{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9995388034044694,
  "eval_steps": 500,
  "global_step": 745,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013416628233617039,
      "grad_norm": 7.392911121308514,
      "learning_rate": 0.0,
      "loss": 1.5209,
      "step": 1
    },
    {
      "epoch": 0.0026833256467234078,
      "grad_norm": 7.112473452836917,
      "learning_rate": 4.347826086956522e-07,
      "loss": 1.5143,
      "step": 2
    },
    {
      "epoch": 0.004024988470085112,
      "grad_norm": 7.177785683559338,
      "learning_rate": 8.695652173913044e-07,
      "loss": 1.4845,
      "step": 3
    },
    {
      "epoch": 0.0053666512934468155,
      "grad_norm": 7.134127065979566,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 1.5244,
      "step": 4
    },
    {
      "epoch": 0.006708314116808519,
      "grad_norm": 7.03887090554238,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 1.4937,
      "step": 5
    },
    {
      "epoch": 0.008049976940170223,
      "grad_norm": 6.431004514232511,
      "learning_rate": 2.173913043478261e-06,
      "loss": 1.4744,
      "step": 6
    },
    {
      "epoch": 0.009391639763531927,
      "grad_norm": 6.329636636805787,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 1.4811,
      "step": 7
    },
    {
      "epoch": 0.010733302586893631,
      "grad_norm": 5.167818404189778,
      "learning_rate": 3.043478260869566e-06,
      "loss": 1.4749,
      "step": 8
    },
    {
      "epoch": 0.012074965410255335,
      "grad_norm": 4.756114436376914,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 1.4369,
      "step": 9
    },
    {
      "epoch": 0.013416628233617039,
      "grad_norm": 3.1923565422063027,
      "learning_rate": 3.91304347826087e-06,
      "loss": 1.4196,
      "step": 10
    },
    {
      "epoch": 0.014758291056978743,
      "grad_norm": 3.7393628201726337,
      "learning_rate": 4.347826086956522e-06,
      "loss": 1.3848,
      "step": 11
    },
    {
      "epoch": 0.016099953880340447,
      "grad_norm": 3.820597086197902,
      "learning_rate": 4.782608695652174e-06,
      "loss": 1.3935,
      "step": 12
    },
    {
      "epoch": 0.017441616703702152,
      "grad_norm": 3.6939756049433,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 1.3457,
      "step": 13
    },
    {
      "epoch": 0.018783279527063854,
      "grad_norm": 4.7768162647135535,
      "learning_rate": 5.652173913043479e-06,
      "loss": 1.3411,
      "step": 14
    },
    {
      "epoch": 0.02012494235042556,
      "grad_norm": 5.708989901082793,
      "learning_rate": 6.086956521739132e-06,
      "loss": 1.3803,
      "step": 15
    },
    {
      "epoch": 0.021466605173787262,
      "grad_norm": 5.129788229516496,
      "learning_rate": 6.521739130434783e-06,
      "loss": 1.3514,
      "step": 16
    },
    {
      "epoch": 0.022808267997148968,
      "grad_norm": 3.957950873594244,
      "learning_rate": 6.956521739130435e-06,
      "loss": 1.3337,
      "step": 17
    },
    {
      "epoch": 0.02414993082051067,
      "grad_norm": 3.2380219205405716,
      "learning_rate": 7.391304347826087e-06,
      "loss": 1.3109,
      "step": 18
    },
    {
      "epoch": 0.025491593643872375,
      "grad_norm": 3.3372390298941963,
      "learning_rate": 7.82608695652174e-06,
      "loss": 1.3058,
      "step": 19
    },
    {
      "epoch": 0.026833256467234078,
      "grad_norm": 3.5871280327198463,
      "learning_rate": 8.260869565217392e-06,
      "loss": 1.3026,
      "step": 20
    },
    {
      "epoch": 0.028174919290595783,
      "grad_norm": 3.22326237900213,
      "learning_rate": 8.695652173913044e-06,
      "loss": 1.2918,
      "step": 21
    },
    {
      "epoch": 0.029516582113957485,
      "grad_norm": 2.5369193377611867,
      "learning_rate": 9.130434782608697e-06,
      "loss": 1.2958,
      "step": 22
    },
    {
      "epoch": 0.03085824493731919,
      "grad_norm": 2.6224230361852827,
      "learning_rate": 9.565217391304349e-06,
      "loss": 1.2228,
      "step": 23
    },
    {
      "epoch": 0.03219990776068089,
      "grad_norm": 2.582819063813051,
      "learning_rate": 1e-05,
      "loss": 1.2521,
      "step": 24
    },
    {
      "epoch": 0.0335415705840426,
      "grad_norm": 2.2985705568510957,
      "learning_rate": 9.99995279777441e-06,
      "loss": 1.2341,
      "step": 25
    },
    {
      "epoch": 0.034883233407404304,
      "grad_norm": 2.0127817582932037,
      "learning_rate": 9.999811191988863e-06,
      "loss": 1.2559,
      "step": 26
    },
    {
      "epoch": 0.036224896230766,
      "grad_norm": 1.9823343228538544,
      "learning_rate": 9.999575185316994e-06,
      "loss": 1.2264,
      "step": 27
    },
    {
      "epoch": 0.03756655905412771,
      "grad_norm": 1.792201819633752,
      "learning_rate": 9.999244782214828e-06,
      "loss": 1.2036,
      "step": 28
    },
    {
      "epoch": 0.038908221877489414,
      "grad_norm": 1.7372048174282044,
      "learning_rate": 9.998819988920665e-06,
      "loss": 1.1851,
      "step": 29
    },
    {
      "epoch": 0.04024988470085112,
      "grad_norm": 1.78600915759922,
      "learning_rate": 9.998300813454981e-06,
      "loss": 1.2168,
      "step": 30
    },
    {
      "epoch": 0.04159154752421282,
      "grad_norm": 1.7629065705331128,
      "learning_rate": 9.997687265620274e-06,
      "loss": 1.1855,
      "step": 31
    },
    {
      "epoch": 0.042933210347574524,
      "grad_norm": 1.644529086310792,
      "learning_rate": 9.996979357000869e-06,
      "loss": 1.2073,
      "step": 32
    },
    {
      "epoch": 0.04427487317093623,
      "grad_norm": 2.5577223830781333,
      "learning_rate": 9.996177100962714e-06,
      "loss": 1.1595,
      "step": 33
    },
    {
      "epoch": 0.045616535994297935,
      "grad_norm": 1.7300882554626944,
      "learning_rate": 9.995280512653116e-06,
      "loss": 1.1688,
      "step": 34
    },
    {
      "epoch": 0.046958198817659634,
      "grad_norm": 1.5646971960903784,
      "learning_rate": 9.99428960900046e-06,
      "loss": 1.1868,
      "step": 35
    },
    {
      "epoch": 0.04829986164102134,
      "grad_norm": 1.685967317065919,
      "learning_rate": 9.99320440871389e-06,
      "loss": 1.1642,
      "step": 36
    },
    {
      "epoch": 0.049641524464383045,
      "grad_norm": 1.633989524662171,
      "learning_rate": 9.992024932282955e-06,
      "loss": 1.1843,
      "step": 37
    },
    {
      "epoch": 0.05098318728774475,
      "grad_norm": 1.5510371509273424,
      "learning_rate": 9.990751201977217e-06,
      "loss": 1.1819,
      "step": 38
    },
    {
      "epoch": 0.05232485011110645,
      "grad_norm": 1.5181337823607957,
      "learning_rate": 9.98938324184584e-06,
      "loss": 1.1381,
      "step": 39
    },
    {
      "epoch": 0.053666512934468155,
      "grad_norm": 1.5147318061378112,
      "learning_rate": 9.987921077717127e-06,
      "loss": 1.1375,
      "step": 40
    },
    {
      "epoch": 0.05500817575782986,
      "grad_norm": 1.855980773594425,
      "learning_rate": 9.98636473719804e-06,
      "loss": 1.1451,
      "step": 41
    },
    {
      "epoch": 0.056349838581191566,
      "grad_norm": 1.6054934665141092,
      "learning_rate": 9.984714249673676e-06,
      "loss": 1.1447,
      "step": 42
    },
    {
      "epoch": 0.057691501404553265,
      "grad_norm": 1.447759664276557,
      "learning_rate": 9.982969646306704e-06,
      "loss": 1.1369,
      "step": 43
    },
    {
      "epoch": 0.05903316422791497,
      "grad_norm": 1.4941544438032464,
      "learning_rate": 9.98113096003679e-06,
      "loss": 1.1397,
      "step": 44
    },
    {
      "epoch": 0.060374827051276676,
      "grad_norm": 1.6610959253342235,
      "learning_rate": 9.979198225579968e-06,
      "loss": 1.1345,
      "step": 45
    },
    {
      "epoch": 0.06171648987463838,
      "grad_norm": 1.5390215688694597,
      "learning_rate": 9.97717147942799e-06,
      "loss": 1.119,
      "step": 46
    },
    {
      "epoch": 0.06305815269800008,
      "grad_norm": 1.510282897486931,
      "learning_rate": 9.97505075984762e-06,
      "loss": 1.1376,
      "step": 47
    },
    {
      "epoch": 0.06439981552136179,
      "grad_norm": 1.4415829320619717,
      "learning_rate": 9.972836106879936e-06,
      "loss": 1.1324,
      "step": 48
    },
    {
      "epoch": 0.06574147834472349,
      "grad_norm": 1.72947543283011,
      "learning_rate": 9.970527562339554e-06,
      "loss": 1.1269,
      "step": 49
    },
    {
      "epoch": 0.0670831411680852,
      "grad_norm": 1.7207642200980233,
      "learning_rate": 9.968125169813855e-06,
      "loss": 1.1116,
      "step": 50
    },
    {
      "epoch": 0.0684248039914469,
      "grad_norm": 1.571820205098199,
      "learning_rate": 9.965628974662145e-06,
      "loss": 1.1258,
      "step": 51
    },
    {
      "epoch": 0.06976646681480861,
      "grad_norm": 1.5781999464670455,
      "learning_rate": 9.963039024014811e-06,
      "loss": 1.114,
      "step": 52
    },
    {
      "epoch": 0.0711081296381703,
      "grad_norm": 1.4961557725112893,
      "learning_rate": 9.96035536677243e-06,
      "loss": 1.1138,
      "step": 53
    },
    {
      "epoch": 0.072449792461532,
      "grad_norm": 1.7347021506143052,
      "learning_rate": 9.957578053604837e-06,
      "loss": 1.1141,
      "step": 54
    },
    {
      "epoch": 0.07379145528489371,
      "grad_norm": 1.438812778229738,
      "learning_rate": 9.954707136950176e-06,
      "loss": 1.1103,
      "step": 55
    },
    {
      "epoch": 0.07513311810825542,
      "grad_norm": 1.545484642307709,
      "learning_rate": 9.951742671013914e-06,
      "loss": 1.09,
      "step": 56
    },
    {
      "epoch": 0.07647478093161712,
      "grad_norm": 1.4676495438249457,
      "learning_rate": 9.9486847117678e-06,
      "loss": 1.0811,
      "step": 57
    },
    {
      "epoch": 0.07781644375497883,
      "grad_norm": 1.4950763559495723,
      "learning_rate": 9.945533316948833e-06,
      "loss": 1.0785,
      "step": 58
    },
    {
      "epoch": 0.07915810657834053,
      "grad_norm": 1.8715712469761752,
      "learning_rate": 9.942288546058148e-06,
      "loss": 1.0837,
      "step": 59
    },
    {
      "epoch": 0.08049976940170224,
      "grad_norm": 1.692239129693084,
      "learning_rate": 9.938950460359912e-06,
      "loss": 1.0914,
      "step": 60
    },
    {
      "epoch": 0.08184143222506395,
      "grad_norm": 1.4201822302921803,
      "learning_rate": 9.935519122880152e-06,
      "loss": 1.0812,
      "step": 61
    },
    {
      "epoch": 0.08318309504842564,
      "grad_norm": 1.503736803352269,
      "learning_rate": 9.931994598405576e-06,
      "loss": 1.1036,
      "step": 62
    },
    {
      "epoch": 0.08452475787178734,
      "grad_norm": 1.6032922425382767,
      "learning_rate": 9.928376953482343e-06,
      "loss": 1.0798,
      "step": 63
    },
    {
      "epoch": 0.08586642069514905,
      "grad_norm": 1.5854938428466847,
      "learning_rate": 9.924666256414812e-06,
      "loss": 1.0911,
      "step": 64
    },
    {
      "epoch": 0.08720808351851075,
      "grad_norm": 1.4675504699964215,
      "learning_rate": 9.920862577264242e-06,
      "loss": 1.0762,
      "step": 65
    },
    {
      "epoch": 0.08854974634187246,
      "grad_norm": 2.091734701831164,
      "learning_rate": 9.916965987847485e-06,
      "loss": 1.0704,
      "step": 66
    },
    {
      "epoch": 0.08989140916523417,
      "grad_norm": 1.6501153418242909,
      "learning_rate": 9.912976561735617e-06,
      "loss": 1.059,
      "step": 67
    },
    {
      "epoch": 0.09123307198859587,
      "grad_norm": 1.5290291395938629,
      "learning_rate": 9.908894374252556e-06,
      "loss": 1.074,
      "step": 68
    },
    {
      "epoch": 0.09257473481195758,
      "grad_norm": 1.8118777602247247,
      "learning_rate": 9.904719502473635e-06,
      "loss": 1.0862,
      "step": 69
    },
    {
      "epoch": 0.09391639763531927,
      "grad_norm": 1.7204147419517413,
      "learning_rate": 9.900452025224148e-06,
      "loss": 1.0707,
      "step": 70
    },
    {
      "epoch": 0.09525806045868097,
      "grad_norm": 1.6239428850208912,
      "learning_rate": 9.896092023077866e-06,
      "loss": 1.0767,
      "step": 71
    },
    {
      "epoch": 0.09659972328204268,
      "grad_norm": 1.8341002603819232,
      "learning_rate": 9.891639578355511e-06,
      "loss": 1.0725,
      "step": 72
    },
    {
      "epoch": 0.09794138610540438,
      "grad_norm": 1.8558854123237023,
      "learning_rate": 9.887094775123203e-06,
      "loss": 1.0863,
      "step": 73
    },
    {
      "epoch": 0.09928304892876609,
      "grad_norm": 1.622711136214513,
      "learning_rate": 9.882457699190874e-06,
      "loss": 1.0702,
      "step": 74
    },
    {
      "epoch": 0.1006247117521278,
      "grad_norm": 1.495142178664832,
      "learning_rate": 9.877728438110645e-06,
      "loss": 1.0701,
      "step": 75
    },
    {
      "epoch": 0.1019663745754895,
      "grad_norm": 1.4506371213965268,
      "learning_rate": 9.872907081175175e-06,
      "loss": 1.0829,
      "step": 76
    },
    {
      "epoch": 0.10330803739885121,
      "grad_norm": 1.8034795976596587,
      "learning_rate": 9.867993719415974e-06,
      "loss": 1.0654,
      "step": 77
    },
    {
      "epoch": 0.1046497002222129,
      "grad_norm": 1.785781379811563,
      "learning_rate": 9.86298844560169e-06,
      "loss": 1.0514,
      "step": 78
    },
    {
      "epoch": 0.1059913630455746,
      "grad_norm": 1.7535465757365365,
      "learning_rate": 9.857891354236342e-06,
      "loss": 1.0564,
      "step": 79
    },
    {
      "epoch": 0.10733302586893631,
      "grad_norm": 1.7337828430288411,
      "learning_rate": 9.852702541557559e-06,
      "loss": 1.0683,
      "step": 80
    },
    {
      "epoch": 0.10867468869229802,
      "grad_norm": 1.6693391553415382,
      "learning_rate": 9.847422105534739e-06,
      "loss": 1.0431,
      "step": 81
    },
    {
      "epoch": 0.11001635151565972,
      "grad_norm": 1.474045813243068,
      "learning_rate": 9.842050145867219e-06,
      "loss": 1.0564,
      "step": 82
    },
    {
      "epoch": 0.11135801433902143,
      "grad_norm": 2.7863753262947206,
      "learning_rate": 9.836586763982376e-06,
      "loss": 1.0567,
      "step": 83
    },
    {
      "epoch": 0.11269967716238313,
      "grad_norm": 1.5338677331929418,
      "learning_rate": 9.831032063033726e-06,
      "loss": 1.0308,
      "step": 84
    },
    {
      "epoch": 0.11404133998574484,
      "grad_norm": 2.179252254257764,
      "learning_rate": 9.825386147898967e-06,
      "loss": 1.0653,
      "step": 85
    },
    {
      "epoch": 0.11538300280910653,
      "grad_norm": 1.6643070967595677,
      "learning_rate": 9.819649125178004e-06,
      "loss": 1.0586,
      "step": 86
    },
    {
      "epoch": 0.11672466563246824,
      "grad_norm": 1.734693819240776,
      "learning_rate": 9.813821103190932e-06,
      "loss": 1.0671,
      "step": 87
    },
    {
      "epoch": 0.11806632845582994,
      "grad_norm": 1.9047365235771927,
      "learning_rate": 9.807902191975996e-06,
      "loss": 1.0563,
      "step": 88
    },
    {
      "epoch": 0.11940799127919165,
      "grad_norm": 1.7159087250602358,
      "learning_rate": 9.801892503287507e-06,
      "loss": 1.049,
      "step": 89
    },
    {
      "epoch": 0.12074965410255335,
      "grad_norm": 1.7365191389572339,
      "learning_rate": 9.795792150593739e-06,
      "loss": 1.0668,
      "step": 90
    },
    {
      "epoch": 0.12209131692591506,
      "grad_norm": 1.61649848507611,
      "learning_rate": 9.789601249074781e-06,
      "loss": 1.0365,
      "step": 91
    },
    {
      "epoch": 0.12343297974927676,
      "grad_norm": 1.525103477384651,
      "learning_rate": 9.783319915620365e-06,
      "loss": 1.0753,
      "step": 92
    },
    {
      "epoch": 0.12477464257263847,
      "grad_norm": 1.5468595731679113,
      "learning_rate": 9.776948268827658e-06,
      "loss": 1.0377,
      "step": 93
    },
    {
      "epoch": 0.12611630539600016,
      "grad_norm": 1.5143092484316483,
      "learning_rate": 9.770486428999026e-06,
      "loss": 1.0398,
      "step": 94
    },
    {
      "epoch": 0.12745796821936187,
      "grad_norm": 2.098110190522861,
      "learning_rate": 9.763934518139754e-06,
      "loss": 1.0321,
      "step": 95
    },
    {
      "epoch": 0.12879963104272357,
      "grad_norm": 1.596350127992628,
      "learning_rate": 9.757292659955755e-06,
      "loss": 1.0465,
      "step": 96
    },
    {
      "epoch": 0.13014129386608528,
      "grad_norm": 1.5349488332845023,
      "learning_rate": 9.750560979851222e-06,
      "loss": 1.0405,
      "step": 97
    },
    {
      "epoch": 0.13148295668944698,
      "grad_norm": 1.6094024991622389,
      "learning_rate": 9.743739604926268e-06,
      "loss": 1.0397,
      "step": 98
    },
    {
      "epoch": 0.1328246195128087,
      "grad_norm": 1.4888314569282421,
      "learning_rate": 9.736828663974527e-06,
      "loss": 1.026,
      "step": 99
    },
    {
      "epoch": 0.1341662823361704,
      "grad_norm": 1.5305760875880698,
      "learning_rate": 9.729828287480713e-06,
      "loss": 1.04,
      "step": 100
    },
    {
      "epoch": 0.1355079451595321,
      "grad_norm": 1.6171171257521297,
      "learning_rate": 9.722738607618171e-06,
      "loss": 1.0467,
      "step": 101
    },
    {
      "epoch": 0.1368496079828938,
      "grad_norm": 1.8275927631260815,
      "learning_rate": 9.715559758246363e-06,
      "loss": 1.0424,
      "step": 102
    },
    {
      "epoch": 0.1381912708062555,
      "grad_norm": 1.5808349943913245,
      "learning_rate": 9.70829187490836e-06,
      "loss": 1.052,
      "step": 103
    },
    {
      "epoch": 0.13953293362961722,
      "grad_norm": 1.599275155249567,
      "learning_rate": 9.700935094828267e-06,
      "loss": 1.0433,
      "step": 104
    },
    {
      "epoch": 0.14087459645297892,
      "grad_norm": 1.4748747611236446,
      "learning_rate": 9.693489556908641e-06,
      "loss": 1.0512,
      "step": 105
    },
    {
      "epoch": 0.1422162592763406,
      "grad_norm": 1.6060077466070208,
      "learning_rate": 9.68595540172787e-06,
      "loss": 1.01,
      "step": 106
    },
    {
      "epoch": 0.1435579220997023,
      "grad_norm": 1.8066414228236412,
      "learning_rate": 9.678332771537506e-06,
      "loss": 1.0172,
      "step": 107
    },
    {
      "epoch": 0.144899584923064,
      "grad_norm": 1.5192038597826958,
      "learning_rate": 9.670621810259596e-06,
      "loss": 1.0243,
      "step": 108
    },
    {
      "epoch": 0.14624124774642572,
      "grad_norm": 1.8566706929318617,
      "learning_rate": 9.662822663483952e-06,
      "loss": 1.0386,
      "step": 109
    },
    {
      "epoch": 0.14758291056978742,
      "grad_norm": 1.5109073208750536,
      "learning_rate": 9.654935478465409e-06,
      "loss": 1.0134,
      "step": 110
    },
    {
      "epoch": 0.14892457339314913,
      "grad_norm": 1.5248271536361229,
      "learning_rate": 9.646960404121042e-06,
      "loss": 1.0294,
      "step": 111
    },
    {
      "epoch": 0.15026623621651083,
      "grad_norm": 1.4414336657282556,
      "learning_rate": 9.638897591027355e-06,
      "loss": 1.0259,
      "step": 112
    },
    {
      "epoch": 0.15160789903987254,
      "grad_norm": 1.5329634677484316,
      "learning_rate": 9.630747191417433e-06,
      "loss": 1.0087,
      "step": 113
    },
    {
      "epoch": 0.15294956186323425,
      "grad_norm": 1.7617577737233274,
      "learning_rate": 9.62250935917808e-06,
      "loss": 1.022,
      "step": 114
    },
    {
      "epoch": 0.15429122468659595,
      "grad_norm": 1.492081973940454,
      "learning_rate": 9.614184249846903e-06,
      "loss": 1.0083,
      "step": 115
    },
    {
      "epoch": 0.15563288750995766,
      "grad_norm": 1.6947020263672135,
      "learning_rate": 9.605772020609376e-06,
      "loss": 1.0448,
      "step": 116
    },
    {
      "epoch": 0.15697455033331936,
      "grad_norm": 1.517976041310006,
      "learning_rate": 9.597272830295877e-06,
      "loss": 1.0273,
      "step": 117
    },
    {
      "epoch": 0.15831621315668107,
      "grad_norm": 1.6697358355322918,
      "learning_rate": 9.58868683937868e-06,
      "loss": 1.0388,
      "step": 118
    },
    {
      "epoch": 0.15965787598004277,
      "grad_norm": 1.628205377705706,
      "learning_rate": 9.580014209968946e-06,
      "loss": 0.9951,
      "step": 119
    },
    {
      "epoch": 0.16099953880340448,
      "grad_norm": 1.5991384820195875,
      "learning_rate": 9.571255105813632e-06,
      "loss": 1.0445,
      "step": 120
    },
    {
      "epoch": 0.16234120162676619,
      "grad_norm": 1.5239720573296276,
      "learning_rate": 9.562409692292425e-06,
      "loss": 1.0476,
      "step": 121
    },
    {
      "epoch": 0.1636828644501279,
      "grad_norm": 1.5397990887970099,
      "learning_rate": 9.553478136414606e-06,
      "loss": 1.0131,
      "step": 122
    },
    {
      "epoch": 0.16502452727348957,
      "grad_norm": 1.682926751705613,
      "learning_rate": 9.544460606815901e-06,
      "loss": 1.0215,
      "step": 123
    },
    {
      "epoch": 0.16636619009685127,
      "grad_norm": 1.5374236614326178,
      "learning_rate": 9.535357273755296e-06,
      "loss": 1.022,
      "step": 124
    },
    {
      "epoch": 0.16770785292021298,
      "grad_norm": 1.5125873221214663,
      "learning_rate": 9.526168309111827e-06,
      "loss": 1.0325,
      "step": 125
    },
    {
      "epoch": 0.16904951574357469,
      "grad_norm": 1.5020363203895621,
      "learning_rate": 9.516893886381324e-06,
      "loss": 1.003,
      "step": 126
    },
    {
      "epoch": 0.1703911785669364,
      "grad_norm": 1.5253980013545938,
      "learning_rate": 9.507534180673142e-06,
      "loss": 1.0121,
      "step": 127
    },
    {
      "epoch": 0.1717328413902981,
      "grad_norm": 1.5838384344412881,
      "learning_rate": 9.498089368706862e-06,
      "loss": 1.0173,
      "step": 128
    },
    {
      "epoch": 0.1730745042136598,
      "grad_norm": 1.5526982969322647,
      "learning_rate": 9.488559628808939e-06,
      "loss": 1.0172,
      "step": 129
    },
    {
      "epoch": 0.1744161670370215,
      "grad_norm": 1.6505573103371025,
      "learning_rate": 9.478945140909346e-06,
      "loss": 0.9963,
      "step": 130
    },
    {
      "epoch": 0.1757578298603832,
      "grad_norm": 2.4994210597949533,
      "learning_rate": 9.469246086538175e-06,
      "loss": 1.0379,
      "step": 131
    },
    {
      "epoch": 0.17709949268374492,
      "grad_norm": 1.6133167321425232,
      "learning_rate": 9.459462648822209e-06,
      "loss": 1.0112,
      "step": 132
    },
    {
      "epoch": 0.17844115550710662,
      "grad_norm": 1.6250631419898731,
      "learning_rate": 9.449595012481455e-06,
      "loss": 1.0472,
      "step": 133
    },
    {
      "epoch": 0.17978281833046833,
      "grad_norm": 1.7608651873887804,
      "learning_rate": 9.439643363825678e-06,
      "loss": 1.0345,
      "step": 134
    },
    {
      "epoch": 0.18112448115383004,
      "grad_norm": 1.4462780649585327,
      "learning_rate": 9.429607890750863e-06,
      "loss": 1.0169,
      "step": 135
    },
    {
      "epoch": 0.18246614397719174,
      "grad_norm": 1.5346110788641523,
      "learning_rate": 9.419488782735671e-06,
      "loss": 1.0252,
      "step": 136
    },
    {
      "epoch": 0.18380780680055345,
      "grad_norm": 1.4786881677807142,
      "learning_rate": 9.409286230837876e-06,
      "loss": 0.9736,
      "step": 137
    },
    {
      "epoch": 0.18514946962391515,
      "grad_norm": 1.7544449884490962,
      "learning_rate": 9.399000427690736e-06,
      "loss": 1.001,
      "step": 138
    },
    {
      "epoch": 0.18649113244727683,
      "grad_norm": 1.5443323941244775,
      "learning_rate": 9.388631567499373e-06,
      "loss": 1.0171,
      "step": 139
    },
    {
      "epoch": 0.18783279527063854,
      "grad_norm": 1.5112277113219468,
      "learning_rate": 9.378179846037096e-06,
      "loss": 0.9966,
      "step": 140
    },
    {
      "epoch": 0.18917445809400024,
      "grad_norm": 1.4469628985744891,
      "learning_rate": 9.367645460641716e-06,
      "loss": 0.9999,
      "step": 141
    },
    {
      "epoch": 0.19051612091736195,
      "grad_norm": 1.509159997421376,
      "learning_rate": 9.357028610211802e-06,
      "loss": 1.0039,
      "step": 142
    },
    {
      "epoch": 0.19185778374072365,
      "grad_norm": 1.9033920658610284,
      "learning_rate": 9.346329495202944e-06,
      "loss": 0.9927,
      "step": 143
    },
    {
      "epoch": 0.19319944656408536,
      "grad_norm": 1.8697851005515826,
      "learning_rate": 9.335548317623957e-06,
      "loss": 0.9862,
      "step": 144
    },
    {
      "epoch": 0.19454110938744706,
      "grad_norm": 1.727575777874671,
      "learning_rate": 9.324685281033073e-06,
      "loss": 0.9985,
      "step": 145
    },
    {
      "epoch": 0.19588277221080877,
      "grad_norm": 1.4973664146911145,
      "learning_rate": 9.313740590534093e-06,
      "loss": 1.0107,
      "step": 146
    },
    {
      "epoch": 0.19722443503417048,
      "grad_norm": 1.5511498919072226,
      "learning_rate": 9.302714452772515e-06,
      "loss": 1.0211,
      "step": 147
    },
    {
      "epoch": 0.19856609785753218,
      "grad_norm": 1.7549997830763724,
      "learning_rate": 9.291607075931638e-06,
      "loss": 1.0127,
      "step": 148
    },
    {
      "epoch": 0.1999077606808939,
      "grad_norm": 1.5905298977716988,
      "learning_rate": 9.280418669728625e-06,
      "loss": 0.9755,
      "step": 149
    },
    {
      "epoch": 0.2012494235042556,
      "grad_norm": 1.7118974196904215,
      "learning_rate": 9.269149445410545e-06,
      "loss": 0.9962,
      "step": 150
    },
    {
      "epoch": 0.2025910863276173,
      "grad_norm": 1.7000772782219813,
      "learning_rate": 9.257799615750384e-06,
      "loss": 1.017,
      "step": 151
    },
    {
      "epoch": 0.203932749150979,
      "grad_norm": 1.9810883869642395,
      "learning_rate": 9.246369395043033e-06,
      "loss": 0.996,
      "step": 152
    },
    {
      "epoch": 0.2052744119743407,
      "grad_norm": 1.5890509537547852,
      "learning_rate": 9.234858999101232e-06,
      "loss": 1.0012,
      "step": 153
    },
    {
      "epoch": 0.20661607479770241,
      "grad_norm": 1.5976972038680295,
      "learning_rate": 9.223268645251504e-06,
      "loss": 0.978,
      "step": 154
    },
    {
      "epoch": 0.2079577376210641,
      "grad_norm": 1.5846128874339207,
      "learning_rate": 9.211598552330047e-06,
      "loss": 1.015,
      "step": 155
    },
    {
      "epoch": 0.2092994004444258,
      "grad_norm": 1.6624623820573463,
      "learning_rate": 9.199848940678607e-06,
      "loss": 0.9916,
      "step": 156
    },
    {
      "epoch": 0.2106410632677875,
      "grad_norm": 1.519781144208935,
      "learning_rate": 9.188020032140308e-06,
      "loss": 0.9759,
      "step": 157
    },
    {
      "epoch": 0.2119827260911492,
      "grad_norm": 1.7030331960727392,
      "learning_rate": 9.17611205005548e-06,
      "loss": 1.0027,
      "step": 158
    },
    {
      "epoch": 0.21332438891451092,
      "grad_norm": 2.3227823126647986,
      "learning_rate": 9.164125219257419e-06,
      "loss": 1.0084,
      "step": 159
    },
    {
      "epoch": 0.21466605173787262,
      "grad_norm": 1.5950630942235617,
      "learning_rate": 9.152059766068164e-06,
      "loss": 1.0171,
      "step": 160
    },
    {
      "epoch": 0.21600771456123433,
      "grad_norm": 1.5097514171757036,
      "learning_rate": 9.139915918294213e-06,
      "loss": 1.0052,
      "step": 161
    },
    {
      "epoch": 0.21734937738459603,
      "grad_norm": 1.668749714662694,
      "learning_rate": 9.127693905222223e-06,
      "loss": 0.9837,
      "step": 162
    },
    {
      "epoch": 0.21869104020795774,
      "grad_norm": 1.613501913429001,
      "learning_rate": 9.115393957614679e-06,
      "loss": 1.0006,
      "step": 163
    },
    {
      "epoch": 0.22003270303131944,
      "grad_norm": 1.548280495817311,
      "learning_rate": 9.103016307705546e-06,
      "loss": 1.0172,
      "step": 164
    },
    {
      "epoch": 0.22137436585468115,
      "grad_norm": 1.6799688152981618,
      "learning_rate": 9.09056118919587e-06,
      "loss": 1.0173,
      "step": 165
    },
    {
      "epoch": 0.22271602867804285,
      "grad_norm": 1.4892884530388883,
      "learning_rate": 9.078028837249378e-06,
      "loss": 0.9944,
      "step": 166
    },
    {
      "epoch": 0.22405769150140456,
      "grad_norm": 1.5421729124892896,
      "learning_rate": 9.065419488488029e-06,
      "loss": 1.013,
      "step": 167
    },
    {
      "epoch": 0.22539935432476627,
      "grad_norm": 1.8129370572182915,
      "learning_rate": 9.052733380987555e-06,
      "loss": 1.0076,
      "step": 168
    },
    {
      "epoch": 0.22674101714812797,
      "grad_norm": 1.76369666967373,
      "learning_rate": 9.03997075427296e-06,
      "loss": 1.0034,
      "step": 169
    },
    {
      "epoch": 0.22808267997148968,
      "grad_norm": 1.470527664070282,
      "learning_rate": 9.027131849313995e-06,
      "loss": 0.968,
      "step": 170
    },
    {
      "epoch": 0.22942434279485138,
      "grad_norm": 1.5038389192074497,
      "learning_rate": 9.014216908520619e-06,
      "loss": 0.9939,
      "step": 171
    },
    {
      "epoch": 0.23076600561821306,
      "grad_norm": 1.4538596123560616,
      "learning_rate": 9.001226175738409e-06,
      "loss": 0.9873,
      "step": 172
    },
    {
      "epoch": 0.23210766844157477,
      "grad_norm": 1.498912180760362,
      "learning_rate": 8.988159896243967e-06,
      "loss": 1.0123,
      "step": 173
    },
    {
      "epoch": 0.23344933126493647,
      "grad_norm": 1.4063632583533416,
      "learning_rate": 8.975018316740278e-06,
      "loss": 0.9543,
      "step": 174
    },
    {
      "epoch": 0.23479099408829818,
      "grad_norm": 1.4042961321205751,
      "learning_rate": 8.961801685352068e-06,
      "loss": 0.9941,
      "step": 175
    },
    {
      "epoch": 0.23613265691165988,
      "grad_norm": 1.5241960865995388,
      "learning_rate": 8.948510251621096e-06,
      "loss": 0.9907,
      "step": 176
    },
    {
      "epoch": 0.2374743197350216,
      "grad_norm": 1.6001855683643875,
      "learning_rate": 8.93514426650147e-06,
      "loss": 0.9726,
      "step": 177
    },
    {
      "epoch": 0.2388159825583833,
      "grad_norm": 1.6999875656481869,
      "learning_rate": 8.921703982354886e-06,
      "loss": 0.9979,
      "step": 178
    },
    {
      "epoch": 0.240157645381745,
      "grad_norm": 1.5296332672890027,
      "learning_rate": 8.908189652945871e-06,
      "loss": 1.0019,
      "step": 179
    },
    {
      "epoch": 0.2414993082051067,
      "grad_norm": 1.664067570536287,
      "learning_rate": 8.894601533437e-06,
      "loss": 0.9698,
      "step": 180
    },
    {
      "epoch": 0.2428409710284684,
      "grad_norm": 2.102340889974168,
      "learning_rate": 8.880939880384061e-06,
      "loss": 0.999,
      "step": 181
    },
    {
      "epoch": 0.24418263385183012,
      "grad_norm": 1.7753482498612518,
      "learning_rate": 8.867204951731227e-06,
      "loss": 0.9975,
      "step": 182
    },
    {
      "epoch": 0.24552429667519182,
      "grad_norm": 1.6954057018557689,
      "learning_rate": 8.853397006806183e-06,
      "loss": 0.9595,
      "step": 183
    },
    {
      "epoch": 0.24686595949855353,
      "grad_norm": 1.6220416310433559,
      "learning_rate": 8.839516306315216e-06,
      "loss": 0.9772,
      "step": 184
    },
    {
      "epoch": 0.24820762232191523,
      "grad_norm": 1.4344699633715252,
      "learning_rate": 8.825563112338308e-06,
      "loss": 0.9915,
      "step": 185
    },
    {
      "epoch": 0.24954928514527694,
      "grad_norm": 1.5652210870327667,
      "learning_rate": 8.811537688324187e-06,
      "loss": 0.9884,
      "step": 186
    },
    {
      "epoch": 0.25089094796863864,
      "grad_norm": 1.5480421042178967,
      "learning_rate": 8.797440299085344e-06,
      "loss": 0.9912,
      "step": 187
    },
    {
      "epoch": 0.2522326107920003,
      "grad_norm": 1.5815223622058714,
      "learning_rate": 8.783271210793034e-06,
      "loss": 0.9712,
      "step": 188
    },
    {
      "epoch": 0.25357427361536206,
      "grad_norm": 1.6906168223241804,
      "learning_rate": 8.769030690972262e-06,
      "loss": 0.9751,
      "step": 189
    },
    {
      "epoch": 0.25491593643872373,
      "grad_norm": 1.6673317361003395,
      "learning_rate": 8.754719008496718e-06,
      "loss": 0.9857,
      "step": 190
    },
    {
      "epoch": 0.25625759926208547,
      "grad_norm": 1.5742573500825576,
      "learning_rate": 8.740336433583704e-06,
      "loss": 0.988,
      "step": 191
    },
    {
      "epoch": 0.25759926208544714,
      "grad_norm": 1.5087573984013583,
      "learning_rate": 8.725883237789046e-06,
      "loss": 0.987,
      "step": 192
    },
    {
      "epoch": 0.2589409249088089,
      "grad_norm": 1.6052629099050266,
      "learning_rate": 8.711359694001941e-06,
      "loss": 0.9853,
      "step": 193
    },
    {
      "epoch": 0.26028258773217056,
      "grad_norm": 16.444996469677516,
      "learning_rate": 8.696766076439826e-06,
      "loss": 0.9776,
      "step": 194
    },
    {
      "epoch": 0.2616242505555323,
      "grad_norm": 1.9102037969030812,
      "learning_rate": 8.682102660643196e-06,
      "loss": 1.0172,
      "step": 195
    },
    {
      "epoch": 0.26296591337889397,
      "grad_norm": 2.9756746121109314,
      "learning_rate": 8.667369723470393e-06,
      "loss": 0.9947,
      "step": 196
    },
    {
      "epoch": 0.26430757620225565,
      "grad_norm": 1.6204007709317712,
      "learning_rate": 8.652567543092385e-06,
      "loss": 0.9799,
      "step": 197
    },
    {
      "epoch": 0.2656492390256174,
      "grad_norm": 1.5303211756166764,
      "learning_rate": 8.637696398987517e-06,
      "loss": 0.9783,
      "step": 198
    },
    {
      "epoch": 0.26699090184897906,
      "grad_norm": 1.6840365379900046,
      "learning_rate": 8.622756571936229e-06,
      "loss": 0.9736,
      "step": 199
    },
    {
      "epoch": 0.2683325646723408,
      "grad_norm": 1.683788142218529,
      "learning_rate": 8.607748344015752e-06,
      "loss": 0.9702,
      "step": 200
    },
    {
      "epoch": 0.26967422749570247,
      "grad_norm": 1.457488972413562,
      "learning_rate": 8.592671998594794e-06,
      "loss": 0.9806,
      "step": 201
    },
    {
      "epoch": 0.2710158903190642,
      "grad_norm": 1.547971757217976,
      "learning_rate": 8.577527820328176e-06,
      "loss": 0.9634,
      "step": 202
    },
    {
      "epoch": 0.2723575531424259,
      "grad_norm": 1.3316553082460927,
      "learning_rate": 8.562316095151468e-06,
      "loss": 0.9512,
      "step": 203
    },
    {
      "epoch": 0.2736992159657876,
      "grad_norm": 1.5661643370029357,
      "learning_rate": 8.54703711027558e-06,
      "loss": 0.9855,
      "step": 204
    },
    {
      "epoch": 0.2750408787891493,
      "grad_norm": 1.4810435360919245,
      "learning_rate": 8.53169115418135e-06,
      "loss": 0.9658,
      "step": 205
    },
    {
      "epoch": 0.276382541612511,
      "grad_norm": 1.5394045600056434,
      "learning_rate": 8.516278516614092e-06,
      "loss": 0.9734,
      "step": 206
    },
    {
      "epoch": 0.2777242044358727,
      "grad_norm": 1.4871363595645701,
      "learning_rate": 8.50079948857812e-06,
      "loss": 0.9701,
      "step": 207
    },
    {
      "epoch": 0.27906586725923443,
      "grad_norm": 1.6508266936645926,
      "learning_rate": 8.485254362331268e-06,
      "loss": 0.9891,
      "step": 208
    },
    {
      "epoch": 0.2804075300825961,
      "grad_norm": 1.4599758131575469,
      "learning_rate": 8.469643431379355e-06,
      "loss": 0.9689,
      "step": 209
    },
    {
      "epoch": 0.28174919290595785,
      "grad_norm": 1.4858135833977075,
      "learning_rate": 8.453966990470656e-06,
      "loss": 0.9735,
      "step": 210
    },
    {
      "epoch": 0.2830908557293195,
      "grad_norm": 1.5170470501397098,
      "learning_rate": 8.438225335590334e-06,
      "loss": 0.9745,
      "step": 211
    },
    {
      "epoch": 0.2844325185526812,
      "grad_norm": 1.5641985728172763,
      "learning_rate": 8.422418763954841e-06,
      "loss": 0.9618,
      "step": 212
    },
    {
      "epoch": 0.28577418137604294,
      "grad_norm": 1.6783440611678353,
      "learning_rate": 8.406547574006326e-06,
      "loss": 0.9783,
      "step": 213
    },
    {
      "epoch": 0.2871158441994046,
      "grad_norm": 1.579120726370794,
      "learning_rate": 8.390612065406983e-06,
      "loss": 0.9567,
      "step": 214
    },
    {
      "epoch": 0.28845750702276635,
      "grad_norm": 1.5939955792145337,
      "learning_rate": 8.374612539033398e-06,
      "loss": 0.9746,
      "step": 215
    },
    {
      "epoch": 0.289799169846128,
      "grad_norm": 1.4721387285954308,
      "learning_rate": 8.358549296970877e-06,
      "loss": 0.9691,
      "step": 216
    },
    {
      "epoch": 0.29114083266948976,
      "grad_norm": 1.5457271723104407,
      "learning_rate": 8.342422642507727e-06,
      "loss": 0.9878,
      "step": 217
    },
    {
      "epoch": 0.29248249549285144,
      "grad_norm": 1.3692684340326473,
      "learning_rate": 8.32623288012954e-06,
      "loss": 0.9594,
      "step": 218
    },
    {
      "epoch": 0.29382415831621317,
      "grad_norm": 1.5163724984197753,
      "learning_rate": 8.309980315513444e-06,
      "loss": 0.972,
      "step": 219
    },
    {
      "epoch": 0.29516582113957485,
      "grad_norm": 1.6114780037861143,
      "learning_rate": 8.29366525552233e-06,
      "loss": 0.9603,
      "step": 220
    },
    {
      "epoch": 0.2965074839629366,
      "grad_norm": 1.4762572081466976,
      "learning_rate": 8.27728800819905e-06,
      "loss": 0.9727,
      "step": 221
    },
    {
      "epoch": 0.29784914678629826,
      "grad_norm": 1.4684362587718065,
      "learning_rate": 8.260848882760616e-06,
      "loss": 0.9785,
      "step": 222
    },
    {
      "epoch": 0.29919080960966,
      "grad_norm": 1.5477609027947252,
      "learning_rate": 8.24434818959235e-06,
      "loss": 0.9457,
      "step": 223
    },
    {
      "epoch": 0.30053247243302167,
      "grad_norm": 1.4186396792517242,
      "learning_rate": 8.22778624024203e-06,
      "loss": 0.9743,
      "step": 224
    },
    {
      "epoch": 0.3018741352563834,
      "grad_norm": 1.507163119567676,
      "learning_rate": 8.211163347414005e-06,
      "loss": 0.9723,
      "step": 225
    },
    {
      "epoch": 0.3032157980797451,
      "grad_norm": 2.061573031593893,
      "learning_rate": 8.194479824963284e-06,
      "loss": 0.9695,
      "step": 226
    },
    {
      "epoch": 0.3045574609031068,
      "grad_norm": 1.7101381416686336,
      "learning_rate": 8.177735987889628e-06,
      "loss": 0.9639,
      "step": 227
    },
    {
      "epoch": 0.3058991237264685,
      "grad_norm": 1.4858650389281058,
      "learning_rate": 8.160932152331587e-06,
      "loss": 0.9707,
      "step": 228
    },
    {
      "epoch": 0.30724078654983017,
      "grad_norm": 1.6965951272902517,
      "learning_rate": 8.144068635560533e-06,
      "loss": 0.9893,
      "step": 229
    },
    {
      "epoch": 0.3085824493731919,
      "grad_norm": 1.388920287350121,
      "learning_rate": 8.127145755974676e-06,
      "loss": 0.9717,
      "step": 230
    },
    {
      "epoch": 0.3099241121965536,
      "grad_norm": 1.5510344042881328,
      "learning_rate": 8.11016383309305e-06,
      "loss": 0.9637,
      "step": 231
    },
    {
      "epoch": 0.3112657750199153,
      "grad_norm": 1.3951378486816117,
      "learning_rate": 8.093123187549475e-06,
      "loss": 0.9798,
      "step": 232
    },
    {
      "epoch": 0.312607437843277,
      "grad_norm": 1.3958446772921527,
      "learning_rate": 8.07602414108651e-06,
      "loss": 0.9683,
      "step": 233
    },
    {
      "epoch": 0.3139491006666387,
      "grad_norm": 1.466094331759861,
      "learning_rate": 8.058867016549372e-06,
      "loss": 0.9707,
      "step": 234
    },
    {
      "epoch": 0.3152907634900004,
      "grad_norm": 1.8036674074115169,
      "learning_rate": 8.04165213787985e-06,
      "loss": 0.9489,
      "step": 235
    },
    {
      "epoch": 0.31663242631336214,
      "grad_norm": 1.5957322487948662,
      "learning_rate": 8.024379830110176e-06,
      "loss": 0.9452,
      "step": 236
    },
    {
      "epoch": 0.3179740891367238,
      "grad_norm": 1.4680343088229408,
      "learning_rate": 8.007050419356898e-06,
      "loss": 0.9683,
      "step": 237
    },
    {
      "epoch": 0.31931575196008555,
      "grad_norm": 1.5328747678300108,
      "learning_rate": 7.989664232814718e-06,
      "loss": 0.9741,
      "step": 238
    },
    {
      "epoch": 0.3206574147834472,
      "grad_norm": 1.374414004947906,
      "learning_rate": 7.972221598750316e-06,
      "loss": 0.9535,
      "step": 239
    },
    {
      "epoch": 0.32199907760680896,
      "grad_norm": 1.3346744700214859,
      "learning_rate": 7.95472284649615e-06,
      "loss": 0.9654,
      "step": 240
    },
    {
      "epoch": 0.32334074043017064,
      "grad_norm": 1.4669968920031307,
      "learning_rate": 7.937168306444243e-06,
      "loss": 0.9507,
      "step": 241
    },
    {
      "epoch": 0.32468240325353237,
      "grad_norm": 1.5766148114998875,
      "learning_rate": 7.919558310039937e-06,
      "loss": 0.965,
      "step": 242
    },
    {
      "epoch": 0.32602406607689405,
      "grad_norm": 1.4875838167859463,
      "learning_rate": 7.90189318977564e-06,
      "loss": 0.9717,
      "step": 243
    },
    {
      "epoch": 0.3273657289002558,
      "grad_norm": 1.7148717900145576,
      "learning_rate": 7.884173279184552e-06,
      "loss": 0.9787,
      "step": 244
    },
    {
      "epoch": 0.32870739172361746,
      "grad_norm": 1.6903792913472557,
      "learning_rate": 7.866398912834357e-06,
      "loss": 0.981,
      "step": 245
    },
    {
      "epoch": 0.33004905454697914,
      "grad_norm": 1.5756216364889237,
      "learning_rate": 7.848570426320918e-06,
      "loss": 0.9705,
      "step": 246
    },
    {
      "epoch": 0.33139071737034087,
      "grad_norm": 1.3681608333123005,
      "learning_rate": 7.830688156261927e-06,
      "loss": 0.9794,
      "step": 247
    },
    {
      "epoch": 0.33273238019370255,
      "grad_norm": 1.4816370422331064,
      "learning_rate": 7.812752440290568e-06,
      "loss": 0.9524,
      "step": 248
    },
    {
      "epoch": 0.3340740430170643,
      "grad_norm": 1.5218600193401652,
      "learning_rate": 7.794763617049124e-06,
      "loss": 0.9618,
      "step": 249
    },
    {
      "epoch": 0.33541570584042596,
      "grad_norm": 1.4743104016838262,
      "learning_rate": 7.776722026182588e-06,
      "loss": 0.9744,
      "step": 250
    },
    {
      "epoch": 0.3367573686637877,
      "grad_norm": 1.5788453081737652,
      "learning_rate": 7.758628008332262e-06,
      "loss": 0.9665,
      "step": 251
    },
    {
      "epoch": 0.33809903148714937,
      "grad_norm": 1.5865308150096813,
      "learning_rate": 7.740481905129307e-06,
      "loss": 0.9609,
      "step": 252
    },
    {
      "epoch": 0.3394406943105111,
      "grad_norm": 1.3752322651899374,
      "learning_rate": 7.722284059188308e-06,
      "loss": 0.9454,
      "step": 253
    },
    {
      "epoch": 0.3407823571338728,
      "grad_norm": 1.5628688780654367,
      "learning_rate": 7.704034814100798e-06,
      "loss": 0.9453,
      "step": 254
    },
    {
      "epoch": 0.3421240199572345,
      "grad_norm": 1.3995041103502648,
      "learning_rate": 7.685734514428767e-06,
      "loss": 0.9712,
      "step": 255
    },
    {
      "epoch": 0.3434656827805962,
      "grad_norm": 1.5469450835726162,
      "learning_rate": 7.667383505698165e-06,
      "loss": 0.9602,
      "step": 256
    },
    {
      "epoch": 0.3448073456039579,
      "grad_norm": 1.4254233488403434,
      "learning_rate": 7.648982134392378e-06,
      "loss": 0.9476,
      "step": 257
    },
    {
      "epoch": 0.3461490084273196,
      "grad_norm": 1.4496312090906014,
      "learning_rate": 7.630530747945672e-06,
      "loss": 0.9447,
      "step": 258
    },
    {
      "epoch": 0.34749067125068134,
      "grad_norm": 1.5432996608177838,
      "learning_rate": 7.612029694736652e-06,
      "loss": 0.953,
      "step": 259
    },
    {
      "epoch": 0.348832334074043,
      "grad_norm": 1.4528733449175,
      "learning_rate": 7.593479324081672e-06,
      "loss": 0.9779,
      "step": 260
    },
    {
      "epoch": 0.3501739968974047,
      "grad_norm": 1.585347953236908,
      "learning_rate": 7.574879986228245e-06,
      "loss": 0.9596,
      "step": 261
    },
    {
      "epoch": 0.3515156597207664,
      "grad_norm": 2.2500708161081877,
      "learning_rate": 7.556232032348429e-06,
      "loss": 0.9627,
      "step": 262
    },
    {
      "epoch": 0.3528573225441281,
      "grad_norm": 1.5763755196223852,
      "learning_rate": 7.537535814532191e-06,
      "loss": 0.9607,
      "step": 263
    },
    {
      "epoch": 0.35419898536748984,
      "grad_norm": 1.771846845155408,
      "learning_rate": 7.518791685780769e-06,
      "loss": 0.9434,
      "step": 264
    },
    {
      "epoch": 0.3555406481908515,
      "grad_norm": 1.5224566041251197,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.9691,
      "step": 265
    },
    {
      "epoch": 0.35688231101421325,
      "grad_norm": 1.4640993296882345,
      "learning_rate": 7.481161111993642e-06,
      "loss": 0.9595,
      "step": 266
    },
    {
      "epoch": 0.3582239738375749,
      "grad_norm": 1.6809152959115363,
      "learning_rate": 7.462275377456671e-06,
      "loss": 0.9563,
      "step": 267
    },
    {
      "epoch": 0.35956563666093666,
      "grad_norm": 1.6675443282439029,
      "learning_rate": 7.443343152968566e-06,
      "loss": 0.9785,
      "step": 268
    },
    {
      "epoch": 0.36090729948429834,
      "grad_norm": 1.4792935920603045,
      "learning_rate": 7.424364795986582e-06,
      "loss": 0.9609,
      "step": 269
    },
    {
      "epoch": 0.36224896230766007,
      "grad_norm": 1.5483209952088646,
      "learning_rate": 7.405340664838994e-06,
      "loss": 0.9485,
      "step": 270
    },
    {
      "epoch": 0.36359062513102175,
      "grad_norm": 1.4340034591760151,
      "learning_rate": 7.38627111871833e-06,
      "loss": 0.9534,
      "step": 271
    },
    {
      "epoch": 0.3649322879543835,
      "grad_norm": 1.613866321606026,
      "learning_rate": 7.3671565176746025e-06,
      "loss": 0.9725,
      "step": 272
    },
    {
      "epoch": 0.36627395077774516,
      "grad_norm": 1.4824243681022105,
      "learning_rate": 7.3479972226084925e-06,
      "loss": 0.958,
      "step": 273
    },
    {
      "epoch": 0.3676156136011069,
      "grad_norm": 1.4427286368890009,
      "learning_rate": 7.328793595264549e-06,
      "loss": 0.9435,
      "step": 274
    },
    {
      "epoch": 0.36895727642446857,
      "grad_norm": 1.3805561786232567,
      "learning_rate": 7.309545998224351e-06,
      "loss": 0.9458,
      "step": 275
    },
    {
      "epoch": 0.3702989392478303,
      "grad_norm": 1.8516381642259134,
      "learning_rate": 7.290254794899665e-06,
      "loss": 0.9678,
      "step": 276
    },
    {
      "epoch": 0.371640602071192,
      "grad_norm": 1.5815466257148956,
      "learning_rate": 7.270920349525584e-06,
      "loss": 0.9599,
      "step": 277
    },
    {
      "epoch": 0.37298226489455366,
      "grad_norm": 1.3960905267997115,
      "learning_rate": 7.2515430271536495e-06,
      "loss": 0.9484,
      "step": 278
    },
    {
      "epoch": 0.3743239277179154,
      "grad_norm": 1.5277524276047343,
      "learning_rate": 7.232123193644957e-06,
      "loss": 0.9603,
      "step": 279
    },
    {
      "epoch": 0.3756655905412771,
      "grad_norm": 1.3938189518246926,
      "learning_rate": 7.212661215663252e-06,
      "loss": 0.9435,
      "step": 280
    },
    {
      "epoch": 0.3770072533646388,
      "grad_norm": 1.5608424878074654,
      "learning_rate": 7.193157460668005e-06,
      "loss": 0.9385,
      "step": 281
    },
    {
      "epoch": 0.3783489161880005,
      "grad_norm": 1.3994956171633623,
      "learning_rate": 7.173612296907473e-06,
      "loss": 0.9508,
      "step": 282
    },
    {
      "epoch": 0.3796905790113622,
      "grad_norm": 1.4666586737314051,
      "learning_rate": 7.154026093411747e-06,
      "loss": 0.9671,
      "step": 283
    },
    {
      "epoch": 0.3810322418347239,
      "grad_norm": 1.4337201807932214,
      "learning_rate": 7.134399219985786e-06,
      "loss": 0.9666,
      "step": 284
    },
    {
      "epoch": 0.38237390465808563,
      "grad_norm": 1.4728179477625383,
      "learning_rate": 7.114732047202433e-06,
      "loss": 0.9671,
      "step": 285
    },
    {
      "epoch": 0.3837155674814473,
      "grad_norm": 1.360568477796925,
      "learning_rate": 7.095024946395419e-06,
      "loss": 0.9571,
      "step": 286
    },
    {
      "epoch": 0.38505723030480904,
      "grad_norm": 1.5631552939635314,
      "learning_rate": 7.075278289652349e-06,
      "loss": 0.9562,
      "step": 287
    },
    {
      "epoch": 0.3863988931281707,
      "grad_norm": 1.566734610660415,
      "learning_rate": 7.055492449807684e-06,
      "loss": 0.9368,
      "step": 288
    },
    {
      "epoch": 0.38774055595153245,
      "grad_norm": 1.4515435817786102,
      "learning_rate": 7.035667800435694e-06,
      "loss": 0.9416,
      "step": 289
    },
    {
      "epoch": 0.38908221877489413,
      "grad_norm": 1.5035435190437387,
      "learning_rate": 7.015804715843404e-06,
      "loss": 0.9462,
      "step": 290
    },
    {
      "epoch": 0.39042388159825586,
      "grad_norm": 1.6249008432001628,
      "learning_rate": 6.995903571063541e-06,
      "loss": 0.9507,
      "step": 291
    },
    {
      "epoch": 0.39176554442161754,
      "grad_norm": 1.5168673161442563,
      "learning_rate": 6.975964741847427e-06,
      "loss": 0.9626,
      "step": 292
    },
    {
      "epoch": 0.3931072072449793,
      "grad_norm": 1.4864183283216745,
      "learning_rate": 6.955988604657914e-06,
      "loss": 0.9436,
      "step": 293
    },
    {
      "epoch": 0.39444887006834095,
      "grad_norm": 1.398353146411846,
      "learning_rate": 6.935975536662254e-06,
      "loss": 0.943,
      "step": 294
    },
    {
      "epoch": 0.39579053289170263,
      "grad_norm": 1.4161271859037246,
      "learning_rate": 6.915925915724987e-06,
      "loss": 0.928,
      "step": 295
    },
    {
      "epoch": 0.39713219571506436,
      "grad_norm": 1.6833911607161969,
      "learning_rate": 6.895840120400804e-06,
      "loss": 0.9329,
      "step": 296
    },
    {
      "epoch": 0.39847385853842604,
      "grad_norm": 1.487710183205637,
      "learning_rate": 6.875718529927404e-06,
      "loss": 0.9466,
      "step": 297
    },
    {
      "epoch": 0.3998155213617878,
      "grad_norm": 1.549467448109303,
      "learning_rate": 6.855561524218326e-06,
      "loss": 0.9821,
      "step": 298
    },
    {
      "epoch": 0.40115718418514945,
      "grad_norm": 1.3745294880173118,
      "learning_rate": 6.835369483855783e-06,
      "loss": 0.9161,
      "step": 299
    },
    {
      "epoch": 0.4024988470085112,
      "grad_norm": 1.4224517987313379,
      "learning_rate": 6.815142790083473e-06,
      "loss": 0.9469,
      "step": 300
    },
    {
      "epoch": 0.40384050983187286,
      "grad_norm": 1.5056244514305563,
      "learning_rate": 6.79488182479938e-06,
      "loss": 0.9662,
      "step": 301
    },
    {
      "epoch": 0.4051821726552346,
      "grad_norm": 1.4054898081789005,
      "learning_rate": 6.774586970548567e-06,
      "loss": 0.9489,
      "step": 302
    },
    {
      "epoch": 0.4065238354785963,
      "grad_norm": 1.4511626851951822,
      "learning_rate": 6.754258610515949e-06,
      "loss": 0.9496,
      "step": 303
    },
    {
      "epoch": 0.407865498301958,
      "grad_norm": 3.3319586578353655,
      "learning_rate": 6.73389712851906e-06,
      "loss": 0.9405,
      "step": 304
    },
    {
      "epoch": 0.4092071611253197,
      "grad_norm": 2.1898520431778246,
      "learning_rate": 6.713502909000808e-06,
      "loss": 0.923,
      "step": 305
    },
    {
      "epoch": 0.4105488239486814,
      "grad_norm": 1.4488374448148915,
      "learning_rate": 6.6930763370222104e-06,
      "loss": 0.9751,
      "step": 306
    },
    {
      "epoch": 0.4118904867720431,
      "grad_norm": 1.4191601427296872,
      "learning_rate": 6.672617798255135e-06,
      "loss": 0.9398,
      "step": 307
    },
    {
      "epoch": 0.41323214959540483,
      "grad_norm": 1.428689176709015,
      "learning_rate": 6.652127678975002e-06,
      "loss": 0.941,
      "step": 308
    },
    {
      "epoch": 0.4145738124187665,
      "grad_norm": 1.4795595116700593,
      "learning_rate": 6.631606366053507e-06,
      "loss": 0.9466,
      "step": 309
    },
    {
      "epoch": 0.4159154752421282,
      "grad_norm": 1.51175705146348,
      "learning_rate": 6.6110542469513065e-06,
      "loss": 0.9384,
      "step": 310
    },
    {
      "epoch": 0.4172571380654899,
      "grad_norm": 1.4180114835407127,
      "learning_rate": 6.590471709710704e-06,
      "loss": 0.9413,
      "step": 311
    },
    {
      "epoch": 0.4185988008888516,
      "grad_norm": 1.4988278891464717,
      "learning_rate": 6.5698591429483286e-06,
      "loss": 0.9516,
      "step": 312
    },
    {
      "epoch": 0.41994046371221333,
      "grad_norm": 1.5791863550483363,
      "learning_rate": 6.549216935847786e-06,
      "loss": 0.9372,
      "step": 313
    },
    {
      "epoch": 0.421282126535575,
      "grad_norm": 1.620573210066323,
      "learning_rate": 6.52854547815233e-06,
      "loss": 0.9286,
      "step": 314
    },
    {
      "epoch": 0.42262378935893674,
      "grad_norm": 1.4040060176512321,
      "learning_rate": 6.507845160157476e-06,
      "loss": 0.951,
      "step": 315
    },
    {
      "epoch": 0.4239654521822984,
      "grad_norm": 1.6033577516875774,
      "learning_rate": 6.487116372703663e-06,
      "loss": 0.9546,
      "step": 316
    },
    {
      "epoch": 0.42530711500566015,
      "grad_norm": 1.530825634158125,
      "learning_rate": 6.466359507168849e-06,
      "loss": 0.9407,
      "step": 317
    },
    {
      "epoch": 0.42664877782902183,
      "grad_norm": 1.462628543689146,
      "learning_rate": 6.445574955461134e-06,
      "loss": 0.9346,
      "step": 318
    },
    {
      "epoch": 0.42799044065238356,
      "grad_norm": 1.4846164727489912,
      "learning_rate": 6.424763110011356e-06,
      "loss": 0.9512,
      "step": 319
    },
    {
      "epoch": 0.42933210347574524,
      "grad_norm": 1.6348569990232538,
      "learning_rate": 6.403924363765686e-06,
      "loss": 0.9426,
      "step": 320
    },
    {
      "epoch": 0.430673766299107,
      "grad_norm": 1.3033058347347706,
      "learning_rate": 6.383059110178205e-06,
      "loss": 0.9367,
      "step": 321
    },
    {
      "epoch": 0.43201542912246865,
      "grad_norm": 1.3976319401939663,
      "learning_rate": 6.362167743203474e-06,
      "loss": 0.9478,
      "step": 322
    },
    {
      "epoch": 0.4333570919458304,
      "grad_norm": 2.240174630411269,
      "learning_rate": 6.3412506572891e-06,
      "loss": 0.9493,
      "step": 323
    },
    {
      "epoch": 0.43469875476919206,
      "grad_norm": 1.3498981951443567,
      "learning_rate": 6.320308247368285e-06,
      "loss": 0.9352,
      "step": 324
    },
    {
      "epoch": 0.4360404175925538,
      "grad_norm": 1.5036960555030219,
      "learning_rate": 6.2993409088523774e-06,
      "loss": 0.9537,
      "step": 325
    },
    {
      "epoch": 0.4373820804159155,
      "grad_norm": 1.4501884398705591,
      "learning_rate": 6.278349037623388e-06,
      "loss": 0.9389,
      "step": 326
    },
    {
      "epoch": 0.43872374323927715,
      "grad_norm": 1.8643264437967426,
      "learning_rate": 6.2573330300265375e-06,
      "loss": 0.9589,
      "step": 327
    },
    {
      "epoch": 0.4400654060626389,
      "grad_norm": 1.6346684956995083,
      "learning_rate": 6.236293282862756e-06,
      "loss": 0.95,
      "step": 328
    },
    {
      "epoch": 0.44140706888600056,
      "grad_norm": 1.468976590448429,
      "learning_rate": 6.215230193381203e-06,
      "loss": 0.9173,
      "step": 329
    },
    {
      "epoch": 0.4427487317093623,
      "grad_norm": 1.570877242399236,
      "learning_rate": 6.1941441592717564e-06,
      "loss": 0.9437,
      "step": 330
    },
    {
      "epoch": 0.444090394532724,
      "grad_norm": 1.4652863203353141,
      "learning_rate": 6.173035578657513e-06,
      "loss": 0.9316,
      "step": 331
    },
    {
      "epoch": 0.4454320573560857,
      "grad_norm": 1.3924224386568063,
      "learning_rate": 6.151904850087265e-06,
      "loss": 0.9435,
      "step": 332
    },
    {
      "epoch": 0.4467737201794474,
      "grad_norm": 1.4651748132936537,
      "learning_rate": 6.130752372527981e-06,
      "loss": 0.9252,
      "step": 333
    },
    {
      "epoch": 0.4481153830028091,
      "grad_norm": 1.4286663324788682,
      "learning_rate": 6.109578545357268e-06,
      "loss": 0.9457,
      "step": 334
    },
    {
      "epoch": 0.4494570458261708,
      "grad_norm": 1.740501083768386,
      "learning_rate": 6.088383768355829e-06,
      "loss": 0.9432,
      "step": 335
    },
    {
      "epoch": 0.45079870864953253,
      "grad_norm": 1.4961395872275942,
      "learning_rate": 6.067168441699927e-06,
      "loss": 0.9676,
      "step": 336
    },
    {
      "epoch": 0.4521403714728942,
      "grad_norm": 1.3752313492049115,
      "learning_rate": 6.045932965953813e-06,
      "loss": 0.9433,
      "step": 337
    },
    {
      "epoch": 0.45348203429625594,
      "grad_norm": 1.3820484972229559,
      "learning_rate": 6.024677742062176e-06,
      "loss": 0.944,
      "step": 338
    },
    {
      "epoch": 0.4548236971196176,
      "grad_norm": 2.0893492032041827,
      "learning_rate": 6.0034031713425636e-06,
      "loss": 0.9434,
      "step": 339
    },
    {
      "epoch": 0.45616535994297935,
      "grad_norm": 1.549871037483782,
      "learning_rate": 5.98210965547781e-06,
      "loss": 0.9748,
      "step": 340
    },
    {
      "epoch": 0.45750702276634103,
      "grad_norm": 1.5298386519614102,
      "learning_rate": 5.960797596508453e-06,
      "loss": 0.9508,
      "step": 341
    },
    {
      "epoch": 0.45884868558970276,
      "grad_norm": 1.413672937879993,
      "learning_rate": 5.939467396825137e-06,
      "loss": 0.9397,
      "step": 342
    },
    {
      "epoch": 0.46019034841306444,
      "grad_norm": 1.4329767730242147,
      "learning_rate": 5.918119459161021e-06,
      "loss": 0.9666,
      "step": 343
    },
    {
      "epoch": 0.4615320112364261,
      "grad_norm": 1.3700806796060583,
      "learning_rate": 5.896754186584173e-06,
      "loss": 0.9415,
      "step": 344
    },
    {
      "epoch": 0.46287367405978785,
      "grad_norm": 1.6410590057880081,
      "learning_rate": 5.875371982489959e-06,
      "loss": 0.9405,
      "step": 345
    },
    {
      "epoch": 0.46421533688314953,
      "grad_norm": 1.4182156518065911,
      "learning_rate": 5.8539732505934266e-06,
      "loss": 0.9198,
      "step": 346
    },
    {
      "epoch": 0.46555699970651127,
      "grad_norm": 1.4338567504194042,
      "learning_rate": 5.832558394921688e-06,
      "loss": 0.9207,
      "step": 347
    },
    {
      "epoch": 0.46689866252987294,
      "grad_norm": 1.4666500410104721,
      "learning_rate": 5.811127819806277e-06,
      "loss": 0.9286,
      "step": 348
    },
    {
      "epoch": 0.4682403253532347,
      "grad_norm": 1.3913442891866983,
      "learning_rate": 5.7896819298755346e-06,
      "loss": 0.9398,
      "step": 349
    },
    {
      "epoch": 0.46958198817659635,
      "grad_norm": 1.944673612051866,
      "learning_rate": 5.768221130046954e-06,
      "loss": 0.9165,
      "step": 350
    },
    {
      "epoch": 0.4709236509999581,
      "grad_norm": 1.9768029373235239,
      "learning_rate": 5.746745825519539e-06,
      "loss": 0.9324,
      "step": 351
    },
    {
      "epoch": 0.47226531382331977,
      "grad_norm": 1.522618389240632,
      "learning_rate": 5.725256421766158e-06,
      "loss": 0.9353,
      "step": 352
    },
    {
      "epoch": 0.4736069766466815,
      "grad_norm": 1.3101805254428442,
      "learning_rate": 5.703753324525885e-06,
      "loss": 0.9294,
      "step": 353
    },
    {
      "epoch": 0.4749486394700432,
      "grad_norm": 1.4723714938359946,
      "learning_rate": 5.682236939796337e-06,
      "loss": 0.9263,
      "step": 354
    },
    {
      "epoch": 0.4762903022934049,
      "grad_norm": 1.4879677997320575,
      "learning_rate": 5.660707673826016e-06,
      "loss": 0.9187,
      "step": 355
    },
    {
      "epoch": 0.4776319651167666,
      "grad_norm": 1.3341527207031307,
      "learning_rate": 5.639165933106626e-06,
      "loss": 0.9324,
      "step": 356
    },
    {
      "epoch": 0.4789736279401283,
      "grad_norm": 1.4232186684864636,
      "learning_rate": 5.617612124365411e-06,
      "loss": 0.945,
      "step": 357
    },
    {
      "epoch": 0.48031529076349,
      "grad_norm": 1.5777179718248369,
      "learning_rate": 5.596046654557467e-06,
      "loss": 0.9165,
      "step": 358
    },
    {
      "epoch": 0.4816569535868517,
      "grad_norm": 1.4509824338743527,
      "learning_rate": 5.574469930858062e-06,
      "loss": 0.9367,
      "step": 359
    },
    {
      "epoch": 0.4829986164102134,
      "grad_norm": 1.43523407700473,
      "learning_rate": 5.55288236065495e-06,
      "loss": 0.9381,
      "step": 360
    },
    {
      "epoch": 0.4843402792335751,
      "grad_norm": 1.2941141889852268,
      "learning_rate": 5.5312843515406725e-06,
      "loss": 0.9294,
      "step": 361
    },
    {
      "epoch": 0.4856819420569368,
      "grad_norm": 3.499123796935252,
      "learning_rate": 5.509676311304869e-06,
      "loss": 0.9485,
      "step": 362
    },
    {
      "epoch": 0.4870236048802985,
      "grad_norm": 1.4060828658646398,
      "learning_rate": 5.4880586479265774e-06,
      "loss": 0.9334,
      "step": 363
    },
    {
      "epoch": 0.48836526770366023,
      "grad_norm": 1.5632703720203085,
      "learning_rate": 5.466431769566525e-06,
      "loss": 0.9317,
      "step": 364
    },
    {
      "epoch": 0.4897069305270219,
      "grad_norm": 1.3794201306987215,
      "learning_rate": 5.4447960845594286e-06,
      "loss": 0.9581,
      "step": 365
    },
    {
      "epoch": 0.49104859335038364,
      "grad_norm": 1.4139859532152406,
      "learning_rate": 5.423152001406282e-06,
      "loss": 0.9272,
      "step": 366
    },
    {
      "epoch": 0.4923902561737453,
      "grad_norm": 1.3607081182885088,
      "learning_rate": 5.401499928766644e-06,
      "loss": 0.9083,
      "step": 367
    },
    {
      "epoch": 0.49373191899710706,
      "grad_norm": 1.3139396936274126,
      "learning_rate": 5.379840275450922e-06,
      "loss": 0.9125,
      "step": 368
    },
    {
      "epoch": 0.49507358182046873,
      "grad_norm": 1.720881483335334,
      "learning_rate": 5.358173450412649e-06,
      "loss": 0.9284,
      "step": 369
    },
    {
      "epoch": 0.49641524464383047,
      "grad_norm": 1.4961481743297107,
      "learning_rate": 5.3364998627407746e-06,
      "loss": 0.9333,
      "step": 370
    },
    {
      "epoch": 0.49775690746719214,
      "grad_norm": 1.291001773361284,
      "learning_rate": 5.314819921651928e-06,
      "loss": 0.9297,
      "step": 371
    },
    {
      "epoch": 0.4990985702905539,
      "grad_norm": 1.4874335774584129,
      "learning_rate": 5.293134036482697e-06,
      "loss": 0.926,
      "step": 372
    },
    {
      "epoch": 0.5004402331139156,
      "grad_norm": 1.3249150101616787,
      "learning_rate": 5.271442616681898e-06,
      "loss": 0.9477,
      "step": 373
    },
    {
      "epoch": 0.5017818959372773,
      "grad_norm": 1.4314814462436254,
      "learning_rate": 5.249746071802849e-06,
      "loss": 0.9158,
      "step": 374
    },
    {
      "epoch": 0.503123558760639,
      "grad_norm": 1.4850215366796362,
      "learning_rate": 5.228044811495632e-06,
      "loss": 0.9139,
      "step": 375
    },
    {
      "epoch": 0.5044652215840006,
      "grad_norm": 1.4346788643421962,
      "learning_rate": 5.2063392454993585e-06,
      "loss": 0.9275,
      "step": 376
    },
    {
      "epoch": 0.5058068844073623,
      "grad_norm": 1.3913557969729673,
      "learning_rate": 5.184629783634441e-06,
      "loss": 0.9252,
      "step": 377
    },
    {
      "epoch": 0.5071485472307241,
      "grad_norm": 1.3664085740854548,
      "learning_rate": 5.162916835794843e-06,
      "loss": 0.9318,
      "step": 378
    },
    {
      "epoch": 0.5084902100540858,
      "grad_norm": 1.32531510438658,
      "learning_rate": 5.141200811940352e-06,
      "loss": 0.9163,
      "step": 379
    },
    {
      "epoch": 0.5098318728774475,
      "grad_norm": 1.4833763087565515,
      "learning_rate": 5.119482122088828e-06,
      "loss": 0.9437,
      "step": 380
    },
    {
      "epoch": 0.5111735357008091,
      "grad_norm": 1.455536383105933,
      "learning_rate": 5.097761176308471e-06,
      "loss": 0.9353,
      "step": 381
    },
    {
      "epoch": 0.5125151985241709,
      "grad_norm": 1.4018575883339595,
      "learning_rate": 5.076038384710077e-06,
      "loss": 0.9378,
      "step": 382
    },
    {
      "epoch": 0.5138568613475326,
      "grad_norm": 1.529443325382905,
      "learning_rate": 5.054314157439286e-06,
      "loss": 0.9518,
      "step": 383
    },
    {
      "epoch": 0.5151985241708943,
      "grad_norm": 1.3333546144124035,
      "learning_rate": 5.032588904668851e-06,
      "loss": 0.9121,
      "step": 384
    },
    {
      "epoch": 0.516540186994256,
      "grad_norm": 1.4024032505431785,
      "learning_rate": 5.010863036590885e-06,
      "loss": 0.92,
      "step": 385
    },
    {
      "epoch": 0.5178818498176178,
      "grad_norm": 1.3464025585023112,
      "learning_rate": 4.989136963409117e-06,
      "loss": 0.933,
      "step": 386
    },
    {
      "epoch": 0.5192235126409794,
      "grad_norm": 1.380359478548665,
      "learning_rate": 4.967411095331149e-06,
      "loss": 0.9157,
      "step": 387
    },
    {
      "epoch": 0.5205651754643411,
      "grad_norm": 1.4753034718726812,
      "learning_rate": 4.945685842560715e-06,
      "loss": 0.9145,
      "step": 388
    },
    {
      "epoch": 0.5219068382877028,
      "grad_norm": 1.6175730490704796,
      "learning_rate": 4.923961615289924e-06,
      "loss": 0.9269,
      "step": 389
    },
    {
      "epoch": 0.5232485011110646,
      "grad_norm": 1.3437231689169467,
      "learning_rate": 4.9022388236915306e-06,
      "loss": 0.953,
      "step": 390
    },
    {
      "epoch": 0.5245901639344263,
      "grad_norm": 1.3084234752713977,
      "learning_rate": 4.880517877911173e-06,
      "loss": 0.9183,
      "step": 391
    },
    {
      "epoch": 0.5259318267577879,
      "grad_norm": 1.5145801721587038,
      "learning_rate": 4.858799188059651e-06,
      "loss": 0.9367,
      "step": 392
    },
    {
      "epoch": 0.5272734895811496,
      "grad_norm": 1.4077045421921577,
      "learning_rate": 4.837083164205159e-06,
      "loss": 0.9345,
      "step": 393
    },
    {
      "epoch": 0.5286151524045113,
      "grad_norm": 1.4157771272811692,
      "learning_rate": 4.815370216365562e-06,
      "loss": 0.942,
      "step": 394
    },
    {
      "epoch": 0.5299568152278731,
      "grad_norm": 1.2955482603917654,
      "learning_rate": 4.793660754500644e-06,
      "loss": 0.944,
      "step": 395
    },
    {
      "epoch": 0.5312984780512348,
      "grad_norm": 1.4171045612753872,
      "learning_rate": 4.771955188504371e-06,
      "loss": 0.9277,
      "step": 396
    },
    {
      "epoch": 0.5326401408745964,
      "grad_norm": 1.7506953727580603,
      "learning_rate": 4.75025392819715e-06,
      "loss": 0.9593,
      "step": 397
    },
    {
      "epoch": 0.5339818036979581,
      "grad_norm": 1.3630627845724446,
      "learning_rate": 4.7285573833181016e-06,
      "loss": 0.929,
      "step": 398
    },
    {
      "epoch": 0.5353234665213199,
      "grad_norm": 1.4941901177118797,
      "learning_rate": 4.7068659635173034e-06,
      "loss": 0.938,
      "step": 399
    },
    {
      "epoch": 0.5366651293446816,
      "grad_norm": 1.3837582089973754,
      "learning_rate": 4.685180078348072e-06,
      "loss": 0.9272,
      "step": 400
    },
    {
      "epoch": 0.5380067921680433,
      "grad_norm": 1.6311738523492914,
      "learning_rate": 4.663500137259226e-06,
      "loss": 0.9257,
      "step": 401
    },
    {
      "epoch": 0.5393484549914049,
      "grad_norm": 1.7826160967355433,
      "learning_rate": 4.641826549587352e-06,
      "loss": 0.9209,
      "step": 402
    },
    {
      "epoch": 0.5406901178147667,
      "grad_norm": 1.4426572030982026,
      "learning_rate": 4.62015972454908e-06,
      "loss": 0.9494,
      "step": 403
    },
    {
      "epoch": 0.5420317806381284,
      "grad_norm": 1.5231558399676566,
      "learning_rate": 4.598500071233358e-06,
      "loss": 0.9165,
      "step": 404
    },
    {
      "epoch": 0.5433734434614901,
      "grad_norm": 1.2912843216495555,
      "learning_rate": 4.57684799859372e-06,
      "loss": 0.9327,
      "step": 405
    },
    {
      "epoch": 0.5447151062848518,
      "grad_norm": 1.3285466026775592,
      "learning_rate": 4.555203915440573e-06,
      "loss": 0.9158,
      "step": 406
    },
    {
      "epoch": 0.5460567691082135,
      "grad_norm": 1.4181531717310327,
      "learning_rate": 4.533568230433477e-06,
      "loss": 0.8959,
      "step": 407
    },
    {
      "epoch": 0.5473984319315752,
      "grad_norm": 1.3759849226932546,
      "learning_rate": 4.511941352073424e-06,
      "loss": 0.9203,
      "step": 408
    },
    {
      "epoch": 0.5487400947549369,
      "grad_norm": 1.5030252203208716,
      "learning_rate": 4.4903236886951315e-06,
      "loss": 0.9218,
      "step": 409
    },
    {
      "epoch": 0.5500817575782986,
      "grad_norm": 1.4120496050942768,
      "learning_rate": 4.468715648459329e-06,
      "loss": 0.9271,
      "step": 410
    },
    {
      "epoch": 0.5514234204016603,
      "grad_norm": 1.6150466333531532,
      "learning_rate": 4.447117639345052e-06,
      "loss": 0.9352,
      "step": 411
    },
    {
      "epoch": 0.552765083225022,
      "grad_norm": 1.3313158710584112,
      "learning_rate": 4.42553006914194e-06,
      "loss": 0.9424,
      "step": 412
    },
    {
      "epoch": 0.5541067460483837,
      "grad_norm": 1.4173966801171765,
      "learning_rate": 4.403953345442535e-06,
      "loss": 0.9059,
      "step": 413
    },
    {
      "epoch": 0.5554484088717454,
      "grad_norm": 1.4219127140605912,
      "learning_rate": 4.382387875634592e-06,
      "loss": 0.8982,
      "step": 414
    },
    {
      "epoch": 0.5567900716951071,
      "grad_norm": 1.4807768234573824,
      "learning_rate": 4.360834066893376e-06,
      "loss": 0.9112,
      "step": 415
    },
    {
      "epoch": 0.5581317345184689,
      "grad_norm": 1.4934876617789734,
      "learning_rate": 4.339292326173987e-06,
      "loss": 0.9394,
      "step": 416
    },
    {
      "epoch": 0.5594733973418305,
      "grad_norm": 1.2964568666580691,
      "learning_rate": 4.317763060203665e-06,
      "loss": 0.9087,
      "step": 417
    },
    {
      "epoch": 0.5608150601651922,
      "grad_norm": 1.5071257220649843,
      "learning_rate": 4.296246675474119e-06,
      "loss": 0.9273,
      "step": 418
    },
    {
      "epoch": 0.5621567229885539,
      "grad_norm": 1.389386966271088,
      "learning_rate": 4.274743578233846e-06,
      "loss": 0.9264,
      "step": 419
    },
    {
      "epoch": 0.5634983858119157,
      "grad_norm": 1.7541684817806704,
      "learning_rate": 4.253254174480462e-06,
      "loss": 0.9148,
      "step": 420
    },
    {
      "epoch": 0.5648400486352774,
      "grad_norm": 1.6521682610990946,
      "learning_rate": 4.231778869953047e-06,
      "loss": 0.9324,
      "step": 421
    },
    {
      "epoch": 0.566181711458639,
      "grad_norm": 1.4036410472707417,
      "learning_rate": 4.210318070124465e-06,
      "loss": 0.9175,
      "step": 422
    },
    {
      "epoch": 0.5675233742820007,
      "grad_norm": 1.4760260423378482,
      "learning_rate": 4.188872180193723e-06,
      "loss": 0.9301,
      "step": 423
    },
    {
      "epoch": 0.5688650371053624,
      "grad_norm": 1.4915112006125035,
      "learning_rate": 4.167441605078314e-06,
      "loss": 0.9192,
      "step": 424
    },
    {
      "epoch": 0.5702066999287242,
      "grad_norm": 1.4529343235561802,
      "learning_rate": 4.146026749406574e-06,
      "loss": 0.9237,
      "step": 425
    },
    {
      "epoch": 0.5715483627520859,
      "grad_norm": 1.5441157122225924,
      "learning_rate": 4.124628017510043e-06,
      "loss": 0.9382,
      "step": 426
    },
    {
      "epoch": 0.5728900255754475,
      "grad_norm": 1.4800738041606645,
      "learning_rate": 4.10324581341583e-06,
      "loss": 0.9259,
      "step": 427
    },
    {
      "epoch": 0.5742316883988092,
      "grad_norm": 1.244794516407526,
      "learning_rate": 4.081880540838981e-06,
      "loss": 0.9147,
      "step": 428
    },
    {
      "epoch": 0.575573351222171,
      "grad_norm": 1.9868472108932922,
      "learning_rate": 4.060532603174865e-06,
      "loss": 0.9249,
      "step": 429
    },
    {
      "epoch": 0.5769150140455327,
      "grad_norm": 2.043155859552025,
      "learning_rate": 4.039202403491548e-06,
      "loss": 0.914,
      "step": 430
    },
    {
      "epoch": 0.5782566768688944,
      "grad_norm": 1.4635852855825318,
      "learning_rate": 4.0178903445221905e-06,
      "loss": 0.9268,
      "step": 431
    },
    {
      "epoch": 0.579598339692256,
      "grad_norm": 1.4320508916527583,
      "learning_rate": 3.996596828657437e-06,
      "loss": 0.9288,
      "step": 432
    },
    {
      "epoch": 0.5809400025156178,
      "grad_norm": 1.6077840908976693,
      "learning_rate": 3.975322257937825e-06,
      "loss": 0.93,
      "step": 433
    },
    {
      "epoch": 0.5822816653389795,
      "grad_norm": 1.3697545321413767,
      "learning_rate": 3.9540670340461875e-06,
      "loss": 0.9037,
      "step": 434
    },
    {
      "epoch": 0.5836233281623412,
      "grad_norm": 1.509438713977569,
      "learning_rate": 3.932831558300074e-06,
      "loss": 0.9233,
      "step": 435
    },
    {
      "epoch": 0.5849649909857029,
      "grad_norm": 1.4813174251922685,
      "learning_rate": 3.911616231644172e-06,
      "loss": 0.9214,
      "step": 436
    },
    {
      "epoch": 0.5863066538090647,
      "grad_norm": 1.3554393686833108,
      "learning_rate": 3.8904214546427355e-06,
      "loss": 0.9217,
      "step": 437
    },
    {
      "epoch": 0.5876483166324263,
      "grad_norm": 1.4327778536512215,
      "learning_rate": 3.869247627472021e-06,
      "loss": 0.9381,
      "step": 438
    },
    {
      "epoch": 0.588989979455788,
      "grad_norm": 1.4341323896035245,
      "learning_rate": 3.848095149912737e-06,
      "loss": 0.9346,
      "step": 439
    },
    {
      "epoch": 0.5903316422791497,
      "grad_norm": 1.4824059212863905,
      "learning_rate": 3.82696442134249e-06,
      "loss": 0.9436,
      "step": 440
    },
    {
      "epoch": 0.5916733051025114,
      "grad_norm": 1.4570331826042282,
      "learning_rate": 3.8058558407282465e-06,
      "loss": 0.9106,
      "step": 441
    },
    {
      "epoch": 0.5930149679258732,
      "grad_norm": 1.308814793199382,
      "learning_rate": 3.7847698066187975e-06,
      "loss": 0.9165,
      "step": 442
    },
    {
      "epoch": 0.5943566307492348,
      "grad_norm": 1.378524196980017,
      "learning_rate": 3.7637067171372437e-06,
      "loss": 0.9233,
      "step": 443
    },
    {
      "epoch": 0.5956982935725965,
      "grad_norm": 1.4319296775411137,
      "learning_rate": 3.742666969973463e-06,
      "loss": 0.9328,
      "step": 444
    },
    {
      "epoch": 0.5970399563959582,
      "grad_norm": 1.3595089447285513,
      "learning_rate": 3.721650962376612e-06,
      "loss": 0.9237,
      "step": 445
    },
    {
      "epoch": 0.59838161921932,
      "grad_norm": 1.3018802984460596,
      "learning_rate": 3.7006590911476242e-06,
      "loss": 0.9075,
      "step": 446
    },
    {
      "epoch": 0.5997232820426817,
      "grad_norm": 1.3741192831894566,
      "learning_rate": 3.6796917526317153e-06,
      "loss": 0.9242,
      "step": 447
    },
    {
      "epoch": 0.6010649448660433,
      "grad_norm": 1.5039382896600555,
      "learning_rate": 3.6587493427109015e-06,
      "loss": 0.9005,
      "step": 448
    },
    {
      "epoch": 0.602406607689405,
      "grad_norm": 2.594160580170954,
      "learning_rate": 3.6378322567965284e-06,
      "loss": 0.9419,
      "step": 449
    },
    {
      "epoch": 0.6037482705127668,
      "grad_norm": 1.5181333868507627,
      "learning_rate": 3.6169408898217973e-06,
      "loss": 0.9126,
      "step": 450
    },
    {
      "epoch": 0.6050899333361285,
      "grad_norm": 1.2916775825580151,
      "learning_rate": 3.596075636234315e-06,
      "loss": 0.9203,
      "step": 451
    },
    {
      "epoch": 0.6064315961594902,
      "grad_norm": 1.3505432254066787,
      "learning_rate": 3.575236889988646e-06,
      "loss": 0.9098,
      "step": 452
    },
    {
      "epoch": 0.6077732589828518,
      "grad_norm": 1.4144638543833534,
      "learning_rate": 3.554425044538868e-06,
      "loss": 0.9233,
      "step": 453
    },
    {
      "epoch": 0.6091149218062136,
      "grad_norm": 1.374872140252449,
      "learning_rate": 3.5336404928311518e-06,
      "loss": 0.9023,
      "step": 454
    },
    {
      "epoch": 0.6104565846295753,
      "grad_norm": 1.361948134387405,
      "learning_rate": 3.5128836272963377e-06,
      "loss": 0.9235,
      "step": 455
    },
    {
      "epoch": 0.611798247452937,
      "grad_norm": 1.5371710757092296,
      "learning_rate": 3.4921548398425246e-06,
      "loss": 0.9306,
      "step": 456
    },
    {
      "epoch": 0.6131399102762987,
      "grad_norm": 1.3466506760555965,
      "learning_rate": 3.4714545218476727e-06,
      "loss": 0.922,
      "step": 457
    },
    {
      "epoch": 0.6144815730996603,
      "grad_norm": 1.4262251212000858,
      "learning_rate": 3.4507830641522144e-06,
      "loss": 0.9236,
      "step": 458
    },
    {
      "epoch": 0.6158232359230221,
      "grad_norm": 1.3179672517423986,
      "learning_rate": 3.430140857051675e-06,
      "loss": 0.9231,
      "step": 459
    },
    {
      "epoch": 0.6171648987463838,
      "grad_norm": 1.2130034269192478,
      "learning_rate": 3.4095282902892982e-06,
      "loss": 0.9127,
      "step": 460
    },
    {
      "epoch": 0.6185065615697455,
      "grad_norm": 1.251798831435258,
      "learning_rate": 3.388945753048697e-06,
      "loss": 0.8986,
      "step": 461
    },
    {
      "epoch": 0.6198482243931072,
      "grad_norm": 1.3580683945289906,
      "learning_rate": 3.3683936339464957e-06,
      "loss": 0.9437,
      "step": 462
    },
    {
      "epoch": 0.621189887216469,
      "grad_norm": 1.4318866735343905,
      "learning_rate": 3.3478723210250006e-06,
      "loss": 0.911,
      "step": 463
    },
    {
      "epoch": 0.6225315500398306,
      "grad_norm": 1.4492417004844953,
      "learning_rate": 3.3273822017448683e-06,
      "loss": 0.9233,
      "step": 464
    },
    {
      "epoch": 0.6238732128631923,
      "grad_norm": 1.32995831571142,
      "learning_rate": 3.306923662977789e-06,
      "loss": 0.9054,
      "step": 465
    },
    {
      "epoch": 0.625214875686554,
      "grad_norm": 1.3917139306065034,
      "learning_rate": 3.2864970909991927e-06,
      "loss": 0.9088,
      "step": 466
    },
    {
      "epoch": 0.6265565385099158,
      "grad_norm": 1.326251399568166,
      "learning_rate": 3.2661028714809405e-06,
      "loss": 0.9164,
      "step": 467
    },
    {
      "epoch": 0.6278982013332775,
      "grad_norm": 1.362806929059327,
      "learning_rate": 3.2457413894840516e-06,
      "loss": 0.8841,
      "step": 468
    },
    {
      "epoch": 0.6292398641566391,
      "grad_norm": 1.5947392064771095,
      "learning_rate": 3.225413029451434e-06,
      "loss": 0.9128,
      "step": 469
    },
    {
      "epoch": 0.6305815269800008,
      "grad_norm": 1.816586346006587,
      "learning_rate": 3.20511817520062e-06,
      "loss": 0.8956,
      "step": 470
    },
    {
      "epoch": 0.6319231898033626,
      "grad_norm": 1.414680996959683,
      "learning_rate": 3.184857209916528e-06,
      "loss": 0.9163,
      "step": 471
    },
    {
      "epoch": 0.6332648526267243,
      "grad_norm": 1.3021277778693485,
      "learning_rate": 3.1646305161442183e-06,
      "loss": 0.9021,
      "step": 472
    },
    {
      "epoch": 0.634606515450086,
      "grad_norm": 1.7791142325104827,
      "learning_rate": 3.144438475781676e-06,
      "loss": 0.9261,
      "step": 473
    },
    {
      "epoch": 0.6359481782734476,
      "grad_norm": 1.5515962833422756,
      "learning_rate": 3.1242814700725977e-06,
      "loss": 0.919,
      "step": 474
    },
    {
      "epoch": 0.6372898410968093,
      "grad_norm": 1.3625300760427628,
      "learning_rate": 3.1041598795991967e-06,
      "loss": 0.9244,
      "step": 475
    },
    {
      "epoch": 0.6386315039201711,
      "grad_norm": 1.2420275075734377,
      "learning_rate": 3.0840740842750145e-06,
      "loss": 0.9062,
      "step": 476
    },
    {
      "epoch": 0.6399731667435328,
      "grad_norm": 2.2279397674834205,
      "learning_rate": 3.064024463337747e-06,
      "loss": 0.9084,
      "step": 477
    },
    {
      "epoch": 0.6413148295668945,
      "grad_norm": 1.4768686245810154,
      "learning_rate": 3.044011395342087e-06,
      "loss": 0.9296,
      "step": 478
    },
    {
      "epoch": 0.6426564923902561,
      "grad_norm": 1.255495856439165,
      "learning_rate": 3.0240352581525735e-06,
      "loss": 0.9226,
      "step": 479
    },
    {
      "epoch": 0.6439981552136179,
      "grad_norm": 1.3311864368234478,
      "learning_rate": 3.0040964289364618e-06,
      "loss": 0.9174,
      "step": 480
    },
    {
      "epoch": 0.6453398180369796,
      "grad_norm": 1.4180453328625617,
      "learning_rate": 2.984195284156597e-06,
      "loss": 0.9018,
      "step": 481
    },
    {
      "epoch": 0.6466814808603413,
      "grad_norm": 1.454713670842971,
      "learning_rate": 2.964332199564309e-06,
      "loss": 0.9171,
      "step": 482
    },
    {
      "epoch": 0.648023143683703,
      "grad_norm": 1.3635198053975708,
      "learning_rate": 2.944507550192318e-06,
      "loss": 0.927,
      "step": 483
    },
    {
      "epoch": 0.6493648065070647,
      "grad_norm": 1.279867079899654,
      "learning_rate": 2.9247217103476527e-06,
      "loss": 0.9028,
      "step": 484
    },
    {
      "epoch": 0.6507064693304264,
      "grad_norm": 1.445182274301436,
      "learning_rate": 2.904975053604584e-06,
      "loss": 0.9098,
      "step": 485
    },
    {
      "epoch": 0.6520481321537881,
      "grad_norm": 1.397463498237059,
      "learning_rate": 2.885267952797569e-06,
      "loss": 0.9216,
      "step": 486
    },
    {
      "epoch": 0.6533897949771498,
      "grad_norm": 1.4525135526307655,
      "learning_rate": 2.865600780014216e-06,
      "loss": 0.9289,
      "step": 487
    },
    {
      "epoch": 0.6547314578005116,
      "grad_norm": 1.6042241879134849,
      "learning_rate": 2.8459739065882537e-06,
      "loss": 0.9245,
      "step": 488
    },
    {
      "epoch": 0.6560731206238732,
      "grad_norm": 1.4445341641370009,
      "learning_rate": 2.826387703092528e-06,
      "loss": 0.9208,
      "step": 489
    },
    {
      "epoch": 0.6574147834472349,
      "grad_norm": 2.448108575133241,
      "learning_rate": 2.8068425393319965e-06,
      "loss": 0.9254,
      "step": 490
    },
    {
      "epoch": 0.6587564462705966,
      "grad_norm": 1.4669057051144683,
      "learning_rate": 2.7873387843367494e-06,
      "loss": 0.8776,
      "step": 491
    },
    {
      "epoch": 0.6600981090939583,
      "grad_norm": 1.4674760507750413,
      "learning_rate": 2.7678768063550454e-06,
      "loss": 0.9223,
      "step": 492
    },
    {
      "epoch": 0.6614397719173201,
      "grad_norm": 1.363895109336093,
      "learning_rate": 2.748456972846353e-06,
      "loss": 0.9282,
      "step": 493
    },
    {
      "epoch": 0.6627814347406817,
      "grad_norm": 1.2507979608624489,
      "learning_rate": 2.7290796504744184e-06,
      "loss": 0.9114,
      "step": 494
    },
    {
      "epoch": 0.6641230975640434,
      "grad_norm": 1.357510836988213,
      "learning_rate": 2.7097452051003375e-06,
      "loss": 0.9045,
      "step": 495
    },
    {
      "epoch": 0.6654647603874051,
      "grad_norm": 1.352458443420182,
      "learning_rate": 2.69045400177565e-06,
      "loss": 0.9277,
      "step": 496
    },
    {
      "epoch": 0.6668064232107669,
      "grad_norm": 1.3923375575907286,
      "learning_rate": 2.6712064047354515e-06,
      "loss": 0.9063,
      "step": 497
    },
    {
      "epoch": 0.6681480860341286,
      "grad_norm": 1.6237983373253484,
      "learning_rate": 2.6520027773915075e-06,
      "loss": 0.9255,
      "step": 498
    },
    {
      "epoch": 0.6694897488574902,
      "grad_norm": 1.5156068144660368,
      "learning_rate": 2.632843482325398e-06,
      "loss": 0.9217,
      "step": 499
    },
    {
      "epoch": 0.6708314116808519,
      "grad_norm": 1.2845138737825397,
      "learning_rate": 2.6137288812816695e-06,
      "loss": 0.9361,
      "step": 500
    },
    {
      "epoch": 0.6721730745042137,
      "grad_norm": 1.3411532615795354,
      "learning_rate": 2.594659335161008e-06,
      "loss": 0.8976,
      "step": 501
    },
    {
      "epoch": 0.6735147373275754,
      "grad_norm": 1.3161918955262504,
      "learning_rate": 2.5756352040134193e-06,
      "loss": 0.9032,
      "step": 502
    },
    {
      "epoch": 0.6748564001509371,
      "grad_norm": 1.2808948711754613,
      "learning_rate": 2.5566568470314345e-06,
      "loss": 0.876,
      "step": 503
    },
    {
      "epoch": 0.6761980629742987,
      "grad_norm": 1.2683456388164283,
      "learning_rate": 2.5377246225433306e-06,
      "loss": 0.9133,
      "step": 504
    },
    {
      "epoch": 0.6775397257976605,
      "grad_norm": 1.4083617008244946,
      "learning_rate": 2.5188388880063595e-06,
      "loss": 0.8979,
      "step": 505
    },
    {
      "epoch": 0.6788813886210222,
      "grad_norm": 1.3198518834409616,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.9163,
      "step": 506
    },
    {
      "epoch": 0.6802230514443839,
      "grad_norm": 1.2627840763757283,
      "learning_rate": 2.481208314219233e-06,
      "loss": 0.8808,
      "step": 507
    },
    {
      "epoch": 0.6815647142677456,
      "grad_norm": 1.5321601221415106,
      "learning_rate": 2.462464185467811e-06,
      "loss": 0.9326,
      "step": 508
    },
    {
      "epoch": 0.6829063770911072,
      "grad_norm": 1.4336529527711754,
      "learning_rate": 2.4437679676515735e-06,
      "loss": 0.9273,
      "step": 509
    },
    {
      "epoch": 0.684248039914469,
      "grad_norm": 1.301319565756621,
      "learning_rate": 2.4251200137717545e-06,
      "loss": 0.9144,
      "step": 510
    },
    {
      "epoch": 0.6855897027378307,
      "grad_norm": 1.450888662732102,
      "learning_rate": 2.4065206759183284e-06,
      "loss": 0.9163,
      "step": 511
    },
    {
      "epoch": 0.6869313655611924,
      "grad_norm": 1.5629774444609523,
      "learning_rate": 2.387970305263349e-06,
      "loss": 0.9079,
      "step": 512
    },
    {
      "epoch": 0.6882730283845541,
      "grad_norm": 1.419799211127572,
      "learning_rate": 2.3694692520543293e-06,
      "loss": 0.9176,
      "step": 513
    },
    {
      "epoch": 0.6896146912079159,
      "grad_norm": 1.3850306280239242,
      "learning_rate": 2.3510178656076245e-06,
      "loss": 0.9203,
      "step": 514
    },
    {
      "epoch": 0.6909563540312775,
      "grad_norm": 1.4448336778378281,
      "learning_rate": 2.3326164943018353e-06,
      "loss": 0.9224,
      "step": 515
    },
    {
      "epoch": 0.6922980168546392,
      "grad_norm": 1.5296983904894126,
      "learning_rate": 2.3142654855712353e-06,
      "loss": 0.9003,
      "step": 516
    },
    {
      "epoch": 0.6936396796780009,
      "grad_norm": 1.511761151648842,
      "learning_rate": 2.295965185899205e-06,
      "loss": 0.9116,
      "step": 517
    },
    {
      "epoch": 0.6949813425013627,
      "grad_norm": 1.2910704562876771,
      "learning_rate": 2.277715940811692e-06,
      "loss": 0.9047,
      "step": 518
    },
    {
      "epoch": 0.6963230053247244,
      "grad_norm": 1.2875242298776879,
      "learning_rate": 2.259518094870693e-06,
      "loss": 0.9144,
      "step": 519
    },
    {
      "epoch": 0.697664668148086,
      "grad_norm": 1.3530863365538197,
      "learning_rate": 2.241371991667739e-06,
      "loss": 0.9102,
      "step": 520
    },
    {
      "epoch": 0.6990063309714477,
      "grad_norm": 1.7870727665566009,
      "learning_rate": 2.2232779738174126e-06,
      "loss": 0.9147,
      "step": 521
    },
    {
      "epoch": 0.7003479937948094,
      "grad_norm": 1.2954206040170222,
      "learning_rate": 2.2052363829508776e-06,
      "loss": 0.9151,
      "step": 522
    },
    {
      "epoch": 0.7016896566181712,
      "grad_norm": 1.2518430383457515,
      "learning_rate": 2.1872475597094323e-06,
      "loss": 0.9096,
      "step": 523
    },
    {
      "epoch": 0.7030313194415329,
      "grad_norm": 1.2480131310096585,
      "learning_rate": 2.1693118437380727e-06,
      "loss": 0.9363,
      "step": 524
    },
    {
      "epoch": 0.7043729822648945,
      "grad_norm": 1.2545817095323775,
      "learning_rate": 2.151429573679084e-06,
      "loss": 0.887,
      "step": 525
    },
    {
      "epoch": 0.7057146450882562,
      "grad_norm": 1.3238327616598924,
      "learning_rate": 2.133601087165644e-06,
      "loss": 0.9247,
      "step": 526
    },
    {
      "epoch": 0.707056307911618,
      "grad_norm": 1.2204792393828976,
      "learning_rate": 2.1158267208154497e-06,
      "loss": 0.9005,
      "step": 527
    },
    {
      "epoch": 0.7083979707349797,
      "grad_norm": 1.8336414137905717,
      "learning_rate": 2.098106810224362e-06,
      "loss": 0.9074,
      "step": 528
    },
    {
      "epoch": 0.7097396335583414,
      "grad_norm": 1.3213871439009308,
      "learning_rate": 2.080441689960066e-06,
      "loss": 0.9003,
      "step": 529
    },
    {
      "epoch": 0.711081296381703,
      "grad_norm": 1.5959521305555897,
      "learning_rate": 2.0628316935557595e-06,
      "loss": 0.9092,
      "step": 530
    },
    {
      "epoch": 0.7124229592050648,
      "grad_norm": 1.22786859984951,
      "learning_rate": 2.0452771535038518e-06,
      "loss": 0.9035,
      "step": 531
    },
    {
      "epoch": 0.7137646220284265,
      "grad_norm": 1.2187462970114127,
      "learning_rate": 2.0277784012496865e-06,
      "loss": 0.9129,
      "step": 532
    },
    {
      "epoch": 0.7151062848517882,
      "grad_norm": 1.3041611731183413,
      "learning_rate": 2.010335767185283e-06,
      "loss": 0.9262,
      "step": 533
    },
    {
      "epoch": 0.7164479476751499,
      "grad_norm": 1.2636919161262752,
      "learning_rate": 1.9929495806431024e-06,
      "loss": 0.9065,
      "step": 534
    },
    {
      "epoch": 0.7177896104985116,
      "grad_norm": 1.4815138266505345,
      "learning_rate": 1.9756201698898246e-06,
      "loss": 0.9026,
      "step": 535
    },
    {
      "epoch": 0.7191312733218733,
      "grad_norm": 1.3422685731450896,
      "learning_rate": 1.958347862120151e-06,
      "loss": 0.8931,
      "step": 536
    },
    {
      "epoch": 0.720472936145235,
      "grad_norm": 1.2736773153988343,
      "learning_rate": 1.9411329834506286e-06,
      "loss": 0.9189,
      "step": 537
    },
    {
      "epoch": 0.7218145989685967,
      "grad_norm": 1.337143595246076,
      "learning_rate": 1.923975858913492e-06,
      "loss": 0.8973,
      "step": 538
    },
    {
      "epoch": 0.7231562617919584,
      "grad_norm": 1.384362022652147,
      "learning_rate": 1.9068768124505265e-06,
      "loss": 0.9113,
      "step": 539
    },
    {
      "epoch": 0.7244979246153201,
      "grad_norm": 1.3520897803445997,
      "learning_rate": 1.8898361669069497e-06,
      "loss": 0.9008,
      "step": 540
    },
    {
      "epoch": 0.7258395874386818,
      "grad_norm": 1.2796126814289661,
      "learning_rate": 1.8728542440253232e-06,
      "loss": 0.8952,
      "step": 541
    },
    {
      "epoch": 0.7271812502620435,
      "grad_norm": 1.2835543746380815,
      "learning_rate": 1.8559313644394677e-06,
      "loss": 0.9304,
      "step": 542
    },
    {
      "epoch": 0.7285229130854052,
      "grad_norm": 1.4028148598362826,
      "learning_rate": 1.8390678476684143e-06,
      "loss": 0.911,
      "step": 543
    },
    {
      "epoch": 0.729864575908767,
      "grad_norm": 1.7653722051119831,
      "learning_rate": 1.8222640121103724e-06,
      "loss": 0.9123,
      "step": 544
    },
    {
      "epoch": 0.7312062387321286,
      "grad_norm": 1.761958414062295,
      "learning_rate": 1.8055201750367163e-06,
      "loss": 0.8993,
      "step": 545
    },
    {
      "epoch": 0.7325479015554903,
      "grad_norm": 1.4012619406304154,
      "learning_rate": 1.7888366525859968e-06,
      "loss": 0.9051,
      "step": 546
    },
    {
      "epoch": 0.733889564378852,
      "grad_norm": 1.2707818756082008,
      "learning_rate": 1.7722137597579698e-06,
      "loss": 0.9121,
      "step": 547
    },
    {
      "epoch": 0.7352312272022138,
      "grad_norm": 1.2914565300838898,
      "learning_rate": 1.7556518104076497e-06,
      "loss": 0.9213,
      "step": 548
    },
    {
      "epoch": 0.7365728900255755,
      "grad_norm": 1.2704401187441536,
      "learning_rate": 1.7391511172393849e-06,
      "loss": 0.8885,
      "step": 549
    },
    {
      "epoch": 0.7379145528489371,
      "grad_norm": 1.2933241809008478,
      "learning_rate": 1.7227119918009516e-06,
      "loss": 0.9147,
      "step": 550
    },
    {
      "epoch": 0.7392562156722988,
      "grad_norm": 1.2465347261670734,
      "learning_rate": 1.706334744477673e-06,
      "loss": 0.9163,
      "step": 551
    },
    {
      "epoch": 0.7405978784956606,
      "grad_norm": 1.6435915576463558,
      "learning_rate": 1.6900196844865575e-06,
      "loss": 0.9097,
      "step": 552
    },
    {
      "epoch": 0.7419395413190223,
      "grad_norm": 1.490867528585643,
      "learning_rate": 1.6737671198704625e-06,
      "loss": 0.9133,
      "step": 553
    },
    {
      "epoch": 0.743281204142384,
      "grad_norm": 1.2585765466201615,
      "learning_rate": 1.657577357492277e-06,
      "loss": 0.9059,
      "step": 554
    },
    {
      "epoch": 0.7446228669657456,
      "grad_norm": 1.301408224002227,
      "learning_rate": 1.6414507030291249e-06,
      "loss": 0.9005,
      "step": 555
    },
    {
      "epoch": 0.7459645297891073,
      "grad_norm": 1.2519314889957627,
      "learning_rate": 1.6253874609666033e-06,
      "loss": 0.9055,
      "step": 556
    },
    {
      "epoch": 0.7473061926124691,
      "grad_norm": 1.2463068016774805,
      "learning_rate": 1.609387934593019e-06,
      "loss": 0.9102,
      "step": 557
    },
    {
      "epoch": 0.7486478554358308,
      "grad_norm": 1.2928441743340047,
      "learning_rate": 1.5934524259936757e-06,
      "loss": 0.899,
      "step": 558
    },
    {
      "epoch": 0.7499895182591925,
      "grad_norm": 1.4106688088793764,
      "learning_rate": 1.5775812360451598e-06,
      "loss": 0.9166,
      "step": 559
    },
    {
      "epoch": 0.7513311810825541,
      "grad_norm": 1.2624680052602721,
      "learning_rate": 1.5617746644096681e-06,
      "loss": 0.9004,
      "step": 560
    },
    {
      "epoch": 0.7526728439059159,
      "grad_norm": 1.3318884983797796,
      "learning_rate": 1.5460330095293447e-06,
      "loss": 0.8796,
      "step": 561
    },
    {
      "epoch": 0.7540145067292776,
      "grad_norm": 1.2843333775104402,
      "learning_rate": 1.5303565686206452e-06,
      "loss": 0.8792,
      "step": 562
    },
    {
      "epoch": 0.7553561695526393,
      "grad_norm": 1.5718745193782164,
      "learning_rate": 1.514745637668733e-06,
      "loss": 0.9076,
      "step": 563
    },
    {
      "epoch": 0.756697832376001,
      "grad_norm": 1.3199412911960229,
      "learning_rate": 1.4992005114218805e-06,
      "loss": 0.8985,
      "step": 564
    },
    {
      "epoch": 0.7580394951993628,
      "grad_norm": 1.5059559020554452,
      "learning_rate": 1.4837214833859099e-06,
      "loss": 0.9307,
      "step": 565
    },
    {
      "epoch": 0.7593811580227244,
      "grad_norm": 1.5423291317821797,
      "learning_rate": 1.4683088458186512e-06,
      "loss": 0.9048,
      "step": 566
    },
    {
      "epoch": 0.7607228208460861,
      "grad_norm": 2.410355060033747,
      "learning_rate": 1.4529628897244214e-06,
      "loss": 0.908,
      "step": 567
    },
    {
      "epoch": 0.7620644836694478,
      "grad_norm": 1.3044233422661902,
      "learning_rate": 1.4376839048485335e-06,
      "loss": 0.8743,
      "step": 568
    },
    {
      "epoch": 0.7634061464928096,
      "grad_norm": 1.4222830676115559,
      "learning_rate": 1.422472179671825e-06,
      "loss": 0.9205,
      "step": 569
    },
    {
      "epoch": 0.7647478093161713,
      "grad_norm": 1.3510955600944414,
      "learning_rate": 1.4073280014052077e-06,
      "loss": 0.9395,
      "step": 570
    },
    {
      "epoch": 0.7660894721395329,
      "grad_norm": 1.2345880688085022,
      "learning_rate": 1.3922516559842498e-06,
      "loss": 0.9,
      "step": 571
    },
    {
      "epoch": 0.7674311349628946,
      "grad_norm": 1.3057826578608795,
      "learning_rate": 1.3772434280637737e-06,
      "loss": 0.9206,
      "step": 572
    },
    {
      "epoch": 0.7687727977862563,
      "grad_norm": 1.715585212025323,
      "learning_rate": 1.3623036010124845e-06,
      "loss": 0.897,
      "step": 573
    },
    {
      "epoch": 0.7701144606096181,
      "grad_norm": 1.3750530619640935,
      "learning_rate": 1.347432456907617e-06,
      "loss": 0.8954,
      "step": 574
    },
    {
      "epoch": 0.7714561234329798,
      "grad_norm": 1.3395553063427457,
      "learning_rate": 1.3326302765296096e-06,
      "loss": 0.9226,
      "step": 575
    },
    {
      "epoch": 0.7727977862563414,
      "grad_norm": 1.4774831354066553,
      "learning_rate": 1.3178973393568055e-06,
      "loss": 0.9076,
      "step": 576
    },
    {
      "epoch": 0.7741394490797031,
      "grad_norm": 1.2722467324017726,
      "learning_rate": 1.3032339235601749e-06,
      "loss": 0.8837,
      "step": 577
    },
    {
      "epoch": 0.7754811119030649,
      "grad_norm": 1.4000195103653286,
      "learning_rate": 1.2886403059980602e-06,
      "loss": 0.8932,
      "step": 578
    },
    {
      "epoch": 0.7768227747264266,
      "grad_norm": 1.4528378906964872,
      "learning_rate": 1.2741167622109557e-06,
      "loss": 0.8949,
      "step": 579
    },
    {
      "epoch": 0.7781644375497883,
      "grad_norm": 1.2457551580747528,
      "learning_rate": 1.259663566416296e-06,
      "loss": 0.9079,
      "step": 580
    },
    {
      "epoch": 0.7795061003731499,
      "grad_norm": 1.9720344272287313,
      "learning_rate": 1.2452809915032843e-06,
      "loss": 0.9189,
      "step": 581
    },
    {
      "epoch": 0.7808477631965117,
      "grad_norm": 1.270969125072813,
      "learning_rate": 1.2309693090277392e-06,
      "loss": 0.9054,
      "step": 582
    },
    {
      "epoch": 0.7821894260198734,
      "grad_norm": 1.2484963540929594,
      "learning_rate": 1.2167287892069662e-06,
      "loss": 0.9052,
      "step": 583
    },
    {
      "epoch": 0.7835310888432351,
      "grad_norm": 1.3514106246708202,
      "learning_rate": 1.202559700914656e-06,
      "loss": 0.9307,
      "step": 584
    },
    {
      "epoch": 0.7848727516665968,
      "grad_norm": 1.579563869491601,
      "learning_rate": 1.1884623116758121e-06,
      "loss": 0.9126,
      "step": 585
    },
    {
      "epoch": 0.7862144144899585,
      "grad_norm": 1.240608735446977,
      "learning_rate": 1.1744368876616913e-06,
      "loss": 0.8825,
      "step": 586
    },
    {
      "epoch": 0.7875560773133202,
      "grad_norm": 1.3475713796286928,
      "learning_rate": 1.1604836936847852e-06,
      "loss": 0.9356,
      "step": 587
    },
    {
      "epoch": 0.7888977401366819,
      "grad_norm": 1.3683933074436645,
      "learning_rate": 1.1466029931938182e-06,
      "loss": 0.9034,
      "step": 588
    },
    {
      "epoch": 0.7902394029600436,
      "grad_norm": 1.2934899729517133,
      "learning_rate": 1.1327950482687727e-06,
      "loss": 0.9292,
      "step": 589
    },
    {
      "epoch": 0.7915810657834053,
      "grad_norm": 1.5461960865270319,
      "learning_rate": 1.1190601196159406e-06,
      "loss": 0.8924,
      "step": 590
    },
    {
      "epoch": 0.792922728606767,
      "grad_norm": 1.377453051187854,
      "learning_rate": 1.1053984665630025e-06,
      "loss": 0.8993,
      "step": 591
    },
    {
      "epoch": 0.7942643914301287,
      "grad_norm": 1.2032585780385652,
      "learning_rate": 1.0918103470541297e-06,
      "loss": 0.9156,
      "step": 592
    },
    {
      "epoch": 0.7956060542534904,
      "grad_norm": 1.2989113276114335,
      "learning_rate": 1.0782960176451164e-06,
      "loss": 0.8946,
      "step": 593
    },
    {
      "epoch": 0.7969477170768521,
      "grad_norm": 1.244245982468071,
      "learning_rate": 1.064855733498531e-06,
      "loss": 0.9097,
      "step": 594
    },
    {
      "epoch": 0.7982893799002139,
      "grad_norm": 1.3280190656349367,
      "learning_rate": 1.051489748378905e-06,
      "loss": 0.91,
      "step": 595
    },
    {
      "epoch": 0.7996310427235755,
      "grad_norm": 1.3113261246318444,
      "learning_rate": 1.0381983146479352e-06,
      "loss": 0.8962,
      "step": 596
    },
    {
      "epoch": 0.8009727055469372,
      "grad_norm": 1.331281912168968,
      "learning_rate": 1.024981683259723e-06,
      "loss": 0.9035,
      "step": 597
    },
    {
      "epoch": 0.8023143683702989,
      "grad_norm": 1.6549084680343515,
      "learning_rate": 1.0118401037560354e-06,
      "loss": 0.9181,
      "step": 598
    },
    {
      "epoch": 0.8036560311936607,
      "grad_norm": 1.2586619576665807,
      "learning_rate": 9.987738242615924e-07,
      "loss": 0.8948,
      "step": 599
    },
    {
      "epoch": 0.8049976940170224,
      "grad_norm": 1.309953776979878,
      "learning_rate": 9.857830914793827e-07,
      "loss": 0.901,
      "step": 600
    },
    {
      "epoch": 0.806339356840384,
      "grad_norm": 1.2433181727297853,
      "learning_rate": 9.72868150686005e-07,
      "loss": 0.9124,
      "step": 601
    },
    {
      "epoch": 0.8076810196637457,
      "grad_norm": 1.1983249619659875,
      "learning_rate": 9.600292457270416e-07,
      "loss": 0.8996,
      "step": 602
    },
    {
      "epoch": 0.8090226824871075,
      "grad_norm": 1.2523970949426702,
      "learning_rate": 9.472666190124457e-07,
      "loss": 0.9258,
      "step": 603
    },
    {
      "epoch": 0.8103643453104692,
      "grad_norm": 1.1859328500508861,
      "learning_rate": 9.345805115119722e-07,
      "loss": 0.9052,
      "step": 604
    },
    {
      "epoch": 0.8117060081338309,
      "grad_norm": 1.3744025315238007,
      "learning_rate": 9.219711627506239e-07,
      "loss": 0.9139,
      "step": 605
    },
    {
      "epoch": 0.8130476709571925,
      "grad_norm": 1.192545092485379,
      "learning_rate": 9.094388108041302e-07,
      "loss": 0.9057,
      "step": 606
    },
    {
      "epoch": 0.8143893337805542,
      "grad_norm": 1.3291859242744368,
      "learning_rate": 8.969836922944536e-07,
      "loss": 0.902,
      "step": 607
    },
    {
      "epoch": 0.815730996603916,
      "grad_norm": 1.2686568646382825,
      "learning_rate": 8.846060423853198e-07,
      "loss": 0.8994,
      "step": 608
    },
    {
      "epoch": 0.8170726594272777,
      "grad_norm": 1.4200335252538632,
      "learning_rate": 8.723060947777778e-07,
      "loss": 0.9067,
      "step": 609
    },
    {
      "epoch": 0.8184143222506394,
      "grad_norm": 1.5265855250125198,
      "learning_rate": 8.600840817057877e-07,
      "loss": 0.91,
      "step": 610
    },
    {
      "epoch": 0.819755985074001,
      "grad_norm": 1.3838763518737331,
      "learning_rate": 8.479402339318372e-07,
      "loss": 0.906,
      "step": 611
    },
    {
      "epoch": 0.8210976478973628,
      "grad_norm": 1.2817254846160717,
      "learning_rate": 8.358747807425827e-07,
      "loss": 0.9106,
      "step": 612
    },
    {
      "epoch": 0.8224393107207245,
      "grad_norm": 1.3157115651741536,
      "learning_rate": 8.238879499445224e-07,
      "loss": 0.9109,
      "step": 613
    },
    {
      "epoch": 0.8237809735440862,
      "grad_norm": 2.370081298583623,
      "learning_rate": 8.119799678596918e-07,
      "loss": 0.8968,
      "step": 614
    },
    {
      "epoch": 0.8251226363674479,
      "grad_norm": 1.163618291049878,
      "learning_rate": 8.001510593213946e-07,
      "loss": 0.9041,
      "step": 615
    },
    {
      "epoch": 0.8264642991908097,
      "grad_norm": 1.1913159812321341,
      "learning_rate": 7.884014476699541e-07,
      "loss": 0.9075,
      "step": 616
    },
    {
      "epoch": 0.8278059620141713,
      "grad_norm": 1.3224358409661805,
      "learning_rate": 7.767313547484979e-07,
      "loss": 0.9177,
      "step": 617
    },
    {
      "epoch": 0.829147624837533,
      "grad_norm": 1.5777854124699655,
      "learning_rate": 7.651410008987698e-07,
      "loss": 0.9069,
      "step": 618
    },
    {
      "epoch": 0.8304892876608947,
      "grad_norm": 1.2325653285194271,
      "learning_rate": 7.536306049569686e-07,
      "loss": 0.9089,
      "step": 619
    },
    {
      "epoch": 0.8318309504842564,
      "grad_norm": 1.2072805741779278,
      "learning_rate": 7.422003842496167e-07,
      "loss": 0.9009,
      "step": 620
    },
    {
      "epoch": 0.8331726133076182,
      "grad_norm": 1.2312294895632327,
      "learning_rate": 7.308505545894567e-07,
      "loss": 0.9114,
      "step": 621
    },
    {
      "epoch": 0.8345142761309798,
      "grad_norm": 1.2389500012630505,
      "learning_rate": 7.195813302713761e-07,
      "loss": 0.8884,
      "step": 622
    },
    {
      "epoch": 0.8358559389543415,
      "grad_norm": 1.2610850320698639,
      "learning_rate": 7.083929240683618e-07,
      "loss": 0.9106,
      "step": 623
    },
    {
      "epoch": 0.8371976017777032,
      "grad_norm": 1.5973319577372662,
      "learning_rate": 6.972855472274853e-07,
      "loss": 0.9064,
      "step": 624
    },
    {
      "epoch": 0.838539264601065,
      "grad_norm": 1.3932907377877346,
      "learning_rate": 6.862594094659086e-07,
      "loss": 0.9242,
      "step": 625
    },
    {
      "epoch": 0.8398809274244267,
      "grad_norm": 1.2694477207687995,
      "learning_rate": 6.753147189669279e-07,
      "loss": 0.9001,
      "step": 626
    },
    {
      "epoch": 0.8412225902477883,
      "grad_norm": 1.4356848330613432,
      "learning_rate": 6.644516823760439e-07,
      "loss": 0.8741,
      "step": 627
    },
    {
      "epoch": 0.84256425307115,
      "grad_norm": 1.3465335502199314,
      "learning_rate": 6.536705047970566e-07,
      "loss": 0.8948,
      "step": 628
    },
    {
      "epoch": 0.8439059158945118,
      "grad_norm": 1.393718175289126,
      "learning_rate": 6.429713897881984e-07,
      "loss": 0.9163,
      "step": 629
    },
    {
      "epoch": 0.8452475787178735,
      "grad_norm": 1.2433002110204077,
      "learning_rate": 6.323545393582847e-07,
      "loss": 0.9088,
      "step": 630
    },
    {
      "epoch": 0.8465892415412352,
      "grad_norm": 1.3283975816353129,
      "learning_rate": 6.218201539629032e-07,
      "loss": 0.9233,
      "step": 631
    },
    {
      "epoch": 0.8479309043645968,
      "grad_norm": 1.211143540762075,
      "learning_rate": 6.113684325006286e-07,
      "loss": 0.9134,
      "step": 632
    },
    {
      "epoch": 0.8492725671879586,
      "grad_norm": 1.2705343275484424,
      "learning_rate": 6.009995723092655e-07,
      "loss": 0.9185,
      "step": 633
    },
    {
      "epoch": 0.8506142300113203,
      "grad_norm": 1.3957017176759934,
      "learning_rate": 5.907137691621256e-07,
      "loss": 0.9024,
      "step": 634
    },
    {
      "epoch": 0.851955892834682,
      "grad_norm": 1.3527649104045,
      "learning_rate": 5.805112172643296e-07,
      "loss": 0.9044,
      "step": 635
    },
    {
      "epoch": 0.8532975556580437,
      "grad_norm": 1.261750149081806,
      "learning_rate": 5.703921092491393e-07,
      "loss": 0.9221,
      "step": 636
    },
    {
      "epoch": 0.8546392184814053,
      "grad_norm": 1.3055595818097723,
      "learning_rate": 5.603566361743229e-07,
      "loss": 0.908,
      "step": 637
    },
    {
      "epoch": 0.8559808813047671,
      "grad_norm": 1.4280255737874952,
      "learning_rate": 5.504049875185458e-07,
      "loss": 0.8873,
      "step": 638
    },
    {
      "epoch": 0.8573225441281288,
      "grad_norm": 1.2949770398691693,
      "learning_rate": 5.405373511777939e-07,
      "loss": 0.9042,
      "step": 639
    },
    {
      "epoch": 0.8586642069514905,
      "grad_norm": 1.2916262406127108,
      "learning_rate": 5.307539134618256e-07,
      "loss": 0.8978,
      "step": 640
    },
    {
      "epoch": 0.8600058697748522,
      "grad_norm": 1.3529046928197803,
      "learning_rate": 5.21054859090655e-07,
      "loss": 0.8892,
      "step": 641
    },
    {
      "epoch": 0.861347532598214,
      "grad_norm": 1.282419973354587,
      "learning_rate": 5.114403711910631e-07,
      "loss": 0.9179,
      "step": 642
    },
    {
      "epoch": 0.8626891954215756,
      "grad_norm": 1.3420938264698856,
      "learning_rate": 5.019106312931399e-07,
      "loss": 0.9155,
      "step": 643
    },
    {
      "epoch": 0.8640308582449373,
      "grad_norm": 1.498460165413374,
      "learning_rate": 4.924658193268595e-07,
      "loss": 0.9068,
      "step": 644
    },
    {
      "epoch": 0.865372521068299,
      "grad_norm": 1.3390119903047981,
      "learning_rate": 4.831061136186787e-07,
      "loss": 0.9028,
      "step": 645
    },
    {
      "epoch": 0.8667141838916608,
      "grad_norm": 1.3139597887405536,
      "learning_rate": 4.7383169088817407e-07,
      "loss": 0.8899,
      "step": 646
    },
    {
      "epoch": 0.8680558467150224,
      "grad_norm": 1.2345562311100178,
      "learning_rate": 4.646427262447034e-07,
      "loss": 0.9113,
      "step": 647
    },
    {
      "epoch": 0.8693975095383841,
      "grad_norm": 1.2810485954457775,
      "learning_rate": 4.555393931841001e-07,
      "loss": 0.9059,
      "step": 648
    },
    {
      "epoch": 0.8707391723617458,
      "grad_norm": 1.4327167922145774,
      "learning_rate": 4.4652186358539573e-07,
      "loss": 0.8924,
      "step": 649
    },
    {
      "epoch": 0.8720808351851076,
      "grad_norm": 1.4173807953026714,
      "learning_rate": 4.3759030770757606e-07,
      "loss": 0.9062,
      "step": 650
    },
    {
      "epoch": 0.8734224980084693,
      "grad_norm": 1.2049321963818846,
      "learning_rate": 4.287448941863692e-07,
      "loss": 0.9088,
      "step": 651
    },
    {
      "epoch": 0.874764160831831,
      "grad_norm": 1.2315593362133814,
      "learning_rate": 4.1998579003105553e-07,
      "loss": 0.8958,
      "step": 652
    },
    {
      "epoch": 0.8761058236551926,
      "grad_norm": 1.2723610492335584,
      "learning_rate": 4.1131316062131944e-07,
      "loss": 0.91,
      "step": 653
    },
    {
      "epoch": 0.8774474864785543,
      "grad_norm": 1.2008077017130754,
      "learning_rate": 4.0272716970412516e-07,
      "loss": 0.9006,
      "step": 654
    },
    {
      "epoch": 0.8787891493019161,
      "grad_norm": 1.20510795448871,
      "learning_rate": 3.9422797939062505e-07,
      "loss": 0.9049,
      "step": 655
    },
    {
      "epoch": 0.8801308121252778,
      "grad_norm": 1.5762215699625823,
      "learning_rate": 3.858157501530974e-07,
      "loss": 0.8951,
      "step": 656
    },
    {
      "epoch": 0.8814724749486395,
      "grad_norm": 1.3538553053953095,
      "learning_rate": 3.7749064082191976e-07,
      "loss": 0.8873,
      "step": 657
    },
    {
      "epoch": 0.8828141377720011,
      "grad_norm": 1.3062999793490424,
      "learning_rate": 3.692528085825675e-07,
      "loss": 0.8937,
      "step": 658
    },
    {
      "epoch": 0.8841558005953629,
      "grad_norm": 1.2757504728646711,
      "learning_rate": 3.6110240897264727e-07,
      "loss": 0.9226,
      "step": 659
    },
    {
      "epoch": 0.8854974634187246,
      "grad_norm": 2.1158439115081573,
      "learning_rate": 3.53039595878959e-07,
      "loss": 0.8921,
      "step": 660
    },
    {
      "epoch": 0.8868391262420863,
      "grad_norm": 1.5194549778504565,
      "learning_rate": 3.450645215345921e-07,
      "loss": 0.9407,
      "step": 661
    },
    {
      "epoch": 0.888180789065448,
      "grad_norm": 1.1961113555111094,
      "learning_rate": 3.3717733651604967e-07,
      "loss": 0.9054,
      "step": 662
    },
    {
      "epoch": 0.8895224518888097,
      "grad_norm": 1.2448619199933089,
      "learning_rate": 3.2937818974040637e-07,
      "loss": 0.8897,
      "step": 663
    },
    {
      "epoch": 0.8908641147121714,
      "grad_norm": 1.2228819784956602,
      "learning_rate": 3.216672284624961e-07,
      "loss": 0.8962,
      "step": 664
    },
    {
      "epoch": 0.8922057775355331,
      "grad_norm": 1.2302573473318397,
      "learning_rate": 3.140445982721324e-07,
      "loss": 0.8973,
      "step": 665
    },
    {
      "epoch": 0.8935474403588948,
      "grad_norm": 1.232916261356426,
      "learning_rate": 3.0651044309136016e-07,
      "loss": 0.8963,
      "step": 666
    },
    {
      "epoch": 0.8948891031822566,
      "grad_norm": 1.3879794683807776,
      "learning_rate": 2.990649051717348e-07,
      "loss": 0.895,
      "step": 667
    },
    {
      "epoch": 0.8962307660056182,
      "grad_norm": 1.3094666189869715,
      "learning_rate": 2.917081250916415e-07,
      "loss": 0.911,
      "step": 668
    },
    {
      "epoch": 0.8975724288289799,
      "grad_norm": 2.1986551459756885,
      "learning_rate": 2.844402417536374e-07,
      "loss": 0.9189,
      "step": 669
    },
    {
      "epoch": 0.8989140916523416,
      "grad_norm": 1.247107904720917,
      "learning_rate": 2.772613923818301e-07,
      "loss": 0.9175,
      "step": 670
    },
    {
      "epoch": 0.9002557544757033,
      "grad_norm": 1.2265952393127215,
      "learning_rate": 2.701717125192865e-07,
      "loss": 0.8806,
      "step": 671
    },
    {
      "epoch": 0.9015974172990651,
      "grad_norm": 1.1972138465547026,
      "learning_rate": 2.631713360254734e-07,
      "loss": 0.8968,
      "step": 672
    },
    {
      "epoch": 0.9029390801224267,
      "grad_norm": 1.808757745752747,
      "learning_rate": 2.562603950737319e-07,
      "loss": 0.9065,
      "step": 673
    },
    {
      "epoch": 0.9042807429457884,
      "grad_norm": 1.377558178542473,
      "learning_rate": 2.494390201487795e-07,
      "loss": 0.9007,
      "step": 674
    },
    {
      "epoch": 0.9056224057691501,
      "grad_norm": 1.4113034956571244,
      "learning_rate": 2.4270734004424643e-07,
      "loss": 0.9001,
      "step": 675
    },
    {
      "epoch": 0.9069640685925119,
      "grad_norm": 1.2994213819698734,
      "learning_rate": 2.3606548186024702e-07,
      "loss": 0.9088,
      "step": 676
    },
    {
      "epoch": 0.9083057314158736,
      "grad_norm": 1.3098020957658707,
      "learning_rate": 2.295135710009755e-07,
      "loss": 0.9084,
      "step": 677
    },
    {
      "epoch": 0.9096473942392352,
      "grad_norm": 1.2442584029120756,
      "learning_rate": 2.2305173117234236e-07,
      "loss": 0.899,
      "step": 678
    },
    {
      "epoch": 0.9109890570625969,
      "grad_norm": 1.16625420459152,
      "learning_rate": 2.166800843796357e-07,
      "loss": 0.9098,
      "step": 679
    },
    {
      "epoch": 0.9123307198859587,
      "grad_norm": 1.1965287831013598,
      "learning_rate": 2.1039875092521978e-07,
      "loss": 0.8831,
      "step": 680
    },
    {
      "epoch": 0.9136723827093204,
      "grad_norm": 1.369985402269482,
      "learning_rate": 2.042078494062616e-07,
      "loss": 0.8911,
      "step": 681
    },
    {
      "epoch": 0.9150140455326821,
      "grad_norm": 1.1922261874260034,
      "learning_rate": 1.9810749671249353e-07,
      "loss": 0.8933,
      "step": 682
    },
    {
      "epoch": 0.9163557083560437,
      "grad_norm": 1.1791520718486093,
      "learning_rate": 1.920978080240049e-07,
      "loss": 0.9004,
      "step": 683
    },
    {
      "epoch": 0.9176973711794055,
      "grad_norm": 1.252972974339097,
      "learning_rate": 1.861788968090683e-07,
      "loss": 0.9109,
      "step": 684
    },
    {
      "epoch": 0.9190390340027672,
      "grad_norm": 1.210490712141151,
      "learning_rate": 1.8035087482199676e-07,
      "loss": 0.9012,
      "step": 685
    },
    {
      "epoch": 0.9203806968261289,
      "grad_norm": 1.3085615609680596,
      "learning_rate": 1.7461385210103377e-07,
      "loss": 0.9226,
      "step": 686
    },
    {
      "epoch": 0.9217223596494906,
      "grad_norm": 1.245267818500846,
      "learning_rate": 1.68967936966275e-07,
      "loss": 0.9068,
      "step": 687
    },
    {
      "epoch": 0.9230640224728522,
      "grad_norm": 1.401189629029748,
      "learning_rate": 1.6341323601762548e-07,
      "loss": 0.9151,
      "step": 688
    },
    {
      "epoch": 0.924405685296214,
      "grad_norm": 1.5915899329635883,
      "learning_rate": 1.579498541327834e-07,
      "loss": 0.8845,
      "step": 689
    },
    {
      "epoch": 0.9257473481195757,
      "grad_norm": 1.2173104542610342,
      "learning_rate": 1.5257789446526172e-07,
      "loss": 0.9003,
      "step": 690
    },
    {
      "epoch": 0.9270890109429374,
      "grad_norm": 1.2187813291272873,
      "learning_rate": 1.4729745844244302e-07,
      "loss": 0.9071,
      "step": 691
    },
    {
      "epoch": 0.9284306737662991,
      "grad_norm": 1.3906288260053012,
      "learning_rate": 1.4210864576365891e-07,
      "loss": 0.9,
      "step": 692
    },
    {
      "epoch": 0.9297723365896609,
      "grad_norm": 1.1874827126453622,
      "learning_rate": 1.3701155439831249e-07,
      "loss": 0.8927,
      "step": 693
    },
    {
      "epoch": 0.9311139994130225,
      "grad_norm": 1.2552339140761175,
      "learning_rate": 1.320062805840261e-07,
      "loss": 0.8997,
      "step": 694
    },
    {
      "epoch": 0.9324556622363842,
      "grad_norm": 1.2017347339152205,
      "learning_rate": 1.27092918824826e-07,
      "loss": 0.9131,
      "step": 695
    },
    {
      "epoch": 0.9337973250597459,
      "grad_norm": 1.3162069081963632,
      "learning_rate": 1.2227156188935552e-07,
      "loss": 0.9098,
      "step": 696
    },
    {
      "epoch": 0.9351389878831077,
      "grad_norm": 1.2473748551780643,
      "learning_rate": 1.1754230080912588e-07,
      "loss": 0.905,
      "step": 697
    },
    {
      "epoch": 0.9364806507064694,
      "grad_norm": 1.203625706135816,
      "learning_rate": 1.12905224876797e-07,
      "loss": 0.9108,
      "step": 698
    },
    {
      "epoch": 0.937822313529831,
      "grad_norm": 1.2488624887801947,
      "learning_rate": 1.0836042164448945e-07,
      "loss": 0.8915,
      "step": 699
    },
    {
      "epoch": 0.9391639763531927,
      "grad_norm": 1.2552228912980437,
      "learning_rate": 1.0390797692213516e-07,
      "loss": 0.8811,
      "step": 700
    },
    {
      "epoch": 0.9405056391765545,
      "grad_norm": 1.8280475750688994,
      "learning_rate": 9.954797477585376e-08,
      "loss": 0.9137,
      "step": 701
    },
    {
      "epoch": 0.9418473019999162,
      "grad_norm": 1.2071040198044656,
      "learning_rate": 9.528049752636714e-08,
      "loss": 0.9038,
      "step": 702
    },
    {
      "epoch": 0.9431889648232779,
      "grad_norm": 1.2674976745179707,
      "learning_rate": 9.110562574744519e-08,
      "loss": 0.9179,
      "step": 703
    },
    {
      "epoch": 0.9445306276466395,
      "grad_norm": 1.2141438354939842,
      "learning_rate": 8.702343826438364e-08,
      "loss": 0.9125,
      "step": 704
    },
    {
      "epoch": 0.9458722904700012,
      "grad_norm": 1.5780774480631814,
      "learning_rate": 8.303401215251583e-08,
      "loss": 0.8969,
      "step": 705
    },
    {
      "epoch": 0.947213953293363,
      "grad_norm": 1.310428875382451,
      "learning_rate": 7.913742273575886e-08,
      "loss": 0.8908,
      "step": 706
    },
    {
      "epoch": 0.9485556161167247,
      "grad_norm": 1.1664669903330205,
      "learning_rate": 7.533374358518974e-08,
      "loss": 0.9021,
      "step": 707
    },
    {
      "epoch": 0.9498972789400864,
      "grad_norm": 1.3412312405769142,
      "learning_rate": 7.16230465176565e-08,
      "loss": 0.9197,
      "step": 708
    },
    {
      "epoch": 0.951238941763448,
      "grad_norm": 1.1188094863277196,
      "learning_rate": 6.80054015944237e-08,
      "loss": 0.9124,
      "step": 709
    },
    {
      "epoch": 0.9525806045868098,
      "grad_norm": 1.2264327425075252,
      "learning_rate": 6.448087711984796e-08,
      "loss": 0.9218,
      "step": 710
    },
    {
      "epoch": 0.9539222674101715,
      "grad_norm": 1.2728923335730993,
      "learning_rate": 6.104953964008897e-08,
      "loss": 0.9032,
      "step": 711
    },
    {
      "epoch": 0.9552639302335332,
      "grad_norm": 1.3013816313631277,
      "learning_rate": 5.7711453941852736e-08,
      "loss": 0.9178,
      "step": 712
    },
    {
      "epoch": 0.9566055930568949,
      "grad_norm": 1.2072792472853993,
      "learning_rate": 5.446668305116865e-08,
      "loss": 0.9038,
      "step": 713
    },
    {
      "epoch": 0.9579472558802566,
      "grad_norm": 1.2293045934199316,
      "learning_rate": 5.1315288232201e-08,
      "loss": 0.9061,
      "step": 714
    },
    {
      "epoch": 0.9592889187036183,
      "grad_norm": 1.2938337912720965,
      "learning_rate": 4.825732898608826e-08,
      "loss": 0.9242,
      "step": 715
    },
    {
      "epoch": 0.96063058152698,
      "grad_norm": 1.2703267261050453,
      "learning_rate": 4.529286304982394e-08,
      "loss": 0.9166,
      "step": 716
    },
    {
      "epoch": 0.9619722443503417,
      "grad_norm": 1.1662442465906284,
      "learning_rate": 4.2421946395164174e-08,
      "loss": 0.9148,
      "step": 717
    },
    {
      "epoch": 0.9633139071737034,
      "grad_norm": 1.325544575496581,
      "learning_rate": 3.964463322757017e-08,
      "loss": 0.8958,
      "step": 718
    },
    {
      "epoch": 0.9646555699970651,
      "grad_norm": 1.3596657315473117,
      "learning_rate": 3.696097598518855e-08,
      "loss": 0.9062,
      "step": 719
    },
    {
      "epoch": 0.9659972328204268,
      "grad_norm": 1.2269416950521412,
      "learning_rate": 3.437102533785541e-08,
      "loss": 0.8801,
      "step": 720
    },
    {
      "epoch": 0.9673388956437885,
      "grad_norm": 1.2795185757363334,
      "learning_rate": 3.187483018614601e-08,
      "loss": 0.9012,
      "step": 721
    },
    {
      "epoch": 0.9686805584671502,
      "grad_norm": 1.2125366854934336,
      "learning_rate": 2.9472437660446605e-08,
      "loss": 0.9038,
      "step": 722
    },
    {
      "epoch": 0.970022221290512,
      "grad_norm": 1.2636902281710753,
      "learning_rate": 2.7163893120066288e-08,
      "loss": 0.9043,
      "step": 723
    },
    {
      "epoch": 0.9713638841138736,
      "grad_norm": 1.149530053551484,
      "learning_rate": 2.4949240152381536e-08,
      "loss": 0.9169,
      "step": 724
    },
    {
      "epoch": 0.9727055469372353,
      "grad_norm": 1.2416891050976275,
      "learning_rate": 2.2828520572011902e-08,
      "loss": 0.8992,
      "step": 725
    },
    {
      "epoch": 0.974047209760597,
      "grad_norm": 1.2065853371634816,
      "learning_rate": 2.0801774420031172e-08,
      "loss": 0.8896,
      "step": 726
    },
    {
      "epoch": 0.9753888725839588,
      "grad_norm": 1.3596342258401561,
      "learning_rate": 1.8869039963210766e-08,
      "loss": 0.9199,
      "step": 727
    },
    {
      "epoch": 0.9767305354073205,
      "grad_norm": 1.1746597148679385,
      "learning_rate": 1.7030353693298086e-08,
      "loss": 0.9148,
      "step": 728
    },
    {
      "epoch": 0.9780721982306821,
      "grad_norm": 1.4018834174480932,
      "learning_rate": 1.5285750326325953e-08,
      "loss": 0.9202,
      "step": 729
    },
    {
      "epoch": 0.9794138610540438,
      "grad_norm": 1.1702727095870482,
      "learning_rate": 1.3635262801960369e-08,
      "loss": 0.8974,
      "step": 730
    },
    {
      "epoch": 0.9807555238774056,
      "grad_norm": 1.318215740498902,
      "learning_rate": 1.2078922282873773e-08,
      "loss": 0.8951,
      "step": 731
    },
    {
      "epoch": 0.9820971867007673,
      "grad_norm": 1.221980524500853,
      "learning_rate": 1.0616758154161633e-08,
      "loss": 0.8947,
      "step": 732
    },
    {
      "epoch": 0.983438849524129,
      "grad_norm": 1.148112060120653,
      "learning_rate": 9.248798022784001e-09,
      "loss": 0.8849,
      "step": 733
    },
    {
      "epoch": 0.9847805123474906,
      "grad_norm": 1.2053514295055652,
      "learning_rate": 7.975067717045926e-09,
      "loss": 0.9117,
      "step": 734
    },
    {
      "epoch": 0.9861221751708523,
      "grad_norm": 1.3574322437625919,
      "learning_rate": 6.7955912861095155e-09,
      "loss": 0.918,
      "step": 735
    },
    {
      "epoch": 0.9874638379942141,
      "grad_norm": 1.3563827012833976,
      "learning_rate": 5.71039099953985e-09,
      "loss": 0.8986,
      "step": 736
    },
    {
      "epoch": 0.9888055008175758,
      "grad_norm": 1.16006022136295,
      "learning_rate": 4.719487346884211e-09,
      "loss": 0.887,
      "step": 737
    },
    {
      "epoch": 0.9901471636409375,
      "grad_norm": 1.3750630982606136,
      "learning_rate": 3.822899037286276e-09,
      "loss": 0.8995,
      "step": 738
    },
    {
      "epoch": 0.9914888264642991,
      "grad_norm": 2.131815418813632,
      "learning_rate": 3.0206429991314067e-09,
      "loss": 0.9266,
      "step": 739
    },
    {
      "epoch": 0.9928304892876609,
      "grad_norm": 1.206285530756936,
      "learning_rate": 2.3127343797269e-09,
      "loss": 0.8972,
      "step": 740
    },
    {
      "epoch": 0.9941721521110226,
      "grad_norm": 1.2109701207143455,
      "learning_rate": 1.6991865450188827e-09,
      "loss": 0.8776,
      "step": 741
    },
    {
      "epoch": 0.9955138149343843,
      "grad_norm": 1.2479774386935207,
      "learning_rate": 1.1800110793358521e-09,
      "loss": 0.8874,
      "step": 742
    },
    {
      "epoch": 0.996855477757746,
      "grad_norm": 1.279395306153215,
      "learning_rate": 7.552177851732901e-10,
      "loss": 0.9076,
      "step": 743
    },
    {
      "epoch": 0.9981971405811078,
      "grad_norm": 1.2092532151319206,
      "learning_rate": 4.2481468300603625e-10,
      "loss": 0.9312,
      "step": 744
    },
    {
      "epoch": 0.9995388034044694,
      "grad_norm": 1.2355454119392744,
      "learning_rate": 1.8880801113951853e-10,
      "loss": 0.887,
      "step": 745
    },
    {
      "epoch": 0.9995388034044694,
      "step": 745,
      "total_flos": 1422828632670208.0,
      "train_loss": 0.05509326341968255,
      "train_runtime": 2517.145,
      "train_samples_per_second": 151.606,
      "train_steps_per_second": 0.296
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 746,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1422828632670208.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}