078,
      "learning_rate": 5.753593025351928e-07,
      "loss": 0.8673,
      "step": 29125
    },
    {
      "epoch": 0.45350167747359243,
      "grad_norm": 3.525341033935547,
      "learning_rate": 5.752773635305878e-07,
      "loss": 0.7404,
      "step": 29130
    },
    {
      "epoch": 0.4535795184755618,
      "grad_norm": 3.208970308303833,
      "learning_rate": 5.751954245259828e-07,
      "loss": 0.7721,
      "step": 29135
    },
    {
      "epoch": 0.4536573594775312,
      "grad_norm": 3.5546417236328125,
      "learning_rate": 5.751134855213778e-07,
      "loss": 0.8085,
      "step": 29140
    },
    {
      "epoch": 0.45373520047950056,
      "grad_norm": 3.8426883220672607,
      "learning_rate": 5.750315465167729e-07,
      "loss": 0.8667,
      "step": 29145
    },
    {
      "epoch": 0.45381304148146995,
      "grad_norm": 4.148952484130859,
      "learning_rate": 5.74949607512168e-07,
      "loss": 0.8004,
      "step": 29150
    },
    {
      "epoch": 0.45389088248343934,
      "grad_norm": 4.1008992195129395,
      "learning_rate": 5.748676685075629e-07,
      "loss": 0.7331,
      "step": 29155
    },
    {
      "epoch": 0.45396872348540873,
      "grad_norm": 5.24363899230957,
      "learning_rate": 5.74785729502958e-07,
      "loss": 0.8068,
      "step": 29160
    },
    {
      "epoch": 0.45404656448737807,
      "grad_norm": 3.7679924964904785,
      "learning_rate": 5.747037904983531e-07,
      "loss": 0.7042,
      "step": 29165
    },
    {
      "epoch": 0.45412440548934746,
      "grad_norm": 6.450275897979736,
      "learning_rate": 5.746218514937479e-07,
      "loss": 0.725,
      "step": 29170
    },
    {
      "epoch": 0.45420224649131685,
      "grad_norm": 4.289034366607666,
      "learning_rate": 5.74539912489143e-07,
      "loss": 0.8314,
      "step": 29175
    },
    {
      "epoch": 0.4542800874932862,
      "grad_norm": 4.317148208618164,
      "learning_rate": 5.744579734845381e-07,
      "loss": 0.8628,
      "step": 29180
    },
    {
      "epoch": 0.4543579284952556,
      "grad_norm": 3.4826083183288574,
      "learning_rate": 5.743760344799332e-07,
      "loss": 0.7807,
      "step": 29185
    },
    {
      "epoch": 0.454435769497225,
      "grad_norm": 4.1573872566223145,
      "learning_rate": 5.742940954753281e-07,
      "loss": 0.7241,
      "step": 29190
    },
    {
      "epoch": 0.45451361049919436,
      "grad_norm": 2.6551735401153564,
      "learning_rate": 5.742121564707232e-07,
      "loss": 0.7656,
      "step": 29195
    },
    {
      "epoch": 0.4545914515011637,
      "grad_norm": 3.7584831714630127,
      "learning_rate": 5.741302174661183e-07,
      "loss": 0.693,
      "step": 29200
    },
    {
      "epoch": 0.4546692925031331,
      "grad_norm": 3.4537553787231445,
      "learning_rate": 5.740482784615132e-07,
      "loss": 0.7632,
      "step": 29205
    },
    {
      "epoch": 0.4547471335051025,
      "grad_norm": 9.524008750915527,
      "learning_rate": 5.739663394569082e-07,
      "loss": 0.781,
      "step": 29210
    },
    {
      "epoch": 0.4548249745070719,
      "grad_norm": 4.735363960266113,
      "learning_rate": 5.738844004523033e-07,
      "loss": 0.7214,
      "step": 29215
    },
    {
      "epoch": 0.4549028155090412,
      "grad_norm": 4.976006031036377,
      "learning_rate": 5.738024614476983e-07,
      "loss": 0.6781,
      "step": 29220
    },
    {
      "epoch": 0.4549806565110106,
      "grad_norm": 2.9076504707336426,
      "learning_rate": 5.737205224430933e-07,
      "loss": 0.7562,
      "step": 29225
    },
    {
      "epoch": 0.45505849751298,
      "grad_norm": 6.729519844055176,
      "learning_rate": 5.736385834384884e-07,
      "loss": 0.7656,
      "step": 29230
    },
    {
      "epoch": 0.4551363385149494,
      "grad_norm": 3.2870726585388184,
      "learning_rate": 5.735566444338834e-07,
      "loss": 0.753,
      "step": 29235
    },
    {
      "epoch": 0.4552141795169187,
      "grad_norm": 4.06284236907959,
      "learning_rate": 5.734747054292784e-07,
      "loss": 0.8596,
      "step": 29240
    },
    {
      "epoch": 0.4552920205188881,
      "grad_norm": 3.725203514099121,
      "learning_rate": 5.733927664246735e-07,
      "loss": 0.6136,
      "step": 29245
    },
    {
      "epoch": 0.4553698615208575,
      "grad_norm": 2.6347107887268066,
      "learning_rate": 5.733108274200684e-07,
      "loss": 0.7469,
      "step": 29250
    },
    {
      "epoch": 0.45544770252282685,
      "grad_norm": 4.7031636238098145,
      "learning_rate": 5.732288884154635e-07,
      "loss": 0.8537,
      "step": 29255
    },
    {
      "epoch": 0.45552554352479624,
      "grad_norm": 2.4778549671173096,
      "learning_rate": 5.731469494108585e-07,
      "loss": 0.8185,
      "step": 29260
    },
    {
      "epoch": 0.45560338452676563,
      "grad_norm": 4.985352516174316,
      "learning_rate": 5.730650104062535e-07,
      "loss": 0.8419,
      "step": 29265
    },
    {
      "epoch": 0.455681225528735,
      "grad_norm": 3.5948269367218018,
      "learning_rate": 5.729830714016486e-07,
      "loss": 0.7006,
      "step": 29270
    },
    {
      "epoch": 0.45575906653070436,
      "grad_norm": 3.4470608234405518,
      "learning_rate": 5.729011323970437e-07,
      "loss": 0.797,
      "step": 29275
    },
    {
      "epoch": 0.45583690753267375,
      "grad_norm": 4.510948657989502,
      "learning_rate": 5.728191933924386e-07,
      "loss": 0.5894,
      "step": 29280
    },
    {
      "epoch": 0.45591474853464314,
      "grad_norm": 3.0016355514526367,
      "learning_rate": 5.727372543878337e-07,
      "loss": 0.7484,
      "step": 29285
    },
    {
      "epoch": 0.45599258953661254,
      "grad_norm": 3.7839488983154297,
      "learning_rate": 5.726553153832287e-07,
      "loss": 0.6431,
      "step": 29290
    },
    {
      "epoch": 0.4560704305385819,
      "grad_norm": 3.340162515640259,
      "learning_rate": 5.725733763786236e-07,
      "loss": 0.734,
      "step": 29295
    },
    {
      "epoch": 0.45614827154055126,
      "grad_norm": 3.7294671535491943,
      "learning_rate": 5.724914373740187e-07,
      "loss": 0.7254,
      "step": 29300
    },
    {
      "epoch": 0.45622611254252066,
      "grad_norm": 3.578665018081665,
      "learning_rate": 5.724094983694138e-07,
      "loss": 0.7948,
      "step": 29305
    },
    {
      "epoch": 0.45630395354449005,
      "grad_norm": 4.261101722717285,
      "learning_rate": 5.723275593648089e-07,
      "loss": 0.7266,
      "step": 29310
    },
    {
      "epoch": 0.4563817945464594,
      "grad_norm": 5.133454322814941,
      "learning_rate": 5.722456203602038e-07,
      "loss": 0.7848,
      "step": 29315
    },
    {
      "epoch": 0.4564596355484288,
      "grad_norm": 2.6158576011657715,
      "learning_rate": 5.721636813555989e-07,
      "loss": 0.7003,
      "step": 29320
    },
    {
      "epoch": 0.45653747655039817,
      "grad_norm": 3.4494142532348633,
      "learning_rate": 5.72081742350994e-07,
      "loss": 0.7098,
      "step": 29325
    },
    {
      "epoch": 0.45661531755236756,
      "grad_norm": 4.859612941741943,
      "learning_rate": 5.71999803346389e-07,
      "loss": 0.8115,
      "step": 29330
    },
    {
      "epoch": 0.4566931585543369,
      "grad_norm": 6.412881374359131,
      "learning_rate": 5.719178643417839e-07,
      "loss": 0.7827,
      "step": 29335
    },
    {
      "epoch": 0.4567709995563063,
      "grad_norm": 2.4222824573516846,
      "learning_rate": 5.71835925337179e-07,
      "loss": 0.7147,
      "step": 29340
    },
    {
      "epoch": 0.4568488405582757,
      "grad_norm": 3.288569927215576,
      "learning_rate": 5.71753986332574e-07,
      "loss": 0.8637,
      "step": 29345
    },
    {
      "epoch": 0.456926681560245,
      "grad_norm": 3.8363094329833984,
      "learning_rate": 5.71672047327969e-07,
      "loss": 0.627,
      "step": 29350
    },
    {
      "epoch": 0.4570045225622144,
      "grad_norm": 2.8272054195404053,
      "learning_rate": 5.715901083233641e-07,
      "loss": 0.7331,
      "step": 29355
    },
    {
      "epoch": 0.4570823635641838,
      "grad_norm": 7.778134822845459,
      "learning_rate": 5.715081693187591e-07,
      "loss": 0.6883,
      "step": 29360
    },
    {
      "epoch": 0.4571602045661532,
      "grad_norm": 2.7845077514648438,
      "learning_rate": 5.714262303141542e-07,
      "loss": 0.782,
      "step": 29365
    },
    {
      "epoch": 0.45723804556812253,
      "grad_norm": 3.7273714542388916,
      "learning_rate": 5.713442913095492e-07,
      "loss": 0.7656,
      "step": 29370
    },
    {
      "epoch": 0.4573158865700919,
      "grad_norm": 4.313918590545654,
      "learning_rate": 5.712623523049441e-07,
      "loss": 0.7244,
      "step": 29375
    },
    {
      "epoch": 0.4573937275720613,
      "grad_norm": 2.5442774295806885,
      "learning_rate": 5.711804133003392e-07,
      "loss": 0.6705,
      "step": 29380
    },
    {
      "epoch": 0.4574715685740307,
      "grad_norm": 2.8726694583892822,
      "learning_rate": 5.710984742957342e-07,
      "loss": 0.7727,
      "step": 29385
    },
    {
      "epoch": 0.45754940957600004,
      "grad_norm": 4.24089241027832,
      "learning_rate": 5.710165352911292e-07,
      "loss": 0.6543,
      "step": 29390
    },
    {
      "epoch": 0.45762725057796944,
      "grad_norm": 2.787386417388916,
      "learning_rate": 5.709345962865243e-07,
      "loss": 0.7181,
      "step": 29395
    },
    {
      "epoch": 0.4577050915799388,
      "grad_norm": 4.498680114746094,
      "learning_rate": 5.708526572819194e-07,
      "loss": 0.846,
      "step": 29400
    },
    {
      "epoch": 0.4577829325819082,
      "grad_norm": 4.32037878036499,
      "learning_rate": 5.707707182773143e-07,
      "loss": 0.8141,
      "step": 29405
    },
    {
      "epoch": 0.45786077358387756,
      "grad_norm": 3.402519464492798,
      "learning_rate": 5.706887792727094e-07,
      "loss": 0.8875,
      "step": 29410
    },
    {
      "epoch": 0.45793861458584695,
      "grad_norm": 3.7643373012542725,
      "learning_rate": 5.706068402681044e-07,
      "loss": 0.7522,
      "step": 29415
    },
    {
      "epoch": 0.45801645558781634,
      "grad_norm": 3.1787290573120117,
      "learning_rate": 5.705249012634993e-07,
      "loss": 0.6755,
      "step": 29420
    },
    {
      "epoch": 0.4580942965897857,
      "grad_norm": 6.672213554382324,
      "learning_rate": 5.704429622588944e-07,
      "loss": 0.8007,
      "step": 29425
    },
    {
      "epoch": 0.45817213759175507,
      "grad_norm": 4.505547523498535,
      "learning_rate": 5.703610232542895e-07,
      "loss": 0.8463,
      "step": 29430
    },
    {
      "epoch": 0.45824997859372446,
      "grad_norm": 3.9626431465148926,
      "learning_rate": 5.702790842496846e-07,
      "loss": 0.7459,
      "step": 29435
    },
    {
      "epoch": 0.45832781959569385,
      "grad_norm": 3.782198905944824,
      "learning_rate": 5.701971452450795e-07,
      "loss": 0.847,
      "step": 29440
    },
    {
      "epoch": 0.4584056605976632,
      "grad_norm": 6.641538143157959,
      "learning_rate": 5.701152062404746e-07,
      "loss": 0.8242,
      "step": 29445
    },
    {
      "epoch": 0.4584835015996326,
      "grad_norm": 3.9570329189300537,
      "learning_rate": 5.700332672358697e-07,
      "loss": 0.6582,
      "step": 29450
    },
    {
      "epoch": 0.458561342601602,
      "grad_norm": 2.9618704319000244,
      "learning_rate": 5.699513282312645e-07,
      "loss": 0.6949,
      "step": 29455
    },
    {
      "epoch": 0.45863918360357137,
      "grad_norm": 3.5960450172424316,
      "learning_rate": 5.698693892266596e-07,
      "loss": 0.7812,
      "step": 29460
    },
    {
      "epoch": 0.4587170246055407,
      "grad_norm": 3.4689462184906006,
      "learning_rate": 5.697874502220547e-07,
      "loss": 0.7432,
      "step": 29465
    },
    {
      "epoch": 0.4587948656075101,
      "grad_norm": 3.1630213260650635,
      "learning_rate": 5.697055112174497e-07,
      "loss": 0.7249,
      "step": 29470
    },
    {
      "epoch": 0.4588727066094795,
      "grad_norm": 4.015108585357666,
      "learning_rate": 5.696235722128447e-07,
      "loss": 0.7426,
      "step": 29475
    },
    {
      "epoch": 0.4589505476114489,
      "grad_norm": 3.334855079650879,
      "learning_rate": 5.695416332082398e-07,
      "loss": 0.6557,
      "step": 29480
    },
    {
      "epoch": 0.4590283886134182,
      "grad_norm": 5.416299819946289,
      "learning_rate": 5.694596942036348e-07,
      "loss": 0.8126,
      "step": 29485
    },
    {
      "epoch": 0.4591062296153876,
      "grad_norm": 5.627259254455566,
      "learning_rate": 5.693777551990299e-07,
      "loss": 0.7561,
      "step": 29490
    },
    {
      "epoch": 0.459184070617357,
      "grad_norm": 4.5271477699279785,
      "learning_rate": 5.692958161944248e-07,
      "loss": 0.6919,
      "step": 29495
    },
    {
      "epoch": 0.4592619116193264,
      "grad_norm": 5.4130167961120605,
      "learning_rate": 5.692138771898198e-07,
      "loss": 0.8622,
      "step": 29500
    },
    {
      "epoch": 0.4593397526212957,
      "grad_norm": 4.641055583953857,
      "learning_rate": 5.691319381852149e-07,
      "loss": 0.7803,
      "step": 29505
    },
    {
      "epoch": 0.4594175936232651,
      "grad_norm": 5.009194850921631,
      "learning_rate": 5.6904999918061e-07,
      "loss": 0.8414,
      "step": 29510
    },
    {
      "epoch": 0.4594954346252345,
      "grad_norm": 3.12497878074646,
      "learning_rate": 5.689680601760049e-07,
      "loss": 0.7733,
      "step": 29515
    },
    {
      "epoch": 0.45957327562720385,
      "grad_norm": 4.629786014556885,
      "learning_rate": 5.688861211714e-07,
      "loss": 0.8534,
      "step": 29520
    },
    {
      "epoch": 0.45965111662917324,
      "grad_norm": 3.6191458702087402,
      "learning_rate": 5.688041821667951e-07,
      "loss": 0.8469,
      "step": 29525
    },
    {
      "epoch": 0.45972895763114263,
      "grad_norm": 3.2484846115112305,
      "learning_rate": 5.6872224316219e-07,
      "loss": 0.83,
      "step": 29530
    },
    {
      "epoch": 0.459806798633112,
      "grad_norm": 5.290815830230713,
      "learning_rate": 5.68640304157585e-07,
      "loss": 0.6943,
      "step": 29535
    },
    {
      "epoch": 0.45988463963508136,
      "grad_norm": 4.702110290527344,
      "learning_rate": 5.685583651529801e-07,
      "loss": 0.8024,
      "step": 29540
    },
    {
      "epoch": 0.45996248063705075,
      "grad_norm": 5.266806125640869,
      "learning_rate": 5.68476426148375e-07,
      "loss": 0.7295,
      "step": 29545
    },
    {
      "epoch": 0.46004032163902014,
      "grad_norm": 3.227226972579956,
      "learning_rate": 5.683944871437701e-07,
      "loss": 0.7055,
      "step": 29550
    },
    {
      "epoch": 0.46011816264098954,
      "grad_norm": 3.443768262863159,
      "learning_rate": 5.683125481391652e-07,
      "loss": 0.7523,
      "step": 29555
    },
    {
      "epoch": 0.4601960036429589,
      "grad_norm": 3.3546595573425293,
      "learning_rate": 5.682306091345603e-07,
      "loss": 0.7309,
      "step": 29560
    },
    {
      "epoch": 0.46027384464492826,
      "grad_norm": 6.671410083770752,
      "learning_rate": 5.681486701299552e-07,
      "loss": 0.8336,
      "step": 29565
    },
    {
      "epoch": 0.46035168564689766,
      "grad_norm": 6.067061424255371,
      "learning_rate": 5.680667311253503e-07,
      "loss": 0.7152,
      "step": 29570
    },
    {
      "epoch": 0.46042952664886705,
      "grad_norm": 8.262660026550293,
      "learning_rate": 5.679847921207453e-07,
      "loss": 0.7065,
      "step": 29575
    },
    {
      "epoch": 0.4605073676508364,
      "grad_norm": 2.8983047008514404,
      "learning_rate": 5.679028531161403e-07,
      "loss": 0.8039,
      "step": 29580
    },
    {
      "epoch": 0.4605852086528058,
      "grad_norm": 3.577810525894165,
      "learning_rate": 5.678209141115353e-07,
      "loss": 0.7898,
      "step": 29585
    },
    {
      "epoch": 0.46066304965477517,
      "grad_norm": 3.0140891075134277,
      "learning_rate": 5.677389751069304e-07,
      "loss": 0.8619,
      "step": 29590
    },
    {
      "epoch": 0.4607408906567445,
      "grad_norm": 3.443023204803467,
      "learning_rate": 5.676570361023254e-07,
      "loss": 0.7451,
      "step": 29595
    },
    {
      "epoch": 0.4608187316587139,
      "grad_norm": 7.99015998840332,
      "learning_rate": 5.675750970977205e-07,
      "loss": 0.8006,
      "step": 29600
    },
    {
      "epoch": 0.4608965726606833,
      "grad_norm": 4.371585845947266,
      "learning_rate": 5.674931580931155e-07,
      "loss": 0.7726,
      "step": 29605
    },
    {
      "epoch": 0.4609744136626527,
      "grad_norm": 3.129302501678467,
      "learning_rate": 5.674112190885105e-07,
      "loss": 0.7019,
      "step": 29610
    },
    {
      "epoch": 0.461052254664622,
      "grad_norm": 2.687493324279785,
      "learning_rate": 5.673292800839056e-07,
      "loss": 0.6881,
      "step": 29615
    },
    {
      "epoch": 0.4611300956665914,
      "grad_norm": 9.908051490783691,
      "learning_rate": 5.672473410793005e-07,
      "loss": 0.8048,
      "step": 29620
    },
    {
      "epoch": 0.4612079366685608,
      "grad_norm": 4.389907360076904,
      "learning_rate": 5.671654020746955e-07,
      "loss": 0.7758,
      "step": 29625
    },
    {
      "epoch": 0.4612857776705302,
      "grad_norm": 3.2980127334594727,
      "learning_rate": 5.670834630700906e-07,
      "loss": 0.7126,
      "step": 29630
    },
    {
      "epoch": 0.46136361867249953,
      "grad_norm": 3.6595046520233154,
      "learning_rate": 5.670015240654857e-07,
      "loss": 0.8298,
      "step": 29635
    },
    {
      "epoch": 0.4614414596744689,
      "grad_norm": 10.296154022216797,
      "learning_rate": 5.669195850608806e-07,
      "loss": 0.6761,
      "step": 29640
    },
    {
      "epoch": 0.4615193006764383,
      "grad_norm": 3.4469966888427734,
      "learning_rate": 5.668376460562757e-07,
      "loss": 0.7524,
      "step": 29645
    },
    {
      "epoch": 0.4615971416784077,
      "grad_norm": 3.731187582015991,
      "learning_rate": 5.667557070516708e-07,
      "loss": 0.7712,
      "step": 29650
    },
    {
      "epoch": 0.46167498268037704,
      "grad_norm": 2.4918253421783447,
      "learning_rate": 5.666737680470657e-07,
      "loss": 0.6676,
      "step": 29655
    },
    {
      "epoch": 0.46175282368234644,
      "grad_norm": 9.466460227966309,
      "learning_rate": 5.665918290424607e-07,
      "loss": 0.8928,
      "step": 29660
    },
    {
      "epoch": 0.46183066468431583,
      "grad_norm": 2.682614803314209,
      "learning_rate": 5.665098900378558e-07,
      "loss": 0.6856,
      "step": 29665
    },
    {
      "epoch": 0.4619085056862852,
      "grad_norm": 3.7968413829803467,
      "learning_rate": 5.664279510332508e-07,
      "loss": 0.8142,
      "step": 29670
    },
    {
      "epoch": 0.46198634668825456,
      "grad_norm": 3.2053468227386475,
      "learning_rate": 5.663460120286458e-07,
      "loss": 0.8449,
      "step": 29675
    },
    {
      "epoch": 0.46206418769022395,
      "grad_norm": 3.584822177886963,
      "learning_rate": 5.662640730240409e-07,
      "loss": 0.7932,
      "step": 29680
    },
    {
      "epoch": 0.46214202869219334,
      "grad_norm": 5.949429512023926,
      "learning_rate": 5.66182134019436e-07,
      "loss": 0.7658,
      "step": 29685
    },
    {
      "epoch": 0.4622198696941627,
      "grad_norm": 5.678579330444336,
      "learning_rate": 5.66100195014831e-07,
      "loss": 0.8584,
      "step": 29690
    },
    {
      "epoch": 0.46229771069613207,
      "grad_norm": 3.8462507724761963,
      "learning_rate": 5.66018256010226e-07,
      "loss": 0.856,
      "step": 29695
    },
    {
      "epoch": 0.46237555169810146,
      "grad_norm": 2.270792245864868,
      "learning_rate": 5.65936317005621e-07,
      "loss": 0.6093,
      "step": 29700
    },
    {
      "epoch": 0.46245339270007085,
      "grad_norm": 2.746656894683838,
      "learning_rate": 5.65854378001016e-07,
      "loss": 0.7386,
      "step": 29705
    },
    {
      "epoch": 0.4625312337020402,
      "grad_norm": 4.786650657653809,
      "learning_rate": 5.65772438996411e-07,
      "loss": 0.8235,
      "step": 29710
    },
    {
      "epoch": 0.4626090747040096,
      "grad_norm": 7.245391368865967,
      "learning_rate": 5.656904999918061e-07,
      "loss": 0.7279,
      "step": 29715
    },
    {
      "epoch": 0.462686915705979,
      "grad_norm": 5.278894901275635,
      "learning_rate": 5.656085609872011e-07,
      "loss": 0.8377,
      "step": 29720
    },
    {
      "epoch": 0.46276475670794837,
      "grad_norm": 3.9391064643859863,
      "learning_rate": 5.655266219825962e-07,
      "loss": 0.7426,
      "step": 29725
    },
    {
      "epoch": 0.4628425977099177,
      "grad_norm": 5.838922500610352,
      "learning_rate": 5.654446829779912e-07,
      "loss": 0.7045,
      "step": 29730
    },
    {
      "epoch": 0.4629204387118871,
      "grad_norm": 3.234442710876465,
      "learning_rate": 5.653627439733862e-07,
      "loss": 0.8333,
      "step": 29735
    },
    {
      "epoch": 0.4629982797138565,
      "grad_norm": 2.542278528213501,
      "learning_rate": 5.652808049687812e-07,
      "loss": 0.6799,
      "step": 29740
    },
    {
      "epoch": 0.4630761207158259,
      "grad_norm": 4.046177387237549,
      "learning_rate": 5.651988659641762e-07,
      "loss": 0.7613,
      "step": 29745
    },
    {
      "epoch": 0.4631539617177952,
      "grad_norm": 3.80517315864563,
      "learning_rate": 5.651169269595712e-07,
      "loss": 0.7418,
      "step": 29750
    },
    {
      "epoch": 0.4632318027197646,
      "grad_norm": 5.722404956817627,
      "learning_rate": 5.650349879549663e-07,
      "loss": 0.7325,
      "step": 29755
    },
    {
      "epoch": 0.463309643721734,
      "grad_norm": 3.710421085357666,
      "learning_rate": 5.649530489503614e-07,
      "loss": 0.845,
      "step": 29760
    },
    {
      "epoch": 0.4633874847237034,
      "grad_norm": 3.7477540969848633,
      "learning_rate": 5.648711099457563e-07,
      "loss": 0.6633,
      "step": 29765
    },
    {
      "epoch": 0.4634653257256727,
      "grad_norm": 3.7613565921783447,
      "learning_rate": 5.647891709411514e-07,
      "loss": 0.8116,
      "step": 29770
    },
    {
      "epoch": 0.4635431667276421,
      "grad_norm": 3.897918224334717,
      "learning_rate": 5.647072319365465e-07,
      "loss": 0.7957,
      "step": 29775
    },
    {
      "epoch": 0.4636210077296115,
      "grad_norm": 4.354563236236572,
      "learning_rate": 5.646252929319413e-07,
      "loss": 0.8552,
      "step": 29780
    },
    {
      "epoch": 0.46369884873158085,
      "grad_norm": 4.190893173217773,
      "learning_rate": 5.645433539273364e-07,
      "loss": 0.6798,
      "step": 29785
    },
    {
      "epoch": 0.46377668973355024,
      "grad_norm": 6.748042583465576,
      "learning_rate": 5.644614149227315e-07,
      "loss": 0.7531,
      "step": 29790
    },
    {
      "epoch": 0.46385453073551963,
      "grad_norm": 3.1443727016448975,
      "learning_rate": 5.643794759181265e-07,
      "loss": 0.6858,
      "step": 29795
    },
    {
      "epoch": 0.463932371737489,
      "grad_norm": 4.152675151824951,
      "learning_rate": 5.642975369135215e-07,
      "loss": 0.7706,
      "step": 29800
    },
    {
      "epoch": 0.46401021273945836,
      "grad_norm": 3.2192883491516113,
      "learning_rate": 5.642155979089166e-07,
      "loss": 0.8153,
      "step": 29805
    },
    {
      "epoch": 0.46408805374142775,
      "grad_norm": 3.3782994747161865,
      "learning_rate": 5.641336589043117e-07,
      "loss": 0.6968,
      "step": 29810
    },
    {
      "epoch": 0.46416589474339714,
      "grad_norm": 2.9708139896392822,
      "learning_rate": 5.640517198997067e-07,
      "loss": 0.7539,
      "step": 29815
    },
    {
      "epoch": 0.46424373574536654,
      "grad_norm": 2.698129177093506,
      "learning_rate": 5.639697808951016e-07,
      "loss": 0.7498,
      "step": 29820
    },
    {
      "epoch": 0.4643215767473359,
      "grad_norm": 3.8263444900512695,
      "learning_rate": 5.638878418904967e-07,
      "loss": 0.6767,
      "step": 29825
    },
    {
      "epoch": 0.46439941774930527,
      "grad_norm": 3.469433069229126,
      "learning_rate": 5.638059028858917e-07,
      "loss": 0.6213,
      "step": 29830
    },
    {
      "epoch": 0.46447725875127466,
      "grad_norm": 10.169317245483398,
      "learning_rate": 5.637239638812867e-07,
      "loss": 0.7964,
      "step": 29835
    },
    {
      "epoch": 0.46455509975324405,
      "grad_norm": 3.737732410430908,
      "learning_rate": 5.636420248766818e-07,
      "loss": 0.6369,
      "step": 29840
    },
    {
      "epoch": 0.4646329407552134,
      "grad_norm": 4.020811080932617,
      "learning_rate": 5.635600858720768e-07,
      "loss": 0.7707,
      "step": 29845
    },
    {
      "epoch": 0.4647107817571828,
      "grad_norm": 3.8647196292877197,
      "learning_rate": 5.634781468674719e-07,
      "loss": 0.68,
      "step": 29850
    },
    {
      "epoch": 0.46478862275915217,
      "grad_norm": 8.446564674377441,
      "learning_rate": 5.633962078628669e-07,
      "loss": 0.8873,
      "step": 29855
    },
    {
      "epoch": 0.4648664637611215,
      "grad_norm": 3.6714611053466797,
      "learning_rate": 5.633142688582618e-07,
      "loss": 0.6804,
      "step": 29860
    },
    {
      "epoch": 0.4649443047630909,
      "grad_norm": 4.163028717041016,
      "learning_rate": 5.632323298536569e-07,
      "loss": 0.7578,
      "step": 29865
    },
    {
      "epoch": 0.4650221457650603,
      "grad_norm": 3.9823803901672363,
      "learning_rate": 5.63150390849052e-07,
      "loss": 0.8175,
      "step": 29870
    },
    {
      "epoch": 0.4650999867670297,
      "grad_norm": 3.8202476501464844,
      "learning_rate": 5.630684518444469e-07,
      "loss": 0.734,
      "step": 29875
    },
    {
      "epoch": 0.465177827768999,
      "grad_norm": 8.29537296295166,
      "learning_rate": 5.62986512839842e-07,
      "loss": 0.7623,
      "step": 29880
    },
    {
      "epoch": 0.4652556687709684,
      "grad_norm": 3.22029709815979,
      "learning_rate": 5.629045738352371e-07,
      "loss": 0.9522,
      "step": 29885
    },
    {
      "epoch": 0.4653335097729378,
      "grad_norm": 3.0242271423339844,
      "learning_rate": 5.62822634830632e-07,
      "loss": 0.7163,
      "step": 29890
    },
    {
      "epoch": 0.4654113507749072,
      "grad_norm": 3.0410945415496826,
      "learning_rate": 5.627406958260271e-07,
      "loss": 0.7557,
      "step": 29895
    },
    {
      "epoch": 0.46548919177687653,
      "grad_norm": 4.94197940826416,
      "learning_rate": 5.626587568214221e-07,
      "loss": 0.6429,
      "step": 29900
    },
    {
      "epoch": 0.4655670327788459,
      "grad_norm": 3.7214863300323486,
      "learning_rate": 5.62576817816817e-07,
      "loss": 0.7183,
      "step": 29905
    },
    {
      "epoch": 0.4656448737808153,
      "grad_norm": 4.76154899597168,
      "learning_rate": 5.624948788122121e-07,
      "loss": 0.7022,
      "step": 29910
    },
    {
      "epoch": 0.4657227147827847,
      "grad_norm": 3.7771337032318115,
      "learning_rate": 5.624129398076072e-07,
      "loss": 0.6775,
      "step": 29915
    },
    {
      "epoch": 0.46580055578475404,
      "grad_norm": 2.8825271129608154,
      "learning_rate": 5.623310008030022e-07,
      "loss": 0.7057,
      "step": 29920
    },
    {
      "epoch": 0.46587839678672344,
      "grad_norm": 3.221827507019043,
      "learning_rate": 5.622490617983972e-07,
      "loss": 0.8056,
      "step": 29925
    },
    {
      "epoch": 0.46595623778869283,
      "grad_norm": 6.3704938888549805,
      "learning_rate": 5.621671227937923e-07,
      "loss": 0.8132,
      "step": 29930
    },
    {
      "epoch": 0.4660340787906622,
      "grad_norm": 3.94346284866333,
      "learning_rate": 5.620851837891874e-07,
      "loss": 0.6742,
      "step": 29935
    },
    {
      "epoch": 0.46611191979263156,
      "grad_norm": 2.9604880809783936,
      "learning_rate": 5.620032447845824e-07,
      "loss": 0.7277,
      "step": 29940
    },
    {
      "epoch": 0.46618976079460095,
      "grad_norm": 6.1392083168029785,
      "learning_rate": 5.619213057799773e-07,
      "loss": 0.7164,
      "step": 29945
    },
    {
      "epoch": 0.46626760179657034,
      "grad_norm": 2.6434457302093506,
      "learning_rate": 5.618393667753724e-07,
      "loss": 0.7129,
      "step": 29950
    },
    {
      "epoch": 0.4663454427985397,
      "grad_norm": 2.7201597690582275,
      "learning_rate": 5.617574277707674e-07,
      "loss": 0.7211,
      "step": 29955
    },
    {
      "epoch": 0.46642328380050907,
      "grad_norm": 3.1504993438720703,
      "learning_rate": 5.616754887661625e-07,
      "loss": 0.7811,
      "step": 29960
    },
    {
      "epoch": 0.46650112480247846,
      "grad_norm": 4.6782097816467285,
      "learning_rate": 5.615935497615575e-07,
      "loss": 0.7411,
      "step": 29965
    },
    {
      "epoch": 0.46657896580444785,
      "grad_norm": 3.7057347297668457,
      "learning_rate": 5.615116107569525e-07,
      "loss": 0.822,
      "step": 29970
    },
    {
      "epoch": 0.4666568068064172,
      "grad_norm": 2.954636573791504,
      "learning_rate": 5.614296717523476e-07,
      "loss": 0.7547,
      "step": 29975
    },
    {
      "epoch": 0.4667346478083866,
      "grad_norm": 3.5806500911712646,
      "learning_rate": 5.613477327477426e-07,
      "loss": 0.7754,
      "step": 29980
    },
    {
      "epoch": 0.466812488810356,
      "grad_norm": 4.228501319885254,
      "learning_rate": 5.612657937431375e-07,
      "loss": 0.7847,
      "step": 29985
    },
    {
      "epoch": 0.46689032981232537,
      "grad_norm": 3.845973014831543,
      "learning_rate": 5.611838547385326e-07,
      "loss": 0.7468,
      "step": 29990
    },
    {
      "epoch": 0.4669681708142947,
      "grad_norm": 4.924877166748047,
      "learning_rate": 5.611019157339277e-07,
      "loss": 0.8489,
      "step": 29995
    },
    {
      "epoch": 0.4670460118162641,
      "grad_norm": 2.9617180824279785,
      "learning_rate": 5.610199767293226e-07,
      "loss": 0.7707,
      "step": 30000
    },
    {
      "epoch": 0.4671238528182335,
      "grad_norm": 3.4927444458007812,
      "learning_rate": 5.609380377247177e-07,
      "loss": 0.6909,
      "step": 30005
    },
    {
      "epoch": 0.4672016938202029,
      "grad_norm": 4.621715068817139,
      "learning_rate": 5.608560987201128e-07,
      "loss": 0.76,
      "step": 30010
    },
    {
      "epoch": 0.4672795348221722,
      "grad_norm": 3.0478920936584473,
      "learning_rate": 5.607741597155077e-07,
      "loss": 0.6639,
      "step": 30015
    },
    {
      "epoch": 0.4673573758241416,
      "grad_norm": 4.733149528503418,
      "learning_rate": 5.606922207109028e-07,
      "loss": 0.6859,
      "step": 30020
    },
    {
      "epoch": 0.467435216826111,
      "grad_norm": 5.031196594238281,
      "learning_rate": 5.606102817062978e-07,
      "loss": 0.7383,
      "step": 30025
    },
    {
      "epoch": 0.46751305782808034,
      "grad_norm": 8.476208686828613,
      "learning_rate": 5.605283427016928e-07,
      "loss": 0.8552,
      "step": 30030
    },
    {
      "epoch": 0.46759089883004973,
      "grad_norm": 2.9002037048339844,
      "learning_rate": 5.604464036970878e-07,
      "loss": 0.7467,
      "step": 30035
    },
    {
      "epoch": 0.4676687398320191,
      "grad_norm": 2.8706724643707275,
      "learning_rate": 5.603644646924829e-07,
      "loss": 0.6344,
      "step": 30040
    },
    {
      "epoch": 0.4677465808339885,
      "grad_norm": 3.0804450511932373,
      "learning_rate": 5.602825256878779e-07,
      "loss": 0.7574,
      "step": 30045
    },
    {
      "epoch": 0.46782442183595785,
      "grad_norm": 3.774770498275757,
      "learning_rate": 5.60200586683273e-07,
      "loss": 0.6704,
      "step": 30050
    },
    {
      "epoch": 0.46790226283792724,
      "grad_norm": 3.883472204208374,
      "learning_rate": 5.60118647678668e-07,
      "loss": 0.7171,
      "step": 30055
    },
    {
      "epoch": 0.46798010383989663,
      "grad_norm": 4.072752952575684,
      "learning_rate": 5.600367086740631e-07,
      "loss": 0.8177,
      "step": 30060
    },
    {
      "epoch": 0.468057944841866,
      "grad_norm": 5.7671122550964355,
      "learning_rate": 5.59954769669458e-07,
      "loss": 0.6957,
      "step": 30065
    },
    {
      "epoch": 0.46813578584383536,
      "grad_norm": 3.559356927871704,
      "learning_rate": 5.59872830664853e-07,
      "loss": 0.6976,
      "step": 30070
    },
    {
      "epoch": 0.46821362684580475,
      "grad_norm": 5.881770133972168,
      "learning_rate": 5.597908916602481e-07,
      "loss": 0.7266,
      "step": 30075
    },
    {
      "epoch": 0.46829146784777415,
      "grad_norm": 4.246764183044434,
      "learning_rate": 5.597089526556431e-07,
      "loss": 0.8223,
      "step": 30080
    },
    {
      "epoch": 0.46836930884974354,
      "grad_norm": 5.646158218383789,
      "learning_rate": 5.596270136510382e-07,
      "loss": 0.6786,
      "step": 30085
    },
    {
      "epoch": 0.4684471498517129,
      "grad_norm": 10.228924751281738,
      "learning_rate": 5.595450746464332e-07,
      "loss": 0.6733,
      "step": 30090
    },
    {
      "epoch": 0.46852499085368227,
      "grad_norm": 3.891977071762085,
      "learning_rate": 5.594631356418282e-07,
      "loss": 0.7445,
      "step": 30095
    },
    {
      "epoch": 0.46860283185565166,
      "grad_norm": 4.490929126739502,
      "learning_rate": 5.593811966372233e-07,
      "loss": 0.7698,
      "step": 30100
    },
    {
      "epoch": 0.46868067285762105,
      "grad_norm": 3.159205436706543,
      "learning_rate": 5.592992576326182e-07,
      "loss": 0.7899,
      "step": 30105
    },
    {
      "epoch": 0.4687585138595904,
      "grad_norm": 3.0567328929901123,
      "learning_rate": 5.592173186280132e-07,
      "loss": 0.7076,
      "step": 30110
    },
    {
      "epoch": 0.4688363548615598,
      "grad_norm": 3.1571202278137207,
      "learning_rate": 5.591353796234083e-07,
      "loss": 0.7043,
      "step": 30115
    },
    {
      "epoch": 0.46891419586352917,
      "grad_norm": 3.532790422439575,
      "learning_rate": 5.590534406188034e-07,
      "loss": 0.6927,
      "step": 30120
    },
    {
      "epoch": 0.4689920368654985,
      "grad_norm": 6.347319602966309,
      "learning_rate": 5.589715016141983e-07,
      "loss": 0.762,
      "step": 30125
    },
    {
      "epoch": 0.4690698778674679,
      "grad_norm": 3.0561578273773193,
      "learning_rate": 5.588895626095934e-07,
      "loss": 0.8156,
      "step": 30130
    },
    {
      "epoch": 0.4691477188694373,
      "grad_norm": 10.545469284057617,
      "learning_rate": 5.588076236049885e-07,
      "loss": 0.8252,
      "step": 30135
    },
    {
      "epoch": 0.4692255598714067,
      "grad_norm": 8.552352905273438,
      "learning_rate": 5.587256846003835e-07,
      "loss": 0.7887,
      "step": 30140
    },
    {
      "epoch": 0.469303400873376,
      "grad_norm": 3.7188873291015625,
      "learning_rate": 5.586437455957784e-07,
      "loss": 0.7125,
      "step": 30145
    },
    {
      "epoch": 0.4693812418753454,
      "grad_norm": 3.9613687992095947,
      "learning_rate": 5.585618065911735e-07,
      "loss": 0.7146,
      "step": 30150
    },
    {
      "epoch": 0.4694590828773148,
      "grad_norm": 4.090216636657715,
      "learning_rate": 5.584798675865685e-07,
      "loss": 0.7674,
      "step": 30155
    },
    {
      "epoch": 0.4695369238792842,
      "grad_norm": 5.443205833435059,
      "learning_rate": 5.583979285819635e-07,
      "loss": 0.7739,
      "step": 30160
    },
    {
      "epoch": 0.46961476488125353,
      "grad_norm": 3.365614891052246,
      "learning_rate": 5.583159895773586e-07,
      "loss": 0.8041,
      "step": 30165
    },
    {
      "epoch": 0.4696926058832229,
      "grad_norm": 2.3033275604248047,
      "learning_rate": 5.582340505727536e-07,
      "loss": 0.6029,
      "step": 30170
    },
    {
      "epoch": 0.4697704468851923,
      "grad_norm": 2.6599738597869873,
      "learning_rate": 5.581521115681487e-07,
      "loss": 0.7552,
      "step": 30175
    },
    {
      "epoch": 0.4698482878871617,
      "grad_norm": 4.373157978057861,
      "learning_rate": 5.580701725635437e-07,
      "loss": 0.7585,
      "step": 30180
    },
    {
      "epoch": 0.46992612888913105,
      "grad_norm": 2.9015092849731445,
      "learning_rate": 5.579882335589386e-07,
      "loss": 0.8133,
      "step": 30185
    },
    {
      "epoch": 0.47000396989110044,
      "grad_norm": 3.2715601921081543,
      "learning_rate": 5.579062945543337e-07,
      "loss": 0.732,
      "step": 30190
    },
    {
      "epoch": 0.47008181089306983,
      "grad_norm": 3.6906094551086426,
      "learning_rate": 5.578243555497287e-07,
      "loss": 0.793,
      "step": 30195
    },
    {
      "epoch": 0.47015965189503917,
      "grad_norm": 5.613792419433594,
      "learning_rate": 5.577424165451238e-07,
      "loss": 0.8115,
      "step": 30200
    },
    {
      "epoch": 0.47023749289700856,
      "grad_norm": 3.553741455078125,
      "learning_rate": 5.576604775405188e-07,
      "loss": 0.8386,
      "step": 30205
    },
    {
      "epoch": 0.47031533389897795,
      "grad_norm": 3.9397878646850586,
      "learning_rate": 5.575785385359139e-07,
      "loss": 0.6509,
      "step": 30210
    },
    {
      "epoch": 0.47039317490094734,
      "grad_norm": 5.246465682983398,
      "learning_rate": 5.574965995313089e-07,
      "loss": 0.7575,
      "step": 30215
    },
    {
      "epoch": 0.4704710159029167,
      "grad_norm": 3.3713631629943848,
      "learning_rate": 5.574146605267039e-07,
      "loss": 0.696,
      "step": 30220
    },
    {
      "epoch": 0.47054885690488607,
      "grad_norm": 11.992700576782227,
      "learning_rate": 5.573327215220989e-07,
      "loss": 0.7919,
      "step": 30225
    },
    {
      "epoch": 0.47062669790685546,
      "grad_norm": 2.8566079139709473,
      "learning_rate": 5.57250782517494e-07,
      "loss": 0.7537,
      "step": 30230
    },
    {
      "epoch": 0.47070453890882485,
      "grad_norm": 2.114238977432251,
      "learning_rate": 5.571688435128889e-07,
      "loss": 0.7511,
      "step": 30235
    },
    {
      "epoch": 0.4707823799107942,
      "grad_norm": 3.9447951316833496,
      "learning_rate": 5.57086904508284e-07,
      "loss": 0.8093,
      "step": 30240
    },
    {
      "epoch": 0.4708602209127636,
      "grad_norm": 2.5769100189208984,
      "learning_rate": 5.570049655036791e-07,
      "loss": 0.8382,
      "step": 30245
    },
    {
      "epoch": 0.470938061914733,
      "grad_norm": 3.8322553634643555,
      "learning_rate": 5.56923026499074e-07,
      "loss": 0.8452,
      "step": 30250
    },
    {
      "epoch": 0.47101590291670237,
      "grad_norm": 4.5541276931762695,
      "learning_rate": 5.568410874944691e-07,
      "loss": 0.752,
      "step": 30255
    },
    {
      "epoch": 0.4710937439186717,
      "grad_norm": 4.292836666107178,
      "learning_rate": 5.567591484898642e-07,
      "loss": 0.7293,
      "step": 30260
    },
    {
      "epoch": 0.4711715849206411,
      "grad_norm": 4.974549293518066,
      "learning_rate": 5.566772094852592e-07,
      "loss": 0.6776,
      "step": 30265
    },
    {
      "epoch": 0.4712494259226105,
      "grad_norm": 3.6732537746429443,
      "learning_rate": 5.565952704806541e-07,
      "loss": 0.718,
      "step": 30270
    },
    {
      "epoch": 0.4713272669245799,
      "grad_norm": 3.264256000518799,
      "learning_rate": 5.565133314760492e-07,
      "loss": 0.7018,
      "step": 30275
    },
    {
      "epoch": 0.4714051079265492,
      "grad_norm": 3.919670581817627,
      "learning_rate": 5.564313924714442e-07,
      "loss": 0.8198,
      "step": 30280
    },
    {
      "epoch": 0.4714829489285186,
      "grad_norm": 5.429051876068115,
      "learning_rate": 5.563494534668392e-07,
      "loss": 0.7415,
      "step": 30285
    },
    {
      "epoch": 0.471560789930488,
      "grad_norm": 3.516733169555664,
      "learning_rate": 5.562675144622343e-07,
      "loss": 0.8319,
      "step": 30290
    },
    {
      "epoch": 0.47163863093245734,
      "grad_norm": 2.9769673347473145,
      "learning_rate": 5.561855754576293e-07,
      "loss": 0.8558,
      "step": 30295
    },
    {
      "epoch": 0.47171647193442673,
      "grad_norm": 3.807605504989624,
      "learning_rate": 5.561036364530244e-07,
      "loss": 0.7925,
      "step": 30300
    },
    {
      "epoch": 0.4717943129363961,
      "grad_norm": 5.5716729164123535,
      "learning_rate": 5.560216974484194e-07,
      "loss": 0.7343,
      "step": 30305
    },
    {
      "epoch": 0.4718721539383655,
      "grad_norm": 3.192689895629883,
      "learning_rate": 5.559397584438143e-07,
      "loss": 0.911,
      "step": 30310
    },
    {
      "epoch": 0.47194999494033485,
      "grad_norm": 5.447216510772705,
      "learning_rate": 5.558578194392094e-07,
      "loss": 0.7356,
      "step": 30315
    },
    {
      "epoch": 0.47202783594230424,
      "grad_norm": 3.8497579097747803,
      "learning_rate": 5.557758804346045e-07,
      "loss": 0.7136,
      "step": 30320
    },
    {
      "epoch": 0.47210567694427363,
      "grad_norm": 3.0262906551361084,
      "learning_rate": 5.556939414299995e-07,
      "loss": 0.611,
      "step": 30325
    },
    {
      "epoch": 0.472183517946243,
      "grad_norm": 3.951643705368042,
      "learning_rate": 5.556120024253945e-07,
      "loss": 0.7164,
      "step": 30330
    },
    {
      "epoch": 0.47226135894821236,
      "grad_norm": 2.9380064010620117,
      "learning_rate": 5.555300634207896e-07,
      "loss": 0.9327,
      "step": 30335
    },
    {
      "epoch": 0.47233919995018175,
      "grad_norm": 3.5171658992767334,
      "learning_rate": 5.554481244161846e-07,
      "loss": 0.8163,
      "step": 30340
    },
    {
      "epoch": 0.47241704095215115,
      "grad_norm": 3.827981472015381,
      "learning_rate": 5.553661854115796e-07,
      "loss": 0.7316,
      "step": 30345
    },
    {
      "epoch": 0.47249488195412054,
      "grad_norm": 6.381930828094482,
      "learning_rate": 5.552842464069746e-07,
      "loss": 0.8498,
      "step": 30350
    },
    {
      "epoch": 0.4725727229560899,
      "grad_norm": 5.095180988311768,
      "learning_rate": 5.552023074023697e-07,
      "loss": 0.8397,
      "step": 30355
    },
    {
      "epoch": 0.47265056395805927,
      "grad_norm": 3.2798590660095215,
      "learning_rate": 5.551203683977646e-07,
      "loss": 0.8183,
      "step": 30360
    },
    {
      "epoch": 0.47272840496002866,
      "grad_norm": 6.906369686126709,
      "learning_rate": 5.550384293931597e-07,
      "loss": 0.7851,
      "step": 30365
    },
    {
      "epoch": 0.472806245961998,
      "grad_norm": 5.3679070472717285,
      "learning_rate": 5.549564903885548e-07,
      "loss": 0.7946,
      "step": 30370
    },
    {
      "epoch": 0.4728840869639674,
      "grad_norm": 3.064115524291992,
      "learning_rate": 5.548745513839497e-07,
      "loss": 0.7399,
      "step": 30375
    },
    {
      "epoch": 0.4729619279659368,
      "grad_norm": 6.3130598068237305,
      "learning_rate": 5.547926123793448e-07,
      "loss": 0.838,
      "step": 30380
    },
    {
      "epoch": 0.47303976896790617,
      "grad_norm": 2.8328700065612793,
      "learning_rate": 5.547106733747399e-07,
      "loss": 0.8108,
      "step": 30385
    },
    {
      "epoch": 0.4731176099698755,
      "grad_norm": 5.87791109085083,
      "learning_rate": 5.546287343701348e-07,
      "loss": 0.7566,
      "step": 30390
    },
    {
      "epoch": 0.4731954509718449,
      "grad_norm": 3.817066192626953,
      "learning_rate": 5.545467953655298e-07,
      "loss": 0.8262,
      "step": 30395
    },
    {
      "epoch": 0.4732732919738143,
      "grad_norm": 3.3358867168426514,
      "learning_rate": 5.544648563609249e-07,
      "loss": 0.7697,
      "step": 30400
    },
    {
      "epoch": 0.4733511329757837,
      "grad_norm": 4.308083534240723,
      "learning_rate": 5.543829173563199e-07,
      "loss": 0.7029,
      "step": 30405
    },
    {
      "epoch": 0.473428973977753,
      "grad_norm": 2.3288745880126953,
      "learning_rate": 5.54300978351715e-07,
      "loss": 0.6473,
      "step": 30410
    },
    {
      "epoch": 0.4735068149797224,
      "grad_norm": 2.8548166751861572,
      "learning_rate": 5.5421903934711e-07,
      "loss": 0.7697,
      "step": 30415
    },
    {
      "epoch": 0.4735846559816918,
      "grad_norm": 7.335264682769775,
      "learning_rate": 5.54137100342505e-07,
      "loss": 0.7965,
      "step": 30420
    },
    {
      "epoch": 0.4736624969836612,
      "grad_norm": 5.75967264175415,
      "learning_rate": 5.540551613379001e-07,
      "loss": 0.7229,
      "step": 30425
    },
    {
      "epoch": 0.47374033798563053,
      "grad_norm": 3.3753843307495117,
      "learning_rate": 5.53973222333295e-07,
      "loss": 0.7804,
      "step": 30430
    },
    {
      "epoch": 0.4738181789875999,
      "grad_norm": 10.285598754882812,
      "learning_rate": 5.5389128332869e-07,
      "loss": 0.8052,
      "step": 30435
    },
    {
      "epoch": 0.4738960199895693,
      "grad_norm": 2.9903228282928467,
      "learning_rate": 5.538093443240851e-07,
      "loss": 0.634,
      "step": 30440
    },
    {
      "epoch": 0.4739738609915387,
      "grad_norm": 4.23531436920166,
      "learning_rate": 5.537274053194802e-07,
      "loss": 0.7934,
      "step": 30445
    },
    {
      "epoch": 0.47405170199350805,
      "grad_norm": 5.449930191040039,
      "learning_rate": 5.536454663148752e-07,
      "loss": 0.7772,
      "step": 30450
    },
    {
      "epoch": 0.47412954299547744,
      "grad_norm": 3.573413372039795,
      "learning_rate": 5.535635273102702e-07,
      "loss": 0.7654,
      "step": 30455
    },
    {
      "epoch": 0.47420738399744683,
      "grad_norm": 8.743206977844238,
      "learning_rate": 5.534815883056653e-07,
      "loss": 0.7891,
      "step": 30460
    },
    {
      "epoch": 0.47428522499941617,
      "grad_norm": 3.1340866088867188,
      "learning_rate": 5.533996493010604e-07,
      "loss": 0.7283,
      "step": 30465
    },
    {
      "epoch": 0.47436306600138556,
      "grad_norm": 14.64851188659668,
      "learning_rate": 5.533177102964552e-07,
      "loss": 0.8084,
      "step": 30470
    },
    {
      "epoch": 0.47444090700335495,
      "grad_norm": 4.152097225189209,
      "learning_rate": 5.532357712918503e-07,
      "loss": 0.955,
      "step": 30475
    },
    {
      "epoch": 0.47451874800532434,
      "grad_norm": 2.982809066772461,
      "learning_rate": 5.531538322872454e-07,
      "loss": 0.6385,
      "step": 30480
    },
    {
      "epoch": 0.4745965890072937,
      "grad_norm": 5.353337287902832,
      "learning_rate": 5.530718932826403e-07,
      "loss": 0.669,
      "step": 30485
    },
    {
      "epoch": 0.47467443000926307,
      "grad_norm": 4.827667713165283,
      "learning_rate": 5.529899542780354e-07,
      "loss": 0.6335,
      "step": 30490
    },
    {
      "epoch": 0.47475227101123246,
      "grad_norm": 4.295917510986328,
      "learning_rate": 5.529080152734305e-07,
      "loss": 0.8188,
      "step": 30495
    },
    {
      "epoch": 0.47483011201320185,
      "grad_norm": 2.7371466159820557,
      "learning_rate": 5.528260762688255e-07,
      "loss": 0.7194,
      "step": 30500
    },
    {
      "epoch": 0.4749079530151712,
      "grad_norm": 3.5925276279449463,
      "learning_rate": 5.527441372642205e-07,
      "loss": 0.7934,
      "step": 30505
    },
    {
      "epoch": 0.4749857940171406,
      "grad_norm": 3.6323416233062744,
      "learning_rate": 5.526621982596155e-07,
      "loss": 0.7258,
      "step": 30510
    },
    {
      "epoch": 0.47506363501911,
      "grad_norm": 3.0073904991149902,
      "learning_rate": 5.525802592550105e-07,
      "loss": 0.756,
      "step": 30515
    },
    {
      "epoch": 0.47514147602107937,
      "grad_norm": 2.839306116104126,
      "learning_rate": 5.524983202504055e-07,
      "loss": 0.7201,
      "step": 30520
    },
    {
      "epoch": 0.4752193170230487,
      "grad_norm": 4.088250160217285,
      "learning_rate": 5.524163812458006e-07,
      "loss": 0.8,
      "step": 30525
    },
    {
      "epoch": 0.4752971580250181,
      "grad_norm": 4.9360246658325195,
      "learning_rate": 5.523344422411956e-07,
      "loss": 0.7328,
      "step": 30530
    },
    {
      "epoch": 0.4753749990269875,
      "grad_norm": 2.4090027809143066,
      "learning_rate": 5.522525032365907e-07,
      "loss": 0.8216,
      "step": 30535
    },
    {
      "epoch": 0.4754528400289569,
      "grad_norm": 3.4636001586914062,
      "learning_rate": 5.521705642319857e-07,
      "loss": 0.8851,
      "step": 30540
    },
    {
      "epoch": 0.4755306810309262,
      "grad_norm": 3.5605151653289795,
      "learning_rate": 5.520886252273807e-07,
      "loss": 0.7622,
      "step": 30545
    },
    {
      "epoch": 0.4756085220328956,
      "grad_norm": 3.146246910095215,
      "learning_rate": 5.520066862227757e-07,
      "loss": 0.769,
      "step": 30550
    },
    {
      "epoch": 0.475686363034865,
      "grad_norm": 3.3484089374542236,
      "learning_rate": 5.519247472181707e-07,
      "loss": 0.7818,
      "step": 30555
    },
    {
      "epoch": 0.47576420403683434,
      "grad_norm": 6.414142608642578,
      "learning_rate": 5.518428082135657e-07,
      "loss": 0.8276,
      "step": 30560
    },
    {
      "epoch": 0.47584204503880373,
      "grad_norm": 3.6248252391815186,
      "learning_rate": 5.517608692089608e-07,
      "loss": 0.7878,
      "step": 30565
    },
    {
      "epoch": 0.4759198860407731,
      "grad_norm": 5.4479498863220215,
      "learning_rate": 5.516789302043559e-07,
      "loss": 0.6737,
      "step": 30570
    },
    {
      "epoch": 0.4759977270427425,
      "grad_norm": 2.6424448490142822,
      "learning_rate": 5.515969911997509e-07,
      "loss": 0.7526,
      "step": 30575
    },
    {
      "epoch": 0.47607556804471185,
      "grad_norm": 5.836863040924072,
      "learning_rate": 5.515150521951459e-07,
      "loss": 0.8099,
      "step": 30580
    },
    {
      "epoch": 0.47615340904668124,
      "grad_norm": 5.8411970138549805,
      "learning_rate": 5.51433113190541e-07,
      "loss": 0.7849,
      "step": 30585
    },
    {
      "epoch": 0.47623125004865063,
      "grad_norm": 5.983549118041992,
      "learning_rate": 5.513511741859361e-07,
      "loss": 0.7344,
      "step": 30590
    },
    {
      "epoch": 0.47630909105062,
      "grad_norm": 2.5778143405914307,
      "learning_rate": 5.512692351813309e-07,
      "loss": 0.627,
      "step": 30595
    },
    {
      "epoch": 0.47638693205258936,
      "grad_norm": 4.768832683563232,
      "learning_rate": 5.51187296176726e-07,
      "loss": 0.9032,
      "step": 30600
    },
    {
      "epoch": 0.47646477305455875,
      "grad_norm": 4.830402374267578,
      "learning_rate": 5.511053571721211e-07,
      "loss": 0.6845,
      "step": 30605
    },
    {
      "epoch": 0.47654261405652815,
      "grad_norm": 4.313559532165527,
      "learning_rate": 5.51023418167516e-07,
      "loss": 0.7451,
      "step": 30610
    },
    {
      "epoch": 0.47662045505849754,
      "grad_norm": 2.9489803314208984,
      "learning_rate": 5.509414791629111e-07,
      "loss": 0.6711,
      "step": 30615
    },
    {
      "epoch": 0.4766982960604669,
      "grad_norm": 6.420879364013672,
      "learning_rate": 5.508595401583062e-07,
      "loss": 0.8271,
      "step": 30620
    },
    {
      "epoch": 0.47677613706243627,
      "grad_norm": 5.491647720336914,
      "learning_rate": 5.507776011537012e-07,
      "loss": 0.7068,
      "step": 30625
    },
    {
      "epoch": 0.47685397806440566,
      "grad_norm": 3.9549171924591064,
      "learning_rate": 5.506956621490962e-07,
      "loss": 0.7564,
      "step": 30630
    },
    {
      "epoch": 0.476931819066375,
      "grad_norm": 3.1556475162506104,
      "learning_rate": 5.506137231444912e-07,
      "loss": 0.7131,
      "step": 30635
    },
    {
      "epoch": 0.4770096600683444,
      "grad_norm": 5.831211566925049,
      "learning_rate": 5.505317841398862e-07,
      "loss": 0.7348,
      "step": 30640
    },
    {
      "epoch": 0.4770875010703138,
      "grad_norm": 3.6472585201263428,
      "learning_rate": 5.504498451352812e-07,
      "loss": 0.7808,
      "step": 30645
    },
    {
      "epoch": 0.47716534207228317,
      "grad_norm": 4.638653755187988,
      "learning_rate": 5.503679061306763e-07,
      "loss": 0.7586,
      "step": 30650
    },
    {
      "epoch": 0.4772431830742525,
      "grad_norm": 4.502923965454102,
      "learning_rate": 5.502859671260713e-07,
      "loss": 0.759,
      "step": 30655
    },
    {
      "epoch": 0.4773210240762219,
      "grad_norm": 3.446194887161255,
      "learning_rate": 5.502040281214664e-07,
      "loss": 0.722,
      "step": 30660
    },
    {
      "epoch": 0.4773988650781913,
      "grad_norm": 4.650200366973877,
      "learning_rate": 5.501220891168614e-07,
      "loss": 0.7882,
      "step": 30665
    },
    {
      "epoch": 0.4774767060801607,
      "grad_norm": 4.709731578826904,
      "learning_rate": 5.500401501122564e-07,
      "loss": 0.7775,
      "step": 30670
    },
    {
      "epoch": 0.47755454708213,
      "grad_norm": 3.4137790203094482,
      "learning_rate": 5.499582111076514e-07,
      "loss": 0.792,
      "step": 30675
    },
    {
      "epoch": 0.4776323880840994,
      "grad_norm": 3.171194553375244,
      "learning_rate": 5.498762721030465e-07,
      "loss": 0.6756,
      "step": 30680
    },
    {
      "epoch": 0.4777102290860688,
      "grad_norm": 3.039259433746338,
      "learning_rate": 5.497943330984414e-07,
      "loss": 0.7454,
      "step": 30685
    },
    {
      "epoch": 0.4777880700880382,
      "grad_norm": 6.83022403717041,
      "learning_rate": 5.497123940938365e-07,
      "loss": 0.6923,
      "step": 30690
    },
    {
      "epoch": 0.47786591109000753,
      "grad_norm": 10.1714506149292,
      "learning_rate": 5.496304550892316e-07,
      "loss": 0.6514,
      "step": 30695
    },
    {
      "epoch": 0.4779437520919769,
      "grad_norm": 3.7122445106506348,
      "learning_rate": 5.495485160846266e-07,
      "loss": 0.7135,
      "step": 30700
    },
    {
      "epoch": 0.4780215930939463,
      "grad_norm": 7.862659931182861,
      "learning_rate": 5.494665770800216e-07,
      "loss": 0.8054,
      "step": 30705
    },
    {
      "epoch": 0.4780994340959157,
      "grad_norm": 2.792076587677002,
      "learning_rate": 5.493846380754167e-07,
      "loss": 0.7684,
      "step": 30710
    },
    {
      "epoch": 0.47817727509788505,
      "grad_norm": 3.739151954650879,
      "learning_rate": 5.493026990708117e-07,
      "loss": 0.7436,
      "step": 30715
    },
    {
      "epoch": 0.47825511609985444,
      "grad_norm": 3.2284679412841797,
      "learning_rate": 5.492207600662066e-07,
      "loss": 0.7335,
      "step": 30720
    },
    {
      "epoch": 0.47833295710182383,
      "grad_norm": 2.885590076446533,
      "learning_rate": 5.491388210616017e-07,
      "loss": 0.7096,
      "step": 30725
    },
    {
      "epoch": 0.47841079810379317,
      "grad_norm": 3.765489101409912,
      "learning_rate": 5.490568820569968e-07,
      "loss": 0.7399,
      "step": 30730
    },
    {
      "epoch": 0.47848863910576256,
      "grad_norm": 4.197171688079834,
      "learning_rate": 5.489749430523917e-07,
      "loss": 0.7573,
      "step": 30735
    },
    {
      "epoch": 0.47856648010773195,
      "grad_norm": 7.4482340812683105,
      "learning_rate": 5.488930040477868e-07,
      "loss": 0.6916,
      "step": 30740
    },
    {
      "epoch": 0.47864432110970134,
      "grad_norm": 3.448819398880005,
      "learning_rate": 5.488110650431819e-07,
      "loss": 0.8392,
      "step": 30745
    },
    {
      "epoch": 0.4787221621116707,
      "grad_norm": 3.0407896041870117,
      "learning_rate": 5.487291260385769e-07,
      "loss": 0.7925,
      "step": 30750
    },
    {
      "epoch": 0.47880000311364007,
      "grad_norm": 2.9351015090942383,
      "learning_rate": 5.486471870339718e-07,
      "loss": 0.6561,
      "step": 30755
    },
    {
      "epoch": 0.47887784411560946,
      "grad_norm": 2.733675241470337,
      "learning_rate": 5.485652480293669e-07,
      "loss": 0.7624,
      "step": 30760
    },
    {
      "epoch": 0.47895568511757886,
      "grad_norm": 3.5775351524353027,
      "learning_rate": 5.484833090247619e-07,
      "loss": 0.7758,
      "step": 30765
    },
    {
      "epoch": 0.4790335261195482,
      "grad_norm": 3.9379889965057373,
      "learning_rate": 5.48401370020157e-07,
      "loss": 0.6032,
      "step": 30770
    },
    {
      "epoch": 0.4791113671215176,
      "grad_norm": 3.853231191635132,
      "learning_rate": 5.48319431015552e-07,
      "loss": 0.8215,
      "step": 30775
    },
    {
      "epoch": 0.479189208123487,
      "grad_norm": 4.0149054527282715,
      "learning_rate": 5.48237492010947e-07,
      "loss": 0.8128,
      "step": 30780
    },
    {
      "epoch": 0.47926704912545637,
      "grad_norm": 3.5337696075439453,
      "learning_rate": 5.481555530063421e-07,
      "loss": 0.7253,
      "step": 30785
    },
    {
      "epoch": 0.4793448901274257,
      "grad_norm": 3.3882758617401123,
      "learning_rate": 5.480736140017371e-07,
      "loss": 0.7262,
      "step": 30790
    },
    {
      "epoch": 0.4794227311293951,
      "grad_norm": 9.372142791748047,
      "learning_rate": 5.47991674997132e-07,
      "loss": 0.7098,
      "step": 30795
    },
    {
      "epoch": 0.4795005721313645,
      "grad_norm": 2.7920899391174316,
      "learning_rate": 5.479097359925271e-07,
      "loss": 0.7135,
      "step": 30800
    },
    {
      "epoch": 0.4795784131333338,
      "grad_norm": 2.811682939529419,
      "learning_rate": 5.478277969879222e-07,
      "loss": 0.7409,
      "step": 30805
    },
    {
      "epoch": 0.4796562541353032,
      "grad_norm": 5.5391058921813965,
      "learning_rate": 5.477458579833171e-07,
      "loss": 0.7571,
      "step": 30810
    },
    {
      "epoch": 0.4797340951372726,
      "grad_norm": 3.4567179679870605,
      "learning_rate": 5.476639189787122e-07,
      "loss": 0.8027,
      "step": 30815
    },
    {
      "epoch": 0.479811936139242,
      "grad_norm": 3.0708186626434326,
      "learning_rate": 5.475819799741073e-07,
      "loss": 0.6636,
      "step": 30820
    },
    {
      "epoch": 0.47988977714121134,
      "grad_norm": 5.144400119781494,
      "learning_rate": 5.475000409695024e-07,
      "loss": 0.7276,
      "step": 30825
    },
    {
      "epoch": 0.47996761814318073,
      "grad_norm": 4.826434135437012,
      "learning_rate": 5.474181019648973e-07,
      "loss": 0.8631,
      "step": 30830
    },
    {
      "epoch": 0.4800454591451501,
      "grad_norm": 7.047203540802002,
      "learning_rate": 5.473361629602923e-07,
      "loss": 0.7419,
      "step": 30835
    },
    {
      "epoch": 0.4801233001471195,
      "grad_norm": 3.959611415863037,
      "learning_rate": 5.472542239556874e-07,
      "loss": 0.6898,
      "step": 30840
    },
    {
      "epoch": 0.48020114114908885,
      "grad_norm": 2.473358631134033,
      "learning_rate": 5.471722849510823e-07,
      "loss": 0.7219,
      "step": 30845
    },
    {
      "epoch": 0.48027898215105824,
      "grad_norm": 3.1444664001464844,
      "learning_rate": 5.470903459464774e-07,
      "loss": 0.7472,
      "step": 30850
    },
    {
      "epoch": 0.48035682315302763,
      "grad_norm": 7.360410690307617,
      "learning_rate": 5.470084069418725e-07,
      "loss": 0.7435,
      "step": 30855
    },
    {
      "epoch": 0.480434664154997,
      "grad_norm": 2.833648681640625,
      "learning_rate": 5.469264679372675e-07,
      "loss": 0.6979,
      "step": 30860
    },
    {
      "epoch": 0.48051250515696636,
      "grad_norm": 5.742299556732178,
      "learning_rate": 5.468445289326625e-07,
      "loss": 0.8322,
      "step": 30865
    },
    {
      "epoch": 0.48059034615893576,
      "grad_norm": 4.930783748626709,
      "learning_rate": 5.467625899280576e-07,
      "loss": 0.7067,
      "step": 30870
    },
    {
      "epoch": 0.48066818716090515,
      "grad_norm": 4.286282062530518,
      "learning_rate": 5.466806509234525e-07,
      "loss": 0.8284,
      "step": 30875
    },
    {
      "epoch": 0.48074602816287454,
      "grad_norm": 3.72047758102417,
      "learning_rate": 5.465987119188475e-07,
      "loss": 0.7451,
      "step": 30880
    },
    {
      "epoch": 0.4808238691648439,
      "grad_norm": 4.538388729095459,
      "learning_rate": 5.465167729142426e-07,
      "loss": 0.7357,
      "step": 30885
    },
    {
      "epoch": 0.48090171016681327,
      "grad_norm": 7.441522598266602,
      "learning_rate": 5.464348339096376e-07,
      "loss": 0.7003,
      "step": 30890
    },
    {
      "epoch": 0.48097955116878266,
      "grad_norm": 4.495595932006836,
      "learning_rate": 5.463528949050327e-07,
      "loss": 0.7833,
      "step": 30895
    },
    {
      "epoch": 0.481057392170752,
      "grad_norm": 3.37333345413208,
      "learning_rate": 5.462709559004277e-07,
      "loss": 0.8412,
      "step": 30900
    },
    {
      "epoch": 0.4811352331727214,
      "grad_norm": 5.76720666885376,
      "learning_rate": 5.461890168958227e-07,
      "loss": 0.7545,
      "step": 30905
    },
    {
      "epoch": 0.4812130741746908,
      "grad_norm": 4.569162368774414,
      "learning_rate": 5.461070778912178e-07,
      "loss": 0.6286,
      "step": 30910
    },
    {
      "epoch": 0.4812909151766602,
      "grad_norm": 8.438356399536133,
      "learning_rate": 5.460251388866129e-07,
      "loss": 0.7995,
      "step": 30915
    },
    {
      "epoch": 0.4813687561786295,
      "grad_norm": 4.326777935028076,
      "learning_rate": 5.459431998820077e-07,
      "loss": 0.846,
      "step": 30920
    },
    {
      "epoch": 0.4814465971805989,
      "grad_norm": 4.1380934715271,
      "learning_rate": 5.458612608774028e-07,
      "loss": 0.7604,
      "step": 30925
    },
    {
      "epoch": 0.4815244381825683,
      "grad_norm": 4.7531657218933105,
      "learning_rate": 5.457793218727979e-07,
      "loss": 0.6723,
      "step": 30930
    },
    {
      "epoch": 0.4816022791845377,
      "grad_norm": 3.197828531265259,
      "learning_rate": 5.456973828681928e-07,
      "loss": 0.7805,
      "step": 30935
    },
    {
      "epoch": 0.481680120186507,
      "grad_norm": 4.3347625732421875,
      "learning_rate": 5.456154438635879e-07,
      "loss": 0.6928,
      "step": 30940
    },
    {
      "epoch": 0.4817579611884764,
      "grad_norm": 10.630887985229492,
      "learning_rate": 5.45533504858983e-07,
      "loss": 0.8336,
      "step": 30945
    },
    {
      "epoch": 0.4818358021904458,
      "grad_norm": 3.9803647994995117,
      "learning_rate": 5.454515658543781e-07,
      "loss": 0.7157,
      "step": 30950
    },
    {
      "epoch": 0.4819136431924152,
      "grad_norm": 5.097218990325928,
      "learning_rate": 5.45369626849773e-07,
      "loss": 0.6674,
      "step": 30955
    },
    {
      "epoch": 0.48199148419438453,
      "grad_norm": 2.698425531387329,
      "learning_rate": 5.45287687845168e-07,
      "loss": 0.6791,
      "step": 30960
    },
    {
      "epoch": 0.4820693251963539,
      "grad_norm": 2.6709539890289307,
      "learning_rate": 5.452057488405631e-07,
      "loss": 0.7338,
      "step": 30965
    },
    {
      "epoch": 0.4821471661983233,
      "grad_norm": 3.578798770904541,
      "learning_rate": 5.45123809835958e-07,
      "loss": 0.7476,
      "step": 30970
    },
    {
      "epoch": 0.48222500720029265,
      "grad_norm": 5.683715343475342,
      "learning_rate": 5.450418708313531e-07,
      "loss": 0.8027,
      "step": 30975
    },
    {
      "epoch": 0.48230284820226205,
      "grad_norm": 3.637927293777466,
      "learning_rate": 5.449599318267482e-07,
      "loss": 0.6785,
      "step": 30980
    },
    {
      "epoch": 0.48238068920423144,
      "grad_norm": 4.570502758026123,
      "learning_rate": 5.448779928221432e-07,
      "loss": 0.6699,
      "step": 30985
    },
    {
      "epoch": 0.48245853020620083,
      "grad_norm": 3.4548423290252686,
      "learning_rate": 5.447960538175382e-07,
      "loss": 0.7416,
      "step": 30990
    },
    {
      "epoch": 0.48253637120817017,
      "grad_norm": 6.130021095275879,
      "learning_rate": 5.447141148129333e-07,
      "loss": 0.8533,
      "step": 30995
    },
    {
      "epoch": 0.48261421221013956,
      "grad_norm": 5.062192440032959,
      "learning_rate": 5.446321758083282e-07,
      "loss": 0.7889,
      "step": 31000
    },
    {
      "epoch": 0.48269205321210895,
      "grad_norm": 3.6712381839752197,
      "learning_rate": 5.445502368037233e-07,
      "loss": 0.7601,
      "step": 31005
    },
    {
      "epoch": 0.48276989421407834,
      "grad_norm": 12.077085494995117,
      "learning_rate": 5.444682977991183e-07,
      "loss": 0.805,
      "step": 31010
    },
    {
      "epoch": 0.4828477352160477,
      "grad_norm": 10.729565620422363,
      "learning_rate": 5.443863587945133e-07,
      "loss": 0.7932,
      "step": 31015
    },
    {
      "epoch": 0.48292557621801707,
      "grad_norm": 3.317769765853882,
      "learning_rate": 5.443044197899084e-07,
      "loss": 0.708,
      "step": 31020
    },
    {
      "epoch": 0.48300341721998646,
      "grad_norm": 4.708921432495117,
      "learning_rate": 5.442224807853034e-07,
      "loss": 0.7724,
      "step": 31025
    },
    {
      "epoch": 0.48308125822195586,
      "grad_norm": 3.357910394668579,
      "learning_rate": 5.441405417806984e-07,
      "loss": 0.805,
      "step": 31030
    },
    {
      "epoch": 0.4831590992239252,
      "grad_norm": 3.439021348953247,
      "learning_rate": 5.440586027760935e-07,
      "loss": 0.7144,
      "step": 31035
    },
    {
      "epoch": 0.4832369402258946,
      "grad_norm": 2.7874014377593994,
      "learning_rate": 5.439766637714885e-07,
      "loss": 0.7368,
      "step": 31040
    },
    {
      "epoch": 0.483314781227864,
      "grad_norm": 3.7799317836761475,
      "learning_rate": 5.438947247668834e-07,
      "loss": 0.7697,
      "step": 31045
    },
    {
      "epoch": 0.48339262222983337,
      "grad_norm": 6.024733066558838,
      "learning_rate": 5.438127857622785e-07,
      "loss": 0.8239,
      "step": 31050
    },
    {
      "epoch": 0.4834704632318027,
      "grad_norm": 4.746781349182129,
      "learning_rate": 5.437308467576736e-07,
      "loss": 0.7274,
      "step": 31055
    },
    {
      "epoch": 0.4835483042337721,
      "grad_norm": 5.800485610961914,
      "learning_rate": 5.436489077530685e-07,
      "loss": 0.7112,
      "step": 31060
    },
    {
      "epoch": 0.4836261452357415,
      "grad_norm": 6.829309463500977,
      "learning_rate": 5.435669687484636e-07,
      "loss": 0.7123,
      "step": 31065
    },
    {
      "epoch": 0.4837039862377108,
      "grad_norm": 16.465612411499023,
      "learning_rate": 5.434850297438587e-07,
      "loss": 0.708,
      "step": 31070
    },
    {
      "epoch": 0.4837818272396802,
      "grad_norm": 2.9724323749542236,
      "learning_rate": 5.434030907392538e-07,
      "loss": 0.7997,
      "step": 31075
    },
    {
      "epoch": 0.4838596682416496,
      "grad_norm": 8.170181274414062,
      "learning_rate": 5.433211517346486e-07,
      "loss": 0.918,
      "step": 31080
    },
    {
      "epoch": 0.483937509243619,
      "grad_norm": 3.666452407836914,
      "learning_rate": 5.432392127300437e-07,
      "loss": 0.7276,
      "step": 31085
    },
    {
      "epoch": 0.48401535024558834,
      "grad_norm": 4.632342338562012,
      "learning_rate": 5.431572737254388e-07,
      "loss": 0.7994,
      "step": 31090
    },
    {
      "epoch": 0.48409319124755773,
      "grad_norm": 3.7869715690612793,
      "learning_rate": 5.430753347208338e-07,
      "loss": 0.7466,
      "step": 31095
    },
    {
      "epoch": 0.4841710322495271,
      "grad_norm": 9.879354476928711,
      "learning_rate": 5.429933957162288e-07,
      "loss": 0.6935,
      "step": 31100
    },
    {
      "epoch": 0.4842488732514965,
      "grad_norm": 4.113779067993164,
      "learning_rate": 5.429114567116239e-07,
      "loss": 0.7551,
      "step": 31105
    },
    {
      "epoch": 0.48432671425346585,
      "grad_norm": 4.171604156494141,
      "learning_rate": 5.428295177070189e-07,
      "loss": 0.7731,
      "step": 31110
    },
    {
      "epoch": 0.48440455525543524,
      "grad_norm": 2.5860118865966797,
      "learning_rate": 5.427475787024139e-07,
      "loss": 0.7611,
      "step": 31115
    },
    {
      "epoch": 0.48448239625740463,
      "grad_norm": 3.8882453441619873,
      "learning_rate": 5.426656396978089e-07,
      "loss": 0.7583,
      "step": 31120
    },
    {
      "epoch": 0.484560237259374,
      "grad_norm": 9.416706085205078,
      "learning_rate": 5.425837006932039e-07,
      "loss": 0.7568,
      "step": 31125
    },
    {
      "epoch": 0.48463807826134336,
      "grad_norm": 4.6626787185668945,
      "learning_rate": 5.42501761688599e-07,
      "loss": 0.6208,
      "step": 31130
    },
    {
      "epoch": 0.48471591926331276,
      "grad_norm": 4.40387487411499,
      "learning_rate": 5.42419822683994e-07,
      "loss": 0.7322,
      "step": 31135
    },
    {
      "epoch": 0.48479376026528215,
      "grad_norm": 3.26765775680542,
      "learning_rate": 5.42337883679389e-07,
      "loss": 0.8519,
      "step": 31140
    },
    {
      "epoch": 0.48487160126725154,
      "grad_norm": 4.323541641235352,
      "learning_rate": 5.422559446747841e-07,
      "loss": 0.8036,
      "step": 31145
    },
    {
      "epoch": 0.4849494422692209,
      "grad_norm": 7.4432573318481445,
      "learning_rate": 5.421740056701792e-07,
      "loss": 0.6929,
      "step": 31150
    },
    {
      "epoch": 0.48502728327119027,
      "grad_norm": 4.1317596435546875,
      "learning_rate": 5.420920666655741e-07,
      "loss": 0.7799,
      "step": 31155
    },
    {
      "epoch": 0.48510512427315966,
      "grad_norm": 6.6668195724487305,
      "learning_rate": 5.420101276609691e-07,
      "loss": 0.6856,
      "step": 31160
    },
    {
      "epoch": 0.485182965275129,
      "grad_norm": 2.612093687057495,
      "learning_rate": 5.419281886563642e-07,
      "loss": 0.7167,
      "step": 31165
    },
    {
      "epoch": 0.4852608062770984,
      "grad_norm": 2.933692455291748,
      "learning_rate": 5.418462496517591e-07,
      "loss": 0.7604,
      "step": 31170
    },
    {
      "epoch": 0.4853386472790678,
      "grad_norm": 4.096435546875,
      "learning_rate": 5.417643106471542e-07,
      "loss": 0.7378,
      "step": 31175
    },
    {
      "epoch": 0.4854164882810372,
      "grad_norm": 3.6913399696350098,
      "learning_rate": 5.416823716425493e-07,
      "loss": 0.7262,
      "step": 31180
    },
    {
      "epoch": 0.4854943292830065,
      "grad_norm": 3.1496493816375732,
      "learning_rate": 5.416004326379443e-07,
      "loss": 0.7041,
      "step": 31185
    },
    {
      "epoch": 0.4855721702849759,
      "grad_norm": 7.3737945556640625,
      "learning_rate": 5.415184936333393e-07,
      "loss": 0.793,
      "step": 31190
    },
    {
      "epoch": 0.4856500112869453,
      "grad_norm": 7.973015785217285,
      "learning_rate": 5.414365546287344e-07,
      "loss": 0.6608,
      "step": 31195
    },
    {
      "epoch": 0.4857278522889147,
      "grad_norm": 3.815272331237793,
      "learning_rate": 5.413546156241294e-07,
      "loss": 0.7244,
      "step": 31200
    },
    {
      "epoch": 0.485805693290884,
      "grad_norm": 4.9943108558654785,
      "learning_rate": 5.412726766195243e-07,
      "loss": 0.7415,
      "step": 31205
    },
    {
      "epoch": 0.4858835342928534,
      "grad_norm": 4.6898884773254395,
      "learning_rate": 5.411907376149194e-07,
      "loss": 0.8064,
      "step": 31210
    },
    {
      "epoch": 0.4859613752948228,
      "grad_norm": 3.4858808517456055,
      "learning_rate": 5.411087986103145e-07,
      "loss": 0.8227,
      "step": 31215
    },
    {
      "epoch": 0.4860392162967922,
      "grad_norm": 8.153918266296387,
      "learning_rate": 5.410268596057095e-07,
      "loss": 0.7075,
      "step": 31220
    },
    {
      "epoch": 0.48611705729876153,
      "grad_norm": 2.3834216594696045,
      "learning_rate": 5.409449206011045e-07,
      "loss": 0.7394,
      "step": 31225
    },
    {
      "epoch": 0.4861948983007309,
      "grad_norm": 4.020205497741699,
      "learning_rate": 5.408629815964996e-07,
      "loss": 0.6235,
      "step": 31230
    },
    {
      "epoch": 0.4862727393027003,
      "grad_norm": 3.7188174724578857,
      "learning_rate": 5.407810425918946e-07,
      "loss": 0.8095,
      "step": 31235
    },
    {
      "epoch": 0.48635058030466966,
      "grad_norm": 3.206958770751953,
      "learning_rate": 5.406991035872897e-07,
      "loss": 0.6759,
      "step": 31240
    },
    {
      "epoch": 0.48642842130663905,
      "grad_norm": 5.961366653442383,
      "learning_rate": 5.406171645826846e-07,
      "loss": 0.8132,
      "step": 31245
    },
    {
      "epoch": 0.48650626230860844,
      "grad_norm": 3.109642505645752,
      "learning_rate": 5.405352255780796e-07,
      "loss": 0.666,
      "step": 31250
    },
    {
      "epoch": 0.48658410331057783,
      "grad_norm": 4.960744380950928,
      "learning_rate": 5.404532865734747e-07,
      "loss": 0.9611,
      "step": 31255
    },
    {
      "epoch": 0.48666194431254717,
      "grad_norm": 2.7765390872955322,
      "learning_rate": 5.403713475688697e-07,
      "loss": 0.7282,
      "step": 31260
    },
    {
      "epoch": 0.48673978531451656,
      "grad_norm": 3.7915613651275635,
      "learning_rate": 5.402894085642647e-07,
      "loss": 0.7299,
      "step": 31265
    },
    {
      "epoch": 0.48681762631648595,
      "grad_norm": 6.055993556976318,
      "learning_rate": 5.402074695596598e-07,
      "loss": 0.902,
      "step": 31270
    },
    {
      "epoch": 0.48689546731845534,
      "grad_norm": 7.130350589752197,
      "learning_rate": 5.401255305550549e-07,
      "loss": 0.8009,
      "step": 31275
    },
    {
      "epoch": 0.4869733083204247,
      "grad_norm": 2.9298675060272217,
      "learning_rate": 5.400435915504498e-07,
      "loss": 0.7191,
      "step": 31280
    },
    {
      "epoch": 0.4870511493223941,
      "grad_norm": 3.6360220909118652,
      "learning_rate": 5.399616525458448e-07,
      "loss": 0.7822,
      "step": 31285
    },
    {
      "epoch": 0.48712899032436346,
      "grad_norm": 6.178891181945801,
      "learning_rate": 5.398797135412399e-07,
      "loss": 0.7254,
      "step": 31290
    },
    {
      "epoch": 0.48720683132633286,
      "grad_norm": 4.0464653968811035,
      "learning_rate": 5.397977745366348e-07,
      "loss": 0.8377,
      "step": 31295
    },
    {
      "epoch": 0.4872846723283022,
      "grad_norm": 3.031558036804199,
      "learning_rate": 5.397158355320299e-07,
      "loss": 0.7598,
      "step": 31300
    },
    {
      "epoch": 0.4873625133302716,
      "grad_norm": 2.3123228549957275,
      "learning_rate": 5.39633896527425e-07,
      "loss": 0.8469,
      "step": 31305
    },
    {
      "epoch": 0.487440354332241,
      "grad_norm": 4.890013217926025,
      "learning_rate": 5.395519575228201e-07,
      "loss": 0.725,
      "step": 31310
    },
    {
      "epoch": 0.48751819533421037,
      "grad_norm": 5.293360233306885,
      "learning_rate": 5.39470018518215e-07,
      "loss": 0.7072,
      "step": 31315
    },
    {
      "epoch": 0.4875960363361797,
      "grad_norm": 3.1986546516418457,
      "learning_rate": 5.393880795136101e-07,
      "loss": 0.691,
      "step": 31320
    },
    {
      "epoch": 0.4876738773381491,
      "grad_norm": 7.940555095672607,
      "learning_rate": 5.393061405090051e-07,
      "loss": 0.8704,
      "step": 31325
    },
    {
      "epoch": 0.4877517183401185,
      "grad_norm": 3.1320574283599854,
      "learning_rate": 5.392242015044e-07,
      "loss": 0.6929,
      "step": 31330
    },
    {
      "epoch": 0.4878295593420878,
      "grad_norm": 3.882742404937744,
      "learning_rate": 5.391422624997951e-07,
      "loss": 0.6421,
      "step": 31335
    },
    {
      "epoch": 0.4879074003440572,
      "grad_norm": 3.4556562900543213,
      "learning_rate": 5.390603234951902e-07,
      "loss": 0.8479,
      "step": 31340
    },
    {
      "epoch": 0.4879852413460266,
      "grad_norm": 3.0240955352783203,
      "learning_rate": 5.389783844905852e-07,
      "loss": 0.7236,
      "step": 31345
    },
    {
      "epoch": 0.488063082347996,
      "grad_norm": 4.038585662841797,
      "learning_rate": 5.388964454859802e-07,
      "loss": 0.7957,
      "step": 31350
    },
    {
      "epoch": 0.48814092334996534,
      "grad_norm": 4.807283878326416,
      "learning_rate": 5.388145064813753e-07,
      "loss": 0.6972,
      "step": 31355
    },
    {
      "epoch": 0.48821876435193473,
      "grad_norm": 4.245325088500977,
      "learning_rate": 5.387325674767703e-07,
      "loss": 0.7279,
      "step": 31360
    },
    {
      "epoch": 0.4882966053539041,
      "grad_norm": 3.8696720600128174,
      "learning_rate": 5.386506284721653e-07,
      "loss": 0.7803,
      "step": 31365
    },
    {
      "epoch": 0.4883744463558735,
      "grad_norm": 4.037196159362793,
      "learning_rate": 5.385686894675603e-07,
      "loss": 0.7562,
      "step": 31370
    },
    {
      "epoch": 0.48845228735784285,
      "grad_norm": 4.754850387573242,
      "learning_rate": 5.384867504629553e-07,
      "loss": 0.6916,
      "step": 31375
    },
    {
      "epoch": 0.48853012835981224,
      "grad_norm": 4.887664318084717,
      "learning_rate": 5.384048114583504e-07,
      "loss": 0.6616,
      "step": 31380
    },
    {
      "epoch": 0.48860796936178164,
      "grad_norm": 4.8289794921875,
      "learning_rate": 5.383228724537454e-07,
      "loss": 0.818,
      "step": 31385
    },
    {
      "epoch": 0.488685810363751,
      "grad_norm": 4.632926940917969,
      "learning_rate": 5.382409334491404e-07,
      "loss": 0.7232,
      "step": 31390
    },
    {
      "epoch": 0.48876365136572036,
      "grad_norm": 6.670058727264404,
      "learning_rate": 5.381589944445355e-07,
      "loss": 0.7375,
      "step": 31395
    },
    {
      "epoch": 0.48884149236768976,
      "grad_norm": 4.1670050621032715,
      "learning_rate": 5.380770554399306e-07,
      "loss": 0.7674,
      "step": 31400
    },
    {
      "epoch": 0.48891933336965915,
      "grad_norm": 4.5522894859313965,
      "learning_rate": 5.379951164353254e-07,
      "loss": 0.7606,
      "step": 31405
    },
    {
      "epoch": 0.4889971743716285,
      "grad_norm": 7.621889114379883,
      "learning_rate": 5.379131774307205e-07,
      "loss": 0.7636,
      "step": 31410
    },
    {
      "epoch": 0.4890750153735979,
      "grad_norm": 2.912235736846924,
      "learning_rate": 5.378312384261156e-07,
      "loss": 0.7287,
      "step": 31415
    },
    {
      "epoch": 0.48915285637556727,
      "grad_norm": 2.1460154056549072,
      "learning_rate": 5.377492994215105e-07,
      "loss": 0.7201,
      "step": 31420
    },
    {
      "epoch": 0.48923069737753666,
      "grad_norm": 6.309132099151611,
      "learning_rate": 5.376673604169056e-07,
      "loss": 0.7114,
      "step": 31425
    },
    {
      "epoch": 0.489308538379506,
      "grad_norm": 8.84521770477295,
      "learning_rate": 5.375854214123007e-07,
      "loss": 0.8207,
      "step": 31430
    },
    {
      "epoch": 0.4893863793814754,
      "grad_norm": 4.781198024749756,
      "learning_rate": 5.375034824076958e-07,
      "loss": 0.7833,
      "step": 31435
    },
    {
      "epoch": 0.4894642203834448,
      "grad_norm": 3.32561993598938,
      "learning_rate": 5.374215434030907e-07,
      "loss": 0.7075,
      "step": 31440
    },
    {
      "epoch": 0.4895420613854142,
      "grad_norm": 7.8017964363098145,
      "learning_rate": 5.373396043984857e-07,
      "loss": 0.7289,
      "step": 31445
    },
    {
      "epoch": 0.4896199023873835,
      "grad_norm": 3.5266835689544678,
      "learning_rate": 5.372576653938808e-07,
      "loss": 0.7816,
      "step": 31450
    },
    {
      "epoch": 0.4896977433893529,
      "grad_norm": 4.4425530433654785,
      "learning_rate": 5.371757263892758e-07,
      "loss": 0.852,
      "step": 31455
    },
    {
      "epoch": 0.4897755843913223,
      "grad_norm": 5.1467084884643555,
      "learning_rate": 5.370937873846708e-07,
      "loss": 0.8646,
      "step": 31460
    },
    {
      "epoch": 0.4898534253932917,
      "grad_norm": 2.8108134269714355,
      "learning_rate": 5.370118483800659e-07,
      "loss": 0.689,
      "step": 31465
    },
    {
      "epoch": 0.489931266395261,
      "grad_norm": 3.2352046966552734,
      "learning_rate": 5.369299093754609e-07,
      "loss": 0.74,
      "step": 31470
    },
    {
      "epoch": 0.4900091073972304,
      "grad_norm": 3.8721742630004883,
      "learning_rate": 5.368479703708559e-07,
      "loss": 0.8161,
      "step": 31475
    },
    {
      "epoch": 0.4900869483991998,
      "grad_norm": 4.493784427642822,
      "learning_rate": 5.36766031366251e-07,
      "loss": 0.6705,
      "step": 31480
    },
    {
      "epoch": 0.4901647894011692,
      "grad_norm": 2.470388412475586,
      "learning_rate": 5.366840923616459e-07,
      "loss": 0.7436,
      "step": 31485
    },
    {
      "epoch": 0.49024263040313854,
      "grad_norm": 3.03464674949646,
      "learning_rate": 5.36602153357041e-07,
      "loss": 0.7599,
      "step": 31490
    },
    {
      "epoch": 0.4903204714051079,
      "grad_norm": 3.56787371635437,
      "learning_rate": 5.36520214352436e-07,
      "loss": 0.689,
      "step": 31495
    },
    {
      "epoch": 0.4903983124070773,
      "grad_norm": 4.144897937774658,
      "learning_rate": 5.36438275347831e-07,
      "loss": 0.7708,
      "step": 31500
    },
    {
      "epoch": 0.49047615340904666,
      "grad_norm": 2.819772481918335,
      "learning_rate": 5.363563363432261e-07,
      "loss": 0.7889,
      "step": 31505
    },
    {
      "epoch": 0.49055399441101605,
      "grad_norm": 4.154952526092529,
      "learning_rate": 5.362743973386212e-07,
      "loss": 0.7629,
      "step": 31510
    },
    {
      "epoch": 0.49063183541298544,
      "grad_norm": 9.503395080566406,
      "learning_rate": 5.361924583340161e-07,
      "loss": 0.6775,
      "step": 31515
    },
    {
      "epoch": 0.49070967641495483,
      "grad_norm": 5.26347541809082,
      "learning_rate": 5.361105193294112e-07,
      "loss": 0.6906,
      "step": 31520
    },
    {
      "epoch": 0.49078751741692417,
      "grad_norm": 6.194119930267334,
      "learning_rate": 5.360285803248063e-07,
      "loss": 0.8668,
      "step": 31525
    },
    {
      "epoch": 0.49086535841889356,
      "grad_norm": 3.416261911392212,
      "learning_rate": 5.359466413202011e-07,
      "loss": 0.7743,
      "step": 31530
    },
    {
      "epoch": 0.49094319942086295,
      "grad_norm": 3.323477029800415,
      "learning_rate": 5.358647023155962e-07,
      "loss": 0.8348,
      "step": 31535
    },
    {
      "epoch": 0.49102104042283234,
      "grad_norm": 3.4117913246154785,
      "learning_rate": 5.357827633109913e-07,
      "loss": 0.6623,
      "step": 31540
    },
    {
      "epoch": 0.4910988814248017,
      "grad_norm": 5.043321132659912,
      "learning_rate": 5.357008243063863e-07,
      "loss": 0.7419,
      "step": 31545
    },
    {
      "epoch": 0.4911767224267711,
      "grad_norm": 5.590872764587402,
      "learning_rate": 5.356188853017813e-07,
      "loss": 0.7832,
      "step": 31550
    },
    {
      "epoch": 0.49125456342874047,
      "grad_norm": 2.906338691711426,
      "learning_rate": 5.355369462971764e-07,
      "loss": 0.7256,
      "step": 31555
    },
    {
      "epoch": 0.49133240443070986,
      "grad_norm": 3.8234970569610596,
      "learning_rate": 5.354550072925715e-07,
      "loss": 0.7802,
      "step": 31560
    },
    {
      "epoch": 0.4914102454326792,
      "grad_norm": 3.9781415462493896,
      "learning_rate": 5.353730682879664e-07,
      "loss": 0.6958,
      "step": 31565
    },
    {
      "epoch": 0.4914880864346486,
      "grad_norm": 7.56910514831543,
      "learning_rate": 5.352911292833614e-07,
      "loss": 0.7566,
      "step": 31570
    },
    {
      "epoch": 0.491565927436618,
      "grad_norm": 2.6581532955169678,
      "learning_rate": 5.352091902787565e-07,
      "loss": 0.7905,
      "step": 31575
    },
    {
      "epoch": 0.4916437684385873,
      "grad_norm": 4.3859758377075195,
      "learning_rate": 5.351272512741515e-07,
      "loss": 0.6935,
      "step": 31580
    },
    {
      "epoch": 0.4917216094405567,
      "grad_norm": 4.9828033447265625,
      "learning_rate": 5.350453122695465e-07,
      "loss": 0.9096,
      "step": 31585
    },
    {
      "epoch": 0.4917994504425261,
      "grad_norm": 4.203456401824951,
      "learning_rate": 5.349633732649416e-07,
      "loss": 0.7372,
      "step": 31590
    },
    {
      "epoch": 0.4918772914444955,
      "grad_norm": 3.4125876426696777,
      "learning_rate": 5.348814342603366e-07,
      "loss": 0.8609,
      "step": 31595
    },
    {
      "epoch": 0.4919551324464648,
      "grad_norm": 3.541236639022827,
      "learning_rate": 5.347994952557317e-07,
      "loss": 0.8178,
      "step": 31600
    },
    {
      "epoch": 0.4920329734484342,
      "grad_norm": 3.6993322372436523,
      "learning_rate": 5.347175562511267e-07,
      "loss": 0.7717,
      "step": 31605
    },
    {
      "epoch": 0.4921108144504036,
      "grad_norm": 3.4722630977630615,
      "learning_rate": 5.346356172465216e-07,
      "loss": 0.7236,
      "step": 31610
    },
    {
      "epoch": 0.492188655452373,
      "grad_norm": 7.224667072296143,
      "learning_rate": 5.345536782419167e-07,
      "loss": 0.7623,
      "step": 31615
    },
    {
      "epoch": 0.49226649645434234,
      "grad_norm": 5.8936076164245605,
      "learning_rate": 5.344717392373117e-07,
      "loss": 0.6921,
      "step": 31620
    },
    {
      "epoch": 0.49234433745631173,
      "grad_norm": 10.682324409484863,
      "learning_rate": 5.343898002327067e-07,
      "loss": 0.7935,
      "step": 31625
    },
    {
      "epoch": 0.4924221784582811,
      "grad_norm": 4.73100471496582,
      "learning_rate": 5.343078612281018e-07,
      "loss": 0.7838,
      "step": 31630
    },
    {
      "epoch": 0.4925000194602505,
      "grad_norm": 8.651354789733887,
      "learning_rate": 5.342259222234969e-07,
      "loss": 0.7807,
      "step": 31635
    },
    {
      "epoch": 0.49257786046221985,
      "grad_norm": 3.073493480682373,
      "learning_rate": 5.341439832188918e-07,
      "loss": 0.7828,
      "step": 31640
    },
    {
      "epoch": 0.49265570146418924,
      "grad_norm": 3.2530338764190674,
      "learning_rate": 5.340620442142869e-07,
      "loss": 0.6507,
      "step": 31645
    },
    {
      "epoch": 0.49273354246615864,
      "grad_norm": 3.5065674781799316,
      "learning_rate": 5.339801052096819e-07,
      "loss": 0.7421,
      "step": 31650
    },
    {
      "epoch": 0.49281138346812803,
      "grad_norm": 9.087255477905273,
      "learning_rate": 5.338981662050768e-07,
      "loss": 0.6605,
      "step": 31655
    },
    {
      "epoch": 0.49288922447009736,
      "grad_norm": 5.483119010925293,
      "learning_rate": 5.338162272004719e-07,
      "loss": 0.8756,
      "step": 31660
    },
    {
      "epoch": 0.49296706547206676,
      "grad_norm": 4.297025203704834,
      "learning_rate": 5.33734288195867e-07,
      "loss": 0.6759,
      "step": 31665
    },
    {
      "epoch": 0.49304490647403615,
      "grad_norm": 3.946556329727173,
      "learning_rate": 5.33652349191262e-07,
      "loss": 0.7081,
      "step": 31670
    },
    {
      "epoch": 0.4931227474760055,
      "grad_norm": 3.3605268001556396,
      "learning_rate": 5.33570410186657e-07,
      "loss": 0.8256,
      "step": 31675
    },
    {
      "epoch": 0.4932005884779749,
      "grad_norm": 3.9473726749420166,
      "learning_rate": 5.334884711820521e-07,
      "loss": 0.7739,
      "step": 31680
    },
    {
      "epoch": 0.49327842947994427,
      "grad_norm": 19.540639877319336,
      "learning_rate": 5.334065321774472e-07,
      "loss": 0.774,
      "step": 31685
    },
    {
      "epoch": 0.49335627048191366,
      "grad_norm": 3.1018152236938477,
      "learning_rate": 5.33324593172842e-07,
      "loss": 0.7293,
      "step": 31690
    },
    {
      "epoch": 0.493434111483883,
      "grad_norm": 9.049464225769043,
      "learning_rate": 5.332426541682371e-07,
      "loss": 0.7499,
      "step": 31695
    },
    {
      "epoch": 0.4935119524858524,
      "grad_norm": 2.831634044647217,
      "learning_rate": 5.331607151636322e-07,
      "loss": 0.8896,
      "step": 31700
    },
    {
      "epoch": 0.4935897934878218,
      "grad_norm": 7.008086681365967,
      "learning_rate": 5.330787761590272e-07,
      "loss": 0.7746,
      "step": 31705
    },
    {
      "epoch": 0.4936676344897912,
      "grad_norm": 5.138648986816406,
      "learning_rate": 5.329968371544222e-07,
      "loss": 0.8179,
      "step": 31710
    },
    {
      "epoch": 0.4937454754917605,
      "grad_norm": 5.441119194030762,
      "learning_rate": 5.329148981498173e-07,
      "loss": 0.8448,
      "step": 31715
    },
    {
      "epoch": 0.4938233164937299,
      "grad_norm": 3.119594097137451,
      "learning_rate": 5.328329591452123e-07,
      "loss": 0.7159,
      "step": 31720
    },
    {
      "epoch": 0.4939011574956993,
      "grad_norm": 3.755413055419922,
      "learning_rate": 5.327510201406074e-07,
      "loss": 0.7007,
      "step": 31725
    },
    {
      "epoch": 0.4939789984976687,
      "grad_norm": 7.097749710083008,
      "learning_rate": 5.326690811360023e-07,
      "loss": 0.7485,
      "step": 31730
    },
    {
      "epoch": 0.494056839499638,
      "grad_norm": 4.017034530639648,
      "learning_rate": 5.325871421313973e-07,
      "loss": 0.8294,
      "step": 31735
    },
    {
      "epoch": 0.4941346805016074,
      "grad_norm": 4.874077320098877,
      "learning_rate": 5.325052031267924e-07,
      "loss": 0.8477,
      "step": 31740
    },
    {
      "epoch": 0.4942125215035768,
      "grad_norm": 4.150436878204346,
      "learning_rate": 5.324232641221874e-07,
      "loss": 0.8444,
      "step": 31745
    },
    {
      "epoch": 0.4942903625055462,
      "grad_norm": 4.164991855621338,
      "learning_rate": 5.323413251175824e-07,
      "loss": 0.7614,
      "step": 31750
    },
    {
      "epoch": 0.49436820350751554,
      "grad_norm": 3.9992542266845703,
      "learning_rate": 5.322593861129775e-07,
      "loss": 0.7788,
      "step": 31755
    },
    {
      "epoch": 0.4944460445094849,
      "grad_norm": 3.340470790863037,
      "learning_rate": 5.321774471083726e-07,
      "loss": 0.6389,
      "step": 31760
    },
    {
      "epoch": 0.4945238855114543,
      "grad_norm": 3.817882537841797,
      "learning_rate": 5.320955081037675e-07,
      "loss": 0.7768,
      "step": 31765
    },
    {
      "epoch": 0.49460172651342366,
      "grad_norm": 3.8433425426483154,
      "learning_rate": 5.320135690991625e-07,
      "loss": 0.8804,
      "step": 31770
    },
    {
      "epoch": 0.49467956751539305,
      "grad_norm": 2.9987850189208984,
      "learning_rate": 5.319316300945576e-07,
      "loss": 0.6936,
      "step": 31775
    },
    {
      "epoch": 0.49475740851736244,
      "grad_norm": 4.682468891143799,
      "learning_rate": 5.318496910899525e-07,
      "loss": 0.7695,
      "step": 31780
    },
    {
      "epoch": 0.49483524951933183,
      "grad_norm": 3.5517895221710205,
      "learning_rate": 5.317677520853476e-07,
      "loss": 0.8234,
      "step": 31785
    },
    {
      "epoch": 0.49491309052130117,
      "grad_norm": 4.163179874420166,
      "learning_rate": 5.316858130807427e-07,
      "loss": 0.8427,
      "step": 31790
    },
    {
      "epoch": 0.49499093152327056,
      "grad_norm": 2.8317389488220215,
      "learning_rate": 5.316038740761377e-07,
      "loss": 0.7408,
      "step": 31795
    },
    {
      "epoch": 0.49506877252523995,
      "grad_norm": 8.891326904296875,
      "learning_rate": 5.315219350715327e-07,
      "loss": 0.7974,
      "step": 31800
    },
    {
      "epoch": 0.49514661352720934,
      "grad_norm": 7.0925798416137695,
      "learning_rate": 5.314399960669278e-07,
      "loss": 0.8076,
      "step": 31805
    },
    {
      "epoch": 0.4952244545291787,
      "grad_norm": 9.016785621643066,
      "learning_rate": 5.313580570623227e-07,
      "loss": 0.7216,
      "step": 31810
    },
    {
      "epoch": 0.4953022955311481,
      "grad_norm": 4.506376266479492,
      "learning_rate": 5.312761180577178e-07,
      "loss": 0.8062,
      "step": 31815
    },
    {
      "epoch": 0.49538013653311747,
      "grad_norm": 3.6910665035247803,
      "learning_rate": 5.311941790531128e-07,
      "loss": 0.6585,
      "step": 31820
    },
    {
      "epoch": 0.49545797753508686,
      "grad_norm": 3.537003993988037,
      "learning_rate": 5.311122400485079e-07,
      "loss": 0.797,
      "step": 31825
    },
    {
      "epoch": 0.4955358185370562,
      "grad_norm": 7.610426425933838,
      "learning_rate": 5.310303010439029e-07,
      "loss": 0.7488,
      "step": 31830
    },
    {
      "epoch": 0.4956136595390256,
      "grad_norm": 3.17529559135437,
      "learning_rate": 5.30948362039298e-07,
      "loss": 0.751,
      "step": 31835
    },
    {
      "epoch": 0.495691500540995,
      "grad_norm": 6.6837053298950195,
      "learning_rate": 5.30866423034693e-07,
      "loss": 0.6983,
      "step": 31840
    },
    {
      "epoch": 0.4957693415429643,
      "grad_norm": 9.221789360046387,
      "learning_rate": 5.30784484030088e-07,
      "loss": 0.7075,
      "step": 31845
    },
    {
      "epoch": 0.4958471825449337,
      "grad_norm": 5.343038082122803,
      "learning_rate": 5.307025450254831e-07,
      "loss": 0.6657,
      "step": 31850
    },
    {
      "epoch": 0.4959250235469031,
      "grad_norm": 4.970161437988281,
      "learning_rate": 5.30620606020878e-07,
      "loss": 0.8162,
      "step": 31855
    },
    {
      "epoch": 0.4960028645488725,
      "grad_norm": 3.1920602321624756,
      "learning_rate": 5.30538667016273e-07,
      "loss": 0.7677,
      "step": 31860
    },
    {
      "epoch": 0.4960807055508418,
      "grad_norm": 3.2420198917388916,
      "learning_rate": 5.304567280116681e-07,
      "loss": 0.6231,
      "step": 31865
    },
    {
      "epoch": 0.4961585465528112,
      "grad_norm": 4.624316692352295,
      "learning_rate": 5.303747890070632e-07,
      "loss": 0.7881,
      "step": 31870
    },
    {
      "epoch": 0.4962363875547806,
      "grad_norm": 2.762160062789917,
      "learning_rate": 5.302928500024581e-07,
      "loss": 0.7534,
      "step": 31875
    },
    {
      "epoch": 0.49631422855675,
      "grad_norm": 4.296076774597168,
      "learning_rate": 5.302109109978532e-07,
      "loss": 0.7561,
      "step": 31880
    },
    {
      "epoch": 0.49639206955871934,
      "grad_norm": 8.022163391113281,
      "learning_rate": 5.301289719932483e-07,
      "loss": 0.8145,
      "step": 31885
    },
    {
      "epoch": 0.49646991056068873,
      "grad_norm": 4.3699798583984375,
      "learning_rate": 5.300470329886432e-07,
      "loss": 0.7022,
      "step": 31890
    },
    {
      "epoch": 0.4965477515626581,
      "grad_norm": 2.9266278743743896,
      "learning_rate": 5.299650939840382e-07,
      "loss": 0.7051,
      "step": 31895
    },
    {
      "epoch": 0.4966255925646275,
      "grad_norm": 3.2048120498657227,
      "learning_rate": 5.298831549794333e-07,
      "loss": 0.8202,
      "step": 31900
    },
    {
      "epoch": 0.49670343356659685,
      "grad_norm": 3.6829893589019775,
      "learning_rate": 5.298012159748283e-07,
      "loss": 0.688,
      "step": 31905
    },
    {
      "epoch": 0.49678127456856624,
      "grad_norm": 5.51873779296875,
      "learning_rate": 5.297192769702233e-07,
      "loss": 0.7433,
      "step": 31910
    },
    {
      "epoch": 0.49685911557053564,
      "grad_norm": 4.74458646774292,
      "learning_rate": 5.296373379656184e-07,
      "loss": 0.7908,
      "step": 31915
    },
    {
      "epoch": 0.49693695657250503,
      "grad_norm": 4.112690448760986,
      "learning_rate": 5.295553989610134e-07,
      "loss": 0.775,
      "step": 31920
    },
    {
      "epoch": 0.49701479757447437,
      "grad_norm": 3.469588279724121,
      "learning_rate": 5.294734599564084e-07,
      "loss": 0.7587,
      "step": 31925
    },
    {
      "epoch": 0.49709263857644376,
      "grad_norm": 7.575070381164551,
      "learning_rate": 5.293915209518035e-07,
      "loss": 0.7629,
      "step": 31930
    },
    {
      "epoch": 0.49717047957841315,
      "grad_norm": 3.291045665740967,
      "learning_rate": 5.293095819471984e-07,
      "loss": 0.7285,
      "step": 31935
    },
    {
      "epoch": 0.4972483205803825,
      "grad_norm": 2.586054563522339,
      "learning_rate": 5.292276429425935e-07,
      "loss": 0.6347,
      "step": 31940
    },
    {
      "epoch": 0.4973261615823519,
      "grad_norm": 3.9647815227508545,
      "learning_rate": 5.291457039379885e-07,
      "loss": 0.7317,
      "step": 31945
    },
    {
      "epoch": 0.49740400258432127,
      "grad_norm": 2.4557106494903564,
      "learning_rate": 5.290637649333836e-07,
      "loss": 0.6102,
      "step": 31950
    },
    {
      "epoch": 0.49748184358629066,
      "grad_norm": 3.1855356693267822,
      "learning_rate": 5.289818259287786e-07,
      "loss": 0.7472,
      "step": 31955
    },
    {
      "epoch": 0.49755968458826,
      "grad_norm": 3.886975049972534,
      "learning_rate": 5.288998869241737e-07,
      "loss": 0.675,
      "step": 31960
    },
    {
      "epoch": 0.4976375255902294,
      "grad_norm": 3.883635997772217,
      "learning_rate": 5.288179479195687e-07,
      "loss": 0.7279,
      "step": 31965
    },
    {
      "epoch": 0.4977153665921988,
      "grad_norm": 6.775463104248047,
      "learning_rate": 5.287360089149637e-07,
      "loss": 0.6957,
      "step": 31970
    },
    {
      "epoch": 0.4977932075941682,
      "grad_norm": 4.4077229499816895,
      "learning_rate": 5.286540699103587e-07,
      "loss": 0.7071,
      "step": 31975
    },
    {
      "epoch": 0.4978710485961375,
      "grad_norm": 3.4646551609039307,
      "learning_rate": 5.285721309057537e-07,
      "loss": 0.8185,
      "step": 31980
    },
    {
      "epoch": 0.4979488895981069,
      "grad_norm": 4.915809631347656,
      "learning_rate": 5.284901919011487e-07,
      "loss": 0.8274,
      "step": 31985
    },
    {
      "epoch": 0.4980267306000763,
      "grad_norm": 4.068741798400879,
      "learning_rate": 5.284082528965438e-07,
      "loss": 0.6597,
      "step": 31990
    },
    {
      "epoch": 0.4981045716020457,
      "grad_norm": 2.2890095710754395,
      "learning_rate": 5.283263138919389e-07,
      "loss": 0.7578,
      "step": 31995
    },
    {
      "epoch": 0.498182412604015,
      "grad_norm": 4.117259502410889,
      "learning_rate": 5.282443748873338e-07,
      "loss": 0.6874,
      "step": 32000
    },
    {
      "epoch": 0.4982602536059844,
      "grad_norm": 4.0782694816589355,
      "learning_rate": 5.281624358827289e-07,
      "loss": 0.7493,
      "step": 32005
    },
    {
      "epoch": 0.4983380946079538,
      "grad_norm": 3.2333414554595947,
      "learning_rate": 5.28080496878124e-07,
      "loss": 0.8256,
      "step": 32010
    },
    {
      "epoch": 0.49841593560992314,
      "grad_norm": 3.106598138809204,
      "learning_rate": 5.279985578735188e-07,
      "loss": 0.6584,
      "step": 32015
    },
    {
      "epoch": 0.49849377661189254,
      "grad_norm": 5.38564395904541,
      "learning_rate": 5.279166188689139e-07,
      "loss": 0.8011,
      "step": 32020
    },
    {
      "epoch": 0.49857161761386193,
      "grad_norm": 3.85125470161438,
      "learning_rate": 5.27834679864309e-07,
      "loss": 0.7249,
      "step": 32025
    },
    {
      "epoch": 0.4986494586158313,
      "grad_norm": 6.149100303649902,
      "learning_rate": 5.27752740859704e-07,
      "loss": 0.8704,
      "step": 32030
    },
    {
      "epoch": 0.49872729961780066,
      "grad_norm": 9.052156448364258,
      "learning_rate": 5.27670801855099e-07,
      "loss": 0.7138,
      "step": 32035
    },
    {
      "epoch": 0.49880514061977005,
      "grad_norm": 4.9207963943481445,
      "learning_rate": 5.275888628504941e-07,
      "loss": 0.7768,
      "step": 32040
    },
    {
      "epoch": 0.49888298162173944,
      "grad_norm": 3.4195189476013184,
      "learning_rate": 5.275069238458891e-07,
      "loss": 0.8106,
      "step": 32045
    },
    {
      "epoch": 0.49896082262370883,
      "grad_norm": 3.9791674613952637,
      "learning_rate": 5.274249848412842e-07,
      "loss": 0.822,
      "step": 32050
    },
    {
      "epoch": 0.49903866362567817,
      "grad_norm": 4.949367046356201,
      "learning_rate": 5.273430458366791e-07,
      "loss": 0.7004,
      "step": 32055
    },
    {
      "epoch": 0.49911650462764756,
      "grad_norm": 5.259047508239746,
      "learning_rate": 5.272611068320741e-07,
      "loss": 0.8256,
      "step": 32060
    },
    {
      "epoch": 0.49919434562961695,
      "grad_norm": 5.216323375701904,
      "learning_rate": 5.271791678274692e-07,
      "loss": 0.7564,
      "step": 32065
    },
    {
      "epoch": 0.49927218663158635,
      "grad_norm": 3.0554418563842773,
      "learning_rate": 5.270972288228642e-07,
      "loss": 0.7153,
      "step": 32070
    },
    {
      "epoch": 0.4993500276335557,
      "grad_norm": 3.82717227935791,
      "learning_rate": 5.270152898182593e-07,
      "loss": 0.7442,
      "step": 32075
    },
    {
      "epoch": 0.4994278686355251,
      "grad_norm": 2.3360207080841064,
      "learning_rate": 5.269333508136543e-07,
      "loss": 0.8531,
      "step": 32080
    },
    {
      "epoch": 0.49950570963749447,
      "grad_norm": 2.934239149093628,
      "learning_rate": 5.268514118090494e-07,
      "loss": 0.7365,
      "step": 32085
    },
    {
      "epoch": 0.49958355063946386,
      "grad_norm": 11.820845603942871,
      "learning_rate": 5.267694728044444e-07,
      "loss": 0.7855,
      "step": 32090
    },
    {
      "epoch": 0.4996613916414332,
      "grad_norm": 3.8256962299346924,
      "learning_rate": 5.266875337998393e-07,
      "loss": 0.5972,
      "step": 32095
    },
    {
      "epoch": 0.4997392326434026,
      "grad_norm": 3.4147393703460693,
      "learning_rate": 5.266055947952344e-07,
      "loss": 0.7143,
      "step": 32100
    },
    {
      "epoch": 0.499817073645372,
      "grad_norm": 3.64489483833313,
      "learning_rate": 5.265236557906294e-07,
      "loss": 0.7499,
      "step": 32105
    },
    {
      "epoch": 0.4998949146473413,
      "grad_norm": 3.4387834072113037,
      "learning_rate": 5.264417167860244e-07,
      "loss": 0.7335,
      "step": 32110
    },
    {
      "epoch": 0.4999727556493107,
      "grad_norm": 3.9227426052093506,
      "learning_rate": 5.263597777814195e-07,
      "loss": 0.7288,
      "step": 32115
    },
    {
      "epoch": 0.50005059665128,
      "grad_norm": 3.8069722652435303,
      "learning_rate": 5.262778387768146e-07,
      "loss": 0.8151,
      "step": 32120
    },
    {
      "epoch": 0.5001284376532494,
      "grad_norm": 2.4354546070098877,
      "learning_rate": 5.261958997722095e-07,
      "loss": 0.719,
      "step": 32125
    },
    {
      "epoch": 0.5002062786552188,
      "grad_norm": 7.43080472946167,
      "learning_rate": 5.261139607676046e-07,
      "loss": 0.8039,
      "step": 32130
    },
    {
      "epoch": 0.5002841196571882,
      "grad_norm": 4.2918925285339355,
      "learning_rate": 5.260320217629996e-07,
      "loss": 0.6927,
      "step": 32135
    },
    {
      "epoch": 0.5003619606591576,
      "grad_norm": 7.021772384643555,
      "learning_rate": 5.259500827583946e-07,
      "loss": 0.9068,
      "step": 32140
    },
    {
      "epoch": 0.500439801661127,
      "grad_norm": 3.5029406547546387,
      "learning_rate": 5.258681437537896e-07,
      "loss": 0.7214,
      "step": 32145
    },
    {
      "epoch": 0.5005176426630964,
      "grad_norm": 2.6251397132873535,
      "learning_rate": 5.257862047491847e-07,
      "loss": 0.7474,
      "step": 32150
    },
    {
      "epoch": 0.5005954836650658,
      "grad_norm": 6.177379131317139,
      "learning_rate": 5.257042657445797e-07,
      "loss": 0.8461,
      "step": 32155
    },
    {
      "epoch": 0.5006733246670351,
      "grad_norm": 6.51218843460083,
      "learning_rate": 5.256223267399747e-07,
      "loss": 0.7309,
      "step": 32160
    },
    {
      "epoch": 0.5007511656690045,
      "grad_norm": 4.144134044647217,
      "learning_rate": 5.255403877353698e-07,
      "loss": 0.702,
      "step": 32165
    },
    {
      "epoch": 0.5008290066709739,
      "grad_norm": 4.2770466804504395,
      "learning_rate": 5.254584487307648e-07,
      "loss": 0.761,
      "step": 32170
    },
    {
      "epoch": 0.5009068476729432,
      "grad_norm": 4.245838642120361,
      "learning_rate": 5.253765097261599e-07,
      "loss": 0.7223,
      "step": 32175
    },
    {
      "epoch": 0.5009846886749126,
      "grad_norm": 2.4867842197418213,
      "learning_rate": 5.252945707215548e-07,
      "loss": 0.7384,
      "step": 32180
    },
    {
      "epoch": 0.501062529676882,
      "grad_norm": 3.8057751655578613,
      "learning_rate": 5.252126317169498e-07,
      "loss": 0.8152,
      "step": 32185
    },
    {
      "epoch": 0.5011403706788514,
      "grad_norm": 3.783435583114624,
      "learning_rate": 5.251306927123449e-07,
      "loss": 0.7791,
      "step": 32190
    },
    {
      "epoch": 0.5012182116808207,
      "grad_norm": 2.6584465503692627,
      "learning_rate": 5.2504875370774e-07,
      "loss": 0.7681,
      "step": 32195
    },
    {
      "epoch": 0.5012960526827901,
      "grad_norm": 7.950997829437256,
      "learning_rate": 5.24966814703135e-07,
      "loss": 0.6929,
      "step": 32200
    },
    {
      "epoch": 0.5013738936847595,
      "grad_norm": 7.176991939544678,
      "learning_rate": 5.2488487569853e-07,
      "loss": 0.925,
      "step": 32205
    },
    {
      "epoch": 0.5014517346867289,
      "grad_norm": 4.086365699768066,
      "learning_rate": 5.248029366939251e-07,
      "loss": 0.8065,
      "step": 32210
    },
    {
      "epoch": 0.5015295756886983,
      "grad_norm": 3.4049980640411377,
      "learning_rate": 5.247209976893201e-07,
      "loss": 0.7936,
      "step": 32215
    },
    {
      "epoch": 0.5016074166906677,
      "grad_norm": 5.651351451873779,
      "learning_rate": 5.24639058684715e-07,
      "loss": 0.7281,
      "step": 32220
    },
    {
      "epoch": 0.501685257692637,
      "grad_norm": 2.9402480125427246,
      "learning_rate": 5.245571196801101e-07,
      "loss": 0.7272,
      "step": 32225
    },
    {
      "epoch": 0.5017630986946064,
      "grad_norm": 5.111139297485352,
      "learning_rate": 5.244751806755052e-07,
      "loss": 0.7083,
      "step": 32230
    },
    {
      "epoch": 0.5018409396965757,
      "grad_norm": 3.0065505504608154,
      "learning_rate": 5.243932416709001e-07,
      "loss": 0.6849,
      "step": 32235
    },
    {
      "epoch": 0.5019187806985451,
      "grad_norm": 4.171654224395752,
      "learning_rate": 5.243113026662952e-07,
      "loss": 0.8409,
      "step": 32240
    },
    {
      "epoch": 0.5019966217005145,
      "grad_norm": 2.5341849327087402,
      "learning_rate": 5.242293636616903e-07,
      "loss": 0.6005,
      "step": 32245
    },
    {
      "epoch": 0.5020744627024839,
      "grad_norm": 2.825509548187256,
      "learning_rate": 5.241474246570852e-07,
      "loss": 0.6274,
      "step": 32250
    },
    {
      "epoch": 0.5021523037044533,
      "grad_norm": 2.3395230770111084,
      "learning_rate": 5.240654856524803e-07,
      "loss": 0.7539,
      "step": 32255
    },
    {
      "epoch": 0.5022301447064227,
      "grad_norm": 3.047313928604126,
      "learning_rate": 5.239835466478753e-07,
      "loss": 0.6721,
      "step": 32260
    },
    {
      "epoch": 0.5023079857083921,
      "grad_norm": 3.283978223800659,
      "learning_rate": 5.239016076432703e-07,
      "loss": 0.8065,
      "step": 32265
    },
    {
      "epoch": 0.5023858267103614,
      "grad_norm": 5.6952714920043945,
      "learning_rate": 5.238196686386653e-07,
      "loss": 0.6988,
      "step": 32270
    },
    {
      "epoch": 0.5024636677123308,
      "grad_norm": 3.755058526992798,
      "learning_rate": 5.237377296340604e-07,
      "loss": 0.8128,
      "step": 32275
    },
    {
      "epoch": 0.5025415087143001,
      "grad_norm": 4.371938228607178,
      "learning_rate": 5.236557906294554e-07,
      "loss": 0.8891,
      "step": 32280
    },
    {
      "epoch": 0.5026193497162695,
      "grad_norm": 4.422750949859619,
      "learning_rate": 5.235738516248505e-07,
      "loss": 0.8391,
      "step": 32285
    },
    {
      "epoch": 0.5026971907182389,
      "grad_norm": 5.548360824584961,
      "learning_rate": 5.234919126202455e-07,
      "loss": 0.5715,
      "step": 32290
    },
    {
      "epoch": 0.5027750317202083,
      "grad_norm": 3.721450090408325,
      "learning_rate": 5.234099736156405e-07,
      "loss": 0.85,
      "step": 32295
    },
    {
      "epoch": 0.5028528727221777,
      "grad_norm": 3.385890007019043,
      "learning_rate": 5.233280346110355e-07,
      "loss": 0.7142,
      "step": 32300
    },
    {
      "epoch": 0.5029307137241471,
      "grad_norm": 4.370761871337891,
      "learning_rate": 5.232460956064305e-07,
      "loss": 0.7114,
      "step": 32305
    },
    {
      "epoch": 0.5030085547261164,
      "grad_norm": 2.715280532836914,
      "learning_rate": 5.231641566018255e-07,
      "loss": 0.6725,
      "step": 32310
    },
    {
      "epoch": 0.5030863957280858,
      "grad_norm": 4.999746322631836,
      "learning_rate": 5.230822175972206e-07,
      "loss": 0.7442,
      "step": 32315
    },
    {
      "epoch": 0.5031642367300552,
      "grad_norm": 3.1971499919891357,
      "learning_rate": 5.230002785926157e-07,
      "loss": 0.6988,
      "step": 32320
    },
    {
      "epoch": 0.5032420777320246,
      "grad_norm": 4.1423726081848145,
      "learning_rate": 5.229183395880107e-07,
      "loss": 0.7695,
      "step": 32325
    },
    {
      "epoch": 0.503319918733994,
      "grad_norm": 7.00227165222168,
      "learning_rate": 5.228364005834057e-07,
      "loss": 0.8118,
      "step": 32330
    },
    {
      "epoch": 0.5033977597359633,
      "grad_norm": 6.409590721130371,
      "learning_rate": 5.227544615788008e-07,
      "loss": 0.6859,
      "step": 32335
    },
    {
      "epoch": 0.5034756007379327,
      "grad_norm": 4.218523979187012,
      "learning_rate": 5.226725225741957e-07,
      "loss": 0.8547,
      "step": 32340
    },
    {
      "epoch": 0.5035534417399021,
      "grad_norm": 3.0705978870391846,
      "learning_rate": 5.225905835695907e-07,
      "loss": 0.687,
      "step": 32345
    },
    {
      "epoch": 0.5036312827418714,
      "grad_norm": 3.9893829822540283,
      "learning_rate": 5.225086445649858e-07,
      "loss": 0.734,
      "step": 32350
    },
    {
      "epoch": 0.5037091237438408,
      "grad_norm": 2.6017770767211914,
      "learning_rate": 5.224267055603809e-07,
      "loss": 0.7793,
      "step": 32355
    },
    {
      "epoch": 0.5037869647458102,
      "grad_norm": 3.0771327018737793,
      "learning_rate": 5.223447665557758e-07,
      "loss": 0.8074,
      "step": 32360
    },
    {
      "epoch": 0.5038648057477796,
      "grad_norm": 4.845789909362793,
      "learning_rate": 5.222628275511709e-07,
      "loss": 0.714,
      "step": 32365
    },
    {
      "epoch": 0.503942646749749,
      "grad_norm": 3.1186251640319824,
      "learning_rate": 5.22180888546566e-07,
      "loss": 0.7193,
      "step": 32370
    },
    {
      "epoch": 0.5040204877517184,
      "grad_norm": 2.8193464279174805,
      "learning_rate": 5.22098949541961e-07,
      "loss": 0.7982,
      "step": 32375
    },
    {
      "epoch": 0.5040983287536878,
      "grad_norm": 5.575962066650391,
      "learning_rate": 5.220170105373559e-07,
      "loss": 0.8848,
      "step": 32380
    },
    {
      "epoch": 0.504176169755657,
      "grad_norm": 4.0155181884765625,
      "learning_rate": 5.21935071532751e-07,
      "loss": 0.6422,
      "step": 32385
    },
    {
      "epoch": 0.5042540107576264,
      "grad_norm": 6.783470153808594,
      "learning_rate": 5.21853132528146e-07,
      "loss": 0.7836,
      "step": 32390
    },
    {
      "epoch": 0.5043318517595958,
      "grad_norm": 6.39102029800415,
      "learning_rate": 5.21771193523541e-07,
      "loss": 0.7666,
      "step": 32395
    },
    {
      "epoch": 0.5044096927615652,
      "grad_norm": 3.3229477405548096,
      "learning_rate": 5.216892545189361e-07,
      "loss": 0.7479,
      "step": 32400
    },
    {
      "epoch": 0.5044875337635346,
      "grad_norm": 4.352229118347168,
      "learning_rate": 5.216073155143311e-07,
      "loss": 0.7953,
      "step": 32405
    },
    {
      "epoch": 0.504565374765504,
      "grad_norm": 3.9259352684020996,
      "learning_rate": 5.215253765097262e-07,
      "loss": 0.7187,
      "step": 32410
    },
    {
      "epoch": 0.5046432157674734,
      "grad_norm": 6.635485649108887,
      "learning_rate": 5.214434375051212e-07,
      "loss": 0.8302,
      "step": 32415
    },
    {
      "epoch": 0.5047210567694428,
      "grad_norm": 2.1339616775512695,
      "learning_rate": 5.213614985005161e-07,
      "loss": 0.6754,
      "step": 32420
    },
    {
      "epoch": 0.5047988977714121,
      "grad_norm": 2.9039275646209717,
      "learning_rate": 5.212795594959112e-07,
      "loss": 0.5958,
      "step": 32425
    },
    {
      "epoch": 0.5048767387733815,
      "grad_norm": 3.4713127613067627,
      "learning_rate": 5.211976204913062e-07,
      "loss": 0.8021,
      "step": 32430
    },
    {
      "epoch": 0.5049545797753509,
      "grad_norm": 4.629164695739746,
      "learning_rate": 5.211156814867012e-07,
      "loss": 0.7895,
      "step": 32435
    },
    {
      "epoch": 0.5050324207773202,
      "grad_norm": 3.68489146232605,
      "learning_rate": 5.210337424820963e-07,
      "loss": 0.7423,
      "step": 32440
    },
    {
      "epoch": 0.5051102617792896,
      "grad_norm": 3.662266492843628,
      "learning_rate": 5.209518034774914e-07,
      "loss": 0.6381,
      "step": 32445
    },
    {
      "epoch": 0.505188102781259,
      "grad_norm": 4.692276477813721,
      "learning_rate": 5.208698644728864e-07,
      "loss": 0.7904,
      "step": 32450
    },
    {
      "epoch": 0.5052659437832284,
      "grad_norm": 5.333191871643066,
      "learning_rate": 5.207879254682814e-07,
      "loss": 0.6801,
      "step": 32455
    },
    {
      "epoch": 0.5053437847851977,
      "grad_norm": 2.522456407546997,
      "learning_rate": 5.207059864636764e-07,
      "loss": 0.7197,
      "step": 32460
    },
    {
      "epoch": 0.5054216257871671,
      "grad_norm": 3.86631178855896,
      "learning_rate": 5.206240474590715e-07,
      "loss": 0.7418,
      "step": 32465
    },
    {
      "epoch": 0.5054994667891365,
      "grad_norm": 3.5432724952697754,
      "learning_rate": 5.205421084544664e-07,
      "loss": 0.7702,
      "step": 32470
    },
    {
      "epoch": 0.5055773077911059,
      "grad_norm": 3.5175721645355225,
      "learning_rate": 5.204601694498615e-07,
      "loss": 0.8118,
      "step": 32475
    },
    {
      "epoch": 0.5056551487930753,
      "grad_norm": 5.169684886932373,
      "learning_rate": 5.203782304452566e-07,
      "loss": 0.7917,
      "step": 32480
    },
    {
      "epoch": 0.5057329897950447,
      "grad_norm": 8.78017520904541,
      "learning_rate": 5.202962914406515e-07,
      "loss": 0.728,
      "step": 32485
    },
    {
      "epoch": 0.505810830797014,
      "grad_norm": 3.1819334030151367,
      "learning_rate": 5.202143524360466e-07,
      "loss": 0.8763,
      "step": 32490
    },
    {
      "epoch": 0.5058886717989834,
      "grad_norm": 4.259034156799316,
      "learning_rate": 5.201324134314417e-07,
      "loss": 0.744,
      "step": 32495
    },
    {
      "epoch": 0.5059665128009527,
      "grad_norm": 4.665919303894043,
      "learning_rate": 5.200504744268367e-07,
      "loss": 0.7096,
      "step": 32500
    },
    {
      "epoch": 0.5060443538029221,
      "grad_norm": 3.8618032932281494,
      "learning_rate": 5.199685354222316e-07,
      "loss": 0.6539,
      "step": 32505
    },
    {
      "epoch": 0.5061221948048915,
      "grad_norm": 3.2800228595733643,
      "learning_rate": 5.198865964176267e-07,
      "loss": 0.7974,
      "step": 32510
    },
    {
      "epoch": 0.5062000358068609,
      "grad_norm": 3.6797404289245605,
      "learning_rate": 5.198046574130217e-07,
      "loss": 0.8237,
      "step": 32515
    },
    {
      "epoch": 0.5062778768088303,
      "grad_norm": 3.470179557800293,
      "learning_rate": 5.197227184084167e-07,
      "loss": 0.8159,
      "step": 32520
    },
    {
      "epoch": 0.5063557178107997,
      "grad_norm": 4.916172981262207,
      "learning_rate": 5.196407794038118e-07,
      "loss": 0.8046,
      "step": 32525
    },
    {
      "epoch": 0.5064335588127691,
      "grad_norm": 2.802386999130249,
      "learning_rate": 5.195588403992068e-07,
      "loss": 0.7528,
      "step": 32530
    },
    {
      "epoch": 0.5065113998147384,
      "grad_norm": 7.837001800537109,
      "learning_rate": 5.194769013946019e-07,
      "loss": 0.6811,
      "step": 32535
    },
    {
      "epoch": 0.5065892408167078,
      "grad_norm": 4.803575038909912,
      "learning_rate": 5.193949623899969e-07,
      "loss": 0.7693,
      "step": 32540
    },
    {
      "epoch": 0.5066670818186771,
      "grad_norm": 3.329767942428589,
      "learning_rate": 5.193130233853918e-07,
      "loss": 0.8052,
      "step": 32545
    },
    {
      "epoch": 0.5067449228206465,
      "grad_norm": 6.904252052307129,
      "learning_rate": 5.192310843807869e-07,
      "loss": 0.7427,
      "step": 32550
    },
    {
      "epoch": 0.5068227638226159,
      "grad_norm": 3.1481244564056396,
      "learning_rate": 5.19149145376182e-07,
      "loss": 0.8476,
      "step": 32555
    },
    {
      "epoch": 0.5069006048245853,
      "grad_norm": 5.598288059234619,
      "learning_rate": 5.190672063715769e-07,
      "loss": 0.6277,
      "step": 32560
    },
    {
      "epoch": 0.5069784458265547,
      "grad_norm": 3.148559331893921,
      "learning_rate": 5.18985267366972e-07,
      "loss": 0.6526,
      "step": 32565
    },
    {
      "epoch": 0.5070562868285241,
      "grad_norm": 7.0116868019104,
      "learning_rate": 5.189033283623671e-07,
      "loss": 0.7195,
      "step": 32570
    },
    {
      "epoch": 0.5071341278304934,
      "grad_norm": 3.9169554710388184,
      "learning_rate": 5.188213893577621e-07,
      "loss": 0.7519,
      "step": 32575
    },
    {
      "epoch": 0.5072119688324628,
      "grad_norm": 3.9739160537719727,
      "learning_rate": 5.187394503531571e-07,
      "loss": 0.703,
      "step": 32580
    },
    {
      "epoch": 0.5072898098344322,
      "grad_norm": 4.622196674346924,
      "learning_rate": 5.186575113485521e-07,
      "loss": 0.7932,
      "step": 32585
    },
    {
      "epoch": 0.5073676508364016,
      "grad_norm": 3.1050753593444824,
      "learning_rate": 5.185755723439472e-07,
      "loss": 0.7937,
      "step": 32590
    },
    {
      "epoch": 0.507445491838371,
      "grad_norm": 4.132336616516113,
      "learning_rate": 5.184936333393421e-07,
      "loss": 0.78,
      "step": 32595
    },
    {
      "epoch": 0.5075233328403403,
      "grad_norm": 4.109409332275391,
      "learning_rate": 5.184116943347372e-07,
      "loss": 0.6706,
      "step": 32600
    },
    {
      "epoch": 0.5076011738423097,
      "grad_norm": 2.7662978172302246,
      "learning_rate": 5.183297553301323e-07,
      "loss": 0.7022,
      "step": 32605
    },
    {
      "epoch": 0.5076790148442791,
      "grad_norm": 5.922182559967041,
      "learning_rate": 5.182478163255272e-07,
      "loss": 0.7174,
      "step": 32610
    },
    {
      "epoch": 0.5077568558462484,
      "grad_norm": 3.725142240524292,
      "learning_rate": 5.181658773209223e-07,
      "loss": 0.8442,
      "step": 32615
    },
    {
      "epoch": 0.5078346968482178,
      "grad_norm": 6.985898494720459,
      "learning_rate": 5.180839383163174e-07,
      "loss": 0.8086,
      "step": 32620
    },
    {
      "epoch": 0.5079125378501872,
      "grad_norm": 5.506172180175781,
      "learning_rate": 5.180019993117123e-07,
      "loss": 0.768,
      "step": 32625
    },
    {
      "epoch": 0.5079903788521566,
      "grad_norm": 3.238842010498047,
      "learning_rate": 5.179200603071073e-07,
      "loss": 0.818,
      "step": 32630
    },
    {
      "epoch": 0.508068219854126,
      "grad_norm": 6.090027809143066,
      "learning_rate": 5.178381213025024e-07,
      "loss": 0.777,
      "step": 32635
    },
    {
      "epoch": 0.5081460608560954,
      "grad_norm": 3.791621446609497,
      "learning_rate": 5.177561822978974e-07,
      "loss": 0.7164,
      "step": 32640
    },
    {
      "epoch": 0.5082239018580648,
      "grad_norm": 3.8375048637390137,
      "learning_rate": 5.176742432932925e-07,
      "loss": 0.7636,
      "step": 32645
    },
    {
      "epoch": 0.508301742860034,
      "grad_norm": 2.983212471008301,
      "learning_rate": 5.175923042886875e-07,
      "loss": 0.7797,
      "step": 32650
    },
    {
      "epoch": 0.5083795838620034,
      "grad_norm": 4.438043594360352,
      "learning_rate": 5.175103652840825e-07,
      "loss": 0.6892,
      "step": 32655
    },
    {
      "epoch": 0.5084574248639728,
      "grad_norm": 4.691067218780518,
      "learning_rate": 5.174284262794776e-07,
      "loss": 0.6904,
      "step": 32660
    },
    {
      "epoch": 0.5085352658659422,
      "grad_norm": 3.3615128993988037,
      "learning_rate": 5.173464872748725e-07,
      "loss": 0.8454,
      "step": 32665
    },
    {
      "epoch": 0.5086131068679116,
      "grad_norm": 3.7227394580841064,
      "learning_rate": 5.172645482702675e-07,
      "loss": 0.6925,
      "step": 32670
    },
    {
      "epoch": 0.508690947869881,
      "grad_norm": 3.647247552871704,
      "learning_rate": 5.171826092656626e-07,
      "loss": 0.7789,
      "step": 32675
    },
    {
      "epoch": 0.5087687888718504,
      "grad_norm": 3.516751766204834,
      "learning_rate": 5.171006702610577e-07,
      "loss": 0.815,
      "step": 32680
    },
    {
      "epoch": 0.5088466298738198,
      "grad_norm": 2.604646921157837,
      "learning_rate": 5.170187312564526e-07,
      "loss": 0.717,
      "step": 32685
    },
    {
      "epoch": 0.5089244708757891,
      "grad_norm": 5.615584373474121,
      "learning_rate": 5.169367922518477e-07,
      "loss": 0.7581,
      "step": 32690
    },
    {
      "epoch": 0.5090023118777585,
      "grad_norm": 2.866041421890259,
      "learning_rate": 5.168548532472428e-07,
      "loss": 0.7033,
      "step": 32695
    },
    {
      "epoch": 0.5090801528797279,
      "grad_norm": 8.870869636535645,
      "learning_rate": 5.167729142426379e-07,
      "loss": 0.7407,
      "step": 32700
    },
    {
      "epoch": 0.5091579938816972,
      "grad_norm": 2.678502082824707,
      "learning_rate": 5.166909752380327e-07,
      "loss": 0.6712,
      "step": 32705
    },
    {
      "epoch": 0.5092358348836666,
      "grad_norm": 6.8338446617126465,
      "learning_rate": 5.166090362334278e-07,
      "loss": 0.6264,
      "step": 32710
    },
    {
      "epoch": 0.509313675885636,
      "grad_norm": 3.769343614578247,
      "learning_rate": 5.165270972288229e-07,
      "loss": 0.7036,
      "step": 32715
    },
    {
      "epoch": 0.5093915168876054,
      "grad_norm": 4.080246448516846,
      "learning_rate": 5.164451582242178e-07,
      "loss": 0.7117,
      "step": 32720
    },
    {
      "epoch": 0.5094693578895747,
      "grad_norm": 2.5553159713745117,
      "learning_rate": 5.163632192196129e-07,
      "loss": 0.6878,
      "step": 32725
    },
    {
      "epoch": 0.5095471988915441,
      "grad_norm": 6.485873222351074,
      "learning_rate": 5.16281280215008e-07,
      "loss": 0.7334,
      "step": 32730
    },
    {
      "epoch": 0.5096250398935135,
      "grad_norm": 3.025084972381592,
      "learning_rate": 5.16199341210403e-07,
      "loss": 0.7958,
      "step": 32735
    },
    {
      "epoch": 0.5097028808954829,
      "grad_norm": 4.261486530303955,
      "learning_rate": 5.16117402205798e-07,
      "loss": 0.6992,
      "step": 32740
    },
    {
      "epoch": 0.5097807218974523,
      "grad_norm": 4.09752082824707,
      "learning_rate": 5.16035463201193e-07,
      "loss": 0.6898,
      "step": 32745
    },
    {
      "epoch": 0.5098585628994217,
      "grad_norm": 3.0008952617645264,
      "learning_rate": 5.15953524196588e-07,
      "loss": 0.7224,
      "step": 32750
    },
    {
      "epoch": 0.509936403901391,
      "grad_norm": 3.8193342685699463,
      "learning_rate": 5.15871585191983e-07,
      "loss": 0.751,
      "step": 32755
    },
    {
      "epoch": 0.5100142449033604,
      "grad_norm": 3.6056969165802,
      "learning_rate": 5.157896461873781e-07,
      "loss": 0.6964,
      "step": 32760
    },
    {
      "epoch": 0.5100920859053297,
      "grad_norm": 3.1360881328582764,
      "learning_rate": 5.157077071827731e-07,
      "loss": 0.7261,
      "step": 32765
    },
    {
      "epoch": 0.5101699269072991,
      "grad_norm": 3.1022145748138428,
      "learning_rate": 5.156257681781682e-07,
      "loss": 0.7165,
      "step": 32770
    },
    {
      "epoch": 0.5102477679092685,
      "grad_norm": 14.015339851379395,
      "learning_rate": 5.155438291735632e-07,
      "loss": 0.7785,
      "step": 32775
    },
    {
      "epoch": 0.5103256089112379,
      "grad_norm": 4.516002178192139,
      "learning_rate": 5.154618901689582e-07,
      "loss": 0.7877,
      "step": 32780
    },
    {
      "epoch": 0.5104034499132073,
      "grad_norm": 5.136211395263672,
      "learning_rate": 5.153799511643532e-07,
      "loss": 0.7094,
      "step": 32785
    },
    {
      "epoch": 0.5104812909151767,
      "grad_norm": 4.295249938964844,
      "learning_rate": 5.152980121597482e-07,
      "loss": 0.7854,
      "step": 32790
    },
    {
      "epoch": 0.5105591319171461,
      "grad_norm": 3.927434206008911,
      "learning_rate": 5.152160731551432e-07,
      "loss": 0.6865,
      "step": 32795
    },
    {
      "epoch": 0.5106369729191154,
      "grad_norm": 2.739743947982788,
      "learning_rate": 5.151341341505383e-07,
      "loss": 0.8186,
      "step": 32800
    },
    {
      "epoch": 0.5107148139210848,
      "grad_norm": 3.321277379989624,
      "learning_rate": 5.150521951459334e-07,
      "loss": 0.6557,
      "step": 32805
    },
    {
      "epoch": 0.5107926549230541,
      "grad_norm": 3.46451473236084,
      "learning_rate": 5.149702561413283e-07,
      "loss": 0.6972,
      "step": 32810
    },
    {
      "epoch": 0.5108704959250235,
      "grad_norm": 4.951078414916992,
      "learning_rate": 5.148883171367234e-07,
      "loss": 0.7714,
      "step": 32815
    },
    {
      "epoch": 0.5109483369269929,
      "grad_norm": 4.750621795654297,
      "learning_rate": 5.148063781321185e-07,
      "loss": 0.8763,
      "step": 32820
    },
    {
      "epoch": 0.5110261779289623,
      "grad_norm": 5.102913856506348,
      "learning_rate": 5.147244391275136e-07,
      "loss": 0.767,
      "step": 32825
    },
    {
      "epoch": 0.5111040189309317,
      "grad_norm": 2.441957950592041,
      "learning_rate": 5.146425001229084e-07,
      "loss": 0.6325,
      "step": 32830
    },
    {
      "epoch": 0.5111818599329011,
      "grad_norm": 3.343223810195923,
      "learning_rate": 5.145605611183035e-07,
      "loss": 0.6433,
      "step": 32835
    },
    {
      "epoch": 0.5112597009348704,
      "grad_norm": 2.606261968612671,
      "learning_rate": 5.144786221136986e-07,
      "loss": 0.7598,
      "step": 32840
    },
    {
      "epoch": 0.5113375419368398,
      "grad_norm": 6.7902116775512695,
      "learning_rate": 5.143966831090935e-07,
      "loss": 0.7675,
      "step": 32845
    },
    {
      "epoch": 0.5114153829388092,
      "grad_norm": 3.269784450531006,
      "learning_rate": 5.143147441044886e-07,
      "loss": 0.7007,
      "step": 32850
    },
    {
      "epoch": 0.5114932239407786,
      "grad_norm": 9.681060791015625,
      "learning_rate": 5.142328050998837e-07,
      "loss": 0.7995,
      "step": 32855
    },
    {
      "epoch": 0.511571064942748,
      "grad_norm": 3.9746077060699463,
      "learning_rate": 5.141508660952787e-07,
      "loss": 0.6747,
      "step": 32860
    },
    {
      "epoch": 0.5116489059447173,
      "grad_norm": 3.9287326335906982,
      "learning_rate": 5.140689270906737e-07,
      "loss": 0.7338,
      "step": 32865
    },
    {
      "epoch": 0.5117267469466867,
      "grad_norm": 3.1837215423583984,
      "learning_rate": 5.139869880860687e-07,
      "loss": 0.7613,
      "step": 32870
    },
    {
      "epoch": 0.511804587948656,
      "grad_norm": 2.7956628799438477,
      "learning_rate": 5.139050490814637e-07,
      "loss": 0.8087,
      "step": 32875
    },
    {
      "epoch": 0.5118824289506254,
      "grad_norm": 4.300243854522705,
      "learning_rate": 5.138231100768587e-07,
      "loss": 0.8063,
      "step": 32880
    },
    {
      "epoch": 0.5119602699525948,
      "grad_norm": 3.549539566040039,
      "learning_rate": 5.137411710722538e-07,
      "loss": 0.6862,
      "step": 32885
    },
    {
      "epoch": 0.5120381109545642,
      "grad_norm": 3.004115343093872,
      "learning_rate": 5.136592320676488e-07,
      "loss": 0.7664,
      "step": 32890
    },
    {
      "epoch": 0.5121159519565336,
      "grad_norm": 3.5236361026763916,
      "learning_rate": 5.135772930630439e-07,
      "loss": 0.7501,
      "step": 32895
    },
    {
      "epoch": 0.512193792958503,
      "grad_norm": 3.880519390106201,
      "learning_rate": 5.134953540584389e-07,
      "loss": 0.6527,
      "step": 32900
    },
    {
      "epoch": 0.5122716339604724,
      "grad_norm": 3.0610344409942627,
      "learning_rate": 5.134134150538339e-07,
      "loss": 0.6841,
      "step": 32905
    },
    {
      "epoch": 0.5123494749624418,
      "grad_norm": 8.654298782348633,
      "learning_rate": 5.133314760492289e-07,
      "loss": 0.7815,
      "step": 32910
    },
    {
      "epoch": 0.512427315964411,
      "grad_norm": 8.904096603393555,
      "learning_rate": 5.13249537044624e-07,
      "loss": 0.8809,
      "step": 32915
    },
    {
      "epoch": 0.5125051569663804,
      "grad_norm": 5.487973213195801,
      "learning_rate": 5.131675980400189e-07,
      "loss": 0.7971,
      "step": 32920
    },
    {
      "epoch": 0.5125829979683498,
      "grad_norm": 4.270529747009277,
      "learning_rate": 5.13085659035414e-07,
      "loss": 0.8249,
      "step": 32925
    },
    {
      "epoch": 0.5126608389703192,
      "grad_norm": 4.06682014465332,
      "learning_rate": 5.130037200308091e-07,
      "loss": 0.7702,
      "step": 32930
    },
    {
      "epoch": 0.5127386799722886,
      "grad_norm": 7.298919200897217,
      "learning_rate": 5.12921781026204e-07,
      "loss": 0.8415,
      "step": 32935
    },
    {
      "epoch": 0.512816520974258,
      "grad_norm": 6.34363317489624,
      "learning_rate": 5.128398420215991e-07,
      "loss": 0.7976,
      "step": 32940
    },
    {
      "epoch": 0.5128943619762274,
      "grad_norm": 3.7378201484680176,
      "learning_rate": 5.127579030169942e-07,
      "loss": 0.8445,
      "step": 32945
    },
    {
      "epoch": 0.5129722029781968,
      "grad_norm": 3.1709158420562744,
      "learning_rate": 5.126759640123891e-07,
      "loss": 0.6926,
      "step": 32950
    },
    {
      "epoch": 0.5130500439801661,
      "grad_norm": 10.128692626953125,
      "learning_rate": 5.125940250077841e-07,
      "loss": 0.7289,
      "step": 32955
    },
    {
      "epoch": 0.5131278849821355,
      "grad_norm": 4.398654937744141,
      "learning_rate": 5.125120860031792e-07,
      "loss": 0.7334,
      "step": 32960
    },
    {
      "epoch": 0.5132057259841049,
      "grad_norm": 3.7946243286132812,
      "learning_rate": 5.124301469985743e-07,
      "loss": 0.6481,
      "step": 32965
    },
    {
      "epoch": 0.5132835669860742,
      "grad_norm": 2.977980375289917,
      "learning_rate": 5.123482079939692e-07,
      "loss": 0.845,
      "step": 32970
    },
    {
      "epoch": 0.5133614079880436,
      "grad_norm": 3.9080305099487305,
      "learning_rate": 5.122662689893643e-07,
      "loss": 0.932,
      "step": 32975
    },
    {
      "epoch": 0.513439248990013,
      "grad_norm": 2.8391921520233154,
      "learning_rate": 5.121843299847594e-07,
      "loss": 0.6624,
      "step": 32980
    },
    {
      "epoch": 0.5135170899919824,
      "grad_norm": 3.742037296295166,
      "learning_rate": 5.121023909801544e-07,
      "loss": 0.6879,
      "step": 32985
    },
    {
      "epoch": 0.5135949309939517,
      "grad_norm": 5.588380813598633,
      "learning_rate": 5.120204519755493e-07,
      "loss": 0.698,
      "step": 32990
    },
    {
      "epoch": 0.5136727719959211,
      "grad_norm": 3.2711422443389893,
      "learning_rate": 5.119385129709444e-07,
      "loss": 0.7561,
      "step": 32995
    },
    {
      "epoch": 0.5137506129978905,
      "grad_norm": 5.483699798583984,
      "learning_rate": 5.118565739663394e-07,
      "loss": 0.7772,
      "step": 33000
    },
    {
      "epoch": 0.5138284539998599,
      "grad_norm": 3.3001506328582764,
      "learning_rate": 5.117746349617345e-07,
      "loss": 0.663,
      "step": 33005
    },
    {
      "epoch": 0.5139062950018293,
      "grad_norm": 2.37276554107666,
      "learning_rate": 5.116926959571295e-07,
      "loss": 0.7179,
      "step": 33010
    },
    {
      "epoch": 0.5139841360037987,
      "grad_norm": 9.586172103881836,
      "learning_rate": 5.116107569525245e-07,
      "loss": 0.6506,
      "step": 33015
    },
    {
      "epoch": 0.5140619770057681,
      "grad_norm": 8.20467472076416,
      "learning_rate": 5.115288179479196e-07,
      "loss": 0.7228,
      "step": 33020
    },
    {
      "epoch": 0.5141398180077374,
      "grad_norm": 5.777222156524658,
      "learning_rate": 5.114468789433146e-07,
      "loss": 0.7303,
      "step": 33025
    },
    {
      "epoch": 0.5142176590097067,
      "grad_norm": 3.091376304626465,
      "learning_rate": 5.113649399387095e-07,
      "loss": 0.6698,
      "step": 33030
    },
    {
      "epoch": 0.5142955000116761,
      "grad_norm": 3.441102981567383,
      "learning_rate": 5.112830009341046e-07,
      "loss": 0.7443,
      "step": 33035
    },
    {
      "epoch": 0.5143733410136455,
      "grad_norm": 4.060532569885254,
      "learning_rate": 5.112010619294997e-07,
      "loss": 0.7735,
      "step": 33040
    },
    {
      "epoch": 0.5144511820156149,
      "grad_norm": 3.060643196105957,
      "learning_rate": 5.111191229248946e-07,
      "loss": 0.7394,
      "step": 33045
    },
    {
      "epoch": 0.5145290230175843,
      "grad_norm": 4.767629623413086,
      "learning_rate": 5.110371839202897e-07,
      "loss": 0.7703,
      "step": 33050
    },
    {
      "epoch": 0.5146068640195537,
      "grad_norm": 3.5270395278930664,
      "learning_rate": 5.109552449156848e-07,
      "loss": 0.8369,
      "step": 33055
    },
    {
      "epoch": 0.5146847050215231,
      "grad_norm": 4.064676761627197,
      "learning_rate": 5.108733059110797e-07,
      "loss": 0.8119,
      "step": 33060
    },
    {
      "epoch": 0.5147625460234924,
      "grad_norm": 2.927370071411133,
      "learning_rate": 5.107913669064748e-07,
      "loss": 0.746,
      "step": 33065
    },
    {
      "epoch": 0.5148403870254618,
      "grad_norm": 3.0840799808502197,
      "learning_rate": 5.107094279018698e-07,
      "loss": 0.683,
      "step": 33070
    },
    {
      "epoch": 0.5149182280274311,
      "grad_norm": 3.753798723220825,
      "learning_rate": 5.106274888972648e-07,
      "loss": 0.8119,
      "step": 33075
    },
    {
      "epoch": 0.5149960690294005,
      "grad_norm": 3.8448100090026855,
      "learning_rate": 5.105455498926598e-07,
      "loss": 0.8285,
      "step": 33080
    },
    {
      "epoch": 0.5150739100313699,
      "grad_norm": 5.474903583526611,
      "learning_rate": 5.104636108880549e-07,
      "loss": 0.8128,
      "step": 33085
    },
    {
      "epoch": 0.5151517510333393,
      "grad_norm": 5.70355749130249,
      "learning_rate": 5.1038167188345e-07,
      "loss": 0.7753,
      "step": 33090
    },
    {
      "epoch": 0.5152295920353087,
      "grad_norm": 3.046039581298828,
      "learning_rate": 5.10299732878845e-07,
      "loss": 0.6484,
      "step": 33095
    },
    {
      "epoch": 0.5153074330372781,
      "grad_norm": 5.240191459655762,
      "learning_rate": 5.1021779387424e-07,
      "loss": 0.762,
      "step": 33100
    },
    {
      "epoch": 0.5153852740392474,
      "grad_norm": 3.799865484237671,
      "learning_rate": 5.101358548696351e-07,
      "loss": 0.8223,
      "step": 33105
    },
    {
      "epoch": 0.5154631150412168,
      "grad_norm": 4.9581427574157715,
      "learning_rate": 5.1005391586503e-07,
      "loss": 0.7598,
      "step": 33110
    },
    {
      "epoch": 0.5155409560431862,
      "grad_norm": 2.5600438117980957,
      "learning_rate": 5.09971976860425e-07,
      "loss": 0.7384,
      "step": 33115
    },
    {
      "epoch": 0.5156187970451556,
      "grad_norm": 2.921954393386841,
      "learning_rate": 5.098900378558201e-07,
      "loss": 0.6934,
      "step": 33120
    },
    {
      "epoch": 0.515696638047125,
      "grad_norm": 5.0808000564575195,
      "learning_rate": 5.098080988512151e-07,
      "loss": 0.7996,
      "step": 33125
    },
    {
      "epoch": 0.5157744790490943,
      "grad_norm": 6.499424934387207,
      "learning_rate": 5.097261598466102e-07,
      "loss": 0.784,
      "step": 33130
    },
    {
      "epoch": 0.5158523200510637,
      "grad_norm": 2.9274024963378906,
      "learning_rate": 5.096442208420052e-07,
      "loss": 0.7781,
      "step": 33135
    },
    {
      "epoch": 0.515930161053033,
      "grad_norm": 2.921600341796875,
      "learning_rate": 5.095622818374002e-07,
      "loss": 0.9012,
      "step": 33140
    },
    {
      "epoch": 0.5160080020550024,
      "grad_norm": 4.178603172302246,
      "learning_rate": 5.094803428327953e-07,
      "loss": 0.7879,
      "step": 33145
    },
    {
      "epoch": 0.5160858430569718,
      "grad_norm": 3.511152505874634,
      "learning_rate": 5.093984038281904e-07,
      "loss": 0.8024,
      "step": 33150
    },
    {
      "epoch": 0.5161636840589412,
      "grad_norm": 2.5039076805114746,
      "learning_rate": 5.093164648235852e-07,
      "loss": 0.7441,
      "step": 33155
    },
    {
      "epoch": 0.5162415250609106,
      "grad_norm": 3.674992561340332,
      "learning_rate": 5.092345258189803e-07,
      "loss": 0.7008,
      "step": 33160
    },
    {
      "epoch": 0.51631936606288,
      "grad_norm": 8.318180084228516,
      "learning_rate": 5.091525868143754e-07,
      "loss": 0.6677,
      "step": 33165
    },
    {
      "epoch": 0.5163972070648494,
      "grad_norm": 3.729417562484741,
      "learning_rate": 5.090706478097703e-07,
      "loss": 0.762,
      "step": 33170
    },
    {
      "epoch": 0.5164750480668188,
      "grad_norm": 3.0601065158843994,
      "learning_rate": 5.089887088051654e-07,
      "loss": 0.7567,
      "step": 33175
    },
    {
      "epoch": 0.516552889068788,
      "grad_norm": 2.9008240699768066,
      "learning_rate": 5.089067698005605e-07,
      "loss": 0.7682,
      "step": 33180
    },
    {
      "epoch": 0.5166307300707574,
      "grad_norm": 3.3500235080718994,
      "learning_rate": 5.088248307959555e-07,
      "loss": 0.7385,
      "step": 33185
    },
    {
      "epoch": 0.5167085710727268,
      "grad_norm": 4.15496301651001,
      "learning_rate": 5.087428917913505e-07,
      "loss": 0.7066,
      "step": 33190
    },
    {
      "epoch": 0.5167864120746962,
      "grad_norm": 3.7832906246185303,
      "learning_rate": 5.086609527867455e-07,
      "loss": 0.7832,
      "step": 33195
    },
    {
      "epoch": 0.5168642530766656,
      "grad_norm": 3.152782440185547,
      "learning_rate": 5.085790137821405e-07,
      "loss": 0.7015,
      "step": 33200
    },
    {
      "epoch": 0.516942094078635,
      "grad_norm": 4.066208362579346,
      "learning_rate": 5.084970747775355e-07,
      "loss": 0.7001,
      "step": 33205
    },
    {
      "epoch": 0.5170199350806044,
      "grad_norm": 3.806546211242676,
      "learning_rate": 5.084151357729306e-07,
      "loss": 0.875,
      "step": 33210
    },
    {
      "epoch": 0.5170977760825738,
      "grad_norm": 5.553750991821289,
      "learning_rate": 5.083331967683257e-07,
      "loss": 0.7033,
      "step": 33215
    },
    {
      "epoch": 0.5171756170845431,
      "grad_norm": 3.6293158531188965,
      "learning_rate": 5.082512577637207e-07,
      "loss": 0.7778,
      "step": 33220
    },
    {
      "epoch": 0.5172534580865125,
      "grad_norm": 3.184992790222168,
      "learning_rate": 5.081693187591157e-07,
      "loss": 0.6758,
      "step": 33225
    },
    {
      "epoch": 0.5173312990884819,
      "grad_norm": 6.356576442718506,
      "learning_rate": 5.080873797545108e-07,
      "loss": 0.6996,
      "step": 33230
    },
    {
      "epoch": 0.5174091400904512,
      "grad_norm": 3.215970277786255,
      "learning_rate": 5.080054407499057e-07,
      "loss": 0.8301,
      "step": 33235
    },
    {
      "epoch": 0.5174869810924206,
      "grad_norm": 3.0766730308532715,
      "learning_rate": 5.079235017453007e-07,
      "loss": 0.7046,
      "step": 33240
    },
    {
      "epoch": 0.51756482209439,
      "grad_norm": 3.9885709285736084,
      "learning_rate": 5.078415627406958e-07,
      "loss": 0.6887,
      "step": 33245
    },
    {
      "epoch": 0.5176426630963594,
      "grad_norm": 4.094118118286133,
      "learning_rate": 5.077596237360908e-07,
      "loss": 0.7376,
      "step": 33250
    },
    {
      "epoch": 0.5177205040983287,
      "grad_norm": 3.650657892227173,
      "learning_rate": 5.076776847314859e-07,
      "loss": 0.6938,
      "step": 33255
    },
    {
      "epoch": 0.5177983451002981,
      "grad_norm": 6.65205192565918,
      "learning_rate": 5.075957457268809e-07,
      "loss": 0.7593,
      "step": 33260
    },
    {
      "epoch": 0.5178761861022675,
      "grad_norm": 5.344778060913086,
      "learning_rate": 5.075138067222759e-07,
      "loss": 0.8098,
      "step": 33265
    },
    {
      "epoch": 0.5179540271042369,
      "grad_norm": 4.021899223327637,
      "learning_rate": 5.07431867717671e-07,
      "loss": 0.7243,
      "step": 33270
    },
    {
      "epoch": 0.5180318681062063,
      "grad_norm": 3.4968960285186768,
      "learning_rate": 5.07349928713066e-07,
      "loss": 0.8098,
      "step": 33275
    },
    {
      "epoch": 0.5181097091081757,
      "grad_norm": 6.943088054656982,
      "learning_rate": 5.072679897084609e-07,
      "loss": 0.5955,
      "step": 33280
    },
    {
      "epoch": 0.5181875501101451,
      "grad_norm": 3.251338243484497,
      "learning_rate": 5.07186050703856e-07,
      "loss": 0.7478,
      "step": 33285
    },
    {
      "epoch": 0.5182653911121144,
      "grad_norm": 4.761424541473389,
      "learning_rate": 5.071041116992511e-07,
      "loss": 0.7041,
      "step": 33290
    },
    {
      "epoch": 0.5183432321140837,
      "grad_norm": 5.8884148597717285,
      "learning_rate": 5.07022172694646e-07,
      "loss": 0.6648,
      "step": 33295
    },
    {
      "epoch": 0.5184210731160531,
      "grad_norm": 3.3632352352142334,
      "learning_rate": 5.069402336900411e-07,
      "loss": 0.6988,
      "step": 33300
    },
    {
      "epoch": 0.5184989141180225,
      "grad_norm": 4.041135787963867,
      "learning_rate": 5.068582946854362e-07,
      "loss": 0.8575,
      "step": 33305
    },
    {
      "epoch": 0.5185767551199919,
      "grad_norm": 4.31258487701416,
      "learning_rate": 5.067763556808312e-07,
      "loss": 0.7794,
      "step": 33310
    },
    {
      "epoch": 0.5186545961219613,
      "grad_norm": 3.8441247940063477,
      "learning_rate": 5.066944166762261e-07,
      "loss": 0.6312,
      "step": 33315
    },
    {
      "epoch": 0.5187324371239307,
      "grad_norm": 3.7504453659057617,
      "learning_rate": 5.066124776716212e-07,
      "loss": 0.7099,
      "step": 33320
    },
    {
      "epoch": 0.5188102781259001,
      "grad_norm": 3.6207613945007324,
      "learning_rate": 5.065305386670162e-07,
      "loss": 0.7878,
      "step": 33325
    },
    {
      "epoch": 0.5188881191278694,
      "grad_norm": 3.9682514667510986,
      "learning_rate": 5.064485996624112e-07,
      "loss": 0.7551,
      "step": 33330
    },
    {
      "epoch": 0.5189659601298388,
      "grad_norm": 2.6843957901000977,
      "learning_rate": 5.063666606578063e-07,
      "loss": 0.8254,
      "step": 33335
    },
    {
      "epoch": 0.5190438011318081,
      "grad_norm": 4.533762454986572,
      "learning_rate": 5.062847216532014e-07,
      "loss": 0.816,
      "step": 33340
    },
    {
      "epoch": 0.5191216421337775,
      "grad_norm": 4.434187889099121,
      "learning_rate": 5.062027826485964e-07,
      "loss": 0.911,
      "step": 33345
    },
    {
      "epoch": 0.5191994831357469,
      "grad_norm": 4.167538166046143,
      "learning_rate": 5.061208436439914e-07,
      "loss": 0.7596,
      "step": 33350
    },
    {
      "epoch": 0.5192773241377163,
      "grad_norm": 4.479653358459473,
      "learning_rate": 5.060389046393864e-07,
      "loss": 0.6277,
      "step": 33355
    },
    {
      "epoch": 0.5193551651396857,
      "grad_norm": 3.301845073699951,
      "learning_rate": 5.059569656347814e-07,
      "loss": 0.6446,
      "step": 33360
    },
    {
      "epoch": 0.5194330061416551,
      "grad_norm": 4.289400577545166,
      "learning_rate": 5.058750266301765e-07,
      "loss": 0.6907,
      "step": 33365
    },
    {
      "epoch": 0.5195108471436244,
      "grad_norm": 3.2664413452148438,
      "learning_rate": 5.057930876255715e-07,
      "loss": 0.7415,
      "step": 33370
    },
    {
      "epoch": 0.5195886881455938,
      "grad_norm": 3.773880958557129,
      "learning_rate": 5.057111486209665e-07,
      "loss": 0.8199,
      "step": 33375
    },
    {
      "epoch": 0.5196665291475632,
      "grad_norm": 5.109145641326904,
      "learning_rate": 5.056292096163616e-07,
      "loss": 0.7693,
      "step": 33380
    },
    {
      "epoch": 0.5197443701495326,
      "grad_norm": 6.951995372772217,
      "learning_rate": 5.055472706117566e-07,
      "loss": 0.8214,
      "step": 33385
    },
    {
      "epoch": 0.519822211151502,
      "grad_norm": 7.535697937011719,
      "learning_rate": 5.054653316071516e-07,
      "loss": 0.8201,
      "step": 33390
    },
    {
      "epoch": 0.5199000521534713,
      "grad_norm": 3.746630907058716,
      "learning_rate": 5.053833926025466e-07,
      "loss": 0.6445,
      "step": 33395
    },
    {
      "epoch": 0.5199778931554407,
      "grad_norm": 3.4224538803100586,
      "learning_rate": 5.053014535979417e-07,
      "loss": 0.7483,
      "step": 33400
    },
    {
      "epoch": 0.52005573415741,
      "grad_norm": 7.043457508087158,
      "learning_rate": 5.052195145933366e-07,
      "loss": 0.7794,
      "step": 33405
    },
    {
      "epoch": 0.5201335751593794,
      "grad_norm": 2.987311601638794,
      "learning_rate": 5.051375755887317e-07,
      "loss": 0.7109,
      "step": 33410
    },
    {
      "epoch": 0.5202114161613488,
      "grad_norm": 3.201394557952881,
      "learning_rate": 5.050556365841268e-07,
      "loss": 0.6764,
      "step": 33415
    },
    {
      "epoch": 0.5202892571633182,
      "grad_norm": 4.287966251373291,
      "learning_rate": 5.049736975795217e-07,
      "loss": 0.8341,
      "step": 33420
    },
    {
      "epoch": 0.5203670981652876,
      "grad_norm": 4.316837310791016,
      "learning_rate": 5.048917585749168e-07,
      "loss": 0.9356,
      "step": 33425
    },
    {
      "epoch": 0.520444939167257,
      "grad_norm": 5.530353546142578,
      "learning_rate": 5.048098195703119e-07,
      "loss": 0.7405,
      "step": 33430
    },
    {
      "epoch": 0.5205227801692264,
      "grad_norm": 4.705060005187988,
      "learning_rate": 5.047278805657068e-07,
      "loss": 0.7871,
      "step": 33435
    },
    {
      "epoch": 0.5206006211711958,
      "grad_norm": 3.583533525466919,
      "learning_rate": 5.046459415611018e-07,
      "loss": 0.7806,
      "step": 33440
    },
    {
      "epoch": 0.520678462173165,
      "grad_norm": 9.026838302612305,
      "learning_rate": 5.045640025564969e-07,
      "loss": 0.7607,
      "step": 33445
    },
    {
      "epoch": 0.5207563031751344,
      "grad_norm": 3.7230563163757324,
      "learning_rate": 5.04482063551892e-07,
      "loss": 0.675,
      "step": 33450
    },
    {
      "epoch": 0.5208341441771038,
      "grad_norm": 5.408843517303467,
      "learning_rate": 5.04400124547287e-07,
      "loss": 0.8425,
      "step": 33455
    },
    {
      "epoch": 0.5209119851790732,
      "grad_norm": 4.309723854064941,
      "learning_rate": 5.04318185542682e-07,
      "loss": 0.6702,
      "step": 33460
    },
    {
      "epoch": 0.5209898261810426,
      "grad_norm": 4.629941940307617,
      "learning_rate": 5.042362465380771e-07,
      "loss": 0.7241,
      "step": 33465
    },
    {
      "epoch": 0.521067667183012,
      "grad_norm": 3.627274751663208,
      "learning_rate": 5.041543075334721e-07,
      "loss": 0.6421,
      "step": 33470
    },
    {
      "epoch": 0.5211455081849814,
      "grad_norm": 5.469413757324219,
      "learning_rate": 5.040723685288671e-07,
      "loss": 0.7054,
      "step": 33475
    },
    {
      "epoch": 0.5212233491869507,
      "grad_norm": 2.723989248275757,
      "learning_rate": 5.039904295242621e-07,
      "loss": 0.761,
      "step": 33480
    },
    {
      "epoch": 0.5213011901889201,
      "grad_norm": 3.7360827922821045,
      "learning_rate": 5.039084905196571e-07,
      "loss": 0.7293,
      "step": 33485
    },
    {
      "epoch": 0.5213790311908895,
      "grad_norm": 4.112342834472656,
      "learning_rate": 5.038265515150522e-07,
      "loss": 0.7907,
      "step": 33490
    },
    {
      "epoch": 0.5214568721928589,
      "grad_norm": 2.8697752952575684,
      "learning_rate": 5.037446125104472e-07,
      "loss": 0.7141,
      "step": 33495
    },
    {
      "epoch": 0.5215347131948282,
      "grad_norm": 5.6111063957214355,
      "learning_rate": 5.036626735058422e-07,
      "loss": 0.767,
      "step": 33500
    },
    {
      "epoch": 0.5216125541967976,
      "grad_norm": 3.967881679534912,
      "learning_rate": 5.035807345012373e-07,
      "loss": 0.7054,
      "step": 33505
    },
    {
      "epoch": 0.521690395198767,
      "grad_norm": 2.8242242336273193,
      "learning_rate": 5.034987954966324e-07,
      "loss": 0.7896,
      "step": 33510
    },
    {
      "epoch": 0.5217682362007364,
      "grad_norm": 3.550645112991333,
      "learning_rate": 5.034168564920273e-07,
      "loss": 0.6847,
      "step": 33515
    },
    {
      "epoch": 0.5218460772027057,
      "grad_norm": 4.227738857269287,
      "learning_rate": 5.033349174874223e-07,
      "loss": 0.6629,
      "step": 33520
    },
    {
      "epoch": 0.5219239182046751,
      "grad_norm": 5.2480363845825195,
      "learning_rate": 5.032529784828174e-07,
      "loss": 0.6852,
      "step": 33525
    },
    {
      "epoch": 0.5220017592066445,
      "grad_norm": 4.925660610198975,
      "learning_rate": 5.031710394782123e-07,
      "loss": 0.7376,
      "step": 33530
    },
    {
      "epoch": 0.5220796002086139,
      "grad_norm": 3.251960039138794,
      "learning_rate": 5.030891004736074e-07,
      "loss": 0.6985,
      "step": 33535
    },
    {
      "epoch": 0.5221574412105833,
      "grad_norm": 2.8661468029022217,
      "learning_rate": 5.030071614690025e-07,
      "loss": 0.8149,
      "step": 33540
    },
    {
      "epoch": 0.5222352822125527,
      "grad_norm": 4.4495391845703125,
      "learning_rate": 5.029252224643975e-07,
      "loss": 0.7139,
      "step": 33545
    },
    {
      "epoch": 0.5223131232145221,
      "grad_norm": 3.4561924934387207,
      "learning_rate": 5.028432834597925e-07,
      "loss": 0.6793,
      "step": 33550
    },
    {
      "epoch": 0.5223909642164914,
      "grad_norm": 2.496107578277588,
      "learning_rate": 5.027613444551876e-07,
      "loss": 0.6727,
      "step": 33555
    },
    {
      "epoch": 0.5224688052184607,
      "grad_norm": 5.665369510650635,
      "learning_rate": 5.026794054505825e-07,
      "loss": 0.6331,
      "step": 33560
    },
    {
      "epoch": 0.5225466462204301,
      "grad_norm": 3.787381172180176,
      "learning_rate": 5.025974664459775e-07,
      "loss": 0.7289,
      "step": 33565
    },
    {
      "epoch": 0.5226244872223995,
      "grad_norm": 3.8780858516693115,
      "learning_rate": 5.025155274413726e-07,
      "loss": 0.6906,
      "step": 33570
    },
    {
      "epoch": 0.5227023282243689,
      "grad_norm": 6.018281936645508,
      "learning_rate": 5.024335884367677e-07,
      "loss": 0.6958,
      "step": 33575
    },
    {
      "epoch": 0.5227801692263383,
      "grad_norm": 2.9917240142822266,
      "learning_rate": 5.023516494321627e-07,
      "loss": 0.561,
      "step": 33580
    },
    {
      "epoch": 0.5228580102283077,
      "grad_norm": 5.007068157196045,
      "learning_rate": 5.022697104275577e-07,
      "loss": 0.7202,
      "step": 33585
    },
    {
      "epoch": 0.5229358512302771,
      "grad_norm": 5.185565948486328,
      "learning_rate": 5.021877714229528e-07,
      "loss": 0.8364,
      "step": 33590
    },
    {
      "epoch": 0.5230136922322464,
      "grad_norm": 3.950190782546997,
      "learning_rate": 5.021058324183478e-07,
      "loss": 0.7624,
      "step": 33595
    },
    {
      "epoch": 0.5230915332342158,
      "grad_norm": 3.9026620388031006,
      "learning_rate": 5.020238934137428e-07,
      "loss": 0.8187,
      "step": 33600
    },
    {
      "epoch": 0.5231693742361851,
      "grad_norm": 4.830386161804199,
      "learning_rate": 5.019419544091378e-07,
      "loss": 0.7512,
      "step": 33605
    },
    {
      "epoch": 0.5232472152381545,
      "grad_norm": 3.33174991607666,
      "learning_rate": 5.018600154045328e-07,
      "loss": 0.7256,
      "step": 33610
    },
    {
      "epoch": 0.5233250562401239,
      "grad_norm": 7.180850982666016,
      "learning_rate": 5.017780763999279e-07,
      "loss": 0.8775,
      "step": 33615
    },
    {
      "epoch": 0.5234028972420933,
      "grad_norm": 2.395169973373413,
      "learning_rate": 5.016961373953229e-07,
      "loss": 0.7788,
      "step": 33620
    },
    {
      "epoch": 0.5234807382440627,
      "grad_norm": 4.609184741973877,
      "learning_rate": 5.016141983907179e-07,
      "loss": 0.7716,
      "step": 33625
    },
    {
      "epoch": 0.5235585792460321,
      "grad_norm": 3.3770880699157715,
      "learning_rate": 5.01532259386113e-07,
      "loss": 0.8541,
      "step": 33630
    },
    {
      "epoch": 0.5236364202480014,
      "grad_norm": 5.100673675537109,
      "learning_rate": 5.014503203815081e-07,
      "loss": 0.6069,
      "step": 33635
    },
    {
      "epoch": 0.5237142612499708,
      "grad_norm": 2.7191946506500244,
      "learning_rate": 5.013683813769029e-07,
      "loss": 0.8191,
      "step": 33640
    },
    {
      "epoch": 0.5237921022519402,
      "grad_norm": 5.074402332305908,
      "learning_rate": 5.01286442372298e-07,
      "loss": 0.8412,
      "step": 33645
    },
    {
      "epoch": 0.5238699432539096,
      "grad_norm": 4.790443420410156,
      "learning_rate": 5.012045033676931e-07,
      "loss": 0.8082,
      "step": 33650
    },
    {
      "epoch": 0.523947784255879,
      "grad_norm": 3.2585701942443848,
      "learning_rate": 5.01122564363088e-07,
      "loss": 0.7198,
      "step": 33655
    },
    {
      "epoch": 0.5240256252578483,
      "grad_norm": 6.707359313964844,
      "learning_rate": 5.010406253584831e-07,
      "loss": 0.7091,
      "step": 33660
    },
    {
      "epoch": 0.5241034662598177,
      "grad_norm": 3.2364871501922607,
      "learning_rate": 5.009586863538782e-07,
      "loss": 0.666,
      "step": 33665
    },
    {
      "epoch": 0.524181307261787,
      "grad_norm": 4.6501336097717285,
      "learning_rate": 5.008767473492732e-07,
      "loss": 0.7659,
      "step": 33670
    },
    {
      "epoch": 0.5242591482637564,
      "grad_norm": 3.908421039581299,
      "learning_rate": 5.007948083446682e-07,
      "loss": 0.6782,
      "step": 33675
    },
    {
      "epoch": 0.5243369892657258,
      "grad_norm": 2.3602583408355713,
      "learning_rate": 5.007128693400632e-07,
      "loss": 0.7121,
      "step": 33680
    },
    {
      "epoch": 0.5244148302676952,
      "grad_norm": 3.5516517162323,
      "learning_rate": 5.006309303354582e-07,
      "loss": 0.756,
      "step": 33685
    },
    {
      "epoch": 0.5244926712696646,
      "grad_norm": 3.839874267578125,
      "learning_rate": 5.005489913308533e-07,
      "loss": 0.8265,
      "step": 33690
    },
    {
      "epoch": 0.524570512271634,
      "grad_norm": 3.4168505668640137,
      "learning_rate": 5.004670523262483e-07,
      "loss": 0.6943,
      "step": 33695
    },
    {
      "epoch": 0.5246483532736034,
      "grad_norm": 4.070676803588867,
      "learning_rate": 5.003851133216434e-07,
      "loss": 0.6514,
      "step": 33700
    },
    {
      "epoch": 0.5247261942755728,
      "grad_norm": 3.966780185699463,
      "learning_rate": 5.003031743170384e-07,
      "loss": 0.8231,
      "step": 33705
    },
    {
      "epoch": 0.524804035277542,
      "grad_norm": 4.187429428100586,
      "learning_rate": 5.002212353124334e-07,
      "loss": 0.7539,
      "step": 33710
    },
    {
      "epoch": 0.5248818762795114,
      "grad_norm": 3.7121341228485107,
      "learning_rate": 5.001392963078285e-07,
      "loss": 0.7261,
      "step": 33715
    },
    {
      "epoch": 0.5249597172814808,
      "grad_norm": 5.748007297515869,
      "learning_rate": 5.000573573032234e-07,
      "loss": 0.6088,
      "step": 33720
    },
    {
      "epoch": 0.5250375582834502,
      "grad_norm": 2.726752519607544,
      "learning_rate": 4.999754182986185e-07,
      "loss": 0.7888,
      "step": 33725
    },
    {
      "epoch": 0.5251153992854196,
      "grad_norm": 2.751572370529175,
      "learning_rate": 4.998934792940135e-07,
      "loss": 0.7888,
      "step": 33730
    },
    {
      "epoch": 0.525193240287389,
      "grad_norm": 5.40326452255249,
      "learning_rate": 4.998115402894085e-07,
      "loss": 0.8053,
      "step": 33735
    },
    {
      "epoch": 0.5252710812893584,
      "grad_norm": 3.117835760116577,
      "learning_rate": 4.997296012848036e-07,
      "loss": 0.7607,
      "step": 33740
    },
    {
      "epoch": 0.5253489222913277,
      "grad_norm": 2.5177292823791504,
      "learning_rate": 4.996476622801987e-07,
      "loss": 0.7732,
      "step": 33745
    },
    {
      "epoch": 0.5254267632932971,
      "grad_norm": 2.90531325340271,
      "learning_rate": 4.995657232755936e-07,
      "loss": 0.7053,
      "step": 33750
    },
    {
      "epoch": 0.5255046042952665,
      "grad_norm": 3.279926061630249,
      "learning_rate": 4.994837842709886e-07,
      "loss": 0.7867,
      "step": 33755
    },
    {
      "epoch": 0.5255824452972359,
      "grad_norm": 5.175026893615723,
      "learning_rate": 4.994018452663837e-07,
      "loss": 0.6786,
      "step": 33760
    },
    {
      "epoch": 0.5256602862992052,
      "grad_norm": 5.240612506866455,
      "learning_rate": 4.993199062617787e-07,
      "loss": 0.7534,
      "step": 33765
    },
    {
      "epoch": 0.5257381273011746,
      "grad_norm": 7.01984977722168,
      "learning_rate": 4.992379672571738e-07,
      "loss": 0.7302,
      "step": 33770
    },
    {
      "epoch": 0.525815968303144,
      "grad_norm": 4.086512565612793,
      "learning_rate": 4.991560282525688e-07,
      "loss": 0.7378,
      "step": 33775
    },
    {
      "epoch": 0.5258938093051134,
      "grad_norm": 3.8922653198242188,
      "learning_rate": 4.990740892479638e-07,
      "loss": 0.7624,
      "step": 33780
    },
    {
      "epoch": 0.5259716503070827,
      "grad_norm": 4.080042362213135,
      "learning_rate": 4.989921502433588e-07,
      "loss": 0.8976,
      "step": 33785
    },
    {
      "epoch": 0.5260494913090521,
      "grad_norm": 4.862214088439941,
      "learning_rate": 4.989102112387539e-07,
      "loss": 0.6663,
      "step": 33790
    },
    {
      "epoch": 0.5261273323110215,
      "grad_norm": 4.107771396636963,
      "learning_rate": 4.988282722341489e-07,
      "loss": 0.8569,
      "step": 33795
    },
    {
      "epoch": 0.5262051733129909,
      "grad_norm": 3.7388980388641357,
      "learning_rate": 4.987463332295439e-07,
      "loss": 0.7354,
      "step": 33800
    },
    {
      "epoch": 0.5262830143149603,
      "grad_norm": 4.098170757293701,
      "learning_rate": 4.986643942249389e-07,
      "loss": 0.8525,
      "step": 33805
    },
    {
      "epoch": 0.5263608553169297,
      "grad_norm": 9.221914291381836,
      "learning_rate": 4.98582455220334e-07,
      "loss": 0.7478,
      "step": 33810
    },
    {
      "epoch": 0.5264386963188991,
      "grad_norm": 5.288509368896484,
      "learning_rate": 4.98500516215729e-07,
      "loss": 0.7935,
      "step": 33815
    },
    {
      "epoch": 0.5265165373208684,
      "grad_norm": 5.312960624694824,
      "learning_rate": 4.98418577211124e-07,
      "loss": 0.7407,
      "step": 33820
    },
    {
      "epoch": 0.5265943783228377,
      "grad_norm": 4.831115245819092,
      "learning_rate": 4.983366382065191e-07,
      "loss": 0.7304,
      "step": 33825
    },
    {
      "epoch": 0.5266722193248071,
      "grad_norm": 2.691194534301758,
      "learning_rate": 4.982546992019141e-07,
      "loss": 0.7169,
      "step": 33830
    },
    {
      "epoch": 0.5267500603267765,
      "grad_norm": 2.9801266193389893,
      "learning_rate": 4.98172760197309e-07,
      "loss": 0.7796,
      "step": 33835
    },
    {
      "epoch": 0.5268279013287459,
      "grad_norm": 4.106133460998535,
      "learning_rate": 4.980908211927041e-07,
      "loss": 0.7945,
      "step": 33840
    },
    {
      "epoch": 0.5269057423307153,
      "grad_norm": 4.447793006896973,
      "learning_rate": 4.980088821880992e-07,
      "loss": 0.8451,
      "step": 33845
    },
    {
      "epoch": 0.5269835833326847,
      "grad_norm": 4.681987285614014,
      "learning_rate": 4.979269431834942e-07,
      "loss": 0.7116,
      "step": 33850
    },
    {
      "epoch": 0.5270614243346541,
      "grad_norm": 3.0986030101776123,
      "learning_rate": 4.978450041788892e-07,
      "loss": 0.7595,
      "step": 33855
    },
    {
      "epoch": 0.5271392653366234,
      "grad_norm": 4.614822864532471,
      "learning_rate": 4.977630651742842e-07,
      "loss": 0.7379,
      "step": 33860
    },
    {
      "epoch": 0.5272171063385928,
      "grad_norm": 3.934755802154541,
      "learning_rate": 4.976811261696793e-07,
      "loss": 0.7232,
      "step": 33865
    },
    {
      "epoch": 0.5272949473405621,
      "grad_norm": 4.753609657287598,
      "learning_rate": 4.975991871650744e-07,
      "loss": 0.7349,
      "step": 33870
    },
    {
      "epoch": 0.5273727883425315,
      "grad_norm": 6.080814838409424,
      "learning_rate": 4.975172481604693e-07,
      "loss": 0.7746,
      "step": 33875
    },
    {
      "epoch": 0.5274506293445009,
      "grad_norm": 3.9943947792053223,
      "learning_rate": 4.974353091558643e-07,
      "loss": 0.8554,
      "step": 33880
    },
    {
      "epoch": 0.5275284703464703,
      "grad_norm": 3.853639602661133,
      "learning_rate": 4.973533701512594e-07,
      "loss": 0.7426,
      "step": 33885
    },
    {
      "epoch": 0.5276063113484397,
      "grad_norm": 5.21864652633667,
      "learning_rate": 4.972714311466544e-07,
      "loss": 0.6765,
      "step": 33890
    },
    {
      "epoch": 0.5276841523504091,
      "grad_norm": 3.789184093475342,
      "learning_rate": 4.971894921420494e-07,
      "loss": 0.7245,
      "step": 33895
    },
    {
      "epoch": 0.5277619933523784,
      "grad_norm": 6.010834217071533,
      "learning_rate": 4.971075531374445e-07,
      "loss": 0.8485,
      "step": 33900
    },
    {
      "epoch": 0.5278398343543478,
      "grad_norm": 7.017256259918213,
      "learning_rate": 4.970256141328395e-07,
      "loss": 0.653,
      "step": 33905
    },
    {
      "epoch": 0.5279176753563172,
      "grad_norm": 4.261816024780273,
      "learning_rate": 4.969436751282345e-07,
      "loss": 0.7694,
      "step": 33910
    },
    {
      "epoch": 0.5279955163582866,
      "grad_norm": 3.779895067214966,
      "learning_rate": 4.968617361236295e-07,
      "loss": 0.8134,
      "step": 33915
    },
    {
      "epoch": 0.528073357360256,
      "grad_norm": 3.1846213340759277,
      "learning_rate": 4.967797971190246e-07,
      "loss": 0.832,
      "step": 33920
    },
    {
      "epoch": 0.5281511983622253,
      "grad_norm": 7.841121196746826,
      "learning_rate": 4.966978581144197e-07,
      "loss": 0.8224,
      "step": 33925
    },
    {
      "epoch": 0.5282290393641947,
      "grad_norm": 3.547245740890503,
      "learning_rate": 4.966159191098146e-07,
      "loss": 0.7235,
      "step": 33930
    },
    {
      "epoch": 0.528306880366164,
      "grad_norm": 3.5102057456970215,
      "learning_rate": 4.965339801052096e-07,
      "loss": 0.6787,
      "step": 33935
    },
    {
      "epoch": 0.5283847213681334,
      "grad_norm": 4.361902713775635,
      "learning_rate": 4.964520411006047e-07,
      "loss": 0.7728,
      "step": 33940
    },
    {
      "epoch": 0.5284625623701028,
      "grad_norm": 2.7773189544677734,
      "learning_rate": 4.963701020959997e-07,
      "loss": 0.7589,
      "step": 33945
    },
    {
      "epoch": 0.5285404033720722,
      "grad_norm": 4.805288314819336,
      "learning_rate": 4.962881630913948e-07,
      "loss": 0.8062,
      "step": 33950
    },
    {
      "epoch": 0.5286182443740416,
      "grad_norm": 3.467458963394165,
      "learning_rate": 4.962062240867898e-07,
      "loss": 0.8285,
      "step": 33955
    },
    {
      "epoch": 0.528696085376011,
      "grad_norm": 6.023409843444824,
      "learning_rate": 4.961242850821848e-07,
      "loss": 0.79,
      "step": 33960
    },
    {
      "epoch": 0.5287739263779804,
      "grad_norm": 3.5086567401885986,
      "learning_rate": 4.960423460775798e-07,
      "loss": 0.6669,
      "step": 33965
    },
    {
      "epoch": 0.5288517673799498,
      "grad_norm": 4.888806343078613,
      "learning_rate": 4.959604070729749e-07,
      "loss": 0.821,
      "step": 33970
    },
    {
      "epoch": 0.528929608381919,
      "grad_norm": 4.864039897918701,
      "learning_rate": 4.958784680683699e-07,
      "loss": 0.6999,
      "step": 33975
    },
    {
      "epoch": 0.5290074493838884,
      "grad_norm": 2.9159862995147705,
      "learning_rate": 4.957965290637649e-07,
      "loss": 0.7069,
      "step": 33980
    },
    {
      "epoch": 0.5290852903858578,
      "grad_norm": 3.2412190437316895,
      "learning_rate": 4.957145900591599e-07,
      "loss": 0.7121,
      "step": 33985
    },
    {
      "epoch": 0.5291631313878272,
      "grad_norm": 3.3627359867095947,
      "learning_rate": 4.95632651054555e-07,
      "loss": 0.6786,
      "step": 33990
    },
    {
      "epoch": 0.5292409723897966,
      "grad_norm": 4.871255874633789,
      "learning_rate": 4.9555071204995e-07,
      "loss": 0.7398,
      "step": 33995
    },
    {
      "epoch": 0.529318813391766,
      "grad_norm": 6.167374610900879,
      "learning_rate": 4.95468773045345e-07,
      "loss": 0.7712,
      "step": 34000
    },
    {
      "epoch": 0.5293966543937354,
      "grad_norm": 5.2571330070495605,
      "learning_rate": 4.9538683404074e-07,
      "loss": 0.798,
      "step": 34005
    },
    {
      "epoch": 0.5294744953957047,
      "grad_norm": 3.9102399349212646,
      "learning_rate": 4.953048950361351e-07,
      "loss": 0.7668,
      "step": 34010
    },
    {
      "epoch": 0.5295523363976741,
      "grad_norm": 4.761969089508057,
      "learning_rate": 4.9522295603153e-07,
      "loss": 0.7581,
      "step": 34015
    },
    {
      "epoch": 0.5296301773996435,
      "grad_norm": 5.375120162963867,
      "learning_rate": 4.951410170269251e-07,
      "loss": 0.6857,
      "step": 34020
    },
    {
      "epoch": 0.5297080184016129,
      "grad_norm": 6.069493770599365,
      "learning_rate": 4.950590780223202e-07,
      "loss": 0.7138,
      "step": 34025
    },
    {
      "epoch": 0.5297858594035822,
      "grad_norm": 8.324358940124512,
      "learning_rate": 4.949771390177152e-07,
      "loss": 0.8482,
      "step": 34030
    },
    {
      "epoch": 0.5298637004055516,
      "grad_norm": 3.760228157043457,
      "learning_rate": 4.948952000131102e-07,
      "loss": 0.7582,
      "step": 34035
    },
    {
      "epoch": 0.529941541407521,
      "grad_norm": 6.663865566253662,
      "learning_rate": 4.948132610085052e-07,
      "loss": 0.6631,
      "step": 34040
    },
    {
      "epoch": 0.5300193824094904,
      "grad_norm": 2.7659616470336914,
      "learning_rate": 4.947313220039003e-07,
      "loss": 0.6816,
      "step": 34045
    },
    {
      "epoch": 0.5300972234114597,
      "grad_norm": 3.1409595012664795,
      "learning_rate": 4.946493829992954e-07,
      "loss": 0.8048,
      "step": 34050
    },
    {
      "epoch": 0.5301750644134291,
      "grad_norm": 2.9819869995117188,
      "learning_rate": 4.945674439946903e-07,
      "loss": 0.7349,
      "step": 34055
    },
    {
      "epoch": 0.5302529054153985,
      "grad_norm": 2.6420555114746094,
      "learning_rate": 4.944855049900853e-07,
      "loss": 0.7039,
      "step": 34060
    },
    {
      "epoch": 0.5303307464173679,
      "grad_norm": 3.09094500541687,
      "learning_rate": 4.944035659854804e-07,
      "loss": 0.6961,
      "step": 34065
    },
    {
      "epoch": 0.5304085874193373,
      "grad_norm": 3.799021005630493,
      "learning_rate": 4.943216269808754e-07,
      "loss": 0.7014,
      "step": 34070
    },
    {
      "epoch": 0.5304864284213067,
      "grad_norm": 3.0272328853607178,
      "learning_rate": 4.942396879762705e-07,
      "loss": 0.7994,
      "step": 34075
    },
    {
      "epoch": 0.5305642694232761,
      "grad_norm": 4.687512397766113,
      "learning_rate": 4.941577489716655e-07,
      "loss": 0.7953,
      "step": 34080
    },
    {
      "epoch": 0.5306421104252453,
      "grad_norm": 5.1567487716674805,
      "learning_rate": 4.940758099670605e-07,
      "loss": 0.7026,
      "step": 34085
    },
    {
      "epoch": 0.5307199514272147,
      "grad_norm": 2.813913106918335,
      "learning_rate": 4.939938709624555e-07,
      "loss": 0.6821,
      "step": 34090
    },
    {
      "epoch": 0.5307977924291841,
      "grad_norm": 4.016183853149414,
      "learning_rate": 4.939119319578506e-07,
      "loss": 0.8151,
      "step": 34095
    },
    {
      "epoch": 0.5308756334311535,
      "grad_norm": 3.9302196502685547,
      "learning_rate": 4.938299929532456e-07,
      "loss": 0.7694,
      "step": 34100
    },
    {
      "epoch": 0.5309534744331229,
      "grad_norm": 3.738638162612915,
      "learning_rate": 4.937480539486407e-07,
      "loss": 0.8491,
      "step": 34105
    },
    {
      "epoch": 0.5310313154350923,
      "grad_norm": 4.642480373382568,
      "learning_rate": 4.936661149440356e-07,
      "loss": 0.8398,
      "step": 34110
    },
    {
      "epoch": 0.5311091564370617,
      "grad_norm": 3.074406385421753,
      "learning_rate": 4.935841759394307e-07,
      "loss": 0.7478,
      "step": 34115
    },
    {
      "epoch": 0.5311869974390311,
      "grad_norm": 3.739227056503296,
      "learning_rate": 4.935022369348257e-07,
      "loss": 0.7715,
      "step": 34120
    },
    {
      "epoch": 0.5312648384410004,
      "grad_norm": 3.512458324432373,
      "learning_rate": 4.934202979302207e-07,
      "loss": 0.8115,
      "step": 34125
    },
    {
      "epoch": 0.5313426794429698,
      "grad_norm": 3.637192726135254,
      "learning_rate": 4.933383589256157e-07,
      "loss": 0.7619,
      "step": 34130
    },
    {
      "epoch": 0.5314205204449391,
      "grad_norm": 4.556889057159424,
      "learning_rate": 4.932564199210108e-07,
      "loss": 0.8131,
      "step": 34135
    },
    {
      "epoch": 0.5314983614469085,
      "grad_norm": 4.971821308135986,
      "learning_rate": 4.931744809164058e-07,
      "loss": 0.6751,
      "step": 34140
    },
    {
      "epoch": 0.5315762024488779,
      "grad_norm": 4.22316312789917,
      "learning_rate": 4.930925419118008e-07,
      "loss": 0.7935,
      "step": 34145
    },
    {
      "epoch": 0.5316540434508473,
      "grad_norm": 4.1751627922058105,
      "learning_rate": 4.930106029071959e-07,
      "loss": 0.7134,
      "step": 34150
    },
    {
      "epoch": 0.5317318844528167,
      "grad_norm": 2.920233964920044,
      "learning_rate": 4.929286639025909e-07,
      "loss": 0.6879,
      "step": 34155
    },
    {
      "epoch": 0.5318097254547861,
      "grad_norm": 3.4221553802490234,
      "learning_rate": 4.928467248979859e-07,
      "loss": 0.6974,
      "step": 34160
    },
    {
      "epoch": 0.5318875664567554,
      "grad_norm": 3.2680606842041016,
      "learning_rate": 4.927647858933809e-07,
      "loss": 0.7553,
      "step": 34165
    },
    {
      "epoch": 0.5319654074587248,
      "grad_norm": 3.437628984451294,
      "learning_rate": 4.92682846888776e-07,
      "loss": 0.8244,
      "step": 34170
    },
    {
      "epoch": 0.5320432484606942,
      "grad_norm": 4.407691955566406,
      "learning_rate": 4.926009078841711e-07,
      "loss": 0.7666,
      "step": 34175
    },
    {
      "epoch": 0.5321210894626636,
      "grad_norm": 2.402296304702759,
      "learning_rate": 4.92518968879566e-07,
      "loss": 0.7088,
      "step": 34180
    },
    {
      "epoch": 0.532198930464633,
      "grad_norm": 4.2170538902282715,
      "learning_rate": 4.92437029874961e-07,
      "loss": 0.8715,
      "step": 34185
    },
    {
      "epoch": 0.5322767714666023,
      "grad_norm": 3.680683135986328,
      "learning_rate": 4.923550908703561e-07,
      "loss": 0.7471,
      "step": 34190
    },
    {
      "epoch": 0.5323546124685717,
      "grad_norm": 2.5467164516448975,
      "learning_rate": 4.922731518657512e-07,
      "loss": 0.6504,
      "step": 34195
    },
    {
      "epoch": 0.532432453470541,
      "grad_norm": 3.2730138301849365,
      "learning_rate": 4.921912128611461e-07,
      "loss": 0.6793,
      "step": 34200
    },
    {
      "epoch": 0.5325102944725104,
      "grad_norm": 5.379147529602051,
      "learning_rate": 4.921092738565412e-07,
      "loss": 0.7836,
      "step": 34205
    },
    {
      "epoch": 0.5325881354744798,
      "grad_norm": 4.6332502365112305,
      "learning_rate": 4.920273348519362e-07,
      "loss": 0.7858,
      "step": 34210
    },
    {
      "epoch": 0.5326659764764492,
      "grad_norm": 3.1085944175720215,
      "learning_rate": 4.919453958473312e-07,
      "loss": 0.699,
      "step": 34215
    },
    {
      "epoch": 0.5327438174784186,
      "grad_norm": 3.467528820037842,
      "learning_rate": 4.918634568427262e-07,
      "loss": 0.7328,
      "step": 34220
    },
    {
      "epoch": 0.532821658480388,
      "grad_norm": 2.9465508460998535,
      "learning_rate": 4.917815178381213e-07,
      "loss": 0.6162,
      "step": 34225
    },
    {
      "epoch": 0.5328994994823574,
      "grad_norm": 16.964603424072266,
      "learning_rate": 4.916995788335164e-07,
      "loss": 0.8362,
      "step": 34230
    },
    {
      "epoch": 0.5329773404843268,
      "grad_norm": 6.224708557128906,
      "learning_rate": 4.916176398289113e-07,
      "loss": 0.7757,
      "step": 34235
    },
    {
      "epoch": 0.533055181486296,
      "grad_norm": 4.03013801574707,
      "learning_rate": 4.915357008243063e-07,
      "loss": 0.735,
      "step": 34240
    },
    {
      "epoch": 0.5331330224882654,
      "grad_norm": 6.9679179191589355,
      "learning_rate": 4.914537618197014e-07,
      "loss": 0.7561,
      "step": 34245
    },
    {
      "epoch": 0.5332108634902348,
      "grad_norm": 6.83951997756958,
      "learning_rate": 4.913718228150964e-07,
      "loss": 0.7246,
      "step": 34250
    },
    {
      "epoch": 0.5332887044922042,
      "grad_norm": 3.1431663036346436,
      "learning_rate": 4.912898838104914e-07,
      "loss": 0.7262,
      "step": 34255
    },
    {
      "epoch": 0.5333665454941736,
      "grad_norm": 8.569406509399414,
      "learning_rate": 4.912079448058865e-07,
      "loss": 0.7619,
      "step": 34260
    },
    {
      "epoch": 0.533444386496143,
      "grad_norm": 4.437221050262451,
      "learning_rate": 4.911260058012815e-07,
      "loss": 0.7373,
      "step": 34265
    },
    {
      "epoch": 0.5335222274981124,
      "grad_norm": 5.242059230804443,
      "learning_rate": 4.910440667966765e-07,
      "loss": 0.8206,
      "step": 34270
    },
    {
      "epoch": 0.5336000685000817,
      "grad_norm": 4.506338119506836,
      "learning_rate": 4.909621277920716e-07,
      "loss": 0.7691,
      "step": 34275
    },
    {
      "epoch": 0.5336779095020511,
      "grad_norm": 6.6809844970703125,
      "learning_rate": 4.908801887874666e-07,
      "loss": 0.7561,
      "step": 34280
    },
    {
      "epoch": 0.5337557505040205,
      "grad_norm": 4.258233070373535,
      "learning_rate": 4.907982497828617e-07,
      "loss": 0.8176,
      "step": 34285
    },
    {
      "epoch": 0.5338335915059899,
      "grad_norm": 3.1951324939727783,
      "learning_rate": 4.907163107782566e-07,
      "loss": 0.7515,
      "step": 34290
    },
    {
      "epoch": 0.5339114325079592,
      "grad_norm": 2.7342288494110107,
      "learning_rate": 4.906343717736517e-07,
      "loss": 0.6938,
      "step": 34295
    },
    {
      "epoch": 0.5339892735099286,
      "grad_norm": 5.18389368057251,
      "learning_rate": 4.905524327690467e-07,
      "loss": 0.7009,
      "step": 34300
    },
    {
      "epoch": 0.534067114511898,
      "grad_norm": 5.2681565284729,
      "learning_rate": 4.904704937644417e-07,
      "loss": 0.7586,
      "step": 34305
    },
    {
      "epoch": 0.5341449555138674,
      "grad_norm": 3.921173095703125,
      "learning_rate": 4.903885547598367e-07,
      "loss": 0.8117,
      "step": 34310
    },
    {
      "epoch": 0.5342227965158367,
      "grad_norm": 3.5203590393066406,
      "learning_rate": 4.903066157552318e-07,
      "loss": 0.7295,
      "step": 34315
    },
    {
      "epoch": 0.5343006375178061,
      "grad_norm": 2.796928644180298,
      "learning_rate": 4.902246767506268e-07,
      "loss": 0.71,
      "step": 34320
    },
    {
      "epoch": 0.5343784785197755,
      "grad_norm": 6.77574348449707,
      "learning_rate": 4.901427377460218e-07,
      "loss": 0.6791,
      "step": 34325
    },
    {
      "epoch": 0.5344563195217449,
      "grad_norm": 4.604251384735107,
      "learning_rate": 4.900607987414169e-07,
      "loss": 0.7152,
      "step": 34330
    },
    {
      "epoch": 0.5345341605237143,
      "grad_norm": 6.5264787673950195,
      "learning_rate": 4.899788597368119e-07,
      "loss": 0.8005,
      "step": 34335
    },
    {
      "epoch": 0.5346120015256837,
      "grad_norm": 5.777383804321289,
      "learning_rate": 4.89896920732207e-07,
      "loss": 0.6615,
      "step": 34340
    },
    {
      "epoch": 0.5346898425276531,
      "grad_norm": 2.7263336181640625,
      "learning_rate": 4.898149817276019e-07,
      "loss": 0.7404,
      "step": 34345
    },
    {
      "epoch": 0.5347676835296223,
      "grad_norm": 2.732365131378174,
      "learning_rate": 4.89733042722997e-07,
      "loss": 0.794,
      "step": 34350
    },
    {
      "epoch": 0.5348455245315917,
      "grad_norm": 3.9584360122680664,
      "learning_rate": 4.896511037183921e-07,
      "loss": 0.8357,
      "step": 34355
    },
    {
      "epoch": 0.5349233655335611,
      "grad_norm": 2.7650790214538574,
      "learning_rate": 4.89569164713787e-07,
      "loss": 0.7242,
      "step": 34360
    },
    {
      "epoch": 0.5350012065355305,
      "grad_norm": 8.609464645385742,
      "learning_rate": 4.89487225709182e-07,
      "loss": 0.8164,
      "step": 34365
    },
    {
      "epoch": 0.5350790475374999,
      "grad_norm": 5.93986177444458,
      "learning_rate": 4.894052867045771e-07,
      "loss": 0.6974,
      "step": 34370
    },
    {
      "epoch": 0.5351568885394693,
      "grad_norm": 5.969693660736084,
      "learning_rate": 4.893233476999722e-07,
      "loss": 0.8275,
      "step": 34375
    },
    {
      "epoch": 0.5352347295414387,
      "grad_norm": 4.8887128829956055,
      "learning_rate": 4.892414086953671e-07,
      "loss": 0.7736,
      "step": 34380
    },
    {
      "epoch": 0.5353125705434081,
      "grad_norm": 4.7833733558654785,
      "learning_rate": 4.891594696907622e-07,
      "loss": 0.7243,
      "step": 34385
    },
    {
      "epoch": 0.5353904115453774,
      "grad_norm": 2.9387848377227783,
      "learning_rate": 4.890775306861572e-07,
      "loss": 0.7368,
      "step": 34390
    },
    {
      "epoch": 0.5354682525473468,
      "grad_norm": 3.597167491912842,
      "learning_rate": 4.889955916815522e-07,
      "loss": 0.7846,
      "step": 34395
    },
    {
      "epoch": 0.5355460935493161,
      "grad_norm": 4.322056293487549,
      "learning_rate": 4.889136526769473e-07,
      "loss": 0.7295,
      "step": 34400
    },
    {
      "epoch": 0.5356239345512855,
      "grad_norm": 4.98111629486084,
      "learning_rate": 4.888317136723423e-07,
      "loss": 0.8081,
      "step": 34405
    },
    {
      "epoch": 0.5357017755532549,
      "grad_norm": 4.356158256530762,
      "learning_rate": 4.887497746677374e-07,
      "loss": 0.6587,
      "step": 34410
    },
    {
      "epoch": 0.5357796165552243,
      "grad_norm": 3.6506540775299072,
      "learning_rate": 4.886678356631323e-07,
      "loss": 0.7822,
      "step": 34415
    },
    {
      "epoch": 0.5358574575571937,
      "grad_norm": 6.972822666168213,
      "learning_rate": 4.885858966585274e-07,
      "loss": 0.88,
      "step": 34420
    },
    {
      "epoch": 0.535935298559163,
      "grad_norm": 3.9018442630767822,
      "learning_rate": 4.885039576539224e-07,
      "loss": 0.7604,
      "step": 34425
    },
    {
      "epoch": 0.5360131395611324,
      "grad_norm": 3.2524847984313965,
      "learning_rate": 4.884220186493174e-07,
      "loss": 0.6683,
      "step": 34430
    },
    {
      "epoch": 0.5360909805631018,
      "grad_norm": 3.507159471511841,
      "learning_rate": 4.883400796447124e-07,
      "loss": 0.7452,
      "step": 34435
    },
    {
      "epoch": 0.5361688215650712,
      "grad_norm": 3.2452006340026855,
      "learning_rate": 4.882581406401075e-07,
      "loss": 0.7997,
      "step": 34440
    },
    {
      "epoch": 0.5362466625670406,
      "grad_norm": 3.787081003189087,
      "learning_rate": 4.881762016355025e-07,
      "loss": 0.76,
      "step": 34445
    },
    {
      "epoch": 0.53632450356901,
      "grad_norm": 4.333800315856934,
      "learning_rate": 4.880942626308975e-07,
      "loss": 0.7125,
      "step": 34450
    },
    {
      "epoch": 0.5364023445709794,
      "grad_norm": 3.9809696674346924,
      "learning_rate": 4.880123236262926e-07,
      "loss": 0.6139,
      "step": 34455
    },
    {
      "epoch": 0.5364801855729487,
      "grad_norm": 5.141919136047363,
      "learning_rate": 4.879303846216876e-07,
      "loss": 0.768,
      "step": 34460
    },
    {
      "epoch": 0.536558026574918,
      "grad_norm": 4.822776794433594,
      "learning_rate": 4.878484456170827e-07,
      "loss": 0.7242,
      "step": 34465
    },
    {
      "epoch": 0.5366358675768874,
      "grad_norm": 5.4447245597839355,
      "learning_rate": 4.877665066124776e-07,
      "loss": 0.8643,
      "step": 34470
    },
    {
      "epoch": 0.5367137085788568,
      "grad_norm": 4.174193859100342,
      "learning_rate": 4.876845676078727e-07,
      "loss": 0.7275,
      "step": 34475
    },
    {
      "epoch": 0.5367915495808262,
      "grad_norm": 7.328827857971191,
      "learning_rate": 4.876026286032678e-07,
      "loss": 0.7615,
      "step": 34480
    },
    {
      "epoch": 0.5368693905827956,
      "grad_norm": 3.143784999847412,
      "learning_rate": 4.875206895986627e-07,
      "loss": 0.7194,
      "step": 34485
    },
    {
      "epoch": 0.536947231584765,
      "grad_norm": 3.402531623840332,
      "learning_rate": 4.874387505940577e-07,
      "loss": 0.6873,
      "step": 34490
    },
    {
      "epoch": 0.5370250725867344,
      "grad_norm": 2.9436774253845215,
      "learning_rate": 4.873568115894528e-07,
      "loss": 0.758,
      "step": 34495
    },
    {
      "epoch": 0.5371029135887038,
      "grad_norm": 4.677123546600342,
      "learning_rate": 4.872748725848479e-07,
      "loss": 0.7642,
      "step": 34500
    },
    {
      "epoch": 0.537180754590673,
      "grad_norm": 3.827331066131592,
      "learning_rate": 4.871929335802428e-07,
      "loss": 0.6658,
      "step": 34505
    },
    {
      "epoch": 0.5372585955926424,
      "grad_norm": 2.3742899894714355,
      "learning_rate": 4.871109945756379e-07,
      "loss": 0.7458,
      "step": 34510
    },
    {
      "epoch": 0.5373364365946118,
      "grad_norm": 3.611633062362671,
      "learning_rate": 4.870290555710329e-07,
      "loss": 0.6413,
      "step": 34515
    },
    {
      "epoch": 0.5374142775965812,
      "grad_norm": 3.2659499645233154,
      "learning_rate": 4.86947116566428e-07,
      "loss": 0.7661,
      "step": 34520
    },
    {
      "epoch": 0.5374921185985506,
      "grad_norm": 3.1166858673095703,
      "learning_rate": 4.868651775618229e-07,
      "loss": 0.8053,
      "step": 34525
    },
    {
      "epoch": 0.53756995960052,
      "grad_norm": 2.5626490116119385,
      "learning_rate": 4.86783238557218e-07,
      "loss": 0.7971,
      "step": 34530
    },
    {
      "epoch": 0.5376478006024894,
      "grad_norm": 3.9417428970336914,
      "learning_rate": 4.867012995526131e-07,
      "loss": 0.7712,
      "step": 34535
    },
    {
      "epoch": 0.5377256416044587,
      "grad_norm": 3.6905341148376465,
      "learning_rate": 4.86619360548008e-07,
      "loss": 0.6981,
      "step": 34540
    },
    {
      "epoch": 0.5378034826064281,
      "grad_norm": 3.240710496902466,
      "learning_rate": 4.86537421543403e-07,
      "loss": 0.7102,
      "step": 34545
    },
    {
      "epoch": 0.5378813236083975,
      "grad_norm": 3.6515583992004395,
      "learning_rate": 4.864554825387981e-07,
      "loss": 0.6909,
      "step": 34550
    },
    {
      "epoch": 0.5379591646103669,
      "grad_norm": 5.71281623840332,
      "learning_rate": 4.863735435341932e-07,
      "loss": 0.7604,
      "step": 34555
    },
    {
      "epoch": 0.5380370056123362,
      "grad_norm": 4.350783824920654,
      "learning_rate": 4.862916045295881e-07,
      "loss": 0.8111,
      "step": 34560
    },
    {
      "epoch": 0.5381148466143056,
      "grad_norm": 6.9427947998046875,
      "learning_rate": 4.862096655249832e-07,
      "loss": 0.7028,
      "step": 34565
    },
    {
      "epoch": 0.538192687616275,
      "grad_norm": 3.016634225845337,
      "learning_rate": 4.861277265203782e-07,
      "loss": 0.7372,
      "step": 34570
    },
    {
      "epoch": 0.5382705286182444,
      "grad_norm": 3.9433889389038086,
      "learning_rate": 4.860457875157732e-07,
      "loss": 0.7026,
      "step": 34575
    },
    {
      "epoch": 0.5383483696202137,
      "grad_norm": 4.612628936767578,
      "learning_rate": 4.859638485111683e-07,
      "loss": 0.7536,
      "step": 34580
    },
    {
      "epoch": 0.5384262106221831,
      "grad_norm": 3.095318078994751,
      "learning_rate": 4.858819095065633e-07,
      "loss": 0.6275,
      "step": 34585
    },
    {
      "epoch": 0.5385040516241525,
      "grad_norm": 3.604900598526001,
      "learning_rate": 4.857999705019584e-07,
      "loss": 0.7881,
      "step": 34590
    },
    {
      "epoch": 0.5385818926261219,
      "grad_norm": 2.7042195796966553,
      "learning_rate": 4.857180314973533e-07,
      "loss": 0.7259,
      "step": 34595
    },
    {
      "epoch": 0.5386597336280913,
      "grad_norm": 5.0977277755737305,
      "learning_rate": 4.856360924927484e-07,
      "loss": 0.819,
      "step": 34600
    },
    {
      "epoch": 0.5387375746300607,
      "grad_norm": 8.040875434875488,
      "learning_rate": 4.855541534881434e-07,
      "loss": 0.6994,
      "step": 34605
    },
    {
      "epoch": 0.5388154156320301,
      "grad_norm": 5.764067649841309,
      "learning_rate": 4.854722144835384e-07,
      "loss": 0.8276,
      "step": 34610
    },
    {
      "epoch": 0.5388932566339993,
      "grad_norm": 5.121054172515869,
      "learning_rate": 4.853902754789334e-07,
      "loss": 0.8116,
      "step": 34615
    },
    {
      "epoch": 0.5389710976359687,
      "grad_norm": 3.794579029083252,
      "learning_rate": 4.853083364743285e-07,
      "loss": 0.7652,
      "step": 34620
    },
    {
      "epoch": 0.5390489386379381,
      "grad_norm": 3.6065151691436768,
      "learning_rate": 4.852263974697235e-07,
      "loss": 0.776,
      "step": 34625
    },
    {
      "epoch": 0.5391267796399075,
      "grad_norm": 6.5509514808654785,
      "learning_rate": 4.851444584651185e-07,
      "loss": 0.7452,
      "step": 34630
    },
    {
      "epoch": 0.5392046206418769,
      "grad_norm": 4.013660907745361,
      "learning_rate": 4.850625194605136e-07,
      "loss": 0.7182,
      "step": 34635
    },
    {
      "epoch": 0.5392824616438463,
      "grad_norm": 4.391438961029053,
      "learning_rate": 4.849805804559086e-07,
      "loss": 0.7149,
      "step": 34640
    },
    {
      "epoch": 0.5393603026458157,
      "grad_norm": 3.6462409496307373,
      "learning_rate": 4.848986414513037e-07,
      "loss": 0.7017,
      "step": 34645
    },
    {
      "epoch": 0.5394381436477851,
      "grad_norm": 4.66319465637207,
      "learning_rate": 4.848167024466986e-07,
      "loss": 0.6852,
      "step": 34650
    },
    {
      "epoch": 0.5395159846497544,
      "grad_norm": 3.302386522293091,
      "learning_rate": 4.847347634420937e-07,
      "loss": 0.7035,
      "step": 34655
    },
    {
      "epoch": 0.5395938256517238,
      "grad_norm": 3.8576533794403076,
      "learning_rate": 4.846528244374888e-07,
      "loss": 0.759,
      "step": 34660
    },
    {
      "epoch": 0.5396716666536931,
      "grad_norm": 2.8729946613311768,
      "learning_rate": 4.845708854328837e-07,
      "loss": 0.8026,
      "step": 34665
    },
    {
      "epoch": 0.5397495076556625,
      "grad_norm": 3.486466407775879,
      "learning_rate": 4.844889464282787e-07,
      "loss": 0.7162,
      "step": 34670
    },
    {
      "epoch": 0.5398273486576319,
      "grad_norm": 5.393845558166504,
      "learning_rate": 4.844070074236738e-07,
      "loss": 0.7111,
      "step": 34675
    },
    {
      "epoch": 0.5399051896596013,
      "grad_norm": 5.521489143371582,
      "learning_rate": 4.843250684190689e-07,
      "loss": 0.7996,
      "step": 34680
    },
    {
      "epoch": 0.5399830306615707,
      "grad_norm": 5.107725620269775,
      "learning_rate": 4.842431294144638e-07,
      "loss": 0.7519,
      "step": 34685
    },
    {
      "epoch": 0.54006087166354,
      "grad_norm": 3.3498599529266357,
      "learning_rate": 4.841611904098589e-07,
      "loss": 0.7244,
      "step": 34690
    },
    {
      "epoch": 0.5401387126655094,
      "grad_norm": 4.134640216827393,
      "learning_rate": 4.840792514052539e-07,
      "loss": 0.6943,
      "step": 34695
    },
    {
      "epoch": 0.5402165536674788,
      "grad_norm": 4.609651565551758,
      "learning_rate": 4.83997312400649e-07,
      "loss": 0.6376,
      "step": 34700
    },
    {
      "epoch": 0.5402943946694482,
      "grad_norm": 3.462824583053589,
      "learning_rate": 4.839153733960439e-07,
      "loss": 0.7674,
      "step": 34705
    },
    {
      "epoch": 0.5403722356714176,
      "grad_norm": 7.129645347595215,
      "learning_rate": 4.83833434391439e-07,
      "loss": 0.8454,
      "step": 34710
    },
    {
      "epoch": 0.540450076673387,
      "grad_norm": 6.26192569732666,
      "learning_rate": 4.837514953868341e-07,
      "loss": 0.7327,
      "step": 34715
    },
    {
      "epoch": 0.5405279176753564,
      "grad_norm": 4.145537853240967,
      "learning_rate": 4.83669556382229e-07,
      "loss": 0.667,
      "step": 34720
    },
    {
      "epoch": 0.5406057586773257,
      "grad_norm": 5.663415908813477,
      "learning_rate": 4.835876173776241e-07,
      "loss": 0.6684,
      "step": 34725
    },
    {
      "epoch": 0.540683599679295,
      "grad_norm": 3.3992977142333984,
      "learning_rate": 4.835056783730191e-07,
      "loss": 0.655,
      "step": 34730
    },
    {
      "epoch": 0.5407614406812644,
      "grad_norm": 4.434875965118408,
      "learning_rate": 4.834237393684142e-07,
      "loss": 0.7699,
      "step": 34735
    },
    {
      "epoch": 0.5408392816832338,
      "grad_norm": 4.936732769012451,
      "learning_rate": 4.833418003638091e-07,
      "loss": 0.7026,
      "step": 34740
    },
    {
      "epoch": 0.5409171226852032,
      "grad_norm": 3.4522392749786377,
      "learning_rate": 4.832598613592042e-07,
      "loss": 0.7315,
      "step": 34745
    },
    {
      "epoch": 0.5409949636871726,
      "grad_norm": 3.5100135803222656,
      "learning_rate": 4.831779223545992e-07,
      "loss": 0.8253,
      "step": 34750
    },
    {
      "epoch": 0.541072804689142,
      "grad_norm": 3.2848777770996094,
      "learning_rate": 4.830959833499942e-07,
      "loss": 0.828,
      "step": 34755
    },
    {
      "epoch": 0.5411506456911114,
      "grad_norm": 2.660513162612915,
      "learning_rate": 4.830140443453893e-07,
      "loss": 0.6735,
      "step": 34760
    },
    {
      "epoch": 0.5412284866930808,
      "grad_norm": 4.688377857208252,
      "learning_rate": 4.829321053407843e-07,
      "loss": 0.845,
      "step": 34765
    },
    {
      "epoch": 0.54130632769505,
      "grad_norm": 6.530161380767822,
      "learning_rate": 4.828501663361794e-07,
      "loss": 0.7827,
      "step": 34770
    },
    {
      "epoch": 0.5413841686970194,
      "grad_norm": 5.069417953491211,
      "learning_rate": 4.827682273315743e-07,
      "loss": 0.7722,
      "step": 34775
    },
    {
      "epoch": 0.5414620096989888,
      "grad_norm": 3.3508880138397217,
      "learning_rate": 4.826862883269694e-07,
      "loss": 0.6536,
      "step": 34780
    },
    {
      "epoch": 0.5415398507009582,
      "grad_norm": 2.8882598876953125,
      "learning_rate": 4.826043493223645e-07,
      "loss": 0.7712,
      "step": 34785
    },
    {
      "epoch": 0.5416176917029276,
      "grad_norm": 6.926908493041992,
      "learning_rate": 4.825224103177594e-07,
      "loss": 0.754,
      "step": 34790
    },
    {
      "epoch": 0.541695532704897,
      "grad_norm": 4.031869411468506,
      "learning_rate": 4.824404713131544e-07,
      "loss": 0.7298,
      "step": 34795
    },
    {
      "epoch": 0.5417733737068664,
      "grad_norm": 3.7199978828430176,
      "learning_rate": 4.823585323085495e-07,
      "loss": 0.6959,
      "step": 34800
    },
    {
      "epoch": 0.5418512147088357,
      "grad_norm": 4.260602951049805,
      "learning_rate": 4.822765933039446e-07,
      "loss": 0.6581,
      "step": 34805
    },
    {
      "epoch": 0.5419290557108051,
      "grad_norm": 4.81985330581665,
      "learning_rate": 4.821946542993395e-07,
      "loss": 0.7457,
      "step": 34810
    },
    {
      "epoch": 0.5420068967127745,
      "grad_norm": 3.6977930068969727,
      "learning_rate": 4.821127152947346e-07,
      "loss": 0.6823,
      "step": 34815
    },
    {
      "epoch": 0.5420847377147439,
      "grad_norm": 7.567504405975342,
      "learning_rate": 4.820307762901296e-07,
      "loss": 0.7397,
      "step": 34820
    },
    {
      "epoch": 0.5421625787167133,
      "grad_norm": 4.73651647567749,
      "learning_rate": 4.819488372855247e-07,
      "loss": 0.733,
      "step": 34825
    },
    {
      "epoch": 0.5422404197186826,
      "grad_norm": 2.929792881011963,
      "learning_rate": 4.818668982809196e-07,
      "loss": 0.6268,
      "step": 34830
    },
    {
      "epoch": 0.542318260720652,
      "grad_norm": 3.2962162494659424,
      "learning_rate": 4.817849592763147e-07,
      "loss": 0.6958,
      "step": 34835
    },
    {
      "epoch": 0.5423961017226214,
      "grad_norm": 4.669434547424316,
      "learning_rate": 4.817030202717098e-07,
      "loss": 0.675,
      "step": 34840
    },
    {
      "epoch": 0.5424739427245907,
      "grad_norm": 3.5002195835113525,
      "learning_rate": 4.816210812671047e-07,
      "loss": 0.7402,
      "step": 34845
    },
    {
      "epoch": 0.5425517837265601,
      "grad_norm": 3.305626392364502,
      "learning_rate": 4.815391422624997e-07,
      "loss": 0.7739,
      "step": 34850
    },
    {
      "epoch": 0.5426296247285295,
      "grad_norm": 3.16043758392334,
      "learning_rate": 4.814572032578948e-07,
      "loss": 0.8376,
      "step": 34855
    },
    {
      "epoch": 0.5427074657304989,
      "grad_norm": 3.852417469024658,
      "learning_rate": 4.813752642532899e-07,
      "loss": 0.8046,
      "step": 34860
    },
    {
      "epoch": 0.5427853067324683,
      "grad_norm": 5.508766174316406,
      "learning_rate": 4.812933252486848e-07,
      "loss": 0.7659,
      "step": 34865
    },
    {
      "epoch": 0.5428631477344377,
      "grad_norm": 3.9380881786346436,
      "learning_rate": 4.812113862440799e-07,
      "loss": 0.842,
      "step": 34870
    },
    {
      "epoch": 0.5429409887364071,
      "grad_norm": 3.6409707069396973,
      "learning_rate": 4.811294472394749e-07,
      "loss": 0.7605,
      "step": 34875
    },
    {
      "epoch": 0.5430188297383763,
      "grad_norm": 9.14228343963623,
      "learning_rate": 4.8104750823487e-07,
      "loss": 0.827,
      "step": 34880
    },
    {
      "epoch": 0.5430966707403457,
      "grad_norm": 4.955946445465088,
      "learning_rate": 4.80965569230265e-07,
      "loss": 0.7739,
      "step": 34885
    },
    {
      "epoch": 0.5431745117423151,
      "grad_norm": 3.202409029006958,
      "learning_rate": 4.8088363022566e-07,
      "loss": 0.682,
      "step": 34890
    },
    {
      "epoch": 0.5432523527442845,
      "grad_norm": 3.1761975288391113,
      "learning_rate": 4.808016912210551e-07,
      "loss": 0.7217,
      "step": 34895
    },
    {
      "epoch": 0.5433301937462539,
      "grad_norm": 3.448140859603882,
      "learning_rate": 4.8071975221645e-07,
      "loss": 0.6654,
      "step": 34900
    },
    {
      "epoch": 0.5434080347482233,
      "grad_norm": 2.804405450820923,
      "learning_rate": 4.806378132118451e-07,
      "loss": 0.7084,
      "step": 34905
    },
    {
      "epoch": 0.5434858757501927,
      "grad_norm": 4.690401077270508,
      "learning_rate": 4.805558742072401e-07,
      "loss": 0.8571,
      "step": 34910
    },
    {
      "epoch": 0.5435637167521621,
      "grad_norm": 7.2704691886901855,
      "learning_rate": 4.804739352026352e-07,
      "loss": 0.7468,
      "step": 34915
    },
    {
      "epoch": 0.5436415577541314,
      "grad_norm": 9.074907302856445,
      "learning_rate": 4.803919961980301e-07,
      "loss": 0.705,
      "step": 34920
    },
    {
      "epoch": 0.5437193987561008,
      "grad_norm": 5.229625225067139,
      "learning_rate": 4.803100571934252e-07,
      "loss": 0.7919,
      "step": 34925
    },
    {
      "epoch": 0.5437972397580702,
      "grad_norm": 7.028835773468018,
      "learning_rate": 4.802281181888202e-07,
      "loss": 0.7022,
      "step": 34930
    },
    {
      "epoch": 0.5438750807600395,
      "grad_norm": 2.235790491104126,
      "learning_rate": 4.801461791842152e-07,
      "loss": 0.7718,
      "step": 34935
    },
    {
      "epoch": 0.5439529217620089,
      "grad_norm": 4.069645404815674,
      "learning_rate": 4.800642401796103e-07,
      "loss": 0.7645,
      "step": 34940
    },
    {
      "epoch": 0.5440307627639783,
      "grad_norm": 3.056535482406616,
      "learning_rate": 4.799823011750053e-07,
      "loss": 0.8369,
      "step": 34945
    },
    {
      "epoch": 0.5441086037659477,
      "grad_norm": 2.7560112476348877,
      "learning_rate": 4.799003621704003e-07,
      "loss": 0.6901,
      "step": 34950
    },
    {
      "epoch": 0.544186444767917,
      "grad_norm": 4.909725189208984,
      "learning_rate": 4.798184231657953e-07,
      "loss": 0.7572,
      "step": 34955
    },
    {
      "epoch": 0.5442642857698864,
      "grad_norm": 6.896103858947754,
      "learning_rate": 4.797364841611904e-07,
      "loss": 0.7185,
      "step": 34960
    },
    {
      "epoch": 0.5443421267718558,
      "grad_norm": 4.164396286010742,
      "learning_rate": 4.796545451565855e-07,
      "loss": 0.8686,
      "step": 34965
    },
    {
      "epoch": 0.5444199677738252,
      "grad_norm": 3.8166489601135254,
      "learning_rate": 4.795726061519805e-07,
      "loss": 0.778,
      "step": 34970
    },
    {
      "epoch": 0.5444978087757946,
      "grad_norm": 2.6334176063537598,
      "learning_rate": 4.794906671473754e-07,
      "loss": 0.626,
      "step": 34975
    },
    {
      "epoch": 0.544575649777764,
      "grad_norm": 3.179680347442627,
      "learning_rate": 4.794087281427705e-07,
      "loss": 0.6826,
      "step": 34980
    },
    {
      "epoch": 0.5446534907797334,
      "grad_norm": 5.611238479614258,
      "learning_rate": 4.793267891381656e-07,
      "loss": 0.8071,
      "step": 34985
    },
    {
      "epoch": 0.5447313317817027,
      "grad_norm": 4.730948448181152,
      "learning_rate": 4.792448501335605e-07,
      "loss": 0.8698,
      "step": 34990
    },
    {
      "epoch": 0.544809172783672,
      "grad_norm": 4.168317794799805,
      "learning_rate": 4.791629111289556e-07,
      "loss": 0.7156,
      "step": 34995
    },
    {
      "epoch": 0.5448870137856414,
      "grad_norm": 3.667410373687744,
      "learning_rate": 4.790809721243506e-07,
      "loss": 0.7894,
      "step": 35000
    },
    {
      "epoch": 0.5449648547876108,
      "grad_norm": 4.574625492095947,
      "learning_rate": 4.789990331197457e-07,
      "loss": 0.67,
      "step": 35005
    },
    {
      "epoch": 0.5450426957895802,
      "grad_norm": 3.318500280380249,
      "learning_rate": 4.789170941151406e-07,
      "loss": 0.7467,
      "step": 35010
    },
    {
      "epoch": 0.5451205367915496,
      "grad_norm": 2.900972604751587,
      "learning_rate": 4.788351551105357e-07,
      "loss": 0.8672,
      "step": 35015
    },
    {
      "epoch": 0.545198377793519,
      "grad_norm": 5.759210109710693,
      "learning_rate": 4.787532161059308e-07,
      "loss": 0.7071,
      "step": 35020
    },
    {
      "epoch": 0.5452762187954884,
      "grad_norm": 3.4469902515411377,
      "learning_rate": 4.786712771013257e-07,
      "loss": 0.6963,
      "step": 35025
    },
    {
      "epoch": 0.5453540597974577,
      "grad_norm": 3.267678737640381,
      "learning_rate": 4.785893380967207e-07,
      "loss": 0.8397,
      "step": 35030
    },
    {
      "epoch": 0.545431900799427,
      "grad_norm": 2.9190237522125244,
      "learning_rate": 4.785073990921158e-07,
      "loss": 0.7837,
      "step": 35035
    },
    {
      "epoch": 0.5455097418013964,
      "grad_norm": 3.1442651748657227,
      "learning_rate": 4.784254600875109e-07,
      "loss": 0.6602,
      "step": 35040
    },
    {
      "epoch": 0.5455875828033658,
      "grad_norm": 7.307344913482666,
      "learning_rate": 4.783435210829058e-07,
      "loss": 0.6971,
      "step": 35045
    },
    {
      "epoch": 0.5456654238053352,
      "grad_norm": 6.406971454620361,
      "learning_rate": 4.782615820783009e-07,
      "loss": 0.6557,
      "step": 35050
    },
    {
      "epoch": 0.5457432648073046,
      "grad_norm": 3.590261936187744,
      "learning_rate": 4.781796430736959e-07,
      "loss": 0.7948,
      "step": 35055
    },
    {
      "epoch": 0.545821105809274,
      "grad_norm": 3.7348368167877197,
      "learning_rate": 4.78097704069091e-07,
      "loss": 0.7336,
      "step": 35060
    },
    {
      "epoch": 0.5458989468112434,
      "grad_norm": 4.120712757110596,
      "learning_rate": 4.78015765064486e-07,
      "loss": 0.7873,
      "step": 35065
    },
    {
      "epoch": 0.5459767878132127,
      "grad_norm": 3.2841999530792236,
      "learning_rate": 4.77933826059881e-07,
      "loss": 0.7699,
      "step": 35070
    },
    {
      "epoch": 0.5460546288151821,
      "grad_norm": 2.770313262939453,
      "learning_rate": 4.77851887055276e-07,
      "loss": 0.6826,
      "step": 35075
    },
    {
      "epoch": 0.5461324698171515,
      "grad_norm": 3.1554601192474365,
      "learning_rate": 4.77769948050671e-07,
      "loss": 0.712,
      "step": 35080
    },
    {
      "epoch": 0.5462103108191209,
      "grad_norm": 3.7615017890930176,
      "learning_rate": 4.776880090460661e-07,
      "loss": 0.7531,
      "step": 35085
    },
    {
      "epoch": 0.5462881518210903,
      "grad_norm": 3.646489143371582,
      "learning_rate": 4.776060700414612e-07,
      "loss": 0.8615,
      "step": 35090
    },
    {
      "epoch": 0.5463659928230596,
      "grad_norm": 7.921229362487793,
      "learning_rate": 4.775241310368562e-07,
      "loss": 0.8111,
      "step": 35095
    },
    {
      "epoch": 0.546443833825029,
      "grad_norm": 7.983255386352539,
      "learning_rate": 4.774421920322511e-07,
      "loss": 0.6833,
      "step": 35100
    },
    {
      "epoch": 0.5465216748269984,
      "grad_norm": 3.5857231616973877,
      "learning_rate": 4.773602530276462e-07,
      "loss": 0.7327,
      "step": 35105
    },
    {
      "epoch": 0.5465995158289677,
      "grad_norm": 2.881507635116577,
      "learning_rate": 4.772783140230413e-07,
      "loss": 0.7643,
      "step": 35110
    },
    {
      "epoch": 0.5466773568309371,
      "grad_norm": 3.721663236618042,
      "learning_rate": 4.771963750184362e-07,
      "loss": 0.7459,
      "step": 35115
    },
    {
      "epoch": 0.5467551978329065,
      "grad_norm": 3.531926393508911,
      "learning_rate": 4.771144360138313e-07,
      "loss": 0.8041,
      "step": 35120
    },
    {
      "epoch": 0.5468330388348759,
      "grad_norm": 3.8428966999053955,
      "learning_rate": 4.770324970092263e-07,
      "loss": 0.8223,
      "step": 35125
    },
    {
      "epoch": 0.5469108798368453,
      "grad_norm": 6.577498435974121,
      "learning_rate": 4.769505580046214e-07,
      "loss": 0.7255,
      "step": 35130
    },
    {
      "epoch": 0.5469887208388147,
      "grad_norm": 5.276305675506592,
      "learning_rate": 4.768686190000163e-07,
      "loss": 0.7634,
      "step": 35135
    },
    {
      "epoch": 0.5470665618407841,
      "grad_norm": 4.058431625366211,
      "learning_rate": 4.767866799954114e-07,
      "loss": 0.6567,
      "step": 35140
    },
    {
      "epoch": 0.5471444028427533,
      "grad_norm": 2.9675657749176025,
      "learning_rate": 4.7670474099080643e-07,
      "loss": 0.7446,
      "step": 35145
    },
    {
      "epoch": 0.5472222438447227,
      "grad_norm": 5.9325642585754395,
      "learning_rate": 4.7662280198620145e-07,
      "loss": 0.7346,
      "step": 35150
    },
    {
      "epoch": 0.5473000848466921,
      "grad_norm": 4.076824188232422,
      "learning_rate": 4.7654086298159647e-07,
      "loss": 0.6397,
      "step": 35155
    },
    {
      "epoch": 0.5473779258486615,
      "grad_norm": 4.646028518676758,
      "learning_rate": 4.764589239769915e-07,
      "loss": 0.9456,
      "step": 35160
    },
    {
      "epoch": 0.5474557668506309,
      "grad_norm": 5.500026226043701,
      "learning_rate": 4.763769849723865e-07,
      "loss": 0.7849,
      "step": 35165
    },
    {
      "epoch": 0.5475336078526003,
      "grad_norm": 2.8168888092041016,
      "learning_rate": 4.762950459677816e-07,
      "loss": 0.7165,
      "step": 35170
    },
    {
      "epoch": 0.5476114488545697,
      "grad_norm": 3.6977267265319824,
      "learning_rate": 4.7621310696317656e-07,
      "loss": 0.706,
      "step": 35175
    },
    {
      "epoch": 0.5476892898565391,
      "grad_norm": 3.1508781909942627,
      "learning_rate": 4.7613116795857164e-07,
      "loss": 0.6513,
      "step": 35180
    },
    {
      "epoch": 0.5477671308585084,
      "grad_norm": 4.078734874725342,
      "learning_rate": 4.7604922895396666e-07,
      "loss": 0.6998,
      "step": 35185
    },
    {
      "epoch": 0.5478449718604778,
      "grad_norm": 5.0504655838012695,
      "learning_rate": 4.759672899493617e-07,
      "loss": 0.6801,
      "step": 35190
    },
    {
      "epoch": 0.5479228128624472,
      "grad_norm": 4.0492401123046875,
      "learning_rate": 4.758853509447567e-07,
      "loss": 0.6976,
      "step": 35195
    },
    {
      "epoch": 0.5480006538644165,
      "grad_norm": 13.436319351196289,
      "learning_rate": 4.758034119401517e-07,
      "loss": 0.8591,
      "step": 35200
    },
    {
      "epoch": 0.5480784948663859,
      "grad_norm": 6.603302478790283,
      "learning_rate": 4.7572147293554675e-07,
      "loss": 0.7792,
      "step": 35205
    },
    {
      "epoch": 0.5481563358683553,
      "grad_norm": 4.230200290679932,
      "learning_rate": 4.756395339309418e-07,
      "loss": 0.8776,
      "step": 35210
    },
    {
      "epoch": 0.5482341768703247,
      "grad_norm": 6.1803154945373535,
      "learning_rate": 4.755575949263368e-07,
      "loss": 0.781,
      "step": 35215
    },
    {
      "epoch": 0.548312017872294,
      "grad_norm": 2.8648955821990967,
      "learning_rate": 4.754756559217318e-07,
      "loss": 0.7023,
      "step": 35220
    },
    {
      "epoch": 0.5483898588742634,
      "grad_norm": 2.5462913513183594,
      "learning_rate": 4.753937169171269e-07,
      "loss": 0.7746,
      "step": 35225
    },
    {
      "epoch": 0.5484676998762328,
      "grad_norm": 2.3983867168426514,
      "learning_rate": 4.753117779125219e-07,
      "loss": 0.8851,
      "step": 35230
    },
    {
      "epoch": 0.5485455408782022,
      "grad_norm": 3.268474817276001,
      "learning_rate": 4.752298389079169e-07,
      "loss": 0.7457,
      "step": 35235
    },
    {
      "epoch": 0.5486233818801716,
      "grad_norm": 4.376709938049316,
      "learning_rate": 4.7514789990331195e-07,
      "loss": 0.6743,
      "step": 35240
    },
    {
      "epoch": 0.548701222882141,
      "grad_norm": 3.808220863342285,
      "learning_rate": 4.75065960898707e-07,
      "loss": 0.7033,
      "step": 35245
    },
    {
      "epoch": 0.5487790638841104,
      "grad_norm": 10.179418563842773,
      "learning_rate": 4.7498402189410205e-07,
      "loss": 0.7714,
      "step": 35250
    },
    {
      "epoch": 0.5488569048860797,
      "grad_norm": 5.67426872253418,
      "learning_rate": 4.74902082889497e-07,
      "loss": 0.7798,
      "step": 35255
    },
    {
      "epoch": 0.548934745888049,
      "grad_norm": 4.134709358215332,
      "learning_rate": 4.7482014388489204e-07,
      "loss": 0.7631,
      "step": 35260
    },
    {
      "epoch": 0.5490125868900184,
      "grad_norm": 3.417891025543213,
      "learning_rate": 4.747382048802871e-07,
      "loss": 0.8475,
      "step": 35265
    },
    {
      "epoch": 0.5490904278919878,
      "grad_norm": 4.4843292236328125,
      "learning_rate": 4.7465626587568214e-07,
      "loss": 0.792,
      "step": 35270
    },
    {
      "epoch": 0.5491682688939572,
      "grad_norm": 4.29249382019043,
      "learning_rate": 4.745743268710771e-07,
      "loss": 0.7624,
      "step": 35275
    },
    {
      "epoch": 0.5492461098959266,
      "grad_norm": 3.2616255283355713,
      "learning_rate": 4.744923878664722e-07,
      "loss": 0.7241,
      "step": 35280
    },
    {
      "epoch": 0.549323950897896,
      "grad_norm": 2.8956432342529297,
      "learning_rate": 4.744104488618672e-07,
      "loss": 0.7975,
      "step": 35285
    },
    {
      "epoch": 0.5494017918998654,
      "grad_norm": 3.0964996814727783,
      "learning_rate": 4.743285098572622e-07,
      "loss": 0.7693,
      "step": 35290
    },
    {
      "epoch": 0.5494796329018347,
      "grad_norm": 4.650728702545166,
      "learning_rate": 4.7424657085265725e-07,
      "loss": 0.746,
      "step": 35295
    },
    {
      "epoch": 0.549557473903804,
      "grad_norm": 4.9863762855529785,
      "learning_rate": 4.7416463184805227e-07,
      "loss": 0.7741,
      "step": 35300
    },
    {
      "epoch": 0.5496353149057734,
      "grad_norm": 3.3851091861724854,
      "learning_rate": 4.7408269284344734e-07,
      "loss": 0.7272,
      "step": 35305
    },
    {
      "epoch": 0.5497131559077428,
      "grad_norm": 3.7128796577453613,
      "learning_rate": 4.7400075383884236e-07,
      "loss": 0.9109,
      "step": 35310
    },
    {
      "epoch": 0.5497909969097122,
      "grad_norm": 3.931364059448242,
      "learning_rate": 4.7391881483423733e-07,
      "loss": 0.7029,
      "step": 35315
    },
    {
      "epoch": 0.5498688379116816,
      "grad_norm": 4.920255184173584,
      "learning_rate": 4.738368758296324e-07,
      "loss": 0.6405,
      "step": 35320
    },
    {
      "epoch": 0.549946678913651,
      "grad_norm": 3.7904481887817383,
      "learning_rate": 4.7375493682502743e-07,
      "loss": 0.7931,
      "step": 35325
    },
    {
      "epoch": 0.5500245199156204,
      "grad_norm": 8.922059059143066,
      "learning_rate": 4.7367299782042245e-07,
      "loss": 0.8099,
      "step": 35330
    },
    {
      "epoch": 0.5501023609175897,
      "grad_norm": 4.095458984375,
      "learning_rate": 4.735910588158175e-07,
      "loss": 0.7621,
      "step": 35335
    },
    {
      "epoch": 0.5501802019195591,
      "grad_norm": 3.7293636798858643,
      "learning_rate": 4.735091198112125e-07,
      "loss": 0.774,
      "step": 35340
    },
    {
      "epoch": 0.5502580429215285,
      "grad_norm": 3.6939613819122314,
      "learning_rate": 4.734271808066075e-07,
      "loss": 0.8025,
      "step": 35345
    },
    {
      "epoch": 0.5503358839234979,
      "grad_norm": 7.435025691986084,
      "learning_rate": 4.733452418020026e-07,
      "loss": 0.7903,
      "step": 35350
    },
    {
      "epoch": 0.5504137249254673,
      "grad_norm": 4.679044723510742,
      "learning_rate": 4.732633027973976e-07,
      "loss": 0.6071,
      "step": 35355
    },
    {
      "epoch": 0.5504915659274366,
      "grad_norm": 4.705112934112549,
      "learning_rate": 4.731813637927926e-07,
      "loss": 0.7942,
      "step": 35360
    },
    {
      "epoch": 0.550569406929406,
      "grad_norm": 3.4172236919403076,
      "learning_rate": 4.7309942478818766e-07,
      "loss": 0.7121,
      "step": 35365
    },
    {
      "epoch": 0.5506472479313754,
      "grad_norm": 5.15944766998291,
      "learning_rate": 4.730174857835827e-07,
      "loss": 0.7341,
      "step": 35370
    },
    {
      "epoch": 0.5507250889333447,
      "grad_norm": 5.608235836029053,
      "learning_rate": 4.7293554677897776e-07,
      "loss": 0.7537,
      "step": 35375
    },
    {
      "epoch": 0.5508029299353141,
      "grad_norm": 2.693959951400757,
      "learning_rate": 4.728536077743727e-07,
      "loss": 0.6694,
      "step": 35380
    },
    {
      "epoch": 0.5508807709372835,
      "grad_norm": 4.778819561004639,
      "learning_rate": 4.7277166876976775e-07,
      "loss": 0.7463,
      "step": 35385
    },
    {
      "epoch": 0.5509586119392529,
      "grad_norm": 2.9228625297546387,
      "learning_rate": 4.726897297651628e-07,
      "loss": 0.8912,
      "step": 35390
    },
    {
      "epoch": 0.5510364529412223,
      "grad_norm": 4.187877178192139,
      "learning_rate": 4.7260779076055784e-07,
      "loss": 0.6986,
      "step": 35395
    },
    {
      "epoch": 0.5511142939431917,
      "grad_norm": 3.2928717136383057,
      "learning_rate": 4.725258517559528e-07,
      "loss": 0.7258,
      "step": 35400
    },
    {
      "epoch": 0.5511921349451611,
      "grad_norm": 4.1774139404296875,
      "learning_rate": 4.724439127513479e-07,
      "loss": 0.7426,
      "step": 35405
    },
    {
      "epoch": 0.5512699759471303,
      "grad_norm": 3.012646436691284,
      "learning_rate": 4.723619737467429e-07,
      "loss": 0.7553,
      "step": 35410
    },
    {
      "epoch": 0.5513478169490997,
      "grad_norm": 5.729185104370117,
      "learning_rate": 4.7228003474213793e-07,
      "loss": 0.7002,
      "step": 35415
    },
    {
      "epoch": 0.5514256579510691,
      "grad_norm": 5.044157981872559,
      "learning_rate": 4.7219809573753295e-07,
      "loss": 0.721,
      "step": 35420
    },
    {
      "epoch": 0.5515034989530385,
      "grad_norm": 3.32820463180542,
      "learning_rate": 4.72116156732928e-07,
      "loss": 0.6615,
      "step": 35425
    },
    {
      "epoch": 0.5515813399550079,
      "grad_norm": 6.591919898986816,
      "learning_rate": 4.7203421772832305e-07,
      "loss": 0.6857,
      "step": 35430
    },
    {
      "epoch": 0.5516591809569773,
      "grad_norm": 3.498028039932251,
      "learning_rate": 4.7195227872371807e-07,
      "loss": 0.8577,
      "step": 35435
    },
    {
      "epoch": 0.5517370219589467,
      "grad_norm": 3.7285568714141846,
      "learning_rate": 4.7187033971911304e-07,
      "loss": 0.7987,
      "step": 35440
    },
    {
      "epoch": 0.5518148629609161,
      "grad_norm": 3.7045302391052246,
      "learning_rate": 4.717884007145081e-07,
      "loss": 0.8236,
      "step": 35445
    },
    {
      "epoch": 0.5518927039628854,
      "grad_norm": 3.1306092739105225,
      "learning_rate": 4.7170646170990314e-07,
      "loss": 0.7398,
      "step": 35450
    },
    {
      "epoch": 0.5519705449648548,
      "grad_norm": 5.378826141357422,
      "learning_rate": 4.7162452270529816e-07,
      "loss": 0.8033,
      "step": 35455
    },
    {
      "epoch": 0.5520483859668242,
      "grad_norm": 3.5858540534973145,
      "learning_rate": 4.715425837006932e-07,
      "loss": 0.8628,
      "step": 35460
    },
    {
      "epoch": 0.5521262269687935,
      "grad_norm": 7.7958221435546875,
      "learning_rate": 4.714606446960882e-07,
      "loss": 0.7457,
      "step": 35465
    },
    {
      "epoch": 0.5522040679707629,
      "grad_norm": 3.3041484355926514,
      "learning_rate": 4.713787056914832e-07,
      "loss": 0.6521,
      "step": 35470
    },
    {
      "epoch": 0.5522819089727323,
      "grad_norm": 2.8264236450195312,
      "learning_rate": 4.712967666868783e-07,
      "loss": 0.7672,
      "step": 35475
    },
    {
      "epoch": 0.5523597499747017,
      "grad_norm": 4.413764476776123,
      "learning_rate": 4.7121482768227327e-07,
      "loss": 0.7427,
      "step": 35480
    },
    {
      "epoch": 0.552437590976671,
      "grad_norm": 8.176310539245605,
      "learning_rate": 4.711328886776683e-07,
      "loss": 0.7619,
      "step": 35485
    },
    {
      "epoch": 0.5525154319786404,
      "grad_norm": 2.707061529159546,
      "learning_rate": 4.7105094967306337e-07,
      "loss": 0.7894,
      "step": 35490
    },
    {
      "epoch": 0.5525932729806098,
      "grad_norm": 7.778326511383057,
      "learning_rate": 4.709690106684584e-07,
      "loss": 0.7476,
      "step": 35495
    },
    {
      "epoch": 0.5526711139825792,
      "grad_norm": 5.544581890106201,
      "learning_rate": 4.708870716638534e-07,
      "loss": 0.8527,
      "step": 35500
    },
    {
      "epoch": 0.5527489549845486,
      "grad_norm": 3.5510692596435547,
      "learning_rate": 4.7080513265924843e-07,
      "loss": 0.6632,
      "step": 35505
    },
    {
      "epoch": 0.552826795986518,
      "grad_norm": 5.6578521728515625,
      "learning_rate": 4.7072319365464345e-07,
      "loss": 0.8103,
      "step": 35510
    },
    {
      "epoch": 0.5529046369884874,
      "grad_norm": 4.125245094299316,
      "learning_rate": 4.7064125465003853e-07,
      "loss": 0.6807,
      "step": 35515
    },
    {
      "epoch": 0.5529824779904567,
      "grad_norm": 2.8284120559692383,
      "learning_rate": 4.705593156454335e-07,
      "loss": 0.6254,
      "step": 35520
    },
    {
      "epoch": 0.553060318992426,
      "grad_norm": 3.43818736076355,
      "learning_rate": 4.704773766408285e-07,
      "loss": 0.743,
      "step": 35525
    },
    {
      "epoch": 0.5531381599943954,
      "grad_norm": 6.09321928024292,
      "learning_rate": 4.703954376362236e-07,
      "loss": 0.7198,
      "step": 35530
    },
    {
      "epoch": 0.5532160009963648,
      "grad_norm": 4.032562732696533,
      "learning_rate": 4.703134986316186e-07,
      "loss": 0.7231,
      "step": 35535
    },
    {
      "epoch": 0.5532938419983342,
      "grad_norm": 4.057165145874023,
      "learning_rate": 4.702315596270136e-07,
      "loss": 0.7627,
      "step": 35540
    },
    {
      "epoch": 0.5533716830003036,
      "grad_norm": 10.285215377807617,
      "learning_rate": 4.7014962062240866e-07,
      "loss": 0.649,
      "step": 35545
    },
    {
      "epoch": 0.553449524002273,
      "grad_norm": 2.8972315788269043,
      "learning_rate": 4.700676816178037e-07,
      "loss": 0.8803,
      "step": 35550
    },
    {
      "epoch": 0.5535273650042424,
      "grad_norm": 8.209697723388672,
      "learning_rate": 4.6998574261319876e-07,
      "loss": 0.8159,
      "step": 35555
    },
    {
      "epoch": 0.5536052060062117,
      "grad_norm": 3.154181957244873,
      "learning_rate": 4.699038036085937e-07,
      "loss": 0.6548,
      "step": 35560
    },
    {
      "epoch": 0.553683047008181,
      "grad_norm": 4.422577381134033,
      "learning_rate": 4.6982186460398875e-07,
      "loss": 0.7339,
      "step": 35565
    },
    {
      "epoch": 0.5537608880101504,
      "grad_norm": 5.073526859283447,
      "learning_rate": 4.697399255993838e-07,
      "loss": 0.6163,
      "step": 35570
    },
    {
      "epoch": 0.5538387290121198,
      "grad_norm": 2.5596282482147217,
      "learning_rate": 4.6965798659477884e-07,
      "loss": 0.6971,
      "step": 35575
    },
    {
      "epoch": 0.5539165700140892,
      "grad_norm": 3.410508871078491,
      "learning_rate": 4.695760475901738e-07,
      "loss": 0.8248,
      "step": 35580
    },
    {
      "epoch": 0.5539944110160586,
      "grad_norm": 4.119227886199951,
      "learning_rate": 4.694941085855689e-07,
      "loss": 0.6794,
      "step": 35585
    },
    {
      "epoch": 0.554072252018028,
      "grad_norm": 4.206475734710693,
      "learning_rate": 4.694121695809639e-07,
      "loss": 0.7681,
      "step": 35590
    },
    {
      "epoch": 0.5541500930199974,
      "grad_norm": 4.35361909866333,
      "learning_rate": 4.6933023057635893e-07,
      "loss": 0.6338,
      "step": 35595
    },
    {
      "epoch": 0.5542279340219667,
      "grad_norm": 6.151709079742432,
      "learning_rate": 4.6924829157175395e-07,
      "loss": 0.6774,
      "step": 35600
    },
    {
      "epoch": 0.5543057750239361,
      "grad_norm": 4.920112609863281,
      "learning_rate": 4.69166352567149e-07,
      "loss": 0.7086,
      "step": 35605
    },
    {
      "epoch": 0.5543836160259055,
      "grad_norm": 3.4265146255493164,
      "learning_rate": 4.69084413562544e-07,
      "loss": 0.7737,
      "step": 35610
    },
    {
      "epoch": 0.5544614570278749,
      "grad_norm": 3.4228551387786865,
      "learning_rate": 4.6900247455793907e-07,
      "loss": 0.7826,
      "step": 35615
    },
    {
      "epoch": 0.5545392980298443,
      "grad_norm": 5.777644634246826,
      "learning_rate": 4.6892053555333404e-07,
      "loss": 0.7021,
      "step": 35620
    },
    {
      "epoch": 0.5546171390318136,
      "grad_norm": 2.789003372192383,
      "learning_rate": 4.688385965487291e-07,
      "loss": 0.8894,
      "step": 35625
    },
    {
      "epoch": 0.554694980033783,
      "grad_norm": 11.459249496459961,
      "learning_rate": 4.6875665754412414e-07,
      "loss": 0.713,
      "step": 35630
    },
    {
      "epoch": 0.5547728210357523,
      "grad_norm": 3.8560914993286133,
      "learning_rate": 4.6867471853951916e-07,
      "loss": 0.8141,
      "step": 35635
    },
    {
      "epoch": 0.5548506620377217,
      "grad_norm": 5.832289218902588,
      "learning_rate": 4.685927795349142e-07,
      "loss": 0.7443,
      "step": 35640
    },
    {
      "epoch": 0.5549285030396911,
      "grad_norm": 5.892776012420654,
      "learning_rate": 4.685108405303092e-07,
      "loss": 0.7216,
      "step": 35645
    },
    {
      "epoch": 0.5550063440416605,
      "grad_norm": 5.076382637023926,
      "learning_rate": 4.684289015257042e-07,
      "loss": 0.7896,
      "step": 35650
    },
    {
      "epoch": 0.5550841850436299,
      "grad_norm": 4.4527201652526855,
      "learning_rate": 4.683469625210993e-07,
      "loss": 0.6978,
      "step": 35655
    },
    {
      "epoch": 0.5551620260455993,
      "grad_norm": 3.76173734664917,
      "learning_rate": 4.6826502351649427e-07,
      "loss": 0.7633,
      "step": 35660
    },
    {
      "epoch": 0.5552398670475687,
      "grad_norm": 5.456421375274658,
      "learning_rate": 4.681830845118893e-07,
      "loss": 0.7423,
      "step": 35665
    },
    {
      "epoch": 0.5553177080495381,
      "grad_norm": 3.7368011474609375,
      "learning_rate": 4.6810114550728437e-07,
      "loss": 0.8582,
      "step": 35670
    },
    {
      "epoch": 0.5553955490515073,
      "grad_norm": 4.8903584480285645,
      "learning_rate": 4.680192065026794e-07,
      "loss": 0.7145,
      "step": 35675
    },
    {
      "epoch": 0.5554733900534767,
      "grad_norm": 2.895900011062622,
      "learning_rate": 4.6793726749807446e-07,
      "loss": 0.6546,
      "step": 35680
    },
    {
      "epoch": 0.5555512310554461,
      "grad_norm": 2.8626956939697266,
      "learning_rate": 4.6785532849346943e-07,
      "loss": 0.6479,
      "step": 35685
    },
    {
      "epoch": 0.5556290720574155,
      "grad_norm": 4.208273887634277,
      "learning_rate": 4.6777338948886445e-07,
      "loss": 0.809,
      "step": 35690
    },
    {
      "epoch": 0.5557069130593849,
      "grad_norm": 5.892264366149902,
      "learning_rate": 4.6769145048425953e-07,
      "loss": 0.7543,
      "step": 35695
    },
    {
      "epoch": 0.5557847540613543,
      "grad_norm": 2.9327714443206787,
      "learning_rate": 4.6760951147965455e-07,
      "loss": 0.6433,
      "step": 35700
    },
    {
      "epoch": 0.5558625950633237,
      "grad_norm": 2.8005001544952393,
      "learning_rate": 4.675275724750495e-07,
      "loss": 0.7508,
      "step": 35705
    },
    {
      "epoch": 0.5559404360652931,
      "grad_norm": 4.737391471862793,
      "learning_rate": 4.674456334704446e-07,
      "loss": 0.7073,
      "step": 35710
    },
    {
      "epoch": 0.5560182770672624,
      "grad_norm": 3.236531972885132,
      "learning_rate": 4.673636944658396e-07,
      "loss": 0.7057,
      "step": 35715
    },
    {
      "epoch": 0.5560961180692318,
      "grad_norm": 5.693810939788818,
      "learning_rate": 4.6728175546123464e-07,
      "loss": 0.8197,
      "step": 35720
    },
    {
      "epoch": 0.5561739590712012,
      "grad_norm": 3.6880037784576416,
      "learning_rate": 4.6719981645662966e-07,
      "loss": 0.8489,
      "step": 35725
    },
    {
      "epoch": 0.5562518000731705,
      "grad_norm": 3.7479710578918457,
      "learning_rate": 4.671178774520247e-07,
      "loss": 0.7616,
      "step": 35730
    },
    {
      "epoch": 0.5563296410751399,
      "grad_norm": 3.7133679389953613,
      "learning_rate": 4.670359384474197e-07,
      "loss": 0.7904,
      "step": 35735
    },
    {
      "epoch": 0.5564074820771093,
      "grad_norm": 3.4112889766693115,
      "learning_rate": 4.669539994428148e-07,
      "loss": 0.7765,
      "step": 35740
    },
    {
      "epoch": 0.5564853230790787,
      "grad_norm": 5.6102070808410645,
      "learning_rate": 4.6687206043820975e-07,
      "loss": 0.7618,
      "step": 35745
    },
    {
      "epoch": 0.556563164081048,
      "grad_norm": 2.569159984588623,
      "learning_rate": 4.667901214336048e-07,
      "loss": 0.8563,
      "step": 35750
    },
    {
      "epoch": 0.5566410050830174,
      "grad_norm": 5.6241374015808105,
      "learning_rate": 4.6670818242899985e-07,
      "loss": 0.6539,
      "step": 35755
    },
    {
      "epoch": 0.5567188460849868,
      "grad_norm": 3.927945137023926,
      "learning_rate": 4.6662624342439487e-07,
      "loss": 0.7972,
      "step": 35760
    },
    {
      "epoch": 0.5567966870869562,
      "grad_norm": 3.401947498321533,
      "learning_rate": 4.665443044197899e-07,
      "loss": 0.76,
      "step": 35765
    },
    {
      "epoch": 0.5568745280889256,
      "grad_norm": 5.090336322784424,
      "learning_rate": 4.664623654151849e-07,
      "loss": 0.8135,
      "step": 35770
    },
    {
      "epoch": 0.556952369090895,
      "grad_norm": 5.554099082946777,
      "learning_rate": 4.6638042641057993e-07,
      "loss": 0.7479,
      "step": 35775
    },
    {
      "epoch": 0.5570302100928644,
      "grad_norm": 2.744924545288086,
      "learning_rate": 4.66298487405975e-07,
      "loss": 0.6804,
      "step": 35780
    },
    {
      "epoch": 0.5571080510948337,
      "grad_norm": 5.976934432983398,
      "learning_rate": 4.6621654840137e-07,
      "loss": 0.6729,
      "step": 35785
    },
    {
      "epoch": 0.557185892096803,
      "grad_norm": 3.4525482654571533,
      "learning_rate": 4.66134609396765e-07,
      "loss": 0.6789,
      "step": 35790
    },
    {
      "epoch": 0.5572637330987724,
      "grad_norm": 5.425734519958496,
      "learning_rate": 4.6605267039216007e-07,
      "loss": 0.7129,
      "step": 35795
    },
    {
      "epoch": 0.5573415741007418,
      "grad_norm": 5.292181015014648,
      "learning_rate": 4.659707313875551e-07,
      "loss": 0.805,
      "step": 35800
    },
    {
      "epoch": 0.5574194151027112,
      "grad_norm": 4.141442775726318,
      "learning_rate": 4.658887923829501e-07,
      "loss": 0.7797,
      "step": 35805
    },
    {
      "epoch": 0.5574972561046806,
      "grad_norm": 3.682215690612793,
      "learning_rate": 4.6580685337834514e-07,
      "loss": 0.8093,
      "step": 35810
    },
    {
      "epoch": 0.55757509710665,
      "grad_norm": 2.8766698837280273,
      "learning_rate": 4.6572491437374016e-07,
      "loss": 0.6752,
      "step": 35815
    },
    {
      "epoch": 0.5576529381086194,
      "grad_norm": 5.1771240234375,
      "learning_rate": 4.6564297536913524e-07,
      "loss": 0.7073,
      "step": 35820
    },
    {
      "epoch": 0.5577307791105887,
      "grad_norm": 3.8695921897888184,
      "learning_rate": 4.655610363645302e-07,
      "loss": 0.6811,
      "step": 35825
    },
    {
      "epoch": 0.557808620112558,
      "grad_norm": 3.362762451171875,
      "learning_rate": 4.6547909735992523e-07,
      "loss": 0.6435,
      "step": 35830
    },
    {
      "epoch": 0.5578864611145274,
      "grad_norm": 6.592752933502197,
      "learning_rate": 4.653971583553203e-07,
      "loss": 0.7716,
      "step": 35835
    },
    {
      "epoch": 0.5579643021164968,
      "grad_norm": 7.647515773773193,
      "learning_rate": 4.653152193507153e-07,
      "loss": 0.7412,
      "step": 35840
    },
    {
      "epoch": 0.5580421431184662,
      "grad_norm": 6.323180198669434,
      "learning_rate": 4.652332803461103e-07,
      "loss": 0.7776,
      "step": 35845
    },
    {
      "epoch": 0.5581199841204356,
      "grad_norm": 4.1566243171691895,
      "learning_rate": 4.6515134134150537e-07,
      "loss": 0.7079,
      "step": 35850
    },
    {
      "epoch": 0.558197825122405,
      "grad_norm": 4.525944232940674,
      "learning_rate": 4.650694023369004e-07,
      "loss": 0.6785,
      "step": 35855
    },
    {
      "epoch": 0.5582756661243744,
      "grad_norm": 7.200018405914307,
      "learning_rate": 4.649874633322954e-07,
      "loss": 0.7066,
      "step": 35860
    },
    {
      "epoch": 0.5583535071263437,
      "grad_norm": 3.141693353652954,
      "learning_rate": 4.6490552432769043e-07,
      "loss": 0.8966,
      "step": 35865
    },
    {
      "epoch": 0.5584313481283131,
      "grad_norm": 4.483099460601807,
      "learning_rate": 4.6482358532308546e-07,
      "loss": 0.785,
      "step": 35870
    },
    {
      "epoch": 0.5585091891302825,
      "grad_norm": 6.768981456756592,
      "learning_rate": 4.6474164631848053e-07,
      "loss": 0.7452,
      "step": 35875
    },
    {
      "epoch": 0.5585870301322519,
      "grad_norm": 8.592025756835938,
      "learning_rate": 4.6465970731387555e-07,
      "loss": 0.9273,
      "step": 35880
    },
    {
      "epoch": 0.5586648711342213,
      "grad_norm": 3.496518135070801,
      "learning_rate": 4.645777683092705e-07,
      "loss": 0.7974,
      "step": 35885
    },
    {
      "epoch": 0.5587427121361906,
      "grad_norm": 4.193195819854736,
      "learning_rate": 4.644958293046656e-07,
      "loss": 0.7666,
      "step": 35890
    },
    {
      "epoch": 0.55882055313816,
      "grad_norm": 3.298384189605713,
      "learning_rate": 4.644138903000606e-07,
      "loss": 0.7261,
      "step": 35895
    },
    {
      "epoch": 0.5588983941401293,
      "grad_norm": 10.040658950805664,
      "learning_rate": 4.6433195129545564e-07,
      "loss": 0.7096,
      "step": 35900
    },
    {
      "epoch": 0.5589762351420987,
      "grad_norm": 2.9459140300750732,
      "learning_rate": 4.6425001229085066e-07,
      "loss": 0.8274,
      "step": 35905
    },
    {
      "epoch": 0.5590540761440681,
      "grad_norm": 3.8296151161193848,
      "learning_rate": 4.641680732862457e-07,
      "loss": 0.783,
      "step": 35910
    },
    {
      "epoch": 0.5591319171460375,
      "grad_norm": 3.2005720138549805,
      "learning_rate": 4.640861342816407e-07,
      "loss": 0.6892,
      "step": 35915
    },
    {
      "epoch": 0.5592097581480069,
      "grad_norm": 3.081958293914795,
      "learning_rate": 4.640041952770358e-07,
      "loss": 0.6255,
      "step": 35920
    },
    {
      "epoch": 0.5592875991499763,
      "grad_norm": 3.5177485942840576,
      "learning_rate": 4.6392225627243075e-07,
      "loss": 0.8508,
      "step": 35925
    },
    {
      "epoch": 0.5593654401519457,
      "grad_norm": 2.8935422897338867,
      "learning_rate": 4.638403172678258e-07,
      "loss": 0.5727,
      "step": 35930
    },
    {
      "epoch": 0.5594432811539151,
      "grad_norm": 3.7642908096313477,
      "learning_rate": 4.6375837826322085e-07,
      "loss": 0.9037,
      "step": 35935
    },
    {
      "epoch": 0.5595211221558843,
      "grad_norm": 2.8425753116607666,
      "learning_rate": 4.6367643925861587e-07,
      "loss": 0.6724,
      "step": 35940
    },
    {
      "epoch": 0.5595989631578537,
      "grad_norm": 4.154412746429443,
      "learning_rate": 4.635945002540109e-07,
      "loss": 0.716,
      "step": 35945
    },
    {
      "epoch": 0.5596768041598231,
      "grad_norm": 8.159989356994629,
      "learning_rate": 4.635125612494059e-07,
      "loss": 0.7407,
      "step": 35950
    },
    {
      "epoch": 0.5597546451617925,
      "grad_norm": 3.139669179916382,
      "learning_rate": 4.6343062224480093e-07,
      "loss": 0.7875,
      "step": 35955
    },
    {
      "epoch": 0.5598324861637619,
      "grad_norm": 3.2319705486297607,
      "learning_rate": 4.63348683240196e-07,
      "loss": 0.6204,
      "step": 35960
    },
    {
      "epoch": 0.5599103271657313,
      "grad_norm": 4.022894382476807,
      "learning_rate": 4.63266744235591e-07,
      "loss": 0.7567,
      "step": 35965
    },
    {
      "epoch": 0.5599881681677007,
      "grad_norm": 4.326730728149414,
      "learning_rate": 4.63184805230986e-07,
      "loss": 0.8875,
      "step": 35970
    },
    {
      "epoch": 0.5600660091696701,
      "grad_norm": 11.820920944213867,
      "learning_rate": 4.631028662263811e-07,
      "loss": 0.8068,
      "step": 35975
    },
    {
      "epoch": 0.5601438501716394,
      "grad_norm": 3.2422702312469482,
      "learning_rate": 4.630209272217761e-07,
      "loss": 0.7086,
      "step": 35980
    },
    {
      "epoch": 0.5602216911736088,
      "grad_norm": 3.7909584045410156,
      "learning_rate": 4.6293898821717107e-07,
      "loss": 0.733,
      "step": 35985
    },
    {
      "epoch": 0.5602995321755782,
      "grad_norm": 4.180745601654053,
      "learning_rate": 4.6285704921256614e-07,
      "loss": 0.7298,
      "step": 35990
    },
    {
      "epoch": 0.5603773731775475,
      "grad_norm": 4.838062763214111,
      "learning_rate": 4.6277511020796116e-07,
      "loss": 0.819,
      "step": 35995
    },
    {
      "epoch": 0.5604552141795169,
      "grad_norm": 5.055113792419434,
      "learning_rate": 4.6269317120335624e-07,
      "loss": 0.7845,
      "step": 36000
    },
    {
      "epoch": 0.5605330551814863,
      "grad_norm": 3.502206325531006,
      "learning_rate": 4.6261123219875126e-07,
      "loss": 0.7345,
      "step": 36005
    },
    {
      "epoch": 0.5606108961834557,
      "grad_norm": 3.6397273540496826,
      "learning_rate": 4.6252929319414623e-07,
      "loss": 0.8646,
      "step": 36010
    },
    {
      "epoch": 0.560688737185425,
      "grad_norm": 8.188968658447266,
      "learning_rate": 4.624473541895413e-07,
      "loss": 0.6948,
      "step": 36015
    },
    {
      "epoch": 0.5607665781873944,
      "grad_norm": 3.284403085708618,
      "learning_rate": 4.623654151849363e-07,
      "loss": 0.7206,
      "step": 36020
    },
    {
      "epoch": 0.5608444191893638,
      "grad_norm": 3.2094428539276123,
      "learning_rate": 4.6228347618033135e-07,
      "loss": 0.8098,
      "step": 36025
    },
    {
      "epoch": 0.5609222601913332,
      "grad_norm": 4.399909973144531,
      "learning_rate": 4.6220153717572637e-07,
      "loss": 0.6655,
      "step": 36030
    },
    {
      "epoch": 0.5610001011933026,
      "grad_norm": 3.221421003341675,
      "learning_rate": 4.621195981711214e-07,
      "loss": 0.9218,
      "step": 36035
    },
    {
      "epoch": 0.561077942195272,
      "grad_norm": 3.4221837520599365,
      "learning_rate": 4.620376591665164e-07,
      "loss": 0.7738,
      "step": 36040
    },
    {
      "epoch": 0.5611557831972414,
      "grad_norm": 3.2252297401428223,
      "learning_rate": 4.619557201619115e-07,
      "loss": 0.7931,
      "step": 36045
    },
    {
      "epoch": 0.5612336241992107,
      "grad_norm": 6.63355827331543,
      "learning_rate": 4.6187378115730646e-07,
      "loss": 0.8244,
      "step": 36050
    },
    {
      "epoch": 0.56131146520118,
      "grad_norm": 2.6410329341888428,
      "learning_rate": 4.6179184215270153e-07,
      "loss": 0.682,
      "step": 36055
    },
    {
      "epoch": 0.5613893062031494,
      "grad_norm": 8.235663414001465,
      "learning_rate": 4.6170990314809655e-07,
      "loss": 0.6645,
      "step": 36060
    },
    {
      "epoch": 0.5614671472051188,
      "grad_norm": 4.745312213897705,
      "learning_rate": 4.616279641434916e-07,
      "loss": 0.8146,
      "step": 36065
    },
    {
      "epoch": 0.5615449882070882,
      "grad_norm": 3.954742431640625,
      "learning_rate": 4.615460251388866e-07,
      "loss": 0.7639,
      "step": 36070
    },
    {
      "epoch": 0.5616228292090576,
      "grad_norm": 3.0466911792755127,
      "learning_rate": 4.614640861342816e-07,
      "loss": 0.7614,
      "step": 36075
    },
    {
      "epoch": 0.561700670211027,
      "grad_norm": 3.648780107498169,
      "learning_rate": 4.6138214712967664e-07,
      "loss": 0.6669,
      "step": 36080
    },
    {
      "epoch": 0.5617785112129964,
      "grad_norm": 3.142124652862549,
      "learning_rate": 4.613002081250717e-07,
      "loss": 0.7453,
      "step": 36085
    },
    {
      "epoch": 0.5618563522149657,
      "grad_norm": 2.845841646194458,
      "learning_rate": 4.612182691204667e-07,
      "loss": 0.7058,
      "step": 36090
    },
    {
      "epoch": 0.561934193216935,
      "grad_norm": 6.611133098602295,
      "learning_rate": 4.611363301158617e-07,
      "loss": 0.7563,
      "step": 36095
    },
    {
      "epoch": 0.5620120342189044,
      "grad_norm": 3.4856228828430176,
      "learning_rate": 4.610543911112568e-07,
      "loss": 0.7575,
      "step": 36100
    },
    {
      "epoch": 0.5620898752208738,
      "grad_norm": 2.692227363586426,
      "learning_rate": 4.609724521066518e-07,
      "loss": 0.8196,
      "step": 36105
    },
    {
      "epoch": 0.5621677162228432,
      "grad_norm": 5.362658500671387,
      "learning_rate": 4.6089051310204677e-07,
      "loss": 0.8346,
      "step": 36110
    },
    {
      "epoch": 0.5622455572248126,
      "grad_norm": 2.9289515018463135,
      "learning_rate": 4.6080857409744185e-07,
      "loss": 0.6121,
      "step": 36115
    },
    {
      "epoch": 0.562323398226782,
      "grad_norm": 4.042474746704102,
      "learning_rate": 4.6072663509283687e-07,
      "loss": 0.628,
      "step": 36120
    },
    {
      "epoch": 0.5624012392287514,
      "grad_norm": 3.661647081375122,
      "learning_rate": 4.6064469608823194e-07,
      "loss": 0.6336,
      "step": 36125
    },
    {
      "epoch": 0.5624790802307207,
      "grad_norm": 3.3725786209106445,
      "learning_rate": 4.605627570836269e-07,
      "loss": 0.8387,
      "step": 36130
    },
    {
      "epoch": 0.5625569212326901,
      "grad_norm": 3.2094225883483887,
      "learning_rate": 4.6048081807902193e-07,
      "loss": 0.7327,
      "step": 36135
    },
    {
      "epoch": 0.5626347622346595,
      "grad_norm": 3.3927948474884033,
      "learning_rate": 4.60398879074417e-07,
      "loss": 0.9017,
      "step": 36140
    },
    {
      "epoch": 0.5627126032366289,
      "grad_norm": 6.644576072692871,
      "learning_rate": 4.6031694006981203e-07,
      "loss": 0.7119,
      "step": 36145
    },
    {
      "epoch": 0.5627904442385983,
      "grad_norm": 2.5468971729278564,
      "learning_rate": 4.60235001065207e-07,
      "loss": 0.6967,
      "step": 36150
    },
    {
      "epoch": 0.5628682852405676,
      "grad_norm": 3.665527105331421,
      "learning_rate": 4.601530620606021e-07,
      "loss": 0.6859,
      "step": 36155
    },
    {
      "epoch": 0.562946126242537,
      "grad_norm": 2.801375150680542,
      "learning_rate": 4.600711230559971e-07,
      "loss": 0.648,
      "step": 36160
    },
    {
      "epoch": 0.5630239672445063,
      "grad_norm": 3.1157076358795166,
      "learning_rate": 4.599891840513921e-07,
      "loss": 0.7933,
      "step": 36165
    },
    {
      "epoch": 0.5631018082464757,
      "grad_norm": 5.971066951751709,
      "learning_rate": 4.5990724504678714e-07,
      "loss": 0.7583,
      "step": 36170
    },
    {
      "epoch": 0.5631796492484451,
      "grad_norm": 3.293128252029419,
      "learning_rate": 4.5982530604218216e-07,
      "loss": 0.7143,
      "step": 36175
    },
    {
      "epoch": 0.5632574902504145,
      "grad_norm": 3.663938522338867,
      "learning_rate": 4.5974336703757724e-07,
      "loss": 0.9017,
      "step": 36180
    },
    {
      "epoch": 0.5633353312523839,
      "grad_norm": 9.356369972229004,
      "learning_rate": 4.5966142803297226e-07,
      "loss": 0.7251,
      "step": 36185
    },
    {
      "epoch": 0.5634131722543533,
      "grad_norm": 7.07928991317749,
      "learning_rate": 4.5957948902836723e-07,
      "loss": 0.7263,
      "step": 36190
    },
    {
      "epoch": 0.5634910132563227,
      "grad_norm": 2.699371099472046,
      "learning_rate": 4.594975500237623e-07,
      "loss": 0.7828,
      "step": 36195
    },
    {
      "epoch": 0.5635688542582921,
      "grad_norm": 3.2599191665649414,
      "learning_rate": 4.594156110191573e-07,
      "loss": 0.7722,
      "step": 36200
    },
    {
      "epoch": 0.5636466952602613,
      "grad_norm": 3.755296468734741,
      "learning_rate": 4.5933367201455235e-07,
      "loss": 0.8832,
      "step": 36205
    },
    {
      "epoch": 0.5637245362622307,
      "grad_norm": 2.2559268474578857,
      "learning_rate": 4.5925173300994737e-07,
      "loss": 0.7918,
      "step": 36210
    },
    {
      "epoch": 0.5638023772642001,
      "grad_norm": 3.1429529190063477,
      "learning_rate": 4.591697940053424e-07,
      "loss": 0.7086,
      "step": 36215
    },
    {
      "epoch": 0.5638802182661695,
      "grad_norm": 9.8464937210083,
      "learning_rate": 4.590878550007374e-07,
      "loss": 0.7214,
      "step": 36220
    },
    {
      "epoch": 0.5639580592681389,
      "grad_norm": 4.964937210083008,
      "learning_rate": 4.590059159961325e-07,
      "loss": 0.7505,
      "step": 36225
    },
    {
      "epoch": 0.5640359002701083,
      "grad_norm": 2.5905959606170654,
      "learning_rate": 4.5892397699152746e-07,
      "loss": 0.6829,
      "step": 36230
    },
    {
      "epoch": 0.5641137412720777,
      "grad_norm": 3.9555907249450684,
      "learning_rate": 4.588420379869225e-07,
      "loss": 0.7328,
      "step": 36235
    },
    {
      "epoch": 0.564191582274047,
      "grad_norm": 2.4797496795654297,
      "learning_rate": 4.5876009898231755e-07,
      "loss": 0.8573,
      "step": 36240
    },
    {
      "epoch": 0.5642694232760164,
      "grad_norm": 5.142312049865723,
      "learning_rate": 4.586781599777126e-07,
      "loss": 0.6387,
      "step": 36245
    },
    {
      "epoch": 0.5643472642779858,
      "grad_norm": 4.1012725830078125,
      "learning_rate": 4.585962209731076e-07,
      "loss": 0.6264,
      "step": 36250
    },
    {
      "epoch": 0.5644251052799552,
      "grad_norm": 3.8164288997650146,
      "learning_rate": 4.585142819685026e-07,
      "loss": 0.7371,
      "step": 36255
    },
    {
      "epoch": 0.5645029462819245,
      "grad_norm": 3.924522876739502,
      "learning_rate": 4.5843234296389764e-07,
      "loss": 0.7154,
      "step": 36260
    },
    {
      "epoch": 0.5645807872838939,
      "grad_norm": 5.388657569885254,
      "learning_rate": 4.583504039592927e-07,
      "loss": 0.6559,
      "step": 36265
    },
    {
      "epoch": 0.5646586282858633,
      "grad_norm": 5.015802383422852,
      "learning_rate": 4.582684649546877e-07,
      "loss": 0.7687,
      "step": 36270
    },
    {
      "epoch": 0.5647364692878327,
      "grad_norm": 4.677270889282227,
      "learning_rate": 4.581865259500827e-07,
      "loss": 0.7392,
      "step": 36275
    },
    {
      "epoch": 0.564814310289802,
      "grad_norm": 5.2854180335998535,
      "learning_rate": 4.581045869454778e-07,
      "loss": 0.6996,
      "step": 36280
    },
    {
      "epoch": 0.5648921512917714,
      "grad_norm": 4.4013447761535645,
      "learning_rate": 4.580226479408728e-07,
      "loss": 0.724,
      "step": 36285
    },
    {
      "epoch": 0.5649699922937408,
      "grad_norm": 4.112987041473389,
      "learning_rate": 4.5794070893626777e-07,
      "loss": 0.6699,
      "step": 36290
    },
    {
      "epoch": 0.5650478332957102,
      "grad_norm": 3.1017234325408936,
      "learning_rate": 4.5785876993166285e-07,
      "loss": 0.7549,
      "step": 36295
    },
    {
      "epoch": 0.5651256742976796,
      "grad_norm": 8.300948143005371,
      "learning_rate": 4.5777683092705787e-07,
      "loss": 0.7753,
      "step": 36300
    },
    {
      "epoch": 0.565203515299649,
      "grad_norm": 3.914881467819214,
      "learning_rate": 4.5769489192245294e-07,
      "loss": 0.7381,
      "step": 36305
    },
    {
      "epoch": 0.5652813563016184,
      "grad_norm": 3.0609562397003174,
      "learning_rate": 4.5761295291784797e-07,
      "loss": 0.7663,
      "step": 36310
    },
    {
      "epoch": 0.5653591973035877,
      "grad_norm": 3.086632490158081,
      "learning_rate": 4.5753101391324294e-07,
      "loss": 0.8005,
      "step": 36315
    },
    {
      "epoch": 0.565437038305557,
      "grad_norm": 6.448633670806885,
      "learning_rate": 4.57449074908638e-07,
      "loss": 0.7684,
      "step": 36320
    },
    {
      "epoch": 0.5655148793075264,
      "grad_norm": 2.8630194664001465,
      "learning_rate": 4.5736713590403303e-07,
      "loss": 0.8223,
      "step": 36325
    },
    {
      "epoch": 0.5655927203094958,
      "grad_norm": 3.115567922592163,
      "learning_rate": 4.5728519689942805e-07,
      "loss": 0.8198,
      "step": 36330
    },
    {
      "epoch": 0.5656705613114652,
      "grad_norm": 2.8401238918304443,
      "learning_rate": 4.572032578948231e-07,
      "loss": 0.6967,
      "step": 36335
    },
    {
      "epoch": 0.5657484023134346,
      "grad_norm": 4.308891296386719,
      "learning_rate": 4.571213188902181e-07,
      "loss": 0.7222,
      "step": 36340
    },
    {
      "epoch": 0.565826243315404,
      "grad_norm": 5.021890163421631,
      "learning_rate": 4.570393798856131e-07,
      "loss": 0.6999,
      "step": 36345
    },
    {
      "epoch": 0.5659040843173734,
      "grad_norm": 3.706099510192871,
      "learning_rate": 4.569574408810082e-07,
      "loss": 0.8203,
      "step": 36350
    },
    {
      "epoch": 0.5659819253193427,
      "grad_norm": 2.858363151550293,
      "learning_rate": 4.5687550187640316e-07,
      "loss": 0.6879,
      "step": 36355
    },
    {
      "epoch": 0.566059766321312,
      "grad_norm": 3.8888208866119385,
      "learning_rate": 4.567935628717982e-07,
      "loss": 0.6652,
      "step": 36360
    },
    {
      "epoch": 0.5661376073232814,
      "grad_norm": 3.7912349700927734,
      "learning_rate": 4.5671162386719326e-07,
      "loss": 0.701,
      "step": 36365
    },
    {
      "epoch": 0.5662154483252508,
      "grad_norm": 2.9038097858428955,
      "learning_rate": 4.566296848625883e-07,
      "loss": 0.7508,
      "step": 36370
    },
    {
      "epoch": 0.5662932893272202,
      "grad_norm": 3.879856824874878,
      "learning_rate": 4.565477458579833e-07,
      "loss": 0.8364,
      "step": 36375
    },
    {
      "epoch": 0.5663711303291896,
      "grad_norm": 3.312741994857788,
      "learning_rate": 4.564658068533783e-07,
      "loss": 0.7005,
      "step": 36380
    },
    {
      "epoch": 0.566448971331159,
      "grad_norm": 6.319469451904297,
      "learning_rate": 4.5638386784877335e-07,
      "loss": 0.8224,
      "step": 36385
    },
    {
      "epoch": 0.5665268123331284,
      "grad_norm": 2.935088872909546,
      "learning_rate": 4.563019288441684e-07,
      "loss": 0.7518,
      "step": 36390
    },
    {
      "epoch": 0.5666046533350977,
      "grad_norm": 4.225389003753662,
      "learning_rate": 4.562199898395634e-07,
      "loss": 0.8553,
      "step": 36395
    },
    {
      "epoch": 0.5666824943370671,
      "grad_norm": 3.4155266284942627,
      "learning_rate": 4.561380508349584e-07,
      "loss": 0.6988,
      "step": 36400
    },
    {
      "epoch": 0.5667603353390365,
      "grad_norm": 2.4987235069274902,
      "learning_rate": 4.560561118303535e-07,
      "loss": 0.7206,
      "step": 36405
    },
    {
      "epoch": 0.5668381763410059,
      "grad_norm": 7.193516731262207,
      "learning_rate": 4.559741728257485e-07,
      "loss": 0.8197,
      "step": 36410
    },
    {
      "epoch": 0.5669160173429753,
      "grad_norm": 10.598738670349121,
      "learning_rate": 4.558922338211435e-07,
      "loss": 0.7825,
      "step": 36415
    },
    {
      "epoch": 0.5669938583449446,
      "grad_norm": 4.58589506149292,
      "learning_rate": 4.5581029481653855e-07,
      "loss": 0.7882,
      "step": 36420
    },
    {
      "epoch": 0.567071699346914,
      "grad_norm": 3.0383994579315186,
      "learning_rate": 4.557283558119336e-07,
      "loss": 0.8274,
      "step": 36425
    },
    {
      "epoch": 0.5671495403488833,
      "grad_norm": 10.662517547607422,
      "learning_rate": 4.5564641680732865e-07,
      "loss": 0.856,
      "step": 36430
    },
    {
      "epoch": 0.5672273813508527,
      "grad_norm": 2.8928160667419434,
      "learning_rate": 4.555644778027236e-07,
      "loss": 0.6723,
      "step": 36435
    },
    {
      "epoch": 0.5673052223528221,
      "grad_norm": 19.3938045501709,
      "learning_rate": 4.5548253879811864e-07,
      "loss": 0.8463,
      "step": 36440
    },
    {
      "epoch": 0.5673830633547915,
      "grad_norm": 5.673638820648193,
      "learning_rate": 4.554005997935137e-07,
      "loss": 0.7606,
      "step": 36445
    },
    {
      "epoch": 0.5674609043567609,
      "grad_norm": 4.36138391494751,
      "learning_rate": 4.5531866078890874e-07,
      "loss": 0.8976,
      "step": 36450
    },
    {
      "epoch": 0.5675387453587303,
      "grad_norm": 2.895538091659546,
      "learning_rate": 4.552367217843037e-07,
      "loss": 0.8334,
      "step": 36455
    },
    {
      "epoch": 0.5676165863606997,
      "grad_norm": 3.3594233989715576,
      "learning_rate": 4.551547827796988e-07,
      "loss": 0.7681,
      "step": 36460
    },
    {
      "epoch": 0.5676944273626691,
      "grad_norm": 2.764101505279541,
      "learning_rate": 4.550728437750938e-07,
      "loss": 0.6884,
      "step": 36465
    },
    {
      "epoch": 0.5677722683646383,
      "grad_norm": 3.9291276931762695,
      "learning_rate": 4.5499090477048883e-07,
      "loss": 0.5588,
      "step": 36470
    },
    {
      "epoch": 0.5678501093666077,
      "grad_norm": 6.066794395446777,
      "learning_rate": 4.5490896576588385e-07,
      "loss": 0.7311,
      "step": 36475
    },
    {
      "epoch": 0.5679279503685771,
      "grad_norm": 2.6867339611053467,
      "learning_rate": 4.5482702676127887e-07,
      "loss": 0.7108,
      "step": 36480
    },
    {
      "epoch": 0.5680057913705465,
      "grad_norm": 8.926070213317871,
      "learning_rate": 4.547450877566739e-07,
      "loss": 0.7099,
      "step": 36485
    },
    {
      "epoch": 0.5680836323725159,
      "grad_norm": 3.7634787559509277,
      "learning_rate": 4.5466314875206897e-07,
      "loss": 0.7889,
      "step": 36490
    },
    {
      "epoch": 0.5681614733744853,
      "grad_norm": 3.0038180351257324,
      "learning_rate": 4.5458120974746394e-07,
      "loss": 0.8108,
      "step": 36495
    },
    {
      "epoch": 0.5682393143764547,
      "grad_norm": 8.85113525390625,
      "learning_rate": 4.54499270742859e-07,
      "loss": 0.7411,
      "step": 36500
    },
    {
      "epoch": 0.568317155378424,
      "grad_norm": 5.962666988372803,
      "learning_rate": 4.5441733173825403e-07,
      "loss": 0.7578,
      "step": 36505
    },
    {
      "epoch": 0.5683949963803934,
      "grad_norm": 3.8949451446533203,
      "learning_rate": 4.5433539273364905e-07,
      "loss": 0.81,
      "step": 36510
    },
    {
      "epoch": 0.5684728373823628,
      "grad_norm": 2.313023090362549,
      "learning_rate": 4.542534537290441e-07,
      "loss": 0.6381,
      "step": 36515
    },
    {
      "epoch": 0.5685506783843322,
      "grad_norm": 3.233367919921875,
      "learning_rate": 4.541715147244391e-07,
      "loss": 0.7886,
      "step": 36520
    },
    {
      "epoch": 0.5686285193863015,
      "grad_norm": 4.123289585113525,
      "learning_rate": 4.540895757198341e-07,
      "loss": 0.7783,
      "step": 36525
    },
    {
      "epoch": 0.5687063603882709,
      "grad_norm": 3.652589797973633,
      "learning_rate": 4.540076367152292e-07,
      "loss": 0.7242,
      "step": 36530
    },
    {
      "epoch": 0.5687842013902403,
      "grad_norm": 3.9108550548553467,
      "learning_rate": 4.5392569771062416e-07,
      "loss": 0.739,
      "step": 36535
    },
    {
      "epoch": 0.5688620423922097,
      "grad_norm": 2.775463581085205,
      "learning_rate": 4.538437587060192e-07,
      "loss": 0.7467,
      "step": 36540
    },
    {
      "epoch": 0.568939883394179,
      "grad_norm": 3.8630290031433105,
      "learning_rate": 4.5376181970141426e-07,
      "loss": 0.7002,
      "step": 36545
    },
    {
      "epoch": 0.5690177243961484,
      "grad_norm": 6.369472503662109,
      "learning_rate": 4.536798806968093e-07,
      "loss": 0.8137,
      "step": 36550
    },
    {
      "epoch": 0.5690955653981178,
      "grad_norm": 3.333348512649536,
      "learning_rate": 4.5359794169220425e-07,
      "loss": 0.7119,
      "step": 36555
    },
    {
      "epoch": 0.5691734064000872,
      "grad_norm": 4.339738368988037,
      "learning_rate": 4.5351600268759933e-07,
      "loss": 0.6578,
      "step": 36560
    },
    {
      "epoch": 0.5692512474020566,
      "grad_norm": 4.5578484535217285,
      "learning_rate": 4.5343406368299435e-07,
      "loss": 0.7251,
      "step": 36565
    },
    {
      "epoch": 0.569329088404026,
      "grad_norm": 7.2544779777526855,
      "learning_rate": 4.533521246783894e-07,
      "loss": 0.7857,
      "step": 36570
    },
    {
      "epoch": 0.5694069294059954,
      "grad_norm": 3.029428005218506,
      "learning_rate": 4.532701856737844e-07,
      "loss": 0.6827,
      "step": 36575
    },
    {
      "epoch": 0.5694847704079647,
      "grad_norm": 3.827564001083374,
      "learning_rate": 4.531882466691794e-07,
      "loss": 0.6693,
      "step": 36580
    },
    {
      "epoch": 0.569562611409934,
      "grad_norm": 3.0939109325408936,
      "learning_rate": 4.531063076645745e-07,
      "loss": 0.8319,
      "step": 36585
    },
    {
      "epoch": 0.5696404524119034,
      "grad_norm": 5.642122268676758,
      "learning_rate": 4.530243686599695e-07,
      "loss": 0.7853,
      "step": 36590
    },
    {
      "epoch": 0.5697182934138728,
      "grad_norm": 3.351140022277832,
      "learning_rate": 4.529424296553645e-07,
      "loss": 0.7637,
      "step": 36595
    },
    {
      "epoch": 0.5697961344158422,
      "grad_norm": 4.401710510253906,
      "learning_rate": 4.5286049065075956e-07,
      "loss": 0.7934,
      "step": 36600
    },
    {
      "epoch": 0.5698739754178116,
      "grad_norm": 3.6507439613342285,
      "learning_rate": 4.527785516461546e-07,
      "loss": 0.7037,
      "step": 36605
    },
    {
      "epoch": 0.569951816419781,
      "grad_norm": 4.337538719177246,
      "learning_rate": 4.526966126415496e-07,
      "loss": 0.8411,
      "step": 36610
    },
    {
      "epoch": 0.5700296574217504,
      "grad_norm": 5.044707298278809,
      "learning_rate": 4.526146736369446e-07,
      "loss": 0.7601,
      "step": 36615
    },
    {
      "epoch": 0.5701074984237197,
      "grad_norm": 3.7404119968414307,
      "learning_rate": 4.5253273463233964e-07,
      "loss": 0.7743,
      "step": 36620
    },
    {
      "epoch": 0.570185339425689,
      "grad_norm": 3.6612942218780518,
      "learning_rate": 4.524507956277347e-07,
      "loss": 0.785,
      "step": 36625
    },
    {
      "epoch": 0.5702631804276584,
      "grad_norm": 4.6756086349487305,
      "learning_rate": 4.5236885662312974e-07,
      "loss": 0.7297,
      "step": 36630
    },
    {
      "epoch": 0.5703410214296278,
      "grad_norm": 7.739816665649414,
      "learning_rate": 4.5228691761852476e-07,
      "loss": 0.7092,
      "step": 36635
    },
    {
      "epoch": 0.5704188624315972,
      "grad_norm": 3.454681396484375,
      "learning_rate": 4.522049786139198e-07,
      "loss": 0.7875,
      "step": 36640
    },
    {
      "epoch": 0.5704967034335666,
      "grad_norm": 12.625911712646484,
      "learning_rate": 4.521230396093148e-07,
      "loss": 0.6998,
      "step": 36645
    },
    {
      "epoch": 0.570574544435536,
      "grad_norm": 2.6346638202667236,
      "learning_rate": 4.5204110060470983e-07,
      "loss": 0.6195,
      "step": 36650
    },
    {
      "epoch": 0.5706523854375054,
      "grad_norm": 10.461753845214844,
      "learning_rate": 4.519591616001049e-07,
      "loss": 0.8626,
      "step": 36655
    },
    {
      "epoch": 0.5707302264394747,
      "grad_norm": 2.7895283699035645,
      "learning_rate": 4.5187722259549987e-07,
      "loss": 0.7243,
      "step": 36660
    },
    {
      "epoch": 0.5708080674414441,
      "grad_norm": 3.8635172843933105,
      "learning_rate": 4.517952835908949e-07,
      "loss": 0.746,
      "step": 36665
    },
    {
      "epoch": 0.5708859084434135,
      "grad_norm": 2.7480740547180176,
      "learning_rate": 4.5171334458628997e-07,
      "loss": 0.7532,
      "step": 36670
    },
    {
      "epoch": 0.5709637494453829,
      "grad_norm": 3.483422040939331,
      "learning_rate": 4.51631405581685e-07,
      "loss": 0.8347,
      "step": 36675
    },
    {
      "epoch": 0.5710415904473523,
      "grad_norm": 4.101770877838135,
      "learning_rate": 4.5154946657707996e-07,
      "loss": 0.879,
      "step": 36680
    },
    {
      "epoch": 0.5711194314493216,
      "grad_norm": 5.707870960235596,
      "learning_rate": 4.5146752757247503e-07,
      "loss": 0.7417,
      "step": 36685
    },
    {
      "epoch": 0.571197272451291,
      "grad_norm": 4.441585063934326,
      "learning_rate": 4.5138558856787006e-07,
      "loss": 0.7012,
      "step": 36690
    },
    {
      "epoch": 0.5712751134532603,
      "grad_norm": 4.356373310089111,
      "learning_rate": 4.5130364956326513e-07,
      "loss": 0.8851,
      "step": 36695
    },
    {
      "epoch": 0.5713529544552297,
      "grad_norm": 4.200759410858154,
      "learning_rate": 4.512217105586601e-07,
      "loss": 0.7008,
      "step": 36700
    },
    {
      "epoch": 0.5714307954571991,
      "grad_norm": 3.489959239959717,
      "learning_rate": 4.511397715540551e-07,
      "loss": 0.7395,
      "step": 36705
    },
    {
      "epoch": 0.5715086364591685,
      "grad_norm": 4.713895797729492,
      "learning_rate": 4.510578325494502e-07,
      "loss": 0.8186,
      "step": 36710
    },
    {
      "epoch": 0.5715864774611379,
      "grad_norm": 2.7973928451538086,
      "learning_rate": 4.509758935448452e-07,
      "loss": 0.7477,
      "step": 36715
    },
    {
      "epoch": 0.5716643184631073,
      "grad_norm": 4.949597358703613,
      "learning_rate": 4.508939545402402e-07,
      "loss": 0.7247,
      "step": 36720
    },
    {
      "epoch": 0.5717421594650767,
      "grad_norm": 2.640023708343506,
      "learning_rate": 4.5081201553563526e-07,
      "loss": 0.6892,
      "step": 36725
    },
    {
      "epoch": 0.5718200004670461,
      "grad_norm": 2.732719898223877,
      "learning_rate": 4.507300765310303e-07,
      "loss": 0.731,
      "step": 36730
    },
    {
      "epoch": 0.5718978414690153,
      "grad_norm": 3.3502542972564697,
      "learning_rate": 4.506481375264253e-07,
      "loss": 0.7149,
      "step": 36735
    },
    {
      "epoch": 0.5719756824709847,
      "grad_norm": 4.023757457733154,
      "learning_rate": 4.5056619852182033e-07,
      "loss": 0.7765,
      "step": 36740
    },
    {
      "epoch": 0.5720535234729541,
      "grad_norm": 3.4681851863861084,
      "learning_rate": 4.5048425951721535e-07,
      "loss": 0.6557,
      "step": 36745
    },
    {
      "epoch": 0.5721313644749235,
      "grad_norm": 4.5781779289245605,
      "learning_rate": 4.504023205126104e-07,
      "loss": 0.8583,
      "step": 36750
    },
    {
      "epoch": 0.5722092054768929,
      "grad_norm": 3.9121389389038086,
      "learning_rate": 4.5032038150800545e-07,
      "loss": 0.7215,
      "step": 36755
    },
    {
      "epoch": 0.5722870464788623,
      "grad_norm": 3.6294186115264893,
      "learning_rate": 4.502384425034004e-07,
      "loss": 0.7242,
      "step": 36760
    },
    {
      "epoch": 0.5723648874808317,
      "grad_norm": 4.858184814453125,
      "learning_rate": 4.501565034987955e-07,
      "loss": 0.6989,
      "step": 36765
    },
    {
      "epoch": 0.572442728482801,
      "grad_norm": 4.430702209472656,
      "learning_rate": 4.500745644941905e-07,
      "loss": 0.7334,
      "step": 36770
    },
    {
      "epoch": 0.5725205694847704,
      "grad_norm": 7.22116231918335,
      "learning_rate": 4.4999262548958553e-07,
      "loss": 0.7647,
      "step": 36775
    },
    {
      "epoch": 0.5725984104867398,
      "grad_norm": 4.143237590789795,
      "learning_rate": 4.4991068648498056e-07,
      "loss": 0.7805,
      "step": 36780
    },
    {
      "epoch": 0.5726762514887092,
      "grad_norm": 3.705885887145996,
      "learning_rate": 4.498287474803756e-07,
      "loss": 0.8186,
      "step": 36785
    },
    {
      "epoch": 0.5727540924906785,
      "grad_norm": 5.623631477355957,
      "learning_rate": 4.497468084757706e-07,
      "loss": 0.6464,
      "step": 36790
    },
    {
      "epoch": 0.5728319334926479,
      "grad_norm": 3.5489749908447266,
      "learning_rate": 4.496648694711657e-07,
      "loss": 0.6619,
      "step": 36795
    },
    {
      "epoch": 0.5729097744946173,
      "grad_norm": 3.7704784870147705,
      "learning_rate": 4.4958293046656064e-07,
      "loss": 0.7794,
      "step": 36800
    },
    {
      "epoch": 0.5729876154965867,
      "grad_norm": 5.847343921661377,
      "learning_rate": 4.4950099146195567e-07,
      "loss": 0.8089,
      "step": 36805
    },
    {
      "epoch": 0.573065456498556,
      "grad_norm": 2.560455560684204,
      "learning_rate": 4.4941905245735074e-07,
      "loss": 0.7306,
      "step": 36810
    },
    {
      "epoch": 0.5731432975005254,
      "grad_norm": 2.626814842224121,
      "learning_rate": 4.4933711345274576e-07,
      "loss": 0.6882,
      "step": 36815
    },
    {
      "epoch": 0.5732211385024948,
      "grad_norm": 5.356245517730713,
      "learning_rate": 4.492551744481408e-07,
      "loss": 0.6765,
      "step": 36820
    },
    {
      "epoch": 0.5732989795044642,
      "grad_norm": 3.0126771926879883,
      "learning_rate": 4.491732354435358e-07,
      "loss": 0.894,
      "step": 36825
    },
    {
      "epoch": 0.5733768205064336,
      "grad_norm": 2.701789617538452,
      "learning_rate": 4.4909129643893083e-07,
      "loss": 0.8137,
      "step": 36830
    },
    {
      "epoch": 0.573454661508403,
      "grad_norm": 3.8949315547943115,
      "learning_rate": 4.490093574343259e-07,
      "loss": 0.7897,
      "step": 36835
    },
    {
      "epoch": 0.5735325025103724,
      "grad_norm": 5.79347562789917,
      "learning_rate": 4.4892741842972087e-07,
      "loss": 0.8444,
      "step": 36840
    },
    {
      "epoch": 0.5736103435123416,
      "grad_norm": 4.316470623016357,
      "learning_rate": 4.488454794251159e-07,
      "loss": 0.5949,
      "step": 36845
    },
    {
      "epoch": 0.573688184514311,
      "grad_norm": 4.749272346496582,
      "learning_rate": 4.4876354042051097e-07,
      "loss": 0.8378,
      "step": 36850
    },
    {
      "epoch": 0.5737660255162804,
      "grad_norm": 3.4906039237976074,
      "learning_rate": 4.48681601415906e-07,
      "loss": 0.8311,
      "step": 36855
    },
    {
      "epoch": 0.5738438665182498,
      "grad_norm": 5.242584228515625,
      "learning_rate": 4.4859966241130096e-07,
      "loss": 0.7472,
      "step": 36860
    },
    {
      "epoch": 0.5739217075202192,
      "grad_norm": 3.814054250717163,
      "learning_rate": 4.4851772340669603e-07,
      "loss": 0.7369,
      "step": 36865
    },
    {
      "epoch": 0.5739995485221886,
      "grad_norm": 4.658022880554199,
      "learning_rate": 4.4843578440209106e-07,
      "loss": 0.8094,
      "step": 36870
    },
    {
      "epoch": 0.574077389524158,
      "grad_norm": 5.94177770614624,
      "learning_rate": 4.4835384539748613e-07,
      "loss": 0.7985,
      "step": 36875
    },
    {
      "epoch": 0.5741552305261274,
      "grad_norm": 3.4186336994171143,
      "learning_rate": 4.482719063928811e-07,
      "loss": 0.673,
      "step": 36880
    },
    {
      "epoch": 0.5742330715280967,
      "grad_norm": 3.850654363632202,
      "learning_rate": 4.481899673882761e-07,
      "loss": 0.7093,
      "step": 36885
    },
    {
      "epoch": 0.574310912530066,
      "grad_norm": 3.9102721214294434,
      "learning_rate": 4.481080283836712e-07,
      "loss": 0.8479,
      "step": 36890
    },
    {
      "epoch": 0.5743887535320354,
      "grad_norm": 4.528449535369873,
      "learning_rate": 4.480260893790662e-07,
      "loss": 0.7836,
      "step": 36895
    },
    {
      "epoch": 0.5744665945340048,
      "grad_norm": 3.059283971786499,
      "learning_rate": 4.479441503744612e-07,
      "loss": 0.8496,
      "step": 36900
    },
    {
      "epoch": 0.5745444355359742,
      "grad_norm": 4.690300464630127,
      "learning_rate": 4.4786221136985626e-07,
      "loss": 0.8301,
      "step": 36905
    },
    {
      "epoch": 0.5746222765379436,
      "grad_norm": 3.0751466751098633,
      "learning_rate": 4.477802723652513e-07,
      "loss": 0.7745,
      "step": 36910
    },
    {
      "epoch": 0.574700117539913,
      "grad_norm": 4.539811134338379,
      "learning_rate": 4.476983333606463e-07,
      "loss": 0.8658,
      "step": 36915
    },
    {
      "epoch": 0.5747779585418824,
      "grad_norm": 5.584836959838867,
      "learning_rate": 4.4761639435604133e-07,
      "loss": 0.6888,
      "step": 36920
    },
    {
      "epoch": 0.5748557995438517,
      "grad_norm": 3.03218150138855,
      "learning_rate": 4.4753445535143635e-07,
      "loss": 0.7256,
      "step": 36925
    },
    {
      "epoch": 0.5749336405458211,
      "grad_norm": 4.958030700683594,
      "learning_rate": 4.4745251634683137e-07,
      "loss": 0.7514,
      "step": 36930
    },
    {
      "epoch": 0.5750114815477905,
      "grad_norm": 3.76651668548584,
      "learning_rate": 4.4737057734222645e-07,
      "loss": 0.7468,
      "step": 36935
    },
    {
      "epoch": 0.5750893225497599,
      "grad_norm": 3.004082679748535,
      "learning_rate": 4.472886383376214e-07,
      "loss": 0.713,
      "step": 36940
    },
    {
      "epoch": 0.5751671635517293,
      "grad_norm": 5.6487016677856445,
      "learning_rate": 4.472066993330165e-07,
      "loss": 0.7217,
      "step": 36945
    },
    {
      "epoch": 0.5752450045536986,
      "grad_norm": 7.2317094802856445,
      "learning_rate": 4.471247603284115e-07,
      "loss": 0.747,
      "step": 36950
    },
    {
      "epoch": 0.575322845555668,
      "grad_norm": 6.422186374664307,
      "learning_rate": 4.4704282132380654e-07,
      "loss": 0.7238,
      "step": 36955
    },
    {
      "epoch": 0.5754006865576373,
      "grad_norm": 3.3757214546203613,
      "learning_rate": 4.469608823192016e-07,
      "loss": 0.7205,
      "step": 36960
    },
    {
      "epoch": 0.5754785275596067,
      "grad_norm": 5.187408447265625,
      "learning_rate": 4.468789433145966e-07,
      "loss": 0.7261,
      "step": 36965
    },
    {
      "epoch": 0.5755563685615761,
      "grad_norm": 3.340165138244629,
      "learning_rate": 4.467970043099916e-07,
      "loss": 0.718,
      "step": 36970
    },
    {
      "epoch": 0.5756342095635455,
      "grad_norm": 3.0176074504852295,
      "learning_rate": 4.467150653053867e-07,
      "loss": 0.7838,
      "step": 36975
    },
    {
      "epoch": 0.5757120505655149,
      "grad_norm": 2.544583559036255,
      "learning_rate": 4.466331263007817e-07,
      "loss": 0.7147,
      "step": 36980
    },
    {
      "epoch": 0.5757898915674843,
      "grad_norm": 4.563441276550293,
      "learning_rate": 4.4655118729617667e-07,
      "loss": 0.7485,
      "step": 36985
    },
    {
      "epoch": 0.5758677325694537,
      "grad_norm": 2.393890857696533,
      "learning_rate": 4.4646924829157174e-07,
      "loss": 0.7681,
      "step": 36990
    },
    {
      "epoch": 0.5759455735714231,
      "grad_norm": 3.0192654132843018,
      "learning_rate": 4.4638730928696676e-07,
      "loss": 0.7312,
      "step": 36995
    },
    {
      "epoch": 0.5760234145733923,
      "grad_norm": 3.4682118892669678,
      "learning_rate": 4.4630537028236184e-07,
      "loss": 0.7897,
      "step": 37000
    },
    {
      "epoch": 0.5761012555753617,
      "grad_norm": 3.0685019493103027,
      "learning_rate": 4.462234312777568e-07,
      "loss": 0.7657,
      "step": 37005
    },
    {
      "epoch": 0.5761790965773311,
      "grad_norm": 3.9035444259643555,
      "learning_rate": 4.4614149227315183e-07,
      "loss": 0.7689,
      "step": 37010
    },
    {
      "epoch": 0.5762569375793005,
      "grad_norm": 3.159846067428589,
      "learning_rate": 4.460595532685469e-07,
      "loss": 0.7192,
      "step": 37015
    },
    {
      "epoch": 0.5763347785812699,
      "grad_norm": 4.263913154602051,
      "learning_rate": 4.459776142639419e-07,
      "loss": 0.7385,
      "step": 37020
    },
    {
      "epoch": 0.5764126195832393,
      "grad_norm": 4.782285690307617,
      "learning_rate": 4.458956752593369e-07,
      "loss": 0.7781,
      "step": 37025
    },
    {
      "epoch": 0.5764904605852087,
      "grad_norm": 2.932845115661621,
      "learning_rate": 4.4581373625473197e-07,
      "loss": 0.8749,
      "step": 37030
    },
    {
      "epoch": 0.576568301587178,
      "grad_norm": 4.05242919921875,
      "learning_rate": 4.45731797250127e-07,
      "loss": 0.8303,
      "step": 37035
    },
    {
      "epoch": 0.5766461425891474,
      "grad_norm": 5.182920455932617,
      "learning_rate": 4.45649858245522e-07,
      "loss": 0.7782,
      "step": 37040
    },
    {
      "epoch": 0.5767239835911168,
      "grad_norm": 4.171004772186279,
      "learning_rate": 4.4556791924091704e-07,
      "loss": 0.7386,
      "step": 37045
    },
    {
      "epoch": 0.5768018245930862,
      "grad_norm": 7.244475364685059,
      "learning_rate": 4.4548598023631206e-07,
      "loss": 0.8995,
      "step": 37050
    },
    {
      "epoch": 0.5768796655950555,
      "grad_norm": 4.518347263336182,
      "learning_rate": 4.4540404123170713e-07,
      "loss": 0.7628,
      "step": 37055
    },
    {
      "epoch": 0.5769575065970249,
      "grad_norm": 4.655910968780518,
      "learning_rate": 4.4532210222710215e-07,
      "loss": 0.6865,
      "step": 37060
    },
    {
      "epoch": 0.5770353475989943,
      "grad_norm": 3.5219624042510986,
      "learning_rate": 4.452401632224971e-07,
      "loss": 0.5456,
      "step": 37065
    },
    {
      "epoch": 0.5771131886009637,
      "grad_norm": 3.5598604679107666,
      "learning_rate": 4.451582242178922e-07,
      "loss": 0.7421,
      "step": 37070
    },
    {
      "epoch": 0.577191029602933,
      "grad_norm": 5.40380859375,
      "learning_rate": 4.450762852132872e-07,
      "loss": 0.6198,
      "step": 37075
    },
    {
      "epoch": 0.5772688706049024,
      "grad_norm": 3.369943618774414,
      "learning_rate": 4.4499434620868224e-07,
      "loss": 0.6487,
      "step": 37080
    },
    {
      "epoch": 0.5773467116068718,
      "grad_norm": 6.176918983459473,
      "learning_rate": 4.4491240720407726e-07,
      "loss": 0.7256,
      "step": 37085
    },
    {
      "epoch": 0.5774245526088412,
      "grad_norm": 4.338992595672607,
      "learning_rate": 4.448304681994723e-07,
      "loss": 0.7784,
      "step": 37090
    },
    {
      "epoch": 0.5775023936108106,
      "grad_norm": 3.8260715007781982,
      "learning_rate": 4.447485291948673e-07,
      "loss": 0.7289,
      "step": 37095
    },
    {
      "epoch": 0.57758023461278,
      "grad_norm": 3.6482412815093994,
      "learning_rate": 4.446665901902624e-07,
      "loss": 0.8233,
      "step": 37100
    },
    {
      "epoch": 0.5776580756147494,
      "grad_norm": 2.651073455810547,
      "learning_rate": 4.4458465118565735e-07,
      "loss": 0.8083,
      "step": 37105
    },
    {
      "epoch": 0.5777359166167186,
      "grad_norm": 2.9311158657073975,
      "learning_rate": 4.4450271218105237e-07,
      "loss": 0.6595,
      "step": 37110
    },
    {
      "epoch": 0.577813757618688,
      "grad_norm": 4.731007099151611,
      "learning_rate": 4.4442077317644745e-07,
      "loss": 0.6787,
      "step": 37115
    },
    {
      "epoch": 0.5778915986206574,
      "grad_norm": 4.16535758972168,
      "learning_rate": 4.4433883417184247e-07,
      "loss": 0.7654,
      "step": 37120
    },
    {
      "epoch": 0.5779694396226268,
      "grad_norm": 5.9839091300964355,
      "learning_rate": 4.442568951672375e-07,
      "loss": 0.8616,
      "step": 37125
    },
    {
      "epoch": 0.5780472806245962,
      "grad_norm": 3.0118939876556396,
      "learning_rate": 4.441749561626325e-07,
      "loss": 0.801,
      "step": 37130
    },
    {
      "epoch": 0.5781251216265656,
      "grad_norm": 14.881279945373535,
      "learning_rate": 4.4409301715802754e-07,
      "loss": 0.8181,
      "step": 37135
    },
    {
      "epoch": 0.578202962628535,
      "grad_norm": 7.762650489807129,
      "learning_rate": 4.440110781534226e-07,
      "loss": 0.7829,
      "step": 37140
    },
    {
      "epoch": 0.5782808036305044,
      "grad_norm": 8.179519653320312,
      "learning_rate": 4.439291391488176e-07,
      "loss": 0.761,
      "step": 37145
    },
    {
      "epoch": 0.5783586446324737,
      "grad_norm": 3.4446332454681396,
      "learning_rate": 4.438472001442126e-07,
      "loss": 0.819,
      "step": 37150
    },
    {
      "epoch": 0.578436485634443,
      "grad_norm": 6.1664533615112305,
      "learning_rate": 4.437652611396077e-07,
      "loss": 0.7547,
      "step": 37155
    },
    {
      "epoch": 0.5785143266364124,
      "grad_norm": 5.532240390777588,
      "learning_rate": 4.436833221350027e-07,
      "loss": 0.8252,
      "step": 37160
    },
    {
      "epoch": 0.5785921676383818,
      "grad_norm": 4.026022434234619,
      "learning_rate": 4.4360138313039767e-07,
      "loss": 0.756,
      "step": 37165
    },
    {
      "epoch": 0.5786700086403512,
      "grad_norm": 3.149070978164673,
      "learning_rate": 4.4351944412579274e-07,
      "loss": 0.6621,
      "step": 37170
    },
    {
      "epoch": 0.5787478496423206,
      "grad_norm": 5.486191749572754,
      "learning_rate": 4.4343750512118776e-07,
      "loss": 0.6947,
      "step": 37175
    },
    {
      "epoch": 0.57882569064429,
      "grad_norm": 4.807506084442139,
      "learning_rate": 4.4335556611658284e-07,
      "loss": 0.692,
      "step": 37180
    },
    {
      "epoch": 0.5789035316462594,
      "grad_norm": 3.8123767375946045,
      "learning_rate": 4.432736271119778e-07,
      "loss": 0.8558,
      "step": 37185
    },
    {
      "epoch": 0.5789813726482287,
      "grad_norm": 3.0400938987731934,
      "learning_rate": 4.4319168810737283e-07,
      "loss": 0.706,
      "step": 37190
    },
    {
      "epoch": 0.5790592136501981,
      "grad_norm": 6.030972957611084,
      "learning_rate": 4.431097491027679e-07,
      "loss": 0.7114,
      "step": 37195
    },
    {
      "epoch": 0.5791370546521675,
      "grad_norm": 4.024885654449463,
      "learning_rate": 4.4302781009816293e-07,
      "loss": 0.6627,
      "step": 37200
    },
    {
      "epoch": 0.5792148956541369,
      "grad_norm": 3.673020124435425,
      "learning_rate": 4.429458710935579e-07,
      "loss": 0.7175,
      "step": 37205
    },
    {
      "epoch": 0.5792927366561063,
      "grad_norm": 8.647665023803711,
      "learning_rate": 4.4286393208895297e-07,
      "loss": 0.7032,
      "step": 37210
    },
    {
      "epoch": 0.5793705776580756,
      "grad_norm": 3.456531047821045,
      "learning_rate": 4.42781993084348e-07,
      "loss": 0.8632,
      "step": 37215
    },
    {
      "epoch": 0.579448418660045,
      "grad_norm": 5.602091312408447,
      "learning_rate": 4.42700054079743e-07,
      "loss": 0.6625,
      "step": 37220
    },
    {
      "epoch": 0.5795262596620143,
      "grad_norm": 2.7934651374816895,
      "learning_rate": 4.4261811507513804e-07,
      "loss": 0.7154,
      "step": 37225
    },
    {
      "epoch": 0.5796041006639837,
      "grad_norm": 5.680755138397217,
      "learning_rate": 4.4253617607053306e-07,
      "loss": 0.6783,
      "step": 37230
    },
    {
      "epoch": 0.5796819416659531,
      "grad_norm": 3.896085262298584,
      "learning_rate": 4.424542370659281e-07,
      "loss": 0.6727,
      "step": 37235
    },
    {
      "epoch": 0.5797597826679225,
      "grad_norm": 2.582078695297241,
      "learning_rate": 4.4237229806132316e-07,
      "loss": 0.8256,
      "step": 37240
    },
    {
      "epoch": 0.5798376236698919,
      "grad_norm": 4.293142318725586,
      "learning_rate": 4.422903590567181e-07,
      "loss": 0.8141,
      "step": 37245
    },
    {
      "epoch": 0.5799154646718613,
      "grad_norm": 8.875615119934082,
      "learning_rate": 4.422084200521132e-07,
      "loss": 0.8712,
      "step": 37250
    },
    {
      "epoch": 0.5799933056738307,
      "grad_norm": 3.492053985595703,
      "learning_rate": 4.421264810475082e-07,
      "loss": 0.7382,
      "step": 37255
    },
    {
      "epoch": 0.5800711466758001,
      "grad_norm": 3.994518756866455,
      "learning_rate": 4.4204454204290324e-07,
      "loss": 0.7867,
      "step": 37260
    },
    {
      "epoch": 0.5801489876777693,
      "grad_norm": 4.340789318084717,
      "learning_rate": 4.419626030382983e-07,
      "loss": 0.7568,
      "step": 37265
    },
    {
      "epoch": 0.5802268286797387,
      "grad_norm": 4.114905834197998,
      "learning_rate": 4.418806640336933e-07,
      "loss": 0.7418,
      "step": 37270
    },
    {
      "epoch": 0.5803046696817081,
      "grad_norm": 3.706138849258423,
      "learning_rate": 4.417987250290883e-07,
      "loss": 0.8484,
      "step": 37275
    },
    {
      "epoch": 0.5803825106836775,
      "grad_norm": 3.9241416454315186,
      "learning_rate": 4.417167860244834e-07,
      "loss": 0.7377,
      "step": 37280
    },
    {
      "epoch": 0.5804603516856469,
      "grad_norm": 7.387014865875244,
      "learning_rate": 4.416348470198784e-07,
      "loss": 0.7096,
      "step": 37285
    },
    {
      "epoch": 0.5805381926876163,
      "grad_norm": 4.157270908355713,
      "learning_rate": 4.415529080152734e-07,
      "loss": 0.7448,
      "step": 37290
    },
    {
      "epoch": 0.5806160336895857,
      "grad_norm": 3.069624423980713,
      "learning_rate": 4.4147096901066845e-07,
      "loss": 0.7976,
      "step": 37295
    },
    {
      "epoch": 0.580693874691555,
      "grad_norm": 4.58493709564209,
      "learning_rate": 4.4138903000606347e-07,
      "loss": 0.6876,
      "step": 37300
    },
    {
      "epoch": 0.5807717156935244,
      "grad_norm": 4.756598472595215,
      "learning_rate": 4.4130709100145855e-07,
      "loss": 0.6747,
      "step": 37305
    },
    {
      "epoch": 0.5808495566954938,
      "grad_norm": 6.113785266876221,
      "learning_rate": 4.412251519968535e-07,
      "loss": 0.8857,
      "step": 37310
    },
    {
      "epoch": 0.5809273976974632,
      "grad_norm": 3.462789535522461,
      "learning_rate": 4.4114321299224854e-07,
      "loss": 0.69,
      "step": 37315
    },
    {
      "epoch": 0.5810052386994325,
      "grad_norm": 3.6441822052001953,
      "learning_rate": 4.410612739876436e-07,
      "loss": 0.8032,
      "step": 37320
    },
    {
      "epoch": 0.5810830797014019,
      "grad_norm": 3.5212314128875732,
      "learning_rate": 4.4097933498303863e-07,
      "loss": 0.7083,
      "step": 37325
    },
    {
      "epoch": 0.5811609207033713,
      "grad_norm": 4.167048454284668,
      "learning_rate": 4.408973959784336e-07,
      "loss": 0.795,
      "step": 37330
    },
    {
      "epoch": 0.5812387617053407,
      "grad_norm": 6.5197577476501465,
      "learning_rate": 4.408154569738287e-07,
      "loss": 0.8436,
      "step": 37335
    },
    {
      "epoch": 0.58131660270731,
      "grad_norm": 6.018814563751221,
      "learning_rate": 4.407335179692237e-07,
      "loss": 0.7592,
      "step": 37340
    },
    {
      "epoch": 0.5813944437092794,
      "grad_norm": 4.134605884552002,
      "learning_rate": 4.406515789646187e-07,
      "loss": 0.7153,
      "step": 37345
    },
    {
      "epoch": 0.5814722847112488,
      "grad_norm": 2.6605520248413086,
      "learning_rate": 4.4056963996001374e-07,
      "loss": 0.5985,
      "step": 37350
    },
    {
      "epoch": 0.5815501257132182,
      "grad_norm": 7.417990684509277,
      "learning_rate": 4.4048770095540877e-07,
      "loss": 0.6614,
      "step": 37355
    },
    {
      "epoch": 0.5816279667151876,
      "grad_norm": 4.777806758880615,
      "learning_rate": 4.404057619508038e-07,
      "loss": 0.7471,
      "step": 37360
    },
    {
      "epoch": 0.581705807717157,
      "grad_norm": 3.4674954414367676,
      "learning_rate": 4.4032382294619886e-07,
      "loss": 0.7947,
      "step": 37365
    },
    {
      "epoch": 0.5817836487191264,
      "grad_norm": 5.817410469055176,
      "learning_rate": 4.4024188394159383e-07,
      "loss": 0.8502,
      "step": 37370
    },
    {
      "epoch": 0.5818614897210956,
      "grad_norm": 3.474041700363159,
      "learning_rate": 4.401599449369889e-07,
      "loss": 0.7065,
      "step": 37375
    },
    {
      "epoch": 0.581939330723065,
      "grad_norm": 3.893080949783325,
      "learning_rate": 4.4007800593238393e-07,
      "loss": 0.7766,
      "step": 37380
    },
    {
      "epoch": 0.5820171717250344,
      "grad_norm": 3.25528621673584,
      "learning_rate": 4.3999606692777895e-07,
      "loss": 0.6987,
      "step": 37385
    },
    {
      "epoch": 0.5820950127270038,
      "grad_norm": 3.759158134460449,
      "learning_rate": 4.3991412792317397e-07,
      "loss": 0.7775,
      "step": 37390
    },
    {
      "epoch": 0.5821728537289732,
      "grad_norm": 4.0417304039001465,
      "learning_rate": 4.39832188918569e-07,
      "loss": 0.7202,
      "step": 37395
    },
    {
      "epoch": 0.5822506947309426,
      "grad_norm": 4.7286882400512695,
      "learning_rate": 4.39750249913964e-07,
      "loss": 0.6888,
      "step": 37400
    },
    {
      "epoch": 0.582328535732912,
      "grad_norm": 5.395811557769775,
      "learning_rate": 4.396683109093591e-07,
      "loss": 0.7301,
      "step": 37405
    },
    {
      "epoch": 0.5824063767348814,
      "grad_norm": 3.345646619796753,
      "learning_rate": 4.3958637190475406e-07,
      "loss": 0.9037,
      "step": 37410
    },
    {
      "epoch": 0.5824842177368507,
      "grad_norm": 3.7008140087127686,
      "learning_rate": 4.395044329001491e-07,
      "loss": 0.8375,
      "step": 37415
    },
    {
      "epoch": 0.58256205873882,
      "grad_norm": 2.7580933570861816,
      "learning_rate": 4.3942249389554416e-07,
      "loss": 0.7805,
      "step": 37420
    },
    {
      "epoch": 0.5826398997407894,
      "grad_norm": 4.5180816650390625,
      "learning_rate": 4.393405548909392e-07,
      "loss": 0.6909,
      "step": 37425
    },
    {
      "epoch": 0.5827177407427588,
      "grad_norm": 2.975207805633545,
      "learning_rate": 4.3925861588633415e-07,
      "loss": 0.7206,
      "step": 37430
    },
    {
      "epoch": 0.5827955817447282,
      "grad_norm": 3.548135280609131,
      "learning_rate": 4.391766768817292e-07,
      "loss": 0.7538,
      "step": 37435
    },
    {
      "epoch": 0.5828734227466976,
      "grad_norm": 5.2094502449035645,
      "learning_rate": 4.3909473787712424e-07,
      "loss": 0.7366,
      "step": 37440
    },
    {
      "epoch": 0.582951263748667,
      "grad_norm": 3.7388124465942383,
      "learning_rate": 4.390127988725193e-07,
      "loss": 0.722,
      "step": 37445
    },
    {
      "epoch": 0.5830291047506363,
      "grad_norm": 3.6782116889953613,
      "learning_rate": 4.389308598679143e-07,
      "loss": 0.771,
      "step": 37450
    },
    {
      "epoch": 0.5831069457526057,
      "grad_norm": 3.524994373321533,
      "learning_rate": 4.388489208633093e-07,
      "loss": 0.7295,
      "step": 37455
    },
    {
      "epoch": 0.5831847867545751,
      "grad_norm": 4.152987480163574,
      "learning_rate": 4.387669818587044e-07,
      "loss": 0.7561,
      "step": 37460
    },
    {
      "epoch": 0.5832626277565445,
      "grad_norm": 5.414799213409424,
      "learning_rate": 4.386850428540994e-07,
      "loss": 0.7211,
      "step": 37465
    },
    {
      "epoch": 0.5833404687585139,
      "grad_norm": 3.277332305908203,
      "learning_rate": 4.386031038494944e-07,
      "loss": 0.7199,
      "step": 37470
    },
    {
      "epoch": 0.5834183097604833,
      "grad_norm": 4.118561267852783,
      "learning_rate": 4.3852116484488945e-07,
      "loss": 0.7523,
      "step": 37475
    },
    {
      "epoch": 0.5834961507624526,
      "grad_norm": 3.351937770843506,
      "learning_rate": 4.3843922584028447e-07,
      "loss": 0.7545,
      "step": 37480
    },
    {
      "epoch": 0.583573991764422,
      "grad_norm": 3.4315693378448486,
      "learning_rate": 4.383572868356795e-07,
      "loss": 0.7757,
      "step": 37485
    },
    {
      "epoch": 0.5836518327663913,
      "grad_norm": 5.651627063751221,
      "learning_rate": 4.382753478310745e-07,
      "loss": 0.6968,
      "step": 37490
    },
    {
      "epoch": 0.5837296737683607,
      "grad_norm": 6.561192035675049,
      "learning_rate": 4.3819340882646954e-07,
      "loss": 0.7423,
      "step": 37495
    },
    {
      "epoch": 0.5838075147703301,
      "grad_norm": 3.9459850788116455,
      "learning_rate": 4.381114698218646e-07,
      "loss": 0.8208,
      "step": 37500
    },
    {
      "epoch": 0.5838853557722995,
      "grad_norm": 4.725958347320557,
      "learning_rate": 4.3802953081725963e-07,
      "loss": 0.6995,
      "step": 37505
    },
    {
      "epoch": 0.5839631967742689,
      "grad_norm": 4.678044319152832,
      "learning_rate": 4.379475918126546e-07,
      "loss": 0.7702,
      "step": 37510
    },
    {
      "epoch": 0.5840410377762383,
      "grad_norm": 4.7829766273498535,
      "learning_rate": 4.378656528080497e-07,
      "loss": 0.7268,
      "step": 37515
    },
    {
      "epoch": 0.5841188787782077,
      "grad_norm": 5.207371711730957,
      "learning_rate": 4.377837138034447e-07,
      "loss": 0.6843,
      "step": 37520
    },
    {
      "epoch": 0.5841967197801771,
      "grad_norm": 3.146820306777954,
      "learning_rate": 4.377017747988397e-07,
      "loss": 0.7551,
      "step": 37525
    },
    {
      "epoch": 0.5842745607821463,
      "grad_norm": 8.804973602294922,
      "learning_rate": 4.3761983579423474e-07,
      "loss": 0.7418,
      "step": 37530
    },
    {
      "epoch": 0.5843524017841157,
      "grad_norm": 2.764760732650757,
      "learning_rate": 4.3753789678962977e-07,
      "loss": 0.8045,
      "step": 37535
    },
    {
      "epoch": 0.5844302427860851,
      "grad_norm": 5.973789215087891,
      "learning_rate": 4.374559577850248e-07,
      "loss": 0.6681,
      "step": 37540
    },
    {
      "epoch": 0.5845080837880545,
      "grad_norm": 3.7041358947753906,
      "learning_rate": 4.3737401878041986e-07,
      "loss": 0.808,
      "step": 37545
    },
    {
      "epoch": 0.5845859247900239,
      "grad_norm": 3.4011168479919434,
      "learning_rate": 4.3729207977581483e-07,
      "loss": 0.7845,
      "step": 37550
    },
    {
      "epoch": 0.5846637657919933,
      "grad_norm": 3.0675768852233887,
      "learning_rate": 4.3721014077120985e-07,
      "loss": 0.7277,
      "step": 37555
    },
    {
      "epoch": 0.5847416067939627,
      "grad_norm": 3.601224422454834,
      "learning_rate": 4.3712820176660493e-07,
      "loss": 0.7585,
      "step": 37560
    },
    {
      "epoch": 0.584819447795932,
      "grad_norm": 4.33906888961792,
      "learning_rate": 4.3704626276199995e-07,
      "loss": 0.7324,
      "step": 37565
    },
    {
      "epoch": 0.5848972887979014,
      "grad_norm": 3.2355477809906006,
      "learning_rate": 4.3696432375739497e-07,
      "loss": 0.8477,
      "step": 37570
    },
    {
      "epoch": 0.5849751297998708,
      "grad_norm": 5.205055236816406,
      "learning_rate": 4.3688238475279e-07,
      "loss": 0.7604,
      "step": 37575
    },
    {
      "epoch": 0.5850529708018402,
      "grad_norm": 3.7320597171783447,
      "learning_rate": 4.36800445748185e-07,
      "loss": 0.6649,
      "step": 37580
    },
    {
      "epoch": 0.5851308118038095,
      "grad_norm": 8.569360733032227,
      "learning_rate": 4.367185067435801e-07,
      "loss": 0.6788,
      "step": 37585
    },
    {
      "epoch": 0.5852086528057789,
      "grad_norm": 2.6943612098693848,
      "learning_rate": 4.366365677389751e-07,
      "loss": 0.7382,
      "step": 37590
    },
    {
      "epoch": 0.5852864938077483,
      "grad_norm": 3.3973402976989746,
      "learning_rate": 4.365546287343701e-07,
      "loss": 0.6629,
      "step": 37595
    },
    {
      "epoch": 0.5853643348097177,
      "grad_norm": 7.558845520019531,
      "learning_rate": 4.3647268972976516e-07,
      "loss": 0.7245,
      "step": 37600
    },
    {
      "epoch": 0.585442175811687,
      "grad_norm": 6.471187591552734,
      "learning_rate": 4.363907507251602e-07,
      "loss": 0.6738,
      "step": 37605
    },
    {
      "epoch": 0.5855200168136564,
      "grad_norm": 4.115442752838135,
      "learning_rate": 4.363088117205552e-07,
      "loss": 0.7249,
      "step": 37610
    },
    {
      "epoch": 0.5855978578156258,
      "grad_norm": 3.6548688411712646,
      "learning_rate": 4.362268727159502e-07,
      "loss": 0.8291,
      "step": 37615
    },
    {
      "epoch": 0.5856756988175952,
      "grad_norm": 2.96681547164917,
      "learning_rate": 4.3614493371134524e-07,
      "loss": 0.7309,
      "step": 37620
    },
    {
      "epoch": 0.5857535398195646,
      "grad_norm": 6.621280193328857,
      "learning_rate": 4.360629947067403e-07,
      "loss": 0.6762,
      "step": 37625
    },
    {
      "epoch": 0.585831380821534,
      "grad_norm": 4.576493263244629,
      "learning_rate": 4.3598105570213534e-07,
      "loss": 0.7641,
      "step": 37630
    },
    {
      "epoch": 0.5859092218235034,
      "grad_norm": 2.862795114517212,
      "learning_rate": 4.358991166975303e-07,
      "loss": 0.671,
      "step": 37635
    },
    {
      "epoch": 0.5859870628254726,
      "grad_norm": 2.21162486076355,
      "learning_rate": 4.358171776929254e-07,
      "loss": 0.7447,
      "step": 37640
    },
    {
      "epoch": 0.586064903827442,
      "grad_norm": 3.079237222671509,
      "learning_rate": 4.357352386883204e-07,
      "loss": 0.8045,
      "step": 37645
    },
    {
      "epoch": 0.5861427448294114,
      "grad_norm": 9.33649730682373,
      "learning_rate": 4.3565329968371543e-07,
      "loss": 0.7791,
      "step": 37650
    },
    {
      "epoch": 0.5862205858313808,
      "grad_norm": 8.775187492370605,
      "learning_rate": 4.3557136067911045e-07,
      "loss": 0.8169,
      "step": 37655
    },
    {
      "epoch": 0.5862984268333502,
      "grad_norm": 4.090290546417236,
      "learning_rate": 4.3548942167450547e-07,
      "loss": 0.6684,
      "step": 37660
    },
    {
      "epoch": 0.5863762678353196,
      "grad_norm": 2.943504571914673,
      "learning_rate": 4.354074826699005e-07,
      "loss": 0.6921,
      "step": 37665
    },
    {
      "epoch": 0.586454108837289,
      "grad_norm": 3.5116372108459473,
      "learning_rate": 4.3532554366529557e-07,
      "loss": 0.7923,
      "step": 37670
    },
    {
      "epoch": 0.5865319498392584,
      "grad_norm": 3.322075843811035,
      "learning_rate": 4.3524360466069054e-07,
      "loss": 0.626,
      "step": 37675
    },
    {
      "epoch": 0.5866097908412277,
      "grad_norm": 3.0946972370147705,
      "learning_rate": 4.3516166565608556e-07,
      "loss": 0.6784,
      "step": 37680
    },
    {
      "epoch": 0.586687631843197,
      "grad_norm": 5.177741050720215,
      "learning_rate": 4.3507972665148064e-07,
      "loss": 0.7429,
      "step": 37685
    },
    {
      "epoch": 0.5867654728451664,
      "grad_norm": 3.523179531097412,
      "learning_rate": 4.3499778764687566e-07,
      "loss": 0.818,
      "step": 37690
    },
    {
      "epoch": 0.5868433138471358,
      "grad_norm": 2.9222826957702637,
      "learning_rate": 4.349158486422707e-07,
      "loss": 0.6916,
      "step": 37695
    },
    {
      "epoch": 0.5869211548491052,
      "grad_norm": 3.770170211791992,
      "learning_rate": 4.348339096376657e-07,
      "loss": 0.6926,
      "step": 37700
    },
    {
      "epoch": 0.5869989958510746,
      "grad_norm": 4.050478458404541,
      "learning_rate": 4.347519706330607e-07,
      "loss": 0.7084,
      "step": 37705
    },
    {
      "epoch": 0.587076836853044,
      "grad_norm": 7.340362548828125,
      "learning_rate": 4.346700316284558e-07,
      "loss": 0.7297,
      "step": 37710
    },
    {
      "epoch": 0.5871546778550133,
      "grad_norm": 5.613221645355225,
      "learning_rate": 4.3458809262385077e-07,
      "loss": 0.8206,
      "step": 37715
    },
    {
      "epoch": 0.5872325188569827,
      "grad_norm": 4.700793743133545,
      "learning_rate": 4.345061536192458e-07,
      "loss": 0.7743,
      "step": 37720
    },
    {
      "epoch": 0.5873103598589521,
      "grad_norm": 3.6294491291046143,
      "learning_rate": 4.3442421461464086e-07,
      "loss": 0.7854,
      "step": 37725
    },
    {
      "epoch": 0.5873882008609215,
      "grad_norm": 3.3486297130584717,
      "learning_rate": 4.343422756100359e-07,
      "loss": 0.7586,
      "step": 37730
    },
    {
      "epoch": 0.5874660418628909,
      "grad_norm": 8.278900146484375,
      "learning_rate": 4.3426033660543085e-07,
      "loss": 0.7377,
      "step": 37735
    },
    {
      "epoch": 0.5875438828648603,
      "grad_norm": 2.991431951522827,
      "learning_rate": 4.3417839760082593e-07,
      "loss": 0.6444,
      "step": 37740
    },
    {
      "epoch": 0.5876217238668296,
      "grad_norm": 3.3825161457061768,
      "learning_rate": 4.3409645859622095e-07,
      "loss": 0.7549,
      "step": 37745
    },
    {
      "epoch": 0.587699564868799,
      "grad_norm": 4.206120491027832,
      "learning_rate": 4.34014519591616e-07,
      "loss": 0.7704,
      "step": 37750
    },
    {
      "epoch": 0.5877774058707683,
      "grad_norm": 5.141931056976318,
      "learning_rate": 4.33932580587011e-07,
      "loss": 0.587,
      "step": 37755
    },
    {
      "epoch": 0.5878552468727377,
      "grad_norm": 3.4665002822875977,
      "learning_rate": 4.33850641582406e-07,
      "loss": 0.7095,
      "step": 37760
    },
    {
      "epoch": 0.5879330878747071,
      "grad_norm": 5.052914619445801,
      "learning_rate": 4.337687025778011e-07,
      "loss": 0.7399,
      "step": 37765
    },
    {
      "epoch": 0.5880109288766765,
      "grad_norm": 3.0287275314331055,
      "learning_rate": 4.336867635731961e-07,
      "loss": 0.8319,
      "step": 37770
    },
    {
      "epoch": 0.5880887698786459,
      "grad_norm": 3.871943950653076,
      "learning_rate": 4.336048245685911e-07,
      "loss": 0.7354,
      "step": 37775
    },
    {
      "epoch": 0.5881666108806153,
      "grad_norm": 3.344891309738159,
      "learning_rate": 4.3352288556398616e-07,
      "loss": 0.8283,
      "step": 37780
    },
    {
      "epoch": 0.5882444518825847,
      "grad_norm": 2.2011942863464355,
      "learning_rate": 4.334409465593812e-07,
      "loss": 0.6723,
      "step": 37785
    },
    {
      "epoch": 0.588322292884554,
      "grad_norm": 5.631941795349121,
      "learning_rate": 4.333590075547762e-07,
      "loss": 0.6831,
      "step": 37790
    },
    {
      "epoch": 0.5884001338865233,
      "grad_norm": 4.11518669128418,
      "learning_rate": 4.332770685501712e-07,
      "loss": 0.899,
      "step": 37795
    },
    {
      "epoch": 0.5884779748884927,
      "grad_norm": 3.968435049057007,
      "learning_rate": 4.3319512954556625e-07,
      "loss": 0.7435,
      "step": 37800
    },
    {
      "epoch": 0.5885558158904621,
      "grad_norm": 3.8502509593963623,
      "learning_rate": 4.3311319054096127e-07,
      "loss": 0.7121,
      "step": 37805
    },
    {
      "epoch": 0.5886336568924315,
      "grad_norm": 2.9485244750976562,
      "learning_rate": 4.3303125153635634e-07,
      "loss": 0.7846,
      "step": 37810
    },
    {
      "epoch": 0.5887114978944009,
      "grad_norm": 6.609480381011963,
      "learning_rate": 4.329493125317513e-07,
      "loss": 0.6589,
      "step": 37815
    },
    {
      "epoch": 0.5887893388963703,
      "grad_norm": 3.2942028045654297,
      "learning_rate": 4.328673735271464e-07,
      "loss": 0.6566,
      "step": 37820
    },
    {
      "epoch": 0.5888671798983397,
      "grad_norm": 3.662158727645874,
      "learning_rate": 4.327854345225414e-07,
      "loss": 0.6489,
      "step": 37825
    },
    {
      "epoch": 0.588945020900309,
      "grad_norm": 3.3512990474700928,
      "learning_rate": 4.3270349551793643e-07,
      "loss": 0.7234,
      "step": 37830
    },
    {
      "epoch": 0.5890228619022784,
      "grad_norm": 3.1668429374694824,
      "learning_rate": 4.3262155651333145e-07,
      "loss": 0.6152,
      "step": 37835
    },
    {
      "epoch": 0.5891007029042478,
      "grad_norm": 3.1730434894561768,
      "learning_rate": 4.325396175087265e-07,
      "loss": 0.8093,
      "step": 37840
    },
    {
      "epoch": 0.5891785439062172,
      "grad_norm": 4.644217491149902,
      "learning_rate": 4.324576785041215e-07,
      "loss": 0.6419,
      "step": 37845
    },
    {
      "epoch": 0.5892563849081865,
      "grad_norm": 2.744152069091797,
      "learning_rate": 4.3237573949951657e-07,
      "loss": 0.7713,
      "step": 37850
    },
    {
      "epoch": 0.5893342259101559,
      "grad_norm": 4.232105255126953,
      "learning_rate": 4.3229380049491154e-07,
      "loss": 0.7279,
      "step": 37855
    },
    {
      "epoch": 0.5894120669121253,
      "grad_norm": 3.6211671829223633,
      "learning_rate": 4.3221186149030656e-07,
      "loss": 0.8913,
      "step": 37860
    },
    {
      "epoch": 0.5894899079140947,
      "grad_norm": 3.2710461616516113,
      "learning_rate": 4.3212992248570164e-07,
      "loss": 0.7391,
      "step": 37865
    },
    {
      "epoch": 0.589567748916064,
      "grad_norm": 3.1929943561553955,
      "learning_rate": 4.3204798348109666e-07,
      "loss": 0.7435,
      "step": 37870
    },
    {
      "epoch": 0.5896455899180334,
      "grad_norm": 2.94234299659729,
      "learning_rate": 4.3196604447649163e-07,
      "loss": 0.7424,
      "step": 37875
    },
    {
      "epoch": 0.5897234309200028,
      "grad_norm": 4.389419078826904,
      "learning_rate": 4.318841054718867e-07,
      "loss": 0.6661,
      "step": 37880
    },
    {
      "epoch": 0.5898012719219722,
      "grad_norm": 3.376418113708496,
      "learning_rate": 4.318021664672817e-07,
      "loss": 0.7287,
      "step": 37885
    },
    {
      "epoch": 0.5898791129239416,
      "grad_norm": 3.9427671432495117,
      "learning_rate": 4.317202274626768e-07,
      "loss": 0.7682,
      "step": 37890
    },
    {
      "epoch": 0.589956953925911,
      "grad_norm": 4.101550579071045,
      "learning_rate": 4.3163828845807177e-07,
      "loss": 0.7208,
      "step": 37895
    },
    {
      "epoch": 0.5900347949278804,
      "grad_norm": 6.7960028648376465,
      "learning_rate": 4.315563494534668e-07,
      "loss": 0.7928,
      "step": 37900
    },
    {
      "epoch": 0.5901126359298496,
      "grad_norm": 2.9020400047302246,
      "learning_rate": 4.3147441044886186e-07,
      "loss": 0.7527,
      "step": 37905
    },
    {
      "epoch": 0.590190476931819,
      "grad_norm": 4.232307434082031,
      "learning_rate": 4.313924714442569e-07,
      "loss": 0.6577,
      "step": 37910
    },
    {
      "epoch": 0.5902683179337884,
      "grad_norm": 4.673219203948975,
      "learning_rate": 4.313105324396519e-07,
      "loss": 0.7203,
      "step": 37915
    },
    {
      "epoch": 0.5903461589357578,
      "grad_norm": 2.7411675453186035,
      "learning_rate": 4.3122859343504693e-07,
      "loss": 0.8011,
      "step": 37920
    },
    {
      "epoch": 0.5904239999377272,
      "grad_norm": 3.2982139587402344,
      "learning_rate": 4.3114665443044195e-07,
      "loss": 0.7415,
      "step": 37925
    },
    {
      "epoch": 0.5905018409396966,
      "grad_norm": 4.5075578689575195,
      "learning_rate": 4.31064715425837e-07,
      "loss": 0.853,
      "step": 37930
    },
    {
      "epoch": 0.590579681941666,
      "grad_norm": 16.39893341064453,
      "learning_rate": 4.3098277642123205e-07,
      "loss": 0.702,
      "step": 37935
    },
    {
      "epoch": 0.5906575229436354,
      "grad_norm": 4.565467834472656,
      "learning_rate": 4.30900837416627e-07,
      "loss": 0.7557,
      "step": 37940
    },
    {
      "epoch": 0.5907353639456047,
      "grad_norm": 4.250036239624023,
      "learning_rate": 4.308188984120221e-07,
      "loss": 0.7132,
      "step": 37945
    },
    {
      "epoch": 0.590813204947574,
      "grad_norm": 3.9342617988586426,
      "learning_rate": 4.307369594074171e-07,
      "loss": 0.7669,
      "step": 37950
    },
    {
      "epoch": 0.5908910459495434,
      "grad_norm": 3.9016637802124023,
      "learning_rate": 4.3065502040281214e-07,
      "loss": 0.7221,
      "step": 37955
    },
    {
      "epoch": 0.5909688869515128,
      "grad_norm": 3.677473783493042,
      "learning_rate": 4.3057308139820716e-07,
      "loss": 0.7168,
      "step": 37960
    },
    {
      "epoch": 0.5910467279534822,
      "grad_norm": 2.8502583503723145,
      "learning_rate": 4.304911423936022e-07,
      "loss": 0.6136,
      "step": 37965
    },
    {
      "epoch": 0.5911245689554516,
      "grad_norm": 4.504276752471924,
      "learning_rate": 4.304092033889972e-07,
      "loss": 0.6625,
      "step": 37970
    },
    {
      "epoch": 0.591202409957421,
      "grad_norm": 2.7609446048736572,
      "learning_rate": 4.303272643843923e-07,
      "loss": 0.7404,
      "step": 37975
    },
    {
      "epoch": 0.5912802509593903,
      "grad_norm": 8.808570861816406,
      "learning_rate": 4.3024532537978725e-07,
      "loss": 0.7961,
      "step": 37980
    },
    {
      "epoch": 0.5913580919613597,
      "grad_norm": 4.347479343414307,
      "learning_rate": 4.3016338637518227e-07,
      "loss": 0.6965,
      "step": 37985
    },
    {
      "epoch": 0.5914359329633291,
      "grad_norm": 5.645142555236816,
      "learning_rate": 4.3008144737057734e-07,
      "loss": 0.8283,
      "step": 37990
    },
    {
      "epoch": 0.5915137739652985,
      "grad_norm": 5.01529598236084,
      "learning_rate": 4.2999950836597236e-07,
      "loss": 0.7753,
      "step": 37995
    },
    {
      "epoch": 0.5915916149672679,
      "grad_norm": 2.8226544857025146,
      "learning_rate": 4.299175693613674e-07,
      "loss": 0.7876,
      "step": 38000
    },
    {
      "epoch": 0.5916694559692373,
      "grad_norm": 3.962336301803589,
      "learning_rate": 4.298356303567624e-07,
      "loss": 0.7058,
      "step": 38005
    },
    {
      "epoch": 0.5917472969712066,
      "grad_norm": 2.903952121734619,
      "learning_rate": 4.2975369135215743e-07,
      "loss": 0.7697,
      "step": 38010
    },
    {
      "epoch": 0.591825137973176,
      "grad_norm": 12.224853515625,
      "learning_rate": 4.296717523475525e-07,
      "loss": 0.6704,
      "step": 38015
    },
    {
      "epoch": 0.5919029789751453,
      "grad_norm": 3.9108035564422607,
      "learning_rate": 4.295898133429475e-07,
      "loss": 0.8118,
      "step": 38020
    },
    {
      "epoch": 0.5919808199771147,
      "grad_norm": 3.355175256729126,
      "learning_rate": 4.295078743383425e-07,
      "loss": 0.8067,
      "step": 38025
    },
    {
      "epoch": 0.5920586609790841,
      "grad_norm": 3.149761915206909,
      "learning_rate": 4.2942593533373757e-07,
      "loss": 0.7996,
      "step": 38030
    },
    {
      "epoch": 0.5921365019810535,
      "grad_norm": 3.1590535640716553,
      "learning_rate": 4.293439963291326e-07,
      "loss": 0.724,
      "step": 38035
    },
    {
      "epoch": 0.5922143429830229,
      "grad_norm": 3.3773910999298096,
      "learning_rate": 4.2926205732452756e-07,
      "loss": 0.7004,
      "step": 38040
    },
    {
      "epoch": 0.5922921839849923,
      "grad_norm": 3.256758213043213,
      "learning_rate": 4.2918011831992264e-07,
      "loss": 0.6831,
      "step": 38045
    },
    {
      "epoch": 0.5923700249869617,
      "grad_norm": 3.5986318588256836,
      "learning_rate": 4.2909817931531766e-07,
      "loss": 0.8466,
      "step": 38050
    },
    {
      "epoch": 0.592447865988931,
      "grad_norm": 4.349609375,
      "learning_rate": 4.290162403107127e-07,
      "loss": 0.8033,
      "step": 38055
    },
    {
      "epoch": 0.5925257069909003,
      "grad_norm": 3.1396384239196777,
      "learning_rate": 4.289343013061077e-07,
      "loss": 0.5902,
      "step": 38060
    },
    {
      "epoch": 0.5926035479928697,
      "grad_norm": 7.653540134429932,
      "learning_rate": 4.288523623015027e-07,
      "loss": 0.7076,
      "step": 38065
    },
    {
      "epoch": 0.5926813889948391,
      "grad_norm": 4.12556791305542,
      "learning_rate": 4.287704232968978e-07,
      "loss": 0.6421,
      "step": 38070
    },
    {
      "epoch": 0.5927592299968085,
      "grad_norm": 3.6118996143341064,
      "learning_rate": 4.286884842922928e-07,
      "loss": 0.8442,
      "step": 38075
    },
    {
      "epoch": 0.5928370709987779,
      "grad_norm": 2.6838951110839844,
      "learning_rate": 4.286065452876878e-07,
      "loss": 0.6586,
      "step": 38080
    },
    {
      "epoch": 0.5929149120007473,
      "grad_norm": 3.188577175140381,
      "learning_rate": 4.2852460628308287e-07,
      "loss": 0.7352,
      "step": 38085
    },
    {
      "epoch": 0.5929927530027167,
      "grad_norm": 3.9639928340911865,
      "learning_rate": 4.284426672784779e-07,
      "loss": 0.7819,
      "step": 38090
    },
    {
      "epoch": 0.593070594004686,
      "grad_norm": 2.9961462020874023,
      "learning_rate": 4.283607282738729e-07,
      "loss": 0.7149,
      "step": 38095
    },
    {
      "epoch": 0.5931484350066554,
      "grad_norm": 3.466834306716919,
      "learning_rate": 4.2827878926926793e-07,
      "loss": 0.5653,
      "step": 38100
    },
    {
      "epoch": 0.5932262760086248,
      "grad_norm": 3.4719815254211426,
      "learning_rate": 4.2819685026466295e-07,
      "loss": 0.8381,
      "step": 38105
    },
    {
      "epoch": 0.5933041170105942,
      "grad_norm": 3.0592329502105713,
      "learning_rate": 4.28114911260058e-07,
      "loss": 0.6628,
      "step": 38110
    },
    {
      "epoch": 0.5933819580125635,
      "grad_norm": 4.619132041931152,
      "learning_rate": 4.2803297225545305e-07,
      "loss": 0.7792,
      "step": 38115
    },
    {
      "epoch": 0.5934597990145329,
      "grad_norm": 3.255004644393921,
      "learning_rate": 4.27951033250848e-07,
      "loss": 0.7402,
      "step": 38120
    },
    {
      "epoch": 0.5935376400165023,
      "grad_norm": 8.541637420654297,
      "learning_rate": 4.278690942462431e-07,
      "loss": 0.6628,
      "step": 38125
    },
    {
      "epoch": 0.5936154810184717,
      "grad_norm": 2.8503732681274414,
      "learning_rate": 4.277871552416381e-07,
      "loss": 0.7862,
      "step": 38130
    },
    {
      "epoch": 0.593693322020441,
      "grad_norm": 4.021406173706055,
      "learning_rate": 4.2770521623703314e-07,
      "loss": 0.7478,
      "step": 38135
    },
    {
      "epoch": 0.5937711630224104,
      "grad_norm": 3.5898277759552,
      "learning_rate": 4.2762327723242816e-07,
      "loss": 0.7514,
      "step": 38140
    },
    {
      "epoch": 0.5938490040243798,
      "grad_norm": 3.7222306728363037,
      "learning_rate": 4.275413382278232e-07,
      "loss": 0.771,
      "step": 38145
    },
    {
      "epoch": 0.5939268450263492,
      "grad_norm": 5.752460956573486,
      "learning_rate": 4.274593992232182e-07,
      "loss": 0.7155,
      "step": 38150
    },
    {
      "epoch": 0.5940046860283186,
      "grad_norm": 10.54542350769043,
      "learning_rate": 4.273774602186133e-07,
      "loss": 0.6828,
      "step": 38155
    },
    {
      "epoch": 0.594082527030288,
      "grad_norm": 4.585755825042725,
      "learning_rate": 4.2729552121400825e-07,
      "loss": 0.8322,
      "step": 38160
    },
    {
      "epoch": 0.5941603680322574,
      "grad_norm": 10.165528297424316,
      "learning_rate": 4.2721358220940327e-07,
      "loss": 0.7807,
      "step": 38165
    },
    {
      "epoch": 0.5942382090342266,
      "grad_norm": 4.18814754486084,
      "learning_rate": 4.2713164320479834e-07,
      "loss": 0.8005,
      "step": 38170
    },
    {
      "epoch": 0.594316050036196,
      "grad_norm": 5.790228843688965,
      "learning_rate": 4.2704970420019337e-07,
      "loss": 0.6989,
      "step": 38175
    },
    {
      "epoch": 0.5943938910381654,
      "grad_norm": 7.689640998840332,
      "learning_rate": 4.2696776519558833e-07,
      "loss": 0.7684,
      "step": 38180
    },
    {
      "epoch": 0.5944717320401348,
      "grad_norm": 4.072168827056885,
      "learning_rate": 4.268858261909834e-07,
      "loss": 0.6586,
      "step": 38185
    },
    {
      "epoch": 0.5945495730421042,
      "grad_norm": 3.7060728073120117,
      "learning_rate": 4.2680388718637843e-07,
      "loss": 0.6983,
      "step": 38190
    },
    {
      "epoch": 0.5946274140440736,
      "grad_norm": 5.167768955230713,
      "learning_rate": 4.267219481817735e-07,
      "loss": 0.789,
      "step": 38195
    },
    {
      "epoch": 0.594705255046043,
      "grad_norm": 3.973193883895874,
      "learning_rate": 4.266400091771685e-07,
      "loss": 0.7989,
      "step": 38200
    },
    {
      "epoch": 0.5947830960480124,
      "grad_norm": 2.94742488861084,
      "learning_rate": 4.265580701725635e-07,
      "loss": 0.7468,
      "step": 38205
    },
    {
      "epoch": 0.5948609370499817,
      "grad_norm": 4.520554542541504,
      "learning_rate": 4.2647613116795857e-07,
      "loss": 0.8302,
      "step": 38210
    },
    {
      "epoch": 0.5949387780519511,
      "grad_norm": 3.7475380897521973,
      "learning_rate": 4.263941921633536e-07,
      "loss": 0.7708,
      "step": 38215
    },
    {
      "epoch": 0.5950166190539204,
      "grad_norm": 4.927328586578369,
      "learning_rate": 4.263122531587486e-07,
      "loss": 0.7209,
      "step": 38220
    },
    {
      "epoch": 0.5950944600558898,
      "grad_norm": 3.1835684776306152,
      "learning_rate": 4.2623031415414364e-07,
      "loss": 0.6958,
      "step": 38225
    },
    {
      "epoch": 0.5951723010578592,
      "grad_norm": 5.740211009979248,
      "learning_rate": 4.2614837514953866e-07,
      "loss": 0.6772,
      "step": 38230
    },
    {
      "epoch": 0.5952501420598286,
      "grad_norm": 3.132599353790283,
      "learning_rate": 4.260664361449337e-07,
      "loss": 0.6546,
      "step": 38235
    },
    {
      "epoch": 0.595327983061798,
      "grad_norm": 3.3599374294281006,
      "learning_rate": 4.2598449714032876e-07,
      "loss": 0.8553,
      "step": 38240
    },
    {
      "epoch": 0.5954058240637673,
      "grad_norm": 6.064815998077393,
      "learning_rate": 4.259025581357237e-07,
      "loss": 0.766,
      "step": 38245
    },
    {
      "epoch": 0.5954836650657367,
      "grad_norm": 5.8122878074646,
      "learning_rate": 4.258206191311188e-07,
      "loss": 0.8169,
      "step": 38250
    },
    {
      "epoch": 0.5955615060677061,
      "grad_norm": 3.881312608718872,
      "learning_rate": 4.257386801265138e-07,
      "loss": 0.828,
      "step": 38255
    },
    {
      "epoch": 0.5956393470696755,
      "grad_norm": 10.445817947387695,
      "learning_rate": 4.2565674112190884e-07,
      "loss": 0.614,
      "step": 38260
    },
    {
      "epoch": 0.5957171880716449,
      "grad_norm": 5.692413330078125,
      "learning_rate": 4.2557480211730387e-07,
      "loss": 0.7249,
      "step": 38265
    },
    {
      "epoch": 0.5957950290736143,
      "grad_norm": 3.82875919342041,
      "learning_rate": 4.254928631126989e-07,
      "loss": 0.7245,
      "step": 38270
    },
    {
      "epoch": 0.5958728700755836,
      "grad_norm": 6.629369258880615,
      "learning_rate": 4.254109241080939e-07,
      "loss": 0.7463,
      "step": 38275
    },
    {
      "epoch": 0.595950711077553,
      "grad_norm": 3.944502353668213,
      "learning_rate": 4.25328985103489e-07,
      "loss": 0.7218,
      "step": 38280
    },
    {
      "epoch": 0.5960285520795223,
      "grad_norm": 5.721776485443115,
      "learning_rate": 4.2524704609888395e-07,
      "loss": 0.8147,
      "step": 38285
    },
    {
      "epoch": 0.5961063930814917,
      "grad_norm": 2.55517315864563,
      "learning_rate": 4.25165107094279e-07,
      "loss": 0.6609,
      "step": 38290
    },
    {
      "epoch": 0.5961842340834611,
      "grad_norm": 3.4020028114318848,
      "learning_rate": 4.2508316808967405e-07,
      "loss": 0.7534,
      "step": 38295
    },
    {
      "epoch": 0.5962620750854305,
      "grad_norm": 5.368770599365234,
      "learning_rate": 4.2500122908506907e-07,
      "loss": 0.7771,
      "step": 38300
    },
    {
      "epoch": 0.5963399160873999,
      "grad_norm": 3.566833257675171,
      "learning_rate": 4.2491929008046404e-07,
      "loss": 0.7308,
      "step": 38305
    },
    {
      "epoch": 0.5964177570893693,
      "grad_norm": 3.544767141342163,
      "learning_rate": 4.248373510758591e-07,
      "loss": 0.7161,
      "step": 38310
    },
    {
      "epoch": 0.5964955980913387,
      "grad_norm": 3.454904556274414,
      "learning_rate": 4.2475541207125414e-07,
      "loss": 0.6666,
      "step": 38315
    },
    {
      "epoch": 0.596573439093308,
      "grad_norm": 3.493147373199463,
      "learning_rate": 4.246734730666492e-07,
      "loss": 0.8424,
      "step": 38320
    },
    {
      "epoch": 0.5966512800952773,
      "grad_norm": 2.8635623455047607,
      "learning_rate": 4.245915340620442e-07,
      "loss": 0.7427,
      "step": 38325
    },
    {
      "epoch": 0.5967291210972467,
      "grad_norm": 1.7464525699615479,
      "learning_rate": 4.245095950574392e-07,
      "loss": 0.6271,
      "step": 38330
    },
    {
      "epoch": 0.5968069620992161,
      "grad_norm": 4.078522682189941,
      "learning_rate": 4.244276560528343e-07,
      "loss": 0.7958,
      "step": 38335
    },
    {
      "epoch": 0.5968848031011855,
      "grad_norm": 4.385714054107666,
      "learning_rate": 4.243457170482293e-07,
      "loss": 0.7871,
      "step": 38340
    },
    {
      "epoch": 0.5969626441031549,
      "grad_norm": 3.4198031425476074,
      "learning_rate": 4.2426377804362427e-07,
      "loss": 0.6849,
      "step": 38345
    },
    {
      "epoch": 0.5970404851051243,
      "grad_norm": 4.621438980102539,
      "learning_rate": 4.2418183903901934e-07,
      "loss": 0.7077,
      "step": 38350
    },
    {
      "epoch": 0.5971183261070937,
      "grad_norm": 4.165358543395996,
      "learning_rate": 4.2409990003441437e-07,
      "loss": 0.7936,
      "step": 38355
    },
    {
      "epoch": 0.597196167109063,
      "grad_norm": 3.5964839458465576,
      "learning_rate": 4.240179610298094e-07,
      "loss": 0.6716,
      "step": 38360
    },
    {
      "epoch": 0.5972740081110324,
      "grad_norm": 6.612133026123047,
      "learning_rate": 4.239360220252044e-07,
      "loss": 0.6268,
      "step": 38365
    },
    {
      "epoch": 0.5973518491130018,
      "grad_norm": 4.084269046783447,
      "learning_rate": 4.2385408302059943e-07,
      "loss": 0.7208,
      "step": 38370
    },
    {
      "epoch": 0.5974296901149712,
      "grad_norm": 6.646197319030762,
      "learning_rate": 4.237721440159945e-07,
      "loss": 0.7889,
      "step": 38375
    },
    {
      "epoch": 0.5975075311169405,
      "grad_norm": 3.1683709621429443,
      "learning_rate": 4.2369020501138953e-07,
      "loss": 0.772,
      "step": 38380
    },
    {
      "epoch": 0.5975853721189099,
      "grad_norm": 5.805420398712158,
      "learning_rate": 4.236082660067845e-07,
      "loss": 0.6657,
      "step": 38385
    },
    {
      "epoch": 0.5976632131208793,
      "grad_norm": 3.2822582721710205,
      "learning_rate": 4.2352632700217957e-07,
      "loss": 0.8141,
      "step": 38390
    },
    {
      "epoch": 0.5977410541228486,
      "grad_norm": 3.1840503215789795,
      "learning_rate": 4.234443879975746e-07,
      "loss": 0.7342,
      "step": 38395
    },
    {
      "epoch": 0.597818895124818,
      "grad_norm": 2.741015672683716,
      "learning_rate": 4.233624489929696e-07,
      "loss": 0.7362,
      "step": 38400
    },
    {
      "epoch": 0.5978967361267874,
      "grad_norm": 3.976806163787842,
      "learning_rate": 4.2328050998836464e-07,
      "loss": 0.7938,
      "step": 38405
    },
    {
      "epoch": 0.5979745771287568,
      "grad_norm": 3.395183563232422,
      "learning_rate": 4.2319857098375966e-07,
      "loss": 0.9398,
      "step": 38410
    },
    {
      "epoch": 0.5980524181307262,
      "grad_norm": 5.811853885650635,
      "learning_rate": 4.231166319791547e-07,
      "loss": 0.7427,
      "step": 38415
    },
    {
      "epoch": 0.5981302591326956,
      "grad_norm": 4.283621311187744,
      "learning_rate": 4.2303469297454976e-07,
      "loss": 0.7345,
      "step": 38420
    },
    {
      "epoch": 0.598208100134665,
      "grad_norm": 2.862078905105591,
      "learning_rate": 4.2295275396994473e-07,
      "loss": 0.7357,
      "step": 38425
    },
    {
      "epoch": 0.5982859411366344,
      "grad_norm": 3.3480498790740967,
      "learning_rate": 4.2287081496533975e-07,
      "loss": 0.8368,
      "step": 38430
    },
    {
      "epoch": 0.5983637821386036,
      "grad_norm": 2.853435516357422,
      "learning_rate": 4.227888759607348e-07,
      "loss": 0.7697,
      "step": 38435
    },
    {
      "epoch": 0.598441623140573,
      "grad_norm": 4.9920454025268555,
      "learning_rate": 4.2270693695612985e-07,
      "loss": 0.8507,
      "step": 38440
    },
    {
      "epoch": 0.5985194641425424,
      "grad_norm": 4.227562427520752,
      "learning_rate": 4.2262499795152487e-07,
      "loss": 0.6868,
      "step": 38445
    },
    {
      "epoch": 0.5985973051445118,
      "grad_norm": 3.785295009613037,
      "learning_rate": 4.225430589469199e-07,
      "loss": 0.8168,
      "step": 38450
    },
    {
      "epoch": 0.5986751461464812,
      "grad_norm": 3.722372531890869,
      "learning_rate": 4.224611199423149e-07,
      "loss": 0.7935,
      "step": 38455
    },
    {
      "epoch": 0.5987529871484506,
      "grad_norm": 3.7080230712890625,
      "learning_rate": 4.2237918093771e-07,
      "loss": 0.8772,
      "step": 38460
    },
    {
      "epoch": 0.59883082815042,
      "grad_norm": 2.860541820526123,
      "learning_rate": 4.2229724193310495e-07,
      "loss": 0.7082,
      "step": 38465
    },
    {
      "epoch": 0.5989086691523894,
      "grad_norm": 3.3840725421905518,
      "learning_rate": 4.222153029285e-07,
      "loss": 0.744,
      "step": 38470
    },
    {
      "epoch": 0.5989865101543587,
      "grad_norm": 4.850223064422607,
      "learning_rate": 4.2213336392389505e-07,
      "loss": 0.7401,
      "step": 38475
    },
    {
      "epoch": 0.5990643511563281,
      "grad_norm": 3.389915704727173,
      "learning_rate": 4.2205142491929007e-07,
      "loss": 0.6984,
      "step": 38480
    },
    {
      "epoch": 0.5991421921582974,
      "grad_norm": 2.9824862480163574,
      "learning_rate": 4.2196948591468504e-07,
      "loss": 0.7049,
      "step": 38485
    },
    {
      "epoch": 0.5992200331602668,
      "grad_norm": 5.369162559509277,
      "learning_rate": 4.218875469100801e-07,
      "loss": 0.6708,
      "step": 38490
    },
    {
      "epoch": 0.5992978741622362,
      "grad_norm": 5.881549835205078,
      "learning_rate": 4.2180560790547514e-07,
      "loss": 0.7282,
      "step": 38495
    },
    {
      "epoch": 0.5993757151642056,
      "grad_norm": 4.575121879577637,
      "learning_rate": 4.217236689008702e-07,
      "loss": 0.7414,
      "step": 38500
    },
    {
      "epoch": 0.599453556166175,
      "grad_norm": 3.1772477626800537,
      "learning_rate": 4.216417298962652e-07,
      "loss": 0.7601,
      "step": 38505
    },
    {
      "epoch": 0.5995313971681443,
      "grad_norm": 4.598190784454346,
      "learning_rate": 4.215597908916602e-07,
      "loss": 0.7718,
      "step": 38510
    },
    {
      "epoch": 0.5996092381701137,
      "grad_norm": 4.588962078094482,
      "learning_rate": 4.214778518870553e-07,
      "loss": 0.7272,
      "step": 38515
    },
    {
      "epoch": 0.5996870791720831,
      "grad_norm": 3.9823856353759766,
      "learning_rate": 4.213959128824503e-07,
      "loss": 0.6968,
      "step": 38520
    },
    {
      "epoch": 0.5997649201740525,
      "grad_norm": 4.227712154388428,
      "learning_rate": 4.2131397387784527e-07,
      "loss": 0.7463,
      "step": 38525
    },
    {
      "epoch": 0.5998427611760219,
      "grad_norm": 6.595142841339111,
      "learning_rate": 4.2123203487324035e-07,
      "loss": 0.66,
      "step": 38530
    },
    {
      "epoch": 0.5999206021779913,
      "grad_norm": 8.815082550048828,
      "learning_rate": 4.2115009586863537e-07,
      "loss": 0.7947,
      "step": 38535
    },
    {
      "epoch": 0.5999984431799606,
      "grad_norm": 7.649603843688965,
      "learning_rate": 4.210681568640304e-07,
      "loss": 0.8412,
      "step": 38540
    },
    {
      "epoch": 0.60007628418193,
      "grad_norm": 2.8193442821502686,
      "learning_rate": 4.2098621785942546e-07,
      "loss": 0.7284,
      "step": 38545
    },
    {
      "epoch": 0.6001541251838993,
      "grad_norm": 4.629957675933838,
      "learning_rate": 4.2090427885482043e-07,
      "loss": 0.6817,
      "step": 38550
    },
    {
      "epoch": 0.6002319661858687,
      "grad_norm": 5.396300792694092,
      "learning_rate": 4.2082233985021546e-07,
      "loss": 0.7943,
      "step": 38555
    },
    {
      "epoch": 0.6003098071878381,
      "grad_norm": 5.600543022155762,
      "learning_rate": 4.2074040084561053e-07,
      "loss": 0.6614,
      "step": 38560
    },
    {
      "epoch": 0.6003876481898075,
      "grad_norm": 4.440396308898926,
      "learning_rate": 4.2065846184100555e-07,
      "loss": 0.8788,
      "step": 38565
    },
    {
      "epoch": 0.6004654891917769,
      "grad_norm": 3.1608381271362305,
      "learning_rate": 4.205765228364006e-07,
      "loss": 0.7529,
      "step": 38570
    },
    {
      "epoch": 0.6005433301937463,
      "grad_norm": 3.8151934146881104,
      "learning_rate": 4.204945838317956e-07,
      "loss": 0.7728,
      "step": 38575
    },
    {
      "epoch": 0.6006211711957157,
      "grad_norm": 4.300529956817627,
      "learning_rate": 4.204126448271906e-07,
      "loss": 0.7308,
      "step": 38580
    },
    {
      "epoch": 0.600699012197685,
      "grad_norm": 3.349766969680786,
      "learning_rate": 4.203307058225857e-07,
      "loss": 0.7378,
      "step": 38585
    },
    {
      "epoch": 0.6007768531996543,
      "grad_norm": 4.454970836639404,
      "learning_rate": 4.2024876681798066e-07,
      "loss": 0.7342,
      "step": 38590
    },
    {
      "epoch": 0.6008546942016237,
      "grad_norm": 2.738116502761841,
      "learning_rate": 4.201668278133757e-07,
      "loss": 0.7099,
      "step": 38595
    },
    {
      "epoch": 0.6009325352035931,
      "grad_norm": 3.0065481662750244,
      "learning_rate": 4.2008488880877076e-07,
      "loss": 0.6516,
      "step": 38600
    },
    {
      "epoch": 0.6010103762055625,
      "grad_norm": 4.0895819664001465,
      "learning_rate": 4.200029498041658e-07,
      "loss": 0.7837,
      "step": 38605
    },
    {
      "epoch": 0.6010882172075319,
      "grad_norm": 5.282425880432129,
      "learning_rate": 4.1992101079956075e-07,
      "loss": 0.6689,
      "step": 38610
    },
    {
      "epoch": 0.6011660582095013,
      "grad_norm": 8.682831764221191,
      "learning_rate": 4.198390717949558e-07,
      "loss": 0.7511,
      "step": 38615
    },
    {
      "epoch": 0.6012438992114707,
      "grad_norm": 7.007259845733643,
      "learning_rate": 4.1975713279035085e-07,
      "loss": 0.814,
      "step": 38620
    },
    {
      "epoch": 0.60132174021344,
      "grad_norm": 6.4824604988098145,
      "learning_rate": 4.196751937857459e-07,
      "loss": 0.7077,
      "step": 38625
    },
    {
      "epoch": 0.6013995812154094,
      "grad_norm": 2.9014294147491455,
      "learning_rate": 4.195932547811409e-07,
      "loss": 0.6718,
      "step": 38630
    },
    {
      "epoch": 0.6014774222173788,
      "grad_norm": 6.2585225105285645,
      "learning_rate": 4.195113157765359e-07,
      "loss": 0.7218,
      "step": 38635
    },
    {
      "epoch": 0.6015552632193482,
      "grad_norm": 3.224862575531006,
      "learning_rate": 4.19429376771931e-07,
      "loss": 0.7297,
      "step": 38640
    },
    {
      "epoch": 0.6016331042213175,
      "grad_norm": 2.8529951572418213,
      "learning_rate": 4.19347437767326e-07,
      "loss": 0.7147,
      "step": 38645
    },
    {
      "epoch": 0.6017109452232869,
      "grad_norm": 3.5962228775024414,
      "learning_rate": 4.19265498762721e-07,
      "loss": 0.7651,
      "step": 38650
    },
    {
      "epoch": 0.6017887862252563,
      "grad_norm": 3.7043309211730957,
      "learning_rate": 4.1918355975811605e-07,
      "loss": 0.7137,
      "step": 38655
    },
    {
      "epoch": 0.6018666272272256,
      "grad_norm": 3.5017330646514893,
      "learning_rate": 4.191016207535111e-07,
      "loss": 0.6791,
      "step": 38660
    },
    {
      "epoch": 0.601944468229195,
      "grad_norm": 4.400074005126953,
      "learning_rate": 4.190196817489061e-07,
      "loss": 0.7241,
      "step": 38665
    },
    {
      "epoch": 0.6020223092311644,
      "grad_norm": 3.518547534942627,
      "learning_rate": 4.189377427443011e-07,
      "loss": 0.7235,
      "step": 38670
    },
    {
      "epoch": 0.6021001502331338,
      "grad_norm": 4.306619167327881,
      "learning_rate": 4.1885580373969614e-07,
      "loss": 0.7333,
      "step": 38675
    },
    {
      "epoch": 0.6021779912351032,
      "grad_norm": 3.960416316986084,
      "learning_rate": 4.1877386473509116e-07,
      "loss": 0.6207,
      "step": 38680
    },
    {
      "epoch": 0.6022558322370726,
      "grad_norm": 3.6650583744049072,
      "learning_rate": 4.1869192573048624e-07,
      "loss": 0.5762,
      "step": 38685
    },
    {
      "epoch": 0.602333673239042,
      "grad_norm": 3.4492640495300293,
      "learning_rate": 4.186099867258812e-07,
      "loss": 0.7181,
      "step": 38690
    },
    {
      "epoch": 0.6024115142410114,
      "grad_norm": 3.1774511337280273,
      "learning_rate": 4.185280477212763e-07,
      "loss": 0.7667,
      "step": 38695
    },
    {
      "epoch": 0.6024893552429806,
      "grad_norm": 2.6972265243530273,
      "learning_rate": 4.184461087166713e-07,
      "loss": 0.7558,
      "step": 38700
    },
    {
      "epoch": 0.60256719624495,
      "grad_norm": 8.99730110168457,
      "learning_rate": 4.183641697120663e-07,
      "loss": 0.7594,
      "step": 38705
    },
    {
      "epoch": 0.6026450372469194,
      "grad_norm": 2.9149889945983887,
      "learning_rate": 4.1828223070746135e-07,
      "loss": 0.6711,
      "step": 38710
    },
    {
      "epoch": 0.6027228782488888,
      "grad_norm": 4.985609531402588,
      "learning_rate": 4.1820029170285637e-07,
      "loss": 0.7436,
      "step": 38715
    },
    {
      "epoch": 0.6028007192508582,
      "grad_norm": 5.0288872718811035,
      "learning_rate": 4.181183526982514e-07,
      "loss": 0.8574,
      "step": 38720
    },
    {
      "epoch": 0.6028785602528276,
      "grad_norm": 2.5508811473846436,
      "learning_rate": 4.1803641369364647e-07,
      "loss": 0.7149,
      "step": 38725
    },
    {
      "epoch": 0.602956401254797,
      "grad_norm": 3.41721773147583,
      "learning_rate": 4.1795447468904143e-07,
      "loss": 0.744,
      "step": 38730
    },
    {
      "epoch": 0.6030342422567664,
      "grad_norm": 3.904279947280884,
      "learning_rate": 4.1787253568443646e-07,
      "loss": 0.7351,
      "step": 38735
    },
    {
      "epoch": 0.6031120832587357,
      "grad_norm": 3.095898389816284,
      "learning_rate": 4.1779059667983153e-07,
      "loss": 0.649,
      "step": 38740
    },
    {
      "epoch": 0.6031899242607051,
      "grad_norm": 2.6285059452056885,
      "learning_rate": 4.1770865767522655e-07,
      "loss": 0.7636,
      "step": 38745
    },
    {
      "epoch": 0.6032677652626744,
      "grad_norm": 8.989908218383789,
      "learning_rate": 4.176267186706215e-07,
      "loss": 0.5928,
      "step": 38750
    },
    {
      "epoch": 0.6033456062646438,
      "grad_norm": 3.068652391433716,
      "learning_rate": 4.175447796660166e-07,
      "loss": 0.6253,
      "step": 38755
    },
    {
      "epoch": 0.6034234472666132,
      "grad_norm": 3.040426015853882,
      "learning_rate": 4.174628406614116e-07,
      "loss": 0.8266,
      "step": 38760
    },
    {
      "epoch": 0.6035012882685826,
      "grad_norm": 5.4839091300964355,
      "learning_rate": 4.173809016568067e-07,
      "loss": 0.6478,
      "step": 38765
    },
    {
      "epoch": 0.603579129270552,
      "grad_norm": 3.8447153568267822,
      "learning_rate": 4.1729896265220166e-07,
      "loss": 0.7919,
      "step": 38770
    },
    {
      "epoch": 0.6036569702725213,
      "grad_norm": 6.046387672424316,
      "learning_rate": 4.172170236475967e-07,
      "loss": 0.7108,
      "step": 38775
    },
    {
      "epoch": 0.6037348112744907,
      "grad_norm": 4.365834712982178,
      "learning_rate": 4.1713508464299176e-07,
      "loss": 0.81,
      "step": 38780
    },
    {
      "epoch": 0.6038126522764601,
      "grad_norm": 3.966287612915039,
      "learning_rate": 4.170531456383868e-07,
      "loss": 0.7566,
      "step": 38785
    },
    {
      "epoch": 0.6038904932784295,
      "grad_norm": 4.648808002471924,
      "learning_rate": 4.1697120663378175e-07,
      "loss": 0.8329,
      "step": 38790
    },
    {
      "epoch": 0.6039683342803989,
      "grad_norm": 3.9006409645080566,
      "learning_rate": 4.168892676291768e-07,
      "loss": 0.7947,
      "step": 38795
    },
    {
      "epoch": 0.6040461752823683,
      "grad_norm": 4.299967288970947,
      "learning_rate": 4.1680732862457185e-07,
      "loss": 0.7439,
      "step": 38800
    },
    {
      "epoch": 0.6041240162843377,
      "grad_norm": 4.394403457641602,
      "learning_rate": 4.1672538961996687e-07,
      "loss": 0.8195,
      "step": 38805
    },
    {
      "epoch": 0.604201857286307,
      "grad_norm": 3.140852689743042,
      "learning_rate": 4.166434506153619e-07,
      "loss": 0.7742,
      "step": 38810
    },
    {
      "epoch": 0.6042796982882763,
      "grad_norm": 4.173416614532471,
      "learning_rate": 4.165615116107569e-07,
      "loss": 0.7203,
      "step": 38815
    },
    {
      "epoch": 0.6043575392902457,
      "grad_norm": 3.6813554763793945,
      "learning_rate": 4.16479572606152e-07,
      "loss": 0.8355,
      "step": 38820
    },
    {
      "epoch": 0.6044353802922151,
      "grad_norm": 5.201940059661865,
      "learning_rate": 4.16397633601547e-07,
      "loss": 0.7961,
      "step": 38825
    },
    {
      "epoch": 0.6045132212941845,
      "grad_norm": 7.682170867919922,
      "learning_rate": 4.16315694596942e-07,
      "loss": 0.7625,
      "step": 38830
    },
    {
      "epoch": 0.6045910622961539,
      "grad_norm": 2.4550082683563232,
      "learning_rate": 4.1623375559233705e-07,
      "loss": 0.6449,
      "step": 38835
    },
    {
      "epoch": 0.6046689032981233,
      "grad_norm": 7.961645126342773,
      "learning_rate": 4.161518165877321e-07,
      "loss": 0.7796,
      "step": 38840
    },
    {
      "epoch": 0.6047467443000927,
      "grad_norm": 6.982649803161621,
      "learning_rate": 4.160698775831271e-07,
      "loss": 0.7343,
      "step": 38845
    },
    {
      "epoch": 0.604824585302062,
      "grad_norm": 4.030280113220215,
      "learning_rate": 4.159879385785221e-07,
      "loss": 0.6107,
      "step": 38850
    },
    {
      "epoch": 0.6049024263040313,
      "grad_norm": 2.7146997451782227,
      "learning_rate": 4.1590599957391714e-07,
      "loss": 0.7827,
      "step": 38855
    },
    {
      "epoch": 0.6049802673060007,
      "grad_norm": 4.76948356628418,
      "learning_rate": 4.1582406056931216e-07,
      "loss": 0.8435,
      "step": 38860
    },
    {
      "epoch": 0.6050581083079701,
      "grad_norm": 11.783685684204102,
      "learning_rate": 4.1574212156470724e-07,
      "loss": 0.6406,
      "step": 38865
    },
    {
      "epoch": 0.6051359493099395,
      "grad_norm": 3.7206830978393555,
      "learning_rate": 4.1566018256010226e-07,
      "loss": 0.616,
      "step": 38870
    },
    {
      "epoch": 0.6052137903119089,
      "grad_norm": 2.9958953857421875,
      "learning_rate": 4.1557824355549723e-07,
      "loss": 0.8191,
      "step": 38875
    },
    {
      "epoch": 0.6052916313138783,
      "grad_norm": 2.738062620162964,
      "learning_rate": 4.154963045508923e-07,
      "loss": 0.6344,
      "step": 38880
    },
    {
      "epoch": 0.6053694723158477,
      "grad_norm": 5.062041282653809,
      "learning_rate": 4.154143655462873e-07,
      "loss": 0.7856,
      "step": 38885
    },
    {
      "epoch": 0.605447313317817,
      "grad_norm": 5.251619815826416,
      "learning_rate": 4.153324265416824e-07,
      "loss": 0.783,
      "step": 38890
    },
    {
      "epoch": 0.6055251543197864,
      "grad_norm": 3.9272677898406982,
      "learning_rate": 4.1525048753707737e-07,
      "loss": 0.851,
      "step": 38895
    },
    {
      "epoch": 0.6056029953217558,
      "grad_norm": 5.4498114585876465,
      "learning_rate": 4.151685485324724e-07,
      "loss": 0.6092,
      "step": 38900
    },
    {
      "epoch": 0.6056808363237252,
      "grad_norm": 9.554481506347656,
      "learning_rate": 4.1508660952786747e-07,
      "loss": 0.7348,
      "step": 38905
    },
    {
      "epoch": 0.6057586773256946,
      "grad_norm": 3.203547954559326,
      "learning_rate": 4.150046705232625e-07,
      "loss": 0.7755,
      "step": 38910
    },
    {
      "epoch": 0.6058365183276639,
      "grad_norm": 5.171728610992432,
      "learning_rate": 4.1492273151865746e-07,
      "loss": 0.8787,
      "step": 38915
    },
    {
      "epoch": 0.6059143593296333,
      "grad_norm": 2.708963632583618,
      "learning_rate": 4.1484079251405253e-07,
      "loss": 0.7856,
      "step": 38920
    },
    {
      "epoch": 0.6059922003316026,
      "grad_norm": 3.6987416744232178,
      "learning_rate": 4.1475885350944755e-07,
      "loss": 0.6439,
      "step": 38925
    },
    {
      "epoch": 0.606070041333572,
      "grad_norm": 4.390050411224365,
      "learning_rate": 4.146769145048426e-07,
      "loss": 0.7572,
      "step": 38930
    },
    {
      "epoch": 0.6061478823355414,
      "grad_norm": 3.213430404663086,
      "learning_rate": 4.145949755002376e-07,
      "loss": 0.7521,
      "step": 38935
    },
    {
      "epoch": 0.6062257233375108,
      "grad_norm": 5.075226783752441,
      "learning_rate": 4.145130364956326e-07,
      "loss": 0.7876,
      "step": 38940
    },
    {
      "epoch": 0.6063035643394802,
      "grad_norm": 3.1709799766540527,
      "learning_rate": 4.144310974910277e-07,
      "loss": 0.8208,
      "step": 38945
    },
    {
      "epoch": 0.6063814053414496,
      "grad_norm": 5.410738945007324,
      "learning_rate": 4.143491584864227e-07,
      "loss": 0.7509,
      "step": 38950
    },
    {
      "epoch": 0.606459246343419,
      "grad_norm": 4.873678207397461,
      "learning_rate": 4.142672194818177e-07,
      "loss": 0.7201,
      "step": 38955
    },
    {
      "epoch": 0.6065370873453884,
      "grad_norm": 3.3184118270874023,
      "learning_rate": 4.1418528047721276e-07,
      "loss": 0.6393,
      "step": 38960
    },
    {
      "epoch": 0.6066149283473576,
      "grad_norm": 3.898770332336426,
      "learning_rate": 4.141033414726078e-07,
      "loss": 0.7036,
      "step": 38965
    },
    {
      "epoch": 0.606692769349327,
      "grad_norm": 7.267730236053467,
      "learning_rate": 4.140214024680028e-07,
      "loss": 0.7932,
      "step": 38970
    },
    {
      "epoch": 0.6067706103512964,
      "grad_norm": 2.8807530403137207,
      "learning_rate": 4.139394634633978e-07,
      "loss": 0.6838,
      "step": 38975
    },
    {
      "epoch": 0.6068484513532658,
      "grad_norm": 5.941493511199951,
      "learning_rate": 4.1385752445879285e-07,
      "loss": 0.6607,
      "step": 38980
    },
    {
      "epoch": 0.6069262923552352,
      "grad_norm": 3.706313133239746,
      "learning_rate": 4.1377558545418787e-07,
      "loss": 0.6396,
      "step": 38985
    },
    {
      "epoch": 0.6070041333572046,
      "grad_norm": 3.172049045562744,
      "learning_rate": 4.1369364644958294e-07,
      "loss": 0.7329,
      "step": 38990
    },
    {
      "epoch": 0.607081974359174,
      "grad_norm": 3.1179158687591553,
      "learning_rate": 4.136117074449779e-07,
      "loss": 0.7829,
      "step": 38995
    },
    {
      "epoch": 0.6071598153611433,
      "grad_norm": 2.904696226119995,
      "learning_rate": 4.1352976844037294e-07,
      "loss": 0.7333,
      "step": 39000
    },
    {
      "epoch": 0.6072376563631127,
      "grad_norm": 3.9968748092651367,
      "learning_rate": 4.13447829435768e-07,
      "loss": 0.7187,
      "step": 39005
    },
    {
      "epoch": 0.6073154973650821,
      "grad_norm": 8.46354866027832,
      "learning_rate": 4.1336589043116303e-07,
      "loss": 0.8063,
      "step": 39010
    },
    {
      "epoch": 0.6073933383670514,
      "grad_norm": 8.95866870880127,
      "learning_rate": 4.1328395142655805e-07,
      "loss": 0.7486,
      "step": 39015
    },
    {
      "epoch": 0.6074711793690208,
      "grad_norm": 7.068979263305664,
      "learning_rate": 4.132020124219531e-07,
      "loss": 0.6715,
      "step": 39020
    },
    {
      "epoch": 0.6075490203709902,
      "grad_norm": 5.791064262390137,
      "learning_rate": 4.131200734173481e-07,
      "loss": 0.6836,
      "step": 39025
    },
    {
      "epoch": 0.6076268613729596,
      "grad_norm": 2.763953685760498,
      "learning_rate": 4.1303813441274317e-07,
      "loss": 0.7173,
      "step": 39030
    },
    {
      "epoch": 0.607704702374929,
      "grad_norm": 3.992194414138794,
      "learning_rate": 4.1295619540813814e-07,
      "loss": 0.7895,
      "step": 39035
    },
    {
      "epoch": 0.6077825433768983,
      "grad_norm": 3.8581206798553467,
      "learning_rate": 4.1287425640353316e-07,
      "loss": 0.7484,
      "step": 39040
    },
    {
      "epoch": 0.6078603843788677,
      "grad_norm": 3.467031955718994,
      "learning_rate": 4.1279231739892824e-07,
      "loss": 0.6839,
      "step": 39045
    },
    {
      "epoch": 0.6079382253808371,
      "grad_norm": 3.9381110668182373,
      "learning_rate": 4.1271037839432326e-07,
      "loss": 0.8818,
      "step": 39050
    },
    {
      "epoch": 0.6080160663828065,
      "grad_norm": 3.0449042320251465,
      "learning_rate": 4.1262843938971823e-07,
      "loss": 0.741,
      "step": 39055
    },
    {
      "epoch": 0.6080939073847759,
      "grad_norm": 4.400388240814209,
      "learning_rate": 4.125465003851133e-07,
      "loss": 0.78,
      "step": 39060
    },
    {
      "epoch": 0.6081717483867453,
      "grad_norm": 3.2337820529937744,
      "learning_rate": 4.124645613805083e-07,
      "loss": 0.7144,
      "step": 39065
    },
    {
      "epoch": 0.6082495893887147,
      "grad_norm": 8.551438331604004,
      "learning_rate": 4.123826223759034e-07,
      "loss": 0.6293,
      "step": 39070
    },
    {
      "epoch": 0.608327430390684,
      "grad_norm": 3.334404706954956,
      "learning_rate": 4.1230068337129837e-07,
      "loss": 0.7764,
      "step": 39075
    },
    {
      "epoch": 0.6084052713926533,
      "grad_norm": 5.051881790161133,
      "learning_rate": 4.122187443666934e-07,
      "loss": 0.7684,
      "step": 39080
    },
    {
      "epoch": 0.6084831123946227,
      "grad_norm": 3.531160593032837,
      "learning_rate": 4.1213680536208847e-07,
      "loss": 0.7966,
      "step": 39085
    },
    {
      "epoch": 0.6085609533965921,
      "grad_norm": 6.416616916656494,
      "learning_rate": 4.120548663574835e-07,
      "loss": 0.6872,
      "step": 39090
    },
    {
      "epoch": 0.6086387943985615,
      "grad_norm": 4.0549540519714355,
      "learning_rate": 4.1197292735287846e-07,
      "loss": 0.6708,
      "step": 39095
    },
    {
      "epoch": 0.6087166354005309,
      "grad_norm": 9.749567031860352,
      "learning_rate": 4.1189098834827353e-07,
      "loss": 0.8016,
      "step": 39100
    },
    {
      "epoch": 0.6087944764025003,
      "grad_norm": 2.7380635738372803,
      "learning_rate": 4.1180904934366855e-07,
      "loss": 0.7255,
      "step": 39105
    },
    {
      "epoch": 0.6088723174044697,
      "grad_norm": 2.5923984050750732,
      "learning_rate": 4.117271103390636e-07,
      "loss": 0.7068,
      "step": 39110
    },
    {
      "epoch": 0.608950158406439,
      "grad_norm": 3.785151481628418,
      "learning_rate": 4.116451713344586e-07,
      "loss": 0.8347,
      "step": 39115
    },
    {
      "epoch": 0.6090279994084083,
      "grad_norm": 5.512859344482422,
      "learning_rate": 4.115632323298536e-07,
      "loss": 0.7567,
      "step": 39120
    },
    {
      "epoch": 0.6091058404103777,
      "grad_norm": 3.9963438510894775,
      "learning_rate": 4.1148129332524864e-07,
      "loss": 0.8284,
      "step": 39125
    },
    {
      "epoch": 0.6091836814123471,
      "grad_norm": 6.539175987243652,
      "learning_rate": 4.113993543206437e-07,
      "loss": 0.7639,
      "step": 39130
    },
    {
      "epoch": 0.6092615224143165,
      "grad_norm": 5.0428667068481445,
      "learning_rate": 4.113174153160387e-07,
      "loss": 0.6662,
      "step": 39135
    },
    {
      "epoch": 0.6093393634162859,
      "grad_norm": 6.604543685913086,
      "learning_rate": 4.1123547631143376e-07,
      "loss": 0.6676,
      "step": 39140
    },
    {
      "epoch": 0.6094172044182553,
      "grad_norm": 6.024013519287109,
      "learning_rate": 4.111535373068288e-07,
      "loss": 0.8965,
      "step": 39145
    },
    {
      "epoch": 0.6094950454202247,
      "grad_norm": 3.7081356048583984,
      "learning_rate": 4.110715983022238e-07,
      "loss": 0.689,
      "step": 39150
    },
    {
      "epoch": 0.609572886422194,
      "grad_norm": 4.272152900695801,
      "learning_rate": 4.1098965929761883e-07,
      "loss": 0.8184,
      "step": 39155
    },
    {
      "epoch": 0.6096507274241634,
      "grad_norm": 3.5037147998809814,
      "learning_rate": 4.1090772029301385e-07,
      "loss": 0.7017,
      "step": 39160
    },
    {
      "epoch": 0.6097285684261328,
      "grad_norm": 3.119861125946045,
      "learning_rate": 4.1082578128840887e-07,
      "loss": 0.8447,
      "step": 39165
    },
    {
      "epoch": 0.6098064094281022,
      "grad_norm": 3.786356210708618,
      "learning_rate": 4.1074384228380395e-07,
      "loss": 0.7679,
      "step": 39170
    },
    {
      "epoch": 0.6098842504300716,
      "grad_norm": 3.382629632949829,
      "learning_rate": 4.106619032791989e-07,
      "loss": 0.6912,
      "step": 39175
    },
    {
      "epoch": 0.6099620914320409,
      "grad_norm": 3.726262092590332,
      "learning_rate": 4.1057996427459394e-07,
      "loss": 0.8807,
      "step": 39180
    },
    {
      "epoch": 0.6100399324340103,
      "grad_norm": 3.108146905899048,
      "learning_rate": 4.10498025269989e-07,
      "loss": 0.7186,
      "step": 39185
    },
    {
      "epoch": 0.6101177734359796,
      "grad_norm": 5.5432024002075195,
      "learning_rate": 4.1041608626538403e-07,
      "loss": 0.675,
      "step": 39190
    },
    {
      "epoch": 0.610195614437949,
      "grad_norm": 6.811224937438965,
      "learning_rate": 4.103341472607791e-07,
      "loss": 0.7813,
      "step": 39195
    },
    {
      "epoch": 0.6102734554399184,
      "grad_norm": 8.459604263305664,
      "learning_rate": 4.102522082561741e-07,
      "loss": 0.73,
      "step": 39200
    },
    {
      "epoch": 0.6103512964418878,
      "grad_norm": 4.586516380310059,
      "learning_rate": 4.101702692515691e-07,
      "loss": 0.7326,
      "step": 39205
    },
    {
      "epoch": 0.6104291374438572,
      "grad_norm": 2.645754098892212,
      "learning_rate": 4.100883302469642e-07,
      "loss": 0.7312,
      "step": 39210
    },
    {
      "epoch": 0.6105069784458266,
      "grad_norm": 4.536027431488037,
      "learning_rate": 4.100063912423592e-07,
      "loss": 0.7188,
      "step": 39215
    },
    {
      "epoch": 0.610584819447796,
      "grad_norm": 3.4416491985321045,
      "learning_rate": 4.0992445223775416e-07,
      "loss": 0.8266,
      "step": 39220
    },
    {
      "epoch": 0.6106626604497654,
      "grad_norm": 4.604366302490234,
      "learning_rate": 4.0984251323314924e-07,
      "loss": 0.7332,
      "step": 39225
    },
    {
      "epoch": 0.6107405014517346,
      "grad_norm": 2.8680050373077393,
      "learning_rate": 4.0976057422854426e-07,
      "loss": 0.7768,
      "step": 39230
    },
    {
      "epoch": 0.610818342453704,
      "grad_norm": 3.424793243408203,
      "learning_rate": 4.096786352239393e-07,
      "loss": 0.7583,
      "step": 39235
    },
    {
      "epoch": 0.6108961834556734,
      "grad_norm": 4.334976673126221,
      "learning_rate": 4.095966962193343e-07,
      "loss": 0.7324,
      "step": 39240
    },
    {
      "epoch": 0.6109740244576428,
      "grad_norm": 3.4201934337615967,
      "learning_rate": 4.0951475721472933e-07,
      "loss": 0.7833,
      "step": 39245
    },
    {
      "epoch": 0.6110518654596122,
      "grad_norm": 5.36106538772583,
      "learning_rate": 4.0943281821012435e-07,
      "loss": 0.7375,
      "step": 39250
    },
    {
      "epoch": 0.6111297064615816,
      "grad_norm": 3.453793525695801,
      "learning_rate": 4.093508792055194e-07,
      "loss": 0.6779,
      "step": 39255
    },
    {
      "epoch": 0.611207547463551,
      "grad_norm": 8.306373596191406,
      "learning_rate": 4.092689402009144e-07,
      "loss": 0.7889,
      "step": 39260
    },
    {
      "epoch": 0.6112853884655203,
      "grad_norm": 2.7198047637939453,
      "learning_rate": 4.0918700119630947e-07,
      "loss": 0.7448,
      "step": 39265
    },
    {
      "epoch": 0.6113632294674897,
      "grad_norm": 4.299631595611572,
      "learning_rate": 4.091050621917045e-07,
      "loss": 0.6995,
      "step": 39270
    },
    {
      "epoch": 0.6114410704694591,
      "grad_norm": 4.193734645843506,
      "learning_rate": 4.090231231870995e-07,
      "loss": 0.7543,
      "step": 39275
    },
    {
      "epoch": 0.6115189114714285,
      "grad_norm": 3.627577066421509,
      "learning_rate": 4.0894118418249453e-07,
      "loss": 0.8473,
      "step": 39280
    },
    {
      "epoch": 0.6115967524733978,
      "grad_norm": 3.0676145553588867,
      "learning_rate": 4.0885924517788956e-07,
      "loss": 0.6746,
      "step": 39285
    },
    {
      "epoch": 0.6116745934753672,
      "grad_norm": 2.919372797012329,
      "learning_rate": 4.087773061732846e-07,
      "loss": 0.7806,
      "step": 39290
    },
    {
      "epoch": 0.6117524344773366,
      "grad_norm": 4.465323448181152,
      "learning_rate": 4.0869536716867965e-07,
      "loss": 0.747,
      "step": 39295
    },
    {
      "epoch": 0.611830275479306,
      "grad_norm": 13.452363967895508,
      "learning_rate": 4.086134281640746e-07,
      "loss": 0.7914,
      "step": 39300
    },
    {
      "epoch": 0.6119081164812753,
      "grad_norm": 2.707576274871826,
      "learning_rate": 4.0853148915946964e-07,
      "loss": 0.7901,
      "step": 39305
    },
    {
      "epoch": 0.6119859574832447,
      "grad_norm": 9.56675910949707,
      "learning_rate": 4.084495501548647e-07,
      "loss": 0.6667,
      "step": 39310
    },
    {
      "epoch": 0.6120637984852141,
      "grad_norm": 3.064061403274536,
      "learning_rate": 4.0836761115025974e-07,
      "loss": 0.7509,
      "step": 39315
    },
    {
      "epoch": 0.6121416394871835,
      "grad_norm": 8.183847427368164,
      "learning_rate": 4.0828567214565476e-07,
      "loss": 0.7302,
      "step": 39320
    },
    {
      "epoch": 0.6122194804891529,
      "grad_norm": 7.25828742980957,
      "learning_rate": 4.082037331410498e-07,
      "loss": 0.8113,
      "step": 39325
    },
    {
      "epoch": 0.6122973214911223,
      "grad_norm": 3.5414304733276367,
      "learning_rate": 4.081217941364448e-07,
      "loss": 0.761,
      "step": 39330
    },
    {
      "epoch": 0.6123751624930917,
      "grad_norm": 2.6410717964172363,
      "learning_rate": 4.080398551318399e-07,
      "loss": 0.7069,
      "step": 39335
    },
    {
      "epoch": 0.612453003495061,
      "grad_norm": 2.6031923294067383,
      "learning_rate": 4.0795791612723485e-07,
      "loss": 0.7424,
      "step": 39340
    },
    {
      "epoch": 0.6125308444970303,
      "grad_norm": 4.200246810913086,
      "learning_rate": 4.0787597712262987e-07,
      "loss": 0.6833,
      "step": 39345
    },
    {
      "epoch": 0.6126086854989997,
      "grad_norm": 4.823015213012695,
      "learning_rate": 4.0779403811802495e-07,
      "loss": 0.7407,
      "step": 39350
    },
    {
      "epoch": 0.6126865265009691,
      "grad_norm": 3.403510808944702,
      "learning_rate": 4.0771209911341997e-07,
      "loss": 0.7739,
      "step": 39355
    },
    {
      "epoch": 0.6127643675029385,
      "grad_norm": 6.34179162979126,
      "learning_rate": 4.0763016010881494e-07,
      "loss": 0.799,
      "step": 39360
    },
    {
      "epoch": 0.6128422085049079,
      "grad_norm": 12.049829483032227,
      "learning_rate": 4.0754822110421e-07,
      "loss": 0.6701,
      "step": 39365
    },
    {
      "epoch": 0.6129200495068773,
      "grad_norm": 4.598782062530518,
      "learning_rate": 4.0746628209960503e-07,
      "loss": 0.7164,
      "step": 39370
    },
    {
      "epoch": 0.6129978905088467,
      "grad_norm": 3.4022891521453857,
      "learning_rate": 4.073843430950001e-07,
      "loss": 0.6705,
      "step": 39375
    },
    {
      "epoch": 0.613075731510816,
      "grad_norm": 3.45033860206604,
      "learning_rate": 4.073024040903951e-07,
      "loss": 0.6834,
      "step": 39380
    },
    {
      "epoch": 0.6131535725127853,
      "grad_norm": 4.8961501121521,
      "learning_rate": 4.072204650857901e-07,
      "loss": 0.6667,
      "step": 39385
    },
    {
      "epoch": 0.6132314135147547,
      "grad_norm": 2.4274744987487793,
      "learning_rate": 4.071385260811852e-07,
      "loss": 0.7223,
      "step": 39390
    },
    {
      "epoch": 0.6133092545167241,
      "grad_norm": 5.743806838989258,
      "learning_rate": 4.070565870765802e-07,
      "loss": 0.8045,
      "step": 39395
    },
    {
      "epoch": 0.6133870955186935,
      "grad_norm": 3.6674177646636963,
      "learning_rate": 4.0697464807197517e-07,
      "loss": 0.7649,
      "step": 39400
    },
    {
      "epoch": 0.6134649365206629,
      "grad_norm": 4.55914306640625,
      "learning_rate": 4.0689270906737024e-07,
      "loss": 0.7797,
      "step": 39405
    },
    {
      "epoch": 0.6135427775226323,
      "grad_norm": 3.4974446296691895,
      "learning_rate": 4.0681077006276526e-07,
      "loss": 0.708,
      "step": 39410
    },
    {
      "epoch": 0.6136206185246017,
      "grad_norm": 6.238338947296143,
      "learning_rate": 4.067288310581603e-07,
      "loss": 0.8547,
      "step": 39415
    },
    {
      "epoch": 0.613698459526571,
      "grad_norm": 7.216058254241943,
      "learning_rate": 4.066468920535553e-07,
      "loss": 0.7372,
      "step": 39420
    },
    {
      "epoch": 0.6137763005285404,
      "grad_norm": 4.798983573913574,
      "learning_rate": 4.0656495304895033e-07,
      "loss": 0.629,
      "step": 39425
    },
    {
      "epoch": 0.6138541415305098,
      "grad_norm": 4.304426670074463,
      "learning_rate": 4.0648301404434535e-07,
      "loss": 0.6898,
      "step": 39430
    },
    {
      "epoch": 0.6139319825324792,
      "grad_norm": 5.827239513397217,
      "learning_rate": 4.064010750397404e-07,
      "loss": 0.7477,
      "step": 39435
    },
    {
      "epoch": 0.6140098235344486,
      "grad_norm": 4.135549545288086,
      "learning_rate": 4.063191360351354e-07,
      "loss": 0.7737,
      "step": 39440
    },
    {
      "epoch": 0.6140876645364179,
      "grad_norm": 3.454678773880005,
      "learning_rate": 4.0623719703053047e-07,
      "loss": 0.6749,
      "step": 39445
    },
    {
      "epoch": 0.6141655055383873,
      "grad_norm": 5.33576774597168,
      "learning_rate": 4.061552580259255e-07,
      "loss": 0.8597,
      "step": 39450
    },
    {
      "epoch": 0.6142433465403566,
      "grad_norm": 5.088623046875,
      "learning_rate": 4.060733190213205e-07,
      "loss": 0.804,
      "step": 39455
    },
    {
      "epoch": 0.614321187542326,
      "grad_norm": 4.081348896026611,
      "learning_rate": 4.0599138001671553e-07,
      "loss": 0.6735,
      "step": 39460
    },
    {
      "epoch": 0.6143990285442954,
      "grad_norm": 4.2015814781188965,
      "learning_rate": 4.0590944101211056e-07,
      "loss": 0.6777,
      "step": 39465
    },
    {
      "epoch": 0.6144768695462648,
      "grad_norm": 4.993946075439453,
      "learning_rate": 4.058275020075056e-07,
      "loss": 0.7869,
      "step": 39470
    },
    {
      "epoch": 0.6145547105482342,
      "grad_norm": 2.431502103805542,
      "learning_rate": 4.0574556300290065e-07,
      "loss": 0.6906,
      "step": 39475
    },
    {
      "epoch": 0.6146325515502036,
      "grad_norm": 3.5009548664093018,
      "learning_rate": 4.056636239982956e-07,
      "loss": 0.7927,
      "step": 39480
    },
    {
      "epoch": 0.614710392552173,
      "grad_norm": 5.608779430389404,
      "learning_rate": 4.0558168499369064e-07,
      "loss": 0.6917,
      "step": 39485
    },
    {
      "epoch": 0.6147882335541424,
      "grad_norm": 4.179677963256836,
      "learning_rate": 4.054997459890857e-07,
      "loss": 0.7198,
      "step": 39490
    },
    {
      "epoch": 0.6148660745561116,
      "grad_norm": 5.763132572174072,
      "learning_rate": 4.0541780698448074e-07,
      "loss": 0.7783,
      "step": 39495
    },
    {
      "epoch": 0.614943915558081,
      "grad_norm": 4.71073055267334,
      "learning_rate": 4.053358679798758e-07,
      "loss": 0.7058,
      "step": 39500
    },
    {
      "epoch": 0.6150217565600504,
      "grad_norm": 5.335464000701904,
      "learning_rate": 4.052539289752708e-07,
      "loss": 0.6371,
      "step": 39505
    },
    {
      "epoch": 0.6150995975620198,
      "grad_norm": 2.279665231704712,
      "learning_rate": 4.051719899706658e-07,
      "loss": 0.7493,
      "step": 39510
    },
    {
      "epoch": 0.6151774385639892,
      "grad_norm": 5.083649158477783,
      "learning_rate": 4.050900509660609e-07,
      "loss": 0.6666,
      "step": 39515
    },
    {
      "epoch": 0.6152552795659586,
      "grad_norm": 4.002017498016357,
      "learning_rate": 4.050081119614559e-07,
      "loss": 0.7291,
      "step": 39520
    },
    {
      "epoch": 0.615333120567928,
      "grad_norm": 4.288511276245117,
      "learning_rate": 4.0492617295685087e-07,
      "loss": 0.8346,
      "step": 39525
    },
    {
      "epoch": 0.6154109615698973,
      "grad_norm": 3.3616411685943604,
      "learning_rate": 4.0484423395224595e-07,
      "loss": 0.7292,
      "step": 39530
    },
    {
      "epoch": 0.6154888025718667,
      "grad_norm": 3.887479782104492,
      "learning_rate": 4.0476229494764097e-07,
      "loss": 0.692,
      "step": 39535
    },
    {
      "epoch": 0.6155666435738361,
      "grad_norm": 2.7648532390594482,
      "learning_rate": 4.04680355943036e-07,
      "loss": 0.6177,
      "step": 39540
    },
    {
      "epoch": 0.6156444845758055,
      "grad_norm": 4.084407806396484,
      "learning_rate": 4.04598416938431e-07,
      "loss": 0.6493,
      "step": 39545
    },
    {
      "epoch": 0.6157223255777748,
      "grad_norm": 3.3023629188537598,
      "learning_rate": 4.0451647793382603e-07,
      "loss": 0.8276,
      "step": 39550
    },
    {
      "epoch": 0.6158001665797442,
      "grad_norm": 5.282424449920654,
      "learning_rate": 4.0443453892922106e-07,
      "loss": 0.6892,
      "step": 39555
    },
    {
      "epoch": 0.6158780075817136,
      "grad_norm": 2.984018325805664,
      "learning_rate": 4.0435259992461613e-07,
      "loss": 0.7198,
      "step": 39560
    },
    {
      "epoch": 0.615955848583683,
      "grad_norm": 5.932767391204834,
      "learning_rate": 4.042706609200111e-07,
      "loss": 0.8458,
      "step": 39565
    },
    {
      "epoch": 0.6160336895856523,
      "grad_norm": 4.437289714813232,
      "learning_rate": 4.041887219154062e-07,
      "loss": 0.6845,
      "step": 39570
    },
    {
      "epoch": 0.6161115305876217,
      "grad_norm": 3.6258370876312256,
      "learning_rate": 4.041067829108012e-07,
      "loss": 0.7115,
      "step": 39575
    },
    {
      "epoch": 0.6161893715895911,
      "grad_norm": 2.8586692810058594,
      "learning_rate": 4.040248439061962e-07,
      "loss": 0.6244,
      "step": 39580
    },
    {
      "epoch": 0.6162672125915605,
      "grad_norm": 3.697373151779175,
      "learning_rate": 4.0394290490159124e-07,
      "loss": 0.8051,
      "step": 39585
    },
    {
      "epoch": 0.6163450535935299,
      "grad_norm": 6.740497589111328,
      "learning_rate": 4.0386096589698626e-07,
      "loss": 0.7598,
      "step": 39590
    },
    {
      "epoch": 0.6164228945954993,
      "grad_norm": 7.085832118988037,
      "learning_rate": 4.037790268923813e-07,
      "loss": 0.7927,
      "step": 39595
    },
    {
      "epoch": 0.6165007355974687,
      "grad_norm": 3.321800470352173,
      "learning_rate": 4.0369708788777636e-07,
      "loss": 0.7902,
      "step": 39600
    },
    {
      "epoch": 0.6165785765994379,
      "grad_norm": 7.254165172576904,
      "learning_rate": 4.0361514888317133e-07,
      "loss": 0.731,
      "step": 39605
    },
    {
      "epoch": 0.6166564176014073,
      "grad_norm": 3.4321482181549072,
      "learning_rate": 4.0353320987856635e-07,
      "loss": 0.7292,
      "step": 39610
    },
    {
      "epoch": 0.6167342586033767,
      "grad_norm": 2.917635917663574,
      "learning_rate": 4.034512708739614e-07,
      "loss": 0.7501,
      "step": 39615
    },
    {
      "epoch": 0.6168120996053461,
      "grad_norm": 3.5796961784362793,
      "learning_rate": 4.0336933186935645e-07,
      "loss": 0.7169,
      "step": 39620
    },
    {
      "epoch": 0.6168899406073155,
      "grad_norm": 3.866138458251953,
      "learning_rate": 4.032873928647514e-07,
      "loss": 0.6251,
      "step": 39625
    },
    {
      "epoch": 0.6169677816092849,
      "grad_norm": 3.7258641719818115,
      "learning_rate": 4.032054538601465e-07,
      "loss": 0.7973,
      "step": 39630
    },
    {
      "epoch": 0.6170456226112543,
      "grad_norm": 2.8312764167785645,
      "learning_rate": 4.031235148555415e-07,
      "loss": 0.8247,
      "step": 39635
    },
    {
      "epoch": 0.6171234636132237,
      "grad_norm": 3.36230206489563,
      "learning_rate": 4.030415758509366e-07,
      "loss": 0.6992,
      "step": 39640
    },
    {
      "epoch": 0.617201304615193,
      "grad_norm": 8.737380027770996,
      "learning_rate": 4.0295963684633156e-07,
      "loss": 0.6648,
      "step": 39645
    },
    {
      "epoch": 0.6172791456171624,
      "grad_norm": 6.160430431365967,
      "learning_rate": 4.028776978417266e-07,
      "loss": 0.7707,
      "step": 39650
    },
    {
      "epoch": 0.6173569866191317,
      "grad_norm": 6.520876407623291,
      "learning_rate": 4.0279575883712165e-07,
      "loss": 0.7785,
      "step": 39655
    },
    {
      "epoch": 0.6174348276211011,
      "grad_norm": 4.372855186462402,
      "learning_rate": 4.027138198325167e-07,
      "loss": 0.7577,
      "step": 39660
    },
    {
      "epoch": 0.6175126686230705,
      "grad_norm": 7.884913921356201,
      "learning_rate": 4.0263188082791164e-07,
      "loss": 0.8251,
      "step": 39665
    },
    {
      "epoch": 0.6175905096250399,
      "grad_norm": 3.289794445037842,
      "learning_rate": 4.025499418233067e-07,
      "loss": 0.6722,
      "step": 39670
    },
    {
      "epoch": 0.6176683506270093,
      "grad_norm": 8.901487350463867,
      "learning_rate": 4.0246800281870174e-07,
      "loss": 0.7341,
      "step": 39675
    },
    {
      "epoch": 0.6177461916289787,
      "grad_norm": 3.3814287185668945,
      "learning_rate": 4.0238606381409676e-07,
      "loss": 0.818,
      "step": 39680
    },
    {
      "epoch": 0.617824032630948,
      "grad_norm": 4.139296531677246,
      "learning_rate": 4.023041248094918e-07,
      "loss": 0.6896,
      "step": 39685
    },
    {
      "epoch": 0.6179018736329174,
      "grad_norm": 3.267014741897583,
      "learning_rate": 4.022221858048868e-07,
      "loss": 0.797,
      "step": 39690
    },
    {
      "epoch": 0.6179797146348868,
      "grad_norm": 5.44881534576416,
      "learning_rate": 4.021402468002819e-07,
      "loss": 0.8084,
      "step": 39695
    },
    {
      "epoch": 0.6180575556368562,
      "grad_norm": 4.601943016052246,
      "learning_rate": 4.020583077956769e-07,
      "loss": 0.825,
      "step": 39700
    },
    {
      "epoch": 0.6181353966388256,
      "grad_norm": 3.6288840770721436,
      "learning_rate": 4.0197636879107187e-07,
      "loss": 0.791,
      "step": 39705
    },
    {
      "epoch": 0.6182132376407949,
      "grad_norm": 3.2632014751434326,
      "learning_rate": 4.0189442978646695e-07,
      "loss": 0.7061,
      "step": 39710
    },
    {
      "epoch": 0.6182910786427643,
      "grad_norm": 3.9054431915283203,
      "learning_rate": 4.0181249078186197e-07,
      "loss": 0.8127,
      "step": 39715
    },
    {
      "epoch": 0.6183689196447336,
      "grad_norm": 4.7405266761779785,
      "learning_rate": 4.01730551777257e-07,
      "loss": 0.8081,
      "step": 39720
    },
    {
      "epoch": 0.618446760646703,
      "grad_norm": 3.8895163536071777,
      "learning_rate": 4.01648612772652e-07,
      "loss": 0.9205,
      "step": 39725
    },
    {
      "epoch": 0.6185246016486724,
      "grad_norm": 3.0387494564056396,
      "learning_rate": 4.0156667376804704e-07,
      "loss": 0.7789,
      "step": 39730
    },
    {
      "epoch": 0.6186024426506418,
      "grad_norm": 5.294724941253662,
      "learning_rate": 4.0148473476344206e-07,
      "loss": 0.7387,
      "step": 39735
    },
    {
      "epoch": 0.6186802836526112,
      "grad_norm": 6.25783634185791,
      "learning_rate": 4.0140279575883713e-07,
      "loss": 0.7435,
      "step": 39740
    },
    {
      "epoch": 0.6187581246545806,
      "grad_norm": 3.4548604488372803,
      "learning_rate": 4.013208567542321e-07,
      "loss": 0.8438,
      "step": 39745
    },
    {
      "epoch": 0.61883596565655,
      "grad_norm": 5.93376350402832,
      "learning_rate": 4.012389177496271e-07,
      "loss": 0.7913,
      "step": 39750
    },
    {
      "epoch": 0.6189138066585194,
      "grad_norm": 3.2155590057373047,
      "learning_rate": 4.011569787450222e-07,
      "loss": 0.6255,
      "step": 39755
    },
    {
      "epoch": 0.6189916476604886,
      "grad_norm": 3.2723147869110107,
      "learning_rate": 4.010750397404172e-07,
      "loss": 0.8119,
      "step": 39760
    },
    {
      "epoch": 0.619069488662458,
      "grad_norm": 3.218729257583618,
      "learning_rate": 4.0099310073581224e-07,
      "loss": 0.8122,
      "step": 39765
    },
    {
      "epoch": 0.6191473296644274,
      "grad_norm": 3.0231781005859375,
      "learning_rate": 4.0091116173120726e-07,
      "loss": 0.7342,
      "step": 39770
    },
    {
      "epoch": 0.6192251706663968,
      "grad_norm": 3.9648866653442383,
      "learning_rate": 4.008292227266023e-07,
      "loss": 0.762,
      "step": 39775
    },
    {
      "epoch": 0.6193030116683662,
      "grad_norm": 3.370361089706421,
      "learning_rate": 4.0074728372199736e-07,
      "loss": 0.7663,
      "step": 39780
    },
    {
      "epoch": 0.6193808526703356,
      "grad_norm": 5.944691181182861,
      "learning_rate": 4.0066534471739233e-07,
      "loss": 0.7712,
      "step": 39785
    },
    {
      "epoch": 0.619458693672305,
      "grad_norm": 3.8935952186584473,
      "learning_rate": 4.0058340571278735e-07,
      "loss": 0.8311,
      "step": 39790
    },
    {
      "epoch": 0.6195365346742743,
      "grad_norm": 3.2222964763641357,
      "learning_rate": 4.0050146670818243e-07,
      "loss": 0.7275,
      "step": 39795
    },
    {
      "epoch": 0.6196143756762437,
      "grad_norm": 10.95319938659668,
      "learning_rate": 4.0041952770357745e-07,
      "loss": 0.8925,
      "step": 39800
    },
    {
      "epoch": 0.6196922166782131,
      "grad_norm": 6.563601493835449,
      "learning_rate": 4.003375886989724e-07,
      "loss": 0.8799,
      "step": 39805
    },
    {
      "epoch": 0.6197700576801825,
      "grad_norm": 4.75308084487915,
      "learning_rate": 4.002556496943675e-07,
      "loss": 0.7904,
      "step": 39810
    },
    {
      "epoch": 0.6198478986821518,
      "grad_norm": 3.868544578552246,
      "learning_rate": 4.001737106897625e-07,
      "loss": 0.6451,
      "step": 39815
    },
    {
      "epoch": 0.6199257396841212,
      "grad_norm": 2.6562716960906982,
      "learning_rate": 4.000917716851576e-07,
      "loss": 0.7751,
      "step": 39820
    },
    {
      "epoch": 0.6200035806860906,
      "grad_norm": 4.331404685974121,
      "learning_rate": 4.000098326805526e-07,
      "loss": 0.8126,
      "step": 39825
    },
    {
      "epoch": 0.62008142168806,
      "grad_norm": 4.255936622619629,
      "learning_rate": 3.999278936759476e-07,
      "loss": 0.7919,
      "step": 39830
    },
    {
      "epoch": 0.6201592626900293,
      "grad_norm": 3.312748670578003,
      "learning_rate": 3.9984595467134265e-07,
      "loss": 0.7443,
      "step": 39835
    },
    {
      "epoch": 0.6202371036919987,
      "grad_norm": 5.283825874328613,
      "learning_rate": 3.997640156667377e-07,
      "loss": 0.7273,
      "step": 39840
    },
    {
      "epoch": 0.6203149446939681,
      "grad_norm": 4.295399188995361,
      "learning_rate": 3.996820766621327e-07,
      "loss": 0.6971,
      "step": 39845
    },
    {
      "epoch": 0.6203927856959375,
      "grad_norm": 5.522240161895752,
      "learning_rate": 3.996001376575277e-07,
      "loss": 0.7391,
      "step": 39850
    },
    {
      "epoch": 0.6204706266979069,
      "grad_norm": 3.4844582080841064,
      "learning_rate": 3.9951819865292274e-07,
      "loss": 0.6574,
      "step": 39855
    },
    {
      "epoch": 0.6205484676998763,
      "grad_norm": 2.8905818462371826,
      "learning_rate": 3.9943625964831776e-07,
      "loss": 0.7196,
      "step": 39860
    },
    {
      "epoch": 0.6206263087018457,
      "grad_norm": 3.960275888442993,
      "learning_rate": 3.9935432064371284e-07,
      "loss": 0.8114,
      "step": 39865
    },
    {
      "epoch": 0.6207041497038149,
      "grad_norm": 4.139747619628906,
      "learning_rate": 3.992723816391078e-07,
      "loss": 0.7407,
      "step": 39870
    },
    {
      "epoch": 0.6207819907057843,
      "grad_norm": 4.535872936248779,
      "learning_rate": 3.9919044263450283e-07,
      "loss": 0.7183,
      "step": 39875
    },
    {
      "epoch": 0.6208598317077537,
      "grad_norm": 3.249366044998169,
      "learning_rate": 3.991085036298979e-07,
      "loss": 0.7061,
      "step": 39880
    },
    {
      "epoch": 0.6209376727097231,
      "grad_norm": 3.505553722381592,
      "learning_rate": 3.9902656462529293e-07,
      "loss": 0.6117,
      "step": 39885
    },
    {
      "epoch": 0.6210155137116925,
      "grad_norm": 3.244825601577759,
      "learning_rate": 3.9894462562068795e-07,
      "loss": 0.7458,
      "step": 39890
    },
    {
      "epoch": 0.6210933547136619,
      "grad_norm": 3.5578088760375977,
      "learning_rate": 3.9886268661608297e-07,
      "loss": 0.7246,
      "step": 39895
    },
    {
      "epoch": 0.6211711957156313,
      "grad_norm": 3.017503261566162,
      "learning_rate": 3.98780747611478e-07,
      "loss": 0.7797,
      "step": 39900
    },
    {
      "epoch": 0.6212490367176007,
      "grad_norm": 5.441774368286133,
      "learning_rate": 3.9869880860687307e-07,
      "loss": 0.7816,
      "step": 39905
    },
    {
      "epoch": 0.62132687771957,
      "grad_norm": 8.501574516296387,
      "learning_rate": 3.9861686960226804e-07,
      "loss": 0.7798,
      "step": 39910
    },
    {
      "epoch": 0.6214047187215394,
      "grad_norm": 3.5010392665863037,
      "learning_rate": 3.9853493059766306e-07,
      "loss": 0.8378,
      "step": 39915
    },
    {
      "epoch": 0.6214825597235087,
      "grad_norm": 8.38367748260498,
      "learning_rate": 3.9845299159305813e-07,
      "loss": 0.6319,
      "step": 39920
    },
    {
      "epoch": 0.6215604007254781,
      "grad_norm": 3.3855247497558594,
      "learning_rate": 3.9837105258845316e-07,
      "loss": 0.7391,
      "step": 39925
    },
    {
      "epoch": 0.6216382417274475,
      "grad_norm": 2.4978036880493164,
      "learning_rate": 3.982891135838481e-07,
      "loss": 0.7405,
      "step": 39930
    },
    {
      "epoch": 0.6217160827294169,
      "grad_norm": 4.2187910079956055,
      "learning_rate": 3.982071745792432e-07,
      "loss": 0.8396,
      "step": 39935
    },
    {
      "epoch": 0.6217939237313863,
      "grad_norm": 21.20357894897461,
      "learning_rate": 3.981252355746382e-07,
      "loss": 0.7914,
      "step": 39940
    },
    {
      "epoch": 0.6218717647333557,
      "grad_norm": 4.16295862197876,
      "learning_rate": 3.980432965700333e-07,
      "loss": 0.7373,
      "step": 39945
    },
    {
      "epoch": 0.621949605735325,
      "grad_norm": 3.6028060913085938,
      "learning_rate": 3.9796135756542826e-07,
      "loss": 0.8967,
      "step": 39950
    },
    {
      "epoch": 0.6220274467372944,
      "grad_norm": 6.922677516937256,
      "learning_rate": 3.978794185608233e-07,
      "loss": 0.6442,
      "step": 39955
    },
    {
      "epoch": 0.6221052877392638,
      "grad_norm": 2.784919023513794,
      "learning_rate": 3.9779747955621836e-07,
      "loss": 0.6532,
      "step": 39960
    },
    {
      "epoch": 0.6221831287412332,
      "grad_norm": 6.433320999145508,
      "learning_rate": 3.977155405516134e-07,
      "loss": 0.7698,
      "step": 39965
    },
    {
      "epoch": 0.6222609697432026,
      "grad_norm": 3.5730509757995605,
      "learning_rate": 3.9763360154700835e-07,
      "loss": 0.8455,
      "step": 39970
    },
    {
      "epoch": 0.6223388107451719,
      "grad_norm": 3.1607539653778076,
      "learning_rate": 3.9755166254240343e-07,
      "loss": 0.6986,
      "step": 39975
    },
    {
      "epoch": 0.6224166517471413,
      "grad_norm": 7.510758399963379,
      "learning_rate": 3.9746972353779845e-07,
      "loss": 0.8297,
      "step": 39980
    },
    {
      "epoch": 0.6224944927491106,
      "grad_norm": 3.3702340126037598,
      "learning_rate": 3.9738778453319347e-07,
      "loss": 0.7475,
      "step": 39985
    },
    {
      "epoch": 0.62257233375108,
      "grad_norm": 2.240464687347412,
      "learning_rate": 3.973058455285885e-07,
      "loss": 0.5728,
      "step": 39990
    },
    {
      "epoch": 0.6226501747530494,
      "grad_norm": 3.49788236618042,
      "learning_rate": 3.972239065239835e-07,
      "loss": 0.8985,
      "step": 39995
    },
    {
      "epoch": 0.6227280157550188,
      "grad_norm": 2.337686777114868,
      "learning_rate": 3.9714196751937854e-07,
      "loss": 0.8044,
      "step": 40000
    },
    {
      "epoch": 0.6228058567569882,
      "grad_norm": 3.218118906021118,
      "learning_rate": 3.970600285147736e-07,
      "loss": 0.7059,
      "step": 40005
    },
    {
      "epoch": 0.6228836977589576,
      "grad_norm": 4.477120876312256,
      "learning_rate": 3.969780895101686e-07,
      "loss": 0.7889,
      "step": 40010
    },
    {
      "epoch": 0.622961538760927,
      "grad_norm": 3.3402247428894043,
      "learning_rate": 3.9689615050556366e-07,
      "loss": 0.8595,
      "step": 40015
    },
    {
      "epoch": 0.6230393797628964,
      "grad_norm": 3.807766914367676,
      "learning_rate": 3.968142115009587e-07,
      "loss": 0.8956,
      "step": 40020
    },
    {
      "epoch": 0.6231172207648656,
      "grad_norm": 5.288766384124756,
      "learning_rate": 3.967322724963537e-07,
      "loss": 0.7516,
      "step": 40025
    },
    {
      "epoch": 0.623195061766835,
      "grad_norm": 6.913799285888672,
      "learning_rate": 3.966503334917487e-07,
      "loss": 0.7289,
      "step": 40030
    },
    {
      "epoch": 0.6232729027688044,
      "grad_norm": 3.516345500946045,
      "learning_rate": 3.9656839448714374e-07,
      "loss": 0.6797,
      "step": 40035
    },
    {
      "epoch": 0.6233507437707738,
      "grad_norm": 3.3447091579437256,
      "learning_rate": 3.9648645548253877e-07,
      "loss": 0.7927,
      "step": 40040
    },
    {
      "epoch": 0.6234285847727432,
      "grad_norm": 5.997098445892334,
      "learning_rate": 3.9640451647793384e-07,
      "loss": 0.8417,
      "step": 40045
    },
    {
      "epoch": 0.6235064257747126,
      "grad_norm": 9.106229782104492,
      "learning_rate": 3.963225774733288e-07,
      "loss": 0.7381,
      "step": 40050
    },
    {
      "epoch": 0.623584266776682,
      "grad_norm": 4.790067195892334,
      "learning_rate": 3.9624063846872383e-07,
      "loss": 0.7961,
      "step": 40055
    },
    {
      "epoch": 0.6236621077786513,
      "grad_norm": 5.3225789070129395,
      "learning_rate": 3.961586994641189e-07,
      "loss": 0.6536,
      "step": 40060
    },
    {
      "epoch": 0.6237399487806207,
      "grad_norm": 16.287736892700195,
      "learning_rate": 3.9607676045951393e-07,
      "loss": 0.6869,
      "step": 40065
    },
    {
      "epoch": 0.6238177897825901,
      "grad_norm": 3.4869582653045654,
      "learning_rate": 3.959948214549089e-07,
      "loss": 0.6698,
      "step": 40070
    },
    {
      "epoch": 0.6238956307845595,
      "grad_norm": 4.292524337768555,
      "learning_rate": 3.9591288245030397e-07,
      "loss": 0.7385,
      "step": 40075
    },
    {
      "epoch": 0.6239734717865288,
      "grad_norm": 3.951045513153076,
      "learning_rate": 3.95830943445699e-07,
      "loss": 0.8316,
      "step": 40080
    },
    {
      "epoch": 0.6240513127884982,
      "grad_norm": 2.5484578609466553,
      "learning_rate": 3.9574900444109407e-07,
      "loss": 0.5927,
      "step": 40085
    },
    {
      "epoch": 0.6241291537904676,
      "grad_norm": 3.3833181858062744,
      "learning_rate": 3.9566706543648904e-07,
      "loss": 0.8005,
      "step": 40090
    },
    {
      "epoch": 0.624206994792437,
      "grad_norm": 3.468243360519409,
      "learning_rate": 3.9558512643188406e-07,
      "loss": 0.8605,
      "step": 40095
    },
    {
      "epoch": 0.6242848357944063,
      "grad_norm": 6.380690097808838,
      "learning_rate": 3.9550318742727913e-07,
      "loss": 0.9081,
      "step": 40100
    },
    {
      "epoch": 0.6243626767963757,
      "grad_norm": 9.10147476196289,
      "learning_rate": 3.9542124842267416e-07,
      "loss": 0.7163,
      "step": 40105
    },
    {
      "epoch": 0.6244405177983451,
      "grad_norm": 3.842978000640869,
      "learning_rate": 3.953393094180691e-07,
      "loss": 0.6746,
      "step": 40110
    },
    {
      "epoch": 0.6245183588003145,
      "grad_norm": 5.561923027038574,
      "learning_rate": 3.952573704134642e-07,
      "loss": 0.6905,
      "step": 40115
    },
    {
      "epoch": 0.6245961998022839,
      "grad_norm": 3.767864227294922,
      "learning_rate": 3.951754314088592e-07,
      "loss": 0.6149,
      "step": 40120
    },
    {
      "epoch": 0.6246740408042533,
      "grad_norm": 2.9838409423828125,
      "learning_rate": 3.9509349240425424e-07,
      "loss": 0.7224,
      "step": 40125
    },
    {
      "epoch": 0.6247518818062227,
      "grad_norm": 3.225837230682373,
      "learning_rate": 3.9501155339964927e-07,
      "loss": 0.8866,
      "step": 40130
    },
    {
      "epoch": 0.6248297228081919,
      "grad_norm": 3.562525749206543,
      "learning_rate": 3.949296143950443e-07,
      "loss": 0.7939,
      "step": 40135
    },
    {
      "epoch": 0.6249075638101613,
      "grad_norm": 4.6887407302856445,
      "learning_rate": 3.9484767539043936e-07,
      "loss": 0.6721,
      "step": 40140
    },
    {
      "epoch": 0.6249854048121307,
      "grad_norm": 3.2326924800872803,
      "learning_rate": 3.947657363858344e-07,
      "loss": 0.6244,
      "step": 40145
    },
    {
      "epoch": 0.6250632458141001,
      "grad_norm": 3.0695505142211914,
      "learning_rate": 3.946837973812294e-07,
      "loss": 0.6486,
      "step": 40150
    },
    {
      "epoch": 0.6251410868160695,
      "grad_norm": 2.9343271255493164,
      "learning_rate": 3.9460185837662443e-07,
      "loss": 0.7571,
      "step": 40155
    },
    {
      "epoch": 0.6252189278180389,
      "grad_norm": 3.6993069648742676,
      "learning_rate": 3.9451991937201945e-07,
      "loss": 0.8822,
      "step": 40160
    },
    {
      "epoch": 0.6252967688200083,
      "grad_norm": 3.3059322834014893,
      "learning_rate": 3.9443798036741447e-07,
      "loss": 0.7544,
      "step": 40165
    },
    {
      "epoch": 0.6253746098219777,
      "grad_norm": 3.429699182510376,
      "learning_rate": 3.9435604136280955e-07,
      "loss": 0.7519,
      "step": 40170
    },
    {
      "epoch": 0.625452450823947,
      "grad_norm": 2.776123046875,
      "learning_rate": 3.942741023582045e-07,
      "loss": 0.7933,
      "step": 40175
    },
    {
      "epoch": 0.6255302918259164,
      "grad_norm": 3.5443432331085205,
      "learning_rate": 3.9419216335359954e-07,
      "loss": 0.697,
      "step": 40180
    },
    {
      "epoch": 0.6256081328278857,
      "grad_norm": 3.1650633811950684,
      "learning_rate": 3.941102243489946e-07,
      "loss": 0.8073,
      "step": 40185
    },
    {
      "epoch": 0.6256859738298551,
      "grad_norm": 4.997108459472656,
      "learning_rate": 3.9402828534438963e-07,
      "loss": 0.791,
      "step": 40190
    },
    {
      "epoch": 0.6257638148318245,
      "grad_norm": 3.0306813716888428,
      "learning_rate": 3.939463463397846e-07,
      "loss": 0.6717,
      "step": 40195
    },
    {
      "epoch": 0.6258416558337939,
      "grad_norm": 3.518197774887085,
      "learning_rate": 3.938644073351797e-07,
      "loss": 0.8226,
      "step": 40200
    },
    {
      "epoch": 0.6259194968357633,
      "grad_norm": 5.217704772949219,
      "learning_rate": 3.937824683305747e-07,
      "loss": 0.7409,
      "step": 40205
    },
    {
      "epoch": 0.6259973378377326,
      "grad_norm": 4.029904842376709,
      "learning_rate": 3.937005293259698e-07,
      "loss": 0.6951,
      "step": 40210
    },
    {
      "epoch": 0.626075178839702,
      "grad_norm": 7.991400718688965,
      "learning_rate": 3.9361859032136474e-07,
      "loss": 0.6417,
      "step": 40215
    },
    {
      "epoch": 0.6261530198416714,
      "grad_norm": 5.952320575714111,
      "learning_rate": 3.9353665131675977e-07,
      "loss": 0.6169,
      "step": 40220
    },
    {
      "epoch": 0.6262308608436408,
      "grad_norm": 3.669065475463867,
      "learning_rate": 3.9345471231215484e-07,
      "loss": 0.8038,
      "step": 40225
    },
    {
      "epoch": 0.6263087018456102,
      "grad_norm": 3.1107237339019775,
      "learning_rate": 3.9337277330754986e-07,
      "loss": 0.695,
      "step": 40230
    },
    {
      "epoch": 0.6263865428475796,
      "grad_norm": 2.8692715167999268,
      "learning_rate": 3.9329083430294483e-07,
      "loss": 0.7712,
      "step": 40235
    },
    {
      "epoch": 0.626464383849549,
      "grad_norm": 3.6877214908599854,
      "learning_rate": 3.932088952983399e-07,
      "loss": 0.8252,
      "step": 40240
    },
    {
      "epoch": 0.6265422248515183,
      "grad_norm": 3.1394309997558594,
      "learning_rate": 3.9312695629373493e-07,
      "loss": 0.8148,
      "step": 40245
    },
    {
      "epoch": 0.6266200658534876,
      "grad_norm": 4.392218589782715,
      "learning_rate": 3.9304501728912995e-07,
      "loss": 0.7376,
      "step": 40250
    },
    {
      "epoch": 0.626697906855457,
      "grad_norm": 4.456888675689697,
      "learning_rate": 3.9296307828452497e-07,
      "loss": 0.7267,
      "step": 40255
    },
    {
      "epoch": 0.6267757478574264,
      "grad_norm": 4.857935428619385,
      "learning_rate": 3.9288113927992e-07,
      "loss": 0.8109,
      "step": 40260
    },
    {
      "epoch": 0.6268535888593958,
      "grad_norm": 4.6022629737854,
      "learning_rate": 3.9279920027531507e-07,
      "loss": 0.8392,
      "step": 40265
    },
    {
      "epoch": 0.6269314298613652,
      "grad_norm": 3.0200412273406982,
      "learning_rate": 3.927172612707101e-07,
      "loss": 0.6964,
      "step": 40270
    },
    {
      "epoch": 0.6270092708633346,
      "grad_norm": 2.824856996536255,
      "learning_rate": 3.9263532226610506e-07,
      "loss": 0.7522,
      "step": 40275
    },
    {
      "epoch": 0.627087111865304,
      "grad_norm": 4.095926761627197,
      "learning_rate": 3.9255338326150014e-07,
      "loss": 0.7757,
      "step": 40280
    },
    {
      "epoch": 0.6271649528672734,
      "grad_norm": 8.60374927520752,
      "learning_rate": 3.9247144425689516e-07,
      "loss": 0.6772,
      "step": 40285
    },
    {
      "epoch": 0.6272427938692426,
      "grad_norm": 6.929055213928223,
      "learning_rate": 3.923895052522902e-07,
      "loss": 0.8012,
      "step": 40290
    },
    {
      "epoch": 0.627320634871212,
      "grad_norm": 2.9979069232940674,
      "learning_rate": 3.923075662476852e-07,
      "loss": 0.7987,
      "step": 40295
    },
    {
      "epoch": 0.6273984758731814,
      "grad_norm": 3.8696656227111816,
      "learning_rate": 3.922256272430802e-07,
      "loss": 0.6649,
      "step": 40300
    },
    {
      "epoch": 0.6274763168751508,
      "grad_norm": 2.688227415084839,
      "learning_rate": 3.9214368823847524e-07,
      "loss": 0.6782,
      "step": 40305
    },
    {
      "epoch": 0.6275541578771202,
      "grad_norm": 5.204583644866943,
      "learning_rate": 3.920617492338703e-07,
      "loss": 0.8622,
      "step": 40310
    },
    {
      "epoch": 0.6276319988790896,
      "grad_norm": 5.594030857086182,
      "learning_rate": 3.919798102292653e-07,
      "loss": 0.7864,
      "step": 40315
    },
    {
      "epoch": 0.627709839881059,
      "grad_norm": 3.16546893119812,
      "learning_rate": 3.918978712246603e-07,
      "loss": 0.7566,
      "step": 40320
    },
    {
      "epoch": 0.6277876808830283,
      "grad_norm": 6.871097087860107,
      "learning_rate": 3.918159322200554e-07,
      "loss": 0.6511,
      "step": 40325
    },
    {
      "epoch": 0.6278655218849977,
      "grad_norm": 6.030157566070557,
      "learning_rate": 3.917339932154504e-07,
      "loss": 0.7636,
      "step": 40330
    },
    {
      "epoch": 0.6279433628869671,
      "grad_norm": 4.084057331085205,
      "learning_rate": 3.9165205421084543e-07,
      "loss": 0.6766,
      "step": 40335
    },
    {
      "epoch": 0.6280212038889365,
      "grad_norm": 4.945893287658691,
      "learning_rate": 3.9157011520624045e-07,
      "loss": 0.6959,
      "step": 40340
    },
    {
      "epoch": 0.6280990448909058,
      "grad_norm": 5.323774814605713,
      "learning_rate": 3.9148817620163547e-07,
      "loss": 0.7112,
      "step": 40345
    },
    {
      "epoch": 0.6281768858928752,
      "grad_norm": 3.700428009033203,
      "learning_rate": 3.9140623719703055e-07,
      "loss": 0.6702,
      "step": 40350
    },
    {
      "epoch": 0.6282547268948446,
      "grad_norm": 3.2915420532226562,
      "learning_rate": 3.913242981924255e-07,
      "loss": 0.8218,
      "step": 40355
    },
    {
      "epoch": 0.628332567896814,
      "grad_norm": 7.17383337020874,
      "learning_rate": 3.9124235918782054e-07,
      "loss": 0.7329,
      "step": 40360
    },
    {
      "epoch": 0.6284104088987833,
      "grad_norm": 3.618509531021118,
      "learning_rate": 3.911604201832156e-07,
      "loss": 0.627,
      "step": 40365
    },
    {
      "epoch": 0.6284882499007527,
      "grad_norm": 3.7692081928253174,
      "learning_rate": 3.9107848117861064e-07,
      "loss": 0.7213,
      "step": 40370
    },
    {
      "epoch": 0.6285660909027221,
      "grad_norm": 4.005009651184082,
      "learning_rate": 3.909965421740056e-07,
      "loss": 0.8247,
      "step": 40375
    },
    {
      "epoch": 0.6286439319046915,
      "grad_norm": 5.181269645690918,
      "learning_rate": 3.909146031694007e-07,
      "loss": 0.8019,
      "step": 40380
    },
    {
      "epoch": 0.6287217729066609,
      "grad_norm": 3.158268451690674,
      "learning_rate": 3.908326641647957e-07,
      "loss": 0.6537,
      "step": 40385
    },
    {
      "epoch": 0.6287996139086303,
      "grad_norm": 4.588606834411621,
      "learning_rate": 3.907507251601908e-07,
      "loss": 0.7534,
      "step": 40390
    },
    {
      "epoch": 0.6288774549105997,
      "grad_norm": 6.237179279327393,
      "learning_rate": 3.9066878615558575e-07,
      "loss": 0.6852,
      "step": 40395
    },
    {
      "epoch": 0.6289552959125689,
      "grad_norm": 5.862989902496338,
      "learning_rate": 3.9058684715098077e-07,
      "loss": 0.6888,
      "step": 40400
    },
    {
      "epoch": 0.6290331369145383,
      "grad_norm": 3.4091105461120605,
      "learning_rate": 3.9050490814637584e-07,
      "loss": 0.7983,
      "step": 40405
    },
    {
      "epoch": 0.6291109779165077,
      "grad_norm": 3.335134267807007,
      "learning_rate": 3.9042296914177086e-07,
      "loss": 0.7171,
      "step": 40410
    },
    {
      "epoch": 0.6291888189184771,
      "grad_norm": 4.2927632331848145,
      "learning_rate": 3.9034103013716583e-07,
      "loss": 0.6753,
      "step": 40415
    },
    {
      "epoch": 0.6292666599204465,
      "grad_norm": 4.237804412841797,
      "learning_rate": 3.902590911325609e-07,
      "loss": 0.7828,
      "step": 40420
    },
    {
      "epoch": 0.6293445009224159,
      "grad_norm": 5.191350936889648,
      "learning_rate": 3.9017715212795593e-07,
      "loss": 0.6736,
      "step": 40425
    },
    {
      "epoch": 0.6294223419243853,
      "grad_norm": 3.641042470932007,
      "learning_rate": 3.9009521312335095e-07,
      "loss": 0.7221,
      "step": 40430
    },
    {
      "epoch": 0.6295001829263547,
      "grad_norm": 3.935760021209717,
      "learning_rate": 3.9001327411874597e-07,
      "loss": 0.7936,
      "step": 40435
    },
    {
      "epoch": 0.629578023928324,
      "grad_norm": 4.079832553863525,
      "learning_rate": 3.89931335114141e-07,
      "loss": 0.8043,
      "step": 40440
    },
    {
      "epoch": 0.6296558649302934,
      "grad_norm": 3.167574882507324,
      "learning_rate": 3.8984939610953607e-07,
      "loss": 0.7413,
      "step": 40445
    },
    {
      "epoch": 0.6297337059322627,
      "grad_norm": 3.470376491546631,
      "learning_rate": 3.897674571049311e-07,
      "loss": 0.843,
      "step": 40450
    },
    {
      "epoch": 0.6298115469342321,
      "grad_norm": 2.76739239692688,
      "learning_rate": 3.896855181003261e-07,
      "loss": 0.5939,
      "step": 40455
    },
    {
      "epoch": 0.6298893879362015,
      "grad_norm": 3.022260904312134,
      "learning_rate": 3.8960357909572114e-07,
      "loss": 0.7274,
      "step": 40460
    },
    {
      "epoch": 0.6299672289381709,
      "grad_norm": 2.899265766143799,
      "learning_rate": 3.8952164009111616e-07,
      "loss": 0.76,
      "step": 40465
    },
    {
      "epoch": 0.6300450699401403,
      "grad_norm": 6.206949234008789,
      "learning_rate": 3.894397010865112e-07,
      "loss": 0.8558,
      "step": 40470
    },
    {
      "epoch": 0.6301229109421096,
      "grad_norm": 7.897107124328613,
      "learning_rate": 3.8935776208190625e-07,
      "loss": 0.7077,
      "step": 40475
    },
    {
      "epoch": 0.630200751944079,
      "grad_norm": 4.208906650543213,
      "learning_rate": 3.892758230773012e-07,
      "loss": 0.7921,
      "step": 40480
    },
    {
      "epoch": 0.6302785929460484,
      "grad_norm": 3.0071067810058594,
      "learning_rate": 3.8919388407269625e-07,
      "loss": 0.751,
      "step": 40485
    },
    {
      "epoch": 0.6303564339480178,
      "grad_norm": 4.027498245239258,
      "learning_rate": 3.891119450680913e-07,
      "loss": 0.7105,
      "step": 40490
    },
    {
      "epoch": 0.6304342749499872,
      "grad_norm": 6.658172607421875,
      "learning_rate": 3.8903000606348634e-07,
      "loss": 0.7643,
      "step": 40495
    },
    {
      "epoch": 0.6305121159519566,
      "grad_norm": 3.7065136432647705,
      "learning_rate": 3.889480670588813e-07,
      "loss": 0.7669,
      "step": 40500
    },
    {
      "epoch": 0.630589956953926,
      "grad_norm": 9.350017547607422,
      "learning_rate": 3.888661280542764e-07,
      "loss": 0.8772,
      "step": 40505
    },
    {
      "epoch": 0.6306677979558953,
      "grad_norm": 2.9904589653015137,
      "learning_rate": 3.887841890496714e-07,
      "loss": 0.7617,
      "step": 40510
    },
    {
      "epoch": 0.6307456389578646,
      "grad_norm": 4.993828773498535,
      "learning_rate": 3.887022500450665e-07,
      "loss": 0.732,
      "step": 40515
    },
    {
      "epoch": 0.630823479959834,
      "grad_norm": 3.3674392700195312,
      "learning_rate": 3.8862031104046145e-07,
      "loss": 0.8483,
      "step": 40520
    },
    {
      "epoch": 0.6309013209618034,
      "grad_norm": 3.686980962753296,
      "learning_rate": 3.885383720358565e-07,
      "loss": 0.7834,
      "step": 40525
    },
    {
      "epoch": 0.6309791619637728,
      "grad_norm": 3.8560752868652344,
      "learning_rate": 3.8845643303125155e-07,
      "loss": 0.7889,
      "step": 40530
    },
    {
      "epoch": 0.6310570029657422,
      "grad_norm": 5.640587329864502,
      "learning_rate": 3.8837449402664657e-07,
      "loss": 0.7848,
      "step": 40535
    },
    {
      "epoch": 0.6311348439677116,
      "grad_norm": 3.313509464263916,
      "learning_rate": 3.8829255502204154e-07,
      "loss": 0.7816,
      "step": 40540
    },
    {
      "epoch": 0.631212684969681,
      "grad_norm": 4.138770580291748,
      "learning_rate": 3.882106160174366e-07,
      "loss": 0.8436,
      "step": 40545
    },
    {
      "epoch": 0.6312905259716504,
      "grad_norm": 3.3513598442077637,
      "learning_rate": 3.8812867701283164e-07,
      "loss": 0.7845,
      "step": 40550
    },
    {
      "epoch": 0.6313683669736196,
      "grad_norm": 3.892230749130249,
      "learning_rate": 3.8804673800822666e-07,
      "loss": 0.7974,
      "step": 40555
    },
    {
      "epoch": 0.631446207975589,
      "grad_norm": 3.050997018814087,
      "learning_rate": 3.879647990036217e-07,
      "loss": 0.7668,
      "step": 40560
    },
    {
      "epoch": 0.6315240489775584,
      "grad_norm": 3.7937769889831543,
      "learning_rate": 3.878828599990167e-07,
      "loss": 0.743,
      "step": 40565
    },
    {
      "epoch": 0.6316018899795278,
      "grad_norm": 3.8931117057800293,
      "learning_rate": 3.878009209944118e-07,
      "loss": 0.6814,
      "step": 40570
    },
    {
      "epoch": 0.6316797309814972,
      "grad_norm": 3.432852268218994,
      "learning_rate": 3.877189819898068e-07,
      "loss": 0.8528,
      "step": 40575
    },
    {
      "epoch": 0.6317575719834666,
      "grad_norm": 3.757092237472534,
      "learning_rate": 3.8763704298520177e-07,
      "loss": 0.7059,
      "step": 40580
    },
    {
      "epoch": 0.631835412985436,
      "grad_norm": 3.3738479614257812,
      "learning_rate": 3.8755510398059684e-07,
      "loss": 0.5819,
      "step": 40585
    },
    {
      "epoch": 0.6319132539874053,
      "grad_norm": 3.5792367458343506,
      "learning_rate": 3.8747316497599186e-07,
      "loss": 0.6964,
      "step": 40590
    },
    {
      "epoch": 0.6319910949893747,
      "grad_norm": 6.4384355545043945,
      "learning_rate": 3.873912259713869e-07,
      "loss": 0.7403,
      "step": 40595
    },
    {
      "epoch": 0.6320689359913441,
      "grad_norm": 5.846506595611572,
      "learning_rate": 3.873092869667819e-07,
      "loss": 0.8256,
      "step": 40600
    },
    {
      "epoch": 0.6321467769933135,
      "grad_norm": 3.9439003467559814,
      "learning_rate": 3.8722734796217693e-07,
      "loss": 0.6557,
      "step": 40605
    },
    {
      "epoch": 0.6322246179952828,
      "grad_norm": 3.5198142528533936,
      "learning_rate": 3.8714540895757195e-07,
      "loss": 0.7078,
      "step": 40610
    },
    {
      "epoch": 0.6323024589972522,
      "grad_norm": 4.58626651763916,
      "learning_rate": 3.8706346995296703e-07,
      "loss": 0.7363,
      "step": 40615
    },
    {
      "epoch": 0.6323802999992216,
      "grad_norm": 2.5278806686401367,
      "learning_rate": 3.86981530948362e-07,
      "loss": 0.7069,
      "step": 40620
    },
    {
      "epoch": 0.632458141001191,
      "grad_norm": 7.54095458984375,
      "learning_rate": 3.86899591943757e-07,
      "loss": 0.6845,
      "step": 40625
    },
    {
      "epoch": 0.6325359820031603,
      "grad_norm": 4.589016914367676,
      "learning_rate": 3.868176529391521e-07,
      "loss": 0.6386,
      "step": 40630
    },
    {
      "epoch": 0.6326138230051297,
      "grad_norm": 6.310573577880859,
      "learning_rate": 3.867357139345471e-07,
      "loss": 0.7521,
      "step": 40635
    },
    {
      "epoch": 0.6326916640070991,
      "grad_norm": 3.7803823947906494,
      "learning_rate": 3.8665377492994214e-07,
      "loss": 0.7377,
      "step": 40640
    },
    {
      "epoch": 0.6327695050090685,
      "grad_norm": 5.0447869300842285,
      "learning_rate": 3.8657183592533716e-07,
      "loss": 0.7122,
      "step": 40645
    },
    {
      "epoch": 0.6328473460110379,
      "grad_norm": 3.7057087421417236,
      "learning_rate": 3.864898969207322e-07,
      "loss": 0.7596,
      "step": 40650
    },
    {
      "epoch": 0.6329251870130073,
      "grad_norm": 3.5439815521240234,
      "learning_rate": 3.8640795791612726e-07,
      "loss": 0.7252,
      "step": 40655
    },
    {
      "epoch": 0.6330030280149767,
      "grad_norm": 4.328409194946289,
      "learning_rate": 3.863260189115222e-07,
      "loss": 0.6509,
      "step": 40660
    },
    {
      "epoch": 0.6330808690169459,
      "grad_norm": 3.001586437225342,
      "learning_rate": 3.8624407990691725e-07,
      "loss": 0.7342,
      "step": 40665
    },
    {
      "epoch": 0.6331587100189153,
      "grad_norm": 2.8196706771850586,
      "learning_rate": 3.861621409023123e-07,
      "loss": 0.7477,
      "step": 40670
    },
    {
      "epoch": 0.6332365510208847,
      "grad_norm": 2.282076358795166,
      "learning_rate": 3.8608020189770734e-07,
      "loss": 0.8071,
      "step": 40675
    },
    {
      "epoch": 0.6333143920228541,
      "grad_norm": 2.6239631175994873,
      "learning_rate": 3.859982628931023e-07,
      "loss": 0.7183,
      "step": 40680
    },
    {
      "epoch": 0.6333922330248235,
      "grad_norm": 3.9009649753570557,
      "learning_rate": 3.859163238884974e-07,
      "loss": 0.7964,
      "step": 40685
    },
    {
      "epoch": 0.6334700740267929,
      "grad_norm": 4.065356731414795,
      "learning_rate": 3.858343848838924e-07,
      "loss": 0.8236,
      "step": 40690
    },
    {
      "epoch": 0.6335479150287623,
      "grad_norm": 4.366590976715088,
      "learning_rate": 3.857524458792875e-07,
      "loss": 0.6739,
      "step": 40695
    },
    {
      "epoch": 0.6336257560307317,
      "grad_norm": 4.591556549072266,
      "learning_rate": 3.8567050687468245e-07,
      "loss": 0.7698,
      "step": 40700
    },
    {
      "epoch": 0.633703597032701,
      "grad_norm": 3.1845710277557373,
      "learning_rate": 3.855885678700775e-07,
      "loss": 0.8821,
      "step": 40705
    },
    {
      "epoch": 0.6337814380346704,
      "grad_norm": 5.980571269989014,
      "learning_rate": 3.8550662886547255e-07,
      "loss": 0.8006,
      "step": 40710
    },
    {
      "epoch": 0.6338592790366397,
      "grad_norm": 3.735835075378418,
      "learning_rate": 3.8542468986086757e-07,
      "loss": 0.777,
      "step": 40715
    },
    {
      "epoch": 0.6339371200386091,
      "grad_norm": 3.5226235389709473,
      "learning_rate": 3.8534275085626254e-07,
      "loss": 0.6366,
      "step": 40720
    },
    {
      "epoch": 0.6340149610405785,
      "grad_norm": 3.1956546306610107,
      "learning_rate": 3.852608118516576e-07,
      "loss": 0.6348,
      "step": 40725
    },
    {
      "epoch": 0.6340928020425479,
      "grad_norm": 3.7543811798095703,
      "learning_rate": 3.8517887284705264e-07,
      "loss": 0.7981,
      "step": 40730
    },
    {
      "epoch": 0.6341706430445173,
      "grad_norm": 4.9216532707214355,
      "learning_rate": 3.8509693384244766e-07,
      "loss": 0.6529,
      "step": 40735
    },
    {
      "epoch": 0.6342484840464866,
      "grad_norm": 3.8284928798675537,
      "learning_rate": 3.850149948378427e-07,
      "loss": 0.7224,
      "step": 40740
    },
    {
      "epoch": 0.634326325048456,
      "grad_norm": 3.8260271549224854,
      "learning_rate": 3.849330558332377e-07,
      "loss": 0.7891,
      "step": 40745
    },
    {
      "epoch": 0.6344041660504254,
      "grad_norm": 3.1684346199035645,
      "learning_rate": 3.848511168286327e-07,
      "loss": 0.6806,
      "step": 40750
    },
    {
      "epoch": 0.6344820070523948,
      "grad_norm": 3.509564161300659,
      "learning_rate": 3.847691778240278e-07,
      "loss": 0.6681,
      "step": 40755
    },
    {
      "epoch": 0.6345598480543642,
      "grad_norm": 5.915243625640869,
      "learning_rate": 3.8468723881942277e-07,
      "loss": 0.622,
      "step": 40760
    },
    {
      "epoch": 0.6346376890563336,
      "grad_norm": 7.517695426940918,
      "learning_rate": 3.8460529981481784e-07,
      "loss": 0.9134,
      "step": 40765
    },
    {
      "epoch": 0.634715530058303,
      "grad_norm": 4.937188625335693,
      "learning_rate": 3.8452336081021287e-07,
      "loss": 0.7916,
      "step": 40770
    },
    {
      "epoch": 0.6347933710602723,
      "grad_norm": 4.206146717071533,
      "learning_rate": 3.844414218056079e-07,
      "loss": 0.692,
      "step": 40775
    },
    {
      "epoch": 0.6348712120622416,
      "grad_norm": 5.193788528442383,
      "learning_rate": 3.8435948280100296e-07,
      "loss": 0.6778,
      "step": 40780
    },
    {
      "epoch": 0.634949053064211,
      "grad_norm": 5.39143180847168,
      "learning_rate": 3.8427754379639793e-07,
      "loss": 0.7384,
      "step": 40785
    },
    {
      "epoch": 0.6350268940661804,
      "grad_norm": 3.1364071369171143,
      "learning_rate": 3.8419560479179295e-07,
      "loss": 0.6757,
      "step": 40790
    },
    {
      "epoch": 0.6351047350681498,
      "grad_norm": 4.111540794372559,
      "learning_rate": 3.8411366578718803e-07,
      "loss": 0.7512,
      "step": 40795
    },
    {
      "epoch": 0.6351825760701192,
      "grad_norm": 3.3314995765686035,
      "learning_rate": 3.8403172678258305e-07,
      "loss": 0.6687,
      "step": 40800
    },
    {
      "epoch": 0.6352604170720886,
      "grad_norm": 3.5224087238311768,
      "learning_rate": 3.83949787777978e-07,
      "loss": 0.8106,
      "step": 40805
    },
    {
      "epoch": 0.635338258074058,
      "grad_norm": 5.40964937210083,
      "learning_rate": 3.838678487733731e-07,
      "loss": 0.8678,
      "step": 40810
    },
    {
      "epoch": 0.6354160990760273,
      "grad_norm": 5.482173442840576,
      "learning_rate": 3.837859097687681e-07,
      "loss": 0.7475,
      "step": 40815
    },
    {
      "epoch": 0.6354939400779966,
      "grad_norm": 3.804060220718384,
      "learning_rate": 3.837039707641632e-07,
      "loss": 0.8096,
      "step": 40820
    },
    {
      "epoch": 0.635571781079966,
      "grad_norm": 4.039361476898193,
      "learning_rate": 3.8362203175955816e-07,
      "loss": 0.8757,
      "step": 40825
    },
    {
      "epoch": 0.6356496220819354,
      "grad_norm": 2.9846019744873047,
      "learning_rate": 3.835400927549532e-07,
      "loss": 0.6399,
      "step": 40830
    },
    {
      "epoch": 0.6357274630839048,
      "grad_norm": 2.612964153289795,
      "learning_rate": 3.8345815375034826e-07,
      "loss": 0.7426,
      "step": 40835
    },
    {
      "epoch": 0.6358053040858742,
      "grad_norm": 6.423083305358887,
      "learning_rate": 3.833762147457433e-07,
      "loss": 0.7515,
      "step": 40840
    },
    {
      "epoch": 0.6358831450878436,
      "grad_norm": 3.9829013347625732,
      "learning_rate": 3.8329427574113825e-07,
      "loss": 0.8745,
      "step": 40845
    },
    {
      "epoch": 0.635960986089813,
      "grad_norm": 3.694988250732422,
      "learning_rate": 3.832123367365333e-07,
      "loss": 0.6924,
      "step": 40850
    },
    {
      "epoch": 0.6360388270917823,
      "grad_norm": 3.950003147125244,
      "learning_rate": 3.8313039773192834e-07,
      "loss": 0.7075,
      "step": 40855
    },
    {
      "epoch": 0.6361166680937517,
      "grad_norm": 5.668179512023926,
      "learning_rate": 3.8304845872732337e-07,
      "loss": 0.7368,
      "step": 40860
    },
    {
      "epoch": 0.6361945090957211,
      "grad_norm": 3.2212798595428467,
      "learning_rate": 3.829665197227184e-07,
      "loss": 0.8055,
      "step": 40865
    },
    {
      "epoch": 0.6362723500976905,
      "grad_norm": 2.7427401542663574,
      "learning_rate": 3.828845807181134e-07,
      "loss": 0.7053,
      "step": 40870
    },
    {
      "epoch": 0.6363501910996598,
      "grad_norm": 4.0167765617370605,
      "learning_rate": 3.8280264171350843e-07,
      "loss": 0.8324,
      "step": 40875
    },
    {
      "epoch": 0.6364280321016292,
      "grad_norm": 5.455165386199951,
      "learning_rate": 3.827207027089035e-07,
      "loss": 0.6968,
      "step": 40880
    },
    {
      "epoch": 0.6365058731035986,
      "grad_norm": 3.318476676940918,
      "learning_rate": 3.826387637042985e-07,
      "loss": 0.7899,
      "step": 40885
    },
    {
      "epoch": 0.636583714105568,
      "grad_norm": 6.358670234680176,
      "learning_rate": 3.8255682469969355e-07,
      "loss": 0.6713,
      "step": 40890
    },
    {
      "epoch": 0.6366615551075373,
      "grad_norm": 2.8279776573181152,
      "learning_rate": 3.8247488569508857e-07,
      "loss": 0.7264,
      "step": 40895
    },
    {
      "epoch": 0.6367393961095067,
      "grad_norm": 4.945094108581543,
      "learning_rate": 3.823929466904836e-07,
      "loss": 0.751,
      "step": 40900
    },
    {
      "epoch": 0.6368172371114761,
      "grad_norm": 3.8329458236694336,
      "learning_rate": 3.823110076858786e-07,
      "loss": 0.7672,
      "step": 40905
    },
    {
      "epoch": 0.6368950781134455,
      "grad_norm": 3.484246015548706,
      "learning_rate": 3.8222906868127364e-07,
      "loss": 0.7879,
      "step": 40910
    },
    {
      "epoch": 0.6369729191154149,
      "grad_norm": 4.3278326988220215,
      "learning_rate": 3.8214712967666866e-07,
      "loss": 0.7225,
      "step": 40915
    },
    {
      "epoch": 0.6370507601173843,
      "grad_norm": 3.4526877403259277,
      "learning_rate": 3.8206519067206373e-07,
      "loss": 0.7351,
      "step": 40920
    },
    {
      "epoch": 0.6371286011193537,
      "grad_norm": 4.027953147888184,
      "learning_rate": 3.819832516674587e-07,
      "loss": 0.7365,
      "step": 40925
    },
    {
      "epoch": 0.6372064421213229,
      "grad_norm": 4.375374794006348,
      "learning_rate": 3.819013126628537e-07,
      "loss": 0.6889,
      "step": 40930
    },
    {
      "epoch": 0.6372842831232923,
      "grad_norm": 2.2958669662475586,
      "learning_rate": 3.818193736582488e-07,
      "loss": 0.8449,
      "step": 40935
    },
    {
      "epoch": 0.6373621241252617,
      "grad_norm": 2.8625123500823975,
      "learning_rate": 3.817374346536438e-07,
      "loss": 0.7953,
      "step": 40940
    },
    {
      "epoch": 0.6374399651272311,
      "grad_norm": 2.859424114227295,
      "learning_rate": 3.816554956490388e-07,
      "loss": 0.5904,
      "step": 40945
    },
    {
      "epoch": 0.6375178061292005,
      "grad_norm": 3.511207103729248,
      "learning_rate": 3.8157355664443387e-07,
      "loss": 0.7364,
      "step": 40950
    },
    {
      "epoch": 0.6375956471311699,
      "grad_norm": 4.3205037117004395,
      "learning_rate": 3.814916176398289e-07,
      "loss": 0.853,
      "step": 40955
    },
    {
      "epoch": 0.6376734881331393,
      "grad_norm": 5.253109931945801,
      "learning_rate": 3.8140967863522396e-07,
      "loss": 0.8611,
      "step": 40960
    },
    {
      "epoch": 0.6377513291351087,
      "grad_norm": 3.9629220962524414,
      "learning_rate": 3.8132773963061893e-07,
      "loss": 0.6616,
      "step": 40965
    },
    {
      "epoch": 0.637829170137078,
      "grad_norm": 5.628442287445068,
      "learning_rate": 3.8124580062601395e-07,
      "loss": 0.7785,
      "step": 40970
    },
    {
      "epoch": 0.6379070111390474,
      "grad_norm": 5.085522651672363,
      "learning_rate": 3.8116386162140903e-07,
      "loss": 0.7416,
      "step": 40975
    },
    {
      "epoch": 0.6379848521410167,
      "grad_norm": 6.248499393463135,
      "learning_rate": 3.8108192261680405e-07,
      "loss": 0.6665,
      "step": 40980
    },
    {
      "epoch": 0.6380626931429861,
      "grad_norm": 2.8380696773529053,
      "learning_rate": 3.80999983612199e-07,
      "loss": 0.7803,
      "step": 40985
    },
    {
      "epoch": 0.6381405341449555,
      "grad_norm": 7.096277713775635,
      "learning_rate": 3.809180446075941e-07,
      "loss": 0.7391,
      "step": 40990
    },
    {
      "epoch": 0.6382183751469249,
      "grad_norm": 4.337358474731445,
      "learning_rate": 3.808361056029891e-07,
      "loss": 0.6891,
      "step": 40995
    },
    {
      "epoch": 0.6382962161488943,
      "grad_norm": 4.121712684631348,
      "learning_rate": 3.8075416659838414e-07,
      "loss": 0.6397,
      "step": 41000
    },
    {
      "epoch": 0.6383740571508636,
      "grad_norm": 2.667370080947876,
      "learning_rate": 3.8067222759377916e-07,
      "loss": 0.774,
      "step": 41005
    },
    {
      "epoch": 0.638451898152833,
      "grad_norm": 3.145350456237793,
      "learning_rate": 3.805902885891742e-07,
      "loss": 0.7745,
      "step": 41010
    },
    {
      "epoch": 0.6385297391548024,
      "grad_norm": 20.301923751831055,
      "learning_rate": 3.8050834958456926e-07,
      "loss": 0.6526,
      "step": 41015
    },
    {
      "epoch": 0.6386075801567718,
      "grad_norm": 3.7136478424072266,
      "learning_rate": 3.804264105799643e-07,
      "loss": 0.7197,
      "step": 41020
    },
    {
      "epoch": 0.6386854211587412,
      "grad_norm": 3.3507494926452637,
      "learning_rate": 3.8034447157535925e-07,
      "loss": 0.8046,
      "step": 41025
    },
    {
      "epoch": 0.6387632621607106,
      "grad_norm": 3.205886125564575,
      "learning_rate": 3.802625325707543e-07,
      "loss": 0.7732,
      "step": 41030
    },
    {
      "epoch": 0.63884110316268,
      "grad_norm": 2.7827250957489014,
      "learning_rate": 3.8018059356614934e-07,
      "loss": 0.8102,
      "step": 41035
    },
    {
      "epoch": 0.6389189441646493,
      "grad_norm": 2.609689950942993,
      "learning_rate": 3.8009865456154437e-07,
      "loss": 0.7843,
      "step": 41040
    },
    {
      "epoch": 0.6389967851666186,
      "grad_norm": 2.975839853286743,
      "learning_rate": 3.800167155569394e-07,
      "loss": 0.7461,
      "step": 41045
    },
    {
      "epoch": 0.639074626168588,
      "grad_norm": 2.665325403213501,
      "learning_rate": 3.799347765523344e-07,
      "loss": 0.7107,
      "step": 41050
    },
    {
      "epoch": 0.6391524671705574,
      "grad_norm": 4.908530235290527,
      "learning_rate": 3.7985283754772943e-07,
      "loss": 0.7814,
      "step": 41055
    },
    {
      "epoch": 0.6392303081725268,
      "grad_norm": 3.772562026977539,
      "learning_rate": 3.797708985431245e-07,
      "loss": 0.8548,
      "step": 41060
    },
    {
      "epoch": 0.6393081491744962,
      "grad_norm": 2.678597927093506,
      "learning_rate": 3.796889595385195e-07,
      "loss": 0.6945,
      "step": 41065
    },
    {
      "epoch": 0.6393859901764656,
      "grad_norm": 3.4535462856292725,
      "learning_rate": 3.796070205339145e-07,
      "loss": 0.7645,
      "step": 41070
    },
    {
      "epoch": 0.639463831178435,
      "grad_norm": 3.465233087539673,
      "learning_rate": 3.7952508152930957e-07,
      "loss": 0.7246,
      "step": 41075
    },
    {
      "epoch": 0.6395416721804043,
      "grad_norm": 5.990328788757324,
      "learning_rate": 3.794431425247046e-07,
      "loss": 0.7978,
      "step": 41080
    },
    {
      "epoch": 0.6396195131823736,
      "grad_norm": 3.200066089630127,
      "learning_rate": 3.793612035200996e-07,
      "loss": 0.6928,
      "step": 41085
    },
    {
      "epoch": 0.639697354184343,
      "grad_norm": 3.42942214012146,
      "learning_rate": 3.7927926451549464e-07,
      "loss": 0.7283,
      "step": 41090
    },
    {
      "epoch": 0.6397751951863124,
      "grad_norm": 4.317032337188721,
      "learning_rate": 3.7919732551088966e-07,
      "loss": 0.8526,
      "step": 41095
    },
    {
      "epoch": 0.6398530361882818,
      "grad_norm": 3.1574923992156982,
      "learning_rate": 3.7911538650628474e-07,
      "loss": 0.6491,
      "step": 41100
    },
    {
      "epoch": 0.6399308771902512,
      "grad_norm": 3.3248939514160156,
      "learning_rate": 3.7903344750167976e-07,
      "loss": 0.8398,
      "step": 41105
    },
    {
      "epoch": 0.6400087181922206,
      "grad_norm": 3.4290640354156494,
      "learning_rate": 3.7895150849707473e-07,
      "loss": 0.7049,
      "step": 41110
    },
    {
      "epoch": 0.64008655919419,
      "grad_norm": 3.967106580734253,
      "learning_rate": 3.788695694924698e-07,
      "loss": 0.7374,
      "step": 41115
    },
    {
      "epoch": 0.6401644001961593,
      "grad_norm": 2.9278724193573,
      "learning_rate": 3.787876304878648e-07,
      "loss": 0.6212,
      "step": 41120
    },
    {
      "epoch": 0.6402422411981287,
      "grad_norm": 4.077934265136719,
      "learning_rate": 3.7870569148325985e-07,
      "loss": 0.8014,
      "step": 41125
    },
    {
      "epoch": 0.6403200822000981,
      "grad_norm": 3.7736330032348633,
      "learning_rate": 3.7862375247865487e-07,
      "loss": 0.7178,
      "step": 41130
    },
    {
      "epoch": 0.6403979232020675,
      "grad_norm": 4.08732795715332,
      "learning_rate": 3.785418134740499e-07,
      "loss": 0.7347,
      "step": 41135
    },
    {
      "epoch": 0.6404757642040368,
      "grad_norm": 2.5996291637420654,
      "learning_rate": 3.7845987446944496e-07,
      "loss": 0.701,
      "step": 41140
    },
    {
      "epoch": 0.6405536052060062,
      "grad_norm": 3.507585287094116,
      "learning_rate": 3.7837793546484e-07,
      "loss": 0.813,
      "step": 41145
    },
    {
      "epoch": 0.6406314462079756,
      "grad_norm": 4.664650917053223,
      "learning_rate": 3.7829599646023495e-07,
      "loss": 0.7854,
      "step": 41150
    },
    {
      "epoch": 0.6407092872099449,
      "grad_norm": 5.394179821014404,
      "learning_rate": 3.7821405745563003e-07,
      "loss": 0.8412,
      "step": 41155
    },
    {
      "epoch": 0.6407871282119143,
      "grad_norm": 5.198697090148926,
      "learning_rate": 3.7813211845102505e-07,
      "loss": 0.6561,
      "step": 41160
    },
    {
      "epoch": 0.6408649692138837,
      "grad_norm": 3.1176159381866455,
      "learning_rate": 3.780501794464201e-07,
      "loss": 0.7566,
      "step": 41165
    },
    {
      "epoch": 0.6409428102158531,
      "grad_norm": 8.87204360961914,
      "learning_rate": 3.779682404418151e-07,
      "loss": 0.8029,
      "step": 41170
    },
    {
      "epoch": 0.6410206512178225,
      "grad_norm": 9.095691680908203,
      "learning_rate": 3.778863014372101e-07,
      "loss": 0.8171,
      "step": 41175
    },
    {
      "epoch": 0.6410984922197919,
      "grad_norm": 3.1503560543060303,
      "learning_rate": 3.7780436243260514e-07,
      "loss": 0.7838,
      "step": 41180
    },
    {
      "epoch": 0.6411763332217613,
      "grad_norm": 6.014164924621582,
      "learning_rate": 3.777224234280002e-07,
      "loss": 0.7167,
      "step": 41185
    },
    {
      "epoch": 0.6412541742237307,
      "grad_norm": 3.09187650680542,
      "learning_rate": 3.776404844233952e-07,
      "loss": 0.7156,
      "step": 41190
    },
    {
      "epoch": 0.6413320152256999,
      "grad_norm": 4.680224418640137,
      "learning_rate": 3.775585454187902e-07,
      "loss": 0.735,
      "step": 41195
    },
    {
      "epoch": 0.6414098562276693,
      "grad_norm": 6.020374774932861,
      "learning_rate": 3.774766064141853e-07,
      "loss": 0.7429,
      "step": 41200
    },
    {
      "epoch": 0.6414876972296387,
      "grad_norm": 2.7733442783355713,
      "learning_rate": 3.773946674095803e-07,
      "loss": 0.7341,
      "step": 41205
    },
    {
      "epoch": 0.6415655382316081,
      "grad_norm": 5.184880256652832,
      "learning_rate": 3.773127284049753e-07,
      "loss": 0.8862,
      "step": 41210
    },
    {
      "epoch": 0.6416433792335775,
      "grad_norm": 3.5138251781463623,
      "learning_rate": 3.7723078940037035e-07,
      "loss": 0.6502,
      "step": 41215
    },
    {
      "epoch": 0.6417212202355469,
      "grad_norm": 4.8179216384887695,
      "learning_rate": 3.7714885039576537e-07,
      "loss": 0.6162,
      "step": 41220
    },
    {
      "epoch": 0.6417990612375163,
      "grad_norm": 2.8230583667755127,
      "learning_rate": 3.7706691139116044e-07,
      "loss": 0.7652,
      "step": 41225
    },
    {
      "epoch": 0.6418769022394857,
      "grad_norm": 4.103043556213379,
      "learning_rate": 3.769849723865554e-07,
      "loss": 0.729,
      "step": 41230
    },
    {
      "epoch": 0.641954743241455,
      "grad_norm": 4.729548931121826,
      "learning_rate": 3.7690303338195043e-07,
      "loss": 0.8516,
      "step": 41235
    },
    {
      "epoch": 0.6420325842434244,
      "grad_norm": 2.9785947799682617,
      "learning_rate": 3.768210943773455e-07,
      "loss": 0.6564,
      "step": 41240
    },
    {
      "epoch": 0.6421104252453937,
      "grad_norm": 5.850048542022705,
      "learning_rate": 3.7673915537274053e-07,
      "loss": 0.8891,
      "step": 41245
    },
    {
      "epoch": 0.6421882662473631,
      "grad_norm": 2.8443448543548584,
      "learning_rate": 3.766572163681355e-07,
      "loss": 0.7996,
      "step": 41250
    },
    {
      "epoch": 0.6422661072493325,
      "grad_norm": 2.6573002338409424,
      "learning_rate": 3.765752773635306e-07,
      "loss": 0.818,
      "step": 41255
    },
    {
      "epoch": 0.6423439482513019,
      "grad_norm": 5.465129852294922,
      "learning_rate": 3.764933383589256e-07,
      "loss": 0.8232,
      "step": 41260
    },
    {
      "epoch": 0.6424217892532713,
      "grad_norm": 3.68987774848938,
      "learning_rate": 3.7641139935432067e-07,
      "loss": 0.7938,
      "step": 41265
    },
    {
      "epoch": 0.6424996302552406,
      "grad_norm": 3.1878132820129395,
      "learning_rate": 3.7632946034971564e-07,
      "loss": 0.7344,
      "step": 41270
    },
    {
      "epoch": 0.64257747125721,
      "grad_norm": 3.211620569229126,
      "learning_rate": 3.7624752134511066e-07,
      "loss": 0.7418,
      "step": 41275
    },
    {
      "epoch": 0.6426553122591794,
      "grad_norm": 6.801905632019043,
      "learning_rate": 3.7616558234050574e-07,
      "loss": 0.7728,
      "step": 41280
    },
    {
      "epoch": 0.6427331532611488,
      "grad_norm": 3.71543550491333,
      "learning_rate": 3.7608364333590076e-07,
      "loss": 0.8921,
      "step": 41285
    },
    {
      "epoch": 0.6428109942631182,
      "grad_norm": 4.771504878997803,
      "learning_rate": 3.7600170433129573e-07,
      "loss": 0.7281,
      "step": 41290
    },
    {
      "epoch": 0.6428888352650876,
      "grad_norm": 4.5773773193359375,
      "learning_rate": 3.759197653266908e-07,
      "loss": 0.7639,
      "step": 41295
    },
    {
      "epoch": 0.642966676267057,
      "grad_norm": 3.7005128860473633,
      "learning_rate": 3.758378263220858e-07,
      "loss": 0.8759,
      "step": 41300
    },
    {
      "epoch": 0.6430445172690263,
      "grad_norm": 3.6319329738616943,
      "learning_rate": 3.7575588731748085e-07,
      "loss": 0.81,
      "step": 41305
    },
    {
      "epoch": 0.6431223582709956,
      "grad_norm": 3.043229103088379,
      "learning_rate": 3.7567394831287587e-07,
      "loss": 0.8246,
      "step": 41310
    },
    {
      "epoch": 0.643200199272965,
      "grad_norm": 5.534616470336914,
      "learning_rate": 3.755920093082709e-07,
      "loss": 0.7222,
      "step": 41315
    },
    {
      "epoch": 0.6432780402749344,
      "grad_norm": 3.7299044132232666,
      "learning_rate": 3.755100703036659e-07,
      "loss": 0.7275,
      "step": 41320
    },
    {
      "epoch": 0.6433558812769038,
      "grad_norm": 4.139654636383057,
      "learning_rate": 3.75428131299061e-07,
      "loss": 0.7892,
      "step": 41325
    },
    {
      "epoch": 0.6434337222788732,
      "grad_norm": 4.817360877990723,
      "learning_rate": 3.7534619229445596e-07,
      "loss": 0.8544,
      "step": 41330
    },
    {
      "epoch": 0.6435115632808426,
      "grad_norm": 2.358349561691284,
      "learning_rate": 3.7526425328985103e-07,
      "loss": 0.6768,
      "step": 41335
    },
    {
      "epoch": 0.643589404282812,
      "grad_norm": 8.75611400604248,
      "learning_rate": 3.7518231428524605e-07,
      "loss": 0.6918,
      "step": 41340
    },
    {
      "epoch": 0.6436672452847813,
      "grad_norm": 4.5237717628479,
      "learning_rate": 3.751003752806411e-07,
      "loss": 0.7109,
      "step": 41345
    },
    {
      "epoch": 0.6437450862867506,
      "grad_norm": 5.845875263214111,
      "learning_rate": 3.750184362760361e-07,
      "loss": 0.7326,
      "step": 41350
    },
    {
      "epoch": 0.64382292728872,
      "grad_norm": 2.752054214477539,
      "learning_rate": 3.749364972714311e-07,
      "loss": 0.8133,
      "step": 41355
    },
    {
      "epoch": 0.6439007682906894,
      "grad_norm": 4.462615489959717,
      "learning_rate": 3.7485455826682614e-07,
      "loss": 0.7082,
      "step": 41360
    },
    {
      "epoch": 0.6439786092926588,
      "grad_norm": 3.4426181316375732,
      "learning_rate": 3.747726192622212e-07,
      "loss": 0.8882,
      "step": 41365
    },
    {
      "epoch": 0.6440564502946282,
      "grad_norm": 2.756438970565796,
      "learning_rate": 3.746906802576162e-07,
      "loss": 0.6704,
      "step": 41370
    },
    {
      "epoch": 0.6441342912965976,
      "grad_norm": 3.256645679473877,
      "learning_rate": 3.746087412530112e-07,
      "loss": 0.825,
      "step": 41375
    },
    {
      "epoch": 0.644212132298567,
      "grad_norm": 4.9434494972229,
      "learning_rate": 3.745268022484063e-07,
      "loss": 0.6623,
      "step": 41380
    },
    {
      "epoch": 0.6442899733005363,
      "grad_norm": 4.180547714233398,
      "learning_rate": 3.744448632438013e-07,
      "loss": 0.7228,
      "step": 41385
    },
    {
      "epoch": 0.6443678143025057,
      "grad_norm": 3.7599496841430664,
      "learning_rate": 3.7436292423919627e-07,
      "loss": 0.6444,
      "step": 41390
    },
    {
      "epoch": 0.6444456553044751,
      "grad_norm": 6.147278308868408,
      "learning_rate": 3.7428098523459135e-07,
      "loss": 0.8069,
      "step": 41395
    },
    {
      "epoch": 0.6445234963064445,
      "grad_norm": 3.9845097064971924,
      "learning_rate": 3.7419904622998637e-07,
      "loss": 0.6398,
      "step": 41400
    },
    {
      "epoch": 0.6446013373084138,
      "grad_norm": 2.9971117973327637,
      "learning_rate": 3.7411710722538144e-07,
      "loss": 0.6859,
      "step": 41405
    },
    {
      "epoch": 0.6446791783103832,
      "grad_norm": 4.785406589508057,
      "learning_rate": 3.7403516822077647e-07,
      "loss": 0.7475,
      "step": 41410
    },
    {
      "epoch": 0.6447570193123526,
      "grad_norm": 6.1959967613220215,
      "learning_rate": 3.7395322921617143e-07,
      "loss": 0.638,
      "step": 41415
    },
    {
      "epoch": 0.6448348603143219,
      "grad_norm": 3.410834789276123,
      "learning_rate": 3.738712902115665e-07,
      "loss": 0.7414,
      "step": 41420
    },
    {
      "epoch": 0.6449127013162913,
      "grad_norm": 3.7868876457214355,
      "learning_rate": 3.7378935120696153e-07,
      "loss": 0.8007,
      "step": 41425
    },
    {
      "epoch": 0.6449905423182607,
      "grad_norm": 4.072096347808838,
      "learning_rate": 3.7370741220235655e-07,
      "loss": 0.7928,
      "step": 41430
    },
    {
      "epoch": 0.6450683833202301,
      "grad_norm": 4.204719066619873,
      "learning_rate": 3.736254731977516e-07,
      "loss": 0.6942,
      "step": 41435
    },
    {
      "epoch": 0.6451462243221995,
      "grad_norm": 7.717184543609619,
      "learning_rate": 3.735435341931466e-07,
      "loss": 0.7897,
      "step": 41440
    },
    {
      "epoch": 0.6452240653241689,
      "grad_norm": 4.062209606170654,
      "learning_rate": 3.734615951885416e-07,
      "loss": 0.7914,
      "step": 41445
    },
    {
      "epoch": 0.6453019063261383,
      "grad_norm": 3.9234747886657715,
      "learning_rate": 3.733796561839367e-07,
      "loss": 0.7791,
      "step": 41450
    },
    {
      "epoch": 0.6453797473281077,
      "grad_norm": 2.828282117843628,
      "learning_rate": 3.7329771717933166e-07,
      "loss": 0.8083,
      "step": 41455
    },
    {
      "epoch": 0.6454575883300769,
      "grad_norm": 3.420192241668701,
      "learning_rate": 3.7321577817472674e-07,
      "loss": 0.7211,
      "step": 41460
    },
    {
      "epoch": 0.6455354293320463,
      "grad_norm": 2.9371538162231445,
      "learning_rate": 3.7313383917012176e-07,
      "loss": 0.8243,
      "step": 41465
    },
    {
      "epoch": 0.6456132703340157,
      "grad_norm": 2.778998851776123,
      "learning_rate": 3.730519001655168e-07,
      "loss": 0.7216,
      "step": 41470
    },
    {
      "epoch": 0.6456911113359851,
      "grad_norm": 5.02791166305542,
      "learning_rate": 3.729699611609118e-07,
      "loss": 0.8293,
      "step": 41475
    },
    {
      "epoch": 0.6457689523379545,
      "grad_norm": 5.255801677703857,
      "learning_rate": 3.728880221563068e-07,
      "loss": 0.7546,
      "step": 41480
    },
    {
      "epoch": 0.6458467933399239,
      "grad_norm": 3.1657016277313232,
      "learning_rate": 3.7280608315170185e-07,
      "loss": 0.8566,
      "step": 41485
    },
    {
      "epoch": 0.6459246343418933,
      "grad_norm": 3.7271106243133545,
      "learning_rate": 3.727241441470969e-07,
      "loss": 0.7184,
      "step": 41490
    },
    {
      "epoch": 0.6460024753438627,
      "grad_norm": 4.049560546875,
      "learning_rate": 3.726422051424919e-07,
      "loss": 0.7134,
      "step": 41495
    },
    {
      "epoch": 0.646080316345832,
      "grad_norm": 3.0601391792297363,
      "learning_rate": 3.725602661378869e-07,
      "loss": 0.7361,
      "step": 41500
    },
    {
      "epoch": 0.6461581573478014,
      "grad_norm": 3.703916072845459,
      "learning_rate": 3.72478327133282e-07,
      "loss": 0.687,
      "step": 41505
    },
    {
      "epoch": 0.6462359983497707,
      "grad_norm": 3.676582098007202,
      "learning_rate": 3.72396388128677e-07,
      "loss": 0.7388,
      "step": 41510
    },
    {
      "epoch": 0.6463138393517401,
      "grad_norm": 3.463157892227173,
      "learning_rate": 3.7231444912407203e-07,
      "loss": 0.8198,
      "step": 41515
    },
    {
      "epoch": 0.6463916803537095,
      "grad_norm": 3.4636313915252686,
      "learning_rate": 3.7223251011946705e-07,
      "loss": 0.831,
      "step": 41520
    },
    {
      "epoch": 0.6464695213556789,
      "grad_norm": 2.7257275581359863,
      "learning_rate": 3.721505711148621e-07,
      "loss": 0.7211,
      "step": 41525
    },
    {
      "epoch": 0.6465473623576483,
      "grad_norm": 4.406435966491699,
      "learning_rate": 3.7206863211025715e-07,
      "loss": 0.6394,
      "step": 41530
    },
    {
      "epoch": 0.6466252033596176,
      "grad_norm": 8.30848503112793,
      "learning_rate": 3.719866931056521e-07,
      "loss": 0.7649,
      "step": 41535
    },
    {
      "epoch": 0.646703044361587,
      "grad_norm": 3.769920825958252,
      "learning_rate": 3.7190475410104714e-07,
      "loss": 0.8148,
      "step": 41540
    },
    {
      "epoch": 0.6467808853635564,
      "grad_norm": 7.23143196105957,
      "learning_rate": 3.718228150964422e-07,
      "loss": 0.8884,
      "step": 41545
    },
    {
      "epoch": 0.6468587263655258,
      "grad_norm": 9.43487548828125,
      "learning_rate": 3.7174087609183724e-07,
      "loss": 0.7994,
      "step": 41550
    },
    {
      "epoch": 0.6469365673674952,
      "grad_norm": 3.7569494247436523,
      "learning_rate": 3.716589370872322e-07,
      "loss": 0.7463,
      "step": 41555
    },
    {
      "epoch": 0.6470144083694646,
      "grad_norm": 3.0045650005340576,
      "learning_rate": 3.715769980826273e-07,
      "loss": 0.6621,
      "step": 41560
    },
    {
      "epoch": 0.647092249371434,
      "grad_norm": 3.0652050971984863,
      "learning_rate": 3.714950590780223e-07,
      "loss": 0.7819,
      "step": 41565
    },
    {
      "epoch": 0.6471700903734033,
      "grad_norm": 4.574825763702393,
      "learning_rate": 3.714131200734173e-07,
      "loss": 0.6192,
      "step": 41570
    },
    {
      "epoch": 0.6472479313753726,
      "grad_norm": 5.208134651184082,
      "learning_rate": 3.7133118106881235e-07,
      "loss": 0.7024,
      "step": 41575
    },
    {
      "epoch": 0.647325772377342,
      "grad_norm": 4.49965238571167,
      "learning_rate": 3.7124924206420737e-07,
      "loss": 0.6485,
      "step": 41580
    },
    {
      "epoch": 0.6474036133793114,
      "grad_norm": 7.903379440307617,
      "learning_rate": 3.7116730305960244e-07,
      "loss": 0.7819,
      "step": 41585
    },
    {
      "epoch": 0.6474814543812808,
      "grad_norm": 3.4983954429626465,
      "learning_rate": 3.7108536405499747e-07,
      "loss": 0.7338,
      "step": 41590
    },
    {
      "epoch": 0.6475592953832502,
      "grad_norm": 7.198625087738037,
      "learning_rate": 3.7100342505039244e-07,
      "loss": 0.655,
      "step": 41595
    },
    {
      "epoch": 0.6476371363852196,
      "grad_norm": 3.290994644165039,
      "learning_rate": 3.709214860457875e-07,
      "loss": 0.7121,
      "step": 41600
    },
    {
      "epoch": 0.647714977387189,
      "grad_norm": 6.0432939529418945,
      "learning_rate": 3.7083954704118253e-07,
      "loss": 0.7179,
      "step": 41605
    },
    {
      "epoch": 0.6477928183891583,
      "grad_norm": 7.8545098304748535,
      "learning_rate": 3.7075760803657755e-07,
      "loss": 0.6716,
      "step": 41610
    },
    {
      "epoch": 0.6478706593911276,
      "grad_norm": 2.9196503162384033,
      "learning_rate": 3.706756690319726e-07,
      "loss": 0.6076,
      "step": 41615
    },
    {
      "epoch": 0.647948500393097,
      "grad_norm": 3.5943706035614014,
      "learning_rate": 3.705937300273676e-07,
      "loss": 0.8717,
      "step": 41620
    },
    {
      "epoch": 0.6480263413950664,
      "grad_norm": 9.273316383361816,
      "learning_rate": 3.705117910227626e-07,
      "loss": 0.7407,
      "step": 41625
    },
    {
      "epoch": 0.6481041823970358,
      "grad_norm": 4.871614933013916,
      "learning_rate": 3.704298520181577e-07,
      "loss": 0.6444,
      "step": 41630
    },
    {
      "epoch": 0.6481820233990052,
      "grad_norm": 3.648860454559326,
      "learning_rate": 3.7034791301355266e-07,
      "loss": 0.6217,
      "step": 41635
    },
    {
      "epoch": 0.6482598644009746,
      "grad_norm": 3.1219499111175537,
      "learning_rate": 3.7026597400894774e-07,
      "loss": 0.7049,
      "step": 41640
    },
    {
      "epoch": 0.648337705402944,
      "grad_norm": 3.1243622303009033,
      "learning_rate": 3.7018403500434276e-07,
      "loss": 0.6518,
      "step": 41645
    },
    {
      "epoch": 0.6484155464049133,
      "grad_norm": 3.3632712364196777,
      "learning_rate": 3.701020959997378e-07,
      "loss": 0.9175,
      "step": 41650
    },
    {
      "epoch": 0.6484933874068827,
      "grad_norm": 4.511447429656982,
      "learning_rate": 3.700201569951328e-07,
      "loss": 0.6795,
      "step": 41655
    },
    {
      "epoch": 0.6485712284088521,
      "grad_norm": 4.381533145904541,
      "learning_rate": 3.699382179905278e-07,
      "loss": 0.7924,
      "step": 41660
    },
    {
      "epoch": 0.6486490694108215,
      "grad_norm": 2.662503242492676,
      "learning_rate": 3.6985627898592285e-07,
      "loss": 0.7492,
      "step": 41665
    },
    {
      "epoch": 0.6487269104127908,
      "grad_norm": 5.151483535766602,
      "learning_rate": 3.697743399813179e-07,
      "loss": 0.6498,
      "step": 41670
    },
    {
      "epoch": 0.6488047514147602,
      "grad_norm": 4.378106117248535,
      "learning_rate": 3.696924009767129e-07,
      "loss": 0.7353,
      "step": 41675
    },
    {
      "epoch": 0.6488825924167296,
      "grad_norm": 7.105069160461426,
      "learning_rate": 3.696104619721079e-07,
      "loss": 0.731,
      "step": 41680
    },
    {
      "epoch": 0.6489604334186989,
      "grad_norm": 2.963292360305786,
      "learning_rate": 3.69528522967503e-07,
      "loss": 0.7669,
      "step": 41685
    },
    {
      "epoch": 0.6490382744206683,
      "grad_norm": 2.9259302616119385,
      "learning_rate": 3.69446583962898e-07,
      "loss": 0.732,
      "step": 41690
    },
    {
      "epoch": 0.6491161154226377,
      "grad_norm": 3.850205659866333,
      "learning_rate": 3.69364644958293e-07,
      "loss": 0.7504,
      "step": 41695
    },
    {
      "epoch": 0.6491939564246071,
      "grad_norm": 4.258008003234863,
      "learning_rate": 3.6928270595368805e-07,
      "loss": 0.7247,
      "step": 41700
    },
    {
      "epoch": 0.6492717974265765,
      "grad_norm": 4.296056270599365,
      "learning_rate": 3.692007669490831e-07,
      "loss": 0.7835,
      "step": 41705
    },
    {
      "epoch": 0.6493496384285459,
      "grad_norm": 3.3367154598236084,
      "learning_rate": 3.6911882794447815e-07,
      "loss": 0.7313,
      "step": 41710
    },
    {
      "epoch": 0.6494274794305153,
      "grad_norm": 3.4741368293762207,
      "learning_rate": 3.690368889398731e-07,
      "loss": 0.7694,
      "step": 41715
    },
    {
      "epoch": 0.6495053204324847,
      "grad_norm": 4.587907314300537,
      "learning_rate": 3.6895494993526814e-07,
      "loss": 0.6975,
      "step": 41720
    },
    {
      "epoch": 0.6495831614344539,
      "grad_norm": 4.50137996673584,
      "learning_rate": 3.688730109306632e-07,
      "loss": 0.7359,
      "step": 41725
    },
    {
      "epoch": 0.6496610024364233,
      "grad_norm": 3.25174617767334,
      "learning_rate": 3.6879107192605824e-07,
      "loss": 0.7159,
      "step": 41730
    },
    {
      "epoch": 0.6497388434383927,
      "grad_norm": 4.774181842803955,
      "learning_rate": 3.6870913292145326e-07,
      "loss": 0.7679,
      "step": 41735
    },
    {
      "epoch": 0.6498166844403621,
      "grad_norm": 3.1152217388153076,
      "learning_rate": 3.686271939168483e-07,
      "loss": 0.8033,
      "step": 41740
    },
    {
      "epoch": 0.6498945254423315,
      "grad_norm": 8.285780906677246,
      "learning_rate": 3.685452549122433e-07,
      "loss": 0.702,
      "step": 41745
    },
    {
      "epoch": 0.6499723664443009,
      "grad_norm": 3.4480533599853516,
      "learning_rate": 3.6846331590763833e-07,
      "loss": 0.6887,
      "step": 41750
    },
    {
      "epoch": 0.6500502074462703,
      "grad_norm": 3.216451406478882,
      "learning_rate": 3.683813769030334e-07,
      "loss": 0.7539,
      "step": 41755
    },
    {
      "epoch": 0.6501280484482396,
      "grad_norm": 6.662845611572266,
      "learning_rate": 3.6829943789842837e-07,
      "loss": 0.7491,
      "step": 41760
    },
    {
      "epoch": 0.650205889450209,
      "grad_norm": 4.225223064422607,
      "learning_rate": 3.6821749889382345e-07,
      "loss": 0.7048,
      "step": 41765
    },
    {
      "epoch": 0.6502837304521784,
      "grad_norm": 4.446674823760986,
      "learning_rate": 3.6813555988921847e-07,
      "loss": 0.8285,
      "step": 41770
    },
    {
      "epoch": 0.6503615714541477,
      "grad_norm": 3.2333836555480957,
      "learning_rate": 3.680536208846135e-07,
      "loss": 0.6692,
      "step": 41775
    },
    {
      "epoch": 0.6504394124561171,
      "grad_norm": 3.695831537246704,
      "learning_rate": 3.679716818800085e-07,
      "loss": 0.7122,
      "step": 41780
    },
    {
      "epoch": 0.6505172534580865,
      "grad_norm": 4.92869234085083,
      "learning_rate": 3.6788974287540353e-07,
      "loss": 0.7876,
      "step": 41785
    },
    {
      "epoch": 0.6505950944600559,
      "grad_norm": 2.8876078128814697,
      "learning_rate": 3.6780780387079855e-07,
      "loss": 0.7078,
      "step": 41790
    },
    {
      "epoch": 0.6506729354620253,
      "grad_norm": 2.819228172302246,
      "learning_rate": 3.6772586486619363e-07,
      "loss": 0.9205,
      "step": 41795
    },
    {
      "epoch": 0.6507507764639946,
      "grad_norm": 3.2575864791870117,
      "learning_rate": 3.676439258615886e-07,
      "loss": 0.7838,
      "step": 41800
    },
    {
      "epoch": 0.650828617465964,
      "grad_norm": 8.494056701660156,
      "learning_rate": 3.675619868569836e-07,
      "loss": 0.7015,
      "step": 41805
    },
    {
      "epoch": 0.6509064584679334,
      "grad_norm": 2.9342498779296875,
      "learning_rate": 3.674800478523787e-07,
      "loss": 0.6235,
      "step": 41810
    },
    {
      "epoch": 0.6509842994699028,
      "grad_norm": 3.7202892303466797,
      "learning_rate": 3.673981088477737e-07,
      "loss": 0.7026,
      "step": 41815
    },
    {
      "epoch": 0.6510621404718722,
      "grad_norm": 4.095865249633789,
      "learning_rate": 3.673161698431687e-07,
      "loss": 0.6946,
      "step": 41820
    },
    {
      "epoch": 0.6511399814738416,
      "grad_norm": 3.789100170135498,
      "learning_rate": 3.6723423083856376e-07,
      "loss": 0.7497,
      "step": 41825
    },
    {
      "epoch": 0.651217822475811,
      "grad_norm": 6.1070404052734375,
      "learning_rate": 3.671522918339588e-07,
      "loss": 0.6864,
      "step": 41830
    },
    {
      "epoch": 0.6512956634777803,
      "grad_norm": 4.715911388397217,
      "learning_rate": 3.6707035282935386e-07,
      "loss": 0.69,
      "step": 41835
    },
    {
      "epoch": 0.6513735044797496,
      "grad_norm": 3.7059707641601562,
      "learning_rate": 3.6698841382474883e-07,
      "loss": 0.6751,
      "step": 41840
    },
    {
      "epoch": 0.651451345481719,
      "grad_norm": 5.946600437164307,
      "learning_rate": 3.6690647482014385e-07,
      "loss": 0.7607,
      "step": 41845
    },
    {
      "epoch": 0.6515291864836884,
      "grad_norm": 4.869922161102295,
      "learning_rate": 3.668245358155389e-07,
      "loss": 0.7324,
      "step": 41850
    },
    {
      "epoch": 0.6516070274856578,
      "grad_norm": 4.6485371589660645,
      "learning_rate": 3.6674259681093395e-07,
      "loss": 0.6471,
      "step": 41855
    },
    {
      "epoch": 0.6516848684876272,
      "grad_norm": 3.9355266094207764,
      "learning_rate": 3.666606578063289e-07,
      "loss": 0.695,
      "step": 41860
    },
    {
      "epoch": 0.6517627094895966,
      "grad_norm": 5.446768760681152,
      "learning_rate": 3.66578718801724e-07,
      "loss": 0.6813,
      "step": 41865
    },
    {
      "epoch": 0.651840550491566,
      "grad_norm": 3.152959108352661,
      "learning_rate": 3.66496779797119e-07,
      "loss": 0.8176,
      "step": 41870
    },
    {
      "epoch": 0.6519183914935353,
      "grad_norm": 4.807401657104492,
      "learning_rate": 3.6641484079251403e-07,
      "loss": 0.8442,
      "step": 41875
    },
    {
      "epoch": 0.6519962324955046,
      "grad_norm": 3.055206537246704,
      "learning_rate": 3.6633290178790906e-07,
      "loss": 0.7896,
      "step": 41880
    },
    {
      "epoch": 0.652074073497474,
      "grad_norm": 4.317441463470459,
      "learning_rate": 3.662509627833041e-07,
      "loss": 0.8539,
      "step": 41885
    },
    {
      "epoch": 0.6521519144994434,
      "grad_norm": 3.622631311416626,
      "learning_rate": 3.6616902377869915e-07,
      "loss": 0.7422,
      "step": 41890
    },
    {
      "epoch": 0.6522297555014128,
      "grad_norm": 3.086956024169922,
      "learning_rate": 3.660870847740942e-07,
      "loss": 0.7291,
      "step": 41895
    },
    {
      "epoch": 0.6523075965033822,
      "grad_norm": 3.2881956100463867,
      "learning_rate": 3.6600514576948914e-07,
      "loss": 0.7022,
      "step": 41900
    },
    {
      "epoch": 0.6523854375053516,
      "grad_norm": 3.072791337966919,
      "learning_rate": 3.659232067648842e-07,
      "loss": 0.789,
      "step": 41905
    },
    {
      "epoch": 0.652463278507321,
      "grad_norm": 3.8537228107452393,
      "learning_rate": 3.6584126776027924e-07,
      "loss": 0.7397,
      "step": 41910
    },
    {
      "epoch": 0.6525411195092903,
      "grad_norm": 4.116491794586182,
      "learning_rate": 3.6575932875567426e-07,
      "loss": 0.7978,
      "step": 41915
    },
    {
      "epoch": 0.6526189605112597,
      "grad_norm": 6.148048400878906,
      "learning_rate": 3.656773897510693e-07,
      "loss": 0.8242,
      "step": 41920
    },
    {
      "epoch": 0.6526968015132291,
      "grad_norm": 2.608595848083496,
      "learning_rate": 3.655954507464643e-07,
      "loss": 0.6957,
      "step": 41925
    },
    {
      "epoch": 0.6527746425151985,
      "grad_norm": 2.877262830734253,
      "learning_rate": 3.6551351174185933e-07,
      "loss": 0.7416,
      "step": 41930
    },
    {
      "epoch": 0.6528524835171678,
      "grad_norm": 3.9800829887390137,
      "learning_rate": 3.654315727372544e-07,
      "loss": 0.7843,
      "step": 41935
    },
    {
      "epoch": 0.6529303245191372,
      "grad_norm": 3.1504392623901367,
      "learning_rate": 3.6534963373264937e-07,
      "loss": 0.7037,
      "step": 41940
    },
    {
      "epoch": 0.6530081655211066,
      "grad_norm": 3.605436325073242,
      "learning_rate": 3.652676947280444e-07,
      "loss": 0.6171,
      "step": 41945
    },
    {
      "epoch": 0.6530860065230759,
      "grad_norm": 5.8045430183410645,
      "learning_rate": 3.6518575572343947e-07,
      "loss": 0.7362,
      "step": 41950
    },
    {
      "epoch": 0.6531638475250453,
      "grad_norm": 3.90343976020813,
      "learning_rate": 3.651038167188345e-07,
      "loss": 0.7088,
      "step": 41955
    },
    {
      "epoch": 0.6532416885270147,
      "grad_norm": 5.188382625579834,
      "learning_rate": 3.650218777142295e-07,
      "loss": 0.7038,
      "step": 41960
    },
    {
      "epoch": 0.6533195295289841,
      "grad_norm": 3.0929410457611084,
      "learning_rate": 3.6493993870962453e-07,
      "loss": 0.6536,
      "step": 41965
    },
    {
      "epoch": 0.6533973705309535,
      "grad_norm": 4.426760196685791,
      "learning_rate": 3.6485799970501956e-07,
      "loss": 0.7766,
      "step": 41970
    },
    {
      "epoch": 0.6534752115329229,
      "grad_norm": 2.853370428085327,
      "learning_rate": 3.6477606070041463e-07,
      "loss": 0.7648,
      "step": 41975
    },
    {
      "epoch": 0.6535530525348923,
      "grad_norm": 3.432647943496704,
      "learning_rate": 3.646941216958096e-07,
      "loss": 0.6937,
      "step": 41980
    },
    {
      "epoch": 0.6536308935368617,
      "grad_norm": 3.2468276023864746,
      "learning_rate": 3.646121826912046e-07,
      "loss": 0.6011,
      "step": 41985
    },
    {
      "epoch": 0.6537087345388309,
      "grad_norm": 3.4072225093841553,
      "learning_rate": 3.645302436865997e-07,
      "loss": 0.6954,
      "step": 41990
    },
    {
      "epoch": 0.6537865755408003,
      "grad_norm": 5.404689311981201,
      "learning_rate": 3.644483046819947e-07,
      "loss": 0.7833,
      "step": 41995
    },
    {
      "epoch": 0.6538644165427697,
      "grad_norm": 4.833832740783691,
      "learning_rate": 3.643663656773897e-07,
      "loss": 0.6809,
      "step": 42000
    },
    {
      "epoch": 0.6539422575447391,
      "grad_norm": 4.284120082855225,
      "learning_rate": 3.6428442667278476e-07,
      "loss": 0.6648,
      "step": 42005
    },
    {
      "epoch": 0.6540200985467085,
      "grad_norm": 3.297881841659546,
      "learning_rate": 3.642024876681798e-07,
      "loss": 0.7674,
      "step": 42010
    },
    {
      "epoch": 0.6540979395486779,
      "grad_norm": 3.799231767654419,
      "learning_rate": 3.6412054866357486e-07,
      "loss": 0.6944,
      "step": 42015
    },
    {
      "epoch": 0.6541757805506473,
      "grad_norm": 3.5449180603027344,
      "learning_rate": 3.6403860965896983e-07,
      "loss": 0.6633,
      "step": 42020
    },
    {
      "epoch": 0.6542536215526166,
      "grad_norm": 3.2312800884246826,
      "learning_rate": 3.6395667065436485e-07,
      "loss": 0.8045,
      "step": 42025
    },
    {
      "epoch": 0.654331462554586,
      "grad_norm": 3.43339467048645,
      "learning_rate": 3.638747316497599e-07,
      "loss": 0.6772,
      "step": 42030
    },
    {
      "epoch": 0.6544093035565554,
      "grad_norm": 3.0115959644317627,
      "learning_rate": 3.6379279264515495e-07,
      "loss": 0.7473,
      "step": 42035
    },
    {
      "epoch": 0.6544871445585247,
      "grad_norm": 3.6355514526367188,
      "learning_rate": 3.637108536405499e-07,
      "loss": 0.6699,
      "step": 42040
    },
    {
      "epoch": 0.6545649855604941,
      "grad_norm": 4.174508571624756,
      "learning_rate": 3.63628914635945e-07,
      "loss": 0.7371,
      "step": 42045
    },
    {
      "epoch": 0.6546428265624635,
      "grad_norm": 4.423547744750977,
      "learning_rate": 3.6354697563134e-07,
      "loss": 0.7849,
      "step": 42050
    },
    {
      "epoch": 0.6547206675644329,
      "grad_norm": 2.88385009765625,
      "learning_rate": 3.6346503662673503e-07,
      "loss": 0.7238,
      "step": 42055
    },
    {
      "epoch": 0.6547985085664023,
      "grad_norm": 2.82192325592041,
      "learning_rate": 3.633830976221301e-07,
      "loss": 0.65,
      "step": 42060
    },
    {
      "epoch": 0.6548763495683716,
      "grad_norm": 3.2152857780456543,
      "learning_rate": 3.633011586175251e-07,
      "loss": 0.7023,
      "step": 42065
    },
    {
      "epoch": 0.654954190570341,
      "grad_norm": 3.7792491912841797,
      "learning_rate": 3.632192196129201e-07,
      "loss": 0.7197,
      "step": 42070
    },
    {
      "epoch": 0.6550320315723104,
      "grad_norm": 2.9485833644866943,
      "learning_rate": 3.631372806083152e-07,
      "loss": 0.6239,
      "step": 42075
    },
    {
      "epoch": 0.6551098725742798,
      "grad_norm": 3.3722829818725586,
      "learning_rate": 3.630553416037102e-07,
      "loss": 0.6995,
      "step": 42080
    },
    {
      "epoch": 0.6551877135762492,
      "grad_norm": 4.927794933319092,
      "learning_rate": 3.629734025991052e-07,
      "loss": 0.6768,
      "step": 42085
    },
    {
      "epoch": 0.6552655545782186,
      "grad_norm": 4.0036773681640625,
      "learning_rate": 3.6289146359450024e-07,
      "loss": 0.7702,
      "step": 42090
    },
    {
      "epoch": 0.655343395580188,
      "grad_norm": 3.7197577953338623,
      "learning_rate": 3.6280952458989526e-07,
      "loss": 0.8438,
      "step": 42095
    },
    {
      "epoch": 0.6554212365821573,
      "grad_norm": 5.001380443572998,
      "learning_rate": 3.6272758558529034e-07,
      "loss": 0.8231,
      "step": 42100
    },
    {
      "epoch": 0.6554990775841266,
      "grad_norm": 3.9659597873687744,
      "learning_rate": 3.626456465806853e-07,
      "loss": 0.6143,
      "step": 42105
    },
    {
      "epoch": 0.655576918586096,
      "grad_norm": 4.938342094421387,
      "learning_rate": 3.6256370757608033e-07,
      "loss": 0.6764,
      "step": 42110
    },
    {
      "epoch": 0.6556547595880654,
      "grad_norm": 4.644030570983887,
      "learning_rate": 3.624817685714754e-07,
      "loss": 0.7302,
      "step": 42115
    },
    {
      "epoch": 0.6557326005900348,
      "grad_norm": 4.74369478225708,
      "learning_rate": 3.623998295668704e-07,
      "loss": 0.6247,
      "step": 42120
    },
    {
      "epoch": 0.6558104415920042,
      "grad_norm": 2.8321585655212402,
      "learning_rate": 3.623178905622654e-07,
      "loss": 0.7748,
      "step": 42125
    },
    {
      "epoch": 0.6558882825939736,
      "grad_norm": 3.30884051322937,
      "learning_rate": 3.6223595155766047e-07,
      "loss": 0.741,
      "step": 42130
    },
    {
      "epoch": 0.655966123595943,
      "grad_norm": 3.768672227859497,
      "learning_rate": 3.621540125530555e-07,
      "loss": 0.7431,
      "step": 42135
    },
    {
      "epoch": 0.6560439645979123,
      "grad_norm": 22.723468780517578,
      "learning_rate": 3.6207207354845057e-07,
      "loss": 0.7879,
      "step": 42140
    },
    {
      "epoch": 0.6561218055998816,
      "grad_norm": 3.341456651687622,
      "learning_rate": 3.6199013454384553e-07,
      "loss": 0.8022,
      "step": 42145
    },
    {
      "epoch": 0.656199646601851,
      "grad_norm": 4.197352886199951,
      "learning_rate": 3.6190819553924056e-07,
      "loss": 0.6938,
      "step": 42150
    },
    {
      "epoch": 0.6562774876038204,
      "grad_norm": 4.481323719024658,
      "learning_rate": 3.6182625653463563e-07,
      "loss": 0.755,
      "step": 42155
    },
    {
      "epoch": 0.6563553286057898,
      "grad_norm": 3.8212993144989014,
      "learning_rate": 3.6174431753003065e-07,
      "loss": 0.7068,
      "step": 42160
    },
    {
      "epoch": 0.6564331696077592,
      "grad_norm": 3.098504066467285,
      "learning_rate": 3.616623785254256e-07,
      "loss": 0.7263,
      "step": 42165
    },
    {
      "epoch": 0.6565110106097286,
      "grad_norm": 4.782130718231201,
      "learning_rate": 3.615804395208207e-07,
      "loss": 0.6579,
      "step": 42170
    },
    {
      "epoch": 0.656588851611698,
      "grad_norm": 4.623470306396484,
      "learning_rate": 3.614985005162157e-07,
      "loss": 0.6868,
      "step": 42175
    },
    {
      "epoch": 0.6566666926136673,
      "grad_norm": 5.866987705230713,
      "learning_rate": 3.6141656151161074e-07,
      "loss": 0.8275,
      "step": 42180
    },
    {
      "epoch": 0.6567445336156367,
      "grad_norm": 3.9063549041748047,
      "learning_rate": 3.6133462250700576e-07,
      "loss": 0.8131,
      "step": 42185
    },
    {
      "epoch": 0.6568223746176061,
      "grad_norm": 3.1429500579833984,
      "learning_rate": 3.612526835024008e-07,
      "loss": 0.7201,
      "step": 42190
    },
    {
      "epoch": 0.6569002156195755,
      "grad_norm": 3.863953113555908,
      "learning_rate": 3.611707444977958e-07,
      "loss": 0.6816,
      "step": 42195
    },
    {
      "epoch": 0.6569780566215448,
      "grad_norm": 5.157500743865967,
      "learning_rate": 3.610888054931909e-07,
      "loss": 0.7436,
      "step": 42200
    },
    {
      "epoch": 0.6570558976235142,
      "grad_norm": 2.552596092224121,
      "learning_rate": 3.6100686648858585e-07,
      "loss": 0.6922,
      "step": 42205
    },
    {
      "epoch": 0.6571337386254836,
      "grad_norm": 2.2045555114746094,
      "learning_rate": 3.609249274839809e-07,
      "loss": 0.7557,
      "step": 42210
    },
    {
      "epoch": 0.6572115796274529,
      "grad_norm": 2.9920003414154053,
      "learning_rate": 3.6084298847937595e-07,
      "loss": 0.6921,
      "step": 42215
    },
    {
      "epoch": 0.6572894206294223,
      "grad_norm": 6.989012241363525,
      "learning_rate": 3.6076104947477097e-07,
      "loss": 0.7613,
      "step": 42220
    },
    {
      "epoch": 0.6573672616313917,
      "grad_norm": 3.2715814113616943,
      "learning_rate": 3.60679110470166e-07,
      "loss": 0.7753,
      "step": 42225
    },
    {
      "epoch": 0.6574451026333611,
      "grad_norm": 4.237335205078125,
      "learning_rate": 3.60597171465561e-07,
      "loss": 0.7278,
      "step": 42230
    },
    {
      "epoch": 0.6575229436353305,
      "grad_norm": 6.869584560394287,
      "learning_rate": 3.6051523246095604e-07,
      "loss": 0.7193,
      "step": 42235
    },
    {
      "epoch": 0.6576007846372999,
      "grad_norm": 4.894192218780518,
      "learning_rate": 3.604332934563511e-07,
      "loss": 0.7488,
      "step": 42240
    },
    {
      "epoch": 0.6576786256392693,
      "grad_norm": 5.429255962371826,
      "learning_rate": 3.603513544517461e-07,
      "loss": 0.6728,
      "step": 42245
    },
    {
      "epoch": 0.6577564666412387,
      "grad_norm": 4.802512168884277,
      "learning_rate": 3.602694154471411e-07,
      "loss": 0.7803,
      "step": 42250
    },
    {
      "epoch": 0.6578343076432079,
      "grad_norm": 3.386401891708374,
      "learning_rate": 3.601874764425362e-07,
      "loss": 0.6954,
      "step": 42255
    },
    {
      "epoch": 0.6579121486451773,
      "grad_norm": 3.4135310649871826,
      "learning_rate": 3.601055374379312e-07,
      "loss": 0.8275,
      "step": 42260
    },
    {
      "epoch": 0.6579899896471467,
      "grad_norm": 3.3160500526428223,
      "learning_rate": 3.6002359843332617e-07,
      "loss": 0.6474,
      "step": 42265
    },
    {
      "epoch": 0.6580678306491161,
      "grad_norm": 3.212941884994507,
      "learning_rate": 3.5994165942872124e-07,
      "loss": 0.709,
      "step": 42270
    },
    {
      "epoch": 0.6581456716510855,
      "grad_norm": 2.708740472793579,
      "learning_rate": 3.5985972042411626e-07,
      "loss": 0.7397,
      "step": 42275
    },
    {
      "epoch": 0.6582235126530549,
      "grad_norm": 2.608743190765381,
      "learning_rate": 3.5977778141951134e-07,
      "loss": 0.7344,
      "step": 42280
    },
    {
      "epoch": 0.6583013536550243,
      "grad_norm": 5.463550090789795,
      "learning_rate": 3.596958424149063e-07,
      "loss": 0.7503,
      "step": 42285
    },
    {
      "epoch": 0.6583791946569936,
      "grad_norm": 3.688950300216675,
      "learning_rate": 3.5961390341030133e-07,
      "loss": 0.6855,
      "step": 42290
    },
    {
      "epoch": 0.658457035658963,
      "grad_norm": 3.595453977584839,
      "learning_rate": 3.595319644056964e-07,
      "loss": 0.5953,
      "step": 42295
    },
    {
      "epoch": 0.6585348766609324,
      "grad_norm": 6.927585601806641,
      "learning_rate": 3.594500254010914e-07,
      "loss": 0.683,
      "step": 42300
    },
    {
      "epoch": 0.6586127176629017,
      "grad_norm": 3.900540351867676,
      "learning_rate": 3.593680863964864e-07,
      "loss": 0.756,
      "step": 42305
    },
    {
      "epoch": 0.6586905586648711,
      "grad_norm": 5.130527019500732,
      "learning_rate": 3.5928614739188147e-07,
      "loss": 0.722,
      "step": 42310
    },
    {
      "epoch": 0.6587683996668405,
      "grad_norm": 2.8873579502105713,
      "learning_rate": 3.592042083872765e-07,
      "loss": 0.7066,
      "step": 42315
    },
    {
      "epoch": 0.6588462406688099,
      "grad_norm": 2.9646666049957275,
      "learning_rate": 3.591222693826715e-07,
      "loss": 0.7051,
      "step": 42320
    },
    {
      "epoch": 0.6589240816707793,
      "grad_norm": 5.481849193572998,
      "learning_rate": 3.5904033037806654e-07,
      "loss": 0.7098,
      "step": 42325
    },
    {
      "epoch": 0.6590019226727486,
      "grad_norm": 4.012901306152344,
      "learning_rate": 3.5895839137346156e-07,
      "loss": 0.7527,
      "step": 42330
    },
    {
      "epoch": 0.659079763674718,
      "grad_norm": 3.042602777481079,
      "learning_rate": 3.5887645236885663e-07,
      "loss": 0.5895,
      "step": 42335
    },
    {
      "epoch": 0.6591576046766874,
      "grad_norm": 2.9244980812072754,
      "learning_rate": 3.5879451336425165e-07,
      "loss": 0.6704,
      "step": 42340
    },
    {
      "epoch": 0.6592354456786568,
      "grad_norm": 6.3195271492004395,
      "learning_rate": 3.587125743596466e-07,
      "loss": 0.854,
      "step": 42345
    },
    {
      "epoch": 0.6593132866806262,
      "grad_norm": 2.914604902267456,
      "learning_rate": 3.586306353550417e-07,
      "loss": 0.7588,
      "step": 42350
    },
    {
      "epoch": 0.6593911276825956,
      "grad_norm": 4.81829833984375,
      "learning_rate": 3.585486963504367e-07,
      "loss": 0.678,
      "step": 42355
    },
    {
      "epoch": 0.659468968684565,
      "grad_norm": 2.6965172290802,
      "learning_rate": 3.5846675734583174e-07,
      "loss": 0.6722,
      "step": 42360
    },
    {
      "epoch": 0.6595468096865342,
      "grad_norm": 3.6982998847961426,
      "learning_rate": 3.583848183412268e-07,
      "loss": 0.6844,
      "step": 42365
    },
    {
      "epoch": 0.6596246506885036,
      "grad_norm": 3.6694717407226562,
      "learning_rate": 3.583028793366218e-07,
      "loss": 0.7262,
      "step": 42370
    },
    {
      "epoch": 0.659702491690473,
      "grad_norm": 4.186232089996338,
      "learning_rate": 3.582209403320168e-07,
      "loss": 0.7321,
      "step": 42375
    },
    {
      "epoch": 0.6597803326924424,
      "grad_norm": 5.403656482696533,
      "learning_rate": 3.581390013274119e-07,
      "loss": 0.726,
      "step": 42380
    },
    {
      "epoch": 0.6598581736944118,
      "grad_norm": 2.7567694187164307,
      "learning_rate": 3.580570623228069e-07,
      "loss": 0.5997,
      "step": 42385
    },
    {
      "epoch": 0.6599360146963812,
      "grad_norm": 5.498208522796631,
      "learning_rate": 3.5797512331820187e-07,
      "loss": 0.8235,
      "step": 42390
    },
    {
      "epoch": 0.6600138556983506,
      "grad_norm": 3.6834511756896973,
      "learning_rate": 3.5789318431359695e-07,
      "loss": 0.6069,
      "step": 42395
    },
    {
      "epoch": 0.66009169670032,
      "grad_norm": 3.7909188270568848,
      "learning_rate": 3.5781124530899197e-07,
      "loss": 0.7007,
      "step": 42400
    },
    {
      "epoch": 0.6601695377022893,
      "grad_norm": 4.241979598999023,
      "learning_rate": 3.5772930630438705e-07,
      "loss": 0.8293,
      "step": 42405
    },
    {
      "epoch": 0.6602473787042586,
      "grad_norm": 3.588465690612793,
      "learning_rate": 3.57647367299782e-07,
      "loss": 0.7962,
      "step": 42410
    },
    {
      "epoch": 0.660325219706228,
      "grad_norm": 4.2016472816467285,
      "learning_rate": 3.5756542829517704e-07,
      "loss": 0.8229,
      "step": 42415
    },
    {
      "epoch": 0.6604030607081974,
      "grad_norm": 5.786152362823486,
      "learning_rate": 3.574834892905721e-07,
      "loss": 0.8555,
      "step": 42420
    },
    {
      "epoch": 0.6604809017101668,
      "grad_norm": 3.8570468425750732,
      "learning_rate": 3.5740155028596713e-07,
      "loss": 0.9254,
      "step": 42425
    },
    {
      "epoch": 0.6605587427121362,
      "grad_norm": 5.963582992553711,
      "learning_rate": 3.573196112813621e-07,
      "loss": 0.8311,
      "step": 42430
    },
    {
      "epoch": 0.6606365837141056,
      "grad_norm": 2.422403335571289,
      "learning_rate": 3.572376722767572e-07,
      "loss": 0.6405,
      "step": 42435
    },
    {
      "epoch": 0.660714424716075,
      "grad_norm": 4.573333263397217,
      "learning_rate": 3.571557332721522e-07,
      "loss": 0.7722,
      "step": 42440
    },
    {
      "epoch": 0.6607922657180443,
      "grad_norm": 2.9828858375549316,
      "learning_rate": 3.570737942675472e-07,
      "loss": 0.6997,
      "step": 42445
    },
    {
      "epoch": 0.6608701067200137,
      "grad_norm": 5.616395473480225,
      "learning_rate": 3.5699185526294224e-07,
      "loss": 0.7569,
      "step": 42450
    },
    {
      "epoch": 0.6609479477219831,
      "grad_norm": 3.751662254333496,
      "learning_rate": 3.5690991625833726e-07,
      "loss": 0.7978,
      "step": 42455
    },
    {
      "epoch": 0.6610257887239525,
      "grad_norm": 7.5746588706970215,
      "learning_rate": 3.5682797725373234e-07,
      "loss": 0.9483,
      "step": 42460
    },
    {
      "epoch": 0.6611036297259218,
      "grad_norm": 2.8900678157806396,
      "learning_rate": 3.5674603824912736e-07,
      "loss": 0.8354,
      "step": 42465
    },
    {
      "epoch": 0.6611814707278912,
      "grad_norm": 4.748530864715576,
      "learning_rate": 3.5666409924452233e-07,
      "loss": 0.6759,
      "step": 42470
    },
    {
      "epoch": 0.6612593117298606,
      "grad_norm": 4.789640426635742,
      "learning_rate": 3.565821602399174e-07,
      "loss": 0.8503,
      "step": 42475
    },
    {
      "epoch": 0.6613371527318299,
      "grad_norm": 5.952419281005859,
      "learning_rate": 3.5650022123531243e-07,
      "loss": 0.7054,
      "step": 42480
    },
    {
      "epoch": 0.6614149937337993,
      "grad_norm": 3.925952434539795,
      "learning_rate": 3.5641828223070745e-07,
      "loss": 0.7565,
      "step": 42485
    },
    {
      "epoch": 0.6614928347357687,
      "grad_norm": 4.081397533416748,
      "learning_rate": 3.5633634322610247e-07,
      "loss": 0.7333,
      "step": 42490
    },
    {
      "epoch": 0.6615706757377381,
      "grad_norm": 3.1329903602600098,
      "learning_rate": 3.562544042214975e-07,
      "loss": 0.8815,
      "step": 42495
    },
    {
      "epoch": 0.6616485167397075,
      "grad_norm": 13.413626670837402,
      "learning_rate": 3.561724652168925e-07,
      "loss": 0.7099,
      "step": 42500
    },
    {
      "epoch": 0.6617263577416769,
      "grad_norm": 3.899742603302002,
      "learning_rate": 3.560905262122876e-07,
      "loss": 0.7069,
      "step": 42505
    },
    {
      "epoch": 0.6618041987436463,
      "grad_norm": 4.178427219390869,
      "learning_rate": 3.5600858720768256e-07,
      "loss": 0.6838,
      "step": 42510
    },
    {
      "epoch": 0.6618820397456157,
      "grad_norm": 3.002199172973633,
      "learning_rate": 3.559266482030776e-07,
      "loss": 0.7671,
      "step": 42515
    },
    {
      "epoch": 0.6619598807475849,
      "grad_norm": 4.241408824920654,
      "learning_rate": 3.5584470919847266e-07,
      "loss": 0.5612,
      "step": 42520
    },
    {
      "epoch": 0.6620377217495543,
      "grad_norm": 3.043455123901367,
      "learning_rate": 3.557627701938677e-07,
      "loss": 0.7932,
      "step": 42525
    },
    {
      "epoch": 0.6621155627515237,
      "grad_norm": 4.029059410095215,
      "learning_rate": 3.556808311892627e-07,
      "loss": 0.677,
      "step": 42530
    },
    {
      "epoch": 0.6621934037534931,
      "grad_norm": 3.905036211013794,
      "learning_rate": 3.555988921846577e-07,
      "loss": 0.7299,
      "step": 42535
    },
    {
      "epoch": 0.6622712447554625,
      "grad_norm": 2.875826597213745,
      "learning_rate": 3.5551695318005274e-07,
      "loss": 0.727,
      "step": 42540
    },
    {
      "epoch": 0.6623490857574319,
      "grad_norm": 3.761948347091675,
      "learning_rate": 3.554350141754478e-07,
      "loss": 0.7414,
      "step": 42545
    },
    {
      "epoch": 0.6624269267594013,
      "grad_norm": 3.4006121158599854,
      "learning_rate": 3.553530751708428e-07,
      "loss": 0.6942,
      "step": 42550
    },
    {
      "epoch": 0.6625047677613706,
      "grad_norm": 3.3827552795410156,
      "learning_rate": 3.552711361662378e-07,
      "loss": 0.7165,
      "step": 42555
    },
    {
      "epoch": 0.66258260876334,
      "grad_norm": 5.64990234375,
      "learning_rate": 3.551891971616329e-07,
      "loss": 0.6898,
      "step": 42560
    },
    {
      "epoch": 0.6626604497653094,
      "grad_norm": 3.4614367485046387,
      "learning_rate": 3.551072581570279e-07,
      "loss": 0.6741,
      "step": 42565
    },
    {
      "epoch": 0.6627382907672787,
      "grad_norm": 3.200205087661743,
      "learning_rate": 3.550253191524229e-07,
      "loss": 0.6995,
      "step": 42570
    },
    {
      "epoch": 0.6628161317692481,
      "grad_norm": 3.2221009731292725,
      "learning_rate": 3.5494338014781795e-07,
      "loss": 0.8052,
      "step": 42575
    },
    {
      "epoch": 0.6628939727712175,
      "grad_norm": 2.8508145809173584,
      "learning_rate": 3.5486144114321297e-07,
      "loss": 0.7606,
      "step": 42580
    },
    {
      "epoch": 0.6629718137731869,
      "grad_norm": 2.874998092651367,
      "learning_rate": 3.5477950213860805e-07,
      "loss": 0.8041,
      "step": 42585
    },
    {
      "epoch": 0.6630496547751563,
      "grad_norm": 4.830933570861816,
      "learning_rate": 3.54697563134003e-07,
      "loss": 0.7546,
      "step": 42590
    },
    {
      "epoch": 0.6631274957771256,
      "grad_norm": 2.573822498321533,
      "learning_rate": 3.5461562412939804e-07,
      "loss": 0.8182,
      "step": 42595
    },
    {
      "epoch": 0.663205336779095,
      "grad_norm": 3.2831828594207764,
      "learning_rate": 3.545336851247931e-07,
      "loss": 0.732,
      "step": 42600
    },
    {
      "epoch": 0.6632831777810644,
      "grad_norm": 4.074308395385742,
      "learning_rate": 3.5445174612018813e-07,
      "loss": 0.7,
      "step": 42605
    },
    {
      "epoch": 0.6633610187830338,
      "grad_norm": 2.7913310527801514,
      "learning_rate": 3.543698071155831e-07,
      "loss": 0.764,
      "step": 42610
    },
    {
      "epoch": 0.6634388597850032,
      "grad_norm": 6.573644161224365,
      "learning_rate": 3.542878681109782e-07,
      "loss": 0.7792,
      "step": 42615
    },
    {
      "epoch": 0.6635167007869726,
      "grad_norm": 3.43157696723938,
      "learning_rate": 3.542059291063732e-07,
      "loss": 0.7257,
      "step": 42620
    },
    {
      "epoch": 0.663594541788942,
      "grad_norm": 3.5137135982513428,
      "learning_rate": 3.541239901017682e-07,
      "loss": 0.7609,
      "step": 42625
    },
    {
      "epoch": 0.6636723827909112,
      "grad_norm": 2.4340147972106934,
      "learning_rate": 3.5404205109716324e-07,
      "loss": 0.7918,
      "step": 42630
    },
    {
      "epoch": 0.6637502237928806,
      "grad_norm": 3.459289073944092,
      "learning_rate": 3.5396011209255827e-07,
      "loss": 0.7204,
      "step": 42635
    },
    {
      "epoch": 0.66382806479485,
      "grad_norm": 4.456779956817627,
      "learning_rate": 3.538781730879533e-07,
      "loss": 0.7488,
      "step": 42640
    },
    {
      "epoch": 0.6639059057968194,
      "grad_norm": 7.907991409301758,
      "learning_rate": 3.5379623408334836e-07,
      "loss": 0.7976,
      "step": 42645
    },
    {
      "epoch": 0.6639837467987888,
      "grad_norm": 3.248631477355957,
      "learning_rate": 3.5371429507874333e-07,
      "loss": 0.7624,
      "step": 42650
    },
    {
      "epoch": 0.6640615878007582,
      "grad_norm": 2.6080527305603027,
      "learning_rate": 3.536323560741384e-07,
      "loss": 0.7549,
      "step": 42655
    },
    {
      "epoch": 0.6641394288027276,
      "grad_norm": 3.279320478439331,
      "learning_rate": 3.5355041706953343e-07,
      "loss": 0.7059,
      "step": 42660
    },
    {
      "epoch": 0.664217269804697,
      "grad_norm": 4.202732563018799,
      "learning_rate": 3.5346847806492845e-07,
      "loss": 0.7511,
      "step": 42665
    },
    {
      "epoch": 0.6642951108066663,
      "grad_norm": 5.315632343292236,
      "learning_rate": 3.5338653906032347e-07,
      "loss": 0.8495,
      "step": 42670
    },
    {
      "epoch": 0.6643729518086356,
      "grad_norm": 4.199154376983643,
      "learning_rate": 3.533046000557185e-07,
      "loss": 0.745,
      "step": 42675
    },
    {
      "epoch": 0.664450792810605,
      "grad_norm": 4.275646209716797,
      "learning_rate": 3.532226610511135e-07,
      "loss": 0.7455,
      "step": 42680
    },
    {
      "epoch": 0.6645286338125744,
      "grad_norm": 4.551513195037842,
      "learning_rate": 3.531407220465086e-07,
      "loss": 0.7422,
      "step": 42685
    },
    {
      "epoch": 0.6646064748145438,
      "grad_norm": 3.6756105422973633,
      "learning_rate": 3.530587830419036e-07,
      "loss": 0.7453,
      "step": 42690
    },
    {
      "epoch": 0.6646843158165132,
      "grad_norm": 6.201836109161377,
      "learning_rate": 3.529768440372986e-07,
      "loss": 0.7873,
      "step": 42695
    },
    {
      "epoch": 0.6647621568184826,
      "grad_norm": 3.776667833328247,
      "learning_rate": 3.5289490503269366e-07,
      "loss": 0.8061,
      "step": 42700
    },
    {
      "epoch": 0.664839997820452,
      "grad_norm": 14.086362838745117,
      "learning_rate": 3.528129660280887e-07,
      "loss": 0.7743,
      "step": 42705
    },
    {
      "epoch": 0.6649178388224213,
      "grad_norm": 2.663594961166382,
      "learning_rate": 3.5273102702348375e-07,
      "loss": 0.7721,
      "step": 42710
    },
    {
      "epoch": 0.6649956798243907,
      "grad_norm": 2.6219124794006348,
      "learning_rate": 3.526490880188787e-07,
      "loss": 0.6726,
      "step": 42715
    },
    {
      "epoch": 0.6650735208263601,
      "grad_norm": 3.523674964904785,
      "learning_rate": 3.5256714901427374e-07,
      "loss": 0.8027,
      "step": 42720
    },
    {
      "epoch": 0.6651513618283295,
      "grad_norm": 3.7221858501434326,
      "learning_rate": 3.524852100096688e-07,
      "loss": 0.6888,
      "step": 42725
    },
    {
      "epoch": 0.6652292028302988,
      "grad_norm": 7.166980743408203,
      "learning_rate": 3.5240327100506384e-07,
      "loss": 0.8595,
      "step": 42730
    },
    {
      "epoch": 0.6653070438322682,
      "grad_norm": 5.685295104980469,
      "learning_rate": 3.523213320004588e-07,
      "loss": 0.7578,
      "step": 42735
    },
    {
      "epoch": 0.6653848848342376,
      "grad_norm": 4.930285453796387,
      "learning_rate": 3.522393929958539e-07,
      "loss": 0.7623,
      "step": 42740
    },
    {
      "epoch": 0.6654627258362069,
      "grad_norm": 3.8471548557281494,
      "learning_rate": 3.521574539912489e-07,
      "loss": 0.7049,
      "step": 42745
    },
    {
      "epoch": 0.6655405668381763,
      "grad_norm": 6.0011491775512695,
      "learning_rate": 3.5207551498664393e-07,
      "loss": 0.7286,
      "step": 42750
    },
    {
      "epoch": 0.6656184078401457,
      "grad_norm": 4.614305019378662,
      "learning_rate": 3.5199357598203895e-07,
      "loss": 0.7855,
      "step": 42755
    },
    {
      "epoch": 0.6656962488421151,
      "grad_norm": 3.696873426437378,
      "learning_rate": 3.5191163697743397e-07,
      "loss": 0.7848,
      "step": 42760
    },
    {
      "epoch": 0.6657740898440845,
      "grad_norm": 6.268959999084473,
      "learning_rate": 3.5182969797282905e-07,
      "loss": 0.8721,
      "step": 42765
    },
    {
      "epoch": 0.6658519308460539,
      "grad_norm": 5.392154693603516,
      "learning_rate": 3.5174775896822407e-07,
      "loss": 0.7521,
      "step": 42770
    },
    {
      "epoch": 0.6659297718480233,
      "grad_norm": 5.628934860229492,
      "learning_rate": 3.5166581996361904e-07,
      "loss": 0.8044,
      "step": 42775
    },
    {
      "epoch": 0.6660076128499927,
      "grad_norm": 3.6572206020355225,
      "learning_rate": 3.515838809590141e-07,
      "loss": 0.8912,
      "step": 42780
    },
    {
      "epoch": 0.6660854538519619,
      "grad_norm": 3.9402010440826416,
      "learning_rate": 3.5150194195440913e-07,
      "loss": 0.8195,
      "step": 42785
    },
    {
      "epoch": 0.6661632948539313,
      "grad_norm": 4.384683609008789,
      "learning_rate": 3.5142000294980416e-07,
      "loss": 0.7946,
      "step": 42790
    },
    {
      "epoch": 0.6662411358559007,
      "grad_norm": 2.881610155105591,
      "learning_rate": 3.513380639451992e-07,
      "loss": 0.6541,
      "step": 42795
    },
    {
      "epoch": 0.6663189768578701,
      "grad_norm": 3.997616767883301,
      "learning_rate": 3.512561249405942e-07,
      "loss": 0.6992,
      "step": 42800
    },
    {
      "epoch": 0.6663968178598395,
      "grad_norm": 3.2822213172912598,
      "learning_rate": 3.511741859359892e-07,
      "loss": 0.6711,
      "step": 42805
    },
    {
      "epoch": 0.6664746588618089,
      "grad_norm": 3.141810655593872,
      "learning_rate": 3.510922469313843e-07,
      "loss": 0.7269,
      "step": 42810
    },
    {
      "epoch": 0.6665524998637783,
      "grad_norm": 3.760939836502075,
      "learning_rate": 3.5101030792677927e-07,
      "loss": 0.7772,
      "step": 42815
    },
    {
      "epoch": 0.6666303408657476,
      "grad_norm": 13.503973007202148,
      "learning_rate": 3.509283689221743e-07,
      "loss": 0.7501,
      "step": 42820
    },
    {
      "epoch": 0.666708181867717,
      "grad_norm": 5.004620552062988,
      "learning_rate": 3.5084642991756936e-07,
      "loss": 0.7956,
      "step": 42825
    },
    {
      "epoch": 0.6667860228696864,
      "grad_norm": 6.332820415496826,
      "learning_rate": 3.507644909129644e-07,
      "loss": 0.7973,
      "step": 42830
    },
    {
      "epoch": 0.6668638638716557,
      "grad_norm": 2.878511905670166,
      "learning_rate": 3.506825519083594e-07,
      "loss": 0.7399,
      "step": 42835
    },
    {
      "epoch": 0.6669417048736251,
      "grad_norm": 3.559971809387207,
      "learning_rate": 3.5060061290375443e-07,
      "loss": 0.8554,
      "step": 42840
    },
    {
      "epoch": 0.6670195458755945,
      "grad_norm": 3.6511693000793457,
      "learning_rate": 3.5051867389914945e-07,
      "loss": 0.7854,
      "step": 42845
    },
    {
      "epoch": 0.6670973868775639,
      "grad_norm": 4.178702354431152,
      "learning_rate": 3.504367348945445e-07,
      "loss": 0.6373,
      "step": 42850
    },
    {
      "epoch": 0.6671752278795333,
      "grad_norm": 2.882660150527954,
      "learning_rate": 3.503547958899395e-07,
      "loss": 0.6487,
      "step": 42855
    },
    {
      "epoch": 0.6672530688815026,
      "grad_norm": 3.491137981414795,
      "learning_rate": 3.502728568853345e-07,
      "loss": 0.7719,
      "step": 42860
    },
    {
      "epoch": 0.667330909883472,
      "grad_norm": 2.9467668533325195,
      "learning_rate": 3.501909178807296e-07,
      "loss": 0.7749,
      "step": 42865
    },
    {
      "epoch": 0.6674087508854414,
      "grad_norm": 5.262530326843262,
      "learning_rate": 3.501089788761246e-07,
      "loss": 0.6352,
      "step": 42870
    },
    {
      "epoch": 0.6674865918874108,
      "grad_norm": 3.5344624519348145,
      "learning_rate": 3.500270398715196e-07,
      "loss": 0.8253,
      "step": 42875
    },
    {
      "epoch": 0.6675644328893802,
      "grad_norm": 3.275409460067749,
      "learning_rate": 3.4994510086691466e-07,
      "loss": 0.7754,
      "step": 42880
    },
    {
      "epoch": 0.6676422738913496,
      "grad_norm": 10.296807289123535,
      "learning_rate": 3.498631618623097e-07,
      "loss": 0.834,
      "step": 42885
    },
    {
      "epoch": 0.667720114893319,
      "grad_norm": 3.546525001525879,
      "learning_rate": 3.4978122285770475e-07,
      "loss": 0.7117,
      "step": 42890
    },
    {
      "epoch": 0.6677979558952882,
      "grad_norm": 4.0889058113098145,
      "learning_rate": 3.496992838530997e-07,
      "loss": 0.7762,
      "step": 42895
    },
    {
      "epoch": 0.6678757968972576,
      "grad_norm": 4.494797229766846,
      "learning_rate": 3.4961734484849474e-07,
      "loss": 0.7342,
      "step": 42900
    },
    {
      "epoch": 0.667953637899227,
      "grad_norm": 2.7845354080200195,
      "learning_rate": 3.495354058438898e-07,
      "loss": 0.7475,
      "step": 42905
    },
    {
      "epoch": 0.6680314789011964,
      "grad_norm": 3.308802843093872,
      "learning_rate": 3.4945346683928484e-07,
      "loss": 0.7619,
      "step": 42910
    },
    {
      "epoch": 0.6681093199031658,
      "grad_norm": 4.663692474365234,
      "learning_rate": 3.493715278346798e-07,
      "loss": 0.7409,
      "step": 42915
    },
    {
      "epoch": 0.6681871609051352,
      "grad_norm": 3.6634132862091064,
      "learning_rate": 3.492895888300749e-07,
      "loss": 0.6258,
      "step": 42920
    },
    {
      "epoch": 0.6682650019071046,
      "grad_norm": 3.3925750255584717,
      "learning_rate": 3.492076498254699e-07,
      "loss": 0.7106,
      "step": 42925
    },
    {
      "epoch": 0.668342842909074,
      "grad_norm": 3.8225646018981934,
      "learning_rate": 3.4912571082086493e-07,
      "loss": 0.7928,
      "step": 42930
    },
    {
      "epoch": 0.6684206839110433,
      "grad_norm": 3.0660603046417236,
      "learning_rate": 3.4904377181625995e-07,
      "loss": 0.6447,
      "step": 42935
    },
    {
      "epoch": 0.6684985249130126,
      "grad_norm": 2.3726539611816406,
      "learning_rate": 3.4896183281165497e-07,
      "loss": 0.7238,
      "step": 42940
    },
    {
      "epoch": 0.668576365914982,
      "grad_norm": 4.054858684539795,
      "learning_rate": 3.4887989380705e-07,
      "loss": 0.7642,
      "step": 42945
    },
    {
      "epoch": 0.6686542069169514,
      "grad_norm": 4.149957180023193,
      "learning_rate": 3.4879795480244507e-07,
      "loss": 0.6925,
      "step": 42950
    },
    {
      "epoch": 0.6687320479189208,
      "grad_norm": 6.917638778686523,
      "learning_rate": 3.4871601579784004e-07,
      "loss": 0.763,
      "step": 42955
    },
    {
      "epoch": 0.6688098889208902,
      "grad_norm": 3.0221285820007324,
      "learning_rate": 3.486340767932351e-07,
      "loss": 0.8442,
      "step": 42960
    },
    {
      "epoch": 0.6688877299228596,
      "grad_norm": 4.234553337097168,
      "learning_rate": 3.4855213778863014e-07,
      "loss": 0.7856,
      "step": 42965
    },
    {
      "epoch": 0.6689655709248289,
      "grad_norm": 2.9445300102233887,
      "learning_rate": 3.4847019878402516e-07,
      "loss": 0.6782,
      "step": 42970
    },
    {
      "epoch": 0.6690434119267983,
      "grad_norm": 3.888066291809082,
      "learning_rate": 3.483882597794202e-07,
      "loss": 0.7499,
      "step": 42975
    },
    {
      "epoch": 0.6691212529287677,
      "grad_norm": 6.645394325256348,
      "learning_rate": 3.483063207748152e-07,
      "loss": 0.8572,
      "step": 42980
    },
    {
      "epoch": 0.6691990939307371,
      "grad_norm": 6.322195529937744,
      "learning_rate": 3.482243817702102e-07,
      "loss": 0.8463,
      "step": 42985
    },
    {
      "epoch": 0.6692769349327065,
      "grad_norm": 2.90338134765625,
      "learning_rate": 3.481424427656053e-07,
      "loss": 0.8826,
      "step": 42990
    },
    {
      "epoch": 0.6693547759346758,
      "grad_norm": 3.12627911567688,
      "learning_rate": 3.4806050376100027e-07,
      "loss": 0.7169,
      "step": 42995
    },
    {
      "epoch": 0.6694326169366452,
      "grad_norm": 2.4262895584106445,
      "learning_rate": 3.479785647563953e-07,
      "loss": 0.6517,
      "step": 43000
    },
    {
      "epoch": 0.6695104579386146,
      "grad_norm": 2.652876853942871,
      "learning_rate": 3.4789662575179036e-07,
      "loss": 0.808,
      "step": 43005
    },
    {
      "epoch": 0.6695882989405839,
      "grad_norm": 2.899812698364258,
      "learning_rate": 3.478146867471854e-07,
      "loss": 0.6674,
      "step": 43010
    },
    {
      "epoch": 0.6696661399425533,
      "grad_norm": 5.46381139755249,
      "learning_rate": 3.4773274774258046e-07,
      "loss": 0.8373,
      "step": 43015
    },
    {
      "epoch": 0.6697439809445227,
      "grad_norm": 6.055871963500977,
      "learning_rate": 3.4765080873797543e-07,
      "loss": 0.7808,
      "step": 43020
    },
    {
      "epoch": 0.6698218219464921,
      "grad_norm": 3.7345707416534424,
      "learning_rate": 3.4756886973337045e-07,
      "loss": 0.8666,
      "step": 43025
    },
    {
      "epoch": 0.6698996629484615,
      "grad_norm": 5.38214635848999,
      "learning_rate": 3.474869307287655e-07,
      "loss": 0.7271,
      "step": 43030
    },
    {
      "epoch": 0.6699775039504309,
      "grad_norm": 3.236685037612915,
      "learning_rate": 3.4740499172416055e-07,
      "loss": 0.7162,
      "step": 43035
    },
    {
      "epoch": 0.6700553449524003,
      "grad_norm": 5.348214626312256,
      "learning_rate": 3.473230527195555e-07,
      "loss": 0.6941,
      "step": 43040
    },
    {
      "epoch": 0.6701331859543697,
      "grad_norm": 3.4515814781188965,
      "learning_rate": 3.472411137149506e-07,
      "loss": 0.6988,
      "step": 43045
    },
    {
      "epoch": 0.6702110269563389,
      "grad_norm": 3.339134931564331,
      "learning_rate": 3.471591747103456e-07,
      "loss": 0.7434,
      "step": 43050
    },
    {
      "epoch": 0.6702888679583083,
      "grad_norm": 2.7067840099334717,
      "learning_rate": 3.4707723570574064e-07,
      "loss": 0.7074,
      "step": 43055
    },
    {
      "epoch": 0.6703667089602777,
      "grad_norm": 5.749795436859131,
      "learning_rate": 3.4699529670113566e-07,
      "loss": 0.6194,
      "step": 43060
    },
    {
      "epoch": 0.6704445499622471,
      "grad_norm": 6.408483028411865,
      "learning_rate": 3.469133576965307e-07,
      "loss": 0.7155,
      "step": 43065
    },
    {
      "epoch": 0.6705223909642165,
      "grad_norm": 4.833917617797852,
      "learning_rate": 3.468314186919257e-07,
      "loss": 0.7323,
      "step": 43070
    },
    {
      "epoch": 0.6706002319661859,
      "grad_norm": 3.122100591659546,
      "learning_rate": 3.467494796873208e-07,
      "loss": 0.7914,
      "step": 43075
    },
    {
      "epoch": 0.6706780729681553,
      "grad_norm": 3.7405431270599365,
      "learning_rate": 3.4666754068271575e-07,
      "loss": 0.8156,
      "step": 43080
    },
    {
      "epoch": 0.6707559139701246,
      "grad_norm": 3.573119878768921,
      "learning_rate": 3.465856016781108e-07,
      "loss": 0.7426,
      "step": 43085
    },
    {
      "epoch": 0.670833754972094,
      "grad_norm": 4.233011722564697,
      "learning_rate": 3.4650366267350584e-07,
      "loss": 0.6949,
      "step": 43090
    },
    {
      "epoch": 0.6709115959740634,
      "grad_norm": 3.332555055618286,
      "learning_rate": 3.4642172366890086e-07,
      "loss": 0.8087,
      "step": 43095
    },
    {
      "epoch": 0.6709894369760327,
      "grad_norm": 3.2821602821350098,
      "learning_rate": 3.463397846642959e-07,
      "loss": 0.6458,
      "step": 43100
    },
    {
      "epoch": 0.6710672779780021,
      "grad_norm": 5.4845452308654785,
      "learning_rate": 3.462578456596909e-07,
      "loss": 0.6964,
      "step": 43105
    },
    {
      "epoch": 0.6711451189799715,
      "grad_norm": 5.266623497009277,
      "learning_rate": 3.4617590665508593e-07,
      "loss": 0.7579,
      "step": 43110
    },
    {
      "epoch": 0.6712229599819409,
      "grad_norm": 4.861994743347168,
      "learning_rate": 3.46093967650481e-07,
      "loss": 0.781,
      "step": 43115
    },
    {
      "epoch": 0.6713008009839103,
      "grad_norm": 2.7779431343078613,
      "learning_rate": 3.46012028645876e-07,
      "loss": 0.781,
      "step": 43120
    },
    {
      "epoch": 0.6713786419858796,
      "grad_norm": 3.6146883964538574,
      "learning_rate": 3.45930089641271e-07,
      "loss": 0.8103,
      "step": 43125
    },
    {
      "epoch": 0.671456482987849,
      "grad_norm": 4.589087009429932,
      "learning_rate": 3.4584815063666607e-07,
      "loss": 0.7659,
      "step": 43130
    },
    {
      "epoch": 0.6715343239898184,
      "grad_norm": 5.363763332366943,
      "learning_rate": 3.457662116320611e-07,
      "loss": 0.8992,
      "step": 43135
    },
    {
      "epoch": 0.6716121649917878,
      "grad_norm": 3.9445576667785645,
      "learning_rate": 3.4568427262745606e-07,
      "loss": 0.7396,
      "step": 43140
    },
    {
      "epoch": 0.6716900059937572,
      "grad_norm": 2.9027767181396484,
      "learning_rate": 3.4560233362285114e-07,
      "loss": 0.7598,
      "step": 43145
    },
    {
      "epoch": 0.6717678469957266,
      "grad_norm": 3.014850378036499,
      "learning_rate": 3.4552039461824616e-07,
      "loss": 0.7495,
      "step": 43150
    },
    {
      "epoch": 0.671845687997696,
      "grad_norm": 3.779632806777954,
      "learning_rate": 3.4543845561364123e-07,
      "loss": 0.705,
      "step": 43155
    },
    {
      "epoch": 0.6719235289996652,
      "grad_norm": 3.0600550174713135,
      "learning_rate": 3.453565166090362e-07,
      "loss": 0.7086,
      "step": 43160
    },
    {
      "epoch": 0.6720013700016346,
      "grad_norm": 4.010749340057373,
      "learning_rate": 3.452745776044312e-07,
      "loss": 0.678,
      "step": 43165
    },
    {
      "epoch": 0.672079211003604,
      "grad_norm": 3.132394313812256,
      "learning_rate": 3.451926385998263e-07,
      "loss": 0.6669,
      "step": 43170
    },
    {
      "epoch": 0.6721570520055734,
      "grad_norm": 3.0093703269958496,
      "learning_rate": 3.451106995952213e-07,
      "loss": 0.6798,
      "step": 43175
    },
    {
      "epoch": 0.6722348930075428,
      "grad_norm": 3.8264517784118652,
      "learning_rate": 3.450287605906163e-07,
      "loss": 0.7501,
      "step": 43180
    },
    {
      "epoch": 0.6723127340095122,
      "grad_norm": 3.6196916103363037,
      "learning_rate": 3.4494682158601136e-07,
      "loss": 0.7758,
      "step": 43185
    },
    {
      "epoch": 0.6723905750114816,
      "grad_norm": 4.23793888092041,
      "learning_rate": 3.448648825814064e-07,
      "loss": 0.7805,
      "step": 43190
    },
    {
      "epoch": 0.672468416013451,
      "grad_norm": 3.5586998462677,
      "learning_rate": 3.447829435768014e-07,
      "loss": 0.7595,
      "step": 43195
    },
    {
      "epoch": 0.6725462570154203,
      "grad_norm": 5.154542922973633,
      "learning_rate": 3.4470100457219643e-07,
      "loss": 0.7866,
      "step": 43200
    },
    {
      "epoch": 0.6726240980173896,
      "grad_norm": 3.478046417236328,
      "learning_rate": 3.4461906556759145e-07,
      "loss": 0.6854,
      "step": 43205
    },
    {
      "epoch": 0.672701939019359,
      "grad_norm": 2.8171770572662354,
      "learning_rate": 3.4453712656298653e-07,
      "loss": 0.7967,
      "step": 43210
    },
    {
      "epoch": 0.6727797800213284,
      "grad_norm": 4.6766839027404785,
      "learning_rate": 3.4445518755838155e-07,
      "loss": 0.6558,
      "step": 43215
    },
    {
      "epoch": 0.6728576210232978,
      "grad_norm": 3.1634795665740967,
      "learning_rate": 3.443732485537765e-07,
      "loss": 0.6413,
      "step": 43220
    },
    {
      "epoch": 0.6729354620252672,
      "grad_norm": 3.4578025341033936,
      "learning_rate": 3.442913095491716e-07,
      "loss": 0.7805,
      "step": 43225
    },
    {
      "epoch": 0.6730133030272366,
      "grad_norm": 3.019021987915039,
      "learning_rate": 3.442093705445666e-07,
      "loss": 0.6858,
      "step": 43230
    },
    {
      "epoch": 0.6730911440292059,
      "grad_norm": 4.485779762268066,
      "learning_rate": 3.4412743153996164e-07,
      "loss": 0.7185,
      "step": 43235
    },
    {
      "epoch": 0.6731689850311753,
      "grad_norm": 2.332998275756836,
      "learning_rate": 3.4404549253535666e-07,
      "loss": 0.6271,
      "step": 43240
    },
    {
      "epoch": 0.6732468260331447,
      "grad_norm": 3.819098949432373,
      "learning_rate": 3.439635535307517e-07,
      "loss": 0.6517,
      "step": 43245
    },
    {
      "epoch": 0.6733246670351141,
      "grad_norm": 10.25633716583252,
      "learning_rate": 3.438816145261467e-07,
      "loss": 0.7328,
      "step": 43250
    },
    {
      "epoch": 0.6734025080370835,
      "grad_norm": 3.5705032348632812,
      "learning_rate": 3.437996755215418e-07,
      "loss": 0.712,
      "step": 43255
    },
    {
      "epoch": 0.6734803490390529,
      "grad_norm": 7.906955242156982,
      "learning_rate": 3.4371773651693675e-07,
      "loss": 0.7311,
      "step": 43260
    },
    {
      "epoch": 0.6735581900410222,
      "grad_norm": 4.550412654876709,
      "learning_rate": 3.4363579751233177e-07,
      "loss": 0.7668,
      "step": 43265
    },
    {
      "epoch": 0.6736360310429916,
      "grad_norm": 3.4469106197357178,
      "learning_rate": 3.4355385850772684e-07,
      "loss": 0.7474,
      "step": 43270
    },
    {
      "epoch": 0.6737138720449609,
      "grad_norm": 3.5301003456115723,
      "learning_rate": 3.4347191950312186e-07,
      "loss": 0.7348,
      "step": 43275
    },
    {
      "epoch": 0.6737917130469303,
      "grad_norm": 3.631756544113159,
      "learning_rate": 3.433899804985169e-07,
      "loss": 0.7588,
      "step": 43280
    },
    {
      "epoch": 0.6738695540488997,
      "grad_norm": 2.4092540740966797,
      "learning_rate": 3.433080414939119e-07,
      "loss": 0.7082,
      "step": 43285
    },
    {
      "epoch": 0.6739473950508691,
      "grad_norm": 4.1742095947265625,
      "learning_rate": 3.4322610248930693e-07,
      "loss": 0.6558,
      "step": 43290
    },
    {
      "epoch": 0.6740252360528385,
      "grad_norm": 4.940985202789307,
      "learning_rate": 3.43144163484702e-07,
      "loss": 0.7471,
      "step": 43295
    },
    {
      "epoch": 0.6741030770548079,
      "grad_norm": 2.777864694595337,
      "learning_rate": 3.43062224480097e-07,
      "loss": 0.7135,
      "step": 43300
    },
    {
      "epoch": 0.6741809180567773,
      "grad_norm": 3.493037223815918,
      "learning_rate": 3.42980285475492e-07,
      "loss": 0.6375,
      "step": 43305
    },
    {
      "epoch": 0.6742587590587467,
      "grad_norm": 4.442993640899658,
      "learning_rate": 3.4289834647088707e-07,
      "loss": 0.7556,
      "step": 43310
    },
    {
      "epoch": 0.6743366000607159,
      "grad_norm": 5.956271648406982,
      "learning_rate": 3.428164074662821e-07,
      "loss": 0.7207,
      "step": 43315
    },
    {
      "epoch": 0.6744144410626853,
      "grad_norm": 4.378262996673584,
      "learning_rate": 3.427344684616771e-07,
      "loss": 0.8042,
      "step": 43320
    },
    {
      "epoch": 0.6744922820646547,
      "grad_norm": 4.4390153884887695,
      "learning_rate": 3.4265252945707214e-07,
      "loss": 0.673,
      "step": 43325
    },
    {
      "epoch": 0.6745701230666241,
      "grad_norm": 3.624469757080078,
      "learning_rate": 3.4257059045246716e-07,
      "loss": 0.6263,
      "step": 43330
    },
    {
      "epoch": 0.6746479640685935,
      "grad_norm": 7.120113372802734,
      "learning_rate": 3.4248865144786223e-07,
      "loss": 0.706,
      "step": 43335
    },
    {
      "epoch": 0.6747258050705629,
      "grad_norm": 3.8985066413879395,
      "learning_rate": 3.4240671244325726e-07,
      "loss": 0.6928,
      "step": 43340
    },
    {
      "epoch": 0.6748036460725323,
      "grad_norm": 4.036379337310791,
      "learning_rate": 3.423247734386522e-07,
      "loss": 0.8131,
      "step": 43345
    },
    {
      "epoch": 0.6748814870745016,
      "grad_norm": 4.213803291320801,
      "learning_rate": 3.422428344340473e-07,
      "loss": 0.8095,
      "step": 43350
    },
    {
      "epoch": 0.674959328076471,
      "grad_norm": 4.406517505645752,
      "learning_rate": 3.421608954294423e-07,
      "loss": 0.7937,
      "step": 43355
    },
    {
      "epoch": 0.6750371690784404,
      "grad_norm": 3.3817994594573975,
      "learning_rate": 3.4207895642483734e-07,
      "loss": 0.714,
      "step": 43360
    },
    {
      "epoch": 0.6751150100804097,
      "grad_norm": 3.5341508388519287,
      "learning_rate": 3.4199701742023237e-07,
      "loss": 0.6855,
      "step": 43365
    },
    {
      "epoch": 0.6751928510823791,
      "grad_norm": 2.9421820640563965,
      "learning_rate": 3.419150784156274e-07,
      "loss": 0.6858,
      "step": 43370
    },
    {
      "epoch": 0.6752706920843485,
      "grad_norm": 4.007630825042725,
      "learning_rate": 3.418331394110224e-07,
      "loss": 0.7431,
      "step": 43375
    },
    {
      "epoch": 0.6753485330863179,
      "grad_norm": 3.779986619949341,
      "learning_rate": 3.417512004064175e-07,
      "loss": 0.817,
      "step": 43380
    },
    {
      "epoch": 0.6754263740882873,
      "grad_norm": 4.010073184967041,
      "learning_rate": 3.4166926140181245e-07,
      "loss": 0.6826,
      "step": 43385
    },
    {
      "epoch": 0.6755042150902566,
      "grad_norm": 5.103675365447998,
      "learning_rate": 3.415873223972075e-07,
      "loss": 0.8209,
      "step": 43390
    },
    {
      "epoch": 0.675582056092226,
      "grad_norm": 6.922098636627197,
      "learning_rate": 3.4150538339260255e-07,
      "loss": 0.7581,
      "step": 43395
    },
    {
      "epoch": 0.6756598970941954,
      "grad_norm": 13.275751113891602,
      "learning_rate": 3.4142344438799757e-07,
      "loss": 0.7829,
      "step": 43400
    },
    {
      "epoch": 0.6757377380961648,
      "grad_norm": 2.8118398189544678,
      "learning_rate": 3.413415053833926e-07,
      "loss": 0.8062,
      "step": 43405
    },
    {
      "epoch": 0.6758155790981342,
      "grad_norm": 4.528495788574219,
      "learning_rate": 3.412595663787876e-07,
      "loss": 0.7837,
      "step": 43410
    },
    {
      "epoch": 0.6758934201001036,
      "grad_norm": 3.620072364807129,
      "learning_rate": 3.4117762737418264e-07,
      "loss": 0.842,
      "step": 43415
    },
    {
      "epoch": 0.675971261102073,
      "grad_norm": 2.6222105026245117,
      "learning_rate": 3.410956883695777e-07,
      "loss": 0.6687,
      "step": 43420
    },
    {
      "epoch": 0.6760491021040422,
      "grad_norm": 4.36367130279541,
      "learning_rate": 3.410137493649727e-07,
      "loss": 0.7233,
      "step": 43425
    },
    {
      "epoch": 0.6761269431060116,
      "grad_norm": 3.889292001724243,
      "learning_rate": 3.409318103603677e-07,
      "loss": 0.7567,
      "step": 43430
    },
    {
      "epoch": 0.676204784107981,
      "grad_norm": 5.09576416015625,
      "learning_rate": 3.408498713557628e-07,
      "loss": 0.7562,
      "step": 43435
    },
    {
      "epoch": 0.6762826251099504,
      "grad_norm": 5.368563175201416,
      "learning_rate": 3.407679323511578e-07,
      "loss": 0.8207,
      "step": 43440
    },
    {
      "epoch": 0.6763604661119198,
      "grad_norm": 7.179274082183838,
      "learning_rate": 3.4068599334655277e-07,
      "loss": 0.7546,
      "step": 43445
    },
    {
      "epoch": 0.6764383071138892,
      "grad_norm": 8.383667945861816,
      "learning_rate": 3.4060405434194784e-07,
      "loss": 0.7146,
      "step": 43450
    },
    {
      "epoch": 0.6765161481158586,
      "grad_norm": 6.962827205657959,
      "learning_rate": 3.4052211533734287e-07,
      "loss": 0.8567,
      "step": 43455
    },
    {
      "epoch": 0.676593989117828,
      "grad_norm": 5.001617908477783,
      "learning_rate": 3.4044017633273794e-07,
      "loss": 0.7063,
      "step": 43460
    },
    {
      "epoch": 0.6766718301197973,
      "grad_norm": 6.039570331573486,
      "learning_rate": 3.403582373281329e-07,
      "loss": 0.7491,
      "step": 43465
    },
    {
      "epoch": 0.6767496711217666,
      "grad_norm": 3.757479190826416,
      "learning_rate": 3.4027629832352793e-07,
      "loss": 0.7221,
      "step": 43470
    },
    {
      "epoch": 0.676827512123736,
      "grad_norm": 3.7640066146850586,
      "learning_rate": 3.40194359318923e-07,
      "loss": 0.7579,
      "step": 43475
    },
    {
      "epoch": 0.6769053531257054,
      "grad_norm": 3.464946985244751,
      "learning_rate": 3.4011242031431803e-07,
      "loss": 0.641,
      "step": 43480
    },
    {
      "epoch": 0.6769831941276748,
      "grad_norm": 4.646814823150635,
      "learning_rate": 3.40030481309713e-07,
      "loss": 0.6759,
      "step": 43485
    },
    {
      "epoch": 0.6770610351296442,
      "grad_norm": 7.891297817230225,
      "learning_rate": 3.3994854230510807e-07,
      "loss": 0.8283,
      "step": 43490
    },
    {
      "epoch": 0.6771388761316136,
      "grad_norm": 3.9876599311828613,
      "learning_rate": 3.398666033005031e-07,
      "loss": 0.6986,
      "step": 43495
    },
    {
      "epoch": 0.6772167171335829,
      "grad_norm": 4.609341621398926,
      "learning_rate": 3.397846642958981e-07,
      "loss": 0.6419,
      "step": 43500
    },
    {
      "epoch": 0.6772945581355523,
      "grad_norm": 5.634749412536621,
      "learning_rate": 3.3970272529129314e-07,
      "loss": 0.8131,
      "step": 43505
    },
    {
      "epoch": 0.6773723991375217,
      "grad_norm": 7.264232635498047,
      "learning_rate": 3.3962078628668816e-07,
      "loss": 0.6808,
      "step": 43510
    },
    {
      "epoch": 0.6774502401394911,
      "grad_norm": 4.910491943359375,
      "learning_rate": 3.395388472820832e-07,
      "loss": 0.8618,
      "step": 43515
    },
    {
      "epoch": 0.6775280811414605,
      "grad_norm": 3.4227261543273926,
      "learning_rate": 3.3945690827747826e-07,
      "loss": 0.6971,
      "step": 43520
    },
    {
      "epoch": 0.6776059221434299,
      "grad_norm": 4.070761203765869,
      "learning_rate": 3.393749692728732e-07,
      "loss": 0.6912,
      "step": 43525
    },
    {
      "epoch": 0.6776837631453992,
      "grad_norm": 3.948561191558838,
      "learning_rate": 3.392930302682683e-07,
      "loss": 0.7503,
      "step": 43530
    },
    {
      "epoch": 0.6777616041473686,
      "grad_norm": 3.0104687213897705,
      "learning_rate": 3.392110912636633e-07,
      "loss": 0.6472,
      "step": 43535
    },
    {
      "epoch": 0.6778394451493379,
      "grad_norm": 5.071345329284668,
      "learning_rate": 3.3912915225905834e-07,
      "loss": 0.7515,
      "step": 43540
    },
    {
      "epoch": 0.6779172861513073,
      "grad_norm": 3.497471332550049,
      "learning_rate": 3.3904721325445337e-07,
      "loss": 0.7017,
      "step": 43545
    },
    {
      "epoch": 0.6779951271532767,
      "grad_norm": 2.9580917358398438,
      "learning_rate": 3.389652742498484e-07,
      "loss": 0.7626,
      "step": 43550
    },
    {
      "epoch": 0.6780729681552461,
      "grad_norm": 2.5114800930023193,
      "learning_rate": 3.388833352452434e-07,
      "loss": 0.7189,
      "step": 43555
    },
    {
      "epoch": 0.6781508091572155,
      "grad_norm": 2.9231464862823486,
      "learning_rate": 3.388013962406385e-07,
      "loss": 0.8666,
      "step": 43560
    },
    {
      "epoch": 0.6782286501591849,
      "grad_norm": 5.19349479675293,
      "learning_rate": 3.3871945723603345e-07,
      "loss": 0.8265,
      "step": 43565
    },
    {
      "epoch": 0.6783064911611543,
      "grad_norm": 4.220258712768555,
      "learning_rate": 3.386375182314285e-07,
      "loss": 0.7265,
      "step": 43570
    },
    {
      "epoch": 0.6783843321631235,
      "grad_norm": 2.9845762252807617,
      "learning_rate": 3.3855557922682355e-07,
      "loss": 0.7085,
      "step": 43575
    },
    {
      "epoch": 0.6784621731650929,
      "grad_norm": 3.5858893394470215,
      "learning_rate": 3.3847364022221857e-07,
      "loss": 0.5985,
      "step": 43580
    },
    {
      "epoch": 0.6785400141670623,
      "grad_norm": 3.960005044937134,
      "learning_rate": 3.3839170121761354e-07,
      "loss": 0.757,
      "step": 43585
    },
    {
      "epoch": 0.6786178551690317,
      "grad_norm": 3.728795051574707,
      "learning_rate": 3.383097622130086e-07,
      "loss": 0.7574,
      "step": 43590
    },
    {
      "epoch": 0.6786956961710011,
      "grad_norm": 3.065697431564331,
      "learning_rate": 3.3822782320840364e-07,
      "loss": 0.704,
      "step": 43595
    },
    {
      "epoch": 0.6787735371729705,
      "grad_norm": 6.750757694244385,
      "learning_rate": 3.381458842037987e-07,
      "loss": 0.7352,
      "step": 43600
    },
    {
      "epoch": 0.6788513781749399,
      "grad_norm": 3.30930757522583,
      "learning_rate": 3.380639451991937e-07,
      "loss": 0.672,
      "step": 43605
    },
    {
      "epoch": 0.6789292191769093,
      "grad_norm": 5.908539295196533,
      "learning_rate": 3.379820061945887e-07,
      "loss": 0.5568,
      "step": 43610
    },
    {
      "epoch": 0.6790070601788786,
      "grad_norm": 3.3994572162628174,
      "learning_rate": 3.379000671899838e-07,
      "loss": 0.8873,
      "step": 43615
    },
    {
      "epoch": 0.679084901180848,
      "grad_norm": 3.162890672683716,
      "learning_rate": 3.378181281853788e-07,
      "loss": 0.8238,
      "step": 43620
    },
    {
      "epoch": 0.6791627421828174,
      "grad_norm": 3.2565619945526123,
      "learning_rate": 3.3773618918077377e-07,
      "loss": 0.8365,
      "step": 43625
    },
    {
      "epoch": 0.6792405831847868,
      "grad_norm": 7.477787494659424,
      "learning_rate": 3.3765425017616884e-07,
      "loss": 0.7684,
      "step": 43630
    },
    {
      "epoch": 0.6793184241867561,
      "grad_norm": 5.025554180145264,
      "learning_rate": 3.3757231117156387e-07,
      "loss": 0.7387,
      "step": 43635
    },
    {
      "epoch": 0.6793962651887255,
      "grad_norm": 6.024256706237793,
      "learning_rate": 3.374903721669589e-07,
      "loss": 0.6729,
      "step": 43640
    },
    {
      "epoch": 0.6794741061906949,
      "grad_norm": 2.858487367630005,
      "learning_rate": 3.3740843316235396e-07,
      "loss": 0.7972,
      "step": 43645
    },
    {
      "epoch": 0.6795519471926643,
      "grad_norm": 3.688282012939453,
      "learning_rate": 3.3732649415774893e-07,
      "loss": 0.7965,
      "step": 43650
    },
    {
      "epoch": 0.6796297881946336,
      "grad_norm": 3.1246793270111084,
      "learning_rate": 3.37244555153144e-07,
      "loss": 0.7615,
      "step": 43655
    },
    {
      "epoch": 0.679707629196603,
      "grad_norm": 3.362187385559082,
      "learning_rate": 3.3716261614853903e-07,
      "loss": 0.7027,
      "step": 43660
    },
    {
      "epoch": 0.6797854701985724,
      "grad_norm": 3.206632375717163,
      "learning_rate": 3.3708067714393405e-07,
      "loss": 0.6452,
      "step": 43665
    },
    {
      "epoch": 0.6798633112005418,
      "grad_norm": 3.090350866317749,
      "learning_rate": 3.3699873813932907e-07,
      "loss": 0.7642,
      "step": 43670
    },
    {
      "epoch": 0.6799411522025112,
      "grad_norm": 3.5976877212524414,
      "learning_rate": 3.369167991347241e-07,
      "loss": 0.7326,
      "step": 43675
    },
    {
      "epoch": 0.6800189932044806,
      "grad_norm": 2.956827163696289,
      "learning_rate": 3.368348601301191e-07,
      "loss": 0.6705,
      "step": 43680
    },
    {
      "epoch": 0.68009683420645,
      "grad_norm": 3.540605306625366,
      "learning_rate": 3.367529211255142e-07,
      "loss": 0.7074,
      "step": 43685
    },
    {
      "epoch": 0.6801746752084192,
      "grad_norm": 4.437077045440674,
      "learning_rate": 3.3667098212090916e-07,
      "loss": 0.6729,
      "step": 43690
    },
    {
      "epoch": 0.6802525162103886,
      "grad_norm": 3.245274305343628,
      "learning_rate": 3.365890431163042e-07,
      "loss": 0.7854,
      "step": 43695
    },
    {
      "epoch": 0.680330357212358,
      "grad_norm": 8.934989929199219,
      "learning_rate": 3.3650710411169926e-07,
      "loss": 0.7487,
      "step": 43700
    },
    {
      "epoch": 0.6804081982143274,
      "grad_norm": 5.963617324829102,
      "learning_rate": 3.364251651070943e-07,
      "loss": 0.7069,
      "step": 43705
    },
    {
      "epoch": 0.6804860392162968,
      "grad_norm": 4.199155807495117,
      "learning_rate": 3.3634322610248925e-07,
      "loss": 0.6427,
      "step": 43710
    },
    {
      "epoch": 0.6805638802182662,
      "grad_norm": 2.8385355472564697,
      "learning_rate": 3.362612870978843e-07,
      "loss": 0.6955,
      "step": 43715
    },
    {
      "epoch": 0.6806417212202356,
      "grad_norm": 3.1126134395599365,
      "learning_rate": 3.3617934809327935e-07,
      "loss": 0.821,
      "step": 43720
    },
    {
      "epoch": 0.680719562222205,
      "grad_norm": 4.306998252868652,
      "learning_rate": 3.360974090886744e-07,
      "loss": 0.6368,
      "step": 43725
    },
    {
      "epoch": 0.6807974032241743,
      "grad_norm": 4.928831100463867,
      "learning_rate": 3.360154700840694e-07,
      "loss": 0.8052,
      "step": 43730
    },
    {
      "epoch": 0.6808752442261436,
      "grad_norm": 4.440091133117676,
      "learning_rate": 3.359335310794644e-07,
      "loss": 0.7625,
      "step": 43735
    },
    {
      "epoch": 0.680953085228113,
      "grad_norm": 4.38109016418457,
      "learning_rate": 3.358515920748595e-07,
      "loss": 0.7292,
      "step": 43740
    },
    {
      "epoch": 0.6810309262300824,
      "grad_norm": 3.510606527328491,
      "learning_rate": 3.357696530702545e-07,
      "loss": 0.6737,
      "step": 43745
    },
    {
      "epoch": 0.6811087672320518,
      "grad_norm": 3.4536354541778564,
      "learning_rate": 3.356877140656495e-07,
      "loss": 0.6703,
      "step": 43750
    },
    {
      "epoch": 0.6811866082340212,
      "grad_norm": 8.014554977416992,
      "learning_rate": 3.3560577506104455e-07,
      "loss": 0.7386,
      "step": 43755
    },
    {
      "epoch": 0.6812644492359906,
      "grad_norm": 3.4720659255981445,
      "learning_rate": 3.3552383605643957e-07,
      "loss": 0.692,
      "step": 43760
    },
    {
      "epoch": 0.6813422902379599,
      "grad_norm": 4.725083351135254,
      "learning_rate": 3.354418970518346e-07,
      "loss": 0.6798,
      "step": 43765
    },
    {
      "epoch": 0.6814201312399293,
      "grad_norm": 3.7027039527893066,
      "learning_rate": 3.353599580472296e-07,
      "loss": 0.8023,
      "step": 43770
    },
    {
      "epoch": 0.6814979722418987,
      "grad_norm": 5.262540817260742,
      "learning_rate": 3.3527801904262464e-07,
      "loss": 0.7445,
      "step": 43775
    },
    {
      "epoch": 0.6815758132438681,
      "grad_norm": 4.353088855743408,
      "learning_rate": 3.351960800380197e-07,
      "loss": 0.7551,
      "step": 43780
    },
    {
      "epoch": 0.6816536542458375,
      "grad_norm": 4.837805271148682,
      "learning_rate": 3.3511414103341474e-07,
      "loss": 0.8032,
      "step": 43785
    },
    {
      "epoch": 0.6817314952478069,
      "grad_norm": 5.764411926269531,
      "learning_rate": 3.350322020288097e-07,
      "loss": 0.6887,
      "step": 43790
    },
    {
      "epoch": 0.6818093362497762,
      "grad_norm": 4.478251934051514,
      "learning_rate": 3.349502630242048e-07,
      "loss": 0.7362,
      "step": 43795
    },
    {
      "epoch": 0.6818871772517456,
      "grad_norm": 4.349662780761719,
      "learning_rate": 3.348683240195998e-07,
      "loss": 0.7636,
      "step": 43800
    },
    {
      "epoch": 0.6819650182537149,
      "grad_norm": 3.5404627323150635,
      "learning_rate": 3.347863850149948e-07,
      "loss": 0.8073,
      "step": 43805
    },
    {
      "epoch": 0.6820428592556843,
      "grad_norm": 5.5866193771362305,
      "learning_rate": 3.3470444601038985e-07,
      "loss": 0.6987,
      "step": 43810
    },
    {
      "epoch": 0.6821207002576537,
      "grad_norm": 3.5849039554595947,
      "learning_rate": 3.3462250700578487e-07,
      "loss": 0.7434,
      "step": 43815
    },
    {
      "epoch": 0.6821985412596231,
      "grad_norm": 2.674553394317627,
      "learning_rate": 3.345405680011799e-07,
      "loss": 0.8175,
      "step": 43820
    },
    {
      "epoch": 0.6822763822615925,
      "grad_norm": 7.1580424308776855,
      "learning_rate": 3.3445862899657496e-07,
      "loss": 0.7435,
      "step": 43825
    },
    {
      "epoch": 0.6823542232635619,
      "grad_norm": 3.3526079654693604,
      "learning_rate": 3.3437668999196993e-07,
      "loss": 0.7863,
      "step": 43830
    },
    {
      "epoch": 0.6824320642655313,
      "grad_norm": 3.2791218757629395,
      "learning_rate": 3.34294750987365e-07,
      "loss": 0.7361,
      "step": 43835
    },
    {
      "epoch": 0.6825099052675005,
      "grad_norm": 3.501462697982788,
      "learning_rate": 3.3421281198276003e-07,
      "loss": 0.6947,
      "step": 43840
    },
    {
      "epoch": 0.6825877462694699,
      "grad_norm": 3.1613030433654785,
      "learning_rate": 3.3413087297815505e-07,
      "loss": 0.76,
      "step": 43845
    },
    {
      "epoch": 0.6826655872714393,
      "grad_norm": 3.8929340839385986,
      "learning_rate": 3.340489339735501e-07,
      "loss": 0.6943,
      "step": 43850
    },
    {
      "epoch": 0.6827434282734087,
      "grad_norm": 6.546085834503174,
      "learning_rate": 3.339669949689451e-07,
      "loss": 0.7168,
      "step": 43855
    },
    {
      "epoch": 0.6828212692753781,
      "grad_norm": 3.1008520126342773,
      "learning_rate": 3.338850559643401e-07,
      "loss": 0.8062,
      "step": 43860
    },
    {
      "epoch": 0.6828991102773475,
      "grad_norm": 2.896251678466797,
      "learning_rate": 3.338031169597352e-07,
      "loss": 0.5897,
      "step": 43865
    },
    {
      "epoch": 0.6829769512793169,
      "grad_norm": 3.2136237621307373,
      "learning_rate": 3.3372117795513016e-07,
      "loss": 0.7186,
      "step": 43870
    },
    {
      "epoch": 0.6830547922812863,
      "grad_norm": 5.651174068450928,
      "learning_rate": 3.336392389505252e-07,
      "loss": 0.7582,
      "step": 43875
    },
    {
      "epoch": 0.6831326332832556,
      "grad_norm": 3.195446014404297,
      "learning_rate": 3.3355729994592026e-07,
      "loss": 0.7101,
      "step": 43880
    },
    {
      "epoch": 0.683210474285225,
      "grad_norm": 3.22259521484375,
      "learning_rate": 3.334753609413153e-07,
      "loss": 0.7423,
      "step": 43885
    },
    {
      "epoch": 0.6832883152871944,
      "grad_norm": 4.570167064666748,
      "learning_rate": 3.3339342193671025e-07,
      "loss": 0.7431,
      "step": 43890
    },
    {
      "epoch": 0.6833661562891638,
      "grad_norm": 7.377659797668457,
      "learning_rate": 3.333114829321053e-07,
      "loss": 0.7747,
      "step": 43895
    },
    {
      "epoch": 0.6834439972911331,
      "grad_norm": 3.4306681156158447,
      "learning_rate": 3.3322954392750035e-07,
      "loss": 0.7873,
      "step": 43900
    },
    {
      "epoch": 0.6835218382931025,
      "grad_norm": 4.068570613861084,
      "learning_rate": 3.331476049228954e-07,
      "loss": 0.7691,
      "step": 43905
    },
    {
      "epoch": 0.6835996792950719,
      "grad_norm": 4.018854141235352,
      "learning_rate": 3.330656659182904e-07,
      "loss": 0.7752,
      "step": 43910
    },
    {
      "epoch": 0.6836775202970413,
      "grad_norm": 3.217566728591919,
      "learning_rate": 3.329837269136854e-07,
      "loss": 0.6996,
      "step": 43915
    },
    {
      "epoch": 0.6837553612990106,
      "grad_norm": 3.3054680824279785,
      "learning_rate": 3.329017879090805e-07,
      "loss": 0.7233,
      "step": 43920
    },
    {
      "epoch": 0.68383320230098,
      "grad_norm": 6.181046009063721,
      "learning_rate": 3.328198489044755e-07,
      "loss": 0.7444,
      "step": 43925
    },
    {
      "epoch": 0.6839110433029494,
      "grad_norm": 3.50191068649292,
      "learning_rate": 3.327379098998705e-07,
      "loss": 0.7106,
      "step": 43930
    },
    {
      "epoch": 0.6839888843049188,
      "grad_norm": 2.750953197479248,
      "learning_rate": 3.3265597089526555e-07,
      "loss": 0.7379,
      "step": 43935
    },
    {
      "epoch": 0.6840667253068882,
      "grad_norm": 3.6753265857696533,
      "learning_rate": 3.325740318906606e-07,
      "loss": 0.7271,
      "step": 43940
    },
    {
      "epoch": 0.6841445663088576,
      "grad_norm": 4.053527355194092,
      "learning_rate": 3.324920928860556e-07,
      "loss": 0.7656,
      "step": 43945
    },
    {
      "epoch": 0.684222407310827,
      "grad_norm": 3.114941120147705,
      "learning_rate": 3.324101538814506e-07,
      "loss": 0.7117,
      "step": 43950
    },
    {
      "epoch": 0.6843002483127962,
      "grad_norm": 4.088073253631592,
      "learning_rate": 3.3232821487684564e-07,
      "loss": 0.7587,
      "step": 43955
    },
    {
      "epoch": 0.6843780893147656,
      "grad_norm": 3.449873447418213,
      "learning_rate": 3.322462758722407e-07,
      "loss": 0.7561,
      "step": 43960
    },
    {
      "epoch": 0.684455930316735,
      "grad_norm": 5.354700088500977,
      "learning_rate": 3.3216433686763574e-07,
      "loss": 0.7336,
      "step": 43965
    },
    {
      "epoch": 0.6845337713187044,
      "grad_norm": 3.158113956451416,
      "learning_rate": 3.3208239786303076e-07,
      "loss": 0.6063,
      "step": 43970
    },
    {
      "epoch": 0.6846116123206738,
      "grad_norm": 3.251915216445923,
      "learning_rate": 3.320004588584258e-07,
      "loss": 0.7067,
      "step": 43975
    },
    {
      "epoch": 0.6846894533226432,
      "grad_norm": 4.2592034339904785,
      "learning_rate": 3.319185198538208e-07,
      "loss": 0.7252,
      "step": 43980
    },
    {
      "epoch": 0.6847672943246126,
      "grad_norm": 4.470844268798828,
      "learning_rate": 3.318365808492158e-07,
      "loss": 0.6236,
      "step": 43985
    },
    {
      "epoch": 0.684845135326582,
      "grad_norm": 2.9213521480560303,
      "learning_rate": 3.317546418446109e-07,
      "loss": 0.7352,
      "step": 43990
    },
    {
      "epoch": 0.6849229763285513,
      "grad_norm": 4.676266193389893,
      "learning_rate": 3.3167270284000587e-07,
      "loss": 0.7444,
      "step": 43995
    },
    {
      "epoch": 0.6850008173305207,
      "grad_norm": 12.755716323852539,
      "learning_rate": 3.315907638354009e-07,
      "loss": 0.6761,
      "step": 44000
    },
    {
      "epoch": 0.68507865833249,
      "grad_norm": 3.301211357116699,
      "learning_rate": 3.3150882483079597e-07,
      "loss": 0.777,
      "step": 44005
    },
    {
      "epoch": 0.6851564993344594,
      "grad_norm": 3.6110363006591797,
      "learning_rate": 3.31426885826191e-07,
      "loss": 0.7253,
      "step": 44010
    },
    {
      "epoch": 0.6852343403364288,
      "grad_norm": 4.971611976623535,
      "learning_rate": 3.3134494682158596e-07,
      "loss": 0.5742,
      "step": 44015
    },
    {
      "epoch": 0.6853121813383982,
      "grad_norm": 3.530388355255127,
      "learning_rate": 3.3126300781698103e-07,
      "loss": 0.7254,
      "step": 44020
    },
    {
      "epoch": 0.6853900223403676,
      "grad_norm": 3.1901206970214844,
      "learning_rate": 3.3118106881237605e-07,
      "loss": 0.6795,
      "step": 44025
    },
    {
      "epoch": 0.6854678633423369,
      "grad_norm": 5.3250203132629395,
      "learning_rate": 3.3109912980777113e-07,
      "loss": 0.8157,
      "step": 44030
    },
    {
      "epoch": 0.6855457043443063,
      "grad_norm": 5.427084922790527,
      "learning_rate": 3.310171908031661e-07,
      "loss": 0.7336,
      "step": 44035
    },
    {
      "epoch": 0.6856235453462757,
      "grad_norm": 3.3820009231567383,
      "learning_rate": 3.309352517985611e-07,
      "loss": 0.6994,
      "step": 44040
    },
    {
      "epoch": 0.6857013863482451,
      "grad_norm": 3.944084405899048,
      "learning_rate": 3.308533127939562e-07,
      "loss": 0.8101,
      "step": 44045
    },
    {
      "epoch": 0.6857792273502145,
      "grad_norm": 3.2355542182922363,
      "learning_rate": 3.307713737893512e-07,
      "loss": 0.7289,
      "step": 44050
    },
    {
      "epoch": 0.6858570683521839,
      "grad_norm": 2.469600200653076,
      "learning_rate": 3.306894347847462e-07,
      "loss": 0.6959,
      "step": 44055
    },
    {
      "epoch": 0.6859349093541532,
      "grad_norm": 3.2230308055877686,
      "learning_rate": 3.3060749578014126e-07,
      "loss": 0.8022,
      "step": 44060
    },
    {
      "epoch": 0.6860127503561226,
      "grad_norm": 4.760956287384033,
      "learning_rate": 3.305255567755363e-07,
      "loss": 0.8172,
      "step": 44065
    },
    {
      "epoch": 0.6860905913580919,
      "grad_norm": 2.544679880142212,
      "learning_rate": 3.304436177709313e-07,
      "loss": 0.708,
      "step": 44070
    },
    {
      "epoch": 0.6861684323600613,
      "grad_norm": 2.8041954040527344,
      "learning_rate": 3.303616787663263e-07,
      "loss": 0.8421,
      "step": 44075
    },
    {
      "epoch": 0.6862462733620307,
      "grad_norm": 4.500549793243408,
      "learning_rate": 3.3027973976172135e-07,
      "loss": 0.6269,
      "step": 44080
    },
    {
      "epoch": 0.6863241143640001,
      "grad_norm": 6.616947650909424,
      "learning_rate": 3.301978007571164e-07,
      "loss": 0.7079,
      "step": 44085
    },
    {
      "epoch": 0.6864019553659695,
      "grad_norm": 3.3033759593963623,
      "learning_rate": 3.3011586175251144e-07,
      "loss": 0.7385,
      "step": 44090
    },
    {
      "epoch": 0.6864797963679389,
      "grad_norm": 8.398449897766113,
      "learning_rate": 3.300339227479064e-07,
      "loss": 0.8163,
      "step": 44095
    },
    {
      "epoch": 0.6865576373699083,
      "grad_norm": 3.5369644165039062,
      "learning_rate": 3.299519837433015e-07,
      "loss": 0.6331,
      "step": 44100
    },
    {
      "epoch": 0.6866354783718776,
      "grad_norm": 2.9129951000213623,
      "learning_rate": 3.298700447386965e-07,
      "loss": 0.8102,
      "step": 44105
    },
    {
      "epoch": 0.6867133193738469,
      "grad_norm": 5.29641580581665,
      "learning_rate": 3.2978810573409153e-07,
      "loss": 0.8412,
      "step": 44110
    },
    {
      "epoch": 0.6867911603758163,
      "grad_norm": 3.8847150802612305,
      "learning_rate": 3.2970616672948655e-07,
      "loss": 0.8218,
      "step": 44115
    },
    {
      "epoch": 0.6868690013777857,
      "grad_norm": 5.402711868286133,
      "learning_rate": 3.296242277248816e-07,
      "loss": 0.8029,
      "step": 44120
    },
    {
      "epoch": 0.6869468423797551,
      "grad_norm": 4.635100364685059,
      "learning_rate": 3.295422887202766e-07,
      "loss": 0.8125,
      "step": 44125
    },
    {
      "epoch": 0.6870246833817245,
      "grad_norm": 3.537489414215088,
      "learning_rate": 3.2946034971567167e-07,
      "loss": 0.8211,
      "step": 44130
    },
    {
      "epoch": 0.6871025243836939,
      "grad_norm": 5.8473687171936035,
      "learning_rate": 3.2937841071106664e-07,
      "loss": 0.7178,
      "step": 44135
    },
    {
      "epoch": 0.6871803653856633,
      "grad_norm": 4.851031303405762,
      "learning_rate": 3.2929647170646166e-07,
      "loss": 0.8057,
      "step": 44140
    },
    {
      "epoch": 0.6872582063876326,
      "grad_norm": 2.883733034133911,
      "learning_rate": 3.2921453270185674e-07,
      "loss": 0.732,
      "step": 44145
    },
    {
      "epoch": 0.687336047389602,
      "grad_norm": 3.854414701461792,
      "learning_rate": 3.2913259369725176e-07,
      "loss": 0.7536,
      "step": 44150
    },
    {
      "epoch": 0.6874138883915714,
      "grad_norm": 3.727436065673828,
      "learning_rate": 3.290506546926468e-07,
      "loss": 0.7661,
      "step": 44155
    },
    {
      "epoch": 0.6874917293935408,
      "grad_norm": 4.10051155090332,
      "learning_rate": 3.289687156880418e-07,
      "loss": 0.8163,
      "step": 44160
    },
    {
      "epoch": 0.6875695703955101,
      "grad_norm": 3.572967529296875,
      "learning_rate": 3.288867766834368e-07,
      "loss": 0.8412,
      "step": 44165
    },
    {
      "epoch": 0.6876474113974795,
      "grad_norm": 2.259187698364258,
      "learning_rate": 3.288048376788319e-07,
      "loss": 0.688,
      "step": 44170
    },
    {
      "epoch": 0.6877252523994489,
      "grad_norm": 4.5084733963012695,
      "learning_rate": 3.2872289867422687e-07,
      "loss": 0.7689,
      "step": 44175
    },
    {
      "epoch": 0.6878030934014182,
      "grad_norm": 9.757917404174805,
      "learning_rate": 3.286409596696219e-07,
      "loss": 0.7128,
      "step": 44180
    },
    {
      "epoch": 0.6878809344033876,
      "grad_norm": 2.805596351623535,
      "learning_rate": 3.2855902066501697e-07,
      "loss": 0.7558,
      "step": 44185
    },
    {
      "epoch": 0.687958775405357,
      "grad_norm": 7.814718723297119,
      "learning_rate": 3.28477081660412e-07,
      "loss": 0.6714,
      "step": 44190
    },
    {
      "epoch": 0.6880366164073264,
      "grad_norm": 4.5613484382629395,
      "learning_rate": 3.2839514265580696e-07,
      "loss": 0.6384,
      "step": 44195
    },
    {
      "epoch": 0.6881144574092958,
      "grad_norm": 5.466517925262451,
      "learning_rate": 3.2831320365120203e-07,
      "loss": 0.7607,
      "step": 44200
    },
    {
      "epoch": 0.6881922984112652,
      "grad_norm": 3.4753341674804688,
      "learning_rate": 3.2823126464659705e-07,
      "loss": 0.6395,
      "step": 44205
    },
    {
      "epoch": 0.6882701394132346,
      "grad_norm": 3.1602120399475098,
      "learning_rate": 3.2814932564199213e-07,
      "loss": 0.7635,
      "step": 44210
    },
    {
      "epoch": 0.688347980415204,
      "grad_norm": 4.730556011199951,
      "learning_rate": 3.280673866373871e-07,
      "loss": 0.6351,
      "step": 44215
    },
    {
      "epoch": 0.6884258214171732,
      "grad_norm": 3.5707106590270996,
      "learning_rate": 3.279854476327821e-07,
      "loss": 0.7174,
      "step": 44220
    },
    {
      "epoch": 0.6885036624191426,
      "grad_norm": 4.608479022979736,
      "learning_rate": 3.279035086281772e-07,
      "loss": 0.7786,
      "step": 44225
    },
    {
      "epoch": 0.688581503421112,
      "grad_norm": 3.546304702758789,
      "learning_rate": 3.278215696235722e-07,
      "loss": 0.7409,
      "step": 44230
    },
    {
      "epoch": 0.6886593444230814,
      "grad_norm": 10.189797401428223,
      "learning_rate": 3.277396306189672e-07,
      "loss": 0.7284,
      "step": 44235
    },
    {
      "epoch": 0.6887371854250508,
      "grad_norm": 3.996938467025757,
      "learning_rate": 3.2765769161436226e-07,
      "loss": 0.7202,
      "step": 44240
    },
    {
      "epoch": 0.6888150264270202,
      "grad_norm": 9.441819190979004,
      "learning_rate": 3.275757526097573e-07,
      "loss": 0.6538,
      "step": 44245
    },
    {
      "epoch": 0.6888928674289896,
      "grad_norm": 3.9540085792541504,
      "learning_rate": 3.274938136051523e-07,
      "loss": 0.7271,
      "step": 44250
    },
    {
      "epoch": 0.688970708430959,
      "grad_norm": 8.33227825164795,
      "learning_rate": 3.274118746005473e-07,
      "loss": 0.6645,
      "step": 44255
    },
    {
      "epoch": 0.6890485494329283,
      "grad_norm": 2.4066879749298096,
      "learning_rate": 3.2732993559594235e-07,
      "loss": 0.6814,
      "step": 44260
    },
    {
      "epoch": 0.6891263904348977,
      "grad_norm": 3.2576639652252197,
      "learning_rate": 3.2724799659133737e-07,
      "loss": 0.7247,
      "step": 44265
    },
    {
      "epoch": 0.689204231436867,
      "grad_norm": 4.605559349060059,
      "learning_rate": 3.2716605758673244e-07,
      "loss": 0.7084,
      "step": 44270
    },
    {
      "epoch": 0.6892820724388364,
      "grad_norm": 3.6177306175231934,
      "learning_rate": 3.270841185821274e-07,
      "loss": 0.7124,
      "step": 44275
    },
    {
      "epoch": 0.6893599134408058,
      "grad_norm": 6.048017501831055,
      "learning_rate": 3.270021795775225e-07,
      "loss": 0.8291,
      "step": 44280
    },
    {
      "epoch": 0.6894377544427752,
      "grad_norm": 2.741009473800659,
      "learning_rate": 3.269202405729175e-07,
      "loss": 0.7037,
      "step": 44285
    },
    {
      "epoch": 0.6895155954447446,
      "grad_norm": 7.66215705871582,
      "learning_rate": 3.2683830156831253e-07,
      "loss": 0.7676,
      "step": 44290
    },
    {
      "epoch": 0.6895934364467139,
      "grad_norm": 3.2019970417022705,
      "learning_rate": 3.267563625637076e-07,
      "loss": 0.6585,
      "step": 44295
    },
    {
      "epoch": 0.6896712774486833,
      "grad_norm": 4.958421230316162,
      "learning_rate": 3.266744235591026e-07,
      "loss": 0.7021,
      "step": 44300
    },
    {
      "epoch": 0.6897491184506527,
      "grad_norm": 7.29365348815918,
      "learning_rate": 3.265924845544976e-07,
      "loss": 0.7973,
      "step": 44305
    },
    {
      "epoch": 0.6898269594526221,
      "grad_norm": 4.6497483253479,
      "learning_rate": 3.2651054554989267e-07,
      "loss": 0.7737,
      "step": 44310
    },
    {
      "epoch": 0.6899048004545915,
      "grad_norm": 7.096072196960449,
      "learning_rate": 3.264286065452877e-07,
      "loss": 0.6901,
      "step": 44315
    },
    {
      "epoch": 0.6899826414565609,
      "grad_norm": 3.5146732330322266,
      "learning_rate": 3.2634666754068266e-07,
      "loss": 0.6929,
      "step": 44320
    },
    {
      "epoch": 0.6900604824585302,
      "grad_norm": 3.3185009956359863,
      "learning_rate": 3.2626472853607774e-07,
      "loss": 0.7101,
      "step": 44325
    },
    {
      "epoch": 0.6901383234604996,
      "grad_norm": 5.031402587890625,
      "learning_rate": 3.2618278953147276e-07,
      "loss": 0.8186,
      "step": 44330
    },
    {
      "epoch": 0.6902161644624689,
      "grad_norm": 5.139355182647705,
      "learning_rate": 3.2610085052686784e-07,
      "loss": 0.8087,
      "step": 44335
    },
    {
      "epoch": 0.6902940054644383,
      "grad_norm": 4.935749053955078,
      "learning_rate": 3.260189115222628e-07,
      "loss": 0.6429,
      "step": 44340
    },
    {
      "epoch": 0.6903718464664077,
      "grad_norm": 3.576235294342041,
      "learning_rate": 3.259369725176578e-07,
      "loss": 0.7493,
      "step": 44345
    },
    {
      "epoch": 0.6904496874683771,
      "grad_norm": 3.9032094478607178,
      "learning_rate": 3.258550335130529e-07,
      "loss": 0.7773,
      "step": 44350
    },
    {
      "epoch": 0.6905275284703465,
      "grad_norm": 3.9932587146759033,
      "learning_rate": 3.257730945084479e-07,
      "loss": 0.7428,
      "step": 44355
    },
    {
      "epoch": 0.6906053694723159,
      "grad_norm": 3.568847179412842,
      "learning_rate": 3.256911555038429e-07,
      "loss": 0.8298,
      "step": 44360
    },
    {
      "epoch": 0.6906832104742853,
      "grad_norm": 2.7744300365448,
      "learning_rate": 3.2560921649923797e-07,
      "loss": 0.7778,
      "step": 44365
    },
    {
      "epoch": 0.6907610514762546,
      "grad_norm": 3.4983668327331543,
      "learning_rate": 3.25527277494633e-07,
      "loss": 0.7291,
      "step": 44370
    },
    {
      "epoch": 0.6908388924782239,
      "grad_norm": 3.0201833248138428,
      "learning_rate": 3.25445338490028e-07,
      "loss": 0.6851,
      "step": 44375
    },
    {
      "epoch": 0.6909167334801933,
      "grad_norm": 3.1214568614959717,
      "learning_rate": 3.2536339948542303e-07,
      "loss": 0.7091,
      "step": 44380
    },
    {
      "epoch": 0.6909945744821627,
      "grad_norm": 7.290832996368408,
      "learning_rate": 3.2528146048081805e-07,
      "loss": 0.6998,
      "step": 44385
    },
    {
      "epoch": 0.6910724154841321,
      "grad_norm": 3.362792730331421,
      "learning_rate": 3.251995214762131e-07,
      "loss": 0.7388,
      "step": 44390
    },
    {
      "epoch": 0.6911502564861015,
      "grad_norm": 5.960023880004883,
      "learning_rate": 3.2511758247160815e-07,
      "loss": 0.7798,
      "step": 44395
    },
    {
      "epoch": 0.6912280974880709,
      "grad_norm": 4.534994602203369,
      "learning_rate": 3.250356434670031e-07,
      "loss": 0.7691,
      "step": 44400
    },
    {
      "epoch": 0.6913059384900403,
      "grad_norm": 6.411965847015381,
      "learning_rate": 3.249537044623982e-07,
      "loss": 0.7399,
      "step": 44405
    },
    {
      "epoch": 0.6913837794920096,
      "grad_norm": 2.57249116897583,
      "learning_rate": 3.248717654577932e-07,
      "loss": 0.5969,
      "step": 44410
    },
    {
      "epoch": 0.691461620493979,
      "grad_norm": 2.758023262023926,
      "learning_rate": 3.2478982645318824e-07,
      "loss": 0.7118,
      "step": 44415
    },
    {
      "epoch": 0.6915394614959484,
      "grad_norm": 3.7951455116271973,
      "learning_rate": 3.2470788744858326e-07,
      "loss": 0.7253,
      "step": 44420
    },
    {
      "epoch": 0.6916173024979178,
      "grad_norm": 4.069206714630127,
      "learning_rate": 3.246259484439783e-07,
      "loss": 0.6401,
      "step": 44425
    },
    {
      "epoch": 0.6916951434998871,
      "grad_norm": 2.7056708335876465,
      "learning_rate": 3.245440094393733e-07,
      "loss": 0.7773,
      "step": 44430
    },
    {
      "epoch": 0.6917729845018565,
      "grad_norm": 3.3103575706481934,
      "learning_rate": 3.244620704347684e-07,
      "loss": 0.6996,
      "step": 44435
    },
    {
      "epoch": 0.6918508255038259,
      "grad_norm": 3.496828079223633,
      "learning_rate": 3.2438013143016335e-07,
      "loss": 0.7546,
      "step": 44440
    },
    {
      "epoch": 0.6919286665057952,
      "grad_norm": 5.84952974319458,
      "learning_rate": 3.2429819242555837e-07,
      "loss": 0.7056,
      "step": 44445
    },
    {
      "epoch": 0.6920065075077646,
      "grad_norm": 4.366995811462402,
      "learning_rate": 3.2421625342095345e-07,
      "loss": 0.775,
      "step": 44450
    },
    {
      "epoch": 0.692084348509734,
      "grad_norm": 9.660295486450195,
      "learning_rate": 3.2413431441634847e-07,
      "loss": 0.7315,
      "step": 44455
    },
    {
      "epoch": 0.6921621895117034,
      "grad_norm": 4.724564552307129,
      "learning_rate": 3.2405237541174344e-07,
      "loss": 0.8931,
      "step": 44460
    },
    {
      "epoch": 0.6922400305136728,
      "grad_norm": 4.917026519775391,
      "learning_rate": 3.239704364071385e-07,
      "loss": 0.6424,
      "step": 44465
    },
    {
      "epoch": 0.6923178715156422,
      "grad_norm": 3.5691773891448975,
      "learning_rate": 3.2388849740253353e-07,
      "loss": 0.7787,
      "step": 44470
    },
    {
      "epoch": 0.6923957125176116,
      "grad_norm": 3.5394139289855957,
      "learning_rate": 3.238065583979286e-07,
      "loss": 0.727,
      "step": 44475
    },
    {
      "epoch": 0.692473553519581,
      "grad_norm": 6.220149040222168,
      "learning_rate": 3.237246193933236e-07,
      "loss": 0.742,
      "step": 44480
    },
    {
      "epoch": 0.6925513945215502,
      "grad_norm": 2.0169625282287598,
      "learning_rate": 3.236426803887186e-07,
      "loss": 0.5853,
      "step": 44485
    },
    {
      "epoch": 0.6926292355235196,
      "grad_norm": 6.046042442321777,
      "learning_rate": 3.235607413841137e-07,
      "loss": 0.8133,
      "step": 44490
    },
    {
      "epoch": 0.692707076525489,
      "grad_norm": 3.403136730194092,
      "learning_rate": 3.234788023795087e-07,
      "loss": 0.7701,
      "step": 44495
    },
    {
      "epoch": 0.6927849175274584,
      "grad_norm": 3.918903112411499,
      "learning_rate": 3.2339686337490366e-07,
      "loss": 0.7225,
      "step": 44500
    },
    {
      "epoch": 0.6928627585294278,
      "grad_norm": 3.1237285137176514,
      "learning_rate": 3.2331492437029874e-07,
      "loss": 0.7761,
      "step": 44505
    },
    {
      "epoch": 0.6929405995313972,
      "grad_norm": 5.299807548522949,
      "learning_rate": 3.2323298536569376e-07,
      "loss": 0.6597,
      "step": 44510
    },
    {
      "epoch": 0.6930184405333666,
      "grad_norm": 6.620615482330322,
      "learning_rate": 3.231510463610888e-07,
      "loss": 0.9013,
      "step": 44515
    },
    {
      "epoch": 0.6930962815353359,
      "grad_norm": 3.3337907791137695,
      "learning_rate": 3.230691073564838e-07,
      "loss": 0.7842,
      "step": 44520
    },
    {
      "epoch": 0.6931741225373053,
      "grad_norm": 4.090300559997559,
      "learning_rate": 3.2298716835187883e-07,
      "loss": 0.7515,
      "step": 44525
    },
    {
      "epoch": 0.6932519635392747,
      "grad_norm": 3.833961009979248,
      "learning_rate": 3.229052293472739e-07,
      "loss": 0.6598,
      "step": 44530
    },
    {
      "epoch": 0.693329804541244,
      "grad_norm": 3.749589681625366,
      "learning_rate": 3.228232903426689e-07,
      "loss": 0.6539,
      "step": 44535
    },
    {
      "epoch": 0.6934076455432134,
      "grad_norm": 3.8142995834350586,
      "learning_rate": 3.227413513380639e-07,
      "loss": 0.7186,
      "step": 44540
    },
    {
      "epoch": 0.6934854865451828,
      "grad_norm": 7.578232288360596,
      "learning_rate": 3.2265941233345897e-07,
      "loss": 0.6983,
      "step": 44545
    },
    {
      "epoch": 0.6935633275471522,
      "grad_norm": 3.4458231925964355,
      "learning_rate": 3.22577473328854e-07,
      "loss": 0.6991,
      "step": 44550
    },
    {
      "epoch": 0.6936411685491216,
      "grad_norm": 5.668604850769043,
      "learning_rate": 3.22495534324249e-07,
      "loss": 0.6614,
      "step": 44555
    },
    {
      "epoch": 0.6937190095510909,
      "grad_norm": 4.282538414001465,
      "learning_rate": 3.2241359531964403e-07,
      "loss": 0.6482,
      "step": 44560
    },
    {
      "epoch": 0.6937968505530603,
      "grad_norm": 3.9799692630767822,
      "learning_rate": 3.2233165631503906e-07,
      "loss": 0.7115,
      "step": 44565
    },
    {
      "epoch": 0.6938746915550297,
      "grad_norm": 4.081216812133789,
      "learning_rate": 3.222497173104341e-07,
      "loss": 0.8214,
      "step": 44570
    },
    {
      "epoch": 0.6939525325569991,
      "grad_norm": 2.8831448554992676,
      "learning_rate": 3.2216777830582915e-07,
      "loss": 0.8191,
      "step": 44575
    },
    {
      "epoch": 0.6940303735589685,
      "grad_norm": 3.6178247928619385,
      "learning_rate": 3.220858393012241e-07,
      "loss": 0.823,
      "step": 44580
    },
    {
      "epoch": 0.6941082145609379,
      "grad_norm": 4.663093090057373,
      "learning_rate": 3.2200390029661914e-07,
      "loss": 0.7436,
      "step": 44585
    },
    {
      "epoch": 0.6941860555629072,
      "grad_norm": 6.621307849884033,
      "learning_rate": 3.219219612920142e-07,
      "loss": 0.7585,
      "step": 44590
    },
    {
      "epoch": 0.6942638965648766,
      "grad_norm": 6.331142425537109,
      "learning_rate": 3.2184002228740924e-07,
      "loss": 0.7268,
      "step": 44595
    },
    {
      "epoch": 0.6943417375668459,
      "grad_norm": 6.911218643188477,
      "learning_rate": 3.217580832828043e-07,
      "loss": 0.9535,
      "step": 44600
    },
    {
      "epoch": 0.6944195785688153,
      "grad_norm": 1.9647390842437744,
      "learning_rate": 3.216761442781993e-07,
      "loss": 0.6814,
      "step": 44605
    },
    {
      "epoch": 0.6944974195707847,
      "grad_norm": 3.6637792587280273,
      "learning_rate": 3.215942052735943e-07,
      "loss": 0.7003,
      "step": 44610
    },
    {
      "epoch": 0.6945752605727541,
      "grad_norm": 3.96181321144104,
      "learning_rate": 3.215122662689894e-07,
      "loss": 0.6898,
      "step": 44615
    },
    {
      "epoch": 0.6946531015747235,
      "grad_norm": 4.402724742889404,
      "learning_rate": 3.214303272643844e-07,
      "loss": 0.8909,
      "step": 44620
    },
    {
      "epoch": 0.6947309425766929,
      "grad_norm": 4.197029113769531,
      "learning_rate": 3.2134838825977937e-07,
      "loss": 0.7661,
      "step": 44625
    },
    {
      "epoch": 0.6948087835786623,
      "grad_norm": 4.183274745941162,
      "learning_rate": 3.2126644925517445e-07,
      "loss": 0.5703,
      "step": 44630
    },
    {
      "epoch": 0.6948866245806316,
      "grad_norm": 3.2585289478302,
      "learning_rate": 3.2118451025056947e-07,
      "loss": 0.7627,
      "step": 44635
    },
    {
      "epoch": 0.6949644655826009,
      "grad_norm": 4.1372222900390625,
      "learning_rate": 3.211025712459645e-07,
      "loss": 0.8537,
      "step": 44640
    },
    {
      "epoch": 0.6950423065845703,
      "grad_norm": 2.8562052249908447,
      "learning_rate": 3.210206322413595e-07,
      "loss": 0.659,
      "step": 44645
    },
    {
      "epoch": 0.6951201475865397,
      "grad_norm": 8.294960021972656,
      "learning_rate": 3.2093869323675453e-07,
      "loss": 0.7505,
      "step": 44650
    },
    {
      "epoch": 0.6951979885885091,
      "grad_norm": 2.843148708343506,
      "learning_rate": 3.208567542321496e-07,
      "loss": 0.6907,
      "step": 44655
    },
    {
      "epoch": 0.6952758295904785,
      "grad_norm": 3.288318634033203,
      "learning_rate": 3.2077481522754463e-07,
      "loss": 0.6858,
      "step": 44660
    },
    {
      "epoch": 0.6953536705924479,
      "grad_norm": 3.941612958908081,
      "learning_rate": 3.206928762229396e-07,
      "loss": 0.6343,
      "step": 44665
    },
    {
      "epoch": 0.6954315115944173,
      "grad_norm": 4.528519630432129,
      "learning_rate": 3.206109372183347e-07,
      "loss": 0.7629,
      "step": 44670
    },
    {
      "epoch": 0.6955093525963866,
      "grad_norm": 3.559549570083618,
      "learning_rate": 3.205289982137297e-07,
      "loss": 0.7864,
      "step": 44675
    },
    {
      "epoch": 0.695587193598356,
      "grad_norm": 4.656329154968262,
      "learning_rate": 3.204470592091247e-07,
      "loss": 0.7389,
      "step": 44680
    },
    {
      "epoch": 0.6956650346003254,
      "grad_norm": 2.7449679374694824,
      "learning_rate": 3.2036512020451974e-07,
      "loss": 0.6378,
      "step": 44685
    },
    {
      "epoch": 0.6957428756022948,
      "grad_norm": 4.100686073303223,
      "learning_rate": 3.2028318119991476e-07,
      "loss": 0.7488,
      "step": 44690
    },
    {
      "epoch": 0.6958207166042641,
      "grad_norm": 3.0010628700256348,
      "learning_rate": 3.202012421953098e-07,
      "loss": 0.6962,
      "step": 44695
    },
    {
      "epoch": 0.6958985576062335,
      "grad_norm": 3.087158203125,
      "learning_rate": 3.2011930319070486e-07,
      "loss": 0.757,
      "step": 44700
    },
    {
      "epoch": 0.6959763986082029,
      "grad_norm": 3.2836010456085205,
      "learning_rate": 3.2003736418609983e-07,
      "loss": 0.6567,
      "step": 44705
    },
    {
      "epoch": 0.6960542396101722,
      "grad_norm": 3.5438196659088135,
      "learning_rate": 3.1995542518149485e-07,
      "loss": 0.5942,
      "step": 44710
    },
    {
      "epoch": 0.6961320806121416,
      "grad_norm": 2.8008615970611572,
      "learning_rate": 3.198734861768899e-07,
      "loss": 0.7626,
      "step": 44715
    },
    {
      "epoch": 0.696209921614111,
      "grad_norm": 4.08805513381958,
      "learning_rate": 3.1979154717228495e-07,
      "loss": 0.8112,
      "step": 44720
    },
    {
      "epoch": 0.6962877626160804,
      "grad_norm": 2.729886531829834,
      "learning_rate": 3.1970960816767997e-07,
      "loss": 0.7778,
      "step": 44725
    },
    {
      "epoch": 0.6963656036180498,
      "grad_norm": 4.44882869720459,
      "learning_rate": 3.19627669163075e-07,
      "loss": 0.7052,
      "step": 44730
    },
    {
      "epoch": 0.6964434446200192,
      "grad_norm": 2.9087371826171875,
      "learning_rate": 3.1954573015847e-07,
      "loss": 0.6236,
      "step": 44735
    },
    {
      "epoch": 0.6965212856219886,
      "grad_norm": 3.8849921226501465,
      "learning_rate": 3.194637911538651e-07,
      "loss": 0.6876,
      "step": 44740
    },
    {
      "epoch": 0.696599126623958,
      "grad_norm": 2.966862678527832,
      "learning_rate": 3.1938185214926006e-07,
      "loss": 0.7456,
      "step": 44745
    },
    {
      "epoch": 0.6966769676259272,
      "grad_norm": 3.9457006454467773,
      "learning_rate": 3.192999131446551e-07,
      "loss": 0.7936,
      "step": 44750
    },
    {
      "epoch": 0.6967548086278966,
      "grad_norm": 3.073709726333618,
      "learning_rate": 3.1921797414005015e-07,
      "loss": 0.7958,
      "step": 44755
    },
    {
      "epoch": 0.696832649629866,
      "grad_norm": 3.9600818157196045,
      "learning_rate": 3.191360351354452e-07,
      "loss": 0.6937,
      "step": 44760
    },
    {
      "epoch": 0.6969104906318354,
      "grad_norm": 4.376247406005859,
      "learning_rate": 3.1905409613084014e-07,
      "loss": 0.7915,
      "step": 44765
    },
    {
      "epoch": 0.6969883316338048,
      "grad_norm": 4.799108505249023,
      "learning_rate": 3.189721571262352e-07,
      "loss": 0.8302,
      "step": 44770
    },
    {
      "epoch": 0.6970661726357742,
      "grad_norm": 3.601184844970703,
      "learning_rate": 3.1889021812163024e-07,
      "loss": 0.7675,
      "step": 44775
    },
    {
      "epoch": 0.6971440136377436,
      "grad_norm": 6.4482421875,
      "learning_rate": 3.188082791170253e-07,
      "loss": 0.6881,
      "step": 44780
    },
    {
      "epoch": 0.6972218546397129,
      "grad_norm": 2.7261383533477783,
      "learning_rate": 3.187263401124203e-07,
      "loss": 0.7419,
      "step": 44785
    },
    {
      "epoch": 0.6972996956416823,
      "grad_norm": 3.5320985317230225,
      "learning_rate": 3.186444011078153e-07,
      "loss": 0.6641,
      "step": 44790
    },
    {
      "epoch": 0.6973775366436517,
      "grad_norm": 3.657058000564575,
      "learning_rate": 3.185624621032104e-07,
      "loss": 0.632,
      "step": 44795
    },
    {
      "epoch": 0.697455377645621,
      "grad_norm": 5.171084880828857,
      "learning_rate": 3.184805230986054e-07,
      "loss": 0.6747,
      "step": 44800
    },
    {
      "epoch": 0.6975332186475904,
      "grad_norm": 4.433236598968506,
      "learning_rate": 3.1839858409400037e-07,
      "loss": 0.7931,
      "step": 44805
    },
    {
      "epoch": 0.6976110596495598,
      "grad_norm": 3.699195384979248,
      "learning_rate": 3.1831664508939545e-07,
      "loss": 0.8182,
      "step": 44810
    },
    {
      "epoch": 0.6976889006515292,
      "grad_norm": 5.8162841796875,
      "learning_rate": 3.1823470608479047e-07,
      "loss": 0.7299,
      "step": 44815
    },
    {
      "epoch": 0.6977667416534986,
      "grad_norm": 3.622817039489746,
      "learning_rate": 3.181527670801855e-07,
      "loss": 0.7539,
      "step": 44820
    },
    {
      "epoch": 0.6978445826554679,
      "grad_norm": 4.663959503173828,
      "learning_rate": 3.180708280755805e-07,
      "loss": 0.6796,
      "step": 44825
    },
    {
      "epoch": 0.6979224236574373,
      "grad_norm": 5.090993404388428,
      "learning_rate": 3.1798888907097553e-07,
      "loss": 0.7707,
      "step": 44830
    },
    {
      "epoch": 0.6980002646594067,
      "grad_norm": 6.511724948883057,
      "learning_rate": 3.1790695006637056e-07,
      "loss": 0.8612,
      "step": 44835
    },
    {
      "epoch": 0.6980781056613761,
      "grad_norm": 3.174237012863159,
      "learning_rate": 3.1782501106176563e-07,
      "loss": 0.7997,
      "step": 44840
    },
    {
      "epoch": 0.6981559466633455,
      "grad_norm": 6.8070783615112305,
      "learning_rate": 3.177430720571606e-07,
      "loss": 0.8806,
      "step": 44845
    },
    {
      "epoch": 0.6982337876653149,
      "grad_norm": 4.633585453033447,
      "learning_rate": 3.176611330525557e-07,
      "loss": 0.8046,
      "step": 44850
    },
    {
      "epoch": 0.6983116286672842,
      "grad_norm": 5.13078498840332,
      "learning_rate": 3.175791940479507e-07,
      "loss": 0.8624,
      "step": 44855
    },
    {
      "epoch": 0.6983894696692536,
      "grad_norm": 3.29133939743042,
      "learning_rate": 3.174972550433457e-07,
      "loss": 0.725,
      "step": 44860
    },
    {
      "epoch": 0.6984673106712229,
      "grad_norm": 2.964341163635254,
      "learning_rate": 3.1741531603874074e-07,
      "loss": 0.8322,
      "step": 44865
    },
    {
      "epoch": 0.6985451516731923,
      "grad_norm": 11.552732467651367,
      "learning_rate": 3.1733337703413576e-07,
      "loss": 0.7734,
      "step": 44870
    },
    {
      "epoch": 0.6986229926751617,
      "grad_norm": 4.12690544128418,
      "learning_rate": 3.172514380295308e-07,
      "loss": 0.6674,
      "step": 44875
    },
    {
      "epoch": 0.6987008336771311,
      "grad_norm": 9.65038776397705,
      "learning_rate": 3.1716949902492586e-07,
      "loss": 0.7262,
      "step": 44880
    },
    {
      "epoch": 0.6987786746791005,
      "grad_norm": 5.294374465942383,
      "learning_rate": 3.1708756002032083e-07,
      "loss": 0.8374,
      "step": 44885
    },
    {
      "epoch": 0.6988565156810699,
      "grad_norm": 4.357327938079834,
      "learning_rate": 3.1700562101571585e-07,
      "loss": 0.7354,
      "step": 44890
    },
    {
      "epoch": 0.6989343566830393,
      "grad_norm": 2.6193816661834717,
      "learning_rate": 3.169236820111109e-07,
      "loss": 0.7096,
      "step": 44895
    },
    {
      "epoch": 0.6990121976850086,
      "grad_norm": 3.466761827468872,
      "learning_rate": 3.1684174300650595e-07,
      "loss": 0.7312,
      "step": 44900
    },
    {
      "epoch": 0.6990900386869779,
      "grad_norm": 3.758063316345215,
      "learning_rate": 3.1675980400190097e-07,
      "loss": 0.7749,
      "step": 44905
    },
    {
      "epoch": 0.6991678796889473,
      "grad_norm": 3.2110955715179443,
      "learning_rate": 3.16677864997296e-07,
      "loss": 0.6738,
      "step": 44910
    },
    {
      "epoch": 0.6992457206909167,
      "grad_norm": 4.206228733062744,
      "learning_rate": 3.16595925992691e-07,
      "loss": 0.7151,
      "step": 44915
    },
    {
      "epoch": 0.6993235616928861,
      "grad_norm": 4.669415473937988,
      "learning_rate": 3.165139869880861e-07,
      "loss": 0.6914,
      "step": 44920
    },
    {
      "epoch": 0.6994014026948555,
      "grad_norm": 5.395704746246338,
      "learning_rate": 3.164320479834811e-07,
      "loss": 0.7092,
      "step": 44925
    },
    {
      "epoch": 0.6994792436968249,
      "grad_norm": 3.740260601043701,
      "learning_rate": 3.163501089788761e-07,
      "loss": 0.6827,
      "step": 44930
    },
    {
      "epoch": 0.6995570846987943,
      "grad_norm": 3.921980381011963,
      "learning_rate": 3.1626816997427115e-07,
      "loss": 0.58,
      "step": 44935
    },
    {
      "epoch": 0.6996349257007636,
      "grad_norm": 4.55852746963501,
      "learning_rate": 3.161862309696662e-07,
      "loss": 0.651,
      "step": 44940
    },
    {
      "epoch": 0.699712766702733,
      "grad_norm": 4.210649490356445,
      "learning_rate": 3.161042919650612e-07,
      "loss": 0.7365,
      "step": 44945
    },
    {
      "epoch": 0.6997906077047024,
      "grad_norm": 3.2003166675567627,
      "learning_rate": 3.160223529604562e-07,
      "loss": 0.8432,
      "step": 44950
    },
    {
      "epoch": 0.6998684487066718,
      "grad_norm": 8.792283058166504,
      "learning_rate": 3.1594041395585124e-07,
      "loss": 0.9186,
      "step": 44955
    },
    {
      "epoch": 0.6999462897086411,
      "grad_norm": 5.730156898498535,
      "learning_rate": 3.1585847495124626e-07,
      "loss": 0.6962,
      "step": 44960
    },
    {
      "epoch": 0.7000241307106105,
      "grad_norm": 2.5939512252807617,
      "learning_rate": 3.1577653594664134e-07,
      "loss": 0.6794,
      "step": 44965
    },
    {
      "epoch": 0.7001019717125799,
      "grad_norm": 4.188800811767578,
      "learning_rate": 3.156945969420363e-07,
      "loss": 0.8593,
      "step": 44970
    },
    {
      "epoch": 0.7001798127145492,
      "grad_norm": 3.341172695159912,
      "learning_rate": 3.156126579374314e-07,
      "loss": 0.7753,
      "step": 44975
    },
    {
      "epoch": 0.7002576537165186,
      "grad_norm": 2.754958152770996,
      "learning_rate": 3.155307189328264e-07,
      "loss": 0.6972,
      "step": 44980
    },
    {
      "epoch": 0.700335494718488,
      "grad_norm": 7.0763936042785645,
      "learning_rate": 3.154487799282214e-07,
      "loss": 0.9043,
      "step": 44985
    },
    {
      "epoch": 0.7004133357204574,
      "grad_norm": 3.0021302700042725,
      "learning_rate": 3.1536684092361645e-07,
      "loss": 0.7113,
      "step": 44990
    },
    {
      "epoch": 0.7004911767224268,
      "grad_norm": 2.9225008487701416,
      "learning_rate": 3.1528490191901147e-07,
      "loss": 0.6894,
      "step": 44995
    },
    {
      "epoch": 0.7005690177243962,
      "grad_norm": 4.19906759262085,
      "learning_rate": 3.152029629144065e-07,
      "loss": 0.7342,
      "step": 45000
    },
    {
      "epoch": 0.7006468587263656,
      "grad_norm": 4.56402063369751,
      "learning_rate": 3.1512102390980157e-07,
      "loss": 0.8398,
      "step": 45005
    },
    {
      "epoch": 0.700724699728335,
      "grad_norm": 2.982168197631836,
      "learning_rate": 3.1503908490519654e-07,
      "loss": 0.6646,
      "step": 45010
    },
    {
      "epoch": 0.7008025407303042,
      "grad_norm": 4.950345516204834,
      "learning_rate": 3.1495714590059156e-07,
      "loss": 0.8003,
      "step": 45015
    },
    {
      "epoch": 0.7008803817322736,
      "grad_norm": 3.263009786605835,
      "learning_rate": 3.1487520689598663e-07,
      "loss": 0.6252,
      "step": 45020
    },
    {
      "epoch": 0.700958222734243,
      "grad_norm": 5.379505157470703,
      "learning_rate": 3.1479326789138165e-07,
      "loss": 0.7242,
      "step": 45025
    },
    {
      "epoch": 0.7010360637362124,
      "grad_norm": 3.0536603927612305,
      "learning_rate": 3.147113288867767e-07,
      "loss": 0.6973,
      "step": 45030
    },
    {
      "epoch": 0.7011139047381818,
      "grad_norm": 3.48382830619812,
      "learning_rate": 3.146293898821717e-07,
      "loss": 0.8481,
      "step": 45035
    },
    {
      "epoch": 0.7011917457401512,
      "grad_norm": 2.684818983078003,
      "learning_rate": 3.145474508775667e-07,
      "loss": 0.7994,
      "step": 45040
    },
    {
      "epoch": 0.7012695867421206,
      "grad_norm": 7.540963649749756,
      "learning_rate": 3.144655118729618e-07,
      "loss": 0.7522,
      "step": 45045
    },
    {
      "epoch": 0.7013474277440899,
      "grad_norm": 12.608531951904297,
      "learning_rate": 3.1438357286835676e-07,
      "loss": 0.8047,
      "step": 45050
    },
    {
      "epoch": 0.7014252687460593,
      "grad_norm": 4.38758659362793,
      "learning_rate": 3.143016338637518e-07,
      "loss": 0.7215,
      "step": 45055
    },
    {
      "epoch": 0.7015031097480287,
      "grad_norm": 2.7637720108032227,
      "learning_rate": 3.1421969485914686e-07,
      "loss": 0.6849,
      "step": 45060
    },
    {
      "epoch": 0.701580950749998,
      "grad_norm": 4.568196773529053,
      "learning_rate": 3.141377558545419e-07,
      "loss": 0.7472,
      "step": 45065
    },
    {
      "epoch": 0.7016587917519674,
      "grad_norm": 3.447049617767334,
      "learning_rate": 3.1405581684993685e-07,
      "loss": 0.6996,
      "step": 45070
    },
    {
      "epoch": 0.7017366327539368,
      "grad_norm": 2.971358299255371,
      "learning_rate": 3.1397387784533193e-07,
      "loss": 0.7863,
      "step": 45075
    },
    {
      "epoch": 0.7018144737559062,
      "grad_norm": 5.023493766784668,
      "learning_rate": 3.1389193884072695e-07,
      "loss": 0.7235,
      "step": 45080
    },
    {
      "epoch": 0.7018923147578756,
      "grad_norm": 4.907797813415527,
      "learning_rate": 3.13809999836122e-07,
      "loss": 0.8026,
      "step": 45085
    },
    {
      "epoch": 0.7019701557598449,
      "grad_norm": 5.246790885925293,
      "learning_rate": 3.13728060831517e-07,
      "loss": 0.8177,
      "step": 45090
    },
    {
      "epoch": 0.7020479967618143,
      "grad_norm": 3.3034846782684326,
      "learning_rate": 3.13646121826912e-07,
      "loss": 0.7011,
      "step": 45095
    },
    {
      "epoch": 0.7021258377637837,
      "grad_norm": 3.2205328941345215,
      "learning_rate": 3.135641828223071e-07,
      "loss": 0.8903,
      "step": 45100
    },
    {
      "epoch": 0.7022036787657531,
      "grad_norm": 6.118496894836426,
      "learning_rate": 3.134822438177021e-07,
      "loss": 0.7767,
      "step": 45105
    },
    {
      "epoch": 0.7022815197677225,
      "grad_norm": 5.2367777824401855,
      "learning_rate": 3.134003048130971e-07,
      "loss": 0.7674,
      "step": 45110
    },
    {
      "epoch": 0.7023593607696919,
      "grad_norm": 8.271392822265625,
      "learning_rate": 3.1331836580849215e-07,
      "loss": 0.691,
      "step": 45115
    },
    {
      "epoch": 0.7024372017716612,
      "grad_norm": 5.897238254547119,
      "learning_rate": 3.132364268038872e-07,
      "loss": 0.7193,
      "step": 45120
    },
    {
      "epoch": 0.7025150427736305,
      "grad_norm": 3.138092279434204,
      "learning_rate": 3.131544877992822e-07,
      "loss": 0.8006,
      "step": 45125
    },
    {
      "epoch": 0.7025928837755999,
      "grad_norm": 2.9634246826171875,
      "learning_rate": 3.130725487946772e-07,
      "loss": 0.6906,
      "step": 45130
    },
    {
      "epoch": 0.7026707247775693,
      "grad_norm": 3.6134426593780518,
      "learning_rate": 3.1299060979007224e-07,
      "loss": 0.8219,
      "step": 45135
    },
    {
      "epoch": 0.7027485657795387,
      "grad_norm": 2.544174909591675,
      "learning_rate": 3.1290867078546726e-07,
      "loss": 0.7499,
      "step": 45140
    },
    {
      "epoch": 0.7028264067815081,
      "grad_norm": 4.649796009063721,
      "learning_rate": 3.1282673178086234e-07,
      "loss": 0.6047,
      "step": 45145
    },
    {
      "epoch": 0.7029042477834775,
      "grad_norm": 3.823444366455078,
      "learning_rate": 3.127447927762573e-07,
      "loss": 0.725,
      "step": 45150
    },
    {
      "epoch": 0.7029820887854469,
      "grad_norm": 3.6117119789123535,
      "learning_rate": 3.126628537716524e-07,
      "loss": 0.7306,
      "step": 45155
    },
    {
      "epoch": 0.7030599297874163,
      "grad_norm": 9.543429374694824,
      "learning_rate": 3.125809147670474e-07,
      "loss": 0.7857,
      "step": 45160
    },
    {
      "epoch": 0.7031377707893856,
      "grad_norm": 2.7647228240966797,
      "learning_rate": 3.1249897576244243e-07,
      "loss": 0.7348,
      "step": 45165
    },
    {
      "epoch": 0.7032156117913549,
      "grad_norm": 6.592708587646484,
      "learning_rate": 3.1241703675783745e-07,
      "loss": 0.7725,
      "step": 45170
    },
    {
      "epoch": 0.7032934527933243,
      "grad_norm": 3.463387966156006,
      "learning_rate": 3.1233509775323247e-07,
      "loss": 0.7784,
      "step": 45175
    },
    {
      "epoch": 0.7033712937952937,
      "grad_norm": 3.4816014766693115,
      "learning_rate": 3.122531587486275e-07,
      "loss": 0.7576,
      "step": 45180
    },
    {
      "epoch": 0.7034491347972631,
      "grad_norm": 3.45517635345459,
      "learning_rate": 3.1217121974402257e-07,
      "loss": 0.7271,
      "step": 45185
    },
    {
      "epoch": 0.7035269757992325,
      "grad_norm": 3.1751227378845215,
      "learning_rate": 3.1208928073941754e-07,
      "loss": 0.8468,
      "step": 45190
    },
    {
      "epoch": 0.7036048168012019,
      "grad_norm": 5.186757564544678,
      "learning_rate": 3.1200734173481256e-07,
      "loss": 0.7713,
      "step": 45195
    },
    {
      "epoch": 0.7036826578031713,
      "grad_norm": 8.4635591506958,
      "learning_rate": 3.1192540273020763e-07,
      "loss": 0.7104,
      "step": 45200
    },
    {
      "epoch": 0.7037604988051406,
      "grad_norm": 4.862208366394043,
      "learning_rate": 3.1184346372560266e-07,
      "loss": 0.64,
      "step": 45205
    },
    {
      "epoch": 0.70383833980711,
      "grad_norm": 3.0354807376861572,
      "learning_rate": 3.117615247209976e-07,
      "loss": 0.7514,
      "step": 45210
    },
    {
      "epoch": 0.7039161808090794,
      "grad_norm": 2.9959049224853516,
      "learning_rate": 3.116795857163927e-07,
      "loss": 0.6536,
      "step": 45215
    },
    {
      "epoch": 0.7039940218110488,
      "grad_norm": 3.4781994819641113,
      "learning_rate": 3.115976467117877e-07,
      "loss": 0.7366,
      "step": 45220
    },
    {
      "epoch": 0.7040718628130181,
      "grad_norm": 9.216371536254883,
      "learning_rate": 3.115157077071828e-07,
      "loss": 0.6573,
      "step": 45225
    },
    {
      "epoch": 0.7041497038149875,
      "grad_norm": 4.113851547241211,
      "learning_rate": 3.1143376870257776e-07,
      "loss": 0.7436,
      "step": 45230
    },
    {
      "epoch": 0.7042275448169569,
      "grad_norm": 5.015747547149658,
      "learning_rate": 3.113518296979728e-07,
      "loss": 0.7468,
      "step": 45235
    },
    {
      "epoch": 0.7043053858189262,
      "grad_norm": 3.8564445972442627,
      "learning_rate": 3.1126989069336786e-07,
      "loss": 0.8319,
      "step": 45240
    },
    {
      "epoch": 0.7043832268208956,
      "grad_norm": 3.4509637355804443,
      "learning_rate": 3.111879516887629e-07,
      "loss": 0.6931,
      "step": 45245
    },
    {
      "epoch": 0.704461067822865,
      "grad_norm": 3.5531883239746094,
      "learning_rate": 3.111060126841579e-07,
      "loss": 0.6236,
      "step": 45250
    },
    {
      "epoch": 0.7045389088248344,
      "grad_norm": 4.095730781555176,
      "learning_rate": 3.1102407367955293e-07,
      "loss": 0.6178,
      "step": 45255
    },
    {
      "epoch": 0.7046167498268038,
      "grad_norm": 4.6644792556762695,
      "learning_rate": 3.1094213467494795e-07,
      "loss": 0.7607,
      "step": 45260
    },
    {
      "epoch": 0.7046945908287732,
      "grad_norm": 4.015369892120361,
      "learning_rate": 3.1086019567034297e-07,
      "loss": 0.759,
      "step": 45265
    },
    {
      "epoch": 0.7047724318307426,
      "grad_norm": 3.1052534580230713,
      "learning_rate": 3.1077825666573805e-07,
      "loss": 0.7348,
      "step": 45270
    },
    {
      "epoch": 0.704850272832712,
      "grad_norm": 4.672604084014893,
      "learning_rate": 3.10696317661133e-07,
      "loss": 0.789,
      "step": 45275
    },
    {
      "epoch": 0.7049281138346812,
      "grad_norm": 3.948845624923706,
      "learning_rate": 3.106143786565281e-07,
      "loss": 0.7193,
      "step": 45280
    },
    {
      "epoch": 0.7050059548366506,
      "grad_norm": 2.839003086090088,
      "learning_rate": 3.105324396519231e-07,
      "loss": 0.8319,
      "step": 45285
    },
    {
      "epoch": 0.70508379583862,
      "grad_norm": 3.9795756340026855,
      "learning_rate": 3.1045050064731813e-07,
      "loss": 0.7594,
      "step": 45290
    },
    {
      "epoch": 0.7051616368405894,
      "grad_norm": 7.408195495605469,
      "learning_rate": 3.1036856164271316e-07,
      "loss": 0.7917,
      "step": 45295
    },
    {
      "epoch": 0.7052394778425588,
      "grad_norm": 5.954751491546631,
      "learning_rate": 3.102866226381082e-07,
      "loss": 0.7412,
      "step": 45300
    },
    {
      "epoch": 0.7053173188445282,
      "grad_norm": 4.557971477508545,
      "learning_rate": 3.102046836335032e-07,
      "loss": 0.7792,
      "step": 45305
    },
    {
      "epoch": 0.7053951598464976,
      "grad_norm": 5.036478042602539,
      "learning_rate": 3.101227446288983e-07,
      "loss": 0.7331,
      "step": 45310
    },
    {
      "epoch": 0.7054730008484669,
      "grad_norm": 3.1639325618743896,
      "learning_rate": 3.1004080562429324e-07,
      "loss": 0.7343,
      "step": 45315
    },
    {
      "epoch": 0.7055508418504363,
      "grad_norm": 2.961461067199707,
      "learning_rate": 3.0995886661968827e-07,
      "loss": 0.613,
      "step": 45320
    },
    {
      "epoch": 0.7056286828524057,
      "grad_norm": 12.662181854248047,
      "learning_rate": 3.0987692761508334e-07,
      "loss": 0.6706,
      "step": 45325
    },
    {
      "epoch": 0.705706523854375,
      "grad_norm": 8.014673233032227,
      "learning_rate": 3.0979498861047836e-07,
      "loss": 0.8589,
      "step": 45330
    },
    {
      "epoch": 0.7057843648563444,
      "grad_norm": 2.823760509490967,
      "learning_rate": 3.0971304960587333e-07,
      "loss": 0.7706,
      "step": 45335
    },
    {
      "epoch": 0.7058622058583138,
      "grad_norm": 4.188385963439941,
      "learning_rate": 3.096311106012684e-07,
      "loss": 0.6794,
      "step": 45340
    },
    {
      "epoch": 0.7059400468602832,
      "grad_norm": 5.733588695526123,
      "learning_rate": 3.0954917159666343e-07,
      "loss": 0.7499,
      "step": 45345
    },
    {
      "epoch": 0.7060178878622526,
      "grad_norm": 2.767557144165039,
      "learning_rate": 3.094672325920585e-07,
      "loss": 0.6186,
      "step": 45350
    },
    {
      "epoch": 0.7060957288642219,
      "grad_norm": 5.1272501945495605,
      "learning_rate": 3.0938529358745347e-07,
      "loss": 0.7788,
      "step": 45355
    },
    {
      "epoch": 0.7061735698661913,
      "grad_norm": 5.058626174926758,
      "learning_rate": 3.093033545828485e-07,
      "loss": 0.8508,
      "step": 45360
    },
    {
      "epoch": 0.7062514108681607,
      "grad_norm": 6.653048992156982,
      "learning_rate": 3.0922141557824357e-07,
      "loss": 0.6942,
      "step": 45365
    },
    {
      "epoch": 0.7063292518701301,
      "grad_norm": 3.692868947982788,
      "learning_rate": 3.091394765736386e-07,
      "loss": 0.6633,
      "step": 45370
    },
    {
      "epoch": 0.7064070928720995,
      "grad_norm": 10.436076164245605,
      "learning_rate": 3.0905753756903356e-07,
      "loss": 0.8079,
      "step": 45375
    },
    {
      "epoch": 0.7064849338740689,
      "grad_norm": 3.6519272327423096,
      "learning_rate": 3.0897559856442863e-07,
      "loss": 0.7221,
      "step": 45380
    },
    {
      "epoch": 0.7065627748760382,
      "grad_norm": 3.799046754837036,
      "learning_rate": 3.0889365955982366e-07,
      "loss": 0.6609,
      "step": 45385
    },
    {
      "epoch": 0.7066406158780075,
      "grad_norm": 6.7345685958862305,
      "learning_rate": 3.088117205552187e-07,
      "loss": 0.6395,
      "step": 45390
    },
    {
      "epoch": 0.7067184568799769,
      "grad_norm": 3.855308771133423,
      "learning_rate": 3.087297815506137e-07,
      "loss": 0.7133,
      "step": 45395
    },
    {
      "epoch": 0.7067962978819463,
      "grad_norm": 2.894350290298462,
      "learning_rate": 3.086478425460087e-07,
      "loss": 0.7765,
      "step": 45400
    },
    {
      "epoch": 0.7068741388839157,
      "grad_norm": 4.465053081512451,
      "learning_rate": 3.085659035414038e-07,
      "loss": 0.755,
      "step": 45405
    },
    {
      "epoch": 0.7069519798858851,
      "grad_norm": 2.6621930599212646,
      "learning_rate": 3.084839645367988e-07,
      "loss": 0.7042,
      "step": 45410
    },
    {
      "epoch": 0.7070298208878545,
      "grad_norm": 4.088991641998291,
      "learning_rate": 3.084020255321938e-07,
      "loss": 0.7653,
      "step": 45415
    },
    {
      "epoch": 0.7071076618898239,
      "grad_norm": 4.188680171966553,
      "learning_rate": 3.0832008652758886e-07,
      "loss": 0.7812,
      "step": 45420
    },
    {
      "epoch": 0.7071855028917933,
      "grad_norm": 3.3172194957733154,
      "learning_rate": 3.082381475229839e-07,
      "loss": 0.7952,
      "step": 45425
    },
    {
      "epoch": 0.7072633438937626,
      "grad_norm": 3.730107545852661,
      "learning_rate": 3.081562085183789e-07,
      "loss": 0.7169,
      "step": 45430
    },
    {
      "epoch": 0.707341184895732,
      "grad_norm": 5.2290496826171875,
      "learning_rate": 3.0807426951377393e-07,
      "loss": 0.7167,
      "step": 45435
    },
    {
      "epoch": 0.7074190258977013,
      "grad_norm": 2.853937864303589,
      "learning_rate": 3.0799233050916895e-07,
      "loss": 0.7418,
      "step": 45440
    },
    {
      "epoch": 0.7074968668996707,
      "grad_norm": 8.731186866760254,
      "learning_rate": 3.0791039150456397e-07,
      "loss": 0.6501,
      "step": 45445
    },
    {
      "epoch": 0.7075747079016401,
      "grad_norm": 3.099287748336792,
      "learning_rate": 3.0782845249995905e-07,
      "loss": 0.8412,
      "step": 45450
    },
    {
      "epoch": 0.7076525489036095,
      "grad_norm": 2.228121757507324,
      "learning_rate": 3.07746513495354e-07,
      "loss": 0.6239,
      "step": 45455
    },
    {
      "epoch": 0.7077303899055789,
      "grad_norm": 5.449446201324463,
      "learning_rate": 3.0766457449074904e-07,
      "loss": 0.6971,
      "step": 45460
    },
    {
      "epoch": 0.7078082309075483,
      "grad_norm": 9.187175750732422,
      "learning_rate": 3.075826354861441e-07,
      "loss": 0.7164,
      "step": 45465
    },
    {
      "epoch": 0.7078860719095176,
      "grad_norm": 4.421596527099609,
      "learning_rate": 3.0750069648153913e-07,
      "loss": 0.7804,
      "step": 45470
    },
    {
      "epoch": 0.707963912911487,
      "grad_norm": 3.50272798538208,
      "learning_rate": 3.0741875747693416e-07,
      "loss": 0.7776,
      "step": 45475
    },
    {
      "epoch": 0.7080417539134564,
      "grad_norm": 4.509073734283447,
      "learning_rate": 3.073368184723292e-07,
      "loss": 0.8449,
      "step": 45480
    },
    {
      "epoch": 0.7081195949154258,
      "grad_norm": 3.4443674087524414,
      "learning_rate": 3.072548794677242e-07,
      "loss": 0.7249,
      "step": 45485
    },
    {
      "epoch": 0.7081974359173951,
      "grad_norm": 2.2308735847473145,
      "learning_rate": 3.071729404631193e-07,
      "loss": 0.7071,
      "step": 45490
    },
    {
      "epoch": 0.7082752769193645,
      "grad_norm": 5.674857139587402,
      "learning_rate": 3.0709100145851424e-07,
      "loss": 0.6811,
      "step": 45495
    },
    {
      "epoch": 0.7083531179213339,
      "grad_norm": 3.3135907649993896,
      "learning_rate": 3.0700906245390927e-07,
      "loss": 0.7141,
      "step": 45500
    },
    {
      "epoch": 0.7084309589233032,
      "grad_norm": 6.646925926208496,
      "learning_rate": 3.0692712344930434e-07,
      "loss": 0.7331,
      "step": 45505
    },
    {
      "epoch": 0.7085087999252726,
      "grad_norm": 6.949533939361572,
      "learning_rate": 3.0684518444469936e-07,
      "loss": 0.8741,
      "step": 45510
    },
    {
      "epoch": 0.708586640927242,
      "grad_norm": 2.832998514175415,
      "learning_rate": 3.0676324544009433e-07,
      "loss": 0.7111,
      "step": 45515
    },
    {
      "epoch": 0.7086644819292114,
      "grad_norm": 3.0167109966278076,
      "learning_rate": 3.066813064354894e-07,
      "loss": 0.7209,
      "step": 45520
    },
    {
      "epoch": 0.7087423229311808,
      "grad_norm": 8.416694641113281,
      "learning_rate": 3.0659936743088443e-07,
      "loss": 0.6188,
      "step": 45525
    },
    {
      "epoch": 0.7088201639331502,
      "grad_norm": 4.774659156799316,
      "learning_rate": 3.065174284262795e-07,
      "loss": 0.7768,
      "step": 45530
    },
    {
      "epoch": 0.7088980049351196,
      "grad_norm": 4.5973358154296875,
      "learning_rate": 3.0643548942167447e-07,
      "loss": 0.72,
      "step": 45535
    },
    {
      "epoch": 0.708975845937089,
      "grad_norm": 2.840872287750244,
      "learning_rate": 3.063535504170695e-07,
      "loss": 0.7967,
      "step": 45540
    },
    {
      "epoch": 0.7090536869390582,
      "grad_norm": 4.631728649139404,
      "learning_rate": 3.0627161141246457e-07,
      "loss": 0.6744,
      "step": 45545
    },
    {
      "epoch": 0.7091315279410276,
      "grad_norm": 3.398324728012085,
      "learning_rate": 3.061896724078596e-07,
      "loss": 0.8776,
      "step": 45550
    },
    {
      "epoch": 0.709209368942997,
      "grad_norm": 3.480661153793335,
      "learning_rate": 3.061077334032546e-07,
      "loss": 0.8824,
      "step": 45555
    },
    {
      "epoch": 0.7092872099449664,
      "grad_norm": 2.9502832889556885,
      "learning_rate": 3.0602579439864964e-07,
      "loss": 0.6202,
      "step": 45560
    },
    {
      "epoch": 0.7093650509469358,
      "grad_norm": 3.6461739540100098,
      "learning_rate": 3.0594385539404466e-07,
      "loss": 0.7491,
      "step": 45565
    },
    {
      "epoch": 0.7094428919489052,
      "grad_norm": 5.5186896324157715,
      "learning_rate": 3.058619163894397e-07,
      "loss": 0.6887,
      "step": 45570
    },
    {
      "epoch": 0.7095207329508746,
      "grad_norm": 7.2920613288879395,
      "learning_rate": 3.0577997738483475e-07,
      "loss": 0.618,
      "step": 45575
    },
    {
      "epoch": 0.7095985739528439,
      "grad_norm": 2.7266523838043213,
      "learning_rate": 3.056980383802297e-07,
      "loss": 0.8774,
      "step": 45580
    },
    {
      "epoch": 0.7096764149548133,
      "grad_norm": 4.741499423980713,
      "learning_rate": 3.0561609937562474e-07,
      "loss": 0.7003,
      "step": 45585
    },
    {
      "epoch": 0.7097542559567827,
      "grad_norm": 3.419884204864502,
      "learning_rate": 3.055341603710198e-07,
      "loss": 0.7541,
      "step": 45590
    },
    {
      "epoch": 0.709832096958752,
      "grad_norm": 3.946743965148926,
      "learning_rate": 3.0545222136641484e-07,
      "loss": 0.7555,
      "step": 45595
    },
    {
      "epoch": 0.7099099379607214,
      "grad_norm": 2.75134539604187,
      "learning_rate": 3.0537028236180986e-07,
      "loss": 0.7112,
      "step": 45600
    },
    {
      "epoch": 0.7099877789626908,
      "grad_norm": 3.9646244049072266,
      "learning_rate": 3.052883433572049e-07,
      "loss": 0.8283,
      "step": 45605
    },
    {
      "epoch": 0.7100656199646602,
      "grad_norm": 3.7716445922851562,
      "learning_rate": 3.052064043525999e-07,
      "loss": 0.7225,
      "step": 45610
    },
    {
      "epoch": 0.7101434609666296,
      "grad_norm": 3.8970305919647217,
      "learning_rate": 3.05124465347995e-07,
      "loss": 0.649,
      "step": 45615
    },
    {
      "epoch": 0.7102213019685989,
      "grad_norm": 5.1552042961120605,
      "learning_rate": 3.0504252634338995e-07,
      "loss": 0.721,
      "step": 45620
    },
    {
      "epoch": 0.7102991429705683,
      "grad_norm": 3.048236131668091,
      "learning_rate": 3.0496058733878497e-07,
      "loss": 0.6965,
      "step": 45625
    },
    {
      "epoch": 0.7103769839725377,
      "grad_norm": 3.584634780883789,
      "learning_rate": 3.0487864833418005e-07,
      "loss": 0.6863,
      "step": 45630
    },
    {
      "epoch": 0.7104548249745071,
      "grad_norm": 4.55698823928833,
      "learning_rate": 3.0479670932957507e-07,
      "loss": 0.7694,
      "step": 45635
    },
    {
      "epoch": 0.7105326659764765,
      "grad_norm": 6.706613540649414,
      "learning_rate": 3.0471477032497004e-07,
      "loss": 0.7666,
      "step": 45640
    },
    {
      "epoch": 0.7106105069784459,
      "grad_norm": 3.1434645652770996,
      "learning_rate": 3.046328313203651e-07,
      "loss": 0.6718,
      "step": 45645
    },
    {
      "epoch": 0.7106883479804152,
      "grad_norm": 4.4039506912231445,
      "learning_rate": 3.0455089231576014e-07,
      "loss": 0.7012,
      "step": 45650
    },
    {
      "epoch": 0.7107661889823845,
      "grad_norm": 3.2558979988098145,
      "learning_rate": 3.044689533111552e-07,
      "loss": 0.8406,
      "step": 45655
    },
    {
      "epoch": 0.7108440299843539,
      "grad_norm": 4.625216007232666,
      "learning_rate": 3.043870143065502e-07,
      "loss": 0.733,
      "step": 45660
    },
    {
      "epoch": 0.7109218709863233,
      "grad_norm": 3.4232177734375,
      "learning_rate": 3.043050753019452e-07,
      "loss": 0.82,
      "step": 45665
    },
    {
      "epoch": 0.7109997119882927,
      "grad_norm": 2.6679632663726807,
      "learning_rate": 3.042231362973403e-07,
      "loss": 0.6596,
      "step": 45670
    },
    {
      "epoch": 0.7110775529902621,
      "grad_norm": 5.6992316246032715,
      "learning_rate": 3.041411972927353e-07,
      "loss": 0.8607,
      "step": 45675
    },
    {
      "epoch": 0.7111553939922315,
      "grad_norm": 3.129899501800537,
      "learning_rate": 3.0405925828813027e-07,
      "loss": 0.8321,
      "step": 45680
    },
    {
      "epoch": 0.7112332349942009,
      "grad_norm": 3.2444095611572266,
      "learning_rate": 3.0397731928352534e-07,
      "loss": 0.6867,
      "step": 45685
    },
    {
      "epoch": 0.7113110759961703,
      "grad_norm": 5.8682050704956055,
      "learning_rate": 3.0389538027892036e-07,
      "loss": 0.6897,
      "step": 45690
    },
    {
      "epoch": 0.7113889169981396,
      "grad_norm": 3.832493782043457,
      "learning_rate": 3.038134412743154e-07,
      "loss": 0.7371,
      "step": 45695
    },
    {
      "epoch": 0.711466758000109,
      "grad_norm": 3.299806833267212,
      "learning_rate": 3.037315022697104e-07,
      "loss": 0.6669,
      "step": 45700
    },
    {
      "epoch": 0.7115445990020783,
      "grad_norm": 3.7874245643615723,
      "learning_rate": 3.0364956326510543e-07,
      "loss": 0.6957,
      "step": 45705
    },
    {
      "epoch": 0.7116224400040477,
      "grad_norm": 4.356656074523926,
      "learning_rate": 3.0356762426050045e-07,
      "loss": 0.7478,
      "step": 45710
    },
    {
      "epoch": 0.7117002810060171,
      "grad_norm": 10.624296188354492,
      "learning_rate": 3.034856852558955e-07,
      "loss": 0.7141,
      "step": 45715
    },
    {
      "epoch": 0.7117781220079865,
      "grad_norm": 3.522219657897949,
      "learning_rate": 3.034037462512905e-07,
      "loss": 0.6995,
      "step": 45720
    },
    {
      "epoch": 0.7118559630099559,
      "grad_norm": 3.2723617553710938,
      "learning_rate": 3.0332180724668557e-07,
      "loss": 0.7557,
      "step": 45725
    },
    {
      "epoch": 0.7119338040119252,
      "grad_norm": 4.269672870635986,
      "learning_rate": 3.032398682420806e-07,
      "loss": 0.6976,
      "step": 45730
    },
    {
      "epoch": 0.7120116450138946,
      "grad_norm": 3.1820530891418457,
      "learning_rate": 3.031579292374756e-07,
      "loss": 0.6657,
      "step": 45735
    },
    {
      "epoch": 0.712089486015864,
      "grad_norm": 3.2613580226898193,
      "learning_rate": 3.0307599023287064e-07,
      "loss": 0.7293,
      "step": 45740
    },
    {
      "epoch": 0.7121673270178334,
      "grad_norm": 2.8417999744415283,
      "learning_rate": 3.0299405122826566e-07,
      "loss": 0.6752,
      "step": 45745
    },
    {
      "epoch": 0.7122451680198028,
      "grad_norm": 3.836299180984497,
      "learning_rate": 3.029121122236607e-07,
      "loss": 0.7937,
      "step": 45750
    },
    {
      "epoch": 0.7123230090217721,
      "grad_norm": 6.739011287689209,
      "learning_rate": 3.0283017321905575e-07,
      "loss": 0.64,
      "step": 45755
    },
    {
      "epoch": 0.7124008500237415,
      "grad_norm": 3.29705548286438,
      "learning_rate": 3.027482342144507e-07,
      "loss": 0.7242,
      "step": 45760
    },
    {
      "epoch": 0.7124786910257109,
      "grad_norm": 6.242274284362793,
      "learning_rate": 3.0266629520984575e-07,
      "loss": 0.7302,
      "step": 45765
    },
    {
      "epoch": 0.7125565320276802,
      "grad_norm": 8.750752449035645,
      "learning_rate": 3.025843562052408e-07,
      "loss": 0.5918,
      "step": 45770
    },
    {
      "epoch": 0.7126343730296496,
      "grad_norm": 3.6574418544769287,
      "learning_rate": 3.0250241720063584e-07,
      "loss": 0.744,
      "step": 45775
    },
    {
      "epoch": 0.712712214031619,
      "grad_norm": 4.188930988311768,
      "learning_rate": 3.024204781960308e-07,
      "loss": 0.7084,
      "step": 45780
    },
    {
      "epoch": 0.7127900550335884,
      "grad_norm": 4.008793830871582,
      "learning_rate": 3.023385391914259e-07,
      "loss": 0.7681,
      "step": 45785
    },
    {
      "epoch": 0.7128678960355578,
      "grad_norm": 4.1495842933654785,
      "learning_rate": 3.022566001868209e-07,
      "loss": 0.8177,
      "step": 45790
    },
    {
      "epoch": 0.7129457370375272,
      "grad_norm": 4.5191450119018555,
      "learning_rate": 3.02174661182216e-07,
      "loss": 0.738,
      "step": 45795
    },
    {
      "epoch": 0.7130235780394966,
      "grad_norm": 2.965454578399658,
      "learning_rate": 3.0209272217761095e-07,
      "loss": 0.6721,
      "step": 45800
    },
    {
      "epoch": 0.713101419041466,
      "grad_norm": 4.043234348297119,
      "learning_rate": 3.02010783173006e-07,
      "loss": 0.8084,
      "step": 45805
    },
    {
      "epoch": 0.7131792600434352,
      "grad_norm": 3.690993309020996,
      "learning_rate": 3.0192884416840105e-07,
      "loss": 0.7557,
      "step": 45810
    },
    {
      "epoch": 0.7132571010454046,
      "grad_norm": 2.961923360824585,
      "learning_rate": 3.0184690516379607e-07,
      "loss": 0.8667,
      "step": 45815
    },
    {
      "epoch": 0.713334942047374,
      "grad_norm": 6.575785160064697,
      "learning_rate": 3.0176496615919104e-07,
      "loss": 0.8457,
      "step": 45820
    },
    {
      "epoch": 0.7134127830493434,
      "grad_norm": 6.5424017906188965,
      "learning_rate": 3.016830271545861e-07,
      "loss": 0.7916,
      "step": 45825
    },
    {
      "epoch": 0.7134906240513128,
      "grad_norm": 3.9216175079345703,
      "learning_rate": 3.0160108814998114e-07,
      "loss": 0.7161,
      "step": 45830
    },
    {
      "epoch": 0.7135684650532822,
      "grad_norm": 4.492792129516602,
      "learning_rate": 3.0151914914537616e-07,
      "loss": 0.72,
      "step": 45835
    },
    {
      "epoch": 0.7136463060552516,
      "grad_norm": 3.7698237895965576,
      "learning_rate": 3.014372101407712e-07,
      "loss": 0.8048,
      "step": 45840
    },
    {
      "epoch": 0.7137241470572209,
      "grad_norm": 4.641737937927246,
      "learning_rate": 3.013552711361662e-07,
      "loss": 0.8014,
      "step": 45845
    },
    {
      "epoch": 0.7138019880591903,
      "grad_norm": 6.043346405029297,
      "learning_rate": 3.012733321315613e-07,
      "loss": 0.7229,
      "step": 45850
    },
    {
      "epoch": 0.7138798290611597,
      "grad_norm": 2.8968422412872314,
      "learning_rate": 3.011913931269563e-07,
      "loss": 0.8128,
      "step": 45855
    },
    {
      "epoch": 0.713957670063129,
      "grad_norm": 3.5968289375305176,
      "learning_rate": 3.0110945412235127e-07,
      "loss": 0.7728,
      "step": 45860
    },
    {
      "epoch": 0.7140355110650984,
      "grad_norm": 3.561190366744995,
      "learning_rate": 3.0102751511774634e-07,
      "loss": 0.6779,
      "step": 45865
    },
    {
      "epoch": 0.7141133520670678,
      "grad_norm": 2.897341728210449,
      "learning_rate": 3.0094557611314136e-07,
      "loss": 0.6611,
      "step": 45870
    },
    {
      "epoch": 0.7141911930690372,
      "grad_norm": 2.8038854598999023,
      "learning_rate": 3.008636371085364e-07,
      "loss": 0.7484,
      "step": 45875
    },
    {
      "epoch": 0.7142690340710066,
      "grad_norm": 2.690054178237915,
      "learning_rate": 3.0078169810393146e-07,
      "loss": 0.757,
      "step": 45880
    },
    {
      "epoch": 0.7143468750729759,
      "grad_norm": 3.0343995094299316,
      "learning_rate": 3.0069975909932643e-07,
      "loss": 0.728,
      "step": 45885
    },
    {
      "epoch": 0.7144247160749453,
      "grad_norm": 5.852138042449951,
      "learning_rate": 3.0061782009472145e-07,
      "loss": 0.7359,
      "step": 45890
    },
    {
      "epoch": 0.7145025570769147,
      "grad_norm": 3.3477582931518555,
      "learning_rate": 3.0053588109011653e-07,
      "loss": 0.7077,
      "step": 45895
    },
    {
      "epoch": 0.7145803980788841,
      "grad_norm": 5.4564900398254395,
      "learning_rate": 3.0045394208551155e-07,
      "loss": 0.7161,
      "step": 45900
    },
    {
      "epoch": 0.7146582390808535,
      "grad_norm": 3.766054391860962,
      "learning_rate": 3.003720030809065e-07,
      "loss": 0.7514,
      "step": 45905
    },
    {
      "epoch": 0.7147360800828229,
      "grad_norm": 2.4148316383361816,
      "learning_rate": 3.002900640763016e-07,
      "loss": 0.6184,
      "step": 45910
    },
    {
      "epoch": 0.7148139210847922,
      "grad_norm": 3.8145718574523926,
      "learning_rate": 3.002081250716966e-07,
      "loss": 0.7312,
      "step": 45915
    },
    {
      "epoch": 0.7148917620867615,
      "grad_norm": 6.638522148132324,
      "learning_rate": 3.001261860670917e-07,
      "loss": 0.6919,
      "step": 45920
    },
    {
      "epoch": 0.7149696030887309,
      "grad_norm": 4.60577392578125,
      "learning_rate": 3.0004424706248666e-07,
      "loss": 0.6918,
      "step": 45925
    },
    {
      "epoch": 0.7150474440907003,
      "grad_norm": 3.5046634674072266,
      "learning_rate": 2.999623080578817e-07,
      "loss": 0.675,
      "step": 45930
    },
    {
      "epoch": 0.7151252850926697,
      "grad_norm": 3.8361122608184814,
      "learning_rate": 2.9988036905327676e-07,
      "loss": 0.748,
      "step": 45935
    },
    {
      "epoch": 0.7152031260946391,
      "grad_norm": 12.049629211425781,
      "learning_rate": 2.997984300486718e-07,
      "loss": 0.7585,
      "step": 45940
    },
    {
      "epoch": 0.7152809670966085,
      "grad_norm": 4.348023414611816,
      "learning_rate": 2.9971649104406675e-07,
      "loss": 0.7237,
      "step": 45945
    },
    {
      "epoch": 0.7153588080985779,
      "grad_norm": 4.42105770111084,
      "learning_rate": 2.996345520394618e-07,
      "loss": 0.5378,
      "step": 45950
    },
    {
      "epoch": 0.7154366491005473,
      "grad_norm": 5.288759708404541,
      "learning_rate": 2.9955261303485684e-07,
      "loss": 0.9099,
      "step": 45955
    },
    {
      "epoch": 0.7155144901025166,
      "grad_norm": 3.6096267700195312,
      "learning_rate": 2.9947067403025187e-07,
      "loss": 0.8167,
      "step": 45960
    },
    {
      "epoch": 0.715592331104486,
      "grad_norm": 3.177370548248291,
      "learning_rate": 2.993887350256469e-07,
      "loss": 0.821,
      "step": 45965
    },
    {
      "epoch": 0.7156701721064553,
      "grad_norm": 4.759136199951172,
      "learning_rate": 2.993067960210419e-07,
      "loss": 0.7975,
      "step": 45970
    },
    {
      "epoch": 0.7157480131084247,
      "grad_norm": 4.752901554107666,
      "learning_rate": 2.99224857016437e-07,
      "loss": 0.7157,
      "step": 45975
    },
    {
      "epoch": 0.7158258541103941,
      "grad_norm": 3.617866277694702,
      "learning_rate": 2.99142918011832e-07,
      "loss": 0.687,
      "step": 45980
    },
    {
      "epoch": 0.7159036951123635,
      "grad_norm": 3.3718085289001465,
      "learning_rate": 2.99060979007227e-07,
      "loss": 0.8469,
      "step": 45985
    },
    {
      "epoch": 0.7159815361143329,
      "grad_norm": 3.502880096435547,
      "learning_rate": 2.9897904000262205e-07,
      "loss": 0.6264,
      "step": 45990
    },
    {
      "epoch": 0.7160593771163022,
      "grad_norm": 3.4871578216552734,
      "learning_rate": 2.9889710099801707e-07,
      "loss": 0.7109,
      "step": 45995
    },
    {
      "epoch": 0.7161372181182716,
      "grad_norm": 3.1216280460357666,
      "learning_rate": 2.988151619934121e-07,
      "loss": 0.7853,
      "step": 46000
    },
    {
      "epoch": 0.716215059120241,
      "grad_norm": 8.77890682220459,
      "learning_rate": 2.987332229888071e-07,
      "loss": 0.6686,
      "step": 46005
    },
    {
      "epoch": 0.7162929001222104,
      "grad_norm": 3.2109861373901367,
      "learning_rate": 2.9865128398420214e-07,
      "loss": 0.8654,
      "step": 46010
    },
    {
      "epoch": 0.7163707411241798,
      "grad_norm": 4.989051818847656,
      "learning_rate": 2.9856934497959716e-07,
      "loss": 0.8135,
      "step": 46015
    },
    {
      "epoch": 0.7164485821261491,
      "grad_norm": 4.763101100921631,
      "learning_rate": 2.9848740597499223e-07,
      "loss": 0.7318,
      "step": 46020
    },
    {
      "epoch": 0.7165264231281185,
      "grad_norm": 4.364039421081543,
      "learning_rate": 2.984054669703872e-07,
      "loss": 0.6428,
      "step": 46025
    },
    {
      "epoch": 0.7166042641300879,
      "grad_norm": 3.4445176124572754,
      "learning_rate": 2.983235279657822e-07,
      "loss": 0.7721,
      "step": 46030
    },
    {
      "epoch": 0.7166821051320572,
      "grad_norm": 7.8140435218811035,
      "learning_rate": 2.982415889611773e-07,
      "loss": 0.7245,
      "step": 46035
    },
    {
      "epoch": 0.7167599461340266,
      "grad_norm": 5.69026517868042,
      "learning_rate": 2.981596499565723e-07,
      "loss": 0.6912,
      "step": 46040
    },
    {
      "epoch": 0.716837787135996,
      "grad_norm": 4.802468299865723,
      "learning_rate": 2.9807771095196734e-07,
      "loss": 0.7006,
      "step": 46045
    },
    {
      "epoch": 0.7169156281379654,
      "grad_norm": 4.980616569519043,
      "learning_rate": 2.9799577194736237e-07,
      "loss": 0.6533,
      "step": 46050
    },
    {
      "epoch": 0.7169934691399348,
      "grad_norm": 4.377559661865234,
      "learning_rate": 2.979138329427574e-07,
      "loss": 0.7744,
      "step": 46055
    },
    {
      "epoch": 0.7170713101419042,
      "grad_norm": 3.531216621398926,
      "learning_rate": 2.9783189393815246e-07,
      "loss": 0.8196,
      "step": 46060
    },
    {
      "epoch": 0.7171491511438736,
      "grad_norm": 5.315894603729248,
      "learning_rate": 2.9774995493354743e-07,
      "loss": 0.6904,
      "step": 46065
    },
    {
      "epoch": 0.717226992145843,
      "grad_norm": 5.120900630950928,
      "learning_rate": 2.9766801592894245e-07,
      "loss": 0.7467,
      "step": 46070
    },
    {
      "epoch": 0.7173048331478122,
      "grad_norm": 6.201507568359375,
      "learning_rate": 2.9758607692433753e-07,
      "loss": 0.7477,
      "step": 46075
    },
    {
      "epoch": 0.7173826741497816,
      "grad_norm": 3.2009425163269043,
      "learning_rate": 2.9750413791973255e-07,
      "loss": 0.8074,
      "step": 46080
    },
    {
      "epoch": 0.717460515151751,
      "grad_norm": 6.0363311767578125,
      "learning_rate": 2.974221989151275e-07,
      "loss": 0.8226,
      "step": 46085
    },
    {
      "epoch": 0.7175383561537204,
      "grad_norm": 2.716510772705078,
      "learning_rate": 2.973402599105226e-07,
      "loss": 0.7605,
      "step": 46090
    },
    {
      "epoch": 0.7176161971556898,
      "grad_norm": 7.154038906097412,
      "learning_rate": 2.972583209059176e-07,
      "loss": 0.6115,
      "step": 46095
    },
    {
      "epoch": 0.7176940381576592,
      "grad_norm": 5.40645170211792,
      "learning_rate": 2.971763819013127e-07,
      "loss": 0.6903,
      "step": 46100
    },
    {
      "epoch": 0.7177718791596286,
      "grad_norm": 2.9864797592163086,
      "learning_rate": 2.9709444289670766e-07,
      "loss": 0.7875,
      "step": 46105
    },
    {
      "epoch": 0.7178497201615979,
      "grad_norm": 3.673088788986206,
      "learning_rate": 2.970125038921027e-07,
      "loss": 0.8726,
      "step": 46110
    },
    {
      "epoch": 0.7179275611635673,
      "grad_norm": 3.035494327545166,
      "learning_rate": 2.9693056488749776e-07,
      "loss": 0.7731,
      "step": 46115
    },
    {
      "epoch": 0.7180054021655367,
      "grad_norm": 3.6323564052581787,
      "learning_rate": 2.968486258828928e-07,
      "loss": 0.6061,
      "step": 46120
    },
    {
      "epoch": 0.718083243167506,
      "grad_norm": 3.6712989807128906,
      "learning_rate": 2.9676668687828775e-07,
      "loss": 0.919,
      "step": 46125
    },
    {
      "epoch": 0.7181610841694754,
      "grad_norm": 3.659567356109619,
      "learning_rate": 2.966847478736828e-07,
      "loss": 0.8989,
      "step": 46130
    },
    {
      "epoch": 0.7182389251714448,
      "grad_norm": 3.6912240982055664,
      "learning_rate": 2.9660280886907784e-07,
      "loss": 0.678,
      "step": 46135
    },
    {
      "epoch": 0.7183167661734142,
      "grad_norm": 3.3548033237457275,
      "learning_rate": 2.9652086986447287e-07,
      "loss": 0.7202,
      "step": 46140
    },
    {
      "epoch": 0.7183946071753836,
      "grad_norm": 4.072553634643555,
      "learning_rate": 2.964389308598679e-07,
      "loss": 0.8447,
      "step": 46145
    },
    {
      "epoch": 0.7184724481773529,
      "grad_norm": 4.41408109664917,
      "learning_rate": 2.963569918552629e-07,
      "loss": 0.7064,
      "step": 46150
    },
    {
      "epoch": 0.7185502891793223,
      "grad_norm": 2.9772348403930664,
      "learning_rate": 2.96275052850658e-07,
      "loss": 0.782,
      "step": 46155
    },
    {
      "epoch": 0.7186281301812917,
      "grad_norm": 5.656728744506836,
      "learning_rate": 2.96193113846053e-07,
      "loss": 0.7285,
      "step": 46160
    },
    {
      "epoch": 0.7187059711832611,
      "grad_norm": 3.468104600906372,
      "learning_rate": 2.96111174841448e-07,
      "loss": 0.6801,
      "step": 46165
    },
    {
      "epoch": 0.7187838121852305,
      "grad_norm": 3.827741861343384,
      "learning_rate": 2.9602923583684305e-07,
      "loss": 0.8037,
      "step": 46170
    },
    {
      "epoch": 0.7188616531871999,
      "grad_norm": 3.1488895416259766,
      "learning_rate": 2.9594729683223807e-07,
      "loss": 0.7624,
      "step": 46175
    },
    {
      "epoch": 0.7189394941891692,
      "grad_norm": 3.471668243408203,
      "learning_rate": 2.958653578276331e-07,
      "loss": 0.6994,
      "step": 46180
    },
    {
      "epoch": 0.7190173351911385,
      "grad_norm": 3.3940954208374023,
      "learning_rate": 2.957834188230281e-07,
      "loss": 0.8313,
      "step": 46185
    },
    {
      "epoch": 0.7190951761931079,
      "grad_norm": 10.389409065246582,
      "learning_rate": 2.9570147981842314e-07,
      "loss": 0.7376,
      "step": 46190
    },
    {
      "epoch": 0.7191730171950773,
      "grad_norm": 5.124993324279785,
      "learning_rate": 2.9561954081381816e-07,
      "loss": 0.7647,
      "step": 46195
    },
    {
      "epoch": 0.7192508581970467,
      "grad_norm": 3.0172483921051025,
      "learning_rate": 2.9553760180921323e-07,
      "loss": 0.7225,
      "step": 46200
    },
    {
      "epoch": 0.7193286991990161,
      "grad_norm": 3.9033315181732178,
      "learning_rate": 2.9545566280460826e-07,
      "loss": 0.6786,
      "step": 46205
    },
    {
      "epoch": 0.7194065402009855,
      "grad_norm": 2.9832382202148438,
      "learning_rate": 2.953737238000032e-07,
      "loss": 0.6426,
      "step": 46210
    },
    {
      "epoch": 0.7194843812029549,
      "grad_norm": 3.657440423965454,
      "learning_rate": 2.952917847953983e-07,
      "loss": 0.6299,
      "step": 46215
    },
    {
      "epoch": 0.7195622222049243,
      "grad_norm": 2.5976953506469727,
      "learning_rate": 2.952098457907933e-07,
      "loss": 0.6277,
      "step": 46220
    },
    {
      "epoch": 0.7196400632068936,
      "grad_norm": 3.467040777206421,
      "learning_rate": 2.951279067861884e-07,
      "loss": 0.7984,
      "step": 46225
    },
    {
      "epoch": 0.719717904208863,
      "grad_norm": 5.004181861877441,
      "learning_rate": 2.9504596778158337e-07,
      "loss": 0.7801,
      "step": 46230
    },
    {
      "epoch": 0.7197957452108323,
      "grad_norm": 2.648336410522461,
      "learning_rate": 2.949640287769784e-07,
      "loss": 0.7237,
      "step": 46235
    },
    {
      "epoch": 0.7198735862128017,
      "grad_norm": 3.476382255554199,
      "learning_rate": 2.9488208977237346e-07,
      "loss": 0.7563,
      "step": 46240
    },
    {
      "epoch": 0.7199514272147711,
      "grad_norm": 3.924221992492676,
      "learning_rate": 2.948001507677685e-07,
      "loss": 0.7305,
      "step": 46245
    },
    {
      "epoch": 0.7200292682167405,
      "grad_norm": 2.8663086891174316,
      "learning_rate": 2.9471821176316345e-07,
      "loss": 0.7111,
      "step": 46250
    },
    {
      "epoch": 0.7201071092187099,
      "grad_norm": 3.1556224822998047,
      "learning_rate": 2.9463627275855853e-07,
      "loss": 0.7834,
      "step": 46255
    },
    {
      "epoch": 0.7201849502206792,
      "grad_norm": 4.04901123046875,
      "learning_rate": 2.9455433375395355e-07,
      "loss": 0.7771,
      "step": 46260
    },
    {
      "epoch": 0.7202627912226486,
      "grad_norm": 4.539096832275391,
      "learning_rate": 2.9447239474934857e-07,
      "loss": 0.6241,
      "step": 46265
    },
    {
      "epoch": 0.720340632224618,
      "grad_norm": 4.580163478851318,
      "learning_rate": 2.943904557447436e-07,
      "loss": 0.7546,
      "step": 46270
    },
    {
      "epoch": 0.7204184732265874,
      "grad_norm": 3.6383652687072754,
      "learning_rate": 2.943085167401386e-07,
      "loss": 0.6945,
      "step": 46275
    },
    {
      "epoch": 0.7204963142285568,
      "grad_norm": 5.334415435791016,
      "learning_rate": 2.942265777355337e-07,
      "loss": 0.8102,
      "step": 46280
    },
    {
      "epoch": 0.7205741552305261,
      "grad_norm": 3.5706520080566406,
      "learning_rate": 2.941446387309287e-07,
      "loss": 0.6862,
      "step": 46285
    },
    {
      "epoch": 0.7206519962324955,
      "grad_norm": 4.339644908905029,
      "learning_rate": 2.940626997263237e-07,
      "loss": 0.7701,
      "step": 46290
    },
    {
      "epoch": 0.7207298372344649,
      "grad_norm": 6.153683662414551,
      "learning_rate": 2.9398076072171876e-07,
      "loss": 0.6716,
      "step": 46295
    },
    {
      "epoch": 0.7208076782364342,
      "grad_norm": 3.4239673614501953,
      "learning_rate": 2.938988217171138e-07,
      "loss": 0.7866,
      "step": 46300
    },
    {
      "epoch": 0.7208855192384036,
      "grad_norm": 4.248059272766113,
      "learning_rate": 2.938168827125088e-07,
      "loss": 0.6646,
      "step": 46305
    },
    {
      "epoch": 0.720963360240373,
      "grad_norm": 2.450856924057007,
      "learning_rate": 2.937349437079038e-07,
      "loss": 0.6932,
      "step": 46310
    },
    {
      "epoch": 0.7210412012423424,
      "grad_norm": 5.8886399269104,
      "learning_rate": 2.9365300470329884e-07,
      "loss": 0.7495,
      "step": 46315
    },
    {
      "epoch": 0.7211190422443118,
      "grad_norm": 19.727489471435547,
      "learning_rate": 2.9357106569869387e-07,
      "loss": 0.7516,
      "step": 46320
    },
    {
      "epoch": 0.7211968832462812,
      "grad_norm": 6.112648963928223,
      "learning_rate": 2.9348912669408894e-07,
      "loss": 0.72,
      "step": 46325
    },
    {
      "epoch": 0.7212747242482506,
      "grad_norm": 3.841801881790161,
      "learning_rate": 2.934071876894839e-07,
      "loss": 0.6678,
      "step": 46330
    },
    {
      "epoch": 0.7213525652502198,
      "grad_norm": 3.1759486198425293,
      "learning_rate": 2.9332524868487893e-07,
      "loss": 0.6912,
      "step": 46335
    },
    {
      "epoch": 0.7214304062521892,
      "grad_norm": 4.838475704193115,
      "learning_rate": 2.93243309680274e-07,
      "loss": 0.767,
      "step": 46340
    },
    {
      "epoch": 0.7215082472541586,
      "grad_norm": 3.322343349456787,
      "learning_rate": 2.9316137067566903e-07,
      "loss": 0.646,
      "step": 46345
    },
    {
      "epoch": 0.721586088256128,
      "grad_norm": 3.1822056770324707,
      "learning_rate": 2.9307943167106405e-07,
      "loss": 0.6525,
      "step": 46350
    },
    {
      "epoch": 0.7216639292580974,
      "grad_norm": 6.648345947265625,
      "learning_rate": 2.9299749266645907e-07,
      "loss": 0.8776,
      "step": 46355
    },
    {
      "epoch": 0.7217417702600668,
      "grad_norm": 3.5682308673858643,
      "learning_rate": 2.929155536618541e-07,
      "loss": 0.8254,
      "step": 46360
    },
    {
      "epoch": 0.7218196112620362,
      "grad_norm": 4.291876316070557,
      "learning_rate": 2.9283361465724917e-07,
      "loss": 0.6616,
      "step": 46365
    },
    {
      "epoch": 0.7218974522640056,
      "grad_norm": 4.587896823883057,
      "learning_rate": 2.9275167565264414e-07,
      "loss": 0.8567,
      "step": 46370
    },
    {
      "epoch": 0.7219752932659749,
      "grad_norm": 4.53436803817749,
      "learning_rate": 2.9266973664803916e-07,
      "loss": 0.7347,
      "step": 46375
    },
    {
      "epoch": 0.7220531342679443,
      "grad_norm": 5.517923831939697,
      "learning_rate": 2.9258779764343424e-07,
      "loss": 0.7014,
      "step": 46380
    },
    {
      "epoch": 0.7221309752699137,
      "grad_norm": 6.124557971954346,
      "learning_rate": 2.9250585863882926e-07,
      "loss": 0.7101,
      "step": 46385
    },
    {
      "epoch": 0.722208816271883,
      "grad_norm": 3.0672316551208496,
      "learning_rate": 2.9242391963422423e-07,
      "loss": 0.7052,
      "step": 46390
    },
    {
      "epoch": 0.7222866572738524,
      "grad_norm": 3.167468309402466,
      "learning_rate": 2.923419806296193e-07,
      "loss": 0.6457,
      "step": 46395
    },
    {
      "epoch": 0.7223644982758218,
      "grad_norm": 11.265911102294922,
      "learning_rate": 2.922600416250143e-07,
      "loss": 0.8083,
      "step": 46400
    },
    {
      "epoch": 0.7224423392777912,
      "grad_norm": 3.510064125061035,
      "learning_rate": 2.921781026204094e-07,
      "loss": 0.9433,
      "step": 46405
    },
    {
      "epoch": 0.7225201802797606,
      "grad_norm": 3.9093453884124756,
      "learning_rate": 2.9209616361580437e-07,
      "loss": 0.6538,
      "step": 46410
    },
    {
      "epoch": 0.7225980212817299,
      "grad_norm": 4.731550216674805,
      "learning_rate": 2.920142246111994e-07,
      "loss": 0.8053,
      "step": 46415
    },
    {
      "epoch": 0.7226758622836993,
      "grad_norm": 2.790996551513672,
      "learning_rate": 2.9193228560659446e-07,
      "loss": 0.9325,
      "step": 46420
    },
    {
      "epoch": 0.7227537032856687,
      "grad_norm": 4.062405586242676,
      "learning_rate": 2.918503466019895e-07,
      "loss": 0.7221,
      "step": 46425
    },
    {
      "epoch": 0.7228315442876381,
      "grad_norm": 2.549067497253418,
      "learning_rate": 2.9176840759738445e-07,
      "loss": 0.7381,
      "step": 46430
    },
    {
      "epoch": 0.7229093852896075,
      "grad_norm": 4.98679780960083,
      "learning_rate": 2.9168646859277953e-07,
      "loss": 0.7358,
      "step": 46435
    },
    {
      "epoch": 0.7229872262915769,
      "grad_norm": 4.959805488586426,
      "learning_rate": 2.9160452958817455e-07,
      "loss": 0.7473,
      "step": 46440
    },
    {
      "epoch": 0.7230650672935462,
      "grad_norm": 3.958557605743408,
      "learning_rate": 2.915225905835696e-07,
      "loss": 0.7163,
      "step": 46445
    },
    {
      "epoch": 0.7231429082955155,
      "grad_norm": 4.476954936981201,
      "learning_rate": 2.914406515789646e-07,
      "loss": 0.6111,
      "step": 46450
    },
    {
      "epoch": 0.7232207492974849,
      "grad_norm": 4.612060070037842,
      "learning_rate": 2.913587125743596e-07,
      "loss": 0.6159,
      "step": 46455
    },
    {
      "epoch": 0.7232985902994543,
      "grad_norm": 4.709410667419434,
      "learning_rate": 2.9127677356975464e-07,
      "loss": 0.6905,
      "step": 46460
    },
    {
      "epoch": 0.7233764313014237,
      "grad_norm": 3.035778045654297,
      "learning_rate": 2.911948345651497e-07,
      "loss": 0.7772,
      "step": 46465
    },
    {
      "epoch": 0.7234542723033931,
      "grad_norm": 2.7283544540405273,
      "learning_rate": 2.911128955605447e-07,
      "loss": 0.7598,
      "step": 46470
    },
    {
      "epoch": 0.7235321133053625,
      "grad_norm": 11.397520065307617,
      "learning_rate": 2.9103095655593976e-07,
      "loss": 0.6694,
      "step": 46475
    },
    {
      "epoch": 0.7236099543073319,
      "grad_norm": 3.0994932651519775,
      "learning_rate": 2.909490175513348e-07,
      "loss": 0.7747,
      "step": 46480
    },
    {
      "epoch": 0.7236877953093013,
      "grad_norm": 2.32393741607666,
      "learning_rate": 2.908670785467298e-07,
      "loss": 0.7043,
      "step": 46485
    },
    {
      "epoch": 0.7237656363112706,
      "grad_norm": 6.357470512390137,
      "learning_rate": 2.907851395421248e-07,
      "loss": 0.6101,
      "step": 46490
    },
    {
      "epoch": 0.72384347731324,
      "grad_norm": 4.971627712249756,
      "learning_rate": 2.9070320053751985e-07,
      "loss": 0.811,
      "step": 46495
    },
    {
      "epoch": 0.7239213183152093,
      "grad_norm": 3.538233995437622,
      "learning_rate": 2.9062126153291487e-07,
      "loss": 0.6834,
      "step": 46500
    },
    {
      "epoch": 0.7239991593171787,
      "grad_norm": 2.4038281440734863,
      "learning_rate": 2.9053932252830994e-07,
      "loss": 0.7625,
      "step": 46505
    },
    {
      "epoch": 0.7240770003191481,
      "grad_norm": 2.7993245124816895,
      "learning_rate": 2.9045738352370496e-07,
      "loss": 0.6383,
      "step": 46510
    },
    {
      "epoch": 0.7241548413211175,
      "grad_norm": 3.45759916305542,
      "learning_rate": 2.9037544451909993e-07,
      "loss": 0.7737,
      "step": 46515
    },
    {
      "epoch": 0.7242326823230869,
      "grad_norm": 4.645711898803711,
      "learning_rate": 2.90293505514495e-07,
      "loss": 0.6953,
      "step": 46520
    },
    {
      "epoch": 0.7243105233250562,
      "grad_norm": 4.082793235778809,
      "learning_rate": 2.9021156650989003e-07,
      "loss": 0.8654,
      "step": 46525
    },
    {
      "epoch": 0.7243883643270256,
      "grad_norm": 2.9457767009735107,
      "learning_rate": 2.901296275052851e-07,
      "loss": 0.6895,
      "step": 46530
    },
    {
      "epoch": 0.724466205328995,
      "grad_norm": 3.705721855163574,
      "learning_rate": 2.900476885006801e-07,
      "loss": 0.7616,
      "step": 46535
    },
    {
      "epoch": 0.7245440463309644,
      "grad_norm": 3.081197738647461,
      "learning_rate": 2.899657494960751e-07,
      "loss": 0.737,
      "step": 46540
    },
    {
      "epoch": 0.7246218873329338,
      "grad_norm": 3.2306511402130127,
      "learning_rate": 2.8988381049147017e-07,
      "loss": 0.7575,
      "step": 46545
    },
    {
      "epoch": 0.7246997283349031,
      "grad_norm": 4.005180358886719,
      "learning_rate": 2.898018714868652e-07,
      "loss": 0.7201,
      "step": 46550
    },
    {
      "epoch": 0.7247775693368725,
      "grad_norm": 3.349125623703003,
      "learning_rate": 2.8971993248226016e-07,
      "loss": 0.7439,
      "step": 46555
    },
    {
      "epoch": 0.7248554103388419,
      "grad_norm": 3.493135452270508,
      "learning_rate": 2.8963799347765524e-07,
      "loss": 0.73,
      "step": 46560
    },
    {
      "epoch": 0.7249332513408112,
      "grad_norm": 3.0322208404541016,
      "learning_rate": 2.8955605447305026e-07,
      "loss": 0.6985,
      "step": 46565
    },
    {
      "epoch": 0.7250110923427806,
      "grad_norm": 4.455844879150391,
      "learning_rate": 2.894741154684453e-07,
      "loss": 0.6944,
      "step": 46570
    },
    {
      "epoch": 0.72508893334475,
      "grad_norm": 4.086513519287109,
      "learning_rate": 2.893921764638403e-07,
      "loss": 0.8659,
      "step": 46575
    },
    {
      "epoch": 0.7251667743467194,
      "grad_norm": 3.976155996322632,
      "learning_rate": 2.893102374592353e-07,
      "loss": 0.7463,
      "step": 46580
    },
    {
      "epoch": 0.7252446153486888,
      "grad_norm": 3.7206223011016846,
      "learning_rate": 2.8922829845463035e-07,
      "loss": 0.7261,
      "step": 46585
    },
    {
      "epoch": 0.7253224563506582,
      "grad_norm": 4.797060966491699,
      "learning_rate": 2.891463594500254e-07,
      "loss": 0.7701,
      "step": 46590
    },
    {
      "epoch": 0.7254002973526276,
      "grad_norm": 2.802661180496216,
      "learning_rate": 2.890644204454204e-07,
      "loss": 0.8242,
      "step": 46595
    },
    {
      "epoch": 0.7254781383545968,
      "grad_norm": 3.3675217628479004,
      "learning_rate": 2.8898248144081546e-07,
      "loss": 0.7434,
      "step": 46600
    },
    {
      "epoch": 0.7255559793565662,
      "grad_norm": 3.00331449508667,
      "learning_rate": 2.889005424362105e-07,
      "loss": 0.6314,
      "step": 46605
    },
    {
      "epoch": 0.7256338203585356,
      "grad_norm": 2.977809429168701,
      "learning_rate": 2.888186034316055e-07,
      "loss": 0.6235,
      "step": 46610
    },
    {
      "epoch": 0.725711661360505,
      "grad_norm": 3.598029136657715,
      "learning_rate": 2.8873666442700053e-07,
      "loss": 0.7042,
      "step": 46615
    },
    {
      "epoch": 0.7257895023624744,
      "grad_norm": 2.827414035797119,
      "learning_rate": 2.8865472542239555e-07,
      "loss": 0.6439,
      "step": 46620
    },
    {
      "epoch": 0.7258673433644438,
      "grad_norm": 5.8172173500061035,
      "learning_rate": 2.885727864177906e-07,
      "loss": 0.894,
      "step": 46625
    },
    {
      "epoch": 0.7259451843664132,
      "grad_norm": 4.755241870880127,
      "learning_rate": 2.8849084741318565e-07,
      "loss": 0.7387,
      "step": 46630
    },
    {
      "epoch": 0.7260230253683826,
      "grad_norm": 3.7556514739990234,
      "learning_rate": 2.884089084085806e-07,
      "loss": 0.73,
      "step": 46635
    },
    {
      "epoch": 0.7261008663703519,
      "grad_norm": 5.084848880767822,
      "learning_rate": 2.8832696940397564e-07,
      "loss": 0.6739,
      "step": 46640
    },
    {
      "epoch": 0.7261787073723213,
      "grad_norm": 2.490741729736328,
      "learning_rate": 2.882450303993707e-07,
      "loss": 0.6692,
      "step": 46645
    },
    {
      "epoch": 0.7262565483742907,
      "grad_norm": 5.148022174835205,
      "learning_rate": 2.8816309139476574e-07,
      "loss": 0.7902,
      "step": 46650
    },
    {
      "epoch": 0.72633438937626,
      "grad_norm": 2.9043350219726562,
      "learning_rate": 2.880811523901607e-07,
      "loss": 0.6688,
      "step": 46655
    },
    {
      "epoch": 0.7264122303782294,
      "grad_norm": 3.710983991622925,
      "learning_rate": 2.879992133855558e-07,
      "loss": 0.7981,
      "step": 46660
    },
    {
      "epoch": 0.7264900713801988,
      "grad_norm": 3.5651676654815674,
      "learning_rate": 2.879172743809508e-07,
      "loss": 0.7387,
      "step": 46665
    },
    {
      "epoch": 0.7265679123821682,
      "grad_norm": 4.111935615539551,
      "learning_rate": 2.878353353763459e-07,
      "loss": 0.7855,
      "step": 46670
    },
    {
      "epoch": 0.7266457533841376,
      "grad_norm": 12.296172142028809,
      "learning_rate": 2.8775339637174085e-07,
      "loss": 0.774,
      "step": 46675
    },
    {
      "epoch": 0.7267235943861069,
      "grad_norm": 3.2710492610931396,
      "learning_rate": 2.8767145736713587e-07,
      "loss": 0.7676,
      "step": 46680
    },
    {
      "epoch": 0.7268014353880763,
      "grad_norm": 3.460404396057129,
      "learning_rate": 2.8758951836253094e-07,
      "loss": 0.7878,
      "step": 46685
    },
    {
      "epoch": 0.7268792763900457,
      "grad_norm": 7.249520301818848,
      "learning_rate": 2.8750757935792597e-07,
      "loss": 0.7464,
      "step": 46690
    },
    {
      "epoch": 0.7269571173920151,
      "grad_norm": 3.977527141571045,
      "learning_rate": 2.8742564035332093e-07,
      "loss": 0.7243,
      "step": 46695
    },
    {
      "epoch": 0.7270349583939845,
      "grad_norm": 3.7443528175354004,
      "learning_rate": 2.87343701348716e-07,
      "loss": 0.6487,
      "step": 46700
    },
    {
      "epoch": 0.7271127993959539,
      "grad_norm": 2.538301944732666,
      "learning_rate": 2.8726176234411103e-07,
      "loss": 0.6813,
      "step": 46705
    },
    {
      "epoch": 0.7271906403979232,
      "grad_norm": 3.631068229675293,
      "learning_rate": 2.8717982333950605e-07,
      "loss": 0.7127,
      "step": 46710
    },
    {
      "epoch": 0.7272684813998925,
      "grad_norm": 2.9710469245910645,
      "learning_rate": 2.870978843349011e-07,
      "loss": 0.6641,
      "step": 46715
    },
    {
      "epoch": 0.7273463224018619,
      "grad_norm": 3.8461720943450928,
      "learning_rate": 2.870159453302961e-07,
      "loss": 0.7981,
      "step": 46720
    },
    {
      "epoch": 0.7274241634038313,
      "grad_norm": 2.877857208251953,
      "learning_rate": 2.8693400632569117e-07,
      "loss": 0.7312,
      "step": 46725
    },
    {
      "epoch": 0.7275020044058007,
      "grad_norm": 3.847768545150757,
      "learning_rate": 2.868520673210862e-07,
      "loss": 0.7558,
      "step": 46730
    },
    {
      "epoch": 0.7275798454077701,
      "grad_norm": 4.2333221435546875,
      "learning_rate": 2.8677012831648116e-07,
      "loss": 0.7451,
      "step": 46735
    },
    {
      "epoch": 0.7276576864097395,
      "grad_norm": 3.6807615756988525,
      "learning_rate": 2.8668818931187624e-07,
      "loss": 0.8138,
      "step": 46740
    },
    {
      "epoch": 0.7277355274117089,
      "grad_norm": 5.777658462524414,
      "learning_rate": 2.8660625030727126e-07,
      "loss": 0.6963,
      "step": 46745
    },
    {
      "epoch": 0.7278133684136783,
      "grad_norm": 3.1798999309539795,
      "learning_rate": 2.865243113026663e-07,
      "loss": 0.8197,
      "step": 46750
    },
    {
      "epoch": 0.7278912094156476,
      "grad_norm": 3.646143913269043,
      "learning_rate": 2.864423722980613e-07,
      "loss": 0.7866,
      "step": 46755
    },
    {
      "epoch": 0.727969050417617,
      "grad_norm": 3.592097759246826,
      "learning_rate": 2.863604332934563e-07,
      "loss": 0.7642,
      "step": 46760
    },
    {
      "epoch": 0.7280468914195863,
      "grad_norm": 3.26198673248291,
      "learning_rate": 2.8627849428885135e-07,
      "loss": 0.7533,
      "step": 46765
    },
    {
      "epoch": 0.7281247324215557,
      "grad_norm": 4.254547595977783,
      "learning_rate": 2.861965552842464e-07,
      "loss": 0.6636,
      "step": 46770
    },
    {
      "epoch": 0.7282025734235251,
      "grad_norm": 5.791482448577881,
      "learning_rate": 2.861146162796414e-07,
      "loss": 0.7692,
      "step": 46775
    },
    {
      "epoch": 0.7282804144254945,
      "grad_norm": 3.4144723415374756,
      "learning_rate": 2.860326772750364e-07,
      "loss": 0.7872,
      "step": 46780
    },
    {
      "epoch": 0.7283582554274639,
      "grad_norm": 3.173515558242798,
      "learning_rate": 2.859507382704315e-07,
      "loss": 0.7189,
      "step": 46785
    },
    {
      "epoch": 0.7284360964294332,
      "grad_norm": 3.3186471462249756,
      "learning_rate": 2.858687992658265e-07,
      "loss": 0.8245,
      "step": 46790
    },
    {
      "epoch": 0.7285139374314026,
      "grad_norm": 3.573747158050537,
      "learning_rate": 2.8578686026122153e-07,
      "loss": 0.689,
      "step": 46795
    },
    {
      "epoch": 0.728591778433372,
      "grad_norm": 3.1744699478149414,
      "learning_rate": 2.8570492125661655e-07,
      "loss": 0.7536,
      "step": 46800
    },
    {
      "epoch": 0.7286696194353414,
      "grad_norm": 5.481973648071289,
      "learning_rate": 2.856229822520116e-07,
      "loss": 0.6991,
      "step": 46805
    },
    {
      "epoch": 0.7287474604373108,
      "grad_norm": 3.6041758060455322,
      "learning_rate": 2.8554104324740665e-07,
      "loss": 0.775,
      "step": 46810
    },
    {
      "epoch": 0.7288253014392801,
      "grad_norm": 2.9685873985290527,
      "learning_rate": 2.854591042428016e-07,
      "loss": 0.7055,
      "step": 46815
    },
    {
      "epoch": 0.7289031424412495,
      "grad_norm": 3.551771879196167,
      "learning_rate": 2.8537716523819664e-07,
      "loss": 0.5994,
      "step": 46820
    },
    {
      "epoch": 0.7289809834432189,
      "grad_norm": 3.4299192428588867,
      "learning_rate": 2.852952262335917e-07,
      "loss": 0.6677,
      "step": 46825
    },
    {
      "epoch": 0.7290588244451882,
      "grad_norm": 3.0535383224487305,
      "learning_rate": 2.8521328722898674e-07,
      "loss": 0.7743,
      "step": 46830
    },
    {
      "epoch": 0.7291366654471576,
      "grad_norm": 3.602445363998413,
      "learning_rate": 2.8513134822438176e-07,
      "loss": 0.7645,
      "step": 46835
    },
    {
      "epoch": 0.729214506449127,
      "grad_norm": 3.3587710857391357,
      "learning_rate": 2.850494092197768e-07,
      "loss": 0.7604,
      "step": 46840
    },
    {
      "epoch": 0.7292923474510964,
      "grad_norm": 5.948568820953369,
      "learning_rate": 2.849674702151718e-07,
      "loss": 0.7222,
      "step": 46845
    },
    {
      "epoch": 0.7293701884530658,
      "grad_norm": 2.6205029487609863,
      "learning_rate": 2.848855312105669e-07,
      "loss": 0.6018,
      "step": 46850
    },
    {
      "epoch": 0.7294480294550352,
      "grad_norm": 6.05385684967041,
      "learning_rate": 2.848035922059619e-07,
      "loss": 0.8153,
      "step": 46855
    },
    {
      "epoch": 0.7295258704570046,
      "grad_norm": 3.174029588699341,
      "learning_rate": 2.8472165320135687e-07,
      "loss": 0.6683,
      "step": 46860
    },
    {
      "epoch": 0.7296037114589738,
      "grad_norm": 3.8151023387908936,
      "learning_rate": 2.8463971419675194e-07,
      "loss": 0.7366,
      "step": 46865
    },
    {
      "epoch": 0.7296815524609432,
      "grad_norm": 4.031579494476318,
      "learning_rate": 2.8455777519214697e-07,
      "loss": 0.6956,
      "step": 46870
    },
    {
      "epoch": 0.7297593934629126,
      "grad_norm": 3.301027297973633,
      "learning_rate": 2.84475836187542e-07,
      "loss": 0.8245,
      "step": 46875
    },
    {
      "epoch": 0.729837234464882,
      "grad_norm": 3.7310054302215576,
      "learning_rate": 2.84393897182937e-07,
      "loss": 0.831,
      "step": 46880
    },
    {
      "epoch": 0.7299150754668514,
      "grad_norm": 2.6273341178894043,
      "learning_rate": 2.8431195817833203e-07,
      "loss": 0.7249,
      "step": 46885
    },
    {
      "epoch": 0.7299929164688208,
      "grad_norm": 4.502962112426758,
      "learning_rate": 2.8423001917372705e-07,
      "loss": 0.682,
      "step": 46890
    },
    {
      "epoch": 0.7300707574707902,
      "grad_norm": 2.847952365875244,
      "learning_rate": 2.8414808016912213e-07,
      "loss": 0.8095,
      "step": 46895
    },
    {
      "epoch": 0.7301485984727596,
      "grad_norm": 5.369004726409912,
      "learning_rate": 2.840661411645171e-07,
      "loss": 0.7259,
      "step": 46900
    },
    {
      "epoch": 0.7302264394747289,
      "grad_norm": 4.038816452026367,
      "learning_rate": 2.839842021599121e-07,
      "loss": 0.7622,
      "step": 46905
    },
    {
      "epoch": 0.7303042804766983,
      "grad_norm": 3.46608829498291,
      "learning_rate": 2.839022631553072e-07,
      "loss": 0.6466,
      "step": 46910
    },
    {
      "epoch": 0.7303821214786677,
      "grad_norm": 3.9980077743530273,
      "learning_rate": 2.838203241507022e-07,
      "loss": 0.7241,
      "step": 46915
    },
    {
      "epoch": 0.730459962480637,
      "grad_norm": 4.407654285430908,
      "learning_rate": 2.8373838514609724e-07,
      "loss": 0.8187,
      "step": 46920
    },
    {
      "epoch": 0.7305378034826064,
      "grad_norm": 4.83888578414917,
      "learning_rate": 2.8365644614149226e-07,
      "loss": 0.6975,
      "step": 46925
    },
    {
      "epoch": 0.7306156444845758,
      "grad_norm": 3.406704902648926,
      "learning_rate": 2.835745071368873e-07,
      "loss": 0.7006,
      "step": 46930
    },
    {
      "epoch": 0.7306934854865452,
      "grad_norm": 3.3480916023254395,
      "learning_rate": 2.8349256813228236e-07,
      "loss": 0.6672,
      "step": 46935
    },
    {
      "epoch": 0.7307713264885145,
      "grad_norm": 3.4775843620300293,
      "learning_rate": 2.834106291276773e-07,
      "loss": 0.6643,
      "step": 46940
    },
    {
      "epoch": 0.7308491674904839,
      "grad_norm": 3.0172197818756104,
      "learning_rate": 2.8332869012307235e-07,
      "loss": 0.6706,
      "step": 46945
    },
    {
      "epoch": 0.7309270084924533,
      "grad_norm": 4.071489334106445,
      "learning_rate": 2.832467511184674e-07,
      "loss": 0.7745,
      "step": 46950
    },
    {
      "epoch": 0.7310048494944227,
      "grad_norm": 4.3749775886535645,
      "learning_rate": 2.8316481211386244e-07,
      "loss": 0.7059,
      "step": 46955
    },
    {
      "epoch": 0.7310826904963921,
      "grad_norm": 4.053957939147949,
      "learning_rate": 2.830828731092574e-07,
      "loss": 0.6982,
      "step": 46960
    },
    {
      "epoch": 0.7311605314983615,
      "grad_norm": 3.3284595012664795,
      "learning_rate": 2.830009341046525e-07,
      "loss": 0.6965,
      "step": 46965
    },
    {
      "epoch": 0.7312383725003309,
      "grad_norm": 3.2319602966308594,
      "learning_rate": 2.829189951000475e-07,
      "loss": 0.8168,
      "step": 46970
    },
    {
      "epoch": 0.7313162135023002,
      "grad_norm": 7.422788143157959,
      "learning_rate": 2.828370560954426e-07,
      "loss": 0.9516,
      "step": 46975
    },
    {
      "epoch": 0.7313940545042695,
      "grad_norm": 5.347562313079834,
      "learning_rate": 2.8275511709083755e-07,
      "loss": 0.6436,
      "step": 46980
    },
    {
      "epoch": 0.7314718955062389,
      "grad_norm": 3.0234568119049072,
      "learning_rate": 2.826731780862326e-07,
      "loss": 0.7171,
      "step": 46985
    },
    {
      "epoch": 0.7315497365082083,
      "grad_norm": 4.647586822509766,
      "learning_rate": 2.8259123908162765e-07,
      "loss": 0.7006,
      "step": 46990
    },
    {
      "epoch": 0.7316275775101777,
      "grad_norm": 3.3549914360046387,
      "learning_rate": 2.8250930007702267e-07,
      "loss": 0.8365,
      "step": 46995
    },
    {
      "epoch": 0.7317054185121471,
      "grad_norm": 4.270804405212402,
      "learning_rate": 2.8242736107241764e-07,
      "loss": 0.6871,
      "step": 47000
    },
    {
      "epoch": 0.7317832595141165,
      "grad_norm": 5.053944110870361,
      "learning_rate": 2.823454220678127e-07,
      "loss": 0.756,
      "step": 47005
    },
    {
      "epoch": 0.7318611005160859,
      "grad_norm": 3.172109365463257,
      "learning_rate": 2.8226348306320774e-07,
      "loss": 0.8108,
      "step": 47010
    },
    {
      "epoch": 0.7319389415180553,
      "grad_norm": 3.203810930252075,
      "learning_rate": 2.8218154405860276e-07,
      "loss": 0.8529,
      "step": 47015
    },
    {
      "epoch": 0.7320167825200246,
      "grad_norm": 4.791949272155762,
      "learning_rate": 2.820996050539978e-07,
      "loss": 0.7162,
      "step": 47020
    },
    {
      "epoch": 0.732094623521994,
      "grad_norm": 2.962249279022217,
      "learning_rate": 2.820176660493928e-07,
      "loss": 0.7423,
      "step": 47025
    },
    {
      "epoch": 0.7321724645239633,
      "grad_norm": 3.710045337677002,
      "learning_rate": 2.819357270447878e-07,
      "loss": 0.672,
      "step": 47030
    },
    {
      "epoch": 0.7322503055259327,
      "grad_norm": 3.500413179397583,
      "learning_rate": 2.818537880401829e-07,
      "loss": 0.8006,
      "step": 47035
    },
    {
      "epoch": 0.7323281465279021,
      "grad_norm": 5.875401496887207,
      "learning_rate": 2.8177184903557787e-07,
      "loss": 0.7423,
      "step": 47040
    },
    {
      "epoch": 0.7324059875298715,
      "grad_norm": 3.7425382137298584,
      "learning_rate": 2.8168991003097295e-07,
      "loss": 0.8016,
      "step": 47045
    },
    {
      "epoch": 0.7324838285318409,
      "grad_norm": 4.828375816345215,
      "learning_rate": 2.8160797102636797e-07,
      "loss": 0.7041,
      "step": 47050
    },
    {
      "epoch": 0.7325616695338102,
      "grad_norm": 2.9381606578826904,
      "learning_rate": 2.81526032021763e-07,
      "loss": 0.7696,
      "step": 47055
    },
    {
      "epoch": 0.7326395105357796,
      "grad_norm": 5.7646613121032715,
      "learning_rate": 2.81444093017158e-07,
      "loss": 0.739,
      "step": 47060
    },
    {
      "epoch": 0.732717351537749,
      "grad_norm": 3.7390706539154053,
      "learning_rate": 2.8136215401255303e-07,
      "loss": 0.6788,
      "step": 47065
    },
    {
      "epoch": 0.7327951925397184,
      "grad_norm": 5.14375114440918,
      "learning_rate": 2.8128021500794805e-07,
      "loss": 0.7309,
      "step": 47070
    },
    {
      "epoch": 0.7328730335416878,
      "grad_norm": 4.093204498291016,
      "learning_rate": 2.8119827600334313e-07,
      "loss": 0.698,
      "step": 47075
    },
    {
      "epoch": 0.7329508745436571,
      "grad_norm": 3.7759382724761963,
      "learning_rate": 2.811163369987381e-07,
      "loss": 0.7748,
      "step": 47080
    },
    {
      "epoch": 0.7330287155456265,
      "grad_norm": 3.904548406600952,
      "learning_rate": 2.810343979941331e-07,
      "loss": 0.8128,
      "step": 47085
    },
    {
      "epoch": 0.7331065565475959,
      "grad_norm": 9.094959259033203,
      "learning_rate": 2.809524589895282e-07,
      "loss": 0.6991,
      "step": 47090
    },
    {
      "epoch": 0.7331843975495652,
      "grad_norm": 8.977814674377441,
      "learning_rate": 2.808705199849232e-07,
      "loss": 0.7016,
      "step": 47095
    },
    {
      "epoch": 0.7332622385515346,
      "grad_norm": 2.9130189418792725,
      "learning_rate": 2.807885809803182e-07,
      "loss": 0.8116,
      "step": 47100
    },
    {
      "epoch": 0.733340079553504,
      "grad_norm": 2.8606650829315186,
      "learning_rate": 2.8070664197571326e-07,
      "loss": 0.7642,
      "step": 47105
    },
    {
      "epoch": 0.7334179205554734,
      "grad_norm": 2.479870080947876,
      "learning_rate": 2.806247029711083e-07,
      "loss": 0.7308,
      "step": 47110
    },
    {
      "epoch": 0.7334957615574428,
      "grad_norm": 3.4306538105010986,
      "learning_rate": 2.8054276396650336e-07,
      "loss": 0.6415,
      "step": 47115
    },
    {
      "epoch": 0.7335736025594122,
      "grad_norm": 2.8478362560272217,
      "learning_rate": 2.8046082496189833e-07,
      "loss": 0.6767,
      "step": 47120
    },
    {
      "epoch": 0.7336514435613816,
      "grad_norm": 3.4778783321380615,
      "learning_rate": 2.8037888595729335e-07,
      "loss": 0.7173,
      "step": 47125
    },
    {
      "epoch": 0.7337292845633508,
      "grad_norm": 3.813336133956909,
      "learning_rate": 2.802969469526884e-07,
      "loss": 0.7851,
      "step": 47130
    },
    {
      "epoch": 0.7338071255653202,
      "grad_norm": 3.2470643520355225,
      "learning_rate": 2.8021500794808345e-07,
      "loss": 0.7574,
      "step": 47135
    },
    {
      "epoch": 0.7338849665672896,
      "grad_norm": 3.961081027984619,
      "learning_rate": 2.801330689434784e-07,
      "loss": 0.8201,
      "step": 47140
    },
    {
      "epoch": 0.733962807569259,
      "grad_norm": 3.000685930252075,
      "learning_rate": 2.800511299388735e-07,
      "loss": 0.6425,
      "step": 47145
    },
    {
      "epoch": 0.7340406485712284,
      "grad_norm": 4.910801887512207,
      "learning_rate": 2.799691909342685e-07,
      "loss": 0.8394,
      "step": 47150
    },
    {
      "epoch": 0.7341184895731978,
      "grad_norm": 3.086811065673828,
      "learning_rate": 2.7988725192966353e-07,
      "loss": 0.7809,
      "step": 47155
    },
    {
      "epoch": 0.7341963305751672,
      "grad_norm": 2.9702835083007812,
      "learning_rate": 2.798053129250586e-07,
      "loss": 0.8669,
      "step": 47160
    },
    {
      "epoch": 0.7342741715771366,
      "grad_norm": 3.781804084777832,
      "learning_rate": 2.797233739204536e-07,
      "loss": 0.8293,
      "step": 47165
    },
    {
      "epoch": 0.7343520125791059,
      "grad_norm": 11.27530574798584,
      "learning_rate": 2.7964143491584865e-07,
      "loss": 0.7841,
      "step": 47170
    },
    {
      "epoch": 0.7344298535810753,
      "grad_norm": 2.859074592590332,
      "learning_rate": 2.795594959112437e-07,
      "loss": 0.7178,
      "step": 47175
    },
    {
      "epoch": 0.7345076945830447,
      "grad_norm": 6.172121524810791,
      "learning_rate": 2.794775569066387e-07,
      "loss": 0.7698,
      "step": 47180
    },
    {
      "epoch": 0.734585535585014,
      "grad_norm": 4.02262020111084,
      "learning_rate": 2.793956179020337e-07,
      "loss": 0.7288,
      "step": 47185
    },
    {
      "epoch": 0.7346633765869834,
      "grad_norm": 5.290551662445068,
      "learning_rate": 2.7931367889742874e-07,
      "loss": 0.7651,
      "step": 47190
    },
    {
      "epoch": 0.7347412175889528,
      "grad_norm": 3.318727493286133,
      "learning_rate": 2.7923173989282376e-07,
      "loss": 0.7852,
      "step": 47195
    },
    {
      "epoch": 0.7348190585909222,
      "grad_norm": 3.1297385692596436,
      "learning_rate": 2.7914980088821884e-07,
      "loss": 0.7468,
      "step": 47200
    },
    {
      "epoch": 0.7348968995928915,
      "grad_norm": 4.752504825592041,
      "learning_rate": 2.790678618836138e-07,
      "loss": 0.81,
      "step": 47205
    },
    {
      "epoch": 0.7349747405948609,
      "grad_norm": 4.348777770996094,
      "learning_rate": 2.7898592287900883e-07,
      "loss": 0.7062,
      "step": 47210
    },
    {
      "epoch": 0.7350525815968303,
      "grad_norm": 3.4660797119140625,
      "learning_rate": 2.789039838744039e-07,
      "loss": 0.8043,
      "step": 47215
    },
    {
      "epoch": 0.7351304225987997,
      "grad_norm": 3.7979254722595215,
      "learning_rate": 2.788220448697989e-07,
      "loss": 0.7914,
      "step": 47220
    },
    {
      "epoch": 0.7352082636007691,
      "grad_norm": 4.6002912521362305,
      "learning_rate": 2.7874010586519395e-07,
      "loss": 0.6584,
      "step": 47225
    },
    {
      "epoch": 0.7352861046027385,
      "grad_norm": 4.253154277801514,
      "learning_rate": 2.7865816686058897e-07,
      "loss": 0.6789,
      "step": 47230
    },
    {
      "epoch": 0.7353639456047079,
      "grad_norm": 11.233471870422363,
      "learning_rate": 2.78576227855984e-07,
      "loss": 0.7888,
      "step": 47235
    },
    {
      "epoch": 0.7354417866066773,
      "grad_norm": 5.823216438293457,
      "learning_rate": 2.7849428885137906e-07,
      "loss": 0.7251,
      "step": 47240
    },
    {
      "epoch": 0.7355196276086465,
      "grad_norm": 4.937676906585693,
      "learning_rate": 2.7841234984677403e-07,
      "loss": 0.6578,
      "step": 47245
    },
    {
      "epoch": 0.7355974686106159,
      "grad_norm": 4.6061110496521,
      "learning_rate": 2.7833041084216906e-07,
      "loss": 0.7806,
      "step": 47250
    },
    {
      "epoch": 0.7356753096125853,
      "grad_norm": 3.425356149673462,
      "learning_rate": 2.7824847183756413e-07,
      "loss": 0.8431,
      "step": 47255
    },
    {
      "epoch": 0.7357531506145547,
      "grad_norm": 6.686736583709717,
      "learning_rate": 2.7816653283295915e-07,
      "loss": 0.8683,
      "step": 47260
    },
    {
      "epoch": 0.7358309916165241,
      "grad_norm": 2.9011332988739014,
      "learning_rate": 2.780845938283541e-07,
      "loss": 0.6877,
      "step": 47265
    },
    {
      "epoch": 0.7359088326184935,
      "grad_norm": 4.304633617401123,
      "learning_rate": 2.780026548237492e-07,
      "loss": 0.7282,
      "step": 47270
    },
    {
      "epoch": 0.7359866736204629,
      "grad_norm": 3.545462131500244,
      "learning_rate": 2.779207158191442e-07,
      "loss": 0.6448,
      "step": 47275
    },
    {
      "epoch": 0.7360645146224322,
      "grad_norm": 5.752591609954834,
      "learning_rate": 2.7783877681453924e-07,
      "loss": 0.6319,
      "step": 47280
    },
    {
      "epoch": 0.7361423556244016,
      "grad_norm": 4.883378982543945,
      "learning_rate": 2.7775683780993426e-07,
      "loss": 0.767,
      "step": 47285
    },
    {
      "epoch": 0.736220196626371,
      "grad_norm": 3.5203464031219482,
      "learning_rate": 2.776748988053293e-07,
      "loss": 0.761,
      "step": 47290
    },
    {
      "epoch": 0.7362980376283403,
      "grad_norm": 2.801938533782959,
      "learning_rate": 2.7759295980072436e-07,
      "loss": 0.6571,
      "step": 47295
    },
    {
      "epoch": 0.7363758786303097,
      "grad_norm": 2.4005565643310547,
      "learning_rate": 2.775110207961194e-07,
      "loss": 0.756,
      "step": 47300
    },
    {
      "epoch": 0.7364537196322791,
      "grad_norm": 4.064172744750977,
      "learning_rate": 2.7742908179151435e-07,
      "loss": 0.7288,
      "step": 47305
    },
    {
      "epoch": 0.7365315606342485,
      "grad_norm": 3.2868990898132324,
      "learning_rate": 2.773471427869094e-07,
      "loss": 0.7411,
      "step": 47310
    },
    {
      "epoch": 0.7366094016362179,
      "grad_norm": 2.138129949569702,
      "learning_rate": 2.7726520378230445e-07,
      "loss": 0.6878,
      "step": 47315
    },
    {
      "epoch": 0.7366872426381872,
      "grad_norm": 4.990584850311279,
      "learning_rate": 2.7718326477769947e-07,
      "loss": 0.7258,
      "step": 47320
    },
    {
      "epoch": 0.7367650836401566,
      "grad_norm": 4.147559642791748,
      "learning_rate": 2.771013257730945e-07,
      "loss": 0.7485,
      "step": 47325
    },
    {
      "epoch": 0.736842924642126,
      "grad_norm": 4.67435884475708,
      "learning_rate": 2.770193867684895e-07,
      "loss": 0.7258,
      "step": 47330
    },
    {
      "epoch": 0.7369207656440954,
      "grad_norm": 3.6716771125793457,
      "learning_rate": 2.7693744776388453e-07,
      "loss": 0.7873,
      "step": 47335
    },
    {
      "epoch": 0.7369986066460648,
      "grad_norm": 3.869614839553833,
      "learning_rate": 2.768555087592796e-07,
      "loss": 0.6794,
      "step": 47340
    },
    {
      "epoch": 0.7370764476480341,
      "grad_norm": 4.422678470611572,
      "learning_rate": 2.767735697546746e-07,
      "loss": 0.7056,
      "step": 47345
    },
    {
      "epoch": 0.7371542886500035,
      "grad_norm": 8.342263221740723,
      "learning_rate": 2.7669163075006965e-07,
      "loss": 0.7318,
      "step": 47350
    },
    {
      "epoch": 0.7372321296519729,
      "grad_norm": 3.36340594291687,
      "learning_rate": 2.766096917454647e-07,
      "loss": 0.7057,
      "step": 47355
    },
    {
      "epoch": 0.7373099706539422,
      "grad_norm": 4.39555025100708,
      "learning_rate": 2.765277527408597e-07,
      "loss": 0.7117,
      "step": 47360
    },
    {
      "epoch": 0.7373878116559116,
      "grad_norm": 5.481449604034424,
      "learning_rate": 2.764458137362547e-07,
      "loss": 0.7779,
      "step": 47365
    },
    {
      "epoch": 0.737465652657881,
      "grad_norm": 3.4881973266601562,
      "learning_rate": 2.7636387473164974e-07,
      "loss": 0.6581,
      "step": 47370
    },
    {
      "epoch": 0.7375434936598504,
      "grad_norm": 6.064071178436279,
      "learning_rate": 2.7628193572704476e-07,
      "loss": 0.6667,
      "step": 47375
    },
    {
      "epoch": 0.7376213346618198,
      "grad_norm": 3.9830546379089355,
      "learning_rate": 2.7619999672243984e-07,
      "loss": 0.7319,
      "step": 47380
    },
    {
      "epoch": 0.7376991756637892,
      "grad_norm": 3.0753324031829834,
      "learning_rate": 2.761180577178348e-07,
      "loss": 0.8208,
      "step": 47385
    },
    {
      "epoch": 0.7377770166657586,
      "grad_norm": 5.100201606750488,
      "learning_rate": 2.7603611871322983e-07,
      "loss": 0.7705,
      "step": 47390
    },
    {
      "epoch": 0.7378548576677278,
      "grad_norm": 21.23238754272461,
      "learning_rate": 2.759541797086249e-07,
      "loss": 0.708,
      "step": 47395
    },
    {
      "epoch": 0.7379326986696972,
      "grad_norm": 4.162858963012695,
      "learning_rate": 2.758722407040199e-07,
      "loss": 0.7625,
      "step": 47400
    },
    {
      "epoch": 0.7380105396716666,
      "grad_norm": 2.6884047985076904,
      "learning_rate": 2.757903016994149e-07,
      "loss": 0.6676,
      "step": 47405
    },
    {
      "epoch": 0.738088380673636,
      "grad_norm": 4.463581562042236,
      "learning_rate": 2.7570836269480997e-07,
      "loss": 0.6305,
      "step": 47410
    },
    {
      "epoch": 0.7381662216756054,
      "grad_norm": 4.976219654083252,
      "learning_rate": 2.75626423690205e-07,
      "loss": 0.7193,
      "step": 47415
    },
    {
      "epoch": 0.7382440626775748,
      "grad_norm": 2.697484254837036,
      "learning_rate": 2.7554448468560007e-07,
      "loss": 0.7284,
      "step": 47420
    },
    {
      "epoch": 0.7383219036795442,
      "grad_norm": 4.643062591552734,
      "learning_rate": 2.7546254568099503e-07,
      "loss": 0.7134,
      "step": 47425
    },
    {
      "epoch": 0.7383997446815136,
      "grad_norm": 3.4359333515167236,
      "learning_rate": 2.7538060667639006e-07,
      "loss": 0.7798,
      "step": 47430
    },
    {
      "epoch": 0.7384775856834829,
      "grad_norm": 6.4495158195495605,
      "learning_rate": 2.7529866767178513e-07,
      "loss": 0.8561,
      "step": 47435
    },
    {
      "epoch": 0.7385554266854523,
      "grad_norm": 4.457662582397461,
      "learning_rate": 2.7521672866718015e-07,
      "loss": 0.7262,
      "step": 47440
    },
    {
      "epoch": 0.7386332676874217,
      "grad_norm": 8.19156551361084,
      "learning_rate": 2.751347896625751e-07,
      "loss": 0.7589,
      "step": 47445
    },
    {
      "epoch": 0.738711108689391,
      "grad_norm": 5.752453327178955,
      "learning_rate": 2.750528506579702e-07,
      "loss": 0.8002,
      "step": 47450
    },
    {
      "epoch": 0.7387889496913604,
      "grad_norm": 2.6278951168060303,
      "learning_rate": 2.749709116533652e-07,
      "loss": 0.7035,
      "step": 47455
    },
    {
      "epoch": 0.7388667906933298,
      "grad_norm": 5.286937236785889,
      "learning_rate": 2.7488897264876024e-07,
      "loss": 0.8191,
      "step": 47460
    },
    {
      "epoch": 0.7389446316952992,
      "grad_norm": 3.1538033485412598,
      "learning_rate": 2.748070336441553e-07,
      "loss": 0.8144,
      "step": 47465
    },
    {
      "epoch": 0.7390224726972685,
      "grad_norm": 3.220646858215332,
      "learning_rate": 2.747250946395503e-07,
      "loss": 0.8253,
      "step": 47470
    },
    {
      "epoch": 0.7391003136992379,
      "grad_norm": 5.327004432678223,
      "learning_rate": 2.7464315563494536e-07,
      "loss": 0.7094,
      "step": 47475
    },
    {
      "epoch": 0.7391781547012073,
      "grad_norm": 7.698801517486572,
      "learning_rate": 2.745612166303404e-07,
      "loss": 0.7853,
      "step": 47480
    },
    {
      "epoch": 0.7392559957031767,
      "grad_norm": 2.4389476776123047,
      "learning_rate": 2.744792776257354e-07,
      "loss": 0.6888,
      "step": 47485
    },
    {
      "epoch": 0.7393338367051461,
      "grad_norm": 6.331624507904053,
      "learning_rate": 2.743973386211304e-07,
      "loss": 0.6881,
      "step": 47490
    },
    {
      "epoch": 0.7394116777071155,
      "grad_norm": 2.6485209465026855,
      "learning_rate": 2.7431539961652545e-07,
      "loss": 0.7445,
      "step": 47495
    },
    {
      "epoch": 0.7394895187090849,
      "grad_norm": 5.753763198852539,
      "learning_rate": 2.7423346061192047e-07,
      "loss": 0.8761,
      "step": 47500
    },
    {
      "epoch": 0.7395673597110543,
      "grad_norm": 4.331512451171875,
      "learning_rate": 2.7415152160731554e-07,
      "loss": 0.8634,
      "step": 47505
    },
    {
      "epoch": 0.7396452007130235,
      "grad_norm": 3.6013667583465576,
      "learning_rate": 2.740695826027105e-07,
      "loss": 0.7974,
      "step": 47510
    },
    {
      "epoch": 0.7397230417149929,
      "grad_norm": 3.3562183380126953,
      "learning_rate": 2.7398764359810553e-07,
      "loss": 0.687,
      "step": 47515
    },
    {
      "epoch": 0.7398008827169623,
      "grad_norm": 2.7636563777923584,
      "learning_rate": 2.739057045935006e-07,
      "loss": 0.7236,
      "step": 47520
    },
    {
      "epoch": 0.7398787237189317,
      "grad_norm": 6.658202648162842,
      "learning_rate": 2.7382376558889563e-07,
      "loss": 0.8165,
      "step": 47525
    },
    {
      "epoch": 0.7399565647209011,
      "grad_norm": 4.762782573699951,
      "learning_rate": 2.737418265842906e-07,
      "loss": 0.6817,
      "step": 47530
    },
    {
      "epoch": 0.7400344057228705,
      "grad_norm": 3.3788833618164062,
      "learning_rate": 2.736598875796857e-07,
      "loss": 0.7067,
      "step": 47535
    },
    {
      "epoch": 0.7401122467248399,
      "grad_norm": 3.0917439460754395,
      "learning_rate": 2.735779485750807e-07,
      "loss": 0.7459,
      "step": 47540
    },
    {
      "epoch": 0.7401900877268092,
      "grad_norm": 3.6282193660736084,
      "learning_rate": 2.7349600957047577e-07,
      "loss": 0.6712,
      "step": 47545
    },
    {
      "epoch": 0.7402679287287786,
      "grad_norm": 6.270857810974121,
      "learning_rate": 2.7341407056587074e-07,
      "loss": 0.7208,
      "step": 47550
    },
    {
      "epoch": 0.740345769730748,
      "grad_norm": 5.801802635192871,
      "learning_rate": 2.7333213156126576e-07,
      "loss": 0.6853,
      "step": 47555
    },
    {
      "epoch": 0.7404236107327173,
      "grad_norm": 4.123294830322266,
      "learning_rate": 2.7325019255666084e-07,
      "loss": 0.7415,
      "step": 47560
    },
    {
      "epoch": 0.7405014517346867,
      "grad_norm": 3.1462278366088867,
      "learning_rate": 2.7316825355205586e-07,
      "loss": 0.7924,
      "step": 47565
    },
    {
      "epoch": 0.7405792927366561,
      "grad_norm": 3.746478796005249,
      "learning_rate": 2.7308631454745083e-07,
      "loss": 0.7074,
      "step": 47570
    },
    {
      "epoch": 0.7406571337386255,
      "grad_norm": 3.6443593502044678,
      "learning_rate": 2.730043755428459e-07,
      "loss": 0.7835,
      "step": 47575
    },
    {
      "epoch": 0.7407349747405949,
      "grad_norm": 2.6422572135925293,
      "learning_rate": 2.729224365382409e-07,
      "loss": 0.6888,
      "step": 47580
    },
    {
      "epoch": 0.7408128157425642,
      "grad_norm": 3.640272855758667,
      "learning_rate": 2.7284049753363595e-07,
      "loss": 0.568,
      "step": 47585
    },
    {
      "epoch": 0.7408906567445336,
      "grad_norm": 4.0122971534729,
      "learning_rate": 2.7275855852903097e-07,
      "loss": 0.8044,
      "step": 47590
    },
    {
      "epoch": 0.740968497746503,
      "grad_norm": 3.5645861625671387,
      "learning_rate": 2.72676619524426e-07,
      "loss": 0.688,
      "step": 47595
    },
    {
      "epoch": 0.7410463387484724,
      "grad_norm": 4.369284629821777,
      "learning_rate": 2.7259468051982107e-07,
      "loss": 0.7223,
      "step": 47600
    },
    {
      "epoch": 0.7411241797504418,
      "grad_norm": 4.465457916259766,
      "learning_rate": 2.725127415152161e-07,
      "loss": 0.676,
      "step": 47605
    },
    {
      "epoch": 0.7412020207524112,
      "grad_norm": 5.196043968200684,
      "learning_rate": 2.7243080251061106e-07,
      "loss": 0.808,
      "step": 47610
    },
    {
      "epoch": 0.7412798617543805,
      "grad_norm": 6.151198387145996,
      "learning_rate": 2.7234886350600613e-07,
      "loss": 0.7302,
      "step": 47615
    },
    {
      "epoch": 0.7413577027563499,
      "grad_norm": 5.993812084197998,
      "learning_rate": 2.7226692450140115e-07,
      "loss": 0.6613,
      "step": 47620
    },
    {
      "epoch": 0.7414355437583192,
      "grad_norm": 2.9085958003997803,
      "learning_rate": 2.721849854967962e-07,
      "loss": 0.6624,
      "step": 47625
    },
    {
      "epoch": 0.7415133847602886,
      "grad_norm": 4.893017292022705,
      "learning_rate": 2.721030464921912e-07,
      "loss": 0.8212,
      "step": 47630
    },
    {
      "epoch": 0.741591225762258,
      "grad_norm": 4.533414363861084,
      "learning_rate": 2.720211074875862e-07,
      "loss": 0.7,
      "step": 47635
    },
    {
      "epoch": 0.7416690667642274,
      "grad_norm": 4.465665340423584,
      "learning_rate": 2.7193916848298124e-07,
      "loss": 0.6888,
      "step": 47640
    },
    {
      "epoch": 0.7417469077661968,
      "grad_norm": 6.0814337730407715,
      "learning_rate": 2.718572294783763e-07,
      "loss": 0.7948,
      "step": 47645
    },
    {
      "epoch": 0.7418247487681662,
      "grad_norm": 3.7214996814727783,
      "learning_rate": 2.717752904737713e-07,
      "loss": 0.6612,
      "step": 47650
    },
    {
      "epoch": 0.7419025897701356,
      "grad_norm": 3.5121705532073975,
      "learning_rate": 2.716933514691663e-07,
      "loss": 0.7303,
      "step": 47655
    },
    {
      "epoch": 0.7419804307721048,
      "grad_norm": 3.631423234939575,
      "learning_rate": 2.716114124645614e-07,
      "loss": 0.7649,
      "step": 47660
    },
    {
      "epoch": 0.7420582717740742,
      "grad_norm": 2.728424310684204,
      "learning_rate": 2.715294734599564e-07,
      "loss": 0.752,
      "step": 47665
    },
    {
      "epoch": 0.7421361127760436,
      "grad_norm": 4.5135498046875,
      "learning_rate": 2.714475344553514e-07,
      "loss": 0.7374,
      "step": 47670
    },
    {
      "epoch": 0.742213953778013,
      "grad_norm": 3.7203385829925537,
      "learning_rate": 2.7136559545074645e-07,
      "loss": 0.7419,
      "step": 47675
    },
    {
      "epoch": 0.7422917947799824,
      "grad_norm": 3.4192516803741455,
      "learning_rate": 2.7128365644614147e-07,
      "loss": 0.7193,
      "step": 47680
    },
    {
      "epoch": 0.7423696357819518,
      "grad_norm": 4.042068958282471,
      "learning_rate": 2.7120171744153654e-07,
      "loss": 0.6969,
      "step": 47685
    },
    {
      "epoch": 0.7424474767839212,
      "grad_norm": 6.505244731903076,
      "learning_rate": 2.711197784369315e-07,
      "loss": 0.7533,
      "step": 47690
    },
    {
      "epoch": 0.7425253177858906,
      "grad_norm": 4.347591876983643,
      "learning_rate": 2.7103783943232654e-07,
      "loss": 0.8151,
      "step": 47695
    },
    {
      "epoch": 0.7426031587878599,
      "grad_norm": 4.266767978668213,
      "learning_rate": 2.709559004277216e-07,
      "loss": 0.6697,
      "step": 47700
    },
    {
      "epoch": 0.7426809997898293,
      "grad_norm": 3.526186466217041,
      "learning_rate": 2.7087396142311663e-07,
      "loss": 0.8113,
      "step": 47705
    },
    {
      "epoch": 0.7427588407917987,
      "grad_norm": 3.870364189147949,
      "learning_rate": 2.707920224185116e-07,
      "loss": 0.7746,
      "step": 47710
    },
    {
      "epoch": 0.742836681793768,
      "grad_norm": 7.055910587310791,
      "learning_rate": 2.707100834139067e-07,
      "loss": 0.731,
      "step": 47715
    },
    {
      "epoch": 0.7429145227957374,
      "grad_norm": 3.7447736263275146,
      "learning_rate": 2.706281444093017e-07,
      "loss": 0.7767,
      "step": 47720
    },
    {
      "epoch": 0.7429923637977068,
      "grad_norm": 4.325788497924805,
      "learning_rate": 2.7054620540469677e-07,
      "loss": 0.7058,
      "step": 47725
    },
    {
      "epoch": 0.7430702047996762,
      "grad_norm": 6.888981819152832,
      "learning_rate": 2.7046426640009174e-07,
      "loss": 0.7805,
      "step": 47730
    },
    {
      "epoch": 0.7431480458016455,
      "grad_norm": 2.768723487854004,
      "learning_rate": 2.7038232739548676e-07,
      "loss": 0.5797,
      "step": 47735
    },
    {
      "epoch": 0.7432258868036149,
      "grad_norm": 3.3143179416656494,
      "learning_rate": 2.7030038839088184e-07,
      "loss": 0.6517,
      "step": 47740
    },
    {
      "epoch": 0.7433037278055843,
      "grad_norm": 6.379628658294678,
      "learning_rate": 2.7021844938627686e-07,
      "loss": 0.6693,
      "step": 47745
    },
    {
      "epoch": 0.7433815688075537,
      "grad_norm": 4.410959720611572,
      "learning_rate": 2.7013651038167183e-07,
      "loss": 0.7422,
      "step": 47750
    },
    {
      "epoch": 0.7434594098095231,
      "grad_norm": 3.036133289337158,
      "learning_rate": 2.700545713770669e-07,
      "loss": 0.8109,
      "step": 47755
    },
    {
      "epoch": 0.7435372508114925,
      "grad_norm": 3.189561128616333,
      "learning_rate": 2.6997263237246193e-07,
      "loss": 0.7201,
      "step": 47760
    },
    {
      "epoch": 0.7436150918134619,
      "grad_norm": 5.351257801055908,
      "learning_rate": 2.6989069336785695e-07,
      "loss": 0.9075,
      "step": 47765
    },
    {
      "epoch": 0.7436929328154313,
      "grad_norm": 2.5677008628845215,
      "learning_rate": 2.6980875436325197e-07,
      "loss": 0.6958,
      "step": 47770
    },
    {
      "epoch": 0.7437707738174005,
      "grad_norm": 6.060825824737549,
      "learning_rate": 2.69726815358647e-07,
      "loss": 0.7556,
      "step": 47775
    },
    {
      "epoch": 0.7438486148193699,
      "grad_norm": 2.898480176925659,
      "learning_rate": 2.69644876354042e-07,
      "loss": 0.7661,
      "step": 47780
    },
    {
      "epoch": 0.7439264558213393,
      "grad_norm": 5.315193176269531,
      "learning_rate": 2.695629373494371e-07,
      "loss": 0.7571,
      "step": 47785
    },
    {
      "epoch": 0.7440042968233087,
      "grad_norm": 2.9084980487823486,
      "learning_rate": 2.694809983448321e-07,
      "loss": 0.747,
      "step": 47790
    },
    {
      "epoch": 0.7440821378252781,
      "grad_norm": 4.630505561828613,
      "learning_rate": 2.6939905934022713e-07,
      "loss": 0.7187,
      "step": 47795
    },
    {
      "epoch": 0.7441599788272475,
      "grad_norm": 2.6790871620178223,
      "learning_rate": 2.6931712033562215e-07,
      "loss": 0.7494,
      "step": 47800
    },
    {
      "epoch": 0.7442378198292169,
      "grad_norm": 4.317978858947754,
      "learning_rate": 2.692351813310172e-07,
      "loss": 0.8152,
      "step": 47805
    },
    {
      "epoch": 0.7443156608311862,
      "grad_norm": 5.493204116821289,
      "learning_rate": 2.6915324232641225e-07,
      "loss": 0.766,
      "step": 47810
    },
    {
      "epoch": 0.7443935018331556,
      "grad_norm": 3.744283437728882,
      "learning_rate": 2.690713033218072e-07,
      "loss": 0.7856,
      "step": 47815
    },
    {
      "epoch": 0.744471342835125,
      "grad_norm": 4.157007217407227,
      "learning_rate": 2.6898936431720224e-07,
      "loss": 0.8028,
      "step": 47820
    },
    {
      "epoch": 0.7445491838370943,
      "grad_norm": 7.936100006103516,
      "learning_rate": 2.689074253125973e-07,
      "loss": 0.7592,
      "step": 47825
    },
    {
      "epoch": 0.7446270248390637,
      "grad_norm": 4.121297359466553,
      "learning_rate": 2.6882548630799234e-07,
      "loss": 0.7274,
      "step": 47830
    },
    {
      "epoch": 0.7447048658410331,
      "grad_norm": 4.276645660400391,
      "learning_rate": 2.687435473033873e-07,
      "loss": 0.7919,
      "step": 47835
    },
    {
      "epoch": 0.7447827068430025,
      "grad_norm": 5.099218368530273,
      "learning_rate": 2.686616082987824e-07,
      "loss": 0.7346,
      "step": 47840
    },
    {
      "epoch": 0.7448605478449719,
      "grad_norm": 3.85038423538208,
      "learning_rate": 2.685796692941774e-07,
      "loss": 0.8233,
      "step": 47845
    },
    {
      "epoch": 0.7449383888469412,
      "grad_norm": 2.833653211593628,
      "learning_rate": 2.684977302895725e-07,
      "loss": 0.6577,
      "step": 47850
    },
    {
      "epoch": 0.7450162298489106,
      "grad_norm": 3.5164794921875,
      "learning_rate": 2.6841579128496745e-07,
      "loss": 0.8341,
      "step": 47855
    },
    {
      "epoch": 0.74509407085088,
      "grad_norm": 5.090096950531006,
      "learning_rate": 2.6833385228036247e-07,
      "loss": 0.8065,
      "step": 47860
    },
    {
      "epoch": 0.7451719118528494,
      "grad_norm": 3.8225555419921875,
      "learning_rate": 2.6825191327575755e-07,
      "loss": 0.7675,
      "step": 47865
    },
    {
      "epoch": 0.7452497528548188,
      "grad_norm": 3.4643301963806152,
      "learning_rate": 2.6816997427115257e-07,
      "loss": 0.6924,
      "step": 47870
    },
    {
      "epoch": 0.7453275938567882,
      "grad_norm": 4.720452785491943,
      "learning_rate": 2.6808803526654754e-07,
      "loss": 0.7063,
      "step": 47875
    },
    {
      "epoch": 0.7454054348587575,
      "grad_norm": 3.2558114528656006,
      "learning_rate": 2.680060962619426e-07,
      "loss": 0.7459,
      "step": 47880
    },
    {
      "epoch": 0.7454832758607268,
      "grad_norm": 3.3760828971862793,
      "learning_rate": 2.6792415725733763e-07,
      "loss": 0.6626,
      "step": 47885
    },
    {
      "epoch": 0.7455611168626962,
      "grad_norm": 3.749676465988159,
      "learning_rate": 2.6784221825273266e-07,
      "loss": 0.7684,
      "step": 47890
    },
    {
      "epoch": 0.7456389578646656,
      "grad_norm": 3.9121060371398926,
      "learning_rate": 2.677602792481277e-07,
      "loss": 0.7392,
      "step": 47895
    },
    {
      "epoch": 0.745716798866635,
      "grad_norm": 6.382185459136963,
      "learning_rate": 2.676783402435227e-07,
      "loss": 0.7559,
      "step": 47900
    },
    {
      "epoch": 0.7457946398686044,
      "grad_norm": 3.7465670108795166,
      "learning_rate": 2.675964012389177e-07,
      "loss": 0.764,
      "step": 47905
    },
    {
      "epoch": 0.7458724808705738,
      "grad_norm": 3.571451425552368,
      "learning_rate": 2.675144622343128e-07,
      "loss": 0.6972,
      "step": 47910
    },
    {
      "epoch": 0.7459503218725432,
      "grad_norm": 2.9850687980651855,
      "learning_rate": 2.6743252322970777e-07,
      "loss": 0.7103,
      "step": 47915
    },
    {
      "epoch": 0.7460281628745126,
      "grad_norm": 4.465003490447998,
      "learning_rate": 2.6735058422510284e-07,
      "loss": 0.7215,
      "step": 47920
    },
    {
      "epoch": 0.7461060038764818,
      "grad_norm": 7.804738998413086,
      "learning_rate": 2.6726864522049786e-07,
      "loss": 0.7554,
      "step": 47925
    },
    {
      "epoch": 0.7461838448784512,
      "grad_norm": 10.050346374511719,
      "learning_rate": 2.671867062158929e-07,
      "loss": 0.7415,
      "step": 47930
    },
    {
      "epoch": 0.7462616858804206,
      "grad_norm": 5.889149188995361,
      "learning_rate": 2.671047672112879e-07,
      "loss": 0.7277,
      "step": 47935
    },
    {
      "epoch": 0.74633952688239,
      "grad_norm": 4.103286266326904,
      "learning_rate": 2.6702282820668293e-07,
      "loss": 0.7393,
      "step": 47940
    },
    {
      "epoch": 0.7464173678843594,
      "grad_norm": 3.0805633068084717,
      "learning_rate": 2.6694088920207795e-07,
      "loss": 0.6215,
      "step": 47945
    },
    {
      "epoch": 0.7464952088863288,
      "grad_norm": 4.746264934539795,
      "learning_rate": 2.66858950197473e-07,
      "loss": 0.7021,
      "step": 47950
    },
    {
      "epoch": 0.7465730498882982,
      "grad_norm": 3.826657295227051,
      "learning_rate": 2.66777011192868e-07,
      "loss": 0.6723,
      "step": 47955
    },
    {
      "epoch": 0.7466508908902676,
      "grad_norm": 2.868925094604492,
      "learning_rate": 2.66695072188263e-07,
      "loss": 0.78,
      "step": 47960
    },
    {
      "epoch": 0.7467287318922369,
      "grad_norm": 3.6088671684265137,
      "learning_rate": 2.666131331836581e-07,
      "loss": 0.65,
      "step": 47965
    },
    {
      "epoch": 0.7468065728942063,
      "grad_norm": 3.4721877574920654,
      "learning_rate": 2.665311941790531e-07,
      "loss": 0.7272,
      "step": 47970
    },
    {
      "epoch": 0.7468844138961757,
      "grad_norm": 2.9733211994171143,
      "learning_rate": 2.664492551744481e-07,
      "loss": 0.6604,
      "step": 47975
    },
    {
      "epoch": 0.746962254898145,
      "grad_norm": 2.990245819091797,
      "learning_rate": 2.6636731616984316e-07,
      "loss": 0.7242,
      "step": 47980
    },
    {
      "epoch": 0.7470400959001144,
      "grad_norm": 5.151874542236328,
      "learning_rate": 2.662853771652382e-07,
      "loss": 0.8054,
      "step": 47985
    },
    {
      "epoch": 0.7471179369020838,
      "grad_norm": 3.8161354064941406,
      "learning_rate": 2.6620343816063325e-07,
      "loss": 0.762,
      "step": 47990
    },
    {
      "epoch": 0.7471957779040532,
      "grad_norm": 3.423837900161743,
      "learning_rate": 2.661214991560282e-07,
      "loss": 0.708,
      "step": 47995
    },
    {
      "epoch": 0.7472736189060225,
      "grad_norm": 3.05196475982666,
      "learning_rate": 2.6603956015142324e-07,
      "loss": 0.7697,
      "step": 48000
    },
    {
      "epoch": 0.7473514599079919,
      "grad_norm": 4.275445938110352,
      "learning_rate": 2.659576211468183e-07,
      "loss": 0.7197,
      "step": 48005
    },
    {
      "epoch": 0.7474293009099613,
      "grad_norm": 5.850835800170898,
      "learning_rate": 2.6587568214221334e-07,
      "loss": 0.74,
      "step": 48010
    },
    {
      "epoch": 0.7475071419119307,
      "grad_norm": 3.436946153640747,
      "learning_rate": 2.657937431376083e-07,
      "loss": 0.6956,
      "step": 48015
    },
    {
      "epoch": 0.7475849829139001,
      "grad_norm": 8.99487018585205,
      "learning_rate": 2.657118041330034e-07,
      "loss": 0.7047,
      "step": 48020
    },
    {
      "epoch": 0.7476628239158695,
      "grad_norm": 3.8055357933044434,
      "learning_rate": 2.656298651283984e-07,
      "loss": 0.6932,
      "step": 48025
    },
    {
      "epoch": 0.7477406649178389,
      "grad_norm": 4.417625904083252,
      "learning_rate": 2.6554792612379343e-07,
      "loss": 0.8841,
      "step": 48030
    },
    {
      "epoch": 0.7478185059198083,
      "grad_norm": 3.533223867416382,
      "learning_rate": 2.6546598711918845e-07,
      "loss": 0.7001,
      "step": 48035
    },
    {
      "epoch": 0.7478963469217775,
      "grad_norm": 6.1243720054626465,
      "learning_rate": 2.6538404811458347e-07,
      "loss": 0.7036,
      "step": 48040
    },
    {
      "epoch": 0.7479741879237469,
      "grad_norm": 3.180238962173462,
      "learning_rate": 2.6530210910997855e-07,
      "loss": 0.6187,
      "step": 48045
    },
    {
      "epoch": 0.7480520289257163,
      "grad_norm": 6.181376934051514,
      "learning_rate": 2.6522017010537357e-07,
      "loss": 0.7504,
      "step": 48050
    },
    {
      "epoch": 0.7481298699276857,
      "grad_norm": 5.24513578414917,
      "learning_rate": 2.6513823110076854e-07,
      "loss": 0.7124,
      "step": 48055
    },
    {
      "epoch": 0.7482077109296551,
      "grad_norm": 4.932600021362305,
      "learning_rate": 2.650562920961636e-07,
      "loss": 0.6336,
      "step": 48060
    },
    {
      "epoch": 0.7482855519316245,
      "grad_norm": 12.562298774719238,
      "learning_rate": 2.6497435309155863e-07,
      "loss": 0.6891,
      "step": 48065
    },
    {
      "epoch": 0.7483633929335939,
      "grad_norm": 4.206464767456055,
      "learning_rate": 2.6489241408695366e-07,
      "loss": 0.8251,
      "step": 48070
    },
    {
      "epoch": 0.7484412339355632,
      "grad_norm": 3.6798508167266846,
      "learning_rate": 2.648104750823487e-07,
      "loss": 0.723,
      "step": 48075
    },
    {
      "epoch": 0.7485190749375326,
      "grad_norm": 3.4506421089172363,
      "learning_rate": 2.647285360777437e-07,
      "loss": 0.8116,
      "step": 48080
    },
    {
      "epoch": 0.748596915939502,
      "grad_norm": 2.9762420654296875,
      "learning_rate": 2.646465970731387e-07,
      "loss": 0.6895,
      "step": 48085
    },
    {
      "epoch": 0.7486747569414713,
      "grad_norm": 4.071290016174316,
      "learning_rate": 2.645646580685338e-07,
      "loss": 0.7713,
      "step": 48090
    },
    {
      "epoch": 0.7487525979434407,
      "grad_norm": 4.152281761169434,
      "learning_rate": 2.6448271906392877e-07,
      "loss": 0.7028,
      "step": 48095
    },
    {
      "epoch": 0.7488304389454101,
      "grad_norm": 6.530187129974365,
      "learning_rate": 2.644007800593238e-07,
      "loss": 0.7878,
      "step": 48100
    },
    {
      "epoch": 0.7489082799473795,
      "grad_norm": 3.586433172225952,
      "learning_rate": 2.6431884105471886e-07,
      "loss": 0.7345,
      "step": 48105
    },
    {
      "epoch": 0.7489861209493489,
      "grad_norm": 6.216477394104004,
      "learning_rate": 2.642369020501139e-07,
      "loss": 0.7603,
      "step": 48110
    },
    {
      "epoch": 0.7490639619513182,
      "grad_norm": 5.067862510681152,
      "learning_rate": 2.6415496304550896e-07,
      "loss": 0.7162,
      "step": 48115
    },
    {
      "epoch": 0.7491418029532876,
      "grad_norm": 4.337597370147705,
      "learning_rate": 2.6407302404090393e-07,
      "loss": 0.7871,
      "step": 48120
    },
    {
      "epoch": 0.749219643955257,
      "grad_norm": 4.733022212982178,
      "learning_rate": 2.6399108503629895e-07,
      "loss": 0.6738,
      "step": 48125
    },
    {
      "epoch": 0.7492974849572264,
      "grad_norm": 2.918813705444336,
      "learning_rate": 2.63909146031694e-07,
      "loss": 0.7073,
      "step": 48130
    },
    {
      "epoch": 0.7493753259591958,
      "grad_norm": 3.363646984100342,
      "learning_rate": 2.6382720702708905e-07,
      "loss": 0.6825,
      "step": 48135
    },
    {
      "epoch": 0.7494531669611652,
      "grad_norm": 3.9906091690063477,
      "learning_rate": 2.63745268022484e-07,
      "loss": 0.703,
      "step": 48140
    },
    {
      "epoch": 0.7495310079631345,
      "grad_norm": 4.613624095916748,
      "learning_rate": 2.636633290178791e-07,
      "loss": 0.709,
      "step": 48145
    },
    {
      "epoch": 0.7496088489651038,
      "grad_norm": 6.9466447830200195,
      "learning_rate": 2.635813900132741e-07,
      "loss": 0.7051,
      "step": 48150
    },
    {
      "epoch": 0.7496866899670732,
      "grad_norm": 3.1652586460113525,
      "learning_rate": 2.6349945100866913e-07,
      "loss": 0.8725,
      "step": 48155
    },
    {
      "epoch": 0.7497645309690426,
      "grad_norm": 9.098777770996094,
      "learning_rate": 2.6341751200406416e-07,
      "loss": 0.7741,
      "step": 48160
    },
    {
      "epoch": 0.749842371971012,
      "grad_norm": 4.554375171661377,
      "learning_rate": 2.633355729994592e-07,
      "loss": 0.7347,
      "step": 48165
    },
    {
      "epoch": 0.7499202129729814,
      "grad_norm": 2.4276487827301025,
      "learning_rate": 2.6325363399485425e-07,
      "loss": 0.6752,
      "step": 48170
    },
    {
      "epoch": 0.7499980539749508,
      "grad_norm": 4.137203216552734,
      "learning_rate": 2.631716949902493e-07,
      "loss": 0.7115,
      "step": 48175
    },
    {
      "epoch": 0.7500758949769202,
      "grad_norm": 3.0894439220428467,
      "learning_rate": 2.6308975598564424e-07,
      "loss": 0.8186,
      "step": 48180
    },
    {
      "epoch": 0.7501537359788896,
      "grad_norm": 2.858708143234253,
      "learning_rate": 2.630078169810393e-07,
      "loss": 0.6801,
      "step": 48185
    },
    {
      "epoch": 0.7502315769808588,
      "grad_norm": 3.822368860244751,
      "learning_rate": 2.6292587797643434e-07,
      "loss": 0.8387,
      "step": 48190
    },
    {
      "epoch": 0.7503094179828282,
      "grad_norm": 6.196854591369629,
      "learning_rate": 2.6284393897182936e-07,
      "loss": 0.6966,
      "step": 48195
    },
    {
      "epoch": 0.7503872589847976,
      "grad_norm": 3.0974886417388916,
      "learning_rate": 2.627619999672244e-07,
      "loss": 0.7811,
      "step": 48200
    },
    {
      "epoch": 0.750465099986767,
      "grad_norm": 3.969660758972168,
      "learning_rate": 2.626800609626194e-07,
      "loss": 0.7258,
      "step": 48205
    },
    {
      "epoch": 0.7505429409887364,
      "grad_norm": 3.3275885581970215,
      "learning_rate": 2.6259812195801443e-07,
      "loss": 0.6986,
      "step": 48210
    },
    {
      "epoch": 0.7506207819907058,
      "grad_norm": 6.462599754333496,
      "learning_rate": 2.625161829534095e-07,
      "loss": 0.8426,
      "step": 48215
    },
    {
      "epoch": 0.7506986229926752,
      "grad_norm": 5.5715765953063965,
      "learning_rate": 2.6243424394880447e-07,
      "loss": 0.6928,
      "step": 48220
    },
    {
      "epoch": 0.7507764639946446,
      "grad_norm": 12.188241004943848,
      "learning_rate": 2.623523049441995e-07,
      "loss": 0.6368,
      "step": 48225
    },
    {
      "epoch": 0.7508543049966139,
      "grad_norm": 4.682760238647461,
      "learning_rate": 2.6227036593959457e-07,
      "loss": 0.8275,
      "step": 48230
    },
    {
      "epoch": 0.7509321459985833,
      "grad_norm": 3.3816568851470947,
      "learning_rate": 2.621884269349896e-07,
      "loss": 0.7562,
      "step": 48235
    },
    {
      "epoch": 0.7510099870005527,
      "grad_norm": 4.146965503692627,
      "learning_rate": 2.621064879303846e-07,
      "loss": 0.7117,
      "step": 48240
    },
    {
      "epoch": 0.751087828002522,
      "grad_norm": 3.4598164558410645,
      "learning_rate": 2.6202454892577964e-07,
      "loss": 0.8756,
      "step": 48245
    },
    {
      "epoch": 0.7511656690044914,
      "grad_norm": 3.8203394412994385,
      "learning_rate": 2.6194260992117466e-07,
      "loss": 0.8172,
      "step": 48250
    },
    {
      "epoch": 0.7512435100064608,
      "grad_norm": 3.4714767932891846,
      "learning_rate": 2.6186067091656973e-07,
      "loss": 0.8811,
      "step": 48255
    },
    {
      "epoch": 0.7513213510084302,
      "grad_norm": 4.726281642913818,
      "learning_rate": 2.617787319119647e-07,
      "loss": 0.7435,
      "step": 48260
    },
    {
      "epoch": 0.7513991920103995,
      "grad_norm": 7.40652322769165,
      "learning_rate": 2.616967929073597e-07,
      "loss": 0.7792,
      "step": 48265
    },
    {
      "epoch": 0.7514770330123689,
      "grad_norm": 4.569562911987305,
      "learning_rate": 2.616148539027548e-07,
      "loss": 0.8253,
      "step": 48270
    },
    {
      "epoch": 0.7515548740143383,
      "grad_norm": 2.924746513366699,
      "learning_rate": 2.615329148981498e-07,
      "loss": 0.6101,
      "step": 48275
    },
    {
      "epoch": 0.7516327150163077,
      "grad_norm": 2.9592092037200928,
      "learning_rate": 2.614509758935448e-07,
      "loss": 0.7404,
      "step": 48280
    },
    {
      "epoch": 0.7517105560182771,
      "grad_norm": 2.8561787605285645,
      "learning_rate": 2.6136903688893986e-07,
      "loss": 0.6884,
      "step": 48285
    },
    {
      "epoch": 0.7517883970202465,
      "grad_norm": 3.988363027572632,
      "learning_rate": 2.612870978843349e-07,
      "loss": 0.7112,
      "step": 48290
    },
    {
      "epoch": 0.7518662380222159,
      "grad_norm": 3.5633351802825928,
      "learning_rate": 2.6120515887972996e-07,
      "loss": 0.6593,
      "step": 48295
    },
    {
      "epoch": 0.7519440790241853,
      "grad_norm": 3.081228494644165,
      "learning_rate": 2.6112321987512493e-07,
      "loss": 0.7692,
      "step": 48300
    },
    {
      "epoch": 0.7520219200261545,
      "grad_norm": 3.9475784301757812,
      "learning_rate": 2.6104128087051995e-07,
      "loss": 0.7671,
      "step": 48305
    },
    {
      "epoch": 0.7520997610281239,
      "grad_norm": 4.52240514755249,
      "learning_rate": 2.60959341865915e-07,
      "loss": 0.7081,
      "step": 48310
    },
    {
      "epoch": 0.7521776020300933,
      "grad_norm": 3.6757264137268066,
      "learning_rate": 2.6087740286131005e-07,
      "loss": 0.7178,
      "step": 48315
    },
    {
      "epoch": 0.7522554430320627,
      "grad_norm": 4.49394416809082,
      "learning_rate": 2.60795463856705e-07,
      "loss": 0.8816,
      "step": 48320
    },
    {
      "epoch": 0.7523332840340321,
      "grad_norm": 6.117863178253174,
      "learning_rate": 2.607135248521001e-07,
      "loss": 0.8477,
      "step": 48325
    },
    {
      "epoch": 0.7524111250360015,
      "grad_norm": 3.8686344623565674,
      "learning_rate": 2.606315858474951e-07,
      "loss": 0.5644,
      "step": 48330
    },
    {
      "epoch": 0.7524889660379709,
      "grad_norm": 4.157417297363281,
      "learning_rate": 2.6054964684289014e-07,
      "loss": 0.777,
      "step": 48335
    },
    {
      "epoch": 0.7525668070399402,
      "grad_norm": 7.127273082733154,
      "learning_rate": 2.6046770783828516e-07,
      "loss": 0.7731,
      "step": 48340
    },
    {
      "epoch": 0.7526446480419096,
      "grad_norm": 8.284809112548828,
      "learning_rate": 2.603857688336802e-07,
      "loss": 0.7154,
      "step": 48345
    },
    {
      "epoch": 0.752722489043879,
      "grad_norm": 2.9273693561553955,
      "learning_rate": 2.603038298290752e-07,
      "loss": 0.8042,
      "step": 48350
    },
    {
      "epoch": 0.7528003300458483,
      "grad_norm": 3.812910318374634,
      "learning_rate": 2.602218908244703e-07,
      "loss": 0.7437,
      "step": 48355
    },
    {
      "epoch": 0.7528781710478177,
      "grad_norm": 4.995997905731201,
      "learning_rate": 2.6013995181986525e-07,
      "loss": 0.7864,
      "step": 48360
    },
    {
      "epoch": 0.7529560120497871,
      "grad_norm": 3.667799949645996,
      "learning_rate": 2.600580128152603e-07,
      "loss": 0.7464,
      "step": 48365
    },
    {
      "epoch": 0.7530338530517565,
      "grad_norm": 2.5335299968719482,
      "learning_rate": 2.5997607381065534e-07,
      "loss": 0.656,
      "step": 48370
    },
    {
      "epoch": 0.7531116940537259,
      "grad_norm": 3.432154655456543,
      "learning_rate": 2.5989413480605036e-07,
      "loss": 0.7162,
      "step": 48375
    },
    {
      "epoch": 0.7531895350556952,
      "grad_norm": 3.534316062927246,
      "learning_rate": 2.598121958014454e-07,
      "loss": 0.6673,
      "step": 48380
    },
    {
      "epoch": 0.7532673760576646,
      "grad_norm": 2.6901772022247314,
      "learning_rate": 2.597302567968404e-07,
      "loss": 0.7534,
      "step": 48385
    },
    {
      "epoch": 0.753345217059634,
      "grad_norm": 6.175843715667725,
      "learning_rate": 2.5964831779223543e-07,
      "loss": 0.6605,
      "step": 48390
    },
    {
      "epoch": 0.7534230580616034,
      "grad_norm": 3.7835397720336914,
      "learning_rate": 2.595663787876305e-07,
      "loss": 0.7025,
      "step": 48395
    },
    {
      "epoch": 0.7535008990635728,
      "grad_norm": 5.323223114013672,
      "learning_rate": 2.594844397830255e-07,
      "loss": 0.7255,
      "step": 48400
    },
    {
      "epoch": 0.7535787400655422,
      "grad_norm": 3.303152084350586,
      "learning_rate": 2.594025007784205e-07,
      "loss": 0.7385,
      "step": 48405
    },
    {
      "epoch": 0.7536565810675115,
      "grad_norm": 4.871014595031738,
      "learning_rate": 2.5932056177381557e-07,
      "loss": 0.6444,
      "step": 48410
    },
    {
      "epoch": 0.7537344220694808,
      "grad_norm": 4.795705318450928,
      "learning_rate": 2.592386227692106e-07,
      "loss": 0.7281,
      "step": 48415
    },
    {
      "epoch": 0.7538122630714502,
      "grad_norm": 2.5301849842071533,
      "learning_rate": 2.591566837646056e-07,
      "loss": 0.7874,
      "step": 48420
    },
    {
      "epoch": 0.7538901040734196,
      "grad_norm": 2.875927448272705,
      "learning_rate": 2.5907474476000064e-07,
      "loss": 0.6898,
      "step": 48425
    },
    {
      "epoch": 0.753967945075389,
      "grad_norm": 3.0883944034576416,
      "learning_rate": 2.5899280575539566e-07,
      "loss": 0.6554,
      "step": 48430
    },
    {
      "epoch": 0.7540457860773584,
      "grad_norm": 3.4719669818878174,
      "learning_rate": 2.5891086675079073e-07,
      "loss": 0.7156,
      "step": 48435
    },
    {
      "epoch": 0.7541236270793278,
      "grad_norm": 3.3315792083740234,
      "learning_rate": 2.5882892774618575e-07,
      "loss": 0.6996,
      "step": 48440
    },
    {
      "epoch": 0.7542014680812972,
      "grad_norm": 5.981139659881592,
      "learning_rate": 2.587469887415807e-07,
      "loss": 0.724,
      "step": 48445
    },
    {
      "epoch": 0.7542793090832666,
      "grad_norm": 3.6438074111938477,
      "learning_rate": 2.586650497369758e-07,
      "loss": 0.7332,
      "step": 48450
    },
    {
      "epoch": 0.7543571500852359,
      "grad_norm": 2.8766863346099854,
      "learning_rate": 2.585831107323708e-07,
      "loss": 0.6718,
      "step": 48455
    },
    {
      "epoch": 0.7544349910872052,
      "grad_norm": 4.070430278778076,
      "learning_rate": 2.5850117172776584e-07,
      "loss": 0.646,
      "step": 48460
    },
    {
      "epoch": 0.7545128320891746,
      "grad_norm": 3.222543478012085,
      "learning_rate": 2.5841923272316086e-07,
      "loss": 0.736,
      "step": 48465
    },
    {
      "epoch": 0.754590673091144,
      "grad_norm": 5.901883602142334,
      "learning_rate": 2.583372937185559e-07,
      "loss": 0.5671,
      "step": 48470
    },
    {
      "epoch": 0.7546685140931134,
      "grad_norm": 5.67926549911499,
      "learning_rate": 2.5825535471395096e-07,
      "loss": 0.8773,
      "step": 48475
    },
    {
      "epoch": 0.7547463550950828,
      "grad_norm": 2.06284499168396,
      "learning_rate": 2.58173415709346e-07,
      "loss": 0.7026,
      "step": 48480
    },
    {
      "epoch": 0.7548241960970522,
      "grad_norm": 3.3181986808776855,
      "learning_rate": 2.5809147670474095e-07,
      "loss": 0.8565,
      "step": 48485
    },
    {
      "epoch": 0.7549020370990215,
      "grad_norm": 3.349825143814087,
      "learning_rate": 2.5800953770013603e-07,
      "loss": 0.7209,
      "step": 48490
    },
    {
      "epoch": 0.7549798781009909,
      "grad_norm": 4.362368106842041,
      "learning_rate": 2.5792759869553105e-07,
      "loss": 0.8218,
      "step": 48495
    },
    {
      "epoch": 0.7550577191029603,
      "grad_norm": 2.1099157333374023,
      "learning_rate": 2.5784565969092607e-07,
      "loss": 0.6931,
      "step": 48500
    },
    {
      "epoch": 0.7551355601049297,
      "grad_norm": 3.0572030544281006,
      "learning_rate": 2.577637206863211e-07,
      "loss": 0.7007,
      "step": 48505
    },
    {
      "epoch": 0.755213401106899,
      "grad_norm": 5.706721305847168,
      "learning_rate": 2.576817816817161e-07,
      "loss": 0.7601,
      "step": 48510
    },
    {
      "epoch": 0.7552912421088684,
      "grad_norm": 4.101369857788086,
      "learning_rate": 2.5759984267711114e-07,
      "loss": 0.7799,
      "step": 48515
    },
    {
      "epoch": 0.7553690831108378,
      "grad_norm": 3.2605559825897217,
      "learning_rate": 2.575179036725062e-07,
      "loss": 0.7163,
      "step": 48520
    },
    {
      "epoch": 0.7554469241128072,
      "grad_norm": 4.863544940948486,
      "learning_rate": 2.574359646679012e-07,
      "loss": 0.7141,
      "step": 48525
    },
    {
      "epoch": 0.7555247651147765,
      "grad_norm": 4.439524173736572,
      "learning_rate": 2.573540256632962e-07,
      "loss": 0.6872,
      "step": 48530
    },
    {
      "epoch": 0.7556026061167459,
      "grad_norm": 5.1897101402282715,
      "learning_rate": 2.572720866586913e-07,
      "loss": 0.6383,
      "step": 48535
    },
    {
      "epoch": 0.7556804471187153,
      "grad_norm": 3.448988676071167,
      "learning_rate": 2.571901476540863e-07,
      "loss": 0.6381,
      "step": 48540
    },
    {
      "epoch": 0.7557582881206847,
      "grad_norm": 4.469512462615967,
      "learning_rate": 2.571082086494813e-07,
      "loss": 0.7273,
      "step": 48545
    },
    {
      "epoch": 0.7558361291226541,
      "grad_norm": 3.9337432384490967,
      "learning_rate": 2.5702626964487634e-07,
      "loss": 0.7939,
      "step": 48550
    },
    {
      "epoch": 0.7559139701246235,
      "grad_norm": 8.526634216308594,
      "learning_rate": 2.5694433064027136e-07,
      "loss": 0.7888,
      "step": 48555
    },
    {
      "epoch": 0.7559918111265929,
      "grad_norm": 3.8542537689208984,
      "learning_rate": 2.5686239163566644e-07,
      "loss": 0.6254,
      "step": 48560
    },
    {
      "epoch": 0.7560696521285623,
      "grad_norm": 4.012295722961426,
      "learning_rate": 2.567804526310614e-07,
      "loss": 0.8641,
      "step": 48565
    },
    {
      "epoch": 0.7561474931305315,
      "grad_norm": 4.594897747039795,
      "learning_rate": 2.5669851362645643e-07,
      "loss": 0.7293,
      "step": 48570
    },
    {
      "epoch": 0.7562253341325009,
      "grad_norm": 3.3328113555908203,
      "learning_rate": 2.566165746218515e-07,
      "loss": 0.7502,
      "step": 48575
    },
    {
      "epoch": 0.7563031751344703,
      "grad_norm": 3.522597312927246,
      "learning_rate": 2.5653463561724653e-07,
      "loss": 0.7814,
      "step": 48580
    },
    {
      "epoch": 0.7563810161364397,
      "grad_norm": 3.1570746898651123,
      "learning_rate": 2.564526966126415e-07,
      "loss": 0.7442,
      "step": 48585
    },
    {
      "epoch": 0.7564588571384091,
      "grad_norm": 3.2185089588165283,
      "learning_rate": 2.5637075760803657e-07,
      "loss": 0.8301,
      "step": 48590
    },
    {
      "epoch": 0.7565366981403785,
      "grad_norm": 4.23042631149292,
      "learning_rate": 2.562888186034316e-07,
      "loss": 0.7493,
      "step": 48595
    },
    {
      "epoch": 0.7566145391423479,
      "grad_norm": 7.4301676750183105,
      "learning_rate": 2.5620687959882667e-07,
      "loss": 0.7323,
      "step": 48600
    },
    {
      "epoch": 0.7566923801443172,
      "grad_norm": 3.3372926712036133,
      "learning_rate": 2.5612494059422164e-07,
      "loss": 0.5809,
      "step": 48605
    },
    {
      "epoch": 0.7567702211462866,
      "grad_norm": 3.1729896068573,
      "learning_rate": 2.5604300158961666e-07,
      "loss": 0.6356,
      "step": 48610
    },
    {
      "epoch": 0.756848062148256,
      "grad_norm": 3.200605869293213,
      "learning_rate": 2.5596106258501173e-07,
      "loss": 0.7315,
      "step": 48615
    },
    {
      "epoch": 0.7569259031502253,
      "grad_norm": 3.502124547958374,
      "learning_rate": 2.5587912358040676e-07,
      "loss": 0.6895,
      "step": 48620
    },
    {
      "epoch": 0.7570037441521947,
      "grad_norm": 4.398038387298584,
      "learning_rate": 2.557971845758017e-07,
      "loss": 0.8461,
      "step": 48625
    },
    {
      "epoch": 0.7570815851541641,
      "grad_norm": 3.0940768718719482,
      "learning_rate": 2.557152455711968e-07,
      "loss": 0.7397,
      "step": 48630
    },
    {
      "epoch": 0.7571594261561335,
      "grad_norm": 4.514076232910156,
      "learning_rate": 2.556333065665918e-07,
      "loss": 0.8103,
      "step": 48635
    },
    {
      "epoch": 0.7572372671581029,
      "grad_norm": 2.814715623855591,
      "learning_rate": 2.5555136756198684e-07,
      "loss": 0.6773,
      "step": 48640
    },
    {
      "epoch": 0.7573151081600722,
      "grad_norm": 3.565088987350464,
      "learning_rate": 2.5546942855738187e-07,
      "loss": 0.6974,
      "step": 48645
    },
    {
      "epoch": 0.7573929491620416,
      "grad_norm": 3.690486192703247,
      "learning_rate": 2.553874895527769e-07,
      "loss": 0.6697,
      "step": 48650
    },
    {
      "epoch": 0.757470790164011,
      "grad_norm": 3.513007879257202,
      "learning_rate": 2.553055505481719e-07,
      "loss": 0.7729,
      "step": 48655
    },
    {
      "epoch": 0.7575486311659804,
      "grad_norm": 3.003901720046997,
      "learning_rate": 2.55223611543567e-07,
      "loss": 0.7619,
      "step": 48660
    },
    {
      "epoch": 0.7576264721679498,
      "grad_norm": 4.0286641120910645,
      "learning_rate": 2.5514167253896195e-07,
      "loss": 0.7308,
      "step": 48665
    },
    {
      "epoch": 0.7577043131699192,
      "grad_norm": 3.933617115020752,
      "learning_rate": 2.5505973353435703e-07,
      "loss": 0.8099,
      "step": 48670
    },
    {
      "epoch": 0.7577821541718885,
      "grad_norm": 3.2504937648773193,
      "learning_rate": 2.5497779452975205e-07,
      "loss": 0.746,
      "step": 48675
    },
    {
      "epoch": 0.7578599951738578,
      "grad_norm": 4.5149922370910645,
      "learning_rate": 2.5489585552514707e-07,
      "loss": 0.7411,
      "step": 48680
    },
    {
      "epoch": 0.7579378361758272,
      "grad_norm": 3.5665762424468994,
      "learning_rate": 2.548139165205421e-07,
      "loss": 0.7092,
      "step": 48685
    },
    {
      "epoch": 0.7580156771777966,
      "grad_norm": 4.673671245574951,
      "learning_rate": 2.547319775159371e-07,
      "loss": 0.7207,
      "step": 48690
    },
    {
      "epoch": 0.758093518179766,
      "grad_norm": 5.287565231323242,
      "learning_rate": 2.5465003851133214e-07,
      "loss": 0.7873,
      "step": 48695
    },
    {
      "epoch": 0.7581713591817354,
      "grad_norm": 3.8340790271759033,
      "learning_rate": 2.545680995067272e-07,
      "loss": 0.758,
      "step": 48700
    },
    {
      "epoch": 0.7582492001837048,
      "grad_norm": 3.269801139831543,
      "learning_rate": 2.544861605021222e-07,
      "loss": 0.6957,
      "step": 48705
    },
    {
      "epoch": 0.7583270411856742,
      "grad_norm": 2.376138925552368,
      "learning_rate": 2.544042214975172e-07,
      "loss": 0.662,
      "step": 48710
    },
    {
      "epoch": 0.7584048821876436,
      "grad_norm": 4.289655685424805,
      "learning_rate": 2.543222824929123e-07,
      "loss": 0.7423,
      "step": 48715
    },
    {
      "epoch": 0.7584827231896129,
      "grad_norm": 2.8355417251586914,
      "learning_rate": 2.542403434883073e-07,
      "loss": 0.8113,
      "step": 48720
    },
    {
      "epoch": 0.7585605641915822,
      "grad_norm": 14.35962963104248,
      "learning_rate": 2.5415840448370227e-07,
      "loss": 0.7237,
      "step": 48725
    },
    {
      "epoch": 0.7586384051935516,
      "grad_norm": 3.510697364807129,
      "learning_rate": 2.5407646547909734e-07,
      "loss": 0.6913,
      "step": 48730
    },
    {
      "epoch": 0.758716246195521,
      "grad_norm": 14.052264213562012,
      "learning_rate": 2.5399452647449237e-07,
      "loss": 0.8477,
      "step": 48735
    },
    {
      "epoch": 0.7587940871974904,
      "grad_norm": 10.11467456817627,
      "learning_rate": 2.5391258746988744e-07,
      "loss": 0.7973,
      "step": 48740
    },
    {
      "epoch": 0.7588719281994598,
      "grad_norm": 3.387511730194092,
      "learning_rate": 2.5383064846528246e-07,
      "loss": 0.6745,
      "step": 48745
    },
    {
      "epoch": 0.7589497692014292,
      "grad_norm": 5.907968521118164,
      "learning_rate": 2.5374870946067743e-07,
      "loss": 0.7116,
      "step": 48750
    },
    {
      "epoch": 0.7590276102033985,
      "grad_norm": 4.174731731414795,
      "learning_rate": 2.536667704560725e-07,
      "loss": 0.7733,
      "step": 48755
    },
    {
      "epoch": 0.7591054512053679,
      "grad_norm": 14.381922721862793,
      "learning_rate": 2.5358483145146753e-07,
      "loss": 0.9245,
      "step": 48760
    },
    {
      "epoch": 0.7591832922073373,
      "grad_norm": 8.29033088684082,
      "learning_rate": 2.5350289244686255e-07,
      "loss": 0.7251,
      "step": 48765
    },
    {
      "epoch": 0.7592611332093067,
      "grad_norm": 3.1846120357513428,
      "learning_rate": 2.5342095344225757e-07,
      "loss": 0.8808,
      "step": 48770
    },
    {
      "epoch": 0.759338974211276,
      "grad_norm": 3.2413275241851807,
      "learning_rate": 2.533390144376526e-07,
      "loss": 0.7913,
      "step": 48775
    },
    {
      "epoch": 0.7594168152132454,
      "grad_norm": 4.933974742889404,
      "learning_rate": 2.532570754330476e-07,
      "loss": 0.7255,
      "step": 48780
    },
    {
      "epoch": 0.7594946562152148,
      "grad_norm": 3.4503118991851807,
      "learning_rate": 2.531751364284427e-07,
      "loss": 0.7494,
      "step": 48785
    },
    {
      "epoch": 0.7595724972171842,
      "grad_norm": 3.1261160373687744,
      "learning_rate": 2.5309319742383766e-07,
      "loss": 0.7142,
      "step": 48790
    },
    {
      "epoch": 0.7596503382191535,
      "grad_norm": 3.5265305042266846,
      "learning_rate": 2.5301125841923273e-07,
      "loss": 0.6944,
      "step": 48795
    },
    {
      "epoch": 0.7597281792211229,
      "grad_norm": 3.851320505142212,
      "learning_rate": 2.5292931941462776e-07,
      "loss": 0.6544,
      "step": 48800
    },
    {
      "epoch": 0.7598060202230923,
      "grad_norm": 3.3573648929595947,
      "learning_rate": 2.528473804100228e-07,
      "loss": 0.7459,
      "step": 48805
    },
    {
      "epoch": 0.7598838612250617,
      "grad_norm": 3.8234846591949463,
      "learning_rate": 2.527654414054178e-07,
      "loss": 0.8468,
      "step": 48810
    },
    {
      "epoch": 0.7599617022270311,
      "grad_norm": 3.99474835395813,
      "learning_rate": 2.526835024008128e-07,
      "loss": 0.7002,
      "step": 48815
    },
    {
      "epoch": 0.7600395432290005,
      "grad_norm": 3.2376253604888916,
      "learning_rate": 2.5260156339620784e-07,
      "loss": 0.7739,
      "step": 48820
    },
    {
      "epoch": 0.7601173842309699,
      "grad_norm": 2.8744630813598633,
      "learning_rate": 2.525196243916029e-07,
      "loss": 0.6676,
      "step": 48825
    },
    {
      "epoch": 0.7601952252329393,
      "grad_norm": 3.240936040878296,
      "learning_rate": 2.524376853869979e-07,
      "loss": 0.6417,
      "step": 48830
    },
    {
      "epoch": 0.7602730662349085,
      "grad_norm": 3.552718162536621,
      "learning_rate": 2.523557463823929e-07,
      "loss": 0.7524,
      "step": 48835
    },
    {
      "epoch": 0.7603509072368779,
      "grad_norm": 4.020523548126221,
      "learning_rate": 2.52273807377788e-07,
      "loss": 0.724,
      "step": 48840
    },
    {
      "epoch": 0.7604287482388473,
      "grad_norm": 5.739832878112793,
      "learning_rate": 2.52191868373183e-07,
      "loss": 0.6617,
      "step": 48845
    },
    {
      "epoch": 0.7605065892408167,
      "grad_norm": 3.2421019077301025,
      "learning_rate": 2.52109929368578e-07,
      "loss": 0.7672,
      "step": 48850
    },
    {
      "epoch": 0.7605844302427861,
      "grad_norm": 3.6157326698303223,
      "learning_rate": 2.5202799036397305e-07,
      "loss": 0.7671,
      "step": 48855
    },
    {
      "epoch": 0.7606622712447555,
      "grad_norm": 4.4948320388793945,
      "learning_rate": 2.5194605135936807e-07,
      "loss": 0.7145,
      "step": 48860
    },
    {
      "epoch": 0.7607401122467249,
      "grad_norm": 3.170400619506836,
      "learning_rate": 2.5186411235476315e-07,
      "loss": 0.5919,
      "step": 48865
    },
    {
      "epoch": 0.7608179532486942,
      "grad_norm": 7.510459899902344,
      "learning_rate": 2.517821733501581e-07,
      "loss": 0.8359,
      "step": 48870
    },
    {
      "epoch": 0.7608957942506636,
      "grad_norm": 2.8163902759552,
      "learning_rate": 2.5170023434555314e-07,
      "loss": 0.7952,
      "step": 48875
    },
    {
      "epoch": 0.760973635252633,
      "grad_norm": 3.6466588973999023,
      "learning_rate": 2.516182953409482e-07,
      "loss": 0.7341,
      "step": 48880
    },
    {
      "epoch": 0.7610514762546023,
      "grad_norm": 4.376975059509277,
      "learning_rate": 2.5153635633634324e-07,
      "loss": 0.7014,
      "step": 48885
    },
    {
      "epoch": 0.7611293172565717,
      "grad_norm": 5.088446140289307,
      "learning_rate": 2.514544173317382e-07,
      "loss": 0.8135,
      "step": 48890
    },
    {
      "epoch": 0.7612071582585411,
      "grad_norm": 4.059525012969971,
      "learning_rate": 2.513724783271333e-07,
      "loss": 0.7638,
      "step": 48895
    },
    {
      "epoch": 0.7612849992605105,
      "grad_norm": 2.8197181224823,
      "learning_rate": 2.512905393225283e-07,
      "loss": 0.7003,
      "step": 48900
    },
    {
      "epoch": 0.7613628402624799,
      "grad_norm": 3.6158533096313477,
      "learning_rate": 2.512086003179233e-07,
      "loss": 0.6531,
      "step": 48905
    },
    {
      "epoch": 0.7614406812644492,
      "grad_norm": 6.0726704597473145,
      "learning_rate": 2.5112666131331834e-07,
      "loss": 0.7545,
      "step": 48910
    },
    {
      "epoch": 0.7615185222664186,
      "grad_norm": 7.751058578491211,
      "learning_rate": 2.5104472230871337e-07,
      "loss": 0.653,
      "step": 48915
    },
    {
      "epoch": 0.761596363268388,
      "grad_norm": 4.0223870277404785,
      "learning_rate": 2.5096278330410844e-07,
      "loss": 0.7714,
      "step": 48920
    },
    {
      "epoch": 0.7616742042703574,
      "grad_norm": 2.679323673248291,
      "learning_rate": 2.5088084429950346e-07,
      "loss": 0.6609,
      "step": 48925
    },
    {
      "epoch": 0.7617520452723268,
      "grad_norm": 3.08990216255188,
      "learning_rate": 2.5079890529489843e-07,
      "loss": 0.5697,
      "step": 48930
    },
    {
      "epoch": 0.7618298862742962,
      "grad_norm": 3.3288497924804688,
      "learning_rate": 2.507169662902935e-07,
      "loss": 0.7623,
      "step": 48935
    },
    {
      "epoch": 0.7619077272762655,
      "grad_norm": 3.6989262104034424,
      "learning_rate": 2.5063502728568853e-07,
      "loss": 0.6164,
      "step": 48940
    },
    {
      "epoch": 0.7619855682782348,
      "grad_norm": 3.573164463043213,
      "learning_rate": 2.5055308828108355e-07,
      "loss": 0.7611,
      "step": 48945
    },
    {
      "epoch": 0.7620634092802042,
      "grad_norm": 4.420063018798828,
      "learning_rate": 2.5047114927647857e-07,
      "loss": 0.6603,
      "step": 48950
    },
    {
      "epoch": 0.7621412502821736,
      "grad_norm": 3.6368908882141113,
      "learning_rate": 2.503892102718736e-07,
      "loss": 0.7827,
      "step": 48955
    },
    {
      "epoch": 0.762219091284143,
      "grad_norm": 3.685645580291748,
      "learning_rate": 2.503072712672686e-07,
      "loss": 0.8144,
      "step": 48960
    },
    {
      "epoch": 0.7622969322861124,
      "grad_norm": 5.540101528167725,
      "learning_rate": 2.502253322626637e-07,
      "loss": 0.7388,
      "step": 48965
    },
    {
      "epoch": 0.7623747732880818,
      "grad_norm": 2.4716320037841797,
      "learning_rate": 2.5014339325805866e-07,
      "loss": 0.6478,
      "step": 48970
    },
    {
      "epoch": 0.7624526142900512,
      "grad_norm": 4.779886722564697,
      "learning_rate": 2.500614542534537e-07,
      "loss": 0.6464,
      "step": 48975
    },
    {
      "epoch": 0.7625304552920206,
      "grad_norm": 5.363478660583496,
      "learning_rate": 2.4997951524884876e-07,
      "loss": 0.7088,
      "step": 48980
    },
    {
      "epoch": 0.7626082962939899,
      "grad_norm": 2.9699647426605225,
      "learning_rate": 2.498975762442438e-07,
      "loss": 0.6759,
      "step": 48985
    },
    {
      "epoch": 0.7626861372959592,
      "grad_norm": 3.7940964698791504,
      "learning_rate": 2.498156372396388e-07,
      "loss": 0.6954,
      "step": 48990
    },
    {
      "epoch": 0.7627639782979286,
      "grad_norm": 2.6180057525634766,
      "learning_rate": 2.497336982350338e-07,
      "loss": 0.738,
      "step": 48995
    },
    {
      "epoch": 0.762841819299898,
      "grad_norm": 3.1249663829803467,
      "learning_rate": 2.4965175923042885e-07,
      "loss": 0.6888,
      "step": 49000
    },
    {
      "epoch": 0.7629196603018674,
      "grad_norm": 3.2859630584716797,
      "learning_rate": 2.4956982022582387e-07,
      "loss": 0.731,
      "step": 49005
    },
    {
      "epoch": 0.7629975013038368,
      "grad_norm": 2.845222234725952,
      "learning_rate": 2.4948788122121894e-07,
      "loss": 0.59,
      "step": 49010
    },
    {
      "epoch": 0.7630753423058062,
      "grad_norm": 4.130507469177246,
      "learning_rate": 2.494059422166139e-07,
      "loss": 0.686,
      "step": 49015
    },
    {
      "epoch": 0.7631531833077755,
      "grad_norm": 2.6772913932800293,
      "learning_rate": 2.49324003212009e-07,
      "loss": 0.8755,
      "step": 49020
    },
    {
      "epoch": 0.7632310243097449,
      "grad_norm": 2.790729522705078,
      "learning_rate": 2.49242064207404e-07,
      "loss": 0.7286,
      "step": 49025
    },
    {
      "epoch": 0.7633088653117143,
      "grad_norm": 4.471197605133057,
      "learning_rate": 2.4916012520279903e-07,
      "loss": 0.9139,
      "step": 49030
    },
    {
      "epoch": 0.7633867063136837,
      "grad_norm": 4.599584102630615,
      "learning_rate": 2.4907818619819405e-07,
      "loss": 0.7516,
      "step": 49035
    },
    {
      "epoch": 0.763464547315653,
      "grad_norm": 3.1316299438476562,
      "learning_rate": 2.4899624719358907e-07,
      "loss": 0.6882,
      "step": 49040
    },
    {
      "epoch": 0.7635423883176224,
      "grad_norm": 4.672937393188477,
      "learning_rate": 2.489143081889841e-07,
      "loss": 0.7255,
      "step": 49045
    },
    {
      "epoch": 0.7636202293195918,
      "grad_norm": 3.3009822368621826,
      "learning_rate": 2.488323691843791e-07,
      "loss": 0.7638,
      "step": 49050
    },
    {
      "epoch": 0.7636980703215612,
      "grad_norm": 5.565947532653809,
      "learning_rate": 2.4875043017977414e-07,
      "loss": 0.664,
      "step": 49055
    },
    {
      "epoch": 0.7637759113235305,
      "grad_norm": 3.697021722793579,
      "learning_rate": 2.486684911751692e-07,
      "loss": 0.7348,
      "step": 49060
    },
    {
      "epoch": 0.7638537523254999,
      "grad_norm": 2.7236220836639404,
      "learning_rate": 2.4858655217056424e-07,
      "loss": 0.6639,
      "step": 49065
    },
    {
      "epoch": 0.7639315933274693,
      "grad_norm": 2.8520610332489014,
      "learning_rate": 2.4850461316595926e-07,
      "loss": 0.7582,
      "step": 49070
    },
    {
      "epoch": 0.7640094343294387,
      "grad_norm": 3.5481178760528564,
      "learning_rate": 2.484226741613543e-07,
      "loss": 0.7012,
      "step": 49075
    },
    {
      "epoch": 0.7640872753314081,
      "grad_norm": 5.888685703277588,
      "learning_rate": 2.483407351567493e-07,
      "loss": 0.6849,
      "step": 49080
    },
    {
      "epoch": 0.7641651163333775,
      "grad_norm": 4.974430084228516,
      "learning_rate": 2.482587961521443e-07,
      "loss": 0.7505,
      "step": 49085
    },
    {
      "epoch": 0.7642429573353469,
      "grad_norm": 4.014532089233398,
      "learning_rate": 2.4817685714753935e-07,
      "loss": 0.8265,
      "step": 49090
    },
    {
      "epoch": 0.7643207983373161,
      "grad_norm": 5.152015209197998,
      "learning_rate": 2.480949181429344e-07,
      "loss": 0.8399,
      "step": 49095
    },
    {
      "epoch": 0.7643986393392855,
      "grad_norm": 4.340140342712402,
      "learning_rate": 2.480129791383294e-07,
      "loss": 0.7328,
      "step": 49100
    },
    {
      "epoch": 0.7644764803412549,
      "grad_norm": 3.7381668090820312,
      "learning_rate": 2.4793104013372446e-07,
      "loss": 0.7847,
      "step": 49105
    },
    {
      "epoch": 0.7645543213432243,
      "grad_norm": 3.4899659156799316,
      "learning_rate": 2.478491011291195e-07,
      "loss": 0.8283,
      "step": 49110
    },
    {
      "epoch": 0.7646321623451937,
      "grad_norm": 5.513186931610107,
      "learning_rate": 2.477671621245145e-07,
      "loss": 0.7563,
      "step": 49115
    },
    {
      "epoch": 0.7647100033471631,
      "grad_norm": 3.7160868644714355,
      "learning_rate": 2.4768522311990953e-07,
      "loss": 0.6455,
      "step": 49120
    },
    {
      "epoch": 0.7647878443491325,
      "grad_norm": 4.990394115447998,
      "learning_rate": 2.4760328411530455e-07,
      "loss": 0.8784,
      "step": 49125
    },
    {
      "epoch": 0.7648656853511019,
      "grad_norm": 5.105183124542236,
      "learning_rate": 2.475213451106996e-07,
      "loss": 0.7584,
      "step": 49130
    },
    {
      "epoch": 0.7649435263530712,
      "grad_norm": 8.322501182556152,
      "learning_rate": 2.4743940610609465e-07,
      "loss": 0.7756,
      "step": 49135
    },
    {
      "epoch": 0.7650213673550406,
      "grad_norm": 2.990161418914795,
      "learning_rate": 2.473574671014896e-07,
      "loss": 0.8368,
      "step": 49140
    },
    {
      "epoch": 0.76509920835701,
      "grad_norm": 4.686463832855225,
      "learning_rate": 2.472755280968847e-07,
      "loss": 0.7305,
      "step": 49145
    },
    {
      "epoch": 0.7651770493589793,
      "grad_norm": 2.918755292892456,
      "learning_rate": 2.471935890922797e-07,
      "loss": 0.8883,
      "step": 49150
    },
    {
      "epoch": 0.7652548903609487,
      "grad_norm": 3.0047659873962402,
      "learning_rate": 2.4711165008767474e-07,
      "loss": 0.7918,
      "step": 49155
    },
    {
      "epoch": 0.7653327313629181,
      "grad_norm": 3.8076112270355225,
      "learning_rate": 2.4702971108306976e-07,
      "loss": 0.7608,
      "step": 49160
    },
    {
      "epoch": 0.7654105723648875,
      "grad_norm": 4.364243507385254,
      "learning_rate": 2.469477720784648e-07,
      "loss": 0.7472,
      "step": 49165
    },
    {
      "epoch": 0.7654884133668569,
      "grad_norm": 6.164710998535156,
      "learning_rate": 2.468658330738598e-07,
      "loss": 0.7975,
      "step": 49170
    },
    {
      "epoch": 0.7655662543688262,
      "grad_norm": 5.021227836608887,
      "learning_rate": 2.467838940692548e-07,
      "loss": 0.723,
      "step": 49175
    },
    {
      "epoch": 0.7656440953707956,
      "grad_norm": 3.571418046951294,
      "learning_rate": 2.4670195506464985e-07,
      "loss": 0.6569,
      "step": 49180
    },
    {
      "epoch": 0.765721936372765,
      "grad_norm": 4.599207878112793,
      "learning_rate": 2.466200160600449e-07,
      "loss": 0.7421,
      "step": 49185
    },
    {
      "epoch": 0.7657997773747344,
      "grad_norm": 3.685793399810791,
      "learning_rate": 2.465380770554399e-07,
      "loss": 0.8346,
      "step": 49190
    },
    {
      "epoch": 0.7658776183767038,
      "grad_norm": 3.0055902004241943,
      "learning_rate": 2.4645613805083496e-07,
      "loss": 0.6574,
      "step": 49195
    },
    {
      "epoch": 0.7659554593786732,
      "grad_norm": 5.164788246154785,
      "learning_rate": 2.4637419904623e-07,
      "loss": 0.7217,
      "step": 49200
    },
    {
      "epoch": 0.7660333003806425,
      "grad_norm": 3.2760889530181885,
      "learning_rate": 2.46292260041625e-07,
      "loss": 0.6944,
      "step": 49205
    },
    {
      "epoch": 0.7661111413826118,
      "grad_norm": 3.6592044830322266,
      "learning_rate": 2.4621032103702003e-07,
      "loss": 0.8182,
      "step": 49210
    },
    {
      "epoch": 0.7661889823845812,
      "grad_norm": 3.5396876335144043,
      "learning_rate": 2.4612838203241505e-07,
      "loss": 0.7534,
      "step": 49215
    },
    {
      "epoch": 0.7662668233865506,
      "grad_norm": 3.7652127742767334,
      "learning_rate": 2.460464430278101e-07,
      "loss": 0.6839,
      "step": 49220
    },
    {
      "epoch": 0.76634466438852,
      "grad_norm": 5.833047389984131,
      "learning_rate": 2.459645040232051e-07,
      "loss": 0.7933,
      "step": 49225
    },
    {
      "epoch": 0.7664225053904894,
      "grad_norm": 2.6962411403656006,
      "learning_rate": 2.458825650186001e-07,
      "loss": 0.8201,
      "step": 49230
    },
    {
      "epoch": 0.7665003463924588,
      "grad_norm": 3.0270347595214844,
      "learning_rate": 2.458006260139952e-07,
      "loss": 0.7919,
      "step": 49235
    },
    {
      "epoch": 0.7665781873944282,
      "grad_norm": 3.4062485694885254,
      "learning_rate": 2.457186870093902e-07,
      "loss": 0.7235,
      "step": 49240
    },
    {
      "epoch": 0.7666560283963976,
      "grad_norm": 4.263779163360596,
      "learning_rate": 2.4563674800478524e-07,
      "loss": 0.6996,
      "step": 49245
    },
    {
      "epoch": 0.7667338693983669,
      "grad_norm": 4.200270175933838,
      "learning_rate": 2.4555480900018026e-07,
      "loss": 0.7311,
      "step": 49250
    },
    {
      "epoch": 0.7668117104003362,
      "grad_norm": 3.2039566040039062,
      "learning_rate": 2.454728699955753e-07,
      "loss": 0.764,
      "step": 49255
    },
    {
      "epoch": 0.7668895514023056,
      "grad_norm": 4.914599895477295,
      "learning_rate": 2.453909309909703e-07,
      "loss": 0.9022,
      "step": 49260
    },
    {
      "epoch": 0.766967392404275,
      "grad_norm": 4.4890217781066895,
      "learning_rate": 2.453089919863653e-07,
      "loss": 0.6259,
      "step": 49265
    },
    {
      "epoch": 0.7670452334062444,
      "grad_norm": 6.108554363250732,
      "learning_rate": 2.4522705298176035e-07,
      "loss": 0.5756,
      "step": 49270
    },
    {
      "epoch": 0.7671230744082138,
      "grad_norm": 2.5483131408691406,
      "learning_rate": 2.451451139771554e-07,
      "loss": 0.7174,
      "step": 49275
    },
    {
      "epoch": 0.7672009154101832,
      "grad_norm": 2.879093647003174,
      "learning_rate": 2.450631749725504e-07,
      "loss": 0.8018,
      "step": 49280
    },
    {
      "epoch": 0.7672787564121525,
      "grad_norm": 3.30706787109375,
      "learning_rate": 2.4498123596794547e-07,
      "loss": 0.7524,
      "step": 49285
    },
    {
      "epoch": 0.7673565974141219,
      "grad_norm": 2.7918338775634766,
      "learning_rate": 2.448992969633405e-07,
      "loss": 0.6778,
      "step": 49290
    },
    {
      "epoch": 0.7674344384160913,
      "grad_norm": 2.9541919231414795,
      "learning_rate": 2.448173579587355e-07,
      "loss": 0.7291,
      "step": 49295
    },
    {
      "epoch": 0.7675122794180607,
      "grad_norm": 3.871837854385376,
      "learning_rate": 2.4473541895413053e-07,
      "loss": 0.6467,
      "step": 49300
    },
    {
      "epoch": 0.76759012042003,
      "grad_norm": 4.733366012573242,
      "learning_rate": 2.4465347994952555e-07,
      "loss": 0.8346,
      "step": 49305
    },
    {
      "epoch": 0.7676679614219994,
      "grad_norm": 3.907397508621216,
      "learning_rate": 2.445715409449206e-07,
      "loss": 0.7166,
      "step": 49310
    },
    {
      "epoch": 0.7677458024239688,
      "grad_norm": 5.186086654663086,
      "learning_rate": 2.444896019403156e-07,
      "loss": 0.7962,
      "step": 49315
    },
    {
      "epoch": 0.7678236434259382,
      "grad_norm": 3.6739299297332764,
      "learning_rate": 2.444076629357106e-07,
      "loss": 0.8069,
      "step": 49320
    },
    {
      "epoch": 0.7679014844279075,
      "grad_norm": 3.5974411964416504,
      "learning_rate": 2.443257239311057e-07,
      "loss": 0.7224,
      "step": 49325
    },
    {
      "epoch": 0.7679793254298769,
      "grad_norm": 9.54343318939209,
      "learning_rate": 2.442437849265007e-07,
      "loss": 0.7356,
      "step": 49330
    },
    {
      "epoch": 0.7680571664318463,
      "grad_norm": 3.015650749206543,
      "learning_rate": 2.4416184592189574e-07,
      "loss": 0.7421,
      "step": 49335
    },
    {
      "epoch": 0.7681350074338157,
      "grad_norm": 3.118992328643799,
      "learning_rate": 2.4407990691729076e-07,
      "loss": 0.7326,
      "step": 49340
    },
    {
      "epoch": 0.7682128484357851,
      "grad_norm": 4.74088191986084,
      "learning_rate": 2.439979679126858e-07,
      "loss": 0.6069,
      "step": 49345
    },
    {
      "epoch": 0.7682906894377545,
      "grad_norm": 5.363069534301758,
      "learning_rate": 2.439160289080808e-07,
      "loss": 0.7663,
      "step": 49350
    },
    {
      "epoch": 0.7683685304397239,
      "grad_norm": 4.66932487487793,
      "learning_rate": 2.438340899034758e-07,
      "loss": 0.7234,
      "step": 49355
    },
    {
      "epoch": 0.7684463714416931,
      "grad_norm": 4.0302886962890625,
      "learning_rate": 2.4375215089887085e-07,
      "loss": 0.6966,
      "step": 49360
    },
    {
      "epoch": 0.7685242124436625,
      "grad_norm": 4.186161518096924,
      "learning_rate": 2.436702118942659e-07,
      "loss": 0.6641,
      "step": 49365
    },
    {
      "epoch": 0.7686020534456319,
      "grad_norm": 2.6642134189605713,
      "learning_rate": 2.435882728896609e-07,
      "loss": 0.7822,
      "step": 49370
    },
    {
      "epoch": 0.7686798944476013,
      "grad_norm": 3.6317873001098633,
      "learning_rate": 2.4350633388505597e-07,
      "loss": 0.8189,
      "step": 49375
    },
    {
      "epoch": 0.7687577354495707,
      "grad_norm": 2.7458817958831787,
      "learning_rate": 2.43424394880451e-07,
      "loss": 0.6657,
      "step": 49380
    },
    {
      "epoch": 0.7688355764515401,
      "grad_norm": 4.093200206756592,
      "learning_rate": 2.43342455875846e-07,
      "loss": 0.845,
      "step": 49385
    },
    {
      "epoch": 0.7689134174535095,
      "grad_norm": 3.205003023147583,
      "learning_rate": 2.4326051687124103e-07,
      "loss": 0.7763,
      "step": 49390
    },
    {
      "epoch": 0.7689912584554789,
      "grad_norm": 4.691277027130127,
      "learning_rate": 2.4317857786663605e-07,
      "loss": 0.6929,
      "step": 49395
    },
    {
      "epoch": 0.7690690994574482,
      "grad_norm": 4.421628475189209,
      "learning_rate": 2.4309663886203113e-07,
      "loss": 0.6755,
      "step": 49400
    },
    {
      "epoch": 0.7691469404594176,
      "grad_norm": 4.70053768157959,
      "learning_rate": 2.430146998574261e-07,
      "loss": 0.8084,
      "step": 49405
    },
    {
      "epoch": 0.769224781461387,
      "grad_norm": 3.8914616107940674,
      "learning_rate": 2.4293276085282117e-07,
      "loss": 0.7845,
      "step": 49410
    },
    {
      "epoch": 0.7693026224633563,
      "grad_norm": 2.8957951068878174,
      "learning_rate": 2.428508218482162e-07,
      "loss": 0.6018,
      "step": 49415
    },
    {
      "epoch": 0.7693804634653257,
      "grad_norm": 3.9543561935424805,
      "learning_rate": 2.427688828436112e-07,
      "loss": 0.6849,
      "step": 49420
    },
    {
      "epoch": 0.7694583044672951,
      "grad_norm": 3.3527772426605225,
      "learning_rate": 2.4268694383900624e-07,
      "loss": 0.6801,
      "step": 49425
    },
    {
      "epoch": 0.7695361454692645,
      "grad_norm": 4.150762557983398,
      "learning_rate": 2.4260500483440126e-07,
      "loss": 0.774,
      "step": 49430
    },
    {
      "epoch": 0.7696139864712339,
      "grad_norm": 2.674226760864258,
      "learning_rate": 2.425230658297963e-07,
      "loss": 0.8886,
      "step": 49435
    },
    {
      "epoch": 0.7696918274732032,
      "grad_norm": 4.9008965492248535,
      "learning_rate": 2.424411268251913e-07,
      "loss": 0.7933,
      "step": 49440
    },
    {
      "epoch": 0.7697696684751726,
      "grad_norm": 4.012134075164795,
      "learning_rate": 2.423591878205863e-07,
      "loss": 0.7125,
      "step": 49445
    },
    {
      "epoch": 0.769847509477142,
      "grad_norm": 3.35235857963562,
      "learning_rate": 2.422772488159814e-07,
      "loss": 0.7396,
      "step": 49450
    },
    {
      "epoch": 0.7699253504791114,
      "grad_norm": 3.148524522781372,
      "learning_rate": 2.421953098113764e-07,
      "loss": 0.7609,
      "step": 49455
    },
    {
      "epoch": 0.7700031914810808,
      "grad_norm": 3.7128608226776123,
      "learning_rate": 2.4211337080677144e-07,
      "loss": 0.7802,
      "step": 49460
    },
    {
      "epoch": 0.7700810324830502,
      "grad_norm": 4.097230911254883,
      "learning_rate": 2.4203143180216647e-07,
      "loss": 0.6692,
      "step": 49465
    },
    {
      "epoch": 0.7701588734850195,
      "grad_norm": 3.4509012699127197,
      "learning_rate": 2.419494927975615e-07,
      "loss": 0.7139,
      "step": 49470
    },
    {
      "epoch": 0.7702367144869888,
      "grad_norm": 3.780954360961914,
      "learning_rate": 2.418675537929565e-07,
      "loss": 0.8454,
      "step": 49475
    },
    {
      "epoch": 0.7703145554889582,
      "grad_norm": 2.8097803592681885,
      "learning_rate": 2.4178561478835153e-07,
      "loss": 0.6498,
      "step": 49480
    },
    {
      "epoch": 0.7703923964909276,
      "grad_norm": 4.118881702423096,
      "learning_rate": 2.4170367578374655e-07,
      "loss": 0.6203,
      "step": 49485
    },
    {
      "epoch": 0.770470237492897,
      "grad_norm": 3.354949474334717,
      "learning_rate": 2.4162173677914163e-07,
      "loss": 0.7029,
      "step": 49490
    },
    {
      "epoch": 0.7705480784948664,
      "grad_norm": 3.898728847503662,
      "learning_rate": 2.415397977745366e-07,
      "loss": 0.8887,
      "step": 49495
    },
    {
      "epoch": 0.7706259194968358,
      "grad_norm": 7.526986598968506,
      "learning_rate": 2.4145785876993167e-07,
      "loss": 0.746,
      "step": 49500
    },
    {
      "epoch": 0.7707037604988052,
      "grad_norm": 3.384589672088623,
      "learning_rate": 2.413759197653267e-07,
      "loss": 0.81,
      "step": 49505
    },
    {
      "epoch": 0.7707816015007746,
      "grad_norm": 3.6285593509674072,
      "learning_rate": 2.412939807607217e-07,
      "loss": 0.7913,
      "step": 49510
    },
    {
      "epoch": 0.7708594425027439,
      "grad_norm": 3.087144613265991,
      "learning_rate": 2.4121204175611674e-07,
      "loss": 0.6348,
      "step": 49515
    },
    {
      "epoch": 0.7709372835047132,
      "grad_norm": 4.38054084777832,
      "learning_rate": 2.4113010275151176e-07,
      "loss": 0.6817,
      "step": 49520
    },
    {
      "epoch": 0.7710151245066826,
      "grad_norm": 7.17706823348999,
      "learning_rate": 2.410481637469068e-07,
      "loss": 0.698,
      "step": 49525
    },
    {
      "epoch": 0.771092965508652,
      "grad_norm": 2.7040164470672607,
      "learning_rate": 2.409662247423018e-07,
      "loss": 0.7024,
      "step": 49530
    },
    {
      "epoch": 0.7711708065106214,
      "grad_norm": 3.4209563732147217,
      "learning_rate": 2.408842857376968e-07,
      "loss": 0.6984,
      "step": 49535
    },
    {
      "epoch": 0.7712486475125908,
      "grad_norm": 5.506772518157959,
      "learning_rate": 2.408023467330919e-07,
      "loss": 0.656,
      "step": 49540
    },
    {
      "epoch": 0.7713264885145602,
      "grad_norm": 4.22564697265625,
      "learning_rate": 2.407204077284869e-07,
      "loss": 0.7452,
      "step": 49545
    },
    {
      "epoch": 0.7714043295165295,
      "grad_norm": 4.780029773712158,
      "learning_rate": 2.4063846872388194e-07,
      "loss": 0.7027,
      "step": 49550
    },
    {
      "epoch": 0.7714821705184989,
      "grad_norm": 4.376079082489014,
      "learning_rate": 2.4055652971927697e-07,
      "loss": 0.6561,
      "step": 49555
    },
    {
      "epoch": 0.7715600115204683,
      "grad_norm": 3.5744950771331787,
      "learning_rate": 2.40474590714672e-07,
      "loss": 0.7829,
      "step": 49560
    },
    {
      "epoch": 0.7716378525224377,
      "grad_norm": 3.4532699584960938,
      "learning_rate": 2.40392651710067e-07,
      "loss": 0.7337,
      "step": 49565
    },
    {
      "epoch": 0.771715693524407,
      "grad_norm": 3.7450497150421143,
      "learning_rate": 2.4031071270546203e-07,
      "loss": 0.7955,
      "step": 49570
    },
    {
      "epoch": 0.7717935345263764,
      "grad_norm": 3.3204538822174072,
      "learning_rate": 2.4022877370085705e-07,
      "loss": 0.7952,
      "step": 49575
    },
    {
      "epoch": 0.7718713755283458,
      "grad_norm": 4.233084201812744,
      "learning_rate": 2.4014683469625213e-07,
      "loss": 0.7303,
      "step": 49580
    },
    {
      "epoch": 0.7719492165303152,
      "grad_norm": 7.1850433349609375,
      "learning_rate": 2.400648956916471e-07,
      "loss": 0.7707,
      "step": 49585
    },
    {
      "epoch": 0.7720270575322845,
      "grad_norm": 9.02645206451416,
      "learning_rate": 2.3998295668704217e-07,
      "loss": 0.7987,
      "step": 49590
    },
    {
      "epoch": 0.7721048985342539,
      "grad_norm": 3.249011516571045,
      "learning_rate": 2.399010176824372e-07,
      "loss": 0.6815,
      "step": 49595
    },
    {
      "epoch": 0.7721827395362233,
      "grad_norm": 3.1607813835144043,
      "learning_rate": 2.398190786778322e-07,
      "loss": 0.7439,
      "step": 49600
    },
    {
      "epoch": 0.7722605805381927,
      "grad_norm": 3.5115578174591064,
      "learning_rate": 2.3973713967322724e-07,
      "loss": 0.628,
      "step": 49605
    },
    {
      "epoch": 0.7723384215401621,
      "grad_norm": 6.7505269050598145,
      "learning_rate": 2.3965520066862226e-07,
      "loss": 0.8301,
      "step": 49610
    },
    {
      "epoch": 0.7724162625421315,
      "grad_norm": 6.42640495300293,
      "learning_rate": 2.395732616640173e-07,
      "loss": 0.7068,
      "step": 49615
    },
    {
      "epoch": 0.7724941035441009,
      "grad_norm": 4.064393520355225,
      "learning_rate": 2.394913226594123e-07,
      "loss": 0.7901,
      "step": 49620
    },
    {
      "epoch": 0.7725719445460701,
      "grad_norm": 3.200143575668335,
      "learning_rate": 2.394093836548073e-07,
      "loss": 0.669,
      "step": 49625
    },
    {
      "epoch": 0.7726497855480395,
      "grad_norm": 7.63839054107666,
      "learning_rate": 2.393274446502024e-07,
      "loss": 0.7333,
      "step": 49630
    },
    {
      "epoch": 0.7727276265500089,
      "grad_norm": 4.476169586181641,
      "learning_rate": 2.3924550564559737e-07,
      "loss": 0.7005,
      "step": 49635
    },
    {
      "epoch": 0.7728054675519783,
      "grad_norm": 4.466472148895264,
      "learning_rate": 2.3916356664099244e-07,
      "loss": 0.7224,
      "step": 49640
    },
    {
      "epoch": 0.7728833085539477,
      "grad_norm": 3.8245456218719482,
      "learning_rate": 2.3908162763638747e-07,
      "loss": 0.6995,
      "step": 49645
    },
    {
      "epoch": 0.7729611495559171,
      "grad_norm": 3.7528457641601562,
      "learning_rate": 2.389996886317825e-07,
      "loss": 0.6769,
      "step": 49650
    },
    {
      "epoch": 0.7730389905578865,
      "grad_norm": 3.660797595977783,
      "learning_rate": 2.389177496271775e-07,
      "loss": 0.7021,
      "step": 49655
    },
    {
      "epoch": 0.7731168315598559,
      "grad_norm": 4.458802223205566,
      "learning_rate": 2.3883581062257253e-07,
      "loss": 0.7316,
      "step": 49660
    },
    {
      "epoch": 0.7731946725618252,
      "grad_norm": 4.059835433959961,
      "learning_rate": 2.3875387161796755e-07,
      "loss": 0.6966,
      "step": 49665
    },
    {
      "epoch": 0.7732725135637946,
      "grad_norm": 4.9577741622924805,
      "learning_rate": 2.3867193261336263e-07,
      "loss": 0.6935,
      "step": 49670
    },
    {
      "epoch": 0.773350354565764,
      "grad_norm": 6.608840465545654,
      "learning_rate": 2.385899936087576e-07,
      "loss": 0.7606,
      "step": 49675
    },
    {
      "epoch": 0.7734281955677333,
      "grad_norm": 4.704737186431885,
      "learning_rate": 2.3850805460415267e-07,
      "loss": 0.808,
      "step": 49680
    },
    {
      "epoch": 0.7735060365697027,
      "grad_norm": 4.847075462341309,
      "learning_rate": 2.384261155995477e-07,
      "loss": 0.6661,
      "step": 49685
    },
    {
      "epoch": 0.7735838775716721,
      "grad_norm": 4.502778053283691,
      "learning_rate": 2.3834417659494272e-07,
      "loss": 0.6847,
      "step": 49690
    },
    {
      "epoch": 0.7736617185736415,
      "grad_norm": 4.423571586608887,
      "learning_rate": 2.3826223759033774e-07,
      "loss": 0.7008,
      "step": 49695
    },
    {
      "epoch": 0.7737395595756108,
      "grad_norm": 3.947033405303955,
      "learning_rate": 2.3818029858573276e-07,
      "loss": 0.8182,
      "step": 49700
    },
    {
      "epoch": 0.7738174005775802,
      "grad_norm": 2.6030240058898926,
      "learning_rate": 2.380983595811278e-07,
      "loss": 0.76,
      "step": 49705
    },
    {
      "epoch": 0.7738952415795496,
      "grad_norm": 4.222322940826416,
      "learning_rate": 2.3801642057652283e-07,
      "loss": 0.7191,
      "step": 49710
    },
    {
      "epoch": 0.773973082581519,
      "grad_norm": 3.492258310317993,
      "learning_rate": 2.3793448157191788e-07,
      "loss": 0.8191,
      "step": 49715
    },
    {
      "epoch": 0.7740509235834884,
      "grad_norm": 5.074420928955078,
      "learning_rate": 2.3785254256731288e-07,
      "loss": 0.7033,
      "step": 49720
    },
    {
      "epoch": 0.7741287645854578,
      "grad_norm": 4.124150276184082,
      "learning_rate": 2.3777060356270792e-07,
      "loss": 0.7136,
      "step": 49725
    },
    {
      "epoch": 0.7742066055874272,
      "grad_norm": 3.562244176864624,
      "learning_rate": 2.3768866455810295e-07,
      "loss": 0.7133,
      "step": 49730
    },
    {
      "epoch": 0.7742844465893965,
      "grad_norm": 3.2005350589752197,
      "learning_rate": 2.3760672555349797e-07,
      "loss": 0.74,
      "step": 49735
    },
    {
      "epoch": 0.7743622875913658,
      "grad_norm": 2.8821942806243896,
      "learning_rate": 2.37524786548893e-07,
      "loss": 0.7582,
      "step": 49740
    },
    {
      "epoch": 0.7744401285933352,
      "grad_norm": 2.8762593269348145,
      "learning_rate": 2.3744284754428804e-07,
      "loss": 0.7857,
      "step": 49745
    },
    {
      "epoch": 0.7745179695953046,
      "grad_norm": 3.330601692199707,
      "learning_rate": 2.3736090853968306e-07,
      "loss": 0.7445,
      "step": 49750
    },
    {
      "epoch": 0.774595810597274,
      "grad_norm": 5.1199870109558105,
      "learning_rate": 2.3727896953507808e-07,
      "loss": 0.6995,
      "step": 49755
    },
    {
      "epoch": 0.7746736515992434,
      "grad_norm": 5.179469585418701,
      "learning_rate": 2.371970305304731e-07,
      "loss": 0.7857,
      "step": 49760
    },
    {
      "epoch": 0.7747514926012128,
      "grad_norm": 5.0093889236450195,
      "learning_rate": 2.3711509152586815e-07,
      "loss": 0.7117,
      "step": 49765
    },
    {
      "epoch": 0.7748293336031822,
      "grad_norm": 3.888913869857788,
      "learning_rate": 2.3703315252126315e-07,
      "loss": 0.7589,
      "step": 49770
    },
    {
      "epoch": 0.7749071746051516,
      "grad_norm": 5.228609561920166,
      "learning_rate": 2.369512135166582e-07,
      "loss": 0.7827,
      "step": 49775
    },
    {
      "epoch": 0.7749850156071209,
      "grad_norm": 3.6669371128082275,
      "learning_rate": 2.3686927451205322e-07,
      "loss": 0.8126,
      "step": 49780
    },
    {
      "epoch": 0.7750628566090902,
      "grad_norm": 3.11326265335083,
      "learning_rate": 2.3678733550744827e-07,
      "loss": 0.6694,
      "step": 49785
    },
    {
      "epoch": 0.7751406976110596,
      "grad_norm": 3.2052478790283203,
      "learning_rate": 2.3670539650284326e-07,
      "loss": 0.6642,
      "step": 49790
    },
    {
      "epoch": 0.775218538613029,
      "grad_norm": 9.6849365234375,
      "learning_rate": 2.366234574982383e-07,
      "loss": 0.8009,
      "step": 49795
    },
    {
      "epoch": 0.7752963796149984,
      "grad_norm": 2.8290960788726807,
      "learning_rate": 2.3654151849363333e-07,
      "loss": 0.6712,
      "step": 49800
    },
    {
      "epoch": 0.7753742206169678,
      "grad_norm": 4.102849006652832,
      "learning_rate": 2.3645957948902835e-07,
      "loss": 0.6695,
      "step": 49805
    },
    {
      "epoch": 0.7754520616189372,
      "grad_norm": 3.9562175273895264,
      "learning_rate": 2.3637764048442338e-07,
      "loss": 0.8547,
      "step": 49810
    },
    {
      "epoch": 0.7755299026209065,
      "grad_norm": 15.124236106872559,
      "learning_rate": 2.3629570147981842e-07,
      "loss": 0.7895,
      "step": 49815
    },
    {
      "epoch": 0.7756077436228759,
      "grad_norm": 3.8551747798919678,
      "learning_rate": 2.3621376247521345e-07,
      "loss": 0.7436,
      "step": 49820
    },
    {
      "epoch": 0.7756855846248453,
      "grad_norm": 3.7137210369110107,
      "learning_rate": 2.3613182347060847e-07,
      "loss": 0.7448,
      "step": 49825
    },
    {
      "epoch": 0.7757634256268147,
      "grad_norm": 5.220455169677734,
      "learning_rate": 2.360498844660035e-07,
      "loss": 0.7095,
      "step": 49830
    },
    {
      "epoch": 0.775841266628784,
      "grad_norm": 3.6896097660064697,
      "learning_rate": 2.3596794546139854e-07,
      "loss": 0.7281,
      "step": 49835
    },
    {
      "epoch": 0.7759191076307534,
      "grad_norm": 3.7551612854003906,
      "learning_rate": 2.3588600645679353e-07,
      "loss": 0.8312,
      "step": 49840
    },
    {
      "epoch": 0.7759969486327228,
      "grad_norm": 5.13862419128418,
      "learning_rate": 2.3580406745218858e-07,
      "loss": 0.726,
      "step": 49845
    },
    {
      "epoch": 0.7760747896346922,
      "grad_norm": 4.668825626373291,
      "learning_rate": 2.357221284475836e-07,
      "loss": 0.5577,
      "step": 49850
    },
    {
      "epoch": 0.7761526306366615,
      "grad_norm": 8.246150970458984,
      "learning_rate": 2.3564018944297865e-07,
      "loss": 0.8109,
      "step": 49855
    },
    {
      "epoch": 0.7762304716386309,
      "grad_norm": 3.691365957260132,
      "learning_rate": 2.3555825043837365e-07,
      "loss": 0.6302,
      "step": 49860
    },
    {
      "epoch": 0.7763083126406003,
      "grad_norm": 3.074518918991089,
      "learning_rate": 2.354763114337687e-07,
      "loss": 0.7181,
      "step": 49865
    },
    {
      "epoch": 0.7763861536425697,
      "grad_norm": 3.9329559803009033,
      "learning_rate": 2.3539437242916372e-07,
      "loss": 0.645,
      "step": 49870
    },
    {
      "epoch": 0.7764639946445391,
      "grad_norm": 11.183392524719238,
      "learning_rate": 2.3531243342455877e-07,
      "loss": 0.6514,
      "step": 49875
    },
    {
      "epoch": 0.7765418356465085,
      "grad_norm": 3.556567907333374,
      "learning_rate": 2.3523049441995376e-07,
      "loss": 0.7488,
      "step": 49880
    },
    {
      "epoch": 0.7766196766484779,
      "grad_norm": 3.726628541946411,
      "learning_rate": 2.351485554153488e-07,
      "loss": 0.6458,
      "step": 49885
    },
    {
      "epoch": 0.7766975176504471,
      "grad_norm": 3.969853162765503,
      "learning_rate": 2.3506661641074383e-07,
      "loss": 0.6702,
      "step": 49890
    },
    {
      "epoch": 0.7767753586524165,
      "grad_norm": 7.359721660614014,
      "learning_rate": 2.3498467740613885e-07,
      "loss": 0.7069,
      "step": 49895
    },
    {
      "epoch": 0.7768531996543859,
      "grad_norm": 3.0903115272521973,
      "learning_rate": 2.3490273840153388e-07,
      "loss": 0.6461,
      "step": 49900
    },
    {
      "epoch": 0.7769310406563553,
      "grad_norm": 5.798946380615234,
      "learning_rate": 2.3482079939692892e-07,
      "loss": 0.639,
      "step": 49905
    },
    {
      "epoch": 0.7770088816583247,
      "grad_norm": 3.972919225692749,
      "learning_rate": 2.3473886039232395e-07,
      "loss": 0.7725,
      "step": 49910
    },
    {
      "epoch": 0.7770867226602941,
      "grad_norm": 5.170044898986816,
      "learning_rate": 2.3465692138771897e-07,
      "loss": 0.7,
      "step": 49915
    },
    {
      "epoch": 0.7771645636622635,
      "grad_norm": 3.562562942504883,
      "learning_rate": 2.34574982383114e-07,
      "loss": 0.785,
      "step": 49920
    },
    {
      "epoch": 0.7772424046642329,
      "grad_norm": 4.479970455169678,
      "learning_rate": 2.3449304337850904e-07,
      "loss": 0.715,
      "step": 49925
    },
    {
      "epoch": 0.7773202456662022,
      "grad_norm": 11.058489799499512,
      "learning_rate": 2.3441110437390403e-07,
      "loss": 0.7684,
      "step": 49930
    },
    {
      "epoch": 0.7773980866681716,
      "grad_norm": 6.741491794586182,
      "learning_rate": 2.3432916536929908e-07,
      "loss": 0.8159,
      "step": 49935
    },
    {
      "epoch": 0.777475927670141,
      "grad_norm": 3.341219902038574,
      "learning_rate": 2.342472263646941e-07,
      "loss": 0.8033,
      "step": 49940
    },
    {
      "epoch": 0.7775537686721103,
      "grad_norm": 5.101444721221924,
      "learning_rate": 2.3416528736008915e-07,
      "loss": 0.7389,
      "step": 49945
    },
    {
      "epoch": 0.7776316096740797,
      "grad_norm": 5.403138637542725,
      "learning_rate": 2.3408334835548415e-07,
      "loss": 0.7764,
      "step": 49950
    },
    {
      "epoch": 0.7777094506760491,
      "grad_norm": 5.157065391540527,
      "learning_rate": 2.340014093508792e-07,
      "loss": 0.6497,
      "step": 49955
    },
    {
      "epoch": 0.7777872916780185,
      "grad_norm": 3.57837176322937,
      "learning_rate": 2.3391947034627422e-07,
      "loss": 0.7563,
      "step": 49960
    },
    {
      "epoch": 0.7778651326799878,
      "grad_norm": 2.6519217491149902,
      "learning_rate": 2.3383753134166924e-07,
      "loss": 0.7416,
      "step": 49965
    },
    {
      "epoch": 0.7779429736819572,
      "grad_norm": 3.866227626800537,
      "learning_rate": 2.3375559233706426e-07,
      "loss": 0.8068,
      "step": 49970
    },
    {
      "epoch": 0.7780208146839266,
      "grad_norm": 3.695298194885254,
      "learning_rate": 2.336736533324593e-07,
      "loss": 0.7722,
      "step": 49975
    },
    {
      "epoch": 0.778098655685896,
      "grad_norm": 4.00108528137207,
      "learning_rate": 2.3359171432785433e-07,
      "loss": 0.8402,
      "step": 49980
    },
    {
      "epoch": 0.7781764966878654,
      "grad_norm": 2.86110520362854,
      "learning_rate": 2.3350977532324935e-07,
      "loss": 0.6659,
      "step": 49985
    },
    {
      "epoch": 0.7782543376898348,
      "grad_norm": 3.3430674076080322,
      "learning_rate": 2.3342783631864438e-07,
      "loss": 0.6902,
      "step": 49990
    },
    {
      "epoch": 0.7783321786918042,
      "grad_norm": 4.969819068908691,
      "learning_rate": 2.3334589731403942e-07,
      "loss": 0.7592,
      "step": 49995
    },
    {
      "epoch": 0.7784100196937735,
      "grad_norm": 7.861973762512207,
      "learning_rate": 2.3326395830943442e-07,
      "loss": 0.8062,
      "step": 50000
    },
    {
      "epoch": 0.7784878606957428,
      "grad_norm": 3.1127188205718994,
      "learning_rate": 2.3318201930482947e-07,
      "loss": 0.7005,
      "step": 50005
    },
    {
      "epoch": 0.7785657016977122,
      "grad_norm": 2.897697687149048,
      "learning_rate": 2.331000803002245e-07,
      "loss": 0.6577,
      "step": 50010
    },
    {
      "epoch": 0.7786435426996816,
      "grad_norm": 6.348909854888916,
      "learning_rate": 2.3301814129561954e-07,
      "loss": 0.6591,
      "step": 50015
    },
    {
      "epoch": 0.778721383701651,
      "grad_norm": 3.50740647315979,
      "learning_rate": 2.3293620229101456e-07,
      "loss": 0.6913,
      "step": 50020
    },
    {
      "epoch": 0.7787992247036204,
      "grad_norm": 5.595913410186768,
      "learning_rate": 2.3285426328640958e-07,
      "loss": 0.672,
      "step": 50025
    },
    {
      "epoch": 0.7788770657055898,
      "grad_norm": 5.121108531951904,
      "learning_rate": 2.3277232428180463e-07,
      "loss": 0.7539,
      "step": 50030
    },
    {
      "epoch": 0.7789549067075592,
      "grad_norm": 4.253491401672363,
      "learning_rate": 2.3269038527719965e-07,
      "loss": 0.7361,
      "step": 50035
    },
    {
      "epoch": 0.7790327477095286,
      "grad_norm": 2.9214773178100586,
      "learning_rate": 2.3260844627259467e-07,
      "loss": 0.7782,
      "step": 50040
    },
    {
      "epoch": 0.7791105887114979,
      "grad_norm": 2.8831064701080322,
      "learning_rate": 2.325265072679897e-07,
      "loss": 0.7264,
      "step": 50045
    },
    {
      "epoch": 0.7791884297134672,
      "grad_norm": 10.14620304107666,
      "learning_rate": 2.3244456826338475e-07,
      "loss": 0.764,
      "step": 50050
    },
    {
      "epoch": 0.7792662707154366,
      "grad_norm": 5.952034950256348,
      "learning_rate": 2.3236262925877974e-07,
      "loss": 0.7068,
      "step": 50055
    },
    {
      "epoch": 0.779344111717406,
      "grad_norm": 2.8188583850860596,
      "learning_rate": 2.322806902541748e-07,
      "loss": 0.7606,
      "step": 50060
    },
    {
      "epoch": 0.7794219527193754,
      "grad_norm": 8.575106620788574,
      "learning_rate": 2.321987512495698e-07,
      "loss": 0.8053,
      "step": 50065
    },
    {
      "epoch": 0.7794997937213448,
      "grad_norm": 3.10381817817688,
      "learning_rate": 2.3211681224496486e-07,
      "loss": 0.6229,
      "step": 50070
    },
    {
      "epoch": 0.7795776347233142,
      "grad_norm": 2.627490282058716,
      "learning_rate": 2.3203487324035985e-07,
      "loss": 0.7242,
      "step": 50075
    },
    {
      "epoch": 0.7796554757252835,
      "grad_norm": 3.7412009239196777,
      "learning_rate": 2.319529342357549e-07,
      "loss": 0.777,
      "step": 50080
    },
    {
      "epoch": 0.7797333167272529,
      "grad_norm": 9.328071594238281,
      "learning_rate": 2.3187099523114993e-07,
      "loss": 0.7182,
      "step": 50085
    },
    {
      "epoch": 0.7798111577292223,
      "grad_norm": 3.630995988845825,
      "learning_rate": 2.3178905622654495e-07,
      "loss": 0.8931,
      "step": 50090
    },
    {
      "epoch": 0.7798889987311917,
      "grad_norm": 4.071939468383789,
      "learning_rate": 2.3170711722193997e-07,
      "loss": 0.693,
      "step": 50095
    },
    {
      "epoch": 0.779966839733161,
      "grad_norm": 3.9308433532714844,
      "learning_rate": 2.3162517821733502e-07,
      "loss": 0.7511,
      "step": 50100
    },
    {
      "epoch": 0.7800446807351304,
      "grad_norm": 2.92189359664917,
      "learning_rate": 2.3154323921273004e-07,
      "loss": 0.7667,
      "step": 50105
    },
    {
      "epoch": 0.7801225217370998,
      "grad_norm": 3.7951788902282715,
      "learning_rate": 2.3146130020812506e-07,
      "loss": 0.6494,
      "step": 50110
    },
    {
      "epoch": 0.7802003627390692,
      "grad_norm": 4.280236721038818,
      "learning_rate": 2.3137936120352008e-07,
      "loss": 0.7237,
      "step": 50115
    },
    {
      "epoch": 0.7802782037410385,
      "grad_norm": 3.6984400749206543,
      "learning_rate": 2.3129742219891513e-07,
      "loss": 0.8465,
      "step": 50120
    },
    {
      "epoch": 0.7803560447430079,
      "grad_norm": 5.0458664894104,
      "learning_rate": 2.3121548319431015e-07,
      "loss": 0.662,
      "step": 50125
    },
    {
      "epoch": 0.7804338857449773,
      "grad_norm": 3.9302902221679688,
      "learning_rate": 2.3113354418970518e-07,
      "loss": 0.8048,
      "step": 50130
    },
    {
      "epoch": 0.7805117267469467,
      "grad_norm": 4.581846237182617,
      "learning_rate": 2.310516051851002e-07,
      "loss": 0.6743,
      "step": 50135
    },
    {
      "epoch": 0.7805895677489161,
      "grad_norm": 4.528824806213379,
      "learning_rate": 2.3096966618049525e-07,
      "loss": 0.76,
      "step": 50140
    },
    {
      "epoch": 0.7806674087508855,
      "grad_norm": 3.3941500186920166,
      "learning_rate": 2.3088772717589024e-07,
      "loss": 0.7468,
      "step": 50145
    },
    {
      "epoch": 0.7807452497528549,
      "grad_norm": 5.647637844085693,
      "learning_rate": 2.308057881712853e-07,
      "loss": 0.6492,
      "step": 50150
    },
    {
      "epoch": 0.7808230907548241,
      "grad_norm": 4.289590358734131,
      "learning_rate": 2.307238491666803e-07,
      "loss": 0.6441,
      "step": 50155
    },
    {
      "epoch": 0.7809009317567935,
      "grad_norm": 2.5204265117645264,
      "learning_rate": 2.3064191016207536e-07,
      "loss": 0.6201,
      "step": 50160
    },
    {
      "epoch": 0.7809787727587629,
      "grad_norm": 2.9211478233337402,
      "learning_rate": 2.3055997115747036e-07,
      "loss": 0.6577,
      "step": 50165
    },
    {
      "epoch": 0.7810566137607323,
      "grad_norm": 3.473616123199463,
      "learning_rate": 2.304780321528654e-07,
      "loss": 0.7055,
      "step": 50170
    },
    {
      "epoch": 0.7811344547627017,
      "grad_norm": 6.0367889404296875,
      "learning_rate": 2.3039609314826043e-07,
      "loss": 0.7712,
      "step": 50175
    },
    {
      "epoch": 0.7812122957646711,
      "grad_norm": 4.342297077178955,
      "learning_rate": 2.3031415414365545e-07,
      "loss": 0.8016,
      "step": 50180
    },
    {
      "epoch": 0.7812901367666405,
      "grad_norm": 4.306207656860352,
      "learning_rate": 2.3023221513905047e-07,
      "loss": 0.718,
      "step": 50185
    },
    {
      "epoch": 0.7813679777686099,
      "grad_norm": 4.102506160736084,
      "learning_rate": 2.3015027613444552e-07,
      "loss": 0.6951,
      "step": 50190
    },
    {
      "epoch": 0.7814458187705792,
      "grad_norm": 2.9631118774414062,
      "learning_rate": 2.3006833712984054e-07,
      "loss": 0.6281,
      "step": 50195
    },
    {
      "epoch": 0.7815236597725486,
      "grad_norm": 4.26964807510376,
      "learning_rate": 2.2998639812523556e-07,
      "loss": 0.8587,
      "step": 50200
    },
    {
      "epoch": 0.781601500774518,
      "grad_norm": 5.167074203491211,
      "learning_rate": 2.2990445912063058e-07,
      "loss": 0.7291,
      "step": 50205
    },
    {
      "epoch": 0.7816793417764873,
      "grad_norm": 3.6119847297668457,
      "learning_rate": 2.2982252011602563e-07,
      "loss": 0.7464,
      "step": 50210
    },
    {
      "epoch": 0.7817571827784567,
      "grad_norm": 3.3604848384857178,
      "learning_rate": 2.2974058111142063e-07,
      "loss": 0.8016,
      "step": 50215
    },
    {
      "epoch": 0.7818350237804261,
      "grad_norm": 3.6794660091400146,
      "learning_rate": 2.2965864210681568e-07,
      "loss": 0.6086,
      "step": 50220
    },
    {
      "epoch": 0.7819128647823955,
      "grad_norm": 2.759500741958618,
      "learning_rate": 2.295767031022107e-07,
      "loss": 0.7259,
      "step": 50225
    },
    {
      "epoch": 0.7819907057843648,
      "grad_norm": 4.1494598388671875,
      "learning_rate": 2.2949476409760575e-07,
      "loss": 0.7021,
      "step": 50230
    },
    {
      "epoch": 0.7820685467863342,
      "grad_norm": 5.115108489990234,
      "learning_rate": 2.2941282509300074e-07,
      "loss": 0.8003,
      "step": 50235
    },
    {
      "epoch": 0.7821463877883036,
      "grad_norm": 3.9977917671203613,
      "learning_rate": 2.293308860883958e-07,
      "loss": 0.7861,
      "step": 50240
    },
    {
      "epoch": 0.782224228790273,
      "grad_norm": 4.58198356628418,
      "learning_rate": 2.292489470837908e-07,
      "loss": 0.7084,
      "step": 50245
    },
    {
      "epoch": 0.7823020697922424,
      "grad_norm": 3.113095760345459,
      "learning_rate": 2.2916700807918586e-07,
      "loss": 0.6644,
      "step": 50250
    },
    {
      "epoch": 0.7823799107942118,
      "grad_norm": 3.0590479373931885,
      "learning_rate": 2.2908506907458086e-07,
      "loss": 0.7474,
      "step": 50255
    },
    {
      "epoch": 0.7824577517961812,
      "grad_norm": 3.6873786449432373,
      "learning_rate": 2.290031300699759e-07,
      "loss": 0.7794,
      "step": 50260
    },
    {
      "epoch": 0.7825355927981505,
      "grad_norm": 6.7857537269592285,
      "learning_rate": 2.2892119106537093e-07,
      "loss": 0.7213,
      "step": 50265
    },
    {
      "epoch": 0.7826134338001198,
      "grad_norm": 2.4996869564056396,
      "learning_rate": 2.2883925206076595e-07,
      "loss": 0.817,
      "step": 50270
    },
    {
      "epoch": 0.7826912748020892,
      "grad_norm": 2.2416183948516846,
      "learning_rate": 2.2875731305616097e-07,
      "loss": 0.6899,
      "step": 50275
    },
    {
      "epoch": 0.7827691158040586,
      "grad_norm": 5.684887409210205,
      "learning_rate": 2.2867537405155602e-07,
      "loss": 0.6335,
      "step": 50280
    },
    {
      "epoch": 0.782846956806028,
      "grad_norm": 5.1376633644104,
      "learning_rate": 2.2859343504695104e-07,
      "loss": 0.7635,
      "step": 50285
    },
    {
      "epoch": 0.7829247978079974,
      "grad_norm": 5.694852828979492,
      "learning_rate": 2.2851149604234606e-07,
      "loss": 0.689,
      "step": 50290
    },
    {
      "epoch": 0.7830026388099668,
      "grad_norm": 7.156779766082764,
      "learning_rate": 2.2842955703774108e-07,
      "loss": 0.6812,
      "step": 50295
    },
    {
      "epoch": 0.7830804798119362,
      "grad_norm": 1.7789318561553955,
      "learning_rate": 2.2834761803313613e-07,
      "loss": 0.6074,
      "step": 50300
    },
    {
      "epoch": 0.7831583208139055,
      "grad_norm": 3.300056219100952,
      "learning_rate": 2.2826567902853113e-07,
      "loss": 0.6078,
      "step": 50305
    },
    {
      "epoch": 0.7832361618158749,
      "grad_norm": 2.874091148376465,
      "learning_rate": 2.2818374002392618e-07,
      "loss": 0.6954,
      "step": 50310
    },
    {
      "epoch": 0.7833140028178442,
      "grad_norm": 3.44417667388916,
      "learning_rate": 2.281018010193212e-07,
      "loss": 0.6924,
      "step": 50315
    },
    {
      "epoch": 0.7833918438198136,
      "grad_norm": 4.237050533294678,
      "learning_rate": 2.2801986201471625e-07,
      "loss": 0.7681,
      "step": 50320
    },
    {
      "epoch": 0.783469684821783,
      "grad_norm": 4.542873382568359,
      "learning_rate": 2.2793792301011124e-07,
      "loss": 0.622,
      "step": 50325
    },
    {
      "epoch": 0.7835475258237524,
      "grad_norm": 4.57938814163208,
      "learning_rate": 2.278559840055063e-07,
      "loss": 0.7093,
      "step": 50330
    },
    {
      "epoch": 0.7836253668257218,
      "grad_norm": 3.4819154739379883,
      "learning_rate": 2.2777404500090134e-07,
      "loss": 0.771,
      "step": 50335
    },
    {
      "epoch": 0.7837032078276912,
      "grad_norm": 11.869524955749512,
      "learning_rate": 2.2769210599629633e-07,
      "loss": 0.7801,
      "step": 50340
    },
    {
      "epoch": 0.7837810488296605,
      "grad_norm": 3.398078203201294,
      "learning_rate": 2.2761016699169138e-07,
      "loss": 0.7481,
      "step": 50345
    },
    {
      "epoch": 0.7838588898316299,
      "grad_norm": 5.014096260070801,
      "learning_rate": 2.275282279870864e-07,
      "loss": 0.9331,
      "step": 50350
    },
    {
      "epoch": 0.7839367308335993,
      "grad_norm": 3.8018643856048584,
      "learning_rate": 2.2744628898248145e-07,
      "loss": 0.7099,
      "step": 50355
    },
    {
      "epoch": 0.7840145718355687,
      "grad_norm": 3.769270181655884,
      "learning_rate": 2.2736434997787645e-07,
      "loss": 0.7107,
      "step": 50360
    },
    {
      "epoch": 0.784092412837538,
      "grad_norm": 3.535346746444702,
      "learning_rate": 2.272824109732715e-07,
      "loss": 0.8838,
      "step": 50365
    },
    {
      "epoch": 0.7841702538395074,
      "grad_norm": 3.0127336978912354,
      "learning_rate": 2.2720047196866652e-07,
      "loss": 0.6267,
      "step": 50370
    },
    {
      "epoch": 0.7842480948414768,
      "grad_norm": 2.486269235610962,
      "learning_rate": 2.2711853296406157e-07,
      "loss": 0.7536,
      "step": 50375
    },
    {
      "epoch": 0.7843259358434462,
      "grad_norm": 2.7879626750946045,
      "learning_rate": 2.2703659395945656e-07,
      "loss": 0.7881,
      "step": 50380
    },
    {
      "epoch": 0.7844037768454155,
      "grad_norm": 2.6142940521240234,
      "learning_rate": 2.269546549548516e-07,
      "loss": 0.7409,
      "step": 50385
    },
    {
      "epoch": 0.7844816178473849,
      "grad_norm": 3.2745590209960938,
      "learning_rate": 2.2687271595024663e-07,
      "loss": 0.7731,
      "step": 50390
    },
    {
      "epoch": 0.7845594588493543,
      "grad_norm": 3.7306408882141113,
      "learning_rate": 2.2679077694564165e-07,
      "loss": 0.8055,
      "step": 50395
    },
    {
      "epoch": 0.7846372998513237,
      "grad_norm": 3.555879831314087,
      "learning_rate": 2.2670883794103668e-07,
      "loss": 0.735,
      "step": 50400
    },
    {
      "epoch": 0.7847151408532931,
      "grad_norm": 4.004828453063965,
      "learning_rate": 2.2662689893643173e-07,
      "loss": 0.6169,
      "step": 50405
    },
    {
      "epoch": 0.7847929818552625,
      "grad_norm": 3.3610544204711914,
      "learning_rate": 2.2654495993182675e-07,
      "loss": 0.6198,
      "step": 50410
    },
    {
      "epoch": 0.7848708228572319,
      "grad_norm": 3.2004592418670654,
      "learning_rate": 2.2646302092722177e-07,
      "loss": 0.6907,
      "step": 50415
    },
    {
      "epoch": 0.7849486638592011,
      "grad_norm": 3.380871534347534,
      "learning_rate": 2.263810819226168e-07,
      "loss": 0.6955,
      "step": 50420
    },
    {
      "epoch": 0.7850265048611705,
      "grad_norm": 3.0148487091064453,
      "learning_rate": 2.2629914291801184e-07,
      "loss": 0.6495,
      "step": 50425
    },
    {
      "epoch": 0.7851043458631399,
      "grad_norm": 2.5917248725891113,
      "learning_rate": 2.2621720391340683e-07,
      "loss": 0.6698,
      "step": 50430
    },
    {
      "epoch": 0.7851821868651093,
      "grad_norm": 3.530505418777466,
      "learning_rate": 2.2613526490880188e-07,
      "loss": 0.8331,
      "step": 50435
    },
    {
      "epoch": 0.7852600278670787,
      "grad_norm": 3.5497562885284424,
      "learning_rate": 2.260533259041969e-07,
      "loss": 0.7127,
      "step": 50440
    },
    {
      "epoch": 0.7853378688690481,
      "grad_norm": 3.8401565551757812,
      "learning_rate": 2.2597138689959195e-07,
      "loss": 0.7394,
      "step": 50445
    },
    {
      "epoch": 0.7854157098710175,
      "grad_norm": 3.196229934692383,
      "learning_rate": 2.2588944789498695e-07,
      "loss": 0.7913,
      "step": 50450
    },
    {
      "epoch": 0.7854935508729869,
      "grad_norm": 10.290125846862793,
      "learning_rate": 2.25807508890382e-07,
      "loss": 0.6521,
      "step": 50455
    },
    {
      "epoch": 0.7855713918749562,
      "grad_norm": 3.6701347827911377,
      "learning_rate": 2.2572556988577702e-07,
      "loss": 0.8141,
      "step": 50460
    },
    {
      "epoch": 0.7856492328769256,
      "grad_norm": 3.4584710597991943,
      "learning_rate": 2.2564363088117204e-07,
      "loss": 0.7624,
      "step": 50465
    },
    {
      "epoch": 0.785727073878895,
      "grad_norm": 5.415408611297607,
      "learning_rate": 2.2556169187656706e-07,
      "loss": 0.686,
      "step": 50470
    },
    {
      "epoch": 0.7858049148808643,
      "grad_norm": 4.285885810852051,
      "learning_rate": 2.254797528719621e-07,
      "loss": 0.6757,
      "step": 50475
    },
    {
      "epoch": 0.7858827558828337,
      "grad_norm": 4.286549091339111,
      "learning_rate": 2.2539781386735713e-07,
      "loss": 0.8468,
      "step": 50480
    },
    {
      "epoch": 0.7859605968848031,
      "grad_norm": 3.324176788330078,
      "learning_rate": 2.2531587486275216e-07,
      "loss": 0.7864,
      "step": 50485
    },
    {
      "epoch": 0.7860384378867725,
      "grad_norm": 6.200387954711914,
      "learning_rate": 2.2523393585814718e-07,
      "loss": 0.7385,
      "step": 50490
    },
    {
      "epoch": 0.7861162788887418,
      "grad_norm": 3.967029333114624,
      "learning_rate": 2.2515199685354223e-07,
      "loss": 0.8133,
      "step": 50495
    },
    {
      "epoch": 0.7861941198907112,
      "grad_norm": 6.041956424713135,
      "learning_rate": 2.2507005784893722e-07,
      "loss": 0.7172,
      "step": 50500
    },
    {
      "epoch": 0.7862719608926806,
      "grad_norm": 3.460447072982788,
      "learning_rate": 2.2498811884433227e-07,
      "loss": 0.7837,
      "step": 50505
    },
    {
      "epoch": 0.78634980189465,
      "grad_norm": 5.800002098083496,
      "learning_rate": 2.249061798397273e-07,
      "loss": 0.7167,
      "step": 50510
    },
    {
      "epoch": 0.7864276428966194,
      "grad_norm": 6.605666637420654,
      "learning_rate": 2.2482424083512234e-07,
      "loss": 0.7298,
      "step": 50515
    },
    {
      "epoch": 0.7865054838985888,
      "grad_norm": 9.547050476074219,
      "learning_rate": 2.2474230183051734e-07,
      "loss": 0.6935,
      "step": 50520
    },
    {
      "epoch": 0.7865833249005582,
      "grad_norm": 4.347421169281006,
      "learning_rate": 2.2466036282591238e-07,
      "loss": 0.7255,
      "step": 50525
    },
    {
      "epoch": 0.7866611659025275,
      "grad_norm": 6.003590106964111,
      "learning_rate": 2.245784238213074e-07,
      "loss": 0.8019,
      "step": 50530
    },
    {
      "epoch": 0.7867390069044968,
      "grad_norm": 4.230383396148682,
      "learning_rate": 2.2449648481670245e-07,
      "loss": 0.8184,
      "step": 50535
    },
    {
      "epoch": 0.7868168479064662,
      "grad_norm": 3.783878803253174,
      "learning_rate": 2.2441454581209745e-07,
      "loss": 0.7209,
      "step": 50540
    },
    {
      "epoch": 0.7868946889084356,
      "grad_norm": 8.670601844787598,
      "learning_rate": 2.243326068074925e-07,
      "loss": 0.7162,
      "step": 50545
    },
    {
      "epoch": 0.786972529910405,
      "grad_norm": 4.571006774902344,
      "learning_rate": 2.2425066780288752e-07,
      "loss": 0.6972,
      "step": 50550
    },
    {
      "epoch": 0.7870503709123744,
      "grad_norm": 4.5071187019348145,
      "learning_rate": 2.2416872879828254e-07,
      "loss": 0.6969,
      "step": 50555
    },
    {
      "epoch": 0.7871282119143438,
      "grad_norm": 2.9758307933807373,
      "learning_rate": 2.2408678979367756e-07,
      "loss": 0.7725,
      "step": 50560
    },
    {
      "epoch": 0.7872060529163132,
      "grad_norm": 6.023144721984863,
      "learning_rate": 2.240048507890726e-07,
      "loss": 0.677,
      "step": 50565
    },
    {
      "epoch": 0.7872838939182825,
      "grad_norm": 3.0693228244781494,
      "learning_rate": 2.2392291178446763e-07,
      "loss": 0.7228,
      "step": 50570
    },
    {
      "epoch": 0.7873617349202519,
      "grad_norm": 3.691896915435791,
      "learning_rate": 2.2384097277986266e-07,
      "loss": 0.7701,
      "step": 50575
    },
    {
      "epoch": 0.7874395759222212,
      "grad_norm": 2.9061577320098877,
      "learning_rate": 2.2375903377525768e-07,
      "loss": 0.7507,
      "step": 50580
    },
    {
      "epoch": 0.7875174169241906,
      "grad_norm": 4.131178855895996,
      "learning_rate": 2.2367709477065273e-07,
      "loss": 0.618,
      "step": 50585
    },
    {
      "epoch": 0.78759525792616,
      "grad_norm": 5.043824672698975,
      "learning_rate": 2.2359515576604772e-07,
      "loss": 0.7264,
      "step": 50590
    },
    {
      "epoch": 0.7876730989281294,
      "grad_norm": 3.040057420730591,
      "learning_rate": 2.2351321676144277e-07,
      "loss": 0.8611,
      "step": 50595
    },
    {
      "epoch": 0.7877509399300988,
      "grad_norm": 6.305315017700195,
      "learning_rate": 2.234312777568378e-07,
      "loss": 0.8273,
      "step": 50600
    },
    {
      "epoch": 0.7878287809320682,
      "grad_norm": 2.821653127670288,
      "learning_rate": 2.2334933875223284e-07,
      "loss": 0.688,
      "step": 50605
    },
    {
      "epoch": 0.7879066219340375,
      "grad_norm": 7.883811950683594,
      "learning_rate": 2.2326739974762784e-07,
      "loss": 0.8126,
      "step": 50610
    },
    {
      "epoch": 0.7879844629360069,
      "grad_norm": 2.997893810272217,
      "learning_rate": 2.2318546074302288e-07,
      "loss": 0.8065,
      "step": 50615
    },
    {
      "epoch": 0.7880623039379763,
      "grad_norm": 3.500997304916382,
      "learning_rate": 2.231035217384179e-07,
      "loss": 0.6988,
      "step": 50620
    },
    {
      "epoch": 0.7881401449399457,
      "grad_norm": 4.335129261016846,
      "learning_rate": 2.2302158273381293e-07,
      "loss": 0.7705,
      "step": 50625
    },
    {
      "epoch": 0.7882179859419151,
      "grad_norm": 2.5020034313201904,
      "learning_rate": 2.2293964372920795e-07,
      "loss": 0.7052,
      "step": 50630
    },
    {
      "epoch": 0.7882958269438844,
      "grad_norm": 3.5721755027770996,
      "learning_rate": 2.22857704724603e-07,
      "loss": 0.7901,
      "step": 50635
    },
    {
      "epoch": 0.7883736679458538,
      "grad_norm": 4.055649757385254,
      "learning_rate": 2.2277576571999802e-07,
      "loss": 0.7412,
      "step": 50640
    },
    {
      "epoch": 0.7884515089478231,
      "grad_norm": 3.8442587852478027,
      "learning_rate": 2.2269382671539304e-07,
      "loss": 0.7324,
      "step": 50645
    },
    {
      "epoch": 0.7885293499497925,
      "grad_norm": 7.811295986175537,
      "learning_rate": 2.2261188771078806e-07,
      "loss": 0.7357,
      "step": 50650
    },
    {
      "epoch": 0.7886071909517619,
      "grad_norm": 2.4942703247070312,
      "learning_rate": 2.225299487061831e-07,
      "loss": 0.6278,
      "step": 50655
    },
    {
      "epoch": 0.7886850319537313,
      "grad_norm": 3.407963275909424,
      "learning_rate": 2.2244800970157816e-07,
      "loss": 0.6416,
      "step": 50660
    },
    {
      "epoch": 0.7887628729557007,
      "grad_norm": 6.131340980529785,
      "learning_rate": 2.2236607069697316e-07,
      "loss": 0.7604,
      "step": 50665
    },
    {
      "epoch": 0.7888407139576701,
      "grad_norm": 4.378029823303223,
      "learning_rate": 2.222841316923682e-07,
      "loss": 0.7073,
      "step": 50670
    },
    {
      "epoch": 0.7889185549596395,
      "grad_norm": 4.747143745422363,
      "learning_rate": 2.2220219268776323e-07,
      "loss": 0.81,
      "step": 50675
    },
    {
      "epoch": 0.7889963959616089,
      "grad_norm": 3.3924198150634766,
      "learning_rate": 2.2212025368315825e-07,
      "loss": 0.7349,
      "step": 50680
    },
    {
      "epoch": 0.7890742369635781,
      "grad_norm": 3.0885112285614014,
      "learning_rate": 2.2203831467855327e-07,
      "loss": 0.7088,
      "step": 50685
    },
    {
      "epoch": 0.7891520779655475,
      "grad_norm": 2.941011905670166,
      "learning_rate": 2.2195637567394832e-07,
      "loss": 0.7331,
      "step": 50690
    },
    {
      "epoch": 0.7892299189675169,
      "grad_norm": 4.425881862640381,
      "learning_rate": 2.2187443666934334e-07,
      "loss": 0.7897,
      "step": 50695
    },
    {
      "epoch": 0.7893077599694863,
      "grad_norm": 2.722203493118286,
      "learning_rate": 2.2179249766473836e-07,
      "loss": 0.7037,
      "step": 50700
    },
    {
      "epoch": 0.7893856009714557,
      "grad_norm": 8.49703311920166,
      "learning_rate": 2.2171055866013338e-07,
      "loss": 0.7078,
      "step": 50705
    },
    {
      "epoch": 0.7894634419734251,
      "grad_norm": 3.1682991981506348,
      "learning_rate": 2.2162861965552843e-07,
      "loss": 0.7937,
      "step": 50710
    },
    {
      "epoch": 0.7895412829753945,
      "grad_norm": 5.107608795166016,
      "learning_rate": 2.2154668065092343e-07,
      "loss": 0.7069,
      "step": 50715
    },
    {
      "epoch": 0.7896191239773639,
      "grad_norm": 6.479969024658203,
      "learning_rate": 2.2146474164631848e-07,
      "loss": 0.8183,
      "step": 50720
    },
    {
      "epoch": 0.7896969649793332,
      "grad_norm": 2.5819475650787354,
      "learning_rate": 2.213828026417135e-07,
      "loss": 0.8285,
      "step": 50725
    },
    {
      "epoch": 0.7897748059813026,
      "grad_norm": 3.5713770389556885,
      "learning_rate": 2.2130086363710855e-07,
      "loss": 0.7451,
      "step": 50730
    },
    {
      "epoch": 0.789852646983272,
      "grad_norm": 5.477764129638672,
      "learning_rate": 2.2121892463250354e-07,
      "loss": 0.7214,
      "step": 50735
    },
    {
      "epoch": 0.7899304879852413,
      "grad_norm": 3.7531752586364746,
      "learning_rate": 2.211369856278986e-07,
      "loss": 0.7161,
      "step": 50740
    },
    {
      "epoch": 0.7900083289872107,
      "grad_norm": 8.371977806091309,
      "learning_rate": 2.210550466232936e-07,
      "loss": 0.7258,
      "step": 50745
    },
    {
      "epoch": 0.7900861699891801,
      "grad_norm": 3.199742078781128,
      "learning_rate": 2.2097310761868866e-07,
      "loss": 0.6493,
      "step": 50750
    },
    {
      "epoch": 0.7901640109911495,
      "grad_norm": 6.208636283874512,
      "learning_rate": 2.2089116861408366e-07,
      "loss": 0.767,
      "step": 50755
    },
    {
      "epoch": 0.7902418519931188,
      "grad_norm": 2.5745840072631836,
      "learning_rate": 2.208092296094787e-07,
      "loss": 0.6041,
      "step": 50760
    },
    {
      "epoch": 0.7903196929950882,
      "grad_norm": 3.9380972385406494,
      "learning_rate": 2.2072729060487373e-07,
      "loss": 0.6737,
      "step": 50765
    },
    {
      "epoch": 0.7903975339970576,
      "grad_norm": 4.914278984069824,
      "learning_rate": 2.2064535160026875e-07,
      "loss": 0.7356,
      "step": 50770
    },
    {
      "epoch": 0.790475374999027,
      "grad_norm": 3.993757963180542,
      "learning_rate": 2.2056341259566377e-07,
      "loss": 0.7518,
      "step": 50775
    },
    {
      "epoch": 0.7905532160009964,
      "grad_norm": 3.429353713989258,
      "learning_rate": 2.2048147359105882e-07,
      "loss": 0.6542,
      "step": 50780
    },
    {
      "epoch": 0.7906310570029658,
      "grad_norm": 4.218374729156494,
      "learning_rate": 2.2039953458645384e-07,
      "loss": 0.792,
      "step": 50785
    },
    {
      "epoch": 0.7907088980049352,
      "grad_norm": 3.3340017795562744,
      "learning_rate": 2.2031759558184886e-07,
      "loss": 0.683,
      "step": 50790
    },
    {
      "epoch": 0.7907867390069045,
      "grad_norm": 9.516267776489258,
      "learning_rate": 2.2023565657724388e-07,
      "loss": 0.718,
      "step": 50795
    },
    {
      "epoch": 0.7908645800088738,
      "grad_norm": 3.463284492492676,
      "learning_rate": 2.2015371757263893e-07,
      "loss": 0.7847,
      "step": 50800
    },
    {
      "epoch": 0.7909424210108432,
      "grad_norm": 4.327865123748779,
      "learning_rate": 2.2007177856803393e-07,
      "loss": 0.6948,
      "step": 50805
    },
    {
      "epoch": 0.7910202620128126,
      "grad_norm": 2.4825024604797363,
      "learning_rate": 2.1998983956342898e-07,
      "loss": 0.6798,
      "step": 50810
    },
    {
      "epoch": 0.791098103014782,
      "grad_norm": 14.349628448486328,
      "learning_rate": 2.19907900558824e-07,
      "loss": 0.6736,
      "step": 50815
    },
    {
      "epoch": 0.7911759440167514,
      "grad_norm": 4.309773921966553,
      "learning_rate": 2.1982596155421905e-07,
      "loss": 0.6954,
      "step": 50820
    },
    {
      "epoch": 0.7912537850187208,
      "grad_norm": 9.160594940185547,
      "learning_rate": 2.1974402254961404e-07,
      "loss": 0.7315,
      "step": 50825
    },
    {
      "epoch": 0.7913316260206902,
      "grad_norm": 7.37003231048584,
      "learning_rate": 2.196620835450091e-07,
      "loss": 0.816,
      "step": 50830
    },
    {
      "epoch": 0.7914094670226595,
      "grad_norm": 4.806380271911621,
      "learning_rate": 2.195801445404041e-07,
      "loss": 0.7403,
      "step": 50835
    },
    {
      "epoch": 0.7914873080246289,
      "grad_norm": 2.5450868606567383,
      "learning_rate": 2.1949820553579913e-07,
      "loss": 0.727,
      "step": 50840
    },
    {
      "epoch": 0.7915651490265982,
      "grad_norm": 3.067566156387329,
      "learning_rate": 2.1941626653119416e-07,
      "loss": 0.6604,
      "step": 50845
    },
    {
      "epoch": 0.7916429900285676,
      "grad_norm": 3.718106985092163,
      "learning_rate": 2.193343275265892e-07,
      "loss": 0.7196,
      "step": 50850
    },
    {
      "epoch": 0.791720831030537,
      "grad_norm": 2.4944980144500732,
      "learning_rate": 2.1925238852198423e-07,
      "loss": 0.678,
      "step": 50855
    },
    {
      "epoch": 0.7917986720325064,
      "grad_norm": 3.369215250015259,
      "learning_rate": 2.1917044951737925e-07,
      "loss": 0.7286,
      "step": 50860
    },
    {
      "epoch": 0.7918765130344758,
      "grad_norm": 3.1837239265441895,
      "learning_rate": 2.1908851051277427e-07,
      "loss": 0.6133,
      "step": 50865
    },
    {
      "epoch": 0.7919543540364452,
      "grad_norm": 7.350306034088135,
      "learning_rate": 2.1900657150816932e-07,
      "loss": 0.778,
      "step": 50870
    },
    {
      "epoch": 0.7920321950384145,
      "grad_norm": 5.66510009765625,
      "learning_rate": 2.1892463250356431e-07,
      "loss": 0.8402,
      "step": 50875
    },
    {
      "epoch": 0.7921100360403839,
      "grad_norm": 6.4912824630737305,
      "learning_rate": 2.1884269349895936e-07,
      "loss": 0.6217,
      "step": 50880
    },
    {
      "epoch": 0.7921878770423533,
      "grad_norm": 5.339352607727051,
      "learning_rate": 2.1876075449435439e-07,
      "loss": 0.8406,
      "step": 50885
    },
    {
      "epoch": 0.7922657180443227,
      "grad_norm": 3.505523920059204,
      "learning_rate": 2.1867881548974943e-07,
      "loss": 0.8424,
      "step": 50890
    },
    {
      "epoch": 0.7923435590462921,
      "grad_norm": 3.032679796218872,
      "learning_rate": 2.1859687648514443e-07,
      "loss": 0.7731,
      "step": 50895
    },
    {
      "epoch": 0.7924214000482614,
      "grad_norm": 2.604158878326416,
      "learning_rate": 2.1851493748053948e-07,
      "loss": 0.7013,
      "step": 50900
    },
    {
      "epoch": 0.7924992410502308,
      "grad_norm": 2.955343723297119,
      "learning_rate": 2.184329984759345e-07,
      "loss": 0.7809,
      "step": 50905
    },
    {
      "epoch": 0.7925770820522001,
      "grad_norm": 3.863309860229492,
      "learning_rate": 2.1835105947132955e-07,
      "loss": 0.858,
      "step": 50910
    },
    {
      "epoch": 0.7926549230541695,
      "grad_norm": 4.862393856048584,
      "learning_rate": 2.1826912046672454e-07,
      "loss": 0.8179,
      "step": 50915
    },
    {
      "epoch": 0.7927327640561389,
      "grad_norm": 4.70264196395874,
      "learning_rate": 2.181871814621196e-07,
      "loss": 0.8109,
      "step": 50920
    },
    {
      "epoch": 0.7928106050581083,
      "grad_norm": 2.9560837745666504,
      "learning_rate": 2.1810524245751461e-07,
      "loss": 0.6951,
      "step": 50925
    },
    {
      "epoch": 0.7928884460600777,
      "grad_norm": 4.2055983543396,
      "learning_rate": 2.1802330345290964e-07,
      "loss": 0.655,
      "step": 50930
    },
    {
      "epoch": 0.7929662870620471,
      "grad_norm": 5.784762382507324,
      "learning_rate": 2.1794136444830466e-07,
      "loss": 0.7462,
      "step": 50935
    },
    {
      "epoch": 0.7930441280640165,
      "grad_norm": 3.739675998687744,
      "learning_rate": 2.178594254436997e-07,
      "loss": 0.7119,
      "step": 50940
    },
    {
      "epoch": 0.7931219690659859,
      "grad_norm": 5.453944206237793,
      "learning_rate": 2.1777748643909473e-07,
      "loss": 0.6391,
      "step": 50945
    },
    {
      "epoch": 0.7931998100679551,
      "grad_norm": 7.297431945800781,
      "learning_rate": 2.1769554743448975e-07,
      "loss": 0.7434,
      "step": 50950
    },
    {
      "epoch": 0.7932776510699245,
      "grad_norm": 5.985004901885986,
      "learning_rate": 2.1761360842988477e-07,
      "loss": 0.6696,
      "step": 50955
    },
    {
      "epoch": 0.7933554920718939,
      "grad_norm": 12.377729415893555,
      "learning_rate": 2.1753166942527982e-07,
      "loss": 0.8005,
      "step": 50960
    },
    {
      "epoch": 0.7934333330738633,
      "grad_norm": 5.531628131866455,
      "learning_rate": 2.1744973042067482e-07,
      "loss": 0.6518,
      "step": 50965
    },
    {
      "epoch": 0.7935111740758327,
      "grad_norm": 5.706180572509766,
      "learning_rate": 2.1736779141606986e-07,
      "loss": 0.6931,
      "step": 50970
    },
    {
      "epoch": 0.7935890150778021,
      "grad_norm": 3.0661888122558594,
      "learning_rate": 2.172858524114649e-07,
      "loss": 0.7909,
      "step": 50975
    },
    {
      "epoch": 0.7936668560797715,
      "grad_norm": 3.441998243331909,
      "learning_rate": 2.1720391340685993e-07,
      "loss": 0.7613,
      "step": 50980
    },
    {
      "epoch": 0.7937446970817409,
      "grad_norm": 3.5612378120422363,
      "learning_rate": 2.1712197440225496e-07,
      "loss": 0.6876,
      "step": 50985
    },
    {
      "epoch": 0.7938225380837102,
      "grad_norm": 9.735668182373047,
      "learning_rate": 2.1704003539764998e-07,
      "loss": 0.7944,
      "step": 50990
    },
    {
      "epoch": 0.7939003790856796,
      "grad_norm": 3.5423412322998047,
      "learning_rate": 2.1695809639304503e-07,
      "loss": 0.7179,
      "step": 50995
    },
    {
      "epoch": 0.793978220087649,
      "grad_norm": 3.4396069049835205,
      "learning_rate": 2.1687615738844002e-07,
      "loss": 0.7247,
      "step": 51000
    },
    {
      "epoch": 0.7940560610896183,
      "grad_norm": 3.9558088779449463,
      "learning_rate": 2.1679421838383507e-07,
      "loss": 0.7441,
      "step": 51005
    },
    {
      "epoch": 0.7941339020915877,
      "grad_norm": 2.683441400527954,
      "learning_rate": 2.167122793792301e-07,
      "loss": 0.6756,
      "step": 51010
    },
    {
      "epoch": 0.7942117430935571,
      "grad_norm": 3.4686713218688965,
      "learning_rate": 2.1663034037462514e-07,
      "loss": 0.7552,
      "step": 51015
    },
    {
      "epoch": 0.7942895840955265,
      "grad_norm": 8.928704261779785,
      "learning_rate": 2.1654840137002014e-07,
      "loss": 0.7248,
      "step": 51020
    },
    {
      "epoch": 0.7943674250974958,
      "grad_norm": 4.6985626220703125,
      "learning_rate": 2.1646646236541518e-07,
      "loss": 0.7704,
      "step": 51025
    },
    {
      "epoch": 0.7944452660994652,
      "grad_norm": 3.8064804077148438,
      "learning_rate": 2.163845233608102e-07,
      "loss": 0.6851,
      "step": 51030
    },
    {
      "epoch": 0.7945231071014346,
      "grad_norm": 3.9881961345672607,
      "learning_rate": 2.1630258435620525e-07,
      "loss": 0.7121,
      "step": 51035
    },
    {
      "epoch": 0.794600948103404,
      "grad_norm": 4.960756301879883,
      "learning_rate": 2.1622064535160025e-07,
      "loss": 0.7631,
      "step": 51040
    },
    {
      "epoch": 0.7946787891053734,
      "grad_norm": 5.728940010070801,
      "learning_rate": 2.161387063469953e-07,
      "loss": 0.7288,
      "step": 51045
    },
    {
      "epoch": 0.7947566301073428,
      "grad_norm": 3.5991110801696777,
      "learning_rate": 2.1605676734239032e-07,
      "loss": 0.7767,
      "step": 51050
    },
    {
      "epoch": 0.7948344711093122,
      "grad_norm": 3.3040809631347656,
      "learning_rate": 2.1597482833778534e-07,
      "loss": 0.769,
      "step": 51055
    },
    {
      "epoch": 0.7949123121112815,
      "grad_norm": 4.494775295257568,
      "learning_rate": 2.1589288933318036e-07,
      "loss": 0.7543,
      "step": 51060
    },
    {
      "epoch": 0.7949901531132508,
      "grad_norm": 3.3512914180755615,
      "learning_rate": 2.158109503285754e-07,
      "loss": 0.7121,
      "step": 51065
    },
    {
      "epoch": 0.7950679941152202,
      "grad_norm": 7.9084906578063965,
      "learning_rate": 2.1572901132397043e-07,
      "loss": 0.7546,
      "step": 51070
    },
    {
      "epoch": 0.7951458351171896,
      "grad_norm": 5.355246543884277,
      "learning_rate": 2.1564707231936546e-07,
      "loss": 0.8075,
      "step": 51075
    },
    {
      "epoch": 0.795223676119159,
      "grad_norm": 5.282921314239502,
      "learning_rate": 2.1556513331476048e-07,
      "loss": 0.7708,
      "step": 51080
    },
    {
      "epoch": 0.7953015171211284,
      "grad_norm": 6.666356563568115,
      "learning_rate": 2.1548319431015553e-07,
      "loss": 0.6968,
      "step": 51085
    },
    {
      "epoch": 0.7953793581230978,
      "grad_norm": 7.327302932739258,
      "learning_rate": 2.1540125530555052e-07,
      "loss": 0.7205,
      "step": 51090
    },
    {
      "epoch": 0.7954571991250672,
      "grad_norm": 4.904540061950684,
      "learning_rate": 2.1531931630094557e-07,
      "loss": 0.7455,
      "step": 51095
    },
    {
      "epoch": 0.7955350401270365,
      "grad_norm": 4.4911885261535645,
      "learning_rate": 2.152373772963406e-07,
      "loss": 0.7871,
      "step": 51100
    },
    {
      "epoch": 0.7956128811290059,
      "grad_norm": 3.5192856788635254,
      "learning_rate": 2.1515543829173564e-07,
      "loss": 0.7614,
      "step": 51105
    },
    {
      "epoch": 0.7956907221309752,
      "grad_norm": 3.309195041656494,
      "learning_rate": 2.1507349928713064e-07,
      "loss": 0.7627,
      "step": 51110
    },
    {
      "epoch": 0.7957685631329446,
      "grad_norm": 3.4927847385406494,
      "learning_rate": 2.1499156028252568e-07,
      "loss": 0.7373,
      "step": 51115
    },
    {
      "epoch": 0.795846404134914,
      "grad_norm": 4.3902907371521,
      "learning_rate": 2.149096212779207e-07,
      "loss": 0.6676,
      "step": 51120
    },
    {
      "epoch": 0.7959242451368834,
      "grad_norm": 5.345546722412109,
      "learning_rate": 2.1482768227331573e-07,
      "loss": 0.717,
      "step": 51125
    },
    {
      "epoch": 0.7960020861388528,
      "grad_norm": 4.026414394378662,
      "learning_rate": 2.1474574326871075e-07,
      "loss": 0.7053,
      "step": 51130
    },
    {
      "epoch": 0.7960799271408222,
      "grad_norm": 2.89192533493042,
      "learning_rate": 2.146638042641058e-07,
      "loss": 0.7066,
      "step": 51135
    },
    {
      "epoch": 0.7961577681427915,
      "grad_norm": 3.816721200942993,
      "learning_rate": 2.1458186525950082e-07,
      "loss": 0.8318,
      "step": 51140
    },
    {
      "epoch": 0.7962356091447609,
      "grad_norm": 2.881959915161133,
      "learning_rate": 2.1449992625489584e-07,
      "loss": 0.6767,
      "step": 51145
    },
    {
      "epoch": 0.7963134501467303,
      "grad_norm": 8.211742401123047,
      "learning_rate": 2.1441798725029086e-07,
      "loss": 0.7031,
      "step": 51150
    },
    {
      "epoch": 0.7963912911486997,
      "grad_norm": 3.4175140857696533,
      "learning_rate": 2.143360482456859e-07,
      "loss": 0.7609,
      "step": 51155
    },
    {
      "epoch": 0.7964691321506691,
      "grad_norm": 4.3593268394470215,
      "learning_rate": 2.142541092410809e-07,
      "loss": 0.7888,
      "step": 51160
    },
    {
      "epoch": 0.7965469731526384,
      "grad_norm": 3.8612399101257324,
      "learning_rate": 2.1417217023647596e-07,
      "loss": 0.7268,
      "step": 51165
    },
    {
      "epoch": 0.7966248141546078,
      "grad_norm": 3.903899669647217,
      "learning_rate": 2.1409023123187098e-07,
      "loss": 0.8025,
      "step": 51170
    },
    {
      "epoch": 0.7967026551565771,
      "grad_norm": 4.673237323760986,
      "learning_rate": 2.1400829222726603e-07,
      "loss": 0.6484,
      "step": 51175
    },
    {
      "epoch": 0.7967804961585465,
      "grad_norm": 2.8236684799194336,
      "learning_rate": 2.1392635322266102e-07,
      "loss": 0.7672,
      "step": 51180
    },
    {
      "epoch": 0.7968583371605159,
      "grad_norm": 4.149912357330322,
      "learning_rate": 2.1384441421805607e-07,
      "loss": 0.6628,
      "step": 51185
    },
    {
      "epoch": 0.7969361781624853,
      "grad_norm": 4.502058029174805,
      "learning_rate": 2.137624752134511e-07,
      "loss": 0.7109,
      "step": 51190
    },
    {
      "epoch": 0.7970140191644547,
      "grad_norm": 3.687145709991455,
      "learning_rate": 2.1368053620884614e-07,
      "loss": 0.8219,
      "step": 51195
    },
    {
      "epoch": 0.7970918601664241,
      "grad_norm": 3.250568151473999,
      "learning_rate": 2.1359859720424114e-07,
      "loss": 0.6644,
      "step": 51200
    },
    {
      "epoch": 0.7971697011683935,
      "grad_norm": 3.98972487449646,
      "learning_rate": 2.1351665819963619e-07,
      "loss": 0.6568,
      "step": 51205
    },
    {
      "epoch": 0.7972475421703629,
      "grad_norm": 5.023486137390137,
      "learning_rate": 2.134347191950312e-07,
      "loss": 0.67,
      "step": 51210
    },
    {
      "epoch": 0.7973253831723321,
      "grad_norm": 3.4484915733337402,
      "learning_rate": 2.1335278019042623e-07,
      "loss": 0.6793,
      "step": 51215
    },
    {
      "epoch": 0.7974032241743015,
      "grad_norm": 8.990313529968262,
      "learning_rate": 2.1327084118582125e-07,
      "loss": 0.7992,
      "step": 51220
    },
    {
      "epoch": 0.7974810651762709,
      "grad_norm": 3.9510045051574707,
      "learning_rate": 2.131889021812163e-07,
      "loss": 0.6109,
      "step": 51225
    },
    {
      "epoch": 0.7975589061782403,
      "grad_norm": 2.7473058700561523,
      "learning_rate": 2.1310696317661132e-07,
      "loss": 0.7617,
      "step": 51230
    },
    {
      "epoch": 0.7976367471802097,
      "grad_norm": 4.019311428070068,
      "learning_rate": 2.1302502417200634e-07,
      "loss": 0.7298,
      "step": 51235
    },
    {
      "epoch": 0.7977145881821791,
      "grad_norm": 3.6507010459899902,
      "learning_rate": 2.1294308516740137e-07,
      "loss": 0.751,
      "step": 51240
    },
    {
      "epoch": 0.7977924291841485,
      "grad_norm": 3.6586925983428955,
      "learning_rate": 2.1286114616279641e-07,
      "loss": 0.77,
      "step": 51245
    },
    {
      "epoch": 0.7978702701861178,
      "grad_norm": 3.2544147968292236,
      "learning_rate": 2.127792071581914e-07,
      "loss": 0.7179,
      "step": 51250
    },
    {
      "epoch": 0.7979481111880872,
      "grad_norm": 3.9433209896087646,
      "learning_rate": 2.1269726815358646e-07,
      "loss": 0.7218,
      "step": 51255
    },
    {
      "epoch": 0.7980259521900566,
      "grad_norm": 3.494509696960449,
      "learning_rate": 2.1261532914898148e-07,
      "loss": 0.8186,
      "step": 51260
    },
    {
      "epoch": 0.798103793192026,
      "grad_norm": 3.208751678466797,
      "learning_rate": 2.1253339014437653e-07,
      "loss": 0.6414,
      "step": 51265
    },
    {
      "epoch": 0.7981816341939953,
      "grad_norm": 3.8166394233703613,
      "learning_rate": 2.1245145113977152e-07,
      "loss": 0.7142,
      "step": 51270
    },
    {
      "epoch": 0.7982594751959647,
      "grad_norm": 7.277013301849365,
      "learning_rate": 2.1236951213516657e-07,
      "loss": 0.8647,
      "step": 51275
    },
    {
      "epoch": 0.7983373161979341,
      "grad_norm": 3.3959672451019287,
      "learning_rate": 2.122875731305616e-07,
      "loss": 0.6458,
      "step": 51280
    },
    {
      "epoch": 0.7984151571999035,
      "grad_norm": 4.594758033752441,
      "learning_rate": 2.1220563412595664e-07,
      "loss": 0.68,
      "step": 51285
    },
    {
      "epoch": 0.7984929982018728,
      "grad_norm": 4.924461841583252,
      "learning_rate": 2.1212369512135164e-07,
      "loss": 0.9149,
      "step": 51290
    },
    {
      "epoch": 0.7985708392038422,
      "grad_norm": 3.235504627227783,
      "learning_rate": 2.1204175611674669e-07,
      "loss": 0.6989,
      "step": 51295
    },
    {
      "epoch": 0.7986486802058116,
      "grad_norm": 4.046937942504883,
      "learning_rate": 2.1195981711214173e-07,
      "loss": 0.6602,
      "step": 51300
    },
    {
      "epoch": 0.798726521207781,
      "grad_norm": 3.5406785011291504,
      "learning_rate": 2.1187787810753673e-07,
      "loss": 0.78,
      "step": 51305
    },
    {
      "epoch": 0.7988043622097504,
      "grad_norm": 3.261706590652466,
      "learning_rate": 2.1179593910293178e-07,
      "loss": 0.7363,
      "step": 51310
    },
    {
      "epoch": 0.7988822032117198,
      "grad_norm": 6.311913967132568,
      "learning_rate": 2.117140000983268e-07,
      "loss": 0.682,
      "step": 51315
    },
    {
      "epoch": 0.7989600442136892,
      "grad_norm": 5.185361385345459,
      "learning_rate": 2.1163206109372185e-07,
      "loss": 0.7271,
      "step": 51320
    },
    {
      "epoch": 0.7990378852156585,
      "grad_norm": 3.027522563934326,
      "learning_rate": 2.1155012208911684e-07,
      "loss": 0.7986,
      "step": 51325
    },
    {
      "epoch": 0.7991157262176278,
      "grad_norm": 4.438055992126465,
      "learning_rate": 2.114681830845119e-07,
      "loss": 0.797,
      "step": 51330
    },
    {
      "epoch": 0.7991935672195972,
      "grad_norm": 3.722618818283081,
      "learning_rate": 2.1138624407990691e-07,
      "loss": 0.7118,
      "step": 51335
    },
    {
      "epoch": 0.7992714082215666,
      "grad_norm": 3.5783019065856934,
      "learning_rate": 2.1130430507530194e-07,
      "loss": 0.6472,
      "step": 51340
    },
    {
      "epoch": 0.799349249223536,
      "grad_norm": 3.436788320541382,
      "learning_rate": 2.1122236607069696e-07,
      "loss": 0.7362,
      "step": 51345
    },
    {
      "epoch": 0.7994270902255054,
      "grad_norm": 3.8613643646240234,
      "learning_rate": 2.11140427066092e-07,
      "loss": 0.7997,
      "step": 51350
    },
    {
      "epoch": 0.7995049312274748,
      "grad_norm": 4.697255611419678,
      "learning_rate": 2.1105848806148703e-07,
      "loss": 0.7623,
      "step": 51355
    },
    {
      "epoch": 0.7995827722294442,
      "grad_norm": 4.06143856048584,
      "learning_rate": 2.1097654905688205e-07,
      "loss": 0.7099,
      "step": 51360
    },
    {
      "epoch": 0.7996606132314135,
      "grad_norm": 10.546807289123535,
      "learning_rate": 2.1089461005227707e-07,
      "loss": 0.7277,
      "step": 51365
    },
    {
      "epoch": 0.7997384542333829,
      "grad_norm": 5.692515850067139,
      "learning_rate": 2.1081267104767212e-07,
      "loss": 0.7444,
      "step": 51370
    },
    {
      "epoch": 0.7998162952353522,
      "grad_norm": 3.2277963161468506,
      "learning_rate": 2.1073073204306712e-07,
      "loss": 0.7137,
      "step": 51375
    },
    {
      "epoch": 0.7998941362373216,
      "grad_norm": 3.129011631011963,
      "learning_rate": 2.1064879303846216e-07,
      "loss": 0.6691,
      "step": 51380
    },
    {
      "epoch": 0.799971977239291,
      "grad_norm": 2.842421770095825,
      "learning_rate": 2.1056685403385719e-07,
      "loss": 0.7296,
      "step": 51385
    },
    {
      "epoch": 0.8000498182412604,
      "grad_norm": 9.446612358093262,
      "learning_rate": 2.1048491502925223e-07,
      "loss": 0.6638,
      "step": 51390
    },
    {
      "epoch": 0.8001276592432298,
      "grad_norm": 8.698366165161133,
      "learning_rate": 2.1040297602464723e-07,
      "loss": 0.6694,
      "step": 51395
    },
    {
      "epoch": 0.8002055002451992,
      "grad_norm": 3.92960524559021,
      "learning_rate": 2.1032103702004228e-07,
      "loss": 0.7231,
      "step": 51400
    },
    {
      "epoch": 0.8002833412471685,
      "grad_norm": 3.146808624267578,
      "learning_rate": 2.102390980154373e-07,
      "loss": 0.7285,
      "step": 51405
    },
    {
      "epoch": 0.8003611822491379,
      "grad_norm": 3.277787685394287,
      "learning_rate": 2.1015715901083235e-07,
      "loss": 0.7698,
      "step": 51410
    },
    {
      "epoch": 0.8004390232511073,
      "grad_norm": 5.5949273109436035,
      "learning_rate": 2.1007522000622734e-07,
      "loss": 0.6738,
      "step": 51415
    },
    {
      "epoch": 0.8005168642530767,
      "grad_norm": 4.0006937980651855,
      "learning_rate": 2.099932810016224e-07,
      "loss": 0.7986,
      "step": 51420
    },
    {
      "epoch": 0.8005947052550461,
      "grad_norm": 3.2565877437591553,
      "learning_rate": 2.0991134199701741e-07,
      "loss": 0.7111,
      "step": 51425
    },
    {
      "epoch": 0.8006725462570154,
      "grad_norm": 4.11714506149292,
      "learning_rate": 2.0982940299241244e-07,
      "loss": 0.8312,
      "step": 51430
    },
    {
      "epoch": 0.8007503872589848,
      "grad_norm": 3.549177646636963,
      "learning_rate": 2.0974746398780746e-07,
      "loss": 0.7909,
      "step": 51435
    },
    {
      "epoch": 0.8008282282609541,
      "grad_norm": 3.516306161880493,
      "learning_rate": 2.096655249832025e-07,
      "loss": 0.7454,
      "step": 51440
    },
    {
      "epoch": 0.8009060692629235,
      "grad_norm": 4.265626907348633,
      "learning_rate": 2.0958358597859753e-07,
      "loss": 0.8038,
      "step": 51445
    },
    {
      "epoch": 0.8009839102648929,
      "grad_norm": 9.082649230957031,
      "learning_rate": 2.0950164697399255e-07,
      "loss": 0.72,
      "step": 51450
    },
    {
      "epoch": 0.8010617512668623,
      "grad_norm": 4.604339122772217,
      "learning_rate": 2.0941970796938757e-07,
      "loss": 0.7504,
      "step": 51455
    },
    {
      "epoch": 0.8011395922688317,
      "grad_norm": 4.967108249664307,
      "learning_rate": 2.0933776896478262e-07,
      "loss": 0.7592,
      "step": 51460
    },
    {
      "epoch": 0.8012174332708011,
      "grad_norm": 4.045896053314209,
      "learning_rate": 2.0925582996017762e-07,
      "loss": 0.7434,
      "step": 51465
    },
    {
      "epoch": 0.8012952742727705,
      "grad_norm": 3.4572672843933105,
      "learning_rate": 2.0917389095557266e-07,
      "loss": 0.7058,
      "step": 51470
    },
    {
      "epoch": 0.8013731152747399,
      "grad_norm": 2.8202171325683594,
      "learning_rate": 2.0909195195096769e-07,
      "loss": 0.6592,
      "step": 51475
    },
    {
      "epoch": 0.8014509562767091,
      "grad_norm": 12.622154235839844,
      "learning_rate": 2.0901001294636273e-07,
      "loss": 0.7704,
      "step": 51480
    },
    {
      "epoch": 0.8015287972786785,
      "grad_norm": 4.271938800811768,
      "learning_rate": 2.0892807394175773e-07,
      "loss": 0.6879,
      "step": 51485
    },
    {
      "epoch": 0.8016066382806479,
      "grad_norm": 4.102352619171143,
      "learning_rate": 2.0884613493715278e-07,
      "loss": 0.7485,
      "step": 51490
    },
    {
      "epoch": 0.8016844792826173,
      "grad_norm": 4.173807144165039,
      "learning_rate": 2.087641959325478e-07,
      "loss": 0.8591,
      "step": 51495
    },
    {
      "epoch": 0.8017623202845867,
      "grad_norm": 3.807666063308716,
      "learning_rate": 2.0868225692794282e-07,
      "loss": 0.8112,
      "step": 51500
    },
    {
      "epoch": 0.8018401612865561,
      "grad_norm": 3.6424922943115234,
      "learning_rate": 2.0860031792333784e-07,
      "loss": 0.7123,
      "step": 51505
    },
    {
      "epoch": 0.8019180022885255,
      "grad_norm": 3.8213181495666504,
      "learning_rate": 2.085183789187329e-07,
      "loss": 0.77,
      "step": 51510
    },
    {
      "epoch": 0.8019958432904948,
      "grad_norm": 2.7489006519317627,
      "learning_rate": 2.0843643991412791e-07,
      "loss": 0.7868,
      "step": 51515
    },
    {
      "epoch": 0.8020736842924642,
      "grad_norm": 4.170252799987793,
      "learning_rate": 2.0835450090952294e-07,
      "loss": 0.6539,
      "step": 51520
    },
    {
      "epoch": 0.8021515252944336,
      "grad_norm": 5.151482582092285,
      "learning_rate": 2.0827256190491796e-07,
      "loss": 0.6858,
      "step": 51525
    },
    {
      "epoch": 0.802229366296403,
      "grad_norm": 2.7144248485565186,
      "learning_rate": 2.08190622900313e-07,
      "loss": 0.8334,
      "step": 51530
    },
    {
      "epoch": 0.8023072072983723,
      "grad_norm": 3.2036197185516357,
      "learning_rate": 2.08108683895708e-07,
      "loss": 0.7247,
      "step": 51535
    },
    {
      "epoch": 0.8023850483003417,
      "grad_norm": 3.761096715927124,
      "learning_rate": 2.0802674489110305e-07,
      "loss": 0.8199,
      "step": 51540
    },
    {
      "epoch": 0.8024628893023111,
      "grad_norm": 3.4821853637695312,
      "learning_rate": 2.0794480588649807e-07,
      "loss": 0.6976,
      "step": 51545
    },
    {
      "epoch": 0.8025407303042805,
      "grad_norm": 2.880582571029663,
      "learning_rate": 2.0786286688189312e-07,
      "loss": 0.8628,
      "step": 51550
    },
    {
      "epoch": 0.8026185713062498,
      "grad_norm": 3.653869152069092,
      "learning_rate": 2.0778092787728812e-07,
      "loss": 0.8359,
      "step": 51555
    },
    {
      "epoch": 0.8026964123082192,
      "grad_norm": 3.7833058834075928,
      "learning_rate": 2.0769898887268316e-07,
      "loss": 0.6936,
      "step": 51560
    },
    {
      "epoch": 0.8027742533101886,
      "grad_norm": 2.8439316749572754,
      "learning_rate": 2.076170498680782e-07,
      "loss": 0.6291,
      "step": 51565
    },
    {
      "epoch": 0.802852094312158,
      "grad_norm": 4.462703227996826,
      "learning_rate": 2.0753511086347324e-07,
      "loss": 0.7069,
      "step": 51570
    },
    {
      "epoch": 0.8029299353141274,
      "grad_norm": 4.544771671295166,
      "learning_rate": 2.0745317185886823e-07,
      "loss": 0.6625,
      "step": 51575
    },
    {
      "epoch": 0.8030077763160968,
      "grad_norm": 7.09713077545166,
      "learning_rate": 2.0737123285426328e-07,
      "loss": 0.8227,
      "step": 51580
    },
    {
      "epoch": 0.8030856173180662,
      "grad_norm": 4.412929058074951,
      "learning_rate": 2.072892938496583e-07,
      "loss": 0.724,
      "step": 51585
    },
    {
      "epoch": 0.8031634583200356,
      "grad_norm": 3.181428909301758,
      "learning_rate": 2.0720735484505332e-07,
      "loss": 0.7941,
      "step": 51590
    },
    {
      "epoch": 0.8032412993220048,
      "grad_norm": 3.3188979625701904,
      "learning_rate": 2.0712541584044834e-07,
      "loss": 0.6406,
      "step": 51595
    },
    {
      "epoch": 0.8033191403239742,
      "grad_norm": 4.440804958343506,
      "learning_rate": 2.070434768358434e-07,
      "loss": 0.7126,
      "step": 51600
    },
    {
      "epoch": 0.8033969813259436,
      "grad_norm": 2.3336360454559326,
      "learning_rate": 2.0696153783123842e-07,
      "loss": 0.6746,
      "step": 51605
    },
    {
      "epoch": 0.803474822327913,
      "grad_norm": 5.531771183013916,
      "learning_rate": 2.0687959882663344e-07,
      "loss": 0.7448,
      "step": 51610
    },
    {
      "epoch": 0.8035526633298824,
      "grad_norm": 3.3685901165008545,
      "learning_rate": 2.0679765982202849e-07,
      "loss": 0.7064,
      "step": 51615
    },
    {
      "epoch": 0.8036305043318518,
      "grad_norm": 5.938897609710693,
      "learning_rate": 2.067157208174235e-07,
      "loss": 0.7641,
      "step": 51620
    },
    {
      "epoch": 0.8037083453338212,
      "grad_norm": 4.730757713317871,
      "learning_rate": 2.0663378181281853e-07,
      "loss": 0.8697,
      "step": 51625
    },
    {
      "epoch": 0.8037861863357905,
      "grad_norm": 3.6286628246307373,
      "learning_rate": 2.0655184280821355e-07,
      "loss": 0.626,
      "step": 51630
    },
    {
      "epoch": 0.8038640273377599,
      "grad_norm": 7.627096652984619,
      "learning_rate": 2.064699038036086e-07,
      "loss": 0.5939,
      "step": 51635
    },
    {
      "epoch": 0.8039418683397292,
      "grad_norm": 3.7657241821289062,
      "learning_rate": 2.0638796479900362e-07,
      "loss": 0.7783,
      "step": 51640
    },
    {
      "epoch": 0.8040197093416986,
      "grad_norm": 3.2721621990203857,
      "learning_rate": 2.0630602579439864e-07,
      "loss": 0.6933,
      "step": 51645
    },
    {
      "epoch": 0.804097550343668,
      "grad_norm": 2.920910120010376,
      "learning_rate": 2.0622408678979367e-07,
      "loss": 0.829,
      "step": 51650
    },
    {
      "epoch": 0.8041753913456374,
      "grad_norm": 4.89809513092041,
      "learning_rate": 2.0614214778518871e-07,
      "loss": 0.778,
      "step": 51655
    },
    {
      "epoch": 0.8042532323476068,
      "grad_norm": 5.254788875579834,
      "learning_rate": 2.060602087805837e-07,
      "loss": 0.7564,
      "step": 51660
    },
    {
      "epoch": 0.8043310733495762,
      "grad_norm": 2.574824333190918,
      "learning_rate": 2.0597826977597876e-07,
      "loss": 0.7146,
      "step": 51665
    },
    {
      "epoch": 0.8044089143515455,
      "grad_norm": 4.6628007888793945,
      "learning_rate": 2.0589633077137378e-07,
      "loss": 0.899,
      "step": 51670
    },
    {
      "epoch": 0.8044867553535149,
      "grad_norm": 9.768836975097656,
      "learning_rate": 2.0581439176676883e-07,
      "loss": 0.8784,
      "step": 51675
    },
    {
      "epoch": 0.8045645963554843,
      "grad_norm": 5.513589859008789,
      "learning_rate": 2.0573245276216382e-07,
      "loss": 0.8031,
      "step": 51680
    },
    {
      "epoch": 0.8046424373574537,
      "grad_norm": 2.8976688385009766,
      "learning_rate": 2.0565051375755887e-07,
      "loss": 0.74,
      "step": 51685
    },
    {
      "epoch": 0.8047202783594231,
      "grad_norm": 3.004197359085083,
      "learning_rate": 2.055685747529539e-07,
      "loss": 0.6594,
      "step": 51690
    },
    {
      "epoch": 0.8047981193613924,
      "grad_norm": 4.432955265045166,
      "learning_rate": 2.0548663574834894e-07,
      "loss": 0.7336,
      "step": 51695
    },
    {
      "epoch": 0.8048759603633618,
      "grad_norm": 3.42651629447937,
      "learning_rate": 2.0540469674374394e-07,
      "loss": 0.7948,
      "step": 51700
    },
    {
      "epoch": 0.8049538013653311,
      "grad_norm": 3.4903786182403564,
      "learning_rate": 2.0532275773913899e-07,
      "loss": 0.6307,
      "step": 51705
    },
    {
      "epoch": 0.8050316423673005,
      "grad_norm": 3.3571455478668213,
      "learning_rate": 2.05240818734534e-07,
      "loss": 0.8268,
      "step": 51710
    },
    {
      "epoch": 0.8051094833692699,
      "grad_norm": 3.2211015224456787,
      "learning_rate": 2.0515887972992903e-07,
      "loss": 0.8182,
      "step": 51715
    },
    {
      "epoch": 0.8051873243712393,
      "grad_norm": 3.302103042602539,
      "learning_rate": 2.0507694072532405e-07,
      "loss": 0.8524,
      "step": 51720
    },
    {
      "epoch": 0.8052651653732087,
      "grad_norm": 3.8519070148468018,
      "learning_rate": 2.049950017207191e-07,
      "loss": 0.7614,
      "step": 51725
    },
    {
      "epoch": 0.8053430063751781,
      "grad_norm": 3.621553659439087,
      "learning_rate": 2.0491306271611412e-07,
      "loss": 0.7189,
      "step": 51730
    },
    {
      "epoch": 0.8054208473771475,
      "grad_norm": 3.6923582553863525,
      "learning_rate": 2.0483112371150914e-07,
      "loss": 0.7194,
      "step": 51735
    },
    {
      "epoch": 0.8054986883791169,
      "grad_norm": 3.5743603706359863,
      "learning_rate": 2.0474918470690417e-07,
      "loss": 0.6558,
      "step": 51740
    },
    {
      "epoch": 0.8055765293810861,
      "grad_norm": 2.744717836380005,
      "learning_rate": 2.0466724570229921e-07,
      "loss": 0.7017,
      "step": 51745
    },
    {
      "epoch": 0.8056543703830555,
      "grad_norm": 3.285449981689453,
      "learning_rate": 2.045853066976942e-07,
      "loss": 0.713,
      "step": 51750
    },
    {
      "epoch": 0.8057322113850249,
      "grad_norm": 4.333995819091797,
      "learning_rate": 2.0450336769308926e-07,
      "loss": 0.8342,
      "step": 51755
    },
    {
      "epoch": 0.8058100523869943,
      "grad_norm": 4.934929847717285,
      "learning_rate": 2.0442142868848428e-07,
      "loss": 0.7771,
      "step": 51760
    },
    {
      "epoch": 0.8058878933889637,
      "grad_norm": 3.8606655597686768,
      "learning_rate": 2.0433948968387933e-07,
      "loss": 0.7007,
      "step": 51765
    },
    {
      "epoch": 0.8059657343909331,
      "grad_norm": 3.535297155380249,
      "learning_rate": 2.0425755067927432e-07,
      "loss": 0.7124,
      "step": 51770
    },
    {
      "epoch": 0.8060435753929025,
      "grad_norm": 4.34794282913208,
      "learning_rate": 2.0417561167466937e-07,
      "loss": 0.838,
      "step": 51775
    },
    {
      "epoch": 0.8061214163948718,
      "grad_norm": 3.074516773223877,
      "learning_rate": 2.040936726700644e-07,
      "loss": 0.7086,
      "step": 51780
    },
    {
      "epoch": 0.8061992573968412,
      "grad_norm": 3.7737574577331543,
      "learning_rate": 2.0401173366545942e-07,
      "loss": 0.7551,
      "step": 51785
    },
    {
      "epoch": 0.8062770983988106,
      "grad_norm": 3.0810165405273438,
      "learning_rate": 2.0392979466085444e-07,
      "loss": 0.7246,
      "step": 51790
    },
    {
      "epoch": 0.80635493940078,
      "grad_norm": 3.7942569255828857,
      "learning_rate": 2.0384785565624949e-07,
      "loss": 0.6653,
      "step": 51795
    },
    {
      "epoch": 0.8064327804027493,
      "grad_norm": 2.607365846633911,
      "learning_rate": 2.037659166516445e-07,
      "loss": 0.6899,
      "step": 51800
    },
    {
      "epoch": 0.8065106214047187,
      "grad_norm": 11.3803071975708,
      "learning_rate": 2.0368397764703953e-07,
      "loss": 0.7719,
      "step": 51805
    },
    {
      "epoch": 0.8065884624066881,
      "grad_norm": 4.183210849761963,
      "learning_rate": 2.0360203864243455e-07,
      "loss": 0.8836,
      "step": 51810
    },
    {
      "epoch": 0.8066663034086575,
      "grad_norm": 2.7863495349884033,
      "learning_rate": 2.035200996378296e-07,
      "loss": 0.7001,
      "step": 51815
    },
    {
      "epoch": 0.8067441444106268,
      "grad_norm": 6.1742658615112305,
      "learning_rate": 2.0343816063322462e-07,
      "loss": 0.7511,
      "step": 51820
    },
    {
      "epoch": 0.8068219854125962,
      "grad_norm": 2.61411190032959,
      "learning_rate": 2.0335622162861964e-07,
      "loss": 0.6619,
      "step": 51825
    },
    {
      "epoch": 0.8068998264145656,
      "grad_norm": 6.050604820251465,
      "learning_rate": 2.0327428262401467e-07,
      "loss": 0.7994,
      "step": 51830
    },
    {
      "epoch": 0.806977667416535,
      "grad_norm": 2.8306872844696045,
      "learning_rate": 2.0319234361940971e-07,
      "loss": 0.5754,
      "step": 51835
    },
    {
      "epoch": 0.8070555084185044,
      "grad_norm": 4.8263959884643555,
      "learning_rate": 2.031104046148047e-07,
      "loss": 0.7318,
      "step": 51840
    },
    {
      "epoch": 0.8071333494204738,
      "grad_norm": 2.955113172531128,
      "learning_rate": 2.0302846561019976e-07,
      "loss": 0.6964,
      "step": 51845
    },
    {
      "epoch": 0.8072111904224432,
      "grad_norm": 3.2799088954925537,
      "learning_rate": 2.0294652660559478e-07,
      "loss": 0.7307,
      "step": 51850
    },
    {
      "epoch": 0.8072890314244124,
      "grad_norm": 4.680817127227783,
      "learning_rate": 2.0286458760098983e-07,
      "loss": 0.7497,
      "step": 51855
    },
    {
      "epoch": 0.8073668724263818,
      "grad_norm": 3.59967041015625,
      "learning_rate": 2.0278264859638482e-07,
      "loss": 0.8303,
      "step": 51860
    },
    {
      "epoch": 0.8074447134283512,
      "grad_norm": 4.052394390106201,
      "learning_rate": 2.0270070959177987e-07,
      "loss": 0.7858,
      "step": 51865
    },
    {
      "epoch": 0.8075225544303206,
      "grad_norm": 4.459505081176758,
      "learning_rate": 2.026187705871749e-07,
      "loss": 0.7059,
      "step": 51870
    },
    {
      "epoch": 0.80760039543229,
      "grad_norm": 1.9578063488006592,
      "learning_rate": 2.0253683158256992e-07,
      "loss": 0.7385,
      "step": 51875
    },
    {
      "epoch": 0.8076782364342594,
      "grad_norm": 6.517970085144043,
      "learning_rate": 2.0245489257796494e-07,
      "loss": 0.7385,
      "step": 51880
    },
    {
      "epoch": 0.8077560774362288,
      "grad_norm": 3.448279857635498,
      "learning_rate": 2.0237295357335999e-07,
      "loss": 0.6437,
      "step": 51885
    },
    {
      "epoch": 0.8078339184381982,
      "grad_norm": 3.563441276550293,
      "learning_rate": 2.02291014568755e-07,
      "loss": 0.72,
      "step": 51890
    },
    {
      "epoch": 0.8079117594401675,
      "grad_norm": 2.8551816940307617,
      "learning_rate": 2.0220907556415003e-07,
      "loss": 0.689,
      "step": 51895
    },
    {
      "epoch": 0.8079896004421369,
      "grad_norm": 3.9990174770355225,
      "learning_rate": 2.0212713655954505e-07,
      "loss": 0.728,
      "step": 51900
    },
    {
      "epoch": 0.8080674414441062,
      "grad_norm": 4.831725597381592,
      "learning_rate": 2.020451975549401e-07,
      "loss": 0.8356,
      "step": 51905
    },
    {
      "epoch": 0.8081452824460756,
      "grad_norm": 5.097940444946289,
      "learning_rate": 2.019632585503351e-07,
      "loss": 0.7534,
      "step": 51910
    },
    {
      "epoch": 0.808223123448045,
      "grad_norm": 4.240472316741943,
      "learning_rate": 2.0188131954573014e-07,
      "loss": 0.715,
      "step": 51915
    },
    {
      "epoch": 0.8083009644500144,
      "grad_norm": 4.367761135101318,
      "learning_rate": 2.0179938054112517e-07,
      "loss": 0.777,
      "step": 51920
    },
    {
      "epoch": 0.8083788054519838,
      "grad_norm": 3.029447078704834,
      "learning_rate": 2.0171744153652022e-07,
      "loss": 0.6914,
      "step": 51925
    },
    {
      "epoch": 0.8084566464539532,
      "grad_norm": 2.484893560409546,
      "learning_rate": 2.0163550253191524e-07,
      "loss": 0.7379,
      "step": 51930
    },
    {
      "epoch": 0.8085344874559225,
      "grad_norm": 4.217326641082764,
      "learning_rate": 2.0155356352731026e-07,
      "loss": 0.798,
      "step": 51935
    },
    {
      "epoch": 0.8086123284578919,
      "grad_norm": 5.119178295135498,
      "learning_rate": 2.014716245227053e-07,
      "loss": 0.7594,
      "step": 51940
    },
    {
      "epoch": 0.8086901694598613,
      "grad_norm": 6.977635383605957,
      "learning_rate": 2.0138968551810033e-07,
      "loss": 0.6828,
      "step": 51945
    },
    {
      "epoch": 0.8087680104618307,
      "grad_norm": 5.062507152557373,
      "learning_rate": 2.0130774651349535e-07,
      "loss": 0.6981,
      "step": 51950
    },
    {
      "epoch": 0.8088458514638001,
      "grad_norm": 3.399139881134033,
      "learning_rate": 2.0122580750889037e-07,
      "loss": 0.7241,
      "step": 51955
    },
    {
      "epoch": 0.8089236924657695,
      "grad_norm": 3.078021764755249,
      "learning_rate": 2.0114386850428542e-07,
      "loss": 0.757,
      "step": 51960
    },
    {
      "epoch": 0.8090015334677388,
      "grad_norm": 3.7249794006347656,
      "learning_rate": 2.0106192949968042e-07,
      "loss": 0.6007,
      "step": 51965
    },
    {
      "epoch": 0.8090793744697081,
      "grad_norm": 5.958231449127197,
      "learning_rate": 2.0097999049507547e-07,
      "loss": 0.8399,
      "step": 51970
    },
    {
      "epoch": 0.8091572154716775,
      "grad_norm": 6.256702423095703,
      "learning_rate": 2.008980514904705e-07,
      "loss": 0.6294,
      "step": 51975
    },
    {
      "epoch": 0.8092350564736469,
      "grad_norm": 5.778950214385986,
      "learning_rate": 2.0081611248586554e-07,
      "loss": 0.7046,
      "step": 51980
    },
    {
      "epoch": 0.8093128974756163,
      "grad_norm": 11.596414566040039,
      "learning_rate": 2.0073417348126053e-07,
      "loss": 0.7206,
      "step": 51985
    },
    {
      "epoch": 0.8093907384775857,
      "grad_norm": 4.344325065612793,
      "learning_rate": 2.0065223447665558e-07,
      "loss": 0.7186,
      "step": 51990
    },
    {
      "epoch": 0.8094685794795551,
      "grad_norm": 2.4756743907928467,
      "learning_rate": 2.005702954720506e-07,
      "loss": 0.6297,
      "step": 51995
    },
    {
      "epoch": 0.8095464204815245,
      "grad_norm": 3.2217185497283936,
      "learning_rate": 2.0048835646744562e-07,
      "loss": 0.7459,
      "step": 52000
    },
    {
      "epoch": 0.8096242614834939,
      "grad_norm": 2.9370317459106445,
      "learning_rate": 2.0040641746284065e-07,
      "loss": 0.7124,
      "step": 52005
    },
    {
      "epoch": 0.8097021024854631,
      "grad_norm": 4.673339366912842,
      "learning_rate": 2.003244784582357e-07,
      "loss": 0.7707,
      "step": 52010
    },
    {
      "epoch": 0.8097799434874325,
      "grad_norm": 2.9292311668395996,
      "learning_rate": 2.0024253945363072e-07,
      "loss": 0.7257,
      "step": 52015
    },
    {
      "epoch": 0.8098577844894019,
      "grad_norm": 3.9582130908966064,
      "learning_rate": 2.0016060044902574e-07,
      "loss": 0.7298,
      "step": 52020
    },
    {
      "epoch": 0.8099356254913713,
      "grad_norm": 8.15613842010498,
      "learning_rate": 2.0007866144442076e-07,
      "loss": 0.7154,
      "step": 52025
    },
    {
      "epoch": 0.8100134664933407,
      "grad_norm": 2.8851170539855957,
      "learning_rate": 1.999967224398158e-07,
      "loss": 0.8235,
      "step": 52030
    },
    {
      "epoch": 0.8100913074953101,
      "grad_norm": 4.326573371887207,
      "learning_rate": 1.999147834352108e-07,
      "loss": 0.7067,
      "step": 52035
    },
    {
      "epoch": 0.8101691484972795,
      "grad_norm": 4.022038459777832,
      "learning_rate": 1.9983284443060585e-07,
      "loss": 0.6544,
      "step": 52040
    },
    {
      "epoch": 0.8102469894992488,
      "grad_norm": 4.400419235229492,
      "learning_rate": 1.9975090542600087e-07,
      "loss": 0.7449,
      "step": 52045
    },
    {
      "epoch": 0.8103248305012182,
      "grad_norm": 5.081811904907227,
      "learning_rate": 1.9966896642139592e-07,
      "loss": 0.6684,
      "step": 52050
    },
    {
      "epoch": 0.8104026715031876,
      "grad_norm": 4.464503288269043,
      "learning_rate": 1.9958702741679092e-07,
      "loss": 0.7566,
      "step": 52055
    },
    {
      "epoch": 0.810480512505157,
      "grad_norm": 4.095409393310547,
      "learning_rate": 1.9950508841218597e-07,
      "loss": 0.6994,
      "step": 52060
    },
    {
      "epoch": 0.8105583535071264,
      "grad_norm": 5.934963226318359,
      "learning_rate": 1.99423149407581e-07,
      "loss": 0.6964,
      "step": 52065
    },
    {
      "epoch": 0.8106361945090957,
      "grad_norm": 3.1645925045013428,
      "learning_rate": 1.9934121040297604e-07,
      "loss": 0.696,
      "step": 52070
    },
    {
      "epoch": 0.8107140355110651,
      "grad_norm": 2.980380058288574,
      "learning_rate": 1.9925927139837103e-07,
      "loss": 0.7452,
      "step": 52075
    },
    {
      "epoch": 0.8107918765130345,
      "grad_norm": 4.4078192710876465,
      "learning_rate": 1.9917733239376608e-07,
      "loss": 0.7367,
      "step": 52080
    },
    {
      "epoch": 0.8108697175150038,
      "grad_norm": 3.354391574859619,
      "learning_rate": 1.990953933891611e-07,
      "loss": 0.7631,
      "step": 52085
    },
    {
      "epoch": 0.8109475585169732,
      "grad_norm": 4.8928937911987305,
      "learning_rate": 1.9901345438455612e-07,
      "loss": 0.6983,
      "step": 52090
    },
    {
      "epoch": 0.8110253995189426,
      "grad_norm": 6.480390548706055,
      "learning_rate": 1.9893151537995115e-07,
      "loss": 0.6034,
      "step": 52095
    },
    {
      "epoch": 0.811103240520912,
      "grad_norm": 3.685262680053711,
      "learning_rate": 1.988495763753462e-07,
      "loss": 0.8242,
      "step": 52100
    },
    {
      "epoch": 0.8111810815228814,
      "grad_norm": 4.094353675842285,
      "learning_rate": 1.9876763737074122e-07,
      "loss": 0.6349,
      "step": 52105
    },
    {
      "epoch": 0.8112589225248508,
      "grad_norm": 3.5944437980651855,
      "learning_rate": 1.9868569836613624e-07,
      "loss": 0.6929,
      "step": 52110
    },
    {
      "epoch": 0.8113367635268202,
      "grad_norm": 3.555060625076294,
      "learning_rate": 1.9860375936153126e-07,
      "loss": 0.7067,
      "step": 52115
    },
    {
      "epoch": 0.8114146045287894,
      "grad_norm": 3.6349096298217773,
      "learning_rate": 1.985218203569263e-07,
      "loss": 0.6455,
      "step": 52120
    },
    {
      "epoch": 0.8114924455307588,
      "grad_norm": 3.860948324203491,
      "learning_rate": 1.984398813523213e-07,
      "loss": 0.7823,
      "step": 52125
    },
    {
      "epoch": 0.8115702865327282,
      "grad_norm": 3.028873920440674,
      "learning_rate": 1.9835794234771635e-07,
      "loss": 0.7058,
      "step": 52130
    },
    {
      "epoch": 0.8116481275346976,
      "grad_norm": 3.8320565223693848,
      "learning_rate": 1.9827600334311137e-07,
      "loss": 0.8005,
      "step": 52135
    },
    {
      "epoch": 0.811725968536667,
      "grad_norm": 3.3632454872131348,
      "learning_rate": 1.9819406433850642e-07,
      "loss": 0.8465,
      "step": 52140
    },
    {
      "epoch": 0.8118038095386364,
      "grad_norm": 3.013026714324951,
      "learning_rate": 1.9811212533390142e-07,
      "loss": 0.6041,
      "step": 52145
    },
    {
      "epoch": 0.8118816505406058,
      "grad_norm": 4.278226852416992,
      "learning_rate": 1.9803018632929647e-07,
      "loss": 0.762,
      "step": 52150
    },
    {
      "epoch": 0.8119594915425752,
      "grad_norm": 4.990754127502441,
      "learning_rate": 1.979482473246915e-07,
      "loss": 0.6848,
      "step": 52155
    },
    {
      "epoch": 0.8120373325445445,
      "grad_norm": 2.897615432739258,
      "learning_rate": 1.978663083200865e-07,
      "loss": 0.6322,
      "step": 52160
    },
    {
      "epoch": 0.8121151735465139,
      "grad_norm": 4.446718692779541,
      "learning_rate": 1.9778436931548153e-07,
      "loss": 0.6595,
      "step": 52165
    },
    {
      "epoch": 0.8121930145484832,
      "grad_norm": 6.578906536102295,
      "learning_rate": 1.9770243031087658e-07,
      "loss": 0.8173,
      "step": 52170
    },
    {
      "epoch": 0.8122708555504526,
      "grad_norm": 4.254462242126465,
      "learning_rate": 1.976204913062716e-07,
      "loss": 0.5847,
      "step": 52175
    },
    {
      "epoch": 0.812348696552422,
      "grad_norm": 4.793978214263916,
      "learning_rate": 1.9753855230166662e-07,
      "loss": 0.7679,
      "step": 52180
    },
    {
      "epoch": 0.8124265375543914,
      "grad_norm": 2.870114326477051,
      "learning_rate": 1.9745661329706165e-07,
      "loss": 0.6965,
      "step": 52185
    },
    {
      "epoch": 0.8125043785563608,
      "grad_norm": 4.388121128082275,
      "learning_rate": 1.973746742924567e-07,
      "loss": 0.8108,
      "step": 52190
    },
    {
      "epoch": 0.8125822195583302,
      "grad_norm": 2.573714256286621,
      "learning_rate": 1.972927352878517e-07,
      "loss": 0.6834,
      "step": 52195
    },
    {
      "epoch": 0.8126600605602995,
      "grad_norm": 3.315488576889038,
      "learning_rate": 1.9721079628324674e-07,
      "loss": 0.6398,
      "step": 52200
    },
    {
      "epoch": 0.8127379015622689,
      "grad_norm": 5.868358135223389,
      "learning_rate": 1.9712885727864176e-07,
      "loss": 0.8257,
      "step": 52205
    },
    {
      "epoch": 0.8128157425642383,
      "grad_norm": 2.4226083755493164,
      "learning_rate": 1.970469182740368e-07,
      "loss": 0.6907,
      "step": 52210
    },
    {
      "epoch": 0.8128935835662077,
      "grad_norm": 4.197224140167236,
      "learning_rate": 1.969649792694318e-07,
      "loss": 0.7042,
      "step": 52215
    },
    {
      "epoch": 0.8129714245681771,
      "grad_norm": 3.3440825939178467,
      "learning_rate": 1.9688304026482685e-07,
      "loss": 0.76,
      "step": 52220
    },
    {
      "epoch": 0.8130492655701465,
      "grad_norm": 3.4761550426483154,
      "learning_rate": 1.9680110126022187e-07,
      "loss": 0.6476,
      "step": 52225
    },
    {
      "epoch": 0.8131271065721158,
      "grad_norm": 3.43440580368042,
      "learning_rate": 1.9671916225561692e-07,
      "loss": 0.6588,
      "step": 52230
    },
    {
      "epoch": 0.8132049475740851,
      "grad_norm": 4.573919773101807,
      "learning_rate": 1.9663722325101192e-07,
      "loss": 0.7141,
      "step": 52235
    },
    {
      "epoch": 0.8132827885760545,
      "grad_norm": 4.084165573120117,
      "learning_rate": 1.9655528424640697e-07,
      "loss": 0.7186,
      "step": 52240
    },
    {
      "epoch": 0.8133606295780239,
      "grad_norm": 3.563990831375122,
      "learning_rate": 1.96473345241802e-07,
      "loss": 0.7511,
      "step": 52245
    },
    {
      "epoch": 0.8134384705799933,
      "grad_norm": 2.5278565883636475,
      "learning_rate": 1.96391406237197e-07,
      "loss": 0.7548,
      "step": 52250
    },
    {
      "epoch": 0.8135163115819627,
      "grad_norm": 2.800311326980591,
      "learning_rate": 1.9630946723259206e-07,
      "loss": 0.702,
      "step": 52255
    },
    {
      "epoch": 0.8135941525839321,
      "grad_norm": 2.5655510425567627,
      "learning_rate": 1.9622752822798708e-07,
      "loss": 0.6789,
      "step": 52260
    },
    {
      "epoch": 0.8136719935859015,
      "grad_norm": 3.293060779571533,
      "learning_rate": 1.9614558922338213e-07,
      "loss": 0.6832,
      "step": 52265
    },
    {
      "epoch": 0.8137498345878709,
      "grad_norm": 3.828979969024658,
      "learning_rate": 1.9606365021877712e-07,
      "loss": 0.6402,
      "step": 52270
    },
    {
      "epoch": 0.8138276755898401,
      "grad_norm": 4.5994415283203125,
      "learning_rate": 1.9598171121417217e-07,
      "loss": 0.6863,
      "step": 52275
    },
    {
      "epoch": 0.8139055165918095,
      "grad_norm": 2.5963430404663086,
      "learning_rate": 1.958997722095672e-07,
      "loss": 0.6839,
      "step": 52280
    },
    {
      "epoch": 0.8139833575937789,
      "grad_norm": 4.073090553283691,
      "learning_rate": 1.9581783320496222e-07,
      "loss": 0.7136,
      "step": 52285
    },
    {
      "epoch": 0.8140611985957483,
      "grad_norm": 5.707798957824707,
      "learning_rate": 1.9573589420035724e-07,
      "loss": 0.7443,
      "step": 52290
    },
    {
      "epoch": 0.8141390395977177,
      "grad_norm": 3.1592884063720703,
      "learning_rate": 1.956539551957523e-07,
      "loss": 0.7481,
      "step": 52295
    },
    {
      "epoch": 0.8142168805996871,
      "grad_norm": 3.313603162765503,
      "learning_rate": 1.955720161911473e-07,
      "loss": 0.7437,
      "step": 52300
    },
    {
      "epoch": 0.8142947216016565,
      "grad_norm": 2.950671672821045,
      "learning_rate": 1.9549007718654233e-07,
      "loss": 0.7466,
      "step": 52305
    },
    {
      "epoch": 0.8143725626036258,
      "grad_norm": 5.0073347091674805,
      "learning_rate": 1.9540813818193735e-07,
      "loss": 0.7783,
      "step": 52310
    },
    {
      "epoch": 0.8144504036055952,
      "grad_norm": 7.320366859436035,
      "learning_rate": 1.953261991773324e-07,
      "loss": 0.7318,
      "step": 52315
    },
    {
      "epoch": 0.8145282446075646,
      "grad_norm": 4.105320453643799,
      "learning_rate": 1.952442601727274e-07,
      "loss": 0.7256,
      "step": 52320
    },
    {
      "epoch": 0.814606085609534,
      "grad_norm": 3.2573704719543457,
      "learning_rate": 1.9516232116812245e-07,
      "loss": 0.6774,
      "step": 52325
    },
    {
      "epoch": 0.8146839266115034,
      "grad_norm": 2.620844602584839,
      "learning_rate": 1.9508038216351747e-07,
      "loss": 0.7556,
      "step": 52330
    },
    {
      "epoch": 0.8147617676134727,
      "grad_norm": 3.0829238891601562,
      "learning_rate": 1.9499844315891252e-07,
      "loss": 0.8005,
      "step": 52335
    },
    {
      "epoch": 0.8148396086154421,
      "grad_norm": 4.276255130767822,
      "learning_rate": 1.949165041543075e-07,
      "loss": 0.7347,
      "step": 52340
    },
    {
      "epoch": 0.8149174496174115,
      "grad_norm": 4.506711959838867,
      "learning_rate": 1.9483456514970256e-07,
      "loss": 0.7042,
      "step": 52345
    },
    {
      "epoch": 0.8149952906193808,
      "grad_norm": 2.583613634109497,
      "learning_rate": 1.9475262614509758e-07,
      "loss": 0.681,
      "step": 52350
    },
    {
      "epoch": 0.8150731316213502,
      "grad_norm": 4.825068950653076,
      "learning_rate": 1.9467068714049263e-07,
      "loss": 0.7708,
      "step": 52355
    },
    {
      "epoch": 0.8151509726233196,
      "grad_norm": 5.83071231842041,
      "learning_rate": 1.9458874813588762e-07,
      "loss": 0.761,
      "step": 52360
    },
    {
      "epoch": 0.815228813625289,
      "grad_norm": 3.7539734840393066,
      "learning_rate": 1.9450680913128267e-07,
      "loss": 0.8128,
      "step": 52365
    },
    {
      "epoch": 0.8153066546272584,
      "grad_norm": 6.027698993682861,
      "learning_rate": 1.944248701266777e-07,
      "loss": 0.7527,
      "step": 52370
    },
    {
      "epoch": 0.8153844956292278,
      "grad_norm": 5.1614203453063965,
      "learning_rate": 1.9434293112207272e-07,
      "loss": 0.7518,
      "step": 52375
    },
    {
      "epoch": 0.8154623366311972,
      "grad_norm": 3.716229200363159,
      "learning_rate": 1.9426099211746774e-07,
      "loss": 0.8104,
      "step": 52380
    },
    {
      "epoch": 0.8155401776331664,
      "grad_norm": 4.197135925292969,
      "learning_rate": 1.941790531128628e-07,
      "loss": 0.7065,
      "step": 52385
    },
    {
      "epoch": 0.8156180186351358,
      "grad_norm": 3.5560615062713623,
      "learning_rate": 1.940971141082578e-07,
      "loss": 0.8292,
      "step": 52390
    },
    {
      "epoch": 0.8156958596371052,
      "grad_norm": 3.305527925491333,
      "learning_rate": 1.9401517510365283e-07,
      "loss": 0.8721,
      "step": 52395
    },
    {
      "epoch": 0.8157737006390746,
      "grad_norm": 4.950320243835449,
      "learning_rate": 1.9393323609904785e-07,
      "loss": 0.7289,
      "step": 52400
    },
    {
      "epoch": 0.815851541641044,
      "grad_norm": 4.529018878936768,
      "learning_rate": 1.938512970944429e-07,
      "loss": 0.8598,
      "step": 52405
    },
    {
      "epoch": 0.8159293826430134,
      "grad_norm": 3.4007904529571533,
      "learning_rate": 1.937693580898379e-07,
      "loss": 0.7344,
      "step": 52410
    },
    {
      "epoch": 0.8160072236449828,
      "grad_norm": 4.715421199798584,
      "learning_rate": 1.9368741908523295e-07,
      "loss": 0.8048,
      "step": 52415
    },
    {
      "epoch": 0.8160850646469522,
      "grad_norm": 6.000762462615967,
      "learning_rate": 1.9360548008062797e-07,
      "loss": 0.7555,
      "step": 52420
    },
    {
      "epoch": 0.8161629056489215,
      "grad_norm": 2.742121696472168,
      "learning_rate": 1.9352354107602302e-07,
      "loss": 0.7868,
      "step": 52425
    },
    {
      "epoch": 0.8162407466508909,
      "grad_norm": 8.21036434173584,
      "learning_rate": 1.93441602071418e-07,
      "loss": 0.7047,
      "step": 52430
    },
    {
      "epoch": 0.8163185876528603,
      "grad_norm": 4.338558673858643,
      "learning_rate": 1.9335966306681306e-07,
      "loss": 0.8129,
      "step": 52435
    },
    {
      "epoch": 0.8163964286548296,
      "grad_norm": 5.213375091552734,
      "learning_rate": 1.9327772406220808e-07,
      "loss": 0.815,
      "step": 52440
    },
    {
      "epoch": 0.816474269656799,
      "grad_norm": 6.735192775726318,
      "learning_rate": 1.9319578505760313e-07,
      "loss": 0.7163,
      "step": 52445
    },
    {
      "epoch": 0.8165521106587684,
      "grad_norm": 6.488433361053467,
      "learning_rate": 1.9311384605299813e-07,
      "loss": 0.7691,
      "step": 52450
    },
    {
      "epoch": 0.8166299516607378,
      "grad_norm": 2.9705090522766113,
      "learning_rate": 1.9303190704839317e-07,
      "loss": 0.569,
      "step": 52455
    },
    {
      "epoch": 0.8167077926627071,
      "grad_norm": 4.922258377075195,
      "learning_rate": 1.929499680437882e-07,
      "loss": 0.7649,
      "step": 52460
    },
    {
      "epoch": 0.8167856336646765,
      "grad_norm": 3.0382869243621826,
      "learning_rate": 1.9286802903918322e-07,
      "loss": 0.7851,
      "step": 52465
    },
    {
      "epoch": 0.8168634746666459,
      "grad_norm": 5.332004547119141,
      "learning_rate": 1.9278609003457824e-07,
      "loss": 0.7767,
      "step": 52470
    },
    {
      "epoch": 0.8169413156686153,
      "grad_norm": 3.6386220455169678,
      "learning_rate": 1.927041510299733e-07,
      "loss": 0.6264,
      "step": 52475
    },
    {
      "epoch": 0.8170191566705847,
      "grad_norm": 6.401283264160156,
      "learning_rate": 1.926222120253683e-07,
      "loss": 0.7948,
      "step": 52480
    },
    {
      "epoch": 0.8170969976725541,
      "grad_norm": 5.519702434539795,
      "learning_rate": 1.9254027302076333e-07,
      "loss": 0.7279,
      "step": 52485
    },
    {
      "epoch": 0.8171748386745235,
      "grad_norm": 3.2340781688690186,
      "learning_rate": 1.9245833401615835e-07,
      "loss": 0.7403,
      "step": 52490
    },
    {
      "epoch": 0.8172526796764928,
      "grad_norm": 3.1249566078186035,
      "learning_rate": 1.923763950115534e-07,
      "loss": 0.7675,
      "step": 52495
    },
    {
      "epoch": 0.8173305206784621,
      "grad_norm": 3.621028184890747,
      "learning_rate": 1.922944560069484e-07,
      "loss": 0.7661,
      "step": 52500
    },
    {
      "epoch": 0.8174083616804315,
      "grad_norm": 4.5405988693237305,
      "learning_rate": 1.9221251700234345e-07,
      "loss": 0.7365,
      "step": 52505
    },
    {
      "epoch": 0.8174862026824009,
      "grad_norm": 3.17675518989563,
      "learning_rate": 1.9213057799773847e-07,
      "loss": 0.7723,
      "step": 52510
    },
    {
      "epoch": 0.8175640436843703,
      "grad_norm": 6.033045768737793,
      "learning_rate": 1.9204863899313352e-07,
      "loss": 0.8009,
      "step": 52515
    },
    {
      "epoch": 0.8176418846863397,
      "grad_norm": 6.609638690948486,
      "learning_rate": 1.919666999885285e-07,
      "loss": 0.6599,
      "step": 52520
    },
    {
      "epoch": 0.8177197256883091,
      "grad_norm": 3.5679421424865723,
      "learning_rate": 1.9188476098392356e-07,
      "loss": 0.7161,
      "step": 52525
    },
    {
      "epoch": 0.8177975666902785,
      "grad_norm": 2.8557729721069336,
      "learning_rate": 1.9180282197931858e-07,
      "loss": 0.7568,
      "step": 52530
    },
    {
      "epoch": 0.8178754076922479,
      "grad_norm": 4.037822246551514,
      "learning_rate": 1.917208829747136e-07,
      "loss": 0.6697,
      "step": 52535
    },
    {
      "epoch": 0.8179532486942171,
      "grad_norm": 2.998384475708008,
      "learning_rate": 1.9163894397010863e-07,
      "loss": 0.6062,
      "step": 52540
    },
    {
      "epoch": 0.8180310896961865,
      "grad_norm": 4.6753997802734375,
      "learning_rate": 1.9155700496550367e-07,
      "loss": 0.7585,
      "step": 52545
    },
    {
      "epoch": 0.8181089306981559,
      "grad_norm": 4.046789169311523,
      "learning_rate": 1.914750659608987e-07,
      "loss": 0.7155,
      "step": 52550
    },
    {
      "epoch": 0.8181867717001253,
      "grad_norm": 2.7731034755706787,
      "learning_rate": 1.9139312695629372e-07,
      "loss": 0.7409,
      "step": 52555
    },
    {
      "epoch": 0.8182646127020947,
      "grad_norm": 3.988541841506958,
      "learning_rate": 1.9131118795168874e-07,
      "loss": 0.7902,
      "step": 52560
    },
    {
      "epoch": 0.8183424537040641,
      "grad_norm": 3.3796608448028564,
      "learning_rate": 1.912292489470838e-07,
      "loss": 0.9207,
      "step": 52565
    },
    {
      "epoch": 0.8184202947060335,
      "grad_norm": 3.52365779876709,
      "learning_rate": 1.9114730994247884e-07,
      "loss": 0.7295,
      "step": 52570
    },
    {
      "epoch": 0.8184981357080028,
      "grad_norm": 2.773773670196533,
      "learning_rate": 1.9106537093787383e-07,
      "loss": 0.7501,
      "step": 52575
    },
    {
      "epoch": 0.8185759767099722,
      "grad_norm": 4.223803520202637,
      "learning_rate": 1.9098343193326888e-07,
      "loss": 0.668,
      "step": 52580
    },
    {
      "epoch": 0.8186538177119416,
      "grad_norm": 5.90623664855957,
      "learning_rate": 1.909014929286639e-07,
      "loss": 0.6992,
      "step": 52585
    },
    {
      "epoch": 0.818731658713911,
      "grad_norm": 5.672672271728516,
      "learning_rate": 1.9081955392405892e-07,
      "loss": 0.8155,
      "step": 52590
    },
    {
      "epoch": 0.8188094997158804,
      "grad_norm": 3.8862524032592773,
      "learning_rate": 1.9073761491945395e-07,
      "loss": 0.7405,
      "step": 52595
    },
    {
      "epoch": 0.8188873407178497,
      "grad_norm": 3.685518741607666,
      "learning_rate": 1.90655675914849e-07,
      "loss": 0.7896,
      "step": 52600
    },
    {
      "epoch": 0.8189651817198191,
      "grad_norm": 4.049343585968018,
      "learning_rate": 1.9057373691024402e-07,
      "loss": 0.7253,
      "step": 52605
    },
    {
      "epoch": 0.8190430227217885,
      "grad_norm": 5.41522741317749,
      "learning_rate": 1.9049179790563904e-07,
      "loss": 0.7705,
      "step": 52610
    },
    {
      "epoch": 0.8191208637237578,
      "grad_norm": 4.693664073944092,
      "learning_rate": 1.9040985890103406e-07,
      "loss": 0.6598,
      "step": 52615
    },
    {
      "epoch": 0.8191987047257272,
      "grad_norm": 5.3432297706604,
      "learning_rate": 1.903279198964291e-07,
      "loss": 0.7274,
      "step": 52620
    },
    {
      "epoch": 0.8192765457276966,
      "grad_norm": 3.7844955921173096,
      "learning_rate": 1.902459808918241e-07,
      "loss": 0.7712,
      "step": 52625
    },
    {
      "epoch": 0.819354386729666,
      "grad_norm": 4.020893096923828,
      "learning_rate": 1.9016404188721915e-07,
      "loss": 0.6731,
      "step": 52630
    },
    {
      "epoch": 0.8194322277316354,
      "grad_norm": 4.31149959564209,
      "learning_rate": 1.9008210288261417e-07,
      "loss": 0.6434,
      "step": 52635
    },
    {
      "epoch": 0.8195100687336048,
      "grad_norm": 3.167632818222046,
      "learning_rate": 1.9000016387800922e-07,
      "loss": 0.7416,
      "step": 52640
    },
    {
      "epoch": 0.8195879097355742,
      "grad_norm": 3.084571361541748,
      "learning_rate": 1.8991822487340422e-07,
      "loss": 0.7685,
      "step": 52645
    },
    {
      "epoch": 0.8196657507375434,
      "grad_norm": 5.368929862976074,
      "learning_rate": 1.8983628586879927e-07,
      "loss": 0.7406,
      "step": 52650
    },
    {
      "epoch": 0.8197435917395128,
      "grad_norm": 4.05276346206665,
      "learning_rate": 1.897543468641943e-07,
      "loss": 0.8213,
      "step": 52655
    },
    {
      "epoch": 0.8198214327414822,
      "grad_norm": 3.309346914291382,
      "learning_rate": 1.896724078595893e-07,
      "loss": 0.6572,
      "step": 52660
    },
    {
      "epoch": 0.8198992737434516,
      "grad_norm": 3.132782459259033,
      "learning_rate": 1.8959046885498433e-07,
      "loss": 0.7379,
      "step": 52665
    },
    {
      "epoch": 0.819977114745421,
      "grad_norm": 4.422957420349121,
      "learning_rate": 1.8950852985037938e-07,
      "loss": 0.7039,
      "step": 52670
    },
    {
      "epoch": 0.8200549557473904,
      "grad_norm": 6.01898717880249,
      "learning_rate": 1.894265908457744e-07,
      "loss": 0.7407,
      "step": 52675
    },
    {
      "epoch": 0.8201327967493598,
      "grad_norm": 6.295933723449707,
      "learning_rate": 1.8934465184116942e-07,
      "loss": 0.7581,
      "step": 52680
    },
    {
      "epoch": 0.8202106377513292,
      "grad_norm": 4.258965969085693,
      "learning_rate": 1.8926271283656445e-07,
      "loss": 0.7425,
      "step": 52685
    },
    {
      "epoch": 0.8202884787532985,
      "grad_norm": 3.7884628772735596,
      "learning_rate": 1.891807738319595e-07,
      "loss": 0.7336,
      "step": 52690
    },
    {
      "epoch": 0.8203663197552679,
      "grad_norm": 5.805574417114258,
      "learning_rate": 1.890988348273545e-07,
      "loss": 0.7739,
      "step": 52695
    },
    {
      "epoch": 0.8204441607572373,
      "grad_norm": 3.1947410106658936,
      "learning_rate": 1.8901689582274954e-07,
      "loss": 0.8312,
      "step": 52700
    },
    {
      "epoch": 0.8205220017592066,
      "grad_norm": 2.980165481567383,
      "learning_rate": 1.8893495681814456e-07,
      "loss": 0.6926,
      "step": 52705
    },
    {
      "epoch": 0.820599842761176,
      "grad_norm": 3.4354684352874756,
      "learning_rate": 1.888530178135396e-07,
      "loss": 0.6868,
      "step": 52710
    },
    {
      "epoch": 0.8206776837631454,
      "grad_norm": 3.57425856590271,
      "learning_rate": 1.887710788089346e-07,
      "loss": 0.7762,
      "step": 52715
    },
    {
      "epoch": 0.8207555247651148,
      "grad_norm": 3.742769241333008,
      "learning_rate": 1.8868913980432965e-07,
      "loss": 0.7209,
      "step": 52720
    },
    {
      "epoch": 0.8208333657670841,
      "grad_norm": 3.4661710262298584,
      "learning_rate": 1.8860720079972468e-07,
      "loss": 0.7213,
      "step": 52725
    },
    {
      "epoch": 0.8209112067690535,
      "grad_norm": 4.107504367828369,
      "learning_rate": 1.8852526179511972e-07,
      "loss": 0.6812,
      "step": 52730
    },
    {
      "epoch": 0.8209890477710229,
      "grad_norm": 8.202617645263672,
      "learning_rate": 1.8844332279051472e-07,
      "loss": 0.7202,
      "step": 52735
    },
    {
      "epoch": 0.8210668887729923,
      "grad_norm": 8.054469108581543,
      "learning_rate": 1.8836138378590977e-07,
      "loss": 0.7126,
      "step": 52740
    },
    {
      "epoch": 0.8211447297749617,
      "grad_norm": 2.9791061878204346,
      "learning_rate": 1.882794447813048e-07,
      "loss": 0.6819,
      "step": 52745
    },
    {
      "epoch": 0.8212225707769311,
      "grad_norm": 3.2550740242004395,
      "learning_rate": 1.881975057766998e-07,
      "loss": 0.7569,
      "step": 52750
    },
    {
      "epoch": 0.8213004117789005,
      "grad_norm": 3.3471927642822266,
      "learning_rate": 1.8811556677209483e-07,
      "loss": 0.7668,
      "step": 52755
    },
    {
      "epoch": 0.8213782527808698,
      "grad_norm": 3.070326089859009,
      "learning_rate": 1.8803362776748988e-07,
      "loss": 0.606,
      "step": 52760
    },
    {
      "epoch": 0.8214560937828391,
      "grad_norm": 3.9186534881591797,
      "learning_rate": 1.879516887628849e-07,
      "loss": 0.7587,
      "step": 52765
    },
    {
      "epoch": 0.8215339347848085,
      "grad_norm": 3.200526475906372,
      "learning_rate": 1.8786974975827993e-07,
      "loss": 0.765,
      "step": 52770
    },
    {
      "epoch": 0.8216117757867779,
      "grad_norm": 2.6812970638275146,
      "learning_rate": 1.8778781075367495e-07,
      "loss": 0.5981,
      "step": 52775
    },
    {
      "epoch": 0.8216896167887473,
      "grad_norm": 3.28224515914917,
      "learning_rate": 1.8770587174907e-07,
      "loss": 0.6946,
      "step": 52780
    },
    {
      "epoch": 0.8217674577907167,
      "grad_norm": 4.24637508392334,
      "learning_rate": 1.87623932744465e-07,
      "loss": 0.74,
      "step": 52785
    },
    {
      "epoch": 0.8218452987926861,
      "grad_norm": 5.24114465713501,
      "learning_rate": 1.8754199373986004e-07,
      "loss": 0.6526,
      "step": 52790
    },
    {
      "epoch": 0.8219231397946555,
      "grad_norm": 3.0430965423583984,
      "learning_rate": 1.8746005473525506e-07,
      "loss": 0.7315,
      "step": 52795
    },
    {
      "epoch": 0.8220009807966249,
      "grad_norm": 3.010605812072754,
      "learning_rate": 1.873781157306501e-07,
      "loss": 0.8276,
      "step": 52800
    },
    {
      "epoch": 0.8220788217985942,
      "grad_norm": 3.549471616744995,
      "learning_rate": 1.872961767260451e-07,
      "loss": 0.7306,
      "step": 52805
    },
    {
      "epoch": 0.8221566628005635,
      "grad_norm": 4.324155807495117,
      "learning_rate": 1.8721423772144015e-07,
      "loss": 0.8575,
      "step": 52810
    },
    {
      "epoch": 0.8222345038025329,
      "grad_norm": 3.719729423522949,
      "learning_rate": 1.8713229871683518e-07,
      "loss": 0.7146,
      "step": 52815
    },
    {
      "epoch": 0.8223123448045023,
      "grad_norm": 4.08032751083374,
      "learning_rate": 1.870503597122302e-07,
      "loss": 0.7075,
      "step": 52820
    },
    {
      "epoch": 0.8223901858064717,
      "grad_norm": 3.394622802734375,
      "learning_rate": 1.8696842070762522e-07,
      "loss": 0.6869,
      "step": 52825
    },
    {
      "epoch": 0.8224680268084411,
      "grad_norm": 3.3150534629821777,
      "learning_rate": 1.8688648170302027e-07,
      "loss": 0.8847,
      "step": 52830
    },
    {
      "epoch": 0.8225458678104105,
      "grad_norm": 5.626073360443115,
      "learning_rate": 1.868045426984153e-07,
      "loss": 0.7583,
      "step": 52835
    },
    {
      "epoch": 0.8226237088123798,
      "grad_norm": 3.384359836578369,
      "learning_rate": 1.867226036938103e-07,
      "loss": 0.7006,
      "step": 52840
    },
    {
      "epoch": 0.8227015498143492,
      "grad_norm": 4.348812580108643,
      "learning_rate": 1.8664066468920533e-07,
      "loss": 0.7334,
      "step": 52845
    },
    {
      "epoch": 0.8227793908163186,
      "grad_norm": 3.8175435066223145,
      "learning_rate": 1.8655872568460038e-07,
      "loss": 0.6423,
      "step": 52850
    },
    {
      "epoch": 0.822857231818288,
      "grad_norm": 3.7770535945892334,
      "learning_rate": 1.8647678667999538e-07,
      "loss": 0.7383,
      "step": 52855
    },
    {
      "epoch": 0.8229350728202574,
      "grad_norm": 3.642742872238159,
      "learning_rate": 1.8639484767539043e-07,
      "loss": 0.6864,
      "step": 52860
    },
    {
      "epoch": 0.8230129138222267,
      "grad_norm": 2.8787238597869873,
      "learning_rate": 1.8631290867078545e-07,
      "loss": 0.6871,
      "step": 52865
    },
    {
      "epoch": 0.8230907548241961,
      "grad_norm": 3.260321855545044,
      "learning_rate": 1.862309696661805e-07,
      "loss": 0.6727,
      "step": 52870
    },
    {
      "epoch": 0.8231685958261655,
      "grad_norm": 3.672762155532837,
      "learning_rate": 1.861490306615755e-07,
      "loss": 0.7855,
      "step": 52875
    },
    {
      "epoch": 0.8232464368281348,
      "grad_norm": 3.764991521835327,
      "learning_rate": 1.8606709165697054e-07,
      "loss": 0.6751,
      "step": 52880
    },
    {
      "epoch": 0.8233242778301042,
      "grad_norm": 5.106266975402832,
      "learning_rate": 1.859851526523656e-07,
      "loss": 0.765,
      "step": 52885
    },
    {
      "epoch": 0.8234021188320736,
      "grad_norm": 3.2545130252838135,
      "learning_rate": 1.859032136477606e-07,
      "loss": 0.6327,
      "step": 52890
    },
    {
      "epoch": 0.823479959834043,
      "grad_norm": 3.774287462234497,
      "learning_rate": 1.8582127464315563e-07,
      "loss": 0.8104,
      "step": 52895
    },
    {
      "epoch": 0.8235578008360124,
      "grad_norm": 5.744304180145264,
      "learning_rate": 1.8573933563855065e-07,
      "loss": 0.7437,
      "step": 52900
    },
    {
      "epoch": 0.8236356418379818,
      "grad_norm": 3.160486936569214,
      "learning_rate": 1.856573966339457e-07,
      "loss": 0.6492,
      "step": 52905
    },
    {
      "epoch": 0.8237134828399512,
      "grad_norm": 7.700464248657227,
      "learning_rate": 1.855754576293407e-07,
      "loss": 0.7449,
      "step": 52910
    },
    {
      "epoch": 0.8237913238419204,
      "grad_norm": 2.96140718460083,
      "learning_rate": 1.8549351862473575e-07,
      "loss": 0.7022,
      "step": 52915
    },
    {
      "epoch": 0.8238691648438898,
      "grad_norm": 3.166518211364746,
      "learning_rate": 1.8541157962013077e-07,
      "loss": 0.9095,
      "step": 52920
    },
    {
      "epoch": 0.8239470058458592,
      "grad_norm": 11.438024520874023,
      "learning_rate": 1.8532964061552582e-07,
      "loss": 0.7185,
      "step": 52925
    },
    {
      "epoch": 0.8240248468478286,
      "grad_norm": 3.6097190380096436,
      "learning_rate": 1.852477016109208e-07,
      "loss": 0.7352,
      "step": 52930
    },
    {
      "epoch": 0.824102687849798,
      "grad_norm": 3.1926488876342773,
      "learning_rate": 1.8516576260631586e-07,
      "loss": 0.7675,
      "step": 52935
    },
    {
      "epoch": 0.8241805288517674,
      "grad_norm": 3.3030455112457275,
      "learning_rate": 1.8508382360171088e-07,
      "loss": 0.7996,
      "step": 52940
    },
    {
      "epoch": 0.8242583698537368,
      "grad_norm": 3.920374870300293,
      "learning_rate": 1.850018845971059e-07,
      "loss": 0.7545,
      "step": 52945
    },
    {
      "epoch": 0.8243362108557062,
      "grad_norm": 2.4193532466888428,
      "learning_rate": 1.8491994559250093e-07,
      "loss": 0.6863,
      "step": 52950
    },
    {
      "epoch": 0.8244140518576755,
      "grad_norm": 5.6541972160339355,
      "learning_rate": 1.8483800658789597e-07,
      "loss": 0.7362,
      "step": 52955
    },
    {
      "epoch": 0.8244918928596449,
      "grad_norm": 3.0625786781311035,
      "learning_rate": 1.84756067583291e-07,
      "loss": 0.7754,
      "step": 52960
    },
    {
      "epoch": 0.8245697338616143,
      "grad_norm": 2.762301445007324,
      "learning_rate": 1.8467412857868602e-07,
      "loss": 0.7126,
      "step": 52965
    },
    {
      "epoch": 0.8246475748635836,
      "grad_norm": 3.6609842777252197,
      "learning_rate": 1.8459218957408104e-07,
      "loss": 0.7612,
      "step": 52970
    },
    {
      "epoch": 0.824725415865553,
      "grad_norm": 3.514352321624756,
      "learning_rate": 1.845102505694761e-07,
      "loss": 0.8039,
      "step": 52975
    },
    {
      "epoch": 0.8248032568675224,
      "grad_norm": 2.8552377223968506,
      "learning_rate": 1.844283115648711e-07,
      "loss": 0.7269,
      "step": 52980
    },
    {
      "epoch": 0.8248810978694918,
      "grad_norm": 10.76278018951416,
      "learning_rate": 1.8434637256026613e-07,
      "loss": 0.7054,
      "step": 52985
    },
    {
      "epoch": 0.8249589388714611,
      "grad_norm": 3.4718878269195557,
      "learning_rate": 1.8426443355566115e-07,
      "loss": 0.6688,
      "step": 52990
    },
    {
      "epoch": 0.8250367798734305,
      "grad_norm": 2.8028817176818848,
      "learning_rate": 1.841824945510562e-07,
      "loss": 0.684,
      "step": 52995
    },
    {
      "epoch": 0.8251146208753999,
      "grad_norm": 4.734134674072266,
      "learning_rate": 1.841005555464512e-07,
      "loss": 0.785,
      "step": 53000
    },
    {
      "epoch": 0.8251924618773693,
      "grad_norm": 3.7249114513397217,
      "learning_rate": 1.8401861654184625e-07,
      "loss": 0.7012,
      "step": 53005
    },
    {
      "epoch": 0.8252703028793387,
      "grad_norm": 2.417930841445923,
      "learning_rate": 1.8393667753724127e-07,
      "loss": 0.6351,
      "step": 53010
    },
    {
      "epoch": 0.8253481438813081,
      "grad_norm": 3.044121026992798,
      "learning_rate": 1.8385473853263632e-07,
      "loss": 0.766,
      "step": 53015
    },
    {
      "epoch": 0.8254259848832775,
      "grad_norm": 3.073611259460449,
      "learning_rate": 1.837727995280313e-07,
      "loss": 0.6988,
      "step": 53020
    },
    {
      "epoch": 0.8255038258852468,
      "grad_norm": 2.4141581058502197,
      "learning_rate": 1.8369086052342636e-07,
      "loss": 0.6438,
      "step": 53025
    },
    {
      "epoch": 0.8255816668872161,
      "grad_norm": 4.605281829833984,
      "learning_rate": 1.8360892151882138e-07,
      "loss": 0.7086,
      "step": 53030
    },
    {
      "epoch": 0.8256595078891855,
      "grad_norm": 4.06589937210083,
      "learning_rate": 1.835269825142164e-07,
      "loss": 0.8241,
      "step": 53035
    },
    {
      "epoch": 0.8257373488911549,
      "grad_norm": 4.736262798309326,
      "learning_rate": 1.8344504350961143e-07,
      "loss": 0.677,
      "step": 53040
    },
    {
      "epoch": 0.8258151898931243,
      "grad_norm": 3.6541924476623535,
      "learning_rate": 1.8336310450500648e-07,
      "loss": 0.7223,
      "step": 53045
    },
    {
      "epoch": 0.8258930308950937,
      "grad_norm": 4.166376113891602,
      "learning_rate": 1.832811655004015e-07,
      "loss": 0.7262,
      "step": 53050
    },
    {
      "epoch": 0.8259708718970631,
      "grad_norm": 2.9411706924438477,
      "learning_rate": 1.8319922649579652e-07,
      "loss": 0.7756,
      "step": 53055
    },
    {
      "epoch": 0.8260487128990325,
      "grad_norm": 7.837759494781494,
      "learning_rate": 1.8311728749119154e-07,
      "loss": 0.8287,
      "step": 53060
    },
    {
      "epoch": 0.8261265539010018,
      "grad_norm": 4.58543586730957,
      "learning_rate": 1.830353484865866e-07,
      "loss": 0.752,
      "step": 53065
    },
    {
      "epoch": 0.8262043949029712,
      "grad_norm": 3.5503833293914795,
      "learning_rate": 1.8295340948198158e-07,
      "loss": 0.724,
      "step": 53070
    },
    {
      "epoch": 0.8262822359049405,
      "grad_norm": 4.567082405090332,
      "learning_rate": 1.8287147047737663e-07,
      "loss": 0.6938,
      "step": 53075
    },
    {
      "epoch": 0.8263600769069099,
      "grad_norm": 4.324936389923096,
      "learning_rate": 1.8278953147277165e-07,
      "loss": 0.7609,
      "step": 53080
    },
    {
      "epoch": 0.8264379179088793,
      "grad_norm": 5.083077430725098,
      "learning_rate": 1.827075924681667e-07,
      "loss": 0.7083,
      "step": 53085
    },
    {
      "epoch": 0.8265157589108487,
      "grad_norm": 3.733135223388672,
      "learning_rate": 1.826256534635617e-07,
      "loss": 0.7326,
      "step": 53090
    },
    {
      "epoch": 0.8265935999128181,
      "grad_norm": 3.6276872158050537,
      "learning_rate": 1.8254371445895675e-07,
      "loss": 0.7772,
      "step": 53095
    },
    {
      "epoch": 0.8266714409147875,
      "grad_norm": 3.6290223598480225,
      "learning_rate": 1.8246177545435177e-07,
      "loss": 0.7475,
      "step": 53100
    },
    {
      "epoch": 0.8267492819167568,
      "grad_norm": 2.9519970417022705,
      "learning_rate": 1.8237983644974682e-07,
      "loss": 0.7799,
      "step": 53105
    },
    {
      "epoch": 0.8268271229187262,
      "grad_norm": 5.221888542175293,
      "learning_rate": 1.822978974451418e-07,
      "loss": 0.6587,
      "step": 53110
    },
    {
      "epoch": 0.8269049639206956,
      "grad_norm": 4.268246173858643,
      "learning_rate": 1.8221595844053686e-07,
      "loss": 0.734,
      "step": 53115
    },
    {
      "epoch": 0.826982804922665,
      "grad_norm": 5.316501617431641,
      "learning_rate": 1.8213401943593188e-07,
      "loss": 0.7131,
      "step": 53120
    },
    {
      "epoch": 0.8270606459246344,
      "grad_norm": 4.57843017578125,
      "learning_rate": 1.820520804313269e-07,
      "loss": 0.8263,
      "step": 53125
    },
    {
      "epoch": 0.8271384869266037,
      "grad_norm": 7.864269733428955,
      "learning_rate": 1.8197014142672193e-07,
      "loss": 0.6956,
      "step": 53130
    },
    {
      "epoch": 0.8272163279285731,
      "grad_norm": 5.749431133270264,
      "learning_rate": 1.8188820242211698e-07,
      "loss": 0.7811,
      "step": 53135
    },
    {
      "epoch": 0.8272941689305425,
      "grad_norm": 2.8069663047790527,
      "learning_rate": 1.81806263417512e-07,
      "loss": 0.6181,
      "step": 53140
    },
    {
      "epoch": 0.8273720099325118,
      "grad_norm": 3.510007381439209,
      "learning_rate": 1.8172432441290702e-07,
      "loss": 0.7425,
      "step": 53145
    },
    {
      "epoch": 0.8274498509344812,
      "grad_norm": 3.182915210723877,
      "learning_rate": 1.8164238540830204e-07,
      "loss": 0.7529,
      "step": 53150
    },
    {
      "epoch": 0.8275276919364506,
      "grad_norm": 8.018515586853027,
      "learning_rate": 1.815604464036971e-07,
      "loss": 0.9375,
      "step": 53155
    },
    {
      "epoch": 0.82760553293842,
      "grad_norm": 3.9722671508789062,
      "learning_rate": 1.8147850739909209e-07,
      "loss": 0.812,
      "step": 53160
    },
    {
      "epoch": 0.8276833739403894,
      "grad_norm": 3.782501697540283,
      "learning_rate": 1.8139656839448713e-07,
      "loss": 0.7501,
      "step": 53165
    },
    {
      "epoch": 0.8277612149423588,
      "grad_norm": 4.626613616943359,
      "learning_rate": 1.8131462938988216e-07,
      "loss": 0.7118,
      "step": 53170
    },
    {
      "epoch": 0.8278390559443282,
      "grad_norm": 4.549905300140381,
      "learning_rate": 1.812326903852772e-07,
      "loss": 0.7239,
      "step": 53175
    },
    {
      "epoch": 0.8279168969462974,
      "grad_norm": 5.36144495010376,
      "learning_rate": 1.811507513806722e-07,
      "loss": 0.8575,
      "step": 53180
    },
    {
      "epoch": 0.8279947379482668,
      "grad_norm": 3.889080286026001,
      "learning_rate": 1.8106881237606725e-07,
      "loss": 0.6497,
      "step": 53185
    },
    {
      "epoch": 0.8280725789502362,
      "grad_norm": 3.719517946243286,
      "learning_rate": 1.8098687337146227e-07,
      "loss": 0.7218,
      "step": 53190
    },
    {
      "epoch": 0.8281504199522056,
      "grad_norm": 4.781241416931152,
      "learning_rate": 1.809049343668573e-07,
      "loss": 0.7693,
      "step": 53195
    },
    {
      "epoch": 0.828228260954175,
      "grad_norm": 4.195295810699463,
      "learning_rate": 1.8082299536225231e-07,
      "loss": 0.7882,
      "step": 53200
    },
    {
      "epoch": 0.8283061019561444,
      "grad_norm": 3.2362470626831055,
      "learning_rate": 1.8074105635764736e-07,
      "loss": 0.7539,
      "step": 53205
    },
    {
      "epoch": 0.8283839429581138,
      "grad_norm": 3.200885772705078,
      "learning_rate": 1.806591173530424e-07,
      "loss": 0.7045,
      "step": 53210
    },
    {
      "epoch": 0.8284617839600832,
      "grad_norm": 4.3491129875183105,
      "learning_rate": 1.805771783484374e-07,
      "loss": 0.6972,
      "step": 53215
    },
    {
      "epoch": 0.8285396249620525,
      "grad_norm": 3.919623851776123,
      "learning_rate": 1.8049523934383245e-07,
      "loss": 0.7669,
      "step": 53220
    },
    {
      "epoch": 0.8286174659640219,
      "grad_norm": 3.0189907550811768,
      "learning_rate": 1.8041330033922748e-07,
      "loss": 0.7289,
      "step": 53225
    },
    {
      "epoch": 0.8286953069659913,
      "grad_norm": 2.670377016067505,
      "learning_rate": 1.8033136133462252e-07,
      "loss": 0.7237,
      "step": 53230
    },
    {
      "epoch": 0.8287731479679606,
      "grad_norm": 3.3646020889282227,
      "learning_rate": 1.8024942233001752e-07,
      "loss": 0.64,
      "step": 53235
    },
    {
      "epoch": 0.82885098896993,
      "grad_norm": 3.349203109741211,
      "learning_rate": 1.8016748332541257e-07,
      "loss": 0.6932,
      "step": 53240
    },
    {
      "epoch": 0.8289288299718994,
      "grad_norm": 3.4410057067871094,
      "learning_rate": 1.800855443208076e-07,
      "loss": 0.6801,
      "step": 53245
    },
    {
      "epoch": 0.8290066709738688,
      "grad_norm": 3.4300858974456787,
      "learning_rate": 1.800036053162026e-07,
      "loss": 0.8224,
      "step": 53250
    },
    {
      "epoch": 0.8290845119758381,
      "grad_norm": 4.3226847648620605,
      "learning_rate": 1.7992166631159763e-07,
      "loss": 0.7738,
      "step": 53255
    },
    {
      "epoch": 0.8291623529778075,
      "grad_norm": 2.920203447341919,
      "learning_rate": 1.7983972730699268e-07,
      "loss": 0.7359,
      "step": 53260
    },
    {
      "epoch": 0.8292401939797769,
      "grad_norm": 4.190410614013672,
      "learning_rate": 1.797577883023877e-07,
      "loss": 0.7277,
      "step": 53265
    },
    {
      "epoch": 0.8293180349817463,
      "grad_norm": 2.9433329105377197,
      "learning_rate": 1.7967584929778273e-07,
      "loss": 0.73,
      "step": 53270
    },
    {
      "epoch": 0.8293958759837157,
      "grad_norm": 3.393692970275879,
      "learning_rate": 1.7959391029317775e-07,
      "loss": 0.6672,
      "step": 53275
    },
    {
      "epoch": 0.8294737169856851,
      "grad_norm": 6.385544300079346,
      "learning_rate": 1.795119712885728e-07,
      "loss": 0.7116,
      "step": 53280
    },
    {
      "epoch": 0.8295515579876545,
      "grad_norm": 3.5230276584625244,
      "learning_rate": 1.794300322839678e-07,
      "loss": 0.7814,
      "step": 53285
    },
    {
      "epoch": 0.8296293989896238,
      "grad_norm": 3.9629416465759277,
      "learning_rate": 1.7934809327936284e-07,
      "loss": 0.742,
      "step": 53290
    },
    {
      "epoch": 0.8297072399915931,
      "grad_norm": 3.5776171684265137,
      "learning_rate": 1.7926615427475786e-07,
      "loss": 0.6579,
      "step": 53295
    },
    {
      "epoch": 0.8297850809935625,
      "grad_norm": 2.95524263381958,
      "learning_rate": 1.791842152701529e-07,
      "loss": 0.7979,
      "step": 53300
    },
    {
      "epoch": 0.8298629219955319,
      "grad_norm": 2.6189610958099365,
      "learning_rate": 1.791022762655479e-07,
      "loss": 0.6953,
      "step": 53305
    },
    {
      "epoch": 0.8299407629975013,
      "grad_norm": 4.316525936126709,
      "learning_rate": 1.7902033726094295e-07,
      "loss": 0.7738,
      "step": 53310
    },
    {
      "epoch": 0.8300186039994707,
      "grad_norm": 5.372035980224609,
      "learning_rate": 1.7893839825633798e-07,
      "loss": 0.6645,
      "step": 53315
    },
    {
      "epoch": 0.8300964450014401,
      "grad_norm": 3.205012559890747,
      "learning_rate": 1.78856459251733e-07,
      "loss": 0.8239,
      "step": 53320
    },
    {
      "epoch": 0.8301742860034095,
      "grad_norm": 2.821965217590332,
      "learning_rate": 1.7877452024712802e-07,
      "loss": 0.6411,
      "step": 53325
    },
    {
      "epoch": 0.8302521270053788,
      "grad_norm": 3.06540584564209,
      "learning_rate": 1.7869258124252307e-07,
      "loss": 0.738,
      "step": 53330
    },
    {
      "epoch": 0.8303299680073482,
      "grad_norm": 2.9707372188568115,
      "learning_rate": 1.786106422379181e-07,
      "loss": 0.7189,
      "step": 53335
    },
    {
      "epoch": 0.8304078090093175,
      "grad_norm": 6.991352081298828,
      "learning_rate": 1.785287032333131e-07,
      "loss": 0.7596,
      "step": 53340
    },
    {
      "epoch": 0.8304856500112869,
      "grad_norm": 3.448251247406006,
      "learning_rate": 1.7844676422870813e-07,
      "loss": 0.7634,
      "step": 53345
    },
    {
      "epoch": 0.8305634910132563,
      "grad_norm": 4.381549835205078,
      "learning_rate": 1.7836482522410318e-07,
      "loss": 0.8144,
      "step": 53350
    },
    {
      "epoch": 0.8306413320152257,
      "grad_norm": 4.294335842132568,
      "learning_rate": 1.7828288621949818e-07,
      "loss": 0.7278,
      "step": 53355
    },
    {
      "epoch": 0.8307191730171951,
      "grad_norm": 3.2765400409698486,
      "learning_rate": 1.7820094721489323e-07,
      "loss": 0.7551,
      "step": 53360
    },
    {
      "epoch": 0.8307970140191645,
      "grad_norm": 2.5278005599975586,
      "learning_rate": 1.7811900821028825e-07,
      "loss": 0.739,
      "step": 53365
    },
    {
      "epoch": 0.8308748550211338,
      "grad_norm": 3.301809549331665,
      "learning_rate": 1.780370692056833e-07,
      "loss": 0.6658,
      "step": 53370
    },
    {
      "epoch": 0.8309526960231032,
      "grad_norm": 4.541007041931152,
      "learning_rate": 1.779551302010783e-07,
      "loss": 0.6408,
      "step": 53375
    },
    {
      "epoch": 0.8310305370250726,
      "grad_norm": 3.616001605987549,
      "learning_rate": 1.7787319119647334e-07,
      "loss": 0.6794,
      "step": 53380
    },
    {
      "epoch": 0.831108378027042,
      "grad_norm": 4.942752838134766,
      "learning_rate": 1.7779125219186836e-07,
      "loss": 0.7171,
      "step": 53385
    },
    {
      "epoch": 0.8311862190290114,
      "grad_norm": 4.380762577056885,
      "learning_rate": 1.777093131872634e-07,
      "loss": 0.6946,
      "step": 53390
    },
    {
      "epoch": 0.8312640600309807,
      "grad_norm": 3.2007837295532227,
      "learning_rate": 1.776273741826584e-07,
      "loss": 0.7978,
      "step": 53395
    },
    {
      "epoch": 0.8313419010329501,
      "grad_norm": 7.72484827041626,
      "learning_rate": 1.7754543517805345e-07,
      "loss": 0.7378,
      "step": 53400
    },
    {
      "epoch": 0.8314197420349195,
      "grad_norm": 8.381691932678223,
      "learning_rate": 1.7746349617344848e-07,
      "loss": 0.7577,
      "step": 53405
    },
    {
      "epoch": 0.8314975830368888,
      "grad_norm": 6.203024387359619,
      "learning_rate": 1.773815571688435e-07,
      "loss": 0.7942,
      "step": 53410
    },
    {
      "epoch": 0.8315754240388582,
      "grad_norm": 6.964951038360596,
      "learning_rate": 1.7729961816423852e-07,
      "loss": 0.8361,
      "step": 53415
    },
    {
      "epoch": 0.8316532650408276,
      "grad_norm": 2.9017953872680664,
      "learning_rate": 1.7721767915963357e-07,
      "loss": 0.6455,
      "step": 53420
    },
    {
      "epoch": 0.831731106042797,
      "grad_norm": 4.79205846786499,
      "learning_rate": 1.771357401550286e-07,
      "loss": 0.7218,
      "step": 53425
    },
    {
      "epoch": 0.8318089470447664,
      "grad_norm": 2.7590830326080322,
      "learning_rate": 1.770538011504236e-07,
      "loss": 0.7362,
      "step": 53430
    },
    {
      "epoch": 0.8318867880467358,
      "grad_norm": 8.644283294677734,
      "learning_rate": 1.7697186214581863e-07,
      "loss": 0.7798,
      "step": 53435
    },
    {
      "epoch": 0.8319646290487052,
      "grad_norm": 3.975590944290161,
      "learning_rate": 1.7688992314121368e-07,
      "loss": 0.6902,
      "step": 53440
    },
    {
      "epoch": 0.8320424700506744,
      "grad_norm": 13.37384033203125,
      "learning_rate": 1.7680798413660868e-07,
      "loss": 0.8178,
      "step": 53445
    },
    {
      "epoch": 0.8321203110526438,
      "grad_norm": 2.9015822410583496,
      "learning_rate": 1.7672604513200373e-07,
      "loss": 0.6373,
      "step": 53450
    },
    {
      "epoch": 0.8321981520546132,
      "grad_norm": 11.441720008850098,
      "learning_rate": 1.7664410612739875e-07,
      "loss": 0.7722,
      "step": 53455
    },
    {
      "epoch": 0.8322759930565826,
      "grad_norm": 3.6386003494262695,
      "learning_rate": 1.765621671227938e-07,
      "loss": 0.7141,
      "step": 53460
    },
    {
      "epoch": 0.832353834058552,
      "grad_norm": 3.033367395401001,
      "learning_rate": 1.764802281181888e-07,
      "loss": 0.6627,
      "step": 53465
    },
    {
      "epoch": 0.8324316750605214,
      "grad_norm": 3.076796293258667,
      "learning_rate": 1.7639828911358384e-07,
      "loss": 0.7602,
      "step": 53470
    },
    {
      "epoch": 0.8325095160624908,
      "grad_norm": 2.3928511142730713,
      "learning_rate": 1.7631635010897886e-07,
      "loss": 0.7504,
      "step": 53475
    },
    {
      "epoch": 0.8325873570644602,
      "grad_norm": 3.536756753921509,
      "learning_rate": 1.7623441110437388e-07,
      "loss": 0.73,
      "step": 53480
    },
    {
      "epoch": 0.8326651980664295,
      "grad_norm": 2.9666614532470703,
      "learning_rate": 1.761524720997689e-07,
      "loss": 0.6877,
      "step": 53485
    },
    {
      "epoch": 0.8327430390683989,
      "grad_norm": 3.1436526775360107,
      "learning_rate": 1.7607053309516396e-07,
      "loss": 0.6476,
      "step": 53490
    },
    {
      "epoch": 0.8328208800703683,
      "grad_norm": 3.24334454536438,
      "learning_rate": 1.7598859409055898e-07,
      "loss": 0.7406,
      "step": 53495
    },
    {
      "epoch": 0.8328987210723376,
      "grad_norm": 3.1539793014526367,
      "learning_rate": 1.75906655085954e-07,
      "loss": 0.756,
      "step": 53500
    },
    {
      "epoch": 0.832976562074307,
      "grad_norm": 3.929300308227539,
      "learning_rate": 1.7582471608134902e-07,
      "loss": 0.6669,
      "step": 53505
    },
    {
      "epoch": 0.8330544030762764,
      "grad_norm": 4.61472225189209,
      "learning_rate": 1.7574277707674407e-07,
      "loss": 0.8863,
      "step": 53510
    },
    {
      "epoch": 0.8331322440782458,
      "grad_norm": 4.408937931060791,
      "learning_rate": 1.756608380721391e-07,
      "loss": 0.734,
      "step": 53515
    },
    {
      "epoch": 0.8332100850802151,
      "grad_norm": 3.612572193145752,
      "learning_rate": 1.7557889906753411e-07,
      "loss": 0.5632,
      "step": 53520
    },
    {
      "epoch": 0.8332879260821845,
      "grad_norm": 3.766458749771118,
      "learning_rate": 1.7549696006292916e-07,
      "loss": 0.6437,
      "step": 53525
    },
    {
      "epoch": 0.8333657670841539,
      "grad_norm": 4.642184734344482,
      "learning_rate": 1.7541502105832418e-07,
      "loss": 0.7086,
      "step": 53530
    },
    {
      "epoch": 0.8334436080861233,
      "grad_norm": 5.802420616149902,
      "learning_rate": 1.753330820537192e-07,
      "loss": 0.7479,
      "step": 53535
    },
    {
      "epoch": 0.8335214490880927,
      "grad_norm": 2.5645253658294678,
      "learning_rate": 1.7525114304911423e-07,
      "loss": 0.681,
      "step": 53540
    },
    {
      "epoch": 0.8335992900900621,
      "grad_norm": 2.2696573734283447,
      "learning_rate": 1.7516920404450928e-07,
      "loss": 0.6607,
      "step": 53545
    },
    {
      "epoch": 0.8336771310920315,
      "grad_norm": 4.515111923217773,
      "learning_rate": 1.750872650399043e-07,
      "loss": 0.6546,
      "step": 53550
    },
    {
      "epoch": 0.8337549720940008,
      "grad_norm": 3.2549331188201904,
      "learning_rate": 1.7500532603529932e-07,
      "loss": 0.7289,
      "step": 53555
    },
    {
      "epoch": 0.8338328130959701,
      "grad_norm": 14.888633728027344,
      "learning_rate": 1.7492338703069434e-07,
      "loss": 0.6998,
      "step": 53560
    },
    {
      "epoch": 0.8339106540979395,
      "grad_norm": 3.938843250274658,
      "learning_rate": 1.748414480260894e-07,
      "loss": 0.8348,
      "step": 53565
    },
    {
      "epoch": 0.8339884950999089,
      "grad_norm": 3.511597156524658,
      "learning_rate": 1.7475950902148439e-07,
      "loss": 0.6784,
      "step": 53570
    },
    {
      "epoch": 0.8340663361018783,
      "grad_norm": 4.267786979675293,
      "learning_rate": 1.7467757001687943e-07,
      "loss": 0.7902,
      "step": 53575
    },
    {
      "epoch": 0.8341441771038477,
      "grad_norm": 4.834145545959473,
      "learning_rate": 1.7459563101227446e-07,
      "loss": 0.6849,
      "step": 53580
    },
    {
      "epoch": 0.8342220181058171,
      "grad_norm": 2.9110238552093506,
      "learning_rate": 1.745136920076695e-07,
      "loss": 0.7524,
      "step": 53585
    },
    {
      "epoch": 0.8342998591077865,
      "grad_norm": 3.4037725925445557,
      "learning_rate": 1.744317530030645e-07,
      "loss": 0.7648,
      "step": 53590
    },
    {
      "epoch": 0.8343777001097558,
      "grad_norm": 7.343257427215576,
      "learning_rate": 1.7434981399845955e-07,
      "loss": 0.7395,
      "step": 53595
    },
    {
      "epoch": 0.8344555411117252,
      "grad_norm": 4.509312152862549,
      "learning_rate": 1.7426787499385457e-07,
      "loss": 0.7423,
      "step": 53600
    },
    {
      "epoch": 0.8345333821136945,
      "grad_norm": 4.569262504577637,
      "learning_rate": 1.741859359892496e-07,
      "loss": 0.7474,
      "step": 53605
    },
    {
      "epoch": 0.8346112231156639,
      "grad_norm": 3.8337035179138184,
      "learning_rate": 1.7410399698464461e-07,
      "loss": 0.7439,
      "step": 53610
    },
    {
      "epoch": 0.8346890641176333,
      "grad_norm": 5.247994422912598,
      "learning_rate": 1.7402205798003966e-07,
      "loss": 0.6468,
      "step": 53615
    },
    {
      "epoch": 0.8347669051196027,
      "grad_norm": 10.763262748718262,
      "learning_rate": 1.7394011897543468e-07,
      "loss": 0.7674,
      "step": 53620
    },
    {
      "epoch": 0.8348447461215721,
      "grad_norm": 4.1517333984375,
      "learning_rate": 1.738581799708297e-07,
      "loss": 0.7455,
      "step": 53625
    },
    {
      "epoch": 0.8349225871235415,
      "grad_norm": 2.4810104370117188,
      "learning_rate": 1.7377624096622473e-07,
      "loss": 0.7176,
      "step": 53630
    },
    {
      "epoch": 0.8350004281255108,
      "grad_norm": 8.524070739746094,
      "learning_rate": 1.7369430196161978e-07,
      "loss": 0.7144,
      "step": 53635
    },
    {
      "epoch": 0.8350782691274802,
      "grad_norm": 7.514963150024414,
      "learning_rate": 1.736123629570148e-07,
      "loss": 0.7577,
      "step": 53640
    },
    {
      "epoch": 0.8351561101294496,
      "grad_norm": 4.215222358703613,
      "learning_rate": 1.7353042395240982e-07,
      "loss": 0.692,
      "step": 53645
    },
    {
      "epoch": 0.835233951131419,
      "grad_norm": 4.035255432128906,
      "learning_rate": 1.7344848494780484e-07,
      "loss": 0.7276,
      "step": 53650
    },
    {
      "epoch": 0.8353117921333884,
      "grad_norm": 2.51218318939209,
      "learning_rate": 1.733665459431999e-07,
      "loss": 0.6525,
      "step": 53655
    },
    {
      "epoch": 0.8353896331353577,
      "grad_norm": 3.128560781478882,
      "learning_rate": 1.7328460693859489e-07,
      "loss": 0.7739,
      "step": 53660
    },
    {
      "epoch": 0.8354674741373271,
      "grad_norm": 3.349975824356079,
      "learning_rate": 1.7320266793398993e-07,
      "loss": 0.6481,
      "step": 53665
    },
    {
      "epoch": 0.8355453151392964,
      "grad_norm": 3.0939722061157227,
      "learning_rate": 1.7312072892938496e-07,
      "loss": 0.7335,
      "step": 53670
    },
    {
      "epoch": 0.8356231561412658,
      "grad_norm": 3.4146599769592285,
      "learning_rate": 1.7303878992478e-07,
      "loss": 0.7711,
      "step": 53675
    },
    {
      "epoch": 0.8357009971432352,
      "grad_norm": 4.910788536071777,
      "learning_rate": 1.72956850920175e-07,
      "loss": 0.8482,
      "step": 53680
    },
    {
      "epoch": 0.8357788381452046,
      "grad_norm": 3.1009645462036133,
      "learning_rate": 1.7287491191557005e-07,
      "loss": 0.6524,
      "step": 53685
    },
    {
      "epoch": 0.835856679147174,
      "grad_norm": 2.981515407562256,
      "learning_rate": 1.7279297291096507e-07,
      "loss": 0.8402,
      "step": 53690
    },
    {
      "epoch": 0.8359345201491434,
      "grad_norm": 6.21446418762207,
      "learning_rate": 1.727110339063601e-07,
      "loss": 0.8063,
      "step": 53695
    },
    {
      "epoch": 0.8360123611511128,
      "grad_norm": 5.540207862854004,
      "learning_rate": 1.7262909490175511e-07,
      "loss": 0.8062,
      "step": 53700
    },
    {
      "epoch": 0.8360902021530822,
      "grad_norm": 5.194900035858154,
      "learning_rate": 1.7254715589715016e-07,
      "loss": 0.7097,
      "step": 53705
    },
    {
      "epoch": 0.8361680431550514,
      "grad_norm": 3.1524300575256348,
      "learning_rate": 1.7246521689254518e-07,
      "loss": 0.7798,
      "step": 53710
    },
    {
      "epoch": 0.8362458841570208,
      "grad_norm": 3.017381429672241,
      "learning_rate": 1.723832778879402e-07,
      "loss": 0.6683,
      "step": 53715
    },
    {
      "epoch": 0.8363237251589902,
      "grad_norm": 4.3789825439453125,
      "learning_rate": 1.7230133888333523e-07,
      "loss": 0.7985,
      "step": 53720
    },
    {
      "epoch": 0.8364015661609596,
      "grad_norm": 3.5694572925567627,
      "learning_rate": 1.7221939987873028e-07,
      "loss": 0.7965,
      "step": 53725
    },
    {
      "epoch": 0.836479407162929,
      "grad_norm": 5.385875701904297,
      "learning_rate": 1.7213746087412527e-07,
      "loss": 0.7883,
      "step": 53730
    },
    {
      "epoch": 0.8365572481648984,
      "grad_norm": 5.480561256408691,
      "learning_rate": 1.7205552186952032e-07,
      "loss": 0.7789,
      "step": 53735
    },
    {
      "epoch": 0.8366350891668678,
      "grad_norm": 3.64609956741333,
      "learning_rate": 1.7197358286491534e-07,
      "loss": 0.6761,
      "step": 53740
    },
    {
      "epoch": 0.8367129301688372,
      "grad_norm": 4.346398830413818,
      "learning_rate": 1.718916438603104e-07,
      "loss": 0.6491,
      "step": 53745
    },
    {
      "epoch": 0.8367907711708065,
      "grad_norm": 3.5501697063446045,
      "learning_rate": 1.7180970485570539e-07,
      "loss": 0.8697,
      "step": 53750
    },
    {
      "epoch": 0.8368686121727759,
      "grad_norm": 2.6795217990875244,
      "learning_rate": 1.7172776585110043e-07,
      "loss": 0.6668,
      "step": 53755
    },
    {
      "epoch": 0.8369464531747453,
      "grad_norm": 2.9927480220794678,
      "learning_rate": 1.7164582684649546e-07,
      "loss": 0.6899,
      "step": 53760
    },
    {
      "epoch": 0.8370242941767146,
      "grad_norm": 4.172988414764404,
      "learning_rate": 1.715638878418905e-07,
      "loss": 0.6101,
      "step": 53765
    },
    {
      "epoch": 0.837102135178684,
      "grad_norm": 4.198971271514893,
      "learning_rate": 1.714819488372855e-07,
      "loss": 0.6837,
      "step": 53770
    },
    {
      "epoch": 0.8371799761806534,
      "grad_norm": 3.6658859252929688,
      "learning_rate": 1.7140000983268055e-07,
      "loss": 0.7007,
      "step": 53775
    },
    {
      "epoch": 0.8372578171826228,
      "grad_norm": 3.6414554119110107,
      "learning_rate": 1.7131807082807557e-07,
      "loss": 0.7257,
      "step": 53780
    },
    {
      "epoch": 0.8373356581845921,
      "grad_norm": 4.221417427062988,
      "learning_rate": 1.712361318234706e-07,
      "loss": 0.8251,
      "step": 53785
    },
    {
      "epoch": 0.8374134991865615,
      "grad_norm": 3.2467780113220215,
      "learning_rate": 1.7115419281886561e-07,
      "loss": 0.6414,
      "step": 53790
    },
    {
      "epoch": 0.8374913401885309,
      "grad_norm": 7.528860092163086,
      "learning_rate": 1.7107225381426066e-07,
      "loss": 0.6507,
      "step": 53795
    },
    {
      "epoch": 0.8375691811905003,
      "grad_norm": 6.706357955932617,
      "learning_rate": 1.7099031480965568e-07,
      "loss": 0.7165,
      "step": 53800
    },
    {
      "epoch": 0.8376470221924697,
      "grad_norm": 4.345954895019531,
      "learning_rate": 1.709083758050507e-07,
      "loss": 0.7434,
      "step": 53805
    },
    {
      "epoch": 0.8377248631944391,
      "grad_norm": 3.39530611038208,
      "learning_rate": 1.7082643680044573e-07,
      "loss": 0.8522,
      "step": 53810
    },
    {
      "epoch": 0.8378027041964085,
      "grad_norm": 5.489085674285889,
      "learning_rate": 1.7074449779584078e-07,
      "loss": 0.7407,
      "step": 53815
    },
    {
      "epoch": 0.8378805451983778,
      "grad_norm": 4.326323986053467,
      "learning_rate": 1.7066255879123577e-07,
      "loss": 0.8435,
      "step": 53820
    },
    {
      "epoch": 0.8379583862003471,
      "grad_norm": 4.113926410675049,
      "learning_rate": 1.7058061978663082e-07,
      "loss": 0.7564,
      "step": 53825
    },
    {
      "epoch": 0.8380362272023165,
      "grad_norm": 4.471818447113037,
      "learning_rate": 1.7049868078202584e-07,
      "loss": 0.6765,
      "step": 53830
    },
    {
      "epoch": 0.8381140682042859,
      "grad_norm": 5.135436058044434,
      "learning_rate": 1.704167417774209e-07,
      "loss": 0.8069,
      "step": 53835
    },
    {
      "epoch": 0.8381919092062553,
      "grad_norm": 3.243103265762329,
      "learning_rate": 1.7033480277281589e-07,
      "loss": 0.6836,
      "step": 53840
    },
    {
      "epoch": 0.8382697502082247,
      "grad_norm": 3.2555391788482666,
      "learning_rate": 1.7025286376821094e-07,
      "loss": 0.7741,
      "step": 53845
    },
    {
      "epoch": 0.8383475912101941,
      "grad_norm": 4.274740219116211,
      "learning_rate": 1.7017092476360598e-07,
      "loss": 0.7743,
      "step": 53850
    },
    {
      "epoch": 0.8384254322121635,
      "grad_norm": 3.827758550643921,
      "learning_rate": 1.7008898575900098e-07,
      "loss": 0.729,
      "step": 53855
    },
    {
      "epoch": 0.8385032732141328,
      "grad_norm": 2.5682132244110107,
      "learning_rate": 1.7000704675439603e-07,
      "loss": 0.7628,
      "step": 53860
    },
    {
      "epoch": 0.8385811142161022,
      "grad_norm": 3.225823163986206,
      "learning_rate": 1.6992510774979105e-07,
      "loss": 0.712,
      "step": 53865
    },
    {
      "epoch": 0.8386589552180715,
      "grad_norm": 2.7413926124572754,
      "learning_rate": 1.698431687451861e-07,
      "loss": 0.6646,
      "step": 53870
    },
    {
      "epoch": 0.8387367962200409,
      "grad_norm": 4.771670818328857,
      "learning_rate": 1.697612297405811e-07,
      "loss": 0.6705,
      "step": 53875
    },
    {
      "epoch": 0.8388146372220103,
      "grad_norm": 3.410545587539673,
      "learning_rate": 1.6967929073597614e-07,
      "loss": 0.6591,
      "step": 53880
    },
    {
      "epoch": 0.8388924782239797,
      "grad_norm": 3.6376476287841797,
      "learning_rate": 1.6959735173137116e-07,
      "loss": 0.696,
      "step": 53885
    },
    {
      "epoch": 0.8389703192259491,
      "grad_norm": 3.53463077545166,
      "learning_rate": 1.695154127267662e-07,
      "loss": 0.7654,
      "step": 53890
    },
    {
      "epoch": 0.8390481602279185,
      "grad_norm": 3.6810109615325928,
      "learning_rate": 1.694334737221612e-07,
      "loss": 0.7684,
      "step": 53895
    },
    {
      "epoch": 0.8391260012298878,
      "grad_norm": 4.402599334716797,
      "learning_rate": 1.6935153471755626e-07,
      "loss": 0.6878,
      "step": 53900
    },
    {
      "epoch": 0.8392038422318572,
      "grad_norm": 3.581925630569458,
      "learning_rate": 1.6926959571295128e-07,
      "loss": 0.7768,
      "step": 53905
    },
    {
      "epoch": 0.8392816832338266,
      "grad_norm": 4.308877468109131,
      "learning_rate": 1.691876567083463e-07,
      "loss": 0.8605,
      "step": 53910
    },
    {
      "epoch": 0.839359524235796,
      "grad_norm": 5.019647121429443,
      "learning_rate": 1.6910571770374132e-07,
      "loss": 0.7542,
      "step": 53915
    },
    {
      "epoch": 0.8394373652377654,
      "grad_norm": 6.9310832023620605,
      "learning_rate": 1.6902377869913637e-07,
      "loss": 0.6673,
      "step": 53920
    },
    {
      "epoch": 0.8395152062397347,
      "grad_norm": 2.740588903427124,
      "learning_rate": 1.689418396945314e-07,
      "loss": 0.7727,
      "step": 53925
    },
    {
      "epoch": 0.8395930472417041,
      "grad_norm": 4.196685314178467,
      "learning_rate": 1.6885990068992641e-07,
      "loss": 0.6807,
      "step": 53930
    },
    {
      "epoch": 0.8396708882436734,
      "grad_norm": 3.8610050678253174,
      "learning_rate": 1.6877796168532144e-07,
      "loss": 0.7526,
      "step": 53935
    },
    {
      "epoch": 0.8397487292456428,
      "grad_norm": 3.6284873485565186,
      "learning_rate": 1.6869602268071648e-07,
      "loss": 0.8316,
      "step": 53940
    },
    {
      "epoch": 0.8398265702476122,
      "grad_norm": 3.7958626747131348,
      "learning_rate": 1.6861408367611148e-07,
      "loss": 0.6867,
      "step": 53945
    },
    {
      "epoch": 0.8399044112495816,
      "grad_norm": 5.251313209533691,
      "learning_rate": 1.6853214467150653e-07,
      "loss": 0.8335,
      "step": 53950
    },
    {
      "epoch": 0.839982252251551,
      "grad_norm": 3.8648130893707275,
      "learning_rate": 1.6845020566690155e-07,
      "loss": 0.7956,
      "step": 53955
    },
    {
      "epoch": 0.8400600932535204,
      "grad_norm": 4.290517330169678,
      "learning_rate": 1.683682666622966e-07,
      "loss": 0.7458,
      "step": 53960
    },
    {
      "epoch": 0.8401379342554898,
      "grad_norm": 3.686875104904175,
      "learning_rate": 1.682863276576916e-07,
      "loss": 0.7772,
      "step": 53965
    },
    {
      "epoch": 0.8402157752574592,
      "grad_norm": 3.391662359237671,
      "learning_rate": 1.6820438865308664e-07,
      "loss": 0.7463,
      "step": 53970
    },
    {
      "epoch": 0.8402936162594284,
      "grad_norm": 4.731991767883301,
      "learning_rate": 1.6812244964848166e-07,
      "loss": 0.7398,
      "step": 53975
    },
    {
      "epoch": 0.8403714572613978,
      "grad_norm": 3.0060553550720215,
      "learning_rate": 1.6804051064387669e-07,
      "loss": 0.7053,
      "step": 53980
    },
    {
      "epoch": 0.8404492982633672,
      "grad_norm": 4.913924694061279,
      "learning_rate": 1.679585716392717e-07,
      "loss": 0.8544,
      "step": 53985
    },
    {
      "epoch": 0.8405271392653366,
      "grad_norm": 2.4100394248962402,
      "learning_rate": 1.6787663263466676e-07,
      "loss": 0.6416,
      "step": 53990
    },
    {
      "epoch": 0.840604980267306,
      "grad_norm": 6.283283233642578,
      "learning_rate": 1.6779469363006178e-07,
      "loss": 0.7279,
      "step": 53995
    },
    {
      "epoch": 0.8406828212692754,
      "grad_norm": 3.3576266765594482,
      "learning_rate": 1.677127546254568e-07,
      "loss": 0.7779,
      "step": 54000
    },
    {
      "epoch": 0.8407606622712448,
      "grad_norm": 5.686350345611572,
      "learning_rate": 1.6763081562085182e-07,
      "loss": 0.7096,
      "step": 54005
    },
    {
      "epoch": 0.8408385032732141,
      "grad_norm": 4.298459529876709,
      "learning_rate": 1.6754887661624687e-07,
      "loss": 0.7366,
      "step": 54010
    },
    {
      "epoch": 0.8409163442751835,
      "grad_norm": 3.4432363510131836,
      "learning_rate": 1.6746693761164187e-07,
      "loss": 0.6104,
      "step": 54015
    },
    {
      "epoch": 0.8409941852771529,
      "grad_norm": 2.8809094429016113,
      "learning_rate": 1.6738499860703691e-07,
      "loss": 0.712,
      "step": 54020
    },
    {
      "epoch": 0.8410720262791223,
      "grad_norm": 3.0508673191070557,
      "learning_rate": 1.6730305960243194e-07,
      "loss": 0.7759,
      "step": 54025
    },
    {
      "epoch": 0.8411498672810916,
      "grad_norm": 4.997377395629883,
      "learning_rate": 1.6722112059782698e-07,
      "loss": 0.7084,
      "step": 54030
    },
    {
      "epoch": 0.841227708283061,
      "grad_norm": 3.8306801319122314,
      "learning_rate": 1.6713918159322198e-07,
      "loss": 0.7906,
      "step": 54035
    },
    {
      "epoch": 0.8413055492850304,
      "grad_norm": 3.059521198272705,
      "learning_rate": 1.6705724258861703e-07,
      "loss": 0.7186,
      "step": 54040
    },
    {
      "epoch": 0.8413833902869998,
      "grad_norm": 2.5660817623138428,
      "learning_rate": 1.6697530358401205e-07,
      "loss": 0.6238,
      "step": 54045
    },
    {
      "epoch": 0.8414612312889691,
      "grad_norm": 5.484297752380371,
      "learning_rate": 1.668933645794071e-07,
      "loss": 0.7396,
      "step": 54050
    },
    {
      "epoch": 0.8415390722909385,
      "grad_norm": 3.250892162322998,
      "learning_rate": 1.668114255748021e-07,
      "loss": 0.7055,
      "step": 54055
    },
    {
      "epoch": 0.8416169132929079,
      "grad_norm": 3.01938533782959,
      "learning_rate": 1.6672948657019714e-07,
      "loss": 0.6077,
      "step": 54060
    },
    {
      "epoch": 0.8416947542948773,
      "grad_norm": 3.929241180419922,
      "learning_rate": 1.6664754756559216e-07,
      "loss": 0.6981,
      "step": 54065
    },
    {
      "epoch": 0.8417725952968467,
      "grad_norm": 4.001524925231934,
      "learning_rate": 1.6656560856098719e-07,
      "loss": 0.6676,
      "step": 54070
    },
    {
      "epoch": 0.8418504362988161,
      "grad_norm": 3.402467727661133,
      "learning_rate": 1.664836695563822e-07,
      "loss": 0.7154,
      "step": 54075
    },
    {
      "epoch": 0.8419282773007855,
      "grad_norm": 3.45058274269104,
      "learning_rate": 1.6640173055177726e-07,
      "loss": 0.8047,
      "step": 54080
    },
    {
      "epoch": 0.8420061183027548,
      "grad_norm": 3.853041887283325,
      "learning_rate": 1.6631979154717228e-07,
      "loss": 0.8138,
      "step": 54085
    },
    {
      "epoch": 0.8420839593047241,
      "grad_norm": 4.1795654296875,
      "learning_rate": 1.662378525425673e-07,
      "loss": 0.673,
      "step": 54090
    },
    {
      "epoch": 0.8421618003066935,
      "grad_norm": 3.3027775287628174,
      "learning_rate": 1.6615591353796232e-07,
      "loss": 0.8009,
      "step": 54095
    },
    {
      "epoch": 0.8422396413086629,
      "grad_norm": 4.8819098472595215,
      "learning_rate": 1.6607397453335737e-07,
      "loss": 0.6937,
      "step": 54100
    },
    {
      "epoch": 0.8423174823106323,
      "grad_norm": 2.8995580673217773,
      "learning_rate": 1.6599203552875237e-07,
      "loss": 0.708,
      "step": 54105
    },
    {
      "epoch": 0.8423953233126017,
      "grad_norm": 3.6737935543060303,
      "learning_rate": 1.6591009652414741e-07,
      "loss": 0.8152,
      "step": 54110
    },
    {
      "epoch": 0.8424731643145711,
      "grad_norm": 2.911975860595703,
      "learning_rate": 1.6582815751954244e-07,
      "loss": 0.7492,
      "step": 54115
    },
    {
      "epoch": 0.8425510053165405,
      "grad_norm": 5.798208236694336,
      "learning_rate": 1.6574621851493748e-07,
      "loss": 0.8005,
      "step": 54120
    },
    {
      "epoch": 0.8426288463185098,
      "grad_norm": 4.6597466468811035,
      "learning_rate": 1.6566427951033248e-07,
      "loss": 0.7735,
      "step": 54125
    },
    {
      "epoch": 0.8427066873204792,
      "grad_norm": 7.712786674499512,
      "learning_rate": 1.6558234050572753e-07,
      "loss": 0.6407,
      "step": 54130
    },
    {
      "epoch": 0.8427845283224485,
      "grad_norm": 2.8602709770202637,
      "learning_rate": 1.6550040150112255e-07,
      "loss": 0.655,
      "step": 54135
    },
    {
      "epoch": 0.8428623693244179,
      "grad_norm": 4.437834739685059,
      "learning_rate": 1.6541846249651757e-07,
      "loss": 0.7739,
      "step": 54140
    },
    {
      "epoch": 0.8429402103263873,
      "grad_norm": 2.526226758956909,
      "learning_rate": 1.653365234919126e-07,
      "loss": 0.6741,
      "step": 54145
    },
    {
      "epoch": 0.8430180513283567,
      "grad_norm": 5.441184043884277,
      "learning_rate": 1.6525458448730764e-07,
      "loss": 0.7585,
      "step": 54150
    },
    {
      "epoch": 0.8430958923303261,
      "grad_norm": 8.626420974731445,
      "learning_rate": 1.6517264548270266e-07,
      "loss": 0.7265,
      "step": 54155
    },
    {
      "epoch": 0.8431737333322955,
      "grad_norm": 4.2593231201171875,
      "learning_rate": 1.6509070647809769e-07,
      "loss": 0.7882,
      "step": 54160
    },
    {
      "epoch": 0.8432515743342648,
      "grad_norm": 3.0201570987701416,
      "learning_rate": 1.6500876747349273e-07,
      "loss": 0.79,
      "step": 54165
    },
    {
      "epoch": 0.8433294153362342,
      "grad_norm": 2.9060263633728027,
      "learning_rate": 1.6492682846888776e-07,
      "loss": 0.6436,
      "step": 54170
    },
    {
      "epoch": 0.8434072563382036,
      "grad_norm": 7.522155284881592,
      "learning_rate": 1.648448894642828e-07,
      "loss": 0.848,
      "step": 54175
    },
    {
      "epoch": 0.843485097340173,
      "grad_norm": 2.6187744140625,
      "learning_rate": 1.647629504596778e-07,
      "loss": 0.7272,
      "step": 54180
    },
    {
      "epoch": 0.8435629383421424,
      "grad_norm": 4.149000644683838,
      "learning_rate": 1.6468101145507285e-07,
      "loss": 0.7488,
      "step": 54185
    },
    {
      "epoch": 0.8436407793441117,
      "grad_norm": 3.8222572803497314,
      "learning_rate": 1.6459907245046787e-07,
      "loss": 0.7275,
      "step": 54190
    },
    {
      "epoch": 0.8437186203460811,
      "grad_norm": 3.980178117752075,
      "learning_rate": 1.645171334458629e-07,
      "loss": 0.688,
      "step": 54195
    },
    {
      "epoch": 0.8437964613480504,
      "grad_norm": 2.55849289894104,
      "learning_rate": 1.6443519444125791e-07,
      "loss": 0.7145,
      "step": 54200
    },
    {
      "epoch": 0.8438743023500198,
      "grad_norm": 4.3173956871032715,
      "learning_rate": 1.6435325543665296e-07,
      "loss": 0.7784,
      "step": 54205
    },
    {
      "epoch": 0.8439521433519892,
      "grad_norm": 3.4170172214508057,
      "learning_rate": 1.6427131643204799e-07,
      "loss": 0.7812,
      "step": 54210
    },
    {
      "epoch": 0.8440299843539586,
      "grad_norm": 3.8942880630493164,
      "learning_rate": 1.64189377427443e-07,
      "loss": 0.7983,
      "step": 54215
    },
    {
      "epoch": 0.844107825355928,
      "grad_norm": 6.527953624725342,
      "learning_rate": 1.6410743842283803e-07,
      "loss": 0.6636,
      "step": 54220
    },
    {
      "epoch": 0.8441856663578974,
      "grad_norm": 3.2828967571258545,
      "learning_rate": 1.6402549941823308e-07,
      "loss": 0.7333,
      "step": 54225
    },
    {
      "epoch": 0.8442635073598668,
      "grad_norm": 6.369480133056641,
      "learning_rate": 1.6394356041362807e-07,
      "loss": 0.7414,
      "step": 54230
    },
    {
      "epoch": 0.8443413483618362,
      "grad_norm": 3.0079944133758545,
      "learning_rate": 1.6386162140902312e-07,
      "loss": 0.8128,
      "step": 54235
    },
    {
      "epoch": 0.8444191893638054,
      "grad_norm": 4.320895195007324,
      "learning_rate": 1.6377968240441814e-07,
      "loss": 0.8236,
      "step": 54240
    },
    {
      "epoch": 0.8444970303657748,
      "grad_norm": 5.211856365203857,
      "learning_rate": 1.636977433998132e-07,
      "loss": 0.7227,
      "step": 54245
    },
    {
      "epoch": 0.8445748713677442,
      "grad_norm": 3.3257970809936523,
      "learning_rate": 1.636158043952082e-07,
      "loss": 0.7071,
      "step": 54250
    },
    {
      "epoch": 0.8446527123697136,
      "grad_norm": 2.6674141883850098,
      "learning_rate": 1.6353386539060324e-07,
      "loss": 0.7366,
      "step": 54255
    },
    {
      "epoch": 0.844730553371683,
      "grad_norm": 3.7396609783172607,
      "learning_rate": 1.6345192638599826e-07,
      "loss": 0.7135,
      "step": 54260
    },
    {
      "epoch": 0.8448083943736524,
      "grad_norm": 4.606040954589844,
      "learning_rate": 1.633699873813933e-07,
      "loss": 0.7294,
      "step": 54265
    },
    {
      "epoch": 0.8448862353756218,
      "grad_norm": 5.433845043182373,
      "learning_rate": 1.632880483767883e-07,
      "loss": 0.7741,
      "step": 54270
    },
    {
      "epoch": 0.8449640763775911,
      "grad_norm": 2.4507601261138916,
      "learning_rate": 1.6320610937218335e-07,
      "loss": 0.7687,
      "step": 54275
    },
    {
      "epoch": 0.8450419173795605,
      "grad_norm": 3.2316689491271973,
      "learning_rate": 1.6312417036757837e-07,
      "loss": 0.7967,
      "step": 54280
    },
    {
      "epoch": 0.8451197583815299,
      "grad_norm": 3.8701488971710205,
      "learning_rate": 1.630422313629734e-07,
      "loss": 0.7459,
      "step": 54285
    },
    {
      "epoch": 0.8451975993834993,
      "grad_norm": 4.442934513092041,
      "learning_rate": 1.6296029235836842e-07,
      "loss": 0.7438,
      "step": 54290
    },
    {
      "epoch": 0.8452754403854686,
      "grad_norm": 2.6943540573120117,
      "learning_rate": 1.6287835335376346e-07,
      "loss": 0.7747,
      "step": 54295
    },
    {
      "epoch": 0.845353281387438,
      "grad_norm": 2.8886303901672363,
      "learning_rate": 1.6279641434915849e-07,
      "loss": 0.7388,
      "step": 54300
    },
    {
      "epoch": 0.8454311223894074,
      "grad_norm": 2.605725049972534,
      "learning_rate": 1.627144753445535e-07,
      "loss": 0.8388,
      "step": 54305
    },
    {
      "epoch": 0.8455089633913768,
      "grad_norm": 5.513226509094238,
      "learning_rate": 1.6263253633994853e-07,
      "loss": 0.7468,
      "step": 54310
    },
    {
      "epoch": 0.8455868043933461,
      "grad_norm": 4.820272922515869,
      "learning_rate": 1.6255059733534358e-07,
      "loss": 0.687,
      "step": 54315
    },
    {
      "epoch": 0.8456646453953155,
      "grad_norm": 2.8883254528045654,
      "learning_rate": 1.6246865833073857e-07,
      "loss": 0.5899,
      "step": 54320
    },
    {
      "epoch": 0.8457424863972849,
      "grad_norm": 4.018860816955566,
      "learning_rate": 1.6238671932613362e-07,
      "loss": 0.8434,
      "step": 54325
    },
    {
      "epoch": 0.8458203273992543,
      "grad_norm": 3.574259042739868,
      "learning_rate": 1.6230478032152864e-07,
      "loss": 0.7812,
      "step": 54330
    },
    {
      "epoch": 0.8458981684012237,
      "grad_norm": 3.9640090465545654,
      "learning_rate": 1.622228413169237e-07,
      "loss": 0.7709,
      "step": 54335
    },
    {
      "epoch": 0.8459760094031931,
      "grad_norm": 4.083382606506348,
      "learning_rate": 1.621409023123187e-07,
      "loss": 0.7207,
      "step": 54340
    },
    {
      "epoch": 0.8460538504051625,
      "grad_norm": 7.4051384925842285,
      "learning_rate": 1.6205896330771374e-07,
      "loss": 0.6744,
      "step": 54345
    },
    {
      "epoch": 0.8461316914071318,
      "grad_norm": 4.949417591094971,
      "learning_rate": 1.6197702430310876e-07,
      "loss": 0.7763,
      "step": 54350
    },
    {
      "epoch": 0.8462095324091011,
      "grad_norm": 5.163852691650391,
      "learning_rate": 1.6189508529850378e-07,
      "loss": 0.6985,
      "step": 54355
    },
    {
      "epoch": 0.8462873734110705,
      "grad_norm": 5.4622015953063965,
      "learning_rate": 1.618131462938988e-07,
      "loss": 0.8493,
      "step": 54360
    },
    {
      "epoch": 0.8463652144130399,
      "grad_norm": 3.6427536010742188,
      "learning_rate": 1.6173120728929385e-07,
      "loss": 0.848,
      "step": 54365
    },
    {
      "epoch": 0.8464430554150093,
      "grad_norm": 2.2812554836273193,
      "learning_rate": 1.6164926828468887e-07,
      "loss": 0.7272,
      "step": 54370
    },
    {
      "epoch": 0.8465208964169787,
      "grad_norm": 4.5219268798828125,
      "learning_rate": 1.615673292800839e-07,
      "loss": 0.8017,
      "step": 54375
    },
    {
      "epoch": 0.8465987374189481,
      "grad_norm": 3.0619285106658936,
      "learning_rate": 1.6148539027547892e-07,
      "loss": 0.8338,
      "step": 54380
    },
    {
      "epoch": 0.8466765784209175,
      "grad_norm": 3.548933506011963,
      "learning_rate": 1.6140345127087396e-07,
      "loss": 0.7704,
      "step": 54385
    },
    {
      "epoch": 0.8467544194228868,
      "grad_norm": 3.909421920776367,
      "learning_rate": 1.6132151226626896e-07,
      "loss": 0.6494,
      "step": 54390
    },
    {
      "epoch": 0.8468322604248562,
      "grad_norm": 2.8283865451812744,
      "learning_rate": 1.61239573261664e-07,
      "loss": 0.7387,
      "step": 54395
    },
    {
      "epoch": 0.8469101014268255,
      "grad_norm": 3.719780683517456,
      "learning_rate": 1.6115763425705903e-07,
      "loss": 0.8337,
      "step": 54400
    },
    {
      "epoch": 0.8469879424287949,
      "grad_norm": 3.1348230838775635,
      "learning_rate": 1.6107569525245408e-07,
      "loss": 0.7208,
      "step": 54405
    },
    {
      "epoch": 0.8470657834307643,
      "grad_norm": 5.668137550354004,
      "learning_rate": 1.6099375624784907e-07,
      "loss": 0.7352,
      "step": 54410
    },
    {
      "epoch": 0.8471436244327337,
      "grad_norm": 2.6878905296325684,
      "learning_rate": 1.6091181724324412e-07,
      "loss": 0.7518,
      "step": 54415
    },
    {
      "epoch": 0.8472214654347031,
      "grad_norm": 3.4251387119293213,
      "learning_rate": 1.6082987823863914e-07,
      "loss": 0.7973,
      "step": 54420
    },
    {
      "epoch": 0.8472993064366725,
      "grad_norm": 4.888462543487549,
      "learning_rate": 1.607479392340342e-07,
      "loss": 0.6879,
      "step": 54425
    },
    {
      "epoch": 0.8473771474386418,
      "grad_norm": 2.98319673538208,
      "learning_rate": 1.606660002294292e-07,
      "loss": 0.6041,
      "step": 54430
    },
    {
      "epoch": 0.8474549884406112,
      "grad_norm": 4.2112956047058105,
      "learning_rate": 1.6058406122482424e-07,
      "loss": 0.6797,
      "step": 54435
    },
    {
      "epoch": 0.8475328294425806,
      "grad_norm": 3.399606227874756,
      "learning_rate": 1.6050212222021926e-07,
      "loss": 0.7727,
      "step": 54440
    },
    {
      "epoch": 0.84761067044455,
      "grad_norm": 3.315973997116089,
      "learning_rate": 1.6042018321561428e-07,
      "loss": 0.7074,
      "step": 54445
    },
    {
      "epoch": 0.8476885114465194,
      "grad_norm": 5.498802185058594,
      "learning_rate": 1.603382442110093e-07,
      "loss": 0.713,
      "step": 54450
    },
    {
      "epoch": 0.8477663524484887,
      "grad_norm": 7.100419998168945,
      "learning_rate": 1.6025630520640435e-07,
      "loss": 0.7823,
      "step": 54455
    },
    {
      "epoch": 0.8478441934504581,
      "grad_norm": 3.263831853866577,
      "learning_rate": 1.6017436620179937e-07,
      "loss": 0.7156,
      "step": 54460
    },
    {
      "epoch": 0.8479220344524274,
      "grad_norm": 3.68009877204895,
      "learning_rate": 1.600924271971944e-07,
      "loss": 0.6503,
      "step": 54465
    },
    {
      "epoch": 0.8479998754543968,
      "grad_norm": 4.636446475982666,
      "learning_rate": 1.6001048819258942e-07,
      "loss": 0.8785,
      "step": 54470
    },
    {
      "epoch": 0.8480777164563662,
      "grad_norm": 3.76599383354187,
      "learning_rate": 1.5992854918798446e-07,
      "loss": 0.7314,
      "step": 54475
    },
    {
      "epoch": 0.8481555574583356,
      "grad_norm": 3.541609525680542,
      "learning_rate": 1.5984661018337949e-07,
      "loss": 0.6667,
      "step": 54480
    },
    {
      "epoch": 0.848233398460305,
      "grad_norm": 3.3275065422058105,
      "learning_rate": 1.597646711787745e-07,
      "loss": 0.8136,
      "step": 54485
    },
    {
      "epoch": 0.8483112394622744,
      "grad_norm": 4.905918598175049,
      "learning_rate": 1.5968273217416956e-07,
      "loss": 0.6724,
      "step": 54490
    },
    {
      "epoch": 0.8483890804642438,
      "grad_norm": 3.3878726959228516,
      "learning_rate": 1.5960079316956458e-07,
      "loss": 0.8711,
      "step": 54495
    },
    {
      "epoch": 0.8484669214662132,
      "grad_norm": 3.004425048828125,
      "learning_rate": 1.595188541649596e-07,
      "loss": 0.6251,
      "step": 54500
    },
    {
      "epoch": 0.8485447624681824,
      "grad_norm": 6.4495649337768555,
      "learning_rate": 1.5943691516035462e-07,
      "loss": 0.7801,
      "step": 54505
    },
    {
      "epoch": 0.8486226034701518,
      "grad_norm": 8.112421035766602,
      "learning_rate": 1.5935497615574967e-07,
      "loss": 0.7884,
      "step": 54510
    },
    {
      "epoch": 0.8487004444721212,
      "grad_norm": 4.245112895965576,
      "learning_rate": 1.5927303715114467e-07,
      "loss": 0.69,
      "step": 54515
    },
    {
      "epoch": 0.8487782854740906,
      "grad_norm": 4.353617191314697,
      "learning_rate": 1.5919109814653971e-07,
      "loss": 0.7293,
      "step": 54520
    },
    {
      "epoch": 0.84885612647606,
      "grad_norm": 4.947503089904785,
      "learning_rate": 1.5910915914193474e-07,
      "loss": 0.7414,
      "step": 54525
    },
    {
      "epoch": 0.8489339674780294,
      "grad_norm": 3.1440436840057373,
      "learning_rate": 1.5902722013732979e-07,
      "loss": 0.7354,
      "step": 54530
    },
    {
      "epoch": 0.8490118084799988,
      "grad_norm": 4.830748558044434,
      "learning_rate": 1.5894528113272478e-07,
      "loss": 0.8158,
      "step": 54535
    },
    {
      "epoch": 0.8490896494819681,
      "grad_norm": 4.846890926361084,
      "learning_rate": 1.5886334212811983e-07,
      "loss": 0.7865,
      "step": 54540
    },
    {
      "epoch": 0.8491674904839375,
      "grad_norm": 4.882929801940918,
      "learning_rate": 1.5878140312351485e-07,
      "loss": 0.7126,
      "step": 54545
    },
    {
      "epoch": 0.8492453314859069,
      "grad_norm": 2.780062675476074,
      "learning_rate": 1.586994641189099e-07,
      "loss": 0.7801,
      "step": 54550
    },
    {
      "epoch": 0.8493231724878763,
      "grad_norm": 10.664603233337402,
      "learning_rate": 1.586175251143049e-07,
      "loss": 0.7049,
      "step": 54555
    },
    {
      "epoch": 0.8494010134898456,
      "grad_norm": 3.8737123012542725,
      "learning_rate": 1.5853558610969994e-07,
      "loss": 0.7117,
      "step": 54560
    },
    {
      "epoch": 0.849478854491815,
      "grad_norm": 2.855785608291626,
      "learning_rate": 1.5845364710509497e-07,
      "loss": 0.7226,
      "step": 54565
    },
    {
      "epoch": 0.8495566954937844,
      "grad_norm": 4.367470741271973,
      "learning_rate": 1.5837170810049e-07,
      "loss": 0.6398,
      "step": 54570
    },
    {
      "epoch": 0.8496345364957538,
      "grad_norm": 4.615283012390137,
      "learning_rate": 1.58289769095885e-07,
      "loss": 0.7782,
      "step": 54575
    },
    {
      "epoch": 0.8497123774977231,
      "grad_norm": 3.5703768730163574,
      "learning_rate": 1.5820783009128006e-07,
      "loss": 0.7362,
      "step": 54580
    },
    {
      "epoch": 0.8497902184996925,
      "grad_norm": 3.48062801361084,
      "learning_rate": 1.5812589108667508e-07,
      "loss": 0.7352,
      "step": 54585
    },
    {
      "epoch": 0.8498680595016619,
      "grad_norm": 7.972591876983643,
      "learning_rate": 1.580439520820701e-07,
      "loss": 0.6853,
      "step": 54590
    },
    {
      "epoch": 0.8499459005036313,
      "grad_norm": 3.703296661376953,
      "learning_rate": 1.5796201307746512e-07,
      "loss": 0.6455,
      "step": 54595
    },
    {
      "epoch": 0.8500237415056007,
      "grad_norm": 4.787045955657959,
      "learning_rate": 1.5788007407286017e-07,
      "loss": 0.6597,
      "step": 54600
    },
    {
      "epoch": 0.8501015825075701,
      "grad_norm": 3.145493745803833,
      "learning_rate": 1.5779813506825517e-07,
      "loss": 0.7485,
      "step": 54605
    },
    {
      "epoch": 0.8501794235095395,
      "grad_norm": 2.9877171516418457,
      "learning_rate": 1.5771619606365022e-07,
      "loss": 0.666,
      "step": 54610
    },
    {
      "epoch": 0.8502572645115087,
      "grad_norm": 3.6364433765411377,
      "learning_rate": 1.5763425705904524e-07,
      "loss": 0.763,
      "step": 54615
    },
    {
      "epoch": 0.8503351055134781,
      "grad_norm": 3.4951741695404053,
      "learning_rate": 1.5755231805444029e-07,
      "loss": 0.7154,
      "step": 54620
    },
    {
      "epoch": 0.8504129465154475,
      "grad_norm": 3.4532077312469482,
      "learning_rate": 1.5747037904983528e-07,
      "loss": 0.7165,
      "step": 54625
    },
    {
      "epoch": 0.8504907875174169,
      "grad_norm": 3.1939377784729004,
      "learning_rate": 1.5738844004523033e-07,
      "loss": 0.7657,
      "step": 54630
    },
    {
      "epoch": 0.8505686285193863,
      "grad_norm": 13.525358200073242,
      "learning_rate": 1.5730650104062535e-07,
      "loss": 0.8449,
      "step": 54635
    },
    {
      "epoch": 0.8506464695213557,
      "grad_norm": 5.139512062072754,
      "learning_rate": 1.5722456203602037e-07,
      "loss": 0.8794,
      "step": 54640
    },
    {
      "epoch": 0.8507243105233251,
      "grad_norm": 3.2967913150787354,
      "learning_rate": 1.571426230314154e-07,
      "loss": 0.7949,
      "step": 54645
    },
    {
      "epoch": 0.8508021515252945,
      "grad_norm": 9.876358985900879,
      "learning_rate": 1.5706068402681044e-07,
      "loss": 0.7852,
      "step": 54650
    },
    {
      "epoch": 0.8508799925272638,
      "grad_norm": 5.027626991271973,
      "learning_rate": 1.5697874502220547e-07,
      "loss": 0.742,
      "step": 54655
    },
    {
      "epoch": 0.8509578335292332,
      "grad_norm": 10.290229797363281,
      "learning_rate": 1.568968060176005e-07,
      "loss": 0.7779,
      "step": 54660
    },
    {
      "epoch": 0.8510356745312025,
      "grad_norm": 3.6981658935546875,
      "learning_rate": 1.568148670129955e-07,
      "loss": 0.8709,
      "step": 54665
    },
    {
      "epoch": 0.8511135155331719,
      "grad_norm": 3.59417462348938,
      "learning_rate": 1.5673292800839056e-07,
      "loss": 0.745,
      "step": 54670
    },
    {
      "epoch": 0.8511913565351413,
      "grad_norm": 4.495562553405762,
      "learning_rate": 1.5665098900378558e-07,
      "loss": 0.7481,
      "step": 54675
    },
    {
      "epoch": 0.8512691975371107,
      "grad_norm": 4.144758224487305,
      "learning_rate": 1.565690499991806e-07,
      "loss": 0.722,
      "step": 54680
    },
    {
      "epoch": 0.8513470385390801,
      "grad_norm": 3.5521469116210938,
      "learning_rate": 1.5648711099457562e-07,
      "loss": 0.7347,
      "step": 54685
    },
    {
      "epoch": 0.8514248795410495,
      "grad_norm": 4.065563678741455,
      "learning_rate": 1.5640517198997067e-07,
      "loss": 0.7973,
      "step": 54690
    },
    {
      "epoch": 0.8515027205430188,
      "grad_norm": 2.650359869003296,
      "learning_rate": 1.5632323298536567e-07,
      "loss": 0.7004,
      "step": 54695
    },
    {
      "epoch": 0.8515805615449882,
      "grad_norm": 4.409617900848389,
      "learning_rate": 1.5624129398076072e-07,
      "loss": 0.7721,
      "step": 54700
    },
    {
      "epoch": 0.8516584025469576,
      "grad_norm": 6.663437366485596,
      "learning_rate": 1.5615935497615574e-07,
      "loss": 0.6898,
      "step": 54705
    },
    {
      "epoch": 0.851736243548927,
      "grad_norm": 4.077977180480957,
      "learning_rate": 1.5607741597155079e-07,
      "loss": 0.6518,
      "step": 54710
    },
    {
      "epoch": 0.8518140845508964,
      "grad_norm": 5.420622825622559,
      "learning_rate": 1.5599547696694578e-07,
      "loss": 0.7956,
      "step": 54715
    },
    {
      "epoch": 0.8518919255528657,
      "grad_norm": 5.301725387573242,
      "learning_rate": 1.5591353796234083e-07,
      "loss": 0.7009,
      "step": 54720
    },
    {
      "epoch": 0.8519697665548351,
      "grad_norm": 3.2651212215423584,
      "learning_rate": 1.5583159895773585e-07,
      "loss": 0.6864,
      "step": 54725
    },
    {
      "epoch": 0.8520476075568044,
      "grad_norm": 2.387659788131714,
      "learning_rate": 1.5574965995313087e-07,
      "loss": 0.7714,
      "step": 54730
    },
    {
      "epoch": 0.8521254485587738,
      "grad_norm": 3.035280466079712,
      "learning_rate": 1.556677209485259e-07,
      "loss": 0.8478,
      "step": 54735
    },
    {
      "epoch": 0.8522032895607432,
      "grad_norm": 2.891357898712158,
      "learning_rate": 1.5558578194392094e-07,
      "loss": 0.7958,
      "step": 54740
    },
    {
      "epoch": 0.8522811305627126,
      "grad_norm": 4.854222774505615,
      "learning_rate": 1.5550384293931597e-07,
      "loss": 0.7606,
      "step": 54745
    },
    {
      "epoch": 0.852358971564682,
      "grad_norm": 4.361767768859863,
      "learning_rate": 1.55421903934711e-07,
      "loss": 0.652,
      "step": 54750
    },
    {
      "epoch": 0.8524368125666514,
      "grad_norm": 3.7960891723632812,
      "learning_rate": 1.55339964930106e-07,
      "loss": 0.7963,
      "step": 54755
    },
    {
      "epoch": 0.8525146535686208,
      "grad_norm": 3.6916964054107666,
      "learning_rate": 1.5525802592550106e-07,
      "loss": 0.7467,
      "step": 54760
    },
    {
      "epoch": 0.8525924945705902,
      "grad_norm": 3.229459524154663,
      "learning_rate": 1.5517608692089605e-07,
      "loss": 0.8421,
      "step": 54765
    },
    {
      "epoch": 0.8526703355725594,
      "grad_norm": 3.339733600616455,
      "learning_rate": 1.550941479162911e-07,
      "loss": 0.8467,
      "step": 54770
    },
    {
      "epoch": 0.8527481765745288,
      "grad_norm": 5.676372528076172,
      "learning_rate": 1.5501220891168612e-07,
      "loss": 0.7827,
      "step": 54775
    },
    {
      "epoch": 0.8528260175764982,
      "grad_norm": 3.0010364055633545,
      "learning_rate": 1.5493026990708117e-07,
      "loss": 0.7579,
      "step": 54780
    },
    {
      "epoch": 0.8529038585784676,
      "grad_norm": 3.73765230178833,
      "learning_rate": 1.5484833090247617e-07,
      "loss": 0.699,
      "step": 54785
    },
    {
      "epoch": 0.852981699580437,
      "grad_norm": 10.715141296386719,
      "learning_rate": 1.5476639189787122e-07,
      "loss": 0.7884,
      "step": 54790
    },
    {
      "epoch": 0.8530595405824064,
      "grad_norm": 2.64835786819458,
      "learning_rate": 1.5468445289326624e-07,
      "loss": 0.7404,
      "step": 54795
    },
    {
      "epoch": 0.8531373815843758,
      "grad_norm": 4.952651500701904,
      "learning_rate": 1.5460251388866129e-07,
      "loss": 0.7057,
      "step": 54800
    },
    {
      "epoch": 0.8532152225863451,
      "grad_norm": 4.2666425704956055,
      "learning_rate": 1.545205748840563e-07,
      "loss": 0.5695,
      "step": 54805
    },
    {
      "epoch": 0.8532930635883145,
      "grad_norm": 4.233102798461914,
      "learning_rate": 1.5443863587945133e-07,
      "loss": 0.7213,
      "step": 54810
    },
    {
      "epoch": 0.8533709045902839,
      "grad_norm": 3.358175277709961,
      "learning_rate": 1.5435669687484638e-07,
      "loss": 0.6951,
      "step": 54815
    },
    {
      "epoch": 0.8534487455922533,
      "grad_norm": 4.623523712158203,
      "learning_rate": 1.5427475787024137e-07,
      "loss": 0.8572,
      "step": 54820
    },
    {
      "epoch": 0.8535265865942226,
      "grad_norm": 8.545863151550293,
      "learning_rate": 1.5419281886563642e-07,
      "loss": 0.7643,
      "step": 54825
    },
    {
      "epoch": 0.853604427596192,
      "grad_norm": 5.161164283752441,
      "learning_rate": 1.5411087986103144e-07,
      "loss": 0.7105,
      "step": 54830
    },
    {
      "epoch": 0.8536822685981614,
      "grad_norm": 4.408745765686035,
      "learning_rate": 1.540289408564265e-07,
      "loss": 0.6944,
      "step": 54835
    },
    {
      "epoch": 0.8537601096001308,
      "grad_norm": 6.125401020050049,
      "learning_rate": 1.539470018518215e-07,
      "loss": 0.7073,
      "step": 54840
    },
    {
      "epoch": 0.8538379506021001,
      "grad_norm": 8.1498384475708,
      "learning_rate": 1.5386506284721654e-07,
      "loss": 0.813,
      "step": 54845
    },
    {
      "epoch": 0.8539157916040695,
      "grad_norm": 3.6147890090942383,
      "learning_rate": 1.5378312384261156e-07,
      "loss": 0.6606,
      "step": 54850
    },
    {
      "epoch": 0.8539936326060389,
      "grad_norm": 5.331207275390625,
      "learning_rate": 1.5370118483800658e-07,
      "loss": 0.6395,
      "step": 54855
    },
    {
      "epoch": 0.8540714736080083,
      "grad_norm": 3.0040628910064697,
      "learning_rate": 1.536192458334016e-07,
      "loss": 0.7448,
      "step": 54860
    },
    {
      "epoch": 0.8541493146099777,
      "grad_norm": 3.7811992168426514,
      "learning_rate": 1.5353730682879665e-07,
      "loss": 0.713,
      "step": 54865
    },
    {
      "epoch": 0.8542271556119471,
      "grad_norm": 2.519601821899414,
      "learning_rate": 1.5345536782419167e-07,
      "loss": 0.6788,
      "step": 54870
    },
    {
      "epoch": 0.8543049966139165,
      "grad_norm": 5.137478351593018,
      "learning_rate": 1.533734288195867e-07,
      "loss": 0.7324,
      "step": 54875
    },
    {
      "epoch": 0.8543828376158857,
      "grad_norm": 4.291418552398682,
      "learning_rate": 1.5329148981498172e-07,
      "loss": 0.7516,
      "step": 54880
    },
    {
      "epoch": 0.8544606786178551,
      "grad_norm": 5.400655746459961,
      "learning_rate": 1.5320955081037676e-07,
      "loss": 0.8454,
      "step": 54885
    },
    {
      "epoch": 0.8545385196198245,
      "grad_norm": 3.409604549407959,
      "learning_rate": 1.5312761180577176e-07,
      "loss": 0.7974,
      "step": 54890
    },
    {
      "epoch": 0.8546163606217939,
      "grad_norm": 4.226031303405762,
      "learning_rate": 1.530456728011668e-07,
      "loss": 0.8761,
      "step": 54895
    },
    {
      "epoch": 0.8546942016237633,
      "grad_norm": 2.83233642578125,
      "learning_rate": 1.5296373379656183e-07,
      "loss": 0.6293,
      "step": 54900
    },
    {
      "epoch": 0.8547720426257327,
      "grad_norm": 3.253648281097412,
      "learning_rate": 1.5288179479195688e-07,
      "loss": 0.6445,
      "step": 54905
    },
    {
      "epoch": 0.8548498836277021,
      "grad_norm": 3.515502691268921,
      "learning_rate": 1.5279985578735187e-07,
      "loss": 0.7084,
      "step": 54910
    },
    {
      "epoch": 0.8549277246296715,
      "grad_norm": 2.712709665298462,
      "learning_rate": 1.5271791678274692e-07,
      "loss": 0.7402,
      "step": 54915
    },
    {
      "epoch": 0.8550055656316408,
      "grad_norm": 4.305067539215088,
      "learning_rate": 1.5263597777814194e-07,
      "loss": 0.7404,
      "step": 54920
    },
    {
      "epoch": 0.8550834066336102,
      "grad_norm": 3.761936664581299,
      "learning_rate": 1.52554038773537e-07,
      "loss": 0.7469,
      "step": 54925
    },
    {
      "epoch": 0.8551612476355795,
      "grad_norm": 4.037980556488037,
      "learning_rate": 1.52472099768932e-07,
      "loss": 0.7569,
      "step": 54930
    },
    {
      "epoch": 0.8552390886375489,
      "grad_norm": 2.856191873550415,
      "learning_rate": 1.5239016076432704e-07,
      "loss": 0.6764,
      "step": 54935
    },
    {
      "epoch": 0.8553169296395183,
      "grad_norm": 6.134860515594482,
      "learning_rate": 1.5230822175972206e-07,
      "loss": 0.7036,
      "step": 54940
    },
    {
      "epoch": 0.8553947706414877,
      "grad_norm": 2.786863088607788,
      "learning_rate": 1.5222628275511708e-07,
      "loss": 0.7487,
      "step": 54945
    },
    {
      "epoch": 0.8554726116434571,
      "grad_norm": 2.915177822113037,
      "learning_rate": 1.521443437505121e-07,
      "loss": 0.7316,
      "step": 54950
    },
    {
      "epoch": 0.8555504526454265,
      "grad_norm": 4.435000419616699,
      "learning_rate": 1.5206240474590715e-07,
      "loss": 0.6013,
      "step": 54955
    },
    {
      "epoch": 0.8556282936473958,
      "grad_norm": 3.174145221710205,
      "learning_rate": 1.5198046574130217e-07,
      "loss": 0.7916,
      "step": 54960
    },
    {
      "epoch": 0.8557061346493652,
      "grad_norm": 3.1707890033721924,
      "learning_rate": 1.518985267366972e-07,
      "loss": 0.761,
      "step": 54965
    },
    {
      "epoch": 0.8557839756513346,
      "grad_norm": 3.182654619216919,
      "learning_rate": 1.5181658773209222e-07,
      "loss": 0.7774,
      "step": 54970
    },
    {
      "epoch": 0.855861816653304,
      "grad_norm": 4.301407337188721,
      "learning_rate": 1.5173464872748727e-07,
      "loss": 0.6817,
      "step": 54975
    },
    {
      "epoch": 0.8559396576552734,
      "grad_norm": 3.3262157440185547,
      "learning_rate": 1.5165270972288226e-07,
      "loss": 0.6745,
      "step": 54980
    },
    {
      "epoch": 0.8560174986572427,
      "grad_norm": 3.3939924240112305,
      "learning_rate": 1.515707707182773e-07,
      "loss": 0.7971,
      "step": 54985
    },
    {
      "epoch": 0.8560953396592121,
      "grad_norm": 3.500687599182129,
      "learning_rate": 1.5148883171367233e-07,
      "loss": 0.6598,
      "step": 54990
    },
    {
      "epoch": 0.8561731806611814,
      "grad_norm": 4.436592102050781,
      "learning_rate": 1.5140689270906738e-07,
      "loss": 0.7623,
      "step": 54995
    },
    {
      "epoch": 0.8562510216631508,
      "grad_norm": 2.8300623893737793,
      "learning_rate": 1.5132495370446237e-07,
      "loss": 0.7736,
      "step": 55000
    },
    {
      "epoch": 0.8563288626651202,
      "grad_norm": 4.101807594299316,
      "learning_rate": 1.5124301469985742e-07,
      "loss": 0.6385,
      "step": 55005
    },
    {
      "epoch": 0.8564067036670896,
      "grad_norm": 4.399646759033203,
      "learning_rate": 1.5116107569525245e-07,
      "loss": 0.729,
      "step": 55010
    },
    {
      "epoch": 0.856484544669059,
      "grad_norm": 7.393691062927246,
      "learning_rate": 1.5107913669064747e-07,
      "loss": 0.8302,
      "step": 55015
    },
    {
      "epoch": 0.8565623856710284,
      "grad_norm": 4.392599582672119,
      "learning_rate": 1.509971976860425e-07,
      "loss": 0.6555,
      "step": 55020
    },
    {
      "epoch": 0.8566402266729978,
      "grad_norm": 7.756677627563477,
      "learning_rate": 1.5091525868143754e-07,
      "loss": 0.7916,
      "step": 55025
    },
    {
      "epoch": 0.8567180676749672,
      "grad_norm": 3.316840887069702,
      "learning_rate": 1.5083331967683256e-07,
      "loss": 0.6954,
      "step": 55030
    },
    {
      "epoch": 0.8567959086769364,
      "grad_norm": 4.337269306182861,
      "learning_rate": 1.5075138067222758e-07,
      "loss": 0.8035,
      "step": 55035
    },
    {
      "epoch": 0.8568737496789058,
      "grad_norm": 2.6424598693847656,
      "learning_rate": 1.506694416676226e-07,
      "loss": 0.6411,
      "step": 55040
    },
    {
      "epoch": 0.8569515906808752,
      "grad_norm": 11.35302734375,
      "learning_rate": 1.5058750266301765e-07,
      "loss": 0.6642,
      "step": 55045
    },
    {
      "epoch": 0.8570294316828446,
      "grad_norm": 3.1498641967773438,
      "learning_rate": 1.5050556365841265e-07,
      "loss": 0.6677,
      "step": 55050
    },
    {
      "epoch": 0.857107272684814,
      "grad_norm": 3.5982506275177,
      "learning_rate": 1.504236246538077e-07,
      "loss": 0.7919,
      "step": 55055
    },
    {
      "epoch": 0.8571851136867834,
      "grad_norm": 8.383096694946289,
      "learning_rate": 1.5034168564920272e-07,
      "loss": 0.6853,
      "step": 55060
    },
    {
      "epoch": 0.8572629546887528,
      "grad_norm": 4.44288444519043,
      "learning_rate": 1.5025974664459777e-07,
      "loss": 0.8215,
      "step": 55065
    },
    {
      "epoch": 0.8573407956907221,
      "grad_norm": 2.9935104846954346,
      "learning_rate": 1.5017780763999276e-07,
      "loss": 0.6519,
      "step": 55070
    },
    {
      "epoch": 0.8574186366926915,
      "grad_norm": 3.2087700366973877,
      "learning_rate": 1.500958686353878e-07,
      "loss": 0.6943,
      "step": 55075
    },
    {
      "epoch": 0.8574964776946609,
      "grad_norm": 2.6927194595336914,
      "learning_rate": 1.5001392963078283e-07,
      "loss": 0.6898,
      "step": 55080
    },
    {
      "epoch": 0.8575743186966303,
      "grad_norm": 4.0809760093688965,
      "learning_rate": 1.4993199062617788e-07,
      "loss": 0.6692,
      "step": 55085
    },
    {
      "epoch": 0.8576521596985996,
      "grad_norm": 2.8808672428131104,
      "learning_rate": 1.4985005162157288e-07,
      "loss": 0.6391,
      "step": 55090
    },
    {
      "epoch": 0.857730000700569,
      "grad_norm": 5.073269367218018,
      "learning_rate": 1.4976811261696792e-07,
      "loss": 0.6768,
      "step": 55095
    },
    {
      "epoch": 0.8578078417025384,
      "grad_norm": 2.849541187286377,
      "learning_rate": 1.4968617361236295e-07,
      "loss": 0.6739,
      "step": 55100
    },
    {
      "epoch": 0.8578856827045078,
      "grad_norm": 4.765288829803467,
      "learning_rate": 1.4960423460775797e-07,
      "loss": 0.7078,
      "step": 55105
    },
    {
      "epoch": 0.8579635237064771,
      "grad_norm": 3.6573383808135986,
      "learning_rate": 1.49522295603153e-07,
      "loss": 0.7987,
      "step": 55110
    },
    {
      "epoch": 0.8580413647084465,
      "grad_norm": 3.339582920074463,
      "learning_rate": 1.4944035659854804e-07,
      "loss": 0.7805,
      "step": 55115
    },
    {
      "epoch": 0.8581192057104159,
      "grad_norm": 8.18569564819336,
      "learning_rate": 1.4935841759394309e-07,
      "loss": 0.7423,
      "step": 55120
    },
    {
      "epoch": 0.8581970467123853,
      "grad_norm": 5.854790687561035,
      "learning_rate": 1.4927647858933808e-07,
      "loss": 0.7654,
      "step": 55125
    },
    {
      "epoch": 0.8582748877143547,
      "grad_norm": 5.12091064453125,
      "learning_rate": 1.4919453958473313e-07,
      "loss": 0.5921,
      "step": 55130
    },
    {
      "epoch": 0.8583527287163241,
      "grad_norm": 4.525527000427246,
      "learning_rate": 1.4911260058012815e-07,
      "loss": 0.7328,
      "step": 55135
    },
    {
      "epoch": 0.8584305697182935,
      "grad_norm": 2.7544233798980713,
      "learning_rate": 1.4903066157552317e-07,
      "loss": 0.7947,
      "step": 55140
    },
    {
      "epoch": 0.8585084107202627,
      "grad_norm": 4.7305755615234375,
      "learning_rate": 1.489487225709182e-07,
      "loss": 0.7495,
      "step": 55145
    },
    {
      "epoch": 0.8585862517222321,
      "grad_norm": 5.8436279296875,
      "learning_rate": 1.4886678356631324e-07,
      "loss": 0.7891,
      "step": 55150
    },
    {
      "epoch": 0.8586640927242015,
      "grad_norm": 3.050924301147461,
      "learning_rate": 1.4878484456170827e-07,
      "loss": 0.7154,
      "step": 55155
    },
    {
      "epoch": 0.8587419337261709,
      "grad_norm": 9.049485206604004,
      "learning_rate": 1.487029055571033e-07,
      "loss": 0.7131,
      "step": 55160
    },
    {
      "epoch": 0.8588197747281403,
      "grad_norm": 3.4681451320648193,
      "learning_rate": 1.486209665524983e-07,
      "loss": 0.7406,
      "step": 55165
    },
    {
      "epoch": 0.8588976157301097,
      "grad_norm": 4.826496601104736,
      "learning_rate": 1.4853902754789336e-07,
      "loss": 0.6703,
      "step": 55170
    },
    {
      "epoch": 0.8589754567320791,
      "grad_norm": 5.596420764923096,
      "learning_rate": 1.4845708854328835e-07,
      "loss": 0.7913,
      "step": 55175
    },
    {
      "epoch": 0.8590532977340485,
      "grad_norm": 4.791665554046631,
      "learning_rate": 1.483751495386834e-07,
      "loss": 0.7972,
      "step": 55180
    },
    {
      "epoch": 0.8591311387360178,
      "grad_norm": 3.8565292358398438,
      "learning_rate": 1.4829321053407842e-07,
      "loss": 0.7427,
      "step": 55185
    },
    {
      "epoch": 0.8592089797379872,
      "grad_norm": 3.639578104019165,
      "learning_rate": 1.4821127152947347e-07,
      "loss": 0.7316,
      "step": 55190
    },
    {
      "epoch": 0.8592868207399565,
      "grad_norm": 2.4290826320648193,
      "learning_rate": 1.4812933252486847e-07,
      "loss": 0.6623,
      "step": 55195
    },
    {
      "epoch": 0.8593646617419259,
      "grad_norm": 2.4454522132873535,
      "learning_rate": 1.4804739352026352e-07,
      "loss": 0.704,
      "step": 55200
    },
    {
      "epoch": 0.8594425027438953,
      "grad_norm": 4.119056224822998,
      "learning_rate": 1.4796545451565854e-07,
      "loss": 0.6869,
      "step": 55205
    },
    {
      "epoch": 0.8595203437458647,
      "grad_norm": 3.544072389602661,
      "learning_rate": 1.4788351551105359e-07,
      "loss": 0.6672,
      "step": 55210
    },
    {
      "epoch": 0.8595981847478341,
      "grad_norm": 2.9093313217163086,
      "learning_rate": 1.4780157650644858e-07,
      "loss": 0.7902,
      "step": 55215
    },
    {
      "epoch": 0.8596760257498034,
      "grad_norm": 3.992914915084839,
      "learning_rate": 1.4771963750184363e-07,
      "loss": 0.8291,
      "step": 55220
    },
    {
      "epoch": 0.8597538667517728,
      "grad_norm": 3.1128478050231934,
      "learning_rate": 1.4763769849723865e-07,
      "loss": 0.8059,
      "step": 55225
    },
    {
      "epoch": 0.8598317077537422,
      "grad_norm": 4.601522445678711,
      "learning_rate": 1.4755575949263367e-07,
      "loss": 0.6242,
      "step": 55230
    },
    {
      "epoch": 0.8599095487557116,
      "grad_norm": 2.4590370655059814,
      "learning_rate": 1.474738204880287e-07,
      "loss": 0.7507,
      "step": 55235
    },
    {
      "epoch": 0.859987389757681,
      "grad_norm": 6.902151107788086,
      "learning_rate": 1.4739188148342374e-07,
      "loss": 0.6838,
      "step": 55240
    },
    {
      "epoch": 0.8600652307596504,
      "grad_norm": 4.296189785003662,
      "learning_rate": 1.4730994247881877e-07,
      "loss": 0.6771,
      "step": 55245
    },
    {
      "epoch": 0.8601430717616197,
      "grad_norm": 3.938120126724243,
      "learning_rate": 1.472280034742138e-07,
      "loss": 0.734,
      "step": 55250
    },
    {
      "epoch": 0.8602209127635891,
      "grad_norm": 3.3813283443450928,
      "learning_rate": 1.471460644696088e-07,
      "loss": 0.7385,
      "step": 55255
    },
    {
      "epoch": 0.8602987537655584,
      "grad_norm": 4.025691032409668,
      "learning_rate": 1.4706412546500386e-07,
      "loss": 0.723,
      "step": 55260
    },
    {
      "epoch": 0.8603765947675278,
      "grad_norm": 2.9535412788391113,
      "learning_rate": 1.4698218646039885e-07,
      "loss": 0.6438,
      "step": 55265
    },
    {
      "epoch": 0.8604544357694972,
      "grad_norm": 2.508185863494873,
      "learning_rate": 1.469002474557939e-07,
      "loss": 0.6207,
      "step": 55270
    },
    {
      "epoch": 0.8605322767714666,
      "grad_norm": 3.4419097900390625,
      "learning_rate": 1.4681830845118892e-07,
      "loss": 0.7637,
      "step": 55275
    },
    {
      "epoch": 0.860610117773436,
      "grad_norm": 3.1141574382781982,
      "learning_rate": 1.4673636944658397e-07,
      "loss": 0.7159,
      "step": 55280
    },
    {
      "epoch": 0.8606879587754054,
      "grad_norm": 3.5909080505371094,
      "learning_rate": 1.4665443044197897e-07,
      "loss": 0.715,
      "step": 55285
    },
    {
      "epoch": 0.8607657997773748,
      "grad_norm": 2.7459375858306885,
      "learning_rate": 1.4657249143737402e-07,
      "loss": 0.7133,
      "step": 55290
    },
    {
      "epoch": 0.8608436407793442,
      "grad_norm": 4.287467002868652,
      "learning_rate": 1.4649055243276904e-07,
      "loss": 0.7388,
      "step": 55295
    },
    {
      "epoch": 0.8609214817813134,
      "grad_norm": 4.2406158447265625,
      "learning_rate": 1.4640861342816406e-07,
      "loss": 0.7688,
      "step": 55300
    },
    {
      "epoch": 0.8609993227832828,
      "grad_norm": 8.215710639953613,
      "learning_rate": 1.4632667442355908e-07,
      "loss": 0.7703,
      "step": 55305
    },
    {
      "epoch": 0.8610771637852522,
      "grad_norm": 3.0250697135925293,
      "learning_rate": 1.4624473541895413e-07,
      "loss": 0.5667,
      "step": 55310
    },
    {
      "epoch": 0.8611550047872216,
      "grad_norm": 3.3984742164611816,
      "learning_rate": 1.4616279641434915e-07,
      "loss": 0.6202,
      "step": 55315
    },
    {
      "epoch": 0.861232845789191,
      "grad_norm": 5.944708824157715,
      "learning_rate": 1.4608085740974417e-07,
      "loss": 0.8236,
      "step": 55320
    },
    {
      "epoch": 0.8613106867911604,
      "grad_norm": 4.218142509460449,
      "learning_rate": 1.459989184051392e-07,
      "loss": 0.7116,
      "step": 55325
    },
    {
      "epoch": 0.8613885277931298,
      "grad_norm": 3.9960126876831055,
      "learning_rate": 1.4591697940053425e-07,
      "loss": 0.7496,
      "step": 55330
    },
    {
      "epoch": 0.8614663687950991,
      "grad_norm": 3.8918092250823975,
      "learning_rate": 1.4583504039592927e-07,
      "loss": 0.7922,
      "step": 55335
    },
    {
      "epoch": 0.8615442097970685,
      "grad_norm": 9.258756637573242,
      "learning_rate": 1.457531013913243e-07,
      "loss": 0.7862,
      "step": 55340
    },
    {
      "epoch": 0.8616220507990379,
      "grad_norm": 2.9384982585906982,
      "learning_rate": 1.456711623867193e-07,
      "loss": 0.8887,
      "step": 55345
    },
    {
      "epoch": 0.8616998918010073,
      "grad_norm": 3.4544479846954346,
      "learning_rate": 1.4558922338211436e-07,
      "loss": 0.7275,
      "step": 55350
    },
    {
      "epoch": 0.8617777328029766,
      "grad_norm": 3.053607940673828,
      "learning_rate": 1.4550728437750935e-07,
      "loss": 0.8091,
      "step": 55355
    },
    {
      "epoch": 0.861855573804946,
      "grad_norm": 4.241771697998047,
      "learning_rate": 1.454253453729044e-07,
      "loss": 0.6841,
      "step": 55360
    },
    {
      "epoch": 0.8619334148069154,
      "grad_norm": 2.811798572540283,
      "learning_rate": 1.4534340636829943e-07,
      "loss": 0.8262,
      "step": 55365
    },
    {
      "epoch": 0.8620112558088848,
      "grad_norm": 3.310666799545288,
      "learning_rate": 1.4526146736369447e-07,
      "loss": 0.741,
      "step": 55370
    },
    {
      "epoch": 0.8620890968108541,
      "grad_norm": 2.9058854579925537,
      "learning_rate": 1.4517952835908947e-07,
      "loss": 0.6889,
      "step": 55375
    },
    {
      "epoch": 0.8621669378128235,
      "grad_norm": 2.676476240158081,
      "learning_rate": 1.4509758935448452e-07,
      "loss": 0.7689,
      "step": 55380
    },
    {
      "epoch": 0.8622447788147929,
      "grad_norm": 3.743199110031128,
      "learning_rate": 1.4501565034987954e-07,
      "loss": 0.7623,
      "step": 55385
    },
    {
      "epoch": 0.8623226198167623,
      "grad_norm": 2.800229549407959,
      "learning_rate": 1.4493371134527456e-07,
      "loss": 0.8164,
      "step": 55390
    },
    {
      "epoch": 0.8624004608187317,
      "grad_norm": 3.6309211254119873,
      "learning_rate": 1.4485177234066958e-07,
      "loss": 0.7098,
      "step": 55395
    },
    {
      "epoch": 0.8624783018207011,
      "grad_norm": 4.40344762802124,
      "learning_rate": 1.4476983333606463e-07,
      "loss": 0.7083,
      "step": 55400
    },
    {
      "epoch": 0.8625561428226705,
      "grad_norm": 3.641892671585083,
      "learning_rate": 1.4468789433145965e-07,
      "loss": 0.7767,
      "step": 55405
    },
    {
      "epoch": 0.8626339838246397,
      "grad_norm": 3.8605992794036865,
      "learning_rate": 1.4460595532685468e-07,
      "loss": 0.6872,
      "step": 55410
    },
    {
      "epoch": 0.8627118248266091,
      "grad_norm": 3.064389705657959,
      "learning_rate": 1.445240163222497e-07,
      "loss": 0.8056,
      "step": 55415
    },
    {
      "epoch": 0.8627896658285785,
      "grad_norm": 4.730130195617676,
      "learning_rate": 1.4444207731764475e-07,
      "loss": 0.7304,
      "step": 55420
    },
    {
      "epoch": 0.8628675068305479,
      "grad_norm": 2.750380516052246,
      "learning_rate": 1.4436013831303974e-07,
      "loss": 0.6293,
      "step": 55425
    },
    {
      "epoch": 0.8629453478325173,
      "grad_norm": 3.466254711151123,
      "learning_rate": 1.442781993084348e-07,
      "loss": 0.7921,
      "step": 55430
    },
    {
      "epoch": 0.8630231888344867,
      "grad_norm": 2.6578359603881836,
      "learning_rate": 1.4419626030382984e-07,
      "loss": 0.7064,
      "step": 55435
    },
    {
      "epoch": 0.8631010298364561,
      "grad_norm": 2.4797942638397217,
      "learning_rate": 1.4411432129922486e-07,
      "loss": 0.7845,
      "step": 55440
    },
    {
      "epoch": 0.8631788708384255,
      "grad_norm": 3.4206879138946533,
      "learning_rate": 1.4403238229461988e-07,
      "loss": 0.6868,
      "step": 55445
    },
    {
      "epoch": 0.8632567118403948,
      "grad_norm": 2.9791643619537354,
      "learning_rate": 1.439504432900149e-07,
      "loss": 0.7328,
      "step": 55450
    },
    {
      "epoch": 0.8633345528423642,
      "grad_norm": 3.4249978065490723,
      "learning_rate": 1.4386850428540995e-07,
      "loss": 0.6485,
      "step": 55455
    },
    {
      "epoch": 0.8634123938443335,
      "grad_norm": 3.370222568511963,
      "learning_rate": 1.4378656528080497e-07,
      "loss": 0.7643,
      "step": 55460
    },
    {
      "epoch": 0.8634902348463029,
      "grad_norm": 6.137143611907959,
      "learning_rate": 1.437046262762e-07,
      "loss": 0.7276,
      "step": 55465
    },
    {
      "epoch": 0.8635680758482723,
      "grad_norm": 3.5836846828460693,
      "learning_rate": 1.4362268727159502e-07,
      "loss": 0.6551,
      "step": 55470
    },
    {
      "epoch": 0.8636459168502417,
      "grad_norm": 3.456150531768799,
      "learning_rate": 1.4354074826699007e-07,
      "loss": 0.7677,
      "step": 55475
    },
    {
      "epoch": 0.8637237578522111,
      "grad_norm": 4.712547779083252,
      "learning_rate": 1.4345880926238506e-07,
      "loss": 0.7032,
      "step": 55480
    },
    {
      "epoch": 0.8638015988541804,
      "grad_norm": 3.1393394470214844,
      "learning_rate": 1.433768702577801e-07,
      "loss": 0.6682,
      "step": 55485
    },
    {
      "epoch": 0.8638794398561498,
      "grad_norm": 3.7645275592803955,
      "learning_rate": 1.4329493125317513e-07,
      "loss": 0.7393,
      "step": 55490
    },
    {
      "epoch": 0.8639572808581192,
      "grad_norm": 5.903180122375488,
      "learning_rate": 1.4321299224857018e-07,
      "loss": 0.7995,
      "step": 55495
    },
    {
      "epoch": 0.8640351218600886,
      "grad_norm": 10.898876190185547,
      "learning_rate": 1.4313105324396518e-07,
      "loss": 0.684,
      "step": 55500
    },
    {
      "epoch": 0.864112962862058,
      "grad_norm": 3.0385472774505615,
      "learning_rate": 1.4304911423936022e-07,
      "loss": 0.647,
      "step": 55505
    },
    {
      "epoch": 0.8641908038640274,
      "grad_norm": 3.985912322998047,
      "learning_rate": 1.4296717523475525e-07,
      "loss": 0.842,
      "step": 55510
    },
    {
      "epoch": 0.8642686448659967,
      "grad_norm": 4.489705562591553,
      "learning_rate": 1.4288523623015027e-07,
      "loss": 0.7235,
      "step": 55515
    },
    {
      "epoch": 0.8643464858679661,
      "grad_norm": 4.359413146972656,
      "learning_rate": 1.428032972255453e-07,
      "loss": 0.7183,
      "step": 55520
    },
    {
      "epoch": 0.8644243268699354,
      "grad_norm": 3.4856655597686768,
      "learning_rate": 1.4272135822094034e-07,
      "loss": 0.6313,
      "step": 55525
    },
    {
      "epoch": 0.8645021678719048,
      "grad_norm": 4.051481246948242,
      "learning_rate": 1.4263941921633536e-07,
      "loss": 0.822,
      "step": 55530
    },
    {
      "epoch": 0.8645800088738742,
      "grad_norm": 5.549462795257568,
      "learning_rate": 1.4255748021173038e-07,
      "loss": 0.5755,
      "step": 55535
    },
    {
      "epoch": 0.8646578498758436,
      "grad_norm": 3.6000595092773438,
      "learning_rate": 1.424755412071254e-07,
      "loss": 0.7372,
      "step": 55540
    },
    {
      "epoch": 0.864735690877813,
      "grad_norm": 6.632894515991211,
      "learning_rate": 1.4239360220252045e-07,
      "loss": 0.6787,
      "step": 55545
    },
    {
      "epoch": 0.8648135318797824,
      "grad_norm": 3.027646541595459,
      "learning_rate": 1.4231166319791545e-07,
      "loss": 0.6883,
      "step": 55550
    },
    {
      "epoch": 0.8648913728817518,
      "grad_norm": 4.2080078125,
      "learning_rate": 1.422297241933105e-07,
      "loss": 0.6912,
      "step": 55555
    },
    {
      "epoch": 0.8649692138837212,
      "grad_norm": 3.200435161590576,
      "learning_rate": 1.4214778518870552e-07,
      "loss": 0.8039,
      "step": 55560
    },
    {
      "epoch": 0.8650470548856904,
      "grad_norm": 3.3274810314178467,
      "learning_rate": 1.4206584618410057e-07,
      "loss": 0.6846,
      "step": 55565
    },
    {
      "epoch": 0.8651248958876598,
      "grad_norm": 4.982847690582275,
      "learning_rate": 1.4198390717949556e-07,
      "loss": 0.7916,
      "step": 55570
    },
    {
      "epoch": 0.8652027368896292,
      "grad_norm": 3.6036062240600586,
      "learning_rate": 1.419019681748906e-07,
      "loss": 0.839,
      "step": 55575
    },
    {
      "epoch": 0.8652805778915986,
      "grad_norm": 9.392388343811035,
      "learning_rate": 1.4182002917028563e-07,
      "loss": 0.694,
      "step": 55580
    },
    {
      "epoch": 0.865358418893568,
      "grad_norm": 3.558556318283081,
      "learning_rate": 1.4173809016568068e-07,
      "loss": 0.6257,
      "step": 55585
    },
    {
      "epoch": 0.8654362598955374,
      "grad_norm": 3.5037834644317627,
      "learning_rate": 1.4165615116107568e-07,
      "loss": 0.7595,
      "step": 55590
    },
    {
      "epoch": 0.8655141008975068,
      "grad_norm": 3.615809440612793,
      "learning_rate": 1.4157421215647072e-07,
      "loss": 0.7004,
      "step": 55595
    },
    {
      "epoch": 0.8655919418994761,
      "grad_norm": 4.321993350982666,
      "learning_rate": 1.4149227315186575e-07,
      "loss": 0.6977,
      "step": 55600
    },
    {
      "epoch": 0.8656697829014455,
      "grad_norm": 4.385680675506592,
      "learning_rate": 1.4141033414726077e-07,
      "loss": 0.8084,
      "step": 55605
    },
    {
      "epoch": 0.8657476239034149,
      "grad_norm": 3.3422958850860596,
      "learning_rate": 1.413283951426558e-07,
      "loss": 0.704,
      "step": 55610
    },
    {
      "epoch": 0.8658254649053843,
      "grad_norm": 2.716703176498413,
      "learning_rate": 1.4124645613805084e-07,
      "loss": 0.8325,
      "step": 55615
    },
    {
      "epoch": 0.8659033059073536,
      "grad_norm": 4.71327543258667,
      "learning_rate": 1.4116451713344586e-07,
      "loss": 0.7536,
      "step": 55620
    },
    {
      "epoch": 0.865981146909323,
      "grad_norm": 2.574082374572754,
      "learning_rate": 1.4108257812884088e-07,
      "loss": 0.7021,
      "step": 55625
    },
    {
      "epoch": 0.8660589879112924,
      "grad_norm": 2.3839879035949707,
      "learning_rate": 1.410006391242359e-07,
      "loss": 0.7161,
      "step": 55630
    },
    {
      "epoch": 0.8661368289132618,
      "grad_norm": 4.1667046546936035,
      "learning_rate": 1.4091870011963095e-07,
      "loss": 0.829,
      "step": 55635
    },
    {
      "epoch": 0.8662146699152311,
      "grad_norm": 15.696386337280273,
      "learning_rate": 1.4083676111502595e-07,
      "loss": 0.8143,
      "step": 55640
    },
    {
      "epoch": 0.8662925109172005,
      "grad_norm": 4.161869049072266,
      "learning_rate": 1.40754822110421e-07,
      "loss": 0.6212,
      "step": 55645
    },
    {
      "epoch": 0.8663703519191699,
      "grad_norm": 3.5564358234405518,
      "learning_rate": 1.4067288310581602e-07,
      "loss": 0.7262,
      "step": 55650
    },
    {
      "epoch": 0.8664481929211393,
      "grad_norm": 5.27222204208374,
      "learning_rate": 1.4059094410121107e-07,
      "loss": 0.7477,
      "step": 55655
    },
    {
      "epoch": 0.8665260339231087,
      "grad_norm": 7.012271881103516,
      "learning_rate": 1.4050900509660606e-07,
      "loss": 0.7196,
      "step": 55660
    },
    {
      "epoch": 0.8666038749250781,
      "grad_norm": 2.7435522079467773,
      "learning_rate": 1.404270660920011e-07,
      "loss": 0.6745,
      "step": 55665
    },
    {
      "epoch": 0.8666817159270475,
      "grad_norm": 5.7318596839904785,
      "learning_rate": 1.4034512708739613e-07,
      "loss": 0.7199,
      "step": 55670
    },
    {
      "epoch": 0.8667595569290167,
      "grad_norm": 3.8594810962677,
      "learning_rate": 1.4026318808279115e-07,
      "loss": 0.6803,
      "step": 55675
    },
    {
      "epoch": 0.8668373979309861,
      "grad_norm": 2.8159642219543457,
      "learning_rate": 1.4018124907818618e-07,
      "loss": 0.8126,
      "step": 55680
    },
    {
      "epoch": 0.8669152389329555,
      "grad_norm": 3.5000338554382324,
      "learning_rate": 1.4009931007358122e-07,
      "loss": 0.6803,
      "step": 55685
    },
    {
      "epoch": 0.8669930799349249,
      "grad_norm": 4.227432727813721,
      "learning_rate": 1.4001737106897625e-07,
      "loss": 0.7505,
      "step": 55690
    },
    {
      "epoch": 0.8670709209368943,
      "grad_norm": 3.299581527709961,
      "learning_rate": 1.3993543206437127e-07,
      "loss": 0.7573,
      "step": 55695
    },
    {
      "epoch": 0.8671487619388637,
      "grad_norm": 5.235525608062744,
      "learning_rate": 1.398534930597663e-07,
      "loss": 0.7684,
      "step": 55700
    },
    {
      "epoch": 0.8672266029408331,
      "grad_norm": 3.0780296325683594,
      "learning_rate": 1.3977155405516134e-07,
      "loss": 0.6887,
      "step": 55705
    },
    {
      "epoch": 0.8673044439428025,
      "grad_norm": 2.9017231464385986,
      "learning_rate": 1.3968961505055633e-07,
      "loss": 0.8499,
      "step": 55710
    },
    {
      "epoch": 0.8673822849447718,
      "grad_norm": 3.424647808074951,
      "learning_rate": 1.3960767604595138e-07,
      "loss": 0.7735,
      "step": 55715
    },
    {
      "epoch": 0.8674601259467412,
      "grad_norm": 4.507081508636475,
      "learning_rate": 1.395257370413464e-07,
      "loss": 0.8214,
      "step": 55720
    },
    {
      "epoch": 0.8675379669487105,
      "grad_norm": 3.0748445987701416,
      "learning_rate": 1.3944379803674145e-07,
      "loss": 0.656,
      "step": 55725
    },
    {
      "epoch": 0.8676158079506799,
      "grad_norm": 2.9701836109161377,
      "learning_rate": 1.3936185903213645e-07,
      "loss": 0.7125,
      "step": 55730
    },
    {
      "epoch": 0.8676936489526493,
      "grad_norm": 6.473572254180908,
      "learning_rate": 1.392799200275315e-07,
      "loss": 0.7031,
      "step": 55735
    },
    {
      "epoch": 0.8677714899546187,
      "grad_norm": 4.237670421600342,
      "learning_rate": 1.3919798102292652e-07,
      "loss": 0.7468,
      "step": 55740
    },
    {
      "epoch": 0.8678493309565881,
      "grad_norm": 4.6277756690979,
      "learning_rate": 1.3911604201832157e-07,
      "loss": 0.7747,
      "step": 55745
    },
    {
      "epoch": 0.8679271719585574,
      "grad_norm": 3.621310234069824,
      "learning_rate": 1.3903410301371656e-07,
      "loss": 0.7883,
      "step": 55750
    },
    {
      "epoch": 0.8680050129605268,
      "grad_norm": 3.7673890590667725,
      "learning_rate": 1.389521640091116e-07,
      "loss": 0.7795,
      "step": 55755
    },
    {
      "epoch": 0.8680828539624962,
      "grad_norm": 2.9341235160827637,
      "learning_rate": 1.3887022500450666e-07,
      "loss": 0.7364,
      "step": 55760
    },
    {
      "epoch": 0.8681606949644656,
      "grad_norm": 3.802682399749756,
      "learning_rate": 1.3878828599990166e-07,
      "loss": 0.6582,
      "step": 55765
    },
    {
      "epoch": 0.868238535966435,
      "grad_norm": 3.1986570358276367,
      "learning_rate": 1.387063469952967e-07,
      "loss": 0.6337,
      "step": 55770
    },
    {
      "epoch": 0.8683163769684044,
      "grad_norm": 3.5000908374786377,
      "learning_rate": 1.3862440799069173e-07,
      "loss": 0.7816,
      "step": 55775
    },
    {
      "epoch": 0.8683942179703737,
      "grad_norm": 2.4482829570770264,
      "learning_rate": 1.3854246898608677e-07,
      "loss": 0.6973,
      "step": 55780
    },
    {
      "epoch": 0.8684720589723431,
      "grad_norm": 4.365492343902588,
      "learning_rate": 1.3846052998148177e-07,
      "loss": 0.6905,
      "step": 55785
    },
    {
      "epoch": 0.8685498999743124,
      "grad_norm": 3.673154354095459,
      "learning_rate": 1.3837859097687682e-07,
      "loss": 0.6773,
      "step": 55790
    },
    {
      "epoch": 0.8686277409762818,
      "grad_norm": 3.0079965591430664,
      "learning_rate": 1.3829665197227184e-07,
      "loss": 0.8333,
      "step": 55795
    },
    {
      "epoch": 0.8687055819782512,
      "grad_norm": 2.6952285766601562,
      "learning_rate": 1.3821471296766686e-07,
      "loss": 0.7482,
      "step": 55800
    },
    {
      "epoch": 0.8687834229802206,
      "grad_norm": 8.179831504821777,
      "learning_rate": 1.3813277396306188e-07,
      "loss": 0.7373,
      "step": 55805
    },
    {
      "epoch": 0.86886126398219,
      "grad_norm": 3.2923731803894043,
      "learning_rate": 1.3805083495845693e-07,
      "loss": 0.6158,
      "step": 55810
    },
    {
      "epoch": 0.8689391049841594,
      "grad_norm": 3.4821431636810303,
      "learning_rate": 1.3796889595385195e-07,
      "loss": 0.6749,
      "step": 55815
    },
    {
      "epoch": 0.8690169459861288,
      "grad_norm": 4.0804033279418945,
      "learning_rate": 1.3788695694924698e-07,
      "loss": 0.7028,
      "step": 55820
    },
    {
      "epoch": 0.8690947869880981,
      "grad_norm": 4.053691387176514,
      "learning_rate": 1.37805017944642e-07,
      "loss": 0.7129,
      "step": 55825
    },
    {
      "epoch": 0.8691726279900674,
      "grad_norm": 4.305898666381836,
      "learning_rate": 1.3772307894003705e-07,
      "loss": 0.7755,
      "step": 55830
    },
    {
      "epoch": 0.8692504689920368,
      "grad_norm": 3.986853837966919,
      "learning_rate": 1.3764113993543204e-07,
      "loss": 0.7348,
      "step": 55835
    },
    {
      "epoch": 0.8693283099940062,
      "grad_norm": 3.5115814208984375,
      "learning_rate": 1.375592009308271e-07,
      "loss": 0.7578,
      "step": 55840
    },
    {
      "epoch": 0.8694061509959756,
      "grad_norm": 4.366875171661377,
      "learning_rate": 1.374772619262221e-07,
      "loss": 0.653,
      "step": 55845
    },
    {
      "epoch": 0.869483991997945,
      "grad_norm": 2.1414430141448975,
      "learning_rate": 1.3739532292161716e-07,
      "loss": 0.7263,
      "step": 55850
    },
    {
      "epoch": 0.8695618329999144,
      "grad_norm": 3.946572780609131,
      "learning_rate": 1.3731338391701216e-07,
      "loss": 0.6897,
      "step": 55855
    },
    {
      "epoch": 0.8696396740018838,
      "grad_norm": 2.8170411586761475,
      "learning_rate": 1.372314449124072e-07,
      "loss": 0.7115,
      "step": 55860
    },
    {
      "epoch": 0.8697175150038531,
      "grad_norm": 3.888749837875366,
      "learning_rate": 1.3714950590780223e-07,
      "loss": 0.8184,
      "step": 55865
    },
    {
      "epoch": 0.8697953560058225,
      "grad_norm": 4.11981725692749,
      "learning_rate": 1.3706756690319727e-07,
      "loss": 0.7687,
      "step": 55870
    },
    {
      "epoch": 0.8698731970077919,
      "grad_norm": 4.707082271575928,
      "learning_rate": 1.3698562789859227e-07,
      "loss": 0.673,
      "step": 55875
    },
    {
      "epoch": 0.8699510380097613,
      "grad_norm": 2.8377814292907715,
      "learning_rate": 1.3690368889398732e-07,
      "loss": 0.7356,
      "step": 55880
    },
    {
      "epoch": 0.8700288790117306,
      "grad_norm": 4.334586143493652,
      "learning_rate": 1.3682174988938234e-07,
      "loss": 0.7811,
      "step": 55885
    },
    {
      "epoch": 0.8701067200137,
      "grad_norm": 5.477593421936035,
      "learning_rate": 1.3673981088477736e-07,
      "loss": 0.7421,
      "step": 55890
    },
    {
      "epoch": 0.8701845610156694,
      "grad_norm": 4.88681173324585,
      "learning_rate": 1.3665787188017238e-07,
      "loss": 0.8315,
      "step": 55895
    },
    {
      "epoch": 0.8702624020176388,
      "grad_norm": 4.551938533782959,
      "learning_rate": 1.3657593287556743e-07,
      "loss": 0.8182,
      "step": 55900
    },
    {
      "epoch": 0.8703402430196081,
      "grad_norm": 6.552649974822998,
      "learning_rate": 1.3649399387096245e-07,
      "loss": 0.675,
      "step": 55905
    },
    {
      "epoch": 0.8704180840215775,
      "grad_norm": 3.202580213546753,
      "learning_rate": 1.3641205486635748e-07,
      "loss": 0.8042,
      "step": 55910
    },
    {
      "epoch": 0.8704959250235469,
      "grad_norm": 4.294753074645996,
      "learning_rate": 1.363301158617525e-07,
      "loss": 0.9279,
      "step": 55915
    },
    {
      "epoch": 0.8705737660255163,
      "grad_norm": 7.415616035461426,
      "learning_rate": 1.3624817685714755e-07,
      "loss": 0.6931,
      "step": 55920
    },
    {
      "epoch": 0.8706516070274857,
      "grad_norm": 2.6882119178771973,
      "learning_rate": 1.3616623785254254e-07,
      "loss": 0.7415,
      "step": 55925
    },
    {
      "epoch": 0.8707294480294551,
      "grad_norm": 5.514176845550537,
      "learning_rate": 1.360842988479376e-07,
      "loss": 0.6791,
      "step": 55930
    },
    {
      "epoch": 0.8708072890314245,
      "grad_norm": 4.352840900421143,
      "learning_rate": 1.360023598433326e-07,
      "loss": 0.8148,
      "step": 55935
    },
    {
      "epoch": 0.8708851300333937,
      "grad_norm": 3.068570137023926,
      "learning_rate": 1.3592042083872766e-07,
      "loss": 0.7307,
      "step": 55940
    },
    {
      "epoch": 0.8709629710353631,
      "grad_norm": 3.8593661785125732,
      "learning_rate": 1.3583848183412266e-07,
      "loss": 0.6512,
      "step": 55945
    },
    {
      "epoch": 0.8710408120373325,
      "grad_norm": 3.224942207336426,
      "learning_rate": 1.357565428295177e-07,
      "loss": 0.6906,
      "step": 55950
    },
    {
      "epoch": 0.8711186530393019,
      "grad_norm": 8.91712474822998,
      "learning_rate": 1.3567460382491273e-07,
      "loss": 0.6612,
      "step": 55955
    },
    {
      "epoch": 0.8711964940412713,
      "grad_norm": 3.7196197509765625,
      "learning_rate": 1.3559266482030777e-07,
      "loss": 0.7928,
      "step": 55960
    },
    {
      "epoch": 0.8712743350432407,
      "grad_norm": 4.432784557342529,
      "learning_rate": 1.3551072581570277e-07,
      "loss": 0.6705,
      "step": 55965
    },
    {
      "epoch": 0.8713521760452101,
      "grad_norm": 3.6705241203308105,
      "learning_rate": 1.3542878681109782e-07,
      "loss": 0.7555,
      "step": 55970
    },
    {
      "epoch": 0.8714300170471795,
      "grad_norm": 4.986262798309326,
      "learning_rate": 1.3534684780649284e-07,
      "loss": 0.8563,
      "step": 55975
    },
    {
      "epoch": 0.8715078580491488,
      "grad_norm": 3.088235378265381,
      "learning_rate": 1.3526490880188786e-07,
      "loss": 0.8529,
      "step": 55980
    },
    {
      "epoch": 0.8715856990511182,
      "grad_norm": 3.2138454914093018,
      "learning_rate": 1.3518296979728288e-07,
      "loss": 0.7211,
      "step": 55985
    },
    {
      "epoch": 0.8716635400530875,
      "grad_norm": 5.268904209136963,
      "learning_rate": 1.3510103079267793e-07,
      "loss": 0.7576,
      "step": 55990
    },
    {
      "epoch": 0.8717413810550569,
      "grad_norm": 7.013116359710693,
      "learning_rate": 1.3501909178807295e-07,
      "loss": 0.676,
      "step": 55995
    },
    {
      "epoch": 0.8718192220570263,
      "grad_norm": 4.143415927886963,
      "learning_rate": 1.3493715278346798e-07,
      "loss": 0.8104,
      "step": 56000
    },
    {
      "epoch": 0.8718970630589957,
      "grad_norm": 3.097675085067749,
      "learning_rate": 1.34855213778863e-07,
      "loss": 0.6907,
      "step": 56005
    },
    {
      "epoch": 0.8719749040609651,
      "grad_norm": 3.2619833946228027,
      "learning_rate": 1.3477327477425805e-07,
      "loss": 0.7382,
      "step": 56010
    },
    {
      "epoch": 0.8720527450629344,
      "grad_norm": 6.240786552429199,
      "learning_rate": 1.3469133576965304e-07,
      "loss": 0.741,
      "step": 56015
    },
    {
      "epoch": 0.8721305860649038,
      "grad_norm": 4.893679141998291,
      "learning_rate": 1.346093967650481e-07,
      "loss": 0.682,
      "step": 56020
    },
    {
      "epoch": 0.8722084270668732,
      "grad_norm": 2.81005597114563,
      "learning_rate": 1.345274577604431e-07,
      "loss": 0.7455,
      "step": 56025
    },
    {
      "epoch": 0.8722862680688426,
      "grad_norm": 4.2380690574646,
      "learning_rate": 1.3444551875583816e-07,
      "loss": 0.6661,
      "step": 56030
    },
    {
      "epoch": 0.872364109070812,
      "grad_norm": 4.241702079772949,
      "learning_rate": 1.3436357975123316e-07,
      "loss": 0.6841,
      "step": 56035
    },
    {
      "epoch": 0.8724419500727814,
      "grad_norm": 9.392199516296387,
      "learning_rate": 1.342816407466282e-07,
      "loss": 0.7873,
      "step": 56040
    },
    {
      "epoch": 0.8725197910747508,
      "grad_norm": 3.9865970611572266,
      "learning_rate": 1.3419970174202323e-07,
      "loss": 0.6816,
      "step": 56045
    },
    {
      "epoch": 0.8725976320767201,
      "grad_norm": 3.2748987674713135,
      "learning_rate": 1.3411776273741825e-07,
      "loss": 0.6683,
      "step": 56050
    },
    {
      "epoch": 0.8726754730786894,
      "grad_norm": 4.261683940887451,
      "learning_rate": 1.3403582373281327e-07,
      "loss": 0.7483,
      "step": 56055
    },
    {
      "epoch": 0.8727533140806588,
      "grad_norm": 4.092185974121094,
      "learning_rate": 1.3395388472820832e-07,
      "loss": 0.7365,
      "step": 56060
    },
    {
      "epoch": 0.8728311550826282,
      "grad_norm": 2.9737579822540283,
      "learning_rate": 1.3387194572360334e-07,
      "loss": 0.7187,
      "step": 56065
    },
    {
      "epoch": 0.8729089960845976,
      "grad_norm": 3.765455722808838,
      "learning_rate": 1.3379000671899836e-07,
      "loss": 0.7089,
      "step": 56070
    },
    {
      "epoch": 0.872986837086567,
      "grad_norm": 6.8706841468811035,
      "learning_rate": 1.337080677143934e-07,
      "loss": 0.6308,
      "step": 56075
    },
    {
      "epoch": 0.8730646780885364,
      "grad_norm": 3.184854745864868,
      "learning_rate": 1.3362612870978843e-07,
      "loss": 0.8605,
      "step": 56080
    },
    {
      "epoch": 0.8731425190905058,
      "grad_norm": 3.254939079284668,
      "learning_rate": 1.3354418970518348e-07,
      "loss": 0.7224,
      "step": 56085
    },
    {
      "epoch": 0.8732203600924751,
      "grad_norm": 2.835361957550049,
      "learning_rate": 1.3346225070057848e-07,
      "loss": 0.7633,
      "step": 56090
    },
    {
      "epoch": 0.8732982010944444,
      "grad_norm": 4.518415927886963,
      "learning_rate": 1.3338031169597353e-07,
      "loss": 0.6333,
      "step": 56095
    },
    {
      "epoch": 0.8733760420964138,
      "grad_norm": 3.182145357131958,
      "learning_rate": 1.3329837269136855e-07,
      "loss": 0.7099,
      "step": 56100
    },
    {
      "epoch": 0.8734538830983832,
      "grad_norm": 3.4830996990203857,
      "learning_rate": 1.3321643368676357e-07,
      "loss": 0.7464,
      "step": 56105
    },
    {
      "epoch": 0.8735317241003526,
      "grad_norm": 4.295426845550537,
      "learning_rate": 1.331344946821586e-07,
      "loss": 0.7753,
      "step": 56110
    },
    {
      "epoch": 0.873609565102322,
      "grad_norm": 5.327083110809326,
      "learning_rate": 1.3305255567755364e-07,
      "loss": 0.7185,
      "step": 56115
    },
    {
      "epoch": 0.8736874061042914,
      "grad_norm": 3.6394269466400146,
      "learning_rate": 1.3297061667294866e-07,
      "loss": 0.8662,
      "step": 56120
    },
    {
      "epoch": 0.8737652471062608,
      "grad_norm": 14.427268981933594,
      "learning_rate": 1.3288867766834368e-07,
      "loss": 0.6917,
      "step": 56125
    },
    {
      "epoch": 0.8738430881082301,
      "grad_norm": 5.970995903015137,
      "learning_rate": 1.328067386637387e-07,
      "loss": 0.7722,
      "step": 56130
    },
    {
      "epoch": 0.8739209291101995,
      "grad_norm": 4.307867050170898,
      "learning_rate": 1.3272479965913375e-07,
      "loss": 0.8104,
      "step": 56135
    },
    {
      "epoch": 0.8739987701121689,
      "grad_norm": 2.941844940185547,
      "learning_rate": 1.3264286065452875e-07,
      "loss": 0.8195,
      "step": 56140
    },
    {
      "epoch": 0.8740766111141383,
      "grad_norm": 4.369835376739502,
      "learning_rate": 1.325609216499238e-07,
      "loss": 0.778,
      "step": 56145
    },
    {
      "epoch": 0.8741544521161076,
      "grad_norm": 3.727729082107544,
      "learning_rate": 1.3247898264531882e-07,
      "loss": 0.6694,
      "step": 56150
    },
    {
      "epoch": 0.874232293118077,
      "grad_norm": 4.051523208618164,
      "learning_rate": 1.3239704364071387e-07,
      "loss": 0.7146,
      "step": 56155
    },
    {
      "epoch": 0.8743101341200464,
      "grad_norm": 3.1586854457855225,
      "learning_rate": 1.3231510463610886e-07,
      "loss": 0.628,
      "step": 56160
    },
    {
      "epoch": 0.8743879751220158,
      "grad_norm": 4.476223945617676,
      "learning_rate": 1.322331656315039e-07,
      "loss": 0.5959,
      "step": 56165
    },
    {
      "epoch": 0.8744658161239851,
      "grad_norm": 4.40348482131958,
      "learning_rate": 1.3215122662689893e-07,
      "loss": 0.8548,
      "step": 56170
    },
    {
      "epoch": 0.8745436571259545,
      "grad_norm": 3.2502453327178955,
      "learning_rate": 1.3206928762229396e-07,
      "loss": 0.7319,
      "step": 56175
    },
    {
      "epoch": 0.8746214981279239,
      "grad_norm": 4.378718376159668,
      "learning_rate": 1.3198734861768898e-07,
      "loss": 0.7203,
      "step": 56180
    },
    {
      "epoch": 0.8746993391298933,
      "grad_norm": 4.191452980041504,
      "learning_rate": 1.3190540961308403e-07,
      "loss": 0.6996,
      "step": 56185
    },
    {
      "epoch": 0.8747771801318627,
      "grad_norm": 6.194228172302246,
      "learning_rate": 1.3182347060847905e-07,
      "loss": 0.7481,
      "step": 56190
    },
    {
      "epoch": 0.8748550211338321,
      "grad_norm": 4.28296422958374,
      "learning_rate": 1.3174153160387407e-07,
      "loss": 0.703,
      "step": 56195
    },
    {
      "epoch": 0.8749328621358015,
      "grad_norm": 3.6209700107574463,
      "learning_rate": 1.316595925992691e-07,
      "loss": 0.7936,
      "step": 56200
    },
    {
      "epoch": 0.8750107031377707,
      "grad_norm": 5.379439830780029,
      "learning_rate": 1.3157765359466414e-07,
      "loss": 0.7584,
      "step": 56205
    },
    {
      "epoch": 0.8750885441397401,
      "grad_norm": 3.270545244216919,
      "learning_rate": 1.3149571459005914e-07,
      "loss": 0.7455,
      "step": 56210
    },
    {
      "epoch": 0.8751663851417095,
      "grad_norm": 4.447718143463135,
      "learning_rate": 1.3141377558545418e-07,
      "loss": 0.7832,
      "step": 56215
    },
    {
      "epoch": 0.8752442261436789,
      "grad_norm": 3.630067825317383,
      "learning_rate": 1.313318365808492e-07,
      "loss": 0.7294,
      "step": 56220
    },
    {
      "epoch": 0.8753220671456483,
      "grad_norm": 3.7737536430358887,
      "learning_rate": 1.3124989757624425e-07,
      "loss": 0.645,
      "step": 56225
    },
    {
      "epoch": 0.8753999081476177,
      "grad_norm": 4.0180230140686035,
      "learning_rate": 1.3116795857163925e-07,
      "loss": 0.7757,
      "step": 56230
    },
    {
      "epoch": 0.8754777491495871,
      "grad_norm": 3.5569779872894287,
      "learning_rate": 1.310860195670343e-07,
      "loss": 0.7689,
      "step": 56235
    },
    {
      "epoch": 0.8755555901515565,
      "grad_norm": 5.653585910797119,
      "learning_rate": 1.3100408056242932e-07,
      "loss": 0.7176,
      "step": 56240
    },
    {
      "epoch": 0.8756334311535258,
      "grad_norm": 3.0365829467773438,
      "learning_rate": 1.3092214155782437e-07,
      "loss": 0.6923,
      "step": 56245
    },
    {
      "epoch": 0.8757112721554952,
      "grad_norm": 6.790417194366455,
      "learning_rate": 1.3084020255321936e-07,
      "loss": 0.7635,
      "step": 56250
    },
    {
      "epoch": 0.8757891131574645,
      "grad_norm": 3.3026535511016846,
      "learning_rate": 1.307582635486144e-07,
      "loss": 0.7142,
      "step": 56255
    },
    {
      "epoch": 0.8758669541594339,
      "grad_norm": 3.466252326965332,
      "learning_rate": 1.3067632454400943e-07,
      "loss": 0.7248,
      "step": 56260
    },
    {
      "epoch": 0.8759447951614033,
      "grad_norm": 4.846683502197266,
      "learning_rate": 1.3059438553940446e-07,
      "loss": 0.7549,
      "step": 56265
    },
    {
      "epoch": 0.8760226361633727,
      "grad_norm": 8.07625961303711,
      "learning_rate": 1.3051244653479948e-07,
      "loss": 0.8072,
      "step": 56270
    },
    {
      "epoch": 0.8761004771653421,
      "grad_norm": 2.6739368438720703,
      "learning_rate": 1.3043050753019453e-07,
      "loss": 0.6652,
      "step": 56275
    },
    {
      "epoch": 0.8761783181673114,
      "grad_norm": 5.897090911865234,
      "learning_rate": 1.3034856852558955e-07,
      "loss": 0.7032,
      "step": 56280
    },
    {
      "epoch": 0.8762561591692808,
      "grad_norm": 4.173807621002197,
      "learning_rate": 1.3026662952098457e-07,
      "loss": 0.7635,
      "step": 56285
    },
    {
      "epoch": 0.8763340001712502,
      "grad_norm": 4.293265342712402,
      "learning_rate": 1.301846905163796e-07,
      "loss": 0.7423,
      "step": 56290
    },
    {
      "epoch": 0.8764118411732196,
      "grad_norm": 7.3962225914001465,
      "learning_rate": 1.3010275151177464e-07,
      "loss": 0.8207,
      "step": 56295
    },
    {
      "epoch": 0.876489682175189,
      "grad_norm": 4.909463405609131,
      "learning_rate": 1.3002081250716964e-07,
      "loss": 0.6994,
      "step": 56300
    },
    {
      "epoch": 0.8765675231771584,
      "grad_norm": 5.596007347106934,
      "learning_rate": 1.2993887350256468e-07,
      "loss": 0.7589,
      "step": 56305
    },
    {
      "epoch": 0.8766453641791278,
      "grad_norm": 7.126872539520264,
      "learning_rate": 1.298569344979597e-07,
      "loss": 0.7024,
      "step": 56310
    },
    {
      "epoch": 0.8767232051810971,
      "grad_norm": 3.1609716415405273,
      "learning_rate": 1.2977499549335475e-07,
      "loss": 0.7732,
      "step": 56315
    },
    {
      "epoch": 0.8768010461830664,
      "grad_norm": 4.965465545654297,
      "learning_rate": 1.2969305648874975e-07,
      "loss": 0.6562,
      "step": 56320
    },
    {
      "epoch": 0.8768788871850358,
      "grad_norm": 3.8195600509643555,
      "learning_rate": 1.296111174841448e-07,
      "loss": 0.6992,
      "step": 56325
    },
    {
      "epoch": 0.8769567281870052,
      "grad_norm": 2.9354135990142822,
      "learning_rate": 1.2952917847953982e-07,
      "loss": 0.7021,
      "step": 56330
    },
    {
      "epoch": 0.8770345691889746,
      "grad_norm": 3.6384711265563965,
      "learning_rate": 1.2944723947493484e-07,
      "loss": 0.7464,
      "step": 56335
    },
    {
      "epoch": 0.877112410190944,
      "grad_norm": 6.809104919433594,
      "learning_rate": 1.2936530047032986e-07,
      "loss": 0.7605,
      "step": 56340
    },
    {
      "epoch": 0.8771902511929134,
      "grad_norm": 6.566957950592041,
      "learning_rate": 1.292833614657249e-07,
      "loss": 0.7663,
      "step": 56345
    },
    {
      "epoch": 0.8772680921948828,
      "grad_norm": 4.310968399047852,
      "learning_rate": 1.2920142246111993e-07,
      "loss": 0.7567,
      "step": 56350
    },
    {
      "epoch": 0.8773459331968521,
      "grad_norm": 4.642413139343262,
      "learning_rate": 1.2911948345651496e-07,
      "loss": 0.7355,
      "step": 56355
    },
    {
      "epoch": 0.8774237741988214,
      "grad_norm": 4.671013355255127,
      "learning_rate": 1.2903754445190998e-07,
      "loss": 0.7694,
      "step": 56360
    },
    {
      "epoch": 0.8775016152007908,
      "grad_norm": 3.729888677597046,
      "learning_rate": 1.2895560544730503e-07,
      "loss": 0.8377,
      "step": 56365
    },
    {
      "epoch": 0.8775794562027602,
      "grad_norm": 2.92903208732605,
      "learning_rate": 1.2887366644270002e-07,
      "loss": 0.7982,
      "step": 56370
    },
    {
      "epoch": 0.8776572972047296,
      "grad_norm": 4.296948432922363,
      "learning_rate": 1.2879172743809507e-07,
      "loss": 0.6987,
      "step": 56375
    },
    {
      "epoch": 0.877735138206699,
      "grad_norm": 3.697868585586548,
      "learning_rate": 1.287097884334901e-07,
      "loss": 0.766,
      "step": 56380
    },
    {
      "epoch": 0.8778129792086684,
      "grad_norm": 2.4828803539276123,
      "learning_rate": 1.2862784942888514e-07,
      "loss": 0.7431,
      "step": 56385
    },
    {
      "epoch": 0.8778908202106378,
      "grad_norm": 4.158952236175537,
      "learning_rate": 1.2854591042428014e-07,
      "loss": 0.7566,
      "step": 56390
    },
    {
      "epoch": 0.8779686612126071,
      "grad_norm": 2.6162633895874023,
      "learning_rate": 1.2846397141967518e-07,
      "loss": 0.712,
      "step": 56395
    },
    {
      "epoch": 0.8780465022145765,
      "grad_norm": 6.182302474975586,
      "learning_rate": 1.2838203241507023e-07,
      "loss": 0.7428,
      "step": 56400
    },
    {
      "epoch": 0.8781243432165459,
      "grad_norm": 6.934322357177734,
      "learning_rate": 1.2830009341046525e-07,
      "loss": 0.7436,
      "step": 56405
    },
    {
      "epoch": 0.8782021842185153,
      "grad_norm": 4.265978813171387,
      "learning_rate": 1.2821815440586028e-07,
      "loss": 0.8366,
      "step": 56410
    },
    {
      "epoch": 0.8782800252204847,
      "grad_norm": 4.213291645050049,
      "learning_rate": 1.281362154012553e-07,
      "loss": 0.8029,
      "step": 56415
    },
    {
      "epoch": 0.878357866222454,
      "grad_norm": 4.935248851776123,
      "learning_rate": 1.2805427639665035e-07,
      "loss": 0.8149,
      "step": 56420
    },
    {
      "epoch": 0.8784357072244234,
      "grad_norm": 3.660370111465454,
      "learning_rate": 1.2797233739204534e-07,
      "loss": 0.6536,
      "step": 56425
    },
    {
      "epoch": 0.8785135482263927,
      "grad_norm": 3.972989320755005,
      "learning_rate": 1.278903983874404e-07,
      "loss": 0.7971,
      "step": 56430
    },
    {
      "epoch": 0.8785913892283621,
      "grad_norm": 3.7157809734344482,
      "learning_rate": 1.278084593828354e-07,
      "loss": 0.7567,
      "step": 56435
    },
    {
      "epoch": 0.8786692302303315,
      "grad_norm": 3.5546178817749023,
      "learning_rate": 1.2772652037823046e-07,
      "loss": 0.7259,
      "step": 56440
    },
    {
      "epoch": 0.8787470712323009,
      "grad_norm": 4.014951229095459,
      "learning_rate": 1.2764458137362546e-07,
      "loss": 0.6875,
      "step": 56445
    },
    {
      "epoch": 0.8788249122342703,
      "grad_norm": 3.1290183067321777,
      "learning_rate": 1.275626423690205e-07,
      "loss": 0.6486,
      "step": 56450
    },
    {
      "epoch": 0.8789027532362397,
      "grad_norm": 7.870703220367432,
      "learning_rate": 1.2748070336441553e-07,
      "loss": 0.8129,
      "step": 56455
    },
    {
      "epoch": 0.8789805942382091,
      "grad_norm": 3.3703393936157227,
      "learning_rate": 1.2739876435981055e-07,
      "loss": 0.7544,
      "step": 56460
    },
    {
      "epoch": 0.8790584352401785,
      "grad_norm": 2.7564656734466553,
      "learning_rate": 1.2731682535520557e-07,
      "loss": 0.8046,
      "step": 56465
    },
    {
      "epoch": 0.8791362762421477,
      "grad_norm": 4.360289096832275,
      "learning_rate": 1.2723488635060062e-07,
      "loss": 0.6865,
      "step": 56470
    },
    {
      "epoch": 0.8792141172441171,
      "grad_norm": 6.860365867614746,
      "learning_rate": 1.2715294734599564e-07,
      "loss": 0.7172,
      "step": 56475
    },
    {
      "epoch": 0.8792919582460865,
      "grad_norm": 9.749326705932617,
      "learning_rate": 1.2707100834139066e-07,
      "loss": 0.6804,
      "step": 56480
    },
    {
      "epoch": 0.8793697992480559,
      "grad_norm": 4.706418514251709,
      "learning_rate": 1.2698906933678569e-07,
      "loss": 0.7289,
      "step": 56485
    },
    {
      "epoch": 0.8794476402500253,
      "grad_norm": 3.534531593322754,
      "learning_rate": 1.2690713033218073e-07,
      "loss": 0.7721,
      "step": 56490
    },
    {
      "epoch": 0.8795254812519947,
      "grad_norm": 3.613826036453247,
      "learning_rate": 1.2682519132757576e-07,
      "loss": 0.761,
      "step": 56495
    },
    {
      "epoch": 0.8796033222539641,
      "grad_norm": 3.6603169441223145,
      "learning_rate": 1.2674325232297078e-07,
      "loss": 0.7185,
      "step": 56500
    },
    {
      "epoch": 0.8796811632559335,
      "grad_norm": 3.8486320972442627,
      "learning_rate": 1.266613133183658e-07,
      "loss": 0.7638,
      "step": 56505
    },
    {
      "epoch": 0.8797590042579028,
      "grad_norm": 3.7550759315490723,
      "learning_rate": 1.2657937431376085e-07,
      "loss": 0.8397,
      "step": 56510
    },
    {
      "epoch": 0.8798368452598722,
      "grad_norm": 5.25081729888916,
      "learning_rate": 1.2649743530915584e-07,
      "loss": 0.6617,
      "step": 56515
    },
    {
      "epoch": 0.8799146862618415,
      "grad_norm": 6.233492374420166,
      "learning_rate": 1.264154963045509e-07,
      "loss": 0.7915,
      "step": 56520
    },
    {
      "epoch": 0.8799925272638109,
      "grad_norm": 13.970528602600098,
      "learning_rate": 1.2633355729994591e-07,
      "loss": 0.8823,
      "step": 56525
    },
    {
      "epoch": 0.8800703682657803,
      "grad_norm": 3.205857038497925,
      "learning_rate": 1.2625161829534096e-07,
      "loss": 0.6427,
      "step": 56530
    },
    {
      "epoch": 0.8801482092677497,
      "grad_norm": 3.788273811340332,
      "learning_rate": 1.2616967929073596e-07,
      "loss": 0.6812,
      "step": 56535
    },
    {
      "epoch": 0.8802260502697191,
      "grad_norm": 4.394341468811035,
      "learning_rate": 1.26087740286131e-07,
      "loss": 0.8216,
      "step": 56540
    },
    {
      "epoch": 0.8803038912716884,
      "grad_norm": 2.830798625946045,
      "learning_rate": 1.2600580128152603e-07,
      "loss": 0.602,
      "step": 56545
    },
    {
      "epoch": 0.8803817322736578,
      "grad_norm": 5.070180416107178,
      "learning_rate": 1.2592386227692105e-07,
      "loss": 0.9052,
      "step": 56550
    },
    {
      "epoch": 0.8804595732756272,
      "grad_norm": 4.9051995277404785,
      "learning_rate": 1.2584192327231607e-07,
      "loss": 0.7805,
      "step": 56555
    },
    {
      "epoch": 0.8805374142775966,
      "grad_norm": 2.7322583198547363,
      "learning_rate": 1.2575998426771112e-07,
      "loss": 0.6982,
      "step": 56560
    },
    {
      "epoch": 0.880615255279566,
      "grad_norm": 5.003993511199951,
      "learning_rate": 1.2567804526310614e-07,
      "loss": 0.652,
      "step": 56565
    },
    {
      "epoch": 0.8806930962815354,
      "grad_norm": 8.174132347106934,
      "learning_rate": 1.2559610625850116e-07,
      "loss": 0.6922,
      "step": 56570
    },
    {
      "epoch": 0.8807709372835048,
      "grad_norm": 3.6196470260620117,
      "learning_rate": 1.2551416725389619e-07,
      "loss": 0.7946,
      "step": 56575
    },
    {
      "epoch": 0.8808487782854741,
      "grad_norm": 7.573304653167725,
      "learning_rate": 1.2543222824929123e-07,
      "loss": 0.7423,
      "step": 56580
    },
    {
      "epoch": 0.8809266192874434,
      "grad_norm": 3.592780828475952,
      "learning_rate": 1.2535028924468623e-07,
      "loss": 0.7413,
      "step": 56585
    },
    {
      "epoch": 0.8810044602894128,
      "grad_norm": 4.712695598602295,
      "learning_rate": 1.2526835024008128e-07,
      "loss": 0.8045,
      "step": 56590
    },
    {
      "epoch": 0.8810823012913822,
      "grad_norm": 4.784394264221191,
      "learning_rate": 1.251864112354763e-07,
      "loss": 0.6904,
      "step": 56595
    },
    {
      "epoch": 0.8811601422933516,
      "grad_norm": 4.1296916007995605,
      "learning_rate": 1.2510447223087135e-07,
      "loss": 0.6871,
      "step": 56600
    },
    {
      "epoch": 0.881237983295321,
      "grad_norm": 4.8244524002075195,
      "learning_rate": 1.2502253322626634e-07,
      "loss": 0.8091,
      "step": 56605
    },
    {
      "epoch": 0.8813158242972904,
      "grad_norm": 3.881037712097168,
      "learning_rate": 1.249405942216614e-07,
      "loss": 0.837,
      "step": 56610
    },
    {
      "epoch": 0.8813936652992598,
      "grad_norm": 11.355805397033691,
      "learning_rate": 1.2485865521705641e-07,
      "loss": 0.7674,
      "step": 56615
    },
    {
      "epoch": 0.8814715063012291,
      "grad_norm": 3.766165018081665,
      "learning_rate": 1.2477671621245146e-07,
      "loss": 0.753,
      "step": 56620
    },
    {
      "epoch": 0.8815493473031984,
      "grad_norm": 3.5065689086914062,
      "learning_rate": 1.2469477720784648e-07,
      "loss": 0.6827,
      "step": 56625
    },
    {
      "epoch": 0.8816271883051678,
      "grad_norm": 3.563176155090332,
      "learning_rate": 1.246128382032415e-07,
      "loss": 0.7451,
      "step": 56630
    },
    {
      "epoch": 0.8817050293071372,
      "grad_norm": 4.187000274658203,
      "learning_rate": 1.2453089919863653e-07,
      "loss": 0.7488,
      "step": 56635
    },
    {
      "epoch": 0.8817828703091066,
      "grad_norm": 4.486247539520264,
      "learning_rate": 1.2444896019403155e-07,
      "loss": 0.6832,
      "step": 56640
    },
    {
      "epoch": 0.881860711311076,
      "grad_norm": 3.5161051750183105,
      "learning_rate": 1.243670211894266e-07,
      "loss": 0.737,
      "step": 56645
    },
    {
      "epoch": 0.8819385523130454,
      "grad_norm": 2.8879921436309814,
      "learning_rate": 1.2428508218482162e-07,
      "loss": 0.7255,
      "step": 56650
    },
    {
      "epoch": 0.8820163933150148,
      "grad_norm": 3.80989146232605,
      "learning_rate": 1.2420314318021664e-07,
      "loss": 0.7932,
      "step": 56655
    },
    {
      "epoch": 0.8820942343169841,
      "grad_norm": 3.5046894550323486,
      "learning_rate": 1.2412120417561166e-07,
      "loss": 0.7262,
      "step": 56660
    },
    {
      "epoch": 0.8821720753189535,
      "grad_norm": 6.638458251953125,
      "learning_rate": 1.2403926517100669e-07,
      "loss": 0.7539,
      "step": 56665
    },
    {
      "epoch": 0.8822499163209229,
      "grad_norm": 9.757596969604492,
      "learning_rate": 1.2395732616640173e-07,
      "loss": 0.7934,
      "step": 56670
    },
    {
      "epoch": 0.8823277573228923,
      "grad_norm": 6.414024829864502,
      "learning_rate": 1.2387538716179676e-07,
      "loss": 0.72,
      "step": 56675
    },
    {
      "epoch": 0.8824055983248617,
      "grad_norm": 5.326852321624756,
      "learning_rate": 1.2379344815719178e-07,
      "loss": 0.7912,
      "step": 56680
    },
    {
      "epoch": 0.882483439326831,
      "grad_norm": 2.4989001750946045,
      "learning_rate": 1.237115091525868e-07,
      "loss": 0.7301,
      "step": 56685
    },
    {
      "epoch": 0.8825612803288004,
      "grad_norm": 3.816626787185669,
      "learning_rate": 1.2362957014798185e-07,
      "loss": 0.6888,
      "step": 56690
    },
    {
      "epoch": 0.8826391213307697,
      "grad_norm": 2.1098105907440186,
      "learning_rate": 1.2354763114337687e-07,
      "loss": 0.6632,
      "step": 56695
    },
    {
      "epoch": 0.8827169623327391,
      "grad_norm": 3.453254461288452,
      "learning_rate": 1.234656921387719e-07,
      "loss": 0.7642,
      "step": 56700
    },
    {
      "epoch": 0.8827948033347085,
      "grad_norm": 2.889141321182251,
      "learning_rate": 1.2338375313416691e-07,
      "loss": 0.7312,
      "step": 56705
    },
    {
      "epoch": 0.8828726443366779,
      "grad_norm": 4.403680324554443,
      "learning_rate": 1.2330181412956194e-07,
      "loss": 0.719,
      "step": 56710
    },
    {
      "epoch": 0.8829504853386473,
      "grad_norm": 4.169454574584961,
      "learning_rate": 1.2321987512495698e-07,
      "loss": 0.699,
      "step": 56715
    },
    {
      "epoch": 0.8830283263406167,
      "grad_norm": 3.4330496788024902,
      "learning_rate": 1.23137936120352e-07,
      "loss": 0.7342,
      "step": 56720
    },
    {
      "epoch": 0.8831061673425861,
      "grad_norm": 3.399306058883667,
      "learning_rate": 1.2305599711574703e-07,
      "loss": 0.7305,
      "step": 56725
    },
    {
      "epoch": 0.8831840083445555,
      "grad_norm": 3.437608003616333,
      "learning_rate": 1.2297405811114205e-07,
      "loss": 0.7385,
      "step": 56730
    },
    {
      "epoch": 0.8832618493465247,
      "grad_norm": 3.1189212799072266,
      "learning_rate": 1.228921191065371e-07,
      "loss": 0.7518,
      "step": 56735
    },
    {
      "epoch": 0.8833396903484941,
      "grad_norm": 6.237093448638916,
      "learning_rate": 1.2281018010193212e-07,
      "loss": 0.6989,
      "step": 56740
    },
    {
      "epoch": 0.8834175313504635,
      "grad_norm": 6.458376407623291,
      "learning_rate": 1.2272824109732714e-07,
      "loss": 0.6061,
      "step": 56745
    },
    {
      "epoch": 0.8834953723524329,
      "grad_norm": 3.678632974624634,
      "learning_rate": 1.2264630209272216e-07,
      "loss": 0.7054,
      "step": 56750
    },
    {
      "epoch": 0.8835732133544023,
      "grad_norm": 3.669140577316284,
      "learning_rate": 1.2256436308811719e-07,
      "loss": 0.8019,
      "step": 56755
    },
    {
      "epoch": 0.8836510543563717,
      "grad_norm": 5.922098636627197,
      "learning_rate": 1.2248242408351223e-07,
      "loss": 0.6787,
      "step": 56760
    },
    {
      "epoch": 0.8837288953583411,
      "grad_norm": 3.541712999343872,
      "learning_rate": 1.2240048507890726e-07,
      "loss": 0.7572,
      "step": 56765
    },
    {
      "epoch": 0.8838067363603105,
      "grad_norm": 5.488617897033691,
      "learning_rate": 1.2231854607430228e-07,
      "loss": 0.6573,
      "step": 56770
    },
    {
      "epoch": 0.8838845773622798,
      "grad_norm": 3.520219326019287,
      "learning_rate": 1.222366070696973e-07,
      "loss": 0.7742,
      "step": 56775
    },
    {
      "epoch": 0.8839624183642492,
      "grad_norm": 3.8097896575927734,
      "learning_rate": 1.2215466806509235e-07,
      "loss": 0.8075,
      "step": 56780
    },
    {
      "epoch": 0.8840402593662186,
      "grad_norm": 3.0948686599731445,
      "learning_rate": 1.2207272906048737e-07,
      "loss": 0.6975,
      "step": 56785
    },
    {
      "epoch": 0.8841181003681879,
      "grad_norm": 3.633786916732788,
      "learning_rate": 1.219907900558824e-07,
      "loss": 0.7258,
      "step": 56790
    },
    {
      "epoch": 0.8841959413701573,
      "grad_norm": 3.823458671569824,
      "learning_rate": 1.2190885105127741e-07,
      "loss": 0.6519,
      "step": 56795
    },
    {
      "epoch": 0.8842737823721267,
      "grad_norm": 4.036854267120361,
      "learning_rate": 1.2182691204667244e-07,
      "loss": 0.7186,
      "step": 56800
    },
    {
      "epoch": 0.8843516233740961,
      "grad_norm": 3.111171245574951,
      "learning_rate": 1.2174497304206748e-07,
      "loss": 0.716,
      "step": 56805
    },
    {
      "epoch": 0.8844294643760654,
      "grad_norm": 2.7838401794433594,
      "learning_rate": 1.216630340374625e-07,
      "loss": 0.6664,
      "step": 56810
    },
    {
      "epoch": 0.8845073053780348,
      "grad_norm": 4.322157382965088,
      "learning_rate": 1.2158109503285753e-07,
      "loss": 0.7421,
      "step": 56815
    },
    {
      "epoch": 0.8845851463800042,
      "grad_norm": 3.3897886276245117,
      "learning_rate": 1.2149915602825255e-07,
      "loss": 0.8067,
      "step": 56820
    },
    {
      "epoch": 0.8846629873819736,
      "grad_norm": 2.6106998920440674,
      "learning_rate": 1.214172170236476e-07,
      "loss": 0.747,
      "step": 56825
    },
    {
      "epoch": 0.884740828383943,
      "grad_norm": 2.5302155017852783,
      "learning_rate": 1.2133527801904262e-07,
      "loss": 0.6596,
      "step": 56830
    },
    {
      "epoch": 0.8848186693859124,
      "grad_norm": 3.1086983680725098,
      "learning_rate": 1.2125333901443764e-07,
      "loss": 0.6574,
      "step": 56835
    },
    {
      "epoch": 0.8848965103878818,
      "grad_norm": 8.846805572509766,
      "learning_rate": 1.2117140000983266e-07,
      "loss": 0.7939,
      "step": 56840
    },
    {
      "epoch": 0.8849743513898511,
      "grad_norm": 2.5658347606658936,
      "learning_rate": 1.210894610052277e-07,
      "loss": 0.861,
      "step": 56845
    },
    {
      "epoch": 0.8850521923918204,
      "grad_norm": 2.762073278427124,
      "learning_rate": 1.2100752200062274e-07,
      "loss": 0.752,
      "step": 56850
    },
    {
      "epoch": 0.8851300333937898,
      "grad_norm": 3.3005340099334717,
      "learning_rate": 1.2092558299601776e-07,
      "loss": 0.7182,
      "step": 56855
    },
    {
      "epoch": 0.8852078743957592,
      "grad_norm": 5.331596374511719,
      "learning_rate": 1.2084364399141278e-07,
      "loss": 0.7882,
      "step": 56860
    },
    {
      "epoch": 0.8852857153977286,
      "grad_norm": 10.077093124389648,
      "learning_rate": 1.207617049868078e-07,
      "loss": 0.7008,
      "step": 56865
    },
    {
      "epoch": 0.885363556399698,
      "grad_norm": 2.869558334350586,
      "learning_rate": 1.2067976598220285e-07,
      "loss": 0.7502,
      "step": 56870
    },
    {
      "epoch": 0.8854413974016674,
      "grad_norm": 3.223896026611328,
      "learning_rate": 1.2059782697759787e-07,
      "loss": 0.6574,
      "step": 56875
    },
    {
      "epoch": 0.8855192384036368,
      "grad_norm": 2.806377649307251,
      "learning_rate": 1.205158879729929e-07,
      "loss": 0.7226,
      "step": 56880
    },
    {
      "epoch": 0.8855970794056061,
      "grad_norm": 3.6280407905578613,
      "learning_rate": 1.2043394896838794e-07,
      "loss": 0.6629,
      "step": 56885
    },
    {
      "epoch": 0.8856749204075755,
      "grad_norm": 3.17048716545105,
      "learning_rate": 1.2035200996378296e-07,
      "loss": 0.802,
      "step": 56890
    },
    {
      "epoch": 0.8857527614095448,
      "grad_norm": 7.640406131744385,
      "learning_rate": 1.2027007095917799e-07,
      "loss": 0.704,
      "step": 56895
    },
    {
      "epoch": 0.8858306024115142,
      "grad_norm": 2.5215060710906982,
      "learning_rate": 1.20188131954573e-07,
      "loss": 0.6762,
      "step": 56900
    },
    {
      "epoch": 0.8859084434134836,
      "grad_norm": 3.224133253097534,
      "learning_rate": 1.2010619294996806e-07,
      "loss": 0.6684,
      "step": 56905
    },
    {
      "epoch": 0.885986284415453,
      "grad_norm": 3.6025681495666504,
      "learning_rate": 1.2002425394536308e-07,
      "loss": 0.6877,
      "step": 56910
    },
    {
      "epoch": 0.8860641254174224,
      "grad_norm": 7.4579997062683105,
      "learning_rate": 1.199423149407581e-07,
      "loss": 0.7146,
      "step": 56915
    },
    {
      "epoch": 0.8861419664193918,
      "grad_norm": 3.743452310562134,
      "learning_rate": 1.1986037593615312e-07,
      "loss": 0.6689,
      "step": 56920
    },
    {
      "epoch": 0.8862198074213611,
      "grad_norm": 6.264161586761475,
      "learning_rate": 1.1977843693154814e-07,
      "loss": 0.753,
      "step": 56925
    },
    {
      "epoch": 0.8862976484233305,
      "grad_norm": 3.5045132637023926,
      "learning_rate": 1.196964979269432e-07,
      "loss": 0.6559,
      "step": 56930
    },
    {
      "epoch": 0.8863754894252999,
      "grad_norm": 4.51326847076416,
      "learning_rate": 1.1961455892233821e-07,
      "loss": 0.7064,
      "step": 56935
    },
    {
      "epoch": 0.8864533304272693,
      "grad_norm": 3.111332654953003,
      "learning_rate": 1.1953261991773324e-07,
      "loss": 0.7702,
      "step": 56940
    },
    {
      "epoch": 0.8865311714292387,
      "grad_norm": 5.088558197021484,
      "learning_rate": 1.1945068091312826e-07,
      "loss": 0.7834,
      "step": 56945
    },
    {
      "epoch": 0.886609012431208,
      "grad_norm": 2.3277926445007324,
      "learning_rate": 1.193687419085233e-07,
      "loss": 0.7298,
      "step": 56950
    },
    {
      "epoch": 0.8866868534331774,
      "grad_norm": 3.061819076538086,
      "learning_rate": 1.1928680290391833e-07,
      "loss": 0.8024,
      "step": 56955
    },
    {
      "epoch": 0.8867646944351467,
      "grad_norm": 4.5664801597595215,
      "learning_rate": 1.1920486389931335e-07,
      "loss": 0.7476,
      "step": 56960
    },
    {
      "epoch": 0.8868425354371161,
      "grad_norm": 4.1422200202941895,
      "learning_rate": 1.1912292489470837e-07,
      "loss": 0.7191,
      "step": 56965
    },
    {
      "epoch": 0.8869203764390855,
      "grad_norm": 3.3067984580993652,
      "learning_rate": 1.1904098589010341e-07,
      "loss": 0.8005,
      "step": 56970
    },
    {
      "epoch": 0.8869982174410549,
      "grad_norm": 5.923746109008789,
      "learning_rate": 1.1895904688549843e-07,
      "loss": 0.7277,
      "step": 56975
    },
    {
      "epoch": 0.8870760584430243,
      "grad_norm": 4.587955474853516,
      "learning_rate": 1.1887710788089346e-07,
      "loss": 0.7218,
      "step": 56980
    },
    {
      "epoch": 0.8871538994449937,
      "grad_norm": 4.317761421203613,
      "learning_rate": 1.1879516887628849e-07,
      "loss": 0.8249,
      "step": 56985
    },
    {
      "epoch": 0.8872317404469631,
      "grad_norm": 5.247656345367432,
      "learning_rate": 1.1871322987168352e-07,
      "loss": 0.6474,
      "step": 56990
    },
    {
      "epoch": 0.8873095814489325,
      "grad_norm": 4.712335586547852,
      "learning_rate": 1.1863129086707854e-07,
      "loss": 0.6847,
      "step": 56995
    },
    {
      "epoch": 0.8873874224509017,
      "grad_norm": 8.797277450561523,
      "learning_rate": 1.1854935186247356e-07,
      "loss": 0.79,
      "step": 57000
    },
    {
      "epoch": 0.8874652634528711,
      "grad_norm": 3.302917718887329,
      "learning_rate": 1.184674128578686e-07,
      "loss": 0.6569,
      "step": 57005
    },
    {
      "epoch": 0.8875431044548405,
      "grad_norm": 4.936827182769775,
      "learning_rate": 1.1838547385326362e-07,
      "loss": 0.7128,
      "step": 57010
    },
    {
      "epoch": 0.8876209454568099,
      "grad_norm": 4.248513221740723,
      "learning_rate": 1.1830353484865866e-07,
      "loss": 0.7805,
      "step": 57015
    },
    {
      "epoch": 0.8876987864587793,
      "grad_norm": 4.584031581878662,
      "learning_rate": 1.1822159584405368e-07,
      "loss": 0.8119,
      "step": 57020
    },
    {
      "epoch": 0.8877766274607487,
      "grad_norm": 3.692356586456299,
      "learning_rate": 1.1813965683944871e-07,
      "loss": 0.8315,
      "step": 57025
    },
    {
      "epoch": 0.8878544684627181,
      "grad_norm": 3.051269292831421,
      "learning_rate": 1.1805771783484374e-07,
      "loss": 0.8108,
      "step": 57030
    },
    {
      "epoch": 0.8879323094646874,
      "grad_norm": 4.377344608306885,
      "learning_rate": 1.1797577883023877e-07,
      "loss": 0.8103,
      "step": 57035
    },
    {
      "epoch": 0.8880101504666568,
      "grad_norm": 4.007160663604736,
      "learning_rate": 1.1789383982563379e-07,
      "loss": 0.8376,
      "step": 57040
    },
    {
      "epoch": 0.8880879914686262,
      "grad_norm": 3.3551008701324463,
      "learning_rate": 1.1781190082102882e-07,
      "loss": 0.7578,
      "step": 57045
    },
    {
      "epoch": 0.8881658324705956,
      "grad_norm": 3.0606186389923096,
      "learning_rate": 1.1772996181642385e-07,
      "loss": 0.7862,
      "step": 57050
    },
    {
      "epoch": 0.8882436734725649,
      "grad_norm": 3.878682851791382,
      "learning_rate": 1.1764802281181887e-07,
      "loss": 0.7477,
      "step": 57055
    },
    {
      "epoch": 0.8883215144745343,
      "grad_norm": 2.6548829078674316,
      "learning_rate": 1.1756608380721391e-07,
      "loss": 0.6785,
      "step": 57060
    },
    {
      "epoch": 0.8883993554765037,
      "grad_norm": 3.9774625301361084,
      "learning_rate": 1.1748414480260893e-07,
      "loss": 0.7874,
      "step": 57065
    },
    {
      "epoch": 0.8884771964784731,
      "grad_norm": 3.7361531257629395,
      "learning_rate": 1.1740220579800396e-07,
      "loss": 0.7675,
      "step": 57070
    },
    {
      "epoch": 0.8885550374804424,
      "grad_norm": 2.9898688793182373,
      "learning_rate": 1.1732026679339899e-07,
      "loss": 0.6899,
      "step": 57075
    },
    {
      "epoch": 0.8886328784824118,
      "grad_norm": 6.212758541107178,
      "learning_rate": 1.1723832778879401e-07,
      "loss": 0.7086,
      "step": 57080
    },
    {
      "epoch": 0.8887107194843812,
      "grad_norm": 2.7542777061462402,
      "learning_rate": 1.1715638878418904e-07,
      "loss": 0.7809,
      "step": 57085
    },
    {
      "epoch": 0.8887885604863506,
      "grad_norm": 2.9145476818084717,
      "learning_rate": 1.1707444977958407e-07,
      "loss": 0.6463,
      "step": 57090
    },
    {
      "epoch": 0.88886640148832,
      "grad_norm": 3.751741886138916,
      "learning_rate": 1.169925107749791e-07,
      "loss": 0.6672,
      "step": 57095
    },
    {
      "epoch": 0.8889442424902894,
      "grad_norm": 4.501866817474365,
      "learning_rate": 1.1691057177037412e-07,
      "loss": 0.8048,
      "step": 57100
    },
    {
      "epoch": 0.8890220834922588,
      "grad_norm": 5.529430389404297,
      "learning_rate": 1.1682863276576916e-07,
      "loss": 0.7633,
      "step": 57105
    },
    {
      "epoch": 0.8890999244942281,
      "grad_norm": 2.4691476821899414,
      "learning_rate": 1.1674669376116418e-07,
      "loss": 0.706,
      "step": 57110
    },
    {
      "epoch": 0.8891777654961974,
      "grad_norm": 4.6680402755737305,
      "learning_rate": 1.1666475475655921e-07,
      "loss": 0.6824,
      "step": 57115
    },
    {
      "epoch": 0.8892556064981668,
      "grad_norm": 3.4516944885253906,
      "learning_rate": 1.1658281575195424e-07,
      "loss": 0.8289,
      "step": 57120
    },
    {
      "epoch": 0.8893334475001362,
      "grad_norm": 3.077443838119507,
      "learning_rate": 1.1650087674734926e-07,
      "loss": 0.694,
      "step": 57125
    },
    {
      "epoch": 0.8894112885021056,
      "grad_norm": 3.3001136779785156,
      "learning_rate": 1.164189377427443e-07,
      "loss": 0.8056,
      "step": 57130
    },
    {
      "epoch": 0.889489129504075,
      "grad_norm": 3.5928335189819336,
      "learning_rate": 1.1633699873813932e-07,
      "loss": 0.7808,
      "step": 57135
    },
    {
      "epoch": 0.8895669705060444,
      "grad_norm": 6.33901834487915,
      "learning_rate": 1.1625505973353435e-07,
      "loss": 0.7493,
      "step": 57140
    },
    {
      "epoch": 0.8896448115080138,
      "grad_norm": 3.04732608795166,
      "learning_rate": 1.1617312072892937e-07,
      "loss": 0.7338,
      "step": 57145
    },
    {
      "epoch": 0.8897226525099831,
      "grad_norm": 3.044984817504883,
      "learning_rate": 1.1609118172432441e-07,
      "loss": 0.7063,
      "step": 57150
    },
    {
      "epoch": 0.8898004935119525,
      "grad_norm": 3.659397840499878,
      "learning_rate": 1.1600924271971943e-07,
      "loss": 0.7321,
      "step": 57155
    },
    {
      "epoch": 0.8898783345139218,
      "grad_norm": 3.509012222290039,
      "learning_rate": 1.1592730371511445e-07,
      "loss": 0.7256,
      "step": 57160
    },
    {
      "epoch": 0.8899561755158912,
      "grad_norm": 2.8368170261383057,
      "learning_rate": 1.1584536471050949e-07,
      "loss": 0.6965,
      "step": 57165
    },
    {
      "epoch": 0.8900340165178606,
      "grad_norm": 3.5517570972442627,
      "learning_rate": 1.1576342570590451e-07,
      "loss": 0.7591,
      "step": 57170
    },
    {
      "epoch": 0.89011185751983,
      "grad_norm": 3.820858955383301,
      "learning_rate": 1.1568148670129954e-07,
      "loss": 0.7155,
      "step": 57175
    },
    {
      "epoch": 0.8901896985217994,
      "grad_norm": 3.7382965087890625,
      "learning_rate": 1.1559954769669457e-07,
      "loss": 0.6968,
      "step": 57180
    },
    {
      "epoch": 0.8902675395237688,
      "grad_norm": 2.9631519317626953,
      "learning_rate": 1.155176086920896e-07,
      "loss": 0.6067,
      "step": 57185
    },
    {
      "epoch": 0.8903453805257381,
      "grad_norm": 4.102830410003662,
      "learning_rate": 1.1543566968748464e-07,
      "loss": 0.7354,
      "step": 57190
    },
    {
      "epoch": 0.8904232215277075,
      "grad_norm": 8.195807456970215,
      "learning_rate": 1.1535373068287967e-07,
      "loss": 0.715,
      "step": 57195
    },
    {
      "epoch": 0.8905010625296769,
      "grad_norm": 6.862563610076904,
      "learning_rate": 1.1527179167827469e-07,
      "loss": 0.7365,
      "step": 57200
    },
    {
      "epoch": 0.8905789035316463,
      "grad_norm": 13.463440895080566,
      "learning_rate": 1.1518985267366971e-07,
      "loss": 0.6565,
      "step": 57205
    },
    {
      "epoch": 0.8906567445336157,
      "grad_norm": 2.881401300430298,
      "learning_rate": 1.1510791366906475e-07,
      "loss": 0.6421,
      "step": 57210
    },
    {
      "epoch": 0.890734585535585,
      "grad_norm": 3.7269399166107178,
      "learning_rate": 1.1502597466445977e-07,
      "loss": 0.6837,
      "step": 57215
    },
    {
      "epoch": 0.8908124265375544,
      "grad_norm": 3.103339672088623,
      "learning_rate": 1.1494403565985481e-07,
      "loss": 0.7574,
      "step": 57220
    },
    {
      "epoch": 0.8908902675395237,
      "grad_norm": 4.818991661071777,
      "learning_rate": 1.1486209665524983e-07,
      "loss": 0.694,
      "step": 57225
    },
    {
      "epoch": 0.8909681085414931,
      "grad_norm": 3.3130946159362793,
      "learning_rate": 1.1478015765064486e-07,
      "loss": 0.7423,
      "step": 57230
    },
    {
      "epoch": 0.8910459495434625,
      "grad_norm": 9.583632469177246,
      "learning_rate": 1.1469821864603989e-07,
      "loss": 0.6837,
      "step": 57235
    },
    {
      "epoch": 0.8911237905454319,
      "grad_norm": 4.035880088806152,
      "learning_rate": 1.1461627964143492e-07,
      "loss": 0.7316,
      "step": 57240
    },
    {
      "epoch": 0.8912016315474013,
      "grad_norm": 3.9882264137268066,
      "learning_rate": 1.1453434063682994e-07,
      "loss": 0.7102,
      "step": 57245
    },
    {
      "epoch": 0.8912794725493707,
      "grad_norm": 3.1678626537323,
      "learning_rate": 1.1445240163222497e-07,
      "loss": 0.6376,
      "step": 57250
    },
    {
      "epoch": 0.8913573135513401,
      "grad_norm": 3.9376437664031982,
      "learning_rate": 1.1437046262762e-07,
      "loss": 0.9326,
      "step": 57255
    },
    {
      "epoch": 0.8914351545533095,
      "grad_norm": 5.098526477813721,
      "learning_rate": 1.1428852362301502e-07,
      "loss": 0.7276,
      "step": 57260
    },
    {
      "epoch": 0.8915129955552787,
      "grad_norm": 2.940840005874634,
      "learning_rate": 1.1420658461841006e-07,
      "loss": 0.6517,
      "step": 57265
    },
    {
      "epoch": 0.8915908365572481,
      "grad_norm": 5.619111061096191,
      "learning_rate": 1.1412464561380508e-07,
      "loss": 0.8323,
      "step": 57270
    },
    {
      "epoch": 0.8916686775592175,
      "grad_norm": 4.063543319702148,
      "learning_rate": 1.1404270660920011e-07,
      "loss": 0.7424,
      "step": 57275
    },
    {
      "epoch": 0.8917465185611869,
      "grad_norm": 3.3016602993011475,
      "learning_rate": 1.1396076760459514e-07,
      "loss": 0.73,
      "step": 57280
    },
    {
      "epoch": 0.8918243595631563,
      "grad_norm": 8.569050788879395,
      "learning_rate": 1.1387882859999016e-07,
      "loss": 0.6642,
      "step": 57285
    },
    {
      "epoch": 0.8919022005651257,
      "grad_norm": 6.699550628662109,
      "learning_rate": 1.137968895953852e-07,
      "loss": 0.7135,
      "step": 57290
    },
    {
      "epoch": 0.8919800415670951,
      "grad_norm": 3.5775251388549805,
      "learning_rate": 1.1371495059078022e-07,
      "loss": 0.7418,
      "step": 57295
    },
    {
      "epoch": 0.8920578825690644,
      "grad_norm": 3.575989246368408,
      "learning_rate": 1.1363301158617525e-07,
      "loss": 0.7592,
      "step": 57300
    },
    {
      "epoch": 0.8921357235710338,
      "grad_norm": 2.6908063888549805,
      "learning_rate": 1.1355107258157027e-07,
      "loss": 0.7138,
      "step": 57305
    },
    {
      "epoch": 0.8922135645730032,
      "grad_norm": 5.4316511154174805,
      "learning_rate": 1.1346913357696531e-07,
      "loss": 0.7351,
      "step": 57310
    },
    {
      "epoch": 0.8922914055749726,
      "grad_norm": 3.52114200592041,
      "learning_rate": 1.1338719457236033e-07,
      "loss": 0.7362,
      "step": 57315
    },
    {
      "epoch": 0.8923692465769419,
      "grad_norm": 2.670501470565796,
      "learning_rate": 1.1330525556775536e-07,
      "loss": 0.6611,
      "step": 57320
    },
    {
      "epoch": 0.8924470875789113,
      "grad_norm": 3.456374168395996,
      "learning_rate": 1.1322331656315039e-07,
      "loss": 0.6104,
      "step": 57325
    },
    {
      "epoch": 0.8925249285808807,
      "grad_norm": 2.4759929180145264,
      "learning_rate": 1.1314137755854541e-07,
      "loss": 0.5994,
      "step": 57330
    },
    {
      "epoch": 0.8926027695828501,
      "grad_norm": 3.068270683288574,
      "learning_rate": 1.1305943855394044e-07,
      "loss": 0.6868,
      "step": 57335
    },
    {
      "epoch": 0.8926806105848194,
      "grad_norm": 3.8718578815460205,
      "learning_rate": 1.1297749954933547e-07,
      "loss": 0.8762,
      "step": 57340
    },
    {
      "epoch": 0.8927584515867888,
      "grad_norm": 3.2491698265075684,
      "learning_rate": 1.128955605447305e-07,
      "loss": 0.7516,
      "step": 57345
    },
    {
      "epoch": 0.8928362925887582,
      "grad_norm": 4.494777679443359,
      "learning_rate": 1.1281362154012552e-07,
      "loss": 0.6716,
      "step": 57350
    },
    {
      "epoch": 0.8929141335907276,
      "grad_norm": 3.561673402786255,
      "learning_rate": 1.1273168253552056e-07,
      "loss": 0.637,
      "step": 57355
    },
    {
      "epoch": 0.892991974592697,
      "grad_norm": 5.209887504577637,
      "learning_rate": 1.1264974353091558e-07,
      "loss": 0.7055,
      "step": 57360
    },
    {
      "epoch": 0.8930698155946664,
      "grad_norm": 5.793180465698242,
      "learning_rate": 1.1256780452631061e-07,
      "loss": 0.6689,
      "step": 57365
    },
    {
      "epoch": 0.8931476565966358,
      "grad_norm": 3.1091668605804443,
      "learning_rate": 1.1248586552170564e-07,
      "loss": 0.712,
      "step": 57370
    },
    {
      "epoch": 0.893225497598605,
      "grad_norm": 3.3181445598602295,
      "learning_rate": 1.1240392651710066e-07,
      "loss": 0.7824,
      "step": 57375
    },
    {
      "epoch": 0.8933033386005744,
      "grad_norm": 4.765718936920166,
      "learning_rate": 1.123219875124957e-07,
      "loss": 0.7563,
      "step": 57380
    },
    {
      "epoch": 0.8933811796025438,
      "grad_norm": 2.513098955154419,
      "learning_rate": 1.1224004850789072e-07,
      "loss": 0.6825,
      "step": 57385
    },
    {
      "epoch": 0.8934590206045132,
      "grad_norm": 3.450626850128174,
      "learning_rate": 1.1215810950328575e-07,
      "loss": 0.7243,
      "step": 57390
    },
    {
      "epoch": 0.8935368616064826,
      "grad_norm": 4.055398941040039,
      "learning_rate": 1.1207617049868077e-07,
      "loss": 0.7077,
      "step": 57395
    },
    {
      "epoch": 0.893614702608452,
      "grad_norm": 2.8612630367279053,
      "learning_rate": 1.1199423149407581e-07,
      "loss": 0.7172,
      "step": 57400
    },
    {
      "epoch": 0.8936925436104214,
      "grad_norm": 1.9220513105392456,
      "learning_rate": 1.1191229248947083e-07,
      "loss": 0.5762,
      "step": 57405
    },
    {
      "epoch": 0.8937703846123908,
      "grad_norm": 2.6306991577148438,
      "learning_rate": 1.1183035348486585e-07,
      "loss": 0.7358,
      "step": 57410
    },
    {
      "epoch": 0.8938482256143601,
      "grad_norm": 4.040414810180664,
      "learning_rate": 1.1174841448026089e-07,
      "loss": 0.764,
      "step": 57415
    },
    {
      "epoch": 0.8939260666163295,
      "grad_norm": 4.7756171226501465,
      "learning_rate": 1.1166647547565591e-07,
      "loss": 0.858,
      "step": 57420
    },
    {
      "epoch": 0.8940039076182988,
      "grad_norm": 4.155393123626709,
      "learning_rate": 1.1158453647105094e-07,
      "loss": 0.698,
      "step": 57425
    },
    {
      "epoch": 0.8940817486202682,
      "grad_norm": 3.0036075115203857,
      "learning_rate": 1.1150259746644597e-07,
      "loss": 0.742,
      "step": 57430
    },
    {
      "epoch": 0.8941595896222376,
      "grad_norm": 5.474459171295166,
      "learning_rate": 1.11420658461841e-07,
      "loss": 0.6298,
      "step": 57435
    },
    {
      "epoch": 0.894237430624207,
      "grad_norm": 3.8892621994018555,
      "learning_rate": 1.1133871945723602e-07,
      "loss": 0.7713,
      "step": 57440
    },
    {
      "epoch": 0.8943152716261764,
      "grad_norm": 2.3584649562835693,
      "learning_rate": 1.1125678045263106e-07,
      "loss": 0.8857,
      "step": 57445
    },
    {
      "epoch": 0.8943931126281458,
      "grad_norm": 6.03839111328125,
      "learning_rate": 1.1117484144802608e-07,
      "loss": 0.7307,
      "step": 57450
    },
    {
      "epoch": 0.8944709536301151,
      "grad_norm": 3.5136821269989014,
      "learning_rate": 1.110929024434211e-07,
      "loss": 0.7985,
      "step": 57455
    },
    {
      "epoch": 0.8945487946320845,
      "grad_norm": 4.049898624420166,
      "learning_rate": 1.1101096343881614e-07,
      "loss": 0.7085,
      "step": 57460
    },
    {
      "epoch": 0.8946266356340539,
      "grad_norm": 8.444622039794922,
      "learning_rate": 1.1092902443421116e-07,
      "loss": 0.778,
      "step": 57465
    },
    {
      "epoch": 0.8947044766360233,
      "grad_norm": 3.844912528991699,
      "learning_rate": 1.108470854296062e-07,
      "loss": 0.7468,
      "step": 57470
    },
    {
      "epoch": 0.8947823176379927,
      "grad_norm": 3.985010862350464,
      "learning_rate": 1.1076514642500122e-07,
      "loss": 0.7209,
      "step": 57475
    },
    {
      "epoch": 0.894860158639962,
      "grad_norm": 4.156490802764893,
      "learning_rate": 1.1068320742039625e-07,
      "loss": 0.7601,
      "step": 57480
    },
    {
      "epoch": 0.8949379996419314,
      "grad_norm": 3.7230217456817627,
      "learning_rate": 1.1060126841579127e-07,
      "loss": 0.9007,
      "step": 57485
    },
    {
      "epoch": 0.8950158406439007,
      "grad_norm": 2.780439853668213,
      "learning_rate": 1.105193294111863e-07,
      "loss": 0.6875,
      "step": 57490
    },
    {
      "epoch": 0.8950936816458701,
      "grad_norm": 3.06766676902771,
      "learning_rate": 1.1043739040658133e-07,
      "loss": 0.6999,
      "step": 57495
    },
    {
      "epoch": 0.8951715226478395,
      "grad_norm": 3.6182456016540527,
      "learning_rate": 1.1035545140197635e-07,
      "loss": 0.7871,
      "step": 57500
    },
    {
      "epoch": 0.8952493636498089,
      "grad_norm": 5.694599151611328,
      "learning_rate": 1.1027351239737139e-07,
      "loss": 0.7144,
      "step": 57505
    },
    {
      "epoch": 0.8953272046517783,
      "grad_norm": 2.853363275527954,
      "learning_rate": 1.1019157339276642e-07,
      "loss": 0.7214,
      "step": 57510
    },
    {
      "epoch": 0.8954050456537477,
      "grad_norm": 2.9803647994995117,
      "learning_rate": 1.1010963438816146e-07,
      "loss": 0.6671,
      "step": 57515
    },
    {
      "epoch": 0.8954828866557171,
      "grad_norm": 3.8740122318267822,
      "learning_rate": 1.1002769538355648e-07,
      "loss": 0.7563,
      "step": 57520
    },
    {
      "epoch": 0.8955607276576865,
      "grad_norm": 6.7562408447265625,
      "learning_rate": 1.0994575637895151e-07,
      "loss": 0.6616,
      "step": 57525
    },
    {
      "epoch": 0.8956385686596557,
      "grad_norm": 5.084529399871826,
      "learning_rate": 1.0986381737434654e-07,
      "loss": 0.6789,
      "step": 57530
    },
    {
      "epoch": 0.8957164096616251,
      "grad_norm": 3.4171345233917236,
      "learning_rate": 1.0978187836974156e-07,
      "loss": 0.7922,
      "step": 57535
    },
    {
      "epoch": 0.8957942506635945,
      "grad_norm": 3.3058269023895264,
      "learning_rate": 1.096999393651366e-07,
      "loss": 0.8544,
      "step": 57540
    },
    {
      "epoch": 0.8958720916655639,
      "grad_norm": 3.4105091094970703,
      "learning_rate": 1.0961800036053162e-07,
      "loss": 0.7771,
      "step": 57545
    },
    {
      "epoch": 0.8959499326675333,
      "grad_norm": 4.137765407562256,
      "learning_rate": 1.0953606135592665e-07,
      "loss": 0.7801,
      "step": 57550
    },
    {
      "epoch": 0.8960277736695027,
      "grad_norm": 3.982168436050415,
      "learning_rate": 1.0945412235132167e-07,
      "loss": 0.8234,
      "step": 57555
    },
    {
      "epoch": 0.8961056146714721,
      "grad_norm": 4.468942642211914,
      "learning_rate": 1.0937218334671671e-07,
      "loss": 0.6583,
      "step": 57560
    },
    {
      "epoch": 0.8961834556734414,
      "grad_norm": 3.5043623447418213,
      "learning_rate": 1.0929024434211173e-07,
      "loss": 0.7077,
      "step": 57565
    },
    {
      "epoch": 0.8962612966754108,
      "grad_norm": 3.372105121612549,
      "learning_rate": 1.0920830533750677e-07,
      "loss": 0.5912,
      "step": 57570
    },
    {
      "epoch": 0.8963391376773802,
      "grad_norm": 4.104984283447266,
      "learning_rate": 1.0912636633290179e-07,
      "loss": 0.6872,
      "step": 57575
    },
    {
      "epoch": 0.8964169786793496,
      "grad_norm": 6.2466840744018555,
      "learning_rate": 1.0904442732829681e-07,
      "loss": 0.7152,
      "step": 57580
    },
    {
      "epoch": 0.8964948196813189,
      "grad_norm": 2.8765738010406494,
      "learning_rate": 1.0896248832369184e-07,
      "loss": 0.7052,
      "step": 57585
    },
    {
      "epoch": 0.8965726606832883,
      "grad_norm": 3.2874560356140137,
      "learning_rate": 1.0888054931908687e-07,
      "loss": 0.7039,
      "step": 57590
    },
    {
      "epoch": 0.8966505016852577,
      "grad_norm": 2.7780866622924805,
      "learning_rate": 1.087986103144819e-07,
      "loss": 0.7393,
      "step": 57595
    },
    {
      "epoch": 0.8967283426872271,
      "grad_norm": 2.8969168663024902,
      "learning_rate": 1.0871667130987692e-07,
      "loss": 0.7339,
      "step": 57600
    },
    {
      "epoch": 0.8968061836891964,
      "grad_norm": 3.7082533836364746,
      "learning_rate": 1.0863473230527196e-07,
      "loss": 0.7219,
      "step": 57605
    },
    {
      "epoch": 0.8968840246911658,
      "grad_norm": 4.138474464416504,
      "learning_rate": 1.0855279330066698e-07,
      "loss": 0.7974,
      "step": 57610
    },
    {
      "epoch": 0.8969618656931352,
      "grad_norm": 4.415271759033203,
      "learning_rate": 1.0847085429606202e-07,
      "loss": 0.758,
      "step": 57615
    },
    {
      "epoch": 0.8970397066951046,
      "grad_norm": 4.883201599121094,
      "learning_rate": 1.0838891529145704e-07,
      "loss": 0.7889,
      "step": 57620
    },
    {
      "epoch": 0.897117547697074,
      "grad_norm": 5.321115970611572,
      "learning_rate": 1.0830697628685206e-07,
      "loss": 0.7447,
      "step": 57625
    },
    {
      "epoch": 0.8971953886990434,
      "grad_norm": 3.164673089981079,
      "learning_rate": 1.082250372822471e-07,
      "loss": 0.7558,
      "step": 57630
    },
    {
      "epoch": 0.8972732297010128,
      "grad_norm": 4.606579780578613,
      "learning_rate": 1.0814309827764212e-07,
      "loss": 0.8639,
      "step": 57635
    },
    {
      "epoch": 0.897351070702982,
      "grad_norm": 5.195619583129883,
      "learning_rate": 1.0806115927303715e-07,
      "loss": 0.7223,
      "step": 57640
    },
    {
      "epoch": 0.8974289117049514,
      "grad_norm": 3.0141377449035645,
      "learning_rate": 1.0797922026843217e-07,
      "loss": 0.7531,
      "step": 57645
    },
    {
      "epoch": 0.8975067527069208,
      "grad_norm": 2.7319235801696777,
      "learning_rate": 1.0789728126382721e-07,
      "loss": 0.6643,
      "step": 57650
    },
    {
      "epoch": 0.8975845937088902,
      "grad_norm": 3.350571870803833,
      "learning_rate": 1.0781534225922223e-07,
      "loss": 0.6992,
      "step": 57655
    },
    {
      "epoch": 0.8976624347108596,
      "grad_norm": 3.239872694015503,
      "learning_rate": 1.0773340325461725e-07,
      "loss": 0.6555,
      "step": 57660
    },
    {
      "epoch": 0.897740275712829,
      "grad_norm": 4.393332481384277,
      "learning_rate": 1.0765146425001229e-07,
      "loss": 0.7276,
      "step": 57665
    },
    {
      "epoch": 0.8978181167147984,
      "grad_norm": 12.68930435180664,
      "learning_rate": 1.0756952524540731e-07,
      "loss": 0.7018,
      "step": 57670
    },
    {
      "epoch": 0.8978959577167678,
      "grad_norm": 4.893350124359131,
      "learning_rate": 1.0748758624080234e-07,
      "loss": 0.6651,
      "step": 57675
    },
    {
      "epoch": 0.8979737987187371,
      "grad_norm": 2.8646328449249268,
      "learning_rate": 1.0740564723619737e-07,
      "loss": 0.7458,
      "step": 57680
    },
    {
      "epoch": 0.8980516397207065,
      "grad_norm": 4.575784206390381,
      "learning_rate": 1.073237082315924e-07,
      "loss": 0.6936,
      "step": 57685
    },
    {
      "epoch": 0.8981294807226758,
      "grad_norm": 4.252884387969971,
      "learning_rate": 1.0724176922698742e-07,
      "loss": 0.7518,
      "step": 57690
    },
    {
      "epoch": 0.8982073217246452,
      "grad_norm": 3.1084601879119873,
      "learning_rate": 1.0715983022238246e-07,
      "loss": 0.721,
      "step": 57695
    },
    {
      "epoch": 0.8982851627266146,
      "grad_norm": 5.915571212768555,
      "learning_rate": 1.0707789121777748e-07,
      "loss": 0.8384,
      "step": 57700
    },
    {
      "epoch": 0.898363003728584,
      "grad_norm": 5.294565200805664,
      "learning_rate": 1.069959522131725e-07,
      "loss": 0.7165,
      "step": 57705
    },
    {
      "epoch": 0.8984408447305534,
      "grad_norm": 3.5233941078186035,
      "learning_rate": 1.0691401320856754e-07,
      "loss": 0.8077,
      "step": 57710
    },
    {
      "epoch": 0.8985186857325228,
      "grad_norm": 7.192102909088135,
      "learning_rate": 1.0683207420396256e-07,
      "loss": 0.7009,
      "step": 57715
    },
    {
      "epoch": 0.8985965267344921,
      "grad_norm": 3.3756299018859863,
      "learning_rate": 1.067501351993576e-07,
      "loss": 0.8308,
      "step": 57720
    },
    {
      "epoch": 0.8986743677364615,
      "grad_norm": 2.656552314758301,
      "learning_rate": 1.0666819619475262e-07,
      "loss": 0.8116,
      "step": 57725
    },
    {
      "epoch": 0.8987522087384309,
      "grad_norm": 5.02744722366333,
      "learning_rate": 1.0658625719014765e-07,
      "loss": 0.7091,
      "step": 57730
    },
    {
      "epoch": 0.8988300497404003,
      "grad_norm": 5.639800071716309,
      "learning_rate": 1.0650431818554267e-07,
      "loss": 0.8012,
      "step": 57735
    },
    {
      "epoch": 0.8989078907423697,
      "grad_norm": 7.515092372894287,
      "learning_rate": 1.064223791809377e-07,
      "loss": 0.7019,
      "step": 57740
    },
    {
      "epoch": 0.898985731744339,
      "grad_norm": 4.749237060546875,
      "learning_rate": 1.0634044017633273e-07,
      "loss": 0.8066,
      "step": 57745
    },
    {
      "epoch": 0.8990635727463084,
      "grad_norm": 4.598395824432373,
      "learning_rate": 1.0625850117172775e-07,
      "loss": 0.7012,
      "step": 57750
    },
    {
      "epoch": 0.8991414137482777,
      "grad_norm": 7.911144256591797,
      "learning_rate": 1.0617656216712279e-07,
      "loss": 0.7742,
      "step": 57755
    },
    {
      "epoch": 0.8992192547502471,
      "grad_norm": 2.734304189682007,
      "learning_rate": 1.0609462316251781e-07,
      "loss": 0.721,
      "step": 57760
    },
    {
      "epoch": 0.8992970957522165,
      "grad_norm": 3.0176610946655273,
      "learning_rate": 1.0601268415791284e-07,
      "loss": 0.6652,
      "step": 57765
    },
    {
      "epoch": 0.8993749367541859,
      "grad_norm": 5.547643184661865,
      "learning_rate": 1.0593074515330787e-07,
      "loss": 0.768,
      "step": 57770
    },
    {
      "epoch": 0.8994527777561553,
      "grad_norm": 5.317986965179443,
      "learning_rate": 1.058488061487029e-07,
      "loss": 0.7725,
      "step": 57775
    },
    {
      "epoch": 0.8995306187581247,
      "grad_norm": 3.079225540161133,
      "learning_rate": 1.0576686714409792e-07,
      "loss": 0.696,
      "step": 57780
    },
    {
      "epoch": 0.8996084597600941,
      "grad_norm": 3.780205249786377,
      "learning_rate": 1.0568492813949295e-07,
      "loss": 0.6366,
      "step": 57785
    },
    {
      "epoch": 0.8996863007620635,
      "grad_norm": 3.322993040084839,
      "learning_rate": 1.0560298913488798e-07,
      "loss": 0.7285,
      "step": 57790
    },
    {
      "epoch": 0.8997641417640327,
      "grad_norm": 3.284734010696411,
      "learning_rate": 1.05521050130283e-07,
      "loss": 0.7905,
      "step": 57795
    },
    {
      "epoch": 0.8998419827660021,
      "grad_norm": 3.1243081092834473,
      "learning_rate": 1.0543911112567804e-07,
      "loss": 0.7119,
      "step": 57800
    },
    {
      "epoch": 0.8999198237679715,
      "grad_norm": 3.482285261154175,
      "learning_rate": 1.0535717212107306e-07,
      "loss": 0.7178,
      "step": 57805
    },
    {
      "epoch": 0.8999976647699409,
      "grad_norm": 3.754448652267456,
      "learning_rate": 1.052752331164681e-07,
      "loss": 0.7431,
      "step": 57810
    },
    {
      "epoch": 0.9000755057719103,
      "grad_norm": 3.3602914810180664,
      "learning_rate": 1.0519329411186312e-07,
      "loss": 0.7424,
      "step": 57815
    },
    {
      "epoch": 0.9001533467738797,
      "grad_norm": 6.402689456939697,
      "learning_rate": 1.0511135510725814e-07,
      "loss": 0.7568,
      "step": 57820
    },
    {
      "epoch": 0.9002311877758491,
      "grad_norm": 3.6967082023620605,
      "learning_rate": 1.0502941610265317e-07,
      "loss": 0.7915,
      "step": 57825
    },
    {
      "epoch": 0.9003090287778184,
      "grad_norm": 3.6651995182037354,
      "learning_rate": 1.0494747709804821e-07,
      "loss": 0.7744,
      "step": 57830
    },
    {
      "epoch": 0.9003868697797878,
      "grad_norm": 3.1688883304595947,
      "learning_rate": 1.0486553809344324e-07,
      "loss": 0.7436,
      "step": 57835
    },
    {
      "epoch": 0.9004647107817572,
      "grad_norm": 4.0572381019592285,
      "learning_rate": 1.0478359908883827e-07,
      "loss": 0.642,
      "step": 57840
    },
    {
      "epoch": 0.9005425517837266,
      "grad_norm": 2.759850263595581,
      "learning_rate": 1.047016600842333e-07,
      "loss": 0.731,
      "step": 57845
    },
    {
      "epoch": 0.900620392785696,
      "grad_norm": 3.2324702739715576,
      "learning_rate": 1.0461972107962832e-07,
      "loss": 0.658,
      "step": 57850
    },
    {
      "epoch": 0.9006982337876653,
      "grad_norm": 4.262421131134033,
      "learning_rate": 1.0453778207502336e-07,
      "loss": 0.8066,
      "step": 57855
    },
    {
      "epoch": 0.9007760747896347,
      "grad_norm": 3.8865559101104736,
      "learning_rate": 1.0445584307041838e-07,
      "loss": 0.7354,
      "step": 57860
    },
    {
      "epoch": 0.9008539157916041,
      "grad_norm": 3.4689502716064453,
      "learning_rate": 1.043739040658134e-07,
      "loss": 0.764,
      "step": 57865
    },
    {
      "epoch": 0.9009317567935734,
      "grad_norm": 4.260305881500244,
      "learning_rate": 1.0429196506120844e-07,
      "loss": 0.7712,
      "step": 57870
    },
    {
      "epoch": 0.9010095977955428,
      "grad_norm": 4.3983073234558105,
      "learning_rate": 1.0421002605660346e-07,
      "loss": 0.7301,
      "step": 57875
    },
    {
      "epoch": 0.9010874387975122,
      "grad_norm": 5.357254981994629,
      "learning_rate": 1.041280870519985e-07,
      "loss": 0.7848,
      "step": 57880
    },
    {
      "epoch": 0.9011652797994816,
      "grad_norm": 3.860236406326294,
      "learning_rate": 1.0404614804739352e-07,
      "loss": 0.7011,
      "step": 57885
    },
    {
      "epoch": 0.901243120801451,
      "grad_norm": 3.3992600440979004,
      "learning_rate": 1.0396420904278855e-07,
      "loss": 0.7439,
      "step": 57890
    },
    {
      "epoch": 0.9013209618034204,
      "grad_norm": 5.4626240730285645,
      "learning_rate": 1.0388227003818357e-07,
      "loss": 0.6146,
      "step": 57895
    },
    {
      "epoch": 0.9013988028053898,
      "grad_norm": 5.185338497161865,
      "learning_rate": 1.0380033103357861e-07,
      "loss": 0.805,
      "step": 57900
    },
    {
      "epoch": 0.901476643807359,
      "grad_norm": 3.381221055984497,
      "learning_rate": 1.0371839202897363e-07,
      "loss": 0.7674,
      "step": 57905
    },
    {
      "epoch": 0.9015544848093284,
      "grad_norm": 3.732612371444702,
      "learning_rate": 1.0363645302436865e-07,
      "loss": 0.6877,
      "step": 57910
    },
    {
      "epoch": 0.9016323258112978,
      "grad_norm": 3.452061653137207,
      "learning_rate": 1.0355451401976369e-07,
      "loss": 0.8248,
      "step": 57915
    },
    {
      "epoch": 0.9017101668132672,
      "grad_norm": 3.045064926147461,
      "learning_rate": 1.0347257501515871e-07,
      "loss": 0.7525,
      "step": 57920
    },
    {
      "epoch": 0.9017880078152366,
      "grad_norm": 9.50442123413086,
      "learning_rate": 1.0339063601055374e-07,
      "loss": 0.6664,
      "step": 57925
    },
    {
      "epoch": 0.901865848817206,
      "grad_norm": 7.321946144104004,
      "learning_rate": 1.0330869700594877e-07,
      "loss": 0.736,
      "step": 57930
    },
    {
      "epoch": 0.9019436898191754,
      "grad_norm": 7.49962043762207,
      "learning_rate": 1.032267580013438e-07,
      "loss": 0.7328,
      "step": 57935
    },
    {
      "epoch": 0.9020215308211448,
      "grad_norm": 3.0252773761749268,
      "learning_rate": 1.0314481899673882e-07,
      "loss": 0.7542,
      "step": 57940
    },
    {
      "epoch": 0.9020993718231141,
      "grad_norm": 3.825117588043213,
      "learning_rate": 1.0306287999213386e-07,
      "loss": 0.7672,
      "step": 57945
    },
    {
      "epoch": 0.9021772128250835,
      "grad_norm": 5.334710597991943,
      "learning_rate": 1.0298094098752888e-07,
      "loss": 0.5988,
      "step": 57950
    },
    {
      "epoch": 0.9022550538270528,
      "grad_norm": 4.147637844085693,
      "learning_rate": 1.028990019829239e-07,
      "loss": 0.7536,
      "step": 57955
    },
    {
      "epoch": 0.9023328948290222,
      "grad_norm": 3.1427197456359863,
      "learning_rate": 1.0281706297831894e-07,
      "loss": 0.7941,
      "step": 57960
    },
    {
      "epoch": 0.9024107358309916,
      "grad_norm": 2.240316390991211,
      "learning_rate": 1.0273512397371396e-07,
      "loss": 0.7519,
      "step": 57965
    },
    {
      "epoch": 0.902488576832961,
      "grad_norm": 6.411386489868164,
      "learning_rate": 1.02653184969109e-07,
      "loss": 0.7278,
      "step": 57970
    },
    {
      "epoch": 0.9025664178349304,
      "grad_norm": 2.4825263023376465,
      "learning_rate": 1.0257124596450402e-07,
      "loss": 0.6694,
      "step": 57975
    },
    {
      "epoch": 0.9026442588368997,
      "grad_norm": 4.469844341278076,
      "learning_rate": 1.0248930695989905e-07,
      "loss": 0.7252,
      "step": 57980
    },
    {
      "epoch": 0.9027220998388691,
      "grad_norm": 4.454055309295654,
      "learning_rate": 1.0240736795529407e-07,
      "loss": 0.7259,
      "step": 57985
    },
    {
      "epoch": 0.9027999408408385,
      "grad_norm": 5.1197967529296875,
      "learning_rate": 1.023254289506891e-07,
      "loss": 0.6249,
      "step": 57990
    },
    {
      "epoch": 0.9028777818428079,
      "grad_norm": 2.979224920272827,
      "learning_rate": 1.0224348994608413e-07,
      "loss": 0.7038,
      "step": 57995
    },
    {
      "epoch": 0.9029556228447773,
      "grad_norm": 4.926724433898926,
      "learning_rate": 1.0216155094147915e-07,
      "loss": 0.6986,
      "step": 58000
    },
    {
      "epoch": 0.9030334638467467,
      "grad_norm": 5.868091583251953,
      "learning_rate": 1.0207961193687419e-07,
      "loss": 0.6848,
      "step": 58005
    },
    {
      "epoch": 0.903111304848716,
      "grad_norm": 5.8929877281188965,
      "learning_rate": 1.0199767293226921e-07,
      "loss": 0.8704,
      "step": 58010
    },
    {
      "epoch": 0.9031891458506854,
      "grad_norm": 5.2499098777771,
      "learning_rate": 1.0191573392766425e-07,
      "loss": 0.7972,
      "step": 58015
    },
    {
      "epoch": 0.9032669868526547,
      "grad_norm": 3.5320451259613037,
      "learning_rate": 1.0183379492305927e-07,
      "loss": 0.6343,
      "step": 58020
    },
    {
      "epoch": 0.9033448278546241,
      "grad_norm": 6.147691249847412,
      "learning_rate": 1.017518559184543e-07,
      "loss": 0.7542,
      "step": 58025
    },
    {
      "epoch": 0.9034226688565935,
      "grad_norm": 5.113248825073242,
      "learning_rate": 1.0166991691384932e-07,
      "loss": 0.6722,
      "step": 58030
    },
    {
      "epoch": 0.9035005098585629,
      "grad_norm": 5.709242820739746,
      "learning_rate": 1.0158797790924435e-07,
      "loss": 0.7202,
      "step": 58035
    },
    {
      "epoch": 0.9035783508605323,
      "grad_norm": 2.543210029602051,
      "learning_rate": 1.0150603890463938e-07,
      "loss": 0.7479,
      "step": 58040
    },
    {
      "epoch": 0.9036561918625017,
      "grad_norm": 5.981240272521973,
      "learning_rate": 1.014240999000344e-07,
      "loss": 0.691,
      "step": 58045
    },
    {
      "epoch": 0.9037340328644711,
      "grad_norm": 3.4087166786193848,
      "learning_rate": 1.0134216089542944e-07,
      "loss": 0.6395,
      "step": 58050
    },
    {
      "epoch": 0.9038118738664405,
      "grad_norm": 3.359178304672241,
      "learning_rate": 1.0126022189082446e-07,
      "loss": 0.6205,
      "step": 58055
    },
    {
      "epoch": 0.9038897148684097,
      "grad_norm": 3.41044282913208,
      "learning_rate": 1.011782828862195e-07,
      "loss": 0.7161,
      "step": 58060
    },
    {
      "epoch": 0.9039675558703791,
      "grad_norm": 3.529592275619507,
      "learning_rate": 1.0109634388161452e-07,
      "loss": 0.7788,
      "step": 58065
    },
    {
      "epoch": 0.9040453968723485,
      "grad_norm": 3.7278499603271484,
      "learning_rate": 1.0101440487700954e-07,
      "loss": 0.8384,
      "step": 58070
    },
    {
      "epoch": 0.9041232378743179,
      "grad_norm": 6.026160717010498,
      "learning_rate": 1.0093246587240457e-07,
      "loss": 0.7795,
      "step": 58075
    },
    {
      "epoch": 0.9042010788762873,
      "grad_norm": 6.313095569610596,
      "learning_rate": 1.008505268677996e-07,
      "loss": 0.6922,
      "step": 58080
    },
    {
      "epoch": 0.9042789198782567,
      "grad_norm": 2.763465404510498,
      "learning_rate": 1.0076858786319463e-07,
      "loss": 0.7402,
      "step": 58085
    },
    {
      "epoch": 0.9043567608802261,
      "grad_norm": 5.519827842712402,
      "learning_rate": 1.0068664885858965e-07,
      "loss": 0.6933,
      "step": 58090
    },
    {
      "epoch": 0.9044346018821954,
      "grad_norm": 5.627171993255615,
      "learning_rate": 1.0060470985398469e-07,
      "loss": 0.6424,
      "step": 58095
    },
    {
      "epoch": 0.9045124428841648,
      "grad_norm": 4.492709159851074,
      "learning_rate": 1.0052277084937971e-07,
      "loss": 0.5783,
      "step": 58100
    },
    {
      "epoch": 0.9045902838861342,
      "grad_norm": 5.980396747589111,
      "learning_rate": 1.0044083184477475e-07,
      "loss": 0.7976,
      "step": 58105
    },
    {
      "epoch": 0.9046681248881036,
      "grad_norm": 4.588638782501221,
      "learning_rate": 1.0035889284016977e-07,
      "loss": 0.7494,
      "step": 58110
    },
    {
      "epoch": 0.904745965890073,
      "grad_norm": 3.3866443634033203,
      "learning_rate": 1.0027695383556479e-07,
      "loss": 0.6496,
      "step": 58115
    },
    {
      "epoch": 0.9048238068920423,
      "grad_norm": 3.664867401123047,
      "learning_rate": 1.0019501483095982e-07,
      "loss": 0.6294,
      "step": 58120
    },
    {
      "epoch": 0.9049016478940117,
      "grad_norm": 10.318333625793457,
      "learning_rate": 1.0011307582635485e-07,
      "loss": 0.8204,
      "step": 58125
    },
    {
      "epoch": 0.9049794888959811,
      "grad_norm": 6.197859287261963,
      "learning_rate": 1.0003113682174988e-07,
      "loss": 0.7917,
      "step": 58130
    },
    {
      "epoch": 0.9050573298979504,
      "grad_norm": 7.352142333984375,
      "learning_rate": 9.99491978171449e-08,
      "loss": 0.7276,
      "step": 58135
    },
    {
      "epoch": 0.9051351708999198,
      "grad_norm": 4.499242305755615,
      "learning_rate": 9.986725881253994e-08,
      "loss": 0.6968,
      "step": 58140
    },
    {
      "epoch": 0.9052130119018892,
      "grad_norm": 3.2552754878997803,
      "learning_rate": 9.978531980793497e-08,
      "loss": 0.6918,
      "step": 58145
    },
    {
      "epoch": 0.9052908529038586,
      "grad_norm": 4.001320838928223,
      "learning_rate": 9.970338080333001e-08,
      "loss": 0.8099,
      "step": 58150
    },
    {
      "epoch": 0.905368693905828,
      "grad_norm": 4.417248725891113,
      "learning_rate": 9.962144179872503e-08,
      "loss": 0.8387,
      "step": 58155
    },
    {
      "epoch": 0.9054465349077974,
      "grad_norm": 3.50768780708313,
      "learning_rate": 9.953950279412005e-08,
      "loss": 0.7843,
      "step": 58160
    },
    {
      "epoch": 0.9055243759097668,
      "grad_norm": 4.8534088134765625,
      "learning_rate": 9.945756378951509e-08,
      "loss": 0.765,
      "step": 58165
    },
    {
      "epoch": 0.905602216911736,
      "grad_norm": 5.621729850769043,
      "learning_rate": 9.937562478491011e-08,
      "loss": 0.6861,
      "step": 58170
    },
    {
      "epoch": 0.9056800579137054,
      "grad_norm": 2.636119842529297,
      "learning_rate": 9.929368578030515e-08,
      "loss": 0.7527,
      "step": 58175
    },
    {
      "epoch": 0.9057578989156748,
      "grad_norm": 2.8544371128082275,
      "learning_rate": 9.921174677570017e-08,
      "loss": 0.7367,
      "step": 58180
    },
    {
      "epoch": 0.9058357399176442,
      "grad_norm": 4.486292839050293,
      "learning_rate": 9.91298077710952e-08,
      "loss": 0.7414,
      "step": 58185
    },
    {
      "epoch": 0.9059135809196136,
      "grad_norm": 2.6603941917419434,
      "learning_rate": 9.904786876649022e-08,
      "loss": 0.6968,
      "step": 58190
    },
    {
      "epoch": 0.905991421921583,
      "grad_norm": 3.227323293685913,
      "learning_rate": 9.896592976188526e-08,
      "loss": 0.6655,
      "step": 58195
    },
    {
      "epoch": 0.9060692629235524,
      "grad_norm": 3.667597532272339,
      "learning_rate": 9.888399075728028e-08,
      "loss": 0.6757,
      "step": 58200
    },
    {
      "epoch": 0.9061471039255218,
      "grad_norm": 3.082690477371216,
      "learning_rate": 9.88020517526753e-08,
      "loss": 0.7363,
      "step": 58205
    },
    {
      "epoch": 0.9062249449274911,
      "grad_norm": 3.9041006565093994,
      "learning_rate": 9.872011274807034e-08,
      "loss": 0.6668,
      "step": 58210
    },
    {
      "epoch": 0.9063027859294605,
      "grad_norm": 2.3522095680236816,
      "learning_rate": 9.863817374346536e-08,
      "loss": 0.7361,
      "step": 58215
    },
    {
      "epoch": 0.9063806269314298,
      "grad_norm": 3.4040238857269287,
      "learning_rate": 9.85562347388604e-08,
      "loss": 0.7012,
      "step": 58220
    },
    {
      "epoch": 0.9064584679333992,
      "grad_norm": 3.912466287612915,
      "learning_rate": 9.847429573425542e-08,
      "loss": 0.6928,
      "step": 58225
    },
    {
      "epoch": 0.9065363089353686,
      "grad_norm": 5.638211727142334,
      "learning_rate": 9.839235672965045e-08,
      "loss": 0.6312,
      "step": 58230
    },
    {
      "epoch": 0.906614149937338,
      "grad_norm": 2.951988458633423,
      "learning_rate": 9.831041772504547e-08,
      "loss": 0.7152,
      "step": 58235
    },
    {
      "epoch": 0.9066919909393074,
      "grad_norm": 3.881145477294922,
      "learning_rate": 9.82284787204405e-08,
      "loss": 0.7712,
      "step": 58240
    },
    {
      "epoch": 0.9067698319412767,
      "grad_norm": 4.305324554443359,
      "learning_rate": 9.814653971583553e-08,
      "loss": 0.7066,
      "step": 58245
    },
    {
      "epoch": 0.9068476729432461,
      "grad_norm": 4.413559913635254,
      "learning_rate": 9.806460071123055e-08,
      "loss": 0.7286,
      "step": 58250
    },
    {
      "epoch": 0.9069255139452155,
      "grad_norm": 6.951462745666504,
      "learning_rate": 9.798266170662559e-08,
      "loss": 0.6964,
      "step": 58255
    },
    {
      "epoch": 0.9070033549471849,
      "grad_norm": 2.6716530323028564,
      "learning_rate": 9.790072270202061e-08,
      "loss": 0.6853,
      "step": 58260
    },
    {
      "epoch": 0.9070811959491543,
      "grad_norm": 3.9456987380981445,
      "learning_rate": 9.781878369741565e-08,
      "loss": 0.6881,
      "step": 58265
    },
    {
      "epoch": 0.9071590369511237,
      "grad_norm": 3.9124202728271484,
      "learning_rate": 9.773684469281067e-08,
      "loss": 0.8139,
      "step": 58270
    },
    {
      "epoch": 0.907236877953093,
      "grad_norm": 4.4014129638671875,
      "learning_rate": 9.76549056882057e-08,
      "loss": 0.7818,
      "step": 58275
    },
    {
      "epoch": 0.9073147189550624,
      "grad_norm": 2.785839796066284,
      "learning_rate": 9.757296668360072e-08,
      "loss": 0.6269,
      "step": 58280
    },
    {
      "epoch": 0.9073925599570317,
      "grad_norm": 8.033772468566895,
      "learning_rate": 9.749102767899575e-08,
      "loss": 0.718,
      "step": 58285
    },
    {
      "epoch": 0.9074704009590011,
      "grad_norm": 2.8669869899749756,
      "learning_rate": 9.740908867439078e-08,
      "loss": 0.7828,
      "step": 58290
    },
    {
      "epoch": 0.9075482419609705,
      "grad_norm": 2.580873727798462,
      "learning_rate": 9.73271496697858e-08,
      "loss": 0.6906,
      "step": 58295
    },
    {
      "epoch": 0.9076260829629399,
      "grad_norm": 6.3554534912109375,
      "learning_rate": 9.724521066518084e-08,
      "loss": 0.7982,
      "step": 58300
    },
    {
      "epoch": 0.9077039239649093,
      "grad_norm": 2.6191442012786865,
      "learning_rate": 9.716327166057586e-08,
      "loss": 0.7532,
      "step": 58305
    },
    {
      "epoch": 0.9077817649668787,
      "grad_norm": 3.511552333831787,
      "learning_rate": 9.70813326559709e-08,
      "loss": 0.76,
      "step": 58310
    },
    {
      "epoch": 0.9078596059688481,
      "grad_norm": 5.064708232879639,
      "learning_rate": 9.699939365136592e-08,
      "loss": 0.8099,
      "step": 58315
    },
    {
      "epoch": 0.9079374469708175,
      "grad_norm": 5.111839294433594,
      "learning_rate": 9.691745464676094e-08,
      "loss": 0.7252,
      "step": 58320
    },
    {
      "epoch": 0.9080152879727867,
      "grad_norm": 4.529272079467773,
      "learning_rate": 9.683551564215597e-08,
      "loss": 0.7593,
      "step": 58325
    },
    {
      "epoch": 0.9080931289747561,
      "grad_norm": 9.634515762329102,
      "learning_rate": 9.6753576637551e-08,
      "loss": 0.7749,
      "step": 58330
    },
    {
      "epoch": 0.9081709699767255,
      "grad_norm": 3.0101301670074463,
      "learning_rate": 9.667163763294603e-08,
      "loss": 0.6909,
      "step": 58335
    },
    {
      "epoch": 0.9082488109786949,
      "grad_norm": 2.89302921295166,
      "learning_rate": 9.658969862834105e-08,
      "loss": 0.7526,
      "step": 58340
    },
    {
      "epoch": 0.9083266519806643,
      "grad_norm": 3.2086291313171387,
      "learning_rate": 9.650775962373609e-08,
      "loss": 0.6973,
      "step": 58345
    },
    {
      "epoch": 0.9084044929826337,
      "grad_norm": 12.192477226257324,
      "learning_rate": 9.642582061913111e-08,
      "loss": 0.7288,
      "step": 58350
    },
    {
      "epoch": 0.9084823339846031,
      "grad_norm": 4.141300201416016,
      "learning_rate": 9.634388161452615e-08,
      "loss": 0.6202,
      "step": 58355
    },
    {
      "epoch": 0.9085601749865724,
      "grad_norm": 6.19830322265625,
      "learning_rate": 9.626194260992117e-08,
      "loss": 0.7938,
      "step": 58360
    },
    {
      "epoch": 0.9086380159885418,
      "grad_norm": 3.8098394870758057,
      "learning_rate": 9.618000360531619e-08,
      "loss": 0.8943,
      "step": 58365
    },
    {
      "epoch": 0.9087158569905112,
      "grad_norm": 4.1256022453308105,
      "learning_rate": 9.609806460071123e-08,
      "loss": 0.7359,
      "step": 58370
    },
    {
      "epoch": 0.9087936979924806,
      "grad_norm": 3.8118479251861572,
      "learning_rate": 9.601612559610625e-08,
      "loss": 0.6619,
      "step": 58375
    },
    {
      "epoch": 0.90887153899445,
      "grad_norm": 4.5867533683776855,
      "learning_rate": 9.593418659150128e-08,
      "loss": 0.7967,
      "step": 58380
    },
    {
      "epoch": 0.9089493799964193,
      "grad_norm": 3.165525436401367,
      "learning_rate": 9.58522475868963e-08,
      "loss": 0.7948,
      "step": 58385
    },
    {
      "epoch": 0.9090272209983887,
      "grad_norm": 3.7150235176086426,
      "learning_rate": 9.577030858229134e-08,
      "loss": 0.7398,
      "step": 58390
    },
    {
      "epoch": 0.9091050620003581,
      "grad_norm": 3.183760404586792,
      "learning_rate": 9.568836957768636e-08,
      "loss": 0.7292,
      "step": 58395
    },
    {
      "epoch": 0.9091829030023274,
      "grad_norm": 3.153104782104492,
      "learning_rate": 9.560643057308138e-08,
      "loss": 0.6758,
      "step": 58400
    },
    {
      "epoch": 0.9092607440042968,
      "grad_norm": 3.3848507404327393,
      "learning_rate": 9.552449156847642e-08,
      "loss": 0.8499,
      "step": 58405
    },
    {
      "epoch": 0.9093385850062662,
      "grad_norm": 3.2458643913269043,
      "learning_rate": 9.544255256387144e-08,
      "loss": 0.7139,
      "step": 58410
    },
    {
      "epoch": 0.9094164260082356,
      "grad_norm": 4.5694427490234375,
      "learning_rate": 9.536061355926648e-08,
      "loss": 0.7097,
      "step": 58415
    },
    {
      "epoch": 0.909494267010205,
      "grad_norm": 3.488931894302368,
      "learning_rate": 9.52786745546615e-08,
      "loss": 0.8824,
      "step": 58420
    },
    {
      "epoch": 0.9095721080121744,
      "grad_norm": 4.268072605133057,
      "learning_rate": 9.519673555005653e-08,
      "loss": 0.7099,
      "step": 58425
    },
    {
      "epoch": 0.9096499490141438,
      "grad_norm": 2.600616693496704,
      "learning_rate": 9.511479654545155e-08,
      "loss": 0.6341,
      "step": 58430
    },
    {
      "epoch": 0.909727790016113,
      "grad_norm": 3.0600786209106445,
      "learning_rate": 9.503285754084659e-08,
      "loss": 0.7161,
      "step": 58435
    },
    {
      "epoch": 0.9098056310180824,
      "grad_norm": 4.341606140136719,
      "learning_rate": 9.495091853624161e-08,
      "loss": 0.6925,
      "step": 58440
    },
    {
      "epoch": 0.9098834720200518,
      "grad_norm": 3.219712734222412,
      "learning_rate": 9.486897953163663e-08,
      "loss": 0.7739,
      "step": 58445
    },
    {
      "epoch": 0.9099613130220212,
      "grad_norm": 3.1366140842437744,
      "learning_rate": 9.478704052703167e-08,
      "loss": 0.5365,
      "step": 58450
    },
    {
      "epoch": 0.9100391540239906,
      "grad_norm": 5.058888912200928,
      "learning_rate": 9.470510152242669e-08,
      "loss": 0.7611,
      "step": 58455
    },
    {
      "epoch": 0.91011699502596,
      "grad_norm": 4.127474308013916,
      "learning_rate": 9.462316251782173e-08,
      "loss": 0.8074,
      "step": 58460
    },
    {
      "epoch": 0.9101948360279294,
      "grad_norm": 4.662910461425781,
      "learning_rate": 9.454122351321676e-08,
      "loss": 0.7476,
      "step": 58465
    },
    {
      "epoch": 0.9102726770298988,
      "grad_norm": 4.339188098907471,
      "learning_rate": 9.44592845086118e-08,
      "loss": 0.7539,
      "step": 58470
    },
    {
      "epoch": 0.9103505180318681,
      "grad_norm": 3.8490047454833984,
      "learning_rate": 9.437734550400682e-08,
      "loss": 0.7577,
      "step": 58475
    },
    {
      "epoch": 0.9104283590338375,
      "grad_norm": 3.9565608501434326,
      "learning_rate": 9.429540649940185e-08,
      "loss": 0.7504,
      "step": 58480
    },
    {
      "epoch": 0.9105062000358068,
      "grad_norm": 3.4936161041259766,
      "learning_rate": 9.421346749479687e-08,
      "loss": 0.7859,
      "step": 58485
    },
    {
      "epoch": 0.9105840410377762,
      "grad_norm": 2.9777259826660156,
      "learning_rate": 9.41315284901919e-08,
      "loss": 0.6685,
      "step": 58490
    },
    {
      "epoch": 0.9106618820397456,
      "grad_norm": 3.8531382083892822,
      "learning_rate": 9.404958948558693e-08,
      "loss": 0.7556,
      "step": 58495
    },
    {
      "epoch": 0.910739723041715,
      "grad_norm": 5.789593696594238,
      "learning_rate": 9.396765048098195e-08,
      "loss": 0.8324,
      "step": 58500
    },
    {
      "epoch": 0.9108175640436844,
      "grad_norm": 2.9693398475646973,
      "learning_rate": 9.388571147637699e-08,
      "loss": 0.6099,
      "step": 58505
    },
    {
      "epoch": 0.9108954050456537,
      "grad_norm": 3.427449941635132,
      "learning_rate": 9.380377247177201e-08,
      "loss": 0.7987,
      "step": 58510
    },
    {
      "epoch": 0.9109732460476231,
      "grad_norm": 3.233821392059326,
      "learning_rate": 9.372183346716705e-08,
      "loss": 0.6871,
      "step": 58515
    },
    {
      "epoch": 0.9110510870495925,
      "grad_norm": 6.446512222290039,
      "learning_rate": 9.363989446256207e-08,
      "loss": 0.7213,
      "step": 58520
    },
    {
      "epoch": 0.9111289280515619,
      "grad_norm": 3.2003166675567627,
      "learning_rate": 9.35579554579571e-08,
      "loss": 0.7788,
      "step": 58525
    },
    {
      "epoch": 0.9112067690535313,
      "grad_norm": 3.066497325897217,
      "learning_rate": 9.347601645335213e-08,
      "loss": 0.7763,
      "step": 58530
    },
    {
      "epoch": 0.9112846100555007,
      "grad_norm": 3.5190863609313965,
      "learning_rate": 9.339407744874715e-08,
      "loss": 0.6687,
      "step": 58535
    },
    {
      "epoch": 0.91136245105747,
      "grad_norm": 3.069927215576172,
      "learning_rate": 9.331213844414218e-08,
      "loss": 0.7235,
      "step": 58540
    },
    {
      "epoch": 0.9114402920594394,
      "grad_norm": 2.61600923538208,
      "learning_rate": 9.32301994395372e-08,
      "loss": 0.6562,
      "step": 58545
    },
    {
      "epoch": 0.9115181330614087,
      "grad_norm": 4.133739948272705,
      "learning_rate": 9.314826043493224e-08,
      "loss": 0.7352,
      "step": 58550
    },
    {
      "epoch": 0.9115959740633781,
      "grad_norm": 2.840712547302246,
      "learning_rate": 9.306632143032726e-08,
      "loss": 0.6957,
      "step": 58555
    },
    {
      "epoch": 0.9116738150653475,
      "grad_norm": 4.327047824859619,
      "learning_rate": 9.29843824257223e-08,
      "loss": 0.6952,
      "step": 58560
    },
    {
      "epoch": 0.9117516560673169,
      "grad_norm": 3.1248562335968018,
      "learning_rate": 9.290244342111732e-08,
      "loss": 0.673,
      "step": 58565
    },
    {
      "epoch": 0.9118294970692863,
      "grad_norm": 5.928950786590576,
      "learning_rate": 9.282050441651234e-08,
      "loss": 0.8007,
      "step": 58570
    },
    {
      "epoch": 0.9119073380712557,
      "grad_norm": 3.9426519870758057,
      "learning_rate": 9.273856541190738e-08,
      "loss": 0.721,
      "step": 58575
    },
    {
      "epoch": 0.9119851790732251,
      "grad_norm": 3.4137015342712402,
      "learning_rate": 9.26566264073024e-08,
      "loss": 0.7417,
      "step": 58580
    },
    {
      "epoch": 0.9120630200751944,
      "grad_norm": 2.535740375518799,
      "learning_rate": 9.257468740269743e-08,
      "loss": 0.7297,
      "step": 58585
    },
    {
      "epoch": 0.9121408610771637,
      "grad_norm": 2.535541534423828,
      "learning_rate": 9.249274839809245e-08,
      "loss": 0.7335,
      "step": 58590
    },
    {
      "epoch": 0.9122187020791331,
      "grad_norm": 9.698796272277832,
      "learning_rate": 9.241080939348749e-08,
      "loss": 0.8419,
      "step": 58595
    },
    {
      "epoch": 0.9122965430811025,
      "grad_norm": 2.6092886924743652,
      "learning_rate": 9.232887038888251e-08,
      "loss": 0.7568,
      "step": 58600
    },
    {
      "epoch": 0.9123743840830719,
      "grad_norm": 3.4766972064971924,
      "learning_rate": 9.224693138427755e-08,
      "loss": 0.746,
      "step": 58605
    },
    {
      "epoch": 0.9124522250850413,
      "grad_norm": 3.9504926204681396,
      "learning_rate": 9.216499237967257e-08,
      "loss": 0.8315,
      "step": 58610
    },
    {
      "epoch": 0.9125300660870107,
      "grad_norm": 3.5546281337738037,
      "learning_rate": 9.208305337506759e-08,
      "loss": 0.7267,
      "step": 58615
    },
    {
      "epoch": 0.9126079070889801,
      "grad_norm": 3.9596612453460693,
      "learning_rate": 9.200111437046263e-08,
      "loss": 0.8531,
      "step": 58620
    },
    {
      "epoch": 0.9126857480909494,
      "grad_norm": 3.7654027938842773,
      "learning_rate": 9.191917536585765e-08,
      "loss": 0.6534,
      "step": 58625
    },
    {
      "epoch": 0.9127635890929188,
      "grad_norm": 9.95692253112793,
      "learning_rate": 9.183723636125268e-08,
      "loss": 0.7526,
      "step": 58630
    },
    {
      "epoch": 0.9128414300948882,
      "grad_norm": 4.322205066680908,
      "learning_rate": 9.17552973566477e-08,
      "loss": 0.736,
      "step": 58635
    },
    {
      "epoch": 0.9129192710968576,
      "grad_norm": 5.849926471710205,
      "learning_rate": 9.167335835204274e-08,
      "loss": 0.6538,
      "step": 58640
    },
    {
      "epoch": 0.912997112098827,
      "grad_norm": 7.7579779624938965,
      "learning_rate": 9.159141934743776e-08,
      "loss": 0.7751,
      "step": 58645
    },
    {
      "epoch": 0.9130749531007963,
      "grad_norm": 7.411488056182861,
      "learning_rate": 9.150948034283278e-08,
      "loss": 0.7926,
      "step": 58650
    },
    {
      "epoch": 0.9131527941027657,
      "grad_norm": 6.523559093475342,
      "learning_rate": 9.142754133822782e-08,
      "loss": 0.7158,
      "step": 58655
    },
    {
      "epoch": 0.9132306351047351,
      "grad_norm": 5.021554946899414,
      "learning_rate": 9.134560233362284e-08,
      "loss": 0.8245,
      "step": 58660
    },
    {
      "epoch": 0.9133084761067044,
      "grad_norm": 2.6120986938476562,
      "learning_rate": 9.126366332901788e-08,
      "loss": 0.6731,
      "step": 58665
    },
    {
      "epoch": 0.9133863171086738,
      "grad_norm": 3.790174961090088,
      "learning_rate": 9.11817243244129e-08,
      "loss": 0.8036,
      "step": 58670
    },
    {
      "epoch": 0.9134641581106432,
      "grad_norm": 3.8716955184936523,
      "learning_rate": 9.109978531980793e-08,
      "loss": 0.6511,
      "step": 58675
    },
    {
      "epoch": 0.9135419991126126,
      "grad_norm": 1.8677877187728882,
      "learning_rate": 9.101784631520295e-08,
      "loss": 0.7898,
      "step": 58680
    },
    {
      "epoch": 0.913619840114582,
      "grad_norm": 5.266423225402832,
      "learning_rate": 9.093590731059799e-08,
      "loss": 0.7659,
      "step": 58685
    },
    {
      "epoch": 0.9136976811165514,
      "grad_norm": 6.625179767608643,
      "learning_rate": 9.085396830599301e-08,
      "loss": 0.6933,
      "step": 58690
    },
    {
      "epoch": 0.9137755221185208,
      "grad_norm": 4.061988353729248,
      "learning_rate": 9.077202930138803e-08,
      "loss": 0.8298,
      "step": 58695
    },
    {
      "epoch": 0.91385336312049,
      "grad_norm": 3.151887893676758,
      "learning_rate": 9.069009029678307e-08,
      "loss": 0.7265,
      "step": 58700
    },
    {
      "epoch": 0.9139312041224594,
      "grad_norm": 3.149014949798584,
      "learning_rate": 9.060815129217809e-08,
      "loss": 0.8198,
      "step": 58705
    },
    {
      "epoch": 0.9140090451244288,
      "grad_norm": 3.819998025894165,
      "learning_rate": 9.052621228757313e-08,
      "loss": 0.6495,
      "step": 58710
    },
    {
      "epoch": 0.9140868861263982,
      "grad_norm": 2.8135783672332764,
      "learning_rate": 9.044427328296815e-08,
      "loss": 0.6832,
      "step": 58715
    },
    {
      "epoch": 0.9141647271283676,
      "grad_norm": 3.6375155448913574,
      "learning_rate": 9.036233427836318e-08,
      "loss": 0.7583,
      "step": 58720
    },
    {
      "epoch": 0.914242568130337,
      "grad_norm": 5.922914981842041,
      "learning_rate": 9.02803952737582e-08,
      "loss": 0.6599,
      "step": 58725
    },
    {
      "epoch": 0.9143204091323064,
      "grad_norm": 9.75033187866211,
      "learning_rate": 9.019845626915324e-08,
      "loss": 0.7327,
      "step": 58730
    },
    {
      "epoch": 0.9143982501342758,
      "grad_norm": 3.3765721321105957,
      "learning_rate": 9.011651726454826e-08,
      "loss": 0.6828,
      "step": 58735
    },
    {
      "epoch": 0.9144760911362451,
      "grad_norm": 3.1056931018829346,
      "learning_rate": 9.003457825994328e-08,
      "loss": 0.7695,
      "step": 58740
    },
    {
      "epoch": 0.9145539321382145,
      "grad_norm": 4.5549821853637695,
      "learning_rate": 8.995263925533832e-08,
      "loss": 0.7501,
      "step": 58745
    },
    {
      "epoch": 0.9146317731401838,
      "grad_norm": 3.000243902206421,
      "learning_rate": 8.987070025073334e-08,
      "loss": 0.6336,
      "step": 58750
    },
    {
      "epoch": 0.9147096141421532,
      "grad_norm": 4.8810553550720215,
      "learning_rate": 8.978876124612838e-08,
      "loss": 0.8825,
      "step": 58755
    },
    {
      "epoch": 0.9147874551441226,
      "grad_norm": 2.601544141769409,
      "learning_rate": 8.97068222415234e-08,
      "loss": 0.7057,
      "step": 58760
    },
    {
      "epoch": 0.914865296146092,
      "grad_norm": 6.225688457489014,
      "learning_rate": 8.962488323691843e-08,
      "loss": 0.7956,
      "step": 58765
    },
    {
      "epoch": 0.9149431371480614,
      "grad_norm": 3.1837265491485596,
      "learning_rate": 8.954294423231346e-08,
      "loss": 0.834,
      "step": 58770
    },
    {
      "epoch": 0.9150209781500307,
      "grad_norm": 4.121196746826172,
      "learning_rate": 8.946100522770848e-08,
      "loss": 0.7259,
      "step": 58775
    },
    {
      "epoch": 0.9150988191520001,
      "grad_norm": 3.1132490634918213,
      "learning_rate": 8.937906622310351e-08,
      "loss": 0.6461,
      "step": 58780
    },
    {
      "epoch": 0.9151766601539695,
      "grad_norm": 9.460606575012207,
      "learning_rate": 8.929712721849855e-08,
      "loss": 0.7597,
      "step": 58785
    },
    {
      "epoch": 0.9152545011559389,
      "grad_norm": 9.755582809448242,
      "learning_rate": 8.921518821389358e-08,
      "loss": 0.796,
      "step": 58790
    },
    {
      "epoch": 0.9153323421579083,
      "grad_norm": 3.0565783977508545,
      "learning_rate": 8.91332492092886e-08,
      "loss": 0.7292,
      "step": 58795
    },
    {
      "epoch": 0.9154101831598777,
      "grad_norm": 8.975831985473633,
      "learning_rate": 8.905131020468364e-08,
      "loss": 0.7328,
      "step": 58800
    },
    {
      "epoch": 0.915488024161847,
      "grad_norm": 3.6419858932495117,
      "learning_rate": 8.896937120007866e-08,
      "loss": 0.6941,
      "step": 58805
    },
    {
      "epoch": 0.9155658651638164,
      "grad_norm": 2.7038753032684326,
      "learning_rate": 8.88874321954737e-08,
      "loss": 0.7807,
      "step": 58810
    },
    {
      "epoch": 0.9156437061657857,
      "grad_norm": 3.928826332092285,
      "learning_rate": 8.880549319086872e-08,
      "loss": 0.7963,
      "step": 58815
    },
    {
      "epoch": 0.9157215471677551,
      "grad_norm": 5.243278503417969,
      "learning_rate": 8.872355418626374e-08,
      "loss": 0.7867,
      "step": 58820
    },
    {
      "epoch": 0.9157993881697245,
      "grad_norm": 3.081068992614746,
      "learning_rate": 8.864161518165878e-08,
      "loss": 0.7269,
      "step": 58825
    },
    {
      "epoch": 0.9158772291716939,
      "grad_norm": 5.017881393432617,
      "learning_rate": 8.85596761770538e-08,
      "loss": 0.8348,
      "step": 58830
    },
    {
      "epoch": 0.9159550701736633,
      "grad_norm": 3.1931824684143066,
      "learning_rate": 8.847773717244883e-08,
      "loss": 0.702,
      "step": 58835
    },
    {
      "epoch": 0.9160329111756327,
      "grad_norm": 2.776602029800415,
      "learning_rate": 8.839579816784385e-08,
      "loss": 0.6676,
      "step": 58840
    },
    {
      "epoch": 0.9161107521776021,
      "grad_norm": 4.102383136749268,
      "learning_rate": 8.831385916323889e-08,
      "loss": 0.7409,
      "step": 58845
    },
    {
      "epoch": 0.9161885931795714,
      "grad_norm": 4.9644927978515625,
      "learning_rate": 8.823192015863391e-08,
      "loss": 0.7641,
      "step": 58850
    },
    {
      "epoch": 0.9162664341815407,
      "grad_norm": 5.688453197479248,
      "learning_rate": 8.814998115402895e-08,
      "loss": 0.7476,
      "step": 58855
    },
    {
      "epoch": 0.9163442751835101,
      "grad_norm": 3.917360782623291,
      "learning_rate": 8.806804214942397e-08,
      "loss": 0.809,
      "step": 58860
    },
    {
      "epoch": 0.9164221161854795,
      "grad_norm": 3.789743423461914,
      "learning_rate": 8.798610314481899e-08,
      "loss": 0.6759,
      "step": 58865
    },
    {
      "epoch": 0.9164999571874489,
      "grad_norm": 5.565342903137207,
      "learning_rate": 8.790416414021403e-08,
      "loss": 0.717,
      "step": 58870
    },
    {
      "epoch": 0.9165777981894183,
      "grad_norm": 2.58754301071167,
      "learning_rate": 8.782222513560905e-08,
      "loss": 0.6451,
      "step": 58875
    },
    {
      "epoch": 0.9166556391913877,
      "grad_norm": 3.9781370162963867,
      "learning_rate": 8.774028613100408e-08,
      "loss": 0.6321,
      "step": 58880
    },
    {
      "epoch": 0.9167334801933571,
      "grad_norm": 4.810916423797607,
      "learning_rate": 8.76583471263991e-08,
      "loss": 0.7008,
      "step": 58885
    },
    {
      "epoch": 0.9168113211953264,
      "grad_norm": 3.817229747772217,
      "learning_rate": 8.757640812179414e-08,
      "loss": 0.8536,
      "step": 58890
    },
    {
      "epoch": 0.9168891621972958,
      "grad_norm": 2.8795440196990967,
      "learning_rate": 8.749446911718916e-08,
      "loss": 0.6707,
      "step": 58895
    },
    {
      "epoch": 0.9169670031992652,
      "grad_norm": 5.976953983306885,
      "learning_rate": 8.741253011258418e-08,
      "loss": 0.6317,
      "step": 58900
    },
    {
      "epoch": 0.9170448442012346,
      "grad_norm": 4.553403854370117,
      "learning_rate": 8.733059110797922e-08,
      "loss": 0.6936,
      "step": 58905
    },
    {
      "epoch": 0.917122685203204,
      "grad_norm": 3.295867919921875,
      "learning_rate": 8.724865210337424e-08,
      "loss": 0.6936,
      "step": 58910
    },
    {
      "epoch": 0.9172005262051733,
      "grad_norm": 4.384651184082031,
      "learning_rate": 8.716671309876928e-08,
      "loss": 0.7535,
      "step": 58915
    },
    {
      "epoch": 0.9172783672071427,
      "grad_norm": 3.719183921813965,
      "learning_rate": 8.70847740941643e-08,
      "loss": 0.7,
      "step": 58920
    },
    {
      "epoch": 0.9173562082091121,
      "grad_norm": 4.573428153991699,
      "learning_rate": 8.700283508955933e-08,
      "loss": 0.7889,
      "step": 58925
    },
    {
      "epoch": 0.9174340492110814,
      "grad_norm": 6.366910457611084,
      "learning_rate": 8.692089608495436e-08,
      "loss": 0.8248,
      "step": 58930
    },
    {
      "epoch": 0.9175118902130508,
      "grad_norm": 5.7622971534729,
      "learning_rate": 8.683895708034939e-08,
      "loss": 0.7476,
      "step": 58935
    },
    {
      "epoch": 0.9175897312150202,
      "grad_norm": 6.677018165588379,
      "learning_rate": 8.675701807574441e-08,
      "loss": 0.694,
      "step": 58940
    },
    {
      "epoch": 0.9176675722169896,
      "grad_norm": 3.68141508102417,
      "learning_rate": 8.667507907113943e-08,
      "loss": 0.7397,
      "step": 58945
    },
    {
      "epoch": 0.917745413218959,
      "grad_norm": 3.7141647338867188,
      "learning_rate": 8.659314006653447e-08,
      "loss": 0.6657,
      "step": 58950
    },
    {
      "epoch": 0.9178232542209284,
      "grad_norm": 4.4624247550964355,
      "learning_rate": 8.651120106192949e-08,
      "loss": 0.6634,
      "step": 58955
    },
    {
      "epoch": 0.9179010952228978,
      "grad_norm": 6.385409832000732,
      "learning_rate": 8.642926205732453e-08,
      "loss": 0.7695,
      "step": 58960
    },
    {
      "epoch": 0.917978936224867,
      "grad_norm": 3.5434324741363525,
      "learning_rate": 8.634732305271955e-08,
      "loss": 0.7006,
      "step": 58965
    },
    {
      "epoch": 0.9180567772268364,
      "grad_norm": 3.23720383644104,
      "learning_rate": 8.626538404811458e-08,
      "loss": 0.7501,
      "step": 58970
    },
    {
      "epoch": 0.9181346182288058,
      "grad_norm": 3.440925359725952,
      "learning_rate": 8.61834450435096e-08,
      "loss": 0.6761,
      "step": 58975
    },
    {
      "epoch": 0.9182124592307752,
      "grad_norm": 3.106787919998169,
      "learning_rate": 8.610150603890463e-08,
      "loss": 0.7642,
      "step": 58980
    },
    {
      "epoch": 0.9182903002327446,
      "grad_norm": 4.5047993659973145,
      "learning_rate": 8.601956703429966e-08,
      "loss": 0.715,
      "step": 58985
    },
    {
      "epoch": 0.918368141234714,
      "grad_norm": 3.270312547683716,
      "learning_rate": 8.593762802969468e-08,
      "loss": 0.6719,
      "step": 58990
    },
    {
      "epoch": 0.9184459822366834,
      "grad_norm": 5.309615135192871,
      "learning_rate": 8.585568902508972e-08,
      "loss": 0.72,
      "step": 58995
    },
    {
      "epoch": 0.9185238232386528,
      "grad_norm": 3.6470792293548584,
      "learning_rate": 8.577375002048474e-08,
      "loss": 0.6754,
      "step": 59000
    },
    {
      "epoch": 0.9186016642406221,
      "grad_norm": 8.898789405822754,
      "learning_rate": 8.569181101587978e-08,
      "loss": 0.841,
      "step": 59005
    },
    {
      "epoch": 0.9186795052425915,
      "grad_norm": 7.527399063110352,
      "learning_rate": 8.56098720112748e-08,
      "loss": 0.7513,
      "step": 59010
    },
    {
      "epoch": 0.9187573462445608,
      "grad_norm": 4.64124870300293,
      "learning_rate": 8.552793300666983e-08,
      "loss": 0.8007,
      "step": 59015
    },
    {
      "epoch": 0.9188351872465302,
      "grad_norm": 4.897219657897949,
      "learning_rate": 8.544599400206486e-08,
      "loss": 0.7576,
      "step": 59020
    },
    {
      "epoch": 0.9189130282484996,
      "grad_norm": 2.9512879848480225,
      "learning_rate": 8.536405499745988e-08,
      "loss": 0.7282,
      "step": 59025
    },
    {
      "epoch": 0.918990869250469,
      "grad_norm": 2.8408710956573486,
      "learning_rate": 8.528211599285491e-08,
      "loss": 0.6845,
      "step": 59030
    },
    {
      "epoch": 0.9190687102524384,
      "grad_norm": 4.560900688171387,
      "learning_rate": 8.520017698824993e-08,
      "loss": 0.7604,
      "step": 59035
    },
    {
      "epoch": 0.9191465512544077,
      "grad_norm": 8.069441795349121,
      "learning_rate": 8.511823798364497e-08,
      "loss": 0.7571,
      "step": 59040
    },
    {
      "epoch": 0.9192243922563771,
      "grad_norm": 8.190315246582031,
      "learning_rate": 8.503629897903999e-08,
      "loss": 0.6589,
      "step": 59045
    },
    {
      "epoch": 0.9193022332583465,
      "grad_norm": 4.292120456695557,
      "learning_rate": 8.495435997443503e-08,
      "loss": 0.6676,
      "step": 59050
    },
    {
      "epoch": 0.9193800742603159,
      "grad_norm": 2.9687325954437256,
      "learning_rate": 8.487242096983005e-08,
      "loss": 0.644,
      "step": 59055
    },
    {
      "epoch": 0.9194579152622853,
      "grad_norm": 4.393004417419434,
      "learning_rate": 8.479048196522508e-08,
      "loss": 0.7152,
      "step": 59060
    },
    {
      "epoch": 0.9195357562642547,
      "grad_norm": 4.12727165222168,
      "learning_rate": 8.47085429606201e-08,
      "loss": 0.7,
      "step": 59065
    },
    {
      "epoch": 0.919613597266224,
      "grad_norm": 5.002000331878662,
      "learning_rate": 8.462660395601513e-08,
      "loss": 0.7029,
      "step": 59070
    },
    {
      "epoch": 0.9196914382681934,
      "grad_norm": 3.6730992794036865,
      "learning_rate": 8.454466495141016e-08,
      "loss": 0.651,
      "step": 59075
    },
    {
      "epoch": 0.9197692792701627,
      "grad_norm": 4.622668743133545,
      "learning_rate": 8.446272594680518e-08,
      "loss": 0.6339,
      "step": 59080
    },
    {
      "epoch": 0.9198471202721321,
      "grad_norm": 2.753437042236328,
      "learning_rate": 8.438078694220022e-08,
      "loss": 0.7181,
      "step": 59085
    },
    {
      "epoch": 0.9199249612741015,
      "grad_norm": 7.516641139984131,
      "learning_rate": 8.429884793759524e-08,
      "loss": 0.8189,
      "step": 59090
    },
    {
      "epoch": 0.9200028022760709,
      "grad_norm": 2.765319585800171,
      "learning_rate": 8.421690893299028e-08,
      "loss": 0.6481,
      "step": 59095
    },
    {
      "epoch": 0.9200806432780403,
      "grad_norm": 4.155617713928223,
      "learning_rate": 8.41349699283853e-08,
      "loss": 0.7134,
      "step": 59100
    },
    {
      "epoch": 0.9201584842800097,
      "grad_norm": 3.989515542984009,
      "learning_rate": 8.405303092378035e-08,
      "loss": 0.7986,
      "step": 59105
    },
    {
      "epoch": 0.9202363252819791,
      "grad_norm": 3.6144373416900635,
      "learning_rate": 8.397109191917537e-08,
      "loss": 0.8002,
      "step": 59110
    },
    {
      "epoch": 0.9203141662839484,
      "grad_norm": 3.4274280071258545,
      "learning_rate": 8.388915291457039e-08,
      "loss": 0.7061,
      "step": 59115
    },
    {
      "epoch": 0.9203920072859177,
      "grad_norm": 3.383967161178589,
      "learning_rate": 8.380721390996543e-08,
      "loss": 0.8472,
      "step": 59120
    },
    {
      "epoch": 0.9204698482878871,
      "grad_norm": 3.304867744445801,
      "learning_rate": 8.372527490536045e-08,
      "loss": 0.8186,
      "step": 59125
    },
    {
      "epoch": 0.9205476892898565,
      "grad_norm": 4.3925652503967285,
      "learning_rate": 8.364333590075548e-08,
      "loss": 0.7098,
      "step": 59130
    },
    {
      "epoch": 0.9206255302918259,
      "grad_norm": 3.505040407180786,
      "learning_rate": 8.35613968961505e-08,
      "loss": 0.7283,
      "step": 59135
    },
    {
      "epoch": 0.9207033712937953,
      "grad_norm": 3.161982774734497,
      "learning_rate": 8.347945789154554e-08,
      "loss": 0.6668,
      "step": 59140
    },
    {
      "epoch": 0.9207812122957647,
      "grad_norm": 4.624584197998047,
      "learning_rate": 8.339751888694056e-08,
      "loss": 0.7555,
      "step": 59145
    },
    {
      "epoch": 0.9208590532977341,
      "grad_norm": 3.0786759853363037,
      "learning_rate": 8.331557988233558e-08,
      "loss": 0.7377,
      "step": 59150
    },
    {
      "epoch": 0.9209368942997034,
      "grad_norm": 5.609769821166992,
      "learning_rate": 8.323364087773062e-08,
      "loss": 0.7075,
      "step": 59155
    },
    {
      "epoch": 0.9210147353016728,
      "grad_norm": 3.481327772140503,
      "learning_rate": 8.315170187312564e-08,
      "loss": 0.7189,
      "step": 59160
    },
    {
      "epoch": 0.9210925763036422,
      "grad_norm": 2.7407467365264893,
      "learning_rate": 8.306976286852068e-08,
      "loss": 0.7408,
      "step": 59165
    },
    {
      "epoch": 0.9211704173056116,
      "grad_norm": 5.946617603302002,
      "learning_rate": 8.29878238639157e-08,
      "loss": 0.6916,
      "step": 59170
    },
    {
      "epoch": 0.921248258307581,
      "grad_norm": 2.7746224403381348,
      "learning_rate": 8.290588485931073e-08,
      "loss": 0.7507,
      "step": 59175
    },
    {
      "epoch": 0.9213260993095503,
      "grad_norm": 7.005921363830566,
      "learning_rate": 8.282394585470576e-08,
      "loss": 0.8117,
      "step": 59180
    },
    {
      "epoch": 0.9214039403115197,
      "grad_norm": 4.125368595123291,
      "learning_rate": 8.274200685010079e-08,
      "loss": 0.71,
      "step": 59185
    },
    {
      "epoch": 0.921481781313489,
      "grad_norm": 4.301252365112305,
      "learning_rate": 8.266006784549581e-08,
      "loss": 0.7494,
      "step": 59190
    },
    {
      "epoch": 0.9215596223154584,
      "grad_norm": 3.3552944660186768,
      "learning_rate": 8.257812884089083e-08,
      "loss": 0.6538,
      "step": 59195
    },
    {
      "epoch": 0.9216374633174278,
      "grad_norm": 4.501960754394531,
      "learning_rate": 8.249618983628587e-08,
      "loss": 0.6788,
      "step": 59200
    },
    {
      "epoch": 0.9217153043193972,
      "grad_norm": 5.4391350746154785,
      "learning_rate": 8.241425083168089e-08,
      "loss": 0.622,
      "step": 59205
    },
    {
      "epoch": 0.9217931453213666,
      "grad_norm": 2.6977121829986572,
      "learning_rate": 8.233231182707593e-08,
      "loss": 0.6712,
      "step": 59210
    },
    {
      "epoch": 0.921870986323336,
      "grad_norm": 3.458757162094116,
      "learning_rate": 8.225037282247095e-08,
      "loss": 0.775,
      "step": 59215
    },
    {
      "epoch": 0.9219488273253054,
      "grad_norm": 3.127300977706909,
      "learning_rate": 8.216843381786598e-08,
      "loss": 0.7869,
      "step": 59220
    },
    {
      "epoch": 0.9220266683272748,
      "grad_norm": 2.838869094848633,
      "learning_rate": 8.2086494813261e-08,
      "loss": 0.769,
      "step": 59225
    },
    {
      "epoch": 0.922104509329244,
      "grad_norm": 4.249427318572998,
      "learning_rate": 8.200455580865603e-08,
      "loss": 0.7634,
      "step": 59230
    },
    {
      "epoch": 0.9221823503312134,
      "grad_norm": 4.134322166442871,
      "learning_rate": 8.192261680405106e-08,
      "loss": 0.7961,
      "step": 59235
    },
    {
      "epoch": 0.9222601913331828,
      "grad_norm": 4.614013671875,
      "learning_rate": 8.184067779944608e-08,
      "loss": 0.6403,
      "step": 59240
    },
    {
      "epoch": 0.9223380323351522,
      "grad_norm": 3.408700704574585,
      "learning_rate": 8.175873879484112e-08,
      "loss": 0.6608,
      "step": 59245
    },
    {
      "epoch": 0.9224158733371216,
      "grad_norm": 3.459613084793091,
      "learning_rate": 8.167679979023614e-08,
      "loss": 0.7365,
      "step": 59250
    },
    {
      "epoch": 0.922493714339091,
      "grad_norm": 7.845703601837158,
      "learning_rate": 8.159486078563118e-08,
      "loss": 0.7776,
      "step": 59255
    },
    {
      "epoch": 0.9225715553410604,
      "grad_norm": 4.305447101593018,
      "learning_rate": 8.15129217810262e-08,
      "loss": 0.7411,
      "step": 59260
    },
    {
      "epoch": 0.9226493963430298,
      "grad_norm": 5.17306661605835,
      "learning_rate": 8.143098277642123e-08,
      "loss": 0.6845,
      "step": 59265
    },
    {
      "epoch": 0.9227272373449991,
      "grad_norm": 4.288595676422119,
      "learning_rate": 8.134904377181626e-08,
      "loss": 0.74,
      "step": 59270
    },
    {
      "epoch": 0.9228050783469685,
      "grad_norm": 5.907822608947754,
      "learning_rate": 8.126710476721128e-08,
      "loss": 0.7407,
      "step": 59275
    },
    {
      "epoch": 0.9228829193489378,
      "grad_norm": 12.026689529418945,
      "learning_rate": 8.118516576260631e-08,
      "loss": 0.7096,
      "step": 59280
    },
    {
      "epoch": 0.9229607603509072,
      "grad_norm": 3.4096920490264893,
      "learning_rate": 8.110322675800133e-08,
      "loss": 0.6252,
      "step": 59285
    },
    {
      "epoch": 0.9230386013528766,
      "grad_norm": 8.090753555297852,
      "learning_rate": 8.102128775339637e-08,
      "loss": 0.8644,
      "step": 59290
    },
    {
      "epoch": 0.923116442354846,
      "grad_norm": 4.933746337890625,
      "learning_rate": 8.093934874879139e-08,
      "loss": 0.7243,
      "step": 59295
    },
    {
      "epoch": 0.9231942833568154,
      "grad_norm": 3.5507819652557373,
      "learning_rate": 8.085740974418643e-08,
      "loss": 0.8037,
      "step": 59300
    },
    {
      "epoch": 0.9232721243587847,
      "grad_norm": 6.970895290374756,
      "learning_rate": 8.077547073958145e-08,
      "loss": 0.7386,
      "step": 59305
    },
    {
      "epoch": 0.9233499653607541,
      "grad_norm": 3.910081624984741,
      "learning_rate": 8.069353173497648e-08,
      "loss": 0.7961,
      "step": 59310
    },
    {
      "epoch": 0.9234278063627235,
      "grad_norm": 2.9712488651275635,
      "learning_rate": 8.06115927303715e-08,
      "loss": 0.7064,
      "step": 59315
    },
    {
      "epoch": 0.9235056473646929,
      "grad_norm": 5.052121639251709,
      "learning_rate": 8.052965372576653e-08,
      "loss": 0.8339,
      "step": 59320
    },
    {
      "epoch": 0.9235834883666623,
      "grad_norm": 3.811683177947998,
      "learning_rate": 8.044771472116156e-08,
      "loss": 0.7465,
      "step": 59325
    },
    {
      "epoch": 0.9236613293686317,
      "grad_norm": 3.1975247859954834,
      "learning_rate": 8.036577571655659e-08,
      "loss": 0.7259,
      "step": 59330
    },
    {
      "epoch": 0.923739170370601,
      "grad_norm": 4.876020431518555,
      "learning_rate": 8.028383671195162e-08,
      "loss": 0.8071,
      "step": 59335
    },
    {
      "epoch": 0.9238170113725704,
      "grad_norm": 6.120541095733643,
      "learning_rate": 8.020189770734664e-08,
      "loss": 0.7299,
      "step": 59340
    },
    {
      "epoch": 0.9238948523745397,
      "grad_norm": 5.237553596496582,
      "learning_rate": 8.011995870274168e-08,
      "loss": 0.7509,
      "step": 59345
    },
    {
      "epoch": 0.9239726933765091,
      "grad_norm": 5.503535270690918,
      "learning_rate": 8.00380196981367e-08,
      "loss": 0.7581,
      "step": 59350
    },
    {
      "epoch": 0.9240505343784785,
      "grad_norm": 4.783160209655762,
      "learning_rate": 7.995608069353172e-08,
      "loss": 0.6958,
      "step": 59355
    },
    {
      "epoch": 0.9241283753804479,
      "grad_norm": 4.756977558135986,
      "learning_rate": 7.987414168892676e-08,
      "loss": 0.7952,
      "step": 59360
    },
    {
      "epoch": 0.9242062163824173,
      "grad_norm": 4.037032604217529,
      "learning_rate": 7.979220268432178e-08,
      "loss": 0.7644,
      "step": 59365
    },
    {
      "epoch": 0.9242840573843867,
      "grad_norm": 4.761673927307129,
      "learning_rate": 7.971026367971681e-08,
      "loss": 0.7823,
      "step": 59370
    },
    {
      "epoch": 0.9243618983863561,
      "grad_norm": 2.684903860092163,
      "learning_rate": 7.962832467511184e-08,
      "loss": 0.7357,
      "step": 59375
    },
    {
      "epoch": 0.9244397393883254,
      "grad_norm": 3.00250244140625,
      "learning_rate": 7.954638567050687e-08,
      "loss": 0.8283,
      "step": 59380
    },
    {
      "epoch": 0.9245175803902947,
      "grad_norm": 3.205409049987793,
      "learning_rate": 7.946444666590189e-08,
      "loss": 0.7168,
      "step": 59385
    },
    {
      "epoch": 0.9245954213922641,
      "grad_norm": 2.821117877960205,
      "learning_rate": 7.938250766129693e-08,
      "loss": 0.8588,
      "step": 59390
    },
    {
      "epoch": 0.9246732623942335,
      "grad_norm": 2.492983102798462,
      "learning_rate": 7.930056865669195e-08,
      "loss": 0.6545,
      "step": 59395
    },
    {
      "epoch": 0.9247511033962029,
      "grad_norm": 5.436090469360352,
      "learning_rate": 7.921862965208697e-08,
      "loss": 0.7322,
      "step": 59400
    },
    {
      "epoch": 0.9248289443981723,
      "grad_norm": 5.916810035705566,
      "learning_rate": 7.913669064748201e-08,
      "loss": 0.6606,
      "step": 59405
    },
    {
      "epoch": 0.9249067854001417,
      "grad_norm": 3.965684652328491,
      "learning_rate": 7.905475164287703e-08,
      "loss": 0.7001,
      "step": 59410
    },
    {
      "epoch": 0.9249846264021111,
      "grad_norm": 4.775380611419678,
      "learning_rate": 7.897281263827206e-08,
      "loss": 0.8301,
      "step": 59415
    },
    {
      "epoch": 0.9250624674040804,
      "grad_norm": 3.342169761657715,
      "learning_rate": 7.88908736336671e-08,
      "loss": 0.6819,
      "step": 59420
    },
    {
      "epoch": 0.9251403084060498,
      "grad_norm": 3.3289217948913574,
      "learning_rate": 7.880893462906213e-08,
      "loss": 0.7558,
      "step": 59425
    },
    {
      "epoch": 0.9252181494080192,
      "grad_norm": 2.702233076095581,
      "learning_rate": 7.872699562445716e-08,
      "loss": 0.8024,
      "step": 59430
    },
    {
      "epoch": 0.9252959904099886,
      "grad_norm": 4.198214530944824,
      "learning_rate": 7.864505661985219e-08,
      "loss": 0.7259,
      "step": 59435
    },
    {
      "epoch": 0.925373831411958,
      "grad_norm": 5.863811492919922,
      "learning_rate": 7.856311761524721e-08,
      "loss": 0.5519,
      "step": 59440
    },
    {
      "epoch": 0.9254516724139273,
      "grad_norm": 3.9226627349853516,
      "learning_rate": 7.848117861064223e-08,
      "loss": 0.8101,
      "step": 59445
    },
    {
      "epoch": 0.9255295134158967,
      "grad_norm": 4.54281759262085,
      "learning_rate": 7.839923960603727e-08,
      "loss": 0.6676,
      "step": 59450
    },
    {
      "epoch": 0.925607354417866,
      "grad_norm": 4.994977951049805,
      "learning_rate": 7.831730060143229e-08,
      "loss": 0.7794,
      "step": 59455
    },
    {
      "epoch": 0.9256851954198354,
      "grad_norm": 3.046161651611328,
      "learning_rate": 7.823536159682733e-08,
      "loss": 0.6813,
      "step": 59460
    },
    {
      "epoch": 0.9257630364218048,
      "grad_norm": 3.5242838859558105,
      "learning_rate": 7.815342259222235e-08,
      "loss": 0.7513,
      "step": 59465
    },
    {
      "epoch": 0.9258408774237742,
      "grad_norm": 4.399619102478027,
      "learning_rate": 7.807148358761738e-08,
      "loss": 0.8195,
      "step": 59470
    },
    {
      "epoch": 0.9259187184257436,
      "grad_norm": 5.425989627838135,
      "learning_rate": 7.79895445830124e-08,
      "loss": 0.8937,
      "step": 59475
    },
    {
      "epoch": 0.925996559427713,
      "grad_norm": 3.40451717376709,
      "learning_rate": 7.790760557840743e-08,
      "loss": 0.6908,
      "step": 59480
    },
    {
      "epoch": 0.9260744004296824,
      "grad_norm": 4.780428409576416,
      "learning_rate": 7.782566657380246e-08,
      "loss": 0.7443,
      "step": 59485
    },
    {
      "epoch": 0.9261522414316518,
      "grad_norm": 2.736248731613159,
      "learning_rate": 7.774372756919749e-08,
      "loss": 0.6633,
      "step": 59490
    },
    {
      "epoch": 0.926230082433621,
      "grad_norm": 3.1158831119537354,
      "learning_rate": 7.766178856459252e-08,
      "loss": 0.8627,
      "step": 59495
    },
    {
      "epoch": 0.9263079234355904,
      "grad_norm": 7.0678391456604,
      "learning_rate": 7.757984955998754e-08,
      "loss": 0.8141,
      "step": 59500
    },
    {
      "epoch": 0.9263857644375598,
      "grad_norm": 4.170321464538574,
      "learning_rate": 7.749791055538258e-08,
      "loss": 0.8501,
      "step": 59505
    },
    {
      "epoch": 0.9264636054395292,
      "grad_norm": 4.052420616149902,
      "learning_rate": 7.74159715507776e-08,
      "loss": 0.7968,
      "step": 59510
    },
    {
      "epoch": 0.9265414464414986,
      "grad_norm": 11.312172889709473,
      "learning_rate": 7.733403254617263e-08,
      "loss": 0.8118,
      "step": 59515
    },
    {
      "epoch": 0.926619287443468,
      "grad_norm": 4.245031356811523,
      "learning_rate": 7.725209354156766e-08,
      "loss": 0.7316,
      "step": 59520
    },
    {
      "epoch": 0.9266971284454374,
      "grad_norm": 3.270986318588257,
      "learning_rate": 7.717015453696268e-08,
      "loss": 0.6416,
      "step": 59525
    },
    {
      "epoch": 0.9267749694474068,
      "grad_norm": 6.0705952644348145,
      "learning_rate": 7.708821553235771e-08,
      "loss": 0.6846,
      "step": 59530
    },
    {
      "epoch": 0.9268528104493761,
      "grad_norm": 3.813494920730591,
      "learning_rate": 7.700627652775274e-08,
      "loss": 0.6584,
      "step": 59535
    },
    {
      "epoch": 0.9269306514513455,
      "grad_norm": 3.0086796283721924,
      "learning_rate": 7.692433752314777e-08,
      "loss": 0.8243,
      "step": 59540
    },
    {
      "epoch": 0.9270084924533148,
      "grad_norm": 3.8762195110321045,
      "learning_rate": 7.684239851854279e-08,
      "loss": 0.7685,
      "step": 59545
    },
    {
      "epoch": 0.9270863334552842,
      "grad_norm": 5.8003435134887695,
      "learning_rate": 7.676045951393783e-08,
      "loss": 0.7316,
      "step": 59550
    },
    {
      "epoch": 0.9271641744572536,
      "grad_norm": 3.6708149909973145,
      "learning_rate": 7.667852050933285e-08,
      "loss": 0.7457,
      "step": 59555
    },
    {
      "epoch": 0.927242015459223,
      "grad_norm": 4.677990436553955,
      "learning_rate": 7.659658150472787e-08,
      "loss": 0.668,
      "step": 59560
    },
    {
      "epoch": 0.9273198564611924,
      "grad_norm": 3.6643497943878174,
      "learning_rate": 7.65146425001229e-08,
      "loss": 0.7388,
      "step": 59565
    },
    {
      "epoch": 0.9273976974631617,
      "grad_norm": 5.502130508422852,
      "learning_rate": 7.643270349551793e-08,
      "loss": 0.7014,
      "step": 59570
    },
    {
      "epoch": 0.9274755384651311,
      "grad_norm": 3.6354691982269287,
      "learning_rate": 7.635076449091296e-08,
      "loss": 0.7446,
      "step": 59575
    },
    {
      "epoch": 0.9275533794671005,
      "grad_norm": 3.874295234680176,
      "learning_rate": 7.626882548630799e-08,
      "loss": 0.7752,
      "step": 59580
    },
    {
      "epoch": 0.9276312204690699,
      "grad_norm": 4.635042667388916,
      "learning_rate": 7.618688648170302e-08,
      "loss": 0.7595,
      "step": 59585
    },
    {
      "epoch": 0.9277090614710393,
      "grad_norm": 3.445505380630493,
      "learning_rate": 7.610494747709804e-08,
      "loss": 0.7403,
      "step": 59590
    },
    {
      "epoch": 0.9277869024730087,
      "grad_norm": 3.3790462017059326,
      "learning_rate": 7.602300847249308e-08,
      "loss": 0.7814,
      "step": 59595
    },
    {
      "epoch": 0.927864743474978,
      "grad_norm": 6.7586798667907715,
      "learning_rate": 7.59410694678881e-08,
      "loss": 0.8517,
      "step": 59600
    },
    {
      "epoch": 0.9279425844769474,
      "grad_norm": 3.1671319007873535,
      "learning_rate": 7.585913046328312e-08,
      "loss": 0.7967,
      "step": 59605
    },
    {
      "epoch": 0.9280204254789167,
      "grad_norm": 3.9603376388549805,
      "learning_rate": 7.577719145867816e-08,
      "loss": 0.783,
      "step": 59610
    },
    {
      "epoch": 0.9280982664808861,
      "grad_norm": 3.4688801765441895,
      "learning_rate": 7.569525245407318e-08,
      "loss": 0.7154,
      "step": 59615
    },
    {
      "epoch": 0.9281761074828555,
      "grad_norm": 3.3755111694335938,
      "learning_rate": 7.561331344946821e-08,
      "loss": 0.6893,
      "step": 59620
    },
    {
      "epoch": 0.9282539484848249,
      "grad_norm": 3.774723768234253,
      "learning_rate": 7.553137444486324e-08,
      "loss": 0.6818,
      "step": 59625
    },
    {
      "epoch": 0.9283317894867943,
      "grad_norm": 12.895103454589844,
      "learning_rate": 7.544943544025827e-08,
      "loss": 0.832,
      "step": 59630
    },
    {
      "epoch": 0.9284096304887637,
      "grad_norm": 4.030149459838867,
      "learning_rate": 7.536749643565329e-08,
      "loss": 0.7679,
      "step": 59635
    },
    {
      "epoch": 0.9284874714907331,
      "grad_norm": 3.354248523712158,
      "learning_rate": 7.528555743104833e-08,
      "loss": 0.7329,
      "step": 59640
    },
    {
      "epoch": 0.9285653124927024,
      "grad_norm": 2.6033735275268555,
      "learning_rate": 7.520361842644335e-08,
      "loss": 0.7442,
      "step": 59645
    },
    {
      "epoch": 0.9286431534946717,
      "grad_norm": 5.564953327178955,
      "learning_rate": 7.512167942183837e-08,
      "loss": 0.8531,
      "step": 59650
    },
    {
      "epoch": 0.9287209944966411,
      "grad_norm": 3.597801923751831,
      "learning_rate": 7.503974041723341e-08,
      "loss": 0.6939,
      "step": 59655
    },
    {
      "epoch": 0.9287988354986105,
      "grad_norm": 6.746609210968018,
      "learning_rate": 7.495780141262843e-08,
      "loss": 0.7497,
      "step": 59660
    },
    {
      "epoch": 0.9288766765005799,
      "grad_norm": 2.999969959259033,
      "learning_rate": 7.487586240802346e-08,
      "loss": 0.7259,
      "step": 59665
    },
    {
      "epoch": 0.9289545175025493,
      "grad_norm": 4.707868576049805,
      "learning_rate": 7.479392340341849e-08,
      "loss": 0.8148,
      "step": 59670
    },
    {
      "epoch": 0.9290323585045187,
      "grad_norm": 2.7683796882629395,
      "learning_rate": 7.471198439881352e-08,
      "loss": 0.6202,
      "step": 59675
    },
    {
      "epoch": 0.9291101995064881,
      "grad_norm": 12.682868957519531,
      "learning_rate": 7.463004539420854e-08,
      "loss": 0.7438,
      "step": 59680
    },
    {
      "epoch": 0.9291880405084574,
      "grad_norm": 2.3525238037109375,
      "learning_rate": 7.454810638960356e-08,
      "loss": 0.7252,
      "step": 59685
    },
    {
      "epoch": 0.9292658815104268,
      "grad_norm": 6.2097015380859375,
      "learning_rate": 7.44661673849986e-08,
      "loss": 0.7245,
      "step": 59690
    },
    {
      "epoch": 0.9293437225123962,
      "grad_norm": 4.363824844360352,
      "learning_rate": 7.438422838039362e-08,
      "loss": 0.8242,
      "step": 59695
    },
    {
      "epoch": 0.9294215635143656,
      "grad_norm": 3.298180103302002,
      "learning_rate": 7.430228937578866e-08,
      "loss": 0.8591,
      "step": 59700
    },
    {
      "epoch": 0.929499404516335,
      "grad_norm": 4.971315383911133,
      "learning_rate": 7.422035037118368e-08,
      "loss": 0.748,
      "step": 59705
    },
    {
      "epoch": 0.9295772455183043,
      "grad_norm": 4.006728649139404,
      "learning_rate": 7.413841136657871e-08,
      "loss": 0.8485,
      "step": 59710
    },
    {
      "epoch": 0.9296550865202737,
      "grad_norm": 6.1140666007995605,
      "learning_rate": 7.405647236197374e-08,
      "loss": 0.7129,
      "step": 59715
    },
    {
      "epoch": 0.929732927522243,
      "grad_norm": 2.7803354263305664,
      "learning_rate": 7.397453335736877e-08,
      "loss": 0.6484,
      "step": 59720
    },
    {
      "epoch": 0.9298107685242124,
      "grad_norm": 3.025056838989258,
      "learning_rate": 7.389259435276379e-08,
      "loss": 0.676,
      "step": 59725
    },
    {
      "epoch": 0.9298886095261818,
      "grad_norm": 7.240305423736572,
      "learning_rate": 7.381065534815882e-08,
      "loss": 0.7475,
      "step": 59730
    },
    {
      "epoch": 0.9299664505281512,
      "grad_norm": 4.079854488372803,
      "learning_rate": 7.372871634355385e-08,
      "loss": 0.7025,
      "step": 59735
    },
    {
      "epoch": 0.9300442915301206,
      "grad_norm": 3.2951784133911133,
      "learning_rate": 7.364677733894889e-08,
      "loss": 0.8087,
      "step": 59740
    },
    {
      "epoch": 0.93012213253209,
      "grad_norm": 3.135417938232422,
      "learning_rate": 7.356483833434392e-08,
      "loss": 0.7006,
      "step": 59745
    },
    {
      "epoch": 0.9301999735340594,
      "grad_norm": 4.740077018737793,
      "learning_rate": 7.348289932973894e-08,
      "loss": 0.8222,
      "step": 59750
    },
    {
      "epoch": 0.9302778145360288,
      "grad_norm": 3.5958807468414307,
      "learning_rate": 7.340096032513398e-08,
      "loss": 0.659,
      "step": 59755
    },
    {
      "epoch": 0.930355655537998,
      "grad_norm": 2.842545747756958,
      "learning_rate": 7.3319021320529e-08,
      "loss": 0.6821,
      "step": 59760
    },
    {
      "epoch": 0.9304334965399674,
      "grad_norm": 3.2747697830200195,
      "learning_rate": 7.323708231592403e-08,
      "loss": 0.6904,
      "step": 59765
    },
    {
      "epoch": 0.9305113375419368,
      "grad_norm": 3.85937237739563,
      "learning_rate": 7.315514331131906e-08,
      "loss": 0.7087,
      "step": 59770
    },
    {
      "epoch": 0.9305891785439062,
      "grad_norm": 4.821270942687988,
      "learning_rate": 7.307320430671408e-08,
      "loss": 0.7236,
      "step": 59775
    },
    {
      "epoch": 0.9306670195458756,
      "grad_norm": 2.3829762935638428,
      "learning_rate": 7.299126530210911e-08,
      "loss": 0.7226,
      "step": 59780
    },
    {
      "epoch": 0.930744860547845,
      "grad_norm": 3.326301336288452,
      "learning_rate": 7.290932629750414e-08,
      "loss": 0.7873,
      "step": 59785
    },
    {
      "epoch": 0.9308227015498144,
      "grad_norm": 3.2910797595977783,
      "learning_rate": 7.282738729289917e-08,
      "loss": 0.6754,
      "step": 59790
    },
    {
      "epoch": 0.9309005425517837,
      "grad_norm": 4.0624260902404785,
      "learning_rate": 7.274544828829419e-08,
      "loss": 0.7837,
      "step": 59795
    },
    {
      "epoch": 0.9309783835537531,
      "grad_norm": 4.673757553100586,
      "learning_rate": 7.266350928368923e-08,
      "loss": 0.6075,
      "step": 59800
    },
    {
      "epoch": 0.9310562245557225,
      "grad_norm": 3.106111526489258,
      "learning_rate": 7.258157027908425e-08,
      "loss": 0.8443,
      "step": 59805
    },
    {
      "epoch": 0.9311340655576918,
      "grad_norm": 6.423622131347656,
      "learning_rate": 7.249963127447927e-08,
      "loss": 0.8282,
      "step": 59810
    },
    {
      "epoch": 0.9312119065596612,
      "grad_norm": 7.8667707443237305,
      "learning_rate": 7.241769226987431e-08,
      "loss": 0.7313,
      "step": 59815
    },
    {
      "epoch": 0.9312897475616306,
      "grad_norm": 6.78395938873291,
      "learning_rate": 7.233575326526933e-08,
      "loss": 0.7897,
      "step": 59820
    },
    {
      "epoch": 0.9313675885636,
      "grad_norm": 7.128668785095215,
      "learning_rate": 7.225381426066436e-08,
      "loss": 0.7727,
      "step": 59825
    },
    {
      "epoch": 0.9314454295655694,
      "grad_norm": 4.4717864990234375,
      "learning_rate": 7.217187525605939e-08,
      "loss": 0.7351,
      "step": 59830
    },
    {
      "epoch": 0.9315232705675387,
      "grad_norm": 4.957362174987793,
      "learning_rate": 7.208993625145442e-08,
      "loss": 0.7752,
      "step": 59835
    },
    {
      "epoch": 0.9316011115695081,
      "grad_norm": 4.816576957702637,
      "learning_rate": 7.200799724684944e-08,
      "loss": 0.7454,
      "step": 59840
    },
    {
      "epoch": 0.9316789525714775,
      "grad_norm": 5.135659694671631,
      "learning_rate": 7.192605824224448e-08,
      "loss": 0.7511,
      "step": 59845
    },
    {
      "epoch": 0.9317567935734469,
      "grad_norm": 4.588697910308838,
      "learning_rate": 7.18441192376395e-08,
      "loss": 0.6411,
      "step": 59850
    },
    {
      "epoch": 0.9318346345754163,
      "grad_norm": 4.156418323516846,
      "learning_rate": 7.176218023303452e-08,
      "loss": 0.6161,
      "step": 59855
    },
    {
      "epoch": 0.9319124755773857,
      "grad_norm": 12.456742286682129,
      "learning_rate": 7.168024122842956e-08,
      "loss": 0.8447,
      "step": 59860
    },
    {
      "epoch": 0.931990316579355,
      "grad_norm": 3.861664295196533,
      "learning_rate": 7.159830222382458e-08,
      "loss": 0.7451,
      "step": 59865
    },
    {
      "epoch": 0.9320681575813244,
      "grad_norm": 3.2084126472473145,
      "learning_rate": 7.151636321921961e-08,
      "loss": 0.7694,
      "step": 59870
    },
    {
      "epoch": 0.9321459985832937,
      "grad_norm": 3.832240343093872,
      "learning_rate": 7.143442421461464e-08,
      "loss": 0.743,
      "step": 59875
    },
    {
      "epoch": 0.9322238395852631,
      "grad_norm": 3.7093605995178223,
      "learning_rate": 7.135248521000967e-08,
      "loss": 0.6472,
      "step": 59880
    },
    {
      "epoch": 0.9323016805872325,
      "grad_norm": 4.808413028717041,
      "learning_rate": 7.127054620540469e-08,
      "loss": 0.6133,
      "step": 59885
    },
    {
      "epoch": 0.9323795215892019,
      "grad_norm": 2.287787437438965,
      "learning_rate": 7.118860720079973e-08,
      "loss": 0.782,
      "step": 59890
    },
    {
      "epoch": 0.9324573625911713,
      "grad_norm": 3.244069814682007,
      "learning_rate": 7.110666819619475e-08,
      "loss": 0.7366,
      "step": 59895
    },
    {
      "epoch": 0.9325352035931407,
      "grad_norm": 3.155674934387207,
      "learning_rate": 7.102472919158977e-08,
      "loss": 0.6837,
      "step": 59900
    },
    {
      "epoch": 0.9326130445951101,
      "grad_norm": 6.534295082092285,
      "learning_rate": 7.094279018698481e-08,
      "loss": 0.7133,
      "step": 59905
    },
    {
      "epoch": 0.9326908855970794,
      "grad_norm": 3.777623414993286,
      "learning_rate": 7.086085118237983e-08,
      "loss": 0.7351,
      "step": 59910
    },
    {
      "epoch": 0.9327687265990487,
      "grad_norm": 3.8745462894439697,
      "learning_rate": 7.077891217777486e-08,
      "loss": 0.7705,
      "step": 59915
    },
    {
      "epoch": 0.9328465676010181,
      "grad_norm": 3.4919307231903076,
      "learning_rate": 7.069697317316989e-08,
      "loss": 0.6687,
      "step": 59920
    },
    {
      "epoch": 0.9329244086029875,
      "grad_norm": 6.292998790740967,
      "learning_rate": 7.061503416856492e-08,
      "loss": 0.743,
      "step": 59925
    },
    {
      "epoch": 0.9330022496049569,
      "grad_norm": 13.423583030700684,
      "learning_rate": 7.053309516395994e-08,
      "loss": 0.6264,
      "step": 59930
    },
    {
      "epoch": 0.9330800906069263,
      "grad_norm": 2.7625844478607178,
      "learning_rate": 7.045115615935497e-08,
      "loss": 0.7059,
      "step": 59935
    },
    {
      "epoch": 0.9331579316088957,
      "grad_norm": 4.623599529266357,
      "learning_rate": 7.036921715475e-08,
      "loss": 0.7444,
      "step": 59940
    },
    {
      "epoch": 0.9332357726108651,
      "grad_norm": 3.9874353408813477,
      "learning_rate": 7.028727815014502e-08,
      "loss": 0.7542,
      "step": 59945
    },
    {
      "epoch": 0.9333136136128344,
      "grad_norm": 3.5604848861694336,
      "learning_rate": 7.020533914554006e-08,
      "loss": 0.716,
      "step": 59950
    },
    {
      "epoch": 0.9333914546148038,
      "grad_norm": 5.212708950042725,
      "learning_rate": 7.012340014093508e-08,
      "loss": 0.7736,
      "step": 59955
    },
    {
      "epoch": 0.9334692956167732,
      "grad_norm": 3.0385992527008057,
      "learning_rate": 7.004146113633011e-08,
      "loss": 0.7235,
      "step": 59960
    },
    {
      "epoch": 0.9335471366187426,
      "grad_norm": 2.5399491786956787,
      "learning_rate": 6.995952213172514e-08,
      "loss": 0.6862,
      "step": 59965
    },
    {
      "epoch": 0.933624977620712,
      "grad_norm": 3.0798141956329346,
      "learning_rate": 6.987758312712017e-08,
      "loss": 0.732,
      "step": 59970
    },
    {
      "epoch": 0.9337028186226813,
      "grad_norm": 3.8062610626220703,
      "learning_rate": 6.97956441225152e-08,
      "loss": 0.8171,
      "step": 59975
    },
    {
      "epoch": 0.9337806596246507,
      "grad_norm": 3.744398593902588,
      "learning_rate": 6.971370511791022e-08,
      "loss": 0.7743,
      "step": 59980
    },
    {
      "epoch": 0.93385850062662,
      "grad_norm": 2.752143621444702,
      "learning_rate": 6.963176611330525e-08,
      "loss": 0.7773,
      "step": 59985
    },
    {
      "epoch": 0.9339363416285894,
      "grad_norm": 3.2912344932556152,
      "learning_rate": 6.954982710870027e-08,
      "loss": 0.7145,
      "step": 59990
    },
    {
      "epoch": 0.9340141826305588,
      "grad_norm": 2.306467294692993,
      "learning_rate": 6.946788810409531e-08,
      "loss": 0.714,
      "step": 59995
    },
    {
      "epoch": 0.9340920236325282,
      "grad_norm": 3.3383665084838867,
      "learning_rate": 6.938594909949033e-08,
      "loss": 0.6706,
      "step": 60000
    },
    {
      "epoch": 0.9341698646344976,
      "grad_norm": 9.248139381408691,
      "learning_rate": 6.930401009488536e-08,
      "loss": 0.6228,
      "step": 60005
    },
    {
      "epoch": 0.934247705636467,
      "grad_norm": 3.5401771068573,
      "learning_rate": 6.922207109028039e-08,
      "loss": 0.6873,
      "step": 60010
    },
    {
      "epoch": 0.9343255466384364,
      "grad_norm": 2.4261281490325928,
      "learning_rate": 6.914013208567541e-08,
      "loss": 0.6656,
      "step": 60015
    },
    {
      "epoch": 0.9344033876404058,
      "grad_norm": 4.198561668395996,
      "learning_rate": 6.905819308107044e-08,
      "loss": 0.573,
      "step": 60020
    },
    {
      "epoch": 0.934481228642375,
      "grad_norm": 10.665823936462402,
      "learning_rate": 6.897625407646547e-08,
      "loss": 0.6781,
      "step": 60025
    },
    {
      "epoch": 0.9345590696443444,
      "grad_norm": 4.847423553466797,
      "learning_rate": 6.88943150718605e-08,
      "loss": 0.5667,
      "step": 60030
    },
    {
      "epoch": 0.9346369106463138,
      "grad_norm": 7.99983024597168,
      "learning_rate": 6.881237606725552e-08,
      "loss": 0.8024,
      "step": 60035
    },
    {
      "epoch": 0.9347147516482832,
      "grad_norm": 5.135144233703613,
      "learning_rate": 6.873043706265056e-08,
      "loss": 0.6649,
      "step": 60040
    },
    {
      "epoch": 0.9347925926502526,
      "grad_norm": 3.149979829788208,
      "learning_rate": 6.864849805804558e-08,
      "loss": 0.6764,
      "step": 60045
    },
    {
      "epoch": 0.934870433652222,
      "grad_norm": 3.3438751697540283,
      "learning_rate": 6.856655905344062e-08,
      "loss": 0.6825,
      "step": 60050
    },
    {
      "epoch": 0.9349482746541914,
      "grad_norm": 4.264763832092285,
      "learning_rate": 6.848462004883564e-08,
      "loss": 0.7203,
      "step": 60055
    },
    {
      "epoch": 0.9350261156561607,
      "grad_norm": 3.3025712966918945,
      "learning_rate": 6.840268104423067e-08,
      "loss": 0.8268,
      "step": 60060
    },
    {
      "epoch": 0.9351039566581301,
      "grad_norm": 3.950042486190796,
      "learning_rate": 6.832074203962571e-08,
      "loss": 0.7455,
      "step": 60065
    },
    {
      "epoch": 0.9351817976600995,
      "grad_norm": 4.041086196899414,
      "learning_rate": 6.823880303502073e-08,
      "loss": 0.7669,
      "step": 60070
    },
    {
      "epoch": 0.9352596386620688,
      "grad_norm": 2.444018840789795,
      "learning_rate": 6.815686403041576e-08,
      "loss": 0.7164,
      "step": 60075
    },
    {
      "epoch": 0.9353374796640382,
      "grad_norm": 5.785805702209473,
      "learning_rate": 6.807492502581079e-08,
      "loss": 0.6538,
      "step": 60080
    },
    {
      "epoch": 0.9354153206660076,
      "grad_norm": 5.092484474182129,
      "learning_rate": 6.799298602120582e-08,
      "loss": 0.7531,
      "step": 60085
    },
    {
      "epoch": 0.935493161667977,
      "grad_norm": 4.167462348937988,
      "learning_rate": 6.791104701660084e-08,
      "loss": 0.7173,
      "step": 60090
    },
    {
      "epoch": 0.9355710026699464,
      "grad_norm": 4.007900714874268,
      "learning_rate": 6.782910801199588e-08,
      "loss": 0.7443,
      "step": 60095
    },
    {
      "epoch": 0.9356488436719157,
      "grad_norm": 3.030791997909546,
      "learning_rate": 6.77471690073909e-08,
      "loss": 0.6167,
      "step": 60100
    },
    {
      "epoch": 0.9357266846738851,
      "grad_norm": 4.159018516540527,
      "learning_rate": 6.766523000278592e-08,
      "loss": 0.7703,
      "step": 60105
    },
    {
      "epoch": 0.9358045256758545,
      "grad_norm": 3.0988101959228516,
      "learning_rate": 6.758329099818096e-08,
      "loss": 0.7125,
      "step": 60110
    },
    {
      "epoch": 0.9358823666778239,
      "grad_norm": 9.885687828063965,
      "learning_rate": 6.750135199357598e-08,
      "loss": 0.8406,
      "step": 60115
    },
    {
      "epoch": 0.9359602076797933,
      "grad_norm": 5.495907783508301,
      "learning_rate": 6.741941298897101e-08,
      "loss": 0.709,
      "step": 60120
    },
    {
      "epoch": 0.9360380486817627,
      "grad_norm": 3.8880996704101562,
      "learning_rate": 6.733747398436604e-08,
      "loss": 0.7182,
      "step": 60125
    },
    {
      "epoch": 0.936115889683732,
      "grad_norm": 3.500849962234497,
      "learning_rate": 6.725553497976107e-08,
      "loss": 0.7391,
      "step": 60130
    },
    {
      "epoch": 0.9361937306857014,
      "grad_norm": 3.9161407947540283,
      "learning_rate": 6.71735959751561e-08,
      "loss": 0.5996,
      "step": 60135
    },
    {
      "epoch": 0.9362715716876707,
      "grad_norm": 3.585925340652466,
      "learning_rate": 6.709165697055112e-08,
      "loss": 0.8066,
      "step": 60140
    },
    {
      "epoch": 0.9363494126896401,
      "grad_norm": 3.880418539047241,
      "learning_rate": 6.700971796594615e-08,
      "loss": 0.7009,
      "step": 60145
    },
    {
      "epoch": 0.9364272536916095,
      "grad_norm": 3.6753780841827393,
      "learning_rate": 6.692777896134117e-08,
      "loss": 0.7396,
      "step": 60150
    },
    {
      "epoch": 0.9365050946935789,
      "grad_norm": 5.68637228012085,
      "learning_rate": 6.684583995673621e-08,
      "loss": 0.7774,
      "step": 60155
    },
    {
      "epoch": 0.9365829356955483,
      "grad_norm": 2.5429999828338623,
      "learning_rate": 6.676390095213123e-08,
      "loss": 0.5958,
      "step": 60160
    },
    {
      "epoch": 0.9366607766975177,
      "grad_norm": 4.886886119842529,
      "learning_rate": 6.668196194752626e-08,
      "loss": 0.7344,
      "step": 60165
    },
    {
      "epoch": 0.9367386176994871,
      "grad_norm": 5.84676456451416,
      "learning_rate": 6.660002294292129e-08,
      "loss": 0.7706,
      "step": 60170
    },
    {
      "epoch": 0.9368164587014564,
      "grad_norm": 5.367573261260986,
      "learning_rate": 6.651808393831632e-08,
      "loss": 0.7104,
      "step": 60175
    },
    {
      "epoch": 0.9368942997034257,
      "grad_norm": 3.554777145385742,
      "learning_rate": 6.643614493371134e-08,
      "loss": 0.7374,
      "step": 60180
    },
    {
      "epoch": 0.9369721407053951,
      "grad_norm": 2.4250099658966064,
      "learning_rate": 6.635420592910637e-08,
      "loss": 0.667,
      "step": 60185
    },
    {
      "epoch": 0.9370499817073645,
      "grad_norm": 3.2194149494171143,
      "learning_rate": 6.62722669245014e-08,
      "loss": 0.7491,
      "step": 60190
    },
    {
      "epoch": 0.9371278227093339,
      "grad_norm": 3.9318976402282715,
      "learning_rate": 6.619032791989642e-08,
      "loss": 0.802,
      "step": 60195
    },
    {
      "epoch": 0.9372056637113033,
      "grad_norm": 7.2495903968811035,
      "learning_rate": 6.610838891529146e-08,
      "loss": 0.727,
      "step": 60200
    },
    {
      "epoch": 0.9372835047132727,
      "grad_norm": 3.3399229049682617,
      "learning_rate": 6.602644991068648e-08,
      "loss": 0.7026,
      "step": 60205
    },
    {
      "epoch": 0.9373613457152421,
      "grad_norm": 4.995519161224365,
      "learning_rate": 6.594451090608152e-08,
      "loss": 0.7109,
      "step": 60210
    },
    {
      "epoch": 0.9374391867172114,
      "grad_norm": 2.5789616107940674,
      "learning_rate": 6.586257190147654e-08,
      "loss": 0.8244,
      "step": 60215
    },
    {
      "epoch": 0.9375170277191808,
      "grad_norm": 2.5637099742889404,
      "learning_rate": 6.578063289687157e-08,
      "loss": 0.5901,
      "step": 60220
    },
    {
      "epoch": 0.9375948687211502,
      "grad_norm": 4.411526679992676,
      "learning_rate": 6.56986938922666e-08,
      "loss": 0.6459,
      "step": 60225
    },
    {
      "epoch": 0.9376727097231196,
      "grad_norm": 2.8018171787261963,
      "learning_rate": 6.561675488766162e-08,
      "loss": 0.6081,
      "step": 60230
    },
    {
      "epoch": 0.937750550725089,
      "grad_norm": 4.329617977142334,
      "learning_rate": 6.553481588305665e-08,
      "loss": 0.7508,
      "step": 60235
    },
    {
      "epoch": 0.9378283917270583,
      "grad_norm": 2.9975616931915283,
      "learning_rate": 6.545287687845167e-08,
      "loss": 0.6076,
      "step": 60240
    },
    {
      "epoch": 0.9379062327290277,
      "grad_norm": 4.602701663970947,
      "learning_rate": 6.537093787384671e-08,
      "loss": 0.6593,
      "step": 60245
    },
    {
      "epoch": 0.937984073730997,
      "grad_norm": 6.0046234130859375,
      "learning_rate": 6.528899886924173e-08,
      "loss": 0.7036,
      "step": 60250
    },
    {
      "epoch": 0.9380619147329664,
      "grad_norm": 3.7514877319335938,
      "learning_rate": 6.520705986463677e-08,
      "loss": 0.688,
      "step": 60255
    },
    {
      "epoch": 0.9381397557349358,
      "grad_norm": 4.866700172424316,
      "learning_rate": 6.512512086003179e-08,
      "loss": 0.6305,
      "step": 60260
    },
    {
      "epoch": 0.9382175967369052,
      "grad_norm": 4.352721691131592,
      "learning_rate": 6.504318185542681e-08,
      "loss": 0.7927,
      "step": 60265
    },
    {
      "epoch": 0.9382954377388746,
      "grad_norm": 3.5884482860565186,
      "learning_rate": 6.496124285082184e-08,
      "loss": 0.6872,
      "step": 60270
    },
    {
      "epoch": 0.938373278740844,
      "grad_norm": 2.8917505741119385,
      "learning_rate": 6.487930384621687e-08,
      "loss": 0.6765,
      "step": 60275
    },
    {
      "epoch": 0.9384511197428134,
      "grad_norm": 3.5155842304229736,
      "learning_rate": 6.47973648416119e-08,
      "loss": 0.7663,
      "step": 60280
    },
    {
      "epoch": 0.9385289607447828,
      "grad_norm": 3.5440704822540283,
      "learning_rate": 6.471542583700692e-08,
      "loss": 0.7533,
      "step": 60285
    },
    {
      "epoch": 0.938606801746752,
      "grad_norm": 2.970611572265625,
      "learning_rate": 6.463348683240196e-08,
      "loss": 0.6884,
      "step": 60290
    },
    {
      "epoch": 0.9386846427487214,
      "grad_norm": 2.762566566467285,
      "learning_rate": 6.455154782779698e-08,
      "loss": 0.7469,
      "step": 60295
    },
    {
      "epoch": 0.9387624837506908,
      "grad_norm": 5.0357537269592285,
      "learning_rate": 6.446960882319202e-08,
      "loss": 0.7587,
      "step": 60300
    },
    {
      "epoch": 0.9388403247526602,
      "grad_norm": 3.75887393951416,
      "learning_rate": 6.438766981858704e-08,
      "loss": 0.6718,
      "step": 60305
    },
    {
      "epoch": 0.9389181657546296,
      "grad_norm": 3.5064969062805176,
      "learning_rate": 6.430573081398206e-08,
      "loss": 0.7277,
      "step": 60310
    },
    {
      "epoch": 0.938996006756599,
      "grad_norm": 9.85833740234375,
      "learning_rate": 6.42237918093771e-08,
      "loss": 0.7548,
      "step": 60315
    },
    {
      "epoch": 0.9390738477585684,
      "grad_norm": 3.1019344329833984,
      "learning_rate": 6.414185280477212e-08,
      "loss": 0.7561,
      "step": 60320
    },
    {
      "epoch": 0.9391516887605377,
      "grad_norm": 4.672630786895752,
      "learning_rate": 6.405991380016715e-08,
      "loss": 0.7252,
      "step": 60325
    },
    {
      "epoch": 0.9392295297625071,
      "grad_norm": 3.08595609664917,
      "learning_rate": 6.397797479556217e-08,
      "loss": 0.7293,
      "step": 60330
    },
    {
      "epoch": 0.9393073707644765,
      "grad_norm": 6.159271240234375,
      "learning_rate": 6.389603579095721e-08,
      "loss": 0.7008,
      "step": 60335
    },
    {
      "epoch": 0.9393852117664458,
      "grad_norm": 3.258626699447632,
      "learning_rate": 6.381409678635223e-08,
      "loss": 0.7105,
      "step": 60340
    },
    {
      "epoch": 0.9394630527684152,
      "grad_norm": 5.65585470199585,
      "learning_rate": 6.373215778174725e-08,
      "loss": 0.8413,
      "step": 60345
    },
    {
      "epoch": 0.9395408937703846,
      "grad_norm": 3.684997797012329,
      "learning_rate": 6.365021877714229e-08,
      "loss": 0.6826,
      "step": 60350
    },
    {
      "epoch": 0.939618734772354,
      "grad_norm": 3.8016915321350098,
      "learning_rate": 6.356827977253731e-08,
      "loss": 0.6835,
      "step": 60355
    },
    {
      "epoch": 0.9396965757743234,
      "grad_norm": 4.181707382202148,
      "learning_rate": 6.348634076793234e-08,
      "loss": 0.7434,
      "step": 60360
    },
    {
      "epoch": 0.9397744167762927,
      "grad_norm": 4.334168910980225,
      "learning_rate": 6.340440176332737e-08,
      "loss": 0.792,
      "step": 60365
    },
    {
      "epoch": 0.9398522577782621,
      "grad_norm": 6.67325496673584,
      "learning_rate": 6.33224627587224e-08,
      "loss": 0.7561,
      "step": 60370
    },
    {
      "epoch": 0.9399300987802315,
      "grad_norm": 3.5024704933166504,
      "learning_rate": 6.324052375411742e-08,
      "loss": 0.6651,
      "step": 60375
    },
    {
      "epoch": 0.9400079397822009,
      "grad_norm": 3.7136452198028564,
      "learning_rate": 6.315858474951247e-08,
      "loss": 0.7489,
      "step": 60380
    },
    {
      "epoch": 0.9400857807841703,
      "grad_norm": 3.683950901031494,
      "learning_rate": 6.30766457449075e-08,
      "loss": 0.6703,
      "step": 60385
    },
    {
      "epoch": 0.9401636217861397,
      "grad_norm": 3.379873275756836,
      "learning_rate": 6.299470674030252e-08,
      "loss": 0.7149,
      "step": 60390
    },
    {
      "epoch": 0.940241462788109,
      "grad_norm": 3.7318007946014404,
      "learning_rate": 6.291276773569755e-08,
      "loss": 0.7274,
      "step": 60395
    },
    {
      "epoch": 0.9403193037900783,
      "grad_norm": 19.889413833618164,
      "learning_rate": 6.283082873109257e-08,
      "loss": 0.6598,
      "step": 60400
    },
    {
      "epoch": 0.9403971447920477,
      "grad_norm": 4.658337593078613,
      "learning_rate": 6.274888972648761e-08,
      "loss": 0.7266,
      "step": 60405
    },
    {
      "epoch": 0.9404749857940171,
      "grad_norm": 2.434563159942627,
      "learning_rate": 6.266695072188263e-08,
      "loss": 0.6691,
      "step": 60410
    },
    {
      "epoch": 0.9405528267959865,
      "grad_norm": 3.378696918487549,
      "learning_rate": 6.258501171727767e-08,
      "loss": 0.7531,
      "step": 60415
    },
    {
      "epoch": 0.9406306677979559,
      "grad_norm": 2.953840970993042,
      "learning_rate": 6.250307271267269e-08,
      "loss": 0.674,
      "step": 60420
    },
    {
      "epoch": 0.9407085087999253,
      "grad_norm": 3.771686553955078,
      "learning_rate": 6.242113370806771e-08,
      "loss": 0.7445,
      "step": 60425
    },
    {
      "epoch": 0.9407863498018947,
      "grad_norm": 3.5361337661743164,
      "learning_rate": 6.233919470346273e-08,
      "loss": 0.7083,
      "step": 60430
    },
    {
      "epoch": 0.9408641908038641,
      "grad_norm": 5.245040416717529,
      "learning_rate": 6.225725569885777e-08,
      "loss": 0.7036,
      "step": 60435
    },
    {
      "epoch": 0.9409420318058334,
      "grad_norm": 4.356694221496582,
      "learning_rate": 6.217531669425279e-08,
      "loss": 0.767,
      "step": 60440
    },
    {
      "epoch": 0.9410198728078027,
      "grad_norm": 3.370082139968872,
      "learning_rate": 6.209337768964782e-08,
      "loss": 0.8115,
      "step": 60445
    },
    {
      "epoch": 0.9410977138097721,
      "grad_norm": 7.7370924949646,
      "learning_rate": 6.201143868504285e-08,
      "loss": 0.6997,
      "step": 60450
    },
    {
      "epoch": 0.9411755548117415,
      "grad_norm": 3.036024332046509,
      "learning_rate": 6.192949968043788e-08,
      "loss": 0.6766,
      "step": 60455
    },
    {
      "epoch": 0.9412533958137109,
      "grad_norm": 3.472533702850342,
      "learning_rate": 6.184756067583292e-08,
      "loss": 0.7663,
      "step": 60460
    },
    {
      "epoch": 0.9413312368156803,
      "grad_norm": 4.009667873382568,
      "learning_rate": 6.176562167122794e-08,
      "loss": 0.8282,
      "step": 60465
    },
    {
      "epoch": 0.9414090778176497,
      "grad_norm": 3.0732386112213135,
      "learning_rate": 6.168368266662297e-08,
      "loss": 0.7176,
      "step": 60470
    },
    {
      "epoch": 0.9414869188196191,
      "grad_norm": 3.399383544921875,
      "learning_rate": 6.1601743662018e-08,
      "loss": 0.7257,
      "step": 60475
    },
    {
      "epoch": 0.9415647598215884,
      "grad_norm": 3.04971981048584,
      "learning_rate": 6.151980465741302e-08,
      "loss": 0.7843,
      "step": 60480
    },
    {
      "epoch": 0.9416426008235578,
      "grad_norm": 10.264163970947266,
      "learning_rate": 6.143786565280805e-08,
      "loss": 0.6442,
      "step": 60485
    },
    {
      "epoch": 0.9417204418255272,
      "grad_norm": 3.6447947025299072,
      "learning_rate": 6.135592664820307e-08,
      "loss": 0.795,
      "step": 60490
    },
    {
      "epoch": 0.9417982828274966,
      "grad_norm": 2.9862782955169678,
      "learning_rate": 6.127398764359811e-08,
      "loss": 0.7109,
      "step": 60495
    },
    {
      "epoch": 0.941876123829466,
      "grad_norm": 4.2692365646362305,
      "learning_rate": 6.119204863899313e-08,
      "loss": 0.7117,
      "step": 60500
    },
    {
      "epoch": 0.9419539648314353,
      "grad_norm": 3.375166654586792,
      "learning_rate": 6.111010963438817e-08,
      "loss": 0.7272,
      "step": 60505
    },
    {
      "epoch": 0.9420318058334047,
      "grad_norm": 7.254694938659668,
      "learning_rate": 6.102817062978319e-08,
      "loss": 0.6869,
      "step": 60510
    },
    {
      "epoch": 0.942109646835374,
      "grad_norm": 3.3353707790374756,
      "learning_rate": 6.094623162517821e-08,
      "loss": 0.7155,
      "step": 60515
    },
    {
      "epoch": 0.9421874878373434,
      "grad_norm": 3.677379846572876,
      "learning_rate": 6.086429262057324e-08,
      "loss": 0.7079,
      "step": 60520
    },
    {
      "epoch": 0.9422653288393128,
      "grad_norm": 6.686629295349121,
      "learning_rate": 6.078235361596827e-08,
      "loss": 0.8071,
      "step": 60525
    },
    {
      "epoch": 0.9423431698412822,
      "grad_norm": 4.760143280029297,
      "learning_rate": 6.07004146113633e-08,
      "loss": 0.6543,
      "step": 60530
    },
    {
      "epoch": 0.9424210108432516,
      "grad_norm": 2.924347400665283,
      "learning_rate": 6.061847560675832e-08,
      "loss": 0.6693,
      "step": 60535
    },
    {
      "epoch": 0.942498851845221,
      "grad_norm": 6.81570291519165,
      "learning_rate": 6.053653660215336e-08,
      "loss": 0.6804,
      "step": 60540
    },
    {
      "epoch": 0.9425766928471904,
      "grad_norm": 2.882485866546631,
      "learning_rate": 6.045459759754838e-08,
      "loss": 0.7872,
      "step": 60545
    },
    {
      "epoch": 0.9426545338491598,
      "grad_norm": 3.9764084815979004,
      "learning_rate": 6.037265859294342e-08,
      "loss": 0.7363,
      "step": 60550
    },
    {
      "epoch": 0.942732374851129,
      "grad_norm": 8.050917625427246,
      "learning_rate": 6.029071958833844e-08,
      "loss": 0.8581,
      "step": 60555
    },
    {
      "epoch": 0.9428102158530984,
      "grad_norm": 8.460494041442871,
      "learning_rate": 6.020878058373346e-08,
      "loss": 0.7788,
      "step": 60560
    },
    {
      "epoch": 0.9428880568550678,
      "grad_norm": 5.507726192474365,
      "learning_rate": 6.01268415791285e-08,
      "loss": 0.7399,
      "step": 60565
    },
    {
      "epoch": 0.9429658978570372,
      "grad_norm": 3.810924768447876,
      "learning_rate": 6.004490257452352e-08,
      "loss": 0.7273,
      "step": 60570
    },
    {
      "epoch": 0.9430437388590066,
      "grad_norm": 3.512171983718872,
      "learning_rate": 5.996296356991855e-08,
      "loss": 0.6983,
      "step": 60575
    },
    {
      "epoch": 0.943121579860976,
      "grad_norm": 4.054419994354248,
      "learning_rate": 5.988102456531357e-08,
      "loss": 0.8036,
      "step": 60580
    },
    {
      "epoch": 0.9431994208629454,
      "grad_norm": 3.460440158843994,
      "learning_rate": 5.979908556070861e-08,
      "loss": 0.6608,
      "step": 60585
    },
    {
      "epoch": 0.9432772618649147,
      "grad_norm": 2.9246089458465576,
      "learning_rate": 5.971714655610363e-08,
      "loss": 0.6229,
      "step": 60590
    },
    {
      "epoch": 0.9433551028668841,
      "grad_norm": 8.033243179321289,
      "learning_rate": 5.963520755149865e-08,
      "loss": 0.685,
      "step": 60595
    },
    {
      "epoch": 0.9434329438688535,
      "grad_norm": 4.185390472412109,
      "learning_rate": 5.955326854689369e-08,
      "loss": 0.6542,
      "step": 60600
    },
    {
      "epoch": 0.9435107848708228,
      "grad_norm": 4.128832817077637,
      "learning_rate": 5.9471329542288717e-08,
      "loss": 0.7829,
      "step": 60605
    },
    {
      "epoch": 0.9435886258727922,
      "grad_norm": 8.899298667907715,
      "learning_rate": 5.9389390537683745e-08,
      "loss": 0.7054,
      "step": 60610
    },
    {
      "epoch": 0.9436664668747616,
      "grad_norm": 6.887772560119629,
      "learning_rate": 5.930745153307877e-08,
      "loss": 0.8326,
      "step": 60615
    },
    {
      "epoch": 0.943744307876731,
      "grad_norm": 3.0375444889068604,
      "learning_rate": 5.92255125284738e-08,
      "loss": 0.722,
      "step": 60620
    },
    {
      "epoch": 0.9438221488787004,
      "grad_norm": 7.020227432250977,
      "learning_rate": 5.914357352386883e-08,
      "loss": 0.6884,
      "step": 60625
    },
    {
      "epoch": 0.9438999898806697,
      "grad_norm": 5.360675811767578,
      "learning_rate": 5.906163451926386e-08,
      "loss": 0.8371,
      "step": 60630
    },
    {
      "epoch": 0.9439778308826391,
      "grad_norm": 4.4451751708984375,
      "learning_rate": 5.897969551465889e-08,
      "loss": 0.7228,
      "step": 60635
    },
    {
      "epoch": 0.9440556718846085,
      "grad_norm": 3.0617239475250244,
      "learning_rate": 5.8897756510053916e-08,
      "loss": 0.8067,
      "step": 60640
    },
    {
      "epoch": 0.9441335128865779,
      "grad_norm": 5.193785667419434,
      "learning_rate": 5.8815817505448945e-08,
      "loss": 0.8165,
      "step": 60645
    },
    {
      "epoch": 0.9442113538885473,
      "grad_norm": 4.7190937995910645,
      "learning_rate": 5.8733878500843973e-08,
      "loss": 0.6705,
      "step": 60650
    },
    {
      "epoch": 0.9442891948905167,
      "grad_norm": 2.360386371612549,
      "learning_rate": 5.8651939496239e-08,
      "loss": 0.7601,
      "step": 60655
    },
    {
      "epoch": 0.944367035892486,
      "grad_norm": 7.502525329589844,
      "learning_rate": 5.8570000491634024e-08,
      "loss": 0.5989,
      "step": 60660
    },
    {
      "epoch": 0.9444448768944553,
      "grad_norm": 5.209514141082764,
      "learning_rate": 5.848806148702905e-08,
      "loss": 0.7294,
      "step": 60665
    },
    {
      "epoch": 0.9445227178964247,
      "grad_norm": 2.716867208480835,
      "learning_rate": 5.840612248242408e-08,
      "loss": 0.7572,
      "step": 60670
    },
    {
      "epoch": 0.9446005588983941,
      "grad_norm": 2.9535562992095947,
      "learning_rate": 5.832418347781911e-08,
      "loss": 0.7317,
      "step": 60675
    },
    {
      "epoch": 0.9446783999003635,
      "grad_norm": 4.763754367828369,
      "learning_rate": 5.824224447321414e-08,
      "loss": 0.6576,
      "step": 60680
    },
    {
      "epoch": 0.9447562409023329,
      "grad_norm": 3.0685250759124756,
      "learning_rate": 5.8160305468609167e-08,
      "loss": 0.7203,
      "step": 60685
    },
    {
      "epoch": 0.9448340819043023,
      "grad_norm": 4.547595024108887,
      "learning_rate": 5.8078366464004195e-08,
      "loss": 0.6974,
      "step": 60690
    },
    {
      "epoch": 0.9449119229062717,
      "grad_norm": 11.188638687133789,
      "learning_rate": 5.7996427459399224e-08,
      "loss": 0.83,
      "step": 60695
    },
    {
      "epoch": 0.9449897639082411,
      "grad_norm": 3.214197874069214,
      "learning_rate": 5.7914488454794246e-08,
      "loss": 0.6841,
      "step": 60700
    },
    {
      "epoch": 0.9450676049102104,
      "grad_norm": 5.486057281494141,
      "learning_rate": 5.7832549450189274e-08,
      "loss": 0.7586,
      "step": 60705
    },
    {
      "epoch": 0.9451454459121797,
      "grad_norm": 3.3812358379364014,
      "learning_rate": 5.77506104455843e-08,
      "loss": 0.6893,
      "step": 60710
    },
    {
      "epoch": 0.9452232869141491,
      "grad_norm": 2.576131820678711,
      "learning_rate": 5.766867144097933e-08,
      "loss": 0.6989,
      "step": 60715
    },
    {
      "epoch": 0.9453011279161185,
      "grad_norm": 3.1304492950439453,
      "learning_rate": 5.758673243637436e-08,
      "loss": 0.7123,
      "step": 60720
    },
    {
      "epoch": 0.9453789689180879,
      "grad_norm": 4.1172075271606445,
      "learning_rate": 5.750479343176939e-08,
      "loss": 0.6566,
      "step": 60725
    },
    {
      "epoch": 0.9454568099200573,
      "grad_norm": 3.27878999710083,
      "learning_rate": 5.742285442716442e-08,
      "loss": 0.7498,
      "step": 60730
    },
    {
      "epoch": 0.9455346509220267,
      "grad_norm": 3.095766544342041,
      "learning_rate": 5.7340915422559445e-08,
      "loss": 0.7812,
      "step": 60735
    },
    {
      "epoch": 0.945612491923996,
      "grad_norm": 4.260377883911133,
      "learning_rate": 5.725897641795447e-08,
      "loss": 0.7966,
      "step": 60740
    },
    {
      "epoch": 0.9456903329259654,
      "grad_norm": 7.270799160003662,
      "learning_rate": 5.7177037413349496e-08,
      "loss": 0.7379,
      "step": 60745
    },
    {
      "epoch": 0.9457681739279348,
      "grad_norm": 4.253818988800049,
      "learning_rate": 5.7095098408744524e-08,
      "loss": 0.6106,
      "step": 60750
    },
    {
      "epoch": 0.9458460149299042,
      "grad_norm": 4.128617286682129,
      "learning_rate": 5.701315940413955e-08,
      "loss": 0.7478,
      "step": 60755
    },
    {
      "epoch": 0.9459238559318736,
      "grad_norm": 2.716435432434082,
      "learning_rate": 5.693122039953458e-08,
      "loss": 0.7,
      "step": 60760
    },
    {
      "epoch": 0.946001696933843,
      "grad_norm": 2.3667564392089844,
      "learning_rate": 5.684928139492961e-08,
      "loss": 0.571,
      "step": 60765
    },
    {
      "epoch": 0.9460795379358123,
      "grad_norm": 3.593963861465454,
      "learning_rate": 5.676734239032464e-08,
      "loss": 0.7649,
      "step": 60770
    },
    {
      "epoch": 0.9461573789377817,
      "grad_norm": 2.866363286972046,
      "learning_rate": 5.6685403385719674e-08,
      "loss": 0.7148,
      "step": 60775
    },
    {
      "epoch": 0.946235219939751,
      "grad_norm": 3.037990093231201,
      "learning_rate": 5.66034643811147e-08,
      "loss": 0.647,
      "step": 60780
    },
    {
      "epoch": 0.9463130609417204,
      "grad_norm": 3.284942865371704,
      "learning_rate": 5.6521525376509724e-08,
      "loss": 0.6865,
      "step": 60785
    },
    {
      "epoch": 0.9463909019436898,
      "grad_norm": 2.9224517345428467,
      "learning_rate": 5.643958637190475e-08,
      "loss": 0.6834,
      "step": 60790
    },
    {
      "epoch": 0.9464687429456592,
      "grad_norm": 5.713934421539307,
      "learning_rate": 5.635764736729978e-08,
      "loss": 0.6687,
      "step": 60795
    },
    {
      "epoch": 0.9465465839476286,
      "grad_norm": 2.975635528564453,
      "learning_rate": 5.627570836269481e-08,
      "loss": 0.6927,
      "step": 60800
    },
    {
      "epoch": 0.946624424949598,
      "grad_norm": 3.9365851879119873,
      "learning_rate": 5.619376935808984e-08,
      "loss": 0.7231,
      "step": 60805
    },
    {
      "epoch": 0.9467022659515674,
      "grad_norm": 3.5268301963806152,
      "learning_rate": 5.611183035348487e-08,
      "loss": 0.7868,
      "step": 60810
    },
    {
      "epoch": 0.9467801069535368,
      "grad_norm": 3.9385900497436523,
      "learning_rate": 5.6029891348879895e-08,
      "loss": 0.7887,
      "step": 60815
    },
    {
      "epoch": 0.946857947955506,
      "grad_norm": 9.695856094360352,
      "learning_rate": 5.5947952344274924e-08,
      "loss": 0.7105,
      "step": 60820
    },
    {
      "epoch": 0.9469357889574754,
      "grad_norm": 2.7372703552246094,
      "learning_rate": 5.5866013339669946e-08,
      "loss": 0.6367,
      "step": 60825
    },
    {
      "epoch": 0.9470136299594448,
      "grad_norm": 9.145118713378906,
      "learning_rate": 5.5784074335064974e-08,
      "loss": 0.66,
      "step": 60830
    },
    {
      "epoch": 0.9470914709614142,
      "grad_norm": 3.727597713470459,
      "learning_rate": 5.570213533046e-08,
      "loss": 0.7635,
      "step": 60835
    },
    {
      "epoch": 0.9471693119633836,
      "grad_norm": 5.046937942504883,
      "learning_rate": 5.562019632585503e-08,
      "loss": 0.715,
      "step": 60840
    },
    {
      "epoch": 0.947247152965353,
      "grad_norm": 5.1223368644714355,
      "learning_rate": 5.553825732125006e-08,
      "loss": 0.7688,
      "step": 60845
    },
    {
      "epoch": 0.9473249939673224,
      "grad_norm": 4.4011549949646,
      "learning_rate": 5.545631831664509e-08,
      "loss": 0.7075,
      "step": 60850
    },
    {
      "epoch": 0.9474028349692917,
      "grad_norm": 2.8516948223114014,
      "learning_rate": 5.537437931204012e-08,
      "loss": 0.7454,
      "step": 60855
    },
    {
      "epoch": 0.9474806759712611,
      "grad_norm": 2.9155054092407227,
      "learning_rate": 5.5292440307435145e-08,
      "loss": 0.7623,
      "step": 60860
    },
    {
      "epoch": 0.9475585169732305,
      "grad_norm": 3.3459784984588623,
      "learning_rate": 5.521050130283017e-08,
      "loss": 0.8718,
      "step": 60865
    },
    {
      "epoch": 0.9476363579751999,
      "grad_norm": 7.657051086425781,
      "learning_rate": 5.5128562298225196e-08,
      "loss": 0.7786,
      "step": 60870
    },
    {
      "epoch": 0.9477141989771692,
      "grad_norm": 3.1906192302703857,
      "learning_rate": 5.5046623293620224e-08,
      "loss": 0.7575,
      "step": 60875
    },
    {
      "epoch": 0.9477920399791386,
      "grad_norm": 3.890854597091675,
      "learning_rate": 5.496468428901525e-08,
      "loss": 0.8057,
      "step": 60880
    },
    {
      "epoch": 0.947869880981108,
      "grad_norm": 3.859893798828125,
      "learning_rate": 5.488274528441028e-08,
      "loss": 0.7228,
      "step": 60885
    },
    {
      "epoch": 0.9479477219830774,
      "grad_norm": 3.5826621055603027,
      "learning_rate": 5.480080627980531e-08,
      "loss": 0.6166,
      "step": 60890
    },
    {
      "epoch": 0.9480255629850467,
      "grad_norm": 6.24113130569458,
      "learning_rate": 5.471886727520034e-08,
      "loss": 0.7488,
      "step": 60895
    },
    {
      "epoch": 0.9481034039870161,
      "grad_norm": 3.788102149963379,
      "learning_rate": 5.463692827059537e-08,
      "loss": 0.7491,
      "step": 60900
    },
    {
      "epoch": 0.9481812449889855,
      "grad_norm": 2.994426965713501,
      "learning_rate": 5.455498926599039e-08,
      "loss": 0.7613,
      "step": 60905
    },
    {
      "epoch": 0.9482590859909549,
      "grad_norm": 2.7034056186676025,
      "learning_rate": 5.447305026138542e-08,
      "loss": 0.6237,
      "step": 60910
    },
    {
      "epoch": 0.9483369269929243,
      "grad_norm": 4.3756489753723145,
      "learning_rate": 5.4391111256780446e-08,
      "loss": 0.7452,
      "step": 60915
    },
    {
      "epoch": 0.9484147679948937,
      "grad_norm": 2.2670297622680664,
      "learning_rate": 5.4309172252175475e-08,
      "loss": 0.591,
      "step": 60920
    },
    {
      "epoch": 0.948492608996863,
      "grad_norm": 8.331941604614258,
      "learning_rate": 5.42272332475705e-08,
      "loss": 0.727,
      "step": 60925
    },
    {
      "epoch": 0.9485704499988323,
      "grad_norm": 2.9130823612213135,
      "learning_rate": 5.414529424296553e-08,
      "loss": 0.6396,
      "step": 60930
    },
    {
      "epoch": 0.9486482910008017,
      "grad_norm": 4.928648948669434,
      "learning_rate": 5.406335523836057e-08,
      "loss": 0.5596,
      "step": 60935
    },
    {
      "epoch": 0.9487261320027711,
      "grad_norm": 3.7244009971618652,
      "learning_rate": 5.3981416233755595e-08,
      "loss": 0.8041,
      "step": 60940
    },
    {
      "epoch": 0.9488039730047405,
      "grad_norm": 4.148005962371826,
      "learning_rate": 5.3899477229150624e-08,
      "loss": 0.774,
      "step": 60945
    },
    {
      "epoch": 0.9488818140067099,
      "grad_norm": 4.985234260559082,
      "learning_rate": 5.3817538224545646e-08,
      "loss": 0.7806,
      "step": 60950
    },
    {
      "epoch": 0.9489596550086793,
      "grad_norm": 3.4454572200775146,
      "learning_rate": 5.3735599219940674e-08,
      "loss": 0.7236,
      "step": 60955
    },
    {
      "epoch": 0.9490374960106487,
      "grad_norm": 4.4720659255981445,
      "learning_rate": 5.36536602153357e-08,
      "loss": 0.6644,
      "step": 60960
    },
    {
      "epoch": 0.9491153370126181,
      "grad_norm": 2.841381311416626,
      "learning_rate": 5.357172121073073e-08,
      "loss": 0.8178,
      "step": 60965
    },
    {
      "epoch": 0.9491931780145874,
      "grad_norm": 5.443511009216309,
      "learning_rate": 5.348978220612576e-08,
      "loss": 0.6954,
      "step": 60970
    },
    {
      "epoch": 0.9492710190165567,
      "grad_norm": 3.809147357940674,
      "learning_rate": 5.340784320152079e-08,
      "loss": 0.7983,
      "step": 60975
    },
    {
      "epoch": 0.9493488600185261,
      "grad_norm": 5.2645158767700195,
      "learning_rate": 5.332590419691582e-08,
      "loss": 0.753,
      "step": 60980
    },
    {
      "epoch": 0.9494267010204955,
      "grad_norm": 5.038782119750977,
      "learning_rate": 5.3243965192310846e-08,
      "loss": 0.7616,
      "step": 60985
    },
    {
      "epoch": 0.9495045420224649,
      "grad_norm": 3.1030428409576416,
      "learning_rate": 5.316202618770587e-08,
      "loss": 0.7348,
      "step": 60990
    },
    {
      "epoch": 0.9495823830244343,
      "grad_norm": 3.654416561126709,
      "learning_rate": 5.3080087183100896e-08,
      "loss": 0.7103,
      "step": 60995
    },
    {
      "epoch": 0.9496602240264037,
      "grad_norm": 3.1049790382385254,
      "learning_rate": 5.2998148178495925e-08,
      "loss": 0.7441,
      "step": 61000
    },
    {
      "epoch": 0.949738065028373,
      "grad_norm": 3.698291301727295,
      "learning_rate": 5.291620917389095e-08,
      "loss": 0.6605,
      "step": 61005
    },
    {
      "epoch": 0.9498159060303424,
      "grad_norm": 3.564986228942871,
      "learning_rate": 5.283427016928598e-08,
      "loss": 0.6767,
      "step": 61010
    },
    {
      "epoch": 0.9498937470323118,
      "grad_norm": 5.105831146240234,
      "learning_rate": 5.275233116468101e-08,
      "loss": 0.6765,
      "step": 61015
    },
    {
      "epoch": 0.9499715880342812,
      "grad_norm": 3.982098340988159,
      "learning_rate": 5.267039216007604e-08,
      "loss": 0.7466,
      "step": 61020
    },
    {
      "epoch": 0.9500494290362506,
      "grad_norm": 5.261353015899658,
      "learning_rate": 5.258845315547107e-08,
      "loss": 0.6988,
      "step": 61025
    },
    {
      "epoch": 0.95012727003822,
      "grad_norm": 3.9734268188476562,
      "learning_rate": 5.250651415086609e-08,
      "loss": 0.6725,
      "step": 61030
    },
    {
      "epoch": 0.9502051110401893,
      "grad_norm": 4.888693332672119,
      "learning_rate": 5.242457514626112e-08,
      "loss": 0.7123,
      "step": 61035
    },
    {
      "epoch": 0.9502829520421587,
      "grad_norm": 4.026755332946777,
      "learning_rate": 5.2342636141656146e-08,
      "loss": 0.762,
      "step": 61040
    },
    {
      "epoch": 0.950360793044128,
      "grad_norm": 4.017175674438477,
      "learning_rate": 5.2260697137051175e-08,
      "loss": 0.8257,
      "step": 61045
    },
    {
      "epoch": 0.9504386340460974,
      "grad_norm": 2.273310422897339,
      "learning_rate": 5.2178758132446203e-08,
      "loss": 0.7676,
      "step": 61050
    },
    {
      "epoch": 0.9505164750480668,
      "grad_norm": 3.6603004932403564,
      "learning_rate": 5.209681912784123e-08,
      "loss": 0.7831,
      "step": 61055
    },
    {
      "epoch": 0.9505943160500362,
      "grad_norm": 2.782618522644043,
      "learning_rate": 5.201488012323626e-08,
      "loss": 0.7278,
      "step": 61060
    },
    {
      "epoch": 0.9506721570520056,
      "grad_norm": 3.536025285720825,
      "learning_rate": 5.193294111863129e-08,
      "loss": 0.7003,
      "step": 61065
    },
    {
      "epoch": 0.950749998053975,
      "grad_norm": 3.5883216857910156,
      "learning_rate": 5.185100211402631e-08,
      "loss": 0.679,
      "step": 61070
    },
    {
      "epoch": 0.9508278390559444,
      "grad_norm": 3.8136370182037354,
      "learning_rate": 5.176906310942134e-08,
      "loss": 0.6561,
      "step": 61075
    },
    {
      "epoch": 0.9509056800579138,
      "grad_norm": 3.21559476852417,
      "learning_rate": 5.168712410481637e-08,
      "loss": 0.7663,
      "step": 61080
    },
    {
      "epoch": 0.950983521059883,
      "grad_norm": 3.495474338531494,
      "learning_rate": 5.1605185100211397e-08,
      "loss": 0.7936,
      "step": 61085
    },
    {
      "epoch": 0.9510613620618524,
      "grad_norm": 3.563025951385498,
      "learning_rate": 5.1523246095606425e-08,
      "loss": 0.8341,
      "step": 61090
    },
    {
      "epoch": 0.9511392030638218,
      "grad_norm": 3.496880054473877,
      "learning_rate": 5.144130709100146e-08,
      "loss": 0.7639,
      "step": 61095
    },
    {
      "epoch": 0.9512170440657912,
      "grad_norm": 4.301735877990723,
      "learning_rate": 5.135936808639649e-08,
      "loss": 0.7607,
      "step": 61100
    },
    {
      "epoch": 0.9512948850677606,
      "grad_norm": 3.604102373123169,
      "learning_rate": 5.127742908179152e-08,
      "loss": 0.7337,
      "step": 61105
    },
    {
      "epoch": 0.95137272606973,
      "grad_norm": 6.088534832000732,
      "learning_rate": 5.1195490077186546e-08,
      "loss": 0.696,
      "step": 61110
    },
    {
      "epoch": 0.9514505670716994,
      "grad_norm": 3.601520299911499,
      "learning_rate": 5.111355107258157e-08,
      "loss": 0.8033,
      "step": 61115
    },
    {
      "epoch": 0.9515284080736687,
      "grad_norm": 3.553291082382202,
      "learning_rate": 5.1031612067976596e-08,
      "loss": 0.6802,
      "step": 61120
    },
    {
      "epoch": 0.9516062490756381,
      "grad_norm": 3.253790855407715,
      "learning_rate": 5.0949673063371625e-08,
      "loss": 0.8059,
      "step": 61125
    },
    {
      "epoch": 0.9516840900776075,
      "grad_norm": 8.609418869018555,
      "learning_rate": 5.0867734058766653e-08,
      "loss": 0.6699,
      "step": 61130
    },
    {
      "epoch": 0.9517619310795769,
      "grad_norm": 3.337374210357666,
      "learning_rate": 5.078579505416168e-08,
      "loss": 0.7827,
      "step": 61135
    },
    {
      "epoch": 0.9518397720815462,
      "grad_norm": 3.055790901184082,
      "learning_rate": 5.070385604955671e-08,
      "loss": 0.7007,
      "step": 61140
    },
    {
      "epoch": 0.9519176130835156,
      "grad_norm": 5.387094974517822,
      "learning_rate": 5.062191704495174e-08,
      "loss": 0.7507,
      "step": 61145
    },
    {
      "epoch": 0.951995454085485,
      "grad_norm": 11.491509437561035,
      "learning_rate": 5.053997804034677e-08,
      "loss": 0.8325,
      "step": 61150
    },
    {
      "epoch": 0.9520732950874544,
      "grad_norm": 6.358437538146973,
      "learning_rate": 5.045803903574179e-08,
      "loss": 0.7126,
      "step": 61155
    },
    {
      "epoch": 0.9521511360894237,
      "grad_norm": 6.507750988006592,
      "learning_rate": 5.037610003113682e-08,
      "loss": 0.7572,
      "step": 61160
    },
    {
      "epoch": 0.9522289770913931,
      "grad_norm": 3.367774486541748,
      "learning_rate": 5.0294161026531847e-08,
      "loss": 0.6486,
      "step": 61165
    },
    {
      "epoch": 0.9523068180933625,
      "grad_norm": 3.410095453262329,
      "learning_rate": 5.0212222021926875e-08,
      "loss": 0.9325,
      "step": 61170
    },
    {
      "epoch": 0.9523846590953319,
      "grad_norm": 2.795126438140869,
      "learning_rate": 5.0130283017321904e-08,
      "loss": 0.7154,
      "step": 61175
    },
    {
      "epoch": 0.9524625000973013,
      "grad_norm": 2.436511278152466,
      "learning_rate": 5.004834401271693e-08,
      "loss": 0.8319,
      "step": 61180
    },
    {
      "epoch": 0.9525403410992707,
      "grad_norm": 9.51063346862793,
      "learning_rate": 4.996640500811196e-08,
      "loss": 0.8207,
      "step": 61185
    },
    {
      "epoch": 0.95261818210124,
      "grad_norm": 3.492804765701294,
      "learning_rate": 4.988446600350699e-08,
      "loss": 0.7454,
      "step": 61190
    },
    {
      "epoch": 0.9526960231032093,
      "grad_norm": 4.95271110534668,
      "learning_rate": 4.980252699890201e-08,
      "loss": 0.7273,
      "step": 61195
    },
    {
      "epoch": 0.9527738641051787,
      "grad_norm": 4.087480545043945,
      "learning_rate": 4.972058799429704e-08,
      "loss": 0.8595,
      "step": 61200
    },
    {
      "epoch": 0.9528517051071481,
      "grad_norm": 3.1123194694519043,
      "learning_rate": 4.963864898969207e-08,
      "loss": 0.6916,
      "step": 61205
    },
    {
      "epoch": 0.9529295461091175,
      "grad_norm": 4.7453155517578125,
      "learning_rate": 4.95567099850871e-08,
      "loss": 0.7746,
      "step": 61210
    },
    {
      "epoch": 0.9530073871110869,
      "grad_norm": 2.608412504196167,
      "learning_rate": 4.9474770980482125e-08,
      "loss": 0.5922,
      "step": 61215
    },
    {
      "epoch": 0.9530852281130563,
      "grad_norm": 4.973897933959961,
      "learning_rate": 4.9392831975877154e-08,
      "loss": 0.7084,
      "step": 61220
    },
    {
      "epoch": 0.9531630691150257,
      "grad_norm": 4.938835620880127,
      "learning_rate": 4.931089297127218e-08,
      "loss": 0.6993,
      "step": 61225
    },
    {
      "epoch": 0.9532409101169951,
      "grad_norm": 2.666975736618042,
      "learning_rate": 4.922895396666721e-08,
      "loss": 0.7336,
      "step": 61230
    },
    {
      "epoch": 0.9533187511189644,
      "grad_norm": 11.152981758117676,
      "learning_rate": 4.914701496206223e-08,
      "loss": 0.7591,
      "step": 61235
    },
    {
      "epoch": 0.9533965921209338,
      "grad_norm": 3.224479913711548,
      "learning_rate": 4.906507595745726e-08,
      "loss": 0.7609,
      "step": 61240
    },
    {
      "epoch": 0.9534744331229031,
      "grad_norm": 4.709319114685059,
      "learning_rate": 4.898313695285229e-08,
      "loss": 0.8188,
      "step": 61245
    },
    {
      "epoch": 0.9535522741248725,
      "grad_norm": 3.092163562774658,
      "learning_rate": 4.890119794824732e-08,
      "loss": 0.6907,
      "step": 61250
    },
    {
      "epoch": 0.9536301151268419,
      "grad_norm": 3.2764902114868164,
      "learning_rate": 4.8819258943642354e-08,
      "loss": 0.7669,
      "step": 61255
    },
    {
      "epoch": 0.9537079561288113,
      "grad_norm": 5.780247688293457,
      "learning_rate": 4.873731993903738e-08,
      "loss": 0.7977,
      "step": 61260
    },
    {
      "epoch": 0.9537857971307807,
      "grad_norm": 3.154219150543213,
      "learning_rate": 4.865538093443241e-08,
      "loss": 0.7132,
      "step": 61265
    },
    {
      "epoch": 0.95386363813275,
      "grad_norm": 3.8740768432617188,
      "learning_rate": 4.857344192982744e-08,
      "loss": 0.7838,
      "step": 61270
    },
    {
      "epoch": 0.9539414791347194,
      "grad_norm": 3.6531760692596436,
      "learning_rate": 4.849150292522247e-08,
      "loss": 0.828,
      "step": 61275
    },
    {
      "epoch": 0.9540193201366888,
      "grad_norm": 4.70329475402832,
      "learning_rate": 4.840956392061749e-08,
      "loss": 0.7336,
      "step": 61280
    },
    {
      "epoch": 0.9540971611386582,
      "grad_norm": 4.0487060546875,
      "learning_rate": 4.832762491601252e-08,
      "loss": 0.6905,
      "step": 61285
    },
    {
      "epoch": 0.9541750021406276,
      "grad_norm": 12.371079444885254,
      "learning_rate": 4.824568591140755e-08,
      "loss": 0.7009,
      "step": 61290
    },
    {
      "epoch": 0.954252843142597,
      "grad_norm": 2.5091800689697266,
      "learning_rate": 4.8163746906802575e-08,
      "loss": 0.6161,
      "step": 61295
    },
    {
      "epoch": 0.9543306841445663,
      "grad_norm": 4.577026844024658,
      "learning_rate": 4.8081807902197604e-08,
      "loss": 0.7379,
      "step": 61300
    },
    {
      "epoch": 0.9544085251465357,
      "grad_norm": 3.498088836669922,
      "learning_rate": 4.799986889759263e-08,
      "loss": 0.701,
      "step": 61305
    },
    {
      "epoch": 0.954486366148505,
      "grad_norm": 3.3878448009490967,
      "learning_rate": 4.791792989298766e-08,
      "loss": 0.6598,
      "step": 61310
    },
    {
      "epoch": 0.9545642071504744,
      "grad_norm": 7.524173736572266,
      "learning_rate": 4.783599088838269e-08,
      "loss": 0.7085,
      "step": 61315
    },
    {
      "epoch": 0.9546420481524438,
      "grad_norm": 4.093320369720459,
      "learning_rate": 4.775405188377771e-08,
      "loss": 0.7049,
      "step": 61320
    },
    {
      "epoch": 0.9547198891544132,
      "grad_norm": 3.133089065551758,
      "learning_rate": 4.767211287917274e-08,
      "loss": 0.7256,
      "step": 61325
    },
    {
      "epoch": 0.9547977301563826,
      "grad_norm": 5.202256679534912,
      "learning_rate": 4.759017387456777e-08,
      "loss": 0.799,
      "step": 61330
    },
    {
      "epoch": 0.954875571158352,
      "grad_norm": 3.1335256099700928,
      "learning_rate": 4.75082348699628e-08,
      "loss": 0.7494,
      "step": 61335
    },
    {
      "epoch": 0.9549534121603214,
      "grad_norm": 6.685616493225098,
      "learning_rate": 4.7426295865357825e-08,
      "loss": 0.6356,
      "step": 61340
    },
    {
      "epoch": 0.9550312531622906,
      "grad_norm": 4.052431583404541,
      "learning_rate": 4.7344356860752854e-08,
      "loss": 0.712,
      "step": 61345
    },
    {
      "epoch": 0.95510909416426,
      "grad_norm": 4.59567928314209,
      "learning_rate": 4.726241785614788e-08,
      "loss": 0.7153,
      "step": 61350
    },
    {
      "epoch": 0.9551869351662294,
      "grad_norm": 4.354411602020264,
      "learning_rate": 4.718047885154291e-08,
      "loss": 0.8799,
      "step": 61355
    },
    {
      "epoch": 0.9552647761681988,
      "grad_norm": 4.16891622543335,
      "learning_rate": 4.709853984693793e-08,
      "loss": 0.6306,
      "step": 61360
    },
    {
      "epoch": 0.9553426171701682,
      "grad_norm": 4.394661903381348,
      "learning_rate": 4.701660084233296e-08,
      "loss": 0.6709,
      "step": 61365
    },
    {
      "epoch": 0.9554204581721376,
      "grad_norm": 12.73591136932373,
      "learning_rate": 4.693466183772799e-08,
      "loss": 0.7084,
      "step": 61370
    },
    {
      "epoch": 0.955498299174107,
      "grad_norm": 3.2436838150024414,
      "learning_rate": 4.685272283312302e-08,
      "loss": 0.6576,
      "step": 61375
    },
    {
      "epoch": 0.9555761401760764,
      "grad_norm": 2.8519794940948486,
      "learning_rate": 4.677078382851805e-08,
      "loss": 0.7072,
      "step": 61380
    },
    {
      "epoch": 0.9556539811780457,
      "grad_norm": 4.175844192504883,
      "learning_rate": 4.6688844823913076e-08,
      "loss": 0.6792,
      "step": 61385
    },
    {
      "epoch": 0.9557318221800151,
      "grad_norm": 2.706840991973877,
      "learning_rate": 4.6606905819308104e-08,
      "loss": 0.7427,
      "step": 61390
    },
    {
      "epoch": 0.9558096631819845,
      "grad_norm": 3.7760231494903564,
      "learning_rate": 4.652496681470313e-08,
      "loss": 0.7528,
      "step": 61395
    },
    {
      "epoch": 0.9558875041839539,
      "grad_norm": 4.803181171417236,
      "learning_rate": 4.6443027810098155e-08,
      "loss": 0.699,
      "step": 61400
    },
    {
      "epoch": 0.9559653451859232,
      "grad_norm": 2.631591320037842,
      "learning_rate": 4.636108880549318e-08,
      "loss": 0.6188,
      "step": 61405
    },
    {
      "epoch": 0.9560431861878926,
      "grad_norm": 3.3875732421875,
      "learning_rate": 4.627914980088821e-08,
      "loss": 0.8263,
      "step": 61410
    },
    {
      "epoch": 0.956121027189862,
      "grad_norm": 4.971041202545166,
      "learning_rate": 4.619721079628325e-08,
      "loss": 0.6858,
      "step": 61415
    },
    {
      "epoch": 0.9561988681918314,
      "grad_norm": 4.145664691925049,
      "learning_rate": 4.6115271791678275e-08,
      "loss": 0.6332,
      "step": 61420
    },
    {
      "epoch": 0.9562767091938007,
      "grad_norm": 4.822089195251465,
      "learning_rate": 4.6033332787073304e-08,
      "loss": 0.6904,
      "step": 61425
    },
    {
      "epoch": 0.9563545501957701,
      "grad_norm": 4.725489616394043,
      "learning_rate": 4.595139378246833e-08,
      "loss": 0.623,
      "step": 61430
    },
    {
      "epoch": 0.9564323911977395,
      "grad_norm": 3.46065092086792,
      "learning_rate": 4.586945477786336e-08,
      "loss": 0.6637,
      "step": 61435
    },
    {
      "epoch": 0.9565102321997089,
      "grad_norm": 4.897166728973389,
      "learning_rate": 4.578751577325839e-08,
      "loss": 0.7719,
      "step": 61440
    },
    {
      "epoch": 0.9565880732016783,
      "grad_norm": 6.850200653076172,
      "learning_rate": 4.570557676865341e-08,
      "loss": 0.6218,
      "step": 61445
    },
    {
      "epoch": 0.9566659142036477,
      "grad_norm": 3.1212306022644043,
      "learning_rate": 4.562363776404844e-08,
      "loss": 0.7035,
      "step": 61450
    },
    {
      "epoch": 0.956743755205617,
      "grad_norm": 4.773056507110596,
      "learning_rate": 4.554169875944347e-08,
      "loss": 0.8585,
      "step": 61455
    },
    {
      "epoch": 0.9568215962075863,
      "grad_norm": 6.659379959106445,
      "learning_rate": 4.54597597548385e-08,
      "loss": 0.7399,
      "step": 61460
    },
    {
      "epoch": 0.9568994372095557,
      "grad_norm": 3.8375802040100098,
      "learning_rate": 4.5377820750233526e-08,
      "loss": 0.751,
      "step": 61465
    },
    {
      "epoch": 0.9569772782115251,
      "grad_norm": 4.342898368835449,
      "learning_rate": 4.5295881745628554e-08,
      "loss": 0.9028,
      "step": 61470
    },
    {
      "epoch": 0.9570551192134945,
      "grad_norm": 5.114435195922852,
      "learning_rate": 4.521394274102358e-08,
      "loss": 0.8256,
      "step": 61475
    },
    {
      "epoch": 0.9571329602154639,
      "grad_norm": 3.834040880203247,
      "learning_rate": 4.513200373641861e-08,
      "loss": 0.6393,
      "step": 61480
    },
    {
      "epoch": 0.9572108012174333,
      "grad_norm": 3.1106698513031006,
      "learning_rate": 4.505006473181363e-08,
      "loss": 0.7319,
      "step": 61485
    },
    {
      "epoch": 0.9572886422194027,
      "grad_norm": 3.632053852081299,
      "learning_rate": 4.496812572720866e-08,
      "loss": 0.6816,
      "step": 61490
    },
    {
      "epoch": 0.9573664832213721,
      "grad_norm": 3.067706823348999,
      "learning_rate": 4.488618672260369e-08,
      "loss": 0.729,
      "step": 61495
    },
    {
      "epoch": 0.9574443242233414,
      "grad_norm": 5.820592880249023,
      "learning_rate": 4.480424771799872e-08,
      "loss": 0.7787,
      "step": 61500
    },
    {
      "epoch": 0.9575221652253108,
      "grad_norm": 4.328648090362549,
      "learning_rate": 4.472230871339375e-08,
      "loss": 0.8145,
      "step": 61505
    },
    {
      "epoch": 0.9576000062272801,
      "grad_norm": 4.0816216468811035,
      "learning_rate": 4.4640369708788776e-08,
      "loss": 0.8837,
      "step": 61510
    },
    {
      "epoch": 0.9576778472292495,
      "grad_norm": 4.673227787017822,
      "learning_rate": 4.4558430704183804e-08,
      "loss": 0.8977,
      "step": 61515
    },
    {
      "epoch": 0.9577556882312189,
      "grad_norm": 4.654181957244873,
      "learning_rate": 4.447649169957883e-08,
      "loss": 0.7998,
      "step": 61520
    },
    {
      "epoch": 0.9578335292331883,
      "grad_norm": 3.2629480361938477,
      "learning_rate": 4.4394552694973855e-08,
      "loss": 0.6891,
      "step": 61525
    },
    {
      "epoch": 0.9579113702351577,
      "grad_norm": 3.2328968048095703,
      "learning_rate": 4.4312613690368883e-08,
      "loss": 0.7459,
      "step": 61530
    },
    {
      "epoch": 0.957989211237127,
      "grad_norm": 5.624979019165039,
      "learning_rate": 4.423067468576391e-08,
      "loss": 0.8867,
      "step": 61535
    },
    {
      "epoch": 0.9580670522390964,
      "grad_norm": 3.482919454574585,
      "learning_rate": 4.414873568115894e-08,
      "loss": 0.6971,
      "step": 61540
    },
    {
      "epoch": 0.9581448932410658,
      "grad_norm": 6.630836486816406,
      "learning_rate": 4.406679667655397e-08,
      "loss": 0.6839,
      "step": 61545
    },
    {
      "epoch": 0.9582227342430352,
      "grad_norm": 7.060342311859131,
      "learning_rate": 4.3984857671949e-08,
      "loss": 0.7147,
      "step": 61550
    },
    {
      "epoch": 0.9583005752450046,
      "grad_norm": 3.237863302230835,
      "learning_rate": 4.3902918667344026e-08,
      "loss": 0.6646,
      "step": 61555
    },
    {
      "epoch": 0.958378416246974,
      "grad_norm": 3.4414734840393066,
      "learning_rate": 4.3820979662739055e-08,
      "loss": 0.7103,
      "step": 61560
    },
    {
      "epoch": 0.9584562572489433,
      "grad_norm": 2.8609018325805664,
      "learning_rate": 4.3739040658134077e-08,
      "loss": 0.6866,
      "step": 61565
    },
    {
      "epoch": 0.9585340982509127,
      "grad_norm": 4.016711235046387,
      "learning_rate": 4.3657101653529105e-08,
      "loss": 0.7242,
      "step": 61570
    },
    {
      "epoch": 0.958611939252882,
      "grad_norm": 5.5695366859436035,
      "learning_rate": 4.357516264892414e-08,
      "loss": 0.7672,
      "step": 61575
    },
    {
      "epoch": 0.9586897802548514,
      "grad_norm": 2.902132987976074,
      "learning_rate": 4.349322364431917e-08,
      "loss": 0.782,
      "step": 61580
    },
    {
      "epoch": 0.9587676212568208,
      "grad_norm": 3.566563367843628,
      "learning_rate": 4.34112846397142e-08,
      "loss": 0.6772,
      "step": 61585
    },
    {
      "epoch": 0.9588454622587902,
      "grad_norm": 4.549349784851074,
      "learning_rate": 4.3329345635109226e-08,
      "loss": 0.5879,
      "step": 61590
    },
    {
      "epoch": 0.9589233032607596,
      "grad_norm": 2.7438368797302246,
      "learning_rate": 4.3247406630504254e-08,
      "loss": 0.7079,
      "step": 61595
    },
    {
      "epoch": 0.959001144262729,
      "grad_norm": 3.8711488246917725,
      "learning_rate": 4.316546762589928e-08,
      "loss": 0.8173,
      "step": 61600
    },
    {
      "epoch": 0.9590789852646984,
      "grad_norm": 2.7590856552124023,
      "learning_rate": 4.308352862129431e-08,
      "loss": 0.7444,
      "step": 61605
    },
    {
      "epoch": 0.9591568262666677,
      "grad_norm": 7.705221176147461,
      "learning_rate": 4.3001589616689333e-08,
      "loss": 0.7881,
      "step": 61610
    },
    {
      "epoch": 0.959234667268637,
      "grad_norm": 5.835484981536865,
      "learning_rate": 4.291965061208436e-08,
      "loss": 0.8049,
      "step": 61615
    },
    {
      "epoch": 0.9593125082706064,
      "grad_norm": 5.138251304626465,
      "learning_rate": 4.283771160747939e-08,
      "loss": 0.728,
      "step": 61620
    },
    {
      "epoch": 0.9593903492725758,
      "grad_norm": 5.027362823486328,
      "learning_rate": 4.275577260287442e-08,
      "loss": 0.659,
      "step": 61625
    },
    {
      "epoch": 0.9594681902745452,
      "grad_norm": 4.935187816619873,
      "learning_rate": 4.267383359826945e-08,
      "loss": 0.8178,
      "step": 61630
    },
    {
      "epoch": 0.9595460312765146,
      "grad_norm": 4.373867988586426,
      "learning_rate": 4.2591894593664476e-08,
      "loss": 0.6977,
      "step": 61635
    },
    {
      "epoch": 0.959623872278484,
      "grad_norm": 7.0862603187561035,
      "learning_rate": 4.2509955589059505e-08,
      "loss": 0.7107,
      "step": 61640
    },
    {
      "epoch": 0.9597017132804534,
      "grad_norm": 4.651089668273926,
      "learning_rate": 4.242801658445453e-08,
      "loss": 0.6052,
      "step": 61645
    },
    {
      "epoch": 0.9597795542824227,
      "grad_norm": 4.010319709777832,
      "learning_rate": 4.2346077579849555e-08,
      "loss": 0.8129,
      "step": 61650
    },
    {
      "epoch": 0.9598573952843921,
      "grad_norm": 8.611007690429688,
      "learning_rate": 4.2264138575244584e-08,
      "loss": 0.6241,
      "step": 61655
    },
    {
      "epoch": 0.9599352362863615,
      "grad_norm": 3.452972888946533,
      "learning_rate": 4.218219957063961e-08,
      "loss": 0.694,
      "step": 61660
    },
    {
      "epoch": 0.9600130772883309,
      "grad_norm": 5.004734039306641,
      "learning_rate": 4.210026056603464e-08,
      "loss": 0.7527,
      "step": 61665
    },
    {
      "epoch": 0.9600909182903002,
      "grad_norm": 5.204343318939209,
      "learning_rate": 4.201832156142967e-08,
      "loss": 0.6932,
      "step": 61670
    },
    {
      "epoch": 0.9601687592922696,
      "grad_norm": 6.440560340881348,
      "learning_rate": 4.19363825568247e-08,
      "loss": 0.7848,
      "step": 61675
    },
    {
      "epoch": 0.960246600294239,
      "grad_norm": 3.8833181858062744,
      "learning_rate": 4.1854443552219726e-08,
      "loss": 0.6834,
      "step": 61680
    },
    {
      "epoch": 0.9603244412962084,
      "grad_norm": 2.8646631240844727,
      "learning_rate": 4.1772504547614755e-08,
      "loss": 0.7226,
      "step": 61685
    },
    {
      "epoch": 0.9604022822981777,
      "grad_norm": 3.3792688846588135,
      "learning_rate": 4.169056554300978e-08,
      "loss": 0.756,
      "step": 61690
    },
    {
      "epoch": 0.9604801233001471,
      "grad_norm": 4.293079376220703,
      "learning_rate": 4.1608626538404805e-08,
      "loss": 0.7492,
      "step": 61695
    },
    {
      "epoch": 0.9605579643021165,
      "grad_norm": 5.779025077819824,
      "learning_rate": 4.1526687533799834e-08,
      "loss": 0.7943,
      "step": 61700
    },
    {
      "epoch": 0.9606358053040859,
      "grad_norm": 3.3563575744628906,
      "learning_rate": 4.144474852919486e-08,
      "loss": 0.7656,
      "step": 61705
    },
    {
      "epoch": 0.9607136463060553,
      "grad_norm": 5.041967868804932,
      "learning_rate": 4.136280952458989e-08,
      "loss": 0.8327,
      "step": 61710
    },
    {
      "epoch": 0.9607914873080247,
      "grad_norm": 3.154578447341919,
      "learning_rate": 4.128087051998492e-08,
      "loss": 0.7495,
      "step": 61715
    },
    {
      "epoch": 0.960869328309994,
      "grad_norm": 4.2200846672058105,
      "learning_rate": 4.119893151537995e-08,
      "loss": 0.9194,
      "step": 61720
    },
    {
      "epoch": 0.9609471693119633,
      "grad_norm": 4.256812572479248,
      "learning_rate": 4.1116992510774977e-08,
      "loss": 0.7627,
      "step": 61725
    },
    {
      "epoch": 0.9610250103139327,
      "grad_norm": 3.4453728199005127,
      "learning_rate": 4.1035053506170005e-08,
      "loss": 0.6221,
      "step": 61730
    },
    {
      "epoch": 0.9611028513159021,
      "grad_norm": 4.984009265899658,
      "learning_rate": 4.0953114501565034e-08,
      "loss": 0.8075,
      "step": 61735
    },
    {
      "epoch": 0.9611806923178715,
      "grad_norm": 2.7461800575256348,
      "learning_rate": 4.087117549696006e-08,
      "loss": 0.7107,
      "step": 61740
    },
    {
      "epoch": 0.9612585333198409,
      "grad_norm": 3.6694533824920654,
      "learning_rate": 4.078923649235509e-08,
      "loss": 0.7475,
      "step": 61745
    },
    {
      "epoch": 0.9613363743218103,
      "grad_norm": 4.434169769287109,
      "learning_rate": 4.070729748775012e-08,
      "loss": 0.7367,
      "step": 61750
    },
    {
      "epoch": 0.9614142153237797,
      "grad_norm": 5.576082229614258,
      "learning_rate": 4.062535848314515e-08,
      "loss": 0.6696,
      "step": 61755
    },
    {
      "epoch": 0.9614920563257491,
      "grad_norm": 3.133444309234619,
      "learning_rate": 4.0543419478540176e-08,
      "loss": 0.6913,
      "step": 61760
    },
    {
      "epoch": 0.9615698973277184,
      "grad_norm": 8.602654457092285,
      "learning_rate": 4.0461480473935205e-08,
      "loss": 0.7875,
      "step": 61765
    },
    {
      "epoch": 0.9616477383296878,
      "grad_norm": 3.1677820682525635,
      "learning_rate": 4.0379541469330233e-08,
      "loss": 0.7582,
      "step": 61770
    },
    {
      "epoch": 0.9617255793316571,
      "grad_norm": 2.988821268081665,
      "learning_rate": 4.0297602464725255e-08,
      "loss": 0.7663,
      "step": 61775
    },
    {
      "epoch": 0.9618034203336265,
      "grad_norm": 3.830688953399658,
      "learning_rate": 4.0215663460120284e-08,
      "loss": 0.6507,
      "step": 61780
    },
    {
      "epoch": 0.9618812613355959,
      "grad_norm": 3.139862298965454,
      "learning_rate": 4.013372445551531e-08,
      "loss": 0.7029,
      "step": 61785
    },
    {
      "epoch": 0.9619591023375653,
      "grad_norm": 2.6087141036987305,
      "learning_rate": 4.005178545091034e-08,
      "loss": 0.6827,
      "step": 61790
    },
    {
      "epoch": 0.9620369433395347,
      "grad_norm": 4.4289727210998535,
      "learning_rate": 3.996984644630537e-08,
      "loss": 0.7391,
      "step": 61795
    },
    {
      "epoch": 0.962114784341504,
      "grad_norm": 9.220122337341309,
      "learning_rate": 3.98879074417004e-08,
      "loss": 0.6268,
      "step": 61800
    },
    {
      "epoch": 0.9621926253434734,
      "grad_norm": 4.405559539794922,
      "learning_rate": 3.9805968437095427e-08,
      "loss": 0.7414,
      "step": 61805
    },
    {
      "epoch": 0.9622704663454428,
      "grad_norm": 3.208700656890869,
      "learning_rate": 3.9724029432490455e-08,
      "loss": 0.6486,
      "step": 61810
    },
    {
      "epoch": 0.9623483073474122,
      "grad_norm": 13.474762916564941,
      "learning_rate": 3.964209042788548e-08,
      "loss": 0.8076,
      "step": 61815
    },
    {
      "epoch": 0.9624261483493816,
      "grad_norm": 3.247781276702881,
      "learning_rate": 3.9560151423280506e-08,
      "loss": 0.7117,
      "step": 61820
    },
    {
      "epoch": 0.962503989351351,
      "grad_norm": 3.0118322372436523,
      "learning_rate": 3.9478212418675534e-08,
      "loss": 0.8493,
      "step": 61825
    },
    {
      "epoch": 0.9625818303533203,
      "grad_norm": 4.207756996154785,
      "learning_rate": 3.939627341407056e-08,
      "loss": 0.7578,
      "step": 61830
    },
    {
      "epoch": 0.9626596713552897,
      "grad_norm": 5.650848388671875,
      "learning_rate": 3.931433440946559e-08,
      "loss": 0.7752,
      "step": 61835
    },
    {
      "epoch": 0.962737512357259,
      "grad_norm": 3.188511848449707,
      "learning_rate": 3.923239540486062e-08,
      "loss": 0.7862,
      "step": 61840
    },
    {
      "epoch": 0.9628153533592284,
      "grad_norm": 8.89487361907959,
      "learning_rate": 3.915045640025565e-08,
      "loss": 0.7285,
      "step": 61845
    },
    {
      "epoch": 0.9628931943611978,
      "grad_norm": 3.72304368019104,
      "learning_rate": 3.906851739565068e-08,
      "loss": 0.6593,
      "step": 61850
    },
    {
      "epoch": 0.9629710353631672,
      "grad_norm": 2.9224071502685547,
      "learning_rate": 3.89865783910457e-08,
      "loss": 0.7366,
      "step": 61855
    },
    {
      "epoch": 0.9630488763651366,
      "grad_norm": 3.330160617828369,
      "learning_rate": 3.890463938644073e-08,
      "loss": 0.6952,
      "step": 61860
    },
    {
      "epoch": 0.963126717367106,
      "grad_norm": 3.11863112449646,
      "learning_rate": 3.8822700381835756e-08,
      "loss": 0.8023,
      "step": 61865
    },
    {
      "epoch": 0.9632045583690754,
      "grad_norm": 4.045502185821533,
      "learning_rate": 3.8740761377230784e-08,
      "loss": 0.7134,
      "step": 61870
    },
    {
      "epoch": 0.9632823993710447,
      "grad_norm": 4.102147102355957,
      "learning_rate": 3.865882237262581e-08,
      "loss": 0.7558,
      "step": 61875
    },
    {
      "epoch": 0.963360240373014,
      "grad_norm": 4.135412216186523,
      "learning_rate": 3.857688336802084e-08,
      "loss": 0.8516,
      "step": 61880
    },
    {
      "epoch": 0.9634380813749834,
      "grad_norm": 6.7324910163879395,
      "learning_rate": 3.849494436341587e-08,
      "loss": 0.8277,
      "step": 61885
    },
    {
      "epoch": 0.9635159223769528,
      "grad_norm": 5.037081718444824,
      "learning_rate": 3.84130053588109e-08,
      "loss": 0.6593,
      "step": 61890
    },
    {
      "epoch": 0.9635937633789222,
      "grad_norm": 3.892110586166382,
      "learning_rate": 3.8331066354205934e-08,
      "loss": 0.814,
      "step": 61895
    },
    {
      "epoch": 0.9636716043808916,
      "grad_norm": 3.2788479328155518,
      "learning_rate": 3.8249127349600955e-08,
      "loss": 0.649,
      "step": 61900
    },
    {
      "epoch": 0.963749445382861,
      "grad_norm": 3.134877920150757,
      "learning_rate": 3.8167188344995984e-08,
      "loss": 0.6909,
      "step": 61905
    },
    {
      "epoch": 0.9638272863848304,
      "grad_norm": 6.512457847595215,
      "learning_rate": 3.808524934039101e-08,
      "loss": 0.8042,
      "step": 61910
    },
    {
      "epoch": 0.9639051273867997,
      "grad_norm": 3.3637704849243164,
      "learning_rate": 3.800331033578604e-08,
      "loss": 0.7911,
      "step": 61915
    },
    {
      "epoch": 0.9639829683887691,
      "grad_norm": 3.298383951187134,
      "learning_rate": 3.792137133118107e-08,
      "loss": 0.7,
      "step": 61920
    },
    {
      "epoch": 0.9640608093907385,
      "grad_norm": 2.9223666191101074,
      "learning_rate": 3.78394323265761e-08,
      "loss": 0.6746,
      "step": 61925
    },
    {
      "epoch": 0.9641386503927079,
      "grad_norm": 3.005000591278076,
      "learning_rate": 3.775749332197113e-08,
      "loss": 0.6692,
      "step": 61930
    },
    {
      "epoch": 0.9642164913946772,
      "grad_norm": 4.368032455444336,
      "learning_rate": 3.7675554317366155e-08,
      "loss": 0.7438,
      "step": 61935
    },
    {
      "epoch": 0.9642943323966466,
      "grad_norm": 4.566869258880615,
      "learning_rate": 3.759361531276118e-08,
      "loss": 0.7503,
      "step": 61940
    },
    {
      "epoch": 0.964372173398616,
      "grad_norm": 6.085843563079834,
      "learning_rate": 3.7511676308156206e-08,
      "loss": 0.776,
      "step": 61945
    },
    {
      "epoch": 0.9644500144005853,
      "grad_norm": 4.022487640380859,
      "learning_rate": 3.7429737303551234e-08,
      "loss": 0.8182,
      "step": 61950
    },
    {
      "epoch": 0.9645278554025547,
      "grad_norm": 2.469804525375366,
      "learning_rate": 3.734779829894626e-08,
      "loss": 0.7241,
      "step": 61955
    },
    {
      "epoch": 0.9646056964045241,
      "grad_norm": 2.8515594005584717,
      "learning_rate": 3.726585929434129e-08,
      "loss": 0.6428,
      "step": 61960
    },
    {
      "epoch": 0.9646835374064935,
      "grad_norm": 4.364387035369873,
      "learning_rate": 3.718392028973632e-08,
      "loss": 0.6732,
      "step": 61965
    },
    {
      "epoch": 0.9647613784084629,
      "grad_norm": 7.198433876037598,
      "learning_rate": 3.710198128513135e-08,
      "loss": 0.754,
      "step": 61970
    },
    {
      "epoch": 0.9648392194104323,
      "grad_norm": 3.843294143676758,
      "learning_rate": 3.702004228052638e-08,
      "loss": 0.7763,
      "step": 61975
    },
    {
      "epoch": 0.9649170604124017,
      "grad_norm": 7.770458698272705,
      "learning_rate": 3.69381032759214e-08,
      "loss": 0.7773,
      "step": 61980
    },
    {
      "epoch": 0.964994901414371,
      "grad_norm": 6.367640495300293,
      "learning_rate": 3.685616427131643e-08,
      "loss": 0.7654,
      "step": 61985
    },
    {
      "epoch": 0.9650727424163403,
      "grad_norm": 4.705018520355225,
      "learning_rate": 3.6774225266711456e-08,
      "loss": 0.7736,
      "step": 61990
    },
    {
      "epoch": 0.9651505834183097,
      "grad_norm": 2.804293155670166,
      "learning_rate": 3.6692286262106484e-08,
      "loss": 0.7421,
      "step": 61995
    },
    {
      "epoch": 0.9652284244202791,
      "grad_norm": 4.38629674911499,
      "learning_rate": 3.661034725750151e-08,
      "loss": 0.7099,
      "step": 62000
    },
    {
      "epoch": 0.9653062654222485,
      "grad_norm": 3.168816566467285,
      "learning_rate": 3.652840825289654e-08,
      "loss": 0.7339,
      "step": 62005
    },
    {
      "epoch": 0.9653841064242179,
      "grad_norm": 2.621918201446533,
      "learning_rate": 3.644646924829157e-08,
      "loss": 0.6453,
      "step": 62010
    },
    {
      "epoch": 0.9654619474261873,
      "grad_norm": 2.912069797515869,
      "learning_rate": 3.63645302436866e-08,
      "loss": 0.7768,
      "step": 62015
    },
    {
      "epoch": 0.9655397884281567,
      "grad_norm": 3.6733758449554443,
      "learning_rate": 3.628259123908163e-08,
      "loss": 0.7805,
      "step": 62020
    },
    {
      "epoch": 0.9656176294301261,
      "grad_norm": 4.522243022918701,
      "learning_rate": 3.620065223447665e-08,
      "loss": 0.7723,
      "step": 62025
    },
    {
      "epoch": 0.9656954704320954,
      "grad_norm": 3.1538119316101074,
      "learning_rate": 3.611871322987168e-08,
      "loss": 0.7774,
      "step": 62030
    },
    {
      "epoch": 0.9657733114340648,
      "grad_norm": 5.245665073394775,
      "learning_rate": 3.6036774225266706e-08,
      "loss": 0.7716,
      "step": 62035
    },
    {
      "epoch": 0.9658511524360341,
      "grad_norm": 3.180101156234741,
      "learning_rate": 3.5954835220661735e-08,
      "loss": 0.6424,
      "step": 62040
    },
    {
      "epoch": 0.9659289934380035,
      "grad_norm": 4.442811012268066,
      "learning_rate": 3.587289621605676e-08,
      "loss": 0.7316,
      "step": 62045
    },
    {
      "epoch": 0.9660068344399729,
      "grad_norm": 3.1652541160583496,
      "learning_rate": 3.579095721145179e-08,
      "loss": 0.5954,
      "step": 62050
    },
    {
      "epoch": 0.9660846754419423,
      "grad_norm": 5.018191814422607,
      "learning_rate": 3.570901820684683e-08,
      "loss": 0.777,
      "step": 62055
    },
    {
      "epoch": 0.9661625164439117,
      "grad_norm": 2.988568067550659,
      "learning_rate": 3.5627079202241855e-08,
      "loss": 0.6197,
      "step": 62060
    },
    {
      "epoch": 0.966240357445881,
      "grad_norm": 6.668415069580078,
      "learning_rate": 3.554514019763688e-08,
      "loss": 0.7587,
      "step": 62065
    },
    {
      "epoch": 0.9663181984478504,
      "grad_norm": 3.4335532188415527,
      "learning_rate": 3.5463201193031906e-08,
      "loss": 0.6669,
      "step": 62070
    },
    {
      "epoch": 0.9663960394498198,
      "grad_norm": 4.056186199188232,
      "learning_rate": 3.5381262188426934e-08,
      "loss": 0.674,
      "step": 62075
    },
    {
      "epoch": 0.9664738804517892,
      "grad_norm": 5.871220588684082,
      "learning_rate": 3.529932318382196e-08,
      "loss": 0.8216,
      "step": 62080
    },
    {
      "epoch": 0.9665517214537586,
      "grad_norm": 4.170464515686035,
      "learning_rate": 3.521738417921699e-08,
      "loss": 0.6614,
      "step": 62085
    },
    {
      "epoch": 0.966629562455728,
      "grad_norm": 3.3711018562316895,
      "learning_rate": 3.513544517461202e-08,
      "loss": 0.691,
      "step": 62090
    },
    {
      "epoch": 0.9667074034576973,
      "grad_norm": 2.7068448066711426,
      "learning_rate": 3.505350617000705e-08,
      "loss": 0.7321,
      "step": 62095
    },
    {
      "epoch": 0.9667852444596667,
      "grad_norm": 4.862526893615723,
      "learning_rate": 3.497156716540208e-08,
      "loss": 0.7368,
      "step": 62100
    },
    {
      "epoch": 0.966863085461636,
      "grad_norm": 3.086148500442505,
      "learning_rate": 3.48896281607971e-08,
      "loss": 0.8158,
      "step": 62105
    },
    {
      "epoch": 0.9669409264636054,
      "grad_norm": 3.653212785720825,
      "learning_rate": 3.480768915619213e-08,
      "loss": 0.6592,
      "step": 62110
    },
    {
      "epoch": 0.9670187674655748,
      "grad_norm": 3.033029317855835,
      "learning_rate": 3.4725750151587156e-08,
      "loss": 0.6534,
      "step": 62115
    },
    {
      "epoch": 0.9670966084675442,
      "grad_norm": 5.138620853424072,
      "learning_rate": 3.4643811146982185e-08,
      "loss": 0.6833,
      "step": 62120
    },
    {
      "epoch": 0.9671744494695136,
      "grad_norm": 2.9363796710968018,
      "learning_rate": 3.456187214237721e-08,
      "loss": 0.722,
      "step": 62125
    },
    {
      "epoch": 0.967252290471483,
      "grad_norm": 3.345398187637329,
      "learning_rate": 3.447993313777224e-08,
      "loss": 0.8369,
      "step": 62130
    },
    {
      "epoch": 0.9673301314734524,
      "grad_norm": 3.6381125450134277,
      "learning_rate": 3.439799413316727e-08,
      "loss": 0.7486,
      "step": 62135
    },
    {
      "epoch": 0.9674079724754217,
      "grad_norm": 6.391420364379883,
      "learning_rate": 3.43160551285623e-08,
      "loss": 0.7525,
      "step": 62140
    },
    {
      "epoch": 0.967485813477391,
      "grad_norm": 5.21929407119751,
      "learning_rate": 3.423411612395732e-08,
      "loss": 0.7747,
      "step": 62145
    },
    {
      "epoch": 0.9675636544793604,
      "grad_norm": 5.025047779083252,
      "learning_rate": 3.415217711935235e-08,
      "loss": 0.8222,
      "step": 62150
    },
    {
      "epoch": 0.9676414954813298,
      "grad_norm": 3.308523654937744,
      "learning_rate": 3.407023811474738e-08,
      "loss": 0.7455,
      "step": 62155
    },
    {
      "epoch": 0.9677193364832992,
      "grad_norm": 3.1985483169555664,
      "learning_rate": 3.3988299110142406e-08,
      "loss": 0.6889,
      "step": 62160
    },
    {
      "epoch": 0.9677971774852686,
      "grad_norm": 4.073370933532715,
      "learning_rate": 3.3906360105537435e-08,
      "loss": 0.794,
      "step": 62165
    },
    {
      "epoch": 0.967875018487238,
      "grad_norm": 3.954864978790283,
      "learning_rate": 3.3824421100932463e-08,
      "loss": 0.6446,
      "step": 62170
    },
    {
      "epoch": 0.9679528594892074,
      "grad_norm": 3.657557249069214,
      "learning_rate": 3.374248209632749e-08,
      "loss": 0.7698,
      "step": 62175
    },
    {
      "epoch": 0.9680307004911767,
      "grad_norm": 3.193450450897217,
      "learning_rate": 3.366054309172252e-08,
      "loss": 0.6746,
      "step": 62180
    },
    {
      "epoch": 0.9681085414931461,
      "grad_norm": 3.39528751373291,
      "learning_rate": 3.357860408711755e-08,
      "loss": 0.728,
      "step": 62185
    },
    {
      "epoch": 0.9681863824951155,
      "grad_norm": 3.459787368774414,
      "learning_rate": 3.349666508251257e-08,
      "loss": 0.7708,
      "step": 62190
    },
    {
      "epoch": 0.9682642234970849,
      "grad_norm": 3.312924385070801,
      "learning_rate": 3.34147260779076e-08,
      "loss": 0.7166,
      "step": 62195
    },
    {
      "epoch": 0.9683420644990542,
      "grad_norm": 3.3126676082611084,
      "learning_rate": 3.333278707330263e-08,
      "loss": 0.7106,
      "step": 62200
    },
    {
      "epoch": 0.9684199055010236,
      "grad_norm": 3.5541508197784424,
      "learning_rate": 3.3250848068697657e-08,
      "loss": 0.6991,
      "step": 62205
    },
    {
      "epoch": 0.968497746502993,
      "grad_norm": 3.609315872192383,
      "learning_rate": 3.316890906409269e-08,
      "loss": 0.6851,
      "step": 62210
    },
    {
      "epoch": 0.9685755875049623,
      "grad_norm": 5.0067830085754395,
      "learning_rate": 3.308697005948772e-08,
      "loss": 0.7383,
      "step": 62215
    },
    {
      "epoch": 0.9686534285069317,
      "grad_norm": 3.432095766067505,
      "learning_rate": 3.300503105488275e-08,
      "loss": 0.7676,
      "step": 62220
    },
    {
      "epoch": 0.9687312695089011,
      "grad_norm": 4.700979709625244,
      "learning_rate": 3.292309205027778e-08,
      "loss": 0.8092,
      "step": 62225
    },
    {
      "epoch": 0.9688091105108705,
      "grad_norm": 3.643568515777588,
      "learning_rate": 3.28411530456728e-08,
      "loss": 0.7598,
      "step": 62230
    },
    {
      "epoch": 0.9688869515128399,
      "grad_norm": 5.066553115844727,
      "learning_rate": 3.275921404106783e-08,
      "loss": 0.869,
      "step": 62235
    },
    {
      "epoch": 0.9689647925148093,
      "grad_norm": 2.724534749984741,
      "learning_rate": 3.2677275036462856e-08,
      "loss": 0.6741,
      "step": 62240
    },
    {
      "epoch": 0.9690426335167787,
      "grad_norm": 3.475245475769043,
      "learning_rate": 3.2595336031857885e-08,
      "loss": 0.7785,
      "step": 62245
    },
    {
      "epoch": 0.969120474518748,
      "grad_norm": 3.5224194526672363,
      "learning_rate": 3.2513397027252913e-08,
      "loss": 0.7695,
      "step": 62250
    },
    {
      "epoch": 0.9691983155207173,
      "grad_norm": 10.599028587341309,
      "learning_rate": 3.243145802264794e-08,
      "loss": 0.7104,
      "step": 62255
    },
    {
      "epoch": 0.9692761565226867,
      "grad_norm": 4.527596950531006,
      "learning_rate": 3.234951901804297e-08,
      "loss": 0.6923,
      "step": 62260
    },
    {
      "epoch": 0.9693539975246561,
      "grad_norm": 4.170860290527344,
      "learning_rate": 3.2267580013438e-08,
      "loss": 0.7429,
      "step": 62265
    },
    {
      "epoch": 0.9694318385266255,
      "grad_norm": 3.125988245010376,
      "learning_rate": 3.218564100883302e-08,
      "loss": 0.6693,
      "step": 62270
    },
    {
      "epoch": 0.9695096795285949,
      "grad_norm": 4.997470855712891,
      "learning_rate": 3.210370200422805e-08,
      "loss": 0.7034,
      "step": 62275
    },
    {
      "epoch": 0.9695875205305643,
      "grad_norm": 4.304283142089844,
      "learning_rate": 3.202176299962308e-08,
      "loss": 0.7218,
      "step": 62280
    },
    {
      "epoch": 0.9696653615325337,
      "grad_norm": 3.913360118865967,
      "learning_rate": 3.1939823995018107e-08,
      "loss": 0.7734,
      "step": 62285
    },
    {
      "epoch": 0.9697432025345031,
      "grad_norm": 3.887709617614746,
      "learning_rate": 3.1857884990413135e-08,
      "loss": 0.6248,
      "step": 62290
    },
    {
      "epoch": 0.9698210435364724,
      "grad_norm": 5.8655476570129395,
      "learning_rate": 3.1775945985808164e-08,
      "loss": 0.718,
      "step": 62295
    },
    {
      "epoch": 0.9698988845384418,
      "grad_norm": 5.206371307373047,
      "learning_rate": 3.169400698120319e-08,
      "loss": 0.71,
      "step": 62300
    },
    {
      "epoch": 0.9699767255404111,
      "grad_norm": 3.066145181655884,
      "learning_rate": 3.161206797659822e-08,
      "loss": 0.6808,
      "step": 62305
    },
    {
      "epoch": 0.9700545665423805,
      "grad_norm": 4.366287708282471,
      "learning_rate": 3.153012897199325e-08,
      "loss": 0.6855,
      "step": 62310
    },
    {
      "epoch": 0.9701324075443499,
      "grad_norm": 3.1865627765655518,
      "learning_rate": 3.144818996738827e-08,
      "loss": 0.6312,
      "step": 62315
    },
    {
      "epoch": 0.9702102485463193,
      "grad_norm": 2.8681423664093018,
      "learning_rate": 3.13662509627833e-08,
      "loss": 0.7275,
      "step": 62320
    },
    {
      "epoch": 0.9702880895482887,
      "grad_norm": 3.913332462310791,
      "learning_rate": 3.128431195817833e-08,
      "loss": 0.7014,
      "step": 62325
    },
    {
      "epoch": 0.970365930550258,
      "grad_norm": 5.0613884925842285,
      "learning_rate": 3.1202372953573363e-08,
      "loss": 0.7312,
      "step": 62330
    },
    {
      "epoch": 0.9704437715522274,
      "grad_norm": 3.43072772026062,
      "learning_rate": 3.1120433948968385e-08,
      "loss": 0.7299,
      "step": 62335
    },
    {
      "epoch": 0.9705216125541968,
      "grad_norm": 6.228733539581299,
      "learning_rate": 3.1038494944363414e-08,
      "loss": 0.7712,
      "step": 62340
    },
    {
      "epoch": 0.9705994535561662,
      "grad_norm": 6.17812967300415,
      "learning_rate": 3.095655593975844e-08,
      "loss": 0.6875,
      "step": 62345
    },
    {
      "epoch": 0.9706772945581356,
      "grad_norm": 7.2182817459106445,
      "learning_rate": 3.087461693515347e-08,
      "loss": 0.6669,
      "step": 62350
    },
    {
      "epoch": 0.970755135560105,
      "grad_norm": 4.578616142272949,
      "learning_rate": 3.07926779305485e-08,
      "loss": 0.702,
      "step": 62355
    },
    {
      "epoch": 0.9708329765620743,
      "grad_norm": 9.548687934875488,
      "learning_rate": 3.071073892594353e-08,
      "loss": 0.8064,
      "step": 62360
    },
    {
      "epoch": 0.9709108175640437,
      "grad_norm": 3.552658796310425,
      "learning_rate": 3.0628799921338556e-08,
      "loss": 0.7995,
      "step": 62365
    },
    {
      "epoch": 0.970988658566013,
      "grad_norm": 3.6431725025177,
      "learning_rate": 3.0546860916733585e-08,
      "loss": 0.6898,
      "step": 62370
    },
    {
      "epoch": 0.9710664995679824,
      "grad_norm": 8.500741004943848,
      "learning_rate": 3.046492191212861e-08,
      "loss": 0.7415,
      "step": 62375
    },
    {
      "epoch": 0.9711443405699518,
      "grad_norm": 4.619112491607666,
      "learning_rate": 3.0382982907523635e-08,
      "loss": 0.7064,
      "step": 62380
    },
    {
      "epoch": 0.9712221815719212,
      "grad_norm": 3.662475109100342,
      "learning_rate": 3.0301043902918664e-08,
      "loss": 0.7726,
      "step": 62385
    },
    {
      "epoch": 0.9713000225738906,
      "grad_norm": 3.5823633670806885,
      "learning_rate": 3.021910489831369e-08,
      "loss": 0.721,
      "step": 62390
    },
    {
      "epoch": 0.97137786357586,
      "grad_norm": 5.308351516723633,
      "learning_rate": 3.013716589370872e-08,
      "loss": 0.7036,
      "step": 62395
    },
    {
      "epoch": 0.9714557045778294,
      "grad_norm": 5.6073126792907715,
      "learning_rate": 3.005522688910375e-08,
      "loss": 0.6546,
      "step": 62400
    },
    {
      "epoch": 0.9715335455797987,
      "grad_norm": 2.9008114337921143,
      "learning_rate": 2.997328788449878e-08,
      "loss": 0.7271,
      "step": 62405
    },
    {
      "epoch": 0.971611386581768,
      "grad_norm": 2.7881388664245605,
      "learning_rate": 2.989134887989381e-08,
      "loss": 0.7555,
      "step": 62410
    },
    {
      "epoch": 0.9716892275837374,
      "grad_norm": 2.9144296646118164,
      "learning_rate": 2.9809409875288835e-08,
      "loss": 0.67,
      "step": 62415
    },
    {
      "epoch": 0.9717670685857068,
      "grad_norm": 3.0425186157226562,
      "learning_rate": 2.9727470870683864e-08,
      "loss": 0.7239,
      "step": 62420
    },
    {
      "epoch": 0.9718449095876762,
      "grad_norm": 4.626834869384766,
      "learning_rate": 2.964553186607889e-08,
      "loss": 0.7366,
      "step": 62425
    },
    {
      "epoch": 0.9719227505896456,
      "grad_norm": 11.01834774017334,
      "learning_rate": 2.9563592861473918e-08,
      "loss": 0.8107,
      "step": 62430
    },
    {
      "epoch": 0.972000591591615,
      "grad_norm": 6.980829238891602,
      "learning_rate": 2.9481653856868946e-08,
      "loss": 0.7616,
      "step": 62435
    },
    {
      "epoch": 0.9720784325935844,
      "grad_norm": 4.72191858291626,
      "learning_rate": 2.9399714852263975e-08,
      "loss": 0.7098,
      "step": 62440
    },
    {
      "epoch": 0.9721562735955537,
      "grad_norm": 3.994136095046997,
      "learning_rate": 2.9317775847659e-08,
      "loss": 0.7952,
      "step": 62445
    },
    {
      "epoch": 0.9722341145975231,
      "grad_norm": 7.765771389007568,
      "learning_rate": 2.923583684305403e-08,
      "loss": 0.818,
      "step": 62450
    },
    {
      "epoch": 0.9723119555994925,
      "grad_norm": 4.596036434173584,
      "learning_rate": 2.9153897838449057e-08,
      "loss": 0.7485,
      "step": 62455
    },
    {
      "epoch": 0.9723897966014619,
      "grad_norm": 6.076943874359131,
      "learning_rate": 2.9071958833844085e-08,
      "loss": 0.7826,
      "step": 62460
    },
    {
      "epoch": 0.9724676376034312,
      "grad_norm": 7.625444412231445,
      "learning_rate": 2.8990019829239114e-08,
      "loss": 0.7486,
      "step": 62465
    },
    {
      "epoch": 0.9725454786054006,
      "grad_norm": 3.8924782276153564,
      "learning_rate": 2.890808082463414e-08,
      "loss": 0.7205,
      "step": 62470
    },
    {
      "epoch": 0.97262331960737,
      "grad_norm": 5.2657551765441895,
      "learning_rate": 2.8826141820029168e-08,
      "loss": 0.6707,
      "step": 62475
    },
    {
      "epoch": 0.9727011606093393,
      "grad_norm": 4.577308177947998,
      "learning_rate": 2.8744202815424196e-08,
      "loss": 0.8697,
      "step": 62480
    },
    {
      "epoch": 0.9727790016113087,
      "grad_norm": 2.9246013164520264,
      "learning_rate": 2.8662263810819225e-08,
      "loss": 0.8095,
      "step": 62485
    },
    {
      "epoch": 0.9728568426132781,
      "grad_norm": 4.5807318687438965,
      "learning_rate": 2.8580324806214253e-08,
      "loss": 0.7362,
      "step": 62490
    },
    {
      "epoch": 0.9729346836152475,
      "grad_norm": 5.0269927978515625,
      "learning_rate": 2.8498385801609282e-08,
      "loss": 0.6921,
      "step": 62495
    },
    {
      "epoch": 0.9730125246172169,
      "grad_norm": 3.7786293029785156,
      "learning_rate": 2.841644679700431e-08,
      "loss": 0.8233,
      "step": 62500
    },
    {
      "epoch": 0.9730903656191863,
      "grad_norm": 2.7942347526550293,
      "learning_rate": 2.833450779239934e-08,
      "loss": 0.7598,
      "step": 62505
    },
    {
      "epoch": 0.9731682066211557,
      "grad_norm": 3.062075138092041,
      "learning_rate": 2.8252568787794364e-08,
      "loss": 0.7102,
      "step": 62510
    },
    {
      "epoch": 0.973246047623125,
      "grad_norm": 3.314852714538574,
      "learning_rate": 2.8170629783189393e-08,
      "loss": 0.7299,
      "step": 62515
    },
    {
      "epoch": 0.9733238886250943,
      "grad_norm": 2.7636983394622803,
      "learning_rate": 2.808869077858442e-08,
      "loss": 0.7852,
      "step": 62520
    },
    {
      "epoch": 0.9734017296270637,
      "grad_norm": 3.4994571208953857,
      "learning_rate": 2.800675177397945e-08,
      "loss": 0.8728,
      "step": 62525
    },
    {
      "epoch": 0.9734795706290331,
      "grad_norm": 2.893392324447632,
      "learning_rate": 2.7924812769374475e-08,
      "loss": 0.707,
      "step": 62530
    },
    {
      "epoch": 0.9735574116310025,
      "grad_norm": 3.860199213027954,
      "learning_rate": 2.7842873764769504e-08,
      "loss": 0.7907,
      "step": 62535
    },
    {
      "epoch": 0.9736352526329719,
      "grad_norm": 4.309029579162598,
      "learning_rate": 2.7760934760164532e-08,
      "loss": 0.6111,
      "step": 62540
    },
    {
      "epoch": 0.9737130936349413,
      "grad_norm": 3.387026786804199,
      "learning_rate": 2.767899575555956e-08,
      "loss": 0.7362,
      "step": 62545
    },
    {
      "epoch": 0.9737909346369107,
      "grad_norm": 4.3330559730529785,
      "learning_rate": 2.7597056750954586e-08,
      "loss": 0.7087,
      "step": 62550
    },
    {
      "epoch": 0.97386877563888,
      "grad_norm": 3.76174259185791,
      "learning_rate": 2.7515117746349614e-08,
      "loss": 0.8111,
      "step": 62555
    },
    {
      "epoch": 0.9739466166408494,
      "grad_norm": 4.260871887207031,
      "learning_rate": 2.7433178741744643e-08,
      "loss": 0.8427,
      "step": 62560
    },
    {
      "epoch": 0.9740244576428188,
      "grad_norm": 3.6713597774505615,
      "learning_rate": 2.735123973713967e-08,
      "loss": 0.7235,
      "step": 62565
    },
    {
      "epoch": 0.9741022986447881,
      "grad_norm": 2.8924989700317383,
      "learning_rate": 2.72693007325347e-08,
      "loss": 0.7194,
      "step": 62570
    },
    {
      "epoch": 0.9741801396467575,
      "grad_norm": 5.195179462432861,
      "learning_rate": 2.718736172792973e-08,
      "loss": 0.6406,
      "step": 62575
    },
    {
      "epoch": 0.9742579806487269,
      "grad_norm": 5.188493251800537,
      "learning_rate": 2.7105422723324757e-08,
      "loss": 0.8761,
      "step": 62580
    },
    {
      "epoch": 0.9743358216506963,
      "grad_norm": 3.376316547393799,
      "learning_rate": 2.7023483718719786e-08,
      "loss": 0.6343,
      "step": 62585
    },
    {
      "epoch": 0.9744136626526657,
      "grad_norm": 7.360108375549316,
      "learning_rate": 2.694154471411481e-08,
      "loss": 0.7854,
      "step": 62590
    },
    {
      "epoch": 0.974491503654635,
      "grad_norm": 4.635800361633301,
      "learning_rate": 2.685960570950984e-08,
      "loss": 0.9125,
      "step": 62595
    },
    {
      "epoch": 0.9745693446566044,
      "grad_norm": 4.3892011642456055,
      "learning_rate": 2.6777666704904868e-08,
      "loss": 0.8082,
      "step": 62600
    },
    {
      "epoch": 0.9746471856585738,
      "grad_norm": 3.565392255783081,
      "learning_rate": 2.6695727700299896e-08,
      "loss": 0.7464,
      "step": 62605
    },
    {
      "epoch": 0.9747250266605432,
      "grad_norm": 3.959977626800537,
      "learning_rate": 2.6613788695694925e-08,
      "loss": 0.7505,
      "step": 62610
    },
    {
      "epoch": 0.9748028676625126,
      "grad_norm": 8.617205619812012,
      "learning_rate": 2.653184969108995e-08,
      "loss": 0.7793,
      "step": 62615
    },
    {
      "epoch": 0.974880708664482,
      "grad_norm": 3.9771218299865723,
      "learning_rate": 2.644991068648498e-08,
      "loss": 0.7915,
      "step": 62620
    },
    {
      "epoch": 0.9749585496664513,
      "grad_norm": 4.495543956756592,
      "learning_rate": 2.6367971681880007e-08,
      "loss": 0.8022,
      "step": 62625
    },
    {
      "epoch": 0.9750363906684207,
      "grad_norm": 5.950969219207764,
      "learning_rate": 2.6286032677275036e-08,
      "loss": 0.7572,
      "step": 62630
    },
    {
      "epoch": 0.97511423167039,
      "grad_norm": 3.735999822616577,
      "learning_rate": 2.620409367267006e-08,
      "loss": 0.7292,
      "step": 62635
    },
    {
      "epoch": 0.9751920726723594,
      "grad_norm": 3.655789375305176,
      "learning_rate": 2.612215466806509e-08,
      "loss": 0.7624,
      "step": 62640
    },
    {
      "epoch": 0.9752699136743288,
      "grad_norm": 3.1905288696289062,
      "learning_rate": 2.6040215663460118e-08,
      "loss": 0.6844,
      "step": 62645
    },
    {
      "epoch": 0.9753477546762982,
      "grad_norm": 5.0477728843688965,
      "learning_rate": 2.595827665885515e-08,
      "loss": 0.6365,
      "step": 62650
    },
    {
      "epoch": 0.9754255956782676,
      "grad_norm": 1.9121930599212646,
      "learning_rate": 2.5876337654250175e-08,
      "loss": 0.6491,
      "step": 62655
    },
    {
      "epoch": 0.975503436680237,
      "grad_norm": 5.306177616119385,
      "learning_rate": 2.5794398649645204e-08,
      "loss": 0.7672,
      "step": 62660
    },
    {
      "epoch": 0.9755812776822064,
      "grad_norm": 2.877636671066284,
      "learning_rate": 2.5712459645040232e-08,
      "loss": 0.7018,
      "step": 62665
    },
    {
      "epoch": 0.9756591186841757,
      "grad_norm": 2.6087710857391357,
      "learning_rate": 2.563052064043526e-08,
      "loss": 0.7076,
      "step": 62670
    },
    {
      "epoch": 0.975736959686145,
      "grad_norm": 4.552338123321533,
      "learning_rate": 2.5548581635830286e-08,
      "loss": 0.7863,
      "step": 62675
    },
    {
      "epoch": 0.9758148006881144,
      "grad_norm": 4.407055854797363,
      "learning_rate": 2.5466642631225315e-08,
      "loss": 0.724,
      "step": 62680
    },
    {
      "epoch": 0.9758926416900838,
      "grad_norm": 2.7933549880981445,
      "learning_rate": 2.5384703626620343e-08,
      "loss": 0.8421,
      "step": 62685
    },
    {
      "epoch": 0.9759704826920532,
      "grad_norm": 3.782397985458374,
      "learning_rate": 2.5302764622015372e-08,
      "loss": 0.7336,
      "step": 62690
    },
    {
      "epoch": 0.9760483236940226,
      "grad_norm": 3.888249158859253,
      "learning_rate": 2.5220825617410397e-08,
      "loss": 0.7541,
      "step": 62695
    },
    {
      "epoch": 0.976126164695992,
      "grad_norm": 3.372640609741211,
      "learning_rate": 2.5138886612805425e-08,
      "loss": 0.8162,
      "step": 62700
    },
    {
      "epoch": 0.9762040056979614,
      "grad_norm": 3.182354211807251,
      "learning_rate": 2.5056947608200454e-08,
      "loss": 0.7594,
      "step": 62705
    },
    {
      "epoch": 0.9762818466999307,
      "grad_norm": 5.268570423126221,
      "learning_rate": 2.4975008603595483e-08,
      "loss": 0.8455,
      "step": 62710
    },
    {
      "epoch": 0.9763596877019001,
      "grad_norm": 7.9366374015808105,
      "learning_rate": 2.4893069598990508e-08,
      "loss": 0.7871,
      "step": 62715
    },
    {
      "epoch": 0.9764375287038695,
      "grad_norm": 5.058634281158447,
      "learning_rate": 2.4811130594385536e-08,
      "loss": 0.7341,
      "step": 62720
    },
    {
      "epoch": 0.9765153697058389,
      "grad_norm": 2.968153715133667,
      "learning_rate": 2.4729191589780565e-08,
      "loss": 0.7007,
      "step": 62725
    },
    {
      "epoch": 0.9765932107078082,
      "grad_norm": 3.1490142345428467,
      "learning_rate": 2.4647252585175597e-08,
      "loss": 0.6759,
      "step": 62730
    },
    {
      "epoch": 0.9766710517097776,
      "grad_norm": 10.289867401123047,
      "learning_rate": 2.4565313580570622e-08,
      "loss": 0.6617,
      "step": 62735
    },
    {
      "epoch": 0.976748892711747,
      "grad_norm": 5.154320240020752,
      "learning_rate": 2.448337457596565e-08,
      "loss": 0.7016,
      "step": 62740
    },
    {
      "epoch": 0.9768267337137163,
      "grad_norm": 5.366113185882568,
      "learning_rate": 2.440143557136068e-08,
      "loss": 0.7587,
      "step": 62745
    },
    {
      "epoch": 0.9769045747156857,
      "grad_norm": 2.269876003265381,
      "learning_rate": 2.4319496566755708e-08,
      "loss": 0.7054,
      "step": 62750
    },
    {
      "epoch": 0.9769824157176551,
      "grad_norm": 5.1343159675598145,
      "learning_rate": 2.4237557562150736e-08,
      "loss": 0.834,
      "step": 62755
    },
    {
      "epoch": 0.9770602567196245,
      "grad_norm": 3.2128067016601562,
      "learning_rate": 2.415561855754576e-08,
      "loss": 0.7423,
      "step": 62760
    },
    {
      "epoch": 0.9771380977215939,
      "grad_norm": 3.1787447929382324,
      "learning_rate": 2.407367955294079e-08,
      "loss": 0.8536,
      "step": 62765
    },
    {
      "epoch": 0.9772159387235633,
      "grad_norm": 4.024409294128418,
      "learning_rate": 2.399174054833582e-08,
      "loss": 0.741,
      "step": 62770
    },
    {
      "epoch": 0.9772937797255327,
      "grad_norm": 6.748385906219482,
      "learning_rate": 2.3909801543730847e-08,
      "loss": 0.6663,
      "step": 62775
    },
    {
      "epoch": 0.977371620727502,
      "grad_norm": 6.359907150268555,
      "learning_rate": 2.3827862539125872e-08,
      "loss": 0.8418,
      "step": 62780
    },
    {
      "epoch": 0.9774494617294713,
      "grad_norm": 6.067844390869141,
      "learning_rate": 2.37459235345209e-08,
      "loss": 0.6753,
      "step": 62785
    },
    {
      "epoch": 0.9775273027314407,
      "grad_norm": 3.778433322906494,
      "learning_rate": 2.366398452991593e-08,
      "loss": 0.7066,
      "step": 62790
    },
    {
      "epoch": 0.9776051437334101,
      "grad_norm": 6.258406639099121,
      "learning_rate": 2.3582045525310958e-08,
      "loss": 0.8201,
      "step": 62795
    },
    {
      "epoch": 0.9776829847353795,
      "grad_norm": 3.151217460632324,
      "learning_rate": 2.3500106520705983e-08,
      "loss": 0.6407,
      "step": 62800
    },
    {
      "epoch": 0.9777608257373489,
      "grad_norm": 4.727827548980713,
      "learning_rate": 2.341816751610101e-08,
      "loss": 0.6798,
      "step": 62805
    },
    {
      "epoch": 0.9778386667393183,
      "grad_norm": 5.343924522399902,
      "learning_rate": 2.3336228511496043e-08,
      "loss": 0.8398,
      "step": 62810
    },
    {
      "epoch": 0.9779165077412877,
      "grad_norm": 2.7041258811950684,
      "learning_rate": 2.3254289506891072e-08,
      "loss": 0.8008,
      "step": 62815
    },
    {
      "epoch": 0.977994348743257,
      "grad_norm": 5.143749713897705,
      "learning_rate": 2.3172350502286097e-08,
      "loss": 0.621,
      "step": 62820
    },
    {
      "epoch": 0.9780721897452264,
      "grad_norm": 2.2376530170440674,
      "learning_rate": 2.3090411497681126e-08,
      "loss": 0.7421,
      "step": 62825
    },
    {
      "epoch": 0.9781500307471958,
      "grad_norm": 6.662097454071045,
      "learning_rate": 2.3008472493076154e-08,
      "loss": 0.7953,
      "step": 62830
    },
    {
      "epoch": 0.9782278717491651,
      "grad_norm": 4.546566963195801,
      "learning_rate": 2.2926533488471183e-08,
      "loss": 0.7198,
      "step": 62835
    },
    {
      "epoch": 0.9783057127511345,
      "grad_norm": 4.479778289794922,
      "learning_rate": 2.2844594483866208e-08,
      "loss": 0.8075,
      "step": 62840
    },
    {
      "epoch": 0.9783835537531039,
      "grad_norm": 2.3127081394195557,
      "learning_rate": 2.2762655479261236e-08,
      "loss": 0.628,
      "step": 62845
    },
    {
      "epoch": 0.9784613947550733,
      "grad_norm": 4.756625175476074,
      "learning_rate": 2.2680716474656265e-08,
      "loss": 0.658,
      "step": 62850
    },
    {
      "epoch": 0.9785392357570427,
      "grad_norm": 4.267798900604248,
      "learning_rate": 2.2598777470051294e-08,
      "loss": 0.7523,
      "step": 62855
    },
    {
      "epoch": 0.978617076759012,
      "grad_norm": 3.4020802974700928,
      "learning_rate": 2.251683846544632e-08,
      "loss": 0.8435,
      "step": 62860
    },
    {
      "epoch": 0.9786949177609814,
      "grad_norm": 5.286039352416992,
      "learning_rate": 2.2434899460841347e-08,
      "loss": 0.7298,
      "step": 62865
    },
    {
      "epoch": 0.9787727587629508,
      "grad_norm": 3.2786128520965576,
      "learning_rate": 2.2352960456236376e-08,
      "loss": 0.6385,
      "step": 62870
    },
    {
      "epoch": 0.9788505997649202,
      "grad_norm": 3.0996387004852295,
      "learning_rate": 2.2271021451631404e-08,
      "loss": 0.6951,
      "step": 62875
    },
    {
      "epoch": 0.9789284407668896,
      "grad_norm": 3.3906211853027344,
      "learning_rate": 2.218908244702643e-08,
      "loss": 0.7171,
      "step": 62880
    },
    {
      "epoch": 0.979006281768859,
      "grad_norm": 7.830942153930664,
      "learning_rate": 2.2107143442421458e-08,
      "loss": 0.7519,
      "step": 62885
    },
    {
      "epoch": 0.9790841227708283,
      "grad_norm": 3.632171392440796,
      "learning_rate": 2.202520443781649e-08,
      "loss": 0.6572,
      "step": 62890
    },
    {
      "epoch": 0.9791619637727977,
      "grad_norm": 4.728374004364014,
      "learning_rate": 2.194326543321152e-08,
      "loss": 0.7584,
      "step": 62895
    },
    {
      "epoch": 0.979239804774767,
      "grad_norm": 3.145098924636841,
      "learning_rate": 2.1861326428606547e-08,
      "loss": 0.6447,
      "step": 62900
    },
    {
      "epoch": 0.9793176457767364,
      "grad_norm": 3.9144887924194336,
      "learning_rate": 2.1779387424001572e-08,
      "loss": 0.7227,
      "step": 62905
    },
    {
      "epoch": 0.9793954867787058,
      "grad_norm": 3.5299084186553955,
      "learning_rate": 2.16974484193966e-08,
      "loss": 0.8285,
      "step": 62910
    },
    {
      "epoch": 0.9794733277806752,
      "grad_norm": 5.97270393371582,
      "learning_rate": 2.161550941479163e-08,
      "loss": 0.7119,
      "step": 62915
    },
    {
      "epoch": 0.9795511687826446,
      "grad_norm": 3.865696430206299,
      "learning_rate": 2.1533570410186658e-08,
      "loss": 0.8266,
      "step": 62920
    },
    {
      "epoch": 0.979629009784614,
      "grad_norm": 3.494713068008423,
      "learning_rate": 2.1451631405581683e-08,
      "loss": 0.7432,
      "step": 62925
    },
    {
      "epoch": 0.9797068507865834,
      "grad_norm": 3.4617433547973633,
      "learning_rate": 2.1369692400976712e-08,
      "loss": 0.655,
      "step": 62930
    },
    {
      "epoch": 0.9797846917885527,
      "grad_norm": 3.6439316272735596,
      "learning_rate": 2.128775339637174e-08,
      "loss": 0.6982,
      "step": 62935
    },
    {
      "epoch": 0.979862532790522,
      "grad_norm": 2.435784339904785,
      "learning_rate": 2.120581439176677e-08,
      "loss": 0.7387,
      "step": 62940
    },
    {
      "epoch": 0.9799403737924914,
      "grad_norm": 5.931297779083252,
      "learning_rate": 2.1123875387161794e-08,
      "loss": 0.8537,
      "step": 62945
    },
    {
      "epoch": 0.9800182147944608,
      "grad_norm": 3.2748823165893555,
      "learning_rate": 2.1041936382556823e-08,
      "loss": 0.6401,
      "step": 62950
    },
    {
      "epoch": 0.9800960557964302,
      "grad_norm": 4.650161266326904,
      "learning_rate": 2.095999737795185e-08,
      "loss": 0.7722,
      "step": 62955
    },
    {
      "epoch": 0.9801738967983996,
      "grad_norm": 2.959012508392334,
      "learning_rate": 2.087805837334688e-08,
      "loss": 0.7011,
      "step": 62960
    },
    {
      "epoch": 0.980251737800369,
      "grad_norm": 6.586910724639893,
      "learning_rate": 2.0796119368741905e-08,
      "loss": 0.7459,
      "step": 62965
    },
    {
      "epoch": 0.9803295788023384,
      "grad_norm": 3.868774890899658,
      "learning_rate": 2.0714180364136937e-08,
      "loss": 0.7572,
      "step": 62970
    },
    {
      "epoch": 0.9804074198043077,
      "grad_norm": 4.615617752075195,
      "learning_rate": 2.0632241359531965e-08,
      "loss": 0.7655,
      "step": 62975
    },
    {
      "epoch": 0.9804852608062771,
      "grad_norm": 3.5439846515655518,
      "learning_rate": 2.0550302354926994e-08,
      "loss": 0.7687,
      "step": 62980
    },
    {
      "epoch": 0.9805631018082465,
      "grad_norm": 3.6441006660461426,
      "learning_rate": 2.046836335032202e-08,
      "loss": 0.6751,
      "step": 62985
    },
    {
      "epoch": 0.9806409428102159,
      "grad_norm": 3.2452664375305176,
      "learning_rate": 2.0386424345717048e-08,
      "loss": 0.6641,
      "step": 62990
    },
    {
      "epoch": 0.9807187838121852,
      "grad_norm": 3.468182325363159,
      "learning_rate": 2.0304485341112076e-08,
      "loss": 0.7197,
      "step": 62995
    },
    {
      "epoch": 0.9807966248141546,
      "grad_norm": 3.146552801132202,
      "learning_rate": 2.0222546336507105e-08,
      "loss": 0.694,
      "step": 63000
    },
    {
      "epoch": 0.980874465816124,
      "grad_norm": 3.003480911254883,
      "learning_rate": 2.014060733190213e-08,
      "loss": 0.7153,
      "step": 63005
    },
    {
      "epoch": 0.9809523068180933,
      "grad_norm": 4.598273754119873,
      "learning_rate": 2.005866832729716e-08,
      "loss": 0.6845,
      "step": 63010
    },
    {
      "epoch": 0.9810301478200627,
      "grad_norm": 3.695894241333008,
      "learning_rate": 1.9976729322692187e-08,
      "loss": 0.6543,
      "step": 63015
    },
    {
      "epoch": 0.9811079888220321,
      "grad_norm": 3.248878002166748,
      "learning_rate": 1.9894790318087215e-08,
      "loss": 0.6973,
      "step": 63020
    },
    {
      "epoch": 0.9811858298240015,
      "grad_norm": 3.1132171154022217,
      "learning_rate": 1.981285131348224e-08,
      "loss": 0.7041,
      "step": 63025
    },
    {
      "epoch": 0.9812636708259709,
      "grad_norm": 3.329530954360962,
      "learning_rate": 1.973091230887727e-08,
      "loss": 0.7864,
      "step": 63030
    },
    {
      "epoch": 0.9813415118279403,
      "grad_norm": 4.599171161651611,
      "learning_rate": 1.9648973304272298e-08,
      "loss": 0.783,
      "step": 63035
    },
    {
      "epoch": 0.9814193528299097,
      "grad_norm": 4.413967132568359,
      "learning_rate": 1.9567034299667326e-08,
      "loss": 0.7563,
      "step": 63040
    },
    {
      "epoch": 0.9814971938318791,
      "grad_norm": 2.903266191482544,
      "learning_rate": 1.948509529506235e-08,
      "loss": 0.7526,
      "step": 63045
    },
    {
      "epoch": 0.9815750348338483,
      "grad_norm": 5.26735258102417,
      "learning_rate": 1.9403156290457383e-08,
      "loss": 0.7082,
      "step": 63050
    },
    {
      "epoch": 0.9816528758358177,
      "grad_norm": 3.434168815612793,
      "learning_rate": 1.9321217285852412e-08,
      "loss": 0.7168,
      "step": 63055
    },
    {
      "epoch": 0.9817307168377871,
      "grad_norm": 3.840156316757202,
      "learning_rate": 1.923927828124744e-08,
      "loss": 0.7439,
      "step": 63060
    },
    {
      "epoch": 0.9818085578397565,
      "grad_norm": 3.7968363761901855,
      "learning_rate": 1.915733927664247e-08,
      "loss": 0.5691,
      "step": 63065
    },
    {
      "epoch": 0.9818863988417259,
      "grad_norm": 5.181361198425293,
      "learning_rate": 1.9075400272037494e-08,
      "loss": 0.79,
      "step": 63070
    },
    {
      "epoch": 0.9819642398436953,
      "grad_norm": 3.5840351581573486,
      "learning_rate": 1.8993461267432523e-08,
      "loss": 0.7164,
      "step": 63075
    },
    {
      "epoch": 0.9820420808456647,
      "grad_norm": 2.9623236656188965,
      "learning_rate": 1.891152226282755e-08,
      "loss": 0.7314,
      "step": 63080
    },
    {
      "epoch": 0.982119921847634,
      "grad_norm": 4.277416706085205,
      "learning_rate": 1.882958325822258e-08,
      "loss": 0.7965,
      "step": 63085
    },
    {
      "epoch": 0.9821977628496034,
      "grad_norm": 4.578347206115723,
      "learning_rate": 1.8747644253617605e-08,
      "loss": 0.7602,
      "step": 63090
    },
    {
      "epoch": 0.9822756038515728,
      "grad_norm": 3.6462652683258057,
      "learning_rate": 1.8665705249012634e-08,
      "loss": 0.7106,
      "step": 63095
    },
    {
      "epoch": 0.9823534448535421,
      "grad_norm": 2.799330234527588,
      "learning_rate": 1.8583766244407662e-08,
      "loss": 0.6982,
      "step": 63100
    },
    {
      "epoch": 0.9824312858555115,
      "grad_norm": 4.43281888961792,
      "learning_rate": 1.850182723980269e-08,
      "loss": 0.8088,
      "step": 63105
    },
    {
      "epoch": 0.9825091268574809,
      "grad_norm": 3.5525224208831787,
      "learning_rate": 1.8419888235197716e-08,
      "loss": 0.6854,
      "step": 63110
    },
    {
      "epoch": 0.9825869678594503,
      "grad_norm": 4.106832027435303,
      "learning_rate": 1.8337949230592744e-08,
      "loss": 0.7306,
      "step": 63115
    },
    {
      "epoch": 0.9826648088614197,
      "grad_norm": 5.8121843338012695,
      "learning_rate": 1.8256010225987773e-08,
      "loss": 0.8192,
      "step": 63120
    },
    {
      "epoch": 0.982742649863389,
      "grad_norm": 3.944413661956787,
      "learning_rate": 1.81740712213828e-08,
      "loss": 0.7457,
      "step": 63125
    },
    {
      "epoch": 0.9828204908653584,
      "grad_norm": 4.050045013427734,
      "learning_rate": 1.809213221677783e-08,
      "loss": 0.7198,
      "step": 63130
    },
    {
      "epoch": 0.9828983318673278,
      "grad_norm": 3.243039608001709,
      "learning_rate": 1.801019321217286e-08,
      "loss": 0.729,
      "step": 63135
    },
    {
      "epoch": 0.9829761728692972,
      "grad_norm": 4.4547553062438965,
      "learning_rate": 1.7928254207567887e-08,
      "loss": 0.7173,
      "step": 63140
    },
    {
      "epoch": 0.9830540138712666,
      "grad_norm": 3.7277109622955322,
      "learning_rate": 1.7846315202962916e-08,
      "loss": 0.7768,
      "step": 63145
    },
    {
      "epoch": 0.983131854873236,
      "grad_norm": 3.2108240127563477,
      "learning_rate": 1.776437619835794e-08,
      "loss": 0.7717,
      "step": 63150
    },
    {
      "epoch": 0.9832096958752053,
      "grad_norm": 2.7831029891967773,
      "learning_rate": 1.768243719375297e-08,
      "loss": 0.7453,
      "step": 63155
    },
    {
      "epoch": 0.9832875368771746,
      "grad_norm": 2.9549577236175537,
      "learning_rate": 1.7600498189147998e-08,
      "loss": 0.7059,
      "step": 63160
    },
    {
      "epoch": 0.983365377879144,
      "grad_norm": 4.167163372039795,
      "learning_rate": 1.7518559184543026e-08,
      "loss": 0.7454,
      "step": 63165
    },
    {
      "epoch": 0.9834432188811134,
      "grad_norm": 2.6713168621063232,
      "learning_rate": 1.7436620179938052e-08,
      "loss": 0.7105,
      "step": 63170
    },
    {
      "epoch": 0.9835210598830828,
      "grad_norm": 4.047041893005371,
      "learning_rate": 1.735468117533308e-08,
      "loss": 0.6727,
      "step": 63175
    },
    {
      "epoch": 0.9835989008850522,
      "grad_norm": 5.343135356903076,
      "learning_rate": 1.727274217072811e-08,
      "loss": 0.7758,
      "step": 63180
    },
    {
      "epoch": 0.9836767418870216,
      "grad_norm": 3.9977025985717773,
      "learning_rate": 1.7190803166123137e-08,
      "loss": 0.7003,
      "step": 63185
    },
    {
      "epoch": 0.983754582888991,
      "grad_norm": 3.4851160049438477,
      "learning_rate": 1.7108864161518163e-08,
      "loss": 0.8484,
      "step": 63190
    },
    {
      "epoch": 0.9838324238909604,
      "grad_norm": 7.829377174377441,
      "learning_rate": 1.702692515691319e-08,
      "loss": 0.656,
      "step": 63195
    },
    {
      "epoch": 0.9839102648929297,
      "grad_norm": 3.5978925228118896,
      "learning_rate": 1.694498615230822e-08,
      "loss": 0.7203,
      "step": 63200
    },
    {
      "epoch": 0.983988105894899,
      "grad_norm": 4.907487392425537,
      "learning_rate": 1.6863047147703248e-08,
      "loss": 0.892,
      "step": 63205
    },
    {
      "epoch": 0.9840659468968684,
      "grad_norm": 4.666064262390137,
      "learning_rate": 1.678110814309828e-08,
      "loss": 0.7717,
      "step": 63210
    },
    {
      "epoch": 0.9841437878988378,
      "grad_norm": 2.898266077041626,
      "learning_rate": 1.6699169138493305e-08,
      "loss": 0.7758,
      "step": 63215
    },
    {
      "epoch": 0.9842216289008072,
      "grad_norm": 2.742621660232544,
      "learning_rate": 1.6617230133888334e-08,
      "loss": 0.6608,
      "step": 63220
    },
    {
      "epoch": 0.9842994699027766,
      "grad_norm": 3.4552276134490967,
      "learning_rate": 1.6535291129283362e-08,
      "loss": 0.7435,
      "step": 63225
    },
    {
      "epoch": 0.984377310904746,
      "grad_norm": 2.249173164367676,
      "learning_rate": 1.645335212467839e-08,
      "loss": 0.7471,
      "step": 63230
    },
    {
      "epoch": 0.9844551519067154,
      "grad_norm": 3.431009292602539,
      "learning_rate": 1.6371413120073416e-08,
      "loss": 0.6535,
      "step": 63235
    },
    {
      "epoch": 0.9845329929086847,
      "grad_norm": 3.8088719844818115,
      "learning_rate": 1.6289474115468445e-08,
      "loss": 0.6581,
      "step": 63240
    },
    {
      "epoch": 0.9846108339106541,
      "grad_norm": 5.0327582359313965,
      "learning_rate": 1.6207535110863473e-08,
      "loss": 0.7829,
      "step": 63245
    },
    {
      "epoch": 0.9846886749126235,
      "grad_norm": 4.2843828201293945,
      "learning_rate": 1.6125596106258502e-08,
      "loss": 0.6666,
      "step": 63250
    },
    {
      "epoch": 0.9847665159145929,
      "grad_norm": 3.4303157329559326,
      "learning_rate": 1.6043657101653527e-08,
      "loss": 0.7,
      "step": 63255
    },
    {
      "epoch": 0.9848443569165622,
      "grad_norm": 4.464983940124512,
      "learning_rate": 1.5961718097048555e-08,
      "loss": 0.6818,
      "step": 63260
    },
    {
      "epoch": 0.9849221979185316,
      "grad_norm": 3.863933563232422,
      "learning_rate": 1.5879779092443584e-08,
      "loss": 0.7808,
      "step": 63265
    },
    {
      "epoch": 0.985000038920501,
      "grad_norm": 7.9665727615356445,
      "learning_rate": 1.5797840087838613e-08,
      "loss": 0.7943,
      "step": 63270
    },
    {
      "epoch": 0.9850778799224703,
      "grad_norm": 3.2777388095855713,
      "learning_rate": 1.5715901083233638e-08,
      "loss": 0.8515,
      "step": 63275
    },
    {
      "epoch": 0.9851557209244397,
      "grad_norm": 2.9687092304229736,
      "learning_rate": 1.5633962078628666e-08,
      "loss": 0.7365,
      "step": 63280
    },
    {
      "epoch": 0.9852335619264091,
      "grad_norm": 3.3816864490509033,
      "learning_rate": 1.5552023074023695e-08,
      "loss": 0.8085,
      "step": 63285
    },
    {
      "epoch": 0.9853114029283785,
      "grad_norm": 2.7601518630981445,
      "learning_rate": 1.5470084069418723e-08,
      "loss": 0.7298,
      "step": 63290
    },
    {
      "epoch": 0.9853892439303479,
      "grad_norm": 4.064361572265625,
      "learning_rate": 1.5388145064813752e-08,
      "loss": 0.687,
      "step": 63295
    },
    {
      "epoch": 0.9854670849323173,
      "grad_norm": 3.289480209350586,
      "learning_rate": 1.530620606020878e-08,
      "loss": 0.7275,
      "step": 63300
    },
    {
      "epoch": 0.9855449259342867,
      "grad_norm": 4.418823719024658,
      "learning_rate": 1.5224267055603806e-08,
      "loss": 0.6412,
      "step": 63305
    },
    {
      "epoch": 0.9856227669362561,
      "grad_norm": 3.4041502475738525,
      "learning_rate": 1.5142328050998838e-08,
      "loss": 0.6441,
      "step": 63310
    },
    {
      "epoch": 0.9857006079382253,
      "grad_norm": 4.295783996582031,
      "learning_rate": 1.5060389046393863e-08,
      "loss": 0.7013,
      "step": 63315
    },
    {
      "epoch": 0.9857784489401947,
      "grad_norm": 2.9157941341400146,
      "learning_rate": 1.497845004178889e-08,
      "loss": 0.7666,
      "step": 63320
    },
    {
      "epoch": 0.9858562899421641,
      "grad_norm": 2.3584704399108887,
      "learning_rate": 1.489651103718392e-08,
      "loss": 0.7332,
      "step": 63325
    },
    {
      "epoch": 0.9859341309441335,
      "grad_norm": 2.76654314994812,
      "learning_rate": 1.4814572032578947e-08,
      "loss": 0.691,
      "step": 63330
    },
    {
      "epoch": 0.9860119719461029,
      "grad_norm": 4.626145839691162,
      "learning_rate": 1.4732633027973975e-08,
      "loss": 0.8145,
      "step": 63335
    },
    {
      "epoch": 0.9860898129480723,
      "grad_norm": 4.060388088226318,
      "learning_rate": 1.4650694023369002e-08,
      "loss": 0.7085,
      "step": 63340
    },
    {
      "epoch": 0.9861676539500417,
      "grad_norm": 3.0222361087799072,
      "learning_rate": 1.456875501876403e-08,
      "loss": 0.7201,
      "step": 63345
    },
    {
      "epoch": 0.986245494952011,
      "grad_norm": 2.3573179244995117,
      "learning_rate": 1.448681601415906e-08,
      "loss": 0.7174,
      "step": 63350
    },
    {
      "epoch": 0.9863233359539804,
      "grad_norm": 3.624553680419922,
      "learning_rate": 1.4404877009554088e-08,
      "loss": 0.6943,
      "step": 63355
    },
    {
      "epoch": 0.9864011769559498,
      "grad_norm": 5.839592933654785,
      "learning_rate": 1.4322938004949116e-08,
      "loss": 0.7545,
      "step": 63360
    },
    {
      "epoch": 0.9864790179579191,
      "grad_norm": 4.504322528839111,
      "learning_rate": 1.4240999000344143e-08,
      "loss": 0.7,
      "step": 63365
    },
    {
      "epoch": 0.9865568589598885,
      "grad_norm": 3.4115378856658936,
      "learning_rate": 1.4159059995739172e-08,
      "loss": 0.6738,
      "step": 63370
    },
    {
      "epoch": 0.9866346999618579,
      "grad_norm": 3.0422797203063965,
      "learning_rate": 1.4077120991134199e-08,
      "loss": 0.6856,
      "step": 63375
    },
    {
      "epoch": 0.9867125409638273,
      "grad_norm": 3.420686721801758,
      "learning_rate": 1.3995181986529227e-08,
      "loss": 0.8186,
      "step": 63380
    },
    {
      "epoch": 0.9867903819657967,
      "grad_norm": 4.220786094665527,
      "learning_rate": 1.3913242981924254e-08,
      "loss": 0.771,
      "step": 63385
    },
    {
      "epoch": 0.986868222967766,
      "grad_norm": 4.191924095153809,
      "learning_rate": 1.3831303977319284e-08,
      "loss": 0.7782,
      "step": 63390
    },
    {
      "epoch": 0.9869460639697354,
      "grad_norm": 5.526742458343506,
      "learning_rate": 1.3749364972714311e-08,
      "loss": 0.7175,
      "step": 63395
    },
    {
      "epoch": 0.9870239049717048,
      "grad_norm": 2.9534404277801514,
      "learning_rate": 1.366742596810934e-08,
      "loss": 0.8195,
      "step": 63400
    },
    {
      "epoch": 0.9871017459736742,
      "grad_norm": 3.153092384338379,
      "learning_rate": 1.3585486963504366e-08,
      "loss": 0.7479,
      "step": 63405
    },
    {
      "epoch": 0.9871795869756436,
      "grad_norm": 5.283040523529053,
      "learning_rate": 1.3503547958899395e-08,
      "loss": 0.785,
      "step": 63410
    },
    {
      "epoch": 0.987257427977613,
      "grad_norm": 6.946015357971191,
      "learning_rate": 1.3421608954294422e-08,
      "loss": 0.7667,
      "step": 63415
    },
    {
      "epoch": 0.9873352689795823,
      "grad_norm": 3.9242262840270996,
      "learning_rate": 1.333966994968945e-08,
      "loss": 0.7316,
      "step": 63420
    },
    {
      "epoch": 0.9874131099815516,
      "grad_norm": 5.770134925842285,
      "learning_rate": 1.3257730945084477e-08,
      "loss": 0.7328,
      "step": 63425
    },
    {
      "epoch": 0.987490950983521,
      "grad_norm": 2.437711715698242,
      "learning_rate": 1.3175791940479508e-08,
      "loss": 0.7378,
      "step": 63430
    },
    {
      "epoch": 0.9875687919854904,
      "grad_norm": 2.459278106689453,
      "learning_rate": 1.3093852935874534e-08,
      "loss": 0.6136,
      "step": 63435
    },
    {
      "epoch": 0.9876466329874598,
      "grad_norm": 3.5294899940490723,
      "learning_rate": 1.3011913931269563e-08,
      "loss": 0.6595,
      "step": 63440
    },
    {
      "epoch": 0.9877244739894292,
      "grad_norm": 4.695808410644531,
      "learning_rate": 1.292997492666459e-08,
      "loss": 0.7258,
      "step": 63445
    },
    {
      "epoch": 0.9878023149913986,
      "grad_norm": 4.942572116851807,
      "learning_rate": 1.2848035922059618e-08,
      "loss": 0.7834,
      "step": 63450
    },
    {
      "epoch": 0.987880155993368,
      "grad_norm": 3.225465774536133,
      "learning_rate": 1.2766096917454645e-08,
      "loss": 0.7237,
      "step": 63455
    },
    {
      "epoch": 0.9879579969953374,
      "grad_norm": 6.471484661102295,
      "learning_rate": 1.2684157912849674e-08,
      "loss": 0.6211,
      "step": 63460
    },
    {
      "epoch": 0.9880358379973067,
      "grad_norm": 2.961005449295044,
      "learning_rate": 1.2602218908244702e-08,
      "loss": 0.7294,
      "step": 63465
    },
    {
      "epoch": 0.988113678999276,
      "grad_norm": 2.908073902130127,
      "learning_rate": 1.2520279903639731e-08,
      "loss": 0.7429,
      "step": 63470
    },
    {
      "epoch": 0.9881915200012454,
      "grad_norm": 3.522310495376587,
      "learning_rate": 1.2438340899034758e-08,
      "loss": 0.8203,
      "step": 63475
    },
    {
      "epoch": 0.9882693610032148,
      "grad_norm": 4.27707576751709,
      "learning_rate": 1.2356401894429786e-08,
      "loss": 0.7404,
      "step": 63480
    },
    {
      "epoch": 0.9883472020051842,
      "grad_norm": 4.352780342102051,
      "learning_rate": 1.2274462889824813e-08,
      "loss": 0.7864,
      "step": 63485
    },
    {
      "epoch": 0.9884250430071536,
      "grad_norm": 3.6105360984802246,
      "learning_rate": 1.2192523885219842e-08,
      "loss": 0.7873,
      "step": 63490
    },
    {
      "epoch": 0.988502884009123,
      "grad_norm": 3.517084836959839,
      "learning_rate": 1.2110584880614869e-08,
      "loss": 0.8336,
      "step": 63495
    },
    {
      "epoch": 0.9885807250110924,
      "grad_norm": 5.928038597106934,
      "learning_rate": 1.2028645876009897e-08,
      "loss": 0.7436,
      "step": 63500
    },
    {
      "epoch": 0.9886585660130617,
      "grad_norm": 4.887537479400635,
      "learning_rate": 1.1946706871404927e-08,
      "loss": 0.7043,
      "step": 63505
    },
    {
      "epoch": 0.9887364070150311,
      "grad_norm": 3.3034863471984863,
      "learning_rate": 1.1864767866799954e-08,
      "loss": 0.7169,
      "step": 63510
    },
    {
      "epoch": 0.9888142480170005,
      "grad_norm": 4.723690509796143,
      "learning_rate": 1.1782828862194983e-08,
      "loss": 0.7158,
      "step": 63515
    },
    {
      "epoch": 0.9888920890189699,
      "grad_norm": 2.7412972450256348,
      "learning_rate": 1.170088985759001e-08,
      "loss": 0.7565,
      "step": 63520
    },
    {
      "epoch": 0.9889699300209392,
      "grad_norm": 4.207612991333008,
      "learning_rate": 1.1618950852985038e-08,
      "loss": 0.6625,
      "step": 63525
    },
    {
      "epoch": 0.9890477710229086,
      "grad_norm": 3.349536895751953,
      "learning_rate": 1.1537011848380065e-08,
      "loss": 0.784,
      "step": 63530
    },
    {
      "epoch": 0.989125612024878,
      "grad_norm": 6.398697376251221,
      "learning_rate": 1.1455072843775094e-08,
      "loss": 0.7222,
      "step": 63535
    },
    {
      "epoch": 0.9892034530268473,
      "grad_norm": 3.0078814029693604,
      "learning_rate": 1.137313383917012e-08,
      "loss": 0.7326,
      "step": 63540
    },
    {
      "epoch": 0.9892812940288167,
      "grad_norm": 3.220118761062622,
      "learning_rate": 1.129119483456515e-08,
      "loss": 0.6291,
      "step": 63545
    },
    {
      "epoch": 0.9893591350307861,
      "grad_norm": 4.881503582000732,
      "learning_rate": 1.1209255829960178e-08,
      "loss": 0.8022,
      "step": 63550
    },
    {
      "epoch": 0.9894369760327555,
      "grad_norm": 5.035337448120117,
      "learning_rate": 1.1127316825355206e-08,
      "loss": 0.7593,
      "step": 63555
    },
    {
      "epoch": 0.9895148170347249,
      "grad_norm": 3.976548671722412,
      "learning_rate": 1.1045377820750233e-08,
      "loss": 0.6964,
      "step": 63560
    },
    {
      "epoch": 0.9895926580366943,
      "grad_norm": 3.393094778060913,
      "learning_rate": 1.0963438816145261e-08,
      "loss": 0.6681,
      "step": 63565
    },
    {
      "epoch": 0.9896704990386637,
      "grad_norm": 4.206606388092041,
      "learning_rate": 1.0881499811540288e-08,
      "loss": 0.7368,
      "step": 63570
    },
    {
      "epoch": 0.9897483400406331,
      "grad_norm": 3.2907967567443848,
      "learning_rate": 1.0799560806935317e-08,
      "loss": 0.7256,
      "step": 63575
    },
    {
      "epoch": 0.9898261810426023,
      "grad_norm": 3.1335582733154297,
      "learning_rate": 1.0717621802330344e-08,
      "loss": 0.7301,
      "step": 63580
    },
    {
      "epoch": 0.9899040220445717,
      "grad_norm": 3.2821850776672363,
      "learning_rate": 1.0635682797725374e-08,
      "loss": 0.6602,
      "step": 63585
    },
    {
      "epoch": 0.9899818630465411,
      "grad_norm": 3.0242841243743896,
      "learning_rate": 1.0553743793120401e-08,
      "loss": 0.7337,
      "step": 63590
    },
    {
      "epoch": 0.9900597040485105,
      "grad_norm": 5.145160675048828,
      "learning_rate": 1.047180478851543e-08,
      "loss": 0.71,
      "step": 63595
    },
    {
      "epoch": 0.9901375450504799,
      "grad_norm": 2.805278778076172,
      "learning_rate": 1.0389865783910456e-08,
      "loss": 0.6654,
      "step": 63600
    },
    {
      "epoch": 0.9902153860524493,
      "grad_norm": 3.444443464279175,
      "learning_rate": 1.0307926779305485e-08,
      "loss": 0.7481,
      "step": 63605
    },
    {
      "epoch": 0.9902932270544187,
      "grad_norm": 3.3366827964782715,
      "learning_rate": 1.0225987774700512e-08,
      "loss": 0.7711,
      "step": 63610
    },
    {
      "epoch": 0.990371068056388,
      "grad_norm": 6.172890663146973,
      "learning_rate": 1.014404877009554e-08,
      "loss": 0.8817,
      "step": 63615
    },
    {
      "epoch": 0.9904489090583574,
      "grad_norm": 3.1443068981170654,
      "learning_rate": 1.0062109765490567e-08,
      "loss": 0.7913,
      "step": 63620
    },
    {
      "epoch": 0.9905267500603268,
      "grad_norm": 5.89206075668335,
      "learning_rate": 9.980170760885597e-09,
      "loss": 0.689,
      "step": 63625
    },
    {
      "epoch": 0.9906045910622961,
      "grad_norm": 3.1322031021118164,
      "learning_rate": 9.898231756280624e-09,
      "loss": 0.7409,
      "step": 63630
    },
    {
      "epoch": 0.9906824320642655,
      "grad_norm": 9.611970901489258,
      "learning_rate": 9.816292751675653e-09,
      "loss": 0.7064,
      "step": 63635
    },
    {
      "epoch": 0.9907602730662349,
      "grad_norm": 3.8289313316345215,
      "learning_rate": 9.73435374707068e-09,
      "loss": 0.8549,
      "step": 63640
    },
    {
      "epoch": 0.9908381140682043,
      "grad_norm": 6.584836006164551,
      "learning_rate": 9.652414742465708e-09,
      "loss": 0.7016,
      "step": 63645
    },
    {
      "epoch": 0.9909159550701737,
      "grad_norm": 4.992436408996582,
      "learning_rate": 9.570475737860735e-09,
      "loss": 0.8704,
      "step": 63650
    },
    {
      "epoch": 0.990993796072143,
      "grad_norm": 3.3906161785125732,
      "learning_rate": 9.488536733255764e-09,
      "loss": 0.7526,
      "step": 63655
    },
    {
      "epoch": 0.9910716370741124,
      "grad_norm": 3.6645352840423584,
      "learning_rate": 9.40659772865079e-09,
      "loss": 0.6264,
      "step": 63660
    },
    {
      "epoch": 0.9911494780760818,
      "grad_norm": 4.086439609527588,
      "learning_rate": 9.32465872404582e-09,
      "loss": 0.7865,
      "step": 63665
    },
    {
      "epoch": 0.9912273190780512,
      "grad_norm": 3.759938955307007,
      "learning_rate": 9.242719719440849e-09,
      "loss": 0.7763,
      "step": 63670
    },
    {
      "epoch": 0.9913051600800206,
      "grad_norm": 5.700982570648193,
      "learning_rate": 9.160780714835876e-09,
      "loss": 0.7765,
      "step": 63675
    },
    {
      "epoch": 0.99138300108199,
      "grad_norm": 3.1751596927642822,
      "learning_rate": 9.078841710230905e-09,
      "loss": 0.8128,
      "step": 63680
    },
    {
      "epoch": 0.9914608420839593,
      "grad_norm": 4.202373027801514,
      "learning_rate": 8.996902705625931e-09,
      "loss": 0.7577,
      "step": 63685
    },
    {
      "epoch": 0.9915386830859286,
      "grad_norm": 2.516160011291504,
      "learning_rate": 8.91496370102096e-09,
      "loss": 0.7517,
      "step": 63690
    },
    {
      "epoch": 0.991616524087898,
      "grad_norm": 4.633602619171143,
      "learning_rate": 8.833024696415987e-09,
      "loss": 0.7085,
      "step": 63695
    },
    {
      "epoch": 0.9916943650898674,
      "grad_norm": 3.636209011077881,
      "learning_rate": 8.751085691811015e-09,
      "loss": 0.6345,
      "step": 63700
    },
    {
      "epoch": 0.9917722060918368,
      "grad_norm": 11.219762802124023,
      "learning_rate": 8.669146687206044e-09,
      "loss": 0.7115,
      "step": 63705
    },
    {
      "epoch": 0.9918500470938062,
      "grad_norm": 2.4765374660491943,
      "learning_rate": 8.587207682601073e-09,
      "loss": 0.721,
      "step": 63710
    },
    {
      "epoch": 0.9919278880957756,
      "grad_norm": 3.6374223232269287,
      "learning_rate": 8.5052686779961e-09,
      "loss": 0.7685,
      "step": 63715
    },
    {
      "epoch": 0.992005729097745,
      "grad_norm": 5.074397087097168,
      "learning_rate": 8.423329673391128e-09,
      "loss": 0.7528,
      "step": 63720
    },
    {
      "epoch": 0.9920835700997144,
      "grad_norm": 5.595132827758789,
      "learning_rate": 8.341390668786155e-09,
      "loss": 0.675,
      "step": 63725
    },
    {
      "epoch": 0.9921614111016837,
      "grad_norm": 3.179172992706299,
      "learning_rate": 8.259451664181183e-09,
      "loss": 0.7011,
      "step": 63730
    },
    {
      "epoch": 0.992239252103653,
      "grad_norm": 4.581373691558838,
      "learning_rate": 8.17751265957621e-09,
      "loss": 0.7318,
      "step": 63735
    },
    {
      "epoch": 0.9923170931056224,
      "grad_norm": 2.5164904594421387,
      "learning_rate": 8.095573654971239e-09,
      "loss": 0.6751,
      "step": 63740
    },
    {
      "epoch": 0.9923949341075918,
      "grad_norm": 4.48675012588501,
      "learning_rate": 8.013634650366267e-09,
      "loss": 0.6641,
      "step": 63745
    },
    {
      "epoch": 0.9924727751095612,
      "grad_norm": 3.9297633171081543,
      "learning_rate": 7.931695645761296e-09,
      "loss": 0.6435,
      "step": 63750
    },
    {
      "epoch": 0.9925506161115306,
      "grad_norm": 9.197453498840332,
      "learning_rate": 7.849756641156323e-09,
      "loss": 0.6989,
      "step": 63755
    },
    {
      "epoch": 0.9926284571135,
      "grad_norm": 4.466760158538818,
      "learning_rate": 7.767817636551351e-09,
      "loss": 0.7459,
      "step": 63760
    },
    {
      "epoch": 0.9927062981154693,
      "grad_norm": 3.9505622386932373,
      "learning_rate": 7.685878631946378e-09,
      "loss": 0.7867,
      "step": 63765
    },
    {
      "epoch": 0.9927841391174387,
      "grad_norm": 5.759352684020996,
      "learning_rate": 7.603939627341407e-09,
      "loss": 0.7709,
      "step": 63770
    },
    {
      "epoch": 0.9928619801194081,
      "grad_norm": 5.855334281921387,
      "learning_rate": 7.522000622736435e-09,
      "loss": 0.7251,
      "step": 63775
    },
    {
      "epoch": 0.9929398211213775,
      "grad_norm": 6.042656421661377,
      "learning_rate": 7.440061618131463e-09,
      "loss": 0.7153,
      "step": 63780
    },
    {
      "epoch": 0.9930176621233469,
      "grad_norm": 11.473960876464844,
      "learning_rate": 7.358122613526491e-09,
      "loss": 0.7768,
      "step": 63785
    },
    {
      "epoch": 0.9930955031253162,
      "grad_norm": 5.3112311363220215,
      "learning_rate": 7.276183608921518e-09,
      "loss": 0.8133,
      "step": 63790
    },
    {
      "epoch": 0.9931733441272856,
      "grad_norm": 3.379352569580078,
      "learning_rate": 7.194244604316547e-09,
      "loss": 0.6207,
      "step": 63795
    },
    {
      "epoch": 0.993251185129255,
      "grad_norm": 2.982384443283081,
      "learning_rate": 7.112305599711575e-09,
      "loss": 0.6845,
      "step": 63800
    },
    {
      "epoch": 0.9933290261312243,
      "grad_norm": 5.671186923980713,
      "learning_rate": 7.030366595106602e-09,
      "loss": 0.6292,
      "step": 63805
    },
    {
      "epoch": 0.9934068671331937,
      "grad_norm": 3.4236390590667725,
      "learning_rate": 6.94842759050163e-09,
      "loss": 0.6867,
      "step": 63810
    },
    {
      "epoch": 0.9934847081351631,
      "grad_norm": 4.0309953689575195,
      "learning_rate": 6.8664885858966586e-09,
      "loss": 0.6085,
      "step": 63815
    },
    {
      "epoch": 0.9935625491371325,
      "grad_norm": 6.011270523071289,
      "learning_rate": 6.784549581291686e-09,
      "loss": 0.7131,
      "step": 63820
    },
    {
      "epoch": 0.9936403901391019,
      "grad_norm": 8.832276344299316,
      "learning_rate": 6.702610576686714e-09,
      "loss": 0.8149,
      "step": 63825
    },
    {
      "epoch": 0.9937182311410713,
      "grad_norm": 3.802452802658081,
      "learning_rate": 6.620671572081742e-09,
      "loss": 0.6564,
      "step": 63830
    },
    {
      "epoch": 0.9937960721430407,
      "grad_norm": 4.601316928863525,
      "learning_rate": 6.53873256747677e-09,
      "loss": 0.7051,
      "step": 63835
    },
    {
      "epoch": 0.9938739131450101,
      "grad_norm": 3.938494920730591,
      "learning_rate": 6.456793562871798e-09,
      "loss": 0.7754,
      "step": 63840
    },
    {
      "epoch": 0.9939517541469793,
      "grad_norm": 3.3142669200897217,
      "learning_rate": 6.374854558266826e-09,
      "loss": 0.8686,
      "step": 63845
    },
    {
      "epoch": 0.9940295951489487,
      "grad_norm": 5.964455604553223,
      "learning_rate": 6.292915553661853e-09,
      "loss": 0.7309,
      "step": 63850
    },
    {
      "epoch": 0.9941074361509181,
      "grad_norm": 2.859832525253296,
      "learning_rate": 6.210976549056882e-09,
      "loss": 0.6887,
      "step": 63855
    },
    {
      "epoch": 0.9941852771528875,
      "grad_norm": 4.688760757446289,
      "learning_rate": 6.12903754445191e-09,
      "loss": 0.7615,
      "step": 63860
    },
    {
      "epoch": 0.9942631181548569,
      "grad_norm": 6.191370010375977,
      "learning_rate": 6.047098539846937e-09,
      "loss": 0.7128,
      "step": 63865
    },
    {
      "epoch": 0.9943409591568263,
      "grad_norm": 4.318521022796631,
      "learning_rate": 5.965159535241965e-09,
      "loss": 0.7158,
      "step": 63870
    },
    {
      "epoch": 0.9944188001587957,
      "grad_norm": 3.096219301223755,
      "learning_rate": 5.8832205306369936e-09,
      "loss": 0.6002,
      "step": 63875
    },
    {
      "epoch": 0.994496641160765,
      "grad_norm": 4.041717529296875,
      "learning_rate": 5.801281526032021e-09,
      "loss": 0.7229,
      "step": 63880
    },
    {
      "epoch": 0.9945744821627344,
      "grad_norm": 2.6838531494140625,
      "learning_rate": 5.719342521427049e-09,
      "loss": 0.6995,
      "step": 63885
    },
    {
      "epoch": 0.9946523231647038,
      "grad_norm": 6.544908046722412,
      "learning_rate": 5.6374035168220775e-09,
      "loss": 0.8077,
      "step": 63890
    },
    {
      "epoch": 0.9947301641666731,
      "grad_norm": 2.88291335105896,
      "learning_rate": 5.555464512217106e-09,
      "loss": 0.7053,
      "step": 63895
    },
    {
      "epoch": 0.9948080051686425,
      "grad_norm": 4.807986259460449,
      "learning_rate": 5.473525507612134e-09,
      "loss": 0.8752,
      "step": 63900
    },
    {
      "epoch": 0.9948858461706119,
      "grad_norm": 3.5011560916900635,
      "learning_rate": 5.3915865030071615e-09,
      "loss": 0.702,
      "step": 63905
    },
    {
      "epoch": 0.9949636871725813,
      "grad_norm": 4.249161720275879,
      "learning_rate": 5.309647498402189e-09,
      "loss": 0.7289,
      "step": 63910
    },
    {
      "epoch": 0.9950415281745507,
      "grad_norm": 3.4885101318359375,
      "learning_rate": 5.227708493797218e-09,
      "loss": 0.7275,
      "step": 63915
    },
    {
      "epoch": 0.99511936917652,
      "grad_norm": 6.597133636474609,
      "learning_rate": 5.1457694891922454e-09,
      "loss": 0.7917,
      "step": 63920
    },
    {
      "epoch": 0.9951972101784894,
      "grad_norm": 5.384420394897461,
      "learning_rate": 5.063830484587273e-09,
      "loss": 0.8161,
      "step": 63925
    },
    {
      "epoch": 0.9952750511804588,
      "grad_norm": 3.701493501663208,
      "learning_rate": 4.981891479982301e-09,
      "loss": 0.7726,
      "step": 63930
    },
    {
      "epoch": 0.9953528921824282,
      "grad_norm": 6.28872537612915,
      "learning_rate": 4.899952475377329e-09,
      "loss": 0.7169,
      "step": 63935
    },
    {
      "epoch": 0.9954307331843976,
      "grad_norm": 7.4044084548950195,
      "learning_rate": 4.818013470772357e-09,
      "loss": 0.7574,
      "step": 63940
    },
    {
      "epoch": 0.995508574186367,
      "grad_norm": 2.8994486331939697,
      "learning_rate": 4.736074466167385e-09,
      "loss": 0.7738,
      "step": 63945
    },
    {
      "epoch": 0.9955864151883363,
      "grad_norm": 7.800544261932373,
      "learning_rate": 4.6541354615624125e-09,
      "loss": 0.7934,
      "step": 63950
    },
    {
      "epoch": 0.9956642561903056,
      "grad_norm": 4.4726033210754395,
      "learning_rate": 4.572196456957441e-09,
      "loss": 0.6513,
      "step": 63955
    },
    {
      "epoch": 0.995742097192275,
      "grad_norm": 5.063044548034668,
      "learning_rate": 4.490257452352469e-09,
      "loss": 0.7686,
      "step": 63960
    },
    {
      "epoch": 0.9958199381942444,
      "grad_norm": 2.8225364685058594,
      "learning_rate": 4.4083184477474965e-09,
      "loss": 0.6312,
      "step": 63965
    },
    {
      "epoch": 0.9958977791962138,
      "grad_norm": 2.7927982807159424,
      "learning_rate": 4.326379443142524e-09,
      "loss": 0.7251,
      "step": 63970
    },
    {
      "epoch": 0.9959756201981832,
      "grad_norm": 3.456210136413574,
      "learning_rate": 4.244440438537553e-09,
      "loss": 0.7761,
      "step": 63975
    },
    {
      "epoch": 0.9960534612001526,
      "grad_norm": 3.3364107608795166,
      "learning_rate": 4.1625014339325804e-09,
      "loss": 0.7261,
      "step": 63980
    },
    {
      "epoch": 0.996131302202122,
      "grad_norm": 3.2743895053863525,
      "learning_rate": 4.080562429327608e-09,
      "loss": 0.8254,
      "step": 63985
    },
    {
      "epoch": 0.9962091432040914,
      "grad_norm": 3.1898353099823,
      "learning_rate": 3.998623424722636e-09,
      "loss": 0.6913,
      "step": 63990
    },
    {
      "epoch": 0.9962869842060607,
      "grad_norm": 3.0031776428222656,
      "learning_rate": 3.916684420117664e-09,
      "loss": 0.7264,
      "step": 63995
    },
    {
      "epoch": 0.99636482520803,
      "grad_norm": 3.2354369163513184,
      "learning_rate": 3.834745415512692e-09,
      "loss": 0.719,
      "step": 64000
    },
    {
      "epoch": 0.9964426662099994,
      "grad_norm": 2.8282511234283447,
      "learning_rate": 3.75280641090772e-09,
      "loss": 0.6692,
      "step": 64005
    },
    {
      "epoch": 0.9965205072119688,
      "grad_norm": 3.7697787284851074,
      "learning_rate": 3.6708674063027483e-09,
      "loss": 0.6193,
      "step": 64010
    },
    {
      "epoch": 0.9965983482139382,
      "grad_norm": 4.990572929382324,
      "learning_rate": 3.588928401697776e-09,
      "loss": 0.7634,
      "step": 64015
    },
    {
      "epoch": 0.9966761892159076,
      "grad_norm": 3.8955929279327393,
      "learning_rate": 3.506989397092804e-09,
      "loss": 0.7274,
      "step": 64020
    },
    {
      "epoch": 0.996754030217877,
      "grad_norm": 3.4396493434906006,
      "learning_rate": 3.425050392487832e-09,
      "loss": 0.6157,
      "step": 64025
    },
    {
      "epoch": 0.9968318712198463,
      "grad_norm": 4.585066318511963,
      "learning_rate": 3.34311138788286e-09,
      "loss": 0.6847,
      "step": 64030
    },
    {
      "epoch": 0.9969097122218157,
      "grad_norm": 2.69274640083313,
      "learning_rate": 3.2611723832778877e-09,
      "loss": 0.6515,
      "step": 64035
    },
    {
      "epoch": 0.9969875532237851,
      "grad_norm": 3.8642847537994385,
      "learning_rate": 3.179233378672916e-09,
      "loss": 0.7596,
      "step": 64040
    },
    {
      "epoch": 0.9970653942257545,
      "grad_norm": 9.793916702270508,
      "learning_rate": 3.0972943740679435e-09,
      "loss": 0.7656,
      "step": 64045
    },
    {
      "epoch": 0.9971432352277239,
      "grad_norm": 4.169534206390381,
      "learning_rate": 3.0153553694629717e-09,
      "loss": 0.7158,
      "step": 64050
    },
    {
      "epoch": 0.9972210762296932,
      "grad_norm": 4.42549467086792,
      "learning_rate": 2.9334163648579994e-09,
      "loss": 0.7762,
      "step": 64055
    },
    {
      "epoch": 0.9972989172316626,
      "grad_norm": 4.962855815887451,
      "learning_rate": 2.8514773602530275e-09,
      "loss": 0.7151,
      "step": 64060
    },
    {
      "epoch": 0.997376758233632,
      "grad_norm": 3.9648306369781494,
      "learning_rate": 2.7695383556480552e-09,
      "loss": 0.7031,
      "step": 64065
    },
    {
      "epoch": 0.9974545992356013,
      "grad_norm": 2.9575376510620117,
      "learning_rate": 2.6875993510430833e-09,
      "loss": 0.6659,
      "step": 64070
    },
    {
      "epoch": 0.9975324402375707,
      "grad_norm": 3.397369861602783,
      "learning_rate": 2.605660346438111e-09,
      "loss": 0.6479,
      "step": 64075
    },
    {
      "epoch": 0.9976102812395401,
      "grad_norm": 3.3934481143951416,
      "learning_rate": 2.5237213418331396e-09,
      "loss": 0.649,
      "step": 64080
    },
    {
      "epoch": 0.9976881222415095,
      "grad_norm": 4.995265483856201,
      "learning_rate": 2.4417823372281673e-09,
      "loss": 0.6933,
      "step": 64085
    },
    {
      "epoch": 0.9977659632434789,
      "grad_norm": 4.784320831298828,
      "learning_rate": 2.3598433326231954e-09,
      "loss": 0.7701,
      "step": 64090
    },
    {
      "epoch": 0.9978438042454483,
      "grad_norm": 4.577213287353516,
      "learning_rate": 2.277904328018223e-09,
      "loss": 0.7319,
      "step": 64095
    },
    {
      "epoch": 0.9979216452474177,
      "grad_norm": 3.550696611404419,
      "learning_rate": 2.1959653234132513e-09,
      "loss": 0.6825,
      "step": 64100
    },
    {
      "epoch": 0.997999486249387,
      "grad_norm": 9.633803367614746,
      "learning_rate": 2.114026318808279e-09,
      "loss": 0.7031,
      "step": 64105
    },
    {
      "epoch": 0.9980773272513563,
      "grad_norm": 3.705124616622925,
      "learning_rate": 2.032087314203307e-09,
      "loss": 0.6904,
      "step": 64110
    },
    {
      "epoch": 0.9981551682533257,
      "grad_norm": 3.8707830905914307,
      "learning_rate": 1.9501483095983352e-09,
      "loss": 0.7224,
      "step": 64115
    },
    {
      "epoch": 0.9982330092552951,
      "grad_norm": 3.631843328475952,
      "learning_rate": 1.868209304993363e-09,
      "loss": 0.7486,
      "step": 64120
    },
    {
      "epoch": 0.9983108502572645,
      "grad_norm": 4.572499752044678,
      "learning_rate": 1.7862703003883908e-09,
      "loss": 0.7388,
      "step": 64125
    },
    {
      "epoch": 0.9983886912592339,
      "grad_norm": 4.41572904586792,
      "learning_rate": 1.7043312957834188e-09,
      "loss": 0.7605,
      "step": 64130
    },
    {
      "epoch": 0.9984665322612033,
      "grad_norm": 3.37788462638855,
      "learning_rate": 1.6223922911784469e-09,
      "loss": 0.7602,
      "step": 64135
    },
    {
      "epoch": 0.9985443732631727,
      "grad_norm": 3.5984556674957275,
      "learning_rate": 1.5404532865734748e-09,
      "loss": 0.6122,
      "step": 64140
    },
    {
      "epoch": 0.998622214265142,
      "grad_norm": 2.9469170570373535,
      "learning_rate": 1.4585142819685027e-09,
      "loss": 0.6616,
      "step": 64145
    },
    {
      "epoch": 0.9987000552671114,
      "grad_norm": 2.5429351329803467,
      "learning_rate": 1.3765752773635306e-09,
      "loss": 0.7318,
      "step": 64150
    },
    {
      "epoch": 0.9987778962690808,
      "grad_norm": 3.565129041671753,
      "learning_rate": 1.2946362727585585e-09,
      "loss": 0.6985,
      "step": 64155
    },
    {
      "epoch": 0.9988557372710501,
      "grad_norm": 3.2046306133270264,
      "learning_rate": 1.2126972681535865e-09,
      "loss": 0.768,
      "step": 64160
    },
    {
      "epoch": 0.9989335782730195,
      "grad_norm": 6.987384796142578,
      "learning_rate": 1.1307582635486144e-09,
      "loss": 0.7084,
      "step": 64165
    },
    {
      "epoch": 0.9990114192749889,
      "grad_norm": 2.9052305221557617,
      "learning_rate": 1.0488192589436425e-09,
      "loss": 0.6981,
      "step": 64170
    },
    {
      "epoch": 0.9990892602769583,
      "grad_norm": 3.3828189373016357,
      "learning_rate": 9.668802543386702e-10,
      "loss": 0.7198,
      "step": 64175
    },
    {
      "epoch": 0.9991671012789277,
      "grad_norm": 3.4840378761291504,
      "learning_rate": 8.849412497336982e-10,
      "loss": 0.6863,
      "step": 64180
    },
    {
      "epoch": 0.999244942280897,
      "grad_norm": 3.523324966430664,
      "learning_rate": 8.030022451287261e-10,
      "loss": 0.7574,
      "step": 64185
    },
    {
      "epoch": 0.9993227832828664,
      "grad_norm": 3.2661666870117188,
      "learning_rate": 7.210632405237541e-10,
      "loss": 0.821,
      "step": 64190
    },
    {
      "epoch": 0.9994006242848358,
      "grad_norm": 3.1247384548187256,
      "learning_rate": 6.39124235918782e-10,
      "loss": 0.7921,
      "step": 64195
    },
    {
      "epoch": 0.9994784652868052,
      "grad_norm": 3.888141632080078,
      "learning_rate": 5.5718523131381e-10,
      "loss": 0.6868,
      "step": 64200
    },
    {
      "epoch": 0.9995563062887746,
      "grad_norm": 5.796787261962891,
      "learning_rate": 4.752462267088379e-10,
      "loss": 0.7397,
      "step": 64205
    },
    {
      "epoch": 0.999634147290744,
      "grad_norm": 7.851982593536377,
      "learning_rate": 3.9330722210386583e-10,
      "loss": 0.7566,
      "step": 64210
    },
    {
      "epoch": 0.9997119882927133,
      "grad_norm": 3.2448983192443848,
      "learning_rate": 3.113682174988938e-10,
      "loss": 0.65,
      "step": 64215
    },
    {
      "epoch": 0.9997898292946826,
      "grad_norm": 4.299083232879639,
      "learning_rate": 2.2942921289392177e-10,
      "loss": 0.8221,
      "step": 64220
    },
    {
      "epoch": 0.999867670296652,
      "grad_norm": 6.4583916664123535,
      "learning_rate": 1.4749020828894971e-10,
      "loss": 0.8213,
      "step": 64225
    },
    {
      "epoch": 0.9999455112986214,
      "grad_norm": 3.6320903301239014,
      "learning_rate": 6.555120368397766e-11,
      "loss": 0.8549,
      "step": 64230
    },
    {
      "epoch": 0.999992215899803,
      "step": 64233,
      "total_flos": 2.57367992338974e+19,
      "train_loss": 0.7666023439348673,
      "train_runtime": 158972.6632,
      "train_samples_per_second": 6.465,
      "train_steps_per_second": 0.404
    }
  ],
  "logging_steps": 5,
  "max_steps": 64233,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 3000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.57367992338974e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}