{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 25000,
  "global_step": 61115,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016362595107584062,
      "grad_norm": 9.953512504451753,
      "learning_rate": 1.636125654450262e-08,
      "loss": 0.8675,
      "step": 10
    },
    {
      "epoch": 0.00032725190215168124,
      "grad_norm": 9.553511923060478,
      "learning_rate": 3.272251308900524e-08,
      "loss": 0.8383,
      "step": 20
    },
    {
      "epoch": 0.0004908778532275219,
      "grad_norm": 11.131209724566652,
      "learning_rate": 4.908376963350786e-08,
      "loss": 0.8342,
      "step": 30
    },
    {
      "epoch": 0.0006545038043033625,
      "grad_norm": 7.5997281783840185,
      "learning_rate": 6.544502617801048e-08,
      "loss": 0.8489,
      "step": 40
    },
    {
      "epoch": 0.0008181297553792031,
      "grad_norm": 13.653268825185362,
      "learning_rate": 8.18062827225131e-08,
      "loss": 0.8987,
      "step": 50
    },
    {
      "epoch": 0.0009817557064550437,
      "grad_norm": 11.055120151769751,
      "learning_rate": 9.816753926701572e-08,
      "loss": 0.784,
      "step": 60
    },
    {
      "epoch": 0.0011453816575308843,
      "grad_norm": 9.500150545973115,
      "learning_rate": 1.1452879581151833e-07,
      "loss": 0.8406,
      "step": 70
    },
    {
      "epoch": 0.001309007608606725,
      "grad_norm": 8.78176130099345,
      "learning_rate": 1.3089005235602095e-07,
      "loss": 0.7883,
      "step": 80
    },
    {
      "epoch": 0.0014726335596825656,
      "grad_norm": 9.285787853972339,
      "learning_rate": 1.4725130890052356e-07,
      "loss": 0.7712,
      "step": 90
    },
    {
      "epoch": 0.0016362595107584062,
      "grad_norm": 10.351765898816028,
      "learning_rate": 1.636125654450262e-07,
      "loss": 0.7958,
      "step": 100
    },
    {
      "epoch": 0.0017998854618342468,
      "grad_norm": 8.451575407818002,
      "learning_rate": 1.7997382198952883e-07,
      "loss": 0.7156,
      "step": 110
    },
    {
      "epoch": 0.0019635114129100874,
      "grad_norm": 8.024542877925896,
      "learning_rate": 1.9633507853403144e-07,
      "loss": 0.7417,
      "step": 120
    },
    {
      "epoch": 0.002127137363985928,
      "grad_norm": 9.133613751637023,
      "learning_rate": 2.1269633507853407e-07,
      "loss": 0.6717,
      "step": 130
    },
    {
      "epoch": 0.0022907633150617687,
      "grad_norm": 5.968510512679798,
      "learning_rate": 2.2905759162303666e-07,
      "loss": 0.6115,
      "step": 140
    },
    {
      "epoch": 0.0024543892661376093,
      "grad_norm": 6.194628606572387,
      "learning_rate": 2.454188481675393e-07,
      "loss": 0.512,
      "step": 150
    },
    {
      "epoch": 0.00261801521721345,
      "grad_norm": 5.464433852845115,
      "learning_rate": 2.617801047120419e-07,
      "loss": 0.4875,
      "step": 160
    },
    {
      "epoch": 0.0027816411682892906,
      "grad_norm": 3.873490318989648,
      "learning_rate": 2.7814136125654456e-07,
      "loss": 0.4097,
      "step": 170
    },
    {
      "epoch": 0.002945267119365131,
      "grad_norm": 4.54004618500498,
      "learning_rate": 2.945026178010471e-07,
      "loss": 0.3477,
      "step": 180
    },
    {
      "epoch": 0.003108893070440972,
      "grad_norm": 3.330129557566636,
      "learning_rate": 3.108638743455498e-07,
      "loss": 0.2964,
      "step": 190
    },
    {
      "epoch": 0.0032725190215168124,
      "grad_norm": 3.5541678440135365,
      "learning_rate": 3.272251308900524e-07,
      "loss": 0.2507,
      "step": 200
    },
    {
      "epoch": 0.003436144972592653,
      "grad_norm": 3.2547986736994554,
      "learning_rate": 3.43586387434555e-07,
      "loss": 0.2122,
      "step": 210
    },
    {
      "epoch": 0.0035997709236684937,
      "grad_norm": 2.683447457033073,
      "learning_rate": 3.5994764397905766e-07,
      "loss": 0.2045,
      "step": 220
    },
    {
      "epoch": 0.0037633968747443343,
      "grad_norm": 3.0304753757748926,
      "learning_rate": 3.7630890052356027e-07,
      "loss": 0.1978,
      "step": 230
    },
    {
      "epoch": 0.003927022825820175,
      "grad_norm": 2.2690815575105,
      "learning_rate": 3.926701570680629e-07,
      "loss": 0.179,
      "step": 240
    },
    {
      "epoch": 0.004090648776896016,
      "grad_norm": 2.4354798410886325,
      "learning_rate": 4.0903141361256543e-07,
      "loss": 0.1471,
      "step": 250
    },
    {
      "epoch": 0.004254274727971856,
      "grad_norm": 2.8505689228875797,
      "learning_rate": 4.2539267015706815e-07,
      "loss": 0.1633,
      "step": 260
    },
    {
      "epoch": 0.004417900679047697,
      "grad_norm": 2.425772445148507,
      "learning_rate": 4.4175392670157076e-07,
      "loss": 0.1502,
      "step": 270
    },
    {
      "epoch": 0.004581526630123537,
      "grad_norm": 1.9156226126443814,
      "learning_rate": 4.581151832460733e-07,
      "loss": 0.1299,
      "step": 280
    },
    {
      "epoch": 0.004745152581199378,
      "grad_norm": 2.7553677242773933,
      "learning_rate": 4.744764397905759e-07,
      "loss": 0.1284,
      "step": 290
    },
    {
      "epoch": 0.004908778532275219,
      "grad_norm": 2.0747187707340426,
      "learning_rate": 4.908376963350786e-07,
      "loss": 0.1341,
      "step": 300
    },
    {
      "epoch": 0.00507240448335106,
      "grad_norm": 2.545379828991952,
      "learning_rate": 5.071989528795812e-07,
      "loss": 0.1278,
      "step": 310
    },
    {
      "epoch": 0.0052360304344269,
      "grad_norm": 1.6684754833198432,
      "learning_rate": 5.235602094240838e-07,
      "loss": 0.1134,
      "step": 320
    },
    {
      "epoch": 0.005399656385502741,
      "grad_norm": 2.5489345776008667,
      "learning_rate": 5.399214659685865e-07,
      "loss": 0.1041,
      "step": 330
    },
    {
      "epoch": 0.005563282336578581,
      "grad_norm": 2.083270341063134,
      "learning_rate": 5.562827225130891e-07,
      "loss": 0.083,
      "step": 340
    },
    {
      "epoch": 0.005726908287654422,
      "grad_norm": 2.561527976097819,
      "learning_rate": 5.726439790575917e-07,
      "loss": 0.1129,
      "step": 350
    },
    {
      "epoch": 0.005890534238730262,
      "grad_norm": 2.332969713459923,
      "learning_rate": 5.890052356020942e-07,
      "loss": 0.102,
      "step": 360
    },
    {
      "epoch": 0.006054160189806103,
      "grad_norm": 3.1385729593806215,
      "learning_rate": 6.053664921465969e-07,
      "loss": 0.1054,
      "step": 370
    },
    {
      "epoch": 0.006217786140881944,
      "grad_norm": 2.8824313153043963,
      "learning_rate": 6.217277486910996e-07,
      "loss": 0.0974,
      "step": 380
    },
    {
      "epoch": 0.006381412091957785,
      "grad_norm": 2.2879714326506653,
      "learning_rate": 6.380890052356021e-07,
      "loss": 0.0753,
      "step": 390
    },
    {
      "epoch": 0.006545038043033625,
      "grad_norm": 1.8575305027963693,
      "learning_rate": 6.544502617801048e-07,
      "loss": 0.1008,
      "step": 400
    },
    {
      "epoch": 0.006708663994109466,
      "grad_norm": 1.8820501103558993,
      "learning_rate": 6.708115183246073e-07,
      "loss": 0.0916,
      "step": 410
    },
    {
      "epoch": 0.006872289945185306,
      "grad_norm": 2.1069617625641377,
      "learning_rate": 6.8717277486911e-07,
      "loss": 0.0899,
      "step": 420
    },
    {
      "epoch": 0.007035915896261147,
      "grad_norm": 2.129401250846256,
      "learning_rate": 7.035340314136126e-07,
      "loss": 0.0752,
      "step": 430
    },
    {
      "epoch": 0.007199541847336987,
      "grad_norm": 2.082577066787224,
      "learning_rate": 7.198952879581153e-07,
      "loss": 0.0923,
      "step": 440
    },
    {
      "epoch": 0.007363167798412828,
      "grad_norm": 2.092195120658152,
      "learning_rate": 7.362565445026179e-07,
      "loss": 0.0877,
      "step": 450
    },
    {
      "epoch": 0.0075267937494886685,
      "grad_norm": 2.38358275909396,
      "learning_rate": 7.526178010471205e-07,
      "loss": 0.0846,
      "step": 460
    },
    {
      "epoch": 0.00769041970056451,
      "grad_norm": 2.046058464990723,
      "learning_rate": 7.689790575916231e-07,
      "loss": 0.0744,
      "step": 470
    },
    {
      "epoch": 0.00785404565164035,
      "grad_norm": 2.255986980690557,
      "learning_rate": 7.853403141361258e-07,
      "loss": 0.0866,
      "step": 480
    },
    {
      "epoch": 0.00801767160271619,
      "grad_norm": 2.2406121886717503,
      "learning_rate": 8.017015706806283e-07,
      "loss": 0.0694,
      "step": 490
    },
    {
      "epoch": 0.008181297553792032,
      "grad_norm": 1.843535855319224,
      "learning_rate": 8.180628272251309e-07,
      "loss": 0.0726,
      "step": 500
    },
    {
      "epoch": 0.008344923504867872,
      "grad_norm": 1.9415564319480298,
      "learning_rate": 8.344240837696335e-07,
      "loss": 0.0649,
      "step": 510
    },
    {
      "epoch": 0.008508549455943712,
      "grad_norm": 2.553942139990562,
      "learning_rate": 8.507853403141363e-07,
      "loss": 0.0846,
      "step": 520
    },
    {
      "epoch": 0.008672175407019552,
      "grad_norm": 2.9189597974150416,
      "learning_rate": 8.671465968586389e-07,
      "loss": 0.0734,
      "step": 530
    },
    {
      "epoch": 0.008835801358095394,
      "grad_norm": 1.4914881504164346,
      "learning_rate": 8.835078534031415e-07,
      "loss": 0.0654,
      "step": 540
    },
    {
      "epoch": 0.008999427309171235,
      "grad_norm": 2.0937713541818836,
      "learning_rate": 8.998691099476441e-07,
      "loss": 0.0584,
      "step": 550
    },
    {
      "epoch": 0.009163053260247075,
      "grad_norm": 2.0007188467312402,
      "learning_rate": 9.162303664921466e-07,
      "loss": 0.0634,
      "step": 560
    },
    {
      "epoch": 0.009326679211322917,
      "grad_norm": 2.5295589149279887,
      "learning_rate": 9.325916230366493e-07,
      "loss": 0.0583,
      "step": 570
    },
    {
      "epoch": 0.009490305162398757,
      "grad_norm": 1.7551442995086026,
      "learning_rate": 9.489528795811518e-07,
      "loss": 0.0715,
      "step": 580
    },
    {
      "epoch": 0.009653931113474597,
      "grad_norm": 2.092845868494887,
      "learning_rate": 9.653141361256545e-07,
      "loss": 0.0611,
      "step": 590
    },
    {
      "epoch": 0.009817557064550437,
      "grad_norm": 2.01934143420376,
      "learning_rate": 9.816753926701572e-07,
      "loss": 0.0541,
      "step": 600
    },
    {
      "epoch": 0.00998118301562628,
      "grad_norm": 1.8566512307485132,
      "learning_rate": 9.980366492146598e-07,
      "loss": 0.0571,
      "step": 610
    },
    {
      "epoch": 0.01014480896670212,
      "grad_norm": 2.5322222748702763,
      "learning_rate": 1.0143979057591625e-06,
      "loss": 0.0573,
      "step": 620
    },
    {
      "epoch": 0.01030843491777796,
      "grad_norm": 2.3213304916444684,
      "learning_rate": 1.030759162303665e-06,
      "loss": 0.0524,
      "step": 630
    },
    {
      "epoch": 0.0104720608688538,
      "grad_norm": 1.7847020518152077,
      "learning_rate": 1.0471204188481676e-06,
      "loss": 0.0693,
      "step": 640
    },
    {
      "epoch": 0.010635686819929642,
      "grad_norm": 1.7525613602373722,
      "learning_rate": 1.0634816753926703e-06,
      "loss": 0.0459,
      "step": 650
    },
    {
      "epoch": 0.010799312771005482,
      "grad_norm": 1.5243208103372825,
      "learning_rate": 1.079842931937173e-06,
      "loss": 0.0512,
      "step": 660
    },
    {
      "epoch": 0.010962938722081322,
      "grad_norm": 2.3214197206038096,
      "learning_rate": 1.0962041884816754e-06,
      "loss": 0.0579,
      "step": 670
    },
    {
      "epoch": 0.011126564673157162,
      "grad_norm": 1.711882984179115,
      "learning_rate": 1.1125654450261783e-06,
      "loss": 0.0593,
      "step": 680
    },
    {
      "epoch": 0.011290190624233004,
      "grad_norm": 1.363766563612011,
      "learning_rate": 1.1289267015706807e-06,
      "loss": 0.0518,
      "step": 690
    },
    {
      "epoch": 0.011453816575308844,
      "grad_norm": 1.6504973222238681,
      "learning_rate": 1.1452879581151834e-06,
      "loss": 0.0534,
      "step": 700
    },
    {
      "epoch": 0.011617442526384684,
      "grad_norm": 1.6642029241468472,
      "learning_rate": 1.161649214659686e-06,
      "loss": 0.0493,
      "step": 710
    },
    {
      "epoch": 0.011781068477460525,
      "grad_norm": 1.7256103586706497,
      "learning_rate": 1.1780104712041885e-06,
      "loss": 0.0556,
      "step": 720
    },
    {
      "epoch": 0.011944694428536367,
      "grad_norm": 1.9415628815918726,
      "learning_rate": 1.1943717277486911e-06,
      "loss": 0.0474,
      "step": 730
    },
    {
      "epoch": 0.012108320379612207,
      "grad_norm": 2.1484877444490977,
      "learning_rate": 1.2107329842931938e-06,
      "loss": 0.0537,
      "step": 740
    },
    {
      "epoch": 0.012271946330688047,
      "grad_norm": 1.5140596944350828,
      "learning_rate": 1.2270942408376965e-06,
      "loss": 0.0567,
      "step": 750
    },
    {
      "epoch": 0.012435572281763887,
      "grad_norm": 1.7353637859431932,
      "learning_rate": 1.2434554973821991e-06,
      "loss": 0.0511,
      "step": 760
    },
    {
      "epoch": 0.012599198232839729,
      "grad_norm": 1.7322299709323383,
      "learning_rate": 1.2598167539267018e-06,
      "loss": 0.0639,
      "step": 770
    },
    {
      "epoch": 0.01276282418391557,
      "grad_norm": 2.1990681288776117,
      "learning_rate": 1.2761780104712042e-06,
      "loss": 0.049,
      "step": 780
    },
    {
      "epoch": 0.01292645013499141,
      "grad_norm": 1.5454137661378569,
      "learning_rate": 1.292539267015707e-06,
      "loss": 0.0599,
      "step": 790
    },
    {
      "epoch": 0.01309007608606725,
      "grad_norm": 2.0334949781020826,
      "learning_rate": 1.3089005235602096e-06,
      "loss": 0.0532,
      "step": 800
    },
    {
      "epoch": 0.013253702037143092,
      "grad_norm": 1.3285573275282347,
      "learning_rate": 1.3252617801047122e-06,
      "loss": 0.0396,
      "step": 810
    },
    {
      "epoch": 0.013417327988218932,
      "grad_norm": 1.6130715610408146,
      "learning_rate": 1.3416230366492147e-06,
      "loss": 0.0558,
      "step": 820
    },
    {
      "epoch": 0.013580953939294772,
      "grad_norm": 2.7581875193798866,
      "learning_rate": 1.3579842931937173e-06,
      "loss": 0.0501,
      "step": 830
    },
    {
      "epoch": 0.013744579890370612,
      "grad_norm": 1.499495311158914,
      "learning_rate": 1.37434554973822e-06,
      "loss": 0.0502,
      "step": 840
    },
    {
      "epoch": 0.013908205841446454,
      "grad_norm": 2.3264096876684013,
      "learning_rate": 1.3907068062827224e-06,
      "loss": 0.043,
      "step": 850
    },
    {
      "epoch": 0.014071831792522294,
      "grad_norm": 1.6531908451109447,
      "learning_rate": 1.407068062827225e-06,
      "loss": 0.0449,
      "step": 860
    },
    {
      "epoch": 0.014235457743598134,
      "grad_norm": 1.788048347180086,
      "learning_rate": 1.4234293193717278e-06,
      "loss": 0.0427,
      "step": 870
    },
    {
      "epoch": 0.014399083694673975,
      "grad_norm": 2.3014140344107545,
      "learning_rate": 1.4397905759162306e-06,
      "loss": 0.0527,
      "step": 880
    },
    {
      "epoch": 0.014562709645749817,
      "grad_norm": 1.4825497780752035,
      "learning_rate": 1.4561518324607333e-06,
      "loss": 0.0354,
      "step": 890
    },
    {
      "epoch": 0.014726335596825657,
      "grad_norm": 1.8486967565995738,
      "learning_rate": 1.4725130890052358e-06,
      "loss": 0.0458,
      "step": 900
    },
    {
      "epoch": 0.014889961547901497,
      "grad_norm": 1.2359362604595727,
      "learning_rate": 1.4888743455497384e-06,
      "loss": 0.037,
      "step": 910
    },
    {
      "epoch": 0.015053587498977337,
      "grad_norm": 1.6318813705255526,
      "learning_rate": 1.505235602094241e-06,
      "loss": 0.0394,
      "step": 920
    },
    {
      "epoch": 0.015217213450053179,
      "grad_norm": 1.022241620000326,
      "learning_rate": 1.5215968586387435e-06,
      "loss": 0.0408,
      "step": 930
    },
    {
      "epoch": 0.01538083940112902,
      "grad_norm": 1.5334145926324694,
      "learning_rate": 1.5379581151832462e-06,
      "loss": 0.0403,
      "step": 940
    },
    {
      "epoch": 0.01554446535220486,
      "grad_norm": 1.4883351942486145,
      "learning_rate": 1.5543193717277488e-06,
      "loss": 0.0433,
      "step": 950
    },
    {
      "epoch": 0.0157080913032807,
      "grad_norm": 1.435001451856834,
      "learning_rate": 1.5706806282722515e-06,
      "loss": 0.0447,
      "step": 960
    },
    {
      "epoch": 0.01587171725435654,
      "grad_norm": 1.973302151367163,
      "learning_rate": 1.587041884816754e-06,
      "loss": 0.0295,
      "step": 970
    },
    {
      "epoch": 0.01603534320543238,
      "grad_norm": 1.133924636051525,
      "learning_rate": 1.6034031413612566e-06,
      "loss": 0.0331,
      "step": 980
    },
    {
      "epoch": 0.016198969156508224,
      "grad_norm": 1.4293649033089593,
      "learning_rate": 1.6197643979057593e-06,
      "loss": 0.0532,
      "step": 990
    },
    {
      "epoch": 0.016362595107584064,
      "grad_norm": 1.5182992486580937,
      "learning_rate": 1.6361256544502617e-06,
      "loss": 0.0296,
      "step": 1000
    },
    {
      "epoch": 0.016526221058659904,
      "grad_norm": 1.9783123474244562,
      "learning_rate": 1.6524869109947644e-06,
      "loss": 0.0378,
      "step": 1010
    },
    {
      "epoch": 0.016689847009735744,
      "grad_norm": 1.3249556318465352,
      "learning_rate": 1.668848167539267e-06,
      "loss": 0.0395,
      "step": 1020
    },
    {
      "epoch": 0.016853472960811584,
      "grad_norm": 1.5923718457847904,
      "learning_rate": 1.6852094240837697e-06,
      "loss": 0.042,
      "step": 1030
    },
    {
      "epoch": 0.017017098911887425,
      "grad_norm": 1.736176595423258,
      "learning_rate": 1.7015706806282726e-06,
      "loss": 0.0425,
      "step": 1040
    },
    {
      "epoch": 0.017180724862963265,
      "grad_norm": 1.6187886173240689,
      "learning_rate": 1.717931937172775e-06,
      "loss": 0.0463,
      "step": 1050
    },
    {
      "epoch": 0.017344350814039105,
      "grad_norm": 1.8876915670056627,
      "learning_rate": 1.7342931937172777e-06,
      "loss": 0.0363,
      "step": 1060
    },
    {
      "epoch": 0.01750797676511495,
      "grad_norm": 1.621982098427535,
      "learning_rate": 1.7506544502617804e-06,
      "loss": 0.0455,
      "step": 1070
    },
    {
      "epoch": 0.01767160271619079,
      "grad_norm": 1.7261554268835486,
      "learning_rate": 1.767015706806283e-06,
      "loss": 0.0377,
      "step": 1080
    },
    {
      "epoch": 0.01783522866726663,
      "grad_norm": 2.0942676313796595,
      "learning_rate": 1.7833769633507855e-06,
      "loss": 0.0369,
      "step": 1090
    },
    {
      "epoch": 0.01799885461834247,
      "grad_norm": 1.2448446537252458,
      "learning_rate": 1.7997382198952881e-06,
      "loss": 0.0471,
      "step": 1100
    },
    {
      "epoch": 0.01816248056941831,
      "grad_norm": 2.5211330945802812,
      "learning_rate": 1.8160994764397908e-06,
      "loss": 0.0316,
      "step": 1110
    },
    {
      "epoch": 0.01832610652049415,
      "grad_norm": 1.1116051684387043,
      "learning_rate": 1.8324607329842933e-06,
      "loss": 0.0301,
      "step": 1120
    },
    {
      "epoch": 0.01848973247156999,
      "grad_norm": 2.052012791543214,
      "learning_rate": 1.848821989528796e-06,
      "loss": 0.0411,
      "step": 1130
    },
    {
      "epoch": 0.018653358422645833,
      "grad_norm": 1.1851667145655387,
      "learning_rate": 1.8651832460732986e-06,
      "loss": 0.0292,
      "step": 1140
    },
    {
      "epoch": 0.018816984373721674,
      "grad_norm": 1.8324052281712595,
      "learning_rate": 1.8815445026178012e-06,
      "loss": 0.0402,
      "step": 1150
    },
    {
      "epoch": 0.018980610324797514,
      "grad_norm": 1.1427010839942016,
      "learning_rate": 1.8979057591623037e-06,
      "loss": 0.0293,
      "step": 1160
    },
    {
      "epoch": 0.019144236275873354,
      "grad_norm": 1.4379419888144533,
      "learning_rate": 1.914267015706806e-06,
      "loss": 0.0288,
      "step": 1170
    },
    {
      "epoch": 0.019307862226949194,
      "grad_norm": 1.2867665085952575,
      "learning_rate": 1.930628272251309e-06,
      "loss": 0.031,
      "step": 1180
    },
    {
      "epoch": 0.019471488178025034,
      "grad_norm": 1.375000512956942,
      "learning_rate": 1.9469895287958115e-06,
      "loss": 0.026,
      "step": 1190
    },
    {
      "epoch": 0.019635114129100874,
      "grad_norm": 1.5209358343311332,
      "learning_rate": 1.9633507853403143e-06,
      "loss": 0.0447,
      "step": 1200
    },
    {
      "epoch": 0.019798740080176715,
      "grad_norm": 1.6955535802183264,
      "learning_rate": 1.979712041884817e-06,
      "loss": 0.0357,
      "step": 1210
    },
    {
      "epoch": 0.01996236603125256,
      "grad_norm": 1.6039777164982636,
      "learning_rate": 1.9960732984293197e-06,
      "loss": 0.0313,
      "step": 1220
    },
    {
      "epoch": 0.0201259919823284,
      "grad_norm": 1.2200956375113978,
      "learning_rate": 2.012434554973822e-06,
      "loss": 0.0349,
      "step": 1230
    },
    {
      "epoch": 0.02028961793340424,
      "grad_norm": 1.6280353166583323,
      "learning_rate": 2.028795811518325e-06,
      "loss": 0.0337,
      "step": 1240
    },
    {
      "epoch": 0.02045324388448008,
      "grad_norm": 1.0386892459979415,
      "learning_rate": 2.0451570680628274e-06,
      "loss": 0.0336,
      "step": 1250
    },
    {
      "epoch": 0.02061686983555592,
      "grad_norm": 1.138067739041246,
      "learning_rate": 2.06151832460733e-06,
      "loss": 0.037,
      "step": 1260
    },
    {
      "epoch": 0.02078049578663176,
      "grad_norm": 2.140608023965435,
      "learning_rate": 2.0778795811518328e-06,
      "loss": 0.0335,
      "step": 1270
    },
    {
      "epoch": 0.0209441217377076,
      "grad_norm": 0.8186344301970785,
      "learning_rate": 2.094240837696335e-06,
      "loss": 0.0316,
      "step": 1280
    },
    {
      "epoch": 0.02110774768878344,
      "grad_norm": 1.2724399370759332,
      "learning_rate": 2.1106020942408377e-06,
      "loss": 0.0374,
      "step": 1290
    },
    {
      "epoch": 0.021271373639859283,
      "grad_norm": 1.873343507142681,
      "learning_rate": 2.1269633507853405e-06,
      "loss": 0.0341,
      "step": 1300
    },
    {
      "epoch": 0.021434999590935123,
      "grad_norm": 1.2812941090395522,
      "learning_rate": 2.143324607329843e-06,
      "loss": 0.0326,
      "step": 1310
    },
    {
      "epoch": 0.021598625542010964,
      "grad_norm": 1.822516644916563,
      "learning_rate": 2.159685863874346e-06,
      "loss": 0.0356,
      "step": 1320
    },
    {
      "epoch": 0.021762251493086804,
      "grad_norm": 1.287142695522933,
      "learning_rate": 2.1760471204188483e-06,
      "loss": 0.0254,
      "step": 1330
    },
    {
      "epoch": 0.021925877444162644,
      "grad_norm": 1.7091995874390806,
      "learning_rate": 2.1924083769633508e-06,
      "loss": 0.0285,
      "step": 1340
    },
    {
      "epoch": 0.022089503395238484,
      "grad_norm": 1.1059669021786156,
      "learning_rate": 2.2087696335078536e-06,
      "loss": 0.0354,
      "step": 1350
    },
    {
      "epoch": 0.022253129346314324,
      "grad_norm": 1.2809574737691118,
      "learning_rate": 2.2251308900523565e-06,
      "loss": 0.0338,
      "step": 1360
    },
    {
      "epoch": 0.022416755297390165,
      "grad_norm": 1.8488392242066378,
      "learning_rate": 2.241492146596859e-06,
      "loss": 0.04,
      "step": 1370
    },
    {
      "epoch": 0.022580381248466008,
      "grad_norm": 1.4938199951428712,
      "learning_rate": 2.2578534031413614e-06,
      "loss": 0.0364,
      "step": 1380
    },
    {
      "epoch": 0.02274400719954185,
      "grad_norm": 1.0027154187115255,
      "learning_rate": 2.2742146596858643e-06,
      "loss": 0.0297,
      "step": 1390
    },
    {
      "epoch": 0.02290763315061769,
      "grad_norm": 1.8665041620870224,
      "learning_rate": 2.2905759162303667e-06,
      "loss": 0.033,
      "step": 1400
    },
    {
      "epoch": 0.02307125910169353,
      "grad_norm": 1.322884025427935,
      "learning_rate": 2.306937172774869e-06,
      "loss": 0.0362,
      "step": 1410
    },
    {
      "epoch": 0.02323488505276937,
      "grad_norm": 1.2034845915472814,
      "learning_rate": 2.323298429319372e-06,
      "loss": 0.0353,
      "step": 1420
    },
    {
      "epoch": 0.02339851100384521,
      "grad_norm": 1.4191131020941135,
      "learning_rate": 2.3396596858638745e-06,
      "loss": 0.0319,
      "step": 1430
    },
    {
      "epoch": 0.02356213695492105,
      "grad_norm": 1.4678436302993285,
      "learning_rate": 2.356020942408377e-06,
      "loss": 0.037,
      "step": 1440
    },
    {
      "epoch": 0.02372576290599689,
      "grad_norm": 1.629487109182382,
      "learning_rate": 2.37238219895288e-06,
      "loss": 0.0261,
      "step": 1450
    },
    {
      "epoch": 0.023889388857072733,
      "grad_norm": 1.0508390211928016,
      "learning_rate": 2.3887434554973823e-06,
      "loss": 0.0333,
      "step": 1460
    },
    {
      "epoch": 0.024053014808148573,
      "grad_norm": 1.651355652791301,
      "learning_rate": 2.405104712041885e-06,
      "loss": 0.0234,
      "step": 1470
    },
    {
      "epoch": 0.024216640759224414,
      "grad_norm": 1.389819287624273,
      "learning_rate": 2.4214659685863876e-06,
      "loss": 0.0319,
      "step": 1480
    },
    {
      "epoch": 0.024380266710300254,
      "grad_norm": 1.031793512867473,
      "learning_rate": 2.43782722513089e-06,
      "loss": 0.026,
      "step": 1490
    },
    {
      "epoch": 0.024543892661376094,
      "grad_norm": 1.2337614675854987,
      "learning_rate": 2.454188481675393e-06,
      "loss": 0.0282,
      "step": 1500
    },
    {
      "epoch": 0.024707518612451934,
      "grad_norm": 2.3099320330721334,
      "learning_rate": 2.4705497382198954e-06,
      "loss": 0.0267,
      "step": 1510
    },
    {
      "epoch": 0.024871144563527774,
      "grad_norm": 1.6455559766834573,
      "learning_rate": 2.4869109947643982e-06,
      "loss": 0.0329,
      "step": 1520
    },
    {
      "epoch": 0.025034770514603615,
      "grad_norm": 1.3886755558928074,
      "learning_rate": 2.5032722513089007e-06,
      "loss": 0.0383,
      "step": 1530
    },
    {
      "epoch": 0.025198396465679458,
      "grad_norm": 0.9877668054313401,
      "learning_rate": 2.5196335078534036e-06,
      "loss": 0.039,
      "step": 1540
    },
    {
      "epoch": 0.0253620224167553,
      "grad_norm": 1.4201776283317427,
      "learning_rate": 2.535994764397906e-06,
      "loss": 0.0316,
      "step": 1550
    },
    {
      "epoch": 0.02552564836783114,
      "grad_norm": 1.938467779668822,
      "learning_rate": 2.5523560209424085e-06,
      "loss": 0.0278,
      "step": 1560
    },
    {
      "epoch": 0.02568927431890698,
      "grad_norm": 1.0006405696606735,
      "learning_rate": 2.5687172774869113e-06,
      "loss": 0.026,
      "step": 1570
    },
    {
      "epoch": 0.02585290026998282,
      "grad_norm": 1.3008512359014655,
      "learning_rate": 2.585078534031414e-06,
      "loss": 0.0259,
      "step": 1580
    },
    {
      "epoch": 0.02601652622105866,
      "grad_norm": 0.9483672511107769,
      "learning_rate": 2.6014397905759162e-06,
      "loss": 0.0329,
      "step": 1590
    },
    {
      "epoch": 0.0261801521721345,
      "grad_norm": 0.851663514207136,
      "learning_rate": 2.617801047120419e-06,
      "loss": 0.0315,
      "step": 1600
    },
    {
      "epoch": 0.02634377812321034,
      "grad_norm": 1.1998289382412684,
      "learning_rate": 2.6341623036649216e-06,
      "loss": 0.029,
      "step": 1610
    },
    {
      "epoch": 0.026507404074286183,
      "grad_norm": 0.9444444251959695,
      "learning_rate": 2.6505235602094244e-06,
      "loss": 0.0263,
      "step": 1620
    },
    {
      "epoch": 0.026671030025362023,
      "grad_norm": 1.388145696381342,
      "learning_rate": 2.666884816753927e-06,
      "loss": 0.0202,
      "step": 1630
    },
    {
      "epoch": 0.026834655976437864,
      "grad_norm": 1.2659599943687638,
      "learning_rate": 2.6832460732984293e-06,
      "loss": 0.027,
      "step": 1640
    },
    {
      "epoch": 0.026998281927513704,
      "grad_norm": 1.8988077488556223,
      "learning_rate": 2.6996073298429322e-06,
      "loss": 0.0328,
      "step": 1650
    },
    {
      "epoch": 0.027161907878589544,
      "grad_norm": 1.579560540759702,
      "learning_rate": 2.7159685863874347e-06,
      "loss": 0.0307,
      "step": 1660
    },
    {
      "epoch": 0.027325533829665384,
      "grad_norm": 1.2309956306342076,
      "learning_rate": 2.732329842931937e-06,
      "loss": 0.0263,
      "step": 1670
    },
    {
      "epoch": 0.027489159780741224,
      "grad_norm": 1.3690398995611774,
      "learning_rate": 2.74869109947644e-06,
      "loss": 0.0309,
      "step": 1680
    },
    {
      "epoch": 0.027652785731817068,
      "grad_norm": 1.252861999046074,
      "learning_rate": 2.7650523560209424e-06,
      "loss": 0.0283,
      "step": 1690
    },
    {
      "epoch": 0.027816411682892908,
      "grad_norm": 1.0414796664294799,
      "learning_rate": 2.781413612565445e-06,
      "loss": 0.0303,
      "step": 1700
    },
    {
      "epoch": 0.02798003763396875,
      "grad_norm": 1.32699206756568,
      "learning_rate": 2.7977748691099478e-06,
      "loss": 0.0303,
      "step": 1710
    },
    {
      "epoch": 0.02814366358504459,
      "grad_norm": 1.070063583830968,
      "learning_rate": 2.81413612565445e-06,
      "loss": 0.0347,
      "step": 1720
    },
    {
      "epoch": 0.02830728953612043,
      "grad_norm": 1.0632958492332283,
      "learning_rate": 2.8304973821989527e-06,
      "loss": 0.0234,
      "step": 1730
    },
    {
      "epoch": 0.02847091548719627,
      "grad_norm": 0.9307340835086102,
      "learning_rate": 2.8468586387434555e-06,
      "loss": 0.0224,
      "step": 1740
    },
    {
      "epoch": 0.02863454143827211,
      "grad_norm": 1.7396266633234245,
      "learning_rate": 2.863219895287958e-06,
      "loss": 0.0232,
      "step": 1750
    },
    {
      "epoch": 0.02879816738934795,
      "grad_norm": 1.108684751334364,
      "learning_rate": 2.8795811518324613e-06,
      "loss": 0.0282,
      "step": 1760
    },
    {
      "epoch": 0.028961793340423793,
      "grad_norm": 1.4441168573409573,
      "learning_rate": 2.8959424083769637e-06,
      "loss": 0.0356,
      "step": 1770
    },
    {
      "epoch": 0.029125419291499633,
      "grad_norm": 1.2174355622706727,
      "learning_rate": 2.9123036649214666e-06,
      "loss": 0.0289,
      "step": 1780
    },
    {
      "epoch": 0.029289045242575473,
      "grad_norm": 1.433182442333968,
      "learning_rate": 2.928664921465969e-06,
      "loss": 0.0254,
      "step": 1790
    },
    {
      "epoch": 0.029452671193651313,
      "grad_norm": 1.0075645389804508,
      "learning_rate": 2.9450261780104715e-06,
      "loss": 0.0275,
      "step": 1800
    },
    {
      "epoch": 0.029616297144727154,
      "grad_norm": 0.9231958449859682,
      "learning_rate": 2.9613874345549744e-06,
      "loss": 0.0157,
      "step": 1810
    },
    {
      "epoch": 0.029779923095802994,
      "grad_norm": 1.3921739289548163,
      "learning_rate": 2.977748691099477e-06,
      "loss": 0.019,
      "step": 1820
    },
    {
      "epoch": 0.029943549046878834,
      "grad_norm": 1.2248008683896874,
      "learning_rate": 2.9941099476439793e-06,
      "loss": 0.0329,
      "step": 1830
    },
    {
      "epoch": 0.030107174997954674,
      "grad_norm": 1.7717617713821525,
      "learning_rate": 3.010471204188482e-06,
      "loss": 0.0235,
      "step": 1840
    },
    {
      "epoch": 0.030270800949030518,
      "grad_norm": 1.0141322764986376,
      "learning_rate": 3.0268324607329846e-06,
      "loss": 0.0286,
      "step": 1850
    },
    {
      "epoch": 0.030434426900106358,
      "grad_norm": 1.477295340198986,
      "learning_rate": 3.043193717277487e-06,
      "loss": 0.0261,
      "step": 1860
    },
    {
      "epoch": 0.030598052851182198,
      "grad_norm": 0.9760848471935755,
      "learning_rate": 3.05955497382199e-06,
      "loss": 0.0359,
      "step": 1870
    },
    {
      "epoch": 0.03076167880225804,
      "grad_norm": 1.2699769943228976,
      "learning_rate": 3.0759162303664924e-06,
      "loss": 0.0323,
      "step": 1880
    },
    {
      "epoch": 0.03092530475333388,
      "grad_norm": 1.091415157395605,
      "learning_rate": 3.0922774869109952e-06,
      "loss": 0.0247,
      "step": 1890
    },
    {
      "epoch": 0.03108893070440972,
      "grad_norm": 1.1175742604918077,
      "learning_rate": 3.1086387434554977e-06,
      "loss": 0.0247,
      "step": 1900
    },
    {
      "epoch": 0.03125255665548556,
      "grad_norm": 1.194345090346912,
      "learning_rate": 3.125e-06,
      "loss": 0.0269,
      "step": 1910
    },
    {
      "epoch": 0.0314161826065614,
      "grad_norm": 1.2676547681009906,
      "learning_rate": 3.141361256544503e-06,
      "loss": 0.0244,
      "step": 1920
    },
    {
      "epoch": 0.03157980855763724,
      "grad_norm": 1.4613241327486248,
      "learning_rate": 3.1577225130890055e-06,
      "loss": 0.0244,
      "step": 1930
    },
    {
      "epoch": 0.03174343450871308,
      "grad_norm": 1.1286075738157584,
      "learning_rate": 3.174083769633508e-06,
      "loss": 0.0237,
      "step": 1940
    },
    {
      "epoch": 0.03190706045978892,
      "grad_norm": 1.212801782645968,
      "learning_rate": 3.190445026178011e-06,
      "loss": 0.019,
      "step": 1950
    },
    {
      "epoch": 0.03207068641086476,
      "grad_norm": 1.1515000735963563,
      "learning_rate": 3.2068062827225132e-06,
      "loss": 0.0276,
      "step": 1960
    },
    {
      "epoch": 0.032234312361940604,
      "grad_norm": 0.8907100707103518,
      "learning_rate": 3.2231675392670157e-06,
      "loss": 0.0198,
      "step": 1970
    },
    {
      "epoch": 0.03239793831301645,
      "grad_norm": 0.9052681867045012,
      "learning_rate": 3.2395287958115186e-06,
      "loss": 0.0292,
      "step": 1980
    },
    {
      "epoch": 0.032561564264092284,
      "grad_norm": 1.353266340697659,
      "learning_rate": 3.255890052356021e-06,
      "loss": 0.0262,
      "step": 1990
    },
    {
      "epoch": 0.03272519021516813,
      "grad_norm": 1.119087184019235,
      "learning_rate": 3.2722513089005235e-06,
      "loss": 0.0255,
      "step": 2000
    },
    {
      "epoch": 0.032888816166243964,
      "grad_norm": 1.371700303957014,
      "learning_rate": 3.2886125654450263e-06,
      "loss": 0.0293,
      "step": 2010
    },
    {
      "epoch": 0.03305244211731981,
      "grad_norm": 0.8600685242121404,
      "learning_rate": 3.304973821989529e-06,
      "loss": 0.029,
      "step": 2020
    },
    {
      "epoch": 0.033216068068395645,
      "grad_norm": 0.995883929435826,
      "learning_rate": 3.3213350785340317e-06,
      "loss": 0.0204,
      "step": 2030
    },
    {
      "epoch": 0.03337969401947149,
      "grad_norm": 1.2753040167006517,
      "learning_rate": 3.337696335078534e-06,
      "loss": 0.0217,
      "step": 2040
    },
    {
      "epoch": 0.03354331997054733,
      "grad_norm": 1.1745927493187636,
      "learning_rate": 3.3540575916230366e-06,
      "loss": 0.0281,
      "step": 2050
    },
    {
      "epoch": 0.03370694592162317,
      "grad_norm": 1.0801097577386816,
      "learning_rate": 3.3704188481675394e-06,
      "loss": 0.0203,
      "step": 2060
    },
    {
      "epoch": 0.03387057187269901,
      "grad_norm": 0.8639756669060888,
      "learning_rate": 3.3867801047120423e-06,
      "loss": 0.0221,
      "step": 2070
    },
    {
      "epoch": 0.03403419782377485,
      "grad_norm": 1.0688842375212482,
      "learning_rate": 3.403141361256545e-06,
      "loss": 0.0209,
      "step": 2080
    },
    {
      "epoch": 0.03419782377485069,
      "grad_norm": 0.9278469578122404,
      "learning_rate": 3.4195026178010476e-06,
      "loss": 0.0209,
      "step": 2090
    },
    {
      "epoch": 0.03436144972592653,
      "grad_norm": 1.028075410668248,
      "learning_rate": 3.43586387434555e-06,
      "loss": 0.0221,
      "step": 2100
    },
    {
      "epoch": 0.03452507567700237,
      "grad_norm": 0.89901911207049,
      "learning_rate": 3.452225130890053e-06,
      "loss": 0.0393,
      "step": 2110
    },
    {
      "epoch": 0.03468870162807821,
      "grad_norm": 1.2591295151007087,
      "learning_rate": 3.4685863874345554e-06,
      "loss": 0.0217,
      "step": 2120
    },
    {
      "epoch": 0.034852327579154053,
      "grad_norm": 1.0461311343194284,
      "learning_rate": 3.484947643979058e-06,
      "loss": 0.0265,
      "step": 2130
    },
    {
      "epoch": 0.0350159535302299,
      "grad_norm": 0.6788058785810165,
      "learning_rate": 3.5013089005235607e-06,
      "loss": 0.0187,
      "step": 2140
    },
    {
      "epoch": 0.035179579481305734,
      "grad_norm": 0.6544390150635634,
      "learning_rate": 3.517670157068063e-06,
      "loss": 0.0295,
      "step": 2150
    },
    {
      "epoch": 0.03534320543238158,
      "grad_norm": 0.8478595266630747,
      "learning_rate": 3.534031413612566e-06,
      "loss": 0.0228,
      "step": 2160
    },
    {
      "epoch": 0.035506831383457414,
      "grad_norm": 0.9519750502818294,
      "learning_rate": 3.5503926701570685e-06,
      "loss": 0.021,
      "step": 2170
    },
    {
      "epoch": 0.03567045733453326,
      "grad_norm": 1.0021700498009167,
      "learning_rate": 3.566753926701571e-06,
      "loss": 0.0247,
      "step": 2180
    },
    {
      "epoch": 0.035834083285609095,
      "grad_norm": 1.7287560435490075,
      "learning_rate": 3.583115183246074e-06,
      "loss": 0.023,
      "step": 2190
    },
    {
      "epoch": 0.03599770923668494,
      "grad_norm": 1.2258058905009668,
      "learning_rate": 3.5994764397905763e-06,
      "loss": 0.0208,
      "step": 2200
    },
    {
      "epoch": 0.03616133518776078,
      "grad_norm": 1.0962938745476274,
      "learning_rate": 3.6158376963350787e-06,
      "loss": 0.0276,
      "step": 2210
    },
    {
      "epoch": 0.03632496113883662,
      "grad_norm": 1.7774300996973011,
      "learning_rate": 3.6321989528795816e-06,
      "loss": 0.0227,
      "step": 2220
    },
    {
      "epoch": 0.03648858708991246,
      "grad_norm": 1.1875072319659314,
      "learning_rate": 3.648560209424084e-06,
      "loss": 0.0288,
      "step": 2230
    },
    {
      "epoch": 0.0366522130409883,
      "grad_norm": 1.301960747116591,
      "learning_rate": 3.6649214659685865e-06,
      "loss": 0.0191,
      "step": 2240
    },
    {
      "epoch": 0.03681583899206414,
      "grad_norm": 0.9252751995193765,
      "learning_rate": 3.6812827225130894e-06,
      "loss": 0.0224,
      "step": 2250
    },
    {
      "epoch": 0.03697946494313998,
      "grad_norm": 1.0411107735174379,
      "learning_rate": 3.697643979057592e-06,
      "loss": 0.02,
      "step": 2260
    },
    {
      "epoch": 0.03714309089421582,
      "grad_norm": 0.6979691382888245,
      "learning_rate": 3.7140052356020943e-06,
      "loss": 0.0209,
      "step": 2270
    },
    {
      "epoch": 0.03730671684529167,
      "grad_norm": 0.6002271976247916,
      "learning_rate": 3.730366492146597e-06,
      "loss": 0.0209,
      "step": 2280
    },
    {
      "epoch": 0.0374703427963675,
      "grad_norm": 1.1463934839259544,
      "learning_rate": 3.7467277486910996e-06,
      "loss": 0.0188,
      "step": 2290
    },
    {
      "epoch": 0.03763396874744335,
      "grad_norm": 1.1663084440395934,
      "learning_rate": 3.7630890052356025e-06,
      "loss": 0.0259,
      "step": 2300
    },
    {
      "epoch": 0.037797594698519184,
      "grad_norm": 0.780042784927533,
      "learning_rate": 3.779450261780105e-06,
      "loss": 0.0235,
      "step": 2310
    },
    {
      "epoch": 0.03796122064959503,
      "grad_norm": 0.7987987794113918,
      "learning_rate": 3.7958115183246074e-06,
      "loss": 0.0206,
      "step": 2320
    },
    {
      "epoch": 0.038124846600670864,
      "grad_norm": 1.2187625658449108,
      "learning_rate": 3.8121727748691103e-06,
      "loss": 0.0258,
      "step": 2330
    },
    {
      "epoch": 0.03828847255174671,
      "grad_norm": 0.978190971220372,
      "learning_rate": 3.828534031413612e-06,
      "loss": 0.0212,
      "step": 2340
    },
    {
      "epoch": 0.038452098502822545,
      "grad_norm": 1.1794199457130456,
      "learning_rate": 3.844895287958115e-06,
      "loss": 0.0221,
      "step": 2350
    },
    {
      "epoch": 0.03861572445389839,
      "grad_norm": 1.0122841297857827,
      "learning_rate": 3.861256544502618e-06,
      "loss": 0.0207,
      "step": 2360
    },
    {
      "epoch": 0.03877935040497423,
      "grad_norm": 0.4307882536507433,
      "learning_rate": 3.877617801047121e-06,
      "loss": 0.0195,
      "step": 2370
    },
    {
      "epoch": 0.03894297635605007,
      "grad_norm": 0.8577245239447127,
      "learning_rate": 3.893979057591623e-06,
      "loss": 0.0202,
      "step": 2380
    },
    {
      "epoch": 0.03910660230712591,
      "grad_norm": 1.285569904245558,
      "learning_rate": 3.910340314136127e-06,
      "loss": 0.018,
      "step": 2390
    },
    {
      "epoch": 0.03927022825820175,
      "grad_norm": 0.91926981178011,
      "learning_rate": 3.926701570680629e-06,
      "loss": 0.017,
      "step": 2400
    },
    {
      "epoch": 0.03943385420927759,
      "grad_norm": 0.9456854575392329,
      "learning_rate": 3.9430628272251315e-06,
      "loss": 0.0176,
      "step": 2410
    },
    {
      "epoch": 0.03959748016035343,
      "grad_norm": 0.9181370052112151,
      "learning_rate": 3.959424083769634e-06,
      "loss": 0.0185,
      "step": 2420
    },
    {
      "epoch": 0.03976110611142927,
      "grad_norm": 1.9430712253794025,
      "learning_rate": 3.9757853403141364e-06,
      "loss": 0.0223,
      "step": 2430
    },
    {
      "epoch": 0.03992473206250512,
      "grad_norm": 1.2290438101625654,
      "learning_rate": 3.992146596858639e-06,
      "loss": 0.0238,
      "step": 2440
    },
    {
      "epoch": 0.04008835801358095,
      "grad_norm": 1.3751276891527615,
      "learning_rate": 4.008507853403142e-06,
      "loss": 0.0247,
      "step": 2450
    },
    {
      "epoch": 0.0402519839646568,
      "grad_norm": 0.8705566094326588,
      "learning_rate": 4.024869109947644e-06,
      "loss": 0.0232,
      "step": 2460
    },
    {
      "epoch": 0.040415609915732634,
      "grad_norm": 1.1334238108261925,
      "learning_rate": 4.041230366492147e-06,
      "loss": 0.0165,
      "step": 2470
    },
    {
      "epoch": 0.04057923586680848,
      "grad_norm": 0.5273715131117723,
      "learning_rate": 4.05759162303665e-06,
      "loss": 0.0164,
      "step": 2480
    },
    {
      "epoch": 0.040742861817884314,
      "grad_norm": 1.2339543352831737,
      "learning_rate": 4.073952879581152e-06,
      "loss": 0.0234,
      "step": 2490
    },
    {
      "epoch": 0.04090648776896016,
      "grad_norm": 0.9815598904057271,
      "learning_rate": 4.090314136125655e-06,
      "loss": 0.0174,
      "step": 2500
    },
    {
      "epoch": 0.041070113720035994,
      "grad_norm": 0.6846770155227239,
      "learning_rate": 4.106675392670158e-06,
      "loss": 0.0228,
      "step": 2510
    },
    {
      "epoch": 0.04123373967111184,
      "grad_norm": 0.9392888212287435,
      "learning_rate": 4.12303664921466e-06,
      "loss": 0.0191,
      "step": 2520
    },
    {
      "epoch": 0.04139736562218768,
      "grad_norm": 0.93492176131368,
      "learning_rate": 4.139397905759163e-06,
      "loss": 0.0199,
      "step": 2530
    },
    {
      "epoch": 0.04156099157326352,
      "grad_norm": 0.9503545660768167,
      "learning_rate": 4.1557591623036655e-06,
      "loss": 0.0212,
      "step": 2540
    },
    {
      "epoch": 0.04172461752433936,
      "grad_norm": 0.9113455239717534,
      "learning_rate": 4.1721204188481675e-06,
      "loss": 0.0197,
      "step": 2550
    },
    {
      "epoch": 0.0418882434754152,
      "grad_norm": 1.004857597317951,
      "learning_rate": 4.18848167539267e-06,
      "loss": 0.0188,
      "step": 2560
    },
    {
      "epoch": 0.04205186942649104,
      "grad_norm": 0.9603709393996345,
      "learning_rate": 4.204842931937173e-06,
      "loss": 0.014,
      "step": 2570
    },
    {
      "epoch": 0.04221549537756688,
      "grad_norm": 1.04941880252467,
      "learning_rate": 4.221204188481675e-06,
      "loss": 0.013,
      "step": 2580
    },
    {
      "epoch": 0.04237912132864272,
      "grad_norm": 1.0081480317799187,
      "learning_rate": 4.237565445026178e-06,
      "loss": 0.0215,
      "step": 2590
    },
    {
      "epoch": 0.04254274727971857,
      "grad_norm": 0.3557005348366723,
      "learning_rate": 4.253926701570681e-06,
      "loss": 0.0154,
      "step": 2600
    },
    {
      "epoch": 0.0427063732307944,
      "grad_norm": 0.5848244053359664,
      "learning_rate": 4.270287958115183e-06,
      "loss": 0.0183,
      "step": 2610
    },
    {
      "epoch": 0.04286999918187025,
      "grad_norm": 1.7544872642593676,
      "learning_rate": 4.286649214659686e-06,
      "loss": 0.0157,
      "step": 2620
    },
    {
      "epoch": 0.043033625132946084,
      "grad_norm": 0.7859301770642318,
      "learning_rate": 4.303010471204189e-06,
      "loss": 0.017,
      "step": 2630
    },
    {
      "epoch": 0.04319725108402193,
      "grad_norm": 0.736964572960593,
      "learning_rate": 4.319371727748692e-06,
      "loss": 0.021,
      "step": 2640
    },
    {
      "epoch": 0.043360877035097764,
      "grad_norm": 1.002093025521673,
      "learning_rate": 4.335732984293194e-06,
      "loss": 0.0178,
      "step": 2650
    },
    {
      "epoch": 0.04352450298617361,
      "grad_norm": 0.6525711180647051,
      "learning_rate": 4.352094240837697e-06,
      "loss": 0.0224,
      "step": 2660
    },
    {
      "epoch": 0.043688128937249444,
      "grad_norm": 1.1254926151656834,
      "learning_rate": 4.3684554973821995e-06,
      "loss": 0.016,
      "step": 2670
    },
    {
      "epoch": 0.04385175488832529,
      "grad_norm": 1.0187827655949202,
      "learning_rate": 4.3848167539267015e-06,
      "loss": 0.0153,
      "step": 2680
    },
    {
      "epoch": 0.04401538083940113,
      "grad_norm": 0.678776487160841,
      "learning_rate": 4.401178010471204e-06,
      "loss": 0.0196,
      "step": 2690
    },
    {
      "epoch": 0.04417900679047697,
      "grad_norm": 0.7657909064193895,
      "learning_rate": 4.417539267015707e-06,
      "loss": 0.0193,
      "step": 2700
    },
    {
      "epoch": 0.04434263274155281,
      "grad_norm": 0.8317235520567553,
      "learning_rate": 4.43390052356021e-06,
      "loss": 0.0193,
      "step": 2710
    },
    {
      "epoch": 0.04450625869262865,
      "grad_norm": 0.7107851144867301,
      "learning_rate": 4.450261780104713e-06,
      "loss": 0.0178,
      "step": 2720
    },
    {
      "epoch": 0.04466988464370449,
      "grad_norm": 0.8717761490682288,
      "learning_rate": 4.466623036649215e-06,
      "loss": 0.0181,
      "step": 2730
    },
    {
      "epoch": 0.04483351059478033,
      "grad_norm": 0.7106786903981227,
      "learning_rate": 4.482984293193718e-06,
      "loss": 0.0164,
      "step": 2740
    },
    {
      "epoch": 0.04499713654585617,
      "grad_norm": 1.5117108534302517,
      "learning_rate": 4.499345549738221e-06,
      "loss": 0.0222,
      "step": 2750
    },
    {
      "epoch": 0.045160762496932016,
      "grad_norm": 0.8624823316930237,
      "learning_rate": 4.515706806282723e-06,
      "loss": 0.0147,
      "step": 2760
    },
    {
      "epoch": 0.04532438844800785,
      "grad_norm": 0.8444141579649685,
      "learning_rate": 4.532068062827226e-06,
      "loss": 0.0173,
      "step": 2770
    },
    {
      "epoch": 0.0454880143990837,
      "grad_norm": 1.470843747434806,
      "learning_rate": 4.5484293193717286e-06,
      "loss": 0.0262,
      "step": 2780
    },
    {
      "epoch": 0.045651640350159534,
      "grad_norm": 0.9429133659470207,
      "learning_rate": 4.564790575916231e-06,
      "loss": 0.0183,
      "step": 2790
    },
    {
      "epoch": 0.04581526630123538,
      "grad_norm": 0.6624127253568172,
      "learning_rate": 4.5811518324607335e-06,
      "loss": 0.0193,
      "step": 2800
    },
    {
      "epoch": 0.045978892252311214,
      "grad_norm": 0.640065560918503,
      "learning_rate": 4.597513089005236e-06,
      "loss": 0.0213,
      "step": 2810
    },
    {
      "epoch": 0.04614251820338706,
      "grad_norm": 1.391760039150221,
      "learning_rate": 4.613874345549738e-06,
      "loss": 0.0148,
      "step": 2820
    },
    {
      "epoch": 0.0463061441544629,
      "grad_norm": 0.956271615184497,
      "learning_rate": 4.630235602094241e-06,
      "loss": 0.0189,
      "step": 2830
    },
    {
      "epoch": 0.04646977010553874,
      "grad_norm": 0.9875201082203795,
      "learning_rate": 4.646596858638744e-06,
      "loss": 0.0187,
      "step": 2840
    },
    {
      "epoch": 0.04663339605661458,
      "grad_norm": 0.6684546748273962,
      "learning_rate": 4.662958115183246e-06,
      "loss": 0.0211,
      "step": 2850
    },
    {
      "epoch": 0.04679702200769042,
      "grad_norm": 0.5601253250399263,
      "learning_rate": 4.679319371727749e-06,
      "loss": 0.0184,
      "step": 2860
    },
    {
      "epoch": 0.04696064795876626,
      "grad_norm": 0.9026348378273591,
      "learning_rate": 4.695680628272252e-06,
      "loss": 0.0201,
      "step": 2870
    },
    {
      "epoch": 0.0471242739098421,
      "grad_norm": 1.2451921632432312,
      "learning_rate": 4.712041884816754e-06,
      "loss": 0.0247,
      "step": 2880
    },
    {
      "epoch": 0.04728789986091794,
      "grad_norm": 0.9593657838522912,
      "learning_rate": 4.728403141361257e-06,
      "loss": 0.0177,
      "step": 2890
    },
    {
      "epoch": 0.04745152581199378,
      "grad_norm": 0.5197250683014698,
      "learning_rate": 4.74476439790576e-06,
      "loss": 0.0206,
      "step": 2900
    },
    {
      "epoch": 0.04761515176306962,
      "grad_norm": 1.0204469510877177,
      "learning_rate": 4.761125654450262e-06,
      "loss": 0.0192,
      "step": 2910
    },
    {
      "epoch": 0.047778777714145466,
      "grad_norm": 1.324166237756431,
      "learning_rate": 4.7774869109947645e-06,
      "loss": 0.0219,
      "step": 2920
    },
    {
      "epoch": 0.0479424036652213,
      "grad_norm": 0.6491258940752783,
      "learning_rate": 4.793848167539267e-06,
      "loss": 0.0157,
      "step": 2930
    },
    {
      "epoch": 0.04810602961629715,
      "grad_norm": 0.5995781781748115,
      "learning_rate": 4.81020942408377e-06,
      "loss": 0.0204,
      "step": 2940
    },
    {
      "epoch": 0.048269655567372984,
      "grad_norm": 0.48262283939141637,
      "learning_rate": 4.826570680628272e-06,
      "loss": 0.0184,
      "step": 2950
    },
    {
      "epoch": 0.04843328151844883,
      "grad_norm": 0.7483504732182222,
      "learning_rate": 4.842931937172775e-06,
      "loss": 0.0163,
      "step": 2960
    },
    {
      "epoch": 0.048596907469524664,
      "grad_norm": 0.8894273497586248,
      "learning_rate": 4.859293193717278e-06,
      "loss": 0.0157,
      "step": 2970
    },
    {
      "epoch": 0.04876053342060051,
      "grad_norm": 0.9784640843077647,
      "learning_rate": 4.87565445026178e-06,
      "loss": 0.0175,
      "step": 2980
    },
    {
      "epoch": 0.04892415937167635,
      "grad_norm": 0.6643778201631815,
      "learning_rate": 4.892015706806283e-06,
      "loss": 0.0172,
      "step": 2990
    },
    {
      "epoch": 0.04908778532275219,
      "grad_norm": 1.0724792461767614,
      "learning_rate": 4.908376963350786e-06,
      "loss": 0.0158,
      "step": 3000
    },
    {
      "epoch": 0.04925141127382803,
      "grad_norm": 0.9780350713415569,
      "learning_rate": 4.924738219895288e-06,
      "loss": 0.0198,
      "step": 3010
    },
    {
      "epoch": 0.04941503722490387,
      "grad_norm": 0.6378674505205143,
      "learning_rate": 4.941099476439791e-06,
      "loss": 0.0164,
      "step": 3020
    },
    {
      "epoch": 0.04957866317597971,
      "grad_norm": 1.0658378022002295,
      "learning_rate": 4.957460732984294e-06,
      "loss": 0.0127,
      "step": 3030
    },
    {
      "epoch": 0.04974228912705555,
      "grad_norm": 0.4829379599897141,
      "learning_rate": 4.9738219895287965e-06,
      "loss": 0.0205,
      "step": 3040
    },
    {
      "epoch": 0.04990591507813139,
      "grad_norm": 0.9270058272193237,
      "learning_rate": 4.990183246073299e-06,
      "loss": 0.0189,
      "step": 3050
    },
    {
      "epoch": 0.05006954102920723,
      "grad_norm": 0.8175719574480865,
      "learning_rate": 5.006544502617801e-06,
      "loss": 0.0214,
      "step": 3060
    },
    {
      "epoch": 0.05023316698028307,
      "grad_norm": 0.7476158427134629,
      "learning_rate": 5.022905759162304e-06,
      "loss": 0.0145,
      "step": 3070
    },
    {
      "epoch": 0.050396792931358916,
      "grad_norm": 0.8211440572711844,
      "learning_rate": 5.039267015706807e-06,
      "loss": 0.0207,
      "step": 3080
    },
    {
      "epoch": 0.05056041888243475,
      "grad_norm": 1.237348947844268,
      "learning_rate": 5.055628272251309e-06,
      "loss": 0.0187,
      "step": 3090
    },
    {
      "epoch": 0.0507240448335106,
      "grad_norm": 0.6781940109031565,
      "learning_rate": 5.071989528795812e-06,
      "loss": 0.0162,
      "step": 3100
    },
    {
      "epoch": 0.05088767078458643,
      "grad_norm": 1.186412411200683,
      "learning_rate": 5.088350785340315e-06,
      "loss": 0.0154,
      "step": 3110
    },
    {
      "epoch": 0.05105129673566228,
      "grad_norm": 0.7883414029868824,
      "learning_rate": 5.104712041884817e-06,
      "loss": 0.0206,
      "step": 3120
    },
    {
      "epoch": 0.051214922686738114,
      "grad_norm": 0.7789990055792931,
      "learning_rate": 5.12107329842932e-06,
      "loss": 0.0165,
      "step": 3130
    },
    {
      "epoch": 0.05137854863781396,
      "grad_norm": 0.4692693724727117,
      "learning_rate": 5.137434554973823e-06,
      "loss": 0.0151,
      "step": 3140
    },
    {
      "epoch": 0.0515421745888898,
      "grad_norm": 0.9530164712691042,
      "learning_rate": 5.153795811518325e-06,
      "loss": 0.0167,
      "step": 3150
    },
    {
      "epoch": 0.05170580053996564,
      "grad_norm": 0.5835788141798162,
      "learning_rate": 5.170157068062828e-06,
      "loss": 0.0187,
      "step": 3160
    },
    {
      "epoch": 0.05186942649104148,
      "grad_norm": 0.6302619895215621,
      "learning_rate": 5.1865183246073305e-06,
      "loss": 0.0135,
      "step": 3170
    },
    {
      "epoch": 0.05203305244211732,
      "grad_norm": 0.6791271946923848,
      "learning_rate": 5.2028795811518325e-06,
      "loss": 0.0203,
      "step": 3180
    },
    {
      "epoch": 0.05219667839319316,
      "grad_norm": 0.781394547868449,
      "learning_rate": 5.219240837696335e-06,
      "loss": 0.0172,
      "step": 3190
    },
    {
      "epoch": 0.052360304344269,
      "grad_norm": 0.9222174635498976,
      "learning_rate": 5.235602094240838e-06,
      "loss": 0.0206,
      "step": 3200
    },
    {
      "epoch": 0.05252393029534484,
      "grad_norm": 0.6031369431049634,
      "learning_rate": 5.251963350785341e-06,
      "loss": 0.0183,
      "step": 3210
    },
    {
      "epoch": 0.05268755624642068,
      "grad_norm": 0.8242313809947889,
      "learning_rate": 5.268324607329843e-06,
      "loss": 0.0194,
      "step": 3220
    },
    {
      "epoch": 0.05285118219749652,
      "grad_norm": 0.9376408899698326,
      "learning_rate": 5.284685863874346e-06,
      "loss": 0.0178,
      "step": 3230
    },
    {
      "epoch": 0.053014808148572366,
      "grad_norm": 0.9334918594622488,
      "learning_rate": 5.301047120418849e-06,
      "loss": 0.0193,
      "step": 3240
    },
    {
      "epoch": 0.0531784340996482,
      "grad_norm": 0.9132822097471964,
      "learning_rate": 5.317408376963351e-06,
      "loss": 0.0163,
      "step": 3250
    },
    {
      "epoch": 0.05334206005072405,
      "grad_norm": 0.9594471109290825,
      "learning_rate": 5.333769633507854e-06,
      "loss": 0.018,
      "step": 3260
    },
    {
      "epoch": 0.05350568600179988,
      "grad_norm": 0.5053382889220102,
      "learning_rate": 5.350130890052357e-06,
      "loss": 0.0175,
      "step": 3270
    },
    {
      "epoch": 0.05366931195287573,
      "grad_norm": 0.36737552990579614,
      "learning_rate": 5.366492146596859e-06,
      "loss": 0.0168,
      "step": 3280
    },
    {
      "epoch": 0.053832937903951564,
      "grad_norm": 0.6303186079532742,
      "learning_rate": 5.3828534031413616e-06,
      "loss": 0.0214,
      "step": 3290
    },
    {
      "epoch": 0.05399656385502741,
      "grad_norm": 0.5951811317541013,
      "learning_rate": 5.3992146596858644e-06,
      "loss": 0.0174,
      "step": 3300
    },
    {
      "epoch": 0.05416018980610325,
      "grad_norm": 0.8921132202079008,
      "learning_rate": 5.4155759162303665e-06,
      "loss": 0.0144,
      "step": 3310
    },
    {
      "epoch": 0.05432381575717909,
      "grad_norm": 0.7604220871075399,
      "learning_rate": 5.431937172774869e-06,
      "loss": 0.0137,
      "step": 3320
    },
    {
      "epoch": 0.05448744170825493,
      "grad_norm": 0.5268241961569924,
      "learning_rate": 5.448298429319372e-06,
      "loss": 0.0141,
      "step": 3330
    },
    {
      "epoch": 0.05465106765933077,
      "grad_norm": 0.5298816285238115,
      "learning_rate": 5.464659685863874e-06,
      "loss": 0.0128,
      "step": 3340
    },
    {
      "epoch": 0.05481469361040661,
      "grad_norm": 0.4953869493948359,
      "learning_rate": 5.481020942408377e-06,
      "loss": 0.0149,
      "step": 3350
    },
    {
      "epoch": 0.05497831956148245,
      "grad_norm": 0.6835023394685612,
      "learning_rate": 5.49738219895288e-06,
      "loss": 0.0132,
      "step": 3360
    },
    {
      "epoch": 0.05514194551255829,
      "grad_norm": 0.7390804600196659,
      "learning_rate": 5.513743455497382e-06,
      "loss": 0.0135,
      "step": 3370
    },
    {
      "epoch": 0.055305571463634136,
      "grad_norm": 1.094216779550603,
      "learning_rate": 5.530104712041885e-06,
      "loss": 0.0247,
      "step": 3380
    },
    {
      "epoch": 0.05546919741470997,
      "grad_norm": 0.4765546557839919,
      "learning_rate": 5.546465968586388e-06,
      "loss": 0.0227,
      "step": 3390
    },
    {
      "epoch": 0.055632823365785816,
      "grad_norm": 0.4804789937460036,
      "learning_rate": 5.56282722513089e-06,
      "loss": 0.0126,
      "step": 3400
    },
    {
      "epoch": 0.05579644931686165,
      "grad_norm": 0.929847189596145,
      "learning_rate": 5.579188481675393e-06,
      "loss": 0.0176,
      "step": 3410
    },
    {
      "epoch": 0.0559600752679375,
      "grad_norm": 0.8139905504662587,
      "learning_rate": 5.5955497382198955e-06,
      "loss": 0.0182,
      "step": 3420
    },
    {
      "epoch": 0.05612370121901333,
      "grad_norm": 0.6658544045988827,
      "learning_rate": 5.6119109947643975e-06,
      "loss": 0.0144,
      "step": 3430
    },
    {
      "epoch": 0.05628732717008918,
      "grad_norm": 0.9705422411034668,
      "learning_rate": 5.6282722513089e-06,
      "loss": 0.0169,
      "step": 3440
    },
    {
      "epoch": 0.056450953121165014,
      "grad_norm": 1.1170219668224348,
      "learning_rate": 5.644633507853403e-06,
      "loss": 0.0185,
      "step": 3450
    },
    {
      "epoch": 0.05661457907224086,
      "grad_norm": 0.7846484169454016,
      "learning_rate": 5.660994764397905e-06,
      "loss": 0.0142,
      "step": 3460
    },
    {
      "epoch": 0.0567782050233167,
      "grad_norm": 0.6837852877664217,
      "learning_rate": 5.677356020942408e-06,
      "loss": 0.0184,
      "step": 3470
    },
    {
      "epoch": 0.05694183097439254,
      "grad_norm": 0.8774040343091403,
      "learning_rate": 5.693717277486911e-06,
      "loss": 0.0144,
      "step": 3480
    },
    {
      "epoch": 0.05710545692546838,
      "grad_norm": 0.9158868072008943,
      "learning_rate": 5.710078534031414e-06,
      "loss": 0.015,
      "step": 3490
    },
    {
      "epoch": 0.05726908287654422,
      "grad_norm": 1.1947732300157077,
      "learning_rate": 5.726439790575916e-06,
      "loss": 0.0135,
      "step": 3500
    },
    {
      "epoch": 0.05743270882762006,
      "grad_norm": 0.7019699246070533,
      "learning_rate": 5.74280104712042e-06,
      "loss": 0.0156,
      "step": 3510
    },
    {
      "epoch": 0.0575963347786959,
      "grad_norm": 0.6006783200064366,
      "learning_rate": 5.7591623036649226e-06,
      "loss": 0.0154,
      "step": 3520
    },
    {
      "epoch": 0.05775996072977174,
      "grad_norm": 0.866441297150528,
      "learning_rate": 5.7755235602094254e-06,
      "loss": 0.0175,
      "step": 3530
    },
    {
      "epoch": 0.057923586680847586,
      "grad_norm": 0.5628684828106116,
      "learning_rate": 5.7918848167539275e-06,
      "loss": 0.0191,
      "step": 3540
    },
    {
      "epoch": 0.05808721263192342,
      "grad_norm": 0.8675233050727733,
      "learning_rate": 5.80824607329843e-06,
      "loss": 0.0173,
      "step": 3550
    },
    {
      "epoch": 0.058250838582999266,
      "grad_norm": 0.9462581119266399,
      "learning_rate": 5.824607329842933e-06,
      "loss": 0.0151,
      "step": 3560
    },
    {
      "epoch": 0.0584144645340751,
      "grad_norm": 0.7358941798708127,
      "learning_rate": 5.840968586387435e-06,
      "loss": 0.0166,
      "step": 3570
    },
    {
      "epoch": 0.058578090485150947,
      "grad_norm": 0.8164273335486668,
      "learning_rate": 5.857329842931938e-06,
      "loss": 0.0171,
      "step": 3580
    },
    {
      "epoch": 0.05874171643622678,
      "grad_norm": 1.1188885583251384,
      "learning_rate": 5.873691099476441e-06,
      "loss": 0.0144,
      "step": 3590
    },
    {
      "epoch": 0.05890534238730263,
      "grad_norm": 0.8654726294460399,
      "learning_rate": 5.890052356020943e-06,
      "loss": 0.0167,
      "step": 3600
    },
    {
      "epoch": 0.059068968338378464,
      "grad_norm": 0.947715811080645,
      "learning_rate": 5.906413612565446e-06,
      "loss": 0.0175,
      "step": 3610
    },
    {
      "epoch": 0.05923259428945431,
      "grad_norm": 0.5313647974355867,
      "learning_rate": 5.922774869109949e-06,
      "loss": 0.0151,
      "step": 3620
    },
    {
      "epoch": 0.05939622024053015,
      "grad_norm": 1.4699055923617934,
      "learning_rate": 5.939136125654451e-06,
      "loss": 0.0213,
      "step": 3630
    },
    {
      "epoch": 0.05955984619160599,
      "grad_norm": 1.0686135179947462,
      "learning_rate": 5.955497382198954e-06,
      "loss": 0.018,
      "step": 3640
    },
    {
      "epoch": 0.05972347214268183,
      "grad_norm": 0.9300579713054162,
      "learning_rate": 5.9718586387434565e-06,
      "loss": 0.0146,
      "step": 3650
    },
    {
      "epoch": 0.05988709809375767,
      "grad_norm": 0.9990855688437372,
      "learning_rate": 5.9882198952879586e-06,
      "loss": 0.0191,
      "step": 3660
    },
    {
      "epoch": 0.06005072404483351,
      "grad_norm": 0.7334955017808976,
      "learning_rate": 6.0045811518324614e-06,
      "loss": 0.0176,
      "step": 3670
    },
    {
      "epoch": 0.06021434999590935,
      "grad_norm": 0.5378412672950716,
      "learning_rate": 6.020942408376964e-06,
      "loss": 0.0136,
      "step": 3680
    },
    {
      "epoch": 0.06037797594698519,
      "grad_norm": 0.773295481868631,
      "learning_rate": 6.037303664921466e-06,
      "loss": 0.0138,
      "step": 3690
    },
    {
      "epoch": 0.060541601898061036,
      "grad_norm": 0.785233922269194,
      "learning_rate": 6.053664921465969e-06,
      "loss": 0.0207,
      "step": 3700
    },
    {
      "epoch": 0.06070522784913687,
      "grad_norm": 0.8094438465472602,
      "learning_rate": 6.070026178010472e-06,
      "loss": 0.0104,
      "step": 3710
    },
    {
      "epoch": 0.060868853800212716,
      "grad_norm": 0.7507120237096889,
      "learning_rate": 6.086387434554974e-06,
      "loss": 0.0206,
      "step": 3720
    },
    {
      "epoch": 0.06103247975128855,
      "grad_norm": 0.6769979296618228,
      "learning_rate": 6.102748691099477e-06,
      "loss": 0.018,
      "step": 3730
    },
    {
      "epoch": 0.061196105702364396,
      "grad_norm": 0.8736448072676698,
      "learning_rate": 6.11910994764398e-06,
      "loss": 0.0161,
      "step": 3740
    },
    {
      "epoch": 0.06135973165344023,
      "grad_norm": 1.446812065317993,
      "learning_rate": 6.135471204188482e-06,
      "loss": 0.0185,
      "step": 3750
    },
    {
      "epoch": 0.06152335760451608,
      "grad_norm": 0.6501649418113474,
      "learning_rate": 6.151832460732985e-06,
      "loss": 0.0179,
      "step": 3760
    },
    {
      "epoch": 0.061686983555591914,
      "grad_norm": 0.553347905930525,
      "learning_rate": 6.168193717277488e-06,
      "loss": 0.0174,
      "step": 3770
    },
    {
      "epoch": 0.06185060950666776,
      "grad_norm": 0.8574760969414789,
      "learning_rate": 6.1845549738219905e-06,
      "loss": 0.0157,
      "step": 3780
    },
    {
      "epoch": 0.0620142354577436,
      "grad_norm": 0.6983145312300888,
      "learning_rate": 6.2009162303664925e-06,
      "loss": 0.0163,
      "step": 3790
    },
    {
      "epoch": 0.06217786140881944,
      "grad_norm": 0.5860764533748031,
      "learning_rate": 6.217277486910995e-06,
      "loss": 0.017,
      "step": 3800
    },
    {
      "epoch": 0.06234148735989528,
      "grad_norm": 0.7520156015999931,
      "learning_rate": 6.233638743455498e-06,
      "loss": 0.0152,
      "step": 3810
    },
    {
      "epoch": 0.06250511331097112,
      "grad_norm": 0.7303343865774456,
      "learning_rate": 6.25e-06,
      "loss": 0.0138,
      "step": 3820
    },
    {
      "epoch": 0.06266873926204695,
      "grad_norm": 0.6678914196391921,
      "learning_rate": 6.266361256544503e-06,
      "loss": 0.0229,
      "step": 3830
    },
    {
      "epoch": 0.0628323652131228,
      "grad_norm": 0.9001683883396626,
      "learning_rate": 6.282722513089006e-06,
      "loss": 0.0155,
      "step": 3840
    },
    {
      "epoch": 0.06299599116419864,
      "grad_norm": 0.65005219765986,
      "learning_rate": 6.299083769633508e-06,
      "loss": 0.0128,
      "step": 3850
    },
    {
      "epoch": 0.06315961711527449,
      "grad_norm": 0.7464912367018244,
      "learning_rate": 6.315445026178011e-06,
      "loss": 0.0202,
      "step": 3860
    },
    {
      "epoch": 0.06332324306635033,
      "grad_norm": 0.4116869081794819,
      "learning_rate": 6.331806282722514e-06,
      "loss": 0.014,
      "step": 3870
    },
    {
      "epoch": 0.06348686901742616,
      "grad_norm": 0.9017957509806717,
      "learning_rate": 6.348167539267016e-06,
      "loss": 0.0116,
      "step": 3880
    },
    {
      "epoch": 0.063650494968502,
      "grad_norm": 0.5816362276964531,
      "learning_rate": 6.364528795811519e-06,
      "loss": 0.0172,
      "step": 3890
    },
    {
      "epoch": 0.06381412091957785,
      "grad_norm": 0.6763658315873537,
      "learning_rate": 6.380890052356022e-06,
      "loss": 0.0106,
      "step": 3900
    },
    {
      "epoch": 0.06397774687065369,
      "grad_norm": 0.49573377401339364,
      "learning_rate": 6.397251308900524e-06,
      "loss": 0.0132,
      "step": 3910
    },
    {
      "epoch": 0.06414137282172952,
      "grad_norm": 0.9714425293283162,
      "learning_rate": 6.4136125654450265e-06,
      "loss": 0.015,
      "step": 3920
    },
    {
      "epoch": 0.06430499877280536,
      "grad_norm": 0.7840714851230691,
      "learning_rate": 6.429973821989529e-06,
      "loss": 0.0174,
      "step": 3930
    },
    {
      "epoch": 0.06446862472388121,
      "grad_norm": 0.8113175088623846,
      "learning_rate": 6.446335078534031e-06,
      "loss": 0.0111,
      "step": 3940
    },
    {
      "epoch": 0.06463225067495705,
      "grad_norm": 0.5547844836094219,
      "learning_rate": 6.462696335078534e-06,
      "loss": 0.0144,
      "step": 3950
    },
    {
      "epoch": 0.0647958766260329,
      "grad_norm": 0.3557549863210959,
      "learning_rate": 6.479057591623037e-06,
      "loss": 0.0136,
      "step": 3960
    },
    {
      "epoch": 0.06495950257710872,
      "grad_norm": 0.7447082306420533,
      "learning_rate": 6.495418848167539e-06,
      "loss": 0.0157,
      "step": 3970
    },
    {
      "epoch": 0.06512312852818457,
      "grad_norm": 0.7294771571097772,
      "learning_rate": 6.511780104712042e-06,
      "loss": 0.0164,
      "step": 3980
    },
    {
      "epoch": 0.06528675447926041,
      "grad_norm": 0.4547296312974022,
      "learning_rate": 6.528141361256545e-06,
      "loss": 0.0153,
      "step": 3990
    },
    {
      "epoch": 0.06545038043033626,
      "grad_norm": 0.9141209918521507,
      "learning_rate": 6.544502617801047e-06,
      "loss": 0.019,
      "step": 4000
    },
    {
      "epoch": 0.0656140063814121,
      "grad_norm": 1.131175695315703,
      "learning_rate": 6.56086387434555e-06,
      "loss": 0.014,
      "step": 4010
    },
    {
      "epoch": 0.06577763233248793,
      "grad_norm": 0.7657730203552764,
      "learning_rate": 6.577225130890053e-06,
      "loss": 0.0141,
      "step": 4020
    },
    {
      "epoch": 0.06594125828356377,
      "grad_norm": 0.5591370124695751,
      "learning_rate": 6.5935863874345556e-06,
      "loss": 0.019,
      "step": 4030
    },
    {
      "epoch": 0.06610488423463962,
      "grad_norm": 0.45728020236342815,
      "learning_rate": 6.609947643979058e-06,
      "loss": 0.0123,
      "step": 4040
    },
    {
      "epoch": 0.06626851018571546,
      "grad_norm": 0.4102907023022425,
      "learning_rate": 6.6263089005235605e-06,
      "loss": 0.0176,
      "step": 4050
    },
    {
      "epoch": 0.06643213613679129,
      "grad_norm": 0.47127324900756756,
      "learning_rate": 6.642670157068063e-06,
      "loss": 0.0128,
      "step": 4060
    },
    {
      "epoch": 0.06659576208786713,
      "grad_norm": 0.5508555023743599,
      "learning_rate": 6.659031413612565e-06,
      "loss": 0.0116,
      "step": 4070
    },
    {
      "epoch": 0.06675938803894298,
      "grad_norm": 0.7590976899691034,
      "learning_rate": 6.675392670157068e-06,
      "loss": 0.0131,
      "step": 4080
    },
    {
      "epoch": 0.06692301399001882,
      "grad_norm": 0.9544608583829555,
      "learning_rate": 6.691753926701571e-06,
      "loss": 0.0167,
      "step": 4090
    },
    {
      "epoch": 0.06708663994109466,
      "grad_norm": 0.3881567337817207,
      "learning_rate": 6.708115183246073e-06,
      "loss": 0.0109,
      "step": 4100
    },
    {
      "epoch": 0.0672502658921705,
      "grad_norm": 0.7505791407631902,
      "learning_rate": 6.724476439790576e-06,
      "loss": 0.0142,
      "step": 4110
    },
    {
      "epoch": 0.06741389184324634,
      "grad_norm": 0.6790782885784833,
      "learning_rate": 6.740837696335079e-06,
      "loss": 0.0198,
      "step": 4120
    },
    {
      "epoch": 0.06757751779432218,
      "grad_norm": 1.4392835399531998,
      "learning_rate": 6.757198952879581e-06,
      "loss": 0.0145,
      "step": 4130
    },
    {
      "epoch": 0.06774114374539802,
      "grad_norm": 0.7630079196631412,
      "learning_rate": 6.773560209424085e-06,
      "loss": 0.0124,
      "step": 4140
    },
    {
      "epoch": 0.06790476969647385,
      "grad_norm": 1.2648593920336735,
      "learning_rate": 6.7899214659685875e-06,
      "loss": 0.0159,
      "step": 4150
    },
    {
      "epoch": 0.0680683956475497,
      "grad_norm": 0.8827105338994375,
      "learning_rate": 6.80628272251309e-06,
      "loss": 0.0123,
      "step": 4160
    },
    {
      "epoch": 0.06823202159862554,
      "grad_norm": 0.3186105235515339,
      "learning_rate": 6.822643979057592e-06,
      "loss": 0.008,
      "step": 4170
    },
    {
      "epoch": 0.06839564754970139,
      "grad_norm": 0.45738580383678074,
      "learning_rate": 6.839005235602095e-06,
      "loss": 0.0162,
      "step": 4180
    },
    {
      "epoch": 0.06855927350077723,
      "grad_norm": 0.7887035488532753,
      "learning_rate": 6.855366492146598e-06,
      "loss": 0.0105,
      "step": 4190
    },
    {
      "epoch": 0.06872289945185306,
      "grad_norm": 0.8042971788596543,
      "learning_rate": 6.8717277486911e-06,
      "loss": 0.0176,
      "step": 4200
    },
    {
      "epoch": 0.0688865254029289,
      "grad_norm": 1.037092702448011,
      "learning_rate": 6.888089005235603e-06,
      "loss": 0.0138,
      "step": 4210
    },
    {
      "epoch": 0.06905015135400475,
      "grad_norm": 0.6317197586946248,
      "learning_rate": 6.904450261780106e-06,
      "loss": 0.0125,
      "step": 4220
    },
    {
      "epoch": 0.06921377730508059,
      "grad_norm": 0.6152060182718515,
      "learning_rate": 6.920811518324608e-06,
      "loss": 0.012,
      "step": 4230
    },
    {
      "epoch": 0.06937740325615642,
      "grad_norm": 0.9609770861603334,
      "learning_rate": 6.937172774869111e-06,
      "loss": 0.0153,
      "step": 4240
    },
    {
      "epoch": 0.06954102920723226,
      "grad_norm": 0.7142544100467115,
      "learning_rate": 6.953534031413614e-06,
      "loss": 0.0308,
      "step": 4250
    },
    {
      "epoch": 0.06970465515830811,
      "grad_norm": 0.4570493097615138,
      "learning_rate": 6.969895287958116e-06,
      "loss": 0.0175,
      "step": 4260
    },
    {
      "epoch": 0.06986828110938395,
      "grad_norm": 0.7537756532597911,
      "learning_rate": 6.986256544502619e-06,
      "loss": 0.0149,
      "step": 4270
    },
    {
      "epoch": 0.0700319070604598,
      "grad_norm": 0.5883813606397793,
      "learning_rate": 7.0026178010471215e-06,
      "loss": 0.0123,
      "step": 4280
    },
    {
      "epoch": 0.07019553301153562,
      "grad_norm": 0.8140805323424636,
      "learning_rate": 7.0189790575916235e-06,
      "loss": 0.0165,
      "step": 4290
    },
    {
      "epoch": 0.07035915896261147,
      "grad_norm": 0.6476029462192842,
      "learning_rate": 7.035340314136126e-06,
      "loss": 0.0121,
      "step": 4300
    },
    {
      "epoch": 0.07052278491368731,
      "grad_norm": 0.7199555226331041,
      "learning_rate": 7.051701570680629e-06,
      "loss": 0.0113,
      "step": 4310
    },
    {
      "epoch": 0.07068641086476316,
      "grad_norm": 0.5367680601761631,
      "learning_rate": 7.068062827225132e-06,
      "loss": 0.0106,
      "step": 4320
    },
    {
      "epoch": 0.070850036815839,
      "grad_norm": 0.8194904953760117,
      "learning_rate": 7.084424083769634e-06,
      "loss": 0.0212,
      "step": 4330
    },
    {
      "epoch": 0.07101366276691483,
      "grad_norm": 0.5779089065110249,
      "learning_rate": 7.100785340314137e-06,
      "loss": 0.0151,
      "step": 4340
    },
    {
      "epoch": 0.07117728871799067,
      "grad_norm": 1.1633273771495134,
      "learning_rate": 7.11714659685864e-06,
      "loss": 0.0105,
      "step": 4350
    },
    {
      "epoch": 0.07134091466906652,
      "grad_norm": 0.7461620935383845,
      "learning_rate": 7.133507853403142e-06,
      "loss": 0.0125,
      "step": 4360
    },
    {
      "epoch": 0.07150454062014236,
      "grad_norm": 0.49373157653633143,
      "learning_rate": 7.149869109947645e-06,
      "loss": 0.0201,
      "step": 4370
    },
    {
      "epoch": 0.07166816657121819,
      "grad_norm": 0.18989488446746042,
      "learning_rate": 7.166230366492148e-06,
      "loss": 0.021,
      "step": 4380
    },
    {
      "epoch": 0.07183179252229403,
      "grad_norm": 0.5059101308509552,
      "learning_rate": 7.18259162303665e-06,
      "loss": 0.0151,
      "step": 4390
    },
    {
      "epoch": 0.07199541847336988,
      "grad_norm": 0.8021459791545428,
      "learning_rate": 7.1989528795811526e-06,
      "loss": 0.015,
      "step": 4400
    },
    {
      "epoch": 0.07215904442444572,
      "grad_norm": 0.6956180294016454,
      "learning_rate": 7.2153141361256554e-06,
      "loss": 0.0094,
      "step": 4410
    },
    {
      "epoch": 0.07232267037552156,
      "grad_norm": 0.37293818447826366,
      "learning_rate": 7.2316753926701575e-06,
      "loss": 0.0143,
      "step": 4420
    },
    {
      "epoch": 0.0724862963265974,
      "grad_norm": 0.572876713233933,
      "learning_rate": 7.24803664921466e-06,
      "loss": 0.0155,
      "step": 4430
    },
    {
      "epoch": 0.07264992227767324,
      "grad_norm": 0.5632004757337942,
      "learning_rate": 7.264397905759163e-06,
      "loss": 0.0142,
      "step": 4440
    },
    {
      "epoch": 0.07281354822874908,
      "grad_norm": 0.9946656856088647,
      "learning_rate": 7.280759162303665e-06,
      "loss": 0.0189,
      "step": 4450
    },
    {
      "epoch": 0.07297717417982492,
      "grad_norm": 0.5564258242381411,
      "learning_rate": 7.297120418848168e-06,
      "loss": 0.0131,
      "step": 4460
    },
    {
      "epoch": 0.07314080013090075,
      "grad_norm": 0.35464786292046324,
      "learning_rate": 7.313481675392671e-06,
      "loss": 0.0131,
      "step": 4470
    },
    {
      "epoch": 0.0733044260819766,
      "grad_norm": 0.3899449351034589,
      "learning_rate": 7.329842931937173e-06,
      "loss": 0.0125,
      "step": 4480
    },
    {
      "epoch": 0.07346805203305244,
      "grad_norm": 0.6709709607965249,
      "learning_rate": 7.346204188481676e-06,
      "loss": 0.0109,
      "step": 4490
    },
    {
      "epoch": 0.07363167798412829,
      "grad_norm": 0.6645920101688736,
      "learning_rate": 7.362565445026179e-06,
      "loss": 0.0129,
      "step": 4500
    },
    {
      "epoch": 0.07379530393520413,
      "grad_norm": 0.6252601823839561,
      "learning_rate": 7.378926701570681e-06,
      "loss": 0.0183,
      "step": 4510
    },
    {
      "epoch": 0.07395892988627996,
      "grad_norm": 0.31433198781455196,
      "learning_rate": 7.395287958115184e-06,
      "loss": 0.0168,
      "step": 4520
    },
    {
      "epoch": 0.0741225558373558,
      "grad_norm": 0.8476864645641823,
      "learning_rate": 7.4116492146596865e-06,
      "loss": 0.0184,
      "step": 4530
    },
    {
      "epoch": 0.07428618178843165,
      "grad_norm": 0.5373989206838607,
      "learning_rate": 7.4280104712041886e-06,
      "loss": 0.0143,
      "step": 4540
    },
    {
      "epoch": 0.07444980773950749,
      "grad_norm": 0.7227427561336182,
      "learning_rate": 7.4443717277486914e-06,
      "loss": 0.013,
      "step": 4550
    },
    {
      "epoch": 0.07461343369058333,
      "grad_norm": 0.6169782553255874,
      "learning_rate": 7.460732984293194e-06,
      "loss": 0.0176,
      "step": 4560
    },
    {
      "epoch": 0.07477705964165916,
      "grad_norm": 0.45290180994966234,
      "learning_rate": 7.477094240837696e-06,
      "loss": 0.012,
      "step": 4570
    },
    {
      "epoch": 0.074940685592735,
      "grad_norm": 0.949344057584366,
      "learning_rate": 7.493455497382199e-06,
      "loss": 0.0123,
      "step": 4580
    },
    {
      "epoch": 0.07510431154381085,
      "grad_norm": 0.9694129051706206,
      "learning_rate": 7.509816753926702e-06,
      "loss": 0.0112,
      "step": 4590
    },
    {
      "epoch": 0.0752679374948867,
      "grad_norm": 0.6480674266712945,
      "learning_rate": 7.526178010471205e-06,
      "loss": 0.0147,
      "step": 4600
    },
    {
      "epoch": 0.07543156344596252,
      "grad_norm": 0.7306346436899508,
      "learning_rate": 7.542539267015707e-06,
      "loss": 0.0134,
      "step": 4610
    },
    {
      "epoch": 0.07559518939703837,
      "grad_norm": 0.570754536863107,
      "learning_rate": 7.55890052356021e-06,
      "loss": 0.012,
      "step": 4620
    },
    {
      "epoch": 0.07575881534811421,
      "grad_norm": 0.8623451210825357,
      "learning_rate": 7.575261780104713e-06,
      "loss": 0.0162,
      "step": 4630
    },
    {
      "epoch": 0.07592244129919005,
      "grad_norm": 0.9131007921259081,
      "learning_rate": 7.591623036649215e-06,
      "loss": 0.018,
      "step": 4640
    },
    {
      "epoch": 0.0760860672502659,
      "grad_norm": 0.45252256982047045,
      "learning_rate": 7.607984293193718e-06,
      "loss": 0.0162,
      "step": 4650
    },
    {
      "epoch": 0.07624969320134173,
      "grad_norm": 0.6699670107713176,
      "learning_rate": 7.6243455497382205e-06,
      "loss": 0.0151,
      "step": 4660
    },
    {
      "epoch": 0.07641331915241757,
      "grad_norm": 0.4494841647265516,
      "learning_rate": 7.640706806282723e-06,
      "loss": 0.0169,
      "step": 4670
    },
    {
      "epoch": 0.07657694510349342,
      "grad_norm": 0.6081696650031454,
      "learning_rate": 7.657068062827225e-06,
      "loss": 0.0137,
      "step": 4680
    },
    {
      "epoch": 0.07674057105456926,
      "grad_norm": 0.6040142336392571,
      "learning_rate": 7.673429319371727e-06,
      "loss": 0.0152,
      "step": 4690
    },
    {
      "epoch": 0.07690419700564509,
      "grad_norm": 0.9129215084422817,
      "learning_rate": 7.68979057591623e-06,
      "loss": 0.0147,
      "step": 4700
    },
    {
      "epoch": 0.07706782295672093,
      "grad_norm": 0.9280713638306605,
      "learning_rate": 7.706151832460733e-06,
      "loss": 0.015,
      "step": 4710
    },
    {
      "epoch": 0.07723144890779678,
      "grad_norm": 0.454699687086561,
      "learning_rate": 7.722513089005236e-06,
      "loss": 0.0115,
      "step": 4720
    },
    {
      "epoch": 0.07739507485887262,
      "grad_norm": 0.4774867966608041,
      "learning_rate": 7.738874345549739e-06,
      "loss": 0.0134,
      "step": 4730
    },
    {
      "epoch": 0.07755870080994846,
      "grad_norm": 0.5458183954795294,
      "learning_rate": 7.755235602094242e-06,
      "loss": 0.0151,
      "step": 4740
    },
    {
      "epoch": 0.0777223267610243,
      "grad_norm": 0.6469101291081161,
      "learning_rate": 7.771596858638743e-06,
      "loss": 0.0206,
      "step": 4750
    },
    {
      "epoch": 0.07788595271210014,
      "grad_norm": 0.42508130812038447,
      "learning_rate": 7.787958115183246e-06,
      "loss": 0.0094,
      "step": 4760
    },
    {
      "epoch": 0.07804957866317598,
      "grad_norm": 0.5617913129927786,
      "learning_rate": 7.804319371727749e-06,
      "loss": 0.0123,
      "step": 4770
    },
    {
      "epoch": 0.07821320461425182,
      "grad_norm": 0.5329401955860894,
      "learning_rate": 7.820680628272253e-06,
      "loss": 0.0101,
      "step": 4780
    },
    {
      "epoch": 0.07837683056532765,
      "grad_norm": 0.6395338737200763,
      "learning_rate": 7.837041884816754e-06,
      "loss": 0.0133,
      "step": 4790
    },
    {
      "epoch": 0.0785404565164035,
      "grad_norm": 0.666131372679514,
      "learning_rate": 7.853403141361257e-06,
      "loss": 0.0121,
      "step": 4800
    },
    {
      "epoch": 0.07870408246747934,
      "grad_norm": 0.8827681161226104,
      "learning_rate": 7.86976439790576e-06,
      "loss": 0.0124,
      "step": 4810
    },
    {
      "epoch": 0.07886770841855519,
      "grad_norm": 0.7232044563691765,
      "learning_rate": 7.886125654450263e-06,
      "loss": 0.0133,
      "step": 4820
    },
    {
      "epoch": 0.07903133436963103,
      "grad_norm": 1.167337803477484,
      "learning_rate": 7.902486910994766e-06,
      "loss": 0.014,
      "step": 4830
    },
    {
      "epoch": 0.07919496032070686,
      "grad_norm": 0.7248225558572633,
      "learning_rate": 7.918848167539269e-06,
      "loss": 0.013,
      "step": 4840
    },
    {
      "epoch": 0.0793585862717827,
      "grad_norm": 0.4863488074299087,
      "learning_rate": 7.93520942408377e-06,
      "loss": 0.0126,
      "step": 4850
    },
    {
      "epoch": 0.07952221222285855,
      "grad_norm": 0.503189764965063,
      "learning_rate": 7.951570680628273e-06,
      "loss": 0.015,
      "step": 4860
    },
    {
      "epoch": 0.07968583817393439,
      "grad_norm": 0.29074322984198236,
      "learning_rate": 7.967931937172776e-06,
      "loss": 0.0115,
      "step": 4870
    },
    {
      "epoch": 0.07984946412501023,
      "grad_norm": 0.4929143644026491,
      "learning_rate": 7.984293193717279e-06,
      "loss": 0.0128,
      "step": 4880
    },
    {
      "epoch": 0.08001309007608606,
      "grad_norm": 0.5715697905468171,
      "learning_rate": 8.000654450261782e-06,
      "loss": 0.0094,
      "step": 4890
    },
    {
      "epoch": 0.0801767160271619,
      "grad_norm": 0.8214363448815072,
      "learning_rate": 8.017015706806284e-06,
      "loss": 0.0144,
      "step": 4900
    },
    {
      "epoch": 0.08034034197823775,
      "grad_norm": 0.5477150888035965,
      "learning_rate": 8.033376963350786e-06,
      "loss": 0.0107,
      "step": 4910
    },
    {
      "epoch": 0.0805039679293136,
      "grad_norm": 0.5947624973163396,
      "learning_rate": 8.049738219895288e-06,
      "loss": 0.0093,
      "step": 4920
    },
    {
      "epoch": 0.08066759388038942,
      "grad_norm": 0.7621795357035333,
      "learning_rate": 8.066099476439791e-06,
      "loss": 0.0142,
      "step": 4930
    },
    {
      "epoch": 0.08083121983146527,
      "grad_norm": 1.1797658516379197,
      "learning_rate": 8.082460732984294e-06,
      "loss": 0.0122,
      "step": 4940
    },
    {
      "epoch": 0.08099484578254111,
      "grad_norm": 0.5267792543844083,
      "learning_rate": 8.098821989528797e-06,
      "loss": 0.012,
      "step": 4950
    },
    {
      "epoch": 0.08115847173361695,
      "grad_norm": 0.5671734433973993,
      "learning_rate": 8.1151832460733e-06,
      "loss": 0.0158,
      "step": 4960
    },
    {
      "epoch": 0.0813220976846928,
      "grad_norm": 0.21432780065511384,
      "learning_rate": 8.131544502617801e-06,
      "loss": 0.0121,
      "step": 4970
    },
    {
      "epoch": 0.08148572363576863,
      "grad_norm": 0.7443064258324583,
      "learning_rate": 8.147905759162304e-06,
      "loss": 0.0123,
      "step": 4980
    },
    {
      "epoch": 0.08164934958684447,
      "grad_norm": 0.563080575786992,
      "learning_rate": 8.164267015706807e-06,
      "loss": 0.0131,
      "step": 4990
    },
    {
      "epoch": 0.08181297553792032,
      "grad_norm": 0.638586722036964,
      "learning_rate": 8.18062827225131e-06,
      "loss": 0.0131,
      "step": 5000
    },
    {
      "epoch": 0.08197660148899616,
      "grad_norm": 0.8550330064027312,
      "learning_rate": 8.196989528795813e-06,
      "loss": 0.0133,
      "step": 5010
    },
    {
      "epoch": 0.08214022744007199,
      "grad_norm": 0.7372420826855306,
      "learning_rate": 8.213350785340315e-06,
      "loss": 0.0093,
      "step": 5020
    },
    {
      "epoch": 0.08230385339114783,
      "grad_norm": 0.6448125244894859,
      "learning_rate": 8.229712041884818e-06,
      "loss": 0.0131,
      "step": 5030
    },
    {
      "epoch": 0.08246747934222368,
      "grad_norm": 0.49065959460000513,
      "learning_rate": 8.24607329842932e-06,
      "loss": 0.014,
      "step": 5040
    },
    {
      "epoch": 0.08263110529329952,
      "grad_norm": 0.5152802313309615,
      "learning_rate": 8.262434554973822e-06,
      "loss": 0.0143,
      "step": 5050
    },
    {
      "epoch": 0.08279473124437536,
      "grad_norm": 0.558065566548443,
      "learning_rate": 8.278795811518325e-06,
      "loss": 0.0122,
      "step": 5060
    },
    {
      "epoch": 0.0829583571954512,
      "grad_norm": 0.5927835160140081,
      "learning_rate": 8.295157068062828e-06,
      "loss": 0.0135,
      "step": 5070
    },
    {
      "epoch": 0.08312198314652704,
      "grad_norm": 0.7597132931129545,
      "learning_rate": 8.311518324607331e-06,
      "loss": 0.012,
      "step": 5080
    },
    {
      "epoch": 0.08328560909760288,
      "grad_norm": 0.579973786729228,
      "learning_rate": 8.327879581151834e-06,
      "loss": 0.013,
      "step": 5090
    },
    {
      "epoch": 0.08344923504867872,
      "grad_norm": 0.504242159574087,
      "learning_rate": 8.344240837696335e-06,
      "loss": 0.0108,
      "step": 5100
    },
    {
      "epoch": 0.08361286099975457,
      "grad_norm": 0.6092329606623748,
      "learning_rate": 8.360602094240838e-06,
      "loss": 0.0118,
      "step": 5110
    },
    {
      "epoch": 0.0837764869508304,
      "grad_norm": 0.5741700793291044,
      "learning_rate": 8.37696335078534e-06,
      "loss": 0.0174,
      "step": 5120
    },
    {
      "epoch": 0.08394011290190624,
      "grad_norm": 0.9029617784371309,
      "learning_rate": 8.393324607329844e-06,
      "loss": 0.0112,
      "step": 5130
    },
    {
      "epoch": 0.08410373885298209,
      "grad_norm": 0.4234557555918461,
      "learning_rate": 8.409685863874347e-06,
      "loss": 0.0149,
      "step": 5140
    },
    {
      "epoch": 0.08426736480405793,
      "grad_norm": 0.5347592564711967,
      "learning_rate": 8.42604712041885e-06,
      "loss": 0.01,
      "step": 5150
    },
    {
      "epoch": 0.08443099075513376,
      "grad_norm": 0.6298103333681973,
      "learning_rate": 8.44240837696335e-06,
      "loss": 0.0169,
      "step": 5160
    },
    {
      "epoch": 0.0845946167062096,
      "grad_norm": 1.0130294513551699,
      "learning_rate": 8.458769633507854e-06,
      "loss": 0.0176,
      "step": 5170
    },
    {
      "epoch": 0.08475824265728545,
      "grad_norm": 0.4329106991631057,
      "learning_rate": 8.475130890052356e-06,
      "loss": 0.0129,
      "step": 5180
    },
    {
      "epoch": 0.08492186860836129,
      "grad_norm": 0.47608299492066497,
      "learning_rate": 8.49149214659686e-06,
      "loss": 0.0108,
      "step": 5190
    },
    {
      "epoch": 0.08508549455943713,
      "grad_norm": 0.5402699114804711,
      "learning_rate": 8.507853403141362e-06,
      "loss": 0.0188,
      "step": 5200
    },
    {
      "epoch": 0.08524912051051296,
      "grad_norm": 0.32003702749057233,
      "learning_rate": 8.524214659685865e-06,
      "loss": 0.0138,
      "step": 5210
    },
    {
      "epoch": 0.0854127464615888,
      "grad_norm": 0.46881971289152213,
      "learning_rate": 8.540575916230366e-06,
      "loss": 0.018,
      "step": 5220
    },
    {
      "epoch": 0.08557637241266465,
      "grad_norm": 0.45995438815927164,
      "learning_rate": 8.556937172774869e-06,
      "loss": 0.0124,
      "step": 5230
    },
    {
      "epoch": 0.0857399983637405,
      "grad_norm": 0.4389298534749428,
      "learning_rate": 8.573298429319372e-06,
      "loss": 0.0137,
      "step": 5240
    },
    {
      "epoch": 0.08590362431481632,
      "grad_norm": 0.4806428743928941,
      "learning_rate": 8.589659685863875e-06,
      "loss": 0.0141,
      "step": 5250
    },
    {
      "epoch": 0.08606725026589217,
      "grad_norm": 0.4536125253959356,
      "learning_rate": 8.606020942408378e-06,
      "loss": 0.0132,
      "step": 5260
    },
    {
      "epoch": 0.08623087621696801,
      "grad_norm": 0.6191264678357633,
      "learning_rate": 8.62238219895288e-06,
      "loss": 0.009,
      "step": 5270
    },
    {
      "epoch": 0.08639450216804385,
      "grad_norm": 0.4229759315390177,
      "learning_rate": 8.638743455497383e-06,
      "loss": 0.0109,
      "step": 5280
    },
    {
      "epoch": 0.0865581281191197,
      "grad_norm": 0.5097145885350656,
      "learning_rate": 8.655104712041885e-06,
      "loss": 0.0125,
      "step": 5290
    },
    {
      "epoch": 0.08672175407019553,
      "grad_norm": 0.5899817657490514,
      "learning_rate": 8.671465968586387e-06,
      "loss": 0.0137,
      "step": 5300
    },
    {
      "epoch": 0.08688538002127137,
      "grad_norm": 0.5235841408712382,
      "learning_rate": 8.68782722513089e-06,
      "loss": 0.0106,
      "step": 5310
    },
    {
      "epoch": 0.08704900597234722,
      "grad_norm": 0.38049642874794304,
      "learning_rate": 8.704188481675393e-06,
      "loss": 0.0152,
      "step": 5320
    },
    {
      "epoch": 0.08721263192342306,
      "grad_norm": 0.4897392377001293,
      "learning_rate": 8.720549738219896e-06,
      "loss": 0.0143,
      "step": 5330
    },
    {
      "epoch": 0.08737625787449889,
      "grad_norm": 0.8889274785827478,
      "learning_rate": 8.736910994764399e-06,
      "loss": 0.0121,
      "step": 5340
    },
    {
      "epoch": 0.08753988382557473,
      "grad_norm": 0.5224856735765664,
      "learning_rate": 8.7532722513089e-06,
      "loss": 0.0109,
      "step": 5350
    },
    {
      "epoch": 0.08770350977665058,
      "grad_norm": 0.45245700203002165,
      "learning_rate": 8.769633507853403e-06,
      "loss": 0.0092,
      "step": 5360
    },
    {
      "epoch": 0.08786713572772642,
      "grad_norm": 0.302319510756231,
      "learning_rate": 8.785994764397906e-06,
      "loss": 0.0121,
      "step": 5370
    },
    {
      "epoch": 0.08803076167880226,
      "grad_norm": 0.5473049093722626,
      "learning_rate": 8.802356020942409e-06,
      "loss": 0.0172,
      "step": 5380
    },
    {
      "epoch": 0.0881943876298781,
      "grad_norm": 0.576704199332457,
      "learning_rate": 8.818717277486912e-06,
      "loss": 0.0126,
      "step": 5390
    },
    {
      "epoch": 0.08835801358095394,
      "grad_norm": 0.5799164147565287,
      "learning_rate": 8.835078534031415e-06,
      "loss": 0.012,
      "step": 5400
    },
    {
      "epoch": 0.08852163953202978,
      "grad_norm": 0.6598336157256204,
      "learning_rate": 8.851439790575916e-06,
      "loss": 0.0107,
      "step": 5410
    },
    {
      "epoch": 0.08868526548310562,
      "grad_norm": 0.46511049748923067,
      "learning_rate": 8.86780104712042e-06,
      "loss": 0.0093,
      "step": 5420
    },
    {
      "epoch": 0.08884889143418147,
      "grad_norm": 0.3611482691432295,
      "learning_rate": 8.884162303664923e-06,
      "loss": 0.0092,
      "step": 5430
    },
    {
      "epoch": 0.0890125173852573,
      "grad_norm": 0.8314071856354412,
      "learning_rate": 8.900523560209426e-06,
      "loss": 0.011,
      "step": 5440
    },
    {
      "epoch": 0.08917614333633314,
      "grad_norm": 0.19772782566529837,
      "learning_rate": 8.916884816753927e-06,
      "loss": 0.0081,
      "step": 5450
    },
    {
      "epoch": 0.08933976928740898,
      "grad_norm": 0.9126046439618061,
      "learning_rate": 8.93324607329843e-06,
      "loss": 0.0139,
      "step": 5460
    },
    {
      "epoch": 0.08950339523848483,
      "grad_norm": 0.4519667591071422,
      "learning_rate": 8.949607329842933e-06,
      "loss": 0.0088,
      "step": 5470
    },
    {
      "epoch": 0.08966702118956066,
      "grad_norm": 0.8499820813009065,
      "learning_rate": 8.965968586387436e-06,
      "loss": 0.0157,
      "step": 5480
    },
    {
      "epoch": 0.0898306471406365,
      "grad_norm": 0.7261175593770764,
      "learning_rate": 8.982329842931939e-06,
      "loss": 0.0101,
      "step": 5490
    },
    {
      "epoch": 0.08999427309171235,
      "grad_norm": 0.43274976938724563,
      "learning_rate": 8.998691099476442e-06,
      "loss": 0.0106,
      "step": 5500
    },
    {
      "epoch": 0.09015789904278819,
      "grad_norm": 0.4231139241430943,
      "learning_rate": 9.015052356020943e-06,
      "loss": 0.0111,
      "step": 5510
    },
    {
      "epoch": 0.09032152499386403,
      "grad_norm": 0.6678215513867426,
      "learning_rate": 9.031413612565446e-06,
      "loss": 0.01,
      "step": 5520
    },
    {
      "epoch": 0.09048515094493986,
      "grad_norm": 0.4431004571020533,
      "learning_rate": 9.047774869109948e-06,
      "loss": 0.0133,
      "step": 5530
    },
    {
      "epoch": 0.0906487768960157,
      "grad_norm": 0.5043903920741631,
      "learning_rate": 9.064136125654451e-06,
      "loss": 0.012,
      "step": 5540
    },
    {
      "epoch": 0.09081240284709155,
      "grad_norm": 0.49879202981327303,
      "learning_rate": 9.080497382198954e-06,
      "loss": 0.0144,
      "step": 5550
    },
    {
      "epoch": 0.0909760287981674,
      "grad_norm": 0.5487611682303718,
      "learning_rate": 9.096858638743457e-06,
      "loss": 0.0109,
      "step": 5560
    },
    {
      "epoch": 0.09113965474924322,
      "grad_norm": 0.5251807533907431,
      "learning_rate": 9.11321989528796e-06,
      "loss": 0.0154,
      "step": 5570
    },
    {
      "epoch": 0.09130328070031907,
      "grad_norm": 0.4825502958915453,
      "learning_rate": 9.129581151832461e-06,
      "loss": 0.0126,
      "step": 5580
    },
    {
      "epoch": 0.09146690665139491,
      "grad_norm": 0.23661115386257375,
      "learning_rate": 9.145942408376964e-06,
      "loss": 0.0104,
      "step": 5590
    },
    {
      "epoch": 0.09163053260247075,
      "grad_norm": 0.5483570770830758,
      "learning_rate": 9.162303664921467e-06,
      "loss": 0.013,
      "step": 5600
    },
    {
      "epoch": 0.0917941585535466,
      "grad_norm": 0.5258580991583125,
      "learning_rate": 9.17866492146597e-06,
      "loss": 0.0103,
      "step": 5610
    },
    {
      "epoch": 0.09195778450462243,
      "grad_norm": 0.47654426598183103,
      "learning_rate": 9.195026178010473e-06,
      "loss": 0.0167,
      "step": 5620
    },
    {
      "epoch": 0.09212141045569827,
      "grad_norm": 0.6901057737770547,
      "learning_rate": 9.211387434554976e-06,
      "loss": 0.013,
      "step": 5630
    },
    {
      "epoch": 0.09228503640677412,
      "grad_norm": 0.5365234514536517,
      "learning_rate": 9.227748691099477e-06,
      "loss": 0.0174,
      "step": 5640
    },
    {
      "epoch": 0.09244866235784996,
      "grad_norm": 0.42164513528728886,
      "learning_rate": 9.24410994764398e-06,
      "loss": 0.0097,
      "step": 5650
    },
    {
      "epoch": 0.0926122883089258,
      "grad_norm": 0.5323929096467224,
      "learning_rate": 9.260471204188482e-06,
      "loss": 0.0092,
      "step": 5660
    },
    {
      "epoch": 0.09277591426000163,
      "grad_norm": 0.6021617778230087,
      "learning_rate": 9.276832460732985e-06,
      "loss": 0.0122,
      "step": 5670
    },
    {
      "epoch": 0.09293954021107748,
      "grad_norm": 0.6100655936588538,
      "learning_rate": 9.293193717277488e-06,
      "loss": 0.0125,
      "step": 5680
    },
    {
      "epoch": 0.09310316616215332,
      "grad_norm": 0.4264614283239258,
      "learning_rate": 9.309554973821991e-06,
      "loss": 0.0101,
      "step": 5690
    },
    {
      "epoch": 0.09326679211322916,
      "grad_norm": 0.3341489359941811,
      "learning_rate": 9.325916230366492e-06,
      "loss": 0.0124,
      "step": 5700
    },
    {
      "epoch": 0.093430418064305,
      "grad_norm": 0.8771706664828531,
      "learning_rate": 9.342277486910995e-06,
      "loss": 0.021,
      "step": 5710
    },
    {
      "epoch": 0.09359404401538084,
      "grad_norm": 0.4024944051729391,
      "learning_rate": 9.358638743455498e-06,
      "loss": 0.0107,
      "step": 5720
    },
    {
      "epoch": 0.09375766996645668,
      "grad_norm": 0.4664819254955415,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.0104,
      "step": 5730
    },
    {
      "epoch": 0.09392129591753252,
      "grad_norm": 0.9557743248862933,
      "learning_rate": 9.391361256544504e-06,
      "loss": 0.0123,
      "step": 5740
    },
    {
      "epoch": 0.09408492186860837,
      "grad_norm": 0.28965646571528225,
      "learning_rate": 9.407722513089007e-06,
      "loss": 0.0111,
      "step": 5750
    },
    {
      "epoch": 0.0942485478196842,
      "grad_norm": 0.534439643638679,
      "learning_rate": 9.424083769633508e-06,
      "loss": 0.0144,
      "step": 5760
    },
    {
      "epoch": 0.09441217377076004,
      "grad_norm": 0.4790844283951084,
      "learning_rate": 9.44044502617801e-06,
      "loss": 0.0124,
      "step": 5770
    },
    {
      "epoch": 0.09457579972183588,
      "grad_norm": 0.44606197288205657,
      "learning_rate": 9.456806282722514e-06,
      "loss": 0.0116,
      "step": 5780
    },
    {
      "epoch": 0.09473942567291173,
      "grad_norm": 0.48794495511419306,
      "learning_rate": 9.473167539267016e-06,
      "loss": 0.0119,
      "step": 5790
    },
    {
      "epoch": 0.09490305162398756,
      "grad_norm": 0.778767665805708,
      "learning_rate": 9.48952879581152e-06,
      "loss": 0.0129,
      "step": 5800
    },
    {
      "epoch": 0.0950666775750634,
      "grad_norm": 0.561248813747974,
      "learning_rate": 9.505890052356022e-06,
      "loss": 0.011,
      "step": 5810
    },
    {
      "epoch": 0.09523030352613925,
      "grad_norm": 0.47345817934506,
      "learning_rate": 9.522251308900523e-06,
      "loss": 0.0114,
      "step": 5820
    },
    {
      "epoch": 0.09539392947721509,
      "grad_norm": 0.33937833950057283,
      "learning_rate": 9.538612565445026e-06,
      "loss": 0.0155,
      "step": 5830
    },
    {
      "epoch": 0.09555755542829093,
      "grad_norm": 0.22233343462297936,
      "learning_rate": 9.554973821989529e-06,
      "loss": 0.01,
      "step": 5840
    },
    {
      "epoch": 0.09572118137936676,
      "grad_norm": 0.4651076170946273,
      "learning_rate": 9.571335078534032e-06,
      "loss": 0.0123,
      "step": 5850
    },
    {
      "epoch": 0.0958848073304426,
      "grad_norm": 0.41037077259507493,
      "learning_rate": 9.587696335078535e-06,
      "loss": 0.0136,
      "step": 5860
    },
    {
      "epoch": 0.09604843328151845,
      "grad_norm": 0.40390270942599565,
      "learning_rate": 9.604057591623038e-06,
      "loss": 0.0139,
      "step": 5870
    },
    {
      "epoch": 0.0962120592325943,
      "grad_norm": 0.3889564696048176,
      "learning_rate": 9.62041884816754e-06,
      "loss": 0.0074,
      "step": 5880
    },
    {
      "epoch": 0.09637568518367012,
      "grad_norm": 0.635535935685924,
      "learning_rate": 9.636780104712042e-06,
      "loss": 0.0096,
      "step": 5890
    },
    {
      "epoch": 0.09653931113474597,
      "grad_norm": 0.5847337118323772,
      "learning_rate": 9.653141361256545e-06,
      "loss": 0.0135,
      "step": 5900
    },
    {
      "epoch": 0.09670293708582181,
      "grad_norm": 0.313850819389411,
      "learning_rate": 9.669502617801048e-06,
      "loss": 0.0116,
      "step": 5910
    },
    {
      "epoch": 0.09686656303689765,
      "grad_norm": 0.478417121078612,
      "learning_rate": 9.68586387434555e-06,
      "loss": 0.0116,
      "step": 5920
    },
    {
      "epoch": 0.0970301889879735,
      "grad_norm": 0.46530486398417226,
      "learning_rate": 9.702225130890053e-06,
      "loss": 0.0085,
      "step": 5930
    },
    {
      "epoch": 0.09719381493904933,
      "grad_norm": 0.4103041985489078,
      "learning_rate": 9.718586387434556e-06,
      "loss": 0.0098,
      "step": 5940
    },
    {
      "epoch": 0.09735744089012517,
      "grad_norm": 0.5348833294739196,
      "learning_rate": 9.734947643979057e-06,
      "loss": 0.0151,
      "step": 5950
    },
    {
      "epoch": 0.09752106684120102,
      "grad_norm": 0.6883291943698158,
      "learning_rate": 9.75130890052356e-06,
      "loss": 0.0134,
      "step": 5960
    },
    {
      "epoch": 0.09768469279227686,
      "grad_norm": 0.6726455179837149,
      "learning_rate": 9.767670157068063e-06,
      "loss": 0.0134,
      "step": 5970
    },
    {
      "epoch": 0.0978483187433527,
      "grad_norm": 0.5053530352646637,
      "learning_rate": 9.784031413612566e-06,
      "loss": 0.0112,
      "step": 5980
    },
    {
      "epoch": 0.09801194469442853,
      "grad_norm": 0.6342174481202657,
      "learning_rate": 9.800392670157069e-06,
      "loss": 0.0088,
      "step": 5990
    },
    {
      "epoch": 0.09817557064550438,
      "grad_norm": 0.6054053449775758,
      "learning_rate": 9.816753926701572e-06,
      "loss": 0.0123,
      "step": 6000
    },
    {
      "epoch": 0.09833919659658022,
      "grad_norm": 0.3275187680578677,
      "learning_rate": 9.833115183246073e-06,
      "loss": 0.0141,
      "step": 6010
    },
    {
      "epoch": 0.09850282254765606,
      "grad_norm": 0.48429147082982177,
      "learning_rate": 9.849476439790576e-06,
      "loss": 0.0113,
      "step": 6020
    },
    {
      "epoch": 0.09866644849873189,
      "grad_norm": 0.24124561144188728,
      "learning_rate": 9.865837696335079e-06,
      "loss": 0.0085,
      "step": 6030
    },
    {
      "epoch": 0.09883007444980774,
      "grad_norm": 0.7436975583632291,
      "learning_rate": 9.882198952879581e-06,
      "loss": 0.0111,
      "step": 6040
    },
    {
      "epoch": 0.09899370040088358,
      "grad_norm": 0.2831754995725805,
      "learning_rate": 9.898560209424084e-06,
      "loss": 0.0103,
      "step": 6050
    },
    {
      "epoch": 0.09915732635195942,
      "grad_norm": 0.45401967463149967,
      "learning_rate": 9.914921465968587e-06,
      "loss": 0.0094,
      "step": 6060
    },
    {
      "epoch": 0.09932095230303527,
      "grad_norm": 0.5556129123150316,
      "learning_rate": 9.93128272251309e-06,
      "loss": 0.0101,
      "step": 6070
    },
    {
      "epoch": 0.0994845782541111,
      "grad_norm": 0.5305462025353201,
      "learning_rate": 9.947643979057593e-06,
      "loss": 0.0094,
      "step": 6080
    },
    {
      "epoch": 0.09964820420518694,
      "grad_norm": 0.4748836491801436,
      "learning_rate": 9.964005235602096e-06,
      "loss": 0.0143,
      "step": 6090
    },
    {
      "epoch": 0.09981183015626278,
      "grad_norm": 0.6552184020761955,
      "learning_rate": 9.980366492146599e-06,
      "loss": 0.0081,
      "step": 6100
    },
    {
      "epoch": 0.09997545610733863,
      "grad_norm": 0.2635466996072233,
      "learning_rate": 9.9967277486911e-06,
      "loss": 0.0103,
      "step": 6110
    },
    {
      "epoch": 0.10013908205841446,
      "grad_norm": 0.49095445238619656,
      "learning_rate": 9.99999947802829e-06,
      "loss": 0.0067,
      "step": 6120
    },
    {
      "epoch": 0.1003027080094903,
      "grad_norm": 0.29907236600780385,
      "learning_rate": 9.999997357518407e-06,
      "loss": 0.0149,
      "step": 6130
    },
    {
      "epoch": 0.10046633396056615,
      "grad_norm": 0.8397672481540065,
      "learning_rate": 9.99999360584781e-06,
      "loss": 0.0115,
      "step": 6140
    },
    {
      "epoch": 0.10062995991164199,
      "grad_norm": 0.9509560141068086,
      "learning_rate": 9.999988223017722e-06,
      "loss": 0.0158,
      "step": 6150
    },
    {
      "epoch": 0.10079358586271783,
      "grad_norm": 0.8010037045142904,
      "learning_rate": 9.9999812090299e-06,
      "loss": 0.0129,
      "step": 6160
    },
    {
      "epoch": 0.10095721181379366,
      "grad_norm": 0.5371395591512428,
      "learning_rate": 9.999972563886632e-06,
      "loss": 0.0136,
      "step": 6170
    },
    {
      "epoch": 0.1011208377648695,
      "grad_norm": 0.7400906954335196,
      "learning_rate": 9.999962287590739e-06,
      "loss": 0.0125,
      "step": 6180
    },
    {
      "epoch": 0.10128446371594535,
      "grad_norm": 0.4837697018772169,
      "learning_rate": 9.999950380145572e-06,
      "loss": 0.0123,
      "step": 6190
    },
    {
      "epoch": 0.1014480896670212,
      "grad_norm": 0.848980352162884,
      "learning_rate": 9.999936841555017e-06,
      "loss": 0.0116,
      "step": 6200
    },
    {
      "epoch": 0.10161171561809704,
      "grad_norm": 0.5772190766488138,
      "learning_rate": 9.999921671823487e-06,
      "loss": 0.0082,
      "step": 6210
    },
    {
      "epoch": 0.10177534156917287,
      "grad_norm": 0.38725082078719,
      "learning_rate": 9.999904870955936e-06,
      "loss": 0.0132,
      "step": 6220
    },
    {
      "epoch": 0.10193896752024871,
      "grad_norm": 0.7523447039597139,
      "learning_rate": 9.999886438957843e-06,
      "loss": 0.0105,
      "step": 6230
    },
    {
      "epoch": 0.10210259347132455,
      "grad_norm": 0.4450305963353815,
      "learning_rate": 9.999866375835221e-06,
      "loss": 0.0109,
      "step": 6240
    },
    {
      "epoch": 0.1022662194224004,
      "grad_norm": 0.4364063537617544,
      "learning_rate": 9.999844681594615e-06,
      "loss": 0.0088,
      "step": 6250
    },
    {
      "epoch": 0.10242984537347623,
      "grad_norm": 0.28014853889608476,
      "learning_rate": 9.9998213562431e-06,
      "loss": 0.0073,
      "step": 6260
    },
    {
      "epoch": 0.10259347132455207,
      "grad_norm": 0.6019459831370338,
      "learning_rate": 9.999796399788292e-06,
      "loss": 0.0102,
      "step": 6270
    },
    {
      "epoch": 0.10275709727562791,
      "grad_norm": 0.5433955101582122,
      "learning_rate": 9.999769812238327e-06,
      "loss": 0.0126,
      "step": 6280
    },
    {
      "epoch": 0.10292072322670376,
      "grad_norm": 0.7334546492569576,
      "learning_rate": 9.999741593601881e-06,
      "loss": 0.0115,
      "step": 6290
    },
    {
      "epoch": 0.1030843491777796,
      "grad_norm": 0.26043582338582244,
      "learning_rate": 9.999711743888158e-06,
      "loss": 0.0096,
      "step": 6300
    },
    {
      "epoch": 0.10324797512885543,
      "grad_norm": 0.4980737594958685,
      "learning_rate": 9.999680263106897e-06,
      "loss": 0.0101,
      "step": 6310
    },
    {
      "epoch": 0.10341160107993128,
      "grad_norm": 0.4178137912171756,
      "learning_rate": 9.99964715126837e-06,
      "loss": 0.0174,
      "step": 6320
    },
    {
      "epoch": 0.10357522703100712,
      "grad_norm": 0.7385962988512482,
      "learning_rate": 9.999612408383376e-06,
      "loss": 0.0137,
      "step": 6330
    },
    {
      "epoch": 0.10373885298208296,
      "grad_norm": 0.39826782711705155,
      "learning_rate": 9.99957603446325e-06,
      "loss": 0.0121,
      "step": 6340
    },
    {
      "epoch": 0.10390247893315879,
      "grad_norm": 0.35298963132412603,
      "learning_rate": 9.999538029519861e-06,
      "loss": 0.0104,
      "step": 6350
    },
    {
      "epoch": 0.10406610488423464,
      "grad_norm": 0.25144306119691906,
      "learning_rate": 9.999498393565603e-06,
      "loss": 0.0138,
      "step": 6360
    },
    {
      "epoch": 0.10422973083531048,
      "grad_norm": 0.5447426393131171,
      "learning_rate": 9.99945712661341e-06,
      "loss": 0.011,
      "step": 6370
    },
    {
      "epoch": 0.10439335678638632,
      "grad_norm": 0.5467705178447493,
      "learning_rate": 9.999414228676745e-06,
      "loss": 0.0088,
      "step": 6380
    },
    {
      "epoch": 0.10455698273746217,
      "grad_norm": 0.3368890437774478,
      "learning_rate": 9.9993696997696e-06,
      "loss": 0.011,
      "step": 6390
    },
    {
      "epoch": 0.104720608688538,
      "grad_norm": 0.7168666946485738,
      "learning_rate": 9.999323539906502e-06,
      "loss": 0.012,
      "step": 6400
    },
    {
      "epoch": 0.10488423463961384,
      "grad_norm": 0.845153784490695,
      "learning_rate": 9.99927574910251e-06,
      "loss": 0.0126,
      "step": 6410
    },
    {
      "epoch": 0.10504786059068968,
      "grad_norm": 0.6430680689317863,
      "learning_rate": 9.99922632737322e-06,
      "loss": 0.0111,
      "step": 6420
    },
    {
      "epoch": 0.10521148654176553,
      "grad_norm": 0.5778579407416131,
      "learning_rate": 9.999175274734748e-06,
      "loss": 0.0122,
      "step": 6430
    },
    {
      "epoch": 0.10537511249284136,
      "grad_norm": 0.3827239307022562,
      "learning_rate": 9.999122591203751e-06,
      "loss": 0.0084,
      "step": 6440
    },
    {
      "epoch": 0.1055387384439172,
      "grad_norm": 0.16237157664712298,
      "learning_rate": 9.999068276797417e-06,
      "loss": 0.0096,
      "step": 6450
    },
    {
      "epoch": 0.10570236439499305,
      "grad_norm": 0.2703578578777984,
      "learning_rate": 9.999012331533466e-06,
      "loss": 0.012,
      "step": 6460
    },
    {
      "epoch": 0.10586599034606889,
      "grad_norm": 0.5032517355744486,
      "learning_rate": 9.998954755430145e-06,
      "loss": 0.0125,
      "step": 6470
    },
    {
      "epoch": 0.10602961629714473,
      "grad_norm": 0.19448086417145694,
      "learning_rate": 9.998895548506244e-06,
      "loss": 0.0104,
      "step": 6480
    },
    {
      "epoch": 0.10619324224822056,
      "grad_norm": 1.1016969952549975,
      "learning_rate": 9.998834710781073e-06,
      "loss": 0.0143,
      "step": 6490
    },
    {
      "epoch": 0.1063568681992964,
      "grad_norm": 0.4227908717273277,
      "learning_rate": 9.99877224227448e-06,
      "loss": 0.008,
      "step": 6500
    },
    {
      "epoch": 0.10652049415037225,
      "grad_norm": 0.3763617186674828,
      "learning_rate": 9.998708143006847e-06,
      "loss": 0.0118,
      "step": 6510
    },
    {
      "epoch": 0.1066841201014481,
      "grad_norm": 0.6715878812180046,
      "learning_rate": 9.998642412999082e-06,
      "loss": 0.0121,
      "step": 6520
    },
    {
      "epoch": 0.10684774605252394,
      "grad_norm": 0.42395096022430195,
      "learning_rate": 9.998575052272629e-06,
      "loss": 0.0106,
      "step": 6530
    },
    {
      "epoch": 0.10701137200359977,
      "grad_norm": 0.6852581249690433,
      "learning_rate": 9.998506060849465e-06,
      "loss": 0.0124,
      "step": 6540
    },
    {
      "epoch": 0.10717499795467561,
      "grad_norm": 0.5030646364589404,
      "learning_rate": 9.998435438752094e-06,
      "loss": 0.0087,
      "step": 6550
    },
    {
      "epoch": 0.10733862390575145,
      "grad_norm": 0.8514707329056839,
      "learning_rate": 9.998363186003557e-06,
      "loss": 0.0092,
      "step": 6560
    },
    {
      "epoch": 0.1075022498568273,
      "grad_norm": 0.5904561981348231,
      "learning_rate": 9.998289302627427e-06,
      "loss": 0.0144,
      "step": 6570
    },
    {
      "epoch": 0.10766587580790313,
      "grad_norm": 0.4136912731015378,
      "learning_rate": 9.998213788647804e-06,
      "loss": 0.0089,
      "step": 6580
    },
    {
      "epoch": 0.10782950175897897,
      "grad_norm": 0.3640618035395279,
      "learning_rate": 9.998136644089325e-06,
      "loss": 0.0132,
      "step": 6590
    },
    {
      "epoch": 0.10799312771005481,
      "grad_norm": 0.41974650154869514,
      "learning_rate": 9.998057868977158e-06,
      "loss": 0.0115,
      "step": 6600
    },
    {
      "epoch": 0.10815675366113066,
      "grad_norm": 0.5997856513093407,
      "learning_rate": 9.997977463336999e-06,
      "loss": 0.0201,
      "step": 6610
    },
    {
      "epoch": 0.1083203796122065,
      "grad_norm": 0.6365756426066118,
      "learning_rate": 9.997895427195081e-06,
      "loss": 0.0103,
      "step": 6620
    },
    {
      "epoch": 0.10848400556328233,
      "grad_norm": 0.5737435619888953,
      "learning_rate": 9.997811760578166e-06,
      "loss": 0.0078,
      "step": 6630
    },
    {
      "epoch": 0.10864763151435818,
      "grad_norm": 0.5124106814131394,
      "learning_rate": 9.99772646351355e-06,
      "loss": 0.0146,
      "step": 6640
    },
    {
      "epoch": 0.10881125746543402,
      "grad_norm": 0.37502799696794586,
      "learning_rate": 9.997639536029057e-06,
      "loss": 0.0071,
      "step": 6650
    },
    {
      "epoch": 0.10897488341650986,
      "grad_norm": 0.3496365780914275,
      "learning_rate": 9.997550978153048e-06,
      "loss": 0.0096,
      "step": 6660
    },
    {
      "epoch": 0.10913850936758569,
      "grad_norm": 0.3543271745805264,
      "learning_rate": 9.997460789914413e-06,
      "loss": 0.0112,
      "step": 6670
    },
    {
      "epoch": 0.10930213531866154,
      "grad_norm": 0.3877645685162003,
      "learning_rate": 9.997368971342575e-06,
      "loss": 0.0097,
      "step": 6680
    },
    {
      "epoch": 0.10946576126973738,
      "grad_norm": 0.5647123121959107,
      "learning_rate": 9.997275522467486e-06,
      "loss": 0.0099,
      "step": 6690
    },
    {
      "epoch": 0.10962938722081322,
      "grad_norm": 0.3579110467335093,
      "learning_rate": 9.997180443319635e-06,
      "loss": 0.0158,
      "step": 6700
    },
    {
      "epoch": 0.10979301317188907,
      "grad_norm": 0.6510512594206854,
      "learning_rate": 9.997083733930035e-06,
      "loss": 0.0095,
      "step": 6710
    },
    {
      "epoch": 0.1099566391229649,
      "grad_norm": 0.7870841605181781,
      "learning_rate": 9.99698539433024e-06,
      "loss": 0.0171,
      "step": 6720
    },
    {
      "epoch": 0.11012026507404074,
      "grad_norm": 0.4305973730050311,
      "learning_rate": 9.996885424552332e-06,
      "loss": 0.0086,
      "step": 6730
    },
    {
      "epoch": 0.11028389102511658,
      "grad_norm": 0.5048125960477226,
      "learning_rate": 9.996783824628921e-06,
      "loss": 0.0087,
      "step": 6740
    },
    {
      "epoch": 0.11044751697619243,
      "grad_norm": 0.2471328786655937,
      "learning_rate": 9.996680594593157e-06,
      "loss": 0.0088,
      "step": 6750
    },
    {
      "epoch": 0.11061114292726827,
      "grad_norm": 0.525254154690201,
      "learning_rate": 9.996575734478711e-06,
      "loss": 0.0106,
      "step": 6760
    },
    {
      "epoch": 0.1107747688783441,
      "grad_norm": 0.3178604990623649,
      "learning_rate": 9.996469244319796e-06,
      "loss": 0.009,
      "step": 6770
    },
    {
      "epoch": 0.11093839482941995,
      "grad_norm": 0.5792632276416242,
      "learning_rate": 9.996361124151152e-06,
      "loss": 0.0095,
      "step": 6780
    },
    {
      "epoch": 0.11110202078049579,
      "grad_norm": 0.4412227804650962,
      "learning_rate": 9.996251374008049e-06,
      "loss": 0.0071,
      "step": 6790
    },
    {
      "epoch": 0.11126564673157163,
      "grad_norm": 0.1037156866228801,
      "learning_rate": 9.996139993926293e-06,
      "loss": 0.0065,
      "step": 6800
    },
    {
      "epoch": 0.11142927268264746,
      "grad_norm": 0.41941191126354493,
      "learning_rate": 9.99602698394222e-06,
      "loss": 0.0156,
      "step": 6810
    },
    {
      "epoch": 0.1115928986337233,
      "grad_norm": 0.4131215503007927,
      "learning_rate": 9.995912344092696e-06,
      "loss": 0.0141,
      "step": 6820
    },
    {
      "epoch": 0.11175652458479915,
      "grad_norm": 0.8765874065031145,
      "learning_rate": 9.995796074415123e-06,
      "loss": 0.013,
      "step": 6830
    },
    {
      "epoch": 0.111920150535875,
      "grad_norm": 0.19639102705439695,
      "learning_rate": 9.995678174947428e-06,
      "loss": 0.0113,
      "step": 6840
    },
    {
      "epoch": 0.11208377648695084,
      "grad_norm": 0.32128292673184494,
      "learning_rate": 9.995558645728076e-06,
      "loss": 0.0106,
      "step": 6850
    },
    {
      "epoch": 0.11224740243802667,
      "grad_norm": 0.48434215561962674,
      "learning_rate": 9.995437486796061e-06,
      "loss": 0.0125,
      "step": 6860
    },
    {
      "epoch": 0.11241102838910251,
      "grad_norm": 0.5729078299929122,
      "learning_rate": 9.99531469819091e-06,
      "loss": 0.0172,
      "step": 6870
    },
    {
      "epoch": 0.11257465434017835,
      "grad_norm": 1.3737458736241044,
      "learning_rate": 9.995190279952678e-06,
      "loss": 0.0124,
      "step": 6880
    },
    {
      "epoch": 0.1127382802912542,
      "grad_norm": 0.2564057594169095,
      "learning_rate": 9.995064232121958e-06,
      "loss": 0.0058,
      "step": 6890
    },
    {
      "epoch": 0.11290190624233003,
      "grad_norm": 0.1331854057011681,
      "learning_rate": 9.994936554739866e-06,
      "loss": 0.0157,
      "step": 6900
    },
    {
      "epoch": 0.11306553219340587,
      "grad_norm": 0.471743040707044,
      "learning_rate": 9.994807247848058e-06,
      "loss": 0.0104,
      "step": 6910
    },
    {
      "epoch": 0.11322915814448171,
      "grad_norm": 0.22887423957587652,
      "learning_rate": 9.994676311488718e-06,
      "loss": 0.0082,
      "step": 6920
    },
    {
      "epoch": 0.11339278409555756,
      "grad_norm": 0.5053753356765128,
      "learning_rate": 9.99454374570456e-06,
      "loss": 0.0106,
      "step": 6930
    },
    {
      "epoch": 0.1135564100466334,
      "grad_norm": 0.49521876621443284,
      "learning_rate": 9.994409550538832e-06,
      "loss": 0.0091,
      "step": 6940
    },
    {
      "epoch": 0.11372003599770923,
      "grad_norm": 0.5875042581289367,
      "learning_rate": 9.994273726035315e-06,
      "loss": 0.0074,
      "step": 6950
    },
    {
      "epoch": 0.11388366194878508,
      "grad_norm": 0.432155588278268,
      "learning_rate": 9.994136272238315e-06,
      "loss": 0.0103,
      "step": 6960
    },
    {
      "epoch": 0.11404728789986092,
      "grad_norm": 0.39031452179190923,
      "learning_rate": 9.993997189192677e-06,
      "loss": 0.0138,
      "step": 6970
    },
    {
      "epoch": 0.11421091385093676,
      "grad_norm": 0.2731926247935071,
      "learning_rate": 9.993856476943774e-06,
      "loss": 0.0122,
      "step": 6980
    },
    {
      "epoch": 0.11437453980201259,
      "grad_norm": 0.5313500920876125,
      "learning_rate": 9.99371413553751e-06,
      "loss": 0.0114,
      "step": 6990
    },
    {
      "epoch": 0.11453816575308844,
      "grad_norm": 0.43063310857539183,
      "learning_rate": 9.993570165020322e-06,
      "loss": 0.0103,
      "step": 7000
    },
    {
      "epoch": 0.11470179170416428,
      "grad_norm": 0.4672469112453224,
      "learning_rate": 9.993424565439179e-06,
      "loss": 0.0085,
      "step": 7010
    },
    {
      "epoch": 0.11486541765524012,
      "grad_norm": 0.526848647461658,
      "learning_rate": 9.993277336841576e-06,
      "loss": 0.0099,
      "step": 7020
    },
    {
      "epoch": 0.11502904360631597,
      "grad_norm": 0.5703599850373706,
      "learning_rate": 9.993128479275547e-06,
      "loss": 0.0113,
      "step": 7030
    },
    {
      "epoch": 0.1151926695573918,
      "grad_norm": 0.5586900064578564,
      "learning_rate": 9.992977992789655e-06,
      "loss": 0.0145,
      "step": 7040
    },
    {
      "epoch": 0.11535629550846764,
      "grad_norm": 0.966459270445536,
      "learning_rate": 9.992825877432992e-06,
      "loss": 0.0099,
      "step": 7050
    },
    {
      "epoch": 0.11551992145954348,
      "grad_norm": 0.20522298031196878,
      "learning_rate": 9.992672133255183e-06,
      "loss": 0.009,
      "step": 7060
    },
    {
      "epoch": 0.11568354741061933,
      "grad_norm": 0.25623169444005783,
      "learning_rate": 9.992516760306386e-06,
      "loss": 0.0123,
      "step": 7070
    },
    {
      "epoch": 0.11584717336169517,
      "grad_norm": 0.5474450895536691,
      "learning_rate": 9.992359758637287e-06,
      "loss": 0.0148,
      "step": 7080
    },
    {
      "epoch": 0.116010799312771,
      "grad_norm": 0.1622325043630987,
      "learning_rate": 9.992201128299103e-06,
      "loss": 0.0118,
      "step": 7090
    },
    {
      "epoch": 0.11617442526384684,
      "grad_norm": 0.5715358739137986,
      "learning_rate": 9.992040869343588e-06,
      "loss": 0.0111,
      "step": 7100
    },
    {
      "epoch": 0.11633805121492269,
      "grad_norm": 0.28223232876817195,
      "learning_rate": 9.991878981823025e-06,
      "loss": 0.0096,
      "step": 7110
    },
    {
      "epoch": 0.11650167716599853,
      "grad_norm": 0.4089288018832493,
      "learning_rate": 9.99171546579022e-06,
      "loss": 0.0109,
      "step": 7120
    },
    {
      "epoch": 0.11666530311707436,
      "grad_norm": 0.5723616032587838,
      "learning_rate": 9.991550321298524e-06,
      "loss": 0.0104,
      "step": 7130
    },
    {
      "epoch": 0.1168289290681502,
      "grad_norm": 0.2778648237647966,
      "learning_rate": 9.991383548401808e-06,
      "loss": 0.0068,
      "step": 7140
    },
    {
      "epoch": 0.11699255501922605,
      "grad_norm": 0.35871926200508614,
      "learning_rate": 9.991215147154483e-06,
      "loss": 0.012,
      "step": 7150
    },
    {
      "epoch": 0.11715618097030189,
      "grad_norm": 0.5135427271198006,
      "learning_rate": 9.991045117611481e-06,
      "loss": 0.0115,
      "step": 7160
    },
    {
      "epoch": 0.11731980692137774,
      "grad_norm": 0.6694865723399268,
      "learning_rate": 9.990873459828279e-06,
      "loss": 0.009,
      "step": 7170
    },
    {
      "epoch": 0.11748343287245357,
      "grad_norm": 0.5027976638256262,
      "learning_rate": 9.99070017386087e-06,
      "loss": 0.0107,
      "step": 7180
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.5759753205342629,
      "learning_rate": 9.990525259765791e-06,
      "loss": 0.0119,
      "step": 7190
    },
    {
      "epoch": 0.11781068477460525,
      "grad_norm": 0.47970539514906335,
      "learning_rate": 9.9903487176001e-06,
      "loss": 0.007,
      "step": 7200
    },
    {
      "epoch": 0.1179743107256811,
      "grad_norm": 0.5529153967992684,
      "learning_rate": 9.990170547421394e-06,
      "loss": 0.0101,
      "step": 7210
    },
    {
      "epoch": 0.11813793667675693,
      "grad_norm": 0.6756920906327544,
      "learning_rate": 9.989990749287795e-06,
      "loss": 0.0096,
      "step": 7220
    },
    {
      "epoch": 0.11830156262783277,
      "grad_norm": 0.5650664639773525,
      "learning_rate": 9.989809323257961e-06,
      "loss": 0.0072,
      "step": 7230
    },
    {
      "epoch": 0.11846518857890861,
      "grad_norm": 0.3517944147814383,
      "learning_rate": 9.989626269391081e-06,
      "loss": 0.012,
      "step": 7240
    },
    {
      "epoch": 0.11862881452998446,
      "grad_norm": 0.6821015905534964,
      "learning_rate": 9.98944158774687e-06,
      "loss": 0.0053,
      "step": 7250
    },
    {
      "epoch": 0.1187924404810603,
      "grad_norm": 0.2038384184246867,
      "learning_rate": 9.989255278385577e-06,
      "loss": 0.0144,
      "step": 7260
    },
    {
      "epoch": 0.11895606643213613,
      "grad_norm": 0.43442665684120113,
      "learning_rate": 9.989067341367985e-06,
      "loss": 0.0091,
      "step": 7270
    },
    {
      "epoch": 0.11911969238321198,
      "grad_norm": 0.3925250349094445,
      "learning_rate": 9.988877776755402e-06,
      "loss": 0.009,
      "step": 7280
    },
    {
      "epoch": 0.11928331833428782,
      "grad_norm": 0.7058911871375118,
      "learning_rate": 9.98868658460967e-06,
      "loss": 0.0116,
      "step": 7290
    },
    {
      "epoch": 0.11944694428536366,
      "grad_norm": 0.23763714170202263,
      "learning_rate": 9.988493764993163e-06,
      "loss": 0.0083,
      "step": 7300
    },
    {
      "epoch": 0.11961057023643949,
      "grad_norm": 0.4119670334210937,
      "learning_rate": 9.988299317968789e-06,
      "loss": 0.0104,
      "step": 7310
    },
    {
      "epoch": 0.11977419618751534,
      "grad_norm": 0.40369454531228166,
      "learning_rate": 9.988103243599976e-06,
      "loss": 0.0102,
      "step": 7320
    },
    {
      "epoch": 0.11993782213859118,
      "grad_norm": 0.37419035482457746,
      "learning_rate": 9.987905541950694e-06,
      "loss": 0.009,
      "step": 7330
    },
    {
      "epoch": 0.12010144808966702,
      "grad_norm": 0.6923309882836086,
      "learning_rate": 9.98770621308544e-06,
      "loss": 0.0088,
      "step": 7340
    },
    {
      "epoch": 0.12026507404074287,
      "grad_norm": 0.7772752064849809,
      "learning_rate": 9.987505257069239e-06,
      "loss": 0.0166,
      "step": 7350
    },
    {
      "epoch": 0.1204286999918187,
      "grad_norm": 0.40883090691405877,
      "learning_rate": 9.987302673967649e-06,
      "loss": 0.011,
      "step": 7360
    },
    {
      "epoch": 0.12059232594289454,
      "grad_norm": 0.6689714658315377,
      "learning_rate": 9.987098463846762e-06,
      "loss": 0.012,
      "step": 7370
    },
    {
      "epoch": 0.12075595189397038,
      "grad_norm": 0.5957023165630145,
      "learning_rate": 9.986892626773197e-06,
      "loss": 0.0089,
      "step": 7380
    },
    {
      "epoch": 0.12091957784504623,
      "grad_norm": 0.6202100805091941,
      "learning_rate": 9.986685162814106e-06,
      "loss": 0.0171,
      "step": 7390
    },
    {
      "epoch": 0.12108320379612207,
      "grad_norm": 0.6673315898986919,
      "learning_rate": 9.986476072037168e-06,
      "loss": 0.0065,
      "step": 7400
    },
    {
      "epoch": 0.1212468297471979,
      "grad_norm": 0.2529659115623473,
      "learning_rate": 9.986265354510594e-06,
      "loss": 0.0079,
      "step": 7410
    },
    {
      "epoch": 0.12141045569827374,
      "grad_norm": 0.3691320759505354,
      "learning_rate": 9.98605301030313e-06,
      "loss": 0.0128,
      "step": 7420
    },
    {
      "epoch": 0.12157408164934959,
      "grad_norm": 0.3451722725340001,
      "learning_rate": 9.985839039484048e-06,
      "loss": 0.0111,
      "step": 7430
    },
    {
      "epoch": 0.12173770760042543,
      "grad_norm": 0.1322355929776524,
      "learning_rate": 9.985623442123154e-06,
      "loss": 0.0074,
      "step": 7440
    },
    {
      "epoch": 0.12190133355150126,
      "grad_norm": 0.1302490846861022,
      "learning_rate": 9.985406218290781e-06,
      "loss": 0.01,
      "step": 7450
    },
    {
      "epoch": 0.1220649595025771,
      "grad_norm": 0.3880780292039577,
      "learning_rate": 9.985187368057794e-06,
      "loss": 0.0123,
      "step": 7460
    },
    {
      "epoch": 0.12222858545365295,
      "grad_norm": 0.21239876531689256,
      "learning_rate": 9.984966891495591e-06,
      "loss": 0.0061,
      "step": 7470
    },
    {
      "epoch": 0.12239221140472879,
      "grad_norm": 0.48722969858836135,
      "learning_rate": 9.984744788676097e-06,
      "loss": 0.0067,
      "step": 7480
    },
    {
      "epoch": 0.12255583735580464,
      "grad_norm": 0.43993669285663584,
      "learning_rate": 9.98452105967177e-06,
      "loss": 0.0085,
      "step": 7490
    },
    {
      "epoch": 0.12271946330688047,
      "grad_norm": 0.6023606851601276,
      "learning_rate": 9.984295704555595e-06,
      "loss": 0.0109,
      "step": 7500
    },
    {
      "epoch": 0.12288308925795631,
      "grad_norm": 0.5471127678796841,
      "learning_rate": 9.984068723401095e-06,
      "loss": 0.0152,
      "step": 7510
    },
    {
      "epoch": 0.12304671520903215,
      "grad_norm": 1.0448909116485974,
      "learning_rate": 9.983840116282315e-06,
      "loss": 0.011,
      "step": 7520
    },
    {
      "epoch": 0.123210341160108,
      "grad_norm": 0.6978748804873199,
      "learning_rate": 9.983609883273835e-06,
      "loss": 0.0072,
      "step": 7530
    },
    {
      "epoch": 0.12337396711118383,
      "grad_norm": 0.5155271110289986,
      "learning_rate": 9.983378024450765e-06,
      "loss": 0.0121,
      "step": 7540
    },
    {
      "epoch": 0.12353759306225967,
      "grad_norm": 0.33877927060760543,
      "learning_rate": 9.983144539888743e-06,
      "loss": 0.0115,
      "step": 7550
    },
    {
      "epoch": 0.12370121901333551,
      "grad_norm": 0.2542065258467828,
      "learning_rate": 9.98290942966394e-06,
      "loss": 0.0094,
      "step": 7560
    },
    {
      "epoch": 0.12386484496441136,
      "grad_norm": 0.5112176686821556,
      "learning_rate": 9.98267269385306e-06,
      "loss": 0.0092,
      "step": 7570
    },
    {
      "epoch": 0.1240284709154872,
      "grad_norm": 0.31840976149965144,
      "learning_rate": 9.982434332533328e-06,
      "loss": 0.0082,
      "step": 7580
    },
    {
      "epoch": 0.12419209686656303,
      "grad_norm": 0.4189409873171635,
      "learning_rate": 9.98219434578251e-06,
      "loss": 0.013,
      "step": 7590
    },
    {
      "epoch": 0.12435572281763888,
      "grad_norm": 0.6527638038825173,
      "learning_rate": 9.981952733678893e-06,
      "loss": 0.0101,
      "step": 7600
    },
    {
      "epoch": 0.12451934876871472,
      "grad_norm": 0.43006567211478064,
      "learning_rate": 9.981709496301303e-06,
      "loss": 0.0095,
      "step": 7610
    },
    {
      "epoch": 0.12468297471979056,
      "grad_norm": 0.6122561192216692,
      "learning_rate": 9.98146463372909e-06,
      "loss": 0.0111,
      "step": 7620
    },
    {
      "epoch": 0.1248466006708664,
      "grad_norm": 0.4927626535205494,
      "learning_rate": 9.981218146042135e-06,
      "loss": 0.0085,
      "step": 7630
    },
    {
      "epoch": 0.12501022662194225,
      "grad_norm": 0.674376503918369,
      "learning_rate": 9.98097003332085e-06,
      "loss": 0.0083,
      "step": 7640
    },
    {
      "epoch": 0.12517385257301808,
      "grad_norm": 0.7374741360877392,
      "learning_rate": 9.980720295646182e-06,
      "loss": 0.0079,
      "step": 7650
    },
    {
      "epoch": 0.1253374785240939,
      "grad_norm": 1.252409135462966,
      "learning_rate": 9.9804689330996e-06,
      "loss": 0.0093,
      "step": 7660
    },
    {
      "epoch": 0.12550110447516977,
      "grad_norm": 0.36368397724751045,
      "learning_rate": 9.980215945763105e-06,
      "loss": 0.0076,
      "step": 7670
    },
    {
      "epoch": 0.1256647304262456,
      "grad_norm": 0.38366978348564307,
      "learning_rate": 9.979961333719234e-06,
      "loss": 0.0118,
      "step": 7680
    },
    {
      "epoch": 0.12582835637732145,
      "grad_norm": 0.4499159108380496,
      "learning_rate": 9.979705097051045e-06,
      "loss": 0.0147,
      "step": 7690
    },
    {
      "epoch": 0.12599198232839728,
      "grad_norm": 0.38279883429941414,
      "learning_rate": 9.979447235842133e-06,
      "loss": 0.0085,
      "step": 7700
    },
    {
      "epoch": 0.1261556082794731,
      "grad_norm": 0.3327792071790146,
      "learning_rate": 9.979187750176622e-06,
      "loss": 0.0066,
      "step": 7710
    },
    {
      "epoch": 0.12631923423054897,
      "grad_norm": 0.3730433963890892,
      "learning_rate": 9.978926640139161e-06,
      "loss": 0.0082,
      "step": 7720
    },
    {
      "epoch": 0.1264828601816248,
      "grad_norm": 0.19140166994682384,
      "learning_rate": 9.978663905814935e-06,
      "loss": 0.0085,
      "step": 7730
    },
    {
      "epoch": 0.12664648613270066,
      "grad_norm": 0.5242504399307966,
      "learning_rate": 9.97839954728966e-06,
      "loss": 0.0104,
      "step": 7740
    },
    {
      "epoch": 0.1268101120837765,
      "grad_norm": 0.8910352378541504,
      "learning_rate": 9.97813356464957e-06,
      "loss": 0.0095,
      "step": 7750
    },
    {
      "epoch": 0.12697373803485232,
      "grad_norm": 0.5119339834424127,
      "learning_rate": 9.977865957981442e-06,
      "loss": 0.0115,
      "step": 7760
    },
    {
      "epoch": 0.12713736398592818,
      "grad_norm": 0.4549393064494094,
      "learning_rate": 9.977596727372579e-06,
      "loss": 0.0128,
      "step": 7770
    },
    {
      "epoch": 0.127300989937004,
      "grad_norm": 0.5718418789768672,
      "learning_rate": 9.97732587291081e-06,
      "loss": 0.0073,
      "step": 7780
    },
    {
      "epoch": 0.12746461588807986,
      "grad_norm": 0.5031150658063366,
      "learning_rate": 9.977053394684497e-06,
      "loss": 0.0112,
      "step": 7790
    },
    {
      "epoch": 0.1276282418391557,
      "grad_norm": 0.33779627638730514,
      "learning_rate": 9.976779292782534e-06,
      "loss": 0.0125,
      "step": 7800
    },
    {
      "epoch": 0.12779186779023152,
      "grad_norm": 0.5834386709489626,
      "learning_rate": 9.976503567294338e-06,
      "loss": 0.0128,
      "step": 7810
    },
    {
      "epoch": 0.12795549374130738,
      "grad_norm": 0.5065300223268314,
      "learning_rate": 9.976226218309861e-06,
      "loss": 0.0056,
      "step": 7820
    },
    {
      "epoch": 0.1281191196923832,
      "grad_norm": 0.39285655863263635,
      "learning_rate": 9.975947245919586e-06,
      "loss": 0.0087,
      "step": 7830
    },
    {
      "epoch": 0.12828274564345904,
      "grad_norm": 0.6351355020650205,
      "learning_rate": 9.975666650214519e-06,
      "loss": 0.0102,
      "step": 7840
    },
    {
      "epoch": 0.1284463715945349,
      "grad_norm": 0.2293904799000577,
      "learning_rate": 9.975384431286199e-06,
      "loss": 0.0061,
      "step": 7850
    },
    {
      "epoch": 0.12860999754561073,
      "grad_norm": 0.29463279243069634,
      "learning_rate": 9.975100589226698e-06,
      "loss": 0.0115,
      "step": 7860
    },
    {
      "epoch": 0.12877362349668658,
      "grad_norm": 0.3681066221815506,
      "learning_rate": 9.974815124128613e-06,
      "loss": 0.0082,
      "step": 7870
    },
    {
      "epoch": 0.12893724944776241,
      "grad_norm": 0.37929925136191706,
      "learning_rate": 9.974528036085073e-06,
      "loss": 0.01,
      "step": 7880
    },
    {
      "epoch": 0.12910087539883824,
      "grad_norm": 0.3984684069318568,
      "learning_rate": 9.974239325189733e-06,
      "loss": 0.0101,
      "step": 7890
    },
    {
      "epoch": 0.1292645013499141,
      "grad_norm": 0.4409635828897977,
      "learning_rate": 9.973948991536782e-06,
      "loss": 0.0115,
      "step": 7900
    },
    {
      "epoch": 0.12942812730098993,
      "grad_norm": 0.5819005272996403,
      "learning_rate": 9.973657035220936e-06,
      "loss": 0.0122,
      "step": 7910
    },
    {
      "epoch": 0.1295917532520658,
      "grad_norm": 0.2964183404331756,
      "learning_rate": 9.973363456337439e-06,
      "loss": 0.0067,
      "step": 7920
    },
    {
      "epoch": 0.12975537920314162,
      "grad_norm": 0.513040956545932,
      "learning_rate": 9.973068254982067e-06,
      "loss": 0.0087,
      "step": 7930
    },
    {
      "epoch": 0.12991900515421745,
      "grad_norm": 0.24495945014448808,
      "learning_rate": 9.972771431251126e-06,
      "loss": 0.0073,
      "step": 7940
    },
    {
      "epoch": 0.1300826311052933,
      "grad_norm": 0.2918552635809396,
      "learning_rate": 9.972472985241446e-06,
      "loss": 0.0101,
      "step": 7950
    },
    {
      "epoch": 0.13024625705636914,
      "grad_norm": 0.4772610801439484,
      "learning_rate": 9.97217291705039e-06,
      "loss": 0.008,
      "step": 7960
    },
    {
      "epoch": 0.130409883007445,
      "grad_norm": 0.28689235850898825,
      "learning_rate": 9.971871226775856e-06,
      "loss": 0.0084,
      "step": 7970
    },
    {
      "epoch": 0.13057350895852082,
      "grad_norm": 0.3211163110539691,
      "learning_rate": 9.971567914516256e-06,
      "loss": 0.0079,
      "step": 7980
    },
    {
      "epoch": 0.13073713490959665,
      "grad_norm": 0.6895975282977052,
      "learning_rate": 9.971262980370548e-06,
      "loss": 0.0107,
      "step": 7990
    },
    {
      "epoch": 0.1309007608606725,
      "grad_norm": 0.4930688444804981,
      "learning_rate": 9.970956424438206e-06,
      "loss": 0.0114,
      "step": 8000
    },
    {
      "epoch": 0.13106438681174834,
      "grad_norm": 0.6669313344247826,
      "learning_rate": 9.970648246819242e-06,
      "loss": 0.0106,
      "step": 8010
    },
    {
      "epoch": 0.1312280127628242,
      "grad_norm": 0.610351134090419,
      "learning_rate": 9.970338447614189e-06,
      "loss": 0.0075,
      "step": 8020
    },
    {
      "epoch": 0.13139163871390003,
      "grad_norm": 0.7731612332068099,
      "learning_rate": 9.97002702692412e-06,
      "loss": 0.0088,
      "step": 8030
    },
    {
      "epoch": 0.13155526466497586,
      "grad_norm": 0.2435463441612163,
      "learning_rate": 9.969713984850626e-06,
      "loss": 0.0082,
      "step": 8040
    },
    {
      "epoch": 0.13171889061605171,
      "grad_norm": 0.3702965455878276,
      "learning_rate": 9.969399321495831e-06,
      "loss": 0.0073,
      "step": 8050
    },
    {
      "epoch": 0.13188251656712754,
      "grad_norm": 0.8532321708797265,
      "learning_rate": 9.969083036962393e-06,
      "loss": 0.0055,
      "step": 8060
    },
    {
      "epoch": 0.13204614251820337,
      "grad_norm": 0.4440392363683305,
      "learning_rate": 9.968765131353488e-06,
      "loss": 0.0077,
      "step": 8070
    },
    {
      "epoch": 0.13220976846927923,
      "grad_norm": 0.47211443767887445,
      "learning_rate": 9.968445604772831e-06,
      "loss": 0.0093,
      "step": 8080
    },
    {
      "epoch": 0.13237339442035506,
      "grad_norm": 0.3864580530825567,
      "learning_rate": 9.968124457324663e-06,
      "loss": 0.0085,
      "step": 8090
    },
    {
      "epoch": 0.13253702037143092,
      "grad_norm": 0.9491863296092485,
      "learning_rate": 9.96780168911375e-06,
      "loss": 0.0069,
      "step": 8100
    },
    {
      "epoch": 0.13270064632250675,
      "grad_norm": 0.8790948819826223,
      "learning_rate": 9.967477300245388e-06,
      "loss": 0.0079,
      "step": 8110
    },
    {
      "epoch": 0.13286427227358258,
      "grad_norm": 0.4761717724864765,
      "learning_rate": 9.967151290825408e-06,
      "loss": 0.0079,
      "step": 8120
    },
    {
      "epoch": 0.13302789822465844,
      "grad_norm": 0.5055332139402398,
      "learning_rate": 9.966823660960162e-06,
      "loss": 0.0094,
      "step": 8130
    },
    {
      "epoch": 0.13319152417573427,
      "grad_norm": 0.5833829337591888,
      "learning_rate": 9.966494410756531e-06,
      "loss": 0.0065,
      "step": 8140
    },
    {
      "epoch": 0.13335515012681012,
      "grad_norm": 0.45025717390775377,
      "learning_rate": 9.96616354032193e-06,
      "loss": 0.0091,
      "step": 8150
    },
    {
      "epoch": 0.13351877607788595,
      "grad_norm": 0.9173263873215725,
      "learning_rate": 9.965831049764302e-06,
      "loss": 0.0124,
      "step": 8160
    },
    {
      "epoch": 0.13368240202896178,
      "grad_norm": 0.5106457676127525,
      "learning_rate": 9.965496939192113e-06,
      "loss": 0.0072,
      "step": 8170
    },
    {
      "epoch": 0.13384602798003764,
      "grad_norm": 0.4423149537287774,
      "learning_rate": 9.965161208714359e-06,
      "loss": 0.0107,
      "step": 8180
    },
    {
      "epoch": 0.13400965393111347,
      "grad_norm": 0.30649577869750927,
      "learning_rate": 9.96482385844057e-06,
      "loss": 0.0077,
      "step": 8190
    },
    {
      "epoch": 0.13417327988218933,
      "grad_norm": 0.41568687447353264,
      "learning_rate": 9.964484888480797e-06,
      "loss": 0.0143,
      "step": 8200
    },
    {
      "epoch": 0.13433690583326516,
      "grad_norm": 0.5147852339554367,
      "learning_rate": 9.964144298945625e-06,
      "loss": 0.0093,
      "step": 8210
    },
    {
      "epoch": 0.134500531784341,
      "grad_norm": 0.40569518941094834,
      "learning_rate": 9.963802089946166e-06,
      "loss": 0.009,
      "step": 8220
    },
    {
      "epoch": 0.13466415773541685,
      "grad_norm": 0.4371374936449431,
      "learning_rate": 9.96345826159406e-06,
      "loss": 0.0096,
      "step": 8230
    },
    {
      "epoch": 0.13482778368649267,
      "grad_norm": 0.3884647954271551,
      "learning_rate": 9.96311281400147e-06,
      "loss": 0.0098,
      "step": 8240
    },
    {
      "epoch": 0.1349914096375685,
      "grad_norm": 0.32999156031145854,
      "learning_rate": 9.962765747281097e-06,
      "loss": 0.0074,
      "step": 8250
    },
    {
      "epoch": 0.13515503558864436,
      "grad_norm": 0.5058323665158481,
      "learning_rate": 9.962417061546166e-06,
      "loss": 0.009,
      "step": 8260
    },
    {
      "epoch": 0.1353186615397202,
      "grad_norm": 0.3371150273031882,
      "learning_rate": 9.962066756910427e-06,
      "loss": 0.0075,
      "step": 8270
    },
    {
      "epoch": 0.13548228749079605,
      "grad_norm": 0.4022507568444807,
      "learning_rate": 9.96171483348816e-06,
      "loss": 0.0077,
      "step": 8280
    },
    {
      "epoch": 0.13564591344187188,
      "grad_norm": 0.6587497644861972,
      "learning_rate": 9.961361291394176e-06,
      "loss": 0.0106,
      "step": 8290
    },
    {
      "epoch": 0.1358095393929477,
      "grad_norm": 0.6405174898465515,
      "learning_rate": 9.961006130743811e-06,
      "loss": 0.0119,
      "step": 8300
    },
    {
      "epoch": 0.13597316534402357,
      "grad_norm": 0.37482383762293103,
      "learning_rate": 9.960649351652931e-06,
      "loss": 0.0138,
      "step": 8310
    },
    {
      "epoch": 0.1361367912950994,
      "grad_norm": 0.2964876803929516,
      "learning_rate": 9.960290954237927e-06,
      "loss": 0.0098,
      "step": 8320
    },
    {
      "epoch": 0.13630041724617525,
      "grad_norm": 0.39809389169501386,
      "learning_rate": 9.959930938615721e-06,
      "loss": 0.0084,
      "step": 8330
    },
    {
      "epoch": 0.13646404319725108,
      "grad_norm": 0.5289905951932483,
      "learning_rate": 9.95956930490376e-06,
      "loss": 0.0095,
      "step": 8340
    },
    {
      "epoch": 0.1366276691483269,
      "grad_norm": 0.24298209907434354,
      "learning_rate": 9.959206053220023e-06,
      "loss": 0.0075,
      "step": 8350
    },
    {
      "epoch": 0.13679129509940277,
      "grad_norm": 0.3546188098797019,
      "learning_rate": 9.958841183683014e-06,
      "loss": 0.0055,
      "step": 8360
    },
    {
      "epoch": 0.1369549210504786,
      "grad_norm": 0.2880727102093247,
      "learning_rate": 9.958474696411763e-06,
      "loss": 0.0064,
      "step": 8370
    },
    {
      "epoch": 0.13711854700155446,
      "grad_norm": 0.6116985426477467,
      "learning_rate": 9.958106591525833e-06,
      "loss": 0.0096,
      "step": 8380
    },
    {
      "epoch": 0.1372821729526303,
      "grad_norm": 0.5298539295490508,
      "learning_rate": 9.95773686914531e-06,
      "loss": 0.0081,
      "step": 8390
    },
    {
      "epoch": 0.13744579890370612,
      "grad_norm": 0.4365517299152412,
      "learning_rate": 9.95736552939081e-06,
      "loss": 0.0106,
      "step": 8400
    },
    {
      "epoch": 0.13760942485478198,
      "grad_norm": 0.47156272106008296,
      "learning_rate": 9.956992572383474e-06,
      "loss": 0.0073,
      "step": 8410
    },
    {
      "epoch": 0.1377730508058578,
      "grad_norm": 0.3090850942231373,
      "learning_rate": 9.956617998244975e-06,
      "loss": 0.0073,
      "step": 8420
    },
    {
      "epoch": 0.13793667675693366,
      "grad_norm": 0.5692173622575482,
      "learning_rate": 9.95624180709751e-06,
      "loss": 0.0108,
      "step": 8430
    },
    {
      "epoch": 0.1381003027080095,
      "grad_norm": 0.21130202639782922,
      "learning_rate": 9.955863999063809e-06,
      "loss": 0.009,
      "step": 8440
    },
    {
      "epoch": 0.13826392865908532,
      "grad_norm": 1.2382408730524779,
      "learning_rate": 9.955484574267118e-06,
      "loss": 0.0086,
      "step": 8450
    },
    {
      "epoch": 0.13842755461016118,
      "grad_norm": 0.4020863358384348,
      "learning_rate": 9.955103532831223e-06,
      "loss": 0.0085,
      "step": 8460
    },
    {
      "epoch": 0.138591180561237,
      "grad_norm": 0.1747274571972482,
      "learning_rate": 9.954720874880429e-06,
      "loss": 0.011,
      "step": 8470
    },
    {
      "epoch": 0.13875480651231284,
      "grad_norm": 0.617611697190206,
      "learning_rate": 9.954336600539572e-06,
      "loss": 0.0089,
      "step": 8480
    },
    {
      "epoch": 0.1389184324633887,
      "grad_norm": 0.5432507166102423,
      "learning_rate": 9.953950709934015e-06,
      "loss": 0.0122,
      "step": 8490
    },
    {
      "epoch": 0.13908205841446453,
      "grad_norm": 0.6456532716126964,
      "learning_rate": 9.95356320318965e-06,
      "loss": 0.008,
      "step": 8500
    },
    {
      "epoch": 0.13924568436554038,
      "grad_norm": 0.6509873850301885,
      "learning_rate": 9.95317408043289e-06,
      "loss": 0.0096,
      "step": 8510
    },
    {
      "epoch": 0.13940931031661621,
      "grad_norm": 0.3961443727664188,
      "learning_rate": 9.952783341790683e-06,
      "loss": 0.0068,
      "step": 8520
    },
    {
      "epoch": 0.13957293626769204,
      "grad_norm": 0.2451510181407701,
      "learning_rate": 9.952390987390499e-06,
      "loss": 0.011,
      "step": 8530
    },
    {
      "epoch": 0.1397365622187679,
      "grad_norm": 0.4924171652507646,
      "learning_rate": 9.951997017360337e-06,
      "loss": 0.0088,
      "step": 8540
    },
    {
      "epoch": 0.13990018816984373,
      "grad_norm": 0.2381077667610896,
      "learning_rate": 9.951601431828724e-06,
      "loss": 0.0089,
      "step": 8550
    },
    {
      "epoch": 0.1400638141209196,
      "grad_norm": 0.4134252017787631,
      "learning_rate": 9.951204230924712e-06,
      "loss": 0.0082,
      "step": 8560
    },
    {
      "epoch": 0.14022744007199542,
      "grad_norm": 0.2219257577304448,
      "learning_rate": 9.950805414777879e-06,
      "loss": 0.0093,
      "step": 8570
    },
    {
      "epoch": 0.14039106602307125,
      "grad_norm": 0.7723556396810369,
      "learning_rate": 9.950404983518334e-06,
      "loss": 0.0097,
      "step": 8580
    },
    {
      "epoch": 0.1405546919741471,
      "grad_norm": 0.7952814794637963,
      "learning_rate": 9.950002937276708e-06,
      "loss": 0.0121,
      "step": 8590
    },
    {
      "epoch": 0.14071831792522294,
      "grad_norm": 0.380027843914542,
      "learning_rate": 9.949599276184165e-06,
      "loss": 0.0109,
      "step": 8600
    },
    {
      "epoch": 0.1408819438762988,
      "grad_norm": 0.37471098615637444,
      "learning_rate": 9.949194000372391e-06,
      "loss": 0.0067,
      "step": 8610
    },
    {
      "epoch": 0.14104556982737462,
      "grad_norm": 0.3440514636109666,
      "learning_rate": 9.948787109973598e-06,
      "loss": 0.0078,
      "step": 8620
    },
    {
      "epoch": 0.14120919577845045,
      "grad_norm": 0.424943435511282,
      "learning_rate": 9.948378605120528e-06,
      "loss": 0.01,
      "step": 8630
    },
    {
      "epoch": 0.1413728217295263,
      "grad_norm": 0.16893986040428946,
      "learning_rate": 9.94796848594645e-06,
      "loss": 0.0125,
      "step": 8640
    },
    {
      "epoch": 0.14153644768060214,
      "grad_norm": 0.15839320773848795,
      "learning_rate": 9.947556752585158e-06,
      "loss": 0.0098,
      "step": 8650
    },
    {
      "epoch": 0.141700073631678,
      "grad_norm": 0.4259802341373496,
      "learning_rate": 9.94714340517097e-06,
      "loss": 0.0073,
      "step": 8660
    },
    {
      "epoch": 0.14186369958275383,
      "grad_norm": 0.6112867083304985,
      "learning_rate": 9.946728443838735e-06,
      "loss": 0.0115,
      "step": 8670
    },
    {
      "epoch": 0.14202732553382966,
      "grad_norm": 0.4635477358331691,
      "learning_rate": 9.946311868723826e-06,
      "loss": 0.0089,
      "step": 8680
    },
    {
      "epoch": 0.14219095148490551,
      "grad_norm": 0.3613685922466621,
      "learning_rate": 9.945893679962146e-06,
      "loss": 0.007,
      "step": 8690
    },
    {
      "epoch": 0.14235457743598134,
      "grad_norm": 0.29280344832715,
      "learning_rate": 9.94547387769012e-06,
      "loss": 0.0101,
      "step": 8700
    },
    {
      "epoch": 0.14251820338705717,
      "grad_norm": 0.3903999029193127,
      "learning_rate": 9.945052462044698e-06,
      "loss": 0.0093,
      "step": 8710
    },
    {
      "epoch": 0.14268182933813303,
      "grad_norm": 0.6670602607382556,
      "learning_rate": 9.944629433163365e-06,
      "loss": 0.0117,
      "step": 8720
    },
    {
      "epoch": 0.14284545528920886,
      "grad_norm": 0.2554563727979691,
      "learning_rate": 9.944204791184123e-06,
      "loss": 0.0146,
      "step": 8730
    },
    {
      "epoch": 0.14300908124028472,
      "grad_norm": 0.5172661304058259,
      "learning_rate": 9.943778536245505e-06,
      "loss": 0.009,
      "step": 8740
    },
    {
      "epoch": 0.14317270719136055,
      "grad_norm": 0.5401029861472235,
      "learning_rate": 9.943350668486568e-06,
      "loss": 0.0105,
      "step": 8750
    },
    {
      "epoch": 0.14333633314243638,
      "grad_norm": 0.35109286699254905,
      "learning_rate": 9.942921188046898e-06,
      "loss": 0.0083,
      "step": 8760
    },
    {
      "epoch": 0.14349995909351224,
      "grad_norm": 0.8117302205244186,
      "learning_rate": 9.942490095066603e-06,
      "loss": 0.0108,
      "step": 8770
    },
    {
      "epoch": 0.14366358504458807,
      "grad_norm": 0.18422880204362338,
      "learning_rate": 9.942057389686323e-06,
      "loss": 0.0067,
      "step": 8780
    },
    {
      "epoch": 0.14382721099566392,
      "grad_norm": 1.0442420844936615,
      "learning_rate": 9.94162307204722e-06,
      "loss": 0.0078,
      "step": 8790
    },
    {
      "epoch": 0.14399083694673975,
      "grad_norm": 0.5708013045768598,
      "learning_rate": 9.941187142290978e-06,
      "loss": 0.0081,
      "step": 8800
    },
    {
      "epoch": 0.14415446289781558,
      "grad_norm": 0.46354191840015313,
      "learning_rate": 9.940749600559816e-06,
      "loss": 0.0056,
      "step": 8810
    },
    {
      "epoch": 0.14431808884889144,
      "grad_norm": 0.19860903530881147,
      "learning_rate": 9.940310446996474e-06,
      "loss": 0.0078,
      "step": 8820
    },
    {
      "epoch": 0.14448171479996727,
      "grad_norm": 0.4625361162855827,
      "learning_rate": 9.939869681744215e-06,
      "loss": 0.0083,
      "step": 8830
    },
    {
      "epoch": 0.14464534075104313,
      "grad_norm": 0.043661821918211566,
      "learning_rate": 9.939427304946832e-06,
      "loss": 0.0053,
      "step": 8840
    },
    {
      "epoch": 0.14480896670211896,
      "grad_norm": 0.5937446254846646,
      "learning_rate": 9.938983316748643e-06,
      "loss": 0.0083,
      "step": 8850
    },
    {
      "epoch": 0.1449725926531948,
      "grad_norm": 0.14140342932585184,
      "learning_rate": 9.938537717294492e-06,
      "loss": 0.0098,
      "step": 8860
    },
    {
      "epoch": 0.14513621860427064,
      "grad_norm": 0.3285297438534549,
      "learning_rate": 9.93809050672975e-06,
      "loss": 0.0077,
      "step": 8870
    },
    {
      "epoch": 0.14529984455534647,
      "grad_norm": 0.16116408629552414,
      "learning_rate": 9.937641685200306e-06,
      "loss": 0.0096,
      "step": 8880
    },
    {
      "epoch": 0.14546347050642233,
      "grad_norm": 0.2635515471787533,
      "learning_rate": 9.937191252852583e-06,
      "loss": 0.0076,
      "step": 8890
    },
    {
      "epoch": 0.14562709645749816,
      "grad_norm": 0.45473968670725573,
      "learning_rate": 9.936739209833528e-06,
      "loss": 0.0072,
      "step": 8900
    },
    {
      "epoch": 0.145790722408574,
      "grad_norm": 0.3446584360047905,
      "learning_rate": 9.93628555629061e-06,
      "loss": 0.0067,
      "step": 8910
    },
    {
      "epoch": 0.14595434835964985,
      "grad_norm": 0.6062019616633063,
      "learning_rate": 9.935830292371827e-06,
      "loss": 0.0093,
      "step": 8920
    },
    {
      "epoch": 0.14611797431072568,
      "grad_norm": 0.21348975147000424,
      "learning_rate": 9.9353734182257e-06,
      "loss": 0.0093,
      "step": 8930
    },
    {
      "epoch": 0.1462816002618015,
      "grad_norm": 0.6169294096149263,
      "learning_rate": 9.934914934001274e-06,
      "loss": 0.0071,
      "step": 8940
    },
    {
      "epoch": 0.14644522621287737,
      "grad_norm": 0.308857607310664,
      "learning_rate": 9.934454839848126e-06,
      "loss": 0.0073,
      "step": 8950
    },
    {
      "epoch": 0.1466088521639532,
      "grad_norm": 0.42150804636863715,
      "learning_rate": 9.93399313591635e-06,
      "loss": 0.0063,
      "step": 8960
    },
    {
      "epoch": 0.14677247811502905,
      "grad_norm": 0.14321617806529321,
      "learning_rate": 9.933529822356568e-06,
      "loss": 0.0082,
      "step": 8970
    },
    {
      "epoch": 0.14693610406610488,
      "grad_norm": 0.4233447666638266,
      "learning_rate": 9.933064899319933e-06,
      "loss": 0.0081,
      "step": 8980
    },
    {
      "epoch": 0.1470997300171807,
      "grad_norm": 0.5036144380917476,
      "learning_rate": 9.932598366958113e-06,
      "loss": 0.0075,
      "step": 8990
    },
    {
      "epoch": 0.14726335596825657,
      "grad_norm": 0.3378704556119975,
      "learning_rate": 9.932130225423307e-06,
      "loss": 0.0067,
      "step": 9000
    },
    {
      "epoch": 0.1474269819193324,
      "grad_norm": 0.44832351842906576,
      "learning_rate": 9.931660474868238e-06,
      "loss": 0.0091,
      "step": 9010
    },
    {
      "epoch": 0.14759060787040826,
      "grad_norm": 0.3124134853489913,
      "learning_rate": 9.931189115446156e-06,
      "loss": 0.0079,
      "step": 9020
    },
    {
      "epoch": 0.1477542338214841,
      "grad_norm": 0.376129076267134,
      "learning_rate": 9.93071614731083e-06,
      "loss": 0.0081,
      "step": 9030
    },
    {
      "epoch": 0.14791785977255992,
      "grad_norm": 0.5592900785105428,
      "learning_rate": 9.930241570616561e-06,
      "loss": 0.0084,
      "step": 9040
    },
    {
      "epoch": 0.14808148572363578,
      "grad_norm": 0.3045366401986836,
      "learning_rate": 9.92976538551817e-06,
      "loss": 0.0087,
      "step": 9050
    },
    {
      "epoch": 0.1482451116747116,
      "grad_norm": 0.2669190363544682,
      "learning_rate": 9.929287592171002e-06,
      "loss": 0.0087,
      "step": 9060
    },
    {
      "epoch": 0.14840873762578746,
      "grad_norm": 0.36518718513976595,
      "learning_rate": 9.92880819073093e-06,
      "loss": 0.01,
      "step": 9070
    },
    {
      "epoch": 0.1485723635768633,
      "grad_norm": 0.588493426328603,
      "learning_rate": 9.928327181354352e-06,
      "loss": 0.0102,
      "step": 9080
    },
    {
      "epoch": 0.14873598952793912,
      "grad_norm": 0.5717443119906511,
      "learning_rate": 9.927844564198186e-06,
      "loss": 0.0062,
      "step": 9090
    },
    {
      "epoch": 0.14889961547901498,
      "grad_norm": 0.9134875475071091,
      "learning_rate": 9.927360339419878e-06,
      "loss": 0.0082,
      "step": 9100
    },
    {
      "epoch": 0.1490632414300908,
      "grad_norm": 0.2879624478961297,
      "learning_rate": 9.9268745071774e-06,
      "loss": 0.0073,
      "step": 9110
    },
    {
      "epoch": 0.14922686738116667,
      "grad_norm": 0.8029359234190678,
      "learning_rate": 9.926387067629243e-06,
      "loss": 0.014,
      "step": 9120
    },
    {
      "epoch": 0.1493904933322425,
      "grad_norm": 0.37907731222138397,
      "learning_rate": 9.925898020934428e-06,
      "loss": 0.0098,
      "step": 9130
    },
    {
      "epoch": 0.14955411928331833,
      "grad_norm": 0.2476944483508969,
      "learning_rate": 9.925407367252497e-06,
      "loss": 0.008,
      "step": 9140
    },
    {
      "epoch": 0.14971774523439418,
      "grad_norm": 0.4204750739847926,
      "learning_rate": 9.924915106743515e-06,
      "loss": 0.0074,
      "step": 9150
    },
    {
      "epoch": 0.14988137118547,
      "grad_norm": 0.7074599029391628,
      "learning_rate": 9.924421239568077e-06,
      "loss": 0.0117,
      "step": 9160
    },
    {
      "epoch": 0.15004499713654584,
      "grad_norm": 0.2677756013199778,
      "learning_rate": 9.923925765887296e-06,
      "loss": 0.0087,
      "step": 9170
    },
    {
      "epoch": 0.1502086230876217,
      "grad_norm": 0.5697800424731361,
      "learning_rate": 9.923428685862814e-06,
      "loss": 0.0079,
      "step": 9180
    },
    {
      "epoch": 0.15037224903869753,
      "grad_norm": 0.8009000925321772,
      "learning_rate": 9.92292999965679e-06,
      "loss": 0.0076,
      "step": 9190
    },
    {
      "epoch": 0.1505358749897734,
      "grad_norm": 0.49303097539862706,
      "learning_rate": 9.922429707431917e-06,
      "loss": 0.0088,
      "step": 9200
    },
    {
      "epoch": 0.15069950094084922,
      "grad_norm": 0.6974439715725193,
      "learning_rate": 9.921927809351402e-06,
      "loss": 0.0075,
      "step": 9210
    },
    {
      "epoch": 0.15086312689192505,
      "grad_norm": 0.19476468926689297,
      "learning_rate": 9.921424305578983e-06,
      "loss": 0.0053,
      "step": 9220
    },
    {
      "epoch": 0.1510267528430009,
      "grad_norm": 0.37303715697780476,
      "learning_rate": 9.920919196278917e-06,
      "loss": 0.0083,
      "step": 9230
    },
    {
      "epoch": 0.15119037879407674,
      "grad_norm": 0.2224756456525914,
      "learning_rate": 9.92041248161599e-06,
      "loss": 0.0065,
      "step": 9240
    },
    {
      "epoch": 0.1513540047451526,
      "grad_norm": 0.21778292086119141,
      "learning_rate": 9.919904161755507e-06,
      "loss": 0.0058,
      "step": 9250
    },
    {
      "epoch": 0.15151763069622842,
      "grad_norm": 0.2713447061718825,
      "learning_rate": 9.919394236863298e-06,
      "loss": 0.0077,
      "step": 9260
    },
    {
      "epoch": 0.15168125664730425,
      "grad_norm": 0.35129680145951364,
      "learning_rate": 9.918882707105717e-06,
      "loss": 0.007,
      "step": 9270
    },
    {
      "epoch": 0.1518448825983801,
      "grad_norm": 0.6536664463782678,
      "learning_rate": 9.918369572649643e-06,
      "loss": 0.006,
      "step": 9280
    },
    {
      "epoch": 0.15200850854945594,
      "grad_norm": 0.47511884991950143,
      "learning_rate": 9.917854833662474e-06,
      "loss": 0.0071,
      "step": 9290
    },
    {
      "epoch": 0.1521721345005318,
      "grad_norm": 0.5483570606410505,
      "learning_rate": 9.917338490312138e-06,
      "loss": 0.0072,
      "step": 9300
    },
    {
      "epoch": 0.15233576045160763,
      "grad_norm": 0.3916762090305378,
      "learning_rate": 9.916820542767082e-06,
      "loss": 0.0122,
      "step": 9310
    },
    {
      "epoch": 0.15249938640268346,
      "grad_norm": 0.4146095761940944,
      "learning_rate": 9.916300991196275e-06,
      "loss": 0.0095,
      "step": 9320
    },
    {
      "epoch": 0.15266301235375931,
      "grad_norm": 0.18649352333468477,
      "learning_rate": 9.915779835769215e-06,
      "loss": 0.0062,
      "step": 9330
    },
    {
      "epoch": 0.15282663830483514,
      "grad_norm": 0.4991207924409798,
      "learning_rate": 9.915257076655915e-06,
      "loss": 0.0104,
      "step": 9340
    },
    {
      "epoch": 0.15299026425591097,
      "grad_norm": 0.574098452266438,
      "learning_rate": 9.91473271402692e-06,
      "loss": 0.0146,
      "step": 9350
    },
    {
      "epoch": 0.15315389020698683,
      "grad_norm": 0.5249645618883285,
      "learning_rate": 9.914206748053294e-06,
      "loss": 0.0088,
      "step": 9360
    },
    {
      "epoch": 0.15331751615806266,
      "grad_norm": 0.550680337462673,
      "learning_rate": 9.913679178906622e-06,
      "loss": 0.0101,
      "step": 9370
    },
    {
      "epoch": 0.15348114210913852,
      "grad_norm": 0.3011147529976332,
      "learning_rate": 9.913150006759016e-06,
      "loss": 0.0089,
      "step": 9380
    },
    {
      "epoch": 0.15364476806021435,
      "grad_norm": 0.5461743807658227,
      "learning_rate": 9.912619231783108e-06,
      "loss": 0.0065,
      "step": 9390
    },
    {
      "epoch": 0.15380839401129018,
      "grad_norm": 0.4394247789462552,
      "learning_rate": 9.912086854152052e-06,
      "loss": 0.0082,
      "step": 9400
    },
    {
      "epoch": 0.15397201996236604,
      "grad_norm": 0.34739636345630137,
      "learning_rate": 9.91155287403953e-06,
      "loss": 0.0096,
      "step": 9410
    },
    {
      "epoch": 0.15413564591344187,
      "grad_norm": 0.5506210556273866,
      "learning_rate": 9.911017291619742e-06,
      "loss": 0.006,
      "step": 9420
    },
    {
      "epoch": 0.15429927186451772,
      "grad_norm": 0.6869432462297704,
      "learning_rate": 9.910480107067412e-06,
      "loss": 0.0092,
      "step": 9430
    },
    {
      "epoch": 0.15446289781559355,
      "grad_norm": 0.3028731878883146,
      "learning_rate": 9.90994132055779e-06,
      "loss": 0.0087,
      "step": 9440
    },
    {
      "epoch": 0.15462652376666938,
      "grad_norm": 0.5596070179186514,
      "learning_rate": 9.909400932266643e-06,
      "loss": 0.01,
      "step": 9450
    },
    {
      "epoch": 0.15479014971774524,
      "grad_norm": 0.6744786881582322,
      "learning_rate": 9.90885894237026e-06,
      "loss": 0.0097,
      "step": 9460
    },
    {
      "epoch": 0.15495377566882107,
      "grad_norm": 0.535044468377577,
      "learning_rate": 9.90831535104546e-06,
      "loss": 0.0091,
      "step": 9470
    },
    {
      "epoch": 0.15511740161989693,
      "grad_norm": 0.5124968937806973,
      "learning_rate": 9.907770158469581e-06,
      "loss": 0.0096,
      "step": 9480
    },
    {
      "epoch": 0.15528102757097276,
      "grad_norm": 0.211693740141867,
      "learning_rate": 9.90722336482048e-06,
      "loss": 0.0094,
      "step": 9490
    },
    {
      "epoch": 0.1554446535220486,
      "grad_norm": 0.56360984591575,
      "learning_rate": 9.906674970276538e-06,
      "loss": 0.0067,
      "step": 9500
    },
    {
      "epoch": 0.15560827947312444,
      "grad_norm": 0.40719023869943,
      "learning_rate": 9.90612497501666e-06,
      "loss": 0.0073,
      "step": 9510
    },
    {
      "epoch": 0.15577190542420027,
      "grad_norm": 0.226362732371626,
      "learning_rate": 9.905573379220274e-06,
      "loss": 0.0052,
      "step": 9520
    },
    {
      "epoch": 0.15593553137527613,
      "grad_norm": 0.16824055838680885,
      "learning_rate": 9.905020183067326e-06,
      "loss": 0.0104,
      "step": 9530
    },
    {
      "epoch": 0.15609915732635196,
      "grad_norm": 0.2439869592434656,
      "learning_rate": 9.904465386738287e-06,
      "loss": 0.0068,
      "step": 9540
    },
    {
      "epoch": 0.1562627832774278,
      "grad_norm": 0.3589889660316506,
      "learning_rate": 9.90390899041415e-06,
      "loss": 0.0079,
      "step": 9550
    },
    {
      "epoch": 0.15642640922850365,
      "grad_norm": 0.581687246316274,
      "learning_rate": 9.903350994276429e-06,
      "loss": 0.0073,
      "step": 9560
    },
    {
      "epoch": 0.15659003517957948,
      "grad_norm": 0.11769508423524805,
      "learning_rate": 9.902791398507162e-06,
      "loss": 0.0099,
      "step": 9570
    },
    {
      "epoch": 0.1567536611306553,
      "grad_norm": 0.2404613429511316,
      "learning_rate": 9.902230203288902e-06,
      "loss": 0.0074,
      "step": 9580
    },
    {
      "epoch": 0.15691728708173117,
      "grad_norm": 0.3787016102887237,
      "learning_rate": 9.901667408804736e-06,
      "loss": 0.01,
      "step": 9590
    },
    {
      "epoch": 0.157080913032807,
      "grad_norm": 0.48468770378930476,
      "learning_rate": 9.901103015238262e-06,
      "loss": 0.0112,
      "step": 9600
    },
    {
      "epoch": 0.15724453898388285,
      "grad_norm": 0.40627445634562653,
      "learning_rate": 9.900537022773605e-06,
      "loss": 0.0078,
      "step": 9610
    },
    {
      "epoch": 0.15740816493495868,
      "grad_norm": 0.25892633792073505,
      "learning_rate": 9.899969431595406e-06,
      "loss": 0.0067,
      "step": 9620
    },
    {
      "epoch": 0.1575717908860345,
      "grad_norm": 0.19838158906940834,
      "learning_rate": 9.899400241888837e-06,
      "loss": 0.0052,
      "step": 9630
    },
    {
      "epoch": 0.15773541683711037,
      "grad_norm": 0.5851929222273135,
      "learning_rate": 9.898829453839584e-06,
      "loss": 0.0117,
      "step": 9640
    },
    {
      "epoch": 0.1578990427881862,
      "grad_norm": 0.26023426338765165,
      "learning_rate": 9.898257067633853e-06,
      "loss": 0.0085,
      "step": 9650
    },
    {
      "epoch": 0.15806266873926206,
      "grad_norm": 0.35403532811040644,
      "learning_rate": 9.897683083458382e-06,
      "loss": 0.0078,
      "step": 9660
    },
    {
      "epoch": 0.1582262946903379,
      "grad_norm": 0.3324359511178061,
      "learning_rate": 9.897107501500418e-06,
      "loss": 0.0098,
      "step": 9670
    },
    {
      "epoch": 0.15838992064141372,
      "grad_norm": 0.48517397962770487,
      "learning_rate": 9.896530321947734e-06,
      "loss": 0.0095,
      "step": 9680
    },
    {
      "epoch": 0.15855354659248957,
      "grad_norm": 0.47146398077149093,
      "learning_rate": 9.895951544988629e-06,
      "loss": 0.0113,
      "step": 9690
    },
    {
      "epoch": 0.1587171725435654,
      "grad_norm": 0.3270636987118544,
      "learning_rate": 9.895371170811915e-06,
      "loss": 0.0077,
      "step": 9700
    },
    {
      "epoch": 0.15888079849464126,
      "grad_norm": 0.5261811419817289,
      "learning_rate": 9.89478919960693e-06,
      "loss": 0.0073,
      "step": 9710
    },
    {
      "epoch": 0.1590444244457171,
      "grad_norm": 0.18257772348583526,
      "learning_rate": 9.89420563156353e-06,
      "loss": 0.0055,
      "step": 9720
    },
    {
      "epoch": 0.15920805039679292,
      "grad_norm": 0.43962987638433737,
      "learning_rate": 9.893620466872097e-06,
      "loss": 0.008,
      "step": 9730
    },
    {
      "epoch": 0.15937167634786878,
      "grad_norm": 0.3663453261638167,
      "learning_rate": 9.893033705723529e-06,
      "loss": 0.0052,
      "step": 9740
    },
    {
      "epoch": 0.1595353022989446,
      "grad_norm": 0.859832464154158,
      "learning_rate": 9.892445348309245e-06,
      "loss": 0.0075,
      "step": 9750
    },
    {
      "epoch": 0.15969892825002047,
      "grad_norm": 0.4476234566964949,
      "learning_rate": 9.89185539482119e-06,
      "loss": 0.0063,
      "step": 9760
    },
    {
      "epoch": 0.1598625542010963,
      "grad_norm": 0.16726253059105453,
      "learning_rate": 9.891263845451823e-06,
      "loss": 0.0076,
      "step": 9770
    },
    {
      "epoch": 0.16002618015217213,
      "grad_norm": 0.1659356537933866,
      "learning_rate": 9.890670700394125e-06,
      "loss": 0.0069,
      "step": 9780
    },
    {
      "epoch": 0.16018980610324798,
      "grad_norm": 0.5108017461498804,
      "learning_rate": 9.890075959841604e-06,
      "loss": 0.0085,
      "step": 9790
    },
    {
      "epoch": 0.1603534320543238,
      "grad_norm": 0.4302016027300465,
      "learning_rate": 9.889479623988281e-06,
      "loss": 0.0095,
      "step": 9800
    },
    {
      "epoch": 0.16051705800539964,
      "grad_norm": 0.37388041313357845,
      "learning_rate": 9.888881693028697e-06,
      "loss": 0.0089,
      "step": 9810
    },
    {
      "epoch": 0.1606806839564755,
      "grad_norm": 0.3038370863097063,
      "learning_rate": 9.888282167157921e-06,
      "loss": 0.0075,
      "step": 9820
    },
    {
      "epoch": 0.16084430990755133,
      "grad_norm": 0.8798560569350088,
      "learning_rate": 9.887681046571535e-06,
      "loss": 0.0092,
      "step": 9830
    },
    {
      "epoch": 0.1610079358586272,
      "grad_norm": 0.39983846609676804,
      "learning_rate": 9.887078331465648e-06,
      "loss": 0.0073,
      "step": 9840
    },
    {
      "epoch": 0.16117156180970302,
      "grad_norm": 0.3288416908084513,
      "learning_rate": 9.886474022036878e-06,
      "loss": 0.0108,
      "step": 9850
    },
    {
      "epoch": 0.16133518776077885,
      "grad_norm": 0.21542943838620762,
      "learning_rate": 9.885868118482376e-06,
      "loss": 0.0097,
      "step": 9860
    },
    {
      "epoch": 0.1614988137118547,
      "grad_norm": 0.7118021771700417,
      "learning_rate": 9.885260620999805e-06,
      "loss": 0.0095,
      "step": 9870
    },
    {
      "epoch": 0.16166243966293053,
      "grad_norm": 0.39639184416249734,
      "learning_rate": 9.884651529787353e-06,
      "loss": 0.0087,
      "step": 9880
    },
    {
      "epoch": 0.1618260656140064,
      "grad_norm": 0.5077956662498805,
      "learning_rate": 9.884040845043721e-06,
      "loss": 0.0041,
      "step": 9890
    },
    {
      "epoch": 0.16198969156508222,
      "grad_norm": 0.4065907434295841,
      "learning_rate": 9.883428566968136e-06,
      "loss": 0.0091,
      "step": 9900
    },
    {
      "epoch": 0.16215331751615805,
      "grad_norm": 0.47204222186076167,
      "learning_rate": 9.882814695760345e-06,
      "loss": 0.0092,
      "step": 9910
    },
    {
      "epoch": 0.1623169434672339,
      "grad_norm": 0.1558916093685597,
      "learning_rate": 9.88219923162061e-06,
      "loss": 0.0068,
      "step": 9920
    },
    {
      "epoch": 0.16248056941830974,
      "grad_norm": 0.20919770550817593,
      "learning_rate": 9.881582174749715e-06,
      "loss": 0.008,
      "step": 9930
    },
    {
      "epoch": 0.1626441953693856,
      "grad_norm": 0.23932957273879332,
      "learning_rate": 9.880963525348966e-06,
      "loss": 0.0098,
      "step": 9940
    },
    {
      "epoch": 0.16280782132046143,
      "grad_norm": 0.2068389087463516,
      "learning_rate": 9.880343283620187e-06,
      "loss": 0.0094,
      "step": 9950
    },
    {
      "epoch": 0.16297144727153726,
      "grad_norm": 0.14699290218634523,
      "learning_rate": 9.879721449765715e-06,
      "loss": 0.0042,
      "step": 9960
    },
    {
      "epoch": 0.16313507322261311,
      "grad_norm": 0.36538710189305573,
      "learning_rate": 9.87909802398842e-06,
      "loss": 0.0091,
      "step": 9970
    },
    {
      "epoch": 0.16329869917368894,
      "grad_norm": 0.33013233856342156,
      "learning_rate": 9.87847300649168e-06,
      "loss": 0.0064,
      "step": 9980
    },
    {
      "epoch": 0.1634623251247648,
      "grad_norm": 0.32196683340121124,
      "learning_rate": 9.877846397479395e-06,
      "loss": 0.0093,
      "step": 9990
    },
    {
      "epoch": 0.16362595107584063,
      "grad_norm": 0.309836342011727,
      "learning_rate": 9.877218197155987e-06,
      "loss": 0.0058,
      "step": 10000
    },
    {
      "epoch": 0.16378957702691646,
      "grad_norm": 0.322799701026625,
      "learning_rate": 9.876588405726395e-06,
      "loss": 0.0099,
      "step": 10010
    },
    {
      "epoch": 0.16395320297799232,
      "grad_norm": 0.25984899724981264,
      "learning_rate": 9.875957023396077e-06,
      "loss": 0.0082,
      "step": 10020
    },
    {
      "epoch": 0.16411682892906815,
      "grad_norm": 0.4110625467661034,
      "learning_rate": 9.875324050371009e-06,
      "loss": 0.0095,
      "step": 10030
    },
    {
      "epoch": 0.16428045488014398,
      "grad_norm": 0.4428619115890564,
      "learning_rate": 9.874689486857688e-06,
      "loss": 0.0079,
      "step": 10040
    },
    {
      "epoch": 0.16444408083121984,
      "grad_norm": 0.2970819924248066,
      "learning_rate": 9.874053333063129e-06,
      "loss": 0.0074,
      "step": 10050
    },
    {
      "epoch": 0.16460770678229567,
      "grad_norm": 0.43446315284570247,
      "learning_rate": 9.873415589194868e-06,
      "loss": 0.0067,
      "step": 10060
    },
    {
      "epoch": 0.16477133273337152,
      "grad_norm": 0.5383267442342676,
      "learning_rate": 9.872776255460957e-06,
      "loss": 0.0089,
      "step": 10070
    },
    {
      "epoch": 0.16493495868444735,
      "grad_norm": 0.3985499659950599,
      "learning_rate": 9.872135332069964e-06,
      "loss": 0.0082,
      "step": 10080
    },
    {
      "epoch": 0.16509858463552318,
      "grad_norm": 0.4268874543393266,
      "learning_rate": 9.871492819230983e-06,
      "loss": 0.008,
      "step": 10090
    },
    {
      "epoch": 0.16526221058659904,
      "grad_norm": 0.6020027020977168,
      "learning_rate": 9.870848717153621e-06,
      "loss": 0.0082,
      "step": 10100
    },
    {
      "epoch": 0.16542583653767487,
      "grad_norm": 0.5374988523185453,
      "learning_rate": 9.870203026048004e-06,
      "loss": 0.0101,
      "step": 10110
    },
    {
      "epoch": 0.16558946248875073,
      "grad_norm": 0.4329202167058763,
      "learning_rate": 9.869555746124777e-06,
      "loss": 0.007,
      "step": 10120
    },
    {
      "epoch": 0.16575308843982656,
      "grad_norm": 0.4535929033520635,
      "learning_rate": 9.868906877595107e-06,
      "loss": 0.0091,
      "step": 10130
    },
    {
      "epoch": 0.1659167143909024,
      "grad_norm": 0.2632666687701405,
      "learning_rate": 9.86825642067067e-06,
      "loss": 0.0065,
      "step": 10140
    },
    {
      "epoch": 0.16608034034197824,
      "grad_norm": 0.36879767811462943,
      "learning_rate": 9.867604375563671e-06,
      "loss": 0.0097,
      "step": 10150
    },
    {
      "epoch": 0.16624396629305407,
      "grad_norm": 0.291099772187893,
      "learning_rate": 9.866950742486829e-06,
      "loss": 0.0087,
      "step": 10160
    },
    {
      "epoch": 0.16640759224412993,
      "grad_norm": 0.2844259989266916,
      "learning_rate": 9.866295521653376e-06,
      "loss": 0.0083,
      "step": 10170
    },
    {
      "epoch": 0.16657121819520576,
      "grad_norm": 0.5125757783765755,
      "learning_rate": 9.865638713277068e-06,
      "loss": 0.01,
      "step": 10180
    },
    {
      "epoch": 0.1667348441462816,
      "grad_norm": 0.46441647891567794,
      "learning_rate": 9.864980317572178e-06,
      "loss": 0.0041,
      "step": 10190
    },
    {
      "epoch": 0.16689847009735745,
      "grad_norm": 0.31501133107570845,
      "learning_rate": 9.864320334753495e-06,
      "loss": 0.0072,
      "step": 10200
    },
    {
      "epoch": 0.16706209604843328,
      "grad_norm": 0.38613445544791253,
      "learning_rate": 9.863658765036328e-06,
      "loss": 0.0075,
      "step": 10210
    },
    {
      "epoch": 0.16722572199950914,
      "grad_norm": 0.17437420262687167,
      "learning_rate": 9.862995608636502e-06,
      "loss": 0.0063,
      "step": 10220
    },
    {
      "epoch": 0.16738934795058497,
      "grad_norm": 0.5324733380918414,
      "learning_rate": 9.862330865770357e-06,
      "loss": 0.0073,
      "step": 10230
    },
    {
      "epoch": 0.1675529739016608,
      "grad_norm": 0.5242492073416941,
      "learning_rate": 9.861664536654757e-06,
      "loss": 0.01,
      "step": 10240
    },
    {
      "epoch": 0.16771659985273665,
      "grad_norm": 0.20473056698530814,
      "learning_rate": 9.86099662150708e-06,
      "loss": 0.0071,
      "step": 10250
    },
    {
      "epoch": 0.16788022580381248,
      "grad_norm": 0.27294945119052566,
      "learning_rate": 9.86032712054522e-06,
      "loss": 0.0074,
      "step": 10260
    },
    {
      "epoch": 0.1680438517548883,
      "grad_norm": 0.3483303015911527,
      "learning_rate": 9.85965603398759e-06,
      "loss": 0.0106,
      "step": 10270
    },
    {
      "epoch": 0.16820747770596417,
      "grad_norm": 0.2328859732347586,
      "learning_rate": 9.85898336205312e-06,
      "loss": 0.0093,
      "step": 10280
    },
    {
      "epoch": 0.16837110365704,
      "grad_norm": 0.3140498645292364,
      "learning_rate": 9.85830910496126e-06,
      "loss": 0.0071,
      "step": 10290
    },
    {
      "epoch": 0.16853472960811586,
      "grad_norm": 0.5199400245549783,
      "learning_rate": 9.857633262931972e-06,
      "loss": 0.0102,
      "step": 10300
    },
    {
      "epoch": 0.1686983555591917,
      "grad_norm": 0.4056932278604273,
      "learning_rate": 9.856955836185736e-06,
      "loss": 0.0083,
      "step": 10310
    },
    {
      "epoch": 0.16886198151026752,
      "grad_norm": 0.12662900147618225,
      "learning_rate": 9.856276824943553e-06,
      "loss": 0.0084,
      "step": 10320
    },
    {
      "epoch": 0.16902560746134337,
      "grad_norm": 0.25840266007922685,
      "learning_rate": 9.855596229426941e-06,
      "loss": 0.01,
      "step": 10330
    },
    {
      "epoch": 0.1691892334124192,
      "grad_norm": 0.4803474933005574,
      "learning_rate": 9.854914049857926e-06,
      "loss": 0.0068,
      "step": 10340
    },
    {
      "epoch": 0.16935285936349506,
      "grad_norm": 0.5893516603927258,
      "learning_rate": 9.854230286459062e-06,
      "loss": 0.0093,
      "step": 10350
    },
    {
      "epoch": 0.1695164853145709,
      "grad_norm": 0.41429833719847015,
      "learning_rate": 9.85354493945341e-06,
      "loss": 0.0084,
      "step": 10360
    },
    {
      "epoch": 0.16968011126564672,
      "grad_norm": 0.14762429963862791,
      "learning_rate": 9.852858009064557e-06,
      "loss": 0.0064,
      "step": 10370
    },
    {
      "epoch": 0.16984373721672258,
      "grad_norm": 0.25709373600296614,
      "learning_rate": 9.852169495516599e-06,
      "loss": 0.0063,
      "step": 10380
    },
    {
      "epoch": 0.1700073631677984,
      "grad_norm": 0.2652830553115959,
      "learning_rate": 9.851479399034153e-06,
      "loss": 0.0127,
      "step": 10390
    },
    {
      "epoch": 0.17017098911887427,
      "grad_norm": 0.8985472934503602,
      "learning_rate": 9.850787719842351e-06,
      "loss": 0.0071,
      "step": 10400
    },
    {
      "epoch": 0.1703346150699501,
      "grad_norm": 0.18685012772041115,
      "learning_rate": 9.850094458166841e-06,
      "loss": 0.0074,
      "step": 10410
    },
    {
      "epoch": 0.17049824102102593,
      "grad_norm": 0.1525854988689108,
      "learning_rate": 9.849399614233785e-06,
      "loss": 0.0064,
      "step": 10420
    },
    {
      "epoch": 0.17066186697210178,
      "grad_norm": 0.4601118642797104,
      "learning_rate": 9.848703188269865e-06,
      "loss": 0.0097,
      "step": 10430
    },
    {
      "epoch": 0.1708254929231776,
      "grad_norm": 0.38392863390009024,
      "learning_rate": 9.848005180502279e-06,
      "loss": 0.0056,
      "step": 10440
    },
    {
      "epoch": 0.17098911887425344,
      "grad_norm": 0.21364795629404754,
      "learning_rate": 9.847305591158738e-06,
      "loss": 0.0052,
      "step": 10450
    },
    {
      "epoch": 0.1711527448253293,
      "grad_norm": 0.1429460745271687,
      "learning_rate": 9.846604420467471e-06,
      "loss": 0.0089,
      "step": 10460
    },
    {
      "epoch": 0.17131637077640513,
      "grad_norm": 0.4801257314849713,
      "learning_rate": 9.845901668657224e-06,
      "loss": 0.0083,
      "step": 10470
    },
    {
      "epoch": 0.171479996727481,
      "grad_norm": 0.6128505151573106,
      "learning_rate": 9.845197335957255e-06,
      "loss": 0.0106,
      "step": 10480
    },
    {
      "epoch": 0.17164362267855682,
      "grad_norm": 0.4370845575132257,
      "learning_rate": 9.84449142259734e-06,
      "loss": 0.0119,
      "step": 10490
    },
    {
      "epoch": 0.17180724862963265,
      "grad_norm": 0.5414983436641663,
      "learning_rate": 9.843783928807774e-06,
      "loss": 0.0055,
      "step": 10500
    },
    {
      "epoch": 0.1719708745807085,
      "grad_norm": 0.3138193315040035,
      "learning_rate": 9.843074854819362e-06,
      "loss": 0.0093,
      "step": 10510
    },
    {
      "epoch": 0.17213450053178433,
      "grad_norm": 0.5963417187091092,
      "learning_rate": 9.842364200863425e-06,
      "loss": 0.009,
      "step": 10520
    },
    {
      "epoch": 0.1722981264828602,
      "grad_norm": 0.5131871421267759,
      "learning_rate": 9.841651967171806e-06,
      "loss": 0.0079,
      "step": 10530
    },
    {
      "epoch": 0.17246175243393602,
      "grad_norm": 0.2833442046353934,
      "learning_rate": 9.840938153976856e-06,
      "loss": 0.0096,
      "step": 10540
    },
    {
      "epoch": 0.17262537838501185,
      "grad_norm": 0.43053219249403707,
      "learning_rate": 9.840222761511442e-06,
      "loss": 0.0089,
      "step": 10550
    },
    {
      "epoch": 0.1727890043360877,
      "grad_norm": 0.18665372389328408,
      "learning_rate": 9.839505790008951e-06,
      "loss": 0.0088,
      "step": 10560
    },
    {
      "epoch": 0.17295263028716354,
      "grad_norm": 0.27656579583746305,
      "learning_rate": 9.838787239703281e-06,
      "loss": 0.0098,
      "step": 10570
    },
    {
      "epoch": 0.1731162562382394,
      "grad_norm": 0.1720743109569887,
      "learning_rate": 9.838067110828848e-06,
      "loss": 0.0068,
      "step": 10580
    },
    {
      "epoch": 0.17327988218931523,
      "grad_norm": 0.527159024174209,
      "learning_rate": 9.837345403620579e-06,
      "loss": 0.0087,
      "step": 10590
    },
    {
      "epoch": 0.17344350814039106,
      "grad_norm": 0.3281411502061594,
      "learning_rate": 9.836622118313918e-06,
      "loss": 0.0097,
      "step": 10600
    },
    {
      "epoch": 0.1736071340914669,
      "grad_norm": 0.26966717061620776,
      "learning_rate": 9.835897255144828e-06,
      "loss": 0.0049,
      "step": 10610
    },
    {
      "epoch": 0.17377076004254274,
      "grad_norm": 0.37903177565896995,
      "learning_rate": 9.83517081434978e-06,
      "loss": 0.0073,
      "step": 10620
    },
    {
      "epoch": 0.1739343859936186,
      "grad_norm": 0.3433705280406555,
      "learning_rate": 9.83444279616576e-06,
      "loss": 0.0057,
      "step": 10630
    },
    {
      "epoch": 0.17409801194469443,
      "grad_norm": 0.4002972917486334,
      "learning_rate": 9.833713200830274e-06,
      "loss": 0.0073,
      "step": 10640
    },
    {
      "epoch": 0.17426163789577026,
      "grad_norm": 0.38881258312647543,
      "learning_rate": 9.832982028581341e-06,
      "loss": 0.0055,
      "step": 10650
    },
    {
      "epoch": 0.17442526384684612,
      "grad_norm": 0.35746573325509196,
      "learning_rate": 9.83224927965749e-06,
      "loss": 0.006,
      "step": 10660
    },
    {
      "epoch": 0.17458888979792195,
      "grad_norm": 0.4625603782162722,
      "learning_rate": 9.83151495429777e-06,
      "loss": 0.0062,
      "step": 10670
    },
    {
      "epoch": 0.17475251574899778,
      "grad_norm": 0.522601737733597,
      "learning_rate": 9.83077905274174e-06,
      "loss": 0.0111,
      "step": 10680
    },
    {
      "epoch": 0.17491614170007364,
      "grad_norm": 0.2978727947406584,
      "learning_rate": 9.830041575229473e-06,
      "loss": 0.0056,
      "step": 10690
    },
    {
      "epoch": 0.17507976765114946,
      "grad_norm": 0.3216176564452935,
      "learning_rate": 9.82930252200156e-06,
      "loss": 0.0079,
      "step": 10700
    },
    {
      "epoch": 0.17524339360222532,
      "grad_norm": 0.2855433876944901,
      "learning_rate": 9.828561893299106e-06,
      "loss": 0.0059,
      "step": 10710
    },
    {
      "epoch": 0.17540701955330115,
      "grad_norm": 0.3647883468307729,
      "learning_rate": 9.827819689363725e-06,
      "loss": 0.0075,
      "step": 10720
    },
    {
      "epoch": 0.17557064550437698,
      "grad_norm": 0.10036363130058357,
      "learning_rate": 9.82707591043755e-06,
      "loss": 0.007,
      "step": 10730
    },
    {
      "epoch": 0.17573427145545284,
      "grad_norm": 0.43802238484688444,
      "learning_rate": 9.826330556763225e-06,
      "loss": 0.0067,
      "step": 10740
    },
    {
      "epoch": 0.17589789740652867,
      "grad_norm": 0.569053628296172,
      "learning_rate": 9.825583628583907e-06,
      "loss": 0.0042,
      "step": 10750
    },
    {
      "epoch": 0.17606152335760453,
      "grad_norm": 0.584931561283576,
      "learning_rate": 9.824835126143269e-06,
      "loss": 0.0085,
      "step": 10760
    },
    {
      "epoch": 0.17622514930868036,
      "grad_norm": 0.2941199978430101,
      "learning_rate": 9.8240850496855e-06,
      "loss": 0.0073,
      "step": 10770
    },
    {
      "epoch": 0.1763887752597562,
      "grad_norm": 0.6186404002665649,
      "learning_rate": 9.823333399455293e-06,
      "loss": 0.0065,
      "step": 10780
    },
    {
      "epoch": 0.17655240121083204,
      "grad_norm": 0.4555420816456537,
      "learning_rate": 9.822580175697864e-06,
      "loss": 0.0057,
      "step": 10790
    },
    {
      "epoch": 0.17671602716190787,
      "grad_norm": 0.14560680073374588,
      "learning_rate": 9.82182537865894e-06,
      "loss": 0.0054,
      "step": 10800
    },
    {
      "epoch": 0.17687965311298373,
      "grad_norm": 0.23037198840494935,
      "learning_rate": 9.821069008584759e-06,
      "loss": 0.0062,
      "step": 10810
    },
    {
      "epoch": 0.17704327906405956,
      "grad_norm": 0.09358854796123231,
      "learning_rate": 9.820311065722072e-06,
      "loss": 0.0088,
      "step": 10820
    },
    {
      "epoch": 0.1772069050151354,
      "grad_norm": 0.10902212768373744,
      "learning_rate": 9.819551550318147e-06,
      "loss": 0.0051,
      "step": 10830
    },
    {
      "epoch": 0.17737053096621125,
      "grad_norm": 0.29398243222901055,
      "learning_rate": 9.818790462620762e-06,
      "loss": 0.0048,
      "step": 10840
    },
    {
      "epoch": 0.17753415691728708,
      "grad_norm": 0.21289324190222603,
      "learning_rate": 9.818027802878204e-06,
      "loss": 0.0079,
      "step": 10850
    },
    {
      "epoch": 0.17769778286836294,
      "grad_norm": 0.2985632379121235,
      "learning_rate": 9.817263571339285e-06,
      "loss": 0.0066,
      "step": 10860
    },
    {
      "epoch": 0.17786140881943877,
      "grad_norm": 0.30960720423749627,
      "learning_rate": 9.816497768253317e-06,
      "loss": 0.0078,
      "step": 10870
    },
    {
      "epoch": 0.1780250347705146,
      "grad_norm": 0.32659478410833875,
      "learning_rate": 9.81573039387013e-06,
      "loss": 0.0066,
      "step": 10880
    },
    {
      "epoch": 0.17818866072159045,
      "grad_norm": 0.2689840558915546,
      "learning_rate": 9.814961448440066e-06,
      "loss": 0.0065,
      "step": 10890
    },
    {
      "epoch": 0.17835228667266628,
      "grad_norm": 0.3050141845385428,
      "learning_rate": 9.814190932213981e-06,
      "loss": 0.0071,
      "step": 10900
    },
    {
      "epoch": 0.1785159126237421,
      "grad_norm": 0.7256849726430792,
      "learning_rate": 9.813418845443242e-06,
      "loss": 0.0072,
      "step": 10910
    },
    {
      "epoch": 0.17867953857481797,
      "grad_norm": 0.27098188971047327,
      "learning_rate": 9.812645188379729e-06,
      "loss": 0.0051,
      "step": 10920
    },
    {
      "epoch": 0.1788431645258938,
      "grad_norm": 0.05073862591655996,
      "learning_rate": 9.811869961275834e-06,
      "loss": 0.0087,
      "step": 10930
    },
    {
      "epoch": 0.17900679047696966,
      "grad_norm": 0.16464891380734534,
      "learning_rate": 9.811093164384459e-06,
      "loss": 0.0057,
      "step": 10940
    },
    {
      "epoch": 0.1791704164280455,
      "grad_norm": 0.4692611579591489,
      "learning_rate": 9.810314797959024e-06,
      "loss": 0.0084,
      "step": 10950
    },
    {
      "epoch": 0.17933404237912132,
      "grad_norm": 0.36374923824547417,
      "learning_rate": 9.809534862253456e-06,
      "loss": 0.0087,
      "step": 10960
    },
    {
      "epoch": 0.17949766833019717,
      "grad_norm": 0.36037975071253286,
      "learning_rate": 9.808753357522193e-06,
      "loss": 0.0084,
      "step": 10970
    },
    {
      "epoch": 0.179661294281273,
      "grad_norm": 0.2744567048343302,
      "learning_rate": 9.80797028402019e-06,
      "loss": 0.0067,
      "step": 10980
    },
    {
      "epoch": 0.17982492023234886,
      "grad_norm": 0.3101395009251681,
      "learning_rate": 9.807185642002907e-06,
      "loss": 0.0076,
      "step": 10990
    },
    {
      "epoch": 0.1799885461834247,
      "grad_norm": 0.27151334583454845,
      "learning_rate": 9.806399431726326e-06,
      "loss": 0.0082,
      "step": 11000
    },
    {
      "epoch": 0.18015217213450052,
      "grad_norm": 0.3093504215518516,
      "learning_rate": 9.805611653446926e-06,
      "loss": 0.0077,
      "step": 11010
    },
    {
      "epoch": 0.18031579808557638,
      "grad_norm": 0.1842202760155649,
      "learning_rate": 9.804822307421713e-06,
      "loss": 0.0086,
      "step": 11020
    },
    {
      "epoch": 0.1804794240366522,
      "grad_norm": 0.17955579595185978,
      "learning_rate": 9.804031393908192e-06,
      "loss": 0.0097,
      "step": 11030
    },
    {
      "epoch": 0.18064304998772807,
      "grad_norm": 0.4727951966746553,
      "learning_rate": 9.80323891316439e-06,
      "loss": 0.0088,
      "step": 11040
    },
    {
      "epoch": 0.1808066759388039,
      "grad_norm": 0.304157099812122,
      "learning_rate": 9.802444865448834e-06,
      "loss": 0.0063,
      "step": 11050
    },
    {
      "epoch": 0.18097030188987973,
      "grad_norm": 0.48102946848628697,
      "learning_rate": 9.80164925102057e-06,
      "loss": 0.0087,
      "step": 11060
    },
    {
      "epoch": 0.18113392784095558,
      "grad_norm": 0.17445604439930787,
      "learning_rate": 9.800852070139154e-06,
      "loss": 0.0083,
      "step": 11070
    },
    {
      "epoch": 0.1812975537920314,
      "grad_norm": 0.2521407754152959,
      "learning_rate": 9.800053323064655e-06,
      "loss": 0.0108,
      "step": 11080
    },
    {
      "epoch": 0.18146117974310727,
      "grad_norm": 0.4284801552123415,
      "learning_rate": 9.799253010057645e-06,
      "loss": 0.0076,
      "step": 11090
    },
    {
      "epoch": 0.1816248056941831,
      "grad_norm": 0.3192531954836076,
      "learning_rate": 9.798451131379213e-06,
      "loss": 0.0048,
      "step": 11100
    },
    {
      "epoch": 0.18178843164525893,
      "grad_norm": 0.4739618682474299,
      "learning_rate": 9.797647687290959e-06,
      "loss": 0.0079,
      "step": 11110
    },
    {
      "epoch": 0.1819520575963348,
      "grad_norm": 0.43335921480928347,
      "learning_rate": 9.796842678054994e-06,
      "loss": 0.0063,
      "step": 11120
    },
    {
      "epoch": 0.18211568354741062,
      "grad_norm": 0.7066375515263355,
      "learning_rate": 9.796036103933935e-06,
      "loss": 0.0049,
      "step": 11130
    },
    {
      "epoch": 0.18227930949848645,
      "grad_norm": 0.4281251377031283,
      "learning_rate": 9.795227965190914e-06,
      "loss": 0.0071,
      "step": 11140
    },
    {
      "epoch": 0.1824429354495623,
      "grad_norm": 0.589647200424727,
      "learning_rate": 9.794418262089571e-06,
      "loss": 0.0051,
      "step": 11150
    },
    {
      "epoch": 0.18260656140063813,
      "grad_norm": 0.21634038640987657,
      "learning_rate": 9.793606994894058e-06,
      "loss": 0.0065,
      "step": 11160
    },
    {
      "epoch": 0.182770187351714,
      "grad_norm": 0.2466957868707691,
      "learning_rate": 9.792794163869038e-06,
      "loss": 0.0063,
      "step": 11170
    },
    {
      "epoch": 0.18293381330278982,
      "grad_norm": 0.2585836539038335,
      "learning_rate": 9.79197976927968e-06,
      "loss": 0.0065,
      "step": 11180
    },
    {
      "epoch": 0.18309743925386565,
      "grad_norm": 0.1315996827887025,
      "learning_rate": 9.791163811391667e-06,
      "loss": 0.0091,
      "step": 11190
    },
    {
      "epoch": 0.1832610652049415,
      "grad_norm": 0.18507857651942677,
      "learning_rate": 9.790346290471192e-06,
      "loss": 0.0041,
      "step": 11200
    },
    {
      "epoch": 0.18342469115601734,
      "grad_norm": 0.36226611268884884,
      "learning_rate": 9.789527206784955e-06,
      "loss": 0.0075,
      "step": 11210
    },
    {
      "epoch": 0.1835883171070932,
      "grad_norm": 0.6361997358448347,
      "learning_rate": 9.788706560600167e-06,
      "loss": 0.0068,
      "step": 11220
    },
    {
      "epoch": 0.18375194305816903,
      "grad_norm": 0.1864986627660527,
      "learning_rate": 9.787884352184552e-06,
      "loss": 0.0049,
      "step": 11230
    },
    {
      "epoch": 0.18391556900924486,
      "grad_norm": 0.42457477005701955,
      "learning_rate": 9.78706058180634e-06,
      "loss": 0.0067,
      "step": 11240
    },
    {
      "epoch": 0.1840791949603207,
      "grad_norm": 0.2995253922746208,
      "learning_rate": 9.78623524973427e-06,
      "loss": 0.0071,
      "step": 11250
    },
    {
      "epoch": 0.18424282091139654,
      "grad_norm": 0.1470295387337036,
      "learning_rate": 9.785408356237595e-06,
      "loss": 0.007,
      "step": 11260
    },
    {
      "epoch": 0.1844064468624724,
      "grad_norm": 0.3706219745456237,
      "learning_rate": 9.78457990158607e-06,
      "loss": 0.0073,
      "step": 11270
    },
    {
      "epoch": 0.18457007281354823,
      "grad_norm": 0.18642248818552987,
      "learning_rate": 9.783749886049968e-06,
      "loss": 0.0059,
      "step": 11280
    },
    {
      "epoch": 0.18473369876462406,
      "grad_norm": 0.3367756237327783,
      "learning_rate": 9.782918309900065e-06,
      "loss": 0.0069,
      "step": 11290
    },
    {
      "epoch": 0.18489732471569992,
      "grad_norm": 0.30037444777426003,
      "learning_rate": 9.782085173407648e-06,
      "loss": 0.0073,
      "step": 11300
    },
    {
      "epoch": 0.18506095066677575,
      "grad_norm": 0.2611671832619326,
      "learning_rate": 9.781250476844512e-06,
      "loss": 0.0109,
      "step": 11310
    },
    {
      "epoch": 0.1852245766178516,
      "grad_norm": 0.2763791821038597,
      "learning_rate": 9.780414220482963e-06,
      "loss": 0.005,
      "step": 11320
    },
    {
      "epoch": 0.18538820256892743,
      "grad_norm": 0.4920780237134297,
      "learning_rate": 9.779576404595816e-06,
      "loss": 0.0068,
      "step": 11330
    },
    {
      "epoch": 0.18555182852000326,
      "grad_norm": 0.34772583793447126,
      "learning_rate": 9.77873702945639e-06,
      "loss": 0.0081,
      "step": 11340
    },
    {
      "epoch": 0.18571545447107912,
      "grad_norm": 0.2402157466953789,
      "learning_rate": 9.777896095338523e-06,
      "loss": 0.0054,
      "step": 11350
    },
    {
      "epoch": 0.18587908042215495,
      "grad_norm": 0.2175962648848062,
      "learning_rate": 9.777053602516548e-06,
      "loss": 0.0074,
      "step": 11360
    },
    {
      "epoch": 0.18604270637323078,
      "grad_norm": 0.4080172542591644,
      "learning_rate": 9.776209551265316e-06,
      "loss": 0.0092,
      "step": 11370
    },
    {
      "epoch": 0.18620633232430664,
      "grad_norm": 0.47385268601911257,
      "learning_rate": 9.775363941860185e-06,
      "loss": 0.0092,
      "step": 11380
    },
    {
      "epoch": 0.18636995827538247,
      "grad_norm": 0.31266517862802007,
      "learning_rate": 9.774516774577019e-06,
      "loss": 0.0101,
      "step": 11390
    },
    {
      "epoch": 0.18653358422645833,
      "grad_norm": 0.3761774585841798,
      "learning_rate": 9.77366804969219e-06,
      "loss": 0.0079,
      "step": 11400
    },
    {
      "epoch": 0.18669721017753416,
      "grad_norm": 0.8707324719372528,
      "learning_rate": 9.772817767482583e-06,
      "loss": 0.0071,
      "step": 11410
    },
    {
      "epoch": 0.18686083612861,
      "grad_norm": 0.725546705686707,
      "learning_rate": 9.771965928225583e-06,
      "loss": 0.0089,
      "step": 11420
    },
    {
      "epoch": 0.18702446207968584,
      "grad_norm": 0.5032077513770665,
      "learning_rate": 9.771112532199091e-06,
      "loss": 0.0069,
      "step": 11430
    },
    {
      "epoch": 0.18718808803076167,
      "grad_norm": 0.5050081611524134,
      "learning_rate": 9.770257579681513e-06,
      "loss": 0.0066,
      "step": 11440
    },
    {
      "epoch": 0.18735171398183753,
      "grad_norm": 0.524208787867849,
      "learning_rate": 9.769401070951758e-06,
      "loss": 0.0104,
      "step": 11450
    },
    {
      "epoch": 0.18751533993291336,
      "grad_norm": 0.41466010769731676,
      "learning_rate": 9.76854300628925e-06,
      "loss": 0.0063,
      "step": 11460
    },
    {
      "epoch": 0.1876789658839892,
      "grad_norm": 0.1771967172885356,
      "learning_rate": 9.767683385973917e-06,
      "loss": 0.007,
      "step": 11470
    },
    {
      "epoch": 0.18784259183506505,
      "grad_norm": 0.1683073529292256,
      "learning_rate": 9.766822210286193e-06,
      "loss": 0.0079,
      "step": 11480
    },
    {
      "epoch": 0.18800621778614088,
      "grad_norm": 0.4341036637448293,
      "learning_rate": 9.765959479507024e-06,
      "loss": 0.0069,
      "step": 11490
    },
    {
      "epoch": 0.18816984373721674,
      "grad_norm": 0.337723694912821,
      "learning_rate": 9.76509519391786e-06,
      "loss": 0.006,
      "step": 11500
    },
    {
      "epoch": 0.18833346968829257,
      "grad_norm": 0.24703291992630738,
      "learning_rate": 9.764229353800658e-06,
      "loss": 0.0059,
      "step": 11510
    },
    {
      "epoch": 0.1884970956393684,
      "grad_norm": 0.44953337718332714,
      "learning_rate": 9.763361959437882e-06,
      "loss": 0.0107,
      "step": 11520
    },
    {
      "epoch": 0.18866072159044425,
      "grad_norm": 0.34415607589276814,
      "learning_rate": 9.762493011112507e-06,
      "loss": 0.0084,
      "step": 11530
    },
    {
      "epoch": 0.18882434754152008,
      "grad_norm": 0.24751949638319196,
      "learning_rate": 9.761622509108011e-06,
      "loss": 0.007,
      "step": 11540
    },
    {
      "epoch": 0.1889879734925959,
      "grad_norm": 0.3063008796272494,
      "learning_rate": 9.760750453708378e-06,
      "loss": 0.0077,
      "step": 11550
    },
    {
      "epoch": 0.18915159944367177,
      "grad_norm": 0.20160061922919376,
      "learning_rate": 9.759876845198103e-06,
      "loss": 0.0073,
      "step": 11560
    },
    {
      "epoch": 0.1893152253947476,
      "grad_norm": 0.5691982020473816,
      "learning_rate": 9.759001683862186e-06,
      "loss": 0.009,
      "step": 11570
    },
    {
      "epoch": 0.18947885134582346,
      "grad_norm": 0.4357055435169999,
      "learning_rate": 9.758124969986129e-06,
      "loss": 0.0075,
      "step": 11580
    },
    {
      "epoch": 0.1896424772968993,
      "grad_norm": 0.08464089722462388,
      "learning_rate": 9.757246703855947e-06,
      "loss": 0.0048,
      "step": 11590
    },
    {
      "epoch": 0.18980610324797512,
      "grad_norm": 0.28062969462398113,
      "learning_rate": 9.75636688575816e-06,
      "loss": 0.0095,
      "step": 11600
    },
    {
      "epoch": 0.18996972919905097,
      "grad_norm": 0.3264916594553321,
      "learning_rate": 9.75548551597979e-06,
      "loss": 0.0052,
      "step": 11610
    },
    {
      "epoch": 0.1901333551501268,
      "grad_norm": 0.17973620222674624,
      "learning_rate": 9.75460259480837e-06,
      "loss": 0.009,
      "step": 11620
    },
    {
      "epoch": 0.19029698110120266,
      "grad_norm": 0.2029561933904139,
      "learning_rate": 9.753718122531937e-06,
      "loss": 0.0052,
      "step": 11630
    },
    {
      "epoch": 0.1904606070522785,
      "grad_norm": 0.3988648618580588,
      "learning_rate": 9.752832099439035e-06,
      "loss": 0.0078,
      "step": 11640
    },
    {
      "epoch": 0.19062423300335432,
      "grad_norm": 0.24359260999830218,
      "learning_rate": 9.751944525818715e-06,
      "loss": 0.0057,
      "step": 11650
    },
    {
      "epoch": 0.19078785895443018,
      "grad_norm": 0.37338015527357743,
      "learning_rate": 9.75105540196053e-06,
      "loss": 0.0057,
      "step": 11660
    },
    {
      "epoch": 0.190951484905506,
      "grad_norm": 0.21174412550323646,
      "learning_rate": 9.750164728154538e-06,
      "loss": 0.0064,
      "step": 11670
    },
    {
      "epoch": 0.19111511085658187,
      "grad_norm": 0.14801117799063032,
      "learning_rate": 9.749272504691312e-06,
      "loss": 0.0055,
      "step": 11680
    },
    {
      "epoch": 0.1912787368076577,
      "grad_norm": 0.31000782565843377,
      "learning_rate": 9.748378731861921e-06,
      "loss": 0.0084,
      "step": 11690
    },
    {
      "epoch": 0.19144236275873353,
      "grad_norm": 0.14206174103190947,
      "learning_rate": 9.74748340995794e-06,
      "loss": 0.0049,
      "step": 11700
    },
    {
      "epoch": 0.19160598870980938,
      "grad_norm": 0.29401582988388164,
      "learning_rate": 9.746586539271457e-06,
      "loss": 0.0078,
      "step": 11710
    },
    {
      "epoch": 0.1917696146608852,
      "grad_norm": 0.4859946997593239,
      "learning_rate": 9.745688120095056e-06,
      "loss": 0.008,
      "step": 11720
    },
    {
      "epoch": 0.19193324061196107,
      "grad_norm": 0.34352281638395943,
      "learning_rate": 9.744788152721834e-06,
      "loss": 0.0056,
      "step": 11730
    },
    {
      "epoch": 0.1920968665630369,
      "grad_norm": 0.2952173557394536,
      "learning_rate": 9.743886637445385e-06,
      "loss": 0.0043,
      "step": 11740
    },
    {
      "epoch": 0.19226049251411273,
      "grad_norm": 0.03682881227578708,
      "learning_rate": 9.742983574559817e-06,
      "loss": 0.0068,
      "step": 11750
    },
    {
      "epoch": 0.1924241184651886,
      "grad_norm": 0.16629063839522457,
      "learning_rate": 9.742078964359736e-06,
      "loss": 0.0053,
      "step": 11760
    },
    {
      "epoch": 0.19258774441626442,
      "grad_norm": 0.7447714376553686,
      "learning_rate": 9.741172807140254e-06,
      "loss": 0.0083,
      "step": 11770
    },
    {
      "epoch": 0.19275137036734025,
      "grad_norm": 0.31920258692730474,
      "learning_rate": 9.74026510319699e-06,
      "loss": 0.0053,
      "step": 11780
    },
    {
      "epoch": 0.1929149963184161,
      "grad_norm": 0.5438699773052527,
      "learning_rate": 9.739355852826069e-06,
      "loss": 0.0051,
      "step": 11790
    },
    {
      "epoch": 0.19307862226949193,
      "grad_norm": 0.22706135112454567,
      "learning_rate": 9.738445056324113e-06,
      "loss": 0.0058,
      "step": 11800
    },
    {
      "epoch": 0.1932422482205678,
      "grad_norm": 0.32765932232844974,
      "learning_rate": 9.737532713988256e-06,
      "loss": 0.0046,
      "step": 11810
    },
    {
      "epoch": 0.19340587417164362,
      "grad_norm": 0.3894588915813887,
      "learning_rate": 9.736618826116133e-06,
      "loss": 0.0087,
      "step": 11820
    },
    {
      "epoch": 0.19356950012271945,
      "grad_norm": 0.20356218035756277,
      "learning_rate": 9.735703393005885e-06,
      "loss": 0.0061,
      "step": 11830
    },
    {
      "epoch": 0.1937331260737953,
      "grad_norm": 0.34566481300505325,
      "learning_rate": 9.734786414956152e-06,
      "loss": 0.0053,
      "step": 11840
    },
    {
      "epoch": 0.19389675202487114,
      "grad_norm": 0.5089015288453581,
      "learning_rate": 9.733867892266087e-06,
      "loss": 0.0066,
      "step": 11850
    },
    {
      "epoch": 0.194060377975947,
      "grad_norm": 0.31411316990389243,
      "learning_rate": 9.732947825235339e-06,
      "loss": 0.0077,
      "step": 11860
    },
    {
      "epoch": 0.19422400392702283,
      "grad_norm": 0.6256280330075266,
      "learning_rate": 9.732026214164063e-06,
      "loss": 0.01,
      "step": 11870
    },
    {
      "epoch": 0.19438762987809866,
      "grad_norm": 0.4237697237391104,
      "learning_rate": 9.73110305935292e-06,
      "loss": 0.008,
      "step": 11880
    },
    {
      "epoch": 0.1945512558291745,
      "grad_norm": 0.28542089516947944,
      "learning_rate": 9.730178361103071e-06,
      "loss": 0.0063,
      "step": 11890
    },
    {
      "epoch": 0.19471488178025034,
      "grad_norm": 0.6879265857677136,
      "learning_rate": 9.729252119716183e-06,
      "loss": 0.0093,
      "step": 11900
    },
    {
      "epoch": 0.1948785077313262,
      "grad_norm": 0.36321127849872575,
      "learning_rate": 9.72832433549443e-06,
      "loss": 0.0062,
      "step": 11910
    },
    {
      "epoch": 0.19504213368240203,
      "grad_norm": 0.16790806420191687,
      "learning_rate": 9.72739500874048e-06,
      "loss": 0.0062,
      "step": 11920
    },
    {
      "epoch": 0.19520575963347786,
      "grad_norm": 0.12708815413892316,
      "learning_rate": 9.72646413975751e-06,
      "loss": 0.0057,
      "step": 11930
    },
    {
      "epoch": 0.19536938558455372,
      "grad_norm": 0.2913581949467308,
      "learning_rate": 9.725531728849202e-06,
      "loss": 0.0062,
      "step": 11940
    },
    {
      "epoch": 0.19553301153562955,
      "grad_norm": 0.12793090218002592,
      "learning_rate": 9.724597776319737e-06,
      "loss": 0.0075,
      "step": 11950
    },
    {
      "epoch": 0.1956966374867054,
      "grad_norm": 0.29278360496416067,
      "learning_rate": 9.7236622824738e-06,
      "loss": 0.0056,
      "step": 11960
    },
    {
      "epoch": 0.19586026343778123,
      "grad_norm": 0.43463834151273584,
      "learning_rate": 9.72272524761658e-06,
      "loss": 0.0076,
      "step": 11970
    },
    {
      "epoch": 0.19602388938885706,
      "grad_norm": 0.40536106608832556,
      "learning_rate": 9.72178667205377e-06,
      "loss": 0.0075,
      "step": 11980
    },
    {
      "epoch": 0.19618751533993292,
      "grad_norm": 0.28355589944274456,
      "learning_rate": 9.72084655609156e-06,
      "loss": 0.0118,
      "step": 11990
    },
    {
      "epoch": 0.19635114129100875,
      "grad_norm": 0.2101001075035258,
      "learning_rate": 9.719904900036648e-06,
      "loss": 0.0067,
      "step": 12000
    },
    {
      "epoch": 0.19651476724208458,
      "grad_norm": 0.11458634054087051,
      "learning_rate": 9.718961704196233e-06,
      "loss": 0.0085,
      "step": 12010
    },
    {
      "epoch": 0.19667839319316044,
      "grad_norm": 0.33031898045802605,
      "learning_rate": 9.718016968878016e-06,
      "loss": 0.008,
      "step": 12020
    },
    {
      "epoch": 0.19684201914423627,
      "grad_norm": 0.19265538490576667,
      "learning_rate": 9.717070694390199e-06,
      "loss": 0.0075,
      "step": 12030
    },
    {
      "epoch": 0.19700564509531213,
      "grad_norm": 0.21289840457184517,
      "learning_rate": 9.716122881041488e-06,
      "loss": 0.0078,
      "step": 12040
    },
    {
      "epoch": 0.19716927104638796,
      "grad_norm": 0.4889128426999637,
      "learning_rate": 9.715173529141092e-06,
      "loss": 0.0085,
      "step": 12050
    },
    {
      "epoch": 0.19733289699746379,
      "grad_norm": 0.2621124078049699,
      "learning_rate": 9.714222638998717e-06,
      "loss": 0.0113,
      "step": 12060
    },
    {
      "epoch": 0.19749652294853964,
      "grad_norm": 0.3385495327414111,
      "learning_rate": 9.713270210924577e-06,
      "loss": 0.0076,
      "step": 12070
    },
    {
      "epoch": 0.19766014889961547,
      "grad_norm": 0.16629378462763933,
      "learning_rate": 9.712316245229383e-06,
      "loss": 0.0067,
      "step": 12080
    },
    {
      "epoch": 0.19782377485069133,
      "grad_norm": 0.4981478954399011,
      "learning_rate": 9.71136074222435e-06,
      "loss": 0.0076,
      "step": 12090
    },
    {
      "epoch": 0.19798740080176716,
      "grad_norm": 0.4360526731889721,
      "learning_rate": 9.710403702221194e-06,
      "loss": 0.0093,
      "step": 12100
    },
    {
      "epoch": 0.198151026752843,
      "grad_norm": 0.31864374836859155,
      "learning_rate": 9.709445125532133e-06,
      "loss": 0.0059,
      "step": 12110
    },
    {
      "epoch": 0.19831465270391885,
      "grad_norm": 0.3657849550095795,
      "learning_rate": 9.708485012469884e-06,
      "loss": 0.0062,
      "step": 12120
    },
    {
      "epoch": 0.19847827865499468,
      "grad_norm": 0.13541145412841896,
      "learning_rate": 9.707523363347669e-06,
      "loss": 0.0046,
      "step": 12130
    },
    {
      "epoch": 0.19864190460607054,
      "grad_norm": 0.32932502068846803,
      "learning_rate": 9.706560178479207e-06,
      "loss": 0.0081,
      "step": 12140
    },
    {
      "epoch": 0.19880553055714636,
      "grad_norm": 0.6144776783137986,
      "learning_rate": 9.705595458178722e-06,
      "loss": 0.0087,
      "step": 12150
    },
    {
      "epoch": 0.1989691565082222,
      "grad_norm": 0.2843430518113607,
      "learning_rate": 9.704629202760935e-06,
      "loss": 0.0059,
      "step": 12160
    },
    {
      "epoch": 0.19913278245929805,
      "grad_norm": 0.2364227673305673,
      "learning_rate": 9.703661412541073e-06,
      "loss": 0.0117,
      "step": 12170
    },
    {
      "epoch": 0.19929640841037388,
      "grad_norm": 5.346360808828941,
      "learning_rate": 9.702692087834856e-06,
      "loss": 0.0084,
      "step": 12180
    },
    {
      "epoch": 0.19946003436144974,
      "grad_norm": 0.175198840295415,
      "learning_rate": 9.701721228958511e-06,
      "loss": 0.0061,
      "step": 12190
    },
    {
      "epoch": 0.19962366031252557,
      "grad_norm": 0.5753284470503537,
      "learning_rate": 9.700748836228766e-06,
      "loss": 0.0074,
      "step": 12200
    },
    {
      "epoch": 0.1997872862636014,
      "grad_norm": 0.14650786177447164,
      "learning_rate": 9.699774909962842e-06,
      "loss": 0.0045,
      "step": 12210
    },
    {
      "epoch": 0.19995091221467726,
      "grad_norm": 1.0074347249256546,
      "learning_rate": 9.698799450478469e-06,
      "loss": 0.0074,
      "step": 12220
    },
    {
      "epoch": 0.2001145381657531,
      "grad_norm": 0.3469719396945667,
      "learning_rate": 9.697822458093872e-06,
      "loss": 0.0078,
      "step": 12230
    },
    {
      "epoch": 0.20027816411682892,
      "grad_norm": 0.5546327667459244,
      "learning_rate": 9.69684393312778e-06,
      "loss": 0.0084,
      "step": 12240
    },
    {
      "epoch": 0.20044179006790477,
      "grad_norm": 0.42755857876549763,
      "learning_rate": 9.695863875899415e-06,
      "loss": 0.0081,
      "step": 12250
    },
    {
      "epoch": 0.2006054160189806,
      "grad_norm": 0.5349918217490497,
      "learning_rate": 9.694882286728506e-06,
      "loss": 0.0047,
      "step": 12260
    },
    {
      "epoch": 0.20076904197005646,
      "grad_norm": 0.3678883635136783,
      "learning_rate": 9.693899165935278e-06,
      "loss": 0.0049,
      "step": 12270
    },
    {
      "epoch": 0.2009326679211323,
      "grad_norm": 0.3788620823741124,
      "learning_rate": 9.692914513840459e-06,
      "loss": 0.0063,
      "step": 12280
    },
    {
      "epoch": 0.20109629387220812,
      "grad_norm": 0.595153655535423,
      "learning_rate": 9.691928330765273e-06,
      "loss": 0.0093,
      "step": 12290
    },
    {
      "epoch": 0.20125991982328398,
      "grad_norm": 0.309871474654151,
      "learning_rate": 9.690940617031442e-06,
      "loss": 0.0057,
      "step": 12300
    },
    {
      "epoch": 0.2014235457743598,
      "grad_norm": 0.4458159092416787,
      "learning_rate": 9.689951372961195e-06,
      "loss": 0.0074,
      "step": 12310
    },
    {
      "epoch": 0.20158717172543567,
      "grad_norm": 0.26059267801767083,
      "learning_rate": 9.688960598877253e-06,
      "loss": 0.0064,
      "step": 12320
    },
    {
      "epoch": 0.2017507976765115,
      "grad_norm": 0.5035292099910589,
      "learning_rate": 9.687968295102837e-06,
      "loss": 0.0078,
      "step": 12330
    },
    {
      "epoch": 0.20191442362758732,
      "grad_norm": 0.31016519161160244,
      "learning_rate": 9.68697446196167e-06,
      "loss": 0.0057,
      "step": 12340
    },
    {
      "epoch": 0.20207804957866318,
      "grad_norm": 0.4823829055947124,
      "learning_rate": 9.685979099777975e-06,
      "loss": 0.0074,
      "step": 12350
    },
    {
      "epoch": 0.202241675529739,
      "grad_norm": 0.3607532471539726,
      "learning_rate": 9.684982208876468e-06,
      "loss": 0.0054,
      "step": 12360
    },
    {
      "epoch": 0.20240530148081487,
      "grad_norm": 0.307286937344267,
      "learning_rate": 9.683983789582368e-06,
      "loss": 0.006,
      "step": 12370
    },
    {
      "epoch": 0.2025689274318907,
      "grad_norm": 0.3021384956652315,
      "learning_rate": 9.682983842221392e-06,
      "loss": 0.0059,
      "step": 12380
    },
    {
      "epoch": 0.20273255338296653,
      "grad_norm": 0.14032912575562123,
      "learning_rate": 9.681982367119755e-06,
      "loss": 0.0128,
      "step": 12390
    },
    {
      "epoch": 0.2028961793340424,
      "grad_norm": 0.25494701953350013,
      "learning_rate": 9.68097936460417e-06,
      "loss": 0.0036,
      "step": 12400
    },
    {
      "epoch": 0.20305980528511822,
      "grad_norm": 0.7215056757533814,
      "learning_rate": 9.679974835001848e-06,
      "loss": 0.0072,
      "step": 12410
    },
    {
      "epoch": 0.20322343123619407,
      "grad_norm": 0.6874998132940427,
      "learning_rate": 9.678968778640501e-06,
      "loss": 0.0066,
      "step": 12420
    },
    {
      "epoch": 0.2033870571872699,
      "grad_norm": 0.6338288676066054,
      "learning_rate": 9.677961195848339e-06,
      "loss": 0.0073,
      "step": 12430
    },
    {
      "epoch": 0.20355068313834573,
      "grad_norm": 0.47786177716347256,
      "learning_rate": 9.67695208695406e-06,
      "loss": 0.0074,
      "step": 12440
    },
    {
      "epoch": 0.2037143090894216,
      "grad_norm": 0.4006828277497043,
      "learning_rate": 9.675941452286876e-06,
      "loss": 0.0057,
      "step": 12450
    },
    {
      "epoch": 0.20387793504049742,
      "grad_norm": 0.5486545919501874,
      "learning_rate": 9.674929292176487e-06,
      "loss": 0.0076,
      "step": 12460
    },
    {
      "epoch": 0.20404156099157325,
      "grad_norm": 0.43607037531822396,
      "learning_rate": 9.67391560695309e-06,
      "loss": 0.0073,
      "step": 12470
    },
    {
      "epoch": 0.2042051869426491,
      "grad_norm": 0.39146743131215267,
      "learning_rate": 9.672900396947385e-06,
      "loss": 0.0066,
      "step": 12480
    },
    {
      "epoch": 0.20436881289372494,
      "grad_norm": 0.35802275456821875,
      "learning_rate": 9.671883662490562e-06,
      "loss": 0.0062,
      "step": 12490
    },
    {
      "epoch": 0.2045324388448008,
      "grad_norm": 0.12251678948981015,
      "learning_rate": 9.670865403914317e-06,
      "loss": 0.0093,
      "step": 12500
    },
    {
      "epoch": 0.20469606479587663,
      "grad_norm": 0.3319458171569489,
      "learning_rate": 9.669845621550835e-06,
      "loss": 0.0078,
      "step": 12510
    },
    {
      "epoch": 0.20485969074695246,
      "grad_norm": 0.42948027025675256,
      "learning_rate": 9.668824315732805e-06,
      "loss": 0.0064,
      "step": 12520
    },
    {
      "epoch": 0.2050233166980283,
      "grad_norm": 0.47347068952592614,
      "learning_rate": 9.667801486793411e-06,
      "loss": 0.0087,
      "step": 12530
    },
    {
      "epoch": 0.20518694264910414,
      "grad_norm": 0.2905391598286487,
      "learning_rate": 9.666777135066328e-06,
      "loss": 0.0074,
      "step": 12540
    },
    {
      "epoch": 0.20535056860018,
      "grad_norm": 0.2237668634623418,
      "learning_rate": 9.665751260885737e-06,
      "loss": 0.0043,
      "step": 12550
    },
    {
      "epoch": 0.20551419455125583,
      "grad_norm": 0.48019144848456347,
      "learning_rate": 9.664723864586308e-06,
      "loss": 0.0066,
      "step": 12560
    },
    {
      "epoch": 0.20567782050233166,
      "grad_norm": 0.5002425281696792,
      "learning_rate": 9.663694946503213e-06,
      "loss": 0.0052,
      "step": 12570
    },
    {
      "epoch": 0.20584144645340752,
      "grad_norm": 0.3639612781704994,
      "learning_rate": 9.662664506972119e-06,
      "loss": 0.0091,
      "step": 12580
    },
    {
      "epoch": 0.20600507240448335,
      "grad_norm": 0.32170670076523167,
      "learning_rate": 9.661632546329187e-06,
      "loss": 0.0067,
      "step": 12590
    },
    {
      "epoch": 0.2061686983555592,
      "grad_norm": 0.431961869402456,
      "learning_rate": 9.660599064911077e-06,
      "loss": 0.0072,
      "step": 12600
    },
    {
      "epoch": 0.20633232430663503,
      "grad_norm": 0.36899834695392764,
      "learning_rate": 9.659564063054943e-06,
      "loss": 0.0061,
      "step": 12610
    },
    {
      "epoch": 0.20649595025771086,
      "grad_norm": 0.26563665919546947,
      "learning_rate": 9.658527541098438e-06,
      "loss": 0.0059,
      "step": 12620
    },
    {
      "epoch": 0.20665957620878672,
      "grad_norm": 0.11193708439377764,
      "learning_rate": 9.657489499379704e-06,
      "loss": 0.0077,
      "step": 12630
    },
    {
      "epoch": 0.20682320215986255,
      "grad_norm": 0.4780575751925386,
      "learning_rate": 9.656449938237391e-06,
      "loss": 0.0076,
      "step": 12640
    },
    {
      "epoch": 0.20698682811093838,
      "grad_norm": 0.23904755872916147,
      "learning_rate": 9.65540885801063e-06,
      "loss": 0.0088,
      "step": 12650
    },
    {
      "epoch": 0.20715045406201424,
      "grad_norm": 0.22872827533944465,
      "learning_rate": 9.654366259039063e-06,
      "loss": 0.0057,
      "step": 12660
    },
    {
      "epoch": 0.20731408001309007,
      "grad_norm": 0.25847687068755604,
      "learning_rate": 9.653322141662813e-06,
      "loss": 0.0092,
      "step": 12670
    },
    {
      "epoch": 0.20747770596416593,
      "grad_norm": 0.532797968329369,
      "learning_rate": 9.652276506222507e-06,
      "loss": 0.0102,
      "step": 12680
    },
    {
      "epoch": 0.20764133191524176,
      "grad_norm": 0.11560189216615746,
      "learning_rate": 9.651229353059266e-06,
      "loss": 0.0053,
      "step": 12690
    },
    {
      "epoch": 0.20780495786631759,
      "grad_norm": 0.3883805671309935,
      "learning_rate": 9.650180682514703e-06,
      "loss": 0.0071,
      "step": 12700
    },
    {
      "epoch": 0.20796858381739344,
      "grad_norm": 0.3029979787429645,
      "learning_rate": 9.64913049493093e-06,
      "loss": 0.0073,
      "step": 12710
    },
    {
      "epoch": 0.20813220976846927,
      "grad_norm": 0.1935482928563381,
      "learning_rate": 9.648078790650553e-06,
      "loss": 0.0041,
      "step": 12720
    },
    {
      "epoch": 0.20829583571954513,
      "grad_norm": 0.09819876469556359,
      "learning_rate": 9.647025570016667e-06,
      "loss": 0.0061,
      "step": 12730
    },
    {
      "epoch": 0.20845946167062096,
      "grad_norm": 0.4154960487791988,
      "learning_rate": 9.645970833372874e-06,
      "loss": 0.0099,
      "step": 12740
    },
    {
      "epoch": 0.2086230876216968,
      "grad_norm": 0.18485296738170212,
      "learning_rate": 9.644914581063258e-06,
      "loss": 0.0067,
      "step": 12750
    },
    {
      "epoch": 0.20878671357277265,
      "grad_norm": 0.5495450320419935,
      "learning_rate": 9.643856813432404e-06,
      "loss": 0.0083,
      "step": 12760
    },
    {
      "epoch": 0.20895033952384848,
      "grad_norm": 0.41755646701200866,
      "learning_rate": 9.64279753082539e-06,
      "loss": 0.0081,
      "step": 12770
    },
    {
      "epoch": 0.20911396547492433,
      "grad_norm": 0.10329182114654682,
      "learning_rate": 9.641736733587788e-06,
      "loss": 0.0051,
      "step": 12780
    },
    {
      "epoch": 0.20927759142600016,
      "grad_norm": 0.3841368710325325,
      "learning_rate": 9.640674422065665e-06,
      "loss": 0.0086,
      "step": 12790
    },
    {
      "epoch": 0.209441217377076,
      "grad_norm": 0.173328307705944,
      "learning_rate": 9.639610596605582e-06,
      "loss": 0.0097,
      "step": 12800
    },
    {
      "epoch": 0.20960484332815185,
      "grad_norm": 0.20769587644032075,
      "learning_rate": 9.63854525755459e-06,
      "loss": 0.0056,
      "step": 12810
    },
    {
      "epoch": 0.20976846927922768,
      "grad_norm": 0.46200802550431713,
      "learning_rate": 9.63747840526024e-06,
      "loss": 0.0064,
      "step": 12820
    },
    {
      "epoch": 0.20993209523030354,
      "grad_norm": 0.19081215819993722,
      "learning_rate": 9.636410040070573e-06,
      "loss": 0.0069,
      "step": 12830
    },
    {
      "epoch": 0.21009572118137937,
      "grad_norm": 0.049189931026139326,
      "learning_rate": 9.635340162334124e-06,
      "loss": 0.0067,
      "step": 12840
    },
    {
      "epoch": 0.2102593471324552,
      "grad_norm": 0.37550403481683076,
      "learning_rate": 9.634268772399922e-06,
      "loss": 0.0069,
      "step": 12850
    },
    {
      "epoch": 0.21042297308353106,
      "grad_norm": 0.1445532454869915,
      "learning_rate": 9.63319587061749e-06,
      "loss": 0.0058,
      "step": 12860
    },
    {
      "epoch": 0.21058659903460689,
      "grad_norm": 0.282937422846647,
      "learning_rate": 9.632121457336839e-06,
      "loss": 0.0037,
      "step": 12870
    },
    {
      "epoch": 0.21075022498568272,
      "grad_norm": 0.25010971169222856,
      "learning_rate": 9.631045532908483e-06,
      "loss": 0.0048,
      "step": 12880
    },
    {
      "epoch": 0.21091385093675857,
      "grad_norm": 0.2563170288396381,
      "learning_rate": 9.629968097683419e-06,
      "loss": 0.0049,
      "step": 12890
    },
    {
      "epoch": 0.2110774768878344,
      "grad_norm": 0.44567158893021547,
      "learning_rate": 9.628889152013144e-06,
      "loss": 0.0075,
      "step": 12900
    },
    {
      "epoch": 0.21124110283891026,
      "grad_norm": 0.5051481386519453,
      "learning_rate": 9.627808696249642e-06,
      "loss": 0.0075,
      "step": 12910
    },
    {
      "epoch": 0.2114047287899861,
      "grad_norm": 0.3005312142160377,
      "learning_rate": 9.626726730745396e-06,
      "loss": 0.0065,
      "step": 12920
    },
    {
      "epoch": 0.21156835474106192,
      "grad_norm": 0.32682279958165067,
      "learning_rate": 9.625643255853377e-06,
      "loss": 0.0045,
      "step": 12930
    },
    {
      "epoch": 0.21173198069213778,
      "grad_norm": 0.31638464574382064,
      "learning_rate": 9.624558271927049e-06,
      "loss": 0.0061,
      "step": 12940
    },
    {
      "epoch": 0.2118956066432136,
      "grad_norm": 0.5055635511008135,
      "learning_rate": 9.623471779320369e-06,
      "loss": 0.0045,
      "step": 12950
    },
    {
      "epoch": 0.21205923259428947,
      "grad_norm": 0.2435266490874863,
      "learning_rate": 9.622383778387784e-06,
      "loss": 0.0065,
      "step": 12960
    },
    {
      "epoch": 0.2122228585453653,
      "grad_norm": 0.3608265970919034,
      "learning_rate": 9.621294269484238e-06,
      "loss": 0.0063,
      "step": 12970
    },
    {
      "epoch": 0.21238648449644112,
      "grad_norm": 0.3488213751205232,
      "learning_rate": 9.620203252965163e-06,
      "loss": 0.0073,
      "step": 12980
    },
    {
      "epoch": 0.21255011044751698,
      "grad_norm": 0.2049900708808698,
      "learning_rate": 9.619110729186485e-06,
      "loss": 0.0087,
      "step": 12990
    },
    {
      "epoch": 0.2127137363985928,
      "grad_norm": 0.39186369361272927,
      "learning_rate": 9.618016698504618e-06,
      "loss": 0.0051,
      "step": 13000
    },
    {
      "epoch": 0.21287736234966867,
      "grad_norm": 0.24222160481711164,
      "learning_rate": 9.616921161276471e-06,
      "loss": 0.0034,
      "step": 13010
    },
    {
      "epoch": 0.2130409883007445,
      "grad_norm": 0.45191656399038793,
      "learning_rate": 9.615824117859445e-06,
      "loss": 0.0045,
      "step": 13020
    },
    {
      "epoch": 0.21320461425182033,
      "grad_norm": 0.2312858761680152,
      "learning_rate": 9.61472556861143e-06,
      "loss": 0.0071,
      "step": 13030
    },
    {
      "epoch": 0.2133682402028962,
      "grad_norm": 0.36388579459186143,
      "learning_rate": 9.61362551389081e-06,
      "loss": 0.007,
      "step": 13040
    },
    {
      "epoch": 0.21353186615397202,
      "grad_norm": 0.3086548382097394,
      "learning_rate": 9.612523954056454e-06,
      "loss": 0.0069,
      "step": 13050
    },
    {
      "epoch": 0.21369549210504787,
      "grad_norm": 0.14314722661602566,
      "learning_rate": 9.611420889467731e-06,
      "loss": 0.0046,
      "step": 13060
    },
    {
      "epoch": 0.2138591180561237,
      "grad_norm": 0.4373123616002356,
      "learning_rate": 9.610316320484494e-06,
      "loss": 0.006,
      "step": 13070
    },
    {
      "epoch": 0.21402274400719953,
      "grad_norm": 0.4871047676305095,
      "learning_rate": 9.60921024746709e-06,
      "loss": 0.0067,
      "step": 13080
    },
    {
      "epoch": 0.2141863699582754,
      "grad_norm": 0.3266083014988156,
      "learning_rate": 9.608102670776354e-06,
      "loss": 0.0051,
      "step": 13090
    },
    {
      "epoch": 0.21434999590935122,
      "grad_norm": 0.2731045691905199,
      "learning_rate": 9.606993590773616e-06,
      "loss": 0.0074,
      "step": 13100
    },
    {
      "epoch": 0.21451362186042705,
      "grad_norm": 0.2764307463807169,
      "learning_rate": 9.605883007820693e-06,
      "loss": 0.0057,
      "step": 13110
    },
    {
      "epoch": 0.2146772478115029,
      "grad_norm": 0.2368996995251964,
      "learning_rate": 9.604770922279893e-06,
      "loss": 0.0047,
      "step": 13120
    },
    {
      "epoch": 0.21484087376257874,
      "grad_norm": 0.24739123282299902,
      "learning_rate": 9.603657334514012e-06,
      "loss": 0.0066,
      "step": 13130
    },
    {
      "epoch": 0.2150044997136546,
      "grad_norm": 0.18433050559270203,
      "learning_rate": 9.60254224488634e-06,
      "loss": 0.0068,
      "step": 13140
    },
    {
      "epoch": 0.21516812566473043,
      "grad_norm": 0.20447111534033818,
      "learning_rate": 9.601425653760657e-06,
      "loss": 0.008,
      "step": 13150
    },
    {
      "epoch": 0.21533175161580626,
      "grad_norm": 0.17175156416557466,
      "learning_rate": 9.600307561501228e-06,
      "loss": 0.0071,
      "step": 13160
    },
    {
      "epoch": 0.2154953775668821,
      "grad_norm": 0.505279682563232,
      "learning_rate": 9.599187968472814e-06,
      "loss": 0.0076,
      "step": 13170
    },
    {
      "epoch": 0.21565900351795794,
      "grad_norm": 0.29735972258051,
      "learning_rate": 9.59806687504066e-06,
      "loss": 0.0083,
      "step": 13180
    },
    {
      "epoch": 0.2158226294690338,
      "grad_norm": 0.24681865111432016,
      "learning_rate": 9.5969442815705e-06,
      "loss": 0.0089,
      "step": 13190
    },
    {
      "epoch": 0.21598625542010963,
      "grad_norm": 0.32140709317686506,
      "learning_rate": 9.59582018842857e-06,
      "loss": 0.0062,
      "step": 13200
    },
    {
      "epoch": 0.21614988137118546,
      "grad_norm": 0.532337821477823,
      "learning_rate": 9.594694595981575e-06,
      "loss": 0.0079,
      "step": 13210
    },
    {
      "epoch": 0.21631350732226132,
      "grad_norm": 0.3852848127046301,
      "learning_rate": 9.593567504596726e-06,
      "loss": 0.006,
      "step": 13220
    },
    {
      "epoch": 0.21647713327333715,
      "grad_norm": 0.36133466685803317,
      "learning_rate": 9.592438914641715e-06,
      "loss": 0.0066,
      "step": 13230
    },
    {
      "epoch": 0.216640759224413,
      "grad_norm": 0.33291897366357015,
      "learning_rate": 9.591308826484722e-06,
      "loss": 0.0058,
      "step": 13240
    },
    {
      "epoch": 0.21680438517548883,
      "grad_norm": 0.32600592817160806,
      "learning_rate": 9.590177240494423e-06,
      "loss": 0.007,
      "step": 13250
    },
    {
      "epoch": 0.21696801112656466,
      "grad_norm": 0.2171981648055445,
      "learning_rate": 9.589044157039972e-06,
      "loss": 0.0074,
      "step": 13260
    },
    {
      "epoch": 0.21713163707764052,
      "grad_norm": 0.17825027067375412,
      "learning_rate": 9.587909576491025e-06,
      "loss": 0.0064,
      "step": 13270
    },
    {
      "epoch": 0.21729526302871635,
      "grad_norm": 0.37501316413056474,
      "learning_rate": 9.586773499217711e-06,
      "loss": 0.0053,
      "step": 13280
    },
    {
      "epoch": 0.2174588889797922,
      "grad_norm": 0.26287798643027416,
      "learning_rate": 9.585635925590661e-06,
      "loss": 0.0062,
      "step": 13290
    },
    {
      "epoch": 0.21762251493086804,
      "grad_norm": 0.18786313071231878,
      "learning_rate": 9.584496855980985e-06,
      "loss": 0.0052,
      "step": 13300
    },
    {
      "epoch": 0.21778614088194387,
      "grad_norm": 0.12355998891252588,
      "learning_rate": 9.583356290760287e-06,
      "loss": 0.0069,
      "step": 13310
    },
    {
      "epoch": 0.21794976683301973,
      "grad_norm": 0.2815342708333563,
      "learning_rate": 9.582214230300653e-06,
      "loss": 0.0037,
      "step": 13320
    },
    {
      "epoch": 0.21811339278409556,
      "grad_norm": 0.5776772098926313,
      "learning_rate": 9.581070674974662e-06,
      "loss": 0.0076,
      "step": 13330
    },
    {
      "epoch": 0.21827701873517139,
      "grad_norm": 0.4677421218474824,
      "learning_rate": 9.579925625155377e-06,
      "loss": 0.0056,
      "step": 13340
    },
    {
      "epoch": 0.21844064468624724,
      "grad_norm": 0.3025267025505492,
      "learning_rate": 9.578779081216355e-06,
      "loss": 0.0081,
      "step": 13350
    },
    {
      "epoch": 0.21860427063732307,
      "grad_norm": 0.37344902339633046,
      "learning_rate": 9.577631043531628e-06,
      "loss": 0.0114,
      "step": 13360
    },
    {
      "epoch": 0.21876789658839893,
      "grad_norm": 0.6941529396153875,
      "learning_rate": 9.57648151247573e-06,
      "loss": 0.0088,
      "step": 13370
    },
    {
      "epoch": 0.21893152253947476,
      "grad_norm": 0.24355546556371185,
      "learning_rate": 9.575330488423672e-06,
      "loss": 0.005,
      "step": 13380
    },
    {
      "epoch": 0.2190951484905506,
      "grad_norm": 0.2342937368232905,
      "learning_rate": 9.574177971750954e-06,
      "loss": 0.0064,
      "step": 13390
    },
    {
      "epoch": 0.21925877444162645,
      "grad_norm": 0.2740934207646199,
      "learning_rate": 9.573023962833567e-06,
      "loss": 0.0067,
      "step": 13400
    },
    {
      "epoch": 0.21942240039270228,
      "grad_norm": 0.5052348012143701,
      "learning_rate": 9.571868462047985e-06,
      "loss": 0.0083,
      "step": 13410
    },
    {
      "epoch": 0.21958602634377813,
      "grad_norm": 0.1356438839628698,
      "learning_rate": 9.57071146977117e-06,
      "loss": 0.0035,
      "step": 13420
    },
    {
      "epoch": 0.21974965229485396,
      "grad_norm": 0.8484093836547718,
      "learning_rate": 9.569552986380568e-06,
      "loss": 0.0101,
      "step": 13430
    },
    {
      "epoch": 0.2199132782459298,
      "grad_norm": 0.21377409613672566,
      "learning_rate": 9.568393012254118e-06,
      "loss": 0.0083,
      "step": 13440
    },
    {
      "epoch": 0.22007690419700565,
      "grad_norm": 0.231030705717587,
      "learning_rate": 9.567231547770237e-06,
      "loss": 0.0066,
      "step": 13450
    },
    {
      "epoch": 0.22024053014808148,
      "grad_norm": 0.2017966744892309,
      "learning_rate": 9.566068593307835e-06,
      "loss": 0.0042,
      "step": 13460
    },
    {
      "epoch": 0.22040415609915734,
      "grad_norm": 0.29284216348492254,
      "learning_rate": 9.564904149246302e-06,
      "loss": 0.007,
      "step": 13470
    },
    {
      "epoch": 0.22056778205023317,
      "grad_norm": 0.30560245920193674,
      "learning_rate": 9.56373821596552e-06,
      "loss": 0.0049,
      "step": 13480
    },
    {
      "epoch": 0.220731408001309,
      "grad_norm": 0.5001219378865481,
      "learning_rate": 9.562570793845856e-06,
      "loss": 0.0067,
      "step": 13490
    },
    {
      "epoch": 0.22089503395238486,
      "grad_norm": 0.3086288716310587,
      "learning_rate": 9.561401883268157e-06,
      "loss": 0.0077,
      "step": 13500
    },
    {
      "epoch": 0.22105865990346069,
      "grad_norm": 0.21121580296196205,
      "learning_rate": 9.56023148461376e-06,
      "loss": 0.0035,
      "step": 13510
    },
    {
      "epoch": 0.22122228585453654,
      "grad_norm": 0.16208485563364608,
      "learning_rate": 9.55905959826449e-06,
      "loss": 0.0077,
      "step": 13520
    },
    {
      "epoch": 0.22138591180561237,
      "grad_norm": 0.6129561213274362,
      "learning_rate": 9.557886224602648e-06,
      "loss": 0.0083,
      "step": 13530
    },
    {
      "epoch": 0.2215495377566882,
      "grad_norm": 0.32513100068890144,
      "learning_rate": 9.556711364011031e-06,
      "loss": 0.0059,
      "step": 13540
    },
    {
      "epoch": 0.22171316370776406,
      "grad_norm": 0.17938448655392583,
      "learning_rate": 9.555535016872918e-06,
      "loss": 0.0041,
      "step": 13550
    },
    {
      "epoch": 0.2218767896588399,
      "grad_norm": 0.17900725186625582,
      "learning_rate": 9.554357183572067e-06,
      "loss": 0.0058,
      "step": 13560
    },
    {
      "epoch": 0.22204041560991572,
      "grad_norm": 0.42399516304112894,
      "learning_rate": 9.553177864492728e-06,
      "loss": 0.0105,
      "step": 13570
    },
    {
      "epoch": 0.22220404156099158,
      "grad_norm": 0.3813361713911305,
      "learning_rate": 9.551997060019634e-06,
      "loss": 0.0053,
      "step": 13580
    },
    {
      "epoch": 0.2223676675120674,
      "grad_norm": 0.2597109251775523,
      "learning_rate": 9.550814770538e-06,
      "loss": 0.006,
      "step": 13590
    },
    {
      "epoch": 0.22253129346314326,
      "grad_norm": 0.42765706090153527,
      "learning_rate": 9.549630996433526e-06,
      "loss": 0.0066,
      "step": 13600
    },
    {
      "epoch": 0.2226949194142191,
      "grad_norm": 0.48884472589107786,
      "learning_rate": 9.548445738092397e-06,
      "loss": 0.0073,
      "step": 13610
    },
    {
      "epoch": 0.22285854536529492,
      "grad_norm": 0.15532669916419392,
      "learning_rate": 9.547258995901286e-06,
      "loss": 0.0075,
      "step": 13620
    },
    {
      "epoch": 0.22302217131637078,
      "grad_norm": 0.157113377991791,
      "learning_rate": 9.546070770247345e-06,
      "loss": 0.0068,
      "step": 13630
    },
    {
      "epoch": 0.2231857972674466,
      "grad_norm": 0.36220899808927415,
      "learning_rate": 9.54488106151821e-06,
      "loss": 0.0048,
      "step": 13640
    },
    {
      "epoch": 0.22334942321852247,
      "grad_norm": 0.28719435288237044,
      "learning_rate": 9.543689870102004e-06,
      "loss": 0.0042,
      "step": 13650
    },
    {
      "epoch": 0.2235130491695983,
      "grad_norm": 0.25210672502219006,
      "learning_rate": 9.542497196387332e-06,
      "loss": 0.0053,
      "step": 13660
    },
    {
      "epoch": 0.22367667512067413,
      "grad_norm": 0.42478007611712965,
      "learning_rate": 9.541303040763282e-06,
      "loss": 0.0057,
      "step": 13670
    },
    {
      "epoch": 0.22384030107175,
      "grad_norm": 0.21816506900661323,
      "learning_rate": 9.540107403619427e-06,
      "loss": 0.0068,
      "step": 13680
    },
    {
      "epoch": 0.22400392702282582,
      "grad_norm": 0.5742875054648906,
      "learning_rate": 9.538910285345823e-06,
      "loss": 0.0051,
      "step": 13690
    },
    {
      "epoch": 0.22416755297390167,
      "grad_norm": 0.2873033248646462,
      "learning_rate": 9.537711686333006e-06,
      "loss": 0.0072,
      "step": 13700
    },
    {
      "epoch": 0.2243311789249775,
      "grad_norm": 0.4977841781484335,
      "learning_rate": 9.536511606972002e-06,
      "loss": 0.0046,
      "step": 13710
    },
    {
      "epoch": 0.22449480487605333,
      "grad_norm": 0.42424990248621275,
      "learning_rate": 9.535310047654311e-06,
      "loss": 0.0051,
      "step": 13720
    },
    {
      "epoch": 0.2246584308271292,
      "grad_norm": 0.34056986370649706,
      "learning_rate": 9.534107008771924e-06,
      "loss": 0.0076,
      "step": 13730
    },
    {
      "epoch": 0.22482205677820502,
      "grad_norm": 0.2657361732948001,
      "learning_rate": 9.53290249071731e-06,
      "loss": 0.0066,
      "step": 13740
    },
    {
      "epoch": 0.22498568272928085,
      "grad_norm": 0.25114005605147427,
      "learning_rate": 9.53169649388342e-06,
      "loss": 0.0055,
      "step": 13750
    },
    {
      "epoch": 0.2251493086803567,
      "grad_norm": 0.2726808588141479,
      "learning_rate": 9.530489018663693e-06,
      "loss": 0.0057,
      "step": 13760
    },
    {
      "epoch": 0.22531293463143254,
      "grad_norm": 0.3530846299195506,
      "learning_rate": 9.529280065452042e-06,
      "loss": 0.0124,
      "step": 13770
    },
    {
      "epoch": 0.2254765605825084,
      "grad_norm": 0.2818722653891957,
      "learning_rate": 9.528069634642871e-06,
      "loss": 0.0049,
      "step": 13780
    },
    {
      "epoch": 0.22564018653358422,
      "grad_norm": 0.42331227165057544,
      "learning_rate": 9.526857726631059e-06,
      "loss": 0.0078,
      "step": 13790
    },
    {
      "epoch": 0.22580381248466005,
      "grad_norm": 0.4485415808800704,
      "learning_rate": 9.525644341811968e-06,
      "loss": 0.0071,
      "step": 13800
    },
    {
      "epoch": 0.2259674384357359,
      "grad_norm": 0.19258206594422,
      "learning_rate": 9.524429480581445e-06,
      "loss": 0.0055,
      "step": 13810
    },
    {
      "epoch": 0.22613106438681174,
      "grad_norm": 0.3897623178419447,
      "learning_rate": 9.52321314333582e-06,
      "loss": 0.0071,
      "step": 13820
    },
    {
      "epoch": 0.2262946903378876,
      "grad_norm": 0.2321282129526326,
      "learning_rate": 9.521995330471898e-06,
      "loss": 0.0065,
      "step": 13830
    },
    {
      "epoch": 0.22645831628896343,
      "grad_norm": 0.22893327276089045,
      "learning_rate": 9.520776042386968e-06,
      "loss": 0.0085,
      "step": 13840
    },
    {
      "epoch": 0.22662194224003926,
      "grad_norm": 0.6576099504685553,
      "learning_rate": 9.519555279478805e-06,
      "loss": 0.0073,
      "step": 13850
    },
    {
      "epoch": 0.22678556819111512,
      "grad_norm": 0.4418565426040817,
      "learning_rate": 9.518333042145658e-06,
      "loss": 0.0065,
      "step": 13860
    },
    {
      "epoch": 0.22694919414219095,
      "grad_norm": 0.3819352573777527,
      "learning_rate": 9.517109330786261e-06,
      "loss": 0.0088,
      "step": 13870
    },
    {
      "epoch": 0.2271128200932668,
      "grad_norm": 0.26155817292046624,
      "learning_rate": 9.51588414579983e-06,
      "loss": 0.0067,
      "step": 13880
    },
    {
      "epoch": 0.22727644604434263,
      "grad_norm": 0.12584201244957097,
      "learning_rate": 9.514657487586059e-06,
      "loss": 0.0059,
      "step": 13890
    },
    {
      "epoch": 0.22744007199541846,
      "grad_norm": 0.1909321313099596,
      "learning_rate": 9.513429356545121e-06,
      "loss": 0.0073,
      "step": 13900
    },
    {
      "epoch": 0.22760369794649432,
      "grad_norm": 0.2909084391114,
      "learning_rate": 9.512199753077677e-06,
      "loss": 0.0066,
      "step": 13910
    },
    {
      "epoch": 0.22776732389757015,
      "grad_norm": 0.2551113146650867,
      "learning_rate": 9.510968677584858e-06,
      "loss": 0.0067,
      "step": 13920
    },
    {
      "epoch": 0.227930949848646,
      "grad_norm": 0.17608195052899103,
      "learning_rate": 9.509736130468285e-06,
      "loss": 0.0076,
      "step": 13930
    },
    {
      "epoch": 0.22809457579972184,
      "grad_norm": 0.11151091951660022,
      "learning_rate": 9.508502112130053e-06,
      "loss": 0.0055,
      "step": 13940
    },
    {
      "epoch": 0.22825820175079767,
      "grad_norm": 0.39610436178992064,
      "learning_rate": 9.507266622972738e-06,
      "loss": 0.0049,
      "step": 13950
    },
    {
      "epoch": 0.22842182770187353,
      "grad_norm": 0.4088406968195816,
      "learning_rate": 9.506029663399397e-06,
      "loss": 0.005,
      "step": 13960
    },
    {
      "epoch": 0.22858545365294936,
      "grad_norm": 0.5556103262128795,
      "learning_rate": 9.504791233813567e-06,
      "loss": 0.0063,
      "step": 13970
    },
    {
      "epoch": 0.22874907960402519,
      "grad_norm": 0.3142193477238952,
      "learning_rate": 9.50355133461926e-06,
      "loss": 0.0057,
      "step": 13980
    },
    {
      "epoch": 0.22891270555510104,
      "grad_norm": 0.34651416136300606,
      "learning_rate": 9.502309966220978e-06,
      "loss": 0.0079,
      "step": 13990
    },
    {
      "epoch": 0.22907633150617687,
      "grad_norm": 0.2128523021083095,
      "learning_rate": 9.50106712902369e-06,
      "loss": 0.0076,
      "step": 14000
    },
    {
      "epoch": 0.22923995745725273,
      "grad_norm": 0.5650278412198942,
      "learning_rate": 9.499822823432851e-06,
      "loss": 0.0056,
      "step": 14010
    },
    {
      "epoch": 0.22940358340832856,
      "grad_norm": 0.4706166143954149,
      "learning_rate": 9.498577049854395e-06,
      "loss": 0.009,
      "step": 14020
    },
    {
      "epoch": 0.2295672093594044,
      "grad_norm": 0.262413434562026,
      "learning_rate": 9.497329808694732e-06,
      "loss": 0.004,
      "step": 14030
    },
    {
      "epoch": 0.22973083531048025,
      "grad_norm": 0.18732504958743518,
      "learning_rate": 9.496081100360753e-06,
      "loss": 0.0075,
      "step": 14040
    },
    {
      "epoch": 0.22989446126155608,
      "grad_norm": 0.34662797689048,
      "learning_rate": 9.494830925259828e-06,
      "loss": 0.006,
      "step": 14050
    },
    {
      "epoch": 0.23005808721263193,
      "grad_norm": 0.3453315017811865,
      "learning_rate": 9.493579283799803e-06,
      "loss": 0.0056,
      "step": 14060
    },
    {
      "epoch": 0.23022171316370776,
      "grad_norm": 0.26983668964754903,
      "learning_rate": 9.492326176389004e-06,
      "loss": 0.0052,
      "step": 14070
    },
    {
      "epoch": 0.2303853391147836,
      "grad_norm": 0.7037108382465352,
      "learning_rate": 9.491071603436235e-06,
      "loss": 0.0118,
      "step": 14080
    },
    {
      "epoch": 0.23054896506585945,
      "grad_norm": 0.2679989652015329,
      "learning_rate": 9.48981556535078e-06,
      "loss": 0.0052,
      "step": 14090
    },
    {
      "epoch": 0.23071259101693528,
      "grad_norm": 0.5462982707730084,
      "learning_rate": 9.488558062542396e-06,
      "loss": 0.0058,
      "step": 14100
    },
    {
      "epoch": 0.23087621696801114,
      "grad_norm": 0.14330988475015488,
      "learning_rate": 9.487299095421325e-06,
      "loss": 0.0035,
      "step": 14110
    },
    {
      "epoch": 0.23103984291908697,
      "grad_norm": 0.37262643538775314,
      "learning_rate": 9.48603866439828e-06,
      "loss": 0.0063,
      "step": 14120
    },
    {
      "epoch": 0.2312034688701628,
      "grad_norm": 0.34006463751875293,
      "learning_rate": 9.484776769884457e-06,
      "loss": 0.0079,
      "step": 14130
    },
    {
      "epoch": 0.23136709482123866,
      "grad_norm": 0.10388228446029518,
      "learning_rate": 9.483513412291522e-06,
      "loss": 0.0066,
      "step": 14140
    },
    {
      "epoch": 0.23153072077231449,
      "grad_norm": 0.09450310499555999,
      "learning_rate": 9.482248592031629e-06,
      "loss": 0.0079,
      "step": 14150
    },
    {
      "epoch": 0.23169434672339034,
      "grad_norm": 0.3040662061620897,
      "learning_rate": 9.480982309517398e-06,
      "loss": 0.005,
      "step": 14160
    },
    {
      "epoch": 0.23185797267446617,
      "grad_norm": 0.16827225121144465,
      "learning_rate": 9.479714565161934e-06,
      "loss": 0.0055,
      "step": 14170
    },
    {
      "epoch": 0.232021598625542,
      "grad_norm": 0.5489804910377039,
      "learning_rate": 9.478445359378816e-06,
      "loss": 0.0086,
      "step": 14180
    },
    {
      "epoch": 0.23218522457661786,
      "grad_norm": 0.5534358293896808,
      "learning_rate": 9.477174692582099e-06,
      "loss": 0.0071,
      "step": 14190
    },
    {
      "epoch": 0.2323488505276937,
      "grad_norm": 0.3129451956399419,
      "learning_rate": 9.475902565186316e-06,
      "loss": 0.0069,
      "step": 14200
    },
    {
      "epoch": 0.23251247647876952,
      "grad_norm": 0.39368717151298316,
      "learning_rate": 9.474628977606477e-06,
      "loss": 0.0083,
      "step": 14210
    },
    {
      "epoch": 0.23267610242984538,
      "grad_norm": 0.11459955028198908,
      "learning_rate": 9.473353930258067e-06,
      "loss": 0.0065,
      "step": 14220
    },
    {
      "epoch": 0.2328397283809212,
      "grad_norm": 0.44143179423929074,
      "learning_rate": 9.472077423557047e-06,
      "loss": 0.0076,
      "step": 14230
    },
    {
      "epoch": 0.23300335433199706,
      "grad_norm": 0.21452633381719527,
      "learning_rate": 9.470799457919855e-06,
      "loss": 0.0058,
      "step": 14240
    },
    {
      "epoch": 0.2331669802830729,
      "grad_norm": 0.29207166733800266,
      "learning_rate": 9.469520033763404e-06,
      "loss": 0.0076,
      "step": 14250
    },
    {
      "epoch": 0.23333060623414872,
      "grad_norm": 0.4625890655592877,
      "learning_rate": 9.468239151505085e-06,
      "loss": 0.0061,
      "step": 14260
    },
    {
      "epoch": 0.23349423218522458,
      "grad_norm": 0.13674762024956413,
      "learning_rate": 9.466956811562762e-06,
      "loss": 0.0068,
      "step": 14270
    },
    {
      "epoch": 0.2336578581363004,
      "grad_norm": 0.3334218910275108,
      "learning_rate": 9.465673014354777e-06,
      "loss": 0.0046,
      "step": 14280
    },
    {
      "epoch": 0.23382148408737627,
      "grad_norm": 0.209128114710864,
      "learning_rate": 9.464387760299946e-06,
      "loss": 0.0052,
      "step": 14290
    },
    {
      "epoch": 0.2339851100384521,
      "grad_norm": 0.5241121573280724,
      "learning_rate": 9.463101049817559e-06,
      "loss": 0.0081,
      "step": 14300
    },
    {
      "epoch": 0.23414873598952793,
      "grad_norm": 0.3870482538755865,
      "learning_rate": 9.461812883327382e-06,
      "loss": 0.0056,
      "step": 14310
    },
    {
      "epoch": 0.23431236194060379,
      "grad_norm": 0.6962889227049897,
      "learning_rate": 9.460523261249659e-06,
      "loss": 0.0103,
      "step": 14320
    },
    {
      "epoch": 0.23447598789167962,
      "grad_norm": 0.26858273092528284,
      "learning_rate": 9.459232184005104e-06,
      "loss": 0.0066,
      "step": 14330
    },
    {
      "epoch": 0.23463961384275547,
      "grad_norm": 0.3292942805736939,
      "learning_rate": 9.457939652014912e-06,
      "loss": 0.0066,
      "step": 14340
    },
    {
      "epoch": 0.2348032397938313,
      "grad_norm": 0.0901567977586052,
      "learning_rate": 9.456645665700744e-06,
      "loss": 0.0086,
      "step": 14350
    },
    {
      "epoch": 0.23496686574490713,
      "grad_norm": 0.30705374505818217,
      "learning_rate": 9.45535022548474e-06,
      "loss": 0.0081,
      "step": 14360
    },
    {
      "epoch": 0.235130491695983,
      "grad_norm": 0.15174008851125237,
      "learning_rate": 9.45405333178952e-06,
      "loss": 0.0061,
      "step": 14370
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.24054074207388504,
      "learning_rate": 9.452754985038168e-06,
      "loss": 0.0076,
      "step": 14380
    },
    {
      "epoch": 0.23545774359813468,
      "grad_norm": 0.20620360373956154,
      "learning_rate": 9.451455185654248e-06,
      "loss": 0.0096,
      "step": 14390
    },
    {
      "epoch": 0.2356213695492105,
      "grad_norm": 0.4700818658116053,
      "learning_rate": 9.450153934061796e-06,
      "loss": 0.0086,
      "step": 14400
    },
    {
      "epoch": 0.23578499550028634,
      "grad_norm": 0.36575437605963235,
      "learning_rate": 9.448851230685325e-06,
      "loss": 0.0068,
      "step": 14410
    },
    {
      "epoch": 0.2359486214513622,
      "grad_norm": 0.360341990905771,
      "learning_rate": 9.447547075949815e-06,
      "loss": 0.0052,
      "step": 14420
    },
    {
      "epoch": 0.23611224740243802,
      "grad_norm": 0.399218996701051,
      "learning_rate": 9.446241470280726e-06,
      "loss": 0.0068,
      "step": 14430
    },
    {
      "epoch": 0.23627587335351385,
      "grad_norm": 0.6245088876773593,
      "learning_rate": 9.444934414103986e-06,
      "loss": 0.0058,
      "step": 14440
    },
    {
      "epoch": 0.2364394993045897,
      "grad_norm": 0.30598324129481186,
      "learning_rate": 9.443625907846002e-06,
      "loss": 0.009,
      "step": 14450
    },
    {
      "epoch": 0.23660312525566554,
      "grad_norm": 0.14514530750061175,
      "learning_rate": 9.44231595193365e-06,
      "loss": 0.0047,
      "step": 14460
    },
    {
      "epoch": 0.2367667512067414,
      "grad_norm": 0.5858242717729784,
      "learning_rate": 9.44100454679428e-06,
      "loss": 0.0058,
      "step": 14470
    },
    {
      "epoch": 0.23693037715781723,
      "grad_norm": 0.36480864996203793,
      "learning_rate": 9.439691692855715e-06,
      "loss": 0.0048,
      "step": 14480
    },
    {
      "epoch": 0.23709400310889306,
      "grad_norm": 1.2661755140068547,
      "learning_rate": 9.43837739054625e-06,
      "loss": 0.0052,
      "step": 14490
    },
    {
      "epoch": 0.23725762905996892,
      "grad_norm": 0.09276635785854584,
      "learning_rate": 9.437061640294651e-06,
      "loss": 0.0034,
      "step": 14500
    },
    {
      "epoch": 0.23742125501104475,
      "grad_norm": 0.0925421417282902,
      "learning_rate": 9.435744442530162e-06,
      "loss": 0.0056,
      "step": 14510
    },
    {
      "epoch": 0.2375848809621206,
      "grad_norm": 0.310637867984842,
      "learning_rate": 9.434425797682493e-06,
      "loss": 0.0052,
      "step": 14520
    },
    {
      "epoch": 0.23774850691319643,
      "grad_norm": 0.3060726422201542,
      "learning_rate": 9.433105706181829e-06,
      "loss": 0.0077,
      "step": 14530
    },
    {
      "epoch": 0.23791213286427226,
      "grad_norm": 0.33467331135817335,
      "learning_rate": 9.431784168458827e-06,
      "loss": 0.0038,
      "step": 14540
    },
    {
      "epoch": 0.23807575881534812,
      "grad_norm": 0.6404301844845167,
      "learning_rate": 9.430461184944615e-06,
      "loss": 0.0054,
      "step": 14550
    },
    {
      "epoch": 0.23823938476642395,
      "grad_norm": 0.9709040983481245,
      "learning_rate": 9.429136756070793e-06,
      "loss": 0.0046,
      "step": 14560
    },
    {
      "epoch": 0.2384030107174998,
      "grad_norm": 0.24878807963288466,
      "learning_rate": 9.427810882269432e-06,
      "loss": 0.0049,
      "step": 14570
    },
    {
      "epoch": 0.23856663666857564,
      "grad_norm": 0.316034350819017,
      "learning_rate": 9.426483563973075e-06,
      "loss": 0.0057,
      "step": 14580
    },
    {
      "epoch": 0.23873026261965147,
      "grad_norm": 0.610862023166375,
      "learning_rate": 9.425154801614737e-06,
      "loss": 0.0089,
      "step": 14590
    },
    {
      "epoch": 0.23889388857072733,
      "grad_norm": 0.09416937810722693,
      "learning_rate": 9.423824595627904e-06,
      "loss": 0.0041,
      "step": 14600
    },
    {
      "epoch": 0.23905751452180315,
      "grad_norm": 0.47276794221891294,
      "learning_rate": 9.422492946446529e-06,
      "loss": 0.008,
      "step": 14610
    },
    {
      "epoch": 0.23922114047287898,
      "grad_norm": 0.23250562888546936,
      "learning_rate": 9.42115985450504e-06,
      "loss": 0.0056,
      "step": 14620
    },
    {
      "epoch": 0.23938476642395484,
      "grad_norm": 0.10945151238661889,
      "learning_rate": 9.419825320238337e-06,
      "loss": 0.0049,
      "step": 14630
    },
    {
      "epoch": 0.23954839237503067,
      "grad_norm": 0.37550509391419934,
      "learning_rate": 9.418489344081788e-06,
      "loss": 0.0055,
      "step": 14640
    },
    {
      "epoch": 0.23971201832610653,
      "grad_norm": 0.3096162291938593,
      "learning_rate": 9.417151926471227e-06,
      "loss": 0.0059,
      "step": 14650
    },
    {
      "epoch": 0.23987564427718236,
      "grad_norm": 0.17577959004882365,
      "learning_rate": 9.415813067842968e-06,
      "loss": 0.008,
      "step": 14660
    },
    {
      "epoch": 0.2400392702282582,
      "grad_norm": 0.19749149213743192,
      "learning_rate": 9.41447276863379e-06,
      "loss": 0.0069,
      "step": 14670
    },
    {
      "epoch": 0.24020289617933405,
      "grad_norm": 0.2311033362831667,
      "learning_rate": 9.413131029280937e-06,
      "loss": 0.0063,
      "step": 14680
    },
    {
      "epoch": 0.24036652213040988,
      "grad_norm": 0.4104589976868849,
      "learning_rate": 9.411787850222131e-06,
      "loss": 0.004,
      "step": 14690
    },
    {
      "epoch": 0.24053014808148573,
      "grad_norm": 0.11229179818877091,
      "learning_rate": 9.41044323189556e-06,
      "loss": 0.0046,
      "step": 14700
    },
    {
      "epoch": 0.24069377403256156,
      "grad_norm": 0.4742399403936828,
      "learning_rate": 9.409097174739885e-06,
      "loss": 0.0077,
      "step": 14710
    },
    {
      "epoch": 0.2408573999836374,
      "grad_norm": 0.34882810854198265,
      "learning_rate": 9.407749679194228e-06,
      "loss": 0.0042,
      "step": 14720
    },
    {
      "epoch": 0.24102102593471325,
      "grad_norm": 0.16425028064596917,
      "learning_rate": 9.406400745698188e-06,
      "loss": 0.0057,
      "step": 14730
    },
    {
      "epoch": 0.24118465188578908,
      "grad_norm": 0.22104259379452276,
      "learning_rate": 9.405050374691832e-06,
      "loss": 0.0051,
      "step": 14740
    },
    {
      "epoch": 0.24134827783686494,
      "grad_norm": 0.41523541428014976,
      "learning_rate": 9.403698566615693e-06,
      "loss": 0.0063,
      "step": 14750
    },
    {
      "epoch": 0.24151190378794077,
      "grad_norm": 0.3978342793878649,
      "learning_rate": 9.402345321910775e-06,
      "loss": 0.0072,
      "step": 14760
    },
    {
      "epoch": 0.2416755297390166,
      "grad_norm": 0.2687176646482882,
      "learning_rate": 9.400990641018549e-06,
      "loss": 0.0063,
      "step": 14770
    },
    {
      "epoch": 0.24183915569009246,
      "grad_norm": 0.26751074241939615,
      "learning_rate": 9.399634524380959e-06,
      "loss": 0.0046,
      "step": 14780
    },
    {
      "epoch": 0.24200278164116829,
      "grad_norm": 0.20211179922679004,
      "learning_rate": 9.398276972440409e-06,
      "loss": 0.0039,
      "step": 14790
    },
    {
      "epoch": 0.24216640759224414,
      "grad_norm": 0.1888648499768089,
      "learning_rate": 9.39691798563978e-06,
      "loss": 0.0072,
      "step": 14800
    },
    {
      "epoch": 0.24233003354331997,
      "grad_norm": 0.40591294721099713,
      "learning_rate": 9.395557564422418e-06,
      "loss": 0.0058,
      "step": 14810
    },
    {
      "epoch": 0.2424936594943958,
      "grad_norm": 0.3004720116766287,
      "learning_rate": 9.394195709232132e-06,
      "loss": 0.0066,
      "step": 14820
    },
    {
      "epoch": 0.24265728544547166,
      "grad_norm": 0.3008383073629911,
      "learning_rate": 9.392832420513208e-06,
      "loss": 0.0096,
      "step": 14830
    },
    {
      "epoch": 0.2428209113965475,
      "grad_norm": 0.15202838911585395,
      "learning_rate": 9.391467698710393e-06,
      "loss": 0.0061,
      "step": 14840
    },
    {
      "epoch": 0.24298453734762332,
      "grad_norm": 0.10839141651285336,
      "learning_rate": 9.390101544268902e-06,
      "loss": 0.0061,
      "step": 14850
    },
    {
      "epoch": 0.24314816329869918,
      "grad_norm": 0.19685018593162384,
      "learning_rate": 9.388733957634422e-06,
      "loss": 0.0072,
      "step": 14860
    },
    {
      "epoch": 0.243311789249775,
      "grad_norm": 0.2182779570417821,
      "learning_rate": 9.387364939253101e-06,
      "loss": 0.0049,
      "step": 14870
    },
    {
      "epoch": 0.24347541520085086,
      "grad_norm": 0.3135913264053835,
      "learning_rate": 9.385994489571557e-06,
      "loss": 0.0047,
      "step": 14880
    },
    {
      "epoch": 0.2436390411519267,
      "grad_norm": 0.17898868179664465,
      "learning_rate": 9.384622609036876e-06,
      "loss": 0.0043,
      "step": 14890
    },
    {
      "epoch": 0.24380266710300252,
      "grad_norm": 0.2480786729202796,
      "learning_rate": 9.383249298096611e-06,
      "loss": 0.004,
      "step": 14900
    },
    {
      "epoch": 0.24396629305407838,
      "grad_norm": 0.20302182619123238,
      "learning_rate": 9.381874557198777e-06,
      "loss": 0.0101,
      "step": 14910
    },
    {
      "epoch": 0.2441299190051542,
      "grad_norm": 0.20275127170617238,
      "learning_rate": 9.380498386791863e-06,
      "loss": 0.0037,
      "step": 14920
    },
    {
      "epoch": 0.24429354495623007,
      "grad_norm": 0.209137415493891,
      "learning_rate": 9.379120787324818e-06,
      "loss": 0.008,
      "step": 14930
    },
    {
      "epoch": 0.2444571709073059,
      "grad_norm": 0.11077655780576559,
      "learning_rate": 9.37774175924706e-06,
      "loss": 0.0048,
      "step": 14940
    },
    {
      "epoch": 0.24462079685838173,
      "grad_norm": 0.31295377068334534,
      "learning_rate": 9.376361303008472e-06,
      "loss": 0.0052,
      "step": 14950
    },
    {
      "epoch": 0.24478442280945759,
      "grad_norm": 0.290928570598236,
      "learning_rate": 9.374979419059402e-06,
      "loss": 0.0062,
      "step": 14960
    },
    {
      "epoch": 0.24494804876053342,
      "grad_norm": 0.2678787881437659,
      "learning_rate": 9.373596107850669e-06,
      "loss": 0.0075,
      "step": 14970
    },
    {
      "epoch": 0.24511167471160927,
      "grad_norm": 0.48207848539764514,
      "learning_rate": 9.37221136983355e-06,
      "loss": 0.0078,
      "step": 14980
    },
    {
      "epoch": 0.2452753006626851,
      "grad_norm": 0.110899440414048,
      "learning_rate": 9.370825205459795e-06,
      "loss": 0.0057,
      "step": 14990
    },
    {
      "epoch": 0.24543892661376093,
      "grad_norm": 0.5391182225649838,
      "learning_rate": 9.369437615181612e-06,
      "loss": 0.0068,
      "step": 15000
    },
    {
      "epoch": 0.2456025525648368,
      "grad_norm": 0.5818721569796923,
      "learning_rate": 9.36804859945168e-06,
      "loss": 0.0037,
      "step": 15010
    },
    {
      "epoch": 0.24576617851591262,
      "grad_norm": 0.3073631380529355,
      "learning_rate": 9.36665815872314e-06,
      "loss": 0.0065,
      "step": 15020
    },
    {
      "epoch": 0.24592980446698848,
      "grad_norm": 0.17585112163546068,
      "learning_rate": 9.3652662934496e-06,
      "loss": 0.0062,
      "step": 15030
    },
    {
      "epoch": 0.2460934304180643,
      "grad_norm": 0.4154698646351483,
      "learning_rate": 9.36387300408513e-06,
      "loss": 0.0076,
      "step": 15040
    },
    {
      "epoch": 0.24625705636914014,
      "grad_norm": 0.2045864347089957,
      "learning_rate": 9.362478291084264e-06,
      "loss": 0.0074,
      "step": 15050
    },
    {
      "epoch": 0.246420682320216,
      "grad_norm": 0.2602914395562712,
      "learning_rate": 9.361082154902006e-06,
      "loss": 0.0051,
      "step": 15060
    },
    {
      "epoch": 0.24658430827129182,
      "grad_norm": 0.1735322801753643,
      "learning_rate": 9.359684595993821e-06,
      "loss": 0.0049,
      "step": 15070
    },
    {
      "epoch": 0.24674793422236765,
      "grad_norm": 0.21999665214678835,
      "learning_rate": 9.358285614815635e-06,
      "loss": 0.0078,
      "step": 15080
    },
    {
      "epoch": 0.2469115601734435,
      "grad_norm": 0.14946872981848264,
      "learning_rate": 9.356885211823844e-06,
      "loss": 0.0041,
      "step": 15090
    },
    {
      "epoch": 0.24707518612451934,
      "grad_norm": 0.23823262693319208,
      "learning_rate": 9.3554833874753e-06,
      "loss": 0.0047,
      "step": 15100
    },
    {
      "epoch": 0.2472388120755952,
      "grad_norm": 0.15335509595211821,
      "learning_rate": 9.35408014222733e-06,
      "loss": 0.0075,
      "step": 15110
    },
    {
      "epoch": 0.24740243802667103,
      "grad_norm": 0.35151024609445214,
      "learning_rate": 9.352675476537712e-06,
      "loss": 0.0047,
      "step": 15120
    },
    {
      "epoch": 0.24756606397774686,
      "grad_norm": 0.3084663704142106,
      "learning_rate": 9.351269390864695e-06,
      "loss": 0.0049,
      "step": 15130
    },
    {
      "epoch": 0.24772968992882272,
      "grad_norm": 0.3124341271759404,
      "learning_rate": 9.34986188566699e-06,
      "loss": 0.0057,
      "step": 15140
    },
    {
      "epoch": 0.24789331587989855,
      "grad_norm": 0.18709997700104167,
      "learning_rate": 9.348452961403775e-06,
      "loss": 0.0044,
      "step": 15150
    },
    {
      "epoch": 0.2480569418309744,
      "grad_norm": 0.29356791694704526,
      "learning_rate": 9.34704261853468e-06,
      "loss": 0.0045,
      "step": 15160
    },
    {
      "epoch": 0.24822056778205023,
      "grad_norm": 0.17566762489209925,
      "learning_rate": 9.345630857519806e-06,
      "loss": 0.0054,
      "step": 15170
    },
    {
      "epoch": 0.24838419373312606,
      "grad_norm": 0.551096575642366,
      "learning_rate": 9.344217678819718e-06,
      "loss": 0.0068,
      "step": 15180
    },
    {
      "epoch": 0.24854781968420192,
      "grad_norm": 0.131767090722387,
      "learning_rate": 9.342803082895438e-06,
      "loss": 0.0064,
      "step": 15190
    },
    {
      "epoch": 0.24871144563527775,
      "grad_norm": 0.2143855045691361,
      "learning_rate": 9.341387070208451e-06,
      "loss": 0.0055,
      "step": 15200
    },
    {
      "epoch": 0.2488750715863536,
      "grad_norm": 0.0788060313310521,
      "learning_rate": 9.339969641220711e-06,
      "loss": 0.0027,
      "step": 15210
    },
    {
      "epoch": 0.24903869753742944,
      "grad_norm": 0.2840929533913191,
      "learning_rate": 9.338550796394628e-06,
      "loss": 0.0055,
      "step": 15220
    },
    {
      "epoch": 0.24920232348850527,
      "grad_norm": 0.17721593948973868,
      "learning_rate": 9.337130536193072e-06,
      "loss": 0.0081,
      "step": 15230
    },
    {
      "epoch": 0.24936594943958112,
      "grad_norm": 0.2716418811540453,
      "learning_rate": 9.33570886107938e-06,
      "loss": 0.005,
      "step": 15240
    },
    {
      "epoch": 0.24952957539065695,
      "grad_norm": 0.715536111658266,
      "learning_rate": 9.334285771517348e-06,
      "loss": 0.004,
      "step": 15250
    },
    {
      "epoch": 0.2496932013417328,
      "grad_norm": 0.371328530470032,
      "learning_rate": 9.332861267971234e-06,
      "loss": 0.0071,
      "step": 15260
    },
    {
      "epoch": 0.24985682729280864,
      "grad_norm": 0.12260695624886554,
      "learning_rate": 9.331435350905756e-06,
      "loss": 0.0048,
      "step": 15270
    },
    {
      "epoch": 0.2500204532438845,
      "grad_norm": 0.3878963374555747,
      "learning_rate": 9.330008020786096e-06,
      "loss": 0.0063,
      "step": 15280
    },
    {
      "epoch": 0.25018407919496033,
      "grad_norm": 0.30388806020351955,
      "learning_rate": 9.328579278077894e-06,
      "loss": 0.0049,
      "step": 15290
    },
    {
      "epoch": 0.25034770514603616,
      "grad_norm": 0.5285349510757692,
      "learning_rate": 9.327149123247253e-06,
      "loss": 0.0059,
      "step": 15300
    },
    {
      "epoch": 0.250511331097112,
      "grad_norm": 0.3340642760538013,
      "learning_rate": 9.325717556760732e-06,
      "loss": 0.005,
      "step": 15310
    },
    {
      "epoch": 0.2506749570481878,
      "grad_norm": 0.2821603859086629,
      "learning_rate": 9.32428457908536e-06,
      "loss": 0.0052,
      "step": 15320
    },
    {
      "epoch": 0.2508385829992637,
      "grad_norm": 0.13198793199892456,
      "learning_rate": 9.322850190688618e-06,
      "loss": 0.0056,
      "step": 15330
    },
    {
      "epoch": 0.25100220895033953,
      "grad_norm": 0.24696458097131938,
      "learning_rate": 9.321414392038448e-06,
      "loss": 0.0046,
      "step": 15340
    },
    {
      "epoch": 0.25116583490141536,
      "grad_norm": 0.6434908805094101,
      "learning_rate": 9.319977183603257e-06,
      "loss": 0.0064,
      "step": 15350
    },
    {
      "epoch": 0.2513294608524912,
      "grad_norm": 0.5082098354708341,
      "learning_rate": 9.318538565851907e-06,
      "loss": 0.009,
      "step": 15360
    },
    {
      "epoch": 0.251493086803567,
      "grad_norm": 0.39458619037577347,
      "learning_rate": 9.317098539253723e-06,
      "loss": 0.0041,
      "step": 15370
    },
    {
      "epoch": 0.2516567127546429,
      "grad_norm": 0.18893194481615558,
      "learning_rate": 9.315657104278486e-06,
      "loss": 0.0052,
      "step": 15380
    },
    {
      "epoch": 0.25182033870571874,
      "grad_norm": 0.3799928632754006,
      "learning_rate": 9.31421426139644e-06,
      "loss": 0.0064,
      "step": 15390
    },
    {
      "epoch": 0.25198396465679457,
      "grad_norm": 0.14133898556258437,
      "learning_rate": 9.31277001107829e-06,
      "loss": 0.0072,
      "step": 15400
    },
    {
      "epoch": 0.2521475906078704,
      "grad_norm": 0.09071703250620583,
      "learning_rate": 9.31132435379519e-06,
      "loss": 0.0044,
      "step": 15410
    },
    {
      "epoch": 0.2523112165589462,
      "grad_norm": 0.362404729237176,
      "learning_rate": 9.309877290018764e-06,
      "loss": 0.0059,
      "step": 15420
    },
    {
      "epoch": 0.2524748425100221,
      "grad_norm": 0.4409186255538609,
      "learning_rate": 9.308428820221096e-06,
      "loss": 0.0042,
      "step": 15430
    },
    {
      "epoch": 0.25263846846109794,
      "grad_norm": 0.2176099883496552,
      "learning_rate": 9.306978944874715e-06,
      "loss": 0.0067,
      "step": 15440
    },
    {
      "epoch": 0.2528020944121738,
      "grad_norm": 0.2963205444288595,
      "learning_rate": 9.305527664452622e-06,
      "loss": 0.0043,
      "step": 15450
    },
    {
      "epoch": 0.2529657203632496,
      "grad_norm": 0.27428931792980327,
      "learning_rate": 9.30407497942827e-06,
      "loss": 0.0097,
      "step": 15460
    },
    {
      "epoch": 0.25312934631432543,
      "grad_norm": 0.23401933485377163,
      "learning_rate": 9.302620890275573e-06,
      "loss": 0.0033,
      "step": 15470
    },
    {
      "epoch": 0.2532929722654013,
      "grad_norm": 0.38879621381988466,
      "learning_rate": 9.3011653974689e-06,
      "loss": 0.0078,
      "step": 15480
    },
    {
      "epoch": 0.25345659821647715,
      "grad_norm": 0.5825043747382485,
      "learning_rate": 9.299708501483082e-06,
      "loss": 0.0097,
      "step": 15490
    },
    {
      "epoch": 0.253620224167553,
      "grad_norm": 0.23484761617081673,
      "learning_rate": 9.298250202793406e-06,
      "loss": 0.0099,
      "step": 15500
    },
    {
      "epoch": 0.2537838501186288,
      "grad_norm": 0.0730527484970993,
      "learning_rate": 9.296790501875613e-06,
      "loss": 0.0071,
      "step": 15510
    },
    {
      "epoch": 0.25394747606970464,
      "grad_norm": 0.18310148817053432,
      "learning_rate": 9.295329399205906e-06,
      "loss": 0.0044,
      "step": 15520
    },
    {
      "epoch": 0.2541111020207805,
      "grad_norm": 0.43931072882735667,
      "learning_rate": 9.293866895260946e-06,
      "loss": 0.0052,
      "step": 15530
    },
    {
      "epoch": 0.25427472797185635,
      "grad_norm": 0.10917023870661761,
      "learning_rate": 9.292402990517847e-06,
      "loss": 0.005,
      "step": 15540
    },
    {
      "epoch": 0.2544383539229322,
      "grad_norm": 0.19653198374575204,
      "learning_rate": 9.290937685454181e-06,
      "loss": 0.0099,
      "step": 15550
    },
    {
      "epoch": 0.254601979874008,
      "grad_norm": 0.5526590209031602,
      "learning_rate": 9.28947098054798e-06,
      "loss": 0.006,
      "step": 15560
    },
    {
      "epoch": 0.25476560582508384,
      "grad_norm": 0.29235142637747696,
      "learning_rate": 9.28800287627773e-06,
      "loss": 0.0037,
      "step": 15570
    },
    {
      "epoch": 0.2549292317761597,
      "grad_norm": 0.22452855765418456,
      "learning_rate": 9.286533373122372e-06,
      "loss": 0.0055,
      "step": 15580
    },
    {
      "epoch": 0.25509285772723556,
      "grad_norm": 0.4555840956003422,
      "learning_rate": 9.28506247156131e-06,
      "loss": 0.0057,
      "step": 15590
    },
    {
      "epoch": 0.2552564836783114,
      "grad_norm": 0.6662472624032246,
      "learning_rate": 9.283590172074396e-06,
      "loss": 0.0055,
      "step": 15600
    },
    {
      "epoch": 0.2554201096293872,
      "grad_norm": 0.39903543442574063,
      "learning_rate": 9.282116475141942e-06,
      "loss": 0.0077,
      "step": 15610
    },
    {
      "epoch": 0.25558373558046305,
      "grad_norm": 0.2221318833007634,
      "learning_rate": 9.280641381244716e-06,
      "loss": 0.0062,
      "step": 15620
    },
    {
      "epoch": 0.2557473615315389,
      "grad_norm": 0.4451123972798245,
      "learning_rate": 9.279164890863942e-06,
      "loss": 0.0043,
      "step": 15630
    },
    {
      "epoch": 0.25591098748261476,
      "grad_norm": 0.27825984329538817,
      "learning_rate": 9.277687004481298e-06,
      "loss": 0.0035,
      "step": 15640
    },
    {
      "epoch": 0.2560746134336906,
      "grad_norm": 0.49818548435411075,
      "learning_rate": 9.276207722578918e-06,
      "loss": 0.0064,
      "step": 15650
    },
    {
      "epoch": 0.2562382393847664,
      "grad_norm": 0.2214312824588999,
      "learning_rate": 9.274727045639394e-06,
      "loss": 0.004,
      "step": 15660
    },
    {
      "epoch": 0.25640186533584225,
      "grad_norm": 0.4074232127254207,
      "learning_rate": 9.273244974145765e-06,
      "loss": 0.0073,
      "step": 15670
    },
    {
      "epoch": 0.2565654912869181,
      "grad_norm": 0.027917935290623194,
      "learning_rate": 9.271761508581538e-06,
      "loss": 0.0052,
      "step": 15680
    },
    {
      "epoch": 0.25672911723799396,
      "grad_norm": 0.08697256400383539,
      "learning_rate": 9.270276649430662e-06,
      "loss": 0.005,
      "step": 15690
    },
    {
      "epoch": 0.2568927431890698,
      "grad_norm": 0.2098381040473996,
      "learning_rate": 9.268790397177548e-06,
      "loss": 0.0046,
      "step": 15700
    },
    {
      "epoch": 0.2570563691401456,
      "grad_norm": 0.247217340379413,
      "learning_rate": 9.26730275230706e-06,
      "loss": 0.0068,
      "step": 15710
    },
    {
      "epoch": 0.25721999509122145,
      "grad_norm": 0.3295195766262875,
      "learning_rate": 9.265813715304515e-06,
      "loss": 0.0066,
      "step": 15720
    },
    {
      "epoch": 0.2573836210422973,
      "grad_norm": 0.44338163580725737,
      "learning_rate": 9.264323286655685e-06,
      "loss": 0.0075,
      "step": 15730
    },
    {
      "epoch": 0.25754724699337317,
      "grad_norm": 0.25224477187620925,
      "learning_rate": 9.262831466846795e-06,
      "loss": 0.0108,
      "step": 15740
    },
    {
      "epoch": 0.257710872944449,
      "grad_norm": 0.36637598598161486,
      "learning_rate": 9.261338256364528e-06,
      "loss": 0.0064,
      "step": 15750
    },
    {
      "epoch": 0.25787449889552483,
      "grad_norm": 0.18728206841760078,
      "learning_rate": 9.259843655696014e-06,
      "loss": 0.0046,
      "step": 15760
    },
    {
      "epoch": 0.25803812484660066,
      "grad_norm": 0.42248866487523834,
      "learning_rate": 9.258347665328843e-06,
      "loss": 0.0057,
      "step": 15770
    },
    {
      "epoch": 0.2582017507976765,
      "grad_norm": 0.19325300019421338,
      "learning_rate": 9.256850285751053e-06,
      "loss": 0.0048,
      "step": 15780
    },
    {
      "epoch": 0.2583653767487524,
      "grad_norm": 0.1890348137555446,
      "learning_rate": 9.255351517451139e-06,
      "loss": 0.0034,
      "step": 15790
    },
    {
      "epoch": 0.2585290026998282,
      "grad_norm": 0.31362968408301894,
      "learning_rate": 9.253851360918048e-06,
      "loss": 0.0047,
      "step": 15800
    },
    {
      "epoch": 0.25869262865090403,
      "grad_norm": 0.04517231788366573,
      "learning_rate": 9.252349816641175e-06,
      "loss": 0.0043,
      "step": 15810
    },
    {
      "epoch": 0.25885625460197986,
      "grad_norm": 0.2170688319977448,
      "learning_rate": 9.250846885110379e-06,
      "loss": 0.0044,
      "step": 15820
    },
    {
      "epoch": 0.2590198805530557,
      "grad_norm": 0.32973057615182727,
      "learning_rate": 9.249342566815962e-06,
      "loss": 0.0046,
      "step": 15830
    },
    {
      "epoch": 0.2591835065041316,
      "grad_norm": 0.2912919858561665,
      "learning_rate": 9.24783686224868e-06,
      "loss": 0.0049,
      "step": 15840
    },
    {
      "epoch": 0.2593471324552074,
      "grad_norm": 0.28108342588539087,
      "learning_rate": 9.246329771899745e-06,
      "loss": 0.0052,
      "step": 15850
    },
    {
      "epoch": 0.25951075840628324,
      "grad_norm": 0.14382390288727165,
      "learning_rate": 9.244821296260816e-06,
      "loss": 0.0047,
      "step": 15860
    },
    {
      "epoch": 0.25967438435735907,
      "grad_norm": 0.40137604861663184,
      "learning_rate": 9.243311435824007e-06,
      "loss": 0.0047,
      "step": 15870
    },
    {
      "epoch": 0.2598380103084349,
      "grad_norm": 0.2998970927302701,
      "learning_rate": 9.241800191081885e-06,
      "loss": 0.0069,
      "step": 15880
    },
    {
      "epoch": 0.2600016362595108,
      "grad_norm": 0.31822855662971483,
      "learning_rate": 9.240287562527465e-06,
      "loss": 0.0041,
      "step": 15890
    },
    {
      "epoch": 0.2601652622105866,
      "grad_norm": 0.39731230282537305,
      "learning_rate": 9.238773550654216e-06,
      "loss": 0.0042,
      "step": 15900
    },
    {
      "epoch": 0.26032888816166244,
      "grad_norm": 0.36608323151151295,
      "learning_rate": 9.23725815595606e-06,
      "loss": 0.0053,
      "step": 15910
    },
    {
      "epoch": 0.26049251411273827,
      "grad_norm": 0.6111761802997118,
      "learning_rate": 9.235741378927362e-06,
      "loss": 0.0102,
      "step": 15920
    },
    {
      "epoch": 0.2606561400638141,
      "grad_norm": 0.267587602821666,
      "learning_rate": 9.23422322006295e-06,
      "loss": 0.0071,
      "step": 15930
    },
    {
      "epoch": 0.26081976601489,
      "grad_norm": 0.24980430542382004,
      "learning_rate": 9.232703679858092e-06,
      "loss": 0.0039,
      "step": 15940
    },
    {
      "epoch": 0.2609833919659658,
      "grad_norm": 0.17696743983110425,
      "learning_rate": 9.231182758808513e-06,
      "loss": 0.0043,
      "step": 15950
    },
    {
      "epoch": 0.26114701791704165,
      "grad_norm": 0.749844863030081,
      "learning_rate": 9.229660457410385e-06,
      "loss": 0.0125,
      "step": 15960
    },
    {
      "epoch": 0.2613106438681175,
      "grad_norm": 0.31216184453003504,
      "learning_rate": 9.228136776160336e-06,
      "loss": 0.0069,
      "step": 15970
    },
    {
      "epoch": 0.2614742698191933,
      "grad_norm": 0.19592320314798167,
      "learning_rate": 9.226611715555435e-06,
      "loss": 0.0051,
      "step": 15980
    },
    {
      "epoch": 0.2616378957702692,
      "grad_norm": 0.3399569954269113,
      "learning_rate": 9.225085276093208e-06,
      "loss": 0.0071,
      "step": 15990
    },
    {
      "epoch": 0.261801521721345,
      "grad_norm": 0.09711338006471965,
      "learning_rate": 9.223557458271631e-06,
      "loss": 0.0065,
      "step": 16000
    },
    {
      "epoch": 0.26196514767242085,
      "grad_norm": 0.1624696169494459,
      "learning_rate": 9.222028262589125e-06,
      "loss": 0.0062,
      "step": 16010
    },
    {
      "epoch": 0.2621287736234967,
      "grad_norm": 0.1464514361768134,
      "learning_rate": 9.220497689544565e-06,
      "loss": 0.0049,
      "step": 16020
    },
    {
      "epoch": 0.2622923995745725,
      "grad_norm": 0.315712020167498,
      "learning_rate": 9.21896573963727e-06,
      "loss": 0.0073,
      "step": 16030
    },
    {
      "epoch": 0.2624560255256484,
      "grad_norm": 0.2340710221026737,
      "learning_rate": 9.217432413367015e-06,
      "loss": 0.0057,
      "step": 16040
    },
    {
      "epoch": 0.2626196514767242,
      "grad_norm": 0.4481334158908465,
      "learning_rate": 9.215897711234019e-06,
      "loss": 0.006,
      "step": 16050
    },
    {
      "epoch": 0.26278327742780005,
      "grad_norm": 0.3681200585180176,
      "learning_rate": 9.214361633738951e-06,
      "loss": 0.0055,
      "step": 16060
    },
    {
      "epoch": 0.2629469033788759,
      "grad_norm": 0.9873727888723629,
      "learning_rate": 9.21282418138293e-06,
      "loss": 0.0042,
      "step": 16070
    },
    {
      "epoch": 0.2631105293299517,
      "grad_norm": 0.2502418228336838,
      "learning_rate": 9.211285354667525e-06,
      "loss": 0.0038,
      "step": 16080
    },
    {
      "epoch": 0.26327415528102754,
      "grad_norm": 0.3852506178246809,
      "learning_rate": 9.209745154094745e-06,
      "loss": 0.0061,
      "step": 16090
    },
    {
      "epoch": 0.26343778123210343,
      "grad_norm": 0.21968144079947483,
      "learning_rate": 9.20820358016706e-06,
      "loss": 0.0076,
      "step": 16100
    },
    {
      "epoch": 0.26360140718317926,
      "grad_norm": 0.5682113116868072,
      "learning_rate": 9.206660633387374e-06,
      "loss": 0.0065,
      "step": 16110
    },
    {
      "epoch": 0.2637650331342551,
      "grad_norm": 0.12985185614621797,
      "learning_rate": 9.205116314259052e-06,
      "loss": 0.0061,
      "step": 16120
    },
    {
      "epoch": 0.2639286590853309,
      "grad_norm": 0.24235605912591063,
      "learning_rate": 9.2035706232859e-06,
      "loss": 0.0057,
      "step": 16130
    },
    {
      "epoch": 0.26409228503640675,
      "grad_norm": 0.3977661903999316,
      "learning_rate": 9.202023560972171e-06,
      "loss": 0.0046,
      "step": 16140
    },
    {
      "epoch": 0.26425591098748263,
      "grad_norm": 0.1100857998918415,
      "learning_rate": 9.200475127822565e-06,
      "loss": 0.0054,
      "step": 16150
    },
    {
      "epoch": 0.26441953693855846,
      "grad_norm": 0.22110769508196634,
      "learning_rate": 9.198925324342234e-06,
      "loss": 0.005,
      "step": 16160
    },
    {
      "epoch": 0.2645831628896343,
      "grad_norm": 0.47260215709791953,
      "learning_rate": 9.197374151036772e-06,
      "loss": 0.0037,
      "step": 16170
    },
    {
      "epoch": 0.2647467888407101,
      "grad_norm": 0.427479677142539,
      "learning_rate": 9.195821608412226e-06,
      "loss": 0.0087,
      "step": 16180
    },
    {
      "epoch": 0.26491041479178595,
      "grad_norm": 0.4068049439722783,
      "learning_rate": 9.194267696975078e-06,
      "loss": 0.0121,
      "step": 16190
    },
    {
      "epoch": 0.26507404074286184,
      "grad_norm": 0.15077933513095784,
      "learning_rate": 9.19271241723227e-06,
      "loss": 0.0029,
      "step": 16200
    },
    {
      "epoch": 0.26523766669393767,
      "grad_norm": 0.40972426930346395,
      "learning_rate": 9.191155769691183e-06,
      "loss": 0.0053,
      "step": 16210
    },
    {
      "epoch": 0.2654012926450135,
      "grad_norm": 0.3646243472243028,
      "learning_rate": 9.189597754859646e-06,
      "loss": 0.0047,
      "step": 16220
    },
    {
      "epoch": 0.2655649185960893,
      "grad_norm": 0.2558549733916791,
      "learning_rate": 9.188038373245934e-06,
      "loss": 0.0066,
      "step": 16230
    },
    {
      "epoch": 0.26572854454716516,
      "grad_norm": 0.18987672564540053,
      "learning_rate": 9.186477625358765e-06,
      "loss": 0.0051,
      "step": 16240
    },
    {
      "epoch": 0.26589217049824104,
      "grad_norm": 0.19000960438179892,
      "learning_rate": 9.184915511707308e-06,
      "loss": 0.0042,
      "step": 16250
    },
    {
      "epoch": 0.2660557964493169,
      "grad_norm": 0.276272599895,
      "learning_rate": 9.183352032801176e-06,
      "loss": 0.0034,
      "step": 16260
    },
    {
      "epoch": 0.2662194224003927,
      "grad_norm": 0.32598802032195595,
      "learning_rate": 9.181787189150425e-06,
      "loss": 0.0064,
      "step": 16270
    },
    {
      "epoch": 0.26638304835146853,
      "grad_norm": 1.1762093269456946,
      "learning_rate": 9.180220981265555e-06,
      "loss": 0.0043,
      "step": 16280
    },
    {
      "epoch": 0.26654667430254436,
      "grad_norm": 0.2927702563467667,
      "learning_rate": 9.178653409657516e-06,
      "loss": 0.0057,
      "step": 16290
    },
    {
      "epoch": 0.26671030025362025,
      "grad_norm": 0.15808257048176252,
      "learning_rate": 9.177084474837702e-06,
      "loss": 0.0051,
      "step": 16300
    },
    {
      "epoch": 0.2668739262046961,
      "grad_norm": 0.25449279001389224,
      "learning_rate": 9.175514177317947e-06,
      "loss": 0.0055,
      "step": 16310
    },
    {
      "epoch": 0.2670375521557719,
      "grad_norm": 0.2658931474874935,
      "learning_rate": 9.173942517610537e-06,
      "loss": 0.0068,
      "step": 16320
    },
    {
      "epoch": 0.26720117810684774,
      "grad_norm": 0.37278190959936935,
      "learning_rate": 9.172369496228195e-06,
      "loss": 0.0065,
      "step": 16330
    },
    {
      "epoch": 0.26736480405792357,
      "grad_norm": 0.139793344097985,
      "learning_rate": 9.170795113684092e-06,
      "loss": 0.0076,
      "step": 16340
    },
    {
      "epoch": 0.26752843000899945,
      "grad_norm": 0.12159953997731686,
      "learning_rate": 9.169219370491842e-06,
      "loss": 0.0043,
      "step": 16350
    },
    {
      "epoch": 0.2676920559600753,
      "grad_norm": 0.6851385185370912,
      "learning_rate": 9.167642267165503e-06,
      "loss": 0.0071,
      "step": 16360
    },
    {
      "epoch": 0.2678556819111511,
      "grad_norm": 0.21222172206097425,
      "learning_rate": 9.16606380421958e-06,
      "loss": 0.0061,
      "step": 16370
    },
    {
      "epoch": 0.26801930786222694,
      "grad_norm": 0.2515939141550328,
      "learning_rate": 9.164483982169014e-06,
      "loss": 0.0069,
      "step": 16380
    },
    {
      "epoch": 0.26818293381330277,
      "grad_norm": 0.2673037322690824,
      "learning_rate": 9.1629028015292e-06,
      "loss": 0.0097,
      "step": 16390
    },
    {
      "epoch": 0.26834655976437866,
      "grad_norm": 0.27115041179488925,
      "learning_rate": 9.161320262815964e-06,
      "loss": 0.0059,
      "step": 16400
    },
    {
      "epoch": 0.2685101857154545,
      "grad_norm": 0.27777013320319616,
      "learning_rate": 9.159736366545585e-06,
      "loss": 0.0045,
      "step": 16410
    },
    {
      "epoch": 0.2686738116665303,
      "grad_norm": 0.4373588297429033,
      "learning_rate": 9.15815111323478e-06,
      "loss": 0.0078,
      "step": 16420
    },
    {
      "epoch": 0.26883743761760615,
      "grad_norm": 0.36616496539841586,
      "learning_rate": 9.156564503400711e-06,
      "loss": 0.0053,
      "step": 16430
    },
    {
      "epoch": 0.269001063568682,
      "grad_norm": 0.3664139287368561,
      "learning_rate": 9.15497653756098e-06,
      "loss": 0.0062,
      "step": 16440
    },
    {
      "epoch": 0.26916468951975786,
      "grad_norm": 0.3282865865169168,
      "learning_rate": 9.153387216233634e-06,
      "loss": 0.0058,
      "step": 16450
    },
    {
      "epoch": 0.2693283154708337,
      "grad_norm": 1.123811052445781,
      "learning_rate": 9.15179653993716e-06,
      "loss": 0.0063,
      "step": 16460
    },
    {
      "epoch": 0.2694919414219095,
      "grad_norm": 0.4781598360290196,
      "learning_rate": 9.150204509190487e-06,
      "loss": 0.0082,
      "step": 16470
    },
    {
      "epoch": 0.26965556737298535,
      "grad_norm": 0.4861164234865872,
      "learning_rate": 9.148611124512989e-06,
      "loss": 0.0084,
      "step": 16480
    },
    {
      "epoch": 0.2698191933240612,
      "grad_norm": 0.13751985791004354,
      "learning_rate": 9.147016386424478e-06,
      "loss": 0.0053,
      "step": 16490
    },
    {
      "epoch": 0.269982819275137,
      "grad_norm": 0.2194043172380987,
      "learning_rate": 9.145420295445211e-06,
      "loss": 0.0058,
      "step": 16500
    },
    {
      "epoch": 0.2701464452262129,
      "grad_norm": 0.4043655161163458,
      "learning_rate": 9.143822852095882e-06,
      "loss": 0.0074,
      "step": 16510
    },
    {
      "epoch": 0.2703100711772887,
      "grad_norm": 0.12532343095695817,
      "learning_rate": 9.142224056897632e-06,
      "loss": 0.0065,
      "step": 16520
    },
    {
      "epoch": 0.27047369712836455,
      "grad_norm": 0.404539253864606,
      "learning_rate": 9.140623910372036e-06,
      "loss": 0.0053,
      "step": 16530
    },
    {
      "epoch": 0.2706373230794404,
      "grad_norm": 0.39385997659071464,
      "learning_rate": 9.139022413041116e-06,
      "loss": 0.004,
      "step": 16540
    },
    {
      "epoch": 0.2708009490305162,
      "grad_norm": 0.30488845326475833,
      "learning_rate": 9.13741956542733e-06,
      "loss": 0.0106,
      "step": 16550
    },
    {
      "epoch": 0.2709645749815921,
      "grad_norm": 0.28862961532336373,
      "learning_rate": 9.13581536805358e-06,
      "loss": 0.0071,
      "step": 16560
    },
    {
      "epoch": 0.27112820093266793,
      "grad_norm": 0.16617848849213243,
      "learning_rate": 9.134209821443207e-06,
      "loss": 0.0048,
      "step": 16570
    },
    {
      "epoch": 0.27129182688374376,
      "grad_norm": 0.30063968274649416,
      "learning_rate": 9.132602926119993e-06,
      "loss": 0.0057,
      "step": 16580
    },
    {
      "epoch": 0.2714554528348196,
      "grad_norm": 0.32286472707696584,
      "learning_rate": 9.130994682608158e-06,
      "loss": 0.0045,
      "step": 16590
    },
    {
      "epoch": 0.2716190787858954,
      "grad_norm": 0.5709931154407546,
      "learning_rate": 9.129385091432363e-06,
      "loss": 0.0056,
      "step": 16600
    },
    {
      "epoch": 0.2717827047369713,
      "grad_norm": 0.2842297296367366,
      "learning_rate": 9.12777415311771e-06,
      "loss": 0.007,
      "step": 16610
    },
    {
      "epoch": 0.27194633068804713,
      "grad_norm": 0.14358795667053467,
      "learning_rate": 9.126161868189737e-06,
      "loss": 0.0037,
      "step": 16620
    },
    {
      "epoch": 0.27210995663912296,
      "grad_norm": 0.061896592319254755,
      "learning_rate": 9.124548237174426e-06,
      "loss": 0.0035,
      "step": 16630
    },
    {
      "epoch": 0.2722735825901988,
      "grad_norm": 0.13172238433050146,
      "learning_rate": 9.122933260598193e-06,
      "loss": 0.0052,
      "step": 16640
    },
    {
      "epoch": 0.2724372085412746,
      "grad_norm": 0.28068527681033073,
      "learning_rate": 9.121316938987897e-06,
      "loss": 0.0053,
      "step": 16650
    },
    {
      "epoch": 0.2726008344923505,
      "grad_norm": 0.37718647786653287,
      "learning_rate": 9.119699272870835e-06,
      "loss": 0.0065,
      "step": 16660
    },
    {
      "epoch": 0.27276446044342634,
      "grad_norm": 0.25271195197285157,
      "learning_rate": 9.118080262774738e-06,
      "loss": 0.0055,
      "step": 16670
    },
    {
      "epoch": 0.27292808639450217,
      "grad_norm": 0.24992086261862945,
      "learning_rate": 9.116459909227786e-06,
      "loss": 0.0029,
      "step": 16680
    },
    {
      "epoch": 0.273091712345578,
      "grad_norm": 0.5070520608668672,
      "learning_rate": 9.114838212758586e-06,
      "loss": 0.0066,
      "step": 16690
    },
    {
      "epoch": 0.2732553382966538,
      "grad_norm": 0.2468771328638775,
      "learning_rate": 9.113215173896186e-06,
      "loss": 0.0069,
      "step": 16700
    },
    {
      "epoch": 0.2734189642477297,
      "grad_norm": 0.3124143941338982,
      "learning_rate": 9.11159079317008e-06,
      "loss": 0.0051,
      "step": 16710
    },
    {
      "epoch": 0.27358259019880554,
      "grad_norm": 0.12410613250248503,
      "learning_rate": 9.10996507111019e-06,
      "loss": 0.0042,
      "step": 16720
    },
    {
      "epoch": 0.27374621614988137,
      "grad_norm": 0.29237440352959504,
      "learning_rate": 9.108338008246878e-06,
      "loss": 0.0093,
      "step": 16730
    },
    {
      "epoch": 0.2739098421009572,
      "grad_norm": 0.23018482446137556,
      "learning_rate": 9.106709605110946e-06,
      "loss": 0.0051,
      "step": 16740
    },
    {
      "epoch": 0.27407346805203303,
      "grad_norm": 0.43116424608691356,
      "learning_rate": 9.10507986223363e-06,
      "loss": 0.0035,
      "step": 16750
    },
    {
      "epoch": 0.2742370940031089,
      "grad_norm": 0.7167975874742553,
      "learning_rate": 9.103448780146609e-06,
      "loss": 0.0047,
      "step": 16760
    },
    {
      "epoch": 0.27440071995418475,
      "grad_norm": 0.32556554865496257,
      "learning_rate": 9.10181635938199e-06,
      "loss": 0.0063,
      "step": 16770
    },
    {
      "epoch": 0.2745643459052606,
      "grad_norm": 0.20405694563286297,
      "learning_rate": 9.100182600472324e-06,
      "loss": 0.0048,
      "step": 16780
    },
    {
      "epoch": 0.2747279718563364,
      "grad_norm": 0.37260164160194664,
      "learning_rate": 9.098547503950595e-06,
      "loss": 0.0059,
      "step": 16790
    },
    {
      "epoch": 0.27489159780741224,
      "grad_norm": 0.29291779516377503,
      "learning_rate": 9.096911070350224e-06,
      "loss": 0.0044,
      "step": 16800
    },
    {
      "epoch": 0.2750552237584881,
      "grad_norm": 0.27726793544459266,
      "learning_rate": 9.09527330020507e-06,
      "loss": 0.0032,
      "step": 16810
    },
    {
      "epoch": 0.27521884970956395,
      "grad_norm": 0.14699445694045887,
      "learning_rate": 9.093634194049426e-06,
      "loss": 0.0041,
      "step": 16820
    },
    {
      "epoch": 0.2753824756606398,
      "grad_norm": 0.33292535913205984,
      "learning_rate": 9.091993752418021e-06,
      "loss": 0.0046,
      "step": 16830
    },
    {
      "epoch": 0.2755461016117156,
      "grad_norm": 0.37292553144646545,
      "learning_rate": 9.09035197584602e-06,
      "loss": 0.0076,
      "step": 16840
    },
    {
      "epoch": 0.27570972756279144,
      "grad_norm": 0.14270971549604855,
      "learning_rate": 9.088708864869022e-06,
      "loss": 0.0029,
      "step": 16850
    },
    {
      "epoch": 0.2758733535138673,
      "grad_norm": 0.3723486209471581,
      "learning_rate": 9.087064420023065e-06,
      "loss": 0.0089,
      "step": 16860
    },
    {
      "epoch": 0.27603697946494316,
      "grad_norm": 0.18615774717755265,
      "learning_rate": 9.085418641844622e-06,
      "loss": 0.0055,
      "step": 16870
    },
    {
      "epoch": 0.276200605416019,
      "grad_norm": 0.2707819121469773,
      "learning_rate": 9.083771530870595e-06,
      "loss": 0.0065,
      "step": 16880
    },
    {
      "epoch": 0.2763642313670948,
      "grad_norm": 0.4581421950572611,
      "learning_rate": 9.082123087638325e-06,
      "loss": 0.0036,
      "step": 16890
    },
    {
      "epoch": 0.27652785731817064,
      "grad_norm": 0.4380214449614211,
      "learning_rate": 9.080473312685592e-06,
      "loss": 0.0088,
      "step": 16900
    },
    {
      "epoch": 0.27669148326924653,
      "grad_norm": 0.42288449154578683,
      "learning_rate": 9.0788222065506e-06,
      "loss": 0.0061,
      "step": 16910
    },
    {
      "epoch": 0.27685510922032236,
      "grad_norm": 0.39832504292257903,
      "learning_rate": 9.077169769771998e-06,
      "loss": 0.0076,
      "step": 16920
    },
    {
      "epoch": 0.2770187351713982,
      "grad_norm": 0.33699388095297433,
      "learning_rate": 9.075516002888862e-06,
      "loss": 0.0042,
      "step": 16930
    },
    {
      "epoch": 0.277182361122474,
      "grad_norm": 0.24703033693729362,
      "learning_rate": 9.073860906440703e-06,
      "loss": 0.0066,
      "step": 16940
    },
    {
      "epoch": 0.27734598707354985,
      "grad_norm": 0.2996451015044408,
      "learning_rate": 9.07220448096747e-06,
      "loss": 0.0068,
      "step": 16950
    },
    {
      "epoch": 0.2775096130246257,
      "grad_norm": 0.35672567459573773,
      "learning_rate": 9.070546727009539e-06,
      "loss": 0.0062,
      "step": 16960
    },
    {
      "epoch": 0.27767323897570156,
      "grad_norm": 0.2796483859279378,
      "learning_rate": 9.068887645107726e-06,
      "loss": 0.0056,
      "step": 16970
    },
    {
      "epoch": 0.2778368649267774,
      "grad_norm": 0.023606022774942986,
      "learning_rate": 9.067227235803277e-06,
      "loss": 0.0076,
      "step": 16980
    },
    {
      "epoch": 0.2780004908778532,
      "grad_norm": 0.27584039758113854,
      "learning_rate": 9.065565499637869e-06,
      "loss": 0.0047,
      "step": 16990
    },
    {
      "epoch": 0.27816411682892905,
      "grad_norm": 0.4360064860040226,
      "learning_rate": 9.063902437153616e-06,
      "loss": 0.004,
      "step": 17000
    },
    {
      "epoch": 0.2783277427800049,
      "grad_norm": 0.21188956130508307,
      "learning_rate": 9.062238048893061e-06,
      "loss": 0.0071,
      "step": 17010
    },
    {
      "epoch": 0.27849136873108077,
      "grad_norm": 0.19395505055327497,
      "learning_rate": 9.060572335399183e-06,
      "loss": 0.0093,
      "step": 17020
    },
    {
      "epoch": 0.2786549946821566,
      "grad_norm": 0.3649470858688348,
      "learning_rate": 9.05890529721539e-06,
      "loss": 0.0033,
      "step": 17030
    },
    {
      "epoch": 0.27881862063323243,
      "grad_norm": 0.033497644451395296,
      "learning_rate": 9.057236934885524e-06,
      "loss": 0.0044,
      "step": 17040
    },
    {
      "epoch": 0.27898224658430826,
      "grad_norm": 0.33870093461295264,
      "learning_rate": 9.05556724895386e-06,
      "loss": 0.0046,
      "step": 17050
    },
    {
      "epoch": 0.2791458725353841,
      "grad_norm": 0.31853956692180563,
      "learning_rate": 9.053896239965104e-06,
      "loss": 0.0062,
      "step": 17060
    },
    {
      "epoch": 0.27930949848646,
      "grad_norm": 0.13327991231249448,
      "learning_rate": 9.05222390846439e-06,
      "loss": 0.0039,
      "step": 17070
    },
    {
      "epoch": 0.2794731244375358,
      "grad_norm": 0.24398948097927978,
      "learning_rate": 9.050550254997288e-06,
      "loss": 0.0057,
      "step": 17080
    },
    {
      "epoch": 0.27963675038861163,
      "grad_norm": 0.19730903980566172,
      "learning_rate": 9.048875280109798e-06,
      "loss": 0.0037,
      "step": 17090
    },
    {
      "epoch": 0.27980037633968746,
      "grad_norm": 0.18928273627153439,
      "learning_rate": 9.047198984348353e-06,
      "loss": 0.0068,
      "step": 17100
    },
    {
      "epoch": 0.2799640022907633,
      "grad_norm": 0.057966889915412885,
      "learning_rate": 9.045521368259813e-06,
      "loss": 0.0048,
      "step": 17110
    },
    {
      "epoch": 0.2801276282418392,
      "grad_norm": 0.39260989032748844,
      "learning_rate": 9.04384243239147e-06,
      "loss": 0.0078,
      "step": 17120
    },
    {
      "epoch": 0.280291254192915,
      "grad_norm": 0.10091180510684265,
      "learning_rate": 9.042162177291047e-06,
      "loss": 0.0044,
      "step": 17130
    },
    {
      "epoch": 0.28045488014399084,
      "grad_norm": 0.2489850493347078,
      "learning_rate": 9.040480603506699e-06,
      "loss": 0.0046,
      "step": 17140
    },
    {
      "epoch": 0.28061850609506667,
      "grad_norm": 0.36502851008245196,
      "learning_rate": 9.038797711587008e-06,
      "loss": 0.01,
      "step": 17150
    },
    {
      "epoch": 0.2807821320461425,
      "grad_norm": 0.28764326182728056,
      "learning_rate": 9.03711350208099e-06,
      "loss": 0.0066,
      "step": 17160
    },
    {
      "epoch": 0.2809457579972184,
      "grad_norm": 0.23908376733862932,
      "learning_rate": 9.035427975538085e-06,
      "loss": 0.0051,
      "step": 17170
    },
    {
      "epoch": 0.2811093839482942,
      "grad_norm": 0.48911992251028985,
      "learning_rate": 9.03374113250817e-06,
      "loss": 0.0058,
      "step": 17180
    },
    {
      "epoch": 0.28127300989937004,
      "grad_norm": 0.41348377410006,
      "learning_rate": 9.032052973541546e-06,
      "loss": 0.0051,
      "step": 17190
    },
    {
      "epoch": 0.28143663585044587,
      "grad_norm": 0.34645790235967133,
      "learning_rate": 9.030363499188945e-06,
      "loss": 0.0077,
      "step": 17200
    },
    {
      "epoch": 0.2816002618015217,
      "grad_norm": 0.4665938835557448,
      "learning_rate": 9.02867271000153e-06,
      "loss": 0.0057,
      "step": 17210
    },
    {
      "epoch": 0.2817638877525976,
      "grad_norm": 0.24234971544515563,
      "learning_rate": 9.026980606530887e-06,
      "loss": 0.0068,
      "step": 17220
    },
    {
      "epoch": 0.2819275137036734,
      "grad_norm": 0.6641114742761444,
      "learning_rate": 9.025287189329037e-06,
      "loss": 0.0046,
      "step": 17230
    },
    {
      "epoch": 0.28209113965474925,
      "grad_norm": 0.0869959670677447,
      "learning_rate": 9.023592458948429e-06,
      "loss": 0.0057,
      "step": 17240
    },
    {
      "epoch": 0.2822547656058251,
      "grad_norm": 0.39168519474225255,
      "learning_rate": 9.021896415941937e-06,
      "loss": 0.0047,
      "step": 17250
    },
    {
      "epoch": 0.2824183915569009,
      "grad_norm": 0.19561469148036173,
      "learning_rate": 9.020199060862866e-06,
      "loss": 0.0055,
      "step": 17260
    },
    {
      "epoch": 0.2825820175079768,
      "grad_norm": 0.1374125961497764,
      "learning_rate": 9.018500394264948e-06,
      "loss": 0.0058,
      "step": 17270
    },
    {
      "epoch": 0.2827456434590526,
      "grad_norm": 0.17806818971856353,
      "learning_rate": 9.01680041670234e-06,
      "loss": 0.0046,
      "step": 17280
    },
    {
      "epoch": 0.28290926941012845,
      "grad_norm": 0.1314765877410827,
      "learning_rate": 9.015099128729635e-06,
      "loss": 0.006,
      "step": 17290
    },
    {
      "epoch": 0.2830728953612043,
      "grad_norm": 0.17452536820361048,
      "learning_rate": 9.013396530901843e-06,
      "loss": 0.0043,
      "step": 17300
    },
    {
      "epoch": 0.2832365213122801,
      "grad_norm": 0.20534089478193973,
      "learning_rate": 9.01169262377441e-06,
      "loss": 0.0036,
      "step": 17310
    },
    {
      "epoch": 0.283400147263356,
      "grad_norm": 0.14910800353288656,
      "learning_rate": 9.009987407903205e-06,
      "loss": 0.0043,
      "step": 17320
    },
    {
      "epoch": 0.2835637732144318,
      "grad_norm": 0.10230165344511472,
      "learning_rate": 9.008280883844523e-06,
      "loss": 0.0044,
      "step": 17330
    },
    {
      "epoch": 0.28372739916550765,
      "grad_norm": 0.3414558865966233,
      "learning_rate": 9.006573052155087e-06,
      "loss": 0.0047,
      "step": 17340
    },
    {
      "epoch": 0.2838910251165835,
      "grad_norm": 0.24668255897968053,
      "learning_rate": 9.004863913392047e-06,
      "loss": 0.0036,
      "step": 17350
    },
    {
      "epoch": 0.2840546510676593,
      "grad_norm": 0.35223751902469175,
      "learning_rate": 9.003153468112982e-06,
      "loss": 0.0044,
      "step": 17360
    },
    {
      "epoch": 0.28421827701873514,
      "grad_norm": 0.40454125751972686,
      "learning_rate": 9.001441716875893e-06,
      "loss": 0.0037,
      "step": 17370
    },
    {
      "epoch": 0.28438190296981103,
      "grad_norm": 0.2614766062667816,
      "learning_rate": 8.999728660239207e-06,
      "loss": 0.0042,
      "step": 17380
    },
    {
      "epoch": 0.28454552892088686,
      "grad_norm": 0.11378411123274325,
      "learning_rate": 8.99801429876178e-06,
      "loss": 0.0061,
      "step": 17390
    },
    {
      "epoch": 0.2847091548719627,
      "grad_norm": 0.12174561884862016,
      "learning_rate": 8.996298633002894e-06,
      "loss": 0.0054,
      "step": 17400
    },
    {
      "epoch": 0.2848727808230385,
      "grad_norm": 0.1211912745790691,
      "learning_rate": 8.994581663522251e-06,
      "loss": 0.0037,
      "step": 17410
    },
    {
      "epoch": 0.28503640677411435,
      "grad_norm": 0.8700669669677659,
      "learning_rate": 8.992863390879983e-06,
      "loss": 0.0061,
      "step": 17420
    },
    {
      "epoch": 0.28520003272519023,
      "grad_norm": 0.46224671200495093,
      "learning_rate": 8.991143815636646e-06,
      "loss": 0.0057,
      "step": 17430
    },
    {
      "epoch": 0.28536365867626606,
      "grad_norm": 0.15800522754695734,
      "learning_rate": 8.989422938353224e-06,
      "loss": 0.0047,
      "step": 17440
    },
    {
      "epoch": 0.2855272846273419,
      "grad_norm": 0.3776115586561121,
      "learning_rate": 8.987700759591117e-06,
      "loss": 0.0067,
      "step": 17450
    },
    {
      "epoch": 0.2856909105784177,
      "grad_norm": 0.18240876229857578,
      "learning_rate": 8.98597727991216e-06,
      "loss": 0.007,
      "step": 17460
    },
    {
      "epoch": 0.28585453652949355,
      "grad_norm": 0.2870768986438059,
      "learning_rate": 8.984252499878605e-06,
      "loss": 0.0051,
      "step": 17470
    },
    {
      "epoch": 0.28601816248056944,
      "grad_norm": 0.29682256498002096,
      "learning_rate": 8.982526420053134e-06,
      "loss": 0.0032,
      "step": 17480
    },
    {
      "epoch": 0.28618178843164527,
      "grad_norm": 0.620859209387552,
      "learning_rate": 8.980799040998849e-06,
      "loss": 0.0045,
      "step": 17490
    },
    {
      "epoch": 0.2863454143827211,
      "grad_norm": 0.14908789786435275,
      "learning_rate": 8.979070363279275e-06,
      "loss": 0.0048,
      "step": 17500
    },
    {
      "epoch": 0.2865090403337969,
      "grad_norm": 0.12589853842307375,
      "learning_rate": 8.977340387458365e-06,
      "loss": 0.0031,
      "step": 17510
    },
    {
      "epoch": 0.28667266628487276,
      "grad_norm": 0.5752820747930553,
      "learning_rate": 8.97560911410049e-06,
      "loss": 0.0082,
      "step": 17520
    },
    {
      "epoch": 0.28683629223594864,
      "grad_norm": 0.2396749134215087,
      "learning_rate": 8.97387654377045e-06,
      "loss": 0.0048,
      "step": 17530
    },
    {
      "epoch": 0.28699991818702447,
      "grad_norm": 0.15013907966603993,
      "learning_rate": 8.972142677033463e-06,
      "loss": 0.0037,
      "step": 17540
    },
    {
      "epoch": 0.2871635441381003,
      "grad_norm": 0.3009543060005969,
      "learning_rate": 8.970407514455177e-06,
      "loss": 0.0052,
      "step": 17550
    },
    {
      "epoch": 0.28732717008917613,
      "grad_norm": 0.5092118824445776,
      "learning_rate": 8.968671056601654e-06,
      "loss": 0.0083,
      "step": 17560
    },
    {
      "epoch": 0.28749079604025196,
      "grad_norm": 0.3151423385490922,
      "learning_rate": 8.966933304039382e-06,
      "loss": 0.0061,
      "step": 17570
    },
    {
      "epoch": 0.28765442199132785,
      "grad_norm": 0.2757763705127762,
      "learning_rate": 8.965194257335278e-06,
      "loss": 0.0065,
      "step": 17580
    },
    {
      "epoch": 0.2878180479424037,
      "grad_norm": 0.31616577434600274,
      "learning_rate": 8.963453917056666e-06,
      "loss": 0.0047,
      "step": 17590
    },
    {
      "epoch": 0.2879816738934795,
      "grad_norm": 0.6927267992672811,
      "learning_rate": 8.961712283771308e-06,
      "loss": 0.0101,
      "step": 17600
    },
    {
      "epoch": 0.28814529984455534,
      "grad_norm": 0.06416194335391504,
      "learning_rate": 8.95996935804738e-06,
      "loss": 0.0043,
      "step": 17610
    },
    {
      "epoch": 0.28830892579563117,
      "grad_norm": 0.27266230434167743,
      "learning_rate": 8.958225140453479e-06,
      "loss": 0.0033,
      "step": 17620
    },
    {
      "epoch": 0.28847255174670705,
      "grad_norm": 0.45188789865407597,
      "learning_rate": 8.956479631558625e-06,
      "loss": 0.0087,
      "step": 17630
    },
    {
      "epoch": 0.2886361776977829,
      "grad_norm": 0.31034738480716423,
      "learning_rate": 8.954732831932262e-06,
      "loss": 0.0049,
      "step": 17640
    },
    {
      "epoch": 0.2887998036488587,
      "grad_norm": 0.3683349327627979,
      "learning_rate": 8.952984742144251e-06,
      "loss": 0.0044,
      "step": 17650
    },
    {
      "epoch": 0.28896342959993454,
      "grad_norm": 0.13618319303300255,
      "learning_rate": 8.951235362764876e-06,
      "loss": 0.0043,
      "step": 17660
    },
    {
      "epoch": 0.28912705555101037,
      "grad_norm": 0.1516569099587365,
      "learning_rate": 8.94948469436484e-06,
      "loss": 0.0094,
      "step": 17670
    },
    {
      "epoch": 0.28929068150208626,
      "grad_norm": 0.28981642376933275,
      "learning_rate": 8.947732737515267e-06,
      "loss": 0.005,
      "step": 17680
    },
    {
      "epoch": 0.2894543074531621,
      "grad_norm": 0.10721092407987674,
      "learning_rate": 8.945979492787704e-06,
      "loss": 0.0044,
      "step": 17690
    },
    {
      "epoch": 0.2896179334042379,
      "grad_norm": 0.09464912754881448,
      "learning_rate": 8.944224960754114e-06,
      "loss": 0.0038,
      "step": 17700
    },
    {
      "epoch": 0.28978155935531374,
      "grad_norm": 0.5280492576363877,
      "learning_rate": 8.942469141986884e-06,
      "loss": 0.0041,
      "step": 17710
    },
    {
      "epoch": 0.2899451853063896,
      "grad_norm": 0.40234931272393704,
      "learning_rate": 8.940712037058818e-06,
      "loss": 0.0046,
      "step": 17720
    },
    {
      "epoch": 0.29010881125746546,
      "grad_norm": 0.2918394805824406,
      "learning_rate": 8.93895364654314e-06,
      "loss": 0.006,
      "step": 17730
    },
    {
      "epoch": 0.2902724372085413,
      "grad_norm": 0.23525890639420918,
      "learning_rate": 8.937193971013494e-06,
      "loss": 0.0062,
      "step": 17740
    },
    {
      "epoch": 0.2904360631596171,
      "grad_norm": 0.19821305982012602,
      "learning_rate": 8.935433011043944e-06,
      "loss": 0.0068,
      "step": 17750
    },
    {
      "epoch": 0.29059968911069295,
      "grad_norm": 0.19061490060542657,
      "learning_rate": 8.93367076720897e-06,
      "loss": 0.0044,
      "step": 17760
    },
    {
      "epoch": 0.2907633150617688,
      "grad_norm": 0.14079871790277276,
      "learning_rate": 8.931907240083475e-06,
      "loss": 0.0059,
      "step": 17770
    },
    {
      "epoch": 0.29092694101284466,
      "grad_norm": 0.2893343061218463,
      "learning_rate": 8.930142430242779e-06,
      "loss": 0.0046,
      "step": 17780
    },
    {
      "epoch": 0.2910905669639205,
      "grad_norm": 0.14988275197951142,
      "learning_rate": 8.928376338262616e-06,
      "loss": 0.0051,
      "step": 17790
    },
    {
      "epoch": 0.2912541929149963,
      "grad_norm": 0.3125567749806149,
      "learning_rate": 8.926608964719146e-06,
      "loss": 0.0034,
      "step": 17800
    },
    {
      "epoch": 0.29141781886607215,
      "grad_norm": 0.16145206350483474,
      "learning_rate": 8.924840310188943e-06,
      "loss": 0.0045,
      "step": 17810
    },
    {
      "epoch": 0.291581444817148,
      "grad_norm": 0.19849447279373644,
      "learning_rate": 8.923070375249e-06,
      "loss": 0.0088,
      "step": 17820
    },
    {
      "epoch": 0.2917450707682238,
      "grad_norm": 0.05015417132354301,
      "learning_rate": 8.921299160476722e-06,
      "loss": 0.0038,
      "step": 17830
    },
    {
      "epoch": 0.2919086967192997,
      "grad_norm": 0.0923168847667189,
      "learning_rate": 8.919526666449942e-06,
      "loss": 0.0035,
      "step": 17840
    },
    {
      "epoch": 0.29207232267037553,
      "grad_norm": 0.3296136564968153,
      "learning_rate": 8.917752893746902e-06,
      "loss": 0.0042,
      "step": 17850
    },
    {
      "epoch": 0.29223594862145136,
      "grad_norm": 0.33119443016129474,
      "learning_rate": 8.915977842946265e-06,
      "loss": 0.0066,
      "step": 17860
    },
    {
      "epoch": 0.2923995745725272,
      "grad_norm": 0.3831678513747912,
      "learning_rate": 8.91420151462711e-06,
      "loss": 0.0039,
      "step": 17870
    },
    {
      "epoch": 0.292563200523603,
      "grad_norm": 0.4927639726226381,
      "learning_rate": 8.912423909368932e-06,
      "loss": 0.0049,
      "step": 17880
    },
    {
      "epoch": 0.2927268264746789,
      "grad_norm": 0.1330237465558743,
      "learning_rate": 8.910645027751646e-06,
      "loss": 0.0052,
      "step": 17890
    },
    {
      "epoch": 0.29289045242575473,
      "grad_norm": 0.2635275296728759,
      "learning_rate": 8.908864870355576e-06,
      "loss": 0.0049,
      "step": 17900
    },
    {
      "epoch": 0.29305407837683056,
      "grad_norm": 0.601864934306598,
      "learning_rate": 8.90708343776147e-06,
      "loss": 0.0077,
      "step": 17910
    },
    {
      "epoch": 0.2932177043279064,
      "grad_norm": 0.27432386746488,
      "learning_rate": 8.905300730550488e-06,
      "loss": 0.0049,
      "step": 17920
    },
    {
      "epoch": 0.2933813302789822,
      "grad_norm": 0.27032056982719216,
      "learning_rate": 8.903516749304206e-06,
      "loss": 0.004,
      "step": 17930
    },
    {
      "epoch": 0.2935449562300581,
      "grad_norm": 0.21771054757089187,
      "learning_rate": 8.90173149460462e-06,
      "loss": 0.0051,
      "step": 17940
    },
    {
      "epoch": 0.29370858218113394,
      "grad_norm": 0.1315058218148815,
      "learning_rate": 8.899944967034132e-06,
      "loss": 0.0052,
      "step": 17950
    },
    {
      "epoch": 0.29387220813220977,
      "grad_norm": 0.13910142824726318,
      "learning_rate": 8.89815716717557e-06,
      "loss": 0.0029,
      "step": 17960
    },
    {
      "epoch": 0.2940358340832856,
      "grad_norm": 0.01675202395006825,
      "learning_rate": 8.896368095612169e-06,
      "loss": 0.0082,
      "step": 17970
    },
    {
      "epoch": 0.2941994600343614,
      "grad_norm": 0.23263561522251786,
      "learning_rate": 8.894577752927582e-06,
      "loss": 0.0055,
      "step": 17980
    },
    {
      "epoch": 0.2943630859854373,
      "grad_norm": 0.22604968974701078,
      "learning_rate": 8.89278613970588e-06,
      "loss": 0.0069,
      "step": 17990
    },
    {
      "epoch": 0.29452671193651314,
      "grad_norm": 0.1791988325137757,
      "learning_rate": 8.890993256531542e-06,
      "loss": 0.0042,
      "step": 18000
    },
    {
      "epoch": 0.29469033788758897,
      "grad_norm": 0.11094865409149303,
      "learning_rate": 8.889199103989464e-06,
      "loss": 0.0042,
      "step": 18010
    },
    {
      "epoch": 0.2948539638386648,
      "grad_norm": 0.36740796795734953,
      "learning_rate": 8.88740368266496e-06,
      "loss": 0.0048,
      "step": 18020
    },
    {
      "epoch": 0.29501758978974063,
      "grad_norm": 0.28781287478582895,
      "learning_rate": 8.885606993143749e-06,
      "loss": 0.0075,
      "step": 18030
    },
    {
      "epoch": 0.2951812157408165,
      "grad_norm": 0.31541973982435617,
      "learning_rate": 8.883809036011976e-06,
      "loss": 0.0034,
      "step": 18040
    },
    {
      "epoch": 0.29534484169189235,
      "grad_norm": 0.32578378859721946,
      "learning_rate": 8.882009811856188e-06,
      "loss": 0.0041,
      "step": 18050
    },
    {
      "epoch": 0.2955084676429682,
      "grad_norm": 0.11586893555456125,
      "learning_rate": 8.88020932126335e-06,
      "loss": 0.0061,
      "step": 18060
    },
    {
      "epoch": 0.295672093594044,
      "grad_norm": 0.02074561200049313,
      "learning_rate": 8.878407564820843e-06,
      "loss": 0.0082,
      "step": 18070
    },
    {
      "epoch": 0.29583571954511984,
      "grad_norm": 0.38095183250575,
      "learning_rate": 8.876604543116455e-06,
      "loss": 0.0053,
      "step": 18080
    },
    {
      "epoch": 0.2959993454961957,
      "grad_norm": 0.36704127972134565,
      "learning_rate": 8.874800256738392e-06,
      "loss": 0.0061,
      "step": 18090
    },
    {
      "epoch": 0.29616297144727155,
      "grad_norm": 0.49207222032707215,
      "learning_rate": 8.87299470627527e-06,
      "loss": 0.0064,
      "step": 18100
    },
    {
      "epoch": 0.2963265973983474,
      "grad_norm": 0.29803018063380854,
      "learning_rate": 8.871187892316118e-06,
      "loss": 0.0066,
      "step": 18110
    },
    {
      "epoch": 0.2964902233494232,
      "grad_norm": 0.3317739189031181,
      "learning_rate": 8.869379815450378e-06,
      "loss": 0.0077,
      "step": 18120
    },
    {
      "epoch": 0.29665384930049904,
      "grad_norm": 0.8672938953239664,
      "learning_rate": 8.867570476267901e-06,
      "loss": 0.0047,
      "step": 18130
    },
    {
      "epoch": 0.2968174752515749,
      "grad_norm": 0.2585070653441631,
      "learning_rate": 8.865759875358954e-06,
      "loss": 0.0071,
      "step": 18140
    },
    {
      "epoch": 0.29698110120265075,
      "grad_norm": 0.24712967476682376,
      "learning_rate": 8.863948013314212e-06,
      "loss": 0.007,
      "step": 18150
    },
    {
      "epoch": 0.2971447271537266,
      "grad_norm": 0.20797324521727528,
      "learning_rate": 8.862134890724766e-06,
      "loss": 0.0061,
      "step": 18160
    },
    {
      "epoch": 0.2973083531048024,
      "grad_norm": 0.2654532173251771,
      "learning_rate": 8.86032050818211e-06,
      "loss": 0.0054,
      "step": 18170
    },
    {
      "epoch": 0.29747197905587824,
      "grad_norm": 0.3023944442755151,
      "learning_rate": 8.858504866278157e-06,
      "loss": 0.0059,
      "step": 18180
    },
    {
      "epoch": 0.29763560500695413,
      "grad_norm": 0.38632815570910545,
      "learning_rate": 8.85668796560523e-06,
      "loss": 0.0062,
      "step": 18190
    },
    {
      "epoch": 0.29779923095802996,
      "grad_norm": 0.1995985927407976,
      "learning_rate": 8.854869806756055e-06,
      "loss": 0.0051,
      "step": 18200
    },
    {
      "epoch": 0.2979628569091058,
      "grad_norm": 0.3106347940007905,
      "learning_rate": 8.853050390323782e-06,
      "loss": 0.0042,
      "step": 18210
    },
    {
      "epoch": 0.2981264828601816,
      "grad_norm": 0.2624950787994254,
      "learning_rate": 8.851229716901956e-06,
      "loss": 0.004,
      "step": 18220
    },
    {
      "epoch": 0.29829010881125745,
      "grad_norm": 0.14497805549842754,
      "learning_rate": 8.849407787084544e-06,
      "loss": 0.004,
      "step": 18230
    },
    {
      "epoch": 0.29845373476233333,
      "grad_norm": 0.24301509614623562,
      "learning_rate": 8.847584601465917e-06,
      "loss": 0.0054,
      "step": 18240
    },
    {
      "epoch": 0.29861736071340916,
      "grad_norm": 0.14482218962449522,
      "learning_rate": 8.845760160640859e-06,
      "loss": 0.0042,
      "step": 18250
    },
    {
      "epoch": 0.298780986664485,
      "grad_norm": 0.21618373291120715,
      "learning_rate": 8.843934465204559e-06,
      "loss": 0.0051,
      "step": 18260
    },
    {
      "epoch": 0.2989446126155608,
      "grad_norm": 0.3438290998089132,
      "learning_rate": 8.842107515752617e-06,
      "loss": 0.0027,
      "step": 18270
    },
    {
      "epoch": 0.29910823856663665,
      "grad_norm": 0.35272352998246265,
      "learning_rate": 8.840279312881047e-06,
      "loss": 0.0043,
      "step": 18280
    },
    {
      "epoch": 0.2992718645177125,
      "grad_norm": 0.438358446710203,
      "learning_rate": 8.838449857186264e-06,
      "loss": 0.0055,
      "step": 18290
    },
    {
      "epoch": 0.29943549046878837,
      "grad_norm": 0.3960303052744374,
      "learning_rate": 8.836619149265098e-06,
      "loss": 0.005,
      "step": 18300
    },
    {
      "epoch": 0.2995991164198642,
      "grad_norm": 0.152214604989409,
      "learning_rate": 8.834787189714782e-06,
      "loss": 0.0054,
      "step": 18310
    },
    {
      "epoch": 0.29976274237094,
      "grad_norm": 0.41348139589715827,
      "learning_rate": 8.832953979132965e-06,
      "loss": 0.0069,
      "step": 18320
    },
    {
      "epoch": 0.29992636832201586,
      "grad_norm": 0.23746624190110746,
      "learning_rate": 8.831119518117697e-06,
      "loss": 0.0057,
      "step": 18330
    },
    {
      "epoch": 0.3000899942730917,
      "grad_norm": 0.23850571342638413,
      "learning_rate": 8.829283807267438e-06,
      "loss": 0.0055,
      "step": 18340
    },
    {
      "epoch": 0.30025362022416757,
      "grad_norm": 0.3862587162446117,
      "learning_rate": 8.827446847181056e-06,
      "loss": 0.0036,
      "step": 18350
    },
    {
      "epoch": 0.3004172461752434,
      "grad_norm": 0.17553908364711773,
      "learning_rate": 8.825608638457827e-06,
      "loss": 0.0064,
      "step": 18360
    },
    {
      "epoch": 0.30058087212631923,
      "grad_norm": 0.11614959115650703,
      "learning_rate": 8.823769181697438e-06,
      "loss": 0.005,
      "step": 18370
    },
    {
      "epoch": 0.30074449807739506,
      "grad_norm": 0.23593007854225115,
      "learning_rate": 8.821928477499971e-06,
      "loss": 0.0054,
      "step": 18380
    },
    {
      "epoch": 0.3009081240284709,
      "grad_norm": 0.34429547443834035,
      "learning_rate": 8.820086526465931e-06,
      "loss": 0.0054,
      "step": 18390
    },
    {
      "epoch": 0.3010717499795468,
      "grad_norm": 0.23218797767641983,
      "learning_rate": 8.818243329196218e-06,
      "loss": 0.0039,
      "step": 18400
    },
    {
      "epoch": 0.3012353759306226,
      "grad_norm": 0.09291590616620059,
      "learning_rate": 8.816398886292144e-06,
      "loss": 0.004,
      "step": 18410
    },
    {
      "epoch": 0.30139900188169844,
      "grad_norm": 0.1762724824173207,
      "learning_rate": 8.814553198355426e-06,
      "loss": 0.0047,
      "step": 18420
    },
    {
      "epoch": 0.30156262783277427,
      "grad_norm": 0.0933406893231049,
      "learning_rate": 8.812706265988184e-06,
      "loss": 0.0077,
      "step": 18430
    },
    {
      "epoch": 0.3017262537838501,
      "grad_norm": 0.16745497530475859,
      "learning_rate": 8.810858089792954e-06,
      "loss": 0.0052,
      "step": 18440
    },
    {
      "epoch": 0.301889879734926,
      "grad_norm": 0.15011709878510496,
      "learning_rate": 8.80900867037266e-06,
      "loss": 0.0044,
      "step": 18450
    },
    {
      "epoch": 0.3020535056860018,
      "grad_norm": 0.41890521116933294,
      "learning_rate": 8.807158008330655e-06,
      "loss": 0.0052,
      "step": 18460
    },
    {
      "epoch": 0.30221713163707764,
      "grad_norm": 0.19639114275751726,
      "learning_rate": 8.805306104270674e-06,
      "loss": 0.0063,
      "step": 18470
    },
    {
      "epoch": 0.30238075758815347,
      "grad_norm": 0.21254538749148008,
      "learning_rate": 8.803452958796873e-06,
      "loss": 0.0037,
      "step": 18480
    },
    {
      "epoch": 0.3025443835392293,
      "grad_norm": 0.3223922441717158,
      "learning_rate": 8.801598572513808e-06,
      "loss": 0.0053,
      "step": 18490
    },
    {
      "epoch": 0.3027080094903052,
      "grad_norm": 0.1489398260998695,
      "learning_rate": 8.79974294602644e-06,
      "loss": 0.0076,
      "step": 18500
    },
    {
      "epoch": 0.302871635441381,
      "grad_norm": 0.20338335397090485,
      "learning_rate": 8.79788607994013e-06,
      "loss": 0.0048,
      "step": 18510
    },
    {
      "epoch": 0.30303526139245685,
      "grad_norm": 0.6816236833588541,
      "learning_rate": 8.796027974860653e-06,
      "loss": 0.0071,
      "step": 18520
    },
    {
      "epoch": 0.3031988873435327,
      "grad_norm": 0.18237243002855794,
      "learning_rate": 8.794168631394182e-06,
      "loss": 0.0043,
      "step": 18530
    },
    {
      "epoch": 0.3033625132946085,
      "grad_norm": 0.34741448150285914,
      "learning_rate": 8.792308050147294e-06,
      "loss": 0.0043,
      "step": 18540
    },
    {
      "epoch": 0.3035261392456844,
      "grad_norm": 1.6548022264429862,
      "learning_rate": 8.790446231726968e-06,
      "loss": 0.0062,
      "step": 18550
    },
    {
      "epoch": 0.3036897651967602,
      "grad_norm": 0.48609516012728515,
      "learning_rate": 8.788583176740593e-06,
      "loss": 0.0059,
      "step": 18560
    },
    {
      "epoch": 0.30385339114783605,
      "grad_norm": 0.7505138221548328,
      "learning_rate": 8.786718885795958e-06,
      "loss": 0.0087,
      "step": 18570
    },
    {
      "epoch": 0.3040170170989119,
      "grad_norm": 0.2441613267717992,
      "learning_rate": 8.784853359501253e-06,
      "loss": 0.005,
      "step": 18580
    },
    {
      "epoch": 0.3041806430499877,
      "grad_norm": 0.18304307885020182,
      "learning_rate": 8.782986598465073e-06,
      "loss": 0.0035,
      "step": 18590
    },
    {
      "epoch": 0.3043442690010636,
      "grad_norm": 0.13016361821385472,
      "learning_rate": 8.781118603296417e-06,
      "loss": 0.0049,
      "step": 18600
    },
    {
      "epoch": 0.3045078949521394,
      "grad_norm": 0.4125757277342999,
      "learning_rate": 8.779249374604684e-06,
      "loss": 0.0053,
      "step": 18610
    },
    {
      "epoch": 0.30467152090321525,
      "grad_norm": 0.3800863376069269,
      "learning_rate": 8.777378912999678e-06,
      "loss": 0.0063,
      "step": 18620
    },
    {
      "epoch": 0.3048351468542911,
      "grad_norm": 0.3328264659703371,
      "learning_rate": 8.775507219091604e-06,
      "loss": 0.007,
      "step": 18630
    },
    {
      "epoch": 0.3049987728053669,
      "grad_norm": 0.4556026664281397,
      "learning_rate": 8.773634293491068e-06,
      "loss": 0.0073,
      "step": 18640
    },
    {
      "epoch": 0.3051623987564428,
      "grad_norm": 0.07479442427530784,
      "learning_rate": 8.77176013680908e-06,
      "loss": 0.0047,
      "step": 18650
    },
    {
      "epoch": 0.30532602470751863,
      "grad_norm": 0.09658526922805287,
      "learning_rate": 8.769884749657049e-06,
      "loss": 0.0072,
      "step": 18660
    },
    {
      "epoch": 0.30548965065859446,
      "grad_norm": 0.14616953519103046,
      "learning_rate": 8.768008132646788e-06,
      "loss": 0.004,
      "step": 18670
    },
    {
      "epoch": 0.3056532766096703,
      "grad_norm": 0.266322465595293,
      "learning_rate": 8.766130286390512e-06,
      "loss": 0.004,
      "step": 18680
    },
    {
      "epoch": 0.3058169025607461,
      "grad_norm": 0.31851643094119897,
      "learning_rate": 8.76425121150083e-06,
      "loss": 0.0049,
      "step": 18690
    },
    {
      "epoch": 0.30598052851182195,
      "grad_norm": 0.12717876830935648,
      "learning_rate": 8.762370908590762e-06,
      "loss": 0.0077,
      "step": 18700
    },
    {
      "epoch": 0.30614415446289783,
      "grad_norm": 0.21868041852126677,
      "learning_rate": 8.760489378273722e-06,
      "loss": 0.0052,
      "step": 18710
    },
    {
      "epoch": 0.30630778041397366,
      "grad_norm": 0.2360320454864102,
      "learning_rate": 8.758606621163524e-06,
      "loss": 0.0047,
      "step": 18720
    },
    {
      "epoch": 0.3064714063650495,
      "grad_norm": 0.1205666298406094,
      "learning_rate": 8.756722637874386e-06,
      "loss": 0.0078,
      "step": 18730
    },
    {
      "epoch": 0.3066350323161253,
      "grad_norm": 0.12478090418601338,
      "learning_rate": 8.754837429020927e-06,
      "loss": 0.0048,
      "step": 18740
    },
    {
      "epoch": 0.30679865826720115,
      "grad_norm": 0.39898851214392134,
      "learning_rate": 8.752950995218157e-06,
      "loss": 0.0053,
      "step": 18750
    },
    {
      "epoch": 0.30696228421827704,
      "grad_norm": 0.07608921786317842,
      "learning_rate": 8.751063337081493e-06,
      "loss": 0.0047,
      "step": 18760
    },
    {
      "epoch": 0.30712591016935287,
      "grad_norm": 0.14223371610305074,
      "learning_rate": 8.749174455226756e-06,
      "loss": 0.0041,
      "step": 18770
    },
    {
      "epoch": 0.3072895361204287,
      "grad_norm": 0.2635544566552066,
      "learning_rate": 8.747284350270153e-06,
      "loss": 0.0053,
      "step": 18780
    },
    {
      "epoch": 0.3074531620715045,
      "grad_norm": 0.459749435014625,
      "learning_rate": 8.745393022828304e-06,
      "loss": 0.0057,
      "step": 18790
    },
    {
      "epoch": 0.30761678802258036,
      "grad_norm": 0.34504988668142356,
      "learning_rate": 8.743500473518214e-06,
      "loss": 0.0031,
      "step": 18800
    },
    {
      "epoch": 0.30778041397365624,
      "grad_norm": 0.4887865601252787,
      "learning_rate": 8.741606702957299e-06,
      "loss": 0.0073,
      "step": 18810
    },
    {
      "epoch": 0.30794403992473207,
      "grad_norm": 0.3289312426956407,
      "learning_rate": 8.739711711763367e-06,
      "loss": 0.0034,
      "step": 18820
    },
    {
      "epoch": 0.3081076658758079,
      "grad_norm": 0.18539507844966194,
      "learning_rate": 8.737815500554622e-06,
      "loss": 0.0047,
      "step": 18830
    },
    {
      "epoch": 0.30827129182688373,
      "grad_norm": 0.1782173454359978,
      "learning_rate": 8.735918069949676e-06,
      "loss": 0.0035,
      "step": 18840
    },
    {
      "epoch": 0.30843491777795956,
      "grad_norm": 0.1774023733353055,
      "learning_rate": 8.734019420567527e-06,
      "loss": 0.0047,
      "step": 18850
    },
    {
      "epoch": 0.30859854372903545,
      "grad_norm": 0.09002331746051315,
      "learning_rate": 8.732119553027577e-06,
      "loss": 0.0043,
      "step": 18860
    },
    {
      "epoch": 0.3087621696801113,
      "grad_norm": 0.34016142042348196,
      "learning_rate": 8.730218467949622e-06,
      "loss": 0.0064,
      "step": 18870
    },
    {
      "epoch": 0.3089257956311871,
      "grad_norm": 0.008464680878899428,
      "learning_rate": 8.728316165953862e-06,
      "loss": 0.0023,
      "step": 18880
    },
    {
      "epoch": 0.30908942158226294,
      "grad_norm": 0.23471549955090934,
      "learning_rate": 8.72641264766089e-06,
      "loss": 0.007,
      "step": 18890
    },
    {
      "epoch": 0.30925304753333877,
      "grad_norm": 0.29956059071569174,
      "learning_rate": 8.724507913691689e-06,
      "loss": 0.0064,
      "step": 18900
    },
    {
      "epoch": 0.30941667348441465,
      "grad_norm": 0.3164203407022074,
      "learning_rate": 8.722601964667648e-06,
      "loss": 0.0058,
      "step": 18910
    },
    {
      "epoch": 0.3095802994354905,
      "grad_norm": 0.29351027790549017,
      "learning_rate": 8.72069480121055e-06,
      "loss": 0.0042,
      "step": 18920
    },
    {
      "epoch": 0.3097439253865663,
      "grad_norm": 0.15280889313344445,
      "learning_rate": 8.71878642394257e-06,
      "loss": 0.0034,
      "step": 18930
    },
    {
      "epoch": 0.30990755133764214,
      "grad_norm": 0.04776895143099245,
      "learning_rate": 8.716876833486289e-06,
      "loss": 0.005,
      "step": 18940
    },
    {
      "epoch": 0.31007117728871797,
      "grad_norm": 0.257234757952196,
      "learning_rate": 8.714966030464672e-06,
      "loss": 0.0049,
      "step": 18950
    },
    {
      "epoch": 0.31023480323979385,
      "grad_norm": 0.5497336065675381,
      "learning_rate": 8.713054015501085e-06,
      "loss": 0.0066,
      "step": 18960
    },
    {
      "epoch": 0.3103984291908697,
      "grad_norm": 0.19806610101213112,
      "learning_rate": 8.711140789219291e-06,
      "loss": 0.0046,
      "step": 18970
    },
    {
      "epoch": 0.3105620551419455,
      "grad_norm": 0.3960724449055792,
      "learning_rate": 8.709226352243443e-06,
      "loss": 0.0057,
      "step": 18980
    },
    {
      "epoch": 0.31072568109302134,
      "grad_norm": 0.2678968686893664,
      "learning_rate": 8.707310705198095e-06,
      "loss": 0.0054,
      "step": 18990
    },
    {
      "epoch": 0.3108893070440972,
      "grad_norm": 0.11846703447912352,
      "learning_rate": 8.705393848708193e-06,
      "loss": 0.0045,
      "step": 19000
    },
    {
      "epoch": 0.31105293299517306,
      "grad_norm": 0.08551642425816144,
      "learning_rate": 8.703475783399076e-06,
      "loss": 0.0055,
      "step": 19010
    },
    {
      "epoch": 0.3112165589462489,
      "grad_norm": 0.3782112803704034,
      "learning_rate": 8.70155650989648e-06,
      "loss": 0.0056,
      "step": 19020
    },
    {
      "epoch": 0.3113801848973247,
      "grad_norm": 0.4696700330745282,
      "learning_rate": 8.699636028826534e-06,
      "loss": 0.0081,
      "step": 19030
    },
    {
      "epoch": 0.31154381084840055,
      "grad_norm": 0.23654459974427444,
      "learning_rate": 8.697714340815761e-06,
      "loss": 0.0038,
      "step": 19040
    },
    {
      "epoch": 0.3117074367994764,
      "grad_norm": 0.3455276061024164,
      "learning_rate": 8.695791446491074e-06,
      "loss": 0.0051,
      "step": 19050
    },
    {
      "epoch": 0.31187106275055226,
      "grad_norm": 0.15878385427065572,
      "learning_rate": 8.693867346479792e-06,
      "loss": 0.004,
      "step": 19060
    },
    {
      "epoch": 0.3120346887016281,
      "grad_norm": 0.5245370086519782,
      "learning_rate": 8.69194204140961e-06,
      "loss": 0.0065,
      "step": 19070
    },
    {
      "epoch": 0.3121983146527039,
      "grad_norm": 0.20725994649395352,
      "learning_rate": 8.69001553190863e-06,
      "loss": 0.0086,
      "step": 19080
    },
    {
      "epoch": 0.31236194060377975,
      "grad_norm": 0.220904064382517,
      "learning_rate": 8.688087818605338e-06,
      "loss": 0.0047,
      "step": 19090
    },
    {
      "epoch": 0.3125255665548556,
      "grad_norm": 0.5675570489871355,
      "learning_rate": 8.686158902128617e-06,
      "loss": 0.0032,
      "step": 19100
    },
    {
      "epoch": 0.31268919250593147,
      "grad_norm": 0.3071925490662874,
      "learning_rate": 8.684228783107744e-06,
      "loss": 0.004,
      "step": 19110
    },
    {
      "epoch": 0.3128528184570073,
      "grad_norm": 0.34065271841656136,
      "learning_rate": 8.682297462172386e-06,
      "loss": 0.0063,
      "step": 19120
    },
    {
      "epoch": 0.3130164444080831,
      "grad_norm": 0.2580444947796707,
      "learning_rate": 8.680364939952601e-06,
      "loss": 0.005,
      "step": 19130
    },
    {
      "epoch": 0.31318007035915896,
      "grad_norm": 0.21422867871123977,
      "learning_rate": 8.67843121707884e-06,
      "loss": 0.0037,
      "step": 19140
    },
    {
      "epoch": 0.3133436963102348,
      "grad_norm": 0.15631003055279277,
      "learning_rate": 8.676496294181947e-06,
      "loss": 0.0053,
      "step": 19150
    },
    {
      "epoch": 0.3135073222613106,
      "grad_norm": 0.21964014531629195,
      "learning_rate": 8.674560171893157e-06,
      "loss": 0.0049,
      "step": 19160
    },
    {
      "epoch": 0.3136709482123865,
      "grad_norm": 0.6720113623825645,
      "learning_rate": 8.672622850844093e-06,
      "loss": 0.0057,
      "step": 19170
    },
    {
      "epoch": 0.31383457416346233,
      "grad_norm": 0.21757784078471265,
      "learning_rate": 8.670684331666775e-06,
      "loss": 0.0048,
      "step": 19180
    },
    {
      "epoch": 0.31399820011453816,
      "grad_norm": 0.2986299670736392,
      "learning_rate": 8.668744614993607e-06,
      "loss": 0.0053,
      "step": 19190
    },
    {
      "epoch": 0.314161826065614,
      "grad_norm": 0.22721085839857583,
      "learning_rate": 8.666803701457389e-06,
      "loss": 0.0059,
      "step": 19200
    },
    {
      "epoch": 0.3143254520166898,
      "grad_norm": 0.402952908259772,
      "learning_rate": 8.66486159169131e-06,
      "loss": 0.0044,
      "step": 19210
    },
    {
      "epoch": 0.3144890779677657,
      "grad_norm": 0.39865063821389984,
      "learning_rate": 8.66291828632895e-06,
      "loss": 0.0067,
      "step": 19220
    },
    {
      "epoch": 0.31465270391884154,
      "grad_norm": 0.09967959660038472,
      "learning_rate": 8.660973786004276e-06,
      "loss": 0.0042,
      "step": 19230
    },
    {
      "epoch": 0.31481632986991737,
      "grad_norm": 0.272280700635327,
      "learning_rate": 8.659028091351647e-06,
      "loss": 0.004,
      "step": 19240
    },
    {
      "epoch": 0.3149799558209932,
      "grad_norm": 0.3624695830039843,
      "learning_rate": 8.657081203005813e-06,
      "loss": 0.005,
      "step": 19250
    },
    {
      "epoch": 0.315143581772069,
      "grad_norm": 0.268045555804144,
      "learning_rate": 8.655133121601911e-06,
      "loss": 0.0057,
      "step": 19260
    },
    {
      "epoch": 0.3153072077231449,
      "grad_norm": 0.2641219001949664,
      "learning_rate": 8.653183847775469e-06,
      "loss": 0.0051,
      "step": 19270
    },
    {
      "epoch": 0.31547083367422074,
      "grad_norm": 0.14383148069089588,
      "learning_rate": 8.651233382162402e-06,
      "loss": 0.0043,
      "step": 19280
    },
    {
      "epoch": 0.31563445962529657,
      "grad_norm": 0.31020297516258566,
      "learning_rate": 8.649281725399014e-06,
      "loss": 0.0035,
      "step": 19290
    },
    {
      "epoch": 0.3157980855763724,
      "grad_norm": 0.2045947405185043,
      "learning_rate": 8.647328878122e-06,
      "loss": 0.0051,
      "step": 19300
    },
    {
      "epoch": 0.31596171152744823,
      "grad_norm": 0.4171154923085513,
      "learning_rate": 8.645374840968443e-06,
      "loss": 0.0065,
      "step": 19310
    },
    {
      "epoch": 0.3161253374785241,
      "grad_norm": 0.24596706126786663,
      "learning_rate": 8.64341961457581e-06,
      "loss": 0.0035,
      "step": 19320
    },
    {
      "epoch": 0.31628896342959995,
      "grad_norm": 0.2999190922434427,
      "learning_rate": 8.641463199581961e-06,
      "loss": 0.0042,
      "step": 19330
    },
    {
      "epoch": 0.3164525893806758,
      "grad_norm": 0.16358232359248806,
      "learning_rate": 8.639505596625142e-06,
      "loss": 0.0062,
      "step": 19340
    },
    {
      "epoch": 0.3166162153317516,
      "grad_norm": 0.30403105146249765,
      "learning_rate": 8.637546806343984e-06,
      "loss": 0.0057,
      "step": 19350
    },
    {
      "epoch": 0.31677984128282743,
      "grad_norm": 0.27359087349567596,
      "learning_rate": 8.635586829377512e-06,
      "loss": 0.0047,
      "step": 19360
    },
    {
      "epoch": 0.3169434672339033,
      "grad_norm": 0.32371888070260113,
      "learning_rate": 8.63362566636513e-06,
      "loss": 0.0058,
      "step": 19370
    },
    {
      "epoch": 0.31710709318497915,
      "grad_norm": 0.493853117321146,
      "learning_rate": 8.631663317946632e-06,
      "loss": 0.0079,
      "step": 19380
    },
    {
      "epoch": 0.317270719136055,
      "grad_norm": 0.6479730582966409,
      "learning_rate": 8.629699784762206e-06,
      "loss": 0.006,
      "step": 19390
    },
    {
      "epoch": 0.3174343450871308,
      "grad_norm": 0.5048300821134012,
      "learning_rate": 8.627735067452411e-06,
      "loss": 0.0046,
      "step": 19400
    },
    {
      "epoch": 0.31759797103820664,
      "grad_norm": 0.40626100384568736,
      "learning_rate": 8.62576916665821e-06,
      "loss": 0.0045,
      "step": 19410
    },
    {
      "epoch": 0.3177615969892825,
      "grad_norm": 0.5322701370781279,
      "learning_rate": 8.623802083020936e-06,
      "loss": 0.005,
      "step": 19420
    },
    {
      "epoch": 0.31792522294035835,
      "grad_norm": 0.19070525098771587,
      "learning_rate": 8.621833817182318e-06,
      "loss": 0.0081,
      "step": 19430
    },
    {
      "epoch": 0.3180888488914342,
      "grad_norm": 0.3130745824220591,
      "learning_rate": 8.619864369784471e-06,
      "loss": 0.0057,
      "step": 19440
    },
    {
      "epoch": 0.31825247484251,
      "grad_norm": 0.3193189382039648,
      "learning_rate": 8.61789374146989e-06,
      "loss": 0.0057,
      "step": 19450
    },
    {
      "epoch": 0.31841610079358584,
      "grad_norm": 0.11979941934246198,
      "learning_rate": 8.615921932881455e-06,
      "loss": 0.0052,
      "step": 19460
    },
    {
      "epoch": 0.31857972674466173,
      "grad_norm": 0.6702571861452684,
      "learning_rate": 8.613948944662436e-06,
      "loss": 0.0058,
      "step": 19470
    },
    {
      "epoch": 0.31874335269573756,
      "grad_norm": 0.4302687150938046,
      "learning_rate": 8.611974777456487e-06,
      "loss": 0.0033,
      "step": 19480
    },
    {
      "epoch": 0.3189069786468134,
      "grad_norm": 0.17134553120860474,
      "learning_rate": 8.609999431907642e-06,
      "loss": 0.0063,
      "step": 19490
    },
    {
      "epoch": 0.3190706045978892,
      "grad_norm": 0.20772018201977024,
      "learning_rate": 8.608022908660327e-06,
      "loss": 0.0059,
      "step": 19500
    },
    {
      "epoch": 0.31923423054896505,
      "grad_norm": 0.2390205838824782,
      "learning_rate": 8.606045208359343e-06,
      "loss": 0.0046,
      "step": 19510
    },
    {
      "epoch": 0.31939785650004093,
      "grad_norm": 0.34693019130854325,
      "learning_rate": 8.604066331649882e-06,
      "loss": 0.0035,
      "step": 19520
    },
    {
      "epoch": 0.31956148245111676,
      "grad_norm": 0.1981509576409543,
      "learning_rate": 8.602086279177518e-06,
      "loss": 0.0052,
      "step": 19530
    },
    {
      "epoch": 0.3197251084021926,
      "grad_norm": 0.5549142472707205,
      "learning_rate": 8.600105051588206e-06,
      "loss": 0.0049,
      "step": 19540
    },
    {
      "epoch": 0.3198887343532684,
      "grad_norm": 0.33226097169147817,
      "learning_rate": 8.59812264952829e-06,
      "loss": 0.0042,
      "step": 19550
    },
    {
      "epoch": 0.32005236030434425,
      "grad_norm": 0.22547435144445466,
      "learning_rate": 8.596139073644487e-06,
      "loss": 0.0033,
      "step": 19560
    },
    {
      "epoch": 0.3202159862554201,
      "grad_norm": 0.06856239155902007,
      "learning_rate": 8.594154324583912e-06,
      "loss": 0.004,
      "step": 19570
    },
    {
      "epoch": 0.32037961220649597,
      "grad_norm": 0.316131174078988,
      "learning_rate": 8.592168402994049e-06,
      "loss": 0.0072,
      "step": 19580
    },
    {
      "epoch": 0.3205432381575718,
      "grad_norm": 0.21167962455686248,
      "learning_rate": 8.590181309522772e-06,
      "loss": 0.0045,
      "step": 19590
    },
    {
      "epoch": 0.3207068641086476,
      "grad_norm": 0.22424023825826722,
      "learning_rate": 8.588193044818332e-06,
      "loss": 0.0062,
      "step": 19600
    },
    {
      "epoch": 0.32087049005972346,
      "grad_norm": 0.19023740462406216,
      "learning_rate": 8.586203609529368e-06,
      "loss": 0.0057,
      "step": 19610
    },
    {
      "epoch": 0.3210341160107993,
      "grad_norm": 0.11670946787342265,
      "learning_rate": 8.5842130043049e-06,
      "loss": 0.0057,
      "step": 19620
    },
    {
      "epoch": 0.32119774196187517,
      "grad_norm": 0.20554451487363826,
      "learning_rate": 8.582221229794323e-06,
      "loss": 0.0071,
      "step": 19630
    },
    {
      "epoch": 0.321361367912951,
      "grad_norm": 0.1120270020156323,
      "learning_rate": 8.580228286647421e-06,
      "loss": 0.0044,
      "step": 19640
    },
    {
      "epoch": 0.32152499386402683,
      "grad_norm": 0.6032586100051742,
      "learning_rate": 8.578234175514354e-06,
      "loss": 0.0067,
      "step": 19650
    },
    {
      "epoch": 0.32168861981510266,
      "grad_norm": 0.4750675502427299,
      "learning_rate": 8.57623889704567e-06,
      "loss": 0.0046,
      "step": 19660
    },
    {
      "epoch": 0.3218522457661785,
      "grad_norm": 0.198568633458702,
      "learning_rate": 8.574242451892289e-06,
      "loss": 0.0047,
      "step": 19670
    },
    {
      "epoch": 0.3220158717172544,
      "grad_norm": 0.33894573873188527,
      "learning_rate": 8.572244840705519e-06,
      "loss": 0.0063,
      "step": 19680
    },
    {
      "epoch": 0.3221794976683302,
      "grad_norm": 0.6828594341380465,
      "learning_rate": 8.570246064137044e-06,
      "loss": 0.0061,
      "step": 19690
    },
    {
      "epoch": 0.32234312361940604,
      "grad_norm": 0.3174630565931213,
      "learning_rate": 8.568246122838928e-06,
      "loss": 0.004,
      "step": 19700
    },
    {
      "epoch": 0.32250674957048187,
      "grad_norm": 0.10652153541117633,
      "learning_rate": 8.56624501746362e-06,
      "loss": 0.0028,
      "step": 19710
    },
    {
      "epoch": 0.3226703755215577,
      "grad_norm": 0.11040291417424263,
      "learning_rate": 8.56424274866394e-06,
      "loss": 0.0045,
      "step": 19720
    },
    {
      "epoch": 0.3228340014726336,
      "grad_norm": 0.16528110684090927,
      "learning_rate": 8.562239317093097e-06,
      "loss": 0.0039,
      "step": 19730
    },
    {
      "epoch": 0.3229976274237094,
      "grad_norm": 0.27694170379314115,
      "learning_rate": 8.560234723404676e-06,
      "loss": 0.0056,
      "step": 19740
    },
    {
      "epoch": 0.32316125337478524,
      "grad_norm": 0.17723179084876442,
      "learning_rate": 8.558228968252636e-06,
      "loss": 0.0056,
      "step": 19750
    },
    {
      "epoch": 0.32332487932586107,
      "grad_norm": 0.16505081513972458,
      "learning_rate": 8.556222052291323e-06,
      "loss": 0.0036,
      "step": 19760
    },
    {
      "epoch": 0.3234885052769369,
      "grad_norm": 0.24193101889973934,
      "learning_rate": 8.554213976175455e-06,
      "loss": 0.0044,
      "step": 19770
    },
    {
      "epoch": 0.3236521312280128,
      "grad_norm": 0.1602473801930467,
      "learning_rate": 8.552204740560135e-06,
      "loss": 0.0025,
      "step": 19780
    },
    {
      "epoch": 0.3238157571790886,
      "grad_norm": 0.40680679208681314,
      "learning_rate": 8.550194346100835e-06,
      "loss": 0.005,
      "step": 19790
    },
    {
      "epoch": 0.32397938313016444,
      "grad_norm": 0.38883432839704213,
      "learning_rate": 8.548182793453417e-06,
      "loss": 0.0066,
      "step": 19800
    },
    {
      "epoch": 0.3241430090812403,
      "grad_norm": 0.7068000441083453,
      "learning_rate": 8.546170083274109e-06,
      "loss": 0.005,
      "step": 19810
    },
    {
      "epoch": 0.3243066350323161,
      "grad_norm": 0.08823285994921172,
      "learning_rate": 8.544156216219525e-06,
      "loss": 0.0047,
      "step": 19820
    },
    {
      "epoch": 0.324470260983392,
      "grad_norm": 0.35512735180526434,
      "learning_rate": 8.542141192946653e-06,
      "loss": 0.005,
      "step": 19830
    },
    {
      "epoch": 0.3246338869344678,
      "grad_norm": 0.4199746661521805,
      "learning_rate": 8.54012501411286e-06,
      "loss": 0.0044,
      "step": 19840
    },
    {
      "epoch": 0.32479751288554365,
      "grad_norm": 0.14131453131454122,
      "learning_rate": 8.538107680375884e-06,
      "loss": 0.0044,
      "step": 19850
    },
    {
      "epoch": 0.3249611388366195,
      "grad_norm": 0.17457541614631383,
      "learning_rate": 8.53608919239385e-06,
      "loss": 0.0061,
      "step": 19860
    },
    {
      "epoch": 0.3251247647876953,
      "grad_norm": 0.29185652303698856,
      "learning_rate": 8.534069550825252e-06,
      "loss": 0.0055,
      "step": 19870
    },
    {
      "epoch": 0.3252883907387712,
      "grad_norm": 0.14926993364488633,
      "learning_rate": 8.532048756328961e-06,
      "loss": 0.0073,
      "step": 19880
    },
    {
      "epoch": 0.325452016689847,
      "grad_norm": 0.1652374687052916,
      "learning_rate": 8.530026809564227e-06,
      "loss": 0.0057,
      "step": 19890
    },
    {
      "epoch": 0.32561564264092285,
      "grad_norm": 0.5973850833255425,
      "learning_rate": 8.528003711190674e-06,
      "loss": 0.0068,
      "step": 19900
    },
    {
      "epoch": 0.3257792685919987,
      "grad_norm": 0.253477523557912,
      "learning_rate": 8.525979461868303e-06,
      "loss": 0.0069,
      "step": 19910
    },
    {
      "epoch": 0.3259428945430745,
      "grad_norm": 0.23226886417651638,
      "learning_rate": 8.523954062257487e-06,
      "loss": 0.0042,
      "step": 19920
    },
    {
      "epoch": 0.3261065204941504,
      "grad_norm": 0.510517293281738,
      "learning_rate": 8.52192751301898e-06,
      "loss": 0.0046,
      "step": 19930
    },
    {
      "epoch": 0.32627014644522623,
      "grad_norm": 0.8339113112366476,
      "learning_rate": 8.519899814813906e-06,
      "loss": 0.0044,
      "step": 19940
    },
    {
      "epoch": 0.32643377239630206,
      "grad_norm": 0.2236550221867777,
      "learning_rate": 8.517870968303764e-06,
      "loss": 0.006,
      "step": 19950
    },
    {
      "epoch": 0.3265973983473779,
      "grad_norm": 0.3328234329723939,
      "learning_rate": 8.515840974150433e-06,
      "loss": 0.0033,
      "step": 19960
    },
    {
      "epoch": 0.3267610242984537,
      "grad_norm": 0.20888639646895898,
      "learning_rate": 8.51380983301616e-06,
      "loss": 0.0057,
      "step": 19970
    },
    {
      "epoch": 0.3269246502495296,
      "grad_norm": 0.2337675852333641,
      "learning_rate": 8.511777545563571e-06,
      "loss": 0.0037,
      "step": 19980
    },
    {
      "epoch": 0.32708827620060543,
      "grad_norm": 0.2739602446724029,
      "learning_rate": 8.509744112455663e-06,
      "loss": 0.0043,
      "step": 19990
    },
    {
      "epoch": 0.32725190215168126,
      "grad_norm": 0.3307902780303389,
      "learning_rate": 8.507709534355803e-06,
      "loss": 0.008,
      "step": 20000
    },
    {
      "epoch": 0.3274155281027571,
      "grad_norm": 0.18152554101054655,
      "learning_rate": 8.505673811927743e-06,
      "loss": 0.0046,
      "step": 20010
    },
    {
      "epoch": 0.3275791540538329,
      "grad_norm": 0.23342479047005812,
      "learning_rate": 8.503636945835601e-06,
      "loss": 0.005,
      "step": 20020
    },
    {
      "epoch": 0.32774278000490875,
      "grad_norm": 0.18481861964914134,
      "learning_rate": 8.501598936743865e-06,
      "loss": 0.0032,
      "step": 20030
    },
    {
      "epoch": 0.32790640595598464,
      "grad_norm": 0.09212070437751925,
      "learning_rate": 8.4995597853174e-06,
      "loss": 0.0058,
      "step": 20040
    },
    {
      "epoch": 0.32807003190706047,
      "grad_norm": 0.320105963998252,
      "learning_rate": 8.497519492221444e-06,
      "loss": 0.0054,
      "step": 20050
    },
    {
      "epoch": 0.3282336578581363,
      "grad_norm": 0.2962106544030525,
      "learning_rate": 8.495478058121605e-06,
      "loss": 0.007,
      "step": 20060
    },
    {
      "epoch": 0.3283972838092121,
      "grad_norm": 0.2129442456594696,
      "learning_rate": 8.493435483683868e-06,
      "loss": 0.0054,
      "step": 20070
    },
    {
      "epoch": 0.32856090976028796,
      "grad_norm": 0.4597597233273598,
      "learning_rate": 8.491391769574585e-06,
      "loss": 0.0027,
      "step": 20080
    },
    {
      "epoch": 0.32872453571136384,
      "grad_norm": 0.6222957106386409,
      "learning_rate": 8.48934691646048e-06,
      "loss": 0.0082,
      "step": 20090
    },
    {
      "epoch": 0.32888816166243967,
      "grad_norm": 0.15574223169742313,
      "learning_rate": 8.487300925008653e-06,
      "loss": 0.0038,
      "step": 20100
    },
    {
      "epoch": 0.3290517876135155,
      "grad_norm": 0.15719296280165462,
      "learning_rate": 8.485253795886572e-06,
      "loss": 0.0043,
      "step": 20110
    },
    {
      "epoch": 0.32921541356459133,
      "grad_norm": 0.1764110861981905,
      "learning_rate": 8.483205529762073e-06,
      "loss": 0.0078,
      "step": 20120
    },
    {
      "epoch": 0.32937903951566716,
      "grad_norm": 0.019104415685411685,
      "learning_rate": 8.48115612730337e-06,
      "loss": 0.007,
      "step": 20130
    },
    {
      "epoch": 0.32954266546674305,
      "grad_norm": 0.48701239465306084,
      "learning_rate": 8.479105589179047e-06,
      "loss": 0.0066,
      "step": 20140
    },
    {
      "epoch": 0.3297062914178189,
      "grad_norm": 0.3089457276742494,
      "learning_rate": 8.47705391605805e-06,
      "loss": 0.0042,
      "step": 20150
    },
    {
      "epoch": 0.3298699173688947,
      "grad_norm": 0.20750493030711897,
      "learning_rate": 8.475001108609706e-06,
      "loss": 0.0057,
      "step": 20160
    },
    {
      "epoch": 0.33003354331997053,
      "grad_norm": 0.4013216626815941,
      "learning_rate": 8.472947167503703e-06,
      "loss": 0.0095,
      "step": 20170
    },
    {
      "epoch": 0.33019716927104636,
      "grad_norm": 0.3473546672034808,
      "learning_rate": 8.470892093410104e-06,
      "loss": 0.0042,
      "step": 20180
    },
    {
      "epoch": 0.33036079522212225,
      "grad_norm": 0.09198941174776062,
      "learning_rate": 8.468835886999343e-06,
      "loss": 0.0052,
      "step": 20190
    },
    {
      "epoch": 0.3305244211731981,
      "grad_norm": 0.407109610862657,
      "learning_rate": 8.466778548942217e-06,
      "loss": 0.005,
      "step": 20200
    },
    {
      "epoch": 0.3306880471242739,
      "grad_norm": 0.2913748401125258,
      "learning_rate": 8.4647200799099e-06,
      "loss": 0.0078,
      "step": 20210
    },
    {
      "epoch": 0.33085167307534974,
      "grad_norm": 0.3484862369288203,
      "learning_rate": 8.46266048057393e-06,
      "loss": 0.0082,
      "step": 20220
    },
    {
      "epoch": 0.33101529902642557,
      "grad_norm": 0.20825568602552483,
      "learning_rate": 8.460599751606213e-06,
      "loss": 0.0039,
      "step": 20230
    },
    {
      "epoch": 0.33117892497750145,
      "grad_norm": 0.23156507017755068,
      "learning_rate": 8.458537893679028e-06,
      "loss": 0.0031,
      "step": 20240
    },
    {
      "epoch": 0.3313425509285773,
      "grad_norm": 0.3992925900372824,
      "learning_rate": 8.456474907465018e-06,
      "loss": 0.0062,
      "step": 20250
    },
    {
      "epoch": 0.3315061768796531,
      "grad_norm": 0.2470117489752518,
      "learning_rate": 8.454410793637197e-06,
      "loss": 0.005,
      "step": 20260
    },
    {
      "epoch": 0.33166980283072894,
      "grad_norm": 0.31271748039786057,
      "learning_rate": 8.452345552868945e-06,
      "loss": 0.003,
      "step": 20270
    },
    {
      "epoch": 0.3318334287818048,
      "grad_norm": 0.15655217538872404,
      "learning_rate": 8.45027918583401e-06,
      "loss": 0.0044,
      "step": 20280
    },
    {
      "epoch": 0.33199705473288066,
      "grad_norm": 0.2243992444036362,
      "learning_rate": 8.448211693206506e-06,
      "loss": 0.0043,
      "step": 20290
    },
    {
      "epoch": 0.3321606806839565,
      "grad_norm": 0.14825698883583144,
      "learning_rate": 8.44614307566092e-06,
      "loss": 0.0062,
      "step": 20300
    },
    {
      "epoch": 0.3323243066350323,
      "grad_norm": 0.31147961259846463,
      "learning_rate": 8.444073333872097e-06,
      "loss": 0.0045,
      "step": 20310
    },
    {
      "epoch": 0.33248793258610815,
      "grad_norm": 0.12602795235252198,
      "learning_rate": 8.44200246851526e-06,
      "loss": 0.0029,
      "step": 20320
    },
    {
      "epoch": 0.332651558537184,
      "grad_norm": 0.28592830211001063,
      "learning_rate": 8.439930480265988e-06,
      "loss": 0.0058,
      "step": 20330
    },
    {
      "epoch": 0.33281518448825986,
      "grad_norm": 0.2798266528724036,
      "learning_rate": 8.43785736980023e-06,
      "loss": 0.0056,
      "step": 20340
    },
    {
      "epoch": 0.3329788104393357,
      "grad_norm": 0.18360385571270163,
      "learning_rate": 8.435783137794302e-06,
      "loss": 0.0037,
      "step": 20350
    },
    {
      "epoch": 0.3331424363904115,
      "grad_norm": 0.2026247069598769,
      "learning_rate": 8.433707784924888e-06,
      "loss": 0.0048,
      "step": 20360
    },
    {
      "epoch": 0.33330606234148735,
      "grad_norm": 0.2030648715923798,
      "learning_rate": 8.431631311869031e-06,
      "loss": 0.004,
      "step": 20370
    },
    {
      "epoch": 0.3334696882925632,
      "grad_norm": 0.44733634321601373,
      "learning_rate": 8.429553719304146e-06,
      "loss": 0.004,
      "step": 20380
    },
    {
      "epoch": 0.33363331424363907,
      "grad_norm": 0.09426474148409458,
      "learning_rate": 8.427475007908013e-06,
      "loss": 0.0053,
      "step": 20390
    },
    {
      "epoch": 0.3337969401947149,
      "grad_norm": 0.18692655710602513,
      "learning_rate": 8.425395178358773e-06,
      "loss": 0.0042,
      "step": 20400
    },
    {
      "epoch": 0.3339605661457907,
      "grad_norm": 0.21142089709592796,
      "learning_rate": 8.423314231334931e-06,
      "loss": 0.0043,
      "step": 20410
    },
    {
      "epoch": 0.33412419209686656,
      "grad_norm": 0.25454891317813116,
      "learning_rate": 8.421232167515362e-06,
      "loss": 0.0074,
      "step": 20420
    },
    {
      "epoch": 0.3342878180479424,
      "grad_norm": 0.22867865250775674,
      "learning_rate": 8.419148987579301e-06,
      "loss": 0.0062,
      "step": 20430
    },
    {
      "epoch": 0.33445144399901827,
      "grad_norm": 0.20283707773724322,
      "learning_rate": 8.417064692206352e-06,
      "loss": 0.004,
      "step": 20440
    },
    {
      "epoch": 0.3346150699500941,
      "grad_norm": 0.35055917738817965,
      "learning_rate": 8.414979282076477e-06,
      "loss": 0.0043,
      "step": 20450
    },
    {
      "epoch": 0.33477869590116993,
      "grad_norm": 0.014783402893192547,
      "learning_rate": 8.412892757870002e-06,
      "loss": 0.0027,
      "step": 20460
    },
    {
      "epoch": 0.33494232185224576,
      "grad_norm": 0.27000256216174645,
      "learning_rate": 8.410805120267623e-06,
      "loss": 0.0035,
      "step": 20470
    },
    {
      "epoch": 0.3351059478033216,
      "grad_norm": 0.3514817262712476,
      "learning_rate": 8.408716369950391e-06,
      "loss": 0.0116,
      "step": 20480
    },
    {
      "epoch": 0.3352695737543974,
      "grad_norm": 0.4262466140975379,
      "learning_rate": 8.406626507599726e-06,
      "loss": 0.0044,
      "step": 20490
    },
    {
      "epoch": 0.3354331997054733,
      "grad_norm": 0.22328168335638882,
      "learning_rate": 8.40453553389741e-06,
      "loss": 0.0053,
      "step": 20500
    },
    {
      "epoch": 0.33559682565654914,
      "grad_norm": 0.3297965358100579,
      "learning_rate": 8.402443449525581e-06,
      "loss": 0.005,
      "step": 20510
    },
    {
      "epoch": 0.33576045160762497,
      "grad_norm": 0.11304855497402304,
      "learning_rate": 8.400350255166751e-06,
      "loss": 0.0051,
      "step": 20520
    },
    {
      "epoch": 0.3359240775587008,
      "grad_norm": 0.1606646568222215,
      "learning_rate": 8.398255951503784e-06,
      "loss": 0.0092,
      "step": 20530
    },
    {
      "epoch": 0.3360877035097766,
      "grad_norm": 0.38636759251863384,
      "learning_rate": 8.396160539219909e-06,
      "loss": 0.0061,
      "step": 20540
    },
    {
      "epoch": 0.3362513294608525,
      "grad_norm": 0.9041239948038072,
      "learning_rate": 8.394064018998718e-06,
      "loss": 0.0051,
      "step": 20550
    },
    {
      "epoch": 0.33641495541192834,
      "grad_norm": 0.2921845595559238,
      "learning_rate": 8.391966391524165e-06,
      "loss": 0.0042,
      "step": 20560
    },
    {
      "epoch": 0.33657858136300417,
      "grad_norm": 0.09000868605860639,
      "learning_rate": 8.389867657480562e-06,
      "loss": 0.0038,
      "step": 20570
    },
    {
      "epoch": 0.33674220731408,
      "grad_norm": 0.18527719679843574,
      "learning_rate": 8.387767817552585e-06,
      "loss": 0.0057,
      "step": 20580
    },
    {
      "epoch": 0.33690583326515583,
      "grad_norm": 0.24264593240171178,
      "learning_rate": 8.385666872425269e-06,
      "loss": 0.0047,
      "step": 20590
    },
    {
      "epoch": 0.3370694592162317,
      "grad_norm": 0.15137008291253395,
      "learning_rate": 8.383564822784012e-06,
      "loss": 0.004,
      "step": 20600
    },
    {
      "epoch": 0.33723308516730754,
      "grad_norm": 0.29583594710949596,
      "learning_rate": 8.381461669314568e-06,
      "loss": 0.0035,
      "step": 20610
    },
    {
      "epoch": 0.3373967111183834,
      "grad_norm": 0.08598990139889533,
      "learning_rate": 8.379357412703053e-06,
      "loss": 0.0034,
      "step": 20620
    },
    {
      "epoch": 0.3375603370694592,
      "grad_norm": 0.2737661261856486,
      "learning_rate": 8.377252053635946e-06,
      "loss": 0.0033,
      "step": 20630
    },
    {
      "epoch": 0.33772396302053503,
      "grad_norm": 0.1277352753469851,
      "learning_rate": 8.375145592800084e-06,
      "loss": 0.0039,
      "step": 20640
    },
    {
      "epoch": 0.3378875889716109,
      "grad_norm": 0.1590029401460674,
      "learning_rate": 8.37303803088266e-06,
      "loss": 0.0054,
      "step": 20650
    },
    {
      "epoch": 0.33805121492268675,
      "grad_norm": 0.21432959612476377,
      "learning_rate": 8.37092936857123e-06,
      "loss": 0.0034,
      "step": 20660
    },
    {
      "epoch": 0.3382148408737626,
      "grad_norm": 0.15784408626267274,
      "learning_rate": 8.368819606553706e-06,
      "loss": 0.0035,
      "step": 20670
    },
    {
      "epoch": 0.3383784668248384,
      "grad_norm": 0.30690303322233475,
      "learning_rate": 8.366708745518364e-06,
      "loss": 0.0063,
      "step": 20680
    },
    {
      "epoch": 0.33854209277591424,
      "grad_norm": 0.1694243694347292,
      "learning_rate": 8.364596786153832e-06,
      "loss": 0.0033,
      "step": 20690
    },
    {
      "epoch": 0.3387057187269901,
      "grad_norm": 0.32450174992590375,
      "learning_rate": 8.3624837291491e-06,
      "loss": 0.0059,
      "step": 20700
    },
    {
      "epoch": 0.33886934467806595,
      "grad_norm": 0.28827172445527954,
      "learning_rate": 8.360369575193517e-06,
      "loss": 0.0046,
      "step": 20710
    },
    {
      "epoch": 0.3390329706291418,
      "grad_norm": 0.0923260024145372,
      "learning_rate": 8.358254324976787e-06,
      "loss": 0.0036,
      "step": 20720
    },
    {
      "epoch": 0.3391965965802176,
      "grad_norm": 0.39116282443741507,
      "learning_rate": 8.356137979188975e-06,
      "loss": 0.0069,
      "step": 20730
    },
    {
      "epoch": 0.33936022253129344,
      "grad_norm": 0.129851264372882,
      "learning_rate": 8.354020538520498e-06,
      "loss": 0.0044,
      "step": 20740
    },
    {
      "epoch": 0.33952384848236933,
      "grad_norm": 0.29309004354245527,
      "learning_rate": 8.351902003662133e-06,
      "loss": 0.004,
      "step": 20750
    },
    {
      "epoch": 0.33968747443344516,
      "grad_norm": 0.15365493840123007,
      "learning_rate": 8.34978237530502e-06,
      "loss": 0.0037,
      "step": 20760
    },
    {
      "epoch": 0.339851100384521,
      "grad_norm": 0.1338658824062015,
      "learning_rate": 8.347661654140646e-06,
      "loss": 0.0038,
      "step": 20770
    },
    {
      "epoch": 0.3400147263355968,
      "grad_norm": 0.1811580203501847,
      "learning_rate": 8.34553984086086e-06,
      "loss": 0.0042,
      "step": 20780
    },
    {
      "epoch": 0.34017835228667265,
      "grad_norm": 0.4052405570666526,
      "learning_rate": 8.343416936157867e-06,
      "loss": 0.0043,
      "step": 20790
    },
    {
      "epoch": 0.34034197823774853,
      "grad_norm": 0.3666725662196123,
      "learning_rate": 8.341292940724224e-06,
      "loss": 0.0063,
      "step": 20800
    },
    {
      "epoch": 0.34050560418882436,
      "grad_norm": 0.13156444635342052,
      "learning_rate": 8.33916785525285e-06,
      "loss": 0.0033,
      "step": 20810
    },
    {
      "epoch": 0.3406692301399002,
      "grad_norm": 0.3843174070954451,
      "learning_rate": 8.337041680437015e-06,
      "loss": 0.0047,
      "step": 20820
    },
    {
      "epoch": 0.340832856090976,
      "grad_norm": 0.10171622277744272,
      "learning_rate": 8.334914416970348e-06,
      "loss": 0.0054,
      "step": 20830
    },
    {
      "epoch": 0.34099648204205185,
      "grad_norm": 0.1730790074187248,
      "learning_rate": 8.332786065546829e-06,
      "loss": 0.0044,
      "step": 20840
    },
    {
      "epoch": 0.34116010799312774,
      "grad_norm": 0.16304753805254601,
      "learning_rate": 8.330656626860794e-06,
      "loss": 0.007,
      "step": 20850
    },
    {
      "epoch": 0.34132373394420357,
      "grad_norm": 0.19238343508409989,
      "learning_rate": 8.328526101606938e-06,
      "loss": 0.0092,
      "step": 20860
    },
    {
      "epoch": 0.3414873598952794,
      "grad_norm": 0.06326117814238459,
      "learning_rate": 8.326394490480306e-06,
      "loss": 0.0043,
      "step": 20870
    },
    {
      "epoch": 0.3416509858463552,
      "grad_norm": 0.1865533200519949,
      "learning_rate": 8.324261794176295e-06,
      "loss": 0.0031,
      "step": 20880
    },
    {
      "epoch": 0.34181461179743106,
      "grad_norm": 0.23144322171024437,
      "learning_rate": 8.322128013390664e-06,
      "loss": 0.0043,
      "step": 20890
    },
    {
      "epoch": 0.3419782377485069,
      "grad_norm": 0.34188691278164557,
      "learning_rate": 8.319993148819519e-06,
      "loss": 0.0115,
      "step": 20900
    },
    {
      "epoch": 0.34214186369958277,
      "grad_norm": 0.5538022011342335,
      "learning_rate": 8.317857201159323e-06,
      "loss": 0.0074,
      "step": 20910
    },
    {
      "epoch": 0.3423054896506586,
      "grad_norm": 0.2117723034746883,
      "learning_rate": 8.315720171106892e-06,
      "loss": 0.0059,
      "step": 20920
    },
    {
      "epoch": 0.34246911560173443,
      "grad_norm": 0.2621165164570185,
      "learning_rate": 8.313582059359392e-06,
      "loss": 0.0054,
      "step": 20930
    },
    {
      "epoch": 0.34263274155281026,
      "grad_norm": 0.26643959795547384,
      "learning_rate": 8.311442866614343e-06,
      "loss": 0.0048,
      "step": 20940
    },
    {
      "epoch": 0.3427963675038861,
      "grad_norm": 0.0915329289616054,
      "learning_rate": 8.309302593569621e-06,
      "loss": 0.0034,
      "step": 20950
    },
    {
      "epoch": 0.342959993454962,
      "grad_norm": 0.13492827383335063,
      "learning_rate": 8.307161240923454e-06,
      "loss": 0.0046,
      "step": 20960
    },
    {
      "epoch": 0.3431236194060378,
      "grad_norm": 0.22177612129928959,
      "learning_rate": 8.305018809374417e-06,
      "loss": 0.0105,
      "step": 20970
    },
    {
      "epoch": 0.34328724535711364,
      "grad_norm": 0.12035781797990085,
      "learning_rate": 8.302875299621442e-06,
      "loss": 0.0046,
      "step": 20980
    },
    {
      "epoch": 0.34345087130818946,
      "grad_norm": 0.1513481217164888,
      "learning_rate": 8.300730712363811e-06,
      "loss": 0.0037,
      "step": 20990
    },
    {
      "epoch": 0.3436144972592653,
      "grad_norm": 0.13055074699320773,
      "learning_rate": 8.298585048301157e-06,
      "loss": 0.0028,
      "step": 21000
    },
    {
      "epoch": 0.3437781232103412,
      "grad_norm": 0.1525672182778529,
      "learning_rate": 8.296438308133467e-06,
      "loss": 0.0054,
      "step": 21010
    },
    {
      "epoch": 0.343941749161417,
      "grad_norm": 0.1143492611670526,
      "learning_rate": 8.294290492561074e-06,
      "loss": 0.0039,
      "step": 21020
    },
    {
      "epoch": 0.34410537511249284,
      "grad_norm": 0.3819506086264186,
      "learning_rate": 8.29214160228467e-06,
      "loss": 0.0037,
      "step": 21030
    },
    {
      "epoch": 0.34426900106356867,
      "grad_norm": 0.19945894008106016,
      "learning_rate": 8.289991638005286e-06,
      "loss": 0.0045,
      "step": 21040
    },
    {
      "epoch": 0.3444326270146445,
      "grad_norm": 0.16224774545919984,
      "learning_rate": 8.287840600424312e-06,
      "loss": 0.0035,
      "step": 21050
    },
    {
      "epoch": 0.3445962529657204,
      "grad_norm": 0.09074054355020106,
      "learning_rate": 8.28568849024349e-06,
      "loss": 0.0042,
      "step": 21060
    },
    {
      "epoch": 0.3447598789167962,
      "grad_norm": 0.3604183568907086,
      "learning_rate": 8.283535308164901e-06,
      "loss": 0.0052,
      "step": 21070
    },
    {
      "epoch": 0.34492350486787204,
      "grad_norm": 0.40051290464463984,
      "learning_rate": 8.281381054890987e-06,
      "loss": 0.0031,
      "step": 21080
    },
    {
      "epoch": 0.3450871308189479,
      "grad_norm": 0.12258653691971061,
      "learning_rate": 8.279225731124536e-06,
      "loss": 0.0029,
      "step": 21090
    },
    {
      "epoch": 0.3452507567700237,
      "grad_norm": 0.16018771266067622,
      "learning_rate": 8.27706933756868e-06,
      "loss": 0.0067,
      "step": 21100
    },
    {
      "epoch": 0.3454143827210996,
      "grad_norm": 0.2427836154846401,
      "learning_rate": 8.274911874926907e-06,
      "loss": 0.0051,
      "step": 21110
    },
    {
      "epoch": 0.3455780086721754,
      "grad_norm": 0.3156056092451682,
      "learning_rate": 8.272753343903052e-06,
      "loss": 0.0044,
      "step": 21120
    },
    {
      "epoch": 0.34574163462325125,
      "grad_norm": 0.5119172317602168,
      "learning_rate": 8.270593745201297e-06,
      "loss": 0.006,
      "step": 21130
    },
    {
      "epoch": 0.3459052605743271,
      "grad_norm": 0.20966712588108682,
      "learning_rate": 8.26843307952617e-06,
      "loss": 0.0037,
      "step": 21140
    },
    {
      "epoch": 0.3460688865254029,
      "grad_norm": 0.14993499489853723,
      "learning_rate": 8.266271347582555e-06,
      "loss": 0.0051,
      "step": 21150
    },
    {
      "epoch": 0.3462325124764788,
      "grad_norm": 0.31698723927179806,
      "learning_rate": 8.264108550075674e-06,
      "loss": 0.004,
      "step": 21160
    },
    {
      "epoch": 0.3463961384275546,
      "grad_norm": 0.136062319828816,
      "learning_rate": 8.261944687711103e-06,
      "loss": 0.006,
      "step": 21170
    },
    {
      "epoch": 0.34655976437863045,
      "grad_norm": 0.23746139087716034,
      "learning_rate": 8.259779761194767e-06,
      "loss": 0.0068,
      "step": 21180
    },
    {
      "epoch": 0.3467233903297063,
      "grad_norm": 0.14484767838760634,
      "learning_rate": 8.25761377123293e-06,
      "loss": 0.005,
      "step": 21190
    },
    {
      "epoch": 0.3468870162807821,
      "grad_norm": 0.2468755209895756,
      "learning_rate": 8.25544671853221e-06,
      "loss": 0.0041,
      "step": 21200
    },
    {
      "epoch": 0.347050642231858,
      "grad_norm": 0.3270319737142135,
      "learning_rate": 8.25327860379957e-06,
      "loss": 0.0062,
      "step": 21210
    },
    {
      "epoch": 0.3472142681829338,
      "grad_norm": 0.18038277250682985,
      "learning_rate": 8.251109427742319e-06,
      "loss": 0.0042,
      "step": 21220
    },
    {
      "epoch": 0.34737789413400966,
      "grad_norm": 0.10596646650280081,
      "learning_rate": 8.248939191068112e-06,
      "loss": 0.0044,
      "step": 21230
    },
    {
      "epoch": 0.3475415200850855,
      "grad_norm": 0.07531578079228332,
      "learning_rate": 8.246767894484951e-06,
      "loss": 0.0032,
      "step": 21240
    },
    {
      "epoch": 0.3477051460361613,
      "grad_norm": 0.17584144291785261,
      "learning_rate": 8.244595538701183e-06,
      "loss": 0.0068,
      "step": 21250
    },
    {
      "epoch": 0.3478687719872372,
      "grad_norm": 0.4641516827922733,
      "learning_rate": 8.2424221244255e-06,
      "loss": 0.0051,
      "step": 21260
    },
    {
      "epoch": 0.34803239793831303,
      "grad_norm": 0.354582958243507,
      "learning_rate": 8.24024765236694e-06,
      "loss": 0.0046,
      "step": 21270
    },
    {
      "epoch": 0.34819602388938886,
      "grad_norm": 0.03288577018364058,
      "learning_rate": 8.238072123234886e-06,
      "loss": 0.0028,
      "step": 21280
    },
    {
      "epoch": 0.3483596498404647,
      "grad_norm": 0.4160287206232713,
      "learning_rate": 8.235895537739068e-06,
      "loss": 0.0037,
      "step": 21290
    },
    {
      "epoch": 0.3485232757915405,
      "grad_norm": 0.3761054668078874,
      "learning_rate": 8.233717896589555e-06,
      "loss": 0.0038,
      "step": 21300
    },
    {
      "epoch": 0.3486869017426164,
      "grad_norm": 0.160046013028521,
      "learning_rate": 8.231539200496767e-06,
      "loss": 0.0049,
      "step": 21310
    },
    {
      "epoch": 0.34885052769369224,
      "grad_norm": 0.1447209997521521,
      "learning_rate": 8.229359450171463e-06,
      "loss": 0.0052,
      "step": 21320
    },
    {
      "epoch": 0.34901415364476807,
      "grad_norm": 0.04679969984870674,
      "learning_rate": 8.227178646324748e-06,
      "loss": 0.0038,
      "step": 21330
    },
    {
      "epoch": 0.3491777795958439,
      "grad_norm": 0.19300305492658276,
      "learning_rate": 8.224996789668072e-06,
      "loss": 0.006,
      "step": 21340
    },
    {
      "epoch": 0.3493414055469197,
      "grad_norm": 0.08702013120550157,
      "learning_rate": 8.222813880913228e-06,
      "loss": 0.0038,
      "step": 21350
    },
    {
      "epoch": 0.34950503149799556,
      "grad_norm": 0.05577931449954347,
      "learning_rate": 8.220629920772347e-06,
      "loss": 0.0059,
      "step": 21360
    },
    {
      "epoch": 0.34966865744907144,
      "grad_norm": 0.061480121994317856,
      "learning_rate": 8.218444909957911e-06,
      "loss": 0.0049,
      "step": 21370
    },
    {
      "epoch": 0.34983228340014727,
      "grad_norm": 0.12283513595437923,
      "learning_rate": 8.216258849182744e-06,
      "loss": 0.0036,
      "step": 21380
    },
    {
      "epoch": 0.3499959093512231,
      "grad_norm": 0.17095257122231972,
      "learning_rate": 8.214071739160004e-06,
      "loss": 0.0105,
      "step": 21390
    },
    {
      "epoch": 0.35015953530229893,
      "grad_norm": 0.284607632213737,
      "learning_rate": 8.211883580603198e-06,
      "loss": 0.0047,
      "step": 21400
    },
    {
      "epoch": 0.35032316125337476,
      "grad_norm": 0.5103060200311446,
      "learning_rate": 8.209694374226175e-06,
      "loss": 0.0053,
      "step": 21410
    },
    {
      "epoch": 0.35048678720445064,
      "grad_norm": 0.35377893651170256,
      "learning_rate": 8.207504120743127e-06,
      "loss": 0.0044,
      "step": 21420
    },
    {
      "epoch": 0.3506504131555265,
      "grad_norm": 0.13316833575020487,
      "learning_rate": 8.205312820868581e-06,
      "loss": 0.0035,
      "step": 21430
    },
    {
      "epoch": 0.3508140391066023,
      "grad_norm": 0.2322430465996936,
      "learning_rate": 8.203120475317415e-06,
      "loss": 0.0024,
      "step": 21440
    },
    {
      "epoch": 0.35097766505767813,
      "grad_norm": 0.23034998353660002,
      "learning_rate": 8.200927084804839e-06,
      "loss": 0.0073,
      "step": 21450
    },
    {
      "epoch": 0.35114129100875396,
      "grad_norm": 0.47361087732602575,
      "learning_rate": 8.198732650046409e-06,
      "loss": 0.0081,
      "step": 21460
    },
    {
      "epoch": 0.35130491695982985,
      "grad_norm": 0.5479816952160357,
      "learning_rate": 8.19653717175802e-06,
      "loss": 0.0055,
      "step": 21470
    },
    {
      "epoch": 0.3514685429109057,
      "grad_norm": 0.4068777016603905,
      "learning_rate": 8.19434065065591e-06,
      "loss": 0.0043,
      "step": 21480
    },
    {
      "epoch": 0.3516321688619815,
      "grad_norm": 0.34150722644298087,
      "learning_rate": 8.192143087456652e-06,
      "loss": 0.0066,
      "step": 21490
    },
    {
      "epoch": 0.35179579481305734,
      "grad_norm": 0.23758385634434,
      "learning_rate": 8.189944482877166e-06,
      "loss": 0.0058,
      "step": 21500
    },
    {
      "epoch": 0.35195942076413317,
      "grad_norm": 0.2786234188022552,
      "learning_rate": 8.187744837634703e-06,
      "loss": 0.0046,
      "step": 21510
    },
    {
      "epoch": 0.35212304671520905,
      "grad_norm": 0.1742955314541393,
      "learning_rate": 8.185544152446866e-06,
      "loss": 0.0032,
      "step": 21520
    },
    {
      "epoch": 0.3522866726662849,
      "grad_norm": 0.30605901548399084,
      "learning_rate": 8.18334242803158e-06,
      "loss": 0.005,
      "step": 21530
    },
    {
      "epoch": 0.3524502986173607,
      "grad_norm": 0.14296784515913666,
      "learning_rate": 8.181139665107125e-06,
      "loss": 0.0044,
      "step": 21540
    },
    {
      "epoch": 0.35261392456843654,
      "grad_norm": 0.08027658208682562,
      "learning_rate": 8.178935864392113e-06,
      "loss": 0.0032,
      "step": 21550
    },
    {
      "epoch": 0.3527775505195124,
      "grad_norm": 0.1932504302687466,
      "learning_rate": 8.176731026605492e-06,
      "loss": 0.004,
      "step": 21560
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.33930967881985696,
      "learning_rate": 8.174525152466555e-06,
      "loss": 0.0036,
      "step": 21570
    },
    {
      "epoch": 0.3531048024216641,
      "grad_norm": 0.45273479046442755,
      "learning_rate": 8.172318242694925e-06,
      "loss": 0.0056,
      "step": 21580
    },
    {
      "epoch": 0.3532684283727399,
      "grad_norm": 0.16908740662601207,
      "learning_rate": 8.170110298010572e-06,
      "loss": 0.0042,
      "step": 21590
    },
    {
      "epoch": 0.35343205432381575,
      "grad_norm": 0.26052789833320583,
      "learning_rate": 8.167901319133798e-06,
      "loss": 0.0039,
      "step": 21600
    },
    {
      "epoch": 0.3535956802748916,
      "grad_norm": 0.12491718638791037,
      "learning_rate": 8.16569130678524e-06,
      "loss": 0.0056,
      "step": 21610
    },
    {
      "epoch": 0.35375930622596746,
      "grad_norm": 0.299587139198521,
      "learning_rate": 8.163480261685878e-06,
      "loss": 0.0061,
      "step": 21620
    },
    {
      "epoch": 0.3539229321770433,
      "grad_norm": 0.1868248117022889,
      "learning_rate": 8.161268184557027e-06,
      "loss": 0.0047,
      "step": 21630
    },
    {
      "epoch": 0.3540865581281191,
      "grad_norm": 0.4087806482608733,
      "learning_rate": 8.159055076120335e-06,
      "loss": 0.0058,
      "step": 21640
    },
    {
      "epoch": 0.35425018407919495,
      "grad_norm": 0.2688705596798527,
      "learning_rate": 8.156840937097793e-06,
      "loss": 0.004,
      "step": 21650
    },
    {
      "epoch": 0.3544138100302708,
      "grad_norm": 0.2886753511100073,
      "learning_rate": 8.154625768211724e-06,
      "loss": 0.0085,
      "step": 21660
    },
    {
      "epoch": 0.35457743598134667,
      "grad_norm": 0.19786059982396528,
      "learning_rate": 8.152409570184785e-06,
      "loss": 0.0061,
      "step": 21670
    },
    {
      "epoch": 0.3547410619324225,
      "grad_norm": 0.38351927578523554,
      "learning_rate": 8.150192343739973e-06,
      "loss": 0.004,
      "step": 21680
    },
    {
      "epoch": 0.3549046878834983,
      "grad_norm": 0.268873636356188,
      "learning_rate": 8.14797408960062e-06,
      "loss": 0.0096,
      "step": 21690
    },
    {
      "epoch": 0.35506831383457416,
      "grad_norm": 0.05659051967088951,
      "learning_rate": 8.145754808490392e-06,
      "loss": 0.0038,
      "step": 21700
    },
    {
      "epoch": 0.35523193978565,
      "grad_norm": 0.1594146607796532,
      "learning_rate": 8.143534501133288e-06,
      "loss": 0.0047,
      "step": 21710
    },
    {
      "epoch": 0.35539556573672587,
      "grad_norm": 0.21505692245827976,
      "learning_rate": 8.141313168253649e-06,
      "loss": 0.0041,
      "step": 21720
    },
    {
      "epoch": 0.3555591916878017,
      "grad_norm": 0.25306539222798285,
      "learning_rate": 8.139090810576138e-06,
      "loss": 0.0045,
      "step": 21730
    },
    {
      "epoch": 0.35572281763887753,
      "grad_norm": 0.13680675289337932,
      "learning_rate": 8.136867428825766e-06,
      "loss": 0.0048,
      "step": 21740
    },
    {
      "epoch": 0.35588644358995336,
      "grad_norm": 0.17531384502122008,
      "learning_rate": 8.134643023727869e-06,
      "loss": 0.0048,
      "step": 21750
    },
    {
      "epoch": 0.3560500695410292,
      "grad_norm": 0.19604067591260385,
      "learning_rate": 8.132417596008121e-06,
      "loss": 0.0042,
      "step": 21760
    },
    {
      "epoch": 0.356213695492105,
      "grad_norm": 0.10671383357160226,
      "learning_rate": 8.130191146392529e-06,
      "loss": 0.0062,
      "step": 21770
    },
    {
      "epoch": 0.3563773214431809,
      "grad_norm": 0.6925741570314006,
      "learning_rate": 8.127963675607434e-06,
      "loss": 0.0056,
      "step": 21780
    },
    {
      "epoch": 0.35654094739425674,
      "grad_norm": 0.20657289278593421,
      "learning_rate": 8.125735184379504e-06,
      "loss": 0.0029,
      "step": 21790
    },
    {
      "epoch": 0.35670457334533257,
      "grad_norm": 0.34744837496155256,
      "learning_rate": 8.12350567343575e-06,
      "loss": 0.0081,
      "step": 21800
    },
    {
      "epoch": 0.3568681992964084,
      "grad_norm": 0.29106948130856475,
      "learning_rate": 8.121275143503507e-06,
      "loss": 0.0038,
      "step": 21810
    },
    {
      "epoch": 0.3570318252474842,
      "grad_norm": 0.2545472237161258,
      "learning_rate": 8.119043595310448e-06,
      "loss": 0.0047,
      "step": 21820
    },
    {
      "epoch": 0.3571954511985601,
      "grad_norm": 0.20660940929341418,
      "learning_rate": 8.116811029584576e-06,
      "loss": 0.0043,
      "step": 21830
    },
    {
      "epoch": 0.35735907714963594,
      "grad_norm": 0.13566472365068155,
      "learning_rate": 8.114577447054222e-06,
      "loss": 0.0041,
      "step": 21840
    },
    {
      "epoch": 0.35752270310071177,
      "grad_norm": 0.08089286338907706,
      "learning_rate": 8.11234284844806e-06,
      "loss": 0.0052,
      "step": 21850
    },
    {
      "epoch": 0.3576863290517876,
      "grad_norm": 0.1498472041518974,
      "learning_rate": 8.110107234495084e-06,
      "loss": 0.0031,
      "step": 21860
    },
    {
      "epoch": 0.35784995500286343,
      "grad_norm": 0.1073039911172813,
      "learning_rate": 8.107870605924624e-06,
      "loss": 0.0043,
      "step": 21870
    },
    {
      "epoch": 0.3580135809539393,
      "grad_norm": 0.2902065332125187,
      "learning_rate": 8.10563296346634e-06,
      "loss": 0.0033,
      "step": 21880
    },
    {
      "epoch": 0.35817720690501514,
      "grad_norm": 0.1879263704293451,
      "learning_rate": 8.103394307850225e-06,
      "loss": 0.0049,
      "step": 21890
    },
    {
      "epoch": 0.358340832856091,
      "grad_norm": 0.541430949456604,
      "learning_rate": 8.101154639806598e-06,
      "loss": 0.0068,
      "step": 21900
    },
    {
      "epoch": 0.3585044588071668,
      "grad_norm": 0.24423798519045195,
      "learning_rate": 8.098913960066115e-06,
      "loss": 0.0052,
      "step": 21910
    },
    {
      "epoch": 0.35866808475824263,
      "grad_norm": 0.27221078320296016,
      "learning_rate": 8.096672269359755e-06,
      "loss": 0.0057,
      "step": 21920
    },
    {
      "epoch": 0.3588317107093185,
      "grad_norm": 0.3056004130605978,
      "learning_rate": 8.094429568418831e-06,
      "loss": 0.0064,
      "step": 21930
    },
    {
      "epoch": 0.35899533666039435,
      "grad_norm": 0.12140122293182473,
      "learning_rate": 8.092185857974986e-06,
      "loss": 0.0036,
      "step": 21940
    },
    {
      "epoch": 0.3591589626114702,
      "grad_norm": 0.2383787650848925,
      "learning_rate": 8.089941138760186e-06,
      "loss": 0.0052,
      "step": 21950
    },
    {
      "epoch": 0.359322588562546,
      "grad_norm": 0.060690717801894466,
      "learning_rate": 8.087695411506738e-06,
      "loss": 0.0029,
      "step": 21960
    },
    {
      "epoch": 0.35948621451362184,
      "grad_norm": 0.13027267796747477,
      "learning_rate": 8.085448676947265e-06,
      "loss": 0.0038,
      "step": 21970
    },
    {
      "epoch": 0.3596498404646977,
      "grad_norm": 0.06777375110268201,
      "learning_rate": 8.083200935814725e-06,
      "loss": 0.0056,
      "step": 21980
    },
    {
      "epoch": 0.35981346641577355,
      "grad_norm": 0.26404782395114973,
      "learning_rate": 8.080952188842407e-06,
      "loss": 0.0049,
      "step": 21990
    },
    {
      "epoch": 0.3599770923668494,
      "grad_norm": 0.07344352482194981,
      "learning_rate": 8.078702436763925e-06,
      "loss": 0.0043,
      "step": 22000
    },
    {
      "epoch": 0.3601407183179252,
      "grad_norm": 0.20181901262248428,
      "learning_rate": 8.076451680313216e-06,
      "loss": 0.0036,
      "step": 22010
    },
    {
      "epoch": 0.36030434426900104,
      "grad_norm": 0.2687343206702619,
      "learning_rate": 8.074199920224555e-06,
      "loss": 0.0051,
      "step": 22020
    },
    {
      "epoch": 0.3604679702200769,
      "grad_norm": 0.12082289643392462,
      "learning_rate": 8.071947157232536e-06,
      "loss": 0.005,
      "step": 22030
    },
    {
      "epoch": 0.36063159617115276,
      "grad_norm": 0.20599436045430314,
      "learning_rate": 8.069693392072082e-06,
      "loss": 0.0053,
      "step": 22040
    },
    {
      "epoch": 0.3607952221222286,
      "grad_norm": 0.03827285256559599,
      "learning_rate": 8.067438625478445e-06,
      "loss": 0.0033,
      "step": 22050
    },
    {
      "epoch": 0.3609588480733044,
      "grad_norm": 0.2147216862900357,
      "learning_rate": 8.065182858187207e-06,
      "loss": 0.0041,
      "step": 22060
    },
    {
      "epoch": 0.36112247402438025,
      "grad_norm": 0.2671410455933201,
      "learning_rate": 8.062926090934264e-06,
      "loss": 0.0059,
      "step": 22070
    },
    {
      "epoch": 0.36128609997545613,
      "grad_norm": 0.45165716419902047,
      "learning_rate": 8.060668324455853e-06,
      "loss": 0.0053,
      "step": 22080
    },
    {
      "epoch": 0.36144972592653196,
      "grad_norm": 0.06592630615694613,
      "learning_rate": 8.05840955948853e-06,
      "loss": 0.004,
      "step": 22090
    },
    {
      "epoch": 0.3616133518776078,
      "grad_norm": 0.21117603204229507,
      "learning_rate": 8.056149796769173e-06,
      "loss": 0.0042,
      "step": 22100
    },
    {
      "epoch": 0.3617769778286836,
      "grad_norm": 0.2091945921456715,
      "learning_rate": 8.053889037034992e-06,
      "loss": 0.0047,
      "step": 22110
    },
    {
      "epoch": 0.36194060377975945,
      "grad_norm": 0.564376330231919,
      "learning_rate": 8.051627281023519e-06,
      "loss": 0.0036,
      "step": 22120
    },
    {
      "epoch": 0.36210422973083534,
      "grad_norm": 0.267186532798065,
      "learning_rate": 8.049364529472616e-06,
      "loss": 0.0061,
      "step": 22130
    },
    {
      "epoch": 0.36226785568191117,
      "grad_norm": 0.41323442866810006,
      "learning_rate": 8.04710078312046e-06,
      "loss": 0.0047,
      "step": 22140
    },
    {
      "epoch": 0.362431481632987,
      "grad_norm": 0.17939918237862795,
      "learning_rate": 8.044836042705561e-06,
      "loss": 0.004,
      "step": 22150
    },
    {
      "epoch": 0.3625951075840628,
      "grad_norm": 0.10576247212815698,
      "learning_rate": 8.042570308966751e-06,
      "loss": 0.0055,
      "step": 22160
    },
    {
      "epoch": 0.36275873353513866,
      "grad_norm": 0.35053850065871706,
      "learning_rate": 8.040303582643185e-06,
      "loss": 0.0039,
      "step": 22170
    },
    {
      "epoch": 0.36292235948621454,
      "grad_norm": 0.20925125664040806,
      "learning_rate": 8.03803586447434e-06,
      "loss": 0.0058,
      "step": 22180
    },
    {
      "epoch": 0.36308598543729037,
      "grad_norm": 0.18325287190555176,
      "learning_rate": 8.035767155200022e-06,
      "loss": 0.0062,
      "step": 22190
    },
    {
      "epoch": 0.3632496113883662,
      "grad_norm": 0.19090124460483998,
      "learning_rate": 8.033497455560359e-06,
      "loss": 0.0042,
      "step": 22200
    },
    {
      "epoch": 0.36341323733944203,
      "grad_norm": 0.1718975501360852,
      "learning_rate": 8.031226766295794e-06,
      "loss": 0.0054,
      "step": 22210
    },
    {
      "epoch": 0.36357686329051786,
      "grad_norm": 0.1990188426857915,
      "learning_rate": 8.028955088147106e-06,
      "loss": 0.0054,
      "step": 22220
    },
    {
      "epoch": 0.3637404892415937,
      "grad_norm": 0.41056218075581696,
      "learning_rate": 8.026682421855386e-06,
      "loss": 0.007,
      "step": 22230
    },
    {
      "epoch": 0.3639041151926696,
      "grad_norm": 0.12829855457467737,
      "learning_rate": 8.02440876816205e-06,
      "loss": 0.0039,
      "step": 22240
    },
    {
      "epoch": 0.3640677411437454,
      "grad_norm": 0.24589932467234477,
      "learning_rate": 8.02213412780884e-06,
      "loss": 0.0059,
      "step": 22250
    },
    {
      "epoch": 0.36423136709482123,
      "grad_norm": 0.08400456262599232,
      "learning_rate": 8.019858501537818e-06,
      "loss": 0.0044,
      "step": 22260
    },
    {
      "epoch": 0.36439499304589706,
      "grad_norm": 0.5791083382979467,
      "learning_rate": 8.017581890091363e-06,
      "loss": 0.0051,
      "step": 22270
    },
    {
      "epoch": 0.3645586189969729,
      "grad_norm": 0.3022679329800277,
      "learning_rate": 8.015304294212183e-06,
      "loss": 0.0041,
      "step": 22280
    },
    {
      "epoch": 0.3647222449480488,
      "grad_norm": 0.2651324712526449,
      "learning_rate": 8.013025714643303e-06,
      "loss": 0.0038,
      "step": 22290
    },
    {
      "epoch": 0.3648858708991246,
      "grad_norm": 0.6173237085122881,
      "learning_rate": 8.010746152128065e-06,
      "loss": 0.0088,
      "step": 22300
    },
    {
      "epoch": 0.36504949685020044,
      "grad_norm": 0.1620804500124975,
      "learning_rate": 8.008465607410143e-06,
      "loss": 0.0039,
      "step": 22310
    },
    {
      "epoch": 0.36521312280127627,
      "grad_norm": 0.09480572820867364,
      "learning_rate": 8.006184081233517e-06,
      "loss": 0.0035,
      "step": 22320
    },
    {
      "epoch": 0.3653767487523521,
      "grad_norm": 0.24982967973314837,
      "learning_rate": 8.003901574342498e-06,
      "loss": 0.0038,
      "step": 22330
    },
    {
      "epoch": 0.365540374703428,
      "grad_norm": 0.1420652940557556,
      "learning_rate": 8.001618087481716e-06,
      "loss": 0.003,
      "step": 22340
    },
    {
      "epoch": 0.3657040006545038,
      "grad_norm": 0.018918582899131352,
      "learning_rate": 7.999333621396115e-06,
      "loss": 0.0065,
      "step": 22350
    },
    {
      "epoch": 0.36586762660557964,
      "grad_norm": 0.20904676014770399,
      "learning_rate": 7.997048176830961e-06,
      "loss": 0.0029,
      "step": 22360
    },
    {
      "epoch": 0.3660312525566555,
      "grad_norm": 0.2919347511958678,
      "learning_rate": 7.994761754531844e-06,
      "loss": 0.0046,
      "step": 22370
    },
    {
      "epoch": 0.3661948785077313,
      "grad_norm": 0.1890912854510077,
      "learning_rate": 7.992474355244663e-06,
      "loss": 0.0048,
      "step": 22380
    },
    {
      "epoch": 0.3663585044588072,
      "grad_norm": 0.27444379879412956,
      "learning_rate": 7.990185979715648e-06,
      "loss": 0.0044,
      "step": 22390
    },
    {
      "epoch": 0.366522130409883,
      "grad_norm": 0.09103498353880408,
      "learning_rate": 7.987896628691336e-06,
      "loss": 0.0027,
      "step": 22400
    },
    {
      "epoch": 0.36668575636095885,
      "grad_norm": 0.11851401053287791,
      "learning_rate": 7.985606302918589e-06,
      "loss": 0.0047,
      "step": 22410
    },
    {
      "epoch": 0.3668493823120347,
      "grad_norm": 0.038468302837450334,
      "learning_rate": 7.983315003144586e-06,
      "loss": 0.0042,
      "step": 22420
    },
    {
      "epoch": 0.3670130082631105,
      "grad_norm": 0.43509016582558974,
      "learning_rate": 7.98102273011682e-06,
      "loss": 0.0038,
      "step": 22430
    },
    {
      "epoch": 0.3671766342141864,
      "grad_norm": 0.07581748607239056,
      "learning_rate": 7.97872948458311e-06,
      "loss": 0.0052,
      "step": 22440
    },
    {
      "epoch": 0.3673402601652622,
      "grad_norm": 0.20175306482536795,
      "learning_rate": 7.976435267291583e-06,
      "loss": 0.0035,
      "step": 22450
    },
    {
      "epoch": 0.36750388611633805,
      "grad_norm": 0.19902981340182438,
      "learning_rate": 7.974140078990686e-06,
      "loss": 0.004,
      "step": 22460
    },
    {
      "epoch": 0.3676675120674139,
      "grad_norm": 0.3350546693634301,
      "learning_rate": 7.971843920429187e-06,
      "loss": 0.0036,
      "step": 22470
    },
    {
      "epoch": 0.3678311380184897,
      "grad_norm": 0.1442479477327456,
      "learning_rate": 7.969546792356164e-06,
      "loss": 0.0062,
      "step": 22480
    },
    {
      "epoch": 0.3679947639695656,
      "grad_norm": 0.2218002598970167,
      "learning_rate": 7.967248695521017e-06,
      "loss": 0.0057,
      "step": 22490
    },
    {
      "epoch": 0.3681583899206414,
      "grad_norm": 0.13058529815057038,
      "learning_rate": 7.964949630673456e-06,
      "loss": 0.0033,
      "step": 22500
    },
    {
      "epoch": 0.36832201587171726,
      "grad_norm": 0.17822038345815083,
      "learning_rate": 7.962649598563513e-06,
      "loss": 0.0072,
      "step": 22510
    },
    {
      "epoch": 0.3684856418227931,
      "grad_norm": 0.07651427925511488,
      "learning_rate": 7.960348599941533e-06,
      "loss": 0.0047,
      "step": 22520
    },
    {
      "epoch": 0.3686492677738689,
      "grad_norm": 0.18947187168978472,
      "learning_rate": 7.958046635558175e-06,
      "loss": 0.0037,
      "step": 22530
    },
    {
      "epoch": 0.3688128937249448,
      "grad_norm": 0.33646478914927486,
      "learning_rate": 7.955743706164414e-06,
      "loss": 0.0043,
      "step": 22540
    },
    {
      "epoch": 0.36897651967602063,
      "grad_norm": 0.14466146040736594,
      "learning_rate": 7.953439812511541e-06,
      "loss": 0.0058,
      "step": 22550
    },
    {
      "epoch": 0.36914014562709646,
      "grad_norm": 0.20571336708909738,
      "learning_rate": 7.951134955351159e-06,
      "loss": 0.0046,
      "step": 22560
    },
    {
      "epoch": 0.3693037715781723,
      "grad_norm": 0.11899569063552054,
      "learning_rate": 7.948829135435188e-06,
      "loss": 0.0043,
      "step": 22570
    },
    {
      "epoch": 0.3694673975292481,
      "grad_norm": 0.09638828990684081,
      "learning_rate": 7.946522353515863e-06,
      "loss": 0.0023,
      "step": 22580
    },
    {
      "epoch": 0.369631023480324,
      "grad_norm": 0.15540463561014836,
      "learning_rate": 7.944214610345727e-06,
      "loss": 0.004,
      "step": 22590
    },
    {
      "epoch": 0.36979464943139984,
      "grad_norm": 0.2736904618129395,
      "learning_rate": 7.941905906677641e-06,
      "loss": 0.0047,
      "step": 22600
    },
    {
      "epoch": 0.36995827538247567,
      "grad_norm": 0.2631068626683752,
      "learning_rate": 7.93959624326478e-06,
      "loss": 0.0047,
      "step": 22610
    },
    {
      "epoch": 0.3701219013335515,
      "grad_norm": 0.4128766923777127,
      "learning_rate": 7.93728562086063e-06,
      "loss": 0.005,
      "step": 22620
    },
    {
      "epoch": 0.3702855272846273,
      "grad_norm": 0.5910514253286557,
      "learning_rate": 7.934974040218995e-06,
      "loss": 0.0074,
      "step": 22630
    },
    {
      "epoch": 0.3704491532357032,
      "grad_norm": 0.13597735266234007,
      "learning_rate": 7.93266150209398e-06,
      "loss": 0.0044,
      "step": 22640
    },
    {
      "epoch": 0.37061277918677904,
      "grad_norm": 0.27802477851363055,
      "learning_rate": 7.930348007240015e-06,
      "loss": 0.0058,
      "step": 22650
    },
    {
      "epoch": 0.37077640513785487,
      "grad_norm": 0.2644343739747211,
      "learning_rate": 7.928033556411835e-06,
      "loss": 0.0069,
      "step": 22660
    },
    {
      "epoch": 0.3709400310889307,
      "grad_norm": 0.2965258882218535,
      "learning_rate": 7.925718150364488e-06,
      "loss": 0.0084,
      "step": 22670
    },
    {
      "epoch": 0.37110365704000653,
      "grad_norm": 0.34343811577731503,
      "learning_rate": 7.923401789853335e-06,
      "loss": 0.0062,
      "step": 22680
    },
    {
      "epoch": 0.37126728299108236,
      "grad_norm": 0.08198107450571357,
      "learning_rate": 7.921084475634047e-06,
      "loss": 0.0038,
      "step": 22690
    },
    {
      "epoch": 0.37143090894215824,
      "grad_norm": 0.1598291298521683,
      "learning_rate": 7.91876620846261e-06,
      "loss": 0.0055,
      "step": 22700
    },
    {
      "epoch": 0.3715945348932341,
      "grad_norm": 0.1320846220285246,
      "learning_rate": 7.916446989095313e-06,
      "loss": 0.003,
      "step": 22710
    },
    {
      "epoch": 0.3717581608443099,
      "grad_norm": 0.34019235100729966,
      "learning_rate": 7.914126818288763e-06,
      "loss": 0.0028,
      "step": 22720
    },
    {
      "epoch": 0.37192178679538573,
      "grad_norm": 0.18809256027740193,
      "learning_rate": 7.911805696799873e-06,
      "loss": 0.0039,
      "step": 22730
    },
    {
      "epoch": 0.37208541274646156,
      "grad_norm": 0.3126583119581344,
      "learning_rate": 7.90948362538587e-06,
      "loss": 0.0026,
      "step": 22740
    },
    {
      "epoch": 0.37224903869753745,
      "grad_norm": 0.19810785043559126,
      "learning_rate": 7.907160604804288e-06,
      "loss": 0.0043,
      "step": 22750
    },
    {
      "epoch": 0.3724126646486133,
      "grad_norm": 0.39099521854285685,
      "learning_rate": 7.904836635812972e-06,
      "loss": 0.0043,
      "step": 22760
    },
    {
      "epoch": 0.3725762905996891,
      "grad_norm": 0.12100920360865144,
      "learning_rate": 7.902511719170072e-06,
      "loss": 0.0032,
      "step": 22770
    },
    {
      "epoch": 0.37273991655076494,
      "grad_norm": 0.37345056479715394,
      "learning_rate": 7.900185855634055e-06,
      "loss": 0.0035,
      "step": 22780
    },
    {
      "epoch": 0.37290354250184077,
      "grad_norm": 0.30403356425097144,
      "learning_rate": 7.897859045963691e-06,
      "loss": 0.0045,
      "step": 22790
    },
    {
      "epoch": 0.37306716845291665,
      "grad_norm": 0.365619122958088,
      "learning_rate": 7.895531290918062e-06,
      "loss": 0.006,
      "step": 22800
    },
    {
      "epoch": 0.3732307944039925,
      "grad_norm": 0.10193440457323168,
      "learning_rate": 7.893202591256558e-06,
      "loss": 0.0043,
      "step": 22810
    },
    {
      "epoch": 0.3733944203550683,
      "grad_norm": 0.18744634394516266,
      "learning_rate": 7.89087294773887e-06,
      "loss": 0.0043,
      "step": 22820
    },
    {
      "epoch": 0.37355804630614414,
      "grad_norm": 0.2933513423447109,
      "learning_rate": 7.888542361125012e-06,
      "loss": 0.0042,
      "step": 22830
    },
    {
      "epoch": 0.37372167225722,
      "grad_norm": 0.3244846918389883,
      "learning_rate": 7.886210832175288e-06,
      "loss": 0.0032,
      "step": 22840
    },
    {
      "epoch": 0.37388529820829586,
      "grad_norm": 0.5707253912789513,
      "learning_rate": 7.883878361650323e-06,
      "loss": 0.0037,
      "step": 22850
    },
    {
      "epoch": 0.3740489241593717,
      "grad_norm": 0.19542601265558684,
      "learning_rate": 7.881544950311042e-06,
      "loss": 0.0058,
      "step": 22860
    },
    {
      "epoch": 0.3742125501104475,
      "grad_norm": 0.14725526847070805,
      "learning_rate": 7.87921059891868e-06,
      "loss": 0.0045,
      "step": 22870
    },
    {
      "epoch": 0.37437617606152335,
      "grad_norm": 0.22371416217253562,
      "learning_rate": 7.876875308234782e-06,
      "loss": 0.0051,
      "step": 22880
    },
    {
      "epoch": 0.3745398020125992,
      "grad_norm": 0.3101014089284409,
      "learning_rate": 7.874539079021188e-06,
      "loss": 0.0031,
      "step": 22890
    },
    {
      "epoch": 0.37470342796367506,
      "grad_norm": 0.11512983161318277,
      "learning_rate": 7.872201912040056e-06,
      "loss": 0.0034,
      "step": 22900
    },
    {
      "epoch": 0.3748670539147509,
      "grad_norm": 0.12509936324613832,
      "learning_rate": 7.869863808053844e-06,
      "loss": 0.0035,
      "step": 22910
    },
    {
      "epoch": 0.3750306798658267,
      "grad_norm": 0.13445543336436533,
      "learning_rate": 7.867524767825319e-06,
      "loss": 0.0037,
      "step": 22920
    },
    {
      "epoch": 0.37519430581690255,
      "grad_norm": 0.19703069588882463,
      "learning_rate": 7.865184792117548e-06,
      "loss": 0.0029,
      "step": 22930
    },
    {
      "epoch": 0.3753579317679784,
      "grad_norm": 0.23687038294884713,
      "learning_rate": 7.86284388169391e-06,
      "loss": 0.0049,
      "step": 22940
    },
    {
      "epoch": 0.37552155771905427,
      "grad_norm": 0.19932318637696375,
      "learning_rate": 7.860502037318082e-06,
      "loss": 0.0034,
      "step": 22950
    },
    {
      "epoch": 0.3756851836701301,
      "grad_norm": 0.2854873014982349,
      "learning_rate": 7.858159259754053e-06,
      "loss": 0.0034,
      "step": 22960
    },
    {
      "epoch": 0.3758488096212059,
      "grad_norm": 0.1490873789145156,
      "learning_rate": 7.85581554976611e-06,
      "loss": 0.0042,
      "step": 22970
    },
    {
      "epoch": 0.37601243557228176,
      "grad_norm": 0.2664544290072381,
      "learning_rate": 7.853470908118849e-06,
      "loss": 0.0032,
      "step": 22980
    },
    {
      "epoch": 0.3761760615233576,
      "grad_norm": 0.40427861844606977,
      "learning_rate": 7.85112533557717e-06,
      "loss": 0.0056,
      "step": 22990
    },
    {
      "epoch": 0.37633968747443347,
      "grad_norm": 0.2796467438660062,
      "learning_rate": 7.848778832906267e-06,
      "loss": 0.0059,
      "step": 23000
    },
    {
      "epoch": 0.3765033134255093,
      "grad_norm": 0.09917683090435539,
      "learning_rate": 7.846431400871652e-06,
      "loss": 0.0043,
      "step": 23010
    },
    {
      "epoch": 0.37666693937658513,
      "grad_norm": 0.07909310318799044,
      "learning_rate": 7.844083040239132e-06,
      "loss": 0.0045,
      "step": 23020
    },
    {
      "epoch": 0.37683056532766096,
      "grad_norm": 0.2939838305605887,
      "learning_rate": 7.841733751774816e-06,
      "loss": 0.004,
      "step": 23030
    },
    {
      "epoch": 0.3769941912787368,
      "grad_norm": 0.1260181168443803,
      "learning_rate": 7.839383536245117e-06,
      "loss": 0.0059,
      "step": 23040
    },
    {
      "epoch": 0.3771578172298127,
      "grad_norm": 0.4634244806399275,
      "learning_rate": 7.837032394416756e-06,
      "loss": 0.005,
      "step": 23050
    },
    {
      "epoch": 0.3773214431808885,
      "grad_norm": 0.1738223323037068,
      "learning_rate": 7.834680327056748e-06,
      "loss": 0.0045,
      "step": 23060
    },
    {
      "epoch": 0.37748506913196433,
      "grad_norm": 0.10492700250068246,
      "learning_rate": 7.832327334932413e-06,
      "loss": 0.003,
      "step": 23070
    },
    {
      "epoch": 0.37764869508304016,
      "grad_norm": 0.05876576792953704,
      "learning_rate": 7.829973418811372e-06,
      "loss": 0.0039,
      "step": 23080
    },
    {
      "epoch": 0.377812321034116,
      "grad_norm": 0.3431615570779223,
      "learning_rate": 7.827618579461553e-06,
      "loss": 0.0055,
      "step": 23090
    },
    {
      "epoch": 0.3779759469851918,
      "grad_norm": 0.48790285877659084,
      "learning_rate": 7.825262817651177e-06,
      "loss": 0.0035,
      "step": 23100
    },
    {
      "epoch": 0.3781395729362677,
      "grad_norm": 0.18904968153564555,
      "learning_rate": 7.822906134148771e-06,
      "loss": 0.0037,
      "step": 23110
    },
    {
      "epoch": 0.37830319888734354,
      "grad_norm": 0.23568565868984,
      "learning_rate": 7.820548529723162e-06,
      "loss": 0.0032,
      "step": 23120
    },
    {
      "epoch": 0.37846682483841937,
      "grad_norm": 0.1474894882955067,
      "learning_rate": 7.818190005143476e-06,
      "loss": 0.0053,
      "step": 23130
    },
    {
      "epoch": 0.3786304507894952,
      "grad_norm": 0.3924760327712476,
      "learning_rate": 7.815830561179137e-06,
      "loss": 0.0037,
      "step": 23140
    },
    {
      "epoch": 0.37879407674057103,
      "grad_norm": 0.3869784810142888,
      "learning_rate": 7.813470198599877e-06,
      "loss": 0.0055,
      "step": 23150
    },
    {
      "epoch": 0.3789577026916469,
      "grad_norm": 0.14721768230352789,
      "learning_rate": 7.811108918175721e-06,
      "loss": 0.0042,
      "step": 23160
    },
    {
      "epoch": 0.37912132864272274,
      "grad_norm": 0.35248357123434165,
      "learning_rate": 7.808746720676991e-06,
      "loss": 0.0045,
      "step": 23170
    },
    {
      "epoch": 0.3792849545937986,
      "grad_norm": 0.36251417848818146,
      "learning_rate": 7.806383606874316e-06,
      "loss": 0.0041,
      "step": 23180
    },
    {
      "epoch": 0.3794485805448744,
      "grad_norm": 0.10912764567282592,
      "learning_rate": 7.804019577538623e-06,
      "loss": 0.0031,
      "step": 23190
    },
    {
      "epoch": 0.37961220649595023,
      "grad_norm": 0.20429003794117925,
      "learning_rate": 7.801654633441127e-06,
      "loss": 0.0036,
      "step": 23200
    },
    {
      "epoch": 0.3797758324470261,
      "grad_norm": 0.27048246653185415,
      "learning_rate": 7.799288775353356e-06,
      "loss": 0.0061,
      "step": 23210
    },
    {
      "epoch": 0.37993945839810195,
      "grad_norm": 0.3452122167508699,
      "learning_rate": 7.796922004047127e-06,
      "loss": 0.005,
      "step": 23220
    },
    {
      "epoch": 0.3801030843491778,
      "grad_norm": 0.15326631573756364,
      "learning_rate": 7.794554320294554e-06,
      "loss": 0.0036,
      "step": 23230
    },
    {
      "epoch": 0.3802667103002536,
      "grad_norm": 0.09747025014605211,
      "learning_rate": 7.792185724868057e-06,
      "loss": 0.0043,
      "step": 23240
    },
    {
      "epoch": 0.38043033625132944,
      "grad_norm": 0.08128891271270045,
      "learning_rate": 7.789816218540347e-06,
      "loss": 0.0031,
      "step": 23250
    },
    {
      "epoch": 0.3805939622024053,
      "grad_norm": 0.15003152295647687,
      "learning_rate": 7.787445802084431e-06,
      "loss": 0.0028,
      "step": 23260
    },
    {
      "epoch": 0.38075758815348115,
      "grad_norm": 0.44895299846778386,
      "learning_rate": 7.785074476273619e-06,
      "loss": 0.0044,
      "step": 23270
    },
    {
      "epoch": 0.380921214104557,
      "grad_norm": 0.32355572764142454,
      "learning_rate": 7.782702241881511e-06,
      "loss": 0.0033,
      "step": 23280
    },
    {
      "epoch": 0.3810848400556328,
      "grad_norm": 0.3883518724584549,
      "learning_rate": 7.780329099682009e-06,
      "loss": 0.0053,
      "step": 23290
    },
    {
      "epoch": 0.38124846600670864,
      "grad_norm": 0.3550676649078261,
      "learning_rate": 7.777955050449309e-06,
      "loss": 0.0039,
      "step": 23300
    },
    {
      "epoch": 0.3814120919577845,
      "grad_norm": 0.2290813706395646,
      "learning_rate": 7.775580094957898e-06,
      "loss": 0.0054,
      "step": 23310
    },
    {
      "epoch": 0.38157571790886036,
      "grad_norm": 0.23169392574836783,
      "learning_rate": 7.773204233982568e-06,
      "loss": 0.0036,
      "step": 23320
    },
    {
      "epoch": 0.3817393438599362,
      "grad_norm": 0.15030269521599784,
      "learning_rate": 7.7708274682984e-06,
      "loss": 0.0033,
      "step": 23330
    },
    {
      "epoch": 0.381902969811012,
      "grad_norm": 0.08322809389609011,
      "learning_rate": 7.76844979868077e-06,
      "loss": 0.0046,
      "step": 23340
    },
    {
      "epoch": 0.38206659576208785,
      "grad_norm": 0.11267620146852048,
      "learning_rate": 7.766071225905355e-06,
      "loss": 0.0043,
      "step": 23350
    },
    {
      "epoch": 0.38223022171316373,
      "grad_norm": 0.24255192026343259,
      "learning_rate": 7.763691750748119e-06,
      "loss": 0.0048,
      "step": 23360
    },
    {
      "epoch": 0.38239384766423956,
      "grad_norm": 0.09673091441195862,
      "learning_rate": 7.761311373985326e-06,
      "loss": 0.0036,
      "step": 23370
    },
    {
      "epoch": 0.3825574736153154,
      "grad_norm": 0.3279771898044794,
      "learning_rate": 7.758930096393527e-06,
      "loss": 0.0047,
      "step": 23380
    },
    {
      "epoch": 0.3827210995663912,
      "grad_norm": 0.1512847121251218,
      "learning_rate": 7.756547918749578e-06,
      "loss": 0.0041,
      "step": 23390
    },
    {
      "epoch": 0.38288472551746705,
      "grad_norm": 0.1024776299819252,
      "learning_rate": 7.754164841830618e-06,
      "loss": 0.0041,
      "step": 23400
    },
    {
      "epoch": 0.38304835146854294,
      "grad_norm": 0.3501307089138783,
      "learning_rate": 7.751780866414085e-06,
      "loss": 0.0038,
      "step": 23410
    },
    {
      "epoch": 0.38321197741961877,
      "grad_norm": 0.13892322813927876,
      "learning_rate": 7.749395993277709e-06,
      "loss": 0.0044,
      "step": 23420
    },
    {
      "epoch": 0.3833756033706946,
      "grad_norm": 0.3676640348418677,
      "learning_rate": 7.747010223199512e-06,
      "loss": 0.007,
      "step": 23430
    },
    {
      "epoch": 0.3835392293217704,
      "grad_norm": 0.13520775097314114,
      "learning_rate": 7.74462355695781e-06,
      "loss": 0.0032,
      "step": 23440
    },
    {
      "epoch": 0.38370285527284625,
      "grad_norm": 0.12508156707270715,
      "learning_rate": 7.74223599533121e-06,
      "loss": 0.006,
      "step": 23450
    },
    {
      "epoch": 0.38386648122392214,
      "grad_norm": 0.30234936687706865,
      "learning_rate": 7.739847539098614e-06,
      "loss": 0.0034,
      "step": 23460
    },
    {
      "epoch": 0.38403010717499797,
      "grad_norm": 0.2774033559512334,
      "learning_rate": 7.73745818903921e-06,
      "loss": 0.0043,
      "step": 23470
    },
    {
      "epoch": 0.3841937331260738,
      "grad_norm": 0.22625481227573882,
      "learning_rate": 7.735067945932485e-06,
      "loss": 0.0048,
      "step": 23480
    },
    {
      "epoch": 0.38435735907714963,
      "grad_norm": 0.14256361069685392,
      "learning_rate": 7.73267681055821e-06,
      "loss": 0.003,
      "step": 23490
    },
    {
      "epoch": 0.38452098502822546,
      "grad_norm": 0.25642228173895215,
      "learning_rate": 7.730284783696454e-06,
      "loss": 0.0038,
      "step": 23500
    },
    {
      "epoch": 0.38468461097930134,
      "grad_norm": 0.25134789739128394,
      "learning_rate": 7.72789186612757e-06,
      "loss": 0.0033,
      "step": 23510
    },
    {
      "epoch": 0.3848482369303772,
      "grad_norm": 0.18375583525134048,
      "learning_rate": 7.725498058632205e-06,
      "loss": 0.0086,
      "step": 23520
    },
    {
      "epoch": 0.385011862881453,
      "grad_norm": 0.32767712827444134,
      "learning_rate": 7.723103361991303e-06,
      "loss": 0.0039,
      "step": 23530
    },
    {
      "epoch": 0.38517548883252883,
      "grad_norm": 0.03978580268010153,
      "learning_rate": 7.720707776986083e-06,
      "loss": 0.0032,
      "step": 23540
    },
    {
      "epoch": 0.38533911478360466,
      "grad_norm": 0.1426115032113511,
      "learning_rate": 7.718311304398065e-06,
      "loss": 0.0056,
      "step": 23550
    },
    {
      "epoch": 0.3855027407346805,
      "grad_norm": 0.18433456232944362,
      "learning_rate": 7.71591394500906e-06,
      "loss": 0.0035,
      "step": 23560
    },
    {
      "epoch": 0.3856663666857564,
      "grad_norm": 0.23622257986749895,
      "learning_rate": 7.713515699601158e-06,
      "loss": 0.0052,
      "step": 23570
    },
    {
      "epoch": 0.3858299926368322,
      "grad_norm": 0.6796815732065643,
      "learning_rate": 7.711116568956747e-06,
      "loss": 0.0045,
      "step": 23580
    },
    {
      "epoch": 0.38599361858790804,
      "grad_norm": 0.37824039067660775,
      "learning_rate": 7.708716553858502e-06,
      "loss": 0.0036,
      "step": 23590
    },
    {
      "epoch": 0.38615724453898387,
      "grad_norm": 0.19477081148150735,
      "learning_rate": 7.706315655089383e-06,
      "loss": 0.0073,
      "step": 23600
    },
    {
      "epoch": 0.3863208704900597,
      "grad_norm": 0.17877146675746802,
      "learning_rate": 7.703913873432643e-06,
      "loss": 0.0047,
      "step": 23610
    },
    {
      "epoch": 0.3864844964411356,
      "grad_norm": 0.13598237158444526,
      "learning_rate": 7.70151120967182e-06,
      "loss": 0.0032,
      "step": 23620
    },
    {
      "epoch": 0.3866481223922114,
      "grad_norm": 0.12345505375641387,
      "learning_rate": 7.699107664590737e-06,
      "loss": 0.0028,
      "step": 23630
    },
    {
      "epoch": 0.38681174834328724,
      "grad_norm": 0.41988597366761515,
      "learning_rate": 7.696703238973516e-06,
      "loss": 0.0052,
      "step": 23640
    },
    {
      "epoch": 0.3869753742943631,
      "grad_norm": 0.1300826422769908,
      "learning_rate": 7.694297933604551e-06,
      "loss": 0.0035,
      "step": 23650
    },
    {
      "epoch": 0.3871390002454389,
      "grad_norm": 0.1656186336176813,
      "learning_rate": 7.691891749268534e-06,
      "loss": 0.0048,
      "step": 23660
    },
    {
      "epoch": 0.3873026261965148,
      "grad_norm": 0.0994984491806116,
      "learning_rate": 7.689484686750439e-06,
      "loss": 0.004,
      "step": 23670
    },
    {
      "epoch": 0.3874662521475906,
      "grad_norm": 0.056285104982103255,
      "learning_rate": 7.687076746835528e-06,
      "loss": 0.0067,
      "step": 23680
    },
    {
      "epoch": 0.38762987809866645,
      "grad_norm": 0.14547735025543546,
      "learning_rate": 7.684667930309348e-06,
      "loss": 0.0051,
      "step": 23690
    },
    {
      "epoch": 0.3877935040497423,
      "grad_norm": 0.05723893840771515,
      "learning_rate": 7.682258237957734e-06,
      "loss": 0.0035,
      "step": 23700
    },
    {
      "epoch": 0.3879571300008181,
      "grad_norm": 0.07706274094493967,
      "learning_rate": 7.679847670566805e-06,
      "loss": 0.0039,
      "step": 23710
    },
    {
      "epoch": 0.388120755951894,
      "grad_norm": 0.16908053425663697,
      "learning_rate": 7.677436228922965e-06,
      "loss": 0.0047,
      "step": 23720
    },
    {
      "epoch": 0.3882843819029698,
      "grad_norm": 0.20682704785446202,
      "learning_rate": 7.675023913812906e-06,
      "loss": 0.0037,
      "step": 23730
    },
    {
      "epoch": 0.38844800785404565,
      "grad_norm": 0.2105514257682979,
      "learning_rate": 7.672610726023602e-06,
      "loss": 0.0025,
      "step": 23740
    },
    {
      "epoch": 0.3886116338051215,
      "grad_norm": 0.08150280789864257,
      "learning_rate": 7.670196666342313e-06,
      "loss": 0.0033,
      "step": 23750
    },
    {
      "epoch": 0.3887752597561973,
      "grad_norm": 0.41261195221404373,
      "learning_rate": 7.667781735556584e-06,
      "loss": 0.0032,
      "step": 23760
    },
    {
      "epoch": 0.3889388857072732,
      "grad_norm": 0.1402550595408482,
      "learning_rate": 7.66536593445424e-06,
      "loss": 0.0058,
      "step": 23770
    },
    {
      "epoch": 0.389102511658349,
      "grad_norm": 0.4559601834077003,
      "learning_rate": 7.6629492638234e-06,
      "loss": 0.004,
      "step": 23780
    },
    {
      "epoch": 0.38926613760942486,
      "grad_norm": 0.2718020626946962,
      "learning_rate": 7.660531724452455e-06,
      "loss": 0.0046,
      "step": 23790
    },
    {
      "epoch": 0.3894297635605007,
      "grad_norm": 0.31167053727800204,
      "learning_rate": 7.658113317130085e-06,
      "loss": 0.006,
      "step": 23800
    },
    {
      "epoch": 0.3895933895115765,
      "grad_norm": 0.14162698070805388,
      "learning_rate": 7.655694042645253e-06,
      "loss": 0.0047,
      "step": 23810
    },
    {
      "epoch": 0.3897570154626524,
      "grad_norm": 0.20374436730966325,
      "learning_rate": 7.653273901787206e-06,
      "loss": 0.0073,
      "step": 23820
    },
    {
      "epoch": 0.38992064141372823,
      "grad_norm": 0.08987071896446418,
      "learning_rate": 7.650852895345469e-06,
      "loss": 0.0046,
      "step": 23830
    },
    {
      "epoch": 0.39008426736480406,
      "grad_norm": 0.17976110940763293,
      "learning_rate": 7.648431024109857e-06,
      "loss": 0.0032,
      "step": 23840
    },
    {
      "epoch": 0.3902478933158799,
      "grad_norm": 0.10488565209290589,
      "learning_rate": 7.646008288870459e-06,
      "loss": 0.0056,
      "step": 23850
    },
    {
      "epoch": 0.3904115192669557,
      "grad_norm": 0.09359887947841258,
      "learning_rate": 7.643584690417648e-06,
      "loss": 0.0041,
      "step": 23860
    },
    {
      "epoch": 0.3905751452180316,
      "grad_norm": 0.18393071032048405,
      "learning_rate": 7.641160229542088e-06,
      "loss": 0.0041,
      "step": 23870
    },
    {
      "epoch": 0.39073877116910744,
      "grad_norm": 0.49723905136469476,
      "learning_rate": 7.638734907034708e-06,
      "loss": 0.0066,
      "step": 23880
    },
    {
      "epoch": 0.39090239712018326,
      "grad_norm": 0.14714969173836634,
      "learning_rate": 7.636308723686732e-06,
      "loss": 0.0032,
      "step": 23890
    },
    {
      "epoch": 0.3910660230712591,
      "grad_norm": 0.1950743796665209,
      "learning_rate": 7.633881680289656e-06,
      "loss": 0.0037,
      "step": 23900
    },
    {
      "epoch": 0.3912296490223349,
      "grad_norm": 0.03623669447808912,
      "learning_rate": 7.631453777635261e-06,
      "loss": 0.0029,
      "step": 23910
    },
    {
      "epoch": 0.3913932749734108,
      "grad_norm": 0.21860182386702248,
      "learning_rate": 7.629025016515609e-06,
      "loss": 0.0024,
      "step": 23920
    },
    {
      "epoch": 0.39155690092448664,
      "grad_norm": 0.08400513348713573,
      "learning_rate": 7.626595397723037e-06,
      "loss": 0.0026,
      "step": 23930
    },
    {
      "epoch": 0.39172052687556247,
      "grad_norm": 0.17091518477496065,
      "learning_rate": 7.624164922050168e-06,
      "loss": 0.0048,
      "step": 23940
    },
    {
      "epoch": 0.3918841528266383,
      "grad_norm": 0.17808107860712336,
      "learning_rate": 7.6217335902899016e-06,
      "loss": 0.0052,
      "step": 23950
    },
    {
      "epoch": 0.39204777877771413,
      "grad_norm": 0.34108256477238896,
      "learning_rate": 7.619301403235415e-06,
      "loss": 0.0035,
      "step": 23960
    },
    {
      "epoch": 0.39221140472878996,
      "grad_norm": 0.7140888700959167,
      "learning_rate": 7.616868361680167e-06,
      "loss": 0.0054,
      "step": 23970
    },
    {
      "epoch": 0.39237503067986584,
      "grad_norm": 0.11399917840730914,
      "learning_rate": 7.614434466417897e-06,
      "loss": 0.0034,
      "step": 23980
    },
    {
      "epoch": 0.3925386566309417,
      "grad_norm": 0.6017156041415024,
      "learning_rate": 7.611999718242615e-06,
      "loss": 0.0056,
      "step": 23990
    },
    {
      "epoch": 0.3927022825820175,
      "grad_norm": 0.3419080837919293,
      "learning_rate": 7.6095641179486194e-06,
      "loss": 0.0038,
      "step": 24000
    },
    {
      "epoch": 0.39286590853309333,
      "grad_norm": 0.2749004027013854,
      "learning_rate": 7.607127666330478e-06,
      "loss": 0.0046,
      "step": 24010
    },
    {
      "epoch": 0.39302953448416916,
      "grad_norm": 0.28278145551575046,
      "learning_rate": 7.6046903641830425e-06,
      "loss": 0.0033,
      "step": 24020
    },
    {
      "epoch": 0.39319316043524505,
      "grad_norm": 0.14586971876666538,
      "learning_rate": 7.6022522123014395e-06,
      "loss": 0.0048,
      "step": 24030
    },
    {
      "epoch": 0.3933567863863209,
      "grad_norm": 0.1305475610924498,
      "learning_rate": 7.599813211481073e-06,
      "loss": 0.004,
      "step": 24040
    },
    {
      "epoch": 0.3935204123373967,
      "grad_norm": 0.26745455795232354,
      "learning_rate": 7.597373362517622e-06,
      "loss": 0.0047,
      "step": 24050
    },
    {
      "epoch": 0.39368403828847254,
      "grad_norm": 0.20922475271325178,
      "learning_rate": 7.594932666207047e-06,
      "loss": 0.0024,
      "step": 24060
    },
    {
      "epoch": 0.39384766423954837,
      "grad_norm": 0.18923570980855386,
      "learning_rate": 7.592491123345577e-06,
      "loss": 0.0031,
      "step": 24070
    },
    {
      "epoch": 0.39401129019062425,
      "grad_norm": 0.20786433011041838,
      "learning_rate": 7.590048734729728e-06,
      "loss": 0.0064,
      "step": 24080
    },
    {
      "epoch": 0.3941749161417001,
      "grad_norm": 0.11546911766867284,
      "learning_rate": 7.587605501156285e-06,
      "loss": 0.0054,
      "step": 24090
    },
    {
      "epoch": 0.3943385420927759,
      "grad_norm": 0.2562306657399679,
      "learning_rate": 7.5851614234223045e-06,
      "loss": 0.0023,
      "step": 24100
    },
    {
      "epoch": 0.39450216804385174,
      "grad_norm": 0.43551874151285025,
      "learning_rate": 7.5827165023251296e-06,
      "loss": 0.004,
      "step": 24110
    },
    {
      "epoch": 0.39466579399492757,
      "grad_norm": 0.17809073202210374,
      "learning_rate": 7.580270738662372e-06,
      "loss": 0.009,
      "step": 24120
    },
    {
      "epoch": 0.39482941994600346,
      "grad_norm": 0.2753493933784868,
      "learning_rate": 7.577824133231915e-06,
      "loss": 0.0032,
      "step": 24130
    },
    {
      "epoch": 0.3949930458970793,
      "grad_norm": 0.3611824089100605,
      "learning_rate": 7.575376686831923e-06,
      "loss": 0.003,
      "step": 24140
    },
    {
      "epoch": 0.3951566718481551,
      "grad_norm": 0.29567921942116165,
      "learning_rate": 7.572928400260832e-06,
      "loss": 0.0037,
      "step": 24150
    },
    {
      "epoch": 0.39532029779923095,
      "grad_norm": 0.2900818948408225,
      "learning_rate": 7.5704792743173525e-06,
      "loss": 0.0075,
      "step": 24160
    },
    {
      "epoch": 0.3954839237503068,
      "grad_norm": 0.46853705096539516,
      "learning_rate": 7.568029309800468e-06,
      "loss": 0.0033,
      "step": 24170
    },
    {
      "epoch": 0.39564754970138266,
      "grad_norm": 0.32214776195764494,
      "learning_rate": 7.565578507509437e-06,
      "loss": 0.003,
      "step": 24180
    },
    {
      "epoch": 0.3958111756524585,
      "grad_norm": 0.27679982521829827,
      "learning_rate": 7.563126868243787e-06,
      "loss": 0.0038,
      "step": 24190
    },
    {
      "epoch": 0.3959748016035343,
      "grad_norm": 0.24731296058713828,
      "learning_rate": 7.560674392803326e-06,
      "loss": 0.0035,
      "step": 24200
    },
    {
      "epoch": 0.39613842755461015,
      "grad_norm": 0.15660282467961975,
      "learning_rate": 7.55822108198813e-06,
      "loss": 0.0056,
      "step": 24210
    },
    {
      "epoch": 0.396302053505686,
      "grad_norm": 0.2343976464655349,
      "learning_rate": 7.555766936598546e-06,
      "loss": 0.0048,
      "step": 24220
    },
    {
      "epoch": 0.39646567945676187,
      "grad_norm": 0.4364698283022281,
      "learning_rate": 7.5533119574351965e-06,
      "loss": 0.0045,
      "step": 24230
    },
    {
      "epoch": 0.3966293054078377,
      "grad_norm": 0.13215060219049593,
      "learning_rate": 7.550856145298977e-06,
      "loss": 0.0041,
      "step": 24240
    },
    {
      "epoch": 0.3967929313589135,
      "grad_norm": 0.24503982076897654,
      "learning_rate": 7.5483995009910505e-06,
      "loss": 0.0033,
      "step": 24250
    },
    {
      "epoch": 0.39695655730998936,
      "grad_norm": 0.4873248907037443,
      "learning_rate": 7.545942025312855e-06,
      "loss": 0.0055,
      "step": 24260
    },
    {
      "epoch": 0.3971201832610652,
      "grad_norm": 0.17597594873880792,
      "learning_rate": 7.543483719066098e-06,
      "loss": 0.0041,
      "step": 24270
    },
    {
      "epoch": 0.39728380921214107,
      "grad_norm": 0.1718056894711159,
      "learning_rate": 7.5410245830527585e-06,
      "loss": 0.0034,
      "step": 24280
    },
    {
      "epoch": 0.3974474351632169,
      "grad_norm": 0.09169126859098309,
      "learning_rate": 7.538564618075086e-06,
      "loss": 0.0025,
      "step": 24290
    },
    {
      "epoch": 0.39761106111429273,
      "grad_norm": 0.2819883474338189,
      "learning_rate": 7.536103824935601e-06,
      "loss": 0.0038,
      "step": 24300
    },
    {
      "epoch": 0.39777468706536856,
      "grad_norm": 0.28658077709015933,
      "learning_rate": 7.533642204437093e-06,
      "loss": 0.0074,
      "step": 24310
    },
    {
      "epoch": 0.3979383130164444,
      "grad_norm": 0.4155216468552145,
      "learning_rate": 7.531179757382623e-06,
      "loss": 0.0044,
      "step": 24320
    },
    {
      "epoch": 0.3981019389675203,
      "grad_norm": 0.18270416196540462,
      "learning_rate": 7.5287164845755204e-06,
      "loss": 0.0025,
      "step": 24330
    },
    {
      "epoch": 0.3982655649185961,
      "grad_norm": 0.11653914065654808,
      "learning_rate": 7.5262523868193825e-06,
      "loss": 0.0044,
      "step": 24340
    },
    {
      "epoch": 0.39842919086967193,
      "grad_norm": 0.11482324604004897,
      "learning_rate": 7.523787464918082e-06,
      "loss": 0.0029,
      "step": 24350
    },
    {
      "epoch": 0.39859281682074776,
      "grad_norm": 0.26748661071605706,
      "learning_rate": 7.521321719675753e-06,
      "loss": 0.0051,
      "step": 24360
    },
    {
      "epoch": 0.3987564427718236,
      "grad_norm": 0.3298675199707605,
      "learning_rate": 7.5188551518968e-06,
      "loss": 0.0041,
      "step": 24370
    },
    {
      "epoch": 0.3989200687228995,
      "grad_norm": 0.14649169546178883,
      "learning_rate": 7.5163877623859015e-06,
      "loss": 0.0041,
      "step": 24380
    },
    {
      "epoch": 0.3990836946739753,
      "grad_norm": 0.23165195179763795,
      "learning_rate": 7.513919551947997e-06,
      "loss": 0.0039,
      "step": 24390
    },
    {
      "epoch": 0.39924732062505114,
      "grad_norm": 0.32030853317824903,
      "learning_rate": 7.511450521388295e-06,
      "loss": 0.0032,
      "step": 24400
    },
    {
      "epoch": 0.39941094657612697,
      "grad_norm": 0.13467004850465902,
      "learning_rate": 7.508980671512276e-06,
      "loss": 0.0022,
      "step": 24410
    },
    {
      "epoch": 0.3995745725272028,
      "grad_norm": 0.08361064432088058,
      "learning_rate": 7.506510003125682e-06,
      "loss": 0.0045,
      "step": 24420
    },
    {
      "epoch": 0.39973819847827863,
      "grad_norm": 0.2592533028169886,
      "learning_rate": 7.504038517034529e-06,
      "loss": 0.0045,
      "step": 24430
    },
    {
      "epoch": 0.3999018244293545,
      "grad_norm": 0.10463397804815186,
      "learning_rate": 7.501566214045091e-06,
      "loss": 0.0021,
      "step": 24440
    },
    {
      "epoch": 0.40006545038043034,
      "grad_norm": 0.09251492135273492,
      "learning_rate": 7.499093094963916e-06,
      "loss": 0.0041,
      "step": 24450
    },
    {
      "epoch": 0.4002290763315062,
      "grad_norm": 0.217362914402295,
      "learning_rate": 7.496619160597814e-06,
      "loss": 0.0036,
      "step": 24460
    },
    {
      "epoch": 0.400392702282582,
      "grad_norm": 0.17417262821399385,
      "learning_rate": 7.494144411753863e-06,
      "loss": 0.0029,
      "step": 24470
    },
    {
      "epoch": 0.40055632823365783,
      "grad_norm": 0.14374684981023111,
      "learning_rate": 7.491668849239408e-06,
      "loss": 0.0038,
      "step": 24480
    },
    {
      "epoch": 0.4007199541847337,
      "grad_norm": 0.1038876909313206,
      "learning_rate": 7.489192473862053e-06,
      "loss": 0.005,
      "step": 24490
    },
    {
      "epoch": 0.40088358013580955,
      "grad_norm": 0.20241202462472102,
      "learning_rate": 7.486715286429674e-06,
      "loss": 0.0036,
      "step": 24500
    },
    {
      "epoch": 0.4010472060868854,
      "grad_norm": 0.08564381258566638,
      "learning_rate": 7.484237287750411e-06,
      "loss": 0.0039,
      "step": 24510
    },
    {
      "epoch": 0.4012108320379612,
      "grad_norm": 0.1384282979778313,
      "learning_rate": 7.4817584786326655e-06,
      "loss": 0.0033,
      "step": 24520
    },
    {
      "epoch": 0.40137445798903704,
      "grad_norm": 0.14918851870267386,
      "learning_rate": 7.4792788598851056e-06,
      "loss": 0.0036,
      "step": 24530
    },
    {
      "epoch": 0.4015380839401129,
      "grad_norm": 0.3912841325922538,
      "learning_rate": 7.476798432316663e-06,
      "loss": 0.0071,
      "step": 24540
    },
    {
      "epoch": 0.40170170989118875,
      "grad_norm": 0.10517057131657231,
      "learning_rate": 7.474317196736534e-06,
      "loss": 0.0046,
      "step": 24550
    },
    {
      "epoch": 0.4018653358422646,
      "grad_norm": 0.438422896905194,
      "learning_rate": 7.471835153954176e-06,
      "loss": 0.0056,
      "step": 24560
    },
    {
      "epoch": 0.4020289617933404,
      "grad_norm": 0.9635172988054806,
      "learning_rate": 7.469352304779314e-06,
      "loss": 0.004,
      "step": 24570
    },
    {
      "epoch": 0.40219258774441624,
      "grad_norm": 0.250769087968856,
      "learning_rate": 7.466868650021932e-06,
      "loss": 0.0042,
      "step": 24580
    },
    {
      "epoch": 0.4023562136954921,
      "grad_norm": 0.42790683545976993,
      "learning_rate": 7.4643841904922774e-06,
      "loss": 0.0035,
      "step": 24590
    },
    {
      "epoch": 0.40251983964656796,
      "grad_norm": 0.5465562104007252,
      "learning_rate": 7.4618989270008645e-06,
      "loss": 0.005,
      "step": 24600
    },
    {
      "epoch": 0.4026834655976438,
      "grad_norm": 0.07521216493912247,
      "learning_rate": 7.459412860358465e-06,
      "loss": 0.0042,
      "step": 24610
    },
    {
      "epoch": 0.4028470915487196,
      "grad_norm": 0.3394244108728312,
      "learning_rate": 7.456925991376112e-06,
      "loss": 0.0016,
      "step": 24620
    },
    {
      "epoch": 0.40301071749979545,
      "grad_norm": 0.16124652601557224,
      "learning_rate": 7.454438320865105e-06,
      "loss": 0.0029,
      "step": 24630
    },
    {
      "epoch": 0.40317434345087133,
      "grad_norm": 0.3146371089502906,
      "learning_rate": 7.451949849637003e-06,
      "loss": 0.0062,
      "step": 24640
    },
    {
      "epoch": 0.40333796940194716,
      "grad_norm": 0.15917060775104497,
      "learning_rate": 7.449460578503623e-06,
      "loss": 0.0045,
      "step": 24650
    },
    {
      "epoch": 0.403501595353023,
      "grad_norm": 0.23178933729667187,
      "learning_rate": 7.446970508277049e-06,
      "loss": 0.0041,
      "step": 24660
    },
    {
      "epoch": 0.4036652213040988,
      "grad_norm": 0.12327364502342168,
      "learning_rate": 7.444479639769619e-06,
      "loss": 0.0059,
      "step": 24670
    },
    {
      "epoch": 0.40382884725517465,
      "grad_norm": 0.21301808310173742,
      "learning_rate": 7.441987973793938e-06,
      "loss": 0.0037,
      "step": 24680
    },
    {
      "epoch": 0.40399247320625054,
      "grad_norm": 0.17798293986555608,
      "learning_rate": 7.4394955111628675e-06,
      "loss": 0.0043,
      "step": 24690
    },
    {
      "epoch": 0.40415609915732637,
      "grad_norm": 0.1242754693210632,
      "learning_rate": 7.437002252689526e-06,
      "loss": 0.0042,
      "step": 24700
    },
    {
      "epoch": 0.4043197251084022,
      "grad_norm": 0.2822948963275418,
      "learning_rate": 7.434508199187297e-06,
      "loss": 0.0039,
      "step": 24710
    },
    {
      "epoch": 0.404483351059478,
      "grad_norm": 0.12324009229018375,
      "learning_rate": 7.432013351469825e-06,
      "loss": 0.0032,
      "step": 24720
    },
    {
      "epoch": 0.40464697701055385,
      "grad_norm": 0.27731210759245295,
      "learning_rate": 7.429517710351003e-06,
      "loss": 0.0036,
      "step": 24730
    },
    {
      "epoch": 0.40481060296162974,
      "grad_norm": 0.11191213090405355,
      "learning_rate": 7.427021276644994e-06,
      "loss": 0.0043,
      "step": 24740
    },
    {
      "epoch": 0.40497422891270557,
      "grad_norm": 0.2670990291241771,
      "learning_rate": 7.424524051166217e-06,
      "loss": 0.0046,
      "step": 24750
    },
    {
      "epoch": 0.4051378548637814,
      "grad_norm": 0.2289976824532421,
      "learning_rate": 7.4220260347293416e-06,
      "loss": 0.0043,
      "step": 24760
    },
    {
      "epoch": 0.40530148081485723,
      "grad_norm": 0.24994441863450337,
      "learning_rate": 7.419527228149308e-06,
      "loss": 0.0033,
      "step": 24770
    },
    {
      "epoch": 0.40546510676593306,
      "grad_norm": 0.1885125850723168,
      "learning_rate": 7.417027632241305e-06,
      "loss": 0.0042,
      "step": 24780
    },
    {
      "epoch": 0.40562873271700894,
      "grad_norm": 0.1365113928687538,
      "learning_rate": 7.414527247820779e-06,
      "loss": 0.0048,
      "step": 24790
    },
    {
      "epoch": 0.4057923586680848,
      "grad_norm": 0.31691579356917443,
      "learning_rate": 7.412026075703442e-06,
      "loss": 0.0036,
      "step": 24800
    },
    {
      "epoch": 0.4059559846191606,
      "grad_norm": 0.2988252530650168,
      "learning_rate": 7.409524116705252e-06,
      "loss": 0.0027,
      "step": 24810
    },
    {
      "epoch": 0.40611961057023643,
      "grad_norm": 0.21014786417115097,
      "learning_rate": 7.407021371642431e-06,
      "loss": 0.0033,
      "step": 24820
    },
    {
      "epoch": 0.40628323652131226,
      "grad_norm": 0.3467988559851425,
      "learning_rate": 7.404517841331456e-06,
      "loss": 0.0064,
      "step": 24830
    },
    {
      "epoch": 0.40644686247238815,
      "grad_norm": 0.21297141036119263,
      "learning_rate": 7.402013526589057e-06,
      "loss": 0.005,
      "step": 24840
    },
    {
      "epoch": 0.406610488423464,
      "grad_norm": 0.27899888130643274,
      "learning_rate": 7.399508428232224e-06,
      "loss": 0.0026,
      "step": 24850
    },
    {
      "epoch": 0.4067741143745398,
      "grad_norm": 0.10119414622072359,
      "learning_rate": 7.397002547078201e-06,
      "loss": 0.0031,
      "step": 24860
    },
    {
      "epoch": 0.40693774032561564,
      "grad_norm": 0.4926613697071245,
      "learning_rate": 7.394495883944488e-06,
      "loss": 0.0042,
      "step": 24870
    },
    {
      "epoch": 0.40710136627669147,
      "grad_norm": 0.09518737309021089,
      "learning_rate": 7.391988439648838e-06,
      "loss": 0.0037,
      "step": 24880
    },
    {
      "epoch": 0.4072649922277673,
      "grad_norm": 0.05776711379574887,
      "learning_rate": 7.389480215009262e-06,
      "loss": 0.0047,
      "step": 24890
    },
    {
      "epoch": 0.4074286181788432,
      "grad_norm": 0.19075215755369676,
      "learning_rate": 7.386971210844025e-06,
      "loss": 0.0028,
      "step": 24900
    },
    {
      "epoch": 0.407592244129919,
      "grad_norm": 0.18034706335636252,
      "learning_rate": 7.384461427971641e-06,
      "loss": 0.0026,
      "step": 24910
    },
    {
      "epoch": 0.40775587008099484,
      "grad_norm": 0.22716767124497111,
      "learning_rate": 7.381950867210885e-06,
      "loss": 0.0035,
      "step": 24920
    },
    {
      "epoch": 0.40791949603207067,
      "grad_norm": 0.1408005002183599,
      "learning_rate": 7.379439529380783e-06,
      "loss": 0.0048,
      "step": 24930
    },
    {
      "epoch": 0.4080831219831465,
      "grad_norm": 0.30906575390049795,
      "learning_rate": 7.376927415300614e-06,
      "loss": 0.0041,
      "step": 24940
    },
    {
      "epoch": 0.4082467479342224,
      "grad_norm": 0.21051147143743668,
      "learning_rate": 7.374414525789912e-06,
      "loss": 0.0049,
      "step": 24950
    },
    {
      "epoch": 0.4084103738852982,
      "grad_norm": 0.1676996881952094,
      "learning_rate": 7.371900861668461e-06,
      "loss": 0.0046,
      "step": 24960
    },
    {
      "epoch": 0.40857399983637405,
      "grad_norm": 0.14949044401428832,
      "learning_rate": 7.369386423756301e-06,
      "loss": 0.0059,
      "step": 24970
    },
    {
      "epoch": 0.4087376257874499,
      "grad_norm": 0.23799930117179977,
      "learning_rate": 7.366871212873721e-06,
      "loss": 0.0055,
      "step": 24980
    },
    {
      "epoch": 0.4089012517385257,
      "grad_norm": 0.15534322283996277,
      "learning_rate": 7.364355229841267e-06,
      "loss": 0.0035,
      "step": 24990
    },
    {
      "epoch": 0.4090648776896016,
      "grad_norm": 0.1295135965334056,
      "learning_rate": 7.361838475479731e-06,
      "loss": 0.0034,
      "step": 25000
    },
    {
      "epoch": 0.4090648776896016,
      "eval_loss": 0.0027813350316137075,
      "eval_runtime": 7.9267,
      "eval_samples_per_second": 25.231,
      "eval_steps_per_second": 6.308,
      "step": 25000
    },
    {
      "epoch": 0.4092285036406774,
      "grad_norm": 0.16366820602490878,
      "learning_rate": 7.35932095061016e-06,
      "loss": 0.0061,
      "step": 25010
    },
    {
      "epoch": 0.40939212959175325,
      "grad_norm": 0.11493003572652542,
      "learning_rate": 7.356802656053854e-06,
      "loss": 0.0037,
      "step": 25020
    },
    {
      "epoch": 0.4095557555428291,
      "grad_norm": 0.22172278090165354,
      "learning_rate": 7.354283592632362e-06,
      "loss": 0.0037,
      "step": 25030
    },
    {
      "epoch": 0.4097193814939049,
      "grad_norm": 0.09709497709916562,
      "learning_rate": 7.351763761167481e-06,
      "loss": 0.006,
      "step": 25040
    },
    {
      "epoch": 0.4098830074449808,
      "grad_norm": 0.28601669696342946,
      "learning_rate": 7.349243162481264e-06,
      "loss": 0.0045,
      "step": 25050
    },
    {
      "epoch": 0.4100466333960566,
      "grad_norm": 0.09213507453239303,
      "learning_rate": 7.346721797396011e-06,
      "loss": 0.0056,
      "step": 25060
    },
    {
      "epoch": 0.41021025934713246,
      "grad_norm": 0.022767298901006292,
      "learning_rate": 7.344199666734272e-06,
      "loss": 0.0026,
      "step": 25070
    },
    {
      "epoch": 0.4103738852982083,
      "grad_norm": 0.44932751718055636,
      "learning_rate": 7.3416767713188495e-06,
      "loss": 0.0046,
      "step": 25080
    },
    {
      "epoch": 0.4105375112492841,
      "grad_norm": 0.05508288098749318,
      "learning_rate": 7.339153111972792e-06,
      "loss": 0.0034,
      "step": 25090
    },
    {
      "epoch": 0.41070113720036,
      "grad_norm": 0.1993481578208381,
      "learning_rate": 7.3366286895193985e-06,
      "loss": 0.0041,
      "step": 25100
    },
    {
      "epoch": 0.41086476315143583,
      "grad_norm": 0.153656181469206,
      "learning_rate": 7.334103504782217e-06,
      "loss": 0.0041,
      "step": 25110
    },
    {
      "epoch": 0.41102838910251166,
      "grad_norm": 0.15019282274719914,
      "learning_rate": 7.331577558585047e-06,
      "loss": 0.0044,
      "step": 25120
    },
    {
      "epoch": 0.4111920150535875,
      "grad_norm": 0.25499043726316334,
      "learning_rate": 7.32905085175193e-06,
      "loss": 0.0027,
      "step": 25130
    },
    {
      "epoch": 0.4113556410046633,
      "grad_norm": 0.08621914453850096,
      "learning_rate": 7.326523385107163e-06,
      "loss": 0.0064,
      "step": 25140
    },
    {
      "epoch": 0.4115192669557392,
      "grad_norm": 0.26592603663902625,
      "learning_rate": 7.323995159475284e-06,
      "loss": 0.0037,
      "step": 25150
    },
    {
      "epoch": 0.41168289290681503,
      "grad_norm": 0.5372821929579726,
      "learning_rate": 7.321466175681085e-06,
      "loss": 0.0042,
      "step": 25160
    },
    {
      "epoch": 0.41184651885789086,
      "grad_norm": 0.3324393080286242,
      "learning_rate": 7.318936434549599e-06,
      "loss": 0.007,
      "step": 25170
    },
    {
      "epoch": 0.4120101448089667,
      "grad_norm": 0.22921505891319766,
      "learning_rate": 7.316405936906112e-06,
      "loss": 0.0046,
      "step": 25180
    },
    {
      "epoch": 0.4121737707600425,
      "grad_norm": 0.12725402319379356,
      "learning_rate": 7.313874683576152e-06,
      "loss": 0.0037,
      "step": 25190
    },
    {
      "epoch": 0.4123373967111184,
      "grad_norm": 0.09153486889125334,
      "learning_rate": 7.311342675385499e-06,
      "loss": 0.0067,
      "step": 25200
    },
    {
      "epoch": 0.41250102266219424,
      "grad_norm": 0.10998413633811534,
      "learning_rate": 7.308809913160172e-06,
      "loss": 0.006,
      "step": 25210
    },
    {
      "epoch": 0.41266464861327007,
      "grad_norm": 0.1389045403795814,
      "learning_rate": 7.306276397726441e-06,
      "loss": 0.0028,
      "step": 25220
    },
    {
      "epoch": 0.4128282745643459,
      "grad_norm": 0.3011361382987698,
      "learning_rate": 7.303742129910822e-06,
      "loss": 0.0057,
      "step": 25230
    },
    {
      "epoch": 0.41299190051542173,
      "grad_norm": 0.1288056882596583,
      "learning_rate": 7.301207110540075e-06,
      "loss": 0.0028,
      "step": 25240
    },
    {
      "epoch": 0.4131555264664976,
      "grad_norm": 0.07245430049306827,
      "learning_rate": 7.298671340441203e-06,
      "loss": 0.0039,
      "step": 25250
    },
    {
      "epoch": 0.41331915241757344,
      "grad_norm": 0.19863024664792595,
      "learning_rate": 7.296134820441458e-06,
      "loss": 0.0042,
      "step": 25260
    },
    {
      "epoch": 0.4134827783686493,
      "grad_norm": 0.32841361805311275,
      "learning_rate": 7.293597551368336e-06,
      "loss": 0.0027,
      "step": 25270
    },
    {
      "epoch": 0.4136464043197251,
      "grad_norm": 0.3205918095557655,
      "learning_rate": 7.291059534049573e-06,
      "loss": 0.0044,
      "step": 25280
    },
    {
      "epoch": 0.41381003027080093,
      "grad_norm": 0.2735797336259977,
      "learning_rate": 7.288520769313154e-06,
      "loss": 0.004,
      "step": 25290
    },
    {
      "epoch": 0.41397365622187676,
      "grad_norm": 0.2548705959056739,
      "learning_rate": 7.2859812579873044e-06,
      "loss": 0.0039,
      "step": 25300
    },
    {
      "epoch": 0.41413728217295265,
      "grad_norm": 0.23120724601638037,
      "learning_rate": 7.283441000900499e-06,
      "loss": 0.0043,
      "step": 25310
    },
    {
      "epoch": 0.4143009081240285,
      "grad_norm": 0.2588090059982368,
      "learning_rate": 7.280899998881448e-06,
      "loss": 0.0053,
      "step": 25320
    },
    {
      "epoch": 0.4144645340751043,
      "grad_norm": 0.11638425555277944,
      "learning_rate": 7.278358252759108e-06,
      "loss": 0.0065,
      "step": 25330
    },
    {
      "epoch": 0.41462816002618014,
      "grad_norm": 0.44621288665326675,
      "learning_rate": 7.275815763362681e-06,
      "loss": 0.0053,
      "step": 25340
    },
    {
      "epoch": 0.41479178597725597,
      "grad_norm": 0.056298070020892176,
      "learning_rate": 7.27327253152161e-06,
      "loss": 0.0036,
      "step": 25350
    },
    {
      "epoch": 0.41495541192833185,
      "grad_norm": 0.07306393984173162,
      "learning_rate": 7.270728558065577e-06,
      "loss": 0.0035,
      "step": 25360
    },
    {
      "epoch": 0.4151190378794077,
      "grad_norm": 0.08300442543002406,
      "learning_rate": 7.268183843824508e-06,
      "loss": 0.0046,
      "step": 25370
    },
    {
      "epoch": 0.4152826638304835,
      "grad_norm": 0.13002235276868931,
      "learning_rate": 7.265638389628573e-06,
      "loss": 0.0037,
      "step": 25380
    },
    {
      "epoch": 0.41544628978155934,
      "grad_norm": 0.3929847956837395,
      "learning_rate": 7.263092196308181e-06,
      "loss": 0.0042,
      "step": 25390
    },
    {
      "epoch": 0.41560991573263517,
      "grad_norm": 0.059800544730971614,
      "learning_rate": 7.260545264693981e-06,
      "loss": 0.003,
      "step": 25400
    },
    {
      "epoch": 0.41577354168371106,
      "grad_norm": 0.07178877133618652,
      "learning_rate": 7.257997595616868e-06,
      "loss": 0.0043,
      "step": 25410
    },
    {
      "epoch": 0.4159371676347869,
      "grad_norm": 0.1400546119060847,
      "learning_rate": 7.255449189907969e-06,
      "loss": 0.0041,
      "step": 25420
    },
    {
      "epoch": 0.4161007935858627,
      "grad_norm": 0.2590985389353461,
      "learning_rate": 7.252900048398659e-06,
      "loss": 0.0031,
      "step": 25430
    },
    {
      "epoch": 0.41626441953693855,
      "grad_norm": 0.2879370643738906,
      "learning_rate": 7.250350171920552e-06,
      "loss": 0.0049,
      "step": 25440
    },
    {
      "epoch": 0.4164280454880144,
      "grad_norm": 0.19251311708620675,
      "learning_rate": 7.2477995613054976e-06,
      "loss": 0.0048,
      "step": 25450
    },
    {
      "epoch": 0.41659167143909026,
      "grad_norm": 0.04774703019911711,
      "learning_rate": 7.245248217385587e-06,
      "loss": 0.0047,
      "step": 25460
    },
    {
      "epoch": 0.4167552973901661,
      "grad_norm": 0.15131050239706828,
      "learning_rate": 7.242696140993152e-06,
      "loss": 0.0039,
      "step": 25470
    },
    {
      "epoch": 0.4169189233412419,
      "grad_norm": 0.1304384226187665,
      "learning_rate": 7.240143332960763e-06,
      "loss": 0.0026,
      "step": 25480
    },
    {
      "epoch": 0.41708254929231775,
      "grad_norm": 0.3406655896680717,
      "learning_rate": 7.237589794121228e-06,
      "loss": 0.0045,
      "step": 25490
    },
    {
      "epoch": 0.4172461752433936,
      "grad_norm": 0.2754758331290669,
      "learning_rate": 7.235035525307594e-06,
      "loss": 0.0047,
      "step": 25500
    },
    {
      "epoch": 0.41740980119446947,
      "grad_norm": 0.12593022257383235,
      "learning_rate": 7.232480527353146e-06,
      "loss": 0.0035,
      "step": 25510
    },
    {
      "epoch": 0.4175734271455453,
      "grad_norm": 0.08595491025536435,
      "learning_rate": 7.2299248010914055e-06,
      "loss": 0.0036,
      "step": 25520
    },
    {
      "epoch": 0.4177370530966211,
      "grad_norm": 0.440752191744352,
      "learning_rate": 7.227368347356136e-06,
      "loss": 0.0036,
      "step": 25530
    },
    {
      "epoch": 0.41790067904769695,
      "grad_norm": 0.1741835725824583,
      "learning_rate": 7.224811166981333e-06,
      "loss": 0.004,
      "step": 25540
    },
    {
      "epoch": 0.4180643049987728,
      "grad_norm": 0.1183051500251292,
      "learning_rate": 7.222253260801232e-06,
      "loss": 0.0042,
      "step": 25550
    },
    {
      "epoch": 0.41822793094984867,
      "grad_norm": 0.29602314694956017,
      "learning_rate": 7.219694629650305e-06,
      "loss": 0.0038,
      "step": 25560
    },
    {
      "epoch": 0.4183915569009245,
      "grad_norm": 0.6777188826272532,
      "learning_rate": 7.217135274363259e-06,
      "loss": 0.0031,
      "step": 25570
    },
    {
      "epoch": 0.41855518285200033,
      "grad_norm": 0.07905566450227594,
      "learning_rate": 7.214575195775039e-06,
      "loss": 0.0041,
      "step": 25580
    },
    {
      "epoch": 0.41871880880307616,
      "grad_norm": 0.05497533982981633,
      "learning_rate": 7.212014394720827e-06,
      "loss": 0.0047,
      "step": 25590
    },
    {
      "epoch": 0.418882434754152,
      "grad_norm": 0.34419156132658113,
      "learning_rate": 7.209452872036036e-06,
      "loss": 0.0082,
      "step": 25600
    },
    {
      "epoch": 0.4190460607052279,
      "grad_norm": 0.12328705480874431,
      "learning_rate": 7.206890628556319e-06,
      "loss": 0.0029,
      "step": 25610
    },
    {
      "epoch": 0.4192096866563037,
      "grad_norm": 0.3380323392210668,
      "learning_rate": 7.2043276651175644e-06,
      "loss": 0.0067,
      "step": 25620
    },
    {
      "epoch": 0.41937331260737953,
      "grad_norm": 0.2995823746266645,
      "learning_rate": 7.201763982555892e-06,
      "loss": 0.0045,
      "step": 25630
    },
    {
      "epoch": 0.41953693855845536,
      "grad_norm": 0.10651008665162544,
      "learning_rate": 7.199199581707655e-06,
      "loss": 0.0039,
      "step": 25640
    },
    {
      "epoch": 0.4197005645095312,
      "grad_norm": 0.16919199590973988,
      "learning_rate": 7.196634463409449e-06,
      "loss": 0.0027,
      "step": 25650
    },
    {
      "epoch": 0.4198641904606071,
      "grad_norm": 0.18006726829413203,
      "learning_rate": 7.1940686284980964e-06,
      "loss": 0.0036,
      "step": 25660
    },
    {
      "epoch": 0.4200278164116829,
      "grad_norm": 0.13336002912538997,
      "learning_rate": 7.191502077810655e-06,
      "loss": 0.0051,
      "step": 25670
    },
    {
      "epoch": 0.42019144236275874,
      "grad_norm": 0.1484632223575381,
      "learning_rate": 7.188934812184416e-06,
      "loss": 0.0032,
      "step": 25680
    },
    {
      "epoch": 0.42035506831383457,
      "grad_norm": 0.13638726658727227,
      "learning_rate": 7.186366832456906e-06,
      "loss": 0.004,
      "step": 25690
    },
    {
      "epoch": 0.4205186942649104,
      "grad_norm": 0.0875012743478111,
      "learning_rate": 7.1837981394658815e-06,
      "loss": 0.0026,
      "step": 25700
    },
    {
      "epoch": 0.4206823202159863,
      "grad_norm": 0.10142134174183257,
      "learning_rate": 7.181228734049335e-06,
      "loss": 0.004,
      "step": 25710
    },
    {
      "epoch": 0.4208459461670621,
      "grad_norm": 0.23113478252167108,
      "learning_rate": 7.178658617045489e-06,
      "loss": 0.0028,
      "step": 25720
    },
    {
      "epoch": 0.42100957211813794,
      "grad_norm": 0.28072939840163075,
      "learning_rate": 7.1760877892927975e-06,
      "loss": 0.0043,
      "step": 25730
    },
    {
      "epoch": 0.42117319806921377,
      "grad_norm": 0.2424052598356967,
      "learning_rate": 7.173516251629946e-06,
      "loss": 0.0023,
      "step": 25740
    },
    {
      "epoch": 0.4213368240202896,
      "grad_norm": 0.23369629464094502,
      "learning_rate": 7.170944004895858e-06,
      "loss": 0.0031,
      "step": 25750
    },
    {
      "epoch": 0.42150044997136543,
      "grad_norm": 0.33863423999794695,
      "learning_rate": 7.168371049929681e-06,
      "loss": 0.0052,
      "step": 25760
    },
    {
      "epoch": 0.4216640759224413,
      "grad_norm": 0.3606761212599277,
      "learning_rate": 7.165797387570794e-06,
      "loss": 0.0037,
      "step": 25770
    },
    {
      "epoch": 0.42182770187351715,
      "grad_norm": 0.15759483747548642,
      "learning_rate": 7.1632230186588135e-06,
      "loss": 0.004,
      "step": 25780
    },
    {
      "epoch": 0.421991327824593,
      "grad_norm": 0.13523477786409308,
      "learning_rate": 7.160647944033579e-06,
      "loss": 0.004,
      "step": 25790
    },
    {
      "epoch": 0.4221549537756688,
      "grad_norm": 0.14815362536092092,
      "learning_rate": 7.15807216453516e-06,
      "loss": 0.0049,
      "step": 25800
    },
    {
      "epoch": 0.42231857972674464,
      "grad_norm": 0.2747708770203793,
      "learning_rate": 7.155495681003865e-06,
      "loss": 0.0037,
      "step": 25810
    },
    {
      "epoch": 0.4224822056778205,
      "grad_norm": 0.08811620193173504,
      "learning_rate": 7.152918494280223e-06,
      "loss": 0.0031,
      "step": 25820
    },
    {
      "epoch": 0.42264583162889635,
      "grad_norm": 0.2074250495338809,
      "learning_rate": 7.150340605204995e-06,
      "loss": 0.0034,
      "step": 25830
    },
    {
      "epoch": 0.4228094575799722,
      "grad_norm": 0.07747833596524517,
      "learning_rate": 7.147762014619175e-06,
      "loss": 0.0042,
      "step": 25840
    },
    {
      "epoch": 0.422973083531048,
      "grad_norm": 0.2780726539107247,
      "learning_rate": 7.145182723363979e-06,
      "loss": 0.005,
      "step": 25850
    },
    {
      "epoch": 0.42313670948212384,
      "grad_norm": 0.13958914640352982,
      "learning_rate": 7.142602732280855e-06,
      "loss": 0.003,
      "step": 25860
    },
    {
      "epoch": 0.4233003354331997,
      "grad_norm": 0.3234123600236006,
      "learning_rate": 7.140022042211483e-06,
      "loss": 0.0042,
      "step": 25870
    },
    {
      "epoch": 0.42346396138427556,
      "grad_norm": 0.15880185611290065,
      "learning_rate": 7.1374406539977655e-06,
      "loss": 0.0034,
      "step": 25880
    },
    {
      "epoch": 0.4236275873353514,
      "grad_norm": 0.11649412886238553,
      "learning_rate": 7.1348585684818336e-06,
      "loss": 0.0026,
      "step": 25890
    },
    {
      "epoch": 0.4237912132864272,
      "grad_norm": 0.26888195828160083,
      "learning_rate": 7.13227578650605e-06,
      "loss": 0.0039,
      "step": 25900
    },
    {
      "epoch": 0.42395483923750305,
      "grad_norm": 0.1690432687754525,
      "learning_rate": 7.1296923089129975e-06,
      "loss": 0.0042,
      "step": 25910
    },
    {
      "epoch": 0.42411846518857893,
      "grad_norm": 0.16282231202630046,
      "learning_rate": 7.127108136545493e-06,
      "loss": 0.0041,
      "step": 25920
    },
    {
      "epoch": 0.42428209113965476,
      "grad_norm": 0.31043797341011037,
      "learning_rate": 7.124523270246577e-06,
      "loss": 0.0043,
      "step": 25930
    },
    {
      "epoch": 0.4244457170907306,
      "grad_norm": 0.07987518740001981,
      "learning_rate": 7.121937710859514e-06,
      "loss": 0.0051,
      "step": 25940
    },
    {
      "epoch": 0.4246093430418064,
      "grad_norm": 0.03108273118593692,
      "learning_rate": 7.119351459227801e-06,
      "loss": 0.0021,
      "step": 25950
    },
    {
      "epoch": 0.42477296899288225,
      "grad_norm": 0.2657725790529966,
      "learning_rate": 7.116764516195154e-06,
      "loss": 0.0041,
      "step": 25960
    },
    {
      "epoch": 0.42493659494395813,
      "grad_norm": 0.25190590215514863,
      "learning_rate": 7.114176882605517e-06,
      "loss": 0.0034,
      "step": 25970
    },
    {
      "epoch": 0.42510022089503396,
      "grad_norm": 0.28143789275703346,
      "learning_rate": 7.111588559303062e-06,
      "loss": 0.0031,
      "step": 25980
    },
    {
      "epoch": 0.4252638468461098,
      "grad_norm": 0.14686164813988473,
      "learning_rate": 7.108999547132181e-06,
      "loss": 0.0026,
      "step": 25990
    },
    {
      "epoch": 0.4254274727971856,
      "grad_norm": 0.3997855765683396,
      "learning_rate": 7.106409846937495e-06,
      "loss": 0.0039,
      "step": 26000
    },
    {
      "epoch": 0.42559109874826145,
      "grad_norm": 0.12174768720184286,
      "learning_rate": 7.103819459563849e-06,
      "loss": 0.0059,
      "step": 26010
    },
    {
      "epoch": 0.42575472469933734,
      "grad_norm": 0.3043690654399745,
      "learning_rate": 7.1012283858563095e-06,
      "loss": 0.0054,
      "step": 26020
    },
    {
      "epoch": 0.42591835065041317,
      "grad_norm": 0.3010284748838983,
      "learning_rate": 7.098636626660167e-06,
      "loss": 0.004,
      "step": 26030
    },
    {
      "epoch": 0.426081976601489,
      "grad_norm": 0.1536985377324005,
      "learning_rate": 7.09604418282094e-06,
      "loss": 0.0026,
      "step": 26040
    },
    {
      "epoch": 0.42624560255256483,
      "grad_norm": 0.2598214488887838,
      "learning_rate": 7.0934510551843685e-06,
      "loss": 0.0047,
      "step": 26050
    },
    {
      "epoch": 0.42640922850364066,
      "grad_norm": 0.08222984498904659,
      "learning_rate": 7.09085724459641e-06,
      "loss": 0.0027,
      "step": 26060
    },
    {
      "epoch": 0.42657285445471654,
      "grad_norm": 0.2771049633768196,
      "learning_rate": 7.088262751903252e-06,
      "loss": 0.0034,
      "step": 26070
    },
    {
      "epoch": 0.4267364804057924,
      "grad_norm": 0.319512606286483,
      "learning_rate": 7.085667577951302e-06,
      "loss": 0.005,
      "step": 26080
    },
    {
      "epoch": 0.4269001063568682,
      "grad_norm": 0.07960582866559789,
      "learning_rate": 7.083071723587187e-06,
      "loss": 0.005,
      "step": 26090
    },
    {
      "epoch": 0.42706373230794403,
      "grad_norm": 0.11816912444471785,
      "learning_rate": 7.080475189657764e-06,
      "loss": 0.0041,
      "step": 26100
    },
    {
      "epoch": 0.42722735825901986,
      "grad_norm": 0.21320226479503496,
      "learning_rate": 7.077877977010101e-06,
      "loss": 0.0029,
      "step": 26110
    },
    {
      "epoch": 0.42739098421009575,
      "grad_norm": 0.21308824686181474,
      "learning_rate": 7.075280086491495e-06,
      "loss": 0.0056,
      "step": 26120
    },
    {
      "epoch": 0.4275546101611716,
      "grad_norm": 0.37611247773284,
      "learning_rate": 7.072681518949462e-06,
      "loss": 0.0045,
      "step": 26130
    },
    {
      "epoch": 0.4277182361122474,
      "grad_norm": 0.23094808106785786,
      "learning_rate": 7.0700822752317385e-06,
      "loss": 0.0047,
      "step": 26140
    },
    {
      "epoch": 0.42788186206332324,
      "grad_norm": 0.14097639312914642,
      "learning_rate": 7.067482356186282e-06,
      "loss": 0.002,
      "step": 26150
    },
    {
      "epoch": 0.42804548801439907,
      "grad_norm": 0.16980402129117889,
      "learning_rate": 7.064881762661269e-06,
      "loss": 0.0037,
      "step": 26160
    },
    {
      "epoch": 0.4282091139654749,
      "grad_norm": 0.2767072071028258,
      "learning_rate": 7.062280495505097e-06,
      "loss": 0.0059,
      "step": 26170
    },
    {
      "epoch": 0.4283727399165508,
      "grad_norm": 0.10576538767518015,
      "learning_rate": 7.059678555566385e-06,
      "loss": 0.0033,
      "step": 26180
    },
    {
      "epoch": 0.4285363658676266,
      "grad_norm": 0.7782760945480418,
      "learning_rate": 7.057075943693969e-06,
      "loss": 0.0036,
      "step": 26190
    },
    {
      "epoch": 0.42869999181870244,
      "grad_norm": 0.059320742372660176,
      "learning_rate": 7.054472660736906e-06,
      "loss": 0.0039,
      "step": 26200
    },
    {
      "epoch": 0.42886361776977827,
      "grad_norm": 0.3857469899120441,
      "learning_rate": 7.051868707544469e-06,
      "loss": 0.0036,
      "step": 26210
    },
    {
      "epoch": 0.4290272437208541,
      "grad_norm": 0.11383051411665328,
      "learning_rate": 7.049264084966153e-06,
      "loss": 0.0046,
      "step": 26220
    },
    {
      "epoch": 0.42919086967193,
      "grad_norm": 0.10196668250279828,
      "learning_rate": 7.046658793851671e-06,
      "loss": 0.0028,
      "step": 26230
    },
    {
      "epoch": 0.4293544956230058,
      "grad_norm": 0.2820621453676008,
      "learning_rate": 7.044052835050952e-06,
      "loss": 0.0029,
      "step": 26240
    },
    {
      "epoch": 0.42951812157408165,
      "grad_norm": 0.1523478188072128,
      "learning_rate": 7.041446209414143e-06,
      "loss": 0.0045,
      "step": 26250
    },
    {
      "epoch": 0.4296817475251575,
      "grad_norm": 0.3534594705835794,
      "learning_rate": 7.0388389177916115e-06,
      "loss": 0.0062,
      "step": 26260
    },
    {
      "epoch": 0.4298453734762333,
      "grad_norm": 0.2071102533757351,
      "learning_rate": 7.036230961033939e-06,
      "loss": 0.0024,
      "step": 26270
    },
    {
      "epoch": 0.4300089994273092,
      "grad_norm": 0.05559876762546286,
      "learning_rate": 7.033622339991925e-06,
      "loss": 0.0028,
      "step": 26280
    },
    {
      "epoch": 0.430172625378385,
      "grad_norm": 0.28636141542971366,
      "learning_rate": 7.031013055516589e-06,
      "loss": 0.0039,
      "step": 26290
    },
    {
      "epoch": 0.43033625132946085,
      "grad_norm": 0.16371011533088622,
      "learning_rate": 7.02840310845916e-06,
      "loss": 0.0033,
      "step": 26300
    },
    {
      "epoch": 0.4304998772805367,
      "grad_norm": 0.2454076254881016,
      "learning_rate": 7.025792499671088e-06,
      "loss": 0.0055,
      "step": 26310
    },
    {
      "epoch": 0.4306635032316125,
      "grad_norm": 0.059874613857367136,
      "learning_rate": 7.023181230004039e-06,
      "loss": 0.0036,
      "step": 26320
    },
    {
      "epoch": 0.4308271291826884,
      "grad_norm": 0.3551563966825225,
      "learning_rate": 7.020569300309894e-06,
      "loss": 0.0044,
      "step": 26330
    },
    {
      "epoch": 0.4309907551337642,
      "grad_norm": 0.1815498421283578,
      "learning_rate": 7.0179567114407466e-06,
      "loss": 0.0063,
      "step": 26340
    },
    {
      "epoch": 0.43115438108484005,
      "grad_norm": 0.300644638111522,
      "learning_rate": 7.0153434642489105e-06,
      "loss": 0.0027,
      "step": 26350
    },
    {
      "epoch": 0.4313180070359159,
      "grad_norm": 0.24561887504609559,
      "learning_rate": 7.012729559586909e-06,
      "loss": 0.0044,
      "step": 26360
    },
    {
      "epoch": 0.4314816329869917,
      "grad_norm": 0.14308150717914173,
      "learning_rate": 7.0101149983074835e-06,
      "loss": 0.0039,
      "step": 26370
    },
    {
      "epoch": 0.4316452589380676,
      "grad_norm": 0.08432327872081744,
      "learning_rate": 7.007499781263589e-06,
      "loss": 0.0049,
      "step": 26380
    },
    {
      "epoch": 0.43180888488914343,
      "grad_norm": 0.1319968120871778,
      "learning_rate": 7.0048839093083924e-06,
      "loss": 0.0025,
      "step": 26390
    },
    {
      "epoch": 0.43197251084021926,
      "grad_norm": 0.3925233589550464,
      "learning_rate": 7.0022673832952755e-06,
      "loss": 0.0039,
      "step": 26400
    },
    {
      "epoch": 0.4321361367912951,
      "grad_norm": 0.11656273847565152,
      "learning_rate": 6.999650204077836e-06,
      "loss": 0.0045,
      "step": 26410
    },
    {
      "epoch": 0.4322997627423709,
      "grad_norm": 0.11793715394093093,
      "learning_rate": 6.99703237250988e-06,
      "loss": 0.0071,
      "step": 26420
    },
    {
      "epoch": 0.4324633886934468,
      "grad_norm": 0.04020616294005387,
      "learning_rate": 6.9944138894454286e-06,
      "loss": 0.0121,
      "step": 26430
    },
    {
      "epoch": 0.43262701464452263,
      "grad_norm": 0.2896521614662805,
      "learning_rate": 6.991794755738718e-06,
      "loss": 0.0032,
      "step": 26440
    },
    {
      "epoch": 0.43279064059559846,
      "grad_norm": 0.28585578637671455,
      "learning_rate": 6.9891749722441905e-06,
      "loss": 0.0036,
      "step": 26450
    },
    {
      "epoch": 0.4329542665466743,
      "grad_norm": 0.08138209778883475,
      "learning_rate": 6.986554539816507e-06,
      "loss": 0.0035,
      "step": 26460
    },
    {
      "epoch": 0.4331178924977501,
      "grad_norm": 0.4606576564019319,
      "learning_rate": 6.983933459310538e-06,
      "loss": 0.0028,
      "step": 26470
    },
    {
      "epoch": 0.433281518448826,
      "grad_norm": 0.050669523940494635,
      "learning_rate": 6.981311731581362e-06,
      "loss": 0.0028,
      "step": 26480
    },
    {
      "epoch": 0.43344514439990184,
      "grad_norm": 0.3367169531022323,
      "learning_rate": 6.978689357484272e-06,
      "loss": 0.0043,
      "step": 26490
    },
    {
      "epoch": 0.43360877035097767,
      "grad_norm": 1.0214040298479903,
      "learning_rate": 6.976066337874774e-06,
      "loss": 0.01,
      "step": 26500
    },
    {
      "epoch": 0.4337723963020535,
      "grad_norm": 0.2398571509653539,
      "learning_rate": 6.973442673608577e-06,
      "loss": 0.0054,
      "step": 26510
    },
    {
      "epoch": 0.4339360222531293,
      "grad_norm": 0.35119390589056326,
      "learning_rate": 6.970818365541609e-06,
      "loss": 0.0032,
      "step": 26520
    },
    {
      "epoch": 0.4340996482042052,
      "grad_norm": 0.2546658645679215,
      "learning_rate": 6.968193414530003e-06,
      "loss": 0.0038,
      "step": 26530
    },
    {
      "epoch": 0.43426327415528104,
      "grad_norm": 0.08359499744380625,
      "learning_rate": 6.965567821430101e-06,
      "loss": 0.0024,
      "step": 26540
    },
    {
      "epoch": 0.4344269001063569,
      "grad_norm": 0.15206130028073248,
      "learning_rate": 6.962941587098459e-06,
      "loss": 0.002,
      "step": 26550
    },
    {
      "epoch": 0.4345905260574327,
      "grad_norm": 0.13538814682661104,
      "learning_rate": 6.960314712391838e-06,
      "loss": 0.0024,
      "step": 26560
    },
    {
      "epoch": 0.43475415200850853,
      "grad_norm": 0.3383678400522058,
      "learning_rate": 6.957687198167209e-06,
      "loss": 0.0037,
      "step": 26570
    },
    {
      "epoch": 0.4349177779595844,
      "grad_norm": 0.4931067122688483,
      "learning_rate": 6.955059045281754e-06,
      "loss": 0.003,
      "step": 26580
    },
    {
      "epoch": 0.43508140391066025,
      "grad_norm": 0.09294194993299032,
      "learning_rate": 6.95243025459286e-06,
      "loss": 0.0057,
      "step": 26590
    },
    {
      "epoch": 0.4352450298617361,
      "grad_norm": 0.21814255096405427,
      "learning_rate": 6.9498008269581215e-06,
      "loss": 0.0042,
      "step": 26600
    },
    {
      "epoch": 0.4354086558128119,
      "grad_norm": 0.0655241806742665,
      "learning_rate": 6.947170763235348e-06,
      "loss": 0.0037,
      "step": 26610
    },
    {
      "epoch": 0.43557228176388774,
      "grad_norm": 0.09426947713383059,
      "learning_rate": 6.944540064282546e-06,
      "loss": 0.0024,
      "step": 26620
    },
    {
      "epoch": 0.43573590771496357,
      "grad_norm": 0.47283577995391657,
      "learning_rate": 6.941908730957938e-06,
      "loss": 0.0096,
      "step": 26630
    },
    {
      "epoch": 0.43589953366603945,
      "grad_norm": 0.1333318790356309,
      "learning_rate": 6.939276764119949e-06,
      "loss": 0.0047,
      "step": 26640
    },
    {
      "epoch": 0.4360631596171153,
      "grad_norm": 0.2404295140637122,
      "learning_rate": 6.936644164627211e-06,
      "loss": 0.004,
      "step": 26650
    },
    {
      "epoch": 0.4362267855681911,
      "grad_norm": 0.24518198351447984,
      "learning_rate": 6.934010933338562e-06,
      "loss": 0.0038,
      "step": 26660
    },
    {
      "epoch": 0.43639041151926694,
      "grad_norm": 0.3185812874595286,
      "learning_rate": 6.931377071113049e-06,
      "loss": 0.0057,
      "step": 26670
    },
    {
      "epoch": 0.43655403747034277,
      "grad_norm": 0.37415758177021174,
      "learning_rate": 6.928742578809923e-06,
      "loss": 0.0037,
      "step": 26680
    },
    {
      "epoch": 0.43671766342141866,
      "grad_norm": 0.11597583755022083,
      "learning_rate": 6.926107457288639e-06,
      "loss": 0.0065,
      "step": 26690
    },
    {
      "epoch": 0.4368812893724945,
      "grad_norm": 0.18726594627044127,
      "learning_rate": 6.923471707408861e-06,
      "loss": 0.003,
      "step": 26700
    },
    {
      "epoch": 0.4370449153235703,
      "grad_norm": 0.11393662229777222,
      "learning_rate": 6.920835330030454e-06,
      "loss": 0.0056,
      "step": 26710
    },
    {
      "epoch": 0.43720854127464615,
      "grad_norm": 0.18781339406360215,
      "learning_rate": 6.91819832601349e-06,
      "loss": 0.0029,
      "step": 26720
    },
    {
      "epoch": 0.437372167225722,
      "grad_norm": 0.36467731302309664,
      "learning_rate": 6.915560696218244e-06,
      "loss": 0.0057,
      "step": 26730
    },
    {
      "epoch": 0.43753579317679786,
      "grad_norm": 0.07667592611856083,
      "learning_rate": 6.912922441505199e-06,
      "loss": 0.0036,
      "step": 26740
    },
    {
      "epoch": 0.4376994191278737,
      "grad_norm": 0.25489474028229125,
      "learning_rate": 6.9102835627350366e-06,
      "loss": 0.0051,
      "step": 26750
    },
    {
      "epoch": 0.4378630450789495,
      "grad_norm": 0.3282044914912616,
      "learning_rate": 6.907644060768643e-06,
      "loss": 0.0068,
      "step": 26760
    },
    {
      "epoch": 0.43802667103002535,
      "grad_norm": 0.45992873172913684,
      "learning_rate": 6.905003936467113e-06,
      "loss": 0.0034,
      "step": 26770
    },
    {
      "epoch": 0.4381902969811012,
      "grad_norm": 0.2532412518026153,
      "learning_rate": 6.902363190691737e-06,
      "loss": 0.004,
      "step": 26780
    },
    {
      "epoch": 0.43835392293217706,
      "grad_norm": 0.3905362384664212,
      "learning_rate": 6.899721824304013e-06,
      "loss": 0.0048,
      "step": 26790
    },
    {
      "epoch": 0.4385175488832529,
      "grad_norm": 0.20911831716236948,
      "learning_rate": 6.897079838165639e-06,
      "loss": 0.009,
      "step": 26800
    },
    {
      "epoch": 0.4386811748343287,
      "grad_norm": 0.1258964154101348,
      "learning_rate": 6.894437233138519e-06,
      "loss": 0.0026,
      "step": 26810
    },
    {
      "epoch": 0.43884480078540455,
      "grad_norm": 0.037943220304833394,
      "learning_rate": 6.891794010084752e-06,
      "loss": 0.004,
      "step": 26820
    },
    {
      "epoch": 0.4390084267364804,
      "grad_norm": 0.2798017272641533,
      "learning_rate": 6.889150169866648e-06,
      "loss": 0.0042,
      "step": 26830
    },
    {
      "epoch": 0.43917205268755627,
      "grad_norm": 0.10511146771166849,
      "learning_rate": 6.886505713346708e-06,
      "loss": 0.0027,
      "step": 26840
    },
    {
      "epoch": 0.4393356786386321,
      "grad_norm": 0.13947022131414216,
      "learning_rate": 6.883860641387642e-06,
      "loss": 0.0052,
      "step": 26850
    },
    {
      "epoch": 0.43949930458970793,
      "grad_norm": 0.08452819925743989,
      "learning_rate": 6.8812149548523575e-06,
      "loss": 0.0031,
      "step": 26860
    },
    {
      "epoch": 0.43966293054078376,
      "grad_norm": 0.1434245223487868,
      "learning_rate": 6.878568654603963e-06,
      "loss": 0.0033,
      "step": 26870
    },
    {
      "epoch": 0.4398265564918596,
      "grad_norm": 0.28453676144312634,
      "learning_rate": 6.8759217415057656e-06,
      "loss": 0.0053,
      "step": 26880
    },
    {
      "epoch": 0.4399901824429355,
      "grad_norm": 0.18478552034540832,
      "learning_rate": 6.873274216421277e-06,
      "loss": 0.0029,
      "step": 26890
    },
    {
      "epoch": 0.4401538083940113,
      "grad_norm": 0.2676192159492127,
      "learning_rate": 6.8706260802142024e-06,
      "loss": 0.0027,
      "step": 26900
    },
    {
      "epoch": 0.44031743434508713,
      "grad_norm": 0.07319288925996524,
      "learning_rate": 6.86797733374845e-06,
      "loss": 0.0035,
      "step": 26910
    },
    {
      "epoch": 0.44048106029616296,
      "grad_norm": 0.29009586763264417,
      "learning_rate": 6.86532797788813e-06,
      "loss": 0.003,
      "step": 26920
    },
    {
      "epoch": 0.4406446862472388,
      "grad_norm": 0.08665517937386834,
      "learning_rate": 6.862678013497542e-06,
      "loss": 0.0072,
      "step": 26930
    },
    {
      "epoch": 0.4408083121983147,
      "grad_norm": 0.2691543553758849,
      "learning_rate": 6.8600274414411935e-06,
      "loss": 0.0043,
      "step": 26940
    },
    {
      "epoch": 0.4409719381493905,
      "grad_norm": 0.10455772810240402,
      "learning_rate": 6.857376262583788e-06,
      "loss": 0.0037,
      "step": 26950
    },
    {
      "epoch": 0.44113556410046634,
      "grad_norm": 0.19673337511626976,
      "learning_rate": 6.854724477790224e-06,
      "loss": 0.0025,
      "step": 26960
    },
    {
      "epoch": 0.44129919005154217,
      "grad_norm": 0.3046458368383548,
      "learning_rate": 6.852072087925596e-06,
      "loss": 0.0056,
      "step": 26970
    },
    {
      "epoch": 0.441462816002618,
      "grad_norm": 0.41068400098969343,
      "learning_rate": 6.849419093855207e-06,
      "loss": 0.0039,
      "step": 26980
    },
    {
      "epoch": 0.4416264419536939,
      "grad_norm": 0.19644071604477897,
      "learning_rate": 6.846765496444543e-06,
      "loss": 0.0039,
      "step": 26990
    },
    {
      "epoch": 0.4417900679047697,
      "grad_norm": 0.441587072847681,
      "learning_rate": 6.844111296559295e-06,
      "loss": 0.0034,
      "step": 27000
    },
    {
      "epoch": 0.44195369385584554,
      "grad_norm": 0.049335780060254096,
      "learning_rate": 6.841456495065352e-06,
      "loss": 0.0026,
      "step": 27010
    },
    {
      "epoch": 0.44211731980692137,
      "grad_norm": 0.22555364982685128,
      "learning_rate": 6.838801092828789e-06,
      "loss": 0.0055,
      "step": 27020
    },
    {
      "epoch": 0.4422809457579972,
      "grad_norm": 0.3506820752376438,
      "learning_rate": 6.8361450907158914e-06,
      "loss": 0.0037,
      "step": 27030
    },
    {
      "epoch": 0.4424445717090731,
      "grad_norm": 0.37252589746462683,
      "learning_rate": 6.833488489593129e-06,
      "loss": 0.0048,
      "step": 27040
    },
    {
      "epoch": 0.4426081976601489,
      "grad_norm": 0.12596561076287582,
      "learning_rate": 6.83083129032717e-06,
      "loss": 0.0032,
      "step": 27050
    },
    {
      "epoch": 0.44277182361122475,
      "grad_norm": 0.36855003586646656,
      "learning_rate": 6.828173493784882e-06,
      "loss": 0.0034,
      "step": 27060
    },
    {
      "epoch": 0.4429354495623006,
      "grad_norm": 0.1187093229332145,
      "learning_rate": 6.825515100833322e-06,
      "loss": 0.0027,
      "step": 27070
    },
    {
      "epoch": 0.4430990755133764,
      "grad_norm": 0.21420625444207794,
      "learning_rate": 6.822856112339743e-06,
      "loss": 0.0044,
      "step": 27080
    },
    {
      "epoch": 0.44326270146445224,
      "grad_norm": 0.13212195982297378,
      "learning_rate": 6.820196529171595e-06,
      "loss": 0.0029,
      "step": 27090
    },
    {
      "epoch": 0.4434263274155281,
      "grad_norm": 0.16991669895329012,
      "learning_rate": 6.81753635219652e-06,
      "loss": 0.0055,
      "step": 27100
    },
    {
      "epoch": 0.44358995336660395,
      "grad_norm": 0.1301498199844634,
      "learning_rate": 6.814875582282351e-06,
      "loss": 0.0033,
      "step": 27110
    },
    {
      "epoch": 0.4437535793176798,
      "grad_norm": 0.27160532100392515,
      "learning_rate": 6.81221422029712e-06,
      "loss": 0.003,
      "step": 27120
    },
    {
      "epoch": 0.4439172052687556,
      "grad_norm": 0.09349616541647653,
      "learning_rate": 6.809552267109048e-06,
      "loss": 0.0039,
      "step": 27130
    },
    {
      "epoch": 0.44408083121983144,
      "grad_norm": 0.0619327774518488,
      "learning_rate": 6.806889723586549e-06,
      "loss": 0.0033,
      "step": 27140
    },
    {
      "epoch": 0.4442444571709073,
      "grad_norm": 0.06738360657410429,
      "learning_rate": 6.804226590598234e-06,
      "loss": 0.0033,
      "step": 27150
    },
    {
      "epoch": 0.44440808312198316,
      "grad_norm": 0.19304807966978316,
      "learning_rate": 6.801562869012901e-06,
      "loss": 0.0031,
      "step": 27160
    },
    {
      "epoch": 0.444571709073059,
      "grad_norm": 0.22817749270864807,
      "learning_rate": 6.7988985596995406e-06,
      "loss": 0.0035,
      "step": 27170
    },
    {
      "epoch": 0.4447353350241348,
      "grad_norm": 0.2501569833202537,
      "learning_rate": 6.796233663527338e-06,
      "loss": 0.0024,
      "step": 27180
    },
    {
      "epoch": 0.44489896097521064,
      "grad_norm": 0.2511496296375615,
      "learning_rate": 6.793568181365671e-06,
      "loss": 0.0037,
      "step": 27190
    },
    {
      "epoch": 0.44506258692628653,
      "grad_norm": 0.2057920450301289,
      "learning_rate": 6.790902114084102e-06,
      "loss": 0.0059,
      "step": 27200
    },
    {
      "epoch": 0.44522621287736236,
      "grad_norm": 0.30349328966832656,
      "learning_rate": 6.78823546255239e-06,
      "loss": 0.0041,
      "step": 27210
    },
    {
      "epoch": 0.4453898388284382,
      "grad_norm": 0.3238361132473224,
      "learning_rate": 6.785568227640482e-06,
      "loss": 0.0044,
      "step": 27220
    },
    {
      "epoch": 0.445553464779514,
      "grad_norm": 0.2539030981669684,
      "learning_rate": 6.7829004102185196e-06,
      "loss": 0.0028,
      "step": 27230
    },
    {
      "epoch": 0.44571709073058985,
      "grad_norm": 0.19106719969786964,
      "learning_rate": 6.780232011156826e-06,
      "loss": 0.0031,
      "step": 27240
    },
    {
      "epoch": 0.44588071668166573,
      "grad_norm": 0.15011373766246566,
      "learning_rate": 6.7775630313259226e-06,
      "loss": 0.0026,
      "step": 27250
    },
    {
      "epoch": 0.44604434263274156,
      "grad_norm": 0.6300576875990713,
      "learning_rate": 6.774893471596515e-06,
      "loss": 0.0062,
      "step": 27260
    },
    {
      "epoch": 0.4462079685838174,
      "grad_norm": 0.12215105585574247,
      "learning_rate": 6.772223332839502e-06,
      "loss": 0.0044,
      "step": 27270
    },
    {
      "epoch": 0.4463715945348932,
      "grad_norm": 0.14687192121190618,
      "learning_rate": 6.769552615925967e-06,
      "loss": 0.0034,
      "step": 27280
    },
    {
      "epoch": 0.44653522048596905,
      "grad_norm": 0.23091678634828572,
      "learning_rate": 6.766881321727187e-06,
      "loss": 0.0036,
      "step": 27290
    },
    {
      "epoch": 0.44669884643704494,
      "grad_norm": 0.12763779831920075,
      "learning_rate": 6.7642094511146206e-06,
      "loss": 0.0029,
      "step": 27300
    },
    {
      "epoch": 0.44686247238812077,
      "grad_norm": 0.1685973600876189,
      "learning_rate": 6.761537004959922e-06,
      "loss": 0.0029,
      "step": 27310
    },
    {
      "epoch": 0.4470260983391966,
      "grad_norm": 0.07115366140522682,
      "learning_rate": 6.758863984134927e-06,
      "loss": 0.0035,
      "step": 27320
    },
    {
      "epoch": 0.44718972429027243,
      "grad_norm": 0.20852679954973063,
      "learning_rate": 6.756190389511661e-06,
      "loss": 0.0022,
      "step": 27330
    },
    {
      "epoch": 0.44735335024134826,
      "grad_norm": 0.1620276853985165,
      "learning_rate": 6.75351622196234e-06,
      "loss": 0.0051,
      "step": 27340
    },
    {
      "epoch": 0.44751697619242414,
      "grad_norm": 0.10082416701852726,
      "learning_rate": 6.7508414823593615e-06,
      "loss": 0.0035,
      "step": 27350
    },
    {
      "epoch": 0.4476806021435,
      "grad_norm": 0.2683457363565609,
      "learning_rate": 6.748166171575312e-06,
      "loss": 0.0036,
      "step": 27360
    },
    {
      "epoch": 0.4478442280945758,
      "grad_norm": 0.08933184564006597,
      "learning_rate": 6.7454902904829665e-06,
      "loss": 0.0037,
      "step": 27370
    },
    {
      "epoch": 0.44800785404565163,
      "grad_norm": 0.3206275654916656,
      "learning_rate": 6.74281383995528e-06,
      "loss": 0.0034,
      "step": 27380
    },
    {
      "epoch": 0.44817147999672746,
      "grad_norm": 0.19322088702913504,
      "learning_rate": 6.740136820865399e-06,
      "loss": 0.0039,
      "step": 27390
    },
    {
      "epoch": 0.44833510594780335,
      "grad_norm": 0.21529193356299864,
      "learning_rate": 6.737459234086659e-06,
      "loss": 0.0035,
      "step": 27400
    },
    {
      "epoch": 0.4484987318988792,
      "grad_norm": 0.16180435444783095,
      "learning_rate": 6.734781080492565e-06,
      "loss": 0.003,
      "step": 27410
    },
    {
      "epoch": 0.448662357849955,
      "grad_norm": 0.1898129571454145,
      "learning_rate": 6.732102360956824e-06,
      "loss": 0.0033,
      "step": 27420
    },
    {
      "epoch": 0.44882598380103084,
      "grad_norm": 0.18386068824888088,
      "learning_rate": 6.729423076353321e-06,
      "loss": 0.002,
      "step": 27430
    },
    {
      "epoch": 0.44898960975210667,
      "grad_norm": 0.2817407603197924,
      "learning_rate": 6.7267432275561226e-06,
      "loss": 0.0018,
      "step": 27440
    },
    {
      "epoch": 0.44915323570318255,
      "grad_norm": 0.31293412475848786,
      "learning_rate": 6.724062815439483e-06,
      "loss": 0.0053,
      "step": 27450
    },
    {
      "epoch": 0.4493168616542584,
      "grad_norm": 0.15001263918443392,
      "learning_rate": 6.721381840877841e-06,
      "loss": 0.0067,
      "step": 27460
    },
    {
      "epoch": 0.4494804876053342,
      "grad_norm": 0.11503062692936089,
      "learning_rate": 6.718700304745813e-06,
      "loss": 0.0054,
      "step": 27470
    },
    {
      "epoch": 0.44964411355641004,
      "grad_norm": 0.011530787489426025,
      "learning_rate": 6.716018207918207e-06,
      "loss": 0.0052,
      "step": 27480
    },
    {
      "epoch": 0.44980773950748587,
      "grad_norm": 0.3394037368213191,
      "learning_rate": 6.713335551270008e-06,
      "loss": 0.0031,
      "step": 27490
    },
    {
      "epoch": 0.4499713654585617,
      "grad_norm": 0.4962577480068975,
      "learning_rate": 6.710652335676385e-06,
      "loss": 0.0029,
      "step": 27500
    },
    {
      "epoch": 0.4501349914096376,
      "grad_norm": 0.13910441443303384,
      "learning_rate": 6.70796856201269e-06,
      "loss": 0.0041,
      "step": 27510
    },
    {
      "epoch": 0.4502986173607134,
      "grad_norm": 0.1728998996073941,
      "learning_rate": 6.705284231154457e-06,
      "loss": 0.0032,
      "step": 27520
    },
    {
      "epoch": 0.45046224331178925,
      "grad_norm": 0.4107431421970128,
      "learning_rate": 6.7025993439774005e-06,
      "loss": 0.0048,
      "step": 27530
    },
    {
      "epoch": 0.4506258692628651,
      "grad_norm": 0.13430169359339686,
      "learning_rate": 6.699913901357419e-06,
      "loss": 0.0043,
      "step": 27540
    },
    {
      "epoch": 0.4507894952139409,
      "grad_norm": 0.28455499650246296,
      "learning_rate": 6.697227904170589e-06,
      "loss": 0.0028,
      "step": 27550
    },
    {
      "epoch": 0.4509531211650168,
      "grad_norm": 0.28013527746554107,
      "learning_rate": 6.6945413532931695e-06,
      "loss": 0.0029,
      "step": 27560
    },
    {
      "epoch": 0.4511167471160926,
      "grad_norm": 0.2546719535210105,
      "learning_rate": 6.6918542496016015e-06,
      "loss": 0.0025,
      "step": 27570
    },
    {
      "epoch": 0.45128037306716845,
      "grad_norm": 0.4091843429159266,
      "learning_rate": 6.689166593972506e-06,
      "loss": 0.0034,
      "step": 27580
    },
    {
      "epoch": 0.4514439990182443,
      "grad_norm": 0.29023836466149916,
      "learning_rate": 6.68647838728268e-06,
      "loss": 0.0106,
      "step": 27590
    },
    {
      "epoch": 0.4516076249693201,
      "grad_norm": 0.10220764048375632,
      "learning_rate": 6.683789630409105e-06,
      "loss": 0.0051,
      "step": 27600
    },
    {
      "epoch": 0.451771250920396,
      "grad_norm": 0.1853279600570272,
      "learning_rate": 6.681100324228942e-06,
      "loss": 0.0034,
      "step": 27610
    },
    {
      "epoch": 0.4519348768714718,
      "grad_norm": 0.29853930768537046,
      "learning_rate": 6.678410469619526e-06,
      "loss": 0.0047,
      "step": 27620
    },
    {
      "epoch": 0.45209850282254765,
      "grad_norm": 0.09633031660729302,
      "learning_rate": 6.675720067458378e-06,
      "loss": 0.0036,
      "step": 27630
    },
    {
      "epoch": 0.4522621287736235,
      "grad_norm": 0.15881005505147477,
      "learning_rate": 6.673029118623192e-06,
      "loss": 0.0041,
      "step": 27640
    },
    {
      "epoch": 0.4524257547246993,
      "grad_norm": 0.4448010176951898,
      "learning_rate": 6.670337623991843e-06,
      "loss": 0.0036,
      "step": 27650
    },
    {
      "epoch": 0.4525893806757752,
      "grad_norm": 0.21669927361220584,
      "learning_rate": 6.667645584442384e-06,
      "loss": 0.0038,
      "step": 27660
    },
    {
      "epoch": 0.45275300662685103,
      "grad_norm": 0.23087343395496432,
      "learning_rate": 6.664953000853045e-06,
      "loss": 0.0041,
      "step": 27670
    },
    {
      "epoch": 0.45291663257792686,
      "grad_norm": 0.11427074319863043,
      "learning_rate": 6.662259874102234e-06,
      "loss": 0.0041,
      "step": 27680
    },
    {
      "epoch": 0.4530802585290027,
      "grad_norm": 0.23530300356438355,
      "learning_rate": 6.659566205068535e-06,
      "loss": 0.0028,
      "step": 27690
    },
    {
      "epoch": 0.4532438844800785,
      "grad_norm": 0.21368281631611508,
      "learning_rate": 6.656871994630711e-06,
      "loss": 0.0038,
      "step": 27700
    },
    {
      "epoch": 0.4534075104311544,
      "grad_norm": 0.18810898599614018,
      "learning_rate": 6.654177243667701e-06,
      "loss": 0.0031,
      "step": 27710
    },
    {
      "epoch": 0.45357113638223023,
      "grad_norm": 0.22847075665782174,
      "learning_rate": 6.651481953058617e-06,
      "loss": 0.0023,
      "step": 27720
    },
    {
      "epoch": 0.45373476233330606,
      "grad_norm": 0.1006610330293393,
      "learning_rate": 6.648786123682753e-06,
      "loss": 0.0031,
      "step": 27730
    },
    {
      "epoch": 0.4538983882843819,
      "grad_norm": 0.15704252599502816,
      "learning_rate": 6.646089756419575e-06,
      "loss": 0.0039,
      "step": 27740
    },
    {
      "epoch": 0.4540620142354577,
      "grad_norm": 0.14882139131445798,
      "learning_rate": 6.6433928521487234e-06,
      "loss": 0.0035,
      "step": 27750
    },
    {
      "epoch": 0.4542256401865336,
      "grad_norm": 0.0617542920675112,
      "learning_rate": 6.640695411750017e-06,
      "loss": 0.0027,
      "step": 27760
    },
    {
      "epoch": 0.45438926613760944,
      "grad_norm": 0.23990166752639389,
      "learning_rate": 6.637997436103449e-06,
      "loss": 0.0049,
      "step": 27770
    },
    {
      "epoch": 0.45455289208868527,
      "grad_norm": 0.1513712189962754,
      "learning_rate": 6.635298926089183e-06,
      "loss": 0.003,
      "step": 27780
    },
    {
      "epoch": 0.4547165180397611,
      "grad_norm": 0.09116766049656864,
      "learning_rate": 6.632599882587564e-06,
      "loss": 0.003,
      "step": 27790
    },
    {
      "epoch": 0.4548801439908369,
      "grad_norm": 0.2635636914762161,
      "learning_rate": 6.629900306479105e-06,
      "loss": 0.0022,
      "step": 27800
    },
    {
      "epoch": 0.4550437699419128,
      "grad_norm": 0.14830682433501483,
      "learning_rate": 6.627200198644495e-06,
      "loss": 0.0041,
      "step": 27810
    },
    {
      "epoch": 0.45520739589298864,
      "grad_norm": 0.11682866365036997,
      "learning_rate": 6.6244995599645965e-06,
      "loss": 0.0039,
      "step": 27820
    },
    {
      "epoch": 0.45537102184406447,
      "grad_norm": 0.18068296123738037,
      "learning_rate": 6.621798391320447e-06,
      "loss": 0.0024,
      "step": 27830
    },
    {
      "epoch": 0.4555346477951403,
      "grad_norm": 0.06292653796356126,
      "learning_rate": 6.61909669359325e-06,
      "loss": 0.0028,
      "step": 27840
    },
    {
      "epoch": 0.45569827374621613,
      "grad_norm": 0.17913351015372944,
      "learning_rate": 6.616394467664393e-06,
      "loss": 0.0048,
      "step": 27850
    },
    {
      "epoch": 0.455861899697292,
      "grad_norm": 0.14321371585553475,
      "learning_rate": 6.613691714415426e-06,
      "loss": 0.0041,
      "step": 27860
    },
    {
      "epoch": 0.45602552564836785,
      "grad_norm": 0.25830690828249425,
      "learning_rate": 6.610988434728074e-06,
      "loss": 0.0037,
      "step": 27870
    },
    {
      "epoch": 0.4561891515994437,
      "grad_norm": 0.3233300514334663,
      "learning_rate": 6.608284629484236e-06,
      "loss": 0.0044,
      "step": 27880
    },
    {
      "epoch": 0.4563527775505195,
      "grad_norm": 0.17197116439186733,
      "learning_rate": 6.605580299565979e-06,
      "loss": 0.0039,
      "step": 27890
    },
    {
      "epoch": 0.45651640350159534,
      "grad_norm": 0.08076616052499506,
      "learning_rate": 6.602875445855544e-06,
      "loss": 0.0028,
      "step": 27900
    },
    {
      "epoch": 0.4566800294526712,
      "grad_norm": 0.22018222274379615,
      "learning_rate": 6.600170069235341e-06,
      "loss": 0.0044,
      "step": 27910
    },
    {
      "epoch": 0.45684365540374705,
      "grad_norm": 0.0943511793138976,
      "learning_rate": 6.597464170587952e-06,
      "loss": 0.0055,
      "step": 27920
    },
    {
      "epoch": 0.4570072813548229,
      "grad_norm": 0.10811601592329687,
      "learning_rate": 6.594757750796126e-06,
      "loss": 0.004,
      "step": 27930
    },
    {
      "epoch": 0.4571709073058987,
      "grad_norm": 0.345361207625253,
      "learning_rate": 6.5920508107427895e-06,
      "loss": 0.0063,
      "step": 27940
    },
    {
      "epoch": 0.45733453325697454,
      "grad_norm": 0.12460512429776673,
      "learning_rate": 6.58934335131103e-06,
      "loss": 0.0039,
      "step": 27950
    },
    {
      "epoch": 0.45749815920805037,
      "grad_norm": 0.4157533186199701,
      "learning_rate": 6.586635373384108e-06,
      "loss": 0.0027,
      "step": 27960
    },
    {
      "epoch": 0.45766178515912626,
      "grad_norm": 0.19605744477493942,
      "learning_rate": 6.583926877845457e-06,
      "loss": 0.0029,
      "step": 27970
    },
    {
      "epoch": 0.4578254111102021,
      "grad_norm": 0.31972212468392336,
      "learning_rate": 6.5812178655786706e-06,
      "loss": 0.0048,
      "step": 27980
    },
    {
      "epoch": 0.4579890370612779,
      "grad_norm": 0.11560751669539406,
      "learning_rate": 6.5785083374675205e-06,
      "loss": 0.0023,
      "step": 27990
    },
    {
      "epoch": 0.45815266301235374,
      "grad_norm": 0.2452682001892355,
      "learning_rate": 6.575798294395942e-06,
      "loss": 0.0028,
      "step": 28000
    },
    {
      "epoch": 0.4583162889634296,
      "grad_norm": 0.24777936951951843,
      "learning_rate": 6.573087737248036e-06,
      "loss": 0.0036,
      "step": 28010
    },
    {
      "epoch": 0.45847991491450546,
      "grad_norm": 0.13406181121718788,
      "learning_rate": 6.570376666908076e-06,
      "loss": 0.0056,
      "step": 28020
    },
    {
      "epoch": 0.4586435408655813,
      "grad_norm": 0.04274947482068868,
      "learning_rate": 6.5676650842605e-06,
      "loss": 0.0037,
      "step": 28030
    },
    {
      "epoch": 0.4588071668166571,
      "grad_norm": 0.23387800396372455,
      "learning_rate": 6.564952990189912e-06,
      "loss": 0.0041,
      "step": 28040
    },
    {
      "epoch": 0.45897079276773295,
      "grad_norm": 0.12753395391179928,
      "learning_rate": 6.56224038558109e-06,
      "loss": 0.0063,
      "step": 28050
    },
    {
      "epoch": 0.4591344187188088,
      "grad_norm": 0.04267521522810831,
      "learning_rate": 6.559527271318968e-06,
      "loss": 0.0027,
      "step": 28060
    },
    {
      "epoch": 0.45929804466988466,
      "grad_norm": 0.19258106104619244,
      "learning_rate": 6.556813648288654e-06,
      "loss": 0.0037,
      "step": 28070
    },
    {
      "epoch": 0.4594616706209605,
      "grad_norm": 0.10709265415044303,
      "learning_rate": 6.55409951737542e-06,
      "loss": 0.0031,
      "step": 28080
    },
    {
      "epoch": 0.4596252965720363,
      "grad_norm": 0.5781668050613947,
      "learning_rate": 6.551384879464701e-06,
      "loss": 0.0052,
      "step": 28090
    },
    {
      "epoch": 0.45978892252311215,
      "grad_norm": 0.05237590097146237,
      "learning_rate": 6.548669735442102e-06,
      "loss": 0.0044,
      "step": 28100
    },
    {
      "epoch": 0.459952548474188,
      "grad_norm": 0.10969787323608506,
      "learning_rate": 6.545954086193388e-06,
      "loss": 0.0043,
      "step": 28110
    },
    {
      "epoch": 0.46011617442526387,
      "grad_norm": 0.3622901438224775,
      "learning_rate": 6.543237932604495e-06,
      "loss": 0.0043,
      "step": 28120
    },
    {
      "epoch": 0.4602798003763397,
      "grad_norm": 0.157251485534472,
      "learning_rate": 6.5405212755615164e-06,
      "loss": 0.0029,
      "step": 28130
    },
    {
      "epoch": 0.46044342632741553,
      "grad_norm": 0.11641807471512758,
      "learning_rate": 6.537804115950717e-06,
      "loss": 0.0032,
      "step": 28140
    },
    {
      "epoch": 0.46060705227849136,
      "grad_norm": 0.08176782794943495,
      "learning_rate": 6.53508645465852e-06,
      "loss": 0.0028,
      "step": 28150
    },
    {
      "epoch": 0.4607706782295672,
      "grad_norm": 0.11294801996424937,
      "learning_rate": 6.5323682925715136e-06,
      "loss": 0.0045,
      "step": 28160
    },
    {
      "epoch": 0.4609343041806431,
      "grad_norm": 0.22385870160449708,
      "learning_rate": 6.5296496305764525e-06,
      "loss": 0.0069,
      "step": 28170
    },
    {
      "epoch": 0.4610979301317189,
      "grad_norm": 0.21869673012488364,
      "learning_rate": 6.526930469560251e-06,
      "loss": 0.0043,
      "step": 28180
    },
    {
      "epoch": 0.46126155608279473,
      "grad_norm": 0.1413104353645075,
      "learning_rate": 6.524210810409988e-06,
      "loss": 0.004,
      "step": 28190
    },
    {
      "epoch": 0.46142518203387056,
      "grad_norm": 0.2351052479553508,
      "learning_rate": 6.521490654012901e-06,
      "loss": 0.0031,
      "step": 28200
    },
    {
      "epoch": 0.4615888079849464,
      "grad_norm": 0.09478239102803084,
      "learning_rate": 6.518770001256396e-06,
      "loss": 0.0042,
      "step": 28210
    },
    {
      "epoch": 0.4617524339360223,
      "grad_norm": 0.1218401218070308,
      "learning_rate": 6.5160488530280385e-06,
      "loss": 0.0023,
      "step": 28220
    },
    {
      "epoch": 0.4619160598870981,
      "grad_norm": 0.06855636961544907,
      "learning_rate": 6.513327210215552e-06,
      "loss": 0.0055,
      "step": 28230
    },
    {
      "epoch": 0.46207968583817394,
      "grad_norm": 0.1261837964054517,
      "learning_rate": 6.510605073706826e-06,
      "loss": 0.0056,
      "step": 28240
    },
    {
      "epoch": 0.46224331178924977,
      "grad_norm": 0.2578633714803832,
      "learning_rate": 6.5078824443899115e-06,
      "loss": 0.0027,
      "step": 28250
    },
    {
      "epoch": 0.4624069377403256,
      "grad_norm": 0.14503415327099703,
      "learning_rate": 6.505159323153013e-06,
      "loss": 0.0032,
      "step": 28260
    },
    {
      "epoch": 0.4625705636914015,
      "grad_norm": 0.09252036412698689,
      "learning_rate": 6.502435710884506e-06,
      "loss": 0.0027,
      "step": 28270
    },
    {
      "epoch": 0.4627341896424773,
      "grad_norm": 0.15244492131846205,
      "learning_rate": 6.499711608472918e-06,
      "loss": 0.0051,
      "step": 28280
    },
    {
      "epoch": 0.46289781559355314,
      "grad_norm": 0.26100027524936276,
      "learning_rate": 6.496987016806938e-06,
      "loss": 0.0024,
      "step": 28290
    },
    {
      "epoch": 0.46306144154462897,
      "grad_norm": 0.1089780942651381,
      "learning_rate": 6.49426193677542e-06,
      "loss": 0.0039,
      "step": 28300
    },
    {
      "epoch": 0.4632250674957048,
      "grad_norm": 0.5848199749858723,
      "learning_rate": 6.491536369267371e-06,
      "loss": 0.0078,
      "step": 28310
    },
    {
      "epoch": 0.4633886934467807,
      "grad_norm": 0.2802170074961866,
      "learning_rate": 6.488810315171956e-06,
      "loss": 0.0035,
      "step": 28320
    },
    {
      "epoch": 0.4635523193978565,
      "grad_norm": 0.35529530886876903,
      "learning_rate": 6.4860837753785066e-06,
      "loss": 0.0047,
      "step": 28330
    },
    {
      "epoch": 0.46371594534893235,
      "grad_norm": 0.2406100141099862,
      "learning_rate": 6.483356750776507e-06,
      "loss": 0.0046,
      "step": 28340
    },
    {
      "epoch": 0.4638795713000082,
      "grad_norm": 0.10982221706496778,
      "learning_rate": 6.480629242255598e-06,
      "loss": 0.003,
      "step": 28350
    },
    {
      "epoch": 0.464043197251084,
      "grad_norm": 0.2851794022887627,
      "learning_rate": 6.4779012507055864e-06,
      "loss": 0.0049,
      "step": 28360
    },
    {
      "epoch": 0.46420682320215984,
      "grad_norm": 0.02129570408499468,
      "learning_rate": 6.4751727770164275e-06,
      "loss": 0.0029,
      "step": 28370
    },
    {
      "epoch": 0.4643704491532357,
      "grad_norm": 0.04854633804719744,
      "learning_rate": 6.472443822078236e-06,
      "loss": 0.0035,
      "step": 28380
    },
    {
      "epoch": 0.46453407510431155,
      "grad_norm": 0.20110606401675876,
      "learning_rate": 6.469714386781288e-06,
      "loss": 0.0026,
      "step": 28390
    },
    {
      "epoch": 0.4646977010553874,
      "grad_norm": 0.6136880304702995,
      "learning_rate": 6.4669844720160145e-06,
      "loss": 0.005,
      "step": 28400
    },
    {
      "epoch": 0.4648613270064632,
      "grad_norm": 0.2631771719340116,
      "learning_rate": 6.464254078672998e-06,
      "loss": 0.0049,
      "step": 28410
    },
    {
      "epoch": 0.46502495295753904,
      "grad_norm": 0.1278674156545175,
      "learning_rate": 6.461523207642986e-06,
      "loss": 0.0037,
      "step": 28420
    },
    {
      "epoch": 0.4651885789086149,
      "grad_norm": 0.1751038952152329,
      "learning_rate": 6.458791859816872e-06,
      "loss": 0.0031,
      "step": 28430
    },
    {
      "epoch": 0.46535220485969075,
      "grad_norm": 0.06645694751595674,
      "learning_rate": 6.4560600360857115e-06,
      "loss": 0.0025,
      "step": 28440
    },
    {
      "epoch": 0.4655158308107666,
      "grad_norm": 0.11945802561504729,
      "learning_rate": 6.4533277373407154e-06,
      "loss": 0.0032,
      "step": 28450
    },
    {
      "epoch": 0.4656794567618424,
      "grad_norm": 0.13495307293374847,
      "learning_rate": 6.450594964473248e-06,
      "loss": 0.003,
      "step": 28460
    },
    {
      "epoch": 0.46584308271291824,
      "grad_norm": 0.04360982806844561,
      "learning_rate": 6.447861718374825e-06,
      "loss": 0.0047,
      "step": 28470
    },
    {
      "epoch": 0.46600670866399413,
      "grad_norm": 0.1541740574734713,
      "learning_rate": 6.445127999937122e-06,
      "loss": 0.0032,
      "step": 28480
    },
    {
      "epoch": 0.46617033461506996,
      "grad_norm": 0.151070101830312,
      "learning_rate": 6.442393810051966e-06,
      "loss": 0.0033,
      "step": 28490
    },
    {
      "epoch": 0.4663339605661458,
      "grad_norm": 0.11840643569788492,
      "learning_rate": 6.439659149611337e-06,
      "loss": 0.0025,
      "step": 28500
    },
    {
      "epoch": 0.4664975865172216,
      "grad_norm": 0.21699485517620457,
      "learning_rate": 6.436924019507371e-06,
      "loss": 0.0051,
      "step": 28510
    },
    {
      "epoch": 0.46666121246829745,
      "grad_norm": 0.11668382574992901,
      "learning_rate": 6.434188420632356e-06,
      "loss": 0.0038,
      "step": 28520
    },
    {
      "epoch": 0.46682483841937333,
      "grad_norm": 0.15067176771456303,
      "learning_rate": 6.431452353878731e-06,
      "loss": 0.004,
      "step": 28530
    },
    {
      "epoch": 0.46698846437044916,
      "grad_norm": 0.034745285684074256,
      "learning_rate": 6.42871582013909e-06,
      "loss": 0.0026,
      "step": 28540
    },
    {
      "epoch": 0.467152090321525,
      "grad_norm": 0.286874910094359,
      "learning_rate": 6.42597882030618e-06,
      "loss": 0.0041,
      "step": 28550
    },
    {
      "epoch": 0.4673157162726008,
      "grad_norm": 0.07062296015363291,
      "learning_rate": 6.423241355272898e-06,
      "loss": 0.0046,
      "step": 28560
    },
    {
      "epoch": 0.46747934222367665,
      "grad_norm": 0.10331640685732608,
      "learning_rate": 6.420503425932292e-06,
      "loss": 0.0053,
      "step": 28570
    },
    {
      "epoch": 0.46764296817475254,
      "grad_norm": 0.45732462070496194,
      "learning_rate": 6.417765033177566e-06,
      "loss": 0.0063,
      "step": 28580
    },
    {
      "epoch": 0.46780659412582837,
      "grad_norm": 0.19433731930437192,
      "learning_rate": 6.415026177902071e-06,
      "loss": 0.0031,
      "step": 28590
    },
    {
      "epoch": 0.4679702200769042,
      "grad_norm": 0.30586951493956016,
      "learning_rate": 6.412286860999308e-06,
      "loss": 0.0027,
      "step": 28600
    },
    {
      "epoch": 0.46813384602798,
      "grad_norm": 0.1288625194013928,
      "learning_rate": 6.409547083362935e-06,
      "loss": 0.0032,
      "step": 28610
    },
    {
      "epoch": 0.46829747197905586,
      "grad_norm": 0.04114386864208688,
      "learning_rate": 6.406806845886752e-06,
      "loss": 0.0043,
      "step": 28620
    },
    {
      "epoch": 0.46846109793013174,
      "grad_norm": 0.0856279354914779,
      "learning_rate": 6.404066149464716e-06,
      "loss": 0.0026,
      "step": 28630
    },
    {
      "epoch": 0.46862472388120757,
      "grad_norm": 0.058458105618210764,
      "learning_rate": 6.401324994990928e-06,
      "loss": 0.0071,
      "step": 28640
    },
    {
      "epoch": 0.4687883498322834,
      "grad_norm": 0.1948822405731856,
      "learning_rate": 6.398583383359642e-06,
      "loss": 0.0051,
      "step": 28650
    },
    {
      "epoch": 0.46895197578335923,
      "grad_norm": 0.3138218844570942,
      "learning_rate": 6.395841315465261e-06,
      "loss": 0.0053,
      "step": 28660
    },
    {
      "epoch": 0.46911560173443506,
      "grad_norm": 0.060558873230134716,
      "learning_rate": 6.3930987922023355e-06,
      "loss": 0.0047,
      "step": 28670
    },
    {
      "epoch": 0.46927922768551095,
      "grad_norm": 0.1294892096926167,
      "learning_rate": 6.390355814465567e-06,
      "loss": 0.002,
      "step": 28680
    },
    {
      "epoch": 0.4694428536365868,
      "grad_norm": 0.31763248895548724,
      "learning_rate": 6.387612383149801e-06,
      "loss": 0.0035,
      "step": 28690
    },
    {
      "epoch": 0.4696064795876626,
      "grad_norm": 0.20096824516147963,
      "learning_rate": 6.3848684991500365e-06,
      "loss": 0.0061,
      "step": 28700
    },
    {
      "epoch": 0.46977010553873844,
      "grad_norm": 0.2515339739204591,
      "learning_rate": 6.382124163361413e-06,
      "loss": 0.0044,
      "step": 28710
    },
    {
      "epoch": 0.46993373148981427,
      "grad_norm": 0.12116647769642865,
      "learning_rate": 6.379379376679225e-06,
      "loss": 0.0044,
      "step": 28720
    },
    {
      "epoch": 0.47009735744089015,
      "grad_norm": 0.19943133060781973,
      "learning_rate": 6.376634139998909e-06,
      "loss": 0.004,
      "step": 28730
    },
    {
      "epoch": 0.470260983391966,
      "grad_norm": 0.12343505694302365,
      "learning_rate": 6.373888454216048e-06,
      "loss": 0.0018,
      "step": 28740
    },
    {
      "epoch": 0.4704246093430418,
      "grad_norm": 0.2743382043042762,
      "learning_rate": 6.371142320226378e-06,
      "loss": 0.006,
      "step": 28750
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.12485617833650155,
      "learning_rate": 6.368395738925774e-06,
      "loss": 0.003,
      "step": 28760
    },
    {
      "epoch": 0.47075186124519347,
      "grad_norm": 0.08022424403163103,
      "learning_rate": 6.36564871121026e-06,
      "loss": 0.0021,
      "step": 28770
    },
    {
      "epoch": 0.47091548719626936,
      "grad_norm": 0.30242117509979616,
      "learning_rate": 6.3629012379760045e-06,
      "loss": 0.0045,
      "step": 28780
    },
    {
      "epoch": 0.4710791131473452,
      "grad_norm": 0.41906582507647294,
      "learning_rate": 6.360153320119325e-06,
      "loss": 0.0045,
      "step": 28790
    },
    {
      "epoch": 0.471242739098421,
      "grad_norm": 0.24362118206744512,
      "learning_rate": 6.357404958536675e-06,
      "loss": 0.0052,
      "step": 28800
    },
    {
      "epoch": 0.47140636504949684,
      "grad_norm": 0.2430525258174969,
      "learning_rate": 6.354656154124664e-06,
      "loss": 0.0028,
      "step": 28810
    },
    {
      "epoch": 0.4715699910005727,
      "grad_norm": 0.5686912092833595,
      "learning_rate": 6.351906907780041e-06,
      "loss": 0.0048,
      "step": 28820
    },
    {
      "epoch": 0.4717336169516485,
      "grad_norm": 0.27074485600419673,
      "learning_rate": 6.349157220399696e-06,
      "loss": 0.004,
      "step": 28830
    },
    {
      "epoch": 0.4718972429027244,
      "grad_norm": 0.09511359161388898,
      "learning_rate": 6.346407092880668e-06,
      "loss": 0.0109,
      "step": 28840
    },
    {
      "epoch": 0.4720608688538002,
      "grad_norm": 0.2617628188305595,
      "learning_rate": 6.343656526120136e-06,
      "loss": 0.003,
      "step": 28850
    },
    {
      "epoch": 0.47222449480487605,
      "grad_norm": 0.2486964205057107,
      "learning_rate": 6.340905521015425e-06,
      "loss": 0.0037,
      "step": 28860
    },
    {
      "epoch": 0.4723881207559519,
      "grad_norm": 0.14914835295600168,
      "learning_rate": 6.3381540784640015e-06,
      "loss": 0.0031,
      "step": 28870
    },
    {
      "epoch": 0.4725517467070277,
      "grad_norm": 0.04547128775840444,
      "learning_rate": 6.335402199363475e-06,
      "loss": 0.0063,
      "step": 28880
    },
    {
      "epoch": 0.4727153726581036,
      "grad_norm": 0.04228099950194356,
      "learning_rate": 6.332649884611596e-06,
      "loss": 0.0037,
      "step": 28890
    },
    {
      "epoch": 0.4728789986091794,
      "grad_norm": 0.12810134150955807,
      "learning_rate": 6.329897135106261e-06,
      "loss": 0.0033,
      "step": 28900
    },
    {
      "epoch": 0.47304262456025525,
      "grad_norm": 0.08144098121994872,
      "learning_rate": 6.327143951745505e-06,
      "loss": 0.0031,
      "step": 28910
    },
    {
      "epoch": 0.4732062505113311,
      "grad_norm": 0.41958007681560294,
      "learning_rate": 6.324390335427503e-06,
      "loss": 0.0077,
      "step": 28920
    },
    {
      "epoch": 0.4733698764624069,
      "grad_norm": 0.20645399551116456,
      "learning_rate": 6.321636287050576e-06,
      "loss": 0.0036,
      "step": 28930
    },
    {
      "epoch": 0.4735335024134828,
      "grad_norm": 0.15725748055727215,
      "learning_rate": 6.318881807513184e-06,
      "loss": 0.0055,
      "step": 28940
    },
    {
      "epoch": 0.47369712836455863,
      "grad_norm": 0.2545139945740543,
      "learning_rate": 6.316126897713926e-06,
      "loss": 0.0054,
      "step": 28950
    },
    {
      "epoch": 0.47386075431563446,
      "grad_norm": 0.27174063945616145,
      "learning_rate": 6.313371558551543e-06,
      "loss": 0.0036,
      "step": 28960
    },
    {
      "epoch": 0.4740243802667103,
      "grad_norm": 0.05672784788987418,
      "learning_rate": 6.310615790924917e-06,
      "loss": 0.0019,
      "step": 28970
    },
    {
      "epoch": 0.4741880062177861,
      "grad_norm": 0.2603619501217855,
      "learning_rate": 6.307859595733065e-06,
      "loss": 0.0029,
      "step": 28980
    },
    {
      "epoch": 0.474351632168862,
      "grad_norm": 0.1939413865923206,
      "learning_rate": 6.30510297387515e-06,
      "loss": 0.005,
      "step": 28990
    },
    {
      "epoch": 0.47451525811993783,
      "grad_norm": 0.13642460594537312,
      "learning_rate": 6.302345926250471e-06,
      "loss": 0.0043,
      "step": 29000
    },
    {
      "epoch": 0.47467888407101366,
      "grad_norm": 0.10448644155649928,
      "learning_rate": 6.299588453758464e-06,
      "loss": 0.0037,
      "step": 29010
    },
    {
      "epoch": 0.4748425100220895,
      "grad_norm": 0.13191174318270216,
      "learning_rate": 6.296830557298706e-06,
      "loss": 0.0032,
      "step": 29020
    },
    {
      "epoch": 0.4750061359731653,
      "grad_norm": 0.09503486937519907,
      "learning_rate": 6.294072237770914e-06,
      "loss": 0.0041,
      "step": 29030
    },
    {
      "epoch": 0.4751697619242412,
      "grad_norm": 0.3307413030033592,
      "learning_rate": 6.29131349607494e-06,
      "loss": 0.0026,
      "step": 29040
    },
    {
      "epoch": 0.47533338787531704,
      "grad_norm": 0.28180421914480325,
      "learning_rate": 6.288554333110774e-06,
      "loss": 0.0043,
      "step": 29050
    },
    {
      "epoch": 0.47549701382639287,
      "grad_norm": 0.3900740015344946,
      "learning_rate": 6.285794749778544e-06,
      "loss": 0.0044,
      "step": 29060
    },
    {
      "epoch": 0.4756606397774687,
      "grad_norm": 0.45925117285410255,
      "learning_rate": 6.283034746978518e-06,
      "loss": 0.0032,
      "step": 29070
    },
    {
      "epoch": 0.4758242657285445,
      "grad_norm": 0.0756195075816101,
      "learning_rate": 6.280274325611093e-06,
      "loss": 0.0028,
      "step": 29080
    },
    {
      "epoch": 0.4759878916796204,
      "grad_norm": 0.278711814137114,
      "learning_rate": 6.277513486576813e-06,
      "loss": 0.0056,
      "step": 29090
    },
    {
      "epoch": 0.47615151763069624,
      "grad_norm": 0.29830645637032976,
      "learning_rate": 6.274752230776348e-06,
      "loss": 0.0028,
      "step": 29100
    },
    {
      "epoch": 0.47631514358177207,
      "grad_norm": 0.09497249015083563,
      "learning_rate": 6.271990559110514e-06,
      "loss": 0.0035,
      "step": 29110
    },
    {
      "epoch": 0.4764787695328479,
      "grad_norm": 0.1344227517782784,
      "learning_rate": 6.269228472480253e-06,
      "loss": 0.0041,
      "step": 29120
    },
    {
      "epoch": 0.47664239548392373,
      "grad_norm": 0.14281697068255028,
      "learning_rate": 6.266465971786651e-06,
      "loss": 0.0043,
      "step": 29130
    },
    {
      "epoch": 0.4768060214349996,
      "grad_norm": 0.37209888683235626,
      "learning_rate": 6.26370305793092e-06,
      "loss": 0.0052,
      "step": 29140
    },
    {
      "epoch": 0.47696964738607545,
      "grad_norm": 0.18622507040117436,
      "learning_rate": 6.260939731814416e-06,
      "loss": 0.0043,
      "step": 29150
    },
    {
      "epoch": 0.4771332733371513,
      "grad_norm": 0.07666704675208601,
      "learning_rate": 6.258175994338625e-06,
      "loss": 0.0027,
      "step": 29160
    },
    {
      "epoch": 0.4772968992882271,
      "grad_norm": 0.08999712346659572,
      "learning_rate": 6.255411846405164e-06,
      "loss": 0.003,
      "step": 29170
    },
    {
      "epoch": 0.47746052523930294,
      "grad_norm": 0.16231717123856423,
      "learning_rate": 6.252647288915792e-06,
      "loss": 0.0043,
      "step": 29180
    },
    {
      "epoch": 0.4776241511903788,
      "grad_norm": 0.1977245389799726,
      "learning_rate": 6.249882322772393e-06,
      "loss": 0.0031,
      "step": 29190
    },
    {
      "epoch": 0.47778777714145465,
      "grad_norm": 0.11919245973263586,
      "learning_rate": 6.24711694887699e-06,
      "loss": 0.0068,
      "step": 29200
    },
    {
      "epoch": 0.4779514030925305,
      "grad_norm": 0.08787363425062858,
      "learning_rate": 6.24435116813174e-06,
      "loss": 0.0029,
      "step": 29210
    },
    {
      "epoch": 0.4781150290436063,
      "grad_norm": 0.2766724736503148,
      "learning_rate": 6.241584981438924e-06,
      "loss": 0.0036,
      "step": 29220
    },
    {
      "epoch": 0.47827865499468214,
      "grad_norm": 0.357116634700747,
      "learning_rate": 6.2388183897009655e-06,
      "loss": 0.0027,
      "step": 29230
    },
    {
      "epoch": 0.47844228094575797,
      "grad_norm": 0.32954126294664965,
      "learning_rate": 6.236051393820418e-06,
      "loss": 0.004,
      "step": 29240
    },
    {
      "epoch": 0.47860590689683385,
      "grad_norm": 0.049986108641716494,
      "learning_rate": 6.233283994699961e-06,
      "loss": 0.0031,
      "step": 29250
    },
    {
      "epoch": 0.4787695328479097,
      "grad_norm": 0.09923681602857796,
      "learning_rate": 6.230516193242411e-06,
      "loss": 0.0054,
      "step": 29260
    },
    {
      "epoch": 0.4789331587989855,
      "grad_norm": 0.14121999737143903,
      "learning_rate": 6.227747990350714e-06,
      "loss": 0.0036,
      "step": 29270
    },
    {
      "epoch": 0.47909678475006134,
      "grad_norm": 0.45301113150038785,
      "learning_rate": 6.2249793869279476e-06,
      "loss": 0.0054,
      "step": 29280
    },
    {
      "epoch": 0.4792604107011372,
      "grad_norm": 0.060497857501768164,
      "learning_rate": 6.222210383877321e-06,
      "loss": 0.0027,
      "step": 29290
    },
    {
      "epoch": 0.47942403665221306,
      "grad_norm": 0.06584707053590898,
      "learning_rate": 6.21944098210217e-06,
      "loss": 0.0029,
      "step": 29300
    },
    {
      "epoch": 0.4795876626032889,
      "grad_norm": 0.11174683355609316,
      "learning_rate": 6.216671182505963e-06,
      "loss": 0.0026,
      "step": 29310
    },
    {
      "epoch": 0.4797512885543647,
      "grad_norm": 0.2155194570913787,
      "learning_rate": 6.2139009859923e-06,
      "loss": 0.0025,
      "step": 29320
    },
    {
      "epoch": 0.47991491450544055,
      "grad_norm": 0.16046648580711412,
      "learning_rate": 6.2111303934649094e-06,
      "loss": 0.0053,
      "step": 29330
    },
    {
      "epoch": 0.4800785404565164,
      "grad_norm": 0.11485991252632435,
      "learning_rate": 6.208359405827644e-06,
      "loss": 0.0035,
      "step": 29340
    },
    {
      "epoch": 0.48024216640759226,
      "grad_norm": 0.05619253121699922,
      "learning_rate": 6.2055880239844935e-06,
      "loss": 0.0051,
      "step": 29350
    },
    {
      "epoch": 0.4804057923586681,
      "grad_norm": 0.21659237643116297,
      "learning_rate": 6.20281624883957e-06,
      "loss": 0.0034,
      "step": 29360
    },
    {
      "epoch": 0.4805694183097439,
      "grad_norm": 0.23762841550220795,
      "learning_rate": 6.200044081297117e-06,
      "loss": 0.004,
      "step": 29370
    },
    {
      "epoch": 0.48073304426081975,
      "grad_norm": 0.21476950374828657,
      "learning_rate": 6.197271522261505e-06,
      "loss": 0.0031,
      "step": 29380
    },
    {
      "epoch": 0.4808966702118956,
      "grad_norm": 0.32604097466934767,
      "learning_rate": 6.194498572637233e-06,
      "loss": 0.0033,
      "step": 29390
    },
    {
      "epoch": 0.48106029616297147,
      "grad_norm": 0.11988467414905409,
      "learning_rate": 6.191725233328925e-06,
      "loss": 0.003,
      "step": 29400
    },
    {
      "epoch": 0.4812239221140473,
      "grad_norm": 0.03519817068283349,
      "learning_rate": 6.188951505241336e-06,
      "loss": 0.0023,
      "step": 29410
    },
    {
      "epoch": 0.4813875480651231,
      "grad_norm": 0.2672677710543047,
      "learning_rate": 6.186177389279344e-06,
      "loss": 0.0041,
      "step": 29420
    },
    {
      "epoch": 0.48155117401619896,
      "grad_norm": 0.09367708117426597,
      "learning_rate": 6.183402886347956e-06,
      "loss": 0.0035,
      "step": 29430
    },
    {
      "epoch": 0.4817147999672748,
      "grad_norm": 0.16038023134905086,
      "learning_rate": 6.180627997352304e-06,
      "loss": 0.0032,
      "step": 29440
    },
    {
      "epoch": 0.4818784259183507,
      "grad_norm": 0.1274330605509065,
      "learning_rate": 6.177852723197648e-06,
      "loss": 0.0025,
      "step": 29450
    },
    {
      "epoch": 0.4820420518694265,
      "grad_norm": 0.0997741782596983,
      "learning_rate": 6.17507706478937e-06,
      "loss": 0.0039,
      "step": 29460
    },
    {
      "epoch": 0.48220567782050233,
      "grad_norm": 0.1274370438987351,
      "learning_rate": 6.172301023032982e-06,
      "loss": 0.0046,
      "step": 29470
    },
    {
      "epoch": 0.48236930377157816,
      "grad_norm": 0.12090327294473836,
      "learning_rate": 6.169524598834115e-06,
      "loss": 0.0053,
      "step": 29480
    },
    {
      "epoch": 0.482532929722654,
      "grad_norm": 0.17800858058758556,
      "learning_rate": 6.166747793098532e-06,
      "loss": 0.0026,
      "step": 29490
    },
    {
      "epoch": 0.4826965556737299,
      "grad_norm": 0.3117835369535704,
      "learning_rate": 6.163970606732114e-06,
      "loss": 0.004,
      "step": 29500
    },
    {
      "epoch": 0.4828601816248057,
      "grad_norm": 0.22595653630859455,
      "learning_rate": 6.1611930406408725e-06,
      "loss": 0.0048,
      "step": 29510
    },
    {
      "epoch": 0.48302380757588154,
      "grad_norm": 0.2665941274469625,
      "learning_rate": 6.158415095730935e-06,
      "loss": 0.0044,
      "step": 29520
    },
    {
      "epoch": 0.48318743352695737,
      "grad_norm": 0.1297586895289029,
      "learning_rate": 6.155636772908559e-06,
      "loss": 0.0032,
      "step": 29530
    },
    {
      "epoch": 0.4833510594780332,
      "grad_norm": 0.1331237484173101,
      "learning_rate": 6.152858073080122e-06,
      "loss": 0.0032,
      "step": 29540
    },
    {
      "epoch": 0.4835146854291091,
      "grad_norm": 0.17594100183081174,
      "learning_rate": 6.150078997152129e-06,
      "loss": 0.0028,
      "step": 29550
    },
    {
      "epoch": 0.4836783113801849,
      "grad_norm": 0.18970555062186414,
      "learning_rate": 6.147299546031199e-06,
      "loss": 0.0045,
      "step": 29560
    },
    {
      "epoch": 0.48384193733126074,
      "grad_norm": 0.08830290239249784,
      "learning_rate": 6.1445197206240835e-06,
      "loss": 0.0038,
      "step": 29570
    },
    {
      "epoch": 0.48400556328233657,
      "grad_norm": 0.147673850374333,
      "learning_rate": 6.141739521837651e-06,
      "loss": 0.0042,
      "step": 29580
    },
    {
      "epoch": 0.4841691892334124,
      "grad_norm": 0.23240585086855328,
      "learning_rate": 6.138958950578888e-06,
      "loss": 0.0037,
      "step": 29590
    },
    {
      "epoch": 0.4843328151844883,
      "grad_norm": 0.13426945198825696,
      "learning_rate": 6.1361780077549115e-06,
      "loss": 0.0059,
      "step": 29600
    },
    {
      "epoch": 0.4844964411355641,
      "grad_norm": 0.3009426394403168,
      "learning_rate": 6.133396694272953e-06,
      "loss": 0.0025,
      "step": 29610
    },
    {
      "epoch": 0.48466006708663995,
      "grad_norm": 0.018990673250287287,
      "learning_rate": 6.130615011040365e-06,
      "loss": 0.0028,
      "step": 29620
    },
    {
      "epoch": 0.4848236930377158,
      "grad_norm": 0.2862320731774738,
      "learning_rate": 6.127832958964626e-06,
      "loss": 0.0037,
      "step": 29630
    },
    {
      "epoch": 0.4849873189887916,
      "grad_norm": 0.13973691724306697,
      "learning_rate": 6.12505053895333e-06,
      "loss": 0.004,
      "step": 29640
    },
    {
      "epoch": 0.4851509449398675,
      "grad_norm": 0.142199815006184,
      "learning_rate": 6.122267751914189e-06,
      "loss": 0.0031,
      "step": 29650
    },
    {
      "epoch": 0.4853145708909433,
      "grad_norm": 0.31804166612815354,
      "learning_rate": 6.119484598755043e-06,
      "loss": 0.0035,
      "step": 29660
    },
    {
      "epoch": 0.48547819684201915,
      "grad_norm": 0.15919738313465603,
      "learning_rate": 6.116701080383846e-06,
      "loss": 0.0035,
      "step": 29670
    },
    {
      "epoch": 0.485641822793095,
      "grad_norm": 0.19162922534280483,
      "learning_rate": 6.113917197708667e-06,
      "loss": 0.0036,
      "step": 29680
    },
    {
      "epoch": 0.4858054487441708,
      "grad_norm": 0.10584382125969954,
      "learning_rate": 6.111132951637705e-06,
      "loss": 0.0057,
      "step": 29690
    },
    {
      "epoch": 0.48596907469524664,
      "grad_norm": 0.13336140197536098,
      "learning_rate": 6.108348343079265e-06,
      "loss": 0.0028,
      "step": 29700
    },
    {
      "epoch": 0.4861327006463225,
      "grad_norm": 0.25763108446364763,
      "learning_rate": 6.105563372941779e-06,
      "loss": 0.0041,
      "step": 29710
    },
    {
      "epoch": 0.48629632659739835,
      "grad_norm": 0.08544780138585156,
      "learning_rate": 6.102778042133796e-06,
      "loss": 0.0044,
      "step": 29720
    },
    {
      "epoch": 0.4864599525484742,
      "grad_norm": 0.19120116610632867,
      "learning_rate": 6.099992351563977e-06,
      "loss": 0.0033,
      "step": 29730
    },
    {
      "epoch": 0.48662357849955,
      "grad_norm": 0.29861792181844093,
      "learning_rate": 6.097206302141106e-06,
      "loss": 0.0044,
      "step": 29740
    },
    {
      "epoch": 0.48678720445062584,
      "grad_norm": 0.9647579872544751,
      "learning_rate": 6.094419894774085e-06,
      "loss": 0.0039,
      "step": 29750
    },
    {
      "epoch": 0.48695083040170173,
      "grad_norm": 0.08814048699267185,
      "learning_rate": 6.091633130371927e-06,
      "loss": 0.0028,
      "step": 29760
    },
    {
      "epoch": 0.48711445635277756,
      "grad_norm": 0.1252353890662789,
      "learning_rate": 6.088846009843764e-06,
      "loss": 0.0028,
      "step": 29770
    },
    {
      "epoch": 0.4872780823038534,
      "grad_norm": 0.4355197354594601,
      "learning_rate": 6.0860585340988485e-06,
      "loss": 0.0054,
      "step": 29780
    },
    {
      "epoch": 0.4874417082549292,
      "grad_norm": 0.16955114705321353,
      "learning_rate": 6.083270704046542e-06,
      "loss": 0.0019,
      "step": 29790
    },
    {
      "epoch": 0.48760533420600505,
      "grad_norm": 0.2540289533131497,
      "learning_rate": 6.0804825205963245e-06,
      "loss": 0.0062,
      "step": 29800
    },
    {
      "epoch": 0.48776896015708093,
      "grad_norm": 0.09725313204975622,
      "learning_rate": 6.077693984657795e-06,
      "loss": 0.0026,
      "step": 29810
    },
    {
      "epoch": 0.48793258610815676,
      "grad_norm": 0.05517603762687097,
      "learning_rate": 6.074905097140659e-06,
      "loss": 0.0026,
      "step": 29820
    },
    {
      "epoch": 0.4880962120592326,
      "grad_norm": 0.13683282067049105,
      "learning_rate": 6.0721158589547455e-06,
      "loss": 0.003,
      "step": 29830
    },
    {
      "epoch": 0.4882598380103084,
      "grad_norm": 0.22452494900506506,
      "learning_rate": 6.069326271009996e-06,
      "loss": 0.0037,
      "step": 29840
    },
    {
      "epoch": 0.48842346396138425,
      "grad_norm": 0.2801635301143507,
      "learning_rate": 6.066536334216458e-06,
      "loss": 0.0034,
      "step": 29850
    },
    {
      "epoch": 0.48858708991246014,
      "grad_norm": 0.16596463883097873,
      "learning_rate": 6.0637460494843035e-06,
      "loss": 0.0035,
      "step": 29860
    },
    {
      "epoch": 0.48875071586353597,
      "grad_norm": 0.43837351174035766,
      "learning_rate": 6.060955417723813e-06,
      "loss": 0.0042,
      "step": 29870
    },
    {
      "epoch": 0.4889143418146118,
      "grad_norm": 0.11003383109654377,
      "learning_rate": 6.058164439845378e-06,
      "loss": 0.0039,
      "step": 29880
    },
    {
      "epoch": 0.4890779677656876,
      "grad_norm": 0.11727728463669576,
      "learning_rate": 6.05537311675951e-06,
      "loss": 0.0034,
      "step": 29890
    },
    {
      "epoch": 0.48924159371676346,
      "grad_norm": 0.13843011908947217,
      "learning_rate": 6.0525814493768266e-06,
      "loss": 0.0027,
      "step": 29900
    },
    {
      "epoch": 0.48940521966783934,
      "grad_norm": 0.15742073352125482,
      "learning_rate": 6.049789438608058e-06,
      "loss": 0.0031,
      "step": 29910
    },
    {
      "epoch": 0.48956884561891517,
      "grad_norm": 0.35526797173624325,
      "learning_rate": 6.0469970853640516e-06,
      "loss": 0.003,
      "step": 29920
    },
    {
      "epoch": 0.489732471569991,
      "grad_norm": 0.5083059073977578,
      "learning_rate": 6.044204390555763e-06,
      "loss": 0.003,
      "step": 29930
    },
    {
      "epoch": 0.48989609752106683,
      "grad_norm": 0.06707101445552398,
      "learning_rate": 6.041411355094255e-06,
      "loss": 0.0021,
      "step": 29940
    },
    {
      "epoch": 0.49005972347214266,
      "grad_norm": 0.19851638352856973,
      "learning_rate": 6.038617979890713e-06,
      "loss": 0.0033,
      "step": 29950
    },
    {
      "epoch": 0.49022334942321855,
      "grad_norm": 0.140088111498455,
      "learning_rate": 6.0358242658564224e-06,
      "loss": 0.002,
      "step": 29960
    },
    {
      "epoch": 0.4903869753742944,
      "grad_norm": 0.10289592596656784,
      "learning_rate": 6.033030213902782e-06,
      "loss": 0.004,
      "step": 29970
    },
    {
      "epoch": 0.4905506013253702,
      "grad_norm": 0.2299604978984176,
      "learning_rate": 6.030235824941304e-06,
      "loss": 0.0037,
      "step": 29980
    },
    {
      "epoch": 0.49071422727644604,
      "grad_norm": 0.3168382458986692,
      "learning_rate": 6.027441099883607e-06,
      "loss": 0.004,
      "step": 29990
    },
    {
      "epoch": 0.49087785322752187,
      "grad_norm": 0.23270788130372036,
      "learning_rate": 6.024646039641422e-06,
      "loss": 0.0021,
      "step": 30000
    },
    {
      "epoch": 0.49104147917859775,
      "grad_norm": 0.3560858473976975,
      "learning_rate": 6.021850645126586e-06,
      "loss": 0.0025,
      "step": 30010
    },
    {
      "epoch": 0.4912051051296736,
      "grad_norm": 0.40369975814106696,
      "learning_rate": 6.019054917251049e-06,
      "loss": 0.0042,
      "step": 30020
    },
    {
      "epoch": 0.4913687310807494,
      "grad_norm": 0.4963912779417439,
      "learning_rate": 6.016258856926868e-06,
      "loss": 0.0032,
      "step": 30030
    },
    {
      "epoch": 0.49153235703182524,
      "grad_norm": 0.23538670014973337,
      "learning_rate": 6.013462465066205e-06,
      "loss": 0.0049,
      "step": 30040
    },
    {
      "epoch": 0.49169598298290107,
      "grad_norm": 0.09055079974655157,
      "learning_rate": 6.010665742581336e-06,
      "loss": 0.0023,
      "step": 30050
    },
    {
      "epoch": 0.49185960893397696,
      "grad_norm": 0.17989733674634537,
      "learning_rate": 6.007868690384643e-06,
      "loss": 0.005,
      "step": 30060
    },
    {
      "epoch": 0.4920232348850528,
      "grad_norm": 0.16300437269371626,
      "learning_rate": 6.005071309388612e-06,
      "loss": 0.0035,
      "step": 30070
    },
    {
      "epoch": 0.4921868608361286,
      "grad_norm": 0.27451328919098483,
      "learning_rate": 6.002273600505842e-06,
      "loss": 0.0031,
      "step": 30080
    },
    {
      "epoch": 0.49235048678720444,
      "grad_norm": 0.09989386063104876,
      "learning_rate": 5.999475564649035e-06,
      "loss": 0.0038,
      "step": 30090
    },
    {
      "epoch": 0.4925141127382803,
      "grad_norm": 0.10637711464226826,
      "learning_rate": 5.996677202731e-06,
      "loss": 0.0022,
      "step": 30100
    },
    {
      "epoch": 0.49267773868935616,
      "grad_norm": 0.29259948733056984,
      "learning_rate": 5.993878515664654e-06,
      "loss": 0.0032,
      "step": 30110
    },
    {
      "epoch": 0.492841364640432,
      "grad_norm": 0.1245250294705668,
      "learning_rate": 5.991079504363019e-06,
      "loss": 0.0031,
      "step": 30120
    },
    {
      "epoch": 0.4930049905915078,
      "grad_norm": 0.16003737625109335,
      "learning_rate": 5.988280169739221e-06,
      "loss": 0.0029,
      "step": 30130
    },
    {
      "epoch": 0.49316861654258365,
      "grad_norm": 0.08849713684105148,
      "learning_rate": 5.985480512706496e-06,
      "loss": 0.0029,
      "step": 30140
    },
    {
      "epoch": 0.4933322424936595,
      "grad_norm": 0.3895293210767575,
      "learning_rate": 5.982680534178182e-06,
      "loss": 0.0046,
      "step": 30150
    },
    {
      "epoch": 0.4934958684447353,
      "grad_norm": 0.2956355961203383,
      "learning_rate": 5.979880235067722e-06,
      "loss": 0.004,
      "step": 30160
    },
    {
      "epoch": 0.4936594943958112,
      "grad_norm": 0.21900161855985223,
      "learning_rate": 5.977079616288666e-06,
      "loss": 0.0034,
      "step": 30170
    },
    {
      "epoch": 0.493823120346887,
      "grad_norm": 0.13293376446720137,
      "learning_rate": 5.974278678754661e-06,
      "loss": 0.0034,
      "step": 30180
    },
    {
      "epoch": 0.49398674629796285,
      "grad_norm": 0.1109432136093818,
      "learning_rate": 5.9714774233794685e-06,
      "loss": 0.0034,
      "step": 30190
    },
    {
      "epoch": 0.4941503722490387,
      "grad_norm": 0.1308311551348442,
      "learning_rate": 5.968675851076949e-06,
      "loss": 0.0038,
      "step": 30200
    },
    {
      "epoch": 0.4943139982001145,
      "grad_norm": 0.12716538908467298,
      "learning_rate": 5.965873962761061e-06,
      "loss": 0.0027,
      "step": 30210
    },
    {
      "epoch": 0.4944776241511904,
      "grad_norm": 0.10859512178270475,
      "learning_rate": 5.963071759345874e-06,
      "loss": 0.0041,
      "step": 30220
    },
    {
      "epoch": 0.49464125010226623,
      "grad_norm": 0.16979859564474095,
      "learning_rate": 5.96026924174556e-06,
      "loss": 0.0039,
      "step": 30230
    },
    {
      "epoch": 0.49480487605334206,
      "grad_norm": 0.10969800358731176,
      "learning_rate": 5.957466410874385e-06,
      "loss": 0.0035,
      "step": 30240
    },
    {
      "epoch": 0.4949685020044179,
      "grad_norm": 0.1816067515391278,
      "learning_rate": 5.954663267646726e-06,
      "loss": 0.003,
      "step": 30250
    },
    {
      "epoch": 0.4951321279554937,
      "grad_norm": 0.10358004337229164,
      "learning_rate": 5.951859812977062e-06,
      "loss": 0.0039,
      "step": 30260
    },
    {
      "epoch": 0.4952957539065696,
      "grad_norm": 0.14893742261748405,
      "learning_rate": 5.949056047779964e-06,
      "loss": 0.0027,
      "step": 30270
    },
    {
      "epoch": 0.49545937985764543,
      "grad_norm": 0.20663819327916472,
      "learning_rate": 5.946251972970115e-06,
      "loss": 0.002,
      "step": 30280
    },
    {
      "epoch": 0.49562300580872126,
      "grad_norm": 0.2512481243531907,
      "learning_rate": 5.9434475894622965e-06,
      "loss": 0.0032,
      "step": 30290
    },
    {
      "epoch": 0.4957866317597971,
      "grad_norm": 0.31103661899216584,
      "learning_rate": 5.9406428981713835e-06,
      "loss": 0.0024,
      "step": 30300
    },
    {
      "epoch": 0.4959502577108729,
      "grad_norm": 0.27810187298058087,
      "learning_rate": 5.937837900012361e-06,
      "loss": 0.0036,
      "step": 30310
    },
    {
      "epoch": 0.4961138836619488,
      "grad_norm": 0.12598607243994947,
      "learning_rate": 5.935032595900311e-06,
      "loss": 0.0032,
      "step": 30320
    },
    {
      "epoch": 0.49627750961302464,
      "grad_norm": 0.23324557797760348,
      "learning_rate": 5.932226986750409e-06,
      "loss": 0.0027,
      "step": 30330
    },
    {
      "epoch": 0.49644113556410047,
      "grad_norm": 0.25153654471727305,
      "learning_rate": 5.929421073477941e-06,
      "loss": 0.004,
      "step": 30340
    },
    {
      "epoch": 0.4966047615151763,
      "grad_norm": 0.16419486902282132,
      "learning_rate": 5.926614856998284e-06,
      "loss": 0.0042,
      "step": 30350
    },
    {
      "epoch": 0.4967683874662521,
      "grad_norm": 0.05139620099146174,
      "learning_rate": 5.923808338226916e-06,
      "loss": 0.0037,
      "step": 30360
    },
    {
      "epoch": 0.496932013417328,
      "grad_norm": 0.07125926653504847,
      "learning_rate": 5.921001518079416e-06,
      "loss": 0.002,
      "step": 30370
    },
    {
      "epoch": 0.49709563936840384,
      "grad_norm": 0.17962100163706446,
      "learning_rate": 5.918194397471459e-06,
      "loss": 0.0038,
      "step": 30380
    },
    {
      "epoch": 0.49725926531947967,
      "grad_norm": 0.10040518531269127,
      "learning_rate": 5.915386977318815e-06,
      "loss": 0.0056,
      "step": 30390
    },
    {
      "epoch": 0.4974228912705555,
      "grad_norm": 0.2686932315579384,
      "learning_rate": 5.912579258537361e-06,
      "loss": 0.004,
      "step": 30400
    },
    {
      "epoch": 0.49758651722163133,
      "grad_norm": 0.10463183117508539,
      "learning_rate": 5.909771242043061e-06,
      "loss": 0.0045,
      "step": 30410
    },
    {
      "epoch": 0.4977501431727072,
      "grad_norm": 0.3012263587092307,
      "learning_rate": 5.906962928751983e-06,
      "loss": 0.0062,
      "step": 30420
    },
    {
      "epoch": 0.49791376912378305,
      "grad_norm": 0.09366529301477908,
      "learning_rate": 5.904154319580289e-06,
      "loss": 0.0031,
      "step": 30430
    },
    {
      "epoch": 0.4980773950748589,
      "grad_norm": 0.28861923478639584,
      "learning_rate": 5.901345415444237e-06,
      "loss": 0.0019,
      "step": 30440
    },
    {
      "epoch": 0.4982410210259347,
      "grad_norm": 0.11662173066465188,
      "learning_rate": 5.898536217260183e-06,
      "loss": 0.0049,
      "step": 30450
    },
    {
      "epoch": 0.49840464697701053,
      "grad_norm": 0.27824799281949225,
      "learning_rate": 5.895726725944579e-06,
      "loss": 0.003,
      "step": 30460
    },
    {
      "epoch": 0.4985682729280864,
      "grad_norm": 0.11498238988639774,
      "learning_rate": 5.892916942413974e-06,
      "loss": 0.0032,
      "step": 30470
    },
    {
      "epoch": 0.49873189887916225,
      "grad_norm": 0.12476545299725358,
      "learning_rate": 5.8901068675850035e-06,
      "loss": 0.0034,
      "step": 30480
    },
    {
      "epoch": 0.4988955248302381,
      "grad_norm": 0.1434869372567432,
      "learning_rate": 5.887296502374411e-06,
      "loss": 0.0023,
      "step": 30490
    },
    {
      "epoch": 0.4990591507813139,
      "grad_norm": 0.1214203514478344,
      "learning_rate": 5.884485847699026e-06,
      "loss": 0.0046,
      "step": 30500
    },
    {
      "epoch": 0.49922277673238974,
      "grad_norm": 0.16798148312028227,
      "learning_rate": 5.8816749044757746e-06,
      "loss": 0.0043,
      "step": 30510
    },
    {
      "epoch": 0.4993864026834656,
      "grad_norm": 0.19686930918908155,
      "learning_rate": 5.878863673621678e-06,
      "loss": 0.0038,
      "step": 30520
    },
    {
      "epoch": 0.49955002863454145,
      "grad_norm": 0.1933806256148756,
      "learning_rate": 5.876052156053851e-06,
      "loss": 0.0025,
      "step": 30530
    },
    {
      "epoch": 0.4997136545856173,
      "grad_norm": 0.197958555043247,
      "learning_rate": 5.8732403526895e-06,
      "loss": 0.0036,
      "step": 30540
    },
    {
      "epoch": 0.4998772805366931,
      "grad_norm": 0.18543167326358576,
      "learning_rate": 5.870428264445926e-06,
      "loss": 0.0046,
      "step": 30550
    },
    {
      "epoch": 0.500040906487769,
      "grad_norm": 0.13615105113344544,
      "learning_rate": 5.867615892240526e-06,
      "loss": 0.0038,
      "step": 30560
    },
    {
      "epoch": 0.5002045324388448,
      "grad_norm": 0.06299064743082385,
      "learning_rate": 5.864803236990783e-06,
      "loss": 0.0046,
      "step": 30570
    },
    {
      "epoch": 0.5003681583899207,
      "grad_norm": 0.15832340558482352,
      "learning_rate": 5.86199029961428e-06,
      "loss": 0.0032,
      "step": 30580
    },
    {
      "epoch": 0.5005317843409964,
      "grad_norm": 0.1285976553180435,
      "learning_rate": 5.859177081028684e-06,
      "loss": 0.0032,
      "step": 30590
    },
    {
      "epoch": 0.5006954102920723,
      "grad_norm": 0.10136167967629879,
      "learning_rate": 5.856363582151761e-06,
      "loss": 0.0027,
      "step": 30600
    },
    {
      "epoch": 0.5008590362431482,
      "grad_norm": 0.150559013114819,
      "learning_rate": 5.853549803901362e-06,
      "loss": 0.0044,
      "step": 30610
    },
    {
      "epoch": 0.501022662194224,
      "grad_norm": 0.1888069140603963,
      "learning_rate": 5.850735747195435e-06,
      "loss": 0.0038,
      "step": 30620
    },
    {
      "epoch": 0.5011862881452999,
      "grad_norm": 0.15832518871632617,
      "learning_rate": 5.847921412952016e-06,
      "loss": 0.0031,
      "step": 30630
    },
    {
      "epoch": 0.5013499140963756,
      "grad_norm": 0.0670852691615687,
      "learning_rate": 5.845106802089231e-06,
      "loss": 0.0048,
      "step": 30640
    },
    {
      "epoch": 0.5015135400474515,
      "grad_norm": 0.2013968173486725,
      "learning_rate": 5.842291915525298e-06,
      "loss": 0.0034,
      "step": 30650
    },
    {
      "epoch": 0.5016771659985274,
      "grad_norm": 0.15507268193424312,
      "learning_rate": 5.839476754178522e-06,
      "loss": 0.0047,
      "step": 30660
    },
    {
      "epoch": 0.5018407919496032,
      "grad_norm": 0.04931863680466435,
      "learning_rate": 5.836661318967301e-06,
      "loss": 0.0039,
      "step": 30670
    },
    {
      "epoch": 0.5020044179006791,
      "grad_norm": 0.1579966332258695,
      "learning_rate": 5.833845610810123e-06,
      "loss": 0.0023,
      "step": 30680
    },
    {
      "epoch": 0.5021680438517548,
      "grad_norm": 0.351839065048057,
      "learning_rate": 5.831029630625557e-06,
      "loss": 0.0028,
      "step": 30690
    },
    {
      "epoch": 0.5023316698028307,
      "grad_norm": 0.1270213332537777,
      "learning_rate": 5.828213379332272e-06,
      "loss": 0.0028,
      "step": 30700
    },
    {
      "epoch": 0.5024952957539066,
      "grad_norm": 0.4442225510575845,
      "learning_rate": 5.825396857849022e-06,
      "loss": 0.0063,
      "step": 30710
    },
    {
      "epoch": 0.5026589217049824,
      "grad_norm": 0.02512476076602587,
      "learning_rate": 5.8225800670946404e-06,
      "loss": 0.0044,
      "step": 30720
    },
    {
      "epoch": 0.5028225476560583,
      "grad_norm": 0.1787766257453066,
      "learning_rate": 5.81976300798806e-06,
      "loss": 0.0029,
      "step": 30730
    },
    {
      "epoch": 0.502986173607134,
      "grad_norm": 0.27963491370934557,
      "learning_rate": 5.816945681448296e-06,
      "loss": 0.0025,
      "step": 30740
    },
    {
      "epoch": 0.5031497995582099,
      "grad_norm": 0.1245197589191199,
      "learning_rate": 5.814128088394451e-06,
      "loss": 0.0034,
      "step": 30750
    },
    {
      "epoch": 0.5033134255092858,
      "grad_norm": 0.044971505181116304,
      "learning_rate": 5.811310229745713e-06,
      "loss": 0.0025,
      "step": 30760
    },
    {
      "epoch": 0.5034770514603616,
      "grad_norm": 0.1287923144313871,
      "learning_rate": 5.808492106421363e-06,
      "loss": 0.0032,
      "step": 30770
    },
    {
      "epoch": 0.5036406774114375,
      "grad_norm": 0.08738496979239627,
      "learning_rate": 5.80567371934076e-06,
      "loss": 0.0024,
      "step": 30780
    },
    {
      "epoch": 0.5038043033625133,
      "grad_norm": 0.14446477682128556,
      "learning_rate": 5.8028550694233525e-06,
      "loss": 0.0033,
      "step": 30790
    },
    {
      "epoch": 0.5039679293135891,
      "grad_norm": 0.3124559208936593,
      "learning_rate": 5.8000361575886796e-06,
      "loss": 0.0034,
      "step": 30800
    },
    {
      "epoch": 0.504131555264665,
      "grad_norm": 0.2838342098305543,
      "learning_rate": 5.797216984756357e-06,
      "loss": 0.0022,
      "step": 30810
    },
    {
      "epoch": 0.5042951812157408,
      "grad_norm": 0.12793453552653467,
      "learning_rate": 5.794397551846092e-06,
      "loss": 0.0023,
      "step": 30820
    },
    {
      "epoch": 0.5044588071668167,
      "grad_norm": 0.09883559668496325,
      "learning_rate": 5.791577859777676e-06,
      "loss": 0.0037,
      "step": 30830
    },
    {
      "epoch": 0.5046224331178925,
      "grad_norm": 0.4943226875179745,
      "learning_rate": 5.78875790947098e-06,
      "loss": 0.0031,
      "step": 30840
    },
    {
      "epoch": 0.5047860590689683,
      "grad_norm": 0.1887230231511555,
      "learning_rate": 5.785937701845966e-06,
      "loss": 0.0027,
      "step": 30850
    },
    {
      "epoch": 0.5049496850200442,
      "grad_norm": 0.16701855268765975,
      "learning_rate": 5.783117237822676e-06,
      "loss": 0.0042,
      "step": 30860
    },
    {
      "epoch": 0.50511331097112,
      "grad_norm": 0.217177156739254,
      "learning_rate": 5.780296518321235e-06,
      "loss": 0.0032,
      "step": 30870
    },
    {
      "epoch": 0.5052769369221959,
      "grad_norm": 0.39187331233316197,
      "learning_rate": 5.777475544261857e-06,
      "loss": 0.0037,
      "step": 30880
    },
    {
      "epoch": 0.5054405628732717,
      "grad_norm": 0.14065489905819717,
      "learning_rate": 5.77465431656483e-06,
      "loss": 0.0027,
      "step": 30890
    },
    {
      "epoch": 0.5056041888243475,
      "grad_norm": 0.11569206773931066,
      "learning_rate": 5.771832836150532e-06,
      "loss": 0.0034,
      "step": 30900
    },
    {
      "epoch": 0.5057678147754234,
      "grad_norm": 0.28828213040861433,
      "learning_rate": 5.769011103939422e-06,
      "loss": 0.0032,
      "step": 30910
    },
    {
      "epoch": 0.5059314407264992,
      "grad_norm": 0.3013538155415872,
      "learning_rate": 5.766189120852037e-06,
      "loss": 0.0023,
      "step": 30920
    },
    {
      "epoch": 0.5060950666775751,
      "grad_norm": 0.13195666409540505,
      "learning_rate": 5.7633668878090024e-06,
      "loss": 0.0029,
      "step": 30930
    },
    {
      "epoch": 0.5062586926286509,
      "grad_norm": 0.069755959637198,
      "learning_rate": 5.760544405731021e-06,
      "loss": 0.0023,
      "step": 30940
    },
    {
      "epoch": 0.5064223185797267,
      "grad_norm": 0.31627300981273365,
      "learning_rate": 5.757721675538876e-06,
      "loss": 0.0031,
      "step": 30950
    },
    {
      "epoch": 0.5065859445308026,
      "grad_norm": 0.18569620792723815,
      "learning_rate": 5.754898698153435e-06,
      "loss": 0.0022,
      "step": 30960
    },
    {
      "epoch": 0.5067495704818784,
      "grad_norm": 0.22106776794500202,
      "learning_rate": 5.752075474495646e-06,
      "loss": 0.0025,
      "step": 30970
    },
    {
      "epoch": 0.5069131964329543,
      "grad_norm": 0.13399851166904211,
      "learning_rate": 5.749252005486531e-06,
      "loss": 0.0031,
      "step": 30980
    },
    {
      "epoch": 0.5070768223840301,
      "grad_norm": 0.1318008389377712,
      "learning_rate": 5.7464282920472e-06,
      "loss": 0.0045,
      "step": 30990
    },
    {
      "epoch": 0.507240448335106,
      "grad_norm": 0.12848614648887258,
      "learning_rate": 5.74360433509884e-06,
      "loss": 0.0036,
      "step": 31000
    },
    {
      "epoch": 0.5074040742861818,
      "grad_norm": 0.135060505859908,
      "learning_rate": 5.740780135562716e-06,
      "loss": 0.0022,
      "step": 31010
    },
    {
      "epoch": 0.5075677002372576,
      "grad_norm": 0.40443599113234835,
      "learning_rate": 5.737955694360173e-06,
      "loss": 0.005,
      "step": 31020
    },
    {
      "epoch": 0.5077313261883335,
      "grad_norm": 0.16996645715037323,
      "learning_rate": 5.7351310124126344e-06,
      "loss": 0.0031,
      "step": 31030
    },
    {
      "epoch": 0.5078949521394093,
      "grad_norm": 0.213626336256897,
      "learning_rate": 5.732306090641604e-06,
      "loss": 0.0059,
      "step": 31040
    },
    {
      "epoch": 0.5080585780904852,
      "grad_norm": 0.4174294523480365,
      "learning_rate": 5.729480929968664e-06,
      "loss": 0.0028,
      "step": 31050
    },
    {
      "epoch": 0.508222204041561,
      "grad_norm": 0.20896201209299195,
      "learning_rate": 5.726655531315468e-06,
      "loss": 0.0048,
      "step": 31060
    },
    {
      "epoch": 0.5083858299926368,
      "grad_norm": 0.21591243393165307,
      "learning_rate": 5.723829895603758e-06,
      "loss": 0.0052,
      "step": 31070
    },
    {
      "epoch": 0.5085494559437127,
      "grad_norm": 0.20850838065178728,
      "learning_rate": 5.721004023755346e-06,
      "loss": 0.0027,
      "step": 31080
    },
    {
      "epoch": 0.5087130818947885,
      "grad_norm": 0.09051915422380967,
      "learning_rate": 5.71817791669212e-06,
      "loss": 0.0036,
      "step": 31090
    },
    {
      "epoch": 0.5088767078458644,
      "grad_norm": 0.10766145060043082,
      "learning_rate": 5.71535157533605e-06,
      "loss": 0.0023,
      "step": 31100
    },
    {
      "epoch": 0.5090403337969402,
      "grad_norm": 0.2917857138845481,
      "learning_rate": 5.712525000609179e-06,
      "loss": 0.0048,
      "step": 31110
    },
    {
      "epoch": 0.509203959748016,
      "grad_norm": 0.0639142994402491,
      "learning_rate": 5.709698193433628e-06,
      "loss": 0.0061,
      "step": 31120
    },
    {
      "epoch": 0.5093675856990919,
      "grad_norm": 0.4365875913618581,
      "learning_rate": 5.706871154731592e-06,
      "loss": 0.0038,
      "step": 31130
    },
    {
      "epoch": 0.5095312116501677,
      "grad_norm": 0.14276256643303722,
      "learning_rate": 5.704043885425343e-06,
      "loss": 0.0038,
      "step": 31140
    },
    {
      "epoch": 0.5096948376012436,
      "grad_norm": 0.12471012400758794,
      "learning_rate": 5.701216386437227e-06,
      "loss": 0.002,
      "step": 31150
    },
    {
      "epoch": 0.5098584635523195,
      "grad_norm": 0.26905194375019686,
      "learning_rate": 5.698388658689667e-06,
      "loss": 0.0026,
      "step": 31160
    },
    {
      "epoch": 0.5100220895033952,
      "grad_norm": 0.13676870817226802,
      "learning_rate": 5.695560703105156e-06,
      "loss": 0.0035,
      "step": 31170
    },
    {
      "epoch": 0.5101857154544711,
      "grad_norm": 0.3097708675748771,
      "learning_rate": 5.6927325206062655e-06,
      "loss": 0.0021,
      "step": 31180
    },
    {
      "epoch": 0.5103493414055469,
      "grad_norm": 0.09127883981909614,
      "learning_rate": 5.689904112115643e-06,
      "loss": 0.0026,
      "step": 31190
    },
    {
      "epoch": 0.5105129673566228,
      "grad_norm": 0.1853852710904266,
      "learning_rate": 5.687075478556005e-06,
      "loss": 0.0039,
      "step": 31200
    },
    {
      "epoch": 0.5106765933076987,
      "grad_norm": 0.10094171984173658,
      "learning_rate": 5.684246620850141e-06,
      "loss": 0.0023,
      "step": 31210
    },
    {
      "epoch": 0.5108402192587744,
      "grad_norm": 0.09760976915300021,
      "learning_rate": 5.681417539920919e-06,
      "loss": 0.0028,
      "step": 31220
    },
    {
      "epoch": 0.5110038452098503,
      "grad_norm": 0.08920536898318436,
      "learning_rate": 5.678588236691276e-06,
      "loss": 0.0027,
      "step": 31230
    },
    {
      "epoch": 0.5111674711609261,
      "grad_norm": 0.17848121080390708,
      "learning_rate": 5.67575871208422e-06,
      "loss": 0.0036,
      "step": 31240
    },
    {
      "epoch": 0.511331097112002,
      "grad_norm": 0.09645307961795557,
      "learning_rate": 5.672928967022835e-06,
      "loss": 0.0025,
      "step": 31250
    },
    {
      "epoch": 0.5114947230630777,
      "grad_norm": 0.13369272767976886,
      "learning_rate": 5.670099002430276e-06,
      "loss": 0.0024,
      "step": 31260
    },
    {
      "epoch": 0.5116583490141536,
      "grad_norm": 0.13316919562347096,
      "learning_rate": 5.667268819229767e-06,
      "loss": 0.0037,
      "step": 31270
    },
    {
      "epoch": 0.5118219749652295,
      "grad_norm": 0.209351912947307,
      "learning_rate": 5.664438418344607e-06,
      "loss": 0.0027,
      "step": 31280
    },
    {
      "epoch": 0.5119856009163053,
      "grad_norm": 0.06621207763966933,
      "learning_rate": 5.661607800698162e-06,
      "loss": 0.0025,
      "step": 31290
    },
    {
      "epoch": 0.5121492268673812,
      "grad_norm": 0.21419253463821036,
      "learning_rate": 5.658776967213872e-06,
      "loss": 0.0038,
      "step": 31300
    },
    {
      "epoch": 0.512312852818457,
      "grad_norm": 0.16431342556027037,
      "learning_rate": 5.65594591881525e-06,
      "loss": 0.0033,
      "step": 31310
    },
    {
      "epoch": 0.5124764787695328,
      "grad_norm": 0.1560009045206467,
      "learning_rate": 5.6531146564258695e-06,
      "loss": 0.0039,
      "step": 31320
    },
    {
      "epoch": 0.5126401047206087,
      "grad_norm": 0.19634692794600694,
      "learning_rate": 5.650283180969384e-06,
      "loss": 0.0039,
      "step": 31330
    },
    {
      "epoch": 0.5128037306716845,
      "grad_norm": 0.2277691507722992,
      "learning_rate": 5.64745149336951e-06,
      "loss": 0.003,
      "step": 31340
    },
    {
      "epoch": 0.5129673566227604,
      "grad_norm": 0.2114546235807009,
      "learning_rate": 5.644619594550036e-06,
      "loss": 0.004,
      "step": 31350
    },
    {
      "epoch": 0.5131309825738362,
      "grad_norm": 0.1879014501585944,
      "learning_rate": 5.6417874854348194e-06,
      "loss": 0.006,
      "step": 31360
    },
    {
      "epoch": 0.513294608524912,
      "grad_norm": 0.20446875146640012,
      "learning_rate": 5.638955166947786e-06,
      "loss": 0.0054,
      "step": 31370
    },
    {
      "epoch": 0.5134582344759879,
      "grad_norm": 0.11165868664950125,
      "learning_rate": 5.636122640012928e-06,
      "loss": 0.0034,
      "step": 31380
    },
    {
      "epoch": 0.5136218604270637,
      "grad_norm": 0.1652247375133814,
      "learning_rate": 5.63328990555431e-06,
      "loss": 0.0021,
      "step": 31390
    },
    {
      "epoch": 0.5137854863781396,
      "grad_norm": 0.6646531052235036,
      "learning_rate": 5.63045696449606e-06,
      "loss": 0.0041,
      "step": 31400
    },
    {
      "epoch": 0.5139491123292154,
      "grad_norm": 0.12452814759001027,
      "learning_rate": 5.6276238177623724e-06,
      "loss": 0.0024,
      "step": 31410
    },
    {
      "epoch": 0.5141127382802912,
      "grad_norm": 0.07746065000798537,
      "learning_rate": 5.624790466277516e-06,
      "loss": 0.005,
      "step": 31420
    },
    {
      "epoch": 0.5142763642313671,
      "grad_norm": 0.0750660248586226,
      "learning_rate": 5.621956910965818e-06,
      "loss": 0.0035,
      "step": 31430
    },
    {
      "epoch": 0.5144399901824429,
      "grad_norm": 0.24576498868817262,
      "learning_rate": 5.619123152751677e-06,
      "loss": 0.0023,
      "step": 31440
    },
    {
      "epoch": 0.5146036161335188,
      "grad_norm": 0.1545274315863196,
      "learning_rate": 5.616289192559556e-06,
      "loss": 0.0044,
      "step": 31450
    },
    {
      "epoch": 0.5147672420845946,
      "grad_norm": 0.13503356061505511,
      "learning_rate": 5.613455031313987e-06,
      "loss": 0.0041,
      "step": 31460
    },
    {
      "epoch": 0.5149308680356705,
      "grad_norm": 0.2873909660629769,
      "learning_rate": 5.610620669939561e-06,
      "loss": 0.0036,
      "step": 31470
    },
    {
      "epoch": 0.5150944939867463,
      "grad_norm": 0.17793425866920332,
      "learning_rate": 5.60778610936094e-06,
      "loss": 0.0049,
      "step": 31480
    },
    {
      "epoch": 0.5152581199378221,
      "grad_norm": 0.08700013255323455,
      "learning_rate": 5.60495135050285e-06,
      "loss": 0.01,
      "step": 31490
    },
    {
      "epoch": 0.515421745888898,
      "grad_norm": 0.3682165078819258,
      "learning_rate": 5.60211639429008e-06,
      "loss": 0.0047,
      "step": 31500
    },
    {
      "epoch": 0.5155853718399738,
      "grad_norm": 0.12089238214760419,
      "learning_rate": 5.599281241647484e-06,
      "loss": 0.0031,
      "step": 31510
    },
    {
      "epoch": 0.5157489977910497,
      "grad_norm": 0.37900731566690254,
      "learning_rate": 5.596445893499982e-06,
      "loss": 0.0029,
      "step": 31520
    },
    {
      "epoch": 0.5159126237421255,
      "grad_norm": 0.44071421310508796,
      "learning_rate": 5.593610350772556e-06,
      "loss": 0.0035,
      "step": 31530
    },
    {
      "epoch": 0.5160762496932013,
      "grad_norm": 0.28480737892729513,
      "learning_rate": 5.590774614390249e-06,
      "loss": 0.0025,
      "step": 31540
    },
    {
      "epoch": 0.5162398756442772,
      "grad_norm": 0.2563751400558066,
      "learning_rate": 5.587938685278172e-06,
      "loss": 0.0043,
      "step": 31550
    },
    {
      "epoch": 0.516403501595353,
      "grad_norm": 0.11935051150569086,
      "learning_rate": 5.585102564361499e-06,
      "loss": 0.0065,
      "step": 31560
    },
    {
      "epoch": 0.5165671275464289,
      "grad_norm": 0.17922696191171508,
      "learning_rate": 5.582266252565458e-06,
      "loss": 0.0024,
      "step": 31570
    },
    {
      "epoch": 0.5167307534975047,
      "grad_norm": 0.08469500777436423,
      "learning_rate": 5.579429750815351e-06,
      "loss": 0.0057,
      "step": 31580
    },
    {
      "epoch": 0.5168943794485805,
      "grad_norm": 0.3688162466844152,
      "learning_rate": 5.576593060036537e-06,
      "loss": 0.0044,
      "step": 31590
    },
    {
      "epoch": 0.5170580053996564,
      "grad_norm": 0.1670144960303306,
      "learning_rate": 5.57375618115443e-06,
      "loss": 0.0043,
      "step": 31600
    },
    {
      "epoch": 0.5172216313507322,
      "grad_norm": 0.22742852578415487,
      "learning_rate": 5.570919115094518e-06,
      "loss": 0.0048,
      "step": 31610
    },
    {
      "epoch": 0.5173852573018081,
      "grad_norm": 0.40427429575846835,
      "learning_rate": 5.568081862782342e-06,
      "loss": 0.0033,
      "step": 31620
    },
    {
      "epoch": 0.517548883252884,
      "grad_norm": 0.06370511116338798,
      "learning_rate": 5.565244425143504e-06,
      "loss": 0.0039,
      "step": 31630
    },
    {
      "epoch": 0.5177125092039597,
      "grad_norm": 0.16427213215346223,
      "learning_rate": 5.562406803103668e-06,
      "loss": 0.0031,
      "step": 31640
    },
    {
      "epoch": 0.5178761351550356,
      "grad_norm": 0.09402786959017584,
      "learning_rate": 5.559568997588558e-06,
      "loss": 0.0028,
      "step": 31650
    },
    {
      "epoch": 0.5180397611061114,
      "grad_norm": 0.36332542976726234,
      "learning_rate": 5.556731009523959e-06,
      "loss": 0.0033,
      "step": 31660
    },
    {
      "epoch": 0.5182033870571873,
      "grad_norm": 0.19619561151328846,
      "learning_rate": 5.553892839835714e-06,
      "loss": 0.0031,
      "step": 31670
    },
    {
      "epoch": 0.5183670130082632,
      "grad_norm": 0.09720159035212074,
      "learning_rate": 5.551054489449725e-06,
      "loss": 0.0037,
      "step": 31680
    },
    {
      "epoch": 0.5185306389593389,
      "grad_norm": 0.3377310656945975,
      "learning_rate": 5.548215959291954e-06,
      "loss": 0.0027,
      "step": 31690
    },
    {
      "epoch": 0.5186942649104148,
      "grad_norm": 0.11641622347522938,
      "learning_rate": 5.545377250288422e-06,
      "loss": 0.0023,
      "step": 31700
    },
    {
      "epoch": 0.5188578908614906,
      "grad_norm": 0.06900096406129809,
      "learning_rate": 5.542538363365208e-06,
      "loss": 0.0037,
      "step": 31710
    },
    {
      "epoch": 0.5190215168125665,
      "grad_norm": 0.13304384519839824,
      "learning_rate": 5.539699299448444e-06,
      "loss": 0.0041,
      "step": 31720
    },
    {
      "epoch": 0.5191851427636424,
      "grad_norm": 0.06589612568895965,
      "learning_rate": 5.53686005946433e-06,
      "loss": 0.0024,
      "step": 31730
    },
    {
      "epoch": 0.5193487687147181,
      "grad_norm": 0.20920091198196153,
      "learning_rate": 5.534020644339118e-06,
      "loss": 0.0025,
      "step": 31740
    },
    {
      "epoch": 0.519512394665794,
      "grad_norm": 0.27992363186227776,
      "learning_rate": 5.5311810549991115e-06,
      "loss": 0.0038,
      "step": 31750
    },
    {
      "epoch": 0.5196760206168698,
      "grad_norm": 0.14871714870407046,
      "learning_rate": 5.5283412923706815e-06,
      "loss": 0.0029,
      "step": 31760
    },
    {
      "epoch": 0.5198396465679457,
      "grad_norm": 0.16493032172596309,
      "learning_rate": 5.525501357380247e-06,
      "loss": 0.0023,
      "step": 31770
    },
    {
      "epoch": 0.5200032725190216,
      "grad_norm": 0.16648262941400113,
      "learning_rate": 5.522661250954289e-06,
      "loss": 0.004,
      "step": 31780
    },
    {
      "epoch": 0.5201668984700973,
      "grad_norm": 0.3121100438063519,
      "learning_rate": 5.519820974019341e-06,
      "loss": 0.0055,
      "step": 31790
    },
    {
      "epoch": 0.5203305244211732,
      "grad_norm": 0.19435441150776234,
      "learning_rate": 5.516980527501994e-06,
      "loss": 0.0031,
      "step": 31800
    },
    {
      "epoch": 0.520494150372249,
      "grad_norm": 0.11350818341588174,
      "learning_rate": 5.51413991232889e-06,
      "loss": 0.0024,
      "step": 31810
    },
    {
      "epoch": 0.5206577763233249,
      "grad_norm": 0.1418992813321056,
      "learning_rate": 5.511299129426735e-06,
      "loss": 0.004,
      "step": 31820
    },
    {
      "epoch": 0.5208214022744008,
      "grad_norm": 0.09775767424980518,
      "learning_rate": 5.508458179722279e-06,
      "loss": 0.0033,
      "step": 31830
    },
    {
      "epoch": 0.5209850282254765,
      "grad_norm": 0.124817127282726,
      "learning_rate": 5.505617064142336e-06,
      "loss": 0.0027,
      "step": 31840
    },
    {
      "epoch": 0.5211486541765524,
      "grad_norm": 0.4482925886691343,
      "learning_rate": 5.502775783613765e-06,
      "loss": 0.0032,
      "step": 31850
    },
    {
      "epoch": 0.5213122801276282,
      "grad_norm": 0.18621437567168356,
      "learning_rate": 5.4999343390634875e-06,
      "loss": 0.0022,
      "step": 31860
    },
    {
      "epoch": 0.5214759060787041,
      "grad_norm": 0.23038293817958985,
      "learning_rate": 5.497092731418473e-06,
      "loss": 0.0044,
      "step": 31870
    },
    {
      "epoch": 0.52163953202978,
      "grad_norm": 0.23885597895309504,
      "learning_rate": 5.494250961605745e-06,
      "loss": 0.004,
      "step": 31880
    },
    {
      "epoch": 0.5218031579808557,
      "grad_norm": 0.13655926035999974,
      "learning_rate": 5.491409030552383e-06,
      "loss": 0.0028,
      "step": 31890
    },
    {
      "epoch": 0.5219667839319316,
      "grad_norm": 0.04216317177801863,
      "learning_rate": 5.488566939185514e-06,
      "loss": 0.0032,
      "step": 31900
    },
    {
      "epoch": 0.5221304098830074,
      "grad_norm": 0.2383644432125547,
      "learning_rate": 5.48572468843232e-06,
      "loss": 0.0036,
      "step": 31910
    },
    {
      "epoch": 0.5222940358340833,
      "grad_norm": 0.14568711990382036,
      "learning_rate": 5.482882279220039e-06,
      "loss": 0.0023,
      "step": 31920
    },
    {
      "epoch": 0.5224576617851592,
      "grad_norm": 0.07362329808097236,
      "learning_rate": 5.480039712475953e-06,
      "loss": 0.0024,
      "step": 31930
    },
    {
      "epoch": 0.522621287736235,
      "grad_norm": 0.1671837891244254,
      "learning_rate": 5.4771969891274e-06,
      "loss": 0.0022,
      "step": 31940
    },
    {
      "epoch": 0.5227849136873108,
      "grad_norm": 0.15838017118807296,
      "learning_rate": 5.474354110101769e-06,
      "loss": 0.0035,
      "step": 31950
    },
    {
      "epoch": 0.5229485396383866,
      "grad_norm": 0.38880547054207254,
      "learning_rate": 5.471511076326498e-06,
      "loss": 0.0032,
      "step": 31960
    },
    {
      "epoch": 0.5231121655894625,
      "grad_norm": 0.030920400618432743,
      "learning_rate": 5.468667888729077e-06,
      "loss": 0.0019,
      "step": 31970
    },
    {
      "epoch": 0.5232757915405384,
      "grad_norm": 0.31725380635774564,
      "learning_rate": 5.4658245482370455e-06,
      "loss": 0.003,
      "step": 31980
    },
    {
      "epoch": 0.5234394174916142,
      "grad_norm": 0.10378538611670388,
      "learning_rate": 5.462981055777992e-06,
      "loss": 0.0024,
      "step": 31990
    },
    {
      "epoch": 0.52360304344269,
      "grad_norm": 0.13202900436006346,
      "learning_rate": 5.460137412279558e-06,
      "loss": 0.0015,
      "step": 32000
    },
    {
      "epoch": 0.5237666693937658,
      "grad_norm": 0.18499624840034487,
      "learning_rate": 5.457293618669431e-06,
      "loss": 0.0029,
      "step": 32010
    },
    {
      "epoch": 0.5239302953448417,
      "grad_norm": 0.14631558094745362,
      "learning_rate": 5.454449675875346e-06,
      "loss": 0.0031,
      "step": 32020
    },
    {
      "epoch": 0.5240939212959176,
      "grad_norm": 0.23860297779401496,
      "learning_rate": 5.4516055848250935e-06,
      "loss": 0.0042,
      "step": 32030
    },
    {
      "epoch": 0.5242575472469934,
      "grad_norm": 0.34970092659224294,
      "learning_rate": 5.448761346446505e-06,
      "loss": 0.004,
      "step": 32040
    },
    {
      "epoch": 0.5244211731980692,
      "grad_norm": 0.16857497169223268,
      "learning_rate": 5.445916961667463e-06,
      "loss": 0.0027,
      "step": 32050
    },
    {
      "epoch": 0.524584799149145,
      "grad_norm": 0.24628985800494357,
      "learning_rate": 5.4430724314158975e-06,
      "loss": 0.0035,
      "step": 32060
    },
    {
      "epoch": 0.5247484251002209,
      "grad_norm": 0.3172309700172557,
      "learning_rate": 5.440227756619788e-06,
      "loss": 0.0045,
      "step": 32070
    },
    {
      "epoch": 0.5249120510512968,
      "grad_norm": 0.16941070640265618,
      "learning_rate": 5.437382938207157e-06,
      "loss": 0.0025,
      "step": 32080
    },
    {
      "epoch": 0.5250756770023726,
      "grad_norm": 0.3504327893132381,
      "learning_rate": 5.434537977106078e-06,
      "loss": 0.0021,
      "step": 32090
    },
    {
      "epoch": 0.5252393029534485,
      "grad_norm": 0.09708043060019855,
      "learning_rate": 5.43169287424467e-06,
      "loss": 0.0033,
      "step": 32100
    },
    {
      "epoch": 0.5254029289045242,
      "grad_norm": 0.11955537377432242,
      "learning_rate": 5.428847630551093e-06,
      "loss": 0.0021,
      "step": 32110
    },
    {
      "epoch": 0.5255665548556001,
      "grad_norm": 0.16419746435046614,
      "learning_rate": 5.4260022469535635e-06,
      "loss": 0.0026,
      "step": 32120
    },
    {
      "epoch": 0.5257301808066759,
      "grad_norm": 0.19862634391131348,
      "learning_rate": 5.423156724380334e-06,
      "loss": 0.0053,
      "step": 32130
    },
    {
      "epoch": 0.5258938067577518,
      "grad_norm": 0.24748169031056463,
      "learning_rate": 5.420311063759704e-06,
      "loss": 0.0033,
      "step": 32140
    },
    {
      "epoch": 0.5260574327088277,
      "grad_norm": 0.03293657902841861,
      "learning_rate": 5.417465266020027e-06,
      "loss": 0.0015,
      "step": 32150
    },
    {
      "epoch": 0.5262210586599034,
      "grad_norm": 0.22575355482585527,
      "learning_rate": 5.4146193320896864e-06,
      "loss": 0.0031,
      "step": 32160
    },
    {
      "epoch": 0.5263846846109793,
      "grad_norm": 0.33363219636371877,
      "learning_rate": 5.411773262897122e-06,
      "loss": 0.0035,
      "step": 32170
    },
    {
      "epoch": 0.5265483105620551,
      "grad_norm": 0.13360375267772587,
      "learning_rate": 5.408927059370813e-06,
      "loss": 0.0046,
      "step": 32180
    },
    {
      "epoch": 0.526711936513131,
      "grad_norm": 0.0537121976658464,
      "learning_rate": 5.406080722439283e-06,
      "loss": 0.0032,
      "step": 32190
    },
    {
      "epoch": 0.5268755624642069,
      "grad_norm": 0.24586689840729387,
      "learning_rate": 5.403234253031098e-06,
      "loss": 0.0046,
      "step": 32200
    },
    {
      "epoch": 0.5270391884152826,
      "grad_norm": 0.14156560396243956,
      "learning_rate": 5.40038765207487e-06,
      "loss": 0.0024,
      "step": 32210
    },
    {
      "epoch": 0.5272028143663585,
      "grad_norm": 0.24342417836894994,
      "learning_rate": 5.39754092049925e-06,
      "loss": 0.003,
      "step": 32220
    },
    {
      "epoch": 0.5273664403174343,
      "grad_norm": 0.31095122634654004,
      "learning_rate": 5.394694059232936e-06,
      "loss": 0.0036,
      "step": 32230
    },
    {
      "epoch": 0.5275300662685102,
      "grad_norm": 0.09999622296454513,
      "learning_rate": 5.3918470692046645e-06,
      "loss": 0.0017,
      "step": 32240
    },
    {
      "epoch": 0.5276936922195861,
      "grad_norm": 0.20965542748728558,
      "learning_rate": 5.388999951343217e-06,
      "loss": 0.0026,
      "step": 32250
    },
    {
      "epoch": 0.5278573181706618,
      "grad_norm": 0.27259757963727593,
      "learning_rate": 5.386152706577414e-06,
      "loss": 0.0032,
      "step": 32260
    },
    {
      "epoch": 0.5280209441217377,
      "grad_norm": 0.19015882541433954,
      "learning_rate": 5.383305335836121e-06,
      "loss": 0.0043,
      "step": 32270
    },
    {
      "epoch": 0.5281845700728135,
      "grad_norm": 0.047006709635032125,
      "learning_rate": 5.38045784004824e-06,
      "loss": 0.004,
      "step": 32280
    },
    {
      "epoch": 0.5283481960238894,
      "grad_norm": 0.1301786403627152,
      "learning_rate": 5.377610220142716e-06,
      "loss": 0.0028,
      "step": 32290
    },
    {
      "epoch": 0.5285118219749653,
      "grad_norm": 0.31751720324993454,
      "learning_rate": 5.374762477048537e-06,
      "loss": 0.004,
      "step": 32300
    },
    {
      "epoch": 0.528675447926041,
      "grad_norm": 0.3061096890655228,
      "learning_rate": 5.371914611694726e-06,
      "loss": 0.0042,
      "step": 32310
    },
    {
      "epoch": 0.5288390738771169,
      "grad_norm": 0.08582772403980092,
      "learning_rate": 5.369066625010352e-06,
      "loss": 0.0029,
      "step": 32320
    },
    {
      "epoch": 0.5290026998281927,
      "grad_norm": 0.15411337365355393,
      "learning_rate": 5.3662185179245175e-06,
      "loss": 0.0044,
      "step": 32330
    },
    {
      "epoch": 0.5291663257792686,
      "grad_norm": 0.11775395357718962,
      "learning_rate": 5.363370291366367e-06,
      "loss": 0.0046,
      "step": 32340
    },
    {
      "epoch": 0.5293299517303445,
      "grad_norm": 0.19408392170722583,
      "learning_rate": 5.360521946265087e-06,
      "loss": 0.0031,
      "step": 32350
    },
    {
      "epoch": 0.5294935776814202,
      "grad_norm": 0.022966091151615253,
      "learning_rate": 5.357673483549896e-06,
      "loss": 0.0017,
      "step": 32360
    },
    {
      "epoch": 0.5296572036324961,
      "grad_norm": 0.21200362027542247,
      "learning_rate": 5.3548249041500575e-06,
      "loss": 0.0032,
      "step": 32370
    },
    {
      "epoch": 0.5298208295835719,
      "grad_norm": 0.14168072438362506,
      "learning_rate": 5.351976208994869e-06,
      "loss": 0.0038,
      "step": 32380
    },
    {
      "epoch": 0.5299844555346478,
      "grad_norm": 0.09675837115436164,
      "learning_rate": 5.349127399013666e-06,
      "loss": 0.0043,
      "step": 32390
    },
    {
      "epoch": 0.5301480814857237,
      "grad_norm": 0.1832838640122271,
      "learning_rate": 5.346278475135824e-06,
      "loss": 0.0029,
      "step": 32400
    },
    {
      "epoch": 0.5303117074367995,
      "grad_norm": 0.21587723371828704,
      "learning_rate": 5.343429438290754e-06,
      "loss": 0.0024,
      "step": 32410
    },
    {
      "epoch": 0.5304753333878753,
      "grad_norm": 0.21866364162163882,
      "learning_rate": 5.3405802894079016e-06,
      "loss": 0.0041,
      "step": 32420
    },
    {
      "epoch": 0.5306389593389511,
      "grad_norm": 0.2158061676519111,
      "learning_rate": 5.337731029416754e-06,
      "loss": 0.002,
      "step": 32430
    },
    {
      "epoch": 0.530802585290027,
      "grad_norm": 0.06003821699663223,
      "learning_rate": 5.334881659246831e-06,
      "loss": 0.0038,
      "step": 32440
    },
    {
      "epoch": 0.5309662112411029,
      "grad_norm": 0.17315046602738104,
      "learning_rate": 5.3320321798276874e-06,
      "loss": 0.0039,
      "step": 32450
    },
    {
      "epoch": 0.5311298371921787,
      "grad_norm": 0.11690745806499225,
      "learning_rate": 5.329182592088918e-06,
      "loss": 0.0041,
      "step": 32460
    },
    {
      "epoch": 0.5312934631432545,
      "grad_norm": 0.38057251490459354,
      "learning_rate": 5.32633289696015e-06,
      "loss": 0.0039,
      "step": 32470
    },
    {
      "epoch": 0.5314570890943303,
      "grad_norm": 0.13089402350070317,
      "learning_rate": 5.323483095371045e-06,
      "loss": 0.0038,
      "step": 32480
    },
    {
      "epoch": 0.5316207150454062,
      "grad_norm": 0.11504406830203273,
      "learning_rate": 5.320633188251302e-06,
      "loss": 0.0025,
      "step": 32490
    },
    {
      "epoch": 0.5317843409964821,
      "grad_norm": 0.29247749249446015,
      "learning_rate": 5.3177831765306485e-06,
      "loss": 0.0042,
      "step": 32500
    },
    {
      "epoch": 0.5319479669475579,
      "grad_norm": 0.1486389625634579,
      "learning_rate": 5.314933061138855e-06,
      "loss": 0.003,
      "step": 32510
    },
    {
      "epoch": 0.5321115928986337,
      "grad_norm": 0.11432697436806108,
      "learning_rate": 5.312082843005723e-06,
      "loss": 0.004,
      "step": 32520
    },
    {
      "epoch": 0.5322752188497095,
      "grad_norm": 0.15569108730077688,
      "learning_rate": 5.309232523061079e-06,
      "loss": 0.0044,
      "step": 32530
    },
    {
      "epoch": 0.5324388448007854,
      "grad_norm": 0.5086813064898466,
      "learning_rate": 5.306382102234793e-06,
      "loss": 0.0041,
      "step": 32540
    },
    {
      "epoch": 0.5326024707518613,
      "grad_norm": 0.04862547931235565,
      "learning_rate": 5.303531581456767e-06,
      "loss": 0.0038,
      "step": 32550
    },
    {
      "epoch": 0.5327660967029371,
      "grad_norm": 0.2553986481414818,
      "learning_rate": 5.300680961656928e-06,
      "loss": 0.0036,
      "step": 32560
    },
    {
      "epoch": 0.532929722654013,
      "grad_norm": 0.05273455354864918,
      "learning_rate": 5.297830243765243e-06,
      "loss": 0.0021,
      "step": 32570
    },
    {
      "epoch": 0.5330933486050887,
      "grad_norm": 0.284042405125996,
      "learning_rate": 5.294979428711708e-06,
      "loss": 0.0027,
      "step": 32580
    },
    {
      "epoch": 0.5332569745561646,
      "grad_norm": 0.06671907906860931,
      "learning_rate": 5.292128517426351e-06,
      "loss": 0.0024,
      "step": 32590
    },
    {
      "epoch": 0.5334206005072405,
      "grad_norm": 0.1413664959204848,
      "learning_rate": 5.289277510839232e-06,
      "loss": 0.0021,
      "step": 32600
    },
    {
      "epoch": 0.5335842264583163,
      "grad_norm": 0.194005344841848,
      "learning_rate": 5.286426409880442e-06,
      "loss": 0.0031,
      "step": 32610
    },
    {
      "epoch": 0.5337478524093922,
      "grad_norm": 0.10959839605150984,
      "learning_rate": 5.283575215480099e-06,
      "loss": 0.0035,
      "step": 32620
    },
    {
      "epoch": 0.5339114783604679,
      "grad_norm": 0.13946966142624195,
      "learning_rate": 5.280723928568359e-06,
      "loss": 0.0027,
      "step": 32630
    },
    {
      "epoch": 0.5340751043115438,
      "grad_norm": 0.042340973387254746,
      "learning_rate": 5.2778725500754e-06,
      "loss": 0.002,
      "step": 32640
    },
    {
      "epoch": 0.5342387302626197,
      "grad_norm": 0.09736978612490262,
      "learning_rate": 5.275021080931436e-06,
      "loss": 0.0023,
      "step": 32650
    },
    {
      "epoch": 0.5344023562136955,
      "grad_norm": 0.17673280573270805,
      "learning_rate": 5.272169522066709e-06,
      "loss": 0.0022,
      "step": 32660
    },
    {
      "epoch": 0.5345659821647714,
      "grad_norm": 0.29721778957767603,
      "learning_rate": 5.2693178744114885e-06,
      "loss": 0.0019,
      "step": 32670
    },
    {
      "epoch": 0.5347296081158471,
      "grad_norm": 0.20032287859769843,
      "learning_rate": 5.266466138896073e-06,
      "loss": 0.0021,
      "step": 32680
    },
    {
      "epoch": 0.534893234066923,
      "grad_norm": 0.19603243451910135,
      "learning_rate": 5.263614316450793e-06,
      "loss": 0.0021,
      "step": 32690
    },
    {
      "epoch": 0.5350568600179989,
      "grad_norm": 0.26549289908160717,
      "learning_rate": 5.260762408006005e-06,
      "loss": 0.0024,
      "step": 32700
    },
    {
      "epoch": 0.5352204859690747,
      "grad_norm": 0.12324822462546113,
      "learning_rate": 5.257910414492091e-06,
      "loss": 0.0018,
      "step": 32710
    },
    {
      "epoch": 0.5353841119201506,
      "grad_norm": 0.34276891316117636,
      "learning_rate": 5.255058336839467e-06,
      "loss": 0.0025,
      "step": 32720
    },
    {
      "epoch": 0.5355477378712263,
      "grad_norm": 0.2577683840495486,
      "learning_rate": 5.25220617597857e-06,
      "loss": 0.0035,
      "step": 32730
    },
    {
      "epoch": 0.5357113638223022,
      "grad_norm": 0.2875364766500727,
      "learning_rate": 5.249353932839869e-06,
      "loss": 0.0027,
      "step": 32740
    },
    {
      "epoch": 0.5358749897733781,
      "grad_norm": 0.12128678196184674,
      "learning_rate": 5.246501608353857e-06,
      "loss": 0.0028,
      "step": 32750
    },
    {
      "epoch": 0.5360386157244539,
      "grad_norm": 0.13333318369324978,
      "learning_rate": 5.243649203451055e-06,
      "loss": 0.0036,
      "step": 32760
    },
    {
      "epoch": 0.5362022416755298,
      "grad_norm": 0.10917143885680675,
      "learning_rate": 5.240796719062008e-06,
      "loss": 0.0032,
      "step": 32770
    },
    {
      "epoch": 0.5363658676266055,
      "grad_norm": 0.12232298157681375,
      "learning_rate": 5.237944156117292e-06,
      "loss": 0.007,
      "step": 32780
    },
    {
      "epoch": 0.5365294935776814,
      "grad_norm": 0.056364958576616234,
      "learning_rate": 5.235091515547502e-06,
      "loss": 0.0018,
      "step": 32790
    },
    {
      "epoch": 0.5366931195287573,
      "grad_norm": 0.0502022622744375,
      "learning_rate": 5.232238798283261e-06,
      "loss": 0.0026,
      "step": 32800
    },
    {
      "epoch": 0.5368567454798331,
      "grad_norm": 0.15367439161022073,
      "learning_rate": 5.2293860052552206e-06,
      "loss": 0.005,
      "step": 32810
    },
    {
      "epoch": 0.537020371430909,
      "grad_norm": 0.0701227352220075,
      "learning_rate": 5.226533137394053e-06,
      "loss": 0.0028,
      "step": 32820
    },
    {
      "epoch": 0.5371839973819847,
      "grad_norm": 0.2674614758520082,
      "learning_rate": 5.223680195630455e-06,
      "loss": 0.0028,
      "step": 32830
    },
    {
      "epoch": 0.5373476233330606,
      "grad_norm": 0.07660975046700641,
      "learning_rate": 5.220827180895149e-06,
      "loss": 0.0033,
      "step": 32840
    },
    {
      "epoch": 0.5375112492841365,
      "grad_norm": 0.28556638746886626,
      "learning_rate": 5.21797409411888e-06,
      "loss": 0.0021,
      "step": 32850
    },
    {
      "epoch": 0.5376748752352123,
      "grad_norm": 0.10768165428977265,
      "learning_rate": 5.215120936232418e-06,
      "loss": 0.0023,
      "step": 32860
    },
    {
      "epoch": 0.5378385011862882,
      "grad_norm": 0.24504134439443845,
      "learning_rate": 5.212267708166555e-06,
      "loss": 0.004,
      "step": 32870
    },
    {
      "epoch": 0.538002127137364,
      "grad_norm": 0.18686983849963693,
      "learning_rate": 5.209414410852106e-06,
      "loss": 0.0031,
      "step": 32880
    },
    {
      "epoch": 0.5381657530884398,
      "grad_norm": 0.06830813591743791,
      "learning_rate": 5.20656104521991e-06,
      "loss": 0.0024,
      "step": 32890
    },
    {
      "epoch": 0.5383293790395157,
      "grad_norm": 0.187470766627642,
      "learning_rate": 5.203707612200824e-06,
      "loss": 0.0041,
      "step": 32900
    },
    {
      "epoch": 0.5384930049905915,
      "grad_norm": 0.17047553536809082,
      "learning_rate": 5.200854112725733e-06,
      "loss": 0.002,
      "step": 32910
    },
    {
      "epoch": 0.5386566309416674,
      "grad_norm": 0.24568116847062507,
      "learning_rate": 5.19800054772554e-06,
      "loss": 0.0025,
      "step": 32920
    },
    {
      "epoch": 0.5388202568927432,
      "grad_norm": 0.10954059333396925,
      "learning_rate": 5.195146918131169e-06,
      "loss": 0.0042,
      "step": 32930
    },
    {
      "epoch": 0.538983882843819,
      "grad_norm": 0.2941907908846616,
      "learning_rate": 5.192293224873568e-06,
      "loss": 0.0043,
      "step": 32940
    },
    {
      "epoch": 0.5391475087948949,
      "grad_norm": 0.1971298247797568,
      "learning_rate": 5.189439468883704e-06,
      "loss": 0.0025,
      "step": 32950
    },
    {
      "epoch": 0.5393111347459707,
      "grad_norm": 0.04797655456154884,
      "learning_rate": 5.1865856510925605e-06,
      "loss": 0.0019,
      "step": 32960
    },
    {
      "epoch": 0.5394747606970466,
      "grad_norm": 0.1638415827656726,
      "learning_rate": 5.183731772431151e-06,
      "loss": 0.0035,
      "step": 32970
    },
    {
      "epoch": 0.5396383866481224,
      "grad_norm": 0.22496829683090672,
      "learning_rate": 5.180877833830499e-06,
      "loss": 0.0025,
      "step": 32980
    },
    {
      "epoch": 0.5398020125991982,
      "grad_norm": 0.17983115314668138,
      "learning_rate": 5.178023836221651e-06,
      "loss": 0.0026,
      "step": 32990
    },
    {
      "epoch": 0.539965638550274,
      "grad_norm": 0.26245625000853895,
      "learning_rate": 5.175169780535677e-06,
      "loss": 0.0028,
      "step": 33000
    },
    {
      "epoch": 0.5401292645013499,
      "grad_norm": 0.6596782126225295,
      "learning_rate": 5.172315667703658e-06,
      "loss": 0.0048,
      "step": 33010
    },
    {
      "epoch": 0.5402928904524258,
      "grad_norm": 0.21311748822093782,
      "learning_rate": 5.1694614986567005e-06,
      "loss": 0.0025,
      "step": 33020
    },
    {
      "epoch": 0.5404565164035016,
      "grad_norm": 0.11242835083113199,
      "learning_rate": 5.166607274325928e-06,
      "loss": 0.0041,
      "step": 33030
    },
    {
      "epoch": 0.5406201423545774,
      "grad_norm": 0.10927477790466357,
      "learning_rate": 5.163752995642476e-06,
      "loss": 0.0028,
      "step": 33040
    },
    {
      "epoch": 0.5407837683056532,
      "grad_norm": 0.27271916460538836,
      "learning_rate": 5.160898663537505e-06,
      "loss": 0.0022,
      "step": 33050
    },
    {
      "epoch": 0.5409473942567291,
      "grad_norm": 0.15440376123770488,
      "learning_rate": 5.158044278942192e-06,
      "loss": 0.0042,
      "step": 33060
    },
    {
      "epoch": 0.541111020207805,
      "grad_norm": 0.04384963533770578,
      "learning_rate": 5.155189842787727e-06,
      "loss": 0.0024,
      "step": 33070
    },
    {
      "epoch": 0.5412746461588808,
      "grad_norm": 0.09458686641321318,
      "learning_rate": 5.15233535600532e-06,
      "loss": 0.0035,
      "step": 33080
    },
    {
      "epoch": 0.5414382721099567,
      "grad_norm": 0.35165263126856267,
      "learning_rate": 5.149480819526199e-06,
      "loss": 0.0022,
      "step": 33090
    },
    {
      "epoch": 0.5416018980610324,
      "grad_norm": 0.1772141395168747,
      "learning_rate": 5.146626234281602e-06,
      "loss": 0.0031,
      "step": 33100
    },
    {
      "epoch": 0.5417655240121083,
      "grad_norm": 0.4200715322836401,
      "learning_rate": 5.143771601202789e-06,
      "loss": 0.003,
      "step": 33110
    },
    {
      "epoch": 0.5419291499631842,
      "grad_norm": 0.109457101425537,
      "learning_rate": 5.140916921221037e-06,
      "loss": 0.0037,
      "step": 33120
    },
    {
      "epoch": 0.54209277591426,
      "grad_norm": 0.1791063740353525,
      "learning_rate": 5.138062195267628e-06,
      "loss": 0.004,
      "step": 33130
    },
    {
      "epoch": 0.5422564018653359,
      "grad_norm": 0.12639814295352572,
      "learning_rate": 5.1352074242738705e-06,
      "loss": 0.0033,
      "step": 33140
    },
    {
      "epoch": 0.5424200278164116,
      "grad_norm": 0.5381983057446634,
      "learning_rate": 5.1323526091710826e-06,
      "loss": 0.0029,
      "step": 33150
    },
    {
      "epoch": 0.5425836537674875,
      "grad_norm": 0.20999813192750963,
      "learning_rate": 5.129497750890595e-06,
      "loss": 0.0042,
      "step": 33160
    },
    {
      "epoch": 0.5427472797185634,
      "grad_norm": 0.1734021565056439,
      "learning_rate": 5.126642850363758e-06,
      "loss": 0.003,
      "step": 33170
    },
    {
      "epoch": 0.5429109056696392,
      "grad_norm": 0.09173405918512315,
      "learning_rate": 5.1237879085219296e-06,
      "loss": 0.0026,
      "step": 33180
    },
    {
      "epoch": 0.5430745316207151,
      "grad_norm": 0.1798198313788904,
      "learning_rate": 5.120932926296486e-06,
      "loss": 0.0033,
      "step": 33190
    },
    {
      "epoch": 0.5432381575717908,
      "grad_norm": 0.28003916371376075,
      "learning_rate": 5.118077904618813e-06,
      "loss": 0.0027,
      "step": 33200
    },
    {
      "epoch": 0.5434017835228667,
      "grad_norm": 0.09071919941396615,
      "learning_rate": 5.115222844420312e-06,
      "loss": 0.0031,
      "step": 33210
    },
    {
      "epoch": 0.5435654094739426,
      "grad_norm": 0.43652298650022087,
      "learning_rate": 5.1123677466323966e-06,
      "loss": 0.0039,
      "step": 33220
    },
    {
      "epoch": 0.5437290354250184,
      "grad_norm": 0.1137658708628742,
      "learning_rate": 5.10951261218649e-06,
      "loss": 0.0014,
      "step": 33230
    },
    {
      "epoch": 0.5438926613760943,
      "grad_norm": 0.23871677391112014,
      "learning_rate": 5.106657442014031e-06,
      "loss": 0.0085,
      "step": 33240
    },
    {
      "epoch": 0.54405628732717,
      "grad_norm": 0.10758978587401452,
      "learning_rate": 5.103802237046467e-06,
      "loss": 0.0032,
      "step": 33250
    },
    {
      "epoch": 0.5442199132782459,
      "grad_norm": 0.24119290786921255,
      "learning_rate": 5.100946998215259e-06,
      "loss": 0.0038,
      "step": 33260
    },
    {
      "epoch": 0.5443835392293218,
      "grad_norm": 0.27739026732622596,
      "learning_rate": 5.09809172645188e-06,
      "loss": 0.0032,
      "step": 33270
    },
    {
      "epoch": 0.5445471651803976,
      "grad_norm": 0.1463538400044755,
      "learning_rate": 5.095236422687807e-06,
      "loss": 0.0021,
      "step": 33280
    },
    {
      "epoch": 0.5447107911314735,
      "grad_norm": 0.13556738603928378,
      "learning_rate": 5.092381087854538e-06,
      "loss": 0.0032,
      "step": 33290
    },
    {
      "epoch": 0.5448744170825492,
      "grad_norm": 0.28503955281491294,
      "learning_rate": 5.0895257228835715e-06,
      "loss": 0.0028,
      "step": 33300
    },
    {
      "epoch": 0.5450380430336251,
      "grad_norm": 0.01337841405799922,
      "learning_rate": 5.086670328706421e-06,
      "loss": 0.0049,
      "step": 33310
    },
    {
      "epoch": 0.545201668984701,
      "grad_norm": 0.12692851218008064,
      "learning_rate": 5.083814906254608e-06,
      "loss": 0.0014,
      "step": 33320
    },
    {
      "epoch": 0.5453652949357768,
      "grad_norm": 0.4945647827718782,
      "learning_rate": 5.080959456459666e-06,
      "loss": 0.0034,
      "step": 33330
    },
    {
      "epoch": 0.5455289208868527,
      "grad_norm": 0.05670437061252929,
      "learning_rate": 5.078103980253132e-06,
      "loss": 0.0052,
      "step": 33340
    },
    {
      "epoch": 0.5456925468379284,
      "grad_norm": 0.1615537427401772,
      "learning_rate": 5.075248478566555e-06,
      "loss": 0.0023,
      "step": 33350
    },
    {
      "epoch": 0.5458561727890043,
      "grad_norm": 0.11367964380546852,
      "learning_rate": 5.072392952331493e-06,
      "loss": 0.0014,
      "step": 33360
    },
    {
      "epoch": 0.5460197987400802,
      "grad_norm": 0.2324827051522824,
      "learning_rate": 5.069537402479511e-06,
      "loss": 0.0035,
      "step": 33370
    },
    {
      "epoch": 0.546183424691156,
      "grad_norm": 0.20424525843626234,
      "learning_rate": 5.06668182994218e-06,
      "loss": 0.0029,
      "step": 33380
    },
    {
      "epoch": 0.5463470506422319,
      "grad_norm": 0.14397939122142903,
      "learning_rate": 5.063826235651082e-06,
      "loss": 0.0037,
      "step": 33390
    },
    {
      "epoch": 0.5465106765933077,
      "grad_norm": 0.09089785812272315,
      "learning_rate": 5.060970620537805e-06,
      "loss": 0.0032,
      "step": 33400
    },
    {
      "epoch": 0.5466743025443835,
      "grad_norm": 0.09960089827377018,
      "learning_rate": 5.058114985533939e-06,
      "loss": 0.0041,
      "step": 33410
    },
    {
      "epoch": 0.5468379284954594,
      "grad_norm": 0.10802924139868546,
      "learning_rate": 5.055259331571087e-06,
      "loss": 0.0029,
      "step": 33420
    },
    {
      "epoch": 0.5470015544465352,
      "grad_norm": 0.229487867972326,
      "learning_rate": 5.052403659580857e-06,
      "loss": 0.0024,
      "step": 33430
    },
    {
      "epoch": 0.5471651803976111,
      "grad_norm": 0.3389646502670947,
      "learning_rate": 5.049547970494858e-06,
      "loss": 0.003,
      "step": 33440
    },
    {
      "epoch": 0.5473288063486869,
      "grad_norm": 0.5227087776264103,
      "learning_rate": 5.04669226524471e-06,
      "loss": 0.0031,
      "step": 33450
    },
    {
      "epoch": 0.5474924322997627,
      "grad_norm": 0.2223338992273839,
      "learning_rate": 5.043836544762039e-06,
      "loss": 0.0023,
      "step": 33460
    },
    {
      "epoch": 0.5476560582508386,
      "grad_norm": 0.1574476280829336,
      "learning_rate": 5.040980809978466e-06,
      "loss": 0.002,
      "step": 33470
    },
    {
      "epoch": 0.5478196842019144,
      "grad_norm": 0.0616870356740467,
      "learning_rate": 5.038125061825633e-06,
      "loss": 0.0024,
      "step": 33480
    },
    {
      "epoch": 0.5479833101529903,
      "grad_norm": 0.23573832514411017,
      "learning_rate": 5.03526930123517e-06,
      "loss": 0.003,
      "step": 33490
    },
    {
      "epoch": 0.5481469361040661,
      "grad_norm": 0.11478678725584612,
      "learning_rate": 5.032413529138721e-06,
      "loss": 0.0031,
      "step": 33500
    },
    {
      "epoch": 0.548310562055142,
      "grad_norm": 0.1893585896738234,
      "learning_rate": 5.029557746467933e-06,
      "loss": 0.0029,
      "step": 33510
    },
    {
      "epoch": 0.5484741880062178,
      "grad_norm": 0.2559089757330635,
      "learning_rate": 5.026701954154451e-06,
      "loss": 0.0026,
      "step": 33520
    },
    {
      "epoch": 0.5486378139572936,
      "grad_norm": 0.13134963795175011,
      "learning_rate": 5.02384615312993e-06,
      "loss": 0.0057,
      "step": 33530
    },
    {
      "epoch": 0.5488014399083695,
      "grad_norm": 0.07364027139468214,
      "learning_rate": 5.020990344326023e-06,
      "loss": 0.0017,
      "step": 33540
    },
    {
      "epoch": 0.5489650658594453,
      "grad_norm": 0.27319302302887216,
      "learning_rate": 5.0181345286743875e-06,
      "loss": 0.0036,
      "step": 33550
    },
    {
      "epoch": 0.5491286918105212,
      "grad_norm": 0.3380040932206558,
      "learning_rate": 5.015278707106682e-06,
      "loss": 0.0027,
      "step": 33560
    },
    {
      "epoch": 0.549292317761597,
      "grad_norm": 0.19913843755402597,
      "learning_rate": 5.012422880554571e-06,
      "loss": 0.003,
      "step": 33570
    },
    {
      "epoch": 0.5494559437126728,
      "grad_norm": 0.13405105052743013,
      "learning_rate": 5.009567049949712e-06,
      "loss": 0.0035,
      "step": 33580
    },
    {
      "epoch": 0.5496195696637487,
      "grad_norm": 0.16442196281602223,
      "learning_rate": 5.006711216223772e-06,
      "loss": 0.0028,
      "step": 33590
    },
    {
      "epoch": 0.5497831956148245,
      "grad_norm": 0.15015553899152415,
      "learning_rate": 5.003855380308419e-06,
      "loss": 0.0025,
      "step": 33600
    },
    {
      "epoch": 0.5499468215659004,
      "grad_norm": 0.11737836370995329,
      "learning_rate": 5.000999543135314e-06,
      "loss": 0.0026,
      "step": 33610
    },
    {
      "epoch": 0.5501104475169762,
      "grad_norm": 0.13744438558998606,
      "learning_rate": 4.998143705636126e-06,
      "loss": 0.0019,
      "step": 33620
    },
    {
      "epoch": 0.550274073468052,
      "grad_norm": 0.127752365199122,
      "learning_rate": 4.995287868742523e-06,
      "loss": 0.0038,
      "step": 33630
    },
    {
      "epoch": 0.5504376994191279,
      "grad_norm": 0.04251525298122103,
      "learning_rate": 4.9924320333861686e-06,
      "loss": 0.0027,
      "step": 33640
    },
    {
      "epoch": 0.5506013253702037,
      "grad_norm": 0.28985697719302506,
      "learning_rate": 4.989576200498728e-06,
      "loss": 0.0037,
      "step": 33650
    },
    {
      "epoch": 0.5507649513212796,
      "grad_norm": 0.19167194301838225,
      "learning_rate": 4.98672037101187e-06,
      "loss": 0.0034,
      "step": 33660
    },
    {
      "epoch": 0.5509285772723554,
      "grad_norm": 0.19859354650862426,
      "learning_rate": 4.983864545857255e-06,
      "loss": 0.0022,
      "step": 33670
    },
    {
      "epoch": 0.5510922032234312,
      "grad_norm": 0.16589850815242665,
      "learning_rate": 4.981008725966545e-06,
      "loss": 0.0027,
      "step": 33680
    },
    {
      "epoch": 0.5512558291745071,
      "grad_norm": 0.08582570900098556,
      "learning_rate": 4.978152912271405e-06,
      "loss": 0.0025,
      "step": 33690
    },
    {
      "epoch": 0.5514194551255829,
      "grad_norm": 0.054666075775777624,
      "learning_rate": 4.97529710570349e-06,
      "loss": 0.0035,
      "step": 33700
    },
    {
      "epoch": 0.5515830810766588,
      "grad_norm": 0.08829880505141141,
      "learning_rate": 4.972441307194455e-06,
      "loss": 0.0016,
      "step": 33710
    },
    {
      "epoch": 0.5517467070277347,
      "grad_norm": 0.25669690909609333,
      "learning_rate": 4.96958551767596e-06,
      "loss": 0.0026,
      "step": 33720
    },
    {
      "epoch": 0.5519103329788104,
      "grad_norm": 0.10134768577943967,
      "learning_rate": 4.966729738079651e-06,
      "loss": 0.0026,
      "step": 33730
    },
    {
      "epoch": 0.5520739589298863,
      "grad_norm": 0.5135401637531692,
      "learning_rate": 4.963873969337174e-06,
      "loss": 0.0048,
      "step": 33740
    },
    {
      "epoch": 0.5522375848809621,
      "grad_norm": 0.09612637754984373,
      "learning_rate": 4.961018212380179e-06,
      "loss": 0.0018,
      "step": 33750
    },
    {
      "epoch": 0.552401210832038,
      "grad_norm": 0.07296451956170297,
      "learning_rate": 4.958162468140302e-06,
      "loss": 0.0021,
      "step": 33760
    },
    {
      "epoch": 0.5525648367831139,
      "grad_norm": 0.0690768341493563,
      "learning_rate": 4.955306737549177e-06,
      "loss": 0.0045,
      "step": 33770
    },
    {
      "epoch": 0.5527284627341896,
      "grad_norm": 0.06613254029036551,
      "learning_rate": 4.9524510215384424e-06,
      "loss": 0.0017,
      "step": 33780
    },
    {
      "epoch": 0.5528920886852655,
      "grad_norm": 0.1404849340239566,
      "learning_rate": 4.949595321039719e-06,
      "loss": 0.0016,
      "step": 33790
    },
    {
      "epoch": 0.5530557146363413,
      "grad_norm": 0.29227668471988766,
      "learning_rate": 4.94673963698463e-06,
      "loss": 0.0042,
      "step": 33800
    },
    {
      "epoch": 0.5532193405874172,
      "grad_norm": 0.07905115053964465,
      "learning_rate": 4.943883970304795e-06,
      "loss": 0.003,
      "step": 33810
    },
    {
      "epoch": 0.5533829665384931,
      "grad_norm": 0.35047513622471355,
      "learning_rate": 4.941028321931821e-06,
      "loss": 0.0036,
      "step": 33820
    },
    {
      "epoch": 0.5535465924895688,
      "grad_norm": 0.19227056667487888,
      "learning_rate": 4.938172692797313e-06,
      "loss": 0.0022,
      "step": 33830
    },
    {
      "epoch": 0.5537102184406447,
      "grad_norm": 0.15843649704867058,
      "learning_rate": 4.9353170838328725e-06,
      "loss": 0.0018,
      "step": 33840
    },
    {
      "epoch": 0.5538738443917205,
      "grad_norm": 0.40886755463128766,
      "learning_rate": 4.9324614959700894e-06,
      "loss": 0.0052,
      "step": 33850
    },
    {
      "epoch": 0.5540374703427964,
      "grad_norm": 0.09701874312030412,
      "learning_rate": 4.929605930140546e-06,
      "loss": 0.0023,
      "step": 33860
    },
    {
      "epoch": 0.5542010962938722,
      "grad_norm": 0.11219644676221913,
      "learning_rate": 4.9267503872758275e-06,
      "loss": 0.0028,
      "step": 33870
    },
    {
      "epoch": 0.554364722244948,
      "grad_norm": 0.20261954232756732,
      "learning_rate": 4.9238948683074975e-06,
      "loss": 0.0044,
      "step": 33880
    },
    {
      "epoch": 0.5545283481960239,
      "grad_norm": 0.10757315193541118,
      "learning_rate": 4.921039374167121e-06,
      "loss": 0.0022,
      "step": 33890
    },
    {
      "epoch": 0.5546919741470997,
      "grad_norm": 0.19342549766791128,
      "learning_rate": 4.918183905786251e-06,
      "loss": 0.0019,
      "step": 33900
    },
    {
      "epoch": 0.5548556000981756,
      "grad_norm": 0.24104349304967013,
      "learning_rate": 4.915328464096436e-06,
      "loss": 0.0029,
      "step": 33910
    },
    {
      "epoch": 0.5550192260492514,
      "grad_norm": 0.3863572505902604,
      "learning_rate": 4.912473050029212e-06,
      "loss": 0.0026,
      "step": 33920
    },
    {
      "epoch": 0.5551828520003272,
      "grad_norm": 0.3262174658297306,
      "learning_rate": 4.909617664516107e-06,
      "loss": 0.0034,
      "step": 33930
    },
    {
      "epoch": 0.5553464779514031,
      "grad_norm": 0.1800889387564932,
      "learning_rate": 4.906762308488641e-06,
      "loss": 0.0031,
      "step": 33940
    },
    {
      "epoch": 0.5555101039024789,
      "grad_norm": 0.09749444915029594,
      "learning_rate": 4.903906982878324e-06,
      "loss": 0.0045,
      "step": 33950
    },
    {
      "epoch": 0.5556737298535548,
      "grad_norm": 0.0666383914937107,
      "learning_rate": 4.901051688616652e-06,
      "loss": 0.0027,
      "step": 33960
    },
    {
      "epoch": 0.5558373558046306,
      "grad_norm": 0.05634549849460267,
      "learning_rate": 4.898196426635119e-06,
      "loss": 0.004,
      "step": 33970
    },
    {
      "epoch": 0.5560009817557064,
      "grad_norm": 0.12841463188721838,
      "learning_rate": 4.895341197865201e-06,
      "loss": 0.0023,
      "step": 33980
    },
    {
      "epoch": 0.5561646077067823,
      "grad_norm": 0.10917926943019311,
      "learning_rate": 4.8924860032383656e-06,
      "loss": 0.0037,
      "step": 33990
    },
    {
      "epoch": 0.5563282336578581,
      "grad_norm": 0.025895706420346245,
      "learning_rate": 4.88963084368607e-06,
      "loss": 0.0033,
      "step": 34000
    },
    {
      "epoch": 0.556491859608934,
      "grad_norm": 0.04506525106912019,
      "learning_rate": 4.886775720139762e-06,
      "loss": 0.003,
      "step": 34010
    },
    {
      "epoch": 0.5566554855600098,
      "grad_norm": 0.05898426144768243,
      "learning_rate": 4.88392063353087e-06,
      "loss": 0.0028,
      "step": 34020
    },
    {
      "epoch": 0.5568191115110857,
      "grad_norm": 0.15042547484089042,
      "learning_rate": 4.881065584790821e-06,
      "loss": 0.003,
      "step": 34030
    },
    {
      "epoch": 0.5569827374621615,
      "grad_norm": 0.1674247857008809,
      "learning_rate": 4.8782105748510205e-06,
      "loss": 0.0029,
      "step": 34040
    },
    {
      "epoch": 0.5571463634132373,
      "grad_norm": 0.23741060395347063,
      "learning_rate": 4.875355604642866e-06,
      "loss": 0.0034,
      "step": 34050
    },
    {
      "epoch": 0.5573099893643132,
      "grad_norm": 0.2556491577071466,
      "learning_rate": 4.872500675097741e-06,
      "loss": 0.002,
      "step": 34060
    },
    {
      "epoch": 0.557473615315389,
      "grad_norm": 0.18550731753118735,
      "learning_rate": 4.869645787147017e-06,
      "loss": 0.0025,
      "step": 34070
    },
    {
      "epoch": 0.5576372412664649,
      "grad_norm": 0.09432895896236418,
      "learning_rate": 4.866790941722048e-06,
      "loss": 0.0041,
      "step": 34080
    },
    {
      "epoch": 0.5578008672175407,
      "grad_norm": 0.1944323765104268,
      "learning_rate": 4.86393613975418e-06,
      "loss": 0.0035,
      "step": 34090
    },
    {
      "epoch": 0.5579644931686165,
      "grad_norm": 0.2422932611855969,
      "learning_rate": 4.861081382174741e-06,
      "loss": 0.0033,
      "step": 34100
    },
    {
      "epoch": 0.5581281191196924,
      "grad_norm": 0.11254669175779865,
      "learning_rate": 4.858226669915042e-06,
      "loss": 0.0031,
      "step": 34110
    },
    {
      "epoch": 0.5582917450707682,
      "grad_norm": 0.10623624942837201,
      "learning_rate": 4.855372003906386e-06,
      "loss": 0.0033,
      "step": 34120
    },
    {
      "epoch": 0.5584553710218441,
      "grad_norm": 0.07576238211488211,
      "learning_rate": 4.852517385080055e-06,
      "loss": 0.0019,
      "step": 34130
    },
    {
      "epoch": 0.55861899697292,
      "grad_norm": 0.20947958705101766,
      "learning_rate": 4.8496628143673196e-06,
      "loss": 0.0036,
      "step": 34140
    },
    {
      "epoch": 0.5587826229239957,
      "grad_norm": 0.08025112510771956,
      "learning_rate": 4.846808292699432e-06,
      "loss": 0.0023,
      "step": 34150
    },
    {
      "epoch": 0.5589462488750716,
      "grad_norm": 0.03933167879665618,
      "learning_rate": 4.843953821007631e-06,
      "loss": 0.002,
      "step": 34160
    },
    {
      "epoch": 0.5591098748261474,
      "grad_norm": 0.14153836070683692,
      "learning_rate": 4.841099400223135e-06,
      "loss": 0.0021,
      "step": 34170
    },
    {
      "epoch": 0.5592735007772233,
      "grad_norm": 0.2761771339525101,
      "learning_rate": 4.83824503127715e-06,
      "loss": 0.0032,
      "step": 34180
    },
    {
      "epoch": 0.5594371267282992,
      "grad_norm": 0.013084796565139185,
      "learning_rate": 4.835390715100864e-06,
      "loss": 0.0025,
      "step": 34190
    },
    {
      "epoch": 0.5596007526793749,
      "grad_norm": 0.17115944385719223,
      "learning_rate": 4.832536452625444e-06,
      "loss": 0.0036,
      "step": 34200
    },
    {
      "epoch": 0.5597643786304508,
      "grad_norm": 0.2139708289428106,
      "learning_rate": 4.829682244782046e-06,
      "loss": 0.0029,
      "step": 34210
    },
    {
      "epoch": 0.5599280045815266,
      "grad_norm": 0.1250146382281352,
      "learning_rate": 4.8268280925018034e-06,
      "loss": 0.0021,
      "step": 34220
    },
    {
      "epoch": 0.5600916305326025,
      "grad_norm": 0.08994884800786558,
      "learning_rate": 4.8239739967158335e-06,
      "loss": 0.0018,
      "step": 34230
    },
    {
      "epoch": 0.5602552564836784,
      "grad_norm": 0.18660576986633415,
      "learning_rate": 4.821119958355233e-06,
      "loss": 0.0029,
      "step": 34240
    },
    {
      "epoch": 0.5604188824347541,
      "grad_norm": 0.08006527165237609,
      "learning_rate": 4.818265978351084e-06,
      "loss": 0.0017,
      "step": 34250
    },
    {
      "epoch": 0.56058250838583,
      "grad_norm": 0.2476745725678214,
      "learning_rate": 4.815412057634445e-06,
      "loss": 0.0026,
      "step": 34260
    },
    {
      "epoch": 0.5607461343369058,
      "grad_norm": 0.060597485547975505,
      "learning_rate": 4.812558197136358e-06,
      "loss": 0.0025,
      "step": 34270
    },
    {
      "epoch": 0.5609097602879817,
      "grad_norm": 0.22124179620506648,
      "learning_rate": 4.8097043977878445e-06,
      "loss": 0.0027,
      "step": 34280
    },
    {
      "epoch": 0.5610733862390576,
      "grad_norm": 0.1700427490529885,
      "learning_rate": 4.806850660519906e-06,
      "loss": 0.0037,
      "step": 34290
    },
    {
      "epoch": 0.5612370121901333,
      "grad_norm": 0.5274311959724609,
      "learning_rate": 4.803996986263523e-06,
      "loss": 0.004,
      "step": 34300
    },
    {
      "epoch": 0.5614006381412092,
      "grad_norm": 0.3455460538436016,
      "learning_rate": 4.801143375949657e-06,
      "loss": 0.0026,
      "step": 34310
    },
    {
      "epoch": 0.561564264092285,
      "grad_norm": 0.38129445487355873,
      "learning_rate": 4.798289830509249e-06,
      "loss": 0.0043,
      "step": 34320
    },
    {
      "epoch": 0.5617278900433609,
      "grad_norm": 0.17586702686796832,
      "learning_rate": 4.7954363508732145e-06,
      "loss": 0.0029,
      "step": 34330
    },
    {
      "epoch": 0.5618915159944368,
      "grad_norm": 0.24283639700390494,
      "learning_rate": 4.792582937972455e-06,
      "loss": 0.005,
      "step": 34340
    },
    {
      "epoch": 0.5620551419455125,
      "grad_norm": 0.1870736059601488,
      "learning_rate": 4.7897295927378435e-06,
      "loss": 0.0041,
      "step": 34350
    },
    {
      "epoch": 0.5622187678965884,
      "grad_norm": 0.2168414949137978,
      "learning_rate": 4.786876316100233e-06,
      "loss": 0.0025,
      "step": 34360
    },
    {
      "epoch": 0.5623823938476642,
      "grad_norm": 0.03875193052343486,
      "learning_rate": 4.784023108990456e-06,
      "loss": 0.0018,
      "step": 34370
    },
    {
      "epoch": 0.5625460197987401,
      "grad_norm": 0.2533499606070049,
      "learning_rate": 4.7811699723393205e-06,
      "loss": 0.0039,
      "step": 34380
    },
    {
      "epoch": 0.562709645749816,
      "grad_norm": 0.15105636757304827,
      "learning_rate": 4.7783169070776095e-06,
      "loss": 0.0058,
      "step": 34390
    },
    {
      "epoch": 0.5628732717008917,
      "grad_norm": 0.04670597080851487,
      "learning_rate": 4.775463914136089e-06,
      "loss": 0.0016,
      "step": 34400
    },
    {
      "epoch": 0.5630368976519676,
      "grad_norm": 0.2940880891102453,
      "learning_rate": 4.772610994445496e-06,
      "loss": 0.0017,
      "step": 34410
    },
    {
      "epoch": 0.5632005236030434,
      "grad_norm": 0.17700238963468307,
      "learning_rate": 4.769758148936543e-06,
      "loss": 0.003,
      "step": 34420
    },
    {
      "epoch": 0.5633641495541193,
      "grad_norm": 0.18867232861400654,
      "learning_rate": 4.766905378539923e-06,
      "loss": 0.0029,
      "step": 34430
    },
    {
      "epoch": 0.5635277755051952,
      "grad_norm": 0.2760582445608577,
      "learning_rate": 4.764052684186302e-06,
      "loss": 0.0027,
      "step": 34440
    },
    {
      "epoch": 0.563691401456271,
      "grad_norm": 0.07304146055666083,
      "learning_rate": 4.761200066806319e-06,
      "loss": 0.0021,
      "step": 34450
    },
    {
      "epoch": 0.5638550274073468,
      "grad_norm": 0.4745721174776518,
      "learning_rate": 4.758347527330591e-06,
      "loss": 0.0053,
      "step": 34460
    },
    {
      "epoch": 0.5640186533584226,
      "grad_norm": 0.22246678722086954,
      "learning_rate": 4.7554950666897094e-06,
      "loss": 0.0032,
      "step": 34470
    },
    {
      "epoch": 0.5641822793094985,
      "grad_norm": 0.10964403914953515,
      "learning_rate": 4.752642685814238e-06,
      "loss": 0.0031,
      "step": 34480
    },
    {
      "epoch": 0.5643459052605744,
      "grad_norm": 0.18020922498121175,
      "learning_rate": 4.749790385634716e-06,
      "loss": 0.0046,
      "step": 34490
    },
    {
      "epoch": 0.5645095312116502,
      "grad_norm": 0.2612613727880996,
      "learning_rate": 4.746938167081656e-06,
      "loss": 0.004,
      "step": 34500
    },
    {
      "epoch": 0.564673157162726,
      "grad_norm": 0.34073258867079126,
      "learning_rate": 4.7440860310855426e-06,
      "loss": 0.0017,
      "step": 34510
    },
    {
      "epoch": 0.5648367831138018,
      "grad_norm": 0.15262172659323311,
      "learning_rate": 4.741233978576837e-06,
      "loss": 0.0031,
      "step": 34520
    },
    {
      "epoch": 0.5650004090648777,
      "grad_norm": 0.2001287798321962,
      "learning_rate": 4.738382010485969e-06,
      "loss": 0.0037,
      "step": 34530
    },
    {
      "epoch": 0.5651640350159536,
      "grad_norm": 0.04479679502364937,
      "learning_rate": 4.7355301277433455e-06,
      "loss": 0.0024,
      "step": 34540
    },
    {
      "epoch": 0.5653276609670294,
      "grad_norm": 0.0442003310591338,
      "learning_rate": 4.732678331279337e-06,
      "loss": 0.0023,
      "step": 34550
    },
    {
      "epoch": 0.5654912869181052,
      "grad_norm": 0.17239141213988352,
      "learning_rate": 4.729826622024297e-06,
      "loss": 0.0034,
      "step": 34560
    },
    {
      "epoch": 0.565654912869181,
      "grad_norm": 0.2931721128750487,
      "learning_rate": 4.726975000908546e-06,
      "loss": 0.0038,
      "step": 34570
    },
    {
      "epoch": 0.5658185388202569,
      "grad_norm": 0.147521649636593,
      "learning_rate": 4.724123468862368e-06,
      "loss": 0.0032,
      "step": 34580
    },
    {
      "epoch": 0.5659821647713328,
      "grad_norm": 0.19929045062751566,
      "learning_rate": 4.721272026816034e-06,
      "loss": 0.0023,
      "step": 34590
    },
    {
      "epoch": 0.5661457907224086,
      "grad_norm": 0.31199198775033304,
      "learning_rate": 4.7184206756997705e-06,
      "loss": 0.0032,
      "step": 34600
    },
    {
      "epoch": 0.5663094166734844,
      "grad_norm": 0.26629633213529075,
      "learning_rate": 4.7155694164437794e-06,
      "loss": 0.0027,
      "step": 34610
    },
    {
      "epoch": 0.5664730426245602,
      "grad_norm": 0.4314375273907548,
      "learning_rate": 4.7127182499782394e-06,
      "loss": 0.0056,
      "step": 34620
    },
    {
      "epoch": 0.5666366685756361,
      "grad_norm": 0.0823529941841386,
      "learning_rate": 4.709867177233288e-06,
      "loss": 0.0013,
      "step": 34630
    },
    {
      "epoch": 0.566800294526712,
      "grad_norm": 0.31904225921200446,
      "learning_rate": 4.707016199139039e-06,
      "loss": 0.0036,
      "step": 34640
    },
    {
      "epoch": 0.5669639204777878,
      "grad_norm": 0.065714048697883,
      "learning_rate": 4.704165316625575e-06,
      "loss": 0.0025,
      "step": 34650
    },
    {
      "epoch": 0.5671275464288636,
      "grad_norm": 0.019652022695815365,
      "learning_rate": 4.701314530622945e-06,
      "loss": 0.0032,
      "step": 34660
    },
    {
      "epoch": 0.5672911723799394,
      "grad_norm": 0.11393755230884864,
      "learning_rate": 4.698463842061164e-06,
      "loss": 0.0043,
      "step": 34670
    },
    {
      "epoch": 0.5674547983310153,
      "grad_norm": 0.23481955593949247,
      "learning_rate": 4.695613251870226e-06,
      "loss": 0.0035,
      "step": 34680
    },
    {
      "epoch": 0.5676184242820912,
      "grad_norm": 0.2430415671057859,
      "learning_rate": 4.69276276098008e-06,
      "loss": 0.0029,
      "step": 34690
    },
    {
      "epoch": 0.567782050233167,
      "grad_norm": 0.09372633377784939,
      "learning_rate": 4.689912370320647e-06,
      "loss": 0.0029,
      "step": 34700
    },
    {
      "epoch": 0.5679456761842429,
      "grad_norm": 0.1737665124954546,
      "learning_rate": 4.687062080821824e-06,
      "loss": 0.0024,
      "step": 34710
    },
    {
      "epoch": 0.5681093021353186,
      "grad_norm": 0.26333387980378437,
      "learning_rate": 4.68421189341346e-06,
      "loss": 0.0023,
      "step": 34720
    },
    {
      "epoch": 0.5682729280863945,
      "grad_norm": 0.15374022618927896,
      "learning_rate": 4.681361809025381e-06,
      "loss": 0.0019,
      "step": 34730
    },
    {
      "epoch": 0.5684365540374703,
      "grad_norm": 0.12217669515144067,
      "learning_rate": 4.6785118285873795e-06,
      "loss": 0.0027,
      "step": 34740
    },
    {
      "epoch": 0.5686001799885462,
      "grad_norm": 0.20773944903862368,
      "learning_rate": 4.675661953029207e-06,
      "loss": 0.002,
      "step": 34750
    },
    {
      "epoch": 0.5687638059396221,
      "grad_norm": 0.23256929227538928,
      "learning_rate": 4.672812183280584e-06,
      "loss": 0.0026,
      "step": 34760
    },
    {
      "epoch": 0.5689274318906978,
      "grad_norm": 0.19365959146065578,
      "learning_rate": 4.669962520271204e-06,
      "loss": 0.004,
      "step": 34770
    },
    {
      "epoch": 0.5690910578417737,
      "grad_norm": 0.23368407210628542,
      "learning_rate": 4.667112964930714e-06,
      "loss": 0.0042,
      "step": 34780
    },
    {
      "epoch": 0.5692546837928495,
      "grad_norm": 0.1469614490142267,
      "learning_rate": 4.6642635181887295e-06,
      "loss": 0.0045,
      "step": 34790
    },
    {
      "epoch": 0.5694183097439254,
      "grad_norm": 0.08581751202717164,
      "learning_rate": 4.6614141809748385e-06,
      "loss": 0.004,
      "step": 34800
    },
    {
      "epoch": 0.5695819356950013,
      "grad_norm": 0.27520429857050993,
      "learning_rate": 4.658564954218581e-06,
      "loss": 0.0031,
      "step": 34810
    },
    {
      "epoch": 0.569745561646077,
      "grad_norm": 0.2552820164942705,
      "learning_rate": 4.6557158388494676e-06,
      "loss": 0.0033,
      "step": 34820
    },
    {
      "epoch": 0.5699091875971529,
      "grad_norm": 0.3122888841833536,
      "learning_rate": 4.652866835796977e-06,
      "loss": 0.0036,
      "step": 34830
    },
    {
      "epoch": 0.5700728135482287,
      "grad_norm": 0.185488168629714,
      "learning_rate": 4.65001794599054e-06,
      "loss": 0.0035,
      "step": 34840
    },
    {
      "epoch": 0.5702364394993046,
      "grad_norm": 0.1091345426121472,
      "learning_rate": 4.647169170359558e-06,
      "loss": 0.0021,
      "step": 34850
    },
    {
      "epoch": 0.5704000654503805,
      "grad_norm": 0.198385665576042,
      "learning_rate": 4.644320509833397e-06,
      "loss": 0.002,
      "step": 34860
    },
    {
      "epoch": 0.5705636914014562,
      "grad_norm": 0.42822651029816045,
      "learning_rate": 4.641471965341378e-06,
      "loss": 0.0058,
      "step": 34870
    },
    {
      "epoch": 0.5707273173525321,
      "grad_norm": 0.18266343768138915,
      "learning_rate": 4.63862353781279e-06,
      "loss": 0.004,
      "step": 34880
    },
    {
      "epoch": 0.5708909433036079,
      "grad_norm": 0.04457514956056342,
      "learning_rate": 4.635775228176882e-06,
      "loss": 0.003,
      "step": 34890
    },
    {
      "epoch": 0.5710545692546838,
      "grad_norm": 0.3484718784970026,
      "learning_rate": 4.6329270373628645e-06,
      "loss": 0.0049,
      "step": 34900
    },
    {
      "epoch": 0.5712181952057597,
      "grad_norm": 0.0647244875524105,
      "learning_rate": 4.63007896629991e-06,
      "loss": 0.0041,
      "step": 34910
    },
    {
      "epoch": 0.5713818211568354,
      "grad_norm": 0.15067403917686212,
      "learning_rate": 4.627231015917149e-06,
      "loss": 0.0017,
      "step": 34920
    },
    {
      "epoch": 0.5715454471079113,
      "grad_norm": 0.138335357204624,
      "learning_rate": 4.624383187143678e-06,
      "loss": 0.0031,
      "step": 34930
    },
    {
      "epoch": 0.5717090730589871,
      "grad_norm": 0.3433662125079007,
      "learning_rate": 4.621535480908551e-06,
      "loss": 0.005,
      "step": 34940
    },
    {
      "epoch": 0.571872699010063,
      "grad_norm": 0.2781255736222882,
      "learning_rate": 4.618687898140777e-06,
      "loss": 0.0018,
      "step": 34950
    },
    {
      "epoch": 0.5720363249611389,
      "grad_norm": 0.16556140968733749,
      "learning_rate": 4.615840439769335e-06,
      "loss": 0.0035,
      "step": 34960
    },
    {
      "epoch": 0.5721999509122146,
      "grad_norm": 0.11192981865341041,
      "learning_rate": 4.612993106723156e-06,
      "loss": 0.0034,
      "step": 34970
    },
    {
      "epoch": 0.5723635768632905,
      "grad_norm": 0.18455268316402856,
      "learning_rate": 4.6101458999311305e-06,
      "loss": 0.0026,
      "step": 34980
    },
    {
      "epoch": 0.5725272028143663,
      "grad_norm": 0.09519673337402712,
      "learning_rate": 4.607298820322112e-06,
      "loss": 0.0022,
      "step": 34990
    },
    {
      "epoch": 0.5726908287654422,
      "grad_norm": 0.28592452873446594,
      "learning_rate": 4.604451868824908e-06,
      "loss": 0.0042,
      "step": 35000
    },
    {
      "epoch": 0.5728544547165181,
      "grad_norm": 0.06808254018368712,
      "learning_rate": 4.601605046368286e-06,
      "loss": 0.0027,
      "step": 35010
    },
    {
      "epoch": 0.5730180806675939,
      "grad_norm": 0.1961892191344398,
      "learning_rate": 4.598758353880973e-06,
      "loss": 0.0031,
      "step": 35020
    },
    {
      "epoch": 0.5731817066186697,
      "grad_norm": 0.004998944998617602,
      "learning_rate": 4.595911792291651e-06,
      "loss": 0.0023,
      "step": 35030
    },
    {
      "epoch": 0.5733453325697455,
      "grad_norm": 0.09886125808936169,
      "learning_rate": 4.59306536252896e-06,
      "loss": 0.0019,
      "step": 35040
    },
    {
      "epoch": 0.5735089585208214,
      "grad_norm": 0.12288144918918424,
      "learning_rate": 4.590219065521498e-06,
      "loss": 0.0028,
      "step": 35050
    },
    {
      "epoch": 0.5736725844718973,
      "grad_norm": 0.16628383535940017,
      "learning_rate": 4.58737290219782e-06,
      "loss": 0.0024,
      "step": 35060
    },
    {
      "epoch": 0.5738362104229731,
      "grad_norm": 0.23073372945036774,
      "learning_rate": 4.584526873486434e-06,
      "loss": 0.0043,
      "step": 35070
    },
    {
      "epoch": 0.5739998363740489,
      "grad_norm": 0.04255723519591567,
      "learning_rate": 4.58168098031581e-06,
      "loss": 0.0024,
      "step": 35080
    },
    {
      "epoch": 0.5741634623251247,
      "grad_norm": 0.29005401785023754,
      "learning_rate": 4.578835223614367e-06,
      "loss": 0.0035,
      "step": 35090
    },
    {
      "epoch": 0.5743270882762006,
      "grad_norm": 0.14690027524514798,
      "learning_rate": 4.575989604310483e-06,
      "loss": 0.0036,
      "step": 35100
    },
    {
      "epoch": 0.5744907142272765,
      "grad_norm": 0.12595642158838552,
      "learning_rate": 4.573144123332494e-06,
      "loss": 0.0026,
      "step": 35110
    },
    {
      "epoch": 0.5746543401783523,
      "grad_norm": 0.2763350987435719,
      "learning_rate": 4.570298781608686e-06,
      "loss": 0.0029,
      "step": 35120
    },
    {
      "epoch": 0.5748179661294281,
      "grad_norm": 0.35190145688776797,
      "learning_rate": 4.5674535800673e-06,
      "loss": 0.0024,
      "step": 35130
    },
    {
      "epoch": 0.5749815920805039,
      "grad_norm": 0.07416830453420056,
      "learning_rate": 4.564608519636535e-06,
      "loss": 0.0026,
      "step": 35140
    },
    {
      "epoch": 0.5751452180315798,
      "grad_norm": 0.4236063947368003,
      "learning_rate": 4.561763601244541e-06,
      "loss": 0.0027,
      "step": 35150
    },
    {
      "epoch": 0.5753088439826557,
      "grad_norm": 0.08428294166818885,
      "learning_rate": 4.55891882581942e-06,
      "loss": 0.0021,
      "step": 35160
    },
    {
      "epoch": 0.5754724699337315,
      "grad_norm": 0.19782619085405367,
      "learning_rate": 4.5560741942892335e-06,
      "loss": 0.003,
      "step": 35170
    },
    {
      "epoch": 0.5756360958848074,
      "grad_norm": 0.44167748705152654,
      "learning_rate": 4.553229707581991e-06,
      "loss": 0.0017,
      "step": 35180
    },
    {
      "epoch": 0.5757997218358831,
      "grad_norm": 0.3324364830142215,
      "learning_rate": 4.550385366625653e-06,
      "loss": 0.0055,
      "step": 35190
    },
    {
      "epoch": 0.575963347786959,
      "grad_norm": 0.3023508666240383,
      "learning_rate": 4.54754117234814e-06,
      "loss": 0.003,
      "step": 35200
    },
    {
      "epoch": 0.5761269737380349,
      "grad_norm": 0.22533043942256858,
      "learning_rate": 4.544697125677317e-06,
      "loss": 0.0047,
      "step": 35210
    },
    {
      "epoch": 0.5762905996891107,
      "grad_norm": 0.20307455718785172,
      "learning_rate": 4.5418532275410045e-06,
      "loss": 0.0031,
      "step": 35220
    },
    {
      "epoch": 0.5764542256401866,
      "grad_norm": 0.08921232291241027,
      "learning_rate": 4.539009478866975e-06,
      "loss": 0.005,
      "step": 35230
    },
    {
      "epoch": 0.5766178515912623,
      "grad_norm": 0.10398808734965202,
      "learning_rate": 4.53616588058295e-06,
      "loss": 0.0025,
      "step": 35240
    },
    {
      "epoch": 0.5767814775423382,
      "grad_norm": 0.06834456850097631,
      "learning_rate": 4.533322433616604e-06,
      "loss": 0.003,
      "step": 35250
    },
    {
      "epoch": 0.5769451034934141,
      "grad_norm": 0.1329335882340543,
      "learning_rate": 4.530479138895561e-06,
      "loss": 0.0018,
      "step": 35260
    },
    {
      "epoch": 0.5771087294444899,
      "grad_norm": 0.18961894567010026,
      "learning_rate": 4.527635997347395e-06,
      "loss": 0.0018,
      "step": 35270
    },
    {
      "epoch": 0.5772723553955658,
      "grad_norm": 0.12992920761495436,
      "learning_rate": 4.524793009899633e-06,
      "loss": 0.0023,
      "step": 35280
    },
    {
      "epoch": 0.5774359813466415,
      "grad_norm": 0.028874766206016186,
      "learning_rate": 4.521950177479745e-06,
      "loss": 0.0024,
      "step": 35290
    },
    {
      "epoch": 0.5775996072977174,
      "grad_norm": 0.11220881517690003,
      "learning_rate": 4.519107501015159e-06,
      "loss": 0.0019,
      "step": 35300
    },
    {
      "epoch": 0.5777632332487933,
      "grad_norm": 0.2345739368095933,
      "learning_rate": 4.516264981433246e-06,
      "loss": 0.0029,
      "step": 35310
    },
    {
      "epoch": 0.5779268591998691,
      "grad_norm": 0.16103023796128926,
      "learning_rate": 4.513422619661328e-06,
      "loss": 0.0031,
      "step": 35320
    },
    {
      "epoch": 0.578090485150945,
      "grad_norm": 0.1860806586145409,
      "learning_rate": 4.510580416626675e-06,
      "loss": 0.0034,
      "step": 35330
    },
    {
      "epoch": 0.5782541111020207,
      "grad_norm": 0.1850667541021402,
      "learning_rate": 4.507738373256507e-06,
      "loss": 0.0036,
      "step": 35340
    },
    {
      "epoch": 0.5784177370530966,
      "grad_norm": 0.10324916870975388,
      "learning_rate": 4.504896490477988e-06,
      "loss": 0.0035,
      "step": 35350
    },
    {
      "epoch": 0.5785813630041725,
      "grad_norm": 0.5119864671505694,
      "learning_rate": 4.502054769218234e-06,
      "loss": 0.0027,
      "step": 35360
    },
    {
      "epoch": 0.5787449889552483,
      "grad_norm": 0.32733033073044265,
      "learning_rate": 4.499213210404305e-06,
      "loss": 0.0034,
      "step": 35370
    },
    {
      "epoch": 0.5789086149063242,
      "grad_norm": 0.4681426877133344,
      "learning_rate": 4.49637181496321e-06,
      "loss": 0.0036,
      "step": 35380
    },
    {
      "epoch": 0.5790722408574,
      "grad_norm": 0.18864255296210705,
      "learning_rate": 4.493530583821904e-06,
      "loss": 0.0037,
      "step": 35390
    },
    {
      "epoch": 0.5792358668084758,
      "grad_norm": 0.21523847610998229,
      "learning_rate": 4.490689517907288e-06,
      "loss": 0.0051,
      "step": 35400
    },
    {
      "epoch": 0.5793994927595517,
      "grad_norm": 0.20821127391843855,
      "learning_rate": 4.487848618146209e-06,
      "loss": 0.0042,
      "step": 35410
    },
    {
      "epoch": 0.5795631187106275,
      "grad_norm": 0.016955066373417756,
      "learning_rate": 4.485007885465462e-06,
      "loss": 0.003,
      "step": 35420
    },
    {
      "epoch": 0.5797267446617034,
      "grad_norm": 0.13746991830458877,
      "learning_rate": 4.482167320791786e-06,
      "loss": 0.0024,
      "step": 35430
    },
    {
      "epoch": 0.5798903706127791,
      "grad_norm": 0.3522013700940443,
      "learning_rate": 4.479326925051862e-06,
      "loss": 0.0041,
      "step": 35440
    },
    {
      "epoch": 0.580053996563855,
      "grad_norm": 0.46507976524046846,
      "learning_rate": 4.476486699172322e-06,
      "loss": 0.0033,
      "step": 35450
    },
    {
      "epoch": 0.5802176225149309,
      "grad_norm": 0.6915992970846698,
      "learning_rate": 4.473646644079738e-06,
      "loss": 0.0029,
      "step": 35460
    },
    {
      "epoch": 0.5803812484660067,
      "grad_norm": 0.020799982935066054,
      "learning_rate": 4.470806760700627e-06,
      "loss": 0.0027,
      "step": 35470
    },
    {
      "epoch": 0.5805448744170826,
      "grad_norm": 0.27726665504849785,
      "learning_rate": 4.467967049961452e-06,
      "loss": 0.0029,
      "step": 35480
    },
    {
      "epoch": 0.5807085003681584,
      "grad_norm": 0.17385091629629307,
      "learning_rate": 4.465127512788619e-06,
      "loss": 0.0031,
      "step": 35490
    },
    {
      "epoch": 0.5808721263192342,
      "grad_norm": 0.07759368622562746,
      "learning_rate": 4.4622881501084745e-06,
      "loss": 0.0021,
      "step": 35500
    },
    {
      "epoch": 0.5810357522703101,
      "grad_norm": 0.015916280396538602,
      "learning_rate": 4.459448962847313e-06,
      "loss": 0.0025,
      "step": 35510
    },
    {
      "epoch": 0.5811993782213859,
      "grad_norm": 0.14307596543750917,
      "learning_rate": 4.456609951931367e-06,
      "loss": 0.0029,
      "step": 35520
    },
    {
      "epoch": 0.5813630041724618,
      "grad_norm": 0.045026270071096046,
      "learning_rate": 4.4537711182868164e-06,
      "loss": 0.0027,
      "step": 35530
    },
    {
      "epoch": 0.5815266301235376,
      "grad_norm": 0.1875311325041476,
      "learning_rate": 4.450932462839775e-06,
      "loss": 0.0029,
      "step": 35540
    },
    {
      "epoch": 0.5816902560746134,
      "grad_norm": 0.046246788565296154,
      "learning_rate": 4.44809398651631e-06,
      "loss": 0.0029,
      "step": 35550
    },
    {
      "epoch": 0.5818538820256893,
      "grad_norm": 0.20613248243906757,
      "learning_rate": 4.445255690242422e-06,
      "loss": 0.0021,
      "step": 35560
    },
    {
      "epoch": 0.5820175079767651,
      "grad_norm": 0.12109330242773472,
      "learning_rate": 4.442417574944052e-06,
      "loss": 0.003,
      "step": 35570
    },
    {
      "epoch": 0.582181133927841,
      "grad_norm": 0.07933825244372066,
      "learning_rate": 4.439579641547089e-06,
      "loss": 0.0037,
      "step": 35580
    },
    {
      "epoch": 0.5823447598789168,
      "grad_norm": 0.20272782117099777,
      "learning_rate": 4.4367418909773595e-06,
      "loss": 0.0026,
      "step": 35590
    },
    {
      "epoch": 0.5825083858299926,
      "grad_norm": 0.25216766311627675,
      "learning_rate": 4.433904324160622e-06,
      "loss": 0.0049,
      "step": 35600
    },
    {
      "epoch": 0.5826720117810685,
      "grad_norm": 0.129372277136597,
      "learning_rate": 4.431066942022591e-06,
      "loss": 0.0029,
      "step": 35610
    },
    {
      "epoch": 0.5828356377321443,
      "grad_norm": 0.03031690936312735,
      "learning_rate": 4.428229745488909e-06,
      "loss": 0.0033,
      "step": 35620
    },
    {
      "epoch": 0.5829992636832202,
      "grad_norm": 0.20997047578296665,
      "learning_rate": 4.425392735485158e-06,
      "loss": 0.005,
      "step": 35630
    },
    {
      "epoch": 0.583162889634296,
      "grad_norm": 0.19467087799413413,
      "learning_rate": 4.4225559129368675e-06,
      "loss": 0.0041,
      "step": 35640
    },
    {
      "epoch": 0.5833265155853719,
      "grad_norm": 0.23683170458369324,
      "learning_rate": 4.419719278769499e-06,
      "loss": 0.004,
      "step": 35650
    },
    {
      "epoch": 0.5834901415364476,
      "grad_norm": 0.0963018572373444,
      "learning_rate": 4.41688283390845e-06,
      "loss": 0.0038,
      "step": 35660
    },
    {
      "epoch": 0.5836537674875235,
      "grad_norm": 0.12351831580258514,
      "learning_rate": 4.414046579279068e-06,
      "loss": 0.0026,
      "step": 35670
    },
    {
      "epoch": 0.5838173934385994,
      "grad_norm": 0.2287933481164842,
      "learning_rate": 4.411210515806627e-06,
      "loss": 0.0026,
      "step": 35680
    },
    {
      "epoch": 0.5839810193896752,
      "grad_norm": 0.14680820117195276,
      "learning_rate": 4.40837464441634e-06,
      "loss": 0.0031,
      "step": 35690
    },
    {
      "epoch": 0.5841446453407511,
      "grad_norm": 0.04870131994647957,
      "learning_rate": 4.405538966033364e-06,
      "loss": 0.0009,
      "step": 35700
    },
    {
      "epoch": 0.5843082712918268,
      "grad_norm": 0.25407334832327355,
      "learning_rate": 4.40270348158279e-06,
      "loss": 0.0035,
      "step": 35710
    },
    {
      "epoch": 0.5844718972429027,
      "grad_norm": 0.30090296303199754,
      "learning_rate": 4.399868191989637e-06,
      "loss": 0.003,
      "step": 35720
    },
    {
      "epoch": 0.5846355231939786,
      "grad_norm": 0.3024970167537329,
      "learning_rate": 4.397033098178876e-06,
      "loss": 0.0027,
      "step": 35730
    },
    {
      "epoch": 0.5847991491450544,
      "grad_norm": 0.062468027634000146,
      "learning_rate": 4.394198201075405e-06,
      "loss": 0.003,
      "step": 35740
    },
    {
      "epoch": 0.5849627750961303,
      "grad_norm": 0.18311095051418277,
      "learning_rate": 4.391363501604054e-06,
      "loss": 0.0029,
      "step": 35750
    },
    {
      "epoch": 0.585126401047206,
      "grad_norm": 0.16001482253822655,
      "learning_rate": 4.388529000689599e-06,
      "loss": 0.0082,
      "step": 35760
    },
    {
      "epoch": 0.5852900269982819,
      "grad_norm": 0.1803767564390275,
      "learning_rate": 4.385694699256745e-06,
      "loss": 0.0037,
      "step": 35770
    },
    {
      "epoch": 0.5854536529493578,
      "grad_norm": 0.06806524486248523,
      "learning_rate": 4.382860598230128e-06,
      "loss": 0.0026,
      "step": 35780
    },
    {
      "epoch": 0.5856172789004336,
      "grad_norm": 0.0750015732753322,
      "learning_rate": 4.380026698534329e-06,
      "loss": 0.0029,
      "step": 35790
    },
    {
      "epoch": 0.5857809048515095,
      "grad_norm": 0.16755614572288596,
      "learning_rate": 4.377193001093858e-06,
      "loss": 0.0037,
      "step": 35800
    },
    {
      "epoch": 0.5859445308025852,
      "grad_norm": 0.10425387318683599,
      "learning_rate": 4.374359506833151e-06,
      "loss": 0.0026,
      "step": 35810
    },
    {
      "epoch": 0.5861081567536611,
      "grad_norm": 0.2508146305261222,
      "learning_rate": 4.371526216676594e-06,
      "loss": 0.0029,
      "step": 35820
    },
    {
      "epoch": 0.586271782704737,
      "grad_norm": 0.19149196682855552,
      "learning_rate": 4.3686931315484944e-06,
      "loss": 0.0016,
      "step": 35830
    },
    {
      "epoch": 0.5864354086558128,
      "grad_norm": 0.10075282915132526,
      "learning_rate": 4.3658602523730945e-06,
      "loss": 0.0023,
      "step": 35840
    },
    {
      "epoch": 0.5865990346068887,
      "grad_norm": 0.0323738266313214,
      "learning_rate": 4.363027580074574e-06,
      "loss": 0.0018,
      "step": 35850
    },
    {
      "epoch": 0.5867626605579644,
      "grad_norm": 0.030786385606715486,
      "learning_rate": 4.3601951155770416e-06,
      "loss": 0.005,
      "step": 35860
    },
    {
      "epoch": 0.5869262865090403,
      "grad_norm": 0.08847758291392628,
      "learning_rate": 4.357362859804537e-06,
      "loss": 0.0017,
      "step": 35870
    },
    {
      "epoch": 0.5870899124601162,
      "grad_norm": 0.06109116090231912,
      "learning_rate": 4.3545308136810325e-06,
      "loss": 0.0028,
      "step": 35880
    },
    {
      "epoch": 0.587253538411192,
      "grad_norm": 0.24084798163202856,
      "learning_rate": 4.3516989781304385e-06,
      "loss": 0.0035,
      "step": 35890
    },
    {
      "epoch": 0.5874171643622679,
      "grad_norm": 0.026981477878423255,
      "learning_rate": 4.348867354076587e-06,
      "loss": 0.0024,
      "step": 35900
    },
    {
      "epoch": 0.5875807903133436,
      "grad_norm": 0.13507826268212794,
      "learning_rate": 4.346035942443244e-06,
      "loss": 0.0021,
      "step": 35910
    },
    {
      "epoch": 0.5877444162644195,
      "grad_norm": 0.028145795378504878,
      "learning_rate": 4.343204744154111e-06,
      "loss": 0.003,
      "step": 35920
    },
    {
      "epoch": 0.5879080422154954,
      "grad_norm": 0.10039567198233354,
      "learning_rate": 4.340373760132816e-06,
      "loss": 0.0021,
      "step": 35930
    },
    {
      "epoch": 0.5880716681665712,
      "grad_norm": 0.18722086296414542,
      "learning_rate": 4.337542991302914e-06,
      "loss": 0.0027,
      "step": 35940
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.0838904958578955,
      "learning_rate": 4.334712438587898e-06,
      "loss": 0.0023,
      "step": 35950
    },
    {
      "epoch": 0.5883989200687229,
      "grad_norm": 0.1269674298723188,
      "learning_rate": 4.331882102911183e-06,
      "loss": 0.0022,
      "step": 35960
    },
    {
      "epoch": 0.5885625460197987,
      "grad_norm": 0.05790536181225277,
      "learning_rate": 4.329051985196116e-06,
      "loss": 0.0026,
      "step": 35970
    },
    {
      "epoch": 0.5887261719708746,
      "grad_norm": 0.13829500902628994,
      "learning_rate": 4.326222086365973e-06,
      "loss": 0.0032,
      "step": 35980
    },
    {
      "epoch": 0.5888897979219504,
      "grad_norm": 0.10219341941910584,
      "learning_rate": 4.323392407343959e-06,
      "loss": 0.0029,
      "step": 35990
    },
    {
      "epoch": 0.5890534238730263,
      "grad_norm": 0.09537531408532889,
      "learning_rate": 4.320562949053206e-06,
      "loss": 0.0022,
      "step": 36000
    },
    {
      "epoch": 0.5892170498241021,
      "grad_norm": 0.13231540796366972,
      "learning_rate": 4.317733712416776e-06,
      "loss": 0.003,
      "step": 36010
    },
    {
      "epoch": 0.5893806757751779,
      "grad_norm": 0.09675361578730217,
      "learning_rate": 4.314904698357657e-06,
      "loss": 0.0021,
      "step": 36020
    },
    {
      "epoch": 0.5895443017262538,
      "grad_norm": 0.17332796963974267,
      "learning_rate": 4.312075907798763e-06,
      "loss": 0.0032,
      "step": 36030
    },
    {
      "epoch": 0.5897079276773296,
      "grad_norm": 0.0934293702770415,
      "learning_rate": 4.3092473416629395e-06,
      "loss": 0.003,
      "step": 36040
    },
    {
      "epoch": 0.5898715536284055,
      "grad_norm": 0.0481067267652945,
      "learning_rate": 4.306419000872955e-06,
      "loss": 0.0035,
      "step": 36050
    },
    {
      "epoch": 0.5900351795794813,
      "grad_norm": 0.16582398036499876,
      "learning_rate": 4.303590886351505e-06,
      "loss": 0.0036,
      "step": 36060
    },
    {
      "epoch": 0.5901988055305571,
      "grad_norm": 0.15322998141371008,
      "learning_rate": 4.300762999021213e-06,
      "loss": 0.0026,
      "step": 36070
    },
    {
      "epoch": 0.590362431481633,
      "grad_norm": 0.10188242147606635,
      "learning_rate": 4.297935339804627e-06,
      "loss": 0.003,
      "step": 36080
    },
    {
      "epoch": 0.5905260574327088,
      "grad_norm": 0.13989161940417744,
      "learning_rate": 4.295107909624219e-06,
      "loss": 0.0017,
      "step": 36090
    },
    {
      "epoch": 0.5906896833837847,
      "grad_norm": 0.02070882726330858,
      "learning_rate": 4.292280709402391e-06,
      "loss": 0.0019,
      "step": 36100
    },
    {
      "epoch": 0.5908533093348605,
      "grad_norm": 0.2668412935401631,
      "learning_rate": 4.289453740061466e-06,
      "loss": 0.0033,
      "step": 36110
    },
    {
      "epoch": 0.5910169352859364,
      "grad_norm": 0.17066280042492676,
      "learning_rate": 4.286627002523691e-06,
      "loss": 0.0023,
      "step": 36120
    },
    {
      "epoch": 0.5911805612370122,
      "grad_norm": 0.1269775365479083,
      "learning_rate": 4.2838004977112425e-06,
      "loss": 0.002,
      "step": 36130
    },
    {
      "epoch": 0.591344187188088,
      "grad_norm": 0.2055963107680671,
      "learning_rate": 4.280974226546215e-06,
      "loss": 0.0035,
      "step": 36140
    },
    {
      "epoch": 0.5915078131391639,
      "grad_norm": 0.13044974868014672,
      "learning_rate": 4.278148189950629e-06,
      "loss": 0.0017,
      "step": 36150
    },
    {
      "epoch": 0.5916714390902397,
      "grad_norm": 0.0914530302327454,
      "learning_rate": 4.275322388846432e-06,
      "loss": 0.0028,
      "step": 36160
    },
    {
      "epoch": 0.5918350650413156,
      "grad_norm": 0.12257619100752548,
      "learning_rate": 4.2724968241554895e-06,
      "loss": 0.0029,
      "step": 36170
    },
    {
      "epoch": 0.5919986909923914,
      "grad_norm": 0.12296267011965546,
      "learning_rate": 4.269671496799592e-06,
      "loss": 0.0052,
      "step": 36180
    },
    {
      "epoch": 0.5921623169434672,
      "grad_norm": 0.2131324547467885,
      "learning_rate": 4.266846407700454e-06,
      "loss": 0.0033,
      "step": 36190
    },
    {
      "epoch": 0.5923259428945431,
      "grad_norm": 0.059827416722857116,
      "learning_rate": 4.264021557779709e-06,
      "loss": 0.0021,
      "step": 36200
    },
    {
      "epoch": 0.5924895688456189,
      "grad_norm": 0.20032620373688537,
      "learning_rate": 4.261196947958916e-06,
      "loss": 0.004,
      "step": 36210
    },
    {
      "epoch": 0.5926531947966948,
      "grad_norm": 0.31232870247444217,
      "learning_rate": 4.258372579159552e-06,
      "loss": 0.0039,
      "step": 36220
    },
    {
      "epoch": 0.5928168207477706,
      "grad_norm": 0.10748511261964865,
      "learning_rate": 4.2555484523030194e-06,
      "loss": 0.0023,
      "step": 36230
    },
    {
      "epoch": 0.5929804466988464,
      "grad_norm": 0.08386966760149098,
      "learning_rate": 4.2527245683106385e-06,
      "loss": 0.0052,
      "step": 36240
    },
    {
      "epoch": 0.5931440726499223,
      "grad_norm": 0.1438195469608127,
      "learning_rate": 4.249900928103652e-06,
      "loss": 0.0039,
      "step": 36250
    },
    {
      "epoch": 0.5933076986009981,
      "grad_norm": 0.08296149716315411,
      "learning_rate": 4.247077532603222e-06,
      "loss": 0.0021,
      "step": 36260
    },
    {
      "epoch": 0.593471324552074,
      "grad_norm": 0.16364012130807842,
      "learning_rate": 4.244254382730433e-06,
      "loss": 0.003,
      "step": 36270
    },
    {
      "epoch": 0.5936349505031498,
      "grad_norm": 0.18309654039073942,
      "learning_rate": 4.2414314794062845e-06,
      "loss": 0.0019,
      "step": 36280
    },
    {
      "epoch": 0.5937985764542256,
      "grad_norm": 0.0879103603041914,
      "learning_rate": 4.238608823551702e-06,
      "loss": 0.0018,
      "step": 36290
    },
    {
      "epoch": 0.5939622024053015,
      "grad_norm": 0.19810480649906168,
      "learning_rate": 4.2357864160875265e-06,
      "loss": 0.0025,
      "step": 36300
    },
    {
      "epoch": 0.5941258283563773,
      "grad_norm": 0.2783089017279225,
      "learning_rate": 4.2329642579345156e-06,
      "loss": 0.0044,
      "step": 36310
    },
    {
      "epoch": 0.5942894543074532,
      "grad_norm": 0.10276683428460374,
      "learning_rate": 4.230142350013353e-06,
      "loss": 0.0022,
      "step": 36320
    },
    {
      "epoch": 0.594453080258529,
      "grad_norm": 0.03892344742568337,
      "learning_rate": 4.227320693244633e-06,
      "loss": 0.0038,
      "step": 36330
    },
    {
      "epoch": 0.5946167062096048,
      "grad_norm": 0.21970769719073138,
      "learning_rate": 4.224499288548871e-06,
      "loss": 0.0041,
      "step": 36340
    },
    {
      "epoch": 0.5947803321606807,
      "grad_norm": 0.15486301289989388,
      "learning_rate": 4.221678136846503e-06,
      "loss": 0.0021,
      "step": 36350
    },
    {
      "epoch": 0.5949439581117565,
      "grad_norm": 0.31586739587149004,
      "learning_rate": 4.218857239057879e-06,
      "loss": 0.0025,
      "step": 36360
    },
    {
      "epoch": 0.5951075840628324,
      "grad_norm": 0.2612885259064541,
      "learning_rate": 4.216036596103265e-06,
      "loss": 0.0033,
      "step": 36370
    },
    {
      "epoch": 0.5952712100139083,
      "grad_norm": 0.1735913451421948,
      "learning_rate": 4.213216208902848e-06,
      "loss": 0.0023,
      "step": 36380
    },
    {
      "epoch": 0.595434835964984,
      "grad_norm": 0.23817832074767442,
      "learning_rate": 4.21039607837673e-06,
      "loss": 0.0023,
      "step": 36390
    },
    {
      "epoch": 0.5955984619160599,
      "grad_norm": 0.2035655389936723,
      "learning_rate": 4.207576205444925e-06,
      "loss": 0.0041,
      "step": 36400
    },
    {
      "epoch": 0.5957620878671357,
      "grad_norm": 0.35237767609710274,
      "learning_rate": 4.204756591027372e-06,
      "loss": 0.0034,
      "step": 36410
    },
    {
      "epoch": 0.5959257138182116,
      "grad_norm": 0.1554371718966407,
      "learning_rate": 4.201937236043916e-06,
      "loss": 0.0033,
      "step": 36420
    },
    {
      "epoch": 0.5960893397692875,
      "grad_norm": 0.03575689776280272,
      "learning_rate": 4.199118141414322e-06,
      "loss": 0.0036,
      "step": 36430
    },
    {
      "epoch": 0.5962529657203632,
      "grad_norm": 0.050627732821681336,
      "learning_rate": 4.196299308058273e-06,
      "loss": 0.0043,
      "step": 36440
    },
    {
      "epoch": 0.5964165916714391,
      "grad_norm": 0.06014265321114109,
      "learning_rate": 4.19348073689536e-06,
      "loss": 0.0038,
      "step": 36450
    },
    {
      "epoch": 0.5965802176225149,
      "grad_norm": 0.13389561860287294,
      "learning_rate": 4.190662428845093e-06,
      "loss": 0.0014,
      "step": 36460
    },
    {
      "epoch": 0.5967438435735908,
      "grad_norm": 0.07033387873776083,
      "learning_rate": 4.187844384826898e-06,
      "loss": 0.0015,
      "step": 36470
    },
    {
      "epoch": 0.5969074695246667,
      "grad_norm": 0.09645725433389785,
      "learning_rate": 4.185026605760109e-06,
      "loss": 0.0014,
      "step": 36480
    },
    {
      "epoch": 0.5970710954757424,
      "grad_norm": 0.1634236932678504,
      "learning_rate": 4.182209092563976e-06,
      "loss": 0.0029,
      "step": 36490
    },
    {
      "epoch": 0.5972347214268183,
      "grad_norm": 0.1833489140401746,
      "learning_rate": 4.1793918461576655e-06,
      "loss": 0.0017,
      "step": 36500
    },
    {
      "epoch": 0.5973983473778941,
      "grad_norm": 0.18765253298952278,
      "learning_rate": 4.176574867460253e-06,
      "loss": 0.0035,
      "step": 36510
    },
    {
      "epoch": 0.59756197332897,
      "grad_norm": 0.0994645623030929,
      "learning_rate": 4.173758157390727e-06,
      "loss": 0.0028,
      "step": 36520
    },
    {
      "epoch": 0.5977255992800458,
      "grad_norm": 0.25489759257490063,
      "learning_rate": 4.1709417168679905e-06,
      "loss": 0.0024,
      "step": 36530
    },
    {
      "epoch": 0.5978892252311216,
      "grad_norm": 0.04038336713327486,
      "learning_rate": 4.168125546810858e-06,
      "loss": 0.0032,
      "step": 36540
    },
    {
      "epoch": 0.5980528511821975,
      "grad_norm": 0.33041963522856893,
      "learning_rate": 4.165309648138054e-06,
      "loss": 0.0047,
      "step": 36550
    },
    {
      "epoch": 0.5982164771332733,
      "grad_norm": 0.09704289692522848,
      "learning_rate": 4.162494021768215e-06,
      "loss": 0.0033,
      "step": 36560
    },
    {
      "epoch": 0.5983801030843492,
      "grad_norm": 0.12082442249109245,
      "learning_rate": 4.159678668619891e-06,
      "loss": 0.0024,
      "step": 36570
    },
    {
      "epoch": 0.598543729035425,
      "grad_norm": 0.008410919028756136,
      "learning_rate": 4.156863589611541e-06,
      "loss": 0.0034,
      "step": 36580
    },
    {
      "epoch": 0.5987073549865009,
      "grad_norm": 0.1445664055030885,
      "learning_rate": 4.154048785661533e-06,
      "loss": 0.0019,
      "step": 36590
    },
    {
      "epoch": 0.5988709809375767,
      "grad_norm": 0.05458038788310649,
      "learning_rate": 4.151234257688149e-06,
      "loss": 0.004,
      "step": 36600
    },
    {
      "epoch": 0.5990346068886525,
      "grad_norm": 0.17373473755129382,
      "learning_rate": 4.148420006609581e-06,
      "loss": 0.0033,
      "step": 36610
    },
    {
      "epoch": 0.5991982328397284,
      "grad_norm": 0.09752425065598318,
      "learning_rate": 4.145606033343921e-06,
      "loss": 0.0021,
      "step": 36620
    },
    {
      "epoch": 0.5993618587908042,
      "grad_norm": 0.05326524387431546,
      "learning_rate": 4.142792338809185e-06,
      "loss": 0.0021,
      "step": 36630
    },
    {
      "epoch": 0.59952548474188,
      "grad_norm": 0.2774742155489998,
      "learning_rate": 4.1399789239232905e-06,
      "loss": 0.0034,
      "step": 36640
    },
    {
      "epoch": 0.5996891106929559,
      "grad_norm": 0.3661841911097322,
      "learning_rate": 4.137165789604059e-06,
      "loss": 0.0022,
      "step": 36650
    },
    {
      "epoch": 0.5998527366440317,
      "grad_norm": 0.3337166481242711,
      "learning_rate": 4.134352936769233e-06,
      "loss": 0.0028,
      "step": 36660
    },
    {
      "epoch": 0.6000163625951076,
      "grad_norm": 0.10019584515669817,
      "learning_rate": 4.131540366336453e-06,
      "loss": 0.0038,
      "step": 36670
    },
    {
      "epoch": 0.6001799885461834,
      "grad_norm": 0.23534840914400726,
      "learning_rate": 4.128728079223267e-06,
      "loss": 0.003,
      "step": 36680
    },
    {
      "epoch": 0.6003436144972593,
      "grad_norm": 0.1622811439425585,
      "learning_rate": 4.125916076347139e-06,
      "loss": 0.0025,
      "step": 36690
    },
    {
      "epoch": 0.6005072404483351,
      "grad_norm": 0.1966225087609585,
      "learning_rate": 4.123104358625434e-06,
      "loss": 0.0033,
      "step": 36700
    },
    {
      "epoch": 0.6006708663994109,
      "grad_norm": 0.0897234155943217,
      "learning_rate": 4.120292926975421e-06,
      "loss": 0.0032,
      "step": 36710
    },
    {
      "epoch": 0.6008344923504868,
      "grad_norm": 0.10885653523189812,
      "learning_rate": 4.117481782314286e-06,
      "loss": 0.002,
      "step": 36720
    },
    {
      "epoch": 0.6009981183015626,
      "grad_norm": 0.041460265582368684,
      "learning_rate": 4.114670925559113e-06,
      "loss": 0.0046,
      "step": 36730
    },
    {
      "epoch": 0.6011617442526385,
      "grad_norm": 0.06796192765697653,
      "learning_rate": 4.11186035762689e-06,
      "loss": 0.0018,
      "step": 36740
    },
    {
      "epoch": 0.6013253702037143,
      "grad_norm": 0.14463025606061014,
      "learning_rate": 4.109050079434522e-06,
      "loss": 0.003,
      "step": 36750
    },
    {
      "epoch": 0.6014889961547901,
      "grad_norm": 0.1994863267544617,
      "learning_rate": 4.106240091898811e-06,
      "loss": 0.0033,
      "step": 36760
    },
    {
      "epoch": 0.601652622105866,
      "grad_norm": 0.1297294055624739,
      "learning_rate": 4.10343039593646e-06,
      "loss": 0.0026,
      "step": 36770
    },
    {
      "epoch": 0.6018162480569418,
      "grad_norm": 0.20390609796758918,
      "learning_rate": 4.100620992464089e-06,
      "loss": 0.0036,
      "step": 36780
    },
    {
      "epoch": 0.6019798740080177,
      "grad_norm": 0.18485992041494434,
      "learning_rate": 4.0978118823982136e-06,
      "loss": 0.0025,
      "step": 36790
    },
    {
      "epoch": 0.6021434999590936,
      "grad_norm": 0.1038961414462875,
      "learning_rate": 4.095003066655255e-06,
      "loss": 0.0017,
      "step": 36800
    },
    {
      "epoch": 0.6023071259101693,
      "grad_norm": 0.4356324242393016,
      "learning_rate": 4.0921945461515425e-06,
      "loss": 0.0023,
      "step": 36810
    },
    {
      "epoch": 0.6024707518612452,
      "grad_norm": 0.07308170975634545,
      "learning_rate": 4.0893863218033066e-06,
      "loss": 0.0023,
      "step": 36820
    },
    {
      "epoch": 0.602634377812321,
      "grad_norm": 0.22754433366114282,
      "learning_rate": 4.086578394526676e-06,
      "loss": 0.003,
      "step": 36830
    },
    {
      "epoch": 0.6027980037633969,
      "grad_norm": 0.08012394365231984,
      "learning_rate": 4.083770765237693e-06,
      "loss": 0.0055,
      "step": 36840
    },
    {
      "epoch": 0.6029616297144728,
      "grad_norm": 0.21745660210956258,
      "learning_rate": 4.080963434852295e-06,
      "loss": 0.0012,
      "step": 36850
    },
    {
      "epoch": 0.6031252556655485,
      "grad_norm": 0.3736937339095875,
      "learning_rate": 4.078156404286323e-06,
      "loss": 0.0048,
      "step": 36860
    },
    {
      "epoch": 0.6032888816166244,
      "grad_norm": 0.07613640999537188,
      "learning_rate": 4.075349674455519e-06,
      "loss": 0.0038,
      "step": 36870
    },
    {
      "epoch": 0.6034525075677002,
      "grad_norm": 0.08246448875083312,
      "learning_rate": 4.0725432462755355e-06,
      "loss": 0.004,
      "step": 36880
    },
    {
      "epoch": 0.6036161335187761,
      "grad_norm": 0.13389691250146835,
      "learning_rate": 4.069737120661914e-06,
      "loss": 0.0035,
      "step": 36890
    },
    {
      "epoch": 0.603779759469852,
      "grad_norm": 0.0678422066898779,
      "learning_rate": 4.066931298530104e-06,
      "loss": 0.0037,
      "step": 36900
    },
    {
      "epoch": 0.6039433854209277,
      "grad_norm": 0.15672036820565613,
      "learning_rate": 4.06412578079546e-06,
      "loss": 0.0037,
      "step": 36910
    },
    {
      "epoch": 0.6041070113720036,
      "grad_norm": 0.20990148390882177,
      "learning_rate": 4.061320568373228e-06,
      "loss": 0.003,
      "step": 36920
    },
    {
      "epoch": 0.6042706373230794,
      "grad_norm": 0.2462615075851671,
      "learning_rate": 4.0585156621785576e-06,
      "loss": 0.0033,
      "step": 36930
    },
    {
      "epoch": 0.6044342632741553,
      "grad_norm": 0.20828173096515765,
      "learning_rate": 4.055711063126506e-06,
      "loss": 0.0035,
      "step": 36940
    },
    {
      "epoch": 0.6045978892252312,
      "grad_norm": 0.10782458570355521,
      "learning_rate": 4.0529067721320204e-06,
      "loss": 0.0031,
      "step": 36950
    },
    {
      "epoch": 0.6047615151763069,
      "grad_norm": 0.12215624783187451,
      "learning_rate": 4.050102790109948e-06,
      "loss": 0.0017,
      "step": 36960
    },
    {
      "epoch": 0.6049251411273828,
      "grad_norm": 0.1804569574945797,
      "learning_rate": 4.047299117975045e-06,
      "loss": 0.0044,
      "step": 36970
    },
    {
      "epoch": 0.6050887670784586,
      "grad_norm": 0.10299280275963474,
      "learning_rate": 4.044495756641956e-06,
      "loss": 0.0041,
      "step": 36980
    },
    {
      "epoch": 0.6052523930295345,
      "grad_norm": 0.2941023808095268,
      "learning_rate": 4.041692707025227e-06,
      "loss": 0.002,
      "step": 36990
    },
    {
      "epoch": 0.6054160189806104,
      "grad_norm": 0.16487873733420882,
      "learning_rate": 4.038889970039307e-06,
      "loss": 0.0027,
      "step": 37000
    },
    {
      "epoch": 0.6055796449316861,
      "grad_norm": 0.2517207590776455,
      "learning_rate": 4.036087546598536e-06,
      "loss": 0.003,
      "step": 37010
    },
    {
      "epoch": 0.605743270882762,
      "grad_norm": 0.07944757172323902,
      "learning_rate": 4.0332854376171546e-06,
      "loss": 0.0029,
      "step": 37020
    },
    {
      "epoch": 0.6059068968338378,
      "grad_norm": 0.1917589337769471,
      "learning_rate": 4.0304836440093065e-06,
      "loss": 0.0024,
      "step": 37030
    },
    {
      "epoch": 0.6060705227849137,
      "grad_norm": 0.1907877079757886,
      "learning_rate": 4.0276821666890235e-06,
      "loss": 0.002,
      "step": 37040
    },
    {
      "epoch": 0.6062341487359896,
      "grad_norm": 0.062424312543617085,
      "learning_rate": 4.0248810065702346e-06,
      "loss": 0.0079,
      "step": 37050
    },
    {
      "epoch": 0.6063977746870653,
      "grad_norm": 0.4235486738652975,
      "learning_rate": 4.022080164566777e-06,
      "loss": 0.0045,
      "step": 37060
    },
    {
      "epoch": 0.6065614006381412,
      "grad_norm": 0.12975289348897504,
      "learning_rate": 4.019279641592371e-06,
      "loss": 0.0029,
      "step": 37070
    },
    {
      "epoch": 0.606725026589217,
      "grad_norm": 0.2539890381047292,
      "learning_rate": 4.016479438560634e-06,
      "loss": 0.0031,
      "step": 37080
    },
    {
      "epoch": 0.6068886525402929,
      "grad_norm": 0.22943395179923234,
      "learning_rate": 4.013679556385092e-06,
      "loss": 0.0016,
      "step": 37090
    },
    {
      "epoch": 0.6070522784913688,
      "grad_norm": 0.08396111337592173,
      "learning_rate": 4.01087999597915e-06,
      "loss": 0.0021,
      "step": 37100
    },
    {
      "epoch": 0.6072159044424446,
      "grad_norm": 0.1399627198997756,
      "learning_rate": 4.0080807582561145e-06,
      "loss": 0.0037,
      "step": 37110
    },
    {
      "epoch": 0.6073795303935204,
      "grad_norm": 0.15299468351126108,
      "learning_rate": 4.005281844129194e-06,
      "loss": 0.0023,
      "step": 37120
    },
    {
      "epoch": 0.6075431563445962,
      "grad_norm": 0.15501451555039059,
      "learning_rate": 4.002483254511479e-06,
      "loss": 0.0023,
      "step": 37130
    },
    {
      "epoch": 0.6077067822956721,
      "grad_norm": 0.17285212226404453,
      "learning_rate": 3.9996849903159585e-06,
      "loss": 0.0021,
      "step": 37140
    },
    {
      "epoch": 0.607870408246748,
      "grad_norm": 0.09784911200727456,
      "learning_rate": 3.996887052455524e-06,
      "loss": 0.0045,
      "step": 37150
    },
    {
      "epoch": 0.6080340341978238,
      "grad_norm": 0.10470677315630754,
      "learning_rate": 3.9940894418429474e-06,
      "loss": 0.0023,
      "step": 37160
    },
    {
      "epoch": 0.6081976601488996,
      "grad_norm": 0.23220666798845344,
      "learning_rate": 3.991292159390898e-06,
      "loss": 0.0022,
      "step": 37170
    },
    {
      "epoch": 0.6083612860999754,
      "grad_norm": 0.5422134926419073,
      "learning_rate": 3.988495206011947e-06,
      "loss": 0.0044,
      "step": 37180
    },
    {
      "epoch": 0.6085249120510513,
      "grad_norm": 0.0967235690257041,
      "learning_rate": 3.985698582618547e-06,
      "loss": 0.0028,
      "step": 37190
    },
    {
      "epoch": 0.6086885380021272,
      "grad_norm": 0.20590916766387377,
      "learning_rate": 3.982902290123044e-06,
      "loss": 0.0027,
      "step": 37200
    },
    {
      "epoch": 0.608852163953203,
      "grad_norm": 0.06860356303644909,
      "learning_rate": 3.980106329437683e-06,
      "loss": 0.0032,
      "step": 37210
    },
    {
      "epoch": 0.6090157899042788,
      "grad_norm": 0.4353390584756546,
      "learning_rate": 3.977310701474596e-06,
      "loss": 0.0078,
      "step": 37220
    },
    {
      "epoch": 0.6091794158553546,
      "grad_norm": 0.22871093474962506,
      "learning_rate": 3.974515407145806e-06,
      "loss": 0.0022,
      "step": 37230
    },
    {
      "epoch": 0.6093430418064305,
      "grad_norm": 0.15099576724593036,
      "learning_rate": 3.971720447363228e-06,
      "loss": 0.0023,
      "step": 37240
    },
    {
      "epoch": 0.6095066677575064,
      "grad_norm": 0.1556828916147556,
      "learning_rate": 3.968925823038671e-06,
      "loss": 0.0023,
      "step": 37250
    },
    {
      "epoch": 0.6096702937085822,
      "grad_norm": 0.07833375696779173,
      "learning_rate": 3.966131535083829e-06,
      "loss": 0.002,
      "step": 37260
    },
    {
      "epoch": 0.609833919659658,
      "grad_norm": 0.39285409207073324,
      "learning_rate": 3.96333758441029e-06,
      "loss": 0.0045,
      "step": 37270
    },
    {
      "epoch": 0.6099975456107338,
      "grad_norm": 0.10784014870158588,
      "learning_rate": 3.9605439719295305e-06,
      "loss": 0.002,
      "step": 37280
    },
    {
      "epoch": 0.6101611715618097,
      "grad_norm": 0.15094745720172337,
      "learning_rate": 3.957750698552919e-06,
      "loss": 0.0037,
      "step": 37290
    },
    {
      "epoch": 0.6103247975128856,
      "grad_norm": 0.08645059213020569,
      "learning_rate": 3.954957765191708e-06,
      "loss": 0.0023,
      "step": 37300
    },
    {
      "epoch": 0.6104884234639614,
      "grad_norm": 0.4757451267718464,
      "learning_rate": 3.952165172757048e-06,
      "loss": 0.0025,
      "step": 37310
    },
    {
      "epoch": 0.6106520494150373,
      "grad_norm": 0.1824344222585368,
      "learning_rate": 3.949372922159968e-06,
      "loss": 0.0026,
      "step": 37320
    },
    {
      "epoch": 0.610815675366113,
      "grad_norm": 0.18145061440349983,
      "learning_rate": 3.946581014311392e-06,
      "loss": 0.002,
      "step": 37330
    },
    {
      "epoch": 0.6109793013171889,
      "grad_norm": 0.21078370551271883,
      "learning_rate": 3.94378945012213e-06,
      "loss": 0.0042,
      "step": 37340
    },
    {
      "epoch": 0.6111429272682648,
      "grad_norm": 0.09180734517427364,
      "learning_rate": 3.940998230502882e-06,
      "loss": 0.0028,
      "step": 37350
    },
    {
      "epoch": 0.6113065532193406,
      "grad_norm": 0.25692979150780904,
      "learning_rate": 3.9382073563642325e-06,
      "loss": 0.0035,
      "step": 37360
    },
    {
      "epoch": 0.6114701791704165,
      "grad_norm": 0.08201816363954675,
      "learning_rate": 3.935416828616656e-06,
      "loss": 0.0027,
      "step": 37370
    },
    {
      "epoch": 0.6116338051214922,
      "grad_norm": 0.23271137943738318,
      "learning_rate": 3.932626648170513e-06,
      "loss": 0.0035,
      "step": 37380
    },
    {
      "epoch": 0.6117974310725681,
      "grad_norm": 0.10211620419731243,
      "learning_rate": 3.929836815936048e-06,
      "loss": 0.0019,
      "step": 37390
    },
    {
      "epoch": 0.6119610570236439,
      "grad_norm": 0.15475262246981503,
      "learning_rate": 3.927047332823397e-06,
      "loss": 0.0013,
      "step": 37400
    },
    {
      "epoch": 0.6121246829747198,
      "grad_norm": 0.20688028125012914,
      "learning_rate": 3.92425819974258e-06,
      "loss": 0.0026,
      "step": 37410
    },
    {
      "epoch": 0.6122883089257957,
      "grad_norm": 0.077587263136665,
      "learning_rate": 3.9214694176035e-06,
      "loss": 0.0018,
      "step": 37420
    },
    {
      "epoch": 0.6124519348768714,
      "grad_norm": 0.09263516544069764,
      "learning_rate": 3.918680987315949e-06,
      "loss": 0.0036,
      "step": 37430
    },
    {
      "epoch": 0.6126155608279473,
      "grad_norm": 0.27478607281761475,
      "learning_rate": 3.915892909789604e-06,
      "loss": 0.0023,
      "step": 37440
    },
    {
      "epoch": 0.6127791867790231,
      "grad_norm": 0.10399753613880765,
      "learning_rate": 3.913105185934023e-06,
      "loss": 0.0016,
      "step": 37450
    },
    {
      "epoch": 0.612942812730099,
      "grad_norm": 0.1632163488952384,
      "learning_rate": 3.910317816658655e-06,
      "loss": 0.004,
      "step": 37460
    },
    {
      "epoch": 0.6131064386811749,
      "grad_norm": 0.166577690581626,
      "learning_rate": 3.907530802872829e-06,
      "loss": 0.0024,
      "step": 37470
    },
    {
      "epoch": 0.6132700646322506,
      "grad_norm": 0.17666681093128778,
      "learning_rate": 3.904744145485758e-06,
      "loss": 0.0024,
      "step": 37480
    },
    {
      "epoch": 0.6134336905833265,
      "grad_norm": 0.27118543042674303,
      "learning_rate": 3.90195784540654e-06,
      "loss": 0.0043,
      "step": 37490
    },
    {
      "epoch": 0.6135973165344023,
      "grad_norm": 0.11806123229635974,
      "learning_rate": 3.899171903544157e-06,
      "loss": 0.0036,
      "step": 37500
    },
    {
      "epoch": 0.6137609424854782,
      "grad_norm": 0.134944003566278,
      "learning_rate": 3.896386320807472e-06,
      "loss": 0.0036,
      "step": 37510
    },
    {
      "epoch": 0.6139245684365541,
      "grad_norm": 0.34464866186388027,
      "learning_rate": 3.893601098105233e-06,
      "loss": 0.0045,
      "step": 37520
    },
    {
      "epoch": 0.6140881943876298,
      "grad_norm": 0.15278754145298523,
      "learning_rate": 3.89081623634607e-06,
      "loss": 0.0017,
      "step": 37530
    },
    {
      "epoch": 0.6142518203387057,
      "grad_norm": 0.018011850841128607,
      "learning_rate": 3.8880317364384935e-06,
      "loss": 0.0041,
      "step": 37540
    },
    {
      "epoch": 0.6144154462897815,
      "grad_norm": 0.1805694074917229,
      "learning_rate": 3.8852475992908974e-06,
      "loss": 0.0019,
      "step": 37550
    },
    {
      "epoch": 0.6145790722408574,
      "grad_norm": 0.011590662245095196,
      "learning_rate": 3.882463825811558e-06,
      "loss": 0.0016,
      "step": 37560
    },
    {
      "epoch": 0.6147426981919333,
      "grad_norm": 0.4635811278447878,
      "learning_rate": 3.879680416908634e-06,
      "loss": 0.0031,
      "step": 37570
    },
    {
      "epoch": 0.614906324143009,
      "grad_norm": 0.1743790650992657,
      "learning_rate": 3.876897373490159e-06,
      "loss": 0.0037,
      "step": 37580
    },
    {
      "epoch": 0.6150699500940849,
      "grad_norm": 0.22982782593495038,
      "learning_rate": 3.874114696464056e-06,
      "loss": 0.0041,
      "step": 37590
    },
    {
      "epoch": 0.6152335760451607,
      "grad_norm": 0.08763199988950265,
      "learning_rate": 3.871332386738122e-06,
      "loss": 0.0068,
      "step": 37600
    },
    {
      "epoch": 0.6153972019962366,
      "grad_norm": 0.24790797675490406,
      "learning_rate": 3.868550445220036e-06,
      "loss": 0.0019,
      "step": 37610
    },
    {
      "epoch": 0.6155608279473125,
      "grad_norm": 0.16530750972333086,
      "learning_rate": 3.865768872817358e-06,
      "loss": 0.0029,
      "step": 37620
    },
    {
      "epoch": 0.6157244538983883,
      "grad_norm": 0.08996404129814316,
      "learning_rate": 3.862987670437528e-06,
      "loss": 0.0021,
      "step": 37630
    },
    {
      "epoch": 0.6158880798494641,
      "grad_norm": 0.33987167523188566,
      "learning_rate": 3.86020683898786e-06,
      "loss": 0.0037,
      "step": 37640
    },
    {
      "epoch": 0.6160517058005399,
      "grad_norm": 0.19902976122416022,
      "learning_rate": 3.857426379375556e-06,
      "loss": 0.0033,
      "step": 37650
    },
    {
      "epoch": 0.6162153317516158,
      "grad_norm": 0.3512830031517978,
      "learning_rate": 3.85464629250769e-06,
      "loss": 0.0029,
      "step": 37660
    },
    {
      "epoch": 0.6163789577026917,
      "grad_norm": 0.1077441435232946,
      "learning_rate": 3.851866579291214e-06,
      "loss": 0.0029,
      "step": 37670
    },
    {
      "epoch": 0.6165425836537675,
      "grad_norm": 0.1277194500869837,
      "learning_rate": 3.849087240632963e-06,
      "loss": 0.0031,
      "step": 37680
    },
    {
      "epoch": 0.6167062096048433,
      "grad_norm": 0.20216431905643192,
      "learning_rate": 3.846308277439647e-06,
      "loss": 0.0021,
      "step": 37690
    },
    {
      "epoch": 0.6168698355559191,
      "grad_norm": 0.13412720432296538,
      "learning_rate": 3.84352969061785e-06,
      "loss": 0.0019,
      "step": 37700
    },
    {
      "epoch": 0.617033461506995,
      "grad_norm": 0.11020329427162362,
      "learning_rate": 3.840751481074042e-06,
      "loss": 0.0034,
      "step": 37710
    },
    {
      "epoch": 0.6171970874580709,
      "grad_norm": 0.07540799013254593,
      "learning_rate": 3.837973649714562e-06,
      "loss": 0.0031,
      "step": 37720
    },
    {
      "epoch": 0.6173607134091467,
      "grad_norm": 0.24201166602564259,
      "learning_rate": 3.835196197445627e-06,
      "loss": 0.0037,
      "step": 37730
    },
    {
      "epoch": 0.6175243393602226,
      "grad_norm": 0.161167806069703,
      "learning_rate": 3.832419125173336e-06,
      "loss": 0.0044,
      "step": 37740
    },
    {
      "epoch": 0.6176879653112983,
      "grad_norm": 0.43729911749337025,
      "learning_rate": 3.829642433803656e-06,
      "loss": 0.0029,
      "step": 37750
    },
    {
      "epoch": 0.6178515912623742,
      "grad_norm": 0.0931949460377228,
      "learning_rate": 3.826866124242433e-06,
      "loss": 0.0023,
      "step": 37760
    },
    {
      "epoch": 0.6180152172134501,
      "grad_norm": 0.2932474484248866,
      "learning_rate": 3.824090197395393e-06,
      "loss": 0.0044,
      "step": 37770
    },
    {
      "epoch": 0.6181788431645259,
      "grad_norm": 0.21375748714345,
      "learning_rate": 3.821314654168129e-06,
      "loss": 0.0036,
      "step": 37780
    },
    {
      "epoch": 0.6183424691156018,
      "grad_norm": 0.06162195006286049,
      "learning_rate": 3.8185394954661135e-06,
      "loss": 0.0022,
      "step": 37790
    },
    {
      "epoch": 0.6185060950666775,
      "grad_norm": 0.009047925373010546,
      "learning_rate": 3.8157647221946944e-06,
      "loss": 0.003,
      "step": 37800
    },
    {
      "epoch": 0.6186697210177534,
      "grad_norm": 0.022962267625751605,
      "learning_rate": 3.812990335259093e-06,
      "loss": 0.0027,
      "step": 37810
    },
    {
      "epoch": 0.6188333469688293,
      "grad_norm": 0.2811568043788382,
      "learning_rate": 3.8102163355644014e-06,
      "loss": 0.0044,
      "step": 37820
    },
    {
      "epoch": 0.6189969729199051,
      "grad_norm": 0.5305987868241939,
      "learning_rate": 3.8074427240155898e-06,
      "loss": 0.003,
      "step": 37830
    },
    {
      "epoch": 0.619160598870981,
      "grad_norm": 0.6526296697879033,
      "learning_rate": 3.8046695015175e-06,
      "loss": 0.0041,
      "step": 37840
    },
    {
      "epoch": 0.6193242248220567,
      "grad_norm": 0.11485414593826405,
      "learning_rate": 3.801896668974847e-06,
      "loss": 0.004,
      "step": 37850
    },
    {
      "epoch": 0.6194878507731326,
      "grad_norm": 0.4085473348055722,
      "learning_rate": 3.7991242272922146e-06,
      "loss": 0.0038,
      "step": 37860
    },
    {
      "epoch": 0.6196514767242085,
      "grad_norm": 0.15413543882918404,
      "learning_rate": 3.7963521773740674e-06,
      "loss": 0.0027,
      "step": 37870
    },
    {
      "epoch": 0.6198151026752843,
      "grad_norm": 0.10574787854796569,
      "learning_rate": 3.793580520124738e-06,
      "loss": 0.0017,
      "step": 37880
    },
    {
      "epoch": 0.6199787286263602,
      "grad_norm": 0.28003193496730594,
      "learning_rate": 3.790809256448425e-06,
      "loss": 0.0039,
      "step": 37890
    },
    {
      "epoch": 0.6201423545774359,
      "grad_norm": 0.4820275381089115,
      "learning_rate": 3.7880383872492106e-06,
      "loss": 0.0024,
      "step": 37900
    },
    {
      "epoch": 0.6203059805285118,
      "grad_norm": 0.14164787902785522,
      "learning_rate": 3.7852679134310394e-06,
      "loss": 0.0036,
      "step": 37910
    },
    {
      "epoch": 0.6204696064795877,
      "grad_norm": 0.2926391412970169,
      "learning_rate": 3.7824978358977267e-06,
      "loss": 0.0025,
      "step": 37920
    },
    {
      "epoch": 0.6206332324306635,
      "grad_norm": 0.2222956132011905,
      "learning_rate": 3.779728155552966e-06,
      "loss": 0.003,
      "step": 37930
    },
    {
      "epoch": 0.6207968583817394,
      "grad_norm": 0.21625047244474455,
      "learning_rate": 3.776958873300315e-06,
      "loss": 0.0064,
      "step": 37940
    },
    {
      "epoch": 0.6209604843328151,
      "grad_norm": 0.08386944922895864,
      "learning_rate": 3.774189990043199e-06,
      "loss": 0.0025,
      "step": 37950
    },
    {
      "epoch": 0.621124110283891,
      "grad_norm": 0.18876617995396233,
      "learning_rate": 3.7714215066849243e-06,
      "loss": 0.0034,
      "step": 37960
    },
    {
      "epoch": 0.6212877362349669,
      "grad_norm": 0.361279451699795,
      "learning_rate": 3.768653424128654e-06,
      "loss": 0.0019,
      "step": 37970
    },
    {
      "epoch": 0.6214513621860427,
      "grad_norm": 0.04661353442884274,
      "learning_rate": 3.765885743277427e-06,
      "loss": 0.002,
      "step": 37980
    },
    {
      "epoch": 0.6216149881371186,
      "grad_norm": 0.14946435935451746,
      "learning_rate": 3.7631184650341545e-06,
      "loss": 0.0031,
      "step": 37990
    },
    {
      "epoch": 0.6217786140881943,
      "grad_norm": 0.15677448053994492,
      "learning_rate": 3.7603515903016077e-06,
      "loss": 0.0022,
      "step": 38000
    },
    {
      "epoch": 0.6219422400392702,
      "grad_norm": 0.14970270745471875,
      "learning_rate": 3.7575851199824288e-06,
      "loss": 0.0025,
      "step": 38010
    },
    {
      "epoch": 0.6221058659903461,
      "grad_norm": 0.10205677266152648,
      "learning_rate": 3.7548190549791365e-06,
      "loss": 0.0017,
      "step": 38020
    },
    {
      "epoch": 0.6222694919414219,
      "grad_norm": 0.08852553786046366,
      "learning_rate": 3.7520533961941056e-06,
      "loss": 0.0027,
      "step": 38030
    },
    {
      "epoch": 0.6224331178924978,
      "grad_norm": 0.12071588883982835,
      "learning_rate": 3.7492881445295824e-06,
      "loss": 0.0032,
      "step": 38040
    },
    {
      "epoch": 0.6225967438435736,
      "grad_norm": 0.09490543561382184,
      "learning_rate": 3.7465233008876874e-06,
      "loss": 0.0033,
      "step": 38050
    },
    {
      "epoch": 0.6227603697946494,
      "grad_norm": 0.07237853015358595,
      "learning_rate": 3.743758866170396e-06,
      "loss": 0.0019,
      "step": 38060
    },
    {
      "epoch": 0.6229239957457253,
      "grad_norm": 0.40755676204257185,
      "learning_rate": 3.740994841279557e-06,
      "loss": 0.0027,
      "step": 38070
    },
    {
      "epoch": 0.6230876216968011,
      "grad_norm": 0.16112462438342146,
      "learning_rate": 3.7382312271168884e-06,
      "loss": 0.0025,
      "step": 38080
    },
    {
      "epoch": 0.623251247647877,
      "grad_norm": 0.044961857739818256,
      "learning_rate": 3.735468024583967e-06,
      "loss": 0.0033,
      "step": 38090
    },
    {
      "epoch": 0.6234148735989528,
      "grad_norm": 0.12512456061487884,
      "learning_rate": 3.7327052345822377e-06,
      "loss": 0.0023,
      "step": 38100
    },
    {
      "epoch": 0.6235784995500286,
      "grad_norm": 0.3653751614848998,
      "learning_rate": 3.729942858013017e-06,
      "loss": 0.0056,
      "step": 38110
    },
    {
      "epoch": 0.6237421255011045,
      "grad_norm": 0.0888992719732976,
      "learning_rate": 3.7271808957774775e-06,
      "loss": 0.0018,
      "step": 38120
    },
    {
      "epoch": 0.6239057514521803,
      "grad_norm": 0.1705172512795876,
      "learning_rate": 3.724419348776659e-06,
      "loss": 0.0015,
      "step": 38130
    },
    {
      "epoch": 0.6240693774032562,
      "grad_norm": 0.01578039865196833,
      "learning_rate": 3.7216582179114725e-06,
      "loss": 0.0031,
      "step": 38140
    },
    {
      "epoch": 0.624233003354332,
      "grad_norm": 0.25284340632351615,
      "learning_rate": 3.7188975040826843e-06,
      "loss": 0.0038,
      "step": 38150
    },
    {
      "epoch": 0.6243966293054078,
      "grad_norm": 0.3067106693432949,
      "learning_rate": 3.7161372081909273e-06,
      "loss": 0.0044,
      "step": 38160
    },
    {
      "epoch": 0.6245602552564837,
      "grad_norm": 0.08615320886033166,
      "learning_rate": 3.7133773311367043e-06,
      "loss": 0.0024,
      "step": 38170
    },
    {
      "epoch": 0.6247238812075595,
      "grad_norm": 0.2344352622695516,
      "learning_rate": 3.710617873820373e-06,
      "loss": 0.0032,
      "step": 38180
    },
    {
      "epoch": 0.6248875071586354,
      "grad_norm": 0.010428522749680256,
      "learning_rate": 3.7078588371421577e-06,
      "loss": 0.0039,
      "step": 38190
    },
    {
      "epoch": 0.6250511331097112,
      "grad_norm": 0.13603768638902353,
      "learning_rate": 3.7051002220021448e-06,
      "loss": 0.002,
      "step": 38200
    },
    {
      "epoch": 0.625214759060787,
      "grad_norm": 0.17730493551653576,
      "learning_rate": 3.702342029300285e-06,
      "loss": 0.0024,
      "step": 38210
    },
    {
      "epoch": 0.6253783850118629,
      "grad_norm": 0.12409248306868036,
      "learning_rate": 3.6995842599363897e-06,
      "loss": 0.0024,
      "step": 38220
    },
    {
      "epoch": 0.6255420109629387,
      "grad_norm": 0.14838497015282662,
      "learning_rate": 3.696826914810131e-06,
      "loss": 0.0018,
      "step": 38230
    },
    {
      "epoch": 0.6257056369140146,
      "grad_norm": 0.05772436206324633,
      "learning_rate": 3.694069994821046e-06,
      "loss": 0.0019,
      "step": 38240
    },
    {
      "epoch": 0.6258692628650904,
      "grad_norm": 0.25101461980211764,
      "learning_rate": 3.6913135008685306e-06,
      "loss": 0.0023,
      "step": 38250
    },
    {
      "epoch": 0.6260328888161663,
      "grad_norm": 0.14784230056014358,
      "learning_rate": 3.6885574338518405e-06,
      "loss": 0.003,
      "step": 38260
    },
    {
      "epoch": 0.626196514767242,
      "grad_norm": 0.04744194891011138,
      "learning_rate": 3.6858017946700957e-06,
      "loss": 0.0022,
      "step": 38270
    },
    {
      "epoch": 0.6263601407183179,
      "grad_norm": 0.3992167168686326,
      "learning_rate": 3.6830465842222747e-06,
      "loss": 0.0034,
      "step": 38280
    },
    {
      "epoch": 0.6265237666693938,
      "grad_norm": 0.08340280859893437,
      "learning_rate": 3.6802918034072142e-06,
      "loss": 0.003,
      "step": 38290
    },
    {
      "epoch": 0.6266873926204696,
      "grad_norm": 0.14859698103047322,
      "learning_rate": 3.6775374531236153e-06,
      "loss": 0.002,
      "step": 38300
    },
    {
      "epoch": 0.6268510185715455,
      "grad_norm": 0.29810720883017816,
      "learning_rate": 3.6747835342700355e-06,
      "loss": 0.0025,
      "step": 38310
    },
    {
      "epoch": 0.6270146445226212,
      "grad_norm": 0.19357298751918414,
      "learning_rate": 3.67203004774489e-06,
      "loss": 0.0043,
      "step": 38320
    },
    {
      "epoch": 0.6271782704736971,
      "grad_norm": 0.13921365232679245,
      "learning_rate": 3.6692769944464575e-06,
      "loss": 0.0024,
      "step": 38330
    },
    {
      "epoch": 0.627341896424773,
      "grad_norm": 0.10467501382692902,
      "learning_rate": 3.666524375272872e-06,
      "loss": 0.0026,
      "step": 38340
    },
    {
      "epoch": 0.6275055223758488,
      "grad_norm": 0.0433689597337994,
      "learning_rate": 3.6637721911221265e-06,
      "loss": 0.0021,
      "step": 38350
    },
    {
      "epoch": 0.6276691483269247,
      "grad_norm": 0.017340680117201867,
      "learning_rate": 3.6610204428920735e-06,
      "loss": 0.0042,
      "step": 38360
    },
    {
      "epoch": 0.6278327742780004,
      "grad_norm": 0.03517518518574745,
      "learning_rate": 3.6582691314804215e-06,
      "loss": 0.0022,
      "step": 38370
    },
    {
      "epoch": 0.6279964002290763,
      "grad_norm": 0.3998061331245079,
      "learning_rate": 3.655518257784736e-06,
      "loss": 0.0025,
      "step": 38380
    },
    {
      "epoch": 0.6281600261801522,
      "grad_norm": 0.11417011733817845,
      "learning_rate": 3.652767822702443e-06,
      "loss": 0.0021,
      "step": 38390
    },
    {
      "epoch": 0.628323652131228,
      "grad_norm": 0.1384373064343689,
      "learning_rate": 3.6500178271308227e-06,
      "loss": 0.002,
      "step": 38400
    },
    {
      "epoch": 0.6284872780823039,
      "grad_norm": 0.08775581040639487,
      "learning_rate": 3.6472682719670118e-06,
      "loss": 0.0033,
      "step": 38410
    },
    {
      "epoch": 0.6286509040333796,
      "grad_norm": 0.10249673582007192,
      "learning_rate": 3.644519158108004e-06,
      "loss": 0.0039,
      "step": 38420
    },
    {
      "epoch": 0.6288145299844555,
      "grad_norm": 0.1935489375264137,
      "learning_rate": 3.641770486450651e-06,
      "loss": 0.003,
      "step": 38430
    },
    {
      "epoch": 0.6289781559355314,
      "grad_norm": 0.06760697358930276,
      "learning_rate": 3.639022257891655e-06,
      "loss": 0.0028,
      "step": 38440
    },
    {
      "epoch": 0.6291417818866072,
      "grad_norm": 0.10566765312791716,
      "learning_rate": 3.63627447332758e-06,
      "loss": 0.0035,
      "step": 38450
    },
    {
      "epoch": 0.6293054078376831,
      "grad_norm": 0.06728765716506759,
      "learning_rate": 3.6335271336548417e-06,
      "loss": 0.0024,
      "step": 38460
    },
    {
      "epoch": 0.6294690337887588,
      "grad_norm": 0.14316234630000466,
      "learning_rate": 3.6307802397697078e-06,
      "loss": 0.0029,
      "step": 38470
    },
    {
      "epoch": 0.6296326597398347,
      "grad_norm": 0.07815642999977915,
      "learning_rate": 3.6280337925683083e-06,
      "loss": 0.0029,
      "step": 38480
    },
    {
      "epoch": 0.6297962856909106,
      "grad_norm": 0.10653821487349018,
      "learning_rate": 3.625287792946621e-06,
      "loss": 0.002,
      "step": 38490
    },
    {
      "epoch": 0.6299599116419864,
      "grad_norm": 0.06881104670043096,
      "learning_rate": 3.6225422418004807e-06,
      "loss": 0.0022,
      "step": 38500
    },
    {
      "epoch": 0.6301235375930623,
      "grad_norm": 0.07557137509277463,
      "learning_rate": 3.619797140025572e-06,
      "loss": 0.0019,
      "step": 38510
    },
    {
      "epoch": 0.630287163544138,
      "grad_norm": 0.03591441450603031,
      "learning_rate": 3.617052488517439e-06,
      "loss": 0.0035,
      "step": 38520
    },
    {
      "epoch": 0.6304507894952139,
      "grad_norm": 0.25097639885870293,
      "learning_rate": 3.614308288171475e-06,
      "loss": 0.002,
      "step": 38530
    },
    {
      "epoch": 0.6306144154462898,
      "grad_norm": 0.1389971610613418,
      "learning_rate": 3.6115645398829248e-06,
      "loss": 0.0034,
      "step": 38540
    },
    {
      "epoch": 0.6307780413973656,
      "grad_norm": 0.17589646894936803,
      "learning_rate": 3.6088212445468907e-06,
      "loss": 0.0022,
      "step": 38550
    },
    {
      "epoch": 0.6309416673484415,
      "grad_norm": 0.08658727822755675,
      "learning_rate": 3.606078403058323e-06,
      "loss": 0.0038,
      "step": 38560
    },
    {
      "epoch": 0.6311052932995173,
      "grad_norm": 0.23045403396874473,
      "learning_rate": 3.6033360163120233e-06,
      "loss": 0.0024,
      "step": 38570
    },
    {
      "epoch": 0.6312689192505931,
      "grad_norm": 0.2830268893794633,
      "learning_rate": 3.6005940852026513e-06,
      "loss": 0.0055,
      "step": 38580
    },
    {
      "epoch": 0.631432545201669,
      "grad_norm": 0.24679752073941583,
      "learning_rate": 3.59785261062471e-06,
      "loss": 0.0041,
      "step": 38590
    },
    {
      "epoch": 0.6315961711527448,
      "grad_norm": 0.06928124901430129,
      "learning_rate": 3.595111593472557e-06,
      "loss": 0.0022,
      "step": 38600
    },
    {
      "epoch": 0.6317597971038207,
      "grad_norm": 0.06582509661013895,
      "learning_rate": 3.5923710346404017e-06,
      "loss": 0.002,
      "step": 38610
    },
    {
      "epoch": 0.6319234230548965,
      "grad_norm": 0.2349798041444064,
      "learning_rate": 3.5896309350223036e-06,
      "loss": 0.0021,
      "step": 38620
    },
    {
      "epoch": 0.6320870490059723,
      "grad_norm": 0.14511668992385693,
      "learning_rate": 3.5868912955121697e-06,
      "loss": 0.0028,
      "step": 38630
    },
    {
      "epoch": 0.6322506749570482,
      "grad_norm": 0.26415147374666026,
      "learning_rate": 3.5841521170037614e-06,
      "loss": 0.0028,
      "step": 38640
    },
    {
      "epoch": 0.632414300908124,
      "grad_norm": 0.15530303630261916,
      "learning_rate": 3.581413400390686e-06,
      "loss": 0.0036,
      "step": 38650
    },
    {
      "epoch": 0.6325779268591999,
      "grad_norm": 0.17375612971469037,
      "learning_rate": 3.5786751465664003e-06,
      "loss": 0.0024,
      "step": 38660
    },
    {
      "epoch": 0.6327415528102757,
      "grad_norm": 0.1385849198586867,
      "learning_rate": 3.5759373564242135e-06,
      "loss": 0.0019,
      "step": 38670
    },
    {
      "epoch": 0.6329051787613516,
      "grad_norm": 0.130304237375768,
      "learning_rate": 3.5732000308572793e-06,
      "loss": 0.0021,
      "step": 38680
    },
    {
      "epoch": 0.6330688047124274,
      "grad_norm": 0.20550082373084577,
      "learning_rate": 3.5704631707586023e-06,
      "loss": 0.0028,
      "step": 38690
    },
    {
      "epoch": 0.6332324306635032,
      "grad_norm": 0.08068544638794777,
      "learning_rate": 3.5677267770210354e-06,
      "loss": 0.0029,
      "step": 38700
    },
    {
      "epoch": 0.6333960566145791,
      "grad_norm": 0.0405596941365278,
      "learning_rate": 3.5649908505372787e-06,
      "loss": 0.0023,
      "step": 38710
    },
    {
      "epoch": 0.6335596825656549,
      "grad_norm": 0.0319238182833129,
      "learning_rate": 3.562255392199877e-06,
      "loss": 0.0025,
      "step": 38720
    },
    {
      "epoch": 0.6337233085167308,
      "grad_norm": 0.06638724826502856,
      "learning_rate": 3.55952040290123e-06,
      "loss": 0.0017,
      "step": 38730
    },
    {
      "epoch": 0.6338869344678066,
      "grad_norm": 0.055841150920700036,
      "learning_rate": 3.5567858835335762e-06,
      "loss": 0.0023,
      "step": 38740
    },
    {
      "epoch": 0.6340505604188824,
      "grad_norm": 0.2755851712666514,
      "learning_rate": 3.554051834989004e-06,
      "loss": 0.0024,
      "step": 38750
    },
    {
      "epoch": 0.6342141863699583,
      "grad_norm": 0.07089017127900447,
      "learning_rate": 3.5513182581594507e-06,
      "loss": 0.0019,
      "step": 38760
    },
    {
      "epoch": 0.6343778123210341,
      "grad_norm": 0.1665739743608836,
      "learning_rate": 3.5485851539366957e-06,
      "loss": 0.0029,
      "step": 38770
    },
    {
      "epoch": 0.63454143827211,
      "grad_norm": 0.33496896780129287,
      "learning_rate": 3.545852523212366e-06,
      "loss": 0.0034,
      "step": 38780
    },
    {
      "epoch": 0.6347050642231858,
      "grad_norm": 0.027694154292521325,
      "learning_rate": 3.5431203668779344e-06,
      "loss": 0.0017,
      "step": 38790
    },
    {
      "epoch": 0.6348686901742616,
      "grad_norm": 0.10418731321733239,
      "learning_rate": 3.5403886858247195e-06,
      "loss": 0.0033,
      "step": 38800
    },
    {
      "epoch": 0.6350323161253375,
      "grad_norm": 0.21186309953856952,
      "learning_rate": 3.537657480943882e-06,
      "loss": 0.0033,
      "step": 38810
    },
    {
      "epoch": 0.6351959420764133,
      "grad_norm": 0.15116136763187724,
      "learning_rate": 3.534926753126431e-06,
      "loss": 0.0035,
      "step": 38820
    },
    {
      "epoch": 0.6353595680274892,
      "grad_norm": 0.11003966879294831,
      "learning_rate": 3.532196503263217e-06,
      "loss": 0.0025,
      "step": 38830
    },
    {
      "epoch": 0.635523193978565,
      "grad_norm": 0.389729405112115,
      "learning_rate": 3.5294667322449376e-06,
      "loss": 0.0026,
      "step": 38840
    },
    {
      "epoch": 0.6356868199296408,
      "grad_norm": 0.06262217513342504,
      "learning_rate": 3.526737440962128e-06,
      "loss": 0.0029,
      "step": 38850
    },
    {
      "epoch": 0.6358504458807167,
      "grad_norm": 0.05695564873693938,
      "learning_rate": 3.524008630305175e-06,
      "loss": 0.0027,
      "step": 38860
    },
    {
      "epoch": 0.6360140718317925,
      "grad_norm": 0.031833906099472446,
      "learning_rate": 3.521280301164306e-06,
      "loss": 0.0026,
      "step": 38870
    },
    {
      "epoch": 0.6361776977828684,
      "grad_norm": 0.07692799187508191,
      "learning_rate": 3.518552454429585e-06,
      "loss": 0.0023,
      "step": 38880
    },
    {
      "epoch": 0.6363413237339443,
      "grad_norm": 0.28107378208070105,
      "learning_rate": 3.5158250909909274e-06,
      "loss": 0.0027,
      "step": 38890
    },
    {
      "epoch": 0.63650494968502,
      "grad_norm": 0.19079093013755902,
      "learning_rate": 3.5130982117380887e-06,
      "loss": 0.0024,
      "step": 38900
    },
    {
      "epoch": 0.6366685756360959,
      "grad_norm": 0.2284570473305802,
      "learning_rate": 3.510371817560659e-06,
      "loss": 0.0028,
      "step": 38910
    },
    {
      "epoch": 0.6368322015871717,
      "grad_norm": 0.41034425449583406,
      "learning_rate": 3.507645909348083e-06,
      "loss": 0.0026,
      "step": 38920
    },
    {
      "epoch": 0.6369958275382476,
      "grad_norm": 0.10690221900288284,
      "learning_rate": 3.504920487989638e-06,
      "loss": 0.0032,
      "step": 38930
    },
    {
      "epoch": 0.6371594534893235,
      "grad_norm": 0.32264827471237356,
      "learning_rate": 3.5021955543744414e-06,
      "loss": 0.0035,
      "step": 38940
    },
    {
      "epoch": 0.6373230794403992,
      "grad_norm": 0.1095029293673731,
      "learning_rate": 3.499471109391458e-06,
      "loss": 0.0029,
      "step": 38950
    },
    {
      "epoch": 0.6374867053914751,
      "grad_norm": 0.06859882506483665,
      "learning_rate": 3.4967471539294897e-06,
      "loss": 0.0027,
      "step": 38960
    },
    {
      "epoch": 0.6376503313425509,
      "grad_norm": 0.2360184927574292,
      "learning_rate": 3.494023688877175e-06,
      "loss": 0.0055,
      "step": 38970
    },
    {
      "epoch": 0.6378139572936268,
      "grad_norm": 0.15935277716643212,
      "learning_rate": 3.491300715123001e-06,
      "loss": 0.0026,
      "step": 38980
    },
    {
      "epoch": 0.6379775832447027,
      "grad_norm": 0.18765905096543423,
      "learning_rate": 3.4885782335552886e-06,
      "loss": 0.0029,
      "step": 38990
    },
    {
      "epoch": 0.6381412091957784,
      "grad_norm": 0.17056334573291837,
      "learning_rate": 3.485856245062196e-06,
      "loss": 0.0026,
      "step": 39000
    },
    {
      "epoch": 0.6383048351468543,
      "grad_norm": 0.12479206908974337,
      "learning_rate": 3.4831347505317277e-06,
      "loss": 0.0047,
      "step": 39010
    },
    {
      "epoch": 0.6384684610979301,
      "grad_norm": 0.05848187346061733,
      "learning_rate": 3.480413750851723e-06,
      "loss": 0.0025,
      "step": 39020
    },
    {
      "epoch": 0.638632087049006,
      "grad_norm": 0.061245253722046145,
      "learning_rate": 3.4776932469098553e-06,
      "loss": 0.0016,
      "step": 39030
    },
    {
      "epoch": 0.6387957130000819,
      "grad_norm": 0.2604992415635095,
      "learning_rate": 3.474973239593646e-06,
      "loss": 0.0039,
      "step": 39040
    },
    {
      "epoch": 0.6389593389511576,
      "grad_norm": 0.15942433710193663,
      "learning_rate": 3.472253729790449e-06,
      "loss": 0.0017,
      "step": 39050
    },
    {
      "epoch": 0.6391229649022335,
      "grad_norm": 0.22631456028260613,
      "learning_rate": 3.469534718387452e-06,
      "loss": 0.002,
      "step": 39060
    },
    {
      "epoch": 0.6392865908533093,
      "grad_norm": 0.15510403091064973,
      "learning_rate": 3.4668162062716893e-06,
      "loss": 0.002,
      "step": 39070
    },
    {
      "epoch": 0.6394502168043852,
      "grad_norm": 0.03259973575364838,
      "learning_rate": 3.4640981943300257e-06,
      "loss": 0.0018,
      "step": 39080
    },
    {
      "epoch": 0.6396138427554611,
      "grad_norm": 0.27997364757017107,
      "learning_rate": 3.4613806834491616e-06,
      "loss": 0.0032,
      "step": 39090
    },
    {
      "epoch": 0.6397774687065368,
      "grad_norm": 0.12568721517320217,
      "learning_rate": 3.458663674515641e-06,
      "loss": 0.0037,
      "step": 39100
    },
    {
      "epoch": 0.6399410946576127,
      "grad_norm": 0.21657601456856848,
      "learning_rate": 3.45594716841584e-06,
      "loss": 0.0031,
      "step": 39110
    },
    {
      "epoch": 0.6401047206086885,
      "grad_norm": 0.16483118280487852,
      "learning_rate": 3.4532311660359663e-06,
      "loss": 0.0014,
      "step": 39120
    },
    {
      "epoch": 0.6402683465597644,
      "grad_norm": 0.10833034929183712,
      "learning_rate": 3.4505156682620723e-06,
      "loss": 0.003,
      "step": 39130
    },
    {
      "epoch": 0.6404319725108402,
      "grad_norm": 0.053708298906984524,
      "learning_rate": 3.44780067598004e-06,
      "loss": 0.0026,
      "step": 39140
    },
    {
      "epoch": 0.640595598461916,
      "grad_norm": 0.08706665174363926,
      "learning_rate": 3.445086190075584e-06,
      "loss": 0.0022,
      "step": 39150
    },
    {
      "epoch": 0.6407592244129919,
      "grad_norm": 0.09663525873904123,
      "learning_rate": 3.4423722114342627e-06,
      "loss": 0.0022,
      "step": 39160
    },
    {
      "epoch": 0.6409228503640677,
      "grad_norm": 0.2617878128758688,
      "learning_rate": 3.4396587409414616e-06,
      "loss": 0.0026,
      "step": 39170
    },
    {
      "epoch": 0.6410864763151436,
      "grad_norm": 0.019585625635606414,
      "learning_rate": 3.436945779482401e-06,
      "loss": 0.0026,
      "step": 39180
    },
    {
      "epoch": 0.6412501022662194,
      "grad_norm": 0.482092772708202,
      "learning_rate": 3.4342333279421365e-06,
      "loss": 0.0023,
      "step": 39190
    },
    {
      "epoch": 0.6414137282172953,
      "grad_norm": 0.10950673329294372,
      "learning_rate": 3.4315213872055608e-06,
      "loss": 0.0025,
      "step": 39200
    },
    {
      "epoch": 0.6415773541683711,
      "grad_norm": 0.03607768986921398,
      "learning_rate": 3.428809958157394e-06,
      "loss": 0.0032,
      "step": 39210
    },
    {
      "epoch": 0.6417409801194469,
      "grad_norm": 0.12558146764801126,
      "learning_rate": 3.4260990416821905e-06,
      "loss": 0.0051,
      "step": 39220
    },
    {
      "epoch": 0.6419046060705228,
      "grad_norm": 0.06545152650964281,
      "learning_rate": 3.423388638664342e-06,
      "loss": 0.0046,
      "step": 39230
    },
    {
      "epoch": 0.6420682320215986,
      "grad_norm": 0.28751749668765947,
      "learning_rate": 3.420678749988069e-06,
      "loss": 0.0045,
      "step": 39240
    },
    {
      "epoch": 0.6422318579726745,
      "grad_norm": 0.09914673240841411,
      "learning_rate": 3.4179693765374217e-06,
      "loss": 0.004,
      "step": 39250
    },
    {
      "epoch": 0.6423954839237503,
      "grad_norm": 0.1624249882707871,
      "learning_rate": 3.415260519196289e-06,
      "loss": 0.0031,
      "step": 39260
    },
    {
      "epoch": 0.6425591098748261,
      "grad_norm": 0.06780680146887115,
      "learning_rate": 3.412552178848386e-06,
      "loss": 0.0016,
      "step": 39270
    },
    {
      "epoch": 0.642722735825902,
      "grad_norm": 0.08071851840400393,
      "learning_rate": 3.4098443563772603e-06,
      "loss": 0.0023,
      "step": 39280
    },
    {
      "epoch": 0.6428863617769778,
      "grad_norm": 0.0865581863579484,
      "learning_rate": 3.4071370526662927e-06,
      "loss": 0.0021,
      "step": 39290
    },
    {
      "epoch": 0.6430499877280537,
      "grad_norm": 0.1206850955827089,
      "learning_rate": 3.404430268598693e-06,
      "loss": 0.0025,
      "step": 39300
    },
    {
      "epoch": 0.6432136136791295,
      "grad_norm": 0.037373664251977204,
      "learning_rate": 3.401724005057499e-06,
      "loss": 0.0021,
      "step": 39310
    },
    {
      "epoch": 0.6433772396302053,
      "grad_norm": 0.2981050405267764,
      "learning_rate": 3.3990182629255846e-06,
      "loss": 0.0031,
      "step": 39320
    },
    {
      "epoch": 0.6435408655812812,
      "grad_norm": 0.2660153177096172,
      "learning_rate": 3.396313043085649e-06,
      "loss": 0.0043,
      "step": 39330
    },
    {
      "epoch": 0.643704491532357,
      "grad_norm": 0.11660320312791532,
      "learning_rate": 3.3936083464202215e-06,
      "loss": 0.0021,
      "step": 39340
    },
    {
      "epoch": 0.6438681174834329,
      "grad_norm": 0.09542051793649162,
      "learning_rate": 3.390904173811663e-06,
      "loss": 0.0025,
      "step": 39350
    },
    {
      "epoch": 0.6440317434345088,
      "grad_norm": 0.11962384436668982,
      "learning_rate": 3.3882005261421614e-06,
      "loss": 0.0014,
      "step": 39360
    },
    {
      "epoch": 0.6441953693855845,
      "grad_norm": 0.17334410337924983,
      "learning_rate": 3.3854974042937327e-06,
      "loss": 0.0027,
      "step": 39370
    },
    {
      "epoch": 0.6443589953366604,
      "grad_norm": 0.08640245323925387,
      "learning_rate": 3.3827948091482254e-06,
      "loss": 0.0027,
      "step": 39380
    },
    {
      "epoch": 0.6445226212877362,
      "grad_norm": 0.19022382429604776,
      "learning_rate": 3.3800927415873118e-06,
      "loss": 0.0034,
      "step": 39390
    },
    {
      "epoch": 0.6446862472388121,
      "grad_norm": 0.13862264139332806,
      "learning_rate": 3.3773912024924926e-06,
      "loss": 0.0015,
      "step": 39400
    },
    {
      "epoch": 0.644849873189888,
      "grad_norm": 0.2905314794592806,
      "learning_rate": 3.3746901927450985e-06,
      "loss": 0.0044,
      "step": 39410
    },
    {
      "epoch": 0.6450134991409637,
      "grad_norm": 0.0864932818181526,
      "learning_rate": 3.3719897132262865e-06,
      "loss": 0.0045,
      "step": 39420
    },
    {
      "epoch": 0.6451771250920396,
      "grad_norm": 0.10044096407664471,
      "learning_rate": 3.3692897648170385e-06,
      "loss": 0.0019,
      "step": 39430
    },
    {
      "epoch": 0.6453407510431154,
      "grad_norm": 0.33443881843169976,
      "learning_rate": 3.3665903483981667e-06,
      "loss": 0.0023,
      "step": 39440
    },
    {
      "epoch": 0.6455043769941913,
      "grad_norm": 0.17473955994237858,
      "learning_rate": 3.363891464850307e-06,
      "loss": 0.0023,
      "step": 39450
    },
    {
      "epoch": 0.6456680029452672,
      "grad_norm": 0.17463385678983154,
      "learning_rate": 3.3611931150539228e-06,
      "loss": 0.004,
      "step": 39460
    },
    {
      "epoch": 0.6458316288963429,
      "grad_norm": 0.21844348098830813,
      "learning_rate": 3.358495299889304e-06,
      "loss": 0.0063,
      "step": 39470
    },
    {
      "epoch": 0.6459952548474188,
      "grad_norm": 0.44237402085899674,
      "learning_rate": 3.355798020236563e-06,
      "loss": 0.0037,
      "step": 39480
    },
    {
      "epoch": 0.6461588807984946,
      "grad_norm": 0.05849516192366982,
      "learning_rate": 3.3531012769756398e-06,
      "loss": 0.002,
      "step": 39490
    },
    {
      "epoch": 0.6463225067495705,
      "grad_norm": 0.23793781468187503,
      "learning_rate": 3.350405070986302e-06,
      "loss": 0.0031,
      "step": 39500
    },
    {
      "epoch": 0.6464861327006464,
      "grad_norm": 0.26682048461196695,
      "learning_rate": 3.347709403148136e-06,
      "loss": 0.007,
      "step": 39510
    },
    {
      "epoch": 0.6466497586517221,
      "grad_norm": 0.16024914869492465,
      "learning_rate": 3.3450142743405577e-06,
      "loss": 0.0024,
      "step": 39520
    },
    {
      "epoch": 0.646813384602798,
      "grad_norm": 0.05431893310284545,
      "learning_rate": 3.342319685442803e-06,
      "loss": 0.0027,
      "step": 39530
    },
    {
      "epoch": 0.6469770105538738,
      "grad_norm": 0.07468792672795713,
      "learning_rate": 3.3396256373339363e-06,
      "loss": 0.0026,
      "step": 39540
    },
    {
      "epoch": 0.6471406365049497,
      "grad_norm": 0.19813218668722152,
      "learning_rate": 3.3369321308928427e-06,
      "loss": 0.0045,
      "step": 39550
    },
    {
      "epoch": 0.6473042624560256,
      "grad_norm": 0.28014089150395327,
      "learning_rate": 3.334239166998229e-06,
      "loss": 0.0025,
      "step": 39560
    },
    {
      "epoch": 0.6474678884071013,
      "grad_norm": 0.09728016720515517,
      "learning_rate": 3.3315467465286288e-06,
      "loss": 0.0036,
      "step": 39570
    },
    {
      "epoch": 0.6476315143581772,
      "grad_norm": 0.06672731755674141,
      "learning_rate": 3.3288548703623975e-06,
      "loss": 0.0022,
      "step": 39580
    },
    {
      "epoch": 0.647795140309253,
      "grad_norm": 0.1940579031224912,
      "learning_rate": 3.3261635393777094e-06,
      "loss": 0.0035,
      "step": 39590
    },
    {
      "epoch": 0.6479587662603289,
      "grad_norm": 0.022336458287932782,
      "learning_rate": 3.323472754452567e-06,
      "loss": 0.0017,
      "step": 39600
    },
    {
      "epoch": 0.6481223922114048,
      "grad_norm": 0.13438386381020884,
      "learning_rate": 3.3207825164647887e-06,
      "loss": 0.0024,
      "step": 39610
    },
    {
      "epoch": 0.6482860181624805,
      "grad_norm": 0.17504290350410256,
      "learning_rate": 3.3180928262920172e-06,
      "loss": 0.0041,
      "step": 39620
    },
    {
      "epoch": 0.6484496441135564,
      "grad_norm": 0.1386345402065505,
      "learning_rate": 3.315403684811718e-06,
      "loss": 0.003,
      "step": 39630
    },
    {
      "epoch": 0.6486132700646322,
      "grad_norm": 0.11579370961462987,
      "learning_rate": 3.3127150929011746e-06,
      "loss": 0.0017,
      "step": 39640
    },
    {
      "epoch": 0.6487768960157081,
      "grad_norm": 0.1202485041560297,
      "learning_rate": 3.310027051437492e-06,
      "loss": 0.0024,
      "step": 39650
    },
    {
      "epoch": 0.648940521966784,
      "grad_norm": 0.12061792889466585,
      "learning_rate": 3.307339561297599e-06,
      "loss": 0.0018,
      "step": 39660
    },
    {
      "epoch": 0.6491041479178598,
      "grad_norm": 0.11712073233399824,
      "learning_rate": 3.3046526233582395e-06,
      "loss": 0.0031,
      "step": 39670
    },
    {
      "epoch": 0.6492677738689356,
      "grad_norm": 0.14256584072034356,
      "learning_rate": 3.3019662384959785e-06,
      "loss": 0.0023,
      "step": 39680
    },
    {
      "epoch": 0.6494313998200114,
      "grad_norm": 0.1709936322373615,
      "learning_rate": 3.2992804075872044e-06,
      "loss": 0.002,
      "step": 39690
    },
    {
      "epoch": 0.6495950257710873,
      "grad_norm": 0.09806191266742119,
      "learning_rate": 3.29659513150812e-06,
      "loss": 0.0018,
      "step": 39700
    },
    {
      "epoch": 0.6497586517221632,
      "grad_norm": 0.12957565176309693,
      "learning_rate": 3.2939104111347486e-06,
      "loss": 0.0022,
      "step": 39710
    },
    {
      "epoch": 0.649922277673239,
      "grad_norm": 0.20161926415889336,
      "learning_rate": 3.2912262473429356e-06,
      "loss": 0.0017,
      "step": 39720
    },
    {
      "epoch": 0.6500859036243148,
      "grad_norm": 0.04588152739432965,
      "learning_rate": 3.28854264100834e-06,
      "loss": 0.0024,
      "step": 39730
    },
    {
      "epoch": 0.6502495295753906,
      "grad_norm": 0.09643820719669284,
      "learning_rate": 3.2858595930064396e-06,
      "loss": 0.0028,
      "step": 39740
    },
    {
      "epoch": 0.6504131555264665,
      "grad_norm": 0.05130347824026865,
      "learning_rate": 3.283177104212534e-06,
      "loss": 0.0043,
      "step": 39750
    },
    {
      "epoch": 0.6505767814775424,
      "grad_norm": 0.02360428084797385,
      "learning_rate": 3.2804951755017367e-06,
      "loss": 0.0025,
      "step": 39760
    },
    {
      "epoch": 0.6507404074286182,
      "grad_norm": 0.2494168643005269,
      "learning_rate": 3.2778138077489787e-06,
      "loss": 0.0023,
      "step": 39770
    },
    {
      "epoch": 0.650904033379694,
      "grad_norm": 0.028880625257212113,
      "learning_rate": 3.27513300182901e-06,
      "loss": 0.0022,
      "step": 39780
    },
    {
      "epoch": 0.6510676593307698,
      "grad_norm": 0.1292486455571358,
      "learning_rate": 3.2724527586163964e-06,
      "loss": 0.0031,
      "step": 39790
    },
    {
      "epoch": 0.6512312852818457,
      "grad_norm": 0.145923383876627,
      "learning_rate": 3.2697730789855177e-06,
      "loss": 0.0035,
      "step": 39800
    },
    {
      "epoch": 0.6513949112329216,
      "grad_norm": 0.2389334119865392,
      "learning_rate": 3.2670939638105735e-06,
      "loss": 0.0027,
      "step": 39810
    },
    {
      "epoch": 0.6515585371839974,
      "grad_norm": 0.05829126606389657,
      "learning_rate": 3.2644154139655785e-06,
      "loss": 0.004,
      "step": 39820
    },
    {
      "epoch": 0.6517221631350733,
      "grad_norm": 0.1287373268218403,
      "learning_rate": 3.2617374303243616e-06,
      "loss": 0.0032,
      "step": 39830
    },
    {
      "epoch": 0.651885789086149,
      "grad_norm": 0.12808499409673615,
      "learning_rate": 3.2590600137605657e-06,
      "loss": 0.002,
      "step": 39840
    },
    {
      "epoch": 0.6520494150372249,
      "grad_norm": 0.07306170449142108,
      "learning_rate": 3.2563831651476536e-06,
      "loss": 0.002,
      "step": 39850
    },
    {
      "epoch": 0.6522130409883008,
      "grad_norm": 0.07775133813934278,
      "learning_rate": 3.253706885358898e-06,
      "loss": 0.0031,
      "step": 39860
    },
    {
      "epoch": 0.6523766669393766,
      "grad_norm": 0.13357317538168761,
      "learning_rate": 3.2510311752673884e-06,
      "loss": 0.0055,
      "step": 39870
    },
    {
      "epoch": 0.6525402928904525,
      "grad_norm": 0.1521847707567769,
      "learning_rate": 3.2483560357460276e-06,
      "loss": 0.003,
      "step": 39880
    },
    {
      "epoch": 0.6527039188415282,
      "grad_norm": 0.4127917503806797,
      "learning_rate": 3.2456814676675334e-06,
      "loss": 0.0033,
      "step": 39890
    },
    {
      "epoch": 0.6528675447926041,
      "grad_norm": 0.18561189078890014,
      "learning_rate": 3.2430074719044345e-06,
      "loss": 0.002,
      "step": 39900
    },
    {
      "epoch": 0.65303117074368,
      "grad_norm": 0.12787073109409958,
      "learning_rate": 3.240334049329077e-06,
      "loss": 0.0044,
      "step": 39910
    },
    {
      "epoch": 0.6531947966947558,
      "grad_norm": 0.2866621415497609,
      "learning_rate": 3.2376612008136173e-06,
      "loss": 0.0034,
      "step": 39920
    },
    {
      "epoch": 0.6533584226458317,
      "grad_norm": 0.15739206094542157,
      "learning_rate": 3.234988927230023e-06,
      "loss": 0.002,
      "step": 39930
    },
    {
      "epoch": 0.6535220485969074,
      "grad_norm": 0.2795931432108628,
      "learning_rate": 3.232317229450078e-06,
      "loss": 0.0021,
      "step": 39940
    },
    {
      "epoch": 0.6536856745479833,
      "grad_norm": 0.09170065126543997,
      "learning_rate": 3.2296461083453768e-06,
      "loss": 0.003,
      "step": 39950
    },
    {
      "epoch": 0.6538493004990592,
      "grad_norm": 0.08503289135452773,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.0032,
      "step": 39960
    },
    {
      "epoch": 0.654012926450135,
      "grad_norm": 0.11842476542051038,
      "learning_rate": 3.224305599647135e-06,
      "loss": 0.0028,
      "step": 39970
    },
    {
      "epoch": 0.6541765524012109,
      "grad_norm": 0.20862033495901083,
      "learning_rate": 3.2216362137958456e-06,
      "loss": 0.0039,
      "step": 39980
    },
    {
      "epoch": 0.6543401783522866,
      "grad_norm": 0.12705672164595538,
      "learning_rate": 3.2189674081042875e-06,
      "loss": 0.0027,
      "step": 39990
    },
    {
      "epoch": 0.6545038043033625,
      "grad_norm": 0.2860372628199809,
      "learning_rate": 3.216299183443118e-06,
      "loss": 0.0017,
      "step": 40000
    },
    {
      "epoch": 0.6546674302544384,
      "grad_norm": 0.23223418361787682,
      "learning_rate": 3.2136315406827967e-06,
      "loss": 0.0023,
      "step": 40010
    },
    {
      "epoch": 0.6548310562055142,
      "grad_norm": 0.012887209491306267,
      "learning_rate": 3.21096448069359e-06,
      "loss": 0.0042,
      "step": 40020
    },
    {
      "epoch": 0.6549946821565901,
      "grad_norm": 0.08575492368848339,
      "learning_rate": 3.208298004345586e-06,
      "loss": 0.0032,
      "step": 40030
    },
    {
      "epoch": 0.6551583081076658,
      "grad_norm": 0.14995352057200062,
      "learning_rate": 3.2056321125086724e-06,
      "loss": 0.002,
      "step": 40040
    },
    {
      "epoch": 0.6553219340587417,
      "grad_norm": 0.06966696589240441,
      "learning_rate": 3.2029668060525467e-06,
      "loss": 0.0018,
      "step": 40050
    },
    {
      "epoch": 0.6554855600098175,
      "grad_norm": 0.05765247420159761,
      "learning_rate": 3.2003020858467224e-06,
      "loss": 0.0025,
      "step": 40060
    },
    {
      "epoch": 0.6556491859608934,
      "grad_norm": 0.2751262712319028,
      "learning_rate": 3.1976379527605163e-06,
      "loss": 0.0034,
      "step": 40070
    },
    {
      "epoch": 0.6558128119119693,
      "grad_norm": 0.08558690181995803,
      "learning_rate": 3.1949744076630505e-06,
      "loss": 0.0024,
      "step": 40080
    },
    {
      "epoch": 0.655976437863045,
      "grad_norm": 0.08569937119368176,
      "learning_rate": 3.192311451423266e-06,
      "loss": 0.0017,
      "step": 40090
    },
    {
      "epoch": 0.6561400638141209,
      "grad_norm": 0.28402650808015234,
      "learning_rate": 3.1896490849099015e-06,
      "loss": 0.0016,
      "step": 40100
    },
    {
      "epoch": 0.6563036897651967,
      "grad_norm": 0.2821033728842668,
      "learning_rate": 3.186987308991505e-06,
      "loss": 0.0019,
      "step": 40110
    },
    {
      "epoch": 0.6564673157162726,
      "grad_norm": 0.06455344900533107,
      "learning_rate": 3.1843261245364383e-06,
      "loss": 0.0021,
      "step": 40120
    },
    {
      "epoch": 0.6566309416673485,
      "grad_norm": 0.06232349812964358,
      "learning_rate": 3.181665532412865e-06,
      "loss": 0.0046,
      "step": 40130
    },
    {
      "epoch": 0.6567945676184243,
      "grad_norm": 0.1712542184772028,
      "learning_rate": 3.1790055334887516e-06,
      "loss": 0.0035,
      "step": 40140
    },
    {
      "epoch": 0.6569581935695001,
      "grad_norm": 0.4522706649109714,
      "learning_rate": 3.1763461286318815e-06,
      "loss": 0.0081,
      "step": 40150
    },
    {
      "epoch": 0.6571218195205759,
      "grad_norm": 0.10160296605985261,
      "learning_rate": 3.173687318709837e-06,
      "loss": 0.0022,
      "step": 40160
    },
    {
      "epoch": 0.6572854454716518,
      "grad_norm": 0.19452493654430852,
      "learning_rate": 3.1710291045900054e-06,
      "loss": 0.003,
      "step": 40170
    },
    {
      "epoch": 0.6574490714227277,
      "grad_norm": 0.08345665002389553,
      "learning_rate": 3.168371487139582e-06,
      "loss": 0.0028,
      "step": 40180
    },
    {
      "epoch": 0.6576126973738035,
      "grad_norm": 0.06317711366288563,
      "learning_rate": 3.1657144672255726e-06,
      "loss": 0.0017,
      "step": 40190
    },
    {
      "epoch": 0.6577763233248793,
      "grad_norm": 0.0568274186794387,
      "learning_rate": 3.1630580457147777e-06,
      "loss": 0.0044,
      "step": 40200
    },
    {
      "epoch": 0.6579399492759551,
      "grad_norm": 0.14384420536684045,
      "learning_rate": 3.1604022234738073e-06,
      "loss": 0.0031,
      "step": 40210
    },
    {
      "epoch": 0.658103575227031,
      "grad_norm": 0.15772566652543338,
      "learning_rate": 3.1577470013690823e-06,
      "loss": 0.0038,
      "step": 40220
    },
    {
      "epoch": 0.6582672011781069,
      "grad_norm": 0.10366514866569605,
      "learning_rate": 3.155092380266817e-06,
      "loss": 0.0021,
      "step": 40230
    },
    {
      "epoch": 0.6584308271291827,
      "grad_norm": 0.1699050099509039,
      "learning_rate": 3.1524383610330345e-06,
      "loss": 0.0026,
      "step": 40240
    },
    {
      "epoch": 0.6585944530802585,
      "grad_norm": 0.07912439335916925,
      "learning_rate": 3.1497849445335654e-06,
      "loss": 0.003,
      "step": 40250
    },
    {
      "epoch": 0.6587580790313343,
      "grad_norm": 0.1522661454787654,
      "learning_rate": 3.1471321316340364e-06,
      "loss": 0.0028,
      "step": 40260
    },
    {
      "epoch": 0.6589217049824102,
      "grad_norm": 0.04008214753582326,
      "learning_rate": 3.1444799231998805e-06,
      "loss": 0.0013,
      "step": 40270
    },
    {
      "epoch": 0.6590853309334861,
      "grad_norm": 0.12155508234919483,
      "learning_rate": 3.1418283200963385e-06,
      "loss": 0.0034,
      "step": 40280
    },
    {
      "epoch": 0.6592489568845619,
      "grad_norm": 0.2074379356953341,
      "learning_rate": 3.139177323188445e-06,
      "loss": 0.0045,
      "step": 40290
    },
    {
      "epoch": 0.6594125828356378,
      "grad_norm": 0.03942891942010023,
      "learning_rate": 3.1365269333410385e-06,
      "loss": 0.0032,
      "step": 40300
    },
    {
      "epoch": 0.6595762087867135,
      "grad_norm": 0.08922873729187883,
      "learning_rate": 3.133877151418769e-06,
      "loss": 0.0023,
      "step": 40310
    },
    {
      "epoch": 0.6597398347377894,
      "grad_norm": 0.07838412449617105,
      "learning_rate": 3.1312279782860755e-06,
      "loss": 0.003,
      "step": 40320
    },
    {
      "epoch": 0.6599034606888653,
      "grad_norm": 0.05000926522499979,
      "learning_rate": 3.1285794148072034e-06,
      "loss": 0.0048,
      "step": 40330
    },
    {
      "epoch": 0.6600670866399411,
      "grad_norm": 0.05765057889722899,
      "learning_rate": 3.1259314618462046e-06,
      "loss": 0.0016,
      "step": 40340
    },
    {
      "epoch": 0.660230712591017,
      "grad_norm": 0.2655843045276707,
      "learning_rate": 3.1232841202669228e-06,
      "loss": 0.0026,
      "step": 40350
    },
    {
      "epoch": 0.6603943385420927,
      "grad_norm": 0.30198475608781566,
      "learning_rate": 3.1206373909330057e-06,
      "loss": 0.0041,
      "step": 40360
    },
    {
      "epoch": 0.6605579644931686,
      "grad_norm": 0.12496489904880949,
      "learning_rate": 3.1179912747079065e-06,
      "loss": 0.0035,
      "step": 40370
    },
    {
      "epoch": 0.6607215904442445,
      "grad_norm": 0.02567933770593977,
      "learning_rate": 3.1153457724548707e-06,
      "loss": 0.0028,
      "step": 40380
    },
    {
      "epoch": 0.6608852163953203,
      "grad_norm": 0.23412548210268688,
      "learning_rate": 3.112700885036945e-06,
      "loss": 0.0029,
      "step": 40390
    },
    {
      "epoch": 0.6610488423463962,
      "grad_norm": 0.13946125132828505,
      "learning_rate": 3.110056613316982e-06,
      "loss": 0.002,
      "step": 40400
    },
    {
      "epoch": 0.6612124682974719,
      "grad_norm": 0.09678302262651918,
      "learning_rate": 3.107412958157625e-06,
      "loss": 0.0021,
      "step": 40410
    },
    {
      "epoch": 0.6613760942485478,
      "grad_norm": 0.19509377889889065,
      "learning_rate": 3.1047699204213176e-06,
      "loss": 0.0021,
      "step": 40420
    },
    {
      "epoch": 0.6615397201996237,
      "grad_norm": 0.21633787104740393,
      "learning_rate": 3.102127500970311e-06,
      "loss": 0.002,
      "step": 40430
    },
    {
      "epoch": 0.6617033461506995,
      "grad_norm": 0.09215322596108254,
      "learning_rate": 3.0994857006666424e-06,
      "loss": 0.0033,
      "step": 40440
    },
    {
      "epoch": 0.6618669721017754,
      "grad_norm": 0.14275816767678096,
      "learning_rate": 3.096844520372152e-06,
      "loss": 0.0013,
      "step": 40450
    },
    {
      "epoch": 0.6620305980528511,
      "grad_norm": 0.3473689006139139,
      "learning_rate": 3.0942039609484833e-06,
      "loss": 0.0025,
      "step": 40460
    },
    {
      "epoch": 0.662194224003927,
      "grad_norm": 0.09610750533745403,
      "learning_rate": 3.091564023257067e-06,
      "loss": 0.0025,
      "step": 40470
    },
    {
      "epoch": 0.6623578499550029,
      "grad_norm": 0.23561934528778736,
      "learning_rate": 3.0889247081591368e-06,
      "loss": 0.0011,
      "step": 40480
    },
    {
      "epoch": 0.6625214759060787,
      "grad_norm": 0.16203910305151026,
      "learning_rate": 3.086286016515726e-06,
      "loss": 0.0019,
      "step": 40490
    },
    {
      "epoch": 0.6626851018571546,
      "grad_norm": 0.19275699794863643,
      "learning_rate": 3.0836479491876582e-06,
      "loss": 0.0024,
      "step": 40500
    },
    {
      "epoch": 0.6628487278082303,
      "grad_norm": 0.32697829671129397,
      "learning_rate": 3.0810105070355567e-06,
      "loss": 0.0048,
      "step": 40510
    },
    {
      "epoch": 0.6630123537593062,
      "grad_norm": 0.36792904683152244,
      "learning_rate": 3.0783736909198393e-06,
      "loss": 0.0021,
      "step": 40520
    },
    {
      "epoch": 0.6631759797103821,
      "grad_norm": 0.21469863595469219,
      "learning_rate": 3.075737501700723e-06,
      "loss": 0.0031,
      "step": 40530
    },
    {
      "epoch": 0.6633396056614579,
      "grad_norm": 0.14359639461928936,
      "learning_rate": 3.0731019402382167e-06,
      "loss": 0.0019,
      "step": 40540
    },
    {
      "epoch": 0.6635032316125338,
      "grad_norm": 0.054395701590773886,
      "learning_rate": 3.070467007392125e-06,
      "loss": 0.0025,
      "step": 40550
    },
    {
      "epoch": 0.6636668575636095,
      "grad_norm": 0.24050371405752213,
      "learning_rate": 3.067832704022049e-06,
      "loss": 0.0031,
      "step": 40560
    },
    {
      "epoch": 0.6638304835146854,
      "grad_norm": 0.1290172505629635,
      "learning_rate": 3.0651990309873846e-06,
      "loss": 0.0039,
      "step": 40570
    },
    {
      "epoch": 0.6639941094657613,
      "grad_norm": 0.03202927829674529,
      "learning_rate": 3.062565989147318e-06,
      "loss": 0.0038,
      "step": 40580
    },
    {
      "epoch": 0.6641577354168371,
      "grad_norm": 0.18586337574070458,
      "learning_rate": 3.0599335793608353e-06,
      "loss": 0.0025,
      "step": 40590
    },
    {
      "epoch": 0.664321361367913,
      "grad_norm": 0.36856125265353057,
      "learning_rate": 3.057301802486714e-06,
      "loss": 0.003,
      "step": 40600
    },
    {
      "epoch": 0.6644849873189888,
      "grad_norm": 0.14686951370124568,
      "learning_rate": 3.0546706593835214e-06,
      "loss": 0.0033,
      "step": 40610
    },
    {
      "epoch": 0.6646486132700646,
      "grad_norm": 0.021412108609240532,
      "learning_rate": 3.0520401509096248e-06,
      "loss": 0.0019,
      "step": 40620
    },
    {
      "epoch": 0.6648122392211405,
      "grad_norm": 0.005325258307676092,
      "learning_rate": 3.04941027792318e-06,
      "loss": 0.0027,
      "step": 40630
    },
    {
      "epoch": 0.6649758651722163,
      "grad_norm": 0.11931541068795616,
      "learning_rate": 3.046781041282135e-06,
      "loss": 0.0023,
      "step": 40640
    },
    {
      "epoch": 0.6651394911232922,
      "grad_norm": 0.17313174219269734,
      "learning_rate": 3.044152441844235e-06,
      "loss": 0.0035,
      "step": 40650
    },
    {
      "epoch": 0.665303117074368,
      "grad_norm": 0.15497922283450521,
      "learning_rate": 3.0415244804670106e-06,
      "loss": 0.0029,
      "step": 40660
    },
    {
      "epoch": 0.6654667430254438,
      "grad_norm": 0.2689743844869192,
      "learning_rate": 3.0388971580077896e-06,
      "loss": 0.0023,
      "step": 40670
    },
    {
      "epoch": 0.6656303689765197,
      "grad_norm": 0.129486847360236,
      "learning_rate": 3.0362704753236905e-06,
      "loss": 0.0038,
      "step": 40680
    },
    {
      "epoch": 0.6657939949275955,
      "grad_norm": 0.11890158294012872,
      "learning_rate": 3.0336444332716195e-06,
      "loss": 0.0028,
      "step": 40690
    },
    {
      "epoch": 0.6659576208786714,
      "grad_norm": 0.16904242009082043,
      "learning_rate": 3.031019032708278e-06,
      "loss": 0.0023,
      "step": 40700
    },
    {
      "epoch": 0.6661212468297472,
      "grad_norm": 0.20878737715978546,
      "learning_rate": 3.0283942744901562e-06,
      "loss": 0.0039,
      "step": 40710
    },
    {
      "epoch": 0.666284872780823,
      "grad_norm": 0.2644785465065048,
      "learning_rate": 3.0257701594735356e-06,
      "loss": 0.003,
      "step": 40720
    },
    {
      "epoch": 0.6664484987318989,
      "grad_norm": 0.12457724055598955,
      "learning_rate": 3.0231466885144856e-06,
      "loss": 0.0024,
      "step": 40730
    },
    {
      "epoch": 0.6666121246829747,
      "grad_norm": 0.14195933205667619,
      "learning_rate": 3.0205238624688692e-06,
      "loss": 0.0025,
      "step": 40740
    },
    {
      "epoch": 0.6667757506340506,
      "grad_norm": 0.05734868462805376,
      "learning_rate": 3.0179016821923357e-06,
      "loss": 0.0014,
      "step": 40750
    },
    {
      "epoch": 0.6669393765851264,
      "grad_norm": 0.07968922096620716,
      "learning_rate": 3.0152801485403254e-06,
      "loss": 0.0016,
      "step": 40760
    },
    {
      "epoch": 0.6671030025362022,
      "grad_norm": 0.0863898649762563,
      "learning_rate": 3.0126592623680682e-06,
      "loss": 0.0017,
      "step": 40770
    },
    {
      "epoch": 0.6672666284872781,
      "grad_norm": 0.3219425469609863,
      "learning_rate": 3.010039024530581e-06,
      "loss": 0.003,
      "step": 40780
    },
    {
      "epoch": 0.6674302544383539,
      "grad_norm": 0.21590436910313154,
      "learning_rate": 3.0074194358826686e-06,
      "loss": 0.0018,
      "step": 40790
    },
    {
      "epoch": 0.6675938803894298,
      "grad_norm": 0.09149583124572988,
      "learning_rate": 3.0048004972789284e-06,
      "loss": 0.0016,
      "step": 40800
    },
    {
      "epoch": 0.6677575063405056,
      "grad_norm": 0.35200166106536224,
      "learning_rate": 3.002182209573741e-06,
      "loss": 0.003,
      "step": 40810
    },
    {
      "epoch": 0.6679211322915815,
      "grad_norm": 0.07196795209815954,
      "learning_rate": 2.9995645736212777e-06,
      "loss": 0.0007,
      "step": 40820
    },
    {
      "epoch": 0.6680847582426573,
      "grad_norm": 0.10749828898044372,
      "learning_rate": 2.9969475902754933e-06,
      "loss": 0.0018,
      "step": 40830
    },
    {
      "epoch": 0.6682483841937331,
      "grad_norm": 0.11516874932859363,
      "learning_rate": 2.9943312603901355e-06,
      "loss": 0.0021,
      "step": 40840
    },
    {
      "epoch": 0.668412010144809,
      "grad_norm": 0.29385288483058314,
      "learning_rate": 2.991715584818734e-06,
      "loss": 0.0071,
      "step": 40850
    },
    {
      "epoch": 0.6685756360958848,
      "grad_norm": 0.07950238268890146,
      "learning_rate": 2.9891005644146064e-06,
      "loss": 0.0033,
      "step": 40860
    },
    {
      "epoch": 0.6687392620469607,
      "grad_norm": 0.19094952208535834,
      "learning_rate": 2.9864862000308583e-06,
      "loss": 0.003,
      "step": 40870
    },
    {
      "epoch": 0.6689028879980365,
      "grad_norm": 0.10810314690855122,
      "learning_rate": 2.983872492520379e-06,
      "loss": 0.0025,
      "step": 40880
    },
    {
      "epoch": 0.6690665139491123,
      "grad_norm": 0.08182233997925421,
      "learning_rate": 2.981259442735843e-06,
      "loss": 0.0021,
      "step": 40890
    },
    {
      "epoch": 0.6692301399001882,
      "grad_norm": 0.01660795710215538,
      "learning_rate": 2.978647051529714e-06,
      "loss": 0.0025,
      "step": 40900
    },
    {
      "epoch": 0.669393765851264,
      "grad_norm": 0.0821574715269833,
      "learning_rate": 2.976035319754238e-06,
      "loss": 0.0019,
      "step": 40910
    },
    {
      "epoch": 0.6695573918023399,
      "grad_norm": 0.270674784852337,
      "learning_rate": 2.973424248261444e-06,
      "loss": 0.002,
      "step": 40920
    },
    {
      "epoch": 0.6697210177534156,
      "grad_norm": 0.13100342881669663,
      "learning_rate": 2.9708138379031503e-06,
      "loss": 0.002,
      "step": 40930
    },
    {
      "epoch": 0.6698846437044915,
      "grad_norm": 0.0820058590816915,
      "learning_rate": 2.9682040895309572e-06,
      "loss": 0.005,
      "step": 40940
    },
    {
      "epoch": 0.6700482696555674,
      "grad_norm": 0.040961428748845265,
      "learning_rate": 2.9655950039962465e-06,
      "loss": 0.0021,
      "step": 40950
    },
    {
      "epoch": 0.6702118956066432,
      "grad_norm": 0.06012620824639548,
      "learning_rate": 2.962986582150189e-06,
      "loss": 0.0074,
      "step": 40960
    },
    {
      "epoch": 0.6703755215577191,
      "grad_norm": 0.1834192949993341,
      "learning_rate": 2.960378824843735e-06,
      "loss": 0.0027,
      "step": 40970
    },
    {
      "epoch": 0.6705391475087948,
      "grad_norm": 0.028166952745558246,
      "learning_rate": 2.9577717329276186e-06,
      "loss": 0.0015,
      "step": 40980
    },
    {
      "epoch": 0.6707027734598707,
      "grad_norm": 0.6164414045027284,
      "learning_rate": 2.9551653072523584e-06,
      "loss": 0.0025,
      "step": 40990
    },
    {
      "epoch": 0.6708663994109466,
      "grad_norm": 0.10695079591084021,
      "learning_rate": 2.9525595486682545e-06,
      "loss": 0.0026,
      "step": 41000
    },
    {
      "epoch": 0.6710300253620224,
      "grad_norm": 0.07195091333837633,
      "learning_rate": 2.9499544580253893e-06,
      "loss": 0.0015,
      "step": 41010
    },
    {
      "epoch": 0.6711936513130983,
      "grad_norm": 0.1884421931087768,
      "learning_rate": 2.9473500361736273e-06,
      "loss": 0.0023,
      "step": 41020
    },
    {
      "epoch": 0.671357277264174,
      "grad_norm": 0.1028414966117629,
      "learning_rate": 2.9447462839626163e-06,
      "loss": 0.0024,
      "step": 41030
    },
    {
      "epoch": 0.6715209032152499,
      "grad_norm": 0.16820385801526425,
      "learning_rate": 2.942143202241782e-06,
      "loss": 0.0024,
      "step": 41040
    },
    {
      "epoch": 0.6716845291663258,
      "grad_norm": 0.2680442443312905,
      "learning_rate": 2.939540791860337e-06,
      "loss": 0.0038,
      "step": 41050
    },
    {
      "epoch": 0.6718481551174016,
      "grad_norm": 0.022807104573804717,
      "learning_rate": 2.936939053667269e-06,
      "loss": 0.0013,
      "step": 41060
    },
    {
      "epoch": 0.6720117810684775,
      "grad_norm": 0.007735201551940222,
      "learning_rate": 2.9343379885113497e-06,
      "loss": 0.0017,
      "step": 41070
    },
    {
      "epoch": 0.6721754070195533,
      "grad_norm": 0.06415022831170143,
      "learning_rate": 2.9317375972411322e-06,
      "loss": 0.0025,
      "step": 41080
    },
    {
      "epoch": 0.6723390329706291,
      "grad_norm": 0.06374971429116409,
      "learning_rate": 2.929137880704947e-06,
      "loss": 0.002,
      "step": 41090
    },
    {
      "epoch": 0.672502658921705,
      "grad_norm": 0.26021841945345336,
      "learning_rate": 2.9265388397509043e-06,
      "loss": 0.0023,
      "step": 41100
    },
    {
      "epoch": 0.6726662848727808,
      "grad_norm": 0.17526730558955822,
      "learning_rate": 2.9239404752268973e-06,
      "loss": 0.0034,
      "step": 41110
    },
    {
      "epoch": 0.6728299108238567,
      "grad_norm": 0.0962750531408298,
      "learning_rate": 2.9213427879805954e-06,
      "loss": 0.0026,
      "step": 41120
    },
    {
      "epoch": 0.6729935367749325,
      "grad_norm": 0.09437558482822837,
      "learning_rate": 2.91874577885945e-06,
      "loss": 0.0027,
      "step": 41130
    },
    {
      "epoch": 0.6731571627260083,
      "grad_norm": 0.010445814699426627,
      "learning_rate": 2.916149448710686e-06,
      "loss": 0.0024,
      "step": 41140
    },
    {
      "epoch": 0.6733207886770842,
      "grad_norm": 0.10779266810784531,
      "learning_rate": 2.913553798381313e-06,
      "loss": 0.0018,
      "step": 41150
    },
    {
      "epoch": 0.67348441462816,
      "grad_norm": 0.1617484228394904,
      "learning_rate": 2.9109588287181144e-06,
      "loss": 0.0029,
      "step": 41160
    },
    {
      "epoch": 0.6736480405792359,
      "grad_norm": 0.10506760160477985,
      "learning_rate": 2.908364540567652e-06,
      "loss": 0.0021,
      "step": 41170
    },
    {
      "epoch": 0.6738116665303117,
      "grad_norm": 0.2934059769692995,
      "learning_rate": 2.905770934776271e-06,
      "loss": 0.0025,
      "step": 41180
    },
    {
      "epoch": 0.6739752924813875,
      "grad_norm": 0.09724591524984555,
      "learning_rate": 2.9031780121900855e-06,
      "loss": 0.0029,
      "step": 41190
    },
    {
      "epoch": 0.6741389184324634,
      "grad_norm": 0.16370221246817585,
      "learning_rate": 2.9005857736549926e-06,
      "loss": 0.0031,
      "step": 41200
    },
    {
      "epoch": 0.6743025443835392,
      "grad_norm": 0.10688248175711239,
      "learning_rate": 2.8979942200166635e-06,
      "loss": 0.0015,
      "step": 41210
    },
    {
      "epoch": 0.6744661703346151,
      "grad_norm": 0.37216064439188923,
      "learning_rate": 2.895403352120546e-06,
      "loss": 0.0032,
      "step": 41220
    },
    {
      "epoch": 0.6746297962856909,
      "grad_norm": 0.289860475146755,
      "learning_rate": 2.8928131708118635e-06,
      "loss": 0.0035,
      "step": 41230
    },
    {
      "epoch": 0.6747934222367667,
      "grad_norm": 0.08497006426129707,
      "learning_rate": 2.890223676935622e-06,
      "loss": 0.0032,
      "step": 41240
    },
    {
      "epoch": 0.6749570481878426,
      "grad_norm": 0.2525358679804982,
      "learning_rate": 2.8876348713365944e-06,
      "loss": 0.0018,
      "step": 41250
    },
    {
      "epoch": 0.6751206741389184,
      "grad_norm": 0.4327332370270823,
      "learning_rate": 2.885046754859333e-06,
      "loss": 0.006,
      "step": 41260
    },
    {
      "epoch": 0.6752843000899943,
      "grad_norm": 0.12277440634157726,
      "learning_rate": 2.8824593283481663e-06,
      "loss": 0.0033,
      "step": 41270
    },
    {
      "epoch": 0.6754479260410701,
      "grad_norm": 0.24532649375838544,
      "learning_rate": 2.8798725926471953e-06,
      "loss": 0.002,
      "step": 41280
    },
    {
      "epoch": 0.675611551992146,
      "grad_norm": 0.11038957527955254,
      "learning_rate": 2.8772865486002955e-06,
      "loss": 0.0045,
      "step": 41290
    },
    {
      "epoch": 0.6757751779432218,
      "grad_norm": 0.11954932068585873,
      "learning_rate": 2.8747011970511207e-06,
      "loss": 0.0027,
      "step": 41300
    },
    {
      "epoch": 0.6759388038942976,
      "grad_norm": 0.09155624218907009,
      "learning_rate": 2.8721165388430956e-06,
      "loss": 0.003,
      "step": 41310
    },
    {
      "epoch": 0.6761024298453735,
      "grad_norm": 0.10862737127753798,
      "learning_rate": 2.869532574819418e-06,
      "loss": 0.0021,
      "step": 41320
    },
    {
      "epoch": 0.6762660557964493,
      "grad_norm": 0.031737075572473454,
      "learning_rate": 2.866949305823062e-06,
      "loss": 0.0031,
      "step": 41330
    },
    {
      "epoch": 0.6764296817475252,
      "grad_norm": 0.1627864680478158,
      "learning_rate": 2.8643667326967715e-06,
      "loss": 0.0048,
      "step": 41340
    },
    {
      "epoch": 0.676593307698601,
      "grad_norm": 0.13272210155579808,
      "learning_rate": 2.8617848562830653e-06,
      "loss": 0.0016,
      "step": 41350
    },
    {
      "epoch": 0.6767569336496768,
      "grad_norm": 0.1289969767177936,
      "learning_rate": 2.8592036774242383e-06,
      "loss": 0.0016,
      "step": 41360
    },
    {
      "epoch": 0.6769205596007527,
      "grad_norm": 0.08914399738044176,
      "learning_rate": 2.8566231969623507e-06,
      "loss": 0.0008,
      "step": 41370
    },
    {
      "epoch": 0.6770841855518285,
      "grad_norm": 0.35248995228486585,
      "learning_rate": 2.854043415739241e-06,
      "loss": 0.0032,
      "step": 41380
    },
    {
      "epoch": 0.6772478115029044,
      "grad_norm": 0.12126033527672908,
      "learning_rate": 2.8514643345965163e-06,
      "loss": 0.0013,
      "step": 41390
    },
    {
      "epoch": 0.6774114374539802,
      "grad_norm": 0.021444309320496187,
      "learning_rate": 2.848885954375556e-06,
      "loss": 0.0022,
      "step": 41400
    },
    {
      "epoch": 0.677575063405056,
      "grad_norm": 0.20079350632874507,
      "learning_rate": 2.846308275917509e-06,
      "loss": 0.0043,
      "step": 41410
    },
    {
      "epoch": 0.6777386893561319,
      "grad_norm": 0.2263365126980432,
      "learning_rate": 2.8437313000633018e-06,
      "loss": 0.0024,
      "step": 41420
    },
    {
      "epoch": 0.6779023153072077,
      "grad_norm": 0.16264446107881175,
      "learning_rate": 2.841155027653625e-06,
      "loss": 0.0018,
      "step": 41430
    },
    {
      "epoch": 0.6780659412582836,
      "grad_norm": 0.09150596021591757,
      "learning_rate": 2.8385794595289407e-06,
      "loss": 0.0018,
      "step": 41440
    },
    {
      "epoch": 0.6782295672093595,
      "grad_norm": 0.17786958743712,
      "learning_rate": 2.8360045965294847e-06,
      "loss": 0.0019,
      "step": 41450
    },
    {
      "epoch": 0.6783931931604352,
      "grad_norm": 0.1974893193267756,
      "learning_rate": 2.8334304394952596e-06,
      "loss": 0.0033,
      "step": 41460
    },
    {
      "epoch": 0.6785568191115111,
      "grad_norm": 0.14526319601973367,
      "learning_rate": 2.8308569892660353e-06,
      "loss": 0.0028,
      "step": 41470
    },
    {
      "epoch": 0.6787204450625869,
      "grad_norm": 0.15592358737235928,
      "learning_rate": 2.8282842466813604e-06,
      "loss": 0.0023,
      "step": 41480
    },
    {
      "epoch": 0.6788840710136628,
      "grad_norm": 0.09372493912025202,
      "learning_rate": 2.825712212580544e-06,
      "loss": 0.0031,
      "step": 41490
    },
    {
      "epoch": 0.6790476969647387,
      "grad_norm": 0.21160221563883158,
      "learning_rate": 2.8231408878026657e-06,
      "loss": 0.0041,
      "step": 41500
    },
    {
      "epoch": 0.6792113229158144,
      "grad_norm": 0.11999611124703345,
      "learning_rate": 2.8205702731865757e-06,
      "loss": 0.0016,
      "step": 41510
    },
    {
      "epoch": 0.6793749488668903,
      "grad_norm": 0.21433712641464028,
      "learning_rate": 2.8180003695708925e-06,
      "loss": 0.0017,
      "step": 41520
    },
    {
      "epoch": 0.6795385748179661,
      "grad_norm": 0.15433427510459735,
      "learning_rate": 2.8154311777939995e-06,
      "loss": 0.0017,
      "step": 41530
    },
    {
      "epoch": 0.679702200769042,
      "grad_norm": 0.18691691966755242,
      "learning_rate": 2.81286269869405e-06,
      "loss": 0.0032,
      "step": 41540
    },
    {
      "epoch": 0.6798658267201179,
      "grad_norm": 0.1360947914601187,
      "learning_rate": 2.8102949331089676e-06,
      "loss": 0.0025,
      "step": 41550
    },
    {
      "epoch": 0.6800294526711936,
      "grad_norm": 0.21745306544946064,
      "learning_rate": 2.8077278818764386e-06,
      "loss": 0.0041,
      "step": 41560
    },
    {
      "epoch": 0.6801930786222695,
      "grad_norm": 0.181486665325196,
      "learning_rate": 2.8051615458339188e-06,
      "loss": 0.0026,
      "step": 41570
    },
    {
      "epoch": 0.6803567045733453,
      "grad_norm": 0.10783998267729482,
      "learning_rate": 2.80259592581863e-06,
      "loss": 0.0017,
      "step": 41580
    },
    {
      "epoch": 0.6805203305244212,
      "grad_norm": 0.3006325954414733,
      "learning_rate": 2.8000310226675593e-06,
      "loss": 0.0022,
      "step": 41590
    },
    {
      "epoch": 0.6806839564754971,
      "grad_norm": 0.17007047018142366,
      "learning_rate": 2.7974668372174594e-06,
      "loss": 0.0025,
      "step": 41600
    },
    {
      "epoch": 0.6808475824265728,
      "grad_norm": 0.08376332811390634,
      "learning_rate": 2.7949033703048555e-06,
      "loss": 0.003,
      "step": 41610
    },
    {
      "epoch": 0.6810112083776487,
      "grad_norm": 0.30359373417007296,
      "learning_rate": 2.792340622766032e-06,
      "loss": 0.0027,
      "step": 41620
    },
    {
      "epoch": 0.6811748343287245,
      "grad_norm": 0.3445363412051261,
      "learning_rate": 2.789778595437034e-06,
      "loss": 0.0022,
      "step": 41630
    },
    {
      "epoch": 0.6813384602798004,
      "grad_norm": 0.07746944132159855,
      "learning_rate": 2.787217289153685e-06,
      "loss": 0.0045,
      "step": 41640
    },
    {
      "epoch": 0.6815020862308763,
      "grad_norm": 0.09365644305325534,
      "learning_rate": 2.784656704751562e-06,
      "loss": 0.0027,
      "step": 41650
    },
    {
      "epoch": 0.681665712181952,
      "grad_norm": 0.030013011260191038,
      "learning_rate": 2.7820968430660102e-06,
      "loss": 0.0022,
      "step": 41660
    },
    {
      "epoch": 0.6818293381330279,
      "grad_norm": 0.08126143794990909,
      "learning_rate": 2.7795377049321413e-06,
      "loss": 0.0025,
      "step": 41670
    },
    {
      "epoch": 0.6819929640841037,
      "grad_norm": 0.13863908921669518,
      "learning_rate": 2.7769792911848304e-06,
      "loss": 0.003,
      "step": 41680
    },
    {
      "epoch": 0.6821565900351796,
      "grad_norm": 0.21547303100062978,
      "learning_rate": 2.774421602658709e-06,
      "loss": 0.0014,
      "step": 41690
    },
    {
      "epoch": 0.6823202159862555,
      "grad_norm": 0.15936463847171142,
      "learning_rate": 2.771864640188182e-06,
      "loss": 0.0015,
      "step": 41700
    },
    {
      "epoch": 0.6824838419373312,
      "grad_norm": 0.1121288519069009,
      "learning_rate": 2.7693084046074127e-06,
      "loss": 0.0021,
      "step": 41710
    },
    {
      "epoch": 0.6826474678884071,
      "grad_norm": 0.08663973554616647,
      "learning_rate": 2.7667528967503244e-06,
      "loss": 0.0017,
      "step": 41720
    },
    {
      "epoch": 0.6828110938394829,
      "grad_norm": 0.033079284330341914,
      "learning_rate": 2.7641981174506118e-06,
      "loss": 0.0014,
      "step": 41730
    },
    {
      "epoch": 0.6829747197905588,
      "grad_norm": 0.30450504500214215,
      "learning_rate": 2.7616440675417257e-06,
      "loss": 0.0017,
      "step": 41740
    },
    {
      "epoch": 0.6831383457416347,
      "grad_norm": 0.12578510081389757,
      "learning_rate": 2.759090747856874e-06,
      "loss": 0.0055,
      "step": 41750
    },
    {
      "epoch": 0.6833019716927105,
      "grad_norm": 0.0365232978479947,
      "learning_rate": 2.7565381592290375e-06,
      "loss": 0.0014,
      "step": 41760
    },
    {
      "epoch": 0.6834655976437863,
      "grad_norm": 0.06815296567336027,
      "learning_rate": 2.753986302490952e-06,
      "loss": 0.0024,
      "step": 41770
    },
    {
      "epoch": 0.6836292235948621,
      "grad_norm": 0.11923673598337081,
      "learning_rate": 2.7514351784751126e-06,
      "loss": 0.0021,
      "step": 41780
    },
    {
      "epoch": 0.683792849545938,
      "grad_norm": 0.02716008884793376,
      "learning_rate": 2.7488847880137833e-06,
      "loss": 0.0043,
      "step": 41790
    },
    {
      "epoch": 0.6839564754970138,
      "grad_norm": 0.1538618945099668,
      "learning_rate": 2.746335131938983e-06,
      "loss": 0.0023,
      "step": 41800
    },
    {
      "epoch": 0.6841201014480897,
      "grad_norm": 0.013631361854169284,
      "learning_rate": 2.743786211082489e-06,
      "loss": 0.002,
      "step": 41810
    },
    {
      "epoch": 0.6842837273991655,
      "grad_norm": 0.07306195999373453,
      "learning_rate": 2.7412380262758407e-06,
      "loss": 0.0034,
      "step": 41820
    },
    {
      "epoch": 0.6844473533502413,
      "grad_norm": 0.22165486790302227,
      "learning_rate": 2.738690578350344e-06,
      "loss": 0.0031,
      "step": 41830
    },
    {
      "epoch": 0.6846109793013172,
      "grad_norm": 0.09818286395401093,
      "learning_rate": 2.7361438681370544e-06,
      "loss": 0.0026,
      "step": 41840
    },
    {
      "epoch": 0.684774605252393,
      "grad_norm": 0.10467488516708244,
      "learning_rate": 2.7335978964667897e-06,
      "loss": 0.0017,
      "step": 41850
    },
    {
      "epoch": 0.6849382312034689,
      "grad_norm": 0.039540836019590214,
      "learning_rate": 2.7310526641701353e-06,
      "loss": 0.0022,
      "step": 41860
    },
    {
      "epoch": 0.6851018571545447,
      "grad_norm": 0.10848181635045313,
      "learning_rate": 2.7285081720774207e-06,
      "loss": 0.0013,
      "step": 41870
    },
    {
      "epoch": 0.6852654831056205,
      "grad_norm": 0.09182993793651825,
      "learning_rate": 2.725964421018743e-06,
      "loss": 0.005,
      "step": 41880
    },
    {
      "epoch": 0.6854291090566964,
      "grad_norm": 0.2313383011637226,
      "learning_rate": 2.7234214118239592e-06,
      "loss": 0.0025,
      "step": 41890
    },
    {
      "epoch": 0.6855927350077722,
      "grad_norm": 0.44247973710128646,
      "learning_rate": 2.720879145322679e-06,
      "loss": 0.0037,
      "step": 41900
    },
    {
      "epoch": 0.6857563609588481,
      "grad_norm": 0.1468770738686299,
      "learning_rate": 2.7183376223442704e-06,
      "loss": 0.0021,
      "step": 41910
    },
    {
      "epoch": 0.685919986909924,
      "grad_norm": 0.06078744822822211,
      "learning_rate": 2.715796843717865e-06,
      "loss": 0.0025,
      "step": 41920
    },
    {
      "epoch": 0.6860836128609997,
      "grad_norm": 0.09079303704546728,
      "learning_rate": 2.7132568102723424e-06,
      "loss": 0.0029,
      "step": 41930
    },
    {
      "epoch": 0.6862472388120756,
      "grad_norm": 0.37809724214213103,
      "learning_rate": 2.7107175228363425e-06,
      "loss": 0.0028,
      "step": 41940
    },
    {
      "epoch": 0.6864108647631514,
      "grad_norm": 0.19551475233796622,
      "learning_rate": 2.708178982238268e-06,
      "loss": 0.002,
      "step": 41950
    },
    {
      "epoch": 0.6865744907142273,
      "grad_norm": 0.09981874090353988,
      "learning_rate": 2.70564118930627e-06,
      "loss": 0.003,
      "step": 41960
    },
    {
      "epoch": 0.6867381166653032,
      "grad_norm": 0.2986000350999538,
      "learning_rate": 2.7031041448682573e-06,
      "loss": 0.0026,
      "step": 41970
    },
    {
      "epoch": 0.6869017426163789,
      "grad_norm": 0.20915521672323764,
      "learning_rate": 2.7005678497519007e-06,
      "loss": 0.002,
      "step": 41980
    },
    {
      "epoch": 0.6870653685674548,
      "grad_norm": 0.2791850969433655,
      "learning_rate": 2.698032304784617e-06,
      "loss": 0.0039,
      "step": 41990
    },
    {
      "epoch": 0.6872289945185306,
      "grad_norm": 0.13210936726214428,
      "learning_rate": 2.695497510793581e-06,
      "loss": 0.0028,
      "step": 42000
    },
    {
      "epoch": 0.6873926204696065,
      "grad_norm": 0.2695609555567078,
      "learning_rate": 2.6929634686057305e-06,
      "loss": 0.0026,
      "step": 42010
    },
    {
      "epoch": 0.6875562464206824,
      "grad_norm": 0.10556648377407195,
      "learning_rate": 2.690430179047749e-06,
      "loss": 0.0017,
      "step": 42020
    },
    {
      "epoch": 0.6877198723717581,
      "grad_norm": 0.07840149888778411,
      "learning_rate": 2.687897642946075e-06,
      "loss": 0.0015,
      "step": 42030
    },
    {
      "epoch": 0.687883498322834,
      "grad_norm": 0.06528244167430895,
      "learning_rate": 2.685365861126911e-06,
      "loss": 0.0032,
      "step": 42040
    },
    {
      "epoch": 0.6880471242739098,
      "grad_norm": 0.24806960049129503,
      "learning_rate": 2.682834834416198e-06,
      "loss": 0.0032,
      "step": 42050
    },
    {
      "epoch": 0.6882107502249857,
      "grad_norm": 0.12819523596603435,
      "learning_rate": 2.68030456363964e-06,
      "loss": 0.0023,
      "step": 42060
    },
    {
      "epoch": 0.6883743761760616,
      "grad_norm": 0.12031307731792824,
      "learning_rate": 2.677775049622697e-06,
      "loss": 0.0031,
      "step": 42070
    },
    {
      "epoch": 0.6885380021271373,
      "grad_norm": 0.10054300019770546,
      "learning_rate": 2.6752462931905764e-06,
      "loss": 0.0039,
      "step": 42080
    },
    {
      "epoch": 0.6887016280782132,
      "grad_norm": 0.2599112448472979,
      "learning_rate": 2.6727182951682383e-06,
      "loss": 0.0039,
      "step": 42090
    },
    {
      "epoch": 0.688865254029289,
      "grad_norm": 0.05321191812424597,
      "learning_rate": 2.670191056380403e-06,
      "loss": 0.0014,
      "step": 42100
    },
    {
      "epoch": 0.6890288799803649,
      "grad_norm": 1.0787030677670457,
      "learning_rate": 2.667664577651531e-06,
      "loss": 0.004,
      "step": 42110
    },
    {
      "epoch": 0.6891925059314408,
      "grad_norm": 0.14782688536671787,
      "learning_rate": 2.665138859805842e-06,
      "loss": 0.0019,
      "step": 42120
    },
    {
      "epoch": 0.6893561318825165,
      "grad_norm": 0.18113953384072565,
      "learning_rate": 2.6626139036673106e-06,
      "loss": 0.0033,
      "step": 42130
    },
    {
      "epoch": 0.6895197578335924,
      "grad_norm": 0.37737101692411124,
      "learning_rate": 2.660089710059657e-06,
      "loss": 0.0038,
      "step": 42140
    },
    {
      "epoch": 0.6896833837846682,
      "grad_norm": 0.06192965118034621,
      "learning_rate": 2.657566279806355e-06,
      "loss": 0.0016,
      "step": 42150
    },
    {
      "epoch": 0.6898470097357441,
      "grad_norm": 0.2419755439107186,
      "learning_rate": 2.6550436137306283e-06,
      "loss": 0.0043,
      "step": 42160
    },
    {
      "epoch": 0.69001063568682,
      "grad_norm": 0.19497767055163226,
      "learning_rate": 2.6525217126554536e-06,
      "loss": 0.0026,
      "step": 42170
    },
    {
      "epoch": 0.6901742616378957,
      "grad_norm": 0.18707493061328978,
      "learning_rate": 2.650000577403555e-06,
      "loss": 0.0013,
      "step": 42180
    },
    {
      "epoch": 0.6903378875889716,
      "grad_norm": 0.22304668472140066,
      "learning_rate": 2.6474802087974072e-06,
      "loss": 0.0047,
      "step": 42190
    },
    {
      "epoch": 0.6905015135400474,
      "grad_norm": 0.1679943475615923,
      "learning_rate": 2.6449606076592394e-06,
      "loss": 0.0029,
      "step": 42200
    },
    {
      "epoch": 0.6906651394911233,
      "grad_norm": 0.15798290736110393,
      "learning_rate": 2.642441774811025e-06,
      "loss": 0.0021,
      "step": 42210
    },
    {
      "epoch": 0.6908287654421992,
      "grad_norm": 0.1559890354693163,
      "learning_rate": 2.639923711074489e-06,
      "loss": 0.0017,
      "step": 42220
    },
    {
      "epoch": 0.690992391393275,
      "grad_norm": 0.21856857513277864,
      "learning_rate": 2.6374064172711046e-06,
      "loss": 0.0032,
      "step": 42230
    },
    {
      "epoch": 0.6911560173443508,
      "grad_norm": 0.14487209534974688,
      "learning_rate": 2.6348898942220954e-06,
      "loss": 0.0013,
      "step": 42240
    },
    {
      "epoch": 0.6913196432954266,
      "grad_norm": 0.057600959260587864,
      "learning_rate": 2.6323741427484294e-06,
      "loss": 0.0011,
      "step": 42250
    },
    {
      "epoch": 0.6914832692465025,
      "grad_norm": 0.20340513087614323,
      "learning_rate": 2.6298591636708304e-06,
      "loss": 0.0027,
      "step": 42260
    },
    {
      "epoch": 0.6916468951975784,
      "grad_norm": 0.08370605370788999,
      "learning_rate": 2.627344957809764e-06,
      "loss": 0.0038,
      "step": 42270
    },
    {
      "epoch": 0.6918105211486542,
      "grad_norm": 0.19153242011097887,
      "learning_rate": 2.6248315259854455e-06,
      "loss": 0.0048,
      "step": 42280
    },
    {
      "epoch": 0.69197414709973,
      "grad_norm": 0.6059768206292153,
      "learning_rate": 2.6223188690178377e-06,
      "loss": 0.0057,
      "step": 42290
    },
    {
      "epoch": 0.6921377730508058,
      "grad_norm": 0.13463119945813823,
      "learning_rate": 2.6198069877266496e-06,
      "loss": 0.0028,
      "step": 42300
    },
    {
      "epoch": 0.6923013990018817,
      "grad_norm": 0.07834107458140202,
      "learning_rate": 2.6172958829313365e-06,
      "loss": 0.0024,
      "step": 42310
    },
    {
      "epoch": 0.6924650249529576,
      "grad_norm": 0.05727839177059226,
      "learning_rate": 2.614785555451107e-06,
      "loss": 0.0025,
      "step": 42320
    },
    {
      "epoch": 0.6926286509040334,
      "grad_norm": 0.25599497018985073,
      "learning_rate": 2.612276006104908e-06,
      "loss": 0.0026,
      "step": 42330
    },
    {
      "epoch": 0.6927922768551092,
      "grad_norm": 0.17929990937914794,
      "learning_rate": 2.609767235711435e-06,
      "loss": 0.0027,
      "step": 42340
    },
    {
      "epoch": 0.692955902806185,
      "grad_norm": 0.027929323260466855,
      "learning_rate": 2.607259245089131e-06,
      "loss": 0.0021,
      "step": 42350
    },
    {
      "epoch": 0.6931195287572609,
      "grad_norm": 0.033992221815899504,
      "learning_rate": 2.604752035056184e-06,
      "loss": 0.003,
      "step": 42360
    },
    {
      "epoch": 0.6932831547083368,
      "grad_norm": 0.10127130861645584,
      "learning_rate": 2.6022456064305234e-06,
      "loss": 0.0025,
      "step": 42370
    },
    {
      "epoch": 0.6934467806594126,
      "grad_norm": 0.21152248927512007,
      "learning_rate": 2.5997399600298325e-06,
      "loss": 0.0018,
      "step": 42380
    },
    {
      "epoch": 0.6936104066104885,
      "grad_norm": 0.046222316072202875,
      "learning_rate": 2.5972350966715317e-06,
      "loss": 0.0019,
      "step": 42390
    },
    {
      "epoch": 0.6937740325615642,
      "grad_norm": 0.17598986019860094,
      "learning_rate": 2.594731017172788e-06,
      "loss": 0.0028,
      "step": 42400
    },
    {
      "epoch": 0.6939376585126401,
      "grad_norm": 0.029921102258220013,
      "learning_rate": 2.592227722350513e-06,
      "loss": 0.0013,
      "step": 42410
    },
    {
      "epoch": 0.694101284463716,
      "grad_norm": 0.22065976748320415,
      "learning_rate": 2.5897252130213633e-06,
      "loss": 0.0022,
      "step": 42420
    },
    {
      "epoch": 0.6942649104147918,
      "grad_norm": 0.1778260313593099,
      "learning_rate": 2.5872234900017367e-06,
      "loss": 0.0022,
      "step": 42430
    },
    {
      "epoch": 0.6944285363658677,
      "grad_norm": 0.07997018527474191,
      "learning_rate": 2.584722554107778e-06,
      "loss": 0.0017,
      "step": 42440
    },
    {
      "epoch": 0.6945921623169434,
      "grad_norm": 0.11294514882758254,
      "learning_rate": 2.5822224061553723e-06,
      "loss": 0.0024,
      "step": 42450
    },
    {
      "epoch": 0.6947557882680193,
      "grad_norm": 0.1605883759224009,
      "learning_rate": 2.5797230469601495e-06,
      "loss": 0.0019,
      "step": 42460
    },
    {
      "epoch": 0.6949194142190952,
      "grad_norm": 0.1357951844986823,
      "learning_rate": 2.577224477337481e-06,
      "loss": 0.0037,
      "step": 42470
    },
    {
      "epoch": 0.695083040170171,
      "grad_norm": 0.09727077421465809,
      "learning_rate": 2.5747266981024797e-06,
      "loss": 0.0035,
      "step": 42480
    },
    {
      "epoch": 0.6952466661212469,
      "grad_norm": 0.07238823904652714,
      "learning_rate": 2.572229710070003e-06,
      "loss": 0.0037,
      "step": 42490
    },
    {
      "epoch": 0.6954102920723226,
      "grad_norm": 0.05850072874101029,
      "learning_rate": 2.5697335140546476e-06,
      "loss": 0.0015,
      "step": 42500
    },
    {
      "epoch": 0.6955739180233985,
      "grad_norm": 0.08006524619896124,
      "learning_rate": 2.5672381108707568e-06,
      "loss": 0.0026,
      "step": 42510
    },
    {
      "epoch": 0.6957375439744744,
      "grad_norm": 0.22582375911499258,
      "learning_rate": 2.5647435013324083e-06,
      "loss": 0.0015,
      "step": 42520
    },
    {
      "epoch": 0.6959011699255502,
      "grad_norm": 0.1593822373487001,
      "learning_rate": 2.562249686253426e-06,
      "loss": 0.0012,
      "step": 42530
    },
    {
      "epoch": 0.6960647958766261,
      "grad_norm": 0.14592473505468992,
      "learning_rate": 2.5597566664473733e-06,
      "loss": 0.0022,
      "step": 42540
    },
    {
      "epoch": 0.6962284218277018,
      "grad_norm": 0.13714616441540808,
      "learning_rate": 2.5572644427275526e-06,
      "loss": 0.0044,
      "step": 42550
    },
    {
      "epoch": 0.6963920477787777,
      "grad_norm": 0.07518848826728511,
      "learning_rate": 2.554773015907007e-06,
      "loss": 0.002,
      "step": 42560
    },
    {
      "epoch": 0.6965556737298536,
      "grad_norm": 0.39159404842972634,
      "learning_rate": 2.5522823867985225e-06,
      "loss": 0.0025,
      "step": 42570
    },
    {
      "epoch": 0.6967192996809294,
      "grad_norm": 0.1929937804922995,
      "learning_rate": 2.549792556214623e-06,
      "loss": 0.0036,
      "step": 42580
    },
    {
      "epoch": 0.6968829256320053,
      "grad_norm": 0.08372516332858088,
      "learning_rate": 2.54730352496757e-06,
      "loss": 0.0027,
      "step": 42590
    },
    {
      "epoch": 0.697046551583081,
      "grad_norm": 0.26972807598467047,
      "learning_rate": 2.544815293869368e-06,
      "loss": 0.0014,
      "step": 42600
    },
    {
      "epoch": 0.6972101775341569,
      "grad_norm": 0.06408195803365016,
      "learning_rate": 2.542327863731756e-06,
      "loss": 0.0031,
      "step": 42610
    },
    {
      "epoch": 0.6973738034852328,
      "grad_norm": 0.14703981151765072,
      "learning_rate": 2.539841235366214e-06,
      "loss": 0.0016,
      "step": 42620
    },
    {
      "epoch": 0.6975374294363086,
      "grad_norm": 0.3291114015317981,
      "learning_rate": 2.5373554095839627e-06,
      "loss": 0.0022,
      "step": 42630
    },
    {
      "epoch": 0.6977010553873845,
      "grad_norm": 0.0974522599479989,
      "learning_rate": 2.5348703871959584e-06,
      "loss": 0.0022,
      "step": 42640
    },
    {
      "epoch": 0.6978646813384602,
      "grad_norm": 0.08170836080619386,
      "learning_rate": 2.5323861690128946e-06,
      "loss": 0.0035,
      "step": 42650
    },
    {
      "epoch": 0.6980283072895361,
      "grad_norm": 0.3298263758887845,
      "learning_rate": 2.529902755845205e-06,
      "loss": 0.0041,
      "step": 42660
    },
    {
      "epoch": 0.6981919332406119,
      "grad_norm": 0.24360244605998008,
      "learning_rate": 2.527420148503058e-06,
      "loss": 0.0017,
      "step": 42670
    },
    {
      "epoch": 0.6983555591916878,
      "grad_norm": 0.12323794976042991,
      "learning_rate": 2.5249383477963584e-06,
      "loss": 0.0021,
      "step": 42680
    },
    {
      "epoch": 0.6985191851427637,
      "grad_norm": 0.08182833676286878,
      "learning_rate": 2.522457354534754e-06,
      "loss": 0.0023,
      "step": 42690
    },
    {
      "epoch": 0.6986828110938395,
      "grad_norm": 0.1596435859906205,
      "learning_rate": 2.5199771695276233e-06,
      "loss": 0.0014,
      "step": 42700
    },
    {
      "epoch": 0.6988464370449153,
      "grad_norm": 0.2531167413612979,
      "learning_rate": 2.5174977935840817e-06,
      "loss": 0.0032,
      "step": 42710
    },
    {
      "epoch": 0.6990100629959911,
      "grad_norm": 0.01499691021840432,
      "learning_rate": 2.515019227512983e-06,
      "loss": 0.0015,
      "step": 42720
    },
    {
      "epoch": 0.699173688947067,
      "grad_norm": 0.24050260370449672,
      "learning_rate": 2.5125414721229152e-06,
      "loss": 0.0026,
      "step": 42730
    },
    {
      "epoch": 0.6993373148981429,
      "grad_norm": 0.008165072452152035,
      "learning_rate": 2.5100645282222004e-06,
      "loss": 0.0024,
      "step": 42740
    },
    {
      "epoch": 0.6995009408492187,
      "grad_norm": 0.213110493539859,
      "learning_rate": 2.507588396618901e-06,
      "loss": 0.0019,
      "step": 42750
    },
    {
      "epoch": 0.6996645668002945,
      "grad_norm": 0.006691400971860885,
      "learning_rate": 2.5051130781208095e-06,
      "loss": 0.0017,
      "step": 42760
    },
    {
      "epoch": 0.6998281927513703,
      "grad_norm": 0.16603569730029816,
      "learning_rate": 2.5026385735354553e-06,
      "loss": 0.0025,
      "step": 42770
    },
    {
      "epoch": 0.6999918187024462,
      "grad_norm": 0.1240409447365724,
      "learning_rate": 2.500164883670101e-06,
      "loss": 0.0023,
      "step": 42780
    },
    {
      "epoch": 0.7001554446535221,
      "grad_norm": 0.14596170231879715,
      "learning_rate": 2.4976920093317442e-06,
      "loss": 0.0013,
      "step": 42790
    },
    {
      "epoch": 0.7003190706045979,
      "grad_norm": 0.06703826203191914,
      "learning_rate": 2.495219951327117e-06,
      "loss": 0.002,
      "step": 42800
    },
    {
      "epoch": 0.7004826965556737,
      "grad_norm": 0.30908993348752073,
      "learning_rate": 2.4927487104626826e-06,
      "loss": 0.0025,
      "step": 42810
    },
    {
      "epoch": 0.7006463225067495,
      "grad_norm": 0.03648786122586192,
      "learning_rate": 2.4902782875446424e-06,
      "loss": 0.0031,
      "step": 42820
    },
    {
      "epoch": 0.7008099484578254,
      "grad_norm": 0.11710588033916008,
      "learning_rate": 2.48780868337893e-06,
      "loss": 0.0018,
      "step": 42830
    },
    {
      "epoch": 0.7009735744089013,
      "grad_norm": 0.11180839243936214,
      "learning_rate": 2.485339898771203e-06,
      "loss": 0.0019,
      "step": 42840
    },
    {
      "epoch": 0.7011372003599771,
      "grad_norm": 0.12161946187742793,
      "learning_rate": 2.4828719345268647e-06,
      "loss": 0.0024,
      "step": 42850
    },
    {
      "epoch": 0.701300826311053,
      "grad_norm": 0.33568839966015046,
      "learning_rate": 2.480404791451043e-06,
      "loss": 0.0026,
      "step": 42860
    },
    {
      "epoch": 0.7014644522621287,
      "grad_norm": 0.06373395615617619,
      "learning_rate": 2.477938470348598e-06,
      "loss": 0.0027,
      "step": 42870
    },
    {
      "epoch": 0.7016280782132046,
      "grad_norm": 0.25675438810550766,
      "learning_rate": 2.4754729720241295e-06,
      "loss": 0.0033,
      "step": 42880
    },
    {
      "epoch": 0.7017917041642805,
      "grad_norm": 0.1859951100071559,
      "learning_rate": 2.473008297281955e-06,
      "loss": 0.0022,
      "step": 42890
    },
    {
      "epoch": 0.7019553301153563,
      "grad_norm": 0.13239680154073993,
      "learning_rate": 2.4705444469261335e-06,
      "loss": 0.0031,
      "step": 42900
    },
    {
      "epoch": 0.7021189560664322,
      "grad_norm": 0.13048337605727764,
      "learning_rate": 2.4680814217604547e-06,
      "loss": 0.0018,
      "step": 42910
    },
    {
      "epoch": 0.7022825820175079,
      "grad_norm": 0.04740025112416992,
      "learning_rate": 2.465619222588436e-06,
      "loss": 0.0025,
      "step": 42920
    },
    {
      "epoch": 0.7024462079685838,
      "grad_norm": 0.09142486136667569,
      "learning_rate": 2.4631578502133246e-06,
      "loss": 0.0015,
      "step": 42930
    },
    {
      "epoch": 0.7026098339196597,
      "grad_norm": 0.15076886995063912,
      "learning_rate": 2.460697305438105e-06,
      "loss": 0.0033,
      "step": 42940
    },
    {
      "epoch": 0.7027734598707355,
      "grad_norm": 0.07234335583239838,
      "learning_rate": 2.45823758906548e-06,
      "loss": 0.0024,
      "step": 42950
    },
    {
      "epoch": 0.7029370858218114,
      "grad_norm": 0.2102063393373935,
      "learning_rate": 2.4557787018978896e-06,
      "loss": 0.0022,
      "step": 42960
    },
    {
      "epoch": 0.7031007117728871,
      "grad_norm": 0.1252135179822643,
      "learning_rate": 2.453320644737506e-06,
      "loss": 0.0016,
      "step": 42970
    },
    {
      "epoch": 0.703264337723963,
      "grad_norm": 0.14203836993027255,
      "learning_rate": 2.4508634183862243e-06,
      "loss": 0.003,
      "step": 42980
    },
    {
      "epoch": 0.7034279636750389,
      "grad_norm": 0.12570227124008562,
      "learning_rate": 2.44840702364567e-06,
      "loss": 0.0044,
      "step": 42990
    },
    {
      "epoch": 0.7035915896261147,
      "grad_norm": 0.10927408205950763,
      "learning_rate": 2.4459514613172036e-06,
      "loss": 0.0011,
      "step": 43000
    },
    {
      "epoch": 0.7037552155771906,
      "grad_norm": 0.15512188159669973,
      "learning_rate": 2.443496732201903e-06,
      "loss": 0.0023,
      "step": 43010
    },
    {
      "epoch": 0.7039188415282663,
      "grad_norm": 0.08240699106798485,
      "learning_rate": 2.4410428371005812e-06,
      "loss": 0.0017,
      "step": 43020
    },
    {
      "epoch": 0.7040824674793422,
      "grad_norm": 0.06914727288266999,
      "learning_rate": 2.4385897768137806e-06,
      "loss": 0.0018,
      "step": 43030
    },
    {
      "epoch": 0.7042460934304181,
      "grad_norm": 0.10836406557590571,
      "learning_rate": 2.436137552141768e-06,
      "loss": 0.0048,
      "step": 43040
    },
    {
      "epoch": 0.7044097193814939,
      "grad_norm": 0.12569508329837245,
      "learning_rate": 2.4336861638845353e-06,
      "loss": 0.002,
      "step": 43050
    },
    {
      "epoch": 0.7045733453325698,
      "grad_norm": 0.09387843603584835,
      "learning_rate": 2.431235612841811e-06,
      "loss": 0.0041,
      "step": 43060
    },
    {
      "epoch": 0.7047369712836455,
      "grad_norm": 0.045913252266143366,
      "learning_rate": 2.4287858998130387e-06,
      "loss": 0.0023,
      "step": 43070
    },
    {
      "epoch": 0.7049005972347214,
      "grad_norm": 0.058679956327134374,
      "learning_rate": 2.426337025597393e-06,
      "loss": 0.002,
      "step": 43080
    },
    {
      "epoch": 0.7050642231857973,
      "grad_norm": 0.06851702085204948,
      "learning_rate": 2.423888990993781e-06,
      "loss": 0.0018,
      "step": 43090
    },
    {
      "epoch": 0.7052278491368731,
      "grad_norm": 0.12662128009270104,
      "learning_rate": 2.4214417968008263e-06,
      "loss": 0.0024,
      "step": 43100
    },
    {
      "epoch": 0.705391475087949,
      "grad_norm": 0.2661755365389091,
      "learning_rate": 2.4189954438168834e-06,
      "loss": 0.0028,
      "step": 43110
    },
    {
      "epoch": 0.7055551010390247,
      "grad_norm": 0.44854412174273567,
      "learning_rate": 2.416549932840036e-06,
      "loss": 0.0022,
      "step": 43120
    },
    {
      "epoch": 0.7057187269901006,
      "grad_norm": 0.013531746978043236,
      "learning_rate": 2.4141052646680836e-06,
      "loss": 0.0027,
      "step": 43130
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.3333458081072214,
      "learning_rate": 2.411661440098558e-06,
      "loss": 0.0014,
      "step": 43140
    },
    {
      "epoch": 0.7060459788922523,
      "grad_norm": 0.056146300361681764,
      "learning_rate": 2.409218459928711e-06,
      "loss": 0.0027,
      "step": 43150
    },
    {
      "epoch": 0.7062096048433282,
      "grad_norm": 0.1947461125292986,
      "learning_rate": 2.4067763249555265e-06,
      "loss": 0.002,
      "step": 43160
    },
    {
      "epoch": 0.706373230794404,
      "grad_norm": 0.12745218924730045,
      "learning_rate": 2.404335035975705e-06,
      "loss": 0.0021,
      "step": 43170
    },
    {
      "epoch": 0.7065368567454798,
      "grad_norm": 0.08093980793897779,
      "learning_rate": 2.401894593785675e-06,
      "loss": 0.0035,
      "step": 43180
    },
    {
      "epoch": 0.7067004826965557,
      "grad_norm": 0.05602852887606721,
      "learning_rate": 2.399454999181587e-06,
      "loss": 0.0021,
      "step": 43190
    },
    {
      "epoch": 0.7068641086476315,
      "grad_norm": 0.16011167735956833,
      "learning_rate": 2.397016252959315e-06,
      "loss": 0.0039,
      "step": 43200
    },
    {
      "epoch": 0.7070277345987074,
      "grad_norm": 0.23870543082287238,
      "learning_rate": 2.3945783559144558e-06,
      "loss": 0.0039,
      "step": 43210
    },
    {
      "epoch": 0.7071913605497832,
      "grad_norm": 0.17849197238636635,
      "learning_rate": 2.3921413088423335e-06,
      "loss": 0.0035,
      "step": 43220
    },
    {
      "epoch": 0.707354986500859,
      "grad_norm": 0.08108283362383914,
      "learning_rate": 2.38970511253799e-06,
      "loss": 0.0081,
      "step": 43230
    },
    {
      "epoch": 0.7075186124519349,
      "grad_norm": 0.10638935923987021,
      "learning_rate": 2.3872697677961916e-06,
      "loss": 0.0021,
      "step": 43240
    },
    {
      "epoch": 0.7076822384030107,
      "grad_norm": 0.1957088501451425,
      "learning_rate": 2.3848352754114257e-06,
      "loss": 0.0016,
      "step": 43250
    },
    {
      "epoch": 0.7078458643540866,
      "grad_norm": 0.04746579754730785,
      "learning_rate": 2.382401636177902e-06,
      "loss": 0.0029,
      "step": 43260
    },
    {
      "epoch": 0.7080094903051624,
      "grad_norm": 0.09525427884736594,
      "learning_rate": 2.3799688508895523e-06,
      "loss": 0.0021,
      "step": 43270
    },
    {
      "epoch": 0.7081731162562382,
      "grad_norm": 0.193376770408006,
      "learning_rate": 2.377536920340032e-06,
      "loss": 0.0017,
      "step": 43280
    },
    {
      "epoch": 0.7083367422073141,
      "grad_norm": 0.005833740717229194,
      "learning_rate": 2.3751058453227135e-06,
      "loss": 0.0019,
      "step": 43290
    },
    {
      "epoch": 0.7085003681583899,
      "grad_norm": 0.15113952709746847,
      "learning_rate": 2.3726756266306923e-06,
      "loss": 0.0018,
      "step": 43300
    },
    {
      "epoch": 0.7086639941094658,
      "grad_norm": 0.12522710726632713,
      "learning_rate": 2.370246265056785e-06,
      "loss": 0.0034,
      "step": 43310
    },
    {
      "epoch": 0.7088276200605416,
      "grad_norm": 0.15463162496647995,
      "learning_rate": 2.3678177613935264e-06,
      "loss": 0.0017,
      "step": 43320
    },
    {
      "epoch": 0.7089912460116174,
      "grad_norm": 0.14780219613726406,
      "learning_rate": 2.365390116433172e-06,
      "loss": 0.0027,
      "step": 43330
    },
    {
      "epoch": 0.7091548719626933,
      "grad_norm": 0.19317825983289047,
      "learning_rate": 2.3629633309677024e-06,
      "loss": 0.005,
      "step": 43340
    },
    {
      "epoch": 0.7093184979137691,
      "grad_norm": 0.22066408561235396,
      "learning_rate": 2.36053740578881e-06,
      "loss": 0.0045,
      "step": 43350
    },
    {
      "epoch": 0.709482123864845,
      "grad_norm": 0.04282996708322745,
      "learning_rate": 2.3581123416879107e-06,
      "loss": 0.0017,
      "step": 43360
    },
    {
      "epoch": 0.7096457498159208,
      "grad_norm": 0.0278570579881955,
      "learning_rate": 2.355688139456139e-06,
      "loss": 0.0017,
      "step": 43370
    },
    {
      "epoch": 0.7098093757669967,
      "grad_norm": 0.06375907651803764,
      "learning_rate": 2.3532647998843484e-06,
      "loss": 0.0021,
      "step": 43380
    },
    {
      "epoch": 0.7099730017180725,
      "grad_norm": 0.2699096096749135,
      "learning_rate": 2.350842323763108e-06,
      "loss": 0.0022,
      "step": 43390
    },
    {
      "epoch": 0.7101366276691483,
      "grad_norm": 0.3456814471074755,
      "learning_rate": 2.3484207118827113e-06,
      "loss": 0.004,
      "step": 43400
    },
    {
      "epoch": 0.7103002536202242,
      "grad_norm": 0.1759268920962095,
      "learning_rate": 2.345999965033165e-06,
      "loss": 0.0032,
      "step": 43410
    },
    {
      "epoch": 0.7104638795713,
      "grad_norm": 0.1529787854355532,
      "learning_rate": 2.343580084004195e-06,
      "loss": 0.0025,
      "step": 43420
    },
    {
      "epoch": 0.7106275055223759,
      "grad_norm": 0.05528188598290255,
      "learning_rate": 2.3411610695852445e-06,
      "loss": 0.0033,
      "step": 43430
    },
    {
      "epoch": 0.7107911314734517,
      "grad_norm": 0.06423704928730894,
      "learning_rate": 2.338742922565474e-06,
      "loss": 0.0014,
      "step": 43440
    },
    {
      "epoch": 0.7109547574245275,
      "grad_norm": 0.013826571404669113,
      "learning_rate": 2.3363256437337596e-06,
      "loss": 0.0018,
      "step": 43450
    },
    {
      "epoch": 0.7111183833756034,
      "grad_norm": 0.2217710459788391,
      "learning_rate": 2.333909233878699e-06,
      "loss": 0.0023,
      "step": 43460
    },
    {
      "epoch": 0.7112820093266792,
      "grad_norm": 0.2506381107981868,
      "learning_rate": 2.3314936937886024e-06,
      "loss": 0.0017,
      "step": 43470
    },
    {
      "epoch": 0.7114456352777551,
      "grad_norm": 0.17950313337282317,
      "learning_rate": 2.329079024251496e-06,
      "loss": 0.0041,
      "step": 43480
    },
    {
      "epoch": 0.711609261228831,
      "grad_norm": 0.15253353251779353,
      "learning_rate": 2.3266652260551236e-06,
      "loss": 0.0025,
      "step": 43490
    },
    {
      "epoch": 0.7117728871799067,
      "grad_norm": 0.20537571719409897,
      "learning_rate": 2.324252299986944e-06,
      "loss": 0.0015,
      "step": 43500
    },
    {
      "epoch": 0.7119365131309826,
      "grad_norm": 0.2387573361634284,
      "learning_rate": 2.321840246834132e-06,
      "loss": 0.0015,
      "step": 43510
    },
    {
      "epoch": 0.7121001390820584,
      "grad_norm": 0.20600349275903465,
      "learning_rate": 2.319429067383574e-06,
      "loss": 0.0029,
      "step": 43520
    },
    {
      "epoch": 0.7122637650331343,
      "grad_norm": 0.15537405002499002,
      "learning_rate": 2.3170187624218797e-06,
      "loss": 0.0028,
      "step": 43530
    },
    {
      "epoch": 0.71242739098421,
      "grad_norm": 0.09589913480007048,
      "learning_rate": 2.314609332735367e-06,
      "loss": 0.0024,
      "step": 43540
    },
    {
      "epoch": 0.7125910169352859,
      "grad_norm": 0.17927289651980752,
      "learning_rate": 2.3122007791100686e-06,
      "loss": 0.002,
      "step": 43550
    },
    {
      "epoch": 0.7127546428863618,
      "grad_norm": 0.10984114397872247,
      "learning_rate": 2.3097931023317333e-06,
      "loss": 0.0017,
      "step": 43560
    },
    {
      "epoch": 0.7129182688374376,
      "grad_norm": 0.08926713175592099,
      "learning_rate": 2.307386303185823e-06,
      "loss": 0.0023,
      "step": 43570
    },
    {
      "epoch": 0.7130818947885135,
      "grad_norm": 0.06438544351367055,
      "learning_rate": 2.3049803824575106e-06,
      "loss": 0.0022,
      "step": 43580
    },
    {
      "epoch": 0.7132455207395892,
      "grad_norm": 0.1881532308514187,
      "learning_rate": 2.3025753409316898e-06,
      "loss": 0.0027,
      "step": 43590
    },
    {
      "epoch": 0.7134091466906651,
      "grad_norm": 0.15189753840136275,
      "learning_rate": 2.300171179392961e-06,
      "loss": 0.0024,
      "step": 43600
    },
    {
      "epoch": 0.713572772641741,
      "grad_norm": 0.06675455958498402,
      "learning_rate": 2.297767898625639e-06,
      "loss": 0.004,
      "step": 43610
    },
    {
      "epoch": 0.7137363985928168,
      "grad_norm": 0.14670566170918492,
      "learning_rate": 2.295365499413752e-06,
      "loss": 0.0024,
      "step": 43620
    },
    {
      "epoch": 0.7139000245438927,
      "grad_norm": 0.0361588632529197,
      "learning_rate": 2.29296398254104e-06,
      "loss": 0.0042,
      "step": 43630
    },
    {
      "epoch": 0.7140636504949684,
      "grad_norm": 0.06195784555230843,
      "learning_rate": 2.2905633487909536e-06,
      "loss": 0.0024,
      "step": 43640
    },
    {
      "epoch": 0.7142272764460443,
      "grad_norm": 0.22346359704624974,
      "learning_rate": 2.2881635989466604e-06,
      "loss": 0.0034,
      "step": 43650
    },
    {
      "epoch": 0.7143909023971202,
      "grad_norm": 0.4799521223345969,
      "learning_rate": 2.2857647337910357e-06,
      "loss": 0.0022,
      "step": 43660
    },
    {
      "epoch": 0.714554528348196,
      "grad_norm": 0.1967031112667705,
      "learning_rate": 2.283366754106665e-06,
      "loss": 0.0019,
      "step": 43670
    },
    {
      "epoch": 0.7147181542992719,
      "grad_norm": 0.056401445992154894,
      "learning_rate": 2.280969660675849e-06,
      "loss": 0.0027,
      "step": 43680
    },
    {
      "epoch": 0.7148817802503477,
      "grad_norm": 0.09302699789301122,
      "learning_rate": 2.2785734542805955e-06,
      "loss": 0.0018,
      "step": 43690
    },
    {
      "epoch": 0.7150454062014235,
      "grad_norm": 0.1151691221586268,
      "learning_rate": 2.2761781357026226e-06,
      "loss": 0.0031,
      "step": 43700
    },
    {
      "epoch": 0.7152090321524994,
      "grad_norm": 0.13517416407079014,
      "learning_rate": 2.273783705723365e-06,
      "loss": 0.0017,
      "step": 43710
    },
    {
      "epoch": 0.7153726581035752,
      "grad_norm": 0.019787468378722174,
      "learning_rate": 2.2713901651239613e-06,
      "loss": 0.0021,
      "step": 43720
    },
    {
      "epoch": 0.7155362840546511,
      "grad_norm": 0.06635075822986672,
      "learning_rate": 2.268997514685261e-06,
      "loss": 0.0018,
      "step": 43730
    },
    {
      "epoch": 0.7156999100057269,
      "grad_norm": 0.13965861928611048,
      "learning_rate": 2.2666057551878247e-06,
      "loss": 0.0031,
      "step": 43740
    },
    {
      "epoch": 0.7158635359568027,
      "grad_norm": 0.14232147466305387,
      "learning_rate": 2.264214887411921e-06,
      "loss": 0.0017,
      "step": 43750
    },
    {
      "epoch": 0.7160271619078786,
      "grad_norm": 0.23181736420671092,
      "learning_rate": 2.2618249121375275e-06,
      "loss": 0.0024,
      "step": 43760
    },
    {
      "epoch": 0.7161907878589544,
      "grad_norm": 0.28469485245830134,
      "learning_rate": 2.2594358301443336e-06,
      "loss": 0.0014,
      "step": 43770
    },
    {
      "epoch": 0.7163544138100303,
      "grad_norm": 0.33087670695465904,
      "learning_rate": 2.257047642211735e-06,
      "loss": 0.0025,
      "step": 43780
    },
    {
      "epoch": 0.7165180397611061,
      "grad_norm": 0.19231458849622177,
      "learning_rate": 2.2546603491188356e-06,
      "loss": 0.0017,
      "step": 43790
    },
    {
      "epoch": 0.716681665712182,
      "grad_norm": 0.0997182112424718,
      "learning_rate": 2.2522739516444435e-06,
      "loss": 0.0028,
      "step": 43800
    },
    {
      "epoch": 0.7168452916632578,
      "grad_norm": 0.126665187207932,
      "learning_rate": 2.2498884505670836e-06,
      "loss": 0.0029,
      "step": 43810
    },
    {
      "epoch": 0.7170089176143336,
      "grad_norm": 0.01911342995733695,
      "learning_rate": 2.2475038466649817e-06,
      "loss": 0.0021,
      "step": 43820
    },
    {
      "epoch": 0.7171725435654095,
      "grad_norm": 0.21444305364064356,
      "learning_rate": 2.2451201407160703e-06,
      "loss": 0.0027,
      "step": 43830
    },
    {
      "epoch": 0.7173361695164853,
      "grad_norm": 0.26301214500054393,
      "learning_rate": 2.242737333497997e-06,
      "loss": 0.0039,
      "step": 43840
    },
    {
      "epoch": 0.7174997954675612,
      "grad_norm": 0.047847911797802664,
      "learning_rate": 2.2403554257881076e-06,
      "loss": 0.0018,
      "step": 43850
    },
    {
      "epoch": 0.717663421418637,
      "grad_norm": 0.24435002096423136,
      "learning_rate": 2.237974418363454e-06,
      "loss": 0.0017,
      "step": 43860
    },
    {
      "epoch": 0.7178270473697128,
      "grad_norm": 0.091333669138778,
      "learning_rate": 2.235594312000802e-06,
      "loss": 0.0017,
      "step": 43870
    },
    {
      "epoch": 0.7179906733207887,
      "grad_norm": 0.3037204446614908,
      "learning_rate": 2.2332151074766186e-06,
      "loss": 0.0027,
      "step": 43880
    },
    {
      "epoch": 0.7181542992718645,
      "grad_norm": 0.0567493362066839,
      "learning_rate": 2.2308368055670745e-06,
      "loss": 0.003,
      "step": 43890
    },
    {
      "epoch": 0.7183179252229404,
      "grad_norm": 0.08440151226310144,
      "learning_rate": 2.228459407048052e-06,
      "loss": 0.0027,
      "step": 43900
    },
    {
      "epoch": 0.7184815511740162,
      "grad_norm": 0.0063384282290316115,
      "learning_rate": 2.2260829126951362e-06,
      "loss": 0.0016,
      "step": 43910
    },
    {
      "epoch": 0.718645177125092,
      "grad_norm": 0.3099517528169145,
      "learning_rate": 2.2237073232836103e-06,
      "loss": 0.0016,
      "step": 43920
    },
    {
      "epoch": 0.7188088030761679,
      "grad_norm": 0.17749750184925078,
      "learning_rate": 2.221332639588474e-06,
      "loss": 0.0035,
      "step": 43930
    },
    {
      "epoch": 0.7189724290272437,
      "grad_norm": 0.05835565075511775,
      "learning_rate": 2.2189588623844245e-06,
      "loss": 0.0011,
      "step": 43940
    },
    {
      "epoch": 0.7191360549783196,
      "grad_norm": 0.10807162065007092,
      "learning_rate": 2.2165859924458617e-06,
      "loss": 0.002,
      "step": 43950
    },
    {
      "epoch": 0.7192996809293954,
      "grad_norm": 0.04893275895119157,
      "learning_rate": 2.214214030546897e-06,
      "loss": 0.0015,
      "step": 43960
    },
    {
      "epoch": 0.7194633068804712,
      "grad_norm": 0.19951019309564152,
      "learning_rate": 2.211842977461341e-06,
      "loss": 0.0026,
      "step": 43970
    },
    {
      "epoch": 0.7196269328315471,
      "grad_norm": 0.11636389971016477,
      "learning_rate": 2.209472833962702e-06,
      "loss": 0.0026,
      "step": 43980
    },
    {
      "epoch": 0.7197905587826229,
      "grad_norm": 0.09482053466583638,
      "learning_rate": 2.2071036008242032e-06,
      "loss": 0.0011,
      "step": 43990
    },
    {
      "epoch": 0.7199541847336988,
      "grad_norm": 0.14965693451034104,
      "learning_rate": 2.2047352788187625e-06,
      "loss": 0.0067,
      "step": 44000
    },
    {
      "epoch": 0.7201178106847747,
      "grad_norm": 0.18236496754725728,
      "learning_rate": 2.202367868719002e-06,
      "loss": 0.0019,
      "step": 44010
    },
    {
      "epoch": 0.7202814366358504,
      "grad_norm": 0.06773068688590426,
      "learning_rate": 2.200001371297251e-06,
      "loss": 0.002,
      "step": 44020
    },
    {
      "epoch": 0.7204450625869263,
      "grad_norm": 0.0680912365484517,
      "learning_rate": 2.197635787325537e-06,
      "loss": 0.01,
      "step": 44030
    },
    {
      "epoch": 0.7206086885380021,
      "grad_norm": 0.10737250671488516,
      "learning_rate": 2.195271117575585e-06,
      "loss": 0.0022,
      "step": 44040
    },
    {
      "epoch": 0.720772314489078,
      "grad_norm": 0.06894446287476105,
      "learning_rate": 2.1929073628188314e-06,
      "loss": 0.0028,
      "step": 44050
    },
    {
      "epoch": 0.7209359404401539,
      "grad_norm": 0.09155080887020478,
      "learning_rate": 2.190544523826408e-06,
      "loss": 0.0015,
      "step": 44060
    },
    {
      "epoch": 0.7210995663912296,
      "grad_norm": 0.08955388247423478,
      "learning_rate": 2.1881826013691475e-06,
      "loss": 0.0031,
      "step": 44070
    },
    {
      "epoch": 0.7212631923423055,
      "grad_norm": 0.08151558600167949,
      "learning_rate": 2.1858215962175887e-06,
      "loss": 0.0028,
      "step": 44080
    },
    {
      "epoch": 0.7214268182933813,
      "grad_norm": 0.12270750869330221,
      "learning_rate": 2.183461509141968e-06,
      "loss": 0.0017,
      "step": 44090
    },
    {
      "epoch": 0.7215904442444572,
      "grad_norm": 0.1086911873077073,
      "learning_rate": 2.1811023409122174e-06,
      "loss": 0.0029,
      "step": 44100
    },
    {
      "epoch": 0.7217540701955331,
      "grad_norm": 0.13031915620520185,
      "learning_rate": 2.1787440922979776e-06,
      "loss": 0.0025,
      "step": 44110
    },
    {
      "epoch": 0.7219176961466088,
      "grad_norm": 0.1918384283654667,
      "learning_rate": 2.176386764068584e-06,
      "loss": 0.0026,
      "step": 44120
    },
    {
      "epoch": 0.7220813220976847,
      "grad_norm": 0.08215371381681229,
      "learning_rate": 2.174030356993075e-06,
      "loss": 0.002,
      "step": 44130
    },
    {
      "epoch": 0.7222449480487605,
      "grad_norm": 0.022897038373005066,
      "learning_rate": 2.1716748718401833e-06,
      "loss": 0.002,
      "step": 44140
    },
    {
      "epoch": 0.7224085739998364,
      "grad_norm": 0.21844538237668543,
      "learning_rate": 2.1693203093783497e-06,
      "loss": 0.0018,
      "step": 44150
    },
    {
      "epoch": 0.7225721999509123,
      "grad_norm": 0.28596291092602744,
      "learning_rate": 2.1669666703757038e-06,
      "loss": 0.0034,
      "step": 44160
    },
    {
      "epoch": 0.722735825901988,
      "grad_norm": 0.06619562484565944,
      "learning_rate": 2.1646139556000783e-06,
      "loss": 0.0026,
      "step": 44170
    },
    {
      "epoch": 0.7228994518530639,
      "grad_norm": 0.0711077445365739,
      "learning_rate": 2.162262165819009e-06,
      "loss": 0.0017,
      "step": 44180
    },
    {
      "epoch": 0.7230630778041397,
      "grad_norm": 0.08008297807818518,
      "learning_rate": 2.1599113017997233e-06,
      "loss": 0.0024,
      "step": 44190
    },
    {
      "epoch": 0.7232267037552156,
      "grad_norm": 0.15186561774405344,
      "learning_rate": 2.1575613643091487e-06,
      "loss": 0.0017,
      "step": 44200
    },
    {
      "epoch": 0.7233903297062915,
      "grad_norm": 0.2122531406948828,
      "learning_rate": 2.1552123541139114e-06,
      "loss": 0.0021,
      "step": 44210
    },
    {
      "epoch": 0.7235539556573672,
      "grad_norm": 0.10312562754713221,
      "learning_rate": 2.1528642719803346e-06,
      "loss": 0.0012,
      "step": 44220
    },
    {
      "epoch": 0.7237175816084431,
      "grad_norm": 0.10026909434049311,
      "learning_rate": 2.150517118674436e-06,
      "loss": 0.0023,
      "step": 44230
    },
    {
      "epoch": 0.7238812075595189,
      "grad_norm": 0.0732554146837231,
      "learning_rate": 2.1481708949619373e-06,
      "loss": 0.0021,
      "step": 44240
    },
    {
      "epoch": 0.7240448335105948,
      "grad_norm": 0.051036274644266494,
      "learning_rate": 2.14582560160825e-06,
      "loss": 0.0029,
      "step": 44250
    },
    {
      "epoch": 0.7242084594616707,
      "grad_norm": 0.10978618056921247,
      "learning_rate": 2.143481239378485e-06,
      "loss": 0.0021,
      "step": 44260
    },
    {
      "epoch": 0.7243720854127464,
      "grad_norm": 0.04916092976035135,
      "learning_rate": 2.1411378090374486e-06,
      "loss": 0.0023,
      "step": 44270
    },
    {
      "epoch": 0.7245357113638223,
      "grad_norm": 0.11548845765242653,
      "learning_rate": 2.138795311349644e-06,
      "loss": 0.0023,
      "step": 44280
    },
    {
      "epoch": 0.7246993373148981,
      "grad_norm": 0.1926979081538654,
      "learning_rate": 2.1364537470792673e-06,
      "loss": 0.0024,
      "step": 44290
    },
    {
      "epoch": 0.724862963265974,
      "grad_norm": 0.24763817412954164,
      "learning_rate": 2.134113116990217e-06,
      "loss": 0.0016,
      "step": 44300
    },
    {
      "epoch": 0.7250265892170499,
      "grad_norm": 0.10005298246392416,
      "learning_rate": 2.1317734218460793e-06,
      "loss": 0.0019,
      "step": 44310
    },
    {
      "epoch": 0.7251902151681257,
      "grad_norm": 0.2012219429862326,
      "learning_rate": 2.1294346624101393e-06,
      "loss": 0.0022,
      "step": 44320
    },
    {
      "epoch": 0.7253538411192015,
      "grad_norm": 0.08353068461496911,
      "learning_rate": 2.127096839445375e-06,
      "loss": 0.0029,
      "step": 44330
    },
    {
      "epoch": 0.7255174670702773,
      "grad_norm": 0.11564180946661307,
      "learning_rate": 2.1247599537144607e-06,
      "loss": 0.0013,
      "step": 44340
    },
    {
      "epoch": 0.7256810930213532,
      "grad_norm": 0.1005973123267607,
      "learning_rate": 2.1224240059797617e-06,
      "loss": 0.0017,
      "step": 44350
    },
    {
      "epoch": 0.7258447189724291,
      "grad_norm": 0.12810613187539827,
      "learning_rate": 2.1200889970033433e-06,
      "loss": 0.0019,
      "step": 44360
    },
    {
      "epoch": 0.7260083449235049,
      "grad_norm": 0.22320887556695676,
      "learning_rate": 2.1177549275469596e-06,
      "loss": 0.0023,
      "step": 44370
    },
    {
      "epoch": 0.7261719708745807,
      "grad_norm": 0.03672780886869358,
      "learning_rate": 2.1154217983720586e-06,
      "loss": 0.0012,
      "step": 44380
    },
    {
      "epoch": 0.7263355968256565,
      "grad_norm": 0.15200116430435626,
      "learning_rate": 2.113089610239783e-06,
      "loss": 0.0022,
      "step": 44390
    },
    {
      "epoch": 0.7264992227767324,
      "grad_norm": 0.1773420716724589,
      "learning_rate": 2.110758363910968e-06,
      "loss": 0.0019,
      "step": 44400
    },
    {
      "epoch": 0.7266628487278082,
      "grad_norm": 0.14074458289167532,
      "learning_rate": 2.108428060146139e-06,
      "loss": 0.0034,
      "step": 44410
    },
    {
      "epoch": 0.7268264746788841,
      "grad_norm": 0.10200750779468538,
      "learning_rate": 2.106098699705521e-06,
      "loss": 0.0021,
      "step": 44420
    },
    {
      "epoch": 0.72699010062996,
      "grad_norm": 0.4328796941513089,
      "learning_rate": 2.103770283349024e-06,
      "loss": 0.0033,
      "step": 44430
    },
    {
      "epoch": 0.7271537265810357,
      "grad_norm": 0.22376031920537298,
      "learning_rate": 2.1014428118362535e-06,
      "loss": 0.0023,
      "step": 44440
    },
    {
      "epoch": 0.7273173525321116,
      "grad_norm": 0.05157085891916257,
      "learning_rate": 2.099116285926505e-06,
      "loss": 0.0025,
      "step": 44450
    },
    {
      "epoch": 0.7274809784831874,
      "grad_norm": 0.2098957254780201,
      "learning_rate": 2.096790706378767e-06,
      "loss": 0.002,
      "step": 44460
    },
    {
      "epoch": 0.7276446044342633,
      "grad_norm": 0.1524086509507011,
      "learning_rate": 2.0944660739517187e-06,
      "loss": 0.0048,
      "step": 44470
    },
    {
      "epoch": 0.7278082303853391,
      "grad_norm": 0.07275966527166236,
      "learning_rate": 2.092142389403728e-06,
      "loss": 0.0013,
      "step": 44480
    },
    {
      "epoch": 0.7279718563364149,
      "grad_norm": 0.14016817379157911,
      "learning_rate": 2.08981965349286e-06,
      "loss": 0.0015,
      "step": 44490
    },
    {
      "epoch": 0.7281354822874908,
      "grad_norm": 0.1047604176529723,
      "learning_rate": 2.087497866976864e-06,
      "loss": 0.0021,
      "step": 44500
    },
    {
      "epoch": 0.7282991082385666,
      "grad_norm": 0.06882833873036522,
      "learning_rate": 2.085177030613183e-06,
      "loss": 0.0026,
      "step": 44510
    },
    {
      "epoch": 0.7284627341896425,
      "grad_norm": 0.1991065278976609,
      "learning_rate": 2.082857145158947e-06,
      "loss": 0.0028,
      "step": 44520
    },
    {
      "epoch": 0.7286263601407184,
      "grad_norm": 0.07120367759215092,
      "learning_rate": 2.080538211370979e-06,
      "loss": 0.0015,
      "step": 44530
    },
    {
      "epoch": 0.7287899860917941,
      "grad_norm": 0.24189413174854393,
      "learning_rate": 2.0782202300057875e-06,
      "loss": 0.0043,
      "step": 44540
    },
    {
      "epoch": 0.72895361204287,
      "grad_norm": 0.13727730179264033,
      "learning_rate": 2.0759032018195768e-06,
      "loss": 0.0019,
      "step": 44550
    },
    {
      "epoch": 0.7291172379939458,
      "grad_norm": 0.08816536843889627,
      "learning_rate": 2.073587127568235e-06,
      "loss": 0.0026,
      "step": 44560
    },
    {
      "epoch": 0.7292808639450217,
      "grad_norm": 0.14618166300666374,
      "learning_rate": 2.07127200800734e-06,
      "loss": 0.0027,
      "step": 44570
    },
    {
      "epoch": 0.7294444898960976,
      "grad_norm": 0.06422692706143053,
      "learning_rate": 2.0689578438921583e-06,
      "loss": 0.0027,
      "step": 44580
    },
    {
      "epoch": 0.7296081158471733,
      "grad_norm": 0.5534561855697258,
      "learning_rate": 2.0666446359776455e-06,
      "loss": 0.003,
      "step": 44590
    },
    {
      "epoch": 0.7297717417982492,
      "grad_norm": 0.12322313180578576,
      "learning_rate": 2.064332385018443e-06,
      "loss": 0.0029,
      "step": 44600
    },
    {
      "epoch": 0.729935367749325,
      "grad_norm": 0.10289307393065446,
      "learning_rate": 2.0620210917688843e-06,
      "loss": 0.0024,
      "step": 44610
    },
    {
      "epoch": 0.7300989937004009,
      "grad_norm": 0.1773294797103362,
      "learning_rate": 2.0597107569829877e-06,
      "loss": 0.0027,
      "step": 44620
    },
    {
      "epoch": 0.7302626196514768,
      "grad_norm": 0.051749257568555446,
      "learning_rate": 2.057401381414459e-06,
      "loss": 0.0017,
      "step": 44630
    },
    {
      "epoch": 0.7304262456025525,
      "grad_norm": 0.08641623436531766,
      "learning_rate": 2.0550929658166895e-06,
      "loss": 0.0018,
      "step": 44640
    },
    {
      "epoch": 0.7305898715536284,
      "grad_norm": 0.16757332908417882,
      "learning_rate": 2.05278551094276e-06,
      "loss": 0.0039,
      "step": 44650
    },
    {
      "epoch": 0.7307534975047042,
      "grad_norm": 0.34034136118993524,
      "learning_rate": 2.050479017545436e-06,
      "loss": 0.0019,
      "step": 44660
    },
    {
      "epoch": 0.7309171234557801,
      "grad_norm": 0.20456944247652428,
      "learning_rate": 2.0481734863771722e-06,
      "loss": 0.0025,
      "step": 44670
    },
    {
      "epoch": 0.731080749406856,
      "grad_norm": 0.16420890629398852,
      "learning_rate": 2.0458689181901063e-06,
      "loss": 0.0017,
      "step": 44680
    },
    {
      "epoch": 0.7312443753579317,
      "grad_norm": 0.23586257079547152,
      "learning_rate": 2.043565313736063e-06,
      "loss": 0.0016,
      "step": 44690
    },
    {
      "epoch": 0.7314080013090076,
      "grad_norm": 0.3829713795652063,
      "learning_rate": 2.041262673766552e-06,
      "loss": 0.0036,
      "step": 44700
    },
    {
      "epoch": 0.7315716272600834,
      "grad_norm": 0.22532166000163395,
      "learning_rate": 2.038960999032769e-06,
      "loss": 0.0016,
      "step": 44710
    },
    {
      "epoch": 0.7317352532111593,
      "grad_norm": 0.04739444860449337,
      "learning_rate": 2.0366602902855937e-06,
      "loss": 0.0018,
      "step": 44720
    },
    {
      "epoch": 0.7318988791622352,
      "grad_norm": 0.20575924543160157,
      "learning_rate": 2.034360548275594e-06,
      "loss": 0.0026,
      "step": 44730
    },
    {
      "epoch": 0.732062505113311,
      "grad_norm": 0.15894139683413838,
      "learning_rate": 2.0320617737530196e-06,
      "loss": 0.0017,
      "step": 44740
    },
    {
      "epoch": 0.7322261310643868,
      "grad_norm": 0.13377660460013746,
      "learning_rate": 2.029763967467804e-06,
      "loss": 0.0032,
      "step": 44750
    },
    {
      "epoch": 0.7323897570154626,
      "grad_norm": 0.06717132701862027,
      "learning_rate": 2.027467130169566e-06,
      "loss": 0.0023,
      "step": 44760
    },
    {
      "epoch": 0.7325533829665385,
      "grad_norm": 0.20909526028426656,
      "learning_rate": 2.025171262607608e-06,
      "loss": 0.0031,
      "step": 44770
    },
    {
      "epoch": 0.7327170089176144,
      "grad_norm": 0.08374260391433279,
      "learning_rate": 2.0228763655309167e-06,
      "loss": 0.0019,
      "step": 44780
    },
    {
      "epoch": 0.7328806348686902,
      "grad_norm": 0.09072123468106098,
      "learning_rate": 2.02058243968816e-06,
      "loss": 0.002,
      "step": 44790
    },
    {
      "epoch": 0.733044260819766,
      "grad_norm": 0.18573295478430615,
      "learning_rate": 2.018289485827694e-06,
      "loss": 0.0019,
      "step": 44800
    },
    {
      "epoch": 0.7332078867708418,
      "grad_norm": 0.11616154655057644,
      "learning_rate": 2.015997504697553e-06,
      "loss": 0.0014,
      "step": 44810
    },
    {
      "epoch": 0.7333715127219177,
      "grad_norm": 0.40644715335569065,
      "learning_rate": 2.013706497045454e-06,
      "loss": 0.0015,
      "step": 44820
    },
    {
      "epoch": 0.7335351386729936,
      "grad_norm": 0.2240162872258777,
      "learning_rate": 2.0114164636187983e-06,
      "loss": 0.0029,
      "step": 44830
    },
    {
      "epoch": 0.7336987646240694,
      "grad_norm": 0.22131508711574477,
      "learning_rate": 2.0091274051646697e-06,
      "loss": 0.0024,
      "step": 44840
    },
    {
      "epoch": 0.7338623905751452,
      "grad_norm": 0.139439179612051,
      "learning_rate": 2.0068393224298304e-06,
      "loss": 0.0027,
      "step": 44850
    },
    {
      "epoch": 0.734026016526221,
      "grad_norm": 0.1280390475124704,
      "learning_rate": 2.0045522161607306e-06,
      "loss": 0.0019,
      "step": 44860
    },
    {
      "epoch": 0.7341896424772969,
      "grad_norm": 0.08854489695305247,
      "learning_rate": 2.0022660871034966e-06,
      "loss": 0.002,
      "step": 44870
    },
    {
      "epoch": 0.7343532684283728,
      "grad_norm": 0.347302081479223,
      "learning_rate": 1.9999809360039372e-06,
      "loss": 0.0042,
      "step": 44880
    },
    {
      "epoch": 0.7345168943794486,
      "grad_norm": 0.0907303455001772,
      "learning_rate": 1.9976967636075434e-06,
      "loss": 0.003,
      "step": 44890
    },
    {
      "epoch": 0.7346805203305244,
      "grad_norm": 0.14189535923614688,
      "learning_rate": 1.9954135706594853e-06,
      "loss": 0.0021,
      "step": 44900
    },
    {
      "epoch": 0.7348441462816002,
      "grad_norm": 0.16516134939959948,
      "learning_rate": 1.993131357904613e-06,
      "loss": 0.0021,
      "step": 44910
    },
    {
      "epoch": 0.7350077722326761,
      "grad_norm": 0.12287025419956622,
      "learning_rate": 1.990850126087461e-06,
      "loss": 0.0031,
      "step": 44920
    },
    {
      "epoch": 0.735171398183752,
      "grad_norm": 0.07019523587990566,
      "learning_rate": 1.9885698759522408e-06,
      "loss": 0.0032,
      "step": 44930
    },
    {
      "epoch": 0.7353350241348278,
      "grad_norm": 0.20799262453246364,
      "learning_rate": 1.986290608242839e-06,
      "loss": 0.0022,
      "step": 44940
    },
    {
      "epoch": 0.7354986500859036,
      "grad_norm": 0.03875321360011786,
      "learning_rate": 1.984012323702832e-06,
      "loss": 0.0018,
      "step": 44950
    },
    {
      "epoch": 0.7356622760369794,
      "grad_norm": 0.068044816275422,
      "learning_rate": 1.9817350230754665e-06,
      "loss": 0.0017,
      "step": 44960
    },
    {
      "epoch": 0.7358259019880553,
      "grad_norm": 0.26818669510729753,
      "learning_rate": 1.9794587071036714e-06,
      "loss": 0.0029,
      "step": 44970
    },
    {
      "epoch": 0.7359895279391312,
      "grad_norm": 0.07581770693558115,
      "learning_rate": 1.977183376530057e-06,
      "loss": 0.0022,
      "step": 44980
    },
    {
      "epoch": 0.736153153890207,
      "grad_norm": 0.07587232075385605,
      "learning_rate": 1.97490903209691e-06,
      "loss": 0.0013,
      "step": 44990
    },
    {
      "epoch": 0.7363167798412829,
      "grad_norm": 0.06944411015081073,
      "learning_rate": 1.97263567454619e-06,
      "loss": 0.0023,
      "step": 45000
    },
    {
      "epoch": 0.7364804057923586,
      "grad_norm": 0.02386021230886774,
      "learning_rate": 1.9703633046195446e-06,
      "loss": 0.0034,
      "step": 45010
    },
    {
      "epoch": 0.7366440317434345,
      "grad_norm": 0.1710096932848367,
      "learning_rate": 1.9680919230582927e-06,
      "loss": 0.0027,
      "step": 45020
    },
    {
      "epoch": 0.7368076576945104,
      "grad_norm": 0.0938919028809799,
      "learning_rate": 1.965821530603431e-06,
      "loss": 0.0025,
      "step": 45030
    },
    {
      "epoch": 0.7369712836455862,
      "grad_norm": 0.07001442061169248,
      "learning_rate": 1.9635521279956383e-06,
      "loss": 0.0018,
      "step": 45040
    },
    {
      "epoch": 0.7371349095966621,
      "grad_norm": 0.06993680248490873,
      "learning_rate": 1.961283715975266e-06,
      "loss": 0.0022,
      "step": 45050
    },
    {
      "epoch": 0.7372985355477378,
      "grad_norm": 0.1694704095026415,
      "learning_rate": 1.95901629528234e-06,
      "loss": 0.0016,
      "step": 45060
    },
    {
      "epoch": 0.7374621614988137,
      "grad_norm": 0.2005480930223308,
      "learning_rate": 1.95674986665657e-06,
      "loss": 0.0024,
      "step": 45070
    },
    {
      "epoch": 0.7376257874498896,
      "grad_norm": 0.1874518209816124,
      "learning_rate": 1.9544844308373367e-06,
      "loss": 0.0032,
      "step": 45080
    },
    {
      "epoch": 0.7377894134009654,
      "grad_norm": 0.16697680388886196,
      "learning_rate": 1.952219988563697e-06,
      "loss": 0.0031,
      "step": 45090
    },
    {
      "epoch": 0.7379530393520413,
      "grad_norm": 0.09093627501903218,
      "learning_rate": 1.9499565405743886e-06,
      "loss": 0.0028,
      "step": 45100
    },
    {
      "epoch": 0.738116665303117,
      "grad_norm": 0.24135580440753351,
      "learning_rate": 1.9476940876078214e-06,
      "loss": 0.0021,
      "step": 45110
    },
    {
      "epoch": 0.7382802912541929,
      "grad_norm": 0.18026667668544955,
      "learning_rate": 1.9454326304020773e-06,
      "loss": 0.0024,
      "step": 45120
    },
    {
      "epoch": 0.7384439172052688,
      "grad_norm": 0.3016966771453516,
      "learning_rate": 1.9431721696949165e-06,
      "loss": 0.003,
      "step": 45130
    },
    {
      "epoch": 0.7386075431563446,
      "grad_norm": 0.15173579693850883,
      "learning_rate": 1.9409127062237777e-06,
      "loss": 0.0018,
      "step": 45140
    },
    {
      "epoch": 0.7387711691074205,
      "grad_norm": 0.11620601530956795,
      "learning_rate": 1.938654240725769e-06,
      "loss": 0.0014,
      "step": 45150
    },
    {
      "epoch": 0.7389347950584962,
      "grad_norm": 0.10542180131383322,
      "learning_rate": 1.936396773937674e-06,
      "loss": 0.0012,
      "step": 45160
    },
    {
      "epoch": 0.7390984210095721,
      "grad_norm": 0.03722660016681537,
      "learning_rate": 1.9341403065959564e-06,
      "loss": 0.0024,
      "step": 45170
    },
    {
      "epoch": 0.739262046960648,
      "grad_norm": 0.18887618947431387,
      "learning_rate": 1.9318848394367423e-06,
      "loss": 0.0022,
      "step": 45180
    },
    {
      "epoch": 0.7394256729117238,
      "grad_norm": 0.15358224706067536,
      "learning_rate": 1.929630373195839e-06,
      "loss": 0.0018,
      "step": 45190
    },
    {
      "epoch": 0.7395892988627997,
      "grad_norm": 0.10649897725800517,
      "learning_rate": 1.9273769086087296e-06,
      "loss": 0.0028,
      "step": 45200
    },
    {
      "epoch": 0.7397529248138754,
      "grad_norm": 0.13976114201848927,
      "learning_rate": 1.9251244464105657e-06,
      "loss": 0.0048,
      "step": 45210
    },
    {
      "epoch": 0.7399165507649513,
      "grad_norm": 0.16145332507081675,
      "learning_rate": 1.9228729873361702e-06,
      "loss": 0.0011,
      "step": 45220
    },
    {
      "epoch": 0.7400801767160272,
      "grad_norm": 0.11590750512284433,
      "learning_rate": 1.920622532120048e-06,
      "loss": 0.0019,
      "step": 45230
    },
    {
      "epoch": 0.740243802667103,
      "grad_norm": 0.24927081755085978,
      "learning_rate": 1.9183730814963658e-06,
      "loss": 0.0022,
      "step": 45240
    },
    {
      "epoch": 0.7404074286181789,
      "grad_norm": 0.15121382289967583,
      "learning_rate": 1.9161246361989657e-06,
      "loss": 0.0022,
      "step": 45250
    },
    {
      "epoch": 0.7405710545692546,
      "grad_norm": 0.147385205806761,
      "learning_rate": 1.9138771969613674e-06,
      "loss": 0.0026,
      "step": 45260
    },
    {
      "epoch": 0.7407346805203305,
      "grad_norm": 0.24503940760113305,
      "learning_rate": 1.911630764516756e-06,
      "loss": 0.0024,
      "step": 45270
    },
    {
      "epoch": 0.7408983064714064,
      "grad_norm": 0.09040951479567234,
      "learning_rate": 1.9093853395979893e-06,
      "loss": 0.0012,
      "step": 45280
    },
    {
      "epoch": 0.7410619324224822,
      "grad_norm": 0.1755614111364217,
      "learning_rate": 1.907140922937602e-06,
      "loss": 0.0026,
      "step": 45290
    },
    {
      "epoch": 0.7412255583735581,
      "grad_norm": 0.08732935330092154,
      "learning_rate": 1.9048975152677907e-06,
      "loss": 0.0012,
      "step": 45300
    },
    {
      "epoch": 0.7413891843246339,
      "grad_norm": 0.3409927076977567,
      "learning_rate": 1.9026551173204272e-06,
      "loss": 0.0047,
      "step": 45310
    },
    {
      "epoch": 0.7415528102757097,
      "grad_norm": 0.36886882739189664,
      "learning_rate": 1.900413729827058e-06,
      "loss": 0.0033,
      "step": 45320
    },
    {
      "epoch": 0.7417164362267855,
      "grad_norm": 0.04912119302042821,
      "learning_rate": 1.898173353518894e-06,
      "loss": 0.0039,
      "step": 45330
    },
    {
      "epoch": 0.7418800621778614,
      "grad_norm": 0.2547902304322046,
      "learning_rate": 1.8959339891268164e-06,
      "loss": 0.0017,
      "step": 45340
    },
    {
      "epoch": 0.7420436881289373,
      "grad_norm": 0.02119320991032157,
      "learning_rate": 1.8936956373813837e-06,
      "loss": 0.002,
      "step": 45350
    },
    {
      "epoch": 0.7422073140800131,
      "grad_norm": 0.045534037443961166,
      "learning_rate": 1.8914582990128133e-06,
      "loss": 0.0017,
      "step": 45360
    },
    {
      "epoch": 0.7423709400310889,
      "grad_norm": 0.11198447139565858,
      "learning_rate": 1.8892219747509978e-06,
      "loss": 0.0035,
      "step": 45370
    },
    {
      "epoch": 0.7425345659821647,
      "grad_norm": 0.23907883237786615,
      "learning_rate": 1.8869866653255015e-06,
      "loss": 0.0025,
      "step": 45380
    },
    {
      "epoch": 0.7426981919332406,
      "grad_norm": 0.17118705542594306,
      "learning_rate": 1.8847523714655525e-06,
      "loss": 0.0023,
      "step": 45390
    },
    {
      "epoch": 0.7428618178843165,
      "grad_norm": 0.24736742017835792,
      "learning_rate": 1.8825190939000482e-06,
      "loss": 0.0029,
      "step": 45400
    },
    {
      "epoch": 0.7430254438353923,
      "grad_norm": 0.10767291834913967,
      "learning_rate": 1.8802868333575614e-06,
      "loss": 0.0013,
      "step": 45410
    },
    {
      "epoch": 0.7431890697864681,
      "grad_norm": 0.09482697240303418,
      "learning_rate": 1.8780555905663223e-06,
      "loss": 0.0016,
      "step": 45420
    },
    {
      "epoch": 0.7433526957375439,
      "grad_norm": 0.1840747034603889,
      "learning_rate": 1.8758253662542336e-06,
      "loss": 0.003,
      "step": 45430
    },
    {
      "epoch": 0.7435163216886198,
      "grad_norm": 0.4689505965538142,
      "learning_rate": 1.8735961611488712e-06,
      "loss": 0.0024,
      "step": 45440
    },
    {
      "epoch": 0.7436799476396957,
      "grad_norm": 0.11113900725758293,
      "learning_rate": 1.8713679759774718e-06,
      "loss": 0.0011,
      "step": 45450
    },
    {
      "epoch": 0.7438435735907715,
      "grad_norm": 0.19390647895564087,
      "learning_rate": 1.8691408114669406e-06,
      "loss": 0.0015,
      "step": 45460
    },
    {
      "epoch": 0.7440071995418474,
      "grad_norm": 0.17661179148210726,
      "learning_rate": 1.866914668343852e-06,
      "loss": 0.001,
      "step": 45470
    },
    {
      "epoch": 0.7441708254929231,
      "grad_norm": 0.24105734218861763,
      "learning_rate": 1.8646895473344446e-06,
      "loss": 0.0021,
      "step": 45480
    },
    {
      "epoch": 0.744334451443999,
      "grad_norm": 0.22231021008987104,
      "learning_rate": 1.8624654491646254e-06,
      "loss": 0.0025,
      "step": 45490
    },
    {
      "epoch": 0.7444980773950749,
      "grad_norm": 0.12027864243007136,
      "learning_rate": 1.8602423745599652e-06,
      "loss": 0.0028,
      "step": 45500
    },
    {
      "epoch": 0.7446617033461507,
      "grad_norm": 0.02764323081565079,
      "learning_rate": 1.8580203242457061e-06,
      "loss": 0.0017,
      "step": 45510
    },
    {
      "epoch": 0.7448253292972266,
      "grad_norm": 0.17601443336291742,
      "learning_rate": 1.8557992989467517e-06,
      "loss": 0.0018,
      "step": 45520
    },
    {
      "epoch": 0.7449889552483023,
      "grad_norm": 0.1671539370895811,
      "learning_rate": 1.8535792993876712e-06,
      "loss": 0.0012,
      "step": 45530
    },
    {
      "epoch": 0.7451525811993782,
      "grad_norm": 0.03842371507904231,
      "learning_rate": 1.8513603262927005e-06,
      "loss": 0.0009,
      "step": 45540
    },
    {
      "epoch": 0.7453162071504541,
      "grad_norm": 0.09876800924544066,
      "learning_rate": 1.8491423803857405e-06,
      "loss": 0.0017,
      "step": 45550
    },
    {
      "epoch": 0.7454798331015299,
      "grad_norm": 0.09453110457444099,
      "learning_rate": 1.8469254623903554e-06,
      "loss": 0.0017,
      "step": 45560
    },
    {
      "epoch": 0.7456434590526058,
      "grad_norm": 0.1488783985696372,
      "learning_rate": 1.844709573029778e-06,
      "loss": 0.003,
      "step": 45570
    },
    {
      "epoch": 0.7458070850036815,
      "grad_norm": 0.17183523010699714,
      "learning_rate": 1.8424947130269027e-06,
      "loss": 0.0022,
      "step": 45580
    },
    {
      "epoch": 0.7459707109547574,
      "grad_norm": 0.004651303189011751,
      "learning_rate": 1.8402808831042873e-06,
      "loss": 0.0043,
      "step": 45590
    },
    {
      "epoch": 0.7461343369058333,
      "grad_norm": 0.18327532512226255,
      "learning_rate": 1.8380680839841553e-06,
      "loss": 0.0016,
      "step": 45600
    },
    {
      "epoch": 0.7462979628569091,
      "grad_norm": 0.09656479761314256,
      "learning_rate": 1.8358563163883924e-06,
      "loss": 0.0024,
      "step": 45610
    },
    {
      "epoch": 0.746461588807985,
      "grad_norm": 0.36363930387145943,
      "learning_rate": 1.8336455810385478e-06,
      "loss": 0.0029,
      "step": 45620
    },
    {
      "epoch": 0.7466252147590607,
      "grad_norm": 0.17106124723869456,
      "learning_rate": 1.8314358786558384e-06,
      "loss": 0.0017,
      "step": 45630
    },
    {
      "epoch": 0.7467888407101366,
      "grad_norm": 0.01420596762759395,
      "learning_rate": 1.8292272099611375e-06,
      "loss": 0.0028,
      "step": 45640
    },
    {
      "epoch": 0.7469524666612125,
      "grad_norm": 0.17650140423031205,
      "learning_rate": 1.827019575674986e-06,
      "loss": 0.0032,
      "step": 45650
    },
    {
      "epoch": 0.7471160926122883,
      "grad_norm": 0.04427962288809695,
      "learning_rate": 1.8248129765175837e-06,
      "loss": 0.0019,
      "step": 45660
    },
    {
      "epoch": 0.7472797185633642,
      "grad_norm": 0.21856651721766107,
      "learning_rate": 1.8226074132087962e-06,
      "loss": 0.0013,
      "step": 45670
    },
    {
      "epoch": 0.74744334451444,
      "grad_norm": 0.11581839091157066,
      "learning_rate": 1.8204028864681467e-06,
      "loss": 0.0019,
      "step": 45680
    },
    {
      "epoch": 0.7476069704655158,
      "grad_norm": 0.04061043625629293,
      "learning_rate": 1.8181993970148272e-06,
      "loss": 0.0009,
      "step": 45690
    },
    {
      "epoch": 0.7477705964165917,
      "grad_norm": 0.10997613632037238,
      "learning_rate": 1.8159969455676845e-06,
      "loss": 0.0011,
      "step": 45700
    },
    {
      "epoch": 0.7479342223676675,
      "grad_norm": 0.4589938775349932,
      "learning_rate": 1.8137955328452307e-06,
      "loss": 0.0026,
      "step": 45710
    },
    {
      "epoch": 0.7480978483187434,
      "grad_norm": 0.18704913241387183,
      "learning_rate": 1.8115951595656366e-06,
      "loss": 0.0023,
      "step": 45720
    },
    {
      "epoch": 0.7482614742698191,
      "grad_norm": 0.1329641459361191,
      "learning_rate": 1.8093958264467359e-06,
      "loss": 0.0034,
      "step": 45730
    },
    {
      "epoch": 0.748425100220895,
      "grad_norm": 0.19816799287925482,
      "learning_rate": 1.8071975342060204e-06,
      "loss": 0.0025,
      "step": 45740
    },
    {
      "epoch": 0.7485887261719709,
      "grad_norm": 0.1267937946325208,
      "learning_rate": 1.8050002835606468e-06,
      "loss": 0.0028,
      "step": 45750
    },
    {
      "epoch": 0.7487523521230467,
      "grad_norm": 0.19350279082836166,
      "learning_rate": 1.8028040752274283e-06,
      "loss": 0.0021,
      "step": 45760
    },
    {
      "epoch": 0.7489159780741226,
      "grad_norm": 0.09875867404983932,
      "learning_rate": 1.8006089099228386e-06,
      "loss": 0.0024,
      "step": 45770
    },
    {
      "epoch": 0.7490796040251984,
      "grad_norm": 0.0848138155360765,
      "learning_rate": 1.7984147883630115e-06,
      "loss": 0.0015,
      "step": 45780
    },
    {
      "epoch": 0.7492432299762742,
      "grad_norm": 0.16643147003561135,
      "learning_rate": 1.7962217112637404e-06,
      "loss": 0.0014,
      "step": 45790
    },
    {
      "epoch": 0.7494068559273501,
      "grad_norm": 0.14499421686238045,
      "learning_rate": 1.7940296793404783e-06,
      "loss": 0.0027,
      "step": 45800
    },
    {
      "epoch": 0.7495704818784259,
      "grad_norm": 0.11939407827616952,
      "learning_rate": 1.7918386933083348e-06,
      "loss": 0.0019,
      "step": 45810
    },
    {
      "epoch": 0.7497341078295018,
      "grad_norm": 0.09886453832912799,
      "learning_rate": 1.7896487538820833e-06,
      "loss": 0.0012,
      "step": 45820
    },
    {
      "epoch": 0.7498977337805776,
      "grad_norm": 0.06467277177554229,
      "learning_rate": 1.7874598617761524e-06,
      "loss": 0.0009,
      "step": 45830
    },
    {
      "epoch": 0.7500613597316534,
      "grad_norm": 0.10456585610589049,
      "learning_rate": 1.7852720177046284e-06,
      "loss": 0.0029,
      "step": 45840
    },
    {
      "epoch": 0.7502249856827293,
      "grad_norm": 0.11273119911729533,
      "learning_rate": 1.7830852223812573e-06,
      "loss": 0.0017,
      "step": 45850
    },
    {
      "epoch": 0.7503886116338051,
      "grad_norm": 0.11300270318524014,
      "learning_rate": 1.7808994765194414e-06,
      "loss": 0.0017,
      "step": 45860
    },
    {
      "epoch": 0.750552237584881,
      "grad_norm": 0.1315058089385145,
      "learning_rate": 1.7787147808322413e-06,
      "loss": 0.0022,
      "step": 45870
    },
    {
      "epoch": 0.7507158635359568,
      "grad_norm": 0.15766203037906645,
      "learning_rate": 1.7765311360323778e-06,
      "loss": 0.0035,
      "step": 45880
    },
    {
      "epoch": 0.7508794894870326,
      "grad_norm": 0.03440059766839715,
      "learning_rate": 1.774348542832225e-06,
      "loss": 0.0019,
      "step": 45890
    },
    {
      "epoch": 0.7510431154381085,
      "grad_norm": 0.09004760283375655,
      "learning_rate": 1.7721670019438153e-06,
      "loss": 0.0027,
      "step": 45900
    },
    {
      "epoch": 0.7512067413891843,
      "grad_norm": 0.12460464571217722,
      "learning_rate": 1.769986514078838e-06,
      "loss": 0.0056,
      "step": 45910
    },
    {
      "epoch": 0.7513703673402602,
      "grad_norm": 0.3541459106115299,
      "learning_rate": 1.7678070799486386e-06,
      "loss": 0.0054,
      "step": 45920
    },
    {
      "epoch": 0.751533993291336,
      "grad_norm": 0.08881843583724262,
      "learning_rate": 1.765628700264217e-06,
      "loss": 0.0023,
      "step": 45930
    },
    {
      "epoch": 0.7516976192424119,
      "grad_norm": 0.04137372932640359,
      "learning_rate": 1.7634513757362343e-06,
      "loss": 0.0031,
      "step": 45940
    },
    {
      "epoch": 0.7518612451934877,
      "grad_norm": 0.029204014591625475,
      "learning_rate": 1.7612751070750033e-06,
      "loss": 0.0018,
      "step": 45950
    },
    {
      "epoch": 0.7520248711445635,
      "grad_norm": 0.1039995498049076,
      "learning_rate": 1.7590998949904924e-06,
      "loss": 0.003,
      "step": 45960
    },
    {
      "epoch": 0.7521884970956394,
      "grad_norm": 0.13248874400385757,
      "learning_rate": 1.7569257401923263e-06,
      "loss": 0.0017,
      "step": 45970
    },
    {
      "epoch": 0.7523521230467152,
      "grad_norm": 0.17715847502014476,
      "learning_rate": 1.7547526433897848e-06,
      "loss": 0.0024,
      "step": 45980
    },
    {
      "epoch": 0.7525157489977911,
      "grad_norm": 0.26151554617946166,
      "learning_rate": 1.7525806052918e-06,
      "loss": 0.0014,
      "step": 45990
    },
    {
      "epoch": 0.7526793749488669,
      "grad_norm": 0.046693960697858757,
      "learning_rate": 1.7504096266069647e-06,
      "loss": 0.0014,
      "step": 46000
    },
    {
      "epoch": 0.7528430008999427,
      "grad_norm": 0.08446428144037361,
      "learning_rate": 1.7482397080435209e-06,
      "loss": 0.0023,
      "step": 46010
    },
    {
      "epoch": 0.7530066268510186,
      "grad_norm": 0.18679126424579465,
      "learning_rate": 1.7460708503093655e-06,
      "loss": 0.0016,
      "step": 46020
    },
    {
      "epoch": 0.7531702528020944,
      "grad_norm": 0.15606888719379577,
      "learning_rate": 1.7439030541120506e-06,
      "loss": 0.0021,
      "step": 46030
    },
    {
      "epoch": 0.7533338787531703,
      "grad_norm": 0.06365374359783149,
      "learning_rate": 1.7417363201587812e-06,
      "loss": 0.0013,
      "step": 46040
    },
    {
      "epoch": 0.7534975047042461,
      "grad_norm": 0.16804757464299622,
      "learning_rate": 1.7395706491564135e-06,
      "loss": 0.0034,
      "step": 46050
    },
    {
      "epoch": 0.7536611306553219,
      "grad_norm": 0.07866634967892441,
      "learning_rate": 1.7374060418114642e-06,
      "loss": 0.0028,
      "step": 46060
    },
    {
      "epoch": 0.7538247566063978,
      "grad_norm": 0.1447043436267361,
      "learning_rate": 1.7352424988300948e-06,
      "loss": 0.0018,
      "step": 46070
    },
    {
      "epoch": 0.7539883825574736,
      "grad_norm": 0.17424757155382511,
      "learning_rate": 1.7330800209181241e-06,
      "loss": 0.0018,
      "step": 46080
    },
    {
      "epoch": 0.7541520085085495,
      "grad_norm": 0.05951055823103956,
      "learning_rate": 1.730918608781022e-06,
      "loss": 0.0012,
      "step": 46090
    },
    {
      "epoch": 0.7543156344596254,
      "grad_norm": 0.1066577868167116,
      "learning_rate": 1.7287582631239114e-06,
      "loss": 0.0018,
      "step": 46100
    },
    {
      "epoch": 0.7544792604107011,
      "grad_norm": 0.06897828734782331,
      "learning_rate": 1.7265989846515657e-06,
      "loss": 0.0015,
      "step": 46110
    },
    {
      "epoch": 0.754642886361777,
      "grad_norm": 0.21023559560074864,
      "learning_rate": 1.7244407740684105e-06,
      "loss": 0.0034,
      "step": 46120
    },
    {
      "epoch": 0.7548065123128528,
      "grad_norm": 0.17842566677001792,
      "learning_rate": 1.7222836320785275e-06,
      "loss": 0.0019,
      "step": 46130
    },
    {
      "epoch": 0.7549701382639287,
      "grad_norm": 0.20762704759038897,
      "learning_rate": 1.7201275593856454e-06,
      "loss": 0.0018,
      "step": 46140
    },
    {
      "epoch": 0.7551337642150046,
      "grad_norm": 0.07038335604206682,
      "learning_rate": 1.7179725566931399e-06,
      "loss": 0.0021,
      "step": 46150
    },
    {
      "epoch": 0.7552973901660803,
      "grad_norm": 0.26129034568728193,
      "learning_rate": 1.7158186247040476e-06,
      "loss": 0.0015,
      "step": 46160
    },
    {
      "epoch": 0.7554610161171562,
      "grad_norm": 0.18913411890710702,
      "learning_rate": 1.7136657641210492e-06,
      "loss": 0.002,
      "step": 46170
    },
    {
      "epoch": 0.755624642068232,
      "grad_norm": 0.0830088014143581,
      "learning_rate": 1.7115139756464761e-06,
      "loss": 0.0016,
      "step": 46180
    },
    {
      "epoch": 0.7557882680193079,
      "grad_norm": 0.1611380596120119,
      "learning_rate": 1.7093632599823136e-06,
      "loss": 0.0009,
      "step": 46190
    },
    {
      "epoch": 0.7559518939703836,
      "grad_norm": 0.07591516709823387,
      "learning_rate": 1.707213617830195e-06,
      "loss": 0.0023,
      "step": 46200
    },
    {
      "epoch": 0.7561155199214595,
      "grad_norm": 0.14892147848480908,
      "learning_rate": 1.7050650498913984e-06,
      "loss": 0.0014,
      "step": 46210
    },
    {
      "epoch": 0.7562791458725354,
      "grad_norm": 0.052792007782790265,
      "learning_rate": 1.702917556866861e-06,
      "loss": 0.0013,
      "step": 46220
    },
    {
      "epoch": 0.7564427718236112,
      "grad_norm": 0.010226413070532027,
      "learning_rate": 1.700771139457163e-06,
      "loss": 0.0013,
      "step": 46230
    },
    {
      "epoch": 0.7566063977746871,
      "grad_norm": 0.10869114769926211,
      "learning_rate": 1.6986257983625326e-06,
      "loss": 0.0028,
      "step": 46240
    },
    {
      "epoch": 0.7567700237257629,
      "grad_norm": 0.17238230178850342,
      "learning_rate": 1.696481534282855e-06,
      "loss": 0.0029,
      "step": 46250
    },
    {
      "epoch": 0.7569336496768387,
      "grad_norm": 0.1389662974472085,
      "learning_rate": 1.6943383479176534e-06,
      "loss": 0.0024,
      "step": 46260
    },
    {
      "epoch": 0.7570972756279146,
      "grad_norm": 0.11471205834050704,
      "learning_rate": 1.692196239966104e-06,
      "loss": 0.0041,
      "step": 46270
    },
    {
      "epoch": 0.7572609015789904,
      "grad_norm": 0.06957039215390441,
      "learning_rate": 1.6900552111270356e-06,
      "loss": 0.0019,
      "step": 46280
    },
    {
      "epoch": 0.7574245275300663,
      "grad_norm": 0.07234541738303123,
      "learning_rate": 1.6879152620989191e-06,
      "loss": 0.0025,
      "step": 46290
    },
    {
      "epoch": 0.7575881534811421,
      "grad_norm": 0.13107428634116425,
      "learning_rate": 1.6857763935798733e-06,
      "loss": 0.0025,
      "step": 46300
    },
    {
      "epoch": 0.7577517794322179,
      "grad_norm": 0.15651518690608868,
      "learning_rate": 1.6836386062676712e-06,
      "loss": 0.0028,
      "step": 46310
    },
    {
      "epoch": 0.7579154053832938,
      "grad_norm": 0.07394710865282514,
      "learning_rate": 1.6815019008597232e-06,
      "loss": 0.0047,
      "step": 46320
    },
    {
      "epoch": 0.7580790313343696,
      "grad_norm": 0.14916797442407503,
      "learning_rate": 1.6793662780530916e-06,
      "loss": 0.0015,
      "step": 46330
    },
    {
      "epoch": 0.7582426572854455,
      "grad_norm": 0.10545076104125851,
      "learning_rate": 1.6772317385444892e-06,
      "loss": 0.0024,
      "step": 46340
    },
    {
      "epoch": 0.7584062832365213,
      "grad_norm": 0.1853139494490722,
      "learning_rate": 1.6750982830302699e-06,
      "loss": 0.0021,
      "step": 46350
    },
    {
      "epoch": 0.7585699091875971,
      "grad_norm": 0.06326378627227601,
      "learning_rate": 1.672965912206434e-06,
      "loss": 0.0019,
      "step": 46360
    },
    {
      "epoch": 0.758733535138673,
      "grad_norm": 0.07604827670661202,
      "learning_rate": 1.6708346267686343e-06,
      "loss": 0.0014,
      "step": 46370
    },
    {
      "epoch": 0.7588971610897488,
      "grad_norm": 0.08526544901432646,
      "learning_rate": 1.668704427412161e-06,
      "loss": 0.0016,
      "step": 46380
    },
    {
      "epoch": 0.7590607870408247,
      "grad_norm": 0.1403100702221909,
      "learning_rate": 1.6665753148319535e-06,
      "loss": 0.0011,
      "step": 46390
    },
    {
      "epoch": 0.7592244129919005,
      "grad_norm": 0.0717180421433381,
      "learning_rate": 1.6644472897226004e-06,
      "loss": 0.0023,
      "step": 46400
    },
    {
      "epoch": 0.7593880389429764,
      "grad_norm": 0.13747307040847662,
      "learning_rate": 1.6623203527783304e-06,
      "loss": 0.003,
      "step": 46410
    },
    {
      "epoch": 0.7595516648940522,
      "grad_norm": 0.03655971584837516,
      "learning_rate": 1.6601945046930184e-06,
      "loss": 0.0023,
      "step": 46420
    },
    {
      "epoch": 0.759715290845128,
      "grad_norm": 0.27527597967681267,
      "learning_rate": 1.6580697461601885e-06,
      "loss": 0.0034,
      "step": 46430
    },
    {
      "epoch": 0.7598789167962039,
      "grad_norm": 0.09193592655360804,
      "learning_rate": 1.6559460778730007e-06,
      "loss": 0.0017,
      "step": 46440
    },
    {
      "epoch": 0.7600425427472797,
      "grad_norm": 0.1123773067565704,
      "learning_rate": 1.653823500524267e-06,
      "loss": 0.004,
      "step": 46450
    },
    {
      "epoch": 0.7602061686983556,
      "grad_norm": 0.1621251570260926,
      "learning_rate": 1.6517020148064384e-06,
      "loss": 0.0021,
      "step": 46460
    },
    {
      "epoch": 0.7603697946494314,
      "grad_norm": 0.04539460286656017,
      "learning_rate": 1.649581621411615e-06,
      "loss": 0.0021,
      "step": 46470
    },
    {
      "epoch": 0.7605334206005072,
      "grad_norm": 0.3136356898861863,
      "learning_rate": 1.647462321031536e-06,
      "loss": 0.0032,
      "step": 46480
    },
    {
      "epoch": 0.7606970465515831,
      "grad_norm": 0.1859689850362368,
      "learning_rate": 1.6453441143575865e-06,
      "loss": 0.0023,
      "step": 46490
    },
    {
      "epoch": 0.7608606725026589,
      "grad_norm": 0.06229952800958062,
      "learning_rate": 1.6432270020807933e-06,
      "loss": 0.0017,
      "step": 46500
    },
    {
      "epoch": 0.7610242984537348,
      "grad_norm": 0.026158627781626156,
      "learning_rate": 1.641110984891827e-06,
      "loss": 0.0018,
      "step": 46510
    },
    {
      "epoch": 0.7611879244048106,
      "grad_norm": 0.11928646256441852,
      "learning_rate": 1.6389960634809998e-06,
      "loss": 0.0017,
      "step": 46520
    },
    {
      "epoch": 0.7613515503558864,
      "grad_norm": 0.3741799477528657,
      "learning_rate": 1.6368822385382694e-06,
      "loss": 0.0036,
      "step": 46530
    },
    {
      "epoch": 0.7615151763069623,
      "grad_norm": 0.1437330767658559,
      "learning_rate": 1.6347695107532329e-06,
      "loss": 0.0023,
      "step": 46540
    },
    {
      "epoch": 0.7616788022580381,
      "grad_norm": 0.09280638125276448,
      "learning_rate": 1.6326578808151305e-06,
      "loss": 0.0015,
      "step": 46550
    },
    {
      "epoch": 0.761842428209114,
      "grad_norm": 0.1178824916932626,
      "learning_rate": 1.6305473494128437e-06,
      "loss": 0.002,
      "step": 46560
    },
    {
      "epoch": 0.7620060541601898,
      "grad_norm": 0.32909055714790286,
      "learning_rate": 1.6284379172348968e-06,
      "loss": 0.0022,
      "step": 46570
    },
    {
      "epoch": 0.7621696801112656,
      "grad_norm": 0.09575210698464756,
      "learning_rate": 1.6263295849694522e-06,
      "loss": 0.0014,
      "step": 46580
    },
    {
      "epoch": 0.7623333060623415,
      "grad_norm": 0.028458873881841106,
      "learning_rate": 1.6242223533043194e-06,
      "loss": 0.0027,
      "step": 46590
    },
    {
      "epoch": 0.7624969320134173,
      "grad_norm": 0.0848806430118753,
      "learning_rate": 1.6221162229269444e-06,
      "loss": 0.0021,
      "step": 46600
    },
    {
      "epoch": 0.7626605579644932,
      "grad_norm": 0.33646600663602283,
      "learning_rate": 1.6200111945244151e-06,
      "loss": 0.0044,
      "step": 46610
    },
    {
      "epoch": 0.762824183915569,
      "grad_norm": 0.21849747987078838,
      "learning_rate": 1.6179072687834586e-06,
      "loss": 0.0039,
      "step": 46620
    },
    {
      "epoch": 0.7629878098666448,
      "grad_norm": 0.12799221768823424,
      "learning_rate": 1.6158044463904448e-06,
      "loss": 0.002,
      "step": 46630
    },
    {
      "epoch": 0.7631514358177207,
      "grad_norm": 0.10123207359588249,
      "learning_rate": 1.6137027280313799e-06,
      "loss": 0.0022,
      "step": 46640
    },
    {
      "epoch": 0.7633150617687965,
      "grad_norm": 0.31255734553364206,
      "learning_rate": 1.6116021143919158e-06,
      "loss": 0.0014,
      "step": 46650
    },
    {
      "epoch": 0.7634786877198724,
      "grad_norm": 0.0796031598281909,
      "learning_rate": 1.6095026061573393e-06,
      "loss": 0.002,
      "step": 46660
    },
    {
      "epoch": 0.7636423136709483,
      "grad_norm": 0.40405296768151544,
      "learning_rate": 1.607404204012577e-06,
      "loss": 0.0016,
      "step": 46670
    },
    {
      "epoch": 0.763805939622024,
      "grad_norm": 0.5970723212634294,
      "learning_rate": 1.6053069086421962e-06,
      "loss": 0.0025,
      "step": 46680
    },
    {
      "epoch": 0.7639695655730999,
      "grad_norm": 0.14978991314871895,
      "learning_rate": 1.6032107207304021e-06,
      "loss": 0.0017,
      "step": 46690
    },
    {
      "epoch": 0.7641331915241757,
      "grad_norm": 0.1483389179165418,
      "learning_rate": 1.6011156409610379e-06,
      "loss": 0.0022,
      "step": 46700
    },
    {
      "epoch": 0.7642968174752516,
      "grad_norm": 0.20166498602157654,
      "learning_rate": 1.5990216700175887e-06,
      "loss": 0.0023,
      "step": 46710
    },
    {
      "epoch": 0.7644604434263275,
      "grad_norm": 0.4141697727184578,
      "learning_rate": 1.5969288085831741e-06,
      "loss": 0.0025,
      "step": 46720
    },
    {
      "epoch": 0.7646240693774032,
      "grad_norm": 0.11035183399177058,
      "learning_rate": 1.5948370573405536e-06,
      "loss": 0.0016,
      "step": 46730
    },
    {
      "epoch": 0.7647876953284791,
      "grad_norm": 0.2544156090677158,
      "learning_rate": 1.5927464169721234e-06,
      "loss": 0.0026,
      "step": 46740
    },
    {
      "epoch": 0.7649513212795549,
      "grad_norm": 0.06960523098979804,
      "learning_rate": 1.5906568881599187e-06,
      "loss": 0.0018,
      "step": 46750
    },
    {
      "epoch": 0.7651149472306308,
      "grad_norm": 0.14514292351443067,
      "learning_rate": 1.5885684715856108e-06,
      "loss": 0.0019,
      "step": 46760
    },
    {
      "epoch": 0.7652785731817067,
      "grad_norm": 0.2008801199666789,
      "learning_rate": 1.586481167930507e-06,
      "loss": 0.0025,
      "step": 46770
    },
    {
      "epoch": 0.7654421991327824,
      "grad_norm": 0.3152734410769143,
      "learning_rate": 1.584394977875557e-06,
      "loss": 0.0032,
      "step": 46780
    },
    {
      "epoch": 0.7656058250838583,
      "grad_norm": 0.11425898740560331,
      "learning_rate": 1.5823099021013411e-06,
      "loss": 0.0024,
      "step": 46790
    },
    {
      "epoch": 0.7657694510349341,
      "grad_norm": 0.2391848987670341,
      "learning_rate": 1.5802259412880783e-06,
      "loss": 0.0019,
      "step": 46800
    },
    {
      "epoch": 0.76593307698601,
      "grad_norm": 0.2137738232451391,
      "learning_rate": 1.5781430961156247e-06,
      "loss": 0.0016,
      "step": 46810
    },
    {
      "epoch": 0.7660967029370859,
      "grad_norm": 0.08676653787845492,
      "learning_rate": 1.5760613672634717e-06,
      "loss": 0.004,
      "step": 46820
    },
    {
      "epoch": 0.7662603288881616,
      "grad_norm": 0.1892549004346875,
      "learning_rate": 1.573980755410744e-06,
      "loss": 0.0019,
      "step": 46830
    },
    {
      "epoch": 0.7664239548392375,
      "grad_norm": 0.05132221080266076,
      "learning_rate": 1.5719012612362089e-06,
      "loss": 0.004,
      "step": 46840
    },
    {
      "epoch": 0.7665875807903133,
      "grad_norm": 0.07532403764829511,
      "learning_rate": 1.5698228854182618e-06,
      "loss": 0.0013,
      "step": 46850
    },
    {
      "epoch": 0.7667512067413892,
      "grad_norm": 0.12189073409820742,
      "learning_rate": 1.5677456286349363e-06,
      "loss": 0.0019,
      "step": 46860
    },
    {
      "epoch": 0.7669148326924651,
      "grad_norm": 0.176215294450804,
      "learning_rate": 1.5656694915639015e-06,
      "loss": 0.0023,
      "step": 46870
    },
    {
      "epoch": 0.7670784586435409,
      "grad_norm": 0.13490310675082046,
      "learning_rate": 1.56359447488246e-06,
      "loss": 0.0017,
      "step": 46880
    },
    {
      "epoch": 0.7672420845946167,
      "grad_norm": 0.07373989019568637,
      "learning_rate": 1.561520579267547e-06,
      "loss": 0.002,
      "step": 46890
    },
    {
      "epoch": 0.7674057105456925,
      "grad_norm": 0.37681341863303214,
      "learning_rate": 1.559447805395738e-06,
      "loss": 0.0028,
      "step": 46900
    },
    {
      "epoch": 0.7675693364967684,
      "grad_norm": 0.10340771957578078,
      "learning_rate": 1.5573761539432376e-06,
      "loss": 0.0019,
      "step": 46910
    },
    {
      "epoch": 0.7677329624478443,
      "grad_norm": 0.10695387173693185,
      "learning_rate": 1.555305625585885e-06,
      "loss": 0.0023,
      "step": 46920
    },
    {
      "epoch": 0.76789658839892,
      "grad_norm": 0.16248780928775,
      "learning_rate": 1.5532362209991536e-06,
      "loss": 0.0021,
      "step": 46930
    },
    {
      "epoch": 0.7680602143499959,
      "grad_norm": 0.24593994420605056,
      "learning_rate": 1.5511679408581499e-06,
      "loss": 0.0036,
      "step": 46940
    },
    {
      "epoch": 0.7682238403010717,
      "grad_norm": 0.09380877185707595,
      "learning_rate": 1.5491007858376122e-06,
      "loss": 0.003,
      "step": 46950
    },
    {
      "epoch": 0.7683874662521476,
      "grad_norm": 0.18750704195515064,
      "learning_rate": 1.5470347566119166e-06,
      "loss": 0.004,
      "step": 46960
    },
    {
      "epoch": 0.7685510922032235,
      "grad_norm": 0.13081269984348015,
      "learning_rate": 1.5449698538550667e-06,
      "loss": 0.0017,
      "step": 46970
    },
    {
      "epoch": 0.7687147181542993,
      "grad_norm": 0.1868569686816854,
      "learning_rate": 1.5429060782407006e-06,
      "loss": 0.0015,
      "step": 46980
    },
    {
      "epoch": 0.7688783441053751,
      "grad_norm": 0.2702879995099629,
      "learning_rate": 1.5408434304420889e-06,
      "loss": 0.0016,
      "step": 46990
    },
    {
      "epoch": 0.7690419700564509,
      "grad_norm": 0.054722269961225894,
      "learning_rate": 1.5387819111321334e-06,
      "loss": 0.0027,
      "step": 47000
    },
    {
      "epoch": 0.7692055960075268,
      "grad_norm": 0.03239142968600071,
      "learning_rate": 1.5367215209833675e-06,
      "loss": 0.0017,
      "step": 47010
    },
    {
      "epoch": 0.7693692219586027,
      "grad_norm": 0.05714288830972208,
      "learning_rate": 1.5346622606679596e-06,
      "loss": 0.0016,
      "step": 47020
    },
    {
      "epoch": 0.7695328479096785,
      "grad_norm": 0.09679205082249472,
      "learning_rate": 1.532604130857706e-06,
      "loss": 0.0027,
      "step": 47030
    },
    {
      "epoch": 0.7696964738607543,
      "grad_norm": 0.05487403771481312,
      "learning_rate": 1.5305471322240346e-06,
      "loss": 0.0015,
      "step": 47040
    },
    {
      "epoch": 0.7698600998118301,
      "grad_norm": 0.012955878998263351,
      "learning_rate": 1.5284912654380046e-06,
      "loss": 0.0058,
      "step": 47050
    },
    {
      "epoch": 0.770023725762906,
      "grad_norm": 0.0852189414542244,
      "learning_rate": 1.5264365311703067e-06,
      "loss": 0.0015,
      "step": 47060
    },
    {
      "epoch": 0.7701873517139818,
      "grad_norm": 0.07220694492826542,
      "learning_rate": 1.5243829300912594e-06,
      "loss": 0.0027,
      "step": 47070
    },
    {
      "epoch": 0.7703509776650577,
      "grad_norm": 0.0778743303377901,
      "learning_rate": 1.522330462870818e-06,
      "loss": 0.0016,
      "step": 47080
    },
    {
      "epoch": 0.7705146036161336,
      "grad_norm": 0.18949959868428612,
      "learning_rate": 1.5202791301785613e-06,
      "loss": 0.0021,
      "step": 47090
    },
    {
      "epoch": 0.7706782295672093,
      "grad_norm": 0.08432579281828237,
      "learning_rate": 1.518228932683702e-06,
      "loss": 0.003,
      "step": 47100
    },
    {
      "epoch": 0.7708418555182852,
      "grad_norm": 0.017073098747401177,
      "learning_rate": 1.5161798710550763e-06,
      "loss": 0.002,
      "step": 47110
    },
    {
      "epoch": 0.771005481469361,
      "grad_norm": 0.19474752412318322,
      "learning_rate": 1.5141319459611581e-06,
      "loss": 0.0018,
      "step": 47120
    },
    {
      "epoch": 0.7711691074204369,
      "grad_norm": 0.05104824485267002,
      "learning_rate": 1.5120851580700463e-06,
      "loss": 0.0034,
      "step": 47130
    },
    {
      "epoch": 0.7713327333715128,
      "grad_norm": 0.11746610436212994,
      "learning_rate": 1.5100395080494673e-06,
      "loss": 0.0023,
      "step": 47140
    },
    {
      "epoch": 0.7714963593225885,
      "grad_norm": 0.08425259258470946,
      "learning_rate": 1.5079949965667818e-06,
      "loss": 0.0023,
      "step": 47150
    },
    {
      "epoch": 0.7716599852736644,
      "grad_norm": 0.17879719345724795,
      "learning_rate": 1.5059516242889743e-06,
      "loss": 0.0015,
      "step": 47160
    },
    {
      "epoch": 0.7718236112247402,
      "grad_norm": 0.05076336929818043,
      "learning_rate": 1.5039093918826553e-06,
      "loss": 0.0014,
      "step": 47170
    },
    {
      "epoch": 0.7719872371758161,
      "grad_norm": 0.17239404813424308,
      "learning_rate": 1.501868300014071e-06,
      "loss": 0.0016,
      "step": 47180
    },
    {
      "epoch": 0.772150863126892,
      "grad_norm": 0.30236201269090357,
      "learning_rate": 1.4998283493490907e-06,
      "loss": 0.0028,
      "step": 47190
    },
    {
      "epoch": 0.7723144890779677,
      "grad_norm": 0.22153207285212329,
      "learning_rate": 1.49778954055321e-06,
      "loss": 0.003,
      "step": 47200
    },
    {
      "epoch": 0.7724781150290436,
      "grad_norm": 0.09080062709685986,
      "learning_rate": 1.4957518742915577e-06,
      "loss": 0.0014,
      "step": 47210
    },
    {
      "epoch": 0.7726417409801194,
      "grad_norm": 0.1146970518033467,
      "learning_rate": 1.4937153512288866e-06,
      "loss": 0.0019,
      "step": 47220
    },
    {
      "epoch": 0.7728053669311953,
      "grad_norm": 0.3137666781138699,
      "learning_rate": 1.4916799720295706e-06,
      "loss": 0.0024,
      "step": 47230
    },
    {
      "epoch": 0.7729689928822712,
      "grad_norm": 0.35709372668557193,
      "learning_rate": 1.4896457373576216e-06,
      "loss": 0.0022,
      "step": 47240
    },
    {
      "epoch": 0.7731326188333469,
      "grad_norm": 0.11045829800627718,
      "learning_rate": 1.4876126478766717e-06,
      "loss": 0.0025,
      "step": 47250
    },
    {
      "epoch": 0.7732962447844228,
      "grad_norm": 0.12589074670583394,
      "learning_rate": 1.4855807042499782e-06,
      "loss": 0.0011,
      "step": 47260
    },
    {
      "epoch": 0.7734598707354986,
      "grad_norm": 0.2564285886702629,
      "learning_rate": 1.4835499071404298e-06,
      "loss": 0.0033,
      "step": 47270
    },
    {
      "epoch": 0.7736234966865745,
      "grad_norm": 0.15687589364817492,
      "learning_rate": 1.4815202572105381e-06,
      "loss": 0.0016,
      "step": 47280
    },
    {
      "epoch": 0.7737871226376504,
      "grad_norm": 0.12472476213384284,
      "learning_rate": 1.479491755122436e-06,
      "loss": 0.0025,
      "step": 47290
    },
    {
      "epoch": 0.7739507485887261,
      "grad_norm": 0.055125679180532,
      "learning_rate": 1.4774644015378913e-06,
      "loss": 0.0021,
      "step": 47300
    },
    {
      "epoch": 0.774114374539802,
      "grad_norm": 0.04131063431432316,
      "learning_rate": 1.4754381971182901e-06,
      "loss": 0.0012,
      "step": 47310
    },
    {
      "epoch": 0.7742780004908778,
      "grad_norm": 0.3233626509152186,
      "learning_rate": 1.473413142524645e-06,
      "loss": 0.0019,
      "step": 47320
    },
    {
      "epoch": 0.7744416264419537,
      "grad_norm": 0.12489828881837006,
      "learning_rate": 1.4713892384175965e-06,
      "loss": 0.0021,
      "step": 47330
    },
    {
      "epoch": 0.7746052523930296,
      "grad_norm": 0.0837679681996852,
      "learning_rate": 1.4693664854574085e-06,
      "loss": 0.0025,
      "step": 47340
    },
    {
      "epoch": 0.7747688783441053,
      "grad_norm": 0.11027716071468342,
      "learning_rate": 1.4673448843039627e-06,
      "loss": 0.0019,
      "step": 47350
    },
    {
      "epoch": 0.7749325042951812,
      "grad_norm": 0.16096349194869602,
      "learning_rate": 1.4653244356167761e-06,
      "loss": 0.0015,
      "step": 47360
    },
    {
      "epoch": 0.775096130246257,
      "grad_norm": 0.06365189842850415,
      "learning_rate": 1.463305140054983e-06,
      "loss": 0.002,
      "step": 47370
    },
    {
      "epoch": 0.7752597561973329,
      "grad_norm": 0.029708642818202652,
      "learning_rate": 1.4612869982773403e-06,
      "loss": 0.0015,
      "step": 47380
    },
    {
      "epoch": 0.7754233821484088,
      "grad_norm": 0.3175061481642342,
      "learning_rate": 1.4592700109422352e-06,
      "loss": 0.0014,
      "step": 47390
    },
    {
      "epoch": 0.7755870080994846,
      "grad_norm": 0.10276834131282311,
      "learning_rate": 1.4572541787076732e-06,
      "loss": 0.0015,
      "step": 47400
    },
    {
      "epoch": 0.7757506340505604,
      "grad_norm": 0.0933544847800168,
      "learning_rate": 1.4552395022312804e-06,
      "loss": 0.0031,
      "step": 47410
    },
    {
      "epoch": 0.7759142600016362,
      "grad_norm": 0.09320693844665237,
      "learning_rate": 1.4532259821703133e-06,
      "loss": 0.0015,
      "step": 47420
    },
    {
      "epoch": 0.7760778859527121,
      "grad_norm": 0.09729233353527017,
      "learning_rate": 1.451213619181645e-06,
      "loss": 0.0014,
      "step": 47430
    },
    {
      "epoch": 0.776241511903788,
      "grad_norm": 0.38660681675956393,
      "learning_rate": 1.4492024139217747e-06,
      "loss": 0.0032,
      "step": 47440
    },
    {
      "epoch": 0.7764051378548638,
      "grad_norm": 0.17431782836530066,
      "learning_rate": 1.4471923670468202e-06,
      "loss": 0.0017,
      "step": 47450
    },
    {
      "epoch": 0.7765687638059396,
      "grad_norm": 0.2299255608305766,
      "learning_rate": 1.445183479212529e-06,
      "loss": 0.0029,
      "step": 47460
    },
    {
      "epoch": 0.7767323897570154,
      "grad_norm": 0.25449690777728545,
      "learning_rate": 1.44317575107426e-06,
      "loss": 0.003,
      "step": 47470
    },
    {
      "epoch": 0.7768960157080913,
      "grad_norm": 0.23810321308908192,
      "learning_rate": 1.4411691832869989e-06,
      "loss": 0.0018,
      "step": 47480
    },
    {
      "epoch": 0.7770596416591672,
      "grad_norm": 0.09106636597409241,
      "learning_rate": 1.4391637765053562e-06,
      "loss": 0.0023,
      "step": 47490
    },
    {
      "epoch": 0.777223267610243,
      "grad_norm": 0.14903263089934696,
      "learning_rate": 1.4371595313835596e-06,
      "loss": 0.0014,
      "step": 47500
    },
    {
      "epoch": 0.7773868935613188,
      "grad_norm": 0.1791936543731685,
      "learning_rate": 1.4351564485754577e-06,
      "loss": 0.0019,
      "step": 47510
    },
    {
      "epoch": 0.7775505195123946,
      "grad_norm": 0.14007063752594243,
      "learning_rate": 1.433154528734521e-06,
      "loss": 0.0016,
      "step": 47520
    },
    {
      "epoch": 0.7777141454634705,
      "grad_norm": 0.21569852442636941,
      "learning_rate": 1.4311537725138402e-06,
      "loss": 0.0021,
      "step": 47530
    },
    {
      "epoch": 0.7778777714145464,
      "grad_norm": 0.1477813097012733,
      "learning_rate": 1.4291541805661264e-06,
      "loss": 0.0044,
      "step": 47540
    },
    {
      "epoch": 0.7780413973656222,
      "grad_norm": 0.06717715660781988,
      "learning_rate": 1.4271557535437125e-06,
      "loss": 0.002,
      "step": 47550
    },
    {
      "epoch": 0.778205023316698,
      "grad_norm": 0.284568101737055,
      "learning_rate": 1.4251584920985494e-06,
      "loss": 0.0023,
      "step": 47560
    },
    {
      "epoch": 0.7783686492677738,
      "grad_norm": 0.14013774290816464,
      "learning_rate": 1.4231623968822083e-06,
      "loss": 0.0022,
      "step": 47570
    },
    {
      "epoch": 0.7785322752188497,
      "grad_norm": 0.038572173764087905,
      "learning_rate": 1.42116746854588e-06,
      "loss": 0.002,
      "step": 47580
    },
    {
      "epoch": 0.7786959011699256,
      "grad_norm": 0.17048032316101788,
      "learning_rate": 1.4191737077403738e-06,
      "loss": 0.0029,
      "step": 47590
    },
    {
      "epoch": 0.7788595271210014,
      "grad_norm": 0.17016912181734972,
      "learning_rate": 1.4171811151161186e-06,
      "loss": 0.0013,
      "step": 47600
    },
    {
      "epoch": 0.7790231530720773,
      "grad_norm": 0.09619362880440897,
      "learning_rate": 1.4151896913231645e-06,
      "loss": 0.0013,
      "step": 47610
    },
    {
      "epoch": 0.779186779023153,
      "grad_norm": 0.24352342912027405,
      "learning_rate": 1.4131994370111773e-06,
      "loss": 0.0026,
      "step": 47620
    },
    {
      "epoch": 0.7793504049742289,
      "grad_norm": 0.20584183996451946,
      "learning_rate": 1.4112103528294418e-06,
      "loss": 0.0016,
      "step": 47630
    },
    {
      "epoch": 0.7795140309253048,
      "grad_norm": 0.1416382089350118,
      "learning_rate": 1.4092224394268629e-06,
      "loss": 0.002,
      "step": 47640
    },
    {
      "epoch": 0.7796776568763806,
      "grad_norm": 0.103123160930211,
      "learning_rate": 1.4072356974519601e-06,
      "loss": 0.0023,
      "step": 47650
    },
    {
      "epoch": 0.7798412828274565,
      "grad_norm": 0.18341025861418078,
      "learning_rate": 1.405250127552873e-06,
      "loss": 0.0018,
      "step": 47660
    },
    {
      "epoch": 0.7800049087785322,
      "grad_norm": 0.10167835698104476,
      "learning_rate": 1.4032657303773606e-06,
      "loss": 0.0017,
      "step": 47670
    },
    {
      "epoch": 0.7801685347296081,
      "grad_norm": 0.19719538253535882,
      "learning_rate": 1.401282506572797e-06,
      "loss": 0.0031,
      "step": 47680
    },
    {
      "epoch": 0.780332160680684,
      "grad_norm": 0.23152140546286598,
      "learning_rate": 1.3993004567861728e-06,
      "loss": 0.0023,
      "step": 47690
    },
    {
      "epoch": 0.7804957866317598,
      "grad_norm": 0.07451915494433714,
      "learning_rate": 1.3973195816640973e-06,
      "loss": 0.0015,
      "step": 47700
    },
    {
      "epoch": 0.7806594125828357,
      "grad_norm": 0.30495971623377205,
      "learning_rate": 1.3953398818527959e-06,
      "loss": 0.0035,
      "step": 47710
    },
    {
      "epoch": 0.7808230385339114,
      "grad_norm": 0.22746142323368346,
      "learning_rate": 1.3933613579981093e-06,
      "loss": 0.0024,
      "step": 47720
    },
    {
      "epoch": 0.7809866644849873,
      "grad_norm": 0.22205970809478168,
      "learning_rate": 1.391384010745498e-06,
      "loss": 0.0026,
      "step": 47730
    },
    {
      "epoch": 0.7811502904360632,
      "grad_norm": 0.12180686757075486,
      "learning_rate": 1.3894078407400362e-06,
      "loss": 0.0037,
      "step": 47740
    },
    {
      "epoch": 0.781313916387139,
      "grad_norm": 0.05990027464909791,
      "learning_rate": 1.3874328486264143e-06,
      "loss": 0.0012,
      "step": 47750
    },
    {
      "epoch": 0.7814775423382149,
      "grad_norm": 0.21658144097064255,
      "learning_rate": 1.3854590350489378e-06,
      "loss": 0.0026,
      "step": 47760
    },
    {
      "epoch": 0.7816411682892906,
      "grad_norm": 0.11016702789422426,
      "learning_rate": 1.3834864006515292e-06,
      "loss": 0.0028,
      "step": 47770
    },
    {
      "epoch": 0.7818047942403665,
      "grad_norm": 0.20510262414746427,
      "learning_rate": 1.381514946077725e-06,
      "loss": 0.0048,
      "step": 47780
    },
    {
      "epoch": 0.7819684201914424,
      "grad_norm": 0.06465576354909051,
      "learning_rate": 1.3795446719706767e-06,
      "loss": 0.0013,
      "step": 47790
    },
    {
      "epoch": 0.7821320461425182,
      "grad_norm": 0.23943426033453444,
      "learning_rate": 1.3775755789731533e-06,
      "loss": 0.0037,
      "step": 47800
    },
    {
      "epoch": 0.7822956720935941,
      "grad_norm": 0.2811009319554033,
      "learning_rate": 1.375607667727536e-06,
      "loss": 0.0034,
      "step": 47810
    },
    {
      "epoch": 0.7824592980446698,
      "grad_norm": 0.15350533999288016,
      "learning_rate": 1.3736409388758203e-06,
      "loss": 0.0029,
      "step": 47820
    },
    {
      "epoch": 0.7826229239957457,
      "grad_norm": 0.11218334863159217,
      "learning_rate": 1.3716753930596177e-06,
      "loss": 0.0013,
      "step": 47830
    },
    {
      "epoch": 0.7827865499468216,
      "grad_norm": 0.06904977146678239,
      "learning_rate": 1.3697110309201518e-06,
      "loss": 0.0024,
      "step": 47840
    },
    {
      "epoch": 0.7829501758978974,
      "grad_norm": 0.06278450412118958,
      "learning_rate": 1.3677478530982597e-06,
      "loss": 0.0018,
      "step": 47850
    },
    {
      "epoch": 0.7831138018489733,
      "grad_norm": 0.09427927705302362,
      "learning_rate": 1.3657858602343972e-06,
      "loss": 0.002,
      "step": 47860
    },
    {
      "epoch": 0.783277427800049,
      "grad_norm": 0.14483117646688543,
      "learning_rate": 1.363825052968627e-06,
      "loss": 0.0023,
      "step": 47870
    },
    {
      "epoch": 0.7834410537511249,
      "grad_norm": 0.13039105429753506,
      "learning_rate": 1.3618654319406287e-06,
      "loss": 0.0026,
      "step": 47880
    },
    {
      "epoch": 0.7836046797022008,
      "grad_norm": 0.3842447508638256,
      "learning_rate": 1.3599069977896939e-06,
      "loss": 0.001,
      "step": 47890
    },
    {
      "epoch": 0.7837683056532766,
      "grad_norm": 0.18417246564249867,
      "learning_rate": 1.3579497511547268e-06,
      "loss": 0.0022,
      "step": 47900
    },
    {
      "epoch": 0.7839319316043525,
      "grad_norm": 0.06308406191247852,
      "learning_rate": 1.3559936926742433e-06,
      "loss": 0.0023,
      "step": 47910
    },
    {
      "epoch": 0.7840955575554283,
      "grad_norm": 0.05545072416061435,
      "learning_rate": 1.354038822986376e-06,
      "loss": 0.0022,
      "step": 47920
    },
    {
      "epoch": 0.7842591835065041,
      "grad_norm": 0.045346089711888325,
      "learning_rate": 1.3520851427288651e-06,
      "loss": 0.0023,
      "step": 47930
    },
    {
      "epoch": 0.7844228094575799,
      "grad_norm": 0.12886169849205031,
      "learning_rate": 1.3501326525390635e-06,
      "loss": 0.0056,
      "step": 47940
    },
    {
      "epoch": 0.7845864354086558,
      "grad_norm": 0.10413595122795785,
      "learning_rate": 1.3481813530539368e-06,
      "loss": 0.0035,
      "step": 47950
    },
    {
      "epoch": 0.7847500613597317,
      "grad_norm": 0.05225019331871401,
      "learning_rate": 1.3462312449100628e-06,
      "loss": 0.0017,
      "step": 47960
    },
    {
      "epoch": 0.7849136873108075,
      "grad_norm": 0.09285602950112436,
      "learning_rate": 1.3442823287436274e-06,
      "loss": 0.0025,
      "step": 47970
    },
    {
      "epoch": 0.7850773132618833,
      "grad_norm": 0.5464102477723739,
      "learning_rate": 1.3423346051904334e-06,
      "loss": 0.002,
      "step": 47980
    },
    {
      "epoch": 0.7852409392129591,
      "grad_norm": 0.3564787486186036,
      "learning_rate": 1.3403880748858894e-06,
      "loss": 0.0025,
      "step": 47990
    },
    {
      "epoch": 0.785404565164035,
      "grad_norm": 0.15288055382589755,
      "learning_rate": 1.3384427384650166e-06,
      "loss": 0.0033,
      "step": 48000
    },
    {
      "epoch": 0.7855681911151109,
      "grad_norm": 0.17280972223040875,
      "learning_rate": 1.3364985965624473e-06,
      "loss": 0.0024,
      "step": 48010
    },
    {
      "epoch": 0.7857318170661867,
      "grad_norm": 0.0591808163754419,
      "learning_rate": 1.3345556498124228e-06,
      "loss": 0.0015,
      "step": 48020
    },
    {
      "epoch": 0.7858954430172626,
      "grad_norm": 0.23056957798171168,
      "learning_rate": 1.332613898848793e-06,
      "loss": 0.0024,
      "step": 48030
    },
    {
      "epoch": 0.7860590689683383,
      "grad_norm": 0.38065396519306094,
      "learning_rate": 1.330673344305024e-06,
      "loss": 0.0032,
      "step": 48040
    },
    {
      "epoch": 0.7862226949194142,
      "grad_norm": 0.09354012381691237,
      "learning_rate": 1.3287339868141851e-06,
      "loss": 0.0013,
      "step": 48050
    },
    {
      "epoch": 0.7863863208704901,
      "grad_norm": 0.062426577917024884,
      "learning_rate": 1.3267958270089582e-06,
      "loss": 0.0008,
      "step": 48060
    },
    {
      "epoch": 0.7865499468215659,
      "grad_norm": 0.14164616216577655,
      "learning_rate": 1.324858865521632e-06,
      "loss": 0.0024,
      "step": 48070
    },
    {
      "epoch": 0.7867135727726418,
      "grad_norm": 0.1378638949127868,
      "learning_rate": 1.3229231029841077e-06,
      "loss": 0.0017,
      "step": 48080
    },
    {
      "epoch": 0.7868771987237175,
      "grad_norm": 0.10423390547344576,
      "learning_rate": 1.3209885400278933e-06,
      "loss": 0.0013,
      "step": 48090
    },
    {
      "epoch": 0.7870408246747934,
      "grad_norm": 0.11497434302093766,
      "learning_rate": 1.3190551772841032e-06,
      "loss": 0.0018,
      "step": 48100
    },
    {
      "epoch": 0.7872044506258693,
      "grad_norm": 0.14612236400511033,
      "learning_rate": 1.3171230153834669e-06,
      "loss": 0.0014,
      "step": 48110
    },
    {
      "epoch": 0.7873680765769451,
      "grad_norm": 0.0469320150030638,
      "learning_rate": 1.3151920549563158e-06,
      "loss": 0.0023,
      "step": 48120
    },
    {
      "epoch": 0.787531702528021,
      "grad_norm": 0.003913219145657117,
      "learning_rate": 1.313262296632592e-06,
      "loss": 0.0012,
      "step": 48130
    },
    {
      "epoch": 0.7876953284790967,
      "grad_norm": 0.08066569413358624,
      "learning_rate": 1.3113337410418453e-06,
      "loss": 0.0015,
      "step": 48140
    },
    {
      "epoch": 0.7878589544301726,
      "grad_norm": 0.13076089636497218,
      "learning_rate": 1.3094063888132319e-06,
      "loss": 0.0017,
      "step": 48150
    },
    {
      "epoch": 0.7880225803812485,
      "grad_norm": 0.06958654413189007,
      "learning_rate": 1.307480240575516e-06,
      "loss": 0.0019,
      "step": 48160
    },
    {
      "epoch": 0.7881862063323243,
      "grad_norm": 0.293656214552158,
      "learning_rate": 1.3055552969570717e-06,
      "loss": 0.0022,
      "step": 48170
    },
    {
      "epoch": 0.7883498322834002,
      "grad_norm": 0.017069204530643426,
      "learning_rate": 1.3036315585858766e-06,
      "loss": 0.0023,
      "step": 48180
    },
    {
      "epoch": 0.7885134582344759,
      "grad_norm": 0.13197021774189005,
      "learning_rate": 1.301709026089516e-06,
      "loss": 0.0017,
      "step": 48190
    },
    {
      "epoch": 0.7886770841855518,
      "grad_norm": 0.022928867645394774,
      "learning_rate": 1.2997877000951825e-06,
      "loss": 0.0034,
      "step": 48200
    },
    {
      "epoch": 0.7888407101366277,
      "grad_norm": 0.0703507963375967,
      "learning_rate": 1.2978675812296742e-06,
      "loss": 0.002,
      "step": 48210
    },
    {
      "epoch": 0.7890043360877035,
      "grad_norm": 0.18970392836297792,
      "learning_rate": 1.2959486701193951e-06,
      "loss": 0.0009,
      "step": 48220
    },
    {
      "epoch": 0.7891679620387794,
      "grad_norm": 0.04444627565394381,
      "learning_rate": 1.294030967390359e-06,
      "loss": 0.0025,
      "step": 48230
    },
    {
      "epoch": 0.7893315879898551,
      "grad_norm": 0.09752282742047552,
      "learning_rate": 1.2921144736681817e-06,
      "loss": 0.0019,
      "step": 48240
    },
    {
      "epoch": 0.789495213940931,
      "grad_norm": 0.5250058983449449,
      "learning_rate": 1.2901991895780818e-06,
      "loss": 0.0016,
      "step": 48250
    },
    {
      "epoch": 0.7896588398920069,
      "grad_norm": 0.017380514340823345,
      "learning_rate": 1.2882851157448906e-06,
      "loss": 0.0028,
      "step": 48260
    },
    {
      "epoch": 0.7898224658430827,
      "grad_norm": 0.08951283146517515,
      "learning_rate": 1.2863722527930401e-06,
      "loss": 0.0029,
      "step": 48270
    },
    {
      "epoch": 0.7899860917941586,
      "grad_norm": 0.15813523519123485,
      "learning_rate": 1.2844606013465655e-06,
      "loss": 0.0019,
      "step": 48280
    },
    {
      "epoch": 0.7901497177452343,
      "grad_norm": 0.07379427184668581,
      "learning_rate": 1.2825501620291136e-06,
      "loss": 0.0016,
      "step": 48290
    },
    {
      "epoch": 0.7903133436963102,
      "grad_norm": 0.23647016431558107,
      "learning_rate": 1.28064093546393e-06,
      "loss": 0.0034,
      "step": 48300
    },
    {
      "epoch": 0.7904769696473861,
      "grad_norm": 0.049473921908787875,
      "learning_rate": 1.2787329222738626e-06,
      "loss": 0.0015,
      "step": 48310
    },
    {
      "epoch": 0.7906405955984619,
      "grad_norm": 0.2953573060433273,
      "learning_rate": 1.2768261230813717e-06,
      "loss": 0.0031,
      "step": 48320
    },
    {
      "epoch": 0.7908042215495378,
      "grad_norm": 0.06211686656672406,
      "learning_rate": 1.274920538508514e-06,
      "loss": 0.002,
      "step": 48330
    },
    {
      "epoch": 0.7909678475006136,
      "grad_norm": 0.12325816364804497,
      "learning_rate": 1.2730161691769526e-06,
      "loss": 0.0025,
      "step": 48340
    },
    {
      "epoch": 0.7911314734516894,
      "grad_norm": 0.1411198920338921,
      "learning_rate": 1.2711130157079566e-06,
      "loss": 0.0034,
      "step": 48350
    },
    {
      "epoch": 0.7912950994027653,
      "grad_norm": 0.20799919714040022,
      "learning_rate": 1.2692110787223965e-06,
      "loss": 0.0035,
      "step": 48360
    },
    {
      "epoch": 0.7914587253538411,
      "grad_norm": 0.09009724020755089,
      "learning_rate": 1.267310358840741e-06,
      "loss": 0.003,
      "step": 48370
    },
    {
      "epoch": 0.791622351304917,
      "grad_norm": 0.2481705526184436,
      "learning_rate": 1.2654108566830702e-06,
      "loss": 0.0028,
      "step": 48380
    },
    {
      "epoch": 0.7917859772559928,
      "grad_norm": 0.07467848485961295,
      "learning_rate": 1.2635125728690627e-06,
      "loss": 0.0027,
      "step": 48390
    },
    {
      "epoch": 0.7919496032070686,
      "grad_norm": 0.05418448507075133,
      "learning_rate": 1.2616155080179982e-06,
      "loss": 0.0012,
      "step": 48400
    },
    {
      "epoch": 0.7921132291581445,
      "grad_norm": 0.2845613768330517,
      "learning_rate": 1.2597196627487629e-06,
      "loss": 0.0017,
      "step": 48410
    },
    {
      "epoch": 0.7922768551092203,
      "grad_norm": 0.11375271010381821,
      "learning_rate": 1.2578250376798434e-06,
      "loss": 0.0014,
      "step": 48420
    },
    {
      "epoch": 0.7924404810602962,
      "grad_norm": 0.1341817335122393,
      "learning_rate": 1.2559316334293248e-06,
      "loss": 0.0029,
      "step": 48430
    },
    {
      "epoch": 0.792604107011372,
      "grad_norm": 0.2924596770633388,
      "learning_rate": 1.254039450614896e-06,
      "loss": 0.0019,
      "step": 48440
    },
    {
      "epoch": 0.7927677329624478,
      "grad_norm": 0.04525819640855382,
      "learning_rate": 1.2521484898538523e-06,
      "loss": 0.0013,
      "step": 48450
    },
    {
      "epoch": 0.7929313589135237,
      "grad_norm": 0.0639154925905611,
      "learning_rate": 1.2502587517630832e-06,
      "loss": 0.0013,
      "step": 48460
    },
    {
      "epoch": 0.7930949848645995,
      "grad_norm": 0.16817728882628474,
      "learning_rate": 1.2483702369590823e-06,
      "loss": 0.0018,
      "step": 48470
    },
    {
      "epoch": 0.7932586108156754,
      "grad_norm": 0.26547309580554723,
      "learning_rate": 1.2464829460579475e-06,
      "loss": 0.0014,
      "step": 48480
    },
    {
      "epoch": 0.7934222367667512,
      "grad_norm": 0.04125464367940636,
      "learning_rate": 1.2445968796753704e-06,
      "loss": 0.0023,
      "step": 48490
    },
    {
      "epoch": 0.793585862717827,
      "grad_norm": 0.13973628683449885,
      "learning_rate": 1.242712038426646e-06,
      "loss": 0.0026,
      "step": 48500
    },
    {
      "epoch": 0.7937494886689029,
      "grad_norm": 0.13808481908338202,
      "learning_rate": 1.2408284229266748e-06,
      "loss": 0.0021,
      "step": 48510
    },
    {
      "epoch": 0.7939131146199787,
      "grad_norm": 0.25530876687539716,
      "learning_rate": 1.23894603378995e-06,
      "loss": 0.0056,
      "step": 48520
    },
    {
      "epoch": 0.7940767405710546,
      "grad_norm": 0.10519822627650355,
      "learning_rate": 1.2370648716305672e-06,
      "loss": 0.0014,
      "step": 48530
    },
    {
      "epoch": 0.7942403665221304,
      "grad_norm": 0.16921569586770094,
      "learning_rate": 1.2351849370622266e-06,
      "loss": 0.002,
      "step": 48540
    },
    {
      "epoch": 0.7944039924732063,
      "grad_norm": 0.27240336912427127,
      "learning_rate": 1.2333062306982186e-06,
      "loss": 0.0033,
      "step": 48550
    },
    {
      "epoch": 0.7945676184242821,
      "grad_norm": 0.2090694531639897,
      "learning_rate": 1.2314287531514385e-06,
      "loss": 0.0033,
      "step": 48560
    },
    {
      "epoch": 0.7947312443753579,
      "grad_norm": 0.15972070993977083,
      "learning_rate": 1.2295525050343832e-06,
      "loss": 0.0028,
      "step": 48570
    },
    {
      "epoch": 0.7948948703264338,
      "grad_norm": 0.12952037646266074,
      "learning_rate": 1.227677486959144e-06,
      "loss": 0.0018,
      "step": 48580
    },
    {
      "epoch": 0.7950584962775096,
      "grad_norm": 0.03591734762951992,
      "learning_rate": 1.2258036995374102e-06,
      "loss": 0.0015,
      "step": 48590
    },
    {
      "epoch": 0.7952221222285855,
      "grad_norm": 0.2014415389307618,
      "learning_rate": 1.2239311433804769e-06,
      "loss": 0.0016,
      "step": 48600
    },
    {
      "epoch": 0.7953857481796613,
      "grad_norm": 0.004998662143198341,
      "learning_rate": 1.2220598190992277e-06,
      "loss": 0.0013,
      "step": 48610
    },
    {
      "epoch": 0.7955493741307371,
      "grad_norm": 0.08218378151299863,
      "learning_rate": 1.220189727304149e-06,
      "loss": 0.0017,
      "step": 48620
    },
    {
      "epoch": 0.795713000081813,
      "grad_norm": 0.226071398313477,
      "learning_rate": 1.2183208686053289e-06,
      "loss": 0.0022,
      "step": 48630
    },
    {
      "epoch": 0.7958766260328888,
      "grad_norm": 0.2575719719662208,
      "learning_rate": 1.2164532436124472e-06,
      "loss": 0.0016,
      "step": 48640
    },
    {
      "epoch": 0.7960402519839647,
      "grad_norm": 0.5916270759039125,
      "learning_rate": 1.2145868529347821e-06,
      "loss": 0.0027,
      "step": 48650
    },
    {
      "epoch": 0.7962038779350405,
      "grad_norm": 0.1025646487770363,
      "learning_rate": 1.212721697181215e-06,
      "loss": 0.0028,
      "step": 48660
    },
    {
      "epoch": 0.7963675038861163,
      "grad_norm": 0.0919091202605032,
      "learning_rate": 1.2108577769602165e-06,
      "loss": 0.0018,
      "step": 48670
    },
    {
      "epoch": 0.7965311298371922,
      "grad_norm": 0.025338708012351826,
      "learning_rate": 1.208995092879856e-06,
      "loss": 0.003,
      "step": 48680
    },
    {
      "epoch": 0.796694755788268,
      "grad_norm": 0.04743675698561767,
      "learning_rate": 1.2071336455478055e-06,
      "loss": 0.0023,
      "step": 48690
    },
    {
      "epoch": 0.7968583817393439,
      "grad_norm": 0.372528354296269,
      "learning_rate": 1.205273435571327e-06,
      "loss": 0.0046,
      "step": 48700
    },
    {
      "epoch": 0.7970220076904198,
      "grad_norm": 0.5094668952787667,
      "learning_rate": 1.2034144635572797e-06,
      "loss": 0.0068,
      "step": 48710
    },
    {
      "epoch": 0.7971856336414955,
      "grad_norm": 0.1817852062929723,
      "learning_rate": 1.2015567301121256e-06,
      "loss": 0.0029,
      "step": 48720
    },
    {
      "epoch": 0.7973492595925714,
      "grad_norm": 0.08401039650203053,
      "learning_rate": 1.1997002358419118e-06,
      "loss": 0.002,
      "step": 48730
    },
    {
      "epoch": 0.7975128855436472,
      "grad_norm": 0.270742669156899,
      "learning_rate": 1.197844981352289e-06,
      "loss": 0.0028,
      "step": 48740
    },
    {
      "epoch": 0.7976765114947231,
      "grad_norm": 0.16051683474496112,
      "learning_rate": 1.1959909672484998e-06,
      "loss": 0.0025,
      "step": 48750
    },
    {
      "epoch": 0.797840137445799,
      "grad_norm": 0.13389212542110335,
      "learning_rate": 1.1941381941353857e-06,
      "loss": 0.0015,
      "step": 48760
    },
    {
      "epoch": 0.7980037633968747,
      "grad_norm": 0.09398602728654756,
      "learning_rate": 1.1922866626173802e-06,
      "loss": 0.0018,
      "step": 48770
    },
    {
      "epoch": 0.7981673893479506,
      "grad_norm": 0.09871255224743344,
      "learning_rate": 1.1904363732985124e-06,
      "loss": 0.003,
      "step": 48780
    },
    {
      "epoch": 0.7983310152990264,
      "grad_norm": 0.0492863277487948,
      "learning_rate": 1.1885873267824072e-06,
      "loss": 0.0018,
      "step": 48790
    },
    {
      "epoch": 0.7984946412501023,
      "grad_norm": 0.07554320419631094,
      "learning_rate": 1.1867395236722828e-06,
      "loss": 0.0018,
      "step": 48800
    },
    {
      "epoch": 0.798658267201178,
      "grad_norm": 0.18350981573709402,
      "learning_rate": 1.1848929645709506e-06,
      "loss": 0.0019,
      "step": 48810
    },
    {
      "epoch": 0.7988218931522539,
      "grad_norm": 0.13345806362801924,
      "learning_rate": 1.1830476500808207e-06,
      "loss": 0.0012,
      "step": 48820
    },
    {
      "epoch": 0.7989855191033298,
      "grad_norm": 0.15621868766177915,
      "learning_rate": 1.1812035808038934e-06,
      "loss": 0.0044,
      "step": 48830
    },
    {
      "epoch": 0.7991491450544056,
      "grad_norm": 0.23928409511538293,
      "learning_rate": 1.1793607573417638e-06,
      "loss": 0.0036,
      "step": 48840
    },
    {
      "epoch": 0.7993127710054815,
      "grad_norm": 0.18891972137459614,
      "learning_rate": 1.1775191802956199e-06,
      "loss": 0.0017,
      "step": 48850
    },
    {
      "epoch": 0.7994763969565573,
      "grad_norm": 0.1426560669229477,
      "learning_rate": 1.1756788502662436e-06,
      "loss": 0.0025,
      "step": 48860
    },
    {
      "epoch": 0.7996400229076331,
      "grad_norm": 0.14488672582204726,
      "learning_rate": 1.1738397678540091e-06,
      "loss": 0.002,
      "step": 48870
    },
    {
      "epoch": 0.799803648858709,
      "grad_norm": 0.11709875183911246,
      "learning_rate": 1.1720019336588873e-06,
      "loss": 0.0021,
      "step": 48880
    },
    {
      "epoch": 0.7999672748097848,
      "grad_norm": 0.10438392829273505,
      "learning_rate": 1.170165348280437e-06,
      "loss": 0.0024,
      "step": 48890
    },
    {
      "epoch": 0.8001309007608607,
      "grad_norm": 0.1089760491860878,
      "learning_rate": 1.1683300123178126e-06,
      "loss": 0.0022,
      "step": 48900
    },
    {
      "epoch": 0.8002945267119365,
      "grad_norm": 0.18184087226355755,
      "learning_rate": 1.16649592636976e-06,
      "loss": 0.0019,
      "step": 48910
    },
    {
      "epoch": 0.8004581526630123,
      "grad_norm": 0.09544216569412499,
      "learning_rate": 1.1646630910346162e-06,
      "loss": 0.003,
      "step": 48920
    },
    {
      "epoch": 0.8006217786140882,
      "grad_norm": 0.10098643535187125,
      "learning_rate": 1.1628315069103108e-06,
      "loss": 0.002,
      "step": 48930
    },
    {
      "epoch": 0.800785404565164,
      "grad_norm": 0.21537063001659107,
      "learning_rate": 1.161001174594369e-06,
      "loss": 0.0026,
      "step": 48940
    },
    {
      "epoch": 0.8009490305162399,
      "grad_norm": 0.04227623803558214,
      "learning_rate": 1.1591720946839014e-06,
      "loss": 0.0007,
      "step": 48950
    },
    {
      "epoch": 0.8011126564673157,
      "grad_norm": 0.15476867935214153,
      "learning_rate": 1.1573442677756142e-06,
      "loss": 0.0024,
      "step": 48960
    },
    {
      "epoch": 0.8012762824183915,
      "grad_norm": 0.29047802757149555,
      "learning_rate": 1.1555176944658035e-06,
      "loss": 0.0041,
      "step": 48970
    },
    {
      "epoch": 0.8014399083694674,
      "grad_norm": 0.018645180535673708,
      "learning_rate": 1.1536923753503566e-06,
      "loss": 0.0012,
      "step": 48980
    },
    {
      "epoch": 0.8016035343205432,
      "grad_norm": 0.0191645132658388,
      "learning_rate": 1.151868311024749e-06,
      "loss": 0.0013,
      "step": 48990
    },
    {
      "epoch": 0.8017671602716191,
      "grad_norm": 0.12958291259007618,
      "learning_rate": 1.150045502084054e-06,
      "loss": 0.0019,
      "step": 49000
    },
    {
      "epoch": 0.8019307862226949,
      "grad_norm": 0.020998683239434567,
      "learning_rate": 1.148223949122928e-06,
      "loss": 0.0032,
      "step": 49010
    },
    {
      "epoch": 0.8020944121737708,
      "grad_norm": 0.03771036832828451,
      "learning_rate": 1.1464036527356209e-06,
      "loss": 0.0019,
      "step": 49020
    },
    {
      "epoch": 0.8022580381248466,
      "grad_norm": 0.08186291505510673,
      "learning_rate": 1.1445846135159727e-06,
      "loss": 0.0035,
      "step": 49030
    },
    {
      "epoch": 0.8024216640759224,
      "grad_norm": 0.2142463727565218,
      "learning_rate": 1.142766832057412e-06,
      "loss": 0.0031,
      "step": 49040
    },
    {
      "epoch": 0.8025852900269983,
      "grad_norm": 0.13872718864960729,
      "learning_rate": 1.1409503089529573e-06,
      "loss": 0.0029,
      "step": 49050
    },
    {
      "epoch": 0.8027489159780741,
      "grad_norm": 0.18749714463965136,
      "learning_rate": 1.1391350447952188e-06,
      "loss": 0.0023,
      "step": 49060
    },
    {
      "epoch": 0.80291254192915,
      "grad_norm": 0.2547333633417284,
      "learning_rate": 1.137321040176394e-06,
      "loss": 0.0014,
      "step": 49070
    },
    {
      "epoch": 0.8030761678802258,
      "grad_norm": 0.42253488851951854,
      "learning_rate": 1.135508295688269e-06,
      "loss": 0.0024,
      "step": 49080
    },
    {
      "epoch": 0.8032397938313016,
      "grad_norm": 0.11363707593928521,
      "learning_rate": 1.1336968119222203e-06,
      "loss": 0.0013,
      "step": 49090
    },
    {
      "epoch": 0.8034034197823775,
      "grad_norm": 0.13872117607886797,
      "learning_rate": 1.131886589469212e-06,
      "loss": 0.003,
      "step": 49100
    },
    {
      "epoch": 0.8035670457334533,
      "grad_norm": 0.3812063477244664,
      "learning_rate": 1.1300776289197979e-06,
      "loss": 0.0013,
      "step": 49110
    },
    {
      "epoch": 0.8037306716845292,
      "grad_norm": 0.03595518080274262,
      "learning_rate": 1.1282699308641166e-06,
      "loss": 0.0017,
      "step": 49120
    },
    {
      "epoch": 0.803894297635605,
      "grad_norm": 0.09008680669170954,
      "learning_rate": 1.126463495891902e-06,
      "loss": 0.0027,
      "step": 49130
    },
    {
      "epoch": 0.8040579235866808,
      "grad_norm": 0.09142113329494993,
      "learning_rate": 1.1246583245924687e-06,
      "loss": 0.0012,
      "step": 49140
    },
    {
      "epoch": 0.8042215495377567,
      "grad_norm": 0.0911831511286142,
      "learning_rate": 1.122854417554723e-06,
      "loss": 0.0026,
      "step": 49150
    },
    {
      "epoch": 0.8043851754888325,
      "grad_norm": 0.15472805919059357,
      "learning_rate": 1.1210517753671568e-06,
      "loss": 0.0023,
      "step": 49160
    },
    {
      "epoch": 0.8045488014399084,
      "grad_norm": 0.13093212419936698,
      "learning_rate": 1.1192503986178515e-06,
      "loss": 0.0042,
      "step": 49170
    },
    {
      "epoch": 0.8047124273909843,
      "grad_norm": 0.10252447479831282,
      "learning_rate": 1.1174502878944715e-06,
      "loss": 0.0014,
      "step": 49180
    },
    {
      "epoch": 0.80487605334206,
      "grad_norm": 0.08114703827453858,
      "learning_rate": 1.1156514437842753e-06,
      "loss": 0.0041,
      "step": 49190
    },
    {
      "epoch": 0.8050396792931359,
      "grad_norm": 0.2855196029497082,
      "learning_rate": 1.1138538668741011e-06,
      "loss": 0.002,
      "step": 49200
    },
    {
      "epoch": 0.8052033052442117,
      "grad_norm": 0.18086581430494428,
      "learning_rate": 1.1120575577503778e-06,
      "loss": 0.0016,
      "step": 49210
    },
    {
      "epoch": 0.8053669311952876,
      "grad_norm": 0.1264583316951292,
      "learning_rate": 1.1102625169991194e-06,
      "loss": 0.002,
      "step": 49220
    },
    {
      "epoch": 0.8055305571463635,
      "grad_norm": 0.12753249715892995,
      "learning_rate": 1.1084687452059256e-06,
      "loss": 0.0021,
      "step": 49230
    },
    {
      "epoch": 0.8056941830974392,
      "grad_norm": 0.10987794054723932,
      "learning_rate": 1.1066762429559812e-06,
      "loss": 0.0018,
      "step": 49240
    },
    {
      "epoch": 0.8058578090485151,
      "grad_norm": 0.44109008749859413,
      "learning_rate": 1.104885010834062e-06,
      "loss": 0.0041,
      "step": 49250
    },
    {
      "epoch": 0.8060214349995909,
      "grad_norm": 0.09642172589835633,
      "learning_rate": 1.1030950494245234e-06,
      "loss": 0.0021,
      "step": 49260
    },
    {
      "epoch": 0.8061850609506668,
      "grad_norm": 0.03951203650870052,
      "learning_rate": 1.1013063593113098e-06,
      "loss": 0.0008,
      "step": 49270
    },
    {
      "epoch": 0.8063486869017427,
      "grad_norm": 0.041398264768328676,
      "learning_rate": 1.0995189410779489e-06,
      "loss": 0.0013,
      "step": 49280
    },
    {
      "epoch": 0.8065123128528184,
      "grad_norm": 0.327907268720022,
      "learning_rate": 1.0977327953075544e-06,
      "loss": 0.0019,
      "step": 49290
    },
    {
      "epoch": 0.8066759388038943,
      "grad_norm": 0.21949316599751556,
      "learning_rate": 1.0959479225828234e-06,
      "loss": 0.0022,
      "step": 49300
    },
    {
      "epoch": 0.8068395647549701,
      "grad_norm": 0.26978019970528955,
      "learning_rate": 1.0941643234860422e-06,
      "loss": 0.0046,
      "step": 49310
    },
    {
      "epoch": 0.807003190706046,
      "grad_norm": 0.07665381958896401,
      "learning_rate": 1.0923819985990768e-06,
      "loss": 0.0021,
      "step": 49320
    },
    {
      "epoch": 0.8071668166571219,
      "grad_norm": 0.2542194490507373,
      "learning_rate": 1.0906009485033786e-06,
      "loss": 0.0014,
      "step": 49330
    },
    {
      "epoch": 0.8073304426081976,
      "grad_norm": 0.05549457868175632,
      "learning_rate": 1.0888211737799847e-06,
      "loss": 0.0013,
      "step": 49340
    },
    {
      "epoch": 0.8074940685592735,
      "grad_norm": 0.166608547736739,
      "learning_rate": 1.0870426750095148e-06,
      "loss": 0.0018,
      "step": 49350
    },
    {
      "epoch": 0.8076576945103493,
      "grad_norm": 0.3178274859363724,
      "learning_rate": 1.0852654527721712e-06,
      "loss": 0.0026,
      "step": 49360
    },
    {
      "epoch": 0.8078213204614252,
      "grad_norm": 0.06130743603463829,
      "learning_rate": 1.0834895076477437e-06,
      "loss": 0.0015,
      "step": 49370
    },
    {
      "epoch": 0.8079849464125011,
      "grad_norm": 0.010331530406221326,
      "learning_rate": 1.0817148402156019e-06,
      "loss": 0.0012,
      "step": 49380
    },
    {
      "epoch": 0.8081485723635768,
      "grad_norm": 0.04341121348287595,
      "learning_rate": 1.0799414510547003e-06,
      "loss": 0.0018,
      "step": 49390
    },
    {
      "epoch": 0.8083121983146527,
      "grad_norm": 0.07123786874704853,
      "learning_rate": 1.0781693407435745e-06,
      "loss": 0.0021,
      "step": 49400
    },
    {
      "epoch": 0.8084758242657285,
      "grad_norm": 0.2624504874115645,
      "learning_rate": 1.0763985098603447e-06,
      "loss": 0.0025,
      "step": 49410
    },
    {
      "epoch": 0.8086394502168044,
      "grad_norm": 0.2501878316786769,
      "learning_rate": 1.074628958982713e-06,
      "loss": 0.0021,
      "step": 49420
    },
    {
      "epoch": 0.8088030761678803,
      "grad_norm": 0.14268568534069873,
      "learning_rate": 1.0728606886879634e-06,
      "loss": 0.0026,
      "step": 49430
    },
    {
      "epoch": 0.808966702118956,
      "grad_norm": 0.06297368975380918,
      "learning_rate": 1.0710936995529646e-06,
      "loss": 0.0007,
      "step": 49440
    },
    {
      "epoch": 0.8091303280700319,
      "grad_norm": 0.11489188480460669,
      "learning_rate": 1.0693279921541665e-06,
      "loss": 0.0022,
      "step": 49450
    },
    {
      "epoch": 0.8092939540211077,
      "grad_norm": 0.1829565295745005,
      "learning_rate": 1.0675635670675954e-06,
      "loss": 0.0018,
      "step": 49460
    },
    {
      "epoch": 0.8094575799721836,
      "grad_norm": 0.1268995703237754,
      "learning_rate": 1.0658004248688674e-06,
      "loss": 0.0014,
      "step": 49470
    },
    {
      "epoch": 0.8096212059232595,
      "grad_norm": 0.06750752138816665,
      "learning_rate": 1.0640385661331758e-06,
      "loss": 0.0015,
      "step": 49480
    },
    {
      "epoch": 0.8097848318743353,
      "grad_norm": 0.0590578051088296,
      "learning_rate": 1.0622779914352943e-06,
      "loss": 0.0014,
      "step": 49490
    },
    {
      "epoch": 0.8099484578254111,
      "grad_norm": 0.15650020542304577,
      "learning_rate": 1.0605187013495815e-06,
      "loss": 0.0029,
      "step": 49500
    },
    {
      "epoch": 0.8101120837764869,
      "grad_norm": 0.19558107522012827,
      "learning_rate": 1.0587606964499753e-06,
      "loss": 0.0024,
      "step": 49510
    },
    {
      "epoch": 0.8102757097275628,
      "grad_norm": 0.10335553112846346,
      "learning_rate": 1.0570039773099894e-06,
      "loss": 0.0016,
      "step": 49520
    },
    {
      "epoch": 0.8104393356786387,
      "grad_norm": 0.02449399370254586,
      "learning_rate": 1.0552485445027261e-06,
      "loss": 0.003,
      "step": 49530
    },
    {
      "epoch": 0.8106029616297145,
      "grad_norm": 0.1363349693400131,
      "learning_rate": 1.0534943986008633e-06,
      "loss": 0.0027,
      "step": 49540
    },
    {
      "epoch": 0.8107665875807903,
      "grad_norm": 0.10317328739452923,
      "learning_rate": 1.051741540176659e-06,
      "loss": 0.0018,
      "step": 49550
    },
    {
      "epoch": 0.8109302135318661,
      "grad_norm": 0.09270617389713492,
      "learning_rate": 1.049989969801956e-06,
      "loss": 0.0018,
      "step": 49560
    },
    {
      "epoch": 0.811093839482942,
      "grad_norm": 0.05002425077595298,
      "learning_rate": 1.0482396880481692e-06,
      "loss": 0.0018,
      "step": 49570
    },
    {
      "epoch": 0.8112574654340179,
      "grad_norm": 0.0860051333214948,
      "learning_rate": 1.046490695486297e-06,
      "loss": 0.0018,
      "step": 49580
    },
    {
      "epoch": 0.8114210913850937,
      "grad_norm": 0.21056914898429296,
      "learning_rate": 1.0447429926869196e-06,
      "loss": 0.0029,
      "step": 49590
    },
    {
      "epoch": 0.8115847173361695,
      "grad_norm": 0.05975768799229619,
      "learning_rate": 1.0429965802201935e-06,
      "loss": 0.001,
      "step": 49600
    },
    {
      "epoch": 0.8117483432872453,
      "grad_norm": 0.04347185584560614,
      "learning_rate": 1.0412514586558531e-06,
      "loss": 0.002,
      "step": 49610
    },
    {
      "epoch": 0.8119119692383212,
      "grad_norm": 0.04928992122041644,
      "learning_rate": 1.0395076285632177e-06,
      "loss": 0.0026,
      "step": 49620
    },
    {
      "epoch": 0.8120755951893971,
      "grad_norm": 0.03852364332709138,
      "learning_rate": 1.0377650905111764e-06,
      "loss": 0.0019,
      "step": 49630
    },
    {
      "epoch": 0.8122392211404729,
      "grad_norm": 0.02202195388414747,
      "learning_rate": 1.0360238450682026e-06,
      "loss": 0.0029,
      "step": 49640
    },
    {
      "epoch": 0.8124028470915488,
      "grad_norm": 0.07055682156013596,
      "learning_rate": 1.0342838928023486e-06,
      "loss": 0.0017,
      "step": 49650
    },
    {
      "epoch": 0.8125664730426245,
      "grad_norm": 0.041976855526547,
      "learning_rate": 1.0325452342812414e-06,
      "loss": 0.0028,
      "step": 49660
    },
    {
      "epoch": 0.8127300989937004,
      "grad_norm": 0.056436283705117885,
      "learning_rate": 1.0308078700720875e-06,
      "loss": 0.0015,
      "step": 49670
    },
    {
      "epoch": 0.8128937249447763,
      "grad_norm": 0.14382317879234516,
      "learning_rate": 1.029071800741674e-06,
      "loss": 0.0018,
      "step": 49680
    },
    {
      "epoch": 0.8130573508958521,
      "grad_norm": 0.11519219643693086,
      "learning_rate": 1.0273370268563598e-06,
      "loss": 0.0011,
      "step": 49690
    },
    {
      "epoch": 0.813220976846928,
      "grad_norm": 0.11587177410544204,
      "learning_rate": 1.0256035489820836e-06,
      "loss": 0.0017,
      "step": 49700
    },
    {
      "epoch": 0.8133846027980037,
      "grad_norm": 0.03465703032699276,
      "learning_rate": 1.0238713676843648e-06,
      "loss": 0.0019,
      "step": 49710
    },
    {
      "epoch": 0.8135482287490796,
      "grad_norm": 0.23581928375421,
      "learning_rate": 1.0221404835282956e-06,
      "loss": 0.0021,
      "step": 49720
    },
    {
      "epoch": 0.8137118547001554,
      "grad_norm": 0.13247542666398174,
      "learning_rate": 1.0204108970785458e-06,
      "loss": 0.0033,
      "step": 49730
    },
    {
      "epoch": 0.8138754806512313,
      "grad_norm": 0.0904144560927875,
      "learning_rate": 1.0186826088993634e-06,
      "loss": 0.0018,
      "step": 49740
    },
    {
      "epoch": 0.8140391066023072,
      "grad_norm": 0.11022979830028298,
      "learning_rate": 1.016955619554571e-06,
      "loss": 0.0014,
      "step": 49750
    },
    {
      "epoch": 0.8142027325533829,
      "grad_norm": 0.28995232148683214,
      "learning_rate": 1.0152299296075686e-06,
      "loss": 0.0031,
      "step": 49760
    },
    {
      "epoch": 0.8143663585044588,
      "grad_norm": 0.061515727976063445,
      "learning_rate": 1.0135055396213306e-06,
      "loss": 0.0012,
      "step": 49770
    },
    {
      "epoch": 0.8145299844555346,
      "grad_norm": 0.17838213642965908,
      "learning_rate": 1.0117824501584117e-06,
      "loss": 0.0027,
      "step": 49780
    },
    {
      "epoch": 0.8146936104066105,
      "grad_norm": 0.0865376384716027,
      "learning_rate": 1.0100606617809383e-06,
      "loss": 0.0024,
      "step": 49790
    },
    {
      "epoch": 0.8148572363576864,
      "grad_norm": 0.129675101660701,
      "learning_rate": 1.0083401750506122e-06,
      "loss": 0.0015,
      "step": 49800
    },
    {
      "epoch": 0.8150208623087621,
      "grad_norm": 0.04522226685261948,
      "learning_rate": 1.0066209905287132e-06,
      "loss": 0.0017,
      "step": 49810
    },
    {
      "epoch": 0.815184488259838,
      "grad_norm": 0.05308396513483676,
      "learning_rate": 1.004903108776094e-06,
      "loss": 0.004,
      "step": 49820
    },
    {
      "epoch": 0.8153481142109138,
      "grad_norm": 0.11263181848090847,
      "learning_rate": 1.0031865303531819e-06,
      "loss": 0.0023,
      "step": 49830
    },
    {
      "epoch": 0.8155117401619897,
      "grad_norm": 0.08759125851902075,
      "learning_rate": 1.001471255819983e-06,
      "loss": 0.0016,
      "step": 49840
    },
    {
      "epoch": 0.8156753661130656,
      "grad_norm": 0.1293976357907533,
      "learning_rate": 9.997572857360739e-07,
      "loss": 0.002,
      "step": 49850
    },
    {
      "epoch": 0.8158389920641413,
      "grad_norm": 0.17874225970660113,
      "learning_rate": 9.98044620660607e-07,
      "loss": 0.0048,
      "step": 49860
    },
    {
      "epoch": 0.8160026180152172,
      "grad_norm": 0.08490026081882147,
      "learning_rate": 9.963332611523096e-07,
      "loss": 0.0019,
      "step": 49870
    },
    {
      "epoch": 0.816166243966293,
      "grad_norm": 0.06825394309634786,
      "learning_rate": 9.946232077694818e-07,
      "loss": 0.0021,
      "step": 49880
    },
    {
      "epoch": 0.8163298699173689,
      "grad_norm": 0.08619646365586696,
      "learning_rate": 9.92914461069997e-07,
      "loss": 0.0025,
      "step": 49890
    },
    {
      "epoch": 0.8164934958684448,
      "grad_norm": 0.11573814133465485,
      "learning_rate": 9.912070216113063e-07,
      "loss": 0.0028,
      "step": 49900
    },
    {
      "epoch": 0.8166571218195205,
      "grad_norm": 0.08048732632430432,
      "learning_rate": 9.895008899504304e-07,
      "loss": 0.0032,
      "step": 49910
    },
    {
      "epoch": 0.8168207477705964,
      "grad_norm": 0.04886100114918426,
      "learning_rate": 9.877960666439646e-07,
      "loss": 0.0023,
      "step": 49920
    },
    {
      "epoch": 0.8169843737216722,
      "grad_norm": 0.06809259882783376,
      "learning_rate": 9.86092552248077e-07,
      "loss": 0.0015,
      "step": 49930
    },
    {
      "epoch": 0.8171479996727481,
      "grad_norm": 0.23727560381540982,
      "learning_rate": 9.843903473185097e-07,
      "loss": 0.0013,
      "step": 49940
    },
    {
      "epoch": 0.817311625623824,
      "grad_norm": 0.07050594527448661,
      "learning_rate": 9.82689452410575e-07,
      "loss": 0.0028,
      "step": 49950
    },
    {
      "epoch": 0.8174752515748998,
      "grad_norm": 0.11806638591375535,
      "learning_rate": 9.80989868079163e-07,
      "loss": 0.0019,
      "step": 49960
    },
    {
      "epoch": 0.8176388775259756,
      "grad_norm": 0.034993116396804566,
      "learning_rate": 9.792915948787313e-07,
      "loss": 0.0014,
      "step": 49970
    },
    {
      "epoch": 0.8178025034770514,
      "grad_norm": 0.2980401910651879,
      "learning_rate": 9.775946333633114e-07,
      "loss": 0.0012,
      "step": 49980
    },
    {
      "epoch": 0.8179661294281273,
      "grad_norm": 0.0583978725264821,
      "learning_rate": 9.758989840865074e-07,
      "loss": 0.0026,
      "step": 49990
    },
    {
      "epoch": 0.8181297553792032,
      "grad_norm": 0.05977471028561325,
      "learning_rate": 9.742046476014943e-07,
      "loss": 0.0017,
      "step": 50000
    },
    {
      "epoch": 0.8181297553792032,
      "eval_loss": 0.0015543417539447546,
      "eval_runtime": 7.5271,
      "eval_samples_per_second": 26.571,
      "eval_steps_per_second": 6.643,
      "step": 50000
    },
    {
      "epoch": 0.818293381330279,
      "grad_norm": 0.2413835624202922,
      "learning_rate": 9.725116244610189e-07,
      "loss": 0.0028,
      "step": 50010
    },
    {
      "epoch": 0.8184570072813548,
      "grad_norm": 0.1578326168667706,
      "learning_rate": 9.708199152174025e-07,
      "loss": 0.0024,
      "step": 50020
    },
    {
      "epoch": 0.8186206332324306,
      "grad_norm": 0.18755369335015012,
      "learning_rate": 9.691295204225332e-07,
      "loss": 0.0027,
      "step": 50030
    },
    {
      "epoch": 0.8187842591835065,
      "grad_norm": 0.09001391610301843,
      "learning_rate": 9.674404406278731e-07,
      "loss": 0.0023,
      "step": 50040
    },
    {
      "epoch": 0.8189478851345824,
      "grad_norm": 0.176816353102448,
      "learning_rate": 9.65752676384455e-07,
      "loss": 0.0019,
      "step": 50050
    },
    {
      "epoch": 0.8191115110856582,
      "grad_norm": 0.016845614263576194,
      "learning_rate": 9.640662282428815e-07,
      "loss": 0.002,
      "step": 50060
    },
    {
      "epoch": 0.819275137036734,
      "grad_norm": 0.07352834614778095,
      "learning_rate": 9.623810967533258e-07,
      "loss": 0.0018,
      "step": 50070
    },
    {
      "epoch": 0.8194387629878098,
      "grad_norm": 0.05401263593021629,
      "learning_rate": 9.606972824655325e-07,
      "loss": 0.0018,
      "step": 50080
    },
    {
      "epoch": 0.8196023889388857,
      "grad_norm": 0.17464752112973664,
      "learning_rate": 9.59014785928818e-07,
      "loss": 0.0028,
      "step": 50090
    },
    {
      "epoch": 0.8197660148899616,
      "grad_norm": 0.2209633555436409,
      "learning_rate": 9.57333607692066e-07,
      "loss": 0.0026,
      "step": 50100
    },
    {
      "epoch": 0.8199296408410374,
      "grad_norm": 0.1290585879361367,
      "learning_rate": 9.55653748303731e-07,
      "loss": 0.0024,
      "step": 50110
    },
    {
      "epoch": 0.8200932667921133,
      "grad_norm": 0.07405520690057404,
      "learning_rate": 9.53975208311837e-07,
      "loss": 0.0014,
      "step": 50120
    },
    {
      "epoch": 0.820256892743189,
      "grad_norm": 0.18147721133148767,
      "learning_rate": 9.522979882639787e-07,
      "loss": 0.0032,
      "step": 50130
    },
    {
      "epoch": 0.8204205186942649,
      "grad_norm": 0.1819848452900205,
      "learning_rate": 9.506220887073181e-07,
      "loss": 0.0026,
      "step": 50140
    },
    {
      "epoch": 0.8205841446453408,
      "grad_norm": 0.10807158000811144,
      "learning_rate": 9.489475101885898e-07,
      "loss": 0.0016,
      "step": 50150
    },
    {
      "epoch": 0.8207477705964166,
      "grad_norm": 0.14902551968406735,
      "learning_rate": 9.472742532540946e-07,
      "loss": 0.0018,
      "step": 50160
    },
    {
      "epoch": 0.8209113965474925,
      "grad_norm": 0.034174935035031044,
      "learning_rate": 9.456023184497026e-07,
      "loss": 0.0019,
      "step": 50170
    },
    {
      "epoch": 0.8210750224985682,
      "grad_norm": 0.10284340835244944,
      "learning_rate": 9.439317063208536e-07,
      "loss": 0.0023,
      "step": 50180
    },
    {
      "epoch": 0.8212386484496441,
      "grad_norm": 0.15983382150203546,
      "learning_rate": 9.42262417412555e-07,
      "loss": 0.0018,
      "step": 50190
    },
    {
      "epoch": 0.82140227440072,
      "grad_norm": 0.09381858035963471,
      "learning_rate": 9.405944522693816e-07,
      "loss": 0.003,
      "step": 50200
    },
    {
      "epoch": 0.8215659003517958,
      "grad_norm": 0.21904266266915087,
      "learning_rate": 9.389278114354794e-07,
      "loss": 0.0017,
      "step": 50210
    },
    {
      "epoch": 0.8217295263028717,
      "grad_norm": 0.14840739811293743,
      "learning_rate": 9.3726249545456e-07,
      "loss": 0.0019,
      "step": 50220
    },
    {
      "epoch": 0.8218931522539474,
      "grad_norm": 0.12881060104468983,
      "learning_rate": 9.355985048699029e-07,
      "loss": 0.0019,
      "step": 50230
    },
    {
      "epoch": 0.8220567782050233,
      "grad_norm": 0.13188153686497928,
      "learning_rate": 9.339358402243559e-07,
      "loss": 0.0019,
      "step": 50240
    },
    {
      "epoch": 0.8222204041560992,
      "grad_norm": 0.09848077177804467,
      "learning_rate": 9.32274502060334e-07,
      "loss": 0.0029,
      "step": 50250
    },
    {
      "epoch": 0.822384030107175,
      "grad_norm": 0.13853652774023031,
      "learning_rate": 9.306144909198173e-07,
      "loss": 0.0014,
      "step": 50260
    },
    {
      "epoch": 0.8225476560582509,
      "grad_norm": 0.06852128138828285,
      "learning_rate": 9.289558073443588e-07,
      "loss": 0.0019,
      "step": 50270
    },
    {
      "epoch": 0.8227112820093266,
      "grad_norm": 0.08092590263386222,
      "learning_rate": 9.272984518750722e-07,
      "loss": 0.0017,
      "step": 50280
    },
    {
      "epoch": 0.8228749079604025,
      "grad_norm": 0.03663578751872039,
      "learning_rate": 9.256424250526413e-07,
      "loss": 0.0014,
      "step": 50290
    },
    {
      "epoch": 0.8230385339114784,
      "grad_norm": 0.11615936047811047,
      "learning_rate": 9.239877274173154e-07,
      "loss": 0.0021,
      "step": 50300
    },
    {
      "epoch": 0.8232021598625542,
      "grad_norm": 0.05123251220931151,
      "learning_rate": 9.223343595089102e-07,
      "loss": 0.0017,
      "step": 50310
    },
    {
      "epoch": 0.8233657858136301,
      "grad_norm": 0.15836925225373494,
      "learning_rate": 9.206823218668065e-07,
      "loss": 0.0031,
      "step": 50320
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.14603176717438798,
      "learning_rate": 9.190316150299549e-07,
      "loss": 0.0021,
      "step": 50330
    },
    {
      "epoch": 0.8236930377157817,
      "grad_norm": 0.2801561075796078,
      "learning_rate": 9.173822395368682e-07,
      "loss": 0.0017,
      "step": 50340
    },
    {
      "epoch": 0.8238566636668576,
      "grad_norm": 0.1514760749592684,
      "learning_rate": 9.157341959256255e-07,
      "loss": 0.0047,
      "step": 50350
    },
    {
      "epoch": 0.8240202896179334,
      "grad_norm": 0.28785075763690343,
      "learning_rate": 9.140874847338726e-07,
      "loss": 0.002,
      "step": 50360
    },
    {
      "epoch": 0.8241839155690093,
      "grad_norm": 0.09648953641175018,
      "learning_rate": 9.124421064988193e-07,
      "loss": 0.0012,
      "step": 50370
    },
    {
      "epoch": 0.824347541520085,
      "grad_norm": 0.0986778610424153,
      "learning_rate": 9.107980617572398e-07,
      "loss": 0.0008,
      "step": 50380
    },
    {
      "epoch": 0.8245111674711609,
      "grad_norm": 0.052513689326658876,
      "learning_rate": 9.091553510454781e-07,
      "loss": 0.0015,
      "step": 50390
    },
    {
      "epoch": 0.8246747934222368,
      "grad_norm": 0.1567370836669226,
      "learning_rate": 9.075139748994371e-07,
      "loss": 0.0019,
      "step": 50400
    },
    {
      "epoch": 0.8248384193733126,
      "grad_norm": 0.2179903254338755,
      "learning_rate": 9.058739338545886e-07,
      "loss": 0.0019,
      "step": 50410
    },
    {
      "epoch": 0.8250020453243885,
      "grad_norm": 0.06885349749541883,
      "learning_rate": 9.042352284459632e-07,
      "loss": 0.002,
      "step": 50420
    },
    {
      "epoch": 0.8251656712754643,
      "grad_norm": 0.026752320460924166,
      "learning_rate": 9.025978592081635e-07,
      "loss": 0.0019,
      "step": 50430
    },
    {
      "epoch": 0.8253292972265401,
      "grad_norm": 0.22818447017191723,
      "learning_rate": 9.009618266753506e-07,
      "loss": 0.0022,
      "step": 50440
    },
    {
      "epoch": 0.825492923177616,
      "grad_norm": 0.06606107578756397,
      "learning_rate": 8.993271313812501e-07,
      "loss": 0.0012,
      "step": 50450
    },
    {
      "epoch": 0.8256565491286918,
      "grad_norm": 0.07415603453216045,
      "learning_rate": 8.976937738591551e-07,
      "loss": 0.0015,
      "step": 50460
    },
    {
      "epoch": 0.8258201750797677,
      "grad_norm": 0.1409778977604642,
      "learning_rate": 8.960617546419192e-07,
      "loss": 0.0032,
      "step": 50470
    },
    {
      "epoch": 0.8259838010308435,
      "grad_norm": 0.16307468584882612,
      "learning_rate": 8.944310742619561e-07,
      "loss": 0.0019,
      "step": 50480
    },
    {
      "epoch": 0.8261474269819193,
      "grad_norm": 0.35255933462664824,
      "learning_rate": 8.928017332512506e-07,
      "loss": 0.0022,
      "step": 50490
    },
    {
      "epoch": 0.8263110529329952,
      "grad_norm": 0.04905779175011822,
      "learning_rate": 8.91173732141345e-07,
      "loss": 0.003,
      "step": 50500
    },
    {
      "epoch": 0.826474678884071,
      "grad_norm": 0.21010202412904783,
      "learning_rate": 8.895470714633442e-07,
      "loss": 0.0025,
      "step": 50510
    },
    {
      "epoch": 0.8266383048351469,
      "grad_norm": 0.04929280082834834,
      "learning_rate": 8.87921751747921e-07,
      "loss": 0.002,
      "step": 50520
    },
    {
      "epoch": 0.8268019307862227,
      "grad_norm": 0.16293924627340495,
      "learning_rate": 8.862977735253059e-07,
      "loss": 0.002,
      "step": 50530
    },
    {
      "epoch": 0.8269655567372985,
      "grad_norm": 0.12614630476395672,
      "learning_rate": 8.846751373252899e-07,
      "loss": 0.002,
      "step": 50540
    },
    {
      "epoch": 0.8271291826883744,
      "grad_norm": 0.056843479618960586,
      "learning_rate": 8.830538436772335e-07,
      "loss": 0.0017,
      "step": 50550
    },
    {
      "epoch": 0.8272928086394502,
      "grad_norm": 0.10827381267991691,
      "learning_rate": 8.814338931100536e-07,
      "loss": 0.0022,
      "step": 50560
    },
    {
      "epoch": 0.8274564345905261,
      "grad_norm": 0.04587476460158164,
      "learning_rate": 8.798152861522291e-07,
      "loss": 0.0014,
      "step": 50570
    },
    {
      "epoch": 0.8276200605416019,
      "grad_norm": 0.09398186214182185,
      "learning_rate": 8.781980233318038e-07,
      "loss": 0.0013,
      "step": 50580
    },
    {
      "epoch": 0.8277836864926778,
      "grad_norm": 0.08724773974785589,
      "learning_rate": 8.765821051763823e-07,
      "loss": 0.0017,
      "step": 50590
    },
    {
      "epoch": 0.8279473124437535,
      "grad_norm": 0.05999215074593843,
      "learning_rate": 8.749675322131246e-07,
      "loss": 0.0014,
      "step": 50600
    },
    {
      "epoch": 0.8281109383948294,
      "grad_norm": 0.09769846021230107,
      "learning_rate": 8.733543049687604e-07,
      "loss": 0.0014,
      "step": 50610
    },
    {
      "epoch": 0.8282745643459053,
      "grad_norm": 0.13252293646277716,
      "learning_rate": 8.71742423969576e-07,
      "loss": 0.0022,
      "step": 50620
    },
    {
      "epoch": 0.8284381902969811,
      "grad_norm": 0.03554623628157254,
      "learning_rate": 8.701318897414174e-07,
      "loss": 0.0017,
      "step": 50630
    },
    {
      "epoch": 0.828601816248057,
      "grad_norm": 0.1463673548574833,
      "learning_rate": 8.685227028096954e-07,
      "loss": 0.0022,
      "step": 50640
    },
    {
      "epoch": 0.8287654421991327,
      "grad_norm": 0.03943670502028477,
      "learning_rate": 8.669148636993785e-07,
      "loss": 0.0019,
      "step": 50650
    },
    {
      "epoch": 0.8289290681502086,
      "grad_norm": 0.2869232369504291,
      "learning_rate": 8.653083729349926e-07,
      "loss": 0.0027,
      "step": 50660
    },
    {
      "epoch": 0.8290926941012845,
      "grad_norm": 0.04908878257764215,
      "learning_rate": 8.637032310406301e-07,
      "loss": 0.0017,
      "step": 50670
    },
    {
      "epoch": 0.8292563200523603,
      "grad_norm": 0.11119104493211474,
      "learning_rate": 8.62099438539939e-07,
      "loss": 0.0019,
      "step": 50680
    },
    {
      "epoch": 0.8294199460034362,
      "grad_norm": 0.31403290954339175,
      "learning_rate": 8.604969959561277e-07,
      "loss": 0.0016,
      "step": 50690
    },
    {
      "epoch": 0.8295835719545119,
      "grad_norm": 0.18905018602389537,
      "learning_rate": 8.588959038119665e-07,
      "loss": 0.0027,
      "step": 50700
    },
    {
      "epoch": 0.8297471979055878,
      "grad_norm": 0.06544412546012782,
      "learning_rate": 8.572961626297832e-07,
      "loss": 0.0018,
      "step": 50710
    },
    {
      "epoch": 0.8299108238566637,
      "grad_norm": 0.1552052366482621,
      "learning_rate": 8.556977729314636e-07,
      "loss": 0.0027,
      "step": 50720
    },
    {
      "epoch": 0.8300744498077395,
      "grad_norm": 0.034206068773142924,
      "learning_rate": 8.541007352384528e-07,
      "loss": 0.0009,
      "step": 50730
    },
    {
      "epoch": 0.8302380757588154,
      "grad_norm": 0.14659575755679302,
      "learning_rate": 8.525050500717597e-07,
      "loss": 0.0018,
      "step": 50740
    },
    {
      "epoch": 0.8304017017098911,
      "grad_norm": 0.40649534229681034,
      "learning_rate": 8.509107179519465e-07,
      "loss": 0.002,
      "step": 50750
    },
    {
      "epoch": 0.830565327660967,
      "grad_norm": 0.060005238814698304,
      "learning_rate": 8.493177393991353e-07,
      "loss": 0.0023,
      "step": 50760
    },
    {
      "epoch": 0.8307289536120429,
      "grad_norm": 0.3410742717349367,
      "learning_rate": 8.477261149330096e-07,
      "loss": 0.0019,
      "step": 50770
    },
    {
      "epoch": 0.8308925795631187,
      "grad_norm": 0.2636283248786855,
      "learning_rate": 8.461358450728063e-07,
      "loss": 0.0019,
      "step": 50780
    },
    {
      "epoch": 0.8310562055141946,
      "grad_norm": 0.16422577700578947,
      "learning_rate": 8.445469303373222e-07,
      "loss": 0.0027,
      "step": 50790
    },
    {
      "epoch": 0.8312198314652703,
      "grad_norm": 0.2892102437909206,
      "learning_rate": 8.429593712449153e-07,
      "loss": 0.0027,
      "step": 50800
    },
    {
      "epoch": 0.8313834574163462,
      "grad_norm": 0.021395587389908512,
      "learning_rate": 8.413731683134968e-07,
      "loss": 0.004,
      "step": 50810
    },
    {
      "epoch": 0.8315470833674221,
      "grad_norm": 0.14529101152450363,
      "learning_rate": 8.397883220605385e-07,
      "loss": 0.0034,
      "step": 50820
    },
    {
      "epoch": 0.8317107093184979,
      "grad_norm": 0.07154327415000333,
      "learning_rate": 8.382048330030679e-07,
      "loss": 0.0011,
      "step": 50830
    },
    {
      "epoch": 0.8318743352695738,
      "grad_norm": 0.12196579815944476,
      "learning_rate": 8.3662270165767e-07,
      "loss": 0.0028,
      "step": 50840
    },
    {
      "epoch": 0.8320379612206495,
      "grad_norm": 0.10066322728156621,
      "learning_rate": 8.350419285404865e-07,
      "loss": 0.0019,
      "step": 50850
    },
    {
      "epoch": 0.8322015871717254,
      "grad_norm": 0.22409745004352574,
      "learning_rate": 8.334625141672187e-07,
      "loss": 0.0017,
      "step": 50860
    },
    {
      "epoch": 0.8323652131228013,
      "grad_norm": 0.11860832310852683,
      "learning_rate": 8.318844590531217e-07,
      "loss": 0.0034,
      "step": 50870
    },
    {
      "epoch": 0.8325288390738771,
      "grad_norm": 0.1651348359192965,
      "learning_rate": 8.30307763713008e-07,
      "loss": 0.0023,
      "step": 50880
    },
    {
      "epoch": 0.832692465024953,
      "grad_norm": 0.4416311078822497,
      "learning_rate": 8.287324286612463e-07,
      "loss": 0.0019,
      "step": 50890
    },
    {
      "epoch": 0.8328560909760288,
      "grad_norm": 0.10979841570852097,
      "learning_rate": 8.271584544117617e-07,
      "loss": 0.0021,
      "step": 50900
    },
    {
      "epoch": 0.8330197169271046,
      "grad_norm": 0.06759634631166857,
      "learning_rate": 8.255858414780349e-07,
      "loss": 0.0029,
      "step": 50910
    },
    {
      "epoch": 0.8331833428781805,
      "grad_norm": 0.18493834740235032,
      "learning_rate": 8.24014590373105e-07,
      "loss": 0.0014,
      "step": 50920
    },
    {
      "epoch": 0.8333469688292563,
      "grad_norm": 0.08861719209062553,
      "learning_rate": 8.224447016095638e-07,
      "loss": 0.002,
      "step": 50930
    },
    {
      "epoch": 0.8335105947803322,
      "grad_norm": 0.04343136704379043,
      "learning_rate": 8.208761756995598e-07,
      "loss": 0.0025,
      "step": 50940
    },
    {
      "epoch": 0.833674220731408,
      "grad_norm": 0.29287910843299075,
      "learning_rate": 8.193090131547971e-07,
      "loss": 0.0029,
      "step": 50950
    },
    {
      "epoch": 0.8338378466824838,
      "grad_norm": 0.15949785879001482,
      "learning_rate": 8.177432144865338e-07,
      "loss": 0.0024,
      "step": 50960
    },
    {
      "epoch": 0.8340014726335597,
      "grad_norm": 0.03290814485614608,
      "learning_rate": 8.161787802055837e-07,
      "loss": 0.0027,
      "step": 50970
    },
    {
      "epoch": 0.8341650985846355,
      "grad_norm": 0.20128565805810664,
      "learning_rate": 8.146157108223179e-07,
      "loss": 0.0013,
      "step": 50980
    },
    {
      "epoch": 0.8343287245357114,
      "grad_norm": 0.2424816390645313,
      "learning_rate": 8.130540068466597e-07,
      "loss": 0.0034,
      "step": 50990
    },
    {
      "epoch": 0.8344923504867872,
      "grad_norm": 0.026862225545815525,
      "learning_rate": 8.114936687880859e-07,
      "loss": 0.0025,
      "step": 51000
    },
    {
      "epoch": 0.834655976437863,
      "grad_norm": 0.13147698999489818,
      "learning_rate": 8.099346971556305e-07,
      "loss": 0.0023,
      "step": 51010
    },
    {
      "epoch": 0.8348196023889389,
      "grad_norm": 0.27115844335841505,
      "learning_rate": 8.083770924578798e-07,
      "loss": 0.0041,
      "step": 51020
    },
    {
      "epoch": 0.8349832283400147,
      "grad_norm": 0.13099361045857733,
      "learning_rate": 8.068208552029739e-07,
      "loss": 0.0029,
      "step": 51030
    },
    {
      "epoch": 0.8351468542910906,
      "grad_norm": 0.09446867495792653,
      "learning_rate": 8.052659858986101e-07,
      "loss": 0.0011,
      "step": 51040
    },
    {
      "epoch": 0.8353104802421664,
      "grad_norm": 0.08345202968882982,
      "learning_rate": 8.03712485052035e-07,
      "loss": 0.0015,
      "step": 51050
    },
    {
      "epoch": 0.8354741061932422,
      "grad_norm": 0.16706391686782135,
      "learning_rate": 8.021603531700517e-07,
      "loss": 0.002,
      "step": 51060
    },
    {
      "epoch": 0.8356377321443181,
      "grad_norm": 0.4072252204280583,
      "learning_rate": 8.006095907590156e-07,
      "loss": 0.0024,
      "step": 51070
    },
    {
      "epoch": 0.8358013580953939,
      "grad_norm": 0.08938189021570003,
      "learning_rate": 7.990601983248353e-07,
      "loss": 0.0018,
      "step": 51080
    },
    {
      "epoch": 0.8359649840464698,
      "grad_norm": 0.07525252949969802,
      "learning_rate": 7.975121763729732e-07,
      "loss": 0.0021,
      "step": 51090
    },
    {
      "epoch": 0.8361286099975456,
      "grad_norm": 0.04587479196833858,
      "learning_rate": 7.95965525408442e-07,
      "loss": 0.0012,
      "step": 51100
    },
    {
      "epoch": 0.8362922359486215,
      "grad_norm": 0.24156113948710264,
      "learning_rate": 7.944202459358125e-07,
      "loss": 0.0031,
      "step": 51110
    },
    {
      "epoch": 0.8364558618996973,
      "grad_norm": 0.17386140419697935,
      "learning_rate": 7.928763384592025e-07,
      "loss": 0.002,
      "step": 51120
    },
    {
      "epoch": 0.8366194878507731,
      "grad_norm": 0.19379535484109214,
      "learning_rate": 7.913338034822859e-07,
      "loss": 0.0031,
      "step": 51130
    },
    {
      "epoch": 0.836783113801849,
      "grad_norm": 0.24590074715695087,
      "learning_rate": 7.897926415082863e-07,
      "loss": 0.0028,
      "step": 51140
    },
    {
      "epoch": 0.8369467397529248,
      "grad_norm": 0.09136406481245382,
      "learning_rate": 7.882528530399813e-07,
      "loss": 0.0026,
      "step": 51150
    },
    {
      "epoch": 0.8371103657040007,
      "grad_norm": 0.0816631799044623,
      "learning_rate": 7.867144385796982e-07,
      "loss": 0.001,
      "step": 51160
    },
    {
      "epoch": 0.8372739916550765,
      "grad_norm": 0.1345884689992293,
      "learning_rate": 7.851773986293198e-07,
      "loss": 0.002,
      "step": 51170
    },
    {
      "epoch": 0.8374376176061523,
      "grad_norm": 0.15480442618999135,
      "learning_rate": 7.836417336902769e-07,
      "loss": 0.003,
      "step": 51180
    },
    {
      "epoch": 0.8376012435572282,
      "grad_norm": 0.052260353636827746,
      "learning_rate": 7.821074442635535e-07,
      "loss": 0.003,
      "step": 51190
    },
    {
      "epoch": 0.837764869508304,
      "grad_norm": 0.03771533629645811,
      "learning_rate": 7.805745308496837e-07,
      "loss": 0.0019,
      "step": 51200
    },
    {
      "epoch": 0.8379284954593799,
      "grad_norm": 0.3151183532356463,
      "learning_rate": 7.790429939487543e-07,
      "loss": 0.0023,
      "step": 51210
    },
    {
      "epoch": 0.8380921214104557,
      "grad_norm": 0.16476186967728582,
      "learning_rate": 7.775128340604005e-07,
      "loss": 0.0014,
      "step": 51220
    },
    {
      "epoch": 0.8382557473615315,
      "grad_norm": 0.04805967136827938,
      "learning_rate": 7.759840516838113e-07,
      "loss": 0.002,
      "step": 51230
    },
    {
      "epoch": 0.8384193733126074,
      "grad_norm": 0.14837128760492765,
      "learning_rate": 7.744566473177256e-07,
      "loss": 0.0032,
      "step": 51240
    },
    {
      "epoch": 0.8385829992636832,
      "grad_norm": 0.08214173793318282,
      "learning_rate": 7.729306214604304e-07,
      "loss": 0.0019,
      "step": 51250
    },
    {
      "epoch": 0.8387466252147591,
      "grad_norm": 0.18382026524054587,
      "learning_rate": 7.714059746097658e-07,
      "loss": 0.0022,
      "step": 51260
    },
    {
      "epoch": 0.838910251165835,
      "grad_norm": 0.037459655610016435,
      "learning_rate": 7.698827072631204e-07,
      "loss": 0.0026,
      "step": 51270
    },
    {
      "epoch": 0.8390738771169107,
      "grad_norm": 0.1136281649777543,
      "learning_rate": 7.68360819917432e-07,
      "loss": 0.002,
      "step": 51280
    },
    {
      "epoch": 0.8392375030679866,
      "grad_norm": 0.058479074561873776,
      "learning_rate": 7.66840313069192e-07,
      "loss": 0.002,
      "step": 51290
    },
    {
      "epoch": 0.8394011290190624,
      "grad_norm": 0.2771937587096952,
      "learning_rate": 7.653211872144373e-07,
      "loss": 0.0018,
      "step": 51300
    },
    {
      "epoch": 0.8395647549701383,
      "grad_norm": 0.2433673238993186,
      "learning_rate": 7.638034428487562e-07,
      "loss": 0.0021,
      "step": 51310
    },
    {
      "epoch": 0.8397283809212142,
      "grad_norm": 0.09490281069297671,
      "learning_rate": 7.622870804672856e-07,
      "loss": 0.0015,
      "step": 51320
    },
    {
      "epoch": 0.8398920068722899,
      "grad_norm": 0.1417323080183743,
      "learning_rate": 7.607721005647123e-07,
      "loss": 0.0013,
      "step": 51330
    },
    {
      "epoch": 0.8400556328233658,
      "grad_norm": 0.15363848675551178,
      "learning_rate": 7.592585036352707e-07,
      "loss": 0.0021,
      "step": 51340
    },
    {
      "epoch": 0.8402192587744416,
      "grad_norm": 0.14441305539114285,
      "learning_rate": 7.577462901727467e-07,
      "loss": 0.0029,
      "step": 51350
    },
    {
      "epoch": 0.8403828847255175,
      "grad_norm": 0.1527856647553139,
      "learning_rate": 7.562354606704725e-07,
      "loss": 0.0017,
      "step": 51360
    },
    {
      "epoch": 0.8405465106765934,
      "grad_norm": 0.11497204405397646,
      "learning_rate": 7.547260156213299e-07,
      "loss": 0.0015,
      "step": 51370
    },
    {
      "epoch": 0.8407101366276691,
      "grad_norm": 0.17506635507093313,
      "learning_rate": 7.532179555177477e-07,
      "loss": 0.0026,
      "step": 51380
    },
    {
      "epoch": 0.840873762578745,
      "grad_norm": 0.015149470299458822,
      "learning_rate": 7.517112808517041e-07,
      "loss": 0.0013,
      "step": 51390
    },
    {
      "epoch": 0.8410373885298208,
      "grad_norm": 0.04639418222344294,
      "learning_rate": 7.502059921147254e-07,
      "loss": 0.002,
      "step": 51400
    },
    {
      "epoch": 0.8412010144808967,
      "grad_norm": 0.07515616992256932,
      "learning_rate": 7.487020897978842e-07,
      "loss": 0.0016,
      "step": 51410
    },
    {
      "epoch": 0.8413646404319726,
      "grad_norm": 0.07860643601847318,
      "learning_rate": 7.471995743918048e-07,
      "loss": 0.0016,
      "step": 51420
    },
    {
      "epoch": 0.8415282663830483,
      "grad_norm": 0.20559525098616413,
      "learning_rate": 7.45698446386654e-07,
      "loss": 0.0027,
      "step": 51430
    },
    {
      "epoch": 0.8416918923341242,
      "grad_norm": 0.07322614402814903,
      "learning_rate": 7.44198706272149e-07,
      "loss": 0.0016,
      "step": 51440
    },
    {
      "epoch": 0.8418555182852,
      "grad_norm": 0.09639931046323989,
      "learning_rate": 7.427003545375533e-07,
      "loss": 0.0014,
      "step": 51450
    },
    {
      "epoch": 0.8420191442362759,
      "grad_norm": 0.07688377279583636,
      "learning_rate": 7.412033916716777e-07,
      "loss": 0.0017,
      "step": 51460
    },
    {
      "epoch": 0.8421827701873517,
      "grad_norm": 0.1581010876398973,
      "learning_rate": 7.397078181628786e-07,
      "loss": 0.0026,
      "step": 51470
    },
    {
      "epoch": 0.8423463961384275,
      "grad_norm": 0.05258380599910729,
      "learning_rate": 7.382136344990631e-07,
      "loss": 0.0016,
      "step": 51480
    },
    {
      "epoch": 0.8425100220895034,
      "grad_norm": 0.25292190662449676,
      "learning_rate": 7.367208411676807e-07,
      "loss": 0.0016,
      "step": 51490
    },
    {
      "epoch": 0.8426736480405792,
      "grad_norm": 0.09837018809512779,
      "learning_rate": 7.35229438655729e-07,
      "loss": 0.0016,
      "step": 51500
    },
    {
      "epoch": 0.8428372739916551,
      "grad_norm": 0.0413775220456556,
      "learning_rate": 7.337394274497517e-07,
      "loss": 0.0023,
      "step": 51510
    },
    {
      "epoch": 0.8430008999427309,
      "grad_norm": 0.17312222856018422,
      "learning_rate": 7.322508080358381e-07,
      "loss": 0.0017,
      "step": 51520
    },
    {
      "epoch": 0.8431645258938067,
      "grad_norm": 0.14487331937591338,
      "learning_rate": 7.307635808996238e-07,
      "loss": 0.0014,
      "step": 51530
    },
    {
      "epoch": 0.8433281518448826,
      "grad_norm": 0.17394547510314537,
      "learning_rate": 7.292777465262918e-07,
      "loss": 0.0037,
      "step": 51540
    },
    {
      "epoch": 0.8434917777959584,
      "grad_norm": 0.1497988030582846,
      "learning_rate": 7.277933054005693e-07,
      "loss": 0.0019,
      "step": 51550
    },
    {
      "epoch": 0.8436554037470343,
      "grad_norm": 0.22204597193606948,
      "learning_rate": 7.263102580067261e-07,
      "loss": 0.002,
      "step": 51560
    },
    {
      "epoch": 0.8438190296981101,
      "grad_norm": 0.035145299111214376,
      "learning_rate": 7.24828604828583e-07,
      "loss": 0.0018,
      "step": 51570
    },
    {
      "epoch": 0.843982655649186,
      "grad_norm": 0.0839146223194527,
      "learning_rate": 7.233483463495023e-07,
      "loss": 0.0018,
      "step": 51580
    },
    {
      "epoch": 0.8441462816002618,
      "grad_norm": 0.1831161009926824,
      "learning_rate": 7.218694830523915e-07,
      "loss": 0.0019,
      "step": 51590
    },
    {
      "epoch": 0.8443099075513376,
      "grad_norm": 0.3867268494242546,
      "learning_rate": 7.203920154197047e-07,
      "loss": 0.003,
      "step": 51600
    },
    {
      "epoch": 0.8444735335024135,
      "grad_norm": 0.0550704573591412,
      "learning_rate": 7.189159439334409e-07,
      "loss": 0.0011,
      "step": 51610
    },
    {
      "epoch": 0.8446371594534893,
      "grad_norm": 0.14741913756905575,
      "learning_rate": 7.174412690751381e-07,
      "loss": 0.003,
      "step": 51620
    },
    {
      "epoch": 0.8448007854045652,
      "grad_norm": 0.14234583793212882,
      "learning_rate": 7.15967991325886e-07,
      "loss": 0.0016,
      "step": 51630
    },
    {
      "epoch": 0.844964411355641,
      "grad_norm": 0.03509612654027745,
      "learning_rate": 7.144961111663151e-07,
      "loss": 0.0022,
      "step": 51640
    },
    {
      "epoch": 0.8451280373067168,
      "grad_norm": 0.04156446712431083,
      "learning_rate": 7.130256290765991e-07,
      "loss": 0.0035,
      "step": 51650
    },
    {
      "epoch": 0.8452916632577927,
      "grad_norm": 0.09579251044275162,
      "learning_rate": 7.115565455364587e-07,
      "loss": 0.0025,
      "step": 51660
    },
    {
      "epoch": 0.8454552892088685,
      "grad_norm": 0.16665627384187398,
      "learning_rate": 7.100888610251561e-07,
      "loss": 0.0047,
      "step": 51670
    },
    {
      "epoch": 0.8456189151599444,
      "grad_norm": 0.12202959560555679,
      "learning_rate": 7.086225760214943e-07,
      "loss": 0.0022,
      "step": 51680
    },
    {
      "epoch": 0.8457825411110202,
      "grad_norm": 0.06183587377263898,
      "learning_rate": 7.071576910038269e-07,
      "loss": 0.0025,
      "step": 51690
    },
    {
      "epoch": 0.845946167062096,
      "grad_norm": 0.02878578287509975,
      "learning_rate": 7.056942064500449e-07,
      "loss": 0.002,
      "step": 51700
    },
    {
      "epoch": 0.8461097930131719,
      "grad_norm": 0.055006659917020535,
      "learning_rate": 7.042321228375842e-07,
      "loss": 0.0016,
      "step": 51710
    },
    {
      "epoch": 0.8462734189642477,
      "grad_norm": 0.1452951845271317,
      "learning_rate": 7.027714406434228e-07,
      "loss": 0.0016,
      "step": 51720
    },
    {
      "epoch": 0.8464370449153236,
      "grad_norm": 0.1035547797587581,
      "learning_rate": 7.013121603440853e-07,
      "loss": 0.0014,
      "step": 51730
    },
    {
      "epoch": 0.8466006708663995,
      "grad_norm": 0.35937694191909625,
      "learning_rate": 6.99854282415634e-07,
      "loss": 0.0033,
      "step": 51740
    },
    {
      "epoch": 0.8467642968174752,
      "grad_norm": 0.08032949091035617,
      "learning_rate": 6.983978073336745e-07,
      "loss": 0.0015,
      "step": 51750
    },
    {
      "epoch": 0.8469279227685511,
      "grad_norm": 0.13558943540384816,
      "learning_rate": 6.969427355733593e-07,
      "loss": 0.003,
      "step": 51760
    },
    {
      "epoch": 0.8470915487196269,
      "grad_norm": 0.18510281005243345,
      "learning_rate": 6.954890676093778e-07,
      "loss": 0.0021,
      "step": 51770
    },
    {
      "epoch": 0.8472551746707028,
      "grad_norm": 0.05198851668884105,
      "learning_rate": 6.940368039159628e-07,
      "loss": 0.0033,
      "step": 51780
    },
    {
      "epoch": 0.8474188006217787,
      "grad_norm": 0.16621986594732618,
      "learning_rate": 6.925859449668932e-07,
      "loss": 0.0016,
      "step": 51790
    },
    {
      "epoch": 0.8475824265728544,
      "grad_norm": 0.3858628864880291,
      "learning_rate": 6.911364912354818e-07,
      "loss": 0.0028,
      "step": 51800
    },
    {
      "epoch": 0.8477460525239303,
      "grad_norm": 0.10725221025586196,
      "learning_rate": 6.896884431945883e-07,
      "loss": 0.0028,
      "step": 51810
    },
    {
      "epoch": 0.8479096784750061,
      "grad_norm": 0.18871399736461378,
      "learning_rate": 6.882418013166148e-07,
      "loss": 0.0027,
      "step": 51820
    },
    {
      "epoch": 0.848073304426082,
      "grad_norm": 0.09924791660572244,
      "learning_rate": 6.867965660735009e-07,
      "loss": 0.0039,
      "step": 51830
    },
    {
      "epoch": 0.8482369303771579,
      "grad_norm": 0.23416178308576668,
      "learning_rate": 6.853527379367287e-07,
      "loss": 0.0017,
      "step": 51840
    },
    {
      "epoch": 0.8484005563282336,
      "grad_norm": 0.21107843270472454,
      "learning_rate": 6.839103173773249e-07,
      "loss": 0.0019,
      "step": 51850
    },
    {
      "epoch": 0.8485641822793095,
      "grad_norm": 0.07236215742971258,
      "learning_rate": 6.824693048658493e-07,
      "loss": 0.0022,
      "step": 51860
    },
    {
      "epoch": 0.8487278082303853,
      "grad_norm": 0.10434310910458901,
      "learning_rate": 6.810297008724076e-07,
      "loss": 0.0019,
      "step": 51870
    },
    {
      "epoch": 0.8488914341814612,
      "grad_norm": 0.31593886723942927,
      "learning_rate": 6.795915058666475e-07,
      "loss": 0.0018,
      "step": 51880
    },
    {
      "epoch": 0.8490550601325371,
      "grad_norm": 0.0794576665774269,
      "learning_rate": 6.781547203177535e-07,
      "loss": 0.002,
      "step": 51890
    },
    {
      "epoch": 0.8492186860836128,
      "grad_norm": 0.09240107521271751,
      "learning_rate": 6.767193446944498e-07,
      "loss": 0.0027,
      "step": 51900
    },
    {
      "epoch": 0.8493823120346887,
      "grad_norm": 0.014270476704582336,
      "learning_rate": 6.752853794650066e-07,
      "loss": 0.0013,
      "step": 51910
    },
    {
      "epoch": 0.8495459379857645,
      "grad_norm": 0.27294100711286085,
      "learning_rate": 6.738528250972254e-07,
      "loss": 0.003,
      "step": 51920
    },
    {
      "epoch": 0.8497095639368404,
      "grad_norm": 0.061180788937362555,
      "learning_rate": 6.724216820584523e-07,
      "loss": 0.0022,
      "step": 51930
    },
    {
      "epoch": 0.8498731898879163,
      "grad_norm": 0.04180615470625443,
      "learning_rate": 6.709919508155743e-07,
      "loss": 0.0017,
      "step": 51940
    },
    {
      "epoch": 0.850036815838992,
      "grad_norm": 0.15860955205243665,
      "learning_rate": 6.695636318350146e-07,
      "loss": 0.002,
      "step": 51950
    },
    {
      "epoch": 0.8502004417900679,
      "grad_norm": 0.17798542217976646,
      "learning_rate": 6.68136725582737e-07,
      "loss": 0.0074,
      "step": 51960
    },
    {
      "epoch": 0.8503640677411437,
      "grad_norm": 0.11193149267452511,
      "learning_rate": 6.66711232524247e-07,
      "loss": 0.0015,
      "step": 51970
    },
    {
      "epoch": 0.8505276936922196,
      "grad_norm": 0.21659725194057805,
      "learning_rate": 6.652871531245824e-07,
      "loss": 0.0025,
      "step": 51980
    },
    {
      "epoch": 0.8506913196432955,
      "grad_norm": 0.11564710038302363,
      "learning_rate": 6.638644878483258e-07,
      "loss": 0.0023,
      "step": 51990
    },
    {
      "epoch": 0.8508549455943712,
      "grad_norm": 0.07542550323447239,
      "learning_rate": 6.624432371595973e-07,
      "loss": 0.0018,
      "step": 52000
    },
    {
      "epoch": 0.8510185715454471,
      "grad_norm": 0.05280554518032804,
      "learning_rate": 6.61023401522054e-07,
      "loss": 0.0016,
      "step": 52010
    },
    {
      "epoch": 0.8511821974965229,
      "grad_norm": 0.06165095218360115,
      "learning_rate": 6.596049813988914e-07,
      "loss": 0.003,
      "step": 52020
    },
    {
      "epoch": 0.8513458234475988,
      "grad_norm": 0.13644096325430377,
      "learning_rate": 6.581879772528465e-07,
      "loss": 0.0017,
      "step": 52030
    },
    {
      "epoch": 0.8515094493986747,
      "grad_norm": 0.201970400449003,
      "learning_rate": 6.567723895461897e-07,
      "loss": 0.0027,
      "step": 52040
    },
    {
      "epoch": 0.8516730753497505,
      "grad_norm": 0.061274663988517754,
      "learning_rate": 6.553582187407321e-07,
      "loss": 0.0029,
      "step": 52050
    },
    {
      "epoch": 0.8518367013008263,
      "grad_norm": 0.10336651176071354,
      "learning_rate": 6.539454652978206e-07,
      "loss": 0.0012,
      "step": 52060
    },
    {
      "epoch": 0.8520003272519021,
      "grad_norm": 0.11743499209845157,
      "learning_rate": 6.525341296783433e-07,
      "loss": 0.0014,
      "step": 52070
    },
    {
      "epoch": 0.852163953202978,
      "grad_norm": 0.07829051609670336,
      "learning_rate": 6.51124212342722e-07,
      "loss": 0.0013,
      "step": 52080
    },
    {
      "epoch": 0.8523275791540539,
      "grad_norm": 0.05080957610540467,
      "learning_rate": 6.49715713750918e-07,
      "loss": 0.0015,
      "step": 52090
    },
    {
      "epoch": 0.8524912051051297,
      "grad_norm": 0.0908703857451172,
      "learning_rate": 6.48308634362429e-07,
      "loss": 0.0015,
      "step": 52100
    },
    {
      "epoch": 0.8526548310562055,
      "grad_norm": 0.23704966484065404,
      "learning_rate": 6.469029746362892e-07,
      "loss": 0.0014,
      "step": 52110
    },
    {
      "epoch": 0.8528184570072813,
      "grad_norm": 0.04828670876426694,
      "learning_rate": 6.454987350310699e-07,
      "loss": 0.0022,
      "step": 52120
    },
    {
      "epoch": 0.8529820829583572,
      "grad_norm": 0.10559746989920246,
      "learning_rate": 6.440959160048804e-07,
      "loss": 0.0024,
      "step": 52130
    },
    {
      "epoch": 0.8531457089094331,
      "grad_norm": 0.015323861212025834,
      "learning_rate": 6.426945180153659e-07,
      "loss": 0.0015,
      "step": 52140
    },
    {
      "epoch": 0.8533093348605089,
      "grad_norm": 0.06530436168572278,
      "learning_rate": 6.412945415197069e-07,
      "loss": 0.0022,
      "step": 52150
    },
    {
      "epoch": 0.8534729608115847,
      "grad_norm": 0.07343875600310258,
      "learning_rate": 6.398959869746213e-07,
      "loss": 0.0016,
      "step": 52160
    },
    {
      "epoch": 0.8536365867626605,
      "grad_norm": 0.21771313246539128,
      "learning_rate": 6.38498854836363e-07,
      "loss": 0.0015,
      "step": 52170
    },
    {
      "epoch": 0.8538002127137364,
      "grad_norm": 0.1402612424262812,
      "learning_rate": 6.371031455607196e-07,
      "loss": 0.0014,
      "step": 52180
    },
    {
      "epoch": 0.8539638386648123,
      "grad_norm": 0.04033303822465881,
      "learning_rate": 6.357088596030198e-07,
      "loss": 0.0011,
      "step": 52190
    },
    {
      "epoch": 0.8541274646158881,
      "grad_norm": 0.05997505408967628,
      "learning_rate": 6.343159974181229e-07,
      "loss": 0.0032,
      "step": 52200
    },
    {
      "epoch": 0.854291090566964,
      "grad_norm": 0.18372273530665725,
      "learning_rate": 6.329245594604261e-07,
      "loss": 0.0014,
      "step": 52210
    },
    {
      "epoch": 0.8544547165180397,
      "grad_norm": 0.28965353286235224,
      "learning_rate": 6.315345461838613e-07,
      "loss": 0.0031,
      "step": 52220
    },
    {
      "epoch": 0.8546183424691156,
      "grad_norm": 0.09484990909992451,
      "learning_rate": 6.301459580418956e-07,
      "loss": 0.0029,
      "step": 52230
    },
    {
      "epoch": 0.8547819684201915,
      "grad_norm": 0.016684374255821185,
      "learning_rate": 6.287587954875302e-07,
      "loss": 0.001,
      "step": 52240
    },
    {
      "epoch": 0.8549455943712673,
      "grad_norm": 0.17546523518277643,
      "learning_rate": 6.273730589733046e-07,
      "loss": 0.003,
      "step": 52250
    },
    {
      "epoch": 0.8551092203223432,
      "grad_norm": 0.08778242690724009,
      "learning_rate": 6.259887489512894e-07,
      "loss": 0.0015,
      "step": 52260
    },
    {
      "epoch": 0.8552728462734189,
      "grad_norm": 0.18069549889107744,
      "learning_rate": 6.246058658730919e-07,
      "loss": 0.0015,
      "step": 52270
    },
    {
      "epoch": 0.8554364722244948,
      "grad_norm": 0.19874591977062805,
      "learning_rate": 6.232244101898533e-07,
      "loss": 0.0016,
      "step": 52280
    },
    {
      "epoch": 0.8556000981755707,
      "grad_norm": 0.05776878507644744,
      "learning_rate": 6.218443823522486e-07,
      "loss": 0.0017,
      "step": 52290
    },
    {
      "epoch": 0.8557637241266465,
      "grad_norm": 0.02236539449108147,
      "learning_rate": 6.204657828104865e-07,
      "loss": 0.0033,
      "step": 52300
    },
    {
      "epoch": 0.8559273500777224,
      "grad_norm": 0.09153008946792006,
      "learning_rate": 6.190886120143131e-07,
      "loss": 0.0012,
      "step": 52310
    },
    {
      "epoch": 0.8560909760287981,
      "grad_norm": 0.1574011274374583,
      "learning_rate": 6.177128704130042e-07,
      "loss": 0.0016,
      "step": 52320
    },
    {
      "epoch": 0.856254601979874,
      "grad_norm": 0.05777198234322402,
      "learning_rate": 6.163385584553721e-07,
      "loss": 0.0013,
      "step": 52330
    },
    {
      "epoch": 0.8564182279309498,
      "grad_norm": 0.16200610508291988,
      "learning_rate": 6.149656765897616e-07,
      "loss": 0.0021,
      "step": 52340
    },
    {
      "epoch": 0.8565818538820257,
      "grad_norm": 0.11574745523289802,
      "learning_rate": 6.135942252640509e-07,
      "loss": 0.0014,
      "step": 52350
    },
    {
      "epoch": 0.8567454798331016,
      "grad_norm": 0.0717968683026434,
      "learning_rate": 6.122242049256505e-07,
      "loss": 0.0014,
      "step": 52360
    },
    {
      "epoch": 0.8569091057841773,
      "grad_norm": 0.15415360375355136,
      "learning_rate": 6.108556160215078e-07,
      "loss": 0.002,
      "step": 52370
    },
    {
      "epoch": 0.8570727317352532,
      "grad_norm": 0.09176524267552756,
      "learning_rate": 6.094884589980993e-07,
      "loss": 0.0027,
      "step": 52380
    },
    {
      "epoch": 0.857236357686329,
      "grad_norm": 0.07484066561072046,
      "learning_rate": 6.081227343014362e-07,
      "loss": 0.0014,
      "step": 52390
    },
    {
      "epoch": 0.8573999836374049,
      "grad_norm": 0.0938109351429162,
      "learning_rate": 6.067584423770623e-07,
      "loss": 0.0019,
      "step": 52400
    },
    {
      "epoch": 0.8575636095884808,
      "grad_norm": 0.0585635815222195,
      "learning_rate": 6.053955836700525e-07,
      "loss": 0.0021,
      "step": 52410
    },
    {
      "epoch": 0.8577272355395565,
      "grad_norm": 0.13194656612859215,
      "learning_rate": 6.040341586250165e-07,
      "loss": 0.0027,
      "step": 52420
    },
    {
      "epoch": 0.8578908614906324,
      "grad_norm": 0.09135163056146216,
      "learning_rate": 6.026741676860937e-07,
      "loss": 0.0024,
      "step": 52430
    },
    {
      "epoch": 0.8580544874417082,
      "grad_norm": 0.05650079473641278,
      "learning_rate": 6.013156112969592e-07,
      "loss": 0.0026,
      "step": 52440
    },
    {
      "epoch": 0.8582181133927841,
      "grad_norm": 0.09003152257225695,
      "learning_rate": 5.999584899008171e-07,
      "loss": 0.0018,
      "step": 52450
    },
    {
      "epoch": 0.85838173934386,
      "grad_norm": 0.06036638937432826,
      "learning_rate": 5.986028039404035e-07,
      "loss": 0.0016,
      "step": 52460
    },
    {
      "epoch": 0.8585453652949357,
      "grad_norm": 0.23854734828025737,
      "learning_rate": 5.972485538579881e-07,
      "loss": 0.0041,
      "step": 52470
    },
    {
      "epoch": 0.8587089912460116,
      "grad_norm": 0.1515366610704092,
      "learning_rate": 5.958957400953702e-07,
      "loss": 0.0017,
      "step": 52480
    },
    {
      "epoch": 0.8588726171970874,
      "grad_norm": 0.22480788600762538,
      "learning_rate": 5.945443630938808e-07,
      "loss": 0.0019,
      "step": 52490
    },
    {
      "epoch": 0.8590362431481633,
      "grad_norm": 0.2984675505008906,
      "learning_rate": 5.931944232943848e-07,
      "loss": 0.0021,
      "step": 52500
    },
    {
      "epoch": 0.8591998690992392,
      "grad_norm": 0.1165462403644148,
      "learning_rate": 5.918459211372752e-07,
      "loss": 0.003,
      "step": 52510
    },
    {
      "epoch": 0.859363495050315,
      "grad_norm": 0.20867340547815147,
      "learning_rate": 5.904988570624765e-07,
      "loss": 0.0021,
      "step": 52520
    },
    {
      "epoch": 0.8595271210013908,
      "grad_norm": 0.19202000377979206,
      "learning_rate": 5.891532315094444e-07,
      "loss": 0.0029,
      "step": 52530
    },
    {
      "epoch": 0.8596907469524666,
      "grad_norm": 0.117843679234529,
      "learning_rate": 5.878090449171664e-07,
      "loss": 0.0013,
      "step": 52540
    },
    {
      "epoch": 0.8598543729035425,
      "grad_norm": 0.14644827473227537,
      "learning_rate": 5.864662977241581e-07,
      "loss": 0.0023,
      "step": 52550
    },
    {
      "epoch": 0.8600179988546184,
      "grad_norm": 0.027084560741157386,
      "learning_rate": 5.851249903684686e-07,
      "loss": 0.0014,
      "step": 52560
    },
    {
      "epoch": 0.8601816248056942,
      "grad_norm": 0.11779208060748769,
      "learning_rate": 5.837851232876756e-07,
      "loss": 0.0021,
      "step": 52570
    },
    {
      "epoch": 0.86034525075677,
      "grad_norm": 0.2130106261048795,
      "learning_rate": 5.824466969188863e-07,
      "loss": 0.0027,
      "step": 52580
    },
    {
      "epoch": 0.8605088767078458,
      "grad_norm": 0.0961055460582622,
      "learning_rate": 5.811097116987386e-07,
      "loss": 0.0017,
      "step": 52590
    },
    {
      "epoch": 0.8606725026589217,
      "grad_norm": 0.19588721980560836,
      "learning_rate": 5.797741680634006e-07,
      "loss": 0.0015,
      "step": 52600
    },
    {
      "epoch": 0.8608361286099976,
      "grad_norm": 0.07420320083143492,
      "learning_rate": 5.78440066448569e-07,
      "loss": 0.0014,
      "step": 52610
    },
    {
      "epoch": 0.8609997545610734,
      "grad_norm": 0.31134719831976676,
      "learning_rate": 5.771074072894728e-07,
      "loss": 0.0056,
      "step": 52620
    },
    {
      "epoch": 0.8611633805121492,
      "grad_norm": 0.024279194574428942,
      "learning_rate": 5.757761910208671e-07,
      "loss": 0.0021,
      "step": 52630
    },
    {
      "epoch": 0.861327006463225,
      "grad_norm": 0.22763556971819193,
      "learning_rate": 5.744464180770376e-07,
      "loss": 0.0025,
      "step": 52640
    },
    {
      "epoch": 0.8614906324143009,
      "grad_norm": 0.07547680740189375,
      "learning_rate": 5.731180888917997e-07,
      "loss": 0.0034,
      "step": 52650
    },
    {
      "epoch": 0.8616542583653768,
      "grad_norm": 0.0724695126313348,
      "learning_rate": 5.717912038984968e-07,
      "loss": 0.0027,
      "step": 52660
    },
    {
      "epoch": 0.8618178843164526,
      "grad_norm": 0.2455345367765211,
      "learning_rate": 5.704657635300015e-07,
      "loss": 0.002,
      "step": 52670
    },
    {
      "epoch": 0.8619815102675285,
      "grad_norm": 0.09865935846515249,
      "learning_rate": 5.691417682187162e-07,
      "loss": 0.0013,
      "step": 52680
    },
    {
      "epoch": 0.8621451362186042,
      "grad_norm": 0.02524261990991333,
      "learning_rate": 5.678192183965708e-07,
      "loss": 0.0018,
      "step": 52690
    },
    {
      "epoch": 0.8623087621696801,
      "grad_norm": 0.22105627442863288,
      "learning_rate": 5.664981144950243e-07,
      "loss": 0.0018,
      "step": 52700
    },
    {
      "epoch": 0.862472388120756,
      "grad_norm": 0.20563804941198235,
      "learning_rate": 5.651784569450603e-07,
      "loss": 0.0019,
      "step": 52710
    },
    {
      "epoch": 0.8626360140718318,
      "grad_norm": 0.24014956448128946,
      "learning_rate": 5.638602461771975e-07,
      "loss": 0.0023,
      "step": 52720
    },
    {
      "epoch": 0.8627996400229077,
      "grad_norm": 0.020529044658343633,
      "learning_rate": 5.625434826214776e-07,
      "loss": 0.0018,
      "step": 52730
    },
    {
      "epoch": 0.8629632659739834,
      "grad_norm": 0.227136884496555,
      "learning_rate": 5.612281667074698e-07,
      "loss": 0.0017,
      "step": 52740
    },
    {
      "epoch": 0.8631268919250593,
      "grad_norm": 0.23116390876608647,
      "learning_rate": 5.599142988642753e-07,
      "loss": 0.0024,
      "step": 52750
    },
    {
      "epoch": 0.8632905178761352,
      "grad_norm": 0.11937413835087557,
      "learning_rate": 5.5860187952052e-07,
      "loss": 0.0017,
      "step": 52760
    },
    {
      "epoch": 0.863454143827211,
      "grad_norm": 0.10569138418165225,
      "learning_rate": 5.572909091043549e-07,
      "loss": 0.003,
      "step": 52770
    },
    {
      "epoch": 0.8636177697782869,
      "grad_norm": 0.15475087380870345,
      "learning_rate": 5.559813880434634e-07,
      "loss": 0.0015,
      "step": 52780
    },
    {
      "epoch": 0.8637813957293626,
      "grad_norm": 0.04565700061160876,
      "learning_rate": 5.546733167650526e-07,
      "loss": 0.0026,
      "step": 52790
    },
    {
      "epoch": 0.8639450216804385,
      "grad_norm": 0.07683103550785202,
      "learning_rate": 5.533666956958567e-07,
      "loss": 0.0016,
      "step": 52800
    },
    {
      "epoch": 0.8641086476315144,
      "grad_norm": 0.14157236883376392,
      "learning_rate": 5.520615252621397e-07,
      "loss": 0.003,
      "step": 52810
    },
    {
      "epoch": 0.8642722735825902,
      "grad_norm": 0.16044370082407694,
      "learning_rate": 5.507578058896906e-07,
      "loss": 0.0025,
      "step": 52820
    },
    {
      "epoch": 0.8644358995336661,
      "grad_norm": 0.08129999725963098,
      "learning_rate": 5.494555380038219e-07,
      "loss": 0.0033,
      "step": 52830
    },
    {
      "epoch": 0.8645995254847418,
      "grad_norm": 0.09973333399020222,
      "learning_rate": 5.481547220293781e-07,
      "loss": 0.0014,
      "step": 52840
    },
    {
      "epoch": 0.8647631514358177,
      "grad_norm": 0.018302630775638176,
      "learning_rate": 5.468553583907265e-07,
      "loss": 0.0023,
      "step": 52850
    },
    {
      "epoch": 0.8649267773868936,
      "grad_norm": 0.08830658747873654,
      "learning_rate": 5.455574475117609e-07,
      "loss": 0.0011,
      "step": 52860
    },
    {
      "epoch": 0.8650904033379694,
      "grad_norm": 0.10012930289047829,
      "learning_rate": 5.44260989815904e-07,
      "loss": 0.0024,
      "step": 52870
    },
    {
      "epoch": 0.8652540292890453,
      "grad_norm": 0.0860343061252671,
      "learning_rate": 5.429659857260994e-07,
      "loss": 0.0023,
      "step": 52880
    },
    {
      "epoch": 0.865417655240121,
      "grad_norm": 0.09569661808726233,
      "learning_rate": 5.416724356648201e-07,
      "loss": 0.0011,
      "step": 52890
    },
    {
      "epoch": 0.8655812811911969,
      "grad_norm": 0.23115574330581126,
      "learning_rate": 5.403803400540647e-07,
      "loss": 0.0019,
      "step": 52900
    },
    {
      "epoch": 0.8657449071422728,
      "grad_norm": 0.15052083013044068,
      "learning_rate": 5.390896993153566e-07,
      "loss": 0.0018,
      "step": 52910
    },
    {
      "epoch": 0.8659085330933486,
      "grad_norm": 0.32993503018569426,
      "learning_rate": 5.378005138697429e-07,
      "loss": 0.0016,
      "step": 52920
    },
    {
      "epoch": 0.8660721590444245,
      "grad_norm": 0.13224367959295272,
      "learning_rate": 5.365127841378009e-07,
      "loss": 0.0013,
      "step": 52930
    },
    {
      "epoch": 0.8662357849955002,
      "grad_norm": 0.06792313523280331,
      "learning_rate": 5.352265105396265e-07,
      "loss": 0.0009,
      "step": 52940
    },
    {
      "epoch": 0.8663994109465761,
      "grad_norm": 0.27876454923053867,
      "learning_rate": 5.339416934948432e-07,
      "loss": 0.003,
      "step": 52950
    },
    {
      "epoch": 0.866563036897652,
      "grad_norm": 0.2061583456972525,
      "learning_rate": 5.326583334226027e-07,
      "loss": 0.0017,
      "step": 52960
    },
    {
      "epoch": 0.8667266628487278,
      "grad_norm": 0.1333709945814969,
      "learning_rate": 5.313764307415775e-07,
      "loss": 0.002,
      "step": 52970
    },
    {
      "epoch": 0.8668902887998037,
      "grad_norm": 0.09200962338132083,
      "learning_rate": 5.300959858699639e-07,
      "loss": 0.0027,
      "step": 52980
    },
    {
      "epoch": 0.8670539147508795,
      "grad_norm": 0.28327725025852285,
      "learning_rate": 5.288169992254876e-07,
      "loss": 0.0028,
      "step": 52990
    },
    {
      "epoch": 0.8672175407019553,
      "grad_norm": 0.0716261702395288,
      "learning_rate": 5.275394712253934e-07,
      "loss": 0.0042,
      "step": 53000
    },
    {
      "epoch": 0.8673811666530312,
      "grad_norm": 0.17379223396592003,
      "learning_rate": 5.26263402286451e-07,
      "loss": 0.0028,
      "step": 53010
    },
    {
      "epoch": 0.867544792604107,
      "grad_norm": 0.09973840359683082,
      "learning_rate": 5.249887928249575e-07,
      "loss": 0.0026,
      "step": 53020
    },
    {
      "epoch": 0.8677084185551829,
      "grad_norm": 0.004321720996621527,
      "learning_rate": 5.237156432567314e-07,
      "loss": 0.0008,
      "step": 53030
    },
    {
      "epoch": 0.8678720445062587,
      "grad_norm": 0.12832616328743654,
      "learning_rate": 5.224439539971149e-07,
      "loss": 0.0025,
      "step": 53040
    },
    {
      "epoch": 0.8680356704573345,
      "grad_norm": 0.1292776939667015,
      "learning_rate": 5.211737254609733e-07,
      "loss": 0.0025,
      "step": 53050
    },
    {
      "epoch": 0.8681992964084104,
      "grad_norm": 0.15244731600139727,
      "learning_rate": 5.199049580626975e-07,
      "loss": 0.0012,
      "step": 53060
    },
    {
      "epoch": 0.8683629223594862,
      "grad_norm": 0.09142084590072416,
      "learning_rate": 5.186376522162001e-07,
      "loss": 0.0011,
      "step": 53070
    },
    {
      "epoch": 0.8685265483105621,
      "grad_norm": 0.04002642863934641,
      "learning_rate": 5.173718083349155e-07,
      "loss": 0.002,
      "step": 53080
    },
    {
      "epoch": 0.8686901742616379,
      "grad_norm": 0.019807730514134005,
      "learning_rate": 5.161074268318051e-07,
      "loss": 0.0027,
      "step": 53090
    },
    {
      "epoch": 0.8688538002127137,
      "grad_norm": 0.14325996360228122,
      "learning_rate": 5.148445081193503e-07,
      "loss": 0.0023,
      "step": 53100
    },
    {
      "epoch": 0.8690174261637896,
      "grad_norm": 0.12808615726841782,
      "learning_rate": 5.135830526095564e-07,
      "loss": 0.002,
      "step": 53110
    },
    {
      "epoch": 0.8691810521148654,
      "grad_norm": 0.02043394104307728,
      "learning_rate": 5.123230607139507e-07,
      "loss": 0.0014,
      "step": 53120
    },
    {
      "epoch": 0.8693446780659413,
      "grad_norm": 0.13202722662127603,
      "learning_rate": 5.110645328435826e-07,
      "loss": 0.0017,
      "step": 53130
    },
    {
      "epoch": 0.8695083040170171,
      "grad_norm": 0.10325424969490757,
      "learning_rate": 5.098074694090238e-07,
      "loss": 0.0015,
      "step": 53140
    },
    {
      "epoch": 0.869671929968093,
      "grad_norm": 0.15111091453422165,
      "learning_rate": 5.085518708203713e-07,
      "loss": 0.0018,
      "step": 53150
    },
    {
      "epoch": 0.8698355559191688,
      "grad_norm": 0.1971086113314265,
      "learning_rate": 5.072977374872412e-07,
      "loss": 0.0043,
      "step": 53160
    },
    {
      "epoch": 0.8699991818702446,
      "grad_norm": 0.16472595260359932,
      "learning_rate": 5.060450698187719e-07,
      "loss": 0.002,
      "step": 53170
    },
    {
      "epoch": 0.8701628078213205,
      "grad_norm": 0.31783381371282987,
      "learning_rate": 5.047938682236242e-07,
      "loss": 0.0016,
      "step": 53180
    },
    {
      "epoch": 0.8703264337723963,
      "grad_norm": 0.1423298218776779,
      "learning_rate": 5.035441331099805e-07,
      "loss": 0.0017,
      "step": 53190
    },
    {
      "epoch": 0.8704900597234722,
      "grad_norm": 0.04422393663621081,
      "learning_rate": 5.022958648855436e-07,
      "loss": 0.0026,
      "step": 53200
    },
    {
      "epoch": 0.8706536856745479,
      "grad_norm": 0.365739191864283,
      "learning_rate": 5.01049063957541e-07,
      "loss": 0.0034,
      "step": 53210
    },
    {
      "epoch": 0.8708173116256238,
      "grad_norm": 0.08494119388996696,
      "learning_rate": 4.998037307327186e-07,
      "loss": 0.0011,
      "step": 53220
    },
    {
      "epoch": 0.8709809375766997,
      "grad_norm": 0.11808998918537728,
      "learning_rate": 4.985598656173446e-07,
      "loss": 0.0053,
      "step": 53230
    },
    {
      "epoch": 0.8711445635277755,
      "grad_norm": 0.21010444009633525,
      "learning_rate": 4.973174690172072e-07,
      "loss": 0.0026,
      "step": 53240
    },
    {
      "epoch": 0.8713081894788514,
      "grad_norm": 0.019209233079003914,
      "learning_rate": 4.960765413376173e-07,
      "loss": 0.0007,
      "step": 53250
    },
    {
      "epoch": 0.8714718154299271,
      "grad_norm": 0.07876334253518731,
      "learning_rate": 4.948370829834037e-07,
      "loss": 0.0017,
      "step": 53260
    },
    {
      "epoch": 0.871635441381003,
      "grad_norm": 0.10374913829084749,
      "learning_rate": 4.935990943589203e-07,
      "loss": 0.0014,
      "step": 53270
    },
    {
      "epoch": 0.8717990673320789,
      "grad_norm": 0.06539122247159852,
      "learning_rate": 4.923625758680378e-07,
      "loss": 0.0018,
      "step": 53280
    },
    {
      "epoch": 0.8719626932831547,
      "grad_norm": 0.2039532097257218,
      "learning_rate": 4.911275279141486e-07,
      "loss": 0.0022,
      "step": 53290
    },
    {
      "epoch": 0.8721263192342306,
      "grad_norm": 0.09022899773557073,
      "learning_rate": 4.898939509001649e-07,
      "loss": 0.0019,
      "step": 53300
    },
    {
      "epoch": 0.8722899451853063,
      "grad_norm": 0.08054252312342947,
      "learning_rate": 4.886618452285202e-07,
      "loss": 0.0028,
      "step": 53310
    },
    {
      "epoch": 0.8724535711363822,
      "grad_norm": 0.3406458898516414,
      "learning_rate": 4.874312113011653e-07,
      "loss": 0.002,
      "step": 53320
    },
    {
      "epoch": 0.8726171970874581,
      "grad_norm": 0.0938125858484772,
      "learning_rate": 4.862020495195752e-07,
      "loss": 0.0036,
      "step": 53330
    },
    {
      "epoch": 0.8727808230385339,
      "grad_norm": 0.21807165308440107,
      "learning_rate": 4.849743602847412e-07,
      "loss": 0.0024,
      "step": 53340
    },
    {
      "epoch": 0.8729444489896098,
      "grad_norm": 0.1649710947303191,
      "learning_rate": 4.837481439971753e-07,
      "loss": 0.002,
      "step": 53350
    },
    {
      "epoch": 0.8731080749406855,
      "grad_norm": 0.2248221708402171,
      "learning_rate": 4.825234010569086e-07,
      "loss": 0.0011,
      "step": 53360
    },
    {
      "epoch": 0.8732717008917614,
      "grad_norm": 0.09772602427703993,
      "learning_rate": 4.813001318634919e-07,
      "loss": 0.0023,
      "step": 53370
    },
    {
      "epoch": 0.8734353268428373,
      "grad_norm": 0.17687873426667197,
      "learning_rate": 4.800783368159951e-07,
      "loss": 0.002,
      "step": 53380
    },
    {
      "epoch": 0.8735989527939131,
      "grad_norm": 0.2108073740248584,
      "learning_rate": 4.788580163130064e-07,
      "loss": 0.0064,
      "step": 53390
    },
    {
      "epoch": 0.873762578744989,
      "grad_norm": 0.04967326737688878,
      "learning_rate": 4.776391707526351e-07,
      "loss": 0.0017,
      "step": 53400
    },
    {
      "epoch": 0.8739262046960647,
      "grad_norm": 0.07946181666706736,
      "learning_rate": 4.764218005325083e-07,
      "loss": 0.0015,
      "step": 53410
    },
    {
      "epoch": 0.8740898306471406,
      "grad_norm": 0.06337516191278485,
      "learning_rate": 4.7520590604977044e-07,
      "loss": 0.0015,
      "step": 53420
    },
    {
      "epoch": 0.8742534565982165,
      "grad_norm": 0.1201558068281623,
      "learning_rate": 4.739914877010865e-07,
      "loss": 0.0012,
      "step": 53430
    },
    {
      "epoch": 0.8744170825492923,
      "grad_norm": 0.14414874058411617,
      "learning_rate": 4.727785458826378e-07,
      "loss": 0.0023,
      "step": 53440
    },
    {
      "epoch": 0.8745807085003682,
      "grad_norm": 0.09523403081110585,
      "learning_rate": 4.7156708099012516e-07,
      "loss": 0.0015,
      "step": 53450
    },
    {
      "epoch": 0.874744334451444,
      "grad_norm": 0.15767513204009692,
      "learning_rate": 4.70357093418769e-07,
      "loss": 0.0014,
      "step": 53460
    },
    {
      "epoch": 0.8749079604025198,
      "grad_norm": 0.14282526627084383,
      "learning_rate": 4.6914858356330585e-07,
      "loss": 0.0031,
      "step": 53470
    },
    {
      "epoch": 0.8750715863535957,
      "grad_norm": 0.19851638681558015,
      "learning_rate": 4.6794155181798973e-07,
      "loss": 0.0028,
      "step": 53480
    },
    {
      "epoch": 0.8752352123046715,
      "grad_norm": 0.13470833926114842,
      "learning_rate": 4.667359985765946e-07,
      "loss": 0.0012,
      "step": 53490
    },
    {
      "epoch": 0.8753988382557474,
      "grad_norm": 0.11728149889212863,
      "learning_rate": 4.6553192423241023e-07,
      "loss": 0.0023,
      "step": 53500
    },
    {
      "epoch": 0.8755624642068232,
      "grad_norm": 0.10738486319585844,
      "learning_rate": 4.6432932917824414e-07,
      "loss": 0.0024,
      "step": 53510
    },
    {
      "epoch": 0.875726090157899,
      "grad_norm": 0.129271038427575,
      "learning_rate": 4.6312821380642303e-07,
      "loss": 0.0014,
      "step": 53520
    },
    {
      "epoch": 0.8758897161089749,
      "grad_norm": 0.38420873380307424,
      "learning_rate": 4.619285785087885e-07,
      "loss": 0.0017,
      "step": 53530
    },
    {
      "epoch": 0.8760533420600507,
      "grad_norm": 0.07116971014817766,
      "learning_rate": 4.6073042367670073e-07,
      "loss": 0.0031,
      "step": 53540
    },
    {
      "epoch": 0.8762169680111266,
      "grad_norm": 0.1665876892809746,
      "learning_rate": 4.595337497010366e-07,
      "loss": 0.0017,
      "step": 53550
    },
    {
      "epoch": 0.8763805939622024,
      "grad_norm": 0.2621137902335014,
      "learning_rate": 4.5833855697218986e-07,
      "loss": 0.0023,
      "step": 53560
    },
    {
      "epoch": 0.8765442199132782,
      "grad_norm": 0.07771633953405434,
      "learning_rate": 4.571448458800698e-07,
      "loss": 0.001,
      "step": 53570
    },
    {
      "epoch": 0.8767078458643541,
      "grad_norm": 0.038610590107402844,
      "learning_rate": 4.559526168141054e-07,
      "loss": 0.0012,
      "step": 53580
    },
    {
      "epoch": 0.8768714718154299,
      "grad_norm": 0.31763014093488184,
      "learning_rate": 4.547618701632395e-07,
      "loss": 0.0015,
      "step": 53590
    },
    {
      "epoch": 0.8770350977665058,
      "grad_norm": 0.17053397122383354,
      "learning_rate": 4.535726063159318e-07,
      "loss": 0.002,
      "step": 53600
    },
    {
      "epoch": 0.8771987237175816,
      "grad_norm": 0.04642860460063473,
      "learning_rate": 4.523848256601593e-07,
      "loss": 0.0024,
      "step": 53610
    },
    {
      "epoch": 0.8773623496686574,
      "grad_norm": 0.2898670144995844,
      "learning_rate": 4.511985285834142e-07,
      "loss": 0.0015,
      "step": 53620
    },
    {
      "epoch": 0.8775259756197333,
      "grad_norm": 0.04486655775339955,
      "learning_rate": 4.5001371547270354e-07,
      "loss": 0.0019,
      "step": 53630
    },
    {
      "epoch": 0.8776896015708091,
      "grad_norm": 0.15206448318997695,
      "learning_rate": 4.488303867145538e-07,
      "loss": 0.0017,
      "step": 53640
    },
    {
      "epoch": 0.877853227521885,
      "grad_norm": 0.32105805752573213,
      "learning_rate": 4.476485426950045e-07,
      "loss": 0.0011,
      "step": 53650
    },
    {
      "epoch": 0.8780168534729608,
      "grad_norm": 0.021227266076589566,
      "learning_rate": 4.46468183799611e-07,
      "loss": 0.0016,
      "step": 53660
    },
    {
      "epoch": 0.8781804794240367,
      "grad_norm": 0.3212763073713288,
      "learning_rate": 4.4528931041344427e-07,
      "loss": 0.0028,
      "step": 53670
    },
    {
      "epoch": 0.8783441053751125,
      "grad_norm": 0.034399933806645556,
      "learning_rate": 4.4411192292109106e-07,
      "loss": 0.0028,
      "step": 53680
    },
    {
      "epoch": 0.8785077313261883,
      "grad_norm": 0.1969253891794196,
      "learning_rate": 4.4293602170665363e-07,
      "loss": 0.0025,
      "step": 53690
    },
    {
      "epoch": 0.8786713572772642,
      "grad_norm": 0.13290308509176238,
      "learning_rate": 4.417616071537478e-07,
      "loss": 0.0015,
      "step": 53700
    },
    {
      "epoch": 0.87883498322834,
      "grad_norm": 0.09727129804254891,
      "learning_rate": 4.405886796455072e-07,
      "loss": 0.0015,
      "step": 53710
    },
    {
      "epoch": 0.8789986091794159,
      "grad_norm": 0.0644514350780173,
      "learning_rate": 4.3941723956457847e-07,
      "loss": 0.0011,
      "step": 53720
    },
    {
      "epoch": 0.8791622351304917,
      "grad_norm": 0.10295240686542464,
      "learning_rate": 4.3824728729312147e-07,
      "loss": 0.0014,
      "step": 53730
    },
    {
      "epoch": 0.8793258610815675,
      "grad_norm": 0.0783082125628668,
      "learning_rate": 4.3707882321281423e-07,
      "loss": 0.0029,
      "step": 53740
    },
    {
      "epoch": 0.8794894870326434,
      "grad_norm": 0.3203794090684449,
      "learning_rate": 4.3591184770484684e-07,
      "loss": 0.002,
      "step": 53750
    },
    {
      "epoch": 0.8796531129837192,
      "grad_norm": 0.2802376506305925,
      "learning_rate": 4.3474636114992306e-07,
      "loss": 0.0034,
      "step": 53760
    },
    {
      "epoch": 0.8798167389347951,
      "grad_norm": 0.046880285925879714,
      "learning_rate": 4.3358236392826546e-07,
      "loss": 0.0023,
      "step": 53770
    },
    {
      "epoch": 0.879980364885871,
      "grad_norm": 0.06307601033764723,
      "learning_rate": 4.3241985641960584e-07,
      "loss": 0.002,
      "step": 53780
    },
    {
      "epoch": 0.8801439908369467,
      "grad_norm": 0.16857441040650556,
      "learning_rate": 4.312588390031902e-07,
      "loss": 0.0026,
      "step": 53790
    },
    {
      "epoch": 0.8803076167880226,
      "grad_norm": 0.01280808876797322,
      "learning_rate": 4.300993120577823e-07,
      "loss": 0.0013,
      "step": 53800
    },
    {
      "epoch": 0.8804712427390984,
      "grad_norm": 0.06734894634459568,
      "learning_rate": 4.2894127596165625e-07,
      "loss": 0.0018,
      "step": 53810
    },
    {
      "epoch": 0.8806348686901743,
      "grad_norm": 0.23419157677032917,
      "learning_rate": 4.2778473109260033e-07,
      "loss": 0.0023,
      "step": 53820
    },
    {
      "epoch": 0.8807984946412502,
      "grad_norm": 0.027255321763394916,
      "learning_rate": 4.2662967782791774e-07,
      "loss": 0.0018,
      "step": 53830
    },
    {
      "epoch": 0.8809621205923259,
      "grad_norm": 0.10411335262980473,
      "learning_rate": 4.2547611654442547e-07,
      "loss": 0.0032,
      "step": 53840
    },
    {
      "epoch": 0.8811257465434018,
      "grad_norm": 0.11376777266235329,
      "learning_rate": 4.2432404761844903e-07,
      "loss": 0.0011,
      "step": 53850
    },
    {
      "epoch": 0.8812893724944776,
      "grad_norm": 0.27601367047008346,
      "learning_rate": 4.231734714258329e-07,
      "loss": 0.0014,
      "step": 53860
    },
    {
      "epoch": 0.8814529984455535,
      "grad_norm": 0.06169260025959098,
      "learning_rate": 4.2202438834193227e-07,
      "loss": 0.0014,
      "step": 53870
    },
    {
      "epoch": 0.8816166243966294,
      "grad_norm": 0.14622464552539996,
      "learning_rate": 4.208767987416135e-07,
      "loss": 0.0019,
      "step": 53880
    },
    {
      "epoch": 0.8817802503477051,
      "grad_norm": 0.21130735399801473,
      "learning_rate": 4.1973070299925924e-07,
      "loss": 0.0023,
      "step": 53890
    },
    {
      "epoch": 0.881943876298781,
      "grad_norm": 0.08930553175493725,
      "learning_rate": 4.185861014887632e-07,
      "loss": 0.0018,
      "step": 53900
    },
    {
      "epoch": 0.8821075022498568,
      "grad_norm": 0.10020663364029199,
      "learning_rate": 4.17442994583529e-07,
      "loss": 0.0023,
      "step": 53910
    },
    {
      "epoch": 0.8822711282009327,
      "grad_norm": 0.05394851757821303,
      "learning_rate": 4.1630138265647766e-07,
      "loss": 0.0012,
      "step": 53920
    },
    {
      "epoch": 0.8824347541520086,
      "grad_norm": 0.029007281318480946,
      "learning_rate": 4.1516126608003806e-07,
      "loss": 0.0021,
      "step": 53930
    },
    {
      "epoch": 0.8825983801030843,
      "grad_norm": 0.18257748327548776,
      "learning_rate": 4.140226452261531e-07,
      "loss": 0.0032,
      "step": 53940
    },
    {
      "epoch": 0.8827620060541602,
      "grad_norm": 0.03882285326528979,
      "learning_rate": 4.128855204662796e-07,
      "loss": 0.0011,
      "step": 53950
    },
    {
      "epoch": 0.882925632005236,
      "grad_norm": 0.17172805248420886,
      "learning_rate": 4.117498921713836e-07,
      "loss": 0.0016,
      "step": 53960
    },
    {
      "epoch": 0.8830892579563119,
      "grad_norm": 0.05375737232576831,
      "learning_rate": 4.1061576071194155e-07,
      "loss": 0.0016,
      "step": 53970
    },
    {
      "epoch": 0.8832528839073878,
      "grad_norm": 0.15040914229733757,
      "learning_rate": 4.094831264579463e-07,
      "loss": 0.0023,
      "step": 53980
    },
    {
      "epoch": 0.8834165098584635,
      "grad_norm": 0.1544821322086737,
      "learning_rate": 4.0835198977889956e-07,
      "loss": 0.0015,
      "step": 53990
    },
    {
      "epoch": 0.8835801358095394,
      "grad_norm": 0.03111987932389926,
      "learning_rate": 4.072223510438128e-07,
      "loss": 0.0013,
      "step": 54000
    },
    {
      "epoch": 0.8837437617606152,
      "grad_norm": 0.08006658021123461,
      "learning_rate": 4.060942106212129e-07,
      "loss": 0.0021,
      "step": 54010
    },
    {
      "epoch": 0.8839073877116911,
      "grad_norm": 0.06874729993856908,
      "learning_rate": 4.0496756887913547e-07,
      "loss": 0.0011,
      "step": 54020
    },
    {
      "epoch": 0.884071013662767,
      "grad_norm": 0.06459473239526173,
      "learning_rate": 4.0384242618512594e-07,
      "loss": 0.0011,
      "step": 54030
    },
    {
      "epoch": 0.8842346396138427,
      "grad_norm": 0.03136468961192377,
      "learning_rate": 4.027187829062423e-07,
      "loss": 0.0026,
      "step": 54040
    },
    {
      "epoch": 0.8843982655649186,
      "grad_norm": 0.1977382234937298,
      "learning_rate": 4.015966394090548e-07,
      "loss": 0.0032,
      "step": 54050
    },
    {
      "epoch": 0.8845618915159944,
      "grad_norm": 0.5797273827607149,
      "learning_rate": 4.0047599605964216e-07,
      "loss": 0.0015,
      "step": 54060
    },
    {
      "epoch": 0.8847255174670703,
      "grad_norm": 0.2302733516883932,
      "learning_rate": 3.9935685322359363e-07,
      "loss": 0.0024,
      "step": 54070
    },
    {
      "epoch": 0.8848891434181462,
      "grad_norm": 0.36832712919217075,
      "learning_rate": 3.982392112660122e-07,
      "loss": 0.0015,
      "step": 54080
    },
    {
      "epoch": 0.885052769369222,
      "grad_norm": 0.24603542305961687,
      "learning_rate": 3.9712307055150666e-07,
      "loss": 0.0017,
      "step": 54090
    },
    {
      "epoch": 0.8852163953202978,
      "grad_norm": 0.05307829525712286,
      "learning_rate": 3.9600843144419756e-07,
      "loss": 0.0037,
      "step": 54100
    },
    {
      "epoch": 0.8853800212713736,
      "grad_norm": 0.17170822710118297,
      "learning_rate": 3.9489529430771833e-07,
      "loss": 0.001,
      "step": 54110
    },
    {
      "epoch": 0.8855436472224495,
      "grad_norm": 0.07291546121777329,
      "learning_rate": 3.9378365950520916e-07,
      "loss": 0.0017,
      "step": 54120
    },
    {
      "epoch": 0.8857072731735253,
      "grad_norm": 0.19535512247320605,
      "learning_rate": 3.926735273993215e-07,
      "loss": 0.0026,
      "step": 54130
    },
    {
      "epoch": 0.8858708991246012,
      "grad_norm": 0.22297715289383083,
      "learning_rate": 3.915648983522158e-07,
      "loss": 0.0013,
      "step": 54140
    },
    {
      "epoch": 0.886034525075677,
      "grad_norm": 0.43808295267869907,
      "learning_rate": 3.9045777272556316e-07,
      "loss": 0.0035,
      "step": 54150
    },
    {
      "epoch": 0.8861981510267528,
      "grad_norm": 0.1879227815058943,
      "learning_rate": 3.893521508805431e-07,
      "loss": 0.0018,
      "step": 54160
    },
    {
      "epoch": 0.8863617769778287,
      "grad_norm": 0.11198352830741005,
      "learning_rate": 3.8824803317784654e-07,
      "loss": 0.002,
      "step": 54170
    },
    {
      "epoch": 0.8865254029289045,
      "grad_norm": 0.05711237659510425,
      "learning_rate": 3.871454199776714e-07,
      "loss": 0.0021,
      "step": 54180
    },
    {
      "epoch": 0.8866890288799804,
      "grad_norm": 0.20370932821518398,
      "learning_rate": 3.8604431163972555e-07,
      "loss": 0.0031,
      "step": 54190
    },
    {
      "epoch": 0.8868526548310562,
      "grad_norm": 0.10589123024168287,
      "learning_rate": 3.8494470852322716e-07,
      "loss": 0.0027,
      "step": 54200
    },
    {
      "epoch": 0.887016280782132,
      "grad_norm": 0.06264924283497242,
      "learning_rate": 3.8384661098690103e-07,
      "loss": 0.0027,
      "step": 54210
    },
    {
      "epoch": 0.8871799067332079,
      "grad_norm": 0.014958898156792496,
      "learning_rate": 3.8275001938898217e-07,
      "loss": 0.0017,
      "step": 54220
    },
    {
      "epoch": 0.8873435326842837,
      "grad_norm": 0.16023502601202608,
      "learning_rate": 3.81654934087215e-07,
      "loss": 0.0016,
      "step": 54230
    },
    {
      "epoch": 0.8875071586353596,
      "grad_norm": 0.2150135482211736,
      "learning_rate": 3.8056135543885153e-07,
      "loss": 0.0018,
      "step": 54240
    },
    {
      "epoch": 0.8876707845864354,
      "grad_norm": 0.04338428722111253,
      "learning_rate": 3.7946928380065254e-07,
      "loss": 0.0017,
      "step": 54250
    },
    {
      "epoch": 0.8878344105375112,
      "grad_norm": 0.16542739333222103,
      "learning_rate": 3.7837871952888683e-07,
      "loss": 0.0028,
      "step": 54260
    },
    {
      "epoch": 0.8879980364885871,
      "grad_norm": 0.03559082868066973,
      "learning_rate": 3.7728966297933156e-07,
      "loss": 0.0027,
      "step": 54270
    },
    {
      "epoch": 0.8881616624396629,
      "grad_norm": 0.14000396759341385,
      "learning_rate": 3.76202114507272e-07,
      "loss": 0.0021,
      "step": 54280
    },
    {
      "epoch": 0.8883252883907388,
      "grad_norm": 0.1545033049940637,
      "learning_rate": 3.75116074467502e-07,
      "loss": 0.0031,
      "step": 54290
    },
    {
      "epoch": 0.8884889143418147,
      "grad_norm": 0.14906306387113252,
      "learning_rate": 3.740315432143232e-07,
      "loss": 0.0017,
      "step": 54300
    },
    {
      "epoch": 0.8886525402928904,
      "grad_norm": 0.08371583695343607,
      "learning_rate": 3.729485211015443e-07,
      "loss": 0.001,
      "step": 54310
    },
    {
      "epoch": 0.8888161662439663,
      "grad_norm": 0.13456061786399692,
      "learning_rate": 3.718670084824827e-07,
      "loss": 0.0033,
      "step": 54320
    },
    {
      "epoch": 0.8889797921950421,
      "grad_norm": 0.24772188772653891,
      "learning_rate": 3.707870057099616e-07,
      "loss": 0.0019,
      "step": 54330
    },
    {
      "epoch": 0.889143418146118,
      "grad_norm": 0.051093700685423474,
      "learning_rate": 3.6970851313631317e-07,
      "loss": 0.0018,
      "step": 54340
    },
    {
      "epoch": 0.8893070440971939,
      "grad_norm": 0.23808305643775388,
      "learning_rate": 3.6863153111337756e-07,
      "loss": 0.0022,
      "step": 54350
    },
    {
      "epoch": 0.8894706700482696,
      "grad_norm": 0.05669352609086232,
      "learning_rate": 3.675560599924999e-07,
      "loss": 0.0022,
      "step": 54360
    },
    {
      "epoch": 0.8896342959993455,
      "grad_norm": 0.24019804719382926,
      "learning_rate": 3.66482100124535e-07,
      "loss": 0.0026,
      "step": 54370
    },
    {
      "epoch": 0.8897979219504213,
      "grad_norm": 0.029697980901882793,
      "learning_rate": 3.6540965185984156e-07,
      "loss": 0.002,
      "step": 54380
    },
    {
      "epoch": 0.8899615479014972,
      "grad_norm": 0.47784508750364657,
      "learning_rate": 3.64338715548288e-07,
      "loss": 0.0035,
      "step": 54390
    },
    {
      "epoch": 0.8901251738525731,
      "grad_norm": 0.10419988491445924,
      "learning_rate": 3.632692915392483e-07,
      "loss": 0.0015,
      "step": 54400
    },
    {
      "epoch": 0.8902887998036488,
      "grad_norm": 0.05081034400484595,
      "learning_rate": 3.622013801816021e-07,
      "loss": 0.0019,
      "step": 54410
    },
    {
      "epoch": 0.8904524257547247,
      "grad_norm": 0.02749762117578119,
      "learning_rate": 3.611349818237375e-07,
      "loss": 0.003,
      "step": 54420
    },
    {
      "epoch": 0.8906160517058005,
      "grad_norm": 0.06531369305662299,
      "learning_rate": 3.60070096813549e-07,
      "loss": 0.0016,
      "step": 54430
    },
    {
      "epoch": 0.8907796776568764,
      "grad_norm": 0.05547720222133388,
      "learning_rate": 3.5900672549843475e-07,
      "loss": 0.0024,
      "step": 54440
    },
    {
      "epoch": 0.8909433036079523,
      "grad_norm": 0.09571061592254314,
      "learning_rate": 3.5794486822530173e-07,
      "loss": 0.0012,
      "step": 54450
    },
    {
      "epoch": 0.891106929559028,
      "grad_norm": 0.21127385161467688,
      "learning_rate": 3.568845253405617e-07,
      "loss": 0.0026,
      "step": 54460
    },
    {
      "epoch": 0.8912705555101039,
      "grad_norm": 0.05100508380499609,
      "learning_rate": 3.5582569719013247e-07,
      "loss": 0.0019,
      "step": 54470
    },
    {
      "epoch": 0.8914341814611797,
      "grad_norm": 0.22273146699160937,
      "learning_rate": 3.547683841194394e-07,
      "loss": 0.004,
      "step": 54480
    },
    {
      "epoch": 0.8915978074122556,
      "grad_norm": 0.1137759305141339,
      "learning_rate": 3.5371258647341145e-07,
      "loss": 0.0023,
      "step": 54490
    },
    {
      "epoch": 0.8917614333633315,
      "grad_norm": 0.13145345088630248,
      "learning_rate": 3.5265830459648375e-07,
      "loss": 0.0038,
      "step": 54500
    },
    {
      "epoch": 0.8919250593144072,
      "grad_norm": 0.26679332576026116,
      "learning_rate": 3.516055388325967e-07,
      "loss": 0.0018,
      "step": 54510
    },
    {
      "epoch": 0.8920886852654831,
      "grad_norm": 0.12013169027463527,
      "learning_rate": 3.505542895251979e-07,
      "loss": 0.0017,
      "step": 54520
    },
    {
      "epoch": 0.8922523112165589,
      "grad_norm": 0.14120997762289345,
      "learning_rate": 3.495045570172362e-07,
      "loss": 0.0015,
      "step": 54530
    },
    {
      "epoch": 0.8924159371676348,
      "grad_norm": 0.12922124181363964,
      "learning_rate": 3.4845634165117103e-07,
      "loss": 0.0016,
      "step": 54540
    },
    {
      "epoch": 0.8925795631187107,
      "grad_norm": 0.35322515244869157,
      "learning_rate": 3.4740964376896334e-07,
      "loss": 0.0034,
      "step": 54550
    },
    {
      "epoch": 0.8927431890697864,
      "grad_norm": 0.0980118241284525,
      "learning_rate": 3.4636446371207945e-07,
      "loss": 0.0026,
      "step": 54560
    },
    {
      "epoch": 0.8929068150208623,
      "grad_norm": 0.06271630709421423,
      "learning_rate": 3.4532080182149095e-07,
      "loss": 0.0018,
      "step": 54570
    },
    {
      "epoch": 0.8930704409719381,
      "grad_norm": 0.09583763064063557,
      "learning_rate": 3.4427865843767395e-07,
      "loss": 0.0011,
      "step": 54580
    },
    {
      "epoch": 0.893234066923014,
      "grad_norm": 0.1465297321123273,
      "learning_rate": 3.4323803390060916e-07,
      "loss": 0.0019,
      "step": 54590
    },
    {
      "epoch": 0.8933976928740899,
      "grad_norm": 0.2375373810117404,
      "learning_rate": 3.421989285497823e-07,
      "loss": 0.002,
      "step": 54600
    },
    {
      "epoch": 0.8935613188251657,
      "grad_norm": 0.18263315913439176,
      "learning_rate": 3.4116134272418333e-07,
      "loss": 0.0026,
      "step": 54610
    },
    {
      "epoch": 0.8937249447762415,
      "grad_norm": 0.05228682221906292,
      "learning_rate": 3.401252767623059e-07,
      "loss": 0.0022,
      "step": 54620
    },
    {
      "epoch": 0.8938885707273173,
      "grad_norm": 0.13450242690776676,
      "learning_rate": 3.390907310021485e-07,
      "loss": 0.0017,
      "step": 54630
    },
    {
      "epoch": 0.8940521966783932,
      "grad_norm": 0.01106225381209532,
      "learning_rate": 3.3805770578121286e-07,
      "loss": 0.0016,
      "step": 54640
    },
    {
      "epoch": 0.8942158226294691,
      "grad_norm": 0.11958272649717445,
      "learning_rate": 3.37026201436505e-07,
      "loss": 0.0041,
      "step": 54650
    },
    {
      "epoch": 0.8943794485805449,
      "grad_norm": 0.13981988791123232,
      "learning_rate": 3.3599621830453623e-07,
      "loss": 0.002,
      "step": 54660
    },
    {
      "epoch": 0.8945430745316207,
      "grad_norm": 0.1086043642402669,
      "learning_rate": 3.3496775672131944e-07,
      "loss": 0.0019,
      "step": 54670
    },
    {
      "epoch": 0.8947067004826965,
      "grad_norm": 0.10485065405472645,
      "learning_rate": 3.339408170223718e-07,
      "loss": 0.001,
      "step": 54680
    },
    {
      "epoch": 0.8948703264337724,
      "grad_norm": 0.047607194239284965,
      "learning_rate": 3.329153995427153e-07,
      "loss": 0.0017,
      "step": 54690
    },
    {
      "epoch": 0.8950339523848483,
      "grad_norm": 0.1282441154058051,
      "learning_rate": 3.31891504616873e-07,
      "loss": 0.0024,
      "step": 54700
    },
    {
      "epoch": 0.8951975783359241,
      "grad_norm": 0.1535722953131099,
      "learning_rate": 3.308691325788732e-07,
      "loss": 0.0029,
      "step": 54710
    },
    {
      "epoch": 0.895361204287,
      "grad_norm": 0.17968293282700348,
      "learning_rate": 3.2984828376224577e-07,
      "loss": 0.0022,
      "step": 54720
    },
    {
      "epoch": 0.8955248302380757,
      "grad_norm": 0.05500806083358326,
      "learning_rate": 3.288289585000265e-07,
      "loss": 0.0021,
      "step": 54730
    },
    {
      "epoch": 0.8956884561891516,
      "grad_norm": 0.1499708842516863,
      "learning_rate": 3.278111571247505e-07,
      "loss": 0.0016,
      "step": 54740
    },
    {
      "epoch": 0.8958520821402275,
      "grad_norm": 0.15846802856555633,
      "learning_rate": 3.267948799684584e-07,
      "loss": 0.0016,
      "step": 54750
    },
    {
      "epoch": 0.8960157080913033,
      "grad_norm": 0.06276388343098598,
      "learning_rate": 3.257801273626926e-07,
      "loss": 0.0017,
      "step": 54760
    },
    {
      "epoch": 0.8961793340423791,
      "grad_norm": 0.040890379777446126,
      "learning_rate": 3.247668996384973e-07,
      "loss": 0.0014,
      "step": 54770
    },
    {
      "epoch": 0.8963429599934549,
      "grad_norm": 0.0876444442901716,
      "learning_rate": 3.237551971264202e-07,
      "loss": 0.0024,
      "step": 54780
    },
    {
      "epoch": 0.8965065859445308,
      "grad_norm": 0.18767107501538152,
      "learning_rate": 3.22745020156513e-07,
      "loss": 0.0017,
      "step": 54790
    },
    {
      "epoch": 0.8966702118956067,
      "grad_norm": 0.19251069475252744,
      "learning_rate": 3.2173636905832693e-07,
      "loss": 0.0022,
      "step": 54800
    },
    {
      "epoch": 0.8968338378466825,
      "grad_norm": 0.14086856538327036,
      "learning_rate": 3.207292441609161e-07,
      "loss": 0.0019,
      "step": 54810
    },
    {
      "epoch": 0.8969974637977584,
      "grad_norm": 0.05757627548559175,
      "learning_rate": 3.197236457928382e-07,
      "loss": 0.0016,
      "step": 54820
    },
    {
      "epoch": 0.8971610897488341,
      "grad_norm": 0.030941850603707167,
      "learning_rate": 3.1871957428215083e-07,
      "loss": 0.0016,
      "step": 54830
    },
    {
      "epoch": 0.89732471569991,
      "grad_norm": 0.1727152217216821,
      "learning_rate": 3.1771702995641474e-07,
      "loss": 0.0018,
      "step": 54840
    },
    {
      "epoch": 0.8974883416509859,
      "grad_norm": 0.21024436962336948,
      "learning_rate": 3.1671601314269385e-07,
      "loss": 0.0032,
      "step": 54850
    },
    {
      "epoch": 0.8976519676020617,
      "grad_norm": 0.14867534929409998,
      "learning_rate": 3.157165241675514e-07,
      "loss": 0.0035,
      "step": 54860
    },
    {
      "epoch": 0.8978155935531376,
      "grad_norm": 0.04390733531272599,
      "learning_rate": 3.14718563357051e-07,
      "loss": 0.0016,
      "step": 54870
    },
    {
      "epoch": 0.8979792195042133,
      "grad_norm": 0.21032225030609492,
      "learning_rate": 3.1372213103676216e-07,
      "loss": 0.0036,
      "step": 54880
    },
    {
      "epoch": 0.8981428454552892,
      "grad_norm": 0.2205845380768619,
      "learning_rate": 3.127272275317522e-07,
      "loss": 0.0015,
      "step": 54890
    },
    {
      "epoch": 0.8983064714063651,
      "grad_norm": 0.10830606000662713,
      "learning_rate": 3.117338531665909e-07,
      "loss": 0.0014,
      "step": 54900
    },
    {
      "epoch": 0.8984700973574409,
      "grad_norm": 0.05644491812245202,
      "learning_rate": 3.107420082653495e-07,
      "loss": 0.0017,
      "step": 54910
    },
    {
      "epoch": 0.8986337233085168,
      "grad_norm": 0.10431068297762724,
      "learning_rate": 3.0975169315160093e-07,
      "loss": 0.0015,
      "step": 54920
    },
    {
      "epoch": 0.8987973492595925,
      "grad_norm": 0.16705474976151624,
      "learning_rate": 3.0876290814841503e-07,
      "loss": 0.0031,
      "step": 54930
    },
    {
      "epoch": 0.8989609752106684,
      "grad_norm": 0.37901055382188303,
      "learning_rate": 3.077756535783677e-07,
      "loss": 0.0024,
      "step": 54940
    },
    {
      "epoch": 0.8991246011617443,
      "grad_norm": 0.058447289944995116,
      "learning_rate": 3.067899297635324e-07,
      "loss": 0.0029,
      "step": 54950
    },
    {
      "epoch": 0.8992882271128201,
      "grad_norm": 0.14604681219932686,
      "learning_rate": 3.0580573702548357e-07,
      "loss": 0.0009,
      "step": 54960
    },
    {
      "epoch": 0.899451853063896,
      "grad_norm": 0.2025788601108559,
      "learning_rate": 3.0482307568529877e-07,
      "loss": 0.0023,
      "step": 54970
    },
    {
      "epoch": 0.8996154790149717,
      "grad_norm": 0.174359620200919,
      "learning_rate": 3.038419460635528e-07,
      "loss": 0.0024,
      "step": 54980
    },
    {
      "epoch": 0.8997791049660476,
      "grad_norm": 0.04537432838368517,
      "learning_rate": 3.028623484803206e-07,
      "loss": 0.0027,
      "step": 54990
    },
    {
      "epoch": 0.8999427309171234,
      "grad_norm": 0.0345357890341505,
      "learning_rate": 3.018842832551799e-07,
      "loss": 0.0018,
      "step": 55000
    },
    {
      "epoch": 0.9001063568681993,
      "grad_norm": 0.09801966779350116,
      "learning_rate": 3.0090775070720724e-07,
      "loss": 0.0025,
      "step": 55010
    },
    {
      "epoch": 0.9002699828192752,
      "grad_norm": 0.07649213164184805,
      "learning_rate": 2.999327511549782e-07,
      "loss": 0.0038,
      "step": 55020
    },
    {
      "epoch": 0.900433608770351,
      "grad_norm": 0.1606547782899063,
      "learning_rate": 2.9895928491656903e-07,
      "loss": 0.0019,
      "step": 55030
    },
    {
      "epoch": 0.9005972347214268,
      "grad_norm": 0.08315474526242514,
      "learning_rate": 2.979873523095583e-07,
      "loss": 0.0017,
      "step": 55040
    },
    {
      "epoch": 0.9007608606725026,
      "grad_norm": 0.1619305372853114,
      "learning_rate": 2.9701695365101926e-07,
      "loss": 0.0022,
      "step": 55050
    },
    {
      "epoch": 0.9009244866235785,
      "grad_norm": 0.21429890882802613,
      "learning_rate": 2.960480892575268e-07,
      "loss": 0.0026,
      "step": 55060
    },
    {
      "epoch": 0.9010881125746544,
      "grad_norm": 0.10352649985733839,
      "learning_rate": 2.9508075944515824e-07,
      "loss": 0.0019,
      "step": 55070
    },
    {
      "epoch": 0.9012517385257302,
      "grad_norm": 0.015935293654937756,
      "learning_rate": 2.9411496452948663e-07,
      "loss": 0.0048,
      "step": 55080
    },
    {
      "epoch": 0.901415364476806,
      "grad_norm": 0.15620151700533172,
      "learning_rate": 2.931507048255849e-07,
      "loss": 0.0019,
      "step": 55090
    },
    {
      "epoch": 0.9015789904278818,
      "grad_norm": 0.0705950916497526,
      "learning_rate": 2.921879806480277e-07,
      "loss": 0.002,
      "step": 55100
    },
    {
      "epoch": 0.9017426163789577,
      "grad_norm": 0.10795043150059214,
      "learning_rate": 2.912267923108847e-07,
      "loss": 0.0023,
      "step": 55110
    },
    {
      "epoch": 0.9019062423300336,
      "grad_norm": 0.17197534170231377,
      "learning_rate": 2.902671401277268e-07,
      "loss": 0.0024,
      "step": 55120
    },
    {
      "epoch": 0.9020698682811094,
      "grad_norm": 0.1261312565539737,
      "learning_rate": 2.8930902441162423e-07,
      "loss": 0.0021,
      "step": 55130
    },
    {
      "epoch": 0.9022334942321852,
      "grad_norm": 0.13054427755160755,
      "learning_rate": 2.883524454751457e-07,
      "loss": 0.0017,
      "step": 55140
    },
    {
      "epoch": 0.902397120183261,
      "grad_norm": 0.11592971792779985,
      "learning_rate": 2.873974036303562e-07,
      "loss": 0.0016,
      "step": 55150
    },
    {
      "epoch": 0.9025607461343369,
      "grad_norm": 0.1159608936726887,
      "learning_rate": 2.8644389918882463e-07,
      "loss": 0.0023,
      "step": 55160
    },
    {
      "epoch": 0.9027243720854128,
      "grad_norm": 0.15366642328978197,
      "learning_rate": 2.8549193246161233e-07,
      "loss": 0.0018,
      "step": 55170
    },
    {
      "epoch": 0.9028879980364886,
      "grad_norm": 0.06744822157575542,
      "learning_rate": 2.845415037592808e-07,
      "loss": 0.0017,
      "step": 55180
    },
    {
      "epoch": 0.9030516239875644,
      "grad_norm": 0.020999347694959196,
      "learning_rate": 2.835926133918937e-07,
      "loss": 0.0022,
      "step": 55190
    },
    {
      "epoch": 0.9032152499386402,
      "grad_norm": 0.11219823826524569,
      "learning_rate": 2.826452616690073e-07,
      "loss": 0.0021,
      "step": 55200
    },
    {
      "epoch": 0.9033788758897161,
      "grad_norm": 0.04788470037606798,
      "learning_rate": 2.816994488996788e-07,
      "loss": 0.0015,
      "step": 55210
    },
    {
      "epoch": 0.903542501840792,
      "grad_norm": 0.14119615719003709,
      "learning_rate": 2.807551753924642e-07,
      "loss": 0.002,
      "step": 55220
    },
    {
      "epoch": 0.9037061277918678,
      "grad_norm": 0.45189869322670817,
      "learning_rate": 2.7981244145541486e-07,
      "loss": 0.0032,
      "step": 55230
    },
    {
      "epoch": 0.9038697537429436,
      "grad_norm": 0.06539300542807684,
      "learning_rate": 2.788712473960803e-07,
      "loss": 0.0014,
      "step": 55240
    },
    {
      "epoch": 0.9040333796940194,
      "grad_norm": 0.08359094918823193,
      "learning_rate": 2.779315935215099e-07,
      "loss": 0.0029,
      "step": 55250
    },
    {
      "epoch": 0.9041970056450953,
      "grad_norm": 0.12224832662660219,
      "learning_rate": 2.7699348013824903e-07,
      "loss": 0.0026,
      "step": 55260
    },
    {
      "epoch": 0.9043606315961712,
      "grad_norm": 0.26172504280552056,
      "learning_rate": 2.7605690755233896e-07,
      "loss": 0.0024,
      "step": 55270
    },
    {
      "epoch": 0.904524257547247,
      "grad_norm": 0.09217145101803007,
      "learning_rate": 2.7512187606932305e-07,
      "loss": 0.0011,
      "step": 55280
    },
    {
      "epoch": 0.9046878834983229,
      "grad_norm": 0.026259947656491672,
      "learning_rate": 2.741883859942357e-07,
      "loss": 0.0011,
      "step": 55290
    },
    {
      "epoch": 0.9048515094493986,
      "grad_norm": 0.13618765604554886,
      "learning_rate": 2.7325643763161213e-07,
      "loss": 0.0012,
      "step": 55300
    },
    {
      "epoch": 0.9050151354004745,
      "grad_norm": 0.07239486736134429,
      "learning_rate": 2.723260312854853e-07,
      "loss": 0.0026,
      "step": 55310
    },
    {
      "epoch": 0.9051787613515504,
      "grad_norm": 0.005495350910934415,
      "learning_rate": 2.7139716725938293e-07,
      "loss": 0.0028,
      "step": 55320
    },
    {
      "epoch": 0.9053423873026262,
      "grad_norm": 0.2488829957549035,
      "learning_rate": 2.704698458563304e-07,
      "loss": 0.0021,
      "step": 55330
    },
    {
      "epoch": 0.9055060132537021,
      "grad_norm": 0.21766250197040005,
      "learning_rate": 2.695440673788513e-07,
      "loss": 0.0025,
      "step": 55340
    },
    {
      "epoch": 0.9056696392047778,
      "grad_norm": 0.10953824545046612,
      "learning_rate": 2.6861983212896236e-07,
      "loss": 0.0014,
      "step": 55350
    },
    {
      "epoch": 0.9058332651558537,
      "grad_norm": 0.031958995251045834,
      "learning_rate": 2.6769714040818073e-07,
      "loss": 0.001,
      "step": 55360
    },
    {
      "epoch": 0.9059968911069296,
      "grad_norm": 0.024051180391602693,
      "learning_rate": 2.667759925175162e-07,
      "loss": 0.001,
      "step": 55370
    },
    {
      "epoch": 0.9061605170580054,
      "grad_norm": 0.08799517109715474,
      "learning_rate": 2.6585638875747885e-07,
      "loss": 0.0026,
      "step": 55380
    },
    {
      "epoch": 0.9063241430090813,
      "grad_norm": 0.37663692924875547,
      "learning_rate": 2.649383294280733e-07,
      "loss": 0.0018,
      "step": 55390
    },
    {
      "epoch": 0.906487768960157,
      "grad_norm": 0.1733224251920871,
      "learning_rate": 2.6402181482879873e-07,
      "loss": 0.0026,
      "step": 55400
    },
    {
      "epoch": 0.9066513949112329,
      "grad_norm": 0.033847806687230954,
      "learning_rate": 2.631068452586533e-07,
      "loss": 0.0013,
      "step": 55410
    },
    {
      "epoch": 0.9068150208623088,
      "grad_norm": 0.07988489999548476,
      "learning_rate": 2.621934210161281e-07,
      "loss": 0.0009,
      "step": 55420
    },
    {
      "epoch": 0.9069786468133846,
      "grad_norm": 0.033828729523273586,
      "learning_rate": 2.612815423992121e-07,
      "loss": 0.0041,
      "step": 55430
    },
    {
      "epoch": 0.9071422727644605,
      "grad_norm": 0.14756675761151994,
      "learning_rate": 2.6037120970539065e-07,
      "loss": 0.0026,
      "step": 55440
    },
    {
      "epoch": 0.9073058987155362,
      "grad_norm": 0.012179204387660587,
      "learning_rate": 2.5946242323164336e-07,
      "loss": 0.0021,
      "step": 55450
    },
    {
      "epoch": 0.9074695246666121,
      "grad_norm": 0.05349848219179725,
      "learning_rate": 2.585551832744448e-07,
      "loss": 0.0013,
      "step": 55460
    },
    {
      "epoch": 0.907633150617688,
      "grad_norm": 0.1337655768884975,
      "learning_rate": 2.576494901297666e-07,
      "loss": 0.0015,
      "step": 55470
    },
    {
      "epoch": 0.9077967765687638,
      "grad_norm": 0.09731400949961659,
      "learning_rate": 2.5674534409307507e-07,
      "loss": 0.0016,
      "step": 55480
    },
    {
      "epoch": 0.9079604025198397,
      "grad_norm": 0.09452974825624491,
      "learning_rate": 2.55842745459331e-07,
      "loss": 0.0012,
      "step": 55490
    },
    {
      "epoch": 0.9081240284709154,
      "grad_norm": 0.021427705701072268,
      "learning_rate": 2.549416945229932e-07,
      "loss": 0.0024,
      "step": 55500
    },
    {
      "epoch": 0.9082876544219913,
      "grad_norm": 0.08505538446844334,
      "learning_rate": 2.5404219157801224e-07,
      "loss": 0.0033,
      "step": 55510
    },
    {
      "epoch": 0.9084512803730672,
      "grad_norm": 0.19727355048470685,
      "learning_rate": 2.5314423691783496e-07,
      "loss": 0.0015,
      "step": 55520
    },
    {
      "epoch": 0.908614906324143,
      "grad_norm": 0.20772209724049945,
      "learning_rate": 2.5224783083540373e-07,
      "loss": 0.0026,
      "step": 55530
    },
    {
      "epoch": 0.9087785322752189,
      "grad_norm": 0.24227787227924222,
      "learning_rate": 2.513529736231551e-07,
      "loss": 0.0029,
      "step": 55540
    },
    {
      "epoch": 0.9089421582262946,
      "grad_norm": 0.0835086290877973,
      "learning_rate": 2.5045966557301957e-07,
      "loss": 0.0018,
      "step": 55550
    },
    {
      "epoch": 0.9091057841773705,
      "grad_norm": 0.04785476658194691,
      "learning_rate": 2.4956790697642396e-07,
      "loss": 0.002,
      "step": 55560
    },
    {
      "epoch": 0.9092694101284464,
      "grad_norm": 0.07602260406299748,
      "learning_rate": 2.486776981242889e-07,
      "loss": 0.0012,
      "step": 55570
    },
    {
      "epoch": 0.9094330360795222,
      "grad_norm": 0.19717844371832224,
      "learning_rate": 2.4778903930702925e-07,
      "loss": 0.0019,
      "step": 55580
    },
    {
      "epoch": 0.9095966620305981,
      "grad_norm": 0.03427350533854744,
      "learning_rate": 2.469019308145537e-07,
      "loss": 0.0015,
      "step": 55590
    },
    {
      "epoch": 0.9097602879816739,
      "grad_norm": 0.19813913703631816,
      "learning_rate": 2.460163729362658e-07,
      "loss": 0.0028,
      "step": 55600
    },
    {
      "epoch": 0.9099239139327497,
      "grad_norm": 0.04017201311209813,
      "learning_rate": 2.451323659610627e-07,
      "loss": 0.0019,
      "step": 55610
    },
    {
      "epoch": 0.9100875398838256,
      "grad_norm": 0.05326045818163813,
      "learning_rate": 2.442499101773377e-07,
      "loss": 0.0015,
      "step": 55620
    },
    {
      "epoch": 0.9102511658349014,
      "grad_norm": 0.008843341391014631,
      "learning_rate": 2.4336900587297485e-07,
      "loss": 0.0018,
      "step": 55630
    },
    {
      "epoch": 0.9104147917859773,
      "grad_norm": 0.07799195820569942,
      "learning_rate": 2.424896533353538e-07,
      "loss": 0.0013,
      "step": 55640
    },
    {
      "epoch": 0.9105784177370531,
      "grad_norm": 0.053930016010225734,
      "learning_rate": 2.416118528513489e-07,
      "loss": 0.0019,
      "step": 55650
    },
    {
      "epoch": 0.9107420436881289,
      "grad_norm": 0.03646383683449795,
      "learning_rate": 2.407356047073256e-07,
      "loss": 0.0022,
      "step": 55660
    },
    {
      "epoch": 0.9109056696392048,
      "grad_norm": 0.1308189607527901,
      "learning_rate": 2.3986090918914515e-07,
      "loss": 0.0019,
      "step": 55670
    },
    {
      "epoch": 0.9110692955902806,
      "grad_norm": 0.05017846308077851,
      "learning_rate": 2.389877665821605e-07,
      "loss": 0.0018,
      "step": 55680
    },
    {
      "epoch": 0.9112329215413565,
      "grad_norm": 0.12846589249825804,
      "learning_rate": 2.3811617717122037e-07,
      "loss": 0.0033,
      "step": 55690
    },
    {
      "epoch": 0.9113965474924323,
      "grad_norm": 0.017885047868760363,
      "learning_rate": 2.3724614124066457e-07,
      "loss": 0.0009,
      "step": 55700
    },
    {
      "epoch": 0.9115601734435081,
      "grad_norm": 0.058179899984852686,
      "learning_rate": 2.3637765907432776e-07,
      "loss": 0.0012,
      "step": 55710
    },
    {
      "epoch": 0.911723799394584,
      "grad_norm": 0.08149246882345132,
      "learning_rate": 2.35510730955536e-07,
      "loss": 0.001,
      "step": 55720
    },
    {
      "epoch": 0.9118874253456598,
      "grad_norm": 0.15611638239376435,
      "learning_rate": 2.3464535716710924e-07,
      "loss": 0.0013,
      "step": 55730
    },
    {
      "epoch": 0.9120510512967357,
      "grad_norm": 0.15634849274687765,
      "learning_rate": 2.337815379913605e-07,
      "loss": 0.0022,
      "step": 55740
    },
    {
      "epoch": 0.9122146772478115,
      "grad_norm": 0.07734116577605984,
      "learning_rate": 2.3291927371009605e-07,
      "loss": 0.0014,
      "step": 55750
    },
    {
      "epoch": 0.9123783031988874,
      "grad_norm": 0.06718425911709938,
      "learning_rate": 2.3205856460461363e-07,
      "loss": 0.002,
      "step": 55760
    },
    {
      "epoch": 0.9125419291499632,
      "grad_norm": 0.02078459824712382,
      "learning_rate": 2.311994109557053e-07,
      "loss": 0.0011,
      "step": 55770
    },
    {
      "epoch": 0.912705555101039,
      "grad_norm": 0.15015139647321193,
      "learning_rate": 2.3034181304365411e-07,
      "loss": 0.0017,
      "step": 55780
    },
    {
      "epoch": 0.9128691810521149,
      "grad_norm": 0.034504031704151936,
      "learning_rate": 2.2948577114823623e-07,
      "loss": 0.0012,
      "step": 55790
    },
    {
      "epoch": 0.9130328070031907,
      "grad_norm": 0.18892323840720948,
      "learning_rate": 2.2863128554871938e-07,
      "loss": 0.0021,
      "step": 55800
    },
    {
      "epoch": 0.9131964329542666,
      "grad_norm": 0.05855581031692049,
      "learning_rate": 2.2777835652386615e-07,
      "loss": 0.0025,
      "step": 55810
    },
    {
      "epoch": 0.9133600589053424,
      "grad_norm": 0.06081122465622154,
      "learning_rate": 2.269269843519284e-07,
      "loss": 0.0025,
      "step": 55820
    },
    {
      "epoch": 0.9135236848564182,
      "grad_norm": 0.19891858434444634,
      "learning_rate": 2.2607716931065171e-07,
      "loss": 0.0019,
      "step": 55830
    },
    {
      "epoch": 0.9136873108074941,
      "grad_norm": 0.12957716396190783,
      "learning_rate": 2.2522891167727267e-07,
      "loss": 0.0018,
      "step": 55840
    },
    {
      "epoch": 0.9138509367585699,
      "grad_norm": 0.2873020674713747,
      "learning_rate": 2.2438221172852105e-07,
      "loss": 0.0033,
      "step": 55850
    },
    {
      "epoch": 0.9140145627096458,
      "grad_norm": 0.18801450198741834,
      "learning_rate": 2.235370697406164e-07,
      "loss": 0.0024,
      "step": 55860
    },
    {
      "epoch": 0.9141781886607215,
      "grad_norm": 0.655952568273636,
      "learning_rate": 2.2269348598927266e-07,
      "loss": 0.0018,
      "step": 55870
    },
    {
      "epoch": 0.9143418146117974,
      "grad_norm": 0.10980673319771161,
      "learning_rate": 2.218514607496941e-07,
      "loss": 0.002,
      "step": 55880
    },
    {
      "epoch": 0.9145054405628733,
      "grad_norm": 0.16711795285296266,
      "learning_rate": 2.21010994296576e-07,
      "loss": 0.0019,
      "step": 55890
    },
    {
      "epoch": 0.9146690665139491,
      "grad_norm": 0.10836437644539305,
      "learning_rate": 2.201720869041063e-07,
      "loss": 0.0017,
      "step": 55900
    },
    {
      "epoch": 0.914832692465025,
      "grad_norm": 0.031611988333894246,
      "learning_rate": 2.1933473884596267e-07,
      "loss": 0.0027,
      "step": 55910
    },
    {
      "epoch": 0.9149963184161007,
      "grad_norm": 0.13354100904584598,
      "learning_rate": 2.1849895039531555e-07,
      "loss": 0.0017,
      "step": 55920
    },
    {
      "epoch": 0.9151599443671766,
      "grad_norm": 0.1121473421304103,
      "learning_rate": 2.1766472182482624e-07,
      "loss": 0.0013,
      "step": 55930
    },
    {
      "epoch": 0.9153235703182525,
      "grad_norm": 0.07901031108362461,
      "learning_rate": 2.168320534066476e-07,
      "loss": 0.001,
      "step": 55940
    },
    {
      "epoch": 0.9154871962693283,
      "grad_norm": 0.12346570943725273,
      "learning_rate": 2.1600094541242234e-07,
      "loss": 0.0021,
      "step": 55950
    },
    {
      "epoch": 0.9156508222204042,
      "grad_norm": 0.07418633428329167,
      "learning_rate": 2.1517139811328523e-07,
      "loss": 0.0033,
      "step": 55960
    },
    {
      "epoch": 0.91581444817148,
      "grad_norm": 0.29340279984858275,
      "learning_rate": 2.1434341177986094e-07,
      "loss": 0.0026,
      "step": 55970
    },
    {
      "epoch": 0.9159780741225558,
      "grad_norm": 0.17036844895899028,
      "learning_rate": 2.135169866822645e-07,
      "loss": 0.0034,
      "step": 55980
    },
    {
      "epoch": 0.9161417000736317,
      "grad_norm": 0.12099077881816002,
      "learning_rate": 2.1269212309010411e-07,
      "loss": 0.0029,
      "step": 55990
    },
    {
      "epoch": 0.9163053260247075,
      "grad_norm": 0.05223029540291728,
      "learning_rate": 2.1186882127247678e-07,
      "loss": 0.0014,
      "step": 56000
    },
    {
      "epoch": 0.9164689519757834,
      "grad_norm": 0.31384412561880726,
      "learning_rate": 2.1104708149796982e-07,
      "loss": 0.003,
      "step": 56010
    },
    {
      "epoch": 0.9166325779268591,
      "grad_norm": 0.0747577346650278,
      "learning_rate": 2.1022690403465994e-07,
      "loss": 0.0028,
      "step": 56020
    },
    {
      "epoch": 0.916796203877935,
      "grad_norm": 0.0721854685751019,
      "learning_rate": 2.0940828915011757e-07,
      "loss": 0.0013,
      "step": 56030
    },
    {
      "epoch": 0.9169598298290109,
      "grad_norm": 0.11573605420316137,
      "learning_rate": 2.0859123711140072e-07,
      "loss": 0.0038,
      "step": 56040
    },
    {
      "epoch": 0.9171234557800867,
      "grad_norm": 0.14292415946337345,
      "learning_rate": 2.077757481850573e-07,
      "loss": 0.0022,
      "step": 56050
    },
    {
      "epoch": 0.9172870817311626,
      "grad_norm": 0.1182264424909099,
      "learning_rate": 2.0696182263712726e-07,
      "loss": 0.0024,
      "step": 56060
    },
    {
      "epoch": 0.9174507076822384,
      "grad_norm": 0.12415764963071088,
      "learning_rate": 2.0614946073313934e-07,
      "loss": 0.0025,
      "step": 56070
    },
    {
      "epoch": 0.9176143336333142,
      "grad_norm": 0.2281540260842942,
      "learning_rate": 2.0533866273811097e-07,
      "loss": 0.0017,
      "step": 56080
    },
    {
      "epoch": 0.9177779595843901,
      "grad_norm": 0.19525872256303528,
      "learning_rate": 2.0452942891655226e-07,
      "loss": 0.0021,
      "step": 56090
    },
    {
      "epoch": 0.9179415855354659,
      "grad_norm": 0.11312533920422914,
      "learning_rate": 2.0372175953246032e-07,
      "loss": 0.0023,
      "step": 56100
    },
    {
      "epoch": 0.9181052114865418,
      "grad_norm": 0.05861596001143664,
      "learning_rate": 2.029156548493233e-07,
      "loss": 0.003,
      "step": 56110
    },
    {
      "epoch": 0.9182688374376176,
      "grad_norm": 0.1453170418948343,
      "learning_rate": 2.0211111513011916e-07,
      "loss": 0.0018,
      "step": 56120
    },
    {
      "epoch": 0.9184324633886934,
      "grad_norm": 0.31667349994033256,
      "learning_rate": 2.0130814063731463e-07,
      "loss": 0.003,
      "step": 56130
    },
    {
      "epoch": 0.9185960893397693,
      "grad_norm": 0.04255089892916259,
      "learning_rate": 2.005067316328646e-07,
      "loss": 0.0016,
      "step": 56140
    },
    {
      "epoch": 0.9187597152908451,
      "grad_norm": 0.1025138155260419,
      "learning_rate": 1.9970688837821662e-07,
      "loss": 0.001,
      "step": 56150
    },
    {
      "epoch": 0.918923341241921,
      "grad_norm": 0.1664173343061674,
      "learning_rate": 1.989086111343036e-07,
      "loss": 0.0013,
      "step": 56160
    },
    {
      "epoch": 0.9190869671929968,
      "grad_norm": 0.0966479358239818,
      "learning_rate": 1.9811190016155002e-07,
      "loss": 0.0013,
      "step": 56170
    },
    {
      "epoch": 0.9192505931440726,
      "grad_norm": 0.16063982165806956,
      "learning_rate": 1.973167557198702e-07,
      "loss": 0.0024,
      "step": 56180
    },
    {
      "epoch": 0.9194142190951485,
      "grad_norm": 0.041822309286628036,
      "learning_rate": 1.9652317806866383e-07,
      "loss": 0.0013,
      "step": 56190
    },
    {
      "epoch": 0.9195778450462243,
      "grad_norm": 0.07067927088531724,
      "learning_rate": 1.957311674668222e-07,
      "loss": 0.0014,
      "step": 56200
    },
    {
      "epoch": 0.9197414709973002,
      "grad_norm": 0.032332041213308044,
      "learning_rate": 1.9494072417272524e-07,
      "loss": 0.0023,
      "step": 56210
    },
    {
      "epoch": 0.919905096948376,
      "grad_norm": 0.11884091322439831,
      "learning_rate": 1.9415184844424063e-07,
      "loss": 0.0026,
      "step": 56220
    },
    {
      "epoch": 0.9200687228994519,
      "grad_norm": 0.2499504673842196,
      "learning_rate": 1.9336454053872522e-07,
      "loss": 0.002,
      "step": 56230
    },
    {
      "epoch": 0.9202323488505277,
      "grad_norm": 0.05339050055431217,
      "learning_rate": 1.9257880071302525e-07,
      "loss": 0.0014,
      "step": 56240
    },
    {
      "epoch": 0.9203959748016035,
      "grad_norm": 0.13663361761044288,
      "learning_rate": 1.9179462922347292e-07,
      "loss": 0.0017,
      "step": 56250
    },
    {
      "epoch": 0.9205596007526794,
      "grad_norm": 0.07428785140305373,
      "learning_rate": 1.9101202632589077e-07,
      "loss": 0.0012,
      "step": 56260
    },
    {
      "epoch": 0.9207232267037552,
      "grad_norm": 0.3614251728057188,
      "learning_rate": 1.902309922755896e-07,
      "loss": 0.0027,
      "step": 56270
    },
    {
      "epoch": 0.9208868526548311,
      "grad_norm": 0.11004066929447891,
      "learning_rate": 1.8945152732736783e-07,
      "loss": 0.0017,
      "step": 56280
    },
    {
      "epoch": 0.9210504786059069,
      "grad_norm": 0.09062584213277752,
      "learning_rate": 1.886736317355109e-07,
      "loss": 0.002,
      "step": 56290
    },
    {
      "epoch": 0.9212141045569827,
      "grad_norm": 0.17483758213472744,
      "learning_rate": 1.8789730575379638e-07,
      "loss": 0.0019,
      "step": 56300
    },
    {
      "epoch": 0.9213777305080586,
      "grad_norm": 0.09132705281069708,
      "learning_rate": 1.871225496354845e-07,
      "loss": 0.0015,
      "step": 56310
    },
    {
      "epoch": 0.9215413564591344,
      "grad_norm": 0.15235146319019063,
      "learning_rate": 1.8634936363332523e-07,
      "loss": 0.0026,
      "step": 56320
    },
    {
      "epoch": 0.9217049824102103,
      "grad_norm": 0.07322562758897258,
      "learning_rate": 1.8557774799955907e-07,
      "loss": 0.0011,
      "step": 56330
    },
    {
      "epoch": 0.9218686083612861,
      "grad_norm": 0.30773583146583877,
      "learning_rate": 1.8480770298591078e-07,
      "loss": 0.002,
      "step": 56340
    },
    {
      "epoch": 0.9220322343123619,
      "grad_norm": 0.16329344491213882,
      "learning_rate": 1.840392288435938e-07,
      "loss": 0.0014,
      "step": 56350
    },
    {
      "epoch": 0.9221958602634378,
      "grad_norm": 0.3489466202091755,
      "learning_rate": 1.8327232582330934e-07,
      "loss": 0.0021,
      "step": 56360
    },
    {
      "epoch": 0.9223594862145136,
      "grad_norm": 0.08177529706484211,
      "learning_rate": 1.8250699417524666e-07,
      "loss": 0.0028,
      "step": 56370
    },
    {
      "epoch": 0.9225231121655895,
      "grad_norm": 0.18051426226661335,
      "learning_rate": 1.8174323414908046e-07,
      "loss": 0.0017,
      "step": 56380
    },
    {
      "epoch": 0.9226867381166654,
      "grad_norm": 0.15806451601488578,
      "learning_rate": 1.809810459939737e-07,
      "loss": 0.0021,
      "step": 56390
    },
    {
      "epoch": 0.9228503640677411,
      "grad_norm": 0.0872565287801616,
      "learning_rate": 1.8022042995857859e-07,
      "loss": 0.0017,
      "step": 56400
    },
    {
      "epoch": 0.923013990018817,
      "grad_norm": 0.12068942960599625,
      "learning_rate": 1.794613862910316e-07,
      "loss": 0.0012,
      "step": 56410
    },
    {
      "epoch": 0.9231776159698928,
      "grad_norm": 0.06756677647445036,
      "learning_rate": 1.7870391523895746e-07,
      "loss": 0.0013,
      "step": 56420
    },
    {
      "epoch": 0.9233412419209687,
      "grad_norm": 0.13204465865906662,
      "learning_rate": 1.7794801704946795e-07,
      "loss": 0.0012,
      "step": 56430
    },
    {
      "epoch": 0.9235048678720446,
      "grad_norm": 0.2603801495279046,
      "learning_rate": 1.7719369196916081e-07,
      "loss": 0.0027,
      "step": 56440
    },
    {
      "epoch": 0.9236684938231203,
      "grad_norm": 0.1140901751491441,
      "learning_rate": 1.7644094024412083e-07,
      "loss": 0.0014,
      "step": 56450
    },
    {
      "epoch": 0.9238321197741962,
      "grad_norm": 0.12178043751776504,
      "learning_rate": 1.7568976211992162e-07,
      "loss": 0.0014,
      "step": 56460
    },
    {
      "epoch": 0.923995745725272,
      "grad_norm": 0.051216740076593374,
      "learning_rate": 1.7494015784162045e-07,
      "loss": 0.0018,
      "step": 56470
    },
    {
      "epoch": 0.9241593716763479,
      "grad_norm": 0.13807739409114403,
      "learning_rate": 1.7419212765376336e-07,
      "loss": 0.003,
      "step": 56480
    },
    {
      "epoch": 0.9243229976274238,
      "grad_norm": 0.15391130662494193,
      "learning_rate": 1.734456718003813e-07,
      "loss": 0.0027,
      "step": 56490
    },
    {
      "epoch": 0.9244866235784995,
      "grad_norm": 0.03434975143523334,
      "learning_rate": 1.7270079052499278e-07,
      "loss": 0.0018,
      "step": 56500
    },
    {
      "epoch": 0.9246502495295754,
      "grad_norm": 0.00658772233653941,
      "learning_rate": 1.7195748407060064e-07,
      "loss": 0.0012,
      "step": 56510
    },
    {
      "epoch": 0.9248138754806512,
      "grad_norm": 0.2947282495327704,
      "learning_rate": 1.7121575267969758e-07,
      "loss": 0.0016,
      "step": 56520
    },
    {
      "epoch": 0.9249775014317271,
      "grad_norm": 0.14051321186154042,
      "learning_rate": 1.7047559659425948e-07,
      "loss": 0.0022,
      "step": 56530
    },
    {
      "epoch": 0.925141127382803,
      "grad_norm": 0.17219877969543573,
      "learning_rate": 1.6973701605574934e-07,
      "loss": 0.0015,
      "step": 56540
    },
    {
      "epoch": 0.9253047533338787,
      "grad_norm": 0.05393988161970387,
      "learning_rate": 1.6900001130511546e-07,
      "loss": 0.0022,
      "step": 56550
    },
    {
      "epoch": 0.9254683792849546,
      "grad_norm": 0.08722251094028792,
      "learning_rate": 1.6826458258279331e-07,
      "loss": 0.0013,
      "step": 56560
    },
    {
      "epoch": 0.9256320052360304,
      "grad_norm": 0.17524374493945982,
      "learning_rate": 1.6753073012870212e-07,
      "loss": 0.002,
      "step": 56570
    },
    {
      "epoch": 0.9257956311871063,
      "grad_norm": 0.11825250771653613,
      "learning_rate": 1.6679845418225038e-07,
      "loss": 0.0024,
      "step": 56580
    },
    {
      "epoch": 0.9259592571381822,
      "grad_norm": 0.1937679049022653,
      "learning_rate": 1.6606775498232863e-07,
      "loss": 0.0019,
      "step": 56590
    },
    {
      "epoch": 0.9261228830892579,
      "grad_norm": 0.3087846680899814,
      "learning_rate": 1.653386327673151e-07,
      "loss": 0.0027,
      "step": 56600
    },
    {
      "epoch": 0.9262865090403338,
      "grad_norm": 0.06636982167504778,
      "learning_rate": 1.6461108777507285e-07,
      "loss": 0.0022,
      "step": 56610
    },
    {
      "epoch": 0.9264501349914096,
      "grad_norm": 0.11071169781810984,
      "learning_rate": 1.638851202429509e-07,
      "loss": 0.002,
      "step": 56620
    },
    {
      "epoch": 0.9266137609424855,
      "grad_norm": 0.08030151167553365,
      "learning_rate": 1.6316073040778203e-07,
      "loss": 0.0014,
      "step": 56630
    },
    {
      "epoch": 0.9267773868935614,
      "grad_norm": 0.01085950884936691,
      "learning_rate": 1.6243791850588774e-07,
      "loss": 0.0027,
      "step": 56640
    },
    {
      "epoch": 0.9269410128446371,
      "grad_norm": 0.15374391783959676,
      "learning_rate": 1.6171668477307113e-07,
      "loss": 0.0025,
      "step": 56650
    },
    {
      "epoch": 0.927104638795713,
      "grad_norm": 0.14697337043700706,
      "learning_rate": 1.609970294446228e-07,
      "loss": 0.0016,
      "step": 56660
    },
    {
      "epoch": 0.9272682647467888,
      "grad_norm": 0.0920212191121231,
      "learning_rate": 1.6027895275531668e-07,
      "loss": 0.0015,
      "step": 56670
    },
    {
      "epoch": 0.9274318906978647,
      "grad_norm": 0.15788819854947958,
      "learning_rate": 1.5956245493941313e-07,
      "loss": 0.0016,
      "step": 56680
    },
    {
      "epoch": 0.9275955166489406,
      "grad_norm": 0.037872452092925085,
      "learning_rate": 1.5884753623065684e-07,
      "loss": 0.001,
      "step": 56690
    },
    {
      "epoch": 0.9277591426000164,
      "grad_norm": 0.1098913508461192,
      "learning_rate": 1.5813419686227682e-07,
      "loss": 0.0017,
      "step": 56700
    },
    {
      "epoch": 0.9279227685510922,
      "grad_norm": 0.10799060137985783,
      "learning_rate": 1.5742243706698857e-07,
      "loss": 0.0018,
      "step": 56710
    },
    {
      "epoch": 0.928086394502168,
      "grad_norm": 0.17065968094713618,
      "learning_rate": 1.5671225707699078e-07,
      "loss": 0.0024,
      "step": 56720
    },
    {
      "epoch": 0.9282500204532439,
      "grad_norm": 0.08845286044398096,
      "learning_rate": 1.5600365712396647e-07,
      "loss": 0.0031,
      "step": 56730
    },
    {
      "epoch": 0.9284136464043197,
      "grad_norm": 0.18631657208935526,
      "learning_rate": 1.5529663743908406e-07,
      "loss": 0.0015,
      "step": 56740
    },
    {
      "epoch": 0.9285772723553956,
      "grad_norm": 0.2756028334416573,
      "learning_rate": 1.5459119825299684e-07,
      "loss": 0.0015,
      "step": 56750
    },
    {
      "epoch": 0.9287408983064714,
      "grad_norm": 0.11596582185376654,
      "learning_rate": 1.538873397958407e-07,
      "loss": 0.0025,
      "step": 56760
    },
    {
      "epoch": 0.9289045242575472,
      "grad_norm": 0.2733233755138023,
      "learning_rate": 1.5318506229723862e-07,
      "loss": 0.0031,
      "step": 56770
    },
    {
      "epoch": 0.9290681502086231,
      "grad_norm": 0.06742790971744204,
      "learning_rate": 1.5248436598629457e-07,
      "loss": 0.0012,
      "step": 56780
    },
    {
      "epoch": 0.9292317761596989,
      "grad_norm": 0.05183414729130159,
      "learning_rate": 1.517852510915996e-07,
      "loss": 0.0018,
      "step": 56790
    },
    {
      "epoch": 0.9293954021107748,
      "grad_norm": 0.24072429414971452,
      "learning_rate": 1.5108771784122735e-07,
      "loss": 0.0022,
      "step": 56800
    },
    {
      "epoch": 0.9295590280618506,
      "grad_norm": 0.1832841627061076,
      "learning_rate": 1.5039176646273467e-07,
      "loss": 0.0019,
      "step": 56810
    },
    {
      "epoch": 0.9297226540129264,
      "grad_norm": 0.17816897464389503,
      "learning_rate": 1.4969739718316445e-07,
      "loss": 0.0017,
      "step": 56820
    },
    {
      "epoch": 0.9298862799640023,
      "grad_norm": 0.17091305679947827,
      "learning_rate": 1.4900461022904212e-07,
      "loss": 0.0013,
      "step": 56830
    },
    {
      "epoch": 0.9300499059150781,
      "grad_norm": 0.1245130229761237,
      "learning_rate": 1.4831340582637688e-07,
      "loss": 0.002,
      "step": 56840
    },
    {
      "epoch": 0.930213531866154,
      "grad_norm": 0.14531875019943838,
      "learning_rate": 1.476237842006628e-07,
      "loss": 0.0014,
      "step": 56850
    },
    {
      "epoch": 0.9303771578172298,
      "grad_norm": 0.12145816635339683,
      "learning_rate": 1.469357455768755e-07,
      "loss": 0.0028,
      "step": 56860
    },
    {
      "epoch": 0.9305407837683056,
      "grad_norm": 0.04004686636789763,
      "learning_rate": 1.4624929017947654e-07,
      "loss": 0.001,
      "step": 56870
    },
    {
      "epoch": 0.9307044097193815,
      "grad_norm": 0.029071426448493906,
      "learning_rate": 1.4556441823240895e-07,
      "loss": 0.0013,
      "step": 56880
    },
    {
      "epoch": 0.9308680356704573,
      "grad_norm": 0.06711716197325754,
      "learning_rate": 1.4488112995910076e-07,
      "loss": 0.0023,
      "step": 56890
    },
    {
      "epoch": 0.9310316616215332,
      "grad_norm": 0.035698899055916326,
      "learning_rate": 1.4419942558246248e-07,
      "loss": 0.0007,
      "step": 56900
    },
    {
      "epoch": 0.931195287572609,
      "grad_norm": 0.13272245030738503,
      "learning_rate": 1.435193053248879e-07,
      "loss": 0.0023,
      "step": 56910
    },
    {
      "epoch": 0.9313589135236848,
      "grad_norm": 0.078569201300077,
      "learning_rate": 1.4284076940825454e-07,
      "loss": 0.0026,
      "step": 56920
    },
    {
      "epoch": 0.9315225394747607,
      "grad_norm": 0.1980841259891301,
      "learning_rate": 1.4216381805392254e-07,
      "loss": 0.0016,
      "step": 56930
    },
    {
      "epoch": 0.9316861654258365,
      "grad_norm": 0.26067245295665536,
      "learning_rate": 1.4148845148273472e-07,
      "loss": 0.0023,
      "step": 56940
    },
    {
      "epoch": 0.9318497913769124,
      "grad_norm": 0.1065616297874259,
      "learning_rate": 1.408146699150187e-07,
      "loss": 0.0017,
      "step": 56950
    },
    {
      "epoch": 0.9320134173279883,
      "grad_norm": 0.11164504871003998,
      "learning_rate": 1.4014247357058363e-07,
      "loss": 0.0025,
      "step": 56960
    },
    {
      "epoch": 0.932177043279064,
      "grad_norm": 0.09350048848984986,
      "learning_rate": 1.394718626687208e-07,
      "loss": 0.0024,
      "step": 56970
    },
    {
      "epoch": 0.9323406692301399,
      "grad_norm": 0.03791663160604899,
      "learning_rate": 1.3880283742820577e-07,
      "loss": 0.0009,
      "step": 56980
    },
    {
      "epoch": 0.9325042951812157,
      "grad_norm": 0.10606463926942465,
      "learning_rate": 1.3813539806729558e-07,
      "loss": 0.0026,
      "step": 56990
    },
    {
      "epoch": 0.9326679211322916,
      "grad_norm": 0.06524063400901291,
      "learning_rate": 1.3746954480373166e-07,
      "loss": 0.0018,
      "step": 57000
    },
    {
      "epoch": 0.9328315470833675,
      "grad_norm": 0.035178293335618324,
      "learning_rate": 1.3680527785473574e-07,
      "loss": 0.0025,
      "step": 57010
    },
    {
      "epoch": 0.9329951730344432,
      "grad_norm": 0.07009348355695444,
      "learning_rate": 1.361425974370134e-07,
      "loss": 0.0016,
      "step": 57020
    },
    {
      "epoch": 0.9331587989855191,
      "grad_norm": 0.14075558565600693,
      "learning_rate": 1.3548150376675396e-07,
      "loss": 0.0015,
      "step": 57030
    },
    {
      "epoch": 0.9333224249365949,
      "grad_norm": 0.08762978664546711,
      "learning_rate": 1.348219970596243e-07,
      "loss": 0.0014,
      "step": 57040
    },
    {
      "epoch": 0.9334860508876708,
      "grad_norm": 0.06341772653663495,
      "learning_rate": 1.341640775307801e-07,
      "loss": 0.0032,
      "step": 57050
    },
    {
      "epoch": 0.9336496768387467,
      "grad_norm": 0.19012312932323627,
      "learning_rate": 1.335077453948541e-07,
      "loss": 0.0032,
      "step": 57060
    },
    {
      "epoch": 0.9338133027898224,
      "grad_norm": 0.20454370362838475,
      "learning_rate": 1.3285300086596287e-07,
      "loss": 0.0018,
      "step": 57070
    },
    {
      "epoch": 0.9339769287408983,
      "grad_norm": 0.015959309125952014,
      "learning_rate": 1.321998441577066e-07,
      "loss": 0.0019,
      "step": 57080
    },
    {
      "epoch": 0.9341405546919741,
      "grad_norm": 0.05012942195603012,
      "learning_rate": 1.3154827548316596e-07,
      "loss": 0.0024,
      "step": 57090
    },
    {
      "epoch": 0.93430418064305,
      "grad_norm": 0.0740420141950303,
      "learning_rate": 1.3089829505490258e-07,
      "loss": 0.0022,
      "step": 57100
    },
    {
      "epoch": 0.9344678065941259,
      "grad_norm": 0.1300506971368136,
      "learning_rate": 1.3024990308496133e-07,
      "loss": 0.0024,
      "step": 57110
    },
    {
      "epoch": 0.9346314325452016,
      "grad_norm": 0.1279669067745699,
      "learning_rate": 1.296030997848696e-07,
      "loss": 0.0028,
      "step": 57120
    },
    {
      "epoch": 0.9347950584962775,
      "grad_norm": 0.11357931591916387,
      "learning_rate": 1.2895788536563425e-07,
      "loss": 0.0022,
      "step": 57130
    },
    {
      "epoch": 0.9349586844473533,
      "grad_norm": 0.0730682195625526,
      "learning_rate": 1.2831426003774629e-07,
      "loss": 0.0015,
      "step": 57140
    },
    {
      "epoch": 0.9351223103984292,
      "grad_norm": 0.07298688772885675,
      "learning_rate": 1.2767222401117662e-07,
      "loss": 0.002,
      "step": 57150
    },
    {
      "epoch": 0.9352859363495051,
      "grad_norm": 0.43135653167278326,
      "learning_rate": 1.270317774953772e-07,
      "loss": 0.0027,
      "step": 57160
    },
    {
      "epoch": 0.9354495623005809,
      "grad_norm": 0.20469288391008236,
      "learning_rate": 1.263929206992842e-07,
      "loss": 0.002,
      "step": 57170
    },
    {
      "epoch": 0.9356131882516567,
      "grad_norm": 0.11269816826346526,
      "learning_rate": 1.2575565383131206e-07,
      "loss": 0.0014,
      "step": 57180
    },
    {
      "epoch": 0.9357768142027325,
      "grad_norm": 0.10954650216225208,
      "learning_rate": 1.251199770993572e-07,
      "loss": 0.0014,
      "step": 57190
    },
    {
      "epoch": 0.9359404401538084,
      "grad_norm": 0.0659310650420113,
      "learning_rate": 1.2448589071079987e-07,
      "loss": 0.0021,
      "step": 57200
    },
    {
      "epoch": 0.9361040661048843,
      "grad_norm": 0.07375968158490528,
      "learning_rate": 1.2385339487249904e-07,
      "loss": 0.0016,
      "step": 57210
    },
    {
      "epoch": 0.93626769205596,
      "grad_norm": 0.02882472831203947,
      "learning_rate": 1.2322248979079355e-07,
      "loss": 0.0015,
      "step": 57220
    },
    {
      "epoch": 0.9364313180070359,
      "grad_norm": 0.1522293991094588,
      "learning_rate": 1.2259317567150654e-07,
      "loss": 0.0019,
      "step": 57230
    },
    {
      "epoch": 0.9365949439581117,
      "grad_norm": 0.1616738125606185,
      "learning_rate": 1.21965452719941e-07,
      "loss": 0.002,
      "step": 57240
    },
    {
      "epoch": 0.9367585699091876,
      "grad_norm": 0.1369339319940151,
      "learning_rate": 1.2133932114087865e-07,
      "loss": 0.002,
      "step": 57250
    },
    {
      "epoch": 0.9369221958602635,
      "grad_norm": 0.2101787684253629,
      "learning_rate": 1.2071478113858558e-07,
      "loss": 0.0025,
      "step": 57260
    },
    {
      "epoch": 0.9370858218113393,
      "grad_norm": 0.3025341491122814,
      "learning_rate": 1.2009183291680714e-07,
      "loss": 0.0033,
      "step": 57270
    },
    {
      "epoch": 0.9372494477624151,
      "grad_norm": 0.19748354962634784,
      "learning_rate": 1.194704766787669e-07,
      "loss": 0.0016,
      "step": 57280
    },
    {
      "epoch": 0.9374130737134909,
      "grad_norm": 0.10334510332518691,
      "learning_rate": 1.1885071262717318e-07,
      "loss": 0.0013,
      "step": 57290
    },
    {
      "epoch": 0.9375766996645668,
      "grad_norm": 0.2849642905969645,
      "learning_rate": 1.1823254096421322e-07,
      "loss": 0.0015,
      "step": 57300
    },
    {
      "epoch": 0.9377403256156427,
      "grad_norm": 0.098187221295278,
      "learning_rate": 1.1761596189155288e-07,
      "loss": 0.0023,
      "step": 57310
    },
    {
      "epoch": 0.9379039515667185,
      "grad_norm": 0.08360113878202924,
      "learning_rate": 1.1700097561034241e-07,
      "loss": 0.0017,
      "step": 57320
    },
    {
      "epoch": 0.9380675775177943,
      "grad_norm": 0.14315025435316453,
      "learning_rate": 1.1638758232120906e-07,
      "loss": 0.0011,
      "step": 57330
    },
    {
      "epoch": 0.9382312034688701,
      "grad_norm": 0.06717658369038658,
      "learning_rate": 1.1577578222426167e-07,
      "loss": 0.0026,
      "step": 57340
    },
    {
      "epoch": 0.938394829419946,
      "grad_norm": 0.10961696660116207,
      "learning_rate": 1.151655755190878e-07,
      "loss": 0.0016,
      "step": 57350
    },
    {
      "epoch": 0.9385584553710219,
      "grad_norm": 0.061201656868999824,
      "learning_rate": 1.1455696240475933e-07,
      "loss": 0.0018,
      "step": 57360
    },
    {
      "epoch": 0.9387220813220977,
      "grad_norm": 0.15769008040150126,
      "learning_rate": 1.1394994307982354e-07,
      "loss": 0.0013,
      "step": 57370
    },
    {
      "epoch": 0.9388857072731736,
      "grad_norm": 0.11343356284083142,
      "learning_rate": 1.1334451774230981e-07,
      "loss": 0.0021,
      "step": 57380
    },
    {
      "epoch": 0.9390493332242493,
      "grad_norm": 0.01264929846501807,
      "learning_rate": 1.1274068658972904e-07,
      "loss": 0.0014,
      "step": 57390
    },
    {
      "epoch": 0.9392129591753252,
      "grad_norm": 0.059587544571843856,
      "learning_rate": 1.1213844981906918e-07,
      "loss": 0.0015,
      "step": 57400
    },
    {
      "epoch": 0.9393765851264011,
      "grad_norm": 0.15512595960570466,
      "learning_rate": 1.1153780762679867e-07,
      "loss": 0.0021,
      "step": 57410
    },
    {
      "epoch": 0.9395402110774769,
      "grad_norm": 0.3498625192855743,
      "learning_rate": 1.1093876020886796e-07,
      "loss": 0.0019,
      "step": 57420
    },
    {
      "epoch": 0.9397038370285528,
      "grad_norm": 0.26820861576025584,
      "learning_rate": 1.1034130776070517e-07,
      "loss": 0.0021,
      "step": 57430
    },
    {
      "epoch": 0.9398674629796285,
      "grad_norm": 0.26564742324063867,
      "learning_rate": 1.0974545047721774e-07,
      "loss": 0.0019,
      "step": 57440
    },
    {
      "epoch": 0.9400310889307044,
      "grad_norm": 0.14831202440308094,
      "learning_rate": 1.091511885527946e-07,
      "loss": 0.0027,
      "step": 57450
    },
    {
      "epoch": 0.9401947148817803,
      "grad_norm": 0.1467088339922101,
      "learning_rate": 1.0855852218130292e-07,
      "loss": 0.002,
      "step": 57460
    },
    {
      "epoch": 0.9403583408328561,
      "grad_norm": 0.22317499697435758,
      "learning_rate": 1.0796745155608912e-07,
      "loss": 0.0046,
      "step": 57470
    },
    {
      "epoch": 0.940521966783932,
      "grad_norm": 0.3008313420863906,
      "learning_rate": 1.0737797686998063e-07,
      "loss": 0.0033,
      "step": 57480
    },
    {
      "epoch": 0.9406855927350077,
      "grad_norm": 0.10023575467983976,
      "learning_rate": 1.0679009831528197e-07,
      "loss": 0.0012,
      "step": 57490
    },
    {
      "epoch": 0.9408492186860836,
      "grad_norm": 0.1560097083887483,
      "learning_rate": 1.0620381608377861e-07,
      "loss": 0.0024,
      "step": 57500
    },
    {
      "epoch": 0.9410128446371595,
      "grad_norm": 0.14623449655897894,
      "learning_rate": 1.0561913036673532e-07,
      "loss": 0.0016,
      "step": 57510
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.1746786265840518,
      "learning_rate": 1.0503604135489454e-07,
      "loss": 0.002,
      "step": 57520
    },
    {
      "epoch": 0.9413400965393112,
      "grad_norm": 0.19754127043718617,
      "learning_rate": 1.0445454923847854e-07,
      "loss": 0.0009,
      "step": 57530
    },
    {
      "epoch": 0.9415037224903869,
      "grad_norm": 0.10687056682983168,
      "learning_rate": 1.0387465420719e-07,
      "loss": 0.0041,
      "step": 57540
    },
    {
      "epoch": 0.9416673484414628,
      "grad_norm": 0.07158143132011872,
      "learning_rate": 1.0329635645020874e-07,
      "loss": 0.0022,
      "step": 57550
    },
    {
      "epoch": 0.9418309743925387,
      "grad_norm": 0.3643703200421384,
      "learning_rate": 1.0271965615619439e-07,
      "loss": 0.0024,
      "step": 57560
    },
    {
      "epoch": 0.9419946003436145,
      "grad_norm": 0.16318978961743827,
      "learning_rate": 1.0214455351328534e-07,
      "loss": 0.0021,
      "step": 57570
    },
    {
      "epoch": 0.9421582262946904,
      "grad_norm": 0.38479551199978956,
      "learning_rate": 1.0157104870909818e-07,
      "loss": 0.0027,
      "step": 57580
    },
    {
      "epoch": 0.9423218522457661,
      "grad_norm": 0.0439078938311911,
      "learning_rate": 1.009991419307288e-07,
      "loss": 0.0016,
      "step": 57590
    },
    {
      "epoch": 0.942485478196842,
      "grad_norm": 0.07117097248671124,
      "learning_rate": 1.0042883336475184e-07,
      "loss": 0.0021,
      "step": 57600
    },
    {
      "epoch": 0.9426491041479178,
      "grad_norm": 0.03976758142853129,
      "learning_rate": 9.986012319722072e-08,
      "loss": 0.0009,
      "step": 57610
    },
    {
      "epoch": 0.9428127300989937,
      "grad_norm": 0.21025729793978354,
      "learning_rate": 9.929301161366645e-08,
      "loss": 0.002,
      "step": 57620
    },
    {
      "epoch": 0.9429763560500696,
      "grad_norm": 0.08384339552263549,
      "learning_rate": 9.872749879909938e-08,
      "loss": 0.0021,
      "step": 57630
    },
    {
      "epoch": 0.9431399820011453,
      "grad_norm": 0.00995640040133507,
      "learning_rate": 9.816358493800859e-08,
      "loss": 0.0017,
      "step": 57640
    },
    {
      "epoch": 0.9433036079522212,
      "grad_norm": 0.16673134761115105,
      "learning_rate": 9.760127021436028e-08,
      "loss": 0.0018,
      "step": 57650
    },
    {
      "epoch": 0.943467233903297,
      "grad_norm": 0.15131448777040565,
      "learning_rate": 9.704055481159935e-08,
      "loss": 0.0019,
      "step": 57660
    },
    {
      "epoch": 0.9436308598543729,
      "grad_norm": 0.17648385860033064,
      "learning_rate": 9.648143891265061e-08,
      "loss": 0.0016,
      "step": 57670
    },
    {
      "epoch": 0.9437944858054488,
      "grad_norm": 0.11145648299233683,
      "learning_rate": 9.59239226999148e-08,
      "loss": 0.0092,
      "step": 57680
    },
    {
      "epoch": 0.9439581117565246,
      "grad_norm": 0.05464597136457868,
      "learning_rate": 9.536800635527199e-08,
      "loss": 0.0015,
      "step": 57690
    },
    {
      "epoch": 0.9441217377076004,
      "grad_norm": 0.0852903111123607,
      "learning_rate": 9.481369006007989e-08,
      "loss": 0.0017,
      "step": 57700
    },
    {
      "epoch": 0.9442853636586762,
      "grad_norm": 0.28332547458117857,
      "learning_rate": 9.426097399517442e-08,
      "loss": 0.0021,
      "step": 57710
    },
    {
      "epoch": 0.9444489896097521,
      "grad_norm": 0.07890118215253496,
      "learning_rate": 9.370985834086965e-08,
      "loss": 0.0033,
      "step": 57720
    },
    {
      "epoch": 0.944612615560828,
      "grad_norm": 0.11731707107158314,
      "learning_rate": 9.316034327695678e-08,
      "loss": 0.0018,
      "step": 57730
    },
    {
      "epoch": 0.9447762415119038,
      "grad_norm": 0.10439705564443469,
      "learning_rate": 9.261242898270628e-08,
      "loss": 0.0015,
      "step": 57740
    },
    {
      "epoch": 0.9449398674629796,
      "grad_norm": 0.2118576170435444,
      "learning_rate": 9.206611563686462e-08,
      "loss": 0.002,
      "step": 57750
    },
    {
      "epoch": 0.9451034934140554,
      "grad_norm": 0.23106391372005236,
      "learning_rate": 9.152140341765758e-08,
      "loss": 0.0032,
      "step": 57760
    },
    {
      "epoch": 0.9452671193651313,
      "grad_norm": 0.13841575980113488,
      "learning_rate": 9.097829250278745e-08,
      "loss": 0.0026,
      "step": 57770
    },
    {
      "epoch": 0.9454307453162072,
      "grad_norm": 0.24185666531752786,
      "learning_rate": 9.043678306943415e-08,
      "loss": 0.0021,
      "step": 57780
    },
    {
      "epoch": 0.945594371267283,
      "grad_norm": 0.04672540598081172,
      "learning_rate": 8.98968752942564e-08,
      "loss": 0.0012,
      "step": 57790
    },
    {
      "epoch": 0.9457579972183588,
      "grad_norm": 0.059538991817699075,
      "learning_rate": 8.935856935338938e-08,
      "loss": 0.0013,
      "step": 57800
    },
    {
      "epoch": 0.9459216231694346,
      "grad_norm": 0.3392578938019102,
      "learning_rate": 8.882186542244541e-08,
      "loss": 0.0031,
      "step": 57810
    },
    {
      "epoch": 0.9460852491205105,
      "grad_norm": 0.31083706253497007,
      "learning_rate": 8.828676367651557e-08,
      "loss": 0.0013,
      "step": 57820
    },
    {
      "epoch": 0.9462488750715864,
      "grad_norm": 0.1860213339422809,
      "learning_rate": 8.775326429016629e-08,
      "loss": 0.0016,
      "step": 57830
    },
    {
      "epoch": 0.9464125010226622,
      "grad_norm": 0.1705466382144579,
      "learning_rate": 8.722136743744226e-08,
      "loss": 0.0023,
      "step": 57840
    },
    {
      "epoch": 0.946576126973738,
      "grad_norm": 0.07463009332992492,
      "learning_rate": 8.669107329186632e-08,
      "loss": 0.002,
      "step": 57850
    },
    {
      "epoch": 0.9467397529248138,
      "grad_norm": 0.07838546247182297,
      "learning_rate": 8.616238202643734e-08,
      "loss": 0.0021,
      "step": 57860
    },
    {
      "epoch": 0.9469033788758897,
      "grad_norm": 0.17287996555879356,
      "learning_rate": 8.563529381363122e-08,
      "loss": 0.0019,
      "step": 57870
    },
    {
      "epoch": 0.9470670048269656,
      "grad_norm": 0.07361842165655666,
      "learning_rate": 8.510980882540154e-08,
      "loss": 0.0034,
      "step": 57880
    },
    {
      "epoch": 0.9472306307780414,
      "grad_norm": 0.11886692029617128,
      "learning_rate": 8.458592723317782e-08,
      "loss": 0.0017,
      "step": 57890
    },
    {
      "epoch": 0.9473942567291173,
      "grad_norm": 0.10832935220816493,
      "learning_rate": 8.406364920786725e-08,
      "loss": 0.001,
      "step": 57900
    },
    {
      "epoch": 0.947557882680193,
      "grad_norm": 0.13630495427533593,
      "learning_rate": 8.354297491985464e-08,
      "loss": 0.0035,
      "step": 57910
    },
    {
      "epoch": 0.9477215086312689,
      "grad_norm": 0.07950426101852105,
      "learning_rate": 8.302390453900022e-08,
      "loss": 0.0025,
      "step": 57920
    },
    {
      "epoch": 0.9478851345823448,
      "grad_norm": 0.1487496874647099,
      "learning_rate": 8.250643823464133e-08,
      "loss": 0.0019,
      "step": 57930
    },
    {
      "epoch": 0.9480487605334206,
      "grad_norm": 0.020843188758355054,
      "learning_rate": 8.199057617559236e-08,
      "loss": 0.001,
      "step": 57940
    },
    {
      "epoch": 0.9482123864844965,
      "grad_norm": 0.2952459486863922,
      "learning_rate": 8.147631853014482e-08,
      "loss": 0.0018,
      "step": 57950
    },
    {
      "epoch": 0.9483760124355722,
      "grad_norm": 0.08148179194821677,
      "learning_rate": 8.096366546606449e-08,
      "loss": 0.0025,
      "step": 57960
    },
    {
      "epoch": 0.9485396383866481,
      "grad_norm": 0.10041543337758739,
      "learning_rate": 8.045261715059704e-08,
      "loss": 0.0013,
      "step": 57970
    },
    {
      "epoch": 0.948703264337724,
      "grad_norm": 0.10200542727009591,
      "learning_rate": 7.994317375046245e-08,
      "loss": 0.0033,
      "step": 57980
    },
    {
      "epoch": 0.9488668902887998,
      "grad_norm": 0.1860606213210743,
      "learning_rate": 7.943533543185721e-08,
      "loss": 0.0016,
      "step": 57990
    },
    {
      "epoch": 0.9490305162398757,
      "grad_norm": 0.24175031837053038,
      "learning_rate": 7.892910236045548e-08,
      "loss": 0.0015,
      "step": 58000
    },
    {
      "epoch": 0.9491941421909514,
      "grad_norm": 0.20399133500850378,
      "learning_rate": 7.842447470140568e-08,
      "loss": 0.0019,
      "step": 58010
    },
    {
      "epoch": 0.9493577681420273,
      "grad_norm": 0.15817768432957002,
      "learning_rate": 7.792145261933448e-08,
      "loss": 0.0022,
      "step": 58020
    },
    {
      "epoch": 0.9495213940931032,
      "grad_norm": 0.16162145096613162,
      "learning_rate": 7.742003627834338e-08,
      "loss": 0.0014,
      "step": 58030
    },
    {
      "epoch": 0.949685020044179,
      "grad_norm": 0.18688942702686606,
      "learning_rate": 7.692022584201154e-08,
      "loss": 0.0021,
      "step": 58040
    },
    {
      "epoch": 0.9498486459952549,
      "grad_norm": 0.11563409014251906,
      "learning_rate": 7.64220214733924e-08,
      "loss": 0.001,
      "step": 58050
    },
    {
      "epoch": 0.9500122719463306,
      "grad_norm": 0.051110764280992774,
      "learning_rate": 7.592542333501651e-08,
      "loss": 0.0011,
      "step": 58060
    },
    {
      "epoch": 0.9501758978974065,
      "grad_norm": 0.03884345266024981,
      "learning_rate": 7.54304315888904e-08,
      "loss": 0.0013,
      "step": 58070
    },
    {
      "epoch": 0.9503395238484824,
      "grad_norm": 0.054317386873103,
      "learning_rate": 7.493704639649602e-08,
      "loss": 0.0023,
      "step": 58080
    },
    {
      "epoch": 0.9505031497995582,
      "grad_norm": 0.07201268639119436,
      "learning_rate": 7.444526791879181e-08,
      "loss": 0.0012,
      "step": 58090
    },
    {
      "epoch": 0.9506667757506341,
      "grad_norm": 0.21490043800639175,
      "learning_rate": 7.395509631621223e-08,
      "loss": 0.0034,
      "step": 58100
    },
    {
      "epoch": 0.9508304017017098,
      "grad_norm": 0.15654913306028245,
      "learning_rate": 7.346653174866603e-08,
      "loss": 0.0013,
      "step": 58110
    },
    {
      "epoch": 0.9509940276527857,
      "grad_norm": 0.14601754224007132,
      "learning_rate": 7.297957437554016e-08,
      "loss": 0.0018,
      "step": 58120
    },
    {
      "epoch": 0.9511576536038616,
      "grad_norm": 0.07126028025654964,
      "learning_rate": 7.249422435569475e-08,
      "loss": 0.0025,
      "step": 58130
    },
    {
      "epoch": 0.9513212795549374,
      "grad_norm": 0.0694428526235549,
      "learning_rate": 7.201048184746706e-08,
      "loss": 0.0022,
      "step": 58140
    },
    {
      "epoch": 0.9514849055060133,
      "grad_norm": 0.08929014996456064,
      "learning_rate": 7.152834700866918e-08,
      "loss": 0.0019,
      "step": 58150
    },
    {
      "epoch": 0.951648531457089,
      "grad_norm": 0.060434368732969986,
      "learning_rate": 7.104781999658972e-08,
      "loss": 0.0014,
      "step": 58160
    },
    {
      "epoch": 0.9518121574081649,
      "grad_norm": 0.15602636410976803,
      "learning_rate": 7.056890096799219e-08,
      "loss": 0.0011,
      "step": 58170
    },
    {
      "epoch": 0.9519757833592408,
      "grad_norm": 0.0996863302610937,
      "learning_rate": 7.009159007911437e-08,
      "loss": 0.0017,
      "step": 58180
    },
    {
      "epoch": 0.9521394093103166,
      "grad_norm": 0.12972064806547276,
      "learning_rate": 6.961588748567116e-08,
      "loss": 0.0021,
      "step": 58190
    },
    {
      "epoch": 0.9523030352613925,
      "grad_norm": 0.08658670534574064,
      "learning_rate": 6.914179334285231e-08,
      "loss": 0.0015,
      "step": 58200
    },
    {
      "epoch": 0.9524666612124683,
      "grad_norm": 0.15722632037480724,
      "learning_rate": 6.866930780532244e-08,
      "loss": 0.0018,
      "step": 58210
    },
    {
      "epoch": 0.9526302871635441,
      "grad_norm": 0.017855789143822184,
      "learning_rate": 6.819843102722156e-08,
      "loss": 0.0019,
      "step": 58220
    },
    {
      "epoch": 0.95279391311462,
      "grad_norm": 0.11670280063917686,
      "learning_rate": 6.772916316216515e-08,
      "loss": 0.0015,
      "step": 58230
    },
    {
      "epoch": 0.9529575390656958,
      "grad_norm": 0.07319547865123004,
      "learning_rate": 6.726150436324297e-08,
      "loss": 0.0016,
      "step": 58240
    },
    {
      "epoch": 0.9531211650167717,
      "grad_norm": 0.1718648617408185,
      "learning_rate": 6.67954547830213e-08,
      "loss": 0.0015,
      "step": 58250
    },
    {
      "epoch": 0.9532847909678475,
      "grad_norm": 0.1570285759968204,
      "learning_rate": 6.633101457353963e-08,
      "loss": 0.0021,
      "step": 58260
    },
    {
      "epoch": 0.9534484169189233,
      "grad_norm": 0.08621283209454726,
      "learning_rate": 6.5868183886314e-08,
      "loss": 0.0022,
      "step": 58270
    },
    {
      "epoch": 0.9536120428699992,
      "grad_norm": 0.1883249645522094,
      "learning_rate": 6.540696287233473e-08,
      "loss": 0.0014,
      "step": 58280
    },
    {
      "epoch": 0.953775668821075,
      "grad_norm": 0.11254000514423784,
      "learning_rate": 6.4947351682067e-08,
      "loss": 0.0014,
      "step": 58290
    },
    {
      "epoch": 0.9539392947721509,
      "grad_norm": 0.08483314696834378,
      "learning_rate": 6.448935046545035e-08,
      "loss": 0.0021,
      "step": 58300
    },
    {
      "epoch": 0.9541029207232267,
      "grad_norm": 0.09377186964172586,
      "learning_rate": 6.403295937190024e-08,
      "loss": 0.0014,
      "step": 58310
    },
    {
      "epoch": 0.9542665466743026,
      "grad_norm": 0.2818619264692372,
      "learning_rate": 6.357817855030535e-08,
      "loss": 0.0021,
      "step": 58320
    },
    {
      "epoch": 0.9544301726253784,
      "grad_norm": 0.1573992950822469,
      "learning_rate": 6.312500814903089e-08,
      "loss": 0.0027,
      "step": 58330
    },
    {
      "epoch": 0.9545937985764542,
      "grad_norm": 0.10990883119366637,
      "learning_rate": 6.267344831591471e-08,
      "loss": 0.0025,
      "step": 58340
    },
    {
      "epoch": 0.9547574245275301,
      "grad_norm": 0.05807698840918103,
      "learning_rate": 6.22234991982712e-08,
      "loss": 0.0031,
      "step": 58350
    },
    {
      "epoch": 0.9549210504786059,
      "grad_norm": 0.13845174000889665,
      "learning_rate": 6.177516094288794e-08,
      "loss": 0.0037,
      "step": 58360
    },
    {
      "epoch": 0.9550846764296818,
      "grad_norm": 0.2171505790129501,
      "learning_rate": 6.132843369602626e-08,
      "loss": 0.0024,
      "step": 58370
    },
    {
      "epoch": 0.9552483023807576,
      "grad_norm": 0.09716249160551478,
      "learning_rate": 6.088331760342514e-08,
      "loss": 0.0028,
      "step": 58380
    },
    {
      "epoch": 0.9554119283318334,
      "grad_norm": 0.1062334325527875,
      "learning_rate": 6.043981281029399e-08,
      "loss": 0.0024,
      "step": 58390
    },
    {
      "epoch": 0.9555755542829093,
      "grad_norm": 0.056018049459040584,
      "learning_rate": 5.999791946131872e-08,
      "loss": 0.0014,
      "step": 58400
    },
    {
      "epoch": 0.9557391802339851,
      "grad_norm": 0.11403456451433759,
      "learning_rate": 5.95576377006607e-08,
      "loss": 0.0009,
      "step": 58410
    },
    {
      "epoch": 0.955902806185061,
      "grad_norm": 0.036328271352262125,
      "learning_rate": 5.91189676719528e-08,
      "loss": 0.0027,
      "step": 58420
    },
    {
      "epoch": 0.9560664321361368,
      "grad_norm": 0.07385231077441959,
      "learning_rate": 5.868190951830277e-08,
      "loss": 0.0022,
      "step": 58430
    },
    {
      "epoch": 0.9562300580872126,
      "grad_norm": 0.09146875614916235,
      "learning_rate": 5.824646338229434e-08,
      "loss": 0.001,
      "step": 58440
    },
    {
      "epoch": 0.9563936840382885,
      "grad_norm": 0.06927950598409106,
      "learning_rate": 5.781262940598331e-08,
      "loss": 0.0025,
      "step": 58450
    },
    {
      "epoch": 0.9565573099893643,
      "grad_norm": 0.05861033656856944,
      "learning_rate": 5.7380407730900365e-08,
      "loss": 0.0014,
      "step": 58460
    },
    {
      "epoch": 0.9567209359404402,
      "grad_norm": 0.09198446793586258,
      "learning_rate": 5.69497984980516e-08,
      "loss": 0.0024,
      "step": 58470
    },
    {
      "epoch": 0.9568845618915159,
      "grad_norm": 0.08272597589262085,
      "learning_rate": 5.6520801847914086e-08,
      "loss": 0.0023,
      "step": 58480
    },
    {
      "epoch": 0.9570481878425918,
      "grad_norm": 0.49210044509054446,
      "learning_rate": 5.609341792044032e-08,
      "loss": 0.002,
      "step": 58490
    },
    {
      "epoch": 0.9572118137936677,
      "grad_norm": 0.1264082969257141,
      "learning_rate": 5.566764685505766e-08,
      "loss": 0.0018,
      "step": 58500
    },
    {
      "epoch": 0.9573754397447435,
      "grad_norm": 0.043639992297095156,
      "learning_rate": 5.52434887906661e-08,
      "loss": 0.0019,
      "step": 58510
    },
    {
      "epoch": 0.9575390656958194,
      "grad_norm": 0.06063293893603418,
      "learning_rate": 5.4820943865639406e-08,
      "loss": 0.0016,
      "step": 58520
    },
    {
      "epoch": 0.9577026916468951,
      "grad_norm": 0.10659193134897706,
      "learning_rate": 5.4400012217826203e-08,
      "loss": 0.0016,
      "step": 58530
    },
    {
      "epoch": 0.957866317597971,
      "grad_norm": 0.055529780267450785,
      "learning_rate": 5.398069398454664e-08,
      "loss": 0.0013,
      "step": 58540
    },
    {
      "epoch": 0.9580299435490469,
      "grad_norm": 0.0952217861096849,
      "learning_rate": 5.3562989302596844e-08,
      "loss": 0.0025,
      "step": 58550
    },
    {
      "epoch": 0.9581935695001227,
      "grad_norm": 0.05072254899464642,
      "learning_rate": 5.3146898308245044e-08,
      "loss": 0.0014,
      "step": 58560
    },
    {
      "epoch": 0.9583571954511986,
      "grad_norm": 0.020095364657897512,
      "learning_rate": 5.273242113723431e-08,
      "loss": 0.0021,
      "step": 58570
    },
    {
      "epoch": 0.9585208214022743,
      "grad_norm": 0.1907754411091576,
      "learning_rate": 5.231955792477983e-08,
      "loss": 0.0037,
      "step": 58580
    },
    {
      "epoch": 0.9586844473533502,
      "grad_norm": 0.045497601401816176,
      "learning_rate": 5.190830880557163e-08,
      "loss": 0.0009,
      "step": 58590
    },
    {
      "epoch": 0.9588480733044261,
      "grad_norm": 0.15124947919922588,
      "learning_rate": 5.149867391377183e-08,
      "loss": 0.001,
      "step": 58600
    },
    {
      "epoch": 0.9590116992555019,
      "grad_norm": 0.09432583288913221,
      "learning_rate": 5.109065338301633e-08,
      "loss": 0.0016,
      "step": 58610
    },
    {
      "epoch": 0.9591753252065778,
      "grad_norm": 0.09533512774151824,
      "learning_rate": 5.068424734641531e-08,
      "loss": 0.0014,
      "step": 58620
    },
    {
      "epoch": 0.9593389511576536,
      "grad_norm": 0.04971137859136583,
      "learning_rate": 5.027945593655159e-08,
      "loss": 0.0019,
      "step": 58630
    },
    {
      "epoch": 0.9595025771087294,
      "grad_norm": 0.14007880921430796,
      "learning_rate": 4.987627928548011e-08,
      "loss": 0.0013,
      "step": 58640
    },
    {
      "epoch": 0.9596662030598053,
      "grad_norm": 0.32399685364011604,
      "learning_rate": 4.9474717524731766e-08,
      "loss": 0.0019,
      "step": 58650
    },
    {
      "epoch": 0.9598298290108811,
      "grad_norm": 0.23621912378973192,
      "learning_rate": 4.9074770785307314e-08,
      "loss": 0.0025,
      "step": 58660
    },
    {
      "epoch": 0.959993454961957,
      "grad_norm": 0.07226511489232992,
      "learning_rate": 4.8676439197683503e-08,
      "loss": 0.002,
      "step": 58670
    },
    {
      "epoch": 0.9601570809130328,
      "grad_norm": 0.12753092311203978,
      "learning_rate": 4.8279722891808045e-08,
      "loss": 0.0028,
      "step": 58680
    },
    {
      "epoch": 0.9603207068641086,
      "grad_norm": 0.1713443720081189,
      "learning_rate": 4.788462199710409e-08,
      "loss": 0.0016,
      "step": 58690
    },
    {
      "epoch": 0.9604843328151845,
      "grad_norm": 0.1619404154763327,
      "learning_rate": 4.749113664246463e-08,
      "loss": 0.0017,
      "step": 58700
    },
    {
      "epoch": 0.9606479587662603,
      "grad_norm": 0.1762528319638496,
      "learning_rate": 4.7099266956258106e-08,
      "loss": 0.0013,
      "step": 58710
    },
    {
      "epoch": 0.9608115847173362,
      "grad_norm": 0.13254920026153588,
      "learning_rate": 4.6709013066325584e-08,
      "loss": 0.0019,
      "step": 58720
    },
    {
      "epoch": 0.960975210668412,
      "grad_norm": 0.1114083813222817,
      "learning_rate": 4.632037509997911e-08,
      "loss": 0.0018,
      "step": 58730
    },
    {
      "epoch": 0.9611388366194878,
      "grad_norm": 0.14792282514812852,
      "learning_rate": 4.593335318400616e-08,
      "loss": 0.002,
      "step": 58740
    },
    {
      "epoch": 0.9613024625705637,
      "grad_norm": 0.07469875358178601,
      "learning_rate": 4.554794744466518e-08,
      "loss": 0.0013,
      "step": 58750
    },
    {
      "epoch": 0.9614660885216395,
      "grad_norm": 0.1330607806646815,
      "learning_rate": 4.516415800768836e-08,
      "loss": 0.0026,
      "step": 58760
    },
    {
      "epoch": 0.9616297144727154,
      "grad_norm": 0.05893461980846049,
      "learning_rate": 4.4781984998280016e-08,
      "loss": 0.0017,
      "step": 58770
    },
    {
      "epoch": 0.9617933404237912,
      "grad_norm": 0.21197573672220962,
      "learning_rate": 4.4401428541117065e-08,
      "loss": 0.0018,
      "step": 58780
    },
    {
      "epoch": 0.961956966374867,
      "grad_norm": 0.035055742843520976,
      "learning_rate": 4.40224887603502e-08,
      "loss": 0.0018,
      "step": 58790
    },
    {
      "epoch": 0.9621205923259429,
      "grad_norm": 0.1767009529574022,
      "learning_rate": 4.364516577960054e-08,
      "loss": 0.0021,
      "step": 58800
    },
    {
      "epoch": 0.9622842182770187,
      "grad_norm": 0.03855386249913598,
      "learning_rate": 4.326945972196406e-08,
      "loss": 0.0011,
      "step": 58810
    },
    {
      "epoch": 0.9624478442280946,
      "grad_norm": 0.04992735238216703,
      "learning_rate": 4.289537071000827e-08,
      "loss": 0.0014,
      "step": 58820
    },
    {
      "epoch": 0.9626114701791704,
      "grad_norm": 0.29948722544256373,
      "learning_rate": 4.252289886577221e-08,
      "loss": 0.0021,
      "step": 58830
    },
    {
      "epoch": 0.9627750961302463,
      "grad_norm": 0.4223010527745787,
      "learning_rate": 4.215204431076869e-08,
      "loss": 0.0021,
      "step": 58840
    },
    {
      "epoch": 0.9629387220813221,
      "grad_norm": 0.5181825178333329,
      "learning_rate": 4.17828071659826e-08,
      "loss": 0.0033,
      "step": 58850
    },
    {
      "epoch": 0.9631023480323979,
      "grad_norm": 0.2576587417736698,
      "learning_rate": 4.1415187551870904e-08,
      "loss": 0.0026,
      "step": 58860
    },
    {
      "epoch": 0.9632659739834738,
      "grad_norm": 0.1288760711502654,
      "learning_rate": 4.1049185588363235e-08,
      "loss": 0.0013,
      "step": 58870
    },
    {
      "epoch": 0.9634295999345496,
      "grad_norm": 0.1157015071951014,
      "learning_rate": 4.068480139486075e-08,
      "loss": 0.003,
      "step": 58880
    },
    {
      "epoch": 0.9635932258856255,
      "grad_norm": 0.09418429479830841,
      "learning_rate": 4.03220350902378e-08,
      "loss": 0.0024,
      "step": 58890
    },
    {
      "epoch": 0.9637568518367013,
      "grad_norm": 0.15711689506567122,
      "learning_rate": 3.996088679284027e-08,
      "loss": 0.0012,
      "step": 58900
    },
    {
      "epoch": 0.9639204777877771,
      "grad_norm": 0.1982201105431582,
      "learning_rate": 3.96013566204867e-08,
      "loss": 0.0019,
      "step": 58910
    },
    {
      "epoch": 0.964084103738853,
      "grad_norm": 0.08307877563910385,
      "learning_rate": 3.9243444690466595e-08,
      "loss": 0.0014,
      "step": 58920
    },
    {
      "epoch": 0.9642477296899288,
      "grad_norm": 0.13529120702789255,
      "learning_rate": 3.8887151119543224e-08,
      "loss": 0.0034,
      "step": 58930
    },
    {
      "epoch": 0.9644113556410047,
      "grad_norm": 0.06915962588247587,
      "learning_rate": 3.8532476023950826e-08,
      "loss": 0.002,
      "step": 58940
    },
    {
      "epoch": 0.9645749815920805,
      "grad_norm": 0.163514832598477,
      "learning_rate": 3.81794195193963e-08,
      "loss": 0.0012,
      "step": 58950
    },
    {
      "epoch": 0.9647386075431563,
      "grad_norm": 0.1861182780364438,
      "learning_rate": 3.7827981721057506e-08,
      "loss": 0.0021,
      "step": 58960
    },
    {
      "epoch": 0.9649022334942322,
      "grad_norm": 0.08265726360160489,
      "learning_rate": 3.747816274358496e-08,
      "loss": 0.0035,
      "step": 58970
    },
    {
      "epoch": 0.965065859445308,
      "grad_norm": 0.09025027947093056,
      "learning_rate": 3.7129962701100695e-08,
      "loss": 0.0018,
      "step": 58980
    },
    {
      "epoch": 0.9652294853963839,
      "grad_norm": 0.026983059154927706,
      "learning_rate": 3.6783381707198864e-08,
      "loss": 0.0021,
      "step": 58990
    },
    {
      "epoch": 0.9653931113474598,
      "grad_norm": 0.10786923256664574,
      "learning_rate": 3.6438419874946226e-08,
      "loss": 0.0017,
      "step": 59000
    },
    {
      "epoch": 0.9655567372985355,
      "grad_norm": 0.18649585935766294,
      "learning_rate": 3.609507731687945e-08,
      "loss": 0.002,
      "step": 59010
    },
    {
      "epoch": 0.9657203632496114,
      "grad_norm": 0.1826757814165257,
      "learning_rate": 3.575335414500836e-08,
      "loss": 0.0019,
      "step": 59020
    },
    {
      "epoch": 0.9658839892006872,
      "grad_norm": 0.14165149828885565,
      "learning_rate": 3.5413250470814345e-08,
      "loss": 0.0016,
      "step": 59030
    },
    {
      "epoch": 0.9660476151517631,
      "grad_norm": 0.09198019882211068,
      "learning_rate": 3.507476640524976e-08,
      "loss": 0.0031,
      "step": 59040
    },
    {
      "epoch": 0.966211241102839,
      "grad_norm": 0.32003227795756034,
      "learning_rate": 3.473790205873906e-08,
      "loss": 0.002,
      "step": 59050
    },
    {
      "epoch": 0.9663748670539147,
      "grad_norm": 0.04296782108078748,
      "learning_rate": 3.4402657541178774e-08,
      "loss": 0.0016,
      "step": 59060
    },
    {
      "epoch": 0.9665384930049906,
      "grad_norm": 0.1058783659322161,
      "learning_rate": 3.406903296193587e-08,
      "loss": 0.0018,
      "step": 59070
    },
    {
      "epoch": 0.9667021189560664,
      "grad_norm": 0.0658202332603944,
      "learning_rate": 3.373702842984994e-08,
      "loss": 0.0017,
      "step": 59080
    },
    {
      "epoch": 0.9668657449071423,
      "grad_norm": 0.05841415716324278,
      "learning_rate": 3.3406644053231575e-08,
      "loss": 0.0014,
      "step": 59090
    },
    {
      "epoch": 0.9670293708582182,
      "grad_norm": 0.14747002710639198,
      "learning_rate": 3.307787993986289e-08,
      "loss": 0.0021,
      "step": 59100
    },
    {
      "epoch": 0.9671929968092939,
      "grad_norm": 0.07797708003476621,
      "learning_rate": 3.275073619699698e-08,
      "loss": 0.0011,
      "step": 59110
    },
    {
      "epoch": 0.9673566227603698,
      "grad_norm": 0.061814869828232005,
      "learning_rate": 3.2425212931358474e-08,
      "loss": 0.0031,
      "step": 59120
    },
    {
      "epoch": 0.9675202487114456,
      "grad_norm": 0.21167295534928415,
      "learning_rate": 3.210131024914465e-08,
      "loss": 0.0012,
      "step": 59130
    },
    {
      "epoch": 0.9676838746625215,
      "grad_norm": 0.19233268342782223,
      "learning_rate": 3.177902825602208e-08,
      "loss": 0.0016,
      "step": 59140
    },
    {
      "epoch": 0.9678475006135974,
      "grad_norm": 0.05341542310600103,
      "learning_rate": 3.1458367057130014e-08,
      "loss": 0.0019,
      "step": 59150
    },
    {
      "epoch": 0.9680111265646731,
      "grad_norm": 0.08800903468986607,
      "learning_rate": 3.113932675707809e-08,
      "loss": 0.0016,
      "step": 59160
    },
    {
      "epoch": 0.968174752515749,
      "grad_norm": 0.07436792341772674,
      "learning_rate": 3.0821907459948065e-08,
      "loss": 0.0015,
      "step": 59170
    },
    {
      "epoch": 0.9683383784668248,
      "grad_norm": 0.08332377943824636,
      "learning_rate": 3.0506109269292096e-08,
      "loss": 0.0014,
      "step": 59180
    },
    {
      "epoch": 0.9685020044179007,
      "grad_norm": 0.04003653261401171,
      "learning_rate": 3.019193228813388e-08,
      "loss": 0.0017,
      "step": 59190
    },
    {
      "epoch": 0.9686656303689766,
      "grad_norm": 0.11065206703349333,
      "learning_rate": 2.987937661896756e-08,
      "loss": 0.0022,
      "step": 59200
    },
    {
      "epoch": 0.9688292563200523,
      "grad_norm": 0.14535483657078666,
      "learning_rate": 2.9568442363759886e-08,
      "loss": 0.0026,
      "step": 59210
    },
    {
      "epoch": 0.9689928822711282,
      "grad_norm": 0.2643876476246675,
      "learning_rate": 2.9259129623946947e-08,
      "loss": 0.0026,
      "step": 59220
    },
    {
      "epoch": 0.969156508222204,
      "grad_norm": 0.3101119129184491,
      "learning_rate": 2.8951438500436356e-08,
      "loss": 0.0028,
      "step": 59230
    },
    {
      "epoch": 0.9693201341732799,
      "grad_norm": 0.1360174006912135,
      "learning_rate": 2.8645369093607822e-08,
      "loss": 0.0012,
      "step": 59240
    },
    {
      "epoch": 0.9694837601243558,
      "grad_norm": 0.08346650051480006,
      "learning_rate": 2.834092150331036e-08,
      "loss": 0.0015,
      "step": 59250
    },
    {
      "epoch": 0.9696473860754315,
      "grad_norm": 0.4930706098389054,
      "learning_rate": 2.8038095828865076e-08,
      "loss": 0.0029,
      "step": 59260
    },
    {
      "epoch": 0.9698110120265074,
      "grad_norm": 0.25214083220587347,
      "learning_rate": 2.7736892169062946e-08,
      "loss": 0.0016,
      "step": 59270
    },
    {
      "epoch": 0.9699746379775832,
      "grad_norm": 0.09960147036513034,
      "learning_rate": 2.743731062216648e-08,
      "loss": 0.0049,
      "step": 59280
    },
    {
      "epoch": 0.9701382639286591,
      "grad_norm": 0.21324623017451128,
      "learning_rate": 2.713935128590861e-08,
      "loss": 0.0016,
      "step": 59290
    },
    {
      "epoch": 0.970301889879735,
      "grad_norm": 0.27334612285223886,
      "learning_rate": 2.684301425749436e-08,
      "loss": 0.0008,
      "step": 59300
    },
    {
      "epoch": 0.9704655158308108,
      "grad_norm": 0.11816460247486872,
      "learning_rate": 2.654829963359751e-08,
      "loss": 0.0021,
      "step": 59310
    },
    {
      "epoch": 0.9706291417818866,
      "grad_norm": 0.04813337233950942,
      "learning_rate": 2.6255207510363922e-08,
      "loss": 0.0021,
      "step": 59320
    },
    {
      "epoch": 0.9707927677329624,
      "grad_norm": 0.17738386614394322,
      "learning_rate": 2.5963737983408788e-08,
      "loss": 0.0038,
      "step": 59330
    },
    {
      "epoch": 0.9709563936840383,
      "grad_norm": 0.24040021250330543,
      "learning_rate": 2.5673891147820485e-08,
      "loss": 0.002,
      "step": 59340
    },
    {
      "epoch": 0.9711200196351142,
      "grad_norm": 0.19466918878162232,
      "learning_rate": 2.538566709815504e-08,
      "loss": 0.0025,
      "step": 59350
    },
    {
      "epoch": 0.97128364558619,
      "grad_norm": 0.17773145149528702,
      "learning_rate": 2.5099065928441134e-08,
      "loss": 0.0034,
      "step": 59360
    },
    {
      "epoch": 0.9714472715372658,
      "grad_norm": 0.05459227072386862,
      "learning_rate": 2.4814087732177306e-08,
      "loss": 0.0012,
      "step": 59370
    },
    {
      "epoch": 0.9716108974883416,
      "grad_norm": 0.08172895539775601,
      "learning_rate": 2.4530732602332518e-08,
      "loss": 0.0021,
      "step": 59380
    },
    {
      "epoch": 0.9717745234394175,
      "grad_norm": 0.14318585848052304,
      "learning_rate": 2.4249000631346166e-08,
      "loss": 0.0017,
      "step": 59390
    },
    {
      "epoch": 0.9719381493904933,
      "grad_norm": 0.04051257625889486,
      "learning_rate": 2.396889191112861e-08,
      "loss": 0.0013,
      "step": 59400
    },
    {
      "epoch": 0.9721017753415692,
      "grad_norm": 0.107637063203466,
      "learning_rate": 2.3690406533060094e-08,
      "loss": 0.0029,
      "step": 59410
    },
    {
      "epoch": 0.972265401292645,
      "grad_norm": 0.11457841291240357,
      "learning_rate": 2.3413544587991833e-08,
      "loss": 0.0024,
      "step": 59420
    },
    {
      "epoch": 0.9724290272437208,
      "grad_norm": 0.2300968102212999,
      "learning_rate": 2.3138306166244907e-08,
      "loss": 0.0019,
      "step": 59430
    },
    {
      "epoch": 0.9725926531947967,
      "grad_norm": 0.18972001386808235,
      "learning_rate": 2.286469135761138e-08,
      "loss": 0.0016,
      "step": 59440
    },
    {
      "epoch": 0.9727562791458725,
      "grad_norm": 0.04985368895327254,
      "learning_rate": 2.2592700251353182e-08,
      "loss": 0.0023,
      "step": 59450
    },
    {
      "epoch": 0.9729199050969484,
      "grad_norm": 0.03357155345479672,
      "learning_rate": 2.2322332936202117e-08,
      "loss": 0.0012,
      "step": 59460
    },
    {
      "epoch": 0.9730835310480243,
      "grad_norm": 0.09113797402212406,
      "learning_rate": 2.2053589500360962e-08,
      "loss": 0.0022,
      "step": 59470
    },
    {
      "epoch": 0.9732471569991,
      "grad_norm": 0.21901168326204895,
      "learning_rate": 2.178647003150236e-08,
      "loss": 0.0022,
      "step": 59480
    },
    {
      "epoch": 0.9734107829501759,
      "grad_norm": 0.2551297975725779,
      "learning_rate": 2.1520974616769942e-08,
      "loss": 0.0025,
      "step": 59490
    },
    {
      "epoch": 0.9735744089012517,
      "grad_norm": 0.0618221506773981,
      "learning_rate": 2.125710334277664e-08,
      "loss": 0.002,
      "step": 59500
    },
    {
      "epoch": 0.9737380348523276,
      "grad_norm": 0.03967075894498542,
      "learning_rate": 2.0994856295604716e-08,
      "loss": 0.0023,
      "step": 59510
    },
    {
      "epoch": 0.9739016608034035,
      "grad_norm": 0.2032083535959931,
      "learning_rate": 2.0734233560809613e-08,
      "loss": 0.0019,
      "step": 59520
    },
    {
      "epoch": 0.9740652867544792,
      "grad_norm": 0.12680773828821879,
      "learning_rate": 2.0475235223413326e-08,
      "loss": 0.0021,
      "step": 59530
    },
    {
      "epoch": 0.9742289127055551,
      "grad_norm": 0.14590244319232062,
      "learning_rate": 2.0217861367909374e-08,
      "loss": 0.0023,
      "step": 59540
    },
    {
      "epoch": 0.9743925386566309,
      "grad_norm": 0.07087200504061775,
      "learning_rate": 1.996211207826282e-08,
      "loss": 0.0019,
      "step": 59550
    },
    {
      "epoch": 0.9745561646077068,
      "grad_norm": 0.12425254022015622,
      "learning_rate": 1.9707987437906918e-08,
      "loss": 0.002,
      "step": 59560
    },
    {
      "epoch": 0.9747197905587827,
      "grad_norm": 0.17336981494620313,
      "learning_rate": 1.9455487529744243e-08,
      "loss": 0.0028,
      "step": 59570
    },
    {
      "epoch": 0.9748834165098584,
      "grad_norm": 0.043014389223232675,
      "learning_rate": 1.9204612436148906e-08,
      "loss": 0.0023,
      "step": 59580
    },
    {
      "epoch": 0.9750470424609343,
      "grad_norm": 0.12651873851608017,
      "learning_rate": 1.8955362238965434e-08,
      "loss": 0.0015,
      "step": 59590
    },
    {
      "epoch": 0.9752106684120101,
      "grad_norm": 0.081356149594276,
      "learning_rate": 1.870773701950601e-08,
      "loss": 0.0011,
      "step": 59600
    },
    {
      "epoch": 0.975374294363086,
      "grad_norm": 0.11469312910784575,
      "learning_rate": 1.84617368585549e-08,
      "loss": 0.0019,
      "step": 59610
    },
    {
      "epoch": 0.9755379203141619,
      "grad_norm": 0.3527551942721497,
      "learning_rate": 1.8217361836365133e-08,
      "loss": 0.0035,
      "step": 59620
    },
    {
      "epoch": 0.9757015462652376,
      "grad_norm": 0.38648847626159505,
      "learning_rate": 1.7974612032659045e-08,
      "loss": 0.0021,
      "step": 59630
    },
    {
      "epoch": 0.9758651722163135,
      "grad_norm": 0.10299573418207243,
      "learning_rate": 1.773348752663051e-08,
      "loss": 0.0015,
      "step": 59640
    },
    {
      "epoch": 0.9760287981673893,
      "grad_norm": 0.10301574535942533,
      "learning_rate": 1.7493988396941054e-08,
      "loss": 0.0019,
      "step": 59650
    },
    {
      "epoch": 0.9761924241184652,
      "grad_norm": 0.11802126826548867,
      "learning_rate": 1.7256114721724282e-08,
      "loss": 0.002,
      "step": 59660
    },
    {
      "epoch": 0.9763560500695411,
      "grad_norm": 0.0030538399335342265,
      "learning_rate": 1.7019866578580903e-08,
      "loss": 0.0014,
      "step": 59670
    },
    {
      "epoch": 0.9765196760206168,
      "grad_norm": 0.136377095464529,
      "learning_rate": 1.6785244044583704e-08,
      "loss": 0.0016,
      "step": 59680
    },
    {
      "epoch": 0.9766833019716927,
      "grad_norm": 0.16419438578215229,
      "learning_rate": 1.6552247196273685e-08,
      "loss": 0.0041,
      "step": 59690
    },
    {
      "epoch": 0.9768469279227685,
      "grad_norm": 0.09723657990337446,
      "learning_rate": 1.6320876109662263e-08,
      "loss": 0.0023,
      "step": 59700
    },
    {
      "epoch": 0.9770105538738444,
      "grad_norm": 0.03841561328032325,
      "learning_rate": 1.609113086022962e-08,
      "loss": 0.0014,
      "step": 59710
    },
    {
      "epoch": 0.9771741798249203,
      "grad_norm": 0.2924383671696704,
      "learning_rate": 1.586301152292691e-08,
      "loss": 0.0028,
      "step": 59720
    },
    {
      "epoch": 0.977337805775996,
      "grad_norm": 0.07734073986322952,
      "learning_rate": 1.563651817217293e-08,
      "loss": 0.002,
      "step": 59730
    },
    {
      "epoch": 0.9775014317270719,
      "grad_norm": 0.16447490146375676,
      "learning_rate": 1.5411650881858032e-08,
      "loss": 0.0015,
      "step": 59740
    },
    {
      "epoch": 0.9776650576781477,
      "grad_norm": 0.011035378061128838,
      "learning_rate": 1.5188409725341303e-08,
      "loss": 0.0023,
      "step": 59750
    },
    {
      "epoch": 0.9778286836292236,
      "grad_norm": 0.22294155272790692,
      "learning_rate": 1.4966794775450045e-08,
      "loss": 0.0017,
      "step": 59760
    },
    {
      "epoch": 0.9779923095802995,
      "grad_norm": 0.0329407718034747,
      "learning_rate": 1.4746806104483647e-08,
      "loss": 0.0013,
      "step": 59770
    },
    {
      "epoch": 0.9781559355313753,
      "grad_norm": 0.19837866998813475,
      "learning_rate": 1.452844378420859e-08,
      "loss": 0.0021,
      "step": 59780
    },
    {
      "epoch": 0.9783195614824511,
      "grad_norm": 0.23418321617723242,
      "learning_rate": 1.4311707885862336e-08,
      "loss": 0.0024,
      "step": 59790
    },
    {
      "epoch": 0.9784831874335269,
      "grad_norm": 0.16701794287448482,
      "learning_rate": 1.4096598480150547e-08,
      "loss": 0.0012,
      "step": 59800
    },
    {
      "epoch": 0.9786468133846028,
      "grad_norm": 0.05403760177995624,
      "learning_rate": 1.3883115637249311e-08,
      "loss": 0.0019,
      "step": 59810
    },
    {
      "epoch": 0.9788104393356787,
      "grad_norm": 0.20010892685513446,
      "learning_rate": 1.3671259426802919e-08,
      "loss": 0.0019,
      "step": 59820
    },
    {
      "epoch": 0.9789740652867545,
      "grad_norm": 0.05099498487915146,
      "learning_rate": 1.3461029917926639e-08,
      "loss": 0.0023,
      "step": 59830
    },
    {
      "epoch": 0.9791376912378303,
      "grad_norm": 0.19569519818610545,
      "learning_rate": 1.3252427179203942e-08,
      "loss": 0.0025,
      "step": 59840
    },
    {
      "epoch": 0.9793013171889061,
      "grad_norm": 0.16812385832725257,
      "learning_rate": 1.304545127868706e-08,
      "loss": 0.0018,
      "step": 59850
    },
    {
      "epoch": 0.979464943139982,
      "grad_norm": 0.37850236999141923,
      "learning_rate": 1.2840102283899203e-08,
      "loss": 0.0019,
      "step": 59860
    },
    {
      "epoch": 0.9796285690910579,
      "grad_norm": 0.06222418048503737,
      "learning_rate": 1.2636380261831227e-08,
      "loss": 0.0017,
      "step": 59870
    },
    {
      "epoch": 0.9797921950421337,
      "grad_norm": 0.02610674765750475,
      "learning_rate": 1.2434285278943858e-08,
      "loss": 0.0013,
      "step": 59880
    },
    {
      "epoch": 0.9799558209932095,
      "grad_norm": 0.0649896475655067,
      "learning_rate": 1.2233817401167136e-08,
      "loss": 0.0023,
      "step": 59890
    },
    {
      "epoch": 0.9801194469442853,
      "grad_norm": 0.14589999646242188,
      "learning_rate": 1.2034976693900414e-08,
      "loss": 0.0017,
      "step": 59900
    },
    {
      "epoch": 0.9802830728953612,
      "grad_norm": 0.1447093633269783,
      "learning_rate": 1.1837763222011245e-08,
      "loss": 0.0022,
      "step": 59910
    },
    {
      "epoch": 0.9804466988464371,
      "grad_norm": 0.08877573551656542,
      "learning_rate": 1.1642177049837612e-08,
      "loss": 0.0014,
      "step": 59920
    },
    {
      "epoch": 0.9806103247975129,
      "grad_norm": 0.17573779159000005,
      "learning_rate": 1.1448218241186249e-08,
      "loss": 0.0019,
      "step": 59930
    },
    {
      "epoch": 0.9807739507485888,
      "grad_norm": 0.009034029920050495,
      "learning_rate": 1.1255886859332099e-08,
      "loss": 0.0022,
      "step": 59940
    },
    {
      "epoch": 0.9809375766996645,
      "grad_norm": 0.14292365039736996,
      "learning_rate": 1.1065182967020526e-08,
      "loss": 0.0016,
      "step": 59950
    },
    {
      "epoch": 0.9811012026507404,
      "grad_norm": 0.05376385954739454,
      "learning_rate": 1.08761066264651e-08,
      "loss": 0.0018,
      "step": 59960
    },
    {
      "epoch": 0.9812648286018163,
      "grad_norm": 0.11353706314926723,
      "learning_rate": 1.0688657899348699e-08,
      "loss": 0.0018,
      "step": 59970
    },
    {
      "epoch": 0.9814284545528921,
      "grad_norm": 0.10294632008534421,
      "learning_rate": 1.0502836846822405e-08,
      "loss": 0.0015,
      "step": 59980
    },
    {
      "epoch": 0.981592080503968,
      "grad_norm": 0.17648528721068663,
      "learning_rate": 1.0318643529508287e-08,
      "loss": 0.0031,
      "step": 59990
    },
    {
      "epoch": 0.9817557064550437,
      "grad_norm": 0.18783365905344576,
      "learning_rate": 1.0136078007494944e-08,
      "loss": 0.003,
      "step": 60000
    },
    {
      "epoch": 0.9819193324061196,
      "grad_norm": 0.48773854582930476,
      "learning_rate": 9.955140340341951e-09,
      "loss": 0.0027,
      "step": 60010
    },
    {
      "epoch": 0.9820829583571955,
      "grad_norm": 0.10260869997809423,
      "learning_rate": 9.775830587077095e-09,
      "loss": 0.0019,
      "step": 60020
    },
    {
      "epoch": 0.9822465843082713,
      "grad_norm": 0.04587447910006072,
      "learning_rate": 9.59814880619636e-09,
      "loss": 0.0021,
      "step": 60030
    },
    {
      "epoch": 0.9824102102593472,
      "grad_norm": 0.06979774753379442,
      "learning_rate": 9.422095055665603e-09,
      "loss": 0.0022,
      "step": 60040
    },
    {
      "epoch": 0.9825738362104229,
      "grad_norm": 0.08442468186483452,
      "learning_rate": 9.247669392919433e-09,
      "loss": 0.0015,
      "step": 60050
    },
    {
      "epoch": 0.9827374621614988,
      "grad_norm": 0.12147376585719137,
      "learning_rate": 9.074871874861224e-09,
      "loss": 0.0027,
      "step": 60060
    },
    {
      "epoch": 0.9829010881125747,
      "grad_norm": 0.04215229422578133,
      "learning_rate": 8.903702557862549e-09,
      "loss": 0.0024,
      "step": 60070
    },
    {
      "epoch": 0.9830647140636505,
      "grad_norm": 0.07337007150459161,
      "learning_rate": 8.734161497764848e-09,
      "loss": 0.001,
      "step": 60080
    },
    {
      "epoch": 0.9832283400147264,
      "grad_norm": 0.08379642415713438,
      "learning_rate": 8.56624874987777e-09,
      "loss": 0.0032,
      "step": 60090
    },
    {
      "epoch": 0.9833919659658021,
      "grad_norm": 0.15789040921661168,
      "learning_rate": 8.399964368979719e-09,
      "loss": 0.002,
      "step": 60100
    },
    {
      "epoch": 0.983555591916878,
      "grad_norm": 0.5568401126340339,
      "learning_rate": 8.23530840931841e-09,
      "loss": 0.0024,
      "step": 60110
    },
    {
      "epoch": 0.9837192178679539,
      "grad_norm": 0.0796814150994545,
      "learning_rate": 8.07228092460921e-09,
      "loss": 0.0023,
      "step": 60120
    },
    {
      "epoch": 0.9838828438190297,
      "grad_norm": 0.1087612451064515,
      "learning_rate": 7.910881968037909e-09,
      "loss": 0.0033,
      "step": 60130
    },
    {
      "epoch": 0.9840464697701056,
      "grad_norm": 0.11558822922658991,
      "learning_rate": 7.751111592257388e-09,
      "loss": 0.0019,
      "step": 60140
    },
    {
      "epoch": 0.9842100957211813,
      "grad_norm": 0.10832463574397937,
      "learning_rate": 7.592969849390398e-09,
      "loss": 0.0009,
      "step": 60150
    },
    {
      "epoch": 0.9843737216722572,
      "grad_norm": 0.19450215338490973,
      "learning_rate": 7.436456791027336e-09,
      "loss": 0.003,
      "step": 60160
    },
    {
      "epoch": 0.9845373476233331,
      "grad_norm": 0.37797237332341194,
      "learning_rate": 7.2815724682279154e-09,
      "loss": 0.0029,
      "step": 60170
    },
    {
      "epoch": 0.9847009735744089,
      "grad_norm": 0.014815793272699846,
      "learning_rate": 7.128316931521162e-09,
      "loss": 0.0012,
      "step": 60180
    },
    {
      "epoch": 0.9848645995254848,
      "grad_norm": 0.08314315539099056,
      "learning_rate": 6.976690230902639e-09,
      "loss": 0.0017,
      "step": 60190
    },
    {
      "epoch": 0.9850282254765605,
      "grad_norm": 0.006542973292016352,
      "learning_rate": 6.826692415838887e-09,
      "loss": 0.0012,
      "step": 60200
    },
    {
      "epoch": 0.9851918514276364,
      "grad_norm": 0.07894588516068819,
      "learning_rate": 6.6783235352640976e-09,
      "loss": 0.0011,
      "step": 60210
    },
    {
      "epoch": 0.9853554773787123,
      "grad_norm": 0.15213493909389453,
      "learning_rate": 6.531583637580663e-09,
      "loss": 0.001,
      "step": 60220
    },
    {
      "epoch": 0.9855191033297881,
      "grad_norm": 0.065571064615504,
      "learning_rate": 6.386472770659735e-09,
      "loss": 0.0022,
      "step": 60230
    },
    {
      "epoch": 0.985682729280864,
      "grad_norm": 0.12963975197460978,
      "learning_rate": 6.242990981841224e-09,
      "loss": 0.0021,
      "step": 60240
    },
    {
      "epoch": 0.9858463552319398,
      "grad_norm": 0.11514803523916374,
      "learning_rate": 6.1011383179337966e-09,
      "loss": 0.0018,
      "step": 60250
    },
    {
      "epoch": 0.9860099811830156,
      "grad_norm": 0.1571361272708824,
      "learning_rate": 5.960914825214326e-09,
      "loss": 0.0018,
      "step": 60260
    },
    {
      "epoch": 0.9861736071340914,
      "grad_norm": 0.12742103712251568,
      "learning_rate": 5.822320549427884e-09,
      "loss": 0.0016,
      "step": 60270
    },
    {
      "epoch": 0.9863372330851673,
      "grad_norm": 0.2458001491079989,
      "learning_rate": 5.685355535788861e-09,
      "loss": 0.0011,
      "step": 60280
    },
    {
      "epoch": 0.9865008590362432,
      "grad_norm": 0.11390066827535633,
      "learning_rate": 5.5500198289792915e-09,
      "loss": 0.004,
      "step": 60290
    },
    {
      "epoch": 0.986664484987319,
      "grad_norm": 0.19573018563892566,
      "learning_rate": 5.41631347314997e-09,
      "loss": 0.0025,
      "step": 60300
    },
    {
      "epoch": 0.9868281109383948,
      "grad_norm": 0.012970873493516008,
      "learning_rate": 5.284236511921004e-09,
      "loss": 0.0028,
      "step": 60310
    },
    {
      "epoch": 0.9869917368894706,
      "grad_norm": 0.055039228827746596,
      "learning_rate": 5.153788988379593e-09,
      "loss": 0.0032,
      "step": 60320
    },
    {
      "epoch": 0.9871553628405465,
      "grad_norm": 0.1409922209437654,
      "learning_rate": 5.0249709450816976e-09,
      "loss": 0.0016,
      "step": 60330
    },
    {
      "epoch": 0.9873189887916224,
      "grad_norm": 0.0653035032829552,
      "learning_rate": 4.897782424052589e-09,
      "loss": 0.0015,
      "step": 60340
    },
    {
      "epoch": 0.9874826147426982,
      "grad_norm": 0.10716871217137415,
      "learning_rate": 4.772223466784631e-09,
      "loss": 0.0019,
      "step": 60350
    },
    {
      "epoch": 0.987646240693774,
      "grad_norm": 0.26245431623881244,
      "learning_rate": 4.648294114239505e-09,
      "loss": 0.0044,
      "step": 60360
    },
    {
      "epoch": 0.9878098666448498,
      "grad_norm": 0.0721433153889858,
      "learning_rate": 4.5259944068470894e-09,
      "loss": 0.0018,
      "step": 60370
    },
    {
      "epoch": 0.9879734925959257,
      "grad_norm": 0.13675506840579402,
      "learning_rate": 4.4053243845049165e-09,
      "loss": 0.002,
      "step": 60380
    },
    {
      "epoch": 0.9881371185470016,
      "grad_norm": 0.0978644226734315,
      "learning_rate": 4.286284086580383e-09,
      "loss": 0.0013,
      "step": 60390
    },
    {
      "epoch": 0.9883007444980774,
      "grad_norm": 0.042673595689395584,
      "learning_rate": 4.168873551907982e-09,
      "loss": 0.0023,
      "step": 60400
    },
    {
      "epoch": 0.9884643704491533,
      "grad_norm": 0.12913790495598768,
      "learning_rate": 4.053092818790405e-09,
      "loss": 0.0019,
      "step": 60410
    },
    {
      "epoch": 0.988627996400229,
      "grad_norm": 0.24522300422463322,
      "learning_rate": 3.9389419249991065e-09,
      "loss": 0.0018,
      "step": 60420
    },
    {
      "epoch": 0.9887916223513049,
      "grad_norm": 0.0409492574790415,
      "learning_rate": 3.8264209077742975e-09,
      "loss": 0.0014,
      "step": 60430
    },
    {
      "epoch": 0.9889552483023808,
      "grad_norm": 0.050325729929773584,
      "learning_rate": 3.715529803823281e-09,
      "loss": 0.0017,
      "step": 60440
    },
    {
      "epoch": 0.9891188742534566,
      "grad_norm": 0.08215262686447887,
      "learning_rate": 3.606268649322675e-09,
      "loss": 0.0014,
      "step": 60450
    },
    {
      "epoch": 0.9892825002045325,
      "grad_norm": 0.16526693680903798,
      "learning_rate": 3.4986374799172995e-09,
      "loss": 0.0017,
      "step": 60460
    },
    {
      "epoch": 0.9894461261556082,
      "grad_norm": 0.1745853996568687,
      "learning_rate": 3.392636330719623e-09,
      "loss": 0.0016,
      "step": 60470
    },
    {
      "epoch": 0.9896097521066841,
      "grad_norm": 0.07733288679227483,
      "learning_rate": 3.2882652363108726e-09,
      "loss": 0.0007,
      "step": 60480
    },
    {
      "epoch": 0.98977337805776,
      "grad_norm": 0.1123718270286566,
      "learning_rate": 3.1855242307399226e-09,
      "loss": 0.0022,
      "step": 60490
    },
    {
      "epoch": 0.9899370040088358,
      "grad_norm": 0.1410498263659475,
      "learning_rate": 3.0844133475238513e-09,
      "loss": 0.0018,
      "step": 60500
    },
    {
      "epoch": 0.9901006299599117,
      "grad_norm": 0.16450532311059904,
      "learning_rate": 2.9849326196496055e-09,
      "loss": 0.0016,
      "step": 60510
    },
    {
      "epoch": 0.9902642559109874,
      "grad_norm": 0.09082114787697831,
      "learning_rate": 2.8870820795695585e-09,
      "loss": 0.0016,
      "step": 60520
    },
    {
      "epoch": 0.9904278818620633,
      "grad_norm": 0.013173648504481422,
      "learning_rate": 2.7908617592065092e-09,
      "loss": 0.0021,
      "step": 60530
    },
    {
      "epoch": 0.9905915078131392,
      "grad_norm": 0.021228948905903194,
      "learning_rate": 2.6962716899509024e-09,
      "loss": 0.0017,
      "step": 60540
    },
    {
      "epoch": 0.990755133764215,
      "grad_norm": 0.2829491886722695,
      "learning_rate": 2.603311902660277e-09,
      "loss": 0.0026,
      "step": 60550
    },
    {
      "epoch": 0.9909187597152909,
      "grad_norm": 0.10691415150689901,
      "learning_rate": 2.5119824276614857e-09,
      "loss": 0.0018,
      "step": 60560
    },
    {
      "epoch": 0.9910823856663666,
      "grad_norm": 0.14732791728354516,
      "learning_rate": 2.4222832947490284e-09,
      "loss": 0.0023,
      "step": 60570
    },
    {
      "epoch": 0.9912460116174425,
      "grad_norm": 0.09089862558597359,
      "learning_rate": 2.3342145331861633e-09,
      "loss": 0.0016,
      "step": 60580
    },
    {
      "epoch": 0.9914096375685184,
      "grad_norm": 0.06804651704965908,
      "learning_rate": 2.247776171703242e-09,
      "loss": 0.002,
      "step": 60590
    },
    {
      "epoch": 0.9915732635195942,
      "grad_norm": 0.5243362934115394,
      "learning_rate": 2.1629682384999295e-09,
      "loss": 0.0039,
      "step": 60600
    },
    {
      "epoch": 0.9917368894706701,
      "grad_norm": 0.039580329474393994,
      "learning_rate": 2.0797907612424286e-09,
      "loss": 0.002,
      "step": 60610
    },
    {
      "epoch": 0.9919005154217458,
      "grad_norm": 0.07388725810098937,
      "learning_rate": 1.9982437670662546e-09,
      "loss": 0.0018,
      "step": 60620
    },
    {
      "epoch": 0.9920641413728217,
      "grad_norm": 0.32566069699409705,
      "learning_rate": 1.9183272825745726e-09,
      "loss": 0.0017,
      "step": 60630
    },
    {
      "epoch": 0.9922277673238976,
      "grad_norm": 0.11709789684202111,
      "learning_rate": 1.840041333838749e-09,
      "loss": 0.0017,
      "step": 60640
    },
    {
      "epoch": 0.9923913932749734,
      "grad_norm": 0.2858933598390796,
      "learning_rate": 1.76338594639891e-09,
      "loss": 0.0017,
      "step": 60650
    },
    {
      "epoch": 0.9925550192260493,
      "grad_norm": 0.23445480320381554,
      "learning_rate": 1.6883611452611636e-09,
      "loss": 0.0031,
      "step": 60660
    },
    {
      "epoch": 0.992718645177125,
      "grad_norm": 0.19399728146755638,
      "learning_rate": 1.614966954902597e-09,
      "loss": 0.0018,
      "step": 60670
    },
    {
      "epoch": 0.9928822711282009,
      "grad_norm": 0.11307517802571326,
      "learning_rate": 1.5432033992651696e-09,
      "loss": 0.0016,
      "step": 60680
    },
    {
      "epoch": 0.9930458970792768,
      "grad_norm": 0.10675311099084006,
      "learning_rate": 1.4730705017618196e-09,
      "loss": 0.0032,
      "step": 60690
    },
    {
      "epoch": 0.9932095230303526,
      "grad_norm": 0.19889546965927152,
      "learning_rate": 1.4045682852714682e-09,
      "loss": 0.0019,
      "step": 60700
    },
    {
      "epoch": 0.9933731489814285,
      "grad_norm": 0.10604112466126486,
      "learning_rate": 1.3376967721417945e-09,
      "loss": 0.002,
      "step": 60710
    },
    {
      "epoch": 0.9935367749325043,
      "grad_norm": 0.06959550731607973,
      "learning_rate": 1.2724559841886808e-09,
      "loss": 0.0018,
      "step": 60720
    },
    {
      "epoch": 0.9937004008835801,
      "grad_norm": 0.10219405662664624,
      "learning_rate": 1.2088459426951026e-09,
      "loss": 0.0021,
      "step": 60730
    },
    {
      "epoch": 0.993864026834656,
      "grad_norm": 0.15125987983040315,
      "learning_rate": 1.146866668413904e-09,
      "loss": 0.0038,
      "step": 60740
    },
    {
      "epoch": 0.9940276527857318,
      "grad_norm": 0.05925288769258631,
      "learning_rate": 1.0865181815633563e-09,
      "loss": 0.0016,
      "step": 60750
    },
    {
      "epoch": 0.9941912787368077,
      "grad_norm": 0.16897204394201365,
      "learning_rate": 1.0278005018327097e-09,
      "loss": 0.0026,
      "step": 60760
    },
    {
      "epoch": 0.9943549046878835,
      "grad_norm": 0.07610001527467722,
      "learning_rate": 9.70713648376087e-10,
      "loss": 0.0024,
      "step": 60770
    },
    {
      "epoch": 0.9945185306389593,
      "grad_norm": 0.11757244247354288,
      "learning_rate": 9.152576398174795e-10,
      "loss": 0.001,
      "step": 60780
    },
    {
      "epoch": 0.9946821565900352,
      "grad_norm": 0.14934542054597172,
      "learning_rate": 8.614324942490815e-10,
      "loss": 0.0035,
      "step": 60790
    },
    {
      "epoch": 0.994845782541111,
      "grad_norm": 0.16757262928104008,
      "learning_rate": 8.092382292296253e-10,
      "loss": 0.003,
      "step": 60800
    },
    {
      "epoch": 0.9950094084921869,
      "grad_norm": 0.33882316409394125,
      "learning_rate": 7.586748617866013e-10,
      "loss": 0.0028,
      "step": 60810
    },
    {
      "epoch": 0.9951730344432627,
      "grad_norm": 0.0746601687953866,
      "learning_rate": 7.097424084157034e-10,
      "loss": 0.0007,
      "step": 60820
    },
    {
      "epoch": 0.9953366603943385,
      "grad_norm": 0.1824925243432742,
      "learning_rate": 6.624408850808284e-10,
      "loss": 0.0031,
      "step": 60830
    },
    {
      "epoch": 0.9955002863454144,
      "grad_norm": 0.09700631805414069,
      "learning_rate": 6.167703072118558e-10,
      "loss": 0.0015,
      "step": 60840
    },
    {
      "epoch": 0.9956639122964902,
      "grad_norm": 0.09849100566872825,
      "learning_rate": 5.72730689709089e-10,
      "loss": 0.0023,
      "step": 60850
    },
    {
      "epoch": 0.9958275382475661,
      "grad_norm": 0.19214481254598562,
      "learning_rate": 5.303220469399239e-10,
      "loss": 0.0035,
      "step": 60860
    },
    {
      "epoch": 0.9959911641986419,
      "grad_norm": 0.11120133144872563,
      "learning_rate": 4.895443927382948e-10,
      "loss": 0.0021,
      "step": 60870
    },
    {
      "epoch": 0.9961547901497178,
      "grad_norm": 0.155889349230826,
      "learning_rate": 4.5039774040800397e-10,
      "loss": 0.0017,
      "step": 60880
    },
    {
      "epoch": 0.9963184161007936,
      "grad_norm": 0.004320524512787298,
      "learning_rate": 4.128821027193919e-10,
      "loss": 0.0009,
      "step": 60890
    },
    {
      "epoch": 0.9964820420518694,
      "grad_norm": 0.4231217069702302,
      "learning_rate": 3.7699749191211223e-10,
      "loss": 0.0013,
      "step": 60900
    },
    {
      "epoch": 0.9966456680029453,
      "grad_norm": 0.012752050307830578,
      "learning_rate": 3.4274391969180143e-10,
      "loss": 0.0027,
      "step": 60910
    },
    {
      "epoch": 0.9968092939540211,
      "grad_norm": 0.00982818521532133,
      "learning_rate": 3.101213972345196e-10,
      "loss": 0.0018,
      "step": 60920
    },
    {
      "epoch": 0.996972919905097,
      "grad_norm": 0.2416885167538227,
      "learning_rate": 2.7912993518119936e-10,
      "loss": 0.003,
      "step": 60930
    },
    {
      "epoch": 0.9971365458561728,
      "grad_norm": 0.15182605977009,
      "learning_rate": 2.4976954364319683e-10,
      "loss": 0.0024,
      "step": 60940
    },
    {
      "epoch": 0.9973001718072486,
      "grad_norm": 0.14261567522183033,
      "learning_rate": 2.220402321989612e-10,
      "loss": 0.002,
      "step": 60950
    },
    {
      "epoch": 0.9974637977583245,
      "grad_norm": 0.11603194992930861,
      "learning_rate": 1.9594200989403455e-10,
      "loss": 0.0035,
      "step": 60960
    },
    {
      "epoch": 0.9976274237094003,
      "grad_norm": 0.25133548517888693,
      "learning_rate": 1.7147488524271728e-10,
      "loss": 0.0012,
      "step": 60970
    },
    {
      "epoch": 0.9977910496604762,
      "grad_norm": 0.06633368950966952,
      "learning_rate": 1.486388662275129e-10,
      "loss": 0.002,
      "step": 60980
    },
    {
      "epoch": 0.997954675611552,
      "grad_norm": 0.16637022913787614,
      "learning_rate": 1.274339602974628e-10,
      "loss": 0.0032,
      "step": 60990
    },
    {
      "epoch": 0.9981183015626278,
      "grad_norm": 0.09767478228395131,
      "learning_rate": 1.0786017437092177e-10,
      "loss": 0.0011,
      "step": 61000
    },
    {
      "epoch": 0.9982819275137037,
      "grad_norm": 0.12156838897465133,
      "learning_rate": 8.991751483278243e-11,
      "loss": 0.0011,
      "step": 61010
    },
    {
      "epoch": 0.9984455534647795,
      "grad_norm": 0.12433531178406149,
      "learning_rate": 7.360598753725079e-11,
      "loss": 0.0019,
      "step": 61020
    },
    {
      "epoch": 0.9986091794158554,
      "grad_norm": 0.13511939950557417,
      "learning_rate": 5.892559780562579e-11,
      "loss": 0.0014,
      "step": 61030
    },
    {
      "epoch": 0.9987728053669312,
      "grad_norm": 0.03609854062444235,
      "learning_rate": 4.587635042629934e-11,
      "loss": 0.0016,
      "step": 61040
    },
    {
      "epoch": 0.998936431318007,
      "grad_norm": 0.07079022175742873,
      "learning_rate": 3.445824965753186e-11,
      "loss": 0.0012,
      "step": 61050
    },
    {
      "epoch": 0.9991000572690829,
      "grad_norm": 0.035937299268860086,
      "learning_rate": 2.467129922356648e-11,
      "loss": 0.001,
      "step": 61060
    },
    {
      "epoch": 0.9992636832201587,
      "grad_norm": 0.18251185390265254,
      "learning_rate": 1.651550231740462e-11,
      "loss": 0.0013,
      "step": 61070
    },
    {
      "epoch": 0.9994273091712346,
      "grad_norm": 0.13268026673927336,
      "learning_rate": 9.990861599140644e-12,
      "loss": 0.002,
      "step": 61080
    },
    {
      "epoch": 0.9995909351223105,
      "grad_norm": 0.184941253110589,
      "learning_rate": 5.097379198737429e-12,
      "loss": 0.0022,
      "step": 61090
    },
    {
      "epoch": 0.9997545610733862,
      "grad_norm": 0.09222311635005549,
      "learning_rate": 1.8350567110303474e-12,
      "loss": 0.0011,
      "step": 61100
    },
    {
      "epoch": 0.9999181870244621,
      "grad_norm": 0.12849870174228717,
      "learning_rate": 2.0389520127839234e-13,
      "loss": 0.0016,
      "step": 61110
    },
    {
      "epoch": 1.0,
      "step": 61115,
      "total_flos": 389042235549696.0,
      "train_loss": 0.008355458706034012,
      "train_runtime": 62709.2593,
      "train_samples_per_second": 7.797,
      "train_steps_per_second": 0.975
    }
  ],
  "logging_steps": 10,
  "max_steps": 61115,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 389042235549696.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}